diff options
| author | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
|---|---|---|
| committer | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
| commit | 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch) | |
| tree | 0bba044c4ce775e45a88a51686b5d9f90697ea9d /drivers/block | |
Linux-2.6.12-rc2
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Diffstat (limited to 'drivers/block')
81 files changed, 70243 insertions, 0 deletions
diff --git a/drivers/block/DAC960.c b/drivers/block/DAC960.c new file mode 100644 index 00000000000..423bbf2000d --- /dev/null +++ b/drivers/block/DAC960.c @@ -0,0 +1,7099 @@ +/* + + Linux Driver for Mylex DAC960/AcceleRAID/eXtremeRAID PCI RAID Controllers + + Copyright 1998-2001 by Leonard N. Zubkoff <lnz@dandelion.com> + + This program is free software; you may redistribute and/or modify it under + the terms of the GNU General Public License Version 2 as published by the + Free Software Foundation. + + This program is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY, without even the implied warranty of MERCHANTABILITY + or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License + for complete details. + +*/ + + +#define DAC960_DriverVersion "2.5.47" +#define DAC960_DriverDate "14 November 2002" + + +#include <linux/module.h> +#include <linux/types.h> +#include <linux/miscdevice.h> +#include <linux/blkdev.h> +#include <linux/bio.h> +#include <linux/completion.h> +#include <linux/delay.h> +#include <linux/genhd.h> +#include <linux/hdreg.h> +#include <linux/blkpg.h> +#include <linux/interrupt.h> +#include <linux/ioport.h> +#include <linux/mm.h> +#include <linux/slab.h> +#include <linux/proc_fs.h> +#include <linux/reboot.h> +#include <linux/spinlock.h> +#include <linux/timer.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <asm/io.h> +#include <asm/uaccess.h> +#include "DAC960.h" + +#define DAC960_GAM_MINOR 252 + + +static DAC960_Controller_T *DAC960_Controllers[DAC960_MaxControllers]; +static int DAC960_ControllerCount; +static struct proc_dir_entry *DAC960_ProcDirectoryEntry; + +static long disk_size(DAC960_Controller_T *p, int drive_nr) +{ + if (p->FirmwareType == DAC960_V1_Controller) { + if (drive_nr >= p->LogicalDriveCount) + return 0; + return p->V1.LogicalDriveInformation[drive_nr]. + LogicalDriveSize; + } else { + DAC960_V2_LogicalDeviceInfo_T *i = + p->V2.LogicalDeviceInformation[drive_nr]; + if (i == NULL) + return 0; + return i->ConfigurableDeviceSize; + } +} + +static int DAC960_open(struct inode *inode, struct file *file) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + DAC960_Controller_T *p = disk->queue->queuedata; + int drive_nr = (long)disk->private_data; + + if (p->FirmwareType == DAC960_V1_Controller) { + if (p->V1.LogicalDriveInformation[drive_nr]. + LogicalDriveState == DAC960_V1_LogicalDrive_Offline) + return -ENXIO; + } else { + DAC960_V2_LogicalDeviceInfo_T *i = + p->V2.LogicalDeviceInformation[drive_nr]; + if (!i || i->LogicalDeviceState == DAC960_V2_LogicalDevice_Offline) + return -ENXIO; + } + + check_disk_change(inode->i_bdev); + + if (!get_capacity(p->disks[drive_nr])) + return -ENXIO; + return 0; +} + +static int DAC960_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + DAC960_Controller_T *p = disk->queue->queuedata; + int drive_nr = (long)disk->private_data; + struct hd_geometry g; + struct hd_geometry __user *loc = (struct hd_geometry __user *)arg; + + if (cmd != HDIO_GETGEO || !loc) + return -EINVAL; + + if (p->FirmwareType == DAC960_V1_Controller) { + g.heads = p->V1.GeometryTranslationHeads; + g.sectors = p->V1.GeometryTranslationSectors; + g.cylinders = p->V1.LogicalDriveInformation[drive_nr]. + LogicalDriveSize / (g.heads * g.sectors); + } else { + DAC960_V2_LogicalDeviceInfo_T *i = + p->V2.LogicalDeviceInformation[drive_nr]; + switch (i->DriveGeometry) { + case DAC960_V2_Geometry_128_32: + g.heads = 128; + g.sectors = 32; + break; + case DAC960_V2_Geometry_255_63: + g.heads = 255; + g.sectors = 63; + break; + default: + DAC960_Error("Illegal Logical Device Geometry %d\n", + p, i->DriveGeometry); + return -EINVAL; + } + + g.cylinders = i->ConfigurableDeviceSize / (g.heads * g.sectors); + } + + g.start = get_start_sect(inode->i_bdev); + + return copy_to_user(loc, &g, sizeof g) ? -EFAULT : 0; +} + +static int DAC960_media_changed(struct gendisk *disk) +{ + DAC960_Controller_T *p = disk->queue->queuedata; + int drive_nr = (long)disk->private_data; + + if (!p->LogicalDriveInitiallyAccessible[drive_nr]) + return 1; + return 0; +} + +static int DAC960_revalidate_disk(struct gendisk *disk) +{ + DAC960_Controller_T *p = disk->queue->queuedata; + int unit = (long)disk->private_data; + + set_capacity(disk, disk_size(p, unit)); + return 0; +} + +static struct block_device_operations DAC960_BlockDeviceOperations = { + .owner = THIS_MODULE, + .open = DAC960_open, + .ioctl = DAC960_ioctl, + .media_changed = DAC960_media_changed, + .revalidate_disk = DAC960_revalidate_disk, +}; + + +/* + DAC960_AnnounceDriver announces the Driver Version and Date, Author's Name, + Copyright Notice, and Electronic Mail Address. +*/ + +static void DAC960_AnnounceDriver(DAC960_Controller_T *Controller) +{ + DAC960_Announce("***** DAC960 RAID Driver Version " + DAC960_DriverVersion " of " + DAC960_DriverDate " *****\n", Controller); + DAC960_Announce("Copyright 1998-2001 by Leonard N. Zubkoff " + "<lnz@dandelion.com>\n", Controller); +} + + +/* + DAC960_Failure prints a standardized error message, and then returns false. +*/ + +static boolean DAC960_Failure(DAC960_Controller_T *Controller, + unsigned char *ErrorMessage) +{ + DAC960_Error("While configuring DAC960 PCI RAID Controller at\n", + Controller); + if (Controller->IO_Address == 0) + DAC960_Error("PCI Bus %d Device %d Function %d I/O Address N/A " + "PCI Address 0x%X\n", Controller, + Controller->Bus, Controller->Device, + Controller->Function, Controller->PCI_Address); + else DAC960_Error("PCI Bus %d Device %d Function %d I/O Address " + "0x%X PCI Address 0x%X\n", Controller, + Controller->Bus, Controller->Device, + Controller->Function, Controller->IO_Address, + Controller->PCI_Address); + DAC960_Error("%s FAILED - DETACHING\n", Controller, ErrorMessage); + return false; +} + +/* + init_dma_loaf() and slice_dma_loaf() are helper functions for + aggregating the dma-mapped memory for a well-known collection of + data structures that are of different lengths. + + These routines don't guarantee any alignment. The caller must + include any space needed for alignment in the sizes of the structures + that are passed in. + */ + +static boolean init_dma_loaf(struct pci_dev *dev, struct dma_loaf *loaf, + size_t len) +{ + void *cpu_addr; + dma_addr_t dma_handle; + + cpu_addr = pci_alloc_consistent(dev, len, &dma_handle); + if (cpu_addr == NULL) + return false; + + loaf->cpu_free = loaf->cpu_base = cpu_addr; + loaf->dma_free =loaf->dma_base = dma_handle; + loaf->length = len; + memset(cpu_addr, 0, len); + return true; +} + +static void *slice_dma_loaf(struct dma_loaf *loaf, size_t len, + dma_addr_t *dma_handle) +{ + void *cpu_end = loaf->cpu_free + len; + void *cpu_addr = loaf->cpu_free; + + if (cpu_end > loaf->cpu_base + loaf->length) + BUG(); + *dma_handle = loaf->dma_free; + loaf->cpu_free = cpu_end; + loaf->dma_free += len; + return cpu_addr; +} + +static void free_dma_loaf(struct pci_dev *dev, struct dma_loaf *loaf_handle) +{ + if (loaf_handle->cpu_base != NULL) + pci_free_consistent(dev, loaf_handle->length, + loaf_handle->cpu_base, loaf_handle->dma_base); +} + + +/* + DAC960_CreateAuxiliaryStructures allocates and initializes the auxiliary + data structures for Controller. It returns true on success and false on + failure. +*/ + +static boolean DAC960_CreateAuxiliaryStructures(DAC960_Controller_T *Controller) +{ + int CommandAllocationLength, CommandAllocationGroupSize; + int CommandsRemaining = 0, CommandIdentifier, CommandGroupByteCount; + void *AllocationPointer = NULL; + void *ScatterGatherCPU = NULL; + dma_addr_t ScatterGatherDMA; + struct pci_pool *ScatterGatherPool; + void *RequestSenseCPU = NULL; + dma_addr_t RequestSenseDMA; + struct pci_pool *RequestSensePool = NULL; + + if (Controller->FirmwareType == DAC960_V1_Controller) + { + CommandAllocationLength = offsetof(DAC960_Command_T, V1.EndMarker); + CommandAllocationGroupSize = DAC960_V1_CommandAllocationGroupSize; + ScatterGatherPool = pci_pool_create("DAC960_V1_ScatterGather", + Controller->PCIDevice, + DAC960_V1_ScatterGatherLimit * sizeof(DAC960_V1_ScatterGatherSegment_T), + sizeof(DAC960_V1_ScatterGatherSegment_T), 0); + if (ScatterGatherPool == NULL) + return DAC960_Failure(Controller, + "AUXILIARY STRUCTURE CREATION (SG)"); + Controller->ScatterGatherPool = ScatterGatherPool; + } + else + { + CommandAllocationLength = offsetof(DAC960_Command_T, V2.EndMarker); + CommandAllocationGroupSize = DAC960_V2_CommandAllocationGroupSize; + ScatterGatherPool = pci_pool_create("DAC960_V2_ScatterGather", + Controller->PCIDevice, + DAC960_V2_ScatterGatherLimit * sizeof(DAC960_V2_ScatterGatherSegment_T), + sizeof(DAC960_V2_ScatterGatherSegment_T), 0); + if (ScatterGatherPool == NULL) + return DAC960_Failure(Controller, + "AUXILIARY STRUCTURE CREATION (SG)"); + RequestSensePool = pci_pool_create("DAC960_V2_RequestSense", + Controller->PCIDevice, sizeof(DAC960_SCSI_RequestSense_T), + sizeof(int), 0); + if (RequestSensePool == NULL) { + pci_pool_destroy(ScatterGatherPool); + return DAC960_Failure(Controller, + "AUXILIARY STRUCTURE CREATION (SG)"); + } + Controller->ScatterGatherPool = ScatterGatherPool; + Controller->V2.RequestSensePool = RequestSensePool; + } + Controller->CommandAllocationGroupSize = CommandAllocationGroupSize; + Controller->FreeCommands = NULL; + for (CommandIdentifier = 1; + CommandIdentifier <= Controller->DriverQueueDepth; + CommandIdentifier++) + { + DAC960_Command_T *Command; + if (--CommandsRemaining <= 0) + { + CommandsRemaining = + Controller->DriverQueueDepth - CommandIdentifier + 1; + if (CommandsRemaining > CommandAllocationGroupSize) + CommandsRemaining = CommandAllocationGroupSize; + CommandGroupByteCount = + CommandsRemaining * CommandAllocationLength; + AllocationPointer = kmalloc(CommandGroupByteCount, GFP_ATOMIC); + if (AllocationPointer == NULL) + return DAC960_Failure(Controller, + "AUXILIARY STRUCTURE CREATION"); + memset(AllocationPointer, 0, CommandGroupByteCount); + } + Command = (DAC960_Command_T *) AllocationPointer; + AllocationPointer += CommandAllocationLength; + Command->CommandIdentifier = CommandIdentifier; + Command->Controller = Controller; + Command->Next = Controller->FreeCommands; + Controller->FreeCommands = Command; + Controller->Commands[CommandIdentifier-1] = Command; + ScatterGatherCPU = pci_pool_alloc(ScatterGatherPool, SLAB_ATOMIC, + &ScatterGatherDMA); + if (ScatterGatherCPU == NULL) + return DAC960_Failure(Controller, "AUXILIARY STRUCTURE CREATION"); + + if (RequestSensePool != NULL) { + RequestSenseCPU = pci_pool_alloc(RequestSensePool, SLAB_ATOMIC, + &RequestSenseDMA); + if (RequestSenseCPU == NULL) { + pci_pool_free(ScatterGatherPool, ScatterGatherCPU, + ScatterGatherDMA); + return DAC960_Failure(Controller, + "AUXILIARY STRUCTURE CREATION"); + } + } + if (Controller->FirmwareType == DAC960_V1_Controller) { + Command->cmd_sglist = Command->V1.ScatterList; + Command->V1.ScatterGatherList = + (DAC960_V1_ScatterGatherSegment_T *)ScatterGatherCPU; + Command->V1.ScatterGatherListDMA = ScatterGatherDMA; + } else { + Command->cmd_sglist = Command->V2.ScatterList; + Command->V2.ScatterGatherList = + (DAC960_V2_ScatterGatherSegment_T *)ScatterGatherCPU; + Command->V2.ScatterGatherListDMA = ScatterGatherDMA; + Command->V2.RequestSense = + (DAC960_SCSI_RequestSense_T *)RequestSenseCPU; + Command->V2.RequestSenseDMA = RequestSenseDMA; + } + } + return true; +} + + +/* + DAC960_DestroyAuxiliaryStructures deallocates the auxiliary data + structures for Controller. +*/ + +static void DAC960_DestroyAuxiliaryStructures(DAC960_Controller_T *Controller) +{ + int i; + struct pci_pool *ScatterGatherPool = Controller->ScatterGatherPool; + struct pci_pool *RequestSensePool = NULL; + void *ScatterGatherCPU; + dma_addr_t ScatterGatherDMA; + void *RequestSenseCPU; + dma_addr_t RequestSenseDMA; + DAC960_Command_T *CommandGroup = NULL; + + + if (Controller->FirmwareType == DAC960_V2_Controller) + RequestSensePool = Controller->V2.RequestSensePool; + + Controller->FreeCommands = NULL; + for (i = 0; i < Controller->DriverQueueDepth; i++) + { + DAC960_Command_T *Command = Controller->Commands[i]; + + if (Command == NULL) + continue; + + if (Controller->FirmwareType == DAC960_V1_Controller) { + ScatterGatherCPU = (void *)Command->V1.ScatterGatherList; + ScatterGatherDMA = Command->V1.ScatterGatherListDMA; + RequestSenseCPU = NULL; + RequestSenseDMA = (dma_addr_t)0; + } else { + ScatterGatherCPU = (void *)Command->V2.ScatterGatherList; + ScatterGatherDMA = Command->V2.ScatterGatherListDMA; + RequestSenseCPU = (void *)Command->V2.RequestSense; + RequestSenseDMA = Command->V2.RequestSenseDMA; + } + if (ScatterGatherCPU != NULL) + pci_pool_free(ScatterGatherPool, ScatterGatherCPU, ScatterGatherDMA); + if (RequestSenseCPU != NULL) + pci_pool_free(RequestSensePool, RequestSenseCPU, RequestSenseDMA); + + if ((Command->CommandIdentifier + % Controller->CommandAllocationGroupSize) == 1) { + /* + * We can't free the group of commands until all of the + * request sense and scatter gather dma structures are free. + * Remember the beginning of the group, but don't free it + * until we've reached the beginning of the next group. + */ + if (CommandGroup != NULL) + kfree(CommandGroup); + CommandGroup = Command; + } + Controller->Commands[i] = NULL; + } + if (CommandGroup != NULL) + kfree(CommandGroup); + + if (Controller->CombinedStatusBuffer != NULL) + { + kfree(Controller->CombinedStatusBuffer); + Controller->CombinedStatusBuffer = NULL; + Controller->CurrentStatusBuffer = NULL; + } + + if (ScatterGatherPool != NULL) + pci_pool_destroy(ScatterGatherPool); + if (Controller->FirmwareType == DAC960_V1_Controller) return; + + if (RequestSensePool != NULL) + pci_pool_destroy(RequestSensePool); + + for (i = 0; i < DAC960_MaxLogicalDrives; i++) + if (Controller->V2.LogicalDeviceInformation[i] != NULL) + { + kfree(Controller->V2.LogicalDeviceInformation[i]); + Controller->V2.LogicalDeviceInformation[i] = NULL; + } + + for (i = 0; i < DAC960_V2_MaxPhysicalDevices; i++) + { + if (Controller->V2.PhysicalDeviceInformation[i] != NULL) + { + kfree(Controller->V2.PhysicalDeviceInformation[i]); + Controller->V2.PhysicalDeviceInformation[i] = NULL; + } + if (Controller->V2.InquiryUnitSerialNumber[i] != NULL) + { + kfree(Controller->V2.InquiryUnitSerialNumber[i]); + Controller->V2.InquiryUnitSerialNumber[i] = NULL; + } + } +} + + +/* + DAC960_V1_ClearCommand clears critical fields of Command for DAC960 V1 + Firmware Controllers. +*/ + +static inline void DAC960_V1_ClearCommand(DAC960_Command_T *Command) +{ + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + memset(CommandMailbox, 0, sizeof(DAC960_V1_CommandMailbox_T)); + Command->V1.CommandStatus = 0; +} + + +/* + DAC960_V2_ClearCommand clears critical fields of Command for DAC960 V2 + Firmware Controllers. +*/ + +static inline void DAC960_V2_ClearCommand(DAC960_Command_T *Command) +{ + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + memset(CommandMailbox, 0, sizeof(DAC960_V2_CommandMailbox_T)); + Command->V2.CommandStatus = 0; +} + + +/* + DAC960_AllocateCommand allocates a Command structure from Controller's + free list. During driver initialization, a special initialization command + has been placed on the free list to guarantee that command allocation can + never fail. +*/ + +static inline DAC960_Command_T *DAC960_AllocateCommand(DAC960_Controller_T + *Controller) +{ + DAC960_Command_T *Command = Controller->FreeCommands; + if (Command == NULL) return NULL; + Controller->FreeCommands = Command->Next; + Command->Next = NULL; + return Command; +} + + +/* + DAC960_DeallocateCommand deallocates Command, returning it to Controller's + free list. +*/ + +static inline void DAC960_DeallocateCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + + Command->Request = NULL; + Command->Next = Controller->FreeCommands; + Controller->FreeCommands = Command; +} + + +/* + DAC960_WaitForCommand waits for a wake_up on Controller's Command Wait Queue. +*/ + +static void DAC960_WaitForCommand(DAC960_Controller_T *Controller) +{ + spin_unlock_irq(&Controller->queue_lock); + __wait_event(Controller->CommandWaitQueue, Controller->FreeCommands); + spin_lock_irq(&Controller->queue_lock); +} + + +/* + DAC960_BA_QueueCommand queues Command for DAC960 BA Series Controllers. +*/ + +static void DAC960_BA_QueueCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_CommandMailbox_T *NextCommandMailbox = + Controller->V2.NextCommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + DAC960_BA_WriteCommandMailbox(NextCommandMailbox, CommandMailbox); + if (Controller->V2.PreviousCommandMailbox1->Words[0] == 0 || + Controller->V2.PreviousCommandMailbox2->Words[0] == 0) + DAC960_BA_MemoryMailboxNewCommand(ControllerBaseAddress); + Controller->V2.PreviousCommandMailbox2 = + Controller->V2.PreviousCommandMailbox1; + Controller->V2.PreviousCommandMailbox1 = NextCommandMailbox; + if (++NextCommandMailbox > Controller->V2.LastCommandMailbox) + NextCommandMailbox = Controller->V2.FirstCommandMailbox; + Controller->V2.NextCommandMailbox = NextCommandMailbox; +} + + +/* + DAC960_LP_QueueCommand queues Command for DAC960 LP Series Controllers. +*/ + +static void DAC960_LP_QueueCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_CommandMailbox_T *NextCommandMailbox = + Controller->V2.NextCommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + DAC960_LP_WriteCommandMailbox(NextCommandMailbox, CommandMailbox); + if (Controller->V2.PreviousCommandMailbox1->Words[0] == 0 || + Controller->V2.PreviousCommandMailbox2->Words[0] == 0) + DAC960_LP_MemoryMailboxNewCommand(ControllerBaseAddress); + Controller->V2.PreviousCommandMailbox2 = + Controller->V2.PreviousCommandMailbox1; + Controller->V2.PreviousCommandMailbox1 = NextCommandMailbox; + if (++NextCommandMailbox > Controller->V2.LastCommandMailbox) + NextCommandMailbox = Controller->V2.FirstCommandMailbox; + Controller->V2.NextCommandMailbox = NextCommandMailbox; +} + + +/* + DAC960_LA_QueueCommandDualMode queues Command for DAC960 LA Series + Controllers with Dual Mode Firmware. +*/ + +static void DAC960_LA_QueueCommandDualMode(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandMailbox_T *NextCommandMailbox = + Controller->V1.NextCommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + DAC960_LA_WriteCommandMailbox(NextCommandMailbox, CommandMailbox); + if (Controller->V1.PreviousCommandMailbox1->Words[0] == 0 || + Controller->V1.PreviousCommandMailbox2->Words[0] == 0) + DAC960_LA_MemoryMailboxNewCommand(ControllerBaseAddress); + Controller->V1.PreviousCommandMailbox2 = + Controller->V1.PreviousCommandMailbox1; + Controller->V1.PreviousCommandMailbox1 = NextCommandMailbox; + if (++NextCommandMailbox > Controller->V1.LastCommandMailbox) + NextCommandMailbox = Controller->V1.FirstCommandMailbox; + Controller->V1.NextCommandMailbox = NextCommandMailbox; +} + + +/* + DAC960_LA_QueueCommandSingleMode queues Command for DAC960 LA Series + Controllers with Single Mode Firmware. +*/ + +static void DAC960_LA_QueueCommandSingleMode(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandMailbox_T *NextCommandMailbox = + Controller->V1.NextCommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + DAC960_LA_WriteCommandMailbox(NextCommandMailbox, CommandMailbox); + if (Controller->V1.PreviousCommandMailbox1->Words[0] == 0 || + Controller->V1.PreviousCommandMailbox2->Words[0] == 0) + DAC960_LA_HardwareMailboxNewCommand(ControllerBaseAddress); + Controller->V1.PreviousCommandMailbox2 = + Controller->V1.PreviousCommandMailbox1; + Controller->V1.PreviousCommandMailbox1 = NextCommandMailbox; + if (++NextCommandMailbox > Controller->V1.LastCommandMailbox) + NextCommandMailbox = Controller->V1.FirstCommandMailbox; + Controller->V1.NextCommandMailbox = NextCommandMailbox; +} + + +/* + DAC960_PG_QueueCommandDualMode queues Command for DAC960 PG Series + Controllers with Dual Mode Firmware. +*/ + +static void DAC960_PG_QueueCommandDualMode(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandMailbox_T *NextCommandMailbox = + Controller->V1.NextCommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + DAC960_PG_WriteCommandMailbox(NextCommandMailbox, CommandMailbox); + if (Controller->V1.PreviousCommandMailbox1->Words[0] == 0 || + Controller->V1.PreviousCommandMailbox2->Words[0] == 0) + DAC960_PG_MemoryMailboxNewCommand(ControllerBaseAddress); + Controller->V1.PreviousCommandMailbox2 = + Controller->V1.PreviousCommandMailbox1; + Controller->V1.PreviousCommandMailbox1 = NextCommandMailbox; + if (++NextCommandMailbox > Controller->V1.LastCommandMailbox) + NextCommandMailbox = Controller->V1.FirstCommandMailbox; + Controller->V1.NextCommandMailbox = NextCommandMailbox; +} + + +/* + DAC960_PG_QueueCommandSingleMode queues Command for DAC960 PG Series + Controllers with Single Mode Firmware. +*/ + +static void DAC960_PG_QueueCommandSingleMode(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandMailbox_T *NextCommandMailbox = + Controller->V1.NextCommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + DAC960_PG_WriteCommandMailbox(NextCommandMailbox, CommandMailbox); + if (Controller->V1.PreviousCommandMailbox1->Words[0] == 0 || + Controller->V1.PreviousCommandMailbox2->Words[0] == 0) + DAC960_PG_HardwareMailboxNewCommand(ControllerBaseAddress); + Controller->V1.PreviousCommandMailbox2 = + Controller->V1.PreviousCommandMailbox1; + Controller->V1.PreviousCommandMailbox1 = NextCommandMailbox; + if (++NextCommandMailbox > Controller->V1.LastCommandMailbox) + NextCommandMailbox = Controller->V1.FirstCommandMailbox; + Controller->V1.NextCommandMailbox = NextCommandMailbox; +} + + +/* + DAC960_PD_QueueCommand queues Command for DAC960 PD Series Controllers. +*/ + +static void DAC960_PD_QueueCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + while (DAC960_PD_MailboxFullP(ControllerBaseAddress)) + udelay(1); + DAC960_PD_WriteCommandMailbox(ControllerBaseAddress, CommandMailbox); + DAC960_PD_NewCommand(ControllerBaseAddress); +} + + +/* + DAC960_P_QueueCommand queues Command for DAC960 P Series Controllers. +*/ + +static void DAC960_P_QueueCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + CommandMailbox->Common.CommandIdentifier = Command->CommandIdentifier; + switch (CommandMailbox->Common.CommandOpcode) + { + case DAC960_V1_Enquiry: + CommandMailbox->Common.CommandOpcode = DAC960_V1_Enquiry_Old; + break; + case DAC960_V1_GetDeviceState: + CommandMailbox->Common.CommandOpcode = DAC960_V1_GetDeviceState_Old; + break; + case DAC960_V1_Read: + CommandMailbox->Common.CommandOpcode = DAC960_V1_Read_Old; + DAC960_PD_To_P_TranslateReadWriteCommand(CommandMailbox); + break; + case DAC960_V1_Write: + CommandMailbox->Common.CommandOpcode = DAC960_V1_Write_Old; + DAC960_PD_To_P_TranslateReadWriteCommand(CommandMailbox); + break; + case DAC960_V1_ReadWithScatterGather: + CommandMailbox->Common.CommandOpcode = + DAC960_V1_ReadWithScatterGather_Old; + DAC960_PD_To_P_TranslateReadWriteCommand(CommandMailbox); + break; + case DAC960_V1_WriteWithScatterGather: + CommandMailbox->Common.CommandOpcode = + DAC960_V1_WriteWithScatterGather_Old; + DAC960_PD_To_P_TranslateReadWriteCommand(CommandMailbox); + break; + default: + break; + } + while (DAC960_PD_MailboxFullP(ControllerBaseAddress)) + udelay(1); + DAC960_PD_WriteCommandMailbox(ControllerBaseAddress, CommandMailbox); + DAC960_PD_NewCommand(ControllerBaseAddress); +} + + +/* + DAC960_ExecuteCommand executes Command and waits for completion. +*/ + +static void DAC960_ExecuteCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + DECLARE_COMPLETION(Completion); + unsigned long flags; + Command->Completion = &Completion; + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_QueueCommand(Command); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + + if (in_interrupt()) + return; + wait_for_completion(&Completion); +} + + +/* + DAC960_V1_ExecuteType3 executes a DAC960 V1 Firmware Controller Type 3 + Command and waits for completion. It returns true on success and false + on failure. +*/ + +static boolean DAC960_V1_ExecuteType3(DAC960_Controller_T *Controller, + DAC960_V1_CommandOpcode_T CommandOpcode, + dma_addr_t DataDMA) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandStatus_T CommandStatus; + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->Type3.CommandOpcode = CommandOpcode; + CommandMailbox->Type3.BusAddress = DataDMA; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V1.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V1_NormalCompletion); +} + + +/* + DAC960_V1_ExecuteTypeB executes a DAC960 V1 Firmware Controller Type 3B + Command and waits for completion. It returns true on success and false + on failure. +*/ + +static boolean DAC960_V1_ExecuteType3B(DAC960_Controller_T *Controller, + DAC960_V1_CommandOpcode_T CommandOpcode, + unsigned char CommandOpcode2, + dma_addr_t DataDMA) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandStatus_T CommandStatus; + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->Type3B.CommandOpcode = CommandOpcode; + CommandMailbox->Type3B.CommandOpcode2 = CommandOpcode2; + CommandMailbox->Type3B.BusAddress = DataDMA; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V1.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V1_NormalCompletion); +} + + +/* + DAC960_V1_ExecuteType3D executes a DAC960 V1 Firmware Controller Type 3D + Command and waits for completion. It returns true on success and false + on failure. +*/ + +static boolean DAC960_V1_ExecuteType3D(DAC960_Controller_T *Controller, + DAC960_V1_CommandOpcode_T CommandOpcode, + unsigned char Channel, + unsigned char TargetID, + dma_addr_t DataDMA) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandStatus_T CommandStatus; + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->Type3D.CommandOpcode = CommandOpcode; + CommandMailbox->Type3D.Channel = Channel; + CommandMailbox->Type3D.TargetID = TargetID; + CommandMailbox->Type3D.BusAddress = DataDMA; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V1.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V1_NormalCompletion); +} + + +/* + DAC960_V2_GeneralInfo executes a DAC960 V2 Firmware General Information + Reading IOCTL Command and waits for completion. It returns true on success + and false on failure. + + Return data in The controller's HealthStatusBuffer, which is dma-able memory +*/ + +static boolean DAC960_V2_GeneralInfo(DAC960_Controller_T *Controller) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_CommandStatus_T CommandStatus; + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->Common.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->Common.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->Common.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->Common.DataTransferSize = sizeof(DAC960_V2_HealthStatusBuffer_T); + CommandMailbox->Common.IOCTL_Opcode = DAC960_V2_GetHealthStatus; + CommandMailbox->Common.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.HealthStatusBufferDMA; + CommandMailbox->Common.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->Common.DataTransferSize; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V2.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V2_NormalCompletion); +} + + +/* + DAC960_V2_ControllerInfo executes a DAC960 V2 Firmware Controller + Information Reading IOCTL Command and waits for completion. It returns + true on success and false on failure. + + Data is returned in the controller's V2.NewControllerInformation dma-able + memory buffer. +*/ + +static boolean DAC960_V2_NewControllerInfo(DAC960_Controller_T *Controller) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_CommandStatus_T CommandStatus; + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->ControllerInfo.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->ControllerInfo.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->ControllerInfo.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->ControllerInfo.DataTransferSize = sizeof(DAC960_V2_ControllerInfo_T); + CommandMailbox->ControllerInfo.ControllerNumber = 0; + CommandMailbox->ControllerInfo.IOCTL_Opcode = DAC960_V2_GetControllerInfo; + CommandMailbox->ControllerInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewControllerInformationDMA; + CommandMailbox->ControllerInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->ControllerInfo.DataTransferSize; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V2.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V2_NormalCompletion); +} + + +/* + DAC960_V2_LogicalDeviceInfo executes a DAC960 V2 Firmware Controller Logical + Device Information Reading IOCTL Command and waits for completion. It + returns true on success and false on failure. + + Data is returned in the controller's V2.NewLogicalDeviceInformation +*/ + +static boolean DAC960_V2_NewLogicalDeviceInfo(DAC960_Controller_T *Controller, + unsigned short LogicalDeviceNumber) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_CommandStatus_T CommandStatus; + + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->LogicalDeviceInfo.CommandOpcode = + DAC960_V2_IOCTL; + CommandMailbox->LogicalDeviceInfo.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->LogicalDeviceInfo.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->LogicalDeviceInfo.DataTransferSize = + sizeof(DAC960_V2_LogicalDeviceInfo_T); + CommandMailbox->LogicalDeviceInfo.LogicalDevice.LogicalDeviceNumber = + LogicalDeviceNumber; + CommandMailbox->LogicalDeviceInfo.IOCTL_Opcode = DAC960_V2_GetLogicalDeviceInfoValid; + CommandMailbox->LogicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewLogicalDeviceInformationDMA; + CommandMailbox->LogicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->LogicalDeviceInfo.DataTransferSize; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V2.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V2_NormalCompletion); +} + + +/* + DAC960_V2_PhysicalDeviceInfo executes a DAC960 V2 Firmware Controller "Read + Physical Device Information" IOCTL Command and waits for completion. It + returns true on success and false on failure. + + The Channel, TargetID, LogicalUnit arguments should be 0 the first time + this function is called for a given controller. This will return data + for the "first" device on that controller. The returned data includes a + Channel, TargetID, LogicalUnit that can be passed in to this routine to + get data for the NEXT device on that controller. + + Data is stored in the controller's V2.NewPhysicalDeviceInfo dma-able + memory buffer. + +*/ + +static boolean DAC960_V2_NewPhysicalDeviceInfo(DAC960_Controller_T *Controller, + unsigned char Channel, + unsigned char TargetID, + unsigned char LogicalUnit) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_CommandStatus_T CommandStatus; + + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->PhysicalDeviceInfo.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->PhysicalDeviceInfo.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->PhysicalDeviceInfo.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->PhysicalDeviceInfo.DataTransferSize = + sizeof(DAC960_V2_PhysicalDeviceInfo_T); + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.LogicalUnit = LogicalUnit; + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.TargetID = TargetID; + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.Channel = Channel; + CommandMailbox->PhysicalDeviceInfo.IOCTL_Opcode = + DAC960_V2_GetPhysicalDeviceInfoValid; + CommandMailbox->PhysicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewPhysicalDeviceInformationDMA; + CommandMailbox->PhysicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->PhysicalDeviceInfo.DataTransferSize; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V2.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V2_NormalCompletion); +} + + +static void DAC960_V2_ConstructNewUnitSerialNumber( + DAC960_Controller_T *Controller, + DAC960_V2_CommandMailbox_T *CommandMailbox, int Channel, int TargetID, + int LogicalUnit) +{ + CommandMailbox->SCSI_10.CommandOpcode = DAC960_V2_SCSI_10_Passthru; + CommandMailbox->SCSI_10.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->SCSI_10.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->SCSI_10.DataTransferSize = + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T); + CommandMailbox->SCSI_10.PhysicalDevice.LogicalUnit = LogicalUnit; + CommandMailbox->SCSI_10.PhysicalDevice.TargetID = TargetID; + CommandMailbox->SCSI_10.PhysicalDevice.Channel = Channel; + CommandMailbox->SCSI_10.CDBLength = 6; + CommandMailbox->SCSI_10.SCSI_CDB[0] = 0x12; /* INQUIRY */ + CommandMailbox->SCSI_10.SCSI_CDB[1] = 1; /* EVPD = 1 */ + CommandMailbox->SCSI_10.SCSI_CDB[2] = 0x80; /* Page Code */ + CommandMailbox->SCSI_10.SCSI_CDB[3] = 0; /* Reserved */ + CommandMailbox->SCSI_10.SCSI_CDB[4] = + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T); + CommandMailbox->SCSI_10.SCSI_CDB[5] = 0; /* Control */ + CommandMailbox->SCSI_10.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewInquiryUnitSerialNumberDMA; + CommandMailbox->SCSI_10.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->SCSI_10.DataTransferSize; +} + + +/* + DAC960_V2_NewUnitSerialNumber executes an SCSI pass-through + Inquiry command to a SCSI device identified by Channel number, + Target id, Logical Unit Number. This function Waits for completion + of the command. + + The return data includes Unit Serial Number information for the + specified device. + + Data is stored in the controller's V2.NewPhysicalDeviceInfo dma-able + memory buffer. +*/ + +static boolean DAC960_V2_NewInquiryUnitSerialNumber(DAC960_Controller_T *Controller, + int Channel, int TargetID, int LogicalUnit) +{ + DAC960_Command_T *Command; + DAC960_V2_CommandMailbox_T *CommandMailbox; + DAC960_V2_CommandStatus_T CommandStatus; + + Command = DAC960_AllocateCommand(Controller); + CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + + DAC960_V2_ConstructNewUnitSerialNumber(Controller, CommandMailbox, + Channel, TargetID, LogicalUnit); + + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V2.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V2_NormalCompletion); +} + + +/* + DAC960_V2_DeviceOperation executes a DAC960 V2 Firmware Controller Device + Operation IOCTL Command and waits for completion. It returns true on + success and false on failure. +*/ + +static boolean DAC960_V2_DeviceOperation(DAC960_Controller_T *Controller, + DAC960_V2_IOCTL_Opcode_T IOCTL_Opcode, + DAC960_V2_OperationDevice_T + OperationDevice) +{ + DAC960_Command_T *Command = DAC960_AllocateCommand(Controller); + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_CommandStatus_T CommandStatus; + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox->DeviceOperation.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->DeviceOperation.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->DeviceOperation.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->DeviceOperation.IOCTL_Opcode = IOCTL_Opcode; + CommandMailbox->DeviceOperation.OperationDevice = OperationDevice; + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V2.CommandStatus; + DAC960_DeallocateCommand(Command); + return (CommandStatus == DAC960_V2_NormalCompletion); +} + + +/* + DAC960_V1_EnableMemoryMailboxInterface enables the Memory Mailbox Interface + for DAC960 V1 Firmware Controllers. + + PD and P controller types have no memory mailbox, but still need the + other dma mapped memory. +*/ + +static boolean DAC960_V1_EnableMemoryMailboxInterface(DAC960_Controller_T + *Controller) +{ + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_HardwareType_T hw_type = Controller->HardwareType; + struct pci_dev *PCI_Device = Controller->PCIDevice; + struct dma_loaf *DmaPages = &Controller->DmaPages; + size_t DmaPagesSize; + size_t CommandMailboxesSize; + size_t StatusMailboxesSize; + + DAC960_V1_CommandMailbox_T *CommandMailboxesMemory; + dma_addr_t CommandMailboxesMemoryDMA; + + DAC960_V1_StatusMailbox_T *StatusMailboxesMemory; + dma_addr_t StatusMailboxesMemoryDMA; + + DAC960_V1_CommandMailbox_T CommandMailbox; + DAC960_V1_CommandStatus_T CommandStatus; + int TimeoutCounter; + int i; + + + if (pci_set_dma_mask(Controller->PCIDevice, DAC690_V1_PciDmaMask)) + return DAC960_Failure(Controller, "DMA mask out of range"); + Controller->BounceBufferLimit = DAC690_V1_PciDmaMask; + + if ((hw_type == DAC960_PD_Controller) || (hw_type == DAC960_P_Controller)) { + CommandMailboxesSize = 0; + StatusMailboxesSize = 0; + } else { + CommandMailboxesSize = DAC960_V1_CommandMailboxCount * sizeof(DAC960_V1_CommandMailbox_T); + StatusMailboxesSize = DAC960_V1_StatusMailboxCount * sizeof(DAC960_V1_StatusMailbox_T); + } + DmaPagesSize = CommandMailboxesSize + StatusMailboxesSize + + sizeof(DAC960_V1_DCDB_T) + sizeof(DAC960_V1_Enquiry_T) + + sizeof(DAC960_V1_ErrorTable_T) + sizeof(DAC960_V1_EventLogEntry_T) + + sizeof(DAC960_V1_RebuildProgress_T) + + sizeof(DAC960_V1_LogicalDriveInformationArray_T) + + sizeof(DAC960_V1_BackgroundInitializationStatus_T) + + sizeof(DAC960_V1_DeviceState_T) + sizeof(DAC960_SCSI_Inquiry_T) + + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T); + + if (!init_dma_loaf(PCI_Device, DmaPages, DmaPagesSize)) + return false; + + + if ((hw_type == DAC960_PD_Controller) || (hw_type == DAC960_P_Controller)) + goto skip_mailboxes; + + CommandMailboxesMemory = slice_dma_loaf(DmaPages, + CommandMailboxesSize, &CommandMailboxesMemoryDMA); + + /* These are the base addresses for the command memory mailbox array */ + Controller->V1.FirstCommandMailbox = CommandMailboxesMemory; + Controller->V1.FirstCommandMailboxDMA = CommandMailboxesMemoryDMA; + + CommandMailboxesMemory += DAC960_V1_CommandMailboxCount - 1; + Controller->V1.LastCommandMailbox = CommandMailboxesMemory; + Controller->V1.NextCommandMailbox = Controller->V1.FirstCommandMailbox; + Controller->V1.PreviousCommandMailbox1 = Controller->V1.LastCommandMailbox; + Controller->V1.PreviousCommandMailbox2 = + Controller->V1.LastCommandMailbox - 1; + + /* These are the base addresses for the status memory mailbox array */ + StatusMailboxesMemory = slice_dma_loaf(DmaPages, + StatusMailboxesSize, &StatusMailboxesMemoryDMA); + + Controller->V1.FirstStatusMailbox = StatusMailboxesMemory; + Controller->V1.FirstStatusMailboxDMA = StatusMailboxesMemoryDMA; + StatusMailboxesMemory += DAC960_V1_StatusMailboxCount - 1; + Controller->V1.LastStatusMailbox = StatusMailboxesMemory; + Controller->V1.NextStatusMailbox = Controller->V1.FirstStatusMailbox; + +skip_mailboxes: + Controller->V1.MonitoringDCDB = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_DCDB_T), + &Controller->V1.MonitoringDCDB_DMA); + + Controller->V1.NewEnquiry = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_Enquiry_T), + &Controller->V1.NewEnquiryDMA); + + Controller->V1.NewErrorTable = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_ErrorTable_T), + &Controller->V1.NewErrorTableDMA); + + Controller->V1.EventLogEntry = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_EventLogEntry_T), + &Controller->V1.EventLogEntryDMA); + + Controller->V1.RebuildProgress = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_RebuildProgress_T), + &Controller->V1.RebuildProgressDMA); + + Controller->V1.NewLogicalDriveInformation = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_LogicalDriveInformationArray_T), + &Controller->V1.NewLogicalDriveInformationDMA); + + Controller->V1.BackgroundInitializationStatus = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_BackgroundInitializationStatus_T), + &Controller->V1.BackgroundInitializationStatusDMA); + + Controller->V1.NewDeviceState = slice_dma_loaf(DmaPages, + sizeof(DAC960_V1_DeviceState_T), + &Controller->V1.NewDeviceStateDMA); + + Controller->V1.NewInquiryStandardData = slice_dma_loaf(DmaPages, + sizeof(DAC960_SCSI_Inquiry_T), + &Controller->V1.NewInquiryStandardDataDMA); + + Controller->V1.NewInquiryUnitSerialNumber = slice_dma_loaf(DmaPages, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T), + &Controller->V1.NewInquiryUnitSerialNumberDMA); + + if ((hw_type == DAC960_PD_Controller) || (hw_type == DAC960_P_Controller)) + return true; + + /* Enable the Memory Mailbox Interface. */ + Controller->V1.DualModeMemoryMailboxInterface = true; + CommandMailbox.TypeX.CommandOpcode = 0x2B; + CommandMailbox.TypeX.CommandIdentifier = 0; + CommandMailbox.TypeX.CommandOpcode2 = 0x14; + CommandMailbox.TypeX.CommandMailboxesBusAddress = + Controller->V1.FirstCommandMailboxDMA; + CommandMailbox.TypeX.StatusMailboxesBusAddress = + Controller->V1.FirstStatusMailboxDMA; +#define TIMEOUT_COUNT 1000000 + + for (i = 0; i < 2; i++) + switch (Controller->HardwareType) + { + case DAC960_LA_Controller: + TimeoutCounter = TIMEOUT_COUNT; + while (--TimeoutCounter >= 0) + { + if (!DAC960_LA_HardwareMailboxFullP(ControllerBaseAddress)) + break; + udelay(10); + } + if (TimeoutCounter < 0) return false; + DAC960_LA_WriteHardwareMailbox(ControllerBaseAddress, &CommandMailbox); + DAC960_LA_HardwareMailboxNewCommand(ControllerBaseAddress); + TimeoutCounter = TIMEOUT_COUNT; + while (--TimeoutCounter >= 0) + { + if (DAC960_LA_HardwareMailboxStatusAvailableP( + ControllerBaseAddress)) + break; + udelay(10); + } + if (TimeoutCounter < 0) return false; + CommandStatus = DAC960_LA_ReadStatusRegister(ControllerBaseAddress); + DAC960_LA_AcknowledgeHardwareMailboxInterrupt(ControllerBaseAddress); + DAC960_LA_AcknowledgeHardwareMailboxStatus(ControllerBaseAddress); + if (CommandStatus == DAC960_V1_NormalCompletion) return true; + Controller->V1.DualModeMemoryMailboxInterface = false; + CommandMailbox.TypeX.CommandOpcode2 = 0x10; + break; + case DAC960_PG_Controller: + TimeoutCounter = TIMEOUT_COUNT; + while (--TimeoutCounter >= 0) + { + if (!DAC960_PG_HardwareMailboxFullP(ControllerBaseAddress)) + break; + udelay(10); + } + if (TimeoutCounter < 0) return false; + DAC960_PG_WriteHardwareMailbox(ControllerBaseAddress, &CommandMailbox); + DAC960_PG_HardwareMailboxNewCommand(ControllerBaseAddress); + + TimeoutCounter = TIMEOUT_COUNT; + while (--TimeoutCounter >= 0) + { + if (DAC960_PG_HardwareMailboxStatusAvailableP( + ControllerBaseAddress)) + break; + udelay(10); + } + if (TimeoutCounter < 0) return false; + CommandStatus = DAC960_PG_ReadStatusRegister(ControllerBaseAddress); + DAC960_PG_AcknowledgeHardwareMailboxInterrupt(ControllerBaseAddress); + DAC960_PG_AcknowledgeHardwareMailboxStatus(ControllerBaseAddress); + if (CommandStatus == DAC960_V1_NormalCompletion) return true; + Controller->V1.DualModeMemoryMailboxInterface = false; + CommandMailbox.TypeX.CommandOpcode2 = 0x10; + break; + default: + DAC960_Failure(Controller, "Unknown Controller Type\n"); + break; + } + return false; +} + + +/* + DAC960_V2_EnableMemoryMailboxInterface enables the Memory Mailbox Interface + for DAC960 V2 Firmware Controllers. + + Aggregate the space needed for the controller's memory mailbox and + the other data structures that will be targets of dma transfers with + the controller. Allocate a dma-mapped region of memory to hold these + structures. Then, save CPU pointers and dma_addr_t values to reference + the structures that are contained in that region. +*/ + +static boolean DAC960_V2_EnableMemoryMailboxInterface(DAC960_Controller_T + *Controller) +{ + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + struct pci_dev *PCI_Device = Controller->PCIDevice; + struct dma_loaf *DmaPages = &Controller->DmaPages; + size_t DmaPagesSize; + size_t CommandMailboxesSize; + size_t StatusMailboxesSize; + + DAC960_V2_CommandMailbox_T *CommandMailboxesMemory; + dma_addr_t CommandMailboxesMemoryDMA; + + DAC960_V2_StatusMailbox_T *StatusMailboxesMemory; + dma_addr_t StatusMailboxesMemoryDMA; + + DAC960_V2_CommandMailbox_T *CommandMailbox; + dma_addr_t CommandMailboxDMA; + DAC960_V2_CommandStatus_T CommandStatus; + + if (pci_set_dma_mask(Controller->PCIDevice, DAC690_V2_PciDmaMask)) + return DAC960_Failure(Controller, "DMA mask out of range"); + Controller->BounceBufferLimit = DAC690_V2_PciDmaMask; + + /* This is a temporary dma mapping, used only in the scope of this function */ + CommandMailbox = + (DAC960_V2_CommandMailbox_T *)pci_alloc_consistent( PCI_Device, + sizeof(DAC960_V2_CommandMailbox_T), &CommandMailboxDMA); + if (CommandMailbox == NULL) + return false; + + CommandMailboxesSize = DAC960_V2_CommandMailboxCount * sizeof(DAC960_V2_CommandMailbox_T); + StatusMailboxesSize = DAC960_V2_StatusMailboxCount * sizeof(DAC960_V2_StatusMailbox_T); + DmaPagesSize = + CommandMailboxesSize + StatusMailboxesSize + + sizeof(DAC960_V2_HealthStatusBuffer_T) + + sizeof(DAC960_V2_ControllerInfo_T) + + sizeof(DAC960_V2_LogicalDeviceInfo_T) + + sizeof(DAC960_V2_PhysicalDeviceInfo_T) + + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T) + + sizeof(DAC960_V2_Event_T) + + sizeof(DAC960_V2_PhysicalToLogicalDevice_T); + + if (!init_dma_loaf(PCI_Device, DmaPages, DmaPagesSize)) { + pci_free_consistent(PCI_Device, sizeof(DAC960_V2_CommandMailbox_T), + CommandMailbox, CommandMailboxDMA); + return false; + } + + CommandMailboxesMemory = slice_dma_loaf(DmaPages, + CommandMailboxesSize, &CommandMailboxesMemoryDMA); + + /* These are the base addresses for the command memory mailbox array */ + Controller->V2.FirstCommandMailbox = CommandMailboxesMemory; + Controller->V2.FirstCommandMailboxDMA = CommandMailboxesMemoryDMA; + + CommandMailboxesMemory += DAC960_V2_CommandMailboxCount - 1; + Controller->V2.LastCommandMailbox = CommandMailboxesMemory; + Controller->V2.NextCommandMailbox = Controller->V2.FirstCommandMailbox; + Controller->V2.PreviousCommandMailbox1 = Controller->V2.LastCommandMailbox; + Controller->V2.PreviousCommandMailbox2 = + Controller->V2.LastCommandMailbox - 1; + + /* These are the base addresses for the status memory mailbox array */ + StatusMailboxesMemory = slice_dma_loaf(DmaPages, + StatusMailboxesSize, &StatusMailboxesMemoryDMA); + + Controller->V2.FirstStatusMailbox = StatusMailboxesMemory; + Controller->V2.FirstStatusMailboxDMA = StatusMailboxesMemoryDMA; + StatusMailboxesMemory += DAC960_V2_StatusMailboxCount - 1; + Controller->V2.LastStatusMailbox = StatusMailboxesMemory; + Controller->V2.NextStatusMailbox = Controller->V2.FirstStatusMailbox; + + Controller->V2.HealthStatusBuffer = slice_dma_loaf(DmaPages, + sizeof(DAC960_V2_HealthStatusBuffer_T), + &Controller->V2.HealthStatusBufferDMA); + + Controller->V2.NewControllerInformation = slice_dma_loaf(DmaPages, + sizeof(DAC960_V2_ControllerInfo_T), + &Controller->V2.NewControllerInformationDMA); + + Controller->V2.NewLogicalDeviceInformation = slice_dma_loaf(DmaPages, + sizeof(DAC960_V2_LogicalDeviceInfo_T), + &Controller->V2.NewLogicalDeviceInformationDMA); + + Controller->V2.NewPhysicalDeviceInformation = slice_dma_loaf(DmaPages, + sizeof(DAC960_V2_PhysicalDeviceInfo_T), + &Controller->V2.NewPhysicalDeviceInformationDMA); + + Controller->V2.NewInquiryUnitSerialNumber = slice_dma_loaf(DmaPages, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T), + &Controller->V2.NewInquiryUnitSerialNumberDMA); + + Controller->V2.Event = slice_dma_loaf(DmaPages, + sizeof(DAC960_V2_Event_T), + &Controller->V2.EventDMA); + + Controller->V2.PhysicalToLogicalDevice = slice_dma_loaf(DmaPages, + sizeof(DAC960_V2_PhysicalToLogicalDevice_T), + &Controller->V2.PhysicalToLogicalDeviceDMA); + + /* + Enable the Memory Mailbox Interface. + + I don't know why we can't just use one of the memory mailboxes + we just allocated to do this, instead of using this temporary one. + Try this change later. + */ + memset(CommandMailbox, 0, sizeof(DAC960_V2_CommandMailbox_T)); + CommandMailbox->SetMemoryMailbox.CommandIdentifier = 1; + CommandMailbox->SetMemoryMailbox.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->SetMemoryMailbox.CommandControlBits.NoAutoRequestSense = true; + CommandMailbox->SetMemoryMailbox.FirstCommandMailboxSizeKB = + (DAC960_V2_CommandMailboxCount * sizeof(DAC960_V2_CommandMailbox_T)) >> 10; + CommandMailbox->SetMemoryMailbox.FirstStatusMailboxSizeKB = + (DAC960_V2_StatusMailboxCount * sizeof(DAC960_V2_StatusMailbox_T)) >> 10; + CommandMailbox->SetMemoryMailbox.SecondCommandMailboxSizeKB = 0; + CommandMailbox->SetMemoryMailbox.SecondStatusMailboxSizeKB = 0; + CommandMailbox->SetMemoryMailbox.RequestSenseSize = 0; + CommandMailbox->SetMemoryMailbox.IOCTL_Opcode = DAC960_V2_SetMemoryMailbox; + CommandMailbox->SetMemoryMailbox.HealthStatusBufferSizeKB = 1; + CommandMailbox->SetMemoryMailbox.HealthStatusBufferBusAddress = + Controller->V2.HealthStatusBufferDMA; + CommandMailbox->SetMemoryMailbox.FirstCommandMailboxBusAddress = + Controller->V2.FirstCommandMailboxDMA; + CommandMailbox->SetMemoryMailbox.FirstStatusMailboxBusAddress = + Controller->V2.FirstStatusMailboxDMA; + switch (Controller->HardwareType) + { + case DAC960_BA_Controller: + while (DAC960_BA_HardwareMailboxFullP(ControllerBaseAddress)) + udelay(1); + DAC960_BA_WriteHardwareMailbox(ControllerBaseAddress, CommandMailboxDMA); + DAC960_BA_HardwareMailboxNewCommand(ControllerBaseAddress); + while (!DAC960_BA_HardwareMailboxStatusAvailableP(ControllerBaseAddress)) + udelay(1); + CommandStatus = DAC960_BA_ReadCommandStatus(ControllerBaseAddress); + DAC960_BA_AcknowledgeHardwareMailboxInterrupt(ControllerBaseAddress); + DAC960_BA_AcknowledgeHardwareMailboxStatus(ControllerBaseAddress); + break; + case DAC960_LP_Controller: + while (DAC960_LP_HardwareMailboxFullP(ControllerBaseAddress)) + udelay(1); + DAC960_LP_WriteHardwareMailbox(ControllerBaseAddress, CommandMailboxDMA); + DAC960_LP_HardwareMailboxNewCommand(ControllerBaseAddress); + while (!DAC960_LP_HardwareMailboxStatusAvailableP(ControllerBaseAddress)) + udelay(1); + CommandStatus = DAC960_LP_ReadCommandStatus(ControllerBaseAddress); + DAC960_LP_AcknowledgeHardwareMailboxInterrupt(ControllerBaseAddress); + DAC960_LP_AcknowledgeHardwareMailboxStatus(ControllerBaseAddress); + break; + default: + DAC960_Failure(Controller, "Unknown Controller Type\n"); + CommandStatus = DAC960_V2_AbormalCompletion; + break; + } + pci_free_consistent(PCI_Device, sizeof(DAC960_V2_CommandMailbox_T), + CommandMailbox, CommandMailboxDMA); + return (CommandStatus == DAC960_V2_NormalCompletion); +} + + +/* + DAC960_V1_ReadControllerConfiguration reads the Configuration Information + from DAC960 V1 Firmware Controllers and initializes the Controller structure. +*/ + +static boolean DAC960_V1_ReadControllerConfiguration(DAC960_Controller_T + *Controller) +{ + DAC960_V1_Enquiry2_T *Enquiry2; + dma_addr_t Enquiry2DMA; + DAC960_V1_Config2_T *Config2; + dma_addr_t Config2DMA; + int LogicalDriveNumber, Channel, TargetID; + struct dma_loaf local_dma; + + if (!init_dma_loaf(Controller->PCIDevice, &local_dma, + sizeof(DAC960_V1_Enquiry2_T) + sizeof(DAC960_V1_Config2_T))) + return DAC960_Failure(Controller, "LOGICAL DEVICE ALLOCATION"); + + Enquiry2 = slice_dma_loaf(&local_dma, sizeof(DAC960_V1_Enquiry2_T), &Enquiry2DMA); + Config2 = slice_dma_loaf(&local_dma, sizeof(DAC960_V1_Config2_T), &Config2DMA); + + if (!DAC960_V1_ExecuteType3(Controller, DAC960_V1_Enquiry, + Controller->V1.NewEnquiryDMA)) { + free_dma_loaf(Controller->PCIDevice, &local_dma); + return DAC960_Failure(Controller, "ENQUIRY"); + } + memcpy(&Controller->V1.Enquiry, Controller->V1.NewEnquiry, + sizeof(DAC960_V1_Enquiry_T)); + + if (!DAC960_V1_ExecuteType3(Controller, DAC960_V1_Enquiry2, Enquiry2DMA)) { + free_dma_loaf(Controller->PCIDevice, &local_dma); + return DAC960_Failure(Controller, "ENQUIRY2"); + } + + if (!DAC960_V1_ExecuteType3(Controller, DAC960_V1_ReadConfig2, Config2DMA)) { + free_dma_loaf(Controller->PCIDevice, &local_dma); + return DAC960_Failure(Controller, "READ CONFIG2"); + } + + if (!DAC960_V1_ExecuteType3(Controller, DAC960_V1_GetLogicalDriveInformation, + Controller->V1.NewLogicalDriveInformationDMA)) { + free_dma_loaf(Controller->PCIDevice, &local_dma); + return DAC960_Failure(Controller, "GET LOGICAL DRIVE INFORMATION"); + } + memcpy(&Controller->V1.LogicalDriveInformation, + Controller->V1.NewLogicalDriveInformation, + sizeof(DAC960_V1_LogicalDriveInformationArray_T)); + + for (Channel = 0; Channel < Enquiry2->ActualChannels; Channel++) + for (TargetID = 0; TargetID < Enquiry2->MaxTargets; TargetID++) { + if (!DAC960_V1_ExecuteType3D(Controller, DAC960_V1_GetDeviceState, + Channel, TargetID, + Controller->V1.NewDeviceStateDMA)) { + free_dma_loaf(Controller->PCIDevice, &local_dma); + return DAC960_Failure(Controller, "GET DEVICE STATE"); + } + memcpy(&Controller->V1.DeviceState[Channel][TargetID], + Controller->V1.NewDeviceState, sizeof(DAC960_V1_DeviceState_T)); + } + /* + Initialize the Controller Model Name and Full Model Name fields. + */ + switch (Enquiry2->HardwareID.SubModel) + { + case DAC960_V1_P_PD_PU: + if (Enquiry2->SCSICapability.BusSpeed == DAC960_V1_Ultra) + strcpy(Controller->ModelName, "DAC960PU"); + else strcpy(Controller->ModelName, "DAC960PD"); + break; + case DAC960_V1_PL: + strcpy(Controller->ModelName, "DAC960PL"); + break; + case DAC960_V1_PG: + strcpy(Controller->ModelName, "DAC960PG"); + break; + case DAC960_V1_PJ: + strcpy(Controller->ModelName, "DAC960PJ"); + break; + case DAC960_V1_PR: + strcpy(Controller->ModelName, "DAC960PR"); + break; + case DAC960_V1_PT: + strcpy(Controller->ModelName, "DAC960PT"); + break; + case DAC960_V1_PTL0: + strcpy(Controller->ModelName, "DAC960PTL0"); + break; + case DAC960_V1_PRL: + strcpy(Controller->ModelName, "DAC960PRL"); + break; + case DAC960_V1_PTL1: + strcpy(Controller->ModelName, "DAC960PTL1"); + break; + case DAC960_V1_1164P: + strcpy(Controller->ModelName, "DAC1164P"); + break; + default: + free_dma_loaf(Controller->PCIDevice, &local_dma); + return DAC960_Failure(Controller, "MODEL VERIFICATION"); + } + strcpy(Controller->FullModelName, "Mylex "); + strcat(Controller->FullModelName, Controller->ModelName); + /* + Initialize the Controller Firmware Version field and verify that it + is a supported firmware version. The supported firmware versions are: + + DAC1164P 5.06 and above + DAC960PTL/PRL/PJ/PG 4.06 and above + DAC960PU/PD/PL 3.51 and above + DAC960PU/PD/PL/P 2.73 and above + */ +#if defined(CONFIG_ALPHA) + /* + DEC Alpha machines were often equipped with DAC960 cards that were + OEMed from Mylex, and had their own custom firmware. Version 2.70, + the last custom FW revision to be released by DEC for these older + controllers, appears to work quite well with this driver. + + Cards tested successfully were several versions each of the PD and + PU, called by DEC the KZPSC and KZPAC, respectively, and having + the Manufacturer Numbers (from Mylex), usually on a sticker on the + back of the board, of: + + KZPSC: D040347 (1-channel) or D040348 (2-channel) or D040349 (3-channel) + KZPAC: D040395 (1-channel) or D040396 (2-channel) or D040397 (3-channel) + */ +# define FIRMWARE_27X "2.70" +#else +# define FIRMWARE_27X "2.73" +#endif + + if (Enquiry2->FirmwareID.MajorVersion == 0) + { + Enquiry2->FirmwareID.MajorVersion = + Controller->V1.Enquiry.MajorFirmwareVersion; + Enquiry2->FirmwareID.MinorVersion = + Controller->V1.Enquiry.MinorFirmwareVersion; + Enquiry2->FirmwareID.FirmwareType = '0'; + Enquiry2->FirmwareID.TurnID = 0; + } + sprintf(Controller->FirmwareVersion, "%d.%02d-%c-%02d", + Enquiry2->FirmwareID.MajorVersion, Enquiry2->FirmwareID.MinorVersion, + Enquiry2->FirmwareID.FirmwareType, Enquiry2->FirmwareID.TurnID); + if (!((Controller->FirmwareVersion[0] == '5' && + strcmp(Controller->FirmwareVersion, "5.06") >= 0) || + (Controller->FirmwareVersion[0] == '4' && + strcmp(Controller->FirmwareVersion, "4.06") >= 0) || + (Controller->FirmwareVersion[0] == '3' && + strcmp(Controller->FirmwareVersion, "3.51") >= 0) || + (Controller->FirmwareVersion[0] == '2' && + strcmp(Controller->FirmwareVersion, FIRMWARE_27X) >= 0))) + { + DAC960_Failure(Controller, "FIRMWARE VERSION VERIFICATION"); + DAC960_Error("Firmware Version = '%s'\n", Controller, + Controller->FirmwareVersion); + free_dma_loaf(Controller->PCIDevice, &local_dma); + return false; + } + /* + Initialize the Controller Channels, Targets, Memory Size, and SAF-TE + Enclosure Management Enabled fields. + */ + Controller->Channels = Enquiry2->ActualChannels; + Controller->Targets = Enquiry2->MaxTargets; + Controller->MemorySize = Enquiry2->MemorySize >> 20; + Controller->V1.SAFTE_EnclosureManagementEnabled = + (Enquiry2->FaultManagementType == DAC960_V1_SAFTE); + /* + Initialize the Controller Queue Depth, Driver Queue Depth, Logical Drive + Count, Maximum Blocks per Command, Controller Scatter/Gather Limit, and + Driver Scatter/Gather Limit. The Driver Queue Depth must be at most one + less than the Controller Queue Depth to allow for an automatic drive + rebuild operation. + */ + Controller->ControllerQueueDepth = Controller->V1.Enquiry.MaxCommands; + Controller->DriverQueueDepth = Controller->ControllerQueueDepth - 1; + if (Controller->DriverQueueDepth > DAC960_MaxDriverQueueDepth) + Controller->DriverQueueDepth = DAC960_MaxDriverQueueDepth; + Controller->LogicalDriveCount = + Controller->V1.Enquiry.NumberOfLogicalDrives; + Controller->MaxBlocksPerCommand = Enquiry2->MaxBlocksPerCommand; + Controller->ControllerScatterGatherLimit = Enquiry2->MaxScatterGatherEntries; + Controller->DriverScatterGatherLimit = + Controller->ControllerScatterGatherLimit; + if (Controller->DriverScatterGatherLimit > DAC960_V1_ScatterGatherLimit) + Controller->DriverScatterGatherLimit = DAC960_V1_ScatterGatherLimit; + /* + Initialize the Stripe Size, Segment Size, and Geometry Translation. + */ + Controller->V1.StripeSize = Config2->BlocksPerStripe * Config2->BlockFactor + >> (10 - DAC960_BlockSizeBits); + Controller->V1.SegmentSize = Config2->BlocksPerCacheLine * Config2->BlockFactor + >> (10 - DAC960_BlockSizeBits); + switch (Config2->DriveGeometry) + { + case DAC960_V1_Geometry_128_32: + Controller->V1.GeometryTranslationHeads = 128; + Controller->V1.GeometryTranslationSectors = 32; + break; + case DAC960_V1_Geometry_255_63: + Controller->V1.GeometryTranslationHeads = 255; + Controller->V1.GeometryTranslationSectors = 63; + break; + default: + free_dma_loaf(Controller->PCIDevice, &local_dma); + return DAC960_Failure(Controller, "CONFIG2 DRIVE GEOMETRY"); + } + /* + Initialize the Background Initialization Status. + */ + if ((Controller->FirmwareVersion[0] == '4' && + strcmp(Controller->FirmwareVersion, "4.08") >= 0) || + (Controller->FirmwareVersion[0] == '5' && + strcmp(Controller->FirmwareVersion, "5.08") >= 0)) + { + Controller->V1.BackgroundInitializationStatusSupported = true; + DAC960_V1_ExecuteType3B(Controller, + DAC960_V1_BackgroundInitializationControl, 0x20, + Controller-> + V1.BackgroundInitializationStatusDMA); + memcpy(&Controller->V1.LastBackgroundInitializationStatus, + Controller->V1.BackgroundInitializationStatus, + sizeof(DAC960_V1_BackgroundInitializationStatus_T)); + } + /* + Initialize the Logical Drive Initially Accessible flag. + */ + for (LogicalDriveNumber = 0; + LogicalDriveNumber < Controller->LogicalDriveCount; + LogicalDriveNumber++) + if (Controller->V1.LogicalDriveInformation + [LogicalDriveNumber].LogicalDriveState != + DAC960_V1_LogicalDrive_Offline) + Controller->LogicalDriveInitiallyAccessible[LogicalDriveNumber] = true; + Controller->V1.LastRebuildStatus = DAC960_V1_NoRebuildOrCheckInProgress; + free_dma_loaf(Controller->PCIDevice, &local_dma); + return true; +} + + +/* + DAC960_V2_ReadControllerConfiguration reads the Configuration Information + from DAC960 V2 Firmware Controllers and initializes the Controller structure. +*/ + +static boolean DAC960_V2_ReadControllerConfiguration(DAC960_Controller_T + *Controller) +{ + DAC960_V2_ControllerInfo_T *ControllerInfo = + &Controller->V2.ControllerInformation; + unsigned short LogicalDeviceNumber = 0; + int ModelNameLength; + + /* Get data into dma-able area, then copy into permanant location */ + if (!DAC960_V2_NewControllerInfo(Controller)) + return DAC960_Failure(Controller, "GET CONTROLLER INFO"); + memcpy(ControllerInfo, Controller->V2.NewControllerInformation, + sizeof(DAC960_V2_ControllerInfo_T)); + + + if (!DAC960_V2_GeneralInfo(Controller)) + return DAC960_Failure(Controller, "GET HEALTH STATUS"); + + /* + Initialize the Controller Model Name and Full Model Name fields. + */ + ModelNameLength = sizeof(ControllerInfo->ControllerName); + if (ModelNameLength > sizeof(Controller->ModelName)-1) + ModelNameLength = sizeof(Controller->ModelName)-1; + memcpy(Controller->ModelName, ControllerInfo->ControllerName, + ModelNameLength); + ModelNameLength--; + while (Controller->ModelName[ModelNameLength] == ' ' || + Controller->ModelName[ModelNameLength] == '\0') + ModelNameLength--; + Controller->ModelName[++ModelNameLength] = '\0'; + strcpy(Controller->FullModelName, "Mylex "); + strcat(Controller->FullModelName, Controller->ModelName); + /* + Initialize the Controller Firmware Version field. + */ + sprintf(Controller->FirmwareVersion, "%d.%02d-%02d", + ControllerInfo->FirmwareMajorVersion, + ControllerInfo->FirmwareMinorVersion, + ControllerInfo->FirmwareTurnNumber); + if (ControllerInfo->FirmwareMajorVersion == 6 && + ControllerInfo->FirmwareMinorVersion == 0 && + ControllerInfo->FirmwareTurnNumber < 1) + { + DAC960_Info("FIRMWARE VERSION %s DOES NOT PROVIDE THE CONTROLLER\n", + Controller, Controller->FirmwareVersion); + DAC960_Info("STATUS MONITORING FUNCTIONALITY NEEDED BY THIS DRIVER.\n", + Controller); + DAC960_Info("PLEASE UPGRADE TO VERSION 6.00-01 OR ABOVE.\n", + Controller); + } + /* + Initialize the Controller Channels, Targets, and Memory Size. + */ + Controller->Channels = ControllerInfo->NumberOfPhysicalChannelsPresent; + Controller->Targets = + ControllerInfo->MaximumTargetsPerChannel + [ControllerInfo->NumberOfPhysicalChannelsPresent-1]; + Controller->MemorySize = ControllerInfo->MemorySizeMB; + /* + Initialize the Controller Queue Depth, Driver Queue Depth, Logical Drive + Count, Maximum Blocks per Command, Controller Scatter/Gather Limit, and + Driver Scatter/Gather Limit. The Driver Queue Depth must be at most one + less than the Controller Queue Depth to allow for an automatic drive + rebuild operation. + */ + Controller->ControllerQueueDepth = ControllerInfo->MaximumParallelCommands; + Controller->DriverQueueDepth = Controller->ControllerQueueDepth - 1; + if (Controller->DriverQueueDepth > DAC960_MaxDriverQueueDepth) + Controller->DriverQueueDepth = DAC960_MaxDriverQueueDepth; + Controller->LogicalDriveCount = ControllerInfo->LogicalDevicesPresent; + Controller->MaxBlocksPerCommand = + ControllerInfo->MaximumDataTransferSizeInBlocks; + Controller->ControllerScatterGatherLimit = + ControllerInfo->MaximumScatterGatherEntries; + Controller->DriverScatterGatherLimit = + Controller->ControllerScatterGatherLimit; + if (Controller->DriverScatterGatherLimit > DAC960_V2_ScatterGatherLimit) + Controller->DriverScatterGatherLimit = DAC960_V2_ScatterGatherLimit; + /* + Initialize the Logical Device Information. + */ + while (true) + { + DAC960_V2_LogicalDeviceInfo_T *NewLogicalDeviceInfo = + Controller->V2.NewLogicalDeviceInformation; + DAC960_V2_LogicalDeviceInfo_T *LogicalDeviceInfo; + DAC960_V2_PhysicalDevice_T PhysicalDevice; + + if (!DAC960_V2_NewLogicalDeviceInfo(Controller, LogicalDeviceNumber)) + break; + LogicalDeviceNumber = NewLogicalDeviceInfo->LogicalDeviceNumber; + if (LogicalDeviceNumber >= DAC960_MaxLogicalDrives) { + DAC960_Error("DAC960: Logical Drive Number %d not supported\n", + Controller, LogicalDeviceNumber); + break; + } + if (NewLogicalDeviceInfo->DeviceBlockSizeInBytes != DAC960_BlockSize) { + DAC960_Error("DAC960: Logical Drive Block Size %d not supported\n", + Controller, NewLogicalDeviceInfo->DeviceBlockSizeInBytes); + LogicalDeviceNumber++; + continue; + } + PhysicalDevice.Controller = 0; + PhysicalDevice.Channel = NewLogicalDeviceInfo->Channel; + PhysicalDevice.TargetID = NewLogicalDeviceInfo->TargetID; + PhysicalDevice.LogicalUnit = NewLogicalDeviceInfo->LogicalUnit; + Controller->V2.LogicalDriveToVirtualDevice[LogicalDeviceNumber] = + PhysicalDevice; + if (NewLogicalDeviceInfo->LogicalDeviceState != + DAC960_V2_LogicalDevice_Offline) + Controller->LogicalDriveInitiallyAccessible[LogicalDeviceNumber] = true; + LogicalDeviceInfo = (DAC960_V2_LogicalDeviceInfo_T *) + kmalloc(sizeof(DAC960_V2_LogicalDeviceInfo_T), GFP_ATOMIC); + if (LogicalDeviceInfo == NULL) + return DAC960_Failure(Controller, "LOGICAL DEVICE ALLOCATION"); + Controller->V2.LogicalDeviceInformation[LogicalDeviceNumber] = + LogicalDeviceInfo; + memcpy(LogicalDeviceInfo, NewLogicalDeviceInfo, + sizeof(DAC960_V2_LogicalDeviceInfo_T)); + LogicalDeviceNumber++; + } + return true; +} + + +/* + DAC960_ReportControllerConfiguration reports the Configuration Information + for Controller. +*/ + +static boolean DAC960_ReportControllerConfiguration(DAC960_Controller_T + *Controller) +{ + DAC960_Info("Configuring Mylex %s PCI RAID Controller\n", + Controller, Controller->ModelName); + DAC960_Info(" Firmware Version: %s, Channels: %d, Memory Size: %dMB\n", + Controller, Controller->FirmwareVersion, + Controller->Channels, Controller->MemorySize); + DAC960_Info(" PCI Bus: %d, Device: %d, Function: %d, I/O Address: ", + Controller, Controller->Bus, + Controller->Device, Controller->Function); + if (Controller->IO_Address == 0) + DAC960_Info("Unassigned\n", Controller); + else DAC960_Info("0x%X\n", Controller, Controller->IO_Address); + DAC960_Info(" PCI Address: 0x%X mapped at 0x%lX, IRQ Channel: %d\n", + Controller, Controller->PCI_Address, + (unsigned long) Controller->BaseAddress, + Controller->IRQ_Channel); + DAC960_Info(" Controller Queue Depth: %d, " + "Maximum Blocks per Command: %d\n", + Controller, Controller->ControllerQueueDepth, + Controller->MaxBlocksPerCommand); + DAC960_Info(" Driver Queue Depth: %d, " + "Scatter/Gather Limit: %d of %d Segments\n", + Controller, Controller->DriverQueueDepth, + Controller->DriverScatterGatherLimit, + Controller->ControllerScatterGatherLimit); + if (Controller->FirmwareType == DAC960_V1_Controller) + { + DAC960_Info(" Stripe Size: %dKB, Segment Size: %dKB, " + "BIOS Geometry: %d/%d\n", Controller, + Controller->V1.StripeSize, + Controller->V1.SegmentSize, + Controller->V1.GeometryTranslationHeads, + Controller->V1.GeometryTranslationSectors); + if (Controller->V1.SAFTE_EnclosureManagementEnabled) + DAC960_Info(" SAF-TE Enclosure Management Enabled\n", Controller); + } + return true; +} + + +/* + DAC960_V1_ReadDeviceConfiguration reads the Device Configuration Information + for DAC960 V1 Firmware Controllers by requesting the SCSI Inquiry and SCSI + Inquiry Unit Serial Number information for each device connected to + Controller. +*/ + +static boolean DAC960_V1_ReadDeviceConfiguration(DAC960_Controller_T + *Controller) +{ + struct dma_loaf local_dma; + + dma_addr_t DCDBs_dma[DAC960_V1_MaxChannels]; + DAC960_V1_DCDB_T *DCDBs_cpu[DAC960_V1_MaxChannels]; + + dma_addr_t SCSI_Inquiry_dma[DAC960_V1_MaxChannels]; + DAC960_SCSI_Inquiry_T *SCSI_Inquiry_cpu[DAC960_V1_MaxChannels]; + + dma_addr_t SCSI_NewInquiryUnitSerialNumberDMA[DAC960_V1_MaxChannels]; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *SCSI_NewInquiryUnitSerialNumberCPU[DAC960_V1_MaxChannels]; + + struct completion Completions[DAC960_V1_MaxChannels]; + unsigned long flags; + int Channel, TargetID; + + if (!init_dma_loaf(Controller->PCIDevice, &local_dma, + DAC960_V1_MaxChannels*(sizeof(DAC960_V1_DCDB_T) + + sizeof(DAC960_SCSI_Inquiry_T) + + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)))) + return DAC960_Failure(Controller, + "DMA ALLOCATION FAILED IN ReadDeviceConfiguration"); + + for (Channel = 0; Channel < Controller->Channels; Channel++) { + DCDBs_cpu[Channel] = slice_dma_loaf(&local_dma, + sizeof(DAC960_V1_DCDB_T), DCDBs_dma + Channel); + SCSI_Inquiry_cpu[Channel] = slice_dma_loaf(&local_dma, + sizeof(DAC960_SCSI_Inquiry_T), + SCSI_Inquiry_dma + Channel); + SCSI_NewInquiryUnitSerialNumberCPU[Channel] = slice_dma_loaf(&local_dma, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T), + SCSI_NewInquiryUnitSerialNumberDMA + Channel); + } + + for (TargetID = 0; TargetID < Controller->Targets; TargetID++) + { + /* + * For each channel, submit a probe for a device on that channel. + * The timeout interval for a device that is present is 10 seconds. + * With this approach, the timeout periods can elapse in parallel + * on each channel. + */ + for (Channel = 0; Channel < Controller->Channels; Channel++) + { + dma_addr_t NewInquiryStandardDataDMA = SCSI_Inquiry_dma[Channel]; + DAC960_V1_DCDB_T *DCDB = DCDBs_cpu[Channel]; + dma_addr_t DCDB_dma = DCDBs_dma[Channel]; + DAC960_Command_T *Command = Controller->Commands[Channel]; + struct completion *Completion = &Completions[Channel]; + + init_completion(Completion); + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + Command->Completion = Completion; + Command->V1.CommandMailbox.Type3.CommandOpcode = DAC960_V1_DCDB; + Command->V1.CommandMailbox.Type3.BusAddress = DCDB_dma; + DCDB->Channel = Channel; + DCDB->TargetID = TargetID; + DCDB->Direction = DAC960_V1_DCDB_DataTransferDeviceToSystem; + DCDB->EarlyStatus = false; + DCDB->Timeout = DAC960_V1_DCDB_Timeout_10_seconds; + DCDB->NoAutomaticRequestSense = false; + DCDB->DisconnectPermitted = true; + DCDB->TransferLength = sizeof(DAC960_SCSI_Inquiry_T); + DCDB->BusAddress = NewInquiryStandardDataDMA; + DCDB->CDBLength = 6; + DCDB->TransferLengthHigh4 = 0; + DCDB->SenseLength = sizeof(DCDB->SenseData); + DCDB->CDB[0] = 0x12; /* INQUIRY */ + DCDB->CDB[1] = 0; /* EVPD = 0 */ + DCDB->CDB[2] = 0; /* Page Code */ + DCDB->CDB[3] = 0; /* Reserved */ + DCDB->CDB[4] = sizeof(DAC960_SCSI_Inquiry_T); + DCDB->CDB[5] = 0; /* Control */ + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_QueueCommand(Command); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + } + /* + * Wait for the problems submitted in the previous loop + * to complete. On the probes that are successful, + * get the serial number of the device that was found. + */ + for (Channel = 0; Channel < Controller->Channels; Channel++) + { + DAC960_SCSI_Inquiry_T *InquiryStandardData = + &Controller->V1.InquiryStandardData[Channel][TargetID]; + DAC960_SCSI_Inquiry_T *NewInquiryStandardData = SCSI_Inquiry_cpu[Channel]; + dma_addr_t NewInquiryUnitSerialNumberDMA = + SCSI_NewInquiryUnitSerialNumberDMA[Channel]; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *NewInquiryUnitSerialNumber = + SCSI_NewInquiryUnitSerialNumberCPU[Channel]; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + &Controller->V1.InquiryUnitSerialNumber[Channel][TargetID]; + DAC960_Command_T *Command = Controller->Commands[Channel]; + DAC960_V1_DCDB_T *DCDB = DCDBs_cpu[Channel]; + struct completion *Completion = &Completions[Channel]; + + wait_for_completion(Completion); + + if (Command->V1.CommandStatus != DAC960_V1_NormalCompletion) { + memset(InquiryStandardData, 0, sizeof(DAC960_SCSI_Inquiry_T)); + InquiryStandardData->PeripheralDeviceType = 0x1F; + continue; + } else + memcpy(InquiryStandardData, NewInquiryStandardData, sizeof(DAC960_SCSI_Inquiry_T)); + + /* Preserve Channel and TargetID values from the previous loop */ + Command->Completion = Completion; + DCDB->TransferLength = sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T); + DCDB->BusAddress = NewInquiryUnitSerialNumberDMA; + DCDB->SenseLength = sizeof(DCDB->SenseData); + DCDB->CDB[0] = 0x12; /* INQUIRY */ + DCDB->CDB[1] = 1; /* EVPD = 1 */ + DCDB->CDB[2] = 0x80; /* Page Code */ + DCDB->CDB[3] = 0; /* Reserved */ + DCDB->CDB[4] = sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T); + DCDB->CDB[5] = 0; /* Control */ + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_QueueCommand(Command); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + wait_for_completion(Completion); + + if (Command->V1.CommandStatus != DAC960_V1_NormalCompletion) { + memset(InquiryUnitSerialNumber, 0, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + InquiryUnitSerialNumber->PeripheralDeviceType = 0x1F; + } else + memcpy(InquiryUnitSerialNumber, NewInquiryUnitSerialNumber, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + } + } + free_dma_loaf(Controller->PCIDevice, &local_dma); + return true; +} + + +/* + DAC960_V2_ReadDeviceConfiguration reads the Device Configuration Information + for DAC960 V2 Firmware Controllers by requesting the Physical Device + Information and SCSI Inquiry Unit Serial Number information for each + device connected to Controller. +*/ + +static boolean DAC960_V2_ReadDeviceConfiguration(DAC960_Controller_T + *Controller) +{ + unsigned char Channel = 0, TargetID = 0, LogicalUnit = 0; + unsigned short PhysicalDeviceIndex = 0; + + while (true) + { + DAC960_V2_PhysicalDeviceInfo_T *NewPhysicalDeviceInfo = + Controller->V2.NewPhysicalDeviceInformation; + DAC960_V2_PhysicalDeviceInfo_T *PhysicalDeviceInfo; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *NewInquiryUnitSerialNumber = + Controller->V2.NewInquiryUnitSerialNumber; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber; + + if (!DAC960_V2_NewPhysicalDeviceInfo(Controller, Channel, TargetID, LogicalUnit)) + break; + + PhysicalDeviceInfo = (DAC960_V2_PhysicalDeviceInfo_T *) + kmalloc(sizeof(DAC960_V2_PhysicalDeviceInfo_T), GFP_ATOMIC); + if (PhysicalDeviceInfo == NULL) + return DAC960_Failure(Controller, "PHYSICAL DEVICE ALLOCATION"); + Controller->V2.PhysicalDeviceInformation[PhysicalDeviceIndex] = + PhysicalDeviceInfo; + memcpy(PhysicalDeviceInfo, NewPhysicalDeviceInfo, + sizeof(DAC960_V2_PhysicalDeviceInfo_T)); + + InquiryUnitSerialNumber = (DAC960_SCSI_Inquiry_UnitSerialNumber_T *) + kmalloc(sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T), GFP_ATOMIC); + if (InquiryUnitSerialNumber == NULL) { + kfree(PhysicalDeviceInfo); + return DAC960_Failure(Controller, "SERIAL NUMBER ALLOCATION"); + } + Controller->V2.InquiryUnitSerialNumber[PhysicalDeviceIndex] = + InquiryUnitSerialNumber; + + Channel = NewPhysicalDeviceInfo->Channel; + TargetID = NewPhysicalDeviceInfo->TargetID; + LogicalUnit = NewPhysicalDeviceInfo->LogicalUnit; + + /* + Some devices do NOT have Unit Serial Numbers. + This command fails for them. But, we still want to + remember those devices are there. Construct a + UnitSerialNumber structure for the failure case. + */ + if (!DAC960_V2_NewInquiryUnitSerialNumber(Controller, Channel, TargetID, LogicalUnit)) { + memset(InquiryUnitSerialNumber, 0, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + InquiryUnitSerialNumber->PeripheralDeviceType = 0x1F; + } else + memcpy(InquiryUnitSerialNumber, NewInquiryUnitSerialNumber, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + + PhysicalDeviceIndex++; + LogicalUnit++; + } + return true; +} + + +/* + DAC960_SanitizeInquiryData sanitizes the Vendor, Model, Revision, and + Product Serial Number fields of the Inquiry Standard Data and Inquiry + Unit Serial Number structures. +*/ + +static void DAC960_SanitizeInquiryData(DAC960_SCSI_Inquiry_T + *InquiryStandardData, + DAC960_SCSI_Inquiry_UnitSerialNumber_T + *InquiryUnitSerialNumber, + unsigned char *Vendor, + unsigned char *Model, + unsigned char *Revision, + unsigned char *SerialNumber) +{ + int SerialNumberLength, i; + if (InquiryStandardData->PeripheralDeviceType == 0x1F) return; + for (i = 0; i < sizeof(InquiryStandardData->VendorIdentification); i++) + { + unsigned char VendorCharacter = + InquiryStandardData->VendorIdentification[i]; + Vendor[i] = (VendorCharacter >= ' ' && VendorCharacter <= '~' + ? VendorCharacter : ' '); + } + Vendor[sizeof(InquiryStandardData->VendorIdentification)] = '\0'; + for (i = 0; i < sizeof(InquiryStandardData->ProductIdentification); i++) + { + unsigned char ModelCharacter = + InquiryStandardData->ProductIdentification[i]; + Model[i] = (ModelCharacter >= ' ' && ModelCharacter <= '~' + ? ModelCharacter : ' '); + } + Model[sizeof(InquiryStandardData->ProductIdentification)] = '\0'; + for (i = 0; i < sizeof(InquiryStandardData->ProductRevisionLevel); i++) + { + unsigned char RevisionCharacter = + InquiryStandardData->ProductRevisionLevel[i]; + Revision[i] = (RevisionCharacter >= ' ' && RevisionCharacter <= '~' + ? RevisionCharacter : ' '); + } + Revision[sizeof(InquiryStandardData->ProductRevisionLevel)] = '\0'; + if (InquiryUnitSerialNumber->PeripheralDeviceType == 0x1F) return; + SerialNumberLength = InquiryUnitSerialNumber->PageLength; + if (SerialNumberLength > + sizeof(InquiryUnitSerialNumber->ProductSerialNumber)) + SerialNumberLength = sizeof(InquiryUnitSerialNumber->ProductSerialNumber); + for (i = 0; i < SerialNumberLength; i++) + { + unsigned char SerialNumberCharacter = + InquiryUnitSerialNumber->ProductSerialNumber[i]; + SerialNumber[i] = + (SerialNumberCharacter >= ' ' && SerialNumberCharacter <= '~' + ? SerialNumberCharacter : ' '); + } + SerialNumber[SerialNumberLength] = '\0'; +} + + +/* + DAC960_V1_ReportDeviceConfiguration reports the Device Configuration + Information for DAC960 V1 Firmware Controllers. +*/ + +static boolean DAC960_V1_ReportDeviceConfiguration(DAC960_Controller_T + *Controller) +{ + int LogicalDriveNumber, Channel, TargetID; + DAC960_Info(" Physical Devices:\n", Controller); + for (Channel = 0; Channel < Controller->Channels; Channel++) + for (TargetID = 0; TargetID < Controller->Targets; TargetID++) + { + DAC960_SCSI_Inquiry_T *InquiryStandardData = + &Controller->V1.InquiryStandardData[Channel][TargetID]; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + &Controller->V1.InquiryUnitSerialNumber[Channel][TargetID]; + DAC960_V1_DeviceState_T *DeviceState = + &Controller->V1.DeviceState[Channel][TargetID]; + DAC960_V1_ErrorTableEntry_T *ErrorEntry = + &Controller->V1.ErrorTable.ErrorTableEntries[Channel][TargetID]; + char Vendor[1+sizeof(InquiryStandardData->VendorIdentification)]; + char Model[1+sizeof(InquiryStandardData->ProductIdentification)]; + char Revision[1+sizeof(InquiryStandardData->ProductRevisionLevel)]; + char SerialNumber[1+sizeof(InquiryUnitSerialNumber + ->ProductSerialNumber)]; + if (InquiryStandardData->PeripheralDeviceType == 0x1F) continue; + DAC960_SanitizeInquiryData(InquiryStandardData, InquiryUnitSerialNumber, + Vendor, Model, Revision, SerialNumber); + DAC960_Info(" %d:%d%s Vendor: %s Model: %s Revision: %s\n", + Controller, Channel, TargetID, (TargetID < 10 ? " " : ""), + Vendor, Model, Revision); + if (InquiryUnitSerialNumber->PeripheralDeviceType != 0x1F) + DAC960_Info(" Serial Number: %s\n", Controller, SerialNumber); + if (DeviceState->Present && + DeviceState->DeviceType == DAC960_V1_DiskType) + { + if (Controller->V1.DeviceResetCount[Channel][TargetID] > 0) + DAC960_Info(" Disk Status: %s, %u blocks, %d resets\n", + Controller, + (DeviceState->DeviceState == DAC960_V1_Device_Dead + ? "Dead" + : DeviceState->DeviceState + == DAC960_V1_Device_WriteOnly + ? "Write-Only" + : DeviceState->DeviceState + == DAC960_V1_Device_Online + ? "Online" : "Standby"), + DeviceState->DiskSize, + Controller->V1.DeviceResetCount[Channel][TargetID]); + else + DAC960_Info(" Disk Status: %s, %u blocks\n", Controller, + (DeviceState->DeviceState == DAC960_V1_Device_Dead + ? "Dead" + : DeviceState->DeviceState + == DAC960_V1_Device_WriteOnly + ? "Write-Only" + : DeviceState->DeviceState + == DAC960_V1_Device_Online + ? "Online" : "Standby"), + DeviceState->DiskSize); + } + if (ErrorEntry->ParityErrorCount > 0 || + ErrorEntry->SoftErrorCount > 0 || + ErrorEntry->HardErrorCount > 0 || + ErrorEntry->MiscErrorCount > 0) + DAC960_Info(" Errors - Parity: %d, Soft: %d, " + "Hard: %d, Misc: %d\n", Controller, + ErrorEntry->ParityErrorCount, + ErrorEntry->SoftErrorCount, + ErrorEntry->HardErrorCount, + ErrorEntry->MiscErrorCount); + } + DAC960_Info(" Logical Drives:\n", Controller); + for (LogicalDriveNumber = 0; + LogicalDriveNumber < Controller->LogicalDriveCount; + LogicalDriveNumber++) + { + DAC960_V1_LogicalDriveInformation_T *LogicalDriveInformation = + &Controller->V1.LogicalDriveInformation[LogicalDriveNumber]; + DAC960_Info(" /dev/rd/c%dd%d: RAID-%d, %s, %u blocks, %s\n", + Controller, Controller->ControllerNumber, LogicalDriveNumber, + LogicalDriveInformation->RAIDLevel, + (LogicalDriveInformation->LogicalDriveState + == DAC960_V1_LogicalDrive_Online + ? "Online" + : LogicalDriveInformation->LogicalDriveState + == DAC960_V1_LogicalDrive_Critical + ? "Critical" : "Offline"), + LogicalDriveInformation->LogicalDriveSize, + (LogicalDriveInformation->WriteBack + ? "Write Back" : "Write Thru")); + } + return true; +} + + +/* + DAC960_V2_ReportDeviceConfiguration reports the Device Configuration + Information for DAC960 V2 Firmware Controllers. +*/ + +static boolean DAC960_V2_ReportDeviceConfiguration(DAC960_Controller_T + *Controller) +{ + int PhysicalDeviceIndex, LogicalDriveNumber; + DAC960_Info(" Physical Devices:\n", Controller); + for (PhysicalDeviceIndex = 0; + PhysicalDeviceIndex < DAC960_V2_MaxPhysicalDevices; + PhysicalDeviceIndex++) + { + DAC960_V2_PhysicalDeviceInfo_T *PhysicalDeviceInfo = + Controller->V2.PhysicalDeviceInformation[PhysicalDeviceIndex]; + DAC960_SCSI_Inquiry_T *InquiryStandardData = + (DAC960_SCSI_Inquiry_T *) &PhysicalDeviceInfo->SCSI_InquiryData; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + Controller->V2.InquiryUnitSerialNumber[PhysicalDeviceIndex]; + char Vendor[1+sizeof(InquiryStandardData->VendorIdentification)]; + char Model[1+sizeof(InquiryStandardData->ProductIdentification)]; + char Revision[1+sizeof(InquiryStandardData->ProductRevisionLevel)]; + char SerialNumber[1+sizeof(InquiryUnitSerialNumber->ProductSerialNumber)]; + if (PhysicalDeviceInfo == NULL) break; + DAC960_SanitizeInquiryData(InquiryStandardData, InquiryUnitSerialNumber, + Vendor, Model, Revision, SerialNumber); + DAC960_Info(" %d:%d%s Vendor: %s Model: %s Revision: %s\n", + Controller, + PhysicalDeviceInfo->Channel, + PhysicalDeviceInfo->TargetID, + (PhysicalDeviceInfo->TargetID < 10 ? " " : ""), + Vendor, Model, Revision); + if (PhysicalDeviceInfo->NegotiatedSynchronousMegaTransfers == 0) + DAC960_Info(" %sAsynchronous\n", Controller, + (PhysicalDeviceInfo->NegotiatedDataWidthBits == 16 + ? "Wide " :"")); + else + DAC960_Info(" %sSynchronous at %d MB/sec\n", Controller, + (PhysicalDeviceInfo->NegotiatedDataWidthBits == 16 + ? "Wide " :""), + (PhysicalDeviceInfo->NegotiatedSynchronousMegaTransfers + * PhysicalDeviceInfo->NegotiatedDataWidthBits/8)); + if (InquiryUnitSerialNumber->PeripheralDeviceType != 0x1F) + DAC960_Info(" Serial Number: %s\n", Controller, SerialNumber); + if (PhysicalDeviceInfo->PhysicalDeviceState == + DAC960_V2_Device_Unconfigured) + continue; + DAC960_Info(" Disk Status: %s, %u blocks\n", Controller, + (PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Online + ? "Online" + : PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Rebuild + ? "Rebuild" + : PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Missing + ? "Missing" + : PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Critical + ? "Critical" + : PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Dead + ? "Dead" + : PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_SuspectedDead + ? "Suspected-Dead" + : PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_CommandedOffline + ? "Commanded-Offline" + : PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Standby + ? "Standby" : "Unknown"), + PhysicalDeviceInfo->ConfigurableDeviceSize); + if (PhysicalDeviceInfo->ParityErrors == 0 && + PhysicalDeviceInfo->SoftErrors == 0 && + PhysicalDeviceInfo->HardErrors == 0 && + PhysicalDeviceInfo->MiscellaneousErrors == 0 && + PhysicalDeviceInfo->CommandTimeouts == 0 && + PhysicalDeviceInfo->Retries == 0 && + PhysicalDeviceInfo->Aborts == 0 && + PhysicalDeviceInfo->PredictedFailuresDetected == 0) + continue; + DAC960_Info(" Errors - Parity: %d, Soft: %d, " + "Hard: %d, Misc: %d\n", Controller, + PhysicalDeviceInfo->ParityErrors, + PhysicalDeviceInfo->SoftErrors, + PhysicalDeviceInfo->HardErrors, + PhysicalDeviceInfo->MiscellaneousErrors); + DAC960_Info(" Timeouts: %d, Retries: %d, " + "Aborts: %d, Predicted: %d\n", Controller, + PhysicalDeviceInfo->CommandTimeouts, + PhysicalDeviceInfo->Retries, + PhysicalDeviceInfo->Aborts, + PhysicalDeviceInfo->PredictedFailuresDetected); + } + DAC960_Info(" Logical Drives:\n", Controller); + for (LogicalDriveNumber = 0; + LogicalDriveNumber < DAC960_MaxLogicalDrives; + LogicalDriveNumber++) + { + DAC960_V2_LogicalDeviceInfo_T *LogicalDeviceInfo = + Controller->V2.LogicalDeviceInformation[LogicalDriveNumber]; + unsigned char *ReadCacheStatus[] = { "Read Cache Disabled", + "Read Cache Enabled", + "Read Ahead Enabled", + "Intelligent Read Ahead Enabled", + "-", "-", "-", "-" }; + unsigned char *WriteCacheStatus[] = { "Write Cache Disabled", + "Logical Device Read Only", + "Write Cache Enabled", + "Intelligent Write Cache Enabled", + "-", "-", "-", "-" }; + unsigned char *GeometryTranslation; + if (LogicalDeviceInfo == NULL) continue; + switch (LogicalDeviceInfo->DriveGeometry) + { + case DAC960_V2_Geometry_128_32: + GeometryTranslation = "128/32"; + break; + case DAC960_V2_Geometry_255_63: + GeometryTranslation = "255/63"; + break; + default: + GeometryTranslation = "Invalid"; + DAC960_Error("Illegal Logical Device Geometry %d\n", + Controller, LogicalDeviceInfo->DriveGeometry); + break; + } + DAC960_Info(" /dev/rd/c%dd%d: RAID-%d, %s, %u blocks\n", + Controller, Controller->ControllerNumber, LogicalDriveNumber, + LogicalDeviceInfo->RAIDLevel, + (LogicalDeviceInfo->LogicalDeviceState + == DAC960_V2_LogicalDevice_Online + ? "Online" + : LogicalDeviceInfo->LogicalDeviceState + == DAC960_V2_LogicalDevice_Critical + ? "Critical" : "Offline"), + LogicalDeviceInfo->ConfigurableDeviceSize); + DAC960_Info(" Logical Device %s, BIOS Geometry: %s\n", + Controller, + (LogicalDeviceInfo->LogicalDeviceControl + .LogicalDeviceInitialized + ? "Initialized" : "Uninitialized"), + GeometryTranslation); + if (LogicalDeviceInfo->StripeSize == 0) + { + if (LogicalDeviceInfo->CacheLineSize == 0) + DAC960_Info(" Stripe Size: N/A, " + "Segment Size: N/A\n", Controller); + else + DAC960_Info(" Stripe Size: N/A, " + "Segment Size: %dKB\n", Controller, + 1 << (LogicalDeviceInfo->CacheLineSize - 2)); + } + else + { + if (LogicalDeviceInfo->CacheLineSize == 0) + DAC960_Info(" Stripe Size: %dKB, " + "Segment Size: N/A\n", Controller, + 1 << (LogicalDeviceInfo->StripeSize - 2)); + else + DAC960_Info(" Stripe Size: %dKB, " + "Segment Size: %dKB\n", Controller, + 1 << (LogicalDeviceInfo->StripeSize - 2), + 1 << (LogicalDeviceInfo->CacheLineSize - 2)); + } + DAC960_Info(" %s, %s\n", Controller, + ReadCacheStatus[ + LogicalDeviceInfo->LogicalDeviceControl.ReadCache], + WriteCacheStatus[ + LogicalDeviceInfo->LogicalDeviceControl.WriteCache]); + if (LogicalDeviceInfo->SoftErrors > 0 || + LogicalDeviceInfo->CommandsFailed > 0 || + LogicalDeviceInfo->DeferredWriteErrors) + DAC960_Info(" Errors - Soft: %d, Failed: %d, " + "Deferred Write: %d\n", Controller, + LogicalDeviceInfo->SoftErrors, + LogicalDeviceInfo->CommandsFailed, + LogicalDeviceInfo->DeferredWriteErrors); + + } + return true; +} + +/* + DAC960_RegisterBlockDevice registers the Block Device structures + associated with Controller. +*/ + +static boolean DAC960_RegisterBlockDevice(DAC960_Controller_T *Controller) +{ + int MajorNumber = DAC960_MAJOR + Controller->ControllerNumber; + int n; + + /* + Register the Block Device Major Number for this DAC960 Controller. + */ + if (register_blkdev(MajorNumber, "dac960") < 0) + return false; + + for (n = 0; n < DAC960_MaxLogicalDrives; n++) { + struct gendisk *disk = Controller->disks[n]; + struct request_queue *RequestQueue; + + /* for now, let all request queues share controller's lock */ + RequestQueue = blk_init_queue(DAC960_RequestFunction,&Controller->queue_lock); + if (!RequestQueue) { + printk("DAC960: failure to allocate request queue\n"); + continue; + } + Controller->RequestQueue[n] = RequestQueue; + blk_queue_bounce_limit(RequestQueue, Controller->BounceBufferLimit); + RequestQueue->queuedata = Controller; + blk_queue_max_hw_segments(RequestQueue, Controller->DriverScatterGatherLimit); + blk_queue_max_phys_segments(RequestQueue, Controller->DriverScatterGatherLimit); + blk_queue_max_sectors(RequestQueue, Controller->MaxBlocksPerCommand); + disk->queue = RequestQueue; + sprintf(disk->disk_name, "rd/c%dd%d", Controller->ControllerNumber, n); + sprintf(disk->devfs_name, "rd/host%d/target%d", Controller->ControllerNumber, n); + disk->major = MajorNumber; + disk->first_minor = n << DAC960_MaxPartitionsBits; + disk->fops = &DAC960_BlockDeviceOperations; + } + /* + Indicate the Block Device Registration completed successfully, + */ + return true; +} + + +/* + DAC960_UnregisterBlockDevice unregisters the Block Device structures + associated with Controller. +*/ + +static void DAC960_UnregisterBlockDevice(DAC960_Controller_T *Controller) +{ + int MajorNumber = DAC960_MAJOR + Controller->ControllerNumber; + int disk; + + /* does order matter when deleting gendisk and cleanup in request queue? */ + for (disk = 0; disk < DAC960_MaxLogicalDrives; disk++) { + del_gendisk(Controller->disks[disk]); + blk_cleanup_queue(Controller->RequestQueue[disk]); + Controller->RequestQueue[disk] = NULL; + } + + /* + Unregister the Block Device Major Number for this DAC960 Controller. + */ + unregister_blkdev(MajorNumber, "dac960"); +} + +/* + DAC960_ComputeGenericDiskInfo computes the values for the Generic Disk + Information Partition Sector Counts and Block Sizes. +*/ + +static void DAC960_ComputeGenericDiskInfo(DAC960_Controller_T *Controller) +{ + int disk; + for (disk = 0; disk < DAC960_MaxLogicalDrives; disk++) + set_capacity(Controller->disks[disk], disk_size(Controller, disk)); +} + +/* + DAC960_ReportErrorStatus reports Controller BIOS Messages passed through + the Error Status Register when the driver performs the BIOS handshaking. + It returns true for fatal errors and false otherwise. +*/ + +static boolean DAC960_ReportErrorStatus(DAC960_Controller_T *Controller, + unsigned char ErrorStatus, + unsigned char Parameter0, + unsigned char Parameter1) +{ + switch (ErrorStatus) + { + case 0x00: + DAC960_Notice("Physical Device %d:%d Not Responding\n", + Controller, Parameter1, Parameter0); + break; + case 0x08: + if (Controller->DriveSpinUpMessageDisplayed) break; + DAC960_Notice("Spinning Up Drives\n", Controller); + Controller->DriveSpinUpMessageDisplayed = true; + break; + case 0x30: + DAC960_Notice("Configuration Checksum Error\n", Controller); + break; + case 0x60: + DAC960_Notice("Mirror Race Recovery Failed\n", Controller); + break; + case 0x70: + DAC960_Notice("Mirror Race Recovery In Progress\n", Controller); + break; + case 0x90: + DAC960_Notice("Physical Device %d:%d COD Mismatch\n", + Controller, Parameter1, Parameter0); + break; + case 0xA0: + DAC960_Notice("Logical Drive Installation Aborted\n", Controller); + break; + case 0xB0: + DAC960_Notice("Mirror Race On A Critical Logical Drive\n", Controller); + break; + case 0xD0: + DAC960_Notice("New Controller Configuration Found\n", Controller); + break; + case 0xF0: + DAC960_Error("Fatal Memory Parity Error for Controller at\n", Controller); + return true; + default: + DAC960_Error("Unknown Initialization Error %02X for Controller at\n", + Controller, ErrorStatus); + return true; + } + return false; +} + + +/* + * DAC960_DetectCleanup releases the resources that were allocated + * during DAC960_DetectController(). DAC960_DetectController can + * has several internal failure points, so not ALL resources may + * have been allocated. It's important to free only + * resources that HAVE been allocated. The code below always + * tests that the resource has been allocated before attempting to + * free it. + */ +static void DAC960_DetectCleanup(DAC960_Controller_T *Controller) +{ + int i; + + /* Free the memory mailbox, status, and related structures */ + free_dma_loaf(Controller->PCIDevice, &Controller->DmaPages); + if (Controller->MemoryMappedAddress) { + switch(Controller->HardwareType) + { + case DAC960_BA_Controller: + DAC960_BA_DisableInterrupts(Controller->BaseAddress); + break; + case DAC960_LP_Controller: + DAC960_LP_DisableInterrupts(Controller->BaseAddress); + break; + case DAC960_LA_Controller: + DAC960_LA_DisableInterrupts(Controller->BaseAddress); + break; + case DAC960_PG_Controller: + DAC960_PG_DisableInterrupts(Controller->BaseAddress); + break; + case DAC960_PD_Controller: + DAC960_PD_DisableInterrupts(Controller->BaseAddress); + break; + case DAC960_P_Controller: + DAC960_PD_DisableInterrupts(Controller->BaseAddress); + break; + } + iounmap(Controller->MemoryMappedAddress); + } + if (Controller->IRQ_Channel) + free_irq(Controller->IRQ_Channel, Controller); + if (Controller->IO_Address) + release_region(Controller->IO_Address, 0x80); + pci_disable_device(Controller->PCIDevice); + for (i = 0; (i < DAC960_MaxLogicalDrives) && Controller->disks[i]; i++) + put_disk(Controller->disks[i]); + DAC960_Controllers[Controller->ControllerNumber] = NULL; + kfree(Controller); +} + + +/* + DAC960_DetectController detects Mylex DAC960/AcceleRAID/eXtremeRAID + PCI RAID Controllers by interrogating the PCI Configuration Space for + Controller Type. +*/ + +static DAC960_Controller_T * +DAC960_DetectController(struct pci_dev *PCI_Device, + const struct pci_device_id *entry) +{ + struct DAC960_privdata *privdata = + (struct DAC960_privdata *)entry->driver_data; + irqreturn_t (*InterruptHandler)(int, void *, struct pt_regs *) = + privdata->InterruptHandler; + unsigned int MemoryWindowSize = privdata->MemoryWindowSize; + DAC960_Controller_T *Controller = NULL; + unsigned char DeviceFunction = PCI_Device->devfn; + unsigned char ErrorStatus, Parameter0, Parameter1; + unsigned int IRQ_Channel; + void __iomem *BaseAddress; + int i; + + Controller = (DAC960_Controller_T *) + kmalloc(sizeof(DAC960_Controller_T), GFP_ATOMIC); + if (Controller == NULL) { + DAC960_Error("Unable to allocate Controller structure for " + "Controller at\n", NULL); + return NULL; + } + memset(Controller, 0, sizeof(DAC960_Controller_T)); + Controller->ControllerNumber = DAC960_ControllerCount; + DAC960_Controllers[DAC960_ControllerCount++] = Controller; + Controller->Bus = PCI_Device->bus->number; + Controller->FirmwareType = privdata->FirmwareType; + Controller->HardwareType = privdata->HardwareType; + Controller->Device = DeviceFunction >> 3; + Controller->Function = DeviceFunction & 0x7; + Controller->PCIDevice = PCI_Device; + strcpy(Controller->FullModelName, "DAC960"); + + if (pci_enable_device(PCI_Device)) + goto Failure; + + switch (Controller->HardwareType) + { + case DAC960_BA_Controller: + Controller->PCI_Address = pci_resource_start(PCI_Device, 0); + break; + case DAC960_LP_Controller: + Controller->PCI_Address = pci_resource_start(PCI_Device, 0); + break; + case DAC960_LA_Controller: + Controller->PCI_Address = pci_resource_start(PCI_Device, 0); + break; + case DAC960_PG_Controller: + Controller->PCI_Address = pci_resource_start(PCI_Device, 0); + break; + case DAC960_PD_Controller: + Controller->IO_Address = pci_resource_start(PCI_Device, 0); + Controller->PCI_Address = pci_resource_start(PCI_Device, 1); + break; + case DAC960_P_Controller: + Controller->IO_Address = pci_resource_start(PCI_Device, 0); + Controller->PCI_Address = pci_resource_start(PCI_Device, 1); + break; + } + + pci_set_drvdata(PCI_Device, (void *)((long)Controller->ControllerNumber)); + for (i = 0; i < DAC960_MaxLogicalDrives; i++) { + Controller->disks[i] = alloc_disk(1<<DAC960_MaxPartitionsBits); + if (!Controller->disks[i]) + goto Failure; + Controller->disks[i]->private_data = (void *)((long)i); + } + init_waitqueue_head(&Controller->CommandWaitQueue); + init_waitqueue_head(&Controller->HealthStatusWaitQueue); + spin_lock_init(&Controller->queue_lock); + DAC960_AnnounceDriver(Controller); + /* + Map the Controller Register Window. + */ + if (MemoryWindowSize < PAGE_SIZE) + MemoryWindowSize = PAGE_SIZE; + Controller->MemoryMappedAddress = + ioremap_nocache(Controller->PCI_Address & PAGE_MASK, MemoryWindowSize); + Controller->BaseAddress = + Controller->MemoryMappedAddress + (Controller->PCI_Address & ~PAGE_MASK); + if (Controller->MemoryMappedAddress == NULL) + { + DAC960_Error("Unable to map Controller Register Window for " + "Controller at\n", Controller); + goto Failure; + } + BaseAddress = Controller->BaseAddress; + switch (Controller->HardwareType) + { + case DAC960_BA_Controller: + DAC960_BA_DisableInterrupts(BaseAddress); + DAC960_BA_AcknowledgeHardwareMailboxStatus(BaseAddress); + udelay(1000); + while (DAC960_BA_InitializationInProgressP(BaseAddress)) + { + if (DAC960_BA_ReadErrorStatus(BaseAddress, &ErrorStatus, + &Parameter0, &Parameter1) && + DAC960_ReportErrorStatus(Controller, ErrorStatus, + Parameter0, Parameter1)) + goto Failure; + udelay(10); + } + if (!DAC960_V2_EnableMemoryMailboxInterface(Controller)) + { + DAC960_Error("Unable to Enable Memory Mailbox Interface " + "for Controller at\n", Controller); + goto Failure; + } + DAC960_BA_EnableInterrupts(BaseAddress); + Controller->QueueCommand = DAC960_BA_QueueCommand; + Controller->ReadControllerConfiguration = + DAC960_V2_ReadControllerConfiguration; + Controller->ReadDeviceConfiguration = + DAC960_V2_ReadDeviceConfiguration; + Controller->ReportDeviceConfiguration = + DAC960_V2_ReportDeviceConfiguration; + Controller->QueueReadWriteCommand = + DAC960_V2_QueueReadWriteCommand; + break; + case DAC960_LP_Controller: + DAC960_LP_DisableInterrupts(BaseAddress); + DAC960_LP_AcknowledgeHardwareMailboxStatus(BaseAddress); + udelay(1000); + while (DAC960_LP_InitializationInProgressP(BaseAddress)) + { + if (DAC960_LP_ReadErrorStatus(BaseAddress, &ErrorStatus, + &Parameter0, &Parameter1) && + DAC960_ReportErrorStatus(Controller, ErrorStatus, + Parameter0, Parameter1)) + goto Failure; + udelay(10); + } + if (!DAC960_V2_EnableMemoryMailboxInterface(Controller)) + { + DAC960_Error("Unable to Enable Memory Mailbox Interface " + "for Controller at\n", Controller); + goto Failure; + } + DAC960_LP_EnableInterrupts(BaseAddress); + Controller->QueueCommand = DAC960_LP_QueueCommand; + Controller->ReadControllerConfiguration = + DAC960_V2_ReadControllerConfiguration; + Controller->ReadDeviceConfiguration = + DAC960_V2_ReadDeviceConfiguration; + Controller->ReportDeviceConfiguration = + DAC960_V2_ReportDeviceConfiguration; + Controller->QueueReadWriteCommand = + DAC960_V2_QueueReadWriteCommand; + break; + case DAC960_LA_Controller: + DAC960_LA_DisableInterrupts(BaseAddress); + DAC960_LA_AcknowledgeHardwareMailboxStatus(BaseAddress); + udelay(1000); + while (DAC960_LA_InitializationInProgressP(BaseAddress)) + { + if (DAC960_LA_ReadErrorStatus(BaseAddress, &ErrorStatus, + &Parameter0, &Parameter1) && + DAC960_ReportErrorStatus(Controller, ErrorStatus, + Parameter0, Parameter1)) + goto Failure; + udelay(10); + } + if (!DAC960_V1_EnableMemoryMailboxInterface(Controller)) + { + DAC960_Error("Unable to Enable Memory Mailbox Interface " + "for Controller at\n", Controller); + goto Failure; + } + DAC960_LA_EnableInterrupts(BaseAddress); + if (Controller->V1.DualModeMemoryMailboxInterface) + Controller->QueueCommand = DAC960_LA_QueueCommandDualMode; + else Controller->QueueCommand = DAC960_LA_QueueCommandSingleMode; + Controller->ReadControllerConfiguration = + DAC960_V1_ReadControllerConfiguration; + Controller->ReadDeviceConfiguration = + DAC960_V1_ReadDeviceConfiguration; + Controller->ReportDeviceConfiguration = + DAC960_V1_ReportDeviceConfiguration; + Controller->QueueReadWriteCommand = + DAC960_V1_QueueReadWriteCommand; + break; + case DAC960_PG_Controller: + DAC960_PG_DisableInterrupts(BaseAddress); + DAC960_PG_AcknowledgeHardwareMailboxStatus(BaseAddress); + udelay(1000); + while (DAC960_PG_InitializationInProgressP(BaseAddress)) + { + if (DAC960_PG_ReadErrorStatus(BaseAddress, &ErrorStatus, + &Parameter0, &Parameter1) && + DAC960_ReportErrorStatus(Controller, ErrorStatus, + Parameter0, Parameter1)) + goto Failure; + udelay(10); + } + if (!DAC960_V1_EnableMemoryMailboxInterface(Controller)) + { + DAC960_Error("Unable to Enable Memory Mailbox Interface " + "for Controller at\n", Controller); + goto Failure; + } + DAC960_PG_EnableInterrupts(BaseAddress); + if (Controller->V1.DualModeMemoryMailboxInterface) + Controller->QueueCommand = DAC960_PG_QueueCommandDualMode; + else Controller->QueueCommand = DAC960_PG_QueueCommandSingleMode; + Controller->ReadControllerConfiguration = + DAC960_V1_ReadControllerConfiguration; + Controller->ReadDeviceConfiguration = + DAC960_V1_ReadDeviceConfiguration; + Controller->ReportDeviceConfiguration = + DAC960_V1_ReportDeviceConfiguration; + Controller->QueueReadWriteCommand = + DAC960_V1_QueueReadWriteCommand; + break; + case DAC960_PD_Controller: + if (!request_region(Controller->IO_Address, 0x80, + Controller->FullModelName)) { + DAC960_Error("IO port 0x%d busy for Controller at\n", + Controller, Controller->IO_Address); + goto Failure; + } + DAC960_PD_DisableInterrupts(BaseAddress); + DAC960_PD_AcknowledgeStatus(BaseAddress); + udelay(1000); + while (DAC960_PD_InitializationInProgressP(BaseAddress)) + { + if (DAC960_PD_ReadErrorStatus(BaseAddress, &ErrorStatus, + &Parameter0, &Parameter1) && + DAC960_ReportErrorStatus(Controller, ErrorStatus, + Parameter0, Parameter1)) + goto Failure; + udelay(10); + } + if (!DAC960_V1_EnableMemoryMailboxInterface(Controller)) + { + DAC960_Error("Unable to allocate DMA mapped memory " + "for Controller at\n", Controller); + goto Failure; + } + DAC960_PD_EnableInterrupts(BaseAddress); + Controller->QueueCommand = DAC960_PD_QueueCommand; + Controller->ReadControllerConfiguration = + DAC960_V1_ReadControllerConfiguration; + Controller->ReadDeviceConfiguration = + DAC960_V1_ReadDeviceConfiguration; + Controller->ReportDeviceConfiguration = + DAC960_V1_ReportDeviceConfiguration; + Controller->QueueReadWriteCommand = + DAC960_V1_QueueReadWriteCommand; + break; + case DAC960_P_Controller: + if (!request_region(Controller->IO_Address, 0x80, + Controller->FullModelName)){ + DAC960_Error("IO port 0x%d busy for Controller at\n", + Controller, Controller->IO_Address); + goto Failure; + } + DAC960_PD_DisableInterrupts(BaseAddress); + DAC960_PD_AcknowledgeStatus(BaseAddress); + udelay(1000); + while (DAC960_PD_InitializationInProgressP(BaseAddress)) + { + if (DAC960_PD_ReadErrorStatus(BaseAddress, &ErrorStatus, + &Parameter0, &Parameter1) && + DAC960_ReportErrorStatus(Controller, ErrorStatus, + Parameter0, Parameter1)) + goto Failure; + udelay(10); + } + if (!DAC960_V1_EnableMemoryMailboxInterface(Controller)) + { + DAC960_Error("Unable to allocate DMA mapped memory" + "for Controller at\n", Controller); + goto Failure; + } + DAC960_PD_EnableInterrupts(BaseAddress); + Controller->QueueCommand = DAC960_P_QueueCommand; + Controller->ReadControllerConfiguration = + DAC960_V1_ReadControllerConfiguration; + Controller->ReadDeviceConfiguration = + DAC960_V1_ReadDeviceConfiguration; + Controller->ReportDeviceConfiguration = + DAC960_V1_ReportDeviceConfiguration; + Controller->QueueReadWriteCommand = + DAC960_V1_QueueReadWriteCommand; + break; + } + /* + Acquire shared access to the IRQ Channel. + */ + IRQ_Channel = PCI_Device->irq; + if (request_irq(IRQ_Channel, InterruptHandler, SA_SHIRQ, + Controller->FullModelName, Controller) < 0) + { + DAC960_Error("Unable to acquire IRQ Channel %d for Controller at\n", + Controller, Controller->IRQ_Channel); + goto Failure; + } + Controller->IRQ_Channel = IRQ_Channel; + Controller->InitialCommand.CommandIdentifier = 1; + Controller->InitialCommand.Controller = Controller; + Controller->Commands[0] = &Controller->InitialCommand; + Controller->FreeCommands = &Controller->InitialCommand; + return Controller; + +Failure: + if (Controller->IO_Address == 0) + DAC960_Error("PCI Bus %d Device %d Function %d I/O Address N/A " + "PCI Address 0x%X\n", Controller, + Controller->Bus, Controller->Device, + Controller->Function, Controller->PCI_Address); + else + DAC960_Error("PCI Bus %d Device %d Function %d I/O Address " + "0x%X PCI Address 0x%X\n", Controller, + Controller->Bus, Controller->Device, + Controller->Function, Controller->IO_Address, + Controller->PCI_Address); + DAC960_DetectCleanup(Controller); + DAC960_ControllerCount--; + return NULL; +} + +/* + DAC960_InitializeController initializes Controller. +*/ + +static boolean +DAC960_InitializeController(DAC960_Controller_T *Controller) +{ + if (DAC960_ReadControllerConfiguration(Controller) && + DAC960_ReportControllerConfiguration(Controller) && + DAC960_CreateAuxiliaryStructures(Controller) && + DAC960_ReadDeviceConfiguration(Controller) && + DAC960_ReportDeviceConfiguration(Controller) && + DAC960_RegisterBlockDevice(Controller)) + { + /* + Initialize the Monitoring Timer. + */ + init_timer(&Controller->MonitoringTimer); + Controller->MonitoringTimer.expires = + jiffies + DAC960_MonitoringTimerInterval; + Controller->MonitoringTimer.data = (unsigned long) Controller; + Controller->MonitoringTimer.function = DAC960_MonitoringTimerFunction; + add_timer(&Controller->MonitoringTimer); + Controller->ControllerInitialized = true; + return true; + } + return false; +} + + +/* + DAC960_FinalizeController finalizes Controller. +*/ + +static void DAC960_FinalizeController(DAC960_Controller_T *Controller) +{ + if (Controller->ControllerInitialized) + { + unsigned long flags; + + /* + * Acquiring and releasing lock here eliminates + * a very low probability race. + * + * The code below allocates controller command structures + * from the free list without holding the controller lock. + * This is safe assuming there is no other activity on + * the controller at the time. + * + * But, there might be a monitoring command still + * in progress. Setting the Shutdown flag while holding + * the lock ensures that there is no monitoring command + * in the interrupt handler currently, and any monitoring + * commands that complete from this time on will NOT return + * their command structure to the free list. + */ + + spin_lock_irqsave(&Controller->queue_lock, flags); + Controller->ShutdownMonitoringTimer = 1; + spin_unlock_irqrestore(&Controller->queue_lock, flags); + + del_timer_sync(&Controller->MonitoringTimer); + if (Controller->FirmwareType == DAC960_V1_Controller) + { + DAC960_Notice("Flushing Cache...", Controller); + DAC960_V1_ExecuteType3(Controller, DAC960_V1_Flush, 0); + DAC960_Notice("done\n", Controller); + + if (Controller->HardwareType == DAC960_PD_Controller) + release_region(Controller->IO_Address, 0x80); + } + else + { + DAC960_Notice("Flushing Cache...", Controller); + DAC960_V2_DeviceOperation(Controller, DAC960_V2_PauseDevice, + DAC960_V2_RAID_Controller); + DAC960_Notice("done\n", Controller); + } + } + DAC960_UnregisterBlockDevice(Controller); + DAC960_DestroyAuxiliaryStructures(Controller); + DAC960_DestroyProcEntries(Controller); + DAC960_DetectCleanup(Controller); +} + + +/* + DAC960_Probe verifies controller's existence and + initializes the DAC960 Driver for that controller. +*/ + +static int +DAC960_Probe(struct pci_dev *dev, const struct pci_device_id *entry) +{ + int disk; + DAC960_Controller_T *Controller; + + if (DAC960_ControllerCount == DAC960_MaxControllers) + { + DAC960_Error("More than %d DAC960 Controllers detected - " + "ignoring from Controller at\n", + NULL, DAC960_MaxControllers); + return -ENODEV; + } + + Controller = DAC960_DetectController(dev, entry); + if (!Controller) + return -ENODEV; + + if (!DAC960_InitializeController(Controller)) { + DAC960_FinalizeController(Controller); + return -ENODEV; + } + + for (disk = 0; disk < DAC960_MaxLogicalDrives; disk++) { + set_capacity(Controller->disks[disk], disk_size(Controller, disk)); + add_disk(Controller->disks[disk]); + } + DAC960_CreateProcEntries(Controller); + return 0; +} + + +/* + DAC960_Finalize finalizes the DAC960 Driver. +*/ + +static void DAC960_Remove(struct pci_dev *PCI_Device) +{ + int Controller_Number = (long)pci_get_drvdata(PCI_Device); + DAC960_Controller_T *Controller = DAC960_Controllers[Controller_Number]; + if (Controller != NULL) + DAC960_FinalizeController(Controller); +} + + +/* + DAC960_V1_QueueReadWriteCommand prepares and queues a Read/Write Command for + DAC960 V1 Firmware Controllers. +*/ + +static void DAC960_V1_QueueReadWriteCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_ScatterGatherSegment_T *ScatterGatherList = + Command->V1.ScatterGatherList; + struct scatterlist *ScatterList = Command->V1.ScatterList; + + DAC960_V1_ClearCommand(Command); + + if (Command->SegmentCount == 1) + { + if (Command->DmaDirection == PCI_DMA_FROMDEVICE) + CommandMailbox->Type5.CommandOpcode = DAC960_V1_Read; + else + CommandMailbox->Type5.CommandOpcode = DAC960_V1_Write; + + CommandMailbox->Type5.LD.TransferLength = Command->BlockCount; + CommandMailbox->Type5.LD.LogicalDriveNumber = Command->LogicalDriveNumber; + CommandMailbox->Type5.LogicalBlockAddress = Command->BlockNumber; + CommandMailbox->Type5.BusAddress = + (DAC960_BusAddress32_T)sg_dma_address(ScatterList); + } + else + { + int i; + + if (Command->DmaDirection == PCI_DMA_FROMDEVICE) + CommandMailbox->Type5.CommandOpcode = DAC960_V1_ReadWithScatterGather; + else + CommandMailbox->Type5.CommandOpcode = DAC960_V1_WriteWithScatterGather; + + CommandMailbox->Type5.LD.TransferLength = Command->BlockCount; + CommandMailbox->Type5.LD.LogicalDriveNumber = Command->LogicalDriveNumber; + CommandMailbox->Type5.LogicalBlockAddress = Command->BlockNumber; + CommandMailbox->Type5.BusAddress = Command->V1.ScatterGatherListDMA; + + CommandMailbox->Type5.ScatterGatherCount = Command->SegmentCount; + + for (i = 0; i < Command->SegmentCount; i++, ScatterList++, ScatterGatherList++) { + ScatterGatherList->SegmentDataPointer = + (DAC960_BusAddress32_T)sg_dma_address(ScatterList); + ScatterGatherList->SegmentByteCount = + (DAC960_ByteCount32_T)sg_dma_len(ScatterList); + } + } + DAC960_QueueCommand(Command); +} + + +/* + DAC960_V2_QueueReadWriteCommand prepares and queues a Read/Write Command for + DAC960 V2 Firmware Controllers. +*/ + +static void DAC960_V2_QueueReadWriteCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + struct scatterlist *ScatterList = Command->V2.ScatterList; + + DAC960_V2_ClearCommand(Command); + + CommandMailbox->SCSI_10.CommandOpcode = DAC960_V2_SCSI_10; + CommandMailbox->SCSI_10.CommandControlBits.DataTransferControllerToHost = + (Command->DmaDirection == PCI_DMA_FROMDEVICE); + CommandMailbox->SCSI_10.DataTransferSize = + Command->BlockCount << DAC960_BlockSizeBits; + CommandMailbox->SCSI_10.RequestSenseBusAddress = Command->V2.RequestSenseDMA; + CommandMailbox->SCSI_10.PhysicalDevice = + Controller->V2.LogicalDriveToVirtualDevice[Command->LogicalDriveNumber]; + CommandMailbox->SCSI_10.RequestSenseSize = sizeof(DAC960_SCSI_RequestSense_T); + CommandMailbox->SCSI_10.CDBLength = 10; + CommandMailbox->SCSI_10.SCSI_CDB[0] = + (Command->DmaDirection == PCI_DMA_FROMDEVICE ? 0x28 : 0x2A); + CommandMailbox->SCSI_10.SCSI_CDB[2] = Command->BlockNumber >> 24; + CommandMailbox->SCSI_10.SCSI_CDB[3] = Command->BlockNumber >> 16; + CommandMailbox->SCSI_10.SCSI_CDB[4] = Command->BlockNumber >> 8; + CommandMailbox->SCSI_10.SCSI_CDB[5] = Command->BlockNumber; + CommandMailbox->SCSI_10.SCSI_CDB[7] = Command->BlockCount >> 8; + CommandMailbox->SCSI_10.SCSI_CDB[8] = Command->BlockCount; + + if (Command->SegmentCount == 1) + { + CommandMailbox->SCSI_10.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + (DAC960_BusAddress64_T)sg_dma_address(ScatterList); + CommandMailbox->SCSI_10.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->SCSI_10.DataTransferSize; + } + else + { + DAC960_V2_ScatterGatherSegment_T *ScatterGatherList; + int i; + + if (Command->SegmentCount > 2) + { + ScatterGatherList = Command->V2.ScatterGatherList; + CommandMailbox->SCSI_10.CommandControlBits + .AdditionalScatterGatherListMemory = true; + CommandMailbox->SCSI_10.DataTransferMemoryAddress + .ExtendedScatterGather.ScatterGatherList0Length = Command->SegmentCount; + CommandMailbox->SCSI_10.DataTransferMemoryAddress + .ExtendedScatterGather.ScatterGatherList0Address = + Command->V2.ScatterGatherListDMA; + } + else + ScatterGatherList = CommandMailbox->SCSI_10.DataTransferMemoryAddress + .ScatterGatherSegments; + + for (i = 0; i < Command->SegmentCount; i++, ScatterList++, ScatterGatherList++) { + ScatterGatherList->SegmentDataPointer = + (DAC960_BusAddress64_T)sg_dma_address(ScatterList); + ScatterGatherList->SegmentByteCount = + (DAC960_ByteCount64_T)sg_dma_len(ScatterList); + } + } + DAC960_QueueCommand(Command); +} + + +static int DAC960_process_queue(DAC960_Controller_T *Controller, struct request_queue *req_q) +{ + struct request *Request; + DAC960_Command_T *Command; + + while(1) { + Request = elv_next_request(req_q); + if (!Request) + return 1; + + Command = DAC960_AllocateCommand(Controller); + if (Command == NULL) + return 0; + + if (rq_data_dir(Request) == READ) { + Command->DmaDirection = PCI_DMA_FROMDEVICE; + Command->CommandType = DAC960_ReadCommand; + } else { + Command->DmaDirection = PCI_DMA_TODEVICE; + Command->CommandType = DAC960_WriteCommand; + } + Command->Completion = Request->waiting; + Command->LogicalDriveNumber = (long)Request->rq_disk->private_data; + Command->BlockNumber = Request->sector; + Command->BlockCount = Request->nr_sectors; + Command->Request = Request; + blkdev_dequeue_request(Request); + Command->SegmentCount = blk_rq_map_sg(req_q, + Command->Request, Command->cmd_sglist); + /* pci_map_sg MAY change the value of SegCount */ + Command->SegmentCount = pci_map_sg(Controller->PCIDevice, Command->cmd_sglist, + Command->SegmentCount, Command->DmaDirection); + + DAC960_QueueReadWriteCommand(Command); + } +} + +/* + DAC960_ProcessRequest attempts to remove one I/O Request from Controller's + I/O Request Queue and queues it to the Controller. WaitForCommand is true if + this function should wait for a Command to become available if necessary. + This function returns true if an I/O Request was queued and false otherwise. +*/ +static void DAC960_ProcessRequest(DAC960_Controller_T *controller) +{ + int i; + + if (!controller->ControllerInitialized) + return; + + /* Do this better later! */ + for (i = controller->req_q_index; i < DAC960_MaxLogicalDrives; i++) { + struct request_queue *req_q = controller->RequestQueue[i]; + + if (req_q == NULL) + continue; + + if (!DAC960_process_queue(controller, req_q)) { + controller->req_q_index = i; + return; + } + } + + if (controller->req_q_index == 0) + return; + + for (i = 0; i < controller->req_q_index; i++) { + struct request_queue *req_q = controller->RequestQueue[i]; + + if (req_q == NULL) + continue; + + if (!DAC960_process_queue(controller, req_q)) { + controller->req_q_index = i; + return; + } + } +} + + +/* + DAC960_queue_partial_rw extracts one bio from the request already + associated with argument command, and construct a new command block to retry I/O + only on that bio. Queue that command to the controller. + + This function re-uses a previously-allocated Command, + there is no failure mode from trying to allocate a command. +*/ + +static void DAC960_queue_partial_rw(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + struct request *Request = Command->Request; + struct request_queue *req_q = Controller->RequestQueue[Command->LogicalDriveNumber]; + + if (Command->DmaDirection == PCI_DMA_FROMDEVICE) + Command->CommandType = DAC960_ReadRetryCommand; + else + Command->CommandType = DAC960_WriteRetryCommand; + + /* + * We could be more efficient with these mapping requests + * and map only the portions that we need. But since this + * code should almost never be called, just go with a + * simple coding. + */ + (void)blk_rq_map_sg(req_q, Command->Request, Command->cmd_sglist); + + (void)pci_map_sg(Controller->PCIDevice, Command->cmd_sglist, 1, Command->DmaDirection); + /* + * Resubmitting the request sector at a time is really tedious. + * But, this should almost never happen. So, we're willing to pay + * this price so that in the end, as much of the transfer is completed + * successfully as possible. + */ + Command->SegmentCount = 1; + Command->BlockNumber = Request->sector; + Command->BlockCount = 1; + DAC960_QueueReadWriteCommand(Command); + return; +} + +/* + DAC960_RequestFunction is the I/O Request Function for DAC960 Controllers. +*/ + +static void DAC960_RequestFunction(struct request_queue *RequestQueue) +{ + DAC960_ProcessRequest(RequestQueue->queuedata); +} + +/* + DAC960_ProcessCompletedBuffer performs completion processing for an + individual Buffer. +*/ + +static inline boolean DAC960_ProcessCompletedRequest(DAC960_Command_T *Command, + boolean SuccessfulIO) +{ + struct request *Request = Command->Request; + int UpToDate; + + UpToDate = 0; + if (SuccessfulIO) + UpToDate = 1; + + pci_unmap_sg(Command->Controller->PCIDevice, Command->cmd_sglist, + Command->SegmentCount, Command->DmaDirection); + + if (!end_that_request_first(Request, UpToDate, Command->BlockCount)) { + + end_that_request_last(Request); + + if (Command->Completion) { + complete(Command->Completion); + Command->Completion = NULL; + } + return true; + } + return false; +} + +/* + DAC960_V1_ReadWriteError prints an appropriate error message for Command + when an error occurs on a Read or Write operation. +*/ + +static void DAC960_V1_ReadWriteError(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + unsigned char *CommandName = "UNKNOWN"; + switch (Command->CommandType) + { + case DAC960_ReadCommand: + case DAC960_ReadRetryCommand: + CommandName = "READ"; + break; + case DAC960_WriteCommand: + case DAC960_WriteRetryCommand: + CommandName = "WRITE"; + break; + case DAC960_MonitoringCommand: + case DAC960_ImmediateCommand: + case DAC960_QueuedCommand: + break; + } + switch (Command->V1.CommandStatus) + { + case DAC960_V1_IrrecoverableDataError: + DAC960_Error("Irrecoverable Data Error on %s:\n", + Controller, CommandName); + break; + case DAC960_V1_LogicalDriveNonexistentOrOffline: + DAC960_Error("Logical Drive Nonexistent or Offline on %s:\n", + Controller, CommandName); + break; + case DAC960_V1_AccessBeyondEndOfLogicalDrive: + DAC960_Error("Attempt to Access Beyond End of Logical Drive " + "on %s:\n", Controller, CommandName); + break; + case DAC960_V1_BadDataEncountered: + DAC960_Error("Bad Data Encountered on %s:\n", Controller, CommandName); + break; + default: + DAC960_Error("Unexpected Error Status %04X on %s:\n", + Controller, Command->V1.CommandStatus, CommandName); + break; + } + DAC960_Error(" /dev/rd/c%dd%d: absolute blocks %u..%u\n", + Controller, Controller->ControllerNumber, + Command->LogicalDriveNumber, Command->BlockNumber, + Command->BlockNumber + Command->BlockCount - 1); +} + + +/* + DAC960_V1_ProcessCompletedCommand performs completion processing for Command + for DAC960 V1 Firmware Controllers. +*/ + +static void DAC960_V1_ProcessCompletedCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + DAC960_CommandType_T CommandType = Command->CommandType; + DAC960_V1_CommandOpcode_T CommandOpcode = + Command->V1.CommandMailbox.Common.CommandOpcode; + DAC960_V1_CommandStatus_T CommandStatus = Command->V1.CommandStatus; + + if (CommandType == DAC960_ReadCommand || + CommandType == DAC960_WriteCommand) + { + +#ifdef FORCE_RETRY_DEBUG + CommandStatus = DAC960_V1_IrrecoverableDataError; +#endif + + if (CommandStatus == DAC960_V1_NormalCompletion) { + + if (!DAC960_ProcessCompletedRequest(Command, true)) + BUG(); + + } else if (CommandStatus == DAC960_V1_IrrecoverableDataError || + CommandStatus == DAC960_V1_BadDataEncountered) + { + /* + * break the command down into pieces and resubmit each + * piece, hoping that some of them will succeed. + */ + DAC960_queue_partial_rw(Command); + return; + } + else + { + if (CommandStatus != DAC960_V1_LogicalDriveNonexistentOrOffline) + DAC960_V1_ReadWriteError(Command); + + if (!DAC960_ProcessCompletedRequest(Command, false)) + BUG(); + } + } + else if (CommandType == DAC960_ReadRetryCommand || + CommandType == DAC960_WriteRetryCommand) + { + boolean normal_completion; +#ifdef FORCE_RETRY_FAILURE_DEBUG + static int retry_count = 1; +#endif + /* + Perform completion processing for the portion that was + retried, and submit the next portion, if any. + */ + normal_completion = true; + if (CommandStatus != DAC960_V1_NormalCompletion) { + normal_completion = false; + if (CommandStatus != DAC960_V1_LogicalDriveNonexistentOrOffline) + DAC960_V1_ReadWriteError(Command); + } + +#ifdef FORCE_RETRY_FAILURE_DEBUG + if (!(++retry_count % 10000)) { + printk("V1 error retry failure test\n"); + normal_completion = false; + DAC960_V1_ReadWriteError(Command); + } +#endif + + if (!DAC960_ProcessCompletedRequest(Command, normal_completion)) { + DAC960_queue_partial_rw(Command); + return; + } + } + + else if (CommandType == DAC960_MonitoringCommand) + { + if (Controller->ShutdownMonitoringTimer) + return; + if (CommandOpcode == DAC960_V1_Enquiry) + { + DAC960_V1_Enquiry_T *OldEnquiry = &Controller->V1.Enquiry; + DAC960_V1_Enquiry_T *NewEnquiry = Controller->V1.NewEnquiry; + unsigned int OldCriticalLogicalDriveCount = + OldEnquiry->CriticalLogicalDriveCount; + unsigned int NewCriticalLogicalDriveCount = + NewEnquiry->CriticalLogicalDriveCount; + if (NewEnquiry->NumberOfLogicalDrives > Controller->LogicalDriveCount) + { + int LogicalDriveNumber = Controller->LogicalDriveCount - 1; + while (++LogicalDriveNumber < NewEnquiry->NumberOfLogicalDrives) + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) " + "Now Exists\n", Controller, + LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber); + Controller->LogicalDriveCount = NewEnquiry->NumberOfLogicalDrives; + DAC960_ComputeGenericDiskInfo(Controller); + } + if (NewEnquiry->NumberOfLogicalDrives < Controller->LogicalDriveCount) + { + int LogicalDriveNumber = NewEnquiry->NumberOfLogicalDrives - 1; + while (++LogicalDriveNumber < Controller->LogicalDriveCount) + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) " + "No Longer Exists\n", Controller, + LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber); + Controller->LogicalDriveCount = NewEnquiry->NumberOfLogicalDrives; + DAC960_ComputeGenericDiskInfo(Controller); + } + if (NewEnquiry->StatusFlags.DeferredWriteError != + OldEnquiry->StatusFlags.DeferredWriteError) + DAC960_Critical("Deferred Write Error Flag is now %s\n", Controller, + (NewEnquiry->StatusFlags.DeferredWriteError + ? "TRUE" : "FALSE")); + if ((NewCriticalLogicalDriveCount > 0 || + NewCriticalLogicalDriveCount != OldCriticalLogicalDriveCount) || + (NewEnquiry->OfflineLogicalDriveCount > 0 || + NewEnquiry->OfflineLogicalDriveCount != + OldEnquiry->OfflineLogicalDriveCount) || + (NewEnquiry->DeadDriveCount > 0 || + NewEnquiry->DeadDriveCount != + OldEnquiry->DeadDriveCount) || + (NewEnquiry->EventLogSequenceNumber != + OldEnquiry->EventLogSequenceNumber) || + Controller->MonitoringTimerCount == 0 || + (jiffies - Controller->SecondaryMonitoringTime + >= DAC960_SecondaryMonitoringInterval)) + { + Controller->V1.NeedLogicalDriveInformation = true; + Controller->V1.NewEventLogSequenceNumber = + NewEnquiry->EventLogSequenceNumber; + Controller->V1.NeedErrorTableInformation = true; + Controller->V1.NeedDeviceStateInformation = true; + Controller->V1.StartDeviceStateScan = true; + Controller->V1.NeedBackgroundInitializationStatus = + Controller->V1.BackgroundInitializationStatusSupported; + Controller->SecondaryMonitoringTime = jiffies; + } + if (NewEnquiry->RebuildFlag == DAC960_V1_StandbyRebuildInProgress || + NewEnquiry->RebuildFlag + == DAC960_V1_BackgroundRebuildInProgress || + OldEnquiry->RebuildFlag == DAC960_V1_StandbyRebuildInProgress || + OldEnquiry->RebuildFlag == DAC960_V1_BackgroundRebuildInProgress) + { + Controller->V1.NeedRebuildProgress = true; + Controller->V1.RebuildProgressFirst = + (NewEnquiry->CriticalLogicalDriveCount < + OldEnquiry->CriticalLogicalDriveCount); + } + if (OldEnquiry->RebuildFlag == DAC960_V1_BackgroundCheckInProgress) + switch (NewEnquiry->RebuildFlag) + { + case DAC960_V1_NoStandbyRebuildOrCheckInProgress: + DAC960_Progress("Consistency Check Completed Successfully\n", + Controller); + break; + case DAC960_V1_StandbyRebuildInProgress: + case DAC960_V1_BackgroundRebuildInProgress: + break; + case DAC960_V1_BackgroundCheckInProgress: + Controller->V1.NeedConsistencyCheckProgress = true; + break; + case DAC960_V1_StandbyRebuildCompletedWithError: + DAC960_Progress("Consistency Check Completed with Error\n", + Controller); + break; + case DAC960_V1_BackgroundRebuildOrCheckFailed_DriveFailed: + DAC960_Progress("Consistency Check Failed - " + "Physical Device Failed\n", Controller); + break; + case DAC960_V1_BackgroundRebuildOrCheckFailed_LogicalDriveFailed: + DAC960_Progress("Consistency Check Failed - " + "Logical Drive Failed\n", Controller); + break; + case DAC960_V1_BackgroundRebuildOrCheckFailed_OtherCauses: + DAC960_Progress("Consistency Check Failed - Other Causes\n", + Controller); + break; + case DAC960_V1_BackgroundRebuildOrCheckSuccessfullyTerminated: + DAC960_Progress("Consistency Check Successfully Terminated\n", + Controller); + break; + } + else if (NewEnquiry->RebuildFlag + == DAC960_V1_BackgroundCheckInProgress) + Controller->V1.NeedConsistencyCheckProgress = true; + Controller->MonitoringAlertMode = + (NewEnquiry->CriticalLogicalDriveCount > 0 || + NewEnquiry->OfflineLogicalDriveCount > 0 || + NewEnquiry->DeadDriveCount > 0); + if (NewEnquiry->RebuildFlag > DAC960_V1_BackgroundCheckInProgress) + { + Controller->V1.PendingRebuildFlag = NewEnquiry->RebuildFlag; + Controller->V1.RebuildFlagPending = true; + } + memcpy(&Controller->V1.Enquiry, &Controller->V1.NewEnquiry, + sizeof(DAC960_V1_Enquiry_T)); + } + else if (CommandOpcode == DAC960_V1_PerformEventLogOperation) + { + static char + *DAC960_EventMessages[] = + { "killed because write recovery failed", + "killed because of SCSI bus reset failure", + "killed because of double check condition", + "killed because it was removed", + "killed because of gross error on SCSI chip", + "killed because of bad tag returned from drive", + "killed because of timeout on SCSI command", + "killed because of reset SCSI command issued from system", + "killed because busy or parity error count exceeded limit", + "killed because of 'kill drive' command from system", + "killed because of selection timeout", + "killed due to SCSI phase sequence error", + "killed due to unknown status" }; + DAC960_V1_EventLogEntry_T *EventLogEntry = + Controller->V1.EventLogEntry; + if (EventLogEntry->SequenceNumber == + Controller->V1.OldEventLogSequenceNumber) + { + unsigned char SenseKey = EventLogEntry->SenseKey; + unsigned char AdditionalSenseCode = + EventLogEntry->AdditionalSenseCode; + unsigned char AdditionalSenseCodeQualifier = + EventLogEntry->AdditionalSenseCodeQualifier; + if (SenseKey == DAC960_SenseKey_VendorSpecific && + AdditionalSenseCode == 0x80 && + AdditionalSenseCodeQualifier < + sizeof(DAC960_EventMessages) / sizeof(char *)) + DAC960_Critical("Physical Device %d:%d %s\n", Controller, + EventLogEntry->Channel, + EventLogEntry->TargetID, + DAC960_EventMessages[ + AdditionalSenseCodeQualifier]); + else if (SenseKey == DAC960_SenseKey_UnitAttention && + AdditionalSenseCode == 0x29) + { + if (Controller->MonitoringTimerCount > 0) + Controller->V1.DeviceResetCount[EventLogEntry->Channel] + [EventLogEntry->TargetID]++; + } + else if (!(SenseKey == DAC960_SenseKey_NoSense || + (SenseKey == DAC960_SenseKey_NotReady && + AdditionalSenseCode == 0x04 && + (AdditionalSenseCodeQualifier == 0x01 || + AdditionalSenseCodeQualifier == 0x02)))) + { + DAC960_Critical("Physical Device %d:%d Error Log: " + "Sense Key = %X, ASC = %02X, ASCQ = %02X\n", + Controller, + EventLogEntry->Channel, + EventLogEntry->TargetID, + SenseKey, + AdditionalSenseCode, + AdditionalSenseCodeQualifier); + DAC960_Critical("Physical Device %d:%d Error Log: " + "Information = %02X%02X%02X%02X " + "%02X%02X%02X%02X\n", + Controller, + EventLogEntry->Channel, + EventLogEntry->TargetID, + EventLogEntry->Information[0], + EventLogEntry->Information[1], + EventLogEntry->Information[2], + EventLogEntry->Information[3], + EventLogEntry->CommandSpecificInformation[0], + EventLogEntry->CommandSpecificInformation[1], + EventLogEntry->CommandSpecificInformation[2], + EventLogEntry->CommandSpecificInformation[3]); + } + } + Controller->V1.OldEventLogSequenceNumber++; + } + else if (CommandOpcode == DAC960_V1_GetErrorTable) + { + DAC960_V1_ErrorTable_T *OldErrorTable = &Controller->V1.ErrorTable; + DAC960_V1_ErrorTable_T *NewErrorTable = Controller->V1.NewErrorTable; + int Channel, TargetID; + for (Channel = 0; Channel < Controller->Channels; Channel++) + for (TargetID = 0; TargetID < Controller->Targets; TargetID++) + { + DAC960_V1_ErrorTableEntry_T *NewErrorEntry = + &NewErrorTable->ErrorTableEntries[Channel][TargetID]; + DAC960_V1_ErrorTableEntry_T *OldErrorEntry = + &OldErrorTable->ErrorTableEntries[Channel][TargetID]; + if ((NewErrorEntry->ParityErrorCount != + OldErrorEntry->ParityErrorCount) || + (NewErrorEntry->SoftErrorCount != + OldErrorEntry->SoftErrorCount) || + (NewErrorEntry->HardErrorCount != + OldErrorEntry->HardErrorCount) || + (NewErrorEntry->MiscErrorCount != + OldErrorEntry->MiscErrorCount)) + DAC960_Critical("Physical Device %d:%d Errors: " + "Parity = %d, Soft = %d, " + "Hard = %d, Misc = %d\n", + Controller, Channel, TargetID, + NewErrorEntry->ParityErrorCount, + NewErrorEntry->SoftErrorCount, + NewErrorEntry->HardErrorCount, + NewErrorEntry->MiscErrorCount); + } + memcpy(&Controller->V1.ErrorTable, Controller->V1.NewErrorTable, + sizeof(DAC960_V1_ErrorTable_T)); + } + else if (CommandOpcode == DAC960_V1_GetDeviceState) + { + DAC960_V1_DeviceState_T *OldDeviceState = + &Controller->V1.DeviceState[Controller->V1.DeviceStateChannel] + [Controller->V1.DeviceStateTargetID]; + DAC960_V1_DeviceState_T *NewDeviceState = + Controller->V1.NewDeviceState; + if (NewDeviceState->DeviceState != OldDeviceState->DeviceState) + DAC960_Critical("Physical Device %d:%d is now %s\n", Controller, + Controller->V1.DeviceStateChannel, + Controller->V1.DeviceStateTargetID, + (NewDeviceState->DeviceState + == DAC960_V1_Device_Dead + ? "DEAD" + : NewDeviceState->DeviceState + == DAC960_V1_Device_WriteOnly + ? "WRITE-ONLY" + : NewDeviceState->DeviceState + == DAC960_V1_Device_Online + ? "ONLINE" : "STANDBY")); + if (OldDeviceState->DeviceState == DAC960_V1_Device_Dead && + NewDeviceState->DeviceState != DAC960_V1_Device_Dead) + { + Controller->V1.NeedDeviceInquiryInformation = true; + Controller->V1.NeedDeviceSerialNumberInformation = true; + Controller->V1.DeviceResetCount + [Controller->V1.DeviceStateChannel] + [Controller->V1.DeviceStateTargetID] = 0; + } + memcpy(OldDeviceState, NewDeviceState, + sizeof(DAC960_V1_DeviceState_T)); + } + else if (CommandOpcode == DAC960_V1_GetLogicalDriveInformation) + { + int LogicalDriveNumber; + for (LogicalDriveNumber = 0; + LogicalDriveNumber < Controller->LogicalDriveCount; + LogicalDriveNumber++) + { + DAC960_V1_LogicalDriveInformation_T *OldLogicalDriveInformation = + &Controller->V1.LogicalDriveInformation[LogicalDriveNumber]; + DAC960_V1_LogicalDriveInformation_T *NewLogicalDriveInformation = + &(*Controller->V1.NewLogicalDriveInformation)[LogicalDriveNumber]; + if (NewLogicalDriveInformation->LogicalDriveState != + OldLogicalDriveInformation->LogicalDriveState) + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) " + "is now %s\n", Controller, + LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, + (NewLogicalDriveInformation->LogicalDriveState + == DAC960_V1_LogicalDrive_Online + ? "ONLINE" + : NewLogicalDriveInformation->LogicalDriveState + == DAC960_V1_LogicalDrive_Critical + ? "CRITICAL" : "OFFLINE")); + if (NewLogicalDriveInformation->WriteBack != + OldLogicalDriveInformation->WriteBack) + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) " + "is now %s\n", Controller, + LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, + (NewLogicalDriveInformation->WriteBack + ? "WRITE BACK" : "WRITE THRU")); + } + memcpy(&Controller->V1.LogicalDriveInformation, + Controller->V1.NewLogicalDriveInformation, + sizeof(DAC960_V1_LogicalDriveInformationArray_T)); + } + else if (CommandOpcode == DAC960_V1_GetRebuildProgress) + { + unsigned int LogicalDriveNumber = + Controller->V1.RebuildProgress->LogicalDriveNumber; + unsigned int LogicalDriveSize = + Controller->V1.RebuildProgress->LogicalDriveSize; + unsigned int BlocksCompleted = + LogicalDriveSize - Controller->V1.RebuildProgress->RemainingBlocks; + if (CommandStatus == DAC960_V1_NoRebuildOrCheckInProgress && + Controller->V1.LastRebuildStatus == DAC960_V1_NormalCompletion) + CommandStatus = DAC960_V1_RebuildSuccessful; + switch (CommandStatus) + { + case DAC960_V1_NormalCompletion: + Controller->EphemeralProgressMessage = true; + DAC960_Progress("Rebuild in Progress: " + "Logical Drive %d (/dev/rd/c%dd%d) " + "%d%% completed\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, + (100 * (BlocksCompleted >> 7)) + / (LogicalDriveSize >> 7)); + Controller->EphemeralProgressMessage = false; + break; + case DAC960_V1_RebuildFailed_LogicalDriveFailure: + DAC960_Progress("Rebuild Failed due to " + "Logical Drive Failure\n", Controller); + break; + case DAC960_V1_RebuildFailed_BadBlocksOnOther: + DAC960_Progress("Rebuild Failed due to " + "Bad Blocks on Other Drives\n", Controller); + break; + case DAC960_V1_RebuildFailed_NewDriveFailed: + DAC960_Progress("Rebuild Failed due to " + "Failure of Drive Being Rebuilt\n", Controller); + break; + case DAC960_V1_NoRebuildOrCheckInProgress: + break; + case DAC960_V1_RebuildSuccessful: + DAC960_Progress("Rebuild Completed Successfully\n", Controller); + break; + case DAC960_V1_RebuildSuccessfullyTerminated: + DAC960_Progress("Rebuild Successfully Terminated\n", Controller); + break; + } + Controller->V1.LastRebuildStatus = CommandStatus; + if (CommandType != DAC960_MonitoringCommand && + Controller->V1.RebuildStatusPending) + { + Command->V1.CommandStatus = Controller->V1.PendingRebuildStatus; + Controller->V1.RebuildStatusPending = false; + } + else if (CommandType == DAC960_MonitoringCommand && + CommandStatus != DAC960_V1_NormalCompletion && + CommandStatus != DAC960_V1_NoRebuildOrCheckInProgress) + { + Controller->V1.PendingRebuildStatus = CommandStatus; + Controller->V1.RebuildStatusPending = true; + } + } + else if (CommandOpcode == DAC960_V1_RebuildStat) + { + unsigned int LogicalDriveNumber = + Controller->V1.RebuildProgress->LogicalDriveNumber; + unsigned int LogicalDriveSize = + Controller->V1.RebuildProgress->LogicalDriveSize; + unsigned int BlocksCompleted = + LogicalDriveSize - Controller->V1.RebuildProgress->RemainingBlocks; + if (CommandStatus == DAC960_V1_NormalCompletion) + { + Controller->EphemeralProgressMessage = true; + DAC960_Progress("Consistency Check in Progress: " + "Logical Drive %d (/dev/rd/c%dd%d) " + "%d%% completed\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, + (100 * (BlocksCompleted >> 7)) + / (LogicalDriveSize >> 7)); + Controller->EphemeralProgressMessage = false; + } + } + else if (CommandOpcode == DAC960_V1_BackgroundInitializationControl) + { + unsigned int LogicalDriveNumber = + Controller->V1.BackgroundInitializationStatus->LogicalDriveNumber; + unsigned int LogicalDriveSize = + Controller->V1.BackgroundInitializationStatus->LogicalDriveSize; + unsigned int BlocksCompleted = + Controller->V1.BackgroundInitializationStatus->BlocksCompleted; + switch (CommandStatus) + { + case DAC960_V1_NormalCompletion: + switch (Controller->V1.BackgroundInitializationStatus->Status) + { + case DAC960_V1_BackgroundInitializationInvalid: + break; + case DAC960_V1_BackgroundInitializationStarted: + DAC960_Progress("Background Initialization Started\n", + Controller); + break; + case DAC960_V1_BackgroundInitializationInProgress: + if (BlocksCompleted == + Controller->V1.LastBackgroundInitializationStatus. + BlocksCompleted && + LogicalDriveNumber == + Controller->V1.LastBackgroundInitializationStatus. + LogicalDriveNumber) + break; + Controller->EphemeralProgressMessage = true; + DAC960_Progress("Background Initialization in Progress: " + "Logical Drive %d (/dev/rd/c%dd%d) " + "%d%% completed\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, + (100 * (BlocksCompleted >> 7)) + / (LogicalDriveSize >> 7)); + Controller->EphemeralProgressMessage = false; + break; + case DAC960_V1_BackgroundInitializationSuspended: + DAC960_Progress("Background Initialization Suspended\n", + Controller); + break; + case DAC960_V1_BackgroundInitializationCancelled: + DAC960_Progress("Background Initialization Cancelled\n", + Controller); + break; + } + memcpy(&Controller->V1.LastBackgroundInitializationStatus, + Controller->V1.BackgroundInitializationStatus, + sizeof(DAC960_V1_BackgroundInitializationStatus_T)); + break; + case DAC960_V1_BackgroundInitSuccessful: + if (Controller->V1.BackgroundInitializationStatus->Status == + DAC960_V1_BackgroundInitializationInProgress) + DAC960_Progress("Background Initialization " + "Completed Successfully\n", Controller); + Controller->V1.BackgroundInitializationStatus->Status = + DAC960_V1_BackgroundInitializationInvalid; + break; + case DAC960_V1_BackgroundInitAborted: + if (Controller->V1.BackgroundInitializationStatus->Status == + DAC960_V1_BackgroundInitializationInProgress) + DAC960_Progress("Background Initialization Aborted\n", + Controller); + Controller->V1.BackgroundInitializationStatus->Status = + DAC960_V1_BackgroundInitializationInvalid; + break; + case DAC960_V1_NoBackgroundInitInProgress: + break; + } + } + else if (CommandOpcode == DAC960_V1_DCDB) + { + /* + This is a bit ugly. + + The InquiryStandardData and + the InquiryUntitSerialNumber information + retrieval operations BOTH use the DAC960_V1_DCDB + commands. the test above can't distinguish between + these two cases. + + Instead, we rely on the order of code later in this + function to ensure that DeviceInquiryInformation commands + are submitted before DeviceSerialNumber commands. + */ + if (Controller->V1.NeedDeviceInquiryInformation) + { + DAC960_SCSI_Inquiry_T *InquiryStandardData = + &Controller->V1.InquiryStandardData + [Controller->V1.DeviceStateChannel] + [Controller->V1.DeviceStateTargetID]; + if (CommandStatus != DAC960_V1_NormalCompletion) + { + memset(InquiryStandardData, 0, + sizeof(DAC960_SCSI_Inquiry_T)); + InquiryStandardData->PeripheralDeviceType = 0x1F; + } + else + memcpy(InquiryStandardData, + Controller->V1.NewInquiryStandardData, + sizeof(DAC960_SCSI_Inquiry_T)); + Controller->V1.NeedDeviceInquiryInformation = false; + } + else if (Controller->V1.NeedDeviceSerialNumberInformation) + { + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + &Controller->V1.InquiryUnitSerialNumber + [Controller->V1.DeviceStateChannel] + [Controller->V1.DeviceStateTargetID]; + if (CommandStatus != DAC960_V1_NormalCompletion) + { + memset(InquiryUnitSerialNumber, 0, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + InquiryUnitSerialNumber->PeripheralDeviceType = 0x1F; + } + else + memcpy(InquiryUnitSerialNumber, + Controller->V1.NewInquiryUnitSerialNumber, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + Controller->V1.NeedDeviceSerialNumberInformation = false; + } + } + /* + Begin submitting new monitoring commands. + */ + if (Controller->V1.NewEventLogSequenceNumber + - Controller->V1.OldEventLogSequenceNumber > 0) + { + Command->V1.CommandMailbox.Type3E.CommandOpcode = + DAC960_V1_PerformEventLogOperation; + Command->V1.CommandMailbox.Type3E.OperationType = + DAC960_V1_GetEventLogEntry; + Command->V1.CommandMailbox.Type3E.OperationQualifier = 1; + Command->V1.CommandMailbox.Type3E.SequenceNumber = + Controller->V1.OldEventLogSequenceNumber; + Command->V1.CommandMailbox.Type3E.BusAddress = + Controller->V1.EventLogEntryDMA; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.NeedErrorTableInformation) + { + Controller->V1.NeedErrorTableInformation = false; + Command->V1.CommandMailbox.Type3.CommandOpcode = + DAC960_V1_GetErrorTable; + Command->V1.CommandMailbox.Type3.BusAddress = + Controller->V1.NewErrorTableDMA; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.NeedRebuildProgress && + Controller->V1.RebuildProgressFirst) + { + Controller->V1.NeedRebuildProgress = false; + Command->V1.CommandMailbox.Type3.CommandOpcode = + DAC960_V1_GetRebuildProgress; + Command->V1.CommandMailbox.Type3.BusAddress = + Controller->V1.RebuildProgressDMA; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.NeedDeviceStateInformation) + { + if (Controller->V1.NeedDeviceInquiryInformation) + { + DAC960_V1_DCDB_T *DCDB = Controller->V1.MonitoringDCDB; + dma_addr_t DCDB_DMA = Controller->V1.MonitoringDCDB_DMA; + + dma_addr_t NewInquiryStandardDataDMA = + Controller->V1.NewInquiryStandardDataDMA; + + Command->V1.CommandMailbox.Type3.CommandOpcode = DAC960_V1_DCDB; + Command->V1.CommandMailbox.Type3.BusAddress = DCDB_DMA; + DCDB->Channel = Controller->V1.DeviceStateChannel; + DCDB->TargetID = Controller->V1.DeviceStateTargetID; + DCDB->Direction = DAC960_V1_DCDB_DataTransferDeviceToSystem; + DCDB->EarlyStatus = false; + DCDB->Timeout = DAC960_V1_DCDB_Timeout_10_seconds; + DCDB->NoAutomaticRequestSense = false; + DCDB->DisconnectPermitted = true; + DCDB->TransferLength = sizeof(DAC960_SCSI_Inquiry_T); + DCDB->BusAddress = NewInquiryStandardDataDMA; + DCDB->CDBLength = 6; + DCDB->TransferLengthHigh4 = 0; + DCDB->SenseLength = sizeof(DCDB->SenseData); + DCDB->CDB[0] = 0x12; /* INQUIRY */ + DCDB->CDB[1] = 0; /* EVPD = 0 */ + DCDB->CDB[2] = 0; /* Page Code */ + DCDB->CDB[3] = 0; /* Reserved */ + DCDB->CDB[4] = sizeof(DAC960_SCSI_Inquiry_T); + DCDB->CDB[5] = 0; /* Control */ + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.NeedDeviceSerialNumberInformation) + { + DAC960_V1_DCDB_T *DCDB = Controller->V1.MonitoringDCDB; + dma_addr_t DCDB_DMA = Controller->V1.MonitoringDCDB_DMA; + dma_addr_t NewInquiryUnitSerialNumberDMA = + Controller->V1.NewInquiryUnitSerialNumberDMA; + + Command->V1.CommandMailbox.Type3.CommandOpcode = DAC960_V1_DCDB; + Command->V1.CommandMailbox.Type3.BusAddress = DCDB_DMA; + DCDB->Channel = Controller->V1.DeviceStateChannel; + DCDB->TargetID = Controller->V1.DeviceStateTargetID; + DCDB->Direction = DAC960_V1_DCDB_DataTransferDeviceToSystem; + DCDB->EarlyStatus = false; + DCDB->Timeout = DAC960_V1_DCDB_Timeout_10_seconds; + DCDB->NoAutomaticRequestSense = false; + DCDB->DisconnectPermitted = true; + DCDB->TransferLength = + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T); + DCDB->BusAddress = NewInquiryUnitSerialNumberDMA; + DCDB->CDBLength = 6; + DCDB->TransferLengthHigh4 = 0; + DCDB->SenseLength = sizeof(DCDB->SenseData); + DCDB->CDB[0] = 0x12; /* INQUIRY */ + DCDB->CDB[1] = 1; /* EVPD = 1 */ + DCDB->CDB[2] = 0x80; /* Page Code */ + DCDB->CDB[3] = 0; /* Reserved */ + DCDB->CDB[4] = sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T); + DCDB->CDB[5] = 0; /* Control */ + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.StartDeviceStateScan) + { + Controller->V1.DeviceStateChannel = 0; + Controller->V1.DeviceStateTargetID = 0; + Controller->V1.StartDeviceStateScan = false; + } + else if (++Controller->V1.DeviceStateTargetID == Controller->Targets) + { + Controller->V1.DeviceStateChannel++; + Controller->V1.DeviceStateTargetID = 0; + } + if (Controller->V1.DeviceStateChannel < Controller->Channels) + { + Controller->V1.NewDeviceState->DeviceState = + DAC960_V1_Device_Dead; + Command->V1.CommandMailbox.Type3D.CommandOpcode = + DAC960_V1_GetDeviceState; + Command->V1.CommandMailbox.Type3D.Channel = + Controller->V1.DeviceStateChannel; + Command->V1.CommandMailbox.Type3D.TargetID = + Controller->V1.DeviceStateTargetID; + Command->V1.CommandMailbox.Type3D.BusAddress = + Controller->V1.NewDeviceStateDMA; + DAC960_QueueCommand(Command); + return; + } + Controller->V1.NeedDeviceStateInformation = false; + } + if (Controller->V1.NeedLogicalDriveInformation) + { + Controller->V1.NeedLogicalDriveInformation = false; + Command->V1.CommandMailbox.Type3.CommandOpcode = + DAC960_V1_GetLogicalDriveInformation; + Command->V1.CommandMailbox.Type3.BusAddress = + Controller->V1.NewLogicalDriveInformationDMA; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.NeedRebuildProgress) + { + Controller->V1.NeedRebuildProgress = false; + Command->V1.CommandMailbox.Type3.CommandOpcode = + DAC960_V1_GetRebuildProgress; + Command->V1.CommandMailbox.Type3.BusAddress = + Controller->V1.RebuildProgressDMA; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.NeedConsistencyCheckProgress) + { + Controller->V1.NeedConsistencyCheckProgress = false; + Command->V1.CommandMailbox.Type3.CommandOpcode = + DAC960_V1_RebuildStat; + Command->V1.CommandMailbox.Type3.BusAddress = + Controller->V1.RebuildProgressDMA; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V1.NeedBackgroundInitializationStatus) + { + Controller->V1.NeedBackgroundInitializationStatus = false; + Command->V1.CommandMailbox.Type3B.CommandOpcode = + DAC960_V1_BackgroundInitializationControl; + Command->V1.CommandMailbox.Type3B.CommandOpcode2 = 0x20; + Command->V1.CommandMailbox.Type3B.BusAddress = + Controller->V1.BackgroundInitializationStatusDMA; + DAC960_QueueCommand(Command); + return; + } + Controller->MonitoringTimerCount++; + Controller->MonitoringTimer.expires = + jiffies + DAC960_MonitoringTimerInterval; + add_timer(&Controller->MonitoringTimer); + } + if (CommandType == DAC960_ImmediateCommand) + { + complete(Command->Completion); + Command->Completion = NULL; + return; + } + if (CommandType == DAC960_QueuedCommand) + { + DAC960_V1_KernelCommand_T *KernelCommand = Command->V1.KernelCommand; + KernelCommand->CommandStatus = Command->V1.CommandStatus; + Command->V1.KernelCommand = NULL; + if (CommandOpcode == DAC960_V1_DCDB) + Controller->V1.DirectCommandActive[KernelCommand->DCDB->Channel] + [KernelCommand->DCDB->TargetID] = + false; + DAC960_DeallocateCommand(Command); + KernelCommand->CompletionFunction(KernelCommand); + return; + } + /* + Queue a Status Monitoring Command to the Controller using the just + completed Command if one was deferred previously due to lack of a + free Command when the Monitoring Timer Function was called. + */ + if (Controller->MonitoringCommandDeferred) + { + Controller->MonitoringCommandDeferred = false; + DAC960_V1_QueueMonitoringCommand(Command); + return; + } + /* + Deallocate the Command. + */ + DAC960_DeallocateCommand(Command); + /* + Wake up any processes waiting on a free Command. + */ + wake_up(&Controller->CommandWaitQueue); +} + + +/* + DAC960_V2_ReadWriteError prints an appropriate error message for Command + when an error occurs on a Read or Write operation. +*/ + +static void DAC960_V2_ReadWriteError(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + unsigned char *SenseErrors[] = { "NO SENSE", "RECOVERED ERROR", + "NOT READY", "MEDIUM ERROR", + "HARDWARE ERROR", "ILLEGAL REQUEST", + "UNIT ATTENTION", "DATA PROTECT", + "BLANK CHECK", "VENDOR-SPECIFIC", + "COPY ABORTED", "ABORTED COMMAND", + "EQUAL", "VOLUME OVERFLOW", + "MISCOMPARE", "RESERVED" }; + unsigned char *CommandName = "UNKNOWN"; + switch (Command->CommandType) + { + case DAC960_ReadCommand: + case DAC960_ReadRetryCommand: + CommandName = "READ"; + break; + case DAC960_WriteCommand: + case DAC960_WriteRetryCommand: + CommandName = "WRITE"; + break; + case DAC960_MonitoringCommand: + case DAC960_ImmediateCommand: + case DAC960_QueuedCommand: + break; + } + DAC960_Error("Error Condition %s on %s:\n", Controller, + SenseErrors[Command->V2.RequestSense->SenseKey], CommandName); + DAC960_Error(" /dev/rd/c%dd%d: absolute blocks %u..%u\n", + Controller, Controller->ControllerNumber, + Command->LogicalDriveNumber, Command->BlockNumber, + Command->BlockNumber + Command->BlockCount - 1); +} + + +/* + DAC960_V2_ReportEvent prints an appropriate message when a Controller Event + occurs. +*/ + +static void DAC960_V2_ReportEvent(DAC960_Controller_T *Controller, + DAC960_V2_Event_T *Event) +{ + DAC960_SCSI_RequestSense_T *RequestSense = + (DAC960_SCSI_RequestSense_T *) &Event->RequestSenseData; + unsigned char MessageBuffer[DAC960_LineBufferSize]; + static struct { int EventCode; unsigned char *EventMessage; } EventList[] = + { /* Physical Device Events (0x0000 - 0x007F) */ + { 0x0001, "P Online" }, + { 0x0002, "P Standby" }, + { 0x0005, "P Automatic Rebuild Started" }, + { 0x0006, "P Manual Rebuild Started" }, + { 0x0007, "P Rebuild Completed" }, + { 0x0008, "P Rebuild Cancelled" }, + { 0x0009, "P Rebuild Failed for Unknown Reasons" }, + { 0x000A, "P Rebuild Failed due to New Physical Device" }, + { 0x000B, "P Rebuild Failed due to Logical Drive Failure" }, + { 0x000C, "S Offline" }, + { 0x000D, "P Found" }, + { 0x000E, "P Removed" }, + { 0x000F, "P Unconfigured" }, + { 0x0010, "P Expand Capacity Started" }, + { 0x0011, "P Expand Capacity Completed" }, + { 0x0012, "P Expand Capacity Failed" }, + { 0x0013, "P Command Timed Out" }, + { 0x0014, "P Command Aborted" }, + { 0x0015, "P Command Retried" }, + { 0x0016, "P Parity Error" }, + { 0x0017, "P Soft Error" }, + { 0x0018, "P Miscellaneous Error" }, + { 0x0019, "P Reset" }, + { 0x001A, "P Active Spare Found" }, + { 0x001B, "P Warm Spare Found" }, + { 0x001C, "S Sense Data Received" }, + { 0x001D, "P Initialization Started" }, + { 0x001E, "P Initialization Completed" }, + { 0x001F, "P Initialization Failed" }, + { 0x0020, "P Initialization Cancelled" }, + { 0x0021, "P Failed because Write Recovery Failed" }, + { 0x0022, "P Failed because SCSI Bus Reset Failed" }, + { 0x0023, "P Failed because of Double Check Condition" }, + { 0x0024, "P Failed because Device Cannot Be Accessed" }, + { 0x0025, "P Failed because of Gross Error on SCSI Processor" }, + { 0x0026, "P Failed because of Bad Tag from Device" }, + { 0x0027, "P Failed because of Command Timeout" }, + { 0x0028, "P Failed because of System Reset" }, + { 0x0029, "P Failed because of Busy Status or Parity Error" }, + { 0x002A, "P Failed because Host Set Device to Failed State" }, + { 0x002B, "P Failed because of Selection Timeout" }, + { 0x002C, "P Failed because of SCSI Bus Phase Error" }, + { 0x002D, "P Failed because Device Returned Unknown Status" }, + { 0x002E, "P Failed because Device Not Ready" }, + { 0x002F, "P Failed because Device Not Found at Startup" }, + { 0x0030, "P Failed because COD Write Operation Failed" }, + { 0x0031, "P Failed because BDT Write Operation Failed" }, + { 0x0039, "P Missing at Startup" }, + { 0x003A, "P Start Rebuild Failed due to Physical Drive Too Small" }, + { 0x003C, "P Temporarily Offline Device Automatically Made Online" }, + { 0x003D, "P Standby Rebuild Started" }, + /* Logical Device Events (0x0080 - 0x00FF) */ + { 0x0080, "M Consistency Check Started" }, + { 0x0081, "M Consistency Check Completed" }, + { 0x0082, "M Consistency Check Cancelled" }, + { 0x0083, "M Consistency Check Completed With Errors" }, + { 0x0084, "M Consistency Check Failed due to Logical Drive Failure" }, + { 0x0085, "M Consistency Check Failed due to Physical Device Failure" }, + { 0x0086, "L Offline" }, + { 0x0087, "L Critical" }, + { 0x0088, "L Online" }, + { 0x0089, "M Automatic Rebuild Started" }, + { 0x008A, "M Manual Rebuild Started" }, + { 0x008B, "M Rebuild Completed" }, + { 0x008C, "M Rebuild Cancelled" }, + { 0x008D, "M Rebuild Failed for Unknown Reasons" }, + { 0x008E, "M Rebuild Failed due to New Physical Device" }, + { 0x008F, "M Rebuild Failed due to Logical Drive Failure" }, + { 0x0090, "M Initialization Started" }, + { 0x0091, "M Initialization Completed" }, + { 0x0092, "M Initialization Cancelled" }, + { 0x0093, "M Initialization Failed" }, + { 0x0094, "L Found" }, + { 0x0095, "L Deleted" }, + { 0x0096, "M Expand Capacity Started" }, + { 0x0097, "M Expand Capacity Completed" }, + { 0x0098, "M Expand Capacity Failed" }, + { 0x0099, "L Bad Block Found" }, + { 0x009A, "L Size Changed" }, + { 0x009B, "L Type Changed" }, + { 0x009C, "L Bad Data Block Found" }, + { 0x009E, "L Read of Data Block in BDT" }, + { 0x009F, "L Write Back Data for Disk Block Lost" }, + { 0x00A0, "L Temporarily Offline RAID-5/3 Drive Made Online" }, + { 0x00A1, "L Temporarily Offline RAID-6/1/0/7 Drive Made Online" }, + { 0x00A2, "L Standby Rebuild Started" }, + /* Fault Management Events (0x0100 - 0x017F) */ + { 0x0140, "E Fan %d Failed" }, + { 0x0141, "E Fan %d OK" }, + { 0x0142, "E Fan %d Not Present" }, + { 0x0143, "E Power Supply %d Failed" }, + { 0x0144, "E Power Supply %d OK" }, + { 0x0145, "E Power Supply %d Not Present" }, + { 0x0146, "E Temperature Sensor %d Temperature Exceeds Safe Limit" }, + { 0x0147, "E Temperature Sensor %d Temperature Exceeds Working Limit" }, + { 0x0148, "E Temperature Sensor %d Temperature Normal" }, + { 0x0149, "E Temperature Sensor %d Not Present" }, + { 0x014A, "E Enclosure Management Unit %d Access Critical" }, + { 0x014B, "E Enclosure Management Unit %d Access OK" }, + { 0x014C, "E Enclosure Management Unit %d Access Offline" }, + /* Controller Events (0x0180 - 0x01FF) */ + { 0x0181, "C Cache Write Back Error" }, + { 0x0188, "C Battery Backup Unit Found" }, + { 0x0189, "C Battery Backup Unit Charge Level Low" }, + { 0x018A, "C Battery Backup Unit Charge Level OK" }, + { 0x0193, "C Installation Aborted" }, + { 0x0195, "C Battery Backup Unit Physically Removed" }, + { 0x0196, "C Memory Error During Warm Boot" }, + { 0x019E, "C Memory Soft ECC Error Corrected" }, + { 0x019F, "C Memory Hard ECC Error Corrected" }, + { 0x01A2, "C Battery Backup Unit Failed" }, + { 0x01AB, "C Mirror Race Recovery Failed" }, + { 0x01AC, "C Mirror Race on Critical Drive" }, + /* Controller Internal Processor Events */ + { 0x0380, "C Internal Controller Hung" }, + { 0x0381, "C Internal Controller Firmware Breakpoint" }, + { 0x0390, "C Internal Controller i960 Processor Specific Error" }, + { 0x03A0, "C Internal Controller StrongARM Processor Specific Error" }, + { 0, "" } }; + int EventListIndex = 0, EventCode; + unsigned char EventType, *EventMessage; + if (Event->EventCode == 0x1C && + RequestSense->SenseKey == DAC960_SenseKey_VendorSpecific && + (RequestSense->AdditionalSenseCode == 0x80 || + RequestSense->AdditionalSenseCode == 0x81)) + Event->EventCode = ((RequestSense->AdditionalSenseCode - 0x80) << 8) | + RequestSense->AdditionalSenseCodeQualifier; + while (true) + { + EventCode = EventList[EventListIndex].EventCode; + if (EventCode == Event->EventCode || EventCode == 0) break; + EventListIndex++; + } + EventType = EventList[EventListIndex].EventMessage[0]; + EventMessage = &EventList[EventListIndex].EventMessage[2]; + if (EventCode == 0) + { + DAC960_Critical("Unknown Controller Event Code %04X\n", + Controller, Event->EventCode); + return; + } + switch (EventType) + { + case 'P': + DAC960_Critical("Physical Device %d:%d %s\n", Controller, + Event->Channel, Event->TargetID, EventMessage); + break; + case 'L': + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) %s\n", Controller, + Event->LogicalUnit, Controller->ControllerNumber, + Event->LogicalUnit, EventMessage); + break; + case 'M': + DAC960_Progress("Logical Drive %d (/dev/rd/c%dd%d) %s\n", Controller, + Event->LogicalUnit, Controller->ControllerNumber, + Event->LogicalUnit, EventMessage); + break; + case 'S': + if (RequestSense->SenseKey == DAC960_SenseKey_NoSense || + (RequestSense->SenseKey == DAC960_SenseKey_NotReady && + RequestSense->AdditionalSenseCode == 0x04 && + (RequestSense->AdditionalSenseCodeQualifier == 0x01 || + RequestSense->AdditionalSenseCodeQualifier == 0x02))) + break; + DAC960_Critical("Physical Device %d:%d %s\n", Controller, + Event->Channel, Event->TargetID, EventMessage); + DAC960_Critical("Physical Device %d:%d Request Sense: " + "Sense Key = %X, ASC = %02X, ASCQ = %02X\n", + Controller, + Event->Channel, + Event->TargetID, + RequestSense->SenseKey, + RequestSense->AdditionalSenseCode, + RequestSense->AdditionalSenseCodeQualifier); + DAC960_Critical("Physical Device %d:%d Request Sense: " + "Information = %02X%02X%02X%02X " + "%02X%02X%02X%02X\n", + Controller, + Event->Channel, + Event->TargetID, + RequestSense->Information[0], + RequestSense->Information[1], + RequestSense->Information[2], + RequestSense->Information[3], + RequestSense->CommandSpecificInformation[0], + RequestSense->CommandSpecificInformation[1], + RequestSense->CommandSpecificInformation[2], + RequestSense->CommandSpecificInformation[3]); + break; + case 'E': + if (Controller->SuppressEnclosureMessages) break; + sprintf(MessageBuffer, EventMessage, Event->LogicalUnit); + DAC960_Critical("Enclosure %d %s\n", Controller, + Event->TargetID, MessageBuffer); + break; + case 'C': + DAC960_Critical("Controller %s\n", Controller, EventMessage); + break; + default: + DAC960_Critical("Unknown Controller Event Code %04X\n", + Controller, Event->EventCode); + break; + } +} + + +/* + DAC960_V2_ReportProgress prints an appropriate progress message for + Logical Device Long Operations. +*/ + +static void DAC960_V2_ReportProgress(DAC960_Controller_T *Controller, + unsigned char *MessageString, + unsigned int LogicalDeviceNumber, + unsigned long BlocksCompleted, + unsigned long LogicalDeviceSize) +{ + Controller->EphemeralProgressMessage = true; + DAC960_Progress("%s in Progress: Logical Drive %d (/dev/rd/c%dd%d) " + "%d%% completed\n", Controller, + MessageString, + LogicalDeviceNumber, + Controller->ControllerNumber, + LogicalDeviceNumber, + (100 * (BlocksCompleted >> 7)) / (LogicalDeviceSize >> 7)); + Controller->EphemeralProgressMessage = false; +} + + +/* + DAC960_V2_ProcessCompletedCommand performs completion processing for Command + for DAC960 V2 Firmware Controllers. +*/ + +static void DAC960_V2_ProcessCompletedCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + DAC960_CommandType_T CommandType = Command->CommandType; + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_IOCTL_Opcode_T CommandOpcode = CommandMailbox->Common.IOCTL_Opcode; + DAC960_V2_CommandStatus_T CommandStatus = Command->V2.CommandStatus; + + if (CommandType == DAC960_ReadCommand || + CommandType == DAC960_WriteCommand) + { + +#ifdef FORCE_RETRY_DEBUG + CommandStatus = DAC960_V2_AbormalCompletion; +#endif + Command->V2.RequestSense->SenseKey = DAC960_SenseKey_MediumError; + + if (CommandStatus == DAC960_V2_NormalCompletion) { + + if (!DAC960_ProcessCompletedRequest(Command, true)) + BUG(); + + } else if (Command->V2.RequestSense->SenseKey == DAC960_SenseKey_MediumError) + { + /* + * break the command down into pieces and resubmit each + * piece, hoping that some of them will succeed. + */ + DAC960_queue_partial_rw(Command); + return; + } + else + { + if (Command->V2.RequestSense->SenseKey != DAC960_SenseKey_NotReady) + DAC960_V2_ReadWriteError(Command); + /* + Perform completion processing for all buffers in this I/O Request. + */ + (void)DAC960_ProcessCompletedRequest(Command, false); + } + } + else if (CommandType == DAC960_ReadRetryCommand || + CommandType == DAC960_WriteRetryCommand) + { + boolean normal_completion; + +#ifdef FORCE_RETRY_FAILURE_DEBUG + static int retry_count = 1; +#endif + /* + Perform completion processing for the portion that was + retried, and submit the next portion, if any. + */ + normal_completion = true; + if (CommandStatus != DAC960_V2_NormalCompletion) { + normal_completion = false; + if (Command->V2.RequestSense->SenseKey != DAC960_SenseKey_NotReady) + DAC960_V2_ReadWriteError(Command); + } + +#ifdef FORCE_RETRY_FAILURE_DEBUG + if (!(++retry_count % 10000)) { + printk("V2 error retry failure test\n"); + normal_completion = false; + DAC960_V2_ReadWriteError(Command); + } +#endif + + if (!DAC960_ProcessCompletedRequest(Command, normal_completion)) { + DAC960_queue_partial_rw(Command); + return; + } + } + else if (CommandType == DAC960_MonitoringCommand) + { + if (Controller->ShutdownMonitoringTimer) + return; + if (CommandOpcode == DAC960_V2_GetControllerInfo) + { + DAC960_V2_ControllerInfo_T *NewControllerInfo = + Controller->V2.NewControllerInformation; + DAC960_V2_ControllerInfo_T *ControllerInfo = + &Controller->V2.ControllerInformation; + Controller->LogicalDriveCount = + NewControllerInfo->LogicalDevicesPresent; + Controller->V2.NeedLogicalDeviceInformation = true; + Controller->V2.NeedPhysicalDeviceInformation = true; + Controller->V2.StartLogicalDeviceInformationScan = true; + Controller->V2.StartPhysicalDeviceInformationScan = true; + Controller->MonitoringAlertMode = + (NewControllerInfo->LogicalDevicesCritical > 0 || + NewControllerInfo->LogicalDevicesOffline > 0 || + NewControllerInfo->PhysicalDisksCritical > 0 || + NewControllerInfo->PhysicalDisksOffline > 0); + memcpy(ControllerInfo, NewControllerInfo, + sizeof(DAC960_V2_ControllerInfo_T)); + } + else if (CommandOpcode == DAC960_V2_GetEvent) + { + if (CommandStatus == DAC960_V2_NormalCompletion) { + DAC960_V2_ReportEvent(Controller, Controller->V2.Event); + } + Controller->V2.NextEventSequenceNumber++; + } + else if (CommandOpcode == DAC960_V2_GetPhysicalDeviceInfoValid && + CommandStatus == DAC960_V2_NormalCompletion) + { + DAC960_V2_PhysicalDeviceInfo_T *NewPhysicalDeviceInfo = + Controller->V2.NewPhysicalDeviceInformation; + unsigned int PhysicalDeviceIndex = Controller->V2.PhysicalDeviceIndex; + DAC960_V2_PhysicalDeviceInfo_T *PhysicalDeviceInfo = + Controller->V2.PhysicalDeviceInformation[PhysicalDeviceIndex]; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + Controller->V2.InquiryUnitSerialNumber[PhysicalDeviceIndex]; + unsigned int DeviceIndex; + while (PhysicalDeviceInfo != NULL && + (NewPhysicalDeviceInfo->Channel > + PhysicalDeviceInfo->Channel || + (NewPhysicalDeviceInfo->Channel == + PhysicalDeviceInfo->Channel && + (NewPhysicalDeviceInfo->TargetID > + PhysicalDeviceInfo->TargetID || + (NewPhysicalDeviceInfo->TargetID == + PhysicalDeviceInfo->TargetID && + NewPhysicalDeviceInfo->LogicalUnit > + PhysicalDeviceInfo->LogicalUnit))))) + { + DAC960_Critical("Physical Device %d:%d No Longer Exists\n", + Controller, + PhysicalDeviceInfo->Channel, + PhysicalDeviceInfo->TargetID); + Controller->V2.PhysicalDeviceInformation + [PhysicalDeviceIndex] = NULL; + Controller->V2.InquiryUnitSerialNumber + [PhysicalDeviceIndex] = NULL; + kfree(PhysicalDeviceInfo); + kfree(InquiryUnitSerialNumber); + for (DeviceIndex = PhysicalDeviceIndex; + DeviceIndex < DAC960_V2_MaxPhysicalDevices - 1; + DeviceIndex++) + { + Controller->V2.PhysicalDeviceInformation[DeviceIndex] = + Controller->V2.PhysicalDeviceInformation[DeviceIndex+1]; + Controller->V2.InquiryUnitSerialNumber[DeviceIndex] = + Controller->V2.InquiryUnitSerialNumber[DeviceIndex+1]; + } + Controller->V2.PhysicalDeviceInformation + [DAC960_V2_MaxPhysicalDevices-1] = NULL; + Controller->V2.InquiryUnitSerialNumber + [DAC960_V2_MaxPhysicalDevices-1] = NULL; + PhysicalDeviceInfo = + Controller->V2.PhysicalDeviceInformation[PhysicalDeviceIndex]; + InquiryUnitSerialNumber = + Controller->V2.InquiryUnitSerialNumber[PhysicalDeviceIndex]; + } + if (PhysicalDeviceInfo == NULL || + (NewPhysicalDeviceInfo->Channel != + PhysicalDeviceInfo->Channel) || + (NewPhysicalDeviceInfo->TargetID != + PhysicalDeviceInfo->TargetID) || + (NewPhysicalDeviceInfo->LogicalUnit != + PhysicalDeviceInfo->LogicalUnit)) + { + PhysicalDeviceInfo = (DAC960_V2_PhysicalDeviceInfo_T *) + kmalloc(sizeof(DAC960_V2_PhysicalDeviceInfo_T), GFP_ATOMIC); + InquiryUnitSerialNumber = + (DAC960_SCSI_Inquiry_UnitSerialNumber_T *) + kmalloc(sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T), + GFP_ATOMIC); + if (InquiryUnitSerialNumber == NULL && + PhysicalDeviceInfo != NULL) + { + kfree(PhysicalDeviceInfo); + PhysicalDeviceInfo = NULL; + } + DAC960_Critical("Physical Device %d:%d Now Exists%s\n", + Controller, + NewPhysicalDeviceInfo->Channel, + NewPhysicalDeviceInfo->TargetID, + (PhysicalDeviceInfo != NULL + ? "" : " - Allocation Failed")); + if (PhysicalDeviceInfo != NULL) + { + memset(PhysicalDeviceInfo, 0, + sizeof(DAC960_V2_PhysicalDeviceInfo_T)); + PhysicalDeviceInfo->PhysicalDeviceState = + DAC960_V2_Device_InvalidState; + memset(InquiryUnitSerialNumber, 0, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + InquiryUnitSerialNumber->PeripheralDeviceType = 0x1F; + for (DeviceIndex = DAC960_V2_MaxPhysicalDevices - 1; + DeviceIndex > PhysicalDeviceIndex; + DeviceIndex--) + { + Controller->V2.PhysicalDeviceInformation[DeviceIndex] = + Controller->V2.PhysicalDeviceInformation[DeviceIndex-1]; + Controller->V2.InquiryUnitSerialNumber[DeviceIndex] = + Controller->V2.InquiryUnitSerialNumber[DeviceIndex-1]; + } + Controller->V2.PhysicalDeviceInformation + [PhysicalDeviceIndex] = + PhysicalDeviceInfo; + Controller->V2.InquiryUnitSerialNumber + [PhysicalDeviceIndex] = + InquiryUnitSerialNumber; + Controller->V2.NeedDeviceSerialNumberInformation = true; + } + } + if (PhysicalDeviceInfo != NULL) + { + if (NewPhysicalDeviceInfo->PhysicalDeviceState != + PhysicalDeviceInfo->PhysicalDeviceState) + DAC960_Critical( + "Physical Device %d:%d is now %s\n", Controller, + NewPhysicalDeviceInfo->Channel, + NewPhysicalDeviceInfo->TargetID, + (NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Online + ? "ONLINE" + : NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Rebuild + ? "REBUILD" + : NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Missing + ? "MISSING" + : NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Critical + ? "CRITICAL" + : NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Dead + ? "DEAD" + : NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_SuspectedDead + ? "SUSPECTED-DEAD" + : NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_CommandedOffline + ? "COMMANDED-OFFLINE" + : NewPhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Standby + ? "STANDBY" : "UNKNOWN")); + if ((NewPhysicalDeviceInfo->ParityErrors != + PhysicalDeviceInfo->ParityErrors) || + (NewPhysicalDeviceInfo->SoftErrors != + PhysicalDeviceInfo->SoftErrors) || + (NewPhysicalDeviceInfo->HardErrors != + PhysicalDeviceInfo->HardErrors) || + (NewPhysicalDeviceInfo->MiscellaneousErrors != + PhysicalDeviceInfo->MiscellaneousErrors) || + (NewPhysicalDeviceInfo->CommandTimeouts != + PhysicalDeviceInfo->CommandTimeouts) || + (NewPhysicalDeviceInfo->Retries != + PhysicalDeviceInfo->Retries) || + (NewPhysicalDeviceInfo->Aborts != + PhysicalDeviceInfo->Aborts) || + (NewPhysicalDeviceInfo->PredictedFailuresDetected != + PhysicalDeviceInfo->PredictedFailuresDetected)) + { + DAC960_Critical("Physical Device %d:%d Errors: " + "Parity = %d, Soft = %d, " + "Hard = %d, Misc = %d\n", + Controller, + NewPhysicalDeviceInfo->Channel, + NewPhysicalDeviceInfo->TargetID, + NewPhysicalDeviceInfo->ParityErrors, + NewPhysicalDeviceInfo->SoftErrors, + NewPhysicalDeviceInfo->HardErrors, + NewPhysicalDeviceInfo->MiscellaneousErrors); + DAC960_Critical("Physical Device %d:%d Errors: " + "Timeouts = %d, Retries = %d, " + "Aborts = %d, Predicted = %d\n", + Controller, + NewPhysicalDeviceInfo->Channel, + NewPhysicalDeviceInfo->TargetID, + NewPhysicalDeviceInfo->CommandTimeouts, + NewPhysicalDeviceInfo->Retries, + NewPhysicalDeviceInfo->Aborts, + NewPhysicalDeviceInfo + ->PredictedFailuresDetected); + } + if ((PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_Dead || + PhysicalDeviceInfo->PhysicalDeviceState + == DAC960_V2_Device_InvalidState) && + NewPhysicalDeviceInfo->PhysicalDeviceState + != DAC960_V2_Device_Dead) + Controller->V2.NeedDeviceSerialNumberInformation = true; + memcpy(PhysicalDeviceInfo, NewPhysicalDeviceInfo, + sizeof(DAC960_V2_PhysicalDeviceInfo_T)); + } + NewPhysicalDeviceInfo->LogicalUnit++; + Controller->V2.PhysicalDeviceIndex++; + } + else if (CommandOpcode == DAC960_V2_GetPhysicalDeviceInfoValid) + { + unsigned int DeviceIndex; + for (DeviceIndex = Controller->V2.PhysicalDeviceIndex; + DeviceIndex < DAC960_V2_MaxPhysicalDevices; + DeviceIndex++) + { + DAC960_V2_PhysicalDeviceInfo_T *PhysicalDeviceInfo = + Controller->V2.PhysicalDeviceInformation[DeviceIndex]; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + Controller->V2.InquiryUnitSerialNumber[DeviceIndex]; + if (PhysicalDeviceInfo == NULL) break; + DAC960_Critical("Physical Device %d:%d No Longer Exists\n", + Controller, + PhysicalDeviceInfo->Channel, + PhysicalDeviceInfo->TargetID); + Controller->V2.PhysicalDeviceInformation[DeviceIndex] = NULL; + Controller->V2.InquiryUnitSerialNumber[DeviceIndex] = NULL; + kfree(PhysicalDeviceInfo); + kfree(InquiryUnitSerialNumber); + } + Controller->V2.NeedPhysicalDeviceInformation = false; + } + else if (CommandOpcode == DAC960_V2_GetLogicalDeviceInfoValid && + CommandStatus == DAC960_V2_NormalCompletion) + { + DAC960_V2_LogicalDeviceInfo_T *NewLogicalDeviceInfo = + Controller->V2.NewLogicalDeviceInformation; + unsigned short LogicalDeviceNumber = + NewLogicalDeviceInfo->LogicalDeviceNumber; + DAC960_V2_LogicalDeviceInfo_T *LogicalDeviceInfo = + Controller->V2.LogicalDeviceInformation[LogicalDeviceNumber]; + if (LogicalDeviceInfo == NULL) + { + DAC960_V2_PhysicalDevice_T PhysicalDevice; + PhysicalDevice.Controller = 0; + PhysicalDevice.Channel = NewLogicalDeviceInfo->Channel; + PhysicalDevice.TargetID = NewLogicalDeviceInfo->TargetID; + PhysicalDevice.LogicalUnit = NewLogicalDeviceInfo->LogicalUnit; + Controller->V2.LogicalDriveToVirtualDevice[LogicalDeviceNumber] = + PhysicalDevice; + LogicalDeviceInfo = (DAC960_V2_LogicalDeviceInfo_T *) + kmalloc(sizeof(DAC960_V2_LogicalDeviceInfo_T), GFP_ATOMIC); + Controller->V2.LogicalDeviceInformation[LogicalDeviceNumber] = + LogicalDeviceInfo; + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) " + "Now Exists%s\n", Controller, + LogicalDeviceNumber, + Controller->ControllerNumber, + LogicalDeviceNumber, + (LogicalDeviceInfo != NULL + ? "" : " - Allocation Failed")); + if (LogicalDeviceInfo != NULL) + { + memset(LogicalDeviceInfo, 0, + sizeof(DAC960_V2_LogicalDeviceInfo_T)); + DAC960_ComputeGenericDiskInfo(Controller); + } + } + if (LogicalDeviceInfo != NULL) + { + unsigned long LogicalDeviceSize = + NewLogicalDeviceInfo->ConfigurableDeviceSize; + if (NewLogicalDeviceInfo->LogicalDeviceState != + LogicalDeviceInfo->LogicalDeviceState) + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) " + "is now %s\n", Controller, + LogicalDeviceNumber, + Controller->ControllerNumber, + LogicalDeviceNumber, + (NewLogicalDeviceInfo->LogicalDeviceState + == DAC960_V2_LogicalDevice_Online + ? "ONLINE" + : NewLogicalDeviceInfo->LogicalDeviceState + == DAC960_V2_LogicalDevice_Critical + ? "CRITICAL" : "OFFLINE")); + if ((NewLogicalDeviceInfo->SoftErrors != + LogicalDeviceInfo->SoftErrors) || + (NewLogicalDeviceInfo->CommandsFailed != + LogicalDeviceInfo->CommandsFailed) || + (NewLogicalDeviceInfo->DeferredWriteErrors != + LogicalDeviceInfo->DeferredWriteErrors)) + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) Errors: " + "Soft = %d, Failed = %d, Deferred Write = %d\n", + Controller, LogicalDeviceNumber, + Controller->ControllerNumber, + LogicalDeviceNumber, + NewLogicalDeviceInfo->SoftErrors, + NewLogicalDeviceInfo->CommandsFailed, + NewLogicalDeviceInfo->DeferredWriteErrors); + if (NewLogicalDeviceInfo->ConsistencyCheckInProgress) + DAC960_V2_ReportProgress(Controller, + "Consistency Check", + LogicalDeviceNumber, + NewLogicalDeviceInfo + ->ConsistencyCheckBlockNumber, + LogicalDeviceSize); + else if (NewLogicalDeviceInfo->RebuildInProgress) + DAC960_V2_ReportProgress(Controller, + "Rebuild", + LogicalDeviceNumber, + NewLogicalDeviceInfo + ->RebuildBlockNumber, + LogicalDeviceSize); + else if (NewLogicalDeviceInfo->BackgroundInitializationInProgress) + DAC960_V2_ReportProgress(Controller, + "Background Initialization", + LogicalDeviceNumber, + NewLogicalDeviceInfo + ->BackgroundInitializationBlockNumber, + LogicalDeviceSize); + else if (NewLogicalDeviceInfo->ForegroundInitializationInProgress) + DAC960_V2_ReportProgress(Controller, + "Foreground Initialization", + LogicalDeviceNumber, + NewLogicalDeviceInfo + ->ForegroundInitializationBlockNumber, + LogicalDeviceSize); + else if (NewLogicalDeviceInfo->DataMigrationInProgress) + DAC960_V2_ReportProgress(Controller, + "Data Migration", + LogicalDeviceNumber, + NewLogicalDeviceInfo + ->DataMigrationBlockNumber, + LogicalDeviceSize); + else if (NewLogicalDeviceInfo->PatrolOperationInProgress) + DAC960_V2_ReportProgress(Controller, + "Patrol Operation", + LogicalDeviceNumber, + NewLogicalDeviceInfo + ->PatrolOperationBlockNumber, + LogicalDeviceSize); + if (LogicalDeviceInfo->BackgroundInitializationInProgress && + !NewLogicalDeviceInfo->BackgroundInitializationInProgress) + DAC960_Progress("Logical Drive %d (/dev/rd/c%dd%d) " + "Background Initialization %s\n", + Controller, + LogicalDeviceNumber, + Controller->ControllerNumber, + LogicalDeviceNumber, + (NewLogicalDeviceInfo->LogicalDeviceControl + .LogicalDeviceInitialized + ? "Completed" : "Failed")); + memcpy(LogicalDeviceInfo, NewLogicalDeviceInfo, + sizeof(DAC960_V2_LogicalDeviceInfo_T)); + } + Controller->V2.LogicalDriveFoundDuringScan + [LogicalDeviceNumber] = true; + NewLogicalDeviceInfo->LogicalDeviceNumber++; + } + else if (CommandOpcode == DAC960_V2_GetLogicalDeviceInfoValid) + { + int LogicalDriveNumber; + for (LogicalDriveNumber = 0; + LogicalDriveNumber < DAC960_MaxLogicalDrives; + LogicalDriveNumber++) + { + DAC960_V2_LogicalDeviceInfo_T *LogicalDeviceInfo = + Controller->V2.LogicalDeviceInformation[LogicalDriveNumber]; + if (LogicalDeviceInfo == NULL || + Controller->V2.LogicalDriveFoundDuringScan + [LogicalDriveNumber]) + continue; + DAC960_Critical("Logical Drive %d (/dev/rd/c%dd%d) " + "No Longer Exists\n", Controller, + LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber); + Controller->V2.LogicalDeviceInformation + [LogicalDriveNumber] = NULL; + kfree(LogicalDeviceInfo); + Controller->LogicalDriveInitiallyAccessible + [LogicalDriveNumber] = false; + DAC960_ComputeGenericDiskInfo(Controller); + } + Controller->V2.NeedLogicalDeviceInformation = false; + } + else if (CommandOpcode == DAC960_V2_SCSI_10_Passthru) + { + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + Controller->V2.InquiryUnitSerialNumber[Controller->V2.PhysicalDeviceIndex - 1]; + + if (CommandStatus != DAC960_V2_NormalCompletion) { + memset(InquiryUnitSerialNumber, + 0, sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + InquiryUnitSerialNumber->PeripheralDeviceType = 0x1F; + } else + memcpy(InquiryUnitSerialNumber, + Controller->V2.NewInquiryUnitSerialNumber, + sizeof(DAC960_SCSI_Inquiry_UnitSerialNumber_T)); + + Controller->V2.NeedDeviceSerialNumberInformation = false; + } + + if (Controller->V2.HealthStatusBuffer->NextEventSequenceNumber + - Controller->V2.NextEventSequenceNumber > 0) + { + CommandMailbox->GetEvent.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->GetEvent.DataTransferSize = sizeof(DAC960_V2_Event_T); + CommandMailbox->GetEvent.EventSequenceNumberHigh16 = + Controller->V2.NextEventSequenceNumber >> 16; + CommandMailbox->GetEvent.ControllerNumber = 0; + CommandMailbox->GetEvent.IOCTL_Opcode = + DAC960_V2_GetEvent; + CommandMailbox->GetEvent.EventSequenceNumberLow16 = + Controller->V2.NextEventSequenceNumber & 0xFFFF; + CommandMailbox->GetEvent.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.EventDMA; + CommandMailbox->GetEvent.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->GetEvent.DataTransferSize; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V2.NeedPhysicalDeviceInformation) + { + if (Controller->V2.NeedDeviceSerialNumberInformation) + { + DAC960_SCSI_Inquiry_UnitSerialNumber_T *InquiryUnitSerialNumber = + Controller->V2.NewInquiryUnitSerialNumber; + InquiryUnitSerialNumber->PeripheralDeviceType = 0x1F; + + DAC960_V2_ConstructNewUnitSerialNumber(Controller, CommandMailbox, + Controller->V2.NewPhysicalDeviceInformation->Channel, + Controller->V2.NewPhysicalDeviceInformation->TargetID, + Controller->V2.NewPhysicalDeviceInformation->LogicalUnit - 1); + + + DAC960_QueueCommand(Command); + return; + } + if (Controller->V2.StartPhysicalDeviceInformationScan) + { + Controller->V2.PhysicalDeviceIndex = 0; + Controller->V2.NewPhysicalDeviceInformation->Channel = 0; + Controller->V2.NewPhysicalDeviceInformation->TargetID = 0; + Controller->V2.NewPhysicalDeviceInformation->LogicalUnit = 0; + Controller->V2.StartPhysicalDeviceInformationScan = false; + } + CommandMailbox->PhysicalDeviceInfo.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->PhysicalDeviceInfo.DataTransferSize = + sizeof(DAC960_V2_PhysicalDeviceInfo_T); + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.LogicalUnit = + Controller->V2.NewPhysicalDeviceInformation->LogicalUnit; + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.TargetID = + Controller->V2.NewPhysicalDeviceInformation->TargetID; + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.Channel = + Controller->V2.NewPhysicalDeviceInformation->Channel; + CommandMailbox->PhysicalDeviceInfo.IOCTL_Opcode = + DAC960_V2_GetPhysicalDeviceInfoValid; + CommandMailbox->PhysicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewPhysicalDeviceInformationDMA; + CommandMailbox->PhysicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->PhysicalDeviceInfo.DataTransferSize; + DAC960_QueueCommand(Command); + return; + } + if (Controller->V2.NeedLogicalDeviceInformation) + { + if (Controller->V2.StartLogicalDeviceInformationScan) + { + int LogicalDriveNumber; + for (LogicalDriveNumber = 0; + LogicalDriveNumber < DAC960_MaxLogicalDrives; + LogicalDriveNumber++) + Controller->V2.LogicalDriveFoundDuringScan + [LogicalDriveNumber] = false; + Controller->V2.NewLogicalDeviceInformation->LogicalDeviceNumber = 0; + Controller->V2.StartLogicalDeviceInformationScan = false; + } + CommandMailbox->LogicalDeviceInfo.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->LogicalDeviceInfo.DataTransferSize = + sizeof(DAC960_V2_LogicalDeviceInfo_T); + CommandMailbox->LogicalDeviceInfo.LogicalDevice.LogicalDeviceNumber = + Controller->V2.NewLogicalDeviceInformation->LogicalDeviceNumber; + CommandMailbox->LogicalDeviceInfo.IOCTL_Opcode = + DAC960_V2_GetLogicalDeviceInfoValid; + CommandMailbox->LogicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewLogicalDeviceInformationDMA; + CommandMailbox->LogicalDeviceInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->LogicalDeviceInfo.DataTransferSize; + DAC960_QueueCommand(Command); + return; + } + Controller->MonitoringTimerCount++; + Controller->MonitoringTimer.expires = + jiffies + DAC960_HealthStatusMonitoringInterval; + add_timer(&Controller->MonitoringTimer); + } + if (CommandType == DAC960_ImmediateCommand) + { + complete(Command->Completion); + Command->Completion = NULL; + return; + } + if (CommandType == DAC960_QueuedCommand) + { + DAC960_V2_KernelCommand_T *KernelCommand = Command->V2.KernelCommand; + KernelCommand->CommandStatus = CommandStatus; + KernelCommand->RequestSenseLength = Command->V2.RequestSenseLength; + KernelCommand->DataTransferLength = Command->V2.DataTransferResidue; + Command->V2.KernelCommand = NULL; + DAC960_DeallocateCommand(Command); + KernelCommand->CompletionFunction(KernelCommand); + return; + } + /* + Queue a Status Monitoring Command to the Controller using the just + completed Command if one was deferred previously due to lack of a + free Command when the Monitoring Timer Function was called. + */ + if (Controller->MonitoringCommandDeferred) + { + Controller->MonitoringCommandDeferred = false; + DAC960_V2_QueueMonitoringCommand(Command); + return; + } + /* + Deallocate the Command. + */ + DAC960_DeallocateCommand(Command); + /* + Wake up any processes waiting on a free Command. + */ + wake_up(&Controller->CommandWaitQueue); +} + + +/* + DAC960_BA_InterruptHandler handles hardware interrupts from DAC960 BA Series + Controllers. +*/ + +static irqreturn_t DAC960_BA_InterruptHandler(int IRQ_Channel, + void *DeviceIdentifier, + struct pt_regs *InterruptRegisters) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) DeviceIdentifier; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V2_StatusMailbox_T *NextStatusMailbox; + unsigned long flags; + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_BA_AcknowledgeInterrupt(ControllerBaseAddress); + NextStatusMailbox = Controller->V2.NextStatusMailbox; + while (NextStatusMailbox->Fields.CommandIdentifier > 0) + { + DAC960_V2_CommandIdentifier_T CommandIdentifier = + NextStatusMailbox->Fields.CommandIdentifier; + DAC960_Command_T *Command = Controller->Commands[CommandIdentifier-1]; + Command->V2.CommandStatus = NextStatusMailbox->Fields.CommandStatus; + Command->V2.RequestSenseLength = + NextStatusMailbox->Fields.RequestSenseLength; + Command->V2.DataTransferResidue = + NextStatusMailbox->Fields.DataTransferResidue; + NextStatusMailbox->Words[0] = 0; + if (++NextStatusMailbox > Controller->V2.LastStatusMailbox) + NextStatusMailbox = Controller->V2.FirstStatusMailbox; + DAC960_V2_ProcessCompletedCommand(Command); + } + Controller->V2.NextStatusMailbox = NextStatusMailbox; + /* + Attempt to remove additional I/O Requests from the Controller's + I/O Request Queue and queue them to the Controller. + */ + DAC960_ProcessRequest(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return IRQ_HANDLED; +} + + +/* + DAC960_LP_InterruptHandler handles hardware interrupts from DAC960 LP Series + Controllers. +*/ + +static irqreturn_t DAC960_LP_InterruptHandler(int IRQ_Channel, + void *DeviceIdentifier, + struct pt_regs *InterruptRegisters) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) DeviceIdentifier; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V2_StatusMailbox_T *NextStatusMailbox; + unsigned long flags; + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_LP_AcknowledgeInterrupt(ControllerBaseAddress); + NextStatusMailbox = Controller->V2.NextStatusMailbox; + while (NextStatusMailbox->Fields.CommandIdentifier > 0) + { + DAC960_V2_CommandIdentifier_T CommandIdentifier = + NextStatusMailbox->Fields.CommandIdentifier; + DAC960_Command_T *Command = Controller->Commands[CommandIdentifier-1]; + Command->V2.CommandStatus = NextStatusMailbox->Fields.CommandStatus; + Command->V2.RequestSenseLength = + NextStatusMailbox->Fields.RequestSenseLength; + Command->V2.DataTransferResidue = + NextStatusMailbox->Fields.DataTransferResidue; + NextStatusMailbox->Words[0] = 0; + if (++NextStatusMailbox > Controller->V2.LastStatusMailbox) + NextStatusMailbox = Controller->V2.FirstStatusMailbox; + DAC960_V2_ProcessCompletedCommand(Command); + } + Controller->V2.NextStatusMailbox = NextStatusMailbox; + /* + Attempt to remove additional I/O Requests from the Controller's + I/O Request Queue and queue them to the Controller. + */ + DAC960_ProcessRequest(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return IRQ_HANDLED; +} + + +/* + DAC960_LA_InterruptHandler handles hardware interrupts from DAC960 LA Series + Controllers. +*/ + +static irqreturn_t DAC960_LA_InterruptHandler(int IRQ_Channel, + void *DeviceIdentifier, + struct pt_regs *InterruptRegisters) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) DeviceIdentifier; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_StatusMailbox_T *NextStatusMailbox; + unsigned long flags; + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_LA_AcknowledgeInterrupt(ControllerBaseAddress); + NextStatusMailbox = Controller->V1.NextStatusMailbox; + while (NextStatusMailbox->Fields.Valid) + { + DAC960_V1_CommandIdentifier_T CommandIdentifier = + NextStatusMailbox->Fields.CommandIdentifier; + DAC960_Command_T *Command = Controller->Commands[CommandIdentifier-1]; + Command->V1.CommandStatus = NextStatusMailbox->Fields.CommandStatus; + NextStatusMailbox->Word = 0; + if (++NextStatusMailbox > Controller->V1.LastStatusMailbox) + NextStatusMailbox = Controller->V1.FirstStatusMailbox; + DAC960_V1_ProcessCompletedCommand(Command); + } + Controller->V1.NextStatusMailbox = NextStatusMailbox; + /* + Attempt to remove additional I/O Requests from the Controller's + I/O Request Queue and queue them to the Controller. + */ + DAC960_ProcessRequest(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return IRQ_HANDLED; +} + + +/* + DAC960_PG_InterruptHandler handles hardware interrupts from DAC960 PG Series + Controllers. +*/ + +static irqreturn_t DAC960_PG_InterruptHandler(int IRQ_Channel, + void *DeviceIdentifier, + struct pt_regs *InterruptRegisters) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) DeviceIdentifier; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + DAC960_V1_StatusMailbox_T *NextStatusMailbox; + unsigned long flags; + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_PG_AcknowledgeInterrupt(ControllerBaseAddress); + NextStatusMailbox = Controller->V1.NextStatusMailbox; + while (NextStatusMailbox->Fields.Valid) + { + DAC960_V1_CommandIdentifier_T CommandIdentifier = + NextStatusMailbox->Fields.CommandIdentifier; + DAC960_Command_T *Command = Controller->Commands[CommandIdentifier-1]; + Command->V1.CommandStatus = NextStatusMailbox->Fields.CommandStatus; + NextStatusMailbox->Word = 0; + if (++NextStatusMailbox > Controller->V1.LastStatusMailbox) + NextStatusMailbox = Controller->V1.FirstStatusMailbox; + DAC960_V1_ProcessCompletedCommand(Command); + } + Controller->V1.NextStatusMailbox = NextStatusMailbox; + /* + Attempt to remove additional I/O Requests from the Controller's + I/O Request Queue and queue them to the Controller. + */ + DAC960_ProcessRequest(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return IRQ_HANDLED; +} + + +/* + DAC960_PD_InterruptHandler handles hardware interrupts from DAC960 PD Series + Controllers. +*/ + +static irqreturn_t DAC960_PD_InterruptHandler(int IRQ_Channel, + void *DeviceIdentifier, + struct pt_regs *InterruptRegisters) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) DeviceIdentifier; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + unsigned long flags; + + spin_lock_irqsave(&Controller->queue_lock, flags); + while (DAC960_PD_StatusAvailableP(ControllerBaseAddress)) + { + DAC960_V1_CommandIdentifier_T CommandIdentifier = + DAC960_PD_ReadStatusCommandIdentifier(ControllerBaseAddress); + DAC960_Command_T *Command = Controller->Commands[CommandIdentifier-1]; + Command->V1.CommandStatus = + DAC960_PD_ReadStatusRegister(ControllerBaseAddress); + DAC960_PD_AcknowledgeInterrupt(ControllerBaseAddress); + DAC960_PD_AcknowledgeStatus(ControllerBaseAddress); + DAC960_V1_ProcessCompletedCommand(Command); + } + /* + Attempt to remove additional I/O Requests from the Controller's + I/O Request Queue and queue them to the Controller. + */ + DAC960_ProcessRequest(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return IRQ_HANDLED; +} + + +/* + DAC960_P_InterruptHandler handles hardware interrupts from DAC960 P Series + Controllers. + + Translations of DAC960_V1_Enquiry and DAC960_V1_GetDeviceState rely + on the data having been placed into DAC960_Controller_T, rather than + an arbitrary buffer. +*/ + +static irqreturn_t DAC960_P_InterruptHandler(int IRQ_Channel, + void *DeviceIdentifier, + struct pt_regs *InterruptRegisters) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) DeviceIdentifier; + void __iomem *ControllerBaseAddress = Controller->BaseAddress; + unsigned long flags; + + spin_lock_irqsave(&Controller->queue_lock, flags); + while (DAC960_PD_StatusAvailableP(ControllerBaseAddress)) + { + DAC960_V1_CommandIdentifier_T CommandIdentifier = + DAC960_PD_ReadStatusCommandIdentifier(ControllerBaseAddress); + DAC960_Command_T *Command = Controller->Commands[CommandIdentifier-1]; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_CommandOpcode_T CommandOpcode = + CommandMailbox->Common.CommandOpcode; + Command->V1.CommandStatus = + DAC960_PD_ReadStatusRegister(ControllerBaseAddress); + DAC960_PD_AcknowledgeInterrupt(ControllerBaseAddress); + DAC960_PD_AcknowledgeStatus(ControllerBaseAddress); + switch (CommandOpcode) + { + case DAC960_V1_Enquiry_Old: + Command->V1.CommandMailbox.Common.CommandOpcode = DAC960_V1_Enquiry; + DAC960_P_To_PD_TranslateEnquiry(Controller->V1.NewEnquiry); + break; + case DAC960_V1_GetDeviceState_Old: + Command->V1.CommandMailbox.Common.CommandOpcode = + DAC960_V1_GetDeviceState; + DAC960_P_To_PD_TranslateDeviceState(Controller->V1.NewDeviceState); + break; + case DAC960_V1_Read_Old: + Command->V1.CommandMailbox.Common.CommandOpcode = DAC960_V1_Read; + DAC960_P_To_PD_TranslateReadWriteCommand(CommandMailbox); + break; + case DAC960_V1_Write_Old: + Command->V1.CommandMailbox.Common.CommandOpcode = DAC960_V1_Write; + DAC960_P_To_PD_TranslateReadWriteCommand(CommandMailbox); + break; + case DAC960_V1_ReadWithScatterGather_Old: + Command->V1.CommandMailbox.Common.CommandOpcode = + DAC960_V1_ReadWithScatterGather; + DAC960_P_To_PD_TranslateReadWriteCommand(CommandMailbox); + break; + case DAC960_V1_WriteWithScatterGather_Old: + Command->V1.CommandMailbox.Common.CommandOpcode = + DAC960_V1_WriteWithScatterGather; + DAC960_P_To_PD_TranslateReadWriteCommand(CommandMailbox); + break; + default: + break; + } + DAC960_V1_ProcessCompletedCommand(Command); + } + /* + Attempt to remove additional I/O Requests from the Controller's + I/O Request Queue and queue them to the Controller. + */ + DAC960_ProcessRequest(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return IRQ_HANDLED; +} + + +/* + DAC960_V1_QueueMonitoringCommand queues a Monitoring Command to DAC960 V1 + Firmware Controllers. +*/ + +static void DAC960_V1_QueueMonitoringCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_MonitoringCommand; + CommandMailbox->Type3.CommandOpcode = DAC960_V1_Enquiry; + CommandMailbox->Type3.BusAddress = Controller->V1.NewEnquiryDMA; + DAC960_QueueCommand(Command); +} + + +/* + DAC960_V2_QueueMonitoringCommand queues a Monitoring Command to DAC960 V2 + Firmware Controllers. +*/ + +static void DAC960_V2_QueueMonitoringCommand(DAC960_Command_T *Command) +{ + DAC960_Controller_T *Controller = Command->Controller; + DAC960_V2_CommandMailbox_T *CommandMailbox = &Command->V2.CommandMailbox; + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_MonitoringCommand; + CommandMailbox->ControllerInfo.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->ControllerInfo.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->ControllerInfo.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->ControllerInfo.DataTransferSize = + sizeof(DAC960_V2_ControllerInfo_T); + CommandMailbox->ControllerInfo.ControllerNumber = 0; + CommandMailbox->ControllerInfo.IOCTL_Opcode = DAC960_V2_GetControllerInfo; + CommandMailbox->ControllerInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewControllerInformationDMA; + CommandMailbox->ControllerInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->ControllerInfo.DataTransferSize; + DAC960_QueueCommand(Command); +} + + +/* + DAC960_MonitoringTimerFunction is the timer function for monitoring + the status of DAC960 Controllers. +*/ + +static void DAC960_MonitoringTimerFunction(unsigned long TimerData) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) TimerData; + DAC960_Command_T *Command; + unsigned long flags; + + if (Controller->FirmwareType == DAC960_V1_Controller) + { + spin_lock_irqsave(&Controller->queue_lock, flags); + /* + Queue a Status Monitoring Command to Controller. + */ + Command = DAC960_AllocateCommand(Controller); + if (Command != NULL) + DAC960_V1_QueueMonitoringCommand(Command); + else Controller->MonitoringCommandDeferred = true; + spin_unlock_irqrestore(&Controller->queue_lock, flags); + } + else + { + DAC960_V2_ControllerInfo_T *ControllerInfo = + &Controller->V2.ControllerInformation; + unsigned int StatusChangeCounter = + Controller->V2.HealthStatusBuffer->StatusChangeCounter; + boolean ForceMonitoringCommand = false; + if (jiffies - Controller->SecondaryMonitoringTime + > DAC960_SecondaryMonitoringInterval) + { + int LogicalDriveNumber; + for (LogicalDriveNumber = 0; + LogicalDriveNumber < DAC960_MaxLogicalDrives; + LogicalDriveNumber++) + { + DAC960_V2_LogicalDeviceInfo_T *LogicalDeviceInfo = + Controller->V2.LogicalDeviceInformation[LogicalDriveNumber]; + if (LogicalDeviceInfo == NULL) continue; + if (!LogicalDeviceInfo->LogicalDeviceControl + .LogicalDeviceInitialized) + { + ForceMonitoringCommand = true; + break; + } + } + Controller->SecondaryMonitoringTime = jiffies; + } + if (StatusChangeCounter == Controller->V2.StatusChangeCounter && + Controller->V2.HealthStatusBuffer->NextEventSequenceNumber + == Controller->V2.NextEventSequenceNumber && + (ControllerInfo->BackgroundInitializationsActive + + ControllerInfo->LogicalDeviceInitializationsActive + + ControllerInfo->PhysicalDeviceInitializationsActive + + ControllerInfo->ConsistencyChecksActive + + ControllerInfo->RebuildsActive + + ControllerInfo->OnlineExpansionsActive == 0 || + jiffies - Controller->PrimaryMonitoringTime + < DAC960_MonitoringTimerInterval) && + !ForceMonitoringCommand) + { + Controller->MonitoringTimer.expires = + jiffies + DAC960_HealthStatusMonitoringInterval; + add_timer(&Controller->MonitoringTimer); + return; + } + Controller->V2.StatusChangeCounter = StatusChangeCounter; + Controller->PrimaryMonitoringTime = jiffies; + + spin_lock_irqsave(&Controller->queue_lock, flags); + /* + Queue a Status Monitoring Command to Controller. + */ + Command = DAC960_AllocateCommand(Controller); + if (Command != NULL) + DAC960_V2_QueueMonitoringCommand(Command); + else Controller->MonitoringCommandDeferred = true; + spin_unlock_irqrestore(&Controller->queue_lock, flags); + /* + Wake up any processes waiting on a Health Status Buffer change. + */ + wake_up(&Controller->HealthStatusWaitQueue); + } +} + +/* + DAC960_CheckStatusBuffer verifies that there is room to hold ByteCount + additional bytes in the Combined Status Buffer and grows the buffer if + necessary. It returns true if there is enough room and false otherwise. +*/ + +static boolean DAC960_CheckStatusBuffer(DAC960_Controller_T *Controller, + unsigned int ByteCount) +{ + unsigned char *NewStatusBuffer; + if (Controller->InitialStatusLength + 1 + + Controller->CurrentStatusLength + ByteCount + 1 <= + Controller->CombinedStatusBufferLength) + return true; + if (Controller->CombinedStatusBufferLength == 0) + { + unsigned int NewStatusBufferLength = DAC960_InitialStatusBufferSize; + while (NewStatusBufferLength < ByteCount) + NewStatusBufferLength *= 2; + Controller->CombinedStatusBuffer = + (unsigned char *) kmalloc(NewStatusBufferLength, GFP_ATOMIC); + if (Controller->CombinedStatusBuffer == NULL) return false; + Controller->CombinedStatusBufferLength = NewStatusBufferLength; + return true; + } + NewStatusBuffer = (unsigned char *) + kmalloc(2 * Controller->CombinedStatusBufferLength, GFP_ATOMIC); + if (NewStatusBuffer == NULL) + { + DAC960_Warning("Unable to expand Combined Status Buffer - Truncating\n", + Controller); + return false; + } + memcpy(NewStatusBuffer, Controller->CombinedStatusBuffer, + Controller->CombinedStatusBufferLength); + kfree(Controller->CombinedStatusBuffer); + Controller->CombinedStatusBuffer = NewStatusBuffer; + Controller->CombinedStatusBufferLength *= 2; + Controller->CurrentStatusBuffer = + &NewStatusBuffer[Controller->InitialStatusLength + 1]; + return true; +} + + +/* + DAC960_Message prints Driver Messages. +*/ + +static void DAC960_Message(DAC960_MessageLevel_T MessageLevel, + unsigned char *Format, + DAC960_Controller_T *Controller, + ...) +{ + static unsigned char Buffer[DAC960_LineBufferSize]; + static boolean BeginningOfLine = true; + va_list Arguments; + int Length = 0; + va_start(Arguments, Controller); + Length = vsprintf(Buffer, Format, Arguments); + va_end(Arguments); + if (Controller == NULL) + printk("%sDAC960#%d: %s", DAC960_MessageLevelMap[MessageLevel], + DAC960_ControllerCount, Buffer); + else if (MessageLevel == DAC960_AnnounceLevel || + MessageLevel == DAC960_InfoLevel) + { + if (!Controller->ControllerInitialized) + { + if (DAC960_CheckStatusBuffer(Controller, Length)) + { + strcpy(&Controller->CombinedStatusBuffer + [Controller->InitialStatusLength], + Buffer); + Controller->InitialStatusLength += Length; + Controller->CurrentStatusBuffer = + &Controller->CombinedStatusBuffer + [Controller->InitialStatusLength + 1]; + } + if (MessageLevel == DAC960_AnnounceLevel) + { + static int AnnouncementLines = 0; + if (++AnnouncementLines <= 2) + printk("%sDAC960: %s", DAC960_MessageLevelMap[MessageLevel], + Buffer); + } + else + { + if (BeginningOfLine) + { + if (Buffer[0] != '\n' || Length > 1) + printk("%sDAC960#%d: %s", + DAC960_MessageLevelMap[MessageLevel], + Controller->ControllerNumber, Buffer); + } + else printk("%s", Buffer); + } + } + else if (DAC960_CheckStatusBuffer(Controller, Length)) + { + strcpy(&Controller->CurrentStatusBuffer[ + Controller->CurrentStatusLength], Buffer); + Controller->CurrentStatusLength += Length; + } + } + else if (MessageLevel == DAC960_ProgressLevel) + { + strcpy(Controller->ProgressBuffer, Buffer); + Controller->ProgressBufferLength = Length; + if (Controller->EphemeralProgressMessage) + { + if (jiffies - Controller->LastProgressReportTime + >= DAC960_ProgressReportingInterval) + { + printk("%sDAC960#%d: %s", DAC960_MessageLevelMap[MessageLevel], + Controller->ControllerNumber, Buffer); + Controller->LastProgressReportTime = jiffies; + } + } + else printk("%sDAC960#%d: %s", DAC960_MessageLevelMap[MessageLevel], + Controller->ControllerNumber, Buffer); + } + else if (MessageLevel == DAC960_UserCriticalLevel) + { + strcpy(&Controller->UserStatusBuffer[Controller->UserStatusLength], + Buffer); + Controller->UserStatusLength += Length; + if (Buffer[0] != '\n' || Length > 1) + printk("%sDAC960#%d: %s", DAC960_MessageLevelMap[MessageLevel], + Controller->ControllerNumber, Buffer); + } + else + { + if (BeginningOfLine) + printk("%sDAC960#%d: %s", DAC960_MessageLevelMap[MessageLevel], + Controller->ControllerNumber, Buffer); + else printk("%s", Buffer); + } + BeginningOfLine = (Buffer[Length-1] == '\n'); +} + + +/* + DAC960_ParsePhysicalDevice parses spaces followed by a Physical Device + Channel:TargetID specification from a User Command string. It updates + Channel and TargetID and returns true on success and false on failure. +*/ + +static boolean DAC960_ParsePhysicalDevice(DAC960_Controller_T *Controller, + char *UserCommandString, + unsigned char *Channel, + unsigned char *TargetID) +{ + char *NewUserCommandString = UserCommandString; + unsigned long XChannel, XTargetID; + while (*UserCommandString == ' ') UserCommandString++; + if (UserCommandString == NewUserCommandString) + return false; + XChannel = simple_strtoul(UserCommandString, &NewUserCommandString, 10); + if (NewUserCommandString == UserCommandString || + *NewUserCommandString != ':' || + XChannel >= Controller->Channels) + return false; + UserCommandString = ++NewUserCommandString; + XTargetID = simple_strtoul(UserCommandString, &NewUserCommandString, 10); + if (NewUserCommandString == UserCommandString || + *NewUserCommandString != '\0' || + XTargetID >= Controller->Targets) + return false; + *Channel = XChannel; + *TargetID = XTargetID; + return true; +} + + +/* + DAC960_ParseLogicalDrive parses spaces followed by a Logical Drive Number + specification from a User Command string. It updates LogicalDriveNumber and + returns true on success and false on failure. +*/ + +static boolean DAC960_ParseLogicalDrive(DAC960_Controller_T *Controller, + char *UserCommandString, + unsigned char *LogicalDriveNumber) +{ + char *NewUserCommandString = UserCommandString; + unsigned long XLogicalDriveNumber; + while (*UserCommandString == ' ') UserCommandString++; + if (UserCommandString == NewUserCommandString) + return false; + XLogicalDriveNumber = + simple_strtoul(UserCommandString, &NewUserCommandString, 10); + if (NewUserCommandString == UserCommandString || + *NewUserCommandString != '\0' || + XLogicalDriveNumber > DAC960_MaxLogicalDrives - 1) + return false; + *LogicalDriveNumber = XLogicalDriveNumber; + return true; +} + + +/* + DAC960_V1_SetDeviceState sets the Device State for a Physical Device for + DAC960 V1 Firmware Controllers. +*/ + +static void DAC960_V1_SetDeviceState(DAC960_Controller_T *Controller, + DAC960_Command_T *Command, + unsigned char Channel, + unsigned char TargetID, + DAC960_V1_PhysicalDeviceState_T + DeviceState, + const unsigned char *DeviceStateString) +{ + DAC960_V1_CommandMailbox_T *CommandMailbox = &Command->V1.CommandMailbox; + CommandMailbox->Type3D.CommandOpcode = DAC960_V1_StartDevice; + CommandMailbox->Type3D.Channel = Channel; + CommandMailbox->Type3D.TargetID = TargetID; + CommandMailbox->Type3D.DeviceState = DeviceState; + CommandMailbox->Type3D.Modifier = 0; + DAC960_ExecuteCommand(Command); + switch (Command->V1.CommandStatus) + { + case DAC960_V1_NormalCompletion: + DAC960_UserCritical("%s of Physical Device %d:%d Succeeded\n", Controller, + DeviceStateString, Channel, TargetID); + break; + case DAC960_V1_UnableToStartDevice: + DAC960_UserCritical("%s of Physical Device %d:%d Failed - " + "Unable to Start Device\n", Controller, + DeviceStateString, Channel, TargetID); + break; + case DAC960_V1_NoDeviceAtAddress: + DAC960_UserCritical("%s of Physical Device %d:%d Failed - " + "No Device at Address\n", Controller, + DeviceStateString, Channel, TargetID); + break; + case DAC960_V1_InvalidChannelOrTargetOrModifier: + DAC960_UserCritical("%s of Physical Device %d:%d Failed - " + "Invalid Channel or Target or Modifier\n", + Controller, DeviceStateString, Channel, TargetID); + break; + case DAC960_V1_ChannelBusy: + DAC960_UserCritical("%s of Physical Device %d:%d Failed - " + "Channel Busy\n", Controller, + DeviceStateString, Channel, TargetID); + break; + default: + DAC960_UserCritical("%s of Physical Device %d:%d Failed - " + "Unexpected Status %04X\n", Controller, + DeviceStateString, Channel, TargetID, + Command->V1.CommandStatus); + break; + } +} + + +/* + DAC960_V1_ExecuteUserCommand executes a User Command for DAC960 V1 Firmware + Controllers. +*/ + +static boolean DAC960_V1_ExecuteUserCommand(DAC960_Controller_T *Controller, + unsigned char *UserCommand) +{ + DAC960_Command_T *Command; + DAC960_V1_CommandMailbox_T *CommandMailbox; + unsigned long flags; + unsigned char Channel, TargetID, LogicalDriveNumber; + + spin_lock_irqsave(&Controller->queue_lock, flags); + while ((Command = DAC960_AllocateCommand(Controller)) == NULL) + DAC960_WaitForCommand(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + Controller->UserStatusLength = 0; + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox = &Command->V1.CommandMailbox; + if (strcmp(UserCommand, "flush-cache") == 0) + { + CommandMailbox->Type3.CommandOpcode = DAC960_V1_Flush; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Cache Flush Completed\n", Controller); + } + else if (strncmp(UserCommand, "kill", 4) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[4], + &Channel, &TargetID)) + { + DAC960_V1_DeviceState_T *DeviceState = + &Controller->V1.DeviceState[Channel][TargetID]; + if (DeviceState->Present && + DeviceState->DeviceType == DAC960_V1_DiskType && + DeviceState->DeviceState != DAC960_V1_Device_Dead) + DAC960_V1_SetDeviceState(Controller, Command, Channel, TargetID, + DAC960_V1_Device_Dead, "Kill"); + else DAC960_UserCritical("Kill of Physical Device %d:%d Illegal\n", + Controller, Channel, TargetID); + } + else if (strncmp(UserCommand, "make-online", 11) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[11], + &Channel, &TargetID)) + { + DAC960_V1_DeviceState_T *DeviceState = + &Controller->V1.DeviceState[Channel][TargetID]; + if (DeviceState->Present && + DeviceState->DeviceType == DAC960_V1_DiskType && + DeviceState->DeviceState == DAC960_V1_Device_Dead) + DAC960_V1_SetDeviceState(Controller, Command, Channel, TargetID, + DAC960_V1_Device_Online, "Make Online"); + else DAC960_UserCritical("Make Online of Physical Device %d:%d Illegal\n", + Controller, Channel, TargetID); + + } + else if (strncmp(UserCommand, "make-standby", 12) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[12], + &Channel, &TargetID)) + { + DAC960_V1_DeviceState_T *DeviceState = + &Controller->V1.DeviceState[Channel][TargetID]; + if (DeviceState->Present && + DeviceState->DeviceType == DAC960_V1_DiskType && + DeviceState->DeviceState == DAC960_V1_Device_Dead) + DAC960_V1_SetDeviceState(Controller, Command, Channel, TargetID, + DAC960_V1_Device_Standby, "Make Standby"); + else DAC960_UserCritical("Make Standby of Physical " + "Device %d:%d Illegal\n", + Controller, Channel, TargetID); + } + else if (strncmp(UserCommand, "rebuild", 7) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[7], + &Channel, &TargetID)) + { + CommandMailbox->Type3D.CommandOpcode = DAC960_V1_RebuildAsync; + CommandMailbox->Type3D.Channel = Channel; + CommandMailbox->Type3D.TargetID = TargetID; + DAC960_ExecuteCommand(Command); + switch (Command->V1.CommandStatus) + { + case DAC960_V1_NormalCompletion: + DAC960_UserCritical("Rebuild of Physical Device %d:%d Initiated\n", + Controller, Channel, TargetID); + break; + case DAC960_V1_AttemptToRebuildOnlineDrive: + DAC960_UserCritical("Rebuild of Physical Device %d:%d Failed - " + "Attempt to Rebuild Online or " + "Unresponsive Drive\n", + Controller, Channel, TargetID); + break; + case DAC960_V1_NewDiskFailedDuringRebuild: + DAC960_UserCritical("Rebuild of Physical Device %d:%d Failed - " + "New Disk Failed During Rebuild\n", + Controller, Channel, TargetID); + break; + case DAC960_V1_InvalidDeviceAddress: + DAC960_UserCritical("Rebuild of Physical Device %d:%d Failed - " + "Invalid Device Address\n", + Controller, Channel, TargetID); + break; + case DAC960_V1_RebuildOrCheckAlreadyInProgress: + DAC960_UserCritical("Rebuild of Physical Device %d:%d Failed - " + "Rebuild or Consistency Check Already " + "in Progress\n", Controller, Channel, TargetID); + break; + default: + DAC960_UserCritical("Rebuild of Physical Device %d:%d Failed - " + "Unexpected Status %04X\n", Controller, + Channel, TargetID, Command->V1.CommandStatus); + break; + } + } + else if (strncmp(UserCommand, "check-consistency", 17) == 0 && + DAC960_ParseLogicalDrive(Controller, &UserCommand[17], + &LogicalDriveNumber)) + { + CommandMailbox->Type3C.CommandOpcode = DAC960_V1_CheckConsistencyAsync; + CommandMailbox->Type3C.LogicalDriveNumber = LogicalDriveNumber; + CommandMailbox->Type3C.AutoRestore = true; + DAC960_ExecuteCommand(Command); + switch (Command->V1.CommandStatus) + { + case DAC960_V1_NormalCompletion: + DAC960_UserCritical("Consistency Check of Logical Drive %d " + "(/dev/rd/c%dd%d) Initiated\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber); + break; + case DAC960_V1_DependentDiskIsDead: + DAC960_UserCritical("Consistency Check of Logical Drive %d " + "(/dev/rd/c%dd%d) Failed - " + "Dependent Physical Device is DEAD\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber); + break; + case DAC960_V1_InvalidOrNonredundantLogicalDrive: + DAC960_UserCritical("Consistency Check of Logical Drive %d " + "(/dev/rd/c%dd%d) Failed - " + "Invalid or Nonredundant Logical Drive\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber); + break; + case DAC960_V1_RebuildOrCheckAlreadyInProgress: + DAC960_UserCritical("Consistency Check of Logical Drive %d " + "(/dev/rd/c%dd%d) Failed - Rebuild or " + "Consistency Check Already in Progress\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber); + break; + default: + DAC960_UserCritical("Consistency Check of Logical Drive %d " + "(/dev/rd/c%dd%d) Failed - " + "Unexpected Status %04X\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, Command->V1.CommandStatus); + break; + } + } + else if (strcmp(UserCommand, "cancel-rebuild") == 0 || + strcmp(UserCommand, "cancel-consistency-check") == 0) + { + /* + the OldRebuildRateConstant is never actually used + once its value is retrieved from the controller. + */ + unsigned char *OldRebuildRateConstant; + dma_addr_t OldRebuildRateConstantDMA; + + OldRebuildRateConstant = pci_alloc_consistent( Controller->PCIDevice, + sizeof(char), &OldRebuildRateConstantDMA); + if (OldRebuildRateConstant == NULL) { + DAC960_UserCritical("Cancellation of Rebuild or " + "Consistency Check Failed - " + "Out of Memory", + Controller); + goto failure; + } + CommandMailbox->Type3R.CommandOpcode = DAC960_V1_RebuildControl; + CommandMailbox->Type3R.RebuildRateConstant = 0xFF; + CommandMailbox->Type3R.BusAddress = OldRebuildRateConstantDMA; + DAC960_ExecuteCommand(Command); + switch (Command->V1.CommandStatus) + { + case DAC960_V1_NormalCompletion: + DAC960_UserCritical("Rebuild or Consistency Check Cancelled\n", + Controller); + break; + default: + DAC960_UserCritical("Cancellation of Rebuild or " + "Consistency Check Failed - " + "Unexpected Status %04X\n", + Controller, Command->V1.CommandStatus); + break; + } +failure: + pci_free_consistent(Controller->PCIDevice, sizeof(char), + OldRebuildRateConstant, OldRebuildRateConstantDMA); + } + else DAC960_UserCritical("Illegal User Command: '%s'\n", + Controller, UserCommand); + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_DeallocateCommand(Command); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return true; +} + + +/* + DAC960_V2_TranslatePhysicalDevice translates a Physical Device Channel and + TargetID into a Logical Device. It returns true on success and false + on failure. +*/ + +static boolean DAC960_V2_TranslatePhysicalDevice(DAC960_Command_T *Command, + unsigned char Channel, + unsigned char TargetID, + unsigned short + *LogicalDeviceNumber) +{ + DAC960_V2_CommandMailbox_T SavedCommandMailbox, *CommandMailbox; + DAC960_Controller_T *Controller = Command->Controller; + + CommandMailbox = &Command->V2.CommandMailbox; + memcpy(&SavedCommandMailbox, CommandMailbox, + sizeof(DAC960_V2_CommandMailbox_T)); + + CommandMailbox->PhysicalDeviceInfo.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->PhysicalDeviceInfo.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->PhysicalDeviceInfo.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->PhysicalDeviceInfo.DataTransferSize = + sizeof(DAC960_V2_PhysicalToLogicalDevice_T); + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.TargetID = TargetID; + CommandMailbox->PhysicalDeviceInfo.PhysicalDevice.Channel = Channel; + CommandMailbox->PhysicalDeviceInfo.IOCTL_Opcode = + DAC960_V2_TranslatePhysicalToLogicalDevice; + CommandMailbox->Common.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.PhysicalToLogicalDeviceDMA; + CommandMailbox->Common.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->Common.DataTransferSize; + + DAC960_ExecuteCommand(Command); + *LogicalDeviceNumber = Controller->V2.PhysicalToLogicalDevice->LogicalDeviceNumber; + + memcpy(CommandMailbox, &SavedCommandMailbox, + sizeof(DAC960_V2_CommandMailbox_T)); + return (Command->V2.CommandStatus == DAC960_V2_NormalCompletion); +} + + +/* + DAC960_V2_ExecuteUserCommand executes a User Command for DAC960 V2 Firmware + Controllers. +*/ + +static boolean DAC960_V2_ExecuteUserCommand(DAC960_Controller_T *Controller, + unsigned char *UserCommand) +{ + DAC960_Command_T *Command; + DAC960_V2_CommandMailbox_T *CommandMailbox; + unsigned long flags; + unsigned char Channel, TargetID, LogicalDriveNumber; + unsigned short LogicalDeviceNumber; + + spin_lock_irqsave(&Controller->queue_lock, flags); + while ((Command = DAC960_AllocateCommand(Controller)) == NULL) + DAC960_WaitForCommand(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + Controller->UserStatusLength = 0; + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox = &Command->V2.CommandMailbox; + CommandMailbox->Common.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->Common.CommandControlBits.DataTransferControllerToHost = true; + CommandMailbox->Common.CommandControlBits.NoAutoRequestSense = true; + if (strcmp(UserCommand, "flush-cache") == 0) + { + CommandMailbox->DeviceOperation.IOCTL_Opcode = DAC960_V2_PauseDevice; + CommandMailbox->DeviceOperation.OperationDevice = + DAC960_V2_RAID_Controller; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Cache Flush Completed\n", Controller); + } + else if (strncmp(UserCommand, "kill", 4) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[4], + &Channel, &TargetID) && + DAC960_V2_TranslatePhysicalDevice(Command, Channel, TargetID, + &LogicalDeviceNumber)) + { + CommandMailbox->SetDeviceState.LogicalDevice.LogicalDeviceNumber = + LogicalDeviceNumber; + CommandMailbox->SetDeviceState.IOCTL_Opcode = + DAC960_V2_SetDeviceState; + CommandMailbox->SetDeviceState.DeviceState.PhysicalDeviceState = + DAC960_V2_Device_Dead; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Kill of Physical Device %d:%d %s\n", + Controller, Channel, TargetID, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Succeeded" : "Failed")); + } + else if (strncmp(UserCommand, "make-online", 11) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[11], + &Channel, &TargetID) && + DAC960_V2_TranslatePhysicalDevice(Command, Channel, TargetID, + &LogicalDeviceNumber)) + { + CommandMailbox->SetDeviceState.LogicalDevice.LogicalDeviceNumber = + LogicalDeviceNumber; + CommandMailbox->SetDeviceState.IOCTL_Opcode = + DAC960_V2_SetDeviceState; + CommandMailbox->SetDeviceState.DeviceState.PhysicalDeviceState = + DAC960_V2_Device_Online; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Make Online of Physical Device %d:%d %s\n", + Controller, Channel, TargetID, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Succeeded" : "Failed")); + } + else if (strncmp(UserCommand, "make-standby", 12) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[12], + &Channel, &TargetID) && + DAC960_V2_TranslatePhysicalDevice(Command, Channel, TargetID, + &LogicalDeviceNumber)) + { + CommandMailbox->SetDeviceState.LogicalDevice.LogicalDeviceNumber = + LogicalDeviceNumber; + CommandMailbox->SetDeviceState.IOCTL_Opcode = + DAC960_V2_SetDeviceState; + CommandMailbox->SetDeviceState.DeviceState.PhysicalDeviceState = + DAC960_V2_Device_Standby; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Make Standby of Physical Device %d:%d %s\n", + Controller, Channel, TargetID, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Succeeded" : "Failed")); + } + else if (strncmp(UserCommand, "rebuild", 7) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[7], + &Channel, &TargetID) && + DAC960_V2_TranslatePhysicalDevice(Command, Channel, TargetID, + &LogicalDeviceNumber)) + { + CommandMailbox->LogicalDeviceInfo.LogicalDevice.LogicalDeviceNumber = + LogicalDeviceNumber; + CommandMailbox->LogicalDeviceInfo.IOCTL_Opcode = + DAC960_V2_RebuildDeviceStart; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Rebuild of Physical Device %d:%d %s\n", + Controller, Channel, TargetID, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Initiated" : "Not Initiated")); + } + else if (strncmp(UserCommand, "cancel-rebuild", 14) == 0 && + DAC960_ParsePhysicalDevice(Controller, &UserCommand[14], + &Channel, &TargetID) && + DAC960_V2_TranslatePhysicalDevice(Command, Channel, TargetID, + &LogicalDeviceNumber)) + { + CommandMailbox->LogicalDeviceInfo.LogicalDevice.LogicalDeviceNumber = + LogicalDeviceNumber; + CommandMailbox->LogicalDeviceInfo.IOCTL_Opcode = + DAC960_V2_RebuildDeviceStop; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Rebuild of Physical Device %d:%d %s\n", + Controller, Channel, TargetID, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Cancelled" : "Not Cancelled")); + } + else if (strncmp(UserCommand, "check-consistency", 17) == 0 && + DAC960_ParseLogicalDrive(Controller, &UserCommand[17], + &LogicalDriveNumber)) + { + CommandMailbox->ConsistencyCheck.LogicalDevice.LogicalDeviceNumber = + LogicalDriveNumber; + CommandMailbox->ConsistencyCheck.IOCTL_Opcode = + DAC960_V2_ConsistencyCheckStart; + CommandMailbox->ConsistencyCheck.RestoreConsistency = true; + CommandMailbox->ConsistencyCheck.InitializedAreaOnly = false; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Consistency Check of Logical Drive %d " + "(/dev/rd/c%dd%d) %s\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Initiated" : "Not Initiated")); + } + else if (strncmp(UserCommand, "cancel-consistency-check", 24) == 0 && + DAC960_ParseLogicalDrive(Controller, &UserCommand[24], + &LogicalDriveNumber)) + { + CommandMailbox->ConsistencyCheck.LogicalDevice.LogicalDeviceNumber = + LogicalDriveNumber; + CommandMailbox->ConsistencyCheck.IOCTL_Opcode = + DAC960_V2_ConsistencyCheckStop; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Consistency Check of Logical Drive %d " + "(/dev/rd/c%dd%d) %s\n", + Controller, LogicalDriveNumber, + Controller->ControllerNumber, + LogicalDriveNumber, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Cancelled" : "Not Cancelled")); + } + else if (strcmp(UserCommand, "perform-discovery") == 0) + { + CommandMailbox->Common.IOCTL_Opcode = DAC960_V2_StartDiscovery; + DAC960_ExecuteCommand(Command); + DAC960_UserCritical("Discovery %s\n", Controller, + (Command->V2.CommandStatus + == DAC960_V2_NormalCompletion + ? "Initiated" : "Not Initiated")); + if (Command->V2.CommandStatus == DAC960_V2_NormalCompletion) + { + CommandMailbox->ControllerInfo.CommandOpcode = DAC960_V2_IOCTL; + CommandMailbox->ControllerInfo.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->ControllerInfo.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->ControllerInfo.DataTransferSize = + sizeof(DAC960_V2_ControllerInfo_T); + CommandMailbox->ControllerInfo.ControllerNumber = 0; + CommandMailbox->ControllerInfo.IOCTL_Opcode = + DAC960_V2_GetControllerInfo; + /* + * How does this NOT race with the queued Monitoring + * usage of this structure? + */ + CommandMailbox->ControllerInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = + Controller->V2.NewControllerInformationDMA; + CommandMailbox->ControllerInfo.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->ControllerInfo.DataTransferSize; + DAC960_ExecuteCommand(Command); + while (Controller->V2.NewControllerInformation->PhysicalScanActive) + { + DAC960_ExecuteCommand(Command); + sleep_on_timeout(&Controller->CommandWaitQueue, HZ); + } + DAC960_UserCritical("Discovery Completed\n", Controller); + } + } + else if (strcmp(UserCommand, "suppress-enclosure-messages") == 0) + Controller->SuppressEnclosureMessages = true; + else DAC960_UserCritical("Illegal User Command: '%s'\n", + Controller, UserCommand); + + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_DeallocateCommand(Command); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + return true; +} + + +/* + DAC960_ProcReadStatus implements reading /proc/rd/status. +*/ + +static int DAC960_ProcReadStatus(char *Page, char **Start, off_t Offset, + int Count, int *EOF, void *Data) +{ + unsigned char *StatusMessage = "OK\n"; + int ControllerNumber, BytesAvailable; + for (ControllerNumber = 0; + ControllerNumber < DAC960_ControllerCount; + ControllerNumber++) + { + DAC960_Controller_T *Controller = DAC960_Controllers[ControllerNumber]; + if (Controller == NULL) continue; + if (Controller->MonitoringAlertMode) + { + StatusMessage = "ALERT\n"; + break; + } + } + BytesAvailable = strlen(StatusMessage) - Offset; + if (Count >= BytesAvailable) + { + Count = BytesAvailable; + *EOF = true; + } + if (Count <= 0) return 0; + *Start = Page; + memcpy(Page, &StatusMessage[Offset], Count); + return Count; +} + + +/* + DAC960_ProcReadInitialStatus implements reading /proc/rd/cN/initial_status. +*/ + +static int DAC960_ProcReadInitialStatus(char *Page, char **Start, off_t Offset, + int Count, int *EOF, void *Data) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) Data; + int BytesAvailable = Controller->InitialStatusLength - Offset; + if (Count >= BytesAvailable) + { + Count = BytesAvailable; + *EOF = true; + } + if (Count <= 0) return 0; + *Start = Page; + memcpy(Page, &Controller->CombinedStatusBuffer[Offset], Count); + return Count; +} + + +/* + DAC960_ProcReadCurrentStatus implements reading /proc/rd/cN/current_status. +*/ + +static int DAC960_ProcReadCurrentStatus(char *Page, char **Start, off_t Offset, + int Count, int *EOF, void *Data) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) Data; + unsigned char *StatusMessage = + "No Rebuild or Consistency Check in Progress\n"; + int ProgressMessageLength = strlen(StatusMessage); + int BytesAvailable; + if (jiffies != Controller->LastCurrentStatusTime) + { + Controller->CurrentStatusLength = 0; + DAC960_AnnounceDriver(Controller); + DAC960_ReportControllerConfiguration(Controller); + DAC960_ReportDeviceConfiguration(Controller); + if (Controller->ProgressBufferLength > 0) + ProgressMessageLength = Controller->ProgressBufferLength; + if (DAC960_CheckStatusBuffer(Controller, 2 + ProgressMessageLength)) + { + unsigned char *CurrentStatusBuffer = Controller->CurrentStatusBuffer; + CurrentStatusBuffer[Controller->CurrentStatusLength++] = ' '; + CurrentStatusBuffer[Controller->CurrentStatusLength++] = ' '; + if (Controller->ProgressBufferLength > 0) + strcpy(&CurrentStatusBuffer[Controller->CurrentStatusLength], + Controller->ProgressBuffer); + else + strcpy(&CurrentStatusBuffer[Controller->CurrentStatusLength], + StatusMessage); + Controller->CurrentStatusLength += ProgressMessageLength; + } + Controller->LastCurrentStatusTime = jiffies; + } + BytesAvailable = Controller->CurrentStatusLength - Offset; + if (Count >= BytesAvailable) + { + Count = BytesAvailable; + *EOF = true; + } + if (Count <= 0) return 0; + *Start = Page; + memcpy(Page, &Controller->CurrentStatusBuffer[Offset], Count); + return Count; +} + + +/* + DAC960_ProcReadUserCommand implements reading /proc/rd/cN/user_command. +*/ + +static int DAC960_ProcReadUserCommand(char *Page, char **Start, off_t Offset, + int Count, int *EOF, void *Data) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) Data; + int BytesAvailable = Controller->UserStatusLength - Offset; + if (Count >= BytesAvailable) + { + Count = BytesAvailable; + *EOF = true; + } + if (Count <= 0) return 0; + *Start = Page; + memcpy(Page, &Controller->UserStatusBuffer[Offset], Count); + return Count; +} + + +/* + DAC960_ProcWriteUserCommand implements writing /proc/rd/cN/user_command. +*/ + +static int DAC960_ProcWriteUserCommand(struct file *file, + const char __user *Buffer, + unsigned long Count, void *Data) +{ + DAC960_Controller_T *Controller = (DAC960_Controller_T *) Data; + unsigned char CommandBuffer[80]; + int Length; + if (Count > sizeof(CommandBuffer)-1) return -EINVAL; + if (copy_from_user(CommandBuffer, Buffer, Count)) return -EFAULT; + CommandBuffer[Count] = '\0'; + Length = strlen(CommandBuffer); + if (CommandBuffer[Length-1] == '\n') + CommandBuffer[--Length] = '\0'; + if (Controller->FirmwareType == DAC960_V1_Controller) + return (DAC960_V1_ExecuteUserCommand(Controller, CommandBuffer) + ? Count : -EBUSY); + else + return (DAC960_V2_ExecuteUserCommand(Controller, CommandBuffer) + ? Count : -EBUSY); +} + + +/* + DAC960_CreateProcEntries creates the /proc/rd/... entries for the + DAC960 Driver. +*/ + +static void DAC960_CreateProcEntries(DAC960_Controller_T *Controller) +{ + struct proc_dir_entry *StatusProcEntry; + struct proc_dir_entry *ControllerProcEntry; + struct proc_dir_entry *UserCommandProcEntry; + + if (DAC960_ProcDirectoryEntry == NULL) { + DAC960_ProcDirectoryEntry = proc_mkdir("rd", NULL); + StatusProcEntry = create_proc_read_entry("status", 0, + DAC960_ProcDirectoryEntry, + DAC960_ProcReadStatus, NULL); + } + + sprintf(Controller->ControllerName, "c%d", Controller->ControllerNumber); + ControllerProcEntry = proc_mkdir(Controller->ControllerName, + DAC960_ProcDirectoryEntry); + create_proc_read_entry("initial_status", 0, ControllerProcEntry, + DAC960_ProcReadInitialStatus, Controller); + create_proc_read_entry("current_status", 0, ControllerProcEntry, + DAC960_ProcReadCurrentStatus, Controller); + UserCommandProcEntry = + create_proc_read_entry("user_command", S_IWUSR | S_IRUSR, + ControllerProcEntry, DAC960_ProcReadUserCommand, + Controller); + UserCommandProcEntry->write_proc = DAC960_ProcWriteUserCommand; + Controller->ControllerProcEntry = ControllerProcEntry; +} + + +/* + DAC960_DestroyProcEntries destroys the /proc/rd/... entries for the + DAC960 Driver. +*/ + +static void DAC960_DestroyProcEntries(DAC960_Controller_T *Controller) +{ + if (Controller->ControllerProcEntry == NULL) + return; + remove_proc_entry("initial_status", Controller->ControllerProcEntry); + remove_proc_entry("current_status", Controller->ControllerProcEntry); + remove_proc_entry("user_command", Controller->ControllerProcEntry); + remove_proc_entry(Controller->ControllerName, DAC960_ProcDirectoryEntry); + Controller->ControllerProcEntry = NULL; +} + +#ifdef DAC960_GAM_MINOR + +/* + * DAC960_gam_ioctl is the ioctl function for performing RAID operations. +*/ + +static int DAC960_gam_ioctl(struct inode *inode, struct file *file, + unsigned int Request, unsigned long Argument) +{ + int ErrorCode = 0; + if (!capable(CAP_SYS_ADMIN)) return -EACCES; + switch (Request) + { + case DAC960_IOCTL_GET_CONTROLLER_COUNT: + return DAC960_ControllerCount; + case DAC960_IOCTL_GET_CONTROLLER_INFO: + { + DAC960_ControllerInfo_T __user *UserSpaceControllerInfo = + (DAC960_ControllerInfo_T __user *) Argument; + DAC960_ControllerInfo_T ControllerInfo; + DAC960_Controller_T *Controller; + int ControllerNumber; + if (UserSpaceControllerInfo == NULL) return -EINVAL; + ErrorCode = get_user(ControllerNumber, + &UserSpaceControllerInfo->ControllerNumber); + if (ErrorCode != 0) return ErrorCode; + if (ControllerNumber < 0 || + ControllerNumber > DAC960_ControllerCount - 1) + return -ENXIO; + Controller = DAC960_Controllers[ControllerNumber]; + if (Controller == NULL) return -ENXIO; + memset(&ControllerInfo, 0, sizeof(DAC960_ControllerInfo_T)); + ControllerInfo.ControllerNumber = ControllerNumber; + ControllerInfo.FirmwareType = Controller->FirmwareType; + ControllerInfo.Channels = Controller->Channels; + ControllerInfo.Targets = Controller->Targets; + ControllerInfo.PCI_Bus = Controller->Bus; + ControllerInfo.PCI_Device = Controller->Device; + ControllerInfo.PCI_Function = Controller->Function; + ControllerInfo.IRQ_Channel = Controller->IRQ_Channel; + ControllerInfo.PCI_Address = Controller->PCI_Address; + strcpy(ControllerInfo.ModelName, Controller->ModelName); + strcpy(ControllerInfo.FirmwareVersion, Controller->FirmwareVersion); + return (copy_to_user(UserSpaceControllerInfo, &ControllerInfo, + sizeof(DAC960_ControllerInfo_T)) ? -EFAULT : 0); + } + case DAC960_IOCTL_V1_EXECUTE_COMMAND: + { + DAC960_V1_UserCommand_T __user *UserSpaceUserCommand = + (DAC960_V1_UserCommand_T __user *) Argument; + DAC960_V1_UserCommand_T UserCommand; + DAC960_Controller_T *Controller; + DAC960_Command_T *Command = NULL; + DAC960_V1_CommandOpcode_T CommandOpcode; + DAC960_V1_CommandStatus_T CommandStatus; + DAC960_V1_DCDB_T DCDB; + DAC960_V1_DCDB_T *DCDB_IOBUF = NULL; + dma_addr_t DCDB_IOBUFDMA; + unsigned long flags; + int ControllerNumber, DataTransferLength; + unsigned char *DataTransferBuffer = NULL; + dma_addr_t DataTransferBufferDMA; + if (UserSpaceUserCommand == NULL) return -EINVAL; + if (copy_from_user(&UserCommand, UserSpaceUserCommand, + sizeof(DAC960_V1_UserCommand_T))) { + ErrorCode = -EFAULT; + goto Failure1a; + } + ControllerNumber = UserCommand.ControllerNumber; + if (ControllerNumber < 0 || + ControllerNumber > DAC960_ControllerCount - 1) + return -ENXIO; + Controller = DAC960_Controllers[ControllerNumber]; + if (Controller == NULL) return -ENXIO; + if (Controller->FirmwareType != DAC960_V1_Controller) return -EINVAL; + CommandOpcode = UserCommand.CommandMailbox.Common.CommandOpcode; + DataTransferLength = UserCommand.DataTransferLength; + if (CommandOpcode & 0x80) return -EINVAL; + if (CommandOpcode == DAC960_V1_DCDB) + { + if (copy_from_user(&DCDB, UserCommand.DCDB, + sizeof(DAC960_V1_DCDB_T))) { + ErrorCode = -EFAULT; + goto Failure1a; + } + if (DCDB.Channel >= DAC960_V1_MaxChannels) return -EINVAL; + if (!((DataTransferLength == 0 && + DCDB.Direction + == DAC960_V1_DCDB_NoDataTransfer) || + (DataTransferLength > 0 && + DCDB.Direction + == DAC960_V1_DCDB_DataTransferDeviceToSystem) || + (DataTransferLength < 0 && + DCDB.Direction + == DAC960_V1_DCDB_DataTransferSystemToDevice))) + return -EINVAL; + if (((DCDB.TransferLengthHigh4 << 16) | DCDB.TransferLength) + != abs(DataTransferLength)) + return -EINVAL; + DCDB_IOBUF = pci_alloc_consistent(Controller->PCIDevice, + sizeof(DAC960_V1_DCDB_T), &DCDB_IOBUFDMA); + if (DCDB_IOBUF == NULL) + return -ENOMEM; + } + if (DataTransferLength > 0) + { + DataTransferBuffer = pci_alloc_consistent(Controller->PCIDevice, + DataTransferLength, &DataTransferBufferDMA); + if (DataTransferBuffer == NULL) { + ErrorCode = -ENOMEM; + goto Failure1; + } + memset(DataTransferBuffer, 0, DataTransferLength); + } + else if (DataTransferLength < 0) + { + DataTransferBuffer = pci_alloc_consistent(Controller->PCIDevice, + -DataTransferLength, &DataTransferBufferDMA); + if (DataTransferBuffer == NULL) { + ErrorCode = -ENOMEM; + goto Failure1; + } + if (copy_from_user(DataTransferBuffer, + UserCommand.DataTransferBuffer, + -DataTransferLength)) { + ErrorCode = -EFAULT; + goto Failure1; + } + } + if (CommandOpcode == DAC960_V1_DCDB) + { + spin_lock_irqsave(&Controller->queue_lock, flags); + while ((Command = DAC960_AllocateCommand(Controller)) == NULL) + DAC960_WaitForCommand(Controller); + while (Controller->V1.DirectCommandActive[DCDB.Channel] + [DCDB.TargetID]) + { + spin_unlock_irq(&Controller->queue_lock); + __wait_event(Controller->CommandWaitQueue, + !Controller->V1.DirectCommandActive + [DCDB.Channel][DCDB.TargetID]); + spin_lock_irq(&Controller->queue_lock); + } + Controller->V1.DirectCommandActive[DCDB.Channel] + [DCDB.TargetID] = true; + spin_unlock_irqrestore(&Controller->queue_lock, flags); + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + memcpy(&Command->V1.CommandMailbox, &UserCommand.CommandMailbox, + sizeof(DAC960_V1_CommandMailbox_T)); + Command->V1.CommandMailbox.Type3.BusAddress = DCDB_IOBUFDMA; + DCDB.BusAddress = DataTransferBufferDMA; + memcpy(DCDB_IOBUF, &DCDB, sizeof(DAC960_V1_DCDB_T)); + } + else + { + spin_lock_irqsave(&Controller->queue_lock, flags); + while ((Command = DAC960_AllocateCommand(Controller)) == NULL) + DAC960_WaitForCommand(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + DAC960_V1_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + memcpy(&Command->V1.CommandMailbox, &UserCommand.CommandMailbox, + sizeof(DAC960_V1_CommandMailbox_T)); + if (DataTransferBuffer != NULL) + Command->V1.CommandMailbox.Type3.BusAddress = + DataTransferBufferDMA; + } + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V1.CommandStatus; + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_DeallocateCommand(Command); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + if (DataTransferLength > 0) + { + if (copy_to_user(UserCommand.DataTransferBuffer, + DataTransferBuffer, DataTransferLength)) { + ErrorCode = -EFAULT; + goto Failure1; + } + } + if (CommandOpcode == DAC960_V1_DCDB) + { + /* + I don't believe Target or Channel in the DCDB_IOBUF + should be any different from the contents of DCDB. + */ + Controller->V1.DirectCommandActive[DCDB.Channel] + [DCDB.TargetID] = false; + if (copy_to_user(UserCommand.DCDB, DCDB_IOBUF, + sizeof(DAC960_V1_DCDB_T))) { + ErrorCode = -EFAULT; + goto Failure1; + } + } + ErrorCode = CommandStatus; + Failure1: + if (DataTransferBuffer != NULL) + pci_free_consistent(Controller->PCIDevice, abs(DataTransferLength), + DataTransferBuffer, DataTransferBufferDMA); + if (DCDB_IOBUF != NULL) + pci_free_consistent(Controller->PCIDevice, sizeof(DAC960_V1_DCDB_T), + DCDB_IOBUF, DCDB_IOBUFDMA); + Failure1a: + return ErrorCode; + } + case DAC960_IOCTL_V2_EXECUTE_COMMAND: + { + DAC960_V2_UserCommand_T __user *UserSpaceUserCommand = + (DAC960_V2_UserCommand_T __user *) Argument; + DAC960_V2_UserCommand_T UserCommand; + DAC960_Controller_T *Controller; + DAC960_Command_T *Command = NULL; + DAC960_V2_CommandMailbox_T *CommandMailbox; + DAC960_V2_CommandStatus_T CommandStatus; + unsigned long flags; + int ControllerNumber, DataTransferLength; + int DataTransferResidue, RequestSenseLength; + unsigned char *DataTransferBuffer = NULL; + dma_addr_t DataTransferBufferDMA; + unsigned char *RequestSenseBuffer = NULL; + dma_addr_t RequestSenseBufferDMA; + if (UserSpaceUserCommand == NULL) return -EINVAL; + if (copy_from_user(&UserCommand, UserSpaceUserCommand, + sizeof(DAC960_V2_UserCommand_T))) { + ErrorCode = -EFAULT; + goto Failure2a; + } + ControllerNumber = UserCommand.ControllerNumber; + if (ControllerNumber < 0 || + ControllerNumber > DAC960_ControllerCount - 1) + return -ENXIO; + Controller = DAC960_Controllers[ControllerNumber]; + if (Controller == NULL) return -ENXIO; + if (Controller->FirmwareType != DAC960_V2_Controller) return -EINVAL; + DataTransferLength = UserCommand.DataTransferLength; + if (DataTransferLength > 0) + { + DataTransferBuffer = pci_alloc_consistent(Controller->PCIDevice, + DataTransferLength, &DataTransferBufferDMA); + if (DataTransferBuffer == NULL) return -ENOMEM; + memset(DataTransferBuffer, 0, DataTransferLength); + } + else if (DataTransferLength < 0) + { + DataTransferBuffer = pci_alloc_consistent(Controller->PCIDevice, + -DataTransferLength, &DataTransferBufferDMA); + if (DataTransferBuffer == NULL) return -ENOMEM; + if (copy_from_user(DataTransferBuffer, + UserCommand.DataTransferBuffer, + -DataTransferLength)) { + ErrorCode = -EFAULT; + goto Failure2; + } + } + RequestSenseLength = UserCommand.RequestSenseLength; + if (RequestSenseLength > 0) + { + RequestSenseBuffer = pci_alloc_consistent(Controller->PCIDevice, + RequestSenseLength, &RequestSenseBufferDMA); + if (RequestSenseBuffer == NULL) + { + ErrorCode = -ENOMEM; + goto Failure2; + } + memset(RequestSenseBuffer, 0, RequestSenseLength); + } + spin_lock_irqsave(&Controller->queue_lock, flags); + while ((Command = DAC960_AllocateCommand(Controller)) == NULL) + DAC960_WaitForCommand(Controller); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + DAC960_V2_ClearCommand(Command); + Command->CommandType = DAC960_ImmediateCommand; + CommandMailbox = &Command->V2.CommandMailbox; + memcpy(CommandMailbox, &UserCommand.CommandMailbox, + sizeof(DAC960_V2_CommandMailbox_T)); + CommandMailbox->Common.CommandControlBits + .AdditionalScatterGatherListMemory = false; + CommandMailbox->Common.CommandControlBits + .NoAutoRequestSense = true; + CommandMailbox->Common.DataTransferSize = 0; + CommandMailbox->Common.DataTransferPageNumber = 0; + memset(&CommandMailbox->Common.DataTransferMemoryAddress, 0, + sizeof(DAC960_V2_DataTransferMemoryAddress_T)); + if (DataTransferLength != 0) + { + if (DataTransferLength > 0) + { + CommandMailbox->Common.CommandControlBits + .DataTransferControllerToHost = true; + CommandMailbox->Common.DataTransferSize = DataTransferLength; + } + else + { + CommandMailbox->Common.CommandControlBits + .DataTransferControllerToHost = false; + CommandMailbox->Common.DataTransferSize = -DataTransferLength; + } + CommandMailbox->Common.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentDataPointer = DataTransferBufferDMA; + CommandMailbox->Common.DataTransferMemoryAddress + .ScatterGatherSegments[0] + .SegmentByteCount = + CommandMailbox->Common.DataTransferSize; + } + if (RequestSenseLength > 0) + { + CommandMailbox->Common.CommandControlBits + .NoAutoRequestSense = false; + CommandMailbox->Common.RequestSenseSize = RequestSenseLength; + CommandMailbox->Common.RequestSenseBusAddress = + RequestSenseBufferDMA; + } + DAC960_ExecuteCommand(Command); + CommandStatus = Command->V2.CommandStatus; + RequestSenseLength = Command->V2.RequestSenseLength; + DataTransferResidue = Command->V2.DataTransferResidue; + spin_lock_irqsave(&Controller->queue_lock, flags); + DAC960_DeallocateCommand(Command); + spin_unlock_irqrestore(&Controller->queue_lock, flags); + if (RequestSenseLength > UserCommand.RequestSenseLength) + RequestSenseLength = UserCommand.RequestSenseLength; + if (copy_to_user(&UserSpaceUserCommand->DataTransferLength, + &DataTransferResidue, + sizeof(DataTransferResidue))) { + ErrorCode = -EFAULT; + goto Failure2; + } + if (copy_to_user(&UserSpaceUserCommand->RequestSenseLength, + &RequestSenseLength, sizeof(RequestSenseLength))) { + ErrorCode = -EFAULT; + goto Failure2; + } + if (DataTransferLength > 0) + { + if (copy_to_user(UserCommand.DataTransferBuffer, + DataTransferBuffer, DataTransferLength)) { + ErrorCode = -EFAULT; + goto Failure2; + } + } + if (RequestSenseLength > 0) + { + if (copy_to_user(UserCommand.RequestSenseBuffer, + RequestSenseBuffer, RequestSenseLength)) { + ErrorCode = -EFAULT; + goto Failure2; + } + } + ErrorCode = CommandStatus; + Failure2: + pci_free_consistent(Controller->PCIDevice, abs(DataTransferLength), + DataTransferBuffer, DataTransferBufferDMA); + if (RequestSenseBuffer != NULL) + pci_free_consistent(Controller->PCIDevice, RequestSenseLength, + RequestSenseBuffer, RequestSenseBufferDMA); + Failure2a: + return ErrorCode; + } + case DAC960_IOCTL_V2_GET_HEALTH_STATUS: + { + DAC960_V2_GetHealthStatus_T __user *UserSpaceGetHealthStatus = + (DAC960_V2_GetHealthStatus_T __user *) Argument; + DAC960_V2_GetHealthStatus_T GetHealthStatus; + DAC960_V2_HealthStatusBuffer_T HealthStatusBuffer; + DAC960_Controller_T *Controller; + int ControllerNumber; + if (UserSpaceGetHealthStatus == NULL) return -EINVAL; + if (copy_from_user(&GetHealthStatus, UserSpaceGetHealthStatus, + sizeof(DAC960_V2_GetHealthStatus_T))) + return -EFAULT; + ControllerNumber = GetHealthStatus.ControllerNumber; + if (ControllerNumber < 0 || + ControllerNumber > DAC960_ControllerCount - 1) + return -ENXIO; + Controller = DAC960_Controllers[ControllerNumber]; + if (Controller == NULL) return -ENXIO; + if (Controller->FirmwareType != DAC960_V2_Controller) return -EINVAL; + if (copy_from_user(&HealthStatusBuffer, + GetHealthStatus.HealthStatusBuffer, + sizeof(DAC960_V2_HealthStatusBuffer_T))) + return -EFAULT; + while (Controller->V2.HealthStatusBuffer->StatusChangeCounter + == HealthStatusBuffer.StatusChangeCounter && + Controller->V2.HealthStatusBuffer->NextEventSequenceNumber + == HealthStatusBuffer.NextEventSequenceNumber) + { + interruptible_sleep_on_timeout(&Controller->HealthStatusWaitQueue, + DAC960_MonitoringTimerInterval); + if (signal_pending(current)) return -EINTR; + } + if (copy_to_user(GetHealthStatus.HealthStatusBuffer, + Controller->V2.HealthStatusBuffer, + sizeof(DAC960_V2_HealthStatusBuffer_T))) + return -EFAULT; + return 0; + } + } + return -EINVAL; +} + +static struct file_operations DAC960_gam_fops = { + .owner = THIS_MODULE, + .ioctl = DAC960_gam_ioctl +}; + +static struct miscdevice DAC960_gam_dev = { + DAC960_GAM_MINOR, + "dac960_gam", + &DAC960_gam_fops +}; + +static int DAC960_gam_init(void) +{ + int ret; + + ret = misc_register(&DAC960_gam_dev); + if (ret) + printk(KERN_ERR "DAC960_gam: can't misc_register on minor %d\n", DAC960_GAM_MINOR); + return ret; +} + +static void DAC960_gam_cleanup(void) +{ + misc_deregister(&DAC960_gam_dev); +} + +#endif /* DAC960_GAM_MINOR */ + +static struct DAC960_privdata DAC960_BA_privdata = { + .HardwareType = DAC960_BA_Controller, + .FirmwareType = DAC960_V2_Controller, + .InterruptHandler = DAC960_BA_InterruptHandler, + .MemoryWindowSize = DAC960_BA_RegisterWindowSize, +}; + +static struct DAC960_privdata DAC960_LP_privdata = { + .HardwareType = DAC960_LP_Controller, + .FirmwareType = DAC960_LP_Controller, + .InterruptHandler = DAC960_LP_InterruptHandler, + .MemoryWindowSize = DAC960_LP_RegisterWindowSize, +}; + +static struct DAC960_privdata DAC960_LA_privdata = { + .HardwareType = DAC960_LA_Controller, + .FirmwareType = DAC960_V1_Controller, + .InterruptHandler = DAC960_LA_InterruptHandler, + .MemoryWindowSize = DAC960_LA_RegisterWindowSize, +}; + +static struct DAC960_privdata DAC960_PG_privdata = { + .HardwareType = DAC960_PG_Controller, + .FirmwareType = DAC960_V1_Controller, + .InterruptHandler = DAC960_PG_InterruptHandler, + .MemoryWindowSize = DAC960_PG_RegisterWindowSize, +}; + +static struct DAC960_privdata DAC960_PD_privdata = { + .HardwareType = DAC960_PD_Controller, + .FirmwareType = DAC960_V1_Controller, + .InterruptHandler = DAC960_PD_InterruptHandler, + .MemoryWindowSize = DAC960_PD_RegisterWindowSize, +}; + +static struct DAC960_privdata DAC960_P_privdata = { + .HardwareType = DAC960_P_Controller, + .FirmwareType = DAC960_V1_Controller, + .InterruptHandler = DAC960_P_InterruptHandler, + .MemoryWindowSize = DAC960_PD_RegisterWindowSize, +}; + +static struct pci_device_id DAC960_id_table[] = { + { + .vendor = PCI_VENDOR_ID_MYLEX, + .device = PCI_DEVICE_ID_MYLEX_DAC960_BA, + .subvendor = PCI_ANY_ID, + .subdevice = PCI_ANY_ID, + .driver_data = (unsigned long) &DAC960_BA_privdata, + }, + { + .vendor = PCI_VENDOR_ID_MYLEX, + .device = PCI_DEVICE_ID_MYLEX_DAC960_LP, + .subvendor = PCI_ANY_ID, + .subdevice = PCI_ANY_ID, + .driver_data = (unsigned long) &DAC960_LP_privdata, + }, + { + .vendor = PCI_VENDOR_ID_DEC, + .device = PCI_DEVICE_ID_DEC_21285, + .subvendor = PCI_VENDOR_ID_MYLEX, + .subdevice = PCI_DEVICE_ID_MYLEX_DAC960_LA, + .driver_data = (unsigned long) &DAC960_LA_privdata, + }, + { + .vendor = PCI_VENDOR_ID_MYLEX, + .device = PCI_DEVICE_ID_MYLEX_DAC960_PG, + .subvendor = PCI_ANY_ID, + .subdevice = PCI_ANY_ID, + .driver_data = (unsigned long) &DAC960_PG_privdata, + }, + { + .vendor = PCI_VENDOR_ID_MYLEX, + .device = PCI_DEVICE_ID_MYLEX_DAC960_PD, + .subvendor = PCI_ANY_ID, + .subdevice = PCI_ANY_ID, + .driver_data = (unsigned long) &DAC960_PD_privdata, + }, + { + .vendor = PCI_VENDOR_ID_MYLEX, + .device = PCI_DEVICE_ID_MYLEX_DAC960_P, + .subvendor = PCI_ANY_ID, + .subdevice = PCI_ANY_ID, + .driver_data = (unsigned long) &DAC960_P_privdata, + }, + {0, }, +}; + +MODULE_DEVICE_TABLE(pci, DAC960_id_table); + +static struct pci_driver DAC960_pci_driver = { + .name = "DAC960", + .id_table = DAC960_id_table, + .probe = DAC960_Probe, + .remove = DAC960_Remove, +}; + +static int DAC960_init_module(void) +{ + int ret; + + ret = pci_module_init(&DAC960_pci_driver); +#ifdef DAC960_GAM_MINOR + if (!ret) + DAC960_gam_init(); +#endif + return ret; +} + +static void DAC960_cleanup_module(void) +{ + int i; + +#ifdef DAC960_GAM_MINOR + DAC960_gam_cleanup(); +#endif + + for (i = 0; i < DAC960_ControllerCount; i++) { + DAC960_Controller_T *Controller = DAC960_Controllers[i]; + if (Controller == NULL) + continue; + DAC960_FinalizeController(Controller); + } + if (DAC960_ProcDirectoryEntry != NULL) { + remove_proc_entry("rd/status", NULL); + remove_proc_entry("rd", NULL); + } + DAC960_ControllerCount = 0; + pci_unregister_driver(&DAC960_pci_driver); +} + +module_init(DAC960_init_module); +module_exit(DAC960_cleanup_module); + +MODULE_LICENSE("GPL"); diff --git a/drivers/block/DAC960.h b/drivers/block/DAC960.h new file mode 100644 index 00000000000..d5e8e7190c9 --- /dev/null +++ b/drivers/block/DAC960.h @@ -0,0 +1,4114 @@ +/* + + Linux Driver for Mylex DAC960/AcceleRAID/eXtremeRAID PCI RAID Controllers + + Copyright 1998-2001 by Leonard N. Zubkoff <lnz@dandelion.com> + + This program is free software; you may redistribute and/or modify it under + the terms of the GNU General Public License Version 2 as published by the + Free Software Foundation. + + This program is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY, without even the implied warranty of MERCHANTABILITY + or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License + for complete details. + + The author respectfully requests that any modifications to this software be + sent directly to him for evaluation and testing. + +*/ + + +/* + Define the maximum number of DAC960 Controllers supported by this driver. +*/ + +#define DAC960_MaxControllers 8 + + +/* + Define the maximum number of Controller Channels supported by DAC960 + V1 and V2 Firmware Controllers. +*/ + +#define DAC960_V1_MaxChannels 3 +#define DAC960_V2_MaxChannels 4 + + +/* + Define the maximum number of Targets per Channel supported by DAC960 + V1 and V2 Firmware Controllers. +*/ + +#define DAC960_V1_MaxTargets 16 +#define DAC960_V2_MaxTargets 128 + + +/* + Define the maximum number of Logical Drives supported by DAC960 + V1 and V2 Firmware Controllers. +*/ + +#define DAC960_MaxLogicalDrives 32 + + +/* + Define the maximum number of Physical Devices supported by DAC960 + V1 and V2 Firmware Controllers. +*/ + +#define DAC960_V1_MaxPhysicalDevices 45 +#define DAC960_V2_MaxPhysicalDevices 272 + +/* + Define the pci dma mask supported by DAC960 V1 and V2 Firmware Controlers + */ + +#define DAC690_V1_PciDmaMask 0xffffffff +#define DAC690_V2_PciDmaMask 0xffffffffffffffffULL + +/* + Define a Boolean data type. +*/ + +typedef enum { false, true } __attribute__ ((packed)) boolean; + + +/* + Define a 32/64 bit I/O Address data type. +*/ + +typedef unsigned long DAC960_IO_Address_T; + + +/* + Define a 32/64 bit PCI Bus Address data type. +*/ + +typedef unsigned long DAC960_PCI_Address_T; + + +/* + Define a 32 bit Bus Address data type. +*/ + +typedef unsigned int DAC960_BusAddress32_T; + + +/* + Define a 64 bit Bus Address data type. +*/ + +typedef unsigned long long DAC960_BusAddress64_T; + + +/* + Define a 32 bit Byte Count data type. +*/ + +typedef unsigned int DAC960_ByteCount32_T; + + +/* + Define a 64 bit Byte Count data type. +*/ + +typedef unsigned long long DAC960_ByteCount64_T; + + +/* + dma_loaf is used by helper routines to divide a region of + dma mapped memory into smaller pieces, where those pieces + are not of uniform size. + */ + +struct dma_loaf { + void *cpu_base; + dma_addr_t dma_base; + size_t length; + void *cpu_free; + dma_addr_t dma_free; +}; + +/* + Define the SCSI INQUIRY Standard Data structure. +*/ + +typedef struct DAC960_SCSI_Inquiry +{ + unsigned char PeripheralDeviceType:5; /* Byte 0 Bits 0-4 */ + unsigned char PeripheralQualifier:3; /* Byte 0 Bits 5-7 */ + unsigned char DeviceTypeModifier:7; /* Byte 1 Bits 0-6 */ + boolean RMB:1; /* Byte 1 Bit 7 */ + unsigned char ANSI_ApprovedVersion:3; /* Byte 2 Bits 0-2 */ + unsigned char ECMA_Version:3; /* Byte 2 Bits 3-5 */ + unsigned char ISO_Version:2; /* Byte 2 Bits 6-7 */ + unsigned char ResponseDataFormat:4; /* Byte 3 Bits 0-3 */ + unsigned char :2; /* Byte 3 Bits 4-5 */ + boolean TrmIOP:1; /* Byte 3 Bit 6 */ + boolean AENC:1; /* Byte 3 Bit 7 */ + unsigned char AdditionalLength; /* Byte 4 */ + unsigned char :8; /* Byte 5 */ + unsigned char :8; /* Byte 6 */ + boolean SftRe:1; /* Byte 7 Bit 0 */ + boolean CmdQue:1; /* Byte 7 Bit 1 */ + boolean :1; /* Byte 7 Bit 2 */ + boolean Linked:1; /* Byte 7 Bit 3 */ + boolean Sync:1; /* Byte 7 Bit 4 */ + boolean WBus16:1; /* Byte 7 Bit 5 */ + boolean WBus32:1; /* Byte 7 Bit 6 */ + boolean RelAdr:1; /* Byte 7 Bit 7 */ + unsigned char VendorIdentification[8]; /* Bytes 8-15 */ + unsigned char ProductIdentification[16]; /* Bytes 16-31 */ + unsigned char ProductRevisionLevel[4]; /* Bytes 32-35 */ +} +DAC960_SCSI_Inquiry_T; + + +/* + Define the SCSI INQUIRY Unit Serial Number structure. +*/ + +typedef struct DAC960_SCSI_Inquiry_UnitSerialNumber +{ + unsigned char PeripheralDeviceType:5; /* Byte 0 Bits 0-4 */ + unsigned char PeripheralQualifier:3; /* Byte 0 Bits 5-7 */ + unsigned char PageCode; /* Byte 1 */ + unsigned char :8; /* Byte 2 */ + unsigned char PageLength; /* Byte 3 */ + unsigned char ProductSerialNumber[28]; /* Bytes 4-31 */ +} +DAC960_SCSI_Inquiry_UnitSerialNumber_T; + + +/* + Define the SCSI REQUEST SENSE Sense Key type. +*/ + +typedef enum +{ + DAC960_SenseKey_NoSense = 0x0, + DAC960_SenseKey_RecoveredError = 0x1, + DAC960_SenseKey_NotReady = 0x2, + DAC960_SenseKey_MediumError = 0x3, + DAC960_SenseKey_HardwareError = 0x4, + DAC960_SenseKey_IllegalRequest = 0x5, + DAC960_SenseKey_UnitAttention = 0x6, + DAC960_SenseKey_DataProtect = 0x7, + DAC960_SenseKey_BlankCheck = 0x8, + DAC960_SenseKey_VendorSpecific = 0x9, + DAC960_SenseKey_CopyAborted = 0xA, + DAC960_SenseKey_AbortedCommand = 0xB, + DAC960_SenseKey_Equal = 0xC, + DAC960_SenseKey_VolumeOverflow = 0xD, + DAC960_SenseKey_Miscompare = 0xE, + DAC960_SenseKey_Reserved = 0xF +} +__attribute__ ((packed)) +DAC960_SCSI_RequestSenseKey_T; + + +/* + Define the SCSI REQUEST SENSE structure. +*/ + +typedef struct DAC960_SCSI_RequestSense +{ + unsigned char ErrorCode:7; /* Byte 0 Bits 0-6 */ + boolean Valid:1; /* Byte 0 Bit 7 */ + unsigned char SegmentNumber; /* Byte 1 */ + DAC960_SCSI_RequestSenseKey_T SenseKey:4; /* Byte 2 Bits 0-3 */ + unsigned char :1; /* Byte 2 Bit 4 */ + boolean ILI:1; /* Byte 2 Bit 5 */ + boolean EOM:1; /* Byte 2 Bit 6 */ + boolean Filemark:1; /* Byte 2 Bit 7 */ + unsigned char Information[4]; /* Bytes 3-6 */ + unsigned char AdditionalSenseLength; /* Byte 7 */ + unsigned char CommandSpecificInformation[4]; /* Bytes 8-11 */ + unsigned char AdditionalSenseCode; /* Byte 12 */ + unsigned char AdditionalSenseCodeQualifier; /* Byte 13 */ +} +DAC960_SCSI_RequestSense_T; + + +/* + Define the DAC960 V1 Firmware Command Opcodes. +*/ + +typedef enum +{ + /* I/O Commands */ + DAC960_V1_ReadExtended = 0x33, + DAC960_V1_WriteExtended = 0x34, + DAC960_V1_ReadAheadExtended = 0x35, + DAC960_V1_ReadExtendedWithScatterGather = 0xB3, + DAC960_V1_WriteExtendedWithScatterGather = 0xB4, + DAC960_V1_Read = 0x36, + DAC960_V1_ReadWithScatterGather = 0xB6, + DAC960_V1_Write = 0x37, + DAC960_V1_WriteWithScatterGather = 0xB7, + DAC960_V1_DCDB = 0x04, + DAC960_V1_DCDBWithScatterGather = 0x84, + DAC960_V1_Flush = 0x0A, + /* Controller Status Related Commands */ + DAC960_V1_Enquiry = 0x53, + DAC960_V1_Enquiry2 = 0x1C, + DAC960_V1_GetLogicalDriveElement = 0x55, + DAC960_V1_GetLogicalDriveInformation = 0x19, + DAC960_V1_IOPortRead = 0x39, + DAC960_V1_IOPortWrite = 0x3A, + DAC960_V1_GetSDStats = 0x3E, + DAC960_V1_GetPDStats = 0x3F, + DAC960_V1_PerformEventLogOperation = 0x72, + /* Device Related Commands */ + DAC960_V1_StartDevice = 0x10, + DAC960_V1_GetDeviceState = 0x50, + DAC960_V1_StopChannel = 0x13, + DAC960_V1_StartChannel = 0x12, + DAC960_V1_ResetChannel = 0x1A, + /* Commands Associated with Data Consistency and Errors */ + DAC960_V1_Rebuild = 0x09, + DAC960_V1_RebuildAsync = 0x16, + DAC960_V1_CheckConsistency = 0x0F, + DAC960_V1_CheckConsistencyAsync = 0x1E, + DAC960_V1_RebuildStat = 0x0C, + DAC960_V1_GetRebuildProgress = 0x27, + DAC960_V1_RebuildControl = 0x1F, + DAC960_V1_ReadBadBlockTable = 0x0B, + DAC960_V1_ReadBadDataTable = 0x25, + DAC960_V1_ClearBadDataTable = 0x26, + DAC960_V1_GetErrorTable = 0x17, + DAC960_V1_AddCapacityAsync = 0x2A, + DAC960_V1_BackgroundInitializationControl = 0x2B, + /* Configuration Related Commands */ + DAC960_V1_ReadConfig2 = 0x3D, + DAC960_V1_WriteConfig2 = 0x3C, + DAC960_V1_ReadConfigurationOnDisk = 0x4A, + DAC960_V1_WriteConfigurationOnDisk = 0x4B, + DAC960_V1_ReadConfiguration = 0x4E, + DAC960_V1_ReadBackupConfiguration = 0x4D, + DAC960_V1_WriteConfiguration = 0x4F, + DAC960_V1_AddConfiguration = 0x4C, + DAC960_V1_ReadConfigurationLabel = 0x48, + DAC960_V1_WriteConfigurationLabel = 0x49, + /* Firmware Upgrade Related Commands */ + DAC960_V1_LoadImage = 0x20, + DAC960_V1_StoreImage = 0x21, + DAC960_V1_ProgramImage = 0x22, + /* Diagnostic Commands */ + DAC960_V1_SetDiagnosticMode = 0x31, + DAC960_V1_RunDiagnostic = 0x32, + /* Subsystem Service Commands */ + DAC960_V1_GetSubsystemData = 0x70, + DAC960_V1_SetSubsystemParameters = 0x71, + /* Version 2.xx Firmware Commands */ + DAC960_V1_Enquiry_Old = 0x05, + DAC960_V1_GetDeviceState_Old = 0x14, + DAC960_V1_Read_Old = 0x02, + DAC960_V1_Write_Old = 0x03, + DAC960_V1_ReadWithScatterGather_Old = 0x82, + DAC960_V1_WriteWithScatterGather_Old = 0x83 +} +__attribute__ ((packed)) +DAC960_V1_CommandOpcode_T; + + +/* + Define the DAC960 V1 Firmware Command Identifier type. +*/ + +typedef unsigned char DAC960_V1_CommandIdentifier_T; + + +/* + Define the DAC960 V1 Firmware Command Status Codes. +*/ + +#define DAC960_V1_NormalCompletion 0x0000 /* Common */ +#define DAC960_V1_CheckConditionReceived 0x0002 /* Common */ +#define DAC960_V1_NoDeviceAtAddress 0x0102 /* Common */ +#define DAC960_V1_InvalidDeviceAddress 0x0105 /* Common */ +#define DAC960_V1_InvalidParameter 0x0105 /* Common */ +#define DAC960_V1_IrrecoverableDataError 0x0001 /* I/O */ +#define DAC960_V1_LogicalDriveNonexistentOrOffline 0x0002 /* I/O */ +#define DAC960_V1_AccessBeyondEndOfLogicalDrive 0x0105 /* I/O */ +#define DAC960_V1_BadDataEncountered 0x010C /* I/O */ +#define DAC960_V1_DeviceBusy 0x0008 /* DCDB */ +#define DAC960_V1_DeviceNonresponsive 0x000E /* DCDB */ +#define DAC960_V1_CommandTerminatedAbnormally 0x000F /* DCDB */ +#define DAC960_V1_UnableToStartDevice 0x0002 /* Device */ +#define DAC960_V1_InvalidChannelOrTargetOrModifier 0x0105 /* Device */ +#define DAC960_V1_ChannelBusy 0x0106 /* Device */ +#define DAC960_V1_ChannelNotStopped 0x0002 /* Device */ +#define DAC960_V1_AttemptToRebuildOnlineDrive 0x0002 /* Consistency */ +#define DAC960_V1_RebuildBadBlocksEncountered 0x0003 /* Consistency */ +#define DAC960_V1_NewDiskFailedDuringRebuild 0x0004 /* Consistency */ +#define DAC960_V1_RebuildOrCheckAlreadyInProgress 0x0106 /* Consistency */ +#define DAC960_V1_DependentDiskIsDead 0x0002 /* Consistency */ +#define DAC960_V1_InconsistentBlocksFound 0x0003 /* Consistency */ +#define DAC960_V1_InvalidOrNonredundantLogicalDrive 0x0105 /* Consistency */ +#define DAC960_V1_NoRebuildOrCheckInProgress 0x0105 /* Consistency */ +#define DAC960_V1_RebuildInProgress_DataValid 0x0000 /* Consistency */ +#define DAC960_V1_RebuildFailed_LogicalDriveFailure 0x0002 /* Consistency */ +#define DAC960_V1_RebuildFailed_BadBlocksOnOther 0x0003 /* Consistency */ +#define DAC960_V1_RebuildFailed_NewDriveFailed 0x0004 /* Consistency */ +#define DAC960_V1_RebuildSuccessful 0x0100 /* Consistency */ +#define DAC960_V1_RebuildSuccessfullyTerminated 0x0107 /* Consistency */ +#define DAC960_V1_BackgroundInitSuccessful 0x0100 /* Consistency */ +#define DAC960_V1_BackgroundInitAborted 0x0005 /* Consistency */ +#define DAC960_V1_NoBackgroundInitInProgress 0x0105 /* Consistency */ +#define DAC960_V1_AddCapacityInProgress 0x0004 /* Consistency */ +#define DAC960_V1_AddCapacityFailedOrSuspended 0x00F4 /* Consistency */ +#define DAC960_V1_Config2ChecksumError 0x0002 /* Configuration */ +#define DAC960_V1_ConfigurationSuspended 0x0106 /* Configuration */ +#define DAC960_V1_FailedToConfigureNVRAM 0x0105 /* Configuration */ +#define DAC960_V1_ConfigurationNotSavedStateChange 0x0106 /* Configuration */ +#define DAC960_V1_SubsystemNotInstalled 0x0001 /* Subsystem */ +#define DAC960_V1_SubsystemFailed 0x0002 /* Subsystem */ +#define DAC960_V1_SubsystemBusy 0x0106 /* Subsystem */ + +typedef unsigned short DAC960_V1_CommandStatus_T; + + +/* + Define the DAC960 V1 Firmware Enquiry Command reply structure. +*/ + +typedef struct DAC960_V1_Enquiry +{ + unsigned char NumberOfLogicalDrives; /* Byte 0 */ + unsigned int :24; /* Bytes 1-3 */ + unsigned int LogicalDriveSizes[32]; /* Bytes 4-131 */ + unsigned short FlashAge; /* Bytes 132-133 */ + struct { + boolean DeferredWriteError:1; /* Byte 134 Bit 0 */ + boolean BatteryLow:1; /* Byte 134 Bit 1 */ + unsigned char :6; /* Byte 134 Bits 2-7 */ + } StatusFlags; + unsigned char :8; /* Byte 135 */ + unsigned char MinorFirmwareVersion; /* Byte 136 */ + unsigned char MajorFirmwareVersion; /* Byte 137 */ + enum { + DAC960_V1_NoStandbyRebuildOrCheckInProgress = 0x00, + DAC960_V1_StandbyRebuildInProgress = 0x01, + DAC960_V1_BackgroundRebuildInProgress = 0x02, + DAC960_V1_BackgroundCheckInProgress = 0x03, + DAC960_V1_StandbyRebuildCompletedWithError = 0xFF, + DAC960_V1_BackgroundRebuildOrCheckFailed_DriveFailed = 0xF0, + DAC960_V1_BackgroundRebuildOrCheckFailed_LogicalDriveFailed = 0xF1, + DAC960_V1_BackgroundRebuildOrCheckFailed_OtherCauses = 0xF2, + DAC960_V1_BackgroundRebuildOrCheckSuccessfullyTerminated = 0xF3 + } __attribute__ ((packed)) RebuildFlag; /* Byte 138 */ + unsigned char MaxCommands; /* Byte 139 */ + unsigned char OfflineLogicalDriveCount; /* Byte 140 */ + unsigned char :8; /* Byte 141 */ + unsigned short EventLogSequenceNumber; /* Bytes 142-143 */ + unsigned char CriticalLogicalDriveCount; /* Byte 144 */ + unsigned int :24; /* Bytes 145-147 */ + unsigned char DeadDriveCount; /* Byte 148 */ + unsigned char :8; /* Byte 149 */ + unsigned char RebuildCount; /* Byte 150 */ + struct { + unsigned char :3; /* Byte 151 Bits 0-2 */ + boolean BatteryBackupUnitPresent:1; /* Byte 151 Bit 3 */ + unsigned char :3; /* Byte 151 Bits 4-6 */ + unsigned char :1; /* Byte 151 Bit 7 */ + } MiscFlags; + struct { + unsigned char TargetID; + unsigned char Channel; + } DeadDrives[21]; /* Bytes 152-194 */ + unsigned char Reserved[62]; /* Bytes 195-255 */ +} +__attribute__ ((packed)) +DAC960_V1_Enquiry_T; + + +/* + Define the DAC960 V1 Firmware Enquiry2 Command reply structure. +*/ + +typedef struct DAC960_V1_Enquiry2 +{ + struct { + enum { + DAC960_V1_P_PD_PU = 0x01, + DAC960_V1_PL = 0x02, + DAC960_V1_PG = 0x10, + DAC960_V1_PJ = 0x11, + DAC960_V1_PR = 0x12, + DAC960_V1_PT = 0x13, + DAC960_V1_PTL0 = 0x14, + DAC960_V1_PRL = 0x15, + DAC960_V1_PTL1 = 0x16, + DAC960_V1_1164P = 0x20 + } __attribute__ ((packed)) SubModel; /* Byte 0 */ + unsigned char ActualChannels; /* Byte 1 */ + enum { + DAC960_V1_FiveChannelBoard = 0x01, + DAC960_V1_ThreeChannelBoard = 0x02, + DAC960_V1_TwoChannelBoard = 0x03, + DAC960_V1_ThreeChannelASIC_DAC = 0x04 + } __attribute__ ((packed)) Model; /* Byte 2 */ + enum { + DAC960_V1_EISA_Controller = 0x01, + DAC960_V1_MicroChannel_Controller = 0x02, + DAC960_V1_PCI_Controller = 0x03, + DAC960_V1_SCSItoSCSI_Controller = 0x08 + } __attribute__ ((packed)) ProductFamily; /* Byte 3 */ + } HardwareID; /* Bytes 0-3 */ + /* MajorVersion.MinorVersion-FirmwareType-TurnID */ + struct { + unsigned char MajorVersion; /* Byte 4 */ + unsigned char MinorVersion; /* Byte 5 */ + unsigned char TurnID; /* Byte 6 */ + char FirmwareType; /* Byte 7 */ + } FirmwareID; /* Bytes 4-7 */ + unsigned char :8; /* Byte 8 */ + unsigned int :24; /* Bytes 9-11 */ + unsigned char ConfiguredChannels; /* Byte 12 */ + unsigned char ActualChannels; /* Byte 13 */ + unsigned char MaxTargets; /* Byte 14 */ + unsigned char MaxTags; /* Byte 15 */ + unsigned char MaxLogicalDrives; /* Byte 16 */ + unsigned char MaxArms; /* Byte 17 */ + unsigned char MaxSpans; /* Byte 18 */ + unsigned char :8; /* Byte 19 */ + unsigned int :32; /* Bytes 20-23 */ + unsigned int MemorySize; /* Bytes 24-27 */ + unsigned int CacheSize; /* Bytes 28-31 */ + unsigned int FlashMemorySize; /* Bytes 32-35 */ + unsigned int NonVolatileMemorySize; /* Bytes 36-39 */ + struct { + enum { + DAC960_V1_RamType_DRAM = 0x0, + DAC960_V1_RamType_EDO = 0x1, + DAC960_V1_RamType_SDRAM = 0x2, + DAC960_V1_RamType_Last = 0x7 + } __attribute__ ((packed)) RamType:3; /* Byte 40 Bits 0-2 */ + enum { + DAC960_V1_ErrorCorrection_None = 0x0, + DAC960_V1_ErrorCorrection_Parity = 0x1, + DAC960_V1_ErrorCorrection_ECC = 0x2, + DAC960_V1_ErrorCorrection_Last = 0x7 + } __attribute__ ((packed)) ErrorCorrection:3; /* Byte 40 Bits 3-5 */ + boolean FastPageMode:1; /* Byte 40 Bit 6 */ + boolean LowPowerMemory:1; /* Byte 40 Bit 7 */ + unsigned char :8; /* Bytes 41 */ + } MemoryType; + unsigned short ClockSpeed; /* Bytes 42-43 */ + unsigned short MemorySpeed; /* Bytes 44-45 */ + unsigned short HardwareSpeed; /* Bytes 46-47 */ + unsigned int :32; /* Bytes 48-51 */ + unsigned int :32; /* Bytes 52-55 */ + unsigned char :8; /* Byte 56 */ + unsigned char :8; /* Byte 57 */ + unsigned short :16; /* Bytes 58-59 */ + unsigned short MaxCommands; /* Bytes 60-61 */ + unsigned short MaxScatterGatherEntries; /* Bytes 62-63 */ + unsigned short MaxDriveCommands; /* Bytes 64-65 */ + unsigned short MaxIODescriptors; /* Bytes 66-67 */ + unsigned short MaxCombinedSectors; /* Bytes 68-69 */ + unsigned char Latency; /* Byte 70 */ + unsigned char :8; /* Byte 71 */ + unsigned char SCSITimeout; /* Byte 72 */ + unsigned char :8; /* Byte 73 */ + unsigned short MinFreeLines; /* Bytes 74-75 */ + unsigned int :32; /* Bytes 76-79 */ + unsigned int :32; /* Bytes 80-83 */ + unsigned char RebuildRateConstant; /* Byte 84 */ + unsigned char :8; /* Byte 85 */ + unsigned char :8; /* Byte 86 */ + unsigned char :8; /* Byte 87 */ + unsigned int :32; /* Bytes 88-91 */ + unsigned int :32; /* Bytes 92-95 */ + unsigned short PhysicalDriveBlockSize; /* Bytes 96-97 */ + unsigned short LogicalDriveBlockSize; /* Bytes 98-99 */ + unsigned short MaxBlocksPerCommand; /* Bytes 100-101 */ + unsigned short BlockFactor; /* Bytes 102-103 */ + unsigned short CacheLineSize; /* Bytes 104-105 */ + struct { + enum { + DAC960_V1_Narrow_8bit = 0x0, + DAC960_V1_Wide_16bit = 0x1, + DAC960_V1_Wide_32bit = 0x2 + } __attribute__ ((packed)) BusWidth:2; /* Byte 106 Bits 0-1 */ + enum { + DAC960_V1_Fast = 0x0, + DAC960_V1_Ultra = 0x1, + DAC960_V1_Ultra2 = 0x2 + } __attribute__ ((packed)) BusSpeed:2; /* Byte 106 Bits 2-3 */ + boolean Differential:1; /* Byte 106 Bit 4 */ + unsigned char :3; /* Byte 106 Bits 5-7 */ + } SCSICapability; + unsigned char :8; /* Byte 107 */ + unsigned int :32; /* Bytes 108-111 */ + unsigned short FirmwareBuildNumber; /* Bytes 112-113 */ + enum { + DAC960_V1_AEMI = 0x01, + DAC960_V1_OEM1 = 0x02, + DAC960_V1_OEM2 = 0x04, + DAC960_V1_OEM3 = 0x08, + DAC960_V1_Conner = 0x10, + DAC960_V1_SAFTE = 0x20 + } __attribute__ ((packed)) FaultManagementType; /* Byte 114 */ + unsigned char :8; /* Byte 115 */ + struct { + boolean Clustering:1; /* Byte 116 Bit 0 */ + boolean MylexOnlineRAIDExpansion:1; /* Byte 116 Bit 1 */ + boolean ReadAhead:1; /* Byte 116 Bit 2 */ + boolean BackgroundInitialization:1; /* Byte 116 Bit 3 */ + unsigned int :28; /* Bytes 116-119 */ + } FirmwareFeatures; + unsigned int :32; /* Bytes 120-123 */ + unsigned int :32; /* Bytes 124-127 */ +} +DAC960_V1_Enquiry2_T; + + +/* + Define the DAC960 V1 Firmware Logical Drive State type. +*/ + +typedef enum +{ + DAC960_V1_LogicalDrive_Online = 0x03, + DAC960_V1_LogicalDrive_Critical = 0x04, + DAC960_V1_LogicalDrive_Offline = 0xFF +} +__attribute__ ((packed)) +DAC960_V1_LogicalDriveState_T; + + +/* + Define the DAC960 V1 Firmware Logical Drive Information structure. +*/ + +typedef struct DAC960_V1_LogicalDriveInformation +{ + unsigned int LogicalDriveSize; /* Bytes 0-3 */ + DAC960_V1_LogicalDriveState_T LogicalDriveState; /* Byte 4 */ + unsigned char RAIDLevel:7; /* Byte 5 Bits 0-6 */ + boolean WriteBack:1; /* Byte 5 Bit 7 */ + unsigned short :16; /* Bytes 6-7 */ +} +DAC960_V1_LogicalDriveInformation_T; + + +/* + Define the DAC960 V1 Firmware Get Logical Drive Information Command + reply structure. +*/ + +typedef DAC960_V1_LogicalDriveInformation_T + DAC960_V1_LogicalDriveInformationArray_T[DAC960_MaxLogicalDrives]; + + +/* + Define the DAC960 V1 Firmware Perform Event Log Operation Types. +*/ + +typedef enum +{ + DAC960_V1_GetEventLogEntry = 0x00 +} +__attribute__ ((packed)) +DAC960_V1_PerformEventLogOpType_T; + + +/* + Define the DAC960 V1 Firmware Get Event Log Entry Command reply structure. +*/ + +typedef struct DAC960_V1_EventLogEntry +{ + unsigned char MessageType; /* Byte 0 */ + unsigned char MessageLength; /* Byte 1 */ + unsigned char TargetID:5; /* Byte 2 Bits 0-4 */ + unsigned char Channel:3; /* Byte 2 Bits 5-7 */ + unsigned char LogicalUnit:6; /* Byte 3 Bits 0-5 */ + unsigned char :2; /* Byte 3 Bits 6-7 */ + unsigned short SequenceNumber; /* Bytes 4-5 */ + unsigned char ErrorCode:7; /* Byte 6 Bits 0-6 */ + boolean Valid:1; /* Byte 6 Bit 7 */ + unsigned char SegmentNumber; /* Byte 7 */ + DAC960_SCSI_RequestSenseKey_T SenseKey:4; /* Byte 8 Bits 0-3 */ + unsigned char :1; /* Byte 8 Bit 4 */ + boolean ILI:1; /* Byte 8 Bit 5 */ + boolean EOM:1; /* Byte 8 Bit 6 */ + boolean Filemark:1; /* Byte 8 Bit 7 */ + unsigned char Information[4]; /* Bytes 9-12 */ + unsigned char AdditionalSenseLength; /* Byte 13 */ + unsigned char CommandSpecificInformation[4]; /* Bytes 14-17 */ + unsigned char AdditionalSenseCode; /* Byte 18 */ + unsigned char AdditionalSenseCodeQualifier; /* Byte 19 */ + unsigned char Dummy[12]; /* Bytes 20-31 */ +} +DAC960_V1_EventLogEntry_T; + + +/* + Define the DAC960 V1 Firmware Physical Device State type. +*/ + +typedef enum +{ + DAC960_V1_Device_Dead = 0x00, + DAC960_V1_Device_WriteOnly = 0x02, + DAC960_V1_Device_Online = 0x03, + DAC960_V1_Device_Standby = 0x10 +} +__attribute__ ((packed)) +DAC960_V1_PhysicalDeviceState_T; + + +/* + Define the DAC960 V1 Firmware Get Device State Command reply structure. + The structure is padded by 2 bytes for compatibility with Version 2.xx + Firmware. +*/ + +typedef struct DAC960_V1_DeviceState +{ + boolean Present:1; /* Byte 0 Bit 0 */ + unsigned char :7; /* Byte 0 Bits 1-7 */ + enum { + DAC960_V1_OtherType = 0x0, + DAC960_V1_DiskType = 0x1, + DAC960_V1_SequentialType = 0x2, + DAC960_V1_CDROM_or_WORM_Type = 0x3 + } __attribute__ ((packed)) DeviceType:2; /* Byte 1 Bits 0-1 */ + boolean :1; /* Byte 1 Bit 2 */ + boolean Fast20:1; /* Byte 1 Bit 3 */ + boolean Sync:1; /* Byte 1 Bit 4 */ + boolean Fast:1; /* Byte 1 Bit 5 */ + boolean Wide:1; /* Byte 1 Bit 6 */ + boolean TaggedQueuingSupported:1; /* Byte 1 Bit 7 */ + DAC960_V1_PhysicalDeviceState_T DeviceState; /* Byte 2 */ + unsigned char :8; /* Byte 3 */ + unsigned char SynchronousMultiplier; /* Byte 4 */ + unsigned char SynchronousOffset:5; /* Byte 5 Bits 0-4 */ + unsigned char :3; /* Byte 5 Bits 5-7 */ + unsigned int DiskSize __attribute__ ((packed)); /* Bytes 6-9 */ + unsigned short :16; /* Bytes 10-11 */ +} +DAC960_V1_DeviceState_T; + + +/* + Define the DAC960 V1 Firmware Get Rebuild Progress Command reply structure. +*/ + +typedef struct DAC960_V1_RebuildProgress +{ + unsigned int LogicalDriveNumber; /* Bytes 0-3 */ + unsigned int LogicalDriveSize; /* Bytes 4-7 */ + unsigned int RemainingBlocks; /* Bytes 8-11 */ +} +DAC960_V1_RebuildProgress_T; + + +/* + Define the DAC960 V1 Firmware Background Initialization Status Command + reply structure. +*/ + +typedef struct DAC960_V1_BackgroundInitializationStatus +{ + unsigned int LogicalDriveSize; /* Bytes 0-3 */ + unsigned int BlocksCompleted; /* Bytes 4-7 */ + unsigned char Reserved1[12]; /* Bytes 8-19 */ + unsigned int LogicalDriveNumber; /* Bytes 20-23 */ + unsigned char RAIDLevel; /* Byte 24 */ + enum { + DAC960_V1_BackgroundInitializationInvalid = 0x00, + DAC960_V1_BackgroundInitializationStarted = 0x02, + DAC960_V1_BackgroundInitializationInProgress = 0x04, + DAC960_V1_BackgroundInitializationSuspended = 0x05, + DAC960_V1_BackgroundInitializationCancelled = 0x06 + } __attribute__ ((packed)) Status; /* Byte 25 */ + unsigned char Reserved2[6]; /* Bytes 26-31 */ +} +DAC960_V1_BackgroundInitializationStatus_T; + + +/* + Define the DAC960 V1 Firmware Error Table Entry structure. +*/ + +typedef struct DAC960_V1_ErrorTableEntry +{ + unsigned char ParityErrorCount; /* Byte 0 */ + unsigned char SoftErrorCount; /* Byte 1 */ + unsigned char HardErrorCount; /* Byte 2 */ + unsigned char MiscErrorCount; /* Byte 3 */ +} +DAC960_V1_ErrorTableEntry_T; + + +/* + Define the DAC960 V1 Firmware Get Error Table Command reply structure. +*/ + +typedef struct DAC960_V1_ErrorTable +{ + DAC960_V1_ErrorTableEntry_T + ErrorTableEntries[DAC960_V1_MaxChannels][DAC960_V1_MaxTargets]; +} +DAC960_V1_ErrorTable_T; + + +/* + Define the DAC960 V1 Firmware Read Config2 Command reply structure. +*/ + +typedef struct DAC960_V1_Config2 +{ + unsigned char :1; /* Byte 0 Bit 0 */ + boolean ActiveNegationEnabled:1; /* Byte 0 Bit 1 */ + unsigned char :5; /* Byte 0 Bits 2-6 */ + boolean NoRescanIfResetReceivedDuringScan:1; /* Byte 0 Bit 7 */ + boolean StorageWorksSupportEnabled:1; /* Byte 1 Bit 0 */ + boolean HewlettPackardSupportEnabled:1; /* Byte 1 Bit 1 */ + boolean NoDisconnectOnFirstCommand:1; /* Byte 1 Bit 2 */ + unsigned char :2; /* Byte 1 Bits 3-4 */ + boolean AEMI_ARM:1; /* Byte 1 Bit 5 */ + boolean AEMI_OFM:1; /* Byte 1 Bit 6 */ + unsigned char :1; /* Byte 1 Bit 7 */ + enum { + DAC960_V1_OEMID_Mylex = 0x00, + DAC960_V1_OEMID_IBM = 0x08, + DAC960_V1_OEMID_HP = 0x0A, + DAC960_V1_OEMID_DEC = 0x0C, + DAC960_V1_OEMID_Siemens = 0x10, + DAC960_V1_OEMID_Intel = 0x12 + } __attribute__ ((packed)) OEMID; /* Byte 2 */ + unsigned char OEMModelNumber; /* Byte 3 */ + unsigned char PhysicalSector; /* Byte 4 */ + unsigned char LogicalSector; /* Byte 5 */ + unsigned char BlockFactor; /* Byte 6 */ + boolean ReadAheadEnabled:1; /* Byte 7 Bit 0 */ + boolean LowBIOSDelay:1; /* Byte 7 Bit 1 */ + unsigned char :2; /* Byte 7 Bits 2-3 */ + boolean ReassignRestrictedToOneSector:1; /* Byte 7 Bit 4 */ + unsigned char :1; /* Byte 7 Bit 5 */ + boolean ForceUnitAccessDuringWriteRecovery:1; /* Byte 7 Bit 6 */ + boolean EnableLeftSymmetricRAID5Algorithm:1; /* Byte 7 Bit 7 */ + unsigned char DefaultRebuildRate; /* Byte 8 */ + unsigned char :8; /* Byte 9 */ + unsigned char BlocksPerCacheLine; /* Byte 10 */ + unsigned char BlocksPerStripe; /* Byte 11 */ + struct { + enum { + DAC960_V1_Async = 0x0, + DAC960_V1_Sync_8MHz = 0x1, + DAC960_V1_Sync_5MHz = 0x2, + DAC960_V1_Sync_10or20MHz = 0x3 /* Byte 11 Bits 0-1 */ + } __attribute__ ((packed)) Speed:2; + boolean Force8Bit:1; /* Byte 11 Bit 2 */ + boolean DisableFast20:1; /* Byte 11 Bit 3 */ + unsigned char :3; /* Byte 11 Bits 4-6 */ + boolean EnableTaggedQueuing:1; /* Byte 11 Bit 7 */ + } __attribute__ ((packed)) ChannelParameters[6]; /* Bytes 12-17 */ + unsigned char SCSIInitiatorID; /* Byte 18 */ + unsigned char :8; /* Byte 19 */ + enum { + DAC960_V1_StartupMode_ControllerSpinUp = 0x00, + DAC960_V1_StartupMode_PowerOnSpinUp = 0x01 + } __attribute__ ((packed)) StartupMode; /* Byte 20 */ + unsigned char SimultaneousDeviceSpinUpCount; /* Byte 21 */ + unsigned char SecondsDelayBetweenSpinUps; /* Byte 22 */ + unsigned char Reserved1[29]; /* Bytes 23-51 */ + boolean BIOSDisabled:1; /* Byte 52 Bit 0 */ + boolean CDROMBootEnabled:1; /* Byte 52 Bit 1 */ + unsigned char :3; /* Byte 52 Bits 2-4 */ + enum { + DAC960_V1_Geometry_128_32 = 0x0, + DAC960_V1_Geometry_255_63 = 0x1, + DAC960_V1_Geometry_Reserved1 = 0x2, + DAC960_V1_Geometry_Reserved2 = 0x3 + } __attribute__ ((packed)) DriveGeometry:2; /* Byte 52 Bits 5-6 */ + unsigned char :1; /* Byte 52 Bit 7 */ + unsigned char Reserved2[9]; /* Bytes 53-61 */ + unsigned short Checksum; /* Bytes 62-63 */ +} +DAC960_V1_Config2_T; + + +/* + Define the DAC960 V1 Firmware DCDB request structure. +*/ + +typedef struct DAC960_V1_DCDB +{ + unsigned char TargetID:4; /* Byte 0 Bits 0-3 */ + unsigned char Channel:4; /* Byte 0 Bits 4-7 */ + enum { + DAC960_V1_DCDB_NoDataTransfer = 0, + DAC960_V1_DCDB_DataTransferDeviceToSystem = 1, + DAC960_V1_DCDB_DataTransferSystemToDevice = 2, + DAC960_V1_DCDB_IllegalDataTransfer = 3 + } __attribute__ ((packed)) Direction:2; /* Byte 1 Bits 0-1 */ + boolean EarlyStatus:1; /* Byte 1 Bit 2 */ + unsigned char :1; /* Byte 1 Bit 3 */ + enum { + DAC960_V1_DCDB_Timeout_24_hours = 0, + DAC960_V1_DCDB_Timeout_10_seconds = 1, + DAC960_V1_DCDB_Timeout_60_seconds = 2, + DAC960_V1_DCDB_Timeout_10_minutes = 3 + } __attribute__ ((packed)) Timeout:2; /* Byte 1 Bits 4-5 */ + boolean NoAutomaticRequestSense:1; /* Byte 1 Bit 6 */ + boolean DisconnectPermitted:1; /* Byte 1 Bit 7 */ + unsigned short TransferLength; /* Bytes 2-3 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 4-7 */ + unsigned char CDBLength:4; /* Byte 8 Bits 0-3 */ + unsigned char TransferLengthHigh4:4; /* Byte 8 Bits 4-7 */ + unsigned char SenseLength; /* Byte 9 */ + unsigned char CDB[12]; /* Bytes 10-21 */ + unsigned char SenseData[64]; /* Bytes 22-85 */ + unsigned char Status; /* Byte 86 */ + unsigned char :8; /* Byte 87 */ +} +DAC960_V1_DCDB_T; + + +/* + Define the DAC960 V1 Firmware Scatter/Gather List Type 1 32 Bit Address + 32 Bit Byte Count structure. +*/ + +typedef struct DAC960_V1_ScatterGatherSegment +{ + DAC960_BusAddress32_T SegmentDataPointer; /* Bytes 0-3 */ + DAC960_ByteCount32_T SegmentByteCount; /* Bytes 4-7 */ +} +DAC960_V1_ScatterGatherSegment_T; + + +/* + Define the 13 Byte DAC960 V1 Firmware Command Mailbox structure. Bytes 13-15 + are not used. The Command Mailbox structure is padded to 16 bytes for + efficient access. +*/ + +typedef union DAC960_V1_CommandMailbox +{ + unsigned int Words[4]; /* Words 0-3 */ + unsigned char Bytes[16]; /* Bytes 0-15 */ + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned char Dummy[14]; /* Bytes 2-15 */ + } __attribute__ ((packed)) Common; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned char Dummy1[6]; /* Bytes 2-7 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 8-11 */ + unsigned char Dummy2[4]; /* Bytes 12-15 */ + } __attribute__ ((packed)) Type3; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned char CommandOpcode2; /* Byte 2 */ + unsigned char Dummy1[5]; /* Bytes 3-7 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 8-11 */ + unsigned char Dummy2[4]; /* Bytes 12-15 */ + } __attribute__ ((packed)) Type3B; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned char Dummy1[5]; /* Bytes 2-6 */ + unsigned char LogicalDriveNumber:6; /* Byte 7 Bits 0-6 */ + boolean AutoRestore:1; /* Byte 7 Bit 7 */ + unsigned char Dummy2[8]; /* Bytes 8-15 */ + } __attribute__ ((packed)) Type3C; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned char Channel; /* Byte 2 */ + unsigned char TargetID; /* Byte 3 */ + DAC960_V1_PhysicalDeviceState_T DeviceState:5; /* Byte 4 Bits 0-4 */ + unsigned char Modifier:3; /* Byte 4 Bits 5-7 */ + unsigned char Dummy1[3]; /* Bytes 5-7 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 8-11 */ + unsigned char Dummy2[4]; /* Bytes 12-15 */ + } __attribute__ ((packed)) Type3D; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + DAC960_V1_PerformEventLogOpType_T OperationType; /* Byte 2 */ + unsigned char OperationQualifier; /* Byte 3 */ + unsigned short SequenceNumber; /* Bytes 4-5 */ + unsigned char Dummy1[2]; /* Bytes 6-7 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 8-11 */ + unsigned char Dummy2[4]; /* Bytes 12-15 */ + } __attribute__ ((packed)) Type3E; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned char Dummy1[2]; /* Bytes 2-3 */ + unsigned char RebuildRateConstant; /* Byte 4 */ + unsigned char Dummy2[3]; /* Bytes 5-7 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 8-11 */ + unsigned char Dummy3[4]; /* Bytes 12-15 */ + } __attribute__ ((packed)) Type3R; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned short TransferLength; /* Bytes 2-3 */ + unsigned int LogicalBlockAddress; /* Bytes 4-7 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 8-11 */ + unsigned char LogicalDriveNumber; /* Byte 12 */ + unsigned char Dummy[3]; /* Bytes 13-15 */ + } __attribute__ ((packed)) Type4; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + struct { + unsigned short TransferLength:11; /* Bytes 2-3 */ + unsigned char LogicalDriveNumber:5; /* Byte 3 Bits 3-7 */ + } __attribute__ ((packed)) LD; + unsigned int LogicalBlockAddress; /* Bytes 4-7 */ + DAC960_BusAddress32_T BusAddress; /* Bytes 8-11 */ + unsigned char ScatterGatherCount:6; /* Byte 12 Bits 0-5 */ + enum { + DAC960_V1_ScatterGather_32BitAddress_32BitByteCount = 0x0, + DAC960_V1_ScatterGather_32BitAddress_16BitByteCount = 0x1, + DAC960_V1_ScatterGather_32BitByteCount_32BitAddress = 0x2, + DAC960_V1_ScatterGather_16BitByteCount_32BitAddress = 0x3 + } __attribute__ ((packed)) ScatterGatherType:2; /* Byte 12 Bits 6-7 */ + unsigned char Dummy[3]; /* Bytes 13-15 */ + } __attribute__ ((packed)) Type5; + struct { + DAC960_V1_CommandOpcode_T CommandOpcode; /* Byte 0 */ + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 1 */ + unsigned char CommandOpcode2; /* Byte 2 */ + unsigned char :8; /* Byte 3 */ + DAC960_BusAddress32_T CommandMailboxesBusAddress; /* Bytes 4-7 */ + DAC960_BusAddress32_T StatusMailboxesBusAddress; /* Bytes 8-11 */ + unsigned char Dummy[4]; /* Bytes 12-15 */ + } __attribute__ ((packed)) TypeX; +} +DAC960_V1_CommandMailbox_T; + + +/* + Define the DAC960 V2 Firmware Command Opcodes. +*/ + +typedef enum +{ + DAC960_V2_MemCopy = 0x01, + DAC960_V2_SCSI_10_Passthru = 0x02, + DAC960_V2_SCSI_255_Passthru = 0x03, + DAC960_V2_SCSI_10 = 0x04, + DAC960_V2_SCSI_256 = 0x05, + DAC960_V2_IOCTL = 0x20 +} +__attribute__ ((packed)) +DAC960_V2_CommandOpcode_T; + + +/* + Define the DAC960 V2 Firmware IOCTL Opcodes. +*/ + +typedef enum +{ + DAC960_V2_GetControllerInfo = 0x01, + DAC960_V2_GetLogicalDeviceInfoValid = 0x03, + DAC960_V2_GetPhysicalDeviceInfoValid = 0x05, + DAC960_V2_GetHealthStatus = 0x11, + DAC960_V2_GetEvent = 0x15, + DAC960_V2_StartDiscovery = 0x81, + DAC960_V2_SetDeviceState = 0x82, + DAC960_V2_RebuildDeviceStart = 0x88, + DAC960_V2_RebuildDeviceStop = 0x89, + DAC960_V2_ConsistencyCheckStart = 0x8C, + DAC960_V2_ConsistencyCheckStop = 0x8D, + DAC960_V2_SetMemoryMailbox = 0x8E, + DAC960_V2_PauseDevice = 0x92, + DAC960_V2_TranslatePhysicalToLogicalDevice = 0xC5 +} +__attribute__ ((packed)) +DAC960_V2_IOCTL_Opcode_T; + + +/* + Define the DAC960 V2 Firmware Command Identifier type. +*/ + +typedef unsigned short DAC960_V2_CommandIdentifier_T; + + +/* + Define the DAC960 V2 Firmware Command Status Codes. +*/ + +#define DAC960_V2_NormalCompletion 0x00 +#define DAC960_V2_AbormalCompletion 0x02 +#define DAC960_V2_DeviceBusy 0x08 +#define DAC960_V2_DeviceNonresponsive 0x0E +#define DAC960_V2_DeviceNonresponsive2 0x0F +#define DAC960_V2_DeviceRevervationConflict 0x18 + +typedef unsigned char DAC960_V2_CommandStatus_T; + + +/* + Define the DAC960 V2 Firmware Memory Type structure. +*/ + +typedef struct DAC960_V2_MemoryType +{ + enum { + DAC960_V2_MemoryType_Reserved = 0x00, + DAC960_V2_MemoryType_DRAM = 0x01, + DAC960_V2_MemoryType_EDRAM = 0x02, + DAC960_V2_MemoryType_EDO = 0x03, + DAC960_V2_MemoryType_SDRAM = 0x04, + DAC960_V2_MemoryType_Last = 0x1F + } __attribute__ ((packed)) MemoryType:5; /* Byte 0 Bits 0-4 */ + boolean :1; /* Byte 0 Bit 5 */ + boolean MemoryParity:1; /* Byte 0 Bit 6 */ + boolean MemoryECC:1; /* Byte 0 Bit 7 */ +} +DAC960_V2_MemoryType_T; + + +/* + Define the DAC960 V2 Firmware Processor Type structure. +*/ + +typedef enum +{ + DAC960_V2_ProcessorType_i960CA = 0x01, + DAC960_V2_ProcessorType_i960RD = 0x02, + DAC960_V2_ProcessorType_i960RN = 0x03, + DAC960_V2_ProcessorType_i960RP = 0x04, + DAC960_V2_ProcessorType_NorthBay = 0x05, + DAC960_V2_ProcessorType_StrongArm = 0x06, + DAC960_V2_ProcessorType_i960RM = 0x07 +} +__attribute__ ((packed)) +DAC960_V2_ProcessorType_T; + + +/* + Define the DAC960 V2 Firmware Get Controller Info reply structure. +*/ + +typedef struct DAC960_V2_ControllerInfo +{ + unsigned char :8; /* Byte 0 */ + enum { + DAC960_V2_SCSI_Bus = 0x00, + DAC960_V2_Fibre_Bus = 0x01, + DAC960_V2_PCI_Bus = 0x03 + } __attribute__ ((packed)) BusInterfaceType; /* Byte 1 */ + enum { + DAC960_V2_DAC960E = 0x01, + DAC960_V2_DAC960M = 0x08, + DAC960_V2_DAC960PD = 0x10, + DAC960_V2_DAC960PL = 0x11, + DAC960_V2_DAC960PU = 0x12, + DAC960_V2_DAC960PE = 0x13, + DAC960_V2_DAC960PG = 0x14, + DAC960_V2_DAC960PJ = 0x15, + DAC960_V2_DAC960PTL0 = 0x16, + DAC960_V2_DAC960PR = 0x17, + DAC960_V2_DAC960PRL = 0x18, + DAC960_V2_DAC960PT = 0x19, + DAC960_V2_DAC1164P = 0x1A, + DAC960_V2_DAC960PTL1 = 0x1B, + DAC960_V2_EXR2000P = 0x1C, + DAC960_V2_EXR3000P = 0x1D, + DAC960_V2_AcceleRAID352 = 0x1E, + DAC960_V2_AcceleRAID170 = 0x1F, + DAC960_V2_AcceleRAID160 = 0x20, + DAC960_V2_DAC960S = 0x60, + DAC960_V2_DAC960SU = 0x61, + DAC960_V2_DAC960SX = 0x62, + DAC960_V2_DAC960SF = 0x63, + DAC960_V2_DAC960SS = 0x64, + DAC960_V2_DAC960FL = 0x65, + DAC960_V2_DAC960LL = 0x66, + DAC960_V2_DAC960FF = 0x67, + DAC960_V2_DAC960HP = 0x68, + DAC960_V2_RAIDBRICK = 0x69, + DAC960_V2_METEOR_FL = 0x6A, + DAC960_V2_METEOR_FF = 0x6B + } __attribute__ ((packed)) ControllerType; /* Byte 2 */ + unsigned char :8; /* Byte 3 */ + unsigned short BusInterfaceSpeedMHz; /* Bytes 4-5 */ + unsigned char BusWidthBits; /* Byte 6 */ + unsigned char FlashCodeTypeOrProductID; /* Byte 7 */ + unsigned char NumberOfHostPortsPresent; /* Byte 8 */ + unsigned char Reserved1[7]; /* Bytes 9-15 */ + unsigned char BusInterfaceName[16]; /* Bytes 16-31 */ + unsigned char ControllerName[16]; /* Bytes 32-47 */ + unsigned char Reserved2[16]; /* Bytes 48-63 */ + /* Firmware Release Information */ + unsigned char FirmwareMajorVersion; /* Byte 64 */ + unsigned char FirmwareMinorVersion; /* Byte 65 */ + unsigned char FirmwareTurnNumber; /* Byte 66 */ + unsigned char FirmwareBuildNumber; /* Byte 67 */ + unsigned char FirmwareReleaseDay; /* Byte 68 */ + unsigned char FirmwareReleaseMonth; /* Byte 69 */ + unsigned char FirmwareReleaseYearHigh2Digits; /* Byte 70 */ + unsigned char FirmwareReleaseYearLow2Digits; /* Byte 71 */ + /* Hardware Release Information */ + unsigned char HardwareRevision; /* Byte 72 */ + unsigned int :24; /* Bytes 73-75 */ + unsigned char HardwareReleaseDay; /* Byte 76 */ + unsigned char HardwareReleaseMonth; /* Byte 77 */ + unsigned char HardwareReleaseYearHigh2Digits; /* Byte 78 */ + unsigned char HardwareReleaseYearLow2Digits; /* Byte 79 */ + /* Hardware Manufacturing Information */ + unsigned char ManufacturingBatchNumber; /* Byte 80 */ + unsigned char :8; /* Byte 81 */ + unsigned char ManufacturingPlantNumber; /* Byte 82 */ + unsigned char :8; /* Byte 83 */ + unsigned char HardwareManufacturingDay; /* Byte 84 */ + unsigned char HardwareManufacturingMonth; /* Byte 85 */ + unsigned char HardwareManufacturingYearHigh2Digits; /* Byte 86 */ + unsigned char HardwareManufacturingYearLow2Digits; /* Byte 87 */ + unsigned char MaximumNumberOfPDDperXLD; /* Byte 88 */ + unsigned char MaximumNumberOfILDperXLD; /* Byte 89 */ + unsigned short NonvolatileMemorySizeKB; /* Bytes 90-91 */ + unsigned char MaximumNumberOfXLD; /* Byte 92 */ + unsigned int :24; /* Bytes 93-95 */ + /* Unique Information per Controller */ + unsigned char ControllerSerialNumber[16]; /* Bytes 96-111 */ + unsigned char Reserved3[16]; /* Bytes 112-127 */ + /* Vendor Information */ + unsigned int :24; /* Bytes 128-130 */ + unsigned char OEM_Code; /* Byte 131 */ + unsigned char VendorName[16]; /* Bytes 132-147 */ + /* Other Physical/Controller/Operation Information */ + boolean BBU_Present:1; /* Byte 148 Bit 0 */ + boolean ActiveActiveClusteringMode:1; /* Byte 148 Bit 1 */ + unsigned char :6; /* Byte 148 Bits 2-7 */ + unsigned char :8; /* Byte 149 */ + unsigned short :16; /* Bytes 150-151 */ + /* Physical Device Scan Information */ + boolean PhysicalScanActive:1; /* Byte 152 Bit 0 */ + unsigned char :7; /* Byte 152 Bits 1-7 */ + unsigned char PhysicalDeviceChannelNumber; /* Byte 153 */ + unsigned char PhysicalDeviceTargetID; /* Byte 154 */ + unsigned char PhysicalDeviceLogicalUnit; /* Byte 155 */ + /* Maximum Command Data Transfer Sizes */ + unsigned short MaximumDataTransferSizeInBlocks; /* Bytes 156-157 */ + unsigned short MaximumScatterGatherEntries; /* Bytes 158-159 */ + /* Logical/Physical Device Counts */ + unsigned short LogicalDevicesPresent; /* Bytes 160-161 */ + unsigned short LogicalDevicesCritical; /* Bytes 162-163 */ + unsigned short LogicalDevicesOffline; /* Bytes 164-165 */ + unsigned short PhysicalDevicesPresent; /* Bytes 166-167 */ + unsigned short PhysicalDisksPresent; /* Bytes 168-169 */ + unsigned short PhysicalDisksCritical; /* Bytes 170-171 */ + unsigned short PhysicalDisksOffline; /* Bytes 172-173 */ + unsigned short MaximumParallelCommands; /* Bytes 174-175 */ + /* Channel and Target ID Information */ + unsigned char NumberOfPhysicalChannelsPresent; /* Byte 176 */ + unsigned char NumberOfVirtualChannelsPresent; /* Byte 177 */ + unsigned char NumberOfPhysicalChannelsPossible; /* Byte 178 */ + unsigned char NumberOfVirtualChannelsPossible; /* Byte 179 */ + unsigned char MaximumTargetsPerChannel[16]; /* Bytes 180-195 */ + unsigned char Reserved4[12]; /* Bytes 196-207 */ + /* Memory/Cache Information */ + unsigned short MemorySizeMB; /* Bytes 208-209 */ + unsigned short CacheSizeMB; /* Bytes 210-211 */ + unsigned int ValidCacheSizeInBytes; /* Bytes 212-215 */ + unsigned int DirtyCacheSizeInBytes; /* Bytes 216-219 */ + unsigned short MemorySpeedMHz; /* Bytes 220-221 */ + unsigned char MemoryDataWidthBits; /* Byte 222 */ + DAC960_V2_MemoryType_T MemoryType; /* Byte 223 */ + unsigned char CacheMemoryTypeName[16]; /* Bytes 224-239 */ + /* Execution Memory Information */ + unsigned short ExecutionMemorySizeMB; /* Bytes 240-241 */ + unsigned short ExecutionL2CacheSizeMB; /* Bytes 242-243 */ + unsigned char Reserved5[8]; /* Bytes 244-251 */ + unsigned short ExecutionMemorySpeedMHz; /* Bytes 252-253 */ + unsigned char ExecutionMemoryDataWidthBits; /* Byte 254 */ + DAC960_V2_MemoryType_T ExecutionMemoryType; /* Byte 255 */ + unsigned char ExecutionMemoryTypeName[16]; /* Bytes 256-271 */ + /* First CPU Type Information */ + unsigned short FirstProcessorSpeedMHz; /* Bytes 272-273 */ + DAC960_V2_ProcessorType_T FirstProcessorType; /* Byte 274 */ + unsigned char FirstProcessorCount; /* Byte 275 */ + unsigned char Reserved6[12]; /* Bytes 276-287 */ + unsigned char FirstProcessorName[16]; /* Bytes 288-303 */ + /* Second CPU Type Information */ + unsigned short SecondProcessorSpeedMHz; /* Bytes 304-305 */ + DAC960_V2_ProcessorType_T SecondProcessorType; /* Byte 306 */ + unsigned char SecondProcessorCount; /* Byte 307 */ + unsigned char Reserved7[12]; /* Bytes 308-319 */ + unsigned char SecondProcessorName[16]; /* Bytes 320-335 */ + /* Debugging/Profiling/Command Time Tracing Information */ + unsigned short CurrentProfilingDataPageNumber; /* Bytes 336-337 */ + unsigned short ProgramsAwaitingProfilingData; /* Bytes 338-339 */ + unsigned short CurrentCommandTimeTraceDataPageNumber; /* Bytes 340-341 */ + unsigned short ProgramsAwaitingCommandTimeTraceData; /* Bytes 342-343 */ + unsigned char Reserved8[8]; /* Bytes 344-351 */ + /* Error Counters on Physical Devices */ + unsigned short PhysicalDeviceBusResets; /* Bytes 352-353 */ + unsigned short PhysicalDeviceParityErrors; /* Bytes 355-355 */ + unsigned short PhysicalDeviceSoftErrors; /* Bytes 356-357 */ + unsigned short PhysicalDeviceCommandsFailed; /* Bytes 358-359 */ + unsigned short PhysicalDeviceMiscellaneousErrors; /* Bytes 360-361 */ + unsigned short PhysicalDeviceCommandTimeouts; /* Bytes 362-363 */ + unsigned short PhysicalDeviceSelectionTimeouts; /* Bytes 364-365 */ + unsigned short PhysicalDeviceRetriesDone; /* Bytes 366-367 */ + unsigned short PhysicalDeviceAbortsDone; /* Bytes 368-369 */ + unsigned short PhysicalDeviceHostCommandAbortsDone; /* Bytes 370-371 */ + unsigned short PhysicalDevicePredictedFailuresDetected; /* Bytes 372-373 */ + unsigned short PhysicalDeviceHostCommandsFailed; /* Bytes 374-375 */ + unsigned short PhysicalDeviceHardErrors; /* Bytes 376-377 */ + unsigned char Reserved9[6]; /* Bytes 378-383 */ + /* Error Counters on Logical Devices */ + unsigned short LogicalDeviceSoftErrors; /* Bytes 384-385 */ + unsigned short LogicalDeviceCommandsFailed; /* Bytes 386-387 */ + unsigned short LogicalDeviceHostCommandAbortsDone; /* Bytes 388-389 */ + unsigned short :16; /* Bytes 390-391 */ + /* Error Counters on Controller */ + unsigned short ControllerMemoryErrors; /* Bytes 392-393 */ + unsigned short ControllerHostCommandAbortsDone; /* Bytes 394-395 */ + unsigned int :32; /* Bytes 396-399 */ + /* Long Duration Activity Information */ + unsigned short BackgroundInitializationsActive; /* Bytes 400-401 */ + unsigned short LogicalDeviceInitializationsActive; /* Bytes 402-403 */ + unsigned short PhysicalDeviceInitializationsActive; /* Bytes 404-405 */ + unsigned short ConsistencyChecksActive; /* Bytes 406-407 */ + unsigned short RebuildsActive; /* Bytes 408-409 */ + unsigned short OnlineExpansionsActive; /* Bytes 410-411 */ + unsigned short PatrolActivitiesActive; /* Bytes 412-413 */ + unsigned short :16; /* Bytes 414-415 */ + /* Flash ROM Information */ + unsigned char FlashType; /* Byte 416 */ + unsigned char :8; /* Byte 417 */ + unsigned short FlashSizeMB; /* Bytes 418-419 */ + unsigned int FlashLimit; /* Bytes 420-423 */ + unsigned int FlashCount; /* Bytes 424-427 */ + unsigned int :32; /* Bytes 428-431 */ + unsigned char FlashTypeName[16]; /* Bytes 432-447 */ + /* Firmware Run Time Information */ + unsigned char RebuildRate; /* Byte 448 */ + unsigned char BackgroundInitializationRate; /* Byte 449 */ + unsigned char ForegroundInitializationRate; /* Byte 450 */ + unsigned char ConsistencyCheckRate; /* Byte 451 */ + unsigned int :32; /* Bytes 452-455 */ + unsigned int MaximumDP; /* Bytes 456-459 */ + unsigned int FreeDP; /* Bytes 460-463 */ + unsigned int MaximumIOP; /* Bytes 464-467 */ + unsigned int FreeIOP; /* Bytes 468-471 */ + unsigned short MaximumCombLengthInBlocks; /* Bytes 472-473 */ + unsigned short NumberOfConfigurationGroups; /* Bytes 474-475 */ + boolean InstallationAbortStatus:1; /* Byte 476 Bit 0 */ + boolean MaintenanceModeStatus:1; /* Byte 476 Bit 1 */ + unsigned int :24; /* Bytes 476-479 */ + unsigned char Reserved10[32]; /* Bytes 480-511 */ + unsigned char Reserved11[512]; /* Bytes 512-1023 */ +} +DAC960_V2_ControllerInfo_T; + + +/* + Define the DAC960 V2 Firmware Logical Device State type. +*/ + +typedef enum +{ + DAC960_V2_LogicalDevice_Online = 0x01, + DAC960_V2_LogicalDevice_Offline = 0x08, + DAC960_V2_LogicalDevice_Critical = 0x09 +} +__attribute__ ((packed)) +DAC960_V2_LogicalDeviceState_T; + + +/* + Define the DAC960 V2 Firmware Get Logical Device Info reply structure. +*/ + +typedef struct DAC960_V2_LogicalDeviceInfo +{ + unsigned char :8; /* Byte 0 */ + unsigned char Channel; /* Byte 1 */ + unsigned char TargetID; /* Byte 2 */ + unsigned char LogicalUnit; /* Byte 3 */ + DAC960_V2_LogicalDeviceState_T LogicalDeviceState; /* Byte 4 */ + unsigned char RAIDLevel; /* Byte 5 */ + unsigned char StripeSize; /* Byte 6 */ + unsigned char CacheLineSize; /* Byte 7 */ + struct { + enum { + DAC960_V2_ReadCacheDisabled = 0x0, + DAC960_V2_ReadCacheEnabled = 0x1, + DAC960_V2_ReadAheadEnabled = 0x2, + DAC960_V2_IntelligentReadAheadEnabled = 0x3, + DAC960_V2_ReadCache_Last = 0x7 + } __attribute__ ((packed)) ReadCache:3; /* Byte 8 Bits 0-2 */ + enum { + DAC960_V2_WriteCacheDisabled = 0x0, + DAC960_V2_LogicalDeviceReadOnly = 0x1, + DAC960_V2_WriteCacheEnabled = 0x2, + DAC960_V2_IntelligentWriteCacheEnabled = 0x3, + DAC960_V2_WriteCache_Last = 0x7 + } __attribute__ ((packed)) WriteCache:3; /* Byte 8 Bits 3-5 */ + boolean :1; /* Byte 8 Bit 6 */ + boolean LogicalDeviceInitialized:1; /* Byte 8 Bit 7 */ + } LogicalDeviceControl; /* Byte 8 */ + /* Logical Device Operations Status */ + boolean ConsistencyCheckInProgress:1; /* Byte 9 Bit 0 */ + boolean RebuildInProgress:1; /* Byte 9 Bit 1 */ + boolean BackgroundInitializationInProgress:1; /* Byte 9 Bit 2 */ + boolean ForegroundInitializationInProgress:1; /* Byte 9 Bit 3 */ + boolean DataMigrationInProgress:1; /* Byte 9 Bit 4 */ + boolean PatrolOperationInProgress:1; /* Byte 9 Bit 5 */ + unsigned char :2; /* Byte 9 Bits 6-7 */ + unsigned char RAID5WriteUpdate; /* Byte 10 */ + unsigned char RAID5Algorithm; /* Byte 11 */ + unsigned short LogicalDeviceNumber; /* Bytes 12-13 */ + /* BIOS Info */ + boolean BIOSDisabled:1; /* Byte 14 Bit 0 */ + boolean CDROMBootEnabled:1; /* Byte 14 Bit 1 */ + boolean DriveCoercionEnabled:1; /* Byte 14 Bit 2 */ + boolean WriteSameDisabled:1; /* Byte 14 Bit 3 */ + boolean HBA_ModeEnabled:1; /* Byte 14 Bit 4 */ + enum { + DAC960_V2_Geometry_128_32 = 0x0, + DAC960_V2_Geometry_255_63 = 0x1, + DAC960_V2_Geometry_Reserved1 = 0x2, + DAC960_V2_Geometry_Reserved2 = 0x3 + } __attribute__ ((packed)) DriveGeometry:2; /* Byte 14 Bits 5-6 */ + boolean SuperReadAheadEnabled:1; /* Byte 14 Bit 7 */ + unsigned char :8; /* Byte 15 */ + /* Error Counters */ + unsigned short SoftErrors; /* Bytes 16-17 */ + unsigned short CommandsFailed; /* Bytes 18-19 */ + unsigned short HostCommandAbortsDone; /* Bytes 20-21 */ + unsigned short DeferredWriteErrors; /* Bytes 22-23 */ + unsigned int :32; /* Bytes 24-27 */ + unsigned int :32; /* Bytes 28-31 */ + /* Device Size Information */ + unsigned short :16; /* Bytes 32-33 */ + unsigned short DeviceBlockSizeInBytes; /* Bytes 34-35 */ + unsigned int OriginalDeviceSize; /* Bytes 36-39 */ + unsigned int ConfigurableDeviceSize; /* Bytes 40-43 */ + unsigned int :32; /* Bytes 44-47 */ + unsigned char LogicalDeviceName[32]; /* Bytes 48-79 */ + unsigned char SCSI_InquiryData[36]; /* Bytes 80-115 */ + unsigned char Reserved1[12]; /* Bytes 116-127 */ + DAC960_ByteCount64_T LastReadBlockNumber; /* Bytes 128-135 */ + DAC960_ByteCount64_T LastWrittenBlockNumber; /* Bytes 136-143 */ + DAC960_ByteCount64_T ConsistencyCheckBlockNumber; /* Bytes 144-151 */ + DAC960_ByteCount64_T RebuildBlockNumber; /* Bytes 152-159 */ + DAC960_ByteCount64_T BackgroundInitializationBlockNumber; /* Bytes 160-167 */ + DAC960_ByteCount64_T ForegroundInitializationBlockNumber; /* Bytes 168-175 */ + DAC960_ByteCount64_T DataMigrationBlockNumber; /* Bytes 176-183 */ + DAC960_ByteCount64_T PatrolOperationBlockNumber; /* Bytes 184-191 */ + unsigned char Reserved2[64]; /* Bytes 192-255 */ +} +DAC960_V2_LogicalDeviceInfo_T; + + +/* + Define the DAC960 V2 Firmware Physical Device State type. +*/ + +typedef enum +{ + DAC960_V2_Device_Unconfigured = 0x00, + DAC960_V2_Device_Online = 0x01, + DAC960_V2_Device_Rebuild = 0x03, + DAC960_V2_Device_Missing = 0x04, + DAC960_V2_Device_Critical = 0x05, + DAC960_V2_Device_Dead = 0x08, + DAC960_V2_Device_SuspectedDead = 0x0C, + DAC960_V2_Device_CommandedOffline = 0x10, + DAC960_V2_Device_Standby = 0x21, + DAC960_V2_Device_InvalidState = 0xFF +} +__attribute__ ((packed)) +DAC960_V2_PhysicalDeviceState_T; + + +/* + Define the DAC960 V2 Firmware Get Physical Device Info reply structure. +*/ + +typedef struct DAC960_V2_PhysicalDeviceInfo +{ + unsigned char :8; /* Byte 0 */ + unsigned char Channel; /* Byte 1 */ + unsigned char TargetID; /* Byte 2 */ + unsigned char LogicalUnit; /* Byte 3 */ + /* Configuration Status Bits */ + boolean PhysicalDeviceFaultTolerant:1; /* Byte 4 Bit 0 */ + boolean PhysicalDeviceConnected:1; /* Byte 4 Bit 1 */ + boolean PhysicalDeviceLocalToController:1; /* Byte 4 Bit 2 */ + unsigned char :5; /* Byte 4 Bits 3-7 */ + /* Multiple Host/Controller Status Bits */ + boolean RemoteHostSystemDead:1; /* Byte 5 Bit 0 */ + boolean RemoteControllerDead:1; /* Byte 5 Bit 1 */ + unsigned char :6; /* Byte 5 Bits 2-7 */ + DAC960_V2_PhysicalDeviceState_T PhysicalDeviceState; /* Byte 6 */ + unsigned char NegotiatedDataWidthBits; /* Byte 7 */ + unsigned short NegotiatedSynchronousMegaTransfers; /* Bytes 8-9 */ + /* Multiported Physical Device Information */ + unsigned char NumberOfPortConnections; /* Byte 10 */ + unsigned char DriveAccessibilityBitmap; /* Byte 11 */ + unsigned int :32; /* Bytes 12-15 */ + unsigned char NetworkAddress[16]; /* Bytes 16-31 */ + unsigned short MaximumTags; /* Bytes 32-33 */ + /* Physical Device Operations Status */ + boolean ConsistencyCheckInProgress:1; /* Byte 34 Bit 0 */ + boolean RebuildInProgress:1; /* Byte 34 Bit 1 */ + boolean MakingDataConsistentInProgress:1; /* Byte 34 Bit 2 */ + boolean PhysicalDeviceInitializationInProgress:1; /* Byte 34 Bit 3 */ + boolean DataMigrationInProgress:1; /* Byte 34 Bit 4 */ + boolean PatrolOperationInProgress:1; /* Byte 34 Bit 5 */ + unsigned char :2; /* Byte 34 Bits 6-7 */ + unsigned char LongOperationStatus; /* Byte 35 */ + unsigned char ParityErrors; /* Byte 36 */ + unsigned char SoftErrors; /* Byte 37 */ + unsigned char HardErrors; /* Byte 38 */ + unsigned char MiscellaneousErrors; /* Byte 39 */ + unsigned char CommandTimeouts; /* Byte 40 */ + unsigned char Retries; /* Byte 41 */ + unsigned char Aborts; /* Byte 42 */ + unsigned char PredictedFailuresDetected; /* Byte 43 */ + unsigned int :32; /* Bytes 44-47 */ + unsigned short :16; /* Bytes 48-49 */ + unsigned short DeviceBlockSizeInBytes; /* Bytes 50-51 */ + unsigned int OriginalDeviceSize; /* Bytes 52-55 */ + unsigned int ConfigurableDeviceSize; /* Bytes 56-59 */ + unsigned int :32; /* Bytes 60-63 */ + unsigned char PhysicalDeviceName[16]; /* Bytes 64-79 */ + unsigned char Reserved1[16]; /* Bytes 80-95 */ + unsigned char Reserved2[32]; /* Bytes 96-127 */ + unsigned char SCSI_InquiryData[36]; /* Bytes 128-163 */ + unsigned char Reserved3[20]; /* Bytes 164-183 */ + unsigned char Reserved4[8]; /* Bytes 184-191 */ + DAC960_ByteCount64_T LastReadBlockNumber; /* Bytes 192-199 */ + DAC960_ByteCount64_T LastWrittenBlockNumber; /* Bytes 200-207 */ + DAC960_ByteCount64_T ConsistencyCheckBlockNumber; /* Bytes 208-215 */ + DAC960_ByteCount64_T RebuildBlockNumber; /* Bytes 216-223 */ + DAC960_ByteCount64_T MakingDataConsistentBlockNumber; /* Bytes 224-231 */ + DAC960_ByteCount64_T DeviceInitializationBlockNumber; /* Bytes 232-239 */ + DAC960_ByteCount64_T DataMigrationBlockNumber; /* Bytes 240-247 */ + DAC960_ByteCount64_T PatrolOperationBlockNumber; /* Bytes 248-255 */ + unsigned char Reserved5[256]; /* Bytes 256-511 */ +} +DAC960_V2_PhysicalDeviceInfo_T; + + +/* + Define the DAC960 V2 Firmware Health Status Buffer structure. +*/ + +typedef struct DAC960_V2_HealthStatusBuffer +{ + unsigned int MicrosecondsFromControllerStartTime; /* Bytes 0-3 */ + unsigned int MillisecondsFromControllerStartTime; /* Bytes 4-7 */ + unsigned int SecondsFrom1January1970; /* Bytes 8-11 */ + unsigned int :32; /* Bytes 12-15 */ + unsigned int StatusChangeCounter; /* Bytes 16-19 */ + unsigned int :32; /* Bytes 20-23 */ + unsigned int DebugOutputMessageBufferIndex; /* Bytes 24-27 */ + unsigned int CodedMessageBufferIndex; /* Bytes 28-31 */ + unsigned int CurrentTimeTracePageNumber; /* Bytes 32-35 */ + unsigned int CurrentProfilerPageNumber; /* Bytes 36-39 */ + unsigned int NextEventSequenceNumber; /* Bytes 40-43 */ + unsigned int :32; /* Bytes 44-47 */ + unsigned char Reserved1[16]; /* Bytes 48-63 */ + unsigned char Reserved2[64]; /* Bytes 64-127 */ +} +DAC960_V2_HealthStatusBuffer_T; + + +/* + Define the DAC960 V2 Firmware Get Event reply structure. +*/ + +typedef struct DAC960_V2_Event +{ + unsigned int EventSequenceNumber; /* Bytes 0-3 */ + unsigned int EventTime; /* Bytes 4-7 */ + unsigned int EventCode; /* Bytes 8-11 */ + unsigned char :8; /* Byte 12 */ + unsigned char Channel; /* Byte 13 */ + unsigned char TargetID; /* Byte 14 */ + unsigned char LogicalUnit; /* Byte 15 */ + unsigned int :32; /* Bytes 16-19 */ + unsigned int EventSpecificParameter; /* Bytes 20-23 */ + unsigned char RequestSenseData[40]; /* Bytes 24-63 */ +} +DAC960_V2_Event_T; + + +/* + Define the DAC960 V2 Firmware Command Control Bits structure. +*/ + +typedef struct DAC960_V2_CommandControlBits +{ + boolean ForceUnitAccess:1; /* Byte 0 Bit 0 */ + boolean DisablePageOut:1; /* Byte 0 Bit 1 */ + boolean :1; /* Byte 0 Bit 2 */ + boolean AdditionalScatterGatherListMemory:1; /* Byte 0 Bit 3 */ + boolean DataTransferControllerToHost:1; /* Byte 0 Bit 4 */ + boolean :1; /* Byte 0 Bit 5 */ + boolean NoAutoRequestSense:1; /* Byte 0 Bit 6 */ + boolean DisconnectProhibited:1; /* Byte 0 Bit 7 */ +} +DAC960_V2_CommandControlBits_T; + + +/* + Define the DAC960 V2 Firmware Command Timeout structure. +*/ + +typedef struct DAC960_V2_CommandTimeout +{ + unsigned char TimeoutValue:6; /* Byte 0 Bits 0-5 */ + enum { + DAC960_V2_TimeoutScale_Seconds = 0, + DAC960_V2_TimeoutScale_Minutes = 1, + DAC960_V2_TimeoutScale_Hours = 2, + DAC960_V2_TimeoutScale_Reserved = 3 + } __attribute__ ((packed)) TimeoutScale:2; /* Byte 0 Bits 6-7 */ +} +DAC960_V2_CommandTimeout_T; + + +/* + Define the DAC960 V2 Firmware Physical Device structure. +*/ + +typedef struct DAC960_V2_PhysicalDevice +{ + unsigned char LogicalUnit; /* Byte 0 */ + unsigned char TargetID; /* Byte 1 */ + unsigned char Channel:3; /* Byte 2 Bits 0-2 */ + unsigned char Controller:5; /* Byte 2 Bits 3-7 */ +} +__attribute__ ((packed)) +DAC960_V2_PhysicalDevice_T; + + +/* + Define the DAC960 V2 Firmware Logical Device structure. +*/ + +typedef struct DAC960_V2_LogicalDevice +{ + unsigned short LogicalDeviceNumber; /* Bytes 0-1 */ + unsigned char :3; /* Byte 2 Bits 0-2 */ + unsigned char Controller:5; /* Byte 2 Bits 3-7 */ +} +__attribute__ ((packed)) +DAC960_V2_LogicalDevice_T; + + +/* + Define the DAC960 V2 Firmware Operation Device type. +*/ + +typedef enum +{ + DAC960_V2_Physical_Device = 0x00, + DAC960_V2_RAID_Device = 0x01, + DAC960_V2_Physical_Channel = 0x02, + DAC960_V2_RAID_Channel = 0x03, + DAC960_V2_Physical_Controller = 0x04, + DAC960_V2_RAID_Controller = 0x05, + DAC960_V2_Configuration_Group = 0x10, + DAC960_V2_Enclosure = 0x11 +} +__attribute__ ((packed)) +DAC960_V2_OperationDevice_T; + + +/* + Define the DAC960 V2 Firmware Translate Physical To Logical Device structure. +*/ + +typedef struct DAC960_V2_PhysicalToLogicalDevice +{ + unsigned short LogicalDeviceNumber; /* Bytes 0-1 */ + unsigned short :16; /* Bytes 2-3 */ + unsigned char PreviousBootController; /* Byte 4 */ + unsigned char PreviousBootChannel; /* Byte 5 */ + unsigned char PreviousBootTargetID; /* Byte 6 */ + unsigned char PreviousBootLogicalUnit; /* Byte 7 */ +} +DAC960_V2_PhysicalToLogicalDevice_T; + + + +/* + Define the DAC960 V2 Firmware Scatter/Gather List Entry structure. +*/ + +typedef struct DAC960_V2_ScatterGatherSegment +{ + DAC960_BusAddress64_T SegmentDataPointer; /* Bytes 0-7 */ + DAC960_ByteCount64_T SegmentByteCount; /* Bytes 8-15 */ +} +DAC960_V2_ScatterGatherSegment_T; + + +/* + Define the DAC960 V2 Firmware Data Transfer Memory Address structure. +*/ + +typedef union DAC960_V2_DataTransferMemoryAddress +{ + DAC960_V2_ScatterGatherSegment_T ScatterGatherSegments[2]; /* Bytes 0-31 */ + struct { + unsigned short ScatterGatherList0Length; /* Bytes 0-1 */ + unsigned short ScatterGatherList1Length; /* Bytes 2-3 */ + unsigned short ScatterGatherList2Length; /* Bytes 4-5 */ + unsigned short :16; /* Bytes 6-7 */ + DAC960_BusAddress64_T ScatterGatherList0Address; /* Bytes 8-15 */ + DAC960_BusAddress64_T ScatterGatherList1Address; /* Bytes 16-23 */ + DAC960_BusAddress64_T ScatterGatherList2Address; /* Bytes 24-31 */ + } ExtendedScatterGather; +} +DAC960_V2_DataTransferMemoryAddress_T; + + +/* + Define the 64 Byte DAC960 V2 Firmware Command Mailbox structure. +*/ + +typedef union DAC960_V2_CommandMailbox +{ + unsigned int Words[16]; /* Words 0-15 */ + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + unsigned int :24; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + unsigned char Reserved[10]; /* Bytes 22-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } Common; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize; /* Bytes 4-7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + DAC960_V2_PhysicalDevice_T PhysicalDevice; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char CDBLength; /* Byte 21 */ + unsigned char SCSI_CDB[10]; /* Bytes 22-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } SCSI_10; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize; /* Bytes 4-7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + DAC960_V2_PhysicalDevice_T PhysicalDevice; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char CDBLength; /* Byte 21 */ + unsigned short :16; /* Bytes 22-23 */ + DAC960_BusAddress64_T SCSI_CDB_BusAddress; /* Bytes 24-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } SCSI_255; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + unsigned short :16; /* Bytes 16-17 */ + unsigned char ControllerNumber; /* Byte 18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + unsigned char Reserved[10]; /* Bytes 22-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } ControllerInfo; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + DAC960_V2_LogicalDevice_T LogicalDevice; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + unsigned char Reserved[10]; /* Bytes 22-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } LogicalDeviceInfo; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + DAC960_V2_PhysicalDevice_T PhysicalDevice; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + unsigned char Reserved[10]; /* Bytes 22-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } PhysicalDeviceInfo; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + unsigned short EventSequenceNumberHigh16; /* Bytes 16-17 */ + unsigned char ControllerNumber; /* Byte 18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + unsigned short EventSequenceNumberLow16; /* Bytes 22-23 */ + unsigned char Reserved[8]; /* Bytes 24-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } GetEvent; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + DAC960_V2_LogicalDevice_T LogicalDevice; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + union { + DAC960_V2_LogicalDeviceState_T LogicalDeviceState; + DAC960_V2_PhysicalDeviceState_T PhysicalDeviceState; + } DeviceState; /* Byte 22 */ + unsigned char Reserved[9]; /* Bytes 23-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } SetDeviceState; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + DAC960_V2_LogicalDevice_T LogicalDevice; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + boolean RestoreConsistency:1; /* Byte 22 Bit 0 */ + boolean InitializedAreaOnly:1; /* Byte 22 Bit 1 */ + unsigned char :6; /* Byte 22 Bits 2-7 */ + unsigned char Reserved[9]; /* Bytes 23-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } ConsistencyCheck; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + unsigned char FirstCommandMailboxSizeKB; /* Byte 4 */ + unsigned char FirstStatusMailboxSizeKB; /* Byte 5 */ + unsigned char SecondCommandMailboxSizeKB; /* Byte 6 */ + unsigned char SecondStatusMailboxSizeKB; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + unsigned int :24; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + unsigned char HealthStatusBufferSizeKB; /* Byte 22 */ + unsigned char :8; /* Byte 23 */ + DAC960_BusAddress64_T HealthStatusBufferBusAddress; /* Bytes 24-31 */ + DAC960_BusAddress64_T FirstCommandMailboxBusAddress; /* Bytes 32-39 */ + DAC960_BusAddress64_T FirstStatusMailboxBusAddress; /* Bytes 40-47 */ + DAC960_BusAddress64_T SecondCommandMailboxBusAddress; /* Bytes 48-55 */ + DAC960_BusAddress64_T SecondStatusMailboxBusAddress; /* Bytes 56-63 */ + } SetMemoryMailbox; + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandOpcode_T CommandOpcode; /* Byte 2 */ + DAC960_V2_CommandControlBits_T CommandControlBits; /* Byte 3 */ + DAC960_ByteCount32_T DataTransferSize:24; /* Bytes 4-6 */ + unsigned char DataTransferPageNumber; /* Byte 7 */ + DAC960_BusAddress64_T RequestSenseBusAddress; /* Bytes 8-15 */ + DAC960_V2_PhysicalDevice_T PhysicalDevice; /* Bytes 16-18 */ + DAC960_V2_CommandTimeout_T CommandTimeout; /* Byte 19 */ + unsigned char RequestSenseSize; /* Byte 20 */ + unsigned char IOCTL_Opcode; /* Byte 21 */ + DAC960_V2_OperationDevice_T OperationDevice; /* Byte 22 */ + unsigned char Reserved[9]; /* Bytes 23-31 */ + DAC960_V2_DataTransferMemoryAddress_T + DataTransferMemoryAddress; /* Bytes 32-63 */ + } DeviceOperation; +} +DAC960_V2_CommandMailbox_T; + + +/* + Define the DAC960 Driver IOCTL requests. +*/ + +#define DAC960_IOCTL_GET_CONTROLLER_COUNT 0xDAC001 +#define DAC960_IOCTL_GET_CONTROLLER_INFO 0xDAC002 +#define DAC960_IOCTL_V1_EXECUTE_COMMAND 0xDAC003 +#define DAC960_IOCTL_V2_EXECUTE_COMMAND 0xDAC004 +#define DAC960_IOCTL_V2_GET_HEALTH_STATUS 0xDAC005 + + +/* + Define the DAC960_IOCTL_GET_CONTROLLER_INFO reply structure. +*/ + +typedef struct DAC960_ControllerInfo +{ + unsigned char ControllerNumber; + unsigned char FirmwareType; + unsigned char Channels; + unsigned char Targets; + unsigned char PCI_Bus; + unsigned char PCI_Device; + unsigned char PCI_Function; + unsigned char IRQ_Channel; + DAC960_PCI_Address_T PCI_Address; + unsigned char ModelName[20]; + unsigned char FirmwareVersion[12]; +} +DAC960_ControllerInfo_T; + + +/* + Define the User Mode DAC960_IOCTL_V1_EXECUTE_COMMAND request structure. +*/ + +typedef struct DAC960_V1_UserCommand +{ + unsigned char ControllerNumber; + DAC960_V1_CommandMailbox_T CommandMailbox; + int DataTransferLength; + void __user *DataTransferBuffer; + DAC960_V1_DCDB_T __user *DCDB; +} +DAC960_V1_UserCommand_T; + + +/* + Define the Kernel Mode DAC960_IOCTL_V1_EXECUTE_COMMAND request structure. +*/ + +typedef struct DAC960_V1_KernelCommand +{ + unsigned char ControllerNumber; + DAC960_V1_CommandMailbox_T CommandMailbox; + int DataTransferLength; + void *DataTransferBuffer; + DAC960_V1_DCDB_T *DCDB; + DAC960_V1_CommandStatus_T CommandStatus; + void (*CompletionFunction)(struct DAC960_V1_KernelCommand *); + void *CompletionData; +} +DAC960_V1_KernelCommand_T; + + +/* + Define the User Mode DAC960_IOCTL_V2_EXECUTE_COMMAND request structure. +*/ + +typedef struct DAC960_V2_UserCommand +{ + unsigned char ControllerNumber; + DAC960_V2_CommandMailbox_T CommandMailbox; + int DataTransferLength; + int RequestSenseLength; + void __user *DataTransferBuffer; + void __user *RequestSenseBuffer; +} +DAC960_V2_UserCommand_T; + + +/* + Define the Kernel Mode DAC960_IOCTL_V2_EXECUTE_COMMAND request structure. +*/ + +typedef struct DAC960_V2_KernelCommand +{ + unsigned char ControllerNumber; + DAC960_V2_CommandMailbox_T CommandMailbox; + int DataTransferLength; + int RequestSenseLength; + void *DataTransferBuffer; + void *RequestSenseBuffer; + DAC960_V2_CommandStatus_T CommandStatus; + void (*CompletionFunction)(struct DAC960_V2_KernelCommand *); + void *CompletionData; +} +DAC960_V2_KernelCommand_T; + + +/* + Define the User Mode DAC960_IOCTL_V2_GET_HEALTH_STATUS request structure. +*/ + +typedef struct DAC960_V2_GetHealthStatus +{ + unsigned char ControllerNumber; + DAC960_V2_HealthStatusBuffer_T __user *HealthStatusBuffer; +} +DAC960_V2_GetHealthStatus_T; + + +/* + Import the Kernel Mode IOCTL interface. +*/ + +extern int DAC960_KernelIOCTL(unsigned int Request, void *Argument); + + +/* + DAC960_DriverVersion protects the private portion of this file. +*/ + +#ifdef DAC960_DriverVersion + + +/* + Define the maximum Driver Queue Depth and Controller Queue Depth supported + by DAC960 V1 and V2 Firmware Controllers. +*/ + +#define DAC960_MaxDriverQueueDepth 511 +#define DAC960_MaxControllerQueueDepth 512 + + +/* + Define the maximum number of Scatter/Gather Segments supported for any + DAC960 V1 and V2 Firmware controller. +*/ + +#define DAC960_V1_ScatterGatherLimit 33 +#define DAC960_V2_ScatterGatherLimit 128 + + +/* + Define the number of Command Mailboxes and Status Mailboxes used by the + DAC960 V1 and V2 Firmware Memory Mailbox Interface. +*/ + +#define DAC960_V1_CommandMailboxCount 256 +#define DAC960_V1_StatusMailboxCount 1024 +#define DAC960_V2_CommandMailboxCount 512 +#define DAC960_V2_StatusMailboxCount 512 + + +/* + Define the DAC960 Controller Monitoring Timer Interval. +*/ + +#define DAC960_MonitoringTimerInterval (10 * HZ) + + +/* + Define the DAC960 Controller Secondary Monitoring Interval. +*/ + +#define DAC960_SecondaryMonitoringInterval (60 * HZ) + + +/* + Define the DAC960 Controller Health Status Monitoring Interval. +*/ + +#define DAC960_HealthStatusMonitoringInterval (1 * HZ) + + +/* + Define the DAC960 Controller Progress Reporting Interval. +*/ + +#define DAC960_ProgressReportingInterval (60 * HZ) + + +/* + Define the maximum number of Partitions allowed for each Logical Drive. +*/ + +#define DAC960_MaxPartitions 8 +#define DAC960_MaxPartitionsBits 3 + +/* + Define the DAC960 Controller fixed Block Size and Block Size Bits. +*/ + +#define DAC960_BlockSize 512 +#define DAC960_BlockSizeBits 9 + + +/* + Define the number of Command structures that should be allocated as a + group to optimize kernel memory allocation. +*/ + +#define DAC960_V1_CommandAllocationGroupSize 11 +#define DAC960_V2_CommandAllocationGroupSize 29 + + +/* + Define the Controller Line Buffer, Progress Buffer, User Message, and + Initial Status Buffer sizes. +*/ + +#define DAC960_LineBufferSize 100 +#define DAC960_ProgressBufferSize 200 +#define DAC960_UserMessageSize 200 +#define DAC960_InitialStatusBufferSize (8192-32) + + +/* + Define the DAC960 Controller Firmware Types. +*/ + +typedef enum +{ + DAC960_V1_Controller = 1, + DAC960_V2_Controller = 2 +} +DAC960_FirmwareType_T; + + +/* + Define the DAC960 Controller Hardware Types. +*/ + +typedef enum +{ + DAC960_BA_Controller = 1, /* eXtremeRAID 2000 */ + DAC960_LP_Controller = 2, /* AcceleRAID 352 */ + DAC960_LA_Controller = 3, /* DAC1164P */ + DAC960_PG_Controller = 4, /* DAC960PTL/PJ/PG */ + DAC960_PD_Controller = 5, /* DAC960PU/PD/PL/P */ + DAC960_P_Controller = 6 /* DAC960PU/PD/PL/P */ +} +DAC960_HardwareType_T; + + +/* + Define the Driver Message Levels. +*/ + +typedef enum DAC960_MessageLevel +{ + DAC960_AnnounceLevel = 0, + DAC960_InfoLevel = 1, + DAC960_NoticeLevel = 2, + DAC960_WarningLevel = 3, + DAC960_ErrorLevel = 4, + DAC960_ProgressLevel = 5, + DAC960_CriticalLevel = 6, + DAC960_UserCriticalLevel = 7 +} +DAC960_MessageLevel_T; + +static char + *DAC960_MessageLevelMap[] = + { KERN_NOTICE, KERN_NOTICE, KERN_NOTICE, KERN_WARNING, + KERN_ERR, KERN_CRIT, KERN_CRIT, KERN_CRIT }; + + +/* + Define Driver Message macros. +*/ + +#define DAC960_Announce(Format, Arguments...) \ + DAC960_Message(DAC960_AnnounceLevel, Format, ##Arguments) + +#define DAC960_Info(Format, Arguments...) \ + DAC960_Message(DAC960_InfoLevel, Format, ##Arguments) + +#define DAC960_Notice(Format, Arguments...) \ + DAC960_Message(DAC960_NoticeLevel, Format, ##Arguments) + +#define DAC960_Warning(Format, Arguments...) \ + DAC960_Message(DAC960_WarningLevel, Format, ##Arguments) + +#define DAC960_Error(Format, Arguments...) \ + DAC960_Message(DAC960_ErrorLevel, Format, ##Arguments) + +#define DAC960_Progress(Format, Arguments...) \ + DAC960_Message(DAC960_ProgressLevel, Format, ##Arguments) + +#define DAC960_Critical(Format, Arguments...) \ + DAC960_Message(DAC960_CriticalLevel, Format, ##Arguments) + +#define DAC960_UserCritical(Format, Arguments...) \ + DAC960_Message(DAC960_UserCriticalLevel, Format, ##Arguments) + + +struct DAC960_privdata { + DAC960_HardwareType_T HardwareType; + DAC960_FirmwareType_T FirmwareType; + irqreturn_t (*InterruptHandler)(int, void *, struct pt_regs *); + unsigned int MemoryWindowSize; +}; + + +/* + Define the DAC960 V1 Firmware Controller Status Mailbox structure. +*/ + +typedef union DAC960_V1_StatusMailbox +{ + unsigned int Word; /* Word 0 */ + struct { + DAC960_V1_CommandIdentifier_T CommandIdentifier; /* Byte 0 */ + unsigned char :7; /* Byte 1 Bits 0-6 */ + boolean Valid:1; /* Byte 1 Bit 7 */ + DAC960_V1_CommandStatus_T CommandStatus; /* Bytes 2-3 */ + } Fields; +} +DAC960_V1_StatusMailbox_T; + + +/* + Define the DAC960 V2 Firmware Controller Status Mailbox structure. +*/ + +typedef union DAC960_V2_StatusMailbox +{ + unsigned int Words[2]; /* Words 0-1 */ + struct { + DAC960_V2_CommandIdentifier_T CommandIdentifier; /* Bytes 0-1 */ + DAC960_V2_CommandStatus_T CommandStatus; /* Byte 2 */ + unsigned char RequestSenseLength; /* Byte 3 */ + int DataTransferResidue; /* Bytes 4-7 */ + } Fields; +} +DAC960_V2_StatusMailbox_T; + + +/* + Define the DAC960 Driver Command Types. +*/ + +typedef enum +{ + DAC960_ReadCommand = 1, + DAC960_WriteCommand = 2, + DAC960_ReadRetryCommand = 3, + DAC960_WriteRetryCommand = 4, + DAC960_MonitoringCommand = 5, + DAC960_ImmediateCommand = 6, + DAC960_QueuedCommand = 7 +} +DAC960_CommandType_T; + + +/* + Define the DAC960 Driver Command structure. +*/ + +typedef struct DAC960_Command +{ + int CommandIdentifier; + DAC960_CommandType_T CommandType; + struct DAC960_Controller *Controller; + struct DAC960_Command *Next; + struct completion *Completion; + unsigned int LogicalDriveNumber; + unsigned int BlockNumber; + unsigned int BlockCount; + unsigned int SegmentCount; + int DmaDirection; + struct scatterlist *cmd_sglist; + struct request *Request; + union { + struct { + DAC960_V1_CommandMailbox_T CommandMailbox; + DAC960_V1_KernelCommand_T *KernelCommand; + DAC960_V1_CommandStatus_T CommandStatus; + DAC960_V1_ScatterGatherSegment_T *ScatterGatherList; + dma_addr_t ScatterGatherListDMA; + struct scatterlist ScatterList[DAC960_V1_ScatterGatherLimit]; + unsigned int EndMarker[0]; + } V1; + struct { + DAC960_V2_CommandMailbox_T CommandMailbox; + DAC960_V2_KernelCommand_T *KernelCommand; + DAC960_V2_CommandStatus_T CommandStatus; + unsigned char RequestSenseLength; + int DataTransferResidue; + DAC960_V2_ScatterGatherSegment_T *ScatterGatherList; + dma_addr_t ScatterGatherListDMA; + DAC960_SCSI_RequestSense_T *RequestSense; + dma_addr_t RequestSenseDMA; + struct scatterlist ScatterList[DAC960_V2_ScatterGatherLimit]; + unsigned int EndMarker[0]; + } V2; + } FW; +} +DAC960_Command_T; + + +/* + Define the DAC960 Driver Controller structure. +*/ + +typedef struct DAC960_Controller +{ + void __iomem *BaseAddress; + void __iomem *MemoryMappedAddress; + DAC960_FirmwareType_T FirmwareType; + DAC960_HardwareType_T HardwareType; + DAC960_IO_Address_T IO_Address; + DAC960_PCI_Address_T PCI_Address; + struct pci_dev *PCIDevice; + unsigned char ControllerNumber; + unsigned char ControllerName[4]; + unsigned char ModelName[20]; + unsigned char FullModelName[28]; + unsigned char FirmwareVersion[12]; + unsigned char Bus; + unsigned char Device; + unsigned char Function; + unsigned char IRQ_Channel; + unsigned char Channels; + unsigned char Targets; + unsigned char MemorySize; + unsigned char LogicalDriveCount; + unsigned short CommandAllocationGroupSize; + unsigned short ControllerQueueDepth; + unsigned short DriverQueueDepth; + unsigned short MaxBlocksPerCommand; + unsigned short ControllerScatterGatherLimit; + unsigned short DriverScatterGatherLimit; + u64 BounceBufferLimit; + unsigned int CombinedStatusBufferLength; + unsigned int InitialStatusLength; + unsigned int CurrentStatusLength; + unsigned int ProgressBufferLength; + unsigned int UserStatusLength; + struct dma_loaf DmaPages; + unsigned long MonitoringTimerCount; + unsigned long PrimaryMonitoringTime; + unsigned long SecondaryMonitoringTime; + unsigned long ShutdownMonitoringTimer; + unsigned long LastProgressReportTime; + unsigned long LastCurrentStatusTime; + boolean ControllerInitialized; + boolean MonitoringCommandDeferred; + boolean EphemeralProgressMessage; + boolean DriveSpinUpMessageDisplayed; + boolean MonitoringAlertMode; + boolean SuppressEnclosureMessages; + struct timer_list MonitoringTimer; + struct gendisk *disks[DAC960_MaxLogicalDrives]; + struct pci_pool *ScatterGatherPool; + DAC960_Command_T *FreeCommands; + unsigned char *CombinedStatusBuffer; + unsigned char *CurrentStatusBuffer; + struct request_queue *RequestQueue[DAC960_MaxLogicalDrives]; + int req_q_index; + spinlock_t queue_lock; + wait_queue_head_t CommandWaitQueue; + wait_queue_head_t HealthStatusWaitQueue; + DAC960_Command_T InitialCommand; + DAC960_Command_T *Commands[DAC960_MaxDriverQueueDepth]; + struct proc_dir_entry *ControllerProcEntry; + boolean LogicalDriveInitiallyAccessible[DAC960_MaxLogicalDrives]; + void (*QueueCommand)(DAC960_Command_T *Command); + boolean (*ReadControllerConfiguration)(struct DAC960_Controller *); + boolean (*ReadDeviceConfiguration)(struct DAC960_Controller *); + boolean (*ReportDeviceConfiguration)(struct DAC960_Controller *); + void (*QueueReadWriteCommand)(DAC960_Command_T *Command); + union { + struct { + unsigned char GeometryTranslationHeads; + unsigned char GeometryTranslationSectors; + unsigned char PendingRebuildFlag; + unsigned short StripeSize; + unsigned short SegmentSize; + unsigned short NewEventLogSequenceNumber; + unsigned short OldEventLogSequenceNumber; + unsigned short DeviceStateChannel; + unsigned short DeviceStateTargetID; + boolean DualModeMemoryMailboxInterface; + boolean BackgroundInitializationStatusSupported; + boolean SAFTE_EnclosureManagementEnabled; + boolean NeedLogicalDriveInformation; + boolean NeedErrorTableInformation; + boolean NeedDeviceStateInformation; + boolean NeedDeviceInquiryInformation; + boolean NeedDeviceSerialNumberInformation; + boolean NeedRebuildProgress; + boolean NeedConsistencyCheckProgress; + boolean NeedBackgroundInitializationStatus; + boolean StartDeviceStateScan; + boolean RebuildProgressFirst; + boolean RebuildFlagPending; + boolean RebuildStatusPending; + + dma_addr_t FirstCommandMailboxDMA; + DAC960_V1_CommandMailbox_T *FirstCommandMailbox; + DAC960_V1_CommandMailbox_T *LastCommandMailbox; + DAC960_V1_CommandMailbox_T *NextCommandMailbox; + DAC960_V1_CommandMailbox_T *PreviousCommandMailbox1; + DAC960_V1_CommandMailbox_T *PreviousCommandMailbox2; + + dma_addr_t FirstStatusMailboxDMA; + DAC960_V1_StatusMailbox_T *FirstStatusMailbox; + DAC960_V1_StatusMailbox_T *LastStatusMailbox; + DAC960_V1_StatusMailbox_T *NextStatusMailbox; + + DAC960_V1_DCDB_T *MonitoringDCDB; + dma_addr_t MonitoringDCDB_DMA; + + DAC960_V1_Enquiry_T Enquiry; + DAC960_V1_Enquiry_T *NewEnquiry; + dma_addr_t NewEnquiryDMA; + + DAC960_V1_ErrorTable_T ErrorTable; + DAC960_V1_ErrorTable_T *NewErrorTable; + dma_addr_t NewErrorTableDMA; + + DAC960_V1_EventLogEntry_T *EventLogEntry; + dma_addr_t EventLogEntryDMA; + + DAC960_V1_RebuildProgress_T *RebuildProgress; + dma_addr_t RebuildProgressDMA; + DAC960_V1_CommandStatus_T LastRebuildStatus; + DAC960_V1_CommandStatus_T PendingRebuildStatus; + + DAC960_V1_LogicalDriveInformationArray_T LogicalDriveInformation; + DAC960_V1_LogicalDriveInformationArray_T *NewLogicalDriveInformation; + dma_addr_t NewLogicalDriveInformationDMA; + + DAC960_V1_BackgroundInitializationStatus_T + *BackgroundInitializationStatus; + dma_addr_t BackgroundInitializationStatusDMA; + DAC960_V1_BackgroundInitializationStatus_T + LastBackgroundInitializationStatus; + + DAC960_V1_DeviceState_T + DeviceState[DAC960_V1_MaxChannels][DAC960_V1_MaxTargets]; + DAC960_V1_DeviceState_T *NewDeviceState; + dma_addr_t NewDeviceStateDMA; + + DAC960_SCSI_Inquiry_T + InquiryStandardData[DAC960_V1_MaxChannels][DAC960_V1_MaxTargets]; + DAC960_SCSI_Inquiry_T *NewInquiryStandardData; + dma_addr_t NewInquiryStandardDataDMA; + + DAC960_SCSI_Inquiry_UnitSerialNumber_T + InquiryUnitSerialNumber[DAC960_V1_MaxChannels][DAC960_V1_MaxTargets]; + DAC960_SCSI_Inquiry_UnitSerialNumber_T *NewInquiryUnitSerialNumber; + dma_addr_t NewInquiryUnitSerialNumberDMA; + + int DeviceResetCount[DAC960_V1_MaxChannels][DAC960_V1_MaxTargets]; + boolean DirectCommandActive[DAC960_V1_MaxChannels][DAC960_V1_MaxTargets]; + } V1; + struct { + unsigned int StatusChangeCounter; + unsigned int NextEventSequenceNumber; + unsigned int PhysicalDeviceIndex; + boolean NeedLogicalDeviceInformation; + boolean NeedPhysicalDeviceInformation; + boolean NeedDeviceSerialNumberInformation; + boolean StartLogicalDeviceInformationScan; + boolean StartPhysicalDeviceInformationScan; + struct pci_pool *RequestSensePool; + + dma_addr_t FirstCommandMailboxDMA; + DAC960_V2_CommandMailbox_T *FirstCommandMailbox; + DAC960_V2_CommandMailbox_T *LastCommandMailbox; + DAC960_V2_CommandMailbox_T *NextCommandMailbox; + DAC960_V2_CommandMailbox_T *PreviousCommandMailbox1; + DAC960_V2_CommandMailbox_T *PreviousCommandMailbox2; + + dma_addr_t FirstStatusMailboxDMA; + DAC960_V2_StatusMailbox_T *FirstStatusMailbox; + DAC960_V2_StatusMailbox_T *LastStatusMailbox; + DAC960_V2_StatusMailbox_T *NextStatusMailbox; + + dma_addr_t HealthStatusBufferDMA; + DAC960_V2_HealthStatusBuffer_T *HealthStatusBuffer; + + DAC960_V2_ControllerInfo_T ControllerInformation; + DAC960_V2_ControllerInfo_T *NewControllerInformation; + dma_addr_t NewControllerInformationDMA; + + DAC960_V2_LogicalDeviceInfo_T + *LogicalDeviceInformation[DAC960_MaxLogicalDrives]; + DAC960_V2_LogicalDeviceInfo_T *NewLogicalDeviceInformation; + dma_addr_t NewLogicalDeviceInformationDMA; + + DAC960_V2_PhysicalDeviceInfo_T + *PhysicalDeviceInformation[DAC960_V2_MaxPhysicalDevices]; + DAC960_V2_PhysicalDeviceInfo_T *NewPhysicalDeviceInformation; + dma_addr_t NewPhysicalDeviceInformationDMA; + + DAC960_SCSI_Inquiry_UnitSerialNumber_T *NewInquiryUnitSerialNumber; + dma_addr_t NewInquiryUnitSerialNumberDMA; + DAC960_SCSI_Inquiry_UnitSerialNumber_T + *InquiryUnitSerialNumber[DAC960_V2_MaxPhysicalDevices]; + + DAC960_V2_Event_T *Event; + dma_addr_t EventDMA; + + DAC960_V2_PhysicalToLogicalDevice_T *PhysicalToLogicalDevice; + dma_addr_t PhysicalToLogicalDeviceDMA; + + DAC960_V2_PhysicalDevice_T + LogicalDriveToVirtualDevice[DAC960_MaxLogicalDrives]; + boolean LogicalDriveFoundDuringScan[DAC960_MaxLogicalDrives]; + } V2; + } FW; + unsigned char ProgressBuffer[DAC960_ProgressBufferSize]; + unsigned char UserStatusBuffer[DAC960_UserMessageSize]; +} +DAC960_Controller_T; + + +/* + Simplify access to Firmware Version Dependent Data Structure Components + and Functions. +*/ + +#define V1 FW.V1 +#define V2 FW.V2 +#define DAC960_QueueCommand(Command) \ + (Controller->QueueCommand)(Command) +#define DAC960_ReadControllerConfiguration(Controller) \ + (Controller->ReadControllerConfiguration)(Controller) +#define DAC960_ReadDeviceConfiguration(Controller) \ + (Controller->ReadDeviceConfiguration)(Controller) +#define DAC960_ReportDeviceConfiguration(Controller) \ + (Controller->ReportDeviceConfiguration)(Controller) +#define DAC960_QueueReadWriteCommand(Command) \ + (Controller->QueueReadWriteCommand)(Command) + +/* + * dma_addr_writeql is provided to write dma_addr_t types + * to a 64-bit pci address space register. The controller + * will accept having the register written as two 32-bit + * values. + * + * In HIGHMEM kernels, dma_addr_t is a 64-bit value. + * without HIGHMEM, dma_addr_t is a 32-bit value. + * + * The compiler should always fix up the assignment + * to u.wq appropriately, depending upon the size of + * dma_addr_t. + */ +static inline +void dma_addr_writeql(dma_addr_t addr, void __iomem *write_address) +{ + union { + u64 wq; + uint wl[2]; + } u; + + u.wq = addr; + + writel(u.wl[0], write_address); + writel(u.wl[1], write_address + 4); +} + +/* + Define the DAC960 BA Series Controller Interface Register Offsets. +*/ + +#define DAC960_BA_RegisterWindowSize 0x80 + +typedef enum +{ + DAC960_BA_InboundDoorBellRegisterOffset = 0x60, + DAC960_BA_OutboundDoorBellRegisterOffset = 0x61, + DAC960_BA_InterruptStatusRegisterOffset = 0x30, + DAC960_BA_InterruptMaskRegisterOffset = 0x34, + DAC960_BA_CommandMailboxBusAddressOffset = 0x50, + DAC960_BA_CommandStatusOffset = 0x58, + DAC960_BA_ErrorStatusRegisterOffset = 0x63 +} +DAC960_BA_RegisterOffsets_T; + + +/* + Define the structure of the DAC960 BA Series Inbound Door Bell Register. +*/ + +typedef union DAC960_BA_InboundDoorBellRegister +{ + unsigned char All; + struct { + boolean HardwareMailboxNewCommand:1; /* Bit 0 */ + boolean AcknowledgeHardwareMailboxStatus:1; /* Bit 1 */ + boolean GenerateInterrupt:1; /* Bit 2 */ + boolean ControllerReset:1; /* Bit 3 */ + boolean MemoryMailboxNewCommand:1; /* Bit 4 */ + unsigned char :3; /* Bits 5-7 */ + } Write; + struct { + boolean HardwareMailboxEmpty:1; /* Bit 0 */ + boolean InitializationNotInProgress:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Read; +} +DAC960_BA_InboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 BA Series Outbound Door Bell Register. +*/ + +typedef union DAC960_BA_OutboundDoorBellRegister +{ + unsigned char All; + struct { + boolean AcknowledgeHardwareMailboxInterrupt:1; /* Bit 0 */ + boolean AcknowledgeMemoryMailboxInterrupt:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Write; + struct { + boolean HardwareMailboxStatusAvailable:1; /* Bit 0 */ + boolean MemoryMailboxStatusAvailable:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Read; +} +DAC960_BA_OutboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 BA Series Interrupt Mask Register. +*/ + +typedef union DAC960_BA_InterruptMaskRegister +{ + unsigned char All; + struct { + unsigned int :2; /* Bits 0-1 */ + boolean DisableInterrupts:1; /* Bit 2 */ + boolean DisableInterruptsI2O:1; /* Bit 3 */ + unsigned int :4; /* Bits 4-7 */ + } Bits; +} +DAC960_BA_InterruptMaskRegister_T; + + +/* + Define the structure of the DAC960 BA Series Error Status Register. +*/ + +typedef union DAC960_BA_ErrorStatusRegister +{ + unsigned char All; + struct { + unsigned int :2; /* Bits 0-1 */ + boolean ErrorStatusPending:1; /* Bit 2 */ + unsigned int :5; /* Bits 3-7 */ + } Bits; +} +DAC960_BA_ErrorStatusRegister_T; + + +/* + Define inline functions to provide an abstraction for reading and writing the + DAC960 BA Series Controller Interface Registers. +*/ + +static inline +void DAC960_BA_HardwareMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.HardwareMailboxNewCommand = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_BA_AcknowledgeHardwareMailboxStatus(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.AcknowledgeHardwareMailboxStatus = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_BA_GenerateInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.GenerateInterrupt = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_BA_ControllerReset(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.ControllerReset = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_BA_MemoryMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.MemoryMailboxNewCommand = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_InboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_BA_HardwareMailboxFullP(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_BA_InboundDoorBellRegisterOffset); + return !InboundDoorBellRegister.Read.HardwareMailboxEmpty; +} + +static inline +boolean DAC960_BA_InitializationInProgressP(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_BA_InboundDoorBellRegisterOffset); + return !InboundDoorBellRegister.Read.InitializationNotInProgress; +} + +static inline +void DAC960_BA_AcknowledgeHardwareMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_BA_AcknowledgeMemoryMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_BA_AcknowledgeInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_BA_OutboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_BA_HardwareMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_BA_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.HardwareMailboxStatusAvailable; +} + +static inline +boolean DAC960_BA_MemoryMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_BA_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.MemoryMailboxStatusAvailable; +} + +static inline +void DAC960_BA_EnableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0xFF; + InterruptMaskRegister.Bits.DisableInterrupts = false; + InterruptMaskRegister.Bits.DisableInterruptsI2O = true; + writeb(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_BA_InterruptMaskRegisterOffset); +} + +static inline +void DAC960_BA_DisableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0xFF; + InterruptMaskRegister.Bits.DisableInterrupts = true; + InterruptMaskRegister.Bits.DisableInterruptsI2O = true; + writeb(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_BA_InterruptMaskRegisterOffset); +} + +static inline +boolean DAC960_BA_InterruptsEnabledP(void __iomem *ControllerBaseAddress) +{ + DAC960_BA_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = + readb(ControllerBaseAddress + DAC960_BA_InterruptMaskRegisterOffset); + return !InterruptMaskRegister.Bits.DisableInterrupts; +} + +static inline +void DAC960_BA_WriteCommandMailbox(DAC960_V2_CommandMailbox_T + *MemoryCommandMailbox, + DAC960_V2_CommandMailbox_T + *CommandMailbox) +{ + memcpy(&MemoryCommandMailbox->Words[1], &CommandMailbox->Words[1], + sizeof(DAC960_V2_CommandMailbox_T) - sizeof(unsigned int)); + wmb(); + MemoryCommandMailbox->Words[0] = CommandMailbox->Words[0]; + mb(); +} + + +static inline +void DAC960_BA_WriteHardwareMailbox(void __iomem *ControllerBaseAddress, + dma_addr_t CommandMailboxDMA) +{ + dma_addr_writeql(CommandMailboxDMA, + ControllerBaseAddress + + DAC960_BA_CommandMailboxBusAddressOffset); +} + +static inline DAC960_V2_CommandIdentifier_T +DAC960_BA_ReadCommandIdentifier(void __iomem *ControllerBaseAddress) +{ + return readw(ControllerBaseAddress + DAC960_BA_CommandStatusOffset); +} + +static inline DAC960_V2_CommandStatus_T +DAC960_BA_ReadCommandStatus(void __iomem *ControllerBaseAddress) +{ + return readw(ControllerBaseAddress + DAC960_BA_CommandStatusOffset + 2); +} + +static inline boolean +DAC960_BA_ReadErrorStatus(void __iomem *ControllerBaseAddress, + unsigned char *ErrorStatus, + unsigned char *Parameter0, + unsigned char *Parameter1) +{ + DAC960_BA_ErrorStatusRegister_T ErrorStatusRegister; + ErrorStatusRegister.All = + readb(ControllerBaseAddress + DAC960_BA_ErrorStatusRegisterOffset); + if (!ErrorStatusRegister.Bits.ErrorStatusPending) return false; + ErrorStatusRegister.Bits.ErrorStatusPending = false; + *ErrorStatus = ErrorStatusRegister.All; + *Parameter0 = + readb(ControllerBaseAddress + DAC960_BA_CommandMailboxBusAddressOffset + 0); + *Parameter1 = + readb(ControllerBaseAddress + DAC960_BA_CommandMailboxBusAddressOffset + 1); + writeb(0xFF, ControllerBaseAddress + DAC960_BA_ErrorStatusRegisterOffset); + return true; +} + + +/* + Define the DAC960 LP Series Controller Interface Register Offsets. +*/ + +#define DAC960_LP_RegisterWindowSize 0x80 + +typedef enum +{ + DAC960_LP_InboundDoorBellRegisterOffset = 0x20, + DAC960_LP_OutboundDoorBellRegisterOffset = 0x2C, + DAC960_LP_InterruptStatusRegisterOffset = 0x30, + DAC960_LP_InterruptMaskRegisterOffset = 0x34, + DAC960_LP_CommandMailboxBusAddressOffset = 0x10, + DAC960_LP_CommandStatusOffset = 0x18, + DAC960_LP_ErrorStatusRegisterOffset = 0x2E +} +DAC960_LP_RegisterOffsets_T; + + +/* + Define the structure of the DAC960 LP Series Inbound Door Bell Register. +*/ + +typedef union DAC960_LP_InboundDoorBellRegister +{ + unsigned char All; + struct { + boolean HardwareMailboxNewCommand:1; /* Bit 0 */ + boolean AcknowledgeHardwareMailboxStatus:1; /* Bit 1 */ + boolean GenerateInterrupt:1; /* Bit 2 */ + boolean ControllerReset:1; /* Bit 3 */ + boolean MemoryMailboxNewCommand:1; /* Bit 4 */ + unsigned char :3; /* Bits 5-7 */ + } Write; + struct { + boolean HardwareMailboxFull:1; /* Bit 0 */ + boolean InitializationInProgress:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Read; +} +DAC960_LP_InboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 LP Series Outbound Door Bell Register. +*/ + +typedef union DAC960_LP_OutboundDoorBellRegister +{ + unsigned char All; + struct { + boolean AcknowledgeHardwareMailboxInterrupt:1; /* Bit 0 */ + boolean AcknowledgeMemoryMailboxInterrupt:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Write; + struct { + boolean HardwareMailboxStatusAvailable:1; /* Bit 0 */ + boolean MemoryMailboxStatusAvailable:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Read; +} +DAC960_LP_OutboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 LP Series Interrupt Mask Register. +*/ + +typedef union DAC960_LP_InterruptMaskRegister +{ + unsigned char All; + struct { + unsigned int :2; /* Bits 0-1 */ + boolean DisableInterrupts:1; /* Bit 2 */ + unsigned int :5; /* Bits 3-7 */ + } Bits; +} +DAC960_LP_InterruptMaskRegister_T; + + +/* + Define the structure of the DAC960 LP Series Error Status Register. +*/ + +typedef union DAC960_LP_ErrorStatusRegister +{ + unsigned char All; + struct { + unsigned int :2; /* Bits 0-1 */ + boolean ErrorStatusPending:1; /* Bit 2 */ + unsigned int :5; /* Bits 3-7 */ + } Bits; +} +DAC960_LP_ErrorStatusRegister_T; + + +/* + Define inline functions to provide an abstraction for reading and writing the + DAC960 LP Series Controller Interface Registers. +*/ + +static inline +void DAC960_LP_HardwareMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.HardwareMailboxNewCommand = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LP_AcknowledgeHardwareMailboxStatus(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.AcknowledgeHardwareMailboxStatus = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LP_GenerateInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.GenerateInterrupt = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LP_ControllerReset(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.ControllerReset = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LP_MemoryMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.MemoryMailboxNewCommand = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_InboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_LP_HardwareMailboxFullP(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LP_InboundDoorBellRegisterOffset); + return InboundDoorBellRegister.Read.HardwareMailboxFull; +} + +static inline +boolean DAC960_LP_InitializationInProgressP(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LP_InboundDoorBellRegisterOffset); + return InboundDoorBellRegister.Read.InitializationInProgress; +} + +static inline +void DAC960_LP_AcknowledgeHardwareMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LP_AcknowledgeMemoryMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LP_AcknowledgeInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LP_OutboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_LP_HardwareMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LP_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.HardwareMailboxStatusAvailable; +} + +static inline +boolean DAC960_LP_MemoryMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LP_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.MemoryMailboxStatusAvailable; +} + +static inline +void DAC960_LP_EnableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0xFF; + InterruptMaskRegister.Bits.DisableInterrupts = false; + writeb(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_LP_InterruptMaskRegisterOffset); +} + +static inline +void DAC960_LP_DisableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0xFF; + InterruptMaskRegister.Bits.DisableInterrupts = true; + writeb(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_LP_InterruptMaskRegisterOffset); +} + +static inline +boolean DAC960_LP_InterruptsEnabledP(void __iomem *ControllerBaseAddress) +{ + DAC960_LP_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = + readb(ControllerBaseAddress + DAC960_LP_InterruptMaskRegisterOffset); + return !InterruptMaskRegister.Bits.DisableInterrupts; +} + +static inline +void DAC960_LP_WriteCommandMailbox(DAC960_V2_CommandMailbox_T + *MemoryCommandMailbox, + DAC960_V2_CommandMailbox_T + *CommandMailbox) +{ + memcpy(&MemoryCommandMailbox->Words[1], &CommandMailbox->Words[1], + sizeof(DAC960_V2_CommandMailbox_T) - sizeof(unsigned int)); + wmb(); + MemoryCommandMailbox->Words[0] = CommandMailbox->Words[0]; + mb(); +} + +static inline +void DAC960_LP_WriteHardwareMailbox(void __iomem *ControllerBaseAddress, + dma_addr_t CommandMailboxDMA) +{ + dma_addr_writeql(CommandMailboxDMA, + ControllerBaseAddress + + DAC960_LP_CommandMailboxBusAddressOffset); +} + +static inline DAC960_V2_CommandIdentifier_T +DAC960_LP_ReadCommandIdentifier(void __iomem *ControllerBaseAddress) +{ + return readw(ControllerBaseAddress + DAC960_LP_CommandStatusOffset); +} + +static inline DAC960_V2_CommandStatus_T +DAC960_LP_ReadCommandStatus(void __iomem *ControllerBaseAddress) +{ + return readw(ControllerBaseAddress + DAC960_LP_CommandStatusOffset + 2); +} + +static inline boolean +DAC960_LP_ReadErrorStatus(void __iomem *ControllerBaseAddress, + unsigned char *ErrorStatus, + unsigned char *Parameter0, + unsigned char *Parameter1) +{ + DAC960_LP_ErrorStatusRegister_T ErrorStatusRegister; + ErrorStatusRegister.All = + readb(ControllerBaseAddress + DAC960_LP_ErrorStatusRegisterOffset); + if (!ErrorStatusRegister.Bits.ErrorStatusPending) return false; + ErrorStatusRegister.Bits.ErrorStatusPending = false; + *ErrorStatus = ErrorStatusRegister.All; + *Parameter0 = + readb(ControllerBaseAddress + DAC960_LP_CommandMailboxBusAddressOffset + 0); + *Parameter1 = + readb(ControllerBaseAddress + DAC960_LP_CommandMailboxBusAddressOffset + 1); + writeb(0xFF, ControllerBaseAddress + DAC960_LP_ErrorStatusRegisterOffset); + return true; +} + + +/* + Define the DAC960 LA Series Controller Interface Register Offsets. +*/ + +#define DAC960_LA_RegisterWindowSize 0x80 + +typedef enum +{ + DAC960_LA_InboundDoorBellRegisterOffset = 0x60, + DAC960_LA_OutboundDoorBellRegisterOffset = 0x61, + DAC960_LA_InterruptMaskRegisterOffset = 0x34, + DAC960_LA_CommandOpcodeRegisterOffset = 0x50, + DAC960_LA_CommandIdentifierRegisterOffset = 0x51, + DAC960_LA_MailboxRegister2Offset = 0x52, + DAC960_LA_MailboxRegister3Offset = 0x53, + DAC960_LA_MailboxRegister4Offset = 0x54, + DAC960_LA_MailboxRegister5Offset = 0x55, + DAC960_LA_MailboxRegister6Offset = 0x56, + DAC960_LA_MailboxRegister7Offset = 0x57, + DAC960_LA_MailboxRegister8Offset = 0x58, + DAC960_LA_MailboxRegister9Offset = 0x59, + DAC960_LA_MailboxRegister10Offset = 0x5A, + DAC960_LA_MailboxRegister11Offset = 0x5B, + DAC960_LA_MailboxRegister12Offset = 0x5C, + DAC960_LA_StatusCommandIdentifierRegOffset = 0x5D, + DAC960_LA_StatusRegisterOffset = 0x5E, + DAC960_LA_ErrorStatusRegisterOffset = 0x63 +} +DAC960_LA_RegisterOffsets_T; + + +/* + Define the structure of the DAC960 LA Series Inbound Door Bell Register. +*/ + +typedef union DAC960_LA_InboundDoorBellRegister +{ + unsigned char All; + struct { + boolean HardwareMailboxNewCommand:1; /* Bit 0 */ + boolean AcknowledgeHardwareMailboxStatus:1; /* Bit 1 */ + boolean GenerateInterrupt:1; /* Bit 2 */ + boolean ControllerReset:1; /* Bit 3 */ + boolean MemoryMailboxNewCommand:1; /* Bit 4 */ + unsigned char :3; /* Bits 5-7 */ + } Write; + struct { + boolean HardwareMailboxEmpty:1; /* Bit 0 */ + boolean InitializationNotInProgress:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Read; +} +DAC960_LA_InboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 LA Series Outbound Door Bell Register. +*/ + +typedef union DAC960_LA_OutboundDoorBellRegister +{ + unsigned char All; + struct { + boolean AcknowledgeHardwareMailboxInterrupt:1; /* Bit 0 */ + boolean AcknowledgeMemoryMailboxInterrupt:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Write; + struct { + boolean HardwareMailboxStatusAvailable:1; /* Bit 0 */ + boolean MemoryMailboxStatusAvailable:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Read; +} +DAC960_LA_OutboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 LA Series Interrupt Mask Register. +*/ + +typedef union DAC960_LA_InterruptMaskRegister +{ + unsigned char All; + struct { + unsigned char :2; /* Bits 0-1 */ + boolean DisableInterrupts:1; /* Bit 2 */ + unsigned char :5; /* Bits 3-7 */ + } Bits; +} +DAC960_LA_InterruptMaskRegister_T; + + +/* + Define the structure of the DAC960 LA Series Error Status Register. +*/ + +typedef union DAC960_LA_ErrorStatusRegister +{ + unsigned char All; + struct { + unsigned int :2; /* Bits 0-1 */ + boolean ErrorStatusPending:1; /* Bit 2 */ + unsigned int :5; /* Bits 3-7 */ + } Bits; +} +DAC960_LA_ErrorStatusRegister_T; + + +/* + Define inline functions to provide an abstraction for reading and writing the + DAC960 LA Series Controller Interface Registers. +*/ + +static inline +void DAC960_LA_HardwareMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.HardwareMailboxNewCommand = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LA_AcknowledgeHardwareMailboxStatus(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.AcknowledgeHardwareMailboxStatus = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LA_GenerateInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.GenerateInterrupt = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LA_ControllerReset(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.ControllerReset = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LA_MemoryMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.MemoryMailboxNewCommand = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_InboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_LA_HardwareMailboxFullP(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LA_InboundDoorBellRegisterOffset); + return !InboundDoorBellRegister.Read.HardwareMailboxEmpty; +} + +static inline +boolean DAC960_LA_InitializationInProgressP(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LA_InboundDoorBellRegisterOffset); + return !InboundDoorBellRegister.Read.InitializationNotInProgress; +} + +static inline +void DAC960_LA_AcknowledgeHardwareMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LA_AcknowledgeMemoryMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_LA_AcknowledgeInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_LA_OutboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_LA_HardwareMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LA_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.HardwareMailboxStatusAvailable; +} + +static inline +boolean DAC960_LA_MemoryMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_LA_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.MemoryMailboxStatusAvailable; +} + +static inline +void DAC960_LA_EnableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0xFF; + InterruptMaskRegister.Bits.DisableInterrupts = false; + writeb(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_LA_InterruptMaskRegisterOffset); +} + +static inline +void DAC960_LA_DisableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0xFF; + InterruptMaskRegister.Bits.DisableInterrupts = true; + writeb(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_LA_InterruptMaskRegisterOffset); +} + +static inline +boolean DAC960_LA_InterruptsEnabledP(void __iomem *ControllerBaseAddress) +{ + DAC960_LA_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = + readb(ControllerBaseAddress + DAC960_LA_InterruptMaskRegisterOffset); + return !InterruptMaskRegister.Bits.DisableInterrupts; +} + +static inline +void DAC960_LA_WriteCommandMailbox(DAC960_V1_CommandMailbox_T + *MemoryCommandMailbox, + DAC960_V1_CommandMailbox_T + *CommandMailbox) +{ + MemoryCommandMailbox->Words[1] = CommandMailbox->Words[1]; + MemoryCommandMailbox->Words[2] = CommandMailbox->Words[2]; + MemoryCommandMailbox->Words[3] = CommandMailbox->Words[3]; + wmb(); + MemoryCommandMailbox->Words[0] = CommandMailbox->Words[0]; + mb(); +} + +static inline +void DAC960_LA_WriteHardwareMailbox(void __iomem *ControllerBaseAddress, + DAC960_V1_CommandMailbox_T *CommandMailbox) +{ + writel(CommandMailbox->Words[0], + ControllerBaseAddress + DAC960_LA_CommandOpcodeRegisterOffset); + writel(CommandMailbox->Words[1], + ControllerBaseAddress + DAC960_LA_MailboxRegister4Offset); + writel(CommandMailbox->Words[2], + ControllerBaseAddress + DAC960_LA_MailboxRegister8Offset); + writeb(CommandMailbox->Bytes[12], + ControllerBaseAddress + DAC960_LA_MailboxRegister12Offset); +} + +static inline DAC960_V1_CommandIdentifier_T +DAC960_LA_ReadStatusCommandIdentifier(void __iomem *ControllerBaseAddress) +{ + return readb(ControllerBaseAddress + + DAC960_LA_StatusCommandIdentifierRegOffset); +} + +static inline DAC960_V1_CommandStatus_T +DAC960_LA_ReadStatusRegister(void __iomem *ControllerBaseAddress) +{ + return readw(ControllerBaseAddress + DAC960_LA_StatusRegisterOffset); +} + +static inline boolean +DAC960_LA_ReadErrorStatus(void __iomem *ControllerBaseAddress, + unsigned char *ErrorStatus, + unsigned char *Parameter0, + unsigned char *Parameter1) +{ + DAC960_LA_ErrorStatusRegister_T ErrorStatusRegister; + ErrorStatusRegister.All = + readb(ControllerBaseAddress + DAC960_LA_ErrorStatusRegisterOffset); + if (!ErrorStatusRegister.Bits.ErrorStatusPending) return false; + ErrorStatusRegister.Bits.ErrorStatusPending = false; + *ErrorStatus = ErrorStatusRegister.All; + *Parameter0 = + readb(ControllerBaseAddress + DAC960_LA_CommandOpcodeRegisterOffset); + *Parameter1 = + readb(ControllerBaseAddress + DAC960_LA_CommandIdentifierRegisterOffset); + writeb(0xFF, ControllerBaseAddress + DAC960_LA_ErrorStatusRegisterOffset); + return true; +} + +/* + Define the DAC960 PG Series Controller Interface Register Offsets. +*/ + +#define DAC960_PG_RegisterWindowSize 0x2000 + +typedef enum +{ + DAC960_PG_InboundDoorBellRegisterOffset = 0x0020, + DAC960_PG_OutboundDoorBellRegisterOffset = 0x002C, + DAC960_PG_InterruptMaskRegisterOffset = 0x0034, + DAC960_PG_CommandOpcodeRegisterOffset = 0x1000, + DAC960_PG_CommandIdentifierRegisterOffset = 0x1001, + DAC960_PG_MailboxRegister2Offset = 0x1002, + DAC960_PG_MailboxRegister3Offset = 0x1003, + DAC960_PG_MailboxRegister4Offset = 0x1004, + DAC960_PG_MailboxRegister5Offset = 0x1005, + DAC960_PG_MailboxRegister6Offset = 0x1006, + DAC960_PG_MailboxRegister7Offset = 0x1007, + DAC960_PG_MailboxRegister8Offset = 0x1008, + DAC960_PG_MailboxRegister9Offset = 0x1009, + DAC960_PG_MailboxRegister10Offset = 0x100A, + DAC960_PG_MailboxRegister11Offset = 0x100B, + DAC960_PG_MailboxRegister12Offset = 0x100C, + DAC960_PG_StatusCommandIdentifierRegOffset = 0x1018, + DAC960_PG_StatusRegisterOffset = 0x101A, + DAC960_PG_ErrorStatusRegisterOffset = 0x103F +} +DAC960_PG_RegisterOffsets_T; + + +/* + Define the structure of the DAC960 PG Series Inbound Door Bell Register. +*/ + +typedef union DAC960_PG_InboundDoorBellRegister +{ + unsigned int All; + struct { + boolean HardwareMailboxNewCommand:1; /* Bit 0 */ + boolean AcknowledgeHardwareMailboxStatus:1; /* Bit 1 */ + boolean GenerateInterrupt:1; /* Bit 2 */ + boolean ControllerReset:1; /* Bit 3 */ + boolean MemoryMailboxNewCommand:1; /* Bit 4 */ + unsigned int :27; /* Bits 5-31 */ + } Write; + struct { + boolean HardwareMailboxFull:1; /* Bit 0 */ + boolean InitializationInProgress:1; /* Bit 1 */ + unsigned int :30; /* Bits 2-31 */ + } Read; +} +DAC960_PG_InboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 PG Series Outbound Door Bell Register. +*/ + +typedef union DAC960_PG_OutboundDoorBellRegister +{ + unsigned int All; + struct { + boolean AcknowledgeHardwareMailboxInterrupt:1; /* Bit 0 */ + boolean AcknowledgeMemoryMailboxInterrupt:1; /* Bit 1 */ + unsigned int :30; /* Bits 2-31 */ + } Write; + struct { + boolean HardwareMailboxStatusAvailable:1; /* Bit 0 */ + boolean MemoryMailboxStatusAvailable:1; /* Bit 1 */ + unsigned int :30; /* Bits 2-31 */ + } Read; +} +DAC960_PG_OutboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 PG Series Interrupt Mask Register. +*/ + +typedef union DAC960_PG_InterruptMaskRegister +{ + unsigned int All; + struct { + unsigned int MessageUnitInterruptMask1:2; /* Bits 0-1 */ + boolean DisableInterrupts:1; /* Bit 2 */ + unsigned int MessageUnitInterruptMask2:5; /* Bits 3-7 */ + unsigned int Reserved0:24; /* Bits 8-31 */ + } Bits; +} +DAC960_PG_InterruptMaskRegister_T; + + +/* + Define the structure of the DAC960 PG Series Error Status Register. +*/ + +typedef union DAC960_PG_ErrorStatusRegister +{ + unsigned char All; + struct { + unsigned int :2; /* Bits 0-1 */ + boolean ErrorStatusPending:1; /* Bit 2 */ + unsigned int :5; /* Bits 3-7 */ + } Bits; +} +DAC960_PG_ErrorStatusRegister_T; + + +/* + Define inline functions to provide an abstraction for reading and writing the + DAC960 PG Series Controller Interface Registers. +*/ + +static inline +void DAC960_PG_HardwareMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.HardwareMailboxNewCommand = true; + writel(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PG_AcknowledgeHardwareMailboxStatus(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.AcknowledgeHardwareMailboxStatus = true; + writel(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PG_GenerateInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.GenerateInterrupt = true; + writel(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PG_ControllerReset(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.ControllerReset = true; + writel(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PG_MemoryMailboxNewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.MemoryMailboxNewCommand = true; + writel(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_InboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_PG_HardwareMailboxFullP(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readl(ControllerBaseAddress + DAC960_PG_InboundDoorBellRegisterOffset); + return InboundDoorBellRegister.Read.HardwareMailboxFull; +} + +static inline +boolean DAC960_PG_InitializationInProgressP(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readl(ControllerBaseAddress + DAC960_PG_InboundDoorBellRegisterOffset); + return InboundDoorBellRegister.Read.InitializationInProgress; +} + +static inline +void DAC960_PG_AcknowledgeHardwareMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + writel(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PG_AcknowledgeMemoryMailboxInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writel(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_OutboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PG_AcknowledgeInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeHardwareMailboxInterrupt = true; + OutboundDoorBellRegister.Write.AcknowledgeMemoryMailboxInterrupt = true; + writel(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PG_OutboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_PG_HardwareMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readl(ControllerBaseAddress + DAC960_PG_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.HardwareMailboxStatusAvailable; +} + +static inline +boolean DAC960_PG_MemoryMailboxStatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readl(ControllerBaseAddress + DAC960_PG_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.MemoryMailboxStatusAvailable; +} + +static inline +void DAC960_PG_EnableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0; + InterruptMaskRegister.Bits.MessageUnitInterruptMask1 = 0x3; + InterruptMaskRegister.Bits.DisableInterrupts = false; + InterruptMaskRegister.Bits.MessageUnitInterruptMask2 = 0x1F; + writel(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_PG_InterruptMaskRegisterOffset); +} + +static inline +void DAC960_PG_DisableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = 0; + InterruptMaskRegister.Bits.MessageUnitInterruptMask1 = 0x3; + InterruptMaskRegister.Bits.DisableInterrupts = true; + InterruptMaskRegister.Bits.MessageUnitInterruptMask2 = 0x1F; + writel(InterruptMaskRegister.All, + ControllerBaseAddress + DAC960_PG_InterruptMaskRegisterOffset); +} + +static inline +boolean DAC960_PG_InterruptsEnabledP(void __iomem *ControllerBaseAddress) +{ + DAC960_PG_InterruptMaskRegister_T InterruptMaskRegister; + InterruptMaskRegister.All = + readl(ControllerBaseAddress + DAC960_PG_InterruptMaskRegisterOffset); + return !InterruptMaskRegister.Bits.DisableInterrupts; +} + +static inline +void DAC960_PG_WriteCommandMailbox(DAC960_V1_CommandMailbox_T + *MemoryCommandMailbox, + DAC960_V1_CommandMailbox_T + *CommandMailbox) +{ + MemoryCommandMailbox->Words[1] = CommandMailbox->Words[1]; + MemoryCommandMailbox->Words[2] = CommandMailbox->Words[2]; + MemoryCommandMailbox->Words[3] = CommandMailbox->Words[3]; + wmb(); + MemoryCommandMailbox->Words[0] = CommandMailbox->Words[0]; + mb(); +} + +static inline +void DAC960_PG_WriteHardwareMailbox(void __iomem *ControllerBaseAddress, + DAC960_V1_CommandMailbox_T *CommandMailbox) +{ + writel(CommandMailbox->Words[0], + ControllerBaseAddress + DAC960_PG_CommandOpcodeRegisterOffset); + writel(CommandMailbox->Words[1], + ControllerBaseAddress + DAC960_PG_MailboxRegister4Offset); + writel(CommandMailbox->Words[2], + ControllerBaseAddress + DAC960_PG_MailboxRegister8Offset); + writeb(CommandMailbox->Bytes[12], + ControllerBaseAddress + DAC960_PG_MailboxRegister12Offset); +} + +static inline DAC960_V1_CommandIdentifier_T +DAC960_PG_ReadStatusCommandIdentifier(void __iomem *ControllerBaseAddress) +{ + return readb(ControllerBaseAddress + + DAC960_PG_StatusCommandIdentifierRegOffset); +} + +static inline DAC960_V1_CommandStatus_T +DAC960_PG_ReadStatusRegister(void __iomem *ControllerBaseAddress) +{ + return readw(ControllerBaseAddress + DAC960_PG_StatusRegisterOffset); +} + +static inline boolean +DAC960_PG_ReadErrorStatus(void __iomem *ControllerBaseAddress, + unsigned char *ErrorStatus, + unsigned char *Parameter0, + unsigned char *Parameter1) +{ + DAC960_PG_ErrorStatusRegister_T ErrorStatusRegister; + ErrorStatusRegister.All = + readb(ControllerBaseAddress + DAC960_PG_ErrorStatusRegisterOffset); + if (!ErrorStatusRegister.Bits.ErrorStatusPending) return false; + ErrorStatusRegister.Bits.ErrorStatusPending = false; + *ErrorStatus = ErrorStatusRegister.All; + *Parameter0 = + readb(ControllerBaseAddress + DAC960_PG_CommandOpcodeRegisterOffset); + *Parameter1 = + readb(ControllerBaseAddress + DAC960_PG_CommandIdentifierRegisterOffset); + writeb(0, ControllerBaseAddress + DAC960_PG_ErrorStatusRegisterOffset); + return true; +} + +/* + Define the DAC960 PD Series Controller Interface Register Offsets. +*/ + +#define DAC960_PD_RegisterWindowSize 0x80 + +typedef enum +{ + DAC960_PD_CommandOpcodeRegisterOffset = 0x00, + DAC960_PD_CommandIdentifierRegisterOffset = 0x01, + DAC960_PD_MailboxRegister2Offset = 0x02, + DAC960_PD_MailboxRegister3Offset = 0x03, + DAC960_PD_MailboxRegister4Offset = 0x04, + DAC960_PD_MailboxRegister5Offset = 0x05, + DAC960_PD_MailboxRegister6Offset = 0x06, + DAC960_PD_MailboxRegister7Offset = 0x07, + DAC960_PD_MailboxRegister8Offset = 0x08, + DAC960_PD_MailboxRegister9Offset = 0x09, + DAC960_PD_MailboxRegister10Offset = 0x0A, + DAC960_PD_MailboxRegister11Offset = 0x0B, + DAC960_PD_MailboxRegister12Offset = 0x0C, + DAC960_PD_StatusCommandIdentifierRegOffset = 0x0D, + DAC960_PD_StatusRegisterOffset = 0x0E, + DAC960_PD_ErrorStatusRegisterOffset = 0x3F, + DAC960_PD_InboundDoorBellRegisterOffset = 0x40, + DAC960_PD_OutboundDoorBellRegisterOffset = 0x41, + DAC960_PD_InterruptEnableRegisterOffset = 0x43 +} +DAC960_PD_RegisterOffsets_T; + + +/* + Define the structure of the DAC960 PD Series Inbound Door Bell Register. +*/ + +typedef union DAC960_PD_InboundDoorBellRegister +{ + unsigned char All; + struct { + boolean NewCommand:1; /* Bit 0 */ + boolean AcknowledgeStatus:1; /* Bit 1 */ + boolean GenerateInterrupt:1; /* Bit 2 */ + boolean ControllerReset:1; /* Bit 3 */ + unsigned char :4; /* Bits 4-7 */ + } Write; + struct { + boolean MailboxFull:1; /* Bit 0 */ + boolean InitializationInProgress:1; /* Bit 1 */ + unsigned char :6; /* Bits 2-7 */ + } Read; +} +DAC960_PD_InboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 PD Series Outbound Door Bell Register. +*/ + +typedef union DAC960_PD_OutboundDoorBellRegister +{ + unsigned char All; + struct { + boolean AcknowledgeInterrupt:1; /* Bit 0 */ + unsigned char :7; /* Bits 1-7 */ + } Write; + struct { + boolean StatusAvailable:1; /* Bit 0 */ + unsigned char :7; /* Bits 1-7 */ + } Read; +} +DAC960_PD_OutboundDoorBellRegister_T; + + +/* + Define the structure of the DAC960 PD Series Interrupt Enable Register. +*/ + +typedef union DAC960_PD_InterruptEnableRegister +{ + unsigned char All; + struct { + boolean EnableInterrupts:1; /* Bit 0 */ + unsigned char :7; /* Bits 1-7 */ + } Bits; +} +DAC960_PD_InterruptEnableRegister_T; + + +/* + Define the structure of the DAC960 PD Series Error Status Register. +*/ + +typedef union DAC960_PD_ErrorStatusRegister +{ + unsigned char All; + struct { + unsigned int :2; /* Bits 0-1 */ + boolean ErrorStatusPending:1; /* Bit 2 */ + unsigned int :5; /* Bits 3-7 */ + } Bits; +} +DAC960_PD_ErrorStatusRegister_T; + + +/* + Define inline functions to provide an abstraction for reading and writing the + DAC960 PD Series Controller Interface Registers. +*/ + +static inline +void DAC960_PD_NewCommand(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.NewCommand = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PD_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PD_AcknowledgeStatus(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.AcknowledgeStatus = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PD_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PD_GenerateInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.GenerateInterrupt = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PD_InboundDoorBellRegisterOffset); +} + +static inline +void DAC960_PD_ControllerReset(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = 0; + InboundDoorBellRegister.Write.ControllerReset = true; + writeb(InboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PD_InboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_PD_MailboxFullP(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_PD_InboundDoorBellRegisterOffset); + return InboundDoorBellRegister.Read.MailboxFull; +} + +static inline +boolean DAC960_PD_InitializationInProgressP(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InboundDoorBellRegister_T InboundDoorBellRegister; + InboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_PD_InboundDoorBellRegisterOffset); + return InboundDoorBellRegister.Read.InitializationInProgress; +} + +static inline +void DAC960_PD_AcknowledgeInterrupt(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = 0; + OutboundDoorBellRegister.Write.AcknowledgeInterrupt = true; + writeb(OutboundDoorBellRegister.All, + ControllerBaseAddress + DAC960_PD_OutboundDoorBellRegisterOffset); +} + +static inline +boolean DAC960_PD_StatusAvailableP(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_OutboundDoorBellRegister_T OutboundDoorBellRegister; + OutboundDoorBellRegister.All = + readb(ControllerBaseAddress + DAC960_PD_OutboundDoorBellRegisterOffset); + return OutboundDoorBellRegister.Read.StatusAvailable; +} + +static inline +void DAC960_PD_EnableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InterruptEnableRegister_T InterruptEnableRegister; + InterruptEnableRegister.All = 0; + InterruptEnableRegister.Bits.EnableInterrupts = true; + writeb(InterruptEnableRegister.All, + ControllerBaseAddress + DAC960_PD_InterruptEnableRegisterOffset); +} + +static inline +void DAC960_PD_DisableInterrupts(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InterruptEnableRegister_T InterruptEnableRegister; + InterruptEnableRegister.All = 0; + InterruptEnableRegister.Bits.EnableInterrupts = false; + writeb(InterruptEnableRegister.All, + ControllerBaseAddress + DAC960_PD_InterruptEnableRegisterOffset); +} + +static inline +boolean DAC960_PD_InterruptsEnabledP(void __iomem *ControllerBaseAddress) +{ + DAC960_PD_InterruptEnableRegister_T InterruptEnableRegister; + InterruptEnableRegister.All = + readb(ControllerBaseAddress + DAC960_PD_InterruptEnableRegisterOffset); + return InterruptEnableRegister.Bits.EnableInterrupts; +} + +static inline +void DAC960_PD_WriteCommandMailbox(void __iomem *ControllerBaseAddress, + DAC960_V1_CommandMailbox_T *CommandMailbox) +{ + writel(CommandMailbox->Words[0], + ControllerBaseAddress + DAC960_PD_CommandOpcodeRegisterOffset); + writel(CommandMailbox->Words[1], + ControllerBaseAddress + DAC960_PD_MailboxRegister4Offset); + writel(CommandMailbox->Words[2], + ControllerBaseAddress + DAC960_PD_MailboxRegister8Offset); + writeb(CommandMailbox->Bytes[12], + ControllerBaseAddress + DAC960_PD_MailboxRegister12Offset); +} + +static inline DAC960_V1_CommandIdentifier_T +DAC960_PD_ReadStatusCommandIdentifier(void __iomem *ControllerBaseAddress) +{ + return readb(ControllerBaseAddress + + DAC960_PD_StatusCommandIdentifierRegOffset); +} + +static inline DAC960_V1_CommandStatus_T +DAC960_PD_ReadStatusRegister(void __iomem *ControllerBaseAddress) +{ + return readw(ControllerBaseAddress + DAC960_PD_StatusRegisterOffset); +} + +static inline boolean +DAC960_PD_ReadErrorStatus(void __iomem *ControllerBaseAddress, + unsigned char *ErrorStatus, + unsigned char *Parameter0, + unsigned char *Parameter1) +{ + DAC960_PD_ErrorStatusRegister_T ErrorStatusRegister; + ErrorStatusRegister.All = + readb(ControllerBaseAddress + DAC960_PD_ErrorStatusRegisterOffset); + if (!ErrorStatusRegister.Bits.ErrorStatusPending) return false; + ErrorStatusRegister.Bits.ErrorStatusPending = false; + *ErrorStatus = ErrorStatusRegister.All; + *Parameter0 = + readb(ControllerBaseAddress + DAC960_PD_CommandOpcodeRegisterOffset); + *Parameter1 = + readb(ControllerBaseAddress + DAC960_PD_CommandIdentifierRegisterOffset); + writeb(0, ControllerBaseAddress + DAC960_PD_ErrorStatusRegisterOffset); + return true; +} + +static inline void DAC960_P_To_PD_TranslateEnquiry(void *Enquiry) +{ + memcpy(Enquiry + 132, Enquiry + 36, 64); + memset(Enquiry + 36, 0, 96); +} + +static inline void DAC960_P_To_PD_TranslateDeviceState(void *DeviceState) +{ + memcpy(DeviceState + 2, DeviceState + 3, 1); + memcpy(DeviceState + 4, DeviceState + 5, 2); + memcpy(DeviceState + 6, DeviceState + 8, 4); +} + +static inline +void DAC960_PD_To_P_TranslateReadWriteCommand(DAC960_V1_CommandMailbox_T + *CommandMailbox) +{ + int LogicalDriveNumber = CommandMailbox->Type5.LD.LogicalDriveNumber; + CommandMailbox->Bytes[3] &= 0x7; + CommandMailbox->Bytes[3] |= CommandMailbox->Bytes[7] << 6; + CommandMailbox->Bytes[7] = LogicalDriveNumber; +} + +static inline +void DAC960_P_To_PD_TranslateReadWriteCommand(DAC960_V1_CommandMailbox_T + *CommandMailbox) +{ + int LogicalDriveNumber = CommandMailbox->Bytes[7]; + CommandMailbox->Bytes[7] = CommandMailbox->Bytes[3] >> 6; + CommandMailbox->Bytes[3] &= 0x7; + CommandMailbox->Bytes[3] |= LogicalDriveNumber << 3; +} + + +/* + Define prototypes for the forward referenced DAC960 Driver Internal Functions. +*/ + +static void DAC960_FinalizeController(DAC960_Controller_T *); +static void DAC960_V1_QueueReadWriteCommand(DAC960_Command_T *); +static void DAC960_V2_QueueReadWriteCommand(DAC960_Command_T *); +static void DAC960_RequestFunction(struct request_queue *); +static irqreturn_t DAC960_BA_InterruptHandler(int, void *, struct pt_regs *); +static irqreturn_t DAC960_LP_InterruptHandler(int, void *, struct pt_regs *); +static irqreturn_t DAC960_LA_InterruptHandler(int, void *, struct pt_regs *); +static irqreturn_t DAC960_PG_InterruptHandler(int, void *, struct pt_regs *); +static irqreturn_t DAC960_PD_InterruptHandler(int, void *, struct pt_regs *); +static irqreturn_t DAC960_P_InterruptHandler(int, void *, struct pt_regs *); +static void DAC960_V1_QueueMonitoringCommand(DAC960_Command_T *); +static void DAC960_V2_QueueMonitoringCommand(DAC960_Command_T *); +static void DAC960_MonitoringTimerFunction(unsigned long); +static void DAC960_Message(DAC960_MessageLevel_T, unsigned char *, + DAC960_Controller_T *, ...); +static void DAC960_CreateProcEntries(DAC960_Controller_T *); +static void DAC960_DestroyProcEntries(DAC960_Controller_T *); + +#endif /* DAC960_DriverVersion */ diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig new file mode 100644 index 00000000000..e83a1e2e8b1 --- /dev/null +++ b/drivers/block/Kconfig @@ -0,0 +1,509 @@ +# +# Block device driver configuration +# + +menu "Block devices" + +config BLK_DEV_FD + tristate "Normal floppy disk support" + depends on (!ARCH_S390 && !M68K && !IA64 && !UML) || Q40 || (SUN3X && BROKEN) + ---help--- + If you want to use the floppy disk drive(s) of your PC under Linux, + say Y. Information about this driver, especially important for IBM + Thinkpad users, is contained in <file:Documentation/floppy.txt>. + That file also contains the location of the Floppy driver FAQ as + well as location of the fdutils package used to configure additional + parameters of the driver at run time. + + To compile this driver as a module, choose M here: the + module will be called floppy. + +config AMIGA_FLOPPY + tristate "Amiga floppy support" + depends on AMIGA + +config ATARI_FLOPPY + tristate "Atari floppy support" + depends on ATARI + +config BLK_DEV_SWIM_IOP + bool "Macintosh IIfx/Quadra 900/Quadra 950 floppy support (EXPERIMENTAL)" + depends on MAC && EXPERIMENTAL && BROKEN + help + Say Y here to support the SWIM (Super Woz Integrated Machine) IOP + floppy controller on the Macintosh IIfx and Quadra 900/950. + +config MAC_FLOPPY + tristate "Support for PowerMac floppy" + depends on PPC_PMAC && !PPC_PMAC64 + help + If you have a SWIM-3 (Super Woz Integrated Machine 3; from Apple) + floppy controller, say Y here. Most commonly found in PowerMacs. + +config BLK_DEV_PS2 + tristate "PS/2 ESDI hard disk support" + depends on MCA && MCA_LEGACY && BROKEN + help + Say Y here if you have a PS/2 machine with a MCA bus and an ESDI + hard disk. + + To compile this driver as a module, choose M here: the + module will be called ps2esdi. + +config AMIGA_Z2RAM + tristate "Amiga Zorro II ramdisk support" + depends on ZORRO + help + This enables support for using Chip RAM and Zorro II RAM as a + ramdisk or as a swap partition. Say Y if you want to include this + driver in the kernel. + + To compile this driver as a module, choose M here: the + module will be called z2ram. + +config ATARI_ACSI + tristate "Atari ACSI support" + depends on ATARI && BROKEN + ---help--- + This enables support for the Atari ACSI interface. The driver + supports hard disks and CD-ROMs, which have 512-byte sectors, or can + be switched to that mode. Due to the ACSI command format, only disks + up to 1 GB are supported. Special support for certain ACSI to SCSI + adapters, which could relax that, isn't included yet. The ACSI + driver is also the basis for certain other drivers for devices + attached to the ACSI bus: Atari SLM laser printer, BioNet-100 + Ethernet, and PAMsNet Ethernet. If you want to use one of these + devices, you need ACSI support, too. + + To compile this driver as a module, choose M here: the + module will be called acsi. + +comment "Some devices (e.g. CD jukebox) support multiple LUNs" + depends on ATARI && ATARI_ACSI + +config ACSI_MULTI_LUN + bool "Probe all LUNs on each ACSI device" + depends on ATARI_ACSI + help + If you have a ACSI device that supports more than one LUN (Logical + Unit Number), e.g. a CD jukebox, you should say Y here so that all + will be found by the ACSI driver. An ACSI device with multiple LUNs + acts logically like multiple ACSI devices. The vast majority of ACSI + devices have only one LUN, and so most people can say N here and + should in fact do so, because it is safer. + +config ATARI_SLM + tristate "Atari SLM laser printer support" + depends on ATARI && ATARI_ACSI!=n + help + If you have an Atari SLM laser printer, say Y to include support for + it in the kernel. Otherwise, say N. This driver is also available as + a module ( = code which can be inserted in and removed from the + running kernel whenever you want). The module will be called + acsi_slm. Be warned: the driver needs much ST-RAM and can cause + problems due to that fact! + +config BLK_DEV_XD + tristate "XT hard disk support" + depends on ISA + help + Very old 8 bit hard disk controllers used in the IBM XT computer + will be supported if you say Y here. + + To compile this driver as a module, choose M here: the + module will be called xd. + + It's pretty unlikely that you have one of these: say N. + +config PARIDE + tristate "Parallel port IDE device support" + depends on PARPORT + ---help--- + There are many external CD-ROM and disk devices that connect through + your computer's parallel port. Most of them are actually IDE devices + using a parallel port IDE adapter. This option enables the PARIDE + subsystem which contains drivers for many of these external drives. + Read <file:Documentation/paride.txt> for more information. + + If you have said Y to the "Parallel-port support" configuration + option, you may share a single port between your printer and other + parallel port devices. Answer Y to build PARIDE support into your + kernel, or M if you would like to build it as a loadable module. If + your parallel port support is in a loadable module, you must build + PARIDE as a module. If you built PARIDE support into your kernel, + you may still build the individual protocol modules and high-level + drivers as loadable modules. If you build this support as a module, + it will be called paride. + + To use the PARIDE support, you must say Y or M here and also to at + least one high-level driver (e.g. "Parallel port IDE disks", + "Parallel port ATAPI CD-ROMs", "Parallel port ATAPI disks" etc.) and + to at least one protocol driver (e.g. "ATEN EH-100 protocol", + "MicroSolutions backpack protocol", "DataStor Commuter protocol" + etc.). + +source "drivers/block/paride/Kconfig" + +config BLK_CPQ_DA + tristate "Compaq SMART2 support" + depends on PCI + help + This is the driver for Compaq Smart Array controllers. Everyone + using these boards should say Y here. See the file + <file:Documentation/cpqarray.txt> for the current list of boards + supported by this driver, and for further information on the use of + this driver. + +config BLK_CPQ_CISS_DA + tristate "Compaq Smart Array 5xxx support" + depends on PCI + help + This is the driver for Compaq Smart Array 5xxx controllers. + Everyone using these boards should say Y here. + See <file:Documentation/cciss.txt> for the current list of + boards supported by this driver, and for further information + on the use of this driver. + +config CISS_SCSI_TAPE + bool "SCSI tape drive support for Smart Array 5xxx" + depends on BLK_CPQ_CISS_DA && SCSI && PROC_FS + help + When enabled (Y), this option allows SCSI tape drives and SCSI medium + changers (tape robots) to be accessed via a Compaq 5xxx array + controller. (See <file:Documentation/cciss.txt> for more details.) + + "SCSI support" and "SCSI tape support" must also be enabled for this + option to work. + + When this option is disabled (N), the SCSI portion of the driver + is not compiled. + +config BLK_DEV_DAC960 + tristate "Mylex DAC960/DAC1100 PCI RAID Controller support" + depends on PCI + help + This driver adds support for the Mylex DAC960, AcceleRAID, and + eXtremeRAID PCI RAID controllers. See the file + <file:Documentation/README.DAC960> for further information about + this driver. + + To compile this driver as a module, choose M here: the + module will be called DAC960. + +config BLK_DEV_UMEM + tristate "Micro Memory MM5415 Battery Backed RAM support (EXPERIMENTAL)" + depends on PCI && EXPERIMENTAL + ---help--- + Saying Y here will include support for the MM5415 family of + battery backed (Non-volatile) RAM cards. + <http://www.umem.com/> + + The cards appear as block devices that can be partitioned into + as many as 15 partitions. + + To compile this driver as a module, choose M here: the + module will be called umem. + + The umem driver has not yet been allocated a MAJOR number, so + one is chosen dynamically. Use "devfs" or look in /proc/devices + for the device number + +config BLK_DEV_UBD + bool "Virtual block device" + depends on UML + ---help--- + The User-Mode Linux port includes a driver called UBD which will let + you access arbitrary files on the host computer as block devices. + Unless you know that you do not need such virtual block devices say + Y here. + +config BLK_DEV_UBD_SYNC + bool "Always do synchronous disk IO for UBD" + depends on BLK_DEV_UBD + ---help--- + Writes to the virtual block device are not immediately written to the + host's disk; this may cause problems if, for example, the User-Mode + Linux 'Virtual Machine' uses a journalling filesystem and the host + computer crashes. + + Synchronous operation (i.e. always writing data to the host's disk + immediately) is configurable on a per-UBD basis by using a special + kernel command line option. Alternatively, you can say Y here to + turn on synchronous operation by default for all block devices. + + If you're running a journalling file system (like reiserfs, for + example) in your virtual machine, you will want to say Y here. If + you care for the safety of the data in your virtual machine, Y is a + wise choice too. In all other cases (for example, if you're just + playing around with User-Mode Linux) you can choose N. + +config BLK_DEV_COW_COMMON + bool + default BLK_DEV_UBD + +config MMAPPER + tristate "Example IO memory driver (BROKEN)" + depends on UML && BROKEN + ---help--- + The User-Mode Linux port can provide support for IO Memory + emulation with this option. This allows a host file to be + specified as an I/O region on the kernel command line. That file + will be mapped into UML's kernel address space where a driver can + locate it and do whatever it wants with the memory, including + providing an interface to it for UML processes to use. + + For more information, see + <http://user-mode-linux.sourceforge.net/iomem.html>. + + If you'd like to be able to provide a simulated IO port space for + User-Mode Linux processes, say Y. If unsure, say N. + +config BLK_DEV_LOOP + tristate "Loopback device support" + ---help--- + Saying Y here will allow you to use a regular file as a block + device; you can then create a file system on that block device and + mount it just as you would mount other block devices such as hard + drive partitions, CD-ROM drives or floppy drives. The loop devices + are block special device files with major number 7 and typically + called /dev/loop0, /dev/loop1 etc. + + This is useful if you want to check an ISO 9660 file system before + burning the CD, or if you want to use floppy images without first + writing them to floppy. Furthermore, some Linux distributions avoid + the need for a dedicated Linux partition by keeping their complete + root file system inside a DOS FAT file using this loop device + driver. + + To use the loop device, you need the losetup utility, found in the + util-linux package, see + <ftp://ftp.kernel.org/pub/linux/utils/util-linux/>. + + The loop device driver can also be used to "hide" a file system in + a disk partition, floppy, or regular file, either using encryption + (scrambling the data) or steganography (hiding the data in the low + bits of, say, a sound file). This is also safe if the file resides + on a remote file server. + + There are several ways of encrypting disks. Some of these require + kernel patches. The vanilla kernel offers the cryptoloop option + and a Device Mapper target (which is superior, as it supports all + file systems). If you want to use the cryptoloop, say Y to both + LOOP and CRYPTOLOOP, and make sure you have a recent (version 2.12 + or later) version of util-linux. Additionally, be aware that + the cryptoloop is not safe for storing journaled filesystems. + + Note that this loop device has nothing to do with the loopback + device used for network connections from the machine to itself. + + To compile this driver as a module, choose M here: the + module will be called loop. + + Most users will answer N here. + +config BLK_DEV_CRYPTOLOOP + tristate "Cryptoloop Support" + select CRYPTO + depends on BLK_DEV_LOOP + ---help--- + Say Y here if you want to be able to use the ciphers that are + provided by the CryptoAPI as loop transformation. This might be + used as hard disk encryption. + + WARNING: This device is not safe for journaled file systems like + ext3 or Reiserfs. Please use the Device Mapper crypto module + instead, which can be configured to be on-disk compatible with the + cryptoloop device. + +config BLK_DEV_NBD + tristate "Network block device support" + depends on NET + ---help--- + Saying Y here will allow your computer to be a client for network + block devices, i.e. it will be able to use block devices exported by + servers (mount file systems on them etc.). Communication between + client and server works over TCP/IP networking, but to the client + program this is hidden: it looks like a regular local file access to + a block device special file such as /dev/nd0. + + Network block devices also allows you to run a block-device in + userland (making server and client physically the same computer, + communicating using the loopback network device). + + Read <file:Documentation/nbd.txt> for more information, especially + about where to find the server code, which runs in user space and + does not need special kernel support. + + Note that this has nothing to do with the network file systems NFS + or Coda; you can say N here even if you intend to use NFS or Coda. + + To compile this driver as a module, choose M here: the + module will be called nbd. + + If unsure, say N. + +config BLK_DEV_SX8 + tristate "Promise SATA SX8 support" + depends on PCI + ---help--- + Saying Y or M here will enable support for the + Promise SATA SX8 controllers. + + Use devices /dev/sx8/$N and /dev/sx8/$Np$M. + +config BLK_DEV_UB + tristate "Low Performance USB Block driver" + depends on USB + help + This driver supports certain USB attached storage devices + such as flash keys. + + Warning: Enabling this cripples the usb-storage driver. + + If unsure, say N. + +config BLK_DEV_RAM + tristate "RAM disk support" + ---help--- + Saying Y here will allow you to use a portion of your RAM memory as + a block device, so that you can make file systems on it, read and + write to it and do all the other things that you can do with normal + block devices (such as hard drives). It is usually used to load and + store a copy of a minimal root file system off of a floppy into RAM + during the initial install of Linux. + + Note that the kernel command line option "ramdisk=XX" is now + obsolete. For details, read <file:Documentation/ramdisk.txt>. + + To compile this driver as a module, choose M here: the + module will be called rd. + + Most normal users won't need the RAM disk functionality, and can + thus say N here. + +config BLK_DEV_RAM_COUNT + int "Default number of RAM disks" if BLK_DEV_RAM + default "16" + help + The default value is 16 RAM disks. Change this if you know what + are doing. If you boot from a filesystem that needs to be extracted + in memory, you will need at least one RAM disk (e.g. root on cramfs). + +config BLK_DEV_RAM_SIZE + int "Default RAM disk size (kbytes)" + depends on BLK_DEV_RAM + default "4096" + help + The default value is 4096 kilobytes. Only change this if you know + what are you doing. If you are using IBM S/390, then set this to + 8192. + +config BLK_DEV_INITRD + bool "Initial RAM disk (initrd) support" + depends on BLK_DEV_RAM=y + help + The initial RAM disk is a RAM disk that is loaded by the boot loader + (loadlin or lilo) and that is mounted as root before the normal boot + procedure. It is typically used to load modules needed to mount the + "real" root file system, etc. See <file:Documentation/initrd.txt> + for details. + +config INITRAMFS_SOURCE + string "Initramfs source file(s)" + default "" + help + This can be either a single cpio archive with a .cpio suffix or a + space-separated list of directories and files for building the + initramfs image. A cpio archive should contain a filesystem archive + to be used as an initramfs image. Directories should contain a + filesystem layout to be included in the initramfs image. Files + should contain entries according to the format described by the + "usr/gen_init_cpio" program in the kernel tree. + + When multiple directories and files are specified then the + initramfs image will be the aggregate of all of them. + + See <file:Documentation/early-userspace/README for more details. + + If you are not sure, leave it blank. + +config INITRAMFS_ROOT_UID + int "User ID to map to 0 (user root)" + depends on INITRAMFS_SOURCE!="" + default "0" + help + This setting is only meaningful if the INITRAMFS_SOURCE is + contains a directory. Setting this user ID (UID) to something + other than "0" will cause all files owned by that UID to be + owned by user root in the initial ramdisk image. + + If you are not sure, leave it set to "0". + +config INITRAMFS_ROOT_GID + int "Group ID to map to 0 (group root)" + depends on INITRAMFS_SOURCE!="" + default "0" + help + This setting is only meaningful if the INITRAMFS_SOURCE is + contains a directory. Setting this group ID (GID) to something + other than "0" will cause all files owned by that GID to be + owned by group root in the initial ramdisk image. + + If you are not sure, leave it set to "0". + +#XXX - it makes sense to enable this only for 32-bit subarch's, not for x86_64 +#for instance. +config LBD + bool "Support for Large Block Devices" + depends on X86 || MIPS32 || PPC32 || ARCH_S390_31 || SUPERH || UML + help + Say Y here if you want to attach large (bigger than 2TB) discs to + your machine, or if you want to have a raid or loopback device + bigger than 2TB. Otherwise say N. + +config CDROM_PKTCDVD + tristate "Packet writing on CD/DVD media" + depends on !UML + help + If you have a CDROM drive that supports packet writing, say Y to + include preliminary support. It should work with any MMC/Mt Fuji + compliant ATAPI or SCSI drive, which is just about any newer CD + writer. + + Currently only writing to CD-RW, DVD-RW and DVD+RW discs is possible. + DVD-RW disks must be in restricted overwrite mode. + + To compile this driver as a module, choose M here: the + module will be called pktcdvd. + +config CDROM_PKTCDVD_BUFFERS + int "Free buffers for data gathering" + depends on CDROM_PKTCDVD + default "8" + help + This controls the maximum number of active concurrent packets. More + concurrent packets can increase write performance, but also require + more memory. Each concurrent packet will require approximately 64Kb + of non-swappable kernel memory, memory which will be allocated at + pktsetup time. + +config CDROM_PKTCDVD_WCACHE + bool "Enable write caching" + depends on CDROM_PKTCDVD + help + If enabled, write caching will be set for the CD-R/W device. For now + this option is dangerous unless the CD-RW media is known good, as we + don't do deferred write error handling yet. + +source "drivers/s390/block/Kconfig" + +source "drivers/block/Kconfig.iosched" + +config ATA_OVER_ETH + tristate "ATA over Ethernet support" + depends on NET + help + This driver provides Support for ATA over Ethernet block + devices like the Coraid EtherDrive (R) Storage Blade. + +endmenu diff --git a/drivers/block/Kconfig.iosched b/drivers/block/Kconfig.iosched new file mode 100644 index 00000000000..6070a480600 --- /dev/null +++ b/drivers/block/Kconfig.iosched @@ -0,0 +1,41 @@ + +menu "IO Schedulers" + +config IOSCHED_NOOP + bool + default y + ---help--- + The no-op I/O scheduler is a minimal scheduler that does basic merging + and sorting. Its main uses include non-disk based block devices like + memory devices, and specialised software or hardware environments + that do their own scheduling and require only minimal assistance from + the kernel. + +config IOSCHED_AS + tristate "Anticipatory I/O scheduler" + default y + ---help--- + The anticipatory I/O scheduler is the default disk scheduler. It is + generally a good choice for most environments, but is quite large and + complex when compared to the deadline I/O scheduler, it can also be + slower in some cases especially some database loads. + +config IOSCHED_DEADLINE + tristate "Deadline I/O scheduler" + default y + ---help--- + The deadline I/O scheduler is simple and compact, and is often as + good as the anticipatory I/O scheduler, and in some database + workloads, better. In the case of a single process performing I/O to + a disk at any one time, its behaviour is almost identical to the + anticipatory I/O scheduler and so is a good choice. + +config IOSCHED_CFQ + tristate "CFQ I/O scheduler" + default y + ---help--- + The CFQ I/O scheduler tries to distribute bandwidth equally + among all processes in the system. It should provide a fair + working environment, suitable for desktop systems. + +endmenu diff --git a/drivers/block/Makefile b/drivers/block/Makefile new file mode 100644 index 00000000000..1cf09a1c065 --- /dev/null +++ b/drivers/block/Makefile @@ -0,0 +1,47 @@ +# +# Makefile for the kernel block device drivers. +# +# 12 June 2000, Christoph Hellwig <hch@infradead.org> +# Rewritten to use lists instead of if-statements. +# +# Note : at this point, these files are compiled on all systems. +# In the future, some of these should be built conditionally. +# + +# +# NOTE that ll_rw_blk.c must come early in linkage order - it starts the +# kblockd threads +# + +obj-y := elevator.o ll_rw_blk.o ioctl.o genhd.o scsi_ioctl.o + +obj-$(CONFIG_IOSCHED_NOOP) += noop-iosched.o +obj-$(CONFIG_IOSCHED_AS) += as-iosched.o +obj-$(CONFIG_IOSCHED_DEADLINE) += deadline-iosched.o +obj-$(CONFIG_IOSCHED_CFQ) += cfq-iosched.o +obj-$(CONFIG_MAC_FLOPPY) += swim3.o +obj-$(CONFIG_BLK_DEV_FD) += floppy.o +obj-$(CONFIG_BLK_DEV_FD98) += floppy98.o +obj-$(CONFIG_AMIGA_FLOPPY) += amiflop.o +obj-$(CONFIG_ATARI_FLOPPY) += ataflop.o +obj-$(CONFIG_BLK_DEV_SWIM_IOP) += swim_iop.o +obj-$(CONFIG_ATARI_ACSI) += acsi.o +obj-$(CONFIG_ATARI_SLM) += acsi_slm.o +obj-$(CONFIG_AMIGA_Z2RAM) += z2ram.o +obj-$(CONFIG_BLK_DEV_RAM) += rd.o +obj-$(CONFIG_BLK_DEV_LOOP) += loop.o +obj-$(CONFIG_BLK_DEV_PS2) += ps2esdi.o +obj-$(CONFIG_BLK_DEV_XD) += xd.o +obj-$(CONFIG_BLK_CPQ_DA) += cpqarray.o +obj-$(CONFIG_BLK_CPQ_CISS_DA) += cciss.o +obj-$(CONFIG_BLK_DEV_DAC960) += DAC960.o +obj-$(CONFIG_CDROM_PKTCDVD) += pktcdvd.o + +obj-$(CONFIG_BLK_DEV_UMEM) += umem.o +obj-$(CONFIG_BLK_DEV_NBD) += nbd.o +obj-$(CONFIG_BLK_DEV_CRYPTOLOOP) += cryptoloop.o + +obj-$(CONFIG_VIODASD) += viodasd.o +obj-$(CONFIG_BLK_DEV_SX8) += sx8.o +obj-$(CONFIG_BLK_DEV_UB) += ub.o + diff --git a/drivers/block/acsi.c b/drivers/block/acsi.c new file mode 100644 index 00000000000..ce933de4808 --- /dev/null +++ b/drivers/block/acsi.c @@ -0,0 +1,1829 @@ +/* + * acsi.c -- Device driver for Atari ACSI hard disks + * + * Copyright 1994 Roman Hodek <Roman.Hodek@informatik.uni-erlangen.de> + * + * Some parts are based on hd.c by Linus Torvalds + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file COPYING in the main directory of this archive for + * more details. + * + */ + +/* + * Still to in this file: + * - If a command ends with an error status (!= 0), the following + * REQUEST SENSE commands (4 to fill the ST-DMA FIFO) are done by + * polling the _IRQ signal (not interrupt-driven). This should be + * avoided in future because it takes up a non-neglectible time in + * the interrupt service routine while interrupts are disabled. + * Maybe a timer interrupt will get lost :-( + */ + +/* + * General notes: + * + * - All ACSI devices (disks, CD-ROMs, ...) use major number 28. + * Minors are organized like it is with SCSI: The upper 4 bits + * identify the device, the lower 4 bits the partition. + * The device numbers (the upper 4 bits) are given in the same + * order as the devices are found on the bus. + * - Up to 8 LUNs are supported for each target (if CONFIG_ACSI_MULTI_LUN + * is defined), but only a total of 16 devices (due to minor + * numbers...). Note that Atari allows only a maximum of 4 targets + * (i.e. controllers, not devices) on the ACSI bus! + * - A optimizing scheme similar to SCSI scatter-gather is implemented. + * - Removable media are supported. After a medium change to device + * is reinitialized (partition check etc.). Also, if the device + * knows the PREVENT/ALLOW MEDIUM REMOVAL command, the door should + * be locked and unlocked when mounting the first or unmounting the + * last filesystem on the device. The code is untested, because I + * don't have a removable hard disk. + * + */ + +#include <linux/config.h> +#include <linux/module.h> +#include <linux/errno.h> +#include <linux/signal.h> +#include <linux/sched.h> +#include <linux/timer.h> +#include <linux/fs.h> +#include <linux/kernel.h> +#include <linux/genhd.h> +#include <linux/delay.h> +#include <linux/mm.h> +#include <linux/major.h> +#include <linux/slab.h> +#include <linux/interrupt.h> +#include <scsi/scsi.h> /* for SCSI_IOCTL_GET_IDLUN */ +typedef void Scsi_Device; /* hack to avoid including scsi.h */ +#include <scsi/scsi_ioctl.h> +#include <linux/hdreg.h> /* for HDIO_GETGEO */ +#include <linux/blkpg.h> +#include <linux/buffer_head.h> +#include <linux/blkdev.h> + +#include <asm/setup.h> +#include <asm/pgtable.h> +#include <asm/system.h> +#include <asm/uaccess.h> +#include <asm/atarihw.h> +#include <asm/atariints.h> +#include <asm/atari_acsi.h> +#include <asm/atari_stdma.h> +#include <asm/atari_stram.h> + +static void (*do_acsi)(void) = NULL; +static struct request_queue *acsi_queue; +#define QUEUE (acsi_queue) +#define CURRENT elv_next_request(acsi_queue) + +#define DEBUG +#undef DEBUG_DETECT +#undef NO_WRITE + +#define MAX_ERRORS 8 /* Max read/write errors/sector */ +#define MAX_LUN 8 /* Max LUNs per target */ +#define MAX_DEV 16 + +#define ACSI_BUFFER_SIZE (16*1024) /* "normal" ACSI buffer size */ +#define ACSI_BUFFER_MINSIZE (2048) /* min. buf size if ext. DMA */ +#define ACSI_BUFFER_SIZE_ORDER 2 /* order size for above */ +#define ACSI_BUFFER_MINSIZE_ORDER 0 /* order size for above */ +#define ACSI_BUFFER_SECTORS (ACSI_BUFFER_SIZE/512) + +#define ACSI_BUFFER_ORDER \ + (ATARIHW_PRESENT(EXTD_DMA) ? \ + ACSI_BUFFER_MINSIZE_ORDER : \ + ACSI_BUFFER_SIZE_ORDER) + +#define ACSI_TIMEOUT (4*HZ) + +/* minimum delay between two commands */ + +#define COMMAND_DELAY 500 + +typedef enum { + NONE, HARDDISK, CDROM +} ACSI_TYPE; + +struct acsi_info_struct { + ACSI_TYPE type; /* type of device */ + unsigned target; /* target number */ + unsigned lun; /* LUN in target controller */ + unsigned removable : 1; /* Flag for removable media */ + unsigned read_only : 1; /* Flag for read only devices */ + unsigned old_atari_disk : 1; /* Is an old Atari disk */ + unsigned changed : 1; /* Medium has been changed */ + unsigned long size; /* #blocks */ + int access_count; +} acsi_info[MAX_DEV]; + +/* + * SENSE KEYS + */ + +#define NO_SENSE 0x00 +#define RECOVERED_ERROR 0x01 +#define NOT_READY 0x02 +#define MEDIUM_ERROR 0x03 +#define HARDWARE_ERROR 0x04 +#define ILLEGAL_REQUEST 0x05 +#define UNIT_ATTENTION 0x06 +#define DATA_PROTECT 0x07 +#define BLANK_CHECK 0x08 +#define COPY_ABORTED 0x0a +#define ABORTED_COMMAND 0x0b +#define VOLUME_OVERFLOW 0x0d +#define MISCOMPARE 0x0e + + +/* + * DEVICE TYPES + */ + +#define TYPE_DISK 0x00 +#define TYPE_TAPE 0x01 +#define TYPE_WORM 0x04 +#define TYPE_ROM 0x05 +#define TYPE_MOD 0x07 +#define TYPE_NO_LUN 0x7f + +/* The data returned by MODE SENSE differ between the old Atari + * hard disks and SCSI disks connected to ACSI. In the following, both + * formats are defined and some macros to operate on them potably. + */ + +typedef struct { + unsigned long dummy[2]; + unsigned long sector_size; + unsigned char format_code; +#define ATARI_SENSE_FORMAT_FIX 1 +#define ATARI_SENSE_FORMAT_CHNG 2 + unsigned char cylinders_h; + unsigned char cylinders_l; + unsigned char heads; + unsigned char reduced_h; + unsigned char reduced_l; + unsigned char precomp_h; + unsigned char precomp_l; + unsigned char landing_zone; + unsigned char steprate; + unsigned char type; +#define ATARI_SENSE_TYPE_FIXCHNG_MASK 4 +#define ATARI_SENSE_TYPE_SOFTHARD_MASK 8 +#define ATARI_SENSE_TYPE_FIX 4 +#define ATARI_SENSE_TYPE_CHNG 0 +#define ATARI_SENSE_TYPE_SOFT 0 +#define ATARI_SENSE_TYPE_HARD 8 + unsigned char sectors; +} ATARI_SENSE_DATA; + +#define ATARI_CAPACITY(sd) \ + (((int)((sd).cylinders_h<<8)|(sd).cylinders_l) * \ + (sd).heads * (sd).sectors) + + +typedef struct { + unsigned char dummy1; + unsigned char medium_type; + unsigned char dummy2; + unsigned char descriptor_size; + unsigned long block_count; + unsigned long sector_size; + /* Page 0 data */ + unsigned char page_code; + unsigned char page_size; + unsigned char page_flags; + unsigned char qualifier; +} SCSI_SENSE_DATA; + +#define SCSI_CAPACITY(sd) ((sd).block_count & 0xffffff) + + +typedef union { + ATARI_SENSE_DATA atari; + SCSI_SENSE_DATA scsi; +} SENSE_DATA; + +#define SENSE_TYPE_UNKNOWN 0 +#define SENSE_TYPE_ATARI 1 +#define SENSE_TYPE_SCSI 2 + +#define SENSE_TYPE(sd) \ + (((sd).atari.dummy[0] == 8 && \ + ((sd).atari.format_code == 1 || \ + (sd).atari.format_code == 2)) ? SENSE_TYPE_ATARI : \ + ((sd).scsi.dummy1 >= 11) ? SENSE_TYPE_SCSI : \ + SENSE_TYPE_UNKNOWN) + +#define CAPACITY(sd) \ + (SENSE_TYPE(sd) == SENSE_TYPE_ATARI ? \ + ATARI_CAPACITY((sd).atari) : \ + SCSI_CAPACITY((sd).scsi)) + +#define SECTOR_SIZE(sd) \ + (SENSE_TYPE(sd) == SENSE_TYPE_ATARI ? \ + (sd).atari.sector_size : \ + (sd).scsi.sector_size & 0xffffff) + +/* Default size if capacity cannot be determined (1 GByte) */ +#define DEFAULT_SIZE 0x1fffff + +#define CARTRCH_STAT(aip,buf) \ + (aip->old_atari_disk ? \ + (((buf)[0] & 0x7f) == 0x28) : \ + ((((buf)[0] & 0x70) == 0x70) ? \ + (((buf)[2] & 0x0f) == 0x06) : \ + (((buf)[0] & 0x0f) == 0x06))) \ + +/* These two are also exported to other drivers that work on the ACSI bus and + * need an ST-RAM buffer. */ +char *acsi_buffer; +unsigned long phys_acsi_buffer; + +static int NDevices; + +static int CurrentNReq; +static int CurrentNSect; +static char *CurrentBuffer; + +static DEFINE_SPINLOCK(acsi_lock); + + +#define SET_TIMER() mod_timer(&acsi_timer, jiffies + ACSI_TIMEOUT) +#define CLEAR_TIMER() del_timer(&acsi_timer) + +static unsigned long STramMask; +#define STRAM_ADDR(a) (((a) & STramMask) == 0) + + + +/* ACSI commands */ + +static char tur_cmd[6] = { 0x00, 0, 0, 0, 0, 0 }; +static char modesense_cmd[6] = { 0x1a, 0, 0, 0, 24, 0 }; +static char modeselect_cmd[6] = { 0x15, 0, 0, 0, 12, 0 }; +static char inquiry_cmd[6] = { 0x12, 0, 0, 0,255, 0 }; +static char reqsense_cmd[6] = { 0x03, 0, 0, 0, 4, 0 }; +static char read_cmd[6] = { 0x08, 0, 0, 0, 0, 0 }; +static char write_cmd[6] = { 0x0a, 0, 0, 0, 0, 0 }; +static char pa_med_rem_cmd[6] = { 0x1e, 0, 0, 0, 0, 0 }; + +#define CMDSET_TARG_LUN(cmd,targ,lun) \ + do { \ + cmd[0] = (cmd[0] & ~0xe0) | (targ)<<5; \ + cmd[1] = (cmd[1] & ~0xe0) | (lun)<<5; \ + } while(0) + +#define CMDSET_BLOCK(cmd,blk) \ + do { \ + unsigned long __blk = (blk); \ + cmd[3] = __blk; __blk >>= 8; \ + cmd[2] = __blk; __blk >>= 8; \ + cmd[1] = (cmd[1] & 0xe0) | (__blk & 0x1f); \ + } while(0) + +#define CMDSET_LEN(cmd,len) \ + do { \ + cmd[4] = (len); \ + } while(0) + +/* ACSI errors (from REQUEST SENSE); There are two tables, one for the + * old Atari disks and one for SCSI on ACSI disks. + */ + +struct acsi_error { + unsigned char code; + const char *text; +} atari_acsi_errors[] = { + { 0x00, "No error (??)" }, + { 0x01, "No index pulses" }, + { 0x02, "Seek not complete" }, + { 0x03, "Write fault" }, + { 0x04, "Drive not ready" }, + { 0x06, "No Track 00 signal" }, + { 0x10, "ECC error in ID field" }, + { 0x11, "Uncorrectable data error" }, + { 0x12, "ID field address mark not found" }, + { 0x13, "Data field address mark not found" }, + { 0x14, "Record not found" }, + { 0x15, "Seek error" }, + { 0x18, "Data check in no retry mode" }, + { 0x19, "ECC error during verify" }, + { 0x1a, "Access to bad block" }, + { 0x1c, "Unformatted or bad format" }, + { 0x20, "Invalid command" }, + { 0x21, "Invalid block address" }, + { 0x23, "Volume overflow" }, + { 0x24, "Invalid argument" }, + { 0x25, "Invalid drive number" }, + { 0x26, "Byte zero parity check" }, + { 0x28, "Cartride changed" }, + { 0x2c, "Error count overflow" }, + { 0x30, "Controller selftest failed" } +}, + + scsi_acsi_errors[] = { + { 0x00, "No error (??)" }, + { 0x01, "Recovered error" }, + { 0x02, "Drive not ready" }, + { 0x03, "Uncorrectable medium error" }, + { 0x04, "Hardware error" }, + { 0x05, "Illegal request" }, + { 0x06, "Unit attention (Reset or cartridge changed)" }, + { 0x07, "Data protection" }, + { 0x08, "Blank check" }, + { 0x0b, "Aborted Command" }, + { 0x0d, "Volume overflow" } +}; + + + +/***************************** Prototypes *****************************/ + +static int acsicmd_dma( const char *cmd, char *buffer, int blocks, int + rwflag, int enable); +static int acsi_reqsense( char *buffer, int targ, int lun); +static void acsi_print_error(const unsigned char *errblk, struct acsi_info_struct *aip); +static irqreturn_t acsi_interrupt (int irq, void *data, struct pt_regs *fp); +static void unexpected_acsi_interrupt( void ); +static void bad_rw_intr( void ); +static void read_intr( void ); +static void write_intr( void); +static void acsi_times_out( unsigned long dummy ); +static void copy_to_acsibuffer( void ); +static void copy_from_acsibuffer( void ); +static void do_end_requests( void ); +static void do_acsi_request( request_queue_t * ); +static void redo_acsi_request( void ); +static int acsi_ioctl( struct inode *inode, struct file *file, unsigned int + cmd, unsigned long arg ); +static int acsi_open( struct inode * inode, struct file * filp ); +static int acsi_release( struct inode * inode, struct file * file ); +static void acsi_prevent_removal(struct acsi_info_struct *aip, int flag ); +static int acsi_change_blk_size( int target, int lun); +static int acsi_mode_sense( int target, int lun, SENSE_DATA *sd ); +static int acsi_revalidate (struct gendisk *disk); + +/************************* End of Prototypes **************************/ + + +struct timer_list acsi_timer = TIMER_INITIALIZER(acsi_times_out, 0, 0); + + +#ifdef CONFIG_ATARI_SLM + +extern int attach_slm( int target, int lun ); +extern int slm_init( void ); + +#endif + + + +/*********************************************************************** + * + * ACSI primitives + * + **********************************************************************/ + + +/* + * The following two functions wait for _IRQ to become Low or High, + * resp., with a timeout. The 'timeout' parameter is in jiffies + * (10ms). + * If the functions are called with timer interrupts on (int level < + * 6), the timeout is based on the 'jiffies' variable to provide exact + * timeouts for device probing etc. + * If interrupts are disabled, the number of tries is based on the + * 'loops_per_jiffy' variable. A rough estimation is sufficient here... + */ + +#define INT_LEVEL \ + ({ unsigned __sr; \ + __asm__ __volatile__ ( "movew %/sr,%0" : "=dm" (__sr) ); \ + (__sr >> 8) & 7; \ + }) + +int acsi_wait_for_IRQ( unsigned timeout ) + +{ + if (INT_LEVEL < 6) { + unsigned long maxjif = jiffies + timeout; + while (time_before(jiffies, maxjif)) + if (!(mfp.par_dt_reg & 0x20)) return( 1 ); + } + else { + long tries = loops_per_jiffy / 8 * timeout; + while( --tries >= 0 ) + if (!(mfp.par_dt_reg & 0x20)) return( 1 ); + } + return( 0 ); /* timeout! */ +} + + +int acsi_wait_for_noIRQ( unsigned timeout ) + +{ + if (INT_LEVEL < 6) { + unsigned long maxjif = jiffies + timeout; + while (time_before(jiffies, maxjif)) + if (mfp.par_dt_reg & 0x20) return( 1 ); + } + else { + long tries = loops_per_jiffy * timeout / 8; + while( tries-- >= 0 ) + if (mfp.par_dt_reg & 0x20) return( 1 ); + } + return( 0 ); /* timeout! */ +} + +static struct timeval start_time; + +void +acsi_delay_start(void) +{ + do_gettimeofday(&start_time); +} + +/* wait from acsi_delay_start to now usec (<1E6) usec */ + +void +acsi_delay_end(long usec) +{ + struct timeval end_time; + long deltau,deltas; + do_gettimeofday(&end_time); + deltau=end_time.tv_usec - start_time.tv_usec; + deltas=end_time.tv_sec - start_time.tv_sec; + if (deltas > 1 || deltas < 0) + return; + if (deltas > 0) + deltau += 1000*1000; + if (deltau >= usec) + return; + udelay(usec-deltau); +} + +/* acsicmd_dma() sends an ACSI command and sets up the DMA to transfer + * 'blocks' blocks of 512 bytes from/to 'buffer'. + * Because the _IRQ signal is used for handshaking the command bytes, + * the ACSI interrupt has to be disabled in this function. If the end + * of the operation should be signalled by a real interrupt, it has to be + * reenabled afterwards. + */ + +static int acsicmd_dma( const char *cmd, char *buffer, int blocks, int rwflag, int enable) + +{ unsigned long flags, paddr; + int i; + +#ifdef NO_WRITE + if (rwflag || *cmd == 0x0a) { + printk( "ACSI: Write commands disabled!\n" ); + return( 0 ); + } +#endif + + rwflag = rwflag ? 0x100 : 0; + paddr = virt_to_phys( buffer ); + + acsi_delay_end(COMMAND_DELAY); + DISABLE_IRQ(); + + local_irq_save(flags); + /* Low on A1 */ + dma_wd.dma_mode_status = 0x88 | rwflag; + MFPDELAY(); + + /* set DMA address */ + dma_wd.dma_lo = (unsigned char)paddr; + paddr >>= 8; + MFPDELAY(); + dma_wd.dma_md = (unsigned char)paddr; + paddr >>= 8; + MFPDELAY(); + if (ATARIHW_PRESENT(EXTD_DMA)) + st_dma_ext_dmahi = (unsigned short)paddr; + else + dma_wd.dma_hi = (unsigned char)paddr; + MFPDELAY(); + local_irq_restore(flags); + + /* send the command bytes except the last */ + for( i = 0; i < 5; ++i ) { + DMA_LONG_WRITE( *cmd++, 0x8a | rwflag ); + udelay(20); + if (!acsi_wait_for_IRQ( HZ/2 )) return( 0 ); /* timeout */ + } + + /* Clear FIFO and switch DMA to correct direction */ + dma_wd.dma_mode_status = 0x92 | (rwflag ^ 0x100); + MFPDELAY(); + dma_wd.dma_mode_status = 0x92 | rwflag; + MFPDELAY(); + + /* How many sectors for DMA */ + dma_wd.fdc_acces_seccount = blocks; + MFPDELAY(); + + /* send last command byte */ + dma_wd.dma_mode_status = 0x8a | rwflag; + MFPDELAY(); + DMA_LONG_WRITE( *cmd++, 0x0a | rwflag ); + if (enable) + ENABLE_IRQ(); + udelay(80); + + return( 1 ); +} + + +/* + * acsicmd_nodma() sends an ACSI command that requires no DMA. + */ + +int acsicmd_nodma( const char *cmd, int enable) + +{ int i; + + acsi_delay_end(COMMAND_DELAY); + DISABLE_IRQ(); + + /* send first command byte */ + dma_wd.dma_mode_status = 0x88; + MFPDELAY(); + DMA_LONG_WRITE( *cmd++, 0x8a ); + udelay(20); + if (!acsi_wait_for_IRQ( HZ/2 )) return( 0 ); /* timeout */ + + /* send the intermediate command bytes */ + for( i = 0; i < 4; ++i ) { + DMA_LONG_WRITE( *cmd++, 0x8a ); + udelay(20); + if (!acsi_wait_for_IRQ( HZ/2 )) return( 0 ); /* timeout */ + } + + /* send last command byte */ + DMA_LONG_WRITE( *cmd++, 0x0a ); + if (enable) + ENABLE_IRQ(); + udelay(80); + + return( 1 ); + /* Note that the ACSI interrupt is still disabled after this + * function. If you want to get the IRQ delivered, enable it manually! + */ +} + + +static int acsi_reqsense( char *buffer, int targ, int lun) + +{ + CMDSET_TARG_LUN( reqsense_cmd, targ, lun); + if (!acsicmd_dma( reqsense_cmd, buffer, 1, 0, 0 )) return( 0 ); + if (!acsi_wait_for_IRQ( 10 )) return( 0 ); + acsi_getstatus(); + if (!acsicmd_nodma( reqsense_cmd, 0 )) return( 0 ); + if (!acsi_wait_for_IRQ( 10 )) return( 0 ); + acsi_getstatus(); + if (!acsicmd_nodma( reqsense_cmd, 0 )) return( 0 ); + if (!acsi_wait_for_IRQ( 10 )) return( 0 ); + acsi_getstatus(); + if (!acsicmd_nodma( reqsense_cmd, 0 )) return( 0 ); + if (!acsi_wait_for_IRQ( 10 )) return( 0 ); + acsi_getstatus(); + dma_cache_maintenance( virt_to_phys(buffer), 16, 0 ); + + return( 1 ); +} + + +/* + * ACSI status phase: get the status byte from the bus + * + * I've seen several times that a 0xff status is read, propably due to + * a timing error. In this case, the procedure is repeated after the + * next _IRQ edge. + */ + +int acsi_getstatus( void ) + +{ int status; + + DISABLE_IRQ(); + for(;;) { + if (!acsi_wait_for_IRQ( 100 )) { + acsi_delay_start(); + return( -1 ); + } + dma_wd.dma_mode_status = 0x8a; + MFPDELAY(); + status = dma_wd.fdc_acces_seccount; + if (status != 0xff) break; +#ifdef DEBUG + printk("ACSI: skipping 0xff status byte\n" ); +#endif + udelay(40); + acsi_wait_for_noIRQ( 20 ); + } + dma_wd.dma_mode_status = 0x80; + udelay(40); + acsi_wait_for_noIRQ( 20 ); + + acsi_delay_start(); + return( status & 0x1f ); /* mask of the device# */ +} + + +#if (defined(CONFIG_ATARI_SLM) || defined(CONFIG_ATARI_SLM_MODULE)) + +/* Receive data in an extended status phase. Needed by SLM printer. */ + +int acsi_extstatus( char *buffer, int cnt ) + +{ int status; + + DISABLE_IRQ(); + udelay(80); + while( cnt-- > 0 ) { + if (!acsi_wait_for_IRQ( 40 )) return( 0 ); + dma_wd.dma_mode_status = 0x8a; + MFPDELAY(); + status = dma_wd.fdc_acces_seccount; + MFPDELAY(); + *buffer++ = status & 0xff; + udelay(40); + } + return( 1 ); +} + + +/* Finish an extended status phase */ + +void acsi_end_extstatus( void ) + +{ + dma_wd.dma_mode_status = 0x80; + udelay(40); + acsi_wait_for_noIRQ( 20 ); + acsi_delay_start(); +} + + +/* Send data in an extended command phase */ + +int acsi_extcmd( unsigned char *buffer, int cnt ) + +{ + while( cnt-- > 0 ) { + DMA_LONG_WRITE( *buffer++, 0x8a ); + udelay(20); + if (!acsi_wait_for_IRQ( HZ/2 )) return( 0 ); /* timeout */ + } + return( 1 ); +} + +#endif + + +static void acsi_print_error(const unsigned char *errblk, struct acsi_info_struct *aip) + +{ int atari_err, i, errcode; + struct acsi_error *arr; + + atari_err = aip->old_atari_disk; + if (atari_err) + errcode = errblk[0] & 0x7f; + else + if ((errblk[0] & 0x70) == 0x70) + errcode = errblk[2] & 0x0f; + else + errcode = errblk[0] & 0x0f; + + printk( KERN_ERR "ACSI error 0x%02x", errcode ); + + if (errblk[0] & 0x80) + printk( " for sector %d", + ((errblk[1] & 0x1f) << 16) | + (errblk[2] << 8) | errblk[0] ); + + arr = atari_err ? atari_acsi_errors : scsi_acsi_errors; + i = atari_err ? sizeof(atari_acsi_errors)/sizeof(*atari_acsi_errors) : + sizeof(scsi_acsi_errors)/sizeof(*scsi_acsi_errors); + + for( --i; i >= 0; --i ) + if (arr[i].code == errcode) break; + if (i >= 0) + printk( ": %s\n", arr[i].text ); +} + +/******************************************************************* + * + * ACSI interrupt routine + * Test, if this is a ACSI interrupt and call the irq handler + * Otherwise ignore this interrupt. + * + *******************************************************************/ + +static irqreturn_t acsi_interrupt(int irq, void *data, struct pt_regs *fp ) + +{ void (*acsi_irq_handler)(void) = do_acsi; + + do_acsi = NULL; + CLEAR_TIMER(); + + if (!acsi_irq_handler) + acsi_irq_handler = unexpected_acsi_interrupt; + acsi_irq_handler(); + return IRQ_HANDLED; +} + + +/****************************************************************** + * + * The Interrupt handlers + * + *******************************************************************/ + + +static void unexpected_acsi_interrupt( void ) + +{ + printk( KERN_WARNING "Unexpected ACSI interrupt\n" ); +} + + +/* This function is called in case of errors. Because we cannot reset + * the ACSI bus or a single device, there is no other choice than + * retrying several times :-( + */ + +static void bad_rw_intr( void ) + +{ + if (!CURRENT) + return; + + if (++CURRENT->errors >= MAX_ERRORS) + end_request(CURRENT, 0); + /* Otherwise just retry */ +} + + +static void read_intr( void ) + +{ int status; + + status = acsi_getstatus(); + if (status != 0) { + struct gendisk *disk = CURRENT->rq_disk; + struct acsi_info_struct *aip = disk->private_data; + printk(KERN_ERR "%s: ", disk->disk_name); + if (!acsi_reqsense(acsi_buffer, aip->target, aip->lun)) + printk( "ACSI error and REQUEST SENSE failed (status=0x%02x)\n", status ); + else { + acsi_print_error(acsi_buffer, aip); + if (CARTRCH_STAT(aip, acsi_buffer)) + aip->changed = 1; + } + ENABLE_IRQ(); + bad_rw_intr(); + redo_acsi_request(); + return; + } + + dma_cache_maintenance( virt_to_phys(CurrentBuffer), CurrentNSect*512, 0 ); + if (CurrentBuffer == acsi_buffer) + copy_from_acsibuffer(); + + do_end_requests(); + redo_acsi_request(); +} + + +static void write_intr(void) + +{ int status; + + status = acsi_getstatus(); + if (status != 0) { + struct gendisk *disk = CURRENT->rq_disk; + struct acsi_info_struct *aip = disk->private_data; + printk( KERN_ERR "%s: ", disk->disk_name); + if (!acsi_reqsense( acsi_buffer, aip->target, aip->lun)) + printk( "ACSI error and REQUEST SENSE failed (status=0x%02x)\n", status ); + else { + acsi_print_error(acsi_buffer, aip); + if (CARTRCH_STAT(aip, acsi_buffer)) + aip->changed = 1; + } + bad_rw_intr(); + redo_acsi_request(); + return; + } + + do_end_requests(); + redo_acsi_request(); +} + + +static void acsi_times_out( unsigned long dummy ) + +{ + DISABLE_IRQ(); + if (!do_acsi) return; + + do_acsi = NULL; + printk( KERN_ERR "ACSI timeout\n" ); + if (!CURRENT) + return; + if (++CURRENT->errors >= MAX_ERRORS) { +#ifdef DEBUG + printk( KERN_ERR "ACSI: too many errors.\n" ); +#endif + end_request(CURRENT, 0); + } + + redo_acsi_request(); +} + + + +/*********************************************************************** + * + * Scatter-gather utility functions + * + ***********************************************************************/ + + +static void copy_to_acsibuffer( void ) + +{ int i; + char *src, *dst; + struct buffer_head *bh; + + src = CURRENT->buffer; + dst = acsi_buffer; + bh = CURRENT->bh; + + if (!bh) + memcpy( dst, src, CurrentNSect*512 ); + else + for( i = 0; i < CurrentNReq; ++i ) { + memcpy( dst, src, bh->b_size ); + dst += bh->b_size; + if ((bh = bh->b_reqnext)) + src = bh->b_data; + } +} + + +static void copy_from_acsibuffer( void ) + +{ int i; + char *src, *dst; + struct buffer_head *bh; + + dst = CURRENT->buffer; + src = acsi_buffer; + bh = CURRENT->bh; + + if (!bh) + memcpy( dst, src, CurrentNSect*512 ); + else + for( i = 0; i < CurrentNReq; ++i ) { + memcpy( dst, src, bh->b_size ); + src += bh->b_size; + if ((bh = bh->b_reqnext)) + dst = bh->b_data; + } +} + + +static void do_end_requests( void ) + +{ int i, n; + + if (!CURRENT->bh) { + CURRENT->nr_sectors -= CurrentNSect; + CURRENT->current_nr_sectors -= CurrentNSect; + CURRENT->sector += CurrentNSect; + if (CURRENT->nr_sectors == 0) + end_request(CURRENT, 1); + } + else { + for( i = 0; i < CurrentNReq; ++i ) { + n = CURRENT->bh->b_size >> 9; + CURRENT->nr_sectors -= n; + CURRENT->current_nr_sectors -= n; + CURRENT->sector += n; + end_request(CURRENT, 1); + } + } +} + + + + +/*********************************************************************** + * + * do_acsi_request and friends + * + ***********************************************************************/ + +static void do_acsi_request( request_queue_t * q ) + +{ + stdma_lock( acsi_interrupt, NULL ); + redo_acsi_request(); +} + + +static void redo_acsi_request( void ) +{ + unsigned block, target, lun, nsect; + char *buffer; + unsigned long pbuffer; + struct buffer_head *bh; + struct gendisk *disk; + struct acsi_info_struct *aip; + + repeat: + CLEAR_TIMER(); + + if (do_acsi) + return; + + if (!CURRENT) { + do_acsi = NULL; + ENABLE_IRQ(); + stdma_release(); + return; + } + + disk = CURRENT->rq_disk; + aip = disk->private_data; + if (CURRENT->bh) { + if (!CURRENT->bh && !buffer_locked(CURRENT->bh)) + panic("ACSI: block not locked"); + } + + block = CURRENT->sector; + if (block+CURRENT->nr_sectors >= get_capacity(disk)) { +#ifdef DEBUG + printk( "%s: attempted access for blocks %d...%ld past end of device at block %ld.\n", + disk->disk_name, + block, block + CURRENT->nr_sectors - 1, + get_capacity(disk)); +#endif + end_request(CURRENT, 0); + goto repeat; + } + if (aip->changed) { + printk( KERN_NOTICE "%s: request denied because cartridge has " + "been changed.\n", disk->disk_name); + end_request(CURRENT, 0); + goto repeat; + } + + target = aip->target; + lun = aip->lun; + + /* Find out how many sectors should be transferred from/to + * consecutive buffers and thus can be done with a single command. + */ + buffer = CURRENT->buffer; + pbuffer = virt_to_phys(buffer); + nsect = CURRENT->current_nr_sectors; + CurrentNReq = 1; + + if ((bh = CURRENT->bh) && bh != CURRENT->bhtail) { + if (!STRAM_ADDR(pbuffer)) { + /* If transfer is done via the ACSI buffer anyway, we can + * assemble as much bh's as fit in the buffer. + */ + while( (bh = bh->b_reqnext) ) { + if (nsect + (bh->b_size>>9) > ACSI_BUFFER_SECTORS) break; + nsect += bh->b_size >> 9; + ++CurrentNReq; + if (bh == CURRENT->bhtail) break; + } + buffer = acsi_buffer; + pbuffer = phys_acsi_buffer; + } + else { + unsigned long pendadr, pnewadr; + pendadr = pbuffer + nsect*512; + while( (bh = bh->b_reqnext) ) { + pnewadr = virt_to_phys(bh->b_data); + if (!STRAM_ADDR(pnewadr) || pendadr != pnewadr) break; + nsect += bh->b_size >> 9; + pendadr = pnewadr + bh->b_size; + ++CurrentNReq; + if (bh == CURRENT->bhtail) break; + } + } + } + else { + if (!STRAM_ADDR(pbuffer)) { + buffer = acsi_buffer; + pbuffer = phys_acsi_buffer; + if (nsect > ACSI_BUFFER_SECTORS) + nsect = ACSI_BUFFER_SECTORS; + } + } + CurrentBuffer = buffer; + CurrentNSect = nsect; + + if (rq_data_dir(CURRENT) == WRITE) { + CMDSET_TARG_LUN( write_cmd, target, lun ); + CMDSET_BLOCK( write_cmd, block ); + CMDSET_LEN( write_cmd, nsect ); + if (buffer == acsi_buffer) + copy_to_acsibuffer(); + dma_cache_maintenance( pbuffer, nsect*512, 1 ); + do_acsi = write_intr; + if (!acsicmd_dma( write_cmd, buffer, nsect, 1, 1)) { + do_acsi = NULL; + printk( KERN_ERR "ACSI (write): Timeout in command block\n" ); + bad_rw_intr(); + goto repeat; + } + SET_TIMER(); + return; + } + if (rq_data_dir(CURRENT) == READ) { + CMDSET_TARG_LUN( read_cmd, target, lun ); + CMDSET_BLOCK( read_cmd, block ); + CMDSET_LEN( read_cmd, nsect ); + do_acsi = read_intr; + if (!acsicmd_dma( read_cmd, buffer, nsect, 0, 1)) { + do_acsi = NULL; + printk( KERN_ERR "ACSI (read): Timeout in command block\n" ); + bad_rw_intr(); + goto repeat; + } + SET_TIMER(); + return; + } + panic("unknown ACSI command"); +} + + + +/*********************************************************************** + * + * Misc functions: ioctl, open, release, check_change, ... + * + ***********************************************************************/ + + +static int acsi_ioctl( struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg ) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + struct acsi_info_struct *aip = disk->private_data; + switch (cmd) { + case HDIO_GETGEO: + /* HDIO_GETGEO is supported more for getting the partition's + * start sector... */ + { struct hd_geometry *geo = (struct hd_geometry *)arg; + /* just fake some geometry here, it's nonsense anyway; to make it + * easy, use Adaptec's usual 64/32 mapping */ + put_user( 64, &geo->heads ); + put_user( 32, &geo->sectors ); + put_user( aip->size >> 11, &geo->cylinders ); + put_user(get_start_sect(inode->i_bdev), &geo->start); + return 0; + } + case SCSI_IOCTL_GET_IDLUN: + /* SCSI compatible GET_IDLUN call to get target's ID and LUN number */ + put_user( aip->target | (aip->lun << 8), + &((Scsi_Idlun *) arg)->dev_id ); + put_user( 0, &((Scsi_Idlun *) arg)->host_unique_id ); + return 0; + default: + return -EINVAL; + } +} + + +/* + * Open a device, check for read-only and lock the medium if it is + * removable. + * + * Changes by Martin Rogge, 9th Aug 1995: + * Check whether check_disk_change (and therefore revalidate_acsidisk) + * was successful. They fail when there is no medium in the drive. + * + * The problem of media being changed during an operation can be + * ignored because of the prevent_removal code. + * + * Added check for the validity of the device number. + * + */ + +static int acsi_open( struct inode * inode, struct file * filp ) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + struct acsi_info_struct *aip = disk->private_data; + + if (aip->access_count == 0 && aip->removable) { +#if 0 + aip->changed = 1; /* safety first */ +#endif + check_disk_change( inode->i_bdev ); + if (aip->changed) /* revalidate was not successful (no medium) */ + return -ENXIO; + acsi_prevent_removal(aip, 1); + } + aip->access_count++; + + if (filp && filp->f_mode) { + check_disk_change( inode->i_bdev ); + if (filp->f_mode & 2) { + if (aip->read_only) { + acsi_release( inode, filp ); + return -EROFS; + } + } + } + + return 0; +} + +/* + * Releasing a block device means we sync() it, so that it can safely + * be forgotten about... + */ + +static int acsi_release( struct inode * inode, struct file * file ) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + struct acsi_info_struct *aip = disk->private_data; + if (--aip->access_count == 0 && aip->removable) + acsi_prevent_removal(aip, 0); + return( 0 ); +} + +/* + * Prevent or allow a media change for removable devices. + */ + +static void acsi_prevent_removal(struct acsi_info_struct *aip, int flag) +{ + stdma_lock( NULL, NULL ); + + CMDSET_TARG_LUN(pa_med_rem_cmd, aip->target, aip->lun); + CMDSET_LEN( pa_med_rem_cmd, flag ); + + if (acsicmd_nodma(pa_med_rem_cmd, 0) && acsi_wait_for_IRQ(3*HZ)) + acsi_getstatus(); + /* Do not report errors -- some devices may not know this command. */ + + ENABLE_IRQ(); + stdma_release(); +} + +static int acsi_media_change(struct gendisk *disk) +{ + struct acsi_info_struct *aip = disk->private_data; + + if (!aip->removable) + return 0; + + if (aip->changed) + /* We can be sure that the medium has been changed -- REQUEST + * SENSE has reported this earlier. + */ + return 1; + + /* If the flag isn't set, make a test by reading block 0. + * If errors happen, it seems to be better to say "changed"... + */ + stdma_lock( NULL, NULL ); + CMDSET_TARG_LUN(read_cmd, aip->target, aip->lun); + CMDSET_BLOCK( read_cmd, 0 ); + CMDSET_LEN( read_cmd, 1 ); + if (acsicmd_dma(read_cmd, acsi_buffer, 1, 0, 0) && + acsi_wait_for_IRQ(3*HZ)) { + if (acsi_getstatus()) { + if (acsi_reqsense(acsi_buffer, aip->target, aip->lun)) { + if (CARTRCH_STAT(aip, acsi_buffer)) + aip->changed = 1; + } + else { + printk( KERN_ERR "%s: REQUEST SENSE failed in test for " + "medium change; assuming a change\n", disk->disk_name ); + aip->changed = 1; + } + } + } + else { + printk( KERN_ERR "%s: Test for medium changed timed out; " + "assuming a change\n", disk->disk_name); + aip->changed = 1; + } + ENABLE_IRQ(); + stdma_release(); + + /* Now, after reading a block, the changed status is surely valid. */ + return aip->changed; +} + + +static int acsi_change_blk_size( int target, int lun) + +{ int i; + + for (i=0; i<12; i++) + acsi_buffer[i] = 0; + + acsi_buffer[3] = 8; + acsi_buffer[10] = 2; + CMDSET_TARG_LUN( modeselect_cmd, target, lun); + + if (!acsicmd_dma( modeselect_cmd, acsi_buffer, 1,1,0) || + !acsi_wait_for_IRQ( 3*HZ ) || + acsi_getstatus() != 0 ) { + return(0); + } + return(1); +} + + +static int acsi_mode_sense( int target, int lun, SENSE_DATA *sd ) + +{ + int page; + + CMDSET_TARG_LUN( modesense_cmd, target, lun ); + for (page=0; page<4; page++) { + modesense_cmd[2] = page; + if (!acsicmd_dma( modesense_cmd, acsi_buffer, 1, 0, 0 ) || + !acsi_wait_for_IRQ( 3*HZ ) || + acsi_getstatus()) + continue; + + /* read twice to jump over the second 16-byte border! */ + udelay(300); + if (acsi_wait_for_noIRQ( 20 ) && + acsicmd_nodma( modesense_cmd, 0 ) && + acsi_wait_for_IRQ( 3*HZ ) && + acsi_getstatus() == 0) + break; + } + if (page == 4) { + return(0); + } + + dma_cache_maintenance( phys_acsi_buffer, sizeof(SENSE_DATA), 0 ); + *sd = *(SENSE_DATA *)acsi_buffer; + + /* Validity check, depending on type of data */ + + switch( SENSE_TYPE(*sd) ) { + + case SENSE_TYPE_ATARI: + if (CAPACITY(*sd) == 0) + goto invalid_sense; + break; + + case SENSE_TYPE_SCSI: + if (sd->scsi.descriptor_size != 8) + goto invalid_sense; + break; + + case SENSE_TYPE_UNKNOWN: + + printk( KERN_ERR "ACSI target %d, lun %d: Cannot interpret " + "sense data\n", target, lun ); + + invalid_sense: + +#ifdef DEBUG + { int i; + printk( "Mode sense data for ACSI target %d, lun %d seem not valid:", + target, lun ); + for( i = 0; i < sizeof(SENSE_DATA); ++i ) + printk( "%02x ", (unsigned char)acsi_buffer[i] ); + printk( "\n" ); + } +#endif + return( 0 ); + } + + return( 1 ); +} + + + +/******************************************************************* + * + * Initialization + * + ********************************************************************/ + + +extern struct block_device_operations acsi_fops; + +static struct gendisk *acsi_gendisk[MAX_DEV]; + +#define MAX_SCSI_DEVICE_CODE 10 + +static const char *const scsi_device_types[MAX_SCSI_DEVICE_CODE] = +{ + "Direct-Access ", + "Sequential-Access", + "Printer ", + "Processor ", + "WORM ", + "CD-ROM ", + "Scanner ", + "Optical Device ", + "Medium Changer ", + "Communications " +}; + +static void print_inquiry(unsigned char *data) +{ + int i; + + printk(KERN_INFO " Vendor: "); + for (i = 8; i < 16; i++) + { + if (data[i] >= 0x20 && i < data[4] + 5) + printk("%c", data[i]); + else + printk(" "); + } + + printk(" Model: "); + for (i = 16; i < 32; i++) + { + if (data[i] >= 0x20 && i < data[4] + 5) + printk("%c", data[i]); + else + printk(" "); + } + + printk(" Rev: "); + for (i = 32; i < 36; i++) + { + if (data[i] >= 0x20 && i < data[4] + 5) + printk("%c", data[i]); + else + printk(" "); + } + + printk("\n"); + + i = data[0] & 0x1f; + + printk(KERN_INFO " Type: %s ", (i < MAX_SCSI_DEVICE_CODE + ? scsi_device_types[i] + : "Unknown ")); + printk(" ANSI SCSI revision: %02x", data[2] & 0x07); + if ((data[2] & 0x07) == 1 && (data[3] & 0x0f) == 1) + printk(" CCS\n"); + else + printk("\n"); +} + + +/* + * Changes by Martin Rogge, 9th Aug 1995: + * acsi_devinit has been taken out of acsi_geninit, because it needs + * to be called from revalidate_acsidisk. The result of request sense + * is now checked for DRIVE NOT READY. + * + * The structure *aip is only valid when acsi_devinit returns + * DEV_SUPPORTED. + * + */ + +#define DEV_NONE 0 +#define DEV_UNKNOWN 1 +#define DEV_SUPPORTED 2 +#define DEV_SLM 3 + +static int acsi_devinit(struct acsi_info_struct *aip) +{ + int status, got_inquiry; + SENSE_DATA sense; + unsigned char reqsense, extsense; + + /*****************************************************************/ + /* Do a TEST UNIT READY command to test the presence of a device */ + /*****************************************************************/ + + CMDSET_TARG_LUN(tur_cmd, aip->target, aip->lun); + if (!acsicmd_nodma(tur_cmd, 0)) { + /* timed out -> no device here */ +#ifdef DEBUG_DETECT + printk("target %d lun %d: timeout\n", aip->target, aip->lun); +#endif + return DEV_NONE; + } + + /*************************/ + /* Read the ACSI status. */ + /*************************/ + + status = acsi_getstatus(); + if (status) { + if (status == 0x12) { + /* The SLM printer should be the only device that + * responds with the error code in the status byte. In + * correct status bytes, bit 4 is never set. + */ + printk( KERN_INFO "Detected SLM printer at id %d lun %d\n", + aip->target, aip->lun); + return DEV_SLM; + } + /* ignore CHECK CONDITION, since some devices send a + UNIT ATTENTION */ + if ((status & 0x1e) != 0x2) { +#ifdef DEBUG_DETECT + printk("target %d lun %d: status %d\n", + aip->target, aip->lun, status); +#endif + return DEV_UNKNOWN; + } + } + + /*******************************/ + /* Do a REQUEST SENSE command. */ + /*******************************/ + + if (!acsi_reqsense(acsi_buffer, aip->target, aip->lun)) { + printk( KERN_WARNING "acsi_reqsense failed\n"); + acsi_buffer[0] = 0; + acsi_buffer[2] = UNIT_ATTENTION; + } + reqsense = acsi_buffer[0]; + extsense = acsi_buffer[2] & 0xf; + if (status) { + if ((reqsense & 0x70) == 0x70) { /* extended sense */ + if (extsense != UNIT_ATTENTION && + extsense != NOT_READY) { +#ifdef DEBUG_DETECT + printk("target %d lun %d: extended sense %d\n", + aip->target, aip->lun, extsense); +#endif + return DEV_UNKNOWN; + } + } + else { + if (reqsense & 0x7f) { +#ifdef DEBUG_DETECT + printk("target %d lun %d: sense %d\n", + aip->target, aip->lun, reqsense); +#endif + return DEV_UNKNOWN; + } + } + } + else + if (reqsense == 0x4) { /* SH204 Bug workaround */ +#ifdef DEBUG_DETECT + printk("target %d lun %d status=0 sense=4\n", + aip->target, aip->lun); +#endif + return DEV_UNKNOWN; + } + + /***********************************************************/ + /* Do an INQUIRY command to get more infos on this device. */ + /***********************************************************/ + + /* Assume default values */ + aip->removable = 1; + aip->read_only = 0; + aip->old_atari_disk = 0; + aip->changed = (extsense == NOT_READY); /* medium inserted? */ + aip->size = DEFAULT_SIZE; + got_inquiry = 0; + /* Fake inquiry result for old atari disks */ + memcpy(acsi_buffer, "\000\000\001\000 Adaptec 40xx" + " ", 40); + CMDSET_TARG_LUN(inquiry_cmd, aip->target, aip->lun); + if (acsicmd_dma(inquiry_cmd, acsi_buffer, 1, 0, 0) && + acsi_getstatus() == 0) { + acsicmd_nodma(inquiry_cmd, 0); + acsi_getstatus(); + dma_cache_maintenance( phys_acsi_buffer, 256, 0 ); + got_inquiry = 1; + aip->removable = !!(acsi_buffer[1] & 0x80); + } + if (aip->type == NONE) /* only at boot time */ + print_inquiry(acsi_buffer); + switch(acsi_buffer[0]) { + case TYPE_DISK: + aip->type = HARDDISK; + break; + case TYPE_ROM: + aip->type = CDROM; + aip->read_only = 1; + break; + default: + return DEV_UNKNOWN; + } + /****************************/ + /* Do a MODE SENSE command. */ + /****************************/ + + if (!acsi_mode_sense(aip->target, aip->lun, &sense)) { + printk( KERN_WARNING "No mode sense data.\n" ); + return DEV_UNKNOWN; + } + if ((SECTOR_SIZE(sense) != 512) && + ((aip->type != CDROM) || + !acsi_change_blk_size(aip->target, aip->lun) || + !acsi_mode_sense(aip->target, aip->lun, &sense) || + (SECTOR_SIZE(sense) != 512))) { + printk( KERN_WARNING "Sector size != 512 not supported.\n" ); + return DEV_UNKNOWN; + } + /* There are disks out there that claim to have 0 sectors... */ + if (CAPACITY(sense)) + aip->size = CAPACITY(sense); /* else keep DEFAULT_SIZE */ + if (!got_inquiry && SENSE_TYPE(sense) == SENSE_TYPE_ATARI) { + /* If INQUIRY failed and the sense data suggest an old + * Atari disk (SH20x, Megafile), the disk is not removable + */ + aip->removable = 0; + aip->old_atari_disk = 1; + } + + /******************/ + /* We've done it. */ + /******************/ + + return DEV_SUPPORTED; +} + +EXPORT_SYMBOL(acsi_delay_start); +EXPORT_SYMBOL(acsi_delay_end); +EXPORT_SYMBOL(acsi_wait_for_IRQ); +EXPORT_SYMBOL(acsi_wait_for_noIRQ); +EXPORT_SYMBOL(acsicmd_nodma); +EXPORT_SYMBOL(acsi_getstatus); +EXPORT_SYMBOL(acsi_buffer); +EXPORT_SYMBOL(phys_acsi_buffer); + +#ifdef CONFIG_ATARI_SLM_MODULE +void acsi_attach_SLMs( int (*attach_func)( int, int ) ); + +EXPORT_SYMBOL(acsi_extstatus); +EXPORT_SYMBOL(acsi_end_extstatus); +EXPORT_SYMBOL(acsi_extcmd); +EXPORT_SYMBOL(acsi_attach_SLMs); + +/* to remember IDs of SLM devices, SLM module is loaded later + * (index is target#, contents is lun#, -1 means "no SLM") */ +int SLM_devices[8]; +#endif + +static struct block_device_operations acsi_fops = { + .owner = THIS_MODULE, + .open = acsi_open, + .release = acsi_release, + .ioctl = acsi_ioctl, + .media_changed = acsi_media_change, + .revalidate_disk= acsi_revalidate, +}; + +#ifdef CONFIG_ATARI_SLM_MODULE +/* call attach_slm() for each device that is a printer; needed for init of SLM + * driver as a module, since it's not yet present if acsi.c is inited and thus + * the bus gets scanned. */ +void acsi_attach_SLMs( int (*attach_func)( int, int ) ) +{ + int i, n = 0; + + for( i = 0; i < 8; ++i ) + if (SLM_devices[i] >= 0) + n += (*attach_func)( i, SLM_devices[i] ); + printk( KERN_INFO "Found %d SLM printer(s) total.\n", n ); +} +#endif /* CONFIG_ATARI_SLM_MODULE */ + + +int acsi_init( void ) +{ + int err = 0; + int i, target, lun; + struct acsi_info_struct *aip; +#ifdef CONFIG_ATARI_SLM + int n_slm = 0; +#endif + if (!MACH_IS_ATARI || !ATARIHW_PRESENT(ACSI)) + return 0; + if (register_blkdev(ACSI_MAJOR, "ad")) { + err = -EBUSY; + goto out1; + } + if (!(acsi_buffer = + (char *)atari_stram_alloc(ACSI_BUFFER_SIZE, "acsi"))) { + err = -ENOMEM; + printk( KERN_ERR "Unable to get ACSI ST-Ram buffer.\n" ); + goto out2; + } + phys_acsi_buffer = virt_to_phys( acsi_buffer ); + STramMask = ATARIHW_PRESENT(EXTD_DMA) ? 0x00000000 : 0xff000000; + + acsi_queue = blk_init_queue(do_acsi_request, &acsi_lock); + if (!acsi_queue) { + err = -ENOMEM; + goto out2a; + } +#ifdef CONFIG_ATARI_SLM + err = slm_init(); +#endif + if (err) + goto out3; + + printk( KERN_INFO "Probing ACSI devices:\n" ); + NDevices = 0; +#ifdef CONFIG_ATARI_SLM_MODULE + for( i = 0; i < 8; ++i ) + SLM_devices[i] = -1; +#endif + stdma_lock(NULL, NULL); + + for (target = 0; target < 8 && NDevices < MAX_DEV; ++target) { + lun = 0; + do { + aip = &acsi_info[NDevices]; + aip->type = NONE; + aip->target = target; + aip->lun = lun; + i = acsi_devinit(aip); + switch (i) { + case DEV_SUPPORTED: + printk( KERN_INFO "Detected "); + switch (aip->type) { + case HARDDISK: + printk("disk"); + break; + case CDROM: + printk("cdrom"); + break; + default: + } + printk(" ad%c at id %d lun %d ", + 'a' + NDevices, target, lun); + if (aip->removable) + printk("(removable) "); + if (aip->read_only) + printk("(read-only) "); + if (aip->size == DEFAULT_SIZE) + printk(" unkown size, using default "); + printk("%ld MByte\n", + (aip->size*512+1024*1024/2)/(1024*1024)); + NDevices++; + break; + case DEV_SLM: +#ifdef CONFIG_ATARI_SLM + n_slm += attach_slm( target, lun ); + break; +#endif +#ifdef CONFIG_ATARI_SLM_MODULE + SLM_devices[target] = lun; + break; +#endif + /* neither of the above: fall through to unknown device */ + case DEV_UNKNOWN: + printk( KERN_INFO "Detected unsupported device at " + "id %d lun %d\n", target, lun); + break; + } + } +#ifdef CONFIG_ACSI_MULTI_LUN + while (i != DEV_NONE && ++lun < MAX_LUN); +#else + while (0); +#endif + } + + /* reenable interrupt */ + ENABLE_IRQ(); + stdma_release(); + +#ifndef CONFIG_ATARI_SLM + printk( KERN_INFO "Found %d ACSI device(s) total.\n", NDevices ); +#else + printk( KERN_INFO "Found %d ACSI device(s) and %d SLM printer(s) total.\n", + NDevices, n_slm ); +#endif + err = -ENOMEM; + for( i = 0; i < NDevices; ++i ) { + acsi_gendisk[i] = alloc_disk(16); + if (!acsi_gendisk[i]) + goto out4; + } + + for( i = 0; i < NDevices; ++i ) { + struct gendisk *disk = acsi_gendisk[i]; + sprintf(disk->disk_name, "ad%c", 'a'+i); + aip = &acsi_info[NDevices]; + sprintf(disk->devfs_name, "ad/target%d/lun%d", aip->target, aip->lun); + disk->major = ACSI_MAJOR; + disk->first_minor = i << 4; + if (acsi_info[i].type != HARDDISK) { + disk->minors = 1; + strcat(disk->devfs_name, "/disc"); + } + disk->fops = &acsi_fops; + disk->private_data = &acsi_info[i]; + set_capacity(disk, acsi_info[i].size); + disk->queue = acsi_queue; + add_disk(disk); + } + return 0; +out4: + while (i--) + put_disk(acsi_gendisk[i]); +out3: + blk_cleanup_queue(acsi_queue); +out2a: + atari_stram_free( acsi_buffer ); +out2: + unregister_blkdev( ACSI_MAJOR, "ad" ); +out1: + return err; +} + + +#ifdef MODULE + +MODULE_LICENSE("GPL"); + +int init_module(void) +{ + int err; + + if ((err = acsi_init())) + return( err ); + printk( KERN_INFO "ACSI driver loaded as module.\n"); + return( 0 ); +} + +void cleanup_module(void) +{ + int i; + del_timer( &acsi_timer ); + blk_cleanup_queue(acsi_queue); + atari_stram_free( acsi_buffer ); + + if (unregister_blkdev( ACSI_MAJOR, "ad" ) != 0) + printk( KERN_ERR "acsi: cleanup_module failed\n"); + + for (i = 0; i < NDevices; i++) { + del_gendisk(acsi_gendisk[i]); + put_disk(acsi_gendisk[i]); + } +} +#endif + +/* + * This routine is called to flush all partitions and partition tables + * for a changed scsi disk, and then re-read the new partition table. + * If we are revalidating a disk because of a media change, then we + * enter with usage == 0. If we are using an ioctl, we automatically have + * usage == 1 (we need an open channel to use an ioctl :-), so this + * is our limit. + * + * Changes by Martin Rogge, 9th Aug 1995: + * got cd-roms to work by calling acsi_devinit. There are only two problems: + * First, if there is no medium inserted, the status will remain "changed". + * That is no problem at all, but our design of three-valued logic (medium + * changed, medium not changed, no medium inserted). + * Secondly the check could fail completely and the drive could deliver + * nonsensical data, which could mess up the acsi_info[] structure. In + * that case we try to make the entry safe. + * + */ + +static int acsi_revalidate(struct gendisk *disk) +{ + struct acsi_info_struct *aip = disk->private_data; + stdma_lock( NULL, NULL ); + if (acsi_devinit(aip) != DEV_SUPPORTED) { + printk( KERN_ERR "ACSI: revalidate failed for target %d lun %d\n", + aip->target, aip->lun); + aip->size = 0; + aip->read_only = 1; + aip->removable = 1; + aip->changed = 1; /* next acsi_open will try again... */ + } + + ENABLE_IRQ(); + stdma_release(); + set_capacity(disk, aip->size); + return 0; +} diff --git a/drivers/block/acsi_slm.c b/drivers/block/acsi_slm.c new file mode 100644 index 00000000000..e3be8c31a74 --- /dev/null +++ b/drivers/block/acsi_slm.c @@ -0,0 +1,1045 @@ +/* + * acsi_slm.c -- Device driver for the Atari SLM laser printer + * + * Copyright 1995 Roman Hodek <Roman.Hodek@informatik.uni-erlangen.de> + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file COPYING in the main directory of this archive for + * more details. + * + */ + +/* + +Notes: + +The major number for SLM printers is 28 (like ACSI), but as a character +device, not block device. The minor number is the number of the printer (if +you have more than one SLM; currently max. 2 (#define-constant) SLMs are +supported). The device can be opened for reading and writing. If reading it, +you get some status infos (MODE SENSE data). Writing mode is used for the data +to be printed. Some ioctls allow to get the printer status and to tune printer +modes and some internal variables. + +A special problem of the SLM driver is the timing and thus the buffering of +the print data. The problem is that all the data for one page must be present +in memory when printing starts, else --when swapping occurs-- the timing could +not be guaranteed. There are several ways to assure this: + + 1) Reserve a buffer of 1196k (maximum page size) statically by + atari_stram_alloc(). The data are collected there until they're complete, + and then printing starts. Since the buffer is reserved, no further + considerations about memory and swapping are needed. So this is the + simplest method, but it needs a lot of memory for just the SLM. + + An striking advantage of this method is (supposed the SLM_CONT_CNT_REPROG + method works, see there), that there are no timing problems with the DMA + anymore. + + 2) The other method would be to reserve the buffer dynamically each time + printing is required. I could think of looking at mem_map where the + largest unallocted ST-RAM area is, taking the area, and then extending it + by swapping out the neighbored pages, until the needed size is reached. + This requires some mm hacking, but seems possible. The only obstacle could + be pages that cannot be swapped out (reserved pages)... + + 3) Another possibility would be to leave the real data in user space and to + work with two dribble buffers of about 32k in the driver: While the one + buffer is DMAed to the SLM, the other can be filled with new data. But + to keep the timing, that requires that the user data remain in memory and + are not swapped out. Requires mm hacking, too, but maybe not so bad as + method 2). + +*/ + +#include <linux/module.h> + +#include <linux/errno.h> +#include <linux/sched.h> +#include <linux/timer.h> +#include <linux/fs.h> +#include <linux/major.h> +#include <linux/kernel.h> +#include <linux/delay.h> +#include <linux/interrupt.h> +#include <linux/time.h> +#include <linux/mm.h> +#include <linux/slab.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/smp_lock.h> + +#include <asm/pgtable.h> +#include <asm/system.h> +#include <asm/uaccess.h> +#include <asm/atarihw.h> +#include <asm/atariints.h> +#include <asm/atari_acsi.h> +#include <asm/atari_stdma.h> +#include <asm/atari_stram.h> +#include <asm/atari_SLM.h> + + +#undef DEBUG + +/* Define this if the page data are continuous in physical memory. That + * requires less reprogramming of the ST-DMA */ +#define SLM_CONTINUOUS_DMA + +/* Use continuous reprogramming of the ST-DMA counter register. This is + * --strictly speaking-- not allowed, Atari recommends not to look at the + * counter register while a DMA is going on. But I don't know if that applies + * only for reading the register, or also writing to it. Writing only works + * fine for me... The advantage is that the timing becomes absolutely + * uncritical: Just update each, say 200ms, the counter reg to its maximum, + * and the DMA will work until the status byte interrupt occurs. + */ +#define SLM_CONT_CNT_REPROG + +#define CMDSET_TARG_LUN(cmd,targ,lun) \ + do { \ + cmd[0] = (cmd[0] & ~0xe0) | (targ)<<5; \ + cmd[1] = (cmd[1] & ~0xe0) | (lun)<<5; \ + } while(0) + +#define START_TIMER(to) mod_timer(&slm_timer, jiffies + (to)) +#define STOP_TIMER() del_timer(&slm_timer) + + +static char slmreqsense_cmd[6] = { 0x03, 0, 0, 0, 0, 0 }; +static char slmprint_cmd[6] = { 0x0a, 0, 0, 0, 0, 0 }; +static char slminquiry_cmd[6] = { 0x12, 0, 0, 0, 0, 0x80 }; +static char slmmsense_cmd[6] = { 0x1a, 0, 0, 0, 255, 0 }; +#if 0 +static char slmmselect_cmd[6] = { 0x15, 0, 0, 0, 0, 0 }; +#endif + + +#define MAX_SLM 2 + +static struct slm { + unsigned target; /* target number */ + unsigned lun; /* LUN in target controller */ + atomic_t wr_ok; /* set to 0 if output part busy */ + atomic_t rd_ok; /* set to 0 if status part busy */ +} slm_info[MAX_SLM]; + +int N_SLM_Printers = 0; + +/* printer buffer */ +static unsigned char *SLMBuffer; /* start of buffer */ +static unsigned char *BufferP; /* current position in buffer */ +static int BufferSize; /* length of buffer for page size */ + +typedef enum { IDLE, FILLING, PRINTING } SLMSTATE; +static SLMSTATE SLMState; +static int SLMBufOwner; /* SLM# currently using the buffer */ + +/* DMA variables */ +#ifndef SLM_CONT_CNT_REPROG +static unsigned long SLMCurAddr; /* current base addr of DMA chunk */ +static unsigned long SLMEndAddr; /* expected end addr */ +static unsigned long SLMSliceSize; /* size of one DMA chunk */ +#endif +static int SLMError; + +/* wait queues */ +static DECLARE_WAIT_QUEUE_HEAD(slm_wait); /* waiting for buffer */ +static DECLARE_WAIT_QUEUE_HEAD(print_wait); /* waiting for printing finished */ + +/* status codes */ +#define SLMSTAT_OK 0x00 +#define SLMSTAT_ORNERY 0x02 +#define SLMSTAT_TONER 0x03 +#define SLMSTAT_WARMUP 0x04 +#define SLMSTAT_PAPER 0x05 +#define SLMSTAT_DRUM 0x06 +#define SLMSTAT_INJAM 0x07 +#define SLMSTAT_THRJAM 0x08 +#define SLMSTAT_OUTJAM 0x09 +#define SLMSTAT_COVER 0x0a +#define SLMSTAT_FUSER 0x0b +#define SLMSTAT_IMAGER 0x0c +#define SLMSTAT_MOTOR 0x0d +#define SLMSTAT_VIDEO 0x0e +#define SLMSTAT_SYSTO 0x10 +#define SLMSTAT_OPCODE 0x12 +#define SLMSTAT_DEVNUM 0x15 +#define SLMSTAT_PARAM 0x1a +#define SLMSTAT_ACSITO 0x1b /* driver defined */ +#define SLMSTAT_NOTALL 0x1c /* driver defined */ + +static char *SLMErrors[] = { + /* 0x00 */ "OK and ready", + /* 0x01 */ NULL, + /* 0x02 */ "ornery printer", + /* 0x03 */ "toner empty", + /* 0x04 */ "warming up", + /* 0x05 */ "paper empty", + /* 0x06 */ "drum empty", + /* 0x07 */ "input jam", + /* 0x08 */ "through jam", + /* 0x09 */ "output jam", + /* 0x0a */ "cover open", + /* 0x0b */ "fuser malfunction", + /* 0x0c */ "imager malfunction", + /* 0x0d */ "motor malfunction", + /* 0x0e */ "video malfunction", + /* 0x0f */ NULL, + /* 0x10 */ "printer system timeout", + /* 0x11 */ NULL, + /* 0x12 */ "invalid operation code", + /* 0x13 */ NULL, + /* 0x14 */ NULL, + /* 0x15 */ "invalid device number", + /* 0x16 */ NULL, + /* 0x17 */ NULL, + /* 0x18 */ NULL, + /* 0x19 */ NULL, + /* 0x1a */ "invalid parameter list", + /* 0x1b */ "ACSI timeout", + /* 0x1c */ "not all printed" +}; + +#define N_ERRORS (sizeof(SLMErrors)/sizeof(*SLMErrors)) + +/* real (driver caused) error? */ +#define IS_REAL_ERROR(x) (x > 0x10) + + +static struct { + char *name; + int w, h; +} StdPageSize[] = { + { "Letter", 2400, 3180 }, + { "Legal", 2400, 4080 }, + { "A4", 2336, 3386 }, + { "B5", 2016, 2914 } +}; + +#define N_STD_SIZES (sizeof(StdPageSize)/sizeof(*StdPageSize)) + +#define SLM_BUFFER_SIZE (2336*3386/8) /* A4 for now */ +#define SLM_DMA_AMOUNT 255 /* #sectors to program the DMA for */ + +#ifdef SLM_CONTINUOUS_DMA +# define SLM_DMA_INT_OFFSET 0 /* DMA goes until seccnt 0, no offs */ +# define SLM_DMA_END_OFFSET 32 /* 32 Byte ST-DMA FIFO */ +# define SLM_SLICE_SIZE(w) (255*512) +#else +# define SLM_DMA_INT_OFFSET 32 /* 32 Byte ST-DMA FIFO */ +# define SLM_DMA_END_OFFSET 32 /* 32 Byte ST-DMA FIFO */ +# define SLM_SLICE_SIZE(w) ((254*512)/(w/8)*(w/8)) +#endif + +/* calculate the number of jiffies to wait for 'n' bytes */ +#ifdef SLM_CONT_CNT_REPROG +#define DMA_TIME_FOR(n) 50 +#define DMA_STARTUP_TIME 0 +#else +#define DMA_TIME_FOR(n) (n/1400-1) +#define DMA_STARTUP_TIME 650 +#endif + +/***************************** Prototypes *****************************/ + +static char *slm_errstr( int stat ); +static int slm_getstats( char *buffer, int device ); +static ssize_t slm_read( struct file* file, char *buf, size_t count, loff_t + *ppos ); +static void start_print( int device ); +static irqreturn_t slm_interrupt(int irc, void *data, struct pt_regs *fp); +static void slm_test_ready( unsigned long dummy ); +static void set_dma_addr( unsigned long paddr ); +static unsigned long get_dma_addr( void ); +static ssize_t slm_write( struct file *file, const char *buf, size_t count, + loff_t *ppos ); +static int slm_ioctl( struct inode *inode, struct file *file, unsigned int + cmd, unsigned long arg ); +static int slm_open( struct inode *inode, struct file *file ); +static int slm_release( struct inode *inode, struct file *file ); +static int slm_req_sense( int device ); +static int slm_mode_sense( int device, char *buffer, int abs_flag ); +#if 0 +static int slm_mode_select( int device, char *buffer, int len, int + default_flag ); +#endif +static int slm_get_pagesize( int device, int *w, int *h ); + +/************************* End of Prototypes **************************/ + + +static struct timer_list slm_timer = TIMER_INITIALIZER(slm_test_ready, 0, 0); + +static struct file_operations slm_fops = { + .owner = THIS_MODULE, + .read = slm_read, + .write = slm_write, + .ioctl = slm_ioctl, + .open = slm_open, + .release = slm_release, +}; + + +/* ---------------------------------------------------------------------- */ +/* Status Functions */ + + +static char *slm_errstr( int stat ) + +{ char *p; + static char str[22]; + + stat &= 0x1f; + if (stat >= 0 && stat < N_ERRORS && (p = SLMErrors[stat])) + return( p ); + sprintf( str, "unknown status 0x%02x", stat ); + return( str ); +} + + +static int slm_getstats( char *buffer, int device ) + +{ int len = 0, stat, i, w, h; + unsigned char buf[256]; + + stat = slm_mode_sense( device, buf, 0 ); + if (IS_REAL_ERROR(stat)) + return( -EIO ); + +#define SHORTDATA(i) ((buf[i] << 8) | buf[i+1]) +#define BOOLDATA(i,mask) ((buf[i] & mask) ? "on" : "off") + + w = SHORTDATA( 3 ); + h = SHORTDATA( 1 ); + + len += sprintf( buffer+len, "Status\t\t%s\n", + slm_errstr( stat ) ); + len += sprintf( buffer+len, "Page Size\t%dx%d", + w, h ); + + for( i = 0; i < N_STD_SIZES; ++i ) { + if (w == StdPageSize[i].w && h == StdPageSize[i].h) + break; + } + if (i < N_STD_SIZES) + len += sprintf( buffer+len, " (%s)", StdPageSize[i].name ); + buffer[len++] = '\n'; + + len += sprintf( buffer+len, "Top/Left Margin\t%d/%d\n", + SHORTDATA( 5 ), SHORTDATA( 7 ) ); + len += sprintf( buffer+len, "Manual Feed\t%s\n", + BOOLDATA( 9, 0x01 ) ); + len += sprintf( buffer+len, "Input Select\t%d\n", + (buf[9] >> 1) & 7 ); + len += sprintf( buffer+len, "Auto Select\t%s\n", + BOOLDATA( 9, 0x10 ) ); + len += sprintf( buffer+len, "Prefeed Paper\t%s\n", + BOOLDATA( 9, 0x20 ) ); + len += sprintf( buffer+len, "Thick Pixels\t%s\n", + BOOLDATA( 9, 0x40 ) ); + len += sprintf( buffer+len, "H/V Resol.\t%d/%d dpi\n", + SHORTDATA( 12 ), SHORTDATA( 10 ) ); + len += sprintf( buffer+len, "System Timeout\t%d\n", + buf[14] ); + len += sprintf( buffer+len, "Scan Time\t%d\n", + SHORTDATA( 15 ) ); + len += sprintf( buffer+len, "Page Count\t%d\n", + SHORTDATA( 17 ) ); + len += sprintf( buffer+len, "In/Out Cap.\t%d/%d\n", + SHORTDATA( 19 ), SHORTDATA( 21 ) ); + len += sprintf( buffer+len, "Stagger Output\t%s\n", + BOOLDATA( 23, 0x01 ) ); + len += sprintf( buffer+len, "Output Select\t%d\n", + (buf[23] >> 1) & 7 ); + len += sprintf( buffer+len, "Duplex Print\t%s\n", + BOOLDATA( 23, 0x10 ) ); + len += sprintf( buffer+len, "Color Sep.\t%s\n", + BOOLDATA( 23, 0x20 ) ); + + return( len ); +} + + +static ssize_t slm_read( struct file *file, char *buf, size_t count, + loff_t *ppos ) + +{ + struct inode *node = file->f_dentry->d_inode; + unsigned long page; + int length; + int end; + + if (count < 0) + return( -EINVAL ); + if (!(page = __get_free_page( GFP_KERNEL ))) + return( -ENOMEM ); + + length = slm_getstats( (char *)page, iminor(node) ); + if (length < 0) { + count = length; + goto out; + } + if (file->f_pos >= length) { + count = 0; + goto out; + } + if (count + file->f_pos > length) + count = length - file->f_pos; + end = count + file->f_pos; + if (copy_to_user(buf, (char *)page + file->f_pos, count)) { + count = -EFAULT; + goto out; + } + file->f_pos = end; +out: free_page( page ); + return( count ); +} + + +/* ---------------------------------------------------------------------- */ +/* Printing */ + + +static void start_print( int device ) + +{ struct slm *sip = &slm_info[device]; + unsigned char *cmd; + unsigned long paddr; + int i; + + stdma_lock( slm_interrupt, NULL ); + + CMDSET_TARG_LUN( slmprint_cmd, sip->target, sip->lun ); + cmd = slmprint_cmd; + paddr = virt_to_phys( SLMBuffer ); + dma_cache_maintenance( paddr, virt_to_phys(BufferP)-paddr, 1 ); + DISABLE_IRQ(); + + /* Low on A1 */ + dma_wd.dma_mode_status = 0x88; + MFPDELAY(); + + /* send the command bytes except the last */ + for( i = 0; i < 5; ++i ) { + DMA_LONG_WRITE( *cmd++, 0x8a ); + udelay(20); + if (!acsi_wait_for_IRQ( HZ/2 )) { + SLMError = 1; + return; /* timeout */ + } + } + /* last command byte */ + DMA_LONG_WRITE( *cmd++, 0x82 ); + MFPDELAY(); + /* set DMA address */ + set_dma_addr( paddr ); + /* program DMA for write and select sector counter reg */ + dma_wd.dma_mode_status = 0x192; + MFPDELAY(); + /* program for 255*512 bytes and start DMA */ + DMA_LONG_WRITE( SLM_DMA_AMOUNT, 0x112 ); + +#ifndef SLM_CONT_CNT_REPROG + SLMCurAddr = paddr; + SLMEndAddr = paddr + SLMSliceSize + SLM_DMA_INT_OFFSET; +#endif + START_TIMER( DMA_STARTUP_TIME + DMA_TIME_FOR( SLMSliceSize )); +#if !defined(SLM_CONT_CNT_REPROG) && defined(DEBUG) + printk( "SLM: CurAddr=%#lx EndAddr=%#lx timer=%ld\n", + SLMCurAddr, SLMEndAddr, DMA_TIME_FOR( SLMSliceSize ) ); +#endif + + ENABLE_IRQ(); +} + + +/* Only called when an error happened or at the end of a page */ + +static irqreturn_t slm_interrupt(int irc, void *data, struct pt_regs *fp) + +{ unsigned long addr; + int stat; + + STOP_TIMER(); + addr = get_dma_addr(); + stat = acsi_getstatus(); + SLMError = (stat < 0) ? SLMSTAT_ACSITO : + (addr < virt_to_phys(BufferP)) ? SLMSTAT_NOTALL : + stat; + + dma_wd.dma_mode_status = 0x80; + MFPDELAY(); +#ifdef DEBUG + printk( "SLM: interrupt, addr=%#lx, error=%d\n", addr, SLMError ); +#endif + + wake_up( &print_wait ); + stdma_release(); + ENABLE_IRQ(); + return IRQ_HANDLED; +} + + +static void slm_test_ready( unsigned long dummy ) + +{ +#ifdef SLM_CONT_CNT_REPROG + /* program for 255*512 bytes again */ + dma_wd.fdc_acces_seccount = SLM_DMA_AMOUNT; + START_TIMER( DMA_TIME_FOR(0) ); +#ifdef DEBUG + printk( "SLM: reprogramming timer for %d jiffies, addr=%#lx\n", + DMA_TIME_FOR(0), get_dma_addr() ); +#endif + +#else /* !SLM_CONT_CNT_REPROG */ + + unsigned long flags, addr; + int d, ti; +#ifdef DEBUG + struct timeval start_tm, end_tm; + int did_wait = 0; +#endif + + local_irq_save(flags); + + addr = get_dma_addr(); + if ((d = SLMEndAddr - addr) > 0) { + local_irq_restore(flags); + + /* slice not yet finished, decide whether to start another timer or to + * busy-wait */ + ti = DMA_TIME_FOR( d ); + if (ti > 0) { +#ifdef DEBUG + printk( "SLM: reprogramming timer for %d jiffies, rest %d bytes\n", + ti, d ); +#endif + START_TIMER( ti ); + return; + } + /* wait for desired end address to be reached */ +#ifdef DEBUG + do_gettimeofday( &start_tm ); + did_wait = 1; +#endif + local_irq_disable(); + while( get_dma_addr() < SLMEndAddr ) + barrier(); + } + + /* slice finished, start next one */ + SLMCurAddr += SLMSliceSize; + +#ifdef SLM_CONTINUOUS_DMA + /* program for 255*512 bytes again */ + dma_wd.fdc_acces_seccount = SLM_DMA_AMOUNT; +#else + /* set DMA address; + * add 2 bytes for the ones in the SLM controller FIFO! */ + set_dma_addr( SLMCurAddr + 2 ); + /* toggle DMA to write and select sector counter reg */ + dma_wd.dma_mode_status = 0x92; + MFPDELAY(); + dma_wd.dma_mode_status = 0x192; + MFPDELAY(); + /* program for 255*512 bytes and start DMA */ + DMA_LONG_WRITE( SLM_DMA_AMOUNT, 0x112 ); +#endif + + local_irq_restore(flags); + +#ifdef DEBUG + if (did_wait) { + int ms; + do_gettimeofday( &end_tm ); + ms = (end_tm.tv_sec*1000000+end_tm.tv_usec) - + (start_tm.tv_sec*1000000+start_tm.tv_usec); + printk( "SLM: did %ld.%ld ms busy waiting for %d bytes\n", + ms/1000, ms%1000, d ); + } + else + printk( "SLM: didn't wait (!)\n" ); +#endif + + if ((unsigned char *)PTOV( SLMCurAddr + SLMSliceSize ) >= BufferP) { + /* will be last slice, no timer necessary */ +#ifdef DEBUG + printk( "SLM: CurAddr=%#lx EndAddr=%#lx last slice -> no timer\n", + SLMCurAddr, SLMEndAddr ); +#endif + } + else { + /* not last slice */ + SLMEndAddr = SLMCurAddr + SLMSliceSize + SLM_DMA_INT_OFFSET; + START_TIMER( DMA_TIME_FOR( SLMSliceSize )); +#ifdef DEBUG + printk( "SLM: CurAddr=%#lx EndAddr=%#lx timer=%ld\n", + SLMCurAddr, SLMEndAddr, DMA_TIME_FOR( SLMSliceSize ) ); +#endif + } +#endif /* SLM_CONT_CNT_REPROG */ +} + + +static void set_dma_addr( unsigned long paddr ) + +{ unsigned long flags; + + local_irq_save(flags); + dma_wd.dma_lo = (unsigned char)paddr; + paddr >>= 8; + MFPDELAY(); + dma_wd.dma_md = (unsigned char)paddr; + paddr >>= 8; + MFPDELAY(); + if (ATARIHW_PRESENT( EXTD_DMA )) + st_dma_ext_dmahi = (unsigned short)paddr; + else + dma_wd.dma_hi = (unsigned char)paddr; + MFPDELAY(); + local_irq_restore(flags); +} + + +static unsigned long get_dma_addr( void ) + +{ unsigned long addr; + + addr = dma_wd.dma_lo & 0xff; + MFPDELAY(); + addr |= (dma_wd.dma_md & 0xff) << 8; + MFPDELAY(); + addr |= (dma_wd.dma_hi & 0xff) << 16; + MFPDELAY(); + + return( addr ); +} + + +static ssize_t slm_write( struct file *file, const char *buf, size_t count, + loff_t *ppos ) + +{ + struct inode *node = file->f_dentry->d_inode; + int device = iminor(node); + int n, filled, w, h; + + while( SLMState == PRINTING || + (SLMState == FILLING && SLMBufOwner != device) ) { + interruptible_sleep_on( &slm_wait ); + if (signal_pending(current)) + return( -ERESTARTSYS ); + } + if (SLMState == IDLE) { + /* first data of page: get current page size */ + if (slm_get_pagesize( device, &w, &h )) + return( -EIO ); + BufferSize = w*h/8; + if (BufferSize > SLM_BUFFER_SIZE) + return( -ENOMEM ); + + SLMState = FILLING; + SLMBufOwner = device; + } + + n = count; + filled = BufferP - SLMBuffer; + if (filled + n > BufferSize) + n = BufferSize - filled; + + if (copy_from_user(BufferP, buf, n)) + return -EFAULT; + BufferP += n; + filled += n; + + if (filled == BufferSize) { + /* Check the paper size again! The user may have switched it in the + * time between starting the data and finishing them. Would end up in + * a trashy page... */ + if (slm_get_pagesize( device, &w, &h )) + return( -EIO ); + if (BufferSize != w*h/8) { + printk( KERN_NOTICE "slm%d: page size changed while printing\n", + device ); + return( -EAGAIN ); + } + + SLMState = PRINTING; + /* choose a slice size that is a multiple of the line size */ +#ifndef SLM_CONT_CNT_REPROG + SLMSliceSize = SLM_SLICE_SIZE(w); +#endif + + start_print( device ); + sleep_on( &print_wait ); + if (SLMError && IS_REAL_ERROR(SLMError)) { + printk( KERN_ERR "slm%d: %s\n", device, slm_errstr(SLMError) ); + n = -EIO; + } + + SLMState = IDLE; + BufferP = SLMBuffer; + wake_up_interruptible( &slm_wait ); + } + + return( n ); +} + + +/* ---------------------------------------------------------------------- */ +/* ioctl Functions */ + + +static int slm_ioctl( struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg ) + +{ int device = iminor(inode), err; + + /* I can think of setting: + * - manual feed + * - paper format + * - copy count + * - ... + * but haven't implemented that yet :-) + * BTW, has anybody better docs about the MODE SENSE/MODE SELECT data? + */ + switch( cmd ) { + + case SLMIORESET: /* reset buffer, i.e. empty the buffer */ + if (!(file->f_mode & 2)) + return( -EINVAL ); + if (SLMState == PRINTING) + return( -EBUSY ); + SLMState = IDLE; + BufferP = SLMBuffer; + wake_up_interruptible( &slm_wait ); + return( 0 ); + + case SLMIOGSTAT: { /* get status */ + int stat; + char *str; + + stat = slm_req_sense( device ); + if (arg) { + str = slm_errstr( stat ); + if (put_user(stat, + (long *)&((struct SLM_status *)arg)->stat)) + return -EFAULT; + if (copy_to_user( ((struct SLM_status *)arg)->str, str, + strlen(str) + 1)) + return -EFAULT; + } + return( stat ); + } + + case SLMIOGPSIZE: { /* get paper size */ + int w, h; + + if ((err = slm_get_pagesize( device, &w, &h ))) return( err ); + + if (put_user(w, (long *)&((struct SLM_paper_size *)arg)->width)) + return -EFAULT; + if (put_user(h, (long *)&((struct SLM_paper_size *)arg)->height)) + return -EFAULT; + return( 0 ); + } + + case SLMIOGMFEED: /* get manual feed */ + return( -EINVAL ); + + case SLMIOSPSIZE: /* set paper size */ + return( -EINVAL ); + + case SLMIOSMFEED: /* set manual feed */ + return( -EINVAL ); + + } + return( -EINVAL ); +} + + +/* ---------------------------------------------------------------------- */ +/* Opening and Closing */ + + +static int slm_open( struct inode *inode, struct file *file ) + +{ int device; + struct slm *sip; + + device = iminor(inode); + if (device >= N_SLM_Printers) + return( -ENXIO ); + sip = &slm_info[device]; + + if (file->f_mode & 2) { + /* open for writing is exclusive */ + if ( !atomic_dec_and_test(&sip->wr_ok) ) { + atomic_inc(&sip->wr_ok); + return( -EBUSY ); + } + } + if (file->f_mode & 1) { + /* open for reading is exclusive */ + if ( !atomic_dec_and_test(&sip->rd_ok) ) { + atomic_inc(&sip->rd_ok); + return( -EBUSY ); + } + } + + return( 0 ); +} + + +static int slm_release( struct inode *inode, struct file *file ) + +{ int device; + struct slm *sip; + + device = iminor(inode); + sip = &slm_info[device]; + + if (file->f_mode & 2) + atomic_inc( &sip->wr_ok ); + if (file->f_mode & 1) + atomic_inc( &sip->rd_ok ); + + return( 0 ); +} + + +/* ---------------------------------------------------------------------- */ +/* ACSI Primitives for the SLM */ + + +static int slm_req_sense( int device ) + +{ int stat, rv; + struct slm *sip = &slm_info[device]; + + stdma_lock( NULL, NULL ); + + CMDSET_TARG_LUN( slmreqsense_cmd, sip->target, sip->lun ); + if (!acsicmd_nodma( slmreqsense_cmd, 0 ) || + (stat = acsi_getstatus()) < 0) + rv = SLMSTAT_ACSITO; + else + rv = stat & 0x1f; + + ENABLE_IRQ(); + stdma_release(); + return( rv ); +} + + +static int slm_mode_sense( int device, char *buffer, int abs_flag ) + +{ unsigned char stat, len; + int rv = 0; + struct slm *sip = &slm_info[device]; + + stdma_lock( NULL, NULL ); + + CMDSET_TARG_LUN( slmmsense_cmd, sip->target, sip->lun ); + slmmsense_cmd[5] = abs_flag ? 0x80 : 0; + if (!acsicmd_nodma( slmmsense_cmd, 0 )) { + rv = SLMSTAT_ACSITO; + goto the_end; + } + + if (!acsi_extstatus( &stat, 1 )) { + acsi_end_extstatus(); + rv = SLMSTAT_ACSITO; + goto the_end; + } + + if (!acsi_extstatus( &len, 1 )) { + acsi_end_extstatus(); + rv = SLMSTAT_ACSITO; + goto the_end; + } + buffer[0] = len; + if (!acsi_extstatus( buffer+1, len )) { + acsi_end_extstatus(); + rv = SLMSTAT_ACSITO; + goto the_end; + } + + acsi_end_extstatus(); + rv = stat & 0x1f; + + the_end: + ENABLE_IRQ(); + stdma_release(); + return( rv ); +} + + +#if 0 +/* currently unused */ +static int slm_mode_select( int device, char *buffer, int len, + int default_flag ) + +{ int stat, rv; + struct slm *sip = &slm_info[device]; + + stdma_lock( NULL, NULL ); + + CMDSET_TARG_LUN( slmmselect_cmd, sip->target, sip->lun ); + slmmselect_cmd[5] = default_flag ? 0x80 : 0; + if (!acsicmd_nodma( slmmselect_cmd, 0 )) { + rv = SLMSTAT_ACSITO; + goto the_end; + } + + if (!default_flag) { + unsigned char c = len; + if (!acsi_extcmd( &c, 1 )) { + rv = SLMSTAT_ACSITO; + goto the_end; + } + if (!acsi_extcmd( buffer, len )) { + rv = SLMSTAT_ACSITO; + goto the_end; + } + } + + stat = acsi_getstatus(); + rv = (stat < 0 ? SLMSTAT_ACSITO : stat); + + the_end: + ENABLE_IRQ(); + stdma_release(); + return( rv ); +} +#endif + + +static int slm_get_pagesize( int device, int *w, int *h ) + +{ char buf[256]; + int stat; + + stat = slm_mode_sense( device, buf, 0 ); + ENABLE_IRQ(); + stdma_release(); + + if (stat != SLMSTAT_OK) + return( -EIO ); + + *w = (buf[3] << 8) | buf[4]; + *h = (buf[1] << 8) | buf[2]; + return( 0 ); +} + + +/* ---------------------------------------------------------------------- */ +/* Initialization */ + + +int attach_slm( int target, int lun ) + +{ static int did_register; + int len; + + if (N_SLM_Printers >= MAX_SLM) { + printk( KERN_WARNING "Too much SLMs\n" ); + return( 0 ); + } + + /* do an INQUIRY */ + udelay(100); + CMDSET_TARG_LUN( slminquiry_cmd, target, lun ); + if (!acsicmd_nodma( slminquiry_cmd, 0 )) { + inq_timeout: + printk( KERN_ERR "SLM inquiry command timed out.\n" ); + inq_fail: + acsi_end_extstatus(); + return( 0 ); + } + /* read status and header of return data */ + if (!acsi_extstatus( SLMBuffer, 6 )) + goto inq_timeout; + + if (SLMBuffer[1] != 2) { /* device type == printer? */ + printk( KERN_ERR "SLM inquiry returned device type != printer\n" ); + goto inq_fail; + } + len = SLMBuffer[5]; + + /* read id string */ + if (!acsi_extstatus( SLMBuffer, len )) + goto inq_timeout; + acsi_end_extstatus(); + SLMBuffer[len] = 0; + + if (!did_register) { + did_register = 1; + } + + slm_info[N_SLM_Printers].target = target; + slm_info[N_SLM_Printers].lun = lun; + atomic_set(&slm_info[N_SLM_Printers].wr_ok, 1 ); + atomic_set(&slm_info[N_SLM_Printers].rd_ok, 1 ); + + printk( KERN_INFO " Printer: %s\n", SLMBuffer ); + printk( KERN_INFO "Detected slm%d at id %d lun %d\n", + N_SLM_Printers, target, lun ); + N_SLM_Printers++; + return( 1 ); +} + +int slm_init( void ) + +{ + int i; + if (register_chrdev( ACSI_MAJOR, "slm", &slm_fops )) { + printk( KERN_ERR "Unable to get major %d for ACSI SLM\n", ACSI_MAJOR ); + return -EBUSY; + } + + if (!(SLMBuffer = atari_stram_alloc( SLM_BUFFER_SIZE, "SLM" ))) { + printk( KERN_ERR "Unable to get SLM ST-Ram buffer.\n" ); + unregister_chrdev( ACSI_MAJOR, "slm" ); + return -ENOMEM; + } + BufferP = SLMBuffer; + SLMState = IDLE; + + devfs_mk_dir("slm"); + for (i = 0; i < MAX_SLM; i++) { + devfs_mk_cdev(MKDEV(ACSI_MAJOR, i), + S_IFCHR|S_IRUSR|S_IWUSR, "slm/%d", i); + } + return 0; +} + +#ifdef MODULE + +/* from acsi.c */ +void acsi_attach_SLMs( int (*attach_func)( int, int ) ); + +int init_module(void) +{ + int err; + + if ((err = slm_init())) + return( err ); + /* This calls attach_slm() for every target/lun where acsi.c detected a + * printer */ + acsi_attach_SLMs( attach_slm ); + return( 0 ); +} + +void cleanup_module(void) +{ + int i; + for (i = 0; i < MAX_SLM; i++) + devfs_remove("slm/%d", i); + devfs_remove("slm"); + if (unregister_chrdev( ACSI_MAJOR, "slm" ) != 0) + printk( KERN_ERR "acsi_slm: cleanup_module failed\n"); + atari_stram_free( SLMBuffer ); +} +#endif diff --git a/drivers/block/amiflop.c b/drivers/block/amiflop.c new file mode 100644 index 00000000000..1468e8cf712 --- /dev/null +++ b/drivers/block/amiflop.c @@ -0,0 +1,1850 @@ +/* + * linux/amiga/amiflop.c + * + * Copyright (C) 1993 Greg Harp + * Portions of this driver are based on code contributed by Brad Pepers + * + * revised 28.5.95 by Joerg Dorchain + * - now no bugs(?) any more for both HD & DD + * - added support for 40 Track 5.25" drives, 80-track hopefully behaves + * like 3.5" dd (no way to test - are there any 5.25" drives out there + * that work on an A4000?) + * - wrote formatting routine (maybe dirty, but works) + * + * june/july 1995 added ms-dos support by Joerg Dorchain + * (portions based on messydos.device and various contributors) + * - currently only 9 and 18 sector disks + * + * - fixed a bug with the internal trackbuffer when using multiple + * disks the same time + * - made formatting a bit safer + * - added command line and machine based default for "silent" df0 + * + * december 1995 adapted for 1.2.13pl4 by Joerg Dorchain + * - works but I think it's inefficient. (look in redo_fd_request) + * But the changes were very efficient. (only three and a half lines) + * + * january 1996 added special ioctl for tracking down read/write problems + * - usage ioctl(d, RAW_TRACK, ptr); the raw track buffer (MFM-encoded data + * is copied to area. (area should be large enough since no checking is + * done - 30K is currently sufficient). return the actual size of the + * trackbuffer + * - replaced udelays() by a timer (CIAA timer B) for the waits + * needed for the disk mechanic. + * + * february 1996 fixed error recovery and multiple disk access + * - both got broken the first time I tampered with the driver :-( + * - still not safe, but better than before + * + * revised Marts 3rd, 1996 by Jes Sorensen for use in the 1.3.28 kernel. + * - Minor changes to accept the kdev_t. + * - Replaced some more udelays with ms_delays. Udelay is just a loop, + * and so the delay will be different depending on the given + * processor :-( + * - The driver could use a major cleanup because of the new + * major/minor handling that came with kdev_t. It seems to work for + * the time being, but I can't guarantee that it will stay like + * that when we start using 16 (24?) bit minors. + * + * restructured jan 1997 by Joerg Dorchain + * - Fixed Bug accessing multiple disks + * - some code cleanup + * - added trackbuffer for each drive to speed things up + * - fixed some race conditions (who finds the next may send it to me ;-) + */ + +#include <linux/module.h> + +#include <linux/fd.h> +#include <linux/hdreg.h> +#include <linux/delay.h> +#include <linux/init.h> +#include <linux/amifdreg.h> +#include <linux/amifd.h> +#include <linux/buffer_head.h> +#include <linux/blkdev.h> +#include <linux/elevator.h> + +#include <asm/setup.h> +#include <asm/uaccess.h> +#include <asm/amigahw.h> +#include <asm/amigaints.h> +#include <asm/irq.h> + +#undef DEBUG /* print _LOTS_ of infos */ + +#define RAW_IOCTL +#ifdef RAW_IOCTL +#define IOCTL_RAW_TRACK 0x5254524B /* 'RTRK' */ +#endif + +/* + * Defines + */ + +/* + * Error codes + */ +#define FD_OK 0 /* operation succeeded */ +#define FD_ERROR -1 /* general error (seek, read, write, etc) */ +#define FD_NOUNIT 1 /* unit does not exist */ +#define FD_UNITBUSY 2 /* unit already active */ +#define FD_NOTACTIVE 3 /* unit is not active */ +#define FD_NOTREADY 4 /* unit is not ready (motor not on/no disk) */ + +#define MFM_NOSYNC 1 +#define MFM_HEADER 2 +#define MFM_DATA 3 +#define MFM_TRACK 4 + +/* + * Floppy ID values + */ +#define FD_NODRIVE 0x00000000 /* response when no unit is present */ +#define FD_DD_3 0xffffffff /* double-density 3.5" (880K) drive */ +#define FD_HD_3 0x55555555 /* high-density 3.5" (1760K) drive */ +#define FD_DD_5 0xaaaaaaaa /* double-density 5.25" (440K) drive */ + +static unsigned long int fd_def_df0 = FD_DD_3; /* default for df0 if it doesn't identify */ + +module_param(fd_def_df0, ulong, 0); +MODULE_LICENSE("GPL"); + +static struct request_queue *floppy_queue; +#define QUEUE (floppy_queue) +#define CURRENT elv_next_request(floppy_queue) + +/* + * Macros + */ +#define MOTOR_ON (ciab.prb &= ~DSKMOTOR) +#define MOTOR_OFF (ciab.prb |= DSKMOTOR) +#define SELECT(mask) (ciab.prb &= ~mask) +#define DESELECT(mask) (ciab.prb |= mask) +#define SELMASK(drive) (1 << (3 + (drive & 3))) + +static struct fd_drive_type drive_types[] = { +/* code name tr he rdsz wrsz sm pc1 pc2 sd st st*/ +/* warning: times are now in milliseconds (ms) */ +{ FD_DD_3, "DD 3.5", 80, 2, 14716, 13630, 1, 80,161, 3, 18, 1}, +{ FD_HD_3, "HD 3.5", 80, 2, 28344, 27258, 2, 80,161, 3, 18, 1}, +{ FD_DD_5, "DD 5.25", 40, 2, 14716, 13630, 1, 40, 81, 6, 30, 2}, +{ FD_NODRIVE, "No Drive", 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} +}; +static int num_dr_types = sizeof(drive_types) / sizeof(drive_types[0]); + +static int amiga_read(int), dos_read(int); +static void amiga_write(int), dos_write(int); +static struct fd_data_type data_types[] = { + { "Amiga", 11 , amiga_read, amiga_write}, + { "MS-Dos", 9, dos_read, dos_write} +}; + +/* current info on each unit */ +static struct amiga_floppy_struct unit[FD_MAX_UNITS]; + +static struct timer_list flush_track_timer[FD_MAX_UNITS]; +static struct timer_list post_write_timer; +static struct timer_list motor_on_timer; +static struct timer_list motor_off_timer[FD_MAX_UNITS]; +static int on_attempts; + +/* Synchronization of FDC access */ +/* request loop (trackbuffer) */ +static volatile int fdc_busy = -1; +static volatile int fdc_nested; +static DECLARE_WAIT_QUEUE_HEAD(fdc_wait); + +static DECLARE_WAIT_QUEUE_HEAD(motor_wait); + +static volatile int selected = -1; /* currently selected drive */ + +static int writepending; +static int writefromint; +static char *raw_buf; + +static DEFINE_SPINLOCK(amiflop_lock); + +#define RAW_BUF_SIZE 30000 /* size of raw disk data */ + +/* + * These are global variables, as that's the easiest way to give + * information to interrupts. They are the data used for the current + * request. + */ +static volatile char block_flag; +static DECLARE_WAIT_QUEUE_HEAD(wait_fd_block); + +/* MS-Dos MFM Coding tables (should go quick and easy) */ +static unsigned char mfmencode[16]={ + 0x2a, 0x29, 0x24, 0x25, 0x12, 0x11, 0x14, 0x15, + 0x4a, 0x49, 0x44, 0x45, 0x52, 0x51, 0x54, 0x55 +}; +static unsigned char mfmdecode[128]; + +/* floppy internal millisecond timer stuff */ +static volatile int ms_busy = -1; +static DECLARE_WAIT_QUEUE_HEAD(ms_wait); +#define MS_TICKS ((amiga_eclock+50)/1000) + +/* + * Note that MAX_ERRORS=X doesn't imply that we retry every bad read + * max X times - some types of errors increase the errorcount by 2 or + * even 3, so we might actually retry only X/2 times before giving up. + */ +#define MAX_ERRORS 12 + +/* Prevent "aliased" accesses. */ +static int fd_ref[4] = { 0,0,0,0 }; +static int fd_device[4] = { 0, 0, 0, 0 }; + +/* + * Here come the actual hardware access and helper functions. + * They are not reentrant and single threaded because all drives + * share the same hardware and the same trackbuffer. + */ + +/* Milliseconds timer */ + +static irqreturn_t ms_isr(int irq, void *dummy, struct pt_regs *fp) +{ + ms_busy = -1; + wake_up(&ms_wait); + return IRQ_HANDLED; +} + +/* all waits are queued up + A more generic routine would do a schedule a la timer.device */ +static void ms_delay(int ms) +{ + unsigned long flags; + int ticks; + if (ms > 0) { + local_irq_save(flags); + while (ms_busy == 0) + sleep_on(&ms_wait); + ms_busy = 0; + local_irq_restore(flags); + ticks = MS_TICKS*ms-1; + ciaa.tblo=ticks%256; + ciaa.tbhi=ticks/256; + ciaa.crb=0x19; /*count eclock, force load, one-shoot, start */ + sleep_on(&ms_wait); + } +} + +/* Hardware semaphore */ + +/* returns true when we would get the semaphore */ +static inline int try_fdc(int drive) +{ + drive &= 3; + return ((fdc_busy < 0) || (fdc_busy == drive)); +} + +static void get_fdc(int drive) +{ + unsigned long flags; + + drive &= 3; +#ifdef DEBUG + printk("get_fdc: drive %d fdc_busy %d fdc_nested %d\n",drive,fdc_busy,fdc_nested); +#endif + local_irq_save(flags); + while (!try_fdc(drive)) + sleep_on(&fdc_wait); + fdc_busy = drive; + fdc_nested++; + local_irq_restore(flags); +} + +static inline void rel_fdc(void) +{ +#ifdef DEBUG + if (fdc_nested == 0) + printk("fd: unmatched rel_fdc\n"); + printk("rel_fdc: fdc_busy %d fdc_nested %d\n",fdc_busy,fdc_nested); +#endif + fdc_nested--; + if (fdc_nested == 0) { + fdc_busy = -1; + wake_up(&fdc_wait); + } +} + +static void fd_select (int drive) +{ + unsigned char prb = ~0; + + drive&=3; +#ifdef DEBUG + printk("selecting %d\n",drive); +#endif + if (drive == selected) + return; + get_fdc(drive); + selected = drive; + + if (unit[drive].track % 2 != 0) + prb &= ~DSKSIDE; + if (unit[drive].motor == 1) + prb &= ~DSKMOTOR; + ciab.prb |= (SELMASK(0)|SELMASK(1)|SELMASK(2)|SELMASK(3)); + ciab.prb = prb; + prb &= ~SELMASK(drive); + ciab.prb = prb; + rel_fdc(); +} + +static void fd_deselect (int drive) +{ + unsigned char prb; + unsigned long flags; + + drive&=3; +#ifdef DEBUG + printk("deselecting %d\n",drive); +#endif + if (drive != selected) { + printk(KERN_WARNING "Deselecting drive %d while %d was selected!\n",drive,selected); + return; + } + + get_fdc(drive); + local_irq_save(flags); + + selected = -1; + + prb = ciab.prb; + prb |= (SELMASK(0)|SELMASK(1)|SELMASK(2)|SELMASK(3)); + ciab.prb = prb; + + local_irq_restore (flags); + rel_fdc(); + +} + +static void motor_on_callback(unsigned long nr) +{ + if (!(ciaa.pra & DSKRDY) || --on_attempts == 0) { + wake_up (&motor_wait); + } else { + motor_on_timer.expires = jiffies + HZ/10; + add_timer(&motor_on_timer); + } +} + +static int fd_motor_on(int nr) +{ + nr &= 3; + + del_timer(motor_off_timer + nr); + + if (!unit[nr].motor) { + unit[nr].motor = 1; + fd_select(nr); + + motor_on_timer.data = nr; + mod_timer(&motor_on_timer, jiffies + HZ/2); + + on_attempts = 10; + sleep_on (&motor_wait); + fd_deselect(nr); + } + + if (on_attempts == 0) { + on_attempts = -1; +#if 0 + printk (KERN_ERR "motor_on failed, turning motor off\n"); + fd_motor_off (nr); + return 0; +#else + printk (KERN_WARNING "DSKRDY not set after 1.5 seconds - assuming drive is spinning notwithstanding\n"); +#endif + } + + return 1; +} + +static void fd_motor_off(unsigned long drive) +{ + long calledfromint; +#ifdef MODULE + long decusecount; + + decusecount = drive & 0x40000000; +#endif + calledfromint = drive & 0x80000000; + drive&=3; + if (calledfromint && !try_fdc(drive)) { + /* We would be blocked in an interrupt, so try again later */ + motor_off_timer[drive].expires = jiffies + 1; + add_timer(motor_off_timer + drive); + return; + } + unit[drive].motor = 0; + fd_select(drive); + udelay (1); + fd_deselect(drive); +} + +static void floppy_off (unsigned int nr) +{ + int drive; + + drive = nr & 3; + /* called this way it is always from interrupt */ + motor_off_timer[drive].data = nr | 0x80000000; + mod_timer(motor_off_timer + drive, jiffies + 3*HZ); +} + +static int fd_calibrate(int drive) +{ + unsigned char prb; + int n; + + drive &= 3; + get_fdc(drive); + if (!fd_motor_on (drive)) + return 0; + fd_select (drive); + prb = ciab.prb; + prb |= DSKSIDE; + prb &= ~DSKDIREC; + ciab.prb = prb; + for (n = unit[drive].type->tracks/2; n != 0; --n) { + if (ciaa.pra & DSKTRACK0) + break; + prb &= ~DSKSTEP; + ciab.prb = prb; + prb |= DSKSTEP; + udelay (2); + ciab.prb = prb; + ms_delay(unit[drive].type->step_delay); + } + ms_delay (unit[drive].type->settle_time); + prb |= DSKDIREC; + n = unit[drive].type->tracks + 20; + for (;;) { + prb &= ~DSKSTEP; + ciab.prb = prb; + prb |= DSKSTEP; + udelay (2); + ciab.prb = prb; + ms_delay(unit[drive].type->step_delay + 1); + if ((ciaa.pra & DSKTRACK0) == 0) + break; + if (--n == 0) { + printk (KERN_ERR "fd%d: calibrate failed, turning motor off\n", drive); + fd_motor_off (drive); + unit[drive].track = -1; + rel_fdc(); + return 0; + } + } + unit[drive].track = 0; + ms_delay(unit[drive].type->settle_time); + + rel_fdc(); + fd_deselect(drive); + return 1; +} + +static int fd_seek(int drive, int track) +{ + unsigned char prb; + int cnt; + +#ifdef DEBUG + printk("seeking drive %d to track %d\n",drive,track); +#endif + drive &= 3; + get_fdc(drive); + if (unit[drive].track == track) { + rel_fdc(); + return 1; + } + if (!fd_motor_on(drive)) { + rel_fdc(); + return 0; + } + if (unit[drive].track < 0 && !fd_calibrate(drive)) { + rel_fdc(); + return 0; + } + + fd_select (drive); + cnt = unit[drive].track/2 - track/2; + prb = ciab.prb; + prb |= DSKSIDE | DSKDIREC; + if (track % 2 != 0) + prb &= ~DSKSIDE; + if (cnt < 0) { + cnt = - cnt; + prb &= ~DSKDIREC; + } + ciab.prb = prb; + if (track % 2 != unit[drive].track % 2) + ms_delay (unit[drive].type->side_time); + unit[drive].track = track; + if (cnt == 0) { + rel_fdc(); + fd_deselect(drive); + return 1; + } + do { + prb &= ~DSKSTEP; + ciab.prb = prb; + prb |= DSKSTEP; + udelay (1); + ciab.prb = prb; + ms_delay (unit[drive].type->step_delay); + } while (--cnt != 0); + ms_delay (unit[drive].type->settle_time); + + rel_fdc(); + fd_deselect(drive); + return 1; +} + +static unsigned long fd_get_drive_id(int drive) +{ + int i; + ulong id = 0; + + drive&=3; + get_fdc(drive); + /* set up for ID */ + MOTOR_ON; + udelay(2); + SELECT(SELMASK(drive)); + udelay(2); + DESELECT(SELMASK(drive)); + udelay(2); + MOTOR_OFF; + udelay(2); + SELECT(SELMASK(drive)); + udelay(2); + DESELECT(SELMASK(drive)); + udelay(2); + + /* loop and read disk ID */ + for (i=0; i<32; i++) { + SELECT(SELMASK(drive)); + udelay(2); + + /* read and store value of DSKRDY */ + id <<= 1; + id |= (ciaa.pra & DSKRDY) ? 0 : 1; /* cia regs are low-active! */ + + DESELECT(SELMASK(drive)); + } + + rel_fdc(); + + /* + * RB: At least A500/A2000's df0: don't identify themselves. + * As every (real) Amiga has at least a 3.5" DD drive as df0: + * we default to that if df0: doesn't identify as a certain + * type. + */ + if(drive == 0 && id == FD_NODRIVE) + { + id = fd_def_df0; + printk(KERN_NOTICE "fd: drive 0 didn't identify, setting default %08lx\n", (ulong)fd_def_df0); + } + /* return the ID value */ + return (id); +} + +static irqreturn_t fd_block_done(int irq, void *dummy, struct pt_regs *fp) +{ + if (block_flag) + custom.dsklen = 0x4000; + + if (block_flag == 2) { /* writing */ + writepending = 2; + post_write_timer.expires = jiffies + 1; /* at least 2 ms */ + post_write_timer.data = selected; + add_timer(&post_write_timer); + } + else { /* reading */ + block_flag = 0; + wake_up (&wait_fd_block); + } + return IRQ_HANDLED; +} + +static void raw_read(int drive) +{ + drive&=3; + get_fdc(drive); + while (block_flag) + sleep_on(&wait_fd_block); + fd_select(drive); + /* setup adkcon bits correctly */ + custom.adkcon = ADK_MSBSYNC; + custom.adkcon = ADK_SETCLR|ADK_WORDSYNC|ADK_FAST; + + custom.dsksync = MFM_SYNC; + + custom.dsklen = 0; + custom.dskptr = (u_char *)ZTWO_PADDR((u_char *)raw_buf); + custom.dsklen = unit[drive].type->read_size/sizeof(short) | DSKLEN_DMAEN; + custom.dsklen = unit[drive].type->read_size/sizeof(short) | DSKLEN_DMAEN; + + block_flag = 1; + + while (block_flag) + sleep_on (&wait_fd_block); + + custom.dsklen = 0; + fd_deselect(drive); + rel_fdc(); +} + +static int raw_write(int drive) +{ + ushort adk; + + drive&=3; + get_fdc(drive); /* corresponds to rel_fdc() in post_write() */ + if ((ciaa.pra & DSKPROT) == 0) { + rel_fdc(); + return 0; + } + while (block_flag) + sleep_on(&wait_fd_block); + fd_select(drive); + /* clear adkcon bits */ + custom.adkcon = ADK_PRECOMP1|ADK_PRECOMP0|ADK_WORDSYNC|ADK_MSBSYNC; + /* set appropriate adkcon bits */ + adk = ADK_SETCLR|ADK_FAST; + if ((ulong)unit[drive].track >= unit[drive].type->precomp2) + adk |= ADK_PRECOMP1; + else if ((ulong)unit[drive].track >= unit[drive].type->precomp1) + adk |= ADK_PRECOMP0; + custom.adkcon = adk; + + custom.dsklen = DSKLEN_WRITE; + custom.dskptr = (u_char *)ZTWO_PADDR((u_char *)raw_buf); + custom.dsklen = unit[drive].type->write_size/sizeof(short) | DSKLEN_DMAEN|DSKLEN_WRITE; + custom.dsklen = unit[drive].type->write_size/sizeof(short) | DSKLEN_DMAEN|DSKLEN_WRITE; + + block_flag = 2; + return 1; +} + +/* + * to be called at least 2ms after the write has finished but before any + * other access to the hardware. + */ +static void post_write (unsigned long drive) +{ +#ifdef DEBUG + printk("post_write for drive %ld\n",drive); +#endif + drive &= 3; + custom.dsklen = 0; + block_flag = 0; + writepending = 0; + writefromint = 0; + unit[drive].dirty = 0; + wake_up(&wait_fd_block); + fd_deselect(drive); + rel_fdc(); /* corresponds to get_fdc() in raw_write */ +} + + +/* + * The following functions are to convert the block contents into raw data + * written to disk and vice versa. + * (Add other formats here ;-)) + */ + +static unsigned long scan_sync(unsigned long raw, unsigned long end) +{ + ushort *ptr = (ushort *)raw, *endp = (ushort *)end; + + while (ptr < endp && *ptr++ != 0x4489) + ; + if (ptr < endp) { + while (*ptr == 0x4489 && ptr < endp) + ptr++; + return (ulong)ptr; + } + return 0; +} + +static inline unsigned long checksum(unsigned long *addr, int len) +{ + unsigned long csum = 0; + + len /= sizeof(*addr); + while (len-- > 0) + csum ^= *addr++; + csum = ((csum>>1) & 0x55555555) ^ (csum & 0x55555555); + + return csum; +} + +static unsigned long decode (unsigned long *data, unsigned long *raw, + int len) +{ + ulong *odd, *even; + + /* convert length from bytes to longwords */ + len >>= 2; + odd = raw; + even = odd + len; + + /* prepare return pointer */ + raw += len * 2; + + do { + *data++ = ((*odd++ & 0x55555555) << 1) | (*even++ & 0x55555555); + } while (--len != 0); + + return (ulong)raw; +} + +struct header { + unsigned char magic; + unsigned char track; + unsigned char sect; + unsigned char ord; + unsigned char labels[16]; + unsigned long hdrchk; + unsigned long datachk; +}; + +static int amiga_read(int drive) +{ + unsigned long raw; + unsigned long end; + int scnt; + unsigned long csum; + struct header hdr; + + drive&=3; + raw = (long) raw_buf; + end = raw + unit[drive].type->read_size; + + for (scnt = 0;scnt < unit[drive].dtype->sects * unit[drive].type->sect_mult; scnt++) { + if (!(raw = scan_sync(raw, end))) { + printk (KERN_INFO "can't find sync for sector %d\n", scnt); + return MFM_NOSYNC; + } + + raw = decode ((ulong *)&hdr.magic, (ulong *)raw, 4); + raw = decode ((ulong *)&hdr.labels, (ulong *)raw, 16); + raw = decode ((ulong *)&hdr.hdrchk, (ulong *)raw, 4); + raw = decode ((ulong *)&hdr.datachk, (ulong *)raw, 4); + csum = checksum((ulong *)&hdr, + (char *)&hdr.hdrchk-(char *)&hdr); + +#ifdef DEBUG + printk ("(%x,%d,%d,%d) (%lx,%lx,%lx,%lx) %lx %lx\n", + hdr.magic, hdr.track, hdr.sect, hdr.ord, + *(ulong *)&hdr.labels[0], *(ulong *)&hdr.labels[4], + *(ulong *)&hdr.labels[8], *(ulong *)&hdr.labels[12], + hdr.hdrchk, hdr.datachk); +#endif + + if (hdr.hdrchk != csum) { + printk(KERN_INFO "MFM_HEADER: %08lx,%08lx\n", hdr.hdrchk, csum); + return MFM_HEADER; + } + + /* verify track */ + if (hdr.track != unit[drive].track) { + printk(KERN_INFO "MFM_TRACK: %d, %d\n", hdr.track, unit[drive].track); + return MFM_TRACK; + } + + raw = decode ((ulong *)(unit[drive].trackbuf + hdr.sect*512), + (ulong *)raw, 512); + csum = checksum((ulong *)(unit[drive].trackbuf + hdr.sect*512), 512); + + if (hdr.datachk != csum) { + printk(KERN_INFO "MFM_DATA: (%x:%d:%d:%d) sc=%d %lx, %lx\n", + hdr.magic, hdr.track, hdr.sect, hdr.ord, scnt, + hdr.datachk, csum); + printk (KERN_INFO "data=(%lx,%lx,%lx,%lx)\n", + ((ulong *)(unit[drive].trackbuf+hdr.sect*512))[0], + ((ulong *)(unit[drive].trackbuf+hdr.sect*512))[1], + ((ulong *)(unit[drive].trackbuf+hdr.sect*512))[2], + ((ulong *)(unit[drive].trackbuf+hdr.sect*512))[3]); + return MFM_DATA; + } + } + + return 0; +} + +static void encode(unsigned long data, unsigned long *dest) +{ + unsigned long data2; + + data &= 0x55555555; + data2 = data ^ 0x55555555; + data |= ((data2 >> 1) | 0x80000000) & (data2 << 1); + + if (*(dest - 1) & 0x00000001) + data &= 0x7FFFFFFF; + + *dest = data; +} + +static void encode_block(unsigned long *dest, unsigned long *src, int len) +{ + int cnt, to_cnt = 0; + unsigned long data; + + /* odd bits */ + for (cnt = 0; cnt < len / 4; cnt++) { + data = src[cnt] >> 1; + encode(data, dest + to_cnt++); + } + + /* even bits */ + for (cnt = 0; cnt < len / 4; cnt++) { + data = src[cnt]; + encode(data, dest + to_cnt++); + } +} + +static unsigned long *putsec(int disk, unsigned long *raw, int cnt) +{ + struct header hdr; + int i; + + disk&=3; + *raw = (raw[-1]&1) ? 0x2AAAAAAA : 0xAAAAAAAA; + raw++; + *raw++ = 0x44894489; + + hdr.magic = 0xFF; + hdr.track = unit[disk].track; + hdr.sect = cnt; + hdr.ord = unit[disk].dtype->sects * unit[disk].type->sect_mult - cnt; + for (i = 0; i < 16; i++) + hdr.labels[i] = 0; + hdr.hdrchk = checksum((ulong *)&hdr, + (char *)&hdr.hdrchk-(char *)&hdr); + hdr.datachk = checksum((ulong *)(unit[disk].trackbuf+cnt*512), 512); + + encode_block(raw, (ulong *)&hdr.magic, 4); + raw += 2; + encode_block(raw, (ulong *)&hdr.labels, 16); + raw += 8; + encode_block(raw, (ulong *)&hdr.hdrchk, 4); + raw += 2; + encode_block(raw, (ulong *)&hdr.datachk, 4); + raw += 2; + encode_block(raw, (ulong *)(unit[disk].trackbuf+cnt*512), 512); + raw += 256; + + return raw; +} + +static void amiga_write(int disk) +{ + unsigned int cnt; + unsigned long *ptr = (unsigned long *)raw_buf; + + disk&=3; + /* gap space */ + for (cnt = 0; cnt < 415 * unit[disk].type->sect_mult; cnt++) + *ptr++ = 0xaaaaaaaa; + + /* sectors */ + for (cnt = 0; cnt < unit[disk].dtype->sects * unit[disk].type->sect_mult; cnt++) + ptr = putsec (disk, ptr, cnt); + *(ushort *)ptr = (ptr[-1]&1) ? 0x2AA8 : 0xAAA8; +} + + +struct dos_header { + unsigned char track, /* 0-80 */ + side, /* 0-1 */ + sec, /* 0-...*/ + len_desc;/* 2 */ + unsigned short crc; /* on 68000 we got an alignment problem, + but this compiler solves it by adding silently + adding a pad byte so data won't fit + and this took about 3h to discover.... */ + unsigned char gap1[22]; /* for longword-alignedness (0x4e) */ +}; + +/* crc routines are borrowed from the messydos-handler */ + +/* excerpt from the messydos-device +; The CRC is computed not only over the actual data, but including +; the SYNC mark (3 * $a1) and the 'ID/DATA - Address Mark' ($fe/$fb). +; As we don't read or encode these fields into our buffers, we have to +; preload the registers containing the CRC with the values they would have +; after stepping over these fields. +; +; How CRCs "really" work: +; +; First, you should regard a bitstring as a series of coefficients of +; polynomials. We calculate with these polynomials in modulo-2 +; arithmetic, in which both add and subtract are done the same as +; exclusive-or. Now, we modify our data (a very long polynomial) in +; such a way that it becomes divisible by the CCITT-standard 16-bit +; 16 12 5 +; polynomial: x + x + x + 1, represented by $11021. The easiest +; way to do this would be to multiply (using proper arithmetic) our +; datablock with $11021. So we have: +; data * $11021 = +; data * ($10000 + $1021) = +; data * $10000 + data * $1021 +; The left part of this is simple: Just add two 0 bytes. But then +; the right part (data $1021) remains difficult and even could have +; a carry into the left part. The solution is to use a modified +; multiplication, which has a result that is not correct, but with +; a difference of any multiple of $11021. We then only need to keep +; the 16 least significant bits of the result. +; +; The following algorithm does this for us: +; +; unsigned char *data, c, crclo, crchi; +; while (not done) { +; c = *data++ + crchi; +; crchi = (@ c) >> 8 + crclo; +; crclo = @ c; +; } +; +; Remember, + is done with EOR, the @ operator is in two tables (high +; and low byte separately), which is calculated as +; +; $1021 * (c & $F0) +; xor $1021 * (c & $0F) +; xor $1021 * (c >> 4) (* is regular multiplication) +; +; +; Anyway, the end result is the same as the remainder of the division of +; the data by $11021. I am afraid I need to study theory a bit more... + + +my only works was to code this from manx to C.... + +*/ + +static ushort dos_crc(void * data_a3, int data_d0, int data_d1, int data_d3) +{ + static unsigned char CRCTable1[] = { + 0x00,0x10,0x20,0x30,0x40,0x50,0x60,0x70,0x81,0x91,0xa1,0xb1,0xc1,0xd1,0xe1,0xf1, + 0x12,0x02,0x32,0x22,0x52,0x42,0x72,0x62,0x93,0x83,0xb3,0xa3,0xd3,0xc3,0xf3,0xe3, + 0x24,0x34,0x04,0x14,0x64,0x74,0x44,0x54,0xa5,0xb5,0x85,0x95,0xe5,0xf5,0xc5,0xd5, + 0x36,0x26,0x16,0x06,0x76,0x66,0x56,0x46,0xb7,0xa7,0x97,0x87,0xf7,0xe7,0xd7,0xc7, + 0x48,0x58,0x68,0x78,0x08,0x18,0x28,0x38,0xc9,0xd9,0xe9,0xf9,0x89,0x99,0xa9,0xb9, + 0x5a,0x4a,0x7a,0x6a,0x1a,0x0a,0x3a,0x2a,0xdb,0xcb,0xfb,0xeb,0x9b,0x8b,0xbb,0xab, + 0x6c,0x7c,0x4c,0x5c,0x2c,0x3c,0x0c,0x1c,0xed,0xfd,0xcd,0xdd,0xad,0xbd,0x8d,0x9d, + 0x7e,0x6e,0x5e,0x4e,0x3e,0x2e,0x1e,0x0e,0xff,0xef,0xdf,0xcf,0xbf,0xaf,0x9f,0x8f, + 0x91,0x81,0xb1,0xa1,0xd1,0xc1,0xf1,0xe1,0x10,0x00,0x30,0x20,0x50,0x40,0x70,0x60, + 0x83,0x93,0xa3,0xb3,0xc3,0xd3,0xe3,0xf3,0x02,0x12,0x22,0x32,0x42,0x52,0x62,0x72, + 0xb5,0xa5,0x95,0x85,0xf5,0xe5,0xd5,0xc5,0x34,0x24,0x14,0x04,0x74,0x64,0x54,0x44, + 0xa7,0xb7,0x87,0x97,0xe7,0xf7,0xc7,0xd7,0x26,0x36,0x06,0x16,0x66,0x76,0x46,0x56, + 0xd9,0xc9,0xf9,0xe9,0x99,0x89,0xb9,0xa9,0x58,0x48,0x78,0x68,0x18,0x08,0x38,0x28, + 0xcb,0xdb,0xeb,0xfb,0x8b,0x9b,0xab,0xbb,0x4a,0x5a,0x6a,0x7a,0x0a,0x1a,0x2a,0x3a, + 0xfd,0xed,0xdd,0xcd,0xbd,0xad,0x9d,0x8d,0x7c,0x6c,0x5c,0x4c,0x3c,0x2c,0x1c,0x0c, + 0xef,0xff,0xcf,0xdf,0xaf,0xbf,0x8f,0x9f,0x6e,0x7e,0x4e,0x5e,0x2e,0x3e,0x0e,0x1e + }; + + static unsigned char CRCTable2[] = { + 0x00,0x21,0x42,0x63,0x84,0xa5,0xc6,0xe7,0x08,0x29,0x4a,0x6b,0x8c,0xad,0xce,0xef, + 0x31,0x10,0x73,0x52,0xb5,0x94,0xf7,0xd6,0x39,0x18,0x7b,0x5a,0xbd,0x9c,0xff,0xde, + 0x62,0x43,0x20,0x01,0xe6,0xc7,0xa4,0x85,0x6a,0x4b,0x28,0x09,0xee,0xcf,0xac,0x8d, + 0x53,0x72,0x11,0x30,0xd7,0xf6,0x95,0xb4,0x5b,0x7a,0x19,0x38,0xdf,0xfe,0x9d,0xbc, + 0xc4,0xe5,0x86,0xa7,0x40,0x61,0x02,0x23,0xcc,0xed,0x8e,0xaf,0x48,0x69,0x0a,0x2b, + 0xf5,0xd4,0xb7,0x96,0x71,0x50,0x33,0x12,0xfd,0xdc,0xbf,0x9e,0x79,0x58,0x3b,0x1a, + 0xa6,0x87,0xe4,0xc5,0x22,0x03,0x60,0x41,0xae,0x8f,0xec,0xcd,0x2a,0x0b,0x68,0x49, + 0x97,0xb6,0xd5,0xf4,0x13,0x32,0x51,0x70,0x9f,0xbe,0xdd,0xfc,0x1b,0x3a,0x59,0x78, + 0x88,0xa9,0xca,0xeb,0x0c,0x2d,0x4e,0x6f,0x80,0xa1,0xc2,0xe3,0x04,0x25,0x46,0x67, + 0xb9,0x98,0xfb,0xda,0x3d,0x1c,0x7f,0x5e,0xb1,0x90,0xf3,0xd2,0x35,0x14,0x77,0x56, + 0xea,0xcb,0xa8,0x89,0x6e,0x4f,0x2c,0x0d,0xe2,0xc3,0xa0,0x81,0x66,0x47,0x24,0x05, + 0xdb,0xfa,0x99,0xb8,0x5f,0x7e,0x1d,0x3c,0xd3,0xf2,0x91,0xb0,0x57,0x76,0x15,0x34, + 0x4c,0x6d,0x0e,0x2f,0xc8,0xe9,0x8a,0xab,0x44,0x65,0x06,0x27,0xc0,0xe1,0x82,0xa3, + 0x7d,0x5c,0x3f,0x1e,0xf9,0xd8,0xbb,0x9a,0x75,0x54,0x37,0x16,0xf1,0xd0,0xb3,0x92, + 0x2e,0x0f,0x6c,0x4d,0xaa,0x8b,0xe8,0xc9,0x26,0x07,0x64,0x45,0xa2,0x83,0xe0,0xc1, + 0x1f,0x3e,0x5d,0x7c,0x9b,0xba,0xd9,0xf8,0x17,0x36,0x55,0x74,0x93,0xb2,0xd1,0xf0 + }; + +/* look at the asm-code - what looks in C a bit strange is almost as good as handmade */ + register int i; + register unsigned char *CRCT1, *CRCT2, *data, c, crch, crcl; + + CRCT1=CRCTable1; + CRCT2=CRCTable2; + data=data_a3; + crcl=data_d1; + crch=data_d0; + for (i=data_d3; i>=0; i--) { + c = (*data++) ^ crch; + crch = CRCT1[c] ^ crcl; + crcl = CRCT2[c]; + } + return (crch<<8)|crcl; +} + +static inline ushort dos_hdr_crc (struct dos_header *hdr) +{ + return dos_crc(&(hdr->track), 0xb2, 0x30, 3); /* precomputed magic */ +} + +static inline ushort dos_data_crc(unsigned char *data) +{ + return dos_crc(data, 0xe2, 0x95 ,511); /* precomputed magic */ +} + +static inline unsigned char dos_decode_byte(ushort word) +{ + register ushort w2; + register unsigned char byte; + register unsigned char *dec = mfmdecode; + + w2=word; + w2>>=8; + w2&=127; + byte = dec[w2]; + byte <<= 4; + w2 = word & 127; + byte |= dec[w2]; + return byte; +} + +static unsigned long dos_decode(unsigned char *data, unsigned short *raw, int len) +{ + int i; + + for (i = 0; i < len; i++) + *data++=dos_decode_byte(*raw++); + return ((ulong)raw); +} + +#ifdef DEBUG +static void dbg(unsigned long ptr) +{ + printk("raw data @%08lx: %08lx, %08lx ,%08lx, %08lx\n", ptr, + ((ulong *)ptr)[0], ((ulong *)ptr)[1], + ((ulong *)ptr)[2], ((ulong *)ptr)[3]); +} +#endif + +static int dos_read(int drive) +{ + unsigned long end; + unsigned long raw; + int scnt; + unsigned short crc,data_crc[2]; + struct dos_header hdr; + + drive&=3; + raw = (long) raw_buf; + end = raw + unit[drive].type->read_size; + + for (scnt=0; scnt < unit[drive].dtype->sects * unit[drive].type->sect_mult; scnt++) { + do { /* search for the right sync of each sec-hdr */ + if (!(raw = scan_sync (raw, end))) { + printk(KERN_INFO "dos_read: no hdr sync on " + "track %d, unit %d for sector %d\n", + unit[drive].track,drive,scnt); + return MFM_NOSYNC; + } +#ifdef DEBUG + dbg(raw); +#endif + } while (*((ushort *)raw)!=0x5554); /* loop usually only once done */ + raw+=2; /* skip over headermark */ + raw = dos_decode((unsigned char *)&hdr,(ushort *) raw,8); + crc = dos_hdr_crc(&hdr); + +#ifdef DEBUG + printk("(%3d,%d,%2d,%d) %x\n", hdr.track, hdr.side, + hdr.sec, hdr.len_desc, hdr.crc); +#endif + + if (crc != hdr.crc) { + printk(KERN_INFO "dos_read: MFM_HEADER %04x,%04x\n", + hdr.crc, crc); + return MFM_HEADER; + } + if (hdr.track != unit[drive].track/unit[drive].type->heads) { + printk(KERN_INFO "dos_read: MFM_TRACK %d, %d\n", + hdr.track, + unit[drive].track/unit[drive].type->heads); + return MFM_TRACK; + } + + if (hdr.side != unit[drive].track%unit[drive].type->heads) { + printk(KERN_INFO "dos_read: MFM_SIDE %d, %d\n", + hdr.side, + unit[drive].track%unit[drive].type->heads); + return MFM_TRACK; + } + + if (hdr.len_desc != 2) { + printk(KERN_INFO "dos_read: unknown sector len " + "descriptor %d\n", hdr.len_desc); + return MFM_DATA; + } +#ifdef DEBUG + printk("hdr accepted\n"); +#endif + if (!(raw = scan_sync (raw, end))) { + printk(KERN_INFO "dos_read: no data sync on track " + "%d, unit %d for sector%d, disk sector %d\n", + unit[drive].track, drive, scnt, hdr.sec); + return MFM_NOSYNC; + } +#ifdef DEBUG + dbg(raw); +#endif + + if (*((ushort *)raw)!=0x5545) { + printk(KERN_INFO "dos_read: no data mark after " + "sync (%d,%d,%d,%d) sc=%d\n", + hdr.track,hdr.side,hdr.sec,hdr.len_desc,scnt); + return MFM_NOSYNC; + } + + raw+=2; /* skip data mark (included in checksum) */ + raw = dos_decode((unsigned char *)(unit[drive].trackbuf + (hdr.sec - 1) * 512), (ushort *) raw, 512); + raw = dos_decode((unsigned char *)data_crc,(ushort *) raw,4); + crc = dos_data_crc(unit[drive].trackbuf + (hdr.sec - 1) * 512); + + if (crc != data_crc[0]) { + printk(KERN_INFO "dos_read: MFM_DATA (%d,%d,%d,%d) " + "sc=%d, %x %x\n", hdr.track, hdr.side, + hdr.sec, hdr.len_desc, scnt,data_crc[0], crc); + printk(KERN_INFO "data=(%lx,%lx,%lx,%lx,...)\n", + ((ulong *)(unit[drive].trackbuf+(hdr.sec-1)*512))[0], + ((ulong *)(unit[drive].trackbuf+(hdr.sec-1)*512))[1], + ((ulong *)(unit[drive].trackbuf+(hdr.sec-1)*512))[2], + ((ulong *)(unit[drive].trackbuf+(hdr.sec-1)*512))[3]); + return MFM_DATA; + } + } + return 0; +} + +static inline ushort dos_encode_byte(unsigned char byte) +{ + register unsigned char *enc, b2, b1; + register ushort word; + + enc=mfmencode; + b1=byte; + b2=b1>>4; + b1&=15; + word=enc[b2] <<8 | enc [b1]; + return (word|((word&(256|64)) ? 0: 128)); +} + +static void dos_encode_block(ushort *dest, unsigned char *src, int len) +{ + int i; + + for (i = 0; i < len; i++) { + *dest=dos_encode_byte(*src++); + *dest|=((dest[-1]&1)||(*dest&0x4000))? 0: 0x8000; + dest++; + } +} + +static unsigned long *ms_putsec(int drive, unsigned long *raw, int cnt) +{ + static struct dos_header hdr={0,0,0,2,0, + {78,78,78,78,78,78,78,78,78,78,78,78,78,78,78,78,78,78,78,78,78,78}}; + int i; + static ushort crc[2]={0,0x4e4e}; + + drive&=3; +/* id gap 1 */ +/* the MFM word before is always 9254 */ + for(i=0;i<6;i++) + *raw++=0xaaaaaaaa; +/* 3 sync + 1 headermark */ + *raw++=0x44894489; + *raw++=0x44895554; + +/* fill in the variable parts of the header */ + hdr.track=unit[drive].track/unit[drive].type->heads; + hdr.side=unit[drive].track%unit[drive].type->heads; + hdr.sec=cnt+1; + hdr.crc=dos_hdr_crc(&hdr); + +/* header (without "magic") and id gap 2*/ + dos_encode_block((ushort *)raw,(unsigned char *) &hdr.track,28); + raw+=14; + +/*id gap 3 */ + for(i=0;i<6;i++) + *raw++=0xaaaaaaaa; + +/* 3 syncs and 1 datamark */ + *raw++=0x44894489; + *raw++=0x44895545; + +/* data */ + dos_encode_block((ushort *)raw, + (unsigned char *)unit[drive].trackbuf+cnt*512,512); + raw+=256; + +/*data crc + jd's special gap (long words :-/) */ + crc[0]=dos_data_crc(unit[drive].trackbuf+cnt*512); + dos_encode_block((ushort *) raw,(unsigned char *)crc,4); + raw+=2; + +/* data gap */ + for(i=0;i<38;i++) + *raw++=0x92549254; + + return raw; /* wrote 652 MFM words */ +} + +static void dos_write(int disk) +{ + int cnt; + unsigned long raw = (unsigned long) raw_buf; + unsigned long *ptr=(unsigned long *)raw; + + disk&=3; +/* really gap4 + indexgap , but we write it first and round it up */ + for (cnt=0;cnt<425;cnt++) + *ptr++=0x92549254; + +/* the following is just guessed */ + if (unit[disk].type->sect_mult==2) /* check for HD-Disks */ + for(cnt=0;cnt<473;cnt++) + *ptr++=0x92549254; + +/* now the index marks...*/ + for (cnt=0;cnt<20;cnt++) + *ptr++=0x92549254; + for (cnt=0;cnt<6;cnt++) + *ptr++=0xaaaaaaaa; + *ptr++=0x52245224; + *ptr++=0x52245552; + for (cnt=0;cnt<20;cnt++) + *ptr++=0x92549254; + +/* sectors */ + for(cnt = 0; cnt < unit[disk].dtype->sects * unit[disk].type->sect_mult; cnt++) + ptr=ms_putsec(disk,ptr,cnt); + + *(ushort *)ptr = 0xaaa8; /* MFM word before is always 0x9254 */ +} + +/* + * Here comes the high level stuff (i.e. the filesystem interface) + * and helper functions. + * Normally this should be the only part that has to be adapted to + * different kernel versions. + */ + +/* FIXME: this assumes the drive is still spinning - + * which is only true if we complete writing a track within three seconds + */ +static void flush_track_callback(unsigned long nr) +{ + nr&=3; + writefromint = 1; + if (!try_fdc(nr)) { + /* we might block in an interrupt, so try again later */ + flush_track_timer[nr].expires = jiffies + 1; + add_timer(flush_track_timer + nr); + return; + } + get_fdc(nr); + (*unit[nr].dtype->write_fkt)(nr); + if (!raw_write(nr)) { + printk (KERN_NOTICE "floppy disk write protected\n"); + writefromint = 0; + writepending = 0; + } + rel_fdc(); +} + +static int non_int_flush_track (unsigned long nr) +{ + unsigned long flags; + + nr&=3; + writefromint = 0; + del_timer(&post_write_timer); + get_fdc(nr); + if (!fd_motor_on(nr)) { + writepending = 0; + rel_fdc(); + return 0; + } + local_irq_save(flags); + if (writepending != 2) { + local_irq_restore(flags); + (*unit[nr].dtype->write_fkt)(nr); + if (!raw_write(nr)) { + printk (KERN_NOTICE "floppy disk write protected " + "in write!\n"); + writepending = 0; + return 0; + } + while (block_flag == 2) + sleep_on (&wait_fd_block); + } + else { + local_irq_restore(flags); + ms_delay(2); /* 2 ms post_write delay */ + post_write(nr); + } + rel_fdc(); + return 1; +} + +static int get_track(int drive, int track) +{ + int error, errcnt; + + drive&=3; + if (unit[drive].track == track) + return 0; + get_fdc(drive); + if (!fd_motor_on(drive)) { + rel_fdc(); + return -1; + } + + if (unit[drive].dirty == 1) { + del_timer (flush_track_timer + drive); + non_int_flush_track (drive); + } + errcnt = 0; + while (errcnt < MAX_ERRORS) { + if (!fd_seek(drive, track)) + return -1; + raw_read(drive); + error = (*unit[drive].dtype->read_fkt)(drive); + if (error == 0) { + rel_fdc(); + return 0; + } + /* Read Error Handling: recalibrate and try again */ + unit[drive].track = -1; + errcnt++; + } + rel_fdc(); + return -1; +} + +static void redo_fd_request(void) +{ + unsigned int cnt, block, track, sector; + int drive; + struct amiga_floppy_struct *floppy; + char *data; + unsigned long flags; + + repeat: + if (!CURRENT) { + /* Nothing left to do */ + return; + } + + floppy = CURRENT->rq_disk->private_data; + drive = floppy - unit; + + /* Here someone could investigate to be more efficient */ + for (cnt = 0; cnt < CURRENT->current_nr_sectors; cnt++) { +#ifdef DEBUG + printk("fd: sector %ld + %d requested for %s\n", + CURRENT->sector,cnt, + (CURRENT->cmd==READ)?"read":"write"); +#endif + block = CURRENT->sector + cnt; + if ((int)block > floppy->blocks) { + end_request(CURRENT, 0); + goto repeat; + } + + track = block / (floppy->dtype->sects * floppy->type->sect_mult); + sector = block % (floppy->dtype->sects * floppy->type->sect_mult); + data = CURRENT->buffer + 512 * cnt; +#ifdef DEBUG + printk("access to track %d, sector %d, with buffer at " + "0x%08lx\n", track, sector, data); +#endif + + if ((rq_data_dir(CURRENT) != READ) && (rq_data_dir(CURRENT) != WRITE)) { + printk(KERN_WARNING "do_fd_request: unknown command\n"); + end_request(CURRENT, 0); + goto repeat; + } + if (get_track(drive, track) == -1) { + end_request(CURRENT, 0); + goto repeat; + } + + switch (rq_data_dir(CURRENT)) { + case READ: + memcpy(data, floppy->trackbuf + sector * 512, 512); + break; + + case WRITE: + memcpy(floppy->trackbuf + sector * 512, data, 512); + + /* keep the drive spinning while writes are scheduled */ + if (!fd_motor_on(drive)) { + end_request(CURRENT, 0); + goto repeat; + } + /* + * setup a callback to write the track buffer + * after a short (1 tick) delay. + */ + local_irq_save(flags); + + floppy->dirty = 1; + /* reset the timer */ + mod_timer (flush_track_timer + drive, jiffies + 1); + local_irq_restore(flags); + break; + } + } + CURRENT->nr_sectors -= CURRENT->current_nr_sectors; + CURRENT->sector += CURRENT->current_nr_sectors; + + end_request(CURRENT, 1); + goto repeat; +} + +static void do_fd_request(request_queue_t * q) +{ + redo_fd_request(); +} + +static int fd_ioctl(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long param) +{ + int drive = iminor(inode) & 3; + static struct floppy_struct getprm; + + switch(cmd){ + case HDIO_GETGEO: + { + struct hd_geometry loc; + loc.heads = unit[drive].type->heads; + loc.sectors = unit[drive].dtype->sects * unit[drive].type->sect_mult; + loc.cylinders = unit[drive].type->tracks; + loc.start = 0; + if (copy_to_user((void *)param, (void *)&loc, + sizeof(struct hd_geometry))) + return -EFAULT; + break; + } + case FDFMTBEG: + get_fdc(drive); + if (fd_ref[drive] > 1) { + rel_fdc(); + return -EBUSY; + } + fsync_bdev(inode->i_bdev); + if (fd_motor_on(drive) == 0) { + rel_fdc(); + return -ENODEV; + } + if (fd_calibrate(drive) == 0) { + rel_fdc(); + return -ENXIO; + } + floppy_off(drive); + rel_fdc(); + break; + case FDFMTTRK: + if (param < unit[drive].type->tracks * unit[drive].type->heads) + { + get_fdc(drive); + if (fd_seek(drive,param) != 0){ + memset(unit[drive].trackbuf, FD_FILL_BYTE, + unit[drive].dtype->sects * unit[drive].type->sect_mult * 512); + non_int_flush_track(drive); + } + floppy_off(drive); + rel_fdc(); + } + else + return -EINVAL; + break; + case FDFMTEND: + floppy_off(drive); + invalidate_bdev(inode->i_bdev, 0); + break; + case FDGETPRM: + memset((void *)&getprm, 0, sizeof (getprm)); + getprm.track=unit[drive].type->tracks; + getprm.head=unit[drive].type->heads; + getprm.sect=unit[drive].dtype->sects * unit[drive].type->sect_mult; + getprm.size=unit[drive].blocks; + if (copy_to_user((void *)param, + (void *)&getprm, + sizeof(struct floppy_struct))) + return -EFAULT; + break; + case FDSETPRM: + case FDDEFPRM: + return -EINVAL; + case FDFLUSH: /* unconditionally, even if not needed */ + del_timer (flush_track_timer + drive); + non_int_flush_track(drive); + break; +#ifdef RAW_IOCTL + case IOCTL_RAW_TRACK: + if (copy_to_user((void *)param, raw_buf, + unit[drive].type->read_size)) + return -EFAULT; + else + return unit[drive].type->read_size; +#endif + default: + printk(KERN_DEBUG "fd_ioctl: unknown cmd %d for drive %d.", + cmd, drive); + return -ENOSYS; + } + return 0; +} + +static void fd_probe(int dev) +{ + unsigned long code; + int type; + int drive; + + drive = dev & 3; + code = fd_get_drive_id(drive); + + /* get drive type */ + for (type = 0; type < num_dr_types; type++) + if (drive_types[type].code == code) + break; + + if (type >= num_dr_types) { + printk(KERN_WARNING "fd_probe: unsupported drive type " + "%08lx found\n", code); + unit[drive].type = &drive_types[num_dr_types-1]; /* FD_NODRIVE */ + return; + } + + unit[drive].type = drive_types + type; + unit[drive].track = -1; + + unit[drive].disk = -1; + unit[drive].motor = 0; + unit[drive].busy = 0; + unit[drive].status = -1; +} + +/* + * floppy_open check for aliasing (/dev/fd0 can be the same as + * /dev/PS0 etc), and disallows simultaneous access to the same + * drive with different device numbers. + */ +static int floppy_open(struct inode *inode, struct file *filp) +{ + int drive = iminor(inode) & 3; + int system = (iminor(inode) & 4) >> 2; + int old_dev; + unsigned long flags; + + old_dev = fd_device[drive]; + + if (fd_ref[drive] && old_dev != system) + return -EBUSY; + + if (filp && filp->f_mode & 3) { + check_disk_change(inode->i_bdev); + if (filp->f_mode & 2 ) { + int wrprot; + + get_fdc(drive); + fd_select (drive); + wrprot = !(ciaa.pra & DSKPROT); + fd_deselect (drive); + rel_fdc(); + + if (wrprot) + return -EROFS; + } + } + + local_irq_save(flags); + fd_ref[drive]++; + fd_device[drive] = system; + local_irq_restore(flags); + + unit[drive].dtype=&data_types[system]; + unit[drive].blocks=unit[drive].type->heads*unit[drive].type->tracks* + data_types[system].sects*unit[drive].type->sect_mult; + set_capacity(unit[drive].gendisk, unit[drive].blocks); + + printk(KERN_INFO "fd%d: accessing %s-disk with %s-layout\n",drive, + unit[drive].type->name, data_types[system].name); + + return 0; +} + +static int floppy_release(struct inode * inode, struct file * filp) +{ + int drive = iminor(inode) & 3; + + if (unit[drive].dirty == 1) { + del_timer (flush_track_timer + drive); + non_int_flush_track (drive); + } + + if (!fd_ref[drive]--) { + printk(KERN_CRIT "floppy_release with fd_ref == 0"); + fd_ref[drive] = 0; + } +#ifdef MODULE +/* the mod_use counter is handled this way */ + floppy_off (drive | 0x40000000); +#endif + return 0; +} + +/* + * floppy-change is never called from an interrupt, so we can relax a bit + * here, sleep etc. Note that floppy-on tries to set current_DOR to point + * to the desired drive, but it will probably not survive the sleep if + * several floppies are used at the same time: thus the loop. + */ +static int amiga_floppy_change(struct gendisk *disk) +{ + struct amiga_floppy_struct *p = disk->private_data; + int drive = p - unit; + int changed; + static int first_time = 1; + + if (first_time) + changed = first_time--; + else { + get_fdc(drive); + fd_select (drive); + changed = !(ciaa.pra & DSKCHANGE); + fd_deselect (drive); + rel_fdc(); + } + + if (changed) { + fd_probe(drive); + p->track = -1; + p->dirty = 0; + writepending = 0; /* if this was true before, too bad! */ + writefromint = 0; + return 1; + } + return 0; +} + +static struct block_device_operations floppy_fops = { + .owner = THIS_MODULE, + .open = floppy_open, + .release = floppy_release, + .ioctl = fd_ioctl, + .media_changed = amiga_floppy_change, +}; + +void __init amiga_floppy_setup (char *str, int *ints) +{ + printk (KERN_INFO "amiflop: Setting default df0 to %x\n", ints[1]); + fd_def_df0 = ints[1]; +} + +static int __init fd_probe_drives(void) +{ + int drive,drives,nomem; + + printk(KERN_INFO "FD: probing units\n" KERN_INFO "found "); + drives=0; + nomem=0; + for(drive=0;drive<FD_MAX_UNITS;drive++) { + struct gendisk *disk; + fd_probe(drive); + if (unit[drive].type->code == FD_NODRIVE) + continue; + disk = alloc_disk(1); + if (!disk) { + unit[drive].type->code = FD_NODRIVE; + continue; + } + unit[drive].gendisk = disk; + drives++; + if ((unit[drive].trackbuf = kmalloc(FLOPPY_MAX_SECTORS * 512, GFP_KERNEL)) == NULL) { + printk("no mem for "); + unit[drive].type = &drive_types[num_dr_types - 1]; /* FD_NODRIVE */ + drives--; + nomem = 1; + } + printk("fd%d ",drive); + disk->major = FLOPPY_MAJOR; + disk->first_minor = drive; + disk->fops = &floppy_fops; + sprintf(disk->disk_name, "fd%d", drive); + disk->private_data = &unit[drive]; + disk->queue = floppy_queue; + set_capacity(disk, 880*2); + add_disk(disk); + } + if ((drives > 0) || (nomem == 0)) { + if (drives == 0) + printk("no drives"); + printk("\n"); + return drives; + } + printk("\n"); + return -ENOMEM; +} + +static struct kobject *floppy_find(dev_t dev, int *part, void *data) +{ + int drive = *part & 3; + if (unit[drive].type->code == FD_NODRIVE) + return NULL; + *part = 0; + return get_disk(unit[drive].gendisk); +} + +int __init amiga_floppy_init(void) +{ + int i, ret; + + if (!AMIGAHW_PRESENT(AMI_FLOPPY)) + return -ENXIO; + + if (register_blkdev(FLOPPY_MAJOR,"fd")) + return -EBUSY; + + /* + * We request DSKPTR, DSKLEN and DSKDATA only, because the other + * floppy registers are too spreaded over the custom register space + */ + ret = -EBUSY; + if (!request_mem_region(CUSTOM_PHYSADDR+0x20, 8, "amiflop [Paula]")) { + printk("fd: cannot get floppy registers\n"); + goto out_blkdev; + } + + ret = -ENOMEM; + if ((raw_buf = (char *)amiga_chip_alloc (RAW_BUF_SIZE, "Floppy")) == + NULL) { + printk("fd: cannot get chip mem buffer\n"); + goto out_memregion; + } + + ret = -EBUSY; + if (request_irq(IRQ_AMIGA_DSKBLK, fd_block_done, 0, "floppy_dma", NULL)) { + printk("fd: cannot get irq for dma\n"); + goto out_irq; + } + + if (request_irq(IRQ_AMIGA_CIAA_TB, ms_isr, 0, "floppy_timer", NULL)) { + printk("fd: cannot get irq for timer\n"); + goto out_irq2; + } + + ret = -ENOMEM; + floppy_queue = blk_init_queue(do_fd_request, &amiflop_lock); + if (!floppy_queue) + goto out_queue; + + ret = -ENXIO; + if (fd_probe_drives() < 1) /* No usable drives */ + goto out_probe; + + blk_register_region(MKDEV(FLOPPY_MAJOR, 0), 256, THIS_MODULE, + floppy_find, NULL, NULL); + + /* initialize variables */ + init_timer(&motor_on_timer); + motor_on_timer.expires = 0; + motor_on_timer.data = 0; + motor_on_timer.function = motor_on_callback; + for (i = 0; i < FD_MAX_UNITS; i++) { + init_timer(&motor_off_timer[i]); + motor_off_timer[i].expires = 0; + motor_off_timer[i].data = i|0x80000000; + motor_off_timer[i].function = fd_motor_off; + init_timer(&flush_track_timer[i]); + flush_track_timer[i].expires = 0; + flush_track_timer[i].data = i; + flush_track_timer[i].function = flush_track_callback; + + unit[i].track = -1; + } + + init_timer(&post_write_timer); + post_write_timer.expires = 0; + post_write_timer.data = 0; + post_write_timer.function = post_write; + + for (i = 0; i < 128; i++) + mfmdecode[i]=255; + for (i = 0; i < 16; i++) + mfmdecode[mfmencode[i]]=i; + + /* make sure that disk DMA is enabled */ + custom.dmacon = DMAF_SETCLR | DMAF_DISK; + + /* init ms timer */ + ciaa.crb = 8; /* one-shot, stop */ + return 0; + +out_probe: + blk_cleanup_queue(floppy_queue); +out_queue: + free_irq(IRQ_AMIGA_CIAA_TB, NULL); +out_irq2: + free_irq(IRQ_AMIGA_DSKBLK, NULL); +out_irq: + amiga_chip_free(raw_buf); +out_memregion: + release_mem_region(CUSTOM_PHYSADDR+0x20, 8); +out_blkdev: + unregister_blkdev(FLOPPY_MAJOR,"fd"); + return ret; +} + +#ifdef MODULE +#include <linux/version.h> + +int init_module(void) +{ + if (!MACH_IS_AMIGA) + return -ENXIO; + return amiga_floppy_init(); +} + +#if 0 /* not safe to unload */ +void cleanup_module(void) +{ + int i; + + for( i = 0; i < FD_MAX_UNITS; i++) { + if (unit[i].type->code != FD_NODRIVE) { + del_gendisk(unit[i].gendisk); + put_disk(unit[i].gendisk); + kfree(unit[i].trackbuf); + } + } + blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); + free_irq(IRQ_AMIGA_CIAA_TB, NULL); + free_irq(IRQ_AMIGA_DSKBLK, NULL); + custom.dmacon = DMAF_DISK; /* disable DMA */ + amiga_chip_free(raw_buf); + blk_cleanup_queue(floppy_queue); + release_mem_region(CUSTOM_PHYSADDR+0x20, 8); + unregister_blkdev(FLOPPY_MAJOR, "fd"); +} +#endif +#endif diff --git a/drivers/block/aoe/Makefile b/drivers/block/aoe/Makefile new file mode 100644 index 00000000000..e76d997183c --- /dev/null +++ b/drivers/block/aoe/Makefile @@ -0,0 +1,6 @@ +# +# Makefile for ATA over Ethernet +# + +obj-$(CONFIG_ATA_OVER_ETH) += aoe.o +aoe-objs := aoeblk.o aoechr.o aoecmd.o aoedev.o aoemain.o aoenet.o diff --git a/drivers/block/aoe/aoe.h b/drivers/block/aoe/aoe.h new file mode 100644 index 00000000000..db78f826d40 --- /dev/null +++ b/drivers/block/aoe/aoe.h @@ -0,0 +1,165 @@ +/* Copyright (c) 2004 Coraid, Inc. See COPYING for GPL terms. */ +#define VERSION "5" +#define AOE_MAJOR 152 +#define DEVICE_NAME "aoe" +#ifndef AOE_PARTITIONS +#define AOE_PARTITIONS 16 +#endif +#define SYSMINOR(aoemajor, aoeminor) ((aoemajor) * 10 + (aoeminor)) +#define AOEMAJOR(sysminor) ((sysminor) / 10) +#define AOEMINOR(sysminor) ((sysminor) % 10) +#define WHITESPACE " \t\v\f\n" + +enum { + AOECMD_ATA, + AOECMD_CFG, + + AOEFL_RSP = (1<<3), + AOEFL_ERR = (1<<2), + + AOEAFL_EXT = (1<<6), + AOEAFL_DEV = (1<<4), + AOEAFL_ASYNC = (1<<1), + AOEAFL_WRITE = (1<<0), + + AOECCMD_READ = 0, + AOECCMD_TEST, + AOECCMD_PTEST, + AOECCMD_SET, + AOECCMD_FSET, + + AOE_HVER = 0x10, +}; + +struct aoe_hdr { + unsigned char dst[6]; + unsigned char src[6]; + unsigned char type[2]; + unsigned char verfl; + unsigned char err; + unsigned char major[2]; + unsigned char minor; + unsigned char cmd; + unsigned char tag[4]; +}; + +struct aoe_atahdr { + unsigned char aflags; + unsigned char errfeat; + unsigned char scnt; + unsigned char cmdstat; + unsigned char lba0; + unsigned char lba1; + unsigned char lba2; + unsigned char lba3; + unsigned char lba4; + unsigned char lba5; + unsigned char res[2]; +}; + +struct aoe_cfghdr { + unsigned char bufcnt[2]; + unsigned char fwver[2]; + unsigned char res; + unsigned char aoeccmd; + unsigned char cslen[2]; +}; + +enum { + DEVFL_UP = 1, /* device is installed in system and ready for AoE->ATA commands */ + DEVFL_TKILL = (1<<1), /* flag for timer to know when to kill self */ + DEVFL_EXT = (1<<2), /* device accepts lba48 commands */ + DEVFL_CLOSEWAIT = (1<<3), /* device is waiting for all closes to revalidate */ + DEVFL_WC_UPDATE = (1<<4), /* this device needs to update write cache status */ + DEVFL_WORKON = (1<<4), + + BUFFL_FAIL = 1, +}; + +enum { + MAXATADATA = 1024, + NPERSHELF = 10, + FREETAG = -1, + MIN_BUFS = 8, +}; + +struct buf { + struct list_head bufs; + ulong flags; + ulong nframesout; + char *bufaddr; + ulong resid; + ulong bv_resid; + sector_t sector; + struct bio *bio; + struct bio_vec *bv; +}; + +struct frame { + int tag; + ulong waited; + struct buf *buf; + char *bufaddr; + int writedatalen; + int ndata; + + /* largest possible */ + unsigned char data[sizeof(struct aoe_hdr) + sizeof(struct aoe_atahdr)]; +}; + +struct aoedev { + struct aoedev *next; + unsigned char addr[6]; /* remote mac addr */ + ushort flags; + ulong sysminor; + ulong aoemajor; + ulong aoeminor; + ulong nopen; /* (bd_openers isn't available without sleeping) */ + ulong rttavg; /* round trip average of requests/responses */ + u16 fw_ver; /* version of blade's firmware */ + struct work_struct work;/* disk create work struct */ + struct gendisk *gd; + request_queue_t blkq; + struct hd_geometry geo; + sector_t ssize; + struct timer_list timer; + spinlock_t lock; + struct net_device *ifp; /* interface ed is attached to */ + struct sk_buff *skblist;/* packets needing to be sent */ + mempool_t *bufpool; /* for deadlock-free Buf allocation */ + struct list_head bufq; /* queue of bios to work on */ + struct buf *inprocess; /* the one we're currently working on */ + ulong lasttag; /* last tag sent */ + ulong nframes; /* number of frames below */ + struct frame *frames; +}; + + +int aoeblk_init(void); +void aoeblk_exit(void); +void aoeblk_gdalloc(void *); +void aoedisk_rm_sysfs(struct aoedev *d); + +int aoechr_init(void); +void aoechr_exit(void); +void aoechr_error(char *); + +void aoecmd_work(struct aoedev *d); +void aoecmd_cfg(ushort, unsigned char); +void aoecmd_ata_rsp(struct sk_buff *); +void aoecmd_cfg_rsp(struct sk_buff *); + +int aoedev_init(void); +void aoedev_exit(void); +struct aoedev *aoedev_bymac(unsigned char *); +void aoedev_downdev(struct aoedev *d); +struct aoedev *aoedev_set(ulong, unsigned char *, struct net_device *, ulong); +int aoedev_busy(void); + +int aoenet_init(void); +void aoenet_exit(void); +void aoenet_xmit(struct sk_buff *); +int is_aoe_netif(struct net_device *ifp); +int set_aoe_iflist(const char __user *str, size_t size); + +u64 mac_addr(char addr[6]); diff --git a/drivers/block/aoe/aoeblk.c b/drivers/block/aoe/aoeblk.c new file mode 100644 index 00000000000..63561b280bc --- /dev/null +++ b/drivers/block/aoe/aoeblk.c @@ -0,0 +1,267 @@ +/* Copyright (c) 2004 Coraid, Inc. See COPYING for GPL terms. */ +/* + * aoeblk.c + * block device routines + */ + +#include <linux/hdreg.h> +#include <linux/blkdev.h> +#include <linux/fs.h> +#include <linux/ioctl.h> +#include <linux/genhd.h> +#include <linux/netdevice.h> +#include "aoe.h" + +static kmem_cache_t *buf_pool_cache; + +/* add attributes for our block devices in sysfs */ +static ssize_t aoedisk_show_state(struct gendisk * disk, char *page) +{ + struct aoedev *d = disk->private_data; + + return snprintf(page, PAGE_SIZE, + "%s%s\n", + (d->flags & DEVFL_UP) ? "up" : "down", + (d->flags & DEVFL_CLOSEWAIT) ? ",closewait" : ""); +} +static ssize_t aoedisk_show_mac(struct gendisk * disk, char *page) +{ + struct aoedev *d = disk->private_data; + + return snprintf(page, PAGE_SIZE, "%012llx\n", + (unsigned long long)mac_addr(d->addr)); +} +static ssize_t aoedisk_show_netif(struct gendisk * disk, char *page) +{ + struct aoedev *d = disk->private_data; + + return snprintf(page, PAGE_SIZE, "%s\n", d->ifp->name); +} + +static struct disk_attribute disk_attr_state = { + .attr = {.name = "state", .mode = S_IRUGO }, + .show = aoedisk_show_state +}; +static struct disk_attribute disk_attr_mac = { + .attr = {.name = "mac", .mode = S_IRUGO }, + .show = aoedisk_show_mac +}; +static struct disk_attribute disk_attr_netif = { + .attr = {.name = "netif", .mode = S_IRUGO }, + .show = aoedisk_show_netif +}; + +static void +aoedisk_add_sysfs(struct aoedev *d) +{ + sysfs_create_file(&d->gd->kobj, &disk_attr_state.attr); + sysfs_create_file(&d->gd->kobj, &disk_attr_mac.attr); + sysfs_create_file(&d->gd->kobj, &disk_attr_netif.attr); +} +void +aoedisk_rm_sysfs(struct aoedev *d) +{ + sysfs_remove_link(&d->gd->kobj, "state"); + sysfs_remove_link(&d->gd->kobj, "mac"); + sysfs_remove_link(&d->gd->kobj, "netif"); +} + +static int +aoeblk_open(struct inode *inode, struct file *filp) +{ + struct aoedev *d; + ulong flags; + + d = inode->i_bdev->bd_disk->private_data; + + spin_lock_irqsave(&d->lock, flags); + if (d->flags & DEVFL_UP) { + d->nopen++; + spin_unlock_irqrestore(&d->lock, flags); + return 0; + } + spin_unlock_irqrestore(&d->lock, flags); + return -ENODEV; +} + +static int +aoeblk_release(struct inode *inode, struct file *filp) +{ + struct aoedev *d; + ulong flags; + + d = inode->i_bdev->bd_disk->private_data; + + spin_lock_irqsave(&d->lock, flags); + + if (--d->nopen == 0 && (d->flags & DEVFL_CLOSEWAIT)) { + d->flags &= ~DEVFL_CLOSEWAIT; + spin_unlock_irqrestore(&d->lock, flags); + aoecmd_cfg(d->aoemajor, d->aoeminor); + return 0; + } + spin_unlock_irqrestore(&d->lock, flags); + + return 0; +} + +static int +aoeblk_make_request(request_queue_t *q, struct bio *bio) +{ + struct aoedev *d; + struct buf *buf; + struct sk_buff *sl; + ulong flags; + + blk_queue_bounce(q, &bio); + + d = bio->bi_bdev->bd_disk->private_data; + buf = mempool_alloc(d->bufpool, GFP_NOIO); + if (buf == NULL) { + printk(KERN_INFO "aoe: aoeblk_make_request: buf allocation " + "failure\n"); + bio_endio(bio, bio->bi_size, -ENOMEM); + return 0; + } + memset(buf, 0, sizeof(*buf)); + INIT_LIST_HEAD(&buf->bufs); + buf->bio = bio; + buf->resid = bio->bi_size; + buf->sector = bio->bi_sector; + buf->bv = buf->bio->bi_io_vec; + buf->bv_resid = buf->bv->bv_len; + buf->bufaddr = page_address(buf->bv->bv_page) + buf->bv->bv_offset; + + spin_lock_irqsave(&d->lock, flags); + + if ((d->flags & DEVFL_UP) == 0) { + printk(KERN_INFO "aoe: aoeblk_make_request: device %ld.%ld is not up\n", + d->aoemajor, d->aoeminor); + spin_unlock_irqrestore(&d->lock, flags); + mempool_free(buf, d->bufpool); + bio_endio(bio, bio->bi_size, -ENXIO); + return 0; + } + + list_add_tail(&buf->bufs, &d->bufq); + aoecmd_work(d); + + sl = d->skblist; + d->skblist = NULL; + + spin_unlock_irqrestore(&d->lock, flags); + + aoenet_xmit(sl); + return 0; +} + +/* This ioctl implementation expects userland to have the device node + * permissions set so that only priviledged users can open an aoe + * block device directly. + */ +static int +aoeblk_ioctl(struct inode *inode, struct file *filp, uint cmd, ulong arg) +{ + struct aoedev *d; + + if (!arg) + return -EINVAL; + + d = inode->i_bdev->bd_disk->private_data; + if ((d->flags & DEVFL_UP) == 0) { + printk(KERN_ERR "aoe: aoeblk_ioctl: disk not up\n"); + return -ENODEV; + } + + if (cmd == HDIO_GETGEO) { + d->geo.start = get_start_sect(inode->i_bdev); + if (!copy_to_user((void __user *) arg, &d->geo, sizeof d->geo)) + return 0; + return -EFAULT; + } + printk(KERN_INFO "aoe: aoeblk_ioctl: unknown ioctl %d\n", cmd); + return -EINVAL; +} + +static struct block_device_operations aoe_bdops = { + .open = aoeblk_open, + .release = aoeblk_release, + .ioctl = aoeblk_ioctl, + .owner = THIS_MODULE, +}; + +/* alloc_disk and add_disk can sleep */ +void +aoeblk_gdalloc(void *vp) +{ + struct aoedev *d = vp; + struct gendisk *gd; + ulong flags; + + gd = alloc_disk(AOE_PARTITIONS); + if (gd == NULL) { + printk(KERN_ERR "aoe: aoeblk_gdalloc: cannot allocate disk " + "structure for %ld.%ld\n", d->aoemajor, d->aoeminor); + spin_lock_irqsave(&d->lock, flags); + d->flags &= ~DEVFL_WORKON; + spin_unlock_irqrestore(&d->lock, flags); + return; + } + + d->bufpool = mempool_create(MIN_BUFS, + mempool_alloc_slab, mempool_free_slab, + buf_pool_cache); + if (d->bufpool == NULL) { + printk(KERN_ERR "aoe: aoeblk_gdalloc: cannot allocate bufpool " + "for %ld.%ld\n", d->aoemajor, d->aoeminor); + put_disk(gd); + spin_lock_irqsave(&d->lock, flags); + d->flags &= ~DEVFL_WORKON; + spin_unlock_irqrestore(&d->lock, flags); + return; + } + + spin_lock_irqsave(&d->lock, flags); + blk_queue_make_request(&d->blkq, aoeblk_make_request); + gd->major = AOE_MAJOR; + gd->first_minor = d->sysminor * AOE_PARTITIONS; + gd->fops = &aoe_bdops; + gd->private_data = d; + gd->capacity = d->ssize; + snprintf(gd->disk_name, sizeof gd->disk_name, "etherd/e%ld.%ld", + d->aoemajor, d->aoeminor); + + gd->queue = &d->blkq; + d->gd = gd; + d->flags &= ~DEVFL_WORKON; + d->flags |= DEVFL_UP; + + spin_unlock_irqrestore(&d->lock, flags); + + add_disk(gd); + aoedisk_add_sysfs(d); + + printk(KERN_INFO "aoe: %012llx e%lu.%lu v%04x has %llu " + "sectors\n", (unsigned long long)mac_addr(d->addr), + d->aoemajor, d->aoeminor, + d->fw_ver, (long long)d->ssize); +} + +void +aoeblk_exit(void) +{ + kmem_cache_destroy(buf_pool_cache); +} + +int __init +aoeblk_init(void) +{ + buf_pool_cache = kmem_cache_create("aoe_bufs", + sizeof(struct buf), + 0, 0, NULL, NULL); + if (buf_pool_cache == NULL) + return -ENOMEM; + + return 0; +} + diff --git a/drivers/block/aoe/aoechr.c b/drivers/block/aoe/aoechr.c new file mode 100644 index 00000000000..14aeca3e2e8 --- /dev/null +++ b/drivers/block/aoe/aoechr.c @@ -0,0 +1,244 @@ +/* Copyright (c) 2004 Coraid, Inc. See COPYING for GPL terms. */ +/* + * aoechr.c + * AoE character device driver + */ + +#include <linux/hdreg.h> +#include <linux/blkdev.h> +#include "aoe.h" + +enum { + //MINOR_STAT = 1, (moved to sysfs) + MINOR_ERR = 2, + MINOR_DISCOVER, + MINOR_INTERFACES, + MSGSZ = 2048, + NARGS = 10, + NMSG = 100, /* message backlog to retain */ +}; + +struct aoe_chardev { + ulong minor; + char name[32]; +}; + +enum { EMFL_VALID = 1 }; + +struct ErrMsg { + short flags; + short len; + char *msg; +}; + +static struct ErrMsg emsgs[NMSG]; +static int emsgs_head_idx, emsgs_tail_idx; +static struct semaphore emsgs_sema; +static spinlock_t emsgs_lock; +static int nblocked_emsgs_readers; +static struct class_simple *aoe_class; +static struct aoe_chardev chardevs[] = { + { MINOR_ERR, "err" }, + { MINOR_DISCOVER, "discover" }, + { MINOR_INTERFACES, "interfaces" }, +}; + +static int +discover(void) +{ + aoecmd_cfg(0xffff, 0xff); + return 0; +} + +static int +interfaces(const char __user *str, size_t size) +{ + if (set_aoe_iflist(str, size)) { + printk(KERN_CRIT + "%s: could not set interface list: %s\n", + __FUNCTION__, "too many interfaces"); + return -EINVAL; + } + return 0; +} + +void +aoechr_error(char *msg) +{ + struct ErrMsg *em; + char *mp; + ulong flags, n; + + n = strlen(msg); + + spin_lock_irqsave(&emsgs_lock, flags); + + em = emsgs + emsgs_tail_idx; + if ((em->flags & EMFL_VALID)) { +bail: spin_unlock_irqrestore(&emsgs_lock, flags); + return; + } + + mp = kmalloc(n, GFP_ATOMIC); + if (mp == NULL) { + printk(KERN_CRIT "aoe: aoechr_error: allocation failure, len=%ld\n", n); + goto bail; + } + + memcpy(mp, msg, n); + em->msg = mp; + em->flags |= EMFL_VALID; + em->len = n; + + emsgs_tail_idx++; + emsgs_tail_idx %= ARRAY_SIZE(emsgs); + + spin_unlock_irqrestore(&emsgs_lock, flags); + + if (nblocked_emsgs_readers) + up(&emsgs_sema); +} + +static ssize_t +aoechr_write(struct file *filp, const char __user *buf, size_t cnt, loff_t *offp) +{ + int ret = -EINVAL; + + switch ((unsigned long) filp->private_data) { + default: + printk(KERN_INFO "aoe: aoechr_write: can't write to that file.\n"); + break; + case MINOR_DISCOVER: + ret = discover(); + break; + case MINOR_INTERFACES: + ret = interfaces(buf, cnt); + break; + } + if (ret == 0) + ret = cnt; + return ret; +} + +static int +aoechr_open(struct inode *inode, struct file *filp) +{ + int n, i; + + n = MINOR(inode->i_rdev); + filp->private_data = (void *) (unsigned long) n; + + for (i = 0; i < ARRAY_SIZE(chardevs); ++i) + if (chardevs[i].minor == n) + return 0; + return -EINVAL; +} + +static int +aoechr_rel(struct inode *inode, struct file *filp) +{ + return 0; +} + +static ssize_t +aoechr_read(struct file *filp, char __user *buf, size_t cnt, loff_t *off) +{ + unsigned long n; + char *mp; + struct ErrMsg *em; + ssize_t len; + ulong flags; + + n = (unsigned long) filp->private_data; + switch (n) { + case MINOR_ERR: + spin_lock_irqsave(&emsgs_lock, flags); +loop: + em = emsgs + emsgs_head_idx; + if ((em->flags & EMFL_VALID) == 0) { + if (filp->f_flags & O_NDELAY) { + spin_unlock_irqrestore(&emsgs_lock, flags); + return -EAGAIN; + } + nblocked_emsgs_readers++; + + spin_unlock_irqrestore(&emsgs_lock, flags); + + n = down_interruptible(&emsgs_sema); + + spin_lock_irqsave(&emsgs_lock, flags); + + nblocked_emsgs_readers--; + + if (n) { + spin_unlock_irqrestore(&emsgs_lock, flags); + return -ERESTARTSYS; + } + goto loop; + } + if (em->len > cnt) { + spin_unlock_irqrestore(&emsgs_lock, flags); + return -EAGAIN; + } + mp = em->msg; + len = em->len; + em->msg = NULL; + em->flags &= ~EMFL_VALID; + + emsgs_head_idx++; + emsgs_head_idx %= ARRAY_SIZE(emsgs); + + spin_unlock_irqrestore(&emsgs_lock, flags); + + n = copy_to_user(buf, mp, len); + kfree(mp); + return n == 0 ? len : -EFAULT; + default: + return -EFAULT; + } +} + +static struct file_operations aoe_fops = { + .write = aoechr_write, + .read = aoechr_read, + .open = aoechr_open, + .release = aoechr_rel, + .owner = THIS_MODULE, +}; + +int __init +aoechr_init(void) +{ + int n, i; + + n = register_chrdev(AOE_MAJOR, "aoechr", &aoe_fops); + if (n < 0) { + printk(KERN_ERR "aoe: aoechr_init: can't register char device\n"); + return n; + } + sema_init(&emsgs_sema, 0); + spin_lock_init(&emsgs_lock); + aoe_class = class_simple_create(THIS_MODULE, "aoe"); + if (IS_ERR(aoe_class)) { + unregister_chrdev(AOE_MAJOR, "aoechr"); + return PTR_ERR(aoe_class); + } + for (i = 0; i < ARRAY_SIZE(chardevs); ++i) + class_simple_device_add(aoe_class, + MKDEV(AOE_MAJOR, chardevs[i].minor), + NULL, chardevs[i].name); + + return 0; +} + +void +aoechr_exit(void) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(chardevs); ++i) + class_simple_device_remove(MKDEV(AOE_MAJOR, chardevs[i].minor)); + class_simple_destroy(aoe_class); + unregister_chrdev(AOE_MAJOR, "aoechr"); +} + diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c new file mode 100644 index 00000000000..fb6d942a456 --- /dev/null +++ b/drivers/block/aoe/aoecmd.c @@ -0,0 +1,629 @@ +/* Copyright (c) 2004 Coraid, Inc. See COPYING for GPL terms. */ +/* + * aoecmd.c + * Filesystem request handling methods + */ + +#include <linux/hdreg.h> +#include <linux/blkdev.h> +#include <linux/skbuff.h> +#include <linux/netdevice.h> +#include "aoe.h" + +#define TIMERTICK (HZ / 10) +#define MINTIMER (2 * TIMERTICK) +#define MAXTIMER (HZ << 1) +#define MAXWAIT (60 * 3) /* After MAXWAIT seconds, give up and fail dev */ + +static struct sk_buff * +new_skb(struct net_device *if_dev, ulong len) +{ + struct sk_buff *skb; + + skb = alloc_skb(len, GFP_ATOMIC); + if (skb) { + skb->nh.raw = skb->mac.raw = skb->data; + skb->dev = if_dev; + skb->protocol = __constant_htons(ETH_P_AOE); + skb->priority = 0; + skb_put(skb, len); + skb->next = skb->prev = NULL; + + /* tell the network layer not to perform IP checksums + * or to get the NIC to do it + */ + skb->ip_summed = CHECKSUM_NONE; + } + return skb; +} + +static struct sk_buff * +skb_prepare(struct aoedev *d, struct frame *f) +{ + struct sk_buff *skb; + char *p; + + skb = new_skb(d->ifp, f->ndata + f->writedatalen); + if (!skb) { + printk(KERN_INFO "aoe: skb_prepare: failure to allocate skb\n"); + return NULL; + } + + p = skb->mac.raw; + memcpy(p, f->data, f->ndata); + + if (f->writedatalen) { + p += sizeof(struct aoe_hdr) + sizeof(struct aoe_atahdr); + memcpy(p, f->bufaddr, f->writedatalen); + } + + return skb; +} + +static struct frame * +getframe(struct aoedev *d, int tag) +{ + struct frame *f, *e; + + f = d->frames; + e = f + d->nframes; + for (; f<e; f++) + if (f->tag == tag) + return f; + return NULL; +} + +/* + * Leave the top bit clear so we have tagspace for userland. + * The bottom 16 bits are the xmit tick for rexmit/rttavg processing. + * This driver reserves tag -1 to mean "unused frame." + */ +static int +newtag(struct aoedev *d) +{ + register ulong n; + + n = jiffies & 0xffff; + return n |= (++d->lasttag & 0x7fff) << 16; +} + +static int +aoehdr_atainit(struct aoedev *d, struct aoe_hdr *h) +{ + u16 type = __constant_cpu_to_be16(ETH_P_AOE); + u16 aoemajor = __cpu_to_be16(d->aoemajor); + u32 host_tag = newtag(d); + u32 tag = __cpu_to_be32(host_tag); + + memcpy(h->src, d->ifp->dev_addr, sizeof h->src); + memcpy(h->dst, d->addr, sizeof h->dst); + memcpy(h->type, &type, sizeof type); + h->verfl = AOE_HVER; + memcpy(h->major, &aoemajor, sizeof aoemajor); + h->minor = d->aoeminor; + h->cmd = AOECMD_ATA; + memcpy(h->tag, &tag, sizeof tag); + + return host_tag; +} + +static void +aoecmd_ata_rw(struct aoedev *d, struct frame *f) +{ + struct aoe_hdr *h; + struct aoe_atahdr *ah; + struct buf *buf; + struct sk_buff *skb; + ulong bcnt; + register sector_t sector; + char writebit, extbit; + + writebit = 0x10; + extbit = 0x4; + + buf = d->inprocess; + + sector = buf->sector; + bcnt = buf->bv_resid; + if (bcnt > MAXATADATA) + bcnt = MAXATADATA; + + /* initialize the headers & frame */ + h = (struct aoe_hdr *) f->data; + ah = (struct aoe_atahdr *) (h+1); + f->ndata = sizeof *h + sizeof *ah; + memset(h, 0, f->ndata); + f->tag = aoehdr_atainit(d, h); + f->waited = 0; + f->buf = buf; + f->bufaddr = buf->bufaddr; + + /* set up ata header */ + ah->scnt = bcnt >> 9; + ah->lba0 = sector; + ah->lba1 = sector >>= 8; + ah->lba2 = sector >>= 8; + ah->lba3 = sector >>= 8; + if (d->flags & DEVFL_EXT) { + ah->aflags |= AOEAFL_EXT; + ah->lba4 = sector >>= 8; + ah->lba5 = sector >>= 8; + } else { + extbit = 0; + ah->lba3 &= 0x0f; + ah->lba3 |= 0xe0; /* LBA bit + obsolete 0xa0 */ + } + + if (bio_data_dir(buf->bio) == WRITE) { + ah->aflags |= AOEAFL_WRITE; + f->writedatalen = bcnt; + } else { + writebit = 0; + f->writedatalen = 0; + } + + ah->cmdstat = WIN_READ | writebit | extbit; + + /* mark all tracking fields and load out */ + buf->nframesout += 1; + buf->bufaddr += bcnt; + buf->bv_resid -= bcnt; +/* printk(KERN_INFO "aoe: bv_resid=%ld\n", buf->bv_resid); */ + buf->resid -= bcnt; + buf->sector += bcnt >> 9; + if (buf->resid == 0) { + d->inprocess = NULL; + } else if (buf->bv_resid == 0) { + buf->bv++; + buf->bv_resid = buf->bv->bv_len; + buf->bufaddr = page_address(buf->bv->bv_page) + buf->bv->bv_offset; + } + + skb = skb_prepare(d, f); + if (skb) { + skb->next = d->skblist; + d->skblist = skb; + } +} + +/* enters with d->lock held */ +void +aoecmd_work(struct aoedev *d) +{ + struct frame *f; + struct buf *buf; +loop: + f = getframe(d, FREETAG); + if (f == NULL) + return; + if (d->inprocess == NULL) { + if (list_empty(&d->bufq)) + return; + buf = container_of(d->bufq.next, struct buf, bufs); + list_del(d->bufq.next); +/*printk(KERN_INFO "aoecmd_work: bi_size=%ld\n", buf->bio->bi_size); */ + d->inprocess = buf; + } + aoecmd_ata_rw(d, f); + goto loop; +} + +static void +rexmit(struct aoedev *d, struct frame *f) +{ + struct sk_buff *skb; + struct aoe_hdr *h; + char buf[128]; + u32 n; + u32 net_tag; + + n = newtag(d); + + snprintf(buf, sizeof buf, + "%15s e%ld.%ld oldtag=%08x@%08lx newtag=%08x\n", + "retransmit", + d->aoemajor, d->aoeminor, f->tag, jiffies, n); + aoechr_error(buf); + + h = (struct aoe_hdr *) f->data; + f->tag = n; + net_tag = __cpu_to_be32(n); + memcpy(h->tag, &net_tag, sizeof net_tag); + + skb = skb_prepare(d, f); + if (skb) { + skb->next = d->skblist; + d->skblist = skb; + } +} + +static int +tsince(int tag) +{ + int n; + + n = jiffies & 0xffff; + n -= tag & 0xffff; + if (n < 0) + n += 1<<16; + return n; +} + +static void +rexmit_timer(ulong vp) +{ + struct aoedev *d; + struct frame *f, *e; + struct sk_buff *sl; + register long timeout; + ulong flags, n; + + d = (struct aoedev *) vp; + sl = NULL; + + /* timeout is always ~150% of the moving average */ + timeout = d->rttavg; + timeout += timeout >> 1; + + spin_lock_irqsave(&d->lock, flags); + + if (d->flags & DEVFL_TKILL) { +tdie: spin_unlock_irqrestore(&d->lock, flags); + return; + } + f = d->frames; + e = f + d->nframes; + for (; f<e; f++) { + if (f->tag != FREETAG && tsince(f->tag) >= timeout) { + n = f->waited += timeout; + n /= HZ; + if (n > MAXWAIT) { /* waited too long. device failure. */ + aoedev_downdev(d); + goto tdie; + } + rexmit(d, f); + } + } + + sl = d->skblist; + d->skblist = NULL; + if (sl) { + n = d->rttavg <<= 1; + if (n > MAXTIMER) + d->rttavg = MAXTIMER; + } + + d->timer.expires = jiffies + TIMERTICK; + add_timer(&d->timer); + + spin_unlock_irqrestore(&d->lock, flags); + + aoenet_xmit(sl); +} + +static void +ataid_complete(struct aoedev *d, unsigned char *id) +{ + u64 ssize; + u16 n; + + /* word 83: command set supported */ + n = __le16_to_cpu(*((u16 *) &id[83<<1])); + + /* word 86: command set/feature enabled */ + n |= __le16_to_cpu(*((u16 *) &id[86<<1])); + + if (n & (1<<10)) { /* bit 10: LBA 48 */ + d->flags |= DEVFL_EXT; + + /* word 100: number lba48 sectors */ + ssize = __le64_to_cpu(*((u64 *) &id[100<<1])); + + /* set as in ide-disk.c:init_idedisk_capacity */ + d->geo.cylinders = ssize; + d->geo.cylinders /= (255 * 63); + d->geo.heads = 255; + d->geo.sectors = 63; + } else { + d->flags &= ~DEVFL_EXT; + + /* number lba28 sectors */ + ssize = __le32_to_cpu(*((u32 *) &id[60<<1])); + + /* NOTE: obsolete in ATA 6 */ + d->geo.cylinders = __le16_to_cpu(*((u16 *) &id[54<<1])); + d->geo.heads = __le16_to_cpu(*((u16 *) &id[55<<1])); + d->geo.sectors = __le16_to_cpu(*((u16 *) &id[56<<1])); + } + d->ssize = ssize; + d->geo.start = 0; + if (d->gd != NULL) { + d->gd->capacity = ssize; + d->flags |= DEVFL_UP; + return; + } + if (d->flags & DEVFL_WORKON) { + printk(KERN_INFO "aoe: ataid_complete: can't schedule work, it's already on! " + "(This really shouldn't happen).\n"); + return; + } + INIT_WORK(&d->work, aoeblk_gdalloc, d); + schedule_work(&d->work); + d->flags |= DEVFL_WORKON; +} + +static void +calc_rttavg(struct aoedev *d, int rtt) +{ + register long n; + + n = rtt; + if (n < MINTIMER) + n = MINTIMER; + else if (n > MAXTIMER) + n = MAXTIMER; + + /* g == .25; cf. Congestion Avoidance and Control, Jacobson & Karels; 1988 */ + n -= d->rttavg; + d->rttavg += n >> 2; +} + +void +aoecmd_ata_rsp(struct sk_buff *skb) +{ + struct aoedev *d; + struct aoe_hdr *hin; + struct aoe_atahdr *ahin, *ahout; + struct frame *f; + struct buf *buf; + struct sk_buff *sl; + register long n; + ulong flags; + char ebuf[128]; + + hin = (struct aoe_hdr *) skb->mac.raw; + d = aoedev_bymac(hin->src); + if (d == NULL) { + snprintf(ebuf, sizeof ebuf, "aoecmd_ata_rsp: ata response " + "for unknown device %d.%d\n", + __be16_to_cpu(*((u16 *) hin->major)), + hin->minor); + aoechr_error(ebuf); + return; + } + + spin_lock_irqsave(&d->lock, flags); + + f = getframe(d, __be32_to_cpu(*((u32 *) hin->tag))); + if (f == NULL) { + spin_unlock_irqrestore(&d->lock, flags); + snprintf(ebuf, sizeof ebuf, + "%15s e%d.%d tag=%08x@%08lx\n", + "unexpected rsp", + __be16_to_cpu(*((u16 *) hin->major)), + hin->minor, + __be32_to_cpu(*((u32 *) hin->tag)), + jiffies); + aoechr_error(ebuf); + return; + } + + calc_rttavg(d, tsince(f->tag)); + + ahin = (struct aoe_atahdr *) (hin+1); + ahout = (struct aoe_atahdr *) (f->data + sizeof(struct aoe_hdr)); + buf = f->buf; + + if (ahin->cmdstat & 0xa9) { /* these bits cleared on success */ + printk(KERN_CRIT "aoe: aoecmd_ata_rsp: ata error cmd=%2.2Xh " + "stat=%2.2Xh from e%ld.%ld\n", + ahout->cmdstat, ahin->cmdstat, + d->aoemajor, d->aoeminor); + if (buf) + buf->flags |= BUFFL_FAIL; + } else { + switch (ahout->cmdstat) { + case WIN_READ: + case WIN_READ_EXT: + n = ahout->scnt << 9; + if (skb->len - sizeof *hin - sizeof *ahin < n) { + printk(KERN_CRIT "aoe: aoecmd_ata_rsp: runt " + "ata data size in read. skb->len=%d\n", + skb->len); + /* fail frame f? just returning will rexmit. */ + spin_unlock_irqrestore(&d->lock, flags); + return; + } + memcpy(f->bufaddr, ahin+1, n); + case WIN_WRITE: + case WIN_WRITE_EXT: + break; + case WIN_IDENTIFY: + if (skb->len - sizeof *hin - sizeof *ahin < 512) { + printk(KERN_INFO "aoe: aoecmd_ata_rsp: runt data size " + "in ataid. skb->len=%d\n", skb->len); + spin_unlock_irqrestore(&d->lock, flags); + return; + } + ataid_complete(d, (char *) (ahin+1)); + /* d->flags |= DEVFL_WC_UPDATE; */ + break; + default: + printk(KERN_INFO "aoe: aoecmd_ata_rsp: unrecognized " + "outbound ata command %2.2Xh for %d.%d\n", + ahout->cmdstat, + __be16_to_cpu(*((u16 *) hin->major)), + hin->minor); + } + } + + if (buf) { + buf->nframesout -= 1; + if (buf->nframesout == 0 && buf->resid == 0) { + n = (buf->flags & BUFFL_FAIL) ? -EIO : 0; + bio_endio(buf->bio, buf->bio->bi_size, n); + mempool_free(buf, d->bufpool); + } + } + + f->buf = NULL; + f->tag = FREETAG; + + aoecmd_work(d); + + sl = d->skblist; + d->skblist = NULL; + + spin_unlock_irqrestore(&d->lock, flags); + + aoenet_xmit(sl); +} + +void +aoecmd_cfg(ushort aoemajor, unsigned char aoeminor) +{ + struct aoe_hdr *h; + struct aoe_cfghdr *ch; + struct sk_buff *skb, *sl; + struct net_device *ifp; + u16 aoe_type = __constant_cpu_to_be16(ETH_P_AOE); + u16 net_aoemajor = __cpu_to_be16(aoemajor); + + sl = NULL; + + read_lock(&dev_base_lock); + for (ifp = dev_base; ifp; dev_put(ifp), ifp = ifp->next) { + dev_hold(ifp); + if (!is_aoe_netif(ifp)) + continue; + + skb = new_skb(ifp, sizeof *h + sizeof *ch); + if (skb == NULL) { + printk(KERN_INFO "aoe: aoecmd_cfg: skb alloc failure\n"); + continue; + } + h = (struct aoe_hdr *) skb->mac.raw; + memset(h, 0, sizeof *h + sizeof *ch); + + memset(h->dst, 0xff, sizeof h->dst); + memcpy(h->src, ifp->dev_addr, sizeof h->src); + memcpy(h->type, &aoe_type, sizeof aoe_type); + h->verfl = AOE_HVER; + memcpy(h->major, &net_aoemajor, sizeof net_aoemajor); + h->minor = aoeminor; + h->cmd = AOECMD_CFG; + + skb->next = sl; + sl = skb; + } + read_unlock(&dev_base_lock); + + aoenet_xmit(sl); +} + +/* + * Since we only call this in one place (and it only prepares one frame) + * we just return the skb. Usually we'd chain it up to the d->skblist. + */ +static struct sk_buff * +aoecmd_ata_id(struct aoedev *d) +{ + struct aoe_hdr *h; + struct aoe_atahdr *ah; + struct frame *f; + struct sk_buff *skb; + + f = getframe(d, FREETAG); + if (f == NULL) { + printk(KERN_CRIT "aoe: aoecmd_ata_id: can't get a frame. " + "This shouldn't happen.\n"); + return NULL; + } + + /* initialize the headers & frame */ + h = (struct aoe_hdr *) f->data; + ah = (struct aoe_atahdr *) (h+1); + f->ndata = sizeof *h + sizeof *ah; + memset(h, 0, f->ndata); + f->tag = aoehdr_atainit(d, h); + f->waited = 0; + f->writedatalen = 0; + + /* this message initializes the device, so we reset the rttavg */ + d->rttavg = MAXTIMER; + + /* set up ata header */ + ah->scnt = 1; + ah->cmdstat = WIN_IDENTIFY; + ah->lba3 = 0xa0; + + skb = skb_prepare(d, f); + + /* we now want to start the rexmit tracking */ + d->flags &= ~DEVFL_TKILL; + d->timer.data = (ulong) d; + d->timer.function = rexmit_timer; + d->timer.expires = jiffies + TIMERTICK; + add_timer(&d->timer); + + return skb; +} + +void +aoecmd_cfg_rsp(struct sk_buff *skb) +{ + struct aoedev *d; + struct aoe_hdr *h; + struct aoe_cfghdr *ch; + ulong flags, bufcnt, sysminor, aoemajor; + struct sk_buff *sl; + enum { MAXFRAMES = 8, MAXSYSMINOR = 255 }; + + h = (struct aoe_hdr *) skb->mac.raw; + ch = (struct aoe_cfghdr *) (h+1); + + /* + * Enough people have their dip switches set backwards to + * warrant a loud message for this special case. + */ + aoemajor = __be16_to_cpu(*((u16 *) h->major)); + if (aoemajor == 0xfff) { + printk(KERN_CRIT "aoe: aoecmd_cfg_rsp: Warning: shelf " + "address is all ones. Check shelf dip switches\n"); + return; + } + + sysminor = SYSMINOR(aoemajor, h->minor); + if (sysminor > MAXSYSMINOR) { + printk(KERN_INFO "aoe: aoecmd_cfg_rsp: sysminor %ld too " + "large\n", sysminor); + return; + } + + bufcnt = __be16_to_cpu(*((u16 *) ch->bufcnt)); + if (bufcnt > MAXFRAMES) /* keep it reasonable */ + bufcnt = MAXFRAMES; + + d = aoedev_set(sysminor, h->src, skb->dev, bufcnt); + if (d == NULL) { + printk(KERN_INFO "aoe: aoecmd_cfg_rsp: device set failure\n"); + return; + } + + spin_lock_irqsave(&d->lock, flags); + + if (d->flags & (DEVFL_UP | DEVFL_CLOSEWAIT)) { + spin_unlock_irqrestore(&d->lock, flags); + return; + } + + d->fw_ver = __be16_to_cpu(*((u16 *) ch->fwver)); + + /* we get here only if the device is new */ + sl = aoecmd_ata_id(d); + + spin_unlock_irqrestore(&d->lock, flags); + + aoenet_xmit(sl); +} + diff --git a/drivers/block/aoe/aoedev.c b/drivers/block/aoe/aoedev.c new file mode 100644 index 00000000000..240abaec159 --- /dev/null +++ b/drivers/block/aoe/aoedev.c @@ -0,0 +1,180 @@ +/* Copyright (c) 2004 Coraid, Inc. See COPYING for GPL terms. */ +/* + * aoedev.c + * AoE device utility functions; maintains device list. + */ + +#include <linux/hdreg.h> +#include <linux/blkdev.h> +#include <linux/netdevice.h> +#include "aoe.h" + +static struct aoedev *devlist; +static spinlock_t devlist_lock; + +struct aoedev * +aoedev_bymac(unsigned char *macaddr) +{ + struct aoedev *d; + ulong flags; + + spin_lock_irqsave(&devlist_lock, flags); + + for (d=devlist; d; d=d->next) + if (!memcmp(d->addr, macaddr, 6)) + break; + + spin_unlock_irqrestore(&devlist_lock, flags); + return d; +} + +/* called with devlist lock held */ +static struct aoedev * +aoedev_newdev(ulong nframes) +{ + struct aoedev *d; + struct frame *f, *e; + + d = kcalloc(1, sizeof *d, GFP_ATOMIC); + if (d == NULL) + return NULL; + f = kcalloc(nframes, sizeof *f, GFP_ATOMIC); + if (f == NULL) { + kfree(d); + return NULL; + } + + d->nframes = nframes; + d->frames = f; + e = f + nframes; + for (; f<e; f++) + f->tag = FREETAG; + + spin_lock_init(&d->lock); + init_timer(&d->timer); + d->bufpool = NULL; /* defer to aoeblk_gdalloc */ + INIT_LIST_HEAD(&d->bufq); + d->next = devlist; + devlist = d; + + return d; +} + +void +aoedev_downdev(struct aoedev *d) +{ + struct frame *f, *e; + struct buf *buf; + struct bio *bio; + + d->flags |= DEVFL_TKILL; + del_timer(&d->timer); + + f = d->frames; + e = f + d->nframes; + for (; f<e; f->tag = FREETAG, f->buf = NULL, f++) { + if (f->tag == FREETAG || f->buf == NULL) + continue; + buf = f->buf; + bio = buf->bio; + if (--buf->nframesout == 0) { + mempool_free(buf, d->bufpool); + bio_endio(bio, bio->bi_size, -EIO); + } + } + d->inprocess = NULL; + + while (!list_empty(&d->bufq)) { + buf = container_of(d->bufq.next, struct buf, bufs); + list_del(d->bufq.next); + bio = buf->bio; + mempool_free(buf, d->bufpool); + bio_endio(bio, bio->bi_size, -EIO); + } + + if (d->nopen) + d->flags |= DEVFL_CLOSEWAIT; + if (d->gd) + d->gd->capacity = 0; + + d->flags &= ~DEVFL_UP; +} + +struct aoedev * +aoedev_set(ulong sysminor, unsigned char *addr, struct net_device *ifp, ulong bufcnt) +{ + struct aoedev *d; + ulong flags; + + spin_lock_irqsave(&devlist_lock, flags); + + for (d=devlist; d; d=d->next) + if (d->sysminor == sysminor + || memcmp(d->addr, addr, sizeof d->addr) == 0) + break; + + if (d == NULL && (d = aoedev_newdev(bufcnt)) == NULL) { + spin_unlock_irqrestore(&devlist_lock, flags); + printk(KERN_INFO "aoe: aoedev_set: aoedev_newdev failure.\n"); + return NULL; + } + + spin_unlock_irqrestore(&devlist_lock, flags); + spin_lock_irqsave(&d->lock, flags); + + d->ifp = ifp; + + if (d->sysminor != sysminor + || memcmp(d->addr, addr, sizeof d->addr) + || (d->flags & DEVFL_UP) == 0) { + aoedev_downdev(d); /* flushes outstanding frames */ + memcpy(d->addr, addr, sizeof d->addr); + d->sysminor = sysminor; + d->aoemajor = AOEMAJOR(sysminor); + d->aoeminor = AOEMINOR(sysminor); + } + + spin_unlock_irqrestore(&d->lock, flags); + return d; +} + +static void +aoedev_freedev(struct aoedev *d) +{ + if (d->gd) { + aoedisk_rm_sysfs(d); + del_gendisk(d->gd); + put_disk(d->gd); + } + kfree(d->frames); + mempool_destroy(d->bufpool); + kfree(d); +} + +void +aoedev_exit(void) +{ + struct aoedev *d; + ulong flags; + + flush_scheduled_work(); + + while ((d = devlist)) { + devlist = d->next; + + spin_lock_irqsave(&d->lock, flags); + aoedev_downdev(d); + spin_unlock_irqrestore(&d->lock, flags); + + del_timer_sync(&d->timer); + aoedev_freedev(d); + } +} + +int __init +aoedev_init(void) +{ + spin_lock_init(&devlist_lock); + return 0; +} + diff --git a/drivers/block/aoe/aoemain.c b/drivers/block/aoe/aoemain.c new file mode 100644 index 00000000000..387588a3f4b --- /dev/null +++ b/drivers/block/aoe/aoemain.c @@ -0,0 +1,112 @@ +/* Copyright (c) 2004 Coraid, Inc. See COPYING for GPL terms. */ +/* + * aoemain.c + * Module initialization routines, discover timer + */ + +#include <linux/hdreg.h> +#include <linux/blkdev.h> +#include <linux/module.h> +#include "aoe.h" + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Sam Hopkins <sah@coraid.com>"); +MODULE_DESCRIPTION("AoE block/char driver for 2.6.[0-9]+"); +MODULE_VERSION(VERSION); + +enum { TINIT, TRUN, TKILL }; + +static void +discover_timer(ulong vp) +{ + static struct timer_list t; + static volatile ulong die; + static spinlock_t lock; + ulong flags; + enum { DTIMERTICK = HZ * 60 }; /* one minute */ + + switch (vp) { + case TINIT: + init_timer(&t); + spin_lock_init(&lock); + t.data = TRUN; + t.function = discover_timer; + die = 0; + case TRUN: + spin_lock_irqsave(&lock, flags); + if (!die) { + t.expires = jiffies + DTIMERTICK; + add_timer(&t); + } + spin_unlock_irqrestore(&lock, flags); + + aoecmd_cfg(0xffff, 0xff); + return; + case TKILL: + spin_lock_irqsave(&lock, flags); + die = 1; + spin_unlock_irqrestore(&lock, flags); + + del_timer_sync(&t); + default: + return; + } +} + +static void +aoe_exit(void) +{ + discover_timer(TKILL); + + aoenet_exit(); + unregister_blkdev(AOE_MAJOR, DEVICE_NAME); + aoechr_exit(); + aoedev_exit(); + aoeblk_exit(); /* free cache after de-allocating bufs */ +} + +static int __init +aoe_init(void) +{ + int ret; + + ret = aoedev_init(); + if (ret) + return ret; + ret = aoechr_init(); + if (ret) + goto chr_fail; + ret = aoeblk_init(); + if (ret) + goto blk_fail; + ret = aoenet_init(); + if (ret) + goto net_fail; + ret = register_blkdev(AOE_MAJOR, DEVICE_NAME); + if (ret < 0) { + printk(KERN_ERR "aoe: aoeblk_init: can't register major\n"); + goto blkreg_fail; + } + + printk(KERN_INFO + "aoe: aoe_init: AoE v2.6-%s initialised.\n", + VERSION); + discover_timer(TINIT); + return 0; + + blkreg_fail: + aoenet_exit(); + net_fail: + aoeblk_exit(); + blk_fail: + aoechr_exit(); + chr_fail: + aoedev_exit(); + + printk(KERN_INFO "aoe: aoe_init: initialisation failure.\n"); + return ret; +} + +module_init(aoe_init); +module_exit(aoe_exit); + diff --git a/drivers/block/aoe/aoenet.c b/drivers/block/aoe/aoenet.c new file mode 100644 index 00000000000..cc1945b8d52 --- /dev/null +++ b/drivers/block/aoe/aoenet.c @@ -0,0 +1,172 @@ +/* Copyright (c) 2004 Coraid, Inc. See COPYING for GPL terms. */ +/* + * aoenet.c + * Ethernet portion of AoE driver + */ + +#include <linux/hdreg.h> +#include <linux/blkdev.h> +#include <linux/netdevice.h> +#include "aoe.h" + +#define NECODES 5 + +static char *aoe_errlist[] = +{ + "no such error", + "unrecognized command code", + "bad argument parameter", + "device unavailable", + "config string present", + "unsupported version" +}; + +enum { + IFLISTSZ = 1024, +}; + +static char aoe_iflist[IFLISTSZ]; + +int +is_aoe_netif(struct net_device *ifp) +{ + register char *p, *q; + register int len; + + if (aoe_iflist[0] == '\0') + return 1; + + for (p = aoe_iflist; *p; p = q + strspn(q, WHITESPACE)) { + q = p + strcspn(p, WHITESPACE); + if (q != p) + len = q - p; + else + len = strlen(p); /* last token in aoe_iflist */ + + if (strlen(ifp->name) == len && !strncmp(ifp->name, p, len)) + return 1; + if (q == p) + break; + } + + return 0; +} + +int +set_aoe_iflist(const char __user *user_str, size_t size) +{ + if (size >= IFLISTSZ) + return -EINVAL; + + if (copy_from_user(aoe_iflist, user_str, size)) { + printk(KERN_INFO "aoe: %s: copy from user failed\n", __FUNCTION__); + return -EFAULT; + } + aoe_iflist[size] = 0x00; + return 0; +} + +u64 +mac_addr(char addr[6]) +{ + u64 n = 0; + char *p = (char *) &n; + + memcpy(p + 2, addr, 6); /* (sizeof addr != 6) */ + + return __be64_to_cpu(n); +} + +static struct sk_buff * +skb_check(struct sk_buff *skb) +{ + if (skb_is_nonlinear(skb)) + if ((skb = skb_share_check(skb, GFP_ATOMIC))) + if (skb_linearize(skb, GFP_ATOMIC) < 0) { + dev_kfree_skb(skb); + return NULL; + } + return skb; +} + +void +aoenet_xmit(struct sk_buff *sl) +{ + struct sk_buff *skb; + + while ((skb = sl)) { + sl = sl->next; + skb->next = skb->prev = NULL; + dev_queue_xmit(skb); + } +} + +/* + * (1) len doesn't include the header by default. I want this. + */ +static int +aoenet_rcv(struct sk_buff *skb, struct net_device *ifp, struct packet_type *pt) +{ + struct aoe_hdr *h; + ulong n; + + skb = skb_check(skb); + if (!skb) + return 0; + + if (!is_aoe_netif(ifp)) + goto exit; + + //skb->len += ETH_HLEN; /* (1) */ + skb_push(skb, ETH_HLEN); /* (1) */ + + h = (struct aoe_hdr *) skb->mac.raw; + n = __be32_to_cpu(*((u32 *) h->tag)); + if ((h->verfl & AOEFL_RSP) == 0 || (n & 1<<31)) + goto exit; + + if (h->verfl & AOEFL_ERR) { + n = h->err; + if (n > NECODES) + n = 0; + if (net_ratelimit()) + printk(KERN_ERR "aoe: aoenet_rcv: error packet from %d.%d; " + "ecode=%d '%s'\n", + __be16_to_cpu(*((u16 *) h->major)), h->minor, + h->err, aoe_errlist[n]); + goto exit; + } + + switch (h->cmd) { + case AOECMD_ATA: + aoecmd_ata_rsp(skb); + break; + case AOECMD_CFG: + aoecmd_cfg_rsp(skb); + break; + default: + printk(KERN_INFO "aoe: aoenet_rcv: unknown cmd %d\n", h->cmd); + } +exit: + dev_kfree_skb(skb); + return 0; +} + +static struct packet_type aoe_pt = { + .type = __constant_htons(ETH_P_AOE), + .func = aoenet_rcv, +}; + +int __init +aoenet_init(void) +{ + dev_add_pack(&aoe_pt); + return 0; +} + +void +aoenet_exit(void) +{ + dev_remove_pack(&aoe_pt); +} + diff --git a/drivers/block/as-iosched.c b/drivers/block/as-iosched.c new file mode 100644 index 00000000000..a9575bb58a5 --- /dev/null +++ b/drivers/block/as-iosched.c @@ -0,0 +1,2136 @@ +/* + * linux/drivers/block/as-iosched.c + * + * Anticipatory & deadline i/o scheduler. + * + * Copyright (C) 2002 Jens Axboe <axboe@suse.de> + * Nick Piggin <piggin@cyberone.com.au> + * + */ +#include <linux/kernel.h> +#include <linux/fs.h> +#include <linux/blkdev.h> +#include <linux/elevator.h> +#include <linux/bio.h> +#include <linux/config.h> +#include <linux/module.h> +#include <linux/slab.h> +#include <linux/init.h> +#include <linux/compiler.h> +#include <linux/hash.h> +#include <linux/rbtree.h> +#include <linux/interrupt.h> + +#define REQ_SYNC 1 +#define REQ_ASYNC 0 + +/* + * See Documentation/block/as-iosched.txt + */ + +/* + * max time before a read is submitted. + */ +#define default_read_expire (HZ / 8) + +/* + * ditto for writes, these limits are not hard, even + * if the disk is capable of satisfying them. + */ +#define default_write_expire (HZ / 4) + +/* + * read_batch_expire describes how long we will allow a stream of reads to + * persist before looking to see whether it is time to switch over to writes. + */ +#define default_read_batch_expire (HZ / 2) + +/* + * write_batch_expire describes how long we want a stream of writes to run for. + * This is not a hard limit, but a target we set for the auto-tuning thingy. + * See, the problem is: we can send a lot of writes to disk cache / TCQ in + * a short amount of time... + */ +#define default_write_batch_expire (HZ / 8) + +/* + * max time we may wait to anticipate a read (default around 6ms) + */ +#define default_antic_expire ((HZ / 150) ? HZ / 150 : 1) + +/* + * Keep track of up to 20ms thinktimes. We can go as big as we like here, + * however huge values tend to interfere and not decay fast enough. A program + * might be in a non-io phase of operation. Waiting on user input for example, + * or doing a lengthy computation. A small penalty can be justified there, and + * will still catch out those processes that constantly have large thinktimes. + */ +#define MAX_THINKTIME (HZ/50UL) + +/* Bits in as_io_context.state */ +enum as_io_states { + AS_TASK_RUNNING=0, /* Process has not exitted */ + AS_TASK_IOSTARTED, /* Process has started some IO */ + AS_TASK_IORUNNING, /* Process has completed some IO */ +}; + +enum anticipation_status { + ANTIC_OFF=0, /* Not anticipating (normal operation) */ + ANTIC_WAIT_REQ, /* The last read has not yet completed */ + ANTIC_WAIT_NEXT, /* Currently anticipating a request vs + last read (which has completed) */ + ANTIC_FINISHED, /* Anticipating but have found a candidate + * or timed out */ +}; + +struct as_data { + /* + * run time data + */ + + struct request_queue *q; /* the "owner" queue */ + + /* + * requests (as_rq s) are present on both sort_list and fifo_list + */ + struct rb_root sort_list[2]; + struct list_head fifo_list[2]; + + struct as_rq *next_arq[2]; /* next in sort order */ + sector_t last_sector[2]; /* last REQ_SYNC & REQ_ASYNC sectors */ + struct list_head *dispatch; /* driver dispatch queue */ + struct list_head *hash; /* request hash */ + + unsigned long exit_prob; /* probability a task will exit while + being waited on */ + unsigned long new_ttime_total; /* mean thinktime on new proc */ + unsigned long new_ttime_mean; + u64 new_seek_total; /* mean seek on new proc */ + sector_t new_seek_mean; + + unsigned long current_batch_expires; + unsigned long last_check_fifo[2]; + int changed_batch; /* 1: waiting for old batch to end */ + int new_batch; /* 1: waiting on first read complete */ + int batch_data_dir; /* current batch REQ_SYNC / REQ_ASYNC */ + int write_batch_count; /* max # of reqs in a write batch */ + int current_write_count; /* how many requests left this batch */ + int write_batch_idled; /* has the write batch gone idle? */ + mempool_t *arq_pool; + + enum anticipation_status antic_status; + unsigned long antic_start; /* jiffies: when it started */ + struct timer_list antic_timer; /* anticipatory scheduling timer */ + struct work_struct antic_work; /* Deferred unplugging */ + struct io_context *io_context; /* Identify the expected process */ + int ioc_finished; /* IO associated with io_context is finished */ + int nr_dispatched; + + /* + * settings that change how the i/o scheduler behaves + */ + unsigned long fifo_expire[2]; + unsigned long batch_expire[2]; + unsigned long antic_expire; +}; + +#define list_entry_fifo(ptr) list_entry((ptr), struct as_rq, fifo) + +/* + * per-request data. + */ +enum arq_state { + AS_RQ_NEW=0, /* New - not referenced and not on any lists */ + AS_RQ_QUEUED, /* In the request queue. It belongs to the + scheduler */ + AS_RQ_DISPATCHED, /* On the dispatch list. It belongs to the + driver now */ + AS_RQ_PRESCHED, /* Debug poisoning for requests being used */ + AS_RQ_REMOVED, + AS_RQ_MERGED, + AS_RQ_POSTSCHED, /* when they shouldn't be */ +}; + +struct as_rq { + /* + * rbtree index, key is the starting offset + */ + struct rb_node rb_node; + sector_t rb_key; + + struct request *request; + + struct io_context *io_context; /* The submitting task */ + + /* + * request hash, key is the ending offset (for back merge lookup) + */ + struct list_head hash; + unsigned int on_hash; + + /* + * expire fifo + */ + struct list_head fifo; + unsigned long expires; + + unsigned int is_sync; + enum arq_state state; +}; + +#define RQ_DATA(rq) ((struct as_rq *) (rq)->elevator_private) + +static kmem_cache_t *arq_pool; + +/* + * IO Context helper functions + */ + +/* Called to deallocate the as_io_context */ +static void free_as_io_context(struct as_io_context *aic) +{ + kfree(aic); +} + +/* Called when the task exits */ +static void exit_as_io_context(struct as_io_context *aic) +{ + WARN_ON(!test_bit(AS_TASK_RUNNING, &aic->state)); + clear_bit(AS_TASK_RUNNING, &aic->state); +} + +static struct as_io_context *alloc_as_io_context(void) +{ + struct as_io_context *ret; + + ret = kmalloc(sizeof(*ret), GFP_ATOMIC); + if (ret) { + ret->dtor = free_as_io_context; + ret->exit = exit_as_io_context; + ret->state = 1 << AS_TASK_RUNNING; + atomic_set(&ret->nr_queued, 0); + atomic_set(&ret->nr_dispatched, 0); + spin_lock_init(&ret->lock); + ret->ttime_total = 0; + ret->ttime_samples = 0; + ret->ttime_mean = 0; + ret->seek_total = 0; + ret->seek_samples = 0; + ret->seek_mean = 0; + } + + return ret; +} + +/* + * If the current task has no AS IO context then create one and initialise it. + * Then take a ref on the task's io context and return it. + */ +static struct io_context *as_get_io_context(void) +{ + struct io_context *ioc = get_io_context(GFP_ATOMIC); + if (ioc && !ioc->aic) { + ioc->aic = alloc_as_io_context(); + if (!ioc->aic) { + put_io_context(ioc); + ioc = NULL; + } + } + return ioc; +} + +/* + * the back merge hash support functions + */ +static const int as_hash_shift = 6; +#define AS_HASH_BLOCK(sec) ((sec) >> 3) +#define AS_HASH_FN(sec) (hash_long(AS_HASH_BLOCK((sec)), as_hash_shift)) +#define AS_HASH_ENTRIES (1 << as_hash_shift) +#define rq_hash_key(rq) ((rq)->sector + (rq)->nr_sectors) +#define list_entry_hash(ptr) list_entry((ptr), struct as_rq, hash) + +static inline void __as_del_arq_hash(struct as_rq *arq) +{ + arq->on_hash = 0; + list_del_init(&arq->hash); +} + +static inline void as_del_arq_hash(struct as_rq *arq) +{ + if (arq->on_hash) + __as_del_arq_hash(arq); +} + +static void as_remove_merge_hints(request_queue_t *q, struct as_rq *arq) +{ + as_del_arq_hash(arq); + + if (q->last_merge == arq->request) + q->last_merge = NULL; +} + +static void as_add_arq_hash(struct as_data *ad, struct as_rq *arq) +{ + struct request *rq = arq->request; + + BUG_ON(arq->on_hash); + + arq->on_hash = 1; + list_add(&arq->hash, &ad->hash[AS_HASH_FN(rq_hash_key(rq))]); +} + +/* + * move hot entry to front of chain + */ +static inline void as_hot_arq_hash(struct as_data *ad, struct as_rq *arq) +{ + struct request *rq = arq->request; + struct list_head *head = &ad->hash[AS_HASH_FN(rq_hash_key(rq))]; + + if (!arq->on_hash) { + WARN_ON(1); + return; + } + + if (arq->hash.prev != head) { + list_del(&arq->hash); + list_add(&arq->hash, head); + } +} + +static struct request *as_find_arq_hash(struct as_data *ad, sector_t offset) +{ + struct list_head *hash_list = &ad->hash[AS_HASH_FN(offset)]; + struct list_head *entry, *next = hash_list->next; + + while ((entry = next) != hash_list) { + struct as_rq *arq = list_entry_hash(entry); + struct request *__rq = arq->request; + + next = entry->next; + + BUG_ON(!arq->on_hash); + + if (!rq_mergeable(__rq)) { + as_remove_merge_hints(ad->q, arq); + continue; + } + + if (rq_hash_key(__rq) == offset) + return __rq; + } + + return NULL; +} + +/* + * rb tree support functions + */ +#define RB_NONE (2) +#define RB_EMPTY(root) ((root)->rb_node == NULL) +#define ON_RB(node) ((node)->rb_color != RB_NONE) +#define RB_CLEAR(node) ((node)->rb_color = RB_NONE) +#define rb_entry_arq(node) rb_entry((node), struct as_rq, rb_node) +#define ARQ_RB_ROOT(ad, arq) (&(ad)->sort_list[(arq)->is_sync]) +#define rq_rb_key(rq) (rq)->sector + +/* + * as_find_first_arq finds the first (lowest sector numbered) request + * for the specified data_dir. Used to sweep back to the start of the disk + * (1-way elevator) after we process the last (highest sector) request. + */ +static struct as_rq *as_find_first_arq(struct as_data *ad, int data_dir) +{ + struct rb_node *n = ad->sort_list[data_dir].rb_node; + + if (n == NULL) + return NULL; + + for (;;) { + if (n->rb_left == NULL) + return rb_entry_arq(n); + + n = n->rb_left; + } +} + +/* + * Add the request to the rb tree if it is unique. If there is an alias (an + * existing request against the same sector), which can happen when using + * direct IO, then return the alias. + */ +static struct as_rq *as_add_arq_rb(struct as_data *ad, struct as_rq *arq) +{ + struct rb_node **p = &ARQ_RB_ROOT(ad, arq)->rb_node; + struct rb_node *parent = NULL; + struct as_rq *__arq; + struct request *rq = arq->request; + + arq->rb_key = rq_rb_key(rq); + + while (*p) { + parent = *p; + __arq = rb_entry_arq(parent); + + if (arq->rb_key < __arq->rb_key) + p = &(*p)->rb_left; + else if (arq->rb_key > __arq->rb_key) + p = &(*p)->rb_right; + else + return __arq; + } + + rb_link_node(&arq->rb_node, parent, p); + rb_insert_color(&arq->rb_node, ARQ_RB_ROOT(ad, arq)); + + return NULL; +} + +static inline void as_del_arq_rb(struct as_data *ad, struct as_rq *arq) +{ + if (!ON_RB(&arq->rb_node)) { + WARN_ON(1); + return; + } + + rb_erase(&arq->rb_node, ARQ_RB_ROOT(ad, arq)); + RB_CLEAR(&arq->rb_node); +} + +static struct request * +as_find_arq_rb(struct as_data *ad, sector_t sector, int data_dir) +{ + struct rb_node *n = ad->sort_list[data_dir].rb_node; + struct as_rq *arq; + + while (n) { + arq = rb_entry_arq(n); + + if (sector < arq->rb_key) + n = n->rb_left; + else if (sector > arq->rb_key) + n = n->rb_right; + else + return arq->request; + } + + return NULL; +} + +/* + * IO Scheduler proper + */ + +#define MAXBACK (1024 * 1024) /* + * Maximum distance the disk will go backward + * for a request. + */ + +#define BACK_PENALTY 2 + +/* + * as_choose_req selects the preferred one of two requests of the same data_dir + * ignoring time - eg. timeouts, which is the job of as_dispatch_request + */ +static struct as_rq * +as_choose_req(struct as_data *ad, struct as_rq *arq1, struct as_rq *arq2) +{ + int data_dir; + sector_t last, s1, s2, d1, d2; + int r1_wrap=0, r2_wrap=0; /* requests are behind the disk head */ + const sector_t maxback = MAXBACK; + + if (arq1 == NULL || arq1 == arq2) + return arq2; + if (arq2 == NULL) + return arq1; + + data_dir = arq1->is_sync; + + last = ad->last_sector[data_dir]; + s1 = arq1->request->sector; + s2 = arq2->request->sector; + + BUG_ON(data_dir != arq2->is_sync); + + /* + * Strict one way elevator _except_ in the case where we allow + * short backward seeks which are biased as twice the cost of a + * similar forward seek. + */ + if (s1 >= last) + d1 = s1 - last; + else if (s1+maxback >= last) + d1 = (last - s1)*BACK_PENALTY; + else { + r1_wrap = 1; + d1 = 0; /* shut up, gcc */ + } + + if (s2 >= last) + d2 = s2 - last; + else if (s2+maxback >= last) + d2 = (last - s2)*BACK_PENALTY; + else { + r2_wrap = 1; + d2 = 0; + } + + /* Found required data */ + if (!r1_wrap && r2_wrap) + return arq1; + else if (!r2_wrap && r1_wrap) + return arq2; + else if (r1_wrap && r2_wrap) { + /* both behind the head */ + if (s1 <= s2) + return arq1; + else + return arq2; + } + + /* Both requests in front of the head */ + if (d1 < d2) + return arq1; + else if (d2 < d1) + return arq2; + else { + if (s1 >= s2) + return arq1; + else + return arq2; + } +} + +/* + * as_find_next_arq finds the next request after @prev in elevator order. + * this with as_choose_req form the basis for how the scheduler chooses + * what request to process next. Anticipation works on top of this. + */ +static struct as_rq *as_find_next_arq(struct as_data *ad, struct as_rq *last) +{ + const int data_dir = last->is_sync; + struct as_rq *ret; + struct rb_node *rbnext = rb_next(&last->rb_node); + struct rb_node *rbprev = rb_prev(&last->rb_node); + struct as_rq *arq_next, *arq_prev; + + BUG_ON(!ON_RB(&last->rb_node)); + + if (rbprev) + arq_prev = rb_entry_arq(rbprev); + else + arq_prev = NULL; + + if (rbnext) + arq_next = rb_entry_arq(rbnext); + else { + arq_next = as_find_first_arq(ad, data_dir); + if (arq_next == last) + arq_next = NULL; + } + + ret = as_choose_req(ad, arq_next, arq_prev); + + return ret; +} + +/* + * anticipatory scheduling functions follow + */ + +/* + * as_antic_expired tells us when we have anticipated too long. + * The funny "absolute difference" math on the elapsed time is to handle + * jiffy wraps, and disks which have been idle for 0x80000000 jiffies. + */ +static int as_antic_expired(struct as_data *ad) +{ + long delta_jif; + + delta_jif = jiffies - ad->antic_start; + if (unlikely(delta_jif < 0)) + delta_jif = -delta_jif; + if (delta_jif < ad->antic_expire) + return 0; + + return 1; +} + +/* + * as_antic_waitnext starts anticipating that a nice request will soon be + * submitted. See also as_antic_waitreq + */ +static void as_antic_waitnext(struct as_data *ad) +{ + unsigned long timeout; + + BUG_ON(ad->antic_status != ANTIC_OFF + && ad->antic_status != ANTIC_WAIT_REQ); + + timeout = ad->antic_start + ad->antic_expire; + + mod_timer(&ad->antic_timer, timeout); + + ad->antic_status = ANTIC_WAIT_NEXT; +} + +/* + * as_antic_waitreq starts anticipating. We don't start timing the anticipation + * until the request that we're anticipating on has finished. This means we + * are timing from when the candidate process wakes up hopefully. + */ +static void as_antic_waitreq(struct as_data *ad) +{ + BUG_ON(ad->antic_status == ANTIC_FINISHED); + if (ad->antic_status == ANTIC_OFF) { + if (!ad->io_context || ad->ioc_finished) + as_antic_waitnext(ad); + else + ad->antic_status = ANTIC_WAIT_REQ; + } +} + +/* + * This is called directly by the functions in this file to stop anticipation. + * We kill the timer and schedule a call to the request_fn asap. + */ +static void as_antic_stop(struct as_data *ad) +{ + int status = ad->antic_status; + + if (status == ANTIC_WAIT_REQ || status == ANTIC_WAIT_NEXT) { + if (status == ANTIC_WAIT_NEXT) + del_timer(&ad->antic_timer); + ad->antic_status = ANTIC_FINISHED; + /* see as_work_handler */ + kblockd_schedule_work(&ad->antic_work); + } +} + +/* + * as_antic_timeout is the timer function set by as_antic_waitnext. + */ +static void as_antic_timeout(unsigned long data) +{ + struct request_queue *q = (struct request_queue *)data; + struct as_data *ad = q->elevator->elevator_data; + unsigned long flags; + + spin_lock_irqsave(q->queue_lock, flags); + if (ad->antic_status == ANTIC_WAIT_REQ + || ad->antic_status == ANTIC_WAIT_NEXT) { + struct as_io_context *aic = ad->io_context->aic; + + ad->antic_status = ANTIC_FINISHED; + kblockd_schedule_work(&ad->antic_work); + + if (aic->ttime_samples == 0) { + /* process anticipated on has exitted or timed out*/ + ad->exit_prob = (7*ad->exit_prob + 256)/8; + } + } + spin_unlock_irqrestore(q->queue_lock, flags); +} + +/* + * as_close_req decides if one request is considered "close" to the + * previous one issued. + */ +static int as_close_req(struct as_data *ad, struct as_rq *arq) +{ + unsigned long delay; /* milliseconds */ + sector_t last = ad->last_sector[ad->batch_data_dir]; + sector_t next = arq->request->sector; + sector_t delta; /* acceptable close offset (in sectors) */ + + if (ad->antic_status == ANTIC_OFF || !ad->ioc_finished) + delay = 0; + else + delay = ((jiffies - ad->antic_start) * 1000) / HZ; + + if (delay <= 1) + delta = 64; + else if (delay <= 20 && delay <= ad->antic_expire) + delta = 64 << (delay-1); + else + return 1; + + return (last - (delta>>1) <= next) && (next <= last + delta); +} + +/* + * as_can_break_anticipation returns true if we have been anticipating this + * request. + * + * It also returns true if the process against which we are anticipating + * submits a write - that's presumably an fsync, O_SYNC write, etc. We want to + * dispatch it ASAP, because we know that application will not be submitting + * any new reads. + * + * If the task which has submitted the request has exitted, break anticipation. + * + * If this task has queued some other IO, do not enter enticipation. + */ +static int as_can_break_anticipation(struct as_data *ad, struct as_rq *arq) +{ + struct io_context *ioc; + struct as_io_context *aic; + sector_t s; + + ioc = ad->io_context; + BUG_ON(!ioc); + + if (arq && ioc == arq->io_context) { + /* request from same process */ + return 1; + } + + if (ad->ioc_finished && as_antic_expired(ad)) { + /* + * In this situation status should really be FINISHED, + * however the timer hasn't had the chance to run yet. + */ + return 1; + } + + aic = ioc->aic; + if (!aic) + return 0; + + if (!test_bit(AS_TASK_RUNNING, &aic->state)) { + /* process anticipated on has exitted */ + if (aic->ttime_samples == 0) + ad->exit_prob = (7*ad->exit_prob + 256)/8; + return 1; + } + + if (atomic_read(&aic->nr_queued) > 0) { + /* process has more requests queued */ + return 1; + } + + if (atomic_read(&aic->nr_dispatched) > 0) { + /* process has more requests dispatched */ + return 1; + } + + if (arq && arq->is_sync == REQ_SYNC && as_close_req(ad, arq)) { + /* + * Found a close request that is not one of ours. + * + * This makes close requests from another process reset + * our thinktime delay. Is generally useful when there are + * two or more cooperating processes working in the same + * area. + */ + spin_lock(&aic->lock); + aic->last_end_request = jiffies; + spin_unlock(&aic->lock); + return 1; + } + + + if (aic->ttime_samples == 0) { + if (ad->new_ttime_mean > ad->antic_expire) + return 1; + if (ad->exit_prob > 128) + return 1; + } else if (aic->ttime_mean > ad->antic_expire) { + /* the process thinks too much between requests */ + return 1; + } + + if (!arq) + return 0; + + if (ad->last_sector[REQ_SYNC] < arq->request->sector) + s = arq->request->sector - ad->last_sector[REQ_SYNC]; + else + s = ad->last_sector[REQ_SYNC] - arq->request->sector; + + if (aic->seek_samples == 0) { + /* + * Process has just started IO. Use past statistics to + * guage success possibility + */ + if (ad->new_seek_mean > s) { + /* this request is better than what we're expecting */ + return 1; + } + + } else { + if (aic->seek_mean > s) { + /* this request is better than what we're expecting */ + return 1; + } + } + + return 0; +} + +/* + * as_can_anticipate indicates weather we should either run arq + * or keep anticipating a better request. + */ +static int as_can_anticipate(struct as_data *ad, struct as_rq *arq) +{ + if (!ad->io_context) + /* + * Last request submitted was a write + */ + return 0; + + if (ad->antic_status == ANTIC_FINISHED) + /* + * Don't restart if we have just finished. Run the next request + */ + return 0; + + if (as_can_break_anticipation(ad, arq)) + /* + * This request is a good candidate. Don't keep anticipating, + * run it. + */ + return 0; + + /* + * OK from here, we haven't finished, and don't have a decent request! + * Status is either ANTIC_OFF so start waiting, + * ANTIC_WAIT_REQ so continue waiting for request to finish + * or ANTIC_WAIT_NEXT so continue waiting for an acceptable request. + * + */ + + return 1; +} + +static void as_update_thinktime(struct as_data *ad, struct as_io_context *aic, unsigned long ttime) +{ + /* fixed point: 1.0 == 1<<8 */ + if (aic->ttime_samples == 0) { + ad->new_ttime_total = (7*ad->new_ttime_total + 256*ttime) / 8; + ad->new_ttime_mean = ad->new_ttime_total / 256; + + ad->exit_prob = (7*ad->exit_prob)/8; + } + aic->ttime_samples = (7*aic->ttime_samples + 256) / 8; + aic->ttime_total = (7*aic->ttime_total + 256*ttime) / 8; + aic->ttime_mean = (aic->ttime_total + 128) / aic->ttime_samples; +} + +static void as_update_seekdist(struct as_data *ad, struct as_io_context *aic, sector_t sdist) +{ + u64 total; + + if (aic->seek_samples == 0) { + ad->new_seek_total = (7*ad->new_seek_total + 256*(u64)sdist)/8; + ad->new_seek_mean = ad->new_seek_total / 256; + } + + /* + * Don't allow the seek distance to get too large from the + * odd fragment, pagein, etc + */ + if (aic->seek_samples <= 60) /* second&third seek */ + sdist = min(sdist, (aic->seek_mean * 4) + 2*1024*1024); + else + sdist = min(sdist, (aic->seek_mean * 4) + 2*1024*64); + + aic->seek_samples = (7*aic->seek_samples + 256) / 8; + aic->seek_total = (7*aic->seek_total + (u64)256*sdist) / 8; + total = aic->seek_total + (aic->seek_samples/2); + do_div(total, aic->seek_samples); + aic->seek_mean = (sector_t)total; +} + +/* + * as_update_iohist keeps a decaying histogram of IO thinktimes, and + * updates @aic->ttime_mean based on that. It is called when a new + * request is queued. + */ +static void as_update_iohist(struct as_data *ad, struct as_io_context *aic, struct request *rq) +{ + struct as_rq *arq = RQ_DATA(rq); + int data_dir = arq->is_sync; + unsigned long thinktime; + sector_t seek_dist; + + if (aic == NULL) + return; + + if (data_dir == REQ_SYNC) { + unsigned long in_flight = atomic_read(&aic->nr_queued) + + atomic_read(&aic->nr_dispatched); + spin_lock(&aic->lock); + if (test_bit(AS_TASK_IORUNNING, &aic->state) || + test_bit(AS_TASK_IOSTARTED, &aic->state)) { + /* Calculate read -> read thinktime */ + if (test_bit(AS_TASK_IORUNNING, &aic->state) + && in_flight == 0) { + thinktime = jiffies - aic->last_end_request; + thinktime = min(thinktime, MAX_THINKTIME-1); + } else + thinktime = 0; + as_update_thinktime(ad, aic, thinktime); + + /* Calculate read -> read seek distance */ + if (aic->last_request_pos < rq->sector) + seek_dist = rq->sector - aic->last_request_pos; + else + seek_dist = aic->last_request_pos - rq->sector; + as_update_seekdist(ad, aic, seek_dist); + } + aic->last_request_pos = rq->sector + rq->nr_sectors; + set_bit(AS_TASK_IOSTARTED, &aic->state); + spin_unlock(&aic->lock); + } +} + +/* + * as_update_arq must be called whenever a request (arq) is added to + * the sort_list. This function keeps caches up to date, and checks if the + * request might be one we are "anticipating" + */ +static void as_update_arq(struct as_data *ad, struct as_rq *arq) +{ + const int data_dir = arq->is_sync; + + /* keep the next_arq cache up to date */ + ad->next_arq[data_dir] = as_choose_req(ad, arq, ad->next_arq[data_dir]); + + /* + * have we been anticipating this request? + * or does it come from the same process as the one we are anticipating + * for? + */ + if (ad->antic_status == ANTIC_WAIT_REQ + || ad->antic_status == ANTIC_WAIT_NEXT) { + if (as_can_break_anticipation(ad, arq)) + as_antic_stop(ad); + } +} + +/* + * Gathers timings and resizes the write batch automatically + */ +static void update_write_batch(struct as_data *ad) +{ + unsigned long batch = ad->batch_expire[REQ_ASYNC]; + long write_time; + + write_time = (jiffies - ad->current_batch_expires) + batch; + if (write_time < 0) + write_time = 0; + + if (write_time > batch && !ad->write_batch_idled) { + if (write_time > batch * 3) + ad->write_batch_count /= 2; + else + ad->write_batch_count--; + } else if (write_time < batch && ad->current_write_count == 0) { + if (batch > write_time * 3) + ad->write_batch_count *= 2; + else + ad->write_batch_count++; + } + + if (ad->write_batch_count < 1) + ad->write_batch_count = 1; +} + +/* + * as_completed_request is to be called when a request has completed and + * returned something to the requesting process, be it an error or data. + */ +static void as_completed_request(request_queue_t *q, struct request *rq) +{ + struct as_data *ad = q->elevator->elevator_data; + struct as_rq *arq = RQ_DATA(rq); + + WARN_ON(!list_empty(&rq->queuelist)); + + if (arq->state == AS_RQ_PRESCHED) { + WARN_ON(arq->io_context); + goto out; + } + + if (arq->state == AS_RQ_MERGED) + goto out_ioc; + + if (arq->state != AS_RQ_REMOVED) { + printk("arq->state %d\n", arq->state); + WARN_ON(1); + goto out; + } + + if (!blk_fs_request(rq)) + goto out; + + if (ad->changed_batch && ad->nr_dispatched == 1) { + kblockd_schedule_work(&ad->antic_work); + ad->changed_batch = 0; + + if (ad->batch_data_dir == REQ_SYNC) + ad->new_batch = 1; + } + WARN_ON(ad->nr_dispatched == 0); + ad->nr_dispatched--; + + /* + * Start counting the batch from when a request of that direction is + * actually serviced. This should help devices with big TCQ windows + * and writeback caches + */ + if (ad->new_batch && ad->batch_data_dir == arq->is_sync) { + update_write_batch(ad); + ad->current_batch_expires = jiffies + + ad->batch_expire[REQ_SYNC]; + ad->new_batch = 0; + } + + if (ad->io_context == arq->io_context && ad->io_context) { + ad->antic_start = jiffies; + ad->ioc_finished = 1; + if (ad->antic_status == ANTIC_WAIT_REQ) { + /* + * We were waiting on this request, now anticipate + * the next one + */ + as_antic_waitnext(ad); + } + } + +out_ioc: + if (!arq->io_context) + goto out; + + if (arq->is_sync == REQ_SYNC) { + struct as_io_context *aic = arq->io_context->aic; + if (aic) { + spin_lock(&aic->lock); + set_bit(AS_TASK_IORUNNING, &aic->state); + aic->last_end_request = jiffies; + spin_unlock(&aic->lock); + } + } + + put_io_context(arq->io_context); +out: + arq->state = AS_RQ_POSTSCHED; +} + +/* + * as_remove_queued_request removes a request from the pre dispatch queue + * without updating refcounts. It is expected the caller will drop the + * reference unless it replaces the request at somepart of the elevator + * (ie. the dispatch queue) + */ +static void as_remove_queued_request(request_queue_t *q, struct request *rq) +{ + struct as_rq *arq = RQ_DATA(rq); + const int data_dir = arq->is_sync; + struct as_data *ad = q->elevator->elevator_data; + + WARN_ON(arq->state != AS_RQ_QUEUED); + + if (arq->io_context && arq->io_context->aic) { + BUG_ON(!atomic_read(&arq->io_context->aic->nr_queued)); + atomic_dec(&arq->io_context->aic->nr_queued); + } + + /* + * Update the "next_arq" cache if we are about to remove its + * entry + */ + if (ad->next_arq[data_dir] == arq) + ad->next_arq[data_dir] = as_find_next_arq(ad, arq); + + list_del_init(&arq->fifo); + as_remove_merge_hints(q, arq); + as_del_arq_rb(ad, arq); +} + +/* + * as_remove_dispatched_request is called to remove a request which has gone + * to the dispatch list. + */ +static void as_remove_dispatched_request(request_queue_t *q, struct request *rq) +{ + struct as_rq *arq = RQ_DATA(rq); + struct as_io_context *aic; + + if (!arq) { + WARN_ON(1); + return; + } + + WARN_ON(arq->state != AS_RQ_DISPATCHED); + WARN_ON(ON_RB(&arq->rb_node)); + if (arq->io_context && arq->io_context->aic) { + aic = arq->io_context->aic; + if (aic) { + WARN_ON(!atomic_read(&aic->nr_dispatched)); + atomic_dec(&aic->nr_dispatched); + } + } +} + +/* + * as_remove_request is called when a driver has finished with a request. + * This should be only called for dispatched requests, but for some reason + * a POWER4 box running hwscan it does not. + */ +static void as_remove_request(request_queue_t *q, struct request *rq) +{ + struct as_rq *arq = RQ_DATA(rq); + + if (unlikely(arq->state == AS_RQ_NEW)) + goto out; + + if (ON_RB(&arq->rb_node)) { + if (arq->state != AS_RQ_QUEUED) { + printk("arq->state %d\n", arq->state); + WARN_ON(1); + goto out; + } + /* + * We'll lose the aliased request(s) here. I don't think this + * will ever happen, but if it does, hopefully someone will + * report it. + */ + WARN_ON(!list_empty(&rq->queuelist)); + as_remove_queued_request(q, rq); + } else { + if (arq->state != AS_RQ_DISPATCHED) { + printk("arq->state %d\n", arq->state); + WARN_ON(1); + goto out; + } + as_remove_dispatched_request(q, rq); + } +out: + arq->state = AS_RQ_REMOVED; +} + +/* + * as_fifo_expired returns 0 if there are no expired reads on the fifo, + * 1 otherwise. It is ratelimited so that we only perform the check once per + * `fifo_expire' interval. Otherwise a large number of expired requests + * would create a hopeless seekstorm. + * + * See as_antic_expired comment. + */ +static int as_fifo_expired(struct as_data *ad, int adir) +{ + struct as_rq *arq; + long delta_jif; + + delta_jif = jiffies - ad->last_check_fifo[adir]; + if (unlikely(delta_jif < 0)) + delta_jif = -delta_jif; + if (delta_jif < ad->fifo_expire[adir]) + return 0; + + ad->last_check_fifo[adir] = jiffies; + + if (list_empty(&ad->fifo_list[adir])) + return 0; + + arq = list_entry_fifo(ad->fifo_list[adir].next); + + return time_after(jiffies, arq->expires); +} + +/* + * as_batch_expired returns true if the current batch has expired. A batch + * is a set of reads or a set of writes. + */ +static inline int as_batch_expired(struct as_data *ad) +{ + if (ad->changed_batch || ad->new_batch) + return 0; + + if (ad->batch_data_dir == REQ_SYNC) + /* TODO! add a check so a complete fifo gets written? */ + return time_after(jiffies, ad->current_batch_expires); + + return time_after(jiffies, ad->current_batch_expires) + || ad->current_write_count == 0; +} + +/* + * move an entry to dispatch queue + */ +static void as_move_to_dispatch(struct as_data *ad, struct as_rq *arq) +{ + struct request *rq = arq->request; + struct list_head *insert; + const int data_dir = arq->is_sync; + + BUG_ON(!ON_RB(&arq->rb_node)); + + as_antic_stop(ad); + ad->antic_status = ANTIC_OFF; + + /* + * This has to be set in order to be correctly updated by + * as_find_next_arq + */ + ad->last_sector[data_dir] = rq->sector + rq->nr_sectors; + + if (data_dir == REQ_SYNC) { + /* In case we have to anticipate after this */ + copy_io_context(&ad->io_context, &arq->io_context); + } else { + if (ad->io_context) { + put_io_context(ad->io_context); + ad->io_context = NULL; + } + + if (ad->current_write_count != 0) + ad->current_write_count--; + } + ad->ioc_finished = 0; + + ad->next_arq[data_dir] = as_find_next_arq(ad, arq); + + /* + * take it off the sort and fifo list, add to dispatch queue + */ + insert = ad->dispatch->prev; + + while (!list_empty(&rq->queuelist)) { + struct request *__rq = list_entry_rq(rq->queuelist.next); + struct as_rq *__arq = RQ_DATA(__rq); + + list_move_tail(&__rq->queuelist, ad->dispatch); + + if (__arq->io_context && __arq->io_context->aic) + atomic_inc(&__arq->io_context->aic->nr_dispatched); + + WARN_ON(__arq->state != AS_RQ_QUEUED); + __arq->state = AS_RQ_DISPATCHED; + + ad->nr_dispatched++; + } + + as_remove_queued_request(ad->q, rq); + WARN_ON(arq->state != AS_RQ_QUEUED); + + list_add(&rq->queuelist, insert); + arq->state = AS_RQ_DISPATCHED; + if (arq->io_context && arq->io_context->aic) + atomic_inc(&arq->io_context->aic->nr_dispatched); + ad->nr_dispatched++; +} + +/* + * as_dispatch_request selects the best request according to + * read/write expire, batch expire, etc, and moves it to the dispatch + * queue. Returns 1 if a request was found, 0 otherwise. + */ +static int as_dispatch_request(struct as_data *ad) +{ + struct as_rq *arq; + const int reads = !list_empty(&ad->fifo_list[REQ_SYNC]); + const int writes = !list_empty(&ad->fifo_list[REQ_ASYNC]); + + /* Signal that the write batch was uncontended, so we can't time it */ + if (ad->batch_data_dir == REQ_ASYNC && !reads) { + if (ad->current_write_count == 0 || !writes) + ad->write_batch_idled = 1; + } + + if (!(reads || writes) + || ad->antic_status == ANTIC_WAIT_REQ + || ad->antic_status == ANTIC_WAIT_NEXT + || ad->changed_batch) + return 0; + + if (!(reads && writes && as_batch_expired(ad)) ) { + /* + * batch is still running or no reads or no writes + */ + arq = ad->next_arq[ad->batch_data_dir]; + + if (ad->batch_data_dir == REQ_SYNC && ad->antic_expire) { + if (as_fifo_expired(ad, REQ_SYNC)) + goto fifo_expired; + + if (as_can_anticipate(ad, arq)) { + as_antic_waitreq(ad); + return 0; + } + } + + if (arq) { + /* we have a "next request" */ + if (reads && !writes) + ad->current_batch_expires = + jiffies + ad->batch_expire[REQ_SYNC]; + goto dispatch_request; + } + } + + /* + * at this point we are not running a batch. select the appropriate + * data direction (read / write) + */ + + if (reads) { + BUG_ON(RB_EMPTY(&ad->sort_list[REQ_SYNC])); + + if (writes && ad->batch_data_dir == REQ_SYNC) + /* + * Last batch was a read, switch to writes + */ + goto dispatch_writes; + + if (ad->batch_data_dir == REQ_ASYNC) { + WARN_ON(ad->new_batch); + ad->changed_batch = 1; + } + ad->batch_data_dir = REQ_SYNC; + arq = list_entry_fifo(ad->fifo_list[ad->batch_data_dir].next); + ad->last_check_fifo[ad->batch_data_dir] = jiffies; + goto dispatch_request; + } + + /* + * the last batch was a read + */ + + if (writes) { +dispatch_writes: + BUG_ON(RB_EMPTY(&ad->sort_list[REQ_ASYNC])); + + if (ad->batch_data_dir == REQ_SYNC) { + ad->changed_batch = 1; + + /* + * new_batch might be 1 when the queue runs out of + * reads. A subsequent submission of a write might + * cause a change of batch before the read is finished. + */ + ad->new_batch = 0; + } + ad->batch_data_dir = REQ_ASYNC; + ad->current_write_count = ad->write_batch_count; + ad->write_batch_idled = 0; + arq = ad->next_arq[ad->batch_data_dir]; + goto dispatch_request; + } + + BUG(); + return 0; + +dispatch_request: + /* + * If a request has expired, service it. + */ + + if (as_fifo_expired(ad, ad->batch_data_dir)) { +fifo_expired: + arq = list_entry_fifo(ad->fifo_list[ad->batch_data_dir].next); + BUG_ON(arq == NULL); + } + + if (ad->changed_batch) { + WARN_ON(ad->new_batch); + + if (ad->nr_dispatched) + return 0; + + if (ad->batch_data_dir == REQ_ASYNC) + ad->current_batch_expires = jiffies + + ad->batch_expire[REQ_ASYNC]; + else + ad->new_batch = 1; + + ad->changed_batch = 0; + } + + /* + * arq is the selected appropriate request. + */ + as_move_to_dispatch(ad, arq); + + return 1; +} + +static struct request *as_next_request(request_queue_t *q) +{ + struct as_data *ad = q->elevator->elevator_data; + struct request *rq = NULL; + + /* + * if there are still requests on the dispatch queue, grab the first + */ + if (!list_empty(ad->dispatch) || as_dispatch_request(ad)) + rq = list_entry_rq(ad->dispatch->next); + + return rq; +} + +/* + * Add arq to a list behind alias + */ +static inline void +as_add_aliased_request(struct as_data *ad, struct as_rq *arq, struct as_rq *alias) +{ + struct request *req = arq->request; + struct list_head *insert = alias->request->queuelist.prev; + + /* + * Transfer list of aliases + */ + while (!list_empty(&req->queuelist)) { + struct request *__rq = list_entry_rq(req->queuelist.next); + struct as_rq *__arq = RQ_DATA(__rq); + + list_move_tail(&__rq->queuelist, &alias->request->queuelist); + + WARN_ON(__arq->state != AS_RQ_QUEUED); + } + + /* + * Another request with the same start sector on the rbtree. + * Link this request to that sector. They are untangled in + * as_move_to_dispatch + */ + list_add(&arq->request->queuelist, insert); + + /* + * Don't want to have to handle merges. + */ + as_remove_merge_hints(ad->q, arq); +} + +/* + * add arq to rbtree and fifo + */ +static void as_add_request(struct as_data *ad, struct as_rq *arq) +{ + struct as_rq *alias; + int data_dir; + + if (rq_data_dir(arq->request) == READ + || current->flags&PF_SYNCWRITE) + arq->is_sync = 1; + else + arq->is_sync = 0; + data_dir = arq->is_sync; + + arq->io_context = as_get_io_context(); + + if (arq->io_context) { + as_update_iohist(ad, arq->io_context->aic, arq->request); + atomic_inc(&arq->io_context->aic->nr_queued); + } + + alias = as_add_arq_rb(ad, arq); + if (!alias) { + /* + * set expire time (only used for reads) and add to fifo list + */ + arq->expires = jiffies + ad->fifo_expire[data_dir]; + list_add_tail(&arq->fifo, &ad->fifo_list[data_dir]); + + if (rq_mergeable(arq->request)) { + as_add_arq_hash(ad, arq); + + if (!ad->q->last_merge) + ad->q->last_merge = arq->request; + } + as_update_arq(ad, arq); /* keep state machine up to date */ + + } else { + as_add_aliased_request(ad, arq, alias); + + /* + * have we been anticipating this request? + * or does it come from the same process as the one we are + * anticipating for? + */ + if (ad->antic_status == ANTIC_WAIT_REQ + || ad->antic_status == ANTIC_WAIT_NEXT) { + if (as_can_break_anticipation(ad, arq)) + as_antic_stop(ad); + } + } + + arq->state = AS_RQ_QUEUED; +} + +static void as_deactivate_request(request_queue_t *q, struct request *rq) +{ + struct as_data *ad = q->elevator->elevator_data; + struct as_rq *arq = RQ_DATA(rq); + + if (arq) { + if (arq->state == AS_RQ_REMOVED) { + arq->state = AS_RQ_DISPATCHED; + if (arq->io_context && arq->io_context->aic) + atomic_inc(&arq->io_context->aic->nr_dispatched); + } + } else + WARN_ON(blk_fs_request(rq) + && (!(rq->flags & (REQ_HARDBARRIER|REQ_SOFTBARRIER))) ); + + /* Stop anticipating - let this request get through */ + as_antic_stop(ad); +} + +/* + * requeue the request. The request has not been completed, nor is it a + * new request, so don't touch accounting. + */ +static void as_requeue_request(request_queue_t *q, struct request *rq) +{ + as_deactivate_request(q, rq); + list_add(&rq->queuelist, &q->queue_head); +} + +/* + * Account a request that is inserted directly onto the dispatch queue. + * arq->io_context->aic->nr_dispatched should not need to be incremented + * because only new requests should come through here: requeues go through + * our explicit requeue handler. + */ +static void as_account_queued_request(struct as_data *ad, struct request *rq) +{ + if (blk_fs_request(rq)) { + struct as_rq *arq = RQ_DATA(rq); + arq->state = AS_RQ_DISPATCHED; + ad->nr_dispatched++; + } +} + +static void +as_insert_request(request_queue_t *q, struct request *rq, int where) +{ + struct as_data *ad = q->elevator->elevator_data; + struct as_rq *arq = RQ_DATA(rq); + + if (arq) { + if (arq->state != AS_RQ_PRESCHED) { + printk("arq->state: %d\n", arq->state); + WARN_ON(1); + } + arq->state = AS_RQ_NEW; + } + + /* barriers must flush the reorder queue */ + if (unlikely(rq->flags & (REQ_SOFTBARRIER | REQ_HARDBARRIER) + && where == ELEVATOR_INSERT_SORT)) { + WARN_ON(1); + where = ELEVATOR_INSERT_BACK; + } + + switch (where) { + case ELEVATOR_INSERT_BACK: + while (ad->next_arq[REQ_SYNC]) + as_move_to_dispatch(ad, ad->next_arq[REQ_SYNC]); + + while (ad->next_arq[REQ_ASYNC]) + as_move_to_dispatch(ad, ad->next_arq[REQ_ASYNC]); + + list_add_tail(&rq->queuelist, ad->dispatch); + as_account_queued_request(ad, rq); + as_antic_stop(ad); + break; + case ELEVATOR_INSERT_FRONT: + list_add(&rq->queuelist, ad->dispatch); + as_account_queued_request(ad, rq); + as_antic_stop(ad); + break; + case ELEVATOR_INSERT_SORT: + BUG_ON(!blk_fs_request(rq)); + as_add_request(ad, arq); + break; + default: + BUG(); + return; + } +} + +/* + * as_queue_empty tells us if there are requests left in the device. It may + * not be the case that a driver can get the next request even if the queue + * is not empty - it is used in the block layer to check for plugging and + * merging opportunities + */ +static int as_queue_empty(request_queue_t *q) +{ + struct as_data *ad = q->elevator->elevator_data; + + if (!list_empty(&ad->fifo_list[REQ_ASYNC]) + || !list_empty(&ad->fifo_list[REQ_SYNC]) + || !list_empty(ad->dispatch)) + return 0; + + return 1; +} + +static struct request * +as_former_request(request_queue_t *q, struct request *rq) +{ + struct as_rq *arq = RQ_DATA(rq); + struct rb_node *rbprev = rb_prev(&arq->rb_node); + struct request *ret = NULL; + + if (rbprev) + ret = rb_entry_arq(rbprev)->request; + + return ret; +} + +static struct request * +as_latter_request(request_queue_t *q, struct request *rq) +{ + struct as_rq *arq = RQ_DATA(rq); + struct rb_node *rbnext = rb_next(&arq->rb_node); + struct request *ret = NULL; + + if (rbnext) + ret = rb_entry_arq(rbnext)->request; + + return ret; +} + +static int +as_merge(request_queue_t *q, struct request **req, struct bio *bio) +{ + struct as_data *ad = q->elevator->elevator_data; + sector_t rb_key = bio->bi_sector + bio_sectors(bio); + struct request *__rq; + int ret; + + /* + * try last_merge to avoid going to hash + */ + ret = elv_try_last_merge(q, bio); + if (ret != ELEVATOR_NO_MERGE) { + __rq = q->last_merge; + goto out_insert; + } + + /* + * see if the merge hash can satisfy a back merge + */ + __rq = as_find_arq_hash(ad, bio->bi_sector); + if (__rq) { + BUG_ON(__rq->sector + __rq->nr_sectors != bio->bi_sector); + + if (elv_rq_merge_ok(__rq, bio)) { + ret = ELEVATOR_BACK_MERGE; + goto out; + } + } + + /* + * check for front merge + */ + __rq = as_find_arq_rb(ad, rb_key, bio_data_dir(bio)); + if (__rq) { + BUG_ON(rb_key != rq_rb_key(__rq)); + + if (elv_rq_merge_ok(__rq, bio)) { + ret = ELEVATOR_FRONT_MERGE; + goto out; + } + } + + return ELEVATOR_NO_MERGE; +out: + if (rq_mergeable(__rq)) + q->last_merge = __rq; +out_insert: + if (ret) { + if (rq_mergeable(__rq)) + as_hot_arq_hash(ad, RQ_DATA(__rq)); + } + *req = __rq; + return ret; +} + +static void as_merged_request(request_queue_t *q, struct request *req) +{ + struct as_data *ad = q->elevator->elevator_data; + struct as_rq *arq = RQ_DATA(req); + + /* + * hash always needs to be repositioned, key is end sector + */ + as_del_arq_hash(arq); + as_add_arq_hash(ad, arq); + + /* + * if the merge was a front merge, we need to reposition request + */ + if (rq_rb_key(req) != arq->rb_key) { + struct as_rq *alias, *next_arq = NULL; + + if (ad->next_arq[arq->is_sync] == arq) + next_arq = as_find_next_arq(ad, arq); + + /* + * Note! We should really be moving any old aliased requests + * off this request and try to insert them into the rbtree. We + * currently don't bother. Ditto the next function. + */ + as_del_arq_rb(ad, arq); + if ((alias = as_add_arq_rb(ad, arq)) ) { + list_del_init(&arq->fifo); + as_add_aliased_request(ad, arq, alias); + if (next_arq) + ad->next_arq[arq->is_sync] = next_arq; + } + /* + * Note! At this stage of this and the next function, our next + * request may not be optimal - eg the request may have "grown" + * behind the disk head. We currently don't bother adjusting. + */ + } + + if (arq->on_hash) + q->last_merge = req; +} + +static void +as_merged_requests(request_queue_t *q, struct request *req, + struct request *next) +{ + struct as_data *ad = q->elevator->elevator_data; + struct as_rq *arq = RQ_DATA(req); + struct as_rq *anext = RQ_DATA(next); + + BUG_ON(!arq); + BUG_ON(!anext); + + /* + * reposition arq (this is the merged request) in hash, and in rbtree + * in case of a front merge + */ + as_del_arq_hash(arq); + as_add_arq_hash(ad, arq); + + if (rq_rb_key(req) != arq->rb_key) { + struct as_rq *alias, *next_arq = NULL; + + if (ad->next_arq[arq->is_sync] == arq) + next_arq = as_find_next_arq(ad, arq); + + as_del_arq_rb(ad, arq); + if ((alias = as_add_arq_rb(ad, arq)) ) { + list_del_init(&arq->fifo); + as_add_aliased_request(ad, arq, alias); + if (next_arq) + ad->next_arq[arq->is_sync] = next_arq; + } + } + + /* + * if anext expires before arq, assign its expire time to arq + * and move into anext position (anext will be deleted) in fifo + */ + if (!list_empty(&arq->fifo) && !list_empty(&anext->fifo)) { + if (time_before(anext->expires, arq->expires)) { + list_move(&arq->fifo, &anext->fifo); + arq->expires = anext->expires; + /* + * Don't copy here but swap, because when anext is + * removed below, it must contain the unused context + */ + swap_io_context(&arq->io_context, &anext->io_context); + } + } + + /* + * Transfer list of aliases + */ + while (!list_empty(&next->queuelist)) { + struct request *__rq = list_entry_rq(next->queuelist.next); + struct as_rq *__arq = RQ_DATA(__rq); + + list_move_tail(&__rq->queuelist, &req->queuelist); + + WARN_ON(__arq->state != AS_RQ_QUEUED); + } + + /* + * kill knowledge of next, this one is a goner + */ + as_remove_queued_request(q, next); + + anext->state = AS_RQ_MERGED; +} + +/* + * This is executed in a "deferred" process context, by kblockd. It calls the + * driver's request_fn so the driver can submit that request. + * + * IMPORTANT! This guy will reenter the elevator, so set up all queue global + * state before calling, and don't rely on any state over calls. + * + * FIXME! dispatch queue is not a queue at all! + */ +static void as_work_handler(void *data) +{ + struct request_queue *q = data; + unsigned long flags; + + spin_lock_irqsave(q->queue_lock, flags); + if (as_next_request(q)) + q->request_fn(q); + spin_unlock_irqrestore(q->queue_lock, flags); +} + +static void as_put_request(request_queue_t *q, struct request *rq) +{ + struct as_data *ad = q->elevator->elevator_data; + struct as_rq *arq = RQ_DATA(rq); + + if (!arq) { + WARN_ON(1); + return; + } + + if (arq->state != AS_RQ_POSTSCHED && arq->state != AS_RQ_PRESCHED) { + printk("arq->state %d\n", arq->state); + WARN_ON(1); + } + + mempool_free(arq, ad->arq_pool); + rq->elevator_private = NULL; +} + +static int as_set_request(request_queue_t *q, struct request *rq, int gfp_mask) +{ + struct as_data *ad = q->elevator->elevator_data; + struct as_rq *arq = mempool_alloc(ad->arq_pool, gfp_mask); + + if (arq) { + memset(arq, 0, sizeof(*arq)); + RB_CLEAR(&arq->rb_node); + arq->request = rq; + arq->state = AS_RQ_PRESCHED; + arq->io_context = NULL; + INIT_LIST_HEAD(&arq->hash); + arq->on_hash = 0; + INIT_LIST_HEAD(&arq->fifo); + rq->elevator_private = arq; + return 0; + } + + return 1; +} + +static int as_may_queue(request_queue_t *q, int rw) +{ + int ret = ELV_MQUEUE_MAY; + struct as_data *ad = q->elevator->elevator_data; + struct io_context *ioc; + if (ad->antic_status == ANTIC_WAIT_REQ || + ad->antic_status == ANTIC_WAIT_NEXT) { + ioc = as_get_io_context(); + if (ad->io_context == ioc) + ret = ELV_MQUEUE_MUST; + put_io_context(ioc); + } + + return ret; +} + +static void as_exit_queue(elevator_t *e) +{ + struct as_data *ad = e->elevator_data; + + del_timer_sync(&ad->antic_timer); + kblockd_flush(); + + BUG_ON(!list_empty(&ad->fifo_list[REQ_SYNC])); + BUG_ON(!list_empty(&ad->fifo_list[REQ_ASYNC])); + + mempool_destroy(ad->arq_pool); + put_io_context(ad->io_context); + kfree(ad->hash); + kfree(ad); +} + +/* + * initialize elevator private data (as_data), and alloc a arq for + * each request on the free lists + */ +static int as_init_queue(request_queue_t *q, elevator_t *e) +{ + struct as_data *ad; + int i; + + if (!arq_pool) + return -ENOMEM; + + ad = kmalloc(sizeof(*ad), GFP_KERNEL); + if (!ad) + return -ENOMEM; + memset(ad, 0, sizeof(*ad)); + + ad->q = q; /* Identify what queue the data belongs to */ + + ad->hash = kmalloc(sizeof(struct list_head)*AS_HASH_ENTRIES,GFP_KERNEL); + if (!ad->hash) { + kfree(ad); + return -ENOMEM; + } + + ad->arq_pool = mempool_create(BLKDEV_MIN_RQ, mempool_alloc_slab, mempool_free_slab, arq_pool); + if (!ad->arq_pool) { + kfree(ad->hash); + kfree(ad); + return -ENOMEM; + } + + /* anticipatory scheduling helpers */ + ad->antic_timer.function = as_antic_timeout; + ad->antic_timer.data = (unsigned long)q; + init_timer(&ad->antic_timer); + INIT_WORK(&ad->antic_work, as_work_handler, q); + + for (i = 0; i < AS_HASH_ENTRIES; i++) + INIT_LIST_HEAD(&ad->hash[i]); + + INIT_LIST_HEAD(&ad->fifo_list[REQ_SYNC]); + INIT_LIST_HEAD(&ad->fifo_list[REQ_ASYNC]); + ad->sort_list[REQ_SYNC] = RB_ROOT; + ad->sort_list[REQ_ASYNC] = RB_ROOT; + ad->dispatch = &q->queue_head; + ad->fifo_expire[REQ_SYNC] = default_read_expire; + ad->fifo_expire[REQ_ASYNC] = default_write_expire; + ad->antic_expire = default_antic_expire; + ad->batch_expire[REQ_SYNC] = default_read_batch_expire; + ad->batch_expire[REQ_ASYNC] = default_write_batch_expire; + e->elevator_data = ad; + + ad->current_batch_expires = jiffies + ad->batch_expire[REQ_SYNC]; + ad->write_batch_count = ad->batch_expire[REQ_ASYNC] / 10; + if (ad->write_batch_count < 2) + ad->write_batch_count = 2; + + return 0; +} + +/* + * sysfs parts below + */ +struct as_fs_entry { + struct attribute attr; + ssize_t (*show)(struct as_data *, char *); + ssize_t (*store)(struct as_data *, const char *, size_t); +}; + +static ssize_t +as_var_show(unsigned int var, char *page) +{ + var = (var * 1000) / HZ; + return sprintf(page, "%d\n", var); +} + +static ssize_t +as_var_store(unsigned long *var, const char *page, size_t count) +{ + unsigned long tmp; + char *p = (char *) page; + + tmp = simple_strtoul(p, &p, 10); + if (tmp != 0) { + tmp = (tmp * HZ) / 1000; + if (tmp == 0) + tmp = 1; + } + *var = tmp; + return count; +} + +static ssize_t as_est_show(struct as_data *ad, char *page) +{ + int pos = 0; + + pos += sprintf(page+pos, "%lu %% exit probability\n", 100*ad->exit_prob/256); + pos += sprintf(page+pos, "%lu ms new thinktime\n", ad->new_ttime_mean); + pos += sprintf(page+pos, "%llu sectors new seek distance\n", (unsigned long long)ad->new_seek_mean); + + return pos; +} + +#define SHOW_FUNCTION(__FUNC, __VAR) \ +static ssize_t __FUNC(struct as_data *ad, char *page) \ +{ \ + return as_var_show(jiffies_to_msecs((__VAR)), (page)); \ +} +SHOW_FUNCTION(as_readexpire_show, ad->fifo_expire[REQ_SYNC]); +SHOW_FUNCTION(as_writeexpire_show, ad->fifo_expire[REQ_ASYNC]); +SHOW_FUNCTION(as_anticexpire_show, ad->antic_expire); +SHOW_FUNCTION(as_read_batchexpire_show, ad->batch_expire[REQ_SYNC]); +SHOW_FUNCTION(as_write_batchexpire_show, ad->batch_expire[REQ_ASYNC]); +#undef SHOW_FUNCTION + +#define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX) \ +static ssize_t __FUNC(struct as_data *ad, const char *page, size_t count) \ +{ \ + int ret = as_var_store(__PTR, (page), count); \ + if (*(__PTR) < (MIN)) \ + *(__PTR) = (MIN); \ + else if (*(__PTR) > (MAX)) \ + *(__PTR) = (MAX); \ + *(__PTR) = msecs_to_jiffies(*(__PTR)); \ + return ret; \ +} +STORE_FUNCTION(as_readexpire_store, &ad->fifo_expire[REQ_SYNC], 0, INT_MAX); +STORE_FUNCTION(as_writeexpire_store, &ad->fifo_expire[REQ_ASYNC], 0, INT_MAX); +STORE_FUNCTION(as_anticexpire_store, &ad->antic_expire, 0, INT_MAX); +STORE_FUNCTION(as_read_batchexpire_store, + &ad->batch_expire[REQ_SYNC], 0, INT_MAX); +STORE_FUNCTION(as_write_batchexpire_store, + &ad->batch_expire[REQ_ASYNC], 0, INT_MAX); +#undef STORE_FUNCTION + +static struct as_fs_entry as_est_entry = { + .attr = {.name = "est_time", .mode = S_IRUGO }, + .show = as_est_show, +}; +static struct as_fs_entry as_readexpire_entry = { + .attr = {.name = "read_expire", .mode = S_IRUGO | S_IWUSR }, + .show = as_readexpire_show, + .store = as_readexpire_store, +}; +static struct as_fs_entry as_writeexpire_entry = { + .attr = {.name = "write_expire", .mode = S_IRUGO | S_IWUSR }, + .show = as_writeexpire_show, + .store = as_writeexpire_store, +}; +static struct as_fs_entry as_anticexpire_entry = { + .attr = {.name = "antic_expire", .mode = S_IRUGO | S_IWUSR }, + .show = as_anticexpire_show, + .store = as_anticexpire_store, +}; +static struct as_fs_entry as_read_batchexpire_entry = { + .attr = {.name = "read_batch_expire", .mode = S_IRUGO | S_IWUSR }, + .show = as_read_batchexpire_show, + .store = as_read_batchexpire_store, +}; +static struct as_fs_entry as_write_batchexpire_entry = { + .attr = {.name = "write_batch_expire", .mode = S_IRUGO | S_IWUSR }, + .show = as_write_batchexpire_show, + .store = as_write_batchexpire_store, +}; + +static struct attribute *default_attrs[] = { + &as_est_entry.attr, + &as_readexpire_entry.attr, + &as_writeexpire_entry.attr, + &as_anticexpire_entry.attr, + &as_read_batchexpire_entry.attr, + &as_write_batchexpire_entry.attr, + NULL, +}; + +#define to_as(atr) container_of((atr), struct as_fs_entry, attr) + +static ssize_t +as_attr_show(struct kobject *kobj, struct attribute *attr, char *page) +{ + elevator_t *e = container_of(kobj, elevator_t, kobj); + struct as_fs_entry *entry = to_as(attr); + + if (!entry->show) + return 0; + + return entry->show(e->elevator_data, page); +} + +static ssize_t +as_attr_store(struct kobject *kobj, struct attribute *attr, + const char *page, size_t length) +{ + elevator_t *e = container_of(kobj, elevator_t, kobj); + struct as_fs_entry *entry = to_as(attr); + + if (!entry->store) + return -EINVAL; + + return entry->store(e->elevator_data, page, length); +} + +static struct sysfs_ops as_sysfs_ops = { + .show = as_attr_show, + .store = as_attr_store, +}; + +static struct kobj_type as_ktype = { + .sysfs_ops = &as_sysfs_ops, + .default_attrs = default_attrs, +}; + +static struct elevator_type iosched_as = { + .ops = { + .elevator_merge_fn = as_merge, + .elevator_merged_fn = as_merged_request, + .elevator_merge_req_fn = as_merged_requests, + .elevator_next_req_fn = as_next_request, + .elevator_add_req_fn = as_insert_request, + .elevator_remove_req_fn = as_remove_request, + .elevator_requeue_req_fn = as_requeue_request, + .elevator_deactivate_req_fn = as_deactivate_request, + .elevator_queue_empty_fn = as_queue_empty, + .elevator_completed_req_fn = as_completed_request, + .elevator_former_req_fn = as_former_request, + .elevator_latter_req_fn = as_latter_request, + .elevator_set_req_fn = as_set_request, + .elevator_put_req_fn = as_put_request, + .elevator_may_queue_fn = as_may_queue, + .elevator_init_fn = as_init_queue, + .elevator_exit_fn = as_exit_queue, + }, + + .elevator_ktype = &as_ktype, + .elevator_name = "anticipatory", + .elevator_owner = THIS_MODULE, +}; + +static int __init as_init(void) +{ + int ret; + + arq_pool = kmem_cache_create("as_arq", sizeof(struct as_rq), + 0, 0, NULL, NULL); + if (!arq_pool) + return -ENOMEM; + + ret = elv_register(&iosched_as); + if (!ret) { + /* + * don't allow AS to get unregistered, since we would have + * to browse all tasks in the system and release their + * as_io_context first + */ + __module_get(THIS_MODULE); + return 0; + } + + kmem_cache_destroy(arq_pool); + return ret; +} + +static void __exit as_exit(void) +{ + kmem_cache_destroy(arq_pool); + elv_unregister(&iosched_as); +} + +module_init(as_init); +module_exit(as_exit); + +MODULE_AUTHOR("Nick Piggin"); +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("anticipatory IO scheduler"); diff --git a/drivers/block/ataflop.c b/drivers/block/ataflop.c new file mode 100644 index 00000000000..db05a5a99f3 --- /dev/null +++ b/drivers/block/ataflop.c @@ -0,0 +1,2006 @@ +/* + * drivers/block/ataflop.c + * + * Copyright (C) 1993 Greg Harp + * Atari Support by Bjoern Brauel, Roman Hodek + * + * Big cleanup Sep 11..14 1994 Roman Hodek: + * - Driver now works interrupt driven + * - Support for two drives; should work, but I cannot test that :-( + * - Reading is done in whole tracks and buffered to speed up things + * - Disk change detection and drive deselecting after motor-off + * similar to TOS + * - Autodetection of disk format (DD/HD); untested yet, because I + * don't have an HD drive :-( + * + * Fixes Nov 13 1994 Martin Schaller: + * - Autodetection works now + * - Support for 5 1/4'' disks + * - Removed drive type (unknown on atari) + * - Do seeks with 8 Mhz + * + * Changes by Andreas Schwab: + * - After errors in multiple read mode try again reading single sectors + * (Feb 1995): + * - Clean up error handling + * - Set blk_size for proper size checking + * - Initialize track register when testing presence of floppy + * - Implement some ioctl's + * + * Changes by Torsten Lang: + * - When probing the floppies we should add the FDCCMDADD_H flag since + * the FDC will otherwise wait forever when no disk is inserted... + * + * ++ Freddi Aschwanden (fa) 20.9.95 fixes for medusa: + * - MFPDELAY() after each FDC access -> atari + * - more/other disk formats + * - DMA to the block buffer directly if we have a 32bit DMA + * - for medusa, the step rate is always 3ms + * - on medusa, use only cache_push() + * Roman: + * - Make disk format numbering independent from minors + * - Let user set max. supported drive type (speeds up format + * detection, saves buffer space) + * + * Roman 10/15/95: + * - implement some more ioctls + * - disk formatting + * + * Andreas 95/12/12: + * - increase gap size at start of track for HD/ED disks + * + * Michael (MSch) 11/07/96: + * - implemented FDSETPRM and FDDEFPRM ioctl + * + * Andreas (97/03/19): + * - implemented missing BLK* ioctls + * + * Things left to do: + * - Formatting + * - Maybe a better strategy for disk change detection (does anyone + * know one?) + */ + +#include <linux/module.h> + +#include <linux/fd.h> +#include <linux/delay.h> +#include <linux/init.h> +#include <linux/blkdev.h> + +#include <asm/atafd.h> +#include <asm/atafdreg.h> +#include <asm/atariints.h> +#include <asm/atari_stdma.h> +#include <asm/atari_stram.h> + +#define FD_MAX_UNITS 2 + +#undef DEBUG + +static struct request_queue *floppy_queue; + +#define QUEUE (floppy_queue) +#define CURRENT elv_next_request(floppy_queue) + +/* Disk types: DD, HD, ED */ +static struct atari_disk_type { + const char *name; + unsigned spt; /* sectors per track */ + unsigned blocks; /* total number of blocks */ + unsigned fdc_speed; /* fdc_speed setting */ + unsigned stretch; /* track doubling ? */ +} disk_type[] = { + { "d360", 9, 720, 0, 0}, /* 0: 360kB diskette */ + { "D360", 9, 720, 0, 1}, /* 1: 360kb in 720k or 1.2MB drive */ + { "D720", 9,1440, 0, 0}, /* 2: 720kb in 720k or 1.2MB drive */ + { "D820", 10,1640, 0, 0}, /* 3: DD disk with 82 tracks/10 sectors */ +/* formats above are probed for type DD */ +#define MAX_TYPE_DD 3 + { "h1200",15,2400, 3, 0}, /* 4: 1.2MB diskette */ + { "H1440",18,2880, 3, 0}, /* 5: 1.4 MB diskette (HD) */ + { "H1640",20,3280, 3, 0}, /* 6: 1.64MB diskette (fat HD) 82 tr 20 sec */ +/* formats above are probed for types DD and HD */ +#define MAX_TYPE_HD 6 + { "E2880",36,5760, 3, 0}, /* 7: 2.8 MB diskette (ED) */ + { "E3280",40,6560, 3, 0}, /* 8: 3.2 MB diskette (fat ED) 82 tr 40 sec */ +/* formats above are probed for types DD, HD and ED */ +#define MAX_TYPE_ED 8 +/* types below are never autoprobed */ + { "H1680",21,3360, 3, 0}, /* 9: 1.68MB diskette (fat HD) 80 tr 21 sec */ + { "h410",10,820, 0, 1}, /* 10: 410k diskette 41 tr 10 sec, stretch */ + { "h1476",18,2952, 3, 0}, /* 11: 1.48MB diskette 82 tr 18 sec */ + { "H1722",21,3444, 3, 0}, /* 12: 1.72MB diskette 82 tr 21 sec */ + { "h420",10,840, 0, 1}, /* 13: 420k diskette 42 tr 10 sec, stretch */ + { "H830",10,1660, 0, 0}, /* 14: 820k diskette 83 tr 10 sec */ + { "h1494",18,2952, 3, 0}, /* 15: 1.49MB diskette 83 tr 18 sec */ + { "H1743",21,3486, 3, 0}, /* 16: 1.74MB diskette 83 tr 21 sec */ + { "h880",11,1760, 0, 0}, /* 17: 880k diskette 80 tr 11 sec */ + { "D1040",13,2080, 0, 0}, /* 18: 1.04MB diskette 80 tr 13 sec */ + { "D1120",14,2240, 0, 0}, /* 19: 1.12MB diskette 80 tr 14 sec */ + { "h1600",20,3200, 3, 0}, /* 20: 1.60MB diskette 80 tr 20 sec */ + { "H1760",22,3520, 3, 0}, /* 21: 1.76MB diskette 80 tr 22 sec */ + { "H1920",24,3840, 3, 0}, /* 22: 1.92MB diskette 80 tr 24 sec */ + { "E3200",40,6400, 3, 0}, /* 23: 3.2MB diskette 80 tr 40 sec */ + { "E3520",44,7040, 3, 0}, /* 24: 3.52MB diskette 80 tr 44 sec */ + { "E3840",48,7680, 3, 0}, /* 25: 3.84MB diskette 80 tr 48 sec */ + { "H1840",23,3680, 3, 0}, /* 26: 1.84MB diskette 80 tr 23 sec */ + { "D800",10,1600, 0, 0}, /* 27: 800k diskette 80 tr 10 sec */ +}; + +static int StartDiskType[] = { + MAX_TYPE_DD, + MAX_TYPE_HD, + MAX_TYPE_ED +}; + +#define TYPE_DD 0 +#define TYPE_HD 1 +#define TYPE_ED 2 + +static int DriveType = TYPE_HD; + +static DEFINE_SPINLOCK(ataflop_lock); + +/* Array for translating minors into disk formats */ +static struct { + int index; + unsigned drive_types; +} minor2disktype[] = { + { 0, TYPE_DD }, /* 1: d360 */ + { 4, TYPE_HD }, /* 2: h1200 */ + { 1, TYPE_DD }, /* 3: D360 */ + { 2, TYPE_DD }, /* 4: D720 */ + { 1, TYPE_DD }, /* 5: h360 = D360 */ + { 2, TYPE_DD }, /* 6: h720 = D720 */ + { 5, TYPE_HD }, /* 7: H1440 */ + { 7, TYPE_ED }, /* 8: E2880 */ +/* some PC formats :-) */ + { 8, TYPE_ED }, /* 9: E3280 <- was "CompaQ" == E2880 for PC */ + { 5, TYPE_HD }, /* 10: h1440 = H1440 */ + { 9, TYPE_HD }, /* 11: H1680 */ + { 10, TYPE_DD }, /* 12: h410 */ + { 3, TYPE_DD }, /* 13: H820 <- == D820, 82x10 */ + { 11, TYPE_HD }, /* 14: h1476 */ + { 12, TYPE_HD }, /* 15: H1722 */ + { 13, TYPE_DD }, /* 16: h420 */ + { 14, TYPE_DD }, /* 17: H830 */ + { 15, TYPE_HD }, /* 18: h1494 */ + { 16, TYPE_HD }, /* 19: H1743 */ + { 17, TYPE_DD }, /* 20: h880 */ + { 18, TYPE_DD }, /* 21: D1040 */ + { 19, TYPE_DD }, /* 22: D1120 */ + { 20, TYPE_HD }, /* 23: h1600 */ + { 21, TYPE_HD }, /* 24: H1760 */ + { 22, TYPE_HD }, /* 25: H1920 */ + { 23, TYPE_ED }, /* 26: E3200 */ + { 24, TYPE_ED }, /* 27: E3520 */ + { 25, TYPE_ED }, /* 28: E3840 */ + { 26, TYPE_HD }, /* 29: H1840 */ + { 27, TYPE_DD }, /* 30: D800 */ + { 6, TYPE_HD }, /* 31: H1640 <- was H1600 == h1600 for PC */ +}; + +#define NUM_DISK_MINORS (sizeof(minor2disktype)/sizeof(*minor2disktype)) + +/* + * Maximum disk size (in kilobytes). This default is used whenever the + * current disk size is unknown. + */ +#define MAX_DISK_SIZE 3280 + +/* + * MSch: User-provided type information. 'drive' points to + * the respective entry of this array. Set by FDSETPRM ioctls. + */ +static struct atari_disk_type user_params[FD_MAX_UNITS]; + +/* + * User-provided permanent type information. 'drive' points to + * the respective entry of this array. Set by FDDEFPRM ioctls, + * restored upon disk change by floppy_revalidate() if valid (as seen by + * default_params[].blocks > 0 - a bit in unit[].flags might be used for this?) + */ +static struct atari_disk_type default_params[FD_MAX_UNITS]; + +/* current info on each unit */ +static struct atari_floppy_struct { + int connected; /* !=0 : drive is connected */ + int autoprobe; /* !=0 : do autoprobe */ + + struct atari_disk_type *disktype; /* current type of disk */ + + int track; /* current head position or -1 if + unknown */ + unsigned int steprate; /* steprate setting */ + unsigned int wpstat; /* current state of WP signal (for + disk change detection) */ + int flags; /* flags */ + struct gendisk *disk; + int ref; + int type; +} unit[FD_MAX_UNITS]; + +#define UD unit[drive] +#define UDT unit[drive].disktype +#define SUD unit[SelectedDrive] +#define SUDT unit[SelectedDrive].disktype + + +#define FDC_READ(reg) ({ \ + /* unsigned long __flags; */ \ + unsigned short __val; \ + /* local_irq_save(__flags); */ \ + dma_wd.dma_mode_status = 0x80 | (reg); \ + udelay(25); \ + __val = dma_wd.fdc_acces_seccount; \ + MFPDELAY(); \ + /* local_irq_restore(__flags); */ \ + __val & 0xff; \ +}) + +#define FDC_WRITE(reg,val) \ + do { \ + /* unsigned long __flags; */ \ + /* local_irq_save(__flags); */ \ + dma_wd.dma_mode_status = 0x80 | (reg); \ + udelay(25); \ + dma_wd.fdc_acces_seccount = (val); \ + MFPDELAY(); \ + /* local_irq_restore(__flags); */ \ + } while(0) + + +/* Buffering variables: + * First, there is a DMA buffer in ST-RAM that is used for floppy DMA + * operations. Second, a track buffer is used to cache a whole track + * of the disk to save read operations. These are two separate buffers + * because that allows write operations without clearing the track buffer. + */ + +static int MaxSectors[] = { + 11, 22, 44 +}; +static int BufferSize[] = { + 15*512, 30*512, 60*512 +}; + +#define BUFFER_SIZE (BufferSize[DriveType]) + +unsigned char *DMABuffer; /* buffer for writes */ +static unsigned long PhysDMABuffer; /* physical address */ + +static int UseTrackbuffer = -1; /* Do track buffering? */ +MODULE_PARM(UseTrackbuffer, "i"); + +unsigned char *TrackBuffer; /* buffer for reads */ +static unsigned long PhysTrackBuffer; /* physical address */ +static int BufferDrive, BufferSide, BufferTrack; +static int read_track; /* non-zero if we are reading whole tracks */ + +#define SECTOR_BUFFER(sec) (TrackBuffer + ((sec)-1)*512) +#define IS_BUFFERED(drive,side,track) \ + (BufferDrive == (drive) && BufferSide == (side) && BufferTrack == (track)) + +/* + * These are global variables, as that's the easiest way to give + * information to interrupts. They are the data used for the current + * request. + */ +static int SelectedDrive = 0; +static int ReqCmd, ReqBlock; +static int ReqSide, ReqTrack, ReqSector, ReqCnt; +static int HeadSettleFlag = 0; +static unsigned char *ReqData, *ReqBuffer; +static int MotorOn = 0, MotorOffTrys; +static int IsFormatting = 0, FormatError; + +static int UserSteprate[FD_MAX_UNITS] = { -1, -1 }; +MODULE_PARM(UserSteprate, "1-" __MODULE_STRING(FD_MAX_UNITS) "i"); + +/* Synchronization of FDC access. */ +static volatile int fdc_busy = 0; +static DECLARE_WAIT_QUEUE_HEAD(fdc_wait); +static DECLARE_WAIT_QUEUE_HEAD(format_wait); + +static unsigned long changed_floppies = 0xff, fake_change = 0; +#define CHECK_CHANGE_DELAY HZ/2 + +#define FD_MOTOR_OFF_DELAY (3*HZ) +#define FD_MOTOR_OFF_MAXTRY (10*20) + +#define FLOPPY_TIMEOUT (6*HZ) +#define RECALIBRATE_ERRORS 4 /* After this many errors the drive + * will be recalibrated. */ +#define MAX_ERRORS 8 /* After this many errors the driver + * will give up. */ + + +/* + * The driver is trying to determine the correct media format + * while Probing is set. fd_rwsec_done() clears it after a + * successful access. + */ +static int Probing = 0; + +/* This flag is set when a dummy seek is necessary to make the WP + * status bit accessible. + */ +static int NeedSeek = 0; + + +#ifdef DEBUG +#define DPRINT(a) printk a +#else +#define DPRINT(a) +#endif + +/***************************** Prototypes *****************************/ + +static void fd_select_side( int side ); +static void fd_select_drive( int drive ); +static void fd_deselect( void ); +static void fd_motor_off_timer( unsigned long dummy ); +static void check_change( unsigned long dummy ); +static irqreturn_t floppy_irq (int irq, void *dummy, struct pt_regs *fp); +static void fd_error( void ); +static int do_format(int drive, int type, struct atari_format_descr *desc); +static void do_fd_action( int drive ); +static void fd_calibrate( void ); +static void fd_calibrate_done( int status ); +static void fd_seek( void ); +static void fd_seek_done( int status ); +static void fd_rwsec( void ); +static void fd_readtrack_check( unsigned long dummy ); +static void fd_rwsec_done( int status ); +static void fd_rwsec_done1(int status); +static void fd_writetrack( void ); +static void fd_writetrack_done( int status ); +static void fd_times_out( unsigned long dummy ); +static void finish_fdc( void ); +static void finish_fdc_done( int dummy ); +static void setup_req_params( int drive ); +static void redo_fd_request( void); +static int fd_ioctl( struct inode *inode, struct file *filp, unsigned int + cmd, unsigned long param); +static void fd_probe( int drive ); +static int fd_test_drive_present( int drive ); +static void config_types( void ); +static int floppy_open( struct inode *inode, struct file *filp ); +static int floppy_release( struct inode * inode, struct file * filp ); + +/************************* End of Prototypes **************************/ + +static struct timer_list motor_off_timer = + TIMER_INITIALIZER(fd_motor_off_timer, 0, 0); +static struct timer_list readtrack_timer = + TIMER_INITIALIZER(fd_readtrack_check, 0, 0); + +static struct timer_list timeout_timer = + TIMER_INITIALIZER(fd_times_out, 0, 0); + +static struct timer_list fd_timer = + TIMER_INITIALIZER(check_change, 0, 0); + +static inline void start_motor_off_timer(void) +{ + mod_timer(&motor_off_timer, jiffies + FD_MOTOR_OFF_DELAY); + MotorOffTrys = 0; +} + +static inline void start_check_change_timer( void ) +{ + mod_timer(&fd_timer, jiffies + CHECK_CHANGE_DELAY); +} + +static inline void start_timeout(void) +{ + mod_timer(&timeout_timer, jiffies + FLOPPY_TIMEOUT); +} + +static inline void stop_timeout(void) +{ + del_timer(&timeout_timer); +} + +/* Select the side to use. */ + +static void fd_select_side( int side ) +{ + unsigned long flags; + + /* protect against various other ints mucking around with the PSG */ + local_irq_save(flags); + + sound_ym.rd_data_reg_sel = 14; /* Select PSG Port A */ + sound_ym.wd_data = (side == 0) ? sound_ym.rd_data_reg_sel | 0x01 : + sound_ym.rd_data_reg_sel & 0xfe; + + local_irq_restore(flags); +} + + +/* Select a drive, update the FDC's track register and set the correct + * clock speed for this disk's type. + */ + +static void fd_select_drive( int drive ) +{ + unsigned long flags; + unsigned char tmp; + + if (drive == SelectedDrive) + return; + + /* protect against various other ints mucking around with the PSG */ + local_irq_save(flags); + sound_ym.rd_data_reg_sel = 14; /* Select PSG Port A */ + tmp = sound_ym.rd_data_reg_sel; + sound_ym.wd_data = (tmp | DSKDRVNONE) & ~(drive == 0 ? DSKDRV0 : DSKDRV1); + atari_dont_touch_floppy_select = 1; + local_irq_restore(flags); + + /* restore track register to saved value */ + FDC_WRITE( FDCREG_TRACK, UD.track ); + udelay(25); + + /* select 8/16 MHz */ + if (UDT) + if (ATARIHW_PRESENT(FDCSPEED)) + dma_wd.fdc_speed = UDT->fdc_speed; + + SelectedDrive = drive; +} + + +/* Deselect both drives. */ + +static void fd_deselect( void ) +{ + unsigned long flags; + + /* protect against various other ints mucking around with the PSG */ + local_irq_save(flags); + atari_dont_touch_floppy_select = 0; + sound_ym.rd_data_reg_sel=14; /* Select PSG Port A */ + sound_ym.wd_data = (sound_ym.rd_data_reg_sel | + (MACH_IS_FALCON ? 3 : 7)); /* no drives selected */ + /* On Falcon, the drive B select line is used on the printer port, so + * leave it alone... */ + SelectedDrive = -1; + local_irq_restore(flags); +} + + +/* This timer function deselects the drives when the FDC switched the + * motor off. The deselection cannot happen earlier because the FDC + * counts the index signals, which arrive only if one drive is selected. + */ + +static void fd_motor_off_timer( unsigned long dummy ) +{ + unsigned char status; + + if (SelectedDrive < 0) + /* no drive selected, needn't deselect anyone */ + return; + + if (stdma_islocked()) + goto retry; + + status = FDC_READ( FDCREG_STATUS ); + + if (!(status & 0x80)) { + /* motor already turned off by FDC -> deselect drives */ + MotorOn = 0; + fd_deselect(); + return; + } + /* not yet off, try again */ + + retry: + /* Test again later; if tested too often, it seems there is no disk + * in the drive and the FDC will leave the motor on forever (or, + * at least until a disk is inserted). So we'll test only twice + * per second from then on... + */ + mod_timer(&motor_off_timer, + jiffies + (MotorOffTrys++ < FD_MOTOR_OFF_MAXTRY ? HZ/20 : HZ/2)); +} + + +/* This function is repeatedly called to detect disk changes (as good + * as possible) and keep track of the current state of the write protection. + */ + +static void check_change( unsigned long dummy ) +{ + static int drive = 0; + + unsigned long flags; + unsigned char old_porta; + int stat; + + if (++drive > 1 || !UD.connected) + drive = 0; + + /* protect against various other ints mucking around with the PSG */ + local_irq_save(flags); + + if (!stdma_islocked()) { + sound_ym.rd_data_reg_sel = 14; + old_porta = sound_ym.rd_data_reg_sel; + sound_ym.wd_data = (old_porta | DSKDRVNONE) & + ~(drive == 0 ? DSKDRV0 : DSKDRV1); + stat = !!(FDC_READ( FDCREG_STATUS ) & FDCSTAT_WPROT); + sound_ym.wd_data = old_porta; + + if (stat != UD.wpstat) { + DPRINT(( "wpstat[%d] = %d\n", drive, stat )); + UD.wpstat = stat; + set_bit (drive, &changed_floppies); + } + } + local_irq_restore(flags); + + start_check_change_timer(); +} + + +/* Handling of the Head Settling Flag: This flag should be set after each + * seek operation, because we don't use seeks with verify. + */ + +static inline void set_head_settle_flag(void) +{ + HeadSettleFlag = FDCCMDADD_E; +} + +static inline int get_head_settle_flag(void) +{ + int tmp = HeadSettleFlag; + HeadSettleFlag = 0; + return( tmp ); +} + +static inline void copy_buffer(void *from, void *to) +{ + ulong *p1 = (ulong *)from, *p2 = (ulong *)to; + int cnt; + + for (cnt = 512/4; cnt; cnt--) + *p2++ = *p1++; +} + + + + +/* General Interrupt Handling */ + +static void (*FloppyIRQHandler)( int status ) = NULL; + +static irqreturn_t floppy_irq (int irq, void *dummy, struct pt_regs *fp) +{ + unsigned char status; + void (*handler)( int ); + + handler = xchg(&FloppyIRQHandler, NULL); + + if (handler) { + nop(); + status = FDC_READ( FDCREG_STATUS ); + DPRINT(("FDC irq, status = %02x handler = %08lx\n",status,(unsigned long)handler)); + handler( status ); + } + else { + DPRINT(("FDC irq, no handler\n")); + } + return IRQ_HANDLED; +} + + +/* Error handling: If some error happened, retry some times, then + * recalibrate, then try again, and fail after MAX_ERRORS. + */ + +static void fd_error( void ) +{ + if (IsFormatting) { + IsFormatting = 0; + FormatError = 1; + wake_up( &format_wait ); + return; + } + + if (!CURRENT) + return; + + CURRENT->errors++; + if (CURRENT->errors >= MAX_ERRORS) { + printk(KERN_ERR "fd%d: too many errors.\n", SelectedDrive ); + end_request(CURRENT, 0); + } + else if (CURRENT->errors == RECALIBRATE_ERRORS) { + printk(KERN_WARNING "fd%d: recalibrating\n", SelectedDrive ); + if (SelectedDrive != -1) + SUD.track = -1; + } + redo_fd_request(); +} + + + +#define SET_IRQ_HANDLER(proc) do { FloppyIRQHandler = (proc); } while(0) + + +/* ---------- Formatting ---------- */ + +#define FILL(n,val) \ + do { \ + memset( p, val, n ); \ + p += n; \ + } while(0) + +static int do_format(int drive, int type, struct atari_format_descr *desc) +{ + unsigned char *p; + int sect, nsect; + unsigned long flags; + + DPRINT(("do_format( dr=%d tr=%d he=%d offs=%d )\n", + drive, desc->track, desc->head, desc->sect_offset )); + + local_irq_save(flags); + while( fdc_busy ) sleep_on( &fdc_wait ); + fdc_busy = 1; + stdma_lock(floppy_irq, NULL); + atari_turnon_irq( IRQ_MFP_FDC ); /* should be already, just to be sure */ + local_irq_restore(flags); + + if (type) { + if (--type >= NUM_DISK_MINORS || + minor2disktype[type].drive_types > DriveType) { + redo_fd_request(); + return -EINVAL; + } + type = minor2disktype[type].index; + UDT = &disk_type[type]; + } + + if (!UDT || desc->track >= UDT->blocks/UDT->spt/2 || desc->head >= 2) { + redo_fd_request(); + return -EINVAL; + } + + nsect = UDT->spt; + p = TrackBuffer; + /* The track buffer is used for the raw track data, so its + contents become invalid! */ + BufferDrive = -1; + /* stop deselect timer */ + del_timer( &motor_off_timer ); + + FILL( 60 * (nsect / 9), 0x4e ); + for( sect = 0; sect < nsect; ++sect ) { + FILL( 12, 0 ); + FILL( 3, 0xf5 ); + *p++ = 0xfe; + *p++ = desc->track; + *p++ = desc->head; + *p++ = (nsect + sect - desc->sect_offset) % nsect + 1; + *p++ = 2; + *p++ = 0xf7; + FILL( 22, 0x4e ); + FILL( 12, 0 ); + FILL( 3, 0xf5 ); + *p++ = 0xfb; + FILL( 512, 0xe5 ); + *p++ = 0xf7; + FILL( 40, 0x4e ); + } + FILL( TrackBuffer+BUFFER_SIZE-p, 0x4e ); + + IsFormatting = 1; + FormatError = 0; + ReqTrack = desc->track; + ReqSide = desc->head; + do_fd_action( drive ); + + sleep_on( &format_wait ); + + redo_fd_request(); + return( FormatError ? -EIO : 0 ); +} + + +/* do_fd_action() is the general procedure for a fd request: All + * required parameter settings (drive select, side select, track + * position) are checked and set if needed. For each of these + * parameters and the actual reading or writing exist two functions: + * one that starts the setting (or skips it if possible) and one + * callback for the "done" interrupt. Each done func calls the next + * set function to propagate the request down to fd_rwsec_done(). + */ + +static void do_fd_action( int drive ) +{ + DPRINT(("do_fd_action\n")); + + if (UseTrackbuffer && !IsFormatting) { + repeat: + if (IS_BUFFERED( drive, ReqSide, ReqTrack )) { + if (ReqCmd == READ) { + copy_buffer( SECTOR_BUFFER(ReqSector), ReqData ); + if (++ReqCnt < CURRENT->current_nr_sectors) { + /* read next sector */ + setup_req_params( drive ); + goto repeat; + } + else { + /* all sectors finished */ + CURRENT->nr_sectors -= CURRENT->current_nr_sectors; + CURRENT->sector += CURRENT->current_nr_sectors; + end_request(CURRENT, 1); + redo_fd_request(); + return; + } + } + else { + /* cmd == WRITE, pay attention to track buffer + * consistency! */ + copy_buffer( ReqData, SECTOR_BUFFER(ReqSector) ); + } + } + } + + if (SelectedDrive != drive) + fd_select_drive( drive ); + + if (UD.track == -1) + fd_calibrate(); + else if (UD.track != ReqTrack << UDT->stretch) + fd_seek(); + else if (IsFormatting) + fd_writetrack(); + else + fd_rwsec(); +} + + +/* Seek to track 0 if the current track is unknown */ + +static void fd_calibrate( void ) +{ + if (SUD.track >= 0) { + fd_calibrate_done( 0 ); + return; + } + + if (ATARIHW_PRESENT(FDCSPEED)) + dma_wd.fdc_speed = 0; /* always seek with 8 Mhz */; + DPRINT(("fd_calibrate\n")); + SET_IRQ_HANDLER( fd_calibrate_done ); + /* we can't verify, since the speed may be incorrect */ + FDC_WRITE( FDCREG_CMD, FDCCMD_RESTORE | SUD.steprate ); + + NeedSeek = 1; + MotorOn = 1; + start_timeout(); + /* wait for IRQ */ +} + + +static void fd_calibrate_done( int status ) +{ + DPRINT(("fd_calibrate_done()\n")); + stop_timeout(); + + /* set the correct speed now */ + if (ATARIHW_PRESENT(FDCSPEED)) + dma_wd.fdc_speed = SUDT->fdc_speed; + if (status & FDCSTAT_RECNF) { + printk(KERN_ERR "fd%d: restore failed\n", SelectedDrive ); + fd_error(); + } + else { + SUD.track = 0; + fd_seek(); + } +} + + +/* Seek the drive to the requested track. The drive must have been + * calibrated at some point before this. + */ + +static void fd_seek( void ) +{ + if (SUD.track == ReqTrack << SUDT->stretch) { + fd_seek_done( 0 ); + return; + } + + if (ATARIHW_PRESENT(FDCSPEED)) { + dma_wd.fdc_speed = 0; /* always seek witch 8 Mhz */ + MFPDELAY(); + } + + DPRINT(("fd_seek() to track %d\n",ReqTrack)); + FDC_WRITE( FDCREG_DATA, ReqTrack << SUDT->stretch); + udelay(25); + SET_IRQ_HANDLER( fd_seek_done ); + FDC_WRITE( FDCREG_CMD, FDCCMD_SEEK | SUD.steprate ); + + MotorOn = 1; + set_head_settle_flag(); + start_timeout(); + /* wait for IRQ */ +} + + +static void fd_seek_done( int status ) +{ + DPRINT(("fd_seek_done()\n")); + stop_timeout(); + + /* set the correct speed */ + if (ATARIHW_PRESENT(FDCSPEED)) + dma_wd.fdc_speed = SUDT->fdc_speed; + if (status & FDCSTAT_RECNF) { + printk(KERN_ERR "fd%d: seek error (to track %d)\n", + SelectedDrive, ReqTrack ); + /* we don't know exactly which track we are on now! */ + SUD.track = -1; + fd_error(); + } + else { + SUD.track = ReqTrack << SUDT->stretch; + NeedSeek = 0; + if (IsFormatting) + fd_writetrack(); + else + fd_rwsec(); + } +} + + +/* This does the actual reading/writing after positioning the head + * over the correct track. + */ + +static int MultReadInProgress = 0; + + +static void fd_rwsec( void ) +{ + unsigned long paddr, flags; + unsigned int rwflag, old_motoron; + unsigned int track; + + DPRINT(("fd_rwsec(), Sec=%d, Access=%c\n",ReqSector, ReqCmd == WRITE ? 'w' : 'r' )); + if (ReqCmd == WRITE) { + if (ATARIHW_PRESENT(EXTD_DMA)) { + paddr = virt_to_phys(ReqData); + } + else { + copy_buffer( ReqData, DMABuffer ); + paddr = PhysDMABuffer; + } + dma_cache_maintenance( paddr, 512, 1 ); + rwflag = 0x100; + } + else { + if (read_track) + paddr = PhysTrackBuffer; + else + paddr = ATARIHW_PRESENT(EXTD_DMA) ? + virt_to_phys(ReqData) : PhysDMABuffer; + rwflag = 0; + } + + fd_select_side( ReqSide ); + + /* Start sector of this operation */ + FDC_WRITE( FDCREG_SECTOR, read_track ? 1 : ReqSector ); + MFPDELAY(); + /* Cheat for track if stretch != 0 */ + if (SUDT->stretch) { + track = FDC_READ( FDCREG_TRACK); + MFPDELAY(); + FDC_WRITE( FDCREG_TRACK, track >> SUDT->stretch); + } + udelay(25); + + /* Setup DMA */ + local_irq_save(flags); + dma_wd.dma_lo = (unsigned char)paddr; + MFPDELAY(); + paddr >>= 8; + dma_wd.dma_md = (unsigned char)paddr; + MFPDELAY(); + paddr >>= 8; + if (ATARIHW_PRESENT(EXTD_DMA)) + st_dma_ext_dmahi = (unsigned short)paddr; + else + dma_wd.dma_hi = (unsigned char)paddr; + MFPDELAY(); + local_irq_restore(flags); + + /* Clear FIFO and switch DMA to correct mode */ + dma_wd.dma_mode_status = 0x90 | rwflag; + MFPDELAY(); + dma_wd.dma_mode_status = 0x90 | (rwflag ^ 0x100); + MFPDELAY(); + dma_wd.dma_mode_status = 0x90 | rwflag; + MFPDELAY(); + + /* How many sectors for DMA */ + dma_wd.fdc_acces_seccount = read_track ? SUDT->spt : 1; + + udelay(25); + + /* Start operation */ + dma_wd.dma_mode_status = FDCSELREG_STP | rwflag; + udelay(25); + SET_IRQ_HANDLER( fd_rwsec_done ); + dma_wd.fdc_acces_seccount = + (get_head_settle_flag() | + (rwflag ? FDCCMD_WRSEC : (FDCCMD_RDSEC | (read_track ? FDCCMDADD_M : 0)))); + + old_motoron = MotorOn; + MotorOn = 1; + NeedSeek = 1; + /* wait for interrupt */ + + if (read_track) { + /* If reading a whole track, wait about one disk rotation and + * then check if all sectors are read. The FDC will even + * search for the first non-existent sector and need 1 sec to + * recognise that it isn't present :-( + */ + MultReadInProgress = 1; + mod_timer(&readtrack_timer, + /* 1 rot. + 5 rot.s if motor was off */ + jiffies + HZ/5 + (old_motoron ? 0 : HZ)); + } + start_timeout(); +} + + +static void fd_readtrack_check( unsigned long dummy ) +{ + unsigned long flags, addr, addr2; + + local_irq_save(flags); + + if (!MultReadInProgress) { + /* This prevents a race condition that could arise if the + * interrupt is triggered while the calling of this timer + * callback function takes place. The IRQ function then has + * already cleared 'MultReadInProgress' when flow of control + * gets here. + */ + local_irq_restore(flags); + return; + } + + /* get the current DMA address */ + /* ++ f.a. read twice to avoid being fooled by switcher */ + addr = 0; + do { + addr2 = addr; + addr = dma_wd.dma_lo & 0xff; + MFPDELAY(); + addr |= (dma_wd.dma_md & 0xff) << 8; + MFPDELAY(); + if (ATARIHW_PRESENT( EXTD_DMA )) + addr |= (st_dma_ext_dmahi & 0xffff) << 16; + else + addr |= (dma_wd.dma_hi & 0xff) << 16; + MFPDELAY(); + } while(addr != addr2); + + if (addr >= PhysTrackBuffer + SUDT->spt*512) { + /* already read enough data, force an FDC interrupt to stop + * the read operation + */ + SET_IRQ_HANDLER( NULL ); + MultReadInProgress = 0; + local_irq_restore(flags); + DPRINT(("fd_readtrack_check(): done\n")); + FDC_WRITE( FDCREG_CMD, FDCCMD_FORCI ); + udelay(25); + + /* No error until now -- the FDC would have interrupted + * otherwise! + */ + fd_rwsec_done1(0); + } + else { + /* not yet finished, wait another tenth rotation */ + local_irq_restore(flags); + DPRINT(("fd_readtrack_check(): not yet finished\n")); + mod_timer(&readtrack_timer, jiffies + HZ/5/10); + } +} + + +static void fd_rwsec_done( int status ) +{ + DPRINT(("fd_rwsec_done()\n")); + + if (read_track) { + del_timer(&readtrack_timer); + if (!MultReadInProgress) + return; + MultReadInProgress = 0; + } + fd_rwsec_done1(status); +} + +static void fd_rwsec_done1(int status) +{ + unsigned int track; + + stop_timeout(); + + /* Correct the track if stretch != 0 */ + if (SUDT->stretch) { + track = FDC_READ( FDCREG_TRACK); + MFPDELAY(); + FDC_WRITE( FDCREG_TRACK, track << SUDT->stretch); + } + + if (!UseTrackbuffer) { + dma_wd.dma_mode_status = 0x90; + MFPDELAY(); + if (!(dma_wd.dma_mode_status & 0x01)) { + printk(KERN_ERR "fd%d: DMA error\n", SelectedDrive ); + goto err_end; + } + } + MFPDELAY(); + + if (ReqCmd == WRITE && (status & FDCSTAT_WPROT)) { + printk(KERN_NOTICE "fd%d: is write protected\n", SelectedDrive ); + goto err_end; + } + if ((status & FDCSTAT_RECNF) && + /* RECNF is no error after a multiple read when the FDC + searched for a non-existent sector! */ + !(read_track && FDC_READ(FDCREG_SECTOR) > SUDT->spt)) { + if (Probing) { + if (SUDT > disk_type) { + if (SUDT[-1].blocks > ReqBlock) { + /* try another disk type */ + SUDT--; + set_capacity(unit[SelectedDrive].disk, + SUDT->blocks); + } else + Probing = 0; + } + else { + if (SUD.flags & FTD_MSG) + printk(KERN_INFO "fd%d: Auto-detected floppy type %s\n", + SelectedDrive, SUDT->name ); + Probing=0; + } + } else { +/* record not found, but not probing. Maybe stretch wrong ? Restart probing */ + if (SUD.autoprobe) { + SUDT = disk_type + StartDiskType[DriveType]; + set_capacity(unit[SelectedDrive].disk, + SUDT->blocks); + Probing = 1; + } + } + if (Probing) { + if (ATARIHW_PRESENT(FDCSPEED)) { + dma_wd.fdc_speed = SUDT->fdc_speed; + MFPDELAY(); + } + setup_req_params( SelectedDrive ); + BufferDrive = -1; + do_fd_action( SelectedDrive ); + return; + } + + printk(KERN_ERR "fd%d: sector %d not found (side %d, track %d)\n", + SelectedDrive, FDC_READ (FDCREG_SECTOR), ReqSide, ReqTrack ); + goto err_end; + } + if (status & FDCSTAT_CRC) { + printk(KERN_ERR "fd%d: CRC error (side %d, track %d, sector %d)\n", + SelectedDrive, ReqSide, ReqTrack, FDC_READ (FDCREG_SECTOR) ); + goto err_end; + } + if (status & FDCSTAT_LOST) { + printk(KERN_ERR "fd%d: lost data (side %d, track %d, sector %d)\n", + SelectedDrive, ReqSide, ReqTrack, FDC_READ (FDCREG_SECTOR) ); + goto err_end; + } + + Probing = 0; + + if (ReqCmd == READ) { + if (!read_track) { + void *addr; + addr = ATARIHW_PRESENT( EXTD_DMA ) ? ReqData : DMABuffer; + dma_cache_maintenance( virt_to_phys(addr), 512, 0 ); + if (!ATARIHW_PRESENT( EXTD_DMA )) + copy_buffer (addr, ReqData); + } else { + dma_cache_maintenance( PhysTrackBuffer, MaxSectors[DriveType] * 512, 0 ); + BufferDrive = SelectedDrive; + BufferSide = ReqSide; + BufferTrack = ReqTrack; + copy_buffer (SECTOR_BUFFER (ReqSector), ReqData); + } + } + + if (++ReqCnt < CURRENT->current_nr_sectors) { + /* read next sector */ + setup_req_params( SelectedDrive ); + do_fd_action( SelectedDrive ); + } + else { + /* all sectors finished */ + CURRENT->nr_sectors -= CURRENT->current_nr_sectors; + CURRENT->sector += CURRENT->current_nr_sectors; + end_request(CURRENT, 1); + redo_fd_request(); + } + return; + + err_end: + BufferDrive = -1; + fd_error(); +} + + +static void fd_writetrack( void ) +{ + unsigned long paddr, flags; + unsigned int track; + + DPRINT(("fd_writetrack() Tr=%d Si=%d\n", ReqTrack, ReqSide )); + + paddr = PhysTrackBuffer; + dma_cache_maintenance( paddr, BUFFER_SIZE, 1 ); + + fd_select_side( ReqSide ); + + /* Cheat for track if stretch != 0 */ + if (SUDT->stretch) { + track = FDC_READ( FDCREG_TRACK); + MFPDELAY(); + FDC_WRITE(FDCREG_TRACK,track >> SUDT->stretch); + } + udelay(40); + + /* Setup DMA */ + local_irq_save(flags); + dma_wd.dma_lo = (unsigned char)paddr; + MFPDELAY(); + paddr >>= 8; + dma_wd.dma_md = (unsigned char)paddr; + MFPDELAY(); + paddr >>= 8; + if (ATARIHW_PRESENT( EXTD_DMA )) + st_dma_ext_dmahi = (unsigned short)paddr; + else + dma_wd.dma_hi = (unsigned char)paddr; + MFPDELAY(); + local_irq_restore(flags); + + /* Clear FIFO and switch DMA to correct mode */ + dma_wd.dma_mode_status = 0x190; + MFPDELAY(); + dma_wd.dma_mode_status = 0x90; + MFPDELAY(); + dma_wd.dma_mode_status = 0x190; + MFPDELAY(); + + /* How many sectors for DMA */ + dma_wd.fdc_acces_seccount = BUFFER_SIZE/512; + udelay(40); + + /* Start operation */ + dma_wd.dma_mode_status = FDCSELREG_STP | 0x100; + udelay(40); + SET_IRQ_HANDLER( fd_writetrack_done ); + dma_wd.fdc_acces_seccount = FDCCMD_WRTRA | get_head_settle_flag(); + + MotorOn = 1; + start_timeout(); + /* wait for interrupt */ +} + + +static void fd_writetrack_done( int status ) +{ + DPRINT(("fd_writetrack_done()\n")); + + stop_timeout(); + + if (status & FDCSTAT_WPROT) { + printk(KERN_NOTICE "fd%d: is write protected\n", SelectedDrive ); + goto err_end; + } + if (status & FDCSTAT_LOST) { + printk(KERN_ERR "fd%d: lost data (side %d, track %d)\n", + SelectedDrive, ReqSide, ReqTrack ); + goto err_end; + } + + wake_up( &format_wait ); + return; + + err_end: + fd_error(); +} + +static void fd_times_out( unsigned long dummy ) +{ + atari_disable_irq( IRQ_MFP_FDC ); + if (!FloppyIRQHandler) goto end; /* int occurred after timer was fired, but + * before we came here... */ + + SET_IRQ_HANDLER( NULL ); + /* If the timeout occurred while the readtrack_check timer was + * active, we need to cancel it, else bad things will happen */ + if (UseTrackbuffer) + del_timer( &readtrack_timer ); + FDC_WRITE( FDCREG_CMD, FDCCMD_FORCI ); + udelay( 25 ); + + printk(KERN_ERR "floppy timeout\n" ); + fd_error(); + end: + atari_enable_irq( IRQ_MFP_FDC ); +} + + +/* The (noop) seek operation here is needed to make the WP bit in the + * FDC status register accessible for check_change. If the last disk + * operation would have been a RDSEC, this bit would always read as 0 + * no matter what :-( To save time, the seek goes to the track we're + * already on. + */ + +static void finish_fdc( void ) +{ + if (!NeedSeek) { + finish_fdc_done( 0 ); + } + else { + DPRINT(("finish_fdc: dummy seek started\n")); + FDC_WRITE (FDCREG_DATA, SUD.track); + SET_IRQ_HANDLER( finish_fdc_done ); + FDC_WRITE (FDCREG_CMD, FDCCMD_SEEK); + MotorOn = 1; + start_timeout(); + /* we must wait for the IRQ here, because the ST-DMA + is released immediately afterwards and the interrupt + may be delivered to the wrong driver. */ + } +} + + +static void finish_fdc_done( int dummy ) +{ + unsigned long flags; + + DPRINT(("finish_fdc_done entered\n")); + stop_timeout(); + NeedSeek = 0; + + if (timer_pending(&fd_timer) && time_before(fd_timer.expires, jiffies + 5)) + /* If the check for a disk change is done too early after this + * last seek command, the WP bit still reads wrong :-(( + */ + mod_timer(&fd_timer, jiffies + 5); + else + start_check_change_timer(); + start_motor_off_timer(); + + local_irq_save(flags); + stdma_release(); + fdc_busy = 0; + wake_up( &fdc_wait ); + local_irq_restore(flags); + + DPRINT(("finish_fdc() finished\n")); +} + +/* The detection of disk changes is a dark chapter in Atari history :-( + * Because the "Drive ready" signal isn't present in the Atari + * hardware, one has to rely on the "Write Protect". This works fine, + * as long as no write protected disks are used. TOS solves this + * problem by introducing tri-state logic ("maybe changed") and + * looking at the serial number in block 0. This isn't possible for + * Linux, since the floppy driver can't make assumptions about the + * filesystem used on the disk and thus the contents of block 0. I've + * chosen the method to always say "The disk was changed" if it is + * unsure whether it was. This implies that every open or mount + * invalidates the disk buffers if you work with write protected + * disks. But at least this is better than working with incorrect data + * due to unrecognised disk changes. + */ + +static int check_floppy_change(struct gendisk *disk) +{ + struct atari_floppy_struct *p = disk->private_data; + unsigned int drive = p - unit; + if (test_bit (drive, &fake_change)) { + /* simulated change (e.g. after formatting) */ + return 1; + } + if (test_bit (drive, &changed_floppies)) { + /* surely changed (the WP signal changed at least once) */ + return 1; + } + if (UD.wpstat) { + /* WP is on -> could be changed: to be sure, buffers should be + * invalidated... + */ + return 1; + } + + return 0; +} + +static int floppy_revalidate(struct gendisk *disk) +{ + struct atari_floppy_struct *p = disk->private_data; + unsigned int drive = p - unit; + + if (test_bit(drive, &changed_floppies) || + test_bit(drive, &fake_change) || + p->disktype == 0) { + if (UD.flags & FTD_MSG) + printk(KERN_ERR "floppy: clear format %p!\n", UDT); + BufferDrive = -1; + clear_bit(drive, &fake_change); + clear_bit(drive, &changed_floppies); + /* MSch: clearing geometry makes sense only for autoprobe + formats, for 'permanent user-defined' parameter: + restore default_params[] here if flagged valid! */ + if (default_params[drive].blocks == 0) + UDT = 0; + else + UDT = &default_params[drive]; + } + return 0; +} + + +/* This sets up the global variables describing the current request. */ + +static void setup_req_params( int drive ) +{ + int block = ReqBlock + ReqCnt; + + ReqTrack = block / UDT->spt; + ReqSector = block - ReqTrack * UDT->spt + 1; + ReqSide = ReqTrack & 1; + ReqTrack >>= 1; + ReqData = ReqBuffer + 512 * ReqCnt; + + if (UseTrackbuffer) + read_track = (ReqCmd == READ && CURRENT->errors == 0); + else + read_track = 0; + + DPRINT(("Request params: Si=%d Tr=%d Se=%d Data=%08lx\n",ReqSide, + ReqTrack, ReqSector, (unsigned long)ReqData )); +} + + +static void redo_fd_request(void) +{ + int drive, type; + struct atari_floppy_struct *floppy; + + DPRINT(("redo_fd_request: CURRENT=%p dev=%s CURRENT->sector=%ld\n", + CURRENT, CURRENT ? CURRENT->rq_disk->disk_name : "", + CURRENT ? CURRENT->sector : 0 )); + + IsFormatting = 0; + +repeat: + + if (!CURRENT) + goto the_end; + + floppy = CURRENT->rq_disk->private_data; + drive = floppy - unit; + type = floppy->type; + + if (!UD.connected) { + /* drive not connected */ + printk(KERN_ERR "Unknown Device: fd%d\n", drive ); + end_request(CURRENT, 0); + goto repeat; + } + + if (type == 0) { + if (!UDT) { + Probing = 1; + UDT = disk_type + StartDiskType[DriveType]; + set_capacity(floppy->disk, UDT->blocks); + UD.autoprobe = 1; + } + } + else { + /* user supplied disk type */ + if (--type >= NUM_DISK_MINORS) { + printk(KERN_WARNING "fd%d: invalid disk format", drive ); + end_request(CURRENT, 0); + goto repeat; + } + if (minor2disktype[type].drive_types > DriveType) { + printk(KERN_WARNING "fd%d: unsupported disk format", drive ); + end_request(CURRENT, 0); + goto repeat; + } + type = minor2disktype[type].index; + UDT = &disk_type[type]; + set_capacity(floppy->disk, UDT->blocks); + UD.autoprobe = 0; + } + + if (CURRENT->sector + 1 > UDT->blocks) { + end_request(CURRENT, 0); + goto repeat; + } + + /* stop deselect timer */ + del_timer( &motor_off_timer ); + + ReqCnt = 0; + ReqCmd = rq_data_dir(CURRENT); + ReqBlock = CURRENT->sector; + ReqBuffer = CURRENT->buffer; + setup_req_params( drive ); + do_fd_action( drive ); + + return; + + the_end: + finish_fdc(); +} + + +void do_fd_request(request_queue_t * q) +{ + unsigned long flags; + + DPRINT(("do_fd_request for pid %d\n",current->pid)); + while( fdc_busy ) sleep_on( &fdc_wait ); + fdc_busy = 1; + stdma_lock(floppy_irq, NULL); + + atari_disable_irq( IRQ_MFP_FDC ); + local_save_flags(flags); /* The request function is called with ints + local_irq_disable(); * disabled... so must save the IPL for later */ + redo_fd_request(); + local_irq_restore(flags); + atari_enable_irq( IRQ_MFP_FDC ); +} + +static int fd_ioctl(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long param) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + struct atari_floppy_struct *floppy = disk->private_data; + int drive = floppy - unit; + int type = floppy->type; + struct atari_format_descr fmt_desc; + struct atari_disk_type *dtp; + struct floppy_struct getprm; + int settype; + struct floppy_struct setprm; + + switch (cmd) { + case FDGETPRM: + if (type) { + if (--type >= NUM_DISK_MINORS) + return -ENODEV; + if (minor2disktype[type].drive_types > DriveType) + return -ENODEV; + type = minor2disktype[type].index; + dtp = &disk_type[type]; + if (UD.flags & FTD_MSG) + printk (KERN_ERR "floppy%d: found dtp %p name %s!\n", + drive, dtp, dtp->name); + } + else { + if (!UDT) + return -ENXIO; + else + dtp = UDT; + } + memset((void *)&getprm, 0, sizeof(getprm)); + getprm.size = dtp->blocks; + getprm.sect = dtp->spt; + getprm.head = 2; + getprm.track = dtp->blocks/dtp->spt/2; + getprm.stretch = dtp->stretch; + if (copy_to_user((void *)param, &getprm, sizeof(getprm))) + return -EFAULT; + return 0; + } + switch (cmd) { + case FDSETPRM: + case FDDEFPRM: + /* + * MSch 7/96: simple 'set geometry' case: just set the + * 'default' device params (minor == 0). + * Currently, the drive geometry is cleared after each + * disk change and subsequent revalidate()! simple + * implementation of FDDEFPRM: save geometry from a + * FDDEFPRM call and restore it in floppy_revalidate() ! + */ + + /* get the parameters from user space */ + if (floppy->ref != 1 && floppy->ref != -1) + return -EBUSY; + if (copy_from_user(&setprm, (void *) param, sizeof(setprm))) + return -EFAULT; + /* + * first of all: check for floppy change and revalidate, + * or the next access will revalidate - and clear UDT :-( + */ + + if (check_floppy_change(disk)) + floppy_revalidate(disk); + + if (UD.flags & FTD_MSG) + printk (KERN_INFO "floppy%d: setting size %d spt %d str %d!\n", + drive, setprm.size, setprm.sect, setprm.stretch); + + /* what if type > 0 here? Overwrite specified entry ? */ + if (type) { + /* refuse to re-set a predefined type for now */ + redo_fd_request(); + return -EINVAL; + } + + /* + * type == 0: first look for a matching entry in the type list, + * and set the UD.disktype field to use the perdefined entry. + * TODO: add user-defined format to head of autoprobe list ? + * Useful to include the user-type for future autodetection! + */ + + for (settype = 0; settype < NUM_DISK_MINORS; settype++) { + int setidx = 0; + if (minor2disktype[settype].drive_types > DriveType) { + /* skip this one, invalid for drive ... */ + continue; + } + setidx = minor2disktype[settype].index; + dtp = &disk_type[setidx]; + + /* found matching entry ?? */ + if ( dtp->blocks == setprm.size + && dtp->spt == setprm.sect + && dtp->stretch == setprm.stretch ) { + if (UD.flags & FTD_MSG) + printk (KERN_INFO "floppy%d: setting %s %p!\n", + drive, dtp->name, dtp); + UDT = dtp; + set_capacity(floppy->disk, UDT->blocks); + + if (cmd == FDDEFPRM) { + /* save settings as permanent default type */ + default_params[drive].name = dtp->name; + default_params[drive].spt = dtp->spt; + default_params[drive].blocks = dtp->blocks; + default_params[drive].fdc_speed = dtp->fdc_speed; + default_params[drive].stretch = dtp->stretch; + } + + return 0; + } + + } + + /* no matching disk type found above - setting user_params */ + + if (cmd == FDDEFPRM) { + /* set permanent type */ + dtp = &default_params[drive]; + } else + /* set user type (reset by disk change!) */ + dtp = &user_params[drive]; + + dtp->name = "user format"; + dtp->blocks = setprm.size; + dtp->spt = setprm.sect; + if (setprm.sect > 14) + dtp->fdc_speed = 3; + else + dtp->fdc_speed = 0; + dtp->stretch = setprm.stretch; + + if (UD.flags & FTD_MSG) + printk (KERN_INFO "floppy%d: blk %d spt %d str %d!\n", + drive, dtp->blocks, dtp->spt, dtp->stretch); + + /* sanity check */ + if (!dtp || setprm.track != dtp->blocks/dtp->spt/2 || + setprm.head != 2) { + redo_fd_request(); + return -EINVAL; + } + + UDT = dtp; + set_capacity(floppy->disk, UDT->blocks); + + return 0; + case FDMSGON: + UD.flags |= FTD_MSG; + return 0; + case FDMSGOFF: + UD.flags &= ~FTD_MSG; + return 0; + case FDSETEMSGTRESH: + return -EINVAL; + case FDFMTBEG: + return 0; + case FDFMTTRK: + if (floppy->ref != 1 && floppy->ref != -1) + return -EBUSY; + if (copy_from_user(&fmt_desc, (void *) param, sizeof(fmt_desc))) + return -EFAULT; + return do_format(drive, type, &fmt_desc); + case FDCLRPRM: + UDT = NULL; + /* MSch: invalidate default_params */ + default_params[drive].blocks = 0; + set_capacity(floppy->disk, MAX_DISK_SIZE * 2); + case FDFMTEND: + case FDFLUSH: + /* invalidate the buffer track to force a reread */ + BufferDrive = -1; + set_bit(drive, &fake_change); + check_disk_change(inode->i_bdev); + return 0; + default: + return -EINVAL; + } +} + + +/* Initialize the 'unit' variable for drive 'drive' */ + +static void __init fd_probe( int drive ) +{ + UD.connected = 0; + UDT = NULL; + + if (!fd_test_drive_present( drive )) + return; + + UD.connected = 1; + UD.track = 0; + switch( UserSteprate[drive] ) { + case 2: + UD.steprate = FDCSTEP_2; + break; + case 3: + UD.steprate = FDCSTEP_3; + break; + case 6: + UD.steprate = FDCSTEP_6; + break; + case 12: + UD.steprate = FDCSTEP_12; + break; + default: /* should be -1 for "not set by user" */ + if (ATARIHW_PRESENT( FDCSPEED ) || MACH_IS_MEDUSA) + UD.steprate = FDCSTEP_3; + else + UD.steprate = FDCSTEP_6; + break; + } + MotorOn = 1; /* from probe restore operation! */ +} + + +/* This function tests the physical presence of a floppy drive (not + * whether a disk is inserted). This is done by issuing a restore + * command, waiting max. 2 seconds (that should be enough to move the + * head across the whole disk) and looking at the state of the "TR00" + * signal. This should now be raised if there is a drive connected + * (and there is no hardware failure :-) Otherwise, the drive is + * declared absent. + */ + +static int __init fd_test_drive_present( int drive ) +{ + unsigned long timeout; + unsigned char status; + int ok; + + if (drive >= (MACH_IS_FALCON ? 1 : 2)) return( 0 ); + fd_select_drive( drive ); + + /* disable interrupt temporarily */ + atari_turnoff_irq( IRQ_MFP_FDC ); + FDC_WRITE (FDCREG_TRACK, 0xff00); + FDC_WRITE( FDCREG_CMD, FDCCMD_RESTORE | FDCCMDADD_H | FDCSTEP_6 ); + + timeout = jiffies + 2*HZ+HZ/2; + while (time_before(jiffies, timeout)) + if (!(mfp.par_dt_reg & 0x20)) + break; + + status = FDC_READ( FDCREG_STATUS ); + ok = (status & FDCSTAT_TR00) != 0; + + /* force interrupt to abort restore operation (FDC would try + * about 50 seconds!) */ + FDC_WRITE( FDCREG_CMD, FDCCMD_FORCI ); + udelay(500); + status = FDC_READ( FDCREG_STATUS ); + udelay(20); + + if (ok) { + /* dummy seek command to make WP bit accessible */ + FDC_WRITE( FDCREG_DATA, 0 ); + FDC_WRITE( FDCREG_CMD, FDCCMD_SEEK ); + while( mfp.par_dt_reg & 0x20 ) + ; + status = FDC_READ( FDCREG_STATUS ); + } + + atari_turnon_irq( IRQ_MFP_FDC ); + return( ok ); +} + + +/* Look how many and which kind of drives are connected. If there are + * floppies, additionally start the disk-change and motor-off timers. + */ + +static void __init config_types( void ) +{ + int drive, cnt = 0; + + /* for probing drives, set the FDC speed to 8 MHz */ + if (ATARIHW_PRESENT(FDCSPEED)) + dma_wd.fdc_speed = 0; + + printk(KERN_INFO "Probing floppy drive(s):\n"); + for( drive = 0; drive < FD_MAX_UNITS; drive++ ) { + fd_probe( drive ); + if (UD.connected) { + printk(KERN_INFO "fd%d\n", drive); + ++cnt; + } + } + + if (FDC_READ( FDCREG_STATUS ) & FDCSTAT_BUSY) { + /* If FDC is still busy from probing, give it another FORCI + * command to abort the operation. If this isn't done, the FDC + * will interrupt later and its IRQ line stays low, because + * the status register isn't read. And this will block any + * interrupts on this IRQ line :-( + */ + FDC_WRITE( FDCREG_CMD, FDCCMD_FORCI ); + udelay(500); + FDC_READ( FDCREG_STATUS ); + udelay(20); + } + + if (cnt > 0) { + start_motor_off_timer(); + if (cnt == 1) fd_select_drive( 0 ); + start_check_change_timer(); + } +} + +/* + * floppy_open check for aliasing (/dev/fd0 can be the same as + * /dev/PS0 etc), and disallows simultaneous access to the same + * drive with different device numbers. + */ + +static int floppy_open( struct inode *inode, struct file *filp ) +{ + struct atari_floppy_struct *p = inode->i_bdev->bd_disk->private_data; + int type = iminor(inode) >> 2; + + DPRINT(("fd_open: type=%d\n",type)); + if (p->ref && p->type != type) + return -EBUSY; + + if (p->ref == -1 || (p->ref && filp->f_flags & O_EXCL)) + return -EBUSY; + + if (filp->f_flags & O_EXCL) + p->ref = -1; + else + p->ref++; + + p->type = type; + + if (filp->f_flags & O_NDELAY) + return 0; + + if (filp->f_mode & 3) { + check_disk_change(inode->i_bdev); + if (filp->f_mode & 2) { + if (p->wpstat) { + if (p->ref < 0) + p->ref = 0; + else + p->ref--; + floppy_release(inode, filp); + return -EROFS; + } + } + } + return 0; +} + + +static int floppy_release( struct inode * inode, struct file * filp ) +{ + struct atari_floppy_struct *p = inode->i_bdev->bd_disk->private_data; + if (p->ref < 0) + p->ref = 0; + else if (!p->ref--) { + printk(KERN_ERR "floppy_release with fd_ref == 0"); + p->ref = 0; + } + return 0; +} + +static struct block_device_operations floppy_fops = { + .owner = THIS_MODULE, + .open = floppy_open, + .release = floppy_release, + .ioctl = fd_ioctl, + .media_changed = check_floppy_change, + .revalidate_disk= floppy_revalidate, +}; + +static struct kobject *floppy_find(dev_t dev, int *part, void *data) +{ + int drive = *part & 3; + int type = *part >> 2; + if (drive >= FD_MAX_UNITS || type > NUM_DISK_MINORS) + return NULL; + *part = 0; + return get_disk(unit[drive].disk); +} + +static int __init atari_floppy_init (void) +{ + int i; + + if (!MACH_IS_ATARI) + /* Amiga, Mac, ... don't have Atari-compatible floppy :-) */ + return -ENXIO; + + if (MACH_IS_HADES) + /* Hades doesn't have Atari-compatible floppy */ + return -ENXIO; + + if (register_blkdev(FLOPPY_MAJOR,"fd")) + return -EBUSY; + + for (i = 0; i < FD_MAX_UNITS; i++) { + unit[i].disk = alloc_disk(1); + if (!unit[i].disk) + goto Enomem; + } + + if (UseTrackbuffer < 0) + /* not set by user -> use default: for now, we turn + track buffering off for all Medusas, though it + could be used with ones that have a counter + card. But the test is too hard :-( */ + UseTrackbuffer = !MACH_IS_MEDUSA; + + /* initialize variables */ + SelectedDrive = -1; + BufferDrive = -1; + + DMABuffer = atari_stram_alloc(BUFFER_SIZE+512, "ataflop"); + if (!DMABuffer) { + printk(KERN_ERR "atari_floppy_init: cannot get dma buffer\n"); + goto Enomem; + } + TrackBuffer = DMABuffer + 512; + PhysDMABuffer = virt_to_phys(DMABuffer); + PhysTrackBuffer = virt_to_phys(TrackBuffer); + BufferDrive = BufferSide = BufferTrack = -1; + + floppy_queue = blk_init_queue(do_fd_request, &ataflop_lock); + if (!floppy_queue) + goto Enomem; + + for (i = 0; i < FD_MAX_UNITS; i++) { + unit[i].track = -1; + unit[i].flags = 0; + unit[i].disk->major = FLOPPY_MAJOR; + unit[i].disk->first_minor = i; + sprintf(unit[i].disk->disk_name, "fd%d", i); + unit[i].disk->fops = &floppy_fops; + unit[i].disk->private_data = &unit[i]; + unit[i].disk->queue = floppy_queue; + set_capacity(unit[i].disk, MAX_DISK_SIZE * 2); + add_disk(unit[i].disk); + } + + blk_register_region(MKDEV(FLOPPY_MAJOR, 0), 256, THIS_MODULE, + floppy_find, NULL, NULL); + + printk(KERN_INFO "Atari floppy driver: max. %cD, %strack buffering\n", + DriveType == 0 ? 'D' : DriveType == 1 ? 'H' : 'E', + UseTrackbuffer ? "" : "no "); + config_types(); + + return 0; +Enomem: + while (i--) + put_disk(unit[i].disk); + if (floppy_queue) + blk_cleanup_queue(floppy_queue); + unregister_blkdev(FLOPPY_MAJOR, "fd"); + return -ENOMEM; +} + + +void __init atari_floppy_setup( char *str, int *ints ) +{ + int i; + + if (ints[0] < 1) { + printk(KERN_ERR "ataflop_setup: no arguments!\n" ); + return; + } + else if (ints[0] > 2+FD_MAX_UNITS) { + printk(KERN_ERR "ataflop_setup: too many arguments\n" ); + } + + if (ints[1] < 0 || ints[1] > 2) + printk(KERN_ERR "ataflop_setup: bad drive type\n" ); + else + DriveType = ints[1]; + + if (ints[0] >= 2) + UseTrackbuffer = (ints[2] > 0); + + for( i = 3; i <= ints[0] && i-3 < FD_MAX_UNITS; ++i ) { + if (ints[i] != 2 && ints[i] != 3 && ints[i] != 6 && ints[i] != 12) + printk(KERN_ERR "ataflop_setup: bad steprate\n" ); + else + UserSteprate[i-3] = ints[i]; + } +} + +static void atari_floppy_exit(void) +{ + int i; + blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); + for (i = 0; i < FD_MAX_UNITS; i++) { + del_gendisk(unit[i].disk); + put_disk(unit[i].disk); + } + unregister_blkdev(FLOPPY_MAJOR, "fd"); + + blk_cleanup_queue(floppy_queue); + del_timer_sync(&fd_timer); + atari_stram_free( DMABuffer ); +} + +module_init(atari_floppy_init) +module_exit(atari_floppy_exit) + +MODULE_LICENSE("GPL"); diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c new file mode 100644 index 00000000000..8f7c1a1ed7f --- /dev/null +++ b/drivers/block/cciss.c @@ -0,0 +1,2976 @@ +/* + * Disk Array driver for HP SA 5xxx and 6xxx Controllers + * Copyright 2000, 2002 Hewlett-Packard Development Company, L.P. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + */ + +#include <linux/config.h> /* CONFIG_PROC_FS */ +#include <linux/module.h> +#include <linux/interrupt.h> +#include <linux/types.h> +#include <linux/pci.h> +#include <linux/kernel.h> +#include <linux/slab.h> +#include <linux/delay.h> +#include <linux/major.h> +#include <linux/fs.h> +#include <linux/bio.h> +#include <linux/blkpg.h> +#include <linux/timer.h> +#include <linux/proc_fs.h> +#include <linux/init.h> +#include <linux/hdreg.h> +#include <linux/spinlock.h> +#include <linux/compat.h> +#include <asm/uaccess.h> +#include <asm/io.h> + +#include <linux/blkdev.h> +#include <linux/genhd.h> +#include <linux/completion.h> + +#define CCISS_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin)) +#define DRIVER_NAME "HP CISS Driver (v 2.6.6)" +#define DRIVER_VERSION CCISS_DRIVER_VERSION(2,6,6) + +/* Embedded module documentation macros - see modules.h */ +MODULE_AUTHOR("Hewlett-Packard Company"); +MODULE_DESCRIPTION("Driver for HP Controller SA5xxx SA6xxx version 2.6.6"); +MODULE_SUPPORTED_DEVICE("HP SA5i SA5i+ SA532 SA5300 SA5312 SA641 SA642 SA6400" + " SA6i P600 P800 E400"); +MODULE_LICENSE("GPL"); + +#include "cciss_cmd.h" +#include "cciss.h" +#include <linux/cciss_ioctl.h> + +/* define the PCI info for the cards we can control */ +static const struct pci_device_id cciss_pci_device_id[] = { + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISS, + 0x0E11, 0x4070, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, + 0x0E11, 0x4080, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, + 0x0E11, 0x4082, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, + 0x0E11, 0x4083, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, + 0x0E11, 0x409A, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, + 0x0E11, 0x409B, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, + 0x0E11, 0x409C, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, + 0x0E11, 0x409D, 0, 0, 0}, + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, + 0x0E11, 0x4091, 0, 0, 0}, + { PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSA, + 0x103C, 0x3225, 0, 0, 0}, + { PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSB, + 0x103c, 0x3223, 0, 0, 0}, + { PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSB, + 0x103c, 0x3231, 0, 0, 0}, + {0,} +}; +MODULE_DEVICE_TABLE(pci, cciss_pci_device_id); + +#define NR_PRODUCTS (sizeof(products)/sizeof(struct board_type)) + +/* board_id = Subsystem Device ID & Vendor ID + * product = Marketing Name for the board + * access = Address of the struct of function pointers + */ +static struct board_type products[] = { + { 0x40700E11, "Smart Array 5300", &SA5_access }, + { 0x40800E11, "Smart Array 5i", &SA5B_access}, + { 0x40820E11, "Smart Array 532", &SA5B_access}, + { 0x40830E11, "Smart Array 5312", &SA5B_access}, + { 0x409A0E11, "Smart Array 641", &SA5_access}, + { 0x409B0E11, "Smart Array 642", &SA5_access}, + { 0x409C0E11, "Smart Array 6400", &SA5_access}, + { 0x409D0E11, "Smart Array 6400 EM", &SA5_access}, + { 0x40910E11, "Smart Array 6i", &SA5_access}, + { 0x3225103C, "Smart Array P600", &SA5_access}, + { 0x3223103C, "Smart Array P800", &SA5_access}, + { 0x3231103C, "Smart Array E400", &SA5_access}, +}; + +/* How long to wait (in millesconds) for board to go into simple mode */ +#define MAX_CONFIG_WAIT 30000 +#define MAX_IOCTL_CONFIG_WAIT 1000 + +/*define how many times we will try a command because of bus resets */ +#define MAX_CMD_RETRIES 3 + +#define READ_AHEAD 1024 +#define NR_CMDS 384 /* #commands that can be outstanding */ +#define MAX_CTLR 32 + +/* Originally cciss driver only supports 8 major numbers */ +#define MAX_CTLR_ORIG 8 + + +#define CCISS_DMA_MASK 0xFFFFFFFF /* 32 bit DMA */ + +static ctlr_info_t *hba[MAX_CTLR]; + +static void do_cciss_request(request_queue_t *q); +static int cciss_open(struct inode *inode, struct file *filep); +static int cciss_release(struct inode *inode, struct file *filep); +static int cciss_ioctl(struct inode *inode, struct file *filep, + unsigned int cmd, unsigned long arg); + +static int revalidate_allvol(ctlr_info_t *host); +static int cciss_revalidate(struct gendisk *disk); +static int deregister_disk(struct gendisk *disk); +static int register_new_disk(ctlr_info_t *h); + +static void cciss_getgeometry(int cntl_num); + +static void start_io( ctlr_info_t *h); +static int sendcmd( __u8 cmd, int ctlr, void *buff, size_t size, + unsigned int use_unit_num, unsigned int log_unit, __u8 page_code, + unsigned char *scsi3addr, int cmd_type); + +#ifdef CONFIG_PROC_FS +static int cciss_proc_get_info(char *buffer, char **start, off_t offset, + int length, int *eof, void *data); +static void cciss_procinit(int i); +#else +static void cciss_procinit(int i) {} +#endif /* CONFIG_PROC_FS */ + +#ifdef CONFIG_COMPAT +static long cciss_compat_ioctl(struct file *f, unsigned cmd, unsigned long arg); +#endif + +static struct block_device_operations cciss_fops = { + .owner = THIS_MODULE, + .open = cciss_open, + .release = cciss_release, + .ioctl = cciss_ioctl, +#ifdef CONFIG_COMPAT + .compat_ioctl = cciss_compat_ioctl, +#endif + .revalidate_disk= cciss_revalidate, +}; + +/* + * Enqueuing and dequeuing functions for cmdlists. + */ +static inline void addQ(CommandList_struct **Qptr, CommandList_struct *c) +{ + if (*Qptr == NULL) { + *Qptr = c; + c->next = c->prev = c; + } else { + c->prev = (*Qptr)->prev; + c->next = (*Qptr); + (*Qptr)->prev->next = c; + (*Qptr)->prev = c; + } +} + +static inline CommandList_struct *removeQ(CommandList_struct **Qptr, + CommandList_struct *c) +{ + if (c && c->next != c) { + if (*Qptr == c) *Qptr = c->next; + c->prev->next = c->next; + c->next->prev = c->prev; + } else { + *Qptr = NULL; + } + return c; +} + +#include "cciss_scsi.c" /* For SCSI tape support */ + +#ifdef CONFIG_PROC_FS + +/* + * Report information about this controller. + */ +#define ENG_GIG 1000000000 +#define ENG_GIG_FACTOR (ENG_GIG/512) +#define RAID_UNKNOWN 6 +static const char *raid_label[] = {"0","4","1(1+0)","5","5+1","ADG", + "UNKNOWN"}; + +static struct proc_dir_entry *proc_cciss; + +static int cciss_proc_get_info(char *buffer, char **start, off_t offset, + int length, int *eof, void *data) +{ + off_t pos = 0; + off_t len = 0; + int size, i, ctlr; + ctlr_info_t *h = (ctlr_info_t*)data; + drive_info_struct *drv; + unsigned long flags; + sector_t vol_sz, vol_sz_frac; + + ctlr = h->ctlr; + + /* prevent displaying bogus info during configuration + * or deconfiguration of a logical volume + */ + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + if (h->busy_configuring) { + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + return -EBUSY; + } + h->busy_configuring = 1; + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + + size = sprintf(buffer, "%s: HP %s Controller\n" + "Board ID: 0x%08lx\n" + "Firmware Version: %c%c%c%c\n" + "IRQ: %d\n" + "Logical drives: %d\n" + "Current Q depth: %d\n" + "Current # commands on controller: %d\n" + "Max Q depth since init: %d\n" + "Max # commands on controller since init: %d\n" + "Max SG entries since init: %d\n\n", + h->devname, + h->product_name, + (unsigned long)h->board_id, + h->firm_ver[0], h->firm_ver[1], h->firm_ver[2], h->firm_ver[3], + (unsigned int)h->intr, + h->num_luns, + h->Qdepth, h->commands_outstanding, + h->maxQsinceinit, h->max_outstanding, h->maxSG); + + pos += size; len += size; + cciss_proc_tape_report(ctlr, buffer, &pos, &len); + for(i=0; i<=h->highest_lun; i++) { + + drv = &h->drv[i]; + if (drv->block_size == 0) + continue; + + vol_sz = drv->nr_blocks; + vol_sz_frac = sector_div(vol_sz, ENG_GIG_FACTOR); + vol_sz_frac *= 100; + sector_div(vol_sz_frac, ENG_GIG_FACTOR); + + if (drv->raid_level > 5) + drv->raid_level = RAID_UNKNOWN; + size = sprintf(buffer+len, "cciss/c%dd%d:" + "\t%4u.%02uGB\tRAID %s\n", + ctlr, i, (int)vol_sz, (int)vol_sz_frac, + raid_label[drv->raid_level]); + pos += size; len += size; + } + + *eof = 1; + *start = buffer+offset; + len -= offset; + if (len>length) + len = length; + h->busy_configuring = 0; + return len; +} + +static int +cciss_proc_write(struct file *file, const char __user *buffer, + unsigned long count, void *data) +{ + unsigned char cmd[80]; + int len; +#ifdef CONFIG_CISS_SCSI_TAPE + ctlr_info_t *h = (ctlr_info_t *) data; + int rc; +#endif + + if (count > sizeof(cmd)-1) return -EINVAL; + if (copy_from_user(cmd, buffer, count)) return -EFAULT; + cmd[count] = '\0'; + len = strlen(cmd); // above 3 lines ensure safety + if (len && cmd[len-1] == '\n') + cmd[--len] = '\0'; +# ifdef CONFIG_CISS_SCSI_TAPE + if (strcmp("engage scsi", cmd)==0) { + rc = cciss_engage_scsi(h->ctlr); + if (rc != 0) return -rc; + return count; + } + /* might be nice to have "disengage" too, but it's not + safely possible. (only 1 module use count, lock issues.) */ +# endif + return -EINVAL; +} + +/* + * Get us a file in /proc/cciss that says something about each controller. + * Create /proc/cciss if it doesn't exist yet. + */ +static void __devinit cciss_procinit(int i) +{ + struct proc_dir_entry *pde; + + if (proc_cciss == NULL) { + proc_cciss = proc_mkdir("cciss", proc_root_driver); + if (!proc_cciss) + return; + } + + pde = create_proc_read_entry(hba[i]->devname, + S_IWUSR | S_IRUSR | S_IRGRP | S_IROTH, + proc_cciss, cciss_proc_get_info, hba[i]); + pde->write_proc = cciss_proc_write; +} +#endif /* CONFIG_PROC_FS */ + +/* + * For operations that cannot sleep, a command block is allocated at init, + * and managed by cmd_alloc() and cmd_free() using a simple bitmap to track + * which ones are free or in use. For operations that can wait for kmalloc + * to possible sleep, this routine can be called with get_from_pool set to 0. + * cmd_free() MUST be called with a got_from_pool set to 0 if cmd_alloc was. + */ +static CommandList_struct * cmd_alloc(ctlr_info_t *h, int get_from_pool) +{ + CommandList_struct *c; + int i; + u64bit temp64; + dma_addr_t cmd_dma_handle, err_dma_handle; + + if (!get_from_pool) + { + c = (CommandList_struct *) pci_alloc_consistent( + h->pdev, sizeof(CommandList_struct), &cmd_dma_handle); + if(c==NULL) + return NULL; + memset(c, 0, sizeof(CommandList_struct)); + + c->err_info = (ErrorInfo_struct *)pci_alloc_consistent( + h->pdev, sizeof(ErrorInfo_struct), + &err_dma_handle); + + if (c->err_info == NULL) + { + pci_free_consistent(h->pdev, + sizeof(CommandList_struct), c, cmd_dma_handle); + return NULL; + } + memset(c->err_info, 0, sizeof(ErrorInfo_struct)); + } else /* get it out of the controllers pool */ + { + do { + i = find_first_zero_bit(h->cmd_pool_bits, NR_CMDS); + if (i == NR_CMDS) + return NULL; + } while(test_and_set_bit(i & (BITS_PER_LONG - 1), h->cmd_pool_bits+(i/BITS_PER_LONG)) != 0); +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "cciss: using command buffer %d\n", i); +#endif + c = h->cmd_pool + i; + memset(c, 0, sizeof(CommandList_struct)); + cmd_dma_handle = h->cmd_pool_dhandle + + i*sizeof(CommandList_struct); + c->err_info = h->errinfo_pool + i; + memset(c->err_info, 0, sizeof(ErrorInfo_struct)); + err_dma_handle = h->errinfo_pool_dhandle + + i*sizeof(ErrorInfo_struct); + h->nr_allocs++; + } + + c->busaddr = (__u32) cmd_dma_handle; + temp64.val = (__u64) err_dma_handle; + c->ErrDesc.Addr.lower = temp64.val32.lower; + c->ErrDesc.Addr.upper = temp64.val32.upper; + c->ErrDesc.Len = sizeof(ErrorInfo_struct); + + c->ctlr = h->ctlr; + return c; + + +} + +/* + * Frees a command block that was previously allocated with cmd_alloc(). + */ +static void cmd_free(ctlr_info_t *h, CommandList_struct *c, int got_from_pool) +{ + int i; + u64bit temp64; + + if( !got_from_pool) + { + temp64.val32.lower = c->ErrDesc.Addr.lower; + temp64.val32.upper = c->ErrDesc.Addr.upper; + pci_free_consistent(h->pdev, sizeof(ErrorInfo_struct), + c->err_info, (dma_addr_t) temp64.val); + pci_free_consistent(h->pdev, sizeof(CommandList_struct), + c, (dma_addr_t) c->busaddr); + } else + { + i = c - h->cmd_pool; + clear_bit(i&(BITS_PER_LONG-1), h->cmd_pool_bits+(i/BITS_PER_LONG)); + h->nr_frees++; + } +} + +static inline ctlr_info_t *get_host(struct gendisk *disk) +{ + return disk->queue->queuedata; +} + +static inline drive_info_struct *get_drv(struct gendisk *disk) +{ + return disk->private_data; +} + +/* + * Open. Make sure the device is really there. + */ +static int cciss_open(struct inode *inode, struct file *filep) +{ + ctlr_info_t *host = get_host(inode->i_bdev->bd_disk); + drive_info_struct *drv = get_drv(inode->i_bdev->bd_disk); + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "cciss_open %s\n", inode->i_bdev->bd_disk->disk_name); +#endif /* CCISS_DEBUG */ + + /* + * Root is allowed to open raw volume zero even if it's not configured + * so array config can still work. Root is also allowed to open any + * volume that has a LUN ID, so it can issue IOCTL to reread the + * disk information. I don't think I really like this + * but I'm already using way to many device nodes to claim another one + * for "raw controller". + */ + if (drv->nr_blocks == 0) { + if (iminor(inode) != 0) { /* not node 0? */ + /* if not node 0 make sure it is a partition = 0 */ + if (iminor(inode) & 0x0f) { + return -ENXIO; + /* if it is, make sure we have a LUN ID */ + } else if (drv->LunID == 0) { + return -ENXIO; + } + } + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + } + drv->usage_count++; + host->usage_count++; + return 0; +} +/* + * Close. Sync first. + */ +static int cciss_release(struct inode *inode, struct file *filep) +{ + ctlr_info_t *host = get_host(inode->i_bdev->bd_disk); + drive_info_struct *drv = get_drv(inode->i_bdev->bd_disk); + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "cciss_release %s\n", inode->i_bdev->bd_disk->disk_name); +#endif /* CCISS_DEBUG */ + + drv->usage_count--; + host->usage_count--; + return 0; +} + +#ifdef CONFIG_COMPAT + +static int do_ioctl(struct file *f, unsigned cmd, unsigned long arg) +{ + int ret; + lock_kernel(); + ret = cciss_ioctl(f->f_dentry->d_inode, f, cmd, arg); + unlock_kernel(); + return ret; +} + +static int cciss_ioctl32_passthru(struct file *f, unsigned cmd, unsigned long arg); +static int cciss_ioctl32_big_passthru(struct file *f, unsigned cmd, unsigned long arg); + +static long cciss_compat_ioctl(struct file *f, unsigned cmd, unsigned long arg) +{ + switch (cmd) { + case CCISS_GETPCIINFO: + case CCISS_GETINTINFO: + case CCISS_SETINTINFO: + case CCISS_GETNODENAME: + case CCISS_SETNODENAME: + case CCISS_GETHEARTBEAT: + case CCISS_GETBUSTYPES: + case CCISS_GETFIRMVER: + case CCISS_GETDRIVVER: + case CCISS_REVALIDVOLS: + case CCISS_DEREGDISK: + case CCISS_REGNEWDISK: + case CCISS_REGNEWD: + case CCISS_RESCANDISK: + case CCISS_GETLUNINFO: + return do_ioctl(f, cmd, arg); + + case CCISS_PASSTHRU32: + return cciss_ioctl32_passthru(f, cmd, arg); + case CCISS_BIG_PASSTHRU32: + return cciss_ioctl32_big_passthru(f, cmd, arg); + + default: + return -ENOIOCTLCMD; + } +} + +static int cciss_ioctl32_passthru(struct file *f, unsigned cmd, unsigned long arg) +{ + IOCTL32_Command_struct __user *arg32 = + (IOCTL32_Command_struct __user *) arg; + IOCTL_Command_struct arg64; + IOCTL_Command_struct __user *p = compat_alloc_user_space(sizeof(arg64)); + int err; + u32 cp; + + err = 0; + err |= copy_from_user(&arg64.LUN_info, &arg32->LUN_info, sizeof(arg64.LUN_info)); + err |= copy_from_user(&arg64.Request, &arg32->Request, sizeof(arg64.Request)); + err |= copy_from_user(&arg64.error_info, &arg32->error_info, sizeof(arg64.error_info)); + err |= get_user(arg64.buf_size, &arg32->buf_size); + err |= get_user(cp, &arg32->buf); + arg64.buf = compat_ptr(cp); + err |= copy_to_user(p, &arg64, sizeof(arg64)); + + if (err) + return -EFAULT; + + err = do_ioctl(f, CCISS_PASSTHRU, (unsigned long) p); + if (err) + return err; + err |= copy_in_user(&arg32->error_info, &p->error_info, sizeof(arg32->error_info)); + if (err) + return -EFAULT; + return err; +} + +static int cciss_ioctl32_big_passthru(struct file *file, unsigned cmd, unsigned long arg) +{ + BIG_IOCTL32_Command_struct __user *arg32 = + (BIG_IOCTL32_Command_struct __user *) arg; + BIG_IOCTL_Command_struct arg64; + BIG_IOCTL_Command_struct __user *p = compat_alloc_user_space(sizeof(arg64)); + int err; + u32 cp; + + err = 0; + err |= copy_from_user(&arg64.LUN_info, &arg32->LUN_info, sizeof(arg64.LUN_info)); + err |= copy_from_user(&arg64.Request, &arg32->Request, sizeof(arg64.Request)); + err |= copy_from_user(&arg64.error_info, &arg32->error_info, sizeof(arg64.error_info)); + err |= get_user(arg64.buf_size, &arg32->buf_size); + err |= get_user(arg64.malloc_size, &arg32->malloc_size); + err |= get_user(cp, &arg32->buf); + arg64.buf = compat_ptr(cp); + err |= copy_to_user(p, &arg64, sizeof(arg64)); + + if (err) + return -EFAULT; + + err = do_ioctl(file, CCISS_BIG_PASSTHRU, (unsigned long) p); + if (err) + return err; + err |= copy_in_user(&arg32->error_info, &p->error_info, sizeof(arg32->error_info)); + if (err) + return -EFAULT; + return err; +} +#endif +/* + * ioctl + */ +static int cciss_ioctl(struct inode *inode, struct file *filep, + unsigned int cmd, unsigned long arg) +{ + struct block_device *bdev = inode->i_bdev; + struct gendisk *disk = bdev->bd_disk; + ctlr_info_t *host = get_host(disk); + drive_info_struct *drv = get_drv(disk); + int ctlr = host->ctlr; + void __user *argp = (void __user *)arg; + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "cciss_ioctl: Called with cmd=%x %lx\n", cmd, arg); +#endif /* CCISS_DEBUG */ + + switch(cmd) { + case HDIO_GETGEO: + { + struct hd_geometry driver_geo; + if (drv->cylinders) { + driver_geo.heads = drv->heads; + driver_geo.sectors = drv->sectors; + driver_geo.cylinders = drv->cylinders; + } else + return -ENXIO; + driver_geo.start= get_start_sect(inode->i_bdev); + if (copy_to_user(argp, &driver_geo, sizeof(struct hd_geometry))) + return -EFAULT; + return(0); + } + + case CCISS_GETPCIINFO: + { + cciss_pci_info_struct pciinfo; + + if (!arg) return -EINVAL; + pciinfo.bus = host->pdev->bus->number; + pciinfo.dev_fn = host->pdev->devfn; + pciinfo.board_id = host->board_id; + if (copy_to_user(argp, &pciinfo, sizeof( cciss_pci_info_struct ))) + return -EFAULT; + return(0); + } + case CCISS_GETINTINFO: + { + cciss_coalint_struct intinfo; + if (!arg) return -EINVAL; + intinfo.delay = readl(&host->cfgtable->HostWrite.CoalIntDelay); + intinfo.count = readl(&host->cfgtable->HostWrite.CoalIntCount); + if (copy_to_user(argp, &intinfo, sizeof( cciss_coalint_struct ))) + return -EFAULT; + return(0); + } + case CCISS_SETINTINFO: + { + cciss_coalint_struct intinfo; + unsigned long flags; + int i; + + if (!arg) return -EINVAL; + if (!capable(CAP_SYS_ADMIN)) return -EPERM; + if (copy_from_user(&intinfo, argp, sizeof( cciss_coalint_struct))) + return -EFAULT; + if ( (intinfo.delay == 0 ) && (intinfo.count == 0)) + + { +// printk("cciss_ioctl: delay and count cannot be 0\n"); + return( -EINVAL); + } + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + /* Update the field, and then ring the doorbell */ + writel( intinfo.delay, + &(host->cfgtable->HostWrite.CoalIntDelay)); + writel( intinfo.count, + &(host->cfgtable->HostWrite.CoalIntCount)); + writel( CFGTBL_ChangeReq, host->vaddr + SA5_DOORBELL); + + for(i=0;i<MAX_IOCTL_CONFIG_WAIT;i++) { + if (!(readl(host->vaddr + SA5_DOORBELL) + & CFGTBL_ChangeReq)) + break; + /* delay and try again */ + udelay(1000); + } + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + if (i >= MAX_IOCTL_CONFIG_WAIT) + return -EAGAIN; + return(0); + } + case CCISS_GETNODENAME: + { + NodeName_type NodeName; + int i; + + if (!arg) return -EINVAL; + for(i=0;i<16;i++) + NodeName[i] = readb(&host->cfgtable->ServerName[i]); + if (copy_to_user(argp, NodeName, sizeof( NodeName_type))) + return -EFAULT; + return(0); + } + case CCISS_SETNODENAME: + { + NodeName_type NodeName; + unsigned long flags; + int i; + + if (!arg) return -EINVAL; + if (!capable(CAP_SYS_ADMIN)) return -EPERM; + + if (copy_from_user(NodeName, argp, sizeof( NodeName_type))) + return -EFAULT; + + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + + /* Update the field, and then ring the doorbell */ + for(i=0;i<16;i++) + writeb( NodeName[i], &host->cfgtable->ServerName[i]); + + writel( CFGTBL_ChangeReq, host->vaddr + SA5_DOORBELL); + + for(i=0;i<MAX_IOCTL_CONFIG_WAIT;i++) { + if (!(readl(host->vaddr + SA5_DOORBELL) + & CFGTBL_ChangeReq)) + break; + /* delay and try again */ + udelay(1000); + } + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + if (i >= MAX_IOCTL_CONFIG_WAIT) + return -EAGAIN; + return(0); + } + + case CCISS_GETHEARTBEAT: + { + Heartbeat_type heartbeat; + + if (!arg) return -EINVAL; + heartbeat = readl(&host->cfgtable->HeartBeat); + if (copy_to_user(argp, &heartbeat, sizeof( Heartbeat_type))) + return -EFAULT; + return(0); + } + case CCISS_GETBUSTYPES: + { + BusTypes_type BusTypes; + + if (!arg) return -EINVAL; + BusTypes = readl(&host->cfgtable->BusTypes); + if (copy_to_user(argp, &BusTypes, sizeof( BusTypes_type) )) + return -EFAULT; + return(0); + } + case CCISS_GETFIRMVER: + { + FirmwareVer_type firmware; + + if (!arg) return -EINVAL; + memcpy(firmware, host->firm_ver, 4); + + if (copy_to_user(argp, firmware, sizeof( FirmwareVer_type))) + return -EFAULT; + return(0); + } + case CCISS_GETDRIVVER: + { + DriverVer_type DriverVer = DRIVER_VERSION; + + if (!arg) return -EINVAL; + + if (copy_to_user(argp, &DriverVer, sizeof( DriverVer_type) )) + return -EFAULT; + return(0); + } + + case CCISS_REVALIDVOLS: + if (bdev != bdev->bd_contains || drv != host->drv) + return -ENXIO; + return revalidate_allvol(host); + + case CCISS_GETLUNINFO: { + LogvolInfo_struct luninfo; + int i; + + luninfo.LunID = drv->LunID; + luninfo.num_opens = drv->usage_count; + luninfo.num_parts = 0; + /* count partitions 1 to 15 with sizes > 0 */ + for (i = 0; i < MAX_PART - 1; i++) { + if (!disk->part[i]) + continue; + if (disk->part[i]->nr_sects != 0) + luninfo.num_parts++; + } + if (copy_to_user(argp, &luninfo, + sizeof(LogvolInfo_struct))) + return -EFAULT; + return(0); + } + case CCISS_DEREGDISK: + return deregister_disk(disk); + + case CCISS_REGNEWD: + return register_new_disk(host); + + case CCISS_PASSTHRU: + { + IOCTL_Command_struct iocommand; + CommandList_struct *c; + char *buff = NULL; + u64bit temp64; + unsigned long flags; + DECLARE_COMPLETION(wait); + + if (!arg) return -EINVAL; + + if (!capable(CAP_SYS_RAWIO)) return -EPERM; + + if (copy_from_user(&iocommand, argp, sizeof( IOCTL_Command_struct) )) + return -EFAULT; + if((iocommand.buf_size < 1) && + (iocommand.Request.Type.Direction != XFER_NONE)) + { + return -EINVAL; + } +#if 0 /* 'buf_size' member is 16-bits, and always smaller than kmalloc limit */ + /* Check kmalloc limits */ + if(iocommand.buf_size > 128000) + return -EINVAL; +#endif + if(iocommand.buf_size > 0) + { + buff = kmalloc(iocommand.buf_size, GFP_KERNEL); + if( buff == NULL) + return -EFAULT; + } + if (iocommand.Request.Type.Direction == XFER_WRITE) + { + /* Copy the data into the buffer we created */ + if (copy_from_user(buff, iocommand.buf, iocommand.buf_size)) + { + kfree(buff); + return -EFAULT; + } + } else { + memset(buff, 0, iocommand.buf_size); + } + if ((c = cmd_alloc(host , 0)) == NULL) + { + kfree(buff); + return -ENOMEM; + } + // Fill in the command type + c->cmd_type = CMD_IOCTL_PEND; + // Fill in Command Header + c->Header.ReplyQueue = 0; // unused in simple mode + if( iocommand.buf_size > 0) // buffer to fill + { + c->Header.SGList = 1; + c->Header.SGTotal= 1; + } else // no buffers to fill + { + c->Header.SGList = 0; + c->Header.SGTotal= 0; + } + c->Header.LUN = iocommand.LUN_info; + c->Header.Tag.lower = c->busaddr; // use the kernel address the cmd block for tag + + // Fill in Request block + c->Request = iocommand.Request; + + // Fill in the scatter gather information + if (iocommand.buf_size > 0 ) + { + temp64.val = pci_map_single( host->pdev, buff, + iocommand.buf_size, + PCI_DMA_BIDIRECTIONAL); + c->SG[0].Addr.lower = temp64.val32.lower; + c->SG[0].Addr.upper = temp64.val32.upper; + c->SG[0].Len = iocommand.buf_size; + c->SG[0].Ext = 0; // we are not chaining + } + c->waiting = &wait; + + /* Put the request on the tail of the request queue */ + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + addQ(&host->reqQ, c); + host->Qdepth++; + start_io(host); + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + + wait_for_completion(&wait); + + /* unlock the buffers from DMA */ + temp64.val32.lower = c->SG[0].Addr.lower; + temp64.val32.upper = c->SG[0].Addr.upper; + pci_unmap_single( host->pdev, (dma_addr_t) temp64.val, + iocommand.buf_size, PCI_DMA_BIDIRECTIONAL); + + /* Copy the error information out */ + iocommand.error_info = *(c->err_info); + if ( copy_to_user(argp, &iocommand, sizeof( IOCTL_Command_struct) ) ) + { + kfree(buff); + cmd_free(host, c, 0); + return( -EFAULT); + } + + if (iocommand.Request.Type.Direction == XFER_READ) + { + /* Copy the data out of the buffer we created */ + if (copy_to_user(iocommand.buf, buff, iocommand.buf_size)) + { + kfree(buff); + cmd_free(host, c, 0); + return -EFAULT; + } + } + kfree(buff); + cmd_free(host, c, 0); + return(0); + } + case CCISS_BIG_PASSTHRU: { + BIG_IOCTL_Command_struct *ioc; + CommandList_struct *c; + unsigned char **buff = NULL; + int *buff_size = NULL; + u64bit temp64; + unsigned long flags; + BYTE sg_used = 0; + int status = 0; + int i; + DECLARE_COMPLETION(wait); + __u32 left; + __u32 sz; + BYTE __user *data_ptr; + + if (!arg) + return -EINVAL; + if (!capable(CAP_SYS_RAWIO)) + return -EPERM; + ioc = (BIG_IOCTL_Command_struct *) + kmalloc(sizeof(*ioc), GFP_KERNEL); + if (!ioc) { + status = -ENOMEM; + goto cleanup1; + } + if (copy_from_user(ioc, argp, sizeof(*ioc))) { + status = -EFAULT; + goto cleanup1; + } + if ((ioc->buf_size < 1) && + (ioc->Request.Type.Direction != XFER_NONE)) { + status = -EINVAL; + goto cleanup1; + } + /* Check kmalloc limits using all SGs */ + if (ioc->malloc_size > MAX_KMALLOC_SIZE) { + status = -EINVAL; + goto cleanup1; + } + if (ioc->buf_size > ioc->malloc_size * MAXSGENTRIES) { + status = -EINVAL; + goto cleanup1; + } + buff = (unsigned char **) kmalloc(MAXSGENTRIES * + sizeof(char *), GFP_KERNEL); + if (!buff) { + status = -ENOMEM; + goto cleanup1; + } + memset(buff, 0, MAXSGENTRIES); + buff_size = (int *) kmalloc(MAXSGENTRIES * sizeof(int), + GFP_KERNEL); + if (!buff_size) { + status = -ENOMEM; + goto cleanup1; + } + left = ioc->buf_size; + data_ptr = ioc->buf; + while (left) { + sz = (left > ioc->malloc_size) ? ioc->malloc_size : left; + buff_size[sg_used] = sz; + buff[sg_used] = kmalloc(sz, GFP_KERNEL); + if (buff[sg_used] == NULL) { + status = -ENOMEM; + goto cleanup1; + } + if (ioc->Request.Type.Direction == XFER_WRITE && + copy_from_user(buff[sg_used], data_ptr, sz)) { + status = -ENOMEM; + goto cleanup1; + } else { + memset(buff[sg_used], 0, sz); + } + left -= sz; + data_ptr += sz; + sg_used++; + } + if ((c = cmd_alloc(host , 0)) == NULL) { + status = -ENOMEM; + goto cleanup1; + } + c->cmd_type = CMD_IOCTL_PEND; + c->Header.ReplyQueue = 0; + + if( ioc->buf_size > 0) { + c->Header.SGList = sg_used; + c->Header.SGTotal= sg_used; + } else { + c->Header.SGList = 0; + c->Header.SGTotal= 0; + } + c->Header.LUN = ioc->LUN_info; + c->Header.Tag.lower = c->busaddr; + + c->Request = ioc->Request; + if (ioc->buf_size > 0 ) { + int i; + for(i=0; i<sg_used; i++) { + temp64.val = pci_map_single( host->pdev, buff[i], + buff_size[i], + PCI_DMA_BIDIRECTIONAL); + c->SG[i].Addr.lower = temp64.val32.lower; + c->SG[i].Addr.upper = temp64.val32.upper; + c->SG[i].Len = buff_size[i]; + c->SG[i].Ext = 0; /* we are not chaining */ + } + } + c->waiting = &wait; + /* Put the request on the tail of the request queue */ + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + addQ(&host->reqQ, c); + host->Qdepth++; + start_io(host); + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + wait_for_completion(&wait); + /* unlock the buffers from DMA */ + for(i=0; i<sg_used; i++) { + temp64.val32.lower = c->SG[i].Addr.lower; + temp64.val32.upper = c->SG[i].Addr.upper; + pci_unmap_single( host->pdev, (dma_addr_t) temp64.val, + buff_size[i], PCI_DMA_BIDIRECTIONAL); + } + /* Copy the error information out */ + ioc->error_info = *(c->err_info); + if (copy_to_user(argp, ioc, sizeof(*ioc))) { + cmd_free(host, c, 0); + status = -EFAULT; + goto cleanup1; + } + if (ioc->Request.Type.Direction == XFER_READ) { + /* Copy the data out of the buffer we created */ + BYTE __user *ptr = ioc->buf; + for(i=0; i< sg_used; i++) { + if (copy_to_user(ptr, buff[i], buff_size[i])) { + cmd_free(host, c, 0); + status = -EFAULT; + goto cleanup1; + } + ptr += buff_size[i]; + } + } + cmd_free(host, c, 0); + status = 0; +cleanup1: + if (buff) { + for(i=0; i<sg_used; i++) + if(buff[i] != NULL) + kfree(buff[i]); + kfree(buff); + } + if (buff_size) + kfree(buff_size); + if (ioc) + kfree(ioc); + return(status); + } + default: + return -ENOTTY; + } + +} + +/* + * revalidate_allvol is for online array config utilities. After a + * utility reconfigures the drives in the array, it can use this function + * (through an ioctl) to make the driver zap any previous disk structs for + * that controller and get new ones. + * + * Right now I'm using the getgeometry() function to do this, but this + * function should probably be finer grained and allow you to revalidate one + * particualar logical volume (instead of all of them on a particular + * controller). + */ +static int revalidate_allvol(ctlr_info_t *host) +{ + int ctlr = host->ctlr, i; + unsigned long flags; + + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + if (host->usage_count > 1) { + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + printk(KERN_WARNING "cciss: Device busy for volume" + " revalidation (usage=%d)\n", host->usage_count); + return -EBUSY; + } + host->usage_count++; + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + + for(i=0; i< NWD; i++) { + struct gendisk *disk = host->gendisk[i]; + if (disk->flags & GENHD_FL_UP) + del_gendisk(disk); + } + + /* + * Set the partition and block size structures for all volumes + * on this controller to zero. We will reread all of this data + */ + memset(host->drv, 0, sizeof(drive_info_struct) + * CISS_MAX_LUN); + /* + * Tell the array controller not to give us any interrupts while + * we check the new geometry. Then turn interrupts back on when + * we're done. + */ + host->access.set_intr_mask(host, CCISS_INTR_OFF); + cciss_getgeometry(ctlr); + host->access.set_intr_mask(host, CCISS_INTR_ON); + + /* Loop through each real device */ + for (i = 0; i < NWD; i++) { + struct gendisk *disk = host->gendisk[i]; + drive_info_struct *drv = &(host->drv[i]); + /* we must register the controller even if no disks exist */ + /* this is for the online array utilities */ + if (!drv->heads && i) + continue; + blk_queue_hardsect_size(host->queue, drv->block_size); + set_capacity(disk, drv->nr_blocks); + add_disk(disk); + } + host->usage_count--; + return 0; +} + +static int deregister_disk(struct gendisk *disk) +{ + unsigned long flags; + ctlr_info_t *h = get_host(disk); + drive_info_struct *drv = get_drv(disk); + int ctlr = h->ctlr; + + if (!capable(CAP_SYS_RAWIO)) + return -EPERM; + + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + /* make sure logical volume is NOT is use */ + if( drv->usage_count > 1) { + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + return -EBUSY; + } + drv->usage_count++; + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + + /* invalidate the devices and deregister the disk */ + if (disk->flags & GENHD_FL_UP) + del_gendisk(disk); + /* check to see if it was the last disk */ + if (drv == h->drv + h->highest_lun) { + /* if so, find the new hightest lun */ + int i, newhighest =-1; + for(i=0; i<h->highest_lun; i++) { + /* if the disk has size > 0, it is available */ + if (h->drv[i].nr_blocks) + newhighest = i; + } + h->highest_lun = newhighest; + + } + --h->num_luns; + /* zero out the disk size info */ + drv->nr_blocks = 0; + drv->block_size = 0; + drv->cylinders = 0; + drv->LunID = 0; + return(0); +} +static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, + size_t size, + unsigned int use_unit_num, /* 0: address the controller, + 1: address logical volume log_unit, + 2: periph device address is scsi3addr */ + unsigned int log_unit, __u8 page_code, unsigned char *scsi3addr, + int cmd_type) +{ + ctlr_info_t *h= hba[ctlr]; + u64bit buff_dma_handle; + int status = IO_OK; + + c->cmd_type = CMD_IOCTL_PEND; + c->Header.ReplyQueue = 0; + if( buff != NULL) { + c->Header.SGList = 1; + c->Header.SGTotal= 1; + } else { + c->Header.SGList = 0; + c->Header.SGTotal= 0; + } + c->Header.Tag.lower = c->busaddr; + + c->Request.Type.Type = cmd_type; + if (cmd_type == TYPE_CMD) { + switch(cmd) { + case CISS_INQUIRY: + /* If the logical unit number is 0 then, this is going + to controller so It's a physical command + mode = 0 target = 0. So we have nothing to write. + otherwise, if use_unit_num == 1, + mode = 1(volume set addressing) target = LUNID + otherwise, if use_unit_num == 2, + mode = 0(periph dev addr) target = scsi3addr */ + if (use_unit_num == 1) { + c->Header.LUN.LogDev.VolId= + h->drv[log_unit].LunID; + c->Header.LUN.LogDev.Mode = 1; + } else if (use_unit_num == 2) { + memcpy(c->Header.LUN.LunAddrBytes,scsi3addr,8); + c->Header.LUN.LogDev.Mode = 0; + } + /* are we trying to read a vital product page */ + if(page_code != 0) { + c->Request.CDB[1] = 0x01; + c->Request.CDB[2] = page_code; + } + c->Request.CDBLen = 6; + c->Request.Type.Attribute = ATTR_SIMPLE; + c->Request.Type.Direction = XFER_READ; + c->Request.Timeout = 0; + c->Request.CDB[0] = CISS_INQUIRY; + c->Request.CDB[4] = size & 0xFF; + break; + case CISS_REPORT_LOG: + case CISS_REPORT_PHYS: + /* Talking to controller so It's a physical command + mode = 00 target = 0. Nothing to write. + */ + c->Request.CDBLen = 12; + c->Request.Type.Attribute = ATTR_SIMPLE; + c->Request.Type.Direction = XFER_READ; + c->Request.Timeout = 0; + c->Request.CDB[0] = cmd; + c->Request.CDB[6] = (size >> 24) & 0xFF; //MSB + c->Request.CDB[7] = (size >> 16) & 0xFF; + c->Request.CDB[8] = (size >> 8) & 0xFF; + c->Request.CDB[9] = size & 0xFF; + break; + + case CCISS_READ_CAPACITY: + c->Header.LUN.LogDev.VolId = h->drv[log_unit].LunID; + c->Header.LUN.LogDev.Mode = 1; + c->Request.CDBLen = 10; + c->Request.Type.Attribute = ATTR_SIMPLE; + c->Request.Type.Direction = XFER_READ; + c->Request.Timeout = 0; + c->Request.CDB[0] = cmd; + break; + case CCISS_CACHE_FLUSH: + c->Request.CDBLen = 12; + c->Request.Type.Attribute = ATTR_SIMPLE; + c->Request.Type.Direction = XFER_WRITE; + c->Request.Timeout = 0; + c->Request.CDB[0] = BMIC_WRITE; + c->Request.CDB[6] = BMIC_CACHE_FLUSH; + break; + default: + printk(KERN_WARNING + "cciss%d: Unknown Command 0x%c\n", ctlr, cmd); + return(IO_ERROR); + } + } else if (cmd_type == TYPE_MSG) { + switch (cmd) { + case 3: /* No-Op message */ + c->Request.CDBLen = 1; + c->Request.Type.Attribute = ATTR_SIMPLE; + c->Request.Type.Direction = XFER_WRITE; + c->Request.Timeout = 0; + c->Request.CDB[0] = cmd; + break; + default: + printk(KERN_WARNING + "cciss%d: unknown message type %d\n", + ctlr, cmd); + return IO_ERROR; + } + } else { + printk(KERN_WARNING + "cciss%d: unknown command type %d\n", ctlr, cmd_type); + return IO_ERROR; + } + /* Fill in the scatter gather information */ + if (size > 0) { + buff_dma_handle.val = (__u64) pci_map_single(h->pdev, + buff, size, PCI_DMA_BIDIRECTIONAL); + c->SG[0].Addr.lower = buff_dma_handle.val32.lower; + c->SG[0].Addr.upper = buff_dma_handle.val32.upper; + c->SG[0].Len = size; + c->SG[0].Ext = 0; /* we are not chaining */ + } + return status; +} +static int sendcmd_withirq(__u8 cmd, + int ctlr, + void *buff, + size_t size, + unsigned int use_unit_num, + unsigned int log_unit, + __u8 page_code, + int cmd_type) +{ + ctlr_info_t *h = hba[ctlr]; + CommandList_struct *c; + u64bit buff_dma_handle; + unsigned long flags; + int return_status; + DECLARE_COMPLETION(wait); + + if ((c = cmd_alloc(h , 0)) == NULL) + return -ENOMEM; + return_status = fill_cmd(c, cmd, ctlr, buff, size, use_unit_num, + log_unit, page_code, NULL, cmd_type); + if (return_status != IO_OK) { + cmd_free(h, c, 0); + return return_status; + } +resend_cmd2: + c->waiting = &wait; + + /* Put the request on the tail of the queue and send it */ + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + addQ(&h->reqQ, c); + h->Qdepth++; + start_io(h); + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + + wait_for_completion(&wait); + + if(c->err_info->CommandStatus != 0) + { /* an error has occurred */ + switch(c->err_info->CommandStatus) + { + case CMD_TARGET_STATUS: + printk(KERN_WARNING "cciss: cmd %p has " + " completed with errors\n", c); + if( c->err_info->ScsiStatus) + { + printk(KERN_WARNING "cciss: cmd %p " + "has SCSI Status = %x\n", + c, + c->err_info->ScsiStatus); + } + + break; + case CMD_DATA_UNDERRUN: + case CMD_DATA_OVERRUN: + /* expected for inquire and report lun commands */ + break; + case CMD_INVALID: + printk(KERN_WARNING "cciss: Cmd %p is " + "reported invalid\n", c); + return_status = IO_ERROR; + break; + case CMD_PROTOCOL_ERR: + printk(KERN_WARNING "cciss: cmd %p has " + "protocol error \n", c); + return_status = IO_ERROR; + break; +case CMD_HARDWARE_ERR: + printk(KERN_WARNING "cciss: cmd %p had " + " hardware error\n", c); + return_status = IO_ERROR; + break; + case CMD_CONNECTION_LOST: + printk(KERN_WARNING "cciss: cmd %p had " + "connection lost\n", c); + return_status = IO_ERROR; + break; + case CMD_ABORTED: + printk(KERN_WARNING "cciss: cmd %p was " + "aborted\n", c); + return_status = IO_ERROR; + break; + case CMD_ABORT_FAILED: + printk(KERN_WARNING "cciss: cmd %p reports " + "abort failed\n", c); + return_status = IO_ERROR; + break; + case CMD_UNSOLICITED_ABORT: + printk(KERN_WARNING + "cciss%d: unsolicited abort %p\n", + ctlr, c); + if (c->retry_count < MAX_CMD_RETRIES) { + printk(KERN_WARNING + "cciss%d: retrying %p\n", + ctlr, c); + c->retry_count++; + /* erase the old error information */ + memset(c->err_info, 0, + sizeof(ErrorInfo_struct)); + return_status = IO_OK; + INIT_COMPLETION(wait); + goto resend_cmd2; + } + return_status = IO_ERROR; + break; + default: + printk(KERN_WARNING "cciss: cmd %p returned " + "unknown status %x\n", c, + c->err_info->CommandStatus); + return_status = IO_ERROR; + } + } + /* unlock the buffers from DMA */ + pci_unmap_single( h->pdev, (dma_addr_t) buff_dma_handle.val, + size, PCI_DMA_BIDIRECTIONAL); + cmd_free(h, c, 0); + return(return_status); + +} +static void cciss_geometry_inquiry(int ctlr, int logvol, + int withirq, unsigned int total_size, + unsigned int block_size, InquiryData_struct *inq_buff, + drive_info_struct *drv) +{ + int return_code; + memset(inq_buff, 0, sizeof(InquiryData_struct)); + if (withirq) + return_code = sendcmd_withirq(CISS_INQUIRY, ctlr, + inq_buff, sizeof(*inq_buff), 1, logvol ,0xC1, TYPE_CMD); + else + return_code = sendcmd(CISS_INQUIRY, ctlr, inq_buff, + sizeof(*inq_buff), 1, logvol ,0xC1, NULL, TYPE_CMD); + if (return_code == IO_OK) { + if(inq_buff->data_byte[8] == 0xFF) { + printk(KERN_WARNING + "cciss: reading geometry failed, volume " + "does not support reading geometry\n"); + drv->block_size = block_size; + drv->nr_blocks = total_size; + drv->heads = 255; + drv->sectors = 32; // Sectors per track + drv->cylinders = total_size / 255 / 32; + } else { + unsigned int t; + + drv->block_size = block_size; + drv->nr_blocks = total_size; + drv->heads = inq_buff->data_byte[6]; + drv->sectors = inq_buff->data_byte[7]; + drv->cylinders = (inq_buff->data_byte[4] & 0xff) << 8; + drv->cylinders += inq_buff->data_byte[5]; + drv->raid_level = inq_buff->data_byte[8]; + t = drv->heads * drv->sectors; + if (t > 1) { + drv->cylinders = total_size/t; + } + } + } else { /* Get geometry failed */ + printk(KERN_WARNING "cciss: reading geometry failed\n"); + } + printk(KERN_INFO " heads= %d, sectors= %d, cylinders= %d\n\n", + drv->heads, drv->sectors, drv->cylinders); +} +static void +cciss_read_capacity(int ctlr, int logvol, ReadCapdata_struct *buf, + int withirq, unsigned int *total_size, unsigned int *block_size) +{ + int return_code; + memset(buf, 0, sizeof(*buf)); + if (withirq) + return_code = sendcmd_withirq(CCISS_READ_CAPACITY, + ctlr, buf, sizeof(*buf), 1, logvol, 0, TYPE_CMD); + else + return_code = sendcmd(CCISS_READ_CAPACITY, + ctlr, buf, sizeof(*buf), 1, logvol, 0, NULL, TYPE_CMD); + if (return_code == IO_OK) { + *total_size = be32_to_cpu(*((__be32 *) &buf->total_size[0]))+1; + *block_size = be32_to_cpu(*((__be32 *) &buf->block_size[0])); + } else { /* read capacity command failed */ + printk(KERN_WARNING "cciss: read capacity failed\n"); + *total_size = 0; + *block_size = BLOCK_SIZE; + } + printk(KERN_INFO " blocks= %u block_size= %d\n", + *total_size, *block_size); + return; +} + +static int register_new_disk(ctlr_info_t *h) +{ + struct gendisk *disk; + int ctlr = h->ctlr; + int i; + int num_luns; + int logvol; + int new_lun_found = 0; + int new_lun_index = 0; + int free_index_found = 0; + int free_index = 0; + ReportLunData_struct *ld_buff = NULL; + ReadCapdata_struct *size_buff = NULL; + InquiryData_struct *inq_buff = NULL; + int return_code; + int listlength = 0; + __u32 lunid = 0; + unsigned int block_size; + unsigned int total_size; + + if (!capable(CAP_SYS_RAWIO)) + return -EPERM; + /* if we have no space in our disk array left to add anything */ + if( h->num_luns >= CISS_MAX_LUN) + return -EINVAL; + + ld_buff = kmalloc(sizeof(ReportLunData_struct), GFP_KERNEL); + if (ld_buff == NULL) + goto mem_msg; + memset(ld_buff, 0, sizeof(ReportLunData_struct)); + size_buff = kmalloc(sizeof( ReadCapdata_struct), GFP_KERNEL); + if (size_buff == NULL) + goto mem_msg; + inq_buff = kmalloc(sizeof( InquiryData_struct), GFP_KERNEL); + if (inq_buff == NULL) + goto mem_msg; + + return_code = sendcmd_withirq(CISS_REPORT_LOG, ctlr, ld_buff, + sizeof(ReportLunData_struct), 0, 0, 0, TYPE_CMD); + + if( return_code == IO_OK) + { + + // printk("LUN Data\n--------------------------\n"); + + listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[0])) << 24; + listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[1])) << 16; + listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[2])) << 8; + listlength |= 0xff & (unsigned int)(ld_buff->LUNListLength[3]); + } else /* reading number of logical volumes failed */ + { + printk(KERN_WARNING "cciss: report logical volume" + " command failed\n"); + listlength = 0; + goto free_err; + } + num_luns = listlength / 8; // 8 bytes pre entry + if (num_luns > CISS_MAX_LUN) + { + num_luns = CISS_MAX_LUN; + } +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "Length = %x %x %x %x = %d\n", ld_buff->LUNListLength[0], + ld_buff->LUNListLength[1], ld_buff->LUNListLength[2], + ld_buff->LUNListLength[3], num_luns); +#endif + for(i=0; i< num_luns; i++) + { + int j; + int lunID_found = 0; + + lunid = (0xff & (unsigned int)(ld_buff->LUN[i][3])) << 24; + lunid |= (0xff & (unsigned int)(ld_buff->LUN[i][2])) << 16; + lunid |= (0xff & (unsigned int)(ld_buff->LUN[i][1])) << 8; + lunid |= 0xff & (unsigned int)(ld_buff->LUN[i][0]); + + /* check to see if this is a new lun */ + for(j=0; j <= h->highest_lun; j++) + { +#ifdef CCISS_DEBUG + printk("Checking %d %x against %x\n", j,h->drv[j].LunID, + lunid); +#endif /* CCISS_DEBUG */ + if (h->drv[j].LunID == lunid) + { + lunID_found = 1; + break; + } + + } + if( lunID_found == 1) + continue; + else + { /* It is the new lun we have been looking for */ +#ifdef CCISS_DEBUG + printk("new lun found at %d\n", i); +#endif /* CCISS_DEBUG */ + new_lun_index = i; + new_lun_found = 1; + break; + } + } + if (!new_lun_found) + { + printk(KERN_WARNING "cciss: New Logical Volume not found\n"); + goto free_err; + } + /* Now find the free index */ + for(i=0; i <CISS_MAX_LUN; i++) + { +#ifdef CCISS_DEBUG + printk("Checking Index %d\n", i); +#endif /* CCISS_DEBUG */ + if(h->drv[i].LunID == 0) + { +#ifdef CCISS_DEBUG + printk("free index found at %d\n", i); +#endif /* CCISS_DEBUG */ + free_index_found = 1; + free_index = i; + break; + } + } + if (!free_index_found) + { + printk(KERN_WARNING "cciss: unable to find free slot for disk\n"); + goto free_err; + } + + logvol = free_index; + h->drv[logvol].LunID = lunid; + /* there could be gaps in lun numbers, track hightest */ + if(h->highest_lun < lunid) + h->highest_lun = logvol; + cciss_read_capacity(ctlr, logvol, size_buff, 1, + &total_size, &block_size); + cciss_geometry_inquiry(ctlr, logvol, 1, total_size, block_size, + inq_buff, &h->drv[logvol]); + h->drv[logvol].usage_count = 0; + ++h->num_luns; + /* setup partitions per disk */ + disk = h->gendisk[logvol]; + set_capacity(disk, h->drv[logvol].nr_blocks); + /* if it's the controller it's already added */ + if(logvol) + add_disk(disk); +freeret: + kfree(ld_buff); + kfree(size_buff); + kfree(inq_buff); + return (logvol); +mem_msg: + printk(KERN_ERR "cciss: out of memory\n"); +free_err: + logvol = -1; + goto freeret; +} + +static int cciss_revalidate(struct gendisk *disk) +{ + ctlr_info_t *h = get_host(disk); + drive_info_struct *drv = get_drv(disk); + int logvol; + int FOUND=0; + unsigned int block_size; + unsigned int total_size; + ReadCapdata_struct *size_buff = NULL; + InquiryData_struct *inq_buff = NULL; + + for(logvol=0; logvol < CISS_MAX_LUN; logvol++) + { + if(h->drv[logvol].LunID == drv->LunID) { + FOUND=1; + break; + } + } + + if (!FOUND) return 1; + + size_buff = kmalloc(sizeof( ReadCapdata_struct), GFP_KERNEL); + if (size_buff == NULL) + { + printk(KERN_WARNING "cciss: out of memory\n"); + return 1; + } + inq_buff = kmalloc(sizeof( InquiryData_struct), GFP_KERNEL); + if (inq_buff == NULL) + { + printk(KERN_WARNING "cciss: out of memory\n"); + kfree(size_buff); + return 1; + } + + cciss_read_capacity(h->ctlr, logvol, size_buff, 1, &total_size, &block_size); + cciss_geometry_inquiry(h->ctlr, logvol, 1, total_size, block_size, inq_buff, drv); + + blk_queue_hardsect_size(h->queue, drv->block_size); + set_capacity(disk, drv->nr_blocks); + + kfree(size_buff); + kfree(inq_buff); + return 0; +} + +/* + * Wait polling for a command to complete. + * The memory mapped FIFO is polled for the completion. + * Used only at init time, interrupts from the HBA are disabled. + */ +static unsigned long pollcomplete(int ctlr) +{ + unsigned long done; + int i; + + /* Wait (up to 20 seconds) for a command to complete */ + + for (i = 20 * HZ; i > 0; i--) { + done = hba[ctlr]->access.command_completed(hba[ctlr]); + if (done == FIFO_EMPTY) { + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(1); + } else + return (done); + } + /* Invalid address to tell caller we ran out of time */ + return 1; +} +/* + * Send a command to the controller, and wait for it to complete. + * Only used at init time. + */ +static int sendcmd( + __u8 cmd, + int ctlr, + void *buff, + size_t size, + unsigned int use_unit_num, /* 0: address the controller, + 1: address logical volume log_unit, + 2: periph device address is scsi3addr */ + unsigned int log_unit, + __u8 page_code, + unsigned char *scsi3addr, + int cmd_type) +{ + CommandList_struct *c; + int i; + unsigned long complete; + ctlr_info_t *info_p= hba[ctlr]; + u64bit buff_dma_handle; + int status; + + if ((c = cmd_alloc(info_p, 1)) == NULL) { + printk(KERN_WARNING "cciss: unable to get memory"); + return(IO_ERROR); + } + status = fill_cmd(c, cmd, ctlr, buff, size, use_unit_num, + log_unit, page_code, scsi3addr, cmd_type); + if (status != IO_OK) { + cmd_free(info_p, c, 1); + return status; + } +resend_cmd1: + /* + * Disable interrupt + */ +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "cciss: turning intr off\n"); +#endif /* CCISS_DEBUG */ + info_p->access.set_intr_mask(info_p, CCISS_INTR_OFF); + + /* Make sure there is room in the command FIFO */ + /* Actually it should be completely empty at this time. */ + for (i = 200000; i > 0; i--) + { + /* if fifo isn't full go */ + if (!(info_p->access.fifo_full(info_p))) + { + + break; + } + udelay(10); + printk(KERN_WARNING "cciss cciss%d: SendCmd FIFO full," + " waiting!\n", ctlr); + } + /* + * Send the cmd + */ + info_p->access.submit_command(info_p, c); + complete = pollcomplete(ctlr); + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "cciss: command completed\n"); +#endif /* CCISS_DEBUG */ + + if (complete != 1) { + if ( (complete & CISS_ERROR_BIT) + && (complete & ~CISS_ERROR_BIT) == c->busaddr) + { + /* if data overrun or underun on Report command + ignore it + */ + if (((c->Request.CDB[0] == CISS_REPORT_LOG) || + (c->Request.CDB[0] == CISS_REPORT_PHYS) || + (c->Request.CDB[0] == CISS_INQUIRY)) && + ((c->err_info->CommandStatus == + CMD_DATA_OVERRUN) || + (c->err_info->CommandStatus == + CMD_DATA_UNDERRUN) + )) + { + complete = c->busaddr; + } else { + if (c->err_info->CommandStatus == + CMD_UNSOLICITED_ABORT) { + printk(KERN_WARNING "cciss%d: " + "unsolicited abort %p\n", + ctlr, c); + if (c->retry_count < MAX_CMD_RETRIES) { + printk(KERN_WARNING + "cciss%d: retrying %p\n", + ctlr, c); + c->retry_count++; + /* erase the old error */ + /* information */ + memset(c->err_info, 0, + sizeof(ErrorInfo_struct)); + goto resend_cmd1; + } else { + printk(KERN_WARNING + "cciss%d: retried %p too " + "many times\n", ctlr, c); + status = IO_ERROR; + goto cleanup1; + } + } + printk(KERN_WARNING "ciss ciss%d: sendcmd" + " Error %x \n", ctlr, + c->err_info->CommandStatus); + printk(KERN_WARNING "ciss ciss%d: sendcmd" + " offensive info\n" + " size %x\n num %x value %x\n", ctlr, + c->err_info->MoreErrInfo.Invalid_Cmd.offense_size, + c->err_info->MoreErrInfo.Invalid_Cmd.offense_num, + c->err_info->MoreErrInfo.Invalid_Cmd.offense_value); + status = IO_ERROR; + goto cleanup1; + } + } + if (complete != c->busaddr) { + printk( KERN_WARNING "cciss cciss%d: SendCmd " + "Invalid command list address returned! (%lx)\n", + ctlr, complete); + status = IO_ERROR; + goto cleanup1; + } + } else { + printk( KERN_WARNING + "cciss cciss%d: SendCmd Timeout out, " + "No command list address returned!\n", + ctlr); + status = IO_ERROR; + } + +cleanup1: + /* unlock the data buffer from DMA */ + pci_unmap_single(info_p->pdev, (dma_addr_t) buff_dma_handle.val, + size, PCI_DMA_BIDIRECTIONAL); + cmd_free(info_p, c, 1); + return (status); +} +/* + * Map (physical) PCI mem into (virtual) kernel space + */ +static void __iomem *remap_pci_mem(ulong base, ulong size) +{ + ulong page_base = ((ulong) base) & PAGE_MASK; + ulong page_offs = ((ulong) base) - page_base; + void __iomem *page_remapped = ioremap(page_base, page_offs+size); + + return page_remapped ? (page_remapped + page_offs) : NULL; +} + +/* + * Takes jobs of the Q and sends them to the hardware, then puts it on + * the Q to wait for completion. + */ +static void start_io( ctlr_info_t *h) +{ + CommandList_struct *c; + + while(( c = h->reqQ) != NULL ) + { + /* can't do anything if fifo is full */ + if ((h->access.fifo_full(h))) { + printk(KERN_WARNING "cciss: fifo full\n"); + break; + } + + /* Get the frist entry from the Request Q */ + removeQ(&(h->reqQ), c); + h->Qdepth--; + + /* Tell the controller execute command */ + h->access.submit_command(h, c); + + /* Put job onto the completed Q */ + addQ (&(h->cmpQ), c); + } +} + +static inline void complete_buffers(struct bio *bio, int status) +{ + while (bio) { + struct bio *xbh = bio->bi_next; + int nr_sectors = bio_sectors(bio); + + bio->bi_next = NULL; + blk_finished_io(len); + bio_endio(bio, nr_sectors << 9, status ? 0 : -EIO); + bio = xbh; + } + +} +/* Assumes that CCISS_LOCK(h->ctlr) is held. */ +/* Zeros out the error record and then resends the command back */ +/* to the controller */ +static inline void resend_cciss_cmd( ctlr_info_t *h, CommandList_struct *c) +{ + /* erase the old error information */ + memset(c->err_info, 0, sizeof(ErrorInfo_struct)); + + /* add it to software queue and then send it to the controller */ + addQ(&(h->reqQ),c); + h->Qdepth++; + if(h->Qdepth > h->maxQsinceinit) + h->maxQsinceinit = h->Qdepth; + + start_io(h); +} +/* checks the status of the job and calls complete buffers to mark all + * buffers for the completed job. + */ +static inline void complete_command( ctlr_info_t *h, CommandList_struct *cmd, + int timeout) +{ + int status = 1; + int i; + int retry_cmd = 0; + u64bit temp64; + + if (timeout) + status = 0; + + if(cmd->err_info->CommandStatus != 0) + { /* an error has occurred */ + switch(cmd->err_info->CommandStatus) + { + unsigned char sense_key; + case CMD_TARGET_STATUS: + status = 0; + + if( cmd->err_info->ScsiStatus == 0x02) + { + printk(KERN_WARNING "cciss: cmd %p " + "has CHECK CONDITION " + " byte 2 = 0x%x\n", cmd, + cmd->err_info->SenseInfo[2] + ); + /* check the sense key */ + sense_key = 0xf & + cmd->err_info->SenseInfo[2]; + /* no status or recovered error */ + if((sense_key == 0x0) || + (sense_key == 0x1)) + { + status = 1; + } + } else + { + printk(KERN_WARNING "cciss: cmd %p " + "has SCSI Status 0x%x\n", + cmd, cmd->err_info->ScsiStatus); + } + break; + case CMD_DATA_UNDERRUN: + printk(KERN_WARNING "cciss: cmd %p has" + " completed with data underrun " + "reported\n", cmd); + break; + case CMD_DATA_OVERRUN: + printk(KERN_WARNING "cciss: cmd %p has" + " completed with data overrun " + "reported\n", cmd); + break; + case CMD_INVALID: + printk(KERN_WARNING "cciss: cmd %p is " + "reported invalid\n", cmd); + status = 0; + break; + case CMD_PROTOCOL_ERR: + printk(KERN_WARNING "cciss: cmd %p has " + "protocol error \n", cmd); + status = 0; + break; + case CMD_HARDWARE_ERR: + printk(KERN_WARNING "cciss: cmd %p had " + " hardware error\n", cmd); + status = 0; + break; + case CMD_CONNECTION_LOST: + printk(KERN_WARNING "cciss: cmd %p had " + "connection lost\n", cmd); + status=0; + break; + case CMD_ABORTED: + printk(KERN_WARNING "cciss: cmd %p was " + "aborted\n", cmd); + status=0; + break; + case CMD_ABORT_FAILED: + printk(KERN_WARNING "cciss: cmd %p reports " + "abort failed\n", cmd); + status=0; + break; + case CMD_UNSOLICITED_ABORT: + printk(KERN_WARNING "cciss%d: unsolicited " + "abort %p\n", h->ctlr, cmd); + if (cmd->retry_count < MAX_CMD_RETRIES) { + retry_cmd=1; + printk(KERN_WARNING + "cciss%d: retrying %p\n", + h->ctlr, cmd); + cmd->retry_count++; + } else + printk(KERN_WARNING + "cciss%d: %p retried too " + "many times\n", h->ctlr, cmd); + status=0; + break; + case CMD_TIMEOUT: + printk(KERN_WARNING "cciss: cmd %p timedout\n", + cmd); + status=0; + break; + default: + printk(KERN_WARNING "cciss: cmd %p returned " + "unknown status %x\n", cmd, + cmd->err_info->CommandStatus); + status=0; + } + } + /* We need to return this command */ + if(retry_cmd) { + resend_cciss_cmd(h,cmd); + return; + } + /* command did not need to be retried */ + /* unmap the DMA mapping for all the scatter gather elements */ + for(i=0; i<cmd->Header.SGList; i++) { + temp64.val32.lower = cmd->SG[i].Addr.lower; + temp64.val32.upper = cmd->SG[i].Addr.upper; + pci_unmap_page(hba[cmd->ctlr]->pdev, + temp64.val, cmd->SG[i].Len, + (cmd->Request.Type.Direction == XFER_READ) ? + PCI_DMA_FROMDEVICE : PCI_DMA_TODEVICE); + } + complete_buffers(cmd->rq->bio, status); + +#ifdef CCISS_DEBUG + printk("Done with %p\n", cmd->rq); +#endif /* CCISS_DEBUG */ + + end_that_request_last(cmd->rq); + cmd_free(h,cmd,1); +} + +/* + * Get a request and submit it to the controller. + */ +static void do_cciss_request(request_queue_t *q) +{ + ctlr_info_t *h= q->queuedata; + CommandList_struct *c; + int start_blk, seg; + struct request *creq; + u64bit temp64; + struct scatterlist tmp_sg[MAXSGENTRIES]; + drive_info_struct *drv; + int i, dir; + + /* We call start_io here in case there is a command waiting on the + * queue that has not been sent. + */ + if (blk_queue_plugged(q)) + goto startio; + +queue: + creq = elv_next_request(q); + if (!creq) + goto startio; + + if (creq->nr_phys_segments > MAXSGENTRIES) + BUG(); + + if (( c = cmd_alloc(h, 1)) == NULL) + goto full; + + blkdev_dequeue_request(creq); + + spin_unlock_irq(q->queue_lock); + + c->cmd_type = CMD_RWREQ; + c->rq = creq; + + /* fill in the request */ + drv = creq->rq_disk->private_data; + c->Header.ReplyQueue = 0; // unused in simple mode + c->Header.Tag.lower = c->busaddr; // use the physical address the cmd block for tag + c->Header.LUN.LogDev.VolId= drv->LunID; + c->Header.LUN.LogDev.Mode = 1; + c->Request.CDBLen = 10; // 12 byte commands not in FW yet; + c->Request.Type.Type = TYPE_CMD; // It is a command. + c->Request.Type.Attribute = ATTR_SIMPLE; + c->Request.Type.Direction = + (rq_data_dir(creq) == READ) ? XFER_READ: XFER_WRITE; + c->Request.Timeout = 0; // Don't time out + c->Request.CDB[0] = (rq_data_dir(creq) == READ) ? CCISS_READ : CCISS_WRITE; + start_blk = creq->sector; +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "ciss: sector =%d nr_sectors=%d\n",(int) creq->sector, + (int) creq->nr_sectors); +#endif /* CCISS_DEBUG */ + + seg = blk_rq_map_sg(q, creq, tmp_sg); + + /* get the DMA records for the setup */ + if (c->Request.Type.Direction == XFER_READ) + dir = PCI_DMA_FROMDEVICE; + else + dir = PCI_DMA_TODEVICE; + + for (i=0; i<seg; i++) + { + c->SG[i].Len = tmp_sg[i].length; + temp64.val = (__u64) pci_map_page(h->pdev, tmp_sg[i].page, + tmp_sg[i].offset, tmp_sg[i].length, + dir); + c->SG[i].Addr.lower = temp64.val32.lower; + c->SG[i].Addr.upper = temp64.val32.upper; + c->SG[i].Ext = 0; // we are not chaining + } + /* track how many SG entries we are using */ + if( seg > h->maxSG) + h->maxSG = seg; + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "cciss: Submitting %d sectors in %d segments\n", creq->nr_sectors, seg); +#endif /* CCISS_DEBUG */ + + c->Header.SGList = c->Header.SGTotal = seg; + c->Request.CDB[1]= 0; + c->Request.CDB[2]= (start_blk >> 24) & 0xff; //MSB + c->Request.CDB[3]= (start_blk >> 16) & 0xff; + c->Request.CDB[4]= (start_blk >> 8) & 0xff; + c->Request.CDB[5]= start_blk & 0xff; + c->Request.CDB[6]= 0; // (sect >> 24) & 0xff; MSB + c->Request.CDB[7]= (creq->nr_sectors >> 8) & 0xff; + c->Request.CDB[8]= creq->nr_sectors & 0xff; + c->Request.CDB[9] = c->Request.CDB[11] = c->Request.CDB[12] = 0; + + spin_lock_irq(q->queue_lock); + + addQ(&(h->reqQ),c); + h->Qdepth++; + if(h->Qdepth > h->maxQsinceinit) + h->maxQsinceinit = h->Qdepth; + + goto queue; +full: + blk_stop_queue(q); +startio: + /* We will already have the driver lock here so not need + * to lock it. + */ + start_io(h); +} + +static irqreturn_t do_cciss_intr(int irq, void *dev_id, struct pt_regs *regs) +{ + ctlr_info_t *h = dev_id; + CommandList_struct *c; + unsigned long flags; + __u32 a, a1; + int j; + int start_queue = h->next_to_run; + + /* Is this interrupt for us? */ + if (( h->access.intr_pending(h) == 0) || (h->interrupts_enabled == 0)) + return IRQ_NONE; + + /* + * If there are completed commands in the completion queue, + * we had better do something about it. + */ + spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags); + while( h->access.intr_pending(h)) + { + while((a = h->access.command_completed(h)) != FIFO_EMPTY) + { + a1 = a; + a &= ~3; + if ((c = h->cmpQ) == NULL) + { + printk(KERN_WARNING "cciss: Completion of %08lx ignored\n", (unsigned long)a1); + continue; + } + while(c->busaddr != a) { + c = c->next; + if (c == h->cmpQ) + break; + } + /* + * If we've found the command, take it off the + * completion Q and free it + */ + if (c->busaddr == a) { + removeQ(&h->cmpQ, c); + if (c->cmd_type == CMD_RWREQ) { + complete_command(h, c, 0); + } else if (c->cmd_type == CMD_IOCTL_PEND) { + complete(c->waiting); + } +# ifdef CONFIG_CISS_SCSI_TAPE + else if (c->cmd_type == CMD_SCSI) + complete_scsi_command(c, 0, a1); +# endif + continue; + } + } + } + + /* check to see if we have maxed out the number of commands that can + * be placed on the queue. If so then exit. We do this check here + * in case the interrupt we serviced was from an ioctl and did not + * free any new commands. + */ + if ((find_first_zero_bit(h->cmd_pool_bits, NR_CMDS)) == NR_CMDS) + goto cleanup; + + /* We have room on the queue for more commands. Now we need to queue + * them up. We will also keep track of the next queue to run so + * that every queue gets a chance to be started first. + */ + for (j=0; j < NWD; j++){ + int curr_queue = (start_queue + j) % NWD; + /* make sure the disk has been added and the drive is real + * because this can be called from the middle of init_one. + */ + if(!(h->gendisk[curr_queue]->queue) || + !(h->drv[curr_queue].heads)) + continue; + blk_start_queue(h->gendisk[curr_queue]->queue); + + /* check to see if we have maxed out the number of commands + * that can be placed on the queue. + */ + if ((find_first_zero_bit(h->cmd_pool_bits, NR_CMDS)) == NR_CMDS) + { + if (curr_queue == start_queue){ + h->next_to_run = (start_queue + 1) % NWD; + goto cleanup; + } else { + h->next_to_run = curr_queue; + goto cleanup; + } + } else { + curr_queue = (curr_queue + 1) % NWD; + } + } + +cleanup: + spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags); + return IRQ_HANDLED; +} + +/* + * We cannot read the structure directly, for portablity we must use + * the io functions. + * This is for debug only. + */ +#ifdef CCISS_DEBUG +static void print_cfg_table( CfgTable_struct *tb) +{ + int i; + char temp_name[17]; + + printk("Controller Configuration information\n"); + printk("------------------------------------\n"); + for(i=0;i<4;i++) + temp_name[i] = readb(&(tb->Signature[i])); + temp_name[4]='\0'; + printk(" Signature = %s\n", temp_name); + printk(" Spec Number = %d\n", readl(&(tb->SpecValence))); + printk(" Transport methods supported = 0x%x\n", + readl(&(tb-> TransportSupport))); + printk(" Transport methods active = 0x%x\n", + readl(&(tb->TransportActive))); + printk(" Requested transport Method = 0x%x\n", + readl(&(tb->HostWrite.TransportRequest))); + printk(" Coalese Interrupt Delay = 0x%x\n", + readl(&(tb->HostWrite.CoalIntDelay))); + printk(" Coalese Interrupt Count = 0x%x\n", + readl(&(tb->HostWrite.CoalIntCount))); + printk(" Max outstanding commands = 0x%d\n", + readl(&(tb->CmdsOutMax))); + printk(" Bus Types = 0x%x\n", readl(&(tb-> BusTypes))); + for(i=0;i<16;i++) + temp_name[i] = readb(&(tb->ServerName[i])); + temp_name[16] = '\0'; + printk(" Server Name = %s\n", temp_name); + printk(" Heartbeat Counter = 0x%x\n\n\n", + readl(&(tb->HeartBeat))); +} +#endif /* CCISS_DEBUG */ + +static void release_io_mem(ctlr_info_t *c) +{ + /* if IO mem was not protected do nothing */ + if( c->io_mem_addr == 0) + return; + release_region(c->io_mem_addr, c->io_mem_length); + c->io_mem_addr = 0; + c->io_mem_length = 0; +} + +static int find_PCI_BAR_index(struct pci_dev *pdev, + unsigned long pci_bar_addr) +{ + int i, offset, mem_type, bar_type; + if (pci_bar_addr == PCI_BASE_ADDRESS_0) /* looking for BAR zero? */ + return 0; + offset = 0; + for (i=0; i<DEVICE_COUNT_RESOURCE; i++) { + bar_type = pci_resource_flags(pdev, i) & + PCI_BASE_ADDRESS_SPACE; + if (bar_type == PCI_BASE_ADDRESS_SPACE_IO) + offset += 4; + else { + mem_type = pci_resource_flags(pdev, i) & + PCI_BASE_ADDRESS_MEM_TYPE_MASK; + switch (mem_type) { + case PCI_BASE_ADDRESS_MEM_TYPE_32: + case PCI_BASE_ADDRESS_MEM_TYPE_1M: + offset += 4; /* 32 bit */ + break; + case PCI_BASE_ADDRESS_MEM_TYPE_64: + offset += 8; + break; + default: /* reserved in PCI 2.2 */ + printk(KERN_WARNING "Base address is invalid\n"); + return -1; + break; + } + } + if (offset == pci_bar_addr - PCI_BASE_ADDRESS_0) + return i+1; + } + return -1; +} + +static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev) +{ + ushort subsystem_vendor_id, subsystem_device_id, command; + __u32 board_id, scratchpad = 0; + __u64 cfg_offset; + __u32 cfg_base_addr; + __u64 cfg_base_addr_index; + int i; + + /* check to see if controller has been disabled */ + /* BEFORE trying to enable it */ + (void) pci_read_config_word(pdev, PCI_COMMAND,&command); + if(!(command & 0x02)) + { + printk(KERN_WARNING "cciss: controller appears to be disabled\n"); + return(-1); + } + + if (pci_enable_device(pdev)) + { + printk(KERN_ERR "cciss: Unable to Enable PCI device\n"); + return( -1); + } + if (pci_set_dma_mask(pdev, CCISS_DMA_MASK ) != 0) + { + printk(KERN_ERR "cciss: Unable to set DMA mask\n"); + return(-1); + } + + subsystem_vendor_id = pdev->subsystem_vendor; + subsystem_device_id = pdev->subsystem_device; + board_id = (((__u32) (subsystem_device_id << 16) & 0xffff0000) | + subsystem_vendor_id); + + /* search for our IO range so we can protect it */ + for(i=0; i<DEVICE_COUNT_RESOURCE; i++) + { + /* is this an IO range */ + if( pci_resource_flags(pdev, i) & 0x01 ) { + c->io_mem_addr = pci_resource_start(pdev, i); + c->io_mem_length = pci_resource_end(pdev, i) - + pci_resource_start(pdev, i) +1; +#ifdef CCISS_DEBUG + printk("IO value found base_addr[%d] %lx %lx\n", i, + c->io_mem_addr, c->io_mem_length); +#endif /* CCISS_DEBUG */ + /* register the IO range */ + if(!request_region( c->io_mem_addr, + c->io_mem_length, "cciss")) + { + printk(KERN_WARNING "cciss I/O memory range already in use addr=%lx length=%ld\n", + c->io_mem_addr, c->io_mem_length); + c->io_mem_addr= 0; + c->io_mem_length = 0; + } + break; + } + } + +#ifdef CCISS_DEBUG + printk("command = %x\n", command); + printk("irq = %x\n", pdev->irq); + printk("board_id = %x\n", board_id); +#endif /* CCISS_DEBUG */ + + c->intr = pdev->irq; + + /* + * Memory base addr is first addr , the second points to the config + * table + */ + + c->paddr = pci_resource_start(pdev, 0); /* addressing mode bits already removed */ +#ifdef CCISS_DEBUG + printk("address 0 = %x\n", c->paddr); +#endif /* CCISS_DEBUG */ + c->vaddr = remap_pci_mem(c->paddr, 200); + + /* Wait for the board to become ready. (PCI hotplug needs this.) + * We poll for up to 120 secs, once per 100ms. */ + for (i=0; i < 1200; i++) { + scratchpad = readl(c->vaddr + SA5_SCRATCHPAD_OFFSET); + if (scratchpad == CCISS_FIRMWARE_READY) + break; + set_current_state(TASK_INTERRUPTIBLE); + schedule_timeout(HZ / 10); /* wait 100ms */ + } + if (scratchpad != CCISS_FIRMWARE_READY) { + printk(KERN_WARNING "cciss: Board not ready. Timed out.\n"); + return -1; + } + + /* get the address index number */ + cfg_base_addr = readl(c->vaddr + SA5_CTCFG_OFFSET); + cfg_base_addr &= (__u32) 0x0000ffff; +#ifdef CCISS_DEBUG + printk("cfg base address = %x\n", cfg_base_addr); +#endif /* CCISS_DEBUG */ + cfg_base_addr_index = + find_PCI_BAR_index(pdev, cfg_base_addr); +#ifdef CCISS_DEBUG + printk("cfg base address index = %x\n", cfg_base_addr_index); +#endif /* CCISS_DEBUG */ + if (cfg_base_addr_index == -1) { + printk(KERN_WARNING "cciss: Cannot find cfg_base_addr_index\n"); + release_io_mem(c); + return -1; + } + + cfg_offset = readl(c->vaddr + SA5_CTMEM_OFFSET); +#ifdef CCISS_DEBUG + printk("cfg offset = %x\n", cfg_offset); +#endif /* CCISS_DEBUG */ + c->cfgtable = remap_pci_mem(pci_resource_start(pdev, + cfg_base_addr_index) + cfg_offset, + sizeof(CfgTable_struct)); + c->board_id = board_id; + +#ifdef CCISS_DEBUG + print_cfg_table(c->cfgtable); +#endif /* CCISS_DEBUG */ + + for(i=0; i<NR_PRODUCTS; i++) { + if (board_id == products[i].board_id) { + c->product_name = products[i].product_name; + c->access = *(products[i].access); + break; + } + } + if (i == NR_PRODUCTS) { + printk(KERN_WARNING "cciss: Sorry, I don't know how" + " to access the Smart Array controller %08lx\n", + (unsigned long)board_id); + return -1; + } + if ( (readb(&c->cfgtable->Signature[0]) != 'C') || + (readb(&c->cfgtable->Signature[1]) != 'I') || + (readb(&c->cfgtable->Signature[2]) != 'S') || + (readb(&c->cfgtable->Signature[3]) != 'S') ) + { + printk("Does not appear to be a valid CISS config table\n"); + return -1; + } + +#ifdef CONFIG_X86 +{ + /* Need to enable prefetch in the SCSI core for 6400 in x86 */ + __u32 prefetch; + prefetch = readl(&(c->cfgtable->SCSI_Prefetch)); + prefetch |= 0x100; + writel(prefetch, &(c->cfgtable->SCSI_Prefetch)); +} +#endif + +#ifdef CCISS_DEBUG + printk("Trying to put board into Simple mode\n"); +#endif /* CCISS_DEBUG */ + c->max_commands = readl(&(c->cfgtable->CmdsOutMax)); + /* Update the field, and then ring the doorbell */ + writel( CFGTBL_Trans_Simple, + &(c->cfgtable->HostWrite.TransportRequest)); + writel( CFGTBL_ChangeReq, c->vaddr + SA5_DOORBELL); + + /* under certain very rare conditions, this can take awhile. + * (e.g.: hot replace a failed 144GB drive in a RAID 5 set right + * as we enter this code.) */ + for(i=0;i<MAX_CONFIG_WAIT;i++) { + if (!(readl(c->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq)) + break; + /* delay and try again */ + set_current_state(TASK_INTERRUPTIBLE); + schedule_timeout(10); + } + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "I counter got to %d %x\n", i, readl(c->vaddr + SA5_DOORBELL)); +#endif /* CCISS_DEBUG */ +#ifdef CCISS_DEBUG + print_cfg_table(c->cfgtable); +#endif /* CCISS_DEBUG */ + + if (!(readl(&(c->cfgtable->TransportActive)) & CFGTBL_Trans_Simple)) + { + printk(KERN_WARNING "cciss: unable to get board into" + " simple mode\n"); + return -1; + } + return 0; + +} + +/* + * Gets information about the local volumes attached to the controller. + */ +static void cciss_getgeometry(int cntl_num) +{ + ReportLunData_struct *ld_buff; + ReadCapdata_struct *size_buff; + InquiryData_struct *inq_buff; + int return_code; + int i; + int listlength = 0; + __u32 lunid = 0; + int block_size; + int total_size; + + ld_buff = kmalloc(sizeof(ReportLunData_struct), GFP_KERNEL); + if (ld_buff == NULL) + { + printk(KERN_ERR "cciss: out of memory\n"); + return; + } + memset(ld_buff, 0, sizeof(ReportLunData_struct)); + size_buff = kmalloc(sizeof( ReadCapdata_struct), GFP_KERNEL); + if (size_buff == NULL) + { + printk(KERN_ERR "cciss: out of memory\n"); + kfree(ld_buff); + return; + } + inq_buff = kmalloc(sizeof( InquiryData_struct), GFP_KERNEL); + if (inq_buff == NULL) + { + printk(KERN_ERR "cciss: out of memory\n"); + kfree(ld_buff); + kfree(size_buff); + return; + } + /* Get the firmware version */ + return_code = sendcmd(CISS_INQUIRY, cntl_num, inq_buff, + sizeof(InquiryData_struct), 0, 0 ,0, NULL, TYPE_CMD); + if (return_code == IO_OK) + { + hba[cntl_num]->firm_ver[0] = inq_buff->data_byte[32]; + hba[cntl_num]->firm_ver[1] = inq_buff->data_byte[33]; + hba[cntl_num]->firm_ver[2] = inq_buff->data_byte[34]; + hba[cntl_num]->firm_ver[3] = inq_buff->data_byte[35]; + } else /* send command failed */ + { + printk(KERN_WARNING "cciss: unable to determine firmware" + " version of controller\n"); + } + /* Get the number of logical volumes */ + return_code = sendcmd(CISS_REPORT_LOG, cntl_num, ld_buff, + sizeof(ReportLunData_struct), 0, 0, 0, NULL, TYPE_CMD); + + if( return_code == IO_OK) + { +#ifdef CCISS_DEBUG + printk("LUN Data\n--------------------------\n"); +#endif /* CCISS_DEBUG */ + + listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[0])) << 24; + listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[1])) << 16; + listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[2])) << 8; + listlength |= 0xff & (unsigned int)(ld_buff->LUNListLength[3]); + } else /* reading number of logical volumes failed */ + { + printk(KERN_WARNING "cciss: report logical volume" + " command failed\n"); + listlength = 0; + } + hba[cntl_num]->num_luns = listlength / 8; // 8 bytes pre entry + if (hba[cntl_num]->num_luns > CISS_MAX_LUN) + { + printk(KERN_ERR "ciss: only %d number of logical volumes supported\n", + CISS_MAX_LUN); + hba[cntl_num]->num_luns = CISS_MAX_LUN; + } +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "Length = %x %x %x %x = %d\n", ld_buff->LUNListLength[0], + ld_buff->LUNListLength[1], ld_buff->LUNListLength[2], + ld_buff->LUNListLength[3], hba[cntl_num]->num_luns); +#endif /* CCISS_DEBUG */ + + hba[cntl_num]->highest_lun = hba[cntl_num]->num_luns-1; + for(i=0; i< hba[cntl_num]->num_luns; i++) + { + + lunid = (0xff & (unsigned int)(ld_buff->LUN[i][3])) << 24; + lunid |= (0xff & (unsigned int)(ld_buff->LUN[i][2])) << 16; + lunid |= (0xff & (unsigned int)(ld_buff->LUN[i][1])) << 8; + lunid |= 0xff & (unsigned int)(ld_buff->LUN[i][0]); + + hba[cntl_num]->drv[i].LunID = lunid; + + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "LUN[%d]: %x %x %x %x = %x\n", i, + ld_buff->LUN[i][0], ld_buff->LUN[i][1],ld_buff->LUN[i][2], + ld_buff->LUN[i][3], hba[cntl_num]->drv[i].LunID); +#endif /* CCISS_DEBUG */ + cciss_read_capacity(cntl_num, i, size_buff, 0, + &total_size, &block_size); + cciss_geometry_inquiry(cntl_num, i, 0, total_size, block_size, + inq_buff, &hba[cntl_num]->drv[i]); + } + kfree(ld_buff); + kfree(size_buff); + kfree(inq_buff); +} + +/* Function to find the first free pointer into our hba[] array */ +/* Returns -1 if no free entries are left. */ +static int alloc_cciss_hba(void) +{ + struct gendisk *disk[NWD]; + int i, n; + for (n = 0; n < NWD; n++) { + disk[n] = alloc_disk(1 << NWD_SHIFT); + if (!disk[n]) + goto out; + } + + for(i=0; i< MAX_CTLR; i++) { + if (!hba[i]) { + ctlr_info_t *p; + p = kmalloc(sizeof(ctlr_info_t), GFP_KERNEL); + if (!p) + goto Enomem; + memset(p, 0, sizeof(ctlr_info_t)); + for (n = 0; n < NWD; n++) + p->gendisk[n] = disk[n]; + hba[i] = p; + return i; + } + } + printk(KERN_WARNING "cciss: This driver supports a maximum" + " of %d controllers.\n", MAX_CTLR); + goto out; +Enomem: + printk(KERN_ERR "cciss: out of memory.\n"); +out: + while (n--) + put_disk(disk[n]); + return -1; +} + +static void free_hba(int i) +{ + ctlr_info_t *p = hba[i]; + int n; + + hba[i] = NULL; + for (n = 0; n < NWD; n++) + put_disk(p->gendisk[n]); + kfree(p); +} + +/* + * This is it. Find all the controllers and register them. I really hate + * stealing all these major device numbers. + * returns the number of block devices registered. + */ +static int __devinit cciss_init_one(struct pci_dev *pdev, + const struct pci_device_id *ent) +{ + request_queue_t *q; + int i; + int j; + int rc; + + printk(KERN_DEBUG "cciss: Device 0x%x has been found at" + " bus %d dev %d func %d\n", + pdev->device, pdev->bus->number, PCI_SLOT(pdev->devfn), + PCI_FUNC(pdev->devfn)); + i = alloc_cciss_hba(); + if(i < 0) + return (-1); + if (cciss_pci_init(hba[i], pdev) != 0) + goto clean1; + + sprintf(hba[i]->devname, "cciss%d", i); + hba[i]->ctlr = i; + hba[i]->pdev = pdev; + + /* configure PCI DMA stuff */ + if (!pci_set_dma_mask(pdev, 0xffffffffffffffffULL)) + printk("cciss: using DAC cycles\n"); + else if (!pci_set_dma_mask(pdev, 0xffffffff)) + printk("cciss: not using DAC cycles\n"); + else { + printk("cciss: no suitable DMA available\n"); + goto clean1; + } + + /* + * register with the major number, or get a dynamic major number + * by passing 0 as argument. This is done for greater than + * 8 controller support. + */ + if (i < MAX_CTLR_ORIG) + hba[i]->major = MAJOR_NR + i; + rc = register_blkdev(hba[i]->major, hba[i]->devname); + if(rc == -EBUSY || rc == -EINVAL) { + printk(KERN_ERR + "cciss: Unable to get major number %d for %s " + "on hba %d\n", hba[i]->major, hba[i]->devname, i); + goto clean1; + } + else { + if (i >= MAX_CTLR_ORIG) + hba[i]->major = rc; + } + + /* make sure the board interrupts are off */ + hba[i]->access.set_intr_mask(hba[i], CCISS_INTR_OFF); + if( request_irq(hba[i]->intr, do_cciss_intr, + SA_INTERRUPT | SA_SHIRQ | SA_SAMPLE_RANDOM, + hba[i]->devname, hba[i])) { + printk(KERN_ERR "cciss: Unable to get irq %d for %s\n", + hba[i]->intr, hba[i]->devname); + goto clean2; + } + hba[i]->cmd_pool_bits = kmalloc(((NR_CMDS+BITS_PER_LONG-1)/BITS_PER_LONG)*sizeof(unsigned long), GFP_KERNEL); + hba[i]->cmd_pool = (CommandList_struct *)pci_alloc_consistent( + hba[i]->pdev, NR_CMDS * sizeof(CommandList_struct), + &(hba[i]->cmd_pool_dhandle)); + hba[i]->errinfo_pool = (ErrorInfo_struct *)pci_alloc_consistent( + hba[i]->pdev, NR_CMDS * sizeof( ErrorInfo_struct), + &(hba[i]->errinfo_pool_dhandle)); + if((hba[i]->cmd_pool_bits == NULL) + || (hba[i]->cmd_pool == NULL) + || (hba[i]->errinfo_pool == NULL)) { + printk( KERN_ERR "cciss: out of memory"); + goto clean4; + } + + spin_lock_init(&hba[i]->lock); + q = blk_init_queue(do_cciss_request, &hba[i]->lock); + if (!q) + goto clean4; + + q->backing_dev_info.ra_pages = READ_AHEAD; + hba[i]->queue = q; + q->queuedata = hba[i]; + + /* Initialize the pdev driver private data. + have it point to hba[i]. */ + pci_set_drvdata(pdev, hba[i]); + /* command and error info recs zeroed out before + they are used */ + memset(hba[i]->cmd_pool_bits, 0, ((NR_CMDS+BITS_PER_LONG-1)/BITS_PER_LONG)*sizeof(unsigned long)); + +#ifdef CCISS_DEBUG + printk(KERN_DEBUG "Scanning for drives on controller cciss%d\n",i); +#endif /* CCISS_DEBUG */ + + cciss_getgeometry(i); + + cciss_scsi_setup(i); + + /* Turn the interrupts on so we can service requests */ + hba[i]->access.set_intr_mask(hba[i], CCISS_INTR_ON); + + cciss_procinit(i); + + blk_queue_bounce_limit(q, hba[i]->pdev->dma_mask); + + /* This is a hardware imposed limit. */ + blk_queue_max_hw_segments(q, MAXSGENTRIES); + + /* This is a limit in the driver and could be eliminated. */ + blk_queue_max_phys_segments(q, MAXSGENTRIES); + + blk_queue_max_sectors(q, 512); + + + for(j=0; j<NWD; j++) { + drive_info_struct *drv = &(hba[i]->drv[j]); + struct gendisk *disk = hba[i]->gendisk[j]; + + sprintf(disk->disk_name, "cciss/c%dd%d", i, j); + sprintf(disk->devfs_name, "cciss/host%d/target%d", i, j); + disk->major = hba[i]->major; + disk->first_minor = j << NWD_SHIFT; + disk->fops = &cciss_fops; + disk->queue = hba[i]->queue; + disk->private_data = drv; + /* we must register the controller even if no disks exist */ + /* this is for the online array utilities */ + if(!drv->heads && j) + continue; + blk_queue_hardsect_size(hba[i]->queue, drv->block_size); + set_capacity(disk, drv->nr_blocks); + add_disk(disk); + } + return(1); + +clean4: + if(hba[i]->cmd_pool_bits) + kfree(hba[i]->cmd_pool_bits); + if(hba[i]->cmd_pool) + pci_free_consistent(hba[i]->pdev, + NR_CMDS * sizeof(CommandList_struct), + hba[i]->cmd_pool, hba[i]->cmd_pool_dhandle); + if(hba[i]->errinfo_pool) + pci_free_consistent(hba[i]->pdev, + NR_CMDS * sizeof( ErrorInfo_struct), + hba[i]->errinfo_pool, + hba[i]->errinfo_pool_dhandle); + free_irq(hba[i]->intr, hba[i]); +clean2: + unregister_blkdev(hba[i]->major, hba[i]->devname); +clean1: + release_io_mem(hba[i]); + free_hba(i); + return(-1); +} + +static void __devexit cciss_remove_one (struct pci_dev *pdev) +{ + ctlr_info_t *tmp_ptr; + int i, j; + char flush_buf[4]; + int return_code; + + if (pci_get_drvdata(pdev) == NULL) + { + printk( KERN_ERR "cciss: Unable to remove device \n"); + return; + } + tmp_ptr = pci_get_drvdata(pdev); + i = tmp_ptr->ctlr; + if (hba[i] == NULL) + { + printk(KERN_ERR "cciss: device appears to " + "already be removed \n"); + return; + } + /* Turn board interrupts off and send the flush cache command */ + /* sendcmd will turn off interrupt, and send the flush... + * To write all data in the battery backed cache to disks */ + memset(flush_buf, 0, 4); + return_code = sendcmd(CCISS_CACHE_FLUSH, i, flush_buf, 4, 0, 0, 0, NULL, + TYPE_CMD); + if(return_code != IO_OK) + { + printk(KERN_WARNING "Error Flushing cache on controller %d\n", + i); + } + free_irq(hba[i]->intr, hba[i]); + pci_set_drvdata(pdev, NULL); + iounmap(hba[i]->vaddr); + cciss_unregister_scsi(i); /* unhook from SCSI subsystem */ + unregister_blkdev(hba[i]->major, hba[i]->devname); + remove_proc_entry(hba[i]->devname, proc_cciss); + + /* remove it from the disk list */ + for (j = 0; j < NWD; j++) { + struct gendisk *disk = hba[i]->gendisk[j]; + if (disk->flags & GENHD_FL_UP) + del_gendisk(disk); + } + + blk_cleanup_queue(hba[i]->queue); + pci_free_consistent(hba[i]->pdev, NR_CMDS * sizeof(CommandList_struct), + hba[i]->cmd_pool, hba[i]->cmd_pool_dhandle); + pci_free_consistent(hba[i]->pdev, NR_CMDS * sizeof( ErrorInfo_struct), + hba[i]->errinfo_pool, hba[i]->errinfo_pool_dhandle); + kfree(hba[i]->cmd_pool_bits); + release_io_mem(hba[i]); + free_hba(i); +} + +static struct pci_driver cciss_pci_driver = { + .name = "cciss", + .probe = cciss_init_one, + .remove = __devexit_p(cciss_remove_one), + .id_table = cciss_pci_device_id, /* id_table */ +}; + +/* + * This is it. Register the PCI driver information for the cards we control + * the OS will call our registered routines when it finds one of our cards. + */ +static int __init cciss_init(void) +{ + printk(KERN_INFO DRIVER_NAME "\n"); + + /* Register for our PCI devices */ + return pci_module_init(&cciss_pci_driver); +} + +static void __exit cciss_cleanup(void) +{ + int i; + + pci_unregister_driver(&cciss_pci_driver); + /* double check that all controller entrys have been removed */ + for (i=0; i< MAX_CTLR; i++) + { + if (hba[i] != NULL) + { + printk(KERN_WARNING "cciss: had to remove" + " controller %d\n", i); + cciss_remove_one(hba[i]->pdev); + } + } + remove_proc_entry("cciss", proc_root_driver); +} + +module_init(cciss_init); +module_exit(cciss_cleanup); diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h new file mode 100644 index 00000000000..8fb19206edd --- /dev/null +++ b/drivers/block/cciss.h @@ -0,0 +1,266 @@ +#ifndef CCISS_H +#define CCISS_H + +#include <linux/genhd.h> + +#include "cciss_cmd.h" + + +#define NWD 16 +#define NWD_SHIFT 4 +#define MAX_PART (1 << NWD_SHIFT) + +#define IO_OK 0 +#define IO_ERROR 1 + +#define MAJOR_NR COMPAQ_CISS_MAJOR + +struct ctlr_info; +typedef struct ctlr_info ctlr_info_t; + +struct access_method { + void (*submit_command)(ctlr_info_t *h, CommandList_struct *c); + void (*set_intr_mask)(ctlr_info_t *h, unsigned long val); + unsigned long (*fifo_full)(ctlr_info_t *h); + unsigned long (*intr_pending)(ctlr_info_t *h); + unsigned long (*command_completed)(ctlr_info_t *h); +}; +typedef struct _drive_info_struct +{ + __u32 LunID; + int usage_count; + sector_t nr_blocks; + int block_size; + int heads; + int sectors; + int cylinders; + int raid_level; +} drive_info_struct; + +struct ctlr_info +{ + int ctlr; + char devname[8]; + char *product_name; + char firm_ver[4]; // Firmware version + struct pci_dev *pdev; + __u32 board_id; + void __iomem *vaddr; + unsigned long paddr; + unsigned long io_mem_addr; + unsigned long io_mem_length; + CfgTable_struct __iomem *cfgtable; + unsigned int intr; + int interrupts_enabled; + int major; + int max_commands; + int commands_outstanding; + int max_outstanding; /* Debug */ + int num_luns; + int highest_lun; + int usage_count; /* number of opens all all minor devices */ + + // information about each logical volume + drive_info_struct drv[CISS_MAX_LUN]; + + struct access_method access; + + /* queue and queue Info */ + CommandList_struct *reqQ; + CommandList_struct *cmpQ; + unsigned int Qdepth; + unsigned int maxQsinceinit; + unsigned int maxSG; + spinlock_t lock; + struct request_queue *queue; + + //* pointers to command and error info pool */ + CommandList_struct *cmd_pool; + dma_addr_t cmd_pool_dhandle; + ErrorInfo_struct *errinfo_pool; + dma_addr_t errinfo_pool_dhandle; + unsigned long *cmd_pool_bits; + int nr_allocs; + int nr_frees; + int busy_configuring; + + /* This element holds the zero based queue number of the last + * queue to be started. It is used for fairness. + */ + int next_to_run; + + // Disk structures we need to pass back + struct gendisk *gendisk[NWD]; +#ifdef CONFIG_CISS_SCSI_TAPE + void *scsi_ctlr; /* ptr to structure containing scsi related stuff */ +#endif +}; + +/* Defining the diffent access_menthods */ +/* + * Memory mapped FIFO interface (SMART 53xx cards) + */ +#define SA5_DOORBELL 0x20 +#define SA5_REQUEST_PORT_OFFSET 0x40 +#define SA5_REPLY_INTR_MASK_OFFSET 0x34 +#define SA5_REPLY_PORT_OFFSET 0x44 +#define SA5_INTR_STATUS 0x30 +#define SA5_SCRATCHPAD_OFFSET 0xB0 + +#define SA5_CTCFG_OFFSET 0xB4 +#define SA5_CTMEM_OFFSET 0xB8 + +#define SA5_INTR_OFF 0x08 +#define SA5B_INTR_OFF 0x04 +#define SA5_INTR_PENDING 0x08 +#define SA5B_INTR_PENDING 0x04 +#define FIFO_EMPTY 0xffffffff +#define CCISS_FIRMWARE_READY 0xffff0000 /* value in scratchpad register */ + +#define CISS_ERROR_BIT 0x02 + +#define CCISS_INTR_ON 1 +#define CCISS_INTR_OFF 0 +/* + Send the command to the hardware +*/ +static void SA5_submit_command( ctlr_info_t *h, CommandList_struct *c) +{ +#ifdef CCISS_DEBUG + printk("Sending %x - down to controller\n", c->busaddr ); +#endif /* CCISS_DEBUG */ + writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET); + h->commands_outstanding++; + if ( h->commands_outstanding > h->max_outstanding) + h->max_outstanding = h->commands_outstanding; +} + +/* + * This card is the opposite of the other cards. + * 0 turns interrupts on... + * 0x08 turns them off... + */ +static void SA5_intr_mask(ctlr_info_t *h, unsigned long val) +{ + if (val) + { /* Turn interrupts on */ + h->interrupts_enabled = 1; + writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + } else /* Turn them off */ + { + h->interrupts_enabled = 0; + writel( SA5_INTR_OFF, + h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + } +} +/* + * This card is the opposite of the other cards. + * 0 turns interrupts on... + * 0x04 turns them off... + */ +static void SA5B_intr_mask(ctlr_info_t *h, unsigned long val) +{ + if (val) + { /* Turn interrupts on */ + h->interrupts_enabled = 1; + writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + } else /* Turn them off */ + { + h->interrupts_enabled = 0; + writel( SA5B_INTR_OFF, + h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + } +} +/* + * Returns true if fifo is full. + * + */ +static unsigned long SA5_fifo_full(ctlr_info_t *h) +{ + if( h->commands_outstanding >= h->max_commands) + return(1); + else + return(0); + +} +/* + * returns value read from hardware. + * returns FIFO_EMPTY if there is nothing to read + */ +static unsigned long SA5_completed(ctlr_info_t *h) +{ + unsigned long register_value + = readl(h->vaddr + SA5_REPLY_PORT_OFFSET); + if(register_value != FIFO_EMPTY) + { + h->commands_outstanding--; +#ifdef CCISS_DEBUG + printk("cciss: Read %lx back from board\n", register_value); +#endif /* CCISS_DEBUG */ + } +#ifdef CCISS_DEBUG + else + { + printk("cciss: FIFO Empty read\n"); + } +#endif + return ( register_value); + +} +/* + * Returns true if an interrupt is pending.. + */ +static unsigned long SA5_intr_pending(ctlr_info_t *h) +{ + unsigned long register_value = + readl(h->vaddr + SA5_INTR_STATUS); +#ifdef CCISS_DEBUG + printk("cciss: intr_pending %lx\n", register_value); +#endif /* CCISS_DEBUG */ + if( register_value & SA5_INTR_PENDING) + return 1; + return 0 ; +} + +/* + * Returns true if an interrupt is pending.. + */ +static unsigned long SA5B_intr_pending(ctlr_info_t *h) +{ + unsigned long register_value = + readl(h->vaddr + SA5_INTR_STATUS); +#ifdef CCISS_DEBUG + printk("cciss: intr_pending %lx\n", register_value); +#endif /* CCISS_DEBUG */ + if( register_value & SA5B_INTR_PENDING) + return 1; + return 0 ; +} + + +static struct access_method SA5_access = { + SA5_submit_command, + SA5_intr_mask, + SA5_fifo_full, + SA5_intr_pending, + SA5_completed, +}; + +static struct access_method SA5B_access = { + SA5_submit_command, + SA5B_intr_mask, + SA5_fifo_full, + SA5B_intr_pending, + SA5_completed, +}; + +struct board_type { + __u32 board_id; + char *product_name; + struct access_method *access; +}; + +#define CCISS_LOCK(i) (hba[i]->queue->queue_lock) + +#endif /* CCISS_H */ + diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h new file mode 100644 index 00000000000..a88a8881762 --- /dev/null +++ b/drivers/block/cciss_cmd.h @@ -0,0 +1,271 @@ +#ifndef CCISS_CMD_H +#define CCISS_CMD_H +//########################################################################### +//DEFINES +//########################################################################### +#define CISS_VERSION "1.00" + +//general boundary defintions +#define SENSEINFOBYTES 32//note that this value may vary between host implementations +#define MAXSGENTRIES 31 +#define MAXREPLYQS 256 + +//Command Status value +#define CMD_SUCCESS 0x0000 +#define CMD_TARGET_STATUS 0x0001 +#define CMD_DATA_UNDERRUN 0x0002 +#define CMD_DATA_OVERRUN 0x0003 +#define CMD_INVALID 0x0004 +#define CMD_PROTOCOL_ERR 0x0005 +#define CMD_HARDWARE_ERR 0x0006 +#define CMD_CONNECTION_LOST 0x0007 +#define CMD_ABORTED 0x0008 +#define CMD_ABORT_FAILED 0x0009 +#define CMD_UNSOLICITED_ABORT 0x000A +#define CMD_TIMEOUT 0x000B +#define CMD_UNABORTABLE 0x000C + +//transfer direction +#define XFER_NONE 0x00 +#define XFER_WRITE 0x01 +#define XFER_READ 0x02 +#define XFER_RSVD 0x03 + +//task attribute +#define ATTR_UNTAGGED 0x00 +#define ATTR_SIMPLE 0x04 +#define ATTR_HEADOFQUEUE 0x05 +#define ATTR_ORDERED 0x06 +#define ATTR_ACA 0x07 + +//cdb type +#define TYPE_CMD 0x00 +#define TYPE_MSG 0x01 + +//config space register offsets +#define CFG_VENDORID 0x00 +#define CFG_DEVICEID 0x02 +#define CFG_I2OBAR 0x10 +#define CFG_MEM1BAR 0x14 + +//i2o space register offsets +#define I2O_IBDB_SET 0x20 +#define I2O_IBDB_CLEAR 0x70 +#define I2O_INT_STATUS 0x30 +#define I2O_INT_MASK 0x34 +#define I2O_IBPOST_Q 0x40 +#define I2O_OBPOST_Q 0x44 + +//Configuration Table +#define CFGTBL_ChangeReq 0x00000001l +#define CFGTBL_AccCmds 0x00000001l + +#define CFGTBL_Trans_Simple 0x00000002l + +#define CFGTBL_BusType_Ultra2 0x00000001l +#define CFGTBL_BusType_Ultra3 0x00000002l +#define CFGTBL_BusType_Fibre1G 0x00000100l +#define CFGTBL_BusType_Fibre2G 0x00000200l +typedef struct _vals32 +{ + __u32 lower; + __u32 upper; +} vals32; + +typedef union _u64bit +{ + vals32 val32; + __u64 val; +} u64bit; + +// Type defs used in the following structs +#define BYTE __u8 +#define WORD __u16 +#define HWORD __u16 +#define DWORD __u32 +#define QWORD vals32 + +//########################################################################### +//STRUCTURES +//########################################################################### +#define CISS_MAX_LUN 16 +#define CISS_MAX_PHYS_LUN 1024 +// SCSI-3 Cmmands + +#pragma pack(1) + +#define CISS_INQUIRY 0x12 +//Date returned +typedef struct _InquiryData_struct +{ + BYTE data_byte[36]; +} InquiryData_struct; + +#define CISS_REPORT_LOG 0xc2 /* Report Logical LUNs */ +#define CISS_REPORT_PHYS 0xc3 /* Report Physical LUNs */ +// Data returned +typedef struct _ReportLUNdata_struct +{ + BYTE LUNListLength[4]; + DWORD reserved; + BYTE LUN[CISS_MAX_LUN][8]; +} ReportLunData_struct; + +#define CCISS_READ_CAPACITY 0x25 /* Read Capacity */ +typedef struct _ReadCapdata_struct +{ + BYTE total_size[4]; // Total size in blocks + BYTE block_size[4]; // Size of blocks in bytes +} ReadCapdata_struct; + +// 12 byte commands not implemented in firmware yet. +// #define CCISS_READ 0xa8 // Read(12) +// #define CCISS_WRITE 0xaa // Write(12) + #define CCISS_READ 0x28 // Read(10) + #define CCISS_WRITE 0x2a // Write(10) + +// BMIC commands +#define BMIC_READ 0x26 +#define BMIC_WRITE 0x27 +#define BMIC_CACHE_FLUSH 0xc2 +#define CCISS_CACHE_FLUSH 0x01 //C2 was already being used by CCISS + +//Command List Structure +typedef union _SCSI3Addr_struct { + struct { + BYTE Dev; + BYTE Bus:6; + BYTE Mode:2; // b00 + } PeripDev; + struct { + BYTE DevLSB; + BYTE DevMSB:6; + BYTE Mode:2; // b01 + } LogDev; + struct { + BYTE Dev:5; + BYTE Bus:3; + BYTE Targ:6; + BYTE Mode:2; // b10 + } LogUnit; +} SCSI3Addr_struct; + +typedef struct _PhysDevAddr_struct { + DWORD TargetId:24; + DWORD Bus:6; + DWORD Mode:2; + SCSI3Addr_struct Target[2]; //2 level target device addr +} PhysDevAddr_struct; + +typedef struct _LogDevAddr_struct { + DWORD VolId:30; + DWORD Mode:2; + BYTE reserved[4]; +} LogDevAddr_struct; + +typedef union _LUNAddr_struct { + BYTE LunAddrBytes[8]; + SCSI3Addr_struct SCSI3Lun[4]; + PhysDevAddr_struct PhysDev; + LogDevAddr_struct LogDev; +} LUNAddr_struct; + +typedef struct _CommandListHeader_struct { + BYTE ReplyQueue; + BYTE SGList; + HWORD SGTotal; + QWORD Tag; + LUNAddr_struct LUN; +} CommandListHeader_struct; +typedef struct _RequestBlock_struct { + BYTE CDBLen; + struct { + BYTE Type:3; + BYTE Attribute:3; + BYTE Direction:2; + } Type; + HWORD Timeout; + BYTE CDB[16]; +} RequestBlock_struct; +typedef struct _ErrDescriptor_struct { + QWORD Addr; + DWORD Len; +} ErrDescriptor_struct; +typedef struct _SGDescriptor_struct { + QWORD Addr; + DWORD Len; + DWORD Ext; +} SGDescriptor_struct; + +typedef union _MoreErrInfo_struct{ + struct { + BYTE Reserved[3]; + BYTE Type; + DWORD ErrorInfo; + }Common_Info; + struct{ + BYTE Reserved[2]; + BYTE offense_size;//size of offending entry + BYTE offense_num; //byte # of offense 0-base + DWORD offense_value; + }Invalid_Cmd; +}MoreErrInfo_struct; +typedef struct _ErrorInfo_struct { + BYTE ScsiStatus; + BYTE SenseLen; + HWORD CommandStatus; + DWORD ResidualCnt; + MoreErrInfo_struct MoreErrInfo; + BYTE SenseInfo[SENSEINFOBYTES]; +} ErrorInfo_struct; + +/* Command types */ +#define CMD_RWREQ 0x00 +#define CMD_IOCTL_PEND 0x01 +#define CMD_SCSI 0x03 +#define CMD_MSG_DONE 0x04 +#define CMD_MSG_TIMEOUT 0x05 + +typedef struct _CommandList_struct { + CommandListHeader_struct Header; + RequestBlock_struct Request; + ErrDescriptor_struct ErrDesc; + SGDescriptor_struct SG[MAXSGENTRIES]; + /* information associated with the command */ + __u32 busaddr; /* physical address of this record */ + ErrorInfo_struct * err_info; /* pointer to the allocated mem */ + int ctlr; + int cmd_type; + struct _CommandList_struct *prev; + struct _CommandList_struct *next; + struct request * rq; + struct completion *waiting; + int retry_count; +#ifdef CONFIG_CISS_SCSI_TAPE + void * scsi_cmd; +#endif +} CommandList_struct; + +//Configuration Table Structure +typedef struct _HostWrite_struct { + DWORD TransportRequest; + DWORD Reserved; + DWORD CoalIntDelay; + DWORD CoalIntCount; +} HostWrite_struct; + +typedef struct _CfgTable_struct { + BYTE Signature[4]; + DWORD SpecValence; + DWORD TransportSupport; + DWORD TransportActive; + HostWrite_struct HostWrite; + DWORD CmdsOutMax; + DWORD BusTypes; + DWORD Reserved; + BYTE ServerName[16]; + DWORD HeartBeat; + DWORD SCSI_Prefetch; +} CfgTable_struct; +#pragma pack() +#endif // CCISS_CMD_H diff --git a/drivers/block/cciss_scsi.c b/drivers/block/cciss_scsi.c new file mode 100644 index 00000000000..f16e3caed58 --- /dev/null +++ b/drivers/block/cciss_scsi.c @@ -0,0 +1,1417 @@ +/* + * Disk Array driver for Compaq SA53xx Controllers, SCSI Tape module + * Copyright 2001 Compaq Computer Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + * Author: Stephen M. Cameron + */ +#ifdef CONFIG_CISS_SCSI_TAPE + +/* Here we have code to present the driver as a scsi driver + as it is simultaneously presented as a block driver. The + reason for doing this is to allow access to SCSI tape drives + through the array controller. Note in particular, neither + physical nor logical disks are presented through the scsi layer. */ + +#include <scsi/scsi.h> +#include <scsi/scsi_cmnd.h> +#include <scsi/scsi_device.h> +#include <scsi/scsi_host.h> +#include <asm/atomic.h> +#include <linux/timer.h> +#include <linux/completion.h> + +#include "cciss_scsi.h" + +/* some prototypes... */ +static int sendcmd( + __u8 cmd, + int ctlr, + void *buff, + size_t size, + unsigned int use_unit_num, /* 0: address the controller, + 1: address logical volume log_unit, + 2: address is in scsi3addr */ + unsigned int log_unit, + __u8 page_code, + unsigned char *scsi3addr, + int cmd_type); + + +static int cciss_scsi_proc_info( + struct Scsi_Host *sh, + char *buffer, /* data buffer */ + char **start, /* where data in buffer starts */ + off_t offset, /* offset from start of imaginary file */ + int length, /* length of data in buffer */ + int func); /* 0 == read, 1 == write */ + +static int cciss_scsi_queue_command (struct scsi_cmnd *cmd, + void (* done)(struct scsi_cmnd *)); + +static struct cciss_scsi_hba_t ccissscsi[MAX_CTLR] = { + { .name = "cciss0", .ndevices = 0 }, + { .name = "cciss1", .ndevices = 0 }, + { .name = "cciss2", .ndevices = 0 }, + { .name = "cciss3", .ndevices = 0 }, + { .name = "cciss4", .ndevices = 0 }, + { .name = "cciss5", .ndevices = 0 }, + { .name = "cciss6", .ndevices = 0 }, + { .name = "cciss7", .ndevices = 0 }, +}; + +static struct scsi_host_template cciss_driver_template = { + .module = THIS_MODULE, + .name = "cciss", + .proc_name = "cciss", + .proc_info = cciss_scsi_proc_info, + .queuecommand = cciss_scsi_queue_command, + .can_queue = SCSI_CCISS_CAN_QUEUE, + .this_id = 7, + .sg_tablesize = MAXSGENTRIES, + .cmd_per_lun = 1, + .use_clustering = DISABLE_CLUSTERING, +}; + +#pragma pack(1) +struct cciss_scsi_cmd_stack_elem_t { + CommandList_struct cmd; + ErrorInfo_struct Err; + __u32 busaddr; +}; + +#pragma pack() + +#define CMD_STACK_SIZE (SCSI_CCISS_CAN_QUEUE * \ + CCISS_MAX_SCSI_DEVS_PER_HBA + 2) + // plus two for init time usage + +#pragma pack(1) +struct cciss_scsi_cmd_stack_t { + struct cciss_scsi_cmd_stack_elem_t *pool; + struct cciss_scsi_cmd_stack_elem_t *elem[CMD_STACK_SIZE]; + dma_addr_t cmd_pool_handle; + int top; +}; +#pragma pack() + +struct cciss_scsi_adapter_data_t { + struct Scsi_Host *scsi_host; + struct cciss_scsi_cmd_stack_t cmd_stack; + int registered; + spinlock_t lock; // to protect ccissscsi[ctlr]; +}; + +#define CPQ_TAPE_LOCK(ctlr, flags) spin_lock_irqsave( \ + &(((struct cciss_scsi_adapter_data_t *) \ + hba[ctlr]->scsi_ctlr)->lock), flags); +#define CPQ_TAPE_UNLOCK(ctlr, flags) spin_unlock_irqrestore( \ + &(((struct cciss_scsi_adapter_data_t *) \ + hba[ctlr]->scsi_ctlr)->lock), flags); + +static CommandList_struct * +scsi_cmd_alloc(ctlr_info_t *h) +{ + /* assume only one process in here at a time, locking done by caller. */ + /* use CCISS_LOCK(ctlr) */ + /* might be better to rewrite how we allocate scsi commands in a way that */ + /* needs no locking at all. */ + + /* take the top memory chunk off the stack and return it, if any. */ + struct cciss_scsi_cmd_stack_elem_t *c; + struct cciss_scsi_adapter_data_t *sa; + struct cciss_scsi_cmd_stack_t *stk; + u64bit temp64; + + sa = (struct cciss_scsi_adapter_data_t *) h->scsi_ctlr; + stk = &sa->cmd_stack; + + if (stk->top < 0) + return NULL; + c = stk->elem[stk->top]; + /* memset(c, 0, sizeof(*c)); */ + memset(&c->cmd, 0, sizeof(c->cmd)); + memset(&c->Err, 0, sizeof(c->Err)); + /* set physical addr of cmd and addr of scsi parameters */ + c->cmd.busaddr = c->busaddr; + /* (__u32) (stk->cmd_pool_handle + + (sizeof(struct cciss_scsi_cmd_stack_elem_t)*stk->top)); */ + + temp64.val = (__u64) (c->busaddr + sizeof(CommandList_struct)); + /* (__u64) (stk->cmd_pool_handle + + (sizeof(struct cciss_scsi_cmd_stack_elem_t)*stk->top) + + sizeof(CommandList_struct)); */ + stk->top--; + c->cmd.ErrDesc.Addr.lower = temp64.val32.lower; + c->cmd.ErrDesc.Addr.upper = temp64.val32.upper; + c->cmd.ErrDesc.Len = sizeof(ErrorInfo_struct); + + c->cmd.ctlr = h->ctlr; + c->cmd.err_info = &c->Err; + + return (CommandList_struct *) c; +} + +static void +scsi_cmd_free(ctlr_info_t *h, CommandList_struct *cmd) +{ + /* assume only one process in here at a time, locking done by caller. */ + /* use CCISS_LOCK(ctlr) */ + /* drop the free memory chunk on top of the stack. */ + + struct cciss_scsi_adapter_data_t *sa; + struct cciss_scsi_cmd_stack_t *stk; + + sa = (struct cciss_scsi_adapter_data_t *) h->scsi_ctlr; + stk = &sa->cmd_stack; + if (stk->top >= CMD_STACK_SIZE) { + printk("cciss: scsi_cmd_free called too many times.\n"); + BUG(); + } + stk->top++; + stk->elem[stk->top] = (struct cciss_scsi_cmd_stack_elem_t *) cmd; +} + +static int +scsi_cmd_stack_setup(int ctlr, struct cciss_scsi_adapter_data_t *sa) +{ + int i; + struct cciss_scsi_cmd_stack_t *stk; + size_t size; + + stk = &sa->cmd_stack; + size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * CMD_STACK_SIZE; + + // pci_alloc_consistent guarantees 32-bit DMA address will + // be used + + stk->pool = (struct cciss_scsi_cmd_stack_elem_t *) + pci_alloc_consistent(hba[ctlr]->pdev, size, &stk->cmd_pool_handle); + + if (stk->pool == NULL) { + printk("stk->pool is null\n"); + return -1; + } + + for (i=0; i<CMD_STACK_SIZE; i++) { + stk->elem[i] = &stk->pool[i]; + stk->elem[i]->busaddr = (__u32) (stk->cmd_pool_handle + + (sizeof(struct cciss_scsi_cmd_stack_elem_t) * i)); + } + stk->top = CMD_STACK_SIZE-1; + return 0; +} + +static void +scsi_cmd_stack_free(int ctlr) +{ + struct cciss_scsi_adapter_data_t *sa; + struct cciss_scsi_cmd_stack_t *stk; + size_t size; + + sa = (struct cciss_scsi_adapter_data_t *) hba[ctlr]->scsi_ctlr; + stk = &sa->cmd_stack; + if (stk->top != CMD_STACK_SIZE-1) { + printk( "cciss: %d scsi commands are still outstanding.\n", + CMD_STACK_SIZE - stk->top); + // BUG(); + printk("WE HAVE A BUG HERE!!! stk=0x%p\n", stk); + } + size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * CMD_STACK_SIZE; + + pci_free_consistent(hba[ctlr]->pdev, size, stk->pool, stk->cmd_pool_handle); + stk->pool = NULL; +} + +/* scsi_device_types comes from scsi.h */ +#define DEVICETYPE(n) (n<0 || n>MAX_SCSI_DEVICE_CODE) ? \ + "Unknown" : scsi_device_types[n] + +#if 0 +static int xmargin=8; +static int amargin=60; + +static void +print_bytes (unsigned char *c, int len, int hex, int ascii) +{ + + int i; + unsigned char *x; + + if (hex) + { + x = c; + for (i=0;i<len;i++) + { + if ((i % xmargin) == 0 && i>0) printk("\n"); + if ((i % xmargin) == 0) printk("0x%04x:", i); + printk(" %02x", *x); + x++; + } + printk("\n"); + } + if (ascii) + { + x = c; + for (i=0;i<len;i++) + { + if ((i % amargin) == 0 && i>0) printk("\n"); + if ((i % amargin) == 0) printk("0x%04x:", i); + if (*x > 26 && *x < 128) printk("%c", *x); + else printk("."); + x++; + } + printk("\n"); + } +} + +static void +print_cmd(CommandList_struct *cp) +{ + printk("queue:%d\n", cp->Header.ReplyQueue); + printk("sglist:%d\n", cp->Header.SGList); + printk("sgtot:%d\n", cp->Header.SGTotal); + printk("Tag:0x%08x/0x%08x\n", cp->Header.Tag.upper, + cp->Header.Tag.lower); + printk("LUN:0x%02x%02x%02x%02x%02x%02x%02x%02x\n", + cp->Header.LUN.LunAddrBytes[0], + cp->Header.LUN.LunAddrBytes[1], + cp->Header.LUN.LunAddrBytes[2], + cp->Header.LUN.LunAddrBytes[3], + cp->Header.LUN.LunAddrBytes[4], + cp->Header.LUN.LunAddrBytes[5], + cp->Header.LUN.LunAddrBytes[6], + cp->Header.LUN.LunAddrBytes[7]); + printk("CDBLen:%d\n", cp->Request.CDBLen); + printk("Type:%d\n",cp->Request.Type.Type); + printk("Attr:%d\n",cp->Request.Type.Attribute); + printk(" Dir:%d\n",cp->Request.Type.Direction); + printk("Timeout:%d\n",cp->Request.Timeout); + printk( "CDB: %02x %02x %02x %02x %02x %02x %02x %02x" + " %02x %02x %02x %02x %02x %02x %02x %02x\n", + cp->Request.CDB[0], cp->Request.CDB[1], + cp->Request.CDB[2], cp->Request.CDB[3], + cp->Request.CDB[4], cp->Request.CDB[5], + cp->Request.CDB[6], cp->Request.CDB[7], + cp->Request.CDB[8], cp->Request.CDB[9], + cp->Request.CDB[10], cp->Request.CDB[11], + cp->Request.CDB[12], cp->Request.CDB[13], + cp->Request.CDB[14], cp->Request.CDB[15]), + printk("edesc.Addr: 0x%08x/0%08x, Len = %d\n", + cp->ErrDesc.Addr.upper, cp->ErrDesc.Addr.lower, + cp->ErrDesc.Len); + printk("sgs..........Errorinfo:\n"); + printk("scsistatus:%d\n", cp->err_info->ScsiStatus); + printk("senselen:%d\n", cp->err_info->SenseLen); + printk("cmd status:%d\n", cp->err_info->CommandStatus); + printk("resid cnt:%d\n", cp->err_info->ResidualCnt); + printk("offense size:%d\n", cp->err_info->MoreErrInfo.Invalid_Cmd.offense_size); + printk("offense byte:%d\n", cp->err_info->MoreErrInfo.Invalid_Cmd.offense_num); + printk("offense value:%d\n", cp->err_info->MoreErrInfo.Invalid_Cmd.offense_value); + +} + +#endif + +static int +find_bus_target_lun(int ctlr, int *bus, int *target, int *lun) +{ + /* finds an unused bus, target, lun for a new device */ + /* assumes hba[ctlr]->scsi_ctlr->lock is held */ + int i, found=0; + unsigned char target_taken[CCISS_MAX_SCSI_DEVS_PER_HBA]; + + memset(&target_taken[0], 0, CCISS_MAX_SCSI_DEVS_PER_HBA); + + target_taken[SELF_SCSI_ID] = 1; + for (i=0;i<ccissscsi[ctlr].ndevices;i++) + target_taken[ccissscsi[ctlr].dev[i].target] = 1; + + for (i=0;i<CCISS_MAX_SCSI_DEVS_PER_HBA;i++) { + if (!target_taken[i]) { + *bus = 0; *target=i; *lun = 0; found=1; + break; + } + } + return (!found); +} + +static int +cciss_scsi_add_entry(int ctlr, int hostno, + unsigned char *scsi3addr, int devtype) +{ + /* assumes hba[ctlr]->scsi_ctlr->lock is held */ + int n = ccissscsi[ctlr].ndevices; + struct cciss_scsi_dev_t *sd; + + if (n >= CCISS_MAX_SCSI_DEVS_PER_HBA) { + printk("cciss%d: Too many devices, " + "some will be inaccessible.\n", ctlr); + return -1; + } + sd = &ccissscsi[ctlr].dev[n]; + if (find_bus_target_lun(ctlr, &sd->bus, &sd->target, &sd->lun) != 0) + return -1; + memcpy(&sd->scsi3addr[0], scsi3addr, 8); + sd->devtype = devtype; + ccissscsi[ctlr].ndevices++; + + /* initially, (before registering with scsi layer) we don't + know our hostno and we don't want to print anything first + time anyway (the scsi layer's inquiries will show that info) */ + if (hostno != -1) + printk("cciss%d: %s device c%db%dt%dl%d added.\n", + ctlr, DEVICETYPE(sd->devtype), hostno, + sd->bus, sd->target, sd->lun); + return 0; +} + +static void +cciss_scsi_remove_entry(int ctlr, int hostno, int entry) +{ + /* assumes hba[ctlr]->scsi_ctlr->lock is held */ + int i; + struct cciss_scsi_dev_t sd; + + if (entry < 0 || entry >= CCISS_MAX_SCSI_DEVS_PER_HBA) return; + sd = ccissscsi[ctlr].dev[entry]; + for (i=entry;i<ccissscsi[ctlr].ndevices-1;i++) + ccissscsi[ctlr].dev[i] = ccissscsi[ctlr].dev[i+1]; + ccissscsi[ctlr].ndevices--; + printk("cciss%d: %s device c%db%dt%dl%d removed.\n", + ctlr, DEVICETYPE(sd.devtype), hostno, + sd.bus, sd.target, sd.lun); +} + + +#define SCSI3ADDR_EQ(a,b) ( \ + (a)[7] == (b)[7] && \ + (a)[6] == (b)[6] && \ + (a)[5] == (b)[5] && \ + (a)[4] == (b)[4] && \ + (a)[3] == (b)[3] && \ + (a)[2] == (b)[2] && \ + (a)[1] == (b)[1] && \ + (a)[0] == (b)[0]) + +static int +adjust_cciss_scsi_table(int ctlr, int hostno, + struct cciss_scsi_dev_t sd[], int nsds) +{ + /* sd contains scsi3 addresses and devtypes, but + bus target and lun are not filled in. This funciton + takes what's in sd to be the current and adjusts + ccissscsi[] to be in line with what's in sd. */ + + int i,j, found, changes=0; + struct cciss_scsi_dev_t *csd; + unsigned long flags; + + CPQ_TAPE_LOCK(ctlr, flags); + + /* find any devices in ccissscsi[] that are not in + sd[] and remove them from ccissscsi[] */ + + i = 0; + while(i<ccissscsi[ctlr].ndevices) { + csd = &ccissscsi[ctlr].dev[i]; + found=0; + for (j=0;j<nsds;j++) { + if (SCSI3ADDR_EQ(sd[j].scsi3addr, + csd->scsi3addr)) { + if (sd[j].devtype == csd->devtype) + found=2; + else + found=1; + break; + } + } + + if (found == 0) { /* device no longer present. */ + changes++; + /* printk("cciss%d: %s device c%db%dt%dl%d removed.\n", + ctlr, DEVICETYPE(csd->devtype), hostno, + csd->bus, csd->target, csd->lun); */ + cciss_scsi_remove_entry(ctlr, hostno, i); + /* note, i not incremented */ + } + else if (found == 1) { /* device is different kind */ + changes++; + printk("cciss%d: device c%db%dt%dl%d type changed " + "(device type now %s).\n", + ctlr, hostno, csd->bus, csd->target, csd->lun, + DEVICETYPE(csd->devtype)); + csd->devtype = sd[j].devtype; + i++; /* so just move along. */ + } else /* device is same as it ever was, */ + i++; /* so just move along. */ + } + + /* Now, make sure every device listed in sd[] is also + listed in ccissscsi[], adding them if they aren't found */ + + for (i=0;i<nsds;i++) { + found=0; + for (j=0;j<ccissscsi[ctlr].ndevices;j++) { + csd = &ccissscsi[ctlr].dev[j]; + if (SCSI3ADDR_EQ(sd[i].scsi3addr, + csd->scsi3addr)) { + if (sd[i].devtype == csd->devtype) + found=2; /* found device */ + else + found=1; /* found a bug. */ + break; + } + } + if (!found) { + changes++; + if (cciss_scsi_add_entry(ctlr, hostno, + &sd[i].scsi3addr[0], sd[i].devtype) != 0) + break; + } else if (found == 1) { + /* should never happen... */ + changes++; + printk("cciss%d: device unexpectedly changed type\n", + ctlr); + /* but if it does happen, we just ignore that device */ + } + } + CPQ_TAPE_UNLOCK(ctlr, flags); + + if (!changes) + printk("cciss%d: No device changes detected.\n", ctlr); + + return 0; +} + +static int +lookup_scsi3addr(int ctlr, int bus, int target, int lun, char *scsi3addr) +{ + int i; + struct cciss_scsi_dev_t *sd; + unsigned long flags; + + CPQ_TAPE_LOCK(ctlr, flags); + for (i=0;i<ccissscsi[ctlr].ndevices;i++) { + sd = &ccissscsi[ctlr].dev[i]; + if (sd->bus == bus && + sd->target == target && + sd->lun == lun) { + memcpy(scsi3addr, &sd->scsi3addr[0], 8); + CPQ_TAPE_UNLOCK(ctlr, flags); + return 0; + } + } + CPQ_TAPE_UNLOCK(ctlr, flags); + return -1; +} + +static void +cciss_scsi_setup(int cntl_num) +{ + struct cciss_scsi_adapter_data_t * shba; + + ccissscsi[cntl_num].ndevices = 0; + shba = (struct cciss_scsi_adapter_data_t *) + kmalloc(sizeof(*shba), GFP_KERNEL); + if (shba == NULL) + return; + shba->scsi_host = NULL; + spin_lock_init(&shba->lock); + shba->registered = 0; + if (scsi_cmd_stack_setup(cntl_num, shba) != 0) { + kfree(shba); + shba = NULL; + } + hba[cntl_num]->scsi_ctlr = (void *) shba; + return; +} + +static void +complete_scsi_command( CommandList_struct *cp, int timeout, __u32 tag) +{ + struct scsi_cmnd *cmd; + ctlr_info_t *ctlr; + u64bit addr64; + ErrorInfo_struct *ei; + + ei = cp->err_info; + + /* First, see if it was a message rather than a command */ + if (cp->Request.Type.Type == TYPE_MSG) { + cp->cmd_type = CMD_MSG_DONE; + return; + } + + cmd = (struct scsi_cmnd *) cp->scsi_cmd; + ctlr = hba[cp->ctlr]; + + /* undo the DMA mappings */ + + if (cmd->use_sg) { + pci_unmap_sg(ctlr->pdev, + cmd->buffer, cmd->use_sg, + cmd->sc_data_direction); + } + else if (cmd->request_bufflen) { + addr64.val32.lower = cp->SG[0].Addr.lower; + addr64.val32.upper = cp->SG[0].Addr.upper; + pci_unmap_single(ctlr->pdev, (dma_addr_t) addr64.val, + cmd->request_bufflen, + cmd->sc_data_direction); + } + + cmd->result = (DID_OK << 16); /* host byte */ + cmd->result |= (COMMAND_COMPLETE << 8); /* msg byte */ + /* cmd->result |= (GOOD < 1); */ /* status byte */ + + cmd->result |= (ei->ScsiStatus); + /* printk("Scsistatus is 0x%02x\n", ei->ScsiStatus); */ + + /* copy the sense data whether we need to or not. */ + + memcpy(cmd->sense_buffer, ei->SenseInfo, + ei->SenseLen > SCSI_SENSE_BUFFERSIZE ? + SCSI_SENSE_BUFFERSIZE : + ei->SenseLen); + cmd->resid = ei->ResidualCnt; + + if(ei->CommandStatus != 0) + { /* an error has occurred */ + switch(ei->CommandStatus) + { + case CMD_TARGET_STATUS: + /* Pass it up to the upper layers... */ + if( ei->ScsiStatus) + { +#if 0 + printk(KERN_WARNING "cciss: cmd %p " + "has SCSI Status = %x\n", + cp, + ei->ScsiStatus); +#endif + cmd->result |= (ei->ScsiStatus < 1); + } + else { /* scsi status is zero??? How??? */ + + /* Ordinarily, this case should never happen, but there is a bug + in some released firmware revisions that allows it to happen + if, for example, a 4100 backplane loses power and the tape + drive is in it. We assume that it's a fatal error of some + kind because we can't show that it wasn't. We will make it + look like selection timeout since that is the most common + reason for this to occur, and it's severe enough. */ + + cmd->result = DID_NO_CONNECT << 16; + } + break; + case CMD_DATA_UNDERRUN: /* let mid layer handle it. */ + break; + case CMD_DATA_OVERRUN: + printk(KERN_WARNING "cciss: cp %p has" + " completed with data overrun " + "reported\n", cp); + break; + case CMD_INVALID: { + /* print_bytes(cp, sizeof(*cp), 1, 0); + print_cmd(cp); */ + /* We get CMD_INVALID if you address a non-existent tape drive instead + of a selection timeout (no response). You will see this if you yank + out a tape drive, then try to access it. This is kind of a shame + because it means that any other CMD_INVALID (e.g. driver bug) will + get interpreted as a missing target. */ + cmd->result = DID_NO_CONNECT << 16; + } + break; + case CMD_PROTOCOL_ERR: + printk(KERN_WARNING "cciss: cp %p has " + "protocol error \n", cp); + break; + case CMD_HARDWARE_ERR: + cmd->result = DID_ERROR << 16; + printk(KERN_WARNING "cciss: cp %p had " + " hardware error\n", cp); + break; + case CMD_CONNECTION_LOST: + cmd->result = DID_ERROR << 16; + printk(KERN_WARNING "cciss: cp %p had " + "connection lost\n", cp); + break; + case CMD_ABORTED: + cmd->result = DID_ABORT << 16; + printk(KERN_WARNING "cciss: cp %p was " + "aborted\n", cp); + break; + case CMD_ABORT_FAILED: + cmd->result = DID_ERROR << 16; + printk(KERN_WARNING "cciss: cp %p reports " + "abort failed\n", cp); + break; + case CMD_UNSOLICITED_ABORT: + cmd->result = DID_ABORT << 16; + printk(KERN_WARNING "cciss: cp %p aborted " + "do to an unsolicited abort\n", cp); + break; + case CMD_TIMEOUT: + cmd->result = DID_TIME_OUT << 16; + printk(KERN_WARNING "cciss: cp %p timedout\n", + cp); + break; + default: + cmd->result = DID_ERROR << 16; + printk(KERN_WARNING "cciss: cp %p returned " + "unknown status %x\n", cp, + ei->CommandStatus); + } + } + // printk("c:%p:c%db%dt%dl%d ", cmd, ctlr->ctlr, cmd->channel, + // cmd->target, cmd->lun); + cmd->scsi_done(cmd); + scsi_cmd_free(ctlr, cp); +} + +static int +cciss_scsi_detect(int ctlr) +{ + struct Scsi_Host *sh; + int error; + + sh = scsi_host_alloc(&cciss_driver_template, sizeof(struct ctlr_info *)); + if (sh == NULL) + goto fail; + sh->io_port = 0; // good enough? FIXME, + sh->n_io_port = 0; // I don't think we use these two... + sh->this_id = SELF_SCSI_ID; + + ((struct cciss_scsi_adapter_data_t *) + hba[ctlr]->scsi_ctlr)->scsi_host = (void *) sh; + sh->hostdata[0] = (unsigned long) hba[ctlr]; + sh->irq = hba[ctlr]->intr; + sh->unique_id = sh->irq; + error = scsi_add_host(sh, &hba[ctlr]->pdev->dev); + if (error) + goto fail_host_put; + scsi_scan_host(sh); + return 1; + + fail_host_put: + scsi_host_put(sh); + fail: + return 0; +} + +static void +cciss_unmap_one(struct pci_dev *pdev, + CommandList_struct *cp, + size_t buflen, + int data_direction) +{ + u64bit addr64; + + addr64.val32.lower = cp->SG[0].Addr.lower; + addr64.val32.upper = cp->SG[0].Addr.upper; + pci_unmap_single(pdev, (dma_addr_t) addr64.val, buflen, data_direction); +} + +static void +cciss_map_one(struct pci_dev *pdev, + CommandList_struct *cp, + unsigned char *buf, + size_t buflen, + int data_direction) +{ + __u64 addr64; + + addr64 = (__u64) pci_map_single(pdev, buf, buflen, data_direction); + cp->SG[0].Addr.lower = + (__u32) (addr64 & (__u64) 0x00000000FFFFFFFF); + cp->SG[0].Addr.upper = + (__u32) ((addr64 >> 32) & (__u64) 0x00000000FFFFFFFF); + cp->SG[0].Len = buflen; + cp->Header.SGList = (__u8) 1; /* no. SGs contig in this cmd */ + cp->Header.SGTotal = (__u16) 1; /* total sgs in this cmd list */ +} + +static int +cciss_scsi_do_simple_cmd(ctlr_info_t *c, + CommandList_struct *cp, + unsigned char *scsi3addr, + unsigned char *cdb, + unsigned char cdblen, + unsigned char *buf, int bufsize, + int direction) +{ + unsigned long flags; + DECLARE_COMPLETION(wait); + + cp->cmd_type = CMD_IOCTL_PEND; // treat this like an ioctl + cp->scsi_cmd = NULL; + cp->Header.ReplyQueue = 0; // unused in simple mode + memcpy(&cp->Header.LUN, scsi3addr, sizeof(cp->Header.LUN)); + cp->Header.Tag.lower = cp->busaddr; // Use k. address of cmd as tag + // Fill in the request block... + + /* printk("Using scsi3addr 0x%02x%0x2%0x2%0x2%0x2%0x2%0x2%0x2\n", + scsi3addr[0], scsi3addr[1], scsi3addr[2], scsi3addr[3], + scsi3addr[4], scsi3addr[5], scsi3addr[6], scsi3addr[7]); */ + + memset(cp->Request.CDB, 0, sizeof(cp->Request.CDB)); + memcpy(cp->Request.CDB, cdb, cdblen); + cp->Request.Timeout = 0; + cp->Request.CDBLen = cdblen; + cp->Request.Type.Type = TYPE_CMD; + cp->Request.Type.Attribute = ATTR_SIMPLE; + cp->Request.Type.Direction = direction; + + /* Fill in the SG list and do dma mapping */ + cciss_map_one(c->pdev, cp, (unsigned char *) buf, + bufsize, DMA_FROM_DEVICE); + + cp->waiting = &wait; + + /* Put the request on the tail of the request queue */ + spin_lock_irqsave(CCISS_LOCK(c->ctlr), flags); + addQ(&c->reqQ, cp); + c->Qdepth++; + start_io(c); + spin_unlock_irqrestore(CCISS_LOCK(c->ctlr), flags); + + wait_for_completion(&wait); + + /* undo the dma mapping */ + cciss_unmap_one(c->pdev, cp, bufsize, DMA_FROM_DEVICE); + return(0); +} + +static void +cciss_scsi_interpret_error(CommandList_struct *cp) +{ + ErrorInfo_struct *ei; + + ei = cp->err_info; + switch(ei->CommandStatus) + { + case CMD_TARGET_STATUS: + printk(KERN_WARNING "cciss: cmd %p has " + "completed with errors\n", cp); + printk(KERN_WARNING "cciss: cmd %p " + "has SCSI Status = %x\n", + cp, + ei->ScsiStatus); + if (ei->ScsiStatus == 0) + printk(KERN_WARNING + "cciss:SCSI status is abnormally zero. " + "(probably indicates selection timeout " + "reported incorrectly due to a known " + "firmware bug, circa July, 2001.)\n"); + break; + case CMD_DATA_UNDERRUN: /* let mid layer handle it. */ + printk("UNDERRUN\n"); + break; + case CMD_DATA_OVERRUN: + printk(KERN_WARNING "cciss: cp %p has" + " completed with data overrun " + "reported\n", cp); + break; + case CMD_INVALID: { + /* controller unfortunately reports SCSI passthru's */ + /* to non-existent targets as invalid commands. */ + printk(KERN_WARNING "cciss: cp %p is " + "reported invalid (probably means " + "target device no longer present)\n", + cp); + /* print_bytes((unsigned char *) cp, sizeof(*cp), 1, 0); + print_cmd(cp); */ + } + break; + case CMD_PROTOCOL_ERR: + printk(KERN_WARNING "cciss: cp %p has " + "protocol error \n", cp); + break; + case CMD_HARDWARE_ERR: + /* cmd->result = DID_ERROR << 16; */ + printk(KERN_WARNING "cciss: cp %p had " + " hardware error\n", cp); + break; + case CMD_CONNECTION_LOST: + printk(KERN_WARNING "cciss: cp %p had " + "connection lost\n", cp); + break; + case CMD_ABORTED: + printk(KERN_WARNING "cciss: cp %p was " + "aborted\n", cp); + break; + case CMD_ABORT_FAILED: + printk(KERN_WARNING "cciss: cp %p reports " + "abort failed\n", cp); + break; + case CMD_UNSOLICITED_ABORT: + printk(KERN_WARNING "cciss: cp %p aborted " + "do to an unsolicited abort\n", cp); + break; + case CMD_TIMEOUT: + printk(KERN_WARNING "cciss: cp %p timedout\n", + cp); + break; + default: + printk(KERN_WARNING "cciss: cp %p returned " + "unknown status %x\n", cp, + ei->CommandStatus); + } +} + +static int +cciss_scsi_do_inquiry(ctlr_info_t *c, unsigned char *scsi3addr, + InquiryData_struct *buf) +{ + int rc; + CommandList_struct *cp; + char cdb[6]; + ErrorInfo_struct *ei; + unsigned long flags; + + spin_lock_irqsave(CCISS_LOCK(c->ctlr), flags); + cp = scsi_cmd_alloc(c); + spin_unlock_irqrestore(CCISS_LOCK(c->ctlr), flags); + + if (cp == NULL) { /* trouble... */ + printk("cmd_alloc returned NULL!\n"); + return -1; + } + + ei = cp->err_info; + + cdb[0] = CISS_INQUIRY; + cdb[1] = 0; + cdb[2] = 0; + cdb[3] = 0; + cdb[4] = sizeof(*buf) & 0xff; + cdb[5] = 0; + rc = cciss_scsi_do_simple_cmd(c, cp, scsi3addr, cdb, + 6, (unsigned char *) buf, + sizeof(*buf), XFER_READ); + + if (rc != 0) return rc; /* something went wrong */ + + if (ei->CommandStatus != 0 && + ei->CommandStatus != CMD_DATA_UNDERRUN) { + cciss_scsi_interpret_error(cp); + rc = -1; + } + spin_lock_irqsave(CCISS_LOCK(c->ctlr), flags); + scsi_cmd_free(c, cp); + spin_unlock_irqrestore(CCISS_LOCK(c->ctlr), flags); + return rc; +} + +static int +cciss_scsi_do_report_phys_luns(ctlr_info_t *c, + ReportLunData_struct *buf, int bufsize) +{ + int rc; + CommandList_struct *cp; + unsigned char cdb[12]; + unsigned char scsi3addr[8]; + ErrorInfo_struct *ei; + unsigned long flags; + + spin_lock_irqsave(CCISS_LOCK(c->ctlr), flags); + cp = scsi_cmd_alloc(c); + spin_unlock_irqrestore(CCISS_LOCK(c->ctlr), flags); + if (cp == NULL) { /* trouble... */ + printk("cmd_alloc returned NULL!\n"); + return -1; + } + + memset(&scsi3addr[0], 0, 8); /* address the controller */ + cdb[0] = CISS_REPORT_PHYS; + cdb[1] = 0; + cdb[2] = 0; + cdb[3] = 0; + cdb[4] = 0; + cdb[5] = 0; + cdb[6] = (bufsize >> 24) & 0xFF; //MSB + cdb[7] = (bufsize >> 16) & 0xFF; + cdb[8] = (bufsize >> 8) & 0xFF; + cdb[9] = bufsize & 0xFF; + cdb[10] = 0; + cdb[11] = 0; + + rc = cciss_scsi_do_simple_cmd(c, cp, scsi3addr, + cdb, 12, + (unsigned char *) buf, + bufsize, XFER_READ); + + if (rc != 0) return rc; /* something went wrong */ + + ei = cp->err_info; + if (ei->CommandStatus != 0 && + ei->CommandStatus != CMD_DATA_UNDERRUN) { + cciss_scsi_interpret_error(cp); + rc = -1; + } + spin_lock_irqsave(CCISS_LOCK(c->ctlr), flags); + scsi_cmd_free(c, cp); + spin_unlock_irqrestore(CCISS_LOCK(c->ctlr), flags); + return rc; +} + +static void +cciss_update_non_disk_devices(int cntl_num, int hostno) +{ + /* the idea here is we could get notified from /proc + that some devices have changed, so we do a report + physical luns cmd, and adjust our list of devices + accordingly. (We can't rely on the scsi-mid layer just + doing inquiries, because the "busses" that the scsi + mid-layer probes are totally fabricated by this driver, + so new devices wouldn't show up. + + the scsi3addr's of devices won't change so long as the + adapter is not reset. That means we can rescan and + tell which devices we already know about, vs. new + devices, vs. disappearing devices. + + Also, if you yank out a tape drive, then put in a disk + in it's place, (say, a configured volume from another + array controller for instance) _don't_ poke this driver + (so it thinks it's still a tape, but _do_ poke the scsi + mid layer, so it does an inquiry... the scsi mid layer + will see the physical disk. This would be bad. Need to + think about how to prevent that. One idea would be to + snoop all scsi responses and if an inquiry repsonse comes + back that reports a disk, chuck it an return selection + timeout instead and adjust our table... Not sure i like + that though. + + */ + + ReportLunData_struct *ld_buff; + InquiryData_struct *inq_buff; + unsigned char scsi3addr[8]; + ctlr_info_t *c; + __u32 num_luns=0; + unsigned char *ch; + /* unsigned char found[CCISS_MAX_SCSI_DEVS_PER_HBA]; */ + struct cciss_scsi_dev_t currentsd[CCISS_MAX_SCSI_DEVS_PER_HBA]; + int ncurrent=0; + int reportlunsize = sizeof(*ld_buff) + CISS_MAX_PHYS_LUN * 8; + int i; + + c = (ctlr_info_t *) hba[cntl_num]; + ld_buff = kmalloc(reportlunsize, GFP_KERNEL); + if (ld_buff == NULL) { + printk(KERN_ERR "cciss: out of memory\n"); + return; + } + memset(ld_buff, 0, reportlunsize); + inq_buff = kmalloc(sizeof( InquiryData_struct), GFP_KERNEL); + if (inq_buff == NULL) { + printk(KERN_ERR "cciss: out of memory\n"); + kfree(ld_buff); + return; + } + + if (cciss_scsi_do_report_phys_luns(c, ld_buff, reportlunsize) == 0) { + ch = &ld_buff->LUNListLength[0]; + num_luns = ((ch[0]<<24) | (ch[1]<<16) | (ch[2]<<8) | ch[3]) / 8; + if (num_luns > CISS_MAX_PHYS_LUN) { + printk(KERN_WARNING + "cciss: Maximum physical LUNs (%d) exceeded. " + "%d LUNs ignored.\n", CISS_MAX_PHYS_LUN, + num_luns - CISS_MAX_PHYS_LUN); + num_luns = CISS_MAX_PHYS_LUN; + } + } + else { + printk(KERN_ERR "cciss: Report physical LUNs failed.\n"); + goto out; + } + + + /* adjust our table of devices */ + for(i=0; i<num_luns; i++) + { + int devtype; + + /* for each physical lun, do an inquiry */ + if (ld_buff->LUN[i][3] & 0xC0) continue; + memset(inq_buff, 0, sizeof(InquiryData_struct)); + memcpy(&scsi3addr[0], &ld_buff->LUN[i][0], 8); + + if (cciss_scsi_do_inquiry(hba[cntl_num], + scsi3addr, inq_buff) != 0) + { + /* Inquiry failed (msg printed already) */ + devtype = 0; /* so we will skip this device. */ + } else /* what kind of device is this? */ + devtype = (inq_buff->data_byte[0] & 0x1f); + + switch (devtype) + { + case 0x01: /* sequential access, (tape) */ + case 0x08: /* medium changer */ + if (ncurrent >= CCISS_MAX_SCSI_DEVS_PER_HBA) { + printk(KERN_INFO "cciss%d: %s ignored, " + "too many devices.\n", cntl_num, + DEVICETYPE(devtype)); + break; + } + memcpy(¤tsd[ncurrent].scsi3addr[0], + &scsi3addr[0], 8); + currentsd[ncurrent].devtype = devtype; + currentsd[ncurrent].bus = -1; + currentsd[ncurrent].target = -1; + currentsd[ncurrent].lun = -1; + ncurrent++; + break; + default: + break; + } + } + + adjust_cciss_scsi_table(cntl_num, hostno, currentsd, ncurrent); +out: + kfree(inq_buff); + kfree(ld_buff); + return; +} + +static int +is_keyword(char *ptr, int len, char *verb) // Thanks to ncr53c8xx.c +{ + int verb_len = strlen(verb); + if (len >= verb_len && !memcmp(verb,ptr,verb_len)) + return verb_len; + else + return 0; +} + +static int +cciss_scsi_user_command(int ctlr, int hostno, char *buffer, int length) +{ + int arg_len; + + if ((arg_len = is_keyword(buffer, length, "rescan")) != 0) + cciss_update_non_disk_devices(ctlr, hostno); + else + return -EINVAL; + return length; +} + + +static int +cciss_scsi_proc_info(struct Scsi_Host *sh, + char *buffer, /* data buffer */ + char **start, /* where data in buffer starts */ + off_t offset, /* offset from start of imaginary file */ + int length, /* length of data in buffer */ + int func) /* 0 == read, 1 == write */ +{ + + int buflen, datalen; + ctlr_info_t *ci; + int cntl_num; + + + ci = (ctlr_info_t *) sh->hostdata[0]; + if (ci == NULL) /* This really shouldn't ever happen. */ + return -EINVAL; + + cntl_num = ci->ctlr; /* Get our index into the hba[] array */ + + if (func == 0) { /* User is reading from /proc/scsi/ciss*?/?* */ + buflen = sprintf(buffer, "hostnum=%d\n", sh->host_no); + + datalen = buflen - offset; + if (datalen < 0) { /* they're reading past EOF. */ + datalen = 0; + *start = buffer+buflen; + } else + *start = buffer + offset; + return(datalen); + } else /* User is writing to /proc/scsi/cciss*?/?* ... */ + return cciss_scsi_user_command(cntl_num, sh->host_no, + buffer, length); +} + +/* cciss_scatter_gather takes a struct scsi_cmnd, (cmd), and does the pci + dma mapping and fills in the scatter gather entries of the + cciss command, cp. */ + +static void +cciss_scatter_gather(struct pci_dev *pdev, + CommandList_struct *cp, + struct scsi_cmnd *cmd) +{ + unsigned int use_sg, nsegs=0, len; + struct scatterlist *scatter = (struct scatterlist *) cmd->buffer; + __u64 addr64; + + /* is it just one virtual address? */ + if (!cmd->use_sg) { + if (cmd->request_bufflen) { /* anything to xfer? */ + + addr64 = (__u64) pci_map_single(pdev, + cmd->request_buffer, + cmd->request_bufflen, + cmd->sc_data_direction); + + cp->SG[0].Addr.lower = + (__u32) (addr64 & (__u64) 0x00000000FFFFFFFF); + cp->SG[0].Addr.upper = + (__u32) ((addr64 >> 32) & (__u64) 0x00000000FFFFFFFF); + cp->SG[0].Len = cmd->request_bufflen; + nsegs=1; + } + } /* else, must be a list of virtual addresses.... */ + else if (cmd->use_sg <= MAXSGENTRIES) { /* not too many addrs? */ + + use_sg = pci_map_sg(pdev, cmd->buffer, cmd->use_sg, + cmd->sc_data_direction); + + for (nsegs=0; nsegs < use_sg; nsegs++) { + addr64 = (__u64) sg_dma_address(&scatter[nsegs]); + len = sg_dma_len(&scatter[nsegs]); + cp->SG[nsegs].Addr.lower = + (__u32) (addr64 & (__u64) 0x00000000FFFFFFFF); + cp->SG[nsegs].Addr.upper = + (__u32) ((addr64 >> 32) & (__u64) 0x00000000FFFFFFFF); + cp->SG[nsegs].Len = len; + cp->SG[nsegs].Ext = 0; // we are not chaining + } + } else BUG(); + + cp->Header.SGList = (__u8) nsegs; /* no. SGs contig in this cmd */ + cp->Header.SGTotal = (__u16) nsegs; /* total sgs in this cmd list */ + return; +} + + +static int +cciss_scsi_queue_command (struct scsi_cmnd *cmd, void (* done)(struct scsi_cmnd *)) +{ + ctlr_info_t **c; + int ctlr, rc; + unsigned char scsi3addr[8]; + CommandList_struct *cp; + unsigned long flags; + + // Get the ptr to our adapter structure (hba[i]) out of cmd->host. + // We violate cmd->host privacy here. (Is there another way?) + c = (ctlr_info_t **) &cmd->device->host->hostdata[0]; + ctlr = (*c)->ctlr; + + rc = lookup_scsi3addr(ctlr, cmd->device->channel, cmd->device->id, + cmd->device->lun, scsi3addr); + if (rc != 0) { + /* the scsi nexus does not match any that we presented... */ + /* pretend to mid layer that we got selection timeout */ + cmd->result = DID_NO_CONNECT << 16; + done(cmd); + /* we might want to think about registering controller itself + as a processor device on the bus so sg binds to it. */ + return 0; + } + + /* printk("cciss_queue_command, p=%p, cmd=0x%02x, c%db%dt%dl%d\n", + cmd, cmd->cmnd[0], ctlr, cmd->channel, cmd->target, cmd->lun);*/ + // printk("q:%p:c%db%dt%dl%d ", cmd, ctlr, cmd->channel, + // cmd->target, cmd->lun); + + /* Ok, we have a reasonable scsi nexus, so send the cmd down, and + see what the device thinks of it. */ + + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + cp = scsi_cmd_alloc(*c); + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + if (cp == NULL) { /* trouble... */ + printk("scsi_cmd_alloc returned NULL!\n"); + /* FIXME: next 3 lines are -> BAD! <- */ + cmd->result = DID_NO_CONNECT << 16; + done(cmd); + return 0; + } + + // Fill in the command list header + + cmd->scsi_done = done; // save this for use by completion code + + // save cp in case we have to abort it + cmd->host_scribble = (unsigned char *) cp; + + cp->cmd_type = CMD_SCSI; + cp->scsi_cmd = cmd; + cp->Header.ReplyQueue = 0; // unused in simple mode + memcpy(&cp->Header.LUN.LunAddrBytes[0], &scsi3addr[0], 8); + cp->Header.Tag.lower = cp->busaddr; // Use k. address of cmd as tag + + // Fill in the request block... + + cp->Request.Timeout = 0; + memset(cp->Request.CDB, 0, sizeof(cp->Request.CDB)); + if (cmd->cmd_len > sizeof(cp->Request.CDB)) BUG(); + cp->Request.CDBLen = cmd->cmd_len; + memcpy(cp->Request.CDB, cmd->cmnd, cmd->cmd_len); + cp->Request.Type.Type = TYPE_CMD; + cp->Request.Type.Attribute = ATTR_SIMPLE; + switch(cmd->sc_data_direction) + { + case DMA_TO_DEVICE: cp->Request.Type.Direction = XFER_WRITE; break; + case DMA_FROM_DEVICE: cp->Request.Type.Direction = XFER_READ; break; + case DMA_NONE: cp->Request.Type.Direction = XFER_NONE; break; + case DMA_BIDIRECTIONAL: + // This can happen if a buggy application does a scsi passthru + // and sets both inlen and outlen to non-zero. ( see + // ../scsi/scsi_ioctl.c:scsi_ioctl_send_command() ) + + cp->Request.Type.Direction = XFER_RSVD; + // This is technically wrong, and cciss controllers should + // reject it with CMD_INVALID, which is the most correct + // response, but non-fibre backends appear to let it + // slide by, and give the same results as if this field + // were set correctly. Either way is acceptable for + // our purposes here. + + break; + + default: + printk("cciss: unknown data direction: %d\n", + cmd->sc_data_direction); + BUG(); + break; + } + + cciss_scatter_gather((*c)->pdev, cp, cmd); // Fill the SG list + + /* Put the request on the tail of the request queue */ + + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + addQ(&(*c)->reqQ, cp); + (*c)->Qdepth++; + start_io(*c); + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + + /* the cmd'll come back via intr handler in complete_scsi_command() */ + return 0; +} + +static void +cciss_unregister_scsi(int ctlr) +{ + struct cciss_scsi_adapter_data_t *sa; + struct cciss_scsi_cmd_stack_t *stk; + unsigned long flags; + + /* we are being forcibly unloaded, and may not refuse. */ + + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + sa = (struct cciss_scsi_adapter_data_t *) hba[ctlr]->scsi_ctlr; + stk = &sa->cmd_stack; + + /* if we weren't ever actually registered, don't unregister */ + if (sa->registered) { + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + scsi_remove_host(sa->scsi_host); + scsi_host_put(sa->scsi_host); + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + } + + /* set scsi_host to NULL so our detect routine will + find us on register */ + sa->scsi_host = NULL; + scsi_cmd_stack_free(ctlr); + kfree(sa); + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); +} + +static int +cciss_register_scsi(int ctlr) +{ + unsigned long flags; + + CPQ_TAPE_LOCK(ctlr, flags); + + /* Since this is really a block driver, the SCSI core may not be + initialized at init time, in which case, calling scsi_register_host + would hang. Instead, we do it later, via /proc filesystem + and rc scripts, when we know SCSI core is good to go. */ + + /* Only register if SCSI devices are detected. */ + if (ccissscsi[ctlr].ndevices != 0) { + ((struct cciss_scsi_adapter_data_t *) + hba[ctlr]->scsi_ctlr)->registered = 1; + CPQ_TAPE_UNLOCK(ctlr, flags); + return cciss_scsi_detect(ctlr); + } + CPQ_TAPE_UNLOCK(ctlr, flags); + printk(KERN_INFO + "cciss%d: No appropriate SCSI device detected, " + "SCSI subsystem not engaged.\n", ctlr); + return 0; +} + +static int +cciss_engage_scsi(int ctlr) +{ + struct cciss_scsi_adapter_data_t *sa; + struct cciss_scsi_cmd_stack_t *stk; + unsigned long flags; + + spin_lock_irqsave(CCISS_LOCK(ctlr), flags); + sa = (struct cciss_scsi_adapter_data_t *) hba[ctlr]->scsi_ctlr; + stk = &sa->cmd_stack; + + if (((struct cciss_scsi_adapter_data_t *) + hba[ctlr]->scsi_ctlr)->registered) { + printk("cciss%d: SCSI subsystem already engaged.\n", ctlr); + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + return ENXIO; + } + spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags); + cciss_update_non_disk_devices(ctlr, -1); + cciss_register_scsi(ctlr); + return 0; +} + +static void +cciss_proc_tape_report(int ctlr, unsigned char *buffer, off_t *pos, off_t *len) +{ + unsigned long flags; + int size; + + *pos = *pos -1; *len = *len - 1; // cut off the last trailing newline + + CPQ_TAPE_LOCK(ctlr, flags); + size = sprintf(buffer + *len, + " Sequential access devices: %d\n\n", + ccissscsi[ctlr].ndevices); + CPQ_TAPE_UNLOCK(ctlr, flags); + *pos += size; *len += size; +} + +#else /* no CONFIG_CISS_SCSI_TAPE */ + +/* If no tape support, then these become defined out of existence */ + +#define cciss_scsi_setup(cntl_num) +#define cciss_unregister_scsi(ctlr) +#define cciss_register_scsi(ctlr) +#define cciss_proc_tape_report(ctlr, buffer, pos, len) + +#endif /* CONFIG_CISS_SCSI_TAPE */ diff --git a/drivers/block/cciss_scsi.h b/drivers/block/cciss_scsi.h new file mode 100644 index 00000000000..5e7e06c07d6 --- /dev/null +++ b/drivers/block/cciss_scsi.h @@ -0,0 +1,79 @@ +/* + * Disk Array driver for Compaq SA53xx Controllers, SCSI Tape module + * Copyright 2001 Compaq Computer Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + */ +#ifdef CONFIG_CISS_SCSI_TAPE +#ifndef _CCISS_SCSI_H_ +#define _CCISS_SCSI_H_ + +#include <scsi/scsicam.h> /* possibly irrelevant, since we don't show disks */ + + // the scsi id of the adapter... +#define SELF_SCSI_ID 15 + // 15 is somewhat arbitrary, since the scsi-2 bus + // that's presented by the driver to the OS is + // fabricated. The "real" scsi-3 bus the + // hardware presents is fabricated too. + // The actual, honest-to-goodness physical + // bus that the devices are attached to is not + // addressible natively, and may in fact turn + // out to be not scsi at all. + +#define SCSI_CCISS_CAN_QUEUE 2 + +/* + +Note, cmd_per_lun could give us some trouble, so I'm setting it very low. +Likewise, SCSI_CCISS_CAN_QUEUE is set very conservatively. + +If the upper scsi layer tries to track how many commands we have +outstanding, it will be operating under the misapprehension that it is +the only one sending us requests. We also have the block interface, +which is where most requests must surely come from, so the upper layer's +notion of how many requests we have outstanding will be wrong most or +all of the time. + +Note, the normal SCSI mid-layer error handling doesn't work well +for this driver because 1) it takes the io_request_lock before +calling error handlers and uses a local variable to store flags, +so the io_request_lock cannot be released and interrupts enabled +inside the error handlers, and, the error handlers cannot poll +for command completion because they might get commands from the +block half of the driver completing, and not know what to do +with them. That's what we get for making a hybrid scsi/block +driver, I suppose. + +*/ + +struct cciss_scsi_dev_t { + int devtype; + int bus, target, lun; /* as presented to the OS */ + unsigned char scsi3addr[8]; /* as presented to the HW */ +}; + +struct cciss_scsi_hba_t { + char *name; + int ndevices; +#define CCISS_MAX_SCSI_DEVS_PER_HBA 16 + struct cciss_scsi_dev_t dev[CCISS_MAX_SCSI_DEVS_PER_HBA]; +}; + +#endif /* _CCISS_SCSI_H_ */ +#endif /* CONFIG_CISS_SCSI_TAPE */ diff --git a/drivers/block/cfq-iosched.c b/drivers/block/cfq-iosched.c new file mode 100644 index 00000000000..0ef7a0065ec --- /dev/null +++ b/drivers/block/cfq-iosched.c @@ -0,0 +1,1856 @@ +/* + * linux/drivers/block/cfq-iosched.c + * + * CFQ, or complete fairness queueing, disk scheduler. + * + * Based on ideas from a previously unfinished io + * scheduler (round robin per-process disk scheduling) and Andrea Arcangeli. + * + * Copyright (C) 2003 Jens Axboe <axboe@suse.de> + */ +#include <linux/kernel.h> +#include <linux/fs.h> +#include <linux/blkdev.h> +#include <linux/elevator.h> +#include <linux/bio.h> +#include <linux/config.h> +#include <linux/module.h> +#include <linux/slab.h> +#include <linux/init.h> +#include <linux/compiler.h> +#include <linux/hash.h> +#include <linux/rbtree.h> +#include <linux/mempool.h> + +static unsigned long max_elapsed_crq; +static unsigned long max_elapsed_dispatch; + +/* + * tunables + */ +static int cfq_quantum = 4; /* max queue in one round of service */ +static int cfq_queued = 8; /* minimum rq allocate limit per-queue*/ +static int cfq_service = HZ; /* period over which service is avg */ +static int cfq_fifo_expire_r = HZ / 2; /* fifo timeout for sync requests */ +static int cfq_fifo_expire_w = 5 * HZ; /* fifo timeout for async requests */ +static int cfq_fifo_rate = HZ / 8; /* fifo expiry rate */ +static int cfq_back_max = 16 * 1024; /* maximum backwards seek, in KiB */ +static int cfq_back_penalty = 2; /* penalty of a backwards seek */ + +/* + * for the hash of cfqq inside the cfqd + */ +#define CFQ_QHASH_SHIFT 6 +#define CFQ_QHASH_ENTRIES (1 << CFQ_QHASH_SHIFT) +#define list_entry_qhash(entry) hlist_entry((entry), struct cfq_queue, cfq_hash) + +/* + * for the hash of crq inside the cfqq + */ +#define CFQ_MHASH_SHIFT 6 +#define CFQ_MHASH_BLOCK(sec) ((sec) >> 3) +#define CFQ_MHASH_ENTRIES (1 << CFQ_MHASH_SHIFT) +#define CFQ_MHASH_FN(sec) hash_long(CFQ_MHASH_BLOCK(sec), CFQ_MHASH_SHIFT) +#define rq_hash_key(rq) ((rq)->sector + (rq)->nr_sectors) +#define list_entry_hash(ptr) hlist_entry((ptr), struct cfq_rq, hash) + +#define list_entry_cfqq(ptr) list_entry((ptr), struct cfq_queue, cfq_list) + +#define RQ_DATA(rq) (rq)->elevator_private + +/* + * rb-tree defines + */ +#define RB_NONE (2) +#define RB_EMPTY(node) ((node)->rb_node == NULL) +#define RB_CLEAR_COLOR(node) (node)->rb_color = RB_NONE +#define RB_CLEAR(node) do { \ + (node)->rb_parent = NULL; \ + RB_CLEAR_COLOR((node)); \ + (node)->rb_right = NULL; \ + (node)->rb_left = NULL; \ +} while (0) +#define RB_CLEAR_ROOT(root) ((root)->rb_node = NULL) +#define ON_RB(node) ((node)->rb_color != RB_NONE) +#define rb_entry_crq(node) rb_entry((node), struct cfq_rq, rb_node) +#define rq_rb_key(rq) (rq)->sector + +/* + * threshold for switching off non-tag accounting + */ +#define CFQ_MAX_TAG (4) + +/* + * sort key types and names + */ +enum { + CFQ_KEY_PGID, + CFQ_KEY_TGID, + CFQ_KEY_UID, + CFQ_KEY_GID, + CFQ_KEY_LAST, +}; + +static char *cfq_key_types[] = { "pgid", "tgid", "uid", "gid", NULL }; + +static kmem_cache_t *crq_pool; +static kmem_cache_t *cfq_pool; +static kmem_cache_t *cfq_ioc_pool; + +struct cfq_data { + struct list_head rr_list; + struct list_head empty_list; + + struct hlist_head *cfq_hash; + struct hlist_head *crq_hash; + + /* queues on rr_list (ie they have pending requests */ + unsigned int busy_queues; + + unsigned int max_queued; + + atomic_t ref; + + int key_type; + + mempool_t *crq_pool; + + request_queue_t *queue; + + sector_t last_sector; + + int rq_in_driver; + + /* + * tunables, see top of file + */ + unsigned int cfq_quantum; + unsigned int cfq_queued; + unsigned int cfq_fifo_expire_r; + unsigned int cfq_fifo_expire_w; + unsigned int cfq_fifo_batch_expire; + unsigned int cfq_back_penalty; + unsigned int cfq_back_max; + unsigned int find_best_crq; + + unsigned int cfq_tagged; +}; + +struct cfq_queue { + /* reference count */ + atomic_t ref; + /* parent cfq_data */ + struct cfq_data *cfqd; + /* hash of mergeable requests */ + struct hlist_node cfq_hash; + /* hash key */ + unsigned long key; + /* whether queue is on rr (or empty) list */ + int on_rr; + /* on either rr or empty list of cfqd */ + struct list_head cfq_list; + /* sorted list of pending requests */ + struct rb_root sort_list; + /* if fifo isn't expired, next request to serve */ + struct cfq_rq *next_crq; + /* requests queued in sort_list */ + int queued[2]; + /* currently allocated requests */ + int allocated[2]; + /* fifo list of requests in sort_list */ + struct list_head fifo[2]; + /* last time fifo expired */ + unsigned long last_fifo_expire; + + int key_type; + + unsigned long service_start; + unsigned long service_used; + + unsigned int max_rate; + + /* number of requests that have been handed to the driver */ + int in_flight; + /* number of currently allocated requests */ + int alloc_limit[2]; +}; + +struct cfq_rq { + struct rb_node rb_node; + sector_t rb_key; + struct request *request; + struct hlist_node hash; + + struct cfq_queue *cfq_queue; + struct cfq_io_context *io_context; + + unsigned long service_start; + unsigned long queue_start; + + unsigned int in_flight : 1; + unsigned int accounted : 1; + unsigned int is_sync : 1; + unsigned int is_write : 1; +}; + +static struct cfq_queue *cfq_find_cfq_hash(struct cfq_data *, unsigned long); +static void cfq_dispatch_sort(request_queue_t *, struct cfq_rq *); +static void cfq_update_next_crq(struct cfq_rq *); +static void cfq_put_cfqd(struct cfq_data *cfqd); + +/* + * what the fairness is based on (ie how processes are grouped and + * differentiated) + */ +static inline unsigned long +cfq_hash_key(struct cfq_data *cfqd, struct task_struct *tsk) +{ + /* + * optimize this so that ->key_type is the offset into the struct + */ + switch (cfqd->key_type) { + case CFQ_KEY_PGID: + return process_group(tsk); + default: + case CFQ_KEY_TGID: + return tsk->tgid; + case CFQ_KEY_UID: + return tsk->uid; + case CFQ_KEY_GID: + return tsk->gid; + } +} + +/* + * lots of deadline iosched dupes, can be abstracted later... + */ +static inline void cfq_del_crq_hash(struct cfq_rq *crq) +{ + hlist_del_init(&crq->hash); +} + +static void cfq_remove_merge_hints(request_queue_t *q, struct cfq_rq *crq) +{ + cfq_del_crq_hash(crq); + + if (q->last_merge == crq->request) + q->last_merge = NULL; + + cfq_update_next_crq(crq); +} + +static inline void cfq_add_crq_hash(struct cfq_data *cfqd, struct cfq_rq *crq) +{ + const int hash_idx = CFQ_MHASH_FN(rq_hash_key(crq->request)); + + BUG_ON(!hlist_unhashed(&crq->hash)); + + hlist_add_head(&crq->hash, &cfqd->crq_hash[hash_idx]); +} + +static struct request *cfq_find_rq_hash(struct cfq_data *cfqd, sector_t offset) +{ + struct hlist_head *hash_list = &cfqd->crq_hash[CFQ_MHASH_FN(offset)]; + struct hlist_node *entry, *next; + + hlist_for_each_safe(entry, next, hash_list) { + struct cfq_rq *crq = list_entry_hash(entry); + struct request *__rq = crq->request; + + BUG_ON(hlist_unhashed(&crq->hash)); + + if (!rq_mergeable(__rq)) { + cfq_del_crq_hash(crq); + continue; + } + + if (rq_hash_key(__rq) == offset) + return __rq; + } + + return NULL; +} + +/* + * Lifted from AS - choose which of crq1 and crq2 that is best served now. + * We choose the request that is closest to the head right now. Distance + * behind the head are penalized and only allowed to a certain extent. + */ +static struct cfq_rq * +cfq_choose_req(struct cfq_data *cfqd, struct cfq_rq *crq1, struct cfq_rq *crq2) +{ + sector_t last, s1, s2, d1 = 0, d2 = 0; + int r1_wrap = 0, r2_wrap = 0; /* requests are behind the disk head */ + unsigned long back_max; + + if (crq1 == NULL || crq1 == crq2) + return crq2; + if (crq2 == NULL) + return crq1; + + s1 = crq1->request->sector; + s2 = crq2->request->sector; + + last = cfqd->last_sector; + +#if 0 + if (!list_empty(&cfqd->queue->queue_head)) { + struct list_head *entry = &cfqd->queue->queue_head; + unsigned long distance = ~0UL; + struct request *rq; + + while ((entry = entry->prev) != &cfqd->queue->queue_head) { + rq = list_entry_rq(entry); + + if (blk_barrier_rq(rq)) + break; + + if (distance < abs(s1 - rq->sector + rq->nr_sectors)) { + distance = abs(s1 - rq->sector +rq->nr_sectors); + last = rq->sector + rq->nr_sectors; + } + if (distance < abs(s2 - rq->sector + rq->nr_sectors)) { + distance = abs(s2 - rq->sector +rq->nr_sectors); + last = rq->sector + rq->nr_sectors; + } + } + } +#endif + + /* + * by definition, 1KiB is 2 sectors + */ + back_max = cfqd->cfq_back_max * 2; + + /* + * Strict one way elevator _except_ in the case where we allow + * short backward seeks which are biased as twice the cost of a + * similar forward seek. + */ + if (s1 >= last) + d1 = s1 - last; + else if (s1 + back_max >= last) + d1 = (last - s1) * cfqd->cfq_back_penalty; + else + r1_wrap = 1; + + if (s2 >= last) + d2 = s2 - last; + else if (s2 + back_max >= last) + d2 = (last - s2) * cfqd->cfq_back_penalty; + else + r2_wrap = 1; + + /* Found required data */ + if (!r1_wrap && r2_wrap) + return crq1; + else if (!r2_wrap && r1_wrap) + return crq2; + else if (r1_wrap && r2_wrap) { + /* both behind the head */ + if (s1 <= s2) + return crq1; + else + return crq2; + } + + /* Both requests in front of the head */ + if (d1 < d2) + return crq1; + else if (d2 < d1) + return crq2; + else { + if (s1 >= s2) + return crq1; + else + return crq2; + } +} + +/* + * would be nice to take fifo expire time into account as well + */ +static struct cfq_rq * +cfq_find_next_crq(struct cfq_data *cfqd, struct cfq_queue *cfqq, + struct cfq_rq *last) +{ + struct cfq_rq *crq_next = NULL, *crq_prev = NULL; + struct rb_node *rbnext, *rbprev; + + if (!ON_RB(&last->rb_node)) + return NULL; + + if ((rbnext = rb_next(&last->rb_node)) == NULL) + rbnext = rb_first(&cfqq->sort_list); + + rbprev = rb_prev(&last->rb_node); + + if (rbprev) + crq_prev = rb_entry_crq(rbprev); + if (rbnext) + crq_next = rb_entry_crq(rbnext); + + return cfq_choose_req(cfqd, crq_next, crq_prev); +} + +static void cfq_update_next_crq(struct cfq_rq *crq) +{ + struct cfq_queue *cfqq = crq->cfq_queue; + + if (cfqq->next_crq == crq) + cfqq->next_crq = cfq_find_next_crq(cfqq->cfqd, cfqq, crq); +} + +static int cfq_check_sort_rr_list(struct cfq_queue *cfqq) +{ + struct list_head *head = &cfqq->cfqd->rr_list; + struct list_head *next, *prev; + + /* + * list might still be ordered + */ + next = cfqq->cfq_list.next; + if (next != head) { + struct cfq_queue *cnext = list_entry_cfqq(next); + + if (cfqq->service_used > cnext->service_used) + return 1; + } + + prev = cfqq->cfq_list.prev; + if (prev != head) { + struct cfq_queue *cprev = list_entry_cfqq(prev); + + if (cfqq->service_used < cprev->service_used) + return 1; + } + + return 0; +} + +static void cfq_sort_rr_list(struct cfq_queue *cfqq, int new_queue) +{ + struct list_head *entry = &cfqq->cfqd->rr_list; + + if (!cfqq->on_rr) + return; + if (!new_queue && !cfq_check_sort_rr_list(cfqq)) + return; + + list_del(&cfqq->cfq_list); + + /* + * sort by our mean service_used, sub-sort by in-flight requests + */ + while ((entry = entry->prev) != &cfqq->cfqd->rr_list) { + struct cfq_queue *__cfqq = list_entry_cfqq(entry); + + if (cfqq->service_used > __cfqq->service_used) + break; + else if (cfqq->service_used == __cfqq->service_used) { + struct list_head *prv; + + while ((prv = entry->prev) != &cfqq->cfqd->rr_list) { + __cfqq = list_entry_cfqq(prv); + + WARN_ON(__cfqq->service_used > cfqq->service_used); + if (cfqq->service_used != __cfqq->service_used) + break; + if (cfqq->in_flight > __cfqq->in_flight) + break; + + entry = prv; + } + } + } + + list_add(&cfqq->cfq_list, entry); +} + +/* + * add to busy list of queues for service, trying to be fair in ordering + * the pending list according to requests serviced + */ +static inline void +cfq_add_cfqq_rr(struct cfq_data *cfqd, struct cfq_queue *cfqq) +{ + /* + * it's currently on the empty list + */ + cfqq->on_rr = 1; + cfqd->busy_queues++; + + if (time_after(jiffies, cfqq->service_start + cfq_service)) + cfqq->service_used >>= 3; + + cfq_sort_rr_list(cfqq, 1); +} + +static inline void +cfq_del_cfqq_rr(struct cfq_data *cfqd, struct cfq_queue *cfqq) +{ + list_move(&cfqq->cfq_list, &cfqd->empty_list); + cfqq->on_rr = 0; + + BUG_ON(!cfqd->busy_queues); + cfqd->busy_queues--; +} + +/* + * rb tree support functions + */ +static inline void cfq_del_crq_rb(struct cfq_rq *crq) +{ + struct cfq_queue *cfqq = crq->cfq_queue; + + if (ON_RB(&crq->rb_node)) { + struct cfq_data *cfqd = cfqq->cfqd; + + BUG_ON(!cfqq->queued[crq->is_sync]); + + cfq_update_next_crq(crq); + + cfqq->queued[crq->is_sync]--; + rb_erase(&crq->rb_node, &cfqq->sort_list); + RB_CLEAR_COLOR(&crq->rb_node); + + if (RB_EMPTY(&cfqq->sort_list) && cfqq->on_rr) + cfq_del_cfqq_rr(cfqd, cfqq); + } +} + +static struct cfq_rq * +__cfq_add_crq_rb(struct cfq_rq *crq) +{ + struct rb_node **p = &crq->cfq_queue->sort_list.rb_node; + struct rb_node *parent = NULL; + struct cfq_rq *__crq; + + while (*p) { + parent = *p; + __crq = rb_entry_crq(parent); + + if (crq->rb_key < __crq->rb_key) + p = &(*p)->rb_left; + else if (crq->rb_key > __crq->rb_key) + p = &(*p)->rb_right; + else + return __crq; + } + + rb_link_node(&crq->rb_node, parent, p); + return NULL; +} + +static void cfq_add_crq_rb(struct cfq_rq *crq) +{ + struct cfq_queue *cfqq = crq->cfq_queue; + struct cfq_data *cfqd = cfqq->cfqd; + struct request *rq = crq->request; + struct cfq_rq *__alias; + + crq->rb_key = rq_rb_key(rq); + cfqq->queued[crq->is_sync]++; + + /* + * looks a little odd, but the first insert might return an alias. + * if that happens, put the alias on the dispatch list + */ + while ((__alias = __cfq_add_crq_rb(crq)) != NULL) + cfq_dispatch_sort(cfqd->queue, __alias); + + rb_insert_color(&crq->rb_node, &cfqq->sort_list); + + if (!cfqq->on_rr) + cfq_add_cfqq_rr(cfqd, cfqq); + + /* + * check if this request is a better next-serve candidate + */ + cfqq->next_crq = cfq_choose_req(cfqd, cfqq->next_crq, crq); +} + +static inline void +cfq_reposition_crq_rb(struct cfq_queue *cfqq, struct cfq_rq *crq) +{ + if (ON_RB(&crq->rb_node)) { + rb_erase(&crq->rb_node, &cfqq->sort_list); + cfqq->queued[crq->is_sync]--; + } + + cfq_add_crq_rb(crq); +} + +static struct request * +cfq_find_rq_rb(struct cfq_data *cfqd, sector_t sector) +{ + const unsigned long key = cfq_hash_key(cfqd, current); + struct cfq_queue *cfqq = cfq_find_cfq_hash(cfqd, key); + struct rb_node *n; + + if (!cfqq) + goto out; + + n = cfqq->sort_list.rb_node; + while (n) { + struct cfq_rq *crq = rb_entry_crq(n); + + if (sector < crq->rb_key) + n = n->rb_left; + else if (sector > crq->rb_key) + n = n->rb_right; + else + return crq->request; + } + +out: + return NULL; +} + +static void cfq_deactivate_request(request_queue_t *q, struct request *rq) +{ + struct cfq_rq *crq = RQ_DATA(rq); + + if (crq) { + struct cfq_queue *cfqq = crq->cfq_queue; + + if (cfqq->cfqd->cfq_tagged) { + cfqq->service_used--; + cfq_sort_rr_list(cfqq, 0); + } + + if (crq->accounted) { + crq->accounted = 0; + cfqq->cfqd->rq_in_driver--; + } + } +} + +/* + * make sure the service time gets corrected on reissue of this request + */ +static void cfq_requeue_request(request_queue_t *q, struct request *rq) +{ + cfq_deactivate_request(q, rq); + list_add(&rq->queuelist, &q->queue_head); +} + +static void cfq_remove_request(request_queue_t *q, struct request *rq) +{ + struct cfq_rq *crq = RQ_DATA(rq); + + if (crq) { + cfq_remove_merge_hints(q, crq); + list_del_init(&rq->queuelist); + + if (crq->cfq_queue) + cfq_del_crq_rb(crq); + } +} + +static int +cfq_merge(request_queue_t *q, struct request **req, struct bio *bio) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct request *__rq; + int ret; + + ret = elv_try_last_merge(q, bio); + if (ret != ELEVATOR_NO_MERGE) { + __rq = q->last_merge; + goto out_insert; + } + + __rq = cfq_find_rq_hash(cfqd, bio->bi_sector); + if (__rq) { + BUG_ON(__rq->sector + __rq->nr_sectors != bio->bi_sector); + + if (elv_rq_merge_ok(__rq, bio)) { + ret = ELEVATOR_BACK_MERGE; + goto out; + } + } + + __rq = cfq_find_rq_rb(cfqd, bio->bi_sector + bio_sectors(bio)); + if (__rq) { + if (elv_rq_merge_ok(__rq, bio)) { + ret = ELEVATOR_FRONT_MERGE; + goto out; + } + } + + return ELEVATOR_NO_MERGE; +out: + q->last_merge = __rq; +out_insert: + *req = __rq; + return ret; +} + +static void cfq_merged_request(request_queue_t *q, struct request *req) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct cfq_rq *crq = RQ_DATA(req); + + cfq_del_crq_hash(crq); + cfq_add_crq_hash(cfqd, crq); + + if (ON_RB(&crq->rb_node) && (rq_rb_key(req) != crq->rb_key)) { + struct cfq_queue *cfqq = crq->cfq_queue; + + cfq_update_next_crq(crq); + cfq_reposition_crq_rb(cfqq, crq); + } + + q->last_merge = req; +} + +static void +cfq_merged_requests(request_queue_t *q, struct request *rq, + struct request *next) +{ + struct cfq_rq *crq = RQ_DATA(rq); + struct cfq_rq *cnext = RQ_DATA(next); + + cfq_merged_request(q, rq); + + if (!list_empty(&rq->queuelist) && !list_empty(&next->queuelist)) { + if (time_before(cnext->queue_start, crq->queue_start)) { + list_move(&rq->queuelist, &next->queuelist); + crq->queue_start = cnext->queue_start; + } + } + + cfq_update_next_crq(cnext); + cfq_remove_request(q, next); +} + +/* + * we dispatch cfqd->cfq_quantum requests in total from the rr_list queues, + * this function sector sorts the selected request to minimize seeks. we start + * at cfqd->last_sector, not 0. + */ +static void cfq_dispatch_sort(request_queue_t *q, struct cfq_rq *crq) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct cfq_queue *cfqq = crq->cfq_queue; + struct list_head *head = &q->queue_head, *entry = head; + struct request *__rq; + sector_t last; + + cfq_del_crq_rb(crq); + cfq_remove_merge_hints(q, crq); + list_del(&crq->request->queuelist); + + last = cfqd->last_sector; + while ((entry = entry->prev) != head) { + __rq = list_entry_rq(entry); + + if (blk_barrier_rq(crq->request)) + break; + if (!blk_fs_request(crq->request)) + break; + + if (crq->request->sector > __rq->sector) + break; + if (__rq->sector > last && crq->request->sector < last) { + last = crq->request->sector; + break; + } + } + + cfqd->last_sector = last; + crq->in_flight = 1; + cfqq->in_flight++; + list_add(&crq->request->queuelist, entry); +} + +/* + * return expired entry, or NULL to just start from scratch in rbtree + */ +static inline struct cfq_rq *cfq_check_fifo(struct cfq_queue *cfqq) +{ + struct cfq_data *cfqd = cfqq->cfqd; + const int reads = !list_empty(&cfqq->fifo[0]); + const int writes = !list_empty(&cfqq->fifo[1]); + unsigned long now = jiffies; + struct cfq_rq *crq; + + if (time_before(now, cfqq->last_fifo_expire + cfqd->cfq_fifo_batch_expire)) + return NULL; + + crq = RQ_DATA(list_entry(cfqq->fifo[0].next, struct request, queuelist)); + if (reads && time_after(now, crq->queue_start + cfqd->cfq_fifo_expire_r)) { + cfqq->last_fifo_expire = now; + return crq; + } + + crq = RQ_DATA(list_entry(cfqq->fifo[1].next, struct request, queuelist)); + if (writes && time_after(now, crq->queue_start + cfqd->cfq_fifo_expire_w)) { + cfqq->last_fifo_expire = now; + return crq; + } + + return NULL; +} + +/* + * dispatch a single request from given queue + */ +static inline void +cfq_dispatch_request(request_queue_t *q, struct cfq_data *cfqd, + struct cfq_queue *cfqq) +{ + struct cfq_rq *crq; + + /* + * follow expired path, else get first next available + */ + if ((crq = cfq_check_fifo(cfqq)) == NULL) { + if (cfqd->find_best_crq) + crq = cfqq->next_crq; + else + crq = rb_entry_crq(rb_first(&cfqq->sort_list)); + } + + cfqd->last_sector = crq->request->sector + crq->request->nr_sectors; + + /* + * finally, insert request into driver list + */ + cfq_dispatch_sort(q, crq); +} + +static int cfq_dispatch_requests(request_queue_t *q, int max_dispatch) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct cfq_queue *cfqq; + struct list_head *entry, *tmp; + int queued, busy_queues, first_round; + + if (list_empty(&cfqd->rr_list)) + return 0; + + queued = 0; + first_round = 1; +restart: + busy_queues = 0; + list_for_each_safe(entry, tmp, &cfqd->rr_list) { + cfqq = list_entry_cfqq(entry); + + BUG_ON(RB_EMPTY(&cfqq->sort_list)); + + /* + * first round of queueing, only select from queues that + * don't already have io in-flight + */ + if (first_round && cfqq->in_flight) + continue; + + cfq_dispatch_request(q, cfqd, cfqq); + + if (!RB_EMPTY(&cfqq->sort_list)) + busy_queues++; + + queued++; + } + + if ((queued < max_dispatch) && (busy_queues || first_round)) { + first_round = 0; + goto restart; + } + + return queued; +} + +static inline void cfq_account_dispatch(struct cfq_rq *crq) +{ + struct cfq_queue *cfqq = crq->cfq_queue; + struct cfq_data *cfqd = cfqq->cfqd; + unsigned long now, elapsed; + + if (!blk_fs_request(crq->request)) + return; + + /* + * accounted bit is necessary since some drivers will call + * elv_next_request() many times for the same request (eg ide) + */ + if (crq->accounted) + return; + + now = jiffies; + if (cfqq->service_start == ~0UL) + cfqq->service_start = now; + + /* + * on drives with tagged command queueing, command turn-around time + * doesn't necessarily reflect the time spent processing this very + * command inside the drive. so do the accounting differently there, + * by just sorting on the number of requests + */ + if (cfqd->cfq_tagged) { + if (time_after(now, cfqq->service_start + cfq_service)) { + cfqq->service_start = now; + cfqq->service_used /= 10; + } + + cfqq->service_used++; + cfq_sort_rr_list(cfqq, 0); + } + + elapsed = now - crq->queue_start; + if (elapsed > max_elapsed_dispatch) + max_elapsed_dispatch = elapsed; + + crq->accounted = 1; + crq->service_start = now; + + if (++cfqd->rq_in_driver >= CFQ_MAX_TAG && !cfqd->cfq_tagged) { + cfqq->cfqd->cfq_tagged = 1; + printk("cfq: depth %d reached, tagging now on\n", CFQ_MAX_TAG); + } +} + +static inline void +cfq_account_completion(struct cfq_queue *cfqq, struct cfq_rq *crq) +{ + struct cfq_data *cfqd = cfqq->cfqd; + + if (!crq->accounted) + return; + + WARN_ON(!cfqd->rq_in_driver); + cfqd->rq_in_driver--; + + if (!cfqd->cfq_tagged) { + unsigned long now = jiffies; + unsigned long duration = now - crq->service_start; + + if (time_after(now, cfqq->service_start + cfq_service)) { + cfqq->service_start = now; + cfqq->service_used >>= 3; + } + + cfqq->service_used += duration; + cfq_sort_rr_list(cfqq, 0); + + if (duration > max_elapsed_crq) + max_elapsed_crq = duration; + } +} + +static struct request *cfq_next_request(request_queue_t *q) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct request *rq; + + if (!list_empty(&q->queue_head)) { + struct cfq_rq *crq; +dispatch: + rq = list_entry_rq(q->queue_head.next); + + if ((crq = RQ_DATA(rq)) != NULL) { + cfq_remove_merge_hints(q, crq); + cfq_account_dispatch(crq); + } + + return rq; + } + + if (cfq_dispatch_requests(q, cfqd->cfq_quantum)) + goto dispatch; + + return NULL; +} + +/* + * task holds one reference to the queue, dropped when task exits. each crq + * in-flight on this queue also holds a reference, dropped when crq is freed. + * + * queue lock must be held here. + */ +static void cfq_put_queue(struct cfq_queue *cfqq) +{ + BUG_ON(!atomic_read(&cfqq->ref)); + + if (!atomic_dec_and_test(&cfqq->ref)) + return; + + BUG_ON(rb_first(&cfqq->sort_list)); + BUG_ON(cfqq->on_rr); + + cfq_put_cfqd(cfqq->cfqd); + + /* + * it's on the empty list and still hashed + */ + list_del(&cfqq->cfq_list); + hlist_del(&cfqq->cfq_hash); + kmem_cache_free(cfq_pool, cfqq); +} + +static inline struct cfq_queue * +__cfq_find_cfq_hash(struct cfq_data *cfqd, unsigned long key, const int hashval) +{ + struct hlist_head *hash_list = &cfqd->cfq_hash[hashval]; + struct hlist_node *entry, *next; + + hlist_for_each_safe(entry, next, hash_list) { + struct cfq_queue *__cfqq = list_entry_qhash(entry); + + if (__cfqq->key == key) + return __cfqq; + } + + return NULL; +} + +static struct cfq_queue * +cfq_find_cfq_hash(struct cfq_data *cfqd, unsigned long key) +{ + return __cfq_find_cfq_hash(cfqd, key, hash_long(key, CFQ_QHASH_SHIFT)); +} + +static inline void +cfq_rehash_cfqq(struct cfq_data *cfqd, struct cfq_queue **cfqq, + struct cfq_io_context *cic) +{ + unsigned long hashkey = cfq_hash_key(cfqd, current); + unsigned long hashval = hash_long(hashkey, CFQ_QHASH_SHIFT); + struct cfq_queue *__cfqq; + unsigned long flags; + + spin_lock_irqsave(cfqd->queue->queue_lock, flags); + + hlist_del(&(*cfqq)->cfq_hash); + + __cfqq = __cfq_find_cfq_hash(cfqd, hashkey, hashval); + if (!__cfqq || __cfqq == *cfqq) { + __cfqq = *cfqq; + hlist_add_head(&__cfqq->cfq_hash, &cfqd->cfq_hash[hashval]); + __cfqq->key_type = cfqd->key_type; + } else { + atomic_inc(&__cfqq->ref); + cic->cfqq = __cfqq; + cfq_put_queue(*cfqq); + *cfqq = __cfqq; + } + + cic->cfqq = __cfqq; + spin_unlock_irqrestore(cfqd->queue->queue_lock, flags); +} + +static void cfq_free_io_context(struct cfq_io_context *cic) +{ + kmem_cache_free(cfq_ioc_pool, cic); +} + +/* + * locking hierarchy is: io_context lock -> queue locks + */ +static void cfq_exit_io_context(struct cfq_io_context *cic) +{ + struct cfq_queue *cfqq = cic->cfqq; + struct list_head *entry = &cic->list; + request_queue_t *q; + unsigned long flags; + + /* + * put the reference this task is holding to the various queues + */ + spin_lock_irqsave(&cic->ioc->lock, flags); + while ((entry = cic->list.next) != &cic->list) { + struct cfq_io_context *__cic; + + __cic = list_entry(entry, struct cfq_io_context, list); + list_del(entry); + + q = __cic->cfqq->cfqd->queue; + spin_lock(q->queue_lock); + cfq_put_queue(__cic->cfqq); + spin_unlock(q->queue_lock); + } + + q = cfqq->cfqd->queue; + spin_lock(q->queue_lock); + cfq_put_queue(cfqq); + spin_unlock(q->queue_lock); + + cic->cfqq = NULL; + spin_unlock_irqrestore(&cic->ioc->lock, flags); +} + +static struct cfq_io_context *cfq_alloc_io_context(int gfp_flags) +{ + struct cfq_io_context *cic = kmem_cache_alloc(cfq_ioc_pool, gfp_flags); + + if (cic) { + cic->dtor = cfq_free_io_context; + cic->exit = cfq_exit_io_context; + INIT_LIST_HEAD(&cic->list); + cic->cfqq = NULL; + } + + return cic; +} + +/* + * Setup general io context and cfq io context. There can be several cfq + * io contexts per general io context, if this process is doing io to more + * than one device managed by cfq. Note that caller is holding a reference to + * cfqq, so we don't need to worry about it disappearing + */ +static struct cfq_io_context * +cfq_get_io_context(struct cfq_queue **cfqq, int gfp_flags) +{ + struct cfq_data *cfqd = (*cfqq)->cfqd; + struct cfq_queue *__cfqq = *cfqq; + struct cfq_io_context *cic; + struct io_context *ioc; + + might_sleep_if(gfp_flags & __GFP_WAIT); + + ioc = get_io_context(gfp_flags); + if (!ioc) + return NULL; + + if ((cic = ioc->cic) == NULL) { + cic = cfq_alloc_io_context(gfp_flags); + + if (cic == NULL) + goto err; + + ioc->cic = cic; + cic->ioc = ioc; + cic->cfqq = __cfqq; + atomic_inc(&__cfqq->ref); + } else { + struct cfq_io_context *__cic; + unsigned long flags; + + /* + * since the first cic on the list is actually the head + * itself, need to check this here or we'll duplicate an + * cic per ioc for no reason + */ + if (cic->cfqq == __cfqq) + goto out; + + /* + * cic exists, check if we already are there. linear search + * should be ok here, the list will usually not be more than + * 1 or a few entries long + */ + spin_lock_irqsave(&ioc->lock, flags); + list_for_each_entry(__cic, &cic->list, list) { + /* + * this process is already holding a reference to + * this queue, so no need to get one more + */ + if (__cic->cfqq == __cfqq) { + cic = __cic; + spin_unlock_irqrestore(&ioc->lock, flags); + goto out; + } + } + spin_unlock_irqrestore(&ioc->lock, flags); + + /* + * nope, process doesn't have a cic assoicated with this + * cfqq yet. get a new one and add to list + */ + __cic = cfq_alloc_io_context(gfp_flags); + if (__cic == NULL) + goto err; + + __cic->ioc = ioc; + __cic->cfqq = __cfqq; + atomic_inc(&__cfqq->ref); + spin_lock_irqsave(&ioc->lock, flags); + list_add(&__cic->list, &cic->list); + spin_unlock_irqrestore(&ioc->lock, flags); + + cic = __cic; + *cfqq = __cfqq; + } + +out: + /* + * if key_type has been changed on the fly, we lazily rehash + * each queue at lookup time + */ + if ((*cfqq)->key_type != cfqd->key_type) + cfq_rehash_cfqq(cfqd, cfqq, cic); + + return cic; +err: + put_io_context(ioc); + return NULL; +} + +static struct cfq_queue * +__cfq_get_queue(struct cfq_data *cfqd, unsigned long key, int gfp_mask) +{ + const int hashval = hash_long(key, CFQ_QHASH_SHIFT); + struct cfq_queue *cfqq, *new_cfqq = NULL; + +retry: + cfqq = __cfq_find_cfq_hash(cfqd, key, hashval); + + if (!cfqq) { + if (new_cfqq) { + cfqq = new_cfqq; + new_cfqq = NULL; + } else if (gfp_mask & __GFP_WAIT) { + spin_unlock_irq(cfqd->queue->queue_lock); + new_cfqq = kmem_cache_alloc(cfq_pool, gfp_mask); + spin_lock_irq(cfqd->queue->queue_lock); + goto retry; + } else + goto out; + + memset(cfqq, 0, sizeof(*cfqq)); + + INIT_HLIST_NODE(&cfqq->cfq_hash); + INIT_LIST_HEAD(&cfqq->cfq_list); + RB_CLEAR_ROOT(&cfqq->sort_list); + INIT_LIST_HEAD(&cfqq->fifo[0]); + INIT_LIST_HEAD(&cfqq->fifo[1]); + + cfqq->key = key; + hlist_add_head(&cfqq->cfq_hash, &cfqd->cfq_hash[hashval]); + atomic_set(&cfqq->ref, 0); + cfqq->cfqd = cfqd; + atomic_inc(&cfqd->ref); + cfqq->key_type = cfqd->key_type; + cfqq->service_start = ~0UL; + } + + if (new_cfqq) + kmem_cache_free(cfq_pool, new_cfqq); + + atomic_inc(&cfqq->ref); +out: + WARN_ON((gfp_mask & __GFP_WAIT) && !cfqq); + return cfqq; +} + +static void cfq_enqueue(struct cfq_data *cfqd, struct cfq_rq *crq) +{ + crq->is_sync = 0; + if (rq_data_dir(crq->request) == READ || current->flags & PF_SYNCWRITE) + crq->is_sync = 1; + + cfq_add_crq_rb(crq); + crq->queue_start = jiffies; + + list_add_tail(&crq->request->queuelist, &crq->cfq_queue->fifo[crq->is_sync]); +} + +static void +cfq_insert_request(request_queue_t *q, struct request *rq, int where) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct cfq_rq *crq = RQ_DATA(rq); + + switch (where) { + case ELEVATOR_INSERT_BACK: + while (cfq_dispatch_requests(q, cfqd->cfq_quantum)) + ; + list_add_tail(&rq->queuelist, &q->queue_head); + break; + case ELEVATOR_INSERT_FRONT: + list_add(&rq->queuelist, &q->queue_head); + break; + case ELEVATOR_INSERT_SORT: + BUG_ON(!blk_fs_request(rq)); + cfq_enqueue(cfqd, crq); + break; + default: + printk("%s: bad insert point %d\n", __FUNCTION__,where); + return; + } + + if (rq_mergeable(rq)) { + cfq_add_crq_hash(cfqd, crq); + + if (!q->last_merge) + q->last_merge = rq; + } +} + +static int cfq_queue_empty(request_queue_t *q) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + + return list_empty(&q->queue_head) && list_empty(&cfqd->rr_list); +} + +static void cfq_completed_request(request_queue_t *q, struct request *rq) +{ + struct cfq_rq *crq = RQ_DATA(rq); + struct cfq_queue *cfqq; + + if (unlikely(!blk_fs_request(rq))) + return; + + cfqq = crq->cfq_queue; + + if (crq->in_flight) { + WARN_ON(!cfqq->in_flight); + cfqq->in_flight--; + } + + cfq_account_completion(cfqq, crq); +} + +static struct request * +cfq_former_request(request_queue_t *q, struct request *rq) +{ + struct cfq_rq *crq = RQ_DATA(rq); + struct rb_node *rbprev = rb_prev(&crq->rb_node); + + if (rbprev) + return rb_entry_crq(rbprev)->request; + + return NULL; +} + +static struct request * +cfq_latter_request(request_queue_t *q, struct request *rq) +{ + struct cfq_rq *crq = RQ_DATA(rq); + struct rb_node *rbnext = rb_next(&crq->rb_node); + + if (rbnext) + return rb_entry_crq(rbnext)->request; + + return NULL; +} + +static int cfq_may_queue(request_queue_t *q, int rw) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct cfq_queue *cfqq; + int ret = ELV_MQUEUE_MAY; + + if (current->flags & PF_MEMALLOC) + return ELV_MQUEUE_MAY; + + cfqq = cfq_find_cfq_hash(cfqd, cfq_hash_key(cfqd, current)); + if (cfqq) { + int limit = cfqd->max_queued; + + if (cfqq->allocated[rw] < cfqd->cfq_queued) + return ELV_MQUEUE_MUST; + + if (cfqd->busy_queues) + limit = q->nr_requests / cfqd->busy_queues; + + if (limit < cfqd->cfq_queued) + limit = cfqd->cfq_queued; + else if (limit > cfqd->max_queued) + limit = cfqd->max_queued; + + if (cfqq->allocated[rw] >= limit) { + if (limit > cfqq->alloc_limit[rw]) + cfqq->alloc_limit[rw] = limit; + + ret = ELV_MQUEUE_NO; + } + } + + return ret; +} + +static void cfq_check_waiters(request_queue_t *q, struct cfq_queue *cfqq) +{ + struct request_list *rl = &q->rq; + const int write = waitqueue_active(&rl->wait[WRITE]); + const int read = waitqueue_active(&rl->wait[READ]); + + if (read && cfqq->allocated[READ] < cfqq->alloc_limit[READ]) + wake_up(&rl->wait[READ]); + if (write && cfqq->allocated[WRITE] < cfqq->alloc_limit[WRITE]) + wake_up(&rl->wait[WRITE]); +} + +/* + * queue lock held here + */ +static void cfq_put_request(request_queue_t *q, struct request *rq) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct cfq_rq *crq = RQ_DATA(rq); + + if (crq) { + struct cfq_queue *cfqq = crq->cfq_queue; + + BUG_ON(q->last_merge == rq); + BUG_ON(!hlist_unhashed(&crq->hash)); + + if (crq->io_context) + put_io_context(crq->io_context->ioc); + + BUG_ON(!cfqq->allocated[crq->is_write]); + cfqq->allocated[crq->is_write]--; + + mempool_free(crq, cfqd->crq_pool); + rq->elevator_private = NULL; + + smp_mb(); + cfq_check_waiters(q, cfqq); + cfq_put_queue(cfqq); + } +} + +/* + * Allocate cfq data structures associated with this request. A queue and + */ +static int cfq_set_request(request_queue_t *q, struct request *rq, int gfp_mask) +{ + struct cfq_data *cfqd = q->elevator->elevator_data; + struct cfq_io_context *cic; + const int rw = rq_data_dir(rq); + struct cfq_queue *cfqq, *saved_cfqq; + struct cfq_rq *crq; + unsigned long flags; + + might_sleep_if(gfp_mask & __GFP_WAIT); + + spin_lock_irqsave(q->queue_lock, flags); + + cfqq = __cfq_get_queue(cfqd, cfq_hash_key(cfqd, current), gfp_mask); + if (!cfqq) + goto out_lock; + +repeat: + if (cfqq->allocated[rw] >= cfqd->max_queued) + goto out_lock; + + cfqq->allocated[rw]++; + spin_unlock_irqrestore(q->queue_lock, flags); + + /* + * if hashing type has changed, the cfq_queue might change here. + */ + saved_cfqq = cfqq; + cic = cfq_get_io_context(&cfqq, gfp_mask); + if (!cic) + goto err; + + /* + * repeat allocation checks on queue change + */ + if (unlikely(saved_cfqq != cfqq)) { + spin_lock_irqsave(q->queue_lock, flags); + saved_cfqq->allocated[rw]--; + goto repeat; + } + + crq = mempool_alloc(cfqd->crq_pool, gfp_mask); + if (crq) { + RB_CLEAR(&crq->rb_node); + crq->rb_key = 0; + crq->request = rq; + INIT_HLIST_NODE(&crq->hash); + crq->cfq_queue = cfqq; + crq->io_context = cic; + crq->service_start = crq->queue_start = 0; + crq->in_flight = crq->accounted = crq->is_sync = 0; + crq->is_write = rw; + rq->elevator_private = crq; + cfqq->alloc_limit[rw] = 0; + return 0; + } + + put_io_context(cic->ioc); +err: + spin_lock_irqsave(q->queue_lock, flags); + cfqq->allocated[rw]--; + cfq_put_queue(cfqq); +out_lock: + spin_unlock_irqrestore(q->queue_lock, flags); + return 1; +} + +static void cfq_put_cfqd(struct cfq_data *cfqd) +{ + request_queue_t *q = cfqd->queue; + + if (!atomic_dec_and_test(&cfqd->ref)) + return; + + blk_put_queue(q); + + mempool_destroy(cfqd->crq_pool); + kfree(cfqd->crq_hash); + kfree(cfqd->cfq_hash); + kfree(cfqd); +} + +static void cfq_exit_queue(elevator_t *e) +{ + cfq_put_cfqd(e->elevator_data); +} + +static int cfq_init_queue(request_queue_t *q, elevator_t *e) +{ + struct cfq_data *cfqd; + int i; + + cfqd = kmalloc(sizeof(*cfqd), GFP_KERNEL); + if (!cfqd) + return -ENOMEM; + + memset(cfqd, 0, sizeof(*cfqd)); + INIT_LIST_HEAD(&cfqd->rr_list); + INIT_LIST_HEAD(&cfqd->empty_list); + + cfqd->crq_hash = kmalloc(sizeof(struct hlist_head) * CFQ_MHASH_ENTRIES, GFP_KERNEL); + if (!cfqd->crq_hash) + goto out_crqhash; + + cfqd->cfq_hash = kmalloc(sizeof(struct hlist_head) * CFQ_QHASH_ENTRIES, GFP_KERNEL); + if (!cfqd->cfq_hash) + goto out_cfqhash; + + cfqd->crq_pool = mempool_create(BLKDEV_MIN_RQ, mempool_alloc_slab, mempool_free_slab, crq_pool); + if (!cfqd->crq_pool) + goto out_crqpool; + + for (i = 0; i < CFQ_MHASH_ENTRIES; i++) + INIT_HLIST_HEAD(&cfqd->crq_hash[i]); + for (i = 0; i < CFQ_QHASH_ENTRIES; i++) + INIT_HLIST_HEAD(&cfqd->cfq_hash[i]); + + e->elevator_data = cfqd; + + cfqd->queue = q; + atomic_inc(&q->refcnt); + + /* + * just set it to some high value, we want anyone to be able to queue + * some requests. fairness is handled differently + */ + q->nr_requests = 1024; + cfqd->max_queued = q->nr_requests / 16; + q->nr_batching = cfq_queued; + cfqd->key_type = CFQ_KEY_TGID; + cfqd->find_best_crq = 1; + atomic_set(&cfqd->ref, 1); + + cfqd->cfq_queued = cfq_queued; + cfqd->cfq_quantum = cfq_quantum; + cfqd->cfq_fifo_expire_r = cfq_fifo_expire_r; + cfqd->cfq_fifo_expire_w = cfq_fifo_expire_w; + cfqd->cfq_fifo_batch_expire = cfq_fifo_rate; + cfqd->cfq_back_max = cfq_back_max; + cfqd->cfq_back_penalty = cfq_back_penalty; + + return 0; +out_crqpool: + kfree(cfqd->cfq_hash); +out_cfqhash: + kfree(cfqd->crq_hash); +out_crqhash: + kfree(cfqd); + return -ENOMEM; +} + +static void cfq_slab_kill(void) +{ + if (crq_pool) + kmem_cache_destroy(crq_pool); + if (cfq_pool) + kmem_cache_destroy(cfq_pool); + if (cfq_ioc_pool) + kmem_cache_destroy(cfq_ioc_pool); +} + +static int __init cfq_slab_setup(void) +{ + crq_pool = kmem_cache_create("crq_pool", sizeof(struct cfq_rq), 0, 0, + NULL, NULL); + if (!crq_pool) + goto fail; + + cfq_pool = kmem_cache_create("cfq_pool", sizeof(struct cfq_queue), 0, 0, + NULL, NULL); + if (!cfq_pool) + goto fail; + + cfq_ioc_pool = kmem_cache_create("cfq_ioc_pool", + sizeof(struct cfq_io_context), 0, 0, NULL, NULL); + if (!cfq_ioc_pool) + goto fail; + + return 0; +fail: + cfq_slab_kill(); + return -ENOMEM; +} + + +/* + * sysfs parts below --> + */ +struct cfq_fs_entry { + struct attribute attr; + ssize_t (*show)(struct cfq_data *, char *); + ssize_t (*store)(struct cfq_data *, const char *, size_t); +}; + +static ssize_t +cfq_var_show(unsigned int var, char *page) +{ + return sprintf(page, "%d\n", var); +} + +static ssize_t +cfq_var_store(unsigned int *var, const char *page, size_t count) +{ + char *p = (char *) page; + + *var = simple_strtoul(p, &p, 10); + return count; +} + +static ssize_t +cfq_clear_elapsed(struct cfq_data *cfqd, const char *page, size_t count) +{ + max_elapsed_dispatch = max_elapsed_crq = 0; + return count; +} + +static ssize_t +cfq_set_key_type(struct cfq_data *cfqd, const char *page, size_t count) +{ + spin_lock_irq(cfqd->queue->queue_lock); + if (!strncmp(page, "pgid", 4)) + cfqd->key_type = CFQ_KEY_PGID; + else if (!strncmp(page, "tgid", 4)) + cfqd->key_type = CFQ_KEY_TGID; + else if (!strncmp(page, "uid", 3)) + cfqd->key_type = CFQ_KEY_UID; + else if (!strncmp(page, "gid", 3)) + cfqd->key_type = CFQ_KEY_GID; + spin_unlock_irq(cfqd->queue->queue_lock); + return count; +} + +static ssize_t +cfq_read_key_type(struct cfq_data *cfqd, char *page) +{ + ssize_t len = 0; + int i; + + for (i = CFQ_KEY_PGID; i < CFQ_KEY_LAST; i++) { + if (cfqd->key_type == i) + len += sprintf(page+len, "[%s] ", cfq_key_types[i]); + else + len += sprintf(page+len, "%s ", cfq_key_types[i]); + } + len += sprintf(page+len, "\n"); + return len; +} + +#define SHOW_FUNCTION(__FUNC, __VAR, __CONV) \ +static ssize_t __FUNC(struct cfq_data *cfqd, char *page) \ +{ \ + unsigned int __data = __VAR; \ + if (__CONV) \ + __data = jiffies_to_msecs(__data); \ + return cfq_var_show(__data, (page)); \ +} +SHOW_FUNCTION(cfq_quantum_show, cfqd->cfq_quantum, 0); +SHOW_FUNCTION(cfq_queued_show, cfqd->cfq_queued, 0); +SHOW_FUNCTION(cfq_fifo_expire_r_show, cfqd->cfq_fifo_expire_r, 1); +SHOW_FUNCTION(cfq_fifo_expire_w_show, cfqd->cfq_fifo_expire_w, 1); +SHOW_FUNCTION(cfq_fifo_batch_expire_show, cfqd->cfq_fifo_batch_expire, 1); +SHOW_FUNCTION(cfq_find_best_show, cfqd->find_best_crq, 0); +SHOW_FUNCTION(cfq_back_max_show, cfqd->cfq_back_max, 0); +SHOW_FUNCTION(cfq_back_penalty_show, cfqd->cfq_back_penalty, 0); +#undef SHOW_FUNCTION + +#define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX, __CONV) \ +static ssize_t __FUNC(struct cfq_data *cfqd, const char *page, size_t count) \ +{ \ + unsigned int __data; \ + int ret = cfq_var_store(&__data, (page), count); \ + if (__data < (MIN)) \ + __data = (MIN); \ + else if (__data > (MAX)) \ + __data = (MAX); \ + if (__CONV) \ + *(__PTR) = msecs_to_jiffies(__data); \ + else \ + *(__PTR) = __data; \ + return ret; \ +} +STORE_FUNCTION(cfq_quantum_store, &cfqd->cfq_quantum, 1, UINT_MAX, 0); +STORE_FUNCTION(cfq_queued_store, &cfqd->cfq_queued, 1, UINT_MAX, 0); +STORE_FUNCTION(cfq_fifo_expire_r_store, &cfqd->cfq_fifo_expire_r, 1, UINT_MAX, 1); +STORE_FUNCTION(cfq_fifo_expire_w_store, &cfqd->cfq_fifo_expire_w, 1, UINT_MAX, 1); +STORE_FUNCTION(cfq_fifo_batch_expire_store, &cfqd->cfq_fifo_batch_expire, 0, UINT_MAX, 1); +STORE_FUNCTION(cfq_find_best_store, &cfqd->find_best_crq, 0, 1, 0); +STORE_FUNCTION(cfq_back_max_store, &cfqd->cfq_back_max, 0, UINT_MAX, 0); +STORE_FUNCTION(cfq_back_penalty_store, &cfqd->cfq_back_penalty, 1, UINT_MAX, 0); +#undef STORE_FUNCTION + +static struct cfq_fs_entry cfq_quantum_entry = { + .attr = {.name = "quantum", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_quantum_show, + .store = cfq_quantum_store, +}; +static struct cfq_fs_entry cfq_queued_entry = { + .attr = {.name = "queued", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_queued_show, + .store = cfq_queued_store, +}; +static struct cfq_fs_entry cfq_fifo_expire_r_entry = { + .attr = {.name = "fifo_expire_sync", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_fifo_expire_r_show, + .store = cfq_fifo_expire_r_store, +}; +static struct cfq_fs_entry cfq_fifo_expire_w_entry = { + .attr = {.name = "fifo_expire_async", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_fifo_expire_w_show, + .store = cfq_fifo_expire_w_store, +}; +static struct cfq_fs_entry cfq_fifo_batch_expire_entry = { + .attr = {.name = "fifo_batch_expire", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_fifo_batch_expire_show, + .store = cfq_fifo_batch_expire_store, +}; +static struct cfq_fs_entry cfq_find_best_entry = { + .attr = {.name = "find_best_crq", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_find_best_show, + .store = cfq_find_best_store, +}; +static struct cfq_fs_entry cfq_back_max_entry = { + .attr = {.name = "back_seek_max", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_back_max_show, + .store = cfq_back_max_store, +}; +static struct cfq_fs_entry cfq_back_penalty_entry = { + .attr = {.name = "back_seek_penalty", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_back_penalty_show, + .store = cfq_back_penalty_store, +}; +static struct cfq_fs_entry cfq_clear_elapsed_entry = { + .attr = {.name = "clear_elapsed", .mode = S_IWUSR }, + .store = cfq_clear_elapsed, +}; +static struct cfq_fs_entry cfq_key_type_entry = { + .attr = {.name = "key_type", .mode = S_IRUGO | S_IWUSR }, + .show = cfq_read_key_type, + .store = cfq_set_key_type, +}; + +static struct attribute *default_attrs[] = { + &cfq_quantum_entry.attr, + &cfq_queued_entry.attr, + &cfq_fifo_expire_r_entry.attr, + &cfq_fifo_expire_w_entry.attr, + &cfq_fifo_batch_expire_entry.attr, + &cfq_key_type_entry.attr, + &cfq_find_best_entry.attr, + &cfq_back_max_entry.attr, + &cfq_back_penalty_entry.attr, + &cfq_clear_elapsed_entry.attr, + NULL, +}; + +#define to_cfq(atr) container_of((atr), struct cfq_fs_entry, attr) + +static ssize_t +cfq_attr_show(struct kobject *kobj, struct attribute *attr, char *page) +{ + elevator_t *e = container_of(kobj, elevator_t, kobj); + struct cfq_fs_entry *entry = to_cfq(attr); + + if (!entry->show) + return 0; + + return entry->show(e->elevator_data, page); +} + +static ssize_t +cfq_attr_store(struct kobject *kobj, struct attribute *attr, + const char *page, size_t length) +{ + elevator_t *e = container_of(kobj, elevator_t, kobj); + struct cfq_fs_entry *entry = to_cfq(attr); + + if (!entry->store) + return -EINVAL; + + return entry->store(e->elevator_data, page, length); +} + +static struct sysfs_ops cfq_sysfs_ops = { + .show = cfq_attr_show, + .store = cfq_attr_store, +}; + +static struct kobj_type cfq_ktype = { + .sysfs_ops = &cfq_sysfs_ops, + .default_attrs = default_attrs, +}; + +static struct elevator_type iosched_cfq = { + .ops = { + .elevator_merge_fn = cfq_merge, + .elevator_merged_fn = cfq_merged_request, + .elevator_merge_req_fn = cfq_merged_requests, + .elevator_next_req_fn = cfq_next_request, + .elevator_add_req_fn = cfq_insert_request, + .elevator_remove_req_fn = cfq_remove_request, + .elevator_requeue_req_fn = cfq_requeue_request, + .elevator_deactivate_req_fn = cfq_deactivate_request, + .elevator_queue_empty_fn = cfq_queue_empty, + .elevator_completed_req_fn = cfq_completed_request, + .elevator_former_req_fn = cfq_former_request, + .elevator_latter_req_fn = cfq_latter_request, + .elevator_set_req_fn = cfq_set_request, + .elevator_put_req_fn = cfq_put_request, + .elevator_may_queue_fn = cfq_may_queue, + .elevator_init_fn = cfq_init_queue, + .elevator_exit_fn = cfq_exit_queue, + }, + .elevator_ktype = &cfq_ktype, + .elevator_name = "cfq", + .elevator_owner = THIS_MODULE, +}; + +static int __init cfq_init(void) +{ + int ret; + + if (cfq_slab_setup()) + return -ENOMEM; + + ret = elv_register(&iosched_cfq); + if (!ret) { + __module_get(THIS_MODULE); + return 0; + } + + cfq_slab_kill(); + return ret; +} + +static void __exit cfq_exit(void) +{ + cfq_slab_kill(); + elv_unregister(&iosched_cfq); +} + +module_init(cfq_init); +module_exit(cfq_exit); + +MODULE_AUTHOR("Jens Axboe"); +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("Completely Fair Queueing IO scheduler"); diff --git a/drivers/block/cpqarray.c b/drivers/block/cpqarray.c new file mode 100644 index 00000000000..cf1822a6361 --- /dev/null +++ b/drivers/block/cpqarray.c @@ -0,0 +1,1850 @@ +/* + * Disk Array driver for Compaq SMART2 Controllers + * Copyright 1998 Compaq Computer Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + */ +#include <linux/config.h> /* CONFIG_PROC_FS */ +#include <linux/module.h> +#include <linux/types.h> +#include <linux/pci.h> +#include <linux/bio.h> +#include <linux/interrupt.h> +#include <linux/kernel.h> +#include <linux/slab.h> +#include <linux/delay.h> +#include <linux/major.h> +#include <linux/fs.h> +#include <linux/blkpg.h> +#include <linux/timer.h> +#include <linux/proc_fs.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/init.h> +#include <linux/hdreg.h> +#include <linux/spinlock.h> +#include <linux/blkdev.h> +#include <linux/genhd.h> +#include <asm/uaccess.h> +#include <asm/io.h> + + +#define SMART2_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin)) + +#define DRIVER_NAME "Compaq SMART2 Driver (v 2.6.0)" +#define DRIVER_VERSION SMART2_DRIVER_VERSION(2,6,0) + +/* Embedded module documentation macros - see modules.h */ +/* Original author Chris Frantz - Compaq Computer Corporation */ +MODULE_AUTHOR("Compaq Computer Corporation"); +MODULE_DESCRIPTION("Driver for Compaq Smart2 Array Controllers version 2.6.0"); +MODULE_LICENSE("GPL"); + +#include "cpqarray.h" +#include "ida_cmd.h" +#include "smart1,2.h" +#include "ida_ioctl.h" + +#define READ_AHEAD 128 +#define NR_CMDS 128 /* This could probably go as high as ~400 */ + +#define MAX_CTLR 8 +#define CTLR_SHIFT 8 + +#define CPQARRAY_DMA_MASK 0xFFFFFFFF /* 32 bit DMA */ + +static int nr_ctlr; +static ctlr_info_t *hba[MAX_CTLR]; + +static int eisa[8]; + +#define NR_PRODUCTS (sizeof(products)/sizeof(struct board_type)) + +/* board_id = Subsystem Device ID & Vendor ID + * product = Marketing Name for the board + * access = Address of the struct of function pointers + */ +static struct board_type products[] = { + { 0x0040110E, "IDA", &smart1_access }, + { 0x0140110E, "IDA-2", &smart1_access }, + { 0x1040110E, "IAES", &smart1_access }, + { 0x2040110E, "SMART", &smart1_access }, + { 0x3040110E, "SMART-2/E", &smart2e_access }, + { 0x40300E11, "SMART-2/P", &smart2_access }, + { 0x40310E11, "SMART-2SL", &smart2_access }, + { 0x40320E11, "Smart Array 3200", &smart2_access }, + { 0x40330E11, "Smart Array 3100ES", &smart2_access }, + { 0x40340E11, "Smart Array 221", &smart2_access }, + { 0x40400E11, "Integrated Array", &smart4_access }, + { 0x40480E11, "Compaq Raid LC2", &smart4_access }, + { 0x40500E11, "Smart Array 4200", &smart4_access }, + { 0x40510E11, "Smart Array 4250ES", &smart4_access }, + { 0x40580E11, "Smart Array 431", &smart4_access }, +}; + +/* define the PCI info for the PCI cards this driver can control */ +static const struct pci_device_id cpqarray_pci_device_id[] = +{ + { PCI_VENDOR_ID_DEC, PCI_DEVICE_ID_COMPAQ_42XX, + 0x0E11, 0x4058, 0, 0, 0}, /* SA431 */ + { PCI_VENDOR_ID_DEC, PCI_DEVICE_ID_COMPAQ_42XX, + 0x0E11, 0x4051, 0, 0, 0}, /* SA4250ES */ + { PCI_VENDOR_ID_DEC, PCI_DEVICE_ID_COMPAQ_42XX, + 0x0E11, 0x4050, 0, 0, 0}, /* SA4200 */ + { PCI_VENDOR_ID_NCR, PCI_DEVICE_ID_NCR_53C1510, + 0x0E11, 0x4048, 0, 0, 0}, /* LC2 */ + { PCI_VENDOR_ID_NCR, PCI_DEVICE_ID_NCR_53C1510, + 0x0E11, 0x4040, 0, 0, 0}, /* Integrated Array */ + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_SMART2P, + 0x0E11, 0x4034, 0, 0, 0}, /* SA 221 */ + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_SMART2P, + 0x0E11, 0x4033, 0, 0, 0}, /* SA 3100ES*/ + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_SMART2P, + 0x0E11, 0x4032, 0, 0, 0}, /* SA 3200*/ + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_SMART2P, + 0x0E11, 0x4031, 0, 0, 0}, /* SA 2SL*/ + { PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_SMART2P, + 0x0E11, 0x4030, 0, 0, 0}, /* SA 2P */ + { 0 } +}; + +MODULE_DEVICE_TABLE(pci, cpqarray_pci_device_id); + +static struct gendisk *ida_gendisk[MAX_CTLR][NWD]; + +/* Debug... */ +#define DBG(s) do { s } while(0) +/* Debug (general info)... */ +#define DBGINFO(s) do { } while(0) +/* Debug Paranoid... */ +#define DBGP(s) do { } while(0) +/* Debug Extra Paranoid... */ +#define DBGPX(s) do { } while(0) + +static int cpqarray_pci_init(ctlr_info_t *c, struct pci_dev *pdev); +static void __iomem *remap_pci_mem(ulong base, ulong size); +static int cpqarray_eisa_detect(void); +static int pollcomplete(int ctlr); +static void getgeometry(int ctlr); +static void start_fwbk(int ctlr); + +static cmdlist_t * cmd_alloc(ctlr_info_t *h, int get_from_pool); +static void cmd_free(ctlr_info_t *h, cmdlist_t *c, int got_from_pool); + +static void free_hba(int i); +static int alloc_cpqarray_hba(void); + +static int sendcmd( + __u8 cmd, + int ctlr, + void *buff, + size_t size, + unsigned int blk, + unsigned int blkcnt, + unsigned int log_unit ); + +static int ida_open(struct inode *inode, struct file *filep); +static int ida_release(struct inode *inode, struct file *filep); +static int ida_ioctl(struct inode *inode, struct file *filep, unsigned int cmd, unsigned long arg); +static int ida_ctlr_ioctl(ctlr_info_t *h, int dsk, ida_ioctl_t *io); + +static void do_ida_request(request_queue_t *q); +static void start_io(ctlr_info_t *h); + +static inline void addQ(cmdlist_t **Qptr, cmdlist_t *c); +static inline cmdlist_t *removeQ(cmdlist_t **Qptr, cmdlist_t *c); +static inline void complete_buffers(struct bio *bio, int ok); +static inline void complete_command(cmdlist_t *cmd, int timeout); + +static irqreturn_t do_ida_intr(int irq, void *dev_id, struct pt_regs * regs); +static void ida_timer(unsigned long tdata); +static int ida_revalidate(struct gendisk *disk); +static int revalidate_allvol(ctlr_info_t *host); +static int cpqarray_register_ctlr(int ctlr, struct pci_dev *pdev); + +#ifdef CONFIG_PROC_FS +static void ida_procinit(int i); +static int ida_proc_get_info(char *buffer, char **start, off_t offset, int length, int *eof, void *data); +#else +static void ida_procinit(int i) {} +#endif + +static inline drv_info_t *get_drv(struct gendisk *disk) +{ + return disk->private_data; +} + +static inline ctlr_info_t *get_host(struct gendisk *disk) +{ + return disk->queue->queuedata; +} + + +static struct block_device_operations ida_fops = { + .owner = THIS_MODULE, + .open = ida_open, + .release = ida_release, + .ioctl = ida_ioctl, + .revalidate_disk= ida_revalidate, +}; + + +#ifdef CONFIG_PROC_FS + +static struct proc_dir_entry *proc_array; + +/* + * Get us a file in /proc/array that says something about each controller. + * Create /proc/array if it doesn't exist yet. + */ +static void __init ida_procinit(int i) +{ + if (proc_array == NULL) { + proc_array = proc_mkdir("cpqarray", proc_root_driver); + if (!proc_array) return; + } + + create_proc_read_entry(hba[i]->devname, 0, proc_array, + ida_proc_get_info, hba[i]); +} + +/* + * Report information about this controller. + */ +static int ida_proc_get_info(char *buffer, char **start, off_t offset, int length, int *eof, void *data) +{ + off_t pos = 0; + off_t len = 0; + int size, i, ctlr; + ctlr_info_t *h = (ctlr_info_t*)data; + drv_info_t *drv; +#ifdef CPQ_PROC_PRINT_QUEUES + cmdlist_t *c; + unsigned long flags; +#endif + + ctlr = h->ctlr; + size = sprintf(buffer, "%s: Compaq %s Controller\n" + " Board ID: 0x%08lx\n" + " Firmware Revision: %c%c%c%c\n" + " Controller Sig: 0x%08lx\n" + " Memory Address: 0x%08lx\n" + " I/O Port: 0x%04x\n" + " IRQ: %d\n" + " Logical drives: %d\n" + " Physical drives: %d\n\n" + " Current Q depth: %d\n" + " Max Q depth since init: %d\n\n", + h->devname, + h->product_name, + (unsigned long)h->board_id, + h->firm_rev[0], h->firm_rev[1], h->firm_rev[2], h->firm_rev[3], + (unsigned long)h->ctlr_sig, (unsigned long)h->vaddr, + (unsigned int) h->io_mem_addr, (unsigned int)h->intr, + h->log_drives, h->phys_drives, + h->Qdepth, h->maxQsinceinit); + + pos += size; len += size; + + size = sprintf(buffer+len, "Logical Drive Info:\n"); + pos += size; len += size; + + for(i=0; i<h->log_drives; i++) { + drv = &h->drv[i]; + size = sprintf(buffer+len, "ida/c%dd%d: blksz=%d nr_blks=%d\n", + ctlr, i, drv->blk_size, drv->nr_blks); + pos += size; len += size; + } + +#ifdef CPQ_PROC_PRINT_QUEUES + spin_lock_irqsave(IDA_LOCK(h->ctlr), flags); + size = sprintf(buffer+len, "\nCurrent Queues:\n"); + pos += size; len += size; + + c = h->reqQ; + size = sprintf(buffer+len, "reqQ = %p", c); pos += size; len += size; + if (c) c=c->next; + while(c && c != h->reqQ) { + size = sprintf(buffer+len, "->%p", c); + pos += size; len += size; + c=c->next; + } + + c = h->cmpQ; + size = sprintf(buffer+len, "\ncmpQ = %p", c); pos += size; len += size; + if (c) c=c->next; + while(c && c != h->cmpQ) { + size = sprintf(buffer+len, "->%p", c); + pos += size; len += size; + c=c->next; + } + + size = sprintf(buffer+len, "\n"); pos += size; len += size; + spin_unlock_irqrestore(IDA_LOCK(h->ctlr), flags); +#endif + size = sprintf(buffer+len, "nr_allocs = %d\nnr_frees = %d\n", + h->nr_allocs, h->nr_frees); + pos += size; len += size; + + *eof = 1; + *start = buffer+offset; + len -= offset; + if (len>length) + len = length; + return len; +} +#endif /* CONFIG_PROC_FS */ + +module_param_array(eisa, int, NULL, 0); + +static void release_io_mem(ctlr_info_t *c) +{ + /* if IO mem was not protected do nothing */ + if( c->io_mem_addr == 0) + return; + release_region(c->io_mem_addr, c->io_mem_length); + c->io_mem_addr = 0; + c->io_mem_length = 0; +} + +static void __devexit cpqarray_remove_one(int i) +{ + int j; + char buff[4]; + + /* sendcmd will turn off interrupt, and send the flush... + * To write all data in the battery backed cache to disks + * no data returned, but don't want to send NULL to sendcmd */ + if( sendcmd(FLUSH_CACHE, i, buff, 4, 0, 0, 0)) + { + printk(KERN_WARNING "Unable to flush cache on controller %d\n", + i); + } + free_irq(hba[i]->intr, hba[i]); + iounmap(hba[i]->vaddr); + unregister_blkdev(COMPAQ_SMART2_MAJOR+i, hba[i]->devname); + del_timer(&hba[i]->timer); + remove_proc_entry(hba[i]->devname, proc_array); + pci_free_consistent(hba[i]->pci_dev, + NR_CMDS * sizeof(cmdlist_t), (hba[i]->cmd_pool), + hba[i]->cmd_pool_dhandle); + kfree(hba[i]->cmd_pool_bits); + for(j = 0; j < NWD; j++) { + if (ida_gendisk[i][j]->flags & GENHD_FL_UP) + del_gendisk(ida_gendisk[i][j]); + devfs_remove("ida/c%dd%d",i,j); + put_disk(ida_gendisk[i][j]); + } + blk_cleanup_queue(hba[i]->queue); + release_io_mem(hba[i]); + free_hba(i); +} + +static void __devexit cpqarray_remove_one_pci (struct pci_dev *pdev) +{ + int i; + ctlr_info_t *tmp_ptr; + + if (pci_get_drvdata(pdev) == NULL) { + printk( KERN_ERR "cpqarray: Unable to remove device \n"); + return; + } + + tmp_ptr = pci_get_drvdata(pdev); + i = tmp_ptr->ctlr; + if (hba[i] == NULL) { + printk(KERN_ERR "cpqarray: controller %d appears to have" + "already been removed \n", i); + return; + } + pci_set_drvdata(pdev, NULL); + + cpqarray_remove_one(i); +} + +/* removing an instance that was not removed automatically.. + * must be an eisa card. + */ +static void __devexit cpqarray_remove_one_eisa (int i) +{ + if (hba[i] == NULL) { + printk(KERN_ERR "cpqarray: controller %d appears to have" + "already been removed \n", i); + return; + } + cpqarray_remove_one(i); +} + +/* pdev is NULL for eisa */ +static int cpqarray_register_ctlr( int i, struct pci_dev *pdev) +{ + request_queue_t *q; + int j; + + /* + * register block devices + * Find disks and fill in structs + * Get an interrupt, set the Q depth and get into /proc + */ + + /* If this successful it should insure that we are the only */ + /* instance of the driver */ + if (register_blkdev(COMPAQ_SMART2_MAJOR+i, hba[i]->devname)) { + goto Enomem4; + } + hba[i]->access.set_intr_mask(hba[i], 0); + if (request_irq(hba[i]->intr, do_ida_intr, + SA_INTERRUPT|SA_SHIRQ|SA_SAMPLE_RANDOM, + hba[i]->devname, hba[i])) + { + printk(KERN_ERR "cpqarray: Unable to get irq %d for %s\n", + hba[i]->intr, hba[i]->devname); + goto Enomem3; + } + + for (j=0; j<NWD; j++) { + ida_gendisk[i][j] = alloc_disk(1 << NWD_SHIFT); + if (!ida_gendisk[i][j]) + goto Enomem2; + } + + hba[i]->cmd_pool = (cmdlist_t *)pci_alloc_consistent( + hba[i]->pci_dev, NR_CMDS * sizeof(cmdlist_t), + &(hba[i]->cmd_pool_dhandle)); + hba[i]->cmd_pool_bits = kmalloc( + ((NR_CMDS+BITS_PER_LONG-1)/BITS_PER_LONG)*sizeof(unsigned long), + GFP_KERNEL); + + if (!hba[i]->cmd_pool_bits || !hba[i]->cmd_pool) + goto Enomem1; + + memset(hba[i]->cmd_pool, 0, NR_CMDS * sizeof(cmdlist_t)); + memset(hba[i]->cmd_pool_bits, 0, ((NR_CMDS+BITS_PER_LONG-1)/BITS_PER_LONG)*sizeof(unsigned long)); + printk(KERN_INFO "cpqarray: Finding drives on %s", + hba[i]->devname); + + spin_lock_init(&hba[i]->lock); + q = blk_init_queue(do_ida_request, &hba[i]->lock); + if (!q) + goto Enomem1; + + hba[i]->queue = q; + q->queuedata = hba[i]; + + getgeometry(i); + start_fwbk(i); + + ida_procinit(i); + + if (pdev) + blk_queue_bounce_limit(q, hba[i]->pci_dev->dma_mask); + + /* This is a hardware imposed limit. */ + blk_queue_max_hw_segments(q, SG_MAX); + + /* This is a driver limit and could be eliminated. */ + blk_queue_max_phys_segments(q, SG_MAX); + + init_timer(&hba[i]->timer); + hba[i]->timer.expires = jiffies + IDA_TIMER; + hba[i]->timer.data = (unsigned long)hba[i]; + hba[i]->timer.function = ida_timer; + add_timer(&hba[i]->timer); + + /* Enable IRQ now that spinlock and rate limit timer are set up */ + hba[i]->access.set_intr_mask(hba[i], FIFO_NOT_EMPTY); + + for(j=0; j<NWD; j++) { + struct gendisk *disk = ida_gendisk[i][j]; + drv_info_t *drv = &hba[i]->drv[j]; + sprintf(disk->disk_name, "ida/c%dd%d", i, j); + disk->major = COMPAQ_SMART2_MAJOR + i; + disk->first_minor = j<<NWD_SHIFT; + disk->fops = &ida_fops; + if (j && !drv->nr_blks) + continue; + blk_queue_hardsect_size(hba[i]->queue, drv->blk_size); + set_capacity(disk, drv->nr_blks); + disk->queue = hba[i]->queue; + disk->private_data = drv; + add_disk(disk); + } + + /* done ! */ + return(i); + +Enomem1: + nr_ctlr = i; + kfree(hba[i]->cmd_pool_bits); + if (hba[i]->cmd_pool) + pci_free_consistent(hba[i]->pci_dev, NR_CMDS*sizeof(cmdlist_t), + hba[i]->cmd_pool, hba[i]->cmd_pool_dhandle); +Enomem2: + while (j--) { + put_disk(ida_gendisk[i][j]); + ida_gendisk[i][j] = NULL; + } + free_irq(hba[i]->intr, hba[i]); +Enomem3: + unregister_blkdev(COMPAQ_SMART2_MAJOR+i, hba[i]->devname); +Enomem4: + if (pdev) + pci_set_drvdata(pdev, NULL); + release_io_mem(hba[i]); + free_hba(i); + + printk( KERN_ERR "cpqarray: out of memory"); + + return -1; +} + +static int __init cpqarray_init_one( struct pci_dev *pdev, + const struct pci_device_id *ent) +{ + int i; + + printk(KERN_DEBUG "cpqarray: Device 0x%x has been found at" + " bus %d dev %d func %d\n", + pdev->device, pdev->bus->number, PCI_SLOT(pdev->devfn), + PCI_FUNC(pdev->devfn)); + i = alloc_cpqarray_hba(); + if( i < 0 ) + return (-1); + memset(hba[i], 0, sizeof(ctlr_info_t)); + sprintf(hba[i]->devname, "ida%d", i); + hba[i]->ctlr = i; + /* Initialize the pdev driver private data */ + pci_set_drvdata(pdev, hba[i]); + + if (cpqarray_pci_init(hba[i], pdev) != 0) { + pci_set_drvdata(pdev, NULL); + release_io_mem(hba[i]); + free_hba(i); + return -1; + } + + return (cpqarray_register_ctlr(i, pdev)); +} + +static struct pci_driver cpqarray_pci_driver = { + .name = "cpqarray", + .probe = cpqarray_init_one, + .remove = __devexit_p(cpqarray_remove_one_pci), + .id_table = cpqarray_pci_device_id, +}; + +/* + * This is it. Find all the controllers and register them. + * returns the number of block devices registered. + */ +static int __init cpqarray_init(void) +{ + int num_cntlrs_reg = 0; + int i; + int rc = 0; + + /* detect controllers */ + printk(DRIVER_NAME "\n"); + + rc = pci_register_driver(&cpqarray_pci_driver); + if (rc) + return rc; + cpqarray_eisa_detect(); + + for (i=0; i < MAX_CTLR; i++) { + if (hba[i] != NULL) + num_cntlrs_reg++; + } + + return(num_cntlrs_reg); +} + +/* Function to find the first free pointer into our hba[] array */ +/* Returns -1 if no free entries are left. */ +static int alloc_cpqarray_hba(void) +{ + int i; + + for(i=0; i< MAX_CTLR; i++) { + if (hba[i] == NULL) { + hba[i] = kmalloc(sizeof(ctlr_info_t), GFP_KERNEL); + if(hba[i]==NULL) { + printk(KERN_ERR "cpqarray: out of memory.\n"); + return (-1); + } + return (i); + } + } + printk(KERN_WARNING "cpqarray: This driver supports a maximum" + " of 8 controllers.\n"); + return(-1); +} + +static void free_hba(int i) +{ + kfree(hba[i]); + hba[i]=NULL; +} + +/* + * Find the IO address of the controller, its IRQ and so forth. Fill + * in some basic stuff into the ctlr_info_t structure. + */ +static int cpqarray_pci_init(ctlr_info_t *c, struct pci_dev *pdev) +{ + ushort vendor_id, device_id, command; + unchar cache_line_size, latency_timer; + unchar irq, revision; + unsigned long addr[6]; + __u32 board_id; + + int i; + + c->pci_dev = pdev; + if (pci_enable_device(pdev)) { + printk(KERN_ERR "cpqarray: Unable to Enable PCI device\n"); + return -1; + } + vendor_id = pdev->vendor; + device_id = pdev->device; + irq = pdev->irq; + + for(i=0; i<6; i++) + addr[i] = pci_resource_start(pdev, i); + + if (pci_set_dma_mask(pdev, CPQARRAY_DMA_MASK) != 0) + { + printk(KERN_ERR "cpqarray: Unable to set DMA mask\n"); + return -1; + } + + pci_read_config_word(pdev, PCI_COMMAND, &command); + pci_read_config_byte(pdev, PCI_CLASS_REVISION, &revision); + pci_read_config_byte(pdev, PCI_CACHE_LINE_SIZE, &cache_line_size); + pci_read_config_byte(pdev, PCI_LATENCY_TIMER, &latency_timer); + + pci_read_config_dword(pdev, 0x2c, &board_id); + + /* check to see if controller has been disabled */ + if(!(command & 0x02)) { + printk(KERN_WARNING + "cpqarray: controller appears to be disabled\n"); + return(-1); + } + +DBGINFO( + printk("vendor_id = %x\n", vendor_id); + printk("device_id = %x\n", device_id); + printk("command = %x\n", command); + for(i=0; i<6; i++) + printk("addr[%d] = %lx\n", i, addr[i]); + printk("revision = %x\n", revision); + printk("irq = %x\n", irq); + printk("cache_line_size = %x\n", cache_line_size); + printk("latency_timer = %x\n", latency_timer); + printk("board_id = %x\n", board_id); +); + + c->intr = irq; + + for(i=0; i<6; i++) { + if (pci_resource_flags(pdev, i) & PCI_BASE_ADDRESS_SPACE_IO) + { /* IO space */ + c->io_mem_addr = addr[i]; + c->io_mem_length = pci_resource_end(pdev, i) + - pci_resource_start(pdev, i) + 1; + if(!request_region( c->io_mem_addr, c->io_mem_length, + "cpqarray")) + { + printk( KERN_WARNING "cpqarray I/O memory range already in use addr %lx length = %ld\n", c->io_mem_addr, c->io_mem_length); + c->io_mem_addr = 0; + c->io_mem_length = 0; + } + break; + } + } + + c->paddr = 0; + for(i=0; i<6; i++) + if (!(pci_resource_flags(pdev, i) & + PCI_BASE_ADDRESS_SPACE_IO)) { + c->paddr = pci_resource_start (pdev, i); + break; + } + if (!c->paddr) + return -1; + c->vaddr = remap_pci_mem(c->paddr, 128); + if (!c->vaddr) + return -1; + c->board_id = board_id; + + for(i=0; i<NR_PRODUCTS; i++) { + if (board_id == products[i].board_id) { + c->product_name = products[i].product_name; + c->access = *(products[i].access); + break; + } + } + if (i == NR_PRODUCTS) { + printk(KERN_WARNING "cpqarray: Sorry, I don't know how" + " to access the SMART Array controller %08lx\n", + (unsigned long)board_id); + return -1; + } + + return 0; +} + +/* + * Map (physical) PCI mem into (virtual) kernel space + */ +static void __iomem *remap_pci_mem(ulong base, ulong size) +{ + ulong page_base = ((ulong) base) & PAGE_MASK; + ulong page_offs = ((ulong) base) - page_base; + void __iomem *page_remapped = ioremap(page_base, page_offs+size); + + return (page_remapped ? (page_remapped + page_offs) : NULL); +} + +#ifndef MODULE +/* + * Config string is a comma separated set of i/o addresses of EISA cards. + */ +static int cpqarray_setup(char *str) +{ + int i, ints[9]; + + (void)get_options(str, ARRAY_SIZE(ints), ints); + + for(i=0; i<ints[0] && i<8; i++) + eisa[i] = ints[i+1]; + return 1; +} + +__setup("smart2=", cpqarray_setup); + +#endif + +/* + * Find an EISA controller's signature. Set up an hba if we find it. + */ +static int cpqarray_eisa_detect(void) +{ + int i=0, j; + __u32 board_id; + int intr; + int ctlr; + int num_ctlr = 0; + + while(i<8 && eisa[i]) { + ctlr = alloc_cpqarray_hba(); + if(ctlr == -1) + break; + board_id = inl(eisa[i]+0xC80); + for(j=0; j < NR_PRODUCTS; j++) + if (board_id == products[j].board_id) + break; + + if (j == NR_PRODUCTS) { + printk(KERN_WARNING "cpqarray: Sorry, I don't know how" + " to access the SMART Array controller %08lx\n", (unsigned long)board_id); + continue; + } + + memset(hba[ctlr], 0, sizeof(ctlr_info_t)); + hba[ctlr]->io_mem_addr = eisa[i]; + hba[ctlr]->io_mem_length = 0x7FF; + if(!request_region(hba[ctlr]->io_mem_addr, + hba[ctlr]->io_mem_length, + "cpqarray")) + { + printk(KERN_WARNING "cpqarray: I/O range already in " + "use addr = %lx length = %ld\n", + hba[ctlr]->io_mem_addr, + hba[ctlr]->io_mem_length); + free_hba(ctlr); + continue; + } + + /* + * Read the config register to find our interrupt + */ + intr = inb(eisa[i]+0xCC0) >> 4; + if (intr & 1) intr = 11; + else if (intr & 2) intr = 10; + else if (intr & 4) intr = 14; + else if (intr & 8) intr = 15; + + hba[ctlr]->intr = intr; + sprintf(hba[ctlr]->devname, "ida%d", nr_ctlr); + hba[ctlr]->product_name = products[j].product_name; + hba[ctlr]->access = *(products[j].access); + hba[ctlr]->ctlr = ctlr; + hba[ctlr]->board_id = board_id; + hba[ctlr]->pci_dev = NULL; /* not PCI */ + +DBGINFO( + printk("i = %d, j = %d\n", i, j); + printk("irq = %x\n", intr); + printk("product name = %s\n", products[j].product_name); + printk("board_id = %x\n", board_id); +); + + num_ctlr++; + i++; + + if (cpqarray_register_ctlr(ctlr, NULL) == -1) + printk(KERN_WARNING + "cpqarray: Can't register EISA controller %d\n", + ctlr); + + } + + return num_ctlr; +} + +/* + * Open. Make sure the device is really there. + */ +static int ida_open(struct inode *inode, struct file *filep) +{ + drv_info_t *drv = get_drv(inode->i_bdev->bd_disk); + ctlr_info_t *host = get_host(inode->i_bdev->bd_disk); + + DBGINFO(printk("ida_open %s\n", inode->i_bdev->bd_disk->disk_name)); + /* + * Root is allowed to open raw volume zero even if it's not configured + * so array config can still work. I don't think I really like this, + * but I'm already using way to many device nodes to claim another one + * for "raw controller". + */ + if (!drv->nr_blks) { + if (!capable(CAP_SYS_RAWIO)) + return -ENXIO; + if (!capable(CAP_SYS_ADMIN) && drv != host->drv) + return -ENXIO; + } + host->usage_count++; + return 0; +} + +/* + * Close. Sync first. + */ +static int ida_release(struct inode *inode, struct file *filep) +{ + ctlr_info_t *host = get_host(inode->i_bdev->bd_disk); + host->usage_count--; + return 0; +} + +/* + * Enqueuing and dequeuing functions for cmdlists. + */ +static inline void addQ(cmdlist_t **Qptr, cmdlist_t *c) +{ + if (*Qptr == NULL) { + *Qptr = c; + c->next = c->prev = c; + } else { + c->prev = (*Qptr)->prev; + c->next = (*Qptr); + (*Qptr)->prev->next = c; + (*Qptr)->prev = c; + } +} + +static inline cmdlist_t *removeQ(cmdlist_t **Qptr, cmdlist_t *c) +{ + if (c && c->next != c) { + if (*Qptr == c) *Qptr = c->next; + c->prev->next = c->next; + c->next->prev = c->prev; + } else { + *Qptr = NULL; + } + return c; +} + +/* + * Get a request and submit it to the controller. + * This routine needs to grab all the requests it possibly can from the + * req Q and submit them. Interrupts are off (and need to be off) when you + * are in here (either via the dummy do_ida_request functions or by being + * called from the interrupt handler + */ +static void do_ida_request(request_queue_t *q) +{ + ctlr_info_t *h = q->queuedata; + cmdlist_t *c; + struct request *creq; + struct scatterlist tmp_sg[SG_MAX]; + int i, dir, seg; + + if (blk_queue_plugged(q)) + goto startio; + +queue_next: + creq = elv_next_request(q); + if (!creq) + goto startio; + + if (creq->nr_phys_segments > SG_MAX) + BUG(); + + if ((c = cmd_alloc(h,1)) == NULL) + goto startio; + + blkdev_dequeue_request(creq); + + c->ctlr = h->ctlr; + c->hdr.unit = (drv_info_t *)(creq->rq_disk->private_data) - h->drv; + c->hdr.size = sizeof(rblk_t) >> 2; + c->size += sizeof(rblk_t); + + c->req.hdr.blk = creq->sector; + c->rq = creq; +DBGPX( + printk("sector=%d, nr_sectors=%d\n", creq->sector, creq->nr_sectors); +); + seg = blk_rq_map_sg(q, creq, tmp_sg); + + /* Now do all the DMA Mappings */ + if (rq_data_dir(creq) == READ) + dir = PCI_DMA_FROMDEVICE; + else + dir = PCI_DMA_TODEVICE; + for( i=0; i < seg; i++) + { + c->req.sg[i].size = tmp_sg[i].length; + c->req.sg[i].addr = (__u32) pci_map_page(h->pci_dev, + tmp_sg[i].page, + tmp_sg[i].offset, + tmp_sg[i].length, dir); + } +DBGPX( printk("Submitting %d sectors in %d segments\n", creq->nr_sectors, seg); ); + c->req.hdr.sg_cnt = seg; + c->req.hdr.blk_cnt = creq->nr_sectors; + c->req.hdr.cmd = (rq_data_dir(creq) == READ) ? IDA_READ : IDA_WRITE; + c->type = CMD_RWREQ; + + /* Put the request on the tail of the request queue */ + addQ(&h->reqQ, c); + h->Qdepth++; + if (h->Qdepth > h->maxQsinceinit) + h->maxQsinceinit = h->Qdepth; + + goto queue_next; + +startio: + start_io(h); +} + +/* + * start_io submits everything on a controller's request queue + * and moves it to the completion queue. + * + * Interrupts had better be off if you're in here + */ +static void start_io(ctlr_info_t *h) +{ + cmdlist_t *c; + + while((c = h->reqQ) != NULL) { + /* Can't do anything if we're busy */ + if (h->access.fifo_full(h) == 0) + return; + + /* Get the first entry from the request Q */ + removeQ(&h->reqQ, c); + h->Qdepth--; + + /* Tell the controller to do our bidding */ + h->access.submit_command(h, c); + + /* Get onto the completion Q */ + addQ(&h->cmpQ, c); + } +} + +static inline void complete_buffers(struct bio *bio, int ok) +{ + struct bio *xbh; + while(bio) { + int nr_sectors = bio_sectors(bio); + + xbh = bio->bi_next; + bio->bi_next = NULL; + + blk_finished_io(nr_sectors); + bio_endio(bio, nr_sectors << 9, ok ? 0 : -EIO); + + bio = xbh; + } +} +/* + * Mark all buffers that cmd was responsible for + */ +static inline void complete_command(cmdlist_t *cmd, int timeout) +{ + int ok=1; + int i, ddir; + + if (cmd->req.hdr.rcode & RCODE_NONFATAL && + (hba[cmd->ctlr]->misc_tflags & MISC_NONFATAL_WARN) == 0) { + printk(KERN_NOTICE "Non Fatal error on ida/c%dd%d\n", + cmd->ctlr, cmd->hdr.unit); + hba[cmd->ctlr]->misc_tflags |= MISC_NONFATAL_WARN; + } + if (cmd->req.hdr.rcode & RCODE_FATAL) { + printk(KERN_WARNING "Fatal error on ida/c%dd%d\n", + cmd->ctlr, cmd->hdr.unit); + ok = 0; + } + if (cmd->req.hdr.rcode & RCODE_INVREQ) { + printk(KERN_WARNING "Invalid request on ida/c%dd%d = (cmd=%x sect=%d cnt=%d sg=%d ret=%x)\n", + cmd->ctlr, cmd->hdr.unit, cmd->req.hdr.cmd, + cmd->req.hdr.blk, cmd->req.hdr.blk_cnt, + cmd->req.hdr.sg_cnt, cmd->req.hdr.rcode); + ok = 0; + } + if (timeout) ok = 0; + /* unmap the DMA mapping for all the scatter gather elements */ + if (cmd->req.hdr.cmd == IDA_READ) + ddir = PCI_DMA_FROMDEVICE; + else + ddir = PCI_DMA_TODEVICE; + for(i=0; i<cmd->req.hdr.sg_cnt; i++) + pci_unmap_page(hba[cmd->ctlr]->pci_dev, cmd->req.sg[i].addr, + cmd->req.sg[i].size, ddir); + + complete_buffers(cmd->rq->bio, ok); + + DBGPX(printk("Done with %p\n", cmd->rq);); + end_that_request_last(cmd->rq); +} + +/* + * The controller will interrupt us upon completion of commands. + * Find the command on the completion queue, remove it, tell the OS and + * try to queue up more IO + */ +static irqreturn_t do_ida_intr(int irq, void *dev_id, struct pt_regs *regs) +{ + ctlr_info_t *h = dev_id; + cmdlist_t *c; + unsigned long istat; + unsigned long flags; + __u32 a,a1; + + istat = h->access.intr_pending(h); + /* Is this interrupt for us? */ + if (istat == 0) + return IRQ_NONE; + + /* + * If there are completed commands in the completion queue, + * we had better do something about it. + */ + spin_lock_irqsave(IDA_LOCK(h->ctlr), flags); + if (istat & FIFO_NOT_EMPTY) { + while((a = h->access.command_completed(h))) { + a1 = a; a &= ~3; + if ((c = h->cmpQ) == NULL) + { + printk(KERN_WARNING "cpqarray: Completion of %08lx ignored\n", (unsigned long)a1); + continue; + } + while(c->busaddr != a) { + c = c->next; + if (c == h->cmpQ) + break; + } + /* + * If we've found the command, take it off the + * completion Q and free it + */ + if (c->busaddr == a) { + removeQ(&h->cmpQ, c); + /* Check for invalid command. + * Controller returns command error, + * But rcode = 0. + */ + + if((a1 & 0x03) && (c->req.hdr.rcode == 0)) + { + c->req.hdr.rcode = RCODE_INVREQ; + } + if (c->type == CMD_RWREQ) { + complete_command(c, 0); + cmd_free(h, c, 1); + } else if (c->type == CMD_IOCTL_PEND) { + c->type = CMD_IOCTL_DONE; + } + continue; + } + } + } + + /* + * See if we can queue up some more IO + */ + do_ida_request(h->queue); + spin_unlock_irqrestore(IDA_LOCK(h->ctlr), flags); + return IRQ_HANDLED; +} + +/* + * This timer was for timing out requests that haven't happened after + * IDA_TIMEOUT. That wasn't such a good idea. This timer is used to + * reset a flags structure so we don't flood the user with + * "Non-Fatal error" messages. + */ +static void ida_timer(unsigned long tdata) +{ + ctlr_info_t *h = (ctlr_info_t*)tdata; + + h->timer.expires = jiffies + IDA_TIMER; + add_timer(&h->timer); + h->misc_tflags = 0; +} + +/* + * ida_ioctl does some miscellaneous stuff like reporting drive geometry, + * setting readahead and submitting commands from userspace to the controller. + */ +static int ida_ioctl(struct inode *inode, struct file *filep, unsigned int cmd, unsigned long arg) +{ + drv_info_t *drv = get_drv(inode->i_bdev->bd_disk); + ctlr_info_t *host = get_host(inode->i_bdev->bd_disk); + int error; + int diskinfo[4]; + struct hd_geometry __user *geo = (struct hd_geometry __user *)arg; + ida_ioctl_t __user *io = (ida_ioctl_t __user *)arg; + ida_ioctl_t *my_io; + + switch(cmd) { + case HDIO_GETGEO: + if (drv->cylinders) { + diskinfo[0] = drv->heads; + diskinfo[1] = drv->sectors; + diskinfo[2] = drv->cylinders; + } else { + diskinfo[0] = 0xff; + diskinfo[1] = 0x3f; + diskinfo[2] = drv->nr_blks / (0xff*0x3f); + } + put_user(diskinfo[0], &geo->heads); + put_user(diskinfo[1], &geo->sectors); + put_user(diskinfo[2], &geo->cylinders); + put_user(get_start_sect(inode->i_bdev), &geo->start); + return 0; + case IDAGETDRVINFO: + if (copy_to_user(&io->c.drv, drv, sizeof(drv_info_t))) + return -EFAULT; + return 0; + case IDAPASSTHRU: + if (!capable(CAP_SYS_RAWIO)) + return -EPERM; + my_io = kmalloc(sizeof(ida_ioctl_t), GFP_KERNEL); + if (!my_io) + return -ENOMEM; + error = -EFAULT; + if (copy_from_user(my_io, io, sizeof(*my_io))) + goto out_passthru; + error = ida_ctlr_ioctl(host, drv - host->drv, my_io); + if (error) + goto out_passthru; + error = -EFAULT; + if (copy_to_user(io, my_io, sizeof(*my_io))) + goto out_passthru; + error = 0; +out_passthru: + kfree(my_io); + return error; + case IDAGETCTLRSIG: + if (!arg) return -EINVAL; + put_user(host->ctlr_sig, (int __user *)arg); + return 0; + case IDAREVALIDATEVOLS: + if (iminor(inode) != 0) + return -ENXIO; + return revalidate_allvol(host); + case IDADRIVERVERSION: + if (!arg) return -EINVAL; + put_user(DRIVER_VERSION, (unsigned long __user *)arg); + return 0; + case IDAGETPCIINFO: + { + + ida_pci_info_struct pciinfo; + + if (!arg) return -EINVAL; + pciinfo.bus = host->pci_dev->bus->number; + pciinfo.dev_fn = host->pci_dev->devfn; + pciinfo.board_id = host->board_id; + if(copy_to_user((void __user *) arg, &pciinfo, + sizeof( ida_pci_info_struct))) + return -EFAULT; + return(0); + } + + default: + return -EINVAL; + } + +} +/* + * ida_ctlr_ioctl is for passing commands to the controller from userspace. + * The command block (io) has already been copied to kernel space for us, + * however, any elements in the sglist need to be copied to kernel space + * or copied back to userspace. + * + * Only root may perform a controller passthru command, however I'm not doing + * any serious sanity checking on the arguments. Doing an IDA_WRITE_MEDIA and + * putting a 64M buffer in the sglist is probably a *bad* idea. + */ +static int ida_ctlr_ioctl(ctlr_info_t *h, int dsk, ida_ioctl_t *io) +{ + int ctlr = h->ctlr; + cmdlist_t *c; + void *p = NULL; + unsigned long flags; + int error; + + if ((c = cmd_alloc(h, 0)) == NULL) + return -ENOMEM; + c->ctlr = ctlr; + c->hdr.unit = (io->unit & UNITVALID) ? (io->unit & ~UNITVALID) : dsk; + c->hdr.size = sizeof(rblk_t) >> 2; + c->size += sizeof(rblk_t); + + c->req.hdr.cmd = io->cmd; + c->req.hdr.blk = io->blk; + c->req.hdr.blk_cnt = io->blk_cnt; + c->type = CMD_IOCTL_PEND; + + /* Pre submit processing */ + switch(io->cmd) { + case PASSTHRU_A: + p = kmalloc(io->sg[0].size, GFP_KERNEL); + if (!p) + { + error = -ENOMEM; + cmd_free(h, c, 0); + return(error); + } + if (copy_from_user(p, io->sg[0].addr, io->sg[0].size)) { + kfree(p); + cmd_free(h, c, 0); + return -EFAULT; + } + c->req.hdr.blk = pci_map_single(h->pci_dev, &(io->c), + sizeof(ida_ioctl_t), + PCI_DMA_BIDIRECTIONAL); + c->req.sg[0].size = io->sg[0].size; + c->req.sg[0].addr = pci_map_single(h->pci_dev, p, + c->req.sg[0].size, PCI_DMA_BIDIRECTIONAL); + c->req.hdr.sg_cnt = 1; + break; + case IDA_READ: + case READ_FLASH_ROM: + case SENSE_CONTROLLER_PERFORMANCE: + p = kmalloc(io->sg[0].size, GFP_KERNEL); + if (!p) + { + error = -ENOMEM; + cmd_free(h, c, 0); + return(error); + } + + c->req.sg[0].size = io->sg[0].size; + c->req.sg[0].addr = pci_map_single(h->pci_dev, p, + c->req.sg[0].size, PCI_DMA_BIDIRECTIONAL); + c->req.hdr.sg_cnt = 1; + break; + case IDA_WRITE: + case IDA_WRITE_MEDIA: + case DIAG_PASS_THRU: + case COLLECT_BUFFER: + case WRITE_FLASH_ROM: + p = kmalloc(io->sg[0].size, GFP_KERNEL); + if (!p) + { + error = -ENOMEM; + cmd_free(h, c, 0); + return(error); + } + if (copy_from_user(p, io->sg[0].addr, io->sg[0].size)) { + kfree(p); + cmd_free(h, c, 0); + return -EFAULT; + } + c->req.sg[0].size = io->sg[0].size; + c->req.sg[0].addr = pci_map_single(h->pci_dev, p, + c->req.sg[0].size, PCI_DMA_BIDIRECTIONAL); + c->req.hdr.sg_cnt = 1; + break; + default: + c->req.sg[0].size = sizeof(io->c); + c->req.sg[0].addr = pci_map_single(h->pci_dev,&io->c, + c->req.sg[0].size, PCI_DMA_BIDIRECTIONAL); + c->req.hdr.sg_cnt = 1; + } + + /* Put the request on the tail of the request queue */ + spin_lock_irqsave(IDA_LOCK(ctlr), flags); + addQ(&h->reqQ, c); + h->Qdepth++; + start_io(h); + spin_unlock_irqrestore(IDA_LOCK(ctlr), flags); + + /* Wait for completion */ + while(c->type != CMD_IOCTL_DONE) + schedule(); + + /* Unmap the DMA */ + pci_unmap_single(h->pci_dev, c->req.sg[0].addr, c->req.sg[0].size, + PCI_DMA_BIDIRECTIONAL); + /* Post submit processing */ + switch(io->cmd) { + case PASSTHRU_A: + pci_unmap_single(h->pci_dev, c->req.hdr.blk, + sizeof(ida_ioctl_t), + PCI_DMA_BIDIRECTIONAL); + case IDA_READ: + case DIAG_PASS_THRU: + case SENSE_CONTROLLER_PERFORMANCE: + case READ_FLASH_ROM: + if (copy_to_user(io->sg[0].addr, p, io->sg[0].size)) { + kfree(p); + return -EFAULT; + } + /* fall through and free p */ + case IDA_WRITE: + case IDA_WRITE_MEDIA: + case COLLECT_BUFFER: + case WRITE_FLASH_ROM: + kfree(p); + break; + default:; + /* Nothing to do */ + } + + io->rcode = c->req.hdr.rcode; + cmd_free(h, c, 0); + return(0); +} + +/* + * Commands are pre-allocated in a large block. Here we use a simple bitmap + * scheme to suballocte them to the driver. Operations that are not time + * critical (and can wait for kmalloc and possibly sleep) can pass in NULL + * as the first argument to get a new command. + */ +static cmdlist_t * cmd_alloc(ctlr_info_t *h, int get_from_pool) +{ + cmdlist_t * c; + int i; + dma_addr_t cmd_dhandle; + + if (!get_from_pool) { + c = (cmdlist_t*)pci_alloc_consistent(h->pci_dev, + sizeof(cmdlist_t), &cmd_dhandle); + if(c==NULL) + return NULL; + } else { + do { + i = find_first_zero_bit(h->cmd_pool_bits, NR_CMDS); + if (i == NR_CMDS) + return NULL; + } while(test_and_set_bit(i&(BITS_PER_LONG-1), h->cmd_pool_bits+(i/BITS_PER_LONG)) != 0); + c = h->cmd_pool + i; + cmd_dhandle = h->cmd_pool_dhandle + i*sizeof(cmdlist_t); + h->nr_allocs++; + } + + memset(c, 0, sizeof(cmdlist_t)); + c->busaddr = cmd_dhandle; + return c; +} + +static void cmd_free(ctlr_info_t *h, cmdlist_t *c, int got_from_pool) +{ + int i; + + if (!got_from_pool) { + pci_free_consistent(h->pci_dev, sizeof(cmdlist_t), c, + c->busaddr); + } else { + i = c - h->cmd_pool; + clear_bit(i&(BITS_PER_LONG-1), h->cmd_pool_bits+(i/BITS_PER_LONG)); + h->nr_frees++; + } +} + +/*********************************************************************** + name: sendcmd + Send a command to an IDA using the memory mapped FIFO interface + and wait for it to complete. + This routine should only be called at init time. +***********************************************************************/ +static int sendcmd( + __u8 cmd, + int ctlr, + void *buff, + size_t size, + unsigned int blk, + unsigned int blkcnt, + unsigned int log_unit ) +{ + cmdlist_t *c; + int complete; + unsigned long temp; + unsigned long i; + ctlr_info_t *info_p = hba[ctlr]; + + c = cmd_alloc(info_p, 1); + if(!c) + return IO_ERROR; + c->ctlr = ctlr; + c->hdr.unit = log_unit; + c->hdr.prio = 0; + c->hdr.size = sizeof(rblk_t) >> 2; + c->size += sizeof(rblk_t); + + /* The request information. */ + c->req.hdr.next = 0; + c->req.hdr.rcode = 0; + c->req.bp = 0; + c->req.hdr.sg_cnt = 1; + c->req.hdr.reserved = 0; + + if (size == 0) + c->req.sg[0].size = 512; + else + c->req.sg[0].size = size; + + c->req.hdr.blk = blk; + c->req.hdr.blk_cnt = blkcnt; + c->req.hdr.cmd = (unsigned char) cmd; + c->req.sg[0].addr = (__u32) pci_map_single(info_p->pci_dev, + buff, c->req.sg[0].size, PCI_DMA_BIDIRECTIONAL); + /* + * Disable interrupt + */ + info_p->access.set_intr_mask(info_p, 0); + /* Make sure there is room in the command FIFO */ + /* Actually it should be completely empty at this time. */ + for (i = 200000; i > 0; i--) { + temp = info_p->access.fifo_full(info_p); + if (temp != 0) { + break; + } + udelay(10); +DBG( + printk(KERN_WARNING "cpqarray ida%d: idaSendPciCmd FIFO full," + " waiting!\n", ctlr); +); + } + /* + * Send the cmd + */ + info_p->access.submit_command(info_p, c); + complete = pollcomplete(ctlr); + + pci_unmap_single(info_p->pci_dev, (dma_addr_t) c->req.sg[0].addr, + c->req.sg[0].size, PCI_DMA_BIDIRECTIONAL); + if (complete != 1) { + if (complete != c->busaddr) { + printk( KERN_WARNING + "cpqarray ida%d: idaSendPciCmd " + "Invalid command list address returned! (%08lx)\n", + ctlr, (unsigned long)complete); + cmd_free(info_p, c, 1); + return (IO_ERROR); + } + } else { + printk( KERN_WARNING + "cpqarray ida%d: idaSendPciCmd Timeout out, " + "No command list address returned!\n", + ctlr); + cmd_free(info_p, c, 1); + return (IO_ERROR); + } + + if (c->req.hdr.rcode & 0x00FE) { + if (!(c->req.hdr.rcode & BIG_PROBLEM)) { + printk( KERN_WARNING + "cpqarray ida%d: idaSendPciCmd, error: " + "Controller failed at init time " + "cmd: 0x%x, return code = 0x%x\n", + ctlr, c->req.hdr.cmd, c->req.hdr.rcode); + + cmd_free(info_p, c, 1); + return (IO_ERROR); + } + } + cmd_free(info_p, c, 1); + return (IO_OK); +} + +/* + * revalidate_allvol is for online array config utilities. After a + * utility reconfigures the drives in the array, it can use this function + * (through an ioctl) to make the driver zap any previous disk structs for + * that controller and get new ones. + * + * Right now I'm using the getgeometry() function to do this, but this + * function should probably be finer grained and allow you to revalidate one + * particualar logical volume (instead of all of them on a particular + * controller). + */ +static int revalidate_allvol(ctlr_info_t *host) +{ + int ctlr = host->ctlr; + int i; + unsigned long flags; + + spin_lock_irqsave(IDA_LOCK(ctlr), flags); + if (host->usage_count > 1) { + spin_unlock_irqrestore(IDA_LOCK(ctlr), flags); + printk(KERN_WARNING "cpqarray: Device busy for volume" + " revalidation (usage=%d)\n", host->usage_count); + return -EBUSY; + } + host->usage_count++; + spin_unlock_irqrestore(IDA_LOCK(ctlr), flags); + + /* + * Set the partition and block size structures for all volumes + * on this controller to zero. We will reread all of this data + */ + set_capacity(ida_gendisk[ctlr][0], 0); + for (i = 1; i < NWD; i++) { + struct gendisk *disk = ida_gendisk[ctlr][i]; + if (disk->flags & GENHD_FL_UP) + del_gendisk(disk); + } + memset(host->drv, 0, sizeof(drv_info_t)*NWD); + + /* + * Tell the array controller not to give us any interrupts while + * we check the new geometry. Then turn interrupts back on when + * we're done. + */ + host->access.set_intr_mask(host, 0); + getgeometry(ctlr); + host->access.set_intr_mask(host, FIFO_NOT_EMPTY); + + for(i=0; i<NWD; i++) { + struct gendisk *disk = ida_gendisk[ctlr][i]; + drv_info_t *drv = &host->drv[i]; + if (i && !drv->nr_blks) + continue; + blk_queue_hardsect_size(host->queue, drv->blk_size); + set_capacity(disk, drv->nr_blks); + disk->queue = host->queue; + disk->private_data = drv; + if (i) + add_disk(disk); + } + + host->usage_count--; + return 0; +} + +static int ida_revalidate(struct gendisk *disk) +{ + drv_info_t *drv = disk->private_data; + set_capacity(disk, drv->nr_blks); + return 0; +} + +/******************************************************************** + name: pollcomplete + Wait polling for a command to complete. + The memory mapped FIFO is polled for the completion. + Used only at init time, interrupts disabled. + ********************************************************************/ +static int pollcomplete(int ctlr) +{ + int done; + int i; + + /* Wait (up to 2 seconds) for a command to complete */ + + for (i = 200000; i > 0; i--) { + done = hba[ctlr]->access.command_completed(hba[ctlr]); + if (done == 0) { + udelay(10); /* a short fixed delay */ + } else + return (done); + } + /* Invalid address to tell caller we ran out of time */ + return 1; +} +/***************************************************************** + start_fwbk + Starts controller firmwares background processing. + Currently only the Integrated Raid controller needs this done. + If the PCI mem address registers are written to after this, + data corruption may occur +*****************************************************************/ +static void start_fwbk(int ctlr) +{ + id_ctlr_t *id_ctlr_buf; + int ret_code; + + if( (hba[ctlr]->board_id != 0x40400E11) + && (hba[ctlr]->board_id != 0x40480E11) ) + + /* Not a Integrated Raid, so there is nothing for us to do */ + return; + printk(KERN_DEBUG "cpqarray: Starting firmware's background" + " processing\n"); + /* Command does not return anything, but idasend command needs a + buffer */ + id_ctlr_buf = (id_ctlr_t *)kmalloc(sizeof(id_ctlr_t), GFP_KERNEL); + if(id_ctlr_buf==NULL) + { + printk(KERN_WARNING "cpqarray: Out of memory. " + "Unable to start background processing.\n"); + return; + } + ret_code = sendcmd(RESUME_BACKGROUND_ACTIVITY, ctlr, + id_ctlr_buf, 0, 0, 0, 0); + if(ret_code != IO_OK) + printk(KERN_WARNING "cpqarray: Unable to start" + " background processing\n"); + + kfree(id_ctlr_buf); +} +/***************************************************************** + getgeometry + Get ida logical volume geometry from the controller + This is a large bit of code which once existed in two flavors, + It is used only at init time. +*****************************************************************/ +static void getgeometry(int ctlr) +{ + id_log_drv_t *id_ldrive; + id_ctlr_t *id_ctlr_buf; + sense_log_drv_stat_t *id_lstatus_buf; + config_t *sense_config_buf; + unsigned int log_unit, log_index; + int ret_code, size; + drv_info_t *drv; + ctlr_info_t *info_p = hba[ctlr]; + int i; + + info_p->log_drv_map = 0; + + id_ldrive = (id_log_drv_t *)kmalloc(sizeof(id_log_drv_t), GFP_KERNEL); + if(id_ldrive == NULL) + { + printk( KERN_ERR "cpqarray: out of memory.\n"); + return; + } + + id_ctlr_buf = (id_ctlr_t *)kmalloc(sizeof(id_ctlr_t), GFP_KERNEL); + if(id_ctlr_buf == NULL) + { + kfree(id_ldrive); + printk( KERN_ERR "cpqarray: out of memory.\n"); + return; + } + + id_lstatus_buf = (sense_log_drv_stat_t *)kmalloc(sizeof(sense_log_drv_stat_t), GFP_KERNEL); + if(id_lstatus_buf == NULL) + { + kfree(id_ctlr_buf); + kfree(id_ldrive); + printk( KERN_ERR "cpqarray: out of memory.\n"); + return; + } + + sense_config_buf = (config_t *)kmalloc(sizeof(config_t), GFP_KERNEL); + if(sense_config_buf == NULL) + { + kfree(id_lstatus_buf); + kfree(id_ctlr_buf); + kfree(id_ldrive); + printk( KERN_ERR "cpqarray: out of memory.\n"); + return; + } + + memset(id_ldrive, 0, sizeof(id_log_drv_t)); + memset(id_ctlr_buf, 0, sizeof(id_ctlr_t)); + memset(id_lstatus_buf, 0, sizeof(sense_log_drv_stat_t)); + memset(sense_config_buf, 0, sizeof(config_t)); + + info_p->phys_drives = 0; + info_p->log_drv_map = 0; + info_p->drv_assign_map = 0; + info_p->drv_spare_map = 0; + info_p->mp_failed_drv_map = 0; /* only initialized here */ + /* Get controllers info for this logical drive */ + ret_code = sendcmd(ID_CTLR, ctlr, id_ctlr_buf, 0, 0, 0, 0); + if (ret_code == IO_ERROR) { + /* + * If can't get controller info, set the logical drive map to 0, + * so the idastubopen will fail on all logical drives + * on the controller. + */ + /* Free all the buffers and return */ + printk(KERN_ERR "cpqarray: error sending ID controller\n"); + kfree(sense_config_buf); + kfree(id_lstatus_buf); + kfree(id_ctlr_buf); + kfree(id_ldrive); + return; + } + + info_p->log_drives = id_ctlr_buf->nr_drvs; + for(i=0;i<4;i++) + info_p->firm_rev[i] = id_ctlr_buf->firm_rev[i]; + info_p->ctlr_sig = id_ctlr_buf->cfg_sig; + + printk(" (%s)\n", info_p->product_name); + /* + * Initialize logical drive map to zero + */ + log_index = 0; + /* + * Get drive geometry for all logical drives + */ + if (id_ctlr_buf->nr_drvs > 16) + printk(KERN_WARNING "cpqarray ida%d: This driver supports " + "16 logical drives per controller.\n. " + " Additional drives will not be " + "detected\n", ctlr); + + for (log_unit = 0; + (log_index < id_ctlr_buf->nr_drvs) + && (log_unit < NWD); + log_unit++) { + struct gendisk *disk = ida_gendisk[ctlr][log_unit]; + + size = sizeof(sense_log_drv_stat_t); + + /* + Send "Identify logical drive status" cmd + */ + ret_code = sendcmd(SENSE_LOG_DRV_STAT, + ctlr, id_lstatus_buf, size, 0, 0, log_unit); + if (ret_code == IO_ERROR) { + /* + If can't get logical drive status, set + the logical drive map to 0, so the + idastubopen will fail for all logical drives + on the controller. + */ + info_p->log_drv_map = 0; + printk( KERN_WARNING + "cpqarray ida%d: idaGetGeometry - Controller" + " failed to report status of logical drive %d\n" + "Access to this controller has been disabled\n", + ctlr, log_unit); + /* Free all the buffers and return */ + kfree(sense_config_buf); + kfree(id_lstatus_buf); + kfree(id_ctlr_buf); + kfree(id_ldrive); + return; + } + /* + Make sure the logical drive is configured + */ + if (id_lstatus_buf->status != LOG_NOT_CONF) { + ret_code = sendcmd(ID_LOG_DRV, ctlr, id_ldrive, + sizeof(id_log_drv_t), 0, 0, log_unit); + /* + If error, the bit for this + logical drive won't be set and + idastubopen will return error. + */ + if (ret_code != IO_ERROR) { + drv = &info_p->drv[log_unit]; + drv->blk_size = id_ldrive->blk_size; + drv->nr_blks = id_ldrive->nr_blks; + drv->cylinders = id_ldrive->drv.cyl; + drv->heads = id_ldrive->drv.heads; + drv->sectors = id_ldrive->drv.sect_per_track; + info_p->log_drv_map |= (1 << log_unit); + + printk(KERN_INFO "cpqarray ida/c%dd%d: blksz=%d nr_blks=%d\n", + ctlr, log_unit, drv->blk_size, drv->nr_blks); + ret_code = sendcmd(SENSE_CONFIG, + ctlr, sense_config_buf, + sizeof(config_t), 0, 0, log_unit); + if (ret_code == IO_ERROR) { + info_p->log_drv_map = 0; + /* Free all the buffers and return */ + printk(KERN_ERR "cpqarray: error sending sense config\n"); + kfree(sense_config_buf); + kfree(id_lstatus_buf); + kfree(id_ctlr_buf); + kfree(id_ldrive); + return; + + } + + sprintf(disk->devfs_name, "ida/c%dd%d", ctlr, log_unit); + + info_p->phys_drives = + sense_config_buf->ctlr_phys_drv; + info_p->drv_assign_map + |= sense_config_buf->drv_asgn_map; + info_p->drv_assign_map + |= sense_config_buf->spare_asgn_map; + info_p->drv_spare_map + |= sense_config_buf->spare_asgn_map; + } /* end of if no error on id_ldrive */ + log_index = log_index + 1; + } /* end of if logical drive configured */ + } /* end of for log_unit */ + kfree(sense_config_buf); + kfree(id_ldrive); + kfree(id_lstatus_buf); + kfree(id_ctlr_buf); + return; + +} + +static void __exit cpqarray_exit(void) +{ + int i; + + pci_unregister_driver(&cpqarray_pci_driver); + + /* Double check that all controller entries have been removed */ + for(i=0; i<MAX_CTLR; i++) { + if (hba[i] != NULL) { + printk(KERN_WARNING "cpqarray: Removing EISA " + "controller %d\n", i); + cpqarray_remove_one_eisa(i); + } + } + + devfs_remove("ida"); + remove_proc_entry("cpqarray", proc_root_driver); +} + +module_init(cpqarray_init) +module_exit(cpqarray_exit) diff --git a/drivers/block/cpqarray.h b/drivers/block/cpqarray.h new file mode 100644 index 00000000000..be73e9d579c --- /dev/null +++ b/drivers/block/cpqarray.h @@ -0,0 +1,126 @@ +/* + * Disk Array driver for Compaq SMART2 Controllers + * Copyright 1998 Compaq Computer Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + * If you want to make changes, improve or add functionality to this + * driver, you'll probably need the Compaq Array Controller Interface + * Specificiation (Document number ECG086/1198) + */ +#ifndef CPQARRAY_H +#define CPQARRAY_H + +#ifdef __KERNEL__ +#include <linux/blkdev.h> +#include <linux/slab.h> +#include <linux/proc_fs.h> +#include <linux/timer.h> +#endif + +#include "ida_cmd.h" + +#define IO_OK 0 +#define IO_ERROR 1 +#define NWD 16 +#define NWD_SHIFT 4 + +#define IDA_TIMER (5*HZ) +#define IDA_TIMEOUT (10*HZ) + +#define MISC_NONFATAL_WARN 0x01 + +typedef struct { + unsigned blk_size; + unsigned nr_blks; + unsigned cylinders; + unsigned heads; + unsigned sectors; + int usage_count; +} drv_info_t; + +#ifdef __KERNEL__ + +struct ctlr_info; +typedef struct ctlr_info ctlr_info_t; + +struct access_method { + void (*submit_command)(ctlr_info_t *h, cmdlist_t *c); + void (*set_intr_mask)(ctlr_info_t *h, unsigned long val); + unsigned long (*fifo_full)(ctlr_info_t *h); + unsigned long (*intr_pending)(ctlr_info_t *h); + unsigned long (*command_completed)(ctlr_info_t *h); +}; + +struct board_type { + __u32 board_id; + char *product_name; + struct access_method *access; +}; + +struct ctlr_info { + int ctlr; + char devname[8]; + __u32 log_drv_map; + __u32 drv_assign_map; + __u32 drv_spare_map; + __u32 mp_failed_drv_map; + + char firm_rev[4]; + int ctlr_sig; + + int log_drives; + int phys_drives; + + struct pci_dev *pci_dev; /* NULL if EISA */ + __u32 board_id; + char *product_name; + + void __iomem *vaddr; + unsigned long paddr; + unsigned long io_mem_addr; + unsigned long io_mem_length; + int intr; + int usage_count; + drv_info_t drv[NWD]; + struct proc_dir_entry *proc; + + struct access_method access; + + cmdlist_t *reqQ; + cmdlist_t *cmpQ; + cmdlist_t *cmd_pool; + dma_addr_t cmd_pool_dhandle; + unsigned long *cmd_pool_bits; + struct request_queue *queue; + spinlock_t lock; + + unsigned int Qdepth; + unsigned int maxQsinceinit; + + unsigned int nr_requests; + unsigned int nr_allocs; + unsigned int nr_frees; + struct timer_list timer; + unsigned int misc_tflags; +}; + +#define IDA_LOCK(i) (&hba[i]->lock) + +#endif + +#endif /* CPQARRAY_H */ diff --git a/drivers/block/cryptoloop.c b/drivers/block/cryptoloop.c new file mode 100644 index 00000000000..5be6f998d8c --- /dev/null +++ b/drivers/block/cryptoloop.c @@ -0,0 +1,268 @@ +/* + Linux loop encryption enabling module + + Copyright (C) 2002 Herbert Valerio Riedel <hvr@gnu.org> + Copyright (C) 2003 Fruhwirth Clemens <clemens@endorphin.org> + + This module is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This module is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this module; if not, write to the Free Software + Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include <linux/module.h> + +#include <linux/init.h> +#include <linux/string.h> +#include <linux/crypto.h> +#include <linux/blkdev.h> +#include <linux/loop.h> +#include <asm/semaphore.h> +#include <asm/uaccess.h> + +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("loop blockdevice transferfunction adaptor / CryptoAPI"); +MODULE_AUTHOR("Herbert Valerio Riedel <hvr@gnu.org>"); + +#define LOOP_IV_SECTOR_BITS 9 +#define LOOP_IV_SECTOR_SIZE (1 << LOOP_IV_SECTOR_BITS) + +static int +cryptoloop_init(struct loop_device *lo, const struct loop_info64 *info) +{ + int err = -EINVAL; + char cms[LO_NAME_SIZE]; /* cipher-mode string */ + char *cipher; + char *mode; + char *cmsp = cms; /* c-m string pointer */ + struct crypto_tfm *tfm = NULL; + + /* encryption breaks for non sector aligned offsets */ + + if (info->lo_offset % LOOP_IV_SECTOR_SIZE) + goto out; + + strncpy(cms, info->lo_crypt_name, LO_NAME_SIZE); + cms[LO_NAME_SIZE - 1] = 0; + cipher = strsep(&cmsp, "-"); + mode = strsep(&cmsp, "-"); + + if (mode == NULL || strcmp(mode, "cbc") == 0) + tfm = crypto_alloc_tfm(cipher, CRYPTO_TFM_MODE_CBC); + else if (strcmp(mode, "ecb") == 0) + tfm = crypto_alloc_tfm(cipher, CRYPTO_TFM_MODE_ECB); + if (tfm == NULL) + return -EINVAL; + + err = tfm->crt_u.cipher.cit_setkey(tfm, info->lo_encrypt_key, + info->lo_encrypt_key_size); + + if (err != 0) + goto out_free_tfm; + + lo->key_data = tfm; + return 0; + + out_free_tfm: + crypto_free_tfm(tfm); + + out: + return err; +} + + +typedef int (*encdec_ecb_t)(struct crypto_tfm *tfm, + struct scatterlist *sg_out, + struct scatterlist *sg_in, + unsigned int nsg); + + +static int +cryptoloop_transfer_ecb(struct loop_device *lo, int cmd, + struct page *raw_page, unsigned raw_off, + struct page *loop_page, unsigned loop_off, + int size, sector_t IV) +{ + struct crypto_tfm *tfm = (struct crypto_tfm *) lo->key_data; + struct scatterlist sg_out = { NULL, }; + struct scatterlist sg_in = { NULL, }; + + encdec_ecb_t encdecfunc; + struct page *in_page, *out_page; + unsigned in_offs, out_offs; + + if (cmd == READ) { + in_page = raw_page; + in_offs = raw_off; + out_page = loop_page; + out_offs = loop_off; + encdecfunc = tfm->crt_u.cipher.cit_decrypt; + } else { + in_page = loop_page; + in_offs = loop_off; + out_page = raw_page; + out_offs = raw_off; + encdecfunc = tfm->crt_u.cipher.cit_encrypt; + } + + while (size > 0) { + const int sz = min(size, LOOP_IV_SECTOR_SIZE); + + sg_in.page = in_page; + sg_in.offset = in_offs; + sg_in.length = sz; + + sg_out.page = out_page; + sg_out.offset = out_offs; + sg_out.length = sz; + + encdecfunc(tfm, &sg_out, &sg_in, sz); + + size -= sz; + in_offs += sz; + out_offs += sz; + } + + return 0; +} + +typedef int (*encdec_cbc_t)(struct crypto_tfm *tfm, + struct scatterlist *sg_out, + struct scatterlist *sg_in, + unsigned int nsg, u8 *iv); + +static int +cryptoloop_transfer_cbc(struct loop_device *lo, int cmd, + struct page *raw_page, unsigned raw_off, + struct page *loop_page, unsigned loop_off, + int size, sector_t IV) +{ + struct crypto_tfm *tfm = (struct crypto_tfm *) lo->key_data; + struct scatterlist sg_out = { NULL, }; + struct scatterlist sg_in = { NULL, }; + + encdec_cbc_t encdecfunc; + struct page *in_page, *out_page; + unsigned in_offs, out_offs; + + if (cmd == READ) { + in_page = raw_page; + in_offs = raw_off; + out_page = loop_page; + out_offs = loop_off; + encdecfunc = tfm->crt_u.cipher.cit_decrypt_iv; + } else { + in_page = loop_page; + in_offs = loop_off; + out_page = raw_page; + out_offs = raw_off; + encdecfunc = tfm->crt_u.cipher.cit_encrypt_iv; + } + + while (size > 0) { + const int sz = min(size, LOOP_IV_SECTOR_SIZE); + u32 iv[4] = { 0, }; + iv[0] = cpu_to_le32(IV & 0xffffffff); + + sg_in.page = in_page; + sg_in.offset = in_offs; + sg_in.length = sz; + + sg_out.page = out_page; + sg_out.offset = out_offs; + sg_out.length = sz; + + encdecfunc(tfm, &sg_out, &sg_in, sz, (u8 *)iv); + + IV++; + size -= sz; + in_offs += sz; + out_offs += sz; + } + + return 0; +} + +static int +cryptoloop_transfer(struct loop_device *lo, int cmd, + struct page *raw_page, unsigned raw_off, + struct page *loop_page, unsigned loop_off, + int size, sector_t IV) +{ + struct crypto_tfm *tfm = (struct crypto_tfm *) lo->key_data; + if(tfm->crt_cipher.cit_mode == CRYPTO_TFM_MODE_ECB) + { + lo->transfer = cryptoloop_transfer_ecb; + return cryptoloop_transfer_ecb(lo, cmd, raw_page, raw_off, + loop_page, loop_off, size, IV); + } + if(tfm->crt_cipher.cit_mode == CRYPTO_TFM_MODE_CBC) + { + lo->transfer = cryptoloop_transfer_cbc; + return cryptoloop_transfer_cbc(lo, cmd, raw_page, raw_off, + loop_page, loop_off, size, IV); + } + + /* This is not supposed to happen */ + + printk( KERN_ERR "cryptoloop: unsupported cipher mode in cryptoloop_transfer!\n"); + return -EINVAL; +} + +static int +cryptoloop_ioctl(struct loop_device *lo, int cmd, unsigned long arg) +{ + return -EINVAL; +} + +static int +cryptoloop_release(struct loop_device *lo) +{ + struct crypto_tfm *tfm = (struct crypto_tfm *) lo->key_data; + if (tfm != NULL) { + crypto_free_tfm(tfm); + lo->key_data = NULL; + return 0; + } + printk(KERN_ERR "cryptoloop_release(): tfm == NULL?\n"); + return -EINVAL; +} + +static struct loop_func_table cryptoloop_funcs = { + .number = LO_CRYPT_CRYPTOAPI, + .init = cryptoloop_init, + .ioctl = cryptoloop_ioctl, + .transfer = cryptoloop_transfer, + .release = cryptoloop_release, + .owner = THIS_MODULE +}; + +static int __init +init_cryptoloop(void) +{ + int rc = loop_register_transfer(&cryptoloop_funcs); + + if (rc) + printk(KERN_ERR "cryptoloop: loop_register_transfer failed\n"); + return rc; +} + +static void __exit +cleanup_cryptoloop(void) +{ + if (loop_unregister_transfer(LO_CRYPT_CRYPTOAPI)) + printk(KERN_ERR + "cryptoloop: loop_unregister_transfer failed\n"); +} + +module_init(init_cryptoloop); +module_exit(cleanup_cryptoloop); diff --git a/drivers/block/deadline-iosched.c b/drivers/block/deadline-iosched.c new file mode 100644 index 00000000000..d63d34c671f --- /dev/null +++ b/drivers/block/deadline-iosched.c @@ -0,0 +1,967 @@ +/* + * linux/drivers/block/deadline-iosched.c + * + * Deadline i/o scheduler. + * + * Copyright (C) 2002 Jens Axboe <axboe@suse.de> + */ +#include <linux/kernel.h> +#include <linux/fs.h> +#include <linux/blkdev.h> +#include <linux/elevator.h> +#include <linux/bio.h> +#include <linux/config.h> +#include <linux/module.h> +#include <linux/slab.h> +#include <linux/init.h> +#include <linux/compiler.h> +#include <linux/hash.h> +#include <linux/rbtree.h> + +/* + * See Documentation/block/deadline-iosched.txt + */ +static int read_expire = HZ / 2; /* max time before a read is submitted. */ +static int write_expire = 5 * HZ; /* ditto for writes, these limits are SOFT! */ +static int writes_starved = 2; /* max times reads can starve a write */ +static int fifo_batch = 16; /* # of sequential requests treated as one + by the above parameters. For throughput. */ + +static const int deadline_hash_shift = 5; +#define DL_HASH_BLOCK(sec) ((sec) >> 3) +#define DL_HASH_FN(sec) (hash_long(DL_HASH_BLOCK((sec)), deadline_hash_shift)) +#define DL_HASH_ENTRIES (1 << deadline_hash_shift) +#define rq_hash_key(rq) ((rq)->sector + (rq)->nr_sectors) +#define list_entry_hash(ptr) list_entry((ptr), struct deadline_rq, hash) +#define ON_HASH(drq) (drq)->on_hash + +struct deadline_data { + /* + * run time data + */ + + /* + * requests (deadline_rq s) are present on both sort_list and fifo_list + */ + struct rb_root sort_list[2]; + struct list_head fifo_list[2]; + + /* + * next in sort order. read, write or both are NULL + */ + struct deadline_rq *next_drq[2]; + struct list_head *dispatch; /* driver dispatch queue */ + struct list_head *hash; /* request hash */ + unsigned int batching; /* number of sequential requests made */ + sector_t last_sector; /* head position */ + unsigned int starved; /* times reads have starved writes */ + + /* + * settings that change how the i/o scheduler behaves + */ + int fifo_expire[2]; + int fifo_batch; + int writes_starved; + int front_merges; + + mempool_t *drq_pool; +}; + +/* + * pre-request data. + */ +struct deadline_rq { + /* + * rbtree index, key is the starting offset + */ + struct rb_node rb_node; + sector_t rb_key; + + struct request *request; + + /* + * request hash, key is the ending offset (for back merge lookup) + */ + struct list_head hash; + char on_hash; + + /* + * expire fifo + */ + struct list_head fifo; + unsigned long expires; +}; + +static void deadline_move_request(struct deadline_data *dd, struct deadline_rq *drq); + +static kmem_cache_t *drq_pool; + +#define RQ_DATA(rq) ((struct deadline_rq *) (rq)->elevator_private) + +/* + * the back merge hash support functions + */ +static inline void __deadline_del_drq_hash(struct deadline_rq *drq) +{ + drq->on_hash = 0; + list_del_init(&drq->hash); +} + +static inline void deadline_del_drq_hash(struct deadline_rq *drq) +{ + if (ON_HASH(drq)) + __deadline_del_drq_hash(drq); +} + +static void +deadline_remove_merge_hints(request_queue_t *q, struct deadline_rq *drq) +{ + deadline_del_drq_hash(drq); + + if (q->last_merge == drq->request) + q->last_merge = NULL; +} + +static inline void +deadline_add_drq_hash(struct deadline_data *dd, struct deadline_rq *drq) +{ + struct request *rq = drq->request; + + BUG_ON(ON_HASH(drq)); + + drq->on_hash = 1; + list_add(&drq->hash, &dd->hash[DL_HASH_FN(rq_hash_key(rq))]); +} + +/* + * move hot entry to front of chain + */ +static inline void +deadline_hot_drq_hash(struct deadline_data *dd, struct deadline_rq *drq) +{ + struct request *rq = drq->request; + struct list_head *head = &dd->hash[DL_HASH_FN(rq_hash_key(rq))]; + + if (ON_HASH(drq) && drq->hash.prev != head) { + list_del(&drq->hash); + list_add(&drq->hash, head); + } +} + +static struct request * +deadline_find_drq_hash(struct deadline_data *dd, sector_t offset) +{ + struct list_head *hash_list = &dd->hash[DL_HASH_FN(offset)]; + struct list_head *entry, *next = hash_list->next; + + while ((entry = next) != hash_list) { + struct deadline_rq *drq = list_entry_hash(entry); + struct request *__rq = drq->request; + + next = entry->next; + + BUG_ON(!ON_HASH(drq)); + + if (!rq_mergeable(__rq)) { + __deadline_del_drq_hash(drq); + continue; + } + + if (rq_hash_key(__rq) == offset) + return __rq; + } + + return NULL; +} + +/* + * rb tree support functions + */ +#define RB_NONE (2) +#define RB_EMPTY(root) ((root)->rb_node == NULL) +#define ON_RB(node) ((node)->rb_color != RB_NONE) +#define RB_CLEAR(node) ((node)->rb_color = RB_NONE) +#define rb_entry_drq(node) rb_entry((node), struct deadline_rq, rb_node) +#define DRQ_RB_ROOT(dd, drq) (&(dd)->sort_list[rq_data_dir((drq)->request)]) +#define rq_rb_key(rq) (rq)->sector + +static struct deadline_rq * +__deadline_add_drq_rb(struct deadline_data *dd, struct deadline_rq *drq) +{ + struct rb_node **p = &DRQ_RB_ROOT(dd, drq)->rb_node; + struct rb_node *parent = NULL; + struct deadline_rq *__drq; + + while (*p) { + parent = *p; + __drq = rb_entry_drq(parent); + + if (drq->rb_key < __drq->rb_key) + p = &(*p)->rb_left; + else if (drq->rb_key > __drq->rb_key) + p = &(*p)->rb_right; + else + return __drq; + } + + rb_link_node(&drq->rb_node, parent, p); + return NULL; +} + +static void +deadline_add_drq_rb(struct deadline_data *dd, struct deadline_rq *drq) +{ + struct deadline_rq *__alias; + + drq->rb_key = rq_rb_key(drq->request); + +retry: + __alias = __deadline_add_drq_rb(dd, drq); + if (!__alias) { + rb_insert_color(&drq->rb_node, DRQ_RB_ROOT(dd, drq)); + return; + } + + deadline_move_request(dd, __alias); + goto retry; +} + +static inline void +deadline_del_drq_rb(struct deadline_data *dd, struct deadline_rq *drq) +{ + const int data_dir = rq_data_dir(drq->request); + + if (dd->next_drq[data_dir] == drq) { + struct rb_node *rbnext = rb_next(&drq->rb_node); + + dd->next_drq[data_dir] = NULL; + if (rbnext) + dd->next_drq[data_dir] = rb_entry_drq(rbnext); + } + + if (ON_RB(&drq->rb_node)) { + rb_erase(&drq->rb_node, DRQ_RB_ROOT(dd, drq)); + RB_CLEAR(&drq->rb_node); + } +} + +static struct request * +deadline_find_drq_rb(struct deadline_data *dd, sector_t sector, int data_dir) +{ + struct rb_node *n = dd->sort_list[data_dir].rb_node; + struct deadline_rq *drq; + + while (n) { + drq = rb_entry_drq(n); + + if (sector < drq->rb_key) + n = n->rb_left; + else if (sector > drq->rb_key) + n = n->rb_right; + else + return drq->request; + } + + return NULL; +} + +/* + * deadline_find_first_drq finds the first (lowest sector numbered) request + * for the specified data_dir. Used to sweep back to the start of the disk + * (1-way elevator) after we process the last (highest sector) request. + */ +static struct deadline_rq * +deadline_find_first_drq(struct deadline_data *dd, int data_dir) +{ + struct rb_node *n = dd->sort_list[data_dir].rb_node; + + for (;;) { + if (n->rb_left == NULL) + return rb_entry_drq(n); + + n = n->rb_left; + } +} + +/* + * add drq to rbtree and fifo + */ +static inline void +deadline_add_request(struct request_queue *q, struct request *rq) +{ + struct deadline_data *dd = q->elevator->elevator_data; + struct deadline_rq *drq = RQ_DATA(rq); + + const int data_dir = rq_data_dir(drq->request); + + deadline_add_drq_rb(dd, drq); + /* + * set expire time (only used for reads) and add to fifo list + */ + drq->expires = jiffies + dd->fifo_expire[data_dir]; + list_add_tail(&drq->fifo, &dd->fifo_list[data_dir]); + + if (rq_mergeable(rq)) { + deadline_add_drq_hash(dd, drq); + + if (!q->last_merge) + q->last_merge = rq; + } +} + +/* + * remove rq from rbtree, fifo, and hash + */ +static void deadline_remove_request(request_queue_t *q, struct request *rq) +{ + struct deadline_rq *drq = RQ_DATA(rq); + + if (drq) { + struct deadline_data *dd = q->elevator->elevator_data; + + list_del_init(&drq->fifo); + deadline_remove_merge_hints(q, drq); + deadline_del_drq_rb(dd, drq); + } +} + +static int +deadline_merge(request_queue_t *q, struct request **req, struct bio *bio) +{ + struct deadline_data *dd = q->elevator->elevator_data; + struct request *__rq; + int ret; + + /* + * try last_merge to avoid going to hash + */ + ret = elv_try_last_merge(q, bio); + if (ret != ELEVATOR_NO_MERGE) { + __rq = q->last_merge; + goto out_insert; + } + + /* + * see if the merge hash can satisfy a back merge + */ + __rq = deadline_find_drq_hash(dd, bio->bi_sector); + if (__rq) { + BUG_ON(__rq->sector + __rq->nr_sectors != bio->bi_sector); + + if (elv_rq_merge_ok(__rq, bio)) { + ret = ELEVATOR_BACK_MERGE; + goto out; + } + } + + /* + * check for front merge + */ + if (dd->front_merges) { + sector_t rb_key = bio->bi_sector + bio_sectors(bio); + + __rq = deadline_find_drq_rb(dd, rb_key, bio_data_dir(bio)); + if (__rq) { + BUG_ON(rb_key != rq_rb_key(__rq)); + + if (elv_rq_merge_ok(__rq, bio)) { + ret = ELEVATOR_FRONT_MERGE; + goto out; + } + } + } + + return ELEVATOR_NO_MERGE; +out: + q->last_merge = __rq; +out_insert: + if (ret) + deadline_hot_drq_hash(dd, RQ_DATA(__rq)); + *req = __rq; + return ret; +} + +static void deadline_merged_request(request_queue_t *q, struct request *req) +{ + struct deadline_data *dd = q->elevator->elevator_data; + struct deadline_rq *drq = RQ_DATA(req); + + /* + * hash always needs to be repositioned, key is end sector + */ + deadline_del_drq_hash(drq); + deadline_add_drq_hash(dd, drq); + + /* + * if the merge was a front merge, we need to reposition request + */ + if (rq_rb_key(req) != drq->rb_key) { + deadline_del_drq_rb(dd, drq); + deadline_add_drq_rb(dd, drq); + } + + q->last_merge = req; +} + +static void +deadline_merged_requests(request_queue_t *q, struct request *req, + struct request *next) +{ + struct deadline_data *dd = q->elevator->elevator_data; + struct deadline_rq *drq = RQ_DATA(req); + struct deadline_rq *dnext = RQ_DATA(next); + + BUG_ON(!drq); + BUG_ON(!dnext); + + /* + * reposition drq (this is the merged request) in hash, and in rbtree + * in case of a front merge + */ + deadline_del_drq_hash(drq); + deadline_add_drq_hash(dd, drq); + + if (rq_rb_key(req) != drq->rb_key) { + deadline_del_drq_rb(dd, drq); + deadline_add_drq_rb(dd, drq); + } + + /* + * if dnext expires before drq, assign its expire time to drq + * and move into dnext position (dnext will be deleted) in fifo + */ + if (!list_empty(&drq->fifo) && !list_empty(&dnext->fifo)) { + if (time_before(dnext->expires, drq->expires)) { + list_move(&drq->fifo, &dnext->fifo); + drq->expires = dnext->expires; + } + } + + /* + * kill knowledge of next, this one is a goner + */ + deadline_remove_request(q, next); +} + +/* + * move request from sort list to dispatch queue. + */ +static inline void +deadline_move_to_dispatch(struct deadline_data *dd, struct deadline_rq *drq) +{ + request_queue_t *q = drq->request->q; + + deadline_remove_request(q, drq->request); + list_add_tail(&drq->request->queuelist, dd->dispatch); +} + +/* + * move an entry to dispatch queue + */ +static void +deadline_move_request(struct deadline_data *dd, struct deadline_rq *drq) +{ + const int data_dir = rq_data_dir(drq->request); + struct rb_node *rbnext = rb_next(&drq->rb_node); + + dd->next_drq[READ] = NULL; + dd->next_drq[WRITE] = NULL; + + if (rbnext) + dd->next_drq[data_dir] = rb_entry_drq(rbnext); + + dd->last_sector = drq->request->sector + drq->request->nr_sectors; + + /* + * take it off the sort and fifo list, move + * to dispatch queue + */ + deadline_move_to_dispatch(dd, drq); +} + +#define list_entry_fifo(ptr) list_entry((ptr), struct deadline_rq, fifo) + +/* + * deadline_check_fifo returns 0 if there are no expired reads on the fifo, + * 1 otherwise. Requires !list_empty(&dd->fifo_list[data_dir]) + */ +static inline int deadline_check_fifo(struct deadline_data *dd, int ddir) +{ + struct deadline_rq *drq = list_entry_fifo(dd->fifo_list[ddir].next); + + /* + * drq is expired! + */ + if (time_after(jiffies, drq->expires)) + return 1; + + return 0; +} + +/* + * deadline_dispatch_requests selects the best request according to + * read/write expire, fifo_batch, etc + */ +static int deadline_dispatch_requests(struct deadline_data *dd) +{ + const int reads = !list_empty(&dd->fifo_list[READ]); + const int writes = !list_empty(&dd->fifo_list[WRITE]); + struct deadline_rq *drq; + int data_dir, other_dir; + + /* + * batches are currently reads XOR writes + */ + drq = NULL; + + if (dd->next_drq[READ]) + drq = dd->next_drq[READ]; + + if (dd->next_drq[WRITE]) + drq = dd->next_drq[WRITE]; + + if (drq) { + /* we have a "next request" */ + + if (dd->last_sector != drq->request->sector) + /* end the batch on a non sequential request */ + dd->batching += dd->fifo_batch; + + if (dd->batching < dd->fifo_batch) + /* we are still entitled to batch */ + goto dispatch_request; + } + + /* + * at this point we are not running a batch. select the appropriate + * data direction (read / write) + */ + + if (reads) { + BUG_ON(RB_EMPTY(&dd->sort_list[READ])); + + if (writes && (dd->starved++ >= dd->writes_starved)) + goto dispatch_writes; + + data_dir = READ; + other_dir = WRITE; + + goto dispatch_find_request; + } + + /* + * there are either no reads or writes have been starved + */ + + if (writes) { +dispatch_writes: + BUG_ON(RB_EMPTY(&dd->sort_list[WRITE])); + + dd->starved = 0; + + data_dir = WRITE; + other_dir = READ; + + goto dispatch_find_request; + } + + return 0; + +dispatch_find_request: + /* + * we are not running a batch, find best request for selected data_dir + */ + if (deadline_check_fifo(dd, data_dir)) { + /* An expired request exists - satisfy it */ + dd->batching = 0; + drq = list_entry_fifo(dd->fifo_list[data_dir].next); + + } else if (dd->next_drq[data_dir]) { + /* + * The last req was the same dir and we have a next request in + * sort order. No expired requests so continue on from here. + */ + drq = dd->next_drq[data_dir]; + } else { + /* + * The last req was the other direction or we have run out of + * higher-sectored requests. Go back to the lowest sectored + * request (1 way elevator) and start a new batch. + */ + dd->batching = 0; + drq = deadline_find_first_drq(dd, data_dir); + } + +dispatch_request: + /* + * drq is the selected appropriate request. + */ + dd->batching++; + deadline_move_request(dd, drq); + + return 1; +} + +static struct request *deadline_next_request(request_queue_t *q) +{ + struct deadline_data *dd = q->elevator->elevator_data; + struct request *rq; + + /* + * if there are still requests on the dispatch queue, grab the first one + */ + if (!list_empty(dd->dispatch)) { +dispatch: + rq = list_entry_rq(dd->dispatch->next); + return rq; + } + + if (deadline_dispatch_requests(dd)) + goto dispatch; + + return NULL; +} + +static void +deadline_insert_request(request_queue_t *q, struct request *rq, int where) +{ + struct deadline_data *dd = q->elevator->elevator_data; + + /* barriers must flush the reorder queue */ + if (unlikely(rq->flags & (REQ_SOFTBARRIER | REQ_HARDBARRIER) + && where == ELEVATOR_INSERT_SORT)) + where = ELEVATOR_INSERT_BACK; + + switch (where) { + case ELEVATOR_INSERT_BACK: + while (deadline_dispatch_requests(dd)) + ; + list_add_tail(&rq->queuelist, dd->dispatch); + break; + case ELEVATOR_INSERT_FRONT: + list_add(&rq->queuelist, dd->dispatch); + break; + case ELEVATOR_INSERT_SORT: + BUG_ON(!blk_fs_request(rq)); + deadline_add_request(q, rq); + break; + default: + printk("%s: bad insert point %d\n", __FUNCTION__,where); + return; + } +} + +static int deadline_queue_empty(request_queue_t *q) +{ + struct deadline_data *dd = q->elevator->elevator_data; + + if (!list_empty(&dd->fifo_list[WRITE]) + || !list_empty(&dd->fifo_list[READ]) + || !list_empty(dd->dispatch)) + return 0; + + return 1; +} + +static struct request * +deadline_former_request(request_queue_t *q, struct request *rq) +{ + struct deadline_rq *drq = RQ_DATA(rq); + struct rb_node *rbprev = rb_prev(&drq->rb_node); + + if (rbprev) + return rb_entry_drq(rbprev)->request; + + return NULL; +} + +static struct request * +deadline_latter_request(request_queue_t *q, struct request *rq) +{ + struct deadline_rq *drq = RQ_DATA(rq); + struct rb_node *rbnext = rb_next(&drq->rb_node); + + if (rbnext) + return rb_entry_drq(rbnext)->request; + + return NULL; +} + +static void deadline_exit_queue(elevator_t *e) +{ + struct deadline_data *dd = e->elevator_data; + + BUG_ON(!list_empty(&dd->fifo_list[READ])); + BUG_ON(!list_empty(&dd->fifo_list[WRITE])); + + mempool_destroy(dd->drq_pool); + kfree(dd->hash); + kfree(dd); +} + +/* + * initialize elevator private data (deadline_data), and alloc a drq for + * each request on the free lists + */ +static int deadline_init_queue(request_queue_t *q, elevator_t *e) +{ + struct deadline_data *dd; + int i; + + if (!drq_pool) + return -ENOMEM; + + dd = kmalloc(sizeof(*dd), GFP_KERNEL); + if (!dd) + return -ENOMEM; + memset(dd, 0, sizeof(*dd)); + + dd->hash = kmalloc(sizeof(struct list_head)*DL_HASH_ENTRIES,GFP_KERNEL); + if (!dd->hash) { + kfree(dd); + return -ENOMEM; + } + + dd->drq_pool = mempool_create(BLKDEV_MIN_RQ, mempool_alloc_slab, mempool_free_slab, drq_pool); + if (!dd->drq_pool) { + kfree(dd->hash); + kfree(dd); + return -ENOMEM; + } + + for (i = 0; i < DL_HASH_ENTRIES; i++) + INIT_LIST_HEAD(&dd->hash[i]); + + INIT_LIST_HEAD(&dd->fifo_list[READ]); + INIT_LIST_HEAD(&dd->fifo_list[WRITE]); + dd->sort_list[READ] = RB_ROOT; + dd->sort_list[WRITE] = RB_ROOT; + dd->dispatch = &q->queue_head; + dd->fifo_expire[READ] = read_expire; + dd->fifo_expire[WRITE] = write_expire; + dd->writes_starved = writes_starved; + dd->front_merges = 1; + dd->fifo_batch = fifo_batch; + e->elevator_data = dd; + return 0; +} + +static void deadline_put_request(request_queue_t *q, struct request *rq) +{ + struct deadline_data *dd = q->elevator->elevator_data; + struct deadline_rq *drq = RQ_DATA(rq); + + if (drq) { + mempool_free(drq, dd->drq_pool); + rq->elevator_private = NULL; + } +} + +static int +deadline_set_request(request_queue_t *q, struct request *rq, int gfp_mask) +{ + struct deadline_data *dd = q->elevator->elevator_data; + struct deadline_rq *drq; + + drq = mempool_alloc(dd->drq_pool, gfp_mask); + if (drq) { + memset(drq, 0, sizeof(*drq)); + RB_CLEAR(&drq->rb_node); + drq->request = rq; + + INIT_LIST_HEAD(&drq->hash); + drq->on_hash = 0; + + INIT_LIST_HEAD(&drq->fifo); + + rq->elevator_private = drq; + return 0; + } + + return 1; +} + +/* + * sysfs parts below + */ +struct deadline_fs_entry { + struct attribute attr; + ssize_t (*show)(struct deadline_data *, char *); + ssize_t (*store)(struct deadline_data *, const char *, size_t); +}; + +static ssize_t +deadline_var_show(int var, char *page) +{ + return sprintf(page, "%d\n", var); +} + +static ssize_t +deadline_var_store(int *var, const char *page, size_t count) +{ + char *p = (char *) page; + + *var = simple_strtol(p, &p, 10); + return count; +} + +#define SHOW_FUNCTION(__FUNC, __VAR, __CONV) \ +static ssize_t __FUNC(struct deadline_data *dd, char *page) \ +{ \ + int __data = __VAR; \ + if (__CONV) \ + __data = jiffies_to_msecs(__data); \ + return deadline_var_show(__data, (page)); \ +} +SHOW_FUNCTION(deadline_readexpire_show, dd->fifo_expire[READ], 1); +SHOW_FUNCTION(deadline_writeexpire_show, dd->fifo_expire[WRITE], 1); +SHOW_FUNCTION(deadline_writesstarved_show, dd->writes_starved, 0); +SHOW_FUNCTION(deadline_frontmerges_show, dd->front_merges, 0); +SHOW_FUNCTION(deadline_fifobatch_show, dd->fifo_batch, 0); +#undef SHOW_FUNCTION + +#define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX, __CONV) \ +static ssize_t __FUNC(struct deadline_data *dd, const char *page, size_t count) \ +{ \ + int __data; \ + int ret = deadline_var_store(&__data, (page), count); \ + if (__data < (MIN)) \ + __data = (MIN); \ + else if (__data > (MAX)) \ + __data = (MAX); \ + if (__CONV) \ + *(__PTR) = msecs_to_jiffies(__data); \ + else \ + *(__PTR) = __data; \ + return ret; \ +} +STORE_FUNCTION(deadline_readexpire_store, &dd->fifo_expire[READ], 0, INT_MAX, 1); +STORE_FUNCTION(deadline_writeexpire_store, &dd->fifo_expire[WRITE], 0, INT_MAX, 1); +STORE_FUNCTION(deadline_writesstarved_store, &dd->writes_starved, INT_MIN, INT_MAX, 0); +STORE_FUNCTION(deadline_frontmerges_store, &dd->front_merges, 0, 1, 0); +STORE_FUNCTION(deadline_fifobatch_store, &dd->fifo_batch, 0, INT_MAX, 0); +#undef STORE_FUNCTION + +static struct deadline_fs_entry deadline_readexpire_entry = { + .attr = {.name = "read_expire", .mode = S_IRUGO | S_IWUSR }, + .show = deadline_readexpire_show, + .store = deadline_readexpire_store, +}; +static struct deadline_fs_entry deadline_writeexpire_entry = { + .attr = {.name = "write_expire", .mode = S_IRUGO | S_IWUSR }, + .show = deadline_writeexpire_show, + .store = deadline_writeexpire_store, +}; +static struct deadline_fs_entry deadline_writesstarved_entry = { + .attr = {.name = "writes_starved", .mode = S_IRUGO | S_IWUSR }, + .show = deadline_writesstarved_show, + .store = deadline_writesstarved_store, +}; +static struct deadline_fs_entry deadline_frontmerges_entry = { + .attr = {.name = "front_merges", .mode = S_IRUGO | S_IWUSR }, + .show = deadline_frontmerges_show, + .store = deadline_frontmerges_store, +}; +static struct deadline_fs_entry deadline_fifobatch_entry = { + .attr = {.name = "fifo_batch", .mode = S_IRUGO | S_IWUSR }, + .show = deadline_fifobatch_show, + .store = deadline_fifobatch_store, +}; + +static struct attribute *default_attrs[] = { + &deadline_readexpire_entry.attr, + &deadline_writeexpire_entry.attr, + &deadline_writesstarved_entry.attr, + &deadline_frontmerges_entry.attr, + &deadline_fifobatch_entry.attr, + NULL, +}; + +#define to_deadline(atr) container_of((atr), struct deadline_fs_entry, attr) + +static ssize_t +deadline_attr_show(struct kobject *kobj, struct attribute *attr, char *page) +{ + elevator_t *e = container_of(kobj, elevator_t, kobj); + struct deadline_fs_entry *entry = to_deadline(attr); + + if (!entry->show) + return 0; + + return entry->show(e->elevator_data, page); +} + +static ssize_t +deadline_attr_store(struct kobject *kobj, struct attribute *attr, + const char *page, size_t length) +{ + elevator_t *e = container_of(kobj, elevator_t, kobj); + struct deadline_fs_entry *entry = to_deadline(attr); + + if (!entry->store) + return -EINVAL; + + return entry->store(e->elevator_data, page, length); +} + +static struct sysfs_ops deadline_sysfs_ops = { + .show = deadline_attr_show, + .store = deadline_attr_store, +}; + +static struct kobj_type deadline_ktype = { + .sysfs_ops = &deadline_sysfs_ops, + .default_attrs = default_attrs, +}; + +static struct elevator_type iosched_deadline = { + .ops = { + .elevator_merge_fn = deadline_merge, + .elevator_merged_fn = deadline_merged_request, + .elevator_merge_req_fn = deadline_merged_requests, + .elevator_next_req_fn = deadline_next_request, + .elevator_add_req_fn = deadline_insert_request, + .elevator_remove_req_fn = deadline_remove_request, + .elevator_queue_empty_fn = deadline_queue_empty, + .elevator_former_req_fn = deadline_former_request, + .elevator_latter_req_fn = deadline_latter_request, + .elevator_set_req_fn = deadline_set_request, + .elevator_put_req_fn = deadline_put_request, + .elevator_init_fn = deadline_init_queue, + .elevator_exit_fn = deadline_exit_queue, + }, + + .elevator_ktype = &deadline_ktype, + .elevator_name = "deadline", + .elevator_owner = THIS_MODULE, +}; + +static int __init deadline_init(void) +{ + int ret; + + drq_pool = kmem_cache_create("deadline_drq", sizeof(struct deadline_rq), + 0, 0, NULL, NULL); + + if (!drq_pool) + return -ENOMEM; + + ret = elv_register(&iosched_deadline); + if (ret) + kmem_cache_destroy(drq_pool); + + return ret; +} + +static void __exit deadline_exit(void) +{ + kmem_cache_destroy(drq_pool); + elv_unregister(&iosched_deadline); +} + +module_init(deadline_init); +module_exit(deadline_exit); + +MODULE_AUTHOR("Jens Axboe"); +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("deadline IO scheduler"); diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c new file mode 100644 index 00000000000..6b79b431462 --- /dev/null +++ b/drivers/block/elevator.c @@ -0,0 +1,705 @@ +/* + * linux/drivers/block/elevator.c + * + * Block device elevator/IO-scheduler. + * + * Copyright (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE + * + * 30042000 Jens Axboe <axboe@suse.de> : + * + * Split the elevator a bit so that it is possible to choose a different + * one or even write a new "plug in". There are three pieces: + * - elevator_fn, inserts a new request in the queue list + * - elevator_merge_fn, decides whether a new buffer can be merged with + * an existing request + * - elevator_dequeue_fn, called when a request is taken off the active list + * + * 20082000 Dave Jones <davej@suse.de> : + * Removed tests for max-bomb-segments, which was breaking elvtune + * when run without -bN + * + * Jens: + * - Rework again to work with bio instead of buffer_heads + * - loose bi_dev comparisons, partition handling is right now + * - completely modularize elevator setup and teardown + * + */ +#include <linux/kernel.h> +#include <linux/fs.h> +#include <linux/blkdev.h> +#include <linux/elevator.h> +#include <linux/bio.h> +#include <linux/config.h> +#include <linux/module.h> +#include <linux/slab.h> +#include <linux/init.h> +#include <linux/compiler.h> + +#include <asm/uaccess.h> + +static DEFINE_SPINLOCK(elv_list_lock); +static LIST_HEAD(elv_list); + +/* + * can we safely merge with this request? + */ +inline int elv_rq_merge_ok(struct request *rq, struct bio *bio) +{ + if (!rq_mergeable(rq)) + return 0; + + /* + * different data direction or already started, don't merge + */ + if (bio_data_dir(bio) != rq_data_dir(rq)) + return 0; + + /* + * same device and no special stuff set, merge is ok + */ + if (rq->rq_disk == bio->bi_bdev->bd_disk && + !rq->waiting && !rq->special) + return 1; + + return 0; +} +EXPORT_SYMBOL(elv_rq_merge_ok); + +inline int elv_try_merge(struct request *__rq, struct bio *bio) +{ + int ret = ELEVATOR_NO_MERGE; + + /* + * we can merge and sequence is ok, check if it's possible + */ + if (elv_rq_merge_ok(__rq, bio)) { + if (__rq->sector + __rq->nr_sectors == bio->bi_sector) + ret = ELEVATOR_BACK_MERGE; + else if (__rq->sector - bio_sectors(bio) == bio->bi_sector) + ret = ELEVATOR_FRONT_MERGE; + } + + return ret; +} +EXPORT_SYMBOL(elv_try_merge); + +inline int elv_try_last_merge(request_queue_t *q, struct bio *bio) +{ + if (q->last_merge) + return elv_try_merge(q->last_merge, bio); + + return ELEVATOR_NO_MERGE; +} +EXPORT_SYMBOL(elv_try_last_merge); + +static struct elevator_type *elevator_find(const char *name) +{ + struct elevator_type *e = NULL; + struct list_head *entry; + + spin_lock_irq(&elv_list_lock); + list_for_each(entry, &elv_list) { + struct elevator_type *__e; + + __e = list_entry(entry, struct elevator_type, list); + + if (!strcmp(__e->elevator_name, name)) { + e = __e; + break; + } + } + spin_unlock_irq(&elv_list_lock); + + return e; +} + +static void elevator_put(struct elevator_type *e) +{ + module_put(e->elevator_owner); +} + +static struct elevator_type *elevator_get(const char *name) +{ + struct elevator_type *e = elevator_find(name); + + if (!e) + return NULL; + if (!try_module_get(e->elevator_owner)) + return NULL; + + return e; +} + +static int elevator_attach(request_queue_t *q, struct elevator_type *e, + struct elevator_queue *eq) +{ + int ret = 0; + + memset(eq, 0, sizeof(*eq)); + eq->ops = &e->ops; + eq->elevator_type = e; + + INIT_LIST_HEAD(&q->queue_head); + q->last_merge = NULL; + q->elevator = eq; + + if (eq->ops->elevator_init_fn) + ret = eq->ops->elevator_init_fn(q, eq); + + return ret; +} + +static char chosen_elevator[16]; + +static void elevator_setup_default(void) +{ + /* + * check if default is set and exists + */ + if (chosen_elevator[0] && elevator_find(chosen_elevator)) + return; + +#if defined(CONFIG_IOSCHED_AS) + strcpy(chosen_elevator, "anticipatory"); +#elif defined(CONFIG_IOSCHED_DEADLINE) + strcpy(chosen_elevator, "deadline"); +#elif defined(CONFIG_IOSCHED_CFQ) + strcpy(chosen_elevator, "cfq"); +#elif defined(CONFIG_IOSCHED_NOOP) + strcpy(chosen_elevator, "noop"); +#else +#error "You must build at least 1 IO scheduler into the kernel" +#endif +} + +static int __init elevator_setup(char *str) +{ + strncpy(chosen_elevator, str, sizeof(chosen_elevator) - 1); + return 0; +} + +__setup("elevator=", elevator_setup); + +int elevator_init(request_queue_t *q, char *name) +{ + struct elevator_type *e = NULL; + struct elevator_queue *eq; + int ret = 0; + + elevator_setup_default(); + + if (!name) + name = chosen_elevator; + + e = elevator_get(name); + if (!e) + return -EINVAL; + + eq = kmalloc(sizeof(struct elevator_queue), GFP_KERNEL); + if (!eq) { + elevator_put(e->elevator_type); + return -ENOMEM; + } + + ret = elevator_attach(q, e, eq); + if (ret) { + kfree(eq); + elevator_put(e->elevator_type); + } + + return ret; +} + +void elevator_exit(elevator_t *e) +{ + if (e->ops->elevator_exit_fn) + e->ops->elevator_exit_fn(e); + + elevator_put(e->elevator_type); + e->elevator_type = NULL; + kfree(e); +} + +static int elevator_global_init(void) +{ + return 0; +} + +int elv_merge(request_queue_t *q, struct request **req, struct bio *bio) +{ + elevator_t *e = q->elevator; + + if (e->ops->elevator_merge_fn) + return e->ops->elevator_merge_fn(q, req, bio); + + return ELEVATOR_NO_MERGE; +} + +void elv_merged_request(request_queue_t *q, struct request *rq) +{ + elevator_t *e = q->elevator; + + if (e->ops->elevator_merged_fn) + e->ops->elevator_merged_fn(q, rq); +} + +void elv_merge_requests(request_queue_t *q, struct request *rq, + struct request *next) +{ + elevator_t *e = q->elevator; + + if (q->last_merge == next) + q->last_merge = NULL; + + if (e->ops->elevator_merge_req_fn) + e->ops->elevator_merge_req_fn(q, rq, next); +} + +/* + * For careful internal use by the block layer. Essentially the same as + * a requeue in that it tells the io scheduler that this request is not + * active in the driver or hardware anymore, but we don't want the request + * added back to the scheduler. Function is not exported. + */ +void elv_deactivate_request(request_queue_t *q, struct request *rq) +{ + elevator_t *e = q->elevator; + + /* + * it already went through dequeue, we need to decrement the + * in_flight count again + */ + if (blk_account_rq(rq)) + q->in_flight--; + + rq->flags &= ~REQ_STARTED; + + if (e->ops->elevator_deactivate_req_fn) + e->ops->elevator_deactivate_req_fn(q, rq); +} + +void elv_requeue_request(request_queue_t *q, struct request *rq) +{ + elv_deactivate_request(q, rq); + + /* + * if this is the flush, requeue the original instead and drop the flush + */ + if (rq->flags & REQ_BAR_FLUSH) { + clear_bit(QUEUE_FLAG_FLUSH, &q->queue_flags); + rq = rq->end_io_data; + } + + /* + * if iosched has an explicit requeue hook, then use that. otherwise + * just put the request at the front of the queue + */ + if (q->elevator->ops->elevator_requeue_req_fn) + q->elevator->ops->elevator_requeue_req_fn(q, rq); + else + __elv_add_request(q, rq, ELEVATOR_INSERT_FRONT, 0); +} + +void __elv_add_request(request_queue_t *q, struct request *rq, int where, + int plug) +{ + /* + * barriers implicitly indicate back insertion + */ + if (rq->flags & (REQ_SOFTBARRIER | REQ_HARDBARRIER) && + where == ELEVATOR_INSERT_SORT) + where = ELEVATOR_INSERT_BACK; + + if (plug) + blk_plug_device(q); + + rq->q = q; + + if (!test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags)) { + q->elevator->ops->elevator_add_req_fn(q, rq, where); + + if (blk_queue_plugged(q)) { + int nrq = q->rq.count[READ] + q->rq.count[WRITE] + - q->in_flight; + + if (nrq == q->unplug_thresh) + __generic_unplug_device(q); + } + } else + /* + * if drain is set, store the request "locally". when the drain + * is finished, the requests will be handed ordered to the io + * scheduler + */ + list_add_tail(&rq->queuelist, &q->drain_list); +} + +void elv_add_request(request_queue_t *q, struct request *rq, int where, + int plug) +{ + unsigned long flags; + + spin_lock_irqsave(q->queue_lock, flags); + __elv_add_request(q, rq, where, plug); + spin_unlock_irqrestore(q->queue_lock, flags); +} + +static inline struct request *__elv_next_request(request_queue_t *q) +{ + struct request *rq = q->elevator->ops->elevator_next_req_fn(q); + + /* + * if this is a barrier write and the device has to issue a + * flush sequence to support it, check how far we are + */ + if (rq && blk_fs_request(rq) && blk_barrier_rq(rq)) { + BUG_ON(q->ordered == QUEUE_ORDERED_NONE); + + if (q->ordered == QUEUE_ORDERED_FLUSH && + !blk_barrier_preflush(rq)) + rq = blk_start_pre_flush(q, rq); + } + + return rq; +} + +struct request *elv_next_request(request_queue_t *q) +{ + struct request *rq; + int ret; + + while ((rq = __elv_next_request(q)) != NULL) { + /* + * just mark as started even if we don't start it, a request + * that has been delayed should not be passed by new incoming + * requests + */ + rq->flags |= REQ_STARTED; + + if (rq == q->last_merge) + q->last_merge = NULL; + + if ((rq->flags & REQ_DONTPREP) || !q->prep_rq_fn) + break; + + ret = q->prep_rq_fn(q, rq); + if (ret == BLKPREP_OK) { + break; + } else if (ret == BLKPREP_DEFER) { + rq = NULL; + break; + } else if (ret == BLKPREP_KILL) { + int nr_bytes = rq->hard_nr_sectors << 9; + + if (!nr_bytes) + nr_bytes = rq->data_len; + + blkdev_dequeue_request(rq); + rq->flags |= REQ_QUIET; + end_that_request_chunk(rq, 0, nr_bytes); + end_that_request_last(rq); + } else { + printk(KERN_ERR "%s: bad return=%d\n", __FUNCTION__, + ret); + break; + } + } + + return rq; +} + +void elv_remove_request(request_queue_t *q, struct request *rq) +{ + elevator_t *e = q->elevator; + + /* + * the time frame between a request being removed from the lists + * and to it is freed is accounted as io that is in progress at + * the driver side. note that we only account requests that the + * driver has seen (REQ_STARTED set), to avoid false accounting + * for request-request merges + */ + if (blk_account_rq(rq)) + q->in_flight++; + + /* + * the main clearing point for q->last_merge is on retrieval of + * request by driver (it calls elv_next_request()), but it _can_ + * also happen here if a request is added to the queue but later + * deleted without ever being given to driver (merged with another + * request). + */ + if (rq == q->last_merge) + q->last_merge = NULL; + + if (e->ops->elevator_remove_req_fn) + e->ops->elevator_remove_req_fn(q, rq); +} + +int elv_queue_empty(request_queue_t *q) +{ + elevator_t *e = q->elevator; + + if (e->ops->elevator_queue_empty_fn) + return e->ops->elevator_queue_empty_fn(q); + + return list_empty(&q->queue_head); +} + +struct request *elv_latter_request(request_queue_t *q, struct request *rq) +{ + struct list_head *next; + + elevator_t *e = q->elevator; + + if (e->ops->elevator_latter_req_fn) + return e->ops->elevator_latter_req_fn(q, rq); + + next = rq->queuelist.next; + if (next != &q->queue_head && next != &rq->queuelist) + return list_entry_rq(next); + + return NULL; +} + +struct request *elv_former_request(request_queue_t *q, struct request *rq) +{ + struct list_head *prev; + + elevator_t *e = q->elevator; + + if (e->ops->elevator_former_req_fn) + return e->ops->elevator_former_req_fn(q, rq); + + prev = rq->queuelist.prev; + if (prev != &q->queue_head && prev != &rq->queuelist) + return list_entry_rq(prev); + + return NULL; +} + +int elv_set_request(request_queue_t *q, struct request *rq, int gfp_mask) +{ + elevator_t *e = q->elevator; + + if (e->ops->elevator_set_req_fn) + return e->ops->elevator_set_req_fn(q, rq, gfp_mask); + + rq->elevator_private = NULL; + return 0; +} + +void elv_put_request(request_queue_t *q, struct request *rq) +{ + elevator_t *e = q->elevator; + + if (e->ops->elevator_put_req_fn) + e->ops->elevator_put_req_fn(q, rq); +} + +int elv_may_queue(request_queue_t *q, int rw) +{ + elevator_t *e = q->elevator; + + if (e->ops->elevator_may_queue_fn) + return e->ops->elevator_may_queue_fn(q, rw); + + return ELV_MQUEUE_MAY; +} + +void elv_completed_request(request_queue_t *q, struct request *rq) +{ + elevator_t *e = q->elevator; + + /* + * request is released from the driver, io must be done + */ + if (blk_account_rq(rq)) + q->in_flight--; + + if (e->ops->elevator_completed_req_fn) + e->ops->elevator_completed_req_fn(q, rq); +} + +int elv_register_queue(struct request_queue *q) +{ + elevator_t *e = q->elevator; + + e->kobj.parent = kobject_get(&q->kobj); + if (!e->kobj.parent) + return -EBUSY; + + snprintf(e->kobj.name, KOBJ_NAME_LEN, "%s", "iosched"); + e->kobj.ktype = e->elevator_type->elevator_ktype; + + return kobject_register(&e->kobj); +} + +void elv_unregister_queue(struct request_queue *q) +{ + if (q) { + elevator_t *e = q->elevator; + kobject_unregister(&e->kobj); + kobject_put(&q->kobj); + } +} + +int elv_register(struct elevator_type *e) +{ + if (elevator_find(e->elevator_name)) + BUG(); + + spin_lock_irq(&elv_list_lock); + list_add_tail(&e->list, &elv_list); + spin_unlock_irq(&elv_list_lock); + + printk(KERN_INFO "io scheduler %s registered", e->elevator_name); + if (!strcmp(e->elevator_name, chosen_elevator)) + printk(" (default)"); + printk("\n"); + return 0; +} +EXPORT_SYMBOL_GPL(elv_register); + +void elv_unregister(struct elevator_type *e) +{ + spin_lock_irq(&elv_list_lock); + list_del_init(&e->list); + spin_unlock_irq(&elv_list_lock); +} +EXPORT_SYMBOL_GPL(elv_unregister); + +/* + * switch to new_e io scheduler. be careful not to introduce deadlocks - + * we don't free the old io scheduler, before we have allocated what we + * need for the new one. this way we have a chance of going back to the old + * one, if the new one fails init for some reason. we also do an intermediate + * switch to noop to ensure safety with stack-allocated requests, since they + * don't originate from the block layer allocator. noop is safe here, because + * it never needs to touch the elevator itself for completion events. DRAIN + * flags will make sure we don't touch it for additions either. + */ +static void elevator_switch(request_queue_t *q, struct elevator_type *new_e) +{ + elevator_t *e = kmalloc(sizeof(elevator_t), GFP_KERNEL); + struct elevator_type *noop_elevator = NULL; + elevator_t *old_elevator; + + if (!e) + goto error; + + /* + * first step, drain requests from the block freelist + */ + blk_wait_queue_drained(q, 0); + + /* + * unregister old elevator data + */ + elv_unregister_queue(q); + old_elevator = q->elevator; + + /* + * next step, switch to noop since it uses no private rq structures + * and doesn't allocate any memory for anything. then wait for any + * non-fs requests in-flight + */ + noop_elevator = elevator_get("noop"); + spin_lock_irq(q->queue_lock); + elevator_attach(q, noop_elevator, e); + spin_unlock_irq(q->queue_lock); + + blk_wait_queue_drained(q, 1); + + /* + * attach and start new elevator + */ + if (elevator_attach(q, new_e, e)) + goto fail; + + if (elv_register_queue(q)) + goto fail_register; + + /* + * finally exit old elevator and start queue again + */ + elevator_exit(old_elevator); + blk_finish_queue_drain(q); + elevator_put(noop_elevator); + return; + +fail_register: + /* + * switch failed, exit the new io scheduler and reattach the old + * one again (along with re-adding the sysfs dir) + */ + elevator_exit(e); +fail: + q->elevator = old_elevator; + elv_register_queue(q); + blk_finish_queue_drain(q); +error: + if (noop_elevator) + elevator_put(noop_elevator); + elevator_put(new_e); + printk(KERN_ERR "elevator: switch to %s failed\n",new_e->elevator_name); +} + +ssize_t elv_iosched_store(request_queue_t *q, const char *name, size_t count) +{ + char elevator_name[ELV_NAME_MAX]; + struct elevator_type *e; + + memset(elevator_name, 0, sizeof(elevator_name)); + strncpy(elevator_name, name, sizeof(elevator_name)); + + if (elevator_name[strlen(elevator_name) - 1] == '\n') + elevator_name[strlen(elevator_name) - 1] = '\0'; + + e = elevator_get(elevator_name); + if (!e) { + printk(KERN_ERR "elevator: type %s not found\n", elevator_name); + return -EINVAL; + } + + if (!strcmp(elevator_name, q->elevator->elevator_type->elevator_name)) + return count; + + elevator_switch(q, e); + return count; +} + +ssize_t elv_iosched_show(request_queue_t *q, char *name) +{ + elevator_t *e = q->elevator; + struct elevator_type *elv = e->elevator_type; + struct list_head *entry; + int len = 0; + + spin_lock_irq(q->queue_lock); + list_for_each(entry, &elv_list) { + struct elevator_type *__e; + + __e = list_entry(entry, struct elevator_type, list); + if (!strcmp(elv->elevator_name, __e->elevator_name)) + len += sprintf(name+len, "[%s] ", elv->elevator_name); + else + len += sprintf(name+len, "%s ", __e->elevator_name); + } + spin_unlock_irq(q->queue_lock); + + len += sprintf(len+name, "\n"); + return len; +} + +module_init(elevator_global_init); + +EXPORT_SYMBOL(elv_add_request); +EXPORT_SYMBOL(__elv_add_request); +EXPORT_SYMBOL(elv_requeue_request); +EXPORT_SYMBOL(elv_next_request); +EXPORT_SYMBOL(elv_remove_request); +EXPORT_SYMBOL(elv_queue_empty); +EXPORT_SYMBOL(elv_completed_request); +EXPORT_SYMBOL(elevator_exit); +EXPORT_SYMBOL(elevator_init); diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c new file mode 100644 index 00000000000..42dfa281a88 --- /dev/null +++ b/drivers/block/floppy.c @@ -0,0 +1,4638 @@ +/* + * linux/drivers/block/floppy.c + * + * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 1993, 1994 Alain Knaff + * Copyright (C) 1998 Alan Cox + */ +/* + * 02.12.91 - Changed to static variables to indicate need for reset + * and recalibrate. This makes some things easier (output_byte reset + * checking etc), and means less interrupt jumping in case of errors, + * so the code is hopefully easier to understand. + */ + +/* + * This file is certainly a mess. I've tried my best to get it working, + * but I don't like programming floppies, and I have only one anyway. + * Urgel. I should check for more errors, and do more graceful error + * recovery. Seems there are problems with several drives. I've tried to + * correct them. No promises. + */ + +/* + * As with hd.c, all routines within this file can (and will) be called + * by interrupts, so extreme caution is needed. A hardware interrupt + * handler may not sleep, or a kernel panic will happen. Thus I cannot + * call "floppy-on" directly, but have to set a special timer interrupt + * etc. + */ + +/* + * 28.02.92 - made track-buffering routines, based on the routines written + * by entropy@wintermute.wpi.edu (Lawrence Foard). Linus. + */ + +/* + * Automatic floppy-detection and formatting written by Werner Almesberger + * (almesber@nessie.cs.id.ethz.ch), who also corrected some problems with + * the floppy-change signal detection. + */ + +/* + * 1992/7/22 -- Hennus Bergman: Added better error reporting, fixed + * FDC data overrun bug, added some preliminary stuff for vertical + * recording support. + * + * 1992/9/17: Added DMA allocation & DMA functions. -- hhb. + * + * TODO: Errors are still not counted properly. + */ + +/* 1992/9/20 + * Modifications for ``Sector Shifting'' by Rob Hooft (hooft@chem.ruu.nl) + * modeled after the freeware MS-DOS program fdformat/88 V1.8 by + * Christoph H. Hochst\"atter. + * I have fixed the shift values to the ones I always use. Maybe a new + * ioctl() should be created to be able to modify them. + * There is a bug in the driver that makes it impossible to format a + * floppy as the first thing after bootup. + */ + +/* + * 1993/4/29 -- Linus -- cleaned up the timer handling in the kernel, and + * this helped the floppy driver as well. Much cleaner, and still seems to + * work. + */ + +/* 1994/6/24 --bbroad-- added the floppy table entries and made + * minor modifications to allow 2.88 floppies to be run. + */ + +/* 1994/7/13 -- Paul Vojta -- modified the probing code to allow three or more + * disk types. + */ + +/* + * 1994/8/8 -- Alain Knaff -- Switched to fdpatch driver: Support for bigger + * format bug fixes, but unfortunately some new bugs too... + */ + +/* 1994/9/17 -- Koen Holtman -- added logging of physical floppy write + * errors to allow safe writing by specialized programs. + */ + +/* 1995/4/24 -- Dan Fandrich -- added support for Commodore 1581 3.5" disks + * by defining bit 1 of the "stretch" parameter to mean put sectors on the + * opposite side of the disk, leaving the sector IDs alone (i.e. Commodore's + * drives are "upside-down"). + */ + +/* + * 1995/8/26 -- Andreas Busse -- added Mips support. + */ + +/* + * 1995/10/18 -- Ralf Baechle -- Portability cleanup; move machine dependent + * features to asm/floppy.h. + */ + +/* + * 1998/05/07 -- Russell King -- More portability cleanups; moved definition of + * interrupt and dma channel to asm/floppy.h. Cleaned up some formatting & + * use of '0' for NULL. + */ + +/* + * 1998/06/07 -- Alan Cox -- Merged the 2.0.34 fixes for resource allocation + * failures. + */ + +/* + * 1998/09/20 -- David Weinehall -- Added slow-down code for buggy PS/2-drives. + */ + +/* + * 1999/08/13 -- Paul Slootman -- floppy stopped working on Alpha after 24 + * days, 6 hours, 32 minutes and 32 seconds (i.e. MAXINT jiffies; ints were + * being used to store jiffies, which are unsigned longs). + */ + +/* + * 2000/08/28 -- Arnaldo Carvalho de Melo <acme@conectiva.com.br> + * - get rid of check_region + * - s/suser/capable/ + */ + +/* + * 2001/08/26 -- Paul Gortmaker - fix insmod oops on machines with no + * floppy controller (lingering task on list after module is gone... boom.) + */ + +/* + * 2002/02/07 -- Anton Altaparmakov - Fix io ports reservation to correct range + * (0x3f2-0x3f5, 0x3f7). This fix is a bit of a hack but the proper fix + * requires many non-obvious changes in arch dependent code. + */ + +/* 2003/07/28 -- Daniele Bellucci <bellucda@tiscali.it>. + * Better audit of register_blkdev. + */ + +#define FLOPPY_SANITY_CHECK +#undef FLOPPY_SILENT_DCL_CLEAR + +#define REALLY_SLOW_IO + +#define DEBUGT 2 +#define DCL_DEBUG /* debug disk change line */ + +/* do print messages for unexpected interrupts */ +static int print_unex = 1; +#include <linux/module.h> +#include <linux/sched.h> +#include <linux/fs.h> +#include <linux/kernel.h> +#include <linux/timer.h> +#include <linux/workqueue.h> +#define FDPATCHES +#include <linux/fdreg.h> + +/* + * 1998/1/21 -- Richard Gooch <rgooch@atnf.csiro.au> -- devfs support + */ + +#include <linux/fd.h> +#include <linux/hdreg.h> + +#include <linux/errno.h> +#include <linux/slab.h> +#include <linux/mm.h> +#include <linux/bio.h> +#include <linux/string.h> +#include <linux/fcntl.h> +#include <linux/delay.h> +#include <linux/mc146818rtc.h> /* CMOS defines */ +#include <linux/ioport.h> +#include <linux/interrupt.h> +#include <linux/init.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/device.h> +#include <linux/buffer_head.h> /* for invalidate_buffers() */ + +/* + * PS/2 floppies have much slower step rates than regular floppies. + * It's been recommended that take about 1/4 of the default speed + * in some more extreme cases. + */ +static int slow_floppy; + +#include <asm/dma.h> +#include <asm/irq.h> +#include <asm/system.h> +#include <asm/io.h> +#include <asm/uaccess.h> + +static int FLOPPY_IRQ = 6; +static int FLOPPY_DMA = 2; +static int can_use_virtual_dma = 2; +/* ======= + * can use virtual DMA: + * 0 = use of virtual DMA disallowed by config + * 1 = use of virtual DMA prescribed by config + * 2 = no virtual DMA preference configured. By default try hard DMA, + * but fall back on virtual DMA when not enough memory available + */ + +static int use_virtual_dma; +/* ======= + * use virtual DMA + * 0 using hard DMA + * 1 using virtual DMA + * This variable is set to virtual when a DMA mem problem arises, and + * reset back in floppy_grab_irq_and_dma. + * It is not safe to reset it in other circumstances, because the floppy + * driver may have several buffers in use at once, and we do currently not + * record each buffers capabilities + */ + +static DEFINE_SPINLOCK(floppy_lock); +static struct completion device_release; + +static unsigned short virtual_dma_port = 0x3f0; +irqreturn_t floppy_interrupt(int irq, void *dev_id, struct pt_regs *regs); +static int set_dor(int fdc, char mask, char data); +static void register_devfs_entries(int drive) __init; + +#define K_64 0x10000 /* 64KB */ + +/* the following is the mask of allowed drives. By default units 2 and + * 3 of both floppy controllers are disabled, because switching on the + * motor of these drives causes system hangs on some PCI computers. drive + * 0 is the low bit (0x1), and drive 7 is the high bit (0x80). Bits are on if + * a drive is allowed. + * + * NOTE: This must come before we include the arch floppy header because + * some ports reference this variable from there. -DaveM + */ + +static int allowed_drive_mask = 0x33; + +#include <asm/floppy.h> + +static int irqdma_allocated; + +#define LOCAL_END_REQUEST +#define DEVICE_NAME "floppy" + +#include <linux/blkdev.h> +#include <linux/blkpg.h> +#include <linux/cdrom.h> /* for the compatibility eject ioctl */ +#include <linux/completion.h> + +static struct request *current_req; +static struct request_queue *floppy_queue; +static void do_fd_request(request_queue_t * q); + +#ifndef fd_get_dma_residue +#define fd_get_dma_residue() get_dma_residue(FLOPPY_DMA) +#endif + +/* Dma Memory related stuff */ + +#ifndef fd_dma_mem_free +#define fd_dma_mem_free(addr, size) free_pages(addr, get_order(size)) +#endif + +#ifndef fd_dma_mem_alloc +#define fd_dma_mem_alloc(size) __get_dma_pages(GFP_KERNEL,get_order(size)) +#endif + +static inline void fallback_on_nodma_alloc(char **addr, size_t l) +{ +#ifdef FLOPPY_CAN_FALLBACK_ON_NODMA + if (*addr) + return; /* we have the memory */ + if (can_use_virtual_dma != 2) + return; /* no fallback allowed */ + printk + ("DMA memory shortage. Temporarily falling back on virtual DMA\n"); + *addr = (char *)nodma_mem_alloc(l); +#else + return; +#endif +} + +/* End dma memory related stuff */ + +static unsigned long fake_change; +static int initialising = 1; + +#define ITYPE(x) (((x)>>2) & 0x1f) +#define TOMINOR(x) ((x & 3) | ((x & 4) << 5)) +#define UNIT(x) ((x) & 0x03) /* drive on fdc */ +#define FDC(x) (((x) & 0x04) >> 2) /* fdc of drive */ +#define REVDRIVE(fdc, unit) ((unit) + ((fdc) << 2)) + /* reverse mapping from unit and fdc to drive */ +#define DP (&drive_params[current_drive]) +#define DRS (&drive_state[current_drive]) +#define DRWE (&write_errors[current_drive]) +#define FDCS (&fdc_state[fdc]) +#define CLEARF(x) (clear_bit(x##_BIT, &DRS->flags)) +#define SETF(x) (set_bit(x##_BIT, &DRS->flags)) +#define TESTF(x) (test_bit(x##_BIT, &DRS->flags)) + +#define UDP (&drive_params[drive]) +#define UDRS (&drive_state[drive]) +#define UDRWE (&write_errors[drive]) +#define UFDCS (&fdc_state[FDC(drive)]) +#define UCLEARF(x) (clear_bit(x##_BIT, &UDRS->flags)) +#define USETF(x) (set_bit(x##_BIT, &UDRS->flags)) +#define UTESTF(x) (test_bit(x##_BIT, &UDRS->flags)) + +#define DPRINT(format, args...) printk(DEVICE_NAME "%d: " format, current_drive , ## args) + +#define PH_HEAD(floppy,head) (((((floppy)->stretch & 2) >>1) ^ head) << 2) +#define STRETCH(floppy) ((floppy)->stretch & FD_STRETCH) + +#define CLEARSTRUCT(x) memset((x), 0, sizeof(*(x))) + +/* read/write */ +#define COMMAND raw_cmd->cmd[0] +#define DR_SELECT raw_cmd->cmd[1] +#define TRACK raw_cmd->cmd[2] +#define HEAD raw_cmd->cmd[3] +#define SECTOR raw_cmd->cmd[4] +#define SIZECODE raw_cmd->cmd[5] +#define SECT_PER_TRACK raw_cmd->cmd[6] +#define GAP raw_cmd->cmd[7] +#define SIZECODE2 raw_cmd->cmd[8] +#define NR_RW 9 + +/* format */ +#define F_SIZECODE raw_cmd->cmd[2] +#define F_SECT_PER_TRACK raw_cmd->cmd[3] +#define F_GAP raw_cmd->cmd[4] +#define F_FILL raw_cmd->cmd[5] +#define NR_F 6 + +/* + * Maximum disk size (in kilobytes). This default is used whenever the + * current disk size is unknown. + * [Now it is rather a minimum] + */ +#define MAX_DISK_SIZE 4 /* 3984 */ + +/* + * globals used by 'result()' + */ +#define MAX_REPLIES 16 +static unsigned char reply_buffer[MAX_REPLIES]; +static int inr; /* size of reply buffer, when called from interrupt */ +#define ST0 (reply_buffer[0]) +#define ST1 (reply_buffer[1]) +#define ST2 (reply_buffer[2]) +#define ST3 (reply_buffer[0]) /* result of GETSTATUS */ +#define R_TRACK (reply_buffer[3]) +#define R_HEAD (reply_buffer[4]) +#define R_SECTOR (reply_buffer[5]) +#define R_SIZECODE (reply_buffer[6]) + +#define SEL_DLY (2*HZ/100) + +/* + * this struct defines the different floppy drive types. + */ +static struct { + struct floppy_drive_params params; + const char *name; /* name printed while booting */ +} default_drive_params[] = { +/* NOTE: the time values in jiffies should be in msec! + CMOS drive type + | Maximum data rate supported by drive type + | | Head load time, msec + | | | Head unload time, msec (not used) + | | | | Step rate interval, usec + | | | | | Time needed for spinup time (jiffies) + | | | | | | Timeout for spinning down (jiffies) + | | | | | | | Spindown offset (where disk stops) + | | | | | | | | Select delay + | | | | | | | | | RPS + | | | | | | | | | | Max number of tracks + | | | | | | | | | | | Interrupt timeout + | | | | | | | | | | | | Max nonintlv. sectors + | | | | | | | | | | | | | -Max Errors- flags */ +{{0, 500, 16, 16, 8000, 1*HZ, 3*HZ, 0, SEL_DLY, 5, 80, 3*HZ, 20, {3,1,2,0,2}, 0, + 0, { 7, 4, 8, 2, 1, 5, 3,10}, 3*HZ/2, 0 }, "unknown" }, + +{{1, 300, 16, 16, 8000, 1*HZ, 3*HZ, 0, SEL_DLY, 5, 40, 3*HZ, 17, {3,1,2,0,2}, 0, + 0, { 1, 0, 0, 0, 0, 0, 0, 0}, 3*HZ/2, 1 }, "360K PC" }, /*5 1/4 360 KB PC*/ + +{{2, 500, 16, 16, 6000, 4*HZ/10, 3*HZ, 14, SEL_DLY, 6, 83, 3*HZ, 17, {3,1,2,0,2}, 0, + 0, { 2, 5, 6,23,10,20,12, 0}, 3*HZ/2, 2 }, "1.2M" }, /*5 1/4 HD AT*/ + +{{3, 250, 16, 16, 3000, 1*HZ, 3*HZ, 0, SEL_DLY, 5, 83, 3*HZ, 20, {3,1,2,0,2}, 0, + 0, { 4,22,21,30, 3, 0, 0, 0}, 3*HZ/2, 4 }, "720k" }, /*3 1/2 DD*/ + +{{4, 500, 16, 16, 4000, 4*HZ/10, 3*HZ, 10, SEL_DLY, 5, 83, 3*HZ, 20, {3,1,2,0,2}, 0, + 0, { 7, 4,25,22,31,21,29,11}, 3*HZ/2, 7 }, "1.44M" }, /*3 1/2 HD*/ + +{{5, 1000, 15, 8, 3000, 4*HZ/10, 3*HZ, 10, SEL_DLY, 5, 83, 3*HZ, 40, {3,1,2,0,2}, 0, + 0, { 7, 8, 4,25,28,22,31,21}, 3*HZ/2, 8 }, "2.88M AMI BIOS" }, /*3 1/2 ED*/ + +{{6, 1000, 15, 8, 3000, 4*HZ/10, 3*HZ, 10, SEL_DLY, 5, 83, 3*HZ, 40, {3,1,2,0,2}, 0, + 0, { 7, 8, 4,25,28,22,31,21}, 3*HZ/2, 8 }, "2.88M" } /*3 1/2 ED*/ +/* | --autodetected formats--- | | | + * read_track | | Name printed when booting + * | Native format + * Frequency of disk change checks */ +}; + +static struct floppy_drive_params drive_params[N_DRIVE]; +static struct floppy_drive_struct drive_state[N_DRIVE]; +static struct floppy_write_errors write_errors[N_DRIVE]; +static struct timer_list motor_off_timer[N_DRIVE]; +static struct gendisk *disks[N_DRIVE]; +static struct block_device *opened_bdev[N_DRIVE]; +static DECLARE_MUTEX(open_lock); +static struct floppy_raw_cmd *raw_cmd, default_raw_cmd; + +/* + * This struct defines the different floppy types. + * + * Bit 0 of 'stretch' tells if the tracks need to be doubled for some + * types (e.g. 360kB diskette in 1.2MB drive, etc.). Bit 1 of 'stretch' + * tells if the disk is in Commodore 1581 format, which means side 0 sectors + * are located on side 1 of the disk but with a side 0 ID, and vice-versa. + * This is the same as the Sharp MZ-80 5.25" CP/M disk format, except that the + * 1581's logical side 0 is on physical side 1, whereas the Sharp's logical + * side 0 is on physical side 0 (but with the misnamed sector IDs). + * 'stretch' should probably be renamed to something more general, like + * 'options'. Other parameters should be self-explanatory (see also + * setfdprm(8)). + */ +/* + Size + | Sectors per track + | | Head + | | | Tracks + | | | | Stretch + | | | | | Gap 1 size + | | | | | | Data rate, | 0x40 for perp + | | | | | | | Spec1 (stepping rate, head unload + | | | | | | | | /fmt gap (gap2) */ +static struct floppy_struct floppy_type[32] = { + { 0, 0,0, 0,0,0x00,0x00,0x00,0x00,NULL }, /* 0 no testing */ + { 720, 9,2,40,0,0x2A,0x02,0xDF,0x50,"d360" }, /* 1 360KB PC */ + { 2400,15,2,80,0,0x1B,0x00,0xDF,0x54,"h1200" }, /* 2 1.2MB AT */ + { 720, 9,1,80,0,0x2A,0x02,0xDF,0x50,"D360" }, /* 3 360KB SS 3.5" */ + { 1440, 9,2,80,0,0x2A,0x02,0xDF,0x50,"D720" }, /* 4 720KB 3.5" */ + { 720, 9,2,40,1,0x23,0x01,0xDF,0x50,"h360" }, /* 5 360KB AT */ + { 1440, 9,2,80,0,0x23,0x01,0xDF,0x50,"h720" }, /* 6 720KB AT */ + { 2880,18,2,80,0,0x1B,0x00,0xCF,0x6C,"H1440" }, /* 7 1.44MB 3.5" */ + { 5760,36,2,80,0,0x1B,0x43,0xAF,0x54,"E2880" }, /* 8 2.88MB 3.5" */ + { 6240,39,2,80,0,0x1B,0x43,0xAF,0x28,"E3120" }, /* 9 3.12MB 3.5" */ + + { 2880,18,2,80,0,0x25,0x00,0xDF,0x02,"h1440" }, /* 10 1.44MB 5.25" */ + { 3360,21,2,80,0,0x1C,0x00,0xCF,0x0C,"H1680" }, /* 11 1.68MB 3.5" */ + { 820,10,2,41,1,0x25,0x01,0xDF,0x2E,"h410" }, /* 12 410KB 5.25" */ + { 1640,10,2,82,0,0x25,0x02,0xDF,0x2E,"H820" }, /* 13 820KB 3.5" */ + { 2952,18,2,82,0,0x25,0x00,0xDF,0x02,"h1476" }, /* 14 1.48MB 5.25" */ + { 3444,21,2,82,0,0x25,0x00,0xDF,0x0C,"H1722" }, /* 15 1.72MB 3.5" */ + { 840,10,2,42,1,0x25,0x01,0xDF,0x2E,"h420" }, /* 16 420KB 5.25" */ + { 1660,10,2,83,0,0x25,0x02,0xDF,0x2E,"H830" }, /* 17 830KB 3.5" */ + { 2988,18,2,83,0,0x25,0x00,0xDF,0x02,"h1494" }, /* 18 1.49MB 5.25" */ + { 3486,21,2,83,0,0x25,0x00,0xDF,0x0C,"H1743" }, /* 19 1.74 MB 3.5" */ + + { 1760,11,2,80,0,0x1C,0x09,0xCF,0x00,"h880" }, /* 20 880KB 5.25" */ + { 2080,13,2,80,0,0x1C,0x01,0xCF,0x00,"D1040" }, /* 21 1.04MB 3.5" */ + { 2240,14,2,80,0,0x1C,0x19,0xCF,0x00,"D1120" }, /* 22 1.12MB 3.5" */ + { 3200,20,2,80,0,0x1C,0x20,0xCF,0x2C,"h1600" }, /* 23 1.6MB 5.25" */ + { 3520,22,2,80,0,0x1C,0x08,0xCF,0x2e,"H1760" }, /* 24 1.76MB 3.5" */ + { 3840,24,2,80,0,0x1C,0x20,0xCF,0x00,"H1920" }, /* 25 1.92MB 3.5" */ + { 6400,40,2,80,0,0x25,0x5B,0xCF,0x00,"E3200" }, /* 26 3.20MB 3.5" */ + { 7040,44,2,80,0,0x25,0x5B,0xCF,0x00,"E3520" }, /* 27 3.52MB 3.5" */ + { 7680,48,2,80,0,0x25,0x63,0xCF,0x00,"E3840" }, /* 28 3.84MB 3.5" */ + + { 3680,23,2,80,0,0x1C,0x10,0xCF,0x00,"H1840" }, /* 29 1.84MB 3.5" */ + { 1600,10,2,80,0,0x25,0x02,0xDF,0x2E,"D800" }, /* 30 800KB 3.5" */ + { 3200,20,2,80,0,0x1C,0x00,0xCF,0x2C,"H1600" }, /* 31 1.6MB 3.5" */ +}; + +#define NUMBER(x) (sizeof(x) / sizeof(*(x))) +#define SECTSIZE (_FD_SECTSIZE(*floppy)) + +/* Auto-detection: Disk type used until the next media change occurs. */ +static struct floppy_struct *current_type[N_DRIVE]; + +/* + * User-provided type information. current_type points to + * the respective entry of this array. + */ +static struct floppy_struct user_params[N_DRIVE]; + +static sector_t floppy_sizes[256]; + +/* + * The driver is trying to determine the correct media format + * while probing is set. rw_interrupt() clears it after a + * successful access. + */ +static int probing; + +/* Synchronization of FDC access. */ +#define FD_COMMAND_NONE -1 +#define FD_COMMAND_ERROR 2 +#define FD_COMMAND_OKAY 3 + +static volatile int command_status = FD_COMMAND_NONE; +static unsigned long fdc_busy; +static DECLARE_WAIT_QUEUE_HEAD(fdc_wait); +static DECLARE_WAIT_QUEUE_HEAD(command_done); + +#define NO_SIGNAL (!interruptible || !signal_pending(current)) +#define CALL(x) if ((x) == -EINTR) return -EINTR +#define ECALL(x) if ((ret = (x))) return ret; +#define _WAIT(x,i) CALL(ret=wait_til_done((x),i)) +#define WAIT(x) _WAIT((x),interruptible) +#define IWAIT(x) _WAIT((x),1) + +/* Errors during formatting are counted here. */ +static int format_errors; + +/* Format request descriptor. */ +static struct format_descr format_req; + +/* + * Rate is 0 for 500kb/s, 1 for 300kbps, 2 for 250kbps + * Spec1 is 0xSH, where S is stepping rate (F=1ms, E=2ms, D=3ms etc), + * H is head unload time (1=16ms, 2=32ms, etc) + */ + +/* + * Track buffer + * Because these are written to by the DMA controller, they must + * not contain a 64k byte boundary crossing, or data will be + * corrupted/lost. + */ +static char *floppy_track_buffer; +static int max_buffer_sectors; + +static int *errors; +typedef void (*done_f) (int); +static struct cont_t { + void (*interrupt) (void); /* this is called after the interrupt of the + * main command */ + void (*redo) (void); /* this is called to retry the operation */ + void (*error) (void); /* this is called to tally an error */ + done_f done; /* this is called to say if the operation has + * succeeded/failed */ +} *cont; + +static void floppy_ready(void); +static void floppy_start(void); +static void process_fd_request(void); +static void recalibrate_floppy(void); +static void floppy_shutdown(unsigned long); + +static int floppy_grab_irq_and_dma(void); +static void floppy_release_irq_and_dma(void); + +/* + * The "reset" variable should be tested whenever an interrupt is scheduled, + * after the commands have been sent. This is to ensure that the driver doesn't + * get wedged when the interrupt doesn't come because of a failed command. + * reset doesn't need to be tested before sending commands, because + * output_byte is automatically disabled when reset is set. + */ +#define CHECK_RESET { if (FDCS->reset){ reset_fdc(); return; } } +static void reset_fdc(void); + +/* + * These are global variables, as that's the easiest way to give + * information to interrupts. They are the data used for the current + * request. + */ +#define NO_TRACK -1 +#define NEED_1_RECAL -2 +#define NEED_2_RECAL -3 + +static int usage_count; + +/* buffer related variables */ +static int buffer_track = -1; +static int buffer_drive = -1; +static int buffer_min = -1; +static int buffer_max = -1; + +/* fdc related variables, should end up in a struct */ +static struct floppy_fdc_state fdc_state[N_FDC]; +static int fdc; /* current fdc */ + +static struct floppy_struct *_floppy = floppy_type; +static unsigned char current_drive; +static long current_count_sectors; +static unsigned char fsector_t; /* sector in track */ +static unsigned char in_sector_offset; /* offset within physical sector, + * expressed in units of 512 bytes */ + +#ifndef fd_eject +static inline int fd_eject(int drive) +{ + return -EINVAL; +} +#endif + +/* + * Debugging + * ========= + */ +#ifdef DEBUGT +static long unsigned debugtimer; + +static inline void set_debugt(void) +{ + debugtimer = jiffies; +} + +static inline void debugt(const char *message) +{ + if (DP->flags & DEBUGT) + printk("%s dtime=%lu\n", message, jiffies - debugtimer); +} +#else +static inline void set_debugt(void) { } +static inline void debugt(const char *message) { } +#endif /* DEBUGT */ + +typedef void (*timeout_fn) (unsigned long); +static struct timer_list fd_timeout = TIMER_INITIALIZER(floppy_shutdown, 0, 0); + +static const char *timeout_message; + +#ifdef FLOPPY_SANITY_CHECK +static void is_alive(const char *message) +{ + /* this routine checks whether the floppy driver is "alive" */ + if (test_bit(0, &fdc_busy) && command_status < 2 + && !timer_pending(&fd_timeout)) { + DPRINT("timeout handler died: %s\n", message); + } +} +#endif + +static void (*do_floppy) (void) = NULL; + +#ifdef FLOPPY_SANITY_CHECK + +#define OLOGSIZE 20 + +static void (*lasthandler) (void); +static unsigned long interruptjiffies; +static unsigned long resultjiffies; +static int resultsize; +static unsigned long lastredo; + +static struct output_log { + unsigned char data; + unsigned char status; + unsigned long jiffies; +} output_log[OLOGSIZE]; + +static int output_log_pos; +#endif + +#define current_reqD -1 +#define MAXTIMEOUT -2 + +static void __reschedule_timeout(int drive, const char *message, int marg) +{ + if (drive == current_reqD) + drive = current_drive; + del_timer(&fd_timeout); + if (drive < 0 || drive > N_DRIVE) { + fd_timeout.expires = jiffies + 20UL * HZ; + drive = 0; + } else + fd_timeout.expires = jiffies + UDP->timeout; + add_timer(&fd_timeout); + if (UDP->flags & FD_DEBUG) { + DPRINT("reschedule timeout "); + printk(message, marg); + printk("\n"); + } + timeout_message = message; +} + +static void reschedule_timeout(int drive, const char *message, int marg) +{ + unsigned long flags; + + spin_lock_irqsave(&floppy_lock, flags); + __reschedule_timeout(drive, message, marg); + spin_unlock_irqrestore(&floppy_lock, flags); +} + +#define INFBOUND(a,b) (a)=max_t(int, a, b) + +#define SUPBOUND(a,b) (a)=min_t(int, a, b) + +/* + * Bottom half floppy driver. + * ========================== + * + * This part of the file contains the code talking directly to the hardware, + * and also the main service loop (seek-configure-spinup-command) + */ + +/* + * disk change. + * This routine is responsible for maintaining the FD_DISK_CHANGE flag, + * and the last_checked date. + * + * last_checked is the date of the last check which showed 'no disk change' + * FD_DISK_CHANGE is set under two conditions: + * 1. The floppy has been changed after some i/o to that floppy already + * took place. + * 2. No floppy disk is in the drive. This is done in order to ensure that + * requests are quickly flushed in case there is no disk in the drive. It + * follows that FD_DISK_CHANGE can only be cleared if there is a disk in + * the drive. + * + * For 1., maxblock is observed. Maxblock is 0 if no i/o has taken place yet. + * For 2., FD_DISK_NEWCHANGE is watched. FD_DISK_NEWCHANGE is cleared on + * each seek. If a disk is present, the disk change line should also be + * cleared on each seek. Thus, if FD_DISK_NEWCHANGE is clear, but the disk + * change line is set, this means either that no disk is in the drive, or + * that it has been removed since the last seek. + * + * This means that we really have a third possibility too: + * The floppy has been changed after the last seek. + */ + +static int disk_change(int drive) +{ + int fdc = FDC(drive); +#ifdef FLOPPY_SANITY_CHECK + if (jiffies - UDRS->select_date < UDP->select_delay) + DPRINT("WARNING disk change called early\n"); + if (!(FDCS->dor & (0x10 << UNIT(drive))) || + (FDCS->dor & 3) != UNIT(drive) || fdc != FDC(drive)) { + DPRINT("probing disk change on unselected drive\n"); + DPRINT("drive=%d fdc=%d dor=%x\n", drive, FDC(drive), + (unsigned int)FDCS->dor); + } +#endif + +#ifdef DCL_DEBUG + if (UDP->flags & FD_DEBUG) { + DPRINT("checking disk change line for drive %d\n", drive); + DPRINT("jiffies=%lu\n", jiffies); + DPRINT("disk change line=%x\n", fd_inb(FD_DIR) & 0x80); + DPRINT("flags=%lx\n", UDRS->flags); + } +#endif + if (UDP->flags & FD_BROKEN_DCL) + return UTESTF(FD_DISK_CHANGED); + if ((fd_inb(FD_DIR) ^ UDP->flags) & 0x80) { + USETF(FD_VERIFY); /* verify write protection */ + if (UDRS->maxblock) { + /* mark it changed */ + USETF(FD_DISK_CHANGED); + } + + /* invalidate its geometry */ + if (UDRS->keep_data >= 0) { + if ((UDP->flags & FTD_MSG) && + current_type[drive] != NULL) + DPRINT("Disk type is undefined after " + "disk change\n"); + current_type[drive] = NULL; + floppy_sizes[TOMINOR(drive)] = MAX_DISK_SIZE << 1; + } + + /*USETF(FD_DISK_NEWCHANGE); */ + return 1; + } else { + UDRS->last_checked = jiffies; + UCLEARF(FD_DISK_NEWCHANGE); + } + return 0; +} + +static inline int is_selected(int dor, int unit) +{ + return ((dor & (0x10 << unit)) && (dor & 3) == unit); +} + +static int set_dor(int fdc, char mask, char data) +{ + register unsigned char drive, unit, newdor, olddor; + + if (FDCS->address == -1) + return -1; + + olddor = FDCS->dor; + newdor = (olddor & mask) | data; + if (newdor != olddor) { + unit = olddor & 0x3; + if (is_selected(olddor, unit) && !is_selected(newdor, unit)) { + drive = REVDRIVE(fdc, unit); +#ifdef DCL_DEBUG + if (UDP->flags & FD_DEBUG) { + DPRINT("calling disk change from set_dor\n"); + } +#endif + disk_change(drive); + } + FDCS->dor = newdor; + fd_outb(newdor, FD_DOR); + + unit = newdor & 0x3; + if (!is_selected(olddor, unit) && is_selected(newdor, unit)) { + drive = REVDRIVE(fdc, unit); + UDRS->select_date = jiffies; + } + } + /* + * We should propagate failures to grab the resources back + * nicely from here. Actually we ought to rewrite the fd + * driver some day too. + */ + if (newdor & FLOPPY_MOTOR_MASK) + floppy_grab_irq_and_dma(); + if (olddor & FLOPPY_MOTOR_MASK) + floppy_release_irq_and_dma(); + return olddor; +} + +static void twaddle(void) +{ + if (DP->select_delay) + return; + fd_outb(FDCS->dor & ~(0x10 << UNIT(current_drive)), FD_DOR); + fd_outb(FDCS->dor, FD_DOR); + DRS->select_date = jiffies; +} + +/* reset all driver information about the current fdc. This is needed after + * a reset, and after a raw command. */ +static void reset_fdc_info(int mode) +{ + int drive; + + FDCS->spec1 = FDCS->spec2 = -1; + FDCS->need_configure = 1; + FDCS->perp_mode = 1; + FDCS->rawcmd = 0; + for (drive = 0; drive < N_DRIVE; drive++) + if (FDC(drive) == fdc && (mode || UDRS->track != NEED_1_RECAL)) + UDRS->track = NEED_2_RECAL; +} + +/* selects the fdc and drive, and enables the fdc's input/dma. */ +static void set_fdc(int drive) +{ + if (drive >= 0 && drive < N_DRIVE) { + fdc = FDC(drive); + current_drive = drive; + } + if (fdc != 1 && fdc != 0) { + printk("bad fdc value\n"); + return; + } + set_dor(fdc, ~0, 8); +#if N_FDC > 1 + set_dor(1 - fdc, ~8, 0); +#endif + if (FDCS->rawcmd == 2) + reset_fdc_info(1); + if (fd_inb(FD_STATUS) != STATUS_READY) + FDCS->reset = 1; +} + +/* locks the driver */ +static int _lock_fdc(int drive, int interruptible, int line) +{ + if (!usage_count) { + printk(KERN_ERR + "Trying to lock fdc while usage count=0 at line %d\n", + line); + return -1; + } + if (floppy_grab_irq_and_dma() == -1) + return -EBUSY; + + if (test_and_set_bit(0, &fdc_busy)) { + DECLARE_WAITQUEUE(wait, current); + add_wait_queue(&fdc_wait, &wait); + + for (;;) { + set_current_state(TASK_INTERRUPTIBLE); + + if (!test_and_set_bit(0, &fdc_busy)) + break; + + schedule(); + + if (!NO_SIGNAL) { + remove_wait_queue(&fdc_wait, &wait); + return -EINTR; + } + } + + set_current_state(TASK_RUNNING); + remove_wait_queue(&fdc_wait, &wait); + } + command_status = FD_COMMAND_NONE; + + __reschedule_timeout(drive, "lock fdc", 0); + set_fdc(drive); + return 0; +} + +#define lock_fdc(drive,interruptible) _lock_fdc(drive,interruptible, __LINE__) + +#define LOCK_FDC(drive,interruptible) \ +if (lock_fdc(drive,interruptible)) return -EINTR; + +/* unlocks the driver */ +static inline void unlock_fdc(void) +{ + unsigned long flags; + + raw_cmd = NULL; + if (!test_bit(0, &fdc_busy)) + DPRINT("FDC access conflict!\n"); + + if (do_floppy) + DPRINT("device interrupt still active at FDC release: %p!\n", + do_floppy); + command_status = FD_COMMAND_NONE; + spin_lock_irqsave(&floppy_lock, flags); + del_timer(&fd_timeout); + cont = NULL; + clear_bit(0, &fdc_busy); + if (elv_next_request(floppy_queue)) + do_fd_request(floppy_queue); + spin_unlock_irqrestore(&floppy_lock, flags); + floppy_release_irq_and_dma(); + wake_up(&fdc_wait); +} + +/* switches the motor off after a given timeout */ +static void motor_off_callback(unsigned long nr) +{ + unsigned char mask = ~(0x10 << UNIT(nr)); + + set_dor(FDC(nr), mask, 0); +} + +/* schedules motor off */ +static void floppy_off(unsigned int drive) +{ + unsigned long volatile delta; + register int fdc = FDC(drive); + + if (!(FDCS->dor & (0x10 << UNIT(drive)))) + return; + + del_timer(motor_off_timer + drive); + + /* make spindle stop in a position which minimizes spinup time + * next time */ + if (UDP->rps) { + delta = jiffies - UDRS->first_read_date + HZ - + UDP->spindown_offset; + delta = ((delta * UDP->rps) % HZ) / UDP->rps; + motor_off_timer[drive].expires = + jiffies + UDP->spindown - delta; + } + add_timer(motor_off_timer + drive); +} + +/* + * cycle through all N_DRIVE floppy drives, for disk change testing. + * stopping at current drive. This is done before any long operation, to + * be sure to have up to date disk change information. + */ +static void scandrives(void) +{ + int i, drive, saved_drive; + + if (DP->select_delay) + return; + + saved_drive = current_drive; + for (i = 0; i < N_DRIVE; i++) { + drive = (saved_drive + i + 1) % N_DRIVE; + if (UDRS->fd_ref == 0 || UDP->select_delay != 0) + continue; /* skip closed drives */ + set_fdc(drive); + if (!(set_dor(fdc, ~3, UNIT(drive) | (0x10 << UNIT(drive))) & + (0x10 << UNIT(drive)))) + /* switch the motor off again, if it was off to + * begin with */ + set_dor(fdc, ~(0x10 << UNIT(drive)), 0); + } + set_fdc(saved_drive); +} + +static void empty(void) +{ +} + +static DECLARE_WORK(floppy_work, NULL, NULL); + +static void schedule_bh(void (*handler) (void)) +{ + PREPARE_WORK(&floppy_work, (void (*)(void *))handler, NULL); + schedule_work(&floppy_work); +} + +static struct timer_list fd_timer = TIMER_INITIALIZER(NULL, 0, 0); + +static void cancel_activity(void) +{ + unsigned long flags; + + spin_lock_irqsave(&floppy_lock, flags); + do_floppy = NULL; + PREPARE_WORK(&floppy_work, (void *)empty, NULL); + del_timer(&fd_timer); + spin_unlock_irqrestore(&floppy_lock, flags); +} + +/* this function makes sure that the disk stays in the drive during the + * transfer */ +static void fd_watchdog(void) +{ +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT("calling disk change from watchdog\n"); + } +#endif + + if (disk_change(current_drive)) { + DPRINT("disk removed during i/o\n"); + cancel_activity(); + cont->done(0); + reset_fdc(); + } else { + del_timer(&fd_timer); + fd_timer.function = (timeout_fn) fd_watchdog; + fd_timer.expires = jiffies + HZ / 10; + add_timer(&fd_timer); + } +} + +static void main_command_interrupt(void) +{ + del_timer(&fd_timer); + cont->interrupt(); +} + +/* waits for a delay (spinup or select) to pass */ +static int fd_wait_for_completion(unsigned long delay, timeout_fn function) +{ + if (FDCS->reset) { + reset_fdc(); /* do the reset during sleep to win time + * if we don't need to sleep, it's a good + * occasion anyways */ + return 1; + } + + if ((signed)(jiffies - delay) < 0) { + del_timer(&fd_timer); + fd_timer.function = function; + fd_timer.expires = delay; + add_timer(&fd_timer); + return 1; + } + return 0; +} + +static DEFINE_SPINLOCK(floppy_hlt_lock); +static int hlt_disabled; +static void floppy_disable_hlt(void) +{ + unsigned long flags; + + spin_lock_irqsave(&floppy_hlt_lock, flags); + if (!hlt_disabled) { + hlt_disabled = 1; +#ifdef HAVE_DISABLE_HLT + disable_hlt(); +#endif + } + spin_unlock_irqrestore(&floppy_hlt_lock, flags); +} + +static void floppy_enable_hlt(void) +{ + unsigned long flags; + + spin_lock_irqsave(&floppy_hlt_lock, flags); + if (hlt_disabled) { + hlt_disabled = 0; +#ifdef HAVE_DISABLE_HLT + enable_hlt(); +#endif + } + spin_unlock_irqrestore(&floppy_hlt_lock, flags); +} + +static void setup_DMA(void) +{ + unsigned long f; + +#ifdef FLOPPY_SANITY_CHECK + if (raw_cmd->length == 0) { + int i; + + printk("zero dma transfer size:"); + for (i = 0; i < raw_cmd->cmd_count; i++) + printk("%x,", raw_cmd->cmd[i]); + printk("\n"); + cont->done(0); + FDCS->reset = 1; + return; + } + if (((unsigned long)raw_cmd->kernel_data) % 512) { + printk("non aligned address: %p\n", raw_cmd->kernel_data); + cont->done(0); + FDCS->reset = 1; + return; + } +#endif + f = claim_dma_lock(); + fd_disable_dma(); +#ifdef fd_dma_setup + if (fd_dma_setup(raw_cmd->kernel_data, raw_cmd->length, + (raw_cmd->flags & FD_RAW_READ) ? + DMA_MODE_READ : DMA_MODE_WRITE, FDCS->address) < 0) { + release_dma_lock(f); + cont->done(0); + FDCS->reset = 1; + return; + } + release_dma_lock(f); +#else + fd_clear_dma_ff(); + fd_cacheflush(raw_cmd->kernel_data, raw_cmd->length); + fd_set_dma_mode((raw_cmd->flags & FD_RAW_READ) ? + DMA_MODE_READ : DMA_MODE_WRITE); + fd_set_dma_addr(raw_cmd->kernel_data); + fd_set_dma_count(raw_cmd->length); + virtual_dma_port = FDCS->address; + fd_enable_dma(); + release_dma_lock(f); +#endif + floppy_disable_hlt(); +} + +static void show_floppy(void); + +/* waits until the fdc becomes ready */ +static int wait_til_ready(void) +{ + int counter, status; + if (FDCS->reset) + return -1; + for (counter = 0; counter < 10000; counter++) { + status = fd_inb(FD_STATUS); + if (status & STATUS_READY) + return status; + } + if (!initialising) { + DPRINT("Getstatus times out (%x) on fdc %d\n", status, fdc); + show_floppy(); + } + FDCS->reset = 1; + return -1; +} + +/* sends a command byte to the fdc */ +static int output_byte(char byte) +{ + int status; + + if ((status = wait_til_ready()) < 0) + return -1; + if ((status & (STATUS_READY | STATUS_DIR | STATUS_DMA)) == STATUS_READY) { + fd_outb(byte, FD_DATA); +#ifdef FLOPPY_SANITY_CHECK + output_log[output_log_pos].data = byte; + output_log[output_log_pos].status = status; + output_log[output_log_pos].jiffies = jiffies; + output_log_pos = (output_log_pos + 1) % OLOGSIZE; +#endif + return 0; + } + FDCS->reset = 1; + if (!initialising) { + DPRINT("Unable to send byte %x to FDC. Fdc=%x Status=%x\n", + byte, fdc, status); + show_floppy(); + } + return -1; +} + +#define LAST_OUT(x) if (output_byte(x)<0){ reset_fdc();return;} + +/* gets the response from the fdc */ +static int result(void) +{ + int i, status = 0; + + for (i = 0; i < MAX_REPLIES; i++) { + if ((status = wait_til_ready()) < 0) + break; + status &= STATUS_DIR | STATUS_READY | STATUS_BUSY | STATUS_DMA; + if ((status & ~STATUS_BUSY) == STATUS_READY) { +#ifdef FLOPPY_SANITY_CHECK + resultjiffies = jiffies; + resultsize = i; +#endif + return i; + } + if (status == (STATUS_DIR | STATUS_READY | STATUS_BUSY)) + reply_buffer[i] = fd_inb(FD_DATA); + else + break; + } + if (!initialising) { + DPRINT + ("get result error. Fdc=%d Last status=%x Read bytes=%d\n", + fdc, status, i); + show_floppy(); + } + FDCS->reset = 1; + return -1; +} + +#define MORE_OUTPUT -2 +/* does the fdc need more output? */ +static int need_more_output(void) +{ + int status; + if ((status = wait_til_ready()) < 0) + return -1; + if ((status & (STATUS_READY | STATUS_DIR | STATUS_DMA)) == STATUS_READY) + return MORE_OUTPUT; + return result(); +} + +/* Set perpendicular mode as required, based on data rate, if supported. + * 82077 Now tested. 1Mbps data rate only possible with 82077-1. + */ +static inline void perpendicular_mode(void) +{ + unsigned char perp_mode; + + if (raw_cmd->rate & 0x40) { + switch (raw_cmd->rate & 3) { + case 0: + perp_mode = 2; + break; + case 3: + perp_mode = 3; + break; + default: + DPRINT("Invalid data rate for perpendicular mode!\n"); + cont->done(0); + FDCS->reset = 1; /* convenient way to return to + * redo without to much hassle (deep + * stack et al. */ + return; + } + } else + perp_mode = 0; + + if (FDCS->perp_mode == perp_mode) + return; + if (FDCS->version >= FDC_82077_ORIG) { + output_byte(FD_PERPENDICULAR); + output_byte(perp_mode); + FDCS->perp_mode = perp_mode; + } else if (perp_mode) { + DPRINT("perpendicular mode not supported by this FDC.\n"); + } +} /* perpendicular_mode */ + +static int fifo_depth = 0xa; +static int no_fifo; + +static int fdc_configure(void) +{ + /* Turn on FIFO */ + output_byte(FD_CONFIGURE); + if (need_more_output() != MORE_OUTPUT) + return 0; + output_byte(0); + output_byte(0x10 | (no_fifo & 0x20) | (fifo_depth & 0xf)); + output_byte(0); /* pre-compensation from track + 0 upwards */ + return 1; +} + +#define NOMINAL_DTR 500 + +/* Issue a "SPECIFY" command to set the step rate time, head unload time, + * head load time, and DMA disable flag to values needed by floppy. + * + * The value "dtr" is the data transfer rate in Kbps. It is needed + * to account for the data rate-based scaling done by the 82072 and 82077 + * FDC types. This parameter is ignored for other types of FDCs (i.e. + * 8272a). + * + * Note that changing the data transfer rate has a (probably deleterious) + * effect on the parameters subject to scaling for 82072/82077 FDCs, so + * fdc_specify is called again after each data transfer rate + * change. + * + * srt: 1000 to 16000 in microseconds + * hut: 16 to 240 milliseconds + * hlt: 2 to 254 milliseconds + * + * These values are rounded up to the next highest available delay time. + */ +static void fdc_specify(void) +{ + unsigned char spec1, spec2; + unsigned long srt, hlt, hut; + unsigned long dtr = NOMINAL_DTR; + unsigned long scale_dtr = NOMINAL_DTR; + int hlt_max_code = 0x7f; + int hut_max_code = 0xf; + + if (FDCS->need_configure && FDCS->version >= FDC_82072A) { + fdc_configure(); + FDCS->need_configure = 0; + /*DPRINT("FIFO enabled\n"); */ + } + + switch (raw_cmd->rate & 0x03) { + case 3: + dtr = 1000; + break; + case 1: + dtr = 300; + if (FDCS->version >= FDC_82078) { + /* chose the default rate table, not the one + * where 1 = 2 Mbps */ + output_byte(FD_DRIVESPEC); + if (need_more_output() == MORE_OUTPUT) { + output_byte(UNIT(current_drive)); + output_byte(0xc0); + } + } + break; + case 2: + dtr = 250; + break; + } + + if (FDCS->version >= FDC_82072) { + scale_dtr = dtr; + hlt_max_code = 0x00; /* 0==256msec*dtr0/dtr (not linear!) */ + hut_max_code = 0x0; /* 0==256msec*dtr0/dtr (not linear!) */ + } + + /* Convert step rate from microseconds to milliseconds and 4 bits */ + srt = 16 - (DP->srt * scale_dtr / 1000 + NOMINAL_DTR - 1) / NOMINAL_DTR; + if (slow_floppy) { + srt = srt / 4; + } + SUPBOUND(srt, 0xf); + INFBOUND(srt, 0); + + hlt = (DP->hlt * scale_dtr / 2 + NOMINAL_DTR - 1) / NOMINAL_DTR; + if (hlt < 0x01) + hlt = 0x01; + else if (hlt > 0x7f) + hlt = hlt_max_code; + + hut = (DP->hut * scale_dtr / 16 + NOMINAL_DTR - 1) / NOMINAL_DTR; + if (hut < 0x1) + hut = 0x1; + else if (hut > 0xf) + hut = hut_max_code; + + spec1 = (srt << 4) | hut; + spec2 = (hlt << 1) | (use_virtual_dma & 1); + + /* If these parameters did not change, just return with success */ + if (FDCS->spec1 != spec1 || FDCS->spec2 != spec2) { + /* Go ahead and set spec1 and spec2 */ + output_byte(FD_SPECIFY); + output_byte(FDCS->spec1 = spec1); + output_byte(FDCS->spec2 = spec2); + } +} /* fdc_specify */ + +/* Set the FDC's data transfer rate on behalf of the specified drive. + * NOTE: with 82072/82077 FDCs, changing the data rate requires a reissue + * of the specify command (i.e. using the fdc_specify function). + */ +static int fdc_dtr(void) +{ + /* If data rate not already set to desired value, set it. */ + if ((raw_cmd->rate & 3) == FDCS->dtr) + return 0; + + /* Set dtr */ + fd_outb(raw_cmd->rate & 3, FD_DCR); + + /* TODO: some FDC/drive combinations (C&T 82C711 with TEAC 1.2MB) + * need a stabilization period of several milliseconds to be + * enforced after data rate changes before R/W operations. + * Pause 5 msec to avoid trouble. (Needs to be 2 jiffies) + */ + FDCS->dtr = raw_cmd->rate & 3; + return (fd_wait_for_completion(jiffies + 2UL * HZ / 100, + (timeout_fn) floppy_ready)); +} /* fdc_dtr */ + +static void tell_sector(void) +{ + printk(": track %d, head %d, sector %d, size %d", + R_TRACK, R_HEAD, R_SECTOR, R_SIZECODE); +} /* tell_sector */ + +/* + * OK, this error interpreting routine is called after a + * DMA read/write has succeeded + * or failed, so we check the results, and copy any buffers. + * hhb: Added better error reporting. + * ak: Made this into a separate routine. + */ +static int interpret_errors(void) +{ + char bad; + + if (inr != 7) { + DPRINT("-- FDC reply error"); + FDCS->reset = 1; + return 1; + } + + /* check IC to find cause of interrupt */ + switch (ST0 & ST0_INTR) { + case 0x40: /* error occurred during command execution */ + if (ST1 & ST1_EOC) + return 0; /* occurs with pseudo-DMA */ + bad = 1; + if (ST1 & ST1_WP) { + DPRINT("Drive is write protected\n"); + CLEARF(FD_DISK_WRITABLE); + cont->done(0); + bad = 2; + } else if (ST1 & ST1_ND) { + SETF(FD_NEED_TWADDLE); + } else if (ST1 & ST1_OR) { + if (DP->flags & FTD_MSG) + DPRINT("Over/Underrun - retrying\n"); + bad = 0; + } else if (*errors >= DP->max_errors.reporting) { + DPRINT(""); + if (ST0 & ST0_ECE) { + printk("Recalibrate failed!"); + } else if (ST2 & ST2_CRC) { + printk("data CRC error"); + tell_sector(); + } else if (ST1 & ST1_CRC) { + printk("CRC error"); + tell_sector(); + } else if ((ST1 & (ST1_MAM | ST1_ND)) + || (ST2 & ST2_MAM)) { + if (!probing) { + printk("sector not found"); + tell_sector(); + } else + printk("probe failed..."); + } else if (ST2 & ST2_WC) { /* seek error */ + printk("wrong cylinder"); + } else if (ST2 & ST2_BC) { /* cylinder marked as bad */ + printk("bad cylinder"); + } else { + printk + ("unknown error. ST[0..2] are: 0x%x 0x%x 0x%x", + ST0, ST1, ST2); + tell_sector(); + } + printk("\n"); + + } + if (ST2 & ST2_WC || ST2 & ST2_BC) + /* wrong cylinder => recal */ + DRS->track = NEED_2_RECAL; + return bad; + case 0x80: /* invalid command given */ + DPRINT("Invalid FDC command given!\n"); + cont->done(0); + return 2; + case 0xc0: + DPRINT("Abnormal termination caused by polling\n"); + cont->error(); + return 2; + default: /* (0) Normal command termination */ + return 0; + } +} + +/* + * This routine is called when everything should be correctly set up + * for the transfer (i.e. floppy motor is on, the correct floppy is + * selected, and the head is sitting on the right track). + */ +static void setup_rw_floppy(void) +{ + int i, r, flags, dflags; + unsigned long ready_date; + timeout_fn function; + + flags = raw_cmd->flags; + if (flags & (FD_RAW_READ | FD_RAW_WRITE)) + flags |= FD_RAW_INTR; + + if ((flags & FD_RAW_SPIN) && !(flags & FD_RAW_NO_MOTOR)) { + ready_date = DRS->spinup_date + DP->spinup; + /* If spinup will take a long time, rerun scandrives + * again just before spinup completion. Beware that + * after scandrives, we must again wait for selection. + */ + if ((signed)(ready_date - jiffies) > DP->select_delay) { + ready_date -= DP->select_delay; + function = (timeout_fn) floppy_start; + } else + function = (timeout_fn) setup_rw_floppy; + + /* wait until the floppy is spinning fast enough */ + if (fd_wait_for_completion(ready_date, function)) + return; + } + dflags = DRS->flags; + + if ((flags & FD_RAW_READ) || (flags & FD_RAW_WRITE)) + setup_DMA(); + + if (flags & FD_RAW_INTR) + do_floppy = main_command_interrupt; + + r = 0; + for (i = 0; i < raw_cmd->cmd_count; i++) + r |= output_byte(raw_cmd->cmd[i]); + + debugt("rw_command: "); + + if (r) { + cont->error(); + reset_fdc(); + return; + } + + if (!(flags & FD_RAW_INTR)) { + inr = result(); + cont->interrupt(); + } else if (flags & FD_RAW_NEED_DISK) + fd_watchdog(); +} + +static int blind_seek; + +/* + * This is the routine called after every seek (or recalibrate) interrupt + * from the floppy controller. + */ +static void seek_interrupt(void) +{ + debugt("seek interrupt:"); + if (inr != 2 || (ST0 & 0xF8) != 0x20) { + DPRINT("seek failed\n"); + DRS->track = NEED_2_RECAL; + cont->error(); + cont->redo(); + return; + } + if (DRS->track >= 0 && DRS->track != ST1 && !blind_seek) { +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT + ("clearing NEWCHANGE flag because of effective seek\n"); + DPRINT("jiffies=%lu\n", jiffies); + } +#endif + CLEARF(FD_DISK_NEWCHANGE); /* effective seek */ + DRS->select_date = jiffies; + } + DRS->track = ST1; + floppy_ready(); +} + +static void check_wp(void) +{ + if (TESTF(FD_VERIFY)) { + /* check write protection */ + output_byte(FD_GETSTATUS); + output_byte(UNIT(current_drive)); + if (result() != 1) { + FDCS->reset = 1; + return; + } + CLEARF(FD_VERIFY); + CLEARF(FD_NEED_TWADDLE); +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT("checking whether disk is write protected\n"); + DPRINT("wp=%x\n", ST3 & 0x40); + } +#endif + if (!(ST3 & 0x40)) + SETF(FD_DISK_WRITABLE); + else + CLEARF(FD_DISK_WRITABLE); + } +} + +static void seek_floppy(void) +{ + int track; + + blind_seek = 0; + +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT("calling disk change from seek\n"); + } +#endif + + if (!TESTF(FD_DISK_NEWCHANGE) && + disk_change(current_drive) && (raw_cmd->flags & FD_RAW_NEED_DISK)) { + /* the media changed flag should be cleared after the seek. + * If it isn't, this means that there is really no disk in + * the drive. + */ + SETF(FD_DISK_CHANGED); + cont->done(0); + cont->redo(); + return; + } + if (DRS->track <= NEED_1_RECAL) { + recalibrate_floppy(); + return; + } else if (TESTF(FD_DISK_NEWCHANGE) && + (raw_cmd->flags & FD_RAW_NEED_DISK) && + (DRS->track <= NO_TRACK || DRS->track == raw_cmd->track)) { + /* we seek to clear the media-changed condition. Does anybody + * know a more elegant way, which works on all drives? */ + if (raw_cmd->track) + track = raw_cmd->track - 1; + else { + if (DP->flags & FD_SILENT_DCL_CLEAR) { + set_dor(fdc, ~(0x10 << UNIT(current_drive)), 0); + blind_seek = 1; + raw_cmd->flags |= FD_RAW_NEED_SEEK; + } + track = 1; + } + } else { + check_wp(); + if (raw_cmd->track != DRS->track && + (raw_cmd->flags & FD_RAW_NEED_SEEK)) + track = raw_cmd->track; + else { + setup_rw_floppy(); + return; + } + } + + do_floppy = seek_interrupt; + output_byte(FD_SEEK); + output_byte(UNIT(current_drive)); + LAST_OUT(track); + debugt("seek command:"); +} + +static void recal_interrupt(void) +{ + debugt("recal interrupt:"); + if (inr != 2) + FDCS->reset = 1; + else if (ST0 & ST0_ECE) { + switch (DRS->track) { + case NEED_1_RECAL: + debugt("recal interrupt need 1 recal:"); + /* after a second recalibrate, we still haven't + * reached track 0. Probably no drive. Raise an + * error, as failing immediately might upset + * computers possessed by the Devil :-) */ + cont->error(); + cont->redo(); + return; + case NEED_2_RECAL: + debugt("recal interrupt need 2 recal:"); + /* If we already did a recalibrate, + * and we are not at track 0, this + * means we have moved. (The only way + * not to move at recalibration is to + * be already at track 0.) Clear the + * new change flag */ +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT + ("clearing NEWCHANGE flag because of second recalibrate\n"); + } +#endif + + CLEARF(FD_DISK_NEWCHANGE); + DRS->select_date = jiffies; + /* fall through */ + default: + debugt("recal interrupt default:"); + /* Recalibrate moves the head by at + * most 80 steps. If after one + * recalibrate we don't have reached + * track 0, this might mean that we + * started beyond track 80. Try + * again. */ + DRS->track = NEED_1_RECAL; + break; + } + } else + DRS->track = ST1; + floppy_ready(); +} + +static void print_result(char *message, int inr) +{ + int i; + + DPRINT("%s ", message); + if (inr >= 0) + for (i = 0; i < inr; i++) + printk("repl[%d]=%x ", i, reply_buffer[i]); + printk("\n"); +} + +/* interrupt handler. Note that this can be called externally on the Sparc */ +irqreturn_t floppy_interrupt(int irq, void *dev_id, struct pt_regs *regs) +{ + void (*handler) (void) = do_floppy; + int do_print; + unsigned long f; + + lasthandler = handler; + interruptjiffies = jiffies; + + f = claim_dma_lock(); + fd_disable_dma(); + release_dma_lock(f); + + floppy_enable_hlt(); + do_floppy = NULL; + if (fdc >= N_FDC || FDCS->address == -1) { + /* we don't even know which FDC is the culprit */ + printk("DOR0=%x\n", fdc_state[0].dor); + printk("floppy interrupt on bizarre fdc %d\n", fdc); + printk("handler=%p\n", handler); + is_alive("bizarre fdc"); + return IRQ_NONE; + } + + FDCS->reset = 0; + /* We have to clear the reset flag here, because apparently on boxes + * with level triggered interrupts (PS/2, Sparc, ...), it is needed to + * emit SENSEI's to clear the interrupt line. And FDCS->reset blocks the + * emission of the SENSEI's. + * It is OK to emit floppy commands because we are in an interrupt + * handler here, and thus we have to fear no interference of other + * activity. + */ + + do_print = !handler && print_unex && !initialising; + + inr = result(); + if (do_print) + print_result("unexpected interrupt", inr); + if (inr == 0) { + int max_sensei = 4; + do { + output_byte(FD_SENSEI); + inr = result(); + if (do_print) + print_result("sensei", inr); + max_sensei--; + } while ((ST0 & 0x83) != UNIT(current_drive) && inr == 2 + && max_sensei); + } + if (!handler) { + FDCS->reset = 1; + return IRQ_NONE; + } + schedule_bh(handler); + is_alive("normal interrupt end"); + + /* FIXME! Was it really for us? */ + return IRQ_HANDLED; +} + +static void recalibrate_floppy(void) +{ + debugt("recalibrate floppy:"); + do_floppy = recal_interrupt; + output_byte(FD_RECALIBRATE); + LAST_OUT(UNIT(current_drive)); +} + +/* + * Must do 4 FD_SENSEIs after reset because of ``drive polling''. + */ +static void reset_interrupt(void) +{ + debugt("reset interrupt:"); + result(); /* get the status ready for set_fdc */ + if (FDCS->reset) { + printk("reset set in interrupt, calling %p\n", cont->error); + cont->error(); /* a reset just after a reset. BAD! */ + } + cont->redo(); +} + +/* + * reset is done by pulling bit 2 of DOR low for a while (old FDCs), + * or by setting the self clearing bit 7 of STATUS (newer FDCs) + */ +static void reset_fdc(void) +{ + unsigned long flags; + + do_floppy = reset_interrupt; + FDCS->reset = 0; + reset_fdc_info(0); + + /* Pseudo-DMA may intercept 'reset finished' interrupt. */ + /* Irrelevant for systems with true DMA (i386). */ + + flags = claim_dma_lock(); + fd_disable_dma(); + release_dma_lock(flags); + + if (FDCS->version >= FDC_82072A) + fd_outb(0x80 | (FDCS->dtr & 3), FD_STATUS); + else { + fd_outb(FDCS->dor & ~0x04, FD_DOR); + udelay(FD_RESET_DELAY); + fd_outb(FDCS->dor, FD_DOR); + } +} + +static void show_floppy(void) +{ + int i; + + printk("\n"); + printk("floppy driver state\n"); + printk("-------------------\n"); + printk("now=%lu last interrupt=%lu diff=%lu last called handler=%p\n", + jiffies, interruptjiffies, jiffies - interruptjiffies, + lasthandler); + +#ifdef FLOPPY_SANITY_CHECK + printk("timeout_message=%s\n", timeout_message); + printk("last output bytes:\n"); + for (i = 0; i < OLOGSIZE; i++) + printk("%2x %2x %lu\n", + output_log[(i + output_log_pos) % OLOGSIZE].data, + output_log[(i + output_log_pos) % OLOGSIZE].status, + output_log[(i + output_log_pos) % OLOGSIZE].jiffies); + printk("last result at %lu\n", resultjiffies); + printk("last redo_fd_request at %lu\n", lastredo); + for (i = 0; i < resultsize; i++) { + printk("%2x ", reply_buffer[i]); + } + printk("\n"); +#endif + + printk("status=%x\n", fd_inb(FD_STATUS)); + printk("fdc_busy=%lu\n", fdc_busy); + if (do_floppy) + printk("do_floppy=%p\n", do_floppy); + if (floppy_work.pending) + printk("floppy_work.func=%p\n", floppy_work.func); + if (timer_pending(&fd_timer)) + printk("fd_timer.function=%p\n", fd_timer.function); + if (timer_pending(&fd_timeout)) { + printk("timer_function=%p\n", fd_timeout.function); + printk("expires=%lu\n", fd_timeout.expires - jiffies); + printk("now=%lu\n", jiffies); + } + printk("cont=%p\n", cont); + printk("current_req=%p\n", current_req); + printk("command_status=%d\n", command_status); + printk("\n"); +} + +static void floppy_shutdown(unsigned long data) +{ + unsigned long flags; + + if (!initialising) + show_floppy(); + cancel_activity(); + + floppy_enable_hlt(); + + flags = claim_dma_lock(); + fd_disable_dma(); + release_dma_lock(flags); + + /* avoid dma going to a random drive after shutdown */ + + if (!initialising) + DPRINT("floppy timeout called\n"); + FDCS->reset = 1; + if (cont) { + cont->done(0); + cont->redo(); /* this will recall reset when needed */ + } else { + printk("no cont in shutdown!\n"); + process_fd_request(); + } + is_alive("floppy shutdown"); +} + +/*typedef void (*timeout_fn)(unsigned long);*/ + +/* start motor, check media-changed condition and write protection */ +static int start_motor(void (*function) (void)) +{ + int mask, data; + + mask = 0xfc; + data = UNIT(current_drive); + if (!(raw_cmd->flags & FD_RAW_NO_MOTOR)) { + if (!(FDCS->dor & (0x10 << UNIT(current_drive)))) { + set_debugt(); + /* no read since this drive is running */ + DRS->first_read_date = 0; + /* note motor start time if motor is not yet running */ + DRS->spinup_date = jiffies; + data |= (0x10 << UNIT(current_drive)); + } + } else if (FDCS->dor & (0x10 << UNIT(current_drive))) + mask &= ~(0x10 << UNIT(current_drive)); + + /* starts motor and selects floppy */ + del_timer(motor_off_timer + current_drive); + set_dor(fdc, mask, data); + + /* wait_for_completion also schedules reset if needed. */ + return (fd_wait_for_completion(DRS->select_date + DP->select_delay, + (timeout_fn) function)); +} + +static void floppy_ready(void) +{ + CHECK_RESET; + if (start_motor(floppy_ready)) + return; + if (fdc_dtr()) + return; + +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT("calling disk change from floppy_ready\n"); + } +#endif + if (!(raw_cmd->flags & FD_RAW_NO_MOTOR) && + disk_change(current_drive) && !DP->select_delay) + twaddle(); /* this clears the dcl on certain drive/controller + * combinations */ + +#ifdef fd_chose_dma_mode + if ((raw_cmd->flags & FD_RAW_READ) || (raw_cmd->flags & FD_RAW_WRITE)) { + unsigned long flags = claim_dma_lock(); + fd_chose_dma_mode(raw_cmd->kernel_data, raw_cmd->length); + release_dma_lock(flags); + } +#endif + + if (raw_cmd->flags & (FD_RAW_NEED_SEEK | FD_RAW_NEED_DISK)) { + perpendicular_mode(); + fdc_specify(); /* must be done here because of hut, hlt ... */ + seek_floppy(); + } else { + if ((raw_cmd->flags & FD_RAW_READ) || + (raw_cmd->flags & FD_RAW_WRITE)) + fdc_specify(); + setup_rw_floppy(); + } +} + +static void floppy_start(void) +{ + reschedule_timeout(current_reqD, "floppy start", 0); + + scandrives(); +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT("setting NEWCHANGE in floppy_start\n"); + } +#endif + SETF(FD_DISK_NEWCHANGE); + floppy_ready(); +} + +/* + * ======================================================================== + * here ends the bottom half. Exported routines are: + * floppy_start, floppy_off, floppy_ready, lock_fdc, unlock_fdc, set_fdc, + * start_motor, reset_fdc, reset_fdc_info, interpret_errors. + * Initialization also uses output_byte, result, set_dor, floppy_interrupt + * and set_dor. + * ======================================================================== + */ +/* + * General purpose continuations. + * ============================== + */ + +static void do_wakeup(void) +{ + reschedule_timeout(MAXTIMEOUT, "do wakeup", 0); + cont = NULL; + command_status += 2; + wake_up(&command_done); +} + +static struct cont_t wakeup_cont = { + .interrupt = empty, + .redo = do_wakeup, + .error = empty, + .done = (done_f) empty +}; + +static struct cont_t intr_cont = { + .interrupt = empty, + .redo = process_fd_request, + .error = empty, + .done = (done_f) empty +}; + +static int wait_til_done(void (*handler) (void), int interruptible) +{ + int ret; + + schedule_bh(handler); + + if (command_status < 2 && NO_SIGNAL) { + DECLARE_WAITQUEUE(wait, current); + + add_wait_queue(&command_done, &wait); + for (;;) { + set_current_state(interruptible ? + TASK_INTERRUPTIBLE : + TASK_UNINTERRUPTIBLE); + + if (command_status >= 2 || !NO_SIGNAL) + break; + + is_alive("wait_til_done"); + + schedule(); + } + + set_current_state(TASK_RUNNING); + remove_wait_queue(&command_done, &wait); + } + + if (command_status < 2) { + cancel_activity(); + cont = &intr_cont; + reset_fdc(); + return -EINTR; + } + + if (FDCS->reset) + command_status = FD_COMMAND_ERROR; + if (command_status == FD_COMMAND_OKAY) + ret = 0; + else + ret = -EIO; + command_status = FD_COMMAND_NONE; + return ret; +} + +static void generic_done(int result) +{ + command_status = result; + cont = &wakeup_cont; +} + +static void generic_success(void) +{ + cont->done(1); +} + +static void generic_failure(void) +{ + cont->done(0); +} + +static void success_and_wakeup(void) +{ + generic_success(); + cont->redo(); +} + +/* + * formatting and rw support. + * ========================== + */ + +static int next_valid_format(void) +{ + int probed_format; + + probed_format = DRS->probed_format; + while (1) { + if (probed_format >= 8 || !DP->autodetect[probed_format]) { + DRS->probed_format = 0; + return 1; + } + if (floppy_type[DP->autodetect[probed_format]].sect) { + DRS->probed_format = probed_format; + return 0; + } + probed_format++; + } +} + +static void bad_flp_intr(void) +{ + int err_count; + + if (probing) { + DRS->probed_format++; + if (!next_valid_format()) + return; + } + err_count = ++(*errors); + INFBOUND(DRWE->badness, err_count); + if (err_count > DP->max_errors.abort) + cont->done(0); + if (err_count > DP->max_errors.reset) + FDCS->reset = 1; + else if (err_count > DP->max_errors.recal) + DRS->track = NEED_2_RECAL; +} + +static void set_floppy(int drive) +{ + int type = ITYPE(UDRS->fd_device); + if (type) + _floppy = floppy_type + type; + else + _floppy = current_type[drive]; +} + +/* + * formatting support. + * =================== + */ +static void format_interrupt(void) +{ + switch (interpret_errors()) { + case 1: + cont->error(); + case 2: + break; + case 0: + cont->done(1); + } + cont->redo(); +} + +#define CODE2SIZE (ssize = ((1 << SIZECODE) + 3) >> 2) +#define FM_MODE(x,y) ((y) & ~(((x)->rate & 0x80) >>1)) +#define CT(x) ((x) | 0xc0) +static void setup_format_params(int track) +{ + struct fparm { + unsigned char track, head, sect, size; + } *here = (struct fparm *)floppy_track_buffer; + int il, n; + int count, head_shift, track_shift; + + raw_cmd = &default_raw_cmd; + raw_cmd->track = track; + + raw_cmd->flags = FD_RAW_WRITE | FD_RAW_INTR | FD_RAW_SPIN | + FD_RAW_NEED_DISK | FD_RAW_NEED_SEEK; + raw_cmd->rate = _floppy->rate & 0x43; + raw_cmd->cmd_count = NR_F; + COMMAND = FM_MODE(_floppy, FD_FORMAT); + DR_SELECT = UNIT(current_drive) + PH_HEAD(_floppy, format_req.head); + F_SIZECODE = FD_SIZECODE(_floppy); + F_SECT_PER_TRACK = _floppy->sect << 2 >> F_SIZECODE; + F_GAP = _floppy->fmt_gap; + F_FILL = FD_FILL_BYTE; + + raw_cmd->kernel_data = floppy_track_buffer; + raw_cmd->length = 4 * F_SECT_PER_TRACK; + + /* allow for about 30ms for data transport per track */ + head_shift = (F_SECT_PER_TRACK + 5) / 6; + + /* a ``cylinder'' is two tracks plus a little stepping time */ + track_shift = 2 * head_shift + 3; + + /* position of logical sector 1 on this track */ + n = (track_shift * format_req.track + head_shift * format_req.head) + % F_SECT_PER_TRACK; + + /* determine interleave */ + il = 1; + if (_floppy->fmt_gap < 0x22) + il++; + + /* initialize field */ + for (count = 0; count < F_SECT_PER_TRACK; ++count) { + here[count].track = format_req.track; + here[count].head = format_req.head; + here[count].sect = 0; + here[count].size = F_SIZECODE; + } + /* place logical sectors */ + for (count = 1; count <= F_SECT_PER_TRACK; ++count) { + here[n].sect = count; + n = (n + il) % F_SECT_PER_TRACK; + if (here[n].sect) { /* sector busy, find next free sector */ + ++n; + if (n >= F_SECT_PER_TRACK) { + n -= F_SECT_PER_TRACK; + while (here[n].sect) + ++n; + } + } + } + if (_floppy->stretch & FD_ZEROBASED) { + for (count = 0; count < F_SECT_PER_TRACK; count++) + here[count].sect--; + } +} + +static void redo_format(void) +{ + buffer_track = -1; + setup_format_params(format_req.track << STRETCH(_floppy)); + floppy_start(); + debugt("queue format request"); +} + +static struct cont_t format_cont = { + .interrupt = format_interrupt, + .redo = redo_format, + .error = bad_flp_intr, + .done = generic_done +}; + +static int do_format(int drive, struct format_descr *tmp_format_req) +{ + int ret; + + LOCK_FDC(drive, 1); + set_floppy(drive); + if (!_floppy || + _floppy->track > DP->tracks || + tmp_format_req->track >= _floppy->track || + tmp_format_req->head >= _floppy->head || + (_floppy->sect << 2) % (1 << FD_SIZECODE(_floppy)) || + !_floppy->fmt_gap) { + process_fd_request(); + return -EINVAL; + } + format_req = *tmp_format_req; + format_errors = 0; + cont = &format_cont; + errors = &format_errors; + IWAIT(redo_format); + process_fd_request(); + return ret; +} + +/* + * Buffer read/write and support + * ============================= + */ + +static void floppy_end_request(struct request *req, int uptodate) +{ + unsigned int nr_sectors = current_count_sectors; + + /* current_count_sectors can be zero if transfer failed */ + if (!uptodate) + nr_sectors = req->current_nr_sectors; + if (end_that_request_first(req, uptodate, nr_sectors)) + return; + add_disk_randomness(req->rq_disk); + floppy_off((long)req->rq_disk->private_data); + blkdev_dequeue_request(req); + end_that_request_last(req); + + /* We're done with the request */ + current_req = NULL; +} + +/* new request_done. Can handle physical sectors which are smaller than a + * logical buffer */ +static void request_done(int uptodate) +{ + struct request_queue *q = floppy_queue; + struct request *req = current_req; + unsigned long flags; + int block; + + probing = 0; + reschedule_timeout(MAXTIMEOUT, "request done %d", uptodate); + + if (!req) { + printk("floppy.c: no request in request_done\n"); + return; + } + + if (uptodate) { + /* maintain values for invalidation on geometry + * change */ + block = current_count_sectors + req->sector; + INFBOUND(DRS->maxblock, block); + if (block > _floppy->sect) + DRS->maxtrack = 1; + + /* unlock chained buffers */ + spin_lock_irqsave(q->queue_lock, flags); + floppy_end_request(req, 1); + spin_unlock_irqrestore(q->queue_lock, flags); + } else { + if (rq_data_dir(req) == WRITE) { + /* record write error information */ + DRWE->write_errors++; + if (DRWE->write_errors == 1) { + DRWE->first_error_sector = req->sector; + DRWE->first_error_generation = DRS->generation; + } + DRWE->last_error_sector = req->sector; + DRWE->last_error_generation = DRS->generation; + } + spin_lock_irqsave(q->queue_lock, flags); + floppy_end_request(req, 0); + spin_unlock_irqrestore(q->queue_lock, flags); + } +} + +/* Interrupt handler evaluating the result of the r/w operation */ +static void rw_interrupt(void) +{ + int nr_sectors, ssize, eoc, heads; + + if (R_HEAD >= 2) { + /* some Toshiba floppy controllers occasionnally seem to + * return bogus interrupts after read/write operations, which + * can be recognized by a bad head number (>= 2) */ + return; + } + + if (!DRS->first_read_date) + DRS->first_read_date = jiffies; + + nr_sectors = 0; + CODE2SIZE; + + if (ST1 & ST1_EOC) + eoc = 1; + else + eoc = 0; + + if (COMMAND & 0x80) + heads = 2; + else + heads = 1; + + nr_sectors = (((R_TRACK - TRACK) * heads + + R_HEAD - HEAD) * SECT_PER_TRACK + + R_SECTOR - SECTOR + eoc) << SIZECODE >> 2; + +#ifdef FLOPPY_SANITY_CHECK + if (nr_sectors / ssize > + (in_sector_offset + current_count_sectors + ssize - 1) / ssize) { + DPRINT("long rw: %x instead of %lx\n", + nr_sectors, current_count_sectors); + printk("rs=%d s=%d\n", R_SECTOR, SECTOR); + printk("rh=%d h=%d\n", R_HEAD, HEAD); + printk("rt=%d t=%d\n", R_TRACK, TRACK); + printk("heads=%d eoc=%d\n", heads, eoc); + printk("spt=%d st=%d ss=%d\n", SECT_PER_TRACK, + fsector_t, ssize); + printk("in_sector_offset=%d\n", in_sector_offset); + } +#endif + + nr_sectors -= in_sector_offset; + INFBOUND(nr_sectors, 0); + SUPBOUND(current_count_sectors, nr_sectors); + + switch (interpret_errors()) { + case 2: + cont->redo(); + return; + case 1: + if (!current_count_sectors) { + cont->error(); + cont->redo(); + return; + } + break; + case 0: + if (!current_count_sectors) { + cont->redo(); + return; + } + current_type[current_drive] = _floppy; + floppy_sizes[TOMINOR(current_drive)] = _floppy->size; + break; + } + + if (probing) { + if (DP->flags & FTD_MSG) + DPRINT("Auto-detected floppy type %s in fd%d\n", + _floppy->name, current_drive); + current_type[current_drive] = _floppy; + floppy_sizes[TOMINOR(current_drive)] = _floppy->size; + probing = 0; + } + + if (CT(COMMAND) != FD_READ || + raw_cmd->kernel_data == current_req->buffer) { + /* transfer directly from buffer */ + cont->done(1); + } else if (CT(COMMAND) == FD_READ) { + buffer_track = raw_cmd->track; + buffer_drive = current_drive; + INFBOUND(buffer_max, nr_sectors + fsector_t); + } + cont->redo(); +} + +/* Compute maximal contiguous buffer size. */ +static int buffer_chain_size(void) +{ + struct bio *bio; + struct bio_vec *bv; + int size, i; + char *base; + + base = bio_data(current_req->bio); + size = 0; + + rq_for_each_bio(bio, current_req) { + bio_for_each_segment(bv, bio, i) { + if (page_address(bv->bv_page) + bv->bv_offset != + base + size) + break; + + size += bv->bv_len; + } + } + + return size >> 9; +} + +/* Compute the maximal transfer size */ +static int transfer_size(int ssize, int max_sector, int max_size) +{ + SUPBOUND(max_sector, fsector_t + max_size); + + /* alignment */ + max_sector -= (max_sector % _floppy->sect) % ssize; + + /* transfer size, beginning not aligned */ + current_count_sectors = max_sector - fsector_t; + + return max_sector; +} + +/* + * Move data from/to the track buffer to/from the buffer cache. + */ +static void copy_buffer(int ssize, int max_sector, int max_sector_2) +{ + int remaining; /* number of transferred 512-byte sectors */ + struct bio_vec *bv; + struct bio *bio; + char *buffer, *dma_buffer; + int size, i; + + max_sector = transfer_size(ssize, + min(max_sector, max_sector_2), + current_req->nr_sectors); + + if (current_count_sectors <= 0 && CT(COMMAND) == FD_WRITE && + buffer_max > fsector_t + current_req->nr_sectors) + current_count_sectors = min_t(int, buffer_max - fsector_t, + current_req->nr_sectors); + + remaining = current_count_sectors << 9; +#ifdef FLOPPY_SANITY_CHECK + if ((remaining >> 9) > current_req->nr_sectors && + CT(COMMAND) == FD_WRITE) { + DPRINT("in copy buffer\n"); + printk("current_count_sectors=%ld\n", current_count_sectors); + printk("remaining=%d\n", remaining >> 9); + printk("current_req->nr_sectors=%ld\n", + current_req->nr_sectors); + printk("current_req->current_nr_sectors=%u\n", + current_req->current_nr_sectors); + printk("max_sector=%d\n", max_sector); + printk("ssize=%d\n", ssize); + } +#endif + + buffer_max = max(max_sector, buffer_max); + + dma_buffer = floppy_track_buffer + ((fsector_t - buffer_min) << 9); + + size = current_req->current_nr_sectors << 9; + + rq_for_each_bio(bio, current_req) { + bio_for_each_segment(bv, bio, i) { + if (!remaining) + break; + + size = bv->bv_len; + SUPBOUND(size, remaining); + + buffer = page_address(bv->bv_page) + bv->bv_offset; +#ifdef FLOPPY_SANITY_CHECK + if (dma_buffer + size > + floppy_track_buffer + (max_buffer_sectors << 10) || + dma_buffer < floppy_track_buffer) { + DPRINT("buffer overrun in copy buffer %d\n", + (int)((floppy_track_buffer - + dma_buffer) >> 9)); + printk("fsector_t=%d buffer_min=%d\n", + fsector_t, buffer_min); + printk("current_count_sectors=%ld\n", + current_count_sectors); + if (CT(COMMAND) == FD_READ) + printk("read\n"); + if (CT(COMMAND) == FD_WRITE) + printk("write\n"); + break; + } + if (((unsigned long)buffer) % 512) + DPRINT("%p buffer not aligned\n", buffer); +#endif + if (CT(COMMAND) == FD_READ) + memcpy(buffer, dma_buffer, size); + else + memcpy(dma_buffer, buffer, size); + + remaining -= size; + dma_buffer += size; + } + } +#ifdef FLOPPY_SANITY_CHECK + if (remaining) { + if (remaining > 0) + max_sector -= remaining >> 9; + DPRINT("weirdness: remaining %d\n", remaining >> 9); + } +#endif +} + +#if 0 +static inline int check_dma_crossing(char *start, + unsigned long length, char *message) +{ + if (CROSS_64KB(start, length)) { + printk("DMA xfer crosses 64KB boundary in %s %p-%p\n", + message, start, start + length); + return 1; + } else + return 0; +} +#endif + +/* work around a bug in pseudo DMA + * (on some FDCs) pseudo DMA does not stop when the CPU stops + * sending data. Hence we need a different way to signal the + * transfer length: We use SECT_PER_TRACK. Unfortunately, this + * does not work with MT, hence we can only transfer one head at + * a time + */ +static void virtualdmabug_workaround(void) +{ + int hard_sectors, end_sector; + + if (CT(COMMAND) == FD_WRITE) { + COMMAND &= ~0x80; /* switch off multiple track mode */ + + hard_sectors = raw_cmd->length >> (7 + SIZECODE); + end_sector = SECTOR + hard_sectors - 1; +#ifdef FLOPPY_SANITY_CHECK + if (end_sector > SECT_PER_TRACK) { + printk("too many sectors %d > %d\n", + end_sector, SECT_PER_TRACK); + return; + } +#endif + SECT_PER_TRACK = end_sector; /* make sure SECT_PER_TRACK points + * to end of transfer */ + } +} + +/* + * Formulate a read/write request. + * this routine decides where to load the data (directly to buffer, or to + * tmp floppy area), how much data to load (the size of the buffer, the whole + * track, or a single sector) + * All floppy_track_buffer handling goes in here. If we ever add track buffer + * allocation on the fly, it should be done here. No other part should need + * modification. + */ + +static int make_raw_rw_request(void) +{ + int aligned_sector_t; + int max_sector, max_size, tracksize, ssize; + + if (max_buffer_sectors == 0) { + printk("VFS: Block I/O scheduled on unopened device\n"); + return 0; + } + + set_fdc((long)current_req->rq_disk->private_data); + + raw_cmd = &default_raw_cmd; + raw_cmd->flags = FD_RAW_SPIN | FD_RAW_NEED_DISK | FD_RAW_NEED_DISK | + FD_RAW_NEED_SEEK; + raw_cmd->cmd_count = NR_RW; + if (rq_data_dir(current_req) == READ) { + raw_cmd->flags |= FD_RAW_READ; + COMMAND = FM_MODE(_floppy, FD_READ); + } else if (rq_data_dir(current_req) == WRITE) { + raw_cmd->flags |= FD_RAW_WRITE; + COMMAND = FM_MODE(_floppy, FD_WRITE); + } else { + DPRINT("make_raw_rw_request: unknown command\n"); + return 0; + } + + max_sector = _floppy->sect * _floppy->head; + + TRACK = (int)current_req->sector / max_sector; + fsector_t = (int)current_req->sector % max_sector; + if (_floppy->track && TRACK >= _floppy->track) { + if (current_req->current_nr_sectors & 1) { + current_count_sectors = 1; + return 1; + } else + return 0; + } + HEAD = fsector_t / _floppy->sect; + + if (((_floppy->stretch & (FD_SWAPSIDES | FD_ZEROBASED)) || + TESTF(FD_NEED_TWADDLE)) && fsector_t < _floppy->sect) + max_sector = _floppy->sect; + + /* 2M disks have phantom sectors on the first track */ + if ((_floppy->rate & FD_2M) && (!TRACK) && (!HEAD)) { + max_sector = 2 * _floppy->sect / 3; + if (fsector_t >= max_sector) { + current_count_sectors = + min_t(int, _floppy->sect - fsector_t, + current_req->nr_sectors); + return 1; + } + SIZECODE = 2; + } else + SIZECODE = FD_SIZECODE(_floppy); + raw_cmd->rate = _floppy->rate & 0x43; + if ((_floppy->rate & FD_2M) && (TRACK || HEAD) && raw_cmd->rate == 2) + raw_cmd->rate = 1; + + if (SIZECODE) + SIZECODE2 = 0xff; + else + SIZECODE2 = 0x80; + raw_cmd->track = TRACK << STRETCH(_floppy); + DR_SELECT = UNIT(current_drive) + PH_HEAD(_floppy, HEAD); + GAP = _floppy->gap; + CODE2SIZE; + SECT_PER_TRACK = _floppy->sect << 2 >> SIZECODE; + SECTOR = ((fsector_t % _floppy->sect) << 2 >> SIZECODE) + + ((_floppy->stretch & FD_ZEROBASED) ? 0 : 1); + + /* tracksize describes the size which can be filled up with sectors + * of size ssize. + */ + tracksize = _floppy->sect - _floppy->sect % ssize; + if (tracksize < _floppy->sect) { + SECT_PER_TRACK++; + if (tracksize <= fsector_t % _floppy->sect) + SECTOR--; + + /* if we are beyond tracksize, fill up using smaller sectors */ + while (tracksize <= fsector_t % _floppy->sect) { + while (tracksize + ssize > _floppy->sect) { + SIZECODE--; + ssize >>= 1; + } + SECTOR++; + SECT_PER_TRACK++; + tracksize += ssize; + } + max_sector = HEAD * _floppy->sect + tracksize; + } else if (!TRACK && !HEAD && !(_floppy->rate & FD_2M) && probing) { + max_sector = _floppy->sect; + } else if (!HEAD && CT(COMMAND) == FD_WRITE) { + /* for virtual DMA bug workaround */ + max_sector = _floppy->sect; + } + + in_sector_offset = (fsector_t % _floppy->sect) % ssize; + aligned_sector_t = fsector_t - in_sector_offset; + max_size = current_req->nr_sectors; + if ((raw_cmd->track == buffer_track) && + (current_drive == buffer_drive) && + (fsector_t >= buffer_min) && (fsector_t < buffer_max)) { + /* data already in track buffer */ + if (CT(COMMAND) == FD_READ) { + copy_buffer(1, max_sector, buffer_max); + return 1; + } + } else if (in_sector_offset || current_req->nr_sectors < ssize) { + if (CT(COMMAND) == FD_WRITE) { + if (fsector_t + current_req->nr_sectors > ssize && + fsector_t + current_req->nr_sectors < ssize + ssize) + max_size = ssize + ssize; + else + max_size = ssize; + } + raw_cmd->flags &= ~FD_RAW_WRITE; + raw_cmd->flags |= FD_RAW_READ; + COMMAND = FM_MODE(_floppy, FD_READ); + } else if ((unsigned long)current_req->buffer < MAX_DMA_ADDRESS) { + unsigned long dma_limit; + int direct, indirect; + + indirect = + transfer_size(ssize, max_sector, + max_buffer_sectors * 2) - fsector_t; + + /* + * Do NOT use minimum() here---MAX_DMA_ADDRESS is 64 bits wide + * on a 64 bit machine! + */ + max_size = buffer_chain_size(); + dma_limit = + (MAX_DMA_ADDRESS - + ((unsigned long)current_req->buffer)) >> 9; + if ((unsigned long)max_size > dma_limit) { + max_size = dma_limit; + } + /* 64 kb boundaries */ + if (CROSS_64KB(current_req->buffer, max_size << 9)) + max_size = (K_64 - + ((unsigned long)current_req->buffer) % + K_64) >> 9; + direct = transfer_size(ssize, max_sector, max_size) - fsector_t; + /* + * We try to read tracks, but if we get too many errors, we + * go back to reading just one sector at a time. + * + * This means we should be able to read a sector even if there + * are other bad sectors on this track. + */ + if (!direct || + (indirect * 2 > direct * 3 && + *errors < DP->max_errors.read_track && + /*!TESTF(FD_NEED_TWADDLE) && */ + ((!probing + || (DP->read_track & (1 << DRS->probed_format)))))) { + max_size = current_req->nr_sectors; + } else { + raw_cmd->kernel_data = current_req->buffer; + raw_cmd->length = current_count_sectors << 9; + if (raw_cmd->length == 0) { + DPRINT + ("zero dma transfer attempted from make_raw_request\n"); + DPRINT("indirect=%d direct=%d fsector_t=%d", + indirect, direct, fsector_t); + return 0; + } +/* check_dma_crossing(raw_cmd->kernel_data, + raw_cmd->length, + "end of make_raw_request [1]");*/ + + virtualdmabug_workaround(); + return 2; + } + } + + if (CT(COMMAND) == FD_READ) + max_size = max_sector; /* unbounded */ + + /* claim buffer track if needed */ + if (buffer_track != raw_cmd->track || /* bad track */ + buffer_drive != current_drive || /* bad drive */ + fsector_t > buffer_max || + fsector_t < buffer_min || + ((CT(COMMAND) == FD_READ || + (!in_sector_offset && current_req->nr_sectors >= ssize)) && + max_sector > 2 * max_buffer_sectors + buffer_min && + max_size + fsector_t > 2 * max_buffer_sectors + buffer_min) + /* not enough space */ + ) { + buffer_track = -1; + buffer_drive = current_drive; + buffer_max = buffer_min = aligned_sector_t; + } + raw_cmd->kernel_data = floppy_track_buffer + + ((aligned_sector_t - buffer_min) << 9); + + if (CT(COMMAND) == FD_WRITE) { + /* copy write buffer to track buffer. + * if we get here, we know that the write + * is either aligned or the data already in the buffer + * (buffer will be overwritten) */ +#ifdef FLOPPY_SANITY_CHECK + if (in_sector_offset && buffer_track == -1) + DPRINT("internal error offset !=0 on write\n"); +#endif + buffer_track = raw_cmd->track; + buffer_drive = current_drive; + copy_buffer(ssize, max_sector, + 2 * max_buffer_sectors + buffer_min); + } else + transfer_size(ssize, max_sector, + 2 * max_buffer_sectors + buffer_min - + aligned_sector_t); + + /* round up current_count_sectors to get dma xfer size */ + raw_cmd->length = in_sector_offset + current_count_sectors; + raw_cmd->length = ((raw_cmd->length - 1) | (ssize - 1)) + 1; + raw_cmd->length <<= 9; +#ifdef FLOPPY_SANITY_CHECK + /*check_dma_crossing(raw_cmd->kernel_data, raw_cmd->length, + "end of make_raw_request"); */ + if ((raw_cmd->length < current_count_sectors << 9) || + (raw_cmd->kernel_data != current_req->buffer && + CT(COMMAND) == FD_WRITE && + (aligned_sector_t + (raw_cmd->length >> 9) > buffer_max || + aligned_sector_t < buffer_min)) || + raw_cmd->length % (128 << SIZECODE) || + raw_cmd->length <= 0 || current_count_sectors <= 0) { + DPRINT("fractionary current count b=%lx s=%lx\n", + raw_cmd->length, current_count_sectors); + if (raw_cmd->kernel_data != current_req->buffer) + printk("addr=%d, length=%ld\n", + (int)((raw_cmd->kernel_data - + floppy_track_buffer) >> 9), + current_count_sectors); + printk("st=%d ast=%d mse=%d msi=%d\n", + fsector_t, aligned_sector_t, max_sector, max_size); + printk("ssize=%x SIZECODE=%d\n", ssize, SIZECODE); + printk("command=%x SECTOR=%d HEAD=%d, TRACK=%d\n", + COMMAND, SECTOR, HEAD, TRACK); + printk("buffer drive=%d\n", buffer_drive); + printk("buffer track=%d\n", buffer_track); + printk("buffer_min=%d\n", buffer_min); + printk("buffer_max=%d\n", buffer_max); + return 0; + } + + if (raw_cmd->kernel_data != current_req->buffer) { + if (raw_cmd->kernel_data < floppy_track_buffer || + current_count_sectors < 0 || + raw_cmd->length < 0 || + raw_cmd->kernel_data + raw_cmd->length > + floppy_track_buffer + (max_buffer_sectors << 10)) { + DPRINT("buffer overrun in schedule dma\n"); + printk("fsector_t=%d buffer_min=%d current_count=%ld\n", + fsector_t, buffer_min, raw_cmd->length >> 9); + printk("current_count_sectors=%ld\n", + current_count_sectors); + if (CT(COMMAND) == FD_READ) + printk("read\n"); + if (CT(COMMAND) == FD_WRITE) + printk("write\n"); + return 0; + } + } else if (raw_cmd->length > current_req->nr_sectors << 9 || + current_count_sectors > current_req->nr_sectors) { + DPRINT("buffer overrun in direct transfer\n"); + return 0; + } else if (raw_cmd->length < current_count_sectors << 9) { + DPRINT("more sectors than bytes\n"); + printk("bytes=%ld\n", raw_cmd->length >> 9); + printk("sectors=%ld\n", current_count_sectors); + } + if (raw_cmd->length == 0) { + DPRINT("zero dma transfer attempted from make_raw_request\n"); + return 0; + } +#endif + + virtualdmabug_workaround(); + return 2; +} + +static void redo_fd_request(void) +{ +#define REPEAT {request_done(0); continue; } + int drive; + int tmp; + + lastredo = jiffies; + if (current_drive < N_DRIVE) + floppy_off(current_drive); + + for (;;) { + if (!current_req) { + struct request *req; + + spin_lock_irq(floppy_queue->queue_lock); + req = elv_next_request(floppy_queue); + spin_unlock_irq(floppy_queue->queue_lock); + if (!req) { + do_floppy = NULL; + unlock_fdc(); + return; + } + current_req = req; + } + drive = (long)current_req->rq_disk->private_data; + set_fdc(drive); + reschedule_timeout(current_reqD, "redo fd request", 0); + + set_floppy(drive); + raw_cmd = &default_raw_cmd; + raw_cmd->flags = 0; + if (start_motor(redo_fd_request)) + return; + disk_change(current_drive); + if (test_bit(current_drive, &fake_change) || + TESTF(FD_DISK_CHANGED)) { + DPRINT("disk absent or changed during operation\n"); + REPEAT; + } + if (!_floppy) { /* Autodetection */ + if (!probing) { + DRS->probed_format = 0; + if (next_valid_format()) { + DPRINT("no autodetectable formats\n"); + _floppy = NULL; + REPEAT; + } + } + probing = 1; + _floppy = + floppy_type + DP->autodetect[DRS->probed_format]; + } else + probing = 0; + errors = &(current_req->errors); + tmp = make_raw_rw_request(); + if (tmp < 2) { + request_done(tmp); + continue; + } + + if (TESTF(FD_NEED_TWADDLE)) + twaddle(); + schedule_bh(floppy_start); + debugt("queue fd request"); + return; + } +#undef REPEAT +} + +static struct cont_t rw_cont = { + .interrupt = rw_interrupt, + .redo = redo_fd_request, + .error = bad_flp_intr, + .done = request_done +}; + +static void process_fd_request(void) +{ + cont = &rw_cont; + schedule_bh(redo_fd_request); +} + +static void do_fd_request(request_queue_t * q) +{ + if (max_buffer_sectors == 0) { + printk("VFS: do_fd_request called on non-open device\n"); + return; + } + + if (usage_count == 0) { + printk("warning: usage count=0, current_req=%p exiting\n", + current_req); + printk("sect=%ld flags=%lx\n", (long)current_req->sector, + current_req->flags); + return; + } + if (test_bit(0, &fdc_busy)) { + /* fdc busy, this new request will be treated when the + current one is done */ + is_alive("do fd request, old request running"); + return; + } + lock_fdc(MAXTIMEOUT, 0); + process_fd_request(); + is_alive("do fd request"); +} + +static struct cont_t poll_cont = { + .interrupt = success_and_wakeup, + .redo = floppy_ready, + .error = generic_failure, + .done = generic_done +}; + +static int poll_drive(int interruptible, int flag) +{ + int ret; + /* no auto-sense, just clear dcl */ + raw_cmd = &default_raw_cmd; + raw_cmd->flags = flag; + raw_cmd->track = 0; + raw_cmd->cmd_count = 0; + cont = &poll_cont; +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT("setting NEWCHANGE in poll_drive\n"); + } +#endif + SETF(FD_DISK_NEWCHANGE); + WAIT(floppy_ready); + return ret; +} + +/* + * User triggered reset + * ==================== + */ + +static void reset_intr(void) +{ + printk("weird, reset interrupt called\n"); +} + +static struct cont_t reset_cont = { + .interrupt = reset_intr, + .redo = success_and_wakeup, + .error = generic_failure, + .done = generic_done +}; + +static int user_reset_fdc(int drive, int arg, int interruptible) +{ + int ret; + + ret = 0; + LOCK_FDC(drive, interruptible); + if (arg == FD_RESET_ALWAYS) + FDCS->reset = 1; + if (FDCS->reset) { + cont = &reset_cont; + WAIT(reset_fdc); + } + process_fd_request(); + return ret; +} + +/* + * Misc Ioctl's and support + * ======================== + */ +static inline int fd_copyout(void __user *param, const void *address, + unsigned long size) +{ + return copy_to_user(param, address, size) ? -EFAULT : 0; +} + +static inline int fd_copyin(void __user *param, void *address, unsigned long size) +{ + return copy_from_user(address, param, size) ? -EFAULT : 0; +} + +#define _COPYOUT(x) (copy_to_user((void __user *)param, &(x), sizeof(x)) ? -EFAULT : 0) +#define _COPYIN(x) (copy_from_user(&(x), (void __user *)param, sizeof(x)) ? -EFAULT : 0) + +#define COPYOUT(x) ECALL(_COPYOUT(x)) +#define COPYIN(x) ECALL(_COPYIN(x)) + +static inline const char *drive_name(int type, int drive) +{ + struct floppy_struct *floppy; + + if (type) + floppy = floppy_type + type; + else { + if (UDP->native_format) + floppy = floppy_type + UDP->native_format; + else + return "(null)"; + } + if (floppy->name) + return floppy->name; + else + return "(null)"; +} + +/* raw commands */ +static void raw_cmd_done(int flag) +{ + int i; + + if (!flag) { + raw_cmd->flags |= FD_RAW_FAILURE; + raw_cmd->flags |= FD_RAW_HARDFAILURE; + } else { + raw_cmd->reply_count = inr; + if (raw_cmd->reply_count > MAX_REPLIES) + raw_cmd->reply_count = 0; + for (i = 0; i < raw_cmd->reply_count; i++) + raw_cmd->reply[i] = reply_buffer[i]; + + if (raw_cmd->flags & (FD_RAW_READ | FD_RAW_WRITE)) { + unsigned long flags; + flags = claim_dma_lock(); + raw_cmd->length = fd_get_dma_residue(); + release_dma_lock(flags); + } + + if ((raw_cmd->flags & FD_RAW_SOFTFAILURE) && + (!raw_cmd->reply_count || (raw_cmd->reply[0] & 0xc0))) + raw_cmd->flags |= FD_RAW_FAILURE; + + if (disk_change(current_drive)) + raw_cmd->flags |= FD_RAW_DISK_CHANGE; + else + raw_cmd->flags &= ~FD_RAW_DISK_CHANGE; + if (raw_cmd->flags & FD_RAW_NO_MOTOR_AFTER) + motor_off_callback(current_drive); + + if (raw_cmd->next && + (!(raw_cmd->flags & FD_RAW_FAILURE) || + !(raw_cmd->flags & FD_RAW_STOP_IF_FAILURE)) && + ((raw_cmd->flags & FD_RAW_FAILURE) || + !(raw_cmd->flags & FD_RAW_STOP_IF_SUCCESS))) { + raw_cmd = raw_cmd->next; + return; + } + } + generic_done(flag); +} + +static struct cont_t raw_cmd_cont = { + .interrupt = success_and_wakeup, + .redo = floppy_start, + .error = generic_failure, + .done = raw_cmd_done +}; + +static inline int raw_cmd_copyout(int cmd, char __user *param, + struct floppy_raw_cmd *ptr) +{ + int ret; + + while (ptr) { + COPYOUT(*ptr); + param += sizeof(struct floppy_raw_cmd); + if ((ptr->flags & FD_RAW_READ) && ptr->buffer_length) { + if (ptr->length >= 0 + && ptr->length <= ptr->buffer_length) + ECALL(fd_copyout + (ptr->data, ptr->kernel_data, + ptr->buffer_length - ptr->length)); + } + ptr = ptr->next; + } + return 0; +} + +static void raw_cmd_free(struct floppy_raw_cmd **ptr) +{ + struct floppy_raw_cmd *next, *this; + + this = *ptr; + *ptr = NULL; + while (this) { + if (this->buffer_length) { + fd_dma_mem_free((unsigned long)this->kernel_data, + this->buffer_length); + this->buffer_length = 0; + } + next = this->next; + kfree(this); + this = next; + } +} + +static inline int raw_cmd_copyin(int cmd, char __user *param, + struct floppy_raw_cmd **rcmd) +{ + struct floppy_raw_cmd *ptr; + int ret; + int i; + + *rcmd = NULL; + while (1) { + ptr = (struct floppy_raw_cmd *) + kmalloc(sizeof(struct floppy_raw_cmd), GFP_USER); + if (!ptr) + return -ENOMEM; + *rcmd = ptr; + COPYIN(*ptr); + ptr->next = NULL; + ptr->buffer_length = 0; + param += sizeof(struct floppy_raw_cmd); + if (ptr->cmd_count > 33) + /* the command may now also take up the space + * initially intended for the reply & the + * reply count. Needed for long 82078 commands + * such as RESTORE, which takes ... 17 command + * bytes. Murphy's law #137: When you reserve + * 16 bytes for a structure, you'll one day + * discover that you really need 17... + */ + return -EINVAL; + + for (i = 0; i < 16; i++) + ptr->reply[i] = 0; + ptr->resultcode = 0; + ptr->kernel_data = NULL; + + if (ptr->flags & (FD_RAW_READ | FD_RAW_WRITE)) { + if (ptr->length <= 0) + return -EINVAL; + ptr->kernel_data = + (char *)fd_dma_mem_alloc(ptr->length); + fallback_on_nodma_alloc(&ptr->kernel_data, ptr->length); + if (!ptr->kernel_data) + return -ENOMEM; + ptr->buffer_length = ptr->length; + } + if (ptr->flags & FD_RAW_WRITE) + ECALL(fd_copyin(ptr->data, ptr->kernel_data, + ptr->length)); + rcmd = &(ptr->next); + if (!(ptr->flags & FD_RAW_MORE)) + return 0; + ptr->rate &= 0x43; + } +} + +static int raw_cmd_ioctl(int cmd, void __user *param) +{ + int drive, ret, ret2; + struct floppy_raw_cmd *my_raw_cmd; + + if (FDCS->rawcmd <= 1) + FDCS->rawcmd = 1; + for (drive = 0; drive < N_DRIVE; drive++) { + if (FDC(drive) != fdc) + continue; + if (drive == current_drive) { + if (UDRS->fd_ref > 1) { + FDCS->rawcmd = 2; + break; + } + } else if (UDRS->fd_ref) { + FDCS->rawcmd = 2; + break; + } + } + + if (FDCS->reset) + return -EIO; + + ret = raw_cmd_copyin(cmd, param, &my_raw_cmd); + if (ret) { + raw_cmd_free(&my_raw_cmd); + return ret; + } + + raw_cmd = my_raw_cmd; + cont = &raw_cmd_cont; + ret = wait_til_done(floppy_start, 1); +#ifdef DCL_DEBUG + if (DP->flags & FD_DEBUG) { + DPRINT("calling disk change from raw_cmd ioctl\n"); + } +#endif + + if (ret != -EINTR && FDCS->reset) + ret = -EIO; + + DRS->track = NO_TRACK; + + ret2 = raw_cmd_copyout(cmd, param, my_raw_cmd); + if (!ret) + ret = ret2; + raw_cmd_free(&my_raw_cmd); + return ret; +} + +static int invalidate_drive(struct block_device *bdev) +{ + /* invalidate the buffer track to force a reread */ + set_bit((long)bdev->bd_disk->private_data, &fake_change); + process_fd_request(); + check_disk_change(bdev); + return 0; +} + +static inline int set_geometry(unsigned int cmd, struct floppy_struct *g, + int drive, int type, struct block_device *bdev) +{ + int cnt; + + /* sanity checking for parameters. */ + if (g->sect <= 0 || + g->head <= 0 || + g->track <= 0 || g->track > UDP->tracks >> STRETCH(g) || + /* check if reserved bits are set */ + (g->stretch & ~(FD_STRETCH | FD_SWAPSIDES | FD_ZEROBASED)) != 0) + return -EINVAL; + if (type) { + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + down(&open_lock); + LOCK_FDC(drive, 1); + floppy_type[type] = *g; + floppy_type[type].name = "user format"; + for (cnt = type << 2; cnt < (type << 2) + 4; cnt++) + floppy_sizes[cnt] = floppy_sizes[cnt + 0x80] = + floppy_type[type].size + 1; + process_fd_request(); + for (cnt = 0; cnt < N_DRIVE; cnt++) { + struct block_device *bdev = opened_bdev[cnt]; + if (!bdev || ITYPE(drive_state[cnt].fd_device) != type) + continue; + __invalidate_device(bdev, 0); + } + up(&open_lock); + } else { + int oldStretch; + LOCK_FDC(drive, 1); + if (cmd != FDDEFPRM) + /* notice a disk change immediately, else + * we lose our settings immediately*/ + CALL(poll_drive(1, FD_RAW_NEED_DISK)); + oldStretch = g->stretch; + user_params[drive] = *g; + if (buffer_drive == drive) + SUPBOUND(buffer_max, user_params[drive].sect); + current_type[drive] = &user_params[drive]; + floppy_sizes[drive] = user_params[drive].size; + if (cmd == FDDEFPRM) + DRS->keep_data = -1; + else + DRS->keep_data = 1; + /* invalidation. Invalidate only when needed, i.e. + * when there are already sectors in the buffer cache + * whose number will change. This is useful, because + * mtools often changes the geometry of the disk after + * looking at the boot block */ + if (DRS->maxblock > user_params[drive].sect || + DRS->maxtrack || + ((user_params[drive].sect ^ oldStretch) & + (FD_SWAPSIDES | FD_ZEROBASED))) + invalidate_drive(bdev); + else + process_fd_request(); + } + return 0; +} + +/* handle obsolete ioctl's */ +static int ioctl_table[] = { + FDCLRPRM, + FDSETPRM, + FDDEFPRM, + FDGETPRM, + FDMSGON, + FDMSGOFF, + FDFMTBEG, + FDFMTTRK, + FDFMTEND, + FDSETEMSGTRESH, + FDFLUSH, + FDSETMAXERRS, + FDGETMAXERRS, + FDGETDRVTYP, + FDSETDRVPRM, + FDGETDRVPRM, + FDGETDRVSTAT, + FDPOLLDRVSTAT, + FDRESET, + FDGETFDCSTAT, + FDWERRORCLR, + FDWERRORGET, + FDRAWCMD, + FDEJECT, + FDTWADDLE +}; + +static inline int normalize_ioctl(int *cmd, int *size) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(ioctl_table); i++) { + if ((*cmd & 0xffff) == (ioctl_table[i] & 0xffff)) { + *size = _IOC_SIZE(*cmd); + *cmd = ioctl_table[i]; + if (*size > _IOC_SIZE(*cmd)) { + printk("ioctl not yet supported\n"); + return -EFAULT; + } + return 0; + } + } + return -EINVAL; +} + +static int get_floppy_geometry(int drive, int type, struct floppy_struct **g) +{ + if (type) + *g = &floppy_type[type]; + else { + LOCK_FDC(drive, 0); + CALL(poll_drive(0, 0)); + process_fd_request(); + *g = current_type[drive]; + } + if (!*g) + return -ENODEV; + return 0; +} + +static int fd_ioctl(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long param) +{ +#define FD_IOCTL_ALLOWED ((filp) && (filp)->private_data) +#define OUT(c,x) case c: outparam = (const char *) (x); break +#define IN(c,x,tag) case c: *(x) = inparam. tag ; return 0 + + int drive = (long)inode->i_bdev->bd_disk->private_data; + int i, type = ITYPE(UDRS->fd_device); + int ret; + int size; + union inparam { + struct floppy_struct g; /* geometry */ + struct format_descr f; + struct floppy_max_errors max_errors; + struct floppy_drive_params dp; + } inparam; /* parameters coming from user space */ + const char *outparam; /* parameters passed back to user space */ + + /* convert compatibility eject ioctls into floppy eject ioctl. + * We do this in order to provide a means to eject floppy disks before + * installing the new fdutils package */ + if (cmd == CDROMEJECT || /* CD-ROM eject */ + cmd == 0x6470 /* SunOS floppy eject */ ) { + DPRINT("obsolete eject ioctl\n"); + DPRINT("please use floppycontrol --eject\n"); + cmd = FDEJECT; + } + + /* generic block device ioctls */ + switch (cmd) { + /* the following have been inspired by the corresponding + * code for other block devices. */ + struct floppy_struct *g; + case HDIO_GETGEO: + { + struct hd_geometry loc; + ECALL(get_floppy_geometry(drive, type, &g)); + loc.heads = g->head; + loc.sectors = g->sect; + loc.cylinders = g->track; + loc.start = 0; + return _COPYOUT(loc); + } + } + + /* convert the old style command into a new style command */ + if ((cmd & 0xff00) == 0x0200) { + ECALL(normalize_ioctl(&cmd, &size)); + } else + return -EINVAL; + + /* permission checks */ + if (((cmd & 0x40) && !FD_IOCTL_ALLOWED) || + ((cmd & 0x80) && !capable(CAP_SYS_ADMIN))) + return -EPERM; + + /* copyin */ + CLEARSTRUCT(&inparam); + if (_IOC_DIR(cmd) & _IOC_WRITE) + ECALL(fd_copyin((void __user *)param, &inparam, size)) + + switch (cmd) { + case FDEJECT: + if (UDRS->fd_ref != 1) + /* somebody else has this drive open */ + return -EBUSY; + LOCK_FDC(drive, 1); + + /* do the actual eject. Fails on + * non-Sparc architectures */ + ret = fd_eject(UNIT(drive)); + + USETF(FD_DISK_CHANGED); + USETF(FD_VERIFY); + process_fd_request(); + return ret; + case FDCLRPRM: + LOCK_FDC(drive, 1); + current_type[drive] = NULL; + floppy_sizes[drive] = MAX_DISK_SIZE << 1; + UDRS->keep_data = 0; + return invalidate_drive(inode->i_bdev); + case FDSETPRM: + case FDDEFPRM: + return set_geometry(cmd, &inparam.g, + drive, type, inode->i_bdev); + case FDGETPRM: + ECALL(get_floppy_geometry(drive, type, + (struct floppy_struct **) + &outparam)); + break; + + case FDMSGON: + UDP->flags |= FTD_MSG; + return 0; + case FDMSGOFF: + UDP->flags &= ~FTD_MSG; + return 0; + + case FDFMTBEG: + LOCK_FDC(drive, 1); + CALL(poll_drive(1, FD_RAW_NEED_DISK)); + ret = UDRS->flags; + process_fd_request(); + if (ret & FD_VERIFY) + return -ENODEV; + if (!(ret & FD_DISK_WRITABLE)) + return -EROFS; + return 0; + case FDFMTTRK: + if (UDRS->fd_ref != 1) + return -EBUSY; + return do_format(drive, &inparam.f); + case FDFMTEND: + case FDFLUSH: + LOCK_FDC(drive, 1); + return invalidate_drive(inode->i_bdev); + + case FDSETEMSGTRESH: + UDP->max_errors.reporting = + (unsigned short)(param & 0x0f); + return 0; + OUT(FDGETMAXERRS, &UDP->max_errors); + IN(FDSETMAXERRS, &UDP->max_errors, max_errors); + + case FDGETDRVTYP: + outparam = drive_name(type, drive); + SUPBOUND(size, strlen(outparam) + 1); + break; + + IN(FDSETDRVPRM, UDP, dp); + OUT(FDGETDRVPRM, UDP); + + case FDPOLLDRVSTAT: + LOCK_FDC(drive, 1); + CALL(poll_drive(1, FD_RAW_NEED_DISK)); + process_fd_request(); + /* fall through */ + OUT(FDGETDRVSTAT, UDRS); + + case FDRESET: + return user_reset_fdc(drive, (int)param, 1); + + OUT(FDGETFDCSTAT, UFDCS); + + case FDWERRORCLR: + CLEARSTRUCT(UDRWE); + return 0; + OUT(FDWERRORGET, UDRWE); + + case FDRAWCMD: + if (type) + return -EINVAL; + LOCK_FDC(drive, 1); + set_floppy(drive); + CALL(i = raw_cmd_ioctl(cmd, (void __user *)param)); + process_fd_request(); + return i; + + case FDTWADDLE: + LOCK_FDC(drive, 1); + twaddle(); + process_fd_request(); + return 0; + + default: + return -EINVAL; + } + + if (_IOC_DIR(cmd) & _IOC_READ) + return fd_copyout((void __user *)param, outparam, size); + else + return 0; +#undef OUT +#undef IN +} + +static void __init config_types(void) +{ + int first = 1; + int drive; + + /* read drive info out of physical CMOS */ + drive = 0; + if (!UDP->cmos) + UDP->cmos = FLOPPY0_TYPE; + drive = 1; + if (!UDP->cmos && FLOPPY1_TYPE) + UDP->cmos = FLOPPY1_TYPE; + + /* XXX */ + /* additional physical CMOS drive detection should go here */ + + for (drive = 0; drive < N_DRIVE; drive++) { + unsigned int type = UDP->cmos; + struct floppy_drive_params *params; + const char *name = NULL; + static char temparea[32]; + + if (type < NUMBER(default_drive_params)) { + params = &default_drive_params[type].params; + if (type) { + name = default_drive_params[type].name; + allowed_drive_mask |= 1 << drive; + } else + allowed_drive_mask &= ~(1 << drive); + } else { + params = &default_drive_params[0].params; + sprintf(temparea, "unknown type %d (usb?)", type); + name = temparea; + } + if (name) { + const char *prepend = ","; + if (first) { + prepend = KERN_INFO "Floppy drive(s):"; + first = 0; + } + printk("%s fd%d is %s", prepend, drive, name); + register_devfs_entries(drive); + } + *UDP = *params; + } + if (!first) + printk("\n"); +} + +static int floppy_release(struct inode *inode, struct file *filp) +{ + int drive = (long)inode->i_bdev->bd_disk->private_data; + + down(&open_lock); + if (UDRS->fd_ref < 0) + UDRS->fd_ref = 0; + else if (!UDRS->fd_ref--) { + DPRINT("floppy_release with fd_ref == 0"); + UDRS->fd_ref = 0; + } + if (!UDRS->fd_ref) + opened_bdev[drive] = NULL; + floppy_release_irq_and_dma(); + up(&open_lock); + return 0; +} + +/* + * floppy_open check for aliasing (/dev/fd0 can be the same as + * /dev/PS0 etc), and disallows simultaneous access to the same + * drive with different device numbers. + */ +static int floppy_open(struct inode *inode, struct file *filp) +{ + int drive = (long)inode->i_bdev->bd_disk->private_data; + int old_dev; + int try; + int res = -EBUSY; + char *tmp; + + filp->private_data = (void *)0; + down(&open_lock); + old_dev = UDRS->fd_device; + if (opened_bdev[drive] && opened_bdev[drive] != inode->i_bdev) + goto out2; + + if (!UDRS->fd_ref && (UDP->flags & FD_BROKEN_DCL)) { + USETF(FD_DISK_CHANGED); + USETF(FD_VERIFY); + } + + if (UDRS->fd_ref == -1 || (UDRS->fd_ref && (filp->f_flags & O_EXCL))) + goto out2; + + if (floppy_grab_irq_and_dma()) + goto out2; + + if (filp->f_flags & O_EXCL) + UDRS->fd_ref = -1; + else + UDRS->fd_ref++; + + opened_bdev[drive] = inode->i_bdev; + + res = -ENXIO; + + if (!floppy_track_buffer) { + /* if opening an ED drive, reserve a big buffer, + * else reserve a small one */ + if ((UDP->cmos == 6) || (UDP->cmos == 5)) + try = 64; /* Only 48 actually useful */ + else + try = 32; /* Only 24 actually useful */ + + tmp = (char *)fd_dma_mem_alloc(1024 * try); + if (!tmp && !floppy_track_buffer) { + try >>= 1; /* buffer only one side */ + INFBOUND(try, 16); + tmp = (char *)fd_dma_mem_alloc(1024 * try); + } + if (!tmp && !floppy_track_buffer) { + fallback_on_nodma_alloc(&tmp, 2048 * try); + } + if (!tmp && !floppy_track_buffer) { + DPRINT("Unable to allocate DMA memory\n"); + goto out; + } + if (floppy_track_buffer) { + if (tmp) + fd_dma_mem_free((unsigned long)tmp, try * 1024); + } else { + buffer_min = buffer_max = -1; + floppy_track_buffer = tmp; + max_buffer_sectors = try; + } + } + + UDRS->fd_device = iminor(inode); + set_capacity(disks[drive], floppy_sizes[iminor(inode)]); + if (old_dev != -1 && old_dev != iminor(inode)) { + if (buffer_drive == drive) + buffer_track = -1; + } + + /* Allow ioctls if we have write-permissions even if read-only open. + * Needed so that programs such as fdrawcmd still can work on write + * protected disks */ + if (filp->f_mode & 2 + || permission(filp->f_dentry->d_inode, 2, NULL) == 0) + filp->private_data = (void *)8; + + if (UFDCS->rawcmd == 1) + UFDCS->rawcmd = 2; + + if (!(filp->f_flags & O_NDELAY)) { + if (filp->f_mode & 3) { + UDRS->last_checked = 0; + check_disk_change(inode->i_bdev); + if (UTESTF(FD_DISK_CHANGED)) + goto out; + } + res = -EROFS; + if ((filp->f_mode & 2) && !(UTESTF(FD_DISK_WRITABLE))) + goto out; + } + up(&open_lock); + return 0; +out: + if (UDRS->fd_ref < 0) + UDRS->fd_ref = 0; + else + UDRS->fd_ref--; + if (!UDRS->fd_ref) + opened_bdev[drive] = NULL; + floppy_release_irq_and_dma(); +out2: + up(&open_lock); + return res; +} + +/* + * Check if the disk has been changed or if a change has been faked. + */ +static int check_floppy_change(struct gendisk *disk) +{ + int drive = (long)disk->private_data; + + if (UTESTF(FD_DISK_CHANGED) || UTESTF(FD_VERIFY)) + return 1; + + if (UDP->checkfreq < (int)(jiffies - UDRS->last_checked)) { + if (floppy_grab_irq_and_dma()) { + return 1; + } + + lock_fdc(drive, 0); + poll_drive(0, 0); + process_fd_request(); + floppy_release_irq_and_dma(); + } + + if (UTESTF(FD_DISK_CHANGED) || + UTESTF(FD_VERIFY) || + test_bit(drive, &fake_change) || + (!ITYPE(UDRS->fd_device) && !current_type[drive])) + return 1; + return 0; +} + +/* + * This implements "read block 0" for floppy_revalidate(). + * Needed for format autodetection, checking whether there is + * a disk in the drive, and whether that disk is writable. + */ + +static int floppy_rb0_complete(struct bio *bio, unsigned int bytes_done, + int err) +{ + if (bio->bi_size) + return 1; + + complete((struct completion *)bio->bi_private); + return 0; +} + +static int __floppy_read_block_0(struct block_device *bdev) +{ + struct bio bio; + struct bio_vec bio_vec; + struct completion complete; + struct page *page; + size_t size; + + page = alloc_page(GFP_NOIO); + if (!page) { + process_fd_request(); + return -ENOMEM; + } + + size = bdev->bd_block_size; + if (!size) + size = 1024; + + bio_init(&bio); + bio.bi_io_vec = &bio_vec; + bio_vec.bv_page = page; + bio_vec.bv_len = size; + bio_vec.bv_offset = 0; + bio.bi_vcnt = 1; + bio.bi_idx = 0; + bio.bi_size = size; + bio.bi_bdev = bdev; + bio.bi_sector = 0; + init_completion(&complete); + bio.bi_private = &complete; + bio.bi_end_io = floppy_rb0_complete; + + submit_bio(READ, &bio); + generic_unplug_device(bdev_get_queue(bdev)); + process_fd_request(); + wait_for_completion(&complete); + + __free_page(page); + + return 0; +} + +/* revalidate the floppy disk, i.e. trigger format autodetection by reading + * the bootblock (block 0). "Autodetection" is also needed to check whether + * there is a disk in the drive at all... Thus we also do it for fixed + * geometry formats */ +static int floppy_revalidate(struct gendisk *disk) +{ + int drive = (long)disk->private_data; +#define NO_GEOM (!current_type[drive] && !ITYPE(UDRS->fd_device)) + int cf; + int res = 0; + + if (UTESTF(FD_DISK_CHANGED) || + UTESTF(FD_VERIFY) || test_bit(drive, &fake_change) || NO_GEOM) { + if (usage_count == 0) { + printk("VFS: revalidate called on non-open device.\n"); + return -EFAULT; + } + lock_fdc(drive, 0); + cf = UTESTF(FD_DISK_CHANGED) || UTESTF(FD_VERIFY); + if (!(cf || test_bit(drive, &fake_change) || NO_GEOM)) { + process_fd_request(); /*already done by another thread */ + return 0; + } + UDRS->maxblock = 0; + UDRS->maxtrack = 0; + if (buffer_drive == drive) + buffer_track = -1; + clear_bit(drive, &fake_change); + UCLEARF(FD_DISK_CHANGED); + if (cf) + UDRS->generation++; + if (NO_GEOM) { + /* auto-sensing */ + res = __floppy_read_block_0(opened_bdev[drive]); + } else { + if (cf) + poll_drive(0, FD_RAW_NEED_DISK); + process_fd_request(); + } + } + set_capacity(disk, floppy_sizes[UDRS->fd_device]); + return res; +} + +static struct block_device_operations floppy_fops = { + .owner = THIS_MODULE, + .open = floppy_open, + .release = floppy_release, + .ioctl = fd_ioctl, + .media_changed = check_floppy_change, + .revalidate_disk = floppy_revalidate, +}; +static char *table[] = { + "", "d360", "h1200", "u360", "u720", "h360", "h720", + "u1440", "u2880", "CompaQ", "h1440", "u1680", "h410", + "u820", "h1476", "u1722", "h420", "u830", "h1494", "u1743", + "h880", "u1040", "u1120", "h1600", "u1760", "u1920", + "u3200", "u3520", "u3840", "u1840", "u800", "u1600", + NULL +}; +static int t360[] = { 1, 0 }, + t1200[] = { 2, 5, 6, 10, 12, 14, 16, 18, 20, 23, 0 }, + t3in[] = { 8, 9, 26, 27, 28, 7, 11, 15, 19, 24, 25, 29, 31, 3, 4, 13, + 17, 21, 22, 30, 0 }; +static int *table_sup[] = + { NULL, t360, t1200, t3in + 5 + 8, t3in + 5, t3in, t3in }; + +static void __init register_devfs_entries(int drive) +{ + int base_minor = (drive < 4) ? drive : (124 + drive); + + if (UDP->cmos < NUMBER(default_drive_params)) { + int i = 0; + do { + int minor = base_minor + (table_sup[UDP->cmos][i] << 2); + + devfs_mk_bdev(MKDEV(FLOPPY_MAJOR, minor), + S_IFBLK | S_IRUSR | S_IWUSR | S_IRGRP | + S_IWGRP, "floppy/%d%s", drive, + table[table_sup[UDP->cmos][i]]); + } while (table_sup[UDP->cmos][i++]); + } +} + +/* + * Floppy Driver initialization + * ============================= + */ + +/* Determine the floppy disk controller type */ +/* This routine was written by David C. Niemi */ +static char __init get_fdc_version(void) +{ + int r; + + output_byte(FD_DUMPREGS); /* 82072 and better know DUMPREGS */ + if (FDCS->reset) + return FDC_NONE; + if ((r = result()) <= 0x00) + return FDC_NONE; /* No FDC present ??? */ + if ((r == 1) && (reply_buffer[0] == 0x80)) { + printk(KERN_INFO "FDC %d is an 8272A\n", fdc); + return FDC_8272A; /* 8272a/765 don't know DUMPREGS */ + } + if (r != 10) { + printk + ("FDC %d init: DUMPREGS: unexpected return of %d bytes.\n", + fdc, r); + return FDC_UNKNOWN; + } + + if (!fdc_configure()) { + printk(KERN_INFO "FDC %d is an 82072\n", fdc); + return FDC_82072; /* 82072 doesn't know CONFIGURE */ + } + + output_byte(FD_PERPENDICULAR); + if (need_more_output() == MORE_OUTPUT) { + output_byte(0); + } else { + printk(KERN_INFO "FDC %d is an 82072A\n", fdc); + return FDC_82072A; /* 82072A as found on Sparcs. */ + } + + output_byte(FD_UNLOCK); + r = result(); + if ((r == 1) && (reply_buffer[0] == 0x80)) { + printk(KERN_INFO "FDC %d is a pre-1991 82077\n", fdc); + return FDC_82077_ORIG; /* Pre-1991 82077, doesn't know + * LOCK/UNLOCK */ + } + if ((r != 1) || (reply_buffer[0] != 0x00)) { + printk("FDC %d init: UNLOCK: unexpected return of %d bytes.\n", + fdc, r); + return FDC_UNKNOWN; + } + output_byte(FD_PARTID); + r = result(); + if (r != 1) { + printk("FDC %d init: PARTID: unexpected return of %d bytes.\n", + fdc, r); + return FDC_UNKNOWN; + } + if (reply_buffer[0] == 0x80) { + printk(KERN_INFO "FDC %d is a post-1991 82077\n", fdc); + return FDC_82077; /* Revised 82077AA passes all the tests */ + } + switch (reply_buffer[0] >> 5) { + case 0x0: + /* Either a 82078-1 or a 82078SL running at 5Volt */ + printk(KERN_INFO "FDC %d is an 82078.\n", fdc); + return FDC_82078; + case 0x1: + printk(KERN_INFO "FDC %d is a 44pin 82078\n", fdc); + return FDC_82078; + case 0x2: + printk(KERN_INFO "FDC %d is a S82078B\n", fdc); + return FDC_S82078B; + case 0x3: + printk(KERN_INFO "FDC %d is a National Semiconductor PC87306\n", + fdc); + return FDC_87306; + default: + printk(KERN_INFO + "FDC %d init: 82078 variant with unknown PARTID=%d.\n", + fdc, reply_buffer[0] >> 5); + return FDC_82078_UNKN; + } +} /* get_fdc_version */ + +/* lilo configuration */ + +static void __init floppy_set_flags(int *ints, int param, int param2) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(default_drive_params); i++) { + if (param) + default_drive_params[i].params.flags |= param2; + else + default_drive_params[i].params.flags &= ~param2; + } + DPRINT("%s flag 0x%x\n", param2 ? "Setting" : "Clearing", param); +} + +static void __init daring(int *ints, int param, int param2) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(default_drive_params); i++) { + if (param) { + default_drive_params[i].params.select_delay = 0; + default_drive_params[i].params.flags |= + FD_SILENT_DCL_CLEAR; + } else { + default_drive_params[i].params.select_delay = + 2 * HZ / 100; + default_drive_params[i].params.flags &= + ~FD_SILENT_DCL_CLEAR; + } + } + DPRINT("Assuming %s floppy hardware\n", param ? "standard" : "broken"); +} + +static void __init set_cmos(int *ints, int dummy, int dummy2) +{ + int current_drive = 0; + + if (ints[0] != 2) { + DPRINT("wrong number of parameters for CMOS\n"); + return; + } + current_drive = ints[1]; + if (current_drive < 0 || current_drive >= 8) { + DPRINT("bad drive for set_cmos\n"); + return; + } +#if N_FDC > 1 + if (current_drive >= 4 && !FDC2) + FDC2 = 0x370; +#endif + DP->cmos = ints[2]; + DPRINT("setting CMOS code to %d\n", ints[2]); +} + +static struct param_table { + const char *name; + void (*fn) (int *ints, int param, int param2); + int *var; + int def_param; + int param2; +} config_params[] __initdata = { + {"allowed_drive_mask", NULL, &allowed_drive_mask, 0xff, 0}, /* obsolete */ + {"all_drives", NULL, &allowed_drive_mask, 0xff, 0}, /* obsolete */ + {"asus_pci", NULL, &allowed_drive_mask, 0x33, 0}, + {"irq", NULL, &FLOPPY_IRQ, 6, 0}, + {"dma", NULL, &FLOPPY_DMA, 2, 0}, + {"daring", daring, NULL, 1, 0}, +#if N_FDC > 1 + {"two_fdc", NULL, &FDC2, 0x370, 0}, + {"one_fdc", NULL, &FDC2, 0, 0}, +#endif + {"thinkpad", floppy_set_flags, NULL, 1, FD_INVERTED_DCL}, + {"broken_dcl", floppy_set_flags, NULL, 1, FD_BROKEN_DCL}, + {"messages", floppy_set_flags, NULL, 1, FTD_MSG}, + {"silent_dcl_clear", floppy_set_flags, NULL, 1, FD_SILENT_DCL_CLEAR}, + {"debug", floppy_set_flags, NULL, 1, FD_DEBUG}, + {"nodma", NULL, &can_use_virtual_dma, 1, 0}, + {"omnibook", NULL, &can_use_virtual_dma, 1, 0}, + {"yesdma", NULL, &can_use_virtual_dma, 0, 0}, + {"fifo_depth", NULL, &fifo_depth, 0xa, 0}, + {"nofifo", NULL, &no_fifo, 0x20, 0}, + {"usefifo", NULL, &no_fifo, 0, 0}, + {"cmos", set_cmos, NULL, 0, 0}, + {"slow", NULL, &slow_floppy, 1, 0}, + {"unexpected_interrupts", NULL, &print_unex, 1, 0}, + {"no_unexpected_interrupts", NULL, &print_unex, 0, 0}, + {"L40SX", NULL, &print_unex, 0, 0} + + EXTRA_FLOPPY_PARAMS +}; + +static int __init floppy_setup(char *str) +{ + int i; + int param; + int ints[11]; + + str = get_options(str, ARRAY_SIZE(ints), ints); + if (str) { + for (i = 0; i < ARRAY_SIZE(config_params); i++) { + if (strcmp(str, config_params[i].name) == 0) { + if (ints[0]) + param = ints[1]; + else + param = config_params[i].def_param; + if (config_params[i].fn) + config_params[i]. + fn(ints, param, + config_params[i].param2); + if (config_params[i].var) { + DPRINT("%s=%d\n", str, param); + *config_params[i].var = param; + } + return 1; + } + } + } + if (str) { + DPRINT("unknown floppy option [%s]\n", str); + + DPRINT("allowed options are:"); + for (i = 0; i < ARRAY_SIZE(config_params); i++) + printk(" %s", config_params[i].name); + printk("\n"); + } else + DPRINT("botched floppy option\n"); + DPRINT("Read Documentation/floppy.txt\n"); + return 0; +} + +static int have_no_fdc = -ENODEV; + +static void floppy_device_release(struct device *dev) +{ + complete(&device_release); +} + +static struct platform_device floppy_device = { + .name = "floppy", + .id = 0, + .dev = { + .release = floppy_device_release, + } +}; + +static struct kobject *floppy_find(dev_t dev, int *part, void *data) +{ + int drive = (*part & 3) | ((*part & 0x80) >> 5); + if (drive >= N_DRIVE || + !(allowed_drive_mask & (1 << drive)) || + fdc_state[FDC(drive)].version == FDC_NONE) + return NULL; + if (((*part >> 2) & 0x1f) >= NUMBER(floppy_type)) + return NULL; + *part = 0; + return get_disk(disks[drive]); +} + +static int __init floppy_init(void) +{ + int i, unit, drive; + int err, dr; + + raw_cmd = NULL; + + for (dr = 0; dr < N_DRIVE; dr++) { + disks[dr] = alloc_disk(1); + if (!disks[dr]) { + err = -ENOMEM; + goto out_put_disk; + } + + disks[dr]->major = FLOPPY_MAJOR; + disks[dr]->first_minor = TOMINOR(dr); + disks[dr]->fops = &floppy_fops; + sprintf(disks[dr]->disk_name, "fd%d", dr); + + init_timer(&motor_off_timer[dr]); + motor_off_timer[dr].data = dr; + motor_off_timer[dr].function = motor_off_callback; + } + + devfs_mk_dir("floppy"); + + err = register_blkdev(FLOPPY_MAJOR, "fd"); + if (err) + goto out_devfs_remove; + + floppy_queue = blk_init_queue(do_fd_request, &floppy_lock); + if (!floppy_queue) { + err = -ENOMEM; + goto out_unreg_blkdev; + } + blk_queue_max_sectors(floppy_queue, 64); + + blk_register_region(MKDEV(FLOPPY_MAJOR, 0), 256, THIS_MODULE, + floppy_find, NULL, NULL); + + for (i = 0; i < 256; i++) + if (ITYPE(i)) + floppy_sizes[i] = floppy_type[ITYPE(i)].size; + else + floppy_sizes[i] = MAX_DISK_SIZE << 1; + + reschedule_timeout(MAXTIMEOUT, "floppy init", MAXTIMEOUT); + config_types(); + + for (i = 0; i < N_FDC; i++) { + fdc = i; + CLEARSTRUCT(FDCS); + FDCS->dtr = -1; + FDCS->dor = 0x4; +#if defined(__sparc__) || defined(__mc68000__) + /*sparcs/sun3x don't have a DOR reset which we can fall back on to */ +#ifdef __mc68000__ + if (MACH_IS_SUN3X) +#endif + FDCS->version = FDC_82072A; +#endif + } + + use_virtual_dma = can_use_virtual_dma & 1; +#if defined(CONFIG_PPC64) + if (check_legacy_ioport(FDC1)) { + del_timer(&fd_timeout); + err = -ENODEV; + goto out_unreg_region; + } +#endif + fdc_state[0].address = FDC1; + if (fdc_state[0].address == -1) { + del_timer(&fd_timeout); + err = -ENODEV; + goto out_unreg_region; + } +#if N_FDC > 1 + fdc_state[1].address = FDC2; +#endif + + fdc = 0; /* reset fdc in case of unexpected interrupt */ + err = floppy_grab_irq_and_dma(); + if (err) { + del_timer(&fd_timeout); + err = -EBUSY; + goto out_unreg_region; + } + + /* initialise drive state */ + for (drive = 0; drive < N_DRIVE; drive++) { + CLEARSTRUCT(UDRS); + CLEARSTRUCT(UDRWE); + USETF(FD_DISK_NEWCHANGE); + USETF(FD_DISK_CHANGED); + USETF(FD_VERIFY); + UDRS->fd_device = -1; + floppy_track_buffer = NULL; + max_buffer_sectors = 0; + } + /* + * Small 10 msec delay to let through any interrupt that + * initialization might have triggered, to not + * confuse detection: + */ + msleep(10); + + for (i = 0; i < N_FDC; i++) { + fdc = i; + FDCS->driver_version = FD_DRIVER_VERSION; + for (unit = 0; unit < 4; unit++) + FDCS->track[unit] = 0; + if (FDCS->address == -1) + continue; + FDCS->rawcmd = 2; + if (user_reset_fdc(-1, FD_RESET_ALWAYS, 0)) { + /* free ioports reserved by floppy_grab_irq_and_dma() */ + release_region(FDCS->address + 2, 4); + release_region(FDCS->address + 7, 1); + FDCS->address = -1; + FDCS->version = FDC_NONE; + continue; + } + /* Try to determine the floppy controller type */ + FDCS->version = get_fdc_version(); + if (FDCS->version == FDC_NONE) { + /* free ioports reserved by floppy_grab_irq_and_dma() */ + release_region(FDCS->address + 2, 4); + release_region(FDCS->address + 7, 1); + FDCS->address = -1; + continue; + } + if (can_use_virtual_dma == 2 && FDCS->version < FDC_82072A) + can_use_virtual_dma = 0; + + have_no_fdc = 0; + /* Not all FDCs seem to be able to handle the version command + * properly, so force a reset for the standard FDC clones, + * to avoid interrupt garbage. + */ + user_reset_fdc(-1, FD_RESET_ALWAYS, 0); + } + fdc = 0; + del_timer(&fd_timeout); + current_drive = 0; + floppy_release_irq_and_dma(); + initialising = 0; + if (have_no_fdc) { + DPRINT("no floppy controllers found\n"); + err = have_no_fdc; + goto out_flush_work; + } + + err = platform_device_register(&floppy_device); + if (err) + goto out_flush_work; + + for (drive = 0; drive < N_DRIVE; drive++) { + if (!(allowed_drive_mask & (1 << drive))) + continue; + if (fdc_state[FDC(drive)].version == FDC_NONE) + continue; + /* to be cleaned up... */ + disks[drive]->private_data = (void *)(long)drive; + disks[drive]->queue = floppy_queue; + disks[drive]->flags |= GENHD_FL_REMOVABLE; + disks[drive]->driverfs_dev = &floppy_device.dev; + add_disk(disks[drive]); + } + + return 0; + +out_flush_work: + flush_scheduled_work(); + if (usage_count) + floppy_release_irq_and_dma(); +out_unreg_region: + blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); + blk_cleanup_queue(floppy_queue); +out_unreg_blkdev: + unregister_blkdev(FLOPPY_MAJOR, "fd"); +out_devfs_remove: + devfs_remove("floppy"); +out_put_disk: + while (dr--) { + del_timer(&motor_off_timer[dr]); + put_disk(disks[dr]); + } + return err; +} + +static DEFINE_SPINLOCK(floppy_usage_lock); + +static int floppy_grab_irq_and_dma(void) +{ + unsigned long flags; + + spin_lock_irqsave(&floppy_usage_lock, flags); + if (usage_count++) { + spin_unlock_irqrestore(&floppy_usage_lock, flags); + return 0; + } + spin_unlock_irqrestore(&floppy_usage_lock, flags); + if (fd_request_irq()) { + DPRINT("Unable to grab IRQ%d for the floppy driver\n", + FLOPPY_IRQ); + spin_lock_irqsave(&floppy_usage_lock, flags); + usage_count--; + spin_unlock_irqrestore(&floppy_usage_lock, flags); + return -1; + } + if (fd_request_dma()) { + DPRINT("Unable to grab DMA%d for the floppy driver\n", + FLOPPY_DMA); + fd_free_irq(); + spin_lock_irqsave(&floppy_usage_lock, flags); + usage_count--; + spin_unlock_irqrestore(&floppy_usage_lock, flags); + return -1; + } + + for (fdc = 0; fdc < N_FDC; fdc++) { + if (FDCS->address != -1) { + if (!request_region(FDCS->address + 2, 4, "floppy")) { + DPRINT("Floppy io-port 0x%04lx in use\n", + FDCS->address + 2); + goto cleanup1; + } + if (!request_region(FDCS->address + 7, 1, "floppy DIR")) { + DPRINT("Floppy io-port 0x%04lx in use\n", + FDCS->address + 7); + goto cleanup2; + } + /* address + 6 is reserved, and may be taken by IDE. + * Unfortunately, Adaptec doesn't know this :-(, */ + } + } + for (fdc = 0; fdc < N_FDC; fdc++) { + if (FDCS->address != -1) { + reset_fdc_info(1); + fd_outb(FDCS->dor, FD_DOR); + } + } + fdc = 0; + set_dor(0, ~0, 8); /* avoid immediate interrupt */ + + for (fdc = 0; fdc < N_FDC; fdc++) + if (FDCS->address != -1) + fd_outb(FDCS->dor, FD_DOR); + /* + * The driver will try and free resources and relies on us + * to know if they were allocated or not. + */ + fdc = 0; + irqdma_allocated = 1; + return 0; +cleanup2: + release_region(FDCS->address + 2, 4); +cleanup1: + fd_free_irq(); + fd_free_dma(); + while (--fdc >= 0) { + release_region(FDCS->address + 2, 4); + release_region(FDCS->address + 7, 1); + } + spin_lock_irqsave(&floppy_usage_lock, flags); + usage_count--; + spin_unlock_irqrestore(&floppy_usage_lock, flags); + return -1; +} + +static void floppy_release_irq_and_dma(void) +{ + int old_fdc; +#ifdef FLOPPY_SANITY_CHECK +#ifndef __sparc__ + int drive; +#endif +#endif + long tmpsize; + unsigned long tmpaddr; + unsigned long flags; + + spin_lock_irqsave(&floppy_usage_lock, flags); + if (--usage_count) { + spin_unlock_irqrestore(&floppy_usage_lock, flags); + return; + } + spin_unlock_irqrestore(&floppy_usage_lock, flags); + if (irqdma_allocated) { + fd_disable_dma(); + fd_free_dma(); + fd_free_irq(); + irqdma_allocated = 0; + } + set_dor(0, ~0, 8); +#if N_FDC > 1 + set_dor(1, ~8, 0); +#endif + floppy_enable_hlt(); + + if (floppy_track_buffer && max_buffer_sectors) { + tmpsize = max_buffer_sectors * 1024; + tmpaddr = (unsigned long)floppy_track_buffer; + floppy_track_buffer = NULL; + max_buffer_sectors = 0; + buffer_min = buffer_max = -1; + fd_dma_mem_free(tmpaddr, tmpsize); + } +#ifdef FLOPPY_SANITY_CHECK +#ifndef __sparc__ + for (drive = 0; drive < N_FDC * 4; drive++) + if (timer_pending(motor_off_timer + drive)) + printk("motor off timer %d still active\n", drive); +#endif + + if (timer_pending(&fd_timeout)) + printk("floppy timer still active:%s\n", timeout_message); + if (timer_pending(&fd_timer)) + printk("auxiliary floppy timer still active\n"); + if (floppy_work.pending) + printk("work still pending\n"); +#endif + old_fdc = fdc; + for (fdc = 0; fdc < N_FDC; fdc++) + if (FDCS->address != -1) { + release_region(FDCS->address + 2, 4); + release_region(FDCS->address + 7, 1); + } + fdc = old_fdc; +} + +#ifdef MODULE + +static char *floppy; + +static void unregister_devfs_entries(int drive) +{ + int i; + + if (UDP->cmos < NUMBER(default_drive_params)) { + i = 0; + do { + devfs_remove("floppy/%d%s", drive, + table[table_sup[UDP->cmos][i]]); + } while (table_sup[UDP->cmos][i++]); + } +} + +static void __init parse_floppy_cfg_string(char *cfg) +{ + char *ptr; + + while (*cfg) { + for (ptr = cfg; *cfg && *cfg != ' ' && *cfg != '\t'; cfg++) ; + if (*cfg) { + *cfg = '\0'; + cfg++; + } + if (*ptr) + floppy_setup(ptr); + } +} + +int init_module(void) +{ + if (floppy) + parse_floppy_cfg_string(floppy); + return floppy_init(); +} + +void cleanup_module(void) +{ + int drive; + + init_completion(&device_release); + blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); + unregister_blkdev(FLOPPY_MAJOR, "fd"); + + for (drive = 0; drive < N_DRIVE; drive++) { + del_timer_sync(&motor_off_timer[drive]); + + if ((allowed_drive_mask & (1 << drive)) && + fdc_state[FDC(drive)].version != FDC_NONE) { + del_gendisk(disks[drive]); + unregister_devfs_entries(drive); + } + put_disk(disks[drive]); + } + platform_device_unregister(&floppy_device); + devfs_remove("floppy"); + + del_timer_sync(&fd_timeout); + del_timer_sync(&fd_timer); + blk_cleanup_queue(floppy_queue); + + if (usage_count) + floppy_release_irq_and_dma(); + + /* eject disk, if any */ + fd_eject(0); + + wait_for_completion(&device_release); +} + +module_param(floppy, charp, 0); +module_param(FLOPPY_IRQ, int, 0); +module_param(FLOPPY_DMA, int, 0); +MODULE_AUTHOR("Alain L. Knaff"); +MODULE_SUPPORTED_DEVICE("fd"); +MODULE_LICENSE("GPL"); + +#else + +__setup("floppy=", floppy_setup); +module_init(floppy_init) +#endif + +MODULE_ALIAS_BLOCKDEV_MAJOR(FLOPPY_MAJOR); diff --git a/drivers/block/genhd.c b/drivers/block/genhd.c new file mode 100644 index 00000000000..ab4db71375e --- /dev/null +++ b/drivers/block/genhd.c @@ -0,0 +1,685 @@ +/* + * gendisk handling + */ + +#include <linux/config.h> +#include <linux/module.h> +#include <linux/fs.h> +#include <linux/genhd.h> +#include <linux/kernel.h> +#include <linux/blkdev.h> +#include <linux/init.h> +#include <linux/spinlock.h> +#include <linux/seq_file.h> +#include <linux/slab.h> +#include <linux/kmod.h> +#include <linux/kobj_map.h> + +#define MAX_PROBE_HASH 255 /* random */ + +static struct subsystem block_subsys; + +static DECLARE_MUTEX(block_subsys_sem); + +/* + * Can be deleted altogether. Later. + * + */ +static struct blk_major_name { + struct blk_major_name *next; + int major; + char name[16]; +} *major_names[MAX_PROBE_HASH]; + +/* index in the above - for now: assume no multimajor ranges */ +static inline int major_to_index(int major) +{ + return major % MAX_PROBE_HASH; +} + +#ifdef CONFIG_PROC_FS +/* get block device names in somewhat random order */ +int get_blkdev_list(char *p) +{ + struct blk_major_name *n; + int i, len; + + len = sprintf(p, "\nBlock devices:\n"); + + down(&block_subsys_sem); + for (i = 0; i < ARRAY_SIZE(major_names); i++) { + for (n = major_names[i]; n; n = n->next) + len += sprintf(p+len, "%3d %s\n", + n->major, n->name); + } + up(&block_subsys_sem); + + return len; +} +#endif + +int register_blkdev(unsigned int major, const char *name) +{ + struct blk_major_name **n, *p; + int index, ret = 0; + + down(&block_subsys_sem); + + /* temporary */ + if (major == 0) { + for (index = ARRAY_SIZE(major_names)-1; index > 0; index--) { + if (major_names[index] == NULL) + break; + } + + if (index == 0) { + printk("register_blkdev: failed to get major for %s\n", + name); + ret = -EBUSY; + goto out; + } + major = index; + ret = major; + } + + p = kmalloc(sizeof(struct blk_major_name), GFP_KERNEL); + if (p == NULL) { + ret = -ENOMEM; + goto out; + } + + p->major = major; + strlcpy(p->name, name, sizeof(p->name)); + p->next = NULL; + index = major_to_index(major); + + for (n = &major_names[index]; *n; n = &(*n)->next) { + if ((*n)->major == major) + break; + } + if (!*n) + *n = p; + else + ret = -EBUSY; + + if (ret < 0) { + printk("register_blkdev: cannot get major %d for %s\n", + major, name); + kfree(p); + } +out: + up(&block_subsys_sem); + return ret; +} + +EXPORT_SYMBOL(register_blkdev); + +/* todo: make void - error printk here */ +int unregister_blkdev(unsigned int major, const char *name) +{ + struct blk_major_name **n; + struct blk_major_name *p = NULL; + int index = major_to_index(major); + int ret = 0; + + down(&block_subsys_sem); + for (n = &major_names[index]; *n; n = &(*n)->next) + if ((*n)->major == major) + break; + if (!*n || strcmp((*n)->name, name)) + ret = -EINVAL; + else { + p = *n; + *n = p->next; + } + up(&block_subsys_sem); + kfree(p); + + return ret; +} + +EXPORT_SYMBOL(unregister_blkdev); + +static struct kobj_map *bdev_map; + +/* + * Register device numbers dev..(dev+range-1) + * range must be nonzero + * The hash chain is sorted on range, so that subranges can override. + */ +void blk_register_region(dev_t dev, unsigned long range, struct module *module, + struct kobject *(*probe)(dev_t, int *, void *), + int (*lock)(dev_t, void *), void *data) +{ + kobj_map(bdev_map, dev, range, module, probe, lock, data); +} + +EXPORT_SYMBOL(blk_register_region); + +void blk_unregister_region(dev_t dev, unsigned long range) +{ + kobj_unmap(bdev_map, dev, range); +} + +EXPORT_SYMBOL(blk_unregister_region); + +static struct kobject *exact_match(dev_t dev, int *part, void *data) +{ + struct gendisk *p = data; + return &p->kobj; +} + +static int exact_lock(dev_t dev, void *data) +{ + struct gendisk *p = data; + + if (!get_disk(p)) + return -1; + return 0; +} + +/** + * add_disk - add partitioning information to kernel list + * @disk: per-device partitioning information + * + * This function registers the partitioning information in @disk + * with the kernel. + */ +void add_disk(struct gendisk *disk) +{ + disk->flags |= GENHD_FL_UP; + blk_register_region(MKDEV(disk->major, disk->first_minor), + disk->minors, NULL, exact_match, exact_lock, disk); + register_disk(disk); + blk_register_queue(disk); +} + +EXPORT_SYMBOL(add_disk); +EXPORT_SYMBOL(del_gendisk); /* in partitions/check.c */ + +void unlink_gendisk(struct gendisk *disk) +{ + blk_unregister_queue(disk); + blk_unregister_region(MKDEV(disk->major, disk->first_minor), + disk->minors); +} + +#define to_disk(obj) container_of(obj,struct gendisk,kobj) + +/** + * get_gendisk - get partitioning information for a given device + * @dev: device to get partitioning information for + * + * This function gets the structure containing partitioning + * information for the given device @dev. + */ +struct gendisk *get_gendisk(dev_t dev, int *part) +{ + struct kobject *kobj = kobj_lookup(bdev_map, dev, part); + return kobj ? to_disk(kobj) : NULL; +} + +#ifdef CONFIG_PROC_FS +/* iterator */ +static void *part_start(struct seq_file *part, loff_t *pos) +{ + struct list_head *p; + loff_t l = *pos; + + down(&block_subsys_sem); + list_for_each(p, &block_subsys.kset.list) + if (!l--) + return list_entry(p, struct gendisk, kobj.entry); + return NULL; +} + +static void *part_next(struct seq_file *part, void *v, loff_t *pos) +{ + struct list_head *p = ((struct gendisk *)v)->kobj.entry.next; + ++*pos; + return p==&block_subsys.kset.list ? NULL : + list_entry(p, struct gendisk, kobj.entry); +} + +static void part_stop(struct seq_file *part, void *v) +{ + up(&block_subsys_sem); +} + +static int show_partition(struct seq_file *part, void *v) +{ + struct gendisk *sgp = v; + int n; + char buf[BDEVNAME_SIZE]; + + if (&sgp->kobj.entry == block_subsys.kset.list.next) + seq_puts(part, "major minor #blocks name\n\n"); + + /* Don't show non-partitionable removeable devices or empty devices */ + if (!get_capacity(sgp) || + (sgp->minors == 1 && (sgp->flags & GENHD_FL_REMOVABLE))) + return 0; + if (sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO) + return 0; + + /* show the full disk and all non-0 size partitions of it */ + seq_printf(part, "%4d %4d %10llu %s\n", + sgp->major, sgp->first_minor, + (unsigned long long)get_capacity(sgp) >> 1, + disk_name(sgp, 0, buf)); + for (n = 0; n < sgp->minors - 1; n++) { + if (!sgp->part[n]) + continue; + if (sgp->part[n]->nr_sects == 0) + continue; + seq_printf(part, "%4d %4d %10llu %s\n", + sgp->major, n + 1 + sgp->first_minor, + (unsigned long long)sgp->part[n]->nr_sects >> 1 , + disk_name(sgp, n + 1, buf)); + } + + return 0; +} + +struct seq_operations partitions_op = { + .start =part_start, + .next = part_next, + .stop = part_stop, + .show = show_partition +}; +#endif + + +extern int blk_dev_init(void); + +static struct kobject *base_probe(dev_t dev, int *part, void *data) +{ + if (request_module("block-major-%d-%d", MAJOR(dev), MINOR(dev)) > 0) + /* Make old-style 2.4 aliases work */ + request_module("block-major-%d", MAJOR(dev)); + return NULL; +} + +static int __init genhd_device_init(void) +{ + bdev_map = kobj_map_init(base_probe, &block_subsys_sem); + blk_dev_init(); + subsystem_register(&block_subsys); + return 0; +} + +subsys_initcall(genhd_device_init); + + + +/* + * kobject & sysfs bindings for block devices + */ +static ssize_t disk_attr_show(struct kobject *kobj, struct attribute *attr, + char *page) +{ + struct gendisk *disk = to_disk(kobj); + struct disk_attribute *disk_attr = + container_of(attr,struct disk_attribute,attr); + ssize_t ret = 0; + + if (disk_attr->show) + ret = disk_attr->show(disk,page); + return ret; +} + +static struct sysfs_ops disk_sysfs_ops = { + .show = &disk_attr_show, +}; + +static ssize_t disk_dev_read(struct gendisk * disk, char *page) +{ + dev_t base = MKDEV(disk->major, disk->first_minor); + return print_dev_t(page, base); +} +static ssize_t disk_range_read(struct gendisk * disk, char *page) +{ + return sprintf(page, "%d\n", disk->minors); +} +static ssize_t disk_removable_read(struct gendisk * disk, char *page) +{ + return sprintf(page, "%d\n", + (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0)); + +} +static ssize_t disk_size_read(struct gendisk * disk, char *page) +{ + return sprintf(page, "%llu\n", (unsigned long long)get_capacity(disk)); +} + +static ssize_t disk_stats_read(struct gendisk * disk, char *page) +{ + preempt_disable(); + disk_round_stats(disk); + preempt_enable(); + return sprintf(page, + "%8u %8u %8llu %8u " + "%8u %8u %8llu %8u " + "%8u %8u %8u" + "\n", + disk_stat_read(disk, reads), disk_stat_read(disk, read_merges), + (unsigned long long)disk_stat_read(disk, read_sectors), + jiffies_to_msecs(disk_stat_read(disk, read_ticks)), + disk_stat_read(disk, writes), + disk_stat_read(disk, write_merges), + (unsigned long long)disk_stat_read(disk, write_sectors), + jiffies_to_msecs(disk_stat_read(disk, write_ticks)), + disk->in_flight, + jiffies_to_msecs(disk_stat_read(disk, io_ticks)), + jiffies_to_msecs(disk_stat_read(disk, time_in_queue))); +} +static struct disk_attribute disk_attr_dev = { + .attr = {.name = "dev", .mode = S_IRUGO }, + .show = disk_dev_read +}; +static struct disk_attribute disk_attr_range = { + .attr = {.name = "range", .mode = S_IRUGO }, + .show = disk_range_read +}; +static struct disk_attribute disk_attr_removable = { + .attr = {.name = "removable", .mode = S_IRUGO }, + .show = disk_removable_read +}; +static struct disk_attribute disk_attr_size = { + .attr = {.name = "size", .mode = S_IRUGO }, + .show = disk_size_read +}; +static struct disk_attribute disk_attr_stat = { + .attr = {.name = "stat", .mode = S_IRUGO }, + .show = disk_stats_read +}; + +static struct attribute * default_attrs[] = { + &disk_attr_dev.attr, + &disk_attr_range.attr, + &disk_attr_removable.attr, + &disk_attr_size.attr, + &disk_attr_stat.attr, + NULL, +}; + +static void disk_release(struct kobject * kobj) +{ + struct gendisk *disk = to_disk(kobj); + kfree(disk->random); + kfree(disk->part); + free_disk_stats(disk); + kfree(disk); +} + +static struct kobj_type ktype_block = { + .release = disk_release, + .sysfs_ops = &disk_sysfs_ops, + .default_attrs = default_attrs, +}; + +extern struct kobj_type ktype_part; + +static int block_hotplug_filter(struct kset *kset, struct kobject *kobj) +{ + struct kobj_type *ktype = get_ktype(kobj); + + return ((ktype == &ktype_block) || (ktype == &ktype_part)); +} + +static int block_hotplug(struct kset *kset, struct kobject *kobj, char **envp, + int num_envp, char *buffer, int buffer_size) +{ + struct kobj_type *ktype = get_ktype(kobj); + struct device *physdev; + struct gendisk *disk; + struct hd_struct *part; + int length = 0; + int i = 0; + + if (ktype == &ktype_block) { + disk = container_of(kobj, struct gendisk, kobj); + add_hotplug_env_var(envp, num_envp, &i, buffer, buffer_size, + &length, "MINOR=%u", disk->first_minor); + } else if (ktype == &ktype_part) { + disk = container_of(kobj->parent, struct gendisk, kobj); + part = container_of(kobj, struct hd_struct, kobj); + add_hotplug_env_var(envp, num_envp, &i, buffer, buffer_size, + &length, "MINOR=%u", + disk->first_minor + part->partno); + } else + return 0; + + add_hotplug_env_var(envp, num_envp, &i, buffer, buffer_size, &length, + "MAJOR=%u", disk->major); + + /* add physical device, backing this device */ + physdev = disk->driverfs_dev; + if (physdev) { + char *path = kobject_get_path(&physdev->kobj, GFP_KERNEL); + + add_hotplug_env_var(envp, num_envp, &i, buffer, buffer_size, + &length, "PHYSDEVPATH=%s", path); + kfree(path); + + if (physdev->bus) + add_hotplug_env_var(envp, num_envp, &i, + buffer, buffer_size, &length, + "PHYSDEVBUS=%s", + physdev->bus->name); + + if (physdev->driver) + add_hotplug_env_var(envp, num_envp, &i, + buffer, buffer_size, &length, + "PHYSDEVDRIVER=%s", + physdev->driver->name); + } + + /* terminate, set to next free slot, shrink available space */ + envp[i] = NULL; + envp = &envp[i]; + num_envp -= i; + buffer = &buffer[length]; + buffer_size -= length; + + return 0; +} + +static struct kset_hotplug_ops block_hotplug_ops = { + .filter = block_hotplug_filter, + .hotplug = block_hotplug, +}; + +/* declare block_subsys. */ +static decl_subsys(block, &ktype_block, &block_hotplug_ops); + + +/* + * aggregate disk stat collector. Uses the same stats that the sysfs + * entries do, above, but makes them available through one seq_file. + * Watching a few disks may be efficient through sysfs, but watching + * all of them will be more efficient through this interface. + * + * The output looks suspiciously like /proc/partitions with a bunch of + * extra fields. + */ + +/* iterator */ +static void *diskstats_start(struct seq_file *part, loff_t *pos) +{ + loff_t k = *pos; + struct list_head *p; + + down(&block_subsys_sem); + list_for_each(p, &block_subsys.kset.list) + if (!k--) + return list_entry(p, struct gendisk, kobj.entry); + return NULL; +} + +static void *diskstats_next(struct seq_file *part, void *v, loff_t *pos) +{ + struct list_head *p = ((struct gendisk *)v)->kobj.entry.next; + ++*pos; + return p==&block_subsys.kset.list ? NULL : + list_entry(p, struct gendisk, kobj.entry); +} + +static void diskstats_stop(struct seq_file *part, void *v) +{ + up(&block_subsys_sem); +} + +static int diskstats_show(struct seq_file *s, void *v) +{ + struct gendisk *gp = v; + char buf[BDEVNAME_SIZE]; + int n = 0; + + /* + if (&sgp->kobj.entry == block_subsys.kset.list.next) + seq_puts(s, "major minor name" + " rio rmerge rsect ruse wio wmerge " + "wsect wuse running use aveq" + "\n\n"); + */ + + preempt_disable(); + disk_round_stats(gp); + preempt_enable(); + seq_printf(s, "%4d %4d %s %u %u %llu %u %u %u %llu %u %u %u %u\n", + gp->major, n + gp->first_minor, disk_name(gp, n, buf), + disk_stat_read(gp, reads), disk_stat_read(gp, read_merges), + (unsigned long long)disk_stat_read(gp, read_sectors), + jiffies_to_msecs(disk_stat_read(gp, read_ticks)), + disk_stat_read(gp, writes), disk_stat_read(gp, write_merges), + (unsigned long long)disk_stat_read(gp, write_sectors), + jiffies_to_msecs(disk_stat_read(gp, write_ticks)), + gp->in_flight, + jiffies_to_msecs(disk_stat_read(gp, io_ticks)), + jiffies_to_msecs(disk_stat_read(gp, time_in_queue))); + + /* now show all non-0 size partitions of it */ + for (n = 0; n < gp->minors - 1; n++) { + struct hd_struct *hd = gp->part[n]; + + if (hd && hd->nr_sects) + seq_printf(s, "%4d %4d %s %u %u %u %u\n", + gp->major, n + gp->first_minor + 1, + disk_name(gp, n + 1, buf), + hd->reads, hd->read_sectors, + hd->writes, hd->write_sectors); + } + + return 0; +} + +struct seq_operations diskstats_op = { + .start = diskstats_start, + .next = diskstats_next, + .stop = diskstats_stop, + .show = diskstats_show +}; + + +struct gendisk *alloc_disk(int minors) +{ + struct gendisk *disk = kmalloc(sizeof(struct gendisk), GFP_KERNEL); + if (disk) { + memset(disk, 0, sizeof(struct gendisk)); + if (!init_disk_stats(disk)) { + kfree(disk); + return NULL; + } + if (minors > 1) { + int size = (minors - 1) * sizeof(struct hd_struct *); + disk->part = kmalloc(size, GFP_KERNEL); + if (!disk->part) { + kfree(disk); + return NULL; + } + memset(disk->part, 0, size); + } + disk->minors = minors; + kobj_set_kset_s(disk,block_subsys); + kobject_init(&disk->kobj); + rand_initialize_disk(disk); + } + return disk; +} + +EXPORT_SYMBOL(alloc_disk); + +struct kobject *get_disk(struct gendisk *disk) +{ + struct module *owner; + struct kobject *kobj; + + if (!disk->fops) + return NULL; + owner = disk->fops->owner; + if (owner && !try_module_get(owner)) + return NULL; + kobj = kobject_get(&disk->kobj); + if (kobj == NULL) { + module_put(owner); + return NULL; + } + return kobj; + +} + +EXPORT_SYMBOL(get_disk); + +void put_disk(struct gendisk *disk) +{ + if (disk) + kobject_put(&disk->kobj); +} + +EXPORT_SYMBOL(put_disk); + +void set_device_ro(struct block_device *bdev, int flag) +{ + if (bdev->bd_contains != bdev) + bdev->bd_part->policy = flag; + else + bdev->bd_disk->policy = flag; +} + +EXPORT_SYMBOL(set_device_ro); + +void set_disk_ro(struct gendisk *disk, int flag) +{ + int i; + disk->policy = flag; + for (i = 0; i < disk->minors - 1; i++) + if (disk->part[i]) disk->part[i]->policy = flag; +} + +EXPORT_SYMBOL(set_disk_ro); + +int bdev_read_only(struct block_device *bdev) +{ + if (!bdev) + return 0; + else if (bdev->bd_contains != bdev) + return bdev->bd_part->policy; + else + return bdev->bd_disk->policy; +} + +EXPORT_SYMBOL(bdev_read_only); + +int invalidate_partition(struct gendisk *disk, int index) +{ + int res = 0; + struct block_device *bdev = bdget_disk(disk, index); + if (bdev) { + res = __invalidate_device(bdev, 1); + bdput(bdev); + } + return res; +} + +EXPORT_SYMBOL(invalidate_partition); diff --git a/drivers/block/ida_cmd.h b/drivers/block/ida_cmd.h new file mode 100644 index 00000000000..98b5746b308 --- /dev/null +++ b/drivers/block/ida_cmd.h @@ -0,0 +1,349 @@ +/* + * Disk Array driver for Compaq SMART2 Controllers + * Copyright 1998 Compaq Computer Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + */ +#ifndef ARRAYCMD_H +#define ARRAYCMD_H + +#include <asm/types.h> +#if 0 +#include <linux/blkdev.h> +#endif + +/* for the Smart Array 42XX cards */ +#define S42XX_REQUEST_PORT_OFFSET 0x40 +#define S42XX_REPLY_INTR_MASK_OFFSET 0x34 +#define S42XX_REPLY_PORT_OFFSET 0x44 +#define S42XX_INTR_STATUS 0x30 + +#define S42XX_INTR_OFF 0x08 +#define S42XX_INTR_PENDING 0x08 + +#define COMMAND_FIFO 0x04 +#define COMMAND_COMPLETE_FIFO 0x08 +#define INTR_MASK 0x0C +#define INTR_STATUS 0x10 +#define INTR_PENDING 0x14 + +#define FIFO_NOT_EMPTY 0x01 +#define FIFO_NOT_FULL 0x02 + +#define BIG_PROBLEM 0x40 +#define LOG_NOT_CONF 2 + +#pragma pack(1) +typedef struct { + __u32 size; + __u32 addr; +} sg_t; + +#define RCODE_NONFATAL 0x02 +#define RCODE_FATAL 0x04 +#define RCODE_INVREQ 0x10 +typedef struct { + __u16 next; + __u8 cmd; + __u8 rcode; + __u32 blk; + __u16 blk_cnt; + __u8 sg_cnt; + __u8 reserved; +} rhdr_t; + +#define SG_MAX 32 +typedef struct { + rhdr_t hdr; + sg_t sg[SG_MAX]; + __u32 bp; +} rblk_t; + +typedef struct { + __u8 unit; + __u8 prio; + __u16 size; +} chdr_t; + +#define CMD_RWREQ 0x00 +#define CMD_IOCTL_PEND 0x01 +#define CMD_IOCTL_DONE 0x02 + +typedef struct cmdlist { + chdr_t hdr; + rblk_t req; + __u32 size; + int retry_cnt; + __u32 busaddr; + int ctlr; + struct cmdlist *prev; + struct cmdlist *next; + struct request *rq; + int type; +} cmdlist_t; + +#define ID_CTLR 0x11 +typedef struct { + __u8 nr_drvs; + __u32 cfg_sig; + __u8 firm_rev[4]; + __u8 rom_rev[4]; + __u8 hw_rev; + __u32 bb_rev; + __u32 drv_present_map; + __u32 ext_drv_map; + __u32 board_id; + __u8 cfg_error; + __u32 non_disk_bits; + __u8 bad_ram_addr; + __u8 cpu_rev; + __u8 pdpi_rev; + __u8 epic_rev; + __u8 wcxc_rev; + __u8 marketing_rev; + __u8 ctlr_flags; + __u8 host_flags; + __u8 expand_dis; + __u8 scsi_chips; + __u32 max_req_blocks; + __u32 ctlr_clock; + __u8 drvs_per_bus; + __u16 big_drv_present_map[8]; + __u16 big_ext_drv_map[8]; + __u16 big_non_disk_map[8]; + __u16 task_flags; + __u8 icl_bus; + __u8 red_modes; + __u8 cur_red_mode; + __u8 red_ctlr_stat; + __u8 red_fail_reason; + __u8 reserved[403]; +} id_ctlr_t; + +typedef struct { + __u16 cyl; + __u8 heads; + __u8 xsig; + __u8 psectors; + __u16 wpre; + __u8 maxecc; + __u8 drv_ctrl; + __u16 pcyls; + __u8 pheads; + __u16 landz; + __u8 sect_per_track; + __u8 cksum; +} drv_param_t; + +#define ID_LOG_DRV 0x10 +typedef struct { + __u16 blk_size; + __u32 nr_blks; + drv_param_t drv; + __u8 fault_tol; + __u8 reserved; + __u8 bios_disable; +} id_log_drv_t; + +#define ID_LOG_DRV_EXT 0x18 +typedef struct { + __u32 log_drv_id; + __u8 log_drv_label[64]; + __u8 reserved[418]; +} id_log_drv_ext_t; + +#define SENSE_LOG_DRV_STAT 0x12 +typedef struct { + __u8 status; + __u32 fail_map; + __u16 read_err[32]; + __u16 write_err[32]; + __u8 drv_err_data[256]; + __u8 drq_timeout[32]; + __u32 blks_to_recover; + __u8 drv_recovering; + __u16 remap_cnt[32]; + __u32 replace_drv_map; + __u32 act_spare_map; + __u8 spare_stat; + __u8 spare_repl_map[32]; + __u32 repl_ok_map; + __u8 media_exch; + __u8 cache_fail; + __u8 expn_fail; + __u8 unit_flags; + __u16 big_fail_map[8]; + __u16 big_remap_map[128]; + __u16 big_repl_map[8]; + __u16 big_act_spare_map[8]; + __u8 big_spar_repl_map[128]; + __u16 big_repl_ok_map[8]; + __u8 big_drv_rebuild; + __u8 reserved[36]; +} sense_log_drv_stat_t; + +#define START_RECOVER 0x13 + +#define ID_PHYS_DRV 0x15 +typedef struct { + __u8 scsi_bus; + __u8 scsi_id; + __u16 blk_size; + __u32 nr_blks; + __u32 rsvd_blks; + __u8 drv_model[40]; + __u8 drv_sn[40]; + __u8 drv_fw[8]; + __u8 scsi_iq_bits; + __u8 compaq_drv_stmp; + __u8 last_fail; + __u8 phys_drv_flags; + __u8 phys_drv_flags1; + __u8 scsi_lun; + __u8 phys_drv_flags2; + __u8 reserved; + __u32 spi_speed_rules; + __u8 phys_connector[2]; + __u8 phys_box_on_bus; + __u8 phys_bay_in_box; +} id_phys_drv_t; + +#define BLINK_DRV_LEDS 0x16 +typedef struct { + __u32 blink_duration; + __u32 reserved; + __u8 blink[256]; + __u8 reserved1[248]; +} blink_drv_leds_t; + +#define SENSE_BLINK_LEDS 0x17 +typedef struct { + __u32 blink_duration; + __u32 btime_elap; + __u8 blink[256]; + __u8 reserved1[248]; +} sense_blink_leds_t; + +#define IDA_READ 0x20 +#define IDA_WRITE 0x30 +#define IDA_WRITE_MEDIA 0x31 +#define RESET_TO_DIAG 0x40 +#define DIAG_PASS_THRU 0x41 + +#define SENSE_CONFIG 0x50 +#define SET_CONFIG 0x51 +typedef struct { + __u32 cfg_sig; + __u16 compat_port; + __u8 data_dist_mode; + __u8 surf_an_ctrl; + __u16 ctlr_phys_drv; + __u16 log_unit_phys_drv; + __u16 fault_tol_mode; + __u8 phys_drv_param[16]; + drv_param_t drv; + __u32 drv_asgn_map; + __u16 dist_factor; + __u32 spare_asgn_map; + __u8 reserved[6]; + __u16 os; + __u8 ctlr_order; + __u8 extra_info; + __u32 data_offs; + __u8 parity_backedout_write_drvs; + __u8 parity_dist_mode; + __u8 parity_shift_fact; + __u8 bios_disable_flag; + __u32 blks_on_vol; + __u32 blks_per_drv; + __u8 scratch[16]; + __u16 big_drv_map[8]; + __u16 big_spare_map[8]; + __u8 ss_source_vol; + __u8 mix_drv_cap_range; + struct { + __u16 big_drv_map[8]; + __u32 blks_per_drv; + __u16 fault_tol_mode; + __u16 dist_factor; + } MDC_range[4]; + __u8 reserved1[248]; +} config_t; + +#define BYPASS_VOL_STATE 0x52 +#define SS_CREATE_VOL 0x53 +#define CHANGE_CONFIG 0x54 +#define SENSE_ORIG_CONF 0x55 +#define REORDER_LOG_DRV 0x56 +typedef struct { + __u8 old_units[32]; +} reorder_log_drv_t; + +#define LABEL_LOG_DRV 0x57 +typedef struct { + __u8 log_drv_label[64]; +} label_log_drv_t; + +#define SS_TO_VOL 0x58 + +#define SET_SURF_DELAY 0x60 +typedef struct { + __u16 delay; + __u8 reserved[510]; +} surf_delay_t; + +#define SET_OVERHEAT_DELAY 0x61 +typedef struct { + __u16 delay; +} overhead_delay_t; + +#define SET_MP_DELAY +typedef struct { + __u16 delay; + __u8 reserved[510]; +} mp_delay_t; + +#define PASSTHRU_A 0x91 +typedef struct { + __u8 target; + __u8 bus; + __u8 lun; + __u32 timeout; + __u32 flags; + __u8 status; + __u8 error; + __u8 cdb_len; + __u8 sense_error; + __u8 sense_key; + __u32 sense_info; + __u8 sense_code; + __u8 sense_qual; + __u32 residual; + __u8 reserved[4]; + __u8 cdb[12]; +} scsi_param_t; + +#define RESUME_BACKGROUND_ACTIVITY 0x99 +#define SENSE_CONTROLLER_PERFORMANCE 0xa8 +#define FLUSH_CACHE 0xc2 +#define COLLECT_BUFFER 0xd2 +#define READ_FLASH_ROM 0xf6 +#define WRITE_FLASH_ROM 0xf7 +#pragma pack() + +#endif /* ARRAYCMD_H */ diff --git a/drivers/block/ida_ioctl.h b/drivers/block/ida_ioctl.h new file mode 100644 index 00000000000..888fff9caed --- /dev/null +++ b/drivers/block/ida_ioctl.h @@ -0,0 +1,87 @@ +/* + * Disk Array driver for Compaq SMART2 Controllers + * Copyright 1998 Compaq Computer Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + */ +#ifndef IDA_IOCTL_H +#define IDA_IOCTL_H + +#include "ida_cmd.h" +#include "cpqarray.h" + +#define IDAGETDRVINFO 0x27272828 +#define IDAPASSTHRU 0x28282929 +#define IDAGETCTLRSIG 0x29293030 +#define IDAREVALIDATEVOLS 0x30303131 +#define IDADRIVERVERSION 0x31313232 +#define IDAGETPCIINFO 0x32323333 + +typedef struct _ida_pci_info_struct +{ + unsigned char bus; + unsigned char dev_fn; + __u32 board_id; +} ida_pci_info_struct; +/* + * Normally, the ioctl determines the logical unit for this command by + * the major,minor number of the fd passed to ioctl. If you need to send + * a command to a different/nonexistant unit (such as during config), you + * can override the normal behavior by setting the unit valid bit. (Normally, + * it should be zero) The controller the command is sent to is still + * determined by the major number of the open device. + */ + +#define UNITVALID 0x80 +typedef struct { + __u8 cmd; + __u8 rcode; + __u8 unit; + __u32 blk; + __u16 blk_cnt; + +/* currently, sg_cnt is assumed to be 1: only the 0th element of sg is used */ + struct { + void __user *addr; + size_t size; + } sg[SG_MAX]; + int sg_cnt; + + union ctlr_cmds { + drv_info_t drv; + unsigned char buf[1024]; + + id_ctlr_t id_ctlr; + drv_param_t drv_param; + id_log_drv_t id_log_drv; + id_log_drv_ext_t id_log_drv_ext; + sense_log_drv_stat_t sense_log_drv_stat; + id_phys_drv_t id_phys_drv; + blink_drv_leds_t blink_drv_leds; + sense_blink_leds_t sense_blink_leds; + config_t config; + reorder_log_drv_t reorder_log_drv; + label_log_drv_t label_log_drv; + surf_delay_t surf_delay; + overhead_delay_t overhead_delay; + mp_delay_t mp_delay; + scsi_param_t scsi_param; + } c; +} ida_ioctl_t; + +#endif /* IDA_IOCTL_H */ diff --git a/drivers/block/ioctl.c b/drivers/block/ioctl.c new file mode 100644 index 00000000000..5e03f5157ef --- /dev/null +++ b/drivers/block/ioctl.c @@ -0,0 +1,239 @@ +#include <linux/sched.h> /* for capable() */ +#include <linux/blkdev.h> +#include <linux/blkpg.h> +#include <linux/backing-dev.h> +#include <linux/buffer_head.h> +#include <linux/smp_lock.h> +#include <asm/uaccess.h> + +static int blkpg_ioctl(struct block_device *bdev, struct blkpg_ioctl_arg __user *arg) +{ + struct block_device *bdevp; + struct gendisk *disk; + struct blkpg_ioctl_arg a; + struct blkpg_partition p; + long long start, length; + int part; + int i; + + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (copy_from_user(&a, arg, sizeof(struct blkpg_ioctl_arg))) + return -EFAULT; + if (copy_from_user(&p, a.data, sizeof(struct blkpg_partition))) + return -EFAULT; + disk = bdev->bd_disk; + if (bdev != bdev->bd_contains) + return -EINVAL; + part = p.pno; + if (part <= 0 || part >= disk->minors) + return -EINVAL; + switch (a.op) { + case BLKPG_ADD_PARTITION: + start = p.start >> 9; + length = p.length >> 9; + /* check for fit in a hd_struct */ + if (sizeof(sector_t) == sizeof(long) && + sizeof(long long) > sizeof(long)) { + long pstart = start, plength = length; + if (pstart != start || plength != length + || pstart < 0 || plength < 0) + return -EINVAL; + } + /* partition number in use? */ + down(&bdev->bd_sem); + if (disk->part[part - 1]) { + up(&bdev->bd_sem); + return -EBUSY; + } + /* overlap? */ + for (i = 0; i < disk->minors - 1; i++) { + struct hd_struct *s = disk->part[i]; + + if (!s) + continue; + if (!(start+length <= s->start_sect || + start >= s->start_sect + s->nr_sects)) { + up(&bdev->bd_sem); + return -EBUSY; + } + } + /* all seems OK */ + add_partition(disk, part, start, length); + up(&bdev->bd_sem); + return 0; + case BLKPG_DEL_PARTITION: + if (!disk->part[part-1]) + return -ENXIO; + if (disk->part[part - 1]->nr_sects == 0) + return -ENXIO; + bdevp = bdget_disk(disk, part); + if (!bdevp) + return -ENOMEM; + down(&bdevp->bd_sem); + if (bdevp->bd_openers) { + up(&bdevp->bd_sem); + bdput(bdevp); + return -EBUSY; + } + /* all seems OK */ + fsync_bdev(bdevp); + invalidate_bdev(bdevp, 0); + + down(&bdev->bd_sem); + delete_partition(disk, part); + up(&bdev->bd_sem); + up(&bdevp->bd_sem); + bdput(bdevp); + + return 0; + default: + return -EINVAL; + } +} + +static int blkdev_reread_part(struct block_device *bdev) +{ + struct gendisk *disk = bdev->bd_disk; + int res; + + if (disk->minors == 1 || bdev != bdev->bd_contains) + return -EINVAL; + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (down_trylock(&bdev->bd_sem)) + return -EBUSY; + res = rescan_partitions(disk, bdev); + up(&bdev->bd_sem); + return res; +} + +static int put_ushort(unsigned long arg, unsigned short val) +{ + return put_user(val, (unsigned short __user *)arg); +} + +static int put_int(unsigned long arg, int val) +{ + return put_user(val, (int __user *)arg); +} + +static int put_long(unsigned long arg, long val) +{ + return put_user(val, (long __user *)arg); +} + +static int put_ulong(unsigned long arg, unsigned long val) +{ + return put_user(val, (unsigned long __user *)arg); +} + +static int put_u64(unsigned long arg, u64 val) +{ + return put_user(val, (u64 __user *)arg); +} + +int blkdev_ioctl(struct inode *inode, struct file *file, unsigned cmd, + unsigned long arg) +{ + struct block_device *bdev = inode->i_bdev; + struct gendisk *disk = bdev->bd_disk; + struct backing_dev_info *bdi; + int ret, n; + + switch (cmd) { + case BLKRAGET: + case BLKFRAGET: + if (!arg) + return -EINVAL; + bdi = blk_get_backing_dev_info(bdev); + if (bdi == NULL) + return -ENOTTY; + return put_long(arg, (bdi->ra_pages * PAGE_CACHE_SIZE) / 512); + case BLKROGET: + return put_int(arg, bdev_read_only(bdev) != 0); + case BLKBSZGET: /* get the logical block size (cf. BLKSSZGET) */ + return put_int(arg, block_size(bdev)); + case BLKSSZGET: /* get block device hardware sector size */ + return put_int(arg, bdev_hardsect_size(bdev)); + case BLKSECTGET: + return put_ushort(arg, bdev_get_queue(bdev)->max_sectors); + case BLKRASET: + case BLKFRASET: + if(!capable(CAP_SYS_ADMIN)) + return -EACCES; + bdi = blk_get_backing_dev_info(bdev); + if (bdi == NULL) + return -ENOTTY; + bdi->ra_pages = (arg * 512) / PAGE_CACHE_SIZE; + return 0; + case BLKBSZSET: + /* set the logical block size */ + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (!arg) + return -EINVAL; + if (get_user(n, (int __user *) arg)) + return -EFAULT; + if (bd_claim(bdev, file) < 0) + return -EBUSY; + ret = set_blocksize(bdev, n); + bd_release(bdev); + return ret; + case BLKPG: + return blkpg_ioctl(bdev, (struct blkpg_ioctl_arg __user *) arg); + case BLKRRPART: + return blkdev_reread_part(bdev); + case BLKGETSIZE: + if ((bdev->bd_inode->i_size >> 9) > ~0UL) + return -EFBIG; + return put_ulong(arg, bdev->bd_inode->i_size >> 9); + case BLKGETSIZE64: + return put_u64(arg, bdev->bd_inode->i_size); + case BLKFLSBUF: + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (disk->fops->ioctl) { + ret = disk->fops->ioctl(inode, file, cmd, arg); + /* -EINVAL to handle old uncorrected drivers */ + if (ret != -EINVAL && ret != -ENOTTY) + return ret; + } + fsync_bdev(bdev); + invalidate_bdev(bdev, 0); + return 0; + case BLKROSET: + if (disk->fops->ioctl) { + ret = disk->fops->ioctl(inode, file, cmd, arg); + /* -EINVAL to handle old uncorrected drivers */ + if (ret != -EINVAL && ret != -ENOTTY) + return ret; + } + if (!capable(CAP_SYS_ADMIN)) + return -EACCES; + if (get_user(n, (int __user *)(arg))) + return -EFAULT; + set_device_ro(bdev, n); + return 0; + default: + if (disk->fops->ioctl) + return disk->fops->ioctl(inode, file, cmd, arg); + } + return -ENOTTY; +} + +/* Most of the generic ioctls are handled in the normal fallback path. + This assumes the blkdev's low level compat_ioctl always returns + ENOIOCTLCMD for unknown ioctls. */ +long compat_blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg) +{ + struct block_device *bdev = file->f_dentry->d_inode->i_bdev; + struct gendisk *disk = bdev->bd_disk; + int ret = -ENOIOCTLCMD; + if (disk->fops->compat_ioctl) { + lock_kernel(); + ret = disk->fops->compat_ioctl(file, cmd, arg); + unlock_kernel(); + } + return ret; +} diff --git a/drivers/block/ll_rw_blk.c b/drivers/block/ll_rw_blk.c new file mode 100644 index 00000000000..02242e8ba99 --- /dev/null +++ b/drivers/block/ll_rw_blk.c @@ -0,0 +1,3642 @@ +/* + * linux/drivers/block/ll_rw_blk.c + * + * Copyright (C) 1991, 1992 Linus Torvalds + * Copyright (C) 1994, Karl Keyte: Added support for disk statistics + * Elevator latency, (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE + * Queue request tables / lock, selectable elevator, Jens Axboe <axboe@suse.de> + * kernel-doc documentation started by NeilBrown <neilb@cse.unsw.edu.au> - July2000 + * bio rewrite, highmem i/o, etc, Jens Axboe <axboe@suse.de> - may 2001 + */ + +/* + * This handles all read/write requests to block devices + */ +#include <linux/config.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/backing-dev.h> +#include <linux/bio.h> +#include <linux/blkdev.h> +#include <linux/highmem.h> +#include <linux/mm.h> +#include <linux/kernel_stat.h> +#include <linux/string.h> +#include <linux/init.h> +#include <linux/bootmem.h> /* for max_pfn/max_low_pfn */ +#include <linux/completion.h> +#include <linux/slab.h> +#include <linux/swap.h> +#include <linux/writeback.h> + +/* + * for max sense size + */ +#include <scsi/scsi_cmnd.h> + +static void blk_unplug_work(void *data); +static void blk_unplug_timeout(unsigned long data); + +/* + * For the allocated request tables + */ +static kmem_cache_t *request_cachep; + +/* + * For queue allocation + */ +static kmem_cache_t *requestq_cachep; + +/* + * For io context allocations + */ +static kmem_cache_t *iocontext_cachep; + +static wait_queue_head_t congestion_wqh[2] = { + __WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[0]), + __WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[1]) + }; + +/* + * Controlling structure to kblockd + */ +static struct workqueue_struct *kblockd_workqueue; + +unsigned long blk_max_low_pfn, blk_max_pfn; + +EXPORT_SYMBOL(blk_max_low_pfn); +EXPORT_SYMBOL(blk_max_pfn); + +/* Amount of time in which a process may batch requests */ +#define BLK_BATCH_TIME (HZ/50UL) + +/* Number of requests a "batching" process may submit */ +#define BLK_BATCH_REQ 32 + +/* + * Return the threshold (number of used requests) at which the queue is + * considered to be congested. It include a little hysteresis to keep the + * context switch rate down. + */ +static inline int queue_congestion_on_threshold(struct request_queue *q) +{ + return q->nr_congestion_on; +} + +/* + * The threshold at which a queue is considered to be uncongested + */ +static inline int queue_congestion_off_threshold(struct request_queue *q) +{ + return q->nr_congestion_off; +} + +static void blk_queue_congestion_threshold(struct request_queue *q) +{ + int nr; + + nr = q->nr_requests - (q->nr_requests / 8) + 1; + if (nr > q->nr_requests) + nr = q->nr_requests; + q->nr_congestion_on = nr; + + nr = q->nr_requests - (q->nr_requests / 8) - (q->nr_requests / 16) - 1; + if (nr < 1) + nr = 1; + q->nr_congestion_off = nr; +} + +/* + * A queue has just exitted congestion. Note this in the global counter of + * congested queues, and wake up anyone who was waiting for requests to be + * put back. + */ +static void clear_queue_congested(request_queue_t *q, int rw) +{ + enum bdi_state bit; + wait_queue_head_t *wqh = &congestion_wqh[rw]; + + bit = (rw == WRITE) ? BDI_write_congested : BDI_read_congested; + clear_bit(bit, &q->backing_dev_info.state); + smp_mb__after_clear_bit(); + if (waitqueue_active(wqh)) + wake_up(wqh); +} + +/* + * A queue has just entered congestion. Flag that in the queue's VM-visible + * state flags and increment the global gounter of congested queues. + */ +static void set_queue_congested(request_queue_t *q, int rw) +{ + enum bdi_state bit; + + bit = (rw == WRITE) ? BDI_write_congested : BDI_read_congested; + set_bit(bit, &q->backing_dev_info.state); +} + +/** + * blk_get_backing_dev_info - get the address of a queue's backing_dev_info + * @bdev: device + * + * Locates the passed device's request queue and returns the address of its + * backing_dev_info + * + * Will return NULL if the request queue cannot be located. + */ +struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev) +{ + struct backing_dev_info *ret = NULL; + request_queue_t *q = bdev_get_queue(bdev); + + if (q) + ret = &q->backing_dev_info; + return ret; +} + +EXPORT_SYMBOL(blk_get_backing_dev_info); + +void blk_queue_activity_fn(request_queue_t *q, activity_fn *fn, void *data) +{ + q->activity_fn = fn; + q->activity_data = data; +} + +EXPORT_SYMBOL(blk_queue_activity_fn); + +/** + * blk_queue_prep_rq - set a prepare_request function for queue + * @q: queue + * @pfn: prepare_request function + * + * It's possible for a queue to register a prepare_request callback which + * is invoked before the request is handed to the request_fn. The goal of + * the function is to prepare a request for I/O, it can be used to build a + * cdb from the request data for instance. + * + */ +void blk_queue_prep_rq(request_queue_t *q, prep_rq_fn *pfn) +{ + q->prep_rq_fn = pfn; +} + +EXPORT_SYMBOL(blk_queue_prep_rq); + +/** + * blk_queue_merge_bvec - set a merge_bvec function for queue + * @q: queue + * @mbfn: merge_bvec_fn + * + * Usually queues have static limitations on the max sectors or segments that + * we can put in a request. Stacking drivers may have some settings that + * are dynamic, and thus we have to query the queue whether it is ok to + * add a new bio_vec to a bio at a given offset or not. If the block device + * has such limitations, it needs to register a merge_bvec_fn to control + * the size of bio's sent to it. Note that a block device *must* allow a + * single page to be added to an empty bio. The block device driver may want + * to use the bio_split() function to deal with these bio's. By default + * no merge_bvec_fn is defined for a queue, and only the fixed limits are + * honored. + */ +void blk_queue_merge_bvec(request_queue_t *q, merge_bvec_fn *mbfn) +{ + q->merge_bvec_fn = mbfn; +} + +EXPORT_SYMBOL(blk_queue_merge_bvec); + +/** + * blk_queue_make_request - define an alternate make_request function for a device + * @q: the request queue for the device to be affected + * @mfn: the alternate make_request function + * + * Description: + * The normal way for &struct bios to be passed to a device + * driver is for them to be collected into requests on a request + * queue, and then to allow the device driver to select requests + * off that queue when it is ready. This works well for many block + * devices. However some block devices (typically virtual devices + * such as md or lvm) do not benefit from the processing on the + * request queue, and are served best by having the requests passed + * directly to them. This can be achieved by providing a function + * to blk_queue_make_request(). + * + * Caveat: + * The driver that does this *must* be able to deal appropriately + * with buffers in "highmemory". This can be accomplished by either calling + * __bio_kmap_atomic() to get a temporary kernel mapping, or by calling + * blk_queue_bounce() to create a buffer in normal memory. + **/ +void blk_queue_make_request(request_queue_t * q, make_request_fn * mfn) +{ + /* + * set defaults + */ + q->nr_requests = BLKDEV_MAX_RQ; + q->max_phys_segments = MAX_PHYS_SEGMENTS; + q->max_hw_segments = MAX_HW_SEGMENTS; + q->make_request_fn = mfn; + q->backing_dev_info.ra_pages = (VM_MAX_READAHEAD * 1024) / PAGE_CACHE_SIZE; + q->backing_dev_info.state = 0; + q->backing_dev_info.capabilities = BDI_CAP_MAP_COPY; + blk_queue_max_sectors(q, MAX_SECTORS); + blk_queue_hardsect_size(q, 512); + blk_queue_dma_alignment(q, 511); + blk_queue_congestion_threshold(q); + q->nr_batching = BLK_BATCH_REQ; + + q->unplug_thresh = 4; /* hmm */ + q->unplug_delay = (3 * HZ) / 1000; /* 3 milliseconds */ + if (q->unplug_delay == 0) + q->unplug_delay = 1; + + INIT_WORK(&q->unplug_work, blk_unplug_work, q); + + q->unplug_timer.function = blk_unplug_timeout; + q->unplug_timer.data = (unsigned long)q; + + /* + * by default assume old behaviour and bounce for any highmem page + */ + blk_queue_bounce_limit(q, BLK_BOUNCE_HIGH); + + blk_queue_activity_fn(q, NULL, NULL); + + INIT_LIST_HEAD(&q->drain_list); +} + +EXPORT_SYMBOL(blk_queue_make_request); + +static inline void rq_init(request_queue_t *q, struct request *rq) +{ + INIT_LIST_HEAD(&rq->queuelist); + + rq->errors = 0; + rq->rq_status = RQ_ACTIVE; + rq->bio = rq->biotail = NULL; + rq->buffer = NULL; + rq->ref_count = 1; + rq->q = q; + rq->waiting = NULL; + rq->special = NULL; + rq->data_len = 0; + rq->data = NULL; + rq->sense = NULL; + rq->end_io = NULL; + rq->end_io_data = NULL; +} + +/** + * blk_queue_ordered - does this queue support ordered writes + * @q: the request queue + * @flag: see below + * + * Description: + * For journalled file systems, doing ordered writes on a commit + * block instead of explicitly doing wait_on_buffer (which is bad + * for performance) can be a big win. Block drivers supporting this + * feature should call this function and indicate so. + * + **/ +void blk_queue_ordered(request_queue_t *q, int flag) +{ + switch (flag) { + case QUEUE_ORDERED_NONE: + if (q->flush_rq) + kmem_cache_free(request_cachep, q->flush_rq); + q->flush_rq = NULL; + q->ordered = flag; + break; + case QUEUE_ORDERED_TAG: + q->ordered = flag; + break; + case QUEUE_ORDERED_FLUSH: + q->ordered = flag; + if (!q->flush_rq) + q->flush_rq = kmem_cache_alloc(request_cachep, + GFP_KERNEL); + break; + default: + printk("blk_queue_ordered: bad value %d\n", flag); + break; + } +} + +EXPORT_SYMBOL(blk_queue_ordered); + +/** + * blk_queue_issue_flush_fn - set function for issuing a flush + * @q: the request queue + * @iff: the function to be called issuing the flush + * + * Description: + * If a driver supports issuing a flush command, the support is notified + * to the block layer by defining it through this call. + * + **/ +void blk_queue_issue_flush_fn(request_queue_t *q, issue_flush_fn *iff) +{ + q->issue_flush_fn = iff; +} + +EXPORT_SYMBOL(blk_queue_issue_flush_fn); + +/* + * Cache flushing for ordered writes handling + */ +static void blk_pre_flush_end_io(struct request *flush_rq) +{ + struct request *rq = flush_rq->end_io_data; + request_queue_t *q = rq->q; + + rq->flags |= REQ_BAR_PREFLUSH; + + if (!flush_rq->errors) + elv_requeue_request(q, rq); + else { + q->end_flush_fn(q, flush_rq); + clear_bit(QUEUE_FLAG_FLUSH, &q->queue_flags); + q->request_fn(q); + } +} + +static void blk_post_flush_end_io(struct request *flush_rq) +{ + struct request *rq = flush_rq->end_io_data; + request_queue_t *q = rq->q; + + rq->flags |= REQ_BAR_POSTFLUSH; + + q->end_flush_fn(q, flush_rq); + clear_bit(QUEUE_FLAG_FLUSH, &q->queue_flags); + q->request_fn(q); +} + +struct request *blk_start_pre_flush(request_queue_t *q, struct request *rq) +{ + struct request *flush_rq = q->flush_rq; + + BUG_ON(!blk_barrier_rq(rq)); + + if (test_and_set_bit(QUEUE_FLAG_FLUSH, &q->queue_flags)) + return NULL; + + rq_init(q, flush_rq); + flush_rq->elevator_private = NULL; + flush_rq->flags = REQ_BAR_FLUSH; + flush_rq->rq_disk = rq->rq_disk; + flush_rq->rl = NULL; + + /* + * prepare_flush returns 0 if no flush is needed, just mark both + * pre and post flush as done in that case + */ + if (!q->prepare_flush_fn(q, flush_rq)) { + rq->flags |= REQ_BAR_PREFLUSH | REQ_BAR_POSTFLUSH; + clear_bit(QUEUE_FLAG_FLUSH, &q->queue_flags); + return rq; + } + + /* + * some drivers dequeue requests right away, some only after io + * completion. make sure the request is dequeued. + */ + if (!list_empty(&rq->queuelist)) + blkdev_dequeue_request(rq); + + elv_deactivate_request(q, rq); + + flush_rq->end_io_data = rq; + flush_rq->end_io = blk_pre_flush_end_io; + + __elv_add_request(q, flush_rq, ELEVATOR_INSERT_FRONT, 0); + return flush_rq; +} + +static void blk_start_post_flush(request_queue_t *q, struct request *rq) +{ + struct request *flush_rq = q->flush_rq; + + BUG_ON(!blk_barrier_rq(rq)); + + rq_init(q, flush_rq); + flush_rq->elevator_private = NULL; + flush_rq->flags = REQ_BAR_FLUSH; + flush_rq->rq_disk = rq->rq_disk; + flush_rq->rl = NULL; + + if (q->prepare_flush_fn(q, flush_rq)) { + flush_rq->end_io_data = rq; + flush_rq->end_io = blk_post_flush_end_io; + + __elv_add_request(q, flush_rq, ELEVATOR_INSERT_FRONT, 0); + q->request_fn(q); + } +} + +static inline int blk_check_end_barrier(request_queue_t *q, struct request *rq, + int sectors) +{ + if (sectors > rq->nr_sectors) + sectors = rq->nr_sectors; + + rq->nr_sectors -= sectors; + return rq->nr_sectors; +} + +static int __blk_complete_barrier_rq(request_queue_t *q, struct request *rq, + int sectors, int queue_locked) +{ + if (q->ordered != QUEUE_ORDERED_FLUSH) + return 0; + if (!blk_fs_request(rq) || !blk_barrier_rq(rq)) + return 0; + if (blk_barrier_postflush(rq)) + return 0; + + if (!blk_check_end_barrier(q, rq, sectors)) { + unsigned long flags = 0; + + if (!queue_locked) + spin_lock_irqsave(q->queue_lock, flags); + + blk_start_post_flush(q, rq); + + if (!queue_locked) + spin_unlock_irqrestore(q->queue_lock, flags); + } + + return 1; +} + +/** + * blk_complete_barrier_rq - complete possible barrier request + * @q: the request queue for the device + * @rq: the request + * @sectors: number of sectors to complete + * + * Description: + * Used in driver end_io handling to determine whether to postpone + * completion of a barrier request until a post flush has been done. This + * is the unlocked variant, used if the caller doesn't already hold the + * queue lock. + **/ +int blk_complete_barrier_rq(request_queue_t *q, struct request *rq, int sectors) +{ + return __blk_complete_barrier_rq(q, rq, sectors, 0); +} +EXPORT_SYMBOL(blk_complete_barrier_rq); + +/** + * blk_complete_barrier_rq_locked - complete possible barrier request + * @q: the request queue for the device + * @rq: the request + * @sectors: number of sectors to complete + * + * Description: + * See blk_complete_barrier_rq(). This variant must be used if the caller + * holds the queue lock. + **/ +int blk_complete_barrier_rq_locked(request_queue_t *q, struct request *rq, + int sectors) +{ + return __blk_complete_barrier_rq(q, rq, sectors, 1); +} +EXPORT_SYMBOL(blk_complete_barrier_rq_locked); + +/** + * blk_queue_bounce_limit - set bounce buffer limit for queue + * @q: the request queue for the device + * @dma_addr: bus address limit + * + * Description: + * Different hardware can have different requirements as to what pages + * it can do I/O directly to. A low level driver can call + * blk_queue_bounce_limit to have lower memory pages allocated as bounce + * buffers for doing I/O to pages residing above @page. By default + * the block layer sets this to the highest numbered "low" memory page. + **/ +void blk_queue_bounce_limit(request_queue_t *q, u64 dma_addr) +{ + unsigned long bounce_pfn = dma_addr >> PAGE_SHIFT; + + /* + * set appropriate bounce gfp mask -- unfortunately we don't have a + * full 4GB zone, so we have to resort to low memory for any bounces. + * ISA has its own < 16MB zone. + */ + if (bounce_pfn < blk_max_low_pfn) { + BUG_ON(dma_addr < BLK_BOUNCE_ISA); + init_emergency_isa_pool(); + q->bounce_gfp = GFP_NOIO | GFP_DMA; + } else + q->bounce_gfp = GFP_NOIO; + + q->bounce_pfn = bounce_pfn; +} + +EXPORT_SYMBOL(blk_queue_bounce_limit); + +/** + * blk_queue_max_sectors - set max sectors for a request for this queue + * @q: the request queue for the device + * @max_sectors: max sectors in the usual 512b unit + * + * Description: + * Enables a low level driver to set an upper limit on the size of + * received requests. + **/ +void blk_queue_max_sectors(request_queue_t *q, unsigned short max_sectors) +{ + if ((max_sectors << 9) < PAGE_CACHE_SIZE) { + max_sectors = 1 << (PAGE_CACHE_SHIFT - 9); + printk("%s: set to minimum %d\n", __FUNCTION__, max_sectors); + } + + q->max_sectors = q->max_hw_sectors = max_sectors; +} + +EXPORT_SYMBOL(blk_queue_max_sectors); + +/** + * blk_queue_max_phys_segments - set max phys segments for a request for this queue + * @q: the request queue for the device + * @max_segments: max number of segments + * + * Description: + * Enables a low level driver to set an upper limit on the number of + * physical data segments in a request. This would be the largest sized + * scatter list the driver could handle. + **/ +void blk_queue_max_phys_segments(request_queue_t *q, unsigned short max_segments) +{ + if (!max_segments) { + max_segments = 1; + printk("%s: set to minimum %d\n", __FUNCTION__, max_segments); + } + + q->max_phys_segments = max_segments; +} + +EXPORT_SYMBOL(blk_queue_max_phys_segments); + +/** + * blk_queue_max_hw_segments - set max hw segments for a request for this queue + * @q: the request queue for the device + * @max_segments: max number of segments + * + * Description: + * Enables a low level driver to set an upper limit on the number of + * hw data segments in a request. This would be the largest number of + * address/length pairs the host adapter can actually give as once + * to the device. + **/ +void blk_queue_max_hw_segments(request_queue_t *q, unsigned short max_segments) +{ + if (!max_segments) { + max_segments = 1; + printk("%s: set to minimum %d\n", __FUNCTION__, max_segments); + } + + q->max_hw_segments = max_segments; +} + +EXPORT_SYMBOL(blk_queue_max_hw_segments); + +/** + * blk_queue_max_segment_size - set max segment size for blk_rq_map_sg + * @q: the request queue for the device + * @max_size: max size of segment in bytes + * + * Description: + * Enables a low level driver to set an upper limit on the size of a + * coalesced segment + **/ +void blk_queue_max_segment_size(request_queue_t *q, unsigned int max_size) +{ + if (max_size < PAGE_CACHE_SIZE) { + max_size = PAGE_CACHE_SIZE; + printk("%s: set to minimum %d\n", __FUNCTION__, max_size); + } + + q->max_segment_size = max_size; +} + +EXPORT_SYMBOL(blk_queue_max_segment_size); + +/** + * blk_queue_hardsect_size - set hardware sector size for the queue + * @q: the request queue for the device + * @size: the hardware sector size, in bytes + * + * Description: + * This should typically be set to the lowest possible sector size + * that the hardware can operate on (possible without reverting to + * even internal read-modify-write operations). Usually the default + * of 512 covers most hardware. + **/ +void blk_queue_hardsect_size(request_queue_t *q, unsigned short size) +{ + q->hardsect_size = size; +} + +EXPORT_SYMBOL(blk_queue_hardsect_size); + +/* + * Returns the minimum that is _not_ zero, unless both are zero. + */ +#define min_not_zero(l, r) (l == 0) ? r : ((r == 0) ? l : min(l, r)) + +/** + * blk_queue_stack_limits - inherit underlying queue limits for stacked drivers + * @t: the stacking driver (top) + * @b: the underlying device (bottom) + **/ +void blk_queue_stack_limits(request_queue_t *t, request_queue_t *b) +{ + /* zero is "infinity" */ + t->max_sectors = t->max_hw_sectors = + min_not_zero(t->max_sectors,b->max_sectors); + + t->max_phys_segments = min(t->max_phys_segments,b->max_phys_segments); + t->max_hw_segments = min(t->max_hw_segments,b->max_hw_segments); + t->max_segment_size = min(t->max_segment_size,b->max_segment_size); + t->hardsect_size = max(t->hardsect_size,b->hardsect_size); +} + +EXPORT_SYMBOL(blk_queue_stack_limits); + +/** + * blk_queue_segment_boundary - set boundary rules for segment merging + * @q: the request queue for the device + * @mask: the memory boundary mask + **/ +void blk_queue_segment_boundary(request_queue_t *q, unsigned long mask) +{ + if (mask < PAGE_CACHE_SIZE - 1) { + mask = PAGE_CACHE_SIZE - 1; + printk("%s: set to minimum %lx\n", __FUNCTION__, mask); + } + + q->seg_boundary_mask = mask; +} + +EXPORT_SYMBOL(blk_queue_segment_boundary); + +/** + * blk_queue_dma_alignment - set dma length and memory alignment + * @q: the request queue for the device + * @mask: alignment mask + * + * description: + * set required memory and length aligment for direct dma transactions. + * this is used when buiding direct io requests for the queue. + * + **/ +void blk_queue_dma_alignment(request_queue_t *q, int mask) +{ + q->dma_alignment = mask; +} + +EXPORT_SYMBOL(blk_queue_dma_alignment); + +/** + * blk_queue_find_tag - find a request by its tag and queue + * + * @q: The request queue for the device + * @tag: The tag of the request + * + * Notes: + * Should be used when a device returns a tag and you want to match + * it with a request. + * + * no locks need be held. + **/ +struct request *blk_queue_find_tag(request_queue_t *q, int tag) +{ + struct blk_queue_tag *bqt = q->queue_tags; + + if (unlikely(bqt == NULL || tag >= bqt->real_max_depth)) + return NULL; + + return bqt->tag_index[tag]; +} + +EXPORT_SYMBOL(blk_queue_find_tag); + +/** + * __blk_queue_free_tags - release tag maintenance info + * @q: the request queue for the device + * + * Notes: + * blk_cleanup_queue() will take care of calling this function, if tagging + * has been used. So there's no need to call this directly. + **/ +static void __blk_queue_free_tags(request_queue_t *q) +{ + struct blk_queue_tag *bqt = q->queue_tags; + + if (!bqt) + return; + + if (atomic_dec_and_test(&bqt->refcnt)) { + BUG_ON(bqt->busy); + BUG_ON(!list_empty(&bqt->busy_list)); + + kfree(bqt->tag_index); + bqt->tag_index = NULL; + + kfree(bqt->tag_map); + bqt->tag_map = NULL; + + kfree(bqt); + } + + q->queue_tags = NULL; + q->queue_flags &= ~(1 << QUEUE_FLAG_QUEUED); +} + +/** + * blk_queue_free_tags - release tag maintenance info + * @q: the request queue for the device + * + * Notes: + * This is used to disabled tagged queuing to a device, yet leave + * queue in function. + **/ +void blk_queue_free_tags(request_queue_t *q) +{ + clear_bit(QUEUE_FLAG_QUEUED, &q->queue_flags); +} + +EXPORT_SYMBOL(blk_queue_free_tags); + +static int +init_tag_map(request_queue_t *q, struct blk_queue_tag *tags, int depth) +{ + int bits, i; + struct request **tag_index; + unsigned long *tag_map; + + if (depth > q->nr_requests * 2) { + depth = q->nr_requests * 2; + printk(KERN_ERR "%s: adjusted depth to %d\n", + __FUNCTION__, depth); + } + + tag_index = kmalloc(depth * sizeof(struct request *), GFP_ATOMIC); + if (!tag_index) + goto fail; + + bits = (depth / BLK_TAGS_PER_LONG) + 1; + tag_map = kmalloc(bits * sizeof(unsigned long), GFP_ATOMIC); + if (!tag_map) + goto fail; + + memset(tag_index, 0, depth * sizeof(struct request *)); + memset(tag_map, 0, bits * sizeof(unsigned long)); + tags->max_depth = depth; + tags->real_max_depth = bits * BITS_PER_LONG; + tags->tag_index = tag_index; + tags->tag_map = tag_map; + + /* + * set the upper bits if the depth isn't a multiple of the word size + */ + for (i = depth; i < bits * BLK_TAGS_PER_LONG; i++) + __set_bit(i, tag_map); + + return 0; +fail: + kfree(tag_index); + return -ENOMEM; +} + +/** + * blk_queue_init_tags - initialize the queue tag info + * @q: the request queue for the device + * @depth: the maximum queue depth supported + * @tags: the tag to use + **/ +int blk_queue_init_tags(request_queue_t *q, int depth, + struct blk_queue_tag *tags) +{ + int rc; + + BUG_ON(tags && q->queue_tags && tags != q->queue_tags); + + if (!tags && !q->queue_tags) { + tags = kmalloc(sizeof(struct blk_queue_tag), GFP_ATOMIC); + if (!tags) + goto fail; + + if (init_tag_map(q, tags, depth)) + goto fail; + + INIT_LIST_HEAD(&tags->busy_list); + tags->busy = 0; + atomic_set(&tags->refcnt, 1); + } else if (q->queue_tags) { + if ((rc = blk_queue_resize_tags(q, depth))) + return rc; + set_bit(QUEUE_FLAG_QUEUED, &q->queue_flags); + return 0; + } else + atomic_inc(&tags->refcnt); + + /* + * assign it, all done + */ + q->queue_tags = tags; + q->queue_flags |= (1 << QUEUE_FLAG_QUEUED); + return 0; +fail: + kfree(tags); + return -ENOMEM; +} + +EXPORT_SYMBOL(blk_queue_init_tags); + +/** + * blk_queue_resize_tags - change the queueing depth + * @q: the request queue for the device + * @new_depth: the new max command queueing depth + * + * Notes: + * Must be called with the queue lock held. + **/ +int blk_queue_resize_tags(request_queue_t *q, int new_depth) +{ + struct blk_queue_tag *bqt = q->queue_tags; + struct request **tag_index; + unsigned long *tag_map; + int bits, max_depth; + + if (!bqt) + return -ENXIO; + + /* + * don't bother sizing down + */ + if (new_depth <= bqt->real_max_depth) { + bqt->max_depth = new_depth; + return 0; + } + + /* + * save the old state info, so we can copy it back + */ + tag_index = bqt->tag_index; + tag_map = bqt->tag_map; + max_depth = bqt->real_max_depth; + + if (init_tag_map(q, bqt, new_depth)) + return -ENOMEM; + + memcpy(bqt->tag_index, tag_index, max_depth * sizeof(struct request *)); + bits = max_depth / BLK_TAGS_PER_LONG; + memcpy(bqt->tag_map, tag_map, bits * sizeof(unsigned long)); + + kfree(tag_index); + kfree(tag_map); + return 0; +} + +EXPORT_SYMBOL(blk_queue_resize_tags); + +/** + * blk_queue_end_tag - end tag operations for a request + * @q: the request queue for the device + * @rq: the request that has completed + * + * Description: + * Typically called when end_that_request_first() returns 0, meaning + * all transfers have been done for a request. It's important to call + * this function before end_that_request_last(), as that will put the + * request back on the free list thus corrupting the internal tag list. + * + * Notes: + * queue lock must be held. + **/ +void blk_queue_end_tag(request_queue_t *q, struct request *rq) +{ + struct blk_queue_tag *bqt = q->queue_tags; + int tag = rq->tag; + + BUG_ON(tag == -1); + + if (unlikely(tag >= bqt->real_max_depth)) + return; + + if (unlikely(!__test_and_clear_bit(tag, bqt->tag_map))) { + printk("attempt to clear non-busy tag (%d)\n", tag); + return; + } + + list_del_init(&rq->queuelist); + rq->flags &= ~REQ_QUEUED; + rq->tag = -1; + + if (unlikely(bqt->tag_index[tag] == NULL)) + printk("tag %d is missing\n", tag); + + bqt->tag_index[tag] = NULL; + bqt->busy--; +} + +EXPORT_SYMBOL(blk_queue_end_tag); + +/** + * blk_queue_start_tag - find a free tag and assign it + * @q: the request queue for the device + * @rq: the block request that needs tagging + * + * Description: + * This can either be used as a stand-alone helper, or possibly be + * assigned as the queue &prep_rq_fn (in which case &struct request + * automagically gets a tag assigned). Note that this function + * assumes that any type of request can be queued! if this is not + * true for your device, you must check the request type before + * calling this function. The request will also be removed from + * the request queue, so it's the drivers responsibility to readd + * it if it should need to be restarted for some reason. + * + * Notes: + * queue lock must be held. + **/ +int blk_queue_start_tag(request_queue_t *q, struct request *rq) +{ + struct blk_queue_tag *bqt = q->queue_tags; + unsigned long *map = bqt->tag_map; + int tag = 0; + + if (unlikely((rq->flags & REQ_QUEUED))) { + printk(KERN_ERR + "request %p for device [%s] already tagged %d", + rq, rq->rq_disk ? rq->rq_disk->disk_name : "?", rq->tag); + BUG(); + } + + for (map = bqt->tag_map; *map == -1UL; map++) { + tag += BLK_TAGS_PER_LONG; + + if (tag >= bqt->max_depth) + return 1; + } + + tag += ffz(*map); + __set_bit(tag, bqt->tag_map); + + rq->flags |= REQ_QUEUED; + rq->tag = tag; + bqt->tag_index[tag] = rq; + blkdev_dequeue_request(rq); + list_add(&rq->queuelist, &bqt->busy_list); + bqt->busy++; + return 0; +} + +EXPORT_SYMBOL(blk_queue_start_tag); + +/** + * blk_queue_invalidate_tags - invalidate all pending tags + * @q: the request queue for the device + * + * Description: + * Hardware conditions may dictate a need to stop all pending requests. + * In this case, we will safely clear the block side of the tag queue and + * readd all requests to the request queue in the right order. + * + * Notes: + * queue lock must be held. + **/ +void blk_queue_invalidate_tags(request_queue_t *q) +{ + struct blk_queue_tag *bqt = q->queue_tags; + struct list_head *tmp, *n; + struct request *rq; + + list_for_each_safe(tmp, n, &bqt->busy_list) { + rq = list_entry_rq(tmp); + + if (rq->tag == -1) { + printk("bad tag found on list\n"); + list_del_init(&rq->queuelist); + rq->flags &= ~REQ_QUEUED; + } else + blk_queue_end_tag(q, rq); + + rq->flags &= ~REQ_STARTED; + __elv_add_request(q, rq, ELEVATOR_INSERT_BACK, 0); + } +} + +EXPORT_SYMBOL(blk_queue_invalidate_tags); + +static char *rq_flags[] = { + "REQ_RW", + "REQ_FAILFAST", + "REQ_SOFTBARRIER", + "REQ_HARDBARRIER", + "REQ_CMD", + "REQ_NOMERGE", + "REQ_STARTED", + "REQ_DONTPREP", + "REQ_QUEUED", + "REQ_PC", + "REQ_BLOCK_PC", + "REQ_SENSE", + "REQ_FAILED", + "REQ_QUIET", + "REQ_SPECIAL", + "REQ_DRIVE_CMD", + "REQ_DRIVE_TASK", + "REQ_DRIVE_TASKFILE", + "REQ_PREEMPT", + "REQ_PM_SUSPEND", + "REQ_PM_RESUME", + "REQ_PM_SHUTDOWN", +}; + +void blk_dump_rq_flags(struct request *rq, char *msg) +{ + int bit; + + printk("%s: dev %s: flags = ", msg, + rq->rq_disk ? rq->rq_disk->disk_name : "?"); + bit = 0; + do { + if (rq->flags & (1 << bit)) + printk("%s ", rq_flags[bit]); + bit++; + } while (bit < __REQ_NR_BITS); + + printk("\nsector %llu, nr/cnr %lu/%u\n", (unsigned long long)rq->sector, + rq->nr_sectors, + rq->current_nr_sectors); + printk("bio %p, biotail %p, buffer %p, data %p, len %u\n", rq->bio, rq->biotail, rq->buffer, rq->data, rq->data_len); + + if (rq->flags & (REQ_BLOCK_PC | REQ_PC)) { + printk("cdb: "); + for (bit = 0; bit < sizeof(rq->cmd); bit++) + printk("%02x ", rq->cmd[bit]); + printk("\n"); + } +} + +EXPORT_SYMBOL(blk_dump_rq_flags); + +void blk_recount_segments(request_queue_t *q, struct bio *bio) +{ + struct bio_vec *bv, *bvprv = NULL; + int i, nr_phys_segs, nr_hw_segs, seg_size, hw_seg_size, cluster; + int high, highprv = 1; + + if (unlikely(!bio->bi_io_vec)) + return; + + cluster = q->queue_flags & (1 << QUEUE_FLAG_CLUSTER); + hw_seg_size = seg_size = nr_phys_segs = nr_hw_segs = 0; + bio_for_each_segment(bv, bio, i) { + /* + * the trick here is making sure that a high page is never + * considered part of another segment, since that might + * change with the bounce page. + */ + high = page_to_pfn(bv->bv_page) >= q->bounce_pfn; + if (high || highprv) + goto new_hw_segment; + if (cluster) { + if (seg_size + bv->bv_len > q->max_segment_size) + goto new_segment; + if (!BIOVEC_PHYS_MERGEABLE(bvprv, bv)) + goto new_segment; + if (!BIOVEC_SEG_BOUNDARY(q, bvprv, bv)) + goto new_segment; + if (BIOVEC_VIRT_OVERSIZE(hw_seg_size + bv->bv_len)) + goto new_hw_segment; + + seg_size += bv->bv_len; + hw_seg_size += bv->bv_len; + bvprv = bv; + continue; + } +new_segment: + if (BIOVEC_VIRT_MERGEABLE(bvprv, bv) && + !BIOVEC_VIRT_OVERSIZE(hw_seg_size + bv->bv_len)) { + hw_seg_size += bv->bv_len; + } else { +new_hw_segment: + if (hw_seg_size > bio->bi_hw_front_size) + bio->bi_hw_front_size = hw_seg_size; + hw_seg_size = BIOVEC_VIRT_START_SIZE(bv) + bv->bv_len; + nr_hw_segs++; + } + + nr_phys_segs++; + bvprv = bv; + seg_size = bv->bv_len; + highprv = high; + } + if (hw_seg_size > bio->bi_hw_back_size) + bio->bi_hw_back_size = hw_seg_size; + if (nr_hw_segs == 1 && hw_seg_size > bio->bi_hw_front_size) + bio->bi_hw_front_size = hw_seg_size; + bio->bi_phys_segments = nr_phys_segs; + bio->bi_hw_segments = nr_hw_segs; + bio->bi_flags |= (1 << BIO_SEG_VALID); +} + + +int blk_phys_contig_segment(request_queue_t *q, struct bio *bio, + struct bio *nxt) +{ + if (!(q->queue_flags & (1 << QUEUE_FLAG_CLUSTER))) + return 0; + + if (!BIOVEC_PHYS_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt))) + return 0; + if (bio->bi_size + nxt->bi_size > q->max_segment_size) + return 0; + + /* + * bio and nxt are contigous in memory, check if the queue allows + * these two to be merged into one + */ + if (BIO_SEG_BOUNDARY(q, bio, nxt)) + return 1; + + return 0; +} + +EXPORT_SYMBOL(blk_phys_contig_segment); + +int blk_hw_contig_segment(request_queue_t *q, struct bio *bio, + struct bio *nxt) +{ + if (unlikely(!bio_flagged(bio, BIO_SEG_VALID))) + blk_recount_segments(q, bio); + if (unlikely(!bio_flagged(nxt, BIO_SEG_VALID))) + blk_recount_segments(q, nxt); + if (!BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)) || + BIOVEC_VIRT_OVERSIZE(bio->bi_hw_front_size + bio->bi_hw_back_size)) + return 0; + if (bio->bi_size + nxt->bi_size > q->max_segment_size) + return 0; + + return 1; +} + +EXPORT_SYMBOL(blk_hw_contig_segment); + +/* + * map a request to scatterlist, return number of sg entries setup. Caller + * must make sure sg can hold rq->nr_phys_segments entries + */ +int blk_rq_map_sg(request_queue_t *q, struct request *rq, struct scatterlist *sg) +{ + struct bio_vec *bvec, *bvprv; + struct bio *bio; + int nsegs, i, cluster; + + nsegs = 0; + cluster = q->queue_flags & (1 << QUEUE_FLAG_CLUSTER); + + /* + * for each bio in rq + */ + bvprv = NULL; + rq_for_each_bio(bio, rq) { + /* + * for each segment in bio + */ + bio_for_each_segment(bvec, bio, i) { + int nbytes = bvec->bv_len; + + if (bvprv && cluster) { + if (sg[nsegs - 1].length + nbytes > q->max_segment_size) + goto new_segment; + + if (!BIOVEC_PHYS_MERGEABLE(bvprv, bvec)) + goto new_segment; + if (!BIOVEC_SEG_BOUNDARY(q, bvprv, bvec)) + goto new_segment; + + sg[nsegs - 1].length += nbytes; + } else { +new_segment: + memset(&sg[nsegs],0,sizeof(struct scatterlist)); + sg[nsegs].page = bvec->bv_page; + sg[nsegs].length = nbytes; + sg[nsegs].offset = bvec->bv_offset; + + nsegs++; + } + bvprv = bvec; + } /* segments in bio */ + } /* bios in rq */ + + return nsegs; +} + +EXPORT_SYMBOL(blk_rq_map_sg); + +/* + * the standard queue merge functions, can be overridden with device + * specific ones if so desired + */ + +static inline int ll_new_mergeable(request_queue_t *q, + struct request *req, + struct bio *bio) +{ + int nr_phys_segs = bio_phys_segments(q, bio); + + if (req->nr_phys_segments + nr_phys_segs > q->max_phys_segments) { + req->flags |= REQ_NOMERGE; + if (req == q->last_merge) + q->last_merge = NULL; + return 0; + } + + /* + * A hw segment is just getting larger, bump just the phys + * counter. + */ + req->nr_phys_segments += nr_phys_segs; + return 1; +} + +static inline int ll_new_hw_segment(request_queue_t *q, + struct request *req, + struct bio *bio) +{ + int nr_hw_segs = bio_hw_segments(q, bio); + int nr_phys_segs = bio_phys_segments(q, bio); + + if (req->nr_hw_segments + nr_hw_segs > q->max_hw_segments + || req->nr_phys_segments + nr_phys_segs > q->max_phys_segments) { + req->flags |= REQ_NOMERGE; + if (req == q->last_merge) + q->last_merge = NULL; + return 0; + } + + /* + * This will form the start of a new hw segment. Bump both + * counters. + */ + req->nr_hw_segments += nr_hw_segs; + req->nr_phys_segments += nr_phys_segs; + return 1; +} + +static int ll_back_merge_fn(request_queue_t *q, struct request *req, + struct bio *bio) +{ + int len; + + if (req->nr_sectors + bio_sectors(bio) > q->max_sectors) { + req->flags |= REQ_NOMERGE; + if (req == q->last_merge) + q->last_merge = NULL; + return 0; + } + if (unlikely(!bio_flagged(req->biotail, BIO_SEG_VALID))) + blk_recount_segments(q, req->biotail); + if (unlikely(!bio_flagged(bio, BIO_SEG_VALID))) + blk_recount_segments(q, bio); + len = req->biotail->bi_hw_back_size + bio->bi_hw_front_size; + if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(req->biotail), __BVEC_START(bio)) && + !BIOVEC_VIRT_OVERSIZE(len)) { + int mergeable = ll_new_mergeable(q, req, bio); + + if (mergeable) { + if (req->nr_hw_segments == 1) + req->bio->bi_hw_front_size = len; + if (bio->bi_hw_segments == 1) + bio->bi_hw_back_size = len; + } + return mergeable; + } + + return ll_new_hw_segment(q, req, bio); +} + +static int ll_front_merge_fn(request_queue_t *q, struct request *req, + struct bio *bio) +{ + int len; + + if (req->nr_sectors + bio_sectors(bio) > q->max_sectors) { + req->flags |= REQ_NOMERGE; + if (req == q->last_merge) + q->last_merge = NULL; + return 0; + } + len = bio->bi_hw_back_size + req->bio->bi_hw_front_size; + if (unlikely(!bio_flagged(bio, BIO_SEG_VALID))) + blk_recount_segments(q, bio); + if (unlikely(!bio_flagged(req->bio, BIO_SEG_VALID))) + blk_recount_segments(q, req->bio); + if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(req->bio)) && + !BIOVEC_VIRT_OVERSIZE(len)) { + int mergeable = ll_new_mergeable(q, req, bio); + + if (mergeable) { + if (bio->bi_hw_segments == 1) + bio->bi_hw_front_size = len; + if (req->nr_hw_segments == 1) + req->biotail->bi_hw_back_size = len; + } + return mergeable; + } + + return ll_new_hw_segment(q, req, bio); +} + +static int ll_merge_requests_fn(request_queue_t *q, struct request *req, + struct request *next) +{ + int total_phys_segments = req->nr_phys_segments +next->nr_phys_segments; + int total_hw_segments = req->nr_hw_segments + next->nr_hw_segments; + + /* + * First check if the either of the requests are re-queued + * requests. Can't merge them if they are. + */ + if (req->special || next->special) + return 0; + + /* + * Will it become to large? + */ + if ((req->nr_sectors + next->nr_sectors) > q->max_sectors) + return 0; + + total_phys_segments = req->nr_phys_segments + next->nr_phys_segments; + if (blk_phys_contig_segment(q, req->biotail, next->bio)) + total_phys_segments--; + + if (total_phys_segments > q->max_phys_segments) + return 0; + + total_hw_segments = req->nr_hw_segments + next->nr_hw_segments; + if (blk_hw_contig_segment(q, req->biotail, next->bio)) { + int len = req->biotail->bi_hw_back_size + next->bio->bi_hw_front_size; + /* + * propagate the combined length to the end of the requests + */ + if (req->nr_hw_segments == 1) + req->bio->bi_hw_front_size = len; + if (next->nr_hw_segments == 1) + next->biotail->bi_hw_back_size = len; + total_hw_segments--; + } + + if (total_hw_segments > q->max_hw_segments) + return 0; + + /* Merge is OK... */ + req->nr_phys_segments = total_phys_segments; + req->nr_hw_segments = total_hw_segments; + return 1; +} + +/* + * "plug" the device if there are no outstanding requests: this will + * force the transfer to start only after we have put all the requests + * on the list. + * + * This is called with interrupts off and no requests on the queue and + * with the queue lock held. + */ +void blk_plug_device(request_queue_t *q) +{ + WARN_ON(!irqs_disabled()); + + /* + * don't plug a stopped queue, it must be paired with blk_start_queue() + * which will restart the queueing + */ + if (test_bit(QUEUE_FLAG_STOPPED, &q->queue_flags)) + return; + + if (!test_and_set_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags)) + mod_timer(&q->unplug_timer, jiffies + q->unplug_delay); +} + +EXPORT_SYMBOL(blk_plug_device); + +/* + * remove the queue from the plugged list, if present. called with + * queue lock held and interrupts disabled. + */ +int blk_remove_plug(request_queue_t *q) +{ + WARN_ON(!irqs_disabled()); + + if (!test_and_clear_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags)) + return 0; + + del_timer(&q->unplug_timer); + return 1; +} + +EXPORT_SYMBOL(blk_remove_plug); + +/* + * remove the plug and let it rip.. + */ +void __generic_unplug_device(request_queue_t *q) +{ + if (test_bit(QUEUE_FLAG_STOPPED, &q->queue_flags)) + return; + + if (!blk_remove_plug(q)) + return; + + /* + * was plugged, fire request_fn if queue has stuff to do + */ + if (elv_next_request(q)) + q->request_fn(q); +} +EXPORT_SYMBOL(__generic_unplug_device); + +/** + * generic_unplug_device - fire a request queue + * @q: The &request_queue_t in question + * + * Description: + * Linux uses plugging to build bigger requests queues before letting + * the device have at them. If a queue is plugged, the I/O scheduler + * is still adding and merging requests on the queue. Once the queue + * gets unplugged, the request_fn defined for the queue is invoked and + * transfers started. + **/ +void generic_unplug_device(request_queue_t *q) +{ + spin_lock_irq(q->queue_lock); + __generic_unplug_device(q); + spin_unlock_irq(q->queue_lock); +} +EXPORT_SYMBOL(generic_unplug_device); + +static void blk_backing_dev_unplug(struct backing_dev_info *bdi, + struct page *page) +{ + request_queue_t *q = bdi->unplug_io_data; + + /* + * devices don't necessarily have an ->unplug_fn defined + */ + if (q->unplug_fn) + q->unplug_fn(q); +} + +static void blk_unplug_work(void *data) +{ + request_queue_t *q = data; + + q->unplug_fn(q); +} + +static void blk_unplug_timeout(unsigned long data) +{ + request_queue_t *q = (request_queue_t *)data; + + kblockd_schedule_work(&q->unplug_work); +} + +/** + * blk_start_queue - restart a previously stopped queue + * @q: The &request_queue_t in question + * + * Description: + * blk_start_queue() will clear the stop flag on the queue, and call + * the request_fn for the queue if it was in a stopped state when + * entered. Also see blk_stop_queue(). Queue lock must be held. + **/ +void blk_start_queue(request_queue_t *q) +{ + clear_bit(QUEUE_FLAG_STOPPED, &q->queue_flags); + + /* + * one level of recursion is ok and is much faster than kicking + * the unplug handling + */ + if (!test_and_set_bit(QUEUE_FLAG_REENTER, &q->queue_flags)) { + q->request_fn(q); + clear_bit(QUEUE_FLAG_REENTER, &q->queue_flags); + } else { + blk_plug_device(q); + kblockd_schedule_work(&q->unplug_work); + } +} + +EXPORT_SYMBOL(blk_start_queue); + +/** + * blk_stop_queue - stop a queue + * @q: The &request_queue_t in question + * + * Description: + * The Linux block layer assumes that a block driver will consume all + * entries on the request queue when the request_fn strategy is called. + * Often this will not happen, because of hardware limitations (queue + * depth settings). If a device driver gets a 'queue full' response, + * or if it simply chooses not to queue more I/O at one point, it can + * call this function to prevent the request_fn from being called until + * the driver has signalled it's ready to go again. This happens by calling + * blk_start_queue() to restart queue operations. Queue lock must be held. + **/ +void blk_stop_queue(request_queue_t *q) +{ + blk_remove_plug(q); + set_bit(QUEUE_FLAG_STOPPED, &q->queue_flags); +} +EXPORT_SYMBOL(blk_stop_queue); + +/** + * blk_sync_queue - cancel any pending callbacks on a queue + * @q: the queue + * + * Description: + * The block layer may perform asynchronous callback activity + * on a queue, such as calling the unplug function after a timeout. + * A block device may call blk_sync_queue to ensure that any + * such activity is cancelled, thus allowing it to release resources + * the the callbacks might use. The caller must already have made sure + * that its ->make_request_fn will not re-add plugging prior to calling + * this function. + * + */ +void blk_sync_queue(struct request_queue *q) +{ + del_timer_sync(&q->unplug_timer); + kblockd_flush(); +} +EXPORT_SYMBOL(blk_sync_queue); + +/** + * blk_run_queue - run a single device queue + * @q: The queue to run + */ +void blk_run_queue(struct request_queue *q) +{ + unsigned long flags; + + spin_lock_irqsave(q->queue_lock, flags); + blk_remove_plug(q); + q->request_fn(q); + spin_unlock_irqrestore(q->queue_lock, flags); +} +EXPORT_SYMBOL(blk_run_queue); + +/** + * blk_cleanup_queue: - release a &request_queue_t when it is no longer needed + * @q: the request queue to be released + * + * Description: + * blk_cleanup_queue is the pair to blk_init_queue() or + * blk_queue_make_request(). It should be called when a request queue is + * being released; typically when a block device is being de-registered. + * Currently, its primary task it to free all the &struct request + * structures that were allocated to the queue and the queue itself. + * + * Caveat: + * Hopefully the low level driver will have finished any + * outstanding requests first... + **/ +void blk_cleanup_queue(request_queue_t * q) +{ + struct request_list *rl = &q->rq; + + if (!atomic_dec_and_test(&q->refcnt)) + return; + + if (q->elevator) + elevator_exit(q->elevator); + + blk_sync_queue(q); + + if (rl->rq_pool) + mempool_destroy(rl->rq_pool); + + if (q->queue_tags) + __blk_queue_free_tags(q); + + blk_queue_ordered(q, QUEUE_ORDERED_NONE); + + kmem_cache_free(requestq_cachep, q); +} + +EXPORT_SYMBOL(blk_cleanup_queue); + +static int blk_init_free_list(request_queue_t *q) +{ + struct request_list *rl = &q->rq; + + rl->count[READ] = rl->count[WRITE] = 0; + rl->starved[READ] = rl->starved[WRITE] = 0; + init_waitqueue_head(&rl->wait[READ]); + init_waitqueue_head(&rl->wait[WRITE]); + init_waitqueue_head(&rl->drain); + + rl->rq_pool = mempool_create(BLKDEV_MIN_RQ, mempool_alloc_slab, mempool_free_slab, request_cachep); + + if (!rl->rq_pool) + return -ENOMEM; + + return 0; +} + +static int __make_request(request_queue_t *, struct bio *); + +request_queue_t *blk_alloc_queue(int gfp_mask) +{ + request_queue_t *q = kmem_cache_alloc(requestq_cachep, gfp_mask); + + if (!q) + return NULL; + + memset(q, 0, sizeof(*q)); + init_timer(&q->unplug_timer); + atomic_set(&q->refcnt, 1); + + q->backing_dev_info.unplug_io_fn = blk_backing_dev_unplug; + q->backing_dev_info.unplug_io_data = q; + + return q; +} + +EXPORT_SYMBOL(blk_alloc_queue); + +/** + * blk_init_queue - prepare a request queue for use with a block device + * @rfn: The function to be called to process requests that have been + * placed on the queue. + * @lock: Request queue spin lock + * + * Description: + * If a block device wishes to use the standard request handling procedures, + * which sorts requests and coalesces adjacent requests, then it must + * call blk_init_queue(). The function @rfn will be called when there + * are requests on the queue that need to be processed. If the device + * supports plugging, then @rfn may not be called immediately when requests + * are available on the queue, but may be called at some time later instead. + * Plugged queues are generally unplugged when a buffer belonging to one + * of the requests on the queue is needed, or due to memory pressure. + * + * @rfn is not required, or even expected, to remove all requests off the + * queue, but only as many as it can handle at a time. If it does leave + * requests on the queue, it is responsible for arranging that the requests + * get dealt with eventually. + * + * The queue spin lock must be held while manipulating the requests on the + * request queue. + * + * Function returns a pointer to the initialized request queue, or NULL if + * it didn't succeed. + * + * Note: + * blk_init_queue() must be paired with a blk_cleanup_queue() call + * when the block device is deactivated (such as at module unload). + **/ +request_queue_t *blk_init_queue(request_fn_proc *rfn, spinlock_t *lock) +{ + request_queue_t *q = blk_alloc_queue(GFP_KERNEL); + + if (!q) + return NULL; + + if (blk_init_free_list(q)) + goto out_init; + + q->request_fn = rfn; + q->back_merge_fn = ll_back_merge_fn; + q->front_merge_fn = ll_front_merge_fn; + q->merge_requests_fn = ll_merge_requests_fn; + q->prep_rq_fn = NULL; + q->unplug_fn = generic_unplug_device; + q->queue_flags = (1 << QUEUE_FLAG_CLUSTER); + q->queue_lock = lock; + + blk_queue_segment_boundary(q, 0xffffffff); + + blk_queue_make_request(q, __make_request); + blk_queue_max_segment_size(q, MAX_SEGMENT_SIZE); + + blk_queue_max_hw_segments(q, MAX_HW_SEGMENTS); + blk_queue_max_phys_segments(q, MAX_PHYS_SEGMENTS); + + /* + * all done + */ + if (!elevator_init(q, NULL)) { + blk_queue_congestion_threshold(q); + return q; + } + + blk_cleanup_queue(q); +out_init: + kmem_cache_free(requestq_cachep, q); + return NULL; +} + +EXPORT_SYMBOL(blk_init_queue); + +int blk_get_queue(request_queue_t *q) +{ + if (!test_bit(QUEUE_FLAG_DEAD, &q->queue_flags)) { + atomic_inc(&q->refcnt); + return 0; + } + + return 1; +} + +EXPORT_SYMBOL(blk_get_queue); + +static inline void blk_free_request(request_queue_t *q, struct request *rq) +{ + elv_put_request(q, rq); + mempool_free(rq, q->rq.rq_pool); +} + +static inline struct request *blk_alloc_request(request_queue_t *q, int rw, + int gfp_mask) +{ + struct request *rq = mempool_alloc(q->rq.rq_pool, gfp_mask); + + if (!rq) + return NULL; + + /* + * first three bits are identical in rq->flags and bio->bi_rw, + * see bio.h and blkdev.h + */ + rq->flags = rw; + + if (!elv_set_request(q, rq, gfp_mask)) + return rq; + + mempool_free(rq, q->rq.rq_pool); + return NULL; +} + +/* + * ioc_batching returns true if the ioc is a valid batching request and + * should be given priority access to a request. + */ +static inline int ioc_batching(request_queue_t *q, struct io_context *ioc) +{ + if (!ioc) + return 0; + + /* + * Make sure the process is able to allocate at least 1 request + * even if the batch times out, otherwise we could theoretically + * lose wakeups. + */ + return ioc->nr_batch_requests == q->nr_batching || + (ioc->nr_batch_requests > 0 + && time_before(jiffies, ioc->last_waited + BLK_BATCH_TIME)); +} + +/* + * ioc_set_batching sets ioc to be a new "batcher" if it is not one. This + * will cause the process to be a "batcher" on all queues in the system. This + * is the behaviour we want though - once it gets a wakeup it should be given + * a nice run. + */ +void ioc_set_batching(request_queue_t *q, struct io_context *ioc) +{ + if (!ioc || ioc_batching(q, ioc)) + return; + + ioc->nr_batch_requests = q->nr_batching; + ioc->last_waited = jiffies; +} + +static void __freed_request(request_queue_t *q, int rw) +{ + struct request_list *rl = &q->rq; + + if (rl->count[rw] < queue_congestion_off_threshold(q)) + clear_queue_congested(q, rw); + + if (rl->count[rw] + 1 <= q->nr_requests) { + smp_mb(); + if (waitqueue_active(&rl->wait[rw])) + wake_up(&rl->wait[rw]); + + blk_clear_queue_full(q, rw); + } +} + +/* + * A request has just been released. Account for it, update the full and + * congestion status, wake up any waiters. Called under q->queue_lock. + */ +static void freed_request(request_queue_t *q, int rw) +{ + struct request_list *rl = &q->rq; + + rl->count[rw]--; + + __freed_request(q, rw); + + if (unlikely(rl->starved[rw ^ 1])) + __freed_request(q, rw ^ 1); + + if (!rl->count[READ] && !rl->count[WRITE]) { + smp_mb(); + if (unlikely(waitqueue_active(&rl->drain))) + wake_up(&rl->drain); + } +} + +#define blkdev_free_rq(list) list_entry((list)->next, struct request, queuelist) +/* + * Get a free request, queue_lock must not be held + */ +static struct request *get_request(request_queue_t *q, int rw, int gfp_mask) +{ + struct request *rq = NULL; + struct request_list *rl = &q->rq; + struct io_context *ioc = get_io_context(gfp_mask); + + if (unlikely(test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags))) + goto out; + + spin_lock_irq(q->queue_lock); + if (rl->count[rw]+1 >= q->nr_requests) { + /* + * The queue will fill after this allocation, so set it as + * full, and mark this process as "batching". This process + * will be allowed to complete a batch of requests, others + * will be blocked. + */ + if (!blk_queue_full(q, rw)) { + ioc_set_batching(q, ioc); + blk_set_queue_full(q, rw); + } + } + + switch (elv_may_queue(q, rw)) { + case ELV_MQUEUE_NO: + goto rq_starved; + case ELV_MQUEUE_MAY: + break; + case ELV_MQUEUE_MUST: + goto get_rq; + } + + if (blk_queue_full(q, rw) && !ioc_batching(q, ioc)) { + /* + * The queue is full and the allocating process is not a + * "batcher", and not exempted by the IO scheduler + */ + spin_unlock_irq(q->queue_lock); + goto out; + } + +get_rq: + rl->count[rw]++; + rl->starved[rw] = 0; + if (rl->count[rw] >= queue_congestion_on_threshold(q)) + set_queue_congested(q, rw); + spin_unlock_irq(q->queue_lock); + + rq = blk_alloc_request(q, rw, gfp_mask); + if (!rq) { + /* + * Allocation failed presumably due to memory. Undo anything + * we might have messed up. + * + * Allocating task should really be put onto the front of the + * wait queue, but this is pretty rare. + */ + spin_lock_irq(q->queue_lock); + freed_request(q, rw); + + /* + * in the very unlikely event that allocation failed and no + * requests for this direction was pending, mark us starved + * so that freeing of a request in the other direction will + * notice us. another possible fix would be to split the + * rq mempool into READ and WRITE + */ +rq_starved: + if (unlikely(rl->count[rw] == 0)) + rl->starved[rw] = 1; + + spin_unlock_irq(q->queue_lock); + goto out; + } + + if (ioc_batching(q, ioc)) + ioc->nr_batch_requests--; + + rq_init(q, rq); + rq->rl = rl; +out: + put_io_context(ioc); + return rq; +} + +/* + * No available requests for this queue, unplug the device and wait for some + * requests to become available. + */ +static struct request *get_request_wait(request_queue_t *q, int rw) +{ + DEFINE_WAIT(wait); + struct request *rq; + + generic_unplug_device(q); + do { + struct request_list *rl = &q->rq; + + prepare_to_wait_exclusive(&rl->wait[rw], &wait, + TASK_UNINTERRUPTIBLE); + + rq = get_request(q, rw, GFP_NOIO); + + if (!rq) { + struct io_context *ioc; + + io_schedule(); + + /* + * After sleeping, we become a "batching" process and + * will be able to allocate at least one request, and + * up to a big batch of them for a small period time. + * See ioc_batching, ioc_set_batching + */ + ioc = get_io_context(GFP_NOIO); + ioc_set_batching(q, ioc); + put_io_context(ioc); + } + finish_wait(&rl->wait[rw], &wait); + } while (!rq); + + return rq; +} + +struct request *blk_get_request(request_queue_t *q, int rw, int gfp_mask) +{ + struct request *rq; + + BUG_ON(rw != READ && rw != WRITE); + + if (gfp_mask & __GFP_WAIT) + rq = get_request_wait(q, rw); + else + rq = get_request(q, rw, gfp_mask); + + return rq; +} + +EXPORT_SYMBOL(blk_get_request); + +/** + * blk_requeue_request - put a request back on queue + * @q: request queue where request should be inserted + * @rq: request to be inserted + * + * Description: + * Drivers often keep queueing requests until the hardware cannot accept + * more, when that condition happens we need to put the request back + * on the queue. Must be called with queue lock held. + */ +void blk_requeue_request(request_queue_t *q, struct request *rq) +{ + if (blk_rq_tagged(rq)) + blk_queue_end_tag(q, rq); + + elv_requeue_request(q, rq); +} + +EXPORT_SYMBOL(blk_requeue_request); + +/** + * blk_insert_request - insert a special request in to a request queue + * @q: request queue where request should be inserted + * @rq: request to be inserted + * @at_head: insert request at head or tail of queue + * @data: private data + * @reinsert: true if request it a reinsertion of previously processed one + * + * Description: + * Many block devices need to execute commands asynchronously, so they don't + * block the whole kernel from preemption during request execution. This is + * accomplished normally by inserting aritficial requests tagged as + * REQ_SPECIAL in to the corresponding request queue, and letting them be + * scheduled for actual execution by the request queue. + * + * We have the option of inserting the head or the tail of the queue. + * Typically we use the tail for new ioctls and so forth. We use the head + * of the queue for things like a QUEUE_FULL message from a device, or a + * host that is unable to accept a particular command. + */ +void blk_insert_request(request_queue_t *q, struct request *rq, + int at_head, void *data, int reinsert) +{ + unsigned long flags; + + /* + * tell I/O scheduler that this isn't a regular read/write (ie it + * must not attempt merges on this) and that it acts as a soft + * barrier + */ + rq->flags |= REQ_SPECIAL | REQ_SOFTBARRIER; + + rq->special = data; + + spin_lock_irqsave(q->queue_lock, flags); + + /* + * If command is tagged, release the tag + */ + if (reinsert) + blk_requeue_request(q, rq); + else { + int where = ELEVATOR_INSERT_BACK; + + if (at_head) + where = ELEVATOR_INSERT_FRONT; + + if (blk_rq_tagged(rq)) + blk_queue_end_tag(q, rq); + + drive_stat_acct(rq, rq->nr_sectors, 1); + __elv_add_request(q, rq, where, 0); + } + if (blk_queue_plugged(q)) + __generic_unplug_device(q); + else + q->request_fn(q); + spin_unlock_irqrestore(q->queue_lock, flags); +} + +EXPORT_SYMBOL(blk_insert_request); + +/** + * blk_rq_map_user - map user data to a request, for REQ_BLOCK_PC usage + * @q: request queue where request should be inserted + * @rw: READ or WRITE data + * @ubuf: the user buffer + * @len: length of user data + * + * Description: + * Data will be mapped directly for zero copy io, if possible. Otherwise + * a kernel bounce buffer is used. + * + * A matching blk_rq_unmap_user() must be issued at the end of io, while + * still in process context. + * + * Note: The mapped bio may need to be bounced through blk_queue_bounce() + * before being submitted to the device, as pages mapped may be out of + * reach. It's the callers responsibility to make sure this happens. The + * original bio must be passed back in to blk_rq_unmap_user() for proper + * unmapping. + */ +struct request *blk_rq_map_user(request_queue_t *q, int rw, void __user *ubuf, + unsigned int len) +{ + unsigned long uaddr; + struct request *rq; + struct bio *bio; + + if (len > (q->max_sectors << 9)) + return ERR_PTR(-EINVAL); + if ((!len && ubuf) || (len && !ubuf)) + return ERR_PTR(-EINVAL); + + rq = blk_get_request(q, rw, __GFP_WAIT); + if (!rq) + return ERR_PTR(-ENOMEM); + + /* + * if alignment requirement is satisfied, map in user pages for + * direct dma. else, set up kernel bounce buffers + */ + uaddr = (unsigned long) ubuf; + if (!(uaddr & queue_dma_alignment(q)) && !(len & queue_dma_alignment(q))) + bio = bio_map_user(q, NULL, uaddr, len, rw == READ); + else + bio = bio_copy_user(q, uaddr, len, rw == READ); + + if (!IS_ERR(bio)) { + rq->bio = rq->biotail = bio; + blk_rq_bio_prep(q, rq, bio); + + rq->buffer = rq->data = NULL; + rq->data_len = len; + return rq; + } + + /* + * bio is the err-ptr + */ + blk_put_request(rq); + return (struct request *) bio; +} + +EXPORT_SYMBOL(blk_rq_map_user); + +/** + * blk_rq_unmap_user - unmap a request with user data + * @rq: request to be unmapped + * @bio: bio for the request + * @ulen: length of user buffer + * + * Description: + * Unmap a request previously mapped by blk_rq_map_user(). + */ +int blk_rq_unmap_user(struct request *rq, struct bio *bio, unsigned int ulen) +{ + int ret = 0; + + if (bio) { + if (bio_flagged(bio, BIO_USER_MAPPED)) + bio_unmap_user(bio); + else + ret = bio_uncopy_user(bio); + } + + blk_put_request(rq); + return ret; +} + +EXPORT_SYMBOL(blk_rq_unmap_user); + +/** + * blk_execute_rq - insert a request into queue for execution + * @q: queue to insert the request in + * @bd_disk: matching gendisk + * @rq: request to insert + * + * Description: + * Insert a fully prepared request at the back of the io scheduler queue + * for execution. + */ +int blk_execute_rq(request_queue_t *q, struct gendisk *bd_disk, + struct request *rq) +{ + DECLARE_COMPLETION(wait); + char sense[SCSI_SENSE_BUFFERSIZE]; + int err = 0; + + rq->rq_disk = bd_disk; + + /* + * we need an extra reference to the request, so we can look at + * it after io completion + */ + rq->ref_count++; + + if (!rq->sense) { + memset(sense, 0, sizeof(sense)); + rq->sense = sense; + rq->sense_len = 0; + } + + rq->flags |= REQ_NOMERGE; + rq->waiting = &wait; + rq->end_io = blk_end_sync_rq; + elv_add_request(q, rq, ELEVATOR_INSERT_BACK, 1); + generic_unplug_device(q); + wait_for_completion(&wait); + rq->waiting = NULL; + + if (rq->errors) + err = -EIO; + + return err; +} + +EXPORT_SYMBOL(blk_execute_rq); + +/** + * blkdev_issue_flush - queue a flush + * @bdev: blockdev to issue flush for + * @error_sector: error sector + * + * Description: + * Issue a flush for the block device in question. Caller can supply + * room for storing the error offset in case of a flush error, if they + * wish to. Caller must run wait_for_completion() on its own. + */ +int blkdev_issue_flush(struct block_device *bdev, sector_t *error_sector) +{ + request_queue_t *q; + + if (bdev->bd_disk == NULL) + return -ENXIO; + + q = bdev_get_queue(bdev); + if (!q) + return -ENXIO; + if (!q->issue_flush_fn) + return -EOPNOTSUPP; + + return q->issue_flush_fn(q, bdev->bd_disk, error_sector); +} + +EXPORT_SYMBOL(blkdev_issue_flush); + +/** + * blkdev_scsi_issue_flush_fn - issue flush for SCSI devices + * @q: device queue + * @disk: gendisk + * @error_sector: error offset + * + * Description: + * Devices understanding the SCSI command set, can use this function as + * a helper for issuing a cache flush. Note: driver is required to store + * the error offset (in case of error flushing) in ->sector of struct + * request. + */ +int blkdev_scsi_issue_flush_fn(request_queue_t *q, struct gendisk *disk, + sector_t *error_sector) +{ + struct request *rq = blk_get_request(q, WRITE, __GFP_WAIT); + int ret; + + rq->flags |= REQ_BLOCK_PC | REQ_SOFTBARRIER; + rq->sector = 0; + memset(rq->cmd, 0, sizeof(rq->cmd)); + rq->cmd[0] = 0x35; + rq->cmd_len = 12; + rq->data = NULL; + rq->data_len = 0; + rq->timeout = 60 * HZ; + + ret = blk_execute_rq(q, disk, rq); + + if (ret && error_sector) + *error_sector = rq->sector; + + blk_put_request(rq); + return ret; +} + +EXPORT_SYMBOL(blkdev_scsi_issue_flush_fn); + +void drive_stat_acct(struct request *rq, int nr_sectors, int new_io) +{ + int rw = rq_data_dir(rq); + + if (!blk_fs_request(rq) || !rq->rq_disk) + return; + + if (rw == READ) { + __disk_stat_add(rq->rq_disk, read_sectors, nr_sectors); + if (!new_io) + __disk_stat_inc(rq->rq_disk, read_merges); + } else if (rw == WRITE) { + __disk_stat_add(rq->rq_disk, write_sectors, nr_sectors); + if (!new_io) + __disk_stat_inc(rq->rq_disk, write_merges); + } + if (new_io) { + disk_round_stats(rq->rq_disk); + rq->rq_disk->in_flight++; + } +} + +/* + * add-request adds a request to the linked list. + * queue lock is held and interrupts disabled, as we muck with the + * request queue list. + */ +static inline void add_request(request_queue_t * q, struct request * req) +{ + drive_stat_acct(req, req->nr_sectors, 1); + + if (q->activity_fn) + q->activity_fn(q->activity_data, rq_data_dir(req)); + + /* + * elevator indicated where it wants this request to be + * inserted at elevator_merge time + */ + __elv_add_request(q, req, ELEVATOR_INSERT_SORT, 0); +} + +/* + * disk_round_stats() - Round off the performance stats on a struct + * disk_stats. + * + * The average IO queue length and utilisation statistics are maintained + * by observing the current state of the queue length and the amount of + * time it has been in this state for. + * + * Normally, that accounting is done on IO completion, but that can result + * in more than a second's worth of IO being accounted for within any one + * second, leading to >100% utilisation. To deal with that, we call this + * function to do a round-off before returning the results when reading + * /proc/diskstats. This accounts immediately for all queue usage up to + * the current jiffies and restarts the counters again. + */ +void disk_round_stats(struct gendisk *disk) +{ + unsigned long now = jiffies; + + __disk_stat_add(disk, time_in_queue, + disk->in_flight * (now - disk->stamp)); + disk->stamp = now; + + if (disk->in_flight) + __disk_stat_add(disk, io_ticks, (now - disk->stamp_idle)); + disk->stamp_idle = now; +} + +/* + * queue lock must be held + */ +static void __blk_put_request(request_queue_t *q, struct request *req) +{ + struct request_list *rl = req->rl; + + if (unlikely(!q)) + return; + if (unlikely(--req->ref_count)) + return; + + req->rq_status = RQ_INACTIVE; + req->q = NULL; + req->rl = NULL; + + /* + * Request may not have originated from ll_rw_blk. if not, + * it didn't come out of our reserved rq pools + */ + if (rl) { + int rw = rq_data_dir(req); + + elv_completed_request(q, req); + + BUG_ON(!list_empty(&req->queuelist)); + + blk_free_request(q, req); + freed_request(q, rw); + } +} + +void blk_put_request(struct request *req) +{ + /* + * if req->rl isn't set, this request didnt originate from the + * block layer, so it's safe to just disregard it + */ + if (req->rl) { + unsigned long flags; + request_queue_t *q = req->q; + + spin_lock_irqsave(q->queue_lock, flags); + __blk_put_request(q, req); + spin_unlock_irqrestore(q->queue_lock, flags); + } +} + +EXPORT_SYMBOL(blk_put_request); + +/** + * blk_end_sync_rq - executes a completion event on a request + * @rq: request to complete + */ +void blk_end_sync_rq(struct request *rq) +{ + struct completion *waiting = rq->waiting; + + rq->waiting = NULL; + __blk_put_request(rq->q, rq); + + /* + * complete last, if this is a stack request the process (and thus + * the rq pointer) could be invalid right after this complete() + */ + complete(waiting); +} +EXPORT_SYMBOL(blk_end_sync_rq); + +/** + * blk_congestion_wait - wait for a queue to become uncongested + * @rw: READ or WRITE + * @timeout: timeout in jiffies + * + * Waits for up to @timeout jiffies for a queue (any queue) to exit congestion. + * If no queues are congested then just wait for the next request to be + * returned. + */ +long blk_congestion_wait(int rw, long timeout) +{ + long ret; + DEFINE_WAIT(wait); + wait_queue_head_t *wqh = &congestion_wqh[rw]; + + prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE); + ret = io_schedule_timeout(timeout); + finish_wait(wqh, &wait); + return ret; +} + +EXPORT_SYMBOL(blk_congestion_wait); + +/* + * Has to be called with the request spinlock acquired + */ +static int attempt_merge(request_queue_t *q, struct request *req, + struct request *next) +{ + if (!rq_mergeable(req) || !rq_mergeable(next)) + return 0; + + /* + * not contigious + */ + if (req->sector + req->nr_sectors != next->sector) + return 0; + + if (rq_data_dir(req) != rq_data_dir(next) + || req->rq_disk != next->rq_disk + || next->waiting || next->special) + return 0; + + /* + * If we are allowed to merge, then append bio list + * from next to rq and release next. merge_requests_fn + * will have updated segment counts, update sector + * counts here. + */ + if (!q->merge_requests_fn(q, req, next)) + return 0; + + /* + * At this point we have either done a back merge + * or front merge. We need the smaller start_time of + * the merged requests to be the current request + * for accounting purposes. + */ + if (time_after(req->start_time, next->start_time)) + req->start_time = next->start_time; + + req->biotail->bi_next = next->bio; + req->biotail = next->biotail; + + req->nr_sectors = req->hard_nr_sectors += next->hard_nr_sectors; + + elv_merge_requests(q, req, next); + + if (req->rq_disk) { + disk_round_stats(req->rq_disk); + req->rq_disk->in_flight--; + } + + __blk_put_request(q, next); + return 1; +} + +static inline int attempt_back_merge(request_queue_t *q, struct request *rq) +{ + struct request *next = elv_latter_request(q, rq); + + if (next) + return attempt_merge(q, rq, next); + + return 0; +} + +static inline int attempt_front_merge(request_queue_t *q, struct request *rq) +{ + struct request *prev = elv_former_request(q, rq); + + if (prev) + return attempt_merge(q, prev, rq); + + return 0; +} + +/** + * blk_attempt_remerge - attempt to remerge active head with next request + * @q: The &request_queue_t belonging to the device + * @rq: The head request (usually) + * + * Description: + * For head-active devices, the queue can easily be unplugged so quickly + * that proper merging is not done on the front request. This may hurt + * performance greatly for some devices. The block layer cannot safely + * do merging on that first request for these queues, but the driver can + * call this function and make it happen any way. Only the driver knows + * when it is safe to do so. + **/ +void blk_attempt_remerge(request_queue_t *q, struct request *rq) +{ + unsigned long flags; + + spin_lock_irqsave(q->queue_lock, flags); + attempt_back_merge(q, rq); + spin_unlock_irqrestore(q->queue_lock, flags); +} + +EXPORT_SYMBOL(blk_attempt_remerge); + +/* + * Non-locking blk_attempt_remerge variant. + */ +void __blk_attempt_remerge(request_queue_t *q, struct request *rq) +{ + attempt_back_merge(q, rq); +} + +EXPORT_SYMBOL(__blk_attempt_remerge); + +static int __make_request(request_queue_t *q, struct bio *bio) +{ + struct request *req, *freereq = NULL; + int el_ret, rw, nr_sectors, cur_nr_sectors, barrier, err; + sector_t sector; + + sector = bio->bi_sector; + nr_sectors = bio_sectors(bio); + cur_nr_sectors = bio_cur_sectors(bio); + + rw = bio_data_dir(bio); + + /* + * low level driver can indicate that it wants pages above a + * certain limit bounced to low memory (ie for highmem, or even + * ISA dma in theory) + */ + blk_queue_bounce(q, &bio); + + spin_lock_prefetch(q->queue_lock); + + barrier = bio_barrier(bio); + if (barrier && (q->ordered == QUEUE_ORDERED_NONE)) { + err = -EOPNOTSUPP; + goto end_io; + } + +again: + spin_lock_irq(q->queue_lock); + + if (elv_queue_empty(q)) { + blk_plug_device(q); + goto get_rq; + } + if (barrier) + goto get_rq; + + el_ret = elv_merge(q, &req, bio); + switch (el_ret) { + case ELEVATOR_BACK_MERGE: + BUG_ON(!rq_mergeable(req)); + + if (!q->back_merge_fn(q, req, bio)) + break; + + req->biotail->bi_next = bio; + req->biotail = bio; + req->nr_sectors = req->hard_nr_sectors += nr_sectors; + drive_stat_acct(req, nr_sectors, 0); + if (!attempt_back_merge(q, req)) + elv_merged_request(q, req); + goto out; + + case ELEVATOR_FRONT_MERGE: + BUG_ON(!rq_mergeable(req)); + + if (!q->front_merge_fn(q, req, bio)) + break; + + bio->bi_next = req->bio; + req->bio = bio; + + /* + * may not be valid. if the low level driver said + * it didn't need a bounce buffer then it better + * not touch req->buffer either... + */ + req->buffer = bio_data(bio); + req->current_nr_sectors = cur_nr_sectors; + req->hard_cur_sectors = cur_nr_sectors; + req->sector = req->hard_sector = sector; + req->nr_sectors = req->hard_nr_sectors += nr_sectors; + drive_stat_acct(req, nr_sectors, 0); + if (!attempt_front_merge(q, req)) + elv_merged_request(q, req); + goto out; + + /* + * elevator says don't/can't merge. get new request + */ + case ELEVATOR_NO_MERGE: + break; + + default: + printk("elevator returned crap (%d)\n", el_ret); + BUG(); + } + + /* + * Grab a free request from the freelist - if that is empty, check + * if we are doing read ahead and abort instead of blocking for + * a free slot. + */ +get_rq: + if (freereq) { + req = freereq; + freereq = NULL; + } else { + spin_unlock_irq(q->queue_lock); + if ((freereq = get_request(q, rw, GFP_ATOMIC)) == NULL) { + /* + * READA bit set + */ + err = -EWOULDBLOCK; + if (bio_rw_ahead(bio)) + goto end_io; + + freereq = get_request_wait(q, rw); + } + goto again; + } + + req->flags |= REQ_CMD; + + /* + * inherit FAILFAST from bio (for read-ahead, and explicit FAILFAST) + */ + if (bio_rw_ahead(bio) || bio_failfast(bio)) + req->flags |= REQ_FAILFAST; + + /* + * REQ_BARRIER implies no merging, but lets make it explicit + */ + if (barrier) + req->flags |= (REQ_HARDBARRIER | REQ_NOMERGE); + + req->errors = 0; + req->hard_sector = req->sector = sector; + req->hard_nr_sectors = req->nr_sectors = nr_sectors; + req->current_nr_sectors = req->hard_cur_sectors = cur_nr_sectors; + req->nr_phys_segments = bio_phys_segments(q, bio); + req->nr_hw_segments = bio_hw_segments(q, bio); + req->buffer = bio_data(bio); /* see ->buffer comment above */ + req->waiting = NULL; + req->bio = req->biotail = bio; + req->rq_disk = bio->bi_bdev->bd_disk; + req->start_time = jiffies; + + add_request(q, req); +out: + if (freereq) + __blk_put_request(q, freereq); + if (bio_sync(bio)) + __generic_unplug_device(q); + + spin_unlock_irq(q->queue_lock); + return 0; + +end_io: + bio_endio(bio, nr_sectors << 9, err); + return 0; +} + +/* + * If bio->bi_dev is a partition, remap the location + */ +static inline void blk_partition_remap(struct bio *bio) +{ + struct block_device *bdev = bio->bi_bdev; + + if (bdev != bdev->bd_contains) { + struct hd_struct *p = bdev->bd_part; + + switch (bio->bi_rw) { + case READ: + p->read_sectors += bio_sectors(bio); + p->reads++; + break; + case WRITE: + p->write_sectors += bio_sectors(bio); + p->writes++; + break; + } + bio->bi_sector += p->start_sect; + bio->bi_bdev = bdev->bd_contains; + } +} + +void blk_finish_queue_drain(request_queue_t *q) +{ + struct request_list *rl = &q->rq; + struct request *rq; + + spin_lock_irq(q->queue_lock); + clear_bit(QUEUE_FLAG_DRAIN, &q->queue_flags); + + while (!list_empty(&q->drain_list)) { + rq = list_entry_rq(q->drain_list.next); + + list_del_init(&rq->queuelist); + __elv_add_request(q, rq, ELEVATOR_INSERT_BACK, 1); + } + + spin_unlock_irq(q->queue_lock); + + wake_up(&rl->wait[0]); + wake_up(&rl->wait[1]); + wake_up(&rl->drain); +} + +static int wait_drain(request_queue_t *q, struct request_list *rl, int dispatch) +{ + int wait = rl->count[READ] + rl->count[WRITE]; + + if (dispatch) + wait += !list_empty(&q->queue_head); + + return wait; +} + +/* + * We rely on the fact that only requests allocated through blk_alloc_request() + * have io scheduler private data structures associated with them. Any other + * type of request (allocated on stack or through kmalloc()) should not go + * to the io scheduler core, but be attached to the queue head instead. + */ +void blk_wait_queue_drained(request_queue_t *q, int wait_dispatch) +{ + struct request_list *rl = &q->rq; + DEFINE_WAIT(wait); + + spin_lock_irq(q->queue_lock); + set_bit(QUEUE_FLAG_DRAIN, &q->queue_flags); + + while (wait_drain(q, rl, wait_dispatch)) { + prepare_to_wait(&rl->drain, &wait, TASK_UNINTERRUPTIBLE); + + if (wait_drain(q, rl, wait_dispatch)) { + __generic_unplug_device(q); + spin_unlock_irq(q->queue_lock); + io_schedule(); + spin_lock_irq(q->queue_lock); + } + + finish_wait(&rl->drain, &wait); + } + + spin_unlock_irq(q->queue_lock); +} + +/* + * block waiting for the io scheduler being started again. + */ +static inline void block_wait_queue_running(request_queue_t *q) +{ + DEFINE_WAIT(wait); + + while (test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags)) { + struct request_list *rl = &q->rq; + + prepare_to_wait_exclusive(&rl->drain, &wait, + TASK_UNINTERRUPTIBLE); + + /* + * re-check the condition. avoids using prepare_to_wait() + * in the fast path (queue is running) + */ + if (test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags)) + io_schedule(); + + finish_wait(&rl->drain, &wait); + } +} + +static void handle_bad_sector(struct bio *bio) +{ + char b[BDEVNAME_SIZE]; + + printk(KERN_INFO "attempt to access beyond end of device\n"); + printk(KERN_INFO "%s: rw=%ld, want=%Lu, limit=%Lu\n", + bdevname(bio->bi_bdev, b), + bio->bi_rw, + (unsigned long long)bio->bi_sector + bio_sectors(bio), + (long long)(bio->bi_bdev->bd_inode->i_size >> 9)); + + set_bit(BIO_EOF, &bio->bi_flags); +} + +/** + * generic_make_request: hand a buffer to its device driver for I/O + * @bio: The bio describing the location in memory and on the device. + * + * generic_make_request() is used to make I/O requests of block + * devices. It is passed a &struct bio, which describes the I/O that needs + * to be done. + * + * generic_make_request() does not return any status. The + * success/failure status of the request, along with notification of + * completion, is delivered asynchronously through the bio->bi_end_io + * function described (one day) else where. + * + * The caller of generic_make_request must make sure that bi_io_vec + * are set to describe the memory buffer, and that bi_dev and bi_sector are + * set to describe the device address, and the + * bi_end_io and optionally bi_private are set to describe how + * completion notification should be signaled. + * + * generic_make_request and the drivers it calls may use bi_next if this + * bio happens to be merged with someone else, and may change bi_dev and + * bi_sector for remaps as it sees fit. So the values of these fields + * should NOT be depended on after the call to generic_make_request. + */ +void generic_make_request(struct bio *bio) +{ + request_queue_t *q; + sector_t maxsector; + int ret, nr_sectors = bio_sectors(bio); + + might_sleep(); + /* Test device or partition size, when known. */ + maxsector = bio->bi_bdev->bd_inode->i_size >> 9; + if (maxsector) { + sector_t sector = bio->bi_sector; + + if (maxsector < nr_sectors || maxsector - nr_sectors < sector) { + /* + * This may well happen - the kernel calls bread() + * without checking the size of the device, e.g., when + * mounting a device. + */ + handle_bad_sector(bio); + goto end_io; + } + } + + /* + * Resolve the mapping until finished. (drivers are + * still free to implement/resolve their own stacking + * by explicitly returning 0) + * + * NOTE: we don't repeat the blk_size check for each new device. + * Stacking drivers are expected to know what they are doing. + */ + do { + char b[BDEVNAME_SIZE]; + + q = bdev_get_queue(bio->bi_bdev); + if (!q) { + printk(KERN_ERR + "generic_make_request: Trying to access " + "nonexistent block-device %s (%Lu)\n", + bdevname(bio->bi_bdev, b), + (long long) bio->bi_sector); +end_io: + bio_endio(bio, bio->bi_size, -EIO); + break; + } + + if (unlikely(bio_sectors(bio) > q->max_hw_sectors)) { + printk("bio too big device %s (%u > %u)\n", + bdevname(bio->bi_bdev, b), + bio_sectors(bio), + q->max_hw_sectors); + goto end_io; + } + + if (test_bit(QUEUE_FLAG_DEAD, &q->queue_flags)) + goto end_io; + + block_wait_queue_running(q); + + /* + * If this device has partitions, remap block n + * of partition p to block n+start(p) of the disk. + */ + blk_partition_remap(bio); + + ret = q->make_request_fn(q, bio); + } while (ret); +} + +EXPORT_SYMBOL(generic_make_request); + +/** + * submit_bio: submit a bio to the block device layer for I/O + * @rw: whether to %READ or %WRITE, or maybe to %READA (read ahead) + * @bio: The &struct bio which describes the I/O + * + * submit_bio() is very similar in purpose to generic_make_request(), and + * uses that function to do most of the work. Both are fairly rough + * interfaces, @bio must be presetup and ready for I/O. + * + */ +void submit_bio(int rw, struct bio *bio) +{ + int count = bio_sectors(bio); + + BIO_BUG_ON(!bio->bi_size); + BIO_BUG_ON(!bio->bi_io_vec); + bio->bi_rw = rw; + if (rw & WRITE) + mod_page_state(pgpgout, count); + else + mod_page_state(pgpgin, count); + + if (unlikely(block_dump)) { + char b[BDEVNAME_SIZE]; + printk(KERN_DEBUG "%s(%d): %s block %Lu on %s\n", + current->comm, current->pid, + (rw & WRITE) ? "WRITE" : "READ", + (unsigned long long)bio->bi_sector, + bdevname(bio->bi_bdev,b)); + } + + generic_make_request(bio); +} + +EXPORT_SYMBOL(submit_bio); + +void blk_recalc_rq_segments(struct request *rq) +{ + struct bio *bio, *prevbio = NULL; + int nr_phys_segs, nr_hw_segs; + unsigned int phys_size, hw_size; + request_queue_t *q = rq->q; + + if (!rq->bio) + return; + + phys_size = hw_size = nr_phys_segs = nr_hw_segs = 0; + rq_for_each_bio(bio, rq) { + /* Force bio hw/phys segs to be recalculated. */ + bio->bi_flags &= ~(1 << BIO_SEG_VALID); + + nr_phys_segs += bio_phys_segments(q, bio); + nr_hw_segs += bio_hw_segments(q, bio); + if (prevbio) { + int pseg = phys_size + prevbio->bi_size + bio->bi_size; + int hseg = hw_size + prevbio->bi_size + bio->bi_size; + + if (blk_phys_contig_segment(q, prevbio, bio) && + pseg <= q->max_segment_size) { + nr_phys_segs--; + phys_size += prevbio->bi_size + bio->bi_size; + } else + phys_size = 0; + + if (blk_hw_contig_segment(q, prevbio, bio) && + hseg <= q->max_segment_size) { + nr_hw_segs--; + hw_size += prevbio->bi_size + bio->bi_size; + } else + hw_size = 0; + } + prevbio = bio; + } + + rq->nr_phys_segments = nr_phys_segs; + rq->nr_hw_segments = nr_hw_segs; +} + +void blk_recalc_rq_sectors(struct request *rq, int nsect) +{ + if (blk_fs_request(rq)) { + rq->hard_sector += nsect; + rq->hard_nr_sectors -= nsect; + + /* + * Move the I/O submission pointers ahead if required. + */ + if ((rq->nr_sectors >= rq->hard_nr_sectors) && + (rq->sector <= rq->hard_sector)) { + rq->sector = rq->hard_sector; + rq->nr_sectors = rq->hard_nr_sectors; + rq->hard_cur_sectors = bio_cur_sectors(rq->bio); + rq->current_nr_sectors = rq->hard_cur_sectors; + rq->buffer = bio_data(rq->bio); + } + + /* + * if total number of sectors is less than the first segment + * size, something has gone terribly wrong + */ + if (rq->nr_sectors < rq->current_nr_sectors) { + printk("blk: request botched\n"); + rq->nr_sectors = rq->current_nr_sectors; + } + } +} + +static int __end_that_request_first(struct request *req, int uptodate, + int nr_bytes) +{ + int total_bytes, bio_nbytes, error, next_idx = 0; + struct bio *bio; + + /* + * extend uptodate bool to allow < 0 value to be direct io error + */ + error = 0; + if (end_io_error(uptodate)) + error = !uptodate ? -EIO : uptodate; + + /* + * for a REQ_BLOCK_PC request, we want to carry any eventual + * sense key with us all the way through + */ + if (!blk_pc_request(req)) + req->errors = 0; + + if (!uptodate) { + if (blk_fs_request(req) && !(req->flags & REQ_QUIET)) + printk("end_request: I/O error, dev %s, sector %llu\n", + req->rq_disk ? req->rq_disk->disk_name : "?", + (unsigned long long)req->sector); + } + + total_bytes = bio_nbytes = 0; + while ((bio = req->bio) != NULL) { + int nbytes; + + if (nr_bytes >= bio->bi_size) { + req->bio = bio->bi_next; + nbytes = bio->bi_size; + bio_endio(bio, nbytes, error); + next_idx = 0; + bio_nbytes = 0; + } else { + int idx = bio->bi_idx + next_idx; + + if (unlikely(bio->bi_idx >= bio->bi_vcnt)) { + blk_dump_rq_flags(req, "__end_that"); + printk("%s: bio idx %d >= vcnt %d\n", + __FUNCTION__, + bio->bi_idx, bio->bi_vcnt); + break; + } + + nbytes = bio_iovec_idx(bio, idx)->bv_len; + BIO_BUG_ON(nbytes > bio->bi_size); + + /* + * not a complete bvec done + */ + if (unlikely(nbytes > nr_bytes)) { + bio_nbytes += nr_bytes; + total_bytes += nr_bytes; + break; + } + + /* + * advance to the next vector + */ + next_idx++; + bio_nbytes += nbytes; + } + + total_bytes += nbytes; + nr_bytes -= nbytes; + + if ((bio = req->bio)) { + /* + * end more in this run, or just return 'not-done' + */ + if (unlikely(nr_bytes <= 0)) + break; + } + } + + /* + * completely done + */ + if (!req->bio) + return 0; + + /* + * if the request wasn't completed, update state + */ + if (bio_nbytes) { + bio_endio(bio, bio_nbytes, error); + bio->bi_idx += next_idx; + bio_iovec(bio)->bv_offset += nr_bytes; + bio_iovec(bio)->bv_len -= nr_bytes; + } + + blk_recalc_rq_sectors(req, total_bytes >> 9); + blk_recalc_rq_segments(req); + return 1; +} + +/** + * end_that_request_first - end I/O on a request + * @req: the request being processed + * @uptodate: 1 for success, 0 for I/O error, < 0 for specific error + * @nr_sectors: number of sectors to end I/O on + * + * Description: + * Ends I/O on a number of sectors attached to @req, and sets it up + * for the next range of segments (if any) in the cluster. + * + * Return: + * 0 - we are done with this request, call end_that_request_last() + * 1 - still buffers pending for this request + **/ +int end_that_request_first(struct request *req, int uptodate, int nr_sectors) +{ + return __end_that_request_first(req, uptodate, nr_sectors << 9); +} + +EXPORT_SYMBOL(end_that_request_first); + +/** + * end_that_request_chunk - end I/O on a request + * @req: the request being processed + * @uptodate: 1 for success, 0 for I/O error, < 0 for specific error + * @nr_bytes: number of bytes to complete + * + * Description: + * Ends I/O on a number of bytes attached to @req, and sets it up + * for the next range of segments (if any). Like end_that_request_first(), + * but deals with bytes instead of sectors. + * + * Return: + * 0 - we are done with this request, call end_that_request_last() + * 1 - still buffers pending for this request + **/ +int end_that_request_chunk(struct request *req, int uptodate, int nr_bytes) +{ + return __end_that_request_first(req, uptodate, nr_bytes); +} + +EXPORT_SYMBOL(end_that_request_chunk); + +/* + * queue lock must be held + */ +void end_that_request_last(struct request *req) +{ + struct gendisk *disk = req->rq_disk; + + if (unlikely(laptop_mode) && blk_fs_request(req)) + laptop_io_completion(); + + if (disk && blk_fs_request(req)) { + unsigned long duration = jiffies - req->start_time; + switch (rq_data_dir(req)) { + case WRITE: + __disk_stat_inc(disk, writes); + __disk_stat_add(disk, write_ticks, duration); + break; + case READ: + __disk_stat_inc(disk, reads); + __disk_stat_add(disk, read_ticks, duration); + break; + } + disk_round_stats(disk); + disk->in_flight--; + } + if (req->end_io) + req->end_io(req); + else + __blk_put_request(req->q, req); +} + +EXPORT_SYMBOL(end_that_request_last); + +void end_request(struct request *req, int uptodate) +{ + if (!end_that_request_first(req, uptodate, req->hard_cur_sectors)) { + add_disk_randomness(req->rq_disk); + blkdev_dequeue_request(req); + end_that_request_last(req); + } +} + +EXPORT_SYMBOL(end_request); + +void blk_rq_bio_prep(request_queue_t *q, struct request *rq, struct bio *bio) +{ + /* first three bits are identical in rq->flags and bio->bi_rw */ + rq->flags |= (bio->bi_rw & 7); + + rq->nr_phys_segments = bio_phys_segments(q, bio); + rq->nr_hw_segments = bio_hw_segments(q, bio); + rq->current_nr_sectors = bio_cur_sectors(bio); + rq->hard_cur_sectors = rq->current_nr_sectors; + rq->hard_nr_sectors = rq->nr_sectors = bio_sectors(bio); + rq->buffer = bio_data(bio); + + rq->bio = rq->biotail = bio; +} + +EXPORT_SYMBOL(blk_rq_bio_prep); + +int kblockd_schedule_work(struct work_struct *work) +{ + return queue_work(kblockd_workqueue, work); +} + +EXPORT_SYMBOL(kblockd_schedule_work); + +void kblockd_flush(void) +{ + flush_workqueue(kblockd_workqueue); +} +EXPORT_SYMBOL(kblockd_flush); + +int __init blk_dev_init(void) +{ + kblockd_workqueue = create_workqueue("kblockd"); + if (!kblockd_workqueue) + panic("Failed to create kblockd\n"); + + request_cachep = kmem_cache_create("blkdev_requests", + sizeof(struct request), 0, SLAB_PANIC, NULL, NULL); + + requestq_cachep = kmem_cache_create("blkdev_queue", + sizeof(request_queue_t), 0, SLAB_PANIC, NULL, NULL); + + iocontext_cachep = kmem_cache_create("blkdev_ioc", + sizeof(struct io_context), 0, SLAB_PANIC, NULL, NULL); + + blk_max_low_pfn = max_low_pfn; + blk_max_pfn = max_pfn; + + return 0; +} + +/* + * IO Context helper functions + */ +void put_io_context(struct io_context *ioc) +{ + if (ioc == NULL) + return; + + BUG_ON(atomic_read(&ioc->refcount) == 0); + + if (atomic_dec_and_test(&ioc->refcount)) { + if (ioc->aic && ioc->aic->dtor) + ioc->aic->dtor(ioc->aic); + if (ioc->cic && ioc->cic->dtor) + ioc->cic->dtor(ioc->cic); + + kmem_cache_free(iocontext_cachep, ioc); + } +} +EXPORT_SYMBOL(put_io_context); + +/* Called by the exitting task */ +void exit_io_context(void) +{ + unsigned long flags; + struct io_context *ioc; + + local_irq_save(flags); + ioc = current->io_context; + current->io_context = NULL; + local_irq_restore(flags); + + if (ioc->aic && ioc->aic->exit) + ioc->aic->exit(ioc->aic); + if (ioc->cic && ioc->cic->exit) + ioc->cic->exit(ioc->cic); + + put_io_context(ioc); +} + +/* + * If the current task has no IO context then create one and initialise it. + * If it does have a context, take a ref on it. + * + * This is always called in the context of the task which submitted the I/O. + * But weird things happen, so we disable local interrupts to ensure exclusive + * access to *current. + */ +struct io_context *get_io_context(int gfp_flags) +{ + struct task_struct *tsk = current; + unsigned long flags; + struct io_context *ret; + + local_irq_save(flags); + ret = tsk->io_context; + if (ret) + goto out; + + local_irq_restore(flags); + + ret = kmem_cache_alloc(iocontext_cachep, gfp_flags); + if (ret) { + atomic_set(&ret->refcount, 1); + ret->pid = tsk->pid; + ret->last_waited = jiffies; /* doesn't matter... */ + ret->nr_batch_requests = 0; /* because this is 0 */ + ret->aic = NULL; + ret->cic = NULL; + spin_lock_init(&ret->lock); + + local_irq_save(flags); + + /* + * very unlikely, someone raced with us in setting up the task + * io context. free new context and just grab a reference. + */ + if (!tsk->io_context) + tsk->io_context = ret; + else { + kmem_cache_free(iocontext_cachep, ret); + ret = tsk->io_context; + } + +out: + atomic_inc(&ret->refcount); + local_irq_restore(flags); + } + + return ret; +} +EXPORT_SYMBOL(get_io_context); + +void copy_io_context(struct io_context **pdst, struct io_context **psrc) +{ + struct io_context *src = *psrc; + struct io_context *dst = *pdst; + + if (src) { + BUG_ON(atomic_read(&src->refcount) == 0); + atomic_inc(&src->refcount); + put_io_context(dst); + *pdst = src; + } +} +EXPORT_SYMBOL(copy_io_context); + +void swap_io_context(struct io_context **ioc1, struct io_context **ioc2) +{ + struct io_context *temp; + temp = *ioc1; + *ioc1 = *ioc2; + *ioc2 = temp; +} +EXPORT_SYMBOL(swap_io_context); + +/* + * sysfs parts below + */ +struct queue_sysfs_entry { + struct attribute attr; + ssize_t (*show)(struct request_queue *, char *); + ssize_t (*store)(struct request_queue *, const char *, size_t); +}; + +static ssize_t +queue_var_show(unsigned int var, char *page) +{ + return sprintf(page, "%d\n", var); +} + +static ssize_t +queue_var_store(unsigned long *var, const char *page, size_t count) +{ + char *p = (char *) page; + + *var = simple_strtoul(p, &p, 10); + return count; +} + +static ssize_t queue_requests_show(struct request_queue *q, char *page) +{ + return queue_var_show(q->nr_requests, (page)); +} + +static ssize_t +queue_requests_store(struct request_queue *q, const char *page, size_t count) +{ + struct request_list *rl = &q->rq; + + int ret = queue_var_store(&q->nr_requests, page, count); + if (q->nr_requests < BLKDEV_MIN_RQ) + q->nr_requests = BLKDEV_MIN_RQ; + blk_queue_congestion_threshold(q); + + if (rl->count[READ] >= queue_congestion_on_threshold(q)) + set_queue_congested(q, READ); + else if (rl->count[READ] < queue_congestion_off_threshold(q)) + clear_queue_congested(q, READ); + + if (rl->count[WRITE] >= queue_congestion_on_threshold(q)) + set_queue_congested(q, WRITE); + else if (rl->count[WRITE] < queue_congestion_off_threshold(q)) + clear_queue_congested(q, WRITE); + + if (rl->count[READ] >= q->nr_requests) { + blk_set_queue_full(q, READ); + } else if (rl->count[READ]+1 <= q->nr_requests) { + blk_clear_queue_full(q, READ); + wake_up(&rl->wait[READ]); + } + + if (rl->count[WRITE] >= q->nr_requests) { + blk_set_queue_full(q, WRITE); + } else if (rl->count[WRITE]+1 <= q->nr_requests) { + blk_clear_queue_full(q, WRITE); + wake_up(&rl->wait[WRITE]); + } + return ret; +} + +static ssize_t queue_ra_show(struct request_queue *q, char *page) +{ + int ra_kb = q->backing_dev_info.ra_pages << (PAGE_CACHE_SHIFT - 10); + + return queue_var_show(ra_kb, (page)); +} + +static ssize_t +queue_ra_store(struct request_queue *q, const char *page, size_t count) +{ + unsigned long ra_kb; + ssize_t ret = queue_var_store(&ra_kb, page, count); + + spin_lock_irq(q->queue_lock); + if (ra_kb > (q->max_sectors >> 1)) + ra_kb = (q->max_sectors >> 1); + + q->backing_dev_info.ra_pages = ra_kb >> (PAGE_CACHE_SHIFT - 10); + spin_unlock_irq(q->queue_lock); + + return ret; +} + +static ssize_t queue_max_sectors_show(struct request_queue *q, char *page) +{ + int max_sectors_kb = q->max_sectors >> 1; + + return queue_var_show(max_sectors_kb, (page)); +} + +static ssize_t +queue_max_sectors_store(struct request_queue *q, const char *page, size_t count) +{ + unsigned long max_sectors_kb, + max_hw_sectors_kb = q->max_hw_sectors >> 1, + page_kb = 1 << (PAGE_CACHE_SHIFT - 10); + ssize_t ret = queue_var_store(&max_sectors_kb, page, count); + int ra_kb; + + if (max_sectors_kb > max_hw_sectors_kb || max_sectors_kb < page_kb) + return -EINVAL; + /* + * Take the queue lock to update the readahead and max_sectors + * values synchronously: + */ + spin_lock_irq(q->queue_lock); + /* + * Trim readahead window as well, if necessary: + */ + ra_kb = q->backing_dev_info.ra_pages << (PAGE_CACHE_SHIFT - 10); + if (ra_kb > max_sectors_kb) + q->backing_dev_info.ra_pages = + max_sectors_kb >> (PAGE_CACHE_SHIFT - 10); + + q->max_sectors = max_sectors_kb << 1; + spin_unlock_irq(q->queue_lock); + + return ret; +} + +static ssize_t queue_max_hw_sectors_show(struct request_queue *q, char *page) +{ + int max_hw_sectors_kb = q->max_hw_sectors >> 1; + + return queue_var_show(max_hw_sectors_kb, (page)); +} + + +static struct queue_sysfs_entry queue_requests_entry = { + .attr = {.name = "nr_requests", .mode = S_IRUGO | S_IWUSR }, + .show = queue_requests_show, + .store = queue_requests_store, +}; + +static struct queue_sysfs_entry queue_ra_entry = { + .attr = {.name = "read_ahead_kb", .mode = S_IRUGO | S_IWUSR }, + .show = queue_ra_show, + .store = queue_ra_store, +}; + +static struct queue_sysfs_entry queue_max_sectors_entry = { + .attr = {.name = "max_sectors_kb", .mode = S_IRUGO | S_IWUSR }, + .show = queue_max_sectors_show, + .store = queue_max_sectors_store, +}; + +static struct queue_sysfs_entry queue_max_hw_sectors_entry = { + .attr = {.name = "max_hw_sectors_kb", .mode = S_IRUGO }, + .show = queue_max_hw_sectors_show, +}; + +static struct queue_sysfs_entry queue_iosched_entry = { + .attr = {.name = "scheduler", .mode = S_IRUGO | S_IWUSR }, + .show = elv_iosched_show, + .store = elv_iosched_store, +}; + +static struct attribute *default_attrs[] = { + &queue_requests_entry.attr, + &queue_ra_entry.attr, + &queue_max_hw_sectors_entry.attr, + &queue_max_sectors_entry.attr, + &queue_iosched_entry.attr, + NULL, +}; + +#define to_queue(atr) container_of((atr), struct queue_sysfs_entry, attr) + +static ssize_t +queue_attr_show(struct kobject *kobj, struct attribute *attr, char *page) +{ + struct queue_sysfs_entry *entry = to_queue(attr); + struct request_queue *q; + + q = container_of(kobj, struct request_queue, kobj); + if (!entry->show) + return 0; + + return entry->show(q, page); +} + +static ssize_t +queue_attr_store(struct kobject *kobj, struct attribute *attr, + const char *page, size_t length) +{ + struct queue_sysfs_entry *entry = to_queue(attr); + struct request_queue *q; + + q = container_of(kobj, struct request_queue, kobj); + if (!entry->store) + return -EINVAL; + + return entry->store(q, page, length); +} + +static struct sysfs_ops queue_sysfs_ops = { + .show = queue_attr_show, + .store = queue_attr_store, +}; + +struct kobj_type queue_ktype = { + .sysfs_ops = &queue_sysfs_ops, + .default_attrs = default_attrs, +}; + +int blk_register_queue(struct gendisk *disk) +{ + int ret; + + request_queue_t *q = disk->queue; + + if (!q || !q->request_fn) + return -ENXIO; + + q->kobj.parent = kobject_get(&disk->kobj); + if (!q->kobj.parent) + return -EBUSY; + + snprintf(q->kobj.name, KOBJ_NAME_LEN, "%s", "queue"); + q->kobj.ktype = &queue_ktype; + + ret = kobject_register(&q->kobj); + if (ret < 0) + return ret; + + ret = elv_register_queue(q); + if (ret) { + kobject_unregister(&q->kobj); + return ret; + } + + return 0; +} + +void blk_unregister_queue(struct gendisk *disk) +{ + request_queue_t *q = disk->queue; + + if (q && q->request_fn) { + elv_unregister_queue(q); + + kobject_unregister(&q->kobj); + kobject_put(&disk->kobj); + } +} diff --git a/drivers/block/loop.c b/drivers/block/loop.c new file mode 100644 index 00000000000..6f011d0d8e9 --- /dev/null +++ b/drivers/block/loop.c @@ -0,0 +1,1348 @@ +/* + * linux/drivers/block/loop.c + * + * Written by Theodore Ts'o, 3/29/93 + * + * Copyright 1993 by Theodore Ts'o. Redistribution of this file is + * permitted under the GNU General Public License. + * + * DES encryption plus some minor changes by Werner Almesberger, 30-MAY-1993 + * more DES encryption plus IDEA encryption by Nicholas J. Leon, June 20, 1996 + * + * Modularized and updated for 1.1.16 kernel - Mitch Dsouza 28th May 1994 + * Adapted for 1.3.59 kernel - Andries Brouwer, 1 Feb 1996 + * + * Fixed do_loop_request() re-entrancy - Vincent.Renardias@waw.com Mar 20, 1997 + * + * Added devfs support - Richard Gooch <rgooch@atnf.csiro.au> 16-Jan-1998 + * + * Handle sparse backing files correctly - Kenn Humborg, Jun 28, 1998 + * + * Loadable modules and other fixes by AK, 1998 + * + * Make real block number available to downstream transfer functions, enables + * CBC (and relatives) mode encryption requiring unique IVs per data block. + * Reed H. Petty, rhp@draper.net + * + * Maximum number of loop devices now dynamic via max_loop module parameter. + * Russell Kroll <rkroll@exploits.org> 19990701 + * + * Maximum number of loop devices when compiled-in now selectable by passing + * max_loop=<1-255> to the kernel on boot. + * Erik I. Bolsø, <eriki@himolde.no>, Oct 31, 1999 + * + * Completely rewrite request handling to be make_request_fn style and + * non blocking, pushing work to a helper thread. Lots of fixes from + * Al Viro too. + * Jens Axboe <axboe@suse.de>, Nov 2000 + * + * Support up to 256 loop devices + * Heinz Mauelshagen <mge@sistina.com>, Feb 2002 + * + * Support for falling back on the write file operation when the address space + * operations prepare_write and/or commit_write are not available on the + * backing filesystem. + * Anton Altaparmakov, 16 Feb 2005 + * + * Still To Fix: + * - Advisory locking is ignored here. + * - Should use an own CAP_* category instead of CAP_SYS_ADMIN + * + */ + +#include <linux/config.h> +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <linux/sched.h> +#include <linux/fs.h> +#include <linux/file.h> +#include <linux/stat.h> +#include <linux/errno.h> +#include <linux/major.h> +#include <linux/wait.h> +#include <linux/blkdev.h> +#include <linux/blkpg.h> +#include <linux/init.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/smp_lock.h> +#include <linux/swap.h> +#include <linux/slab.h> +#include <linux/loop.h> +#include <linux/suspend.h> +#include <linux/writeback.h> +#include <linux/buffer_head.h> /* for invalidate_bdev() */ +#include <linux/completion.h> +#include <linux/highmem.h> +#include <linux/gfp.h> + +#include <asm/uaccess.h> + +static int max_loop = 8; +static struct loop_device *loop_dev; +static struct gendisk **disks; + +/* + * Transfer functions + */ +static int transfer_none(struct loop_device *lo, int cmd, + struct page *raw_page, unsigned raw_off, + struct page *loop_page, unsigned loop_off, + int size, sector_t real_block) +{ + char *raw_buf = kmap_atomic(raw_page, KM_USER0) + raw_off; + char *loop_buf = kmap_atomic(loop_page, KM_USER1) + loop_off; + + if (cmd == READ) + memcpy(loop_buf, raw_buf, size); + else + memcpy(raw_buf, loop_buf, size); + + kunmap_atomic(raw_buf, KM_USER0); + kunmap_atomic(loop_buf, KM_USER1); + cond_resched(); + return 0; +} + +static int transfer_xor(struct loop_device *lo, int cmd, + struct page *raw_page, unsigned raw_off, + struct page *loop_page, unsigned loop_off, + int size, sector_t real_block) +{ + char *raw_buf = kmap_atomic(raw_page, KM_USER0) + raw_off; + char *loop_buf = kmap_atomic(loop_page, KM_USER1) + loop_off; + char *in, *out, *key; + int i, keysize; + + if (cmd == READ) { + in = raw_buf; + out = loop_buf; + } else { + in = loop_buf; + out = raw_buf; + } + + key = lo->lo_encrypt_key; + keysize = lo->lo_encrypt_key_size; + for (i = 0; i < size; i++) + *out++ = *in++ ^ key[(i & 511) % keysize]; + + kunmap_atomic(raw_buf, KM_USER0); + kunmap_atomic(loop_buf, KM_USER1); + cond_resched(); + return 0; +} + +static int xor_init(struct loop_device *lo, const struct loop_info64 *info) +{ + if (unlikely(info->lo_encrypt_key_size <= 0)) + return -EINVAL; + return 0; +} + +static struct loop_func_table none_funcs = { + .number = LO_CRYPT_NONE, + .transfer = transfer_none, +}; + +static struct loop_func_table xor_funcs = { + .number = LO_CRYPT_XOR, + .transfer = transfer_xor, + .init = xor_init +}; + +/* xfer_funcs[0] is special - its release function is never called */ +static struct loop_func_table *xfer_funcs[MAX_LO_CRYPT] = { + &none_funcs, + &xor_funcs +}; + +static loff_t get_loop_size(struct loop_device *lo, struct file *file) +{ + loff_t size, offset, loopsize; + + /* Compute loopsize in bytes */ + size = i_size_read(file->f_mapping->host); + offset = lo->lo_offset; + loopsize = size - offset; + if (lo->lo_sizelimit > 0 && lo->lo_sizelimit < loopsize) + loopsize = lo->lo_sizelimit; + + /* + * Unfortunately, if we want to do I/O on the device, + * the number of 512-byte sectors has to fit into a sector_t. + */ + return loopsize >> 9; +} + +static int +figure_loop_size(struct loop_device *lo) +{ + loff_t size = get_loop_size(lo, lo->lo_backing_file); + sector_t x = (sector_t)size; + + if (unlikely((loff_t)x != size)) + return -EFBIG; + + set_capacity(disks[lo->lo_number], x); + return 0; +} + +static inline int +lo_do_transfer(struct loop_device *lo, int cmd, + struct page *rpage, unsigned roffs, + struct page *lpage, unsigned loffs, + int size, sector_t rblock) +{ + if (unlikely(!lo->transfer)) + return 0; + + return lo->transfer(lo, cmd, rpage, roffs, lpage, loffs, size, rblock); +} + +/** + * do_lo_send_aops - helper for writing data to a loop device + * + * This is the fast version for backing filesystems which implement the address + * space operations prepare_write and commit_write. + */ +static int do_lo_send_aops(struct loop_device *lo, struct bio_vec *bvec, + int bsize, loff_t pos, struct page *page) +{ + struct file *file = lo->lo_backing_file; /* kudos to NFsckingS */ + struct address_space *mapping = file->f_mapping; + struct address_space_operations *aops = mapping->a_ops; + pgoff_t index; + unsigned offset, bv_offs; + int len, ret = 0; + + down(&mapping->host->i_sem); + index = pos >> PAGE_CACHE_SHIFT; + offset = pos & ((pgoff_t)PAGE_CACHE_SIZE - 1); + bv_offs = bvec->bv_offset; + len = bvec->bv_len; + while (len > 0) { + sector_t IV; + unsigned size; + int transfer_result; + + IV = ((sector_t)index << (PAGE_CACHE_SHIFT - 9))+(offset >> 9); + size = PAGE_CACHE_SIZE - offset; + if (size > len) + size = len; + page = grab_cache_page(mapping, index); + if (unlikely(!page)) + goto fail; + if (unlikely(aops->prepare_write(file, page, offset, + offset + size))) + goto unlock; + transfer_result = lo_do_transfer(lo, WRITE, page, offset, + bvec->bv_page, bv_offs, size, IV); + if (unlikely(transfer_result)) { + char *kaddr; + + /* + * The transfer failed, but we still write the data to + * keep prepare/commit calls balanced. + */ + printk(KERN_ERR "loop: transfer error block %llu\n", + (unsigned long long)index); + kaddr = kmap_atomic(page, KM_USER0); + memset(kaddr + offset, 0, size); + kunmap_atomic(kaddr, KM_USER0); + } + flush_dcache_page(page); + if (unlikely(aops->commit_write(file, page, offset, + offset + size))) + goto unlock; + if (unlikely(transfer_result)) + goto unlock; + bv_offs += size; + len -= size; + offset = 0; + index++; + pos += size; + unlock_page(page); + page_cache_release(page); + } +out: + up(&mapping->host->i_sem); + return ret; +unlock: + unlock_page(page); + page_cache_release(page); +fail: + ret = -1; + goto out; +} + +/** + * __do_lo_send_write - helper for writing data to a loop device + * + * This helper just factors out common code between do_lo_send_direct_write() + * and do_lo_send_write(). + */ +static inline int __do_lo_send_write(struct file *file, + u8 __user *buf, const int len, loff_t pos) +{ + ssize_t bw; + mm_segment_t old_fs = get_fs(); + + set_fs(get_ds()); + bw = file->f_op->write(file, buf, len, &pos); + set_fs(old_fs); + if (likely(bw == len)) + return 0; + printk(KERN_ERR "loop: Write error at byte offset %llu, length %i.\n", + (unsigned long long)pos, len); + if (bw >= 0) + bw = -EIO; + return bw; +} + +/** + * do_lo_send_direct_write - helper for writing data to a loop device + * + * This is the fast, non-transforming version for backing filesystems which do + * not implement the address space operations prepare_write and commit_write. + * It uses the write file operation which should be present on all writeable + * filesystems. + */ +static int do_lo_send_direct_write(struct loop_device *lo, + struct bio_vec *bvec, int bsize, loff_t pos, struct page *page) +{ + ssize_t bw = __do_lo_send_write(lo->lo_backing_file, + (u8 __user *)kmap(bvec->bv_page) + bvec->bv_offset, + bvec->bv_len, pos); + kunmap(bvec->bv_page); + cond_resched(); + return bw; +} + +/** + * do_lo_send_write - helper for writing data to a loop device + * + * This is the slow, transforming version for filesystems which do not + * implement the address space operations prepare_write and commit_write. It + * uses the write file operation which should be present on all writeable + * filesystems. + * + * Using fops->write is slower than using aops->{prepare,commit}_write in the + * transforming case because we need to double buffer the data as we cannot do + * the transformations in place as we do not have direct access to the + * destination pages of the backing file. + */ +static int do_lo_send_write(struct loop_device *lo, struct bio_vec *bvec, + int bsize, loff_t pos, struct page *page) +{ + int ret = lo_do_transfer(lo, WRITE, page, 0, bvec->bv_page, + bvec->bv_offset, bvec->bv_len, pos >> 9); + if (likely(!ret)) + return __do_lo_send_write(lo->lo_backing_file, + (u8 __user *)page_address(page), bvec->bv_len, + pos); + printk(KERN_ERR "loop: Transfer error at byte offset %llu, " + "length %i.\n", (unsigned long long)pos, bvec->bv_len); + if (ret > 0) + ret = -EIO; + return ret; +} + +static int lo_send(struct loop_device *lo, struct bio *bio, int bsize, + loff_t pos) +{ + int (*do_lo_send)(struct loop_device *, struct bio_vec *, int, loff_t, + struct page *page); + struct bio_vec *bvec; + struct page *page = NULL; + int i, ret = 0; + + do_lo_send = do_lo_send_aops; + if (!(lo->lo_flags & LO_FLAGS_USE_AOPS)) { + do_lo_send = do_lo_send_direct_write; + if (lo->transfer != transfer_none) { + page = alloc_page(GFP_NOIO | __GFP_HIGHMEM); + if (unlikely(!page)) + goto fail; + kmap(page); + do_lo_send = do_lo_send_write; + } + } + bio_for_each_segment(bvec, bio, i) { + ret = do_lo_send(lo, bvec, bsize, pos, page); + if (ret < 0) + break; + pos += bvec->bv_len; + } + if (page) { + kunmap(page); + __free_page(page); + } +out: + return ret; +fail: + printk(KERN_ERR "loop: Failed to allocate temporary page for write.\n"); + ret = -ENOMEM; + goto out; +} + +struct lo_read_data { + struct loop_device *lo; + struct page *page; + unsigned offset; + int bsize; +}; + +static int +lo_read_actor(read_descriptor_t *desc, struct page *page, + unsigned long offset, unsigned long size) +{ + unsigned long count = desc->count; + struct lo_read_data *p = desc->arg.data; + struct loop_device *lo = p->lo; + sector_t IV; + + IV = ((sector_t) page->index << (PAGE_CACHE_SHIFT - 9))+(offset >> 9); + + if (size > count) + size = count; + + if (lo_do_transfer(lo, READ, page, offset, p->page, p->offset, size, IV)) { + size = 0; + printk(KERN_ERR "loop: transfer error block %ld\n", + page->index); + desc->error = -EINVAL; + } + + flush_dcache_page(p->page); + + desc->count = count - size; + desc->written += size; + p->offset += size; + return size; +} + +static int +do_lo_receive(struct loop_device *lo, + struct bio_vec *bvec, int bsize, loff_t pos) +{ + struct lo_read_data cookie; + struct file *file; + int retval; + + cookie.lo = lo; + cookie.page = bvec->bv_page; + cookie.offset = bvec->bv_offset; + cookie.bsize = bsize; + file = lo->lo_backing_file; + retval = file->f_op->sendfile(file, &pos, bvec->bv_len, + lo_read_actor, &cookie); + return (retval < 0)? retval: 0; +} + +static int +lo_receive(struct loop_device *lo, struct bio *bio, int bsize, loff_t pos) +{ + struct bio_vec *bvec; + int i, ret = 0; + + bio_for_each_segment(bvec, bio, i) { + ret = do_lo_receive(lo, bvec, bsize, pos); + if (ret < 0) + break; + pos += bvec->bv_len; + } + return ret; +} + +static int do_bio_filebacked(struct loop_device *lo, struct bio *bio) +{ + loff_t pos; + int ret; + + pos = ((loff_t) bio->bi_sector << 9) + lo->lo_offset; + if (bio_rw(bio) == WRITE) + ret = lo_send(lo, bio, lo->lo_blocksize, pos); + else + ret = lo_receive(lo, bio, lo->lo_blocksize, pos); + return ret; +} + +/* + * Add bio to back of pending list + */ +static void loop_add_bio(struct loop_device *lo, struct bio *bio) +{ + unsigned long flags; + + spin_lock_irqsave(&lo->lo_lock, flags); + if (lo->lo_biotail) { + lo->lo_biotail->bi_next = bio; + lo->lo_biotail = bio; + } else + lo->lo_bio = lo->lo_biotail = bio; + spin_unlock_irqrestore(&lo->lo_lock, flags); + + up(&lo->lo_bh_mutex); +} + +/* + * Grab first pending buffer + */ +static struct bio *loop_get_bio(struct loop_device *lo) +{ + struct bio *bio; + + spin_lock_irq(&lo->lo_lock); + if ((bio = lo->lo_bio)) { + if (bio == lo->lo_biotail) + lo->lo_biotail = NULL; + lo->lo_bio = bio->bi_next; + bio->bi_next = NULL; + } + spin_unlock_irq(&lo->lo_lock); + + return bio; +} + +static int loop_make_request(request_queue_t *q, struct bio *old_bio) +{ + struct loop_device *lo = q->queuedata; + int rw = bio_rw(old_bio); + + if (!lo) + goto out; + + spin_lock_irq(&lo->lo_lock); + if (lo->lo_state != Lo_bound) + goto inactive; + atomic_inc(&lo->lo_pending); + spin_unlock_irq(&lo->lo_lock); + + if (rw == WRITE) { + if (lo->lo_flags & LO_FLAGS_READ_ONLY) + goto err; + } else if (rw == READA) { + rw = READ; + } else if (rw != READ) { + printk(KERN_ERR "loop: unknown command (%x)\n", rw); + goto err; + } + loop_add_bio(lo, old_bio); + return 0; +err: + if (atomic_dec_and_test(&lo->lo_pending)) + up(&lo->lo_bh_mutex); +out: + bio_io_error(old_bio, old_bio->bi_size); + return 0; +inactive: + spin_unlock_irq(&lo->lo_lock); + goto out; +} + +/* + * kick off io on the underlying address space + */ +static void loop_unplug(request_queue_t *q) +{ + struct loop_device *lo = q->queuedata; + + clear_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags); + blk_run_address_space(lo->lo_backing_file->f_mapping); +} + +struct switch_request { + struct file *file; + struct completion wait; +}; + +static void do_loop_switch(struct loop_device *, struct switch_request *); + +static inline void loop_handle_bio(struct loop_device *lo, struct bio *bio) +{ + int ret; + + if (unlikely(!bio->bi_bdev)) { + do_loop_switch(lo, bio->bi_private); + bio_put(bio); + } else { + ret = do_bio_filebacked(lo, bio); + bio_endio(bio, bio->bi_size, ret); + } +} + +/* + * worker thread that handles reads/writes to file backed loop devices, + * to avoid blocking in our make_request_fn. it also does loop decrypting + * on reads for block backed loop, as that is too heavy to do from + * b_end_io context where irqs may be disabled. + */ +static int loop_thread(void *data) +{ + struct loop_device *lo = data; + struct bio *bio; + + daemonize("loop%d", lo->lo_number); + + /* + * loop can be used in an encrypted device, + * hence, it mustn't be stopped at all + * because it could be indirectly used during suspension + */ + current->flags |= PF_NOFREEZE; + + set_user_nice(current, -20); + + lo->lo_state = Lo_bound; + atomic_inc(&lo->lo_pending); + + /* + * up sem, we are running + */ + up(&lo->lo_sem); + + for (;;) { + down_interruptible(&lo->lo_bh_mutex); + /* + * could be upped because of tear-down, not because of + * pending work + */ + if (!atomic_read(&lo->lo_pending)) + break; + + bio = loop_get_bio(lo); + if (!bio) { + printk("loop: missing bio\n"); + continue; + } + loop_handle_bio(lo, bio); + + /* + * upped both for pending work and tear-down, lo_pending + * will hit zero then + */ + if (atomic_dec_and_test(&lo->lo_pending)) + break; + } + + up(&lo->lo_sem); + return 0; +} + +/* + * loop_switch performs the hard work of switching a backing store. + * First it needs to flush existing IO, it does this by sending a magic + * BIO down the pipe. The completion of this BIO does the actual switch. + */ +static int loop_switch(struct loop_device *lo, struct file *file) +{ + struct switch_request w; + struct bio *bio = bio_alloc(GFP_KERNEL, 1); + if (!bio) + return -ENOMEM; + init_completion(&w.wait); + w.file = file; + bio->bi_private = &w; + bio->bi_bdev = NULL; + loop_make_request(lo->lo_queue, bio); + wait_for_completion(&w.wait); + return 0; +} + +/* + * Do the actual switch; called from the BIO completion routine + */ +static void do_loop_switch(struct loop_device *lo, struct switch_request *p) +{ + struct file *file = p->file; + struct file *old_file = lo->lo_backing_file; + struct address_space *mapping = file->f_mapping; + + mapping_set_gfp_mask(old_file->f_mapping, lo->old_gfp_mask); + lo->lo_backing_file = file; + lo->lo_blocksize = mapping->host->i_blksize; + lo->old_gfp_mask = mapping_gfp_mask(mapping); + mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS)); + complete(&p->wait); +} + + +/* + * loop_change_fd switched the backing store of a loopback device to + * a new file. This is useful for operating system installers to free up + * the original file and in High Availability environments to switch to + * an alternative location for the content in case of server meltdown. + * This can only work if the loop device is used read-only, and if the + * new backing store is the same size and type as the old backing store. + */ +static int loop_change_fd(struct loop_device *lo, struct file *lo_file, + struct block_device *bdev, unsigned int arg) +{ + struct file *file, *old_file; + struct inode *inode; + int error; + + error = -ENXIO; + if (lo->lo_state != Lo_bound) + goto out; + + /* the loop device has to be read-only */ + error = -EINVAL; + if (!(lo->lo_flags & LO_FLAGS_READ_ONLY)) + goto out; + + error = -EBADF; + file = fget(arg); + if (!file) + goto out; + + inode = file->f_mapping->host; + old_file = lo->lo_backing_file; + + error = -EINVAL; + + if (!S_ISREG(inode->i_mode) && !S_ISBLK(inode->i_mode)) + goto out_putf; + + /* new backing store needs to support loop (eg sendfile) */ + if (!inode->i_fop->sendfile) + goto out_putf; + + /* size of the new backing store needs to be the same */ + if (get_loop_size(lo, file) != get_loop_size(lo, old_file)) + goto out_putf; + + /* and ... switch */ + error = loop_switch(lo, file); + if (error) + goto out_putf; + + fput(old_file); + return 0; + + out_putf: + fput(file); + out: + return error; +} + +static inline int is_loop_device(struct file *file) +{ + struct inode *i = file->f_mapping->host; + + return i && S_ISBLK(i->i_mode) && MAJOR(i->i_rdev) == LOOP_MAJOR; +} + +static int loop_set_fd(struct loop_device *lo, struct file *lo_file, + struct block_device *bdev, unsigned int arg) +{ + struct file *file, *f; + struct inode *inode; + struct address_space *mapping; + unsigned lo_blocksize; + int lo_flags = 0; + int error; + loff_t size; + + /* This is safe, since we have a reference from open(). */ + __module_get(THIS_MODULE); + + error = -EBADF; + file = fget(arg); + if (!file) + goto out; + + error = -EBUSY; + if (lo->lo_state != Lo_unbound) + goto out_putf; + + /* Avoid recursion */ + f = file; + while (is_loop_device(f)) { + struct loop_device *l; + + if (f->f_mapping->host->i_rdev == lo_file->f_mapping->host->i_rdev) + goto out_putf; + + l = f->f_mapping->host->i_bdev->bd_disk->private_data; + if (l->lo_state == Lo_unbound) { + error = -EINVAL; + goto out_putf; + } + f = l->lo_backing_file; + } + + mapping = file->f_mapping; + inode = mapping->host; + + if (!(file->f_mode & FMODE_WRITE)) + lo_flags |= LO_FLAGS_READ_ONLY; + + error = -EINVAL; + if (S_ISREG(inode->i_mode) || S_ISBLK(inode->i_mode)) { + struct address_space_operations *aops = mapping->a_ops; + /* + * If we can't read - sorry. If we only can't write - well, + * it's going to be read-only. + */ + if (!file->f_op->sendfile) + goto out_putf; + if (aops->prepare_write && aops->commit_write) + lo_flags |= LO_FLAGS_USE_AOPS; + if (!(lo_flags & LO_FLAGS_USE_AOPS) && !file->f_op->write) + lo_flags |= LO_FLAGS_READ_ONLY; + + lo_blocksize = inode->i_blksize; + error = 0; + } else { + goto out_putf; + } + + size = get_loop_size(lo, file); + + if ((loff_t)(sector_t)size != size) { + error = -EFBIG; + goto out_putf; + } + + if (!(lo_file->f_mode & FMODE_WRITE)) + lo_flags |= LO_FLAGS_READ_ONLY; + + set_device_ro(bdev, (lo_flags & LO_FLAGS_READ_ONLY) != 0); + + lo->lo_blocksize = lo_blocksize; + lo->lo_device = bdev; + lo->lo_flags = lo_flags; + lo->lo_backing_file = file; + lo->transfer = NULL; + lo->ioctl = NULL; + lo->lo_sizelimit = 0; + lo->old_gfp_mask = mapping_gfp_mask(mapping); + mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS)); + + lo->lo_bio = lo->lo_biotail = NULL; + + /* + * set queue make_request_fn, and add limits based on lower level + * device + */ + blk_queue_make_request(lo->lo_queue, loop_make_request); + lo->lo_queue->queuedata = lo; + lo->lo_queue->unplug_fn = loop_unplug; + + set_capacity(disks[lo->lo_number], size); + bd_set_size(bdev, size << 9); + + set_blocksize(bdev, lo_blocksize); + + kernel_thread(loop_thread, lo, CLONE_KERNEL); + down(&lo->lo_sem); + return 0; + + out_putf: + fput(file); + out: + /* This is safe: open() is still holding a reference. */ + module_put(THIS_MODULE); + return error; +} + +static int +loop_release_xfer(struct loop_device *lo) +{ + int err = 0; + struct loop_func_table *xfer = lo->lo_encryption; + + if (xfer) { + if (xfer->release) + err = xfer->release(lo); + lo->transfer = NULL; + lo->lo_encryption = NULL; + module_put(xfer->owner); + } + return err; +} + +static int +loop_init_xfer(struct loop_device *lo, struct loop_func_table *xfer, + const struct loop_info64 *i) +{ + int err = 0; + + if (xfer) { + struct module *owner = xfer->owner; + + if (!try_module_get(owner)) + return -EINVAL; + if (xfer->init) + err = xfer->init(lo, i); + if (err) + module_put(owner); + else + lo->lo_encryption = xfer; + } + return err; +} + +static int loop_clr_fd(struct loop_device *lo, struct block_device *bdev) +{ + struct file *filp = lo->lo_backing_file; + int gfp = lo->old_gfp_mask; + + if (lo->lo_state != Lo_bound) + return -ENXIO; + + if (lo->lo_refcnt > 1) /* we needed one fd for the ioctl */ + return -EBUSY; + + if (filp == NULL) + return -EINVAL; + + spin_lock_irq(&lo->lo_lock); + lo->lo_state = Lo_rundown; + if (atomic_dec_and_test(&lo->lo_pending)) + up(&lo->lo_bh_mutex); + spin_unlock_irq(&lo->lo_lock); + + down(&lo->lo_sem); + + lo->lo_backing_file = NULL; + + loop_release_xfer(lo); + lo->transfer = NULL; + lo->ioctl = NULL; + lo->lo_device = NULL; + lo->lo_encryption = NULL; + lo->lo_offset = 0; + lo->lo_sizelimit = 0; + lo->lo_encrypt_key_size = 0; + lo->lo_flags = 0; + memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE); + memset(lo->lo_crypt_name, 0, LO_NAME_SIZE); + memset(lo->lo_file_name, 0, LO_NAME_SIZE); + invalidate_bdev(bdev, 0); + set_capacity(disks[lo->lo_number], 0); + bd_set_size(bdev, 0); + mapping_set_gfp_mask(filp->f_mapping, gfp); + lo->lo_state = Lo_unbound; + fput(filp); + /* This is safe: open() is still holding a reference. */ + module_put(THIS_MODULE); + return 0; +} + +static int +loop_set_status(struct loop_device *lo, const struct loop_info64 *info) +{ + int err; + struct loop_func_table *xfer; + + if (lo->lo_encrypt_key_size && lo->lo_key_owner != current->uid && + !capable(CAP_SYS_ADMIN)) + return -EPERM; + if (lo->lo_state != Lo_bound) + return -ENXIO; + if ((unsigned int) info->lo_encrypt_key_size > LO_KEY_SIZE) + return -EINVAL; + + err = loop_release_xfer(lo); + if (err) + return err; + + if (info->lo_encrypt_type) { + unsigned int type = info->lo_encrypt_type; + + if (type >= MAX_LO_CRYPT) + return -EINVAL; + xfer = xfer_funcs[type]; + if (xfer == NULL) + return -EINVAL; + } else + xfer = NULL; + + err = loop_init_xfer(lo, xfer, info); + if (err) + return err; + + if (lo->lo_offset != info->lo_offset || + lo->lo_sizelimit != info->lo_sizelimit) { + lo->lo_offset = info->lo_offset; + lo->lo_sizelimit = info->lo_sizelimit; + if (figure_loop_size(lo)) + return -EFBIG; + } + + memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE); + memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE); + lo->lo_file_name[LO_NAME_SIZE-1] = 0; + lo->lo_crypt_name[LO_NAME_SIZE-1] = 0; + + if (!xfer) + xfer = &none_funcs; + lo->transfer = xfer->transfer; + lo->ioctl = xfer->ioctl; + + lo->lo_encrypt_key_size = info->lo_encrypt_key_size; + lo->lo_init[0] = info->lo_init[0]; + lo->lo_init[1] = info->lo_init[1]; + if (info->lo_encrypt_key_size) { + memcpy(lo->lo_encrypt_key, info->lo_encrypt_key, + info->lo_encrypt_key_size); + lo->lo_key_owner = current->uid; + } + + return 0; +} + +static int +loop_get_status(struct loop_device *lo, struct loop_info64 *info) +{ + struct file *file = lo->lo_backing_file; + struct kstat stat; + int error; + + if (lo->lo_state != Lo_bound) + return -ENXIO; + error = vfs_getattr(file->f_vfsmnt, file->f_dentry, &stat); + if (error) + return error; + memset(info, 0, sizeof(*info)); + info->lo_number = lo->lo_number; + info->lo_device = huge_encode_dev(stat.dev); + info->lo_inode = stat.ino; + info->lo_rdevice = huge_encode_dev(lo->lo_device ? stat.rdev : stat.dev); + info->lo_offset = lo->lo_offset; + info->lo_sizelimit = lo->lo_sizelimit; + info->lo_flags = lo->lo_flags; + memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE); + memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE); + info->lo_encrypt_type = + lo->lo_encryption ? lo->lo_encryption->number : 0; + if (lo->lo_encrypt_key_size && capable(CAP_SYS_ADMIN)) { + info->lo_encrypt_key_size = lo->lo_encrypt_key_size; + memcpy(info->lo_encrypt_key, lo->lo_encrypt_key, + lo->lo_encrypt_key_size); + } + return 0; +} + +static void +loop_info64_from_old(const struct loop_info *info, struct loop_info64 *info64) +{ + memset(info64, 0, sizeof(*info64)); + info64->lo_number = info->lo_number; + info64->lo_device = info->lo_device; + info64->lo_inode = info->lo_inode; + info64->lo_rdevice = info->lo_rdevice; + info64->lo_offset = info->lo_offset; + info64->lo_sizelimit = 0; + info64->lo_encrypt_type = info->lo_encrypt_type; + info64->lo_encrypt_key_size = info->lo_encrypt_key_size; + info64->lo_flags = info->lo_flags; + info64->lo_init[0] = info->lo_init[0]; + info64->lo_init[1] = info->lo_init[1]; + if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI) + memcpy(info64->lo_crypt_name, info->lo_name, LO_NAME_SIZE); + else + memcpy(info64->lo_file_name, info->lo_name, LO_NAME_SIZE); + memcpy(info64->lo_encrypt_key, info->lo_encrypt_key, LO_KEY_SIZE); +} + +static int +loop_info64_to_old(const struct loop_info64 *info64, struct loop_info *info) +{ + memset(info, 0, sizeof(*info)); + info->lo_number = info64->lo_number; + info->lo_device = info64->lo_device; + info->lo_inode = info64->lo_inode; + info->lo_rdevice = info64->lo_rdevice; + info->lo_offset = info64->lo_offset; + info->lo_encrypt_type = info64->lo_encrypt_type; + info->lo_encrypt_key_size = info64->lo_encrypt_key_size; + info->lo_flags = info64->lo_flags; + info->lo_init[0] = info64->lo_init[0]; + info->lo_init[1] = info64->lo_init[1]; + if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI) + memcpy(info->lo_name, info64->lo_crypt_name, LO_NAME_SIZE); + else + memcpy(info->lo_name, info64->lo_file_name, LO_NAME_SIZE); + memcpy(info->lo_encrypt_key, info64->lo_encrypt_key, LO_KEY_SIZE); + + /* error in case values were truncated */ + if (info->lo_device != info64->lo_device || + info->lo_rdevice != info64->lo_rdevice || + info->lo_inode != info64->lo_inode || + info->lo_offset != info64->lo_offset) + return -EOVERFLOW; + + return 0; +} + +static int +loop_set_status_old(struct loop_device *lo, const struct loop_info __user *arg) +{ + struct loop_info info; + struct loop_info64 info64; + + if (copy_from_user(&info, arg, sizeof (struct loop_info))) + return -EFAULT; + loop_info64_from_old(&info, &info64); + return loop_set_status(lo, &info64); +} + +static int +loop_set_status64(struct loop_device *lo, const struct loop_info64 __user *arg) +{ + struct loop_info64 info64; + + if (copy_from_user(&info64, arg, sizeof (struct loop_info64))) + return -EFAULT; + return loop_set_status(lo, &info64); +} + +static int +loop_get_status_old(struct loop_device *lo, struct loop_info __user *arg) { + struct loop_info info; + struct loop_info64 info64; + int err = 0; + + if (!arg) + err = -EINVAL; + if (!err) + err = loop_get_status(lo, &info64); + if (!err) + err = loop_info64_to_old(&info64, &info); + if (!err && copy_to_user(arg, &info, sizeof(info))) + err = -EFAULT; + + return err; +} + +static int +loop_get_status64(struct loop_device *lo, struct loop_info64 __user *arg) { + struct loop_info64 info64; + int err = 0; + + if (!arg) + err = -EINVAL; + if (!err) + err = loop_get_status(lo, &info64); + if (!err && copy_to_user(arg, &info64, sizeof(info64))) + err = -EFAULT; + + return err; +} + +static int lo_ioctl(struct inode * inode, struct file * file, + unsigned int cmd, unsigned long arg) +{ + struct loop_device *lo = inode->i_bdev->bd_disk->private_data; + int err; + + down(&lo->lo_ctl_mutex); + switch (cmd) { + case LOOP_SET_FD: + err = loop_set_fd(lo, file, inode->i_bdev, arg); + break; + case LOOP_CHANGE_FD: + err = loop_change_fd(lo, file, inode->i_bdev, arg); + break; + case LOOP_CLR_FD: + err = loop_clr_fd(lo, inode->i_bdev); + break; + case LOOP_SET_STATUS: + err = loop_set_status_old(lo, (struct loop_info __user *) arg); + break; + case LOOP_GET_STATUS: + err = loop_get_status_old(lo, (struct loop_info __user *) arg); + break; + case LOOP_SET_STATUS64: + err = loop_set_status64(lo, (struct loop_info64 __user *) arg); + break; + case LOOP_GET_STATUS64: + err = loop_get_status64(lo, (struct loop_info64 __user *) arg); + break; + default: + err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL; + } + up(&lo->lo_ctl_mutex); + return err; +} + +static int lo_open(struct inode *inode, struct file *file) +{ + struct loop_device *lo = inode->i_bdev->bd_disk->private_data; + + down(&lo->lo_ctl_mutex); + lo->lo_refcnt++; + up(&lo->lo_ctl_mutex); + + return 0; +} + +static int lo_release(struct inode *inode, struct file *file) +{ + struct loop_device *lo = inode->i_bdev->bd_disk->private_data; + + down(&lo->lo_ctl_mutex); + --lo->lo_refcnt; + up(&lo->lo_ctl_mutex); + + return 0; +} + +static struct block_device_operations lo_fops = { + .owner = THIS_MODULE, + .open = lo_open, + .release = lo_release, + .ioctl = lo_ioctl, +}; + +/* + * And now the modules code and kernel interface. + */ +module_param(max_loop, int, 0); +MODULE_PARM_DESC(max_loop, "Maximum number of loop devices (1-256)"); +MODULE_LICENSE("GPL"); +MODULE_ALIAS_BLOCKDEV_MAJOR(LOOP_MAJOR); + +int loop_register_transfer(struct loop_func_table *funcs) +{ + unsigned int n = funcs->number; + + if (n >= MAX_LO_CRYPT || xfer_funcs[n]) + return -EINVAL; + xfer_funcs[n] = funcs; + return 0; +} + +int loop_unregister_transfer(int number) +{ + unsigned int n = number; + struct loop_device *lo; + struct loop_func_table *xfer; + + if (n == 0 || n >= MAX_LO_CRYPT || (xfer = xfer_funcs[n]) == NULL) + return -EINVAL; + + xfer_funcs[n] = NULL; + + for (lo = &loop_dev[0]; lo < &loop_dev[max_loop]; lo++) { + down(&lo->lo_ctl_mutex); + + if (lo->lo_encryption == xfer) + loop_release_xfer(lo); + + up(&lo->lo_ctl_mutex); + } + + return 0; +} + +EXPORT_SYMBOL(loop_register_transfer); +EXPORT_SYMBOL(loop_unregister_transfer); + +static int __init loop_init(void) +{ + int i; + + if (max_loop < 1 || max_loop > 256) { + printk(KERN_WARNING "loop: invalid max_loop (must be between" + " 1 and 256), using default (8)\n"); + max_loop = 8; + } + + if (register_blkdev(LOOP_MAJOR, "loop")) + return -EIO; + + loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL); + if (!loop_dev) + goto out_mem1; + memset(loop_dev, 0, max_loop * sizeof(struct loop_device)); + + disks = kmalloc(max_loop * sizeof(struct gendisk *), GFP_KERNEL); + if (!disks) + goto out_mem2; + + for (i = 0; i < max_loop; i++) { + disks[i] = alloc_disk(1); + if (!disks[i]) + goto out_mem3; + } + + devfs_mk_dir("loop"); + + for (i = 0; i < max_loop; i++) { + struct loop_device *lo = &loop_dev[i]; + struct gendisk *disk = disks[i]; + + memset(lo, 0, sizeof(*lo)); + lo->lo_queue = blk_alloc_queue(GFP_KERNEL); + if (!lo->lo_queue) + goto out_mem4; + init_MUTEX(&lo->lo_ctl_mutex); + init_MUTEX_LOCKED(&lo->lo_sem); + init_MUTEX_LOCKED(&lo->lo_bh_mutex); + lo->lo_number = i; + spin_lock_init(&lo->lo_lock); + disk->major = LOOP_MAJOR; + disk->first_minor = i; + disk->fops = &lo_fops; + sprintf(disk->disk_name, "loop%d", i); + sprintf(disk->devfs_name, "loop/%d", i); + disk->private_data = lo; + disk->queue = lo->lo_queue; + } + + /* We cannot fail after we call this, so another loop!*/ + for (i = 0; i < max_loop; i++) + add_disk(disks[i]); + printk(KERN_INFO "loop: loaded (max %d devices)\n", max_loop); + return 0; + +out_mem4: + while (i--) + blk_put_queue(loop_dev[i].lo_queue); + devfs_remove("loop"); + i = max_loop; +out_mem3: + while (i--) + put_disk(disks[i]); + kfree(disks); +out_mem2: + kfree(loop_dev); +out_mem1: + unregister_blkdev(LOOP_MAJOR, "loop"); + printk(KERN_ERR "loop: ran out of memory\n"); + return -ENOMEM; +} + +static void loop_exit(void) +{ + int i; + + for (i = 0; i < max_loop; i++) { + del_gendisk(disks[i]); + blk_put_queue(loop_dev[i].lo_queue); + put_disk(disks[i]); + } + devfs_remove("loop"); + if (unregister_blkdev(LOOP_MAJOR, "loop")) + printk(KERN_WARNING "loop: cannot unregister blkdev\n"); + + kfree(disks); + kfree(loop_dev); +} + +module_init(loop_init); +module_exit(loop_exit); + +#ifndef MODULE +static int __init max_loop_setup(char *str) +{ + max_loop = simple_strtol(str, NULL, 0); + return 1; +} + +__setup("max_loop=", max_loop_setup); +#endif diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c new file mode 100644 index 00000000000..efdf04450bf --- /dev/null +++ b/drivers/block/nbd.c @@ -0,0 +1,731 @@ +/* + * Network block device - make block devices work over TCP + * + * Note that you can not swap over this thing, yet. Seems to work but + * deadlocks sometimes - you can not swap over TCP in general. + * + * Copyright 1997-2000 Pavel Machek <pavel@ucw.cz> + * Parts copyright 2001 Steven Whitehouse <steve@chygwyn.com> + * + * (part of code stolen from loop.c) + * + * 97-3-25 compiled 0-th version, not yet tested it + * (it did not work, BTW) (later that day) HEY! it works! + * (bit later) hmm, not that much... 2:00am next day: + * yes, it works, but it gives something like 50kB/sec + * 97-4-01 complete rewrite to make it possible for many requests at + * once to be processed + * 97-4-11 Making protocol independent of endianity etc. + * 97-9-13 Cosmetic changes + * 98-5-13 Attempt to make 64-bit-clean on 64-bit machines + * 99-1-11 Attempt to make 64-bit-clean on 32-bit machines <ankry@mif.pg.gda.pl> + * 01-2-27 Fix to store proper blockcount for kernel (calculated using + * BLOCK_SIZE_BITS, not device blocksize) <aga@permonline.ru> + * 01-3-11 Make nbd work with new Linux block layer code. It now supports + * plugging like all the other block devices. Also added in MSG_MORE to + * reduce number of partial TCP segments sent. <steve@chygwyn.com> + * 01-12-6 Fix deadlock condition by making queue locks independent of + * the transmit lock. <steve@chygwyn.com> + * 02-10-11 Allow hung xmit to be aborted via SIGKILL & various fixes. + * <Paul.Clements@SteelEye.com> <James.Bottomley@SteelEye.com> + * 03-06-22 Make nbd work with new linux 2.5 block layer design. This fixes + * memory corruption from module removal and possible memory corruption + * from sending/receiving disk data. <ldl@aros.net> + * 03-06-23 Cosmetic changes. <ldl@aros.net> + * 03-06-23 Enhance diagnostics support. <ldl@aros.net> + * 03-06-24 Remove unneeded blksize_bits field from nbd_device struct. + * <ldl@aros.net> + * 03-06-24 Cleanup PARANOIA usage & code. <ldl@aros.net> + * 04-02-19 Remove PARANOIA, plus various cleanups (Paul Clements) + * possible FIXME: make set_sock / set_blksize / set_size / do_it one syscall + * why not: would need access_ok and friends, would share yet another + * structure with userland + */ + +#include <linux/major.h> + +#include <linux/blkdev.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/sched.h> +#include <linux/fs.h> +#include <linux/bio.h> +#include <linux/stat.h> +#include <linux/errno.h> +#include <linux/file.h> +#include <linux/ioctl.h> +#include <net/sock.h> + +#include <linux/devfs_fs_kernel.h> + +#include <asm/uaccess.h> +#include <asm/types.h> + +#include <linux/nbd.h> + +#define LO_MAGIC 0x68797548 + +#ifdef NDEBUG +#define dprintk(flags, fmt...) +#else /* NDEBUG */ +#define dprintk(flags, fmt...) do { \ + if (debugflags & (flags)) printk(KERN_DEBUG fmt); \ +} while (0) +#define DBG_IOCTL 0x0004 +#define DBG_INIT 0x0010 +#define DBG_EXIT 0x0020 +#define DBG_BLKDEV 0x0100 +#define DBG_RX 0x0200 +#define DBG_TX 0x0400 +static unsigned int debugflags; +#endif /* NDEBUG */ + +static struct nbd_device nbd_dev[MAX_NBD]; + +/* + * Use just one lock (or at most 1 per NIC). Two arguments for this: + * 1. Each NIC is essentially a synchronization point for all servers + * accessed through that NIC so there's no need to have more locks + * than NICs anyway. + * 2. More locks lead to more "Dirty cache line bouncing" which will slow + * down each lock to the point where they're actually slower than just + * a single lock. + * Thanks go to Jens Axboe and Al Viro for their LKML emails explaining this! + */ +static DEFINE_SPINLOCK(nbd_lock); + +#ifndef NDEBUG +static const char *ioctl_cmd_to_ascii(int cmd) +{ + switch (cmd) { + case NBD_SET_SOCK: return "set-sock"; + case NBD_SET_BLKSIZE: return "set-blksize"; + case NBD_SET_SIZE: return "set-size"; + case NBD_DO_IT: return "do-it"; + case NBD_CLEAR_SOCK: return "clear-sock"; + case NBD_CLEAR_QUE: return "clear-que"; + case NBD_PRINT_DEBUG: return "print-debug"; + case NBD_SET_SIZE_BLOCKS: return "set-size-blocks"; + case NBD_DISCONNECT: return "disconnect"; + case BLKROSET: return "set-read-only"; + case BLKFLSBUF: return "flush-buffer-cache"; + } + return "unknown"; +} + +static const char *nbdcmd_to_ascii(int cmd) +{ + switch (cmd) { + case NBD_CMD_READ: return "read"; + case NBD_CMD_WRITE: return "write"; + case NBD_CMD_DISC: return "disconnect"; + } + return "invalid"; +} +#endif /* NDEBUG */ + +static void nbd_end_request(struct request *req) +{ + int uptodate = (req->errors == 0) ? 1 : 0; + request_queue_t *q = req->q; + unsigned long flags; + + dprintk(DBG_BLKDEV, "%s: request %p: %s\n", req->rq_disk->disk_name, + req, uptodate? "done": "failed"); + + spin_lock_irqsave(q->queue_lock, flags); + if (!end_that_request_first(req, uptodate, req->nr_sectors)) { + end_that_request_last(req); + } + spin_unlock_irqrestore(q->queue_lock, flags); +} + +/* + * Send or receive packet. + */ +static int sock_xmit(struct socket *sock, int send, void *buf, int size, + int msg_flags) +{ + int result; + struct msghdr msg; + struct kvec iov; + unsigned long flags; + sigset_t oldset; + + /* Allow interception of SIGKILL only + * Don't allow other signals to interrupt the transmission */ + spin_lock_irqsave(¤t->sighand->siglock, flags); + oldset = current->blocked; + sigfillset(¤t->blocked); + sigdelsetmask(¤t->blocked, sigmask(SIGKILL)); + recalc_sigpending(); + spin_unlock_irqrestore(¤t->sighand->siglock, flags); + + do { + sock->sk->sk_allocation = GFP_NOIO; + iov.iov_base = buf; + iov.iov_len = size; + msg.msg_name = NULL; + msg.msg_namelen = 0; + msg.msg_control = NULL; + msg.msg_controllen = 0; + msg.msg_namelen = 0; + msg.msg_flags = msg_flags | MSG_NOSIGNAL; + + if (send) + result = kernel_sendmsg(sock, &msg, &iov, 1, size); + else + result = kernel_recvmsg(sock, &msg, &iov, 1, size, 0); + + if (signal_pending(current)) { + siginfo_t info; + spin_lock_irqsave(¤t->sighand->siglock, flags); + printk(KERN_WARNING "nbd (pid %d: %s) got signal %d\n", + current->pid, current->comm, + dequeue_signal(current, ¤t->blocked, &info)); + spin_unlock_irqrestore(¤t->sighand->siglock, flags); + result = -EINTR; + break; + } + + if (result <= 0) { + if (result == 0) + result = -EPIPE; /* short read */ + break; + } + size -= result; + buf += result; + } while (size > 0); + + spin_lock_irqsave(¤t->sighand->siglock, flags); + current->blocked = oldset; + recalc_sigpending(); + spin_unlock_irqrestore(¤t->sighand->siglock, flags); + + return result; +} + +static inline int sock_send_bvec(struct socket *sock, struct bio_vec *bvec, + int flags) +{ + int result; + void *kaddr = kmap(bvec->bv_page); + result = sock_xmit(sock, 1, kaddr + bvec->bv_offset, bvec->bv_len, + flags); + kunmap(bvec->bv_page); + return result; +} + +static int nbd_send_req(struct nbd_device *lo, struct request *req) +{ + int result, i, flags; + struct nbd_request request; + unsigned long size = req->nr_sectors << 9; + struct socket *sock = lo->sock; + + request.magic = htonl(NBD_REQUEST_MAGIC); + request.type = htonl(nbd_cmd(req)); + request.from = cpu_to_be64((u64) req->sector << 9); + request.len = htonl(size); + memcpy(request.handle, &req, sizeof(req)); + + down(&lo->tx_lock); + + if (!sock || !lo->sock) { + printk(KERN_ERR "%s: Attempted send on closed socket\n", + lo->disk->disk_name); + goto error_out; + } + + dprintk(DBG_TX, "%s: request %p: sending control (%s@%llu,%luB)\n", + lo->disk->disk_name, req, + nbdcmd_to_ascii(nbd_cmd(req)), + (unsigned long long)req->sector << 9, + req->nr_sectors << 9); + result = sock_xmit(sock, 1, &request, sizeof(request), + (nbd_cmd(req) == NBD_CMD_WRITE)? MSG_MORE: 0); + if (result <= 0) { + printk(KERN_ERR "%s: Send control failed (result %d)\n", + lo->disk->disk_name, result); + goto error_out; + } + + if (nbd_cmd(req) == NBD_CMD_WRITE) { + struct bio *bio; + /* + * we are really probing at internals to determine + * whether to set MSG_MORE or not... + */ + rq_for_each_bio(bio, req) { + struct bio_vec *bvec; + bio_for_each_segment(bvec, bio, i) { + flags = 0; + if ((i < (bio->bi_vcnt - 1)) || bio->bi_next) + flags = MSG_MORE; + dprintk(DBG_TX, "%s: request %p: sending %d bytes data\n", + lo->disk->disk_name, req, + bvec->bv_len); + result = sock_send_bvec(sock, bvec, flags); + if (result <= 0) { + printk(KERN_ERR "%s: Send data failed (result %d)\n", + lo->disk->disk_name, + result); + goto error_out; + } + } + } + } + up(&lo->tx_lock); + return 0; + +error_out: + up(&lo->tx_lock); + return 1; +} + +static struct request *nbd_find_request(struct nbd_device *lo, char *handle) +{ + struct request *req; + struct list_head *tmp; + struct request *xreq; + + memcpy(&xreq, handle, sizeof(xreq)); + + spin_lock(&lo->queue_lock); + list_for_each(tmp, &lo->queue_head) { + req = list_entry(tmp, struct request, queuelist); + if (req != xreq) + continue; + list_del_init(&req->queuelist); + spin_unlock(&lo->queue_lock); + return req; + } + spin_unlock(&lo->queue_lock); + return NULL; +} + +static inline int sock_recv_bvec(struct socket *sock, struct bio_vec *bvec) +{ + int result; + void *kaddr = kmap(bvec->bv_page); + result = sock_xmit(sock, 0, kaddr + bvec->bv_offset, bvec->bv_len, + MSG_WAITALL); + kunmap(bvec->bv_page); + return result; +} + +/* NULL returned = something went wrong, inform userspace */ +static struct request *nbd_read_stat(struct nbd_device *lo) +{ + int result; + struct nbd_reply reply; + struct request *req; + struct socket *sock = lo->sock; + + reply.magic = 0; + result = sock_xmit(sock, 0, &reply, sizeof(reply), MSG_WAITALL); + if (result <= 0) { + printk(KERN_ERR "%s: Receive control failed (result %d)\n", + lo->disk->disk_name, result); + goto harderror; + } + req = nbd_find_request(lo, reply.handle); + if (req == NULL) { + printk(KERN_ERR "%s: Unexpected reply (%p)\n", + lo->disk->disk_name, reply.handle); + result = -EBADR; + goto harderror; + } + + if (ntohl(reply.magic) != NBD_REPLY_MAGIC) { + printk(KERN_ERR "%s: Wrong magic (0x%lx)\n", + lo->disk->disk_name, + (unsigned long)ntohl(reply.magic)); + result = -EPROTO; + goto harderror; + } + if (ntohl(reply.error)) { + printk(KERN_ERR "%s: Other side returned error (%d)\n", + lo->disk->disk_name, ntohl(reply.error)); + req->errors++; + return req; + } + + dprintk(DBG_RX, "%s: request %p: got reply\n", + lo->disk->disk_name, req); + if (nbd_cmd(req) == NBD_CMD_READ) { + int i; + struct bio *bio; + rq_for_each_bio(bio, req) { + struct bio_vec *bvec; + bio_for_each_segment(bvec, bio, i) { + result = sock_recv_bvec(sock, bvec); + if (result <= 0) { + printk(KERN_ERR "%s: Receive data failed (result %d)\n", + lo->disk->disk_name, + result); + goto harderror; + } + dprintk(DBG_RX, "%s: request %p: got %d bytes data\n", + lo->disk->disk_name, req, bvec->bv_len); + } + } + } + return req; +harderror: + lo->harderror = result; + return NULL; +} + +static void nbd_do_it(struct nbd_device *lo) +{ + struct request *req; + + BUG_ON(lo->magic != LO_MAGIC); + + while ((req = nbd_read_stat(lo)) != NULL) + nbd_end_request(req); + return; +} + +static void nbd_clear_que(struct nbd_device *lo) +{ + struct request *req; + + BUG_ON(lo->magic != LO_MAGIC); + + do { + req = NULL; + spin_lock(&lo->queue_lock); + if (!list_empty(&lo->queue_head)) { + req = list_entry(lo->queue_head.next, struct request, queuelist); + list_del_init(&req->queuelist); + } + spin_unlock(&lo->queue_lock); + if (req) { + req->errors++; + nbd_end_request(req); + } + } while (req); +} + +/* + * We always wait for result of write, for now. It would be nice to make it optional + * in future + * if ((req->cmd == WRITE) && (lo->flags & NBD_WRITE_NOCHK)) + * { printk( "Warning: Ignoring result!\n"); nbd_end_request( req ); } + */ + +static void do_nbd_request(request_queue_t * q) +{ + struct request *req; + + while ((req = elv_next_request(q)) != NULL) { + struct nbd_device *lo; + + blkdev_dequeue_request(req); + dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%lx)\n", + req->rq_disk->disk_name, req, req->flags); + + if (!(req->flags & REQ_CMD)) + goto error_out; + + lo = req->rq_disk->private_data; + + BUG_ON(lo->magic != LO_MAGIC); + + if (!lo->file) { + printk(KERN_ERR "%s: Request when not-ready\n", + lo->disk->disk_name); + goto error_out; + } + nbd_cmd(req) = NBD_CMD_READ; + if (rq_data_dir(req) == WRITE) { + nbd_cmd(req) = NBD_CMD_WRITE; + if (lo->flags & NBD_READ_ONLY) { + printk(KERN_ERR "%s: Write on read-only\n", + lo->disk->disk_name); + goto error_out; + } + } + + req->errors = 0; + spin_unlock_irq(q->queue_lock); + + spin_lock(&lo->queue_lock); + + if (!lo->file) { + spin_unlock(&lo->queue_lock); + printk(KERN_ERR "%s: failed between accept and semaphore, file lost\n", + lo->disk->disk_name); + req->errors++; + nbd_end_request(req); + spin_lock_irq(q->queue_lock); + continue; + } + + list_add(&req->queuelist, &lo->queue_head); + spin_unlock(&lo->queue_lock); + + if (nbd_send_req(lo, req) != 0) { + printk(KERN_ERR "%s: Request send failed\n", + lo->disk->disk_name); + if (nbd_find_request(lo, (char *)&req) != NULL) { + /* we still own req */ + req->errors++; + nbd_end_request(req); + } else /* we're racing with nbd_clear_que */ + printk(KERN_DEBUG "nbd: can't find req\n"); + } + + spin_lock_irq(q->queue_lock); + continue; + +error_out: + req->errors++; + spin_unlock(q->queue_lock); + nbd_end_request(req); + spin_lock(q->queue_lock); + } + return; +} + +static int nbd_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + struct nbd_device *lo = inode->i_bdev->bd_disk->private_data; + int error; + struct request sreq ; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + BUG_ON(lo->magic != LO_MAGIC); + + /* Anyone capable of this syscall can do *real bad* things */ + dprintk(DBG_IOCTL, "%s: nbd_ioctl cmd=%s(0x%x) arg=%lu\n", + lo->disk->disk_name, ioctl_cmd_to_ascii(cmd), cmd, arg); + + switch (cmd) { + case NBD_DISCONNECT: + printk(KERN_INFO "%s: NBD_DISCONNECT\n", lo->disk->disk_name); + sreq.flags = REQ_SPECIAL; + nbd_cmd(&sreq) = NBD_CMD_DISC; + /* + * Set these to sane values in case server implementation + * fails to check the request type first and also to keep + * debugging output cleaner. + */ + sreq.sector = 0; + sreq.nr_sectors = 0; + if (!lo->sock) + return -EINVAL; + nbd_send_req(lo, &sreq); + return 0; + + case NBD_CLEAR_SOCK: + error = 0; + down(&lo->tx_lock); + lo->sock = NULL; + up(&lo->tx_lock); + spin_lock(&lo->queue_lock); + file = lo->file; + lo->file = NULL; + spin_unlock(&lo->queue_lock); + nbd_clear_que(lo); + spin_lock(&lo->queue_lock); + if (!list_empty(&lo->queue_head)) { + printk(KERN_ERR "nbd: disconnect: some requests are in progress -> please try again.\n"); + error = -EBUSY; + } + spin_unlock(&lo->queue_lock); + if (file) + fput(file); + return error; + case NBD_SET_SOCK: + if (lo->file) + return -EBUSY; + error = -EINVAL; + file = fget(arg); + if (file) { + inode = file->f_dentry->d_inode; + if (S_ISSOCK(inode->i_mode)) { + lo->file = file; + lo->sock = SOCKET_I(inode); + error = 0; + } else { + fput(file); + } + } + return error; + case NBD_SET_BLKSIZE: + lo->blksize = arg; + lo->bytesize &= ~(lo->blksize-1); + inode->i_bdev->bd_inode->i_size = lo->bytesize; + set_blocksize(inode->i_bdev, lo->blksize); + set_capacity(lo->disk, lo->bytesize >> 9); + return 0; + case NBD_SET_SIZE: + lo->bytesize = arg & ~(lo->blksize-1); + inode->i_bdev->bd_inode->i_size = lo->bytesize; + set_blocksize(inode->i_bdev, lo->blksize); + set_capacity(lo->disk, lo->bytesize >> 9); + return 0; + case NBD_SET_SIZE_BLOCKS: + lo->bytesize = ((u64) arg) * lo->blksize; + inode->i_bdev->bd_inode->i_size = lo->bytesize; + set_blocksize(inode->i_bdev, lo->blksize); + set_capacity(lo->disk, lo->bytesize >> 9); + return 0; + case NBD_DO_IT: + if (!lo->file) + return -EINVAL; + nbd_do_it(lo); + /* on return tidy up in case we have a signal */ + /* Forcibly shutdown the socket causing all listeners + * to error + * + * FIXME: This code is duplicated from sys_shutdown, but + * there should be a more generic interface rather than + * calling socket ops directly here */ + down(&lo->tx_lock); + if (lo->sock) { + printk(KERN_WARNING "%s: shutting down socket\n", + lo->disk->disk_name); + lo->sock->ops->shutdown(lo->sock, + SEND_SHUTDOWN|RCV_SHUTDOWN); + lo->sock = NULL; + } + up(&lo->tx_lock); + spin_lock(&lo->queue_lock); + file = lo->file; + lo->file = NULL; + spin_unlock(&lo->queue_lock); + nbd_clear_que(lo); + printk(KERN_WARNING "%s: queue cleared\n", lo->disk->disk_name); + if (file) + fput(file); + return lo->harderror; + case NBD_CLEAR_QUE: + down(&lo->tx_lock); + if (lo->sock) { + up(&lo->tx_lock); + return 0; /* probably should be error, but that would + * break "nbd-client -d", so just return 0 */ + } + up(&lo->tx_lock); + nbd_clear_que(lo); + return 0; + case NBD_PRINT_DEBUG: + printk(KERN_INFO "%s: next = %p, prev = %p, head = %p\n", + inode->i_bdev->bd_disk->disk_name, + lo->queue_head.next, lo->queue_head.prev, + &lo->queue_head); + return 0; + } + return -EINVAL; +} + +static struct block_device_operations nbd_fops = +{ + .owner = THIS_MODULE, + .ioctl = nbd_ioctl, +}; + +/* + * And here should be modules and kernel interface + * (Just smiley confuses emacs :-) + */ + +static int __init nbd_init(void) +{ + int err = -ENOMEM; + int i; + + if (sizeof(struct nbd_request) != 28) { + printk(KERN_CRIT "nbd: sizeof nbd_request needs to be 28 in order to work!\n" ); + return -EIO; + } + + for (i = 0; i < MAX_NBD; i++) { + struct gendisk *disk = alloc_disk(1); + if (!disk) + goto out; + nbd_dev[i].disk = disk; + /* + * The new linux 2.5 block layer implementation requires + * every gendisk to have its very own request_queue struct. + * These structs are big so we dynamically allocate them. + */ + disk->queue = blk_init_queue(do_nbd_request, &nbd_lock); + if (!disk->queue) { + put_disk(disk); + goto out; + } + } + + if (register_blkdev(NBD_MAJOR, "nbd")) { + err = -EIO; + goto out; + } + + printk(KERN_INFO "nbd: registered device at major %d\n", NBD_MAJOR); + dprintk(DBG_INIT, "nbd: debugflags=0x%x\n", debugflags); + + devfs_mk_dir("nbd"); + for (i = 0; i < MAX_NBD; i++) { + struct gendisk *disk = nbd_dev[i].disk; + nbd_dev[i].file = NULL; + nbd_dev[i].magic = LO_MAGIC; + nbd_dev[i].flags = 0; + spin_lock_init(&nbd_dev[i].queue_lock); + INIT_LIST_HEAD(&nbd_dev[i].queue_head); + init_MUTEX(&nbd_dev[i].tx_lock); + nbd_dev[i].blksize = 1024; + nbd_dev[i].bytesize = 0x7ffffc00ULL << 10; /* 2TB */ + disk->major = NBD_MAJOR; + disk->first_minor = i; + disk->fops = &nbd_fops; + disk->private_data = &nbd_dev[i]; + disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO; + sprintf(disk->disk_name, "nbd%d", i); + sprintf(disk->devfs_name, "nbd/%d", i); + set_capacity(disk, 0x7ffffc00ULL << 1); /* 2 TB */ + add_disk(disk); + } + + return 0; +out: + while (i--) { + blk_cleanup_queue(nbd_dev[i].disk->queue); + put_disk(nbd_dev[i].disk); + } + return err; +} + +static void __exit nbd_cleanup(void) +{ + int i; + for (i = 0; i < MAX_NBD; i++) { + struct gendisk *disk = nbd_dev[i].disk; + if (disk) { + del_gendisk(disk); + blk_cleanup_queue(disk->queue); + put_disk(disk); + } + } + devfs_remove("nbd"); + unregister_blkdev(NBD_MAJOR, "nbd"); + printk(KERN_INFO "nbd: unregistered device at major %d\n", NBD_MAJOR); +} + +module_init(nbd_init); +module_exit(nbd_cleanup); + +MODULE_DESCRIPTION("Network Block Device"); +MODULE_LICENSE("GPL"); + +#ifndef NDEBUG +module_param(debugflags, int, 0644); +MODULE_PARM_DESC(debugflags, "flags for controlling debug output"); +#endif diff --git a/drivers/block/noop-iosched.c b/drivers/block/noop-iosched.c new file mode 100644 index 00000000000..888c477e02b --- /dev/null +++ b/drivers/block/noop-iosched.c @@ -0,0 +1,104 @@ +/* + * elevator noop + */ +#include <linux/blkdev.h> +#include <linux/elevator.h> +#include <linux/bio.h> +#include <linux/module.h> +#include <linux/init.h> + +/* + * See if we can find a request that this buffer can be coalesced with. + */ +static int elevator_noop_merge(request_queue_t *q, struct request **req, + struct bio *bio) +{ + struct list_head *entry = &q->queue_head; + struct request *__rq; + int ret; + + if ((ret = elv_try_last_merge(q, bio))) { + *req = q->last_merge; + return ret; + } + + while ((entry = entry->prev) != &q->queue_head) { + __rq = list_entry_rq(entry); + + if (__rq->flags & (REQ_SOFTBARRIER | REQ_HARDBARRIER)) + break; + else if (__rq->flags & REQ_STARTED) + break; + + if (!blk_fs_request(__rq)) + continue; + + if ((ret = elv_try_merge(__rq, bio))) { + *req = __rq; + q->last_merge = __rq; + return ret; + } + } + + return ELEVATOR_NO_MERGE; +} + +static void elevator_noop_merge_requests(request_queue_t *q, struct request *req, + struct request *next) +{ + list_del_init(&next->queuelist); +} + +static void elevator_noop_add_request(request_queue_t *q, struct request *rq, + int where) +{ + if (where == ELEVATOR_INSERT_FRONT) + list_add(&rq->queuelist, &q->queue_head); + else + list_add_tail(&rq->queuelist, &q->queue_head); + + /* + * new merges must not precede this barrier + */ + if (rq->flags & REQ_HARDBARRIER) + q->last_merge = NULL; + else if (!q->last_merge) + q->last_merge = rq; +} + +static struct request *elevator_noop_next_request(request_queue_t *q) +{ + if (!list_empty(&q->queue_head)) + return list_entry_rq(q->queue_head.next); + + return NULL; +} + +static struct elevator_type elevator_noop = { + .ops = { + .elevator_merge_fn = elevator_noop_merge, + .elevator_merge_req_fn = elevator_noop_merge_requests, + .elevator_next_req_fn = elevator_noop_next_request, + .elevator_add_req_fn = elevator_noop_add_request, + }, + .elevator_name = "noop", + .elevator_owner = THIS_MODULE, +}; + +static int __init noop_init(void) +{ + return elv_register(&elevator_noop); +} + +static void __exit noop_exit(void) +{ + elv_unregister(&elevator_noop); +} + +module_init(noop_init); +module_exit(noop_exit); + + +MODULE_AUTHOR("Jens Axboe"); +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("No-op IO scheduler"); diff --git a/drivers/block/paride/Kconfig b/drivers/block/paride/Kconfig new file mode 100644 index 00000000000..17ff4056125 --- /dev/null +++ b/drivers/block/paride/Kconfig @@ -0,0 +1,305 @@ +# +# PARIDE configuration +# +# PARIDE doesn't need PARPORT, but if PARPORT is configured as a module, +# PARIDE must also be a module. The bogus CONFIG_PARIDE_PARPORT option +# controls the choices given to the user ... +config PARIDE_PARPORT + tristate + depends on PARIDE!=n + default m if PARPORT=m + default y if PARPORT!=m + +comment "Parallel IDE high-level drivers" + depends on PARIDE + +config PARIDE_PD + tristate "Parallel port IDE disks" + depends on PARIDE + help + This option enables the high-level driver for IDE-type disk devices + connected through a parallel port. If you chose to build PARIDE + support into your kernel, you may answer Y here to build in the + parallel port IDE driver, otherwise you should answer M to build + it as a loadable module. The module will be called pd. You + must also have at least one parallel port protocol driver in your + system. Among the devices supported by this driver are the SyQuest + EZ-135, EZ-230 and SparQ drives, the Avatar Shark and the backpack + hard drives from MicroSolutions. + +config PARIDE_PCD + tristate "Parallel port ATAPI CD-ROMs" + depends on PARIDE + ---help--- + This option enables the high-level driver for ATAPI CD-ROM devices + connected through a parallel port. If you chose to build PARIDE + support into your kernel, you may answer Y here to build in the + parallel port ATAPI CD-ROM driver, otherwise you should answer M to + build it as a loadable module. The module will be called pcd. You + must also have at least one parallel port protocol driver in your + system. Among the devices supported by this driver are the + MicroSolutions backpack CD-ROM drives and the Freecom Power CD. If + you have such a CD-ROM drive, you should also say Y or M to "ISO + 9660 CD-ROM file system support" below, because that's the file + system used on CD-ROMs. + +config PARIDE_PF + tristate "Parallel port ATAPI disks" + depends on PARIDE + help + This option enables the high-level driver for ATAPI disk devices + connected through a parallel port. If you chose to build PARIDE + support into your kernel, you may answer Y here to build in the + parallel port ATAPI disk driver, otherwise you should answer M + to build it as a loadable module. The module will be called pf. + You must also have at least one parallel port protocol driver in + your system. Among the devices supported by this driver are the + MicroSolutions backpack PD/CD drive and the Imation Superdisk + LS-120 drive. + +config PARIDE_PT + tristate "Parallel port ATAPI tapes" + depends on PARIDE + help + This option enables the high-level driver for ATAPI tape devices + connected through a parallel port. If you chose to build PARIDE + support into your kernel, you may answer Y here to build in the + parallel port ATAPI disk driver, otherwise you should answer M + to build it as a loadable module. The module will be called pt. + You must also have at least one parallel port protocol driver in + your system. Among the devices supported by this driver is the + parallel port version of the HP 5GB drive. + +config PARIDE_PG + tristate "Parallel port generic ATAPI devices" + depends on PARIDE + ---help--- + This option enables a special high-level driver for generic ATAPI + devices connected through a parallel port. The driver allows user + programs, such as cdrtools, to send ATAPI commands directly to a + device. + + If you chose to build PARIDE support into your kernel, you may + answer Y here to build in the parallel port generic ATAPI driver, + otherwise you should answer M to build it as a loadable module. The + module will be called pg. + + You must also have at least one parallel port protocol driver in + your system. + + This driver implements an API loosely related to the generic SCSI + driver. See <file:include/linux/pg.h>. for details. + + You can obtain the most recent version of cdrtools from + <ftp://ftp.berlios.de/pub/cdrecord/>. Versions 1.6.1a3 and + later fully support this driver. + +comment "Parallel IDE protocol modules" + depends on PARIDE + +config PARIDE_ATEN + tristate "ATEN EH-100 protocol" + depends on PARIDE + help + This option enables support for the ATEN EH-100 parallel port IDE + protocol. This protocol is used in some inexpensive low performance + parallel port kits made in Hong Kong. If you chose to build PARIDE + support into your kernel, you may answer Y here to build in the + protocol driver, otherwise you should answer M to build it as a + loadable module. The module will be called aten. You must also + have a high-level driver for the type of device that you want to + support. + +config PARIDE_BPCK + tristate "MicroSolutions backpack (Series 5) protocol" + depends on PARIDE + ---help--- + This option enables support for the Micro Solutions BACKPACK + parallel port Series 5 IDE protocol. (Most BACKPACK drives made + before 1999 were Series 5) Series 5 drives will NOT always have the + Series noted on the bottom of the drive. Series 6 drivers will. + + In other words, if your BACKPACK drive doesn't say "Series 6" on the + bottom, enable this option. + + If you chose to build PARIDE support into your kernel, you may + answer Y here to build in the protocol driver, otherwise you should + answer M to build it as a loadable module. The module will be + called bpck. You must also have a high-level driver for the type + of device that you want to support. + +config PARIDE_BPCK6 + tristate "MicroSolutions backpack (Series 6) protocol" + depends on PARIDE && !64BIT + ---help--- + This option enables support for the Micro Solutions BACKPACK + parallel port Series 6 IDE protocol. (Most BACKPACK drives made + after 1999 were Series 6) Series 6 drives will have the Series noted + on the bottom of the drive. Series 5 drivers don't always have it + noted. + + In other words, if your BACKPACK drive says "Series 6" on the + bottom, enable this option. + + If you chose to build PARIDE support into your kernel, you may + answer Y here to build in the protocol driver, otherwise you should + answer M to build it as a loadable module. The module will be + called bpck6. You must also have a high-level driver for the type + of device that you want to support. + +config PARIDE_COMM + tristate "DataStor Commuter protocol" + depends on PARIDE + help + This option enables support for the Commuter parallel port IDE + protocol from DataStor. If you chose to build PARIDE support + into your kernel, you may answer Y here to build in the protocol + driver, otherwise you should answer M to build it as a loadable + module. The module will be called comm. You must also have + a high-level driver for the type of device that you want to support. + +config PARIDE_DSTR + tristate "DataStor EP-2000 protocol" + depends on PARIDE + help + This option enables support for the EP-2000 parallel port IDE + protocol from DataStor. If you chose to build PARIDE support + into your kernel, you may answer Y here to build in the protocol + driver, otherwise you should answer M to build it as a loadable + module. The module will be called dstr. You must also have + a high-level driver for the type of device that you want to support. + +config PARIDE_FIT2 + tristate "FIT TD-2000 protocol" + depends on PARIDE + help + This option enables support for the TD-2000 parallel port IDE + protocol from Fidelity International Technology. This is a simple + (low speed) adapter that is used in some portable hard drives. If + you chose to build PARIDE support into your kernel, you may answer Y + here to build in the protocol driver, otherwise you should answer M + to build it as a loadable module. The module will be called ktti. + You must also have a high-level driver for the type of device that + you want to support. + +config PARIDE_FIT3 + tristate "FIT TD-3000 protocol" + depends on PARIDE + help + This option enables support for the TD-3000 parallel port IDE + protocol from Fidelity International Technology. This protocol is + used in newer models of their portable disk, CD-ROM and PD/CD + devices. If you chose to build PARIDE support into your kernel, you + may answer Y here to build in the protocol driver, otherwise you + should answer M to build it as a loadable module. The module will be + called fit3. You must also have a high-level driver for the type + of device that you want to support. + +config PARIDE_EPAT + tristate "Shuttle EPAT/EPEZ protocol" + depends on PARIDE + help + This option enables support for the EPAT parallel port IDE protocol. + EPAT is a parallel port IDE adapter manufactured by Shuttle + Technology and widely used in devices from major vendors such as + Hewlett-Packard, SyQuest, Imation and Avatar. If you chose to build + PARIDE support into your kernel, you may answer Y here to build in + the protocol driver, otherwise you should answer M to build it as a + loadable module. The module will be called epat. You must also + have a high-level driver for the type of device that you want to + support. + +config PARIDE_EPATC8 + bool "Support c7/c8 chips (EXPERIMENTAL)" + depends on PARIDE_EPAT && EXPERIMENTAL + help + This option enables support for the newer Shuttle EP1284 (aka c7 and + c8) chip. You need this if you are using any recent Imation SuperDisk + (LS-120) drive. + +config PARIDE_EPIA + tristate "Shuttle EPIA protocol" + depends on PARIDE + help + This option enables support for the (obsolete) EPIA parallel port + IDE protocol from Shuttle Technology. This adapter can still be + found in some no-name kits. If you chose to build PARIDE support + into your kernel, you may answer Y here to build in the protocol + driver, otherwise you should answer M to build it as a loadable + module. The module will be called epia. You must also have a + high-level driver for the type of device that you want to support. + +config PARIDE_FRIQ + tristate "Freecom IQ ASIC-2 protocol" + depends on PARIDE + help + This option enables support for version 2 of the Freecom IQ parallel + port IDE adapter. This adapter is used by the Maxell Superdisk + drive. If you chose to build PARIDE support into your kernel, you + may answer Y here to build in the protocol driver, otherwise you + should answer M to build it as a loadable module. The module will be + called friq. You must also have a high-level driver for the type + of device that you want to support. + +config PARIDE_FRPW + tristate "FreeCom power protocol" + depends on PARIDE + help + This option enables support for the Freecom power parallel port IDE + protocol. If you chose to build PARIDE support into your kernel, you + may answer Y here to build in the protocol driver, otherwise you + should answer M to build it as a loadable module. The module will be + called frpw. You must also have a high-level driver for the type + of device that you want to support. + +config PARIDE_KBIC + tristate "KingByte KBIC-951A/971A protocols" + depends on PARIDE + help + This option enables support for the KBIC-951A and KBIC-971A parallel + port IDE protocols from KingByte Information Corp. KingByte's + adapters appear in many no-name portable disk and CD-ROM products, + especially in Europe. If you chose to build PARIDE support into your + kernel, you may answer Y here to build in the protocol driver, + otherwise you should answer M to build it as a loadable module. The + module will be called kbic. You must also have a high-level driver + for the type of device that you want to support. + +config PARIDE_KTTI + tristate "KT PHd protocol" + depends on PARIDE + help + This option enables support for the "PHd" parallel port IDE protocol + from KT Technology. This is a simple (low speed) adapter that is + used in some 2.5" portable hard drives. If you chose to build PARIDE + support into your kernel, you may answer Y here to build in the + protocol driver, otherwise you should answer M to build it as a + loadable module. The module will be called ktti. You must also + have a high-level driver for the type of device that you want to + support. + +config PARIDE_ON20 + tristate "OnSpec 90c20 protocol" + depends on PARIDE + help + This option enables support for the (obsolete) 90c20 parallel port + IDE protocol from OnSpec (often marketed under the ValuStore brand + name). If you chose to build PARIDE support into your kernel, you + may answer Y here to build in the protocol driver, otherwise you + should answer M to build it as a loadable module. The module will + be called on20. You must also have a high-level driver for the + type of device that you want to support. + +config PARIDE_ON26 + tristate "OnSpec 90c26 protocol" + depends on PARIDE + help + This option enables support for the 90c26 parallel port IDE protocol + from OnSpec Electronics (often marketed under the ValuStore brand + name). If you chose to build PARIDE support into your kernel, you + may answer Y here to build in the protocol driver, otherwise you + should answer M to build it as a loadable module. The module will be + called on26. You must also have a high-level driver for the type + of device that you want to support. + +# diff --git a/drivers/block/paride/Makefile b/drivers/block/paride/Makefile new file mode 100644 index 00000000000..a539e004bb7 --- /dev/null +++ b/drivers/block/paride/Makefile @@ -0,0 +1,28 @@ +# +# Makefile for Parallel port IDE device drivers. +# +# 7 October 2000, Bartlomiej Zolnierkiewicz <bkz@linux-ide.org> +# Rewritten to use lists instead of if-statements. +# + +obj-$(CONFIG_PARIDE) += paride.o +obj-$(CONFIG_PARIDE_ATEN) += aten.o +obj-$(CONFIG_PARIDE_BPCK) += bpck.o +obj-$(CONFIG_PARIDE_COMM) += comm.o +obj-$(CONFIG_PARIDE_DSTR) += dstr.o +obj-$(CONFIG_PARIDE_KBIC) += kbic.o +obj-$(CONFIG_PARIDE_EPAT) += epat.o +obj-$(CONFIG_PARIDE_EPIA) += epia.o +obj-$(CONFIG_PARIDE_FRPW) += frpw.o +obj-$(CONFIG_PARIDE_FRIQ) += friq.o +obj-$(CONFIG_PARIDE_FIT2) += fit2.o +obj-$(CONFIG_PARIDE_FIT3) += fit3.o +obj-$(CONFIG_PARIDE_ON20) += on20.o +obj-$(CONFIG_PARIDE_ON26) += on26.o +obj-$(CONFIG_PARIDE_KTTI) += ktti.o +obj-$(CONFIG_PARIDE_BPCK6) += bpck6.o +obj-$(CONFIG_PARIDE_PD) += pd.o +obj-$(CONFIG_PARIDE_PCD) += pcd.o +obj-$(CONFIG_PARIDE_PF) += pf.o +obj-$(CONFIG_PARIDE_PT) += pt.o +obj-$(CONFIG_PARIDE_PG) += pg.o diff --git a/drivers/block/paride/Transition-notes b/drivers/block/paride/Transition-notes new file mode 100644 index 00000000000..70374907c02 --- /dev/null +++ b/drivers/block/paride/Transition-notes @@ -0,0 +1,128 @@ +Lemma 1: + If ps_tq is scheduled, ps_tq_active is 1. ps_tq_int() can be called + only when ps_tq_active is 1. +Proof: All assignments to ps_tq_active and all scheduling of ps_tq happen + under ps_spinlock. There are three places where that can happen: + one in ps_set_intr() (A) and two in ps_tq_int() (B and C). + Consider the sequnce of these events. A can not be preceded by + anything except B, since it is under if (!ps_tq_active) under + ps_spinlock. C is always preceded by B, since we can't reach it + other than through B and we don't drop ps_spinlock between them. + IOW, the sequence is A?(BA|BC|B)*. OTOH, number of B can not exceed + the sum of numbers of A and C, since each call of ps_tq_int() is + the result of ps_tq execution. Therefore, the sequence starts with + A and each B is preceded by either A or C. Moments when we enter + ps_tq_int() are sandwiched between {A,C} and B in that sequence, + since at any time number of B can not exceed the number of these + moments which, in turn, can not exceed the number of A and C. + In other words, the sequence of events is (A or C set ps_tq_active to + 1 and schedule ps_tq, ps_tq is executed, ps_tq_int() is entered, + B resets ps_tq_active)*. + + +consider the following area: + * in do_pd_request1(): to calls of pi_do_claimed() and return in + case when pd_req is NULL. + * in next_request(): to call of do_pd_request1() + * in do_pd_read(): to call of ps_set_intr() + * in do_pd_read_start(): to calls of pi_do_claimed(), next_request() +and ps_set_intr() + * in do_pd_read_drq(): to calls of pi_do_claimed() and next_request() + * in do_pd_write(): to call of ps_set_intr() + * in do_pd_write_start(): to calls of pi_do_claimed(), next_request() +and ps_set_intr() + * in do_pd_write_done(): to calls of pi_do_claimed() and next_request() + * in ps_set_intr(): to check for ps_tq_active and to scheduling + ps_tq if ps_tq_active was 0. + * in ps_tq_int(): from the moment when we get ps_spinlock() to the + return, call of con() or scheduling ps_tq. + * in pi_schedule_claimed() when called from pi_do_claimed() called from + pd.c, everything until returning 1 or setting or setting ->claim_cont + on the path that returns 0 + * in pi_do_claimed() when called from pd.c, everything until the call + of pi_do_claimed() plus the everything until the call of cont() if + pi_do_claimed() has returned 1. + * in pi_wake_up() called for PIA that belongs to pd.c, everything from + the moment when pi_spinlock has been acquired. + +Lemma 2: + 1) at any time at most one thread of execution can be in that area or + be preempted there. + 2) When there is such a thread, pd_busy is set or pd_lock is held by + that thread. + 3) When there is such a thread, ps_tq_active is 0 or ps_spinlock is + held by that thread. + 4) When there is such a thread, all PIA belonging to pd.c have NULL + ->claim_cont or pi_spinlock is held by thread in question. + +Proof: consider the first moment when the above is not true. + +(1) can become not true if some thread enters that area while another is there. + a) do_pd_request1() can be called from next_request() or do_pd_request() + In the first case the thread was already in the area. In the second, + the thread was holding pd_lock and found pd_busy not set, which would + mean that (2) was already not true. + b) ps_set_intr() and pi_schedule_claimed() can be called only from the + area. + c) pi_do_claimed() is called by pd.c only from the area. + d) ps_tq_int() can enter the area only when the thread is holding + ps_spinlock and ps_tq_active is 1 (due to Lemma 1). It means that + (3) was already not true. + e) do_pd_{read,write}* could be called only from the area. The only + case that needs consideration is call from pi_wake_up() and there + we would have to be called for the PIA that got ->claimed_cont + from pd.c. That could happen only if pi_do_claimed() had been + called from pd.c for that PIA, which happens only for PIA belonging + to pd.c. + f) pi_wake_up() can enter the area only when the thread is holding + pi_spinlock and ->claimed_cont is non-NULL for PIA belonging to + pd.c. It means that (4) was already not true. + +(2) can become not true only when pd_lock is released by the thread in question. + Indeed, pd_busy is reset only in the area and thread that resets + it is holding pd_lock. The only place within the area where we + release pd_lock is in pd_next_buf() (called from within the area). + But that code does not reset pd_busy, so pd_busy would have to be + 0 when pd_next_buf() had acquired pd_lock. If it become 0 while + we were acquiring the lock, (1) would be already false, since + the thread that had reset it would be in the area simulateously. + If it was 0 before we tried to acquire pd_lock, (2) would be + already false. + +For similar reasons, (3) can become not true only when ps_spinlock is released +by the thread in question. However, all such places within the area are right +after resetting ps_tq_active to 0. + +(4) is done the same way - all places where we release pi_spinlock within +the area are either after resetting ->claimed_cont to NULL while holding +pi_spinlock, or after not tocuhing ->claimed_cont since acquiring pi_spinlock +also in the area. The only place where ->claimed_cont is made non-NULL is +in the area, under pi_spinlock and we do not release it until after leaving +the area. + +QED. + + +Corollary 1: ps_tq_active can be killed. Indeed, the only place where we +check its value is in ps_set_intr() and if it had been non-zero at that +point, we would have violated either (2.1) (if it was set while ps_set_intr() +was acquiring ps_spinlock) or (2.3) (if it was set when we started to +acquire ps_spinlock). + +Corollary 2: ps_spinlock can be killed. Indeed, Lemma 1 and Lemma 2 show +that the only possible contention is between scheduling ps_tq followed by +immediate release of spinlock and beginning of execution of ps_tq on +another CPU. + +Corollary 3: assignment to pd_busy in do_pd_read_start() and do_pd_write_start() +can be killed. Indeed, we are not holding pd_lock and thus pd_busy is already +1 here. + +Corollary 4: in ps_tq_int() uses of con can be replaced with uses of +ps_continuation, since the latter is changed only from the area. +We don't need to reset it to NULL, since we are guaranteed that there +will be a call of ps_set_intr() before we look at ps_continuation again. +We can remove the check for ps_continuation being NULL for the same +reason - the value is guaranteed to be set by the last ps_set_intr() and +we never pass it NULL. Assignements in the beginning of ps_set_intr() +can be taken to callers as long as they remain within the area. diff --git a/drivers/block/paride/aten.c b/drivers/block/paride/aten.c new file mode 100644 index 00000000000..c4d696d43dc --- /dev/null +++ b/drivers/block/paride/aten.c @@ -0,0 +1,162 @@ +/* + aten.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + aten.c is a low-level protocol driver for the ATEN EH-100 + parallel port adapter. The EH-100 supports 4-bit and 8-bit + modes only. There is also an EH-132 which supports EPP mode + transfers. The EH-132 is not yet supported. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.05 init_proto, release_proto + +*/ + +#define ATEN_VERSION "1.01" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/wait.h> +#include <linux/types.h> +#include <asm/io.h> + +#include "paride.h" + +#define j44(a,b) ((((a>>4)&0x0f)|(b&0xf0))^0x88) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int cont_map[2] = { 0x08, 0x20 }; + +static void aten_write_regr( PIA *pi, int cont, int regr, int val) + +{ int r; + + r = regr + cont_map[cont] + 0x80; + + w0(r); w2(0xe); w2(6); w0(val); w2(7); w2(6); w2(0xc); +} + +static int aten_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, r; + + r = regr + cont_map[cont] + 0x40; + + switch (pi->mode) { + + case 0: w0(r); w2(0xe); w2(6); + w2(7); w2(6); w2(0); + a = r1(); w0(0x10); b = r1(); w2(0xc); + return j44(a,b); + + case 1: r |= 0x10; + w0(r); w2(0xe); w2(6); w0(0xff); + w2(0x27); w2(0x26); w2(0x20); + a = r0(); + w2(0x26); w2(0xc); + return a; + } + return -1; +} + +static void aten_read_block( PIA *pi, char * buf, int count ) + +{ int k, a, b, c, d; + + switch (pi->mode) { + + case 0: w0(0x48); w2(0xe); w2(6); + for (k=0;k<count/2;k++) { + w2(7); w2(6); w2(2); + a = r1(); w0(0x58); b = r1(); + w2(0); d = r1(); w0(0x48); c = r1(); + buf[2*k] = j44(c,d); + buf[2*k+1] = j44(a,b); + } + w2(0xc); + break; + + case 1: w0(0x58); w2(0xe); w2(6); + for (k=0;k<count/2;k++) { + w2(0x27); w2(0x26); w2(0x22); + a = r0(); w2(0x20); b = r0(); + buf[2*k] = b; buf[2*k+1] = a; + } + w2(0x26); w2(0xc); + break; + } +} + +static void aten_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + w0(0x88); w2(0xe); w2(6); + for (k=0;k<count/2;k++) { + w0(buf[2*k+1]); w2(0xe); w2(6); + w0(buf[2*k]); w2(7); w2(6); + } + w2(0xc); +} + +static void aten_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(0xc); +} + +static void aten_disconnect ( PIA *pi ) + +{ w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void aten_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[2] = {"4-bit","8-bit"}; + + printk("%s: aten %s, ATEN EH-100 at 0x%x, ", + pi->device,ATEN_VERSION,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol aten = { + .owner = THIS_MODULE, + .name = "aten", + .max_mode = 2, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = aten_write_regr, + .read_regr = aten_read_regr, + .write_block = aten_write_block, + .read_block = aten_read_block, + .connect = aten_connect, + .disconnect = aten_disconnect, + .log_adapter = aten_log_adapter, +}; + +static int __init aten_init(void) +{ + return pi_register(&aten)-1; +} + +static void __exit aten_exit(void) +{ + pi_unregister( &aten ); +} + +MODULE_LICENSE("GPL"); +module_init(aten_init) +module_exit(aten_exit) diff --git a/drivers/block/paride/bpck.c b/drivers/block/paride/bpck.c new file mode 100644 index 00000000000..d462ff6b139 --- /dev/null +++ b/drivers/block/paride/bpck.c @@ -0,0 +1,477 @@ +/* + bpck.c (c) 1996-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + bpck.c is a low-level protocol driver for the MicroSolutions + "backpack" parallel port IDE adapter. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.05 init_proto, release_proto, pi->delay + 1.02 GRG 1998.08.15 default pi->delay returned to 4 + +*/ + +#define BPCK_VERSION "1.02" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#undef r2 +#undef w2 + +#define PC pi->private +#define r2() (PC=(in_p(2) & 0xff)) +#define w2(byte) {out_p(2,byte); PC = byte;} +#define t2(pat) {PC ^= pat; out_p(2,PC);} +#define e2() {PC &= 0xfe; out_p(2,PC);} +#define o2() {PC |= 1; out_p(2,PC);} + +#define j44(l,h) (((l>>3)&0x7)|((l>>4)&0x8)|((h<<1)&0x70)|(h&0x80)) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set + cont = 2 - use internal bpck register addressing +*/ + +static int cont_map[3] = { 0x40, 0x48, 0 }; + +static int bpck_read_regr( PIA *pi, int cont, int regr ) + +{ int r, l, h; + + r = regr + cont_map[cont]; + + switch (pi->mode) { + + case 0: w0(r & 0xf); w0(r); t2(2); t2(4); + l = r1(); + t2(4); + h = r1(); + return j44(l,h); + + case 1: w0(r & 0xf); w0(r); t2(2); + e2(); t2(0x20); + t2(4); h = r0(); + t2(1); t2(0x20); + return h; + + case 2: + case 3: + case 4: w0(r); w2(9); w2(0); w2(0x20); + h = r4(); + w2(0); + return h; + + } + return -1; +} + +static void bpck_write_regr( PIA *pi, int cont, int regr, int val ) + +{ int r; + + r = regr + cont_map[cont]; + + switch (pi->mode) { + + case 0: + case 1: w0(r); + t2(2); + w0(val); + o2(); t2(4); t2(1); + break; + + case 2: + case 3: + case 4: w0(r); w2(9); w2(0); + w0(val); w2(1); w2(3); w2(0); + break; + + } +} + +/* These macros access the bpck registers in native addressing */ + +#define WR(r,v) bpck_write_regr(pi,2,r,v) +#define RR(r) (bpck_read_regr(pi,2,r)) + +static void bpck_write_block( PIA *pi, char * buf, int count ) + +{ int i; + + switch (pi->mode) { + + case 0: WR(4,0x40); + w0(0x40); t2(2); t2(1); + for (i=0;i<count;i++) { w0(buf[i]); t2(4); } + WR(4,0); + break; + + case 1: WR(4,0x50); + w0(0x40); t2(2); t2(1); + for (i=0;i<count;i++) { w0(buf[i]); t2(4); } + WR(4,0x10); + break; + + case 2: WR(4,0x48); + w0(0x40); w2(9); w2(0); w2(1); + for (i=0;i<count;i++) w4(buf[i]); + w2(0); + WR(4,8); + break; + + case 3: WR(4,0x48); + w0(0x40); w2(9); w2(0); w2(1); + for (i=0;i<count/2;i++) w4w(((u16 *)buf)[i]); + w2(0); + WR(4,8); + break; + + case 4: WR(4,0x48); + w0(0x40); w2(9); w2(0); w2(1); + for (i=0;i<count/4;i++) w4l(((u32 *)buf)[i]); + w2(0); + WR(4,8); + break; + } +} + +static void bpck_read_block( PIA *pi, char * buf, int count ) + +{ int i, l, h; + + switch (pi->mode) { + + case 0: WR(4,0x40); + w0(0x40); t2(2); + for (i=0;i<count;i++) { + t2(4); l = r1(); + t2(4); h = r1(); + buf[i] = j44(l,h); + } + WR(4,0); + break; + + case 1: WR(4,0x50); + w0(0x40); t2(2); t2(0x20); + for(i=0;i<count;i++) { t2(4); buf[i] = r0(); } + t2(1); t2(0x20); + WR(4,0x10); + break; + + case 2: WR(4,0x48); + w0(0x40); w2(9); w2(0); w2(0x20); + for (i=0;i<count;i++) buf[i] = r4(); + w2(0); + WR(4,8); + break; + + case 3: WR(4,0x48); + w0(0x40); w2(9); w2(0); w2(0x20); + for (i=0;i<count/2;i++) ((u16 *)buf)[i] = r4w(); + w2(0); + WR(4,8); + break; + + case 4: WR(4,0x48); + w0(0x40); w2(9); w2(0); w2(0x20); + for (i=0;i<count/4;i++) ((u32 *)buf)[i] = r4l(); + w2(0); + WR(4,8); + break; + + } +} + +static int bpck_probe_unit ( PIA *pi ) + +{ int o1, o0, f7, id; + int t, s; + + id = pi->unit; + s = 0; + w2(4); w2(0xe); r2(); t2(2); + o1 = r1()&0xf8; + o0 = r0(); + w0(255-id); w2(4); w0(id); + t2(8); t2(8); t2(8); + t2(2); t = r1()&0xf8; + f7 = ((id % 8) == 7); + if ((f7) || (t != o1)) { t2(2); s = r1()&0xf8; } + if ((t == o1) && ((!f7) || (s == o1))) { + w2(0x4c); w0(o0); + return 0; + } + t2(8); w0(0); t2(2); w2(0x4c); w0(o0); + return 1; +} + +static void bpck_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + w0(0xff-pi->unit); w2(4); w0(pi->unit); + t2(8); t2(8); t2(8); + t2(2); t2(2); + + switch (pi->mode) { + + case 0: t2(8); WR(4,0); + break; + + case 1: t2(8); WR(4,0x10); + break; + + case 2: + case 3: + case 4: w2(0); WR(4,8); + break; + + } + + WR(5,8); + + if (pi->devtype == PI_PCD) { + WR(0x46,0x10); /* fiddle with ESS logic ??? */ + WR(0x4c,0x38); + WR(0x4d,0x88); + WR(0x46,0xa0); + WR(0x41,0); + WR(0x4e,8); + } +} + +static void bpck_disconnect ( PIA *pi ) + +{ w0(0); + if (pi->mode >= 2) { w2(9); w2(0); } else t2(2); + w2(0x4c); w0(pi->saved_r0); +} + +static void bpck_force_spp ( PIA *pi ) + +/* This fakes the EPP protocol to turn off EPP ... */ + +{ pi->saved_r0 = r0(); + w0(0xff-pi->unit); w2(4); w0(pi->unit); + t2(8); t2(8); t2(8); + t2(2); t2(2); + + w2(0); + w0(4); w2(9); w2(0); + w0(0); w2(1); w2(3); w2(0); + w0(0); w2(9); w2(0); + w2(0x4c); w0(pi->saved_r0); +} + +#define TEST_LEN 16 + +static int bpck_test_proto( PIA *pi, char * scratch, int verbose ) + +{ int i, e, l, h, om; + char buf[TEST_LEN]; + + bpck_force_spp(pi); + + switch (pi->mode) { + + case 0: bpck_connect(pi); + WR(0x13,0x7f); + w0(0x13); t2(2); + for(i=0;i<TEST_LEN;i++) { + t2(4); l = r1(); + t2(4); h = r1(); + buf[i] = j44(l,h); + } + bpck_disconnect(pi); + break; + + case 1: bpck_connect(pi); + WR(0x13,0x7f); + w0(0x13); t2(2); t2(0x20); + for(i=0;i<TEST_LEN;i++) { t2(4); buf[i] = r0(); } + t2(1); t2(0x20); + bpck_disconnect(pi); + break; + + case 2: + case 3: + case 4: om = pi->mode; + pi->mode = 0; + bpck_connect(pi); + WR(7,3); + WR(4,8); + bpck_disconnect(pi); + + pi->mode = om; + bpck_connect(pi); + w0(0x13); w2(9); w2(1); w0(0); w2(3); w2(0); w2(0xe0); + + switch (pi->mode) { + case 2: for (i=0;i<TEST_LEN;i++) buf[i] = r4(); + break; + case 3: for (i=0;i<TEST_LEN/2;i++) ((u16 *)buf)[i] = r4w(); + break; + case 4: for (i=0;i<TEST_LEN/4;i++) ((u32 *)buf)[i] = r4l(); + break; + } + + w2(0); + WR(7,0); + bpck_disconnect(pi); + + break; + + } + + if (verbose) { + printk("%s: bpck: 0x%x unit %d mode %d: ", + pi->device,pi->port,pi->unit,pi->mode); + for (i=0;i<TEST_LEN;i++) printk("%3d",buf[i]); + printk("\n"); + } + + e = 0; + for (i=0;i<TEST_LEN;i++) if (buf[i] != (i+1)) e++; + return e; +} + +static void bpck_read_eeprom ( PIA *pi, char * buf ) + +{ int i,j,k,n,p,v,f, om, od; + + bpck_force_spp(pi); + + om = pi->mode; od = pi->delay; + pi->mode = 0; pi->delay = 6; + + bpck_connect(pi); + + n = 0; + WR(4,0); + for (i=0;i<64;i++) { + WR(6,8); + WR(6,0xc); + p = 0x100; + for (k=0;k<9;k++) { + f = (((i + 0x180) & p) != 0) * 2; + WR(6,f+0xc); + WR(6,f+0xd); + WR(6,f+0xc); + p = (p >> 1); + } + for (j=0;j<2;j++) { + v = 0; + for (k=0;k<8;k++) { + WR(6,0xc); + WR(6,0xd); + WR(6,0xc); + f = RR(0); + v = 2*v + (f == 0x84); + } + buf[2*i+1-j] = v; + } + } + WR(6,8); + WR(6,0); + WR(5,8); + + bpck_disconnect(pi); + + if (om >= 2) { + bpck_connect(pi); + WR(7,3); + WR(4,8); + bpck_disconnect(pi); + } + + pi->mode = om; pi->delay = od; +} + +static int bpck_test_port ( PIA *pi ) /* check for 8-bit port */ + +{ int i, r, m; + + w2(0x2c); i = r0(); w0(255-i); r = r0(); w0(i); + m = -1; + if (r == i) m = 2; + if (r == (255-i)) m = 0; + + w2(0xc); i = r0(); w0(255-i); r = r0(); w0(i); + if (r != (255-i)) m = -1; + + if (m == 0) { w2(6); w2(0xc); r = r0(); w0(0xaa); w0(r); w0(0xaa); } + if (m == 2) { w2(0x26); w2(0xc); } + + if (m == -1) return 0; + return 5; +} + +static void bpck_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[5] = { "4-bit","8-bit","EPP-8", + "EPP-16","EPP-32" }; + +#ifdef DUMP_EEPROM + int i; +#endif + + bpck_read_eeprom(pi,scratch); + +#ifdef DUMP_EEPROM + if (verbose) { + for(i=0;i<128;i++) + if ((scratch[i] < ' ') || (scratch[i] > '~')) + scratch[i] = '.'; + printk("%s: bpck EEPROM: %64.64s\n",pi->device,scratch); + printk("%s: %64.64s\n",pi->device,&scratch[64]); + } +#endif + + printk("%s: bpck %s, backpack %8.8s unit %d", + pi->device,BPCK_VERSION,&scratch[110],pi->unit); + printk(" at 0x%x, mode %d (%s), delay %d\n",pi->port, + pi->mode,mode_string[pi->mode],pi->delay); +} + +static struct pi_protocol bpck = { + .owner = THIS_MODULE, + .name = "bpck", + .max_mode = 5, + .epp_first = 2, + .default_delay = 4, + .max_units = 255, + .write_regr = bpck_write_regr, + .read_regr = bpck_read_regr, + .write_block = bpck_write_block, + .read_block = bpck_read_block, + .connect = bpck_connect, + .disconnect = bpck_disconnect, + .test_port = bpck_test_port, + .probe_unit = bpck_probe_unit, + .test_proto = bpck_test_proto, + .log_adapter = bpck_log_adapter, +}; + +static int __init bpck_init(void) +{ + return pi_register(&bpck)-1; +} + +static void __exit bpck_exit(void) +{ + pi_unregister(&bpck); +} + +MODULE_LICENSE("GPL"); +module_init(bpck_init) +module_exit(bpck_exit) diff --git a/drivers/block/paride/bpck6.c b/drivers/block/paride/bpck6.c new file mode 100644 index 00000000000..08d858ad64d --- /dev/null +++ b/drivers/block/paride/bpck6.c @@ -0,0 +1,282 @@ +/* + backpack.c (c) 2001 Micro Solutions Inc. + Released under the terms of the GNU General Public license + + backpack.c is a low-level protocol driver for the Micro Solutions + "BACKPACK" parallel port IDE adapter + (Works on Series 6 drives) + + Written by: Ken Hahn (linux-dev@micro-solutions.com) + Clive Turvey (linux-dev@micro-solutions.com) + +*/ + +/* + This is Ken's linux wrapper for the PPC library + Version 1.0.0 is the backpack driver for which source is not available + Version 2.0.0 is the first to have source released + Version 2.0.1 is the "Cox-ified" source code + Version 2.0.2 - fixed version string usage, and made ppc functions static +*/ + + +/* PARAMETERS */ +static int verbose; /* set this to 1 to see debugging messages and whatnot */ + +#define BACKPACK_VERSION "2.0.2" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/slab.h> +#include <linux/types.h> +#include <asm/io.h> + +#if defined(CONFIG_PARPORT_MODULE)||defined(CONFIG_PARPORT) +#include <linux/parport.h> +#endif + +#include "ppc6lnx.c" +#include "paride.h" + + + +#define PPCSTRUCT(pi) ((Interface *)(pi->private)) + +/****************************************************************/ +/* + ATAPI CDROM DRIVE REGISTERS +*/ +#define ATAPI_DATA 0 /* data port */ +#define ATAPI_ERROR 1 /* error register (read) */ +#define ATAPI_FEATURES 1 /* feature register (write) */ +#define ATAPI_INT_REASON 2 /* interrupt reason register */ +#define ATAPI_COUNT_LOW 4 /* byte count register (low) */ +#define ATAPI_COUNT_HIGH 5 /* byte count register (high) */ +#define ATAPI_DRIVE_SEL 6 /* drive select register */ +#define ATAPI_STATUS 7 /* status port (read) */ +#define ATAPI_COMMAND 7 /* command port (write) */ +#define ATAPI_ALT_STATUS 0x0e /* alternate status reg (read) */ +#define ATAPI_DEVICE_CONTROL 0x0e /* device control (write) */ +/****************************************************************/ + +static int bpck6_read_regr(PIA *pi, int cont, int reg) +{ + unsigned int out; + + /* check for bad settings */ + if (reg<0 || reg>7 || cont<0 || cont>2) + { + return(-1); + } + out=ppc6_rd_port(PPCSTRUCT(pi),cont?reg|8:reg); + return(out); +} + +static void bpck6_write_regr(PIA *pi, int cont, int reg, int val) +{ + /* check for bad settings */ + if (reg>=0 && reg<=7 && cont>=0 && cont<=1) + { + ppc6_wr_port(PPCSTRUCT(pi),cont?reg|8:reg,(u8)val); + } +} + +static void bpck6_write_block( PIA *pi, char * buf, int len ) +{ + ppc6_wr_port16_blk(PPCSTRUCT(pi),ATAPI_DATA,buf,(u32)len>>1); +} + +static void bpck6_read_block( PIA *pi, char * buf, int len ) +{ + ppc6_rd_port16_blk(PPCSTRUCT(pi),ATAPI_DATA,buf,(u32)len>>1); +} + +static void bpck6_connect ( PIA *pi ) +{ + if(verbose) + { + printk(KERN_DEBUG "connect\n"); + } + + if(pi->mode >=2) + { + PPCSTRUCT(pi)->mode=4+pi->mode-2; + } + else if(pi->mode==1) + { + PPCSTRUCT(pi)->mode=3; + } + else + { + PPCSTRUCT(pi)->mode=1; + } + + ppc6_open(PPCSTRUCT(pi)); + ppc6_wr_extout(PPCSTRUCT(pi),0x3); +} + +static void bpck6_disconnect ( PIA *pi ) +{ + if(verbose) + { + printk("disconnect\n"); + } + ppc6_wr_extout(PPCSTRUCT(pi),0x0); + ppc6_close(PPCSTRUCT(pi)); +} + +static int bpck6_test_port ( PIA *pi ) /* check for 8-bit port */ +{ + if(verbose) + { + printk(KERN_DEBUG "PARPORT indicates modes=%x for lp=0x%lx\n", + ((struct pardevice*)(pi->pardev))->port->modes, + ((struct pardevice *)(pi->pardev))->port->base); + } + + /*copy over duplicate stuff.. initialize state info*/ + PPCSTRUCT(pi)->ppc_id=pi->unit; + PPCSTRUCT(pi)->lpt_addr=pi->port; + +#ifdef CONFIG_PARPORT_PC_MODULE +#define CONFIG_PARPORT_PC +#endif + +#ifdef CONFIG_PARPORT_PC + /* look at the parport device to see if what modes we can use */ + if(((struct pardevice *)(pi->pardev))->port->modes & + (PARPORT_MODE_EPP) + ) + { + return 5; /* Can do EPP*/ + } + else if(((struct pardevice *)(pi->pardev))->port->modes & + (PARPORT_MODE_TRISTATE) + ) + { + return 2; + } + else /*Just flat SPP*/ + { + return 1; + } +#else + /* there is no way of knowing what kind of port we have + default to the highest mode possible */ + return 5; +#endif +} + +static int bpck6_probe_unit ( PIA *pi ) +{ + int out; + + if(verbose) + { + printk(KERN_DEBUG "PROBE UNIT %x on port:%x\n",pi->unit,pi->port); + } + + /*SET PPC UNIT NUMBER*/ + PPCSTRUCT(pi)->ppc_id=pi->unit; + + /*LOWER DOWN TO UNIDIRECTIONAL*/ + PPCSTRUCT(pi)->mode=1; + + out=ppc6_open(PPCSTRUCT(pi)); + + if(verbose) + { + printk(KERN_DEBUG "ppc_open returned %2x\n",out); + } + + if(out) + { + ppc6_close(PPCSTRUCT(pi)); + if(verbose) + { + printk(KERN_DEBUG "leaving probe\n"); + } + return(1); + } + else + { + if(verbose) + { + printk(KERN_DEBUG "Failed open\n"); + } + return(0); + } +} + +static void bpck6_log_adapter( PIA *pi, char * scratch, int verbose ) +{ + char *mode_string[5]= + {"4-bit","8-bit","EPP-8","EPP-16","EPP-32"}; + + printk("%s: BACKPACK Protocol Driver V"BACKPACK_VERSION"\n",pi->device); + printk("%s: Copyright 2001 by Micro Solutions, Inc., DeKalb IL.\n",pi->device); + printk("%s: BACKPACK %s, Micro Solutions BACKPACK Drive at 0x%x\n", + pi->device,BACKPACK_VERSION,pi->port); + printk("%s: Unit: %d Mode:%d (%s) Delay %d\n",pi->device, + pi->unit,pi->mode,mode_string[pi->mode],pi->delay); +} + +static int bpck6_init_proto(PIA *pi) +{ + Interface *p = kmalloc(sizeof(Interface), GFP_KERNEL); + + if (p) { + memset(p, 0, sizeof(Interface)); + pi->private = (unsigned long)p; + return 0; + } + + printk(KERN_ERR "%s: ERROR COULDN'T ALLOCATE MEMORY\n", pi->device); + return -1; +} + +static void bpck6_release_proto(PIA *pi) +{ + kfree((void *)(pi->private)); +} + +static struct pi_protocol bpck6 = { + .owner = THIS_MODULE, + .name = "bpck6", + .max_mode = 5, + .epp_first = 2, /* 2-5 use epp (need 8 ports) */ + .max_units = 255, + .write_regr = bpck6_write_regr, + .read_regr = bpck6_read_regr, + .write_block = bpck6_write_block, + .read_block = bpck6_read_block, + .connect = bpck6_connect, + .disconnect = bpck6_disconnect, + .test_port = bpck6_test_port, + .probe_unit = bpck6_probe_unit, + .log_adapter = bpck6_log_adapter, + .init_proto = bpck6_init_proto, + .release_proto = bpck6_release_proto, +}; + +static int __init bpck6_init(void) +{ + printk(KERN_INFO "bpck6: BACKPACK Protocol Driver V"BACKPACK_VERSION"\n"); + printk(KERN_INFO "bpck6: Copyright 2001 by Micro Solutions, Inc., DeKalb IL. USA\n"); + if(verbose) + printk(KERN_DEBUG "bpck6: verbose debug enabled.\n"); + return pi_register(&bpck6) - 1; +} + +static void __exit bpck6_exit(void) +{ + pi_unregister(&bpck6); +} + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Micro Solutions Inc."); +MODULE_DESCRIPTION("BACKPACK Protocol module, compatible with PARIDE"); +module_param(verbose, bool, 0644); +module_init(bpck6_init) +module_exit(bpck6_exit) diff --git a/drivers/block/paride/comm.c b/drivers/block/paride/comm.c new file mode 100644 index 00000000000..d842956edf7 --- /dev/null +++ b/drivers/block/paride/comm.c @@ -0,0 +1,218 @@ +/* + comm.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + comm.c is a low-level protocol driver for some older models + of the DataStor "Commuter" parallel to IDE adapter. Some of + the parallel port devices marketed by Arista currently + use this adapter. +*/ + +/* Changes: + + 1.01 GRG 1998.05.05 init_proto, release_proto + +*/ + +#define COMM_VERSION "1.01" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +/* mode codes: 0 nybble reads, 8-bit writes + 1 8-bit reads and writes + 2 8-bit EPP mode +*/ + +#define j44(a,b) (((a>>3)&0x0f)|((b<<1)&0xf0)) + +#define P1 w2(5);w2(0xd);w2(0xd);w2(5);w2(4); +#define P2 w2(5);w2(7);w2(7);w2(5);w2(4); + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int cont_map[2] = { 0x08, 0x10 }; + +static int comm_read_regr( PIA *pi, int cont, int regr ) + +{ int l, h, r; + + r = regr + cont_map[cont]; + + switch (pi->mode) { + + case 0: w0(r); P1; w0(0); + w2(6); l = r1(); w0(0x80); h = r1(); w2(4); + return j44(l,h); + + case 1: w0(r+0x20); P1; + w0(0); w2(0x26); h = r0(); w2(4); + return h; + + case 2: + case 3: + case 4: w3(r+0x20); r1(); + w2(0x24); h = r4(); w2(4); + return h; + + } + return -1; +} + +static void comm_write_regr( PIA *pi, int cont, int regr, int val ) + +{ int r; + + r = regr + cont_map[cont]; + + switch (pi->mode) { + + case 0: + case 1: w0(r); P1; w0(val); P2; + break; + + case 2: + case 3: + case 4: w3(r); r1(); w4(val); + break; + } +} + +static void comm_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(4); w0(0xff); w2(6); + w2(4); w0(0xaa); w2(6); + w2(4); w0(0x00); w2(6); + w2(4); w0(0x87); w2(6); + w2(4); w0(0xe0); w2(0xc); w2(0xc); w2(4); +} + +static void comm_disconnect ( PIA *pi ) + +{ w2(0); w2(0); w2(0); w2(4); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void comm_read_block( PIA *pi, char * buf, int count ) + +{ int i, l, h; + + switch (pi->mode) { + + case 0: w0(0x48); P1; + for(i=0;i<count;i++) { + w0(0); w2(6); l = r1(); + w0(0x80); h = r1(); w2(4); + buf[i] = j44(l,h); + } + break; + + case 1: w0(0x68); P1; w0(0); + for(i=0;i<count;i++) { + w2(0x26); buf[i] = r0(); w2(0x24); + } + w2(4); + break; + + case 2: w3(0x68); r1(); w2(0x24); + for (i=0;i<count;i++) buf[i] = r4(); + w2(4); + break; + + case 3: w3(0x68); r1(); w2(0x24); + for (i=0;i<count/2;i++) ((u16 *)buf)[i] = r4w(); + w2(4); + break; + + case 4: w3(0x68); r1(); w2(0x24); + for (i=0;i<count/4;i++) ((u32 *)buf)[i] = r4l(); + w2(4); + break; + + } +} + +/* NB: Watch out for the byte swapped writes ! */ + +static void comm_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + switch (pi->mode) { + + case 0: + case 1: w0(0x68); P1; + for (k=0;k<count;k++) { + w2(5); w0(buf[k^1]); w2(7); + } + w2(5); w2(4); + break; + + case 2: w3(0x48); r1(); + for (k=0;k<count;k++) w4(buf[k^1]); + break; + + case 3: w3(0x48); r1(); + for (k=0;k<count/2;k++) w4w(pi_swab16(buf,k)); + break; + + case 4: w3(0x48); r1(); + for (k=0;k<count/4;k++) w4l(pi_swab32(buf,k)); + break; + + + } +} + +static void comm_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[5] = {"4-bit","8-bit","EPP-8","EPP-16","EPP-32"}; + + printk("%s: comm %s, DataStor Commuter at 0x%x, ", + pi->device,COMM_VERSION,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol comm = { + .owner = THIS_MODULE, + .name = "comm", + .max_mode = 5, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = comm_write_regr, + .read_regr = comm_read_regr, + .write_block = comm_write_block, + .read_block = comm_read_block, + .connect = comm_connect, + .disconnect = comm_disconnect, + .log_adapter = comm_log_adapter, +}; + +static int __init comm_init(void) +{ + return pi_register(&comm)-1; +} + +static void __exit comm_exit(void) +{ + pi_unregister(&comm); +} + +MODULE_LICENSE("GPL"); +module_init(comm_init) +module_exit(comm_exit) diff --git a/drivers/block/paride/dstr.c b/drivers/block/paride/dstr.c new file mode 100644 index 00000000000..04d53bf58e8 --- /dev/null +++ b/drivers/block/paride/dstr.c @@ -0,0 +1,233 @@ +/* + dstr.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + dstr.c is a low-level protocol driver for the + DataStor EP2000 parallel to IDE adapter chip. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 init_proto, release_proto + +*/ + +#define DSTR_VERSION "1.01" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +/* mode codes: 0 nybble reads, 8-bit writes + 1 8-bit reads and writes + 2 8-bit EPP mode + 3 EPP-16 + 4 EPP-32 +*/ + +#define j44(a,b) (((a>>3)&0x07)|((~a>>4)&0x08)|((b<<1)&0x70)|((~b)&0x80)) + +#define P1 w2(5);w2(0xd);w2(5);w2(4); +#define P2 w2(5);w2(7);w2(5);w2(4); +#define P3 w2(6);w2(4);w2(6);w2(4); + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int cont_map[2] = { 0x20, 0x40 }; + +static int dstr_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, r; + + r = regr + cont_map[cont]; + + w0(0x81); P1; + if (pi->mode) { w0(0x11); } else { w0(1); } + P2; w0(r); P1; + + switch (pi->mode) { + + case 0: w2(6); a = r1(); w2(4); w2(6); b = r1(); w2(4); + return j44(a,b); + + case 1: w0(0); w2(0x26); a = r0(); w2(4); + return a; + + case 2: + case 3: + case 4: w2(0x24); a = r4(); w2(4); + return a; + + } + return -1; +} + +static void dstr_write_regr( PIA *pi, int cont, int regr, int val ) + +{ int r; + + r = regr + cont_map[cont]; + + w0(0x81); P1; + if (pi->mode >= 2) { w0(0x11); } else { w0(1); } + P2; w0(r); P1; + + switch (pi->mode) { + + case 0: + case 1: w0(val); w2(5); w2(7); w2(5); w2(4); + break; + + case 2: + case 3: + case 4: w4(val); + break; + } +} + +#define CCP(x) w0(0xff);w2(0xc);w2(4);\ + w0(0xaa);w0(0x55);w0(0);w0(0xff);w0(0x87);w0(0x78);\ + w0(x);w2(5);w2(4); + +static void dstr_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(4); CCP(0xe0); w0(0xff); +} + +static void dstr_disconnect ( PIA *pi ) + +{ CCP(0x30); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void dstr_read_block( PIA *pi, char * buf, int count ) + +{ int k, a, b; + + w0(0x81); P1; + if (pi->mode) { w0(0x19); } else { w0(9); } + P2; w0(0x82); P1; P3; w0(0x20); P1; + + switch (pi->mode) { + + case 0: for (k=0;k<count;k++) { + w2(6); a = r1(); w2(4); + w2(6); b = r1(); w2(4); + buf[k] = j44(a,b); + } + break; + + case 1: w0(0); + for (k=0;k<count;k++) { + w2(0x26); buf[k] = r0(); w2(0x24); + } + w2(4); + break; + + case 2: w2(0x24); + for (k=0;k<count;k++) buf[k] = r4(); + w2(4); + break; + + case 3: w2(0x24); + for (k=0;k<count/2;k++) ((u16 *)buf)[k] = r4w(); + w2(4); + break; + + case 4: w2(0x24); + for (k=0;k<count/4;k++) ((u32 *)buf)[k] = r4l(); + w2(4); + break; + + } +} + +static void dstr_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + w0(0x81); P1; + if (pi->mode) { w0(0x19); } else { w0(9); } + P2; w0(0x82); P1; P3; w0(0x20); P1; + + switch (pi->mode) { + + case 0: + case 1: for (k=0;k<count;k++) { + w2(5); w0(buf[k]); w2(7); + } + w2(5); w2(4); + break; + + case 2: w2(0xc5); + for (k=0;k<count;k++) w4(buf[k]); + w2(0xc4); + break; + + case 3: w2(0xc5); + for (k=0;k<count/2;k++) w4w(((u16 *)buf)[k]); + w2(0xc4); + break; + + case 4: w2(0xc5); + for (k=0;k<count/4;k++) w4l(((u32 *)buf)[k]); + w2(0xc4); + break; + + } +} + + +static void dstr_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[5] = {"4-bit","8-bit","EPP-8", + "EPP-16","EPP-32"}; + + printk("%s: dstr %s, DataStor EP2000 at 0x%x, ", + pi->device,DSTR_VERSION,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol dstr = { + .owner = THIS_MODULE, + .name = "dstr", + .max_mode = 5, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = dstr_write_regr, + .read_regr = dstr_read_regr, + .write_block = dstr_write_block, + .read_block = dstr_read_block, + .connect = dstr_connect, + .disconnect = dstr_disconnect, + .log_adapter = dstr_log_adapter, +}; + +static int __init dstr_init(void) +{ + return pi_register(&dstr)-1; +} + +static void __exit dstr_exit(void) +{ + pi_unregister(&dstr); +} + +MODULE_LICENSE("GPL"); +module_init(dstr_init) +module_exit(dstr_exit) diff --git a/drivers/block/paride/epat.c b/drivers/block/paride/epat.c new file mode 100644 index 00000000000..55d1c0a1fb9 --- /dev/null +++ b/drivers/block/paride/epat.c @@ -0,0 +1,340 @@ +/* + epat.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is the low level protocol driver for the EPAT parallel + to IDE adapter from Shuttle Technologies. This adapter is + used in many popular parallel port disk products such as the + SyQuest EZ drives, the Avatar Shark and the Imation SuperDisk. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 init_proto, release_proto + 1.02 Joshua b. Jore CPP(renamed), epat_connect, epat_disconnect + +*/ + +#define EPAT_VERSION "1.02" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define j44(a,b) (((a>>4)&0x0f)+(b&0xf0)) +#define j53(a,b) (((a>>3)&0x1f)+((b<<4)&0xe0)) + +static int epatc8; + +module_param(epatc8, int, 0); +MODULE_PARM_DESC(epatc8, "support for the Shuttle EP1284 chip, " + "used in any recent Imation SuperDisk (LS-120) drive."); + +/* cont = 0 IDE register file + cont = 1 IDE control registers + cont = 2 internal EPAT registers +*/ + +static int cont_map[3] = { 0x18, 0x10, 0 }; + +static void epat_write_regr( PIA *pi, int cont, int regr, int val) + +{ int r; + + r = regr + cont_map[cont]; + + switch (pi->mode) { + + case 0: + case 1: + case 2: w0(0x60+r); w2(1); w0(val); w2(4); + break; + + case 3: + case 4: + case 5: w3(0x40+r); w4(val); + break; + + } +} + +static int epat_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, r; + + r = regr + cont_map[cont]; + + switch (pi->mode) { + + case 0: w0(r); w2(1); w2(3); + a = r1(); w2(4); b = r1(); + return j44(a,b); + + case 1: w0(0x40+r); w2(1); w2(4); + a = r1(); b = r2(); w0(0xff); + return j53(a,b); + + case 2: w0(0x20+r); w2(1); w2(0x25); + a = r0(); w2(4); + return a; + + case 3: + case 4: + case 5: w3(r); w2(0x24); a = r4(); w2(4); + return a; + + } + return -1; /* never gets here */ +} + +static void epat_read_block( PIA *pi, char * buf, int count ) + +{ int k, ph, a, b; + + switch (pi->mode) { + + case 0: w0(7); w2(1); w2(3); w0(0xff); + ph = 0; + for(k=0;k<count;k++) { + if (k == count-1) w0(0xfd); + w2(6+ph); a = r1(); + if (a & 8) b = a; + else { w2(4+ph); b = r1(); } + buf[k] = j44(a,b); + ph = 1 - ph; + } + w0(0); w2(4); + break; + + case 1: w0(0x47); w2(1); w2(5); w0(0xff); + ph = 0; + for(k=0;k<count;k++) { + if (k == count-1) w0(0xfd); + w2(4+ph); + a = r1(); b = r2(); + buf[k] = j53(a,b); + ph = 1 - ph; + } + w0(0); w2(4); + break; + + case 2: w0(0x27); w2(1); w2(0x25); w0(0); + ph = 0; + for(k=0;k<count-1;k++) { + w2(0x24+ph); + buf[k] = r0(); + ph = 1 - ph; + } + w2(0x26); w2(0x27); buf[count-1] = r0(); + w2(0x25); w2(4); + break; + + case 3: w3(0x80); w2(0x24); + for(k=0;k<count-1;k++) buf[k] = r4(); + w2(4); w3(0xa0); w2(0x24); buf[count-1] = r4(); + w2(4); + break; + + case 4: w3(0x80); w2(0x24); + for(k=0;k<(count/2)-1;k++) ((u16 *)buf)[k] = r4w(); + buf[count-2] = r4(); + w2(4); w3(0xa0); w2(0x24); buf[count-1] = r4(); + w2(4); + break; + + case 5: w3(0x80); w2(0x24); + for(k=0;k<(count/4)-1;k++) ((u32 *)buf)[k] = r4l(); + for(k=count-4;k<count-1;k++) buf[k] = r4(); + w2(4); w3(0xa0); w2(0x24); buf[count-1] = r4(); + w2(4); + break; + + } +} + +static void epat_write_block( PIA *pi, char * buf, int count ) + +{ int ph, k; + + switch (pi->mode) { + + case 0: + case 1: + case 2: w0(0x67); w2(1); w2(5); + ph = 0; + for(k=0;k<count;k++) { + w0(buf[k]); + w2(4+ph); + ph = 1 - ph; + } + w2(7); w2(4); + break; + + case 3: w3(0xc0); + for(k=0;k<count;k++) w4(buf[k]); + w2(4); + break; + + case 4: w3(0xc0); + for(k=0;k<(count/2);k++) w4w(((u16 *)buf)[k]); + w2(4); + break; + + case 5: w3(0xc0); + for(k=0;k<(count/4);k++) w4l(((u32 *)buf)[k]); + w2(4); + break; + + } +} + +/* these macros access the EPAT registers in native addressing */ + +#define WR(r,v) epat_write_regr(pi,2,r,v) +#define RR(r) (epat_read_regr(pi,2,r)) + +/* and these access the IDE task file */ + +#define WRi(r,v) epat_write_regr(pi,0,r,v) +#define RRi(r) (epat_read_regr(pi,0,r)) + +/* FIXME: the CPP stuff should be fixed to handle multiple EPATs on a chain */ + +#define CPP(x) w2(4);w0(0x22);w0(0xaa);w0(0x55);w0(0);w0(0xff);\ + w0(0x87);w0(0x78);w0(x);w2(4);w2(5);w2(4);w0(0xff); + +static void epat_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + + /* Initialize the chip */ + CPP(0); + + if (epatc8) { + CPP(0x40);CPP(0xe0); + w0(0);w2(1);w2(4); + WR(0x8,0x12);WR(0xc,0x14);WR(0x12,0x10); + WR(0xe,0xf);WR(0xf,4); + /* WR(0xe,0xa);WR(0xf,4); */ + WR(0xe,0xd);WR(0xf,0); + /* CPP(0x30); */ + } + + /* Connect to the chip */ + CPP(0xe0); + w0(0);w2(1);w2(4); /* Idle into SPP */ + if (pi->mode >= 3) { + w0(0);w2(1);w2(4);w2(0xc); + /* Request EPP */ + w0(0x40);w2(6);w2(7);w2(4);w2(0xc);w2(4); + } + + if (!epatc8) { + WR(8,0x10); WR(0xc,0x14); WR(0xa,0x38); WR(0x12,0x10); + } +} + +static void epat_disconnect (PIA *pi) +{ CPP(0x30); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static int epat_test_proto( PIA *pi, char * scratch, int verbose ) + +{ int k, j, f, cc; + int e[2] = {0,0}; + + epat_connect(pi); + cc = RR(0xd); + epat_disconnect(pi); + + epat_connect(pi); + for (j=0;j<2;j++) { + WRi(6,0xa0+j*0x10); + for (k=0;k<256;k++) { + WRi(2,k^0xaa); + WRi(3,k^0x55); + if (RRi(2) != (k^0xaa)) e[j]++; + } + } + epat_disconnect(pi); + + f = 0; + epat_connect(pi); + WR(0x13,1); WR(0x13,0); WR(0xa,0x11); + epat_read_block(pi,scratch,512); + + for (k=0;k<256;k++) { + if ((scratch[2*k] & 0xff) != k) f++; + if ((scratch[2*k+1] & 0xff) != (0xff-k)) f++; + } + epat_disconnect(pi); + + if (verbose) { + printk("%s: epat: port 0x%x, mode %d, ccr %x, test=(%d,%d,%d)\n", + pi->device,pi->port,pi->mode,cc,e[0],e[1],f); + } + + return (e[0] && e[1]) || f; +} + +static void epat_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ int ver; + char *mode_string[6] = + {"4-bit","5/3","8-bit","EPP-8","EPP-16","EPP-32"}; + + epat_connect(pi); + WR(0xa,0x38); /* read the version code */ + ver = RR(0xb); + epat_disconnect(pi); + + printk("%s: epat %s, Shuttle EPAT chip %x at 0x%x, ", + pi->device,EPAT_VERSION,ver,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol epat = { + .owner = THIS_MODULE, + .name = "epat", + .max_mode = 6, + .epp_first = 3, + .default_delay = 1, + .max_units = 1, + .write_regr = epat_write_regr, + .read_regr = epat_read_regr, + .write_block = epat_write_block, + .read_block = epat_read_block, + .connect = epat_connect, + .disconnect = epat_disconnect, + .test_proto = epat_test_proto, + .log_adapter = epat_log_adapter, +}; + +static int __init epat_init(void) +{ +#ifdef CONFIG_PARIDE_EPATC8 + epatc8 = 1; +#endif + return pi_register(&epat)-1; +} + +static void __exit epat_exit(void) +{ + pi_unregister(&epat); +} + +MODULE_LICENSE("GPL"); +module_init(epat_init) +module_exit(epat_exit) diff --git a/drivers/block/paride/epia.c b/drivers/block/paride/epia.c new file mode 100644 index 00000000000..0f2e0c292d8 --- /dev/null +++ b/drivers/block/paride/epia.c @@ -0,0 +1,316 @@ +/* + epia.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + epia.c is a low-level protocol driver for Shuttle Technologies + EPIA parallel to IDE adapter chip. This device is now obsolete + and has been replaced with the EPAT chip, which is supported + by epat.c, however, some devices based on EPIA are still + available. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 init_proto, release_proto + 1.02 GRG 1998.06.17 support older versions of EPIA + +*/ + +#define EPIA_VERSION "1.02" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +/* mode codes: 0 nybble reads on port 1, 8-bit writes + 1 5/3 reads on ports 1 & 2, 8-bit writes + 2 8-bit reads and writes + 3 8-bit EPP mode + 4 16-bit EPP + 5 32-bit EPP +*/ + +#define j44(a,b) (((a>>4)&0x0f)+(b&0xf0)) +#define j53(a,b) (((a>>3)&0x1f)+((b<<4)&0xe0)) + +/* cont = 0 IDE register file + cont = 1 IDE control registers +*/ + +static int cont_map[2] = { 0, 0x80 }; + +static int epia_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, r; + + regr += cont_map[cont]; + + switch (pi->mode) { + + case 0: r = regr^0x39; + w0(r); w2(1); w2(3); w0(r); + a = r1(); w2(1); b = r1(); w2(4); + return j44(a,b); + + case 1: r = regr^0x31; + w0(r); w2(1); w0(r&0x37); + w2(3); w2(5); w0(r|0xf0); + a = r1(); b = r2(); w2(4); + return j53(a,b); + + case 2: r = regr^0x29; + w0(r); w2(1); w2(0X21); w2(0x23); + a = r0(); w2(4); + return a; + + case 3: + case 4: + case 5: w3(regr); w2(0x24); a = r4(); w2(4); + return a; + + } + return -1; +} + +static void epia_write_regr( PIA *pi, int cont, int regr, int val) + +{ int r; + + regr += cont_map[cont]; + + switch (pi->mode) { + + case 0: + case 1: + case 2: r = regr^0x19; + w0(r); w2(1); w0(val); w2(3); w2(4); + break; + + case 3: + case 4: + case 5: r = regr^0x40; + w3(r); w4(val); w2(4); + break; + } +} + +#define WR(r,v) epia_write_regr(pi,0,r,v) +#define RR(r) (epia_read_regr(pi,0,r)) + +/* The use of register 0x84 is entirely unclear - it seems to control + some EPP counters ... currently we know about 3 different block + sizes: the standard 512 byte reads and writes, 12 byte writes and + 2048 byte reads (the last two being used in the CDrom drivers. +*/ + +static void epia_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + + w2(4); w0(0xa0); w0(0x50); w0(0xc0); w0(0x30); w0(0xa0); w0(0); + w2(1); w2(4); + if (pi->mode >= 3) { + w0(0xa); w2(1); w2(4); w0(0x82); w2(4); w2(0xc); w2(4); + w2(0x24); w2(0x26); w2(4); + } + WR(0x86,8); +} + +static void epia_disconnect ( PIA *pi ) + +{ /* WR(0x84,0x10); */ + w0(pi->saved_r0); + w2(1); w2(4); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void epia_read_block( PIA *pi, char * buf, int count ) + +{ int k, ph, a, b; + + switch (pi->mode) { + + case 0: w0(0x81); w2(1); w2(3); w0(0xc1); + ph = 1; + for (k=0;k<count;k++) { + w2(2+ph); a = r1(); + w2(4+ph); b = r1(); + buf[k] = j44(a,b); + ph = 1 - ph; + } + w0(0); w2(4); + break; + + case 1: w0(0x91); w2(1); w0(0x10); w2(3); + w0(0x51); w2(5); w0(0xd1); + ph = 1; + for (k=0;k<count;k++) { + w2(4+ph); + a = r1(); b = r2(); + buf[k] = j53(a,b); + ph = 1 - ph; + } + w0(0); w2(4); + break; + + case 2: w0(0x89); w2(1); w2(0x23); w2(0x21); + ph = 1; + for (k=0;k<count;k++) { + w2(0x24+ph); + buf[k] = r0(); + ph = 1 - ph; + } + w2(6); w2(4); + break; + + case 3: if (count > 512) WR(0x84,3); + w3(0); w2(0x24); + for (k=0;k<count;k++) buf[k] = r4(); + w2(4); WR(0x84,0); + break; + + case 4: if (count > 512) WR(0x84,3); + w3(0); w2(0x24); + for (k=0;k<count/2;k++) ((u16 *)buf)[k] = r4w(); + w2(4); WR(0x84,0); + break; + + case 5: if (count > 512) WR(0x84,3); + w3(0); w2(0x24); + for (k=0;k<count/4;k++) ((u32 *)buf)[k] = r4l(); + w2(4); WR(0x84,0); + break; + + } +} + +static void epia_write_block( PIA *pi, char * buf, int count ) + +{ int ph, k, last, d; + + switch (pi->mode) { + + case 0: + case 1: + case 2: w0(0xa1); w2(1); w2(3); w2(1); w2(5); + ph = 0; last = 0x8000; + for (k=0;k<count;k++) { + d = buf[k]; + if (d != last) { last = d; w0(d); } + w2(4+ph); + ph = 1 - ph; + } + w2(7); w2(4); + break; + + case 3: if (count < 512) WR(0x84,1); + w3(0x40); + for (k=0;k<count;k++) w4(buf[k]); + if (count < 512) WR(0x84,0); + break; + + case 4: if (count < 512) WR(0x84,1); + w3(0x40); + for (k=0;k<count/2;k++) w4w(((u16 *)buf)[k]); + if (count < 512) WR(0x84,0); + break; + + case 5: if (count < 512) WR(0x84,1); + w3(0x40); + for (k=0;k<count/4;k++) w4l(((u32 *)buf)[k]); + if (count < 512) WR(0x84,0); + break; + + } + +} + +static int epia_test_proto( PIA *pi, char * scratch, int verbose ) + +{ int j, k, f; + int e[2] = {0,0}; + + epia_connect(pi); + for (j=0;j<2;j++) { + WR(6,0xa0+j*0x10); + for (k=0;k<256;k++) { + WR(2,k^0xaa); + WR(3,k^0x55); + if (RR(2) != (k^0xaa)) e[j]++; + } + WR(2,1); WR(3,1); + } + epia_disconnect(pi); + + f = 0; + epia_connect(pi); + WR(0x84,8); + epia_read_block(pi,scratch,512); + for (k=0;k<256;k++) { + if ((scratch[2*k] & 0xff) != ((k+1) & 0xff)) f++; + if ((scratch[2*k+1] & 0xff) != ((-2-k) & 0xff)) f++; + } + WR(0x84,0); + epia_disconnect(pi); + + if (verbose) { + printk("%s: epia: port 0x%x, mode %d, test=(%d,%d,%d)\n", + pi->device,pi->port,pi->mode,e[0],e[1],f); + } + + return (e[0] && e[1]) || f; + +} + + +static void epia_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[6] = {"4-bit","5/3","8-bit", + "EPP-8","EPP-16","EPP-32"}; + + printk("%s: epia %s, Shuttle EPIA at 0x%x, ", + pi->device,EPIA_VERSION,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol epia = { + .owner = THIS_MODULE, + .name = "epia", + .max_mode = 6, + .epp_first = 3, + .default_delay = 1, + .max_units = 1, + .write_regr = epia_write_regr, + .read_regr = epia_read_regr, + .write_block = epia_write_block, + .read_block = epia_read_block, + .connect = epia_connect, + .disconnect = epia_disconnect, + .test_proto = epia_test_proto, + .log_adapter = epia_log_adapter, +}; + +static int __init epia_init(void) +{ + return pi_register(&epia)-1; +} + +static void __exit epia_exit(void) +{ + pi_unregister(&epia); +} + +MODULE_LICENSE("GPL"); +module_init(epia_init) +module_exit(epia_exit) diff --git a/drivers/block/paride/fit2.c b/drivers/block/paride/fit2.c new file mode 100644 index 00000000000..e0f0691d8bc --- /dev/null +++ b/drivers/block/paride/fit2.c @@ -0,0 +1,151 @@ +/* + fit2.c (c) 1998 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + fit2.c is a low-level protocol driver for the older version + of the Fidelity International Technology parallel port adapter. + This adapter is used in their TransDisk 2000 and older TransDisk + 3000 portable hard-drives. As far as I can tell, this device + supports 4-bit mode _only_. + + Newer models of the FIT products use an enhanced protocol. + The "fit3" protocol module should support current drives. + +*/ + +#define FIT2_VERSION "1.0" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define j44(a,b) (((a>>4)&0x0f)|(b&0xf0)) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set + +NB: The FIT adapter does not appear to use the control registers. +So, we map ALT_STATUS to STATUS and NO-OP writes to the device +control register - this means that IDE reset will not work on these +devices. + +*/ + +static void fit2_write_regr( PIA *pi, int cont, int regr, int val) + +{ if (cont == 1) return; + w2(0xc); w0(regr); w2(4); w0(val); w2(5); w0(0); w2(4); +} + +static int fit2_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, r; + + if (cont) { + if (regr != 6) return 0xff; + r = 7; + } else r = regr + 0x10; + + w2(0xc); w0(r); w2(4); w2(5); + w0(0); a = r1(); + w0(1); b = r1(); + w2(4); + + return j44(a,b); + +} + +static void fit2_read_block( PIA *pi, char * buf, int count ) + +{ int k, a, b, c, d; + + w2(0xc); w0(0x10); + + for (k=0;k<count/4;k++) { + + w2(4); w2(5); + w0(0); a = r1(); w0(1); b = r1(); + w0(3); c = r1(); w0(2); d = r1(); + buf[4*k+0] = j44(a,b); + buf[4*k+1] = j44(d,c); + + w2(4); w2(5); + a = r1(); w0(3); b = r1(); + w0(1); c = r1(); w0(0); d = r1(); + buf[4*k+2] = j44(d,c); + buf[4*k+3] = j44(a,b); + + } + + w2(4); + +} + +static void fit2_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + + w2(0xc); w0(0); + for (k=0;k<count/2;k++) { + w2(4); w0(buf[2*k]); + w2(5); w0(buf[2*k+1]); + } + w2(4); +} + +static void fit2_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(0xcc); +} + +static void fit2_disconnect ( PIA *pi ) + +{ w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void fit2_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ printk("%s: fit2 %s, FIT 2000 adapter at 0x%x, delay %d\n", + pi->device,FIT2_VERSION,pi->port,pi->delay); + +} + +static struct pi_protocol fit2 = { + .owner = THIS_MODULE, + .name = "fit2", + .max_mode = 1, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = fit2_write_regr, + .read_regr = fit2_read_regr, + .write_block = fit2_write_block, + .read_block = fit2_read_block, + .connect = fit2_connect, + .disconnect = fit2_disconnect, + .log_adapter = fit2_log_adapter, +}; + +static int __init fit2_init(void) +{ + return pi_register(&fit2)-1; +} + +static void __exit fit2_exit(void) +{ + pi_unregister(&fit2); +} + +MODULE_LICENSE("GPL"); +module_init(fit2_init) +module_exit(fit2_exit) diff --git a/drivers/block/paride/fit3.c b/drivers/block/paride/fit3.c new file mode 100644 index 00000000000..15400e7bc66 --- /dev/null +++ b/drivers/block/paride/fit3.c @@ -0,0 +1,211 @@ +/* + fit3.c (c) 1998 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + fit3.c is a low-level protocol driver for newer models + of the Fidelity International Technology parallel port adapter. + This adapter is used in their TransDisk 3000 portable + hard-drives, as well as CD-ROM, PD-CD and other devices. + + The TD-2000 and certain older devices use a different protocol. + Try the fit2 protocol module with them. + + NB: The FIT adapters do not appear to support the control + registers. So, we map ALT_STATUS to STATUS and NO-OP writes + to the device control register - this means that IDE reset + will not work on these devices. + +*/ + +#define FIT3_VERSION "1.0" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define j44(a,b) (((a>>3)&0x0f)|((b<<1)&0xf0)) + +#define w7(byte) {out_p(7,byte);} +#define r7() (in_p(7) & 0xff) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set + +*/ + +static void fit3_write_regr( PIA *pi, int cont, int regr, int val) + +{ if (cont == 1) return; + + switch (pi->mode) { + + case 0: + case 1: w2(0xc); w0(regr); w2(0x8); w2(0xc); + w0(val); w2(0xd); + w0(0); w2(0xc); + break; + + case 2: w2(0xc); w0(regr); w2(0x8); w2(0xc); + w4(val); w4(0); + w2(0xc); + break; + + } +} + +static int fit3_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b; + + if (cont) { + if (regr != 6) return 0xff; + regr = 7; + } + + switch (pi->mode) { + + case 0: w2(0xc); w0(regr + 0x10); w2(0x8); w2(0xc); + w2(0xd); a = r1(); + w2(0xf); b = r1(); + w2(0xc); + return j44(a,b); + + case 1: w2(0xc); w0(regr + 0x90); w2(0x8); w2(0xc); + w2(0xec); w2(0xee); w2(0xef); a = r0(); + w2(0xc); + return a; + + case 2: w2(0xc); w0(regr + 0x90); w2(0x8); w2(0xc); + w2(0xec); + a = r4(); b = r4(); + w2(0xc); + return a; + + } + return -1; + +} + +static void fit3_read_block( PIA *pi, char * buf, int count ) + +{ int k, a, b, c, d; + + switch (pi->mode) { + + case 0: w2(0xc); w0(0x10); w2(0x8); w2(0xc); + for (k=0;k<count/2;k++) { + w2(0xd); a = r1(); + w2(0xf); b = r1(); + w2(0xc); c = r1(); + w2(0xe); d = r1(); + buf[2*k ] = j44(a,b); + buf[2*k+1] = j44(c,d); + } + w2(0xc); + break; + + case 1: w2(0xc); w0(0x90); w2(0x8); w2(0xc); + w2(0xec); w2(0xee); + for (k=0;k<count/2;k++) { + w2(0xef); a = r0(); + w2(0xee); b = r0(); + buf[2*k ] = a; + buf[2*k+1] = b; + } + w2(0xec); + w2(0xc); + break; + + case 2: w2(0xc); w0(0x90); w2(0x8); w2(0xc); + w2(0xec); + for (k=0;k<count;k++) buf[k] = r4(); + w2(0xc); + break; + + } +} + +static void fit3_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + switch (pi->mode) { + + case 0: + case 1: w2(0xc); w0(0); w2(0x8); w2(0xc); + for (k=0;k<count/2;k++) { + w0(buf[2*k ]); w2(0xd); + w0(buf[2*k+1]); w2(0xc); + } + break; + + case 2: w2(0xc); w0(0); w2(0x8); w2(0xc); + for (k=0;k<count;k++) w4(buf[k]); + w2(0xc); + break; + } +} + +static void fit3_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(0xc); w0(0); w2(0xa); + if (pi->mode == 2) { + w2(0xc); w0(0x9); w2(0x8); w2(0xc); + } +} + +static void fit3_disconnect ( PIA *pi ) + +{ w2(0xc); w0(0xa); w2(0x8); w2(0xc); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void fit3_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[3] = {"4-bit","8-bit","EPP"}; + + printk("%s: fit3 %s, FIT 3000 adapter at 0x%x, " + "mode %d (%s), delay %d\n", + pi->device,FIT3_VERSION,pi->port, + pi->mode,mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol fit3 = { + .owner = THIS_MODULE, + .name = "fit3", + .max_mode = 3, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = fit3_write_regr, + .read_regr = fit3_read_regr, + .write_block = fit3_write_block, + .read_block = fit3_read_block, + .connect = fit3_connect, + .disconnect = fit3_disconnect, + .log_adapter = fit3_log_adapter, +}; + +static int __init fit3_init(void) +{ + return pi_register(&fit3)-1; +} + +static void __exit fit3_exit(void) +{ + pi_unregister(&fit3); +} + +MODULE_LICENSE("GPL"); +module_init(fit3_init) +module_exit(fit3_exit) diff --git a/drivers/block/paride/friq.c b/drivers/block/paride/friq.c new file mode 100644 index 00000000000..5ea2904d281 --- /dev/null +++ b/drivers/block/paride/friq.c @@ -0,0 +1,276 @@ +/* + friq.c (c) 1998 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License + + friq.c is a low-level protocol driver for the Freecom "IQ" + parallel port IDE adapter. Early versions of this adapter + use the 'frpw' protocol. + + Freecom uses this adapter in a battery powered external + CD-ROM drive. It is also used in LS-120 drives by + Maxell and Panasonic, and other devices. + + The battery powered drive requires software support to + control the power to the drive. This module enables the + drive power when the high level driver (pcd) is loaded + and disables it when the module is unloaded. Note, if + the friq module is built in to the kernel, the power + will never be switched off, so other means should be + used to conserve battery power. + +*/ + +/* Changes: + + 1.01 GRG 1998.12.20 Added support for soft power switch +*/ + +#define FRIQ_VERSION "1.01" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define CMD(x) w2(4);w0(0xff);w0(0xff);w0(0x73);w0(0x73);\ + w0(0xc9);w0(0xc9);w0(0x26);w0(0x26);w0(x);w0(x); + +#define j44(l,h) (((l>>4)&0x0f)|(h&0xf0)) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int cont_map[2] = { 0x08, 0x10 }; + +static int friq_read_regr( PIA *pi, int cont, int regr ) + +{ int h,l,r; + + r = regr + cont_map[cont]; + + CMD(r); + w2(6); l = r1(); + w2(4); h = r1(); + w2(4); + + return j44(l,h); + +} + +static void friq_write_regr( PIA *pi, int cont, int regr, int val) + +{ int r; + + r = regr + cont_map[cont]; + + CMD(r); + w0(val); + w2(5);w2(7);w2(5);w2(4); +} + +static void friq_read_block_int( PIA *pi, char * buf, int count, int regr ) + +{ int h, l, k, ph; + + switch(pi->mode) { + + case 0: CMD(regr); + for (k=0;k<count;k++) { + w2(6); l = r1(); + w2(4); h = r1(); + buf[k] = j44(l,h); + } + w2(4); + break; + + case 1: ph = 2; + CMD(regr+0xc0); + w0(0xff); + for (k=0;k<count;k++) { + w2(0xa4 + ph); + buf[k] = r0(); + ph = 2 - ph; + } + w2(0xac); w2(0xa4); w2(4); + break; + + case 2: CMD(regr+0x80); + for (k=0;k<count-2;k++) buf[k] = r4(); + w2(0xac); w2(0xa4); + buf[count-2] = r4(); + buf[count-1] = r4(); + w2(4); + break; + + case 3: CMD(regr+0x80); + for (k=0;k<(count/2)-1;k++) ((u16 *)buf)[k] = r4w(); + w2(0xac); w2(0xa4); + buf[count-2] = r4(); + buf[count-1] = r4(); + w2(4); + break; + + case 4: CMD(regr+0x80); + for (k=0;k<(count/4)-1;k++) ((u32 *)buf)[k] = r4l(); + buf[count-4] = r4(); + buf[count-3] = r4(); + w2(0xac); w2(0xa4); + buf[count-2] = r4(); + buf[count-1] = r4(); + w2(4); + break; + + } +} + +static void friq_read_block( PIA *pi, char * buf, int count) + +{ friq_read_block_int(pi,buf,count,0x08); +} + +static void friq_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + switch(pi->mode) { + + case 0: + case 1: CMD(8); w2(5); + for (k=0;k<count;k++) { + w0(buf[k]); + w2(7);w2(5); + } + w2(4); + break; + + case 2: CMD(0xc8); w2(5); + for (k=0;k<count;k++) w4(buf[k]); + w2(4); + break; + + case 3: CMD(0xc8); w2(5); + for (k=0;k<count/2;k++) w4w(((u16 *)buf)[k]); + w2(4); + break; + + case 4: CMD(0xc8); w2(5); + for (k=0;k<count/4;k++) w4l(((u32 *)buf)[k]); + w2(4); + break; + } +} + +static void friq_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(4); +} + +static void friq_disconnect ( PIA *pi ) + +{ CMD(0x20); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static int friq_test_proto( PIA *pi, char * scratch, int verbose ) + +{ int j, k, r; + int e[2] = {0,0}; + + pi->saved_r0 = r0(); + w0(0xff); udelay(20); CMD(0x3d); /* turn the power on */ + udelay(500); + w0(pi->saved_r0); + + friq_connect(pi); + for (j=0;j<2;j++) { + friq_write_regr(pi,0,6,0xa0+j*0x10); + for (k=0;k<256;k++) { + friq_write_regr(pi,0,2,k^0xaa); + friq_write_regr(pi,0,3,k^0x55); + if (friq_read_regr(pi,0,2) != (k^0xaa)) e[j]++; + } + } + friq_disconnect(pi); + + friq_connect(pi); + friq_read_block_int(pi,scratch,512,0x10); + r = 0; + for (k=0;k<128;k++) if (scratch[k] != k) r++; + friq_disconnect(pi); + + if (verbose) { + printk("%s: friq: port 0x%x, mode %d, test=(%d,%d,%d)\n", + pi->device,pi->port,pi->mode,e[0],e[1],r); + } + + return (r || (e[0] && e[1])); +} + + +static void friq_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[6] = {"4-bit","8-bit", + "EPP-8","EPP-16","EPP-32"}; + + printk("%s: friq %s, Freecom IQ ASIC-2 adapter at 0x%x, ", pi->device, + FRIQ_VERSION,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + + pi->private = 1; + friq_connect(pi); + CMD(0x9e); /* disable sleep timer */ + friq_disconnect(pi); + +} + +static void friq_release_proto( PIA *pi) +{ + if (pi->private) { /* turn off the power */ + friq_connect(pi); + CMD(0x1d); CMD(0x1e); + friq_disconnect(pi); + pi->private = 0; + } +} + +static struct pi_protocol friq = { + .owner = THIS_MODULE, + .name = "friq", + .max_mode = 5, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = friq_write_regr, + .read_regr = friq_read_regr, + .write_block = friq_write_block, + .read_block = friq_read_block, + .connect = friq_connect, + .disconnect = friq_disconnect, + .test_proto = friq_test_proto, + .log_adapter = friq_log_adapter, + .release_proto = friq_release_proto, +}; + +static int __init friq_init(void) +{ + return pi_register(&friq)-1; +} + +static void __exit friq_exit(void) +{ + pi_unregister(&friq); +} + +MODULE_LICENSE("GPL"); +module_init(friq_init) +module_exit(friq_exit) diff --git a/drivers/block/paride/frpw.c b/drivers/block/paride/frpw.c new file mode 100644 index 00000000000..56b3824b153 --- /dev/null +++ b/drivers/block/paride/frpw.c @@ -0,0 +1,313 @@ +/* + frpw.c (c) 1996-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License + + frpw.c is a low-level protocol driver for the Freecom "Power" + parallel port IDE adapter. + + Some applications of this adapter may require a "printer" reset + prior to loading the driver. This can be done by loading and + unloading the "lp" driver, or it can be done by this driver + if you define FRPW_HARD_RESET. The latter is not recommended + as it may upset devices on other ports. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 init_proto, release_proto + fix chip detect + added EPP-16 and EPP-32 + 1.02 GRG 1998.09.23 added hard reset to initialisation process + 1.03 GRG 1998.12.14 made hard reset conditional + +*/ + +#define FRPW_VERSION "1.03" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define cec4 w2(0xc);w2(0xe);w2(0xe);w2(0xc);w2(4);w2(4);w2(4); +#define j44(l,h) (((l>>4)&0x0f)|(h&0xf0)) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int cont_map[2] = { 0x08, 0x10 }; + +static int frpw_read_regr( PIA *pi, int cont, int regr ) + +{ int h,l,r; + + r = regr + cont_map[cont]; + + w2(4); + w0(r); cec4; + w2(6); l = r1(); + w2(4); h = r1(); + w2(4); + + return j44(l,h); + +} + +static void frpw_write_regr( PIA *pi, int cont, int regr, int val) + +{ int r; + + r = regr + cont_map[cont]; + + w2(4); w0(r); cec4; + w0(val); + w2(5);w2(7);w2(5);w2(4); +} + +static void frpw_read_block_int( PIA *pi, char * buf, int count, int regr ) + +{ int h, l, k, ph; + + switch(pi->mode) { + + case 0: w2(4); w0(regr); cec4; + for (k=0;k<count;k++) { + w2(6); l = r1(); + w2(4); h = r1(); + buf[k] = j44(l,h); + } + w2(4); + break; + + case 1: ph = 2; + w2(4); w0(regr + 0xc0); cec4; + w0(0xff); + for (k=0;k<count;k++) { + w2(0xa4 + ph); + buf[k] = r0(); + ph = 2 - ph; + } + w2(0xac); w2(0xa4); w2(4); + break; + + case 2: w2(4); w0(regr + 0x80); cec4; + for (k=0;k<count;k++) buf[k] = r4(); + w2(0xac); w2(0xa4); + w2(4); + break; + + case 3: w2(4); w0(regr + 0x80); cec4; + for (k=0;k<count-2;k++) buf[k] = r4(); + w2(0xac); w2(0xa4); + buf[count-2] = r4(); + buf[count-1] = r4(); + w2(4); + break; + + case 4: w2(4); w0(regr + 0x80); cec4; + for (k=0;k<(count/2)-1;k++) ((u16 *)buf)[k] = r4w(); + w2(0xac); w2(0xa4); + buf[count-2] = r4(); + buf[count-1] = r4(); + w2(4); + break; + + case 5: w2(4); w0(regr + 0x80); cec4; + for (k=0;k<(count/4)-1;k++) ((u32 *)buf)[k] = r4l(); + buf[count-4] = r4(); + buf[count-3] = r4(); + w2(0xac); w2(0xa4); + buf[count-2] = r4(); + buf[count-1] = r4(); + w2(4); + break; + + } +} + +static void frpw_read_block( PIA *pi, char * buf, int count) + +{ frpw_read_block_int(pi,buf,count,0x08); +} + +static void frpw_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + switch(pi->mode) { + + case 0: + case 1: + case 2: w2(4); w0(8); cec4; w2(5); + for (k=0;k<count;k++) { + w0(buf[k]); + w2(7);w2(5); + } + w2(4); + break; + + case 3: w2(4); w0(0xc8); cec4; w2(5); + for (k=0;k<count;k++) w4(buf[k]); + w2(4); + break; + + case 4: w2(4); w0(0xc8); cec4; w2(5); + for (k=0;k<count/2;k++) w4w(((u16 *)buf)[k]); + w2(4); + break; + + case 5: w2(4); w0(0xc8); cec4; w2(5); + for (k=0;k<count/4;k++) w4l(((u32 *)buf)[k]); + w2(4); + break; + } +} + +static void frpw_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(4); +} + +static void frpw_disconnect ( PIA *pi ) + +{ w2(4); w0(0x20); cec4; + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +/* Stub logic to see if PNP string is available - used to distinguish + between the Xilinx and ASIC implementations of the Freecom adapter. +*/ + +static int frpw_test_pnp ( PIA *pi ) + +/* returns chip_type: 0 = Xilinx, 1 = ASIC */ + +{ int olddelay, a, b; + +#ifdef FRPW_HARD_RESET + w0(0); w2(8); udelay(50); w2(0xc); /* parallel bus reset */ + mdelay(1500); +#endif + + olddelay = pi->delay; + pi->delay = 10; + + pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + + w2(4); w0(4); w2(6); w2(7); + a = r1() & 0xff; w2(4); b = r1() & 0xff; + w2(0xc); w2(0xe); w2(4); + + pi->delay = olddelay; + w0(pi->saved_r0); + w2(pi->saved_r2); + + return ((~a&0x40) && (b&0x40)); +} + +/* We use the pi->private to remember the result of the PNP test. + To make this work, private = port*2 + chip. Yes, I know it's + a hack :-( +*/ + +static int frpw_test_proto( PIA *pi, char * scratch, int verbose ) + +{ int j, k, r; + int e[2] = {0,0}; + + if ((pi->private>>1) != pi->port) + pi->private = frpw_test_pnp(pi) + 2*pi->port; + + if (((pi->private%2) == 0) && (pi->mode > 2)) { + if (verbose) + printk("%s: frpw: Xilinx does not support mode %d\n", + pi->device, pi->mode); + return 1; + } + + if (((pi->private%2) == 1) && (pi->mode == 2)) { + if (verbose) + printk("%s: frpw: ASIC does not support mode 2\n", + pi->device); + return 1; + } + + frpw_connect(pi); + for (j=0;j<2;j++) { + frpw_write_regr(pi,0,6,0xa0+j*0x10); + for (k=0;k<256;k++) { + frpw_write_regr(pi,0,2,k^0xaa); + frpw_write_regr(pi,0,3,k^0x55); + if (frpw_read_regr(pi,0,2) != (k^0xaa)) e[j]++; + } + } + frpw_disconnect(pi); + + frpw_connect(pi); + frpw_read_block_int(pi,scratch,512,0x10); + r = 0; + for (k=0;k<128;k++) if (scratch[k] != k) r++; + frpw_disconnect(pi); + + if (verbose) { + printk("%s: frpw: port 0x%x, chip %ld, mode %d, test=(%d,%d,%d)\n", + pi->device,pi->port,(pi->private%2),pi->mode,e[0],e[1],r); + } + + return (r || (e[0] && e[1])); +} + + +static void frpw_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[6] = {"4-bit","8-bit","EPP", + "EPP-8","EPP-16","EPP-32"}; + + printk("%s: frpw %s, Freecom (%s) adapter at 0x%x, ", pi->device, + FRPW_VERSION,((pi->private%2) == 0)?"Xilinx":"ASIC",pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol frpw = { + .owner = THIS_MODULE, + .name = "frpw", + .max_mode = 6, + .epp_first = 2, + .default_delay = 2, + .max_units = 1, + .write_regr = frpw_write_regr, + .read_regr = frpw_read_regr, + .write_block = frpw_write_block, + .read_block = frpw_read_block, + .connect = frpw_connect, + .disconnect = frpw_disconnect, + .test_proto = frpw_test_proto, + .log_adapter = frpw_log_adapter, +}; + +static int __init frpw_init(void) +{ + return pi_register(&frpw)-1; +} + +static void __exit frpw_exit(void) +{ + pi_unregister(&frpw); +} + +MODULE_LICENSE("GPL"); +module_init(frpw_init) +module_exit(frpw_exit) diff --git a/drivers/block/paride/jumbo b/drivers/block/paride/jumbo new file mode 100644 index 00000000000..e793b9cb7e7 --- /dev/null +++ b/drivers/block/paride/jumbo @@ -0,0 +1,70 @@ +#!/bin/sh +# +# This script can be used to build "jumbo" modules that contain the +# base PARIDE support, one protocol module and one high-level driver. +# +echo -n "High level driver [pcd] : " +read X +HLD=${X:-pcd} +# +echo -n "Protocol module [bpck] : " +read X +PROTO=${X:-bpck} +# +echo -n "Use MODVERSIONS [y] ? " +read X +UMODV=${X:-y} +# +echo -n "For SMP kernel [n] ? " +read X +USMP=${X:-n} +# +echo -n "Support PARPORT [n] ? " +read X +UPARP=${X:-n} +# +echo +# +case $USMP in + y* | Y* ) FSMP="-DCONFIG_SMP" + ;; + *) FSMP="" + ;; +esac +# +MODI="-include ../../../include/linux/modversions.h" +# +case $UMODV in + y* | Y* ) FMODV="-DMODVERSIONS $MODI" + ;; + *) FMODV="" + ;; +esac +# +case $UPARP in + y* | Y* ) FPARP="-DCONFIG_PARPORT" + ;; + *) FPARP="" + ;; +esac +# +TARG=$HLD-$PROTO.o +FPROTO=-DCONFIG_PARIDE_`echo "$PROTO" | tr [a-z] [A-Z]` +FK="-D__KERNEL__ -I ../../../include" +FLCH=-D_LINUX_CONFIG_H +# +echo cc $FK $FSMP $FLCH $FPARP $FPROTO $FMODV -Wall -O2 -o Jb.o -c paride.c +cc $FK $FSMP $FLCH $FPARP $FPROTO $FMODV -Wall -O2 -o Jb.o -c paride.c +# +echo cc $FK $FSMP $FMODV -Wall -O2 -o Jp.o -c $PROTO.c +cc $FK $FSMP $FMODV -Wall -O2 -o Jp.o -c $PROTO.c +# +echo cc $FK $FSMP $FMODV -DMODULE -DPARIDE_JUMBO -Wall -O2 -o Jd.o -c $HLD.c +cc $FK $FSMP $FMODV -DMODULE -DPARIDE_JUMBO -Wall -O2 -o Jd.o -c $HLD.c +# +echo ld -r -o $TARG Jp.o Jb.o Jd.o +ld -r -o $TARG Jp.o Jb.o Jd.o +# +# +rm Jp.o Jb.o Jd.o +# diff --git a/drivers/block/paride/kbic.c b/drivers/block/paride/kbic.c new file mode 100644 index 00000000000..d983bcea76f --- /dev/null +++ b/drivers/block/paride/kbic.c @@ -0,0 +1,297 @@ +/* + kbic.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is a low-level driver for the KBIC-951A and KBIC-971A + parallel to IDE adapter chips from KingByte Information Systems. + + The chips are almost identical, however, the wakeup code + required for the 971A interferes with the correct operation of + the 951A, so this driver registers itself twice, once for + each chip. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 init_proto, release_proto + +*/ + +#define KBIC_VERSION "1.01" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define r12w() (delay_p,inw(pi->port+1)&0xffff) + +#define j44(a,b) ((((a>>4)&0x0f)|(b&0xf0))^0x88) +#define j53(w) (((w>>3)&0x1f)|((w>>4)&0xe0)) + + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int cont_map[2] = { 0x80, 0x40 }; + +static int kbic_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, s; + + s = cont_map[cont]; + + switch (pi->mode) { + + case 0: w0(regr|0x18|s); w2(4); w2(6); w2(4); w2(1); w0(8); + a = r1(); w0(0x28); b = r1(); w2(4); + return j44(a,b); + + case 1: w0(regr|0x38|s); w2(4); w2(6); w2(4); w2(5); w0(8); + a = r12w(); w2(4); + return j53(a); + + case 2: w0(regr|0x08|s); w2(4); w2(6); w2(4); w2(0xa5); w2(0xa1); + a = r0(); w2(4); + return a; + + case 3: + case 4: + case 5: w0(0x20|s); w2(4); w2(6); w2(4); w3(regr); + a = r4(); b = r4(); w2(4); w2(0); w2(4); + return a; + + } + return -1; +} + +static void kbic_write_regr( PIA *pi, int cont, int regr, int val) + +{ int s; + + s = cont_map[cont]; + + switch (pi->mode) { + + case 0: + case 1: + case 2: w0(regr|0x10|s); w2(4); w2(6); w2(4); + w0(val); w2(5); w2(4); + break; + + case 3: + case 4: + case 5: w0(0x20|s); w2(4); w2(6); w2(4); w3(regr); + w4(val); w4(val); + w2(4); w2(0); w2(4); + break; + + } +} + +static void k951_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(4); +} + +static void k951_disconnect ( PIA *pi ) + +{ w0(pi->saved_r0); + w2(pi->saved_r2); +} + +#define CCP(x) w2(0xc4);w0(0xaa);w0(0x55);w0(0);w0(0xff);w0(0x87);\ + w0(0x78);w0(x);w2(0xc5);w2(0xc4);w0(0xff); + +static void k971_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + CCP(0x20); + w2(4); +} + +static void k971_disconnect ( PIA *pi ) + +{ CCP(0x30); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +/* counts must be congruent to 0 MOD 4, but all known applications + have this property. +*/ + +static void kbic_read_block( PIA *pi, char * buf, int count ) + +{ int k, a, b; + + switch (pi->mode) { + + case 0: w0(0x98); w2(4); w2(6); w2(4); + for (k=0;k<count/2;k++) { + w2(1); w0(8); a = r1(); + w0(0x28); b = r1(); + buf[2*k] = j44(a,b); + w2(5); b = r1(); + w0(8); a = r1(); + buf[2*k+1] = j44(a,b); + w2(4); + } + break; + + case 1: w0(0xb8); w2(4); w2(6); w2(4); + for (k=0;k<count/4;k++) { + w0(0xb8); + w2(4); w2(5); + w0(8); buf[4*k] = j53(r12w()); + w0(0xb8); buf[4*k+1] = j53(r12w()); + w2(4); w2(5); + buf[4*k+3] = j53(r12w()); + w0(8); buf[4*k+2] = j53(r12w()); + } + w2(4); + break; + + case 2: w0(0x88); w2(4); w2(6); w2(4); + for (k=0;k<count/2;k++) { + w2(0xa0); w2(0xa1); buf[2*k] = r0(); + w2(0xa5); buf[2*k+1] = r0(); + } + w2(4); + break; + + case 3: w0(0xa0); w2(4); w2(6); w2(4); w3(0); + for (k=0;k<count;k++) buf[k] = r4(); + w2(4); w2(0); w2(4); + break; + + case 4: w0(0xa0); w2(4); w2(6); w2(4); w3(0); + for (k=0;k<count/2;k++) ((u16 *)buf)[k] = r4w(); + w2(4); w2(0); w2(4); + break; + + case 5: w0(0xa0); w2(4); w2(6); w2(4); w3(0); + for (k=0;k<count/4;k++) ((u32 *)buf)[k] = r4l(); + w2(4); w2(0); w2(4); + break; + + + } +} + +static void kbic_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + switch (pi->mode) { + + case 0: + case 1: + case 2: w0(0x90); w2(4); w2(6); w2(4); + for(k=0;k<count/2;k++) { + w0(buf[2*k+1]); w2(0); w2(4); + w0(buf[2*k]); w2(5); w2(4); + } + break; + + case 3: w0(0xa0); w2(4); w2(6); w2(4); w3(0); + for(k=0;k<count/2;k++) { + w4(buf[2*k+1]); + w4(buf[2*k]); + } + w2(4); w2(0); w2(4); + break; + + case 4: w0(0xa0); w2(4); w2(6); w2(4); w3(0); + for(k=0;k<count/2;k++) w4w(pi_swab16(buf,k)); + w2(4); w2(0); w2(4); + break; + + case 5: w0(0xa0); w2(4); w2(6); w2(4); w3(0); + for(k=0;k<count/4;k++) w4l(pi_swab32(buf,k)); + w2(4); w2(0); w2(4); + break; + + } + +} + +static void kbic_log_adapter( PIA *pi, char * scratch, + int verbose, char * chip ) + +{ char *mode_string[6] = {"4-bit","5/3","8-bit", + "EPP-8","EPP_16","EPP-32"}; + + printk("%s: kbic %s, KingByte %s at 0x%x, ", + pi->device,KBIC_VERSION,chip,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static void k951_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ kbic_log_adapter(pi,scratch,verbose,"KBIC-951A"); +} + +static void k971_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ kbic_log_adapter(pi,scratch,verbose,"KBIC-971A"); +} + +static struct pi_protocol k951 = { + .owner = THIS_MODULE, + .name = "k951", + .max_mode = 6, + .epp_first = 3, + .default_delay = 1, + .max_units = 1, + .write_regr = kbic_write_regr, + .read_regr = kbic_read_regr, + .write_block = kbic_write_block, + .read_block = kbic_read_block, + .connect = k951_connect, + .disconnect = k951_disconnect, + .log_adapter = k951_log_adapter, +}; + +static struct pi_protocol k971 = { + .owner = THIS_MODULE, + .name = "k971", + .max_mode = 6, + .epp_first = 3, + .default_delay = 1, + .max_units = 1, + .write_regr = kbic_write_regr, + .read_regr = kbic_read_regr, + .write_block = kbic_write_block, + .read_block = kbic_read_block, + .connect = k971_connect, + .disconnect = k971_disconnect, + .log_adapter = k971_log_adapter, +}; + +static int __init kbic_init(void) +{ + return (pi_register(&k951)||pi_register(&k971))-1; +} + +static void __exit kbic_exit(void) +{ + pi_unregister(&k951); + pi_unregister(&k971); +} + +MODULE_LICENSE("GPL"); +module_init(kbic_init) +module_exit(kbic_exit) diff --git a/drivers/block/paride/ktti.c b/drivers/block/paride/ktti.c new file mode 100644 index 00000000000..6c7edbfba9a --- /dev/null +++ b/drivers/block/paride/ktti.c @@ -0,0 +1,128 @@ +/* + ktti.c (c) 1998 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + ktti.c is a low-level protocol driver for the KT Technology + parallel port adapter. This adapter is used in the "PHd" + portable hard-drives. As far as I can tell, this device + supports 4-bit mode _only_. + +*/ + +#define KTTI_VERSION "1.0" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define j44(a,b) (((a>>4)&0x0f)|(b&0xf0)) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int cont_map[2] = { 0x10, 0x08 }; + +static void ktti_write_regr( PIA *pi, int cont, int regr, int val) + +{ int r; + + r = regr + cont_map[cont]; + + w0(r); w2(0xb); w2(0xa); w2(3); w2(6); + w0(val); w2(3); w0(0); w2(6); w2(0xb); +} + +static int ktti_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, r; + + r = regr + cont_map[cont]; + + w0(r); w2(0xb); w2(0xa); w2(9); w2(0xc); w2(9); + a = r1(); w2(0xc); b = r1(); w2(9); w2(0xc); w2(9); + return j44(a,b); + +} + +static void ktti_read_block( PIA *pi, char * buf, int count ) + +{ int k, a, b; + + for (k=0;k<count/2;k++) { + w0(0x10); w2(0xb); w2(0xa); w2(9); w2(0xc); w2(9); + a = r1(); w2(0xc); b = r1(); w2(9); + buf[2*k] = j44(a,b); + a = r1(); w2(0xc); b = r1(); w2(9); + buf[2*k+1] = j44(a,b); + } +} + +static void ktti_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + for (k=0;k<count/2;k++) { + w0(0x10); w2(0xb); w2(0xa); w2(3); w2(6); + w0(buf[2*k]); w2(3); + w0(buf[2*k+1]); w2(6); + w2(0xb); + } +} + +static void ktti_connect ( PIA *pi ) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + w2(0xb); w2(0xa); w0(0); w2(3); w2(6); +} + +static void ktti_disconnect ( PIA *pi ) + +{ w2(0xb); w2(0xa); w0(0xa0); w2(3); w2(4); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void ktti_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ printk("%s: ktti %s, KT adapter at 0x%x, delay %d\n", + pi->device,KTTI_VERSION,pi->port,pi->delay); + +} + +static struct pi_protocol ktti = { + .owner = THIS_MODULE, + .name = "ktti", + .max_mode = 1, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = ktti_write_regr, + .read_regr = ktti_read_regr, + .write_block = ktti_write_block, + .read_block = ktti_read_block, + .connect = ktti_connect, + .disconnect = ktti_disconnect, + .log_adapter = ktti_log_adapter, +}; + +static int __init ktti_init(void) +{ + return pi_register(&ktti)-1; +} + +static void __exit ktti_exit(void) +{ + pi_unregister(&ktti); +} + +MODULE_LICENSE("GPL"); +module_init(ktti_init) +module_exit(ktti_exit) diff --git a/drivers/block/paride/mkd b/drivers/block/paride/mkd new file mode 100644 index 00000000000..971f099b40a --- /dev/null +++ b/drivers/block/paride/mkd @@ -0,0 +1,30 @@ +#!/bin/bash +# +# mkd -- a script to create the device special files for the PARIDE subsystem +# +# block devices: pd (45), pcd (46), pf (47) +# character devices: pt (96), pg (97) +# +function mkdev { + mknod $1 $2 $3 $4 ; chmod 0660 $1 ; chown root:disk $1 +} +# +function pd { + D=$( printf \\$( printf "x%03x" $[ $1 + 97 ] ) ) + mkdev pd$D b 45 $[ $1 * 16 ] + for P in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 + do mkdev pd$D$P b 45 $[ $1 * 16 + $P ] + done +} +# +cd /dev +# +for u in 0 1 2 3 ; do pd $u ; done +for u in 0 1 2 3 ; do mkdev pcd$u b 46 $u ; done +for u in 0 1 2 3 ; do mkdev pf$u b 47 $u ; done +for u in 0 1 2 3 ; do mkdev pt$u c 96 $u ; done +for u in 0 1 2 3 ; do mkdev npt$u c 96 $[ $u + 128 ] ; done +for u in 0 1 2 3 ; do mkdev pg$u c 97 $u ; done +# +# end of mkd + diff --git a/drivers/block/paride/on20.c b/drivers/block/paride/on20.c new file mode 100644 index 00000000000..9f8e0109680 --- /dev/null +++ b/drivers/block/paride/on20.c @@ -0,0 +1,153 @@ +/* + on20.c (c) 1996-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + on20.c is a low-level protocol driver for the + Onspec 90c20 parallel to IDE adapter. +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 init_proto, release_proto + +*/ + +#define ON20_VERSION "1.01" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +#define op(f) w2(4);w0(f);w2(5);w2(0xd);w2(5);w2(0xd);w2(5);w2(4); +#define vl(v) w2(4);w0(v);w2(5);w2(7);w2(5);w2(4); + +#define j44(a,b) (((a>>4)&0x0f)|(b&0xf0)) + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int on20_read_regr( PIA *pi, int cont, int regr ) + +{ int h,l, r ; + + r = (regr<<2) + 1 + cont; + + op(1); vl(r); op(0); + + switch (pi->mode) { + + case 0: w2(4); w2(6); l = r1(); + w2(4); w2(6); h = r1(); + w2(4); w2(6); w2(4); w2(6); w2(4); + return j44(l,h); + + case 1: w2(4); w2(0x26); r = r0(); + w2(4); w2(0x26); w2(4); + return r; + + } + return -1; +} + +static void on20_write_regr( PIA *pi, int cont, int regr, int val ) + +{ int r; + + r = (regr<<2) + 1 + cont; + + op(1); vl(r); + op(0); vl(val); + op(0); vl(val); +} + +static void on20_connect ( PIA *pi) + +{ pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + + w2(4);w0(0);w2(0xc);w2(4);w2(6);w2(4);w2(6);w2(4); + if (pi->mode) { op(2); vl(8); op(2); vl(9); } + else { op(2); vl(0); op(2); vl(8); } +} + +static void on20_disconnect ( PIA *pi ) + +{ w2(4);w0(7);w2(4);w2(0xc);w2(4); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +static void on20_read_block( PIA *pi, char * buf, int count ) + +{ int k, l, h; + + op(1); vl(1); op(0); + + for (k=0;k<count;k++) + if (pi->mode) { + w2(4); w2(0x26); buf[k] = r0(); + } else { + w2(6); l = r1(); w2(4); + w2(6); h = r1(); w2(4); + buf[k] = j44(l,h); + } + w2(4); +} + +static void on20_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + op(1); vl(1); op(0); + + for (k=0;k<count;k++) { w2(5); w0(buf[k]); w2(7); } + w2(4); +} + +static void on20_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[2] = {"4-bit","8-bit"}; + + printk("%s: on20 %s, OnSpec 90c20 at 0x%x, ", + pi->device,ON20_VERSION,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol on20 = { + .owner = THIS_MODULE, + .name = "on20", + .max_mode = 2, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = on20_write_regr, + .read_regr = on20_read_regr, + .write_block = on20_write_block, + .read_block = on20_read_block, + .connect = on20_connect, + .disconnect = on20_disconnect, + .log_adapter = on20_log_adapter, +}; + +static int __init on20_init(void) +{ + return pi_register(&on20)-1; +} + +static void __exit on20_exit(void) +{ + pi_unregister(&on20); +} + +MODULE_LICENSE("GPL"); +module_init(on20_init) +module_exit(on20_exit) diff --git a/drivers/block/paride/on26.c b/drivers/block/paride/on26.c new file mode 100644 index 00000000000..9f837d9a363 --- /dev/null +++ b/drivers/block/paride/on26.c @@ -0,0 +1,319 @@ +/* + on26.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + on26.c is a low-level protocol driver for the + OnSpec 90c26 parallel to IDE adapter chip. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 init_proto, release_proto + 1.02 GRG 1998.09.23 updates for the -E rev chip + 1.03 GRG 1998.12.14 fix for slave drives + 1.04 GRG 1998.12.20 yet another bug fix + +*/ + +#define ON26_VERSION "1.04" + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/delay.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/wait.h> +#include <asm/io.h> + +#include "paride.h" + +/* mode codes: 0 nybble reads, 8-bit writes + 1 8-bit reads and writes + 2 8-bit EPP mode + 3 EPP-16 + 4 EPP-32 +*/ + +#define j44(a,b) (((a>>4)&0x0f)|(b&0xf0)) + +#define P1 w2(5);w2(0xd);w2(5);w2(0xd);w2(5);w2(4); +#define P2 w2(5);w2(7);w2(5);w2(4); + +/* cont = 0 - access the IDE register file + cont = 1 - access the IDE command set +*/ + +static int on26_read_regr( PIA *pi, int cont, int regr ) + +{ int a, b, r; + + r = (regr<<2) + 1 + cont; + + switch (pi->mode) { + + case 0: w0(1); P1; w0(r); P2; w0(0); P1; + w2(6); a = r1(); w2(4); + w2(6); b = r1(); w2(4); + w2(6); w2(4); w2(6); w2(4); + return j44(a,b); + + case 1: w0(1); P1; w0(r); P2; w0(0); P1; + w2(0x26); a = r0(); w2(4); w2(0x26); w2(4); + return a; + + case 2: + case 3: + case 4: w3(1); w3(1); w2(5); w4(r); w2(4); + w3(0); w3(0); w2(0x24); a = r4(); w2(4); + w2(0x24); r4(); w2(4); + return a; + + } + return -1; +} + +static void on26_write_regr( PIA *pi, int cont, int regr, int val ) + +{ int r; + + r = (regr<<2) + 1 + cont; + + switch (pi->mode) { + + case 0: + case 1: w0(1); P1; w0(r); P2; w0(0); P1; + w0(val); P2; w0(val); P2; + break; + + case 2: + case 3: + case 4: w3(1); w3(1); w2(5); w4(r); w2(4); + w3(0); w3(0); + w2(5); w4(val); w2(4); + w2(5); w4(val); w2(4); + break; + } +} + +#define CCP(x) w0(0xfe);w0(0xaa);w0(0x55);w0(0);w0(0xff);\ + w0(0x87);w0(0x78);w0(x);w2(4);w2(5);w2(4);w0(0xff); + +static void on26_connect ( PIA *pi ) + +{ int x; + + pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + + CCP(0x20); + x = 8; if (pi->mode) x = 9; + + w0(2); P1; w0(8); P2; + w0(2); P1; w0(x); P2; +} + +static void on26_disconnect ( PIA *pi ) + +{ if (pi->mode >= 2) { w3(4); w3(4); w3(4); w3(4); } + else { w0(4); P1; w0(4); P1; } + CCP(0x30); + w0(pi->saved_r0); + w2(pi->saved_r2); +} + +#define RESET_WAIT 200 + +static int on26_test_port( PIA *pi) /* hard reset */ + +{ int i, m, d, x=0, y=0; + + pi->saved_r0 = r0(); + pi->saved_r2 = r2(); + + d = pi->delay; + m = pi->mode; + pi->delay = 5; + pi->mode = 0; + + w2(0xc); + + CCP(0x30); CCP(0); + + w0(0xfe);w0(0xaa);w0(0x55);w0(0);w0(0xff); + i = ((r1() & 0xf0) << 4); w0(0x87); + i |= (r1() & 0xf0); w0(0x78); + w0(0x20);w2(4);w2(5); + i |= ((r1() & 0xf0) >> 4); + w2(4);w0(0xff); + + if (i == 0xb5f) { + + w0(2); P1; w0(0); P2; + w0(3); P1; w0(0); P2; + w0(2); P1; w0(8); P2; udelay(100); + w0(2); P1; w0(0xa); P2; udelay(100); + w0(2); P1; w0(8); P2; udelay(1000); + + on26_write_regr(pi,0,6,0xa0); + + for (i=0;i<RESET_WAIT;i++) { + on26_write_regr(pi,0,6,0xa0); + x = on26_read_regr(pi,0,7); + on26_write_regr(pi,0,6,0xb0); + y = on26_read_regr(pi,0,7); + if (!((x&0x80)||(y&0x80))) break; + mdelay(100); + } + + if (i == RESET_WAIT) + printk("on26: Device reset failed (%x,%x)\n",x,y); + + w0(4); P1; w0(4); P1; + } + + CCP(0x30); + + pi->delay = d; + pi->mode = m; + w0(pi->saved_r0); + w2(pi->saved_r2); + + return 5; +} + + +static void on26_read_block( PIA *pi, char * buf, int count ) + +{ int k, a, b; + + switch (pi->mode) { + + case 0: w0(1); P1; w0(1); P2; w0(2); P1; w0(0x18); P2; w0(0); P1; + udelay(10); + for (k=0;k<count;k++) { + w2(6); a = r1(); + w2(4); b = r1(); + buf[k] = j44(a,b); + } + w0(2); P1; w0(8); P2; + break; + + case 1: w0(1); P1; w0(1); P2; w0(2); P1; w0(0x19); P2; w0(0); P1; + udelay(10); + for (k=0;k<count/2;k++) { + w2(0x26); buf[2*k] = r0(); + w2(0x24); buf[2*k+1] = r0(); + } + w0(2); P1; w0(9); P2; + break; + + case 2: w3(1); w3(1); w2(5); w4(1); w2(4); + w3(0); w3(0); w2(0x24); + udelay(10); + for (k=0;k<count;k++) buf[k] = r4(); + w2(4); + break; + + case 3: w3(1); w3(1); w2(5); w4(1); w2(4); + w3(0); w3(0); w2(0x24); + udelay(10); + for (k=0;k<count/2;k++) ((u16 *)buf)[k] = r4w(); + w2(4); + break; + + case 4: w3(1); w3(1); w2(5); w4(1); w2(4); + w3(0); w3(0); w2(0x24); + udelay(10); + for (k=0;k<count/4;k++) ((u32 *)buf)[k] = r4l(); + w2(4); + break; + + } +} + +static void on26_write_block( PIA *pi, char * buf, int count ) + +{ int k; + + switch (pi->mode) { + + case 0: + case 1: w0(1); P1; w0(1); P2; + w0(2); P1; w0(0x18+pi->mode); P2; w0(0); P1; + udelay(10); + for (k=0;k<count/2;k++) { + w2(5); w0(buf[2*k]); + w2(7); w0(buf[2*k+1]); + } + w2(5); w2(4); + w0(2); P1; w0(8+pi->mode); P2; + break; + + case 2: w3(1); w3(1); w2(5); w4(1); w2(4); + w3(0); w3(0); w2(0xc5); + udelay(10); + for (k=0;k<count;k++) w4(buf[k]); + w2(0xc4); + break; + + case 3: w3(1); w3(1); w2(5); w4(1); w2(4); + w3(0); w3(0); w2(0xc5); + udelay(10); + for (k=0;k<count/2;k++) w4w(((u16 *)buf)[k]); + w2(0xc4); + break; + + case 4: w3(1); w3(1); w2(5); w4(1); w2(4); + w3(0); w3(0); w2(0xc5); + udelay(10); + for (k=0;k<count/4;k++) w4l(((u32 *)buf)[k]); + w2(0xc4); + break; + + } + +} + +static void on26_log_adapter( PIA *pi, char * scratch, int verbose ) + +{ char *mode_string[5] = {"4-bit","8-bit","EPP-8", + "EPP-16","EPP-32"}; + + printk("%s: on26 %s, OnSpec 90c26 at 0x%x, ", + pi->device,ON26_VERSION,pi->port); + printk("mode %d (%s), delay %d\n",pi->mode, + mode_string[pi->mode],pi->delay); + +} + +static struct pi_protocol on26 = { + .owner = THIS_MODULE, + .name = "on26", + .max_mode = 5, + .epp_first = 2, + .default_delay = 1, + .max_units = 1, + .write_regr = on26_write_regr, + .read_regr = on26_read_regr, + .write_block = on26_write_block, + .read_block = on26_read_block, + .connect = on26_connect, + .disconnect = on26_disconnect, + .test_port = on26_test_port, + .log_adapter = on26_log_adapter, +}; + +static int __init on26_init(void) +{ + return pi_register(&on26)-1; +} + +static void __exit on26_exit(void) +{ + pi_unregister(&on26); +} + +MODULE_LICENSE("GPL"); +module_init(on26_init) +module_exit(on26_exit) diff --git a/drivers/block/paride/paride.c b/drivers/block/paride/paride.c new file mode 100644 index 00000000000..1fef136c0e4 --- /dev/null +++ b/drivers/block/paride/paride.c @@ -0,0 +1,467 @@ +/* + paride.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is the base module for the family of device drivers + that support parallel port IDE devices. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.03 Use spinlocks + 1.02 GRG 1998.05.05 init_proto, release_proto, ktti + 1.03 GRG 1998.08.15 eliminate compiler warning + 1.04 GRG 1998.11.28 added support for FRIQ + 1.05 TMW 2000.06.06 use parport_find_number instead of + parport_enumerate + 1.06 TMW 2001.03.26 more sane parport-or-not resource management +*/ + +#define PI_VERSION "1.06" + +#include <linux/module.h> +#include <linux/config.h> +#include <linux/kmod.h> +#include <linux/types.h> +#include <linux/kernel.h> +#include <linux/ioport.h> +#include <linux/string.h> +#include <linux/spinlock.h> +#include <linux/wait.h> + +#ifdef CONFIG_PARPORT_MODULE +#define CONFIG_PARPORT +#endif + +#ifdef CONFIG_PARPORT +#include <linux/parport.h> +#endif + +#include "paride.h" + +MODULE_LICENSE("GPL"); + +#define MAX_PROTOS 32 + +static struct pi_protocol *protocols[MAX_PROTOS]; + +static DEFINE_SPINLOCK(pi_spinlock); + +void pi_write_regr(PIA * pi, int cont, int regr, int val) +{ + pi->proto->write_regr(pi, cont, regr, val); +} + +EXPORT_SYMBOL(pi_write_regr); + +int pi_read_regr(PIA * pi, int cont, int regr) +{ + return pi->proto->read_regr(pi, cont, regr); +} + +EXPORT_SYMBOL(pi_read_regr); + +void pi_write_block(PIA * pi, char *buf, int count) +{ + pi->proto->write_block(pi, buf, count); +} + +EXPORT_SYMBOL(pi_write_block); + +void pi_read_block(PIA * pi, char *buf, int count) +{ + pi->proto->read_block(pi, buf, count); +} + +EXPORT_SYMBOL(pi_read_block); + +#ifdef CONFIG_PARPORT + +static void pi_wake_up(void *p) +{ + PIA *pi = (PIA *) p; + unsigned long flags; + void (*cont) (void) = NULL; + + spin_lock_irqsave(&pi_spinlock, flags); + + if (pi->claim_cont && !parport_claim(pi->pardev)) { + cont = pi->claim_cont; + pi->claim_cont = NULL; + pi->claimed = 1; + } + + spin_unlock_irqrestore(&pi_spinlock, flags); + + wake_up(&(pi->parq)); + + if (cont) + cont(); +} + +#endif + +int pi_schedule_claimed(PIA * pi, void (*cont) (void)) +{ +#ifdef CONFIG_PARPORT + unsigned long flags; + + spin_lock_irqsave(&pi_spinlock, flags); + if (pi->pardev && parport_claim(pi->pardev)) { + pi->claim_cont = cont; + spin_unlock_irqrestore(&pi_spinlock, flags); + return 0; + } + pi->claimed = 1; + spin_unlock_irqrestore(&pi_spinlock, flags); +#endif + return 1; +} +EXPORT_SYMBOL(pi_schedule_claimed); + +void pi_do_claimed(PIA * pi, void (*cont) (void)) +{ + if (pi_schedule_claimed(pi, cont)) + cont(); +} + +EXPORT_SYMBOL(pi_do_claimed); + +static void pi_claim(PIA * pi) +{ + if (pi->claimed) + return; + pi->claimed = 1; +#ifdef CONFIG_PARPORT + if (pi->pardev) + wait_event(pi->parq, + !parport_claim((struct pardevice *) pi->pardev)); +#endif +} + +static void pi_unclaim(PIA * pi) +{ + pi->claimed = 0; +#ifdef CONFIG_PARPORT + if (pi->pardev) + parport_release((struct pardevice *) (pi->pardev)); +#endif +} + +void pi_connect(PIA * pi) +{ + pi_claim(pi); + pi->proto->connect(pi); +} + +EXPORT_SYMBOL(pi_connect); + +void pi_disconnect(PIA * pi) +{ + pi->proto->disconnect(pi); + pi_unclaim(pi); +} + +EXPORT_SYMBOL(pi_disconnect); + +static void pi_unregister_parport(PIA * pi) +{ +#ifdef CONFIG_PARPORT + if (pi->pardev) { + parport_unregister_device((struct pardevice *) (pi->pardev)); + pi->pardev = NULL; + } +#endif +} + +void pi_release(PIA * pi) +{ + pi_unregister_parport(pi); +#ifndef CONFIG_PARPORT + if (pi->reserved) + release_region(pi->port, pi->reserved); +#endif /* !CONFIG_PARPORT */ + if (pi->proto->release_proto) + pi->proto->release_proto(pi); + module_put(pi->proto->owner); +} + +EXPORT_SYMBOL(pi_release); + +static int default_test_proto(PIA * pi, char *scratch, int verbose) +{ + int j, k; + int e[2] = { 0, 0 }; + + pi->proto->connect(pi); + + for (j = 0; j < 2; j++) { + pi_write_regr(pi, 0, 6, 0xa0 + j * 0x10); + for (k = 0; k < 256; k++) { + pi_write_regr(pi, 0, 2, k ^ 0xaa); + pi_write_regr(pi, 0, 3, k ^ 0x55); + if (pi_read_regr(pi, 0, 2) != (k ^ 0xaa)) + e[j]++; + } + } + pi->proto->disconnect(pi); + + if (verbose) + printk("%s: %s: port 0x%x, mode %d, test=(%d,%d)\n", + pi->device, pi->proto->name, pi->port, + pi->mode, e[0], e[1]); + + return (e[0] && e[1]); /* not here if both > 0 */ +} + +static int pi_test_proto(PIA * pi, char *scratch, int verbose) +{ + int res; + + pi_claim(pi); + if (pi->proto->test_proto) + res = pi->proto->test_proto(pi, scratch, verbose); + else + res = default_test_proto(pi, scratch, verbose); + pi_unclaim(pi); + + return res; +} + +int pi_register(PIP * pr) +{ + int k; + + for (k = 0; k < MAX_PROTOS; k++) + if (protocols[k] && !strcmp(pr->name, protocols[k]->name)) { + printk("paride: %s protocol already registered\n", + pr->name); + return 0; + } + k = 0; + while ((k < MAX_PROTOS) && (protocols[k])) + k++; + if (k == MAX_PROTOS) { + printk("paride: protocol table full\n"); + return 0; + } + protocols[k] = pr; + pr->index = k; + printk("paride: %s registered as protocol %d\n", pr->name, k); + return 1; +} + +EXPORT_SYMBOL(pi_register); + +void pi_unregister(PIP * pr) +{ + if (!pr) + return; + if (protocols[pr->index] != pr) { + printk("paride: %s not registered\n", pr->name); + return; + } + protocols[pr->index] = NULL; +} + +EXPORT_SYMBOL(pi_unregister); + +static int pi_register_parport(PIA * pi, int verbose) +{ +#ifdef CONFIG_PARPORT + + struct parport *port; + + port = parport_find_base(pi->port); + if (!port) + return 0; + + pi->pardev = parport_register_device(port, + pi->device, NULL, + pi_wake_up, NULL, 0, (void *) pi); + parport_put_port(port); + if (!pi->pardev) + return 0; + + init_waitqueue_head(&pi->parq); + + if (verbose) + printk("%s: 0x%x is %s\n", pi->device, pi->port, port->name); + + pi->parname = (char *) port->name; +#endif + + return 1; +} + +static int pi_probe_mode(PIA * pi, int max, char *scratch, int verbose) +{ + int best, range; + + if (pi->mode != -1) { + if (pi->mode >= max) + return 0; + range = 3; + if (pi->mode >= pi->proto->epp_first) + range = 8; + if ((range == 8) && (pi->port % 8)) + return 0; + pi->reserved = range; + return (!pi_test_proto(pi, scratch, verbose)); + } + best = -1; + for (pi->mode = 0; pi->mode < max; pi->mode++) { + range = 3; + if (pi->mode >= pi->proto->epp_first) + range = 8; + if ((range == 8) && (pi->port % 8)) + break; + pi->reserved = range; + if (!pi_test_proto(pi, scratch, verbose)) + best = pi->mode; + } + pi->mode = best; + return (best > -1); +} + +static int pi_probe_unit(PIA * pi, int unit, char *scratch, int verbose) +{ + int max, s, e; + + s = unit; + e = s + 1; + + if (s == -1) { + s = 0; + e = pi->proto->max_units; + } + + if (!pi_register_parport(pi, verbose)) + return 0; + + if (pi->proto->test_port) { + pi_claim(pi); + max = pi->proto->test_port(pi); + pi_unclaim(pi); + } else + max = pi->proto->max_mode; + + if (pi->proto->probe_unit) { + pi_claim(pi); + for (pi->unit = s; pi->unit < e; pi->unit++) + if (pi->proto->probe_unit(pi)) { + pi_unclaim(pi); + if (pi_probe_mode(pi, max, scratch, verbose)) + return 1; + pi_unregister_parport(pi); + return 0; + } + pi_unclaim(pi); + pi_unregister_parport(pi); + return 0; + } + + if (!pi_probe_mode(pi, max, scratch, verbose)) { + pi_unregister_parport(pi); + return 0; + } + return 1; + +} + +int pi_init(PIA * pi, int autoprobe, int port, int mode, + int unit, int protocol, int delay, char *scratch, + int devtype, int verbose, char *device) +{ + int p, k, s, e; + int lpts[7] = { 0x3bc, 0x378, 0x278, 0x268, 0x27c, 0x26c, 0 }; + + s = protocol; + e = s + 1; + + if (!protocols[0]) + request_module("paride_protocol"); + + if (autoprobe) { + s = 0; + e = MAX_PROTOS; + } else if ((s < 0) || (s >= MAX_PROTOS) || (port <= 0) || + (!protocols[s]) || (unit < 0) || + (unit >= protocols[s]->max_units)) { + printk("%s: Invalid parameters\n", device); + return 0; + } + + for (p = s; p < e; p++) { + struct pi_protocol *proto = protocols[p]; + if (!proto) + continue; + /* still racy */ + if (!try_module_get(proto->owner)) + continue; + pi->proto = proto; + pi->private = 0; + if (proto->init_proto && proto->init_proto(pi) < 0) { + pi->proto = NULL; + module_put(proto->owner); + continue; + } + if (delay == -1) + pi->delay = pi->proto->default_delay; + else + pi->delay = delay; + pi->devtype = devtype; + pi->device = device; + + pi->parname = NULL; + pi->pardev = NULL; + init_waitqueue_head(&pi->parq); + pi->claimed = 0; + pi->claim_cont = NULL; + + pi->mode = mode; + if (port != -1) { + pi->port = port; + if (pi_probe_unit(pi, unit, scratch, verbose)) + break; + pi->port = 0; + } else { + k = 0; + while ((pi->port = lpts[k++])) + if (pi_probe_unit + (pi, unit, scratch, verbose)) + break; + if (pi->port) + break; + } + if (pi->proto->release_proto) + pi->proto->release_proto(pi); + module_put(proto->owner); + } + + if (!pi->port) { + if (autoprobe) + printk("%s: Autoprobe failed\n", device); + else + printk("%s: Adapter not found\n", device); + return 0; + } +#ifndef CONFIG_PARPORT + if (!request_region(pi->port, pi->reserved, pi->device)) { + printk(KERN_WARNING "paride: Unable to request region 0x%x\n", + pi->port); + return 0; + } +#endif /* !CONFIG_PARPORT */ + + if (pi->parname) + printk("%s: Sharing %s at 0x%x\n", pi->device, + pi->parname, pi->port); + + pi->proto->log_adapter(pi, scratch, verbose); + + return 1; +} + +EXPORT_SYMBOL(pi_init); diff --git a/drivers/block/paride/paride.h b/drivers/block/paride/paride.h new file mode 100644 index 00000000000..c6d98ef09e4 --- /dev/null +++ b/drivers/block/paride/paride.h @@ -0,0 +1,170 @@ +#ifndef __DRIVERS_PARIDE_H__ +#define __DRIVERS_PARIDE_H__ + +/* + paride.h (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GPL. + + This file defines the interface between the high-level parallel + IDE device drivers (pd, pf, pcd, pt) and the adapter chips. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.05 init_proto, release_proto +*/ + +#define PARIDE_H_VERSION "1.01" + +/* Some adapters need to know what kind of device they are in + + Values for devtype: +*/ + +#define PI_PD 0 /* IDE disk */ +#define PI_PCD 1 /* ATAPI CDrom */ +#define PI_PF 2 /* ATAPI disk */ +#define PI_PT 3 /* ATAPI tape */ +#define PI_PG 4 /* ATAPI generic */ + +/* The paride module contains no state, instead the drivers allocate + a pi_adapter data structure and pass it to paride in every operation. + +*/ + +struct pi_adapter { + + struct pi_protocol *proto; /* adapter protocol */ + int port; /* base address of parallel port */ + int mode; /* transfer mode in use */ + int delay; /* adapter delay setting */ + int devtype; /* device type: PI_PD etc. */ + char *device; /* name of driver */ + int unit; /* unit number for chained adapters */ + int saved_r0; /* saved port state */ + int saved_r2; /* saved port state */ + int reserved; /* number of ports reserved */ + unsigned long private; /* for protocol module */ + + wait_queue_head_t parq; /* semaphore for parport sharing */ + void *pardev; /* pointer to pardevice */ + char *parname; /* parport name */ + int claimed; /* parport has already been claimed */ + void (*claim_cont)(void); /* continuation for parport wait */ +}; + +typedef struct pi_adapter PIA; + +/* functions exported by paride to the high level drivers */ + +extern int pi_init(PIA *pi, + int autoprobe, /* 1 to autoprobe */ + int port, /* base port address */ + int mode, /* -1 for autoprobe */ + int unit, /* unit number, if supported */ + int protocol, /* protocol to use */ + int delay, /* -1 to use adapter specific default */ + char * scratch, /* address of 512 byte buffer */ + int devtype, /* device type: PI_PD, PI_PCD, etc ... */ + int verbose, /* log verbose data while probing */ + char *device /* name of the driver */ + ); /* returns 0 on failure, 1 on success */ + +extern void pi_release(PIA *pi); + +/* registers are addressed as (cont,regr) + + cont: 0 for command register file, 1 for control register(s) + regr: 0-7 for register number. + +*/ + +extern void pi_write_regr(PIA *pi, int cont, int regr, int val); + +extern int pi_read_regr(PIA *pi, int cont, int regr); + +extern void pi_write_block(PIA *pi, char * buf, int count); + +extern void pi_read_block(PIA *pi, char * buf, int count); + +extern void pi_connect(PIA *pi); + +extern void pi_disconnect(PIA *pi); + +extern void pi_do_claimed(PIA *pi, void (*cont)(void)); +extern int pi_schedule_claimed(PIA *pi, void (*cont)(void)); + +/* macros and functions exported to the protocol modules */ + +#define delay_p (pi->delay?udelay(pi->delay):(void)0) +#define out_p(offs,byte) outb(byte,pi->port+offs); delay_p; +#define in_p(offs) (delay_p,inb(pi->port+offs)) + +#define w0(byte) {out_p(0,byte);} +#define r0() (in_p(0) & 0xff) +#define w1(byte) {out_p(1,byte);} +#define r1() (in_p(1) & 0xff) +#define w2(byte) {out_p(2,byte);} +#define r2() (in_p(2) & 0xff) +#define w3(byte) {out_p(3,byte);} +#define w4(byte) {out_p(4,byte);} +#define r4() (in_p(4) & 0xff) +#define w4w(data) {outw(data,pi->port+4); delay_p;} +#define w4l(data) {outl(data,pi->port+4); delay_p;} +#define r4w() (delay_p,inw(pi->port+4)&0xffff) +#define r4l() (delay_p,inl(pi->port+4)&0xffffffff) + +static inline u16 pi_swab16( char *b, int k) + +{ union { u16 u; char t[2]; } r; + + r.t[0]=b[2*k+1]; r.t[1]=b[2*k]; + return r.u; +} + +static inline u32 pi_swab32( char *b, int k) + +{ union { u32 u; char f[4]; } r; + + r.f[0]=b[4*k+1]; r.f[1]=b[4*k]; + r.f[2]=b[4*k+3]; r.f[3]=b[4*k+2]; + return r.u; +} + +struct pi_protocol { + + char name[8]; /* name for this protocol */ + int index; /* index into protocol table */ + + int max_mode; /* max mode number */ + int epp_first; /* modes >= this use 8 ports */ + + int default_delay; /* delay parameter if not specified */ + int max_units; /* max chained units probed for */ + + void (*write_regr)(PIA *,int,int,int); + int (*read_regr)(PIA *,int,int); + void (*write_block)(PIA *,char *,int); + void (*read_block)(PIA *,char *,int); + + void (*connect)(PIA *); + void (*disconnect)(PIA *); + + int (*test_port)(PIA *); + int (*probe_unit)(PIA *); + int (*test_proto)(PIA *,char *,int); + void (*log_adapter)(PIA *,char *,int); + + int (*init_proto)(PIA *); + void (*release_proto)(PIA *); + struct module *owner; +}; + +typedef struct pi_protocol PIP; + +extern int pi_register( PIP * ); +extern void pi_unregister ( PIP * ); + +#endif /* __DRIVERS_PARIDE_H__ */ +/* end of paride.h */ diff --git a/drivers/block/paride/pcd.c b/drivers/block/paride/pcd.c new file mode 100644 index 00000000000..7289f67e956 --- /dev/null +++ b/drivers/block/paride/pcd.c @@ -0,0 +1,971 @@ +/* + pcd.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is a high-level driver for parallel port ATAPI CD-ROM + drives based on chips supported by the paride module. + + By default, the driver will autoprobe for a single parallel + port ATAPI CD-ROM drive, but if their individual parameters are + specified, the driver can handle up to 4 drives. + + The behaviour of the pcd driver can be altered by setting + some parameters from the insmod command line. The following + parameters are adjustable: + + drive0 These four arguments can be arrays of + drive1 1-6 integers as follows: + drive2 + drive3 <prt>,<pro>,<uni>,<mod>,<slv>,<dly> + + Where, + + <prt> is the base of the parallel port address for + the corresponding drive. (required) + + <pro> is the protocol number for the adapter that + supports this drive. These numbers are + logged by 'paride' when the protocol modules + are initialised. (0 if not given) + + <uni> for those adapters that support chained + devices, this is the unit selector for the + chain of devices on the given port. It should + be zero for devices that don't support chaining. + (0 if not given) + + <mod> this can be -1 to choose the best mode, or one + of the mode numbers supported by the adapter. + (-1 if not given) + + <slv> ATAPI CD-ROMs can be jumpered to master or slave. + Set this to 0 to choose the master drive, 1 to + choose the slave, -1 (the default) to choose the + first drive found. + + <dly> some parallel ports require the driver to + go more slowly. -1 sets a default value that + should work with the chosen protocol. Otherwise, + set this to a small integer, the larger it is + the slower the port i/o. In some cases, setting + this to zero will speed up the device. (default -1) + + major You may use this parameter to overide the + default major number (46) that this driver + will use. Be sure to change the device + name as well. + + name This parameter is a character string that + contains the name the kernel will use for this + device (in /proc output, for instance). + (default "pcd") + + verbose This parameter controls the amount of logging + that the driver will do. Set it to 0 for + normal operation, 1 to see autoprobe progress + messages, or 2 to see additional debugging + output. (default 0) + + nice This parameter controls the driver's use of + idle CPU time, at the expense of some speed. + + If this driver is built into the kernel, you can use kernel + the following command line parameters, with the same values + as the corresponding module parameters listed above: + + pcd.drive0 + pcd.drive1 + pcd.drive2 + pcd.drive3 + pcd.nice + + In addition, you can use the parameter pcd.disable to disable + the driver entirely. + +*/ + +/* Changes: + + 1.01 GRG 1998.01.24 Added test unit ready support + 1.02 GRG 1998.05.06 Changes to pcd_completion, ready_wait, + and loosen interpretation of ATAPI + standard for clearing error status. + Use spinlocks. Eliminate sti(). + 1.03 GRG 1998.06.16 Eliminated an Ugh + 1.04 GRG 1998.08.15 Added extra debugging, improvements to + pcd_completion, use HZ in loop timing + 1.05 GRG 1998.08.16 Conformed to "Uniform CD-ROM" standard + 1.06 GRG 1998.08.19 Added audio ioctl support + 1.07 GRG 1998.09.24 Increased reset timeout, added jumbo support + +*/ + +#define PCD_VERSION "1.07" +#define PCD_MAJOR 46 +#define PCD_NAME "pcd" +#define PCD_UNITS 4 + +/* Here are things one can override from the insmod command. + Most are autoprobed by paride unless set here. Verbose is off + by default. + +*/ + +static int verbose = 0; +static int major = PCD_MAJOR; +static char *name = PCD_NAME; +static int nice = 0; +static int disable = 0; + +static int drive0[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive1[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive2[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive3[6] = { 0, 0, 0, -1, -1, -1 }; + +static int (*drives[4])[6] = {&drive0, &drive1, &drive2, &drive3}; +static int pcd_drive_count; + +enum {D_PRT, D_PRO, D_UNI, D_MOD, D_SLV, D_DLY}; + +/* end of parameters */ + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/errno.h> +#include <linux/fs.h> +#include <linux/kernel.h> +#include <linux/delay.h> +#include <linux/cdrom.h> +#include <linux/spinlock.h> +#include <linux/blkdev.h> +#include <asm/uaccess.h> + +static spinlock_t pcd_lock; + +module_param(verbose, bool, 0644); +module_param(major, int, 0); +module_param(name, charp, 0); +module_param(nice, int, 0); +module_param_array(drive0, int, NULL, 0); +module_param_array(drive1, int, NULL, 0); +module_param_array(drive2, int, NULL, 0); +module_param_array(drive3, int, NULL, 0); + +#include "paride.h" +#include "pseudo.h" + +#define PCD_RETRIES 5 +#define PCD_TMO 800 /* timeout in jiffies */ +#define PCD_DELAY 50 /* spin delay in uS */ +#define PCD_READY_TMO 20 /* in seconds */ +#define PCD_RESET_TMO 100 /* in tenths of a second */ + +#define PCD_SPIN (1000000*PCD_TMO)/(HZ*PCD_DELAY) + +#define IDE_ERR 0x01 +#define IDE_DRQ 0x08 +#define IDE_READY 0x40 +#define IDE_BUSY 0x80 + +static int pcd_open(struct cdrom_device_info *cdi, int purpose); +static void pcd_release(struct cdrom_device_info *cdi); +static int pcd_drive_status(struct cdrom_device_info *cdi, int slot_nr); +static int pcd_media_changed(struct cdrom_device_info *cdi, int slot_nr); +static int pcd_tray_move(struct cdrom_device_info *cdi, int position); +static int pcd_lock_door(struct cdrom_device_info *cdi, int lock); +static int pcd_drive_reset(struct cdrom_device_info *cdi); +static int pcd_get_mcn(struct cdrom_device_info *cdi, struct cdrom_mcn *mcn); +static int pcd_audio_ioctl(struct cdrom_device_info *cdi, + unsigned int cmd, void *arg); +static int pcd_packet(struct cdrom_device_info *cdi, + struct packet_command *cgc); + +static int pcd_detect(void); +static void pcd_probe_capabilities(void); +static void do_pcd_read_drq(void); +static void do_pcd_request(request_queue_t * q); +static void do_pcd_read(void); + +struct pcd_unit { + struct pi_adapter pia; /* interface to paride layer */ + struct pi_adapter *pi; + int drive; /* master/slave */ + int last_sense; /* result of last request sense */ + int changed; /* media change seen */ + int present; /* does this unit exist ? */ + char *name; /* pcd0, pcd1, etc */ + struct cdrom_device_info info; /* uniform cdrom interface */ + struct gendisk *disk; +}; + +static struct pcd_unit pcd[PCD_UNITS]; + +static char pcd_scratch[64]; +static char pcd_buffer[2048]; /* raw block buffer */ +static int pcd_bufblk = -1; /* block in buffer, in CD units, + -1 for nothing there. See also + pd_unit. + */ + +/* the variables below are used mainly in the I/O request engine, which + processes only one request at a time. +*/ + +static struct pcd_unit *pcd_current; /* current request's drive */ +static struct request *pcd_req; +static int pcd_retries; /* retries on current request */ +static int pcd_busy; /* request being processed ? */ +static int pcd_sector; /* address of next requested sector */ +static int pcd_count; /* number of blocks still to do */ +static char *pcd_buf; /* buffer for request in progress */ + +static int pcd_warned; /* Have we logged a phase warning ? */ + +/* kernel glue structures */ + +static int pcd_block_open(struct inode *inode, struct file *file) +{ + struct pcd_unit *cd = inode->i_bdev->bd_disk->private_data; + return cdrom_open(&cd->info, inode, file); +} + +static int pcd_block_release(struct inode *inode, struct file *file) +{ + struct pcd_unit *cd = inode->i_bdev->bd_disk->private_data; + return cdrom_release(&cd->info, file); +} + +static int pcd_block_ioctl(struct inode *inode, struct file *file, + unsigned cmd, unsigned long arg) +{ + struct pcd_unit *cd = inode->i_bdev->bd_disk->private_data; + return cdrom_ioctl(file, &cd->info, inode, cmd, arg); +} + +static int pcd_block_media_changed(struct gendisk *disk) +{ + struct pcd_unit *cd = disk->private_data; + return cdrom_media_changed(&cd->info); +} + +static struct block_device_operations pcd_bdops = { + .owner = THIS_MODULE, + .open = pcd_block_open, + .release = pcd_block_release, + .ioctl = pcd_block_ioctl, + .media_changed = pcd_block_media_changed, +}; + +static struct cdrom_device_ops pcd_dops = { + .open = pcd_open, + .release = pcd_release, + .drive_status = pcd_drive_status, + .media_changed = pcd_media_changed, + .tray_move = pcd_tray_move, + .lock_door = pcd_lock_door, + .get_mcn = pcd_get_mcn, + .reset = pcd_drive_reset, + .audio_ioctl = pcd_audio_ioctl, + .generic_packet = pcd_packet, + .capability = CDC_CLOSE_TRAY | CDC_OPEN_TRAY | CDC_LOCK | + CDC_MCN | CDC_MEDIA_CHANGED | CDC_RESET | + CDC_PLAY_AUDIO | CDC_GENERIC_PACKET | CDC_CD_R | + CDC_CD_RW, +}; + +static void pcd_init_units(void) +{ + struct pcd_unit *cd; + int unit; + + pcd_drive_count = 0; + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) { + struct gendisk *disk = alloc_disk(1); + if (!disk) + continue; + cd->disk = disk; + cd->pi = &cd->pia; + cd->present = 0; + cd->last_sense = 0; + cd->changed = 1; + cd->drive = (*drives[unit])[D_SLV]; + if ((*drives[unit])[D_PRT]) + pcd_drive_count++; + + cd->name = &cd->info.name[0]; + snprintf(cd->name, sizeof(cd->info.name), "%s%d", name, unit); + cd->info.ops = &pcd_dops; + cd->info.handle = cd; + cd->info.speed = 0; + cd->info.capacity = 1; + cd->info.mask = 0; + disk->major = major; + disk->first_minor = unit; + strcpy(disk->disk_name, cd->name); /* umm... */ + disk->fops = &pcd_bdops; + } +} + +static int pcd_open(struct cdrom_device_info *cdi, int purpose) +{ + struct pcd_unit *cd = cdi->handle; + if (!cd->present) + return -ENODEV; + return 0; +} + +static void pcd_release(struct cdrom_device_info *cdi) +{ +} + +static inline int status_reg(struct pcd_unit *cd) +{ + return pi_read_regr(cd->pi, 1, 6); +} + +static inline int read_reg(struct pcd_unit *cd, int reg) +{ + return pi_read_regr(cd->pi, 0, reg); +} + +static inline void write_reg(struct pcd_unit *cd, int reg, int val) +{ + pi_write_regr(cd->pi, 0, reg, val); +} + +static int pcd_wait(struct pcd_unit *cd, int go, int stop, char *fun, char *msg) +{ + int j, r, e, s, p; + + j = 0; + while ((((r = status_reg(cd)) & go) || (stop && (!(r & stop)))) + && (j++ < PCD_SPIN)) + udelay(PCD_DELAY); + + if ((r & (IDE_ERR & stop)) || (j >= PCD_SPIN)) { + s = read_reg(cd, 7); + e = read_reg(cd, 1); + p = read_reg(cd, 2); + if (j >= PCD_SPIN) + e |= 0x100; + if (fun) + printk("%s: %s %s: alt=0x%x stat=0x%x err=0x%x" + " loop=%d phase=%d\n", + cd->name, fun, msg, r, s, e, j, p); + return (s << 8) + r; + } + return 0; +} + +static int pcd_command(struct pcd_unit *cd, char *cmd, int dlen, char *fun) +{ + pi_connect(cd->pi); + + write_reg(cd, 6, 0xa0 + 0x10 * cd->drive); + + if (pcd_wait(cd, IDE_BUSY | IDE_DRQ, 0, fun, "before command")) { + pi_disconnect(cd->pi); + return -1; + } + + write_reg(cd, 4, dlen % 256); + write_reg(cd, 5, dlen / 256); + write_reg(cd, 7, 0xa0); /* ATAPI packet command */ + + if (pcd_wait(cd, IDE_BUSY, IDE_DRQ, fun, "command DRQ")) { + pi_disconnect(cd->pi); + return -1; + } + + if (read_reg(cd, 2) != 1) { + printk("%s: %s: command phase error\n", cd->name, fun); + pi_disconnect(cd->pi); + return -1; + } + + pi_write_block(cd->pi, cmd, 12); + + return 0; +} + +static int pcd_completion(struct pcd_unit *cd, char *buf, char *fun) +{ + int r, d, p, n, k, j; + + r = -1; + k = 0; + j = 0; + + if (!pcd_wait(cd, IDE_BUSY, IDE_DRQ | IDE_READY | IDE_ERR, + fun, "completion")) { + r = 0; + while (read_reg(cd, 7) & IDE_DRQ) { + d = read_reg(cd, 4) + 256 * read_reg(cd, 5); + n = (d + 3) & 0xfffc; + p = read_reg(cd, 2) & 3; + + if ((p == 2) && (n > 0) && (j == 0)) { + pi_read_block(cd->pi, buf, n); + if (verbose > 1) + printk("%s: %s: Read %d bytes\n", + cd->name, fun, n); + r = 0; + j++; + } else { + if (verbose > 1) + printk + ("%s: %s: Unexpected phase %d, d=%d, k=%d\n", + cd->name, fun, p, d, k); + if ((verbose < 2) && !pcd_warned) { + pcd_warned = 1; + printk + ("%s: WARNING: ATAPI phase errors\n", + cd->name); + } + mdelay(1); + } + if (k++ > PCD_TMO) { + printk("%s: Stuck DRQ\n", cd->name); + break; + } + if (pcd_wait + (cd, IDE_BUSY, IDE_DRQ | IDE_READY | IDE_ERR, fun, + "completion")) { + r = -1; + break; + } + } + } + + pi_disconnect(cd->pi); + + return r; +} + +static void pcd_req_sense(struct pcd_unit *cd, char *fun) +{ + char rs_cmd[12] = { 0x03, 0, 0, 0, 16, 0, 0, 0, 0, 0, 0, 0 }; + char buf[16]; + int r, c; + + r = pcd_command(cd, rs_cmd, 16, "Request sense"); + mdelay(1); + if (!r) + pcd_completion(cd, buf, "Request sense"); + + cd->last_sense = -1; + c = 2; + if (!r) { + if (fun) + printk("%s: %s: Sense key: %x, ASC: %x, ASQ: %x\n", + cd->name, fun, buf[2] & 0xf, buf[12], buf[13]); + c = buf[2] & 0xf; + cd->last_sense = + c | ((buf[12] & 0xff) << 8) | ((buf[13] & 0xff) << 16); + } + if ((c == 2) || (c == 6)) + cd->changed = 1; +} + +static int pcd_atapi(struct pcd_unit *cd, char *cmd, int dlen, char *buf, char *fun) +{ + int r; + + r = pcd_command(cd, cmd, dlen, fun); + mdelay(1); + if (!r) + r = pcd_completion(cd, buf, fun); + if (r) + pcd_req_sense(cd, fun); + + return r; +} + +static int pcd_packet(struct cdrom_device_info *cdi, struct packet_command *cgc) +{ + return pcd_atapi(cdi->handle, cgc->cmd, cgc->buflen, cgc->buffer, + "generic packet"); +} + +#define DBMSG(msg) ((verbose>1)?(msg):NULL) + +static int pcd_media_changed(struct cdrom_device_info *cdi, int slot_nr) +{ + struct pcd_unit *cd = cdi->handle; + int res = cd->changed; + if (res) + cd->changed = 0; + return res; +} + +static int pcd_lock_door(struct cdrom_device_info *cdi, int lock) +{ + char un_cmd[12] = { 0x1e, 0, 0, 0, lock, 0, 0, 0, 0, 0, 0, 0 }; + + return pcd_atapi(cdi->handle, un_cmd, 0, pcd_scratch, + lock ? "lock door" : "unlock door"); +} + +static int pcd_tray_move(struct cdrom_device_info *cdi, int position) +{ + char ej_cmd[12] = { 0x1b, 0, 0, 0, 3 - position, 0, 0, 0, 0, 0, 0, 0 }; + + return pcd_atapi(cdi->handle, ej_cmd, 0, pcd_scratch, + position ? "eject" : "close tray"); +} + +static void pcd_sleep(int cs) +{ + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(cs); +} + +static int pcd_reset(struct pcd_unit *cd) +{ + int i, k, flg; + int expect[5] = { 1, 1, 1, 0x14, 0xeb }; + + pi_connect(cd->pi); + write_reg(cd, 6, 0xa0 + 0x10 * cd->drive); + write_reg(cd, 7, 8); + + pcd_sleep(20 * HZ / 1000); /* delay a bit */ + + k = 0; + while ((k++ < PCD_RESET_TMO) && (status_reg(cd) & IDE_BUSY)) + pcd_sleep(HZ / 10); + + flg = 1; + for (i = 0; i < 5; i++) + flg &= (read_reg(cd, i + 1) == expect[i]); + + if (verbose) { + printk("%s: Reset (%d) signature = ", cd->name, k); + for (i = 0; i < 5; i++) + printk("%3x", read_reg(cd, i + 1)); + if (!flg) + printk(" (incorrect)"); + printk("\n"); + } + + pi_disconnect(cd->pi); + return flg - 1; +} + +static int pcd_drive_reset(struct cdrom_device_info *cdi) +{ + return pcd_reset(cdi->handle); +} + +static int pcd_ready_wait(struct pcd_unit *cd, int tmo) +{ + char tr_cmd[12] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + int k, p; + + k = 0; + while (k < tmo) { + cd->last_sense = 0; + pcd_atapi(cd, tr_cmd, 0, NULL, DBMSG("test unit ready")); + p = cd->last_sense; + if (!p) + return 0; + if (!(((p & 0xffff) == 0x0402) || ((p & 0xff) == 6))) + return p; + k++; + pcd_sleep(HZ); + } + return 0x000020; /* timeout */ +} + +static int pcd_drive_status(struct cdrom_device_info *cdi, int slot_nr) +{ + char rc_cmd[12] = { 0x25, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + struct pcd_unit *cd = cdi->handle; + + if (pcd_ready_wait(cd, PCD_READY_TMO)) + return CDS_DRIVE_NOT_READY; + if (pcd_atapi(cd, rc_cmd, 8, pcd_scratch, DBMSG("check media"))) + return CDS_NO_DISC; + return CDS_DISC_OK; +} + +static int pcd_identify(struct pcd_unit *cd, char *id) +{ + int k, s; + char id_cmd[12] = { 0x12, 0, 0, 0, 36, 0, 0, 0, 0, 0, 0, 0 }; + + pcd_bufblk = -1; + + s = pcd_atapi(cd, id_cmd, 36, pcd_buffer, "identify"); + + if (s) + return -1; + if ((pcd_buffer[0] & 0x1f) != 5) { + if (verbose) + printk("%s: %s is not a CD-ROM\n", + cd->name, cd->drive ? "Slave" : "Master"); + return -1; + } + memcpy(id, pcd_buffer + 16, 16); + id[16] = 0; + k = 16; + while ((k >= 0) && (id[k] <= 0x20)) { + id[k] = 0; + k--; + } + + printk("%s: %s: %s\n", cd->name, cd->drive ? "Slave" : "Master", id); + + return 0; +} + +/* + * returns 0, with id set if drive is detected + * -1, if drive detection failed + */ +static int pcd_probe(struct pcd_unit *cd, int ms, char *id) +{ + if (ms == -1) { + for (cd->drive = 0; cd->drive <= 1; cd->drive++) + if (!pcd_reset(cd) && !pcd_identify(cd, id)) + return 0; + } else { + cd->drive = ms; + if (!pcd_reset(cd) && !pcd_identify(cd, id)) + return 0; + } + return -1; +} + +static void pcd_probe_capabilities(void) +{ + int unit, r; + char buffer[32]; + char cmd[12] = { 0x5a, 1 << 3, 0x2a, 0, 0, 0, 0, 18, 0, 0, 0, 0 }; + struct pcd_unit *cd; + + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) { + if (!cd->present) + continue; + r = pcd_atapi(cd, cmd, 18, buffer, "mode sense capabilities"); + if (r) + continue; + /* we should now have the cap page */ + if ((buffer[11] & 1) == 0) + cd->info.mask |= CDC_CD_R; + if ((buffer[11] & 2) == 0) + cd->info.mask |= CDC_CD_RW; + if ((buffer[12] & 1) == 0) + cd->info.mask |= CDC_PLAY_AUDIO; + if ((buffer[14] & 1) == 0) + cd->info.mask |= CDC_LOCK; + if ((buffer[14] & 8) == 0) + cd->info.mask |= CDC_OPEN_TRAY; + if ((buffer[14] >> 6) == 0) + cd->info.mask |= CDC_CLOSE_TRAY; + } +} + +static int pcd_detect(void) +{ + char id[18]; + int k, unit; + struct pcd_unit *cd; + + printk("%s: %s version %s, major %d, nice %d\n", + name, name, PCD_VERSION, major, nice); + + k = 0; + if (pcd_drive_count == 0) { /* nothing spec'd - so autoprobe for 1 */ + cd = pcd; + if (pi_init(cd->pi, 1, -1, -1, -1, -1, -1, pcd_buffer, + PI_PCD, verbose, cd->name)) { + if (!pcd_probe(cd, -1, id) && cd->disk) { + cd->present = 1; + k++; + } else + pi_release(cd->pi); + } + } else { + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) { + int *conf = *drives[unit]; + if (!conf[D_PRT]) + continue; + if (!pi_init(cd->pi, 0, conf[D_PRT], conf[D_MOD], + conf[D_UNI], conf[D_PRO], conf[D_DLY], + pcd_buffer, PI_PCD, verbose, cd->name)) + continue; + if (!pcd_probe(cd, conf[D_SLV], id) && cd->disk) { + cd->present = 1; + k++; + } else + pi_release(cd->pi); + } + } + if (k) + return 0; + + printk("%s: No CD-ROM drive found\n", name); + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) + put_disk(cd->disk); + return -1; +} + +/* I/O request processing */ +static struct request_queue *pcd_queue; + +static void do_pcd_request(request_queue_t * q) +{ + if (pcd_busy) + return; + while (1) { + pcd_req = elv_next_request(q); + if (!pcd_req) + return; + + if (rq_data_dir(pcd_req) == READ) { + struct pcd_unit *cd = pcd_req->rq_disk->private_data; + if (cd != pcd_current) + pcd_bufblk = -1; + pcd_current = cd; + pcd_sector = pcd_req->sector; + pcd_count = pcd_req->current_nr_sectors; + pcd_buf = pcd_req->buffer; + pcd_busy = 1; + ps_set_intr(do_pcd_read, NULL, 0, nice); + return; + } else + end_request(pcd_req, 0); + } +} + +static inline void next_request(int success) +{ + unsigned long saved_flags; + + spin_lock_irqsave(&pcd_lock, saved_flags); + end_request(pcd_req, success); + pcd_busy = 0; + do_pcd_request(pcd_queue); + spin_unlock_irqrestore(&pcd_lock, saved_flags); +} + +static int pcd_ready(void) +{ + return (((status_reg(pcd_current) & (IDE_BUSY | IDE_DRQ)) == IDE_DRQ)); +} + +static void pcd_transfer(void) +{ + + while (pcd_count && (pcd_sector / 4 == pcd_bufblk)) { + int o = (pcd_sector % 4) * 512; + memcpy(pcd_buf, pcd_buffer + o, 512); + pcd_count--; + pcd_buf += 512; + pcd_sector++; + } +} + +static void pcd_start(void) +{ + int b, i; + char rd_cmd[12] = { 0xa8, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }; + + pcd_bufblk = pcd_sector / 4; + b = pcd_bufblk; + for (i = 0; i < 4; i++) { + rd_cmd[5 - i] = b & 0xff; + b = b >> 8; + } + + if (pcd_command(pcd_current, rd_cmd, 2048, "read block")) { + pcd_bufblk = -1; + next_request(0); + return; + } + + mdelay(1); + + ps_set_intr(do_pcd_read_drq, pcd_ready, PCD_TMO, nice); +} + +static void do_pcd_read(void) +{ + pcd_busy = 1; + pcd_retries = 0; + pcd_transfer(); + if (!pcd_count) { + next_request(1); + return; + } + + pi_do_claimed(pcd_current->pi, pcd_start); +} + +static void do_pcd_read_drq(void) +{ + unsigned long saved_flags; + + if (pcd_completion(pcd_current, pcd_buffer, "read block")) { + if (pcd_retries < PCD_RETRIES) { + mdelay(1); + pcd_retries++; + pi_do_claimed(pcd_current->pi, pcd_start); + return; + } + pcd_bufblk = -1; + next_request(0); + return; + } + + do_pcd_read(); + spin_lock_irqsave(&pcd_lock, saved_flags); + do_pcd_request(pcd_queue); + spin_unlock_irqrestore(&pcd_lock, saved_flags); +} + +/* the audio_ioctl stuff is adapted from sr_ioctl.c */ + +static int pcd_audio_ioctl(struct cdrom_device_info *cdi, unsigned int cmd, void *arg) +{ + struct pcd_unit *cd = cdi->handle; + + switch (cmd) { + + case CDROMREADTOCHDR: + + { + char cmd[12] = + { GPCMD_READ_TOC_PMA_ATIP, 0, 0, 0, 0, 0, 0, 0, 12, + 0, 0, 0 }; + struct cdrom_tochdr *tochdr = + (struct cdrom_tochdr *) arg; + char buffer[32]; + int r; + + r = pcd_atapi(cd, cmd, 12, buffer, "read toc header"); + + tochdr->cdth_trk0 = buffer[2]; + tochdr->cdth_trk1 = buffer[3]; + + return r ? -EIO : 0; + } + + case CDROMREADTOCENTRY: + + { + char cmd[12] = + { GPCMD_READ_TOC_PMA_ATIP, 0, 0, 0, 0, 0, 0, 0, 12, + 0, 0, 0 }; + + struct cdrom_tocentry *tocentry = + (struct cdrom_tocentry *) arg; + unsigned char buffer[32]; + int r; + + cmd[1] = + (tocentry->cdte_format == CDROM_MSF ? 0x02 : 0); + cmd[6] = tocentry->cdte_track; + + r = pcd_atapi(cd, cmd, 12, buffer, "read toc entry"); + + tocentry->cdte_ctrl = buffer[5] & 0xf; + tocentry->cdte_adr = buffer[5] >> 4; + tocentry->cdte_datamode = + (tocentry->cdte_ctrl & 0x04) ? 1 : 0; + if (tocentry->cdte_format == CDROM_MSF) { + tocentry->cdte_addr.msf.minute = buffer[9]; + tocentry->cdte_addr.msf.second = buffer[10]; + tocentry->cdte_addr.msf.frame = buffer[11]; + } else + tocentry->cdte_addr.lba = + (((((buffer[8] << 8) + buffer[9]) << 8) + + buffer[10]) << 8) + buffer[11]; + + return r ? -EIO : 0; + } + + default: + + return -ENOSYS; + } +} + +static int pcd_get_mcn(struct cdrom_device_info *cdi, struct cdrom_mcn *mcn) +{ + char cmd[12] = + { GPCMD_READ_SUBCHANNEL, 0, 0x40, 2, 0, 0, 0, 0, 24, 0, 0, 0 }; + char buffer[32]; + + if (pcd_atapi(cdi->handle, cmd, 24, buffer, "get mcn")) + return -EIO; + + memcpy(mcn->medium_catalog_number, buffer + 9, 13); + mcn->medium_catalog_number[13] = 0; + + return 0; +} + +static int __init pcd_init(void) +{ + struct pcd_unit *cd; + int unit; + + if (disable) + return -1; + + pcd_init_units(); + + if (pcd_detect()) + return -1; + + /* get the atapi capabilities page */ + pcd_probe_capabilities(); + + if (register_blkdev(major, name)) { + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) + put_disk(cd->disk); + return -1; + } + + pcd_queue = blk_init_queue(do_pcd_request, &pcd_lock); + if (!pcd_queue) { + unregister_blkdev(major, name); + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) + put_disk(cd->disk); + return -1; + } + + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) { + if (cd->present) { + register_cdrom(&cd->info); + cd->disk->private_data = cd; + cd->disk->queue = pcd_queue; + add_disk(cd->disk); + } + } + + return 0; +} + +static void __exit pcd_exit(void) +{ + struct pcd_unit *cd; + int unit; + + for (unit = 0, cd = pcd; unit < PCD_UNITS; unit++, cd++) { + if (cd->present) { + del_gendisk(cd->disk); + pi_release(cd->pi); + unregister_cdrom(&cd->info); + } + put_disk(cd->disk); + } + blk_cleanup_queue(pcd_queue); + unregister_blkdev(major, name); +} + +MODULE_LICENSE("GPL"); +module_init(pcd_init) +module_exit(pcd_exit) diff --git a/drivers/block/paride/pd.c b/drivers/block/paride/pd.c new file mode 100644 index 00000000000..202a5a74ad3 --- /dev/null +++ b/drivers/block/paride/pd.c @@ -0,0 +1,950 @@ +/* + pd.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is the high-level driver for parallel port IDE hard + drives based on chips supported by the paride module. + + By default, the driver will autoprobe for a single parallel + port IDE drive, but if their individual parameters are + specified, the driver can handle up to 4 drives. + + The behaviour of the pd driver can be altered by setting + some parameters from the insmod command line. The following + parameters are adjustable: + + drive0 These four arguments can be arrays of + drive1 1-8 integers as follows: + drive2 + drive3 <prt>,<pro>,<uni>,<mod>,<geo>,<sby>,<dly>,<slv> + + Where, + + <prt> is the base of the parallel port address for + the corresponding drive. (required) + + <pro> is the protocol number for the adapter that + supports this drive. These numbers are + logged by 'paride' when the protocol modules + are initialised. (0 if not given) + + <uni> for those adapters that support chained + devices, this is the unit selector for the + chain of devices on the given port. It should + be zero for devices that don't support chaining. + (0 if not given) + + <mod> this can be -1 to choose the best mode, or one + of the mode numbers supported by the adapter. + (-1 if not given) + + <geo> this defaults to 0 to indicate that the driver + should use the CHS geometry provided by the drive + itself. If set to 1, the driver will provide + a logical geometry with 64 heads and 32 sectors + per track, to be consistent with most SCSI + drivers. (0 if not given) + + <sby> set this to zero to disable the power saving + standby mode, if needed. (1 if not given) + + <dly> some parallel ports require the driver to + go more slowly. -1 sets a default value that + should work with the chosen protocol. Otherwise, + set this to a small integer, the larger it is + the slower the port i/o. In some cases, setting + this to zero will speed up the device. (default -1) + + <slv> IDE disks can be jumpered to master or slave. + Set this to 0 to choose the master drive, 1 to + choose the slave, -1 (the default) to choose the + first drive found. + + + major You may use this parameter to overide the + default major number (45) that this driver + will use. Be sure to change the device + name as well. + + name This parameter is a character string that + contains the name the kernel will use for this + device (in /proc output, for instance). + (default "pd") + + cluster The driver will attempt to aggregate requests + for adjacent blocks into larger multi-block + clusters. The maximum cluster size (in 512 + byte sectors) is set with this parameter. + (default 64) + + verbose This parameter controls the amount of logging + that the driver will do. Set it to 0 for + normal operation, 1 to see autoprobe progress + messages, or 2 to see additional debugging + output. (default 0) + + nice This parameter controls the driver's use of + idle CPU time, at the expense of some speed. + + If this driver is built into the kernel, you can use kernel + the following command line parameters, with the same values + as the corresponding module parameters listed above: + + pd.drive0 + pd.drive1 + pd.drive2 + pd.drive3 + pd.cluster + pd.nice + + In addition, you can use the parameter pd.disable to disable + the driver entirely. + +*/ + +/* Changes: + + 1.01 GRG 1997.01.24 Restored pd_reset() + Added eject ioctl + 1.02 GRG 1998.05.06 SMP spinlock changes, + Added slave support + 1.03 GRG 1998.06.16 Eliminate an Ugh. + 1.04 GRG 1998.08.15 Extra debugging, use HZ in loop timing + 1.05 GRG 1998.09.24 Added jumbo support + +*/ + +#define PD_VERSION "1.05" +#define PD_MAJOR 45 +#define PD_NAME "pd" +#define PD_UNITS 4 + +/* Here are things one can override from the insmod command. + Most are autoprobed by paride unless set here. Verbose is off + by default. + +*/ + +static int verbose = 0; +static int major = PD_MAJOR; +static char *name = PD_NAME; +static int cluster = 64; +static int nice = 0; +static int disable = 0; + +static int drive0[8] = { 0, 0, 0, -1, 0, 1, -1, -1 }; +static int drive1[8] = { 0, 0, 0, -1, 0, 1, -1, -1 }; +static int drive2[8] = { 0, 0, 0, -1, 0, 1, -1, -1 }; +static int drive3[8] = { 0, 0, 0, -1, 0, 1, -1, -1 }; + +static int (*drives[4])[8] = {&drive0, &drive1, &drive2, &drive3}; + +enum {D_PRT, D_PRO, D_UNI, D_MOD, D_GEO, D_SBY, D_DLY, D_SLV}; + +/* end of parameters */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/fs.h> +#include <linux/delay.h> +#include <linux/hdreg.h> +#include <linux/cdrom.h> /* for the eject ioctl */ +#include <linux/blkdev.h> +#include <linux/blkpg.h> +#include <asm/uaccess.h> +#include <linux/sched.h> +#include <linux/workqueue.h> + +static DEFINE_SPINLOCK(pd_lock); + +module_param(verbose, bool, 0); +module_param(major, int, 0); +module_param(name, charp, 0); +module_param(cluster, int, 0); +module_param(nice, int, 0); +module_param_array(drive0, int, NULL, 0); +module_param_array(drive1, int, NULL, 0); +module_param_array(drive2, int, NULL, 0); +module_param_array(drive3, int, NULL, 0); + +#include "paride.h" + +#define PD_BITS 4 + +/* numbers for "SCSI" geometry */ + +#define PD_LOG_HEADS 64 +#define PD_LOG_SECTS 32 + +#define PD_ID_OFF 54 +#define PD_ID_LEN 14 + +#define PD_MAX_RETRIES 5 +#define PD_TMO 800 /* interrupt timeout in jiffies */ +#define PD_SPIN_DEL 50 /* spin delay in micro-seconds */ + +#define PD_SPIN (1000000*PD_TMO)/(HZ*PD_SPIN_DEL) + +#define STAT_ERR 0x00001 +#define STAT_INDEX 0x00002 +#define STAT_ECC 0x00004 +#define STAT_DRQ 0x00008 +#define STAT_SEEK 0x00010 +#define STAT_WRERR 0x00020 +#define STAT_READY 0x00040 +#define STAT_BUSY 0x00080 + +#define ERR_AMNF 0x00100 +#define ERR_TK0NF 0x00200 +#define ERR_ABRT 0x00400 +#define ERR_MCR 0x00800 +#define ERR_IDNF 0x01000 +#define ERR_MC 0x02000 +#define ERR_UNC 0x04000 +#define ERR_TMO 0x10000 + +#define IDE_READ 0x20 +#define IDE_WRITE 0x30 +#define IDE_READ_VRFY 0x40 +#define IDE_INIT_DEV_PARMS 0x91 +#define IDE_STANDBY 0x96 +#define IDE_ACKCHANGE 0xdb +#define IDE_DOORLOCK 0xde +#define IDE_DOORUNLOCK 0xdf +#define IDE_IDENTIFY 0xec +#define IDE_EJECT 0xed + +#define PD_NAMELEN 8 + +struct pd_unit { + struct pi_adapter pia; /* interface to paride layer */ + struct pi_adapter *pi; + int access; /* count of active opens ... */ + int capacity; /* Size of this volume in sectors */ + int heads; /* physical geometry */ + int sectors; + int cylinders; + int can_lba; + int drive; /* master=0 slave=1 */ + int changed; /* Have we seen a disk change ? */ + int removable; /* removable media device ? */ + int standby; + int alt_geom; + char name[PD_NAMELEN]; /* pda, pdb, etc ... */ + struct gendisk *gd; +}; + +static struct pd_unit pd[PD_UNITS]; + +static char pd_scratch[512]; /* scratch block buffer */ + +static char *pd_errs[17] = { "ERR", "INDEX", "ECC", "DRQ", "SEEK", "WRERR", + "READY", "BUSY", "AMNF", "TK0NF", "ABRT", "MCR", + "IDNF", "MC", "UNC", "???", "TMO" +}; + +static inline int status_reg(struct pd_unit *disk) +{ + return pi_read_regr(disk->pi, 1, 6); +} + +static inline int read_reg(struct pd_unit *disk, int reg) +{ + return pi_read_regr(disk->pi, 0, reg); +} + +static inline void write_status(struct pd_unit *disk, int val) +{ + pi_write_regr(disk->pi, 1, 6, val); +} + +static inline void write_reg(struct pd_unit *disk, int reg, int val) +{ + pi_write_regr(disk->pi, 0, reg, val); +} + +static inline u8 DRIVE(struct pd_unit *disk) +{ + return 0xa0+0x10*disk->drive; +} + +/* ide command interface */ + +static void pd_print_error(struct pd_unit *disk, char *msg, int status) +{ + int i; + + printk("%s: %s: status = 0x%x =", disk->name, msg, status); + for (i = 0; i < 18; i++) + if (status & (1 << i)) + printk(" %s", pd_errs[i]); + printk("\n"); +} + +static void pd_reset(struct pd_unit *disk) +{ /* called only for MASTER drive */ + write_status(disk, 4); + udelay(50); + write_status(disk, 0); + udelay(250); +} + +#define DBMSG(msg) ((verbose>1)?(msg):NULL) + +static int pd_wait_for(struct pd_unit *disk, int w, char *msg) +{ /* polled wait */ + int k, r, e; + + k = 0; + while (k < PD_SPIN) { + r = status_reg(disk); + k++; + if (((r & w) == w) && !(r & STAT_BUSY)) + break; + udelay(PD_SPIN_DEL); + } + e = (read_reg(disk, 1) << 8) + read_reg(disk, 7); + if (k >= PD_SPIN) + e |= ERR_TMO; + if ((e & (STAT_ERR | ERR_TMO)) && (msg != NULL)) + pd_print_error(disk, msg, e); + return e; +} + +static void pd_send_command(struct pd_unit *disk, int n, int s, int h, int c0, int c1, int func) +{ + write_reg(disk, 6, DRIVE(disk) + h); + write_reg(disk, 1, 0); /* the IDE task file */ + write_reg(disk, 2, n); + write_reg(disk, 3, s); + write_reg(disk, 4, c0); + write_reg(disk, 5, c1); + write_reg(disk, 7, func); + + udelay(1); +} + +static void pd_ide_command(struct pd_unit *disk, int func, int block, int count) +{ + int c1, c0, h, s; + + if (disk->can_lba) { + s = block & 255; + c0 = (block >>= 8) & 255; + c1 = (block >>= 8) & 255; + h = ((block >>= 8) & 15) + 0x40; + } else { + s = (block % disk->sectors) + 1; + h = (block /= disk->sectors) % disk->heads; + c0 = (block /= disk->heads) % 256; + c1 = (block >>= 8); + } + pd_send_command(disk, count, s, h, c0, c1, func); +} + +/* The i/o request engine */ + +enum action {Fail = 0, Ok = 1, Hold, Wait}; + +static struct request *pd_req; /* current request */ +static enum action (*phase)(void); + +static void run_fsm(void); + +static void ps_tq_int( void *data); + +static DECLARE_WORK(fsm_tq, ps_tq_int, NULL); + +static void schedule_fsm(void) +{ + if (!nice) + schedule_work(&fsm_tq); + else + schedule_delayed_work(&fsm_tq, nice-1); +} + +static void ps_tq_int(void *data) +{ + run_fsm(); +} + +static enum action do_pd_io_start(void); +static enum action pd_special(void); +static enum action do_pd_read_start(void); +static enum action do_pd_write_start(void); +static enum action do_pd_read_drq(void); +static enum action do_pd_write_done(void); + +static struct request_queue *pd_queue; +static int pd_claimed; + +static struct pd_unit *pd_current; /* current request's drive */ +static PIA *pi_current; /* current request's PIA */ + +static void run_fsm(void) +{ + while (1) { + enum action res; + unsigned long saved_flags; + int stop = 0; + + if (!phase) { + pd_current = pd_req->rq_disk->private_data; + pi_current = pd_current->pi; + phase = do_pd_io_start; + } + + switch (pd_claimed) { + case 0: + pd_claimed = 1; + if (!pi_schedule_claimed(pi_current, run_fsm)) + return; + case 1: + pd_claimed = 2; + pi_current->proto->connect(pi_current); + } + + switch(res = phase()) { + case Ok: case Fail: + pi_disconnect(pi_current); + pd_claimed = 0; + phase = NULL; + spin_lock_irqsave(&pd_lock, saved_flags); + end_request(pd_req, res); + pd_req = elv_next_request(pd_queue); + if (!pd_req) + stop = 1; + spin_unlock_irqrestore(&pd_lock, saved_flags); + if (stop) + return; + case Hold: + schedule_fsm(); + return; + case Wait: + pi_disconnect(pi_current); + pd_claimed = 0; + } + } +} + +static int pd_retries = 0; /* i/o error retry count */ +static int pd_block; /* address of next requested block */ +static int pd_count; /* number of blocks still to do */ +static int pd_run; /* sectors in current cluster */ +static int pd_cmd; /* current command READ/WRITE */ +static char *pd_buf; /* buffer for request in progress */ + +static enum action do_pd_io_start(void) +{ + if (pd_req->flags & REQ_SPECIAL) { + phase = pd_special; + return pd_special(); + } + + pd_cmd = rq_data_dir(pd_req); + if (pd_cmd == READ || pd_cmd == WRITE) { + pd_block = pd_req->sector; + pd_count = pd_req->current_nr_sectors; + if (pd_block + pd_count > get_capacity(pd_req->rq_disk)) + return Fail; + pd_run = pd_req->nr_sectors; + pd_buf = pd_req->buffer; + pd_retries = 0; + if (pd_cmd == READ) + return do_pd_read_start(); + else + return do_pd_write_start(); + } + return Fail; +} + +static enum action pd_special(void) +{ + enum action (*func)(struct pd_unit *) = pd_req->special; + return func(pd_current); +} + +static int pd_next_buf(void) +{ + unsigned long saved_flags; + + pd_count--; + pd_run--; + pd_buf += 512; + pd_block++; + if (!pd_run) + return 1; + if (pd_count) + return 0; + spin_lock_irqsave(&pd_lock, saved_flags); + end_request(pd_req, 1); + pd_count = pd_req->current_nr_sectors; + pd_buf = pd_req->buffer; + spin_unlock_irqrestore(&pd_lock, saved_flags); + return 0; +} + +static unsigned long pd_timeout; + +static enum action do_pd_read_start(void) +{ + if (pd_wait_for(pd_current, STAT_READY, "do_pd_read") & STAT_ERR) { + if (pd_retries < PD_MAX_RETRIES) { + pd_retries++; + return Wait; + } + return Fail; + } + pd_ide_command(pd_current, IDE_READ, pd_block, pd_run); + phase = do_pd_read_drq; + pd_timeout = jiffies + PD_TMO; + return Hold; +} + +static enum action do_pd_write_start(void) +{ + if (pd_wait_for(pd_current, STAT_READY, "do_pd_write") & STAT_ERR) { + if (pd_retries < PD_MAX_RETRIES) { + pd_retries++; + return Wait; + } + return Fail; + } + pd_ide_command(pd_current, IDE_WRITE, pd_block, pd_run); + while (1) { + if (pd_wait_for(pd_current, STAT_DRQ, "do_pd_write_drq") & STAT_ERR) { + if (pd_retries < PD_MAX_RETRIES) { + pd_retries++; + return Wait; + } + return Fail; + } + pi_write_block(pd_current->pi, pd_buf, 512); + if (pd_next_buf()) + break; + } + phase = do_pd_write_done; + pd_timeout = jiffies + PD_TMO; + return Hold; +} + +static inline int pd_ready(void) +{ + return !(status_reg(pd_current) & STAT_BUSY); +} + +static enum action do_pd_read_drq(void) +{ + if (!pd_ready() && !time_after_eq(jiffies, pd_timeout)) + return Hold; + + while (1) { + if (pd_wait_for(pd_current, STAT_DRQ, "do_pd_read_drq") & STAT_ERR) { + if (pd_retries < PD_MAX_RETRIES) { + pd_retries++; + phase = do_pd_read_start; + return Wait; + } + return Fail; + } + pi_read_block(pd_current->pi, pd_buf, 512); + if (pd_next_buf()) + break; + } + return Ok; +} + +static enum action do_pd_write_done(void) +{ + if (!pd_ready() && !time_after_eq(jiffies, pd_timeout)) + return Hold; + + if (pd_wait_for(pd_current, STAT_READY, "do_pd_write_done") & STAT_ERR) { + if (pd_retries < PD_MAX_RETRIES) { + pd_retries++; + phase = do_pd_write_start; + return Wait; + } + return Fail; + } + return Ok; +} + +/* special io requests */ + +/* According to the ATA standard, the default CHS geometry should be + available following a reset. Some Western Digital drives come up + in a mode where only LBA addresses are accepted until the device + parameters are initialised. +*/ + +static void pd_init_dev_parms(struct pd_unit *disk) +{ + pd_wait_for(disk, 0, DBMSG("before init_dev_parms")); + pd_send_command(disk, disk->sectors, 0, disk->heads - 1, 0, 0, + IDE_INIT_DEV_PARMS); + udelay(300); + pd_wait_for(disk, 0, "Initialise device parameters"); +} + +static enum action pd_door_lock(struct pd_unit *disk) +{ + if (!(pd_wait_for(disk, STAT_READY, "Lock") & STAT_ERR)) { + pd_send_command(disk, 1, 0, 0, 0, 0, IDE_DOORLOCK); + pd_wait_for(disk, STAT_READY, "Lock done"); + } + return Ok; +} + +static enum action pd_door_unlock(struct pd_unit *disk) +{ + if (!(pd_wait_for(disk, STAT_READY, "Lock") & STAT_ERR)) { + pd_send_command(disk, 1, 0, 0, 0, 0, IDE_DOORUNLOCK); + pd_wait_for(disk, STAT_READY, "Lock done"); + } + return Ok; +} + +static enum action pd_eject(struct pd_unit *disk) +{ + pd_wait_for(disk, 0, DBMSG("before unlock on eject")); + pd_send_command(disk, 1, 0, 0, 0, 0, IDE_DOORUNLOCK); + pd_wait_for(disk, 0, DBMSG("after unlock on eject")); + pd_wait_for(disk, 0, DBMSG("before eject")); + pd_send_command(disk, 0, 0, 0, 0, 0, IDE_EJECT); + pd_wait_for(disk, 0, DBMSG("after eject")); + return Ok; +} + +static enum action pd_media_check(struct pd_unit *disk) +{ + int r = pd_wait_for(disk, STAT_READY, DBMSG("before media_check")); + if (!(r & STAT_ERR)) { + pd_send_command(disk, 1, 1, 0, 0, 0, IDE_READ_VRFY); + r = pd_wait_for(disk, STAT_READY, DBMSG("RDY after READ_VRFY")); + } else + disk->changed = 1; /* say changed if other error */ + if (r & ERR_MC) { + disk->changed = 1; + pd_send_command(disk, 1, 0, 0, 0, 0, IDE_ACKCHANGE); + pd_wait_for(disk, STAT_READY, DBMSG("RDY after ACKCHANGE")); + pd_send_command(disk, 1, 1, 0, 0, 0, IDE_READ_VRFY); + r = pd_wait_for(disk, STAT_READY, DBMSG("RDY after VRFY")); + } + return Ok; +} + +static void pd_standby_off(struct pd_unit *disk) +{ + pd_wait_for(disk, 0, DBMSG("before STANDBY")); + pd_send_command(disk, 0, 0, 0, 0, 0, IDE_STANDBY); + pd_wait_for(disk, 0, DBMSG("after STANDBY")); +} + +static enum action pd_identify(struct pd_unit *disk) +{ + int j; + char id[PD_ID_LEN + 1]; + +/* WARNING: here there may be dragons. reset() applies to both drives, + but we call it only on probing the MASTER. This should allow most + common configurations to work, but be warned that a reset can clear + settings on the SLAVE drive. +*/ + + if (disk->drive == 0) + pd_reset(disk); + + write_reg(disk, 6, DRIVE(disk)); + pd_wait_for(disk, 0, DBMSG("before IDENT")); + pd_send_command(disk, 1, 0, 0, 0, 0, IDE_IDENTIFY); + + if (pd_wait_for(disk, STAT_DRQ, DBMSG("IDENT DRQ")) & STAT_ERR) + return Fail; + pi_read_block(disk->pi, pd_scratch, 512); + disk->can_lba = pd_scratch[99] & 2; + disk->sectors = le16_to_cpu(*(u16 *) (pd_scratch + 12)); + disk->heads = le16_to_cpu(*(u16 *) (pd_scratch + 6)); + disk->cylinders = le16_to_cpu(*(u16 *) (pd_scratch + 2)); + if (disk->can_lba) + disk->capacity = le32_to_cpu(*(u32 *) (pd_scratch + 120)); + else + disk->capacity = disk->sectors * disk->heads * disk->cylinders; + + for (j = 0; j < PD_ID_LEN; j++) + id[j ^ 1] = pd_scratch[j + PD_ID_OFF]; + j = PD_ID_LEN - 1; + while ((j >= 0) && (id[j] <= 0x20)) + j--; + j++; + id[j] = 0; + + disk->removable = pd_scratch[0] & 0x80; + + printk("%s: %s, %s, %d blocks [%dM], (%d/%d/%d), %s media\n", + disk->name, id, + disk->drive ? "slave" : "master", + disk->capacity, disk->capacity / 2048, + disk->cylinders, disk->heads, disk->sectors, + disk->removable ? "removable" : "fixed"); + + if (disk->capacity) + pd_init_dev_parms(disk); + if (!disk->standby) + pd_standby_off(disk); + + return Ok; +} + +/* end of io request engine */ + +static void do_pd_request(request_queue_t * q) +{ + if (pd_req) + return; + pd_req = elv_next_request(q); + if (!pd_req) + return; + + schedule_fsm(); +} + +static int pd_special_command(struct pd_unit *disk, + enum action (*func)(struct pd_unit *disk)) +{ + DECLARE_COMPLETION(wait); + struct request rq; + int err = 0; + + memset(&rq, 0, sizeof(rq)); + rq.errors = 0; + rq.rq_status = RQ_ACTIVE; + rq.rq_disk = disk->gd; + rq.ref_count = 1; + rq.waiting = &wait; + rq.end_io = blk_end_sync_rq; + blk_insert_request(disk->gd->queue, &rq, 0, func, 0); + wait_for_completion(&wait); + rq.waiting = NULL; + if (rq.errors) + err = -EIO; + blk_put_request(&rq); + return err; +} + +/* kernel glue structures */ + +static int pd_open(struct inode *inode, struct file *file) +{ + struct pd_unit *disk = inode->i_bdev->bd_disk->private_data; + + disk->access++; + + if (disk->removable) { + pd_special_command(disk, pd_media_check); + pd_special_command(disk, pd_door_lock); + } + return 0; +} + +static int pd_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + struct pd_unit *disk = inode->i_bdev->bd_disk->private_data; + struct hd_geometry __user *geo = (struct hd_geometry __user *) arg; + struct hd_geometry g; + + switch (cmd) { + case CDROMEJECT: + if (disk->access == 1) + pd_special_command(disk, pd_eject); + return 0; + case HDIO_GETGEO: + if (disk->alt_geom) { + g.heads = PD_LOG_HEADS; + g.sectors = PD_LOG_SECTS; + g.cylinders = disk->capacity / (g.heads * g.sectors); + } else { + g.heads = disk->heads; + g.sectors = disk->sectors; + g.cylinders = disk->cylinders; + } + g.start = get_start_sect(inode->i_bdev); + if (copy_to_user(geo, &g, sizeof(struct hd_geometry))) + return -EFAULT; + return 0; + default: + return -EINVAL; + } +} + +static int pd_release(struct inode *inode, struct file *file) +{ + struct pd_unit *disk = inode->i_bdev->bd_disk->private_data; + + if (!--disk->access && disk->removable) + pd_special_command(disk, pd_door_unlock); + + return 0; +} + +static int pd_check_media(struct gendisk *p) +{ + struct pd_unit *disk = p->private_data; + int r; + if (!disk->removable) + return 0; + pd_special_command(disk, pd_media_check); + r = disk->changed; + disk->changed = 0; + return r; +} + +static int pd_revalidate(struct gendisk *p) +{ + struct pd_unit *disk = p->private_data; + if (pd_special_command(disk, pd_identify) == 0) + set_capacity(p, disk->capacity); + else + set_capacity(p, 0); + return 0; +} + +static struct block_device_operations pd_fops = { + .owner = THIS_MODULE, + .open = pd_open, + .release = pd_release, + .ioctl = pd_ioctl, + .media_changed = pd_check_media, + .revalidate_disk= pd_revalidate +}; + +/* probing */ + +static void pd_probe_drive(struct pd_unit *disk) +{ + struct gendisk *p = alloc_disk(1 << PD_BITS); + if (!p) + return; + strcpy(p->disk_name, disk->name); + p->fops = &pd_fops; + p->major = major; + p->first_minor = (disk - pd) << PD_BITS; + disk->gd = p; + p->private_data = disk; + p->queue = pd_queue; + + if (disk->drive == -1) { + for (disk->drive = 0; disk->drive <= 1; disk->drive++) + if (pd_special_command(disk, pd_identify) == 0) + return; + } else if (pd_special_command(disk, pd_identify) == 0) + return; + disk->gd = NULL; + put_disk(p); +} + +static int pd_detect(void) +{ + int found = 0, unit, pd_drive_count = 0; + struct pd_unit *disk; + + for (unit = 0; unit < PD_UNITS; unit++) { + int *parm = *drives[unit]; + struct pd_unit *disk = pd + unit; + disk->pi = &disk->pia; + disk->access = 0; + disk->changed = 1; + disk->capacity = 0; + disk->drive = parm[D_SLV]; + snprintf(disk->name, PD_NAMELEN, "%s%c", name, 'a'+unit); + disk->alt_geom = parm[D_GEO]; + disk->standby = parm[D_SBY]; + if (parm[D_PRT]) + pd_drive_count++; + } + + if (pd_drive_count == 0) { /* nothing spec'd - so autoprobe for 1 */ + disk = pd; + if (pi_init(disk->pi, 1, -1, -1, -1, -1, -1, pd_scratch, + PI_PD, verbose, disk->name)) { + pd_probe_drive(disk); + if (!disk->gd) + pi_release(disk->pi); + } + + } else { + for (unit = 0, disk = pd; unit < PD_UNITS; unit++, disk++) { + int *parm = *drives[unit]; + if (!parm[D_PRT]) + continue; + if (pi_init(disk->pi, 0, parm[D_PRT], parm[D_MOD], + parm[D_UNI], parm[D_PRO], parm[D_DLY], + pd_scratch, PI_PD, verbose, disk->name)) { + pd_probe_drive(disk); + if (!disk->gd) + pi_release(disk->pi); + } + } + } + for (unit = 0, disk = pd; unit < PD_UNITS; unit++, disk++) { + if (disk->gd) { + set_capacity(disk->gd, disk->capacity); + add_disk(disk->gd); + found = 1; + } + } + if (!found) + printk("%s: no valid drive found\n", name); + return found; +} + +static int __init pd_init(void) +{ + if (disable) + goto out1; + + pd_queue = blk_init_queue(do_pd_request, &pd_lock); + if (!pd_queue) + goto out1; + + blk_queue_max_sectors(pd_queue, cluster); + + if (register_blkdev(major, name)) + goto out2; + + printk("%s: %s version %s, major %d, cluster %d, nice %d\n", + name, name, PD_VERSION, major, cluster, nice); + if (!pd_detect()) + goto out3; + + return 0; + +out3: + unregister_blkdev(major, name); +out2: + blk_cleanup_queue(pd_queue); +out1: + return -ENODEV; +} + +static void __exit pd_exit(void) +{ + struct pd_unit *disk; + int unit; + unregister_blkdev(major, name); + for (unit = 0, disk = pd; unit < PD_UNITS; unit++, disk++) { + struct gendisk *p = disk->gd; + if (p) { + disk->gd = NULL; + del_gendisk(p); + put_disk(p); + pi_release(disk->pi); + } + } + blk_cleanup_queue(pd_queue); +} + +MODULE_LICENSE("GPL"); +module_init(pd_init) +module_exit(pd_exit) diff --git a/drivers/block/paride/pf.c b/drivers/block/paride/pf.c new file mode 100644 index 00000000000..060b1f2a91d --- /dev/null +++ b/drivers/block/paride/pf.c @@ -0,0 +1,982 @@ +/* + pf.c (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is the high-level driver for parallel port ATAPI disk + drives based on chips supported by the paride module. + + By default, the driver will autoprobe for a single parallel + port ATAPI disk drive, but if their individual parameters are + specified, the driver can handle up to 4 drives. + + The behaviour of the pf driver can be altered by setting + some parameters from the insmod command line. The following + parameters are adjustable: + + drive0 These four arguments can be arrays of + drive1 1-7 integers as follows: + drive2 + drive3 <prt>,<pro>,<uni>,<mod>,<slv>,<lun>,<dly> + + Where, + + <prt> is the base of the parallel port address for + the corresponding drive. (required) + + <pro> is the protocol number for the adapter that + supports this drive. These numbers are + logged by 'paride' when the protocol modules + are initialised. (0 if not given) + + <uni> for those adapters that support chained + devices, this is the unit selector for the + chain of devices on the given port. It should + be zero for devices that don't support chaining. + (0 if not given) + + <mod> this can be -1 to choose the best mode, or one + of the mode numbers supported by the adapter. + (-1 if not given) + + <slv> ATAPI CDroms can be jumpered to master or slave. + Set this to 0 to choose the master drive, 1 to + choose the slave, -1 (the default) to choose the + first drive found. + + <lun> Some ATAPI devices support multiple LUNs. + One example is the ATAPI PD/CD drive from + Matshita/Panasonic. This device has a + CD drive on LUN 0 and a PD drive on LUN 1. + By default, the driver will search for the + first LUN with a supported device. Set + this parameter to force it to use a specific + LUN. (default -1) + + <dly> some parallel ports require the driver to + go more slowly. -1 sets a default value that + should work with the chosen protocol. Otherwise, + set this to a small integer, the larger it is + the slower the port i/o. In some cases, setting + this to zero will speed up the device. (default -1) + + major You may use this parameter to overide the + default major number (47) that this driver + will use. Be sure to change the device + name as well. + + name This parameter is a character string that + contains the name the kernel will use for this + device (in /proc output, for instance). + (default "pf"). + + cluster The driver will attempt to aggregate requests + for adjacent blocks into larger multi-block + clusters. The maximum cluster size (in 512 + byte sectors) is set with this parameter. + (default 64) + + verbose This parameter controls the amount of logging + that the driver will do. Set it to 0 for + normal operation, 1 to see autoprobe progress + messages, or 2 to see additional debugging + output. (default 0) + + nice This parameter controls the driver's use of + idle CPU time, at the expense of some speed. + + If this driver is built into the kernel, you can use the + following command line parameters, with the same values + as the corresponding module parameters listed above: + + pf.drive0 + pf.drive1 + pf.drive2 + pf.drive3 + pf.cluster + pf.nice + + In addition, you can use the parameter pf.disable to disable + the driver entirely. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.03 Changes for SMP. Eliminate sti(). + Fix for drives that don't clear STAT_ERR + until after next CDB delivered. + Small change in pf_completion to round + up transfer size. + 1.02 GRG 1998.06.16 Eliminated an Ugh + 1.03 GRG 1998.08.16 Use HZ in loop timings, extra debugging + 1.04 GRG 1998.09.24 Added jumbo support + +*/ + +#define PF_VERSION "1.04" +#define PF_MAJOR 47 +#define PF_NAME "pf" +#define PF_UNITS 4 + +/* Here are things one can override from the insmod command. + Most are autoprobed by paride unless set here. Verbose is off + by default. + +*/ + +static int verbose = 0; +static int major = PF_MAJOR; +static char *name = PF_NAME; +static int cluster = 64; +static int nice = 0; +static int disable = 0; + +static int drive0[7] = { 0, 0, 0, -1, -1, -1, -1 }; +static int drive1[7] = { 0, 0, 0, -1, -1, -1, -1 }; +static int drive2[7] = { 0, 0, 0, -1, -1, -1, -1 }; +static int drive3[7] = { 0, 0, 0, -1, -1, -1, -1 }; + +static int (*drives[4])[7] = {&drive0, &drive1, &drive2, &drive3}; +static int pf_drive_count; + +enum {D_PRT, D_PRO, D_UNI, D_MOD, D_SLV, D_LUN, D_DLY}; + +/* end of parameters */ + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/fs.h> +#include <linux/delay.h> +#include <linux/hdreg.h> +#include <linux/cdrom.h> +#include <linux/spinlock.h> +#include <linux/blkdev.h> +#include <linux/blkpg.h> +#include <asm/uaccess.h> + +static spinlock_t pf_spin_lock; + +module_param(verbose, bool, 0644); +module_param(major, int, 0); +module_param(name, charp, 0); +module_param(cluster, int, 0); +module_param(nice, int, 0); +module_param_array(drive0, int, NULL, 0); +module_param_array(drive1, int, NULL, 0); +module_param_array(drive2, int, NULL, 0); +module_param_array(drive3, int, NULL, 0); + +#include "paride.h" +#include "pseudo.h" + +/* constants for faking geometry numbers */ + +#define PF_FD_MAX 8192 /* use FD geometry under this size */ +#define PF_FD_HDS 2 +#define PF_FD_SPT 18 +#define PF_HD_HDS 64 +#define PF_HD_SPT 32 + +#define PF_MAX_RETRIES 5 +#define PF_TMO 800 /* interrupt timeout in jiffies */ +#define PF_SPIN_DEL 50 /* spin delay in micro-seconds */ + +#define PF_SPIN (1000000*PF_TMO)/(HZ*PF_SPIN_DEL) + +#define STAT_ERR 0x00001 +#define STAT_INDEX 0x00002 +#define STAT_ECC 0x00004 +#define STAT_DRQ 0x00008 +#define STAT_SEEK 0x00010 +#define STAT_WRERR 0x00020 +#define STAT_READY 0x00040 +#define STAT_BUSY 0x00080 + +#define ATAPI_REQ_SENSE 0x03 +#define ATAPI_LOCK 0x1e +#define ATAPI_DOOR 0x1b +#define ATAPI_MODE_SENSE 0x5a +#define ATAPI_CAPACITY 0x25 +#define ATAPI_IDENTIFY 0x12 +#define ATAPI_READ_10 0x28 +#define ATAPI_WRITE_10 0x2a + +static int pf_open(struct inode *inode, struct file *file); +static void do_pf_request(request_queue_t * q); +static int pf_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg); + +static int pf_release(struct inode *inode, struct file *file); + +static int pf_detect(void); +static void do_pf_read(void); +static void do_pf_read_start(void); +static void do_pf_write(void); +static void do_pf_write_start(void); +static void do_pf_read_drq(void); +static void do_pf_write_done(void); + +#define PF_NM 0 +#define PF_RO 1 +#define PF_RW 2 + +#define PF_NAMELEN 8 + +struct pf_unit { + struct pi_adapter pia; /* interface to paride layer */ + struct pi_adapter *pi; + int removable; /* removable media device ? */ + int media_status; /* media present ? WP ? */ + int drive; /* drive */ + int lun; + int access; /* count of active opens ... */ + int present; /* device present ? */ + char name[PF_NAMELEN]; /* pf0, pf1, ... */ + struct gendisk *disk; +}; + +static struct pf_unit units[PF_UNITS]; + +static int pf_identify(struct pf_unit *pf); +static void pf_lock(struct pf_unit *pf, int func); +static void pf_eject(struct pf_unit *pf); +static int pf_check_media(struct gendisk *disk); + +static char pf_scratch[512]; /* scratch block buffer */ + +/* the variables below are used mainly in the I/O request engine, which + processes only one request at a time. +*/ + +static int pf_retries = 0; /* i/o error retry count */ +static int pf_busy = 0; /* request being processed ? */ +static struct request *pf_req; /* current request */ +static int pf_block; /* address of next requested block */ +static int pf_count; /* number of blocks still to do */ +static int pf_run; /* sectors in current cluster */ +static int pf_cmd; /* current command READ/WRITE */ +static struct pf_unit *pf_current;/* unit of current request */ +static int pf_mask; /* stopper for pseudo-int */ +static char *pf_buf; /* buffer for request in progress */ + +/* kernel glue structures */ + +static struct block_device_operations pf_fops = { + .owner = THIS_MODULE, + .open = pf_open, + .release = pf_release, + .ioctl = pf_ioctl, + .media_changed = pf_check_media, +}; + +static void __init pf_init_units(void) +{ + struct pf_unit *pf; + int unit; + + pf_drive_count = 0; + for (unit = 0, pf = units; unit < PF_UNITS; unit++, pf++) { + struct gendisk *disk = alloc_disk(1); + if (!disk) + continue; + pf->disk = disk; + pf->pi = &pf->pia; + pf->media_status = PF_NM; + pf->drive = (*drives[unit])[D_SLV]; + pf->lun = (*drives[unit])[D_LUN]; + snprintf(pf->name, PF_NAMELEN, "%s%d", name, unit); + disk->major = major; + disk->first_minor = unit; + strcpy(disk->disk_name, pf->name); + disk->fops = &pf_fops; + if (!(*drives[unit])[D_PRT]) + pf_drive_count++; + } +} + +static int pf_open(struct inode *inode, struct file *file) +{ + struct pf_unit *pf = inode->i_bdev->bd_disk->private_data; + + pf_identify(pf); + + if (pf->media_status == PF_NM) + return -ENODEV; + + if ((pf->media_status == PF_RO) && (file->f_mode & 2)) + return -EROFS; + + pf->access++; + if (pf->removable) + pf_lock(pf, 1); + + return 0; +} + +static int pf_ioctl(struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg) +{ + struct pf_unit *pf = inode->i_bdev->bd_disk->private_data; + struct hd_geometry __user *geo = (struct hd_geometry __user *) arg; + struct hd_geometry g; + sector_t capacity; + + if (cmd == CDROMEJECT) { + if (pf->access == 1) { + pf_eject(pf); + return 0; + } + return -EBUSY; + } + if (cmd != HDIO_GETGEO) + return -EINVAL; + capacity = get_capacity(pf->disk); + if (capacity < PF_FD_MAX) { + g.cylinders = sector_div(capacity, PF_FD_HDS * PF_FD_SPT); + g.heads = PF_FD_HDS; + g.sectors = PF_FD_SPT; + } else { + g.cylinders = sector_div(capacity, PF_HD_HDS * PF_HD_SPT); + g.heads = PF_HD_HDS; + g.sectors = PF_HD_SPT; + } + if (copy_to_user(geo, &g, sizeof(g))) + return -EFAULT; + return 0; +} + +static int pf_release(struct inode *inode, struct file *file) +{ + struct pf_unit *pf = inode->i_bdev->bd_disk->private_data; + + if (pf->access <= 0) + return -EINVAL; + + pf->access--; + + if (!pf->access && pf->removable) + pf_lock(pf, 0); + + return 0; + +} + +static int pf_check_media(struct gendisk *disk) +{ + return 1; +} + +static inline int status_reg(struct pf_unit *pf) +{ + return pi_read_regr(pf->pi, 1, 6); +} + +static inline int read_reg(struct pf_unit *pf, int reg) +{ + return pi_read_regr(pf->pi, 0, reg); +} + +static inline void write_reg(struct pf_unit *pf, int reg, int val) +{ + pi_write_regr(pf->pi, 0, reg, val); +} + +static int pf_wait(struct pf_unit *pf, int go, int stop, char *fun, char *msg) +{ + int j, r, e, s, p; + + j = 0; + while ((((r = status_reg(pf)) & go) || (stop && (!(r & stop)))) + && (j++ < PF_SPIN)) + udelay(PF_SPIN_DEL); + + if ((r & (STAT_ERR & stop)) || (j >= PF_SPIN)) { + s = read_reg(pf, 7); + e = read_reg(pf, 1); + p = read_reg(pf, 2); + if (j >= PF_SPIN) + e |= 0x100; + if (fun) + printk("%s: %s %s: alt=0x%x stat=0x%x err=0x%x" + " loop=%d phase=%d\n", + pf->name, fun, msg, r, s, e, j, p); + return (e << 8) + s; + } + return 0; +} + +static int pf_command(struct pf_unit *pf, char *cmd, int dlen, char *fun) +{ + pi_connect(pf->pi); + + write_reg(pf, 6, 0xa0+0x10*pf->drive); + + if (pf_wait(pf, STAT_BUSY | STAT_DRQ, 0, fun, "before command")) { + pi_disconnect(pf->pi); + return -1; + } + + write_reg(pf, 4, dlen % 256); + write_reg(pf, 5, dlen / 256); + write_reg(pf, 7, 0xa0); /* ATAPI packet command */ + + if (pf_wait(pf, STAT_BUSY, STAT_DRQ, fun, "command DRQ")) { + pi_disconnect(pf->pi); + return -1; + } + + if (read_reg(pf, 2) != 1) { + printk("%s: %s: command phase error\n", pf->name, fun); + pi_disconnect(pf->pi); + return -1; + } + + pi_write_block(pf->pi, cmd, 12); + + return 0; +} + +static int pf_completion(struct pf_unit *pf, char *buf, char *fun) +{ + int r, s, n; + + r = pf_wait(pf, STAT_BUSY, STAT_DRQ | STAT_READY | STAT_ERR, + fun, "completion"); + + if ((read_reg(pf, 2) & 2) && (read_reg(pf, 7) & STAT_DRQ)) { + n = (((read_reg(pf, 4) + 256 * read_reg(pf, 5)) + + 3) & 0xfffc); + pi_read_block(pf->pi, buf, n); + } + + s = pf_wait(pf, STAT_BUSY, STAT_READY | STAT_ERR, fun, "data done"); + + pi_disconnect(pf->pi); + + return (r ? r : s); +} + +static void pf_req_sense(struct pf_unit *pf, int quiet) +{ + char rs_cmd[12] = + { ATAPI_REQ_SENSE, pf->lun << 5, 0, 0, 16, 0, 0, 0, 0, 0, 0, 0 }; + char buf[16]; + int r; + + r = pf_command(pf, rs_cmd, 16, "Request sense"); + mdelay(1); + if (!r) + pf_completion(pf, buf, "Request sense"); + + if ((!r) && (!quiet)) + printk("%s: Sense key: %x, ASC: %x, ASQ: %x\n", + pf->name, buf[2] & 0xf, buf[12], buf[13]); +} + +static int pf_atapi(struct pf_unit *pf, char *cmd, int dlen, char *buf, char *fun) +{ + int r; + + r = pf_command(pf, cmd, dlen, fun); + mdelay(1); + if (!r) + r = pf_completion(pf, buf, fun); + if (r) + pf_req_sense(pf, !fun); + + return r; +} + +#define DBMSG(msg) ((verbose>1)?(msg):NULL) + +static void pf_lock(struct pf_unit *pf, int func) +{ + char lo_cmd[12] = { ATAPI_LOCK, pf->lun << 5, 0, 0, func, 0, 0, 0, 0, 0, 0, 0 }; + + pf_atapi(pf, lo_cmd, 0, pf_scratch, func ? "unlock" : "lock"); +} + +static void pf_eject(struct pf_unit *pf) +{ + char ej_cmd[12] = { ATAPI_DOOR, pf->lun << 5, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0 }; + + pf_lock(pf, 0); + pf_atapi(pf, ej_cmd, 0, pf_scratch, "eject"); +} + +#define PF_RESET_TMO 30 /* in tenths of a second */ + +static void pf_sleep(int cs) +{ + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(cs); +} + +/* the ATAPI standard actually specifies the contents of all 7 registers + after a reset, but the specification is ambiguous concerning the last + two bytes, and different drives interpret the standard differently. + */ + +static int pf_reset(struct pf_unit *pf) +{ + int i, k, flg; + int expect[5] = { 1, 1, 1, 0x14, 0xeb }; + + pi_connect(pf->pi); + write_reg(pf, 6, 0xa0+0x10*pf->drive); + write_reg(pf, 7, 8); + + pf_sleep(20 * HZ / 1000); + + k = 0; + while ((k++ < PF_RESET_TMO) && (status_reg(pf) & STAT_BUSY)) + pf_sleep(HZ / 10); + + flg = 1; + for (i = 0; i < 5; i++) + flg &= (read_reg(pf, i + 1) == expect[i]); + + if (verbose) { + printk("%s: Reset (%d) signature = ", pf->name, k); + for (i = 0; i < 5; i++) + printk("%3x", read_reg(pf, i + 1)); + if (!flg) + printk(" (incorrect)"); + printk("\n"); + } + + pi_disconnect(pf->pi); + return flg - 1; +} + +static void pf_mode_sense(struct pf_unit *pf) +{ + char ms_cmd[12] = + { ATAPI_MODE_SENSE, pf->lun << 5, 0, 0, 0, 0, 0, 0, 8, 0, 0, 0 }; + char buf[8]; + + pf_atapi(pf, ms_cmd, 8, buf, DBMSG("mode sense")); + pf->media_status = PF_RW; + if (buf[3] & 0x80) + pf->media_status = PF_RO; +} + +static void xs(char *buf, char *targ, int offs, int len) +{ + int j, k, l; + + j = 0; + l = 0; + for (k = 0; k < len; k++) + if ((buf[k + offs] != 0x20) || (buf[k + offs] != l)) + l = targ[j++] = buf[k + offs]; + if (l == 0x20) + j--; + targ[j] = 0; +} + +static int xl(char *buf, int offs) +{ + int v, k; + + v = 0; + for (k = 0; k < 4; k++) + v = v * 256 + (buf[k + offs] & 0xff); + return v; +} + +static void pf_get_capacity(struct pf_unit *pf) +{ + char rc_cmd[12] = { ATAPI_CAPACITY, pf->lun << 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + char buf[8]; + int bs; + + if (pf_atapi(pf, rc_cmd, 8, buf, DBMSG("get capacity"))) { + pf->media_status = PF_NM; + return; + } + set_capacity(pf->disk, xl(buf, 0) + 1); + bs = xl(buf, 4); + if (bs != 512) { + set_capacity(pf->disk, 0); + if (verbose) + printk("%s: Drive %d, LUN %d," + " unsupported block size %d\n", + pf->name, pf->drive, pf->lun, bs); + } +} + +static int pf_identify(struct pf_unit *pf) +{ + int dt, s; + char *ms[2] = { "master", "slave" }; + char mf[10], id[18]; + char id_cmd[12] = + { ATAPI_IDENTIFY, pf->lun << 5, 0, 0, 36, 0, 0, 0, 0, 0, 0, 0 }; + char buf[36]; + + s = pf_atapi(pf, id_cmd, 36, buf, "identify"); + if (s) + return -1; + + dt = buf[0] & 0x1f; + if ((dt != 0) && (dt != 7)) { + if (verbose) + printk("%s: Drive %d, LUN %d, unsupported type %d\n", + pf->name, pf->drive, pf->lun, dt); + return -1; + } + + xs(buf, mf, 8, 8); + xs(buf, id, 16, 16); + + pf->removable = (buf[1] & 0x80); + + pf_mode_sense(pf); + pf_mode_sense(pf); + pf_mode_sense(pf); + + pf_get_capacity(pf); + + printk("%s: %s %s, %s LUN %d, type %d", + pf->name, mf, id, ms[pf->drive], pf->lun, dt); + if (pf->removable) + printk(", removable"); + if (pf->media_status == PF_NM) + printk(", no media\n"); + else { + if (pf->media_status == PF_RO) + printk(", RO"); + printk(", %llu blocks\n", + (unsigned long long)get_capacity(pf->disk)); + } + return 0; +} + +/* returns 0, with id set if drive is detected + -1, if drive detection failed +*/ +static int pf_probe(struct pf_unit *pf) +{ + if (pf->drive == -1) { + for (pf->drive = 0; pf->drive <= 1; pf->drive++) + if (!pf_reset(pf)) { + if (pf->lun != -1) + return pf_identify(pf); + else + for (pf->lun = 0; pf->lun < 8; pf->lun++) + if (!pf_identify(pf)) + return 0; + } + } else { + if (pf_reset(pf)) + return -1; + if (pf->lun != -1) + return pf_identify(pf); + for (pf->lun = 0; pf->lun < 8; pf->lun++) + if (!pf_identify(pf)) + return 0; + } + return -1; +} + +static int pf_detect(void) +{ + struct pf_unit *pf = units; + int k, unit; + + printk("%s: %s version %s, major %d, cluster %d, nice %d\n", + name, name, PF_VERSION, major, cluster, nice); + + k = 0; + if (pf_drive_count == 0) { + if (pi_init(pf->pi, 1, -1, -1, -1, -1, -1, pf_scratch, PI_PF, + verbose, pf->name)) { + if (!pf_probe(pf) && pf->disk) { + pf->present = 1; + k++; + } else + pi_release(pf->pi); + } + + } else + for (unit = 0; unit < PF_UNITS; unit++, pf++) { + int *conf = *drives[unit]; + if (!conf[D_PRT]) + continue; + if (pi_init(pf->pi, 0, conf[D_PRT], conf[D_MOD], + conf[D_UNI], conf[D_PRO], conf[D_DLY], + pf_scratch, PI_PF, verbose, pf->name)) { + if (!pf_probe(pf) && pf->disk) { + pf->present = 1; + k++; + } else + pi_release(pf->pi); + } + } + if (k) + return 0; + + printk("%s: No ATAPI disk detected\n", name); + for (pf = units, unit = 0; unit < PF_UNITS; pf++, unit++) + put_disk(pf->disk); + return -1; +} + +/* The i/o request engine */ + +static int pf_start(struct pf_unit *pf, int cmd, int b, int c) +{ + int i; + char io_cmd[12] = { cmd, pf->lun << 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + + for (i = 0; i < 4; i++) { + io_cmd[5 - i] = b & 0xff; + b = b >> 8; + } + + io_cmd[8] = c & 0xff; + io_cmd[7] = (c >> 8) & 0xff; + + i = pf_command(pf, io_cmd, c * 512, "start i/o"); + + mdelay(1); + + return i; +} + +static int pf_ready(void) +{ + return (((status_reg(pf_current) & (STAT_BUSY | pf_mask)) == pf_mask)); +} + +static struct request_queue *pf_queue; + +static void do_pf_request(request_queue_t * q) +{ + if (pf_busy) + return; +repeat: + pf_req = elv_next_request(q); + if (!pf_req) + return; + + pf_current = pf_req->rq_disk->private_data; + pf_block = pf_req->sector; + pf_run = pf_req->nr_sectors; + pf_count = pf_req->current_nr_sectors; + + if (pf_block + pf_count > get_capacity(pf_req->rq_disk)) { + end_request(pf_req, 0); + goto repeat; + } + + pf_cmd = rq_data_dir(pf_req); + pf_buf = pf_req->buffer; + pf_retries = 0; + + pf_busy = 1; + if (pf_cmd == READ) + pi_do_claimed(pf_current->pi, do_pf_read); + else if (pf_cmd == WRITE) + pi_do_claimed(pf_current->pi, do_pf_write); + else { + pf_busy = 0; + end_request(pf_req, 0); + goto repeat; + } +} + +static int pf_next_buf(void) +{ + unsigned long saved_flags; + + pf_count--; + pf_run--; + pf_buf += 512; + pf_block++; + if (!pf_run) + return 0; + if (!pf_count) + return 1; + spin_lock_irqsave(&pf_spin_lock, saved_flags); + end_request(pf_req, 1); + pf_count = pf_req->current_nr_sectors; + pf_buf = pf_req->buffer; + spin_unlock_irqrestore(&pf_spin_lock, saved_flags); + return 1; +} + +static inline void next_request(int success) +{ + unsigned long saved_flags; + + spin_lock_irqsave(&pf_spin_lock, saved_flags); + end_request(pf_req, success); + pf_busy = 0; + do_pf_request(pf_queue); + spin_unlock_irqrestore(&pf_spin_lock, saved_flags); +} + +/* detach from the calling context - in case the spinlock is held */ +static void do_pf_read(void) +{ + ps_set_intr(do_pf_read_start, NULL, 0, nice); +} + +static void do_pf_read_start(void) +{ + pf_busy = 1; + + if (pf_start(pf_current, ATAPI_READ_10, pf_block, pf_run)) { + pi_disconnect(pf_current->pi); + if (pf_retries < PF_MAX_RETRIES) { + pf_retries++; + pi_do_claimed(pf_current->pi, do_pf_read_start); + return; + } + next_request(0); + return; + } + pf_mask = STAT_DRQ; + ps_set_intr(do_pf_read_drq, pf_ready, PF_TMO, nice); +} + +static void do_pf_read_drq(void) +{ + while (1) { + if (pf_wait(pf_current, STAT_BUSY, STAT_DRQ | STAT_ERR, + "read block", "completion") & STAT_ERR) { + pi_disconnect(pf_current->pi); + if (pf_retries < PF_MAX_RETRIES) { + pf_req_sense(pf_current, 0); + pf_retries++; + pi_do_claimed(pf_current->pi, do_pf_read_start); + return; + } + next_request(0); + return; + } + pi_read_block(pf_current->pi, pf_buf, 512); + if (pf_next_buf()) + break; + } + pi_disconnect(pf_current->pi); + next_request(1); +} + +static void do_pf_write(void) +{ + ps_set_intr(do_pf_write_start, NULL, 0, nice); +} + +static void do_pf_write_start(void) +{ + pf_busy = 1; + + if (pf_start(pf_current, ATAPI_WRITE_10, pf_block, pf_run)) { + pi_disconnect(pf_current->pi); + if (pf_retries < PF_MAX_RETRIES) { + pf_retries++; + pi_do_claimed(pf_current->pi, do_pf_write_start); + return; + } + next_request(0); + return; + } + + while (1) { + if (pf_wait(pf_current, STAT_BUSY, STAT_DRQ | STAT_ERR, + "write block", "data wait") & STAT_ERR) { + pi_disconnect(pf_current->pi); + if (pf_retries < PF_MAX_RETRIES) { + pf_retries++; + pi_do_claimed(pf_current->pi, do_pf_write_start); + return; + } + next_request(0); + return; + } + pi_write_block(pf_current->pi, pf_buf, 512); + if (pf_next_buf()) + break; + } + pf_mask = 0; + ps_set_intr(do_pf_write_done, pf_ready, PF_TMO, nice); +} + +static void do_pf_write_done(void) +{ + if (pf_wait(pf_current, STAT_BUSY, 0, "write block", "done") & STAT_ERR) { + pi_disconnect(pf_current->pi); + if (pf_retries < PF_MAX_RETRIES) { + pf_retries++; + pi_do_claimed(pf_current->pi, do_pf_write_start); + return; + } + next_request(0); + return; + } + pi_disconnect(pf_current->pi); + next_request(1); +} + +static int __init pf_init(void) +{ /* preliminary initialisation */ + struct pf_unit *pf; + int unit; + + if (disable) + return -1; + + pf_init_units(); + + if (pf_detect()) + return -1; + pf_busy = 0; + + if (register_blkdev(major, name)) { + for (pf = units, unit = 0; unit < PF_UNITS; pf++, unit++) + put_disk(pf->disk); + return -1; + } + pf_queue = blk_init_queue(do_pf_request, &pf_spin_lock); + if (!pf_queue) { + unregister_blkdev(major, name); + for (pf = units, unit = 0; unit < PF_UNITS; pf++, unit++) + put_disk(pf->disk); + return -1; + } + + blk_queue_max_phys_segments(pf_queue, cluster); + blk_queue_max_hw_segments(pf_queue, cluster); + + for (pf = units, unit = 0; unit < PF_UNITS; pf++, unit++) { + struct gendisk *disk = pf->disk; + + if (!pf->present) + continue; + disk->private_data = pf; + disk->queue = pf_queue; + add_disk(disk); + } + return 0; +} + +static void __exit pf_exit(void) +{ + struct pf_unit *pf; + int unit; + unregister_blkdev(major, name); + for (pf = units, unit = 0; unit < PF_UNITS; pf++, unit++) { + if (!pf->present) + continue; + del_gendisk(pf->disk); + put_disk(pf->disk); + pi_release(pf->pi); + } + blk_cleanup_queue(pf_queue); +} + +MODULE_LICENSE("GPL"); +module_init(pf_init) +module_exit(pf_exit) diff --git a/drivers/block/paride/pg.c b/drivers/block/paride/pg.c new file mode 100644 index 00000000000..dbeb107bb97 --- /dev/null +++ b/drivers/block/paride/pg.c @@ -0,0 +1,723 @@ +/* + pg.c (c) 1998 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + The pg driver provides a simple character device interface for + sending ATAPI commands to a device. With the exception of the + ATAPI reset operation, all operations are performed by a pair + of read and write operations to the appropriate /dev/pgN device. + A write operation delivers a command and any outbound data in + a single buffer. Normally, the write will succeed unless the + device is offline or malfunctioning, or there is already another + command pending. If the write succeeds, it should be followed + immediately by a read operation, to obtain any returned data and + status information. A read will fail if there is no operation + in progress. + + As a special case, the device can be reset with a write operation, + and in this case, no following read is expected, or permitted. + + There are no ioctl() operations. Any single operation + may transfer at most PG_MAX_DATA bytes. Note that the driver must + copy the data through an internal buffer. In keeping with all + current ATAPI devices, command packets are assumed to be exactly + 12 bytes in length. + + To permit future changes to this interface, the headers in the + read and write buffers contain a single character "magic" flag. + Currently this flag must be the character "P". + + By default, the driver will autoprobe for a single parallel + port ATAPI device, but if their individual parameters are + specified, the driver can handle up to 4 devices. + + To use this device, you must have the following device + special files defined: + + /dev/pg0 c 97 0 + /dev/pg1 c 97 1 + /dev/pg2 c 97 2 + /dev/pg3 c 97 3 + + (You'll need to change the 97 to something else if you use + the 'major' parameter to install the driver on a different + major number.) + + The behaviour of the pg driver can be altered by setting + some parameters from the insmod command line. The following + parameters are adjustable: + + drive0 These four arguments can be arrays of + drive1 1-6 integers as follows: + drive2 + drive3 <prt>,<pro>,<uni>,<mod>,<slv>,<dly> + + Where, + + <prt> is the base of the parallel port address for + the corresponding drive. (required) + + <pro> is the protocol number for the adapter that + supports this drive. These numbers are + logged by 'paride' when the protocol modules + are initialised. (0 if not given) + + <uni> for those adapters that support chained + devices, this is the unit selector for the + chain of devices on the given port. It should + be zero for devices that don't support chaining. + (0 if not given) + + <mod> this can be -1 to choose the best mode, or one + of the mode numbers supported by the adapter. + (-1 if not given) + + <slv> ATAPI devices can be jumpered to master or slave. + Set this to 0 to choose the master drive, 1 to + choose the slave, -1 (the default) to choose the + first drive found. + + <dly> some parallel ports require the driver to + go more slowly. -1 sets a default value that + should work with the chosen protocol. Otherwise, + set this to a small integer, the larger it is + the slower the port i/o. In some cases, setting + this to zero will speed up the device. (default -1) + + major You may use this parameter to overide the + default major number (97) that this driver + will use. Be sure to change the device + name as well. + + name This parameter is a character string that + contains the name the kernel will use for this + device (in /proc output, for instance). + (default "pg"). + + verbose This parameter controls the amount of logging + that is done by the driver. Set it to 0 for + quiet operation, to 1 to enable progress + messages while the driver probes for devices, + or to 2 for full debug logging. (default 0) + + If this driver is built into the kernel, you can use + the following command line parameters, with the same values + as the corresponding module parameters listed above: + + pg.drive0 + pg.drive1 + pg.drive2 + pg.drive3 + + In addition, you can use the parameter pg.disable to disable + the driver entirely. + +*/ + +/* Changes: + + 1.01 GRG 1998.06.16 Bug fixes + 1.02 GRG 1998.09.24 Added jumbo support + +*/ + +#define PG_VERSION "1.02" +#define PG_MAJOR 97 +#define PG_NAME "pg" +#define PG_UNITS 4 + +#ifndef PI_PG +#define PI_PG 4 +#endif + +/* Here are things one can override from the insmod command. + Most are autoprobed by paride unless set here. Verbose is 0 + by default. + +*/ + +static int verbose = 0; +static int major = PG_MAJOR; +static char *name = PG_NAME; +static int disable = 0; + +static int drive0[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive1[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive2[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive3[6] = { 0, 0, 0, -1, -1, -1 }; + +static int (*drives[4])[6] = {&drive0, &drive1, &drive2, &drive3}; +static int pg_drive_count; + +enum {D_PRT, D_PRO, D_UNI, D_MOD, D_SLV, D_DLY}; + +/* end of parameters */ + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/fs.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/delay.h> +#include <linux/slab.h> +#include <linux/mtio.h> +#include <linux/pg.h> +#include <linux/device.h> + +#include <asm/uaccess.h> + +module_param(verbose, bool, 0644); +module_param(major, int, 0); +module_param(name, charp, 0); +module_param_array(drive0, int, NULL, 0); +module_param_array(drive1, int, NULL, 0); +module_param_array(drive2, int, NULL, 0); +module_param_array(drive3, int, NULL, 0); + +#include "paride.h" + +#define PG_SPIN_DEL 50 /* spin delay in micro-seconds */ +#define PG_SPIN 200 +#define PG_TMO HZ +#define PG_RESET_TMO 10*HZ + +#define STAT_ERR 0x01 +#define STAT_INDEX 0x02 +#define STAT_ECC 0x04 +#define STAT_DRQ 0x08 +#define STAT_SEEK 0x10 +#define STAT_WRERR 0x20 +#define STAT_READY 0x40 +#define STAT_BUSY 0x80 + +#define ATAPI_IDENTIFY 0x12 + +static int pg_open(struct inode *inode, struct file *file); +static int pg_release(struct inode *inode, struct file *file); +static ssize_t pg_read(struct file *filp, char __user *buf, + size_t count, loff_t * ppos); +static ssize_t pg_write(struct file *filp, const char __user *buf, + size_t count, loff_t * ppos); +static int pg_detect(void); + +#define PG_NAMELEN 8 + +struct pg { + struct pi_adapter pia; /* interface to paride layer */ + struct pi_adapter *pi; + int busy; /* write done, read expected */ + int start; /* jiffies at command start */ + int dlen; /* transfer size requested */ + unsigned long timeout; /* timeout requested */ + int status; /* last sense key */ + int drive; /* drive */ + unsigned long access; /* count of active opens ... */ + int present; /* device present ? */ + char *bufptr; + char name[PG_NAMELEN]; /* pg0, pg1, ... */ +}; + +static struct pg devices[PG_UNITS]; + +static int pg_identify(struct pg *dev, int log); + +static char pg_scratch[512]; /* scratch block buffer */ + +static struct class_simple *pg_class; + +/* kernel glue structures */ + +static struct file_operations pg_fops = { + .owner = THIS_MODULE, + .read = pg_read, + .write = pg_write, + .open = pg_open, + .release = pg_release, +}; + +static void pg_init_units(void) +{ + int unit; + + pg_drive_count = 0; + for (unit = 0; unit < PG_UNITS; unit++) { + int *parm = *drives[unit]; + struct pg *dev = &devices[unit]; + dev->pi = &dev->pia; + clear_bit(0, &dev->access); + dev->busy = 0; + dev->present = 0; + dev->bufptr = NULL; + dev->drive = parm[D_SLV]; + snprintf(dev->name, PG_NAMELEN, "%s%c", name, 'a'+unit); + if (parm[D_PRT]) + pg_drive_count++; + } +} + +static inline int status_reg(struct pg *dev) +{ + return pi_read_regr(dev->pi, 1, 6); +} + +static inline int read_reg(struct pg *dev, int reg) +{ + return pi_read_regr(dev->pi, 0, reg); +} + +static inline void write_reg(struct pg *dev, int reg, int val) +{ + pi_write_regr(dev->pi, 0, reg, val); +} + +static inline u8 DRIVE(struct pg *dev) +{ + return 0xa0+0x10*dev->drive; +} + +static void pg_sleep(int cs) +{ + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(cs); +} + +static int pg_wait(struct pg *dev, int go, int stop, unsigned long tmo, char *msg) +{ + int j, r, e, s, p, to; + + dev->status = 0; + + j = 0; + while ((((r = status_reg(dev)) & go) || (stop && (!(r & stop)))) + && time_before(jiffies, tmo)) { + if (j++ < PG_SPIN) + udelay(PG_SPIN_DEL); + else + pg_sleep(1); + } + + to = time_after_eq(jiffies, tmo); + + if ((r & (STAT_ERR & stop)) || to) { + s = read_reg(dev, 7); + e = read_reg(dev, 1); + p = read_reg(dev, 2); + if (verbose > 1) + printk("%s: %s: stat=0x%x err=0x%x phase=%d%s\n", + dev->name, msg, s, e, p, to ? " timeout" : ""); + if (to) + e |= 0x100; + dev->status = (e >> 4) & 0xff; + return -1; + } + return 0; +} + +static int pg_command(struct pg *dev, char *cmd, int dlen, unsigned long tmo) +{ + int k; + + pi_connect(dev->pi); + + write_reg(dev, 6, DRIVE(dev)); + + if (pg_wait(dev, STAT_BUSY | STAT_DRQ, 0, tmo, "before command")) + goto fail; + + write_reg(dev, 4, dlen % 256); + write_reg(dev, 5, dlen / 256); + write_reg(dev, 7, 0xa0); /* ATAPI packet command */ + + if (pg_wait(dev, STAT_BUSY, STAT_DRQ, tmo, "command DRQ")) + goto fail; + + if (read_reg(dev, 2) != 1) { + printk("%s: command phase error\n", dev->name); + goto fail; + } + + pi_write_block(dev->pi, cmd, 12); + + if (verbose > 1) { + printk("%s: Command sent, dlen=%d packet= ", dev->name, dlen); + for (k = 0; k < 12; k++) + printk("%02x ", cmd[k] & 0xff); + printk("\n"); + } + return 0; +fail: + pi_disconnect(dev->pi); + return -1; +} + +static int pg_completion(struct pg *dev, char *buf, unsigned long tmo) +{ + int r, d, n, p; + + r = pg_wait(dev, STAT_BUSY, STAT_DRQ | STAT_READY | STAT_ERR, + tmo, "completion"); + + dev->dlen = 0; + + while (read_reg(dev, 7) & STAT_DRQ) { + d = (read_reg(dev, 4) + 256 * read_reg(dev, 5)); + n = ((d + 3) & 0xfffc); + p = read_reg(dev, 2) & 3; + if (p == 0) + pi_write_block(dev->pi, buf, n); + if (p == 2) + pi_read_block(dev->pi, buf, n); + if (verbose > 1) + printk("%s: %s %d bytes\n", dev->name, + p ? "Read" : "Write", n); + dev->dlen += (1 - p) * d; + buf += d; + r = pg_wait(dev, STAT_BUSY, STAT_DRQ | STAT_READY | STAT_ERR, + tmo, "completion"); + } + + pi_disconnect(dev->pi); + + return r; +} + +static int pg_reset(struct pg *dev) +{ + int i, k, err; + int expect[5] = { 1, 1, 1, 0x14, 0xeb }; + int got[5]; + + pi_connect(dev->pi); + write_reg(dev, 6, DRIVE(dev)); + write_reg(dev, 7, 8); + + pg_sleep(20 * HZ / 1000); + + k = 0; + while ((k++ < PG_RESET_TMO) && (status_reg(dev) & STAT_BUSY)) + pg_sleep(1); + + for (i = 0; i < 5; i++) + got[i] = read_reg(dev, i + 1); + + err = memcmp(expect, got, sizeof(got)) ? -1 : 0; + + if (verbose) { + printk("%s: Reset (%d) signature = ", dev->name, k); + for (i = 0; i < 5; i++) + printk("%3x", got[i]); + if (err) + printk(" (incorrect)"); + printk("\n"); + } + + pi_disconnect(dev->pi); + return err; +} + +static void xs(char *buf, char *targ, int len) +{ + char l = '\0'; + int k; + + for (k = 0; k < len; k++) { + char c = *buf++; + if (c != ' ' || c != l) + l = *targ++ = c; + } + if (l == ' ') + targ--; + *targ = '\0'; +} + +static int pg_identify(struct pg *dev, int log) +{ + int s; + char *ms[2] = { "master", "slave" }; + char mf[10], id[18]; + char id_cmd[12] = { ATAPI_IDENTIFY, 0, 0, 0, 36, 0, 0, 0, 0, 0, 0, 0 }; + char buf[36]; + + s = pg_command(dev, id_cmd, 36, jiffies + PG_TMO); + if (s) + return -1; + s = pg_completion(dev, buf, jiffies + PG_TMO); + if (s) + return -1; + + if (log) { + xs(buf + 8, mf, 8); + xs(buf + 16, id, 16); + printk("%s: %s %s, %s\n", dev->name, mf, id, ms[dev->drive]); + } + + return 0; +} + +/* + * returns 0, with id set if drive is detected + * -1, if drive detection failed + */ +static int pg_probe(struct pg *dev) +{ + if (dev->drive == -1) { + for (dev->drive = 0; dev->drive <= 1; dev->drive++) + if (!pg_reset(dev)) + return pg_identify(dev, 1); + } else { + if (!pg_reset(dev)) + return pg_identify(dev, 1); + } + return -1; +} + +static int pg_detect(void) +{ + struct pg *dev = &devices[0]; + int k, unit; + + printk("%s: %s version %s, major %d\n", name, name, PG_VERSION, major); + + k = 0; + if (pg_drive_count == 0) { + if (pi_init(dev->pi, 1, -1, -1, -1, -1, -1, pg_scratch, + PI_PG, verbose, dev->name)) { + if (!pg_probe(dev)) { + dev->present = 1; + k++; + } else + pi_release(dev->pi); + } + + } else + for (unit = 0; unit < PG_UNITS; unit++, dev++) { + int *parm = *drives[unit]; + if (!parm[D_PRT]) + continue; + if (pi_init(dev->pi, 0, parm[D_PRT], parm[D_MOD], + parm[D_UNI], parm[D_PRO], parm[D_DLY], + pg_scratch, PI_PG, verbose, dev->name)) { + if (!pg_probe(dev)) { + dev->present = 1; + k++; + } else + pi_release(dev->pi); + } + } + + if (k) + return 0; + + printk("%s: No ATAPI device detected\n", name); + return -1; +} + +static int pg_open(struct inode *inode, struct file *file) +{ + int unit = iminor(inode) & 0x7f; + struct pg *dev = &devices[unit]; + + if ((unit >= PG_UNITS) || (!dev->present)) + return -ENODEV; + + if (test_and_set_bit(0, &dev->access)) + return -EBUSY; + + if (dev->busy) { + pg_reset(dev); + dev->busy = 0; + } + + pg_identify(dev, (verbose > 1)); + + dev->bufptr = kmalloc(PG_MAX_DATA, GFP_KERNEL); + if (dev->bufptr == NULL) { + clear_bit(0, &dev->access); + printk("%s: buffer allocation failed\n", dev->name); + return -ENOMEM; + } + + file->private_data = dev; + + return 0; +} + +static int pg_release(struct inode *inode, struct file *file) +{ + struct pg *dev = file->private_data; + + kfree(dev->bufptr); + dev->bufptr = NULL; + clear_bit(0, &dev->access); + + return 0; +} + +static ssize_t pg_write(struct file *filp, const char __user *buf, size_t count, loff_t *ppos) +{ + struct pg *dev = filp->private_data; + struct pg_write_hdr hdr; + int hs = sizeof (hdr); + + if (dev->busy) + return -EBUSY; + if (count < hs) + return -EINVAL; + + if (copy_from_user(&hdr, buf, hs)) + return -EFAULT; + + if (hdr.magic != PG_MAGIC) + return -EINVAL; + if (hdr.dlen > PG_MAX_DATA) + return -EINVAL; + if ((count - hs) > PG_MAX_DATA) + return -EINVAL; + + if (hdr.func == PG_RESET) { + if (count != hs) + return -EINVAL; + if (pg_reset(dev)) + return -EIO; + return count; + } + + if (hdr.func != PG_COMMAND) + return -EINVAL; + + dev->start = jiffies; + dev->timeout = hdr.timeout * HZ + HZ / 2 + jiffies; + + if (pg_command(dev, hdr.packet, hdr.dlen, jiffies + PG_TMO)) { + if (dev->status & 0x10) + return -ETIME; + return -EIO; + } + + dev->busy = 1; + + if (copy_from_user(dev->bufptr, buf + hs, count - hs)) + return -EFAULT; + return count; +} + +static ssize_t pg_read(struct file *filp, char __user *buf, size_t count, loff_t *ppos) +{ + struct pg *dev = filp->private_data; + struct pg_read_hdr hdr; + int hs = sizeof (hdr); + int copy; + + if (!dev->busy) + return -EINVAL; + if (count < hs) + return -EINVAL; + + dev->busy = 0; + + if (pg_completion(dev, dev->bufptr, dev->timeout)) + if (dev->status & 0x10) + return -ETIME; + + hdr.magic = PG_MAGIC; + hdr.dlen = dev->dlen; + copy = 0; + + if (hdr.dlen < 0) { + hdr.dlen = -1 * hdr.dlen; + copy = hdr.dlen; + if (copy > (count - hs)) + copy = count - hs; + } + + hdr.duration = (jiffies - dev->start + HZ / 2) / HZ; + hdr.scsi = dev->status & 0x0f; + + if (copy_to_user(buf, &hdr, hs)) + return -EFAULT; + if (copy > 0) + if (copy_to_user(buf + hs, dev->bufptr, copy)) + return -EFAULT; + return copy + hs; +} + +static int __init pg_init(void) +{ + int unit, err = 0; + + if (disable){ + err = -1; + goto out; + } + + pg_init_units(); + + if (pg_detect()) { + err = -1; + goto out; + } + + if (register_chrdev(major, name, &pg_fops)) { + printk("pg_init: unable to get major number %d\n", major); + for (unit = 0; unit < PG_UNITS; unit++) { + struct pg *dev = &devices[unit]; + if (dev->present) + pi_release(dev->pi); + } + err = -1; + goto out; + } + pg_class = class_simple_create(THIS_MODULE, "pg"); + if (IS_ERR(pg_class)) { + err = PTR_ERR(pg_class); + goto out_chrdev; + } + devfs_mk_dir("pg"); + for (unit = 0; unit < PG_UNITS; unit++) { + struct pg *dev = &devices[unit]; + if (dev->present) { + class_simple_device_add(pg_class, MKDEV(major, unit), + NULL, "pg%u", unit); + err = devfs_mk_cdev(MKDEV(major, unit), + S_IFCHR | S_IRUSR | S_IWUSR, "pg/%u", + unit); + if (err) + goto out_class; + } + } + err = 0; + goto out; + +out_class: + class_simple_device_remove(MKDEV(major, unit)); + class_simple_destroy(pg_class); +out_chrdev: + unregister_chrdev(major, "pg"); +out: + return err; +} + +static void __exit pg_exit(void) +{ + int unit; + + for (unit = 0; unit < PG_UNITS; unit++) { + struct pg *dev = &devices[unit]; + if (dev->present) { + class_simple_device_remove(MKDEV(major, unit)); + devfs_remove("pg/%u", unit); + } + } + class_simple_destroy(pg_class); + devfs_remove("pg"); + unregister_chrdev(major, name); + + for (unit = 0; unit < PG_UNITS; unit++) { + struct pg *dev = &devices[unit]; + if (dev->present) + pi_release(dev->pi); + } +} + +MODULE_LICENSE("GPL"); +module_init(pg_init) +module_exit(pg_exit) diff --git a/drivers/block/paride/ppc6lnx.c b/drivers/block/paride/ppc6lnx.c new file mode 100644 index 00000000000..5e5521d3b1d --- /dev/null +++ b/drivers/block/paride/ppc6lnx.c @@ -0,0 +1,726 @@ +/* + ppc6lnx.c (c) 2001 Micro Solutions Inc. + Released under the terms of the GNU General Public license + + ppc6lnx.c is a par of the protocol driver for the Micro Solutions + "BACKPACK" parallel port IDE adapter + (Works on Series 6 drives) + +*/ + +//*************************************************************************** + +// PPC 6 Code in C sanitized for LINUX +// Original x86 ASM by Ron, Converted to C by Clive + +//*************************************************************************** + + +#define port_stb 1 +#define port_afd 2 +#define cmd_stb port_afd +#define port_init 4 +#define data_stb port_init +#define port_sel 8 +#define port_int 16 +#define port_dir 0x20 + +#define ECR_EPP 0x80 +#define ECR_BI 0x20 + +//*************************************************************************** + +// 60772 Commands + +#define ACCESS_REG 0x00 +#define ACCESS_PORT 0x40 + +#define ACCESS_READ 0x00 +#define ACCESS_WRITE 0x20 + +// 60772 Command Prefix + +#define CMD_PREFIX_SET 0xe0 // Special command that modifies the next command's operation +#define CMD_PREFIX_RESET 0xc0 // Resets current cmd modifier reg bits + #define PREFIX_IO16 0x01 // perform 16-bit wide I/O + #define PREFIX_FASTWR 0x04 // enable PPC mode fast-write + #define PREFIX_BLK 0x08 // enable block transfer mode + +// 60772 Registers + +#define REG_STATUS 0x00 // status register + #define STATUS_IRQA 0x01 // Peripheral IRQA line + #define STATUS_EEPROM_DO 0x40 // Serial EEPROM data bit +#define REG_VERSION 0x01 // PPC version register (read) +#define REG_HWCFG 0x02 // Hardware Config register +#define REG_RAMSIZE 0x03 // Size of RAM Buffer + #define RAMSIZE_128K 0x02 +#define REG_EEPROM 0x06 // EEPROM control register + #define EEPROM_SK 0x01 // eeprom SK bit + #define EEPROM_DI 0x02 // eeprom DI bit + #define EEPROM_CS 0x04 // eeprom CS bit + #define EEPROM_EN 0x08 // eeprom output enable +#define REG_BLKSIZE 0x08 // Block transfer len (24 bit) + +//*************************************************************************** + +typedef struct ppc_storage { + u16 lpt_addr; // LPT base address + u8 ppc_id; + u8 mode; // operating mode + // 0 = PPC Uni SW + // 1 = PPC Uni FW + // 2 = PPC Bi SW + // 3 = PPC Bi FW + // 4 = EPP Byte + // 5 = EPP Word + // 6 = EPP Dword + u8 ppc_flags; + u8 org_data; // original LPT data port contents + u8 org_ctrl; // original LPT control port contents + u8 cur_ctrl; // current control port contents +} Interface; + +//*************************************************************************** + +// ppc_flags + +#define fifo_wait 0x10 + +//*************************************************************************** + +// DONT CHANGE THESE LEST YOU BREAK EVERYTHING - BIT FIELD DEPENDENCIES + +#define PPCMODE_UNI_SW 0 +#define PPCMODE_UNI_FW 1 +#define PPCMODE_BI_SW 2 +#define PPCMODE_BI_FW 3 +#define PPCMODE_EPP_BYTE 4 +#define PPCMODE_EPP_WORD 5 +#define PPCMODE_EPP_DWORD 6 + +//*************************************************************************** + +static int ppc6_select(Interface *ppc); +static void ppc6_deselect(Interface *ppc); +static void ppc6_send_cmd(Interface *ppc, u8 cmd); +static void ppc6_wr_data_byte(Interface *ppc, u8 data); +static u8 ppc6_rd_data_byte(Interface *ppc); +static u8 ppc6_rd_port(Interface *ppc, u8 port); +static void ppc6_wr_port(Interface *ppc, u8 port, u8 data); +static void ppc6_rd_data_blk(Interface *ppc, u8 *data, long count); +static void ppc6_wait_for_fifo(Interface *ppc); +static void ppc6_wr_data_blk(Interface *ppc, u8 *data, long count); +static void ppc6_rd_port16_blk(Interface *ppc, u8 port, u8 *data, long length); +static void ppc6_wr_port16_blk(Interface *ppc, u8 port, u8 *data, long length); +static void ppc6_wr_extout(Interface *ppc, u8 regdata); +static int ppc6_open(Interface *ppc); +static void ppc6_close(Interface *ppc); + +//*************************************************************************** + +static int ppc6_select(Interface *ppc) +{ + u8 i, j, k; + + i = inb(ppc->lpt_addr + 1); + + if (i & 1) + outb(i, ppc->lpt_addr + 1); + + ppc->org_data = inb(ppc->lpt_addr); + + ppc->org_ctrl = inb(ppc->lpt_addr + 2) & 0x5F; // readback ctrl + + ppc->cur_ctrl = ppc->org_ctrl; + + ppc->cur_ctrl |= port_sel; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + if (ppc->org_data == 'b') + outb('x', ppc->lpt_addr); + + outb('b', ppc->lpt_addr); + outb('p', ppc->lpt_addr); + outb(ppc->ppc_id, ppc->lpt_addr); + outb(~ppc->ppc_id,ppc->lpt_addr); + + ppc->cur_ctrl &= ~port_sel; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + ppc->cur_ctrl = (ppc->cur_ctrl & port_int) | port_init; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + i = ppc->mode & 0x0C; + + if (i == 0) + i = (ppc->mode & 2) | 1; + + outb(i, ppc->lpt_addr); + + ppc->cur_ctrl |= port_sel; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + // DELAY + + ppc->cur_ctrl |= port_afd; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + j = ((i & 0x08) << 4) | ((i & 0x07) << 3); + + k = inb(ppc->lpt_addr + 1) & 0xB8; + + if (j == k) + { + ppc->cur_ctrl &= ~port_afd; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + k = (inb(ppc->lpt_addr + 1) & 0xB8) ^ 0xB8; + + if (j == k) + { + if (i & 4) // EPP + ppc->cur_ctrl &= ~(port_sel | port_init); + else // PPC/ECP + ppc->cur_ctrl &= ~port_sel; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + return(1); + } + } + + outb(ppc->org_ctrl, ppc->lpt_addr + 2); + + outb(ppc->org_data, ppc->lpt_addr); + + return(0); // FAIL +} + +//*************************************************************************** + +static void ppc6_deselect(Interface *ppc) +{ + if (ppc->mode & 4) // EPP + ppc->cur_ctrl |= port_init; + else // PPC/ECP + ppc->cur_ctrl |= port_sel; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + outb(ppc->org_data, ppc->lpt_addr); + + outb((ppc->org_ctrl | port_sel), ppc->lpt_addr + 2); + + outb(ppc->org_ctrl, ppc->lpt_addr + 2); +} + +//*************************************************************************** + +static void ppc6_send_cmd(Interface *ppc, u8 cmd) +{ + switch(ppc->mode) + { + case PPCMODE_UNI_SW : + case PPCMODE_UNI_FW : + case PPCMODE_BI_SW : + case PPCMODE_BI_FW : + { + outb(cmd, ppc->lpt_addr); + + ppc->cur_ctrl ^= cmd_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + break; + } + + case PPCMODE_EPP_BYTE : + case PPCMODE_EPP_WORD : + case PPCMODE_EPP_DWORD : + { + outb(cmd, ppc->lpt_addr + 3); + + break; + } + } +} + +//*************************************************************************** + +static void ppc6_wr_data_byte(Interface *ppc, u8 data) +{ + switch(ppc->mode) + { + case PPCMODE_UNI_SW : + case PPCMODE_UNI_FW : + case PPCMODE_BI_SW : + case PPCMODE_BI_FW : + { + outb(data, ppc->lpt_addr); + + ppc->cur_ctrl ^= data_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + break; + } + + case PPCMODE_EPP_BYTE : + case PPCMODE_EPP_WORD : + case PPCMODE_EPP_DWORD : + { + outb(data, ppc->lpt_addr + 4); + + break; + } + } +} + +//*************************************************************************** + +static u8 ppc6_rd_data_byte(Interface *ppc) +{ + u8 data = 0; + + switch(ppc->mode) + { + case PPCMODE_UNI_SW : + case PPCMODE_UNI_FW : + { + ppc->cur_ctrl = (ppc->cur_ctrl & ~port_stb) ^ data_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + // DELAY + + data = inb(ppc->lpt_addr + 1); + + data = ((data & 0x80) >> 1) | ((data & 0x38) >> 3); + + ppc->cur_ctrl |= port_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + // DELAY + + data |= inb(ppc->lpt_addr + 1) & 0xB8; + + break; + } + + case PPCMODE_BI_SW : + case PPCMODE_BI_FW : + { + ppc->cur_ctrl |= port_dir; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + ppc->cur_ctrl = (ppc->cur_ctrl | port_stb) ^ data_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + data = inb(ppc->lpt_addr); + + ppc->cur_ctrl &= ~port_stb; + + outb(ppc->cur_ctrl,ppc->lpt_addr + 2); + + ppc->cur_ctrl &= ~port_dir; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + break; + } + + case PPCMODE_EPP_BYTE : + case PPCMODE_EPP_WORD : + case PPCMODE_EPP_DWORD : + { + outb((ppc->cur_ctrl | port_dir),ppc->lpt_addr + 2); + + data = inb(ppc->lpt_addr + 4); + + outb(ppc->cur_ctrl,ppc->lpt_addr + 2); + + break; + } + } + + return(data); +} + +//*************************************************************************** + +static u8 ppc6_rd_port(Interface *ppc, u8 port) +{ + ppc6_send_cmd(ppc,(u8)(port | ACCESS_PORT | ACCESS_READ)); + + return(ppc6_rd_data_byte(ppc)); +} + +//*************************************************************************** + +static void ppc6_wr_port(Interface *ppc, u8 port, u8 data) +{ + ppc6_send_cmd(ppc,(u8)(port | ACCESS_PORT | ACCESS_WRITE)); + + ppc6_wr_data_byte(ppc, data); +} + +//*************************************************************************** + +static void ppc6_rd_data_blk(Interface *ppc, u8 *data, long count) +{ + switch(ppc->mode) + { + case PPCMODE_UNI_SW : + case PPCMODE_UNI_FW : + { + while(count) + { + u8 d; + + ppc->cur_ctrl = (ppc->cur_ctrl & ~port_stb) ^ data_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + // DELAY + + d = inb(ppc->lpt_addr + 1); + + d = ((d & 0x80) >> 1) | ((d & 0x38) >> 3); + + ppc->cur_ctrl |= port_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + // DELAY + + d |= inb(ppc->lpt_addr + 1) & 0xB8; + + *data++ = d; + count--; + } + + break; + } + + case PPCMODE_BI_SW : + case PPCMODE_BI_FW : + { + ppc->cur_ctrl |= port_dir; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + ppc->cur_ctrl |= port_stb; + + while(count) + { + ppc->cur_ctrl ^= data_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + *data++ = inb(ppc->lpt_addr); + count--; + } + + ppc->cur_ctrl &= ~port_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + ppc->cur_ctrl &= ~port_dir; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + break; + } + + case PPCMODE_EPP_BYTE : + { + outb((ppc->cur_ctrl | port_dir), ppc->lpt_addr + 2); + + // DELAY + + while(count) + { + *data++ = inb(ppc->lpt_addr + 4); + count--; + } + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + break; + } + + case PPCMODE_EPP_WORD : + { + outb((ppc->cur_ctrl | port_dir), ppc->lpt_addr + 2); + + // DELAY + + while(count > 1) + { + *((u16 *)data) = inw(ppc->lpt_addr + 4); + data += 2; + count -= 2; + } + + while(count) + { + *data++ = inb(ppc->lpt_addr + 4); + count--; + } + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + break; + } + + case PPCMODE_EPP_DWORD : + { + outb((ppc->cur_ctrl | port_dir),ppc->lpt_addr + 2); + + // DELAY + + while(count > 3) + { + *((u32 *)data) = inl(ppc->lpt_addr + 4); + data += 4; + count -= 4; + } + + while(count) + { + *data++ = inb(ppc->lpt_addr + 4); + count--; + } + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + break; + } + } + +} + +//*************************************************************************** + +static void ppc6_wait_for_fifo(Interface *ppc) +{ + int i; + + if (ppc->ppc_flags & fifo_wait) + { + for(i=0; i<20; i++) + inb(ppc->lpt_addr + 1); + } +} + +//*************************************************************************** + +static void ppc6_wr_data_blk(Interface *ppc, u8 *data, long count) +{ + switch(ppc->mode) + { + case PPCMODE_UNI_SW : + case PPCMODE_BI_SW : + { + while(count--) + { + outb(*data++, ppc->lpt_addr); + + ppc->cur_ctrl ^= data_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + } + + break; + } + + case PPCMODE_UNI_FW : + case PPCMODE_BI_FW : + { + u8 this, last; + + ppc6_send_cmd(ppc,(CMD_PREFIX_SET | PREFIX_FASTWR)); + + ppc->cur_ctrl |= port_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + last = *data; + + outb(last, ppc->lpt_addr); + + while(count) + { + this = *data++; + count--; + + if (this == last) + { + ppc->cur_ctrl ^= data_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + } + else + { + outb(this, ppc->lpt_addr); + + last = this; + } + } + + ppc->cur_ctrl &= ~port_stb; + + outb(ppc->cur_ctrl, ppc->lpt_addr + 2); + + ppc6_send_cmd(ppc,(CMD_PREFIX_RESET | PREFIX_FASTWR)); + + break; + } + + case PPCMODE_EPP_BYTE : + { + while(count) + { + outb(*data++,ppc->lpt_addr + 4); + count--; + } + + ppc6_wait_for_fifo(ppc); + + break; + } + + case PPCMODE_EPP_WORD : + { + while(count > 1) + { + outw(*((u16 *)data),ppc->lpt_addr + 4); + data += 2; + count -= 2; + } + + while(count) + { + outb(*data++,ppc->lpt_addr + 4); + count--; + } + + ppc6_wait_for_fifo(ppc); + + break; + } + + case PPCMODE_EPP_DWORD : + { + while(count > 3) + { + outl(*((u32 *)data),ppc->lpt_addr + 4); + data += 4; + count -= 4; + } + + while(count) + { + outb(*data++,ppc->lpt_addr + 4); + count--; + } + + ppc6_wait_for_fifo(ppc); + + break; + } + } +} + +//*************************************************************************** + +static void ppc6_rd_port16_blk(Interface *ppc, u8 port, u8 *data, long length) +{ + length = length << 1; + + ppc6_send_cmd(ppc, (REG_BLKSIZE | ACCESS_REG | ACCESS_WRITE)); + ppc6_wr_data_byte(ppc,(u8)length); + ppc6_wr_data_byte(ppc,(u8)(length >> 8)); + ppc6_wr_data_byte(ppc,0); + + ppc6_send_cmd(ppc, (CMD_PREFIX_SET | PREFIX_IO16 | PREFIX_BLK)); + + ppc6_send_cmd(ppc, (u8)(port | ACCESS_PORT | ACCESS_READ)); + + ppc6_rd_data_blk(ppc, data, length); + + ppc6_send_cmd(ppc, (CMD_PREFIX_RESET | PREFIX_IO16 | PREFIX_BLK)); +} + +//*************************************************************************** + +static void ppc6_wr_port16_blk(Interface *ppc, u8 port, u8 *data, long length) +{ + length = length << 1; + + ppc6_send_cmd(ppc, (REG_BLKSIZE | ACCESS_REG | ACCESS_WRITE)); + ppc6_wr_data_byte(ppc,(u8)length); + ppc6_wr_data_byte(ppc,(u8)(length >> 8)); + ppc6_wr_data_byte(ppc,0); + + ppc6_send_cmd(ppc, (CMD_PREFIX_SET | PREFIX_IO16 | PREFIX_BLK)); + + ppc6_send_cmd(ppc, (u8)(port | ACCESS_PORT | ACCESS_WRITE)); + + ppc6_wr_data_blk(ppc, data, length); + + ppc6_send_cmd(ppc, (CMD_PREFIX_RESET | PREFIX_IO16 | PREFIX_BLK)); +} + +//*************************************************************************** + +static void ppc6_wr_extout(Interface *ppc, u8 regdata) +{ + ppc6_send_cmd(ppc,(REG_VERSION | ACCESS_REG | ACCESS_WRITE)); + + ppc6_wr_data_byte(ppc, (u8)((regdata & 0x03) << 6)); +} + +//*************************************************************************** + +static int ppc6_open(Interface *ppc) +{ + int ret; + + ret = ppc6_select(ppc); + + if (ret == 0) + return(ret); + + ppc->ppc_flags &= ~fifo_wait; + + ppc6_send_cmd(ppc, (ACCESS_REG | ACCESS_WRITE | REG_RAMSIZE)); + ppc6_wr_data_byte(ppc, RAMSIZE_128K); + + ppc6_send_cmd(ppc, (ACCESS_REG | ACCESS_READ | REG_VERSION)); + + if ((ppc6_rd_data_byte(ppc) & 0x3F) == 0x0C) + ppc->ppc_flags |= fifo_wait; + + return(ret); +} + +//*************************************************************************** + +static void ppc6_close(Interface *ppc) +{ + ppc6_deselect(ppc); +} + +//*************************************************************************** + diff --git a/drivers/block/paride/pseudo.h b/drivers/block/paride/pseudo.h new file mode 100644 index 00000000000..932342d7a8e --- /dev/null +++ b/drivers/block/paride/pseudo.h @@ -0,0 +1,102 @@ +/* + pseudo.h (c) 1997-8 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is the "pseudo-interrupt" logic for parallel port drivers. + + This module is #included into each driver. It makes one + function available: + + ps_set_intr( void (*continuation)(void), + int (*ready)(void), + int timeout, + int nice ) + + Which will arrange for ready() to be evaluated frequently and + when either it returns true, or timeout jiffies have passed, + continuation() will be invoked. + + If nice is 1, the test will done approximately once a + jiffy. If nice is 0, the test will also be done whenever + the scheduler runs (by adding it to a task queue). If + nice is greater than 1, the test will be done once every + (nice-1) jiffies. + +*/ + +/* Changes: + + 1.01 1998.05.03 Switched from cli()/sti() to spinlocks + 1.02 1998.12.14 Added support for nice > 1 +*/ + +#define PS_VERSION "1.02" + +#include <linux/sched.h> +#include <linux/workqueue.h> + +static void ps_tq_int( void *data); + +static void (* ps_continuation)(void); +static int (* ps_ready)(void); +static unsigned long ps_timeout; +static int ps_tq_active = 0; +static int ps_nice = 0; + +static DEFINE_SPINLOCK(ps_spinlock __attribute__((unused))); + +static DECLARE_WORK(ps_tq, ps_tq_int, NULL); + +static void ps_set_intr(void (*continuation)(void), + int (*ready)(void), + int timeout, int nice) +{ + unsigned long flags; + + spin_lock_irqsave(&ps_spinlock,flags); + + ps_continuation = continuation; + ps_ready = ready; + ps_timeout = jiffies + timeout; + ps_nice = nice; + + if (!ps_tq_active) { + ps_tq_active = 1; + if (!ps_nice) + schedule_work(&ps_tq); + else + schedule_delayed_work(&ps_tq, ps_nice-1); + } + spin_unlock_irqrestore(&ps_spinlock,flags); +} + +static void ps_tq_int(void *data) +{ + void (*con)(void); + unsigned long flags; + + spin_lock_irqsave(&ps_spinlock,flags); + + con = ps_continuation; + ps_tq_active = 0; + + if (!con) { + spin_unlock_irqrestore(&ps_spinlock,flags); + return; + } + if (!ps_ready || ps_ready() || time_after_eq(jiffies, ps_timeout)) { + ps_continuation = NULL; + spin_unlock_irqrestore(&ps_spinlock,flags); + con(); + return; + } + ps_tq_active = 1; + if (!ps_nice) + schedule_work(&ps_tq); + else + schedule_delayed_work(&ps_tq, ps_nice-1); + spin_unlock_irqrestore(&ps_spinlock,flags); +} + +/* end of pseudo.h */ + diff --git a/drivers/block/paride/pt.c b/drivers/block/paride/pt.c new file mode 100644 index 00000000000..8fbd6922fe0 --- /dev/null +++ b/drivers/block/paride/pt.c @@ -0,0 +1,1024 @@ +/* + pt.c (c) 1998 Grant R. Guenther <grant@torque.net> + Under the terms of the GNU General Public License. + + This is the high-level driver for parallel port ATAPI tape + drives based on chips supported by the paride module. + + The driver implements both rewinding and non-rewinding + devices, filemarks, and the rewind ioctl. It allocates + a small internal "bounce buffer" for each open device, but + otherwise expects buffering and blocking to be done at the + user level. As with most block-structured tapes, short + writes are padded to full tape blocks, so reading back a file + may return more data than was actually written. + + By default, the driver will autoprobe for a single parallel + port ATAPI tape drive, but if their individual parameters are + specified, the driver can handle up to 4 drives. + + The rewinding devices are named /dev/pt0, /dev/pt1, ... + while the non-rewinding devices are /dev/npt0, /dev/npt1, etc. + + The behaviour of the pt driver can be altered by setting + some parameters from the insmod command line. The following + parameters are adjustable: + + drive0 These four arguments can be arrays of + drive1 1-6 integers as follows: + drive2 + drive3 <prt>,<pro>,<uni>,<mod>,<slv>,<dly> + + Where, + + <prt> is the base of the parallel port address for + the corresponding drive. (required) + + <pro> is the protocol number for the adapter that + supports this drive. These numbers are + logged by 'paride' when the protocol modules + are initialised. (0 if not given) + + <uni> for those adapters that support chained + devices, this is the unit selector for the + chain of devices on the given port. It should + be zero for devices that don't support chaining. + (0 if not given) + + <mod> this can be -1 to choose the best mode, or one + of the mode numbers supported by the adapter. + (-1 if not given) + + <slv> ATAPI devices can be jumpered to master or slave. + Set this to 0 to choose the master drive, 1 to + choose the slave, -1 (the default) to choose the + first drive found. + + <dly> some parallel ports require the driver to + go more slowly. -1 sets a default value that + should work with the chosen protocol. Otherwise, + set this to a small integer, the larger it is + the slower the port i/o. In some cases, setting + this to zero will speed up the device. (default -1) + + major You may use this parameter to overide the + default major number (96) that this driver + will use. Be sure to change the device + name as well. + + name This parameter is a character string that + contains the name the kernel will use for this + device (in /proc output, for instance). + (default "pt"). + + verbose This parameter controls the amount of logging + that the driver will do. Set it to 0 for + normal operation, 1 to see autoprobe progress + messages, or 2 to see additional debugging + output. (default 0) + + If this driver is built into the kernel, you can use + the following command line parameters, with the same values + as the corresponding module parameters listed above: + + pt.drive0 + pt.drive1 + pt.drive2 + pt.drive3 + + In addition, you can use the parameter pt.disable to disable + the driver entirely. + +*/ + +/* Changes: + + 1.01 GRG 1998.05.06 Round up transfer size, fix ready_wait, + loosed interpretation of ATAPI standard + for clearing error status. + Eliminate sti(); + 1.02 GRG 1998.06.16 Eliminate an Ugh. + 1.03 GRG 1998.08.15 Adjusted PT_TMO, use HZ in loop timing, + extra debugging + 1.04 GRG 1998.09.24 Repair minor coding error, added jumbo support + +*/ + +#define PT_VERSION "1.04" +#define PT_MAJOR 96 +#define PT_NAME "pt" +#define PT_UNITS 4 + +/* Here are things one can override from the insmod command. + Most are autoprobed by paride unless set here. Verbose is on + by default. + +*/ + +static int verbose = 0; +static int major = PT_MAJOR; +static char *name = PT_NAME; +static int disable = 0; + +static int drive0[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive1[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive2[6] = { 0, 0, 0, -1, -1, -1 }; +static int drive3[6] = { 0, 0, 0, -1, -1, -1 }; + +static int (*drives[4])[6] = {&drive0, &drive1, &drive2, &drive3}; + +#define D_PRT 0 +#define D_PRO 1 +#define D_UNI 2 +#define D_MOD 3 +#define D_SLV 4 +#define D_DLY 5 + +#define DU (*drives[unit]) + +/* end of parameters */ + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/fs.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/delay.h> +#include <linux/slab.h> +#include <linux/mtio.h> +#include <linux/device.h> + +#include <asm/uaccess.h> + +module_param(verbose, bool, 0); +module_param(major, int, 0); +module_param(name, charp, 0); +module_param_array(drive0, int, NULL, 0); +module_param_array(drive1, int, NULL, 0); +module_param_array(drive2, int, NULL, 0); +module_param_array(drive3, int, NULL, 0); + +#include "paride.h" + +#define PT_MAX_RETRIES 5 +#define PT_TMO 3000 /* interrupt timeout in jiffies */ +#define PT_SPIN_DEL 50 /* spin delay in micro-seconds */ +#define PT_RESET_TMO 30 /* 30 seconds */ +#define PT_READY_TMO 60 /* 60 seconds */ +#define PT_REWIND_TMO 1200 /* 20 minutes */ + +#define PT_SPIN ((1000000/(HZ*PT_SPIN_DEL))*PT_TMO) + +#define STAT_ERR 0x00001 +#define STAT_INDEX 0x00002 +#define STAT_ECC 0x00004 +#define STAT_DRQ 0x00008 +#define STAT_SEEK 0x00010 +#define STAT_WRERR 0x00020 +#define STAT_READY 0x00040 +#define STAT_BUSY 0x00080 +#define STAT_SENSE 0x1f000 + +#define ATAPI_TEST_READY 0x00 +#define ATAPI_REWIND 0x01 +#define ATAPI_REQ_SENSE 0x03 +#define ATAPI_READ_6 0x08 +#define ATAPI_WRITE_6 0x0a +#define ATAPI_WFM 0x10 +#define ATAPI_IDENTIFY 0x12 +#define ATAPI_MODE_SENSE 0x1a +#define ATAPI_LOG_SENSE 0x4d + +static int pt_open(struct inode *inode, struct file *file); +static int pt_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg); +static int pt_release(struct inode *inode, struct file *file); +static ssize_t pt_read(struct file *filp, char __user *buf, + size_t count, loff_t * ppos); +static ssize_t pt_write(struct file *filp, const char __user *buf, + size_t count, loff_t * ppos); +static int pt_detect(void); + +/* bits in tape->flags */ + +#define PT_MEDIA 1 +#define PT_WRITE_OK 2 +#define PT_REWIND 4 +#define PT_WRITING 8 +#define PT_READING 16 +#define PT_EOF 32 + +#define PT_NAMELEN 8 +#define PT_BUFSIZE 16384 + +struct pt_unit { + struct pi_adapter pia; /* interface to paride layer */ + struct pi_adapter *pi; + int flags; /* various state flags */ + int last_sense; /* result of last request sense */ + int drive; /* drive */ + atomic_t available; /* 1 if access is available 0 otherwise */ + int bs; /* block size */ + int capacity; /* Size of tape in KB */ + int present; /* device present ? */ + char *bufptr; + char name[PT_NAMELEN]; /* pf0, pf1, ... */ +}; + +static int pt_identify(struct pt_unit *tape); + +static struct pt_unit pt[PT_UNITS]; + +static char pt_scratch[512]; /* scratch block buffer */ + +/* kernel glue structures */ + +static struct file_operations pt_fops = { + .owner = THIS_MODULE, + .read = pt_read, + .write = pt_write, + .ioctl = pt_ioctl, + .open = pt_open, + .release = pt_release, +}; + +/* sysfs class support */ +static struct class_simple *pt_class; + +static inline int status_reg(struct pi_adapter *pi) +{ + return pi_read_regr(pi, 1, 6); +} + +static inline int read_reg(struct pi_adapter *pi, int reg) +{ + return pi_read_regr(pi, 0, reg); +} + +static inline void write_reg(struct pi_adapter *pi, int reg, int val) +{ + pi_write_regr(pi, 0, reg, val); +} + +static inline u8 DRIVE(struct pt_unit *tape) +{ + return 0xa0+0x10*tape->drive; +} + +static int pt_wait(struct pt_unit *tape, int go, int stop, char *fun, char *msg) +{ + int j, r, e, s, p; + struct pi_adapter *pi = tape->pi; + + j = 0; + while ((((r = status_reg(pi)) & go) || (stop && (!(r & stop)))) + && (j++ < PT_SPIN)) + udelay(PT_SPIN_DEL); + + if ((r & (STAT_ERR & stop)) || (j >= PT_SPIN)) { + s = read_reg(pi, 7); + e = read_reg(pi, 1); + p = read_reg(pi, 2); + if (j >= PT_SPIN) + e |= 0x100; + if (fun) + printk("%s: %s %s: alt=0x%x stat=0x%x err=0x%x" + " loop=%d phase=%d\n", + tape->name, fun, msg, r, s, e, j, p); + return (e << 8) + s; + } + return 0; +} + +static int pt_command(struct pt_unit *tape, char *cmd, int dlen, char *fun) +{ + struct pi_adapter *pi = tape->pi; + pi_connect(pi); + + write_reg(pi, 6, DRIVE(tape)); + + if (pt_wait(tape, STAT_BUSY | STAT_DRQ, 0, fun, "before command")) { + pi_disconnect(pi); + return -1; + } + + write_reg(pi, 4, dlen % 256); + write_reg(pi, 5, dlen / 256); + write_reg(pi, 7, 0xa0); /* ATAPI packet command */ + + if (pt_wait(tape, STAT_BUSY, STAT_DRQ, fun, "command DRQ")) { + pi_disconnect(pi); + return -1; + } + + if (read_reg(pi, 2) != 1) { + printk("%s: %s: command phase error\n", tape->name, fun); + pi_disconnect(pi); + return -1; + } + + pi_write_block(pi, cmd, 12); + + return 0; +} + +static int pt_completion(struct pt_unit *tape, char *buf, char *fun) +{ + struct pi_adapter *pi = tape->pi; + int r, s, n, p; + + r = pt_wait(tape, STAT_BUSY, STAT_DRQ | STAT_READY | STAT_ERR, + fun, "completion"); + + if (read_reg(pi, 7) & STAT_DRQ) { + n = (((read_reg(pi, 4) + 256 * read_reg(pi, 5)) + + 3) & 0xfffc); + p = read_reg(pi, 2) & 3; + if (p == 0) + pi_write_block(pi, buf, n); + if (p == 2) + pi_read_block(pi, buf, n); + } + + s = pt_wait(tape, STAT_BUSY, STAT_READY | STAT_ERR, fun, "data done"); + + pi_disconnect(pi); + + return (r ? r : s); +} + +static void pt_req_sense(struct pt_unit *tape, int quiet) +{ + char rs_cmd[12] = { ATAPI_REQ_SENSE, 0, 0, 0, 16, 0, 0, 0, 0, 0, 0, 0 }; + char buf[16]; + int r; + + r = pt_command(tape, rs_cmd, 16, "Request sense"); + mdelay(1); + if (!r) + pt_completion(tape, buf, "Request sense"); + + tape->last_sense = -1; + if (!r) { + if (!quiet) + printk("%s: Sense key: %x, ASC: %x, ASQ: %x\n", + tape->name, buf[2] & 0xf, buf[12], buf[13]); + tape->last_sense = (buf[2] & 0xf) | ((buf[12] & 0xff) << 8) + | ((buf[13] & 0xff) << 16); + } +} + +static int pt_atapi(struct pt_unit *tape, char *cmd, int dlen, char *buf, char *fun) +{ + int r; + + r = pt_command(tape, cmd, dlen, fun); + mdelay(1); + if (!r) + r = pt_completion(tape, buf, fun); + if (r) + pt_req_sense(tape, !fun); + + return r; +} + +static void pt_sleep(int cs) +{ + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(cs); +} + +static int pt_poll_dsc(struct pt_unit *tape, int pause, int tmo, char *msg) +{ + struct pi_adapter *pi = tape->pi; + int k, e, s; + + k = 0; + e = 0; + s = 0; + while (k < tmo) { + pt_sleep(pause); + k++; + pi_connect(pi); + write_reg(pi, 6, DRIVE(tape)); + s = read_reg(pi, 7); + e = read_reg(pi, 1); + pi_disconnect(pi); + if (s & (STAT_ERR | STAT_SEEK)) + break; + } + if ((k >= tmo) || (s & STAT_ERR)) { + if (k >= tmo) + printk("%s: %s DSC timeout\n", tape->name, msg); + else + printk("%s: %s stat=0x%x err=0x%x\n", tape->name, msg, s, + e); + pt_req_sense(tape, 0); + return 0; + } + return 1; +} + +static void pt_media_access_cmd(struct pt_unit *tape, int tmo, char *cmd, char *fun) +{ + if (pt_command(tape, cmd, 0, fun)) { + pt_req_sense(tape, 0); + return; + } + pi_disconnect(tape->pi); + pt_poll_dsc(tape, HZ, tmo, fun); +} + +static void pt_rewind(struct pt_unit *tape) +{ + char rw_cmd[12] = { ATAPI_REWIND, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + + pt_media_access_cmd(tape, PT_REWIND_TMO, rw_cmd, "rewind"); +} + +static void pt_write_fm(struct pt_unit *tape) +{ + char wm_cmd[12] = { ATAPI_WFM, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0 }; + + pt_media_access_cmd(tape, PT_TMO, wm_cmd, "write filemark"); +} + +#define DBMSG(msg) ((verbose>1)?(msg):NULL) + +static int pt_reset(struct pt_unit *tape) +{ + struct pi_adapter *pi = tape->pi; + int i, k, flg; + int expect[5] = { 1, 1, 1, 0x14, 0xeb }; + + pi_connect(pi); + write_reg(pi, 6, DRIVE(tape)); + write_reg(pi, 7, 8); + + pt_sleep(20 * HZ / 1000); + + k = 0; + while ((k++ < PT_RESET_TMO) && (status_reg(pi) & STAT_BUSY)) + pt_sleep(HZ / 10); + + flg = 1; + for (i = 0; i < 5; i++) + flg &= (read_reg(pi, i + 1) == expect[i]); + + if (verbose) { + printk("%s: Reset (%d) signature = ", tape->name, k); + for (i = 0; i < 5; i++) + printk("%3x", read_reg(pi, i + 1)); + if (!flg) + printk(" (incorrect)"); + printk("\n"); + } + + pi_disconnect(pi); + return flg - 1; +} + +static int pt_ready_wait(struct pt_unit *tape, int tmo) +{ + char tr_cmd[12] = { ATAPI_TEST_READY, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + int k, p; + + k = 0; + while (k < tmo) { + tape->last_sense = 0; + pt_atapi(tape, tr_cmd, 0, NULL, DBMSG("test unit ready")); + p = tape->last_sense; + if (!p) + return 0; + if (!(((p & 0xffff) == 0x0402) || ((p & 0xff) == 6))) + return p; + k++; + pt_sleep(HZ); + } + return 0x000020; /* timeout */ +} + +static void xs(char *buf, char *targ, int offs, int len) +{ + int j, k, l; + + j = 0; + l = 0; + for (k = 0; k < len; k++) + if ((buf[k + offs] != 0x20) || (buf[k + offs] != l)) + l = targ[j++] = buf[k + offs]; + if (l == 0x20) + j--; + targ[j] = 0; +} + +static int xn(char *buf, int offs, int size) +{ + int v, k; + + v = 0; + for (k = 0; k < size; k++) + v = v * 256 + (buf[k + offs] & 0xff); + return v; +} + +static int pt_identify(struct pt_unit *tape) +{ + int dt, s; + char *ms[2] = { "master", "slave" }; + char mf[10], id[18]; + char id_cmd[12] = { ATAPI_IDENTIFY, 0, 0, 0, 36, 0, 0, 0, 0, 0, 0, 0 }; + char ms_cmd[12] = + { ATAPI_MODE_SENSE, 0, 0x2a, 0, 36, 0, 0, 0, 0, 0, 0, 0 }; + char ls_cmd[12] = + { ATAPI_LOG_SENSE, 0, 0x71, 0, 0, 0, 0, 0, 36, 0, 0, 0 }; + char buf[36]; + + s = pt_atapi(tape, id_cmd, 36, buf, "identify"); + if (s) + return -1; + + dt = buf[0] & 0x1f; + if (dt != 1) { + if (verbose) + printk("%s: Drive %d, unsupported type %d\n", + tape->name, tape->drive, dt); + return -1; + } + + xs(buf, mf, 8, 8); + xs(buf, id, 16, 16); + + tape->flags = 0; + tape->capacity = 0; + tape->bs = 0; + + if (!pt_ready_wait(tape, PT_READY_TMO)) + tape->flags |= PT_MEDIA; + + if (!pt_atapi(tape, ms_cmd, 36, buf, "mode sense")) { + if (!(buf[2] & 0x80)) + tape->flags |= PT_WRITE_OK; + tape->bs = xn(buf, 10, 2); + } + + if (!pt_atapi(tape, ls_cmd, 36, buf, "log sense")) + tape->capacity = xn(buf, 24, 4); + + printk("%s: %s %s, %s", tape->name, mf, id, ms[tape->drive]); + if (!(tape->flags & PT_MEDIA)) + printk(", no media\n"); + else { + if (!(tape->flags & PT_WRITE_OK)) + printk(", RO"); + printk(", blocksize %d, %d MB\n", tape->bs, tape->capacity / 1024); + } + + return 0; +} + + +/* + * returns 0, with id set if drive is detected + * -1, if drive detection failed + */ +static int pt_probe(struct pt_unit *tape) +{ + if (tape->drive == -1) { + for (tape->drive = 0; tape->drive <= 1; tape->drive++) + if (!pt_reset(tape)) + return pt_identify(tape); + } else { + if (!pt_reset(tape)) + return pt_identify(tape); + } + return -1; +} + +static int pt_detect(void) +{ + struct pt_unit *tape; + int specified = 0, found = 0; + int unit; + + printk("%s: %s version %s, major %d\n", name, name, PT_VERSION, major); + + specified = 0; + for (unit = 0; unit < PT_UNITS; unit++) { + struct pt_unit *tape = &pt[unit]; + tape->pi = &tape->pia; + atomic_set(&tape->available, 1); + tape->flags = 0; + tape->last_sense = 0; + tape->present = 0; + tape->bufptr = NULL; + tape->drive = DU[D_SLV]; + snprintf(tape->name, PT_NAMELEN, "%s%d", name, unit); + if (!DU[D_PRT]) + continue; + specified++; + if (pi_init(tape->pi, 0, DU[D_PRT], DU[D_MOD], DU[D_UNI], + DU[D_PRO], DU[D_DLY], pt_scratch, PI_PT, + verbose, tape->name)) { + if (!pt_probe(tape)) { + tape->present = 1; + found++; + } else + pi_release(tape->pi); + } + } + if (specified == 0) { + tape = pt; + if (pi_init(tape->pi, 1, -1, -1, -1, -1, -1, pt_scratch, + PI_PT, verbose, tape->name)) { + if (!pt_probe(tape)) { + tape->present = 1; + found++; + } else + pi_release(tape->pi); + } + + } + if (found) + return 0; + + printk("%s: No ATAPI tape drive detected\n", name); + return -1; +} + +static int pt_open(struct inode *inode, struct file *file) +{ + int unit = iminor(inode) & 0x7F; + struct pt_unit *tape = pt + unit; + int err; + + if (unit >= PT_UNITS || (!tape->present)) + return -ENODEV; + + err = -EBUSY; + if (!atomic_dec_and_test(&tape->available)) + goto out; + + pt_identify(tape); + + err = -ENODEV; + if (!tape->flags & PT_MEDIA) + goto out; + + err = -EROFS; + if ((!tape->flags & PT_WRITE_OK) && (file->f_mode & 2)) + goto out; + + if (!(iminor(inode) & 128)) + tape->flags |= PT_REWIND; + + err = -ENOMEM; + tape->bufptr = kmalloc(PT_BUFSIZE, GFP_KERNEL); + if (tape->bufptr == NULL) { + printk("%s: buffer allocation failed\n", tape->name); + goto out; + } + + file->private_data = tape; + return 0; + +out: + atomic_inc(&tape->available); + return err; +} + +static int pt_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + struct pt_unit *tape = file->private_data; + struct mtop __user *p = (void __user *)arg; + struct mtop mtop; + + switch (cmd) { + case MTIOCTOP: + if (copy_from_user(&mtop, p, sizeof(struct mtop))) + return -EFAULT; + + switch (mtop.mt_op) { + + case MTREW: + pt_rewind(tape); + return 0; + + case MTWEOF: + pt_write_fm(tape); + return 0; + + default: + printk("%s: Unimplemented mt_op %d\n", tape->name, + mtop.mt_op); + return -EINVAL; + } + + default: + printk("%s: Unimplemented ioctl 0x%x\n", tape->name, cmd); + return -EINVAL; + + } +} + +static int +pt_release(struct inode *inode, struct file *file) +{ + struct pt_unit *tape = file->private_data; + + if (atomic_read(&tape->available) > 1) + return -EINVAL; + + if (tape->flags & PT_WRITING) + pt_write_fm(tape); + + if (tape->flags & PT_REWIND) + pt_rewind(tape); + + kfree(tape->bufptr); + tape->bufptr = NULL; + + atomic_inc(&tape->available); + + return 0; + +} + +static ssize_t pt_read(struct file *filp, char __user *buf, size_t count, loff_t * ppos) +{ + struct pt_unit *tape = filp->private_data; + struct pi_adapter *pi = tape->pi; + char rd_cmd[12] = { ATAPI_READ_6, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + int k, n, r, p, s, t, b; + + if (!(tape->flags & (PT_READING | PT_WRITING))) { + tape->flags |= PT_READING; + if (pt_atapi(tape, rd_cmd, 0, NULL, "start read-ahead")) + return -EIO; + } else if (tape->flags & PT_WRITING) + return -EIO; + + if (tape->flags & PT_EOF) + return 0; + + t = 0; + + while (count > 0) { + + if (!pt_poll_dsc(tape, HZ / 100, PT_TMO, "read")) + return -EIO; + + n = count; + if (n > 32768) + n = 32768; /* max per command */ + b = (n - 1 + tape->bs) / tape->bs; + n = b * tape->bs; /* rounded up to even block */ + + rd_cmd[4] = b; + + r = pt_command(tape, rd_cmd, n, "read"); + + mdelay(1); + + if (r) { + pt_req_sense(tape, 0); + return -EIO; + } + + while (1) { + + r = pt_wait(tape, STAT_BUSY, + STAT_DRQ | STAT_ERR | STAT_READY, + DBMSG("read DRQ"), ""); + + if (r & STAT_SENSE) { + pi_disconnect(pi); + pt_req_sense(tape, 0); + return -EIO; + } + + if (r) + tape->flags |= PT_EOF; + + s = read_reg(pi, 7); + + if (!(s & STAT_DRQ)) + break; + + n = (read_reg(pi, 4) + 256 * read_reg(pi, 5)); + p = (read_reg(pi, 2) & 3); + if (p != 2) { + pi_disconnect(pi); + printk("%s: Phase error on read: %d\n", tape->name, + p); + return -EIO; + } + + while (n > 0) { + k = n; + if (k > PT_BUFSIZE) + k = PT_BUFSIZE; + pi_read_block(pi, tape->bufptr, k); + n -= k; + b = k; + if (b > count) + b = count; + if (copy_to_user(buf + t, tape->bufptr, b)) { + pi_disconnect(pi); + return -EFAULT; + } + t += b; + count -= b; + } + + } + pi_disconnect(pi); + if (tape->flags & PT_EOF) + break; + } + + return t; + +} + +static ssize_t pt_write(struct file *filp, const char __user *buf, size_t count, loff_t * ppos) +{ + struct pt_unit *tape = filp->private_data; + struct pi_adapter *pi = tape->pi; + char wr_cmd[12] = { ATAPI_WRITE_6, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + int k, n, r, p, s, t, b; + + if (!(tape->flags & PT_WRITE_OK)) + return -EROFS; + + if (!(tape->flags & (PT_READING | PT_WRITING))) { + tape->flags |= PT_WRITING; + if (pt_atapi + (tape, wr_cmd, 0, NULL, "start buffer-available mode")) + return -EIO; + } else if (tape->flags & PT_READING) + return -EIO; + + if (tape->flags & PT_EOF) + return -ENOSPC; + + t = 0; + + while (count > 0) { + + if (!pt_poll_dsc(tape, HZ / 100, PT_TMO, "write")) + return -EIO; + + n = count; + if (n > 32768) + n = 32768; /* max per command */ + b = (n - 1 + tape->bs) / tape->bs; + n = b * tape->bs; /* rounded up to even block */ + + wr_cmd[4] = b; + + r = pt_command(tape, wr_cmd, n, "write"); + + mdelay(1); + + if (r) { /* error delivering command only */ + pt_req_sense(tape, 0); + return -EIO; + } + + while (1) { + + r = pt_wait(tape, STAT_BUSY, + STAT_DRQ | STAT_ERR | STAT_READY, + DBMSG("write DRQ"), NULL); + + if (r & STAT_SENSE) { + pi_disconnect(pi); + pt_req_sense(tape, 0); + return -EIO; + } + + if (r) + tape->flags |= PT_EOF; + + s = read_reg(pi, 7); + + if (!(s & STAT_DRQ)) + break; + + n = (read_reg(pi, 4) + 256 * read_reg(pi, 5)); + p = (read_reg(pi, 2) & 3); + if (p != 0) { + pi_disconnect(pi); + printk("%s: Phase error on write: %d \n", + tape->name, p); + return -EIO; + } + + while (n > 0) { + k = n; + if (k > PT_BUFSIZE) + k = PT_BUFSIZE; + b = k; + if (b > count) + b = count; + if (copy_from_user(tape->bufptr, buf + t, b)) { + pi_disconnect(pi); + return -EFAULT; + } + pi_write_block(pi, tape->bufptr, k); + t += b; + count -= b; + n -= k; + } + + } + pi_disconnect(pi); + if (tape->flags & PT_EOF) + break; + } + + return t; +} + +static int __init pt_init(void) +{ + int unit, err = 0; + + if (disable) { + err = -1; + goto out; + } + + if (pt_detect()) { + err = -1; + goto out; + } + + if (register_chrdev(major, name, &pt_fops)) { + printk("pt_init: unable to get major number %d\n", major); + for (unit = 0; unit < PT_UNITS; unit++) + if (pt[unit].present) + pi_release(pt[unit].pi); + err = -1; + goto out; + } + pt_class = class_simple_create(THIS_MODULE, "pt"); + if (IS_ERR(pt_class)) { + err = PTR_ERR(pt_class); + goto out_chrdev; + } + + devfs_mk_dir("pt"); + for (unit = 0; unit < PT_UNITS; unit++) + if (pt[unit].present) { + class_simple_device_add(pt_class, MKDEV(major, unit), + NULL, "pt%d", unit); + err = devfs_mk_cdev(MKDEV(major, unit), + S_IFCHR | S_IRUSR | S_IWUSR, + "pt/%d", unit); + if (err) { + class_simple_device_remove(MKDEV(major, unit)); + goto out_class; + } + class_simple_device_add(pt_class, MKDEV(major, unit + 128), + NULL, "pt%dn", unit); + err = devfs_mk_cdev(MKDEV(major, unit + 128), + S_IFCHR | S_IRUSR | S_IWUSR, + "pt/%dn", unit); + if (err) { + class_simple_device_remove(MKDEV(major, unit + 128)); + goto out_class; + } + } + goto out; + +out_class: + class_simple_destroy(pt_class); +out_chrdev: + unregister_chrdev(major, "pt"); +out: + return err; +} + +static void __exit pt_exit(void) +{ + int unit; + for (unit = 0; unit < PT_UNITS; unit++) + if (pt[unit].present) { + class_simple_device_remove(MKDEV(major, unit)); + devfs_remove("pt/%d", unit); + class_simple_device_remove(MKDEV(major, unit + 128)); + devfs_remove("pt/%dn", unit); + } + class_simple_destroy(pt_class); + devfs_remove("pt"); + unregister_chrdev(major, name); + for (unit = 0; unit < PT_UNITS; unit++) + if (pt[unit].present) + pi_release(pt[unit].pi); +} + +MODULE_LICENSE("GPL"); +module_init(pt_init) +module_exit(pt_exit) diff --git a/drivers/block/pktcdvd.c b/drivers/block/pktcdvd.c new file mode 100644 index 00000000000..1a1fa3ccb91 --- /dev/null +++ b/drivers/block/pktcdvd.c @@ -0,0 +1,2681 @@ +/* + * Copyright (C) 2000 Jens Axboe <axboe@suse.de> + * Copyright (C) 2001-2004 Peter Osterlund <petero2@telia.com> + * + * May be copied or modified under the terms of the GNU General Public + * License. See linux/COPYING for more information. + * + * Packet writing layer for ATAPI and SCSI CD-R, CD-RW, DVD-R, and + * DVD-RW devices (aka an exercise in block layer masturbation) + * + * + * TODO: (circa order of when I will fix it) + * - Only able to write on CD-RW media right now. + * - check host application code on media and set it in write page + * - interface for UDF <-> packet to negotiate a new location when a write + * fails. + * - handle OPC, especially for -RW media + * + * Theory of operation: + * + * We use a custom make_request_fn function that forwards reads directly to + * the underlying CD device. Write requests are either attached directly to + * a live packet_data object, or simply stored sequentially in a list for + * later processing by the kcdrwd kernel thread. This driver doesn't use + * any elevator functionally as defined by the elevator_s struct, but the + * underlying CD device uses a standard elevator. + * + * This strategy makes it possible to do very late merging of IO requests. + * A new bio sent to pkt_make_request can be merged with a live packet_data + * object even if the object is in the data gathering state. + * + *************************************************************************/ + +#define VERSION_CODE "v0.2.0a 2004-07-14 Jens Axboe (axboe@suse.de) and petero2@telia.com" + +#include <linux/pktcdvd.h> +#include <linux/config.h> +#include <linux/module.h> +#include <linux/types.h> +#include <linux/kernel.h> +#include <linux/kthread.h> +#include <linux/errno.h> +#include <linux/spinlock.h> +#include <linux/file.h> +#include <linux/proc_fs.h> +#include <linux/seq_file.h> +#include <linux/miscdevice.h> +#include <linux/suspend.h> +#include <scsi/scsi_cmnd.h> +#include <scsi/scsi_ioctl.h> + +#include <asm/uaccess.h> + +#if PACKET_DEBUG +#define DPRINTK(fmt, args...) printk(KERN_NOTICE fmt, ##args) +#else +#define DPRINTK(fmt, args...) +#endif + +#if PACKET_DEBUG > 1 +#define VPRINTK(fmt, args...) printk(KERN_NOTICE fmt, ##args) +#else +#define VPRINTK(fmt, args...) +#endif + +#define MAX_SPEED 0xffff + +#define ZONE(sector, pd) (((sector) + (pd)->offset) & ~((pd)->settings.size - 1)) + +static struct pktcdvd_device *pkt_devs[MAX_WRITERS]; +static struct proc_dir_entry *pkt_proc; +static int pkt_major; +static struct semaphore ctl_mutex; /* Serialize open/close/setup/teardown */ +static mempool_t *psd_pool; + + +static void pkt_bio_finished(struct pktcdvd_device *pd) +{ + BUG_ON(atomic_read(&pd->cdrw.pending_bios) <= 0); + if (atomic_dec_and_test(&pd->cdrw.pending_bios)) { + VPRINTK("pktcdvd: queue empty\n"); + atomic_set(&pd->iosched.attention, 1); + wake_up(&pd->wqueue); + } +} + +static void pkt_bio_destructor(struct bio *bio) +{ + kfree(bio->bi_io_vec); + kfree(bio); +} + +static struct bio *pkt_bio_alloc(int nr_iovecs) +{ + struct bio_vec *bvl = NULL; + struct bio *bio; + + bio = kmalloc(sizeof(struct bio), GFP_KERNEL); + if (!bio) + goto no_bio; + bio_init(bio); + + bvl = kmalloc(nr_iovecs * sizeof(struct bio_vec), GFP_KERNEL); + if (!bvl) + goto no_bvl; + memset(bvl, 0, nr_iovecs * sizeof(struct bio_vec)); + + bio->bi_max_vecs = nr_iovecs; + bio->bi_io_vec = bvl; + bio->bi_destructor = pkt_bio_destructor; + + return bio; + + no_bvl: + kfree(bio); + no_bio: + return NULL; +} + +/* + * Allocate a packet_data struct + */ +static struct packet_data *pkt_alloc_packet_data(void) +{ + int i; + struct packet_data *pkt; + + pkt = kmalloc(sizeof(struct packet_data), GFP_KERNEL); + if (!pkt) + goto no_pkt; + memset(pkt, 0, sizeof(struct packet_data)); + + pkt->w_bio = pkt_bio_alloc(PACKET_MAX_SIZE); + if (!pkt->w_bio) + goto no_bio; + + for (i = 0; i < PAGES_PER_PACKET; i++) { + pkt->pages[i] = alloc_page(GFP_KERNEL|__GFP_ZERO); + if (!pkt->pages[i]) + goto no_page; + } + + spin_lock_init(&pkt->lock); + + for (i = 0; i < PACKET_MAX_SIZE; i++) { + struct bio *bio = pkt_bio_alloc(1); + if (!bio) + goto no_rd_bio; + pkt->r_bios[i] = bio; + } + + return pkt; + +no_rd_bio: + for (i = 0; i < PACKET_MAX_SIZE; i++) { + struct bio *bio = pkt->r_bios[i]; + if (bio) + bio_put(bio); + } + +no_page: + for (i = 0; i < PAGES_PER_PACKET; i++) + if (pkt->pages[i]) + __free_page(pkt->pages[i]); + bio_put(pkt->w_bio); +no_bio: + kfree(pkt); +no_pkt: + return NULL; +} + +/* + * Free a packet_data struct + */ +static void pkt_free_packet_data(struct packet_data *pkt) +{ + int i; + + for (i = 0; i < PACKET_MAX_SIZE; i++) { + struct bio *bio = pkt->r_bios[i]; + if (bio) + bio_put(bio); + } + for (i = 0; i < PAGES_PER_PACKET; i++) + __free_page(pkt->pages[i]); + bio_put(pkt->w_bio); + kfree(pkt); +} + +static void pkt_shrink_pktlist(struct pktcdvd_device *pd) +{ + struct packet_data *pkt, *next; + + BUG_ON(!list_empty(&pd->cdrw.pkt_active_list)); + + list_for_each_entry_safe(pkt, next, &pd->cdrw.pkt_free_list, list) { + pkt_free_packet_data(pkt); + } +} + +static int pkt_grow_pktlist(struct pktcdvd_device *pd, int nr_packets) +{ + struct packet_data *pkt; + + INIT_LIST_HEAD(&pd->cdrw.pkt_free_list); + INIT_LIST_HEAD(&pd->cdrw.pkt_active_list); + spin_lock_init(&pd->cdrw.active_list_lock); + while (nr_packets > 0) { + pkt = pkt_alloc_packet_data(); + if (!pkt) { + pkt_shrink_pktlist(pd); + return 0; + } + pkt->id = nr_packets; + pkt->pd = pd; + list_add(&pkt->list, &pd->cdrw.pkt_free_list); + nr_packets--; + } + return 1; +} + +static void *pkt_rb_alloc(unsigned int __nocast gfp_mask, void *data) +{ + return kmalloc(sizeof(struct pkt_rb_node), gfp_mask); +} + +static void pkt_rb_free(void *ptr, void *data) +{ + kfree(ptr); +} + +static inline struct pkt_rb_node *pkt_rbtree_next(struct pkt_rb_node *node) +{ + struct rb_node *n = rb_next(&node->rb_node); + if (!n) + return NULL; + return rb_entry(n, struct pkt_rb_node, rb_node); +} + +static inline void pkt_rbtree_erase(struct pktcdvd_device *pd, struct pkt_rb_node *node) +{ + rb_erase(&node->rb_node, &pd->bio_queue); + mempool_free(node, pd->rb_pool); + pd->bio_queue_size--; + BUG_ON(pd->bio_queue_size < 0); +} + +/* + * Find the first node in the pd->bio_queue rb tree with a starting sector >= s. + */ +static struct pkt_rb_node *pkt_rbtree_find(struct pktcdvd_device *pd, sector_t s) +{ + struct rb_node *n = pd->bio_queue.rb_node; + struct rb_node *next; + struct pkt_rb_node *tmp; + + if (!n) { + BUG_ON(pd->bio_queue_size > 0); + return NULL; + } + + for (;;) { + tmp = rb_entry(n, struct pkt_rb_node, rb_node); + if (s <= tmp->bio->bi_sector) + next = n->rb_left; + else + next = n->rb_right; + if (!next) + break; + n = next; + } + + if (s > tmp->bio->bi_sector) { + tmp = pkt_rbtree_next(tmp); + if (!tmp) + return NULL; + } + BUG_ON(s > tmp->bio->bi_sector); + return tmp; +} + +/* + * Insert a node into the pd->bio_queue rb tree. + */ +static void pkt_rbtree_insert(struct pktcdvd_device *pd, struct pkt_rb_node *node) +{ + struct rb_node **p = &pd->bio_queue.rb_node; + struct rb_node *parent = NULL; + sector_t s = node->bio->bi_sector; + struct pkt_rb_node *tmp; + + while (*p) { + parent = *p; + tmp = rb_entry(parent, struct pkt_rb_node, rb_node); + if (s < tmp->bio->bi_sector) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + rb_link_node(&node->rb_node, parent, p); + rb_insert_color(&node->rb_node, &pd->bio_queue); + pd->bio_queue_size++; +} + +/* + * Add a bio to a single linked list defined by its head and tail pointers. + */ +static inline void pkt_add_list_last(struct bio *bio, struct bio **list_head, struct bio **list_tail) +{ + bio->bi_next = NULL; + if (*list_tail) { + BUG_ON((*list_head) == NULL); + (*list_tail)->bi_next = bio; + (*list_tail) = bio; + } else { + BUG_ON((*list_head) != NULL); + (*list_head) = bio; + (*list_tail) = bio; + } +} + +/* + * Remove and return the first bio from a single linked list defined by its + * head and tail pointers. + */ +static inline struct bio *pkt_get_list_first(struct bio **list_head, struct bio **list_tail) +{ + struct bio *bio; + + if (*list_head == NULL) + return NULL; + + bio = *list_head; + *list_head = bio->bi_next; + if (*list_head == NULL) + *list_tail = NULL; + + bio->bi_next = NULL; + return bio; +} + +/* + * Send a packet_command to the underlying block device and + * wait for completion. + */ +static int pkt_generic_packet(struct pktcdvd_device *pd, struct packet_command *cgc) +{ + char sense[SCSI_SENSE_BUFFERSIZE]; + request_queue_t *q; + struct request *rq; + DECLARE_COMPLETION(wait); + int err = 0; + + q = bdev_get_queue(pd->bdev); + + rq = blk_get_request(q, (cgc->data_direction == CGC_DATA_WRITE) ? WRITE : READ, + __GFP_WAIT); + rq->errors = 0; + rq->rq_disk = pd->bdev->bd_disk; + rq->bio = NULL; + rq->buffer = NULL; + rq->timeout = 60*HZ; + rq->data = cgc->buffer; + rq->data_len = cgc->buflen; + rq->sense = sense; + memset(sense, 0, sizeof(sense)); + rq->sense_len = 0; + rq->flags |= REQ_BLOCK_PC | REQ_HARDBARRIER; + if (cgc->quiet) + rq->flags |= REQ_QUIET; + memcpy(rq->cmd, cgc->cmd, CDROM_PACKET_SIZE); + if (sizeof(rq->cmd) > CDROM_PACKET_SIZE) + memset(rq->cmd + CDROM_PACKET_SIZE, 0, sizeof(rq->cmd) - CDROM_PACKET_SIZE); + + rq->ref_count++; + rq->flags |= REQ_NOMERGE; + rq->waiting = &wait; + rq->end_io = blk_end_sync_rq; + elv_add_request(q, rq, ELEVATOR_INSERT_BACK, 1); + generic_unplug_device(q); + wait_for_completion(&wait); + + if (rq->errors) + err = -EIO; + + blk_put_request(rq); + return err; +} + +/* + * A generic sense dump / resolve mechanism should be implemented across + * all ATAPI + SCSI devices. + */ +static void pkt_dump_sense(struct packet_command *cgc) +{ + static char *info[9] = { "No sense", "Recovered error", "Not ready", + "Medium error", "Hardware error", "Illegal request", + "Unit attention", "Data protect", "Blank check" }; + int i; + struct request_sense *sense = cgc->sense; + + printk("pktcdvd:"); + for (i = 0; i < CDROM_PACKET_SIZE; i++) + printk(" %02x", cgc->cmd[i]); + printk(" - "); + + if (sense == NULL) { + printk("no sense\n"); + return; + } + + printk("sense %02x.%02x.%02x", sense->sense_key, sense->asc, sense->ascq); + + if (sense->sense_key > 8) { + printk(" (INVALID)\n"); + return; + } + + printk(" (%s)\n", info[sense->sense_key]); +} + +/* + * flush the drive cache to media + */ +static int pkt_flush_cache(struct pktcdvd_device *pd) +{ + struct packet_command cgc; + + init_cdrom_command(&cgc, NULL, 0, CGC_DATA_NONE); + cgc.cmd[0] = GPCMD_FLUSH_CACHE; + cgc.quiet = 1; + + /* + * the IMMED bit -- we default to not setting it, although that + * would allow a much faster close, this is safer + */ +#if 0 + cgc.cmd[1] = 1 << 1; +#endif + return pkt_generic_packet(pd, &cgc); +} + +/* + * speed is given as the normal factor, e.g. 4 for 4x + */ +static int pkt_set_speed(struct pktcdvd_device *pd, unsigned write_speed, unsigned read_speed) +{ + struct packet_command cgc; + struct request_sense sense; + int ret; + + init_cdrom_command(&cgc, NULL, 0, CGC_DATA_NONE); + cgc.sense = &sense; + cgc.cmd[0] = GPCMD_SET_SPEED; + cgc.cmd[2] = (read_speed >> 8) & 0xff; + cgc.cmd[3] = read_speed & 0xff; + cgc.cmd[4] = (write_speed >> 8) & 0xff; + cgc.cmd[5] = write_speed & 0xff; + + if ((ret = pkt_generic_packet(pd, &cgc))) + pkt_dump_sense(&cgc); + + return ret; +} + +/* + * Queue a bio for processing by the low-level CD device. Must be called + * from process context. + */ +static void pkt_queue_bio(struct pktcdvd_device *pd, struct bio *bio, int high_prio_read) +{ + spin_lock(&pd->iosched.lock); + if (bio_data_dir(bio) == READ) { + pkt_add_list_last(bio, &pd->iosched.read_queue, + &pd->iosched.read_queue_tail); + if (high_prio_read) + pd->iosched.high_prio_read = 1; + } else { + pkt_add_list_last(bio, &pd->iosched.write_queue, + &pd->iosched.write_queue_tail); + } + spin_unlock(&pd->iosched.lock); + + atomic_set(&pd->iosched.attention, 1); + wake_up(&pd->wqueue); +} + +/* + * Process the queued read/write requests. This function handles special + * requirements for CDRW drives: + * - A cache flush command must be inserted before a read request if the + * previous request was a write. + * - Switching between reading and writing is slow, so don't it more often + * than necessary. + * - Set the read speed according to current usage pattern. When only reading + * from the device, it's best to use the highest possible read speed, but + * when switching often between reading and writing, it's better to have the + * same read and write speeds. + * - Reads originating from user space should have higher priority than reads + * originating from pkt_gather_data, because some process is usually waiting + * on reads of the first kind. + */ +static void pkt_iosched_process_queue(struct pktcdvd_device *pd) +{ + request_queue_t *q; + + if (atomic_read(&pd->iosched.attention) == 0) + return; + atomic_set(&pd->iosched.attention, 0); + + q = bdev_get_queue(pd->bdev); + + for (;;) { + struct bio *bio; + int reads_queued, writes_queued, high_prio_read; + + spin_lock(&pd->iosched.lock); + reads_queued = (pd->iosched.read_queue != NULL); + writes_queued = (pd->iosched.write_queue != NULL); + if (!reads_queued) + pd->iosched.high_prio_read = 0; + high_prio_read = pd->iosched.high_prio_read; + spin_unlock(&pd->iosched.lock); + + if (!reads_queued && !writes_queued) + break; + + if (pd->iosched.writing) { + if (high_prio_read || (!writes_queued && reads_queued)) { + if (atomic_read(&pd->cdrw.pending_bios) > 0) { + VPRINTK("pktcdvd: write, waiting\n"); + break; + } + pkt_flush_cache(pd); + pd->iosched.writing = 0; + } + } else { + if (!reads_queued && writes_queued) { + if (atomic_read(&pd->cdrw.pending_bios) > 0) { + VPRINTK("pktcdvd: read, waiting\n"); + break; + } + pd->iosched.writing = 1; + } + } + + spin_lock(&pd->iosched.lock); + if (pd->iosched.writing) { + bio = pkt_get_list_first(&pd->iosched.write_queue, + &pd->iosched.write_queue_tail); + } else { + bio = pkt_get_list_first(&pd->iosched.read_queue, + &pd->iosched.read_queue_tail); + } + spin_unlock(&pd->iosched.lock); + + if (!bio) + continue; + + if (bio_data_dir(bio) == READ) + pd->iosched.successive_reads += bio->bi_size >> 10; + else + pd->iosched.successive_reads = 0; + if (pd->iosched.successive_reads >= HI_SPEED_SWITCH) { + if (pd->read_speed == pd->write_speed) { + pd->read_speed = MAX_SPEED; + pkt_set_speed(pd, pd->write_speed, pd->read_speed); + } + } else { + if (pd->read_speed != pd->write_speed) { + pd->read_speed = pd->write_speed; + pkt_set_speed(pd, pd->write_speed, pd->read_speed); + } + } + + atomic_inc(&pd->cdrw.pending_bios); + generic_make_request(bio); + } +} + +/* + * Special care is needed if the underlying block device has a small + * max_phys_segments value. + */ +static int pkt_set_segment_merging(struct pktcdvd_device *pd, request_queue_t *q) +{ + if ((pd->settings.size << 9) / CD_FRAMESIZE <= q->max_phys_segments) { + /* + * The cdrom device can handle one segment/frame + */ + clear_bit(PACKET_MERGE_SEGS, &pd->flags); + return 0; + } else if ((pd->settings.size << 9) / PAGE_SIZE <= q->max_phys_segments) { + /* + * We can handle this case at the expense of some extra memory + * copies during write operations + */ + set_bit(PACKET_MERGE_SEGS, &pd->flags); + return 0; + } else { + printk("pktcdvd: cdrom max_phys_segments too small\n"); + return -EIO; + } +} + +/* + * Copy CD_FRAMESIZE bytes from src_bio into a destination page + */ +static void pkt_copy_bio_data(struct bio *src_bio, int seg, int offs, struct page *dst_page, int dst_offs) +{ + unsigned int copy_size = CD_FRAMESIZE; + + while (copy_size > 0) { + struct bio_vec *src_bvl = bio_iovec_idx(src_bio, seg); + void *vfrom = kmap_atomic(src_bvl->bv_page, KM_USER0) + + src_bvl->bv_offset + offs; + void *vto = page_address(dst_page) + dst_offs; + int len = min_t(int, copy_size, src_bvl->bv_len - offs); + + BUG_ON(len < 0); + memcpy(vto, vfrom, len); + kunmap_atomic(vfrom, KM_USER0); + + seg++; + offs = 0; + dst_offs += len; + copy_size -= len; + } +} + +/* + * Copy all data for this packet to pkt->pages[], so that + * a) The number of required segments for the write bio is minimized, which + * is necessary for some scsi controllers. + * b) The data can be used as cache to avoid read requests if we receive a + * new write request for the same zone. + */ +static void pkt_make_local_copy(struct packet_data *pkt, struct page **pages, int *offsets) +{ + int f, p, offs; + + /* Copy all data to pkt->pages[] */ + p = 0; + offs = 0; + for (f = 0; f < pkt->frames; f++) { + if (pages[f] != pkt->pages[p]) { + void *vfrom = kmap_atomic(pages[f], KM_USER0) + offsets[f]; + void *vto = page_address(pkt->pages[p]) + offs; + memcpy(vto, vfrom, CD_FRAMESIZE); + kunmap_atomic(vfrom, KM_USER0); + pages[f] = pkt->pages[p]; + offsets[f] = offs; + } else { + BUG_ON(offsets[f] != offs); + } + offs += CD_FRAMESIZE; + if (offs >= PAGE_SIZE) { + BUG_ON(offs > PAGE_SIZE); + offs = 0; + p++; + } + } +} + +static int pkt_end_io_read(struct bio *bio, unsigned int bytes_done, int err) +{ + struct packet_data *pkt = bio->bi_private; + struct pktcdvd_device *pd = pkt->pd; + BUG_ON(!pd); + + if (bio->bi_size) + return 1; + + VPRINTK("pkt_end_io_read: bio=%p sec0=%llx sec=%llx err=%d\n", bio, + (unsigned long long)pkt->sector, (unsigned long long)bio->bi_sector, err); + + if (err) + atomic_inc(&pkt->io_errors); + if (atomic_dec_and_test(&pkt->io_wait)) { + atomic_inc(&pkt->run_sm); + wake_up(&pd->wqueue); + } + pkt_bio_finished(pd); + + return 0; +} + +static int pkt_end_io_packet_write(struct bio *bio, unsigned int bytes_done, int err) +{ + struct packet_data *pkt = bio->bi_private; + struct pktcdvd_device *pd = pkt->pd; + BUG_ON(!pd); + + if (bio->bi_size) + return 1; + + VPRINTK("pkt_end_io_packet_write: id=%d, err=%d\n", pkt->id, err); + + pd->stats.pkt_ended++; + + pkt_bio_finished(pd); + atomic_dec(&pkt->io_wait); + atomic_inc(&pkt->run_sm); + wake_up(&pd->wqueue); + return 0; +} + +/* + * Schedule reads for the holes in a packet + */ +static void pkt_gather_data(struct pktcdvd_device *pd, struct packet_data *pkt) +{ + int frames_read = 0; + struct bio *bio; + int f; + char written[PACKET_MAX_SIZE]; + + BUG_ON(!pkt->orig_bios); + + atomic_set(&pkt->io_wait, 0); + atomic_set(&pkt->io_errors, 0); + + if (pkt->cache_valid) { + VPRINTK("pkt_gather_data: zone %llx cached\n", + (unsigned long long)pkt->sector); + goto out_account; + } + + /* + * Figure out which frames we need to read before we can write. + */ + memset(written, 0, sizeof(written)); + spin_lock(&pkt->lock); + for (bio = pkt->orig_bios; bio; bio = bio->bi_next) { + int first_frame = (bio->bi_sector - pkt->sector) / (CD_FRAMESIZE >> 9); + int num_frames = bio->bi_size / CD_FRAMESIZE; + BUG_ON(first_frame < 0); + BUG_ON(first_frame + num_frames > pkt->frames); + for (f = first_frame; f < first_frame + num_frames; f++) + written[f] = 1; + } + spin_unlock(&pkt->lock); + + /* + * Schedule reads for missing parts of the packet. + */ + for (f = 0; f < pkt->frames; f++) { + int p, offset; + if (written[f]) + continue; + bio = pkt->r_bios[f]; + bio_init(bio); + bio->bi_max_vecs = 1; + bio->bi_sector = pkt->sector + f * (CD_FRAMESIZE >> 9); + bio->bi_bdev = pd->bdev; + bio->bi_end_io = pkt_end_io_read; + bio->bi_private = pkt; + + p = (f * CD_FRAMESIZE) / PAGE_SIZE; + offset = (f * CD_FRAMESIZE) % PAGE_SIZE; + VPRINTK("pkt_gather_data: Adding frame %d, page:%p offs:%d\n", + f, pkt->pages[p], offset); + if (!bio_add_page(bio, pkt->pages[p], CD_FRAMESIZE, offset)) + BUG(); + + atomic_inc(&pkt->io_wait); + bio->bi_rw = READ; + pkt_queue_bio(pd, bio, 0); + frames_read++; + } + +out_account: + VPRINTK("pkt_gather_data: need %d frames for zone %llx\n", + frames_read, (unsigned long long)pkt->sector); + pd->stats.pkt_started++; + pd->stats.secs_rg += frames_read * (CD_FRAMESIZE >> 9); + pd->stats.secs_w += pd->settings.size; +} + +/* + * Find a packet matching zone, or the least recently used packet if + * there is no match. + */ +static struct packet_data *pkt_get_packet_data(struct pktcdvd_device *pd, int zone) +{ + struct packet_data *pkt; + + list_for_each_entry(pkt, &pd->cdrw.pkt_free_list, list) { + if (pkt->sector == zone || pkt->list.next == &pd->cdrw.pkt_free_list) { + list_del_init(&pkt->list); + if (pkt->sector != zone) + pkt->cache_valid = 0; + break; + } + } + return pkt; +} + +static void pkt_put_packet_data(struct pktcdvd_device *pd, struct packet_data *pkt) +{ + if (pkt->cache_valid) { + list_add(&pkt->list, &pd->cdrw.pkt_free_list); + } else { + list_add_tail(&pkt->list, &pd->cdrw.pkt_free_list); + } +} + +/* + * recover a failed write, query for relocation if possible + * + * returns 1 if recovery is possible, or 0 if not + * + */ +static int pkt_start_recovery(struct packet_data *pkt) +{ + /* + * FIXME. We need help from the file system to implement + * recovery handling. + */ + return 0; +#if 0 + struct request *rq = pkt->rq; + struct pktcdvd_device *pd = rq->rq_disk->private_data; + struct block_device *pkt_bdev; + struct super_block *sb = NULL; + unsigned long old_block, new_block; + sector_t new_sector; + + pkt_bdev = bdget(kdev_t_to_nr(pd->pkt_dev)); + if (pkt_bdev) { + sb = get_super(pkt_bdev); + bdput(pkt_bdev); + } + + if (!sb) + return 0; + + if (!sb->s_op || !sb->s_op->relocate_blocks) + goto out; + + old_block = pkt->sector / (CD_FRAMESIZE >> 9); + if (sb->s_op->relocate_blocks(sb, old_block, &new_block)) + goto out; + + new_sector = new_block * (CD_FRAMESIZE >> 9); + pkt->sector = new_sector; + + pkt->bio->bi_sector = new_sector; + pkt->bio->bi_next = NULL; + pkt->bio->bi_flags = 1 << BIO_UPTODATE; + pkt->bio->bi_idx = 0; + + BUG_ON(pkt->bio->bi_rw != (1 << BIO_RW)); + BUG_ON(pkt->bio->bi_vcnt != pkt->frames); + BUG_ON(pkt->bio->bi_size != pkt->frames * CD_FRAMESIZE); + BUG_ON(pkt->bio->bi_end_io != pkt_end_io_packet_write); + BUG_ON(pkt->bio->bi_private != pkt); + + drop_super(sb); + return 1; + +out: + drop_super(sb); + return 0; +#endif +} + +static inline void pkt_set_state(struct packet_data *pkt, enum packet_data_state state) +{ +#if PACKET_DEBUG > 1 + static const char *state_name[] = { + "IDLE", "WAITING", "READ_WAIT", "WRITE_WAIT", "RECOVERY", "FINISHED" + }; + enum packet_data_state old_state = pkt->state; + VPRINTK("pkt %2d : s=%6llx %s -> %s\n", pkt->id, (unsigned long long)pkt->sector, + state_name[old_state], state_name[state]); +#endif + pkt->state = state; +} + +/* + * Scan the work queue to see if we can start a new packet. + * returns non-zero if any work was done. + */ +static int pkt_handle_queue(struct pktcdvd_device *pd) +{ + struct packet_data *pkt, *p; + struct bio *bio = NULL; + sector_t zone = 0; /* Suppress gcc warning */ + struct pkt_rb_node *node, *first_node; + struct rb_node *n; + + VPRINTK("handle_queue\n"); + + atomic_set(&pd->scan_queue, 0); + + if (list_empty(&pd->cdrw.pkt_free_list)) { + VPRINTK("handle_queue: no pkt\n"); + return 0; + } + + /* + * Try to find a zone we are not already working on. + */ + spin_lock(&pd->lock); + first_node = pkt_rbtree_find(pd, pd->current_sector); + if (!first_node) { + n = rb_first(&pd->bio_queue); + if (n) + first_node = rb_entry(n, struct pkt_rb_node, rb_node); + } + node = first_node; + while (node) { + bio = node->bio; + zone = ZONE(bio->bi_sector, pd); + list_for_each_entry(p, &pd->cdrw.pkt_active_list, list) { + if (p->sector == zone) + goto try_next_bio; + } + break; +try_next_bio: + node = pkt_rbtree_next(node); + if (!node) { + n = rb_first(&pd->bio_queue); + if (n) + node = rb_entry(n, struct pkt_rb_node, rb_node); + } + if (node == first_node) + node = NULL; + } + spin_unlock(&pd->lock); + if (!bio) { + VPRINTK("handle_queue: no bio\n"); + return 0; + } + + pkt = pkt_get_packet_data(pd, zone); + BUG_ON(!pkt); + + pd->current_sector = zone + pd->settings.size; + pkt->sector = zone; + pkt->frames = pd->settings.size >> 2; + BUG_ON(pkt->frames > PACKET_MAX_SIZE); + pkt->write_size = 0; + + /* + * Scan work queue for bios in the same zone and link them + * to this packet. + */ + spin_lock(&pd->lock); + VPRINTK("pkt_handle_queue: looking for zone %llx\n", (unsigned long long)zone); + while ((node = pkt_rbtree_find(pd, zone)) != NULL) { + bio = node->bio; + VPRINTK("pkt_handle_queue: found zone=%llx\n", + (unsigned long long)ZONE(bio->bi_sector, pd)); + if (ZONE(bio->bi_sector, pd) != zone) + break; + pkt_rbtree_erase(pd, node); + spin_lock(&pkt->lock); + pkt_add_list_last(bio, &pkt->orig_bios, &pkt->orig_bios_tail); + pkt->write_size += bio->bi_size / CD_FRAMESIZE; + spin_unlock(&pkt->lock); + } + spin_unlock(&pd->lock); + + pkt->sleep_time = max(PACKET_WAIT_TIME, 1); + pkt_set_state(pkt, PACKET_WAITING_STATE); + atomic_set(&pkt->run_sm, 1); + + spin_lock(&pd->cdrw.active_list_lock); + list_add(&pkt->list, &pd->cdrw.pkt_active_list); + spin_unlock(&pd->cdrw.active_list_lock); + + return 1; +} + +/* + * Assemble a bio to write one packet and queue the bio for processing + * by the underlying block device. + */ +static void pkt_start_write(struct pktcdvd_device *pd, struct packet_data *pkt) +{ + struct bio *bio; + struct page *pages[PACKET_MAX_SIZE]; + int offsets[PACKET_MAX_SIZE]; + int f; + int frames_write; + + for (f = 0; f < pkt->frames; f++) { + pages[f] = pkt->pages[(f * CD_FRAMESIZE) / PAGE_SIZE]; + offsets[f] = (f * CD_FRAMESIZE) % PAGE_SIZE; + } + + /* + * Fill-in pages[] and offsets[] with data from orig_bios. + */ + frames_write = 0; + spin_lock(&pkt->lock); + for (bio = pkt->orig_bios; bio; bio = bio->bi_next) { + int segment = bio->bi_idx; + int src_offs = 0; + int first_frame = (bio->bi_sector - pkt->sector) / (CD_FRAMESIZE >> 9); + int num_frames = bio->bi_size / CD_FRAMESIZE; + BUG_ON(first_frame < 0); + BUG_ON(first_frame + num_frames > pkt->frames); + for (f = first_frame; f < first_frame + num_frames; f++) { + struct bio_vec *src_bvl = bio_iovec_idx(bio, segment); + + while (src_offs >= src_bvl->bv_len) { + src_offs -= src_bvl->bv_len; + segment++; + BUG_ON(segment >= bio->bi_vcnt); + src_bvl = bio_iovec_idx(bio, segment); + } + + if (src_bvl->bv_len - src_offs >= CD_FRAMESIZE) { + pages[f] = src_bvl->bv_page; + offsets[f] = src_bvl->bv_offset + src_offs; + } else { + pkt_copy_bio_data(bio, segment, src_offs, + pages[f], offsets[f]); + } + src_offs += CD_FRAMESIZE; + frames_write++; + } + } + pkt_set_state(pkt, PACKET_WRITE_WAIT_STATE); + spin_unlock(&pkt->lock); + + VPRINTK("pkt_start_write: Writing %d frames for zone %llx\n", + frames_write, (unsigned long long)pkt->sector); + BUG_ON(frames_write != pkt->write_size); + + if (test_bit(PACKET_MERGE_SEGS, &pd->flags) || (pkt->write_size < pkt->frames)) { + pkt_make_local_copy(pkt, pages, offsets); + pkt->cache_valid = 1; + } else { + pkt->cache_valid = 0; + } + + /* Start the write request */ + bio_init(pkt->w_bio); + pkt->w_bio->bi_max_vecs = PACKET_MAX_SIZE; + pkt->w_bio->bi_sector = pkt->sector; + pkt->w_bio->bi_bdev = pd->bdev; + pkt->w_bio->bi_end_io = pkt_end_io_packet_write; + pkt->w_bio->bi_private = pkt; + for (f = 0; f < pkt->frames; f++) { + if ((f + 1 < pkt->frames) && (pages[f + 1] == pages[f]) && + (offsets[f + 1] = offsets[f] + CD_FRAMESIZE)) { + if (!bio_add_page(pkt->w_bio, pages[f], CD_FRAMESIZE * 2, offsets[f])) + BUG(); + f++; + } else { + if (!bio_add_page(pkt->w_bio, pages[f], CD_FRAMESIZE, offsets[f])) + BUG(); + } + } + VPRINTK("pktcdvd: vcnt=%d\n", pkt->w_bio->bi_vcnt); + + atomic_set(&pkt->io_wait, 1); + pkt->w_bio->bi_rw = WRITE; + pkt_queue_bio(pd, pkt->w_bio, 0); +} + +static void pkt_finish_packet(struct packet_data *pkt, int uptodate) +{ + struct bio *bio, *next; + + if (!uptodate) + pkt->cache_valid = 0; + + /* Finish all bios corresponding to this packet */ + bio = pkt->orig_bios; + while (bio) { + next = bio->bi_next; + bio->bi_next = NULL; + bio_endio(bio, bio->bi_size, uptodate ? 0 : -EIO); + bio = next; + } + pkt->orig_bios = pkt->orig_bios_tail = NULL; +} + +static void pkt_run_state_machine(struct pktcdvd_device *pd, struct packet_data *pkt) +{ + int uptodate; + + VPRINTK("run_state_machine: pkt %d\n", pkt->id); + + for (;;) { + switch (pkt->state) { + case PACKET_WAITING_STATE: + if ((pkt->write_size < pkt->frames) && (pkt->sleep_time > 0)) + return; + + pkt->sleep_time = 0; + pkt_gather_data(pd, pkt); + pkt_set_state(pkt, PACKET_READ_WAIT_STATE); + break; + + case PACKET_READ_WAIT_STATE: + if (atomic_read(&pkt->io_wait) > 0) + return; + + if (atomic_read(&pkt->io_errors) > 0) { + pkt_set_state(pkt, PACKET_RECOVERY_STATE); + } else { + pkt_start_write(pd, pkt); + } + break; + + case PACKET_WRITE_WAIT_STATE: + if (atomic_read(&pkt->io_wait) > 0) + return; + + if (test_bit(BIO_UPTODATE, &pkt->w_bio->bi_flags)) { + pkt_set_state(pkt, PACKET_FINISHED_STATE); + } else { + pkt_set_state(pkt, PACKET_RECOVERY_STATE); + } + break; + + case PACKET_RECOVERY_STATE: + if (pkt_start_recovery(pkt)) { + pkt_start_write(pd, pkt); + } else { + VPRINTK("No recovery possible\n"); + pkt_set_state(pkt, PACKET_FINISHED_STATE); + } + break; + + case PACKET_FINISHED_STATE: + uptodate = test_bit(BIO_UPTODATE, &pkt->w_bio->bi_flags); + pkt_finish_packet(pkt, uptodate); + return; + + default: + BUG(); + break; + } + } +} + +static void pkt_handle_packets(struct pktcdvd_device *pd) +{ + struct packet_data *pkt, *next; + + VPRINTK("pkt_handle_packets\n"); + + /* + * Run state machine for active packets + */ + list_for_each_entry(pkt, &pd->cdrw.pkt_active_list, list) { + if (atomic_read(&pkt->run_sm) > 0) { + atomic_set(&pkt->run_sm, 0); + pkt_run_state_machine(pd, pkt); + } + } + + /* + * Move no longer active packets to the free list + */ + spin_lock(&pd->cdrw.active_list_lock); + list_for_each_entry_safe(pkt, next, &pd->cdrw.pkt_active_list, list) { + if (pkt->state == PACKET_FINISHED_STATE) { + list_del(&pkt->list); + pkt_put_packet_data(pd, pkt); + pkt_set_state(pkt, PACKET_IDLE_STATE); + atomic_set(&pd->scan_queue, 1); + } + } + spin_unlock(&pd->cdrw.active_list_lock); +} + +static void pkt_count_states(struct pktcdvd_device *pd, int *states) +{ + struct packet_data *pkt; + int i; + + for (i = 0; i <= PACKET_NUM_STATES; i++) + states[i] = 0; + + spin_lock(&pd->cdrw.active_list_lock); + list_for_each_entry(pkt, &pd->cdrw.pkt_active_list, list) { + states[pkt->state]++; + } + spin_unlock(&pd->cdrw.active_list_lock); +} + +/* + * kcdrwd is woken up when writes have been queued for one of our + * registered devices + */ +static int kcdrwd(void *foobar) +{ + struct pktcdvd_device *pd = foobar; + struct packet_data *pkt; + long min_sleep_time, residue; + + set_user_nice(current, -20); + + for (;;) { + DECLARE_WAITQUEUE(wait, current); + + /* + * Wait until there is something to do + */ + add_wait_queue(&pd->wqueue, &wait); + for (;;) { + set_current_state(TASK_INTERRUPTIBLE); + + /* Check if we need to run pkt_handle_queue */ + if (atomic_read(&pd->scan_queue) > 0) + goto work_to_do; + + /* Check if we need to run the state machine for some packet */ + list_for_each_entry(pkt, &pd->cdrw.pkt_active_list, list) { + if (atomic_read(&pkt->run_sm) > 0) + goto work_to_do; + } + + /* Check if we need to process the iosched queues */ + if (atomic_read(&pd->iosched.attention) != 0) + goto work_to_do; + + /* Otherwise, go to sleep */ + if (PACKET_DEBUG > 1) { + int states[PACKET_NUM_STATES]; + pkt_count_states(pd, states); + VPRINTK("kcdrwd: i:%d ow:%d rw:%d ww:%d rec:%d fin:%d\n", + states[0], states[1], states[2], states[3], + states[4], states[5]); + } + + min_sleep_time = MAX_SCHEDULE_TIMEOUT; + list_for_each_entry(pkt, &pd->cdrw.pkt_active_list, list) { + if (pkt->sleep_time && pkt->sleep_time < min_sleep_time) + min_sleep_time = pkt->sleep_time; + } + + generic_unplug_device(bdev_get_queue(pd->bdev)); + + VPRINTK("kcdrwd: sleeping\n"); + residue = schedule_timeout(min_sleep_time); + VPRINTK("kcdrwd: wake up\n"); + + /* make swsusp happy with our thread */ + if (current->flags & PF_FREEZE) + refrigerator(PF_FREEZE); + + list_for_each_entry(pkt, &pd->cdrw.pkt_active_list, list) { + if (!pkt->sleep_time) + continue; + pkt->sleep_time -= min_sleep_time - residue; + if (pkt->sleep_time <= 0) { + pkt->sleep_time = 0; + atomic_inc(&pkt->run_sm); + } + } + + if (signal_pending(current)) { + flush_signals(current); + } + if (kthread_should_stop()) + break; + } +work_to_do: + set_current_state(TASK_RUNNING); + remove_wait_queue(&pd->wqueue, &wait); + + if (kthread_should_stop()) + break; + + /* + * if pkt_handle_queue returns true, we can queue + * another request. + */ + while (pkt_handle_queue(pd)) + ; + + /* + * Handle packet state machine + */ + pkt_handle_packets(pd); + + /* + * Handle iosched queues + */ + pkt_iosched_process_queue(pd); + } + + return 0; +} + +static void pkt_print_settings(struct pktcdvd_device *pd) +{ + printk("pktcdvd: %s packets, ", pd->settings.fp ? "Fixed" : "Variable"); + printk("%u blocks, ", pd->settings.size >> 2); + printk("Mode-%c disc\n", pd->settings.block_mode == 8 ? '1' : '2'); +} + +static int pkt_mode_sense(struct pktcdvd_device *pd, struct packet_command *cgc, int page_code, int page_control) +{ + memset(cgc->cmd, 0, sizeof(cgc->cmd)); + + cgc->cmd[0] = GPCMD_MODE_SENSE_10; + cgc->cmd[2] = page_code | (page_control << 6); + cgc->cmd[7] = cgc->buflen >> 8; + cgc->cmd[8] = cgc->buflen & 0xff; + cgc->data_direction = CGC_DATA_READ; + return pkt_generic_packet(pd, cgc); +} + +static int pkt_mode_select(struct pktcdvd_device *pd, struct packet_command *cgc) +{ + memset(cgc->cmd, 0, sizeof(cgc->cmd)); + memset(cgc->buffer, 0, 2); + cgc->cmd[0] = GPCMD_MODE_SELECT_10; + cgc->cmd[1] = 0x10; /* PF */ + cgc->cmd[7] = cgc->buflen >> 8; + cgc->cmd[8] = cgc->buflen & 0xff; + cgc->data_direction = CGC_DATA_WRITE; + return pkt_generic_packet(pd, cgc); +} + +static int pkt_get_disc_info(struct pktcdvd_device *pd, disc_information *di) +{ + struct packet_command cgc; + int ret; + + /* set up command and get the disc info */ + init_cdrom_command(&cgc, di, sizeof(*di), CGC_DATA_READ); + cgc.cmd[0] = GPCMD_READ_DISC_INFO; + cgc.cmd[8] = cgc.buflen = 2; + cgc.quiet = 1; + + if ((ret = pkt_generic_packet(pd, &cgc))) + return ret; + + /* not all drives have the same disc_info length, so requeue + * packet with the length the drive tells us it can supply + */ + cgc.buflen = be16_to_cpu(di->disc_information_length) + + sizeof(di->disc_information_length); + + if (cgc.buflen > sizeof(disc_information)) + cgc.buflen = sizeof(disc_information); + + cgc.cmd[8] = cgc.buflen; + return pkt_generic_packet(pd, &cgc); +} + +static int pkt_get_track_info(struct pktcdvd_device *pd, __u16 track, __u8 type, track_information *ti) +{ + struct packet_command cgc; + int ret; + + init_cdrom_command(&cgc, ti, 8, CGC_DATA_READ); + cgc.cmd[0] = GPCMD_READ_TRACK_RZONE_INFO; + cgc.cmd[1] = type & 3; + cgc.cmd[4] = (track & 0xff00) >> 8; + cgc.cmd[5] = track & 0xff; + cgc.cmd[8] = 8; + cgc.quiet = 1; + + if ((ret = pkt_generic_packet(pd, &cgc))) + return ret; + + cgc.buflen = be16_to_cpu(ti->track_information_length) + + sizeof(ti->track_information_length); + + if (cgc.buflen > sizeof(track_information)) + cgc.buflen = sizeof(track_information); + + cgc.cmd[8] = cgc.buflen; + return pkt_generic_packet(pd, &cgc); +} + +static int pkt_get_last_written(struct pktcdvd_device *pd, long *last_written) +{ + disc_information di; + track_information ti; + __u32 last_track; + int ret = -1; + + if ((ret = pkt_get_disc_info(pd, &di))) + return ret; + + last_track = (di.last_track_msb << 8) | di.last_track_lsb; + if ((ret = pkt_get_track_info(pd, last_track, 1, &ti))) + return ret; + + /* if this track is blank, try the previous. */ + if (ti.blank) { + last_track--; + if ((ret = pkt_get_track_info(pd, last_track, 1, &ti))) + return ret; + } + + /* if last recorded field is valid, return it. */ + if (ti.lra_v) { + *last_written = be32_to_cpu(ti.last_rec_address); + } else { + /* make it up instead */ + *last_written = be32_to_cpu(ti.track_start) + + be32_to_cpu(ti.track_size); + if (ti.free_blocks) + *last_written -= (be32_to_cpu(ti.free_blocks) + 7); + } + return 0; +} + +/* + * write mode select package based on pd->settings + */ +static int pkt_set_write_settings(struct pktcdvd_device *pd) +{ + struct packet_command cgc; + struct request_sense sense; + write_param_page *wp; + char buffer[128]; + int ret, size; + + /* doesn't apply to DVD+RW or DVD-RAM */ + if ((pd->mmc3_profile == 0x1a) || (pd->mmc3_profile == 0x12)) + return 0; + + memset(buffer, 0, sizeof(buffer)); + init_cdrom_command(&cgc, buffer, sizeof(*wp), CGC_DATA_READ); + cgc.sense = &sense; + if ((ret = pkt_mode_sense(pd, &cgc, GPMODE_WRITE_PARMS_PAGE, 0))) { + pkt_dump_sense(&cgc); + return ret; + } + + size = 2 + ((buffer[0] << 8) | (buffer[1] & 0xff)); + pd->mode_offset = (buffer[6] << 8) | (buffer[7] & 0xff); + if (size > sizeof(buffer)) + size = sizeof(buffer); + + /* + * now get it all + */ + init_cdrom_command(&cgc, buffer, size, CGC_DATA_READ); + cgc.sense = &sense; + if ((ret = pkt_mode_sense(pd, &cgc, GPMODE_WRITE_PARMS_PAGE, 0))) { + pkt_dump_sense(&cgc); + return ret; + } + + /* + * write page is offset header + block descriptor length + */ + wp = (write_param_page *) &buffer[sizeof(struct mode_page_header) + pd->mode_offset]; + + wp->fp = pd->settings.fp; + wp->track_mode = pd->settings.track_mode; + wp->write_type = pd->settings.write_type; + wp->data_block_type = pd->settings.block_mode; + + wp->multi_session = 0; + +#ifdef PACKET_USE_LS + wp->link_size = 7; + wp->ls_v = 1; +#endif + + if (wp->data_block_type == PACKET_BLOCK_MODE1) { + wp->session_format = 0; + wp->subhdr2 = 0x20; + } else if (wp->data_block_type == PACKET_BLOCK_MODE2) { + wp->session_format = 0x20; + wp->subhdr2 = 8; +#if 0 + wp->mcn[0] = 0x80; + memcpy(&wp->mcn[1], PACKET_MCN, sizeof(wp->mcn) - 1); +#endif + } else { + /* + * paranoia + */ + printk("pktcdvd: write mode wrong %d\n", wp->data_block_type); + return 1; + } + wp->packet_size = cpu_to_be32(pd->settings.size >> 2); + + cgc.buflen = cgc.cmd[8] = size; + if ((ret = pkt_mode_select(pd, &cgc))) { + pkt_dump_sense(&cgc); + return ret; + } + + pkt_print_settings(pd); + return 0; +} + +/* + * 0 -- we can write to this track, 1 -- we can't + */ +static int pkt_good_track(track_information *ti) +{ + /* + * only good for CD-RW at the moment, not DVD-RW + */ + + /* + * FIXME: only for FP + */ + if (ti->fp == 0) + return 0; + + /* + * "good" settings as per Mt Fuji. + */ + if (ti->rt == 0 && ti->blank == 0 && ti->packet == 1) + return 0; + + if (ti->rt == 0 && ti->blank == 1 && ti->packet == 1) + return 0; + + if (ti->rt == 1 && ti->blank == 0 && ti->packet == 1) + return 0; + + printk("pktcdvd: bad state %d-%d-%d\n", ti->rt, ti->blank, ti->packet); + return 1; +} + +/* + * 0 -- we can write to this disc, 1 -- we can't + */ +static int pkt_good_disc(struct pktcdvd_device *pd, disc_information *di) +{ + switch (pd->mmc3_profile) { + case 0x0a: /* CD-RW */ + case 0xffff: /* MMC3 not supported */ + break; + case 0x1a: /* DVD+RW */ + case 0x13: /* DVD-RW */ + case 0x12: /* DVD-RAM */ + return 0; + default: + printk("pktcdvd: Wrong disc profile (%x)\n", pd->mmc3_profile); + return 1; + } + + /* + * for disc type 0xff we should probably reserve a new track. + * but i'm not sure, should we leave this to user apps? probably. + */ + if (di->disc_type == 0xff) { + printk("pktcdvd: Unknown disc. No track?\n"); + return 1; + } + + if (di->disc_type != 0x20 && di->disc_type != 0) { + printk("pktcdvd: Wrong disc type (%x)\n", di->disc_type); + return 1; + } + + if (di->erasable == 0) { + printk("pktcdvd: Disc not erasable\n"); + return 1; + } + + if (di->border_status == PACKET_SESSION_RESERVED) { + printk("pktcdvd: Can't write to last track (reserved)\n"); + return 1; + } + + return 0; +} + +static int pkt_probe_settings(struct pktcdvd_device *pd) +{ + struct packet_command cgc; + unsigned char buf[12]; + disc_information di; + track_information ti; + int ret, track; + + init_cdrom_command(&cgc, buf, sizeof(buf), CGC_DATA_READ); + cgc.cmd[0] = GPCMD_GET_CONFIGURATION; + cgc.cmd[8] = 8; + ret = pkt_generic_packet(pd, &cgc); + pd->mmc3_profile = ret ? 0xffff : buf[6] << 8 | buf[7]; + + memset(&di, 0, sizeof(disc_information)); + memset(&ti, 0, sizeof(track_information)); + + if ((ret = pkt_get_disc_info(pd, &di))) { + printk("failed get_disc\n"); + return ret; + } + + if (pkt_good_disc(pd, &di)) + return -ENXIO; + + switch (pd->mmc3_profile) { + case 0x1a: /* DVD+RW */ + printk("pktcdvd: inserted media is DVD+RW\n"); + break; + case 0x13: /* DVD-RW */ + printk("pktcdvd: inserted media is DVD-RW\n"); + break; + case 0x12: /* DVD-RAM */ + printk("pktcdvd: inserted media is DVD-RAM\n"); + break; + default: + printk("pktcdvd: inserted media is CD-R%s\n", di.erasable ? "W" : ""); + break; + } + pd->type = di.erasable ? PACKET_CDRW : PACKET_CDR; + + track = 1; /* (di.last_track_msb << 8) | di.last_track_lsb; */ + if ((ret = pkt_get_track_info(pd, track, 1, &ti))) { + printk("pktcdvd: failed get_track\n"); + return ret; + } + + if (pkt_good_track(&ti)) { + printk("pktcdvd: can't write to this track\n"); + return -ENXIO; + } + + /* + * we keep packet size in 512 byte units, makes it easier to + * deal with request calculations. + */ + pd->settings.size = be32_to_cpu(ti.fixed_packet_size) << 2; + if (pd->settings.size == 0) { + printk("pktcdvd: detected zero packet size!\n"); + pd->settings.size = 128; + } + pd->settings.fp = ti.fp; + pd->offset = (be32_to_cpu(ti.track_start) << 2) & (pd->settings.size - 1); + + if (ti.nwa_v) { + pd->nwa = be32_to_cpu(ti.next_writable); + set_bit(PACKET_NWA_VALID, &pd->flags); + } + + /* + * in theory we could use lra on -RW media as well and just zero + * blocks that haven't been written yet, but in practice that + * is just a no-go. we'll use that for -R, naturally. + */ + if (ti.lra_v) { + pd->lra = be32_to_cpu(ti.last_rec_address); + set_bit(PACKET_LRA_VALID, &pd->flags); + } else { + pd->lra = 0xffffffff; + set_bit(PACKET_LRA_VALID, &pd->flags); + } + + /* + * fine for now + */ + pd->settings.link_loss = 7; + pd->settings.write_type = 0; /* packet */ + pd->settings.track_mode = ti.track_mode; + + /* + * mode1 or mode2 disc + */ + switch (ti.data_mode) { + case PACKET_MODE1: + pd->settings.block_mode = PACKET_BLOCK_MODE1; + break; + case PACKET_MODE2: + pd->settings.block_mode = PACKET_BLOCK_MODE2; + break; + default: + printk("pktcdvd: unknown data mode\n"); + return 1; + } + return 0; +} + +/* + * enable/disable write caching on drive + */ +static int pkt_write_caching(struct pktcdvd_device *pd, int set) +{ + struct packet_command cgc; + struct request_sense sense; + unsigned char buf[64]; + int ret; + + memset(buf, 0, sizeof(buf)); + init_cdrom_command(&cgc, buf, sizeof(buf), CGC_DATA_READ); + cgc.sense = &sense; + cgc.buflen = pd->mode_offset + 12; + + /* + * caching mode page might not be there, so quiet this command + */ + cgc.quiet = 1; + + if ((ret = pkt_mode_sense(pd, &cgc, GPMODE_WCACHING_PAGE, 0))) + return ret; + + buf[pd->mode_offset + 10] |= (!!set << 2); + + cgc.buflen = cgc.cmd[8] = 2 + ((buf[0] << 8) | (buf[1] & 0xff)); + ret = pkt_mode_select(pd, &cgc); + if (ret) { + printk("pktcdvd: write caching control failed\n"); + pkt_dump_sense(&cgc); + } else if (!ret && set) + printk("pktcdvd: enabled write caching on %s\n", pd->name); + return ret; +} + +static int pkt_lock_door(struct pktcdvd_device *pd, int lockflag) +{ + struct packet_command cgc; + + init_cdrom_command(&cgc, NULL, 0, CGC_DATA_NONE); + cgc.cmd[0] = GPCMD_PREVENT_ALLOW_MEDIUM_REMOVAL; + cgc.cmd[4] = lockflag ? 1 : 0; + return pkt_generic_packet(pd, &cgc); +} + +/* + * Returns drive maximum write speed + */ +static int pkt_get_max_speed(struct pktcdvd_device *pd, unsigned *write_speed) +{ + struct packet_command cgc; + struct request_sense sense; + unsigned char buf[256+18]; + unsigned char *cap_buf; + int ret, offset; + + memset(buf, 0, sizeof(buf)); + cap_buf = &buf[sizeof(struct mode_page_header) + pd->mode_offset]; + init_cdrom_command(&cgc, buf, sizeof(buf), CGC_DATA_UNKNOWN); + cgc.sense = &sense; + + ret = pkt_mode_sense(pd, &cgc, GPMODE_CAPABILITIES_PAGE, 0); + if (ret) { + cgc.buflen = pd->mode_offset + cap_buf[1] + 2 + + sizeof(struct mode_page_header); + ret = pkt_mode_sense(pd, &cgc, GPMODE_CAPABILITIES_PAGE, 0); + if (ret) { + pkt_dump_sense(&cgc); + return ret; + } + } + + offset = 20; /* Obsoleted field, used by older drives */ + if (cap_buf[1] >= 28) + offset = 28; /* Current write speed selected */ + if (cap_buf[1] >= 30) { + /* If the drive reports at least one "Logical Unit Write + * Speed Performance Descriptor Block", use the information + * in the first block. (contains the highest speed) + */ + int num_spdb = (cap_buf[30] << 8) + cap_buf[31]; + if (num_spdb > 0) + offset = 34; + } + + *write_speed = (cap_buf[offset] << 8) | cap_buf[offset + 1]; + return 0; +} + +/* These tables from cdrecord - I don't have orange book */ +/* standard speed CD-RW (1-4x) */ +static char clv_to_speed[16] = { + /* 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 */ + 0, 2, 4, 6, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 +}; +/* high speed CD-RW (-10x) */ +static char hs_clv_to_speed[16] = { + /* 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 */ + 0, 2, 4, 6, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 +}; +/* ultra high speed CD-RW */ +static char us_clv_to_speed[16] = { + /* 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 */ + 0, 2, 4, 8, 0, 0,16, 0,24,32,40,48, 0, 0, 0, 0 +}; + +/* + * reads the maximum media speed from ATIP + */ +static int pkt_media_speed(struct pktcdvd_device *pd, unsigned *speed) +{ + struct packet_command cgc; + struct request_sense sense; + unsigned char buf[64]; + unsigned int size, st, sp; + int ret; + + init_cdrom_command(&cgc, buf, 2, CGC_DATA_READ); + cgc.sense = &sense; + cgc.cmd[0] = GPCMD_READ_TOC_PMA_ATIP; + cgc.cmd[1] = 2; + cgc.cmd[2] = 4; /* READ ATIP */ + cgc.cmd[8] = 2; + ret = pkt_generic_packet(pd, &cgc); + if (ret) { + pkt_dump_sense(&cgc); + return ret; + } + size = ((unsigned int) buf[0]<<8) + buf[1] + 2; + if (size > sizeof(buf)) + size = sizeof(buf); + + init_cdrom_command(&cgc, buf, size, CGC_DATA_READ); + cgc.sense = &sense; + cgc.cmd[0] = GPCMD_READ_TOC_PMA_ATIP; + cgc.cmd[1] = 2; + cgc.cmd[2] = 4; + cgc.cmd[8] = size; + ret = pkt_generic_packet(pd, &cgc); + if (ret) { + pkt_dump_sense(&cgc); + return ret; + } + + if (!buf[6] & 0x40) { + printk("pktcdvd: Disc type is not CD-RW\n"); + return 1; + } + if (!buf[6] & 0x4) { + printk("pktcdvd: A1 values on media are not valid, maybe not CDRW?\n"); + return 1; + } + + st = (buf[6] >> 3) & 0x7; /* disc sub-type */ + + sp = buf[16] & 0xf; /* max speed from ATIP A1 field */ + + /* Info from cdrecord */ + switch (st) { + case 0: /* standard speed */ + *speed = clv_to_speed[sp]; + break; + case 1: /* high speed */ + *speed = hs_clv_to_speed[sp]; + break; + case 2: /* ultra high speed */ + *speed = us_clv_to_speed[sp]; + break; + default: + printk("pktcdvd: Unknown disc sub-type %d\n",st); + return 1; + } + if (*speed) { + printk("pktcdvd: Max. media speed: %d\n",*speed); + return 0; + } else { + printk("pktcdvd: Unknown speed %d for sub-type %d\n",sp,st); + return 1; + } +} + +static int pkt_perform_opc(struct pktcdvd_device *pd) +{ + struct packet_command cgc; + struct request_sense sense; + int ret; + + VPRINTK("pktcdvd: Performing OPC\n"); + + init_cdrom_command(&cgc, NULL, 0, CGC_DATA_NONE); + cgc.sense = &sense; + cgc.timeout = 60*HZ; + cgc.cmd[0] = GPCMD_SEND_OPC; + cgc.cmd[1] = 1; + if ((ret = pkt_generic_packet(pd, &cgc))) + pkt_dump_sense(&cgc); + return ret; +} + +static int pkt_open_write(struct pktcdvd_device *pd) +{ + int ret; + unsigned int write_speed, media_write_speed, read_speed; + + if ((ret = pkt_probe_settings(pd))) { + DPRINTK("pktcdvd: %s failed probe\n", pd->name); + return -EIO; + } + + if ((ret = pkt_set_write_settings(pd))) { + DPRINTK("pktcdvd: %s failed saving write settings\n", pd->name); + return -EIO; + } + + pkt_write_caching(pd, USE_WCACHING); + + if ((ret = pkt_get_max_speed(pd, &write_speed))) + write_speed = 16 * 177; + switch (pd->mmc3_profile) { + case 0x13: /* DVD-RW */ + case 0x1a: /* DVD+RW */ + case 0x12: /* DVD-RAM */ + DPRINTK("pktcdvd: write speed %ukB/s\n", write_speed); + break; + default: + if ((ret = pkt_media_speed(pd, &media_write_speed))) + media_write_speed = 16; + write_speed = min(write_speed, media_write_speed * 177); + DPRINTK("pktcdvd: write speed %ux\n", write_speed / 176); + break; + } + read_speed = write_speed; + + if ((ret = pkt_set_speed(pd, write_speed, read_speed))) { + DPRINTK("pktcdvd: %s couldn't set write speed\n", pd->name); + return -EIO; + } + pd->write_speed = write_speed; + pd->read_speed = read_speed; + + if ((ret = pkt_perform_opc(pd))) { + DPRINTK("pktcdvd: %s Optimum Power Calibration failed\n", pd->name); + } + + return 0; +} + +/* + * called at open time. + */ +static int pkt_open_dev(struct pktcdvd_device *pd, int write) +{ + int ret; + long lba; + request_queue_t *q; + + /* + * We need to re-open the cdrom device without O_NONBLOCK to be able + * to read/write from/to it. It is already opened in O_NONBLOCK mode + * so bdget() can't fail. + */ + bdget(pd->bdev->bd_dev); + if ((ret = blkdev_get(pd->bdev, FMODE_READ, O_RDONLY))) + goto out; + + if ((ret = pkt_get_last_written(pd, &lba))) { + printk("pktcdvd: pkt_get_last_written failed\n"); + goto out_putdev; + } + + set_capacity(pd->disk, lba << 2); + set_capacity(pd->bdev->bd_disk, lba << 2); + bd_set_size(pd->bdev, (loff_t)lba << 11); + + q = bdev_get_queue(pd->bdev); + if (write) { + if ((ret = pkt_open_write(pd))) + goto out_putdev; + /* + * Some CDRW drives can not handle writes larger than one packet, + * even if the size is a multiple of the packet size. + */ + spin_lock_irq(q->queue_lock); + blk_queue_max_sectors(q, pd->settings.size); + spin_unlock_irq(q->queue_lock); + set_bit(PACKET_WRITABLE, &pd->flags); + } else { + pkt_set_speed(pd, MAX_SPEED, MAX_SPEED); + clear_bit(PACKET_WRITABLE, &pd->flags); + } + + if ((ret = pkt_set_segment_merging(pd, q))) + goto out_putdev; + + if (write) + printk("pktcdvd: %lukB available on disc\n", lba << 1); + + return 0; + +out_putdev: + blkdev_put(pd->bdev); +out: + return ret; +} + +/* + * called when the device is closed. makes sure that the device flushes + * the internal cache before we close. + */ +static void pkt_release_dev(struct pktcdvd_device *pd, int flush) +{ + if (flush && pkt_flush_cache(pd)) + DPRINTK("pktcdvd: %s not flushing cache\n", pd->name); + + pkt_lock_door(pd, 0); + + pkt_set_speed(pd, MAX_SPEED, MAX_SPEED); + blkdev_put(pd->bdev); +} + +static struct pktcdvd_device *pkt_find_dev_from_minor(int dev_minor) +{ + if (dev_minor >= MAX_WRITERS) + return NULL; + return pkt_devs[dev_minor]; +} + +static int pkt_open(struct inode *inode, struct file *file) +{ + struct pktcdvd_device *pd = NULL; + int ret; + + VPRINTK("pktcdvd: entering open\n"); + + down(&ctl_mutex); + pd = pkt_find_dev_from_minor(iminor(inode)); + if (!pd) { + ret = -ENODEV; + goto out; + } + BUG_ON(pd->refcnt < 0); + + pd->refcnt++; + if (pd->refcnt == 1) { + if (pkt_open_dev(pd, file->f_mode & FMODE_WRITE)) { + ret = -EIO; + goto out_dec; + } + /* + * needed here as well, since ext2 (among others) may change + * the blocksize at mount time + */ + set_blocksize(inode->i_bdev, CD_FRAMESIZE); + } + + up(&ctl_mutex); + return 0; + +out_dec: + pd->refcnt--; +out: + VPRINTK("pktcdvd: failed open (%d)\n", ret); + up(&ctl_mutex); + return ret; +} + +static int pkt_close(struct inode *inode, struct file *file) +{ + struct pktcdvd_device *pd = inode->i_bdev->bd_disk->private_data; + int ret = 0; + + down(&ctl_mutex); + pd->refcnt--; + BUG_ON(pd->refcnt < 0); + if (pd->refcnt == 0) { + int flush = test_bit(PACKET_WRITABLE, &pd->flags); + pkt_release_dev(pd, flush); + } + up(&ctl_mutex); + return ret; +} + + +static void *psd_pool_alloc(unsigned int __nocast gfp_mask, void *data) +{ + return kmalloc(sizeof(struct packet_stacked_data), gfp_mask); +} + +static void psd_pool_free(void *ptr, void *data) +{ + kfree(ptr); +} + +static int pkt_end_io_read_cloned(struct bio *bio, unsigned int bytes_done, int err) +{ + struct packet_stacked_data *psd = bio->bi_private; + struct pktcdvd_device *pd = psd->pd; + + if (bio->bi_size) + return 1; + + bio_put(bio); + bio_endio(psd->bio, psd->bio->bi_size, err); + mempool_free(psd, psd_pool); + pkt_bio_finished(pd); + return 0; +} + +static int pkt_make_request(request_queue_t *q, struct bio *bio) +{ + struct pktcdvd_device *pd; + char b[BDEVNAME_SIZE]; + sector_t zone; + struct packet_data *pkt; + int was_empty, blocked_bio; + struct pkt_rb_node *node; + + pd = q->queuedata; + if (!pd) { + printk("pktcdvd: %s incorrect request queue\n", bdevname(bio->bi_bdev, b)); + goto end_io; + } + + /* + * Clone READ bios so we can have our own bi_end_io callback. + */ + if (bio_data_dir(bio) == READ) { + struct bio *cloned_bio = bio_clone(bio, GFP_NOIO); + struct packet_stacked_data *psd = mempool_alloc(psd_pool, GFP_NOIO); + + psd->pd = pd; + psd->bio = bio; + cloned_bio->bi_bdev = pd->bdev; + cloned_bio->bi_private = psd; + cloned_bio->bi_end_io = pkt_end_io_read_cloned; + pd->stats.secs_r += bio->bi_size >> 9; + pkt_queue_bio(pd, cloned_bio, 1); + return 0; + } + + if (!test_bit(PACKET_WRITABLE, &pd->flags)) { + printk("pktcdvd: WRITE for ro device %s (%llu)\n", + pd->name, (unsigned long long)bio->bi_sector); + goto end_io; + } + + if (!bio->bi_size || (bio->bi_size % CD_FRAMESIZE)) { + printk("pktcdvd: wrong bio size\n"); + goto end_io; + } + + blk_queue_bounce(q, &bio); + + zone = ZONE(bio->bi_sector, pd); + VPRINTK("pkt_make_request: start = %6llx stop = %6llx\n", + (unsigned long long)bio->bi_sector, + (unsigned long long)(bio->bi_sector + bio_sectors(bio))); + + /* Check if we have to split the bio */ + { + struct bio_pair *bp; + sector_t last_zone; + int first_sectors; + + last_zone = ZONE(bio->bi_sector + bio_sectors(bio) - 1, pd); + if (last_zone != zone) { + BUG_ON(last_zone != zone + pd->settings.size); + first_sectors = last_zone - bio->bi_sector; + bp = bio_split(bio, bio_split_pool, first_sectors); + BUG_ON(!bp); + pkt_make_request(q, &bp->bio1); + pkt_make_request(q, &bp->bio2); + bio_pair_release(bp); + return 0; + } + } + + /* + * If we find a matching packet in state WAITING or READ_WAIT, we can + * just append this bio to that packet. + */ + spin_lock(&pd->cdrw.active_list_lock); + blocked_bio = 0; + list_for_each_entry(pkt, &pd->cdrw.pkt_active_list, list) { + if (pkt->sector == zone) { + spin_lock(&pkt->lock); + if ((pkt->state == PACKET_WAITING_STATE) || + (pkt->state == PACKET_READ_WAIT_STATE)) { + pkt_add_list_last(bio, &pkt->orig_bios, + &pkt->orig_bios_tail); + pkt->write_size += bio->bi_size / CD_FRAMESIZE; + if ((pkt->write_size >= pkt->frames) && + (pkt->state == PACKET_WAITING_STATE)) { + atomic_inc(&pkt->run_sm); + wake_up(&pd->wqueue); + } + spin_unlock(&pkt->lock); + spin_unlock(&pd->cdrw.active_list_lock); + return 0; + } else { + blocked_bio = 1; + } + spin_unlock(&pkt->lock); + } + } + spin_unlock(&pd->cdrw.active_list_lock); + + /* + * No matching packet found. Store the bio in the work queue. + */ + node = mempool_alloc(pd->rb_pool, GFP_NOIO); + BUG_ON(!node); + node->bio = bio; + spin_lock(&pd->lock); + BUG_ON(pd->bio_queue_size < 0); + was_empty = (pd->bio_queue_size == 0); + pkt_rbtree_insert(pd, node); + spin_unlock(&pd->lock); + + /* + * Wake up the worker thread. + */ + atomic_set(&pd->scan_queue, 1); + if (was_empty) { + /* This wake_up is required for correct operation */ + wake_up(&pd->wqueue); + } else if (!list_empty(&pd->cdrw.pkt_free_list) && !blocked_bio) { + /* + * This wake up is not required for correct operation, + * but improves performance in some cases. + */ + wake_up(&pd->wqueue); + } + return 0; +end_io: + bio_io_error(bio, bio->bi_size); + return 0; +} + + + +static int pkt_merge_bvec(request_queue_t *q, struct bio *bio, struct bio_vec *bvec) +{ + struct pktcdvd_device *pd = q->queuedata; + sector_t zone = ZONE(bio->bi_sector, pd); + int used = ((bio->bi_sector - zone) << 9) + bio->bi_size; + int remaining = (pd->settings.size << 9) - used; + int remaining2; + + /* + * A bio <= PAGE_SIZE must be allowed. If it crosses a packet + * boundary, pkt_make_request() will split the bio. + */ + remaining2 = PAGE_SIZE - bio->bi_size; + remaining = max(remaining, remaining2); + + BUG_ON(remaining < 0); + return remaining; +} + +static void pkt_init_queue(struct pktcdvd_device *pd) +{ + request_queue_t *q = pd->disk->queue; + + blk_queue_make_request(q, pkt_make_request); + blk_queue_hardsect_size(q, CD_FRAMESIZE); + blk_queue_max_sectors(q, PACKET_MAX_SECTORS); + blk_queue_merge_bvec(q, pkt_merge_bvec); + q->queuedata = pd; +} + +static int pkt_seq_show(struct seq_file *m, void *p) +{ + struct pktcdvd_device *pd = m->private; + char *msg; + char bdev_buf[BDEVNAME_SIZE]; + int states[PACKET_NUM_STATES]; + + seq_printf(m, "Writer %s mapped to %s:\n", pd->name, + bdevname(pd->bdev, bdev_buf)); + + seq_printf(m, "\nSettings:\n"); + seq_printf(m, "\tpacket size:\t\t%dkB\n", pd->settings.size / 2); + + if (pd->settings.write_type == 0) + msg = "Packet"; + else + msg = "Unknown"; + seq_printf(m, "\twrite type:\t\t%s\n", msg); + + seq_printf(m, "\tpacket type:\t\t%s\n", pd->settings.fp ? "Fixed" : "Variable"); + seq_printf(m, "\tlink loss:\t\t%d\n", pd->settings.link_loss); + + seq_printf(m, "\ttrack mode:\t\t%d\n", pd->settings.track_mode); + + if (pd->settings.block_mode == PACKET_BLOCK_MODE1) + msg = "Mode 1"; + else if (pd->settings.block_mode == PACKET_BLOCK_MODE2) + msg = "Mode 2"; + else + msg = "Unknown"; + seq_printf(m, "\tblock mode:\t\t%s\n", msg); + + seq_printf(m, "\nStatistics:\n"); + seq_printf(m, "\tpackets started:\t%lu\n", pd->stats.pkt_started); + seq_printf(m, "\tpackets ended:\t\t%lu\n", pd->stats.pkt_ended); + seq_printf(m, "\twritten:\t\t%lukB\n", pd->stats.secs_w >> 1); + seq_printf(m, "\tread gather:\t\t%lukB\n", pd->stats.secs_rg >> 1); + seq_printf(m, "\tread:\t\t\t%lukB\n", pd->stats.secs_r >> 1); + + seq_printf(m, "\nMisc:\n"); + seq_printf(m, "\treference count:\t%d\n", pd->refcnt); + seq_printf(m, "\tflags:\t\t\t0x%lx\n", pd->flags); + seq_printf(m, "\tread speed:\t\t%ukB/s\n", pd->read_speed); + seq_printf(m, "\twrite speed:\t\t%ukB/s\n", pd->write_speed); + seq_printf(m, "\tstart offset:\t\t%lu\n", pd->offset); + seq_printf(m, "\tmode page offset:\t%u\n", pd->mode_offset); + + seq_printf(m, "\nQueue state:\n"); + seq_printf(m, "\tbios queued:\t\t%d\n", pd->bio_queue_size); + seq_printf(m, "\tbios pending:\t\t%d\n", atomic_read(&pd->cdrw.pending_bios)); + seq_printf(m, "\tcurrent sector:\t\t0x%llx\n", (unsigned long long)pd->current_sector); + + pkt_count_states(pd, states); + seq_printf(m, "\tstate:\t\t\ti:%d ow:%d rw:%d ww:%d rec:%d fin:%d\n", + states[0], states[1], states[2], states[3], states[4], states[5]); + + return 0; +} + +static int pkt_seq_open(struct inode *inode, struct file *file) +{ + return single_open(file, pkt_seq_show, PDE(inode)->data); +} + +static struct file_operations pkt_proc_fops = { + .open = pkt_seq_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release +}; + +static int pkt_new_dev(struct pktcdvd_device *pd, dev_t dev) +{ + int i; + int ret = 0; + char b[BDEVNAME_SIZE]; + struct proc_dir_entry *proc; + struct block_device *bdev; + + if (pd->pkt_dev == dev) { + printk("pktcdvd: Recursive setup not allowed\n"); + return -EBUSY; + } + for (i = 0; i < MAX_WRITERS; i++) { + struct pktcdvd_device *pd2 = pkt_devs[i]; + if (!pd2) + continue; + if (pd2->bdev->bd_dev == dev) { + printk("pktcdvd: %s already setup\n", bdevname(pd2->bdev, b)); + return -EBUSY; + } + if (pd2->pkt_dev == dev) { + printk("pktcdvd: Can't chain pktcdvd devices\n"); + return -EBUSY; + } + } + + bdev = bdget(dev); + if (!bdev) + return -ENOMEM; + ret = blkdev_get(bdev, FMODE_READ, O_RDONLY | O_NONBLOCK); + if (ret) + return ret; + + /* This is safe, since we have a reference from open(). */ + __module_get(THIS_MODULE); + + if (!pkt_grow_pktlist(pd, CONFIG_CDROM_PKTCDVD_BUFFERS)) { + printk("pktcdvd: not enough memory for buffers\n"); + ret = -ENOMEM; + goto out_mem; + } + + pd->bdev = bdev; + set_blocksize(bdev, CD_FRAMESIZE); + + pkt_init_queue(pd); + + atomic_set(&pd->cdrw.pending_bios, 0); + pd->cdrw.thread = kthread_run(kcdrwd, pd, "%s", pd->name); + if (IS_ERR(pd->cdrw.thread)) { + printk("pktcdvd: can't start kernel thread\n"); + ret = -ENOMEM; + goto out_thread; + } + + proc = create_proc_entry(pd->name, 0, pkt_proc); + if (proc) { + proc->data = pd; + proc->proc_fops = &pkt_proc_fops; + } + DPRINTK("pktcdvd: writer %s mapped to %s\n", pd->name, bdevname(bdev, b)); + return 0; + +out_thread: + pkt_shrink_pktlist(pd); +out_mem: + blkdev_put(bdev); + /* This is safe: open() is still holding a reference. */ + module_put(THIS_MODULE); + return ret; +} + +static int pkt_ioctl(struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg) +{ + struct pktcdvd_device *pd = inode->i_bdev->bd_disk->private_data; + + VPRINTK("pkt_ioctl: cmd %x, dev %d:%d\n", cmd, imajor(inode), iminor(inode)); + BUG_ON(!pd); + + switch (cmd) { + /* + * forward selected CDROM ioctls to CD-ROM, for UDF + */ + case CDROMMULTISESSION: + case CDROMREADTOCENTRY: + case CDROM_LAST_WRITTEN: + case CDROM_SEND_PACKET: + case SCSI_IOCTL_SEND_COMMAND: + return ioctl_by_bdev(pd->bdev, cmd, arg); + + case CDROMEJECT: + /* + * The door gets locked when the device is opened, so we + * have to unlock it or else the eject command fails. + */ + pkt_lock_door(pd, 0); + return ioctl_by_bdev(pd->bdev, cmd, arg); + + default: + printk("pktcdvd: Unknown ioctl for %s (%x)\n", pd->name, cmd); + return -ENOTTY; + } + + return 0; +} + +static int pkt_media_changed(struct gendisk *disk) +{ + struct pktcdvd_device *pd = disk->private_data; + struct gendisk *attached_disk; + + if (!pd) + return 0; + if (!pd->bdev) + return 0; + attached_disk = pd->bdev->bd_disk; + if (!attached_disk) + return 0; + return attached_disk->fops->media_changed(attached_disk); +} + +static struct block_device_operations pktcdvd_ops = { + .owner = THIS_MODULE, + .open = pkt_open, + .release = pkt_close, + .ioctl = pkt_ioctl, + .media_changed = pkt_media_changed, +}; + +/* + * Set up mapping from pktcdvd device to CD-ROM device. + */ +static int pkt_setup_dev(struct pkt_ctrl_command *ctrl_cmd) +{ + int idx; + int ret = -ENOMEM; + struct pktcdvd_device *pd; + struct gendisk *disk; + dev_t dev = new_decode_dev(ctrl_cmd->dev); + + for (idx = 0; idx < MAX_WRITERS; idx++) + if (!pkt_devs[idx]) + break; + if (idx == MAX_WRITERS) { + printk("pktcdvd: max %d writers supported\n", MAX_WRITERS); + return -EBUSY; + } + + pd = kmalloc(sizeof(struct pktcdvd_device), GFP_KERNEL); + if (!pd) + return ret; + memset(pd, 0, sizeof(struct pktcdvd_device)); + + pd->rb_pool = mempool_create(PKT_RB_POOL_SIZE, pkt_rb_alloc, pkt_rb_free, NULL); + if (!pd->rb_pool) + goto out_mem; + + disk = alloc_disk(1); + if (!disk) + goto out_mem; + pd->disk = disk; + + spin_lock_init(&pd->lock); + spin_lock_init(&pd->iosched.lock); + sprintf(pd->name, "pktcdvd%d", idx); + init_waitqueue_head(&pd->wqueue); + pd->bio_queue = RB_ROOT; + + disk->major = pkt_major; + disk->first_minor = idx; + disk->fops = &pktcdvd_ops; + disk->flags = GENHD_FL_REMOVABLE; + sprintf(disk->disk_name, "pktcdvd%d", idx); + disk->private_data = pd; + disk->queue = blk_alloc_queue(GFP_KERNEL); + if (!disk->queue) + goto out_mem2; + + pd->pkt_dev = MKDEV(disk->major, disk->first_minor); + ret = pkt_new_dev(pd, dev); + if (ret) + goto out_new_dev; + + add_disk(disk); + pkt_devs[idx] = pd; + ctrl_cmd->pkt_dev = new_encode_dev(pd->pkt_dev); + return 0; + +out_new_dev: + blk_put_queue(disk->queue); +out_mem2: + put_disk(disk); +out_mem: + if (pd->rb_pool) + mempool_destroy(pd->rb_pool); + kfree(pd); + return ret; +} + +/* + * Tear down mapping from pktcdvd device to CD-ROM device. + */ +static int pkt_remove_dev(struct pkt_ctrl_command *ctrl_cmd) +{ + struct pktcdvd_device *pd; + int idx; + dev_t pkt_dev = new_decode_dev(ctrl_cmd->pkt_dev); + + for (idx = 0; idx < MAX_WRITERS; idx++) { + pd = pkt_devs[idx]; + if (pd && (pd->pkt_dev == pkt_dev)) + break; + } + if (idx == MAX_WRITERS) { + DPRINTK("pktcdvd: dev not setup\n"); + return -ENXIO; + } + + if (pd->refcnt > 0) + return -EBUSY; + + if (!IS_ERR(pd->cdrw.thread)) + kthread_stop(pd->cdrw.thread); + + blkdev_put(pd->bdev); + + pkt_shrink_pktlist(pd); + + remove_proc_entry(pd->name, pkt_proc); + DPRINTK("pktcdvd: writer %s unmapped\n", pd->name); + + del_gendisk(pd->disk); + blk_put_queue(pd->disk->queue); + put_disk(pd->disk); + + pkt_devs[idx] = NULL; + mempool_destroy(pd->rb_pool); + kfree(pd); + + /* This is safe: open() is still holding a reference. */ + module_put(THIS_MODULE); + return 0; +} + +static void pkt_get_status(struct pkt_ctrl_command *ctrl_cmd) +{ + struct pktcdvd_device *pd = pkt_find_dev_from_minor(ctrl_cmd->dev_index); + if (pd) { + ctrl_cmd->dev = new_encode_dev(pd->bdev->bd_dev); + ctrl_cmd->pkt_dev = new_encode_dev(pd->pkt_dev); + } else { + ctrl_cmd->dev = 0; + ctrl_cmd->pkt_dev = 0; + } + ctrl_cmd->num_devices = MAX_WRITERS; +} + +static int pkt_ctl_ioctl(struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg) +{ + void __user *argp = (void __user *)arg; + struct pkt_ctrl_command ctrl_cmd; + int ret = 0; + + if (cmd != PACKET_CTRL_CMD) + return -ENOTTY; + + if (copy_from_user(&ctrl_cmd, argp, sizeof(struct pkt_ctrl_command))) + return -EFAULT; + + switch (ctrl_cmd.command) { + case PKT_CTRL_CMD_SETUP: + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + down(&ctl_mutex); + ret = pkt_setup_dev(&ctrl_cmd); + up(&ctl_mutex); + break; + case PKT_CTRL_CMD_TEARDOWN: + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + down(&ctl_mutex); + ret = pkt_remove_dev(&ctrl_cmd); + up(&ctl_mutex); + break; + case PKT_CTRL_CMD_STATUS: + down(&ctl_mutex); + pkt_get_status(&ctrl_cmd); + up(&ctl_mutex); + break; + default: + return -ENOTTY; + } + + if (copy_to_user(argp, &ctrl_cmd, sizeof(struct pkt_ctrl_command))) + return -EFAULT; + return ret; +} + + +static struct file_operations pkt_ctl_fops = { + .ioctl = pkt_ctl_ioctl, + .owner = THIS_MODULE, +}; + +static struct miscdevice pkt_misc = { + .minor = MISC_DYNAMIC_MINOR, + .name = "pktcdvd", + .devfs_name = "pktcdvd/control", + .fops = &pkt_ctl_fops +}; + +static int __init pkt_init(void) +{ + int ret; + + psd_pool = mempool_create(PSD_POOL_SIZE, psd_pool_alloc, psd_pool_free, NULL); + if (!psd_pool) + return -ENOMEM; + + ret = register_blkdev(pkt_major, "pktcdvd"); + if (ret < 0) { + printk("pktcdvd: Unable to register block device\n"); + goto out2; + } + if (!pkt_major) + pkt_major = ret; + + ret = misc_register(&pkt_misc); + if (ret) { + printk("pktcdvd: Unable to register misc device\n"); + goto out; + } + + init_MUTEX(&ctl_mutex); + + pkt_proc = proc_mkdir("pktcdvd", proc_root_driver); + + DPRINTK("pktcdvd: %s\n", VERSION_CODE); + return 0; + +out: + unregister_blkdev(pkt_major, "pktcdvd"); +out2: + mempool_destroy(psd_pool); + return ret; +} + +static void __exit pkt_exit(void) +{ + remove_proc_entry("pktcdvd", proc_root_driver); + misc_deregister(&pkt_misc); + unregister_blkdev(pkt_major, "pktcdvd"); + mempool_destroy(psd_pool); +} + +MODULE_DESCRIPTION("Packet writing layer for CD/DVD drives"); +MODULE_AUTHOR("Jens Axboe <axboe@suse.de>"); +MODULE_LICENSE("GPL"); + +module_init(pkt_init); +module_exit(pkt_exit); diff --git a/drivers/block/ps2esdi.c b/drivers/block/ps2esdi.c new file mode 100644 index 00000000000..29548784cb7 --- /dev/null +++ b/drivers/block/ps2esdi.c @@ -0,0 +1,1092 @@ +/* ps2esdi driver based on assembler code by Arindam Banerji, + written by Peter De Schrijver */ +/* Reassuring note to IBM : This driver was NOT developed by vice-versa + engineering the PS/2's BIOS */ +/* Dedicated to Wannes, Tofke, Ykke, Godot, Killroy and all those + other lovely fish out there... */ +/* This code was written during the long and boring WINA + elections 1994 */ +/* Thanks to Arindam Banerij for giving me the source of his driver */ +/* This code may be freely distributed and modified in any way, + as long as these notes remain intact */ + +/* Revised: 05/07/94 by Arindam Banerji (axb@cse.nd.edu) */ +/* Revised: 09/08/94 by Peter De Schrijver (stud11@cc4.kuleuven.ac.be) + Thanks to Arindam Banerij for sending me the docs of the adapter */ + +/* BA Modified for ThinkPad 720 by Boris Ashkinazi */ +/* (bash@vnet.ibm.com) 08/08/95 */ + +/* Modified further for ThinkPad-720C by Uri Blumenthal */ +/* (uri@watson.ibm.com) Sep 11, 1995 */ + +/* TODO : + + Timeouts + + Get disk parameters + + DMA above 16MB + + reset after read/write error + */ + +#define DEVICE_NAME "PS/2 ESDI" + +#include <linux/config.h> +#include <linux/major.h> +#include <linux/errno.h> +#include <linux/wait.h> +#include <linux/interrupt.h> +#include <linux/fs.h> +#include <linux/kernel.h> +#include <linux/genhd.h> +#include <linux/ps2esdi.h> +#include <linux/blkdev.h> +#include <linux/mca-legacy.h> +#include <linux/init.h> +#include <linux/ioport.h> +#include <linux/module.h> + +#include <asm/system.h> +#include <asm/io.h> +#include <asm/dma.h> +#include <asm/mca_dma.h> +#include <asm/uaccess.h> + +#define PS2ESDI_IRQ 14 +#define MAX_HD 2 +#define MAX_RETRIES 5 +#define MAX_16BIT 65536 +#define ESDI_TIMEOUT 0xf000 +#define ESDI_STAT_TIMEOUT 4 + +#define TYPE_0_CMD_BLK_LENGTH 2 +#define TYPE_1_CMD_BLK_LENGTH 4 + +static void reset_ctrl(void); + +static int ps2esdi_geninit(void); + +static void do_ps2esdi_request(request_queue_t * q); + +static void ps2esdi_readwrite(int cmd, struct request *req); + +static void ps2esdi_fill_cmd_block(u_short * cmd_blk, u_short cmd, +u_short cyl, u_short head, u_short sector, u_short length, u_char drive); + +static int ps2esdi_out_cmd_blk(u_short * cmd_blk); + +static void ps2esdi_prep_dma(char *buffer, u_short length, u_char dma_xmode); + +static irqreturn_t ps2esdi_interrupt_handler(int irq, void *dev_id, + struct pt_regs *regs); +static void (*current_int_handler) (u_int) = NULL; +static void ps2esdi_normal_interrupt_handler(u_int); +static void ps2esdi_initial_reset_int_handler(u_int); +static void ps2esdi_geometry_int_handler(u_int); +static int ps2esdi_ioctl(struct inode *inode, struct file *file, + u_int cmd, u_long arg); + +static int ps2esdi_read_status_words(int num_words, int max_words, u_short * buffer); + +static void dump_cmd_complete_status(u_int int_ret_code); + +static void ps2esdi_get_device_cfg(void); + +static void ps2esdi_reset_timer(unsigned long unused); + +static u_int dma_arb_level; /* DMA arbitration level */ + +static DECLARE_WAIT_QUEUE_HEAD(ps2esdi_int); + +static int no_int_yet; +static int ps2esdi_drives; +static u_short io_base; +static struct timer_list esdi_timer = + TIMER_INITIALIZER(ps2esdi_reset_timer, 0, 0); +static int reset_status; +static int ps2esdi_slot = -1; +static int tp720esdi = 0; /* Is it Integrated ESDI of ThinkPad-720? */ +static int intg_esdi = 0; /* If integrated adapter */ +struct ps2esdi_i_struct { + unsigned int head, sect, cyl, wpcom, lzone, ctl; +}; +static DEFINE_SPINLOCK(ps2esdi_lock); +static struct request_queue *ps2esdi_queue; +static struct request *current_req; + +#if 0 +#if 0 /* try both - I don't know which one is better... UB */ +static struct ps2esdi_i_struct ps2esdi_info[MAX_HD] = +{ + {4, 48, 1553, 0, 0, 0}, + {0, 0, 0, 0, 0, 0}}; +#else +static struct ps2esdi_i_struct ps2esdi_info[MAX_HD] = +{ + {64, 32, 161, 0, 0, 0}, + {0, 0, 0, 0, 0, 0}}; +#endif +#endif +static struct ps2esdi_i_struct ps2esdi_info[MAX_HD] = +{ + {0, 0, 0, 0, 0, 0}, + {0, 0, 0, 0, 0, 0}}; + +static struct block_device_operations ps2esdi_fops = +{ + .owner = THIS_MODULE, + .ioctl = ps2esdi_ioctl, +}; + +static struct gendisk *ps2esdi_gendisk[2]; + +/* initialization routine called by ll_rw_blk.c */ +static int __init ps2esdi_init(void) +{ + + int error = 0; + + /* register the device - pass the name and major number */ + if (register_blkdev(PS2ESDI_MAJOR, "ed")) + return -EBUSY; + + /* set up some global information - indicating device specific info */ + ps2esdi_queue = blk_init_queue(do_ps2esdi_request, &ps2esdi_lock); + if (!ps2esdi_queue) { + unregister_blkdev(PS2ESDI_MAJOR, "ed"); + return -ENOMEM; + } + + /* some minor housekeeping - setup the global gendisk structure */ + error = ps2esdi_geninit(); + if (error) { + printk(KERN_WARNING "PS2ESDI: error initialising" + " device, releasing resources\n"); + unregister_blkdev(PS2ESDI_MAJOR, "ed"); + blk_cleanup_queue(ps2esdi_queue); + return error; + } + return 0; +} /* ps2esdi_init */ + +#ifndef MODULE + +module_init(ps2esdi_init); + +#else + +static int cyl[MAX_HD] = {-1,-1}; +static int head[MAX_HD] = {-1, -1}; +static int sect[MAX_HD] = {-1, -1}; + +module_param(tp720esdi, bool, 0); +module_param_array(cyl, int, NULL, 0); +module_param_array(head, int, NULL, 0); +module_param_array(sect, int, NULL, 0); +MODULE_LICENSE("GPL"); + +int init_module(void) { + int drive; + + for(drive = 0; drive < MAX_HD; drive++) { + struct ps2esdi_i_struct *info = &ps2esdi_info[drive]; + + if (cyl[drive] != -1) { + info->cyl = info->lzone = cyl[drive]; + info->wpcom = 0; + } + if (head[drive] != -1) { + info->head = head[drive]; + info->ctl = (head[drive] > 8 ? 8 : 0); + } + if (sect[drive] != -1) info->sect = sect[drive]; + } + return ps2esdi_init(); +} + +void +cleanup_module(void) { + int i; + if(ps2esdi_slot) { + mca_mark_as_unused(ps2esdi_slot); + mca_set_adapter_procfn(ps2esdi_slot, NULL, NULL); + } + release_region(io_base, 4); + free_dma(dma_arb_level); + free_irq(PS2ESDI_IRQ, &ps2esdi_gendisk); + unregister_blkdev(PS2ESDI_MAJOR, "ed"); + blk_cleanup_queue(ps2esdi_queue); + for (i = 0; i < ps2esdi_drives; i++) { + del_gendisk(ps2esdi_gendisk[i]); + put_disk(ps2esdi_gendisk[i]); + } +} +#endif /* MODULE */ + +/* handles boot time command line parameters */ +void __init tp720_setup(char *str, int *ints) +{ + /* no params, just sets the tp720esdi flag if it exists */ + + printk("%s: TP 720 ESDI flag set\n", DEVICE_NAME); + tp720esdi = 1; +} + +void __init ed_setup(char *str, int *ints) +{ + int hdind = 0; + + /* handles 3 parameters only - corresponding to + 1. Number of cylinders + 2. Number of heads + 3. Sectors/track + */ + + if (ints[0] != 3) + return; + + /* print out the information - seen at boot time */ + printk("%s: ints[0]=%d ints[1]=%d ints[2]=%d ints[3]=%d\n", + DEVICE_NAME, ints[0], ints[1], ints[2], ints[3]); + + /* set the index into device specific information table */ + if (ps2esdi_info[0].head != 0) + hdind = 1; + + /* set up all the device information */ + ps2esdi_info[hdind].head = ints[2]; + ps2esdi_info[hdind].sect = ints[3]; + ps2esdi_info[hdind].cyl = ints[1]; + ps2esdi_info[hdind].wpcom = 0; + ps2esdi_info[hdind].lzone = ints[1]; + ps2esdi_info[hdind].ctl = (ints[2] > 8 ? 8 : 0); +#if 0 /* this may be needed for PS2/Mod.80, but it hurts ThinkPad! */ + ps2esdi_drives = hdind + 1; /* increment index for the next time */ +#endif +} /* ed_setup */ + +static int ps2esdi_getinfo(char *buf, int slot, void *d) +{ + int len = 0; + + len += sprintf(buf + len, "DMA Arbitration Level: %d\n", + dma_arb_level); + len += sprintf(buf + len, "IO Port: %x\n", io_base); + len += sprintf(buf + len, "IRQ: 14\n"); + len += sprintf(buf + len, "Drives: %d\n", ps2esdi_drives); + + return len; +} + +/* ps2 esdi specific initialization - called thru the gendisk chain */ +static int __init ps2esdi_geninit(void) +{ + /* + The first part contains the initialization code + for the ESDI disk subsystem. All we really do + is search for the POS registers of the controller + to do some simple setup operations. First, we + must ensure that the controller is installed, + enabled, and configured as PRIMARY. Then we must + determine the DMA arbitration level being used by + the controller so we can handle data transfer + operations properly. If all of this works, then + we will set the INIT_FLAG to a non-zero value. + */ + + int slot = 0, i, reset_start, reset_end; + u_char status; + unsigned short adapterID; + int error = 0; + + if ((slot = mca_find_adapter(INTG_ESDI_ID, 0)) != MCA_NOTFOUND) { + adapterID = INTG_ESDI_ID; + printk("%s: integrated ESDI adapter found in slot %d\n", + DEVICE_NAME, slot+1); +#ifndef MODULE + mca_set_adapter_name(slot, "PS/2 Integrated ESDI"); +#endif + } else if ((slot = mca_find_adapter(NRML_ESDI_ID, 0)) != -1) { + adapterID = NRML_ESDI_ID; + printk("%s: normal ESDI adapter found in slot %d\n", + DEVICE_NAME, slot+1); + mca_set_adapter_name(slot, "PS/2 ESDI"); + } else { + return -ENODEV; + } + + ps2esdi_slot = slot; + mca_mark_as_used(slot); + mca_set_adapter_procfn(slot, (MCA_ProcFn) ps2esdi_getinfo, NULL); + + /* Found the slot - read the POS register 2 to get the necessary + configuration and status information. POS register 2 has the + following information : + Bit Function + 7 reserved = 0 + 6 arbitration method + 0 - fairness enabled + 1 - fairness disabled, linear priority assignment + 5-2 arbitration level + 1 alternate address + 1 alternate address + 0 - use addresses 0x3510 - 0x3517 + 0 adapter enable + */ + + status = mca_read_stored_pos(slot, 2); + /* is it enabled ? */ + if (!(status & STATUS_ENABLED)) { + printk("%s: ESDI adapter disabled\n", DEVICE_NAME); + error = -ENODEV; + goto err_out1; + } + /* try to grab IRQ, and try to grab a slow IRQ if it fails, so we can + share with the SCSI driver */ + if (request_irq(PS2ESDI_IRQ, ps2esdi_interrupt_handler, + SA_INTERRUPT | SA_SHIRQ, "PS/2 ESDI", &ps2esdi_gendisk) + && request_irq(PS2ESDI_IRQ, ps2esdi_interrupt_handler, + SA_SHIRQ, "PS/2 ESDI", &ps2esdi_gendisk) + ) { + printk("%s: Unable to get IRQ %d\n", DEVICE_NAME, PS2ESDI_IRQ); + error = -EBUSY; + goto err_out1; + } + if (status & STATUS_ALTERNATE) + io_base = ALT_IO_BASE; + else + io_base = PRIMARY_IO_BASE; + + if (!request_region(io_base, 4, "ed")) { + printk(KERN_WARNING"Unable to request region 0x%x\n", io_base); + error = -EBUSY; + goto err_out2; + } + /* get the dma arbitration level */ + dma_arb_level = (status >> 2) & 0xf; + + /* BA */ + printk("%s: DMA arbitration level : %d\n", + DEVICE_NAME, dma_arb_level); + + LITE_ON; + current_int_handler = ps2esdi_initial_reset_int_handler; + reset_ctrl(); + reset_status = 0; + reset_start = jiffies; + while (!reset_status) { + init_timer(&esdi_timer); + esdi_timer.expires = jiffies + HZ; + esdi_timer.data = 0; + add_timer(&esdi_timer); + sleep_on(&ps2esdi_int); + } + reset_end = jiffies; + LITE_OFF; + printk("%s: reset interrupt after %d jiffies, %u.%02u secs\n", + DEVICE_NAME, reset_end - reset_start, (reset_end - reset_start) / HZ, + (reset_end - reset_start) % HZ); + + + /* Integrated ESDI Disk and Controller has only one drive! */ + if (adapterID == INTG_ESDI_ID) {/* if not "normal" PS2 ESDI adapter */ + ps2esdi_drives = 1; /* then we have only one physical disk! */ intg_esdi = 1; + } + + + + /* finally this part sets up some global data structures etc. */ + + ps2esdi_get_device_cfg(); + + /* some annoyance in the above routine returns TWO drives? + Is something else happining in the background? + Regaurdless we fix the # of drives again. AJK */ + /* Integrated ESDI Disk and Controller has only one drive! */ + if (adapterID == INTG_ESDI_ID) /* if not "normal" PS2 ESDI adapter */ + ps2esdi_drives = 1; /* Not three or two, ONE DAMNIT! */ + + current_int_handler = ps2esdi_normal_interrupt_handler; + + if (request_dma(dma_arb_level, "ed") !=0) { + printk(KERN_WARNING "PS2ESDI: Can't request dma-channel %d\n" + ,(int) dma_arb_level); + error = -EBUSY; + goto err_out3; + } + blk_queue_max_sectors(ps2esdi_queue, 128); + + error = -ENOMEM; + for (i = 0; i < ps2esdi_drives; i++) { + struct gendisk *disk = alloc_disk(64); + if (!disk) + goto err_out4; + disk->major = PS2ESDI_MAJOR; + disk->first_minor = i<<6; + sprintf(disk->disk_name, "ed%c", 'a'+i); + sprintf(disk->devfs_name, "ed/target%d", i); + disk->fops = &ps2esdi_fops; + ps2esdi_gendisk[i] = disk; + } + + for (i = 0; i < ps2esdi_drives; i++) { + struct gendisk *disk = ps2esdi_gendisk[i]; + set_capacity(disk, ps2esdi_info[i].head * ps2esdi_info[i].sect * + ps2esdi_info[i].cyl); + disk->queue = ps2esdi_queue; + disk->private_data = &ps2esdi_info[i]; + add_disk(disk); + } + return 0; +err_out4: + while (i--) + put_disk(ps2esdi_gendisk[i]); +err_out3: + release_region(io_base, 4); +err_out2: + free_irq(PS2ESDI_IRQ, &ps2esdi_gendisk); +err_out1: + if(ps2esdi_slot) { + mca_mark_as_unused(ps2esdi_slot); + mca_set_adapter_procfn(ps2esdi_slot, NULL, NULL); + } + return error; +} + +static void __init ps2esdi_get_device_cfg(void) +{ + u_short cmd_blk[TYPE_0_CMD_BLK_LENGTH]; + + /*BA */ printk("%s: Drive 0\n", DEVICE_NAME); + current_int_handler = ps2esdi_geometry_int_handler; + cmd_blk[0] = CMD_GET_DEV_CONFIG | 0x600; + cmd_blk[1] = 0; + no_int_yet = TRUE; + ps2esdi_out_cmd_blk(cmd_blk); + if (no_int_yet) + sleep_on(&ps2esdi_int); + + if (ps2esdi_drives > 1) { + printk("%s: Drive 1\n", DEVICE_NAME); /*BA */ + cmd_blk[0] = CMD_GET_DEV_CONFIG | (1 << 5) | 0x600; + cmd_blk[1] = 0; + no_int_yet = TRUE; + ps2esdi_out_cmd_blk(cmd_blk); + if (no_int_yet) + sleep_on(&ps2esdi_int); + } /* if second physical drive is present */ + return; +} + +/* strategy routine that handles most of the IO requests */ +static void do_ps2esdi_request(request_queue_t * q) +{ + struct request *req; + /* since, this routine is called with interrupts cleared - they + must be before it finishes */ + + req = elv_next_request(q); + if (!req) + return; + +#if 0 + printk("%s:got request. device : %s command : %d sector : %ld count : %ld, buffer: %p\n", + DEVICE_NAME, + req->rq_disk->disk_name, + req->cmd, req->sector, + req->current_nr_sectors, req->buffer); +#endif + + /* check for above 16Mb dmas */ + if (isa_virt_to_bus(req->buffer + req->current_nr_sectors * 512) > 16 * MB) { + printk("%s: DMA above 16MB not supported\n", DEVICE_NAME); + end_request(req, FAIL); + return; + } + + if (req->sector+req->current_nr_sectors > get_capacity(req->rq_disk)) { + printk("Grrr. error. ps2esdi_drives: %d, %llu %llu\n", + ps2esdi_drives, req->sector, + (unsigned long long)get_capacity(req->rq_disk)); + end_request(req, FAIL); + return; + } + + switch (rq_data_dir(req)) { + case READ: + ps2esdi_readwrite(READ, req); + break; + case WRITE: + ps2esdi_readwrite(WRITE, req); + break; + default: + printk("%s: Unknown command\n", req->rq_disk->disk_name); + end_request(req, FAIL); + break; + } /* handle different commands */ +} /* main strategy routine */ + +/* resets the ESDI adapter */ +static void reset_ctrl(void) +{ + + u_long expire; + u_short status; + + /* enable interrupts on the controller */ + status = inb(ESDI_INTRPT); + outb((status & 0xe0) | ATT_EOI, ESDI_ATTN); /* to be sure we don't have + any interrupt pending... */ + outb_p(CTRL_ENABLE_INTR, ESDI_CONTROL); + + /* read the ESDI status port - if the controller is not busy, + simply do a soft reset (fast) - otherwise we'll have to do a + hard (slow) reset. */ + if (!(inb_p(ESDI_STATUS) & STATUS_BUSY)) { + /*BA */ printk("%s: soft reset...\n", DEVICE_NAME); + outb_p(CTRL_SOFT_RESET, ESDI_ATTN); + } + /* soft reset */ + else { + /*BA */ + printk("%s: hard reset...\n", DEVICE_NAME); + outb_p(CTRL_HARD_RESET, ESDI_CONTROL); + expire = jiffies + 2*HZ; + while (time_before(jiffies, expire)); + outb_p(1, ESDI_CONTROL); + } /* hard reset */ + + +} /* reset the controller */ + +/* called by the strategy routine to handle read and write requests */ +static void ps2esdi_readwrite(int cmd, struct request *req) +{ + struct ps2esdi_i_struct *p = req->rq_disk->private_data; + unsigned block = req->sector; + unsigned count = req->current_nr_sectors; + int drive = p - ps2esdi_info; + u_short track, head, cylinder, sector; + u_short cmd_blk[TYPE_1_CMD_BLK_LENGTH]; + + /* do some relevant arithmatic */ + track = block / p->sect; + head = track % p->head; + cylinder = track / p->head; + sector = block % p->sect; + +#if 0 + printk("%s: cyl=%d head=%d sect=%d\n", DEVICE_NAME, cylinder, head, sector); +#endif + /* call the routine that actually fills out a command block */ + ps2esdi_fill_cmd_block + (cmd_blk, + (cmd == READ) ? CMD_READ : CMD_WRITE, + cylinder, head, sector, count, drive); + + /* send the command block to the controller */ + current_req = req; + spin_unlock_irq(&ps2esdi_lock); + if (ps2esdi_out_cmd_blk(cmd_blk)) { + spin_lock_irq(&ps2esdi_lock); + printk("%s: Controller failed\n", DEVICE_NAME); + if ((++req->errors) >= MAX_RETRIES) + end_request(req, FAIL); + } + /* check for failure to put out the command block */ + else { + spin_lock_irq(&ps2esdi_lock); +#if 0 + printk("%s: waiting for xfer\n", DEVICE_NAME); +#endif + /* turn disk lights on */ + LITE_ON; + } + +} /* ps2esdi_readwrite */ + +/* fill out the command block */ +static void ps2esdi_fill_cmd_block(u_short * cmd_blk, u_short cmd, + u_short cyl, u_short head, u_short sector, u_short length, u_char drive) +{ + + cmd_blk[0] = (drive << 5) | cmd; + cmd_blk[1] = length; + cmd_blk[2] = ((cyl & 0x1f) << 11) | (head << 5) | sector; + cmd_blk[3] = (cyl & 0x3E0) >> 5; + +} /* fill out the command block */ + +/* write a command block to the controller */ +static int ps2esdi_out_cmd_blk(u_short * cmd_blk) +{ + + int i; + unsigned long jif; + u_char status; + + /* enable interrupts */ + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + + /* do not write to the controller, if it is busy */ + for (jif = jiffies + ESDI_STAT_TIMEOUT; + time_after(jif, jiffies) && + (inb(ESDI_STATUS) & STATUS_BUSY); ) + ; + +#if 0 + printk("%s: i(1)=%ld\n", DEVICE_NAME, jif); +#endif + + /* if device is still busy - then just time out */ + if (inb(ESDI_STATUS) & STATUS_BUSY) { + printk("%s: ps2esdi_out_cmd timed out (1)\n", DEVICE_NAME); + return ERROR; + } /* timeout ??? */ + /* Set up the attention register in the controller */ + outb(((*cmd_blk) & 0xE0) | 1, ESDI_ATTN); + +#if 0 + printk("%s: sending %d words to controller\n", DEVICE_NAME, (((*cmd_blk) >> 14) + 1) << 1); +#endif + + /* one by one send each word out */ + for (i = (((*cmd_blk) >> 14) + 1) << 1; i; i--) { + status = inb(ESDI_STATUS); + for (jif = jiffies + ESDI_STAT_TIMEOUT; + time_after(jif, jiffies) && (status & STATUS_BUSY) && + (status & STATUS_CMD_INF); status = inb(ESDI_STATUS)); + if ((status & (STATUS_BUSY | STATUS_CMD_INF)) == STATUS_BUSY) { +#if 0 + printk("%s: sending %04X\n", DEVICE_NAME, *cmd_blk); +#endif + outw(*cmd_blk++, ESDI_CMD_INT); + } else { + printk("%s: ps2esdi_out_cmd timed out while sending command (status=%02X)\n", + DEVICE_NAME, status); + return ERROR; + } + } /* send all words out */ + return OK; +} /* send out the commands */ + + +/* prepare for dma - do all the necessary setup */ +static void ps2esdi_prep_dma(char *buffer, u_short length, u_char dma_xmode) +{ + unsigned long flags = claim_dma_lock(); + + mca_disable_dma(dma_arb_level); + + mca_set_dma_addr(dma_arb_level, isa_virt_to_bus(buffer)); + + mca_set_dma_count(dma_arb_level, length * 512 / 2); + + mca_set_dma_mode(dma_arb_level, dma_xmode); + + mca_enable_dma(dma_arb_level); + + release_dma_lock(flags); + +} /* prepare for dma */ + + + +static irqreturn_t ps2esdi_interrupt_handler(int irq, void *dev_id, + struct pt_regs *regs) +{ + u_int int_ret_code; + + if (inb(ESDI_STATUS) & STATUS_INTR) { + int_ret_code = inb(ESDI_INTRPT); + if (current_int_handler) { + /* Disable adapter interrupts till processing is finished */ + outb(CTRL_DISABLE_INTR, ESDI_CONTROL); + current_int_handler(int_ret_code); + } else + printk("%s: help ! No interrupt handler.\n", DEVICE_NAME); + } else { + return IRQ_NONE; + } + return IRQ_HANDLED; +} + +static void ps2esdi_initial_reset_int_handler(u_int int_ret_code) +{ + + switch (int_ret_code & 0xf) { + case INT_RESET: + /*BA */ + printk("%s: initial reset completed.\n", DEVICE_NAME); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + wake_up(&ps2esdi_int); + break; + case INT_ATTN_ERROR: + printk("%s: Attention error. interrupt status : %02X\n", DEVICE_NAME, + int_ret_code); + printk("%s: status: %02x\n", DEVICE_NAME, inb(ESDI_STATUS)); + break; + default: + printk("%s: initial reset handler received interrupt: %02X\n", + DEVICE_NAME, int_ret_code); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + break; + } + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); +} + + +static void ps2esdi_geometry_int_handler(u_int int_ret_code) +{ + u_int status, drive_num; + unsigned long rba; + int i; + + drive_num = int_ret_code >> 5; + switch (int_ret_code & 0xf) { + case INT_CMD_COMPLETE: + for (i = ESDI_TIMEOUT; i && !(inb(ESDI_STATUS) & STATUS_STAT_AVAIL); i--); + if (!(inb(ESDI_STATUS) & STATUS_STAT_AVAIL)) { + printk("%s: timeout reading status word\n", DEVICE_NAME); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + break; + } + status = inw(ESDI_STT_INT); + if ((status & 0x1F) == CMD_GET_DEV_CONFIG) { +#define REPLY_WORDS 5 /* we already read word 0 */ + u_short reply[REPLY_WORDS]; + + if (ps2esdi_read_status_words((status >> 8) - 1, REPLY_WORDS, reply)) { + /*BA */ + printk("%s: Device Configuration Status for drive %u\n", + DEVICE_NAME, drive_num); + + printk("%s: Spares/cyls: %u", DEVICE_NAME, reply[0] >> 8); + + printk + ("Config bits: %s%s%s%s%s\n", + (reply[0] & CONFIG_IS) ? "Invalid Secondary, " : "", + ((reply[0] & CONFIG_ZD) && !(reply[0] & CONFIG_IS)) + ? "Zero Defect, " : "Defects Present, ", + (reply[0] & CONFIG_SF) ? "Skewed Format, " : "", + (reply[0] & CONFIG_FR) ? "Removable, " : "Non-Removable, ", + (reply[0] & CONFIG_RT) ? "No Retries" : "Retries"); + + rba = reply[1] | ((unsigned long) reply[2] << 16); + printk("%s: Number of RBA's: %lu\n", DEVICE_NAME, rba); + + printk("%s: Physical number of cylinders: %u, Sectors/Track: %u, Heads: %u\n", + DEVICE_NAME, reply[3], reply[4] >> 8, reply[4] & 0xff); + + if (!ps2esdi_info[drive_num].head) { + ps2esdi_info[drive_num].head = 64; + ps2esdi_info[drive_num].sect = 32; + ps2esdi_info[drive_num].cyl = rba / (64 * 32); + ps2esdi_info[drive_num].wpcom = 0; + ps2esdi_info[drive_num].lzone = ps2esdi_info[drive_num].cyl; + ps2esdi_info[drive_num].ctl = 8; + if (tp720esdi) { /* store the retrieved parameters */ + ps2esdi_info[0].head = reply[4] & 0Xff; + ps2esdi_info[0].sect = reply[4] >> 8; + ps2esdi_info[0].cyl = reply[3]; + ps2esdi_info[0].wpcom = 0; + ps2esdi_info[0].lzone = reply[3]; + } else { + if (!intg_esdi) + ps2esdi_drives++; + } + } +#ifdef OBSOLETE + if (!ps2esdi_info[drive_num].head) { + ps2esdi_info[drive_num].head = reply[4] & 0Xff; + ps2esdi_info[drive_num].sect = reply[4] >> 8; + ps2esdi_info[drive_num].cyl = reply[3]; + ps2esdi_info[drive_num].wpcom = 0; + ps2esdi_info[drive_num].lzone = reply[3]; + if (tp720esdi) { /* store the retrieved parameters */ + ps2esdi_info[0].head = reply[4] & 0Xff; + ps2esdi_info[0].sect = reply[4] >> 8; + ps2esdi_info[0].cyl = reply[3]; + ps2esdi_info[0].wpcom = 0; + ps2esdi_info[0].lzone = reply[3]; + } else { + ps2esdi_drives++; + } + } +#endif + + } else + printk("%s: failed while getting device config\n", DEVICE_NAME); +#undef REPLY_WORDS + } else + printk("%s: command %02X unknown by geometry handler\n", + DEVICE_NAME, status & 0x1f); + + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + break; + + case INT_ATTN_ERROR: + printk("%s: Attention error. interrupt status : %02X\n", DEVICE_NAME, + int_ret_code); + printk("%s: Device not available\n", DEVICE_NAME); + break; + case INT_CMD_ECC: + case INT_CMD_RETRY: + case INT_CMD_ECC_RETRY: + case INT_CMD_WARNING: + case INT_CMD_ABORT: + case INT_CMD_FAILED: + case INT_DMA_ERR: + case INT_CMD_BLK_ERR: + /*BA */ printk("%s: Whaa. Error occurred...\n", DEVICE_NAME); + dump_cmd_complete_status(int_ret_code); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + break; + default: + printk("%s: Unknown interrupt reason: %02X\n", + DEVICE_NAME, int_ret_code & 0xf); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + break; + } + + wake_up(&ps2esdi_int); + no_int_yet = FALSE; + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + +} + +static void ps2esdi_normal_interrupt_handler(u_int int_ret_code) +{ + unsigned long flags; + u_int status; + u_int ending; + int i; + + switch (int_ret_code & 0x0f) { + case INT_TRANSFER_REQ: + ps2esdi_prep_dma(current_req->buffer, + current_req->current_nr_sectors, + (rq_data_dir(current_req) == READ) + ? MCA_DMA_MODE_16 | MCA_DMA_MODE_WRITE | MCA_DMA_MODE_XFER + : MCA_DMA_MODE_16 | MCA_DMA_MODE_READ); + outb(CTRL_ENABLE_DMA | CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = -1; + break; + + case INT_ATTN_ERROR: + printk("%s: Attention error. interrupt status : %02X\n", DEVICE_NAME, + int_ret_code); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = FAIL; + break; + + case INT_CMD_COMPLETE: + for (i = ESDI_TIMEOUT; i && !(inb(ESDI_STATUS) & STATUS_STAT_AVAIL); i--); + if (!(inb(ESDI_STATUS) & STATUS_STAT_AVAIL)) { + printk("%s: timeout reading status word\n", DEVICE_NAME); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + if ((++current_req->errors) >= MAX_RETRIES) + ending = FAIL; + else + ending = -1; + break; + } + status = inw(ESDI_STT_INT); + switch (status & 0x1F) { + case (CMD_READ & 0xff): + case (CMD_WRITE & 0xff): + LITE_OFF; + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = SUCCES; + break; + default: + printk("%s: interrupt for unknown command %02X\n", + DEVICE_NAME, status & 0x1f); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = -1; + break; + } + break; + case INT_CMD_ECC: + case INT_CMD_RETRY: + case INT_CMD_ECC_RETRY: + LITE_OFF; + dump_cmd_complete_status(int_ret_code); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = SUCCES; + break; + case INT_CMD_WARNING: + case INT_CMD_ABORT: + case INT_CMD_FAILED: + case INT_DMA_ERR: + LITE_OFF; + dump_cmd_complete_status(int_ret_code); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + if ((++current_req->errors) >= MAX_RETRIES) + ending = FAIL; + else + ending = -1; + break; + + case INT_CMD_BLK_ERR: + dump_cmd_complete_status(int_ret_code); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = FAIL; + break; + + case INT_CMD_FORMAT: + printk("%s: huh ? Who issued this format command ?\n" + ,DEVICE_NAME); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = -1; + break; + + case INT_RESET: + /* BA printk("%s: reset completed.\n", DEVICE_NAME) */ ; + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = -1; + break; + + default: + printk("%s: Unknown interrupt reason: %02X\n", + DEVICE_NAME, int_ret_code & 0xf); + outb((int_ret_code & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + ending = -1; + break; + } + if(ending != -1) { + spin_lock_irqsave(&ps2esdi_lock, flags); + end_request(current_req, ending); + current_req = NULL; + do_ps2esdi_request(ps2esdi_queue); + spin_unlock_irqrestore(&ps2esdi_lock, flags); + } +} /* handle interrupts */ + + + +static int ps2esdi_read_status_words(int num_words, + int max_words, + u_short * buffer) +{ + int i; + + for (; max_words && num_words; max_words--, num_words--, buffer++) { + for (i = ESDI_TIMEOUT; i && !(inb(ESDI_STATUS) & STATUS_STAT_AVAIL); i--); + if (!(inb(ESDI_STATUS) & STATUS_STAT_AVAIL)) { + printk("%s: timeout reading status word\n", DEVICE_NAME); + return FAIL; + } + *buffer = inw(ESDI_STT_INT); + } + return SUCCES; +} + + + + +static void dump_cmd_complete_status(u_int int_ret_code) +{ +#define WAIT_FOR_STATUS \ + for(i=ESDI_TIMEOUT;i && !(inb(ESDI_STATUS) & STATUS_STAT_AVAIL);i--); \ + if(!(inb(ESDI_STATUS) & STATUS_STAT_AVAIL)) { \ + printk("%s: timeout reading status word\n",DEVICE_NAME); \ + return; \ + } + + int i, word_count; + u_short stat_word; + u_long rba; + + printk("%s: Device: %u, interrupt ID: %02X\n", + DEVICE_NAME, int_ret_code >> 5, + int_ret_code & 0xf); + + WAIT_FOR_STATUS; + stat_word = inw(ESDI_STT_INT); + word_count = (stat_word >> 8) - 1; + printk("%s: %u status words, command: %02X\n", DEVICE_NAME, word_count, + stat_word & 0xff); + + if (word_count--) { + WAIT_FOR_STATUS; + stat_word = inw(ESDI_STT_INT); + printk("%s: command status code: %02X, command error code: %02X\n", + DEVICE_NAME, stat_word >> 8, stat_word & 0xff); + } + if (word_count--) { + WAIT_FOR_STATUS; + stat_word = inw(ESDI_STT_INT); + printk("%s: device error code: %s%s%s%s%s,%02X\n", DEVICE_NAME, + (stat_word & 0x1000) ? "Ready, " : "Not Ready, ", + (stat_word & 0x0800) ? "Selected, " : "Not Selected, ", + (stat_word & 0x0400) ? "Write Fault, " : "", + (stat_word & 0x0200) ? "Track 0, " : "", + (stat_word & 0x0100) ? "Seek or command complete, " : "", + stat_word >> 8); + } + if (word_count--) { + WAIT_FOR_STATUS; + stat_word = inw(ESDI_STT_INT); + printk("%s: Blocks to do: %u", DEVICE_NAME, stat_word); + } + if (word_count -= 2) { + WAIT_FOR_STATUS; + rba = inw(ESDI_STT_INT); + WAIT_FOR_STATUS; + rba |= inw(ESDI_STT_INT) << 16; + printk(", Last Cyl: %u Head: %u Sector: %u\n", + (u_short) ((rba & 0x1ff80000) >> 11), + (u_short) ((rba & 0x7E0) >> 5), (u_short) (rba & 0x1f)); + } else + printk("\n"); + + if (word_count--) { + WAIT_FOR_STATUS; + stat_word = inw(ESDI_STT_INT); + printk("%s: Blocks required ECC: %u", DEVICE_NAME, stat_word); + } + printk("\n"); + +#undef WAIT_FOR_STATUS + +} + +static int ps2esdi_ioctl(struct inode *inode, + struct file *file, u_int cmd, u_long arg) +{ + struct ps2esdi_i_struct *p = inode->i_bdev->bd_disk->private_data; + struct ps2esdi_geometry geom; + + if (cmd != HDIO_GETGEO) + return -EINVAL; + memset(&geom, 0, sizeof(geom)); + geom.heads = p->head; + geom.sectors = p->sect; + geom.cylinders = p->cyl; + geom.start = get_start_sect(inode->i_bdev); + if (copy_to_user((void __user *)arg, &geom, sizeof(geom))) + return -EFAULT; + return 0; +} + +static void ps2esdi_reset_timer(unsigned long unused) +{ + + int status; + + status = inb(ESDI_INTRPT); + if ((status & 0xf) == INT_RESET) { + outb((status & 0xe0) | ATT_EOI, ESDI_ATTN); + outb(CTRL_ENABLE_INTR, ESDI_CONTROL); + reset_status = 1; + } + wake_up(&ps2esdi_int); +} diff --git a/drivers/block/rd.c b/drivers/block/rd.c new file mode 100644 index 00000000000..145c1fbffe0 --- /dev/null +++ b/drivers/block/rd.c @@ -0,0 +1,515 @@ +/* + * ramdisk.c - Multiple RAM disk driver - gzip-loading version - v. 0.8 beta. + * + * (C) Chad Page, Theodore Ts'o, et. al, 1995. + * + * This RAM disk is designed to have filesystems created on it and mounted + * just like a regular floppy disk. + * + * It also does something suggested by Linus: use the buffer cache as the + * RAM disk data. This makes it possible to dynamically allocate the RAM disk + * buffer - with some consequences I have to deal with as I write this. + * + * This code is based on the original ramdisk.c, written mostly by + * Theodore Ts'o (TYT) in 1991. The code was largely rewritten by + * Chad Page to use the buffer cache to store the RAM disk data in + * 1995; Theodore then took over the driver again, and cleaned it up + * for inclusion in the mainline kernel. + * + * The original CRAMDISK code was written by Richard Lyons, and + * adapted by Chad Page to use the new RAM disk interface. Theodore + * Ts'o rewrote it so that both the compressed RAM disk loader and the + * kernel decompressor uses the same inflate.c codebase. The RAM disk + * loader now also loads into a dynamic (buffer cache based) RAM disk, + * not the old static RAM disk. Support for the old static RAM disk has + * been completely removed. + * + * Loadable module support added by Tom Dyas. + * + * Further cleanups by Chad Page (page0588@sundance.sjsu.edu): + * Cosmetic changes in #ifdef MODULE, code movement, etc. + * When the RAM disk module is removed, free the protected buffers + * Default RAM disk size changed to 2.88 MB + * + * Added initrd: Werner Almesberger & Hans Lermen, Feb '96 + * + * 4/25/96 : Made RAM disk size a parameter (default is now 4 MB) + * - Chad Page + * + * Add support for fs images split across >1 disk, Paul Gortmaker, Mar '98 + * + * Make block size and block size shift for RAM disks a global macro + * and set blk_size for -ENOSPC, Werner Fink <werner@suse.de>, Apr '99 + */ + +#include <linux/config.h> +#include <linux/string.h> +#include <linux/slab.h> +#include <asm/atomic.h> +#include <linux/bio.h> +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <linux/init.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/pagemap.h> +#include <linux/blkdev.h> +#include <linux/genhd.h> +#include <linux/buffer_head.h> /* for invalidate_bdev() */ +#include <linux/backing-dev.h> +#include <linux/blkpg.h> +#include <linux/writeback.h> + +#include <asm/uaccess.h> + +/* Various static variables go here. Most are used only in the RAM disk code. + */ + +static struct gendisk *rd_disks[CONFIG_BLK_DEV_RAM_COUNT]; +static struct block_device *rd_bdev[CONFIG_BLK_DEV_RAM_COUNT];/* Protected device data */ +static struct request_queue *rd_queue[CONFIG_BLK_DEV_RAM_COUNT]; + +/* + * Parameters for the boot-loading of the RAM disk. These are set by + * init/main.c (from arguments to the kernel command line) or from the + * architecture-specific setup routine (from the stored boot sector + * information). + */ +int rd_size = CONFIG_BLK_DEV_RAM_SIZE; /* Size of the RAM disks */ +/* + * It would be very desirable to have a soft-blocksize (that in the case + * of the ramdisk driver is also the hardblocksize ;) of PAGE_SIZE because + * doing that we'll achieve a far better MM footprint. Using a rd_blocksize of + * BLOCK_SIZE in the worst case we'll make PAGE_SIZE/BLOCK_SIZE buffer-pages + * unfreeable. With a rd_blocksize of PAGE_SIZE instead we are sure that only + * 1 page will be protected. Depending on the size of the ramdisk you + * may want to change the ramdisk blocksize to achieve a better or worse MM + * behaviour. The default is still BLOCK_SIZE (needed by rd_load_image that + * supposes the filesystem in the image uses a BLOCK_SIZE blocksize). + */ +static int rd_blocksize = BLOCK_SIZE; /* blocksize of the RAM disks */ + +/* + * Copyright (C) 2000 Linus Torvalds. + * 2000 Transmeta Corp. + * aops copied from ramfs. + */ + +/* + * If a ramdisk page has buffers, some may be uptodate and some may be not. + * To bring the page uptodate we zero out the non-uptodate buffers. The + * page must be locked. + */ +static void make_page_uptodate(struct page *page) +{ + if (page_has_buffers(page)) { + struct buffer_head *bh = page_buffers(page); + struct buffer_head *head = bh; + + do { + if (!buffer_uptodate(bh)) { + memset(bh->b_data, 0, bh->b_size); + /* + * akpm: I'm totally undecided about this. The + * buffer has just been magically brought "up to + * date", but nobody should want to be reading + * it anyway, because it hasn't been used for + * anything yet. It is still in a "not read + * from disk yet" state. + * + * But non-uptodate buffers against an uptodate + * page are against the rules. So do it anyway. + */ + set_buffer_uptodate(bh); + } + } while ((bh = bh->b_this_page) != head); + } else { + memset(page_address(page), 0, PAGE_CACHE_SIZE); + } + flush_dcache_page(page); + SetPageUptodate(page); +} + +static int ramdisk_readpage(struct file *file, struct page *page) +{ + if (!PageUptodate(page)) + make_page_uptodate(page); + unlock_page(page); + return 0; +} + +static int ramdisk_prepare_write(struct file *file, struct page *page, + unsigned offset, unsigned to) +{ + if (!PageUptodate(page)) + make_page_uptodate(page); + return 0; +} + +static int ramdisk_commit_write(struct file *file, struct page *page, + unsigned offset, unsigned to) +{ + set_page_dirty(page); + return 0; +} + +/* + * ->writepage to the the blockdev's mapping has to redirty the page so that the + * VM doesn't go and steal it. We return WRITEPAGE_ACTIVATE so that the VM + * won't try to (pointlessly) write the page again for a while. + * + * Really, these pages should not be on the LRU at all. + */ +static int ramdisk_writepage(struct page *page, struct writeback_control *wbc) +{ + if (!PageUptodate(page)) + make_page_uptodate(page); + SetPageDirty(page); + if (wbc->for_reclaim) + return WRITEPAGE_ACTIVATE; + unlock_page(page); + return 0; +} + +/* + * This is a little speedup thing: short-circuit attempts to write back the + * ramdisk blockdev inode to its non-existent backing store. + */ +static int ramdisk_writepages(struct address_space *mapping, + struct writeback_control *wbc) +{ + return 0; +} + +/* + * ramdisk blockdev pages have their own ->set_page_dirty() because we don't + * want them to contribute to dirty memory accounting. + */ +static int ramdisk_set_page_dirty(struct page *page) +{ + SetPageDirty(page); + return 0; +} + +static struct address_space_operations ramdisk_aops = { + .readpage = ramdisk_readpage, + .prepare_write = ramdisk_prepare_write, + .commit_write = ramdisk_commit_write, + .writepage = ramdisk_writepage, + .set_page_dirty = ramdisk_set_page_dirty, + .writepages = ramdisk_writepages, +}; + +static int rd_blkdev_pagecache_IO(int rw, struct bio_vec *vec, sector_t sector, + struct address_space *mapping) +{ + pgoff_t index = sector >> (PAGE_CACHE_SHIFT - 9); + unsigned int vec_offset = vec->bv_offset; + int offset = (sector << 9) & ~PAGE_CACHE_MASK; + int size = vec->bv_len; + int err = 0; + + do { + int count; + struct page *page; + char *src; + char *dst; + + count = PAGE_CACHE_SIZE - offset; + if (count > size) + count = size; + size -= count; + + page = grab_cache_page(mapping, index); + if (!page) { + err = -ENOMEM; + goto out; + } + + if (!PageUptodate(page)) + make_page_uptodate(page); + + index++; + + if (rw == READ) { + src = kmap_atomic(page, KM_USER0) + offset; + dst = kmap_atomic(vec->bv_page, KM_USER1) + vec_offset; + } else { + src = kmap_atomic(vec->bv_page, KM_USER0) + vec_offset; + dst = kmap_atomic(page, KM_USER1) + offset; + } + offset = 0; + vec_offset += count; + + memcpy(dst, src, count); + + kunmap_atomic(src, KM_USER0); + kunmap_atomic(dst, KM_USER1); + + if (rw == READ) + flush_dcache_page(vec->bv_page); + else + set_page_dirty(page); + unlock_page(page); + put_page(page); + } while (size); + + out: + return err; +} + +/* + * Basically, my strategy here is to set up a buffer-head which can't be + * deleted, and make that my Ramdisk. If the request is outside of the + * allocated size, we must get rid of it... + * + * 19-JAN-1998 Richard Gooch <rgooch@atnf.csiro.au> Added devfs support + * + */ +static int rd_make_request(request_queue_t *q, struct bio *bio) +{ + struct block_device *bdev = bio->bi_bdev; + struct address_space * mapping = bdev->bd_inode->i_mapping; + sector_t sector = bio->bi_sector; + unsigned long len = bio->bi_size >> 9; + int rw = bio_data_dir(bio); + struct bio_vec *bvec; + int ret = 0, i; + + if (sector + len > get_capacity(bdev->bd_disk)) + goto fail; + + if (rw==READA) + rw=READ; + + bio_for_each_segment(bvec, bio, i) { + ret |= rd_blkdev_pagecache_IO(rw, bvec, sector, mapping); + sector += bvec->bv_len >> 9; + } + if (ret) + goto fail; + + bio_endio(bio, bio->bi_size, 0); + return 0; +fail: + bio_io_error(bio, bio->bi_size); + return 0; +} + +static int rd_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + int error; + struct block_device *bdev = inode->i_bdev; + + if (cmd != BLKFLSBUF) + return -ENOTTY; + + /* + * special: we want to release the ramdisk memory, it's not like with + * the other blockdevices where this ioctl only flushes away the buffer + * cache + */ + error = -EBUSY; + down(&bdev->bd_sem); + if (bdev->bd_openers <= 2) { + truncate_inode_pages(bdev->bd_inode->i_mapping, 0); + error = 0; + } + up(&bdev->bd_sem); + return error; +} + +/* + * This is the backing_dev_info for the blockdev inode itself. It doesn't need + * writeback and it does not contribute to dirty memory accounting. + */ +static struct backing_dev_info rd_backing_dev_info = { + .ra_pages = 0, /* No readahead */ + .capabilities = BDI_CAP_NO_ACCT_DIRTY | BDI_CAP_NO_WRITEBACK | BDI_CAP_MAP_COPY, + .unplug_io_fn = default_unplug_io_fn, +}; + +/* + * This is the backing_dev_info for the files which live atop the ramdisk + * "device". These files do need writeback and they do contribute to dirty + * memory accounting. + */ +static struct backing_dev_info rd_file_backing_dev_info = { + .ra_pages = 0, /* No readahead */ + .capabilities = BDI_CAP_MAP_COPY, /* Does contribute to dirty memory */ + .unplug_io_fn = default_unplug_io_fn, +}; + +static int rd_open(struct inode *inode, struct file *filp) +{ + unsigned unit = iminor(inode); + + if (rd_bdev[unit] == NULL) { + struct block_device *bdev = inode->i_bdev; + struct address_space *mapping; + unsigned bsize; + int gfp_mask; + + inode = igrab(bdev->bd_inode); + rd_bdev[unit] = bdev; + bdev->bd_openers++; + bsize = bdev_hardsect_size(bdev); + bdev->bd_block_size = bsize; + inode->i_blkbits = blksize_bits(bsize); + inode->i_size = get_capacity(bdev->bd_disk)<<9; + + mapping = inode->i_mapping; + mapping->a_ops = &ramdisk_aops; + mapping->backing_dev_info = &rd_backing_dev_info; + bdev->bd_inode_backing_dev_info = &rd_file_backing_dev_info; + + /* + * Deep badness. rd_blkdev_pagecache_IO() needs to allocate + * pagecache pages within a request_fn. We cannot recur back + * into the filesytem which is mounted atop the ramdisk, because + * that would deadlock on fs locks. And we really don't want + * to reenter rd_blkdev_pagecache_IO when we're already within + * that function. + * + * So we turn off __GFP_FS and __GFP_IO. + * + * And to give this thing a hope of working, turn on __GFP_HIGH. + * Hopefully, there's enough regular memory allocation going on + * for the page allocator emergency pools to keep the ramdisk + * driver happy. + */ + gfp_mask = mapping_gfp_mask(mapping); + gfp_mask &= ~(__GFP_FS|__GFP_IO); + gfp_mask |= __GFP_HIGH; + mapping_set_gfp_mask(mapping, gfp_mask); + } + + return 0; +} + +static struct block_device_operations rd_bd_op = { + .owner = THIS_MODULE, + .open = rd_open, + .ioctl = rd_ioctl, +}; + +/* + * Before freeing the module, invalidate all of the protected buffers! + */ +static void __exit rd_cleanup(void) +{ + int i; + + for (i = 0; i < CONFIG_BLK_DEV_RAM_COUNT; i++) { + struct block_device *bdev = rd_bdev[i]; + rd_bdev[i] = NULL; + if (bdev) { + invalidate_bdev(bdev, 1); + blkdev_put(bdev); + } + del_gendisk(rd_disks[i]); + put_disk(rd_disks[i]); + blk_cleanup_queue(rd_queue[i]); + } + devfs_remove("rd"); + unregister_blkdev(RAMDISK_MAJOR, "ramdisk"); +} + +/* + * This is the registration and initialization section of the RAM disk driver + */ +static int __init rd_init(void) +{ + int i; + int err = -ENOMEM; + + if (rd_blocksize > PAGE_SIZE || rd_blocksize < 512 || + (rd_blocksize & (rd_blocksize-1))) { + printk("RAMDISK: wrong blocksize %d, reverting to defaults\n", + rd_blocksize); + rd_blocksize = BLOCK_SIZE; + } + + for (i = 0; i < CONFIG_BLK_DEV_RAM_COUNT; i++) { + rd_disks[i] = alloc_disk(1); + if (!rd_disks[i]) + goto out; + } + + if (register_blkdev(RAMDISK_MAJOR, "ramdisk")) { + err = -EIO; + goto out; + } + + devfs_mk_dir("rd"); + + for (i = 0; i < CONFIG_BLK_DEV_RAM_COUNT; i++) { + struct gendisk *disk = rd_disks[i]; + + rd_queue[i] = blk_alloc_queue(GFP_KERNEL); + if (!rd_queue[i]) + goto out_queue; + + blk_queue_make_request(rd_queue[i], &rd_make_request); + blk_queue_hardsect_size(rd_queue[i], rd_blocksize); + + /* rd_size is given in kB */ + disk->major = RAMDISK_MAJOR; + disk->first_minor = i; + disk->fops = &rd_bd_op; + disk->queue = rd_queue[i]; + disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO; + sprintf(disk->disk_name, "ram%d", i); + sprintf(disk->devfs_name, "rd/%d", i); + set_capacity(disk, rd_size * 2); + add_disk(rd_disks[i]); + } + + /* rd_size is given in kB */ + printk("RAMDISK driver initialized: " + "%d RAM disks of %dK size %d blocksize\n", + CONFIG_BLK_DEV_RAM_COUNT, rd_size, rd_blocksize); + + return 0; +out_queue: + unregister_blkdev(RAMDISK_MAJOR, "ramdisk"); +out: + while (i--) { + put_disk(rd_disks[i]); + blk_cleanup_queue(rd_queue[i]); + } + return err; +} + +module_init(rd_init); +module_exit(rd_cleanup); + +/* options - nonmodular */ +#ifndef MODULE +static int __init ramdisk_size(char *str) +{ + rd_size = simple_strtol(str,NULL,0); + return 1; +} +static int __init ramdisk_size2(char *str) /* kludge */ +{ + return ramdisk_size(str); +} +static int __init ramdisk_blocksize(char *str) +{ + rd_blocksize = simple_strtol(str,NULL,0); + return 1; +} +__setup("ramdisk=", ramdisk_size); +__setup("ramdisk_size=", ramdisk_size2); +__setup("ramdisk_blocksize=", ramdisk_blocksize); +#endif + +/* options - modular */ +module_param(rd_size, int, 0); +MODULE_PARM_DESC(rd_size, "Size of each RAM disk in kbytes."); +module_param(rd_blocksize, int, 0); +MODULE_PARM_DESC(rd_blocksize, "Blocksize of each RAM disk in bytes."); +MODULE_ALIAS_BLOCKDEV_MAJOR(RAMDISK_MAJOR); + +MODULE_LICENSE("GPL"); diff --git a/drivers/block/scsi_ioctl.c b/drivers/block/scsi_ioctl.c new file mode 100644 index 00000000000..689527a89de --- /dev/null +++ b/drivers/block/scsi_ioctl.c @@ -0,0 +1,580 @@ +/* + * Copyright (C) 2001 Jens Axboe <axboe@suse.de> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public Licens + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111- + * + */ +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/string.h> +#include <linux/module.h> +#include <linux/blkdev.h> +#include <linux/completion.h> +#include <linux/cdrom.h> +#include <linux/slab.h> +#include <linux/times.h> +#include <asm/uaccess.h> + +#include <scsi/scsi.h> +#include <scsi/scsi_ioctl.h> +#include <scsi/scsi_cmnd.h> + +/* Command group 3 is reserved and should never be used. */ +const unsigned char scsi_command_size[8] = +{ + 6, 10, 10, 12, + 16, 12, 10, 10 +}; + +EXPORT_SYMBOL(scsi_command_size); + +#define BLK_DEFAULT_TIMEOUT (60 * HZ) + +#include <scsi/sg.h> + +static int sg_get_version(int __user *p) +{ + static int sg_version_num = 30527; + return put_user(sg_version_num, p); +} + +static int scsi_get_idlun(request_queue_t *q, int __user *p) +{ + return put_user(0, p); +} + +static int scsi_get_bus(request_queue_t *q, int __user *p) +{ + return put_user(0, p); +} + +static int sg_get_timeout(request_queue_t *q) +{ + return q->sg_timeout / (HZ / USER_HZ); +} + +static int sg_set_timeout(request_queue_t *q, int __user *p) +{ + int timeout, err = get_user(timeout, p); + + if (!err) + q->sg_timeout = timeout * (HZ / USER_HZ); + + return err; +} + +static int sg_get_reserved_size(request_queue_t *q, int __user *p) +{ + return put_user(q->sg_reserved_size, p); +} + +static int sg_set_reserved_size(request_queue_t *q, int __user *p) +{ + int size, err = get_user(size, p); + + if (err) + return err; + + if (size < 0) + return -EINVAL; + if (size > (q->max_sectors << 9)) + size = q->max_sectors << 9; + + q->sg_reserved_size = size; + return 0; +} + +/* + * will always return that we are ATAPI even for a real SCSI drive, I'm not + * so sure this is worth doing anything about (why would you care??) + */ +static int sg_emulated_host(request_queue_t *q, int __user *p) +{ + return put_user(1, p); +} + +#define CMD_READ_SAFE 0x01 +#define CMD_WRITE_SAFE 0x02 +#define CMD_WARNED 0x04 +#define safe_for_read(cmd) [cmd] = CMD_READ_SAFE +#define safe_for_write(cmd) [cmd] = CMD_WRITE_SAFE + +static int verify_command(struct file *file, unsigned char *cmd) +{ + static unsigned char cmd_type[256] = { + + /* Basic read-only commands */ + safe_for_read(TEST_UNIT_READY), + safe_for_read(REQUEST_SENSE), + safe_for_read(READ_6), + safe_for_read(READ_10), + safe_for_read(READ_12), + safe_for_read(READ_16), + safe_for_read(READ_BUFFER), + safe_for_read(READ_LONG), + safe_for_read(INQUIRY), + safe_for_read(MODE_SENSE), + safe_for_read(MODE_SENSE_10), + safe_for_read(LOG_SENSE), + safe_for_read(START_STOP), + safe_for_read(GPCMD_VERIFY_10), + safe_for_read(VERIFY_16), + + /* Audio CD commands */ + safe_for_read(GPCMD_PLAY_CD), + safe_for_read(GPCMD_PLAY_AUDIO_10), + safe_for_read(GPCMD_PLAY_AUDIO_MSF), + safe_for_read(GPCMD_PLAY_AUDIO_TI), + safe_for_read(GPCMD_PAUSE_RESUME), + + /* CD/DVD data reading */ + safe_for_read(GPCMD_READ_BUFFER_CAPACITY), + safe_for_read(GPCMD_READ_CD), + safe_for_read(GPCMD_READ_CD_MSF), + safe_for_read(GPCMD_READ_DISC_INFO), + safe_for_read(GPCMD_READ_CDVD_CAPACITY), + safe_for_read(GPCMD_READ_DVD_STRUCTURE), + safe_for_read(GPCMD_READ_HEADER), + safe_for_read(GPCMD_READ_TRACK_RZONE_INFO), + safe_for_read(GPCMD_READ_SUBCHANNEL), + safe_for_read(GPCMD_READ_TOC_PMA_ATIP), + safe_for_read(GPCMD_REPORT_KEY), + safe_for_read(GPCMD_SCAN), + safe_for_read(GPCMD_GET_CONFIGURATION), + safe_for_read(GPCMD_READ_FORMAT_CAPACITIES), + safe_for_read(GPCMD_GET_EVENT_STATUS_NOTIFICATION), + safe_for_read(GPCMD_GET_PERFORMANCE), + safe_for_read(GPCMD_SEEK), + safe_for_read(GPCMD_STOP_PLAY_SCAN), + + /* Basic writing commands */ + safe_for_write(WRITE_6), + safe_for_write(WRITE_10), + safe_for_write(WRITE_VERIFY), + safe_for_write(WRITE_12), + safe_for_write(WRITE_VERIFY_12), + safe_for_write(WRITE_16), + safe_for_write(WRITE_LONG), + safe_for_write(ERASE), + safe_for_write(GPCMD_MODE_SELECT_10), + safe_for_write(MODE_SELECT), + safe_for_write(LOG_SELECT), + safe_for_write(GPCMD_BLANK), + safe_for_write(GPCMD_CLOSE_TRACK), + safe_for_write(GPCMD_FLUSH_CACHE), + safe_for_write(GPCMD_FORMAT_UNIT), + safe_for_write(GPCMD_REPAIR_RZONE_TRACK), + safe_for_write(GPCMD_RESERVE_RZONE_TRACK), + safe_for_write(GPCMD_SEND_DVD_STRUCTURE), + safe_for_write(GPCMD_SEND_EVENT), + safe_for_write(GPCMD_SEND_KEY), + safe_for_write(GPCMD_SEND_OPC), + safe_for_write(GPCMD_SEND_CUE_SHEET), + safe_for_write(GPCMD_SET_SPEED), + safe_for_write(GPCMD_PREVENT_ALLOW_MEDIUM_REMOVAL), + safe_for_write(GPCMD_LOAD_UNLOAD), + safe_for_write(GPCMD_SET_STREAMING), + }; + unsigned char type = cmd_type[cmd[0]]; + + /* Anybody who can open the device can do a read-safe command */ + if (type & CMD_READ_SAFE) + return 0; + + /* Write-safe commands just require a writable open.. */ + if (type & CMD_WRITE_SAFE) { + if (file->f_mode & FMODE_WRITE) + return 0; + } + + if (!type) { + cmd_type[cmd[0]] = CMD_WARNED; + printk(KERN_WARNING "scsi: unknown opcode 0x%02x\n", cmd[0]); + } + + /* And root can do any command.. */ + if (capable(CAP_SYS_RAWIO)) + return 0; + + /* Otherwise fail it with an "Operation not permitted" */ + return -EPERM; +} + +static int sg_io(struct file *file, request_queue_t *q, + struct gendisk *bd_disk, struct sg_io_hdr *hdr) +{ + unsigned long start_time; + int reading, writing; + struct request *rq; + struct bio *bio; + char sense[SCSI_SENSE_BUFFERSIZE]; + unsigned char cmd[BLK_MAX_CDB]; + + if (hdr->interface_id != 'S') + return -EINVAL; + if (hdr->cmd_len > BLK_MAX_CDB) + return -EINVAL; + if (copy_from_user(cmd, hdr->cmdp, hdr->cmd_len)) + return -EFAULT; + if (verify_command(file, cmd)) + return -EPERM; + + /* + * we'll do that later + */ + if (hdr->iovec_count) + return -EOPNOTSUPP; + + if (hdr->dxfer_len > (q->max_sectors << 9)) + return -EIO; + + reading = writing = 0; + if (hdr->dxfer_len) { + switch (hdr->dxfer_direction) { + default: + return -EINVAL; + case SG_DXFER_TO_FROM_DEV: + reading = 1; + /* fall through */ + case SG_DXFER_TO_DEV: + writing = 1; + break; + case SG_DXFER_FROM_DEV: + reading = 1; + break; + } + + rq = blk_rq_map_user(q, writing ? WRITE : READ, hdr->dxferp, + hdr->dxfer_len); + + if (IS_ERR(rq)) + return PTR_ERR(rq); + } else + rq = blk_get_request(q, READ, __GFP_WAIT); + + /* + * fill in request structure + */ + rq->cmd_len = hdr->cmd_len; + memcpy(rq->cmd, cmd, hdr->cmd_len); + if (sizeof(rq->cmd) != hdr->cmd_len) + memset(rq->cmd + hdr->cmd_len, 0, sizeof(rq->cmd) - hdr->cmd_len); + + memset(sense, 0, sizeof(sense)); + rq->sense = sense; + rq->sense_len = 0; + + rq->flags |= REQ_BLOCK_PC; + bio = rq->bio; + + /* + * bounce this after holding a reference to the original bio, it's + * needed for proper unmapping + */ + if (rq->bio) + blk_queue_bounce(q, &rq->bio); + + rq->timeout = (hdr->timeout * HZ) / 1000; + if (!rq->timeout) + rq->timeout = q->sg_timeout; + if (!rq->timeout) + rq->timeout = BLK_DEFAULT_TIMEOUT; + + start_time = jiffies; + + /* ignore return value. All information is passed back to caller + * (if he doesn't check that is his problem). + * N.B. a non-zero SCSI status is _not_ necessarily an error. + */ + blk_execute_rq(q, bd_disk, rq); + + /* write to all output members */ + hdr->status = 0xff & rq->errors; + hdr->masked_status = status_byte(rq->errors); + hdr->msg_status = msg_byte(rq->errors); + hdr->host_status = host_byte(rq->errors); + hdr->driver_status = driver_byte(rq->errors); + hdr->info = 0; + if (hdr->masked_status || hdr->host_status || hdr->driver_status) + hdr->info |= SG_INFO_CHECK; + hdr->resid = rq->data_len; + hdr->duration = ((jiffies - start_time) * 1000) / HZ; + hdr->sb_len_wr = 0; + + if (rq->sense_len && hdr->sbp) { + int len = min((unsigned int) hdr->mx_sb_len, rq->sense_len); + + if (!copy_to_user(hdr->sbp, rq->sense, len)) + hdr->sb_len_wr = len; + } + + if (blk_rq_unmap_user(rq, bio, hdr->dxfer_len)) + return -EFAULT; + + /* may not have succeeded, but output values written to control + * structure (struct sg_io_hdr). */ + return 0; +} + +#define FORMAT_UNIT_TIMEOUT (2 * 60 * 60 * HZ) +#define START_STOP_TIMEOUT (60 * HZ) +#define MOVE_MEDIUM_TIMEOUT (5 * 60 * HZ) +#define READ_ELEMENT_STATUS_TIMEOUT (5 * 60 * HZ) +#define READ_DEFECT_DATA_TIMEOUT (60 * HZ ) +#define OMAX_SB_LEN 16 /* For backward compatibility */ + +static int sg_scsi_ioctl(struct file *file, request_queue_t *q, + struct gendisk *bd_disk, Scsi_Ioctl_Command __user *sic) +{ + struct request *rq; + int err; + unsigned int in_len, out_len, bytes, opcode, cmdlen; + char *buffer = NULL, sense[SCSI_SENSE_BUFFERSIZE]; + + /* + * get in an out lengths, verify they don't exceed a page worth of data + */ + if (get_user(in_len, &sic->inlen)) + return -EFAULT; + if (get_user(out_len, &sic->outlen)) + return -EFAULT; + if (in_len > PAGE_SIZE || out_len > PAGE_SIZE) + return -EINVAL; + if (get_user(opcode, sic->data)) + return -EFAULT; + + bytes = max(in_len, out_len); + if (bytes) { + buffer = kmalloc(bytes, q->bounce_gfp | GFP_USER| __GFP_NOWARN); + if (!buffer) + return -ENOMEM; + + memset(buffer, 0, bytes); + } + + rq = blk_get_request(q, in_len ? WRITE : READ, __GFP_WAIT); + + cmdlen = COMMAND_SIZE(opcode); + + /* + * get command and data to send to device, if any + */ + err = -EFAULT; + rq->cmd_len = cmdlen; + if (copy_from_user(rq->cmd, sic->data, cmdlen)) + goto error; + + if (copy_from_user(buffer, sic->data + cmdlen, in_len)) + goto error; + + err = verify_command(file, rq->cmd); + if (err) + goto error; + + switch (opcode) { + case SEND_DIAGNOSTIC: + case FORMAT_UNIT: + rq->timeout = FORMAT_UNIT_TIMEOUT; + break; + case START_STOP: + rq->timeout = START_STOP_TIMEOUT; + break; + case MOVE_MEDIUM: + rq->timeout = MOVE_MEDIUM_TIMEOUT; + break; + case READ_ELEMENT_STATUS: + rq->timeout = READ_ELEMENT_STATUS_TIMEOUT; + break; + case READ_DEFECT_DATA: + rq->timeout = READ_DEFECT_DATA_TIMEOUT; + break; + default: + rq->timeout = BLK_DEFAULT_TIMEOUT; + break; + } + + memset(sense, 0, sizeof(sense)); + rq->sense = sense; + rq->sense_len = 0; + + rq->data = buffer; + rq->data_len = bytes; + rq->flags |= REQ_BLOCK_PC; + + blk_execute_rq(q, bd_disk, rq); + err = rq->errors & 0xff; /* only 8 bit SCSI status */ + if (err) { + if (rq->sense_len && rq->sense) { + bytes = (OMAX_SB_LEN > rq->sense_len) ? + rq->sense_len : OMAX_SB_LEN; + if (copy_to_user(sic->data, rq->sense, bytes)) + err = -EFAULT; + } + } else { + if (copy_to_user(sic->data, buffer, out_len)) + err = -EFAULT; + } + +error: + kfree(buffer); + blk_put_request(rq); + return err; +} + +int scsi_cmd_ioctl(struct file *file, struct gendisk *bd_disk, unsigned int cmd, void __user *arg) +{ + request_queue_t *q; + struct request *rq; + int close = 0, err; + + q = bd_disk->queue; + if (!q) + return -ENXIO; + + if (blk_get_queue(q)) + return -ENXIO; + + switch (cmd) { + /* + * new sgv3 interface + */ + case SG_GET_VERSION_NUM: + err = sg_get_version(arg); + break; + case SCSI_IOCTL_GET_IDLUN: + err = scsi_get_idlun(q, arg); + break; + case SCSI_IOCTL_GET_BUS_NUMBER: + err = scsi_get_bus(q, arg); + break; + case SG_SET_TIMEOUT: + err = sg_set_timeout(q, arg); + break; + case SG_GET_TIMEOUT: + err = sg_get_timeout(q); + break; + case SG_GET_RESERVED_SIZE: + err = sg_get_reserved_size(q, arg); + break; + case SG_SET_RESERVED_SIZE: + err = sg_set_reserved_size(q, arg); + break; + case SG_EMULATED_HOST: + err = sg_emulated_host(q, arg); + break; + case SG_IO: { + struct sg_io_hdr hdr; + + err = -EFAULT; + if (copy_from_user(&hdr, arg, sizeof(hdr))) + break; + err = sg_io(file, q, bd_disk, &hdr); + if (err == -EFAULT) + break; + + if (copy_to_user(arg, &hdr, sizeof(hdr))) + err = -EFAULT; + break; + } + case CDROM_SEND_PACKET: { + struct cdrom_generic_command cgc; + struct sg_io_hdr hdr; + + err = -EFAULT; + if (copy_from_user(&cgc, arg, sizeof(cgc))) + break; + cgc.timeout = clock_t_to_jiffies(cgc.timeout); + memset(&hdr, 0, sizeof(hdr)); + hdr.interface_id = 'S'; + hdr.cmd_len = sizeof(cgc.cmd); + hdr.dxfer_len = cgc.buflen; + err = 0; + switch (cgc.data_direction) { + case CGC_DATA_UNKNOWN: + hdr.dxfer_direction = SG_DXFER_UNKNOWN; + break; + case CGC_DATA_WRITE: + hdr.dxfer_direction = SG_DXFER_TO_DEV; + break; + case CGC_DATA_READ: + hdr.dxfer_direction = SG_DXFER_FROM_DEV; + break; + case CGC_DATA_NONE: + hdr.dxfer_direction = SG_DXFER_NONE; + break; + default: + err = -EINVAL; + } + if (err) + break; + + hdr.dxferp = cgc.buffer; + hdr.sbp = cgc.sense; + if (hdr.sbp) + hdr.mx_sb_len = sizeof(struct request_sense); + hdr.timeout = cgc.timeout; + hdr.cmdp = ((struct cdrom_generic_command __user*) arg)->cmd; + hdr.cmd_len = sizeof(cgc.cmd); + + err = sg_io(file, q, bd_disk, &hdr); + if (err == -EFAULT) + break; + + if (hdr.status) + err = -EIO; + + cgc.stat = err; + cgc.buflen = hdr.resid; + if (copy_to_user(arg, &cgc, sizeof(cgc))) + err = -EFAULT; + + break; + } + + /* + * old junk scsi send command ioctl + */ + case SCSI_IOCTL_SEND_COMMAND: + printk(KERN_WARNING "program %s is using a deprecated SCSI ioctl, please convert it to SG_IO\n", current->comm); + err = -EINVAL; + if (!arg) + break; + + err = sg_scsi_ioctl(file, q, bd_disk, arg); + break; + case CDROMCLOSETRAY: + close = 1; + case CDROMEJECT: + rq = blk_get_request(q, WRITE, __GFP_WAIT); + rq->flags |= REQ_BLOCK_PC; + rq->data = NULL; + rq->data_len = 0; + rq->timeout = BLK_DEFAULT_TIMEOUT; + memset(rq->cmd, 0, sizeof(rq->cmd)); + rq->cmd[0] = GPCMD_START_STOP_UNIT; + rq->cmd[4] = 0x02 + (close != 0); + rq->cmd_len = 6; + err = blk_execute_rq(q, bd_disk, rq); + blk_put_request(rq); + break; + default: + err = -ENOTTY; + } + + blk_put_queue(q); + return err; +} + +EXPORT_SYMBOL(scsi_cmd_ioctl); diff --git a/drivers/block/smart1,2.h b/drivers/block/smart1,2.h new file mode 100644 index 00000000000..a0b403a6b4e --- /dev/null +++ b/drivers/block/smart1,2.h @@ -0,0 +1,278 @@ +/* + * Disk Array driver for Compaq SMART2 Controllers + * Copyright 1998 Compaq Computer Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Questions/Comments/Bugfixes to iss_storagedev@hp.com + * + * If you want to make changes, improve or add functionality to this + * driver, you'll probably need the Compaq Array Controller Interface + * Specificiation (Document number ECG086/1198) + */ + +/* + * This file contains the controller communication implementation for + * Compaq SMART-1 and SMART-2 controllers. To the best of my knowledge, + * this should support: + * + * PCI: + * SMART-2/P, SMART-2DH, SMART-2SL, SMART-221, SMART-3100ES, SMART-3200 + * Integerated SMART Array Controller, SMART-4200, SMART-4250ES + * + * EISA: + * SMART-2/E, SMART, IAES, IDA-2, IDA + */ + +/* + * Memory mapped FIFO interface (SMART 42xx cards) + */ +static void smart4_submit_command(ctlr_info_t *h, cmdlist_t *c) +{ + writel(c->busaddr, h->vaddr + S42XX_REQUEST_PORT_OFFSET); +} + +/* + * This card is the opposite of the other cards. + * 0 turns interrupts on... + * 0x08 turns them off... + */ +static void smart4_intr_mask(ctlr_info_t *h, unsigned long val) +{ + if (val) + { /* Turn interrupts on */ + writel(0, h->vaddr + S42XX_REPLY_INTR_MASK_OFFSET); + } else /* Turn them off */ + { + writel( S42XX_INTR_OFF, + h->vaddr + S42XX_REPLY_INTR_MASK_OFFSET); + } +} + +/* + * For older cards FIFO Full = 0. + * On this card 0 means there is room, anything else FIFO Full. + * + */ +static unsigned long smart4_fifo_full(ctlr_info_t *h) +{ + + return (!readl(h->vaddr + S42XX_REQUEST_PORT_OFFSET)); +} + +/* This type of controller returns -1 if the fifo is empty, + * Not 0 like the others. + * And we need to let it know we read a value out + */ +static unsigned long smart4_completed(ctlr_info_t *h) +{ + long register_value + = readl(h->vaddr + S42XX_REPLY_PORT_OFFSET); + + /* Fifo is empty */ + if( register_value == 0xffffffff) + return 0; + + /* Need to let it know we got the reply */ + /* We do this by writing a 0 to the port we just read from */ + writel(0, h->vaddr + S42XX_REPLY_PORT_OFFSET); + + return ((unsigned long) register_value); +} + + /* + * This hardware returns interrupt pending at a different place and + * it does not tell us if the fifo is empty, we will have check + * that by getting a 0 back from the comamnd_completed call. + */ +static unsigned long smart4_intr_pending(ctlr_info_t *h) +{ + unsigned long register_value = + readl(h->vaddr + S42XX_INTR_STATUS); + + if( register_value & S42XX_INTR_PENDING) + return FIFO_NOT_EMPTY; + return 0 ; +} + +static struct access_method smart4_access = { + smart4_submit_command, + smart4_intr_mask, + smart4_fifo_full, + smart4_intr_pending, + smart4_completed, +}; + +/* + * Memory mapped FIFO interface (PCI SMART2 and SMART 3xxx cards) + */ +static void smart2_submit_command(ctlr_info_t *h, cmdlist_t *c) +{ + writel(c->busaddr, h->vaddr + COMMAND_FIFO); +} + +static void smart2_intr_mask(ctlr_info_t *h, unsigned long val) +{ + writel(val, h->vaddr + INTR_MASK); +} + +static unsigned long smart2_fifo_full(ctlr_info_t *h) +{ + return readl(h->vaddr + COMMAND_FIFO); +} + +static unsigned long smart2_completed(ctlr_info_t *h) +{ + return readl(h->vaddr + COMMAND_COMPLETE_FIFO); +} + +static unsigned long smart2_intr_pending(ctlr_info_t *h) +{ + return readl(h->vaddr + INTR_PENDING); +} + +static struct access_method smart2_access = { + smart2_submit_command, + smart2_intr_mask, + smart2_fifo_full, + smart2_intr_pending, + smart2_completed, +}; + +/* + * IO access for SMART-2/E cards + */ +static void smart2e_submit_command(ctlr_info_t *h, cmdlist_t *c) +{ + outl(c->busaddr, h->io_mem_addr + COMMAND_FIFO); +} + +static void smart2e_intr_mask(ctlr_info_t *h, unsigned long val) +{ + outl(val, h->io_mem_addr + INTR_MASK); +} + +static unsigned long smart2e_fifo_full(ctlr_info_t *h) +{ + return inl(h->io_mem_addr + COMMAND_FIFO); +} + +static unsigned long smart2e_completed(ctlr_info_t *h) +{ + return inl(h->io_mem_addr + COMMAND_COMPLETE_FIFO); +} + +static unsigned long smart2e_intr_pending(ctlr_info_t *h) +{ + return inl(h->io_mem_addr + INTR_PENDING); +} + +static struct access_method smart2e_access = { + smart2e_submit_command, + smart2e_intr_mask, + smart2e_fifo_full, + smart2e_intr_pending, + smart2e_completed, +}; + +/* + * IO access for older SMART-1 type cards + */ +#define SMART1_SYSTEM_MASK 0xC8E +#define SMART1_SYSTEM_DOORBELL 0xC8F +#define SMART1_LOCAL_MASK 0xC8C +#define SMART1_LOCAL_DOORBELL 0xC8D +#define SMART1_INTR_MASK 0xC89 +#define SMART1_LISTADDR 0xC90 +#define SMART1_LISTLEN 0xC94 +#define SMART1_TAG 0xC97 +#define SMART1_COMPLETE_ADDR 0xC98 +#define SMART1_LISTSTATUS 0xC9E + +#define CHANNEL_BUSY 0x01 +#define CHANNEL_CLEAR 0x02 + +static void smart1_submit_command(ctlr_info_t *h, cmdlist_t *c) +{ + /* + * This __u16 is actually a bunch of control flags on SMART + * and below. We want them all to be zero. + */ + c->hdr.size = 0; + + outb(CHANNEL_CLEAR, h->io_mem_addr + SMART1_SYSTEM_DOORBELL); + + outl(c->busaddr, h->io_mem_addr + SMART1_LISTADDR); + outw(c->size, h->io_mem_addr + SMART1_LISTLEN); + + outb(CHANNEL_BUSY, h->io_mem_addr + SMART1_LOCAL_DOORBELL); +} + +static void smart1_intr_mask(ctlr_info_t *h, unsigned long val) +{ + if (val == 1) { + outb(0xFD, h->io_mem_addr + SMART1_SYSTEM_DOORBELL); + outb(CHANNEL_BUSY, h->io_mem_addr + SMART1_LOCAL_DOORBELL); + outb(0x01, h->io_mem_addr + SMART1_INTR_MASK); + outb(0x01, h->io_mem_addr + SMART1_SYSTEM_MASK); + } else { + outb(0, h->io_mem_addr + 0xC8E); + } +} + +static unsigned long smart1_fifo_full(ctlr_info_t *h) +{ + unsigned char chan; + chan = inb(h->io_mem_addr + SMART1_SYSTEM_DOORBELL) & CHANNEL_CLEAR; + return chan; +} + +static unsigned long smart1_completed(ctlr_info_t *h) +{ + unsigned char status; + unsigned long cmd; + + if (inb(h->io_mem_addr + SMART1_SYSTEM_DOORBELL) & CHANNEL_BUSY) { + outb(CHANNEL_BUSY, h->io_mem_addr + SMART1_SYSTEM_DOORBELL); + + cmd = inl(h->io_mem_addr + SMART1_COMPLETE_ADDR); + status = inb(h->io_mem_addr + SMART1_LISTSTATUS); + + outb(CHANNEL_CLEAR, h->io_mem_addr + SMART1_LOCAL_DOORBELL); + + /* + * this is x86 (actually compaq x86) only, so it's ok + */ + if (cmd) ((cmdlist_t*)bus_to_virt(cmd))->req.hdr.rcode = status; + } else { + cmd = 0; + } + return cmd; +} + +static unsigned long smart1_intr_pending(ctlr_info_t *h) +{ + unsigned char chan; + chan = inb(h->io_mem_addr + SMART1_SYSTEM_DOORBELL) & CHANNEL_BUSY; + return chan; +} + +static struct access_method smart1_access = { + smart1_submit_command, + smart1_intr_mask, + smart1_fifo_full, + smart1_intr_pending, + smart1_completed, +}; diff --git a/drivers/block/swim3.c b/drivers/block/swim3.c new file mode 100644 index 00000000000..5b09cf154ac --- /dev/null +++ b/drivers/block/swim3.c @@ -0,0 +1,1154 @@ +/* + * Driver for the SWIM3 (Super Woz Integrated Machine 3) + * floppy controller found on Power Macintoshes. + * + * Copyright (C) 1996 Paul Mackerras. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ + +/* + * TODO: + * handle 2 drives + * handle GCR disks + */ + +#include <linux/config.h> +#include <linux/stddef.h> +#include <linux/kernel.h> +#include <linux/sched.h> +#include <linux/timer.h> +#include <linux/delay.h> +#include <linux/fd.h> +#include <linux/ioctl.h> +#include <linux/blkdev.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/interrupt.h> +#include <linux/module.h> +#include <asm/io.h> +#include <asm/dbdma.h> +#include <asm/prom.h> +#include <asm/uaccess.h> +#include <asm/mediabay.h> +#include <asm/machdep.h> +#include <asm/pmac_feature.h> + +static struct request_queue *swim3_queue; +static struct gendisk *disks[2]; +static struct request *fd_req; + +#define MAX_FLOPPIES 2 + +enum swim_state { + idle, + locating, + seeking, + settling, + do_transfer, + jogging, + available, + revalidating, + ejecting +}; + +#define REG(x) unsigned char x; char x ## _pad[15]; + +/* + * The names for these registers mostly represent speculation on my part. + * It will be interesting to see how close they are to the names Apple uses. + */ +struct swim3 { + REG(data); + REG(timer); /* counts down at 1MHz */ + REG(error); + REG(mode); + REG(select); /* controls CA0, CA1, CA2 and LSTRB signals */ + REG(setup); + REG(control); /* writing bits clears them */ + REG(status); /* writing bits sets them in control */ + REG(intr); + REG(nseek); /* # tracks to seek */ + REG(ctrack); /* current track number */ + REG(csect); /* current sector number */ + REG(gap3); /* size of gap 3 in track format */ + REG(sector); /* sector # to read or write */ + REG(nsect); /* # sectors to read or write */ + REG(intr_enable); +}; + +#define control_bic control +#define control_bis status + +/* Bits in select register */ +#define CA_MASK 7 +#define LSTRB 8 + +/* Bits in control register */ +#define DO_SEEK 0x80 +#define FORMAT 0x40 +#define SELECT 0x20 +#define WRITE_SECTORS 0x10 +#define DO_ACTION 0x08 +#define DRIVE2_ENABLE 0x04 +#define DRIVE_ENABLE 0x02 +#define INTR_ENABLE 0x01 + +/* Bits in status register */ +#define FIFO_1BYTE 0x80 +#define FIFO_2BYTE 0x40 +#define ERROR 0x20 +#define DATA 0x08 +#define RDDATA 0x04 +#define INTR_PENDING 0x02 +#define MARK_BYTE 0x01 + +/* Bits in intr and intr_enable registers */ +#define ERROR_INTR 0x20 +#define DATA_CHANGED 0x10 +#define TRANSFER_DONE 0x08 +#define SEEN_SECTOR 0x04 +#define SEEK_DONE 0x02 +#define TIMER_DONE 0x01 + +/* Bits in error register */ +#define ERR_DATA_CRC 0x80 +#define ERR_ADDR_CRC 0x40 +#define ERR_OVERRUN 0x04 +#define ERR_UNDERRUN 0x01 + +/* Bits in setup register */ +#define S_SW_RESET 0x80 +#define S_GCR_WRITE 0x40 +#define S_IBM_DRIVE 0x20 +#define S_TEST_MODE 0x10 +#define S_FCLK_DIV2 0x08 +#define S_GCR 0x04 +#define S_COPY_PROT 0x02 +#define S_INV_WDATA 0x01 + +/* Select values for swim3_action */ +#define SEEK_POSITIVE 0 +#define SEEK_NEGATIVE 4 +#define STEP 1 +#define MOTOR_ON 2 +#define MOTOR_OFF 6 +#define INDEX 3 +#define EJECT 7 +#define SETMFM 9 +#define SETGCR 13 + +/* Select values for swim3_select and swim3_readbit */ +#define STEP_DIR 0 +#define STEPPING 1 +#define MOTOR_ON 2 +#define RELAX 3 /* also eject in progress */ +#define READ_DATA_0 4 +#define TWOMEG_DRIVE 5 +#define SINGLE_SIDED 6 /* drive or diskette is 4MB type? */ +#define DRIVE_PRESENT 7 +#define DISK_IN 8 +#define WRITE_PROT 9 +#define TRACK_ZERO 10 +#define TACHO 11 +#define READ_DATA_1 12 +#define MFM_MODE 13 +#define SEEK_COMPLETE 14 +#define ONEMEG_MEDIA 15 + +/* Definitions of values used in writing and formatting */ +#define DATA_ESCAPE 0x99 +#define GCR_SYNC_EXC 0x3f +#define GCR_SYNC_CONV 0x80 +#define GCR_FIRST_MARK 0xd5 +#define GCR_SECOND_MARK 0xaa +#define GCR_ADDR_MARK "\xd5\xaa\x00" +#define GCR_DATA_MARK "\xd5\xaa\x0b" +#define GCR_SLIP_BYTE "\x27\xaa" +#define GCR_SELF_SYNC "\x3f\xbf\x1e\x34\x3c\x3f" + +#define DATA_99 "\x99\x99" +#define MFM_ADDR_MARK "\x99\xa1\x99\xa1\x99\xa1\x99\xfe" +#define MFM_INDEX_MARK "\x99\xc2\x99\xc2\x99\xc2\x99\xfc" +#define MFM_GAP_LEN 12 + +struct floppy_state { + enum swim_state state; + struct swim3 __iomem *swim3; /* hardware registers */ + struct dbdma_regs __iomem *dma; /* DMA controller registers */ + int swim3_intr; /* interrupt number for SWIM3 */ + int dma_intr; /* interrupt number for DMA channel */ + int cur_cyl; /* cylinder head is on, or -1 */ + int cur_sector; /* last sector we saw go past */ + int req_cyl; /* the cylinder for the current r/w request */ + int head; /* head number ditto */ + int req_sector; /* sector number ditto */ + int scount; /* # sectors we're transferring at present */ + int retries; + int settle_time; + int secpercyl; /* disk geometry information */ + int secpertrack; + int total_secs; + int write_prot; /* 1 if write-protected, 0 if not, -1 dunno */ + struct dbdma_cmd *dma_cmd; + int ref_count; + int expect_cyl; + struct timer_list timeout; + int timeout_pending; + int ejected; + wait_queue_head_t wait; + int wanted; + struct device_node* media_bay; /* NULL when not in bay */ + char dbdma_cmd_space[5 * sizeof(struct dbdma_cmd)]; +}; + +static struct floppy_state floppy_states[MAX_FLOPPIES]; +static int floppy_count = 0; +static DEFINE_SPINLOCK(swim3_lock); + +static unsigned short write_preamble[] = { + 0x4e4e, 0x4e4e, 0x4e4e, 0x4e4e, 0x4e4e, /* gap field */ + 0, 0, 0, 0, 0, 0, /* sync field */ + 0x99a1, 0x99a1, 0x99a1, 0x99fb, /* data address mark */ + 0x990f /* no escape for 512 bytes */ +}; + +static unsigned short write_postamble[] = { + 0x9904, /* insert CRC */ + 0x4e4e, 0x4e4e, + 0x9908, /* stop writing */ + 0, 0, 0, 0, 0, 0 +}; + +static void swim3_select(struct floppy_state *fs, int sel); +static void swim3_action(struct floppy_state *fs, int action); +static int swim3_readbit(struct floppy_state *fs, int bit); +static void do_fd_request(request_queue_t * q); +static void start_request(struct floppy_state *fs); +static void set_timeout(struct floppy_state *fs, int nticks, + void (*proc)(unsigned long)); +static void scan_track(struct floppy_state *fs); +static void seek_track(struct floppy_state *fs, int n); +static void init_dma(struct dbdma_cmd *cp, int cmd, void *buf, int count); +static void setup_transfer(struct floppy_state *fs); +static void act(struct floppy_state *fs); +static void scan_timeout(unsigned long data); +static void seek_timeout(unsigned long data); +static void settle_timeout(unsigned long data); +static void xfer_timeout(unsigned long data); +static irqreturn_t swim3_interrupt(int irq, void *dev_id, struct pt_regs *regs); +/*static void fd_dma_interrupt(int irq, void *dev_id, struct pt_regs *regs);*/ +static int grab_drive(struct floppy_state *fs, enum swim_state state, + int interruptible); +static void release_drive(struct floppy_state *fs); +static int fd_eject(struct floppy_state *fs); +static int floppy_ioctl(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long param); +static int floppy_open(struct inode *inode, struct file *filp); +static int floppy_release(struct inode *inode, struct file *filp); +static int floppy_check_change(struct gendisk *disk); +static int floppy_revalidate(struct gendisk *disk); +static int swim3_add_device(struct device_node *swims); +int swim3_init(void); + +#ifndef CONFIG_PMAC_PBOOK +#define check_media_bay(which, what) 1 +#endif + +static void swim3_select(struct floppy_state *fs, int sel) +{ + struct swim3 __iomem *sw = fs->swim3; + + out_8(&sw->select, RELAX); + if (sel & 8) + out_8(&sw->control_bis, SELECT); + else + out_8(&sw->control_bic, SELECT); + out_8(&sw->select, sel & CA_MASK); +} + +static void swim3_action(struct floppy_state *fs, int action) +{ + struct swim3 __iomem *sw = fs->swim3; + + swim3_select(fs, action); + udelay(1); + out_8(&sw->select, sw->select | LSTRB); + udelay(2); + out_8(&sw->select, sw->select & ~LSTRB); + udelay(1); +} + +static int swim3_readbit(struct floppy_state *fs, int bit) +{ + struct swim3 __iomem *sw = fs->swim3; + int stat; + + swim3_select(fs, bit); + udelay(1); + stat = in_8(&sw->status); + return (stat & DATA) == 0; +} + +static void do_fd_request(request_queue_t * q) +{ + int i; + for(i=0;i<floppy_count;i++) + { + if (floppy_states[i].media_bay && + check_media_bay(floppy_states[i].media_bay, MB_FD)) + continue; + start_request(&floppy_states[i]); + } + sti(); +} + +static void start_request(struct floppy_state *fs) +{ + struct request *req; + unsigned long x; + + if (fs->state == idle && fs->wanted) { + fs->state = available; + wake_up(&fs->wait); + return; + } + while (fs->state == idle && (req = elv_next_request(swim3_queue))) { +#if 0 + printk("do_fd_req: dev=%s cmd=%d sec=%ld nr_sec=%ld buf=%p\n", + req->rq_disk->disk_name, req->cmd, + (long)req->sector, req->nr_sectors, req->buffer); + printk(" rq_status=%d errors=%d current_nr_sectors=%ld\n", + req->rq_status, req->errors, req->current_nr_sectors); +#endif + + if (req->sector < 0 || req->sector >= fs->total_secs) { + end_request(req, 0); + continue; + } + if (req->current_nr_sectors == 0) { + end_request(req, 1); + continue; + } + if (fs->ejected) { + end_request(req, 0); + continue; + } + + if (rq_data_dir(req) == WRITE) { + if (fs->write_prot < 0) + fs->write_prot = swim3_readbit(fs, WRITE_PROT); + if (fs->write_prot) { + end_request(req, 0); + continue; + } + } + + /* Do not remove the cast. req->sector is now a sector_t and + * can be 64 bits, but it will never go past 32 bits for this + * driver anyway, so we can safely cast it down and not have + * to do a 64/32 division + */ + fs->req_cyl = ((long)req->sector) / fs->secpercyl; + x = ((long)req->sector) % fs->secpercyl; + fs->head = x / fs->secpertrack; + fs->req_sector = x % fs->secpertrack + 1; + fd_req = req; + fs->state = do_transfer; + fs->retries = 0; + + act(fs); + } +} + +static void set_timeout(struct floppy_state *fs, int nticks, + void (*proc)(unsigned long)) +{ + unsigned long flags; + + save_flags(flags); cli(); + if (fs->timeout_pending) + del_timer(&fs->timeout); + fs->timeout.expires = jiffies + nticks; + fs->timeout.function = proc; + fs->timeout.data = (unsigned long) fs; + add_timer(&fs->timeout); + fs->timeout_pending = 1; + restore_flags(flags); +} + +static inline void scan_track(struct floppy_state *fs) +{ + struct swim3 __iomem *sw = fs->swim3; + + swim3_select(fs, READ_DATA_0); + in_8(&sw->intr); /* clear SEEN_SECTOR bit */ + in_8(&sw->error); + out_8(&sw->intr_enable, SEEN_SECTOR); + out_8(&sw->control_bis, DO_ACTION); + /* enable intr when track found */ + set_timeout(fs, HZ, scan_timeout); /* enable timeout */ +} + +static inline void seek_track(struct floppy_state *fs, int n) +{ + struct swim3 __iomem *sw = fs->swim3; + + if (n >= 0) { + swim3_action(fs, SEEK_POSITIVE); + sw->nseek = n; + } else { + swim3_action(fs, SEEK_NEGATIVE); + sw->nseek = -n; + } + fs->expect_cyl = (fs->cur_cyl >= 0)? fs->cur_cyl + n: -1; + swim3_select(fs, STEP); + in_8(&sw->error); + /* enable intr when seek finished */ + out_8(&sw->intr_enable, SEEK_DONE); + out_8(&sw->control_bis, DO_SEEK); + set_timeout(fs, 3*HZ, seek_timeout); /* enable timeout */ + fs->settle_time = 0; +} + +static inline void init_dma(struct dbdma_cmd *cp, int cmd, + void *buf, int count) +{ + st_le16(&cp->req_count, count); + st_le16(&cp->command, cmd); + st_le32(&cp->phy_addr, virt_to_bus(buf)); + cp->xfer_status = 0; +} + +static inline void setup_transfer(struct floppy_state *fs) +{ + int n; + struct swim3 __iomem *sw = fs->swim3; + struct dbdma_cmd *cp = fs->dma_cmd; + struct dbdma_regs __iomem *dr = fs->dma; + + if (fd_req->current_nr_sectors <= 0) { + printk(KERN_ERR "swim3: transfer 0 sectors?\n"); + return; + } + if (rq_data_dir(fd_req) == WRITE) + n = 1; + else { + n = fs->secpertrack - fs->req_sector + 1; + if (n > fd_req->current_nr_sectors) + n = fd_req->current_nr_sectors; + } + fs->scount = n; + swim3_select(fs, fs->head? READ_DATA_1: READ_DATA_0); + out_8(&sw->sector, fs->req_sector); + out_8(&sw->nsect, n); + out_8(&sw->gap3, 0); + out_le32(&dr->cmdptr, virt_to_bus(cp)); + if (rq_data_dir(fd_req) == WRITE) { + /* Set up 3 dma commands: write preamble, data, postamble */ + init_dma(cp, OUTPUT_MORE, write_preamble, sizeof(write_preamble)); + ++cp; + init_dma(cp, OUTPUT_MORE, fd_req->buffer, 512); + ++cp; + init_dma(cp, OUTPUT_LAST, write_postamble, sizeof(write_postamble)); + } else { + init_dma(cp, INPUT_LAST, fd_req->buffer, n * 512); + } + ++cp; + out_le16(&cp->command, DBDMA_STOP); + out_8(&sw->control_bic, DO_ACTION | WRITE_SECTORS); + in_8(&sw->error); + out_8(&sw->control_bic, DO_ACTION | WRITE_SECTORS); + if (rq_data_dir(fd_req) == WRITE) + out_8(&sw->control_bis, WRITE_SECTORS); + in_8(&sw->intr); + out_le32(&dr->control, (RUN << 16) | RUN); + /* enable intr when transfer complete */ + out_8(&sw->intr_enable, TRANSFER_DONE); + out_8(&sw->control_bis, DO_ACTION); + set_timeout(fs, 2*HZ, xfer_timeout); /* enable timeout */ +} + +static void act(struct floppy_state *fs) +{ + for (;;) { + switch (fs->state) { + case idle: + return; /* XXX shouldn't get here */ + + case locating: + if (swim3_readbit(fs, TRACK_ZERO)) { + fs->cur_cyl = 0; + if (fs->req_cyl == 0) + fs->state = do_transfer; + else + fs->state = seeking; + break; + } + scan_track(fs); + return; + + case seeking: + if (fs->cur_cyl < 0) { + fs->expect_cyl = -1; + fs->state = locating; + break; + } + if (fs->req_cyl == fs->cur_cyl) { + printk("whoops, seeking 0\n"); + fs->state = do_transfer; + break; + } + seek_track(fs, fs->req_cyl - fs->cur_cyl); + return; + + case settling: + /* check for SEEK_COMPLETE after 30ms */ + fs->settle_time = (HZ + 32) / 33; + set_timeout(fs, fs->settle_time, settle_timeout); + return; + + case do_transfer: + if (fs->cur_cyl != fs->req_cyl) { + if (fs->retries > 5) { + end_request(fd_req, 0); + fs->state = idle; + return; + } + fs->state = seeking; + break; + } + setup_transfer(fs); + return; + + case jogging: + seek_track(fs, -5); + return; + + default: + printk(KERN_ERR"swim3: unknown state %d\n", fs->state); + return; + } + } +} + +static void scan_timeout(unsigned long data) +{ + struct floppy_state *fs = (struct floppy_state *) data; + struct swim3 __iomem *sw = fs->swim3; + + fs->timeout_pending = 0; + out_8(&sw->control_bic, DO_ACTION | WRITE_SECTORS); + out_8(&sw->select, RELAX); + out_8(&sw->intr_enable, 0); + fs->cur_cyl = -1; + if (fs->retries > 5) { + end_request(fd_req, 0); + fs->state = idle; + start_request(fs); + } else { + fs->state = jogging; + act(fs); + } +} + +static void seek_timeout(unsigned long data) +{ + struct floppy_state *fs = (struct floppy_state *) data; + struct swim3 __iomem *sw = fs->swim3; + + fs->timeout_pending = 0; + out_8(&sw->control_bic, DO_SEEK); + out_8(&sw->select, RELAX); + out_8(&sw->intr_enable, 0); + printk(KERN_ERR "swim3: seek timeout\n"); + end_request(fd_req, 0); + fs->state = idle; + start_request(fs); +} + +static void settle_timeout(unsigned long data) +{ + struct floppy_state *fs = (struct floppy_state *) data; + struct swim3 __iomem *sw = fs->swim3; + + fs->timeout_pending = 0; + if (swim3_readbit(fs, SEEK_COMPLETE)) { + out_8(&sw->select, RELAX); + fs->state = locating; + act(fs); + return; + } + out_8(&sw->select, RELAX); + if (fs->settle_time < 2*HZ) { + ++fs->settle_time; + set_timeout(fs, 1, settle_timeout); + return; + } + printk(KERN_ERR "swim3: seek settle timeout\n"); + end_request(fd_req, 0); + fs->state = idle; + start_request(fs); +} + +static void xfer_timeout(unsigned long data) +{ + struct floppy_state *fs = (struct floppy_state *) data; + struct swim3 __iomem *sw = fs->swim3; + struct dbdma_regs __iomem *dr = fs->dma; + struct dbdma_cmd *cp = fs->dma_cmd; + unsigned long s; + int n; + + fs->timeout_pending = 0; + out_le32(&dr->control, RUN << 16); + /* We must wait a bit for dbdma to stop */ + for (n = 0; (in_le32(&dr->status) & ACTIVE) && n < 1000; n++) + udelay(1); + out_8(&sw->intr_enable, 0); + out_8(&sw->control_bic, WRITE_SECTORS | DO_ACTION); + out_8(&sw->select, RELAX); + if (rq_data_dir(fd_req) == WRITE) + ++cp; + if (ld_le16(&cp->xfer_status) != 0) + s = fs->scount - ((ld_le16(&cp->res_count) + 511) >> 9); + else + s = 0; + fd_req->sector += s; + fd_req->current_nr_sectors -= s; + printk(KERN_ERR "swim3: timeout %sing sector %ld\n", + (rq_data_dir(fd_req)==WRITE? "writ": "read"), (long)fd_req->sector); + end_request(fd_req, 0); + fs->state = idle; + start_request(fs); +} + +static irqreturn_t swim3_interrupt(int irq, void *dev_id, struct pt_regs *regs) +{ + struct floppy_state *fs = (struct floppy_state *) dev_id; + struct swim3 __iomem *sw = fs->swim3; + int intr, err, n; + int stat, resid; + struct dbdma_regs __iomem *dr; + struct dbdma_cmd *cp; + + intr = in_8(&sw->intr); + err = (intr & ERROR_INTR)? in_8(&sw->error): 0; + if ((intr & ERROR_INTR) && fs->state != do_transfer) + printk(KERN_ERR "swim3_interrupt, state=%d, dir=%lx, intr=%x, err=%x\n", + fs->state, rq_data_dir(fd_req), intr, err); + switch (fs->state) { + case locating: + if (intr & SEEN_SECTOR) { + out_8(&sw->control_bic, DO_ACTION | WRITE_SECTORS); + out_8(&sw->select, RELAX); + out_8(&sw->intr_enable, 0); + del_timer(&fs->timeout); + fs->timeout_pending = 0; + if (sw->ctrack == 0xff) { + printk(KERN_ERR "swim3: seen sector but cyl=ff?\n"); + fs->cur_cyl = -1; + if (fs->retries > 5) { + end_request(fd_req, 0); + fs->state = idle; + start_request(fs); + } else { + fs->state = jogging; + act(fs); + } + break; + } + fs->cur_cyl = sw->ctrack; + fs->cur_sector = sw->csect; + if (fs->expect_cyl != -1 && fs->expect_cyl != fs->cur_cyl) + printk(KERN_ERR "swim3: expected cyl %d, got %d\n", + fs->expect_cyl, fs->cur_cyl); + fs->state = do_transfer; + act(fs); + } + break; + case seeking: + case jogging: + if (sw->nseek == 0) { + out_8(&sw->control_bic, DO_SEEK); + out_8(&sw->select, RELAX); + out_8(&sw->intr_enable, 0); + del_timer(&fs->timeout); + fs->timeout_pending = 0; + if (fs->state == seeking) + ++fs->retries; + fs->state = settling; + act(fs); + } + break; + case settling: + out_8(&sw->intr_enable, 0); + del_timer(&fs->timeout); + fs->timeout_pending = 0; + act(fs); + break; + case do_transfer: + if ((intr & (ERROR_INTR | TRANSFER_DONE)) == 0) + break; + out_8(&sw->intr_enable, 0); + out_8(&sw->control_bic, WRITE_SECTORS | DO_ACTION); + out_8(&sw->select, RELAX); + del_timer(&fs->timeout); + fs->timeout_pending = 0; + dr = fs->dma; + cp = fs->dma_cmd; + if (rq_data_dir(fd_req) == WRITE) + ++cp; + /* + * Check that the main data transfer has finished. + * On writing, the swim3 sometimes doesn't use + * up all the bytes of the postamble, so we can still + * see DMA active here. That doesn't matter as long + * as all the sector data has been transferred. + */ + if ((intr & ERROR_INTR) == 0 && cp->xfer_status == 0) { + /* wait a little while for DMA to complete */ + for (n = 0; n < 100; ++n) { + if (cp->xfer_status != 0) + break; + udelay(1); + barrier(); + } + } + /* turn off DMA */ + out_le32(&dr->control, (RUN | PAUSE) << 16); + stat = ld_le16(&cp->xfer_status); + resid = ld_le16(&cp->res_count); + if (intr & ERROR_INTR) { + n = fs->scount - 1 - resid / 512; + if (n > 0) { + fd_req->sector += n; + fd_req->current_nr_sectors -= n; + fd_req->buffer += n * 512; + fs->req_sector += n; + } + if (fs->retries < 5) { + ++fs->retries; + act(fs); + } else { + printk("swim3: error %sing block %ld (err=%x)\n", + rq_data_dir(fd_req) == WRITE? "writ": "read", + (long)fd_req->sector, err); + end_request(fd_req, 0); + fs->state = idle; + } + } else { + if ((stat & ACTIVE) == 0 || resid != 0) { + /* musta been an error */ + printk(KERN_ERR "swim3: fd dma: stat=%x resid=%d\n", stat, resid); + printk(KERN_ERR " state=%d, dir=%lx, intr=%x, err=%x\n", + fs->state, rq_data_dir(fd_req), intr, err); + end_request(fd_req, 0); + fs->state = idle; + start_request(fs); + break; + } + fd_req->sector += fs->scount; + fd_req->current_nr_sectors -= fs->scount; + fd_req->buffer += fs->scount * 512; + if (fd_req->current_nr_sectors <= 0) { + end_request(fd_req, 1); + fs->state = idle; + } else { + fs->req_sector += fs->scount; + if (fs->req_sector > fs->secpertrack) { + fs->req_sector -= fs->secpertrack; + if (++fs->head > 1) { + fs->head = 0; + ++fs->req_cyl; + } + } + act(fs); + } + } + if (fs->state == idle) + start_request(fs); + break; + default: + printk(KERN_ERR "swim3: don't know what to do in state %d\n", fs->state); + } + return IRQ_HANDLED; +} + +/* +static void fd_dma_interrupt(int irq, void *dev_id, struct pt_regs *regs) +{ +} +*/ + +static int grab_drive(struct floppy_state *fs, enum swim_state state, + int interruptible) +{ + unsigned long flags; + + save_flags(flags); + cli(); + if (fs->state != idle) { + ++fs->wanted; + while (fs->state != available) { + if (interruptible && signal_pending(current)) { + --fs->wanted; + restore_flags(flags); + return -EINTR; + } + interruptible_sleep_on(&fs->wait); + } + --fs->wanted; + } + fs->state = state; + restore_flags(flags); + return 0; +} + +static void release_drive(struct floppy_state *fs) +{ + unsigned long flags; + + save_flags(flags); + cli(); + fs->state = idle; + start_request(fs); + restore_flags(flags); +} + +static int fd_eject(struct floppy_state *fs) +{ + int err, n; + + err = grab_drive(fs, ejecting, 1); + if (err) + return err; + swim3_action(fs, EJECT); + for (n = 20; n > 0; --n) { + if (signal_pending(current)) { + err = -EINTR; + break; + } + swim3_select(fs, RELAX); + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(1); + if (swim3_readbit(fs, DISK_IN) == 0) + break; + } + swim3_select(fs, RELAX); + udelay(150); + fs->ejected = 1; + release_drive(fs); + return err; +} + +static struct floppy_struct floppy_type = + { 2880,18,2,80,0,0x1B,0x00,0xCF,0x6C,NULL }; /* 7 1.44MB 3.5" */ + +static int floppy_ioctl(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long param) +{ + struct floppy_state *fs = inode->i_bdev->bd_disk->private_data; + int err; + + if ((cmd & 0x80) && !capable(CAP_SYS_ADMIN)) + return -EPERM; + + if (fs->media_bay && check_media_bay(fs->media_bay, MB_FD)) + return -ENXIO; + + switch (cmd) { + case FDEJECT: + if (fs->ref_count != 1) + return -EBUSY; + err = fd_eject(fs); + return err; + case FDGETPRM: + if (copy_to_user((void __user *) param, &floppy_type, + sizeof(struct floppy_struct))) + return -EFAULT; + return 0; + } + return -ENOTTY; +} + +static int floppy_open(struct inode *inode, struct file *filp) +{ + struct floppy_state *fs = inode->i_bdev->bd_disk->private_data; + struct swim3 __iomem *sw = fs->swim3; + int n, err = 0; + + if (fs->ref_count == 0) { + if (fs->media_bay && check_media_bay(fs->media_bay, MB_FD)) + return -ENXIO; + out_8(&sw->setup, S_IBM_DRIVE | S_FCLK_DIV2); + out_8(&sw->control_bic, 0xff); + out_8(&sw->mode, 0x95); + udelay(10); + out_8(&sw->intr_enable, 0); + out_8(&sw->control_bis, DRIVE_ENABLE | INTR_ENABLE); + swim3_action(fs, MOTOR_ON); + fs->write_prot = -1; + fs->cur_cyl = -1; + for (n = 0; n < 2 * HZ; ++n) { + if (n >= HZ/30 && swim3_readbit(fs, SEEK_COMPLETE)) + break; + if (signal_pending(current)) { + err = -EINTR; + break; + } + swim3_select(fs, RELAX); + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(1); + } + if (err == 0 && (swim3_readbit(fs, SEEK_COMPLETE) == 0 + || swim3_readbit(fs, DISK_IN) == 0)) + err = -ENXIO; + swim3_action(fs, SETMFM); + swim3_select(fs, RELAX); + + } else if (fs->ref_count == -1 || filp->f_flags & O_EXCL) + return -EBUSY; + + if (err == 0 && (filp->f_flags & O_NDELAY) == 0 + && (filp->f_mode & 3)) { + check_disk_change(inode->i_bdev); + if (fs->ejected) + err = -ENXIO; + } + + if (err == 0 && (filp->f_mode & 2)) { + if (fs->write_prot < 0) + fs->write_prot = swim3_readbit(fs, WRITE_PROT); + if (fs->write_prot) + err = -EROFS; + } + + if (err) { + if (fs->ref_count == 0) { + swim3_action(fs, MOTOR_OFF); + out_8(&sw->control_bic, DRIVE_ENABLE | INTR_ENABLE); + swim3_select(fs, RELAX); + } + return err; + } + + if (filp->f_flags & O_EXCL) + fs->ref_count = -1; + else + ++fs->ref_count; + + return 0; +} + +static int floppy_release(struct inode *inode, struct file *filp) +{ + struct floppy_state *fs = inode->i_bdev->bd_disk->private_data; + struct swim3 __iomem *sw = fs->swim3; + if (fs->ref_count > 0 && --fs->ref_count == 0) { + swim3_action(fs, MOTOR_OFF); + out_8(&sw->control_bic, 0xff); + swim3_select(fs, RELAX); + } + return 0; +} + +static int floppy_check_change(struct gendisk *disk) +{ + struct floppy_state *fs = disk->private_data; + return fs->ejected; +} + +static int floppy_revalidate(struct gendisk *disk) +{ + struct floppy_state *fs = disk->private_data; + struct swim3 __iomem *sw; + int ret, n; + + if (fs->media_bay && check_media_bay(fs->media_bay, MB_FD)) + return -ENXIO; + + sw = fs->swim3; + grab_drive(fs, revalidating, 0); + out_8(&sw->intr_enable, 0); + out_8(&sw->control_bis, DRIVE_ENABLE); + swim3_action(fs, MOTOR_ON); /* necessary? */ + fs->write_prot = -1; + fs->cur_cyl = -1; + mdelay(1); + for (n = HZ; n > 0; --n) { + if (swim3_readbit(fs, SEEK_COMPLETE)) + break; + if (signal_pending(current)) + break; + swim3_select(fs, RELAX); + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(1); + } + ret = swim3_readbit(fs, SEEK_COMPLETE) == 0 + || swim3_readbit(fs, DISK_IN) == 0; + if (ret) + swim3_action(fs, MOTOR_OFF); + else { + fs->ejected = 0; + swim3_action(fs, SETMFM); + } + swim3_select(fs, RELAX); + + release_drive(fs); + return ret; +} + +static struct block_device_operations floppy_fops = { + .open = floppy_open, + .release = floppy_release, + .ioctl = floppy_ioctl, + .media_changed = floppy_check_change, + .revalidate_disk= floppy_revalidate, +}; + +int swim3_init(void) +{ + struct device_node *swim; + int err = -ENOMEM; + int i; + + devfs_mk_dir("floppy"); + + swim = find_devices("floppy"); + while (swim && (floppy_count < MAX_FLOPPIES)) + { + swim3_add_device(swim); + swim = swim->next; + } + + swim = find_devices("swim3"); + while (swim && (floppy_count < MAX_FLOPPIES)) + { + swim3_add_device(swim); + swim = swim->next; + } + + if (!floppy_count) + return -ENODEV; + + for (i = 0; i < floppy_count; i++) { + disks[i] = alloc_disk(1); + if (!disks[i]) + goto out; + } + + if (register_blkdev(FLOPPY_MAJOR, "fd")) { + err = -EBUSY; + goto out; + } + + swim3_queue = blk_init_queue(do_fd_request, &swim3_lock); + if (!swim3_queue) { + err = -ENOMEM; + goto out_queue; + } + + for (i = 0; i < floppy_count; i++) { + struct gendisk *disk = disks[i]; + disk->major = FLOPPY_MAJOR; + disk->first_minor = i; + disk->fops = &floppy_fops; + disk->private_data = &floppy_states[i]; + disk->queue = swim3_queue; + disk->flags |= GENHD_FL_REMOVABLE; + sprintf(disk->disk_name, "fd%d", i); + sprintf(disk->devfs_name, "floppy/%d", i); + set_capacity(disk, 2880); + add_disk(disk); + } + return 0; + +out_queue: + unregister_blkdev(FLOPPY_MAJOR, "fd"); +out: + while (i--) + put_disk(disks[i]); + /* shouldn't we do something with results of swim_add_device()? */ + return err; +} + +static int swim3_add_device(struct device_node *swim) +{ + struct device_node *mediabay; + struct floppy_state *fs = &floppy_states[floppy_count]; + + if (swim->n_addrs < 2) + { + printk(KERN_INFO "swim3: expecting 2 addrs (n_addrs:%d, n_intrs:%d)\n", + swim->n_addrs, swim->n_intrs); + return -EINVAL; + } + + if (swim->n_intrs < 2) + { + printk(KERN_INFO "swim3: expecting 2 intrs (n_addrs:%d, n_intrs:%d)\n", + swim->n_addrs, swim->n_intrs); + return -EINVAL; + } + + if (!request_OF_resource(swim, 0, NULL)) { + printk(KERN_INFO "swim3: can't request IO resource !\n"); + return -EINVAL; + } + + mediabay = (strcasecmp(swim->parent->type, "media-bay") == 0) ? swim->parent : NULL; + if (mediabay == NULL) + pmac_call_feature(PMAC_FTR_SWIM3_ENABLE, swim, 0, 1); + + memset(fs, 0, sizeof(*fs)); + fs->state = idle; + fs->swim3 = (struct swim3 __iomem *) + ioremap(swim->addrs[0].address, 0x200); + fs->dma = (struct dbdma_regs __iomem *) + ioremap(swim->addrs[1].address, 0x200); + fs->swim3_intr = swim->intrs[0].line; + fs->dma_intr = swim->intrs[1].line; + fs->cur_cyl = -1; + fs->cur_sector = -1; + fs->secpercyl = 36; + fs->secpertrack = 18; + fs->total_secs = 2880; + fs->media_bay = mediabay; + init_waitqueue_head(&fs->wait); + + fs->dma_cmd = (struct dbdma_cmd *) DBDMA_ALIGN(fs->dbdma_cmd_space); + memset(fs->dma_cmd, 0, 2 * sizeof(struct dbdma_cmd)); + st_le16(&fs->dma_cmd[1].command, DBDMA_STOP); + + if (request_irq(fs->swim3_intr, swim3_interrupt, 0, "SWIM3", fs)) { + printk(KERN_ERR "Couldn't get irq %d for SWIM3\n", fs->swim3_intr); + pmac_call_feature(PMAC_FTR_SWIM3_ENABLE, swim, 0, 0); + return -EBUSY; + } +/* + if (request_irq(fs->dma_intr, fd_dma_interrupt, 0, "SWIM3-dma", fs)) { + printk(KERN_ERR "Couldn't get irq %d for SWIM3 DMA", + fs->dma_intr); + pmac_call_feature(PMAC_FTR_SWIM3_ENABLE, swim, 0, 0); + return -EBUSY; + } +*/ + + init_timer(&fs->timeout); + + printk(KERN_INFO "fd%d: SWIM3 floppy controller %s\n", floppy_count, + mediabay ? "in media bay" : ""); + + floppy_count++; + + return 0; +} + +module_init(swim3_init) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Paul Mackerras"); +MODULE_ALIAS_BLOCKDEV_MAJOR(FLOPPY_MAJOR); diff --git a/drivers/block/swim_iop.c b/drivers/block/swim_iop.c new file mode 100644 index 00000000000..a1283f6dc01 --- /dev/null +++ b/drivers/block/swim_iop.c @@ -0,0 +1,579 @@ +/* + * Driver for the SWIM (Super Woz Integrated Machine) IOP + * floppy controller on the Macintosh IIfx and Quadra 900/950 + * + * Written by Joshua M. Thompson (funaho@jurai.org) + * based on the SWIM3 driver (c) 1996 by Paul Mackerras. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + * + * 1999-06-12 (jmt) - Initial implementation. + */ + +/* + * ------------------- + * Theory of Operation + * ------------------- + * + * Since the SWIM IOP is message-driven we implement a simple request queue + * system. One outstanding request may be queued at any given time (this is + * an IOP limitation); only when that request has completed can a new request + * be sent. + */ + +#include <linux/stddef.h> +#include <linux/kernel.h> +#include <linux/sched.h> +#include <linux/timer.h> +#include <linux/delay.h> +#include <linux/fd.h> +#include <linux/ioctl.h> +#include <linux/blkdev.h> +#include <asm/io.h> +#include <asm/uaccess.h> +#include <asm/mac_iop.h> +#include <asm/swim_iop.h> + +#define DRIVER_VERSION "Version 0.1 (1999-06-12)" + +#define MAX_FLOPPIES 4 + +enum swim_state { + idle, + available, + revalidating, + transferring, + ejecting +}; + +struct floppy_state { + enum swim_state state; + int drive_num; /* device number */ + int secpercyl; /* disk geometry information */ + int secpertrack; + int total_secs; + int write_prot; /* 1 if write-protected, 0 if not, -1 dunno */ + int ref_count; + struct timer_list timeout; + int ejected; + struct wait_queue *wait; + int wanted; + int timeout_pending; +}; + +struct swim_iop_req { + int sent; + int complete; + __u8 command[32]; + struct floppy_state *fs; + void (*done)(struct swim_iop_req *); +}; + +static struct swim_iop_req *current_req; +static int floppy_count; + +static struct floppy_state floppy_states[MAX_FLOPPIES]; +static DEFINE_SPINLOCK(swim_iop_lock); + +#define CURRENT elv_next_request(swim_queue) + +static char *drive_names[7] = { + "not installed", /* DRV_NONE */ + "unknown (1)", /* DRV_UNKNOWN */ + "a 400K drive", /* DRV_400K */ + "an 800K drive" /* DRV_800K */ + "unknown (4)", /* ???? */ + "an FDHD", /* DRV_FDHD */ + "unknown (6)", /* ???? */ + "an Apple HD20" /* DRV_HD20 */ +}; + +int swimiop_init(void); +static void swimiop_init_request(struct swim_iop_req *); +static int swimiop_send_request(struct swim_iop_req *); +static void swimiop_receive(struct iop_msg *, struct pt_regs *); +static void swimiop_status_update(int, struct swim_drvstatus *); +static int swimiop_eject(struct floppy_state *fs); + +static int floppy_ioctl(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long param); +static int floppy_open(struct inode *inode, struct file *filp); +static int floppy_release(struct inode *inode, struct file *filp); +static int floppy_check_change(struct gendisk *disk); +static int floppy_revalidate(struct gendisk *disk); +static int grab_drive(struct floppy_state *fs, enum swim_state state, + int interruptible); +static void release_drive(struct floppy_state *fs); +static void set_timeout(struct floppy_state *fs, int nticks, + void (*proc)(unsigned long)); +static void fd_request_timeout(unsigned long); +static void do_fd_request(request_queue_t * q); +static void start_request(struct floppy_state *fs); + +static struct block_device_operations floppy_fops = { + .open = floppy_open, + .release = floppy_release, + .ioctl = floppy_ioctl, + .media_changed = floppy_check_change, + .revalidate_disk= floppy_revalidate, +}; + +static struct request_queue *swim_queue; +/* + * SWIM IOP initialization + */ + +int swimiop_init(void) +{ + volatile struct swim_iop_req req; + struct swimcmd_status *cmd = (struct swimcmd_status *) &req.command[0]; + struct swim_drvstatus *ds = &cmd->status; + struct floppy_state *fs; + int i; + + current_req = NULL; + floppy_count = 0; + + if (!iop_ism_present) + return -ENODEV; + + if (register_blkdev(FLOPPY_MAJOR, "fd")) + return -EBUSY; + + swim_queue = blk_init_queue(do_fd_request, &swim_iop_lock); + if (!swim_queue) { + unregister_blkdev(FLOPPY_MAJOR, "fd"); + return -ENOMEM; + } + + printk("SWIM-IOP: %s by Joshua M. Thompson (funaho@jurai.org)\n", + DRIVER_VERSION); + + if (iop_listen(SWIM_IOP, SWIM_CHAN, swimiop_receive, "SWIM") != 0) { + printk(KERN_ERR "SWIM-IOP: IOP channel already in use; can't initialize.\n"); + unregister_blkdev(FLOPPY_MAJOR, "fd"); + blk_cleanup_queue(swim_queue); + return -EBUSY; + } + + printk(KERN_ERR "SWIM_IOP: probing for installed drives.\n"); + + for (i = 0 ; i < MAX_FLOPPIES ; i++) { + memset(&floppy_states[i], 0, sizeof(struct floppy_state)); + fs = &floppy_states[floppy_count]; + + swimiop_init_request(&req); + cmd->code = CMD_STATUS; + cmd->drive_num = i + 1; + if (swimiop_send_request(&req) != 0) continue; + while (!req.complete); + if (cmd->error != 0) { + printk(KERN_ERR "SWIM-IOP: probe on drive %d returned error %d\n", i, (uint) cmd->error); + continue; + } + if (ds->installed != 0x01) continue; + printk("SWIM-IOP: drive %d is %s (%s, %s, %s, %s)\n", i, + drive_names[ds->info.type], + ds->info.external? "ext" : "int", + ds->info.scsi? "scsi" : "floppy", + ds->info.fixed? "fixed" : "removable", + ds->info.secondary? "secondary" : "primary"); + swimiop_status_update(floppy_count, ds); + fs->state = idle; + + init_timer(&fs->timeout); + floppy_count++; + } + printk("SWIM-IOP: detected %d installed drives.\n", floppy_count); + + for (i = 0; i < floppy_count; i++) { + struct gendisk *disk = alloc_disk(1); + if (!disk) + continue; + disk->major = FLOPPY_MAJOR; + disk->first_minor = i; + disk->fops = &floppy_fops; + sprintf(disk->disk_name, "fd%d", i); + disk->private_data = &floppy_states[i]; + disk->queue = swim_queue; + set_capacity(disk, 2880 * 2); + add_disk(disk); + } + + return 0; +} + +static void swimiop_init_request(struct swim_iop_req *req) +{ + req->sent = 0; + req->complete = 0; + req->done = NULL; +} + +static int swimiop_send_request(struct swim_iop_req *req) +{ + unsigned long flags; + int err; + + /* It's doubtful an interrupt routine would try to send */ + /* a SWIM request, but I'd rather play it safe here. */ + + local_irq_save(flags); + + if (current_req != NULL) { + local_irq_restore(flags); + return -ENOMEM; + } + + current_req = req; + + /* Interrupts should be back on for iop_send_message() */ + + local_irq_restore(flags); + + err = iop_send_message(SWIM_IOP, SWIM_CHAN, (void *) req, + sizeof(req->command), (__u8 *) &req->command[0], + swimiop_receive); + + /* No race condition here; we own current_req at this point */ + + if (err) { + current_req = NULL; + } else { + req->sent = 1; + } + return err; +} + +/* + * Receive a SWIM message from the IOP. + * + * This will be called in two cases: + * + * 1. A message has been successfully sent to the IOP. + * 2. An unsolicited message was received from the IOP. + */ + +void swimiop_receive(struct iop_msg *msg, struct pt_regs *regs) +{ + struct swim_iop_req *req; + struct swimmsg_status *sm; + struct swim_drvstatus *ds; + + req = current_req; + + switch(msg->status) { + case IOP_MSGSTATUS_COMPLETE: + memcpy(&req->command[0], &msg->reply[0], sizeof(req->command)); + req->complete = 1; + if (req->done) (*req->done)(req); + current_req = NULL; + break; + case IOP_MSGSTATUS_UNSOL: + sm = (struct swimmsg_status *) &msg->message[0]; + ds = &sm->status; + swimiop_status_update(sm->drive_num, ds); + iop_complete_message(msg); + break; + } +} + +static void swimiop_status_update(int drive_num, struct swim_drvstatus *ds) +{ + struct floppy_state *fs = &floppy_states[drive_num]; + + fs->write_prot = (ds->write_prot == 0x80); + if ((ds->disk_in_drive != 0x01) && (ds->disk_in_drive != 0x02)) { + fs->ejected = 1; + } else { + fs->ejected = 0; + } + switch(ds->info.type) { + case DRV_400K: + fs->secpercyl = 10; + fs->secpertrack = 10; + fs->total_secs = 800; + break; + case DRV_800K: + fs->secpercyl = 20; + fs->secpertrack = 10; + fs->total_secs = 1600; + break; + case DRV_FDHD: + fs->secpercyl = 36; + fs->secpertrack = 18; + fs->total_secs = 2880; + break; + default: + fs->secpercyl = 0; + fs->secpertrack = 0; + fs->total_secs = 0; + break; + } +} + +static int swimiop_eject(struct floppy_state *fs) +{ + int err, n; + struct swim_iop_req req; + struct swimcmd_eject *cmd = (struct swimcmd_eject *) &req.command[0]; + + err = grab_drive(fs, ejecting, 1); + if (err) return err; + + swimiop_init_request(&req); + cmd->code = CMD_EJECT; + cmd->drive_num = fs->drive_num; + err = swimiop_send_request(&req); + if (err) { + release_drive(fs); + return err; + } + for (n = 2*HZ; n > 0; --n) { + if (req.complete) break; + if (signal_pending(current)) { + err = -EINTR; + break; + } + current->state = TASK_INTERRUPTIBLE; + schedule_timeout(1); + } + release_drive(fs); + return cmd->error; +} + +static struct floppy_struct floppy_type = + { 2880,18,2,80,0,0x1B,0x00,0xCF,0x6C,NULL }; /* 7 1.44MB 3.5" */ + +static int floppy_ioctl(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long param) +{ + struct floppy_state *fs = inode->i_bdev->bd_disk->private_data; + int err; + + if ((cmd & 0x80) && !capable(CAP_SYS_ADMIN)) + return -EPERM; + + switch (cmd) { + case FDEJECT: + if (fs->ref_count != 1) + return -EBUSY; + err = swimiop_eject(fs); + return err; + case FDGETPRM: + if (copy_to_user((void *) param, (void *) &floppy_type, + sizeof(struct floppy_struct))) + return -EFAULT; + return 0; + } + return -ENOTTY; +} + +static int floppy_open(struct inode *inode, struct file *filp) +{ + struct floppy_state *fs = inode->i_bdev->bd_disk->private_data; + + if (fs->ref_count == -1 || filp->f_flags & O_EXCL) + return -EBUSY; + + if ((filp->f_flags & O_NDELAY) == 0 && (filp->f_mode & 3)) { + check_disk_change(inode->i_bdev); + if (fs->ejected) + return -ENXIO; + } + + if ((filp->f_mode & 2) && fs->write_prot) + return -EROFS; + + if (filp->f_flags & O_EXCL) + fs->ref_count = -1; + else + ++fs->ref_count; + + return 0; +} + +static int floppy_release(struct inode *inode, struct file *filp) +{ + struct floppy_state *fs = inode->i_bdev->bd_disk->private_data; + if (fs->ref_count > 0) + fs->ref_count--; + return 0; +} + +static int floppy_check_change(struct gendisk *disk) +{ + struct floppy_state *fs = disk->private_data; + return fs->ejected; +} + +static int floppy_revalidate(struct gendisk *disk) +{ + struct floppy_state *fs = disk->private_data; + grab_drive(fs, revalidating, 0); + /* yadda, yadda */ + release_drive(fs); + return 0; +} + +static void floppy_off(unsigned int nr) +{ +} + +static int grab_drive(struct floppy_state *fs, enum swim_state state, + int interruptible) +{ + unsigned long flags; + + local_irq_save(flags); + if (fs->state != idle) { + ++fs->wanted; + while (fs->state != available) { + if (interruptible && signal_pending(current)) { + --fs->wanted; + local_irq_restore(flags); + return -EINTR; + } + interruptible_sleep_on(&fs->wait); + } + --fs->wanted; + } + fs->state = state; + local_irq_restore(flags); + return 0; +} + +static void release_drive(struct floppy_state *fs) +{ + unsigned long flags; + + local_irq_save(flags); + fs->state = idle; + start_request(fs); + local_irq_restore(flags); +} + +static void set_timeout(struct floppy_state *fs, int nticks, + void (*proc)(unsigned long)) +{ + unsigned long flags; + + local_irq_save(flags); + if (fs->timeout_pending) + del_timer(&fs->timeout); + init_timer(&fs->timeout); + fs->timeout.expires = jiffies + nticks; + fs->timeout.function = proc; + fs->timeout.data = (unsigned long) fs; + add_timer(&fs->timeout); + fs->timeout_pending = 1; + local_irq_restore(flags); +} + +static void do_fd_request(request_queue_t * q) +{ + int i; + + for (i = 0 ; i < floppy_count ; i++) { + start_request(&floppy_states[i]); + } +} + +static void fd_request_complete(struct swim_iop_req *req) +{ + struct floppy_state *fs = req->fs; + struct swimcmd_rw *cmd = (struct swimcmd_rw *) &req->command[0]; + + del_timer(&fs->timeout); + fs->timeout_pending = 0; + fs->state = idle; + if (cmd->error) { + printk(KERN_ERR "SWIM-IOP: error %d on read/write request.\n", cmd->error); + end_request(CURRENT, 0); + } else { + CURRENT->sector += cmd->num_blocks; + CURRENT->current_nr_sectors -= cmd->num_blocks; + if (CURRENT->current_nr_sectors <= 0) { + end_request(CURRENT, 1); + return; + } + } + start_request(fs); +} + +static void fd_request_timeout(unsigned long data) +{ + struct floppy_state *fs = (struct floppy_state *) data; + + fs->timeout_pending = 0; + end_request(CURRENT, 0); + fs->state = idle; +} + +static void start_request(struct floppy_state *fs) +{ + volatile struct swim_iop_req req; + struct swimcmd_rw *cmd = (struct swimcmd_rw *) &req.command[0]; + + if (fs->state == idle && fs->wanted) { + fs->state = available; + wake_up(&fs->wait); + return; + } + while (CURRENT && fs->state == idle) { + if (CURRENT->bh && !buffer_locked(CURRENT->bh)) + panic("floppy: block not locked"); +#if 0 + printk("do_fd_req: dev=%s cmd=%d sec=%ld nr_sec=%ld buf=%p\n", + CURRENT->rq_disk->disk_name, CURRENT->cmd, + CURRENT->sector, CURRENT->nr_sectors, CURRENT->buffer); + printk(" rq_status=%d errors=%d current_nr_sectors=%ld\n", + CURRENT->rq_status, CURRENT->errors, CURRENT->current_nr_sectors); +#endif + + if (CURRENT->sector < 0 || CURRENT->sector >= fs->total_secs) { + end_request(CURRENT, 0); + continue; + } + if (CURRENT->current_nr_sectors == 0) { + end_request(CURRENT, 1); + continue; + } + if (fs->ejected) { + end_request(CURRENT, 0); + continue; + } + + swimiop_init_request(&req); + req.fs = fs; + req.done = fd_request_complete; + + if (CURRENT->cmd == WRITE) { + if (fs->write_prot) { + end_request(CURRENT, 0); + continue; + } + cmd->code = CMD_WRITE; + } else { + cmd->code = CMD_READ; + + } + cmd->drive_num = fs->drive_num; + cmd->buffer = CURRENT->buffer; + cmd->first_block = CURRENT->sector; + cmd->num_blocks = CURRENT->current_nr_sectors; + + if (swimiop_send_request(&req)) { + end_request(CURRENT, 0); + continue; + } + + set_timeout(fs, HZ*CURRENT->current_nr_sectors, + fd_request_timeout); + + fs->state = transferring; + } +} diff --git a/drivers/block/sx8.c b/drivers/block/sx8.c new file mode 100644 index 00000000000..797f5988c2b --- /dev/null +++ b/drivers/block/sx8.c @@ -0,0 +1,1764 @@ +/* + * sx8.c: Driver for Promise SATA SX8 looks-like-I2O hardware + * + * Copyright 2004 Red Hat, Inc. + * + * Author/maintainer: Jeff Garzik <jgarzik@pobox.com> + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/pci.h> +#include <linux/slab.h> +#include <linux/spinlock.h> +#include <linux/blkdev.h> +#include <linux/sched.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/interrupt.h> +#include <linux/compiler.h> +#include <linux/workqueue.h> +#include <linux/bitops.h> +#include <linux/delay.h> +#include <linux/time.h> +#include <linux/hdreg.h> +#include <asm/io.h> +#include <asm/semaphore.h> +#include <asm/uaccess.h> + +MODULE_AUTHOR("Jeff Garzik"); +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("Promise SATA SX8 block driver"); + +#if 0 +#define CARM_DEBUG +#define CARM_VERBOSE_DEBUG +#else +#undef CARM_DEBUG +#undef CARM_VERBOSE_DEBUG +#endif +#undef CARM_NDEBUG + +#define DRV_NAME "sx8" +#define DRV_VERSION "0.8" +#define PFX DRV_NAME ": " + +#define NEXT_RESP(idx) ((idx + 1) % RMSG_Q_LEN) + +/* 0xf is just arbitrary, non-zero noise; this is sorta like poisoning */ +#define TAG_ENCODE(tag) (((tag) << 16) | 0xf) +#define TAG_DECODE(tag) (((tag) >> 16) & 0x1f) +#define TAG_VALID(tag) ((((tag) & 0xf) == 0xf) && (TAG_DECODE(tag) < 32)) + +/* note: prints function name for you */ +#ifdef CARM_DEBUG +#define DPRINTK(fmt, args...) printk(KERN_ERR "%s: " fmt, __FUNCTION__, ## args) +#ifdef CARM_VERBOSE_DEBUG +#define VPRINTK(fmt, args...) printk(KERN_ERR "%s: " fmt, __FUNCTION__, ## args) +#else +#define VPRINTK(fmt, args...) +#endif /* CARM_VERBOSE_DEBUG */ +#else +#define DPRINTK(fmt, args...) +#define VPRINTK(fmt, args...) +#endif /* CARM_DEBUG */ + +#ifdef CARM_NDEBUG +#define assert(expr) +#else +#define assert(expr) \ + if(unlikely(!(expr))) { \ + printk(KERN_ERR "Assertion failed! %s,%s,%s,line=%d\n", \ + #expr,__FILE__,__FUNCTION__,__LINE__); \ + } +#endif + +/* defines only for the constants which don't work well as enums */ +struct carm_host; + +enum { + /* adapter-wide limits */ + CARM_MAX_PORTS = 8, + CARM_SHM_SIZE = (4096 << 7), + CARM_MINORS_PER_MAJOR = 256 / CARM_MAX_PORTS, + CARM_MAX_WAIT_Q = CARM_MAX_PORTS + 1, + + /* command message queue limits */ + CARM_MAX_REQ = 64, /* max command msgs per host */ + CARM_MAX_Q = 1, /* one command at a time */ + CARM_MSG_LOW_WATER = (CARM_MAX_REQ / 4), /* refill mark */ + + /* S/G limits, host-wide and per-request */ + CARM_MAX_REQ_SG = 32, /* max s/g entries per request */ + CARM_SG_BOUNDARY = 0xffffUL, /* s/g segment boundary */ + CARM_MAX_HOST_SG = 600, /* max s/g entries per host */ + CARM_SG_LOW_WATER = (CARM_MAX_HOST_SG / 4), /* re-fill mark */ + + /* hardware registers */ + CARM_IHQP = 0x1c, + CARM_INT_STAT = 0x10, /* interrupt status */ + CARM_INT_MASK = 0x14, /* interrupt mask */ + CARM_HMUC = 0x18, /* host message unit control */ + RBUF_ADDR_LO = 0x20, /* response msg DMA buf low 32 bits */ + RBUF_ADDR_HI = 0x24, /* response msg DMA buf high 32 bits */ + RBUF_BYTE_SZ = 0x28, + CARM_RESP_IDX = 0x2c, + CARM_CMS0 = 0x30, /* command message size reg 0 */ + CARM_LMUC = 0x48, + CARM_HMPHA = 0x6c, + CARM_INITC = 0xb5, + + /* bits in CARM_INT_{STAT,MASK} */ + INT_RESERVED = 0xfffffff0, + INT_WATCHDOG = (1 << 3), /* watchdog timer */ + INT_Q_OVERFLOW = (1 << 2), /* cmd msg q overflow */ + INT_Q_AVAILABLE = (1 << 1), /* cmd msg q has free space */ + INT_RESPONSE = (1 << 0), /* response msg available */ + INT_ACK_MASK = INT_WATCHDOG | INT_Q_OVERFLOW, + INT_DEF_MASK = INT_RESERVED | INT_Q_OVERFLOW | + INT_RESPONSE, + + /* command messages, and related register bits */ + CARM_HAVE_RESP = 0x01, + CARM_MSG_READ = 1, + CARM_MSG_WRITE = 2, + CARM_MSG_VERIFY = 3, + CARM_MSG_GET_CAPACITY = 4, + CARM_MSG_FLUSH = 5, + CARM_MSG_IOCTL = 6, + CARM_MSG_ARRAY = 8, + CARM_MSG_MISC = 9, + CARM_CME = (1 << 2), + CARM_RME = (1 << 1), + CARM_WZBC = (1 << 0), + CARM_RMI = (1 << 0), + CARM_Q_FULL = (1 << 3), + CARM_MSG_SIZE = 288, + CARM_Q_LEN = 48, + + /* CARM_MSG_IOCTL messages */ + CARM_IOC_SCAN_CHAN = 5, /* scan channels for devices */ + CARM_IOC_GET_TCQ = 13, /* get tcq/ncq depth */ + CARM_IOC_SET_TCQ = 14, /* set tcq/ncq depth */ + + IOC_SCAN_CHAN_NODEV = 0x1f, + IOC_SCAN_CHAN_OFFSET = 0x40, + + /* CARM_MSG_ARRAY messages */ + CARM_ARRAY_INFO = 0, + + ARRAY_NO_EXIST = (1 << 31), + + /* response messages */ + RMSG_SZ = 8, /* sizeof(struct carm_response) */ + RMSG_Q_LEN = 48, /* resp. msg list length */ + RMSG_OK = 1, /* bit indicating msg was successful */ + /* length of entire resp. msg buffer */ + RBUF_LEN = RMSG_SZ * RMSG_Q_LEN, + + PDC_SHM_SIZE = (4096 << 7), /* length of entire h/w buffer */ + + /* CARM_MSG_MISC messages */ + MISC_GET_FW_VER = 2, + MISC_ALLOC_MEM = 3, + MISC_SET_TIME = 5, + + /* MISC_GET_FW_VER feature bits */ + FW_VER_4PORT = (1 << 2), /* 1=4 ports, 0=8 ports */ + FW_VER_NON_RAID = (1 << 1), /* 1=non-RAID firmware, 0=RAID */ + FW_VER_ZCR = (1 << 0), /* zero channel RAID (whatever that is) */ + + /* carm_host flags */ + FL_NON_RAID = FW_VER_NON_RAID, + FL_4PORT = FW_VER_4PORT, + FL_FW_VER_MASK = (FW_VER_NON_RAID | FW_VER_4PORT), + FL_DAC = (1 << 16), + FL_DYN_MAJOR = (1 << 17), +}; + +enum scatter_gather_types { + SGT_32BIT = 0, + SGT_64BIT = 1, +}; + +enum host_states { + HST_INVALID, /* invalid state; never used */ + HST_ALLOC_BUF, /* setting up master SHM area */ + HST_ERROR, /* we never leave here */ + HST_PORT_SCAN, /* start dev scan */ + HST_DEV_SCAN_START, /* start per-device probe */ + HST_DEV_SCAN, /* continue per-device probe */ + HST_DEV_ACTIVATE, /* activate devices we found */ + HST_PROBE_FINISHED, /* probe is complete */ + HST_PROBE_START, /* initiate probe */ + HST_SYNC_TIME, /* tell firmware what time it is */ + HST_GET_FW_VER, /* get firmware version, adapter port cnt */ +}; + +#ifdef CARM_DEBUG +static const char *state_name[] = { + "HST_INVALID", + "HST_ALLOC_BUF", + "HST_ERROR", + "HST_PORT_SCAN", + "HST_DEV_SCAN_START", + "HST_DEV_SCAN", + "HST_DEV_ACTIVATE", + "HST_PROBE_FINISHED", + "HST_PROBE_START", + "HST_SYNC_TIME", + "HST_GET_FW_VER", +}; +#endif + +struct carm_port { + unsigned int port_no; + unsigned int n_queued; + struct gendisk *disk; + struct carm_host *host; + + /* attached device characteristics */ + u64 capacity; + char name[41]; + u16 dev_geom_head; + u16 dev_geom_sect; + u16 dev_geom_cyl; +}; + +struct carm_request { + unsigned int tag; + int n_elem; + unsigned int msg_type; + unsigned int msg_subtype; + unsigned int msg_bucket; + struct request *rq; + struct carm_port *port; + struct scatterlist sg[CARM_MAX_REQ_SG]; +}; + +struct carm_host { + unsigned long flags; + void __iomem *mmio; + void *shm; + dma_addr_t shm_dma; + + int major; + int id; + char name[32]; + + spinlock_t lock; + struct pci_dev *pdev; + unsigned int state; + u32 fw_ver; + + request_queue_t *oob_q; + unsigned int n_oob; + + unsigned int hw_sg_used; + + unsigned int resp_idx; + + unsigned int wait_q_prod; + unsigned int wait_q_cons; + request_queue_t *wait_q[CARM_MAX_WAIT_Q]; + + unsigned int n_msgs; + u64 msg_alloc; + struct carm_request req[CARM_MAX_REQ]; + void *msg_base; + dma_addr_t msg_dma; + + int cur_scan_dev; + unsigned long dev_active; + unsigned long dev_present; + struct carm_port port[CARM_MAX_PORTS]; + + struct work_struct fsm_task; + + struct semaphore probe_sem; +}; + +struct carm_response { + __le32 ret_handle; + __le32 status; +} __attribute__((packed)); + +struct carm_msg_sg { + __le32 start; + __le32 len; +} __attribute__((packed)); + +struct carm_msg_rw { + u8 type; + u8 id; + u8 sg_count; + u8 sg_type; + __le32 handle; + __le32 lba; + __le16 lba_count; + __le16 lba_high; + struct carm_msg_sg sg[32]; +} __attribute__((packed)); + +struct carm_msg_allocbuf { + u8 type; + u8 subtype; + u8 n_sg; + u8 sg_type; + __le32 handle; + __le32 addr; + __le32 len; + __le32 evt_pool; + __le32 n_evt; + __le32 rbuf_pool; + __le32 n_rbuf; + __le32 msg_pool; + __le32 n_msg; + struct carm_msg_sg sg[8]; +} __attribute__((packed)); + +struct carm_msg_ioctl { + u8 type; + u8 subtype; + u8 array_id; + u8 reserved1; + __le32 handle; + __le32 data_addr; + u32 reserved2; +} __attribute__((packed)); + +struct carm_msg_sync_time { + u8 type; + u8 subtype; + u16 reserved1; + __le32 handle; + u32 reserved2; + __le32 timestamp; +} __attribute__((packed)); + +struct carm_msg_get_fw_ver { + u8 type; + u8 subtype; + u16 reserved1; + __le32 handle; + __le32 data_addr; + u32 reserved2; +} __attribute__((packed)); + +struct carm_fw_ver { + __le32 version; + u8 features; + u8 reserved1; + u16 reserved2; +} __attribute__((packed)); + +struct carm_array_info { + __le32 size; + + __le16 size_hi; + __le16 stripe_size; + + __le32 mode; + + __le16 stripe_blk_sz; + __le16 reserved1; + + __le16 cyl; + __le16 head; + + __le16 sect; + u8 array_id; + u8 reserved2; + + char name[40]; + + __le32 array_status; + + /* device list continues beyond this point? */ +} __attribute__((packed)); + +static int carm_init_one (struct pci_dev *pdev, const struct pci_device_id *ent); +static void carm_remove_one (struct pci_dev *pdev); +static int carm_bdev_ioctl(struct inode *ino, struct file *fil, + unsigned int cmd, unsigned long arg); + +static struct pci_device_id carm_pci_tbl[] = { + { PCI_VENDOR_ID_PROMISE, 0x8000, PCI_ANY_ID, PCI_ANY_ID, 0, 0, }, + { PCI_VENDOR_ID_PROMISE, 0x8002, PCI_ANY_ID, PCI_ANY_ID, 0, 0, }, + { } /* terminate list */ +}; +MODULE_DEVICE_TABLE(pci, carm_pci_tbl); + +static struct pci_driver carm_driver = { + .name = DRV_NAME, + .id_table = carm_pci_tbl, + .probe = carm_init_one, + .remove = carm_remove_one, +}; + +static struct block_device_operations carm_bd_ops = { + .owner = THIS_MODULE, + .ioctl = carm_bdev_ioctl, +}; + +static unsigned int carm_host_id; +static unsigned long carm_major_alloc; + + + +static int carm_bdev_ioctl(struct inode *ino, struct file *fil, + unsigned int cmd, unsigned long arg) +{ + void __user *usermem = (void __user *) arg; + struct carm_port *port = ino->i_bdev->bd_disk->private_data; + struct hd_geometry geom; + + switch (cmd) { + case HDIO_GETGEO: + if (!usermem) + return -EINVAL; + + geom.heads = (u8) port->dev_geom_head; + geom.sectors = (u8) port->dev_geom_sect; + geom.cylinders = port->dev_geom_cyl; + geom.start = get_start_sect(ino->i_bdev); + + if (copy_to_user(usermem, &geom, sizeof(geom))) + return -EFAULT; + return 0; + + default: + break; + } + + return -EOPNOTSUPP; +} + +static const u32 msg_sizes[] = { 32, 64, 128, CARM_MSG_SIZE }; + +static inline int carm_lookup_bucket(u32 msg_size) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(msg_sizes); i++) + if (msg_size <= msg_sizes[i]) + return i; + + return -ENOENT; +} + +static void carm_init_buckets(void __iomem *mmio) +{ + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(msg_sizes); i++) + writel(msg_sizes[i], mmio + CARM_CMS0 + (4 * i)); +} + +static inline void *carm_ref_msg(struct carm_host *host, + unsigned int msg_idx) +{ + return host->msg_base + (msg_idx * CARM_MSG_SIZE); +} + +static inline dma_addr_t carm_ref_msg_dma(struct carm_host *host, + unsigned int msg_idx) +{ + return host->msg_dma + (msg_idx * CARM_MSG_SIZE); +} + +static int carm_send_msg(struct carm_host *host, + struct carm_request *crq) +{ + void __iomem *mmio = host->mmio; + u32 msg = (u32) carm_ref_msg_dma(host, crq->tag); + u32 cm_bucket = crq->msg_bucket; + u32 tmp; + int rc = 0; + + VPRINTK("ENTER\n"); + + tmp = readl(mmio + CARM_HMUC); + if (tmp & CARM_Q_FULL) { +#if 0 + tmp = readl(mmio + CARM_INT_MASK); + tmp |= INT_Q_AVAILABLE; + writel(tmp, mmio + CARM_INT_MASK); + readl(mmio + CARM_INT_MASK); /* flush */ +#endif + DPRINTK("host msg queue full\n"); + rc = -EBUSY; + } else { + writel(msg | (cm_bucket << 1), mmio + CARM_IHQP); + readl(mmio + CARM_IHQP); /* flush */ + } + + return rc; +} + +static struct carm_request *carm_get_request(struct carm_host *host) +{ + unsigned int i; + + /* obey global hardware limit on S/G entries */ + if (host->hw_sg_used >= (CARM_MAX_HOST_SG - CARM_MAX_REQ_SG)) + return NULL; + + for (i = 0; i < CARM_MAX_Q; i++) + if ((host->msg_alloc & (1ULL << i)) == 0) { + struct carm_request *crq = &host->req[i]; + crq->port = NULL; + crq->n_elem = 0; + + host->msg_alloc |= (1ULL << i); + host->n_msgs++; + + assert(host->n_msgs <= CARM_MAX_REQ); + return crq; + } + + DPRINTK("no request available, returning NULL\n"); + return NULL; +} + +static int carm_put_request(struct carm_host *host, struct carm_request *crq) +{ + assert(crq->tag < CARM_MAX_Q); + + if (unlikely((host->msg_alloc & (1ULL << crq->tag)) == 0)) + return -EINVAL; /* tried to clear a tag that was not active */ + + assert(host->hw_sg_used >= crq->n_elem); + + host->msg_alloc &= ~(1ULL << crq->tag); + host->hw_sg_used -= crq->n_elem; + host->n_msgs--; + + return 0; +} + +static struct carm_request *carm_get_special(struct carm_host *host) +{ + unsigned long flags; + struct carm_request *crq = NULL; + struct request *rq; + int tries = 5000; + + while (tries-- > 0) { + spin_lock_irqsave(&host->lock, flags); + crq = carm_get_request(host); + spin_unlock_irqrestore(&host->lock, flags); + + if (crq) + break; + msleep(10); + } + + if (!crq) + return NULL; + + rq = blk_get_request(host->oob_q, WRITE /* bogus */, GFP_KERNEL); + if (!rq) { + spin_lock_irqsave(&host->lock, flags); + carm_put_request(host, crq); + spin_unlock_irqrestore(&host->lock, flags); + return NULL; + } + + crq->rq = rq; + return crq; +} + +static int carm_array_info (struct carm_host *host, unsigned int array_idx) +{ + struct carm_msg_ioctl *ioc; + unsigned int idx; + u32 msg_data; + dma_addr_t msg_dma; + struct carm_request *crq; + int rc; + + crq = carm_get_special(host); + if (!crq) { + rc = -ENOMEM; + goto err_out; + } + + idx = crq->tag; + + ioc = carm_ref_msg(host, idx); + msg_dma = carm_ref_msg_dma(host, idx); + msg_data = (u32) (msg_dma + sizeof(struct carm_array_info)); + + crq->msg_type = CARM_MSG_ARRAY; + crq->msg_subtype = CARM_ARRAY_INFO; + rc = carm_lookup_bucket(sizeof(struct carm_msg_ioctl) + + sizeof(struct carm_array_info)); + BUG_ON(rc < 0); + crq->msg_bucket = (u32) rc; + + memset(ioc, 0, sizeof(*ioc)); + ioc->type = CARM_MSG_ARRAY; + ioc->subtype = CARM_ARRAY_INFO; + ioc->array_id = (u8) array_idx; + ioc->handle = cpu_to_le32(TAG_ENCODE(idx)); + ioc->data_addr = cpu_to_le32(msg_data); + + spin_lock_irq(&host->lock); + assert(host->state == HST_DEV_SCAN_START || + host->state == HST_DEV_SCAN); + spin_unlock_irq(&host->lock); + + DPRINTK("blk_insert_request, tag == %u\n", idx); + blk_insert_request(host->oob_q, crq->rq, 1, crq, 0); + + return 0; + +err_out: + spin_lock_irq(&host->lock); + host->state = HST_ERROR; + spin_unlock_irq(&host->lock); + return rc; +} + +typedef unsigned int (*carm_sspc_t)(struct carm_host *, unsigned int, void *); + +static int carm_send_special (struct carm_host *host, carm_sspc_t func) +{ + struct carm_request *crq; + struct carm_msg_ioctl *ioc; + void *mem; + unsigned int idx, msg_size; + int rc; + + crq = carm_get_special(host); + if (!crq) + return -ENOMEM; + + idx = crq->tag; + + mem = carm_ref_msg(host, idx); + + msg_size = func(host, idx, mem); + + ioc = mem; + crq->msg_type = ioc->type; + crq->msg_subtype = ioc->subtype; + rc = carm_lookup_bucket(msg_size); + BUG_ON(rc < 0); + crq->msg_bucket = (u32) rc; + + DPRINTK("blk_insert_request, tag == %u\n", idx); + blk_insert_request(host->oob_q, crq->rq, 1, crq, 0); + + return 0; +} + +static unsigned int carm_fill_sync_time(struct carm_host *host, + unsigned int idx, void *mem) +{ + struct timeval tv; + struct carm_msg_sync_time *st = mem; + + do_gettimeofday(&tv); + + memset(st, 0, sizeof(*st)); + st->type = CARM_MSG_MISC; + st->subtype = MISC_SET_TIME; + st->handle = cpu_to_le32(TAG_ENCODE(idx)); + st->timestamp = cpu_to_le32(tv.tv_sec); + + return sizeof(struct carm_msg_sync_time); +} + +static unsigned int carm_fill_alloc_buf(struct carm_host *host, + unsigned int idx, void *mem) +{ + struct carm_msg_allocbuf *ab = mem; + + memset(ab, 0, sizeof(*ab)); + ab->type = CARM_MSG_MISC; + ab->subtype = MISC_ALLOC_MEM; + ab->handle = cpu_to_le32(TAG_ENCODE(idx)); + ab->n_sg = 1; + ab->sg_type = SGT_32BIT; + ab->addr = cpu_to_le32(host->shm_dma + (PDC_SHM_SIZE >> 1)); + ab->len = cpu_to_le32(PDC_SHM_SIZE >> 1); + ab->evt_pool = cpu_to_le32(host->shm_dma + (16 * 1024)); + ab->n_evt = cpu_to_le32(1024); + ab->rbuf_pool = cpu_to_le32(host->shm_dma); + ab->n_rbuf = cpu_to_le32(RMSG_Q_LEN); + ab->msg_pool = cpu_to_le32(host->shm_dma + RBUF_LEN); + ab->n_msg = cpu_to_le32(CARM_Q_LEN); + ab->sg[0].start = cpu_to_le32(host->shm_dma + (PDC_SHM_SIZE >> 1)); + ab->sg[0].len = cpu_to_le32(65536); + + return sizeof(struct carm_msg_allocbuf); +} + +static unsigned int carm_fill_scan_channels(struct carm_host *host, + unsigned int idx, void *mem) +{ + struct carm_msg_ioctl *ioc = mem; + u32 msg_data = (u32) (carm_ref_msg_dma(host, idx) + + IOC_SCAN_CHAN_OFFSET); + + memset(ioc, 0, sizeof(*ioc)); + ioc->type = CARM_MSG_IOCTL; + ioc->subtype = CARM_IOC_SCAN_CHAN; + ioc->handle = cpu_to_le32(TAG_ENCODE(idx)); + ioc->data_addr = cpu_to_le32(msg_data); + + /* fill output data area with "no device" default values */ + mem += IOC_SCAN_CHAN_OFFSET; + memset(mem, IOC_SCAN_CHAN_NODEV, CARM_MAX_PORTS); + + return IOC_SCAN_CHAN_OFFSET + CARM_MAX_PORTS; +} + +static unsigned int carm_fill_get_fw_ver(struct carm_host *host, + unsigned int idx, void *mem) +{ + struct carm_msg_get_fw_ver *ioc = mem; + u32 msg_data = (u32) (carm_ref_msg_dma(host, idx) + sizeof(*ioc)); + + memset(ioc, 0, sizeof(*ioc)); + ioc->type = CARM_MSG_MISC; + ioc->subtype = MISC_GET_FW_VER; + ioc->handle = cpu_to_le32(TAG_ENCODE(idx)); + ioc->data_addr = cpu_to_le32(msg_data); + + return sizeof(struct carm_msg_get_fw_ver) + + sizeof(struct carm_fw_ver); +} + +static inline void carm_end_request_queued(struct carm_host *host, + struct carm_request *crq, + int uptodate) +{ + struct request *req = crq->rq; + int rc; + + rc = end_that_request_first(req, uptodate, req->hard_nr_sectors); + assert(rc == 0); + + end_that_request_last(req); + + rc = carm_put_request(host, crq); + assert(rc == 0); +} + +static inline void carm_push_q (struct carm_host *host, request_queue_t *q) +{ + unsigned int idx = host->wait_q_prod % CARM_MAX_WAIT_Q; + + blk_stop_queue(q); + VPRINTK("STOPPED QUEUE %p\n", q); + + host->wait_q[idx] = q; + host->wait_q_prod++; + BUG_ON(host->wait_q_prod == host->wait_q_cons); /* overrun */ +} + +static inline request_queue_t *carm_pop_q(struct carm_host *host) +{ + unsigned int idx; + + if (host->wait_q_prod == host->wait_q_cons) + return NULL; + + idx = host->wait_q_cons % CARM_MAX_WAIT_Q; + host->wait_q_cons++; + + return host->wait_q[idx]; +} + +static inline void carm_round_robin(struct carm_host *host) +{ + request_queue_t *q = carm_pop_q(host); + if (q) { + blk_start_queue(q); + VPRINTK("STARTED QUEUE %p\n", q); + } +} + +static inline void carm_end_rq(struct carm_host *host, struct carm_request *crq, + int is_ok) +{ + carm_end_request_queued(host, crq, is_ok); + if (CARM_MAX_Q == 1) + carm_round_robin(host); + else if ((host->n_msgs <= CARM_MSG_LOW_WATER) && + (host->hw_sg_used <= CARM_SG_LOW_WATER)) { + carm_round_robin(host); + } +} + +static void carm_oob_rq_fn(request_queue_t *q) +{ + struct carm_host *host = q->queuedata; + struct carm_request *crq; + struct request *rq; + int rc; + + while (1) { + DPRINTK("get req\n"); + rq = elv_next_request(q); + if (!rq) + break; + + blkdev_dequeue_request(rq); + + crq = rq->special; + assert(crq != NULL); + assert(crq->rq == rq); + + crq->n_elem = 0; + + DPRINTK("send req\n"); + rc = carm_send_msg(host, crq); + if (rc) { + blk_requeue_request(q, rq); + carm_push_q(host, q); + return; /* call us again later, eventually */ + } + } +} + +static void carm_rq_fn(request_queue_t *q) +{ + struct carm_port *port = q->queuedata; + struct carm_host *host = port->host; + struct carm_msg_rw *msg; + struct carm_request *crq; + struct request *rq; + struct scatterlist *sg; + int writing = 0, pci_dir, i, n_elem, rc; + u32 tmp; + unsigned int msg_size; + +queue_one_request: + VPRINTK("get req\n"); + rq = elv_next_request(q); + if (!rq) + return; + + crq = carm_get_request(host); + if (!crq) { + carm_push_q(host, q); + return; /* call us again later, eventually */ + } + crq->rq = rq; + + blkdev_dequeue_request(rq); + + if (rq_data_dir(rq) == WRITE) { + writing = 1; + pci_dir = PCI_DMA_TODEVICE; + } else { + pci_dir = PCI_DMA_FROMDEVICE; + } + + /* get scatterlist from block layer */ + sg = &crq->sg[0]; + n_elem = blk_rq_map_sg(q, rq, sg); + if (n_elem <= 0) { + carm_end_rq(host, crq, 0); + return; /* request with no s/g entries? */ + } + + /* map scatterlist to PCI bus addresses */ + n_elem = pci_map_sg(host->pdev, sg, n_elem, pci_dir); + if (n_elem <= 0) { + carm_end_rq(host, crq, 0); + return; /* request with no s/g entries? */ + } + crq->n_elem = n_elem; + crq->port = port; + host->hw_sg_used += n_elem; + + /* + * build read/write message + */ + + VPRINTK("build msg\n"); + msg = (struct carm_msg_rw *) carm_ref_msg(host, crq->tag); + + if (writing) { + msg->type = CARM_MSG_WRITE; + crq->msg_type = CARM_MSG_WRITE; + } else { + msg->type = CARM_MSG_READ; + crq->msg_type = CARM_MSG_READ; + } + + msg->id = port->port_no; + msg->sg_count = n_elem; + msg->sg_type = SGT_32BIT; + msg->handle = cpu_to_le32(TAG_ENCODE(crq->tag)); + msg->lba = cpu_to_le32(rq->sector & 0xffffffff); + tmp = (rq->sector >> 16) >> 16; + msg->lba_high = cpu_to_le16( (u16) tmp ); + msg->lba_count = cpu_to_le16(rq->nr_sectors); + + msg_size = sizeof(struct carm_msg_rw) - sizeof(msg->sg); + for (i = 0; i < n_elem; i++) { + struct carm_msg_sg *carm_sg = &msg->sg[i]; + carm_sg->start = cpu_to_le32(sg_dma_address(&crq->sg[i])); + carm_sg->len = cpu_to_le32(sg_dma_len(&crq->sg[i])); + msg_size += sizeof(struct carm_msg_sg); + } + + rc = carm_lookup_bucket(msg_size); + BUG_ON(rc < 0); + crq->msg_bucket = (u32) rc; + + /* + * queue read/write message to hardware + */ + + VPRINTK("send msg, tag == %u\n", crq->tag); + rc = carm_send_msg(host, crq); + if (rc) { + carm_put_request(host, crq); + blk_requeue_request(q, rq); + carm_push_q(host, q); + return; /* call us again later, eventually */ + } + + goto queue_one_request; +} + +static void carm_handle_array_info(struct carm_host *host, + struct carm_request *crq, u8 *mem, + int is_ok) +{ + struct carm_port *port; + u8 *msg_data = mem + sizeof(struct carm_array_info); + struct carm_array_info *desc = (struct carm_array_info *) msg_data; + u64 lo, hi; + int cur_port; + size_t slen; + + DPRINTK("ENTER\n"); + + carm_end_rq(host, crq, is_ok); + + if (!is_ok) + goto out; + if (le32_to_cpu(desc->array_status) & ARRAY_NO_EXIST) + goto out; + + cur_port = host->cur_scan_dev; + + /* should never occur */ + if ((cur_port < 0) || (cur_port >= CARM_MAX_PORTS)) { + printk(KERN_ERR PFX "BUG: cur_scan_dev==%d, array_id==%d\n", + cur_port, (int) desc->array_id); + goto out; + } + + port = &host->port[cur_port]; + + lo = (u64) le32_to_cpu(desc->size); + hi = (u64) le16_to_cpu(desc->size_hi); + + port->capacity = lo | (hi << 32); + port->dev_geom_head = le16_to_cpu(desc->head); + port->dev_geom_sect = le16_to_cpu(desc->sect); + port->dev_geom_cyl = le16_to_cpu(desc->cyl); + + host->dev_active |= (1 << cur_port); + + strncpy(port->name, desc->name, sizeof(port->name)); + port->name[sizeof(port->name) - 1] = 0; + slen = strlen(port->name); + while (slen && (port->name[slen - 1] == ' ')) { + port->name[slen - 1] = 0; + slen--; + } + + printk(KERN_INFO DRV_NAME "(%s): port %u device %Lu sectors\n", + pci_name(host->pdev), port->port_no, + (unsigned long long) port->capacity); + printk(KERN_INFO DRV_NAME "(%s): port %u device \"%s\"\n", + pci_name(host->pdev), port->port_no, port->name); + +out: + assert(host->state == HST_DEV_SCAN); + schedule_work(&host->fsm_task); +} + +static void carm_handle_scan_chan(struct carm_host *host, + struct carm_request *crq, u8 *mem, + int is_ok) +{ + u8 *msg_data = mem + IOC_SCAN_CHAN_OFFSET; + unsigned int i, dev_count = 0; + int new_state = HST_DEV_SCAN_START; + + DPRINTK("ENTER\n"); + + carm_end_rq(host, crq, is_ok); + + if (!is_ok) { + new_state = HST_ERROR; + goto out; + } + + /* TODO: scan and support non-disk devices */ + for (i = 0; i < 8; i++) + if (msg_data[i] == 0) { /* direct-access device (disk) */ + host->dev_present |= (1 << i); + dev_count++; + } + + printk(KERN_INFO DRV_NAME "(%s): found %u interesting devices\n", + pci_name(host->pdev), dev_count); + +out: + assert(host->state == HST_PORT_SCAN); + host->state = new_state; + schedule_work(&host->fsm_task); +} + +static void carm_handle_generic(struct carm_host *host, + struct carm_request *crq, int is_ok, + int cur_state, int next_state) +{ + DPRINTK("ENTER\n"); + + carm_end_rq(host, crq, is_ok); + + assert(host->state == cur_state); + if (is_ok) + host->state = next_state; + else + host->state = HST_ERROR; + schedule_work(&host->fsm_task); +} + +static inline void carm_handle_rw(struct carm_host *host, + struct carm_request *crq, int is_ok) +{ + int pci_dir; + + VPRINTK("ENTER\n"); + + if (rq_data_dir(crq->rq) == WRITE) + pci_dir = PCI_DMA_TODEVICE; + else + pci_dir = PCI_DMA_FROMDEVICE; + + pci_unmap_sg(host->pdev, &crq->sg[0], crq->n_elem, pci_dir); + + carm_end_rq(host, crq, is_ok); +} + +static inline void carm_handle_resp(struct carm_host *host, + __le32 ret_handle_le, u32 status) +{ + u32 handle = le32_to_cpu(ret_handle_le); + unsigned int msg_idx; + struct carm_request *crq; + int is_ok = (status == RMSG_OK); + u8 *mem; + + VPRINTK("ENTER, handle == 0x%x\n", handle); + + if (unlikely(!TAG_VALID(handle))) { + printk(KERN_ERR DRV_NAME "(%s): BUG: invalid tag 0x%x\n", + pci_name(host->pdev), handle); + return; + } + + msg_idx = TAG_DECODE(handle); + VPRINTK("tag == %u\n", msg_idx); + + crq = &host->req[msg_idx]; + + /* fast path */ + if (likely(crq->msg_type == CARM_MSG_READ || + crq->msg_type == CARM_MSG_WRITE)) { + carm_handle_rw(host, crq, is_ok); + return; + } + + mem = carm_ref_msg(host, msg_idx); + + switch (crq->msg_type) { + case CARM_MSG_IOCTL: { + switch (crq->msg_subtype) { + case CARM_IOC_SCAN_CHAN: + carm_handle_scan_chan(host, crq, mem, is_ok); + break; + default: + /* unknown / invalid response */ + goto err_out; + } + break; + } + + case CARM_MSG_MISC: { + switch (crq->msg_subtype) { + case MISC_ALLOC_MEM: + carm_handle_generic(host, crq, is_ok, + HST_ALLOC_BUF, HST_SYNC_TIME); + break; + case MISC_SET_TIME: + carm_handle_generic(host, crq, is_ok, + HST_SYNC_TIME, HST_GET_FW_VER); + break; + case MISC_GET_FW_VER: { + struct carm_fw_ver *ver = (struct carm_fw_ver *) + mem + sizeof(struct carm_msg_get_fw_ver); + if (is_ok) { + host->fw_ver = le32_to_cpu(ver->version); + host->flags |= (ver->features & FL_FW_VER_MASK); + } + carm_handle_generic(host, crq, is_ok, + HST_GET_FW_VER, HST_PORT_SCAN); + break; + } + default: + /* unknown / invalid response */ + goto err_out; + } + break; + } + + case CARM_MSG_ARRAY: { + switch (crq->msg_subtype) { + case CARM_ARRAY_INFO: + carm_handle_array_info(host, crq, mem, is_ok); + break; + default: + /* unknown / invalid response */ + goto err_out; + } + break; + } + + default: + /* unknown / invalid response */ + goto err_out; + } + + return; + +err_out: + printk(KERN_WARNING DRV_NAME "(%s): BUG: unhandled message type %d/%d\n", + pci_name(host->pdev), crq->msg_type, crq->msg_subtype); + carm_end_rq(host, crq, 0); +} + +static inline void carm_handle_responses(struct carm_host *host) +{ + void __iomem *mmio = host->mmio; + struct carm_response *resp = (struct carm_response *) host->shm; + unsigned int work = 0; + unsigned int idx = host->resp_idx % RMSG_Q_LEN; + + while (1) { + u32 status = le32_to_cpu(resp[idx].status); + + if (status == 0xffffffff) { + VPRINTK("ending response on index %u\n", idx); + writel(idx << 3, mmio + CARM_RESP_IDX); + break; + } + + /* response to a message we sent */ + else if ((status & (1 << 31)) == 0) { + VPRINTK("handling msg response on index %u\n", idx); + carm_handle_resp(host, resp[idx].ret_handle, status); + resp[idx].status = cpu_to_le32(0xffffffff); + } + + /* asynchronous events the hardware throws our way */ + else if ((status & 0xff000000) == (1 << 31)) { + u8 *evt_type_ptr = (u8 *) &resp[idx]; + u8 evt_type = *evt_type_ptr; + printk(KERN_WARNING DRV_NAME "(%s): unhandled event type %d\n", + pci_name(host->pdev), (int) evt_type); + resp[idx].status = cpu_to_le32(0xffffffff); + } + + idx = NEXT_RESP(idx); + work++; + } + + VPRINTK("EXIT, work==%u\n", work); + host->resp_idx += work; +} + +static irqreturn_t carm_interrupt(int irq, void *__host, struct pt_regs *regs) +{ + struct carm_host *host = __host; + void __iomem *mmio; + u32 mask; + int handled = 0; + unsigned long flags; + + if (!host) { + VPRINTK("no host\n"); + return IRQ_NONE; + } + + spin_lock_irqsave(&host->lock, flags); + + mmio = host->mmio; + + /* reading should also clear interrupts */ + mask = readl(mmio + CARM_INT_STAT); + + if (mask == 0 || mask == 0xffffffff) { + VPRINTK("no work, mask == 0x%x\n", mask); + goto out; + } + + if (mask & INT_ACK_MASK) + writel(mask, mmio + CARM_INT_STAT); + + if (unlikely(host->state == HST_INVALID)) { + VPRINTK("not initialized yet, mask = 0x%x\n", mask); + goto out; + } + + if (mask & CARM_HAVE_RESP) { + handled = 1; + carm_handle_responses(host); + } + +out: + spin_unlock_irqrestore(&host->lock, flags); + VPRINTK("EXIT\n"); + return IRQ_RETVAL(handled); +} + +static void carm_fsm_task (void *_data) +{ + struct carm_host *host = _data; + unsigned long flags; + unsigned int state; + int rc, i, next_dev; + int reschedule = 0; + int new_state = HST_INVALID; + + spin_lock_irqsave(&host->lock, flags); + state = host->state; + spin_unlock_irqrestore(&host->lock, flags); + + DPRINTK("ENTER, state == %s\n", state_name[state]); + + switch (state) { + case HST_PROBE_START: + new_state = HST_ALLOC_BUF; + reschedule = 1; + break; + + case HST_ALLOC_BUF: + rc = carm_send_special(host, carm_fill_alloc_buf); + if (rc) { + new_state = HST_ERROR; + reschedule = 1; + } + break; + + case HST_SYNC_TIME: + rc = carm_send_special(host, carm_fill_sync_time); + if (rc) { + new_state = HST_ERROR; + reschedule = 1; + } + break; + + case HST_GET_FW_VER: + rc = carm_send_special(host, carm_fill_get_fw_ver); + if (rc) { + new_state = HST_ERROR; + reschedule = 1; + } + break; + + case HST_PORT_SCAN: + rc = carm_send_special(host, carm_fill_scan_channels); + if (rc) { + new_state = HST_ERROR; + reschedule = 1; + } + break; + + case HST_DEV_SCAN_START: + host->cur_scan_dev = -1; + new_state = HST_DEV_SCAN; + reschedule = 1; + break; + + case HST_DEV_SCAN: + next_dev = -1; + for (i = host->cur_scan_dev + 1; i < CARM_MAX_PORTS; i++) + if (host->dev_present & (1 << i)) { + next_dev = i; + break; + } + + if (next_dev >= 0) { + host->cur_scan_dev = next_dev; + rc = carm_array_info(host, next_dev); + if (rc) { + new_state = HST_ERROR; + reschedule = 1; + } + } else { + new_state = HST_DEV_ACTIVATE; + reschedule = 1; + } + break; + + case HST_DEV_ACTIVATE: { + int activated = 0; + for (i = 0; i < CARM_MAX_PORTS; i++) + if (host->dev_active & (1 << i)) { + struct carm_port *port = &host->port[i]; + struct gendisk *disk = port->disk; + + set_capacity(disk, port->capacity); + add_disk(disk); + activated++; + } + + printk(KERN_INFO DRV_NAME "(%s): %d ports activated\n", + pci_name(host->pdev), activated); + + new_state = HST_PROBE_FINISHED; + reschedule = 1; + break; + } + + case HST_PROBE_FINISHED: + up(&host->probe_sem); + break; + + case HST_ERROR: + /* FIXME: TODO */ + break; + + default: + /* should never occur */ + printk(KERN_ERR PFX "BUG: unknown state %d\n", state); + assert(0); + break; + } + + if (new_state != HST_INVALID) { + spin_lock_irqsave(&host->lock, flags); + host->state = new_state; + spin_unlock_irqrestore(&host->lock, flags); + } + if (reschedule) + schedule_work(&host->fsm_task); +} + +static int carm_init_wait(void __iomem *mmio, u32 bits, unsigned int test_bit) +{ + unsigned int i; + + for (i = 0; i < 50000; i++) { + u32 tmp = readl(mmio + CARM_LMUC); + udelay(100); + + if (test_bit) { + if ((tmp & bits) == bits) + return 0; + } else { + if ((tmp & bits) == 0) + return 0; + } + + cond_resched(); + } + + printk(KERN_ERR PFX "carm_init_wait timeout, bits == 0x%x, test_bit == %s\n", + bits, test_bit ? "yes" : "no"); + return -EBUSY; +} + +static void carm_init_responses(struct carm_host *host) +{ + void __iomem *mmio = host->mmio; + unsigned int i; + struct carm_response *resp = (struct carm_response *) host->shm; + + for (i = 0; i < RMSG_Q_LEN; i++) + resp[i].status = cpu_to_le32(0xffffffff); + + writel(0, mmio + CARM_RESP_IDX); +} + +static int carm_init_host(struct carm_host *host) +{ + void __iomem *mmio = host->mmio; + u32 tmp; + u8 tmp8; + int rc; + + DPRINTK("ENTER\n"); + + writel(0, mmio + CARM_INT_MASK); + + tmp8 = readb(mmio + CARM_INITC); + if (tmp8 & 0x01) { + tmp8 &= ~0x01; + writeb(tmp8, mmio + CARM_INITC); + readb(mmio + CARM_INITC); /* flush */ + + DPRINTK("snooze...\n"); + msleep(5000); + } + + tmp = readl(mmio + CARM_HMUC); + if (tmp & CARM_CME) { + DPRINTK("CME bit present, waiting\n"); + rc = carm_init_wait(mmio, CARM_CME, 1); + if (rc) { + DPRINTK("EXIT, carm_init_wait 1 failed\n"); + return rc; + } + } + if (tmp & CARM_RME) { + DPRINTK("RME bit present, waiting\n"); + rc = carm_init_wait(mmio, CARM_RME, 1); + if (rc) { + DPRINTK("EXIT, carm_init_wait 2 failed\n"); + return rc; + } + } + + tmp &= ~(CARM_RME | CARM_CME); + writel(tmp, mmio + CARM_HMUC); + readl(mmio + CARM_HMUC); /* flush */ + + rc = carm_init_wait(mmio, CARM_RME | CARM_CME, 0); + if (rc) { + DPRINTK("EXIT, carm_init_wait 3 failed\n"); + return rc; + } + + carm_init_buckets(mmio); + + writel(host->shm_dma & 0xffffffff, mmio + RBUF_ADDR_LO); + writel((host->shm_dma >> 16) >> 16, mmio + RBUF_ADDR_HI); + writel(RBUF_LEN, mmio + RBUF_BYTE_SZ); + + tmp = readl(mmio + CARM_HMUC); + tmp |= (CARM_RME | CARM_CME | CARM_WZBC); + writel(tmp, mmio + CARM_HMUC); + readl(mmio + CARM_HMUC); /* flush */ + + rc = carm_init_wait(mmio, CARM_RME | CARM_CME, 1); + if (rc) { + DPRINTK("EXIT, carm_init_wait 4 failed\n"); + return rc; + } + + writel(0, mmio + CARM_HMPHA); + writel(INT_DEF_MASK, mmio + CARM_INT_MASK); + + carm_init_responses(host); + + /* start initialization, probing state machine */ + spin_lock_irq(&host->lock); + assert(host->state == HST_INVALID); + host->state = HST_PROBE_START; + spin_unlock_irq(&host->lock); + schedule_work(&host->fsm_task); + + DPRINTK("EXIT\n"); + return 0; +} + +static int carm_init_disks(struct carm_host *host) +{ + unsigned int i; + int rc = 0; + + for (i = 0; i < CARM_MAX_PORTS; i++) { + struct gendisk *disk; + request_queue_t *q; + struct carm_port *port; + + port = &host->port[i]; + port->host = host; + port->port_no = i; + + disk = alloc_disk(CARM_MINORS_PER_MAJOR); + if (!disk) { + rc = -ENOMEM; + break; + } + + port->disk = disk; + sprintf(disk->disk_name, DRV_NAME "/%u", + (unsigned int) (host->id * CARM_MAX_PORTS) + i); + sprintf(disk->devfs_name, DRV_NAME "/%u_%u", host->id, i); + disk->major = host->major; + disk->first_minor = i * CARM_MINORS_PER_MAJOR; + disk->fops = &carm_bd_ops; + disk->private_data = port; + + q = blk_init_queue(carm_rq_fn, &host->lock); + if (!q) { + rc = -ENOMEM; + break; + } + disk->queue = q; + blk_queue_max_hw_segments(q, CARM_MAX_REQ_SG); + blk_queue_max_phys_segments(q, CARM_MAX_REQ_SG); + blk_queue_segment_boundary(q, CARM_SG_BOUNDARY); + + q->queuedata = port; + } + + return rc; +} + +static void carm_free_disks(struct carm_host *host) +{ + unsigned int i; + + for (i = 0; i < CARM_MAX_PORTS; i++) { + struct gendisk *disk = host->port[i].disk; + if (disk) { + request_queue_t *q = disk->queue; + + if (disk->flags & GENHD_FL_UP) + del_gendisk(disk); + if (q) + blk_cleanup_queue(q); + put_disk(disk); + } + } +} + +static int carm_init_shm(struct carm_host *host) +{ + host->shm = pci_alloc_consistent(host->pdev, CARM_SHM_SIZE, + &host->shm_dma); + if (!host->shm) + return -ENOMEM; + + host->msg_base = host->shm + RBUF_LEN; + host->msg_dma = host->shm_dma + RBUF_LEN; + + memset(host->shm, 0xff, RBUF_LEN); + memset(host->msg_base, 0, PDC_SHM_SIZE - RBUF_LEN); + + return 0; +} + +static int carm_init_one (struct pci_dev *pdev, const struct pci_device_id *ent) +{ + static unsigned int printed_version; + struct carm_host *host; + unsigned int pci_dac; + int rc; + request_queue_t *q; + unsigned int i; + + if (!printed_version++) + printk(KERN_DEBUG DRV_NAME " version " DRV_VERSION "\n"); + + rc = pci_enable_device(pdev); + if (rc) + return rc; + + rc = pci_request_regions(pdev, DRV_NAME); + if (rc) + goto err_out; + +#if IF_64BIT_DMA_IS_POSSIBLE /* grrrr... */ + rc = pci_set_dma_mask(pdev, 0xffffffffffffffffULL); + if (!rc) { + rc = pci_set_consistent_dma_mask(pdev, 0xffffffffffffffffULL); + if (rc) { + printk(KERN_ERR DRV_NAME "(%s): consistent DMA mask failure\n", + pci_name(pdev)); + goto err_out_regions; + } + pci_dac = 1; + } else { +#endif + rc = pci_set_dma_mask(pdev, 0xffffffffULL); + if (rc) { + printk(KERN_ERR DRV_NAME "(%s): DMA mask failure\n", + pci_name(pdev)); + goto err_out_regions; + } + pci_dac = 0; +#if IF_64BIT_DMA_IS_POSSIBLE /* grrrr... */ + } +#endif + + host = kmalloc(sizeof(*host), GFP_KERNEL); + if (!host) { + printk(KERN_ERR DRV_NAME "(%s): memory alloc failure\n", + pci_name(pdev)); + rc = -ENOMEM; + goto err_out_regions; + } + + memset(host, 0, sizeof(*host)); + host->pdev = pdev; + host->flags = pci_dac ? FL_DAC : 0; + spin_lock_init(&host->lock); + INIT_WORK(&host->fsm_task, carm_fsm_task, host); + init_MUTEX_LOCKED(&host->probe_sem); + + for (i = 0; i < ARRAY_SIZE(host->req); i++) + host->req[i].tag = i; + + host->mmio = ioremap(pci_resource_start(pdev, 0), + pci_resource_len(pdev, 0)); + if (!host->mmio) { + printk(KERN_ERR DRV_NAME "(%s): MMIO alloc failure\n", + pci_name(pdev)); + rc = -ENOMEM; + goto err_out_kfree; + } + + rc = carm_init_shm(host); + if (rc) { + printk(KERN_ERR DRV_NAME "(%s): DMA SHM alloc failure\n", + pci_name(pdev)); + goto err_out_iounmap; + } + + q = blk_init_queue(carm_oob_rq_fn, &host->lock); + if (!q) { + printk(KERN_ERR DRV_NAME "(%s): OOB queue alloc failure\n", + pci_name(pdev)); + rc = -ENOMEM; + goto err_out_pci_free; + } + host->oob_q = q; + q->queuedata = host; + + /* + * Figure out which major to use: 160, 161, or dynamic + */ + if (!test_and_set_bit(0, &carm_major_alloc)) + host->major = 160; + else if (!test_and_set_bit(1, &carm_major_alloc)) + host->major = 161; + else + host->flags |= FL_DYN_MAJOR; + + host->id = carm_host_id; + sprintf(host->name, DRV_NAME "%d", carm_host_id); + + rc = register_blkdev(host->major, host->name); + if (rc < 0) + goto err_out_free_majors; + if (host->flags & FL_DYN_MAJOR) + host->major = rc; + + devfs_mk_dir(DRV_NAME); + + rc = carm_init_disks(host); + if (rc) + goto err_out_blkdev_disks; + + pci_set_master(pdev); + + rc = request_irq(pdev->irq, carm_interrupt, SA_SHIRQ, DRV_NAME, host); + if (rc) { + printk(KERN_ERR DRV_NAME "(%s): irq alloc failure\n", + pci_name(pdev)); + goto err_out_blkdev_disks; + } + + rc = carm_init_host(host); + if (rc) + goto err_out_free_irq; + + DPRINTK("waiting for probe_sem\n"); + down(&host->probe_sem); + + printk(KERN_INFO "%s: pci %s, ports %d, io %lx, irq %u, major %d\n", + host->name, pci_name(pdev), (int) CARM_MAX_PORTS, + pci_resource_start(pdev, 0), pdev->irq, host->major); + + carm_host_id++; + pci_set_drvdata(pdev, host); + return 0; + +err_out_free_irq: + free_irq(pdev->irq, host); +err_out_blkdev_disks: + carm_free_disks(host); + unregister_blkdev(host->major, host->name); +err_out_free_majors: + if (host->major == 160) + clear_bit(0, &carm_major_alloc); + else if (host->major == 161) + clear_bit(1, &carm_major_alloc); + blk_cleanup_queue(host->oob_q); +err_out_pci_free: + pci_free_consistent(pdev, CARM_SHM_SIZE, host->shm, host->shm_dma); +err_out_iounmap: + iounmap(host->mmio); +err_out_kfree: + kfree(host); +err_out_regions: + pci_release_regions(pdev); +err_out: + pci_disable_device(pdev); + return rc; +} + +static void carm_remove_one (struct pci_dev *pdev) +{ + struct carm_host *host = pci_get_drvdata(pdev); + + if (!host) { + printk(KERN_ERR PFX "BUG: no host data for PCI(%s)\n", + pci_name(pdev)); + return; + } + + free_irq(pdev->irq, host); + carm_free_disks(host); + devfs_remove(DRV_NAME); + unregister_blkdev(host->major, host->name); + if (host->major == 160) + clear_bit(0, &carm_major_alloc); + else if (host->major == 161) + clear_bit(1, &carm_major_alloc); + blk_cleanup_queue(host->oob_q); + pci_free_consistent(pdev, CARM_SHM_SIZE, host->shm, host->shm_dma); + iounmap(host->mmio); + kfree(host); + pci_release_regions(pdev); + pci_disable_device(pdev); + pci_set_drvdata(pdev, NULL); +} + +static int __init carm_init(void) +{ + return pci_module_init(&carm_driver); +} + +static void __exit carm_exit(void) +{ + pci_unregister_driver(&carm_driver); +} + +module_init(carm_init); +module_exit(carm_exit); + + diff --git a/drivers/block/ub.c b/drivers/block/ub.c new file mode 100644 index 00000000000..ce42889f98f --- /dev/null +++ b/drivers/block/ub.c @@ -0,0 +1,2215 @@ +/* + * The low performance USB storage driver (ub). + * + * Copyright (c) 1999, 2000 Matthew Dharm (mdharm-usb@one-eyed-alien.net) + * Copyright (C) 2004 Pete Zaitcev (zaitcev@yahoo.com) + * + * This work is a part of Linux kernel, is derived from it, + * and is not licensed separately. See file COPYING for details. + * + * TODO (sorted by decreasing priority) + * -- Do resets with usb_device_reset (needs a thread context, use khubd) + * -- set readonly flag for CDs, set removable flag for CF readers + * -- do inquiry and verify we got a disk and not a tape (for LUN mismatch) + * -- support pphaneuf's SDDR-75 with two LUNs (also broken capacity...) + * -- special case some senses, e.g. 3a/0 -> no media present, reduce retries + * -- verify the 13 conditions and do bulk resets + * -- normal pool of commands instead of cmdv[]? + * -- kill last_pipe and simply do two-state clearing on both pipes + * -- verify protocol (bulk) from USB descriptors (maybe...) + * -- highmem and sg + * -- move top_sense and work_bcs into separate allocations (if they survive) + * for cache purists and esoteric architectures. + * -- prune comments, they are too volumnous + * -- Exterminate P3 printks + * -- Resove XXX's + * -- Redo "benh's retries", perhaps have spin-up code to handle them. V:D=? + */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/usb.h> +#include <linux/blkdev.h> +#include <linux/devfs_fs_kernel.h> +#include <linux/timer.h> +#include <scsi/scsi.h> + +#define DRV_NAME "ub" +#define DEVFS_NAME DRV_NAME + +#define UB_MAJOR 180 + +/* + * Definitions which have to be scattered once we understand the layout better. + */ + +/* Transport (despite PR in the name) */ +#define US_PR_BULK 0x50 /* bulk only */ + +/* Protocol */ +#define US_SC_SCSI 0x06 /* Transparent */ + +/* + */ +#define UB_MINORS_PER_MAJOR 8 + +#define UB_MAX_CDB_SIZE 16 /* Corresponds to Bulk */ + +#define UB_SENSE_SIZE 18 + +/* + */ + +/* command block wrapper */ +struct bulk_cb_wrap { + __le32 Signature; /* contains 'USBC' */ + u32 Tag; /* unique per command id */ + __le32 DataTransferLength; /* size of data */ + u8 Flags; /* direction in bit 0 */ + u8 Lun; /* LUN normally 0 */ + u8 Length; /* of of the CDB */ + u8 CDB[UB_MAX_CDB_SIZE]; /* max command */ +}; + +#define US_BULK_CB_WRAP_LEN 31 +#define US_BULK_CB_SIGN 0x43425355 /*spells out USBC */ +#define US_BULK_FLAG_IN 1 +#define US_BULK_FLAG_OUT 0 + +/* command status wrapper */ +struct bulk_cs_wrap { + __le32 Signature; /* should = 'USBS' */ + u32 Tag; /* same as original command */ + __le32 Residue; /* amount not transferred */ + u8 Status; /* see below */ +}; + +#define US_BULK_CS_WRAP_LEN 13 +#define US_BULK_CS_SIGN 0x53425355 /* spells out 'USBS' */ +/* This is for Olympus Camedia digital cameras */ +#define US_BULK_CS_OLYMPUS_SIGN 0x55425355 /* spells out 'USBU' */ +#define US_BULK_STAT_OK 0 +#define US_BULK_STAT_FAIL 1 +#define US_BULK_STAT_PHASE 2 + +/* bulk-only class specific requests */ +#define US_BULK_RESET_REQUEST 0xff +#define US_BULK_GET_MAX_LUN 0xfe + +/* + */ +struct ub_dev; + +#define UB_MAX_REQ_SG 1 +#define UB_MAX_SECTORS 64 + +/* + * A second is more than enough for a 32K transfer (UB_MAX_SECTORS) + * even if a webcam hogs the bus, but some devices need time to spin up. + */ +#define UB_URB_TIMEOUT (HZ*2) +#define UB_DATA_TIMEOUT (HZ*5) /* ZIP does spin-ups in the data phase */ +#define UB_STAT_TIMEOUT (HZ*5) /* Same spinups and eject for a dataless cmd. */ +#define UB_CTRL_TIMEOUT (HZ/2) /* 500ms ought to be enough to clear a stall */ + +/* + * An instance of a SCSI command in transit. + */ +#define UB_DIR_NONE 0 +#define UB_DIR_READ 1 +#define UB_DIR_ILLEGAL2 2 +#define UB_DIR_WRITE 3 + +#define UB_DIR_CHAR(c) (((c)==UB_DIR_WRITE)? 'w': \ + (((c)==UB_DIR_READ)? 'r': 'n')) + +enum ub_scsi_cmd_state { + UB_CMDST_INIT, /* Initial state */ + UB_CMDST_CMD, /* Command submitted */ + UB_CMDST_DATA, /* Data phase */ + UB_CMDST_CLR2STS, /* Clearing before requesting status */ + UB_CMDST_STAT, /* Status phase */ + UB_CMDST_CLEAR, /* Clearing a stall (halt, actually) */ + UB_CMDST_SENSE, /* Sending Request Sense */ + UB_CMDST_DONE /* Final state */ +}; + +static char *ub_scsi_cmd_stname[] = { + ". ", + "Cmd", + "dat", + "c2s", + "sts", + "clr", + "Sen", + "fin" +}; + +struct ub_scsi_cmd { + unsigned char cdb[UB_MAX_CDB_SIZE]; + unsigned char cdb_len; + + unsigned char dir; /* 0 - none, 1 - read, 3 - write. */ + unsigned char trace_index; + enum ub_scsi_cmd_state state; + unsigned int tag; + struct ub_scsi_cmd *next; + + int error; /* Return code - valid upon done */ + unsigned int act_len; /* Return size */ + unsigned char key, asc, ascq; /* May be valid if error==-EIO */ + + int stat_count; /* Retries getting status. */ + + /* + * We do not support transfers from highmem pages + * because the underlying USB framework does not do what we need. + */ + char *data; /* Requested buffer */ + unsigned int len; /* Requested length */ + // struct scatterlist sgv[UB_MAX_REQ_SG]; + + void (*done)(struct ub_dev *, struct ub_scsi_cmd *); + void *back; +}; + +/* + */ +struct ub_capacity { + unsigned long nsec; /* Linux size - 512 byte sectors */ + unsigned int bsize; /* Linux hardsect_size */ + unsigned int bshift; /* Shift between 512 and hard sects */ +}; + +/* + * The SCSI command tracing structure. + */ + +#define SCMD_ST_HIST_SZ 8 +#define SCMD_TRACE_SZ 63 /* Less than 4KB of 61-byte lines */ + +struct ub_scsi_cmd_trace { + int hcur; + unsigned int tag; + unsigned int req_size, act_size; + unsigned char op; + unsigned char dir; + unsigned char key, asc, ascq; + char st_hst[SCMD_ST_HIST_SZ]; +}; + +struct ub_scsi_trace { + int cur; + struct ub_scsi_cmd_trace vec[SCMD_TRACE_SZ]; +}; + +/* + * This is a direct take-off from linux/include/completion.h + * The difference is that I do not wait on this thing, just poll. + * When I want to wait (ub_probe), I just use the stock completion. + * + * Note that INIT_COMPLETION takes no lock. It is correct. But why + * in the bloody hell that thing takes struct instead of pointer to struct + * is quite beyond me. I just copied it from the stock completion. + */ +struct ub_completion { + unsigned int done; + spinlock_t lock; +}; + +static inline void ub_init_completion(struct ub_completion *x) +{ + x->done = 0; + spin_lock_init(&x->lock); +} + +#define UB_INIT_COMPLETION(x) ((x).done = 0) + +static void ub_complete(struct ub_completion *x) +{ + unsigned long flags; + + spin_lock_irqsave(&x->lock, flags); + x->done++; + spin_unlock_irqrestore(&x->lock, flags); +} + +static int ub_is_completed(struct ub_completion *x) +{ + unsigned long flags; + int ret; + + spin_lock_irqsave(&x->lock, flags); + ret = x->done; + spin_unlock_irqrestore(&x->lock, flags); + return ret; +} + +/* + */ +struct ub_scsi_cmd_queue { + int qlen, qmax; + struct ub_scsi_cmd *head, *tail; +}; + +/* + * The UB device instance. + */ +struct ub_dev { + spinlock_t lock; + int id; /* Number among ub's */ + atomic_t poison; /* The USB device is disconnected */ + int openc; /* protected by ub_lock! */ + /* kref is too implicit for our taste */ + unsigned int tagcnt; + int changed; /* Media was changed */ + int removable; + int readonly; + int first_open; /* Kludge. See ub_bd_open. */ + char name[8]; + struct usb_device *dev; + struct usb_interface *intf; + + struct ub_capacity capacity; + struct gendisk *disk; + + unsigned int send_bulk_pipe; /* cached pipe values */ + unsigned int recv_bulk_pipe; + unsigned int send_ctrl_pipe; + unsigned int recv_ctrl_pipe; + + struct tasklet_struct tasklet; + + /* XXX Use Ingo's mempool (once we have more than one) */ + int cmda[1]; + struct ub_scsi_cmd cmdv[1]; + + struct ub_scsi_cmd_queue cmd_queue; + struct ub_scsi_cmd top_rqs_cmd; /* REQUEST SENSE */ + unsigned char top_sense[UB_SENSE_SIZE]; + + struct ub_completion work_done; + struct urb work_urb; + struct timer_list work_timer; + int last_pipe; /* What might need clearing */ + struct bulk_cb_wrap work_bcb; + struct bulk_cs_wrap work_bcs; + struct usb_ctrlrequest work_cr; + + struct ub_scsi_trace tr; +}; + +/* + */ +static void ub_cleanup(struct ub_dev *sc); +static int ub_bd_rq_fn_1(struct ub_dev *sc, struct request *rq); +static int ub_cmd_build_block(struct ub_dev *sc, struct ub_scsi_cmd *cmd, + struct request *rq); +static int ub_cmd_build_packet(struct ub_dev *sc, struct ub_scsi_cmd *cmd, + struct request *rq); +static void ub_rw_cmd_done(struct ub_dev *sc, struct ub_scsi_cmd *cmd); +static void ub_end_rq(struct request *rq, int uptodate); +static int ub_submit_scsi(struct ub_dev *sc, struct ub_scsi_cmd *cmd); +static void ub_urb_complete(struct urb *urb, struct pt_regs *pt); +static void ub_scsi_action(unsigned long _dev); +static void ub_scsi_dispatch(struct ub_dev *sc); +static void ub_scsi_urb_compl(struct ub_dev *sc, struct ub_scsi_cmd *cmd); +static void ub_state_done(struct ub_dev *sc, struct ub_scsi_cmd *cmd, int rc); +static void __ub_state_stat(struct ub_dev *sc, struct ub_scsi_cmd *cmd); +static void ub_state_stat(struct ub_dev *sc, struct ub_scsi_cmd *cmd); +static void ub_state_sense(struct ub_dev *sc, struct ub_scsi_cmd *cmd); +static int ub_submit_clear_stall(struct ub_dev *sc, struct ub_scsi_cmd *cmd, + int stalled_pipe); +static void ub_top_sense_done(struct ub_dev *sc, struct ub_scsi_cmd *scmd); +static int ub_sync_tur(struct ub_dev *sc); +static int ub_sync_read_cap(struct ub_dev *sc, struct ub_capacity *ret); + +/* + */ +static struct usb_device_id ub_usb_ids[] = { + // { USB_DEVICE_VER(0x0781, 0x0002, 0x0009, 0x0009) }, /* SDDR-31 */ + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_SCSI, US_PR_BULK) }, + { } +}; + +MODULE_DEVICE_TABLE(usb, ub_usb_ids); + +/* + * Find me a way to identify "next free minor" for add_disk(), + * and the array disappears the next day. However, the number of + * hosts has something to do with the naming and /proc/partitions. + * This has to be thought out in detail before changing. + * If UB_MAX_HOST was 1000, we'd use a bitmap. Or a better data structure. + */ +#define UB_MAX_HOSTS 26 +static char ub_hostv[UB_MAX_HOSTS]; +static DEFINE_SPINLOCK(ub_lock); /* Locks globals and ->openc */ + +/* + * The SCSI command tracing procedures. + */ + +static void ub_cmdtr_new(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + int n; + struct ub_scsi_cmd_trace *t; + + if ((n = sc->tr.cur + 1) == SCMD_TRACE_SZ) n = 0; + t = &sc->tr.vec[n]; + + memset(t, 0, sizeof(struct ub_scsi_cmd_trace)); + t->tag = cmd->tag; + t->op = cmd->cdb[0]; + t->dir = cmd->dir; + t->req_size = cmd->len; + t->st_hst[0] = cmd->state; + + sc->tr.cur = n; + cmd->trace_index = n; +} + +static void ub_cmdtr_state(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + int n; + struct ub_scsi_cmd_trace *t; + + t = &sc->tr.vec[cmd->trace_index]; + if (t->tag == cmd->tag) { + if ((n = t->hcur + 1) == SCMD_ST_HIST_SZ) n = 0; + t->st_hst[n] = cmd->state; + t->hcur = n; + } +} + +static void ub_cmdtr_act_len(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct ub_scsi_cmd_trace *t; + + t = &sc->tr.vec[cmd->trace_index]; + if (t->tag == cmd->tag) + t->act_size = cmd->act_len; +} + +static void ub_cmdtr_sense(struct ub_dev *sc, struct ub_scsi_cmd *cmd, + unsigned char *sense) +{ + struct ub_scsi_cmd_trace *t; + + t = &sc->tr.vec[cmd->trace_index]; + if (t->tag == cmd->tag) { + t->key = sense[2] & 0x0F; + t->asc = sense[12]; + t->ascq = sense[13]; + } +} + +static ssize_t ub_diag_show(struct device *dev, char *page) +{ + struct usb_interface *intf; + struct ub_dev *sc; + int cnt; + unsigned long flags; + int nc, nh; + int i, j; + struct ub_scsi_cmd_trace *t; + + intf = to_usb_interface(dev); + sc = usb_get_intfdata(intf); + if (sc == NULL) + return 0; + + cnt = 0; + spin_lock_irqsave(&sc->lock, flags); + + cnt += sprintf(page + cnt, + "qlen %d qmax %d changed %d removable %d readonly %d\n", + sc->cmd_queue.qlen, sc->cmd_queue.qmax, + sc->changed, sc->removable, sc->readonly); + + if ((nc = sc->tr.cur + 1) == SCMD_TRACE_SZ) nc = 0; + for (j = 0; j < SCMD_TRACE_SZ; j++) { + t = &sc->tr.vec[nc]; + + cnt += sprintf(page + cnt, "%08x %02x", t->tag, t->op); + if (t->op == REQUEST_SENSE) { + cnt += sprintf(page + cnt, " [sense %x %02x %02x]", + t->key, t->asc, t->ascq); + } else { + cnt += sprintf(page + cnt, " %c", UB_DIR_CHAR(t->dir)); + cnt += sprintf(page + cnt, " [%5d %5d]", + t->req_size, t->act_size); + } + if ((nh = t->hcur + 1) == SCMD_ST_HIST_SZ) nh = 0; + for (i = 0; i < SCMD_ST_HIST_SZ; i++) { + cnt += sprintf(page + cnt, " %s", + ub_scsi_cmd_stname[(int)t->st_hst[nh]]); + if (++nh == SCMD_ST_HIST_SZ) nh = 0; + } + cnt += sprintf(page + cnt, "\n"); + + if (++nc == SCMD_TRACE_SZ) nc = 0; + } + + spin_unlock_irqrestore(&sc->lock, flags); + return cnt; +} + +static DEVICE_ATTR(diag, S_IRUGO, ub_diag_show, NULL); /* N.B. World readable */ + +/* + * The id allocator. + * + * This also stores the host for indexing by minor, which is somewhat dirty. + */ +static int ub_id_get(void) +{ + unsigned long flags; + int i; + + spin_lock_irqsave(&ub_lock, flags); + for (i = 0; i < UB_MAX_HOSTS; i++) { + if (ub_hostv[i] == 0) { + ub_hostv[i] = 1; + spin_unlock_irqrestore(&ub_lock, flags); + return i; + } + } + spin_unlock_irqrestore(&ub_lock, flags); + return -1; +} + +static void ub_id_put(int id) +{ + unsigned long flags; + + if (id < 0 || id >= UB_MAX_HOSTS) { + printk(KERN_ERR DRV_NAME ": bad host ID %d\n", id); + return; + } + + spin_lock_irqsave(&ub_lock, flags); + if (ub_hostv[id] == 0) { + spin_unlock_irqrestore(&ub_lock, flags); + printk(KERN_ERR DRV_NAME ": freeing free host ID %d\n", id); + return; + } + ub_hostv[id] = 0; + spin_unlock_irqrestore(&ub_lock, flags); +} + +/* + * Downcount for deallocation. This rides on two assumptions: + * - once something is poisoned, its refcount cannot grow + * - opens cannot happen at this time (del_gendisk was done) + * If the above is true, we can drop the lock, which we need for + * blk_cleanup_queue(): the silly thing may attempt to sleep. + * [Actually, it never needs to sleep for us, but it calls might_sleep()] + */ +static void ub_put(struct ub_dev *sc) +{ + unsigned long flags; + + spin_lock_irqsave(&ub_lock, flags); + --sc->openc; + if (sc->openc == 0 && atomic_read(&sc->poison)) { + spin_unlock_irqrestore(&ub_lock, flags); + ub_cleanup(sc); + } else { + spin_unlock_irqrestore(&ub_lock, flags); + } +} + +/* + * Final cleanup and deallocation. + */ +static void ub_cleanup(struct ub_dev *sc) +{ + request_queue_t *q; + + /* I don't think queue can be NULL. But... Stolen from sx8.c */ + if ((q = sc->disk->queue) != NULL) + blk_cleanup_queue(q); + + /* + * If we zero disk->private_data BEFORE put_disk, we have to check + * for NULL all over the place in open, release, check_media and + * revalidate, because the block level semaphore is well inside the + * put_disk. But we cannot zero after the call, because *disk is gone. + * The sd.c is blatantly racy in this area. + */ + /* disk->private_data = NULL; */ + put_disk(sc->disk); + sc->disk = NULL; + + ub_id_put(sc->id); + kfree(sc); +} + +/* + * The "command allocator". + */ +static struct ub_scsi_cmd *ub_get_cmd(struct ub_dev *sc) +{ + struct ub_scsi_cmd *ret; + + if (sc->cmda[0]) + return NULL; + ret = &sc->cmdv[0]; + sc->cmda[0] = 1; + return ret; +} + +static void ub_put_cmd(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + if (cmd != &sc->cmdv[0]) { + printk(KERN_WARNING "%s: releasing a foreign cmd %p\n", + sc->name, cmd); + return; + } + if (!sc->cmda[0]) { + printk(KERN_WARNING "%s: releasing a free cmd\n", sc->name); + return; + } + sc->cmda[0] = 0; +} + +/* + * The command queue. + */ +static void ub_cmdq_add(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct ub_scsi_cmd_queue *t = &sc->cmd_queue; + + if (t->qlen++ == 0) { + t->head = cmd; + t->tail = cmd; + } else { + t->tail->next = cmd; + t->tail = cmd; + } + + if (t->qlen > t->qmax) + t->qmax = t->qlen; +} + +static void ub_cmdq_insert(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct ub_scsi_cmd_queue *t = &sc->cmd_queue; + + if (t->qlen++ == 0) { + t->head = cmd; + t->tail = cmd; + } else { + cmd->next = t->head; + t->head = cmd; + } + + if (t->qlen > t->qmax) + t->qmax = t->qlen; +} + +static struct ub_scsi_cmd *ub_cmdq_pop(struct ub_dev *sc) +{ + struct ub_scsi_cmd_queue *t = &sc->cmd_queue; + struct ub_scsi_cmd *cmd; + + if (t->qlen == 0) + return NULL; + if (--t->qlen == 0) + t->tail = NULL; + cmd = t->head; + t->head = cmd->next; + cmd->next = NULL; + return cmd; +} + +#define ub_cmdq_peek(sc) ((sc)->cmd_queue.head) + +/* + * The request function is our main entry point + */ + +static void ub_bd_rq_fn(request_queue_t *q) +{ + struct ub_dev *sc = q->queuedata; + struct request *rq; + + while ((rq = elv_next_request(q)) != NULL) { + if (ub_bd_rq_fn_1(sc, rq) != 0) { + blk_stop_queue(q); + break; + } + } +} + +static int ub_bd_rq_fn_1(struct ub_dev *sc, struct request *rq) +{ + struct ub_scsi_cmd *cmd; + int rc; + + if (atomic_read(&sc->poison) || sc->changed) { + blkdev_dequeue_request(rq); + ub_end_rq(rq, 0); + return 0; + } + + if ((cmd = ub_get_cmd(sc)) == NULL) + return -1; + memset(cmd, 0, sizeof(struct ub_scsi_cmd)); + + blkdev_dequeue_request(rq); + + if (blk_pc_request(rq)) { + rc = ub_cmd_build_packet(sc, cmd, rq); + } else { + rc = ub_cmd_build_block(sc, cmd, rq); + } + if (rc != 0) { + ub_put_cmd(sc, cmd); + ub_end_rq(rq, 0); + blk_start_queue(sc->disk->queue); + return 0; + } + + cmd->state = UB_CMDST_INIT; + cmd->done = ub_rw_cmd_done; + cmd->back = rq; + + cmd->tag = sc->tagcnt++; + if ((rc = ub_submit_scsi(sc, cmd)) != 0) { + ub_put_cmd(sc, cmd); + ub_end_rq(rq, 0); + blk_start_queue(sc->disk->queue); + return 0; + } + + return 0; +} + +static int ub_cmd_build_block(struct ub_dev *sc, struct ub_scsi_cmd *cmd, + struct request *rq) +{ + int ub_dir; +#if 0 /* We use rq->buffer for now */ + struct scatterlist *sg; + int n_elem; +#endif + unsigned int block, nblks; + + if (rq_data_dir(rq) == WRITE) + ub_dir = UB_DIR_WRITE; + else + ub_dir = UB_DIR_READ; + + /* + * get scatterlist from block layer + */ +#if 0 /* We use rq->buffer for now */ + sg = &cmd->sgv[0]; + n_elem = blk_rq_map_sg(q, rq, sg); + if (n_elem <= 0) { + ub_put_cmd(sc, cmd); + ub_end_rq(rq, 0); + blk_start_queue(q); + return 0; /* request with no s/g entries? */ + } + + if (n_elem != 1) { /* Paranoia */ + printk(KERN_WARNING "%s: request with %d segments\n", + sc->name, n_elem); + ub_put_cmd(sc, cmd); + ub_end_rq(rq, 0); + blk_start_queue(q); + return 0; + } +#endif + + /* + * XXX Unfortunately, this check does not work. It is quite possible + * to get bogus non-null rq->buffer if you allow sg by mistake. + */ + if (rq->buffer == NULL) { + /* + * This must not happen if we set the queue right. + * The block level must create bounce buffers for us. + */ + static int do_print = 1; + if (do_print) { + printk(KERN_WARNING "%s: unmapped block request" + " flags 0x%lx sectors %lu\n", + sc->name, rq->flags, rq->nr_sectors); + do_print = 0; + } + return -1; + } + + /* + * build the command + * + * The call to blk_queue_hardsect_size() guarantees that request + * is aligned, but it is given in terms of 512 byte units, always. + */ + block = rq->sector >> sc->capacity.bshift; + nblks = rq->nr_sectors >> sc->capacity.bshift; + + cmd->cdb[0] = (ub_dir == UB_DIR_READ)? READ_10: WRITE_10; + /* 10-byte uses 4 bytes of LBA: 2147483648KB, 2097152MB, 2048GB */ + cmd->cdb[2] = block >> 24; + cmd->cdb[3] = block >> 16; + cmd->cdb[4] = block >> 8; + cmd->cdb[5] = block; + cmd->cdb[7] = nblks >> 8; + cmd->cdb[8] = nblks; + cmd->cdb_len = 10; + + cmd->dir = ub_dir; + cmd->data = rq->buffer; + cmd->len = rq->nr_sectors * 512; + + return 0; +} + +static int ub_cmd_build_packet(struct ub_dev *sc, struct ub_scsi_cmd *cmd, + struct request *rq) +{ + + if (rq->data_len != 0 && rq->data == NULL) { + static int do_print = 1; + if (do_print) { + printk(KERN_WARNING "%s: unmapped packet request" + " flags 0x%lx length %d\n", + sc->name, rq->flags, rq->data_len); + do_print = 0; + } + return -1; + } + + memcpy(&cmd->cdb, rq->cmd, rq->cmd_len); + cmd->cdb_len = rq->cmd_len; + + if (rq->data_len == 0) { + cmd->dir = UB_DIR_NONE; + } else { + if (rq_data_dir(rq) == WRITE) + cmd->dir = UB_DIR_WRITE; + else + cmd->dir = UB_DIR_READ; + } + cmd->data = rq->data; + cmd->len = rq->data_len; + + return 0; +} + +static void ub_rw_cmd_done(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct request *rq = cmd->back; + struct gendisk *disk = sc->disk; + request_queue_t *q = disk->queue; + int uptodate; + + if (blk_pc_request(rq)) { + /* UB_SENSE_SIZE is smaller than SCSI_SENSE_BUFFERSIZE */ + memcpy(rq->sense, sc->top_sense, UB_SENSE_SIZE); + rq->sense_len = UB_SENSE_SIZE; + } + + if (cmd->error == 0) + uptodate = 1; + else + uptodate = 0; + + ub_put_cmd(sc, cmd); + ub_end_rq(rq, uptodate); + blk_start_queue(q); +} + +static void ub_end_rq(struct request *rq, int uptodate) +{ + int rc; + + rc = end_that_request_first(rq, uptodate, rq->hard_nr_sectors); + // assert(rc == 0); + end_that_request_last(rq); +} + +/* + * Submit a regular SCSI operation (not an auto-sense). + * + * The Iron Law of Good Submit Routine is: + * Zero return - callback is done, Nonzero return - callback is not done. + * No exceptions. + * + * Host is assumed locked. + * + * XXX We only support Bulk for the moment. + */ +static int ub_submit_scsi(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + + if (cmd->state != UB_CMDST_INIT || + (cmd->dir != UB_DIR_NONE && cmd->len == 0)) { + return -EINVAL; + } + + ub_cmdq_add(sc, cmd); + /* + * We can call ub_scsi_dispatch(sc) right away here, but it's a little + * safer to jump to a tasklet, in case upper layers do something silly. + */ + tasklet_schedule(&sc->tasklet); + return 0; +} + +/* + * Submit the first URB for the queued command. + * This function does not deal with queueing in any way. + */ +static int ub_scsi_cmd_start(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct bulk_cb_wrap *bcb; + int rc; + + bcb = &sc->work_bcb; + + /* + * ``If the allocation length is eighteen or greater, and a device + * server returns less than eithteen bytes of data, the application + * client should assume that the bytes not transferred would have been + * zeroes had the device server returned those bytes.'' + * + * We zero sense for all commands so that when a packet request + * fails it does not return a stale sense. + */ + memset(&sc->top_sense, 0, UB_SENSE_SIZE); + + /* set up the command wrapper */ + bcb->Signature = cpu_to_le32(US_BULK_CB_SIGN); + bcb->Tag = cmd->tag; /* Endianness is not important */ + bcb->DataTransferLength = cpu_to_le32(cmd->len); + bcb->Flags = (cmd->dir == UB_DIR_READ) ? 0x80 : 0; + bcb->Lun = 0; /* No multi-LUN yet */ + bcb->Length = cmd->cdb_len; + + /* copy the command payload */ + memcpy(bcb->CDB, cmd->cdb, UB_MAX_CDB_SIZE); + + UB_INIT_COMPLETION(sc->work_done); + + sc->last_pipe = sc->send_bulk_pipe; + usb_fill_bulk_urb(&sc->work_urb, sc->dev, sc->send_bulk_pipe, + bcb, US_BULK_CB_WRAP_LEN, ub_urb_complete, sc); + sc->work_urb.transfer_flags = URB_ASYNC_UNLINK; + + /* Fill what we shouldn't be filling, because usb-storage did so. */ + sc->work_urb.actual_length = 0; + sc->work_urb.error_count = 0; + sc->work_urb.status = 0; + + if ((rc = usb_submit_urb(&sc->work_urb, GFP_ATOMIC)) != 0) { + /* XXX Clear stalls */ + printk("ub: cmd #%d start failed (%d)\n", cmd->tag, rc); /* P3 */ + ub_complete(&sc->work_done); + return rc; + } + + sc->work_timer.expires = jiffies + UB_URB_TIMEOUT; + add_timer(&sc->work_timer); + + cmd->state = UB_CMDST_CMD; + ub_cmdtr_state(sc, cmd); + return 0; +} + +/* + * Timeout handler. + */ +static void ub_urb_timeout(unsigned long arg) +{ + struct ub_dev *sc = (struct ub_dev *) arg; + unsigned long flags; + + spin_lock_irqsave(&sc->lock, flags); + usb_unlink_urb(&sc->work_urb); + spin_unlock_irqrestore(&sc->lock, flags); +} + +/* + * Completion routine for the work URB. + * + * This can be called directly from usb_submit_urb (while we have + * the sc->lock taken) and from an interrupt (while we do NOT have + * the sc->lock taken). Therefore, bounce this off to a tasklet. + */ +static void ub_urb_complete(struct urb *urb, struct pt_regs *pt) +{ + struct ub_dev *sc = urb->context; + + ub_complete(&sc->work_done); + tasklet_schedule(&sc->tasklet); +} + +static void ub_scsi_action(unsigned long _dev) +{ + struct ub_dev *sc = (struct ub_dev *) _dev; + unsigned long flags; + + spin_lock_irqsave(&sc->lock, flags); + del_timer(&sc->work_timer); + ub_scsi_dispatch(sc); + spin_unlock_irqrestore(&sc->lock, flags); +} + +static void ub_scsi_dispatch(struct ub_dev *sc) +{ + struct ub_scsi_cmd *cmd; + int rc; + + while ((cmd = ub_cmdq_peek(sc)) != NULL) { + if (cmd->state == UB_CMDST_DONE) { + ub_cmdq_pop(sc); + (*cmd->done)(sc, cmd); + } else if (cmd->state == UB_CMDST_INIT) { + ub_cmdtr_new(sc, cmd); + if ((rc = ub_scsi_cmd_start(sc, cmd)) == 0) + break; + cmd->error = rc; + cmd->state = UB_CMDST_DONE; + ub_cmdtr_state(sc, cmd); + } else { + if (!ub_is_completed(&sc->work_done)) + break; + ub_scsi_urb_compl(sc, cmd); + } + } +} + +static void ub_scsi_urb_compl(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct urb *urb = &sc->work_urb; + struct bulk_cs_wrap *bcs; + int pipe; + int rc; + + if (atomic_read(&sc->poison)) { + /* A little too simplistic, I feel... */ + goto Bad_End; + } + + if (cmd->state == UB_CMDST_CLEAR) { + if (urb->status == -EPIPE) { + /* + * STALL while clearning STALL. + * The control pipe clears itself - nothing to do. + * XXX Might try to reset the device here and retry. + */ + printk(KERN_NOTICE "%s: " + "stall on control pipe for device %u\n", + sc->name, sc->dev->devnum); + goto Bad_End; + } + + /* + * We ignore the result for the halt clear. + */ + + /* reset the endpoint toggle */ + usb_settoggle(sc->dev, usb_pipeendpoint(sc->last_pipe), + usb_pipeout(sc->last_pipe), 0); + + ub_state_sense(sc, cmd); + + } else if (cmd->state == UB_CMDST_CLR2STS) { + if (urb->status == -EPIPE) { + /* + * STALL while clearning STALL. + * The control pipe clears itself - nothing to do. + * XXX Might try to reset the device here and retry. + */ + printk(KERN_NOTICE "%s: " + "stall on control pipe for device %u\n", + sc->name, sc->dev->devnum); + goto Bad_End; + } + + /* + * We ignore the result for the halt clear. + */ + + /* reset the endpoint toggle */ + usb_settoggle(sc->dev, usb_pipeendpoint(sc->last_pipe), + usb_pipeout(sc->last_pipe), 0); + + ub_state_stat(sc, cmd); + + } else if (cmd->state == UB_CMDST_CMD) { + if (urb->status == -EPIPE) { + rc = ub_submit_clear_stall(sc, cmd, sc->last_pipe); + if (rc != 0) { + printk(KERN_NOTICE "%s: " + "unable to submit clear for device %u" + " (code %d)\n", + sc->name, sc->dev->devnum, rc); + /* + * This is typically ENOMEM or some other such shit. + * Retrying is pointless. Just do Bad End on it... + */ + goto Bad_End; + } + cmd->state = UB_CMDST_CLEAR; + ub_cmdtr_state(sc, cmd); + return; + } + if (urb->status != 0) { + printk("ub: cmd #%d cmd status (%d)\n", cmd->tag, urb->status); /* P3 */ + goto Bad_End; + } + if (urb->actual_length != US_BULK_CB_WRAP_LEN) { + printk("ub: cmd #%d xferred %d\n", cmd->tag, urb->actual_length); /* P3 */ + /* XXX Must do reset here to unconfuse the device */ + goto Bad_End; + } + + if (cmd->dir == UB_DIR_NONE) { + ub_state_stat(sc, cmd); + return; + } + + UB_INIT_COMPLETION(sc->work_done); + + if (cmd->dir == UB_DIR_READ) + pipe = sc->recv_bulk_pipe; + else + pipe = sc->send_bulk_pipe; + sc->last_pipe = pipe; + usb_fill_bulk_urb(&sc->work_urb, sc->dev, pipe, + cmd->data, cmd->len, ub_urb_complete, sc); + sc->work_urb.transfer_flags = URB_ASYNC_UNLINK; + sc->work_urb.actual_length = 0; + sc->work_urb.error_count = 0; + sc->work_urb.status = 0; + + if ((rc = usb_submit_urb(&sc->work_urb, GFP_ATOMIC)) != 0) { + /* XXX Clear stalls */ + printk("ub: data #%d submit failed (%d)\n", cmd->tag, rc); /* P3 */ + ub_complete(&sc->work_done); + ub_state_done(sc, cmd, rc); + return; + } + + sc->work_timer.expires = jiffies + UB_DATA_TIMEOUT; + add_timer(&sc->work_timer); + + cmd->state = UB_CMDST_DATA; + ub_cmdtr_state(sc, cmd); + + } else if (cmd->state == UB_CMDST_DATA) { + if (urb->status == -EPIPE) { + rc = ub_submit_clear_stall(sc, cmd, sc->last_pipe); + if (rc != 0) { + printk(KERN_NOTICE "%s: " + "unable to submit clear for device %u" + " (code %d)\n", + sc->name, sc->dev->devnum, rc); + /* + * This is typically ENOMEM or some other such shit. + * Retrying is pointless. Just do Bad End on it... + */ + goto Bad_End; + } + cmd->state = UB_CMDST_CLR2STS; + ub_cmdtr_state(sc, cmd); + return; + } + if (urb->status == -EOVERFLOW) { + /* + * A babble? Failure, but we must transfer CSW now. + */ + cmd->error = -EOVERFLOW; /* A cheap trick... */ + } else { + if (urb->status != 0) + goto Bad_End; + } + + cmd->act_len = urb->actual_length; + ub_cmdtr_act_len(sc, cmd); + + ub_state_stat(sc, cmd); + + } else if (cmd->state == UB_CMDST_STAT) { + if (urb->status == -EPIPE) { + rc = ub_submit_clear_stall(sc, cmd, sc->last_pipe); + if (rc != 0) { + printk(KERN_NOTICE "%s: " + "unable to submit clear for device %u" + " (code %d)\n", + sc->name, sc->dev->devnum, rc); + /* + * This is typically ENOMEM or some other such shit. + * Retrying is pointless. Just do Bad End on it... + */ + goto Bad_End; + } + cmd->state = UB_CMDST_CLEAR; + ub_cmdtr_state(sc, cmd); + return; + } + if (urb->status != 0) + goto Bad_End; + + if (urb->actual_length == 0) { + /* + * Some broken devices add unnecessary zero-length + * packets to the end of their data transfers. + * Such packets show up as 0-length CSWs. If we + * encounter such a thing, try to read the CSW again. + */ + if (++cmd->stat_count >= 4) { + printk(KERN_NOTICE "%s: " + "unable to get CSW on device %u\n", + sc->name, sc->dev->devnum); + goto Bad_End; + } + __ub_state_stat(sc, cmd); + return; + } + + /* + * Check the returned Bulk protocol status. + */ + + bcs = &sc->work_bcs; + rc = le32_to_cpu(bcs->Residue); + if (rc != cmd->len - cmd->act_len) { + /* + * It is all right to transfer less, the caller has + * to check. But it's not all right if the device + * counts disagree with our counts. + */ + /* P3 */ printk("%s: resid %d len %d act %d\n", + sc->name, rc, cmd->len, cmd->act_len); + goto Bad_End; + } + +#if 0 + if (bcs->Signature != cpu_to_le32(US_BULK_CS_SIGN) && + bcs->Signature != cpu_to_le32(US_BULK_CS_OLYMPUS_SIGN)) { + /* Windows ignores signatures, so do we. */ + } +#endif + + if (bcs->Tag != cmd->tag) { + /* + * This usually happens when we disagree with the + * device's microcode about something. For instance, + * a few of them throw this after timeouts. They buffer + * commands and reply at commands we timed out before. + * Without flushing these replies we loop forever. + */ + if (++cmd->stat_count >= 4) { + printk(KERN_NOTICE "%s: " + "tag mismatch orig 0x%x reply 0x%x " + "on device %u\n", + sc->name, cmd->tag, bcs->Tag, + sc->dev->devnum); + goto Bad_End; + } + __ub_state_stat(sc, cmd); + return; + } + + switch (bcs->Status) { + case US_BULK_STAT_OK: + break; + case US_BULK_STAT_FAIL: + ub_state_sense(sc, cmd); + return; + case US_BULK_STAT_PHASE: + /* XXX We must reset the transport here */ + /* P3 */ printk("%s: status PHASE\n", sc->name); + goto Bad_End; + default: + printk(KERN_INFO "%s: unknown CSW status 0x%x\n", + sc->name, bcs->Status); + goto Bad_End; + } + + /* Not zeroing error to preserve a babble indicator */ + cmd->state = UB_CMDST_DONE; + ub_cmdtr_state(sc, cmd); + ub_cmdq_pop(sc); + (*cmd->done)(sc, cmd); + + } else if (cmd->state == UB_CMDST_SENSE) { + ub_state_done(sc, cmd, -EIO); + + } else { + printk(KERN_WARNING "%s: " + "wrong command state %d on device %u\n", + sc->name, cmd->state, sc->dev->devnum); + goto Bad_End; + } + return; + +Bad_End: /* Little Excel is dead */ + ub_state_done(sc, cmd, -EIO); +} + +/* + * Factorization helper for the command state machine: + * Finish the command. + */ +static void ub_state_done(struct ub_dev *sc, struct ub_scsi_cmd *cmd, int rc) +{ + + cmd->error = rc; + cmd->state = UB_CMDST_DONE; + ub_cmdtr_state(sc, cmd); + ub_cmdq_pop(sc); + (*cmd->done)(sc, cmd); +} + +/* + * Factorization helper for the command state machine: + * Submit a CSW read. + */ +static void __ub_state_stat(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + int rc; + + UB_INIT_COMPLETION(sc->work_done); + + sc->last_pipe = sc->recv_bulk_pipe; + usb_fill_bulk_urb(&sc->work_urb, sc->dev, sc->recv_bulk_pipe, + &sc->work_bcs, US_BULK_CS_WRAP_LEN, ub_urb_complete, sc); + sc->work_urb.transfer_flags = URB_ASYNC_UNLINK; + sc->work_urb.actual_length = 0; + sc->work_urb.error_count = 0; + sc->work_urb.status = 0; + + if ((rc = usb_submit_urb(&sc->work_urb, GFP_ATOMIC)) != 0) { + /* XXX Clear stalls */ + printk("%s: CSW #%d submit failed (%d)\n", sc->name, cmd->tag, rc); /* P3 */ + ub_complete(&sc->work_done); + ub_state_done(sc, cmd, rc); + return; + } + + sc->work_timer.expires = jiffies + UB_STAT_TIMEOUT; + add_timer(&sc->work_timer); +} + +/* + * Factorization helper for the command state machine: + * Submit a CSW read and go to STAT state. + */ +static void ub_state_stat(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + __ub_state_stat(sc, cmd); + + cmd->stat_count = 0; + cmd->state = UB_CMDST_STAT; + ub_cmdtr_state(sc, cmd); +} + +/* + * Factorization helper for the command state machine: + * Submit a REQUEST SENSE and go to SENSE state. + */ +static void ub_state_sense(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct ub_scsi_cmd *scmd; + int rc; + + if (cmd->cdb[0] == REQUEST_SENSE) { + rc = -EPIPE; + goto error; + } + + scmd = &sc->top_rqs_cmd; + scmd->cdb[0] = REQUEST_SENSE; + scmd->cdb[4] = UB_SENSE_SIZE; + scmd->cdb_len = 6; + scmd->dir = UB_DIR_READ; + scmd->state = UB_CMDST_INIT; + scmd->data = sc->top_sense; + scmd->len = UB_SENSE_SIZE; + scmd->done = ub_top_sense_done; + scmd->back = cmd; + + scmd->tag = sc->tagcnt++; + + cmd->state = UB_CMDST_SENSE; + ub_cmdtr_state(sc, cmd); + + ub_cmdq_insert(sc, scmd); + return; + +error: + ub_state_done(sc, cmd, rc); +} + +/* + * A helper for the command's state machine: + * Submit a stall clear. + */ +static int ub_submit_clear_stall(struct ub_dev *sc, struct ub_scsi_cmd *cmd, + int stalled_pipe) +{ + int endp; + struct usb_ctrlrequest *cr; + int rc; + + endp = usb_pipeendpoint(stalled_pipe); + if (usb_pipein (stalled_pipe)) + endp |= USB_DIR_IN; + + cr = &sc->work_cr; + cr->bRequestType = USB_RECIP_ENDPOINT; + cr->bRequest = USB_REQ_CLEAR_FEATURE; + cr->wValue = cpu_to_le16(USB_ENDPOINT_HALT); + cr->wIndex = cpu_to_le16(endp); + cr->wLength = cpu_to_le16(0); + + UB_INIT_COMPLETION(sc->work_done); + + usb_fill_control_urb(&sc->work_urb, sc->dev, sc->send_ctrl_pipe, + (unsigned char*) cr, NULL, 0, ub_urb_complete, sc); + sc->work_urb.transfer_flags = URB_ASYNC_UNLINK; + sc->work_urb.actual_length = 0; + sc->work_urb.error_count = 0; + sc->work_urb.status = 0; + + if ((rc = usb_submit_urb(&sc->work_urb, GFP_ATOMIC)) != 0) { + ub_complete(&sc->work_done); + return rc; + } + + sc->work_timer.expires = jiffies + UB_CTRL_TIMEOUT; + add_timer(&sc->work_timer); + return 0; +} + +/* + */ +static void ub_top_sense_done(struct ub_dev *sc, struct ub_scsi_cmd *scmd) +{ + unsigned char *sense = scmd->data; + struct ub_scsi_cmd *cmd; + + /* + * Ignoring scmd->act_len, because the buffer was pre-zeroed. + */ + ub_cmdtr_sense(sc, scmd, sense); + + /* + * Find the command which triggered the unit attention or a check, + * save the sense into it, and advance its state machine. + */ + if ((cmd = ub_cmdq_peek(sc)) == NULL) { + printk(KERN_WARNING "%s: sense done while idle\n", sc->name); + return; + } + if (cmd != scmd->back) { + printk(KERN_WARNING "%s: " + "sense done for wrong command 0x%x on device %u\n", + sc->name, cmd->tag, sc->dev->devnum); + return; + } + if (cmd->state != UB_CMDST_SENSE) { + printk(KERN_WARNING "%s: " + "sense done with bad cmd state %d on device %u\n", + sc->name, cmd->state, sc->dev->devnum); + return; + } + + cmd->key = sense[2] & 0x0F; + cmd->asc = sense[12]; + cmd->ascq = sense[13]; + + ub_scsi_urb_compl(sc, cmd); +} + +#if 0 +/* Determine what the maximum LUN supported is */ +int usb_stor_Bulk_max_lun(struct us_data *us) +{ + int result; + + /* issue the command */ + result = usb_stor_control_msg(us, us->recv_ctrl_pipe, + US_BULK_GET_MAX_LUN, + USB_DIR_IN | USB_TYPE_CLASS | + USB_RECIP_INTERFACE, + 0, us->ifnum, us->iobuf, 1, HZ); + + /* + * Some devices (i.e. Iomega Zip100) need this -- apparently + * the bulk pipes get STALLed when the GetMaxLUN request is + * processed. This is, in theory, harmless to all other devices + * (regardless of if they stall or not). + */ + if (result < 0) { + usb_stor_clear_halt(us, us->recv_bulk_pipe); + usb_stor_clear_halt(us, us->send_bulk_pipe); + } + + US_DEBUGP("GetMaxLUN command result is %d, data is %d\n", + result, us->iobuf[0]); + + /* if we have a successful request, return the result */ + if (result == 1) + return us->iobuf[0]; + + /* return the default -- no LUNs */ + return 0; +} +#endif + +/* + * This is called from a process context. + */ +static void ub_revalidate(struct ub_dev *sc) +{ + + sc->readonly = 0; /* XXX Query this from the device */ + + sc->capacity.nsec = 0; + sc->capacity.bsize = 512; + sc->capacity.bshift = 0; + + if (ub_sync_tur(sc) != 0) + return; /* Not ready */ + sc->changed = 0; + + if (ub_sync_read_cap(sc, &sc->capacity) != 0) { + /* + * The retry here means something is wrong, either with the + * device, with the transport, or with our code. + * We keep this because sd.c has retries for capacity. + */ + if (ub_sync_read_cap(sc, &sc->capacity) != 0) { + sc->capacity.nsec = 0; + sc->capacity.bsize = 512; + sc->capacity.bshift = 0; + } + } +} + +/* + * The open funcion. + * This is mostly needed to keep refcounting, but also to support + * media checks on removable media drives. + */ +static int ub_bd_open(struct inode *inode, struct file *filp) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + struct ub_dev *sc; + unsigned long flags; + int rc; + + if ((sc = disk->private_data) == NULL) + return -ENXIO; + spin_lock_irqsave(&ub_lock, flags); + if (atomic_read(&sc->poison)) { + spin_unlock_irqrestore(&ub_lock, flags); + return -ENXIO; + } + sc->openc++; + spin_unlock_irqrestore(&ub_lock, flags); + + /* + * This is a workaround for a specific problem in our block layer. + * In 2.6.9, register_disk duplicates the code from rescan_partitions. + * However, if we do add_disk with a device which persistently reports + * a changed media, add_disk calls register_disk, which does do_open, + * which will call rescan_paritions for changed media. After that, + * register_disk attempts to do it all again and causes double kobject + * registration and a eventually an oops on module removal. + * + * The bottom line is, Al Viro says that we should not allow + * bdev->bd_invalidated to be set when doing add_disk no matter what. + */ + if (sc->first_open) { + if (sc->changed) { + sc->first_open = 0; + rc = -ENOMEDIUM; + goto err_open; + } + } + + if (sc->removable || sc->readonly) + check_disk_change(inode->i_bdev); + + /* + * The sd.c considers ->media_present and ->changed not equivalent, + * under some pretty murky conditions (a failure of READ CAPACITY). + * We may need it one day. + */ + if (sc->removable && sc->changed && !(filp->f_flags & O_NDELAY)) { + rc = -ENOMEDIUM; + goto err_open; + } + + if (sc->readonly && (filp->f_mode & FMODE_WRITE)) { + rc = -EROFS; + goto err_open; + } + + return 0; + +err_open: + ub_put(sc); + return rc; +} + +/* + */ +static int ub_bd_release(struct inode *inode, struct file *filp) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + struct ub_dev *sc = disk->private_data; + + ub_put(sc); + return 0; +} + +/* + * The ioctl interface. + */ +static int ub_bd_ioctl(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg) +{ + struct gendisk *disk = inode->i_bdev->bd_disk; + void __user *usermem = (void __user *) arg; + + return scsi_cmd_ioctl(filp, disk, cmd, usermem); +} + +/* + * This is called once a new disk was seen by the block layer or by ub_probe(). + * The main onjective here is to discover the features of the media such as + * the capacity, read-only status, etc. USB storage generally does not + * need to be spun up, but if we needed it, this would be the place. + * + * This call can sleep. + * + * The return code is not used. + */ +static int ub_bd_revalidate(struct gendisk *disk) +{ + struct ub_dev *sc = disk->private_data; + + ub_revalidate(sc); + /* This is pretty much a long term P3 */ + if (!atomic_read(&sc->poison)) { /* Cover sc->dev */ + printk(KERN_INFO "%s: device %u capacity nsec %ld bsize %u\n", + sc->name, sc->dev->devnum, + sc->capacity.nsec, sc->capacity.bsize); + } + + /* XXX Support sector size switching like in sr.c */ + blk_queue_hardsect_size(disk->queue, sc->capacity.bsize); + set_capacity(disk, sc->capacity.nsec); + // set_disk_ro(sdkp->disk, sc->readonly); + + return 0; +} + +/* + * The check is called by the block layer to verify if the media + * is still available. It is supposed to be harmless, lightweight and + * non-intrusive in case the media was not changed. + * + * This call can sleep. + * + * The return code is bool! + */ +static int ub_bd_media_changed(struct gendisk *disk) +{ + struct ub_dev *sc = disk->private_data; + + if (!sc->removable) + return 0; + + /* + * We clean checks always after every command, so this is not + * as dangerous as it looks. If the TEST_UNIT_READY fails here, + * the device is actually not ready with operator or software + * intervention required. One dangerous item might be a drive which + * spins itself down, and come the time to write dirty pages, this + * will fail, then block layer discards the data. Since we never + * spin drives up, such devices simply cannot be used with ub anyway. + */ + if (ub_sync_tur(sc) != 0) { + sc->changed = 1; + return 1; + } + + return sc->changed; +} + +static struct block_device_operations ub_bd_fops = { + .owner = THIS_MODULE, + .open = ub_bd_open, + .release = ub_bd_release, + .ioctl = ub_bd_ioctl, + .media_changed = ub_bd_media_changed, + .revalidate_disk = ub_bd_revalidate, +}; + +/* + * Common ->done routine for commands executed synchronously. + */ +static void ub_probe_done(struct ub_dev *sc, struct ub_scsi_cmd *cmd) +{ + struct completion *cop = cmd->back; + complete(cop); +} + +/* + * Test if the device has a check condition on it, synchronously. + */ +static int ub_sync_tur(struct ub_dev *sc) +{ + struct ub_scsi_cmd *cmd; + enum { ALLOC_SIZE = sizeof(struct ub_scsi_cmd) }; + unsigned long flags; + struct completion compl; + int rc; + + init_completion(&compl); + + rc = -ENOMEM; + if ((cmd = kmalloc(ALLOC_SIZE, GFP_KERNEL)) == NULL) + goto err_alloc; + memset(cmd, 0, ALLOC_SIZE); + + cmd->cdb[0] = TEST_UNIT_READY; + cmd->cdb_len = 6; + cmd->dir = UB_DIR_NONE; + cmd->state = UB_CMDST_INIT; + cmd->done = ub_probe_done; + cmd->back = &compl; + + spin_lock_irqsave(&sc->lock, flags); + cmd->tag = sc->tagcnt++; + + rc = ub_submit_scsi(sc, cmd); + spin_unlock_irqrestore(&sc->lock, flags); + + if (rc != 0) { + printk("ub: testing ready: submit error (%d)\n", rc); /* P3 */ + goto err_submit; + } + + wait_for_completion(&compl); + + rc = cmd->error; + + if (rc == -EIO && cmd->key != 0) /* Retries for benh's key */ + rc = cmd->key; + +err_submit: + kfree(cmd); +err_alloc: + return rc; +} + +/* + * Read the SCSI capacity synchronously (for probing). + */ +static int ub_sync_read_cap(struct ub_dev *sc, struct ub_capacity *ret) +{ + struct ub_scsi_cmd *cmd; + char *p; + enum { ALLOC_SIZE = sizeof(struct ub_scsi_cmd) + 8 }; + unsigned long flags; + unsigned int bsize, shift; + unsigned long nsec; + struct completion compl; + int rc; + + init_completion(&compl); + + rc = -ENOMEM; + if ((cmd = kmalloc(ALLOC_SIZE, GFP_KERNEL)) == NULL) + goto err_alloc; + memset(cmd, 0, ALLOC_SIZE); + p = (char *)cmd + sizeof(struct ub_scsi_cmd); + + cmd->cdb[0] = 0x25; + cmd->cdb_len = 10; + cmd->dir = UB_DIR_READ; + cmd->state = UB_CMDST_INIT; + cmd->data = p; + cmd->len = 8; + cmd->done = ub_probe_done; + cmd->back = &compl; + + spin_lock_irqsave(&sc->lock, flags); + cmd->tag = sc->tagcnt++; + + rc = ub_submit_scsi(sc, cmd); + spin_unlock_irqrestore(&sc->lock, flags); + + if (rc != 0) { + printk("ub: reading capacity: submit error (%d)\n", rc); /* P3 */ + goto err_submit; + } + + wait_for_completion(&compl); + + if (cmd->error != 0) { + printk("ub: reading capacity: error %d\n", cmd->error); /* P3 */ + rc = -EIO; + goto err_read; + } + if (cmd->act_len != 8) { + printk("ub: reading capacity: size %d\n", cmd->act_len); /* P3 */ + rc = -EIO; + goto err_read; + } + + /* sd.c special-cases sector size of 0 to mean 512. Needed? Safe? */ + nsec = be32_to_cpu(*(__be32 *)p) + 1; + bsize = be32_to_cpu(*(__be32 *)(p + 4)); + switch (bsize) { + case 512: shift = 0; break; + case 1024: shift = 1; break; + case 2048: shift = 2; break; + case 4096: shift = 3; break; + default: + printk("ub: Bad sector size %u\n", bsize); /* P3 */ + rc = -EDOM; + goto err_inv_bsize; + } + + ret->bsize = bsize; + ret->bshift = shift; + ret->nsec = nsec << shift; + rc = 0; + +err_inv_bsize: +err_read: +err_submit: + kfree(cmd); +err_alloc: + return rc; +} + +/* + */ +static void ub_probe_urb_complete(struct urb *urb, struct pt_regs *pt) +{ + struct completion *cop = urb->context; + complete(cop); +} + +static void ub_probe_timeout(unsigned long arg) +{ + struct completion *cop = (struct completion *) arg; + complete(cop); +} + +/* + * Clear initial stalls. + */ +static int ub_probe_clear_stall(struct ub_dev *sc, int stalled_pipe) +{ + int endp; + struct usb_ctrlrequest *cr; + struct completion compl; + struct timer_list timer; + int rc; + + init_completion(&compl); + + endp = usb_pipeendpoint(stalled_pipe); + if (usb_pipein (stalled_pipe)) + endp |= USB_DIR_IN; + + cr = &sc->work_cr; + cr->bRequestType = USB_RECIP_ENDPOINT; + cr->bRequest = USB_REQ_CLEAR_FEATURE; + cr->wValue = cpu_to_le16(USB_ENDPOINT_HALT); + cr->wIndex = cpu_to_le16(endp); + cr->wLength = cpu_to_le16(0); + + usb_fill_control_urb(&sc->work_urb, sc->dev, sc->send_ctrl_pipe, + (unsigned char*) cr, NULL, 0, ub_probe_urb_complete, &compl); + sc->work_urb.transfer_flags = 0; + sc->work_urb.actual_length = 0; + sc->work_urb.error_count = 0; + sc->work_urb.status = 0; + + if ((rc = usb_submit_urb(&sc->work_urb, GFP_KERNEL)) != 0) { + printk(KERN_WARNING + "%s: Unable to submit a probe clear (%d)\n", sc->name, rc); + return rc; + } + + init_timer(&timer); + timer.function = ub_probe_timeout; + timer.data = (unsigned long) &compl; + timer.expires = jiffies + UB_CTRL_TIMEOUT; + add_timer(&timer); + + wait_for_completion(&compl); + + del_timer_sync(&timer); + usb_kill_urb(&sc->work_urb); + + /* reset the endpoint toggle */ + usb_settoggle(sc->dev, endp, usb_pipeout(sc->last_pipe), 0); + + return 0; +} + +/* + * Get the pipe settings. + */ +static int ub_get_pipes(struct ub_dev *sc, struct usb_device *dev, + struct usb_interface *intf) +{ + struct usb_host_interface *altsetting = intf->cur_altsetting; + struct usb_endpoint_descriptor *ep_in = NULL; + struct usb_endpoint_descriptor *ep_out = NULL; + struct usb_endpoint_descriptor *ep; + int i; + + /* + * Find the endpoints we need. + * We are expecting a minimum of 2 endpoints - in and out (bulk). + * We will ignore any others. + */ + for (i = 0; i < altsetting->desc.bNumEndpoints; i++) { + ep = &altsetting->endpoint[i].desc; + + /* Is it a BULK endpoint? */ + if ((ep->bmAttributes & USB_ENDPOINT_XFERTYPE_MASK) + == USB_ENDPOINT_XFER_BULK) { + /* BULK in or out? */ + if (ep->bEndpointAddress & USB_DIR_IN) + ep_in = ep; + else + ep_out = ep; + } + } + + if (ep_in == NULL || ep_out == NULL) { + printk(KERN_NOTICE "%s: device %u failed endpoint check\n", + sc->name, sc->dev->devnum); + return -EIO; + } + + /* Calculate and store the pipe values */ + sc->send_ctrl_pipe = usb_sndctrlpipe(dev, 0); + sc->recv_ctrl_pipe = usb_rcvctrlpipe(dev, 0); + sc->send_bulk_pipe = usb_sndbulkpipe(dev, + ep_out->bEndpointAddress & USB_ENDPOINT_NUMBER_MASK); + sc->recv_bulk_pipe = usb_rcvbulkpipe(dev, + ep_in->bEndpointAddress & USB_ENDPOINT_NUMBER_MASK); + + return 0; +} + +/* + * Probing is done in the process context, which allows us to cheat + * and not to build a state machine for the discovery. + */ +static int ub_probe(struct usb_interface *intf, + const struct usb_device_id *dev_id) +{ + struct ub_dev *sc; + request_queue_t *q; + struct gendisk *disk; + int rc; + int i; + + rc = -ENOMEM; + if ((sc = kmalloc(sizeof(struct ub_dev), GFP_KERNEL)) == NULL) + goto err_core; + memset(sc, 0, sizeof(struct ub_dev)); + spin_lock_init(&sc->lock); + usb_init_urb(&sc->work_urb); + tasklet_init(&sc->tasklet, ub_scsi_action, (unsigned long)sc); + atomic_set(&sc->poison, 0); + + init_timer(&sc->work_timer); + sc->work_timer.data = (unsigned long) sc; + sc->work_timer.function = ub_urb_timeout; + + ub_init_completion(&sc->work_done); + sc->work_done.done = 1; /* A little yuk, but oh well... */ + + rc = -ENOSR; + if ((sc->id = ub_id_get()) == -1) + goto err_id; + snprintf(sc->name, 8, DRV_NAME "%c", sc->id + 'a'); + + sc->dev = interface_to_usbdev(intf); + sc->intf = intf; + // sc->ifnum = intf->cur_altsetting->desc.bInterfaceNumber; + + usb_set_intfdata(intf, sc); + usb_get_dev(sc->dev); + // usb_get_intf(sc->intf); /* Do we need this? */ + + /* XXX Verify that we can handle the device (from descriptors) */ + + ub_get_pipes(sc, sc->dev, intf); + + if (device_create_file(&sc->intf->dev, &dev_attr_diag) != 0) + goto err_diag; + + /* + * At this point, all USB initialization is done, do upper layer. + * We really hate halfway initialized structures, so from the + * invariants perspective, this ub_dev is fully constructed at + * this point. + */ + + /* + * This is needed to clear toggles. It is a problem only if we do + * `rmmod ub && modprobe ub` without disconnects, but we like that. + */ + ub_probe_clear_stall(sc, sc->recv_bulk_pipe); + ub_probe_clear_stall(sc, sc->send_bulk_pipe); + + /* + * The way this is used by the startup code is a little specific. + * A SCSI check causes a USB stall. Our common case code sees it + * and clears the check, after which the device is ready for use. + * But if a check was not present, any command other than + * TEST_UNIT_READY ends with a lockup (including REQUEST_SENSE). + * + * If we neglect to clear the SCSI check, the first real command fails + * (which is the capacity readout). We clear that and retry, but why + * causing spurious retries for no reason. + * + * Revalidation may start with its own TEST_UNIT_READY, but that one + * has to succeed, so we clear checks with an additional one here. + * In any case it's not our business how revaliadation is implemented. + */ + for (i = 0; i < 3; i++) { /* Retries for benh's key */ + if ((rc = ub_sync_tur(sc)) <= 0) break; + if (rc != 0x6) break; + msleep(10); + } + + sc->removable = 1; /* XXX Query this from the device */ + sc->changed = 1; /* ub_revalidate clears only */ + sc->first_open = 1; + + ub_revalidate(sc); + /* This is pretty much a long term P3 */ + printk(KERN_INFO "%s: device %u capacity nsec %ld bsize %u\n", + sc->name, sc->dev->devnum, sc->capacity.nsec, sc->capacity.bsize); + + /* + * Just one disk per sc currently, but maybe more. + */ + rc = -ENOMEM; + if ((disk = alloc_disk(UB_MINORS_PER_MAJOR)) == NULL) + goto err_diskalloc; + + sc->disk = disk; + sprintf(disk->disk_name, DRV_NAME "%c", sc->id + 'a'); + sprintf(disk->devfs_name, DEVFS_NAME "/%c", sc->id + 'a'); + disk->major = UB_MAJOR; + disk->first_minor = sc->id * UB_MINORS_PER_MAJOR; + disk->fops = &ub_bd_fops; + disk->private_data = sc; + disk->driverfs_dev = &intf->dev; + + rc = -ENOMEM; + if ((q = blk_init_queue(ub_bd_rq_fn, &sc->lock)) == NULL) + goto err_blkqinit; + + disk->queue = q; + + // blk_queue_bounce_limit(q, hba[i]->pdev->dma_mask); + blk_queue_max_hw_segments(q, UB_MAX_REQ_SG); + blk_queue_max_phys_segments(q, UB_MAX_REQ_SG); + // blk_queue_segment_boundary(q, CARM_SG_BOUNDARY); + blk_queue_max_sectors(q, UB_MAX_SECTORS); + blk_queue_hardsect_size(q, sc->capacity.bsize); + + /* + * This is a serious infraction, caused by a deficiency in the + * USB sg interface (usb_sg_wait()). We plan to remove this once + * we get mileage on the driver and can justify a change to USB API. + * See blk_queue_bounce_limit() to understand this part. + * + * XXX And I still need to be aware of the DMA mask in the HC. + */ + q->bounce_pfn = blk_max_low_pfn; + q->bounce_gfp = GFP_NOIO; + + q->queuedata = sc; + + set_capacity(disk, sc->capacity.nsec); + if (sc->removable) + disk->flags |= GENHD_FL_REMOVABLE; + + add_disk(disk); + + return 0; + +err_blkqinit: + put_disk(disk); +err_diskalloc: + device_remove_file(&sc->intf->dev, &dev_attr_diag); +err_diag: + usb_set_intfdata(intf, NULL); + // usb_put_intf(sc->intf); + usb_put_dev(sc->dev); + ub_id_put(sc->id); +err_id: + kfree(sc); +err_core: + return rc; +} + +static void ub_disconnect(struct usb_interface *intf) +{ + struct ub_dev *sc = usb_get_intfdata(intf); + struct gendisk *disk = sc->disk; + unsigned long flags; + + /* + * Prevent ub_bd_release from pulling the rug from under us. + * XXX This is starting to look like a kref. + * XXX Why not to take this ref at probe time? + */ + spin_lock_irqsave(&ub_lock, flags); + sc->openc++; + spin_unlock_irqrestore(&ub_lock, flags); + + /* + * Fence stall clearnings, operations triggered by unlinkings and so on. + * We do not attempt to unlink any URBs, because we do not trust the + * unlink paths in HC drivers. Also, we get -84 upon disconnect anyway. + */ + atomic_set(&sc->poison, 1); + + /* + * Blow away queued commands. + * + * Actually, this never works, because before we get here + * the HCD terminates outstanding URB(s). It causes our + * SCSI command queue to advance, commands fail to submit, + * and the whole queue drains. So, we just use this code to + * print warnings. + */ + spin_lock_irqsave(&sc->lock, flags); + { + struct ub_scsi_cmd *cmd; + int cnt = 0; + while ((cmd = ub_cmdq_pop(sc)) != NULL) { + cmd->error = -ENOTCONN; + cmd->state = UB_CMDST_DONE; + ub_cmdtr_state(sc, cmd); + ub_cmdq_pop(sc); + (*cmd->done)(sc, cmd); + cnt++; + } + if (cnt != 0) { + printk(KERN_WARNING "%s: " + "%d was queued after shutdown\n", sc->name, cnt); + } + } + spin_unlock_irqrestore(&sc->lock, flags); + + /* + * Unregister the upper layer. + */ + if (disk->flags & GENHD_FL_UP) + del_gendisk(disk); + /* + * I wish I could do: + * set_bit(QUEUE_FLAG_DEAD, &q->queue_flags); + * As it is, we rely on our internal poisoning and let + * the upper levels to spin furiously failing all the I/O. + */ + + /* + * Taking a lock on a structure which is about to be freed + * is very nonsensual. Here it is largely a way to do a debug freeze, + * and a bracket which shows where the nonsensual code segment ends. + * + * Testing for -EINPROGRESS is always a bug, so we are bending + * the rules a little. + */ + spin_lock_irqsave(&sc->lock, flags); + if (sc->work_urb.status == -EINPROGRESS) { /* janitors: ignore */ + printk(KERN_WARNING "%s: " + "URB is active after disconnect\n", sc->name); + } + spin_unlock_irqrestore(&sc->lock, flags); + + /* + * There is virtually no chance that other CPU runs times so long + * after ub_urb_complete should have called del_timer, but only if HCD + * didn't forget to deliver a callback on unlink. + */ + del_timer_sync(&sc->work_timer); + + /* + * At this point there must be no commands coming from anyone + * and no URBs left in transit. + */ + + device_remove_file(&sc->intf->dev, &dev_attr_diag); + usb_set_intfdata(intf, NULL); + // usb_put_intf(sc->intf); + sc->intf = NULL; + usb_put_dev(sc->dev); + sc->dev = NULL; + + ub_put(sc); +} + +static struct usb_driver ub_driver = { + .owner = THIS_MODULE, + .name = "ub", + .probe = ub_probe, + .disconnect = ub_disconnect, + .id_table = ub_usb_ids, +}; + +static int __init ub_init(void) +{ + int rc; + + /* P3 */ printk("ub: sizeof ub_scsi_cmd %zu ub_dev %zu\n", + sizeof(struct ub_scsi_cmd), sizeof(struct ub_dev)); + + if ((rc = register_blkdev(UB_MAJOR, DRV_NAME)) != 0) + goto err_regblkdev; + devfs_mk_dir(DEVFS_NAME); + + if ((rc = usb_register(&ub_driver)) != 0) + goto err_register; + + return 0; + +err_register: + devfs_remove(DEVFS_NAME); + unregister_blkdev(UB_MAJOR, DRV_NAME); +err_regblkdev: + return rc; +} + +static void __exit ub_exit(void) +{ + usb_deregister(&ub_driver); + + devfs_remove(DEVFS_NAME); + unregister_blkdev(UB_MAJOR, DRV_NAME); +} + +module_init(ub_init); +module_exit(ub_exit); + +MODULE_LICENSE("GPL"); diff --git a/drivers/block/umem.c b/drivers/block/umem.c new file mode 100644 index 00000000000..0c4c121d2e7 --- /dev/null +++ b/drivers/block/umem.c @@ -0,0 +1,1256 @@ +/* + * mm.c - Micro Memory(tm) PCI memory board block device driver - v2.3 + * + * (C) 2001 San Mehat <nettwerk@valinux.com> + * (C) 2001 Johannes Erdfelt <jerdfelt@valinux.com> + * (C) 2001 NeilBrown <neilb@cse.unsw.edu.au> + * + * This driver for the Micro Memory PCI Memory Module with Battery Backup + * is Copyright Micro Memory Inc 2001-2002. All rights reserved. + * + * This driver is released to the public under the terms of the + * GNU GENERAL PUBLIC LICENSE version 2 + * See the file COPYING for details. + * + * This driver provides a standard block device interface for Micro Memory(tm) + * PCI based RAM boards. + * 10/05/01: Phap Nguyen - Rebuilt the driver + * 10/22/01: Phap Nguyen - v2.1 Added disk partitioning + * 29oct2001:NeilBrown - Use make_request_fn instead of request_fn + * - use stand disk partitioning (so fdisk works). + * 08nov2001:NeilBrown - change driver name from "mm" to "umem" + * - incorporate into main kernel + * 08apr2002:NeilBrown - Move some of interrupt handle to tasklet + * - use spin_lock_bh instead of _irq + * - Never block on make_request. queue + * bh's instead. + * - unregister umem from devfs at mod unload + * - Change version to 2.3 + * 07Nov2001:Phap Nguyen - Select pci read command: 06, 12, 15 (Decimal) + * 07Jan2002: P. Nguyen - Used PCI Memory Write & Invalidate for DMA + * 15May2002:NeilBrown - convert to bio for 2.5 + * 17May2002:NeilBrown - remove init_mem initialisation. Instead detect + * - a sequence of writes that cover the card, and + * - set initialised bit then. + */ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/fs.h> +#include <linux/bio.h> +#include <linux/kernel.h> +#include <linux/mm.h> +#include <linux/mman.h> +#include <linux/ioctl.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/interrupt.h> +#include <linux/smp_lock.h> +#include <linux/timer.h> +#include <linux/pci.h> +#include <linux/slab.h> + +#include <linux/fcntl.h> /* O_ACCMODE */ +#include <linux/hdreg.h> /* HDIO_GETGEO */ + +#include <linux/umem.h> + +#include <asm/uaccess.h> +#include <asm/io.h> + +#define PRINTK(x...) do {} while (0) +#define dprintk(x...) do {} while (0) +/*#define dprintk(x...) printk(x) */ + +#define MM_MAXCARDS 4 +#define MM_RAHEAD 2 /* two sectors */ +#define MM_BLKSIZE 1024 /* 1k blocks */ +#define MM_HARDSECT 512 /* 512-byte hardware sectors */ +#define MM_SHIFT 6 /* max 64 partitions on 4 cards */ + +/* + * Version Information + */ + +#define DRIVER_VERSION "v2.3" +#define DRIVER_AUTHOR "San Mehat, Johannes Erdfelt, NeilBrown" +#define DRIVER_DESC "Micro Memory(tm) PCI memory board block driver" + +static int debug; +/* #define HW_TRACE(x) writeb(x,cards[0].csr_remap + MEMCTRLSTATUS_MAGIC) */ +#define HW_TRACE(x) + +#define DEBUG_LED_ON_TRANSFER 0x01 +#define DEBUG_BATTERY_POLLING 0x02 + +module_param(debug, int, 0644); +MODULE_PARM_DESC(debug, "Debug bitmask"); + +static int pci_read_cmd = 0x0C; /* Read Multiple */ +module_param(pci_read_cmd, int, 0); +MODULE_PARM_DESC(pci_read_cmd, "PCI read command"); + +static int pci_write_cmd = 0x0F; /* Write and Invalidate */ +module_param(pci_write_cmd, int, 0); +MODULE_PARM_DESC(pci_write_cmd, "PCI write command"); + +static int pci_cmds; + +static int major_nr; + +#include <linux/blkdev.h> +#include <linux/blkpg.h> + +struct cardinfo { + int card_number; + struct pci_dev *dev; + + int irq; + + unsigned long csr_base; + unsigned char __iomem *csr_remap; + unsigned long csr_len; +#ifdef CONFIG_MM_MAP_MEMORY + unsigned long mem_base; + unsigned char __iomem *mem_remap; + unsigned long mem_len; +#endif + + unsigned int win_size; /* PCI window size */ + unsigned int mm_size; /* size in kbytes */ + + unsigned int init_size; /* initial segment, in sectors, + * that we know to + * have been written + */ + struct bio *bio, *currentbio, **biotail; + + request_queue_t *queue; + + struct mm_page { + dma_addr_t page_dma; + struct mm_dma_desc *desc; + int cnt, headcnt; + struct bio *bio, **biotail; + } mm_pages[2]; +#define DESC_PER_PAGE ((PAGE_SIZE*2)/sizeof(struct mm_dma_desc)) + + int Active, Ready; + + struct tasklet_struct tasklet; + unsigned int dma_status; + + struct { + int good; + int warned; + unsigned long last_change; + } battery[2]; + + spinlock_t lock; + int check_batteries; + + int flags; +}; + +static struct cardinfo cards[MM_MAXCARDS]; +static struct block_device_operations mm_fops; +static struct timer_list battery_timer; + +static int num_cards = 0; + +static struct gendisk *mm_gendisk[MM_MAXCARDS]; + +static void check_batteries(struct cardinfo *card); + +/* +----------------------------------------------------------------------------------- +-- get_userbit +----------------------------------------------------------------------------------- +*/ +static int get_userbit(struct cardinfo *card, int bit) +{ + unsigned char led; + + led = readb(card->csr_remap + MEMCTRLCMD_LEDCTRL); + return led & bit; +} +/* +----------------------------------------------------------------------------------- +-- set_userbit +----------------------------------------------------------------------------------- +*/ +static int set_userbit(struct cardinfo *card, int bit, unsigned char state) +{ + unsigned char led; + + led = readb(card->csr_remap + MEMCTRLCMD_LEDCTRL); + if (state) + led |= bit; + else + led &= ~bit; + writeb(led, card->csr_remap + MEMCTRLCMD_LEDCTRL); + + return 0; +} +/* +----------------------------------------------------------------------------------- +-- set_led +----------------------------------------------------------------------------------- +*/ +/* + * NOTE: For the power LED, use the LED_POWER_* macros since they differ + */ +static void set_led(struct cardinfo *card, int shift, unsigned char state) +{ + unsigned char led; + + led = readb(card->csr_remap + MEMCTRLCMD_LEDCTRL); + if (state == LED_FLIP) + led ^= (1<<shift); + else { + led &= ~(0x03 << shift); + led |= (state << shift); + } + writeb(led, card->csr_remap + MEMCTRLCMD_LEDCTRL); + +} + +#ifdef MM_DIAG +/* +----------------------------------------------------------------------------------- +-- dump_regs +----------------------------------------------------------------------------------- +*/ +static void dump_regs(struct cardinfo *card) +{ + unsigned char *p; + int i, i1; + + p = card->csr_remap; + for (i = 0; i < 8; i++) { + printk(KERN_DEBUG "%p ", p); + + for (i1 = 0; i1 < 16; i1++) + printk("%02x ", *p++); + + printk("\n"); + } +} +#endif +/* +----------------------------------------------------------------------------------- +-- dump_dmastat +----------------------------------------------------------------------------------- +*/ +static void dump_dmastat(struct cardinfo *card, unsigned int dmastat) +{ + printk(KERN_DEBUG "MM%d*: DMAstat - ", card->card_number); + if (dmastat & DMASCR_ANY_ERR) + printk("ANY_ERR "); + if (dmastat & DMASCR_MBE_ERR) + printk("MBE_ERR "); + if (dmastat & DMASCR_PARITY_ERR_REP) + printk("PARITY_ERR_REP "); + if (dmastat & DMASCR_PARITY_ERR_DET) + printk("PARITY_ERR_DET "); + if (dmastat & DMASCR_SYSTEM_ERR_SIG) + printk("SYSTEM_ERR_SIG "); + if (dmastat & DMASCR_TARGET_ABT) + printk("TARGET_ABT "); + if (dmastat & DMASCR_MASTER_ABT) + printk("MASTER_ABT "); + if (dmastat & DMASCR_CHAIN_COMPLETE) + printk("CHAIN_COMPLETE "); + if (dmastat & DMASCR_DMA_COMPLETE) + printk("DMA_COMPLETE "); + printk("\n"); +} + +/* + * Theory of request handling + * + * Each bio is assigned to one mm_dma_desc - which may not be enough FIXME + * We have two pages of mm_dma_desc, holding about 64 descriptors + * each. These are allocated at init time. + * One page is "Ready" and is either full, or can have request added. + * The other page might be "Active", which DMA is happening on it. + * + * Whenever IO on the active page completes, the Ready page is activated + * and the ex-Active page is clean out and made Ready. + * Otherwise the Ready page is only activated when it becomes full, or + * when mm_unplug_device is called via the unplug_io_fn. + * + * If a request arrives while both pages a full, it is queued, and b_rdev is + * overloaded to record whether it was a read or a write. + * + * The interrupt handler only polls the device to clear the interrupt. + * The processing of the result is done in a tasklet. + */ + +static void mm_start_io(struct cardinfo *card) +{ + /* we have the lock, we know there is + * no IO active, and we know that card->Active + * is set + */ + struct mm_dma_desc *desc; + struct mm_page *page; + int offset; + + /* make the last descriptor end the chain */ + page = &card->mm_pages[card->Active]; + PRINTK("start_io: %d %d->%d\n", card->Active, page->headcnt, page->cnt-1); + desc = &page->desc[page->cnt-1]; + + desc->control_bits |= cpu_to_le32(DMASCR_CHAIN_COMP_EN); + desc->control_bits &= ~cpu_to_le32(DMASCR_CHAIN_EN); + desc->sem_control_bits = desc->control_bits; + + + if (debug & DEBUG_LED_ON_TRANSFER) + set_led(card, LED_REMOVE, LED_ON); + + desc = &page->desc[page->headcnt]; + writel(0, card->csr_remap + DMA_PCI_ADDR); + writel(0, card->csr_remap + DMA_PCI_ADDR + 4); + + writel(0, card->csr_remap + DMA_LOCAL_ADDR); + writel(0, card->csr_remap + DMA_LOCAL_ADDR + 4); + + writel(0, card->csr_remap + DMA_TRANSFER_SIZE); + writel(0, card->csr_remap + DMA_TRANSFER_SIZE + 4); + + writel(0, card->csr_remap + DMA_SEMAPHORE_ADDR); + writel(0, card->csr_remap + DMA_SEMAPHORE_ADDR + 4); + + offset = ((char*)desc) - ((char*)page->desc); + writel(cpu_to_le32((page->page_dma+offset)&0xffffffff), + card->csr_remap + DMA_DESCRIPTOR_ADDR); + /* Force the value to u64 before shifting otherwise >> 32 is undefined C + * and on some ports will do nothing ! */ + writel(cpu_to_le32(((u64)page->page_dma)>>32), + card->csr_remap + DMA_DESCRIPTOR_ADDR + 4); + + /* Go, go, go */ + writel(cpu_to_le32(DMASCR_GO | DMASCR_CHAIN_EN | pci_cmds), + card->csr_remap + DMA_STATUS_CTRL); +} + +static int add_bio(struct cardinfo *card); + +static void activate(struct cardinfo *card) +{ + /* if No page is Active, and Ready is + * not empty, then switch Ready page + * to active and start IO. + * Then add any bh's that are available to Ready + */ + + do { + while (add_bio(card)) + ; + + if (card->Active == -1 && + card->mm_pages[card->Ready].cnt > 0) { + card->Active = card->Ready; + card->Ready = 1-card->Ready; + mm_start_io(card); + } + + } while (card->Active == -1 && add_bio(card)); +} + +static inline void reset_page(struct mm_page *page) +{ + page->cnt = 0; + page->headcnt = 0; + page->bio = NULL; + page->biotail = & page->bio; +} + +static void mm_unplug_device(request_queue_t *q) +{ + struct cardinfo *card = q->queuedata; + unsigned long flags; + + spin_lock_irqsave(&card->lock, flags); + if (blk_remove_plug(q)) + activate(card); + spin_unlock_irqrestore(&card->lock, flags); +} + +/* + * If there is room on Ready page, take + * one bh off list and add it. + * return 1 if there was room, else 0. + */ +static int add_bio(struct cardinfo *card) +{ + struct mm_page *p; + struct mm_dma_desc *desc; + dma_addr_t dma_handle; + int offset; + struct bio *bio; + int rw; + int len; + + bio = card->currentbio; + if (!bio && card->bio) { + card->currentbio = card->bio; + card->bio = card->bio->bi_next; + if (card->bio == NULL) + card->biotail = &card->bio; + card->currentbio->bi_next = NULL; + return 1; + } + if (!bio) + return 0; + + rw = bio_rw(bio); + if (card->mm_pages[card->Ready].cnt >= DESC_PER_PAGE) + return 0; + + len = bio_iovec(bio)->bv_len; + dma_handle = pci_map_page(card->dev, + bio_page(bio), + bio_offset(bio), + len, + (rw==READ) ? + PCI_DMA_FROMDEVICE : PCI_DMA_TODEVICE); + + p = &card->mm_pages[card->Ready]; + desc = &p->desc[p->cnt]; + p->cnt++; + if ((p->biotail) != &bio->bi_next) { + *(p->biotail) = bio; + p->biotail = &(bio->bi_next); + bio->bi_next = NULL; + } + + desc->data_dma_handle = dma_handle; + + desc->pci_addr = cpu_to_le64((u64)desc->data_dma_handle); + desc->local_addr= cpu_to_le64(bio->bi_sector << 9); + desc->transfer_size = cpu_to_le32(len); + offset = ( ((char*)&desc->sem_control_bits) - ((char*)p->desc)); + desc->sem_addr = cpu_to_le64((u64)(p->page_dma+offset)); + desc->zero1 = desc->zero2 = 0; + offset = ( ((char*)(desc+1)) - ((char*)p->desc)); + desc->next_desc_addr = cpu_to_le64(p->page_dma+offset); + desc->control_bits = cpu_to_le32(DMASCR_GO|DMASCR_ERR_INT_EN| + DMASCR_PARITY_INT_EN| + DMASCR_CHAIN_EN | + DMASCR_SEM_EN | + pci_cmds); + if (rw == WRITE) + desc->control_bits |= cpu_to_le32(DMASCR_TRANSFER_READ); + desc->sem_control_bits = desc->control_bits; + + bio->bi_sector += (len>>9); + bio->bi_size -= len; + bio->bi_idx++; + if (bio->bi_idx >= bio->bi_vcnt) + card->currentbio = NULL; + + return 1; +} + +static void process_page(unsigned long data) +{ + /* check if any of the requests in the page are DMA_COMPLETE, + * and deal with them appropriately. + * If we find a descriptor without DMA_COMPLETE in the semaphore, then + * dma must have hit an error on that descriptor, so use dma_status instead + * and assume that all following descriptors must be re-tried. + */ + struct mm_page *page; + struct bio *return_bio=NULL; + struct cardinfo *card = (struct cardinfo *)data; + unsigned int dma_status = card->dma_status; + + spin_lock_bh(&card->lock); + if (card->Active < 0) + goto out_unlock; + page = &card->mm_pages[card->Active]; + + while (page->headcnt < page->cnt) { + struct bio *bio = page->bio; + struct mm_dma_desc *desc = &page->desc[page->headcnt]; + int control = le32_to_cpu(desc->sem_control_bits); + int last=0; + int idx; + + if (!(control & DMASCR_DMA_COMPLETE)) { + control = dma_status; + last=1; + } + page->headcnt++; + idx = bio->bi_phys_segments; + bio->bi_phys_segments++; + if (bio->bi_phys_segments >= bio->bi_vcnt) + page->bio = bio->bi_next; + + pci_unmap_page(card->dev, desc->data_dma_handle, + bio_iovec_idx(bio,idx)->bv_len, + (control& DMASCR_TRANSFER_READ) ? + PCI_DMA_TODEVICE : PCI_DMA_FROMDEVICE); + if (control & DMASCR_HARD_ERROR) { + /* error */ + clear_bit(BIO_UPTODATE, &bio->bi_flags); + printk(KERN_WARNING "MM%d: I/O error on sector %d/%d\n", + card->card_number, + le32_to_cpu(desc->local_addr)>>9, + le32_to_cpu(desc->transfer_size)); + dump_dmastat(card, control); + } else if (test_bit(BIO_RW, &bio->bi_rw) && + le32_to_cpu(desc->local_addr)>>9 == card->init_size) { + card->init_size += le32_to_cpu(desc->transfer_size)>>9; + if (card->init_size>>1 >= card->mm_size) { + printk(KERN_INFO "MM%d: memory now initialised\n", + card->card_number); + set_userbit(card, MEMORY_INITIALIZED, 1); + } + } + if (bio != page->bio) { + bio->bi_next = return_bio; + return_bio = bio; + } + + if (last) break; + } + + if (debug & DEBUG_LED_ON_TRANSFER) + set_led(card, LED_REMOVE, LED_OFF); + + if (card->check_batteries) { + card->check_batteries = 0; + check_batteries(card); + } + if (page->headcnt >= page->cnt) { + reset_page(page); + card->Active = -1; + activate(card); + } else { + /* haven't finished with this one yet */ + PRINTK("do some more\n"); + mm_start_io(card); + } + out_unlock: + spin_unlock_bh(&card->lock); + + while(return_bio) { + struct bio *bio = return_bio; + + return_bio = bio->bi_next; + bio->bi_next = NULL; + bio_endio(bio, bio->bi_size, 0); + } +} + +/* +----------------------------------------------------------------------------------- +-- mm_make_request +----------------------------------------------------------------------------------- +*/ +static int mm_make_request(request_queue_t *q, struct bio *bio) +{ + struct cardinfo *card = q->queuedata; + PRINTK("mm_make_request %ld %d\n", bh->b_rsector, bh->b_size); + + bio->bi_phys_segments = bio->bi_idx; /* count of completed segments*/ + spin_lock_irq(&card->lock); + *card->biotail = bio; + bio->bi_next = NULL; + card->biotail = &bio->bi_next; + blk_plug_device(q); + spin_unlock_irq(&card->lock); + + return 0; +} + +/* +----------------------------------------------------------------------------------- +-- mm_interrupt +----------------------------------------------------------------------------------- +*/ +static irqreturn_t mm_interrupt(int irq, void *__card, struct pt_regs *regs) +{ + struct cardinfo *card = (struct cardinfo *) __card; + unsigned int dma_status; + unsigned short cfg_status; + +HW_TRACE(0x30); + + dma_status = le32_to_cpu(readl(card->csr_remap + DMA_STATUS_CTRL)); + + if (!(dma_status & (DMASCR_ERROR_MASK | DMASCR_CHAIN_COMPLETE))) { + /* interrupt wasn't for me ... */ + return IRQ_NONE; + } + + /* clear COMPLETION interrupts */ + if (card->flags & UM_FLAG_NO_BYTE_STATUS) + writel(cpu_to_le32(DMASCR_DMA_COMPLETE|DMASCR_CHAIN_COMPLETE), + card->csr_remap+ DMA_STATUS_CTRL); + else + writeb((DMASCR_DMA_COMPLETE|DMASCR_CHAIN_COMPLETE) >> 16, + card->csr_remap+ DMA_STATUS_CTRL + 2); + + /* log errors and clear interrupt status */ + if (dma_status & DMASCR_ANY_ERR) { + unsigned int data_log1, data_log2; + unsigned int addr_log1, addr_log2; + unsigned char stat, count, syndrome, check; + + stat = readb(card->csr_remap + MEMCTRLCMD_ERRSTATUS); + + data_log1 = le32_to_cpu(readl(card->csr_remap + ERROR_DATA_LOG)); + data_log2 = le32_to_cpu(readl(card->csr_remap + ERROR_DATA_LOG + 4)); + addr_log1 = le32_to_cpu(readl(card->csr_remap + ERROR_ADDR_LOG)); + addr_log2 = readb(card->csr_remap + ERROR_ADDR_LOG + 4); + + count = readb(card->csr_remap + ERROR_COUNT); + syndrome = readb(card->csr_remap + ERROR_SYNDROME); + check = readb(card->csr_remap + ERROR_CHECK); + + dump_dmastat(card, dma_status); + + if (stat & 0x01) + printk(KERN_ERR "MM%d*: Memory access error detected (err count %d)\n", + card->card_number, count); + if (stat & 0x02) + printk(KERN_ERR "MM%d*: Multi-bit EDC error\n", + card->card_number); + + printk(KERN_ERR "MM%d*: Fault Address 0x%02x%08x, Fault Data 0x%08x%08x\n", + card->card_number, addr_log2, addr_log1, data_log2, data_log1); + printk(KERN_ERR "MM%d*: Fault Check 0x%02x, Fault Syndrome 0x%02x\n", + card->card_number, check, syndrome); + + writeb(0, card->csr_remap + ERROR_COUNT); + } + + if (dma_status & DMASCR_PARITY_ERR_REP) { + printk(KERN_ERR "MM%d*: PARITY ERROR REPORTED\n", card->card_number); + pci_read_config_word(card->dev, PCI_STATUS, &cfg_status); + pci_write_config_word(card->dev, PCI_STATUS, cfg_status); + } + + if (dma_status & DMASCR_PARITY_ERR_DET) { + printk(KERN_ERR "MM%d*: PARITY ERROR DETECTED\n", card->card_number); + pci_read_config_word(card->dev, PCI_STATUS, &cfg_status); + pci_write_config_word(card->dev, PCI_STATUS, cfg_status); + } + + if (dma_status & DMASCR_SYSTEM_ERR_SIG) { + printk(KERN_ERR "MM%d*: SYSTEM ERROR\n", card->card_number); + pci_read_config_word(card->dev, PCI_STATUS, &cfg_status); + pci_write_config_word(card->dev, PCI_STATUS, cfg_status); + } + + if (dma_status & DMASCR_TARGET_ABT) { + printk(KERN_ERR "MM%d*: TARGET ABORT\n", card->card_number); + pci_read_config_word(card->dev, PCI_STATUS, &cfg_status); + pci_write_config_word(card->dev, PCI_STATUS, cfg_status); + } + + if (dma_status & DMASCR_MASTER_ABT) { + printk(KERN_ERR "MM%d*: MASTER ABORT\n", card->card_number); + pci_read_config_word(card->dev, PCI_STATUS, &cfg_status); + pci_write_config_word(card->dev, PCI_STATUS, cfg_status); + } + + /* and process the DMA descriptors */ + card->dma_status = dma_status; + tasklet_schedule(&card->tasklet); + +HW_TRACE(0x36); + + return IRQ_HANDLED; +} +/* +----------------------------------------------------------------------------------- +-- set_fault_to_battery_status +----------------------------------------------------------------------------------- +*/ +/* + * If both batteries are good, no LED + * If either battery has been warned, solid LED + * If both batteries are bad, flash the LED quickly + * If either battery is bad, flash the LED semi quickly + */ +static void set_fault_to_battery_status(struct cardinfo *card) +{ + if (card->battery[0].good && card->battery[1].good) + set_led(card, LED_FAULT, LED_OFF); + else if (card->battery[0].warned || card->battery[1].warned) + set_led(card, LED_FAULT, LED_ON); + else if (!card->battery[0].good && !card->battery[1].good) + set_led(card, LED_FAULT, LED_FLASH_7_0); + else + set_led(card, LED_FAULT, LED_FLASH_3_5); +} + +static void init_battery_timer(void); + + +/* +----------------------------------------------------------------------------------- +-- check_battery +----------------------------------------------------------------------------------- +*/ +static int check_battery(struct cardinfo *card, int battery, int status) +{ + if (status != card->battery[battery].good) { + card->battery[battery].good = !card->battery[battery].good; + card->battery[battery].last_change = jiffies; + + if (card->battery[battery].good) { + printk(KERN_ERR "MM%d: Battery %d now good\n", + card->card_number, battery + 1); + card->battery[battery].warned = 0; + } else + printk(KERN_ERR "MM%d: Battery %d now FAILED\n", + card->card_number, battery + 1); + + return 1; + } else if (!card->battery[battery].good && + !card->battery[battery].warned && + time_after_eq(jiffies, card->battery[battery].last_change + + (HZ * 60 * 60 * 5))) { + printk(KERN_ERR "MM%d: Battery %d still FAILED after 5 hours\n", + card->card_number, battery + 1); + card->battery[battery].warned = 1; + + return 1; + } + + return 0; +} +/* +----------------------------------------------------------------------------------- +-- check_batteries +----------------------------------------------------------------------------------- +*/ +static void check_batteries(struct cardinfo *card) +{ + /* NOTE: this must *never* be called while the card + * is doing (bus-to-card) DMA, or you will need the + * reset switch + */ + unsigned char status; + int ret1, ret2; + + status = readb(card->csr_remap + MEMCTRLSTATUS_BATTERY); + if (debug & DEBUG_BATTERY_POLLING) + printk(KERN_DEBUG "MM%d: checking battery status, 1 = %s, 2 = %s\n", + card->card_number, + (status & BATTERY_1_FAILURE) ? "FAILURE" : "OK", + (status & BATTERY_2_FAILURE) ? "FAILURE" : "OK"); + + ret1 = check_battery(card, 0, !(status & BATTERY_1_FAILURE)); + ret2 = check_battery(card, 1, !(status & BATTERY_2_FAILURE)); + + if (ret1 || ret2) + set_fault_to_battery_status(card); +} + +static void check_all_batteries(unsigned long ptr) +{ + int i; + + for (i = 0; i < num_cards; i++) + if (!(cards[i].flags & UM_FLAG_NO_BATT)) { + struct cardinfo *card = &cards[i]; + spin_lock_bh(&card->lock); + if (card->Active >= 0) + card->check_batteries = 1; + else + check_batteries(card); + spin_unlock_bh(&card->lock); + } + + init_battery_timer(); +} +/* +----------------------------------------------------------------------------------- +-- init_battery_timer +----------------------------------------------------------------------------------- +*/ +static void init_battery_timer(void) +{ + init_timer(&battery_timer); + battery_timer.function = check_all_batteries; + battery_timer.expires = jiffies + (HZ * 60); + add_timer(&battery_timer); +} +/* +----------------------------------------------------------------------------------- +-- del_battery_timer +----------------------------------------------------------------------------------- +*/ +static void del_battery_timer(void) +{ + del_timer(&battery_timer); +} +/* +----------------------------------------------------------------------------------- +-- mm_revalidate +----------------------------------------------------------------------------------- +*/ +/* + * Note no locks taken out here. In a worst case scenario, we could drop + * a chunk of system memory. But that should never happen, since validation + * happens at open or mount time, when locks are held. + * + * That's crap, since doing that while some partitions are opened + * or mounted will give you really nasty results. + */ +static int mm_revalidate(struct gendisk *disk) +{ + struct cardinfo *card = disk->private_data; + set_capacity(disk, card->mm_size << 1); + return 0; +} +/* +----------------------------------------------------------------------------------- +-- mm_ioctl +----------------------------------------------------------------------------------- +*/ +static int mm_ioctl(struct inode *i, struct file *f, unsigned int cmd, unsigned long arg) +{ + if (cmd == HDIO_GETGEO) { + struct cardinfo *card = i->i_bdev->bd_disk->private_data; + int size = card->mm_size * (1024 / MM_HARDSECT); + struct hd_geometry geo; + /* + * get geometry: we have to fake one... trim the size to a + * multiple of 2048 (1M): tell we have 32 sectors, 64 heads, + * whatever cylinders. + */ + geo.heads = 64; + geo.sectors = 32; + geo.start = get_start_sect(i->i_bdev); + geo.cylinders = size / (geo.heads * geo.sectors); + + if (copy_to_user((void __user *) arg, &geo, sizeof(geo))) + return -EFAULT; + return 0; + } + + return -EINVAL; +} +/* +----------------------------------------------------------------------------------- +-- mm_check_change +----------------------------------------------------------------------------------- + Future support for removable devices +*/ +static int mm_check_change(struct gendisk *disk) +{ +/* struct cardinfo *dev = disk->private_data; */ + return 0; +} +/* +----------------------------------------------------------------------------------- +-- mm_fops +----------------------------------------------------------------------------------- +*/ +static struct block_device_operations mm_fops = { + .owner = THIS_MODULE, + .ioctl = mm_ioctl, + .revalidate_disk= mm_revalidate, + .media_changed = mm_check_change, +}; +/* +----------------------------------------------------------------------------------- +-- mm_pci_probe +----------------------------------------------------------------------------------- +*/ +static int __devinit mm_pci_probe(struct pci_dev *dev, const struct pci_device_id *id) +{ + int ret = -ENODEV; + struct cardinfo *card = &cards[num_cards]; + unsigned char mem_present; + unsigned char batt_status; + unsigned int saved_bar, data; + int magic_number; + + if (pci_enable_device(dev) < 0) + return -ENODEV; + + pci_write_config_byte(dev, PCI_LATENCY_TIMER, 0xF8); + pci_set_master(dev); + + card->dev = dev; + card->card_number = num_cards; + + card->csr_base = pci_resource_start(dev, 0); + card->csr_len = pci_resource_len(dev, 0); +#ifdef CONFIG_MM_MAP_MEMORY + card->mem_base = pci_resource_start(dev, 1); + card->mem_len = pci_resource_len(dev, 1); +#endif + + printk(KERN_INFO "Micro Memory(tm) controller #%d found at %02x:%02x (PCI Mem Module (Battery Backup))\n", + card->card_number, dev->bus->number, dev->devfn); + + if (pci_set_dma_mask(dev, 0xffffffffffffffffLL) && + !pci_set_dma_mask(dev, 0xffffffffLL)) { + printk(KERN_WARNING "MM%d: NO suitable DMA found\n",num_cards); + return -ENOMEM; + } + if (!request_mem_region(card->csr_base, card->csr_len, "Micro Memory")) { + printk(KERN_ERR "MM%d: Unable to request memory region\n", card->card_number); + ret = -ENOMEM; + + goto failed_req_csr; + } + + card->csr_remap = ioremap_nocache(card->csr_base, card->csr_len); + if (!card->csr_remap) { + printk(KERN_ERR "MM%d: Unable to remap memory region\n", card->card_number); + ret = -ENOMEM; + + goto failed_remap_csr; + } + + printk(KERN_INFO "MM%d: CSR 0x%08lx -> 0x%p (0x%lx)\n", card->card_number, + card->csr_base, card->csr_remap, card->csr_len); + +#ifdef CONFIG_MM_MAP_MEMORY + if (!request_mem_region(card->mem_base, card->mem_len, "Micro Memory")) { + printk(KERN_ERR "MM%d: Unable to request memory region\n", card->card_number); + ret = -ENOMEM; + + goto failed_req_mem; + } + + if (!(card->mem_remap = ioremap(card->mem_base, cards->mem_len))) { + printk(KERN_ERR "MM%d: Unable to remap memory region\n", card->card_number); + ret = -ENOMEM; + + goto failed_remap_mem; + } + + printk(KERN_INFO "MM%d: MEM 0x%8lx -> 0x%8lx (0x%lx)\n", card->card_number, + card->mem_base, card->mem_remap, card->mem_len); +#else + printk(KERN_INFO "MM%d: MEM area not remapped (CONFIG_MM_MAP_MEMORY not set)\n", + card->card_number); +#endif + switch(card->dev->device) { + case 0x5415: + card->flags |= UM_FLAG_NO_BYTE_STATUS | UM_FLAG_NO_BATTREG; + magic_number = 0x59; + break; + + case 0x5425: + card->flags |= UM_FLAG_NO_BYTE_STATUS; + magic_number = 0x5C; + break; + + case 0x6155: + card->flags |= UM_FLAG_NO_BYTE_STATUS | UM_FLAG_NO_BATTREG | UM_FLAG_NO_BATT; + magic_number = 0x99; + break; + + default: + magic_number = 0x100; + break; + } + + if (readb(card->csr_remap + MEMCTRLSTATUS_MAGIC) != magic_number) { + printk(KERN_ERR "MM%d: Magic number invalid\n", card->card_number); + ret = -ENOMEM; + goto failed_magic; + } + + card->mm_pages[0].desc = pci_alloc_consistent(card->dev, + PAGE_SIZE*2, + &card->mm_pages[0].page_dma); + card->mm_pages[1].desc = pci_alloc_consistent(card->dev, + PAGE_SIZE*2, + &card->mm_pages[1].page_dma); + if (card->mm_pages[0].desc == NULL || + card->mm_pages[1].desc == NULL) { + printk(KERN_ERR "MM%d: alloc failed\n", card->card_number); + goto failed_alloc; + } + reset_page(&card->mm_pages[0]); + reset_page(&card->mm_pages[1]); + card->Ready = 0; /* page 0 is ready */ + card->Active = -1; /* no page is active */ + card->bio = NULL; + card->biotail = &card->bio; + + card->queue = blk_alloc_queue(GFP_KERNEL); + if (!card->queue) + goto failed_alloc; + + blk_queue_make_request(card->queue, mm_make_request); + card->queue->queuedata = card; + card->queue->unplug_fn = mm_unplug_device; + + tasklet_init(&card->tasklet, process_page, (unsigned long)card); + + card->check_batteries = 0; + + mem_present = readb(card->csr_remap + MEMCTRLSTATUS_MEMORY); + switch (mem_present) { + case MEM_128_MB: + card->mm_size = 1024 * 128; + break; + case MEM_256_MB: + card->mm_size = 1024 * 256; + break; + case MEM_512_MB: + card->mm_size = 1024 * 512; + break; + case MEM_1_GB: + card->mm_size = 1024 * 1024; + break; + case MEM_2_GB: + card->mm_size = 1024 * 2048; + break; + default: + card->mm_size = 0; + break; + } + + /* Clear the LED's we control */ + set_led(card, LED_REMOVE, LED_OFF); + set_led(card, LED_FAULT, LED_OFF); + + batt_status = readb(card->csr_remap + MEMCTRLSTATUS_BATTERY); + + card->battery[0].good = !(batt_status & BATTERY_1_FAILURE); + card->battery[1].good = !(batt_status & BATTERY_2_FAILURE); + card->battery[0].last_change = card->battery[1].last_change = jiffies; + + if (card->flags & UM_FLAG_NO_BATT) + printk(KERN_INFO "MM%d: Size %d KB\n", + card->card_number, card->mm_size); + else { + printk(KERN_INFO "MM%d: Size %d KB, Battery 1 %s (%s), Battery 2 %s (%s)\n", + card->card_number, card->mm_size, + (batt_status & BATTERY_1_DISABLED ? "Disabled" : "Enabled"), + card->battery[0].good ? "OK" : "FAILURE", + (batt_status & BATTERY_2_DISABLED ? "Disabled" : "Enabled"), + card->battery[1].good ? "OK" : "FAILURE"); + + set_fault_to_battery_status(card); + } + + pci_read_config_dword(dev, PCI_BASE_ADDRESS_1, &saved_bar); + data = 0xffffffff; + pci_write_config_dword(dev, PCI_BASE_ADDRESS_1, data); + pci_read_config_dword(dev, PCI_BASE_ADDRESS_1, &data); + pci_write_config_dword(dev, PCI_BASE_ADDRESS_1, saved_bar); + data &= 0xfffffff0; + data = ~data; + data += 1; + + card->win_size = data; + + + if (request_irq(dev->irq, mm_interrupt, SA_SHIRQ, "pci-umem", card)) { + printk(KERN_ERR "MM%d: Unable to allocate IRQ\n", card->card_number); + ret = -ENODEV; + + goto failed_req_irq; + } + + card->irq = dev->irq; + printk(KERN_INFO "MM%d: Window size %d bytes, IRQ %d\n", card->card_number, + card->win_size, card->irq); + + spin_lock_init(&card->lock); + + pci_set_drvdata(dev, card); + + if (pci_write_cmd != 0x0F) /* If not Memory Write & Invalidate */ + pci_write_cmd = 0x07; /* then Memory Write command */ + + if (pci_write_cmd & 0x08) { /* use Memory Write and Invalidate */ + unsigned short cfg_command; + pci_read_config_word(dev, PCI_COMMAND, &cfg_command); + cfg_command |= 0x10; /* Memory Write & Invalidate Enable */ + pci_write_config_word(dev, PCI_COMMAND, cfg_command); + } + pci_cmds = (pci_read_cmd << 28) | (pci_write_cmd << 24); + + num_cards++; + + if (!get_userbit(card, MEMORY_INITIALIZED)) { + printk(KERN_INFO "MM%d: memory NOT initialized. Consider over-writing whole device.\n", card->card_number); + card->init_size = 0; + } else { + printk(KERN_INFO "MM%d: memory already initialized\n", card->card_number); + card->init_size = card->mm_size; + } + + /* Enable ECC */ + writeb(EDC_STORE_CORRECT, card->csr_remap + MEMCTRLCMD_ERRCTRL); + + return 0; + + failed_req_irq: + failed_alloc: + if (card->mm_pages[0].desc) + pci_free_consistent(card->dev, PAGE_SIZE*2, + card->mm_pages[0].desc, + card->mm_pages[0].page_dma); + if (card->mm_pages[1].desc) + pci_free_consistent(card->dev, PAGE_SIZE*2, + card->mm_pages[1].desc, + card->mm_pages[1].page_dma); + failed_magic: +#ifdef CONFIG_MM_MAP_MEMORY + iounmap(card->mem_remap); + failed_remap_mem: + release_mem_region(card->mem_base, card->mem_len); + failed_req_mem: +#endif + iounmap(card->csr_remap); + failed_remap_csr: + release_mem_region(card->csr_base, card->csr_len); + failed_req_csr: + + return ret; +} +/* +----------------------------------------------------------------------------------- +-- mm_pci_remove +----------------------------------------------------------------------------------- +*/ +static void mm_pci_remove(struct pci_dev *dev) +{ + struct cardinfo *card = pci_get_drvdata(dev); + + tasklet_kill(&card->tasklet); + iounmap(card->csr_remap); + release_mem_region(card->csr_base, card->csr_len); +#ifdef CONFIG_MM_MAP_MEMORY + iounmap(card->mem_remap); + release_mem_region(card->mem_base, card->mem_len); +#endif + free_irq(card->irq, card); + + if (card->mm_pages[0].desc) + pci_free_consistent(card->dev, PAGE_SIZE*2, + card->mm_pages[0].desc, + card->mm_pages[0].page_dma); + if (card->mm_pages[1].desc) + pci_free_consistent(card->dev, PAGE_SIZE*2, + card->mm_pages[1].desc, + card->mm_pages[1].page_dma); + blk_put_queue(card->queue); +} + +static const struct pci_device_id mm_pci_ids[] = { { + .vendor = PCI_VENDOR_ID_MICRO_MEMORY, + .device = PCI_DEVICE_ID_MICRO_MEMORY_5415CN, + }, { + .vendor = PCI_VENDOR_ID_MICRO_MEMORY, + .device = PCI_DEVICE_ID_MICRO_MEMORY_5425CN, + }, { + .vendor = PCI_VENDOR_ID_MICRO_MEMORY, + .device = PCI_DEVICE_ID_MICRO_MEMORY_6155, + }, { + .vendor = 0x8086, + .device = 0xB555, + .subvendor= 0x1332, + .subdevice= 0x5460, + .class = 0x050000, + .class_mask= 0, + }, { /* end: all zeroes */ } +}; + +MODULE_DEVICE_TABLE(pci, mm_pci_ids); + +static struct pci_driver mm_pci_driver = { + .name = "umem", + .id_table = mm_pci_ids, + .probe = mm_pci_probe, + .remove = mm_pci_remove, +}; +/* +----------------------------------------------------------------------------------- +-- mm_init +----------------------------------------------------------------------------------- +*/ + +static int __init mm_init(void) +{ + int retval, i; + int err; + + printk(KERN_INFO DRIVER_VERSION " : " DRIVER_DESC "\n"); + + retval = pci_module_init(&mm_pci_driver); + if (retval) + return -ENOMEM; + + err = major_nr = register_blkdev(0, "umem"); + if (err < 0) + return -EIO; + + for (i = 0; i < num_cards; i++) { + mm_gendisk[i] = alloc_disk(1 << MM_SHIFT); + if (!mm_gendisk[i]) + goto out; + } + + for (i = 0; i < num_cards; i++) { + struct gendisk *disk = mm_gendisk[i]; + sprintf(disk->disk_name, "umem%c", 'a'+i); + sprintf(disk->devfs_name, "umem/card%d", i); + spin_lock_init(&cards[i].lock); + disk->major = major_nr; + disk->first_minor = i << MM_SHIFT; + disk->fops = &mm_fops; + disk->private_data = &cards[i]; + disk->queue = cards[i].queue; + set_capacity(disk, cards[i].mm_size << 1); + add_disk(disk); + } + + init_battery_timer(); + printk("MM: desc_per_page = %ld\n", DESC_PER_PAGE); +/* printk("mm_init: Done. 10-19-01 9:00\n"); */ + return 0; + +out: + unregister_blkdev(major_nr, "umem"); + while (i--) + put_disk(mm_gendisk[i]); + return -ENOMEM; +} +/* +----------------------------------------------------------------------------------- +-- mm_cleanup +----------------------------------------------------------------------------------- +*/ +static void __exit mm_cleanup(void) +{ + int i; + + del_battery_timer(); + + for (i=0; i < num_cards ; i++) { + del_gendisk(mm_gendisk[i]); + put_disk(mm_gendisk[i]); + } + + pci_unregister_driver(&mm_pci_driver); + + unregister_blkdev(major_nr, "umem"); +} + +module_init(mm_init); +module_exit(mm_cleanup); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL"); diff --git a/drivers/block/viodasd.c b/drivers/block/viodasd.c new file mode 100644 index 00000000000..46e56a25d2c --- /dev/null +++ b/drivers/block/viodasd.c @@ -0,0 +1,846 @@ +/* -*- linux-c -*- + * viodasd.c + * Authors: Dave Boutcher <boutcher@us.ibm.com> + * Ryan Arnold <ryanarn@us.ibm.com> + * Colin Devilbiss <devilbis@us.ibm.com> + * Stephen Rothwell <sfr@au1.ibm.com> + * + * (C) Copyright 2000-2004 IBM Corporation + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License as + * published by the Free Software Foundation; either version 2 of the + * License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * This routine provides access to disk space (termed "DASD" in historical + * IBM terms) owned and managed by an OS/400 partition running on the + * same box as this Linux partition. + * + * All disk operations are performed by sending messages back and forth to + * the OS/400 partition. + */ +#include <linux/major.h> +#include <linux/fs.h> +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/blkdev.h> +#include <linux/genhd.h> +#include <linux/hdreg.h> +#include <linux/errno.h> +#include <linux/init.h> +#include <linux/string.h> +#include <linux/dma-mapping.h> +#include <linux/completion.h> +#include <linux/device.h> +#include <linux/kernel.h> + +#include <asm/uaccess.h> +#include <asm/vio.h> +#include <asm/iSeries/HvTypes.h> +#include <asm/iSeries/HvLpEvent.h> +#include <asm/iSeries/HvLpConfig.h> +#include <asm/iSeries/vio.h> + +MODULE_DESCRIPTION("iSeries Virtual DASD"); +MODULE_AUTHOR("Dave Boutcher"); +MODULE_LICENSE("GPL"); + +/* + * We only support 7 partitions per physical disk....so with minor + * numbers 0-255 we get a maximum of 32 disks. + */ +#define VIOD_GENHD_NAME "iseries/vd" +#define VIOD_GENHD_DEVFS_NAME "iseries/disc" + +#define VIOD_VERS "1.64" + +#define VIOD_KERN_WARNING KERN_WARNING "viod: " +#define VIOD_KERN_INFO KERN_INFO "viod: " + +enum { + PARTITION_SHIFT = 3, + MAX_DISKNO = HVMAXARCHITECTEDVIRTUALDISKS, + MAX_DISK_NAME = sizeof(((struct gendisk *)0)->disk_name) +}; + +static DEFINE_SPINLOCK(viodasd_spinlock); + +#define VIOMAXREQ 16 +#define VIOMAXBLOCKDMA 12 + +#define DEVICE_NO(cell) ((struct viodasd_device *)(cell) - &viodasd_devices[0]) + +struct open_data { + u64 disk_size; + u16 max_disk; + u16 cylinders; + u16 tracks; + u16 sectors; + u16 bytes_per_sector; +}; + +struct rw_data { + u64 offset; + struct { + u32 token; + u32 reserved; + u64 len; + } dma_info[VIOMAXBLOCKDMA]; +}; + +struct vioblocklpevent { + struct HvLpEvent event; + u32 reserved; + u16 version; + u16 sub_result; + u16 disk; + u16 flags; + union { + struct open_data open_data; + struct rw_data rw_data; + u64 changed; + } u; +}; + +#define vioblockflags_ro 0x0001 + +enum vioblocksubtype { + vioblockopen = 0x0001, + vioblockclose = 0x0002, + vioblockread = 0x0003, + vioblockwrite = 0x0004, + vioblockflush = 0x0005, + vioblockcheck = 0x0007 +}; + +struct viodasd_waitevent { + struct completion com; + int rc; + u16 sub_result; + int max_disk; /* open */ +}; + +static const struct vio_error_entry viodasd_err_table[] = { + { 0x0201, EINVAL, "Invalid Range" }, + { 0x0202, EINVAL, "Invalid Token" }, + { 0x0203, EIO, "DMA Error" }, + { 0x0204, EIO, "Use Error" }, + { 0x0205, EIO, "Release Error" }, + { 0x0206, EINVAL, "Invalid Disk" }, + { 0x0207, EBUSY, "Cant Lock" }, + { 0x0208, EIO, "Already Locked" }, + { 0x0209, EIO, "Already Unlocked" }, + { 0x020A, EIO, "Invalid Arg" }, + { 0x020B, EIO, "Bad IFS File" }, + { 0x020C, EROFS, "Read Only Device" }, + { 0x02FF, EIO, "Internal Error" }, + { 0x0000, 0, NULL }, +}; + +/* + * Figure out the biggest I/O request (in sectors) we can accept + */ +#define VIODASD_MAXSECTORS (4096 / 512 * VIOMAXBLOCKDMA) + +/* + * Number of disk I/O requests we've sent to OS/400 + */ +static int num_req_outstanding; + +/* + * This is our internal structure for keeping track of disk devices + */ +struct viodasd_device { + u16 cylinders; + u16 tracks; + u16 sectors; + u16 bytes_per_sector; + u64 size; + int read_only; + spinlock_t q_lock; + struct gendisk *disk; + struct device *dev; +} viodasd_devices[MAX_DISKNO]; + +/* + * External open entry point. + */ +static int viodasd_open(struct inode *ino, struct file *fil) +{ + struct viodasd_device *d = ino->i_bdev->bd_disk->private_data; + HvLpEvent_Rc hvrc; + struct viodasd_waitevent we; + u16 flags = 0; + + if (d->read_only) { + if ((fil != NULL) && (fil->f_mode & FMODE_WRITE)) + return -EROFS; + flags = vioblockflags_ro; + } + + init_completion(&we.com); + + /* Send the open event to OS/400 */ + hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, + HvLpEvent_Type_VirtualIo, + viomajorsubtype_blockio | vioblockopen, + HvLpEvent_AckInd_DoAck, HvLpEvent_AckType_ImmediateAck, + viopath_sourceinst(viopath_hostLp), + viopath_targetinst(viopath_hostLp), + (u64)(unsigned long)&we, VIOVERSION << 16, + ((u64)DEVICE_NO(d) << 48) | ((u64)flags << 32), + 0, 0, 0); + if (hvrc != 0) { + printk(VIOD_KERN_WARNING "HV open failed %d\n", (int)hvrc); + return -EIO; + } + + wait_for_completion(&we.com); + + /* Check the return code */ + if (we.rc != 0) { + const struct vio_error_entry *err = + vio_lookup_rc(viodasd_err_table, we.sub_result); + + printk(VIOD_KERN_WARNING + "bad rc opening disk: %d:0x%04x (%s)\n", + (int)we.rc, we.sub_result, err->msg); + return -EIO; + } + + return 0; +} + +/* + * External release entry point. + */ +static int viodasd_release(struct inode *ino, struct file *fil) +{ + struct viodasd_device *d = ino->i_bdev->bd_disk->private_data; + HvLpEvent_Rc hvrc; + + /* Send the event to OS/400. We DON'T expect a response */ + hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, + HvLpEvent_Type_VirtualIo, + viomajorsubtype_blockio | vioblockclose, + HvLpEvent_AckInd_NoAck, HvLpEvent_AckType_ImmediateAck, + viopath_sourceinst(viopath_hostLp), + viopath_targetinst(viopath_hostLp), + 0, VIOVERSION << 16, + ((u64)DEVICE_NO(d) << 48) /* | ((u64)flags << 32) */, + 0, 0, 0); + if (hvrc != 0) + printk(VIOD_KERN_WARNING "HV close call failed %d\n", + (int)hvrc); + return 0; +} + + +/* External ioctl entry point. + */ +static int viodasd_ioctl(struct inode *ino, struct file *fil, + unsigned int cmd, unsigned long arg) +{ + unsigned char sectors; + unsigned char heads; + unsigned short cylinders; + struct hd_geometry *geo; + struct gendisk *gendisk; + struct viodasd_device *d; + + switch (cmd) { + case HDIO_GETGEO: + geo = (struct hd_geometry *)arg; + if (geo == NULL) + return -EINVAL; + if (!access_ok(VERIFY_WRITE, geo, sizeof(*geo))) + return -EFAULT; + gendisk = ino->i_bdev->bd_disk; + d = gendisk->private_data; + sectors = d->sectors; + if (sectors == 0) + sectors = 32; + heads = d->tracks; + if (heads == 0) + heads = 64; + cylinders = d->cylinders; + if (cylinders == 0) + cylinders = get_capacity(gendisk) / (sectors * heads); + if (__put_user(sectors, &geo->sectors) || + __put_user(heads, &geo->heads) || + __put_user(cylinders, &geo->cylinders) || + __put_user(get_start_sect(ino->i_bdev), &geo->start)) + return -EFAULT; + return 0; + } + + return -EINVAL; +} + +/* + * Our file operations table + */ +static struct block_device_operations viodasd_fops = { + .owner = THIS_MODULE, + .open = viodasd_open, + .release = viodasd_release, + .ioctl = viodasd_ioctl, +}; + +/* + * End a request + */ +static void viodasd_end_request(struct request *req, int uptodate, + int num_sectors) +{ + if (end_that_request_first(req, uptodate, num_sectors)) + return; + add_disk_randomness(req->rq_disk); + end_that_request_last(req); +} + +/* + * Send an actual I/O request to OS/400 + */ +static int send_request(struct request *req) +{ + u64 start; + int direction; + int nsg; + u16 viocmd; + HvLpEvent_Rc hvrc; + struct vioblocklpevent *bevent; + struct scatterlist sg[VIOMAXBLOCKDMA]; + int sgindex; + int statindex; + struct viodasd_device *d; + unsigned long flags; + + start = (u64)req->sector << 9; + + if (rq_data_dir(req) == READ) { + direction = DMA_FROM_DEVICE; + viocmd = viomajorsubtype_blockio | vioblockread; + statindex = 0; + } else { + direction = DMA_TO_DEVICE; + viocmd = viomajorsubtype_blockio | vioblockwrite; + statindex = 1; + } + + d = req->rq_disk->private_data; + + /* Now build the scatter-gather list */ + nsg = blk_rq_map_sg(req->q, req, sg); + nsg = dma_map_sg(d->dev, sg, nsg, direction); + + spin_lock_irqsave(&viodasd_spinlock, flags); + num_req_outstanding++; + + /* This optimization handles a single DMA block */ + if (nsg == 1) + hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, + HvLpEvent_Type_VirtualIo, viocmd, + HvLpEvent_AckInd_DoAck, + HvLpEvent_AckType_ImmediateAck, + viopath_sourceinst(viopath_hostLp), + viopath_targetinst(viopath_hostLp), + (u64)(unsigned long)req, VIOVERSION << 16, + ((u64)DEVICE_NO(d) << 48), start, + ((u64)sg_dma_address(&sg[0])) << 32, + sg_dma_len(&sg[0])); + else { + bevent = (struct vioblocklpevent *) + vio_get_event_buffer(viomajorsubtype_blockio); + if (bevent == NULL) { + printk(VIOD_KERN_WARNING + "error allocating disk event buffer\n"); + goto error_ret; + } + + /* + * Now build up the actual request. Note that we store + * the pointer to the request in the correlation + * token so we can match the response up later + */ + memset(bevent, 0, sizeof(struct vioblocklpevent)); + bevent->event.xFlags.xValid = 1; + bevent->event.xFlags.xFunction = HvLpEvent_Function_Int; + bevent->event.xFlags.xAckInd = HvLpEvent_AckInd_DoAck; + bevent->event.xFlags.xAckType = HvLpEvent_AckType_ImmediateAck; + bevent->event.xType = HvLpEvent_Type_VirtualIo; + bevent->event.xSubtype = viocmd; + bevent->event.xSourceLp = HvLpConfig_getLpIndex(); + bevent->event.xTargetLp = viopath_hostLp; + bevent->event.xSizeMinus1 = + offsetof(struct vioblocklpevent, u.rw_data.dma_info) + + (sizeof(bevent->u.rw_data.dma_info[0]) * nsg) - 1; + bevent->event.xSourceInstanceId = + viopath_sourceinst(viopath_hostLp); + bevent->event.xTargetInstanceId = + viopath_targetinst(viopath_hostLp); + bevent->event.xCorrelationToken = (u64)req; + bevent->version = VIOVERSION; + bevent->disk = DEVICE_NO(d); + bevent->u.rw_data.offset = start; + + /* + * Copy just the dma information from the sg list + * into the request + */ + for (sgindex = 0; sgindex < nsg; sgindex++) { + bevent->u.rw_data.dma_info[sgindex].token = + sg_dma_address(&sg[sgindex]); + bevent->u.rw_data.dma_info[sgindex].len = + sg_dma_len(&sg[sgindex]); + } + + /* Send the request */ + hvrc = HvCallEvent_signalLpEvent(&bevent->event); + vio_free_event_buffer(viomajorsubtype_blockio, bevent); + } + + if (hvrc != HvLpEvent_Rc_Good) { + printk(VIOD_KERN_WARNING + "error sending disk event to OS/400 (rc %d)\n", + (int)hvrc); + goto error_ret; + } + spin_unlock_irqrestore(&viodasd_spinlock, flags); + return 0; + +error_ret: + num_req_outstanding--; + spin_unlock_irqrestore(&viodasd_spinlock, flags); + dma_unmap_sg(d->dev, sg, nsg, direction); + return -1; +} + +/* + * This is the external request processing routine + */ +static void do_viodasd_request(request_queue_t *q) +{ + struct request *req; + + /* + * If we already have the maximum number of requests + * outstanding to OS/400 just bail out. We'll come + * back later. + */ + while (num_req_outstanding < VIOMAXREQ) { + req = elv_next_request(q); + if (req == NULL) + return; + /* dequeue the current request from the queue */ + blkdev_dequeue_request(req); + /* check that request contains a valid command */ + if (!blk_fs_request(req)) { + viodasd_end_request(req, 0, req->hard_nr_sectors); + continue; + } + /* Try sending the request */ + if (send_request(req) != 0) + viodasd_end_request(req, 0, req->hard_nr_sectors); + } +} + +/* + * Probe a single disk and fill in the viodasd_device structure + * for it. + */ +static void probe_disk(struct viodasd_device *d) +{ + HvLpEvent_Rc hvrc; + struct viodasd_waitevent we; + int dev_no = DEVICE_NO(d); + struct gendisk *g; + struct request_queue *q; + u16 flags = 0; + +retry: + init_completion(&we.com); + + /* Send the open event to OS/400 */ + hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, + HvLpEvent_Type_VirtualIo, + viomajorsubtype_blockio | vioblockopen, + HvLpEvent_AckInd_DoAck, HvLpEvent_AckType_ImmediateAck, + viopath_sourceinst(viopath_hostLp), + viopath_targetinst(viopath_hostLp), + (u64)(unsigned long)&we, VIOVERSION << 16, + ((u64)dev_no << 48) | ((u64)flags<< 32), + 0, 0, 0); + if (hvrc != 0) { + printk(VIOD_KERN_WARNING "bad rc on HV open %d\n", (int)hvrc); + return; + } + + wait_for_completion(&we.com); + + if (we.rc != 0) { + if (flags != 0) + return; + /* try again with read only flag set */ + flags = vioblockflags_ro; + goto retry; + } + if (we.max_disk > (MAX_DISKNO - 1)) { + static int warned; + + if (warned == 0) { + warned++; + printk(VIOD_KERN_INFO + "Only examining the first %d " + "of %d disks connected\n", + MAX_DISKNO, we.max_disk + 1); + } + } + + /* Send the close event to OS/400. We DON'T expect a response */ + hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, + HvLpEvent_Type_VirtualIo, + viomajorsubtype_blockio | vioblockclose, + HvLpEvent_AckInd_NoAck, HvLpEvent_AckType_ImmediateAck, + viopath_sourceinst(viopath_hostLp), + viopath_targetinst(viopath_hostLp), + 0, VIOVERSION << 16, + ((u64)dev_no << 48) | ((u64)flags << 32), + 0, 0, 0); + if (hvrc != 0) { + printk(VIOD_KERN_WARNING + "bad rc sending event to OS/400 %d\n", (int)hvrc); + return; + } + /* create the request queue for the disk */ + spin_lock_init(&d->q_lock); + q = blk_init_queue(do_viodasd_request, &d->q_lock); + if (q == NULL) { + printk(VIOD_KERN_WARNING "cannot allocate queue for disk %d\n", + dev_no); + return; + } + g = alloc_disk(1 << PARTITION_SHIFT); + if (g == NULL) { + printk(VIOD_KERN_WARNING + "cannot allocate disk structure for disk %d\n", + dev_no); + blk_cleanup_queue(q); + return; + } + + d->disk = g; + blk_queue_max_hw_segments(q, VIOMAXBLOCKDMA); + blk_queue_max_phys_segments(q, VIOMAXBLOCKDMA); + blk_queue_max_sectors(q, VIODASD_MAXSECTORS); + g->major = VIODASD_MAJOR; + g->first_minor = dev_no << PARTITION_SHIFT; + if (dev_no >= 26) + snprintf(g->disk_name, sizeof(g->disk_name), + VIOD_GENHD_NAME "%c%c", + 'a' + (dev_no / 26) - 1, 'a' + (dev_no % 26)); + else + snprintf(g->disk_name, sizeof(g->disk_name), + VIOD_GENHD_NAME "%c", 'a' + (dev_no % 26)); + snprintf(g->devfs_name, sizeof(g->devfs_name), + "%s%d", VIOD_GENHD_DEVFS_NAME, dev_no); + g->fops = &viodasd_fops; + g->queue = q; + g->private_data = d; + g->driverfs_dev = d->dev; + set_capacity(g, d->size >> 9); + + printk(VIOD_KERN_INFO "disk %d: %lu sectors (%lu MB) " + "CHS=%d/%d/%d sector size %d%s\n", + dev_no, (unsigned long)(d->size >> 9), + (unsigned long)(d->size >> 20), + (int)d->cylinders, (int)d->tracks, + (int)d->sectors, (int)d->bytes_per_sector, + d->read_only ? " (RO)" : ""); + + /* register us in the global list */ + add_disk(g); +} + +/* returns the total number of scatterlist elements converted */ +static int block_event_to_scatterlist(const struct vioblocklpevent *bevent, + struct scatterlist *sg, int *total_len) +{ + int i, numsg; + const struct rw_data *rw_data = &bevent->u.rw_data; + static const int offset = + offsetof(struct vioblocklpevent, u.rw_data.dma_info); + static const int element_size = sizeof(rw_data->dma_info[0]); + + numsg = ((bevent->event.xSizeMinus1 + 1) - offset) / element_size; + if (numsg > VIOMAXBLOCKDMA) + numsg = VIOMAXBLOCKDMA; + + *total_len = 0; + memset(sg, 0, sizeof(sg[0]) * VIOMAXBLOCKDMA); + + for (i = 0; (i < numsg) && (rw_data->dma_info[i].len > 0); ++i) { + sg_dma_address(&sg[i]) = rw_data->dma_info[i].token; + sg_dma_len(&sg[i]) = rw_data->dma_info[i].len; + *total_len += rw_data->dma_info[i].len; + } + return i; +} + +/* + * Restart all queues, starting with the one _after_ the disk given, + * thus reducing the chance of starvation of higher numbered disks. + */ +static void viodasd_restart_all_queues_starting_from(int first_index) +{ + int i; + + for (i = first_index + 1; i < MAX_DISKNO; ++i) + if (viodasd_devices[i].disk) + blk_run_queue(viodasd_devices[i].disk->queue); + for (i = 0; i <= first_index; ++i) + if (viodasd_devices[i].disk) + blk_run_queue(viodasd_devices[i].disk->queue); +} + +/* + * For read and write requests, decrement the number of outstanding requests, + * Free the DMA buffers we allocated. + */ +static int viodasd_handle_read_write(struct vioblocklpevent *bevent) +{ + int num_sg, num_sect, pci_direction, total_len; + struct request *req; + struct scatterlist sg[VIOMAXBLOCKDMA]; + struct HvLpEvent *event = &bevent->event; + unsigned long irq_flags; + struct viodasd_device *d; + int error; + spinlock_t *qlock; + + num_sg = block_event_to_scatterlist(bevent, sg, &total_len); + num_sect = total_len >> 9; + if (event->xSubtype == (viomajorsubtype_blockio | vioblockread)) + pci_direction = DMA_FROM_DEVICE; + else + pci_direction = DMA_TO_DEVICE; + req = (struct request *)bevent->event.xCorrelationToken; + d = req->rq_disk->private_data; + + dma_unmap_sg(d->dev, sg, num_sg, pci_direction); + + /* + * Since this is running in interrupt mode, we need to make sure + * we're not stepping on any global I/O operations + */ + spin_lock_irqsave(&viodasd_spinlock, irq_flags); + num_req_outstanding--; + spin_unlock_irqrestore(&viodasd_spinlock, irq_flags); + + error = event->xRc != HvLpEvent_Rc_Good; + if (error) { + const struct vio_error_entry *err; + err = vio_lookup_rc(viodasd_err_table, bevent->sub_result); + printk(VIOD_KERN_WARNING "read/write error %d:0x%04x (%s)\n", + event->xRc, bevent->sub_result, err->msg); + num_sect = req->hard_nr_sectors; + } + qlock = req->q->queue_lock; + spin_lock_irqsave(qlock, irq_flags); + viodasd_end_request(req, !error, num_sect); + spin_unlock_irqrestore(qlock, irq_flags); + + /* Finally, try to get more requests off of this device's queue */ + viodasd_restart_all_queues_starting_from(DEVICE_NO(d)); + + return 0; +} + +/* This routine handles incoming block LP events */ +static void handle_block_event(struct HvLpEvent *event) +{ + struct vioblocklpevent *bevent = (struct vioblocklpevent *)event; + struct viodasd_waitevent *pwe; + + if (event == NULL) + /* Notification that a partition went away! */ + return; + /* First, we should NEVER get an int here...only acks */ + if (event->xFlags.xFunction == HvLpEvent_Function_Int) { + printk(VIOD_KERN_WARNING + "Yikes! got an int in viodasd event handler!\n"); + if (event->xFlags.xAckInd == HvLpEvent_AckInd_DoAck) { + event->xRc = HvLpEvent_Rc_InvalidSubtype; + HvCallEvent_ackLpEvent(event); + } + } + + switch (event->xSubtype & VIOMINOR_SUBTYPE_MASK) { + case vioblockopen: + /* + * Handle a response to an open request. We get all the + * disk information in the response, so update it. The + * correlation token contains a pointer to a waitevent + * structure that has a completion in it. update the + * return code in the waitevent structure and post the + * completion to wake up the guy who sent the request + */ + pwe = (struct viodasd_waitevent *)event->xCorrelationToken; + pwe->rc = event->xRc; + pwe->sub_result = bevent->sub_result; + if (event->xRc == HvLpEvent_Rc_Good) { + const struct open_data *data = &bevent->u.open_data; + struct viodasd_device *device = + &viodasd_devices[bevent->disk]; + device->read_only = + bevent->flags & vioblockflags_ro; + device->size = data->disk_size; + device->cylinders = data->cylinders; + device->tracks = data->tracks; + device->sectors = data->sectors; + device->bytes_per_sector = data->bytes_per_sector; + pwe->max_disk = data->max_disk; + } + complete(&pwe->com); + break; + case vioblockclose: + break; + case vioblockread: + case vioblockwrite: + viodasd_handle_read_write(bevent); + break; + + default: + printk(VIOD_KERN_WARNING "invalid subtype!"); + if (event->xFlags.xAckInd == HvLpEvent_AckInd_DoAck) { + event->xRc = HvLpEvent_Rc_InvalidSubtype; + HvCallEvent_ackLpEvent(event); + } + } +} + +/* + * Get the driver to reprobe for more disks. + */ +static ssize_t probe_disks(struct device_driver *drv, const char *buf, + size_t count) +{ + struct viodasd_device *d; + + for (d = viodasd_devices; d < &viodasd_devices[MAX_DISKNO]; d++) { + if (d->disk == NULL) + probe_disk(d); + } + return count; +} +static DRIVER_ATTR(probe, S_IWUSR, NULL, probe_disks); + +static int viodasd_probe(struct vio_dev *vdev, const struct vio_device_id *id) +{ + struct viodasd_device *d = &viodasd_devices[vdev->unit_address]; + + d->dev = &vdev->dev; + probe_disk(d); + if (d->disk == NULL) + return -ENODEV; + return 0; +} + +static int viodasd_remove(struct vio_dev *vdev) +{ + struct viodasd_device *d; + + d = &viodasd_devices[vdev->unit_address]; + if (d->disk) { + del_gendisk(d->disk); + blk_cleanup_queue(d->disk->queue); + put_disk(d->disk); + d->disk = NULL; + } + d->dev = NULL; + return 0; +} + +/** + * viodasd_device_table: Used by vio.c to match devices that we + * support. + */ +static struct vio_device_id viodasd_device_table[] __devinitdata = { + { "viodasd", "" }, + { 0, } +}; + +MODULE_DEVICE_TABLE(vio, viodasd_device_table); +static struct vio_driver viodasd_driver = { + .name = "viodasd", + .id_table = viodasd_device_table, + .probe = viodasd_probe, + .remove = viodasd_remove +}; + +/* + * Initialize the whole device driver. Handle module and non-module + * versions + */ +static int __init viodasd_init(void) +{ + int rc; + + /* Try to open to our host lp */ + if (viopath_hostLp == HvLpIndexInvalid) + vio_set_hostlp(); + + if (viopath_hostLp == HvLpIndexInvalid) { + printk(VIOD_KERN_WARNING "invalid hosting partition\n"); + return -EIO; + } + + printk(VIOD_KERN_INFO "vers " VIOD_VERS ", hosting partition %d\n", + viopath_hostLp); + + /* register the block device */ + if (register_blkdev(VIODASD_MAJOR, VIOD_GENHD_NAME)) { + printk(VIOD_KERN_WARNING + "Unable to get major number %d for %s\n", + VIODASD_MAJOR, VIOD_GENHD_NAME); + return -EIO; + } + /* Actually open the path to the hosting partition */ + if (viopath_open(viopath_hostLp, viomajorsubtype_blockio, + VIOMAXREQ + 2)) { + printk(VIOD_KERN_WARNING + "error opening path to host partition %d\n", + viopath_hostLp); + unregister_blkdev(VIODASD_MAJOR, VIOD_GENHD_NAME); + return -EIO; + } + + /* Initialize our request handler */ + vio_setHandler(viomajorsubtype_blockio, handle_block_event); + + rc = vio_register_driver(&viodasd_driver); + if (rc == 0) + driver_create_file(&viodasd_driver.driver, &driver_attr_probe); + return rc; +} +module_init(viodasd_init); + +void viodasd_exit(void) +{ + driver_remove_file(&viodasd_driver.driver, &driver_attr_probe); + vio_unregister_driver(&viodasd_driver); + vio_clearHandler(viomajorsubtype_blockio); + unregister_blkdev(VIODASD_MAJOR, VIOD_GENHD_NAME); + viopath_close(viopath_hostLp, viomajorsubtype_blockio, VIOMAXREQ + 2); +} + +module_exit(viodasd_exit); diff --git a/drivers/block/xd.c b/drivers/block/xd.c new file mode 100644 index 00000000000..1676033da6c --- /dev/null +++ b/drivers/block/xd.c @@ -0,0 +1,1112 @@ +/* + * This file contains the driver for an XT hard disk controller + * (at least the DTC 5150X) for Linux. + * + * Author: Pat Mackinlay, pat@it.com.au + * Date: 29/09/92 + * + * Revised: 01/01/93, ... + * + * Ref: DTC 5150X Controller Specification (thanks to Kevin Fowler, + * kevinf@agora.rain.com) + * Also thanks to: Salvador Abreu, Dave Thaler, Risto Kankkunen and + * Wim Van Dorst. + * + * Revised: 04/04/94 by Risto Kankkunen + * Moved the detection code from xd_init() to xd_geninit() as it needed + * interrupts enabled and Linus didn't want to enable them in that first + * phase. xd_geninit() is the place to do these kinds of things anyway, + * he says. + * + * Modularized: 04/10/96 by Todd Fries, tfries@umr.edu + * + * Revised: 13/12/97 by Andrzej Krzysztofowicz, ankry@mif.pg.gda.pl + * Fixed some problems with disk initialization and module initiation. + * Added support for manual geometry setting (except Seagate controllers) + * in form: + * xd_geo=<cyl_xda>,<head_xda>,<sec_xda>[,<cyl_xdb>,<head_xdb>,<sec_xdb>] + * Recovered DMA access. Abridged messages. Added support for DTC5051CX, + * WD1002-27X & XEBEC controllers. Driver uses now some jumper settings. + * Extended ioctl() support. + * + * Bugfix: 15/02/01, Paul G. - inform queue layer of tiny xd_maxsect. + * + */ + +#include <linux/module.h> +#include <linux/errno.h> +#include <linux/interrupt.h> +#include <linux/mm.h> +#include <linux/fs.h> +#include <linux/kernel.h> +#include <linux/timer.h> +#include <linux/genhd.h> +#include <linux/hdreg.h> +#include <linux/ioport.h> +#include <linux/init.h> +#include <linux/wait.h> +#include <linux/blkdev.h> +#include <linux/blkpg.h> + +#include <asm/system.h> +#include <asm/io.h> +#include <asm/uaccess.h> +#include <asm/dma.h> + +#include "xd.h" + +static void __init do_xd_setup (int *integers); +#ifdef MODULE +static int xd[5] = { -1,-1,-1,-1, }; +#endif + +#define XD_DONT_USE_DMA 0 /* Initial value. may be overriden using + "nodma" module option */ +#define XD_INIT_DISK_DELAY (30*HZ/1000) /* 30 ms delay during disk initialization */ + +/* Above may need to be increased if a problem with the 2nd drive detection + (ST11M controller) or resetting a controller (WD) appears */ + +static XD_INFO xd_info[XD_MAXDRIVES]; + +/* If you try this driver and find that your card is not detected by the driver at bootup, you need to add your BIOS + signature and details to the following list of signatures. A BIOS signature is a string embedded into the first + few bytes of your controller's on-board ROM BIOS. To find out what yours is, use something like MS-DOS's DEBUG + command. Run DEBUG, and then you can examine your BIOS signature with: + + d xxxx:0000 + + where xxxx is the segment of your controller (like C800 or D000 or something). On the ASCII dump at the right, you should + be able to see a string mentioning the manufacturer's copyright etc. Add this string into the table below. The parameters + in the table are, in order: + + offset ; this is the offset (in bytes) from the start of your ROM where the signature starts + signature ; this is the actual text of the signature + xd_?_init_controller ; this is the controller init routine used by your controller + xd_?_init_drive ; this is the drive init routine used by your controller + + The controllers directly supported at the moment are: DTC 5150x, WD 1004A27X, ST11M/R and override. If your controller is + made by the same manufacturer as one of these, try using the same init routines as they do. If that doesn't work, your + best bet is to use the "override" routines. These routines use a "portable" method of getting the disk's geometry, and + may work with your card. If none of these seem to work, try sending me some email and I'll see what I can do <grin>. + + NOTE: You can now specify your XT controller's parameters from the command line in the form xd=TYPE,IRQ,IO,DMA. The driver + should be able to detect your drive's geometry from this info. (eg: xd=0,5,0x320,3 is the "standard"). */ + +#include <asm/page.h> +#define xd_dma_mem_alloc(size) __get_dma_pages(GFP_KERNEL,get_order(size)) +#define xd_dma_mem_free(addr, size) free_pages(addr, get_order(size)) +static char *xd_dma_buffer; + +static XD_SIGNATURE xd_sigs[] __initdata = { + { 0x0000,"Override geometry handler",NULL,xd_override_init_drive,"n unknown" }, /* Pat Mackinlay, pat@it.com.au */ + { 0x0008,"[BXD06 (C) DTC 17-MAY-1985]",xd_dtc_init_controller,xd_dtc5150cx_init_drive," DTC 5150CX" }, /* Andrzej Krzysztofowicz, ankry@mif.pg.gda.pl */ + { 0x000B,"CRD18A Not an IBM rom. (C) Copyright Data Technology Corp. 05/31/88",xd_dtc_init_controller,xd_dtc_init_drive," DTC 5150X" }, /* Todd Fries, tfries@umr.edu */ + { 0x000B,"CXD23A Not an IBM ROM (C)Copyright Data Technology Corp 12/03/88",xd_dtc_init_controller,xd_dtc_init_drive," DTC 5150X" }, /* Pat Mackinlay, pat@it.com.au */ + { 0x0008,"07/15/86(C) Copyright 1986 Western Digital Corp.",xd_wd_init_controller,xd_wd_init_drive," Western Dig. 1002-27X" }, /* Andrzej Krzysztofowicz, ankry@mif.pg.gda.pl */ + { 0x0008,"06/24/88(C) Copyright 1988 Western Digital Corp.",xd_wd_init_controller,xd_wd_init_drive," Western Dig. WDXT-GEN2" }, /* Dan Newcombe, newcombe@aa.csc.peachnet.edu */ + { 0x0015,"SEAGATE ST11 BIOS REVISION",xd_seagate_init_controller,xd_seagate_init_drive," Seagate ST11M/R" }, /* Salvador Abreu, spa@fct.unl.pt */ + { 0x0010,"ST11R BIOS",xd_seagate_init_controller,xd_seagate_init_drive," Seagate ST11M/R" }, /* Risto Kankkunen, risto.kankkunen@cs.helsinki.fi */ + { 0x0010,"ST11 BIOS v1.7",xd_seagate_init_controller,xd_seagate_init_drive," Seagate ST11R" }, /* Alan Hourihane, alanh@fairlite.demon.co.uk */ + { 0x1000,"(c)Copyright 1987 SMS",xd_omti_init_controller,xd_omti_init_drive,"n OMTI 5520" }, /* Dirk Melchers, dirk@merlin.nbg.sub.org */ + { 0x0006,"COPYRIGHT XEBEC (C) 1984",xd_xebec_init_controller,xd_xebec_init_drive," XEBEC" }, /* Andrzej Krzysztofowicz, ankry@mif.pg.gda.pl */ + { 0x0008,"(C) Copyright 1984 Western Digital Corp", xd_wd_init_controller, xd_wd_init_drive," Western Dig. 1002s-wx2" }, + { 0x0008,"(C) Copyright 1986 Western Digital Corporation", xd_wd_init_controller, xd_wd_init_drive," 1986 Western Digital" }, /* jfree@sovereign.org */ +}; + +static unsigned int xd_bases[] __initdata = +{ + 0xC8000, 0xCA000, 0xCC000, + 0xCE000, 0xD0000, 0xD2000, + 0xD4000, 0xD6000, 0xD8000, + 0xDA000, 0xDC000, 0xDE000, + 0xE0000 +}; + +static DEFINE_SPINLOCK(xd_lock); + +static struct gendisk *xd_gendisk[2]; + +static struct block_device_operations xd_fops = { + .owner = THIS_MODULE, + .ioctl = xd_ioctl, +}; +static DECLARE_WAIT_QUEUE_HEAD(xd_wait_int); +static u_char xd_drives, xd_irq = 5, xd_dma = 3, xd_maxsectors; +static u_char xd_override __initdata = 0, xd_type __initdata = 0; +static u_short xd_iobase = 0x320; +static int xd_geo[XD_MAXDRIVES*3] __initdata = { 0, }; + +static volatile int xdc_busy; +static struct timer_list xd_watchdog_int; + +static volatile u_char xd_error; +static int nodma = XD_DONT_USE_DMA; + +static struct request_queue *xd_queue; + +/* xd_init: register the block device number and set up pointer tables */ +static int __init xd_init(void) +{ + u_char i,controller; + unsigned int address; + int err; + +#ifdef MODULE + { + u_char count = 0; + for (i = 4; i > 0; i--) + if (((xd[i] = xd[i-1]) >= 0) && !count) + count = i; + if ((xd[0] = count)) + do_xd_setup(xd); + } +#endif + + init_timer (&xd_watchdog_int); xd_watchdog_int.function = xd_watchdog; + + if (!xd_dma_buffer) + xd_dma_buffer = (char *)xd_dma_mem_alloc(xd_maxsectors * 0x200); + if (!xd_dma_buffer) { + printk(KERN_ERR "xd: Out of memory.\n"); + return -ENOMEM; + } + + err = -EBUSY; + if (register_blkdev(XT_DISK_MAJOR, "xd")) + goto out1; + + err = -ENOMEM; + xd_queue = blk_init_queue(do_xd_request, &xd_lock); + if (!xd_queue) + goto out1a; + + if (xd_detect(&controller,&address)) { + + printk("Detected a%s controller (type %d) at address %06x\n", + xd_sigs[controller].name,controller,address); + if (!request_region(xd_iobase,4,"xd")) { + printk("xd: Ports at 0x%x are not available\n", + xd_iobase); + goto out2; + } + if (controller) + xd_sigs[controller].init_controller(address); + xd_drives = xd_initdrives(xd_sigs[controller].init_drive); + + printk("Detected %d hard drive%s (using IRQ%d & DMA%d)\n", + xd_drives,xd_drives == 1 ? "" : "s",xd_irq,xd_dma); + } + + err = -ENODEV; + if (!xd_drives) + goto out3; + + for (i = 0; i < xd_drives; i++) { + XD_INFO *p = &xd_info[i]; + struct gendisk *disk = alloc_disk(64); + if (!disk) + goto Enomem; + p->unit = i; + disk->major = XT_DISK_MAJOR; + disk->first_minor = i<<6; + sprintf(disk->disk_name, "xd%c", i+'a'); + sprintf(disk->devfs_name, "xd/target%d", i); + disk->fops = &xd_fops; + disk->private_data = p; + disk->queue = xd_queue; + set_capacity(disk, p->heads * p->cylinders * p->sectors); + printk(" %s: CHS=%d/%d/%d\n", disk->disk_name, + p->cylinders, p->heads, p->sectors); + xd_gendisk[i] = disk; + } + + err = -EBUSY; + if (request_irq(xd_irq,xd_interrupt_handler, 0, "XT hard disk", NULL)) { + printk("xd: unable to get IRQ%d\n",xd_irq); + goto out4; + } + + if (request_dma(xd_dma,"xd")) { + printk("xd: unable to get DMA%d\n",xd_dma); + goto out5; + } + + /* xd_maxsectors depends on controller - so set after detection */ + blk_queue_max_sectors(xd_queue, xd_maxsectors); + + for (i = 0; i < xd_drives; i++) + add_disk(xd_gendisk[i]); + + return 0; + +out5: + free_irq(xd_irq, NULL); +out4: + for (i = 0; i < xd_drives; i++) + put_disk(xd_gendisk[i]); +out3: + release_region(xd_iobase,4); +out2: + blk_cleanup_queue(xd_queue); +out1a: + unregister_blkdev(XT_DISK_MAJOR, "xd"); +out1: + if (xd_dma_buffer) + xd_dma_mem_free((unsigned long)xd_dma_buffer, + xd_maxsectors * 0x200); + return err; +Enomem: + err = -ENOMEM; + while (i--) + put_disk(xd_gendisk[i]); + goto out3; +} + +/* xd_detect: scan the possible BIOS ROM locations for the signature strings */ +static u_char __init xd_detect (u_char *controller, unsigned int *address) +{ + int i, j; + + if (xd_override) + { + *controller = xd_type; + *address = 0; + return(1); + } + + for (i = 0; i < (sizeof(xd_bases) / sizeof(xd_bases[0])); i++) { + void __iomem *p = ioremap(xd_bases[i], 0x2000); + if (!p) + continue; + for (j = 1; j < (sizeof(xd_sigs) / sizeof(xd_sigs[0])); j++) { + const char *s = xd_sigs[j].string; + if (check_signature(p + xd_sigs[j].offset, s, strlen(s))) { + *controller = j; + xd_type = j; + *address = xd_bases[i]; + iounmap(p); + return 1; + } + } + iounmap(p); + } + return 0; +} + +/* do_xd_request: handle an incoming request */ +static void do_xd_request (request_queue_t * q) +{ + struct request *req; + + if (xdc_busy) + return; + + while ((req = elv_next_request(q)) != NULL) { + unsigned block = req->sector; + unsigned count = req->nr_sectors; + int rw = rq_data_dir(req); + XD_INFO *disk = req->rq_disk->private_data; + int res = 0; + int retry; + + if (!(req->flags & REQ_CMD)) { + end_request(req, 0); + continue; + } + if (block + count > get_capacity(req->rq_disk)) { + end_request(req, 0); + continue; + } + if (rw != READ && rw != WRITE) { + printk("do_xd_request: unknown request\n"); + end_request(req, 0); + continue; + } + for (retry = 0; (retry < XD_RETRIES) && !res; retry++) + res = xd_readwrite(rw, disk, req->buffer, block, count); + end_request(req, res); /* wrap up, 0 = fail, 1 = success */ + } +} + +/* xd_ioctl: handle device ioctl's */ +static int xd_ioctl (struct inode *inode,struct file *file,u_int cmd,u_long arg) +{ + XD_INFO *p = inode->i_bdev->bd_disk->private_data; + + switch (cmd) { + case HDIO_GETGEO: + { + struct hd_geometry g; + struct hd_geometry __user *geom= (void __user *)arg; + g.heads = p->heads; + g.sectors = p->sectors; + g.cylinders = p->cylinders; + g.start = get_start_sect(inode->i_bdev); + return copy_to_user(geom, &g, sizeof(g)) ? -EFAULT : 0; + } + case HDIO_SET_DMA: + if (!capable(CAP_SYS_ADMIN)) return -EACCES; + if (xdc_busy) return -EBUSY; + nodma = !arg; + if (nodma && xd_dma_buffer) { + xd_dma_mem_free((unsigned long)xd_dma_buffer, + xd_maxsectors * 0x200); + xd_dma_buffer = NULL; + } else if (!nodma && !xd_dma_buffer) { + xd_dma_buffer = (char *)xd_dma_mem_alloc(xd_maxsectors * 0x200); + if (!xd_dma_buffer) { + nodma = XD_DONT_USE_DMA; + return -ENOMEM; + } + } + return 0; + case HDIO_GET_DMA: + return put_user(!nodma, (long __user *) arg); + case HDIO_GET_MULTCOUNT: + return put_user(xd_maxsectors, (long __user *) arg); + default: + return -EINVAL; + } +} + +/* xd_readwrite: handle a read/write request */ +static int xd_readwrite (u_char operation,XD_INFO *p,char *buffer,u_int block,u_int count) +{ + int drive = p->unit; + u_char cmdblk[6],sense[4]; + u_short track,cylinder; + u_char head,sector,control,mode = PIO_MODE,temp; + char **real_buffer; + register int i; + +#ifdef DEBUG_READWRITE + printk("xd_readwrite: operation = %s, drive = %d, buffer = 0x%X, block = %d, count = %d\n",operation == READ ? "read" : "write",drive,buffer,block,count); +#endif /* DEBUG_READWRITE */ + + spin_unlock_irq(&xd_lock); + + control = p->control; + if (!xd_dma_buffer) + xd_dma_buffer = (char *)xd_dma_mem_alloc(xd_maxsectors * 0x200); + while (count) { + temp = count < xd_maxsectors ? count : xd_maxsectors; + + track = block / p->sectors; + head = track % p->heads; + cylinder = track / p->heads; + sector = block % p->sectors; + +#ifdef DEBUG_READWRITE + printk("xd_readwrite: drive = %d, head = %d, cylinder = %d, sector = %d, count = %d\n",drive,head,cylinder,sector,temp); +#endif /* DEBUG_READWRITE */ + + if (xd_dma_buffer) { + mode = xd_setup_dma(operation == READ ? DMA_MODE_READ : DMA_MODE_WRITE,(u_char *)(xd_dma_buffer),temp * 0x200); + real_buffer = &xd_dma_buffer; + for (i=0; i < (temp * 0x200); i++) + xd_dma_buffer[i] = buffer[i]; + } + else + real_buffer = &buffer; + + xd_build(cmdblk,operation == READ ? CMD_READ : CMD_WRITE,drive,head,cylinder,sector,temp & 0xFF,control); + + switch (xd_command(cmdblk,mode,(u_char *)(*real_buffer),(u_char *)(*real_buffer),sense,XD_TIMEOUT)) { + case 1: + printk("xd%c: %s timeout, recalibrating drive\n",'a'+drive,(operation == READ ? "read" : "write")); + xd_recalibrate(drive); + spin_lock_irq(&xd_lock); + return (0); + case 2: + if (sense[0] & 0x30) { + printk("xd%c: %s - ",'a'+drive,(operation == READ ? "reading" : "writing")); + switch ((sense[0] & 0x30) >> 4) { + case 0: printk("drive error, code = 0x%X",sense[0] & 0x0F); + break; + case 1: printk("controller error, code = 0x%X",sense[0] & 0x0F); + break; + case 2: printk("command error, code = 0x%X",sense[0] & 0x0F); + break; + case 3: printk("miscellaneous error, code = 0x%X",sense[0] & 0x0F); + break; + } + } + if (sense[0] & 0x80) + printk(" - CHS = %d/%d/%d\n",((sense[2] & 0xC0) << 2) | sense[3],sense[1] & 0x1F,sense[2] & 0x3F); + /* reported drive number = (sense[1] & 0xE0) >> 5 */ + else + printk(" - no valid disk address\n"); + spin_lock_irq(&xd_lock); + return (0); + } + if (xd_dma_buffer) + for (i=0; i < (temp * 0x200); i++) + buffer[i] = xd_dma_buffer[i]; + + count -= temp, buffer += temp * 0x200, block += temp; + } + spin_lock_irq(&xd_lock); + return (1); +} + +/* xd_recalibrate: recalibrate a given drive and reset controller if necessary */ +static void xd_recalibrate (u_char drive) +{ + u_char cmdblk[6]; + + xd_build(cmdblk,CMD_RECALIBRATE,drive,0,0,0,0,0); + if (xd_command(cmdblk,PIO_MODE,NULL,NULL,NULL,XD_TIMEOUT * 8)) + printk("xd%c: warning! error recalibrating, controller may be unstable\n", 'a'+drive); +} + +/* xd_interrupt_handler: interrupt service routine */ +static irqreturn_t xd_interrupt_handler(int irq, void *dev_id, + struct pt_regs *regs) +{ + if (inb(XD_STATUS) & STAT_INTERRUPT) { /* check if it was our device */ +#ifdef DEBUG_OTHER + printk("xd_interrupt_handler: interrupt detected\n"); +#endif /* DEBUG_OTHER */ + outb(0,XD_CONTROL); /* acknowledge interrupt */ + wake_up(&xd_wait_int); /* and wake up sleeping processes */ + return IRQ_HANDLED; + } + else + printk("xd: unexpected interrupt\n"); + return IRQ_NONE; +} + +/* xd_setup_dma: set up the DMA controller for a data transfer */ +static u_char xd_setup_dma (u_char mode,u_char *buffer,u_int count) +{ + unsigned long f; + + if (nodma) + return (PIO_MODE); + if (((unsigned long) buffer & 0xFFFF0000) != (((unsigned long) buffer + count) & 0xFFFF0000)) { +#ifdef DEBUG_OTHER + printk("xd_setup_dma: using PIO, transfer overlaps 64k boundary\n"); +#endif /* DEBUG_OTHER */ + return (PIO_MODE); + } + + f=claim_dma_lock(); + disable_dma(xd_dma); + clear_dma_ff(xd_dma); + set_dma_mode(xd_dma,mode); + set_dma_addr(xd_dma, (unsigned long) buffer); + set_dma_count(xd_dma,count); + + release_dma_lock(f); + + return (DMA_MODE); /* use DMA and INT */ +} + +/* xd_build: put stuff into an array in a format suitable for the controller */ +static u_char *xd_build (u_char *cmdblk,u_char command,u_char drive,u_char head,u_short cylinder,u_char sector,u_char count,u_char control) +{ + cmdblk[0] = command; + cmdblk[1] = ((drive & 0x07) << 5) | (head & 0x1F); + cmdblk[2] = ((cylinder & 0x300) >> 2) | (sector & 0x3F); + cmdblk[3] = cylinder & 0xFF; + cmdblk[4] = count; + cmdblk[5] = control; + + return (cmdblk); +} + +static void xd_watchdog (unsigned long unused) +{ + xd_error = 1; + wake_up(&xd_wait_int); +} + +/* xd_waitport: waits until port & mask == flags or a timeout occurs. return 1 for a timeout */ +static inline u_char xd_waitport (u_short port,u_char flags,u_char mask,u_long timeout) +{ + u_long expiry = jiffies + timeout; + int success; + + xdc_busy = 1; + while ((success = ((inb(port) & mask) != flags)) && time_before(jiffies, expiry)) { + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(1); + } + xdc_busy = 0; + return (success); +} + +static inline u_int xd_wait_for_IRQ (void) +{ + unsigned long flags; + xd_watchdog_int.expires = jiffies + 8 * HZ; + add_timer(&xd_watchdog_int); + + flags=claim_dma_lock(); + enable_dma(xd_dma); + release_dma_lock(flags); + + sleep_on(&xd_wait_int); + del_timer(&xd_watchdog_int); + xdc_busy = 0; + + flags=claim_dma_lock(); + disable_dma(xd_dma); + release_dma_lock(flags); + + if (xd_error) { + printk("xd: missed IRQ - command aborted\n"); + xd_error = 0; + return (1); + } + return (0); +} + +/* xd_command: handle all data transfers necessary for a single command */ +static u_int xd_command (u_char *command,u_char mode,u_char *indata,u_char *outdata,u_char *sense,u_long timeout) +{ + u_char cmdblk[6],csb,complete = 0; + +#ifdef DEBUG_COMMAND + printk("xd_command: command = 0x%X, mode = 0x%X, indata = 0x%X, outdata = 0x%X, sense = 0x%X\n",command,mode,indata,outdata,sense); +#endif /* DEBUG_COMMAND */ + + outb(0,XD_SELECT); + outb(mode,XD_CONTROL); + + if (xd_waitport(XD_STATUS,STAT_SELECT,STAT_SELECT,timeout)) + return (1); + + while (!complete) { + if (xd_waitport(XD_STATUS,STAT_READY,STAT_READY,timeout)) + return (1); + + switch (inb(XD_STATUS) & (STAT_COMMAND | STAT_INPUT)) { + case 0: + if (mode == DMA_MODE) { + if (xd_wait_for_IRQ()) + return (1); + } else + outb(outdata ? *outdata++ : 0,XD_DATA); + break; + case STAT_INPUT: + if (mode == DMA_MODE) { + if (xd_wait_for_IRQ()) + return (1); + } else + if (indata) + *indata++ = inb(XD_DATA); + else + inb(XD_DATA); + break; + case STAT_COMMAND: + outb(command ? *command++ : 0,XD_DATA); + break; + case STAT_COMMAND | STAT_INPUT: + complete = 1; + break; + } + } + csb = inb(XD_DATA); + + if (xd_waitport(XD_STATUS,0,STAT_SELECT,timeout)) /* wait until deselected */ + return (1); + + if (csb & CSB_ERROR) { /* read sense data if error */ + xd_build(cmdblk,CMD_SENSE,(csb & CSB_LUN) >> 5,0,0,0,0,0); + if (xd_command(cmdblk,0,sense,NULL,NULL,XD_TIMEOUT)) + printk("xd: warning! sense command failed!\n"); + } + +#ifdef DEBUG_COMMAND + printk("xd_command: completed with csb = 0x%X\n",csb); +#endif /* DEBUG_COMMAND */ + + return (csb & CSB_ERROR); +} + +static u_char __init xd_initdrives (void (*init_drive)(u_char drive)) +{ + u_char cmdblk[6],i,count = 0; + + for (i = 0; i < XD_MAXDRIVES; i++) { + xd_build(cmdblk,CMD_TESTREADY,i,0,0,0,0,0); + if (!xd_command(cmdblk,PIO_MODE,NULL,NULL,NULL,XD_TIMEOUT*8)) { + set_current_state(TASK_INTERRUPTIBLE); + schedule_timeout(XD_INIT_DISK_DELAY); + + init_drive(count); + count++; + + set_current_state(TASK_INTERRUPTIBLE); + schedule_timeout(XD_INIT_DISK_DELAY); + } + } + return (count); +} + +static void __init xd_manual_geo_set (u_char drive) +{ + xd_info[drive].heads = (u_char)(xd_geo[3 * drive + 1]); + xd_info[drive].cylinders = (u_short)(xd_geo[3 * drive]); + xd_info[drive].sectors = (u_char)(xd_geo[3 * drive + 2]); +} + +static void __init xd_dtc_init_controller (unsigned int address) +{ + switch (address) { + case 0x00000: + case 0xC8000: break; /*initial: 0x320 */ + case 0xCA000: xd_iobase = 0x324; + case 0xD0000: /*5150CX*/ + case 0xD8000: break; /*5150CX & 5150XL*/ + default: printk("xd_dtc_init_controller: unsupported BIOS address %06x\n",address); + break; + } + xd_maxsectors = 0x01; /* my card seems to have trouble doing multi-block transfers? */ + + outb(0,XD_RESET); /* reset the controller */ +} + + +static void __init xd_dtc5150cx_init_drive (u_char drive) +{ + /* values from controller's BIOS - BIOS chip may be removed */ + static u_short geometry_table[][4] = { + {0x200,8,0x200,0x100}, + {0x267,2,0x267,0x267}, + {0x264,4,0x264,0x80}, + {0x132,4,0x132,0x0}, + {0x132,2,0x80, 0x132}, + {0x177,8,0x177,0x0}, + {0x132,8,0x84, 0x0}, + {}, /* not used */ + {0x132,6,0x80, 0x100}, + {0x200,6,0x100,0x100}, + {0x264,2,0x264,0x80}, + {0x280,4,0x280,0x100}, + {0x2B9,3,0x2B9,0x2B9}, + {0x2B9,5,0x2B9,0x2B9}, + {0x280,6,0x280,0x100}, + {0x132,4,0x132,0x0}}; + u_char n; + + n = inb(XD_JUMPER); + n = (drive ? n : (n >> 2)) & 0x33; + n = (n | (n >> 2)) & 0x0F; + if (xd_geo[3*drive]) + xd_manual_geo_set(drive); + else + if (n != 7) { + xd_info[drive].heads = (u_char)(geometry_table[n][1]); /* heads */ + xd_info[drive].cylinders = geometry_table[n][0]; /* cylinders */ + xd_info[drive].sectors = 17; /* sectors */ +#if 0 + xd_info[drive].rwrite = geometry_table[n][2]; /* reduced write */ + xd_info[drive].precomp = geometry_table[n][3] /* write precomp */ + xd_info[drive].ecc = 0x0B; /* ecc length */ +#endif /* 0 */ + } + else { + printk("xd%c: undetermined drive geometry\n",'a'+drive); + return; + } + xd_info[drive].control = 5; /* control byte */ + xd_setparam(CMD_DTCSETPARAM,drive,xd_info[drive].heads,xd_info[drive].cylinders,geometry_table[n][2],geometry_table[n][3],0x0B); + xd_recalibrate(drive); +} + +static void __init xd_dtc_init_drive (u_char drive) +{ + u_char cmdblk[6],buf[64]; + + xd_build(cmdblk,CMD_DTCGETGEOM,drive,0,0,0,0,0); + if (!xd_command(cmdblk,PIO_MODE,buf,NULL,NULL,XD_TIMEOUT * 2)) { + xd_info[drive].heads = buf[0x0A]; /* heads */ + xd_info[drive].cylinders = ((u_short *) (buf))[0x04]; /* cylinders */ + xd_info[drive].sectors = 17; /* sectors */ + if (xd_geo[3*drive]) + xd_manual_geo_set(drive); +#if 0 + xd_info[drive].rwrite = ((u_short *) (buf + 1))[0x05]; /* reduced write */ + xd_info[drive].precomp = ((u_short *) (buf + 1))[0x06]; /* write precomp */ + xd_info[drive].ecc = buf[0x0F]; /* ecc length */ +#endif /* 0 */ + xd_info[drive].control = 0; /* control byte */ + + xd_setparam(CMD_DTCSETPARAM,drive,xd_info[drive].heads,xd_info[drive].cylinders,((u_short *) (buf + 1))[0x05],((u_short *) (buf + 1))[0x06],buf[0x0F]); + xd_build(cmdblk,CMD_DTCSETSTEP,drive,0,0,0,0,7); + if (xd_command(cmdblk,PIO_MODE,NULL,NULL,NULL,XD_TIMEOUT * 2)) + printk("xd_dtc_init_drive: error setting step rate for xd%c\n", 'a'+drive); + } + else + printk("xd_dtc_init_drive: error reading geometry for xd%c\n", 'a'+drive); +} + +static void __init xd_wd_init_controller (unsigned int address) +{ + switch (address) { + case 0x00000: + case 0xC8000: break; /*initial: 0x320 */ + case 0xCA000: xd_iobase = 0x324; break; + case 0xCC000: xd_iobase = 0x328; break; + case 0xCE000: xd_iobase = 0x32C; break; + case 0xD0000: xd_iobase = 0x328; break; /* ? */ + case 0xD8000: xd_iobase = 0x32C; break; /* ? */ + default: printk("xd_wd_init_controller: unsupported BIOS address %06x\n",address); + break; + } + xd_maxsectors = 0x01; /* this one doesn't wrap properly either... */ + + outb(0,XD_RESET); /* reset the controller */ + + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(XD_INIT_DISK_DELAY); +} + +static void __init xd_wd_init_drive (u_char drive) +{ + /* values from controller's BIOS - BIOS may be disabled */ + static u_short geometry_table[][4] = { + {0x264,4,0x1C2,0x1C2}, /* common part */ + {0x132,4,0x099,0x0}, + {0x267,2,0x1C2,0x1C2}, + {0x267,4,0x1C2,0x1C2}, + + {0x334,6,0x335,0x335}, /* 1004 series RLL */ + {0x30E,4,0x30F,0x3DC}, + {0x30E,2,0x30F,0x30F}, + {0x267,4,0x268,0x268}, + + {0x3D5,5,0x3D6,0x3D6}, /* 1002 series RLL */ + {0x3DB,7,0x3DC,0x3DC}, + {0x264,4,0x265,0x265}, + {0x267,4,0x268,0x268}}; + + u_char cmdblk[6],buf[0x200]; + u_char n = 0,rll,jumper_state,use_jumper_geo; + u_char wd_1002 = (xd_sigs[xd_type].string[7] == '6'); + + jumper_state = ~(inb(0x322)); + if (jumper_state & 0x40) + xd_irq = 9; + rll = (jumper_state & 0x30) ? (0x04 << wd_1002) : 0; + xd_build(cmdblk,CMD_READ,drive,0,0,0,1,0); + if (!xd_command(cmdblk,PIO_MODE,buf,NULL,NULL,XD_TIMEOUT * 2)) { + xd_info[drive].heads = buf[0x1AF]; /* heads */ + xd_info[drive].cylinders = ((u_short *) (buf + 1))[0xD6]; /* cylinders */ + xd_info[drive].sectors = 17; /* sectors */ + if (xd_geo[3*drive]) + xd_manual_geo_set(drive); +#if 0 + xd_info[drive].rwrite = ((u_short *) (buf))[0xD8]; /* reduced write */ + xd_info[drive].wprecomp = ((u_short *) (buf))[0xDA]; /* write precomp */ + xd_info[drive].ecc = buf[0x1B4]; /* ecc length */ +#endif /* 0 */ + xd_info[drive].control = buf[0x1B5]; /* control byte */ + use_jumper_geo = !(xd_info[drive].heads) || !(xd_info[drive].cylinders); + if (xd_geo[3*drive]) { + xd_manual_geo_set(drive); + xd_info[drive].control = rll ? 7 : 5; + } + else if (use_jumper_geo) { + n = (((jumper_state & 0x0F) >> (drive << 1)) & 0x03) | rll; + xd_info[drive].cylinders = geometry_table[n][0]; + xd_info[drive].heads = (u_char)(geometry_table[n][1]); + xd_info[drive].control = rll ? 7 : 5; +#if 0 + xd_info[drive].rwrite = geometry_table[n][2]; + xd_info[drive].wprecomp = geometry_table[n][3]; + xd_info[drive].ecc = 0x0B; +#endif /* 0 */ + } + if (!wd_1002) { + if (use_jumper_geo) + xd_setparam(CMD_WDSETPARAM,drive,xd_info[drive].heads,xd_info[drive].cylinders, + geometry_table[n][2],geometry_table[n][3],0x0B); + else + xd_setparam(CMD_WDSETPARAM,drive,xd_info[drive].heads,xd_info[drive].cylinders, + ((u_short *) (buf))[0xD8],((u_short *) (buf))[0xDA],buf[0x1B4]); + } + /* 1002 based RLL controller requests converted addressing, but reports physical + (physical 26 sec., logical 17 sec.) + 1004 based ???? */ + if (rll & wd_1002) { + if ((xd_info[drive].cylinders *= 26, + xd_info[drive].cylinders /= 17) > 1023) + xd_info[drive].cylinders = 1023; /* 1024 ? */ +#if 0 + xd_info[drive].rwrite *= 26; + xd_info[drive].rwrite /= 17; + xd_info[drive].wprecomp *= 26 + xd_info[drive].wprecomp /= 17; +#endif /* 0 */ + } + } + else + printk("xd_wd_init_drive: error reading geometry for xd%c\n",'a'+drive); + +} + +static void __init xd_seagate_init_controller (unsigned int address) +{ + switch (address) { + case 0x00000: + case 0xC8000: break; /*initial: 0x320 */ + case 0xD0000: xd_iobase = 0x324; break; + case 0xD8000: xd_iobase = 0x328; break; + case 0xE0000: xd_iobase = 0x32C; break; + default: printk("xd_seagate_init_controller: unsupported BIOS address %06x\n",address); + break; + } + xd_maxsectors = 0x40; + + outb(0,XD_RESET); /* reset the controller */ +} + +static void __init xd_seagate_init_drive (u_char drive) +{ + u_char cmdblk[6],buf[0x200]; + + xd_build(cmdblk,CMD_ST11GETGEOM,drive,0,0,0,1,0); + if (!xd_command(cmdblk,PIO_MODE,buf,NULL,NULL,XD_TIMEOUT * 2)) { + xd_info[drive].heads = buf[0x04]; /* heads */ + xd_info[drive].cylinders = (buf[0x02] << 8) | buf[0x03]; /* cylinders */ + xd_info[drive].sectors = buf[0x05]; /* sectors */ + xd_info[drive].control = 0; /* control byte */ + } + else + printk("xd_seagate_init_drive: error reading geometry from xd%c\n", 'a'+drive); +} + +/* Omti support courtesy Dirk Melchers */ +static void __init xd_omti_init_controller (unsigned int address) +{ + switch (address) { + case 0x00000: + case 0xC8000: break; /*initial: 0x320 */ + case 0xD0000: xd_iobase = 0x324; break; + case 0xD8000: xd_iobase = 0x328; break; + case 0xE0000: xd_iobase = 0x32C; break; + default: printk("xd_omti_init_controller: unsupported BIOS address %06x\n",address); + break; + } + + xd_maxsectors = 0x40; + + outb(0,XD_RESET); /* reset the controller */ +} + +static void __init xd_omti_init_drive (u_char drive) +{ + /* gets infos from drive */ + xd_override_init_drive(drive); + + /* set other parameters, Hardcoded, not that nice :-) */ + xd_info[drive].control = 2; +} + +/* Xebec support (AK) */ +static void __init xd_xebec_init_controller (unsigned int address) +{ +/* iobase may be set manually in range 0x300 - 0x33C + irq may be set manually to 2(9),3,4,5,6,7 + dma may be set manually to 1,2,3 + (How to detect them ???) +BIOS address may be set manually in range 0x0 - 0xF8000 +If you need non-standard settings use the xd=... command */ + + switch (address) { + case 0x00000: + case 0xC8000: /* initially: xd_iobase==0x320 */ + case 0xD0000: + case 0xD2000: + case 0xD4000: + case 0xD6000: + case 0xD8000: + case 0xDA000: + case 0xDC000: + case 0xDE000: + case 0xE0000: break; + default: printk("xd_xebec_init_controller: unsupported BIOS address %06x\n",address); + break; + } + + xd_maxsectors = 0x01; + outb(0,XD_RESET); /* reset the controller */ + + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(XD_INIT_DISK_DELAY); +} + +static void __init xd_xebec_init_drive (u_char drive) +{ + /* values from controller's BIOS - BIOS chip may be removed */ + static u_short geometry_table[][5] = { + {0x132,4,0x080,0x080,0x7}, + {0x132,4,0x080,0x080,0x17}, + {0x264,2,0x100,0x100,0x7}, + {0x264,2,0x100,0x100,0x17}, + {0x132,8,0x080,0x080,0x7}, + {0x132,8,0x080,0x080,0x17}, + {0x264,4,0x100,0x100,0x6}, + {0x264,4,0x100,0x100,0x17}, + {0x2BC,5,0x2BC,0x12C,0x6}, + {0x3A5,4,0x3A5,0x3A5,0x7}, + {0x26C,6,0x26C,0x26C,0x7}, + {0x200,8,0x200,0x100,0x17}, + {0x400,5,0x400,0x400,0x7}, + {0x400,6,0x400,0x400,0x7}, + {0x264,8,0x264,0x200,0x17}, + {0x33E,7,0x33E,0x200,0x7}}; + u_char n; + + n = inb(XD_JUMPER) & 0x0F; /* BIOS's drive number: same geometry + is assumed for BOTH drives */ + if (xd_geo[3*drive]) + xd_manual_geo_set(drive); + else { + xd_info[drive].heads = (u_char)(geometry_table[n][1]); /* heads */ + xd_info[drive].cylinders = geometry_table[n][0]; /* cylinders */ + xd_info[drive].sectors = 17; /* sectors */ +#if 0 + xd_info[drive].rwrite = geometry_table[n][2]; /* reduced write */ + xd_info[drive].precomp = geometry_table[n][3] /* write precomp */ + xd_info[drive].ecc = 0x0B; /* ecc length */ +#endif /* 0 */ + } + xd_info[drive].control = geometry_table[n][4]; /* control byte */ + xd_setparam(CMD_XBSETPARAM,drive,xd_info[drive].heads,xd_info[drive].cylinders,geometry_table[n][2],geometry_table[n][3],0x0B); + xd_recalibrate(drive); +} + +/* xd_override_init_drive: this finds disk geometry in a "binary search" style, narrowing in on the "correct" number of heads + etc. by trying values until it gets the highest successful value. Idea courtesy Salvador Abreu (spa@fct.unl.pt). */ +static void __init xd_override_init_drive (u_char drive) +{ + u_short min[] = { 0,0,0 },max[] = { 16,1024,64 },test[] = { 0,0,0 }; + u_char cmdblk[6],i; + + if (xd_geo[3*drive]) + xd_manual_geo_set(drive); + else { + for (i = 0; i < 3; i++) { + while (min[i] != max[i] - 1) { + test[i] = (min[i] + max[i]) / 2; + xd_build(cmdblk,CMD_SEEK,drive,(u_char) test[0],(u_short) test[1],(u_char) test[2],0,0); + if (!xd_command(cmdblk,PIO_MODE,NULL,NULL,NULL,XD_TIMEOUT * 2)) + min[i] = test[i]; + else + max[i] = test[i]; + } + test[i] = min[i]; + } + xd_info[drive].heads = (u_char) min[0] + 1; + xd_info[drive].cylinders = (u_short) min[1] + 1; + xd_info[drive].sectors = (u_char) min[2] + 1; + } + xd_info[drive].control = 0; +} + +/* xd_setup: initialise controller from command line parameters */ +static void __init do_xd_setup (int *integers) +{ + switch (integers[0]) { + case 4: if (integers[4] < 0) + nodma = 1; + else if (integers[4] < 8) + xd_dma = integers[4]; + case 3: if ((integers[3] > 0) && (integers[3] <= 0x3FC)) + xd_iobase = integers[3]; + case 2: if ((integers[2] > 0) && (integers[2] < 16)) + xd_irq = integers[2]; + case 1: xd_override = 1; + if ((integers[1] >= 0) && (integers[1] < (sizeof(xd_sigs) / sizeof(xd_sigs[0])))) + xd_type = integers[1]; + case 0: break; + default:printk("xd: too many parameters for xd\n"); + } + xd_maxsectors = 0x01; +} + +/* xd_setparam: set the drive characteristics */ +static void __init xd_setparam (u_char command,u_char drive,u_char heads,u_short cylinders,u_short rwrite,u_short wprecomp,u_char ecc) +{ + u_char cmdblk[14]; + + xd_build(cmdblk,command,drive,0,0,0,0,0); + cmdblk[6] = (u_char) (cylinders >> 8) & 0x03; + cmdblk[7] = (u_char) (cylinders & 0xFF); + cmdblk[8] = heads & 0x1F; + cmdblk[9] = (u_char) (rwrite >> 8) & 0x03; + cmdblk[10] = (u_char) (rwrite & 0xFF); + cmdblk[11] = (u_char) (wprecomp >> 8) & 0x03; + cmdblk[12] = (u_char) (wprecomp & 0xFF); + cmdblk[13] = ecc; + + /* Some controllers require geometry info as data, not command */ + + if (xd_command(cmdblk,PIO_MODE,NULL,&cmdblk[6],NULL,XD_TIMEOUT * 2)) + printk("xd: error setting characteristics for xd%c\n", 'a'+drive); +} + + +#ifdef MODULE + +module_param_array(xd, int, NULL, 0); +module_param_array(xd_geo, int, NULL, 0); +module_param(nodma, bool, 0); + +MODULE_LICENSE("GPL"); + +void cleanup_module(void) +{ + int i; + unregister_blkdev(XT_DISK_MAJOR, "xd"); + for (i = 0; i < xd_drives; i++) { + del_gendisk(xd_gendisk[i]); + put_disk(xd_gendisk[i]); + } + blk_cleanup_queue(xd_queue); + release_region(xd_iobase,4); + if (xd_drives) { + free_irq(xd_irq, NULL); + free_dma(xd_dma); + if (xd_dma_buffer) + xd_dma_mem_free((unsigned long)xd_dma_buffer, xd_maxsectors * 0x200); + } +} +#else + +static int __init xd_setup (char *str) +{ + int ints[5]; + get_options (str, ARRAY_SIZE (ints), ints); + do_xd_setup (ints); + return 1; +} + +/* xd_manual_geo_init: initialise drive geometry from command line parameters + (used only for WD drives) */ +static int __init xd_manual_geo_init (char *str) +{ + int i, integers[1 + 3*XD_MAXDRIVES]; + + get_options (str, ARRAY_SIZE (integers), integers); + if (integers[0]%3 != 0) { + printk("xd: incorrect number of parameters for xd_geo\n"); + return 1; + } + for (i = 0; (i < integers[0]) && (i < 3*XD_MAXDRIVES); i++) + xd_geo[i] = integers[i+1]; + return 1; +} + +__setup ("xd=", xd_setup); +__setup ("xd_geo=", xd_manual_geo_init); + +#endif /* MODULE */ + +module_init(xd_init); +MODULE_ALIAS_BLOCKDEV_MAJOR(XT_DISK_MAJOR); diff --git a/drivers/block/xd.h b/drivers/block/xd.h new file mode 100644 index 00000000000..71ac2e3dffc --- /dev/null +++ b/drivers/block/xd.h @@ -0,0 +1,135 @@ +#ifndef _LINUX_XD_H +#define _LINUX_XD_H + +/* + * This file contains the definitions for the IO ports and errors etc. for XT hard disk controllers (at least the DTC 5150X). + * + * Author: Pat Mackinlay, pat@it.com.au + * Date: 29/09/92 + * + * Revised: 01/01/93, ... + * + * Ref: DTC 5150X Controller Specification (thanks to Kevin Fowler, kevinf@agora.rain.com) + * Also thanks to: Salvador Abreu, Dave Thaler, Risto Kankkunen and Wim Van Dorst. + */ + +#include <linux/interrupt.h> + +/* XT hard disk controller registers */ +#define XD_DATA (xd_iobase + 0x00) /* data RW register */ +#define XD_RESET (xd_iobase + 0x01) /* reset WO register */ +#define XD_STATUS (xd_iobase + 0x01) /* status RO register */ +#define XD_SELECT (xd_iobase + 0x02) /* select WO register */ +#define XD_JUMPER (xd_iobase + 0x02) /* jumper RO register */ +#define XD_CONTROL (xd_iobase + 0x03) /* DMAE/INTE WO register */ +#define XD_RESERVED (xd_iobase + 0x03) /* reserved */ + +/* XT hard disk controller commands (incomplete list) */ +#define CMD_TESTREADY 0x00 /* test drive ready */ +#define CMD_RECALIBRATE 0x01 /* recalibrate drive */ +#define CMD_SENSE 0x03 /* request sense */ +#define CMD_FORMATDRV 0x04 /* format drive */ +#define CMD_VERIFY 0x05 /* read verify */ +#define CMD_FORMATTRK 0x06 /* format track */ +#define CMD_FORMATBAD 0x07 /* format bad track */ +#define CMD_READ 0x08 /* read */ +#define CMD_WRITE 0x0A /* write */ +#define CMD_SEEK 0x0B /* seek */ + +/* Controller specific commands */ +#define CMD_DTCSETPARAM 0x0C /* set drive parameters (DTC 5150X & CX only?) */ +#define CMD_DTCGETECC 0x0D /* get ecc error length (DTC 5150X only?) */ +#define CMD_DTCREADBUF 0x0E /* read sector buffer (DTC 5150X only?) */ +#define CMD_DTCWRITEBUF 0x0F /* write sector buffer (DTC 5150X only?) */ +#define CMD_DTCREMAPTRK 0x11 /* assign alternate track (DTC 5150X only?) */ +#define CMD_DTCGETPARAM 0xFB /* get drive parameters (DTC 5150X only?) */ +#define CMD_DTCSETSTEP 0xFC /* set step rate (DTC 5150X only?) */ +#define CMD_DTCSETGEOM 0xFE /* set geometry data (DTC 5150X only?) */ +#define CMD_DTCGETGEOM 0xFF /* get geometry data (DTC 5150X only?) */ +#define CMD_ST11GETGEOM 0xF8 /* get geometry data (Seagate ST11R/M only?) */ +#define CMD_WDSETPARAM 0x0C /* set drive parameters (WD 1004A27X only?) */ +#define CMD_XBSETPARAM 0x0C /* set drive parameters (XEBEC only?) */ + +/* Bits for command status byte */ +#define CSB_ERROR 0x02 /* error */ +#define CSB_LUN 0x20 /* logical Unit Number */ + +/* XT hard disk controller status bits */ +#define STAT_READY 0x01 /* controller is ready */ +#define STAT_INPUT 0x02 /* data flowing from controller to host */ +#define STAT_COMMAND 0x04 /* controller in command phase */ +#define STAT_SELECT 0x08 /* controller is selected */ +#define STAT_REQUEST 0x10 /* controller requesting data */ +#define STAT_INTERRUPT 0x20 /* controller requesting interrupt */ + +/* XT hard disk controller control bits */ +#define PIO_MODE 0x00 /* control bits to set for PIO */ +#define DMA_MODE 0x03 /* control bits to set for DMA & interrupt */ + +#define XD_MAXDRIVES 2 /* maximum 2 drives */ +#define XD_TIMEOUT HZ /* 1 second timeout */ +#define XD_RETRIES 4 /* maximum 4 retries */ + +#undef DEBUG /* define for debugging output */ + +#ifdef DEBUG + #define DEBUG_STARTUP /* debug driver initialisation */ + #define DEBUG_OVERRIDE /* debug override geometry detection */ + #define DEBUG_READWRITE /* debug each read/write command */ + #define DEBUG_OTHER /* debug misc. interrupt/DMA stuff */ + #define DEBUG_COMMAND /* debug each controller command */ +#endif /* DEBUG */ + +/* this structure defines the XT drives and their types */ +typedef struct { + u_char heads; + u_short cylinders; + u_char sectors; + u_char control; + int unit; +} XD_INFO; + +/* this structure defines a ROM BIOS signature */ +typedef struct { + unsigned int offset; + const char *string; + void (*init_controller)(unsigned int address); + void (*init_drive)(u_char drive); + const char *name; +} XD_SIGNATURE; + +#ifndef MODULE +static int xd_manual_geo_init (char *command); +#endif /* MODULE */ +static u_char xd_detect (u_char *controller, unsigned int *address); +static u_char xd_initdrives (void (*init_drive)(u_char drive)); + +static void do_xd_request (request_queue_t * q); +static int xd_ioctl (struct inode *inode,struct file *file,unsigned int cmd,unsigned long arg); +static int xd_readwrite (u_char operation,XD_INFO *disk,char *buffer,u_int block,u_int count); +static void xd_recalibrate (u_char drive); + +static irqreturn_t xd_interrupt_handler(int irq, void *dev_id, + struct pt_regs *regs); +static u_char xd_setup_dma (u_char opcode,u_char *buffer,u_int count); +static u_char *xd_build (u_char *cmdblk,u_char command,u_char drive,u_char head,u_short cylinder,u_char sector,u_char count,u_char control); +static void xd_watchdog (unsigned long unused); +static inline u_char xd_waitport (u_short port,u_char flags,u_char mask,u_long timeout); +static u_int xd_command (u_char *command,u_char mode,u_char *indata,u_char *outdata,u_char *sense,u_long timeout); + +/* card specific setup and geometry gathering code */ +static void xd_dtc_init_controller (unsigned int address); +static void xd_dtc5150cx_init_drive (u_char drive); +static void xd_dtc_init_drive (u_char drive); +static void xd_wd_init_controller (unsigned int address); +static void xd_wd_init_drive (u_char drive); +static void xd_seagate_init_controller (unsigned int address); +static void xd_seagate_init_drive (u_char drive); +static void xd_omti_init_controller (unsigned int address); +static void xd_omti_init_drive (u_char drive); +static void xd_xebec_init_controller (unsigned int address); +static void xd_xebec_init_drive (u_char drive); +static void xd_setparam (u_char command,u_char drive,u_char heads,u_short cylinders,u_short rwrite,u_short wprecomp,u_char ecc); +static void xd_override_init_drive (u_char drive); + +#endif /* _LINUX_XD_H */ diff --git a/drivers/block/z2ram.c b/drivers/block/z2ram.c new file mode 100644 index 00000000000..007f6a66243 --- /dev/null +++ b/drivers/block/z2ram.c @@ -0,0 +1,429 @@ +/* +** z2ram - Amiga pseudo-driver to access 16bit-RAM in ZorroII space +** as a block device, to be used as a RAM disk or swap space +** +** Copyright (C) 1994 by Ingo Wilken (Ingo.Wilken@informatik.uni-oldenburg.de) +** +** ++Geert: support for zorro_unused_z2ram, better range checking +** ++roman: translate accesses via an array +** ++Milan: support for ChipRAM usage +** ++yambo: converted to 2.0 kernel +** ++yambo: modularized and support added for 3 minor devices including: +** MAJOR MINOR DESCRIPTION +** ----- ----- ---------------------------------------------- +** 37 0 Use Zorro II and Chip ram +** 37 1 Use only Zorro II ram +** 37 2 Use only Chip ram +** 37 4-7 Use memory list entry 1-4 (first is 0) +** ++jskov: support for 1-4th memory list entry. +** +** Permission to use, copy, modify, and distribute this software and its +** documentation for any purpose and without fee is hereby granted, provided +** that the above copyright notice appear in all copies and that both that +** copyright notice and this permission notice appear in supporting +** documentation. This software is provided "as is" without express or +** implied warranty. +*/ + +#define DEVICE_NAME "Z2RAM" + +#include <linux/major.h> +#include <linux/vmalloc.h> +#include <linux/init.h> +#include <linux/module.h> +#include <linux/blkdev.h> +#include <linux/bitops.h> + +#include <asm/setup.h> +#include <asm/amigahw.h> +#include <asm/pgtable.h> + +#include <linux/zorro.h> + + +extern int m68k_realnum_memory; +extern struct mem_info m68k_memory[NUM_MEMINFO]; + +#define TRUE (1) +#define FALSE (0) + +#define Z2MINOR_COMBINED (0) +#define Z2MINOR_Z2ONLY (1) +#define Z2MINOR_CHIPONLY (2) +#define Z2MINOR_MEMLIST1 (4) +#define Z2MINOR_MEMLIST2 (5) +#define Z2MINOR_MEMLIST3 (6) +#define Z2MINOR_MEMLIST4 (7) +#define Z2MINOR_COUNT (8) /* Move this down when adding a new minor */ + +#define Z2RAM_CHUNK1024 ( Z2RAM_CHUNKSIZE >> 10 ) + +static u_long *z2ram_map = NULL; +static u_long z2ram_size = 0; +static int z2_count = 0; +static int chip_count = 0; +static int list_count = 0; +static int current_device = -1; + +static DEFINE_SPINLOCK(z2ram_lock); + +static struct block_device_operations z2_fops; +static struct gendisk *z2ram_gendisk; + +static void do_z2_request(request_queue_t *q) +{ + struct request *req; + while ((req = elv_next_request(q)) != NULL) { + unsigned long start = req->sector << 9; + unsigned long len = req->current_nr_sectors << 9; + + if (start + len > z2ram_size) { + printk( KERN_ERR DEVICE_NAME ": bad access: block=%lu, count=%u\n", + req->sector, req->current_nr_sectors); + end_request(req, 0); + continue; + } + while (len) { + unsigned long addr = start & Z2RAM_CHUNKMASK; + unsigned long size = Z2RAM_CHUNKSIZE - addr; + if (len < size) + size = len; + addr += z2ram_map[ start >> Z2RAM_CHUNKSHIFT ]; + if (rq_data_dir(req) == READ) + memcpy(req->buffer, (char *)addr, size); + else + memcpy((char *)addr, req->buffer, size); + start += size; + len -= size; + } + end_request(req, 1); + } +} + +static void +get_z2ram( void ) +{ + int i; + + for ( i = 0; i < Z2RAM_SIZE / Z2RAM_CHUNKSIZE; i++ ) + { + if ( test_bit( i, zorro_unused_z2ram ) ) + { + z2_count++; + z2ram_map[ z2ram_size++ ] = + ZTWO_VADDR( Z2RAM_START ) + ( i << Z2RAM_CHUNKSHIFT ); + clear_bit( i, zorro_unused_z2ram ); + } + } + + return; +} + +static void +get_chipram( void ) +{ + + while ( amiga_chip_avail() > ( Z2RAM_CHUNKSIZE * 4 ) ) + { + chip_count++; + z2ram_map[ z2ram_size ] = + (u_long)amiga_chip_alloc( Z2RAM_CHUNKSIZE, "z2ram" ); + + if ( z2ram_map[ z2ram_size ] == 0 ) + { + break; + } + + z2ram_size++; + } + + return; +} + +static int +z2_open( struct inode *inode, struct file *filp ) +{ + int device; + int max_z2_map = ( Z2RAM_SIZE / Z2RAM_CHUNKSIZE ) * + sizeof( z2ram_map[0] ); + int max_chip_map = ( amiga_chip_size / Z2RAM_CHUNKSIZE ) * + sizeof( z2ram_map[0] ); + int rc = -ENOMEM; + + device = iminor(inode); + + if ( current_device != -1 && current_device != device ) + { + rc = -EBUSY; + goto err_out; + } + + if ( current_device == -1 ) + { + z2_count = 0; + chip_count = 0; + list_count = 0; + z2ram_size = 0; + + /* Use a specific list entry. */ + if (device >= Z2MINOR_MEMLIST1 && device <= Z2MINOR_MEMLIST4) { + int index = device - Z2MINOR_MEMLIST1 + 1; + unsigned long size, paddr, vaddr; + + if (index >= m68k_realnum_memory) { + printk( KERN_ERR DEVICE_NAME + ": no such entry in z2ram_map\n" ); + goto err_out; + } + + paddr = m68k_memory[index].addr; + size = m68k_memory[index].size & ~(Z2RAM_CHUNKSIZE-1); + +#ifdef __powerpc__ + /* FIXME: ioremap doesn't build correct memory tables. */ + { + vfree(vmalloc (size)); + } + + vaddr = (unsigned long) __ioremap (paddr, size, + _PAGE_WRITETHRU); + +#else + vaddr = (unsigned long)z_remap_nocache_nonser(paddr, size); +#endif + z2ram_map = + kmalloc((size/Z2RAM_CHUNKSIZE)*sizeof(z2ram_map[0]), + GFP_KERNEL); + if ( z2ram_map == NULL ) + { + printk( KERN_ERR DEVICE_NAME + ": cannot get mem for z2ram_map\n" ); + goto err_out; + } + + while (size) { + z2ram_map[ z2ram_size++ ] = vaddr; + size -= Z2RAM_CHUNKSIZE; + vaddr += Z2RAM_CHUNKSIZE; + list_count++; + } + + if ( z2ram_size != 0 ) + printk( KERN_INFO DEVICE_NAME + ": using %iK List Entry %d Memory\n", + list_count * Z2RAM_CHUNK1024, index ); + } else + + switch ( device ) + { + case Z2MINOR_COMBINED: + + z2ram_map = kmalloc( max_z2_map + max_chip_map, GFP_KERNEL ); + if ( z2ram_map == NULL ) + { + printk( KERN_ERR DEVICE_NAME + ": cannot get mem for z2ram_map\n" ); + goto err_out; + } + + get_z2ram(); + get_chipram(); + + if ( z2ram_size != 0 ) + printk( KERN_INFO DEVICE_NAME + ": using %iK Zorro II RAM and %iK Chip RAM (Total %dK)\n", + z2_count * Z2RAM_CHUNK1024, + chip_count * Z2RAM_CHUNK1024, + ( z2_count + chip_count ) * Z2RAM_CHUNK1024 ); + + break; + + case Z2MINOR_Z2ONLY: + z2ram_map = kmalloc( max_z2_map, GFP_KERNEL ); + if ( z2ram_map == NULL ) + { + printk( KERN_ERR DEVICE_NAME + ": cannot get mem for z2ram_map\n" ); + goto err_out; + } + + get_z2ram(); + + if ( z2ram_size != 0 ) + printk( KERN_INFO DEVICE_NAME + ": using %iK of Zorro II RAM\n", + z2_count * Z2RAM_CHUNK1024 ); + + break; + + case Z2MINOR_CHIPONLY: + z2ram_map = kmalloc( max_chip_map, GFP_KERNEL ); + if ( z2ram_map == NULL ) + { + printk( KERN_ERR DEVICE_NAME + ": cannot get mem for z2ram_map\n" ); + goto err_out; + } + + get_chipram(); + + if ( z2ram_size != 0 ) + printk( KERN_INFO DEVICE_NAME + ": using %iK Chip RAM\n", + chip_count * Z2RAM_CHUNK1024 ); + + break; + + default: + rc = -ENODEV; + goto err_out; + + break; + } + + if ( z2ram_size == 0 ) + { + printk( KERN_NOTICE DEVICE_NAME + ": no unused ZII/Chip RAM found\n" ); + goto err_out_kfree; + } + + current_device = device; + z2ram_size <<= Z2RAM_CHUNKSHIFT; + set_capacity(z2ram_gendisk, z2ram_size >> 9); + } + + return 0; + +err_out_kfree: + kfree( z2ram_map ); +err_out: + return rc; +} + +static int +z2_release( struct inode *inode, struct file *filp ) +{ + if ( current_device == -1 ) + return 0; + + /* + * FIXME: unmap memory + */ + + return 0; +} + +static struct block_device_operations z2_fops = +{ + .owner = THIS_MODULE, + .open = z2_open, + .release = z2_release, +}; + +static struct kobject *z2_find(dev_t dev, int *part, void *data) +{ + *part = 0; + return get_disk(z2ram_gendisk); +} + +static struct request_queue *z2_queue; + +int __init +z2_init(void) +{ + int ret; + + if (!MACH_IS_AMIGA) + return -ENXIO; + + ret = -EBUSY; + if (register_blkdev(Z2RAM_MAJOR, DEVICE_NAME)) + goto err; + + ret = -ENOMEM; + z2ram_gendisk = alloc_disk(1); + if (!z2ram_gendisk) + goto out_disk; + + z2_queue = blk_init_queue(do_z2_request, &z2ram_lock); + if (!z2_queue) + goto out_queue; + + z2ram_gendisk->major = Z2RAM_MAJOR; + z2ram_gendisk->first_minor = 0; + z2ram_gendisk->fops = &z2_fops; + sprintf(z2ram_gendisk->disk_name, "z2ram"); + strcpy(z2ram_gendisk->devfs_name, z2ram_gendisk->disk_name); + + z2ram_gendisk->queue = z2_queue; + add_disk(z2ram_gendisk); + blk_register_region(MKDEV(Z2RAM_MAJOR, 0), Z2MINOR_COUNT, THIS_MODULE, + z2_find, NULL, NULL); + + return 0; + +out_queue: + put_disk(z2ram_gendisk); +out_disk: + unregister_blkdev(Z2RAM_MAJOR, DEVICE_NAME); +err: + return ret; +} + +#if defined(MODULE) + +MODULE_LICENSE("GPL"); + +int +init_module( void ) +{ + int error; + + error = z2_init(); + if ( error == 0 ) + { + printk( KERN_INFO DEVICE_NAME ": loaded as module\n" ); + } + + return error; +} + +void +cleanup_module( void ) +{ + int i, j; + blk_unregister_region(MKDEV(Z2RAM_MAJOR, 0), 256); + if ( unregister_blkdev( Z2RAM_MAJOR, DEVICE_NAME ) != 0 ) + printk( KERN_ERR DEVICE_NAME ": unregister of device failed\n"); + + del_gendisk(z2ram_gendisk); + put_disk(z2ram_gendisk); + blk_cleanup_queue(z2_queue); + + if ( current_device != -1 ) + { + i = 0; + + for ( j = 0 ; j < z2_count; j++ ) + { + set_bit( i++, zorro_unused_z2ram ); + } + + for ( j = 0 ; j < chip_count; j++ ) + { + if ( z2ram_map[ i ] ) + { + amiga_chip_free( (void *) z2ram_map[ i++ ] ); + } + } + + if ( z2ram_map != NULL ) + { + kfree( z2ram_map ); + } + } + + return; +} +#endif |