diff options
Diffstat (limited to 'arch/sparc64')
36 files changed, 1044 insertions, 451 deletions
diff --git a/arch/sparc64/Kconfig b/arch/sparc64/Kconfig index df3eacb5ca1..8acc5cc3862 100644 --- a/arch/sparc64/Kconfig +++ b/arch/sparc64/Kconfig @@ -250,6 +250,26 @@ endchoice endmenu +config NUMA + bool "NUMA support" + +config NODES_SHIFT + int + default "4" + depends on NEED_MULTIPLE_NODES + +# Some NUMA nodes have memory ranges that span +# other nodes. Even though a pfn is valid and +# between a node's start and end pfns, it may not +# reside on that node. See memmap_init_zone() +# for details. +config NODES_SPAN_OTHER_NODES + def_bool y + depends on NEED_MULTIPLE_NODES + +config ARCH_POPULATES_NODE_MAP + def_bool y + config ARCH_SELECT_MEMORY_MODEL def_bool y diff --git a/arch/sparc64/defconfig b/arch/sparc64/defconfig index e1835868ad3..92f79680f70 100644 --- a/arch/sparc64/defconfig +++ b/arch/sparc64/defconfig @@ -1,7 +1,7 @@ # # Automatically generated make config: don't edit -# Linux kernel version: 2.6.25 -# Sun Apr 20 01:33:21 2008 +# Linux kernel version: 2.6.25-numa +# Wed Apr 23 04:49:08 2008 # CONFIG_SPARC=y CONFIG_SPARC64=y @@ -152,6 +152,8 @@ CONFIG_GENERIC_CALIBRATE_DELAY=y CONFIG_HUGETLB_PAGE_SIZE_4MB=y # CONFIG_HUGETLB_PAGE_SIZE_512K is not set # CONFIG_HUGETLB_PAGE_SIZE_64K is not set +# CONFIG_NUMA is not set +CONFIG_ARCH_POPULATES_NODE_MAP=y CONFIG_ARCH_SELECT_MEMORY_MODEL=y CONFIG_ARCH_SPARSEMEM_ENABLE=y CONFIG_ARCH_SPARSEMEM_DEFAULT=y @@ -787,7 +789,6 @@ CONFIG_I2C_ALGOBIT=y # CONFIG_SENSORS_PCF8574 is not set # CONFIG_PCF8575 is not set # CONFIG_SENSORS_PCF8591 is not set -# CONFIG_TPS65010 is not set # CONFIG_SENSORS_MAX6875 is not set # CONFIG_SENSORS_TSL2550 is not set # CONFIG_I2C_DEBUG_CORE is not set @@ -869,6 +870,7 @@ CONFIG_SSB_POSSIBLE=y # Multifunction device drivers # # CONFIG_MFD_SM501 is not set +# CONFIG_HTC_PASIC3 is not set # # Multimedia devices @@ -1219,10 +1221,6 @@ CONFIG_USB_STORAGE=m # CONFIG_NEW_LEDS is not set # CONFIG_INFINIBAND is not set # CONFIG_RTC_CLASS is not set - -# -# Userspace I/O -# # CONFIG_UIO is not set # @@ -1399,6 +1397,7 @@ CONFIG_SCHEDSTATS=y CONFIG_DEBUG_BUGVERBOSE=y # CONFIG_DEBUG_INFO is not set # CONFIG_DEBUG_VM is not set +# CONFIG_DEBUG_WRITECOUNT is not set # CONFIG_DEBUG_LIST is not set # CONFIG_DEBUG_SG is not set # CONFIG_BOOT_PRINTK_DELAY is not set @@ -1425,53 +1424,82 @@ CONFIG_ASYNC_CORE=m CONFIG_ASYNC_MEMCPY=m CONFIG_ASYNC_XOR=m CONFIG_CRYPTO=y + +# +# Crypto core or helper +# CONFIG_CRYPTO_ALGAPI=y CONFIG_CRYPTO_AEAD=y CONFIG_CRYPTO_BLKCIPHER=y -# CONFIG_CRYPTO_SEQIV is not set CONFIG_CRYPTO_HASH=y CONFIG_CRYPTO_MANAGER=y +CONFIG_CRYPTO_GF128MUL=m +CONFIG_CRYPTO_NULL=m +# CONFIG_CRYPTO_CRYPTD is not set +CONFIG_CRYPTO_AUTHENC=y +CONFIG_CRYPTO_TEST=m + +# +# Authenticated Encryption with Associated Data +# +# CONFIG_CRYPTO_CCM is not set +# CONFIG_CRYPTO_GCM is not set +# CONFIG_CRYPTO_SEQIV is not set + +# +# Block modes +# +CONFIG_CRYPTO_CBC=y +# CONFIG_CRYPTO_CTR is not set +# CONFIG_CRYPTO_CTS is not set +CONFIG_CRYPTO_ECB=m +CONFIG_CRYPTO_LRW=m +CONFIG_CRYPTO_PCBC=m +CONFIG_CRYPTO_XTS=m + +# +# Hash modes +# CONFIG_CRYPTO_HMAC=y CONFIG_CRYPTO_XCBC=y -CONFIG_CRYPTO_NULL=m + +# +# Digest +# +CONFIG_CRYPTO_CRC32C=m CONFIG_CRYPTO_MD4=y CONFIG_CRYPTO_MD5=y +CONFIG_CRYPTO_MICHAEL_MIC=m CONFIG_CRYPTO_SHA1=y CONFIG_CRYPTO_SHA256=m CONFIG_CRYPTO_SHA512=m -CONFIG_CRYPTO_WP512=m CONFIG_CRYPTO_TGR192=m -CONFIG_CRYPTO_GF128MUL=m -CONFIG_CRYPTO_ECB=m -CONFIG_CRYPTO_CBC=y -CONFIG_CRYPTO_PCBC=m -CONFIG_CRYPTO_LRW=m -CONFIG_CRYPTO_XTS=m -# CONFIG_CRYPTO_CTR is not set -# CONFIG_CRYPTO_GCM is not set -# CONFIG_CRYPTO_CCM is not set -# CONFIG_CRYPTO_CRYPTD is not set -CONFIG_CRYPTO_DES=y -CONFIG_CRYPTO_FCRYPT=m -CONFIG_CRYPTO_BLOWFISH=m -CONFIG_CRYPTO_TWOFISH=m -CONFIG_CRYPTO_TWOFISH_COMMON=m -CONFIG_CRYPTO_SERPENT=m +CONFIG_CRYPTO_WP512=m + +# +# Ciphers +# CONFIG_CRYPTO_AES=m +CONFIG_CRYPTO_ANUBIS=m +CONFIG_CRYPTO_ARC4=m +CONFIG_CRYPTO_BLOWFISH=m +CONFIG_CRYPTO_CAMELLIA=m CONFIG_CRYPTO_CAST5=m CONFIG_CRYPTO_CAST6=m -CONFIG_CRYPTO_TEA=m -CONFIG_CRYPTO_ARC4=m +CONFIG_CRYPTO_DES=y +CONFIG_CRYPTO_FCRYPT=m CONFIG_CRYPTO_KHAZAD=m -CONFIG_CRYPTO_ANUBIS=m -CONFIG_CRYPTO_SEED=m # CONFIG_CRYPTO_SALSA20 is not set +CONFIG_CRYPTO_SEED=m +CONFIG_CRYPTO_SERPENT=m +CONFIG_CRYPTO_TEA=m +CONFIG_CRYPTO_TWOFISH=m +CONFIG_CRYPTO_TWOFISH_COMMON=m + +# +# Compression +# CONFIG_CRYPTO_DEFLATE=y -CONFIG_CRYPTO_MICHAEL_MIC=m -CONFIG_CRYPTO_CRC32C=m -CONFIG_CRYPTO_CAMELLIA=m -CONFIG_CRYPTO_TEST=m -CONFIG_CRYPTO_AUTHENC=y # CONFIG_CRYPTO_LZO is not set CONFIG_CRYPTO_HW=y # CONFIG_CRYPTO_DEV_HIFN_795X is not set @@ -1492,3 +1520,4 @@ CONFIG_PLIST=y CONFIG_HAS_IOMEM=y CONFIG_HAS_IOPORT=y CONFIG_HAS_DMA=y +CONFIG_HAVE_LMB=y diff --git a/arch/sparc64/kernel/ebus.c b/arch/sparc64/kernel/ebus.c index 04ab81cb4f4..bc263227484 100644 --- a/arch/sparc64/kernel/ebus.c +++ b/arch/sparc64/kernel/ebus.c @@ -396,6 +396,7 @@ static void __init fill_ebus_device(struct device_node *dp, struct linux_ebus_de sd->op = &dev->ofdev; sd->iommu = dev->bus->ofdev.dev.parent->archdata.iommu; sd->stc = dev->bus->ofdev.dev.parent->archdata.stc; + sd->numa_node = dev->bus->ofdev.dev.parent->archdata.numa_node; dev->ofdev.node = dp; dev->ofdev.dev.parent = &dev->bus->ofdev.dev; diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S index fb43c76bdc2..fd06e937ae1 100644 --- a/arch/sparc64/kernel/entry.S +++ b/arch/sparc64/kernel/entry.S @@ -47,7 +47,7 @@ do_fpdis: ba,pt %xcc, etrap 109: or %g7, %lo(109b), %g7 add %g0, %g0, %g0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap 1: TRAP_LOAD_THREAD_REG(%g6, %g1) ldub [%g6 + TI_FPSAVED], %g5 @@ -226,7 +226,7 @@ fp_other_bounce: call do_fpother add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop .globl do_fpother_check_fitos .align 32 @@ -489,7 +489,7 @@ utrap_trap: /* %g3=handler,%g4=level */ call bad_trap add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop invoke_utrap: sllx %g3, 3, %g3 @@ -607,7 +607,7 @@ __spitfire_cee_trap_continue: call spitfire_access_error add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop /* This is the trap handler entry point for ECC correctable * errors. They are corrected, but we listen for the trap @@ -686,7 +686,7 @@ __spitfire_data_access_exception_tl1: call spitfire_data_access_exception_tl1 add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop __spitfire_data_access_exception: rdpr %pstate, %g4 @@ -705,7 +705,7 @@ __spitfire_data_access_exception: call spitfire_data_access_exception add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop .globl __spitfire_insn_access_exception .globl __spitfire_insn_access_exception_tl1 @@ -725,7 +725,7 @@ __spitfire_insn_access_exception_tl1: call spitfire_insn_access_exception_tl1 add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop __spitfire_insn_access_exception: rdpr %pstate, %g4 @@ -743,7 +743,7 @@ __spitfire_insn_access_exception: call spitfire_insn_access_exception add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop /* These get patched into the trap table at boot time * once we know we have a cheetah processor. @@ -937,7 +937,7 @@ do_dcpe_tl1_fatal: call cheetah_plus_parity_error add %sp, PTREGS_OFF, %o1 ba,pt %xcc, rtrap - clr %l6 + nop do_icpe_tl1: rdpr %tl, %g1 ! Save original trap level @@ -979,7 +979,7 @@ do_icpe_tl1_fatal: call cheetah_plus_parity_error add %sp, PTREGS_OFF, %o1 ba,pt %xcc, rtrap - clr %l6 + nop dcpe_icpe_tl1_common: /* Flush D-cache, re-enable D/I caches in DCU and finally @@ -1281,7 +1281,7 @@ __do_privact: call do_privact add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop .globl do_mna do_mna: @@ -1308,7 +1308,7 @@ do_mna: call mem_address_unaligned add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop .globl do_lddfmna do_lddfmna: @@ -1326,7 +1326,7 @@ do_lddfmna: call handle_lddfmna add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop .globl do_stdfmna do_stdfmna: @@ -1344,7 +1344,7 @@ do_stdfmna: call handle_stdfmna add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap - clr %l6 + nop .globl breakpoint_trap breakpoint_trap: @@ -1424,13 +1424,13 @@ sys32_rt_sigreturn: 1: ldx [%curptr + TI_FLAGS], %l5 andcc %l5, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT), %g0 be,pt %icc, rtrap - clr %l6 + nop add %sp, PTREGS_OFF, %o0 call syscall_trace mov 1, %o1 ba,pt %xcc, rtrap - clr %l6 + nop /* This is how fork() was meant to be done, 8 instruction entry. * @@ -1559,7 +1559,7 @@ linux_sparc_syscall32: /* Linux native system calls enter here... */ .align 32 - .globl linux_sparc_syscall, ret_sys_call + .globl linux_sparc_syscall linux_sparc_syscall: /* Direct access to user regs, much faster. */ cmp %g1, NR_SYSCALLS ! IEU1 Group @@ -1605,7 +1605,7 @@ ret_sys_call: bne,pn %icc, linux_syscall_trace2 add %l1, 0x4, %l2 ! npc = npc+4 stx %l1, [%sp + PTREGS_OFF + PT_V9_TPC] - ba,pt %xcc, rtrap_clr_l6 + ba,pt %xcc, rtrap stx %l2, [%sp + PTREGS_OFF + PT_V9_TNPC] 1: @@ -1616,7 +1616,6 @@ ret_sys_call: sub %g0, %o0, %o0 or %g3, %g2, %g3 stx %o0, [%sp + PTREGS_OFF + PT_V9_I0] - mov 1, %l6 stx %g3, [%sp + PTREGS_OFF + PT_V9_TSTATE] bne,pn %icc, linux_syscall_trace2 add %l1, 0x4, %l2 ! npc = npc+4 diff --git a/arch/sparc64/kernel/entry.h b/arch/sparc64/kernel/entry.h index 4a91e9c6d31..32fbab62085 100644 --- a/arch/sparc64/kernel/entry.h +++ b/arch/sparc64/kernel/entry.h @@ -20,7 +20,6 @@ extern void timer_interrupt(int irq, struct pt_regs *regs); extern void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, - int restart_syscall, unsigned long thread_info_flags); extern asmlinkage void syscall_trace(struct pt_regs *regs, diff --git a/arch/sparc64/kernel/etrap.S b/arch/sparc64/kernel/etrap.S index 4b2bf9eb447..b49d3b60bc0 100644 --- a/arch/sparc64/kernel/etrap.S +++ b/arch/sparc64/kernel/etrap.S @@ -53,7 +53,11 @@ etrap_irq: stx %g3, [%g2 + STACKFRAME_SZ + PT_V9_TPC] rd %y, %g3 stx %g1, [%g2 + STACKFRAME_SZ + PT_V9_TNPC] + rdpr %tt, %g1 st %g3, [%g2 + STACKFRAME_SZ + PT_V9_Y] + sethi %hi(PT_REGS_MAGIC), %g3 + or %g3, %g1, %g1 + st %g1, [%g2 + STACKFRAME_SZ + PT_V9_MAGIC] rdpr %cansave, %g1 brnz,pt %g1, etrap_save diff --git a/arch/sparc64/kernel/iommu.c b/arch/sparc64/kernel/iommu.c index 756fa24eeef..2a37a6ca2a1 100644 --- a/arch/sparc64/kernel/iommu.c +++ b/arch/sparc64/kernel/iommu.c @@ -173,9 +173,11 @@ void iommu_range_free(struct iommu *iommu, dma_addr_t dma_addr, unsigned long np } int iommu_table_init(struct iommu *iommu, int tsbsize, - u32 dma_offset, u32 dma_addr_mask) + u32 dma_offset, u32 dma_addr_mask, + int numa_node) { - unsigned long i, tsbbase, order, sz, num_tsb_entries; + unsigned long i, order, sz, num_tsb_entries; + struct page *page; num_tsb_entries = tsbsize / sizeof(iopte_t); @@ -188,11 +190,12 @@ int iommu_table_init(struct iommu *iommu, int tsbsize, /* Allocate and initialize the free area map. */ sz = num_tsb_entries / 8; sz = (sz + 7UL) & ~7UL; - iommu->arena.map = kzalloc(sz, GFP_KERNEL); + iommu->arena.map = kmalloc_node(sz, GFP_KERNEL, numa_node); if (!iommu->arena.map) { printk(KERN_ERR "IOMMU: Error, kmalloc(arena.map) failed.\n"); return -ENOMEM; } + memset(iommu->arena.map, 0, sz); iommu->arena.limit = num_tsb_entries; if (tlb_type != hypervisor) @@ -201,21 +204,23 @@ int iommu_table_init(struct iommu *iommu, int tsbsize, /* Allocate and initialize the dummy page which we * set inactive IO PTEs to point to. */ - iommu->dummy_page = get_zeroed_page(GFP_KERNEL); - if (!iommu->dummy_page) { + page = alloc_pages_node(numa_node, GFP_KERNEL, 0); + if (!page) { printk(KERN_ERR "IOMMU: Error, gfp(dummy_page) failed.\n"); goto out_free_map; } + iommu->dummy_page = (unsigned long) page_address(page); + memset((void *)iommu->dummy_page, 0, PAGE_SIZE); iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); /* Now allocate and setup the IOMMU page table itself. */ order = get_order(tsbsize); - tsbbase = __get_free_pages(GFP_KERNEL, order); - if (!tsbbase) { + page = alloc_pages_node(numa_node, GFP_KERNEL, order); + if (!page) { printk(KERN_ERR "IOMMU: Error, gfp(tsb) failed.\n"); goto out_free_dummy_page; } - iommu->page_table = (iopte_t *)tsbbase; + iommu->page_table = (iopte_t *)page_address(page); for (i = 0; i < num_tsb_entries; i++) iopte_make_dummy(iommu, &iommu->page_table[i]); @@ -276,20 +281,24 @@ static inline void iommu_free_ctx(struct iommu *iommu, int ctx) static void *dma_4u_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_addrp, gfp_t gfp) { + unsigned long flags, order, first_page; struct iommu *iommu; + struct page *page; + int npages, nid; iopte_t *iopte; - unsigned long flags, order, first_page; void *ret; - int npages; size = IO_PAGE_ALIGN(size); order = get_order(size); if (order >= 10) return NULL; - first_page = __get_free_pages(gfp, order); - if (first_page == 0UL) + nid = dev->archdata.numa_node; + page = alloc_pages_node(nid, gfp, order); + if (unlikely(!page)) return NULL; + + first_page = (unsigned long) page_address(page); memset((char *)first_page, 0, PAGE_SIZE << order); iommu = dev->archdata.iommu; diff --git a/arch/sparc64/kernel/isa.c b/arch/sparc64/kernel/isa.c index b5f7b354084..a2af5ed784c 100644 --- a/arch/sparc64/kernel/isa.c +++ b/arch/sparc64/kernel/isa.c @@ -92,6 +92,7 @@ static void __init isa_fill_devices(struct sparc_isa_bridge *isa_br) sd->op = &isa_dev->ofdev; sd->iommu = isa_br->ofdev.dev.parent->archdata.iommu; sd->stc = isa_br->ofdev.dev.parent->archdata.stc; + sd->numa_node = isa_br->ofdev.dev.parent->archdata.numa_node; isa_dev->ofdev.node = dp; isa_dev->ofdev.dev.parent = &isa_br->ofdev.dev; diff --git a/arch/sparc64/kernel/mdesc.c b/arch/sparc64/kernel/mdesc.c index 91008358956..dde52bcf5c6 100644 --- a/arch/sparc64/kernel/mdesc.c +++ b/arch/sparc64/kernel/mdesc.c @@ -1,10 +1,10 @@ /* mdesc.c: Sun4V machine description handling. * - * Copyright (C) 2007 David S. Miller <davem@davemloft.net> + * Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net> */ #include <linux/kernel.h> #include <linux/types.h> -#include <linux/bootmem.h> +#include <linux/lmb.h> #include <linux/log2.h> #include <linux/list.h> #include <linux/slab.h> @@ -84,24 +84,28 @@ static void mdesc_handle_init(struct mdesc_handle *hp, hp->handle_size = handle_size; } -static struct mdesc_handle * __init mdesc_bootmem_alloc(unsigned int mdesc_size) +static struct mdesc_handle * __init mdesc_lmb_alloc(unsigned int mdesc_size) { - struct mdesc_handle *hp; unsigned int handle_size, alloc_size; + struct mdesc_handle *hp; + unsigned long paddr; handle_size = (sizeof(struct mdesc_handle) - sizeof(struct mdesc_hdr) + mdesc_size); alloc_size = PAGE_ALIGN(handle_size); - hp = __alloc_bootmem(alloc_size, PAGE_SIZE, 0UL); - if (hp) - mdesc_handle_init(hp, handle_size, hp); + paddr = lmb_alloc(alloc_size, PAGE_SIZE); + hp = NULL; + if (paddr) { + hp = __va(paddr); + mdesc_handle_init(hp, handle_size, hp); + } return hp; } -static void mdesc_bootmem_free(struct mdesc_handle *hp) +static void mdesc_lmb_free(struct mdesc_handle *hp) { unsigned int alloc_size, handle_size = hp->handle_size; unsigned long start, end; @@ -124,9 +128,9 @@ static void mdesc_bootmem_free(struct mdesc_handle *hp) } } -static struct mdesc_mem_ops bootmem_mdesc_ops = { - .alloc = mdesc_bootmem_alloc, - .free = mdesc_bootmem_free, +static struct mdesc_mem_ops lmb_mdesc_ops = { + .alloc = mdesc_lmb_alloc, + .free = mdesc_lmb_free, }; static struct mdesc_handle *mdesc_kmalloc(unsigned int mdesc_size) @@ -888,7 +892,7 @@ void __init sun4v_mdesc_init(void) printk("MDESC: Size is %lu bytes.\n", len); - hp = mdesc_alloc(len, &bootmem_mdesc_ops); + hp = mdesc_alloc(len, &lmb_mdesc_ops); if (hp == NULL) { prom_printf("MDESC: alloc of %lu bytes failed.\n", len); prom_halt(); diff --git a/arch/sparc64/kernel/of_device.c b/arch/sparc64/kernel/of_device.c index 0fd9db95b89..9e58e8cba1c 100644 --- a/arch/sparc64/kernel/of_device.c +++ b/arch/sparc64/kernel/of_device.c @@ -6,6 +6,7 @@ #include <linux/mod_devicetable.h> #include <linux/slab.h> #include <linux/errno.h> +#include <linux/irq.h> #include <linux/of_device.h> #include <linux/of_platform.h> @@ -660,6 +661,7 @@ static unsigned int __init build_one_device_irq(struct of_device *op, struct device_node *dp = op->node; struct device_node *pp, *ip; unsigned int orig_irq = irq; + int nid; if (irq == 0xffffffff) return irq; @@ -672,7 +674,7 @@ static unsigned int __init build_one_device_irq(struct of_device *op, printk("%s: direct translate %x --> %x\n", dp->full_name, orig_irq, irq); - return irq; + goto out; } /* Something more complicated. Walk up to the root, applying @@ -744,6 +746,14 @@ static unsigned int __init build_one_device_irq(struct of_device *op, printk("%s: Apply IRQ trans [%s] %x --> %x\n", op->node->full_name, ip->full_name, orig_irq, irq); +out: + nid = of_node_to_nid(dp); + if (nid != -1) { + cpumask_t numa_mask = node_to_cpumask(nid); + + irq_set_affinity(irq, numa_mask); + } + return irq; } diff --git a/arch/sparc64/kernel/pci.c b/arch/sparc64/kernel/pci.c index 545356b00e2..49f91276651 100644 --- a/arch/sparc64/kernel/pci.c +++ b/arch/sparc64/kernel/pci.c @@ -369,10 +369,12 @@ struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm, sd->host_controller = pbm; sd->prom_node = node; sd->op = of_find_device_by_node(node); + sd->numa_node = pbm->numa_node; sd = &sd->op->dev.archdata; sd->iommu = pbm->iommu; sd->stc = &pbm->stc; + sd->numa_node = pbm->numa_node; type = of_get_property(node, "device_type", NULL); if (type == NULL) @@ -1159,6 +1161,16 @@ int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma, return 0; } +#ifdef CONFIG_NUMA +int pcibus_to_node(struct pci_bus *pbus) +{ + struct pci_pbm_info *pbm = pbus->sysdata; + + return pbm->numa_node; +} +EXPORT_SYMBOL(pcibus_to_node); +#endif + /* Return the domain nuber for this pci bus */ int pci_domain_nr(struct pci_bus *pbus) diff --git a/arch/sparc64/kernel/pci_fire.c b/arch/sparc64/kernel/pci_fire.c index 7571ed56314..d23bb6f53cd 100644 --- a/arch/sparc64/kernel/pci_fire.c +++ b/arch/sparc64/kernel/pci_fire.c @@ -71,7 +71,8 @@ static int pci_fire_pbm_iommu_init(struct pci_pbm_info *pbm) */ fire_write(iommu->iommu_flushinv, ~(u64)0); - err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask); + err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask, + pbm->numa_node); if (err) return err; @@ -449,6 +450,8 @@ static int __init pci_fire_pbm_init(struct pci_controller_info *p, pbm->next = pci_pbm_root; pci_pbm_root = pbm; + pbm->numa_node = -1; + pbm->scan_bus = pci_fire_scan_bus; pbm->pci_ops = &sun4u_pci_ops; pbm->config_space_reg_bits = 12; diff --git a/arch/sparc64/kernel/pci_impl.h b/arch/sparc64/kernel/pci_impl.h index 4a50da13ce4..218bac4ff79 100644 --- a/arch/sparc64/kernel/pci_impl.h +++ b/arch/sparc64/kernel/pci_impl.h @@ -148,6 +148,8 @@ struct pci_pbm_info { struct pci_bus *pci_bus; void (*scan_bus)(struct pci_pbm_info *); struct pci_ops *pci_ops; + + int numa_node; }; struct pci_controller_info { @@ -161,8 +163,6 @@ extern struct pci_pbm_info *pci_pbm_root; extern int pci_num_pbms; /* PCI bus scanning and fixup support. */ -extern void pci_iommu_table_init(struct iommu *iommu, int tsbsize, - u32 dma_offset, u32 dma_addr_mask); extern void pci_get_pbm_props(struct pci_pbm_info *pbm); extern struct pci_bus *pci_scan_one_pbm(struct pci_pbm_info *pbm); extern void pci_determine_mem_io_space(struct pci_pbm_info *pbm); diff --git a/arch/sparc64/kernel/pci_msi.c b/arch/sparc64/kernel/pci_msi.c index d6d64b44af6..db5e8fd8f67 100644 --- a/arch/sparc64/kernel/pci_msi.c +++ b/arch/sparc64/kernel/pci_msi.c @@ -279,11 +279,17 @@ static int bringup_one_msi_queue(struct pci_pbm_info *pbm, unsigned long devino) { int irq = ops->msiq_build_irq(pbm, msiqid, devino); - int err; + int err, nid; if (irq < 0) return irq; + nid = pbm->numa_node; + if (nid != -1) { + cpumask_t numa_mask = node_to_cpumask(nid); + + irq_set_affinity(irq, numa_mask); + } err = request_irq(irq, sparc64_msiq_interrupt, 0, "MSIQ", &pbm->msiq_irq_cookies[msiqid - pbm->msiq_first]); diff --git a/arch/sparc64/kernel/pci_psycho.c b/arch/sparc64/kernel/pci_psycho.c index 0bad96e5d18..994dbe0603d 100644 --- a/arch/sparc64/kernel/pci_psycho.c +++ b/arch/sparc64/kernel/pci_psycho.c @@ -848,7 +848,8 @@ static int psycho_iommu_init(struct pci_pbm_info *pbm) /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ - err = iommu_table_init(iommu, IO_TSB_SIZE, 0xc0000000, 0xffffffff); + err = iommu_table_init(iommu, IO_TSB_SIZE, 0xc0000000, 0xffffffff, + pbm->numa_node); if (err) return err; @@ -979,6 +980,8 @@ static void __init psycho_pbm_init(struct pci_controller_info *p, pbm->next = pci_pbm_root; pci_pbm_root = pbm; + pbm->numa_node = -1; + pbm->scan_bus = psycho_scan_bus; pbm->pci_ops = &sun4u_pci_ops; pbm->config_space_reg_bits = 8; diff --git a/arch/sparc64/kernel/pci_sabre.c b/arch/sparc64/kernel/pci_sabre.c index 1c5f5fa2339..4c34195baf3 100644 --- a/arch/sparc64/kernel/pci_sabre.c +++ b/arch/sparc64/kernel/pci_sabre.c @@ -704,7 +704,7 @@ static int sabre_iommu_init(struct pci_pbm_info *pbm, * in pci_iommu.c */ err = iommu_table_init(iommu, tsbsize * 1024 * 8, - dvma_offset, dma_mask); + dvma_offset, dma_mask, pbm->numa_node); if (err) return err; @@ -737,6 +737,8 @@ static void __init sabre_pbm_init(struct pci_controller_info *p, pbm->name = dp->full_name; printk("%s: SABRE PCI Bus Module\n", pbm->name); + pbm->numa_node = -1; + pbm->scan_bus = sabre_scan_bus; pbm->pci_ops = &sun4u_pci_ops; pbm->config_space_reg_bits = 8; diff --git a/arch/sparc64/kernel/pci_schizo.c b/arch/sparc64/kernel/pci_schizo.c index e3060936232..615edd9c8e2 100644 --- a/arch/sparc64/kernel/pci_schizo.c +++ b/arch/sparc64/kernel/pci_schizo.c @@ -1220,7 +1220,8 @@ static int schizo_pbm_iommu_init(struct pci_pbm_info *pbm) /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ - err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask); + err = iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask, + pbm->numa_node); if (err) return err; @@ -1379,6 +1380,8 @@ static int __init schizo_pbm_init(struct pci_controller_info *p, pbm->next = pci_pbm_root; pci_pbm_root = pbm; + pbm->numa_node = -1; + pbm->scan_bus = schizo_scan_bus; pbm->pci_ops = &sun4u_pci_ops; pbm->config_space_reg_bits = 8; diff --git a/arch/sparc64/kernel/pci_sun4v.c b/arch/sparc64/kernel/pci_sun4v.c index 01839706bd5..e2bb9790039 100644 --- a/arch/sparc64/kernel/pci_sun4v.c +++ b/arch/sparc64/kernel/pci_sun4v.c @@ -127,10 +127,12 @@ static inline long iommu_batch_end(void) static void *dma_4v_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_addrp, gfp_t gfp) { - struct iommu *iommu; unsigned long flags, order, first_page, npages, n; + struct iommu *iommu; + struct page *page; void *ret; long entry; + int nid; size = IO_PAGE_ALIGN(size); order = get_order(size); @@ -139,10 +141,12 @@ static void *dma_4v_alloc_coherent(struct device *dev, size_t size, npages = size >> IO_PAGE_SHIFT; - first_page = __get_free_pages(gfp, order); - if (unlikely(first_page == 0UL)) + nid = dev->archdata.numa_node; + page = alloc_pages_node(nid, gfp, order); + if (unlikely(!page)) return NULL; + first_page = (unsigned long) page_address(page); memset((char *)first_page, 0, PAGE_SIZE << order); iommu = dev->archdata.iommu; @@ -899,6 +903,8 @@ static void __init pci_sun4v_pbm_init(struct pci_controller_info *p, pbm->next = pci_pbm_root; pci_pbm_root = pbm; + pbm->numa_node = of_node_to_nid(dp); + pbm->scan_bus = pci_sun4v_scan_bus; pbm->pci_ops = &sun4v_pci_ops; pbm->config_space_reg_bits = 12; @@ -913,6 +919,7 @@ static void __init pci_sun4v_pbm_init(struct pci_controller_info *p, pbm->name = dp->full_name; printk("%s: SUN4V PCI Bus Module\n", pbm->name); + printk("%s: On NUMA node %d\n", pbm->name, pbm->numa_node); pci_determine_mem_io_space(pbm); diff --git a/arch/sparc64/kernel/prom.c b/arch/sparc64/kernel/prom.c index 68964ddcde1..ed03a18d3b3 100644 --- a/arch/sparc64/kernel/prom.c +++ b/arch/sparc64/kernel/prom.c @@ -19,8 +19,8 @@ #include <linux/types.h> #include <linux/string.h> #include <linux/mm.h> -#include <linux/bootmem.h> #include <linux/module.h> +#include <linux/lmb.h> #include <asm/prom.h> #include <asm/of_device.h> @@ -122,16 +122,20 @@ int of_find_in_proplist(const char *list, const char *match, int len) } EXPORT_SYMBOL(of_find_in_proplist); -static unsigned int prom_early_allocated; +static unsigned int prom_early_allocated __initdata; static void * __init prom_early_alloc(unsigned long size) { + unsigned long paddr = lmb_alloc(size, SMP_CACHE_BYTES); void *ret; - ret = __alloc_bootmem(size, SMP_CACHE_BYTES, 0UL); - if (ret != NULL) - memset(ret, 0, size); + if (!paddr) { + prom_printf("prom_early_alloc(%lu) failed\n"); + prom_halt(); + } + ret = __va(paddr); + memset(ret, 0, size); prom_early_allocated += size; return ret; diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S index 079d18a11d2..ecf6753b204 100644 --- a/arch/sparc64/kernel/rtrap.S +++ b/arch/sparc64/kernel/rtrap.S @@ -18,12 +18,6 @@ #define RTRAP_PSTATE_IRQOFF (PSTATE_RMO|PSTATE_PEF|PSTATE_PRIV) #define RTRAP_PSTATE_AG_IRQOFF (PSTATE_RMO|PSTATE_PEF|PSTATE_PRIV|PSTATE_AG) - /* Register %l6 keeps track of whether we are returning - * from a system call or not. It is cleared if we call - * do_notify_resume, and it must not be otherwise modified - * until we fully commit to returning to userspace. - */ - .text .align 32 __handle_softirq: @@ -56,14 +50,12 @@ __handle_user_windows: be,pt %xcc, __handle_user_windows_continue nop mov %l5, %o1 - mov %l6, %o2 add %sp, PTREGS_OFF, %o0 - mov %l0, %o3 + mov %l0, %o2 call do_notify_resume wrpr %g0, RTRAP_PSTATE, %pstate wrpr %g0, RTRAP_PSTATE_IRQOFF, %pstate - clr %l6 /* Signal delivery can modify pt_regs tstate, so we must * reload it. */ @@ -99,14 +91,12 @@ __handle_perfctrs: be,pt %xcc, __handle_perfctrs_continue sethi %hi(TSTATE_PEF), %o0 mov %l5, %o1 - mov %l6, %o2 add %sp, PTREGS_OFF, %o0 - mov %l0, %o3 + mov %l0, %o2 call do_notify_resume wrpr %g0, RTRAP_PSTATE, %pstate wrpr %g0, RTRAP_PSTATE_IRQOFF, %pstate - clr %l6 /* Signal delivery can modify pt_regs tstate, so we must * reload it. */ @@ -127,13 +117,11 @@ __handle_userfpu: __handle_signal: mov %l5, %o1 - mov %l6, %o2 add %sp, PTREGS_OFF, %o0 - mov %l0, %o3 + mov %l0, %o2 call do_notify_resume wrpr %g0, RTRAP_PSTATE, %pstate wrpr %g0, RTRAP_PSTATE_IRQOFF, %pstate - clr %l6 /* Signal delivery can modify pt_regs tstate, so we must * reload it. @@ -145,9 +133,8 @@ __handle_signal: andn %l1, %l4, %l1 .align 64 - .globl rtrap_irq, rtrap_clr_l6, rtrap, irqsz_patchme, rtrap_xcall + .globl rtrap_irq, rtrap, irqsz_patchme, rtrap_xcall rtrap_irq: -rtrap_clr_l6: clr %l6 rtrap: #ifndef CONFIG_SMP sethi %hi(per_cpu____cpu_data), %l0 diff --git a/arch/sparc64/kernel/sbus.c b/arch/sparc64/kernel/sbus.c index d1fb13ba02b..fa2827c4a3a 100644 --- a/arch/sparc64/kernel/sbus.c +++ b/arch/sparc64/kernel/sbus.c @@ -544,6 +544,7 @@ static void __init sbus_iommu_init(int __node, struct sbus_bus *sbus) sbus->ofdev.dev.archdata.iommu = iommu; sbus->ofdev.dev.archdata.stc = strbuf; + sbus->ofdev.dev.archdata.numa_node = -1; reg_base = regs + SYSIO_IOMMUREG_BASE; iommu->iommu_control = reg_base + IOMMU_CONTROL; @@ -575,7 +576,7 @@ static void __init sbus_iommu_init(int __node, struct sbus_bus *sbus) sbus->portid, regs); /* Setup for TSB_SIZE=7, TBW_SIZE=0, MMU_DE=1, MMU_EN=1 */ - if (iommu_table_init(iommu, IO_TSB_SIZE, MAP_BASE, 0xffffffff)) + if (iommu_table_init(iommu, IO_TSB_SIZE, MAP_BASE, 0xffffffff, -1)) goto fatal_memory_error; control = upa_readq(iommu->iommu_control); diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index 6acb4c51cfe..da5e6ee0c66 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -82,7 +82,7 @@ unsigned long cmdline_memory_size = 0; static struct console prom_early_console = { .name = "earlyprom", .write = prom_console_write, - .flags = CON_PRINTBUFFER | CON_BOOT, + .flags = CON_PRINTBUFFER | CON_BOOT | CON_ANYTIME, .index = -1, }; @@ -281,6 +281,7 @@ void __init setup_arch(char **cmdline_p) /* Initialize PROM console and command line. */ *cmdline_p = prom_getbootargs(); strcpy(boot_command_line, *cmdline_p); + parse_early_param(); boot_flags_init(*cmdline_p); register_console(&prom_early_console); diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c index 1c47009eb5e..77a3e8592cb 100644 --- a/arch/sparc64/kernel/signal.c +++ b/arch/sparc64/kernel/signal.c @@ -510,15 +510,20 @@ static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs, * want to handle. Thus you cannot kill init even with a SIGKILL even by * mistake. */ -static void do_signal(struct pt_regs *regs, unsigned long orig_i0, int restart_syscall) +static void do_signal(struct pt_regs *regs, unsigned long orig_i0) { - siginfo_t info; struct signal_deliver_cookie cookie; struct k_sigaction ka; - int signr; sigset_t *oldset; + siginfo_t info; + int signr, tt; - cookie.restart_syscall = restart_syscall; + tt = regs->magic & 0x1ff; + if (tt == 0x110 || tt == 0x111 || tt == 0x16d) { + regs->magic &= ~0x1ff; + cookie.restart_syscall = 1; + } else + cookie.restart_syscall = 0; cookie.orig_i0 = orig_i0; if (test_thread_flag(TIF_RESTORE_SIGMASK)) @@ -529,9 +534,8 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0, int restart_s #ifdef CONFIG_SPARC32_COMPAT if (test_thread_flag(TIF_32BIT)) { extern void do_signal32(sigset_t *, struct pt_regs *, - unsigned long, int); - do_signal32(oldset, regs, orig_i0, - cookie.restart_syscall); + struct signal_deliver_cookie *); + do_signal32(oldset, regs, &cookie); return; } #endif @@ -539,7 +543,7 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0, int restart_s signr = get_signal_to_deliver(&info, &ka, regs, &cookie); if (signr > 0) { if (cookie.restart_syscall) - syscall_restart(orig_i0, regs, &ka.sa); + syscall_restart(cookie.orig_i0, regs, &ka.sa); handle_signal(signr, &ka, &info, oldset, regs); /* a signal was successfully delivered; the saved @@ -576,11 +580,10 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0, int restart_s } } -void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, int restart_syscall, - unsigned long thread_info_flags) +void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long thread_info_flags) { if (thread_info_flags & (_TIF_SIGPENDING | _TIF_RESTORE_SIGMASK)) - do_signal(regs, orig_i0, restart_syscall); + do_signal(regs, orig_i0); } void ptrace_signal_deliver(struct pt_regs *regs, void *cookie) diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c index 74e0512f135..43cdec64d9c 100644 --- a/arch/sparc64/kernel/signal32.c +++ b/arch/sparc64/kernel/signal32.c @@ -982,20 +982,16 @@ static inline void syscall_restart32(unsigned long orig_i0, struct pt_regs *regs * mistake. */ void do_signal32(sigset_t *oldset, struct pt_regs * regs, - unsigned long orig_i0, int restart_syscall) + struct signal_deliver_cookie *cookie) { - siginfo_t info; - struct signal_deliver_cookie cookie; struct k_sigaction ka; + siginfo_t info; int signr; - cookie.restart_syscall = restart_syscall; - cookie.orig_i0 = orig_i0; - - signr = get_signal_to_deliver(&info, &ka, regs, &cookie); + signr = get_signal_to_deliver(&info, &ka, regs, cookie); if (signr > 0) { - if (cookie.restart_syscall) - syscall_restart32(orig_i0, regs, &ka.sa); + if (cookie->restart_syscall) + syscall_restart32(cookie->orig_i0, regs, &ka.sa); handle_signal32(signr, &ka, &info, oldset, regs); /* a signal was successfully delivered; the saved @@ -1007,16 +1003,16 @@ void do_signal32(sigset_t *oldset, struct pt_regs * regs, clear_thread_flag(TIF_RESTORE_SIGMASK); return; } - if (cookie.restart_syscall && + if (cookie->restart_syscall && (regs->u_regs[UREG_I0] == ERESTARTNOHAND || regs->u_regs[UREG_I0] == ERESTARTSYS || regs->u_regs[UREG_I0] == ERESTARTNOINTR)) { /* replay the system call when we are done */ - regs->u_regs[UREG_I0] = cookie.orig_i0; + regs->u_regs[UREG_I0] = cookie->orig_i0; regs->tpc -= 4; regs->tnpc -= 4; } - if (cookie.restart_syscall && + if (cookie->restart_syscall && regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) { regs->u_regs[UREG_G1] = __NR_restart_syscall; regs->tpc -= 4; diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index 59f020d69d4..409dd71f273 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -20,7 +20,7 @@ #include <linux/cache.h> #include <linux/jiffies.h> #include <linux/profile.h> -#include <linux/bootmem.h> +#include <linux/lmb.h> #include <asm/head.h> #include <asm/ptrace.h> @@ -866,14 +866,21 @@ void smp_call_function_client(int irq, struct pt_regs *regs) void *info = call_data->info; clear_softint(1 << irq); + + irq_enter(); + + if (!call_data->wait) { + /* let initiator proceed after getting data */ + atomic_inc(&call_data->finished); + } + + func(info); + + irq_exit(); + if (call_data->wait) { /* let initiator proceed only after completion */ - func(info); atomic_inc(&call_data->finished); - } else { - /* let initiator proceed after getting data */ - atomic_inc(&call_data->finished); - func(info); } } @@ -1032,7 +1039,9 @@ void smp_receive_signal(int cpu) void smp_receive_signal_client(int irq, struct pt_regs *regs) { + irq_enter(); clear_softint(1 << irq); + irq_exit(); } void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs) @@ -1040,6 +1049,8 @@ void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs) struct mm_struct *mm; unsigned long flags; + irq_enter(); + clear_softint(1 << irq); /* See if we need to allocate a new TLB context because @@ -1059,6 +1070,8 @@ void smp_new_mmu_context_version_client(int irq, struct pt_regs *regs) load_secondary_context(mm); __flush_tlb_mm(CTX_HWBITS(mm->context), SECONDARY_CONTEXT); + + irq_exit(); } void smp_new_mmu_context_version(void) @@ -1217,6 +1230,8 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs) { clear_softint(1 << irq); + irq_enter(); + preempt_disable(); __asm__ __volatile__("flushw"); @@ -1229,6 +1244,8 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs) prom_world(0); preempt_enable(); + + irq_exit(); } /* /proc/profile writes can call this, don't __init it please. */ @@ -1431,7 +1448,7 @@ EXPORT_SYMBOL(__per_cpu_shift); void __init real_setup_per_cpu_areas(void) { - unsigned long goal, size, i; + unsigned long paddr, goal, size, i; char *ptr; /* Copy section for each CPU (we discard the original) */ @@ -1441,8 +1458,13 @@ void __init real_setup_per_cpu_areas(void) for (size = PAGE_SIZE; size < goal; size <<= 1UL) __per_cpu_shift++; - ptr = alloc_bootmem_pages(size * NR_CPUS); + paddr = lmb_alloc(size * NR_CPUS, PAGE_SIZE); + if (!paddr) { + prom_printf("Cannot allocate per-cpu memory.\n"); + prom_halt(); + } + ptr = __va(paddr); __per_cpu_base = ptr - __per_cpu_start; for (i = 0; i < NR_CPUS; i++, ptr += size) diff --git a/arch/sparc64/kernel/sparc64_ksyms.c b/arch/sparc64/kernel/sparc64_ksyms.c index 38736460b8d..66336590e83 100644 --- a/arch/sparc64/kernel/sparc64_ksyms.c +++ b/arch/sparc64/kernel/sparc64_ksyms.c @@ -68,8 +68,6 @@ extern void *__memscan_zero(void *, size_t); extern void *__memscan_generic(void *, int, size_t); extern int __memcmp(const void *, const void *, __kernel_size_t); extern __kernel_size_t strlen(const char *); -extern void linux_sparc_syscall(void); -extern void rtrap(void); extern void show_regs(struct pt_regs *); extern void syscall_trace(struct pt_regs *, int); extern void sys_sigsuspend(void); diff --git a/arch/sparc64/kernel/stacktrace.c b/arch/sparc64/kernel/stacktrace.c index 84d39e873e8..01b52f561af 100644 --- a/arch/sparc64/kernel/stacktrace.c +++ b/arch/sparc64/kernel/stacktrace.c @@ -20,6 +20,8 @@ void save_stack_trace(struct stack_trace *trace) thread_base = (unsigned long) tp; do { struct reg_window *rw; + struct pt_regs *regs; + unsigned long pc; /* Bogus frame pointer? */ if (fp < (thread_base + sizeof(struct thread_info)) || @@ -27,11 +29,19 @@ void save_stack_trace(struct stack_trace *trace) break; rw = (struct reg_window *) fp; + regs = (struct pt_regs *) (rw + 1); + + if ((regs->magic & ~0x1ff) == PT_REGS_MAGIC) { + pc = regs->tpc; + fp = regs->u_regs[UREG_I6] + STACK_BIAS; + } else { + pc = rw->ins[7]; + fp = rw->ins[6] + STACK_BIAS; + } + if (trace->skip > 0) trace->skip--; else - trace->entries[trace->nr_entries++] = rw->ins[7]; - - fp = rw->ins[6] + STACK_BIAS; + trace->entries[trace->nr_entries++] = pc; } while (trace->nr_entries < trace->max_entries); } diff --git a/arch/sparc64/kernel/sun4v_tlb_miss.S b/arch/sparc64/kernel/sun4v_tlb_miss.S index fd9430562e0..e1fbf8c7578 100644 --- a/arch/sparc64/kernel/sun4v_tlb_miss.S +++ b/arch/sparc64/kernel/sun4v_tlb_miss.S @@ -262,7 +262,7 @@ sun4v_iacc: mov %l5, %o2 call sun4v_insn_access_exception add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap /* Instruction Access Exception, tl1. */ sun4v_iacc_tl1: @@ -278,7 +278,7 @@ sun4v_iacc_tl1: mov %l5, %o2 call sun4v_insn_access_exception_tl1 add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap /* Data Access Exception, tl0. */ sun4v_dacc: @@ -294,7 +294,7 @@ sun4v_dacc: mov %l5, %o2 call sun4v_data_access_exception add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap /* Data Access Exception, tl1. */ sun4v_dacc_tl1: @@ -310,7 +310,7 @@ sun4v_dacc_tl1: mov %l5, %o2 call sun4v_data_access_exception_tl1 add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap /* Memory Address Unaligned. */ sun4v_mna: @@ -344,7 +344,7 @@ sun4v_mna: mov %l5, %o2 call sun4v_do_mna add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap /* Privileged Action. */ sun4v_privact: @@ -352,7 +352,7 @@ sun4v_privact: rd %pc, %g7 call do_privact add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap /* Unaligned ldd float, tl0. */ sun4v_lddfmna: @@ -368,7 +368,7 @@ sun4v_lddfmna: mov %l5, %o2 call handle_lddfmna add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap /* Unaligned std float, tl0. */ sun4v_stdfmna: @@ -384,7 +384,7 @@ sun4v_stdfmna: mov %l5, %o2 call handle_stdfmna add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap #define BRANCH_ALWAYS 0x10680000 #define NOP 0x01000000 diff --git a/arch/sparc64/kernel/sys_sparc.c b/arch/sparc64/kernel/sys_sparc.c index 73ed01ba40d..8d4761f15fa 100644 --- a/arch/sparc64/kernel/sys_sparc.c +++ b/arch/sparc64/kernel/sys_sparc.c @@ -454,8 +454,8 @@ asmlinkage long sys_ipc(unsigned int call, int first, unsigned long second, err = sys_semget(first, (int)second, (int)third); goto out; case SEMCTL: { - err = sys_semctl(first, third, - (int)second | IPC_64, + err = sys_semctl(first, second, + (int)third | IPC_64, (union semun) ptr); goto out; } diff --git a/arch/sparc64/kernel/sysfs.c b/arch/sparc64/kernel/sysfs.c index 52816c7be0b..e885034a6b7 100644 --- a/arch/sparc64/kernel/sysfs.c +++ b/arch/sparc64/kernel/sysfs.c @@ -273,10 +273,22 @@ static void __init check_mmu_stats(void) mmu_stats_supported = 1; } +static void register_nodes(void) +{ +#ifdef CONFIG_NUMA + int i; + + for (i = 0; i < MAX_NUMNODES; i++) + register_one_node(i); +#endif +} + static int __init topology_init(void) { int cpu; + register_nodes(); + check_mmu_stats(); register_cpu_notifier(&sysfs_cpu_nb); diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c index 96da847023f..d9b8d46707d 100644 --- a/arch/sparc64/kernel/traps.c +++ b/arch/sparc64/kernel/traps.c @@ -2091,9 +2091,8 @@ static void user_instruction_dump(unsigned int __user *pc) void show_stack(struct task_struct *tsk, unsigned long *_ksp) { - unsigned long pc, fp, thread_base, ksp; + unsigned long fp, thread_base, ksp; struct thread_info *tp; - struct reg_window *rw; int count = 0; ksp = (unsigned long) _ksp; @@ -2117,15 +2116,27 @@ void show_stack(struct task_struct *tsk, unsigned long *_ksp) printk("\n"); #endif do { + struct reg_window *rw; + struct pt_regs *regs; + unsigned long pc; + /* Bogus frame pointer? */ if (fp < (thread_base + sizeof(struct thread_info)) || fp >= (thread_base + THREAD_SIZE)) break; rw = (struct reg_window *)fp; - pc = rw->ins[7]; + regs = (struct pt_regs *) (rw + 1); + + if ((regs->magic & ~0x1ff) == PT_REGS_MAGIC) { + pc = regs->tpc; + fp = regs->u_regs[UREG_I6] + STACK_BIAS; + } else { + pc = rw->ins[7]; + fp = rw->ins[6] + STACK_BIAS; + } + printk(" [%016lx] ", pc); print_symbol("%s\n", pc); - fp = rw->ins[6] + STACK_BIAS; } while (++count < 16); #ifndef CONFIG_KALLSYMS printk("\n"); diff --git a/arch/sparc64/kernel/tsb.S b/arch/sparc64/kernel/tsb.S index 10adb2fb8ff..c499214b501 100644 --- a/arch/sparc64/kernel/tsb.S +++ b/arch/sparc64/kernel/tsb.S @@ -275,7 +275,7 @@ sparc64_realfault_common: stx %l5, [%g6 + TI_FAULT_ADDR] ! Save fault address call do_sparc64_fault ! Call fault handler add %sp, PTREGS_OFF, %o0 ! Compute pt_regs arg - ba,pt %xcc, rtrap_clr_l6 ! Restore cpu state + ba,pt %xcc, rtrap ! Restore cpu state nop ! Delay slot (fill me) winfix_trampoline: diff --git a/arch/sparc64/kernel/winfixup.S b/arch/sparc64/kernel/winfixup.S index c4aa110a10e..a6b0863c27d 100644 --- a/arch/sparc64/kernel/winfixup.S +++ b/arch/sparc64/kernel/winfixup.S @@ -32,7 +32,7 @@ fill_fixup: rd %pc, %g7 call do_sparc64_fault add %sp, PTREGS_OFF, %o0 - ba,pt %xcc, rtrap_clr_l6 + ba,pt %xcc, rtrap nop /* Be very careful about usage of the trap globals here. @@ -100,7 +100,7 @@ spill_fixup_dax: rd %pc, %g7 call do_sparc64_fault add %sp, PTREGS_OFF, %o0 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap winfix_mna: andn %g3, 0x7f, %g3 @@ -122,12 +122,12 @@ fill_fixup_mna: mov %l4, %o2 call sun4v_do_mna mov %l5, %o1 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap 1: mov %l4, %o1 mov %l5, %o2 call mem_address_unaligned nop - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap winfix_dax: andn %g3, 0x7f, %g3 @@ -150,7 +150,7 @@ fill_fixup_dax: add %sp, PTREGS_OFF, %o0 call sun4v_data_access_exception nop - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap 1: call spitfire_data_access_exception nop - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index f37078d9640..177d8aaeec4 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -24,6 +24,8 @@ #include <linux/cache.h> #include <linux/sort.h> #include <linux/percpu.h> +#include <linux/lmb.h> +#include <linux/mmzone.h> #include <asm/head.h> #include <asm/system.h> @@ -72,9 +74,7 @@ extern struct tsb swapper_4m_tsb[KERNEL_TSB4M_NENTRIES]; #define MAX_BANKS 32 static struct linux_prom64_registers pavail[MAX_BANKS] __initdata; -static struct linux_prom64_registers pavail_rescan[MAX_BANKS] __initdata; static int pavail_ents __initdata; -static int pavail_rescan_ents __initdata; static int cmp_p64(const void *a, const void *b) { @@ -715,285 +715,684 @@ out: smp_new_mmu_context_version(); } -/* Find a free area for the bootmem map, avoiding the kernel image - * and the initial ramdisk. - */ -static unsigned long __init choose_bootmap_pfn(unsigned long start_pfn, - unsigned long end_pfn) +static int numa_enabled = 1; +static int numa_debug; + +static int __init early_numa(char *p) { - unsigned long avoid_start, avoid_end, bootmap_size; - int i; + if (!p) + return 0; + + if (strstr(p, "off")) + numa_enabled = 0; + + if (strstr(p, "debug")) + numa_debug = 1; + + return 0; +} +early_param("numa", early_numa); - bootmap_size = bootmem_bootmap_pages(end_pfn - start_pfn); - bootmap_size <<= PAGE_SHIFT; +#define numadbg(f, a...) \ +do { if (numa_debug) \ + printk(KERN_INFO f, ## a); \ +} while (0) - avoid_start = avoid_end = 0; +static void __init find_ramdisk(unsigned long phys_base) +{ #ifdef CONFIG_BLK_DEV_INITRD - avoid_start = initrd_start; - avoid_end = PAGE_ALIGN(initrd_end); + if (sparc_ramdisk_image || sparc_ramdisk_image64) { + unsigned long ramdisk_image; + + /* Older versions of the bootloader only supported a + * 32-bit physical address for the ramdisk image + * location, stored at sparc_ramdisk_image. Newer + * SILO versions set sparc_ramdisk_image to zero and + * provide a full 64-bit physical address at + * sparc_ramdisk_image64. + */ + ramdisk_image = sparc_ramdisk_image; + if (!ramdisk_image) + ramdisk_image = sparc_ramdisk_image64; + + /* Another bootloader quirk. The bootloader normalizes + * the physical address to KERNBASE, so we have to + * factor that back out and add in the lowest valid + * physical page address to get the true physical address. + */ + ramdisk_image -= KERNBASE; + ramdisk_image += phys_base; + + numadbg("Found ramdisk at physical address 0x%lx, size %u\n", + ramdisk_image, sparc_ramdisk_size); + + initrd_start = ramdisk_image; + initrd_end = ramdisk_image + sparc_ramdisk_size; + + lmb_reserve(initrd_start, initrd_end); + } #endif +} - for (i = 0; i < pavail_ents; i++) { - unsigned long start, end; +struct node_mem_mask { + unsigned long mask; + unsigned long val; + unsigned long bootmem_paddr; +}; +static struct node_mem_mask node_masks[MAX_NUMNODES]; +static int num_node_masks; - start = pavail[i].phys_addr; - end = start + pavail[i].reg_size; +int numa_cpu_lookup_table[NR_CPUS]; +cpumask_t numa_cpumask_lookup_table[MAX_NUMNODES]; - while (start < end) { - if (start >= kern_base && - start < PAGE_ALIGN(kern_base + kern_size)) { - start = PAGE_ALIGN(kern_base + kern_size); - continue; - } - if (start >= avoid_start && start < avoid_end) { - start = avoid_end; - continue; - } +#ifdef CONFIG_NEED_MULTIPLE_NODES +static bootmem_data_t plat_node_bdata[MAX_NUMNODES]; - if ((end - start) < bootmap_size) - break; +struct mdesc_mblock { + u64 base; + u64 size; + u64 offset; /* RA-to-PA */ +}; +static struct mdesc_mblock *mblocks; +static int num_mblocks; - if (start < kern_base && - (start + bootmap_size) > kern_base) { - start = PAGE_ALIGN(kern_base + kern_size); - continue; - } +static unsigned long ra_to_pa(unsigned long addr) +{ + int i; - if (start < avoid_start && - (start + bootmap_size) > avoid_start) { - start = avoid_end; - continue; - } + for (i = 0; i < num_mblocks; i++) { + struct mdesc_mblock *m = &mblocks[i]; - /* OK, it doesn't overlap anything, use it. */ - return start >> PAGE_SHIFT; + if (addr >= m->base && + addr < (m->base + m->size)) { + addr += m->offset; + break; } } - - prom_printf("Cannot find free area for bootmap, aborting.\n"); - prom_halt(); + return addr; } -static void __init trim_pavail(unsigned long *cur_size_p, - unsigned long *end_of_phys_p) +static int find_node(unsigned long addr) { - unsigned long to_trim = *cur_size_p - cmdline_memory_size; - unsigned long avoid_start, avoid_end; int i; - to_trim = PAGE_ALIGN(to_trim); + addr = ra_to_pa(addr); + for (i = 0; i < num_node_masks; i++) { + struct node_mem_mask *p = &node_masks[i]; - avoid_start = avoid_end = 0; -#ifdef CONFIG_BLK_DEV_INITRD - avoid_start = initrd_start; - avoid_end = PAGE_ALIGN(initrd_end); + if ((addr & p->mask) == p->val) + return i; + } + return -1; +} + +static unsigned long nid_range(unsigned long start, unsigned long end, + int *nid) +{ + *nid = find_node(start); + start += PAGE_SIZE; + while (start < end) { + int n = find_node(start); + + if (n != *nid) + break; + start += PAGE_SIZE; + } + + return start; +} +#else +static unsigned long nid_range(unsigned long start, unsigned long end, + int *nid) +{ + *nid = 0; + return end; +} #endif - /* Trim some pavail[] entries in order to satisfy the - * requested "mem=xxx" kernel command line specification. - * - * We must not trim off the kernel image area nor the - * initial ramdisk range (if any). Also, we must not trim - * any pavail[] entry down to zero in order to preserve - * the invariant that all pavail[] entries have a non-zero - * size which is assumed by all of the code in here. - */ - for (i = 0; i < pavail_ents; i++) { - unsigned long start, end, kern_end; - unsigned long trim_low, trim_high, n; +/* This must be invoked after performing all of the necessary + * add_active_range() calls for 'nid'. We need to be able to get + * correct data from get_pfn_range_for_nid(). + */ +static void __init allocate_node_data(int nid) +{ + unsigned long paddr, num_pages, start_pfn, end_pfn; + struct pglist_data *p; + +#ifdef CONFIG_NEED_MULTIPLE_NODES + paddr = lmb_alloc_nid(sizeof(struct pglist_data), + SMP_CACHE_BYTES, nid, nid_range); + if (!paddr) { + prom_printf("Cannot allocate pglist_data for nid[%d]\n", nid); + prom_halt(); + } + NODE_DATA(nid) = __va(paddr); + memset(NODE_DATA(nid), 0, sizeof(struct pglist_data)); - kern_end = PAGE_ALIGN(kern_base + kern_size); + NODE_DATA(nid)->bdata = &plat_node_bdata[nid]; +#endif - trim_low = start = pavail[i].phys_addr; - trim_high = end = start + pavail[i].reg_size; + p = NODE_DATA(nid); - if (kern_base >= start && - kern_base < end) { - trim_low = kern_base; - if (kern_end >= end) - continue; - } - if (kern_end >= start && - kern_end < end) { - trim_high = kern_end; - } - if (avoid_start && - avoid_start >= start && - avoid_start < end) { - if (trim_low > avoid_start) - trim_low = avoid_start; - if (avoid_end >= end) - continue; - } - if (avoid_end && - avoid_end >= start && - avoid_end < end) { - if (trim_high < avoid_end) - trim_high = avoid_end; + get_pfn_range_for_nid(nid, &start_pfn, &end_pfn); + p->node_start_pfn = start_pfn; + p->node_spanned_pages = end_pfn - start_pfn; + + if (p->node_spanned_pages) { + num_pages = bootmem_bootmap_pages(p->node_spanned_pages); + + paddr = lmb_alloc_nid(num_pages << PAGE_SHIFT, PAGE_SIZE, nid, + nid_range); + if (!paddr) { + prom_printf("Cannot allocate bootmap for nid[%d]\n", + nid); + prom_halt(); } + node_masks[nid].bootmem_paddr = paddr; + } +} + +static void init_node_masks_nonnuma(void) +{ + int i; + + numadbg("Initializing tables for non-numa.\n"); + + node_masks[0].mask = node_masks[0].val = 0; + num_node_masks = 1; + + for (i = 0; i < NR_CPUS; i++) + numa_cpu_lookup_table[i] = 0; + + numa_cpumask_lookup_table[0] = CPU_MASK_ALL; +} + +#ifdef CONFIG_NEED_MULTIPLE_NODES +struct pglist_data *node_data[MAX_NUMNODES]; + +EXPORT_SYMBOL(numa_cpu_lookup_table); +EXPORT_SYMBOL(numa_cpumask_lookup_table); +EXPORT_SYMBOL(node_data); + +struct mdesc_mlgroup { + u64 node; + u64 latency; + u64 match; + u64 mask; +}; +static struct mdesc_mlgroup *mlgroups; +static int num_mlgroups; + +static int scan_pio_for_cfg_handle(struct mdesc_handle *md, u64 pio, + u32 cfg_handle) +{ + u64 arc; - if (trim_high <= trim_low) + mdesc_for_each_arc(arc, md, pio, MDESC_ARC_TYPE_FWD) { + u64 target = mdesc_arc_target(md, arc); + const u64 *val; + + val = mdesc_get_property(md, target, + "cfg-handle", NULL); + if (val && *val == cfg_handle) + return 0; + } + return -ENODEV; +} + +static int scan_arcs_for_cfg_handle(struct mdesc_handle *md, u64 grp, + u32 cfg_handle) +{ + u64 arc, candidate, best_latency = ~(u64)0; + + candidate = MDESC_NODE_NULL; + mdesc_for_each_arc(arc, md, grp, MDESC_ARC_TYPE_FWD) { + u64 target = mdesc_arc_target(md, arc); + const char *name = mdesc_node_name(md, target); + const u64 *val; + + if (strcmp(name, "pio-latency-group")) continue; - if (trim_low == start && trim_high == end) { - /* Whole chunk is available for trimming. - * Trim all except one page, in order to keep - * entry non-empty. - */ - n = (end - start) - PAGE_SIZE; - if (n > to_trim) - n = to_trim; - - if (n) { - pavail[i].phys_addr += n; - pavail[i].reg_size -= n; - to_trim -= n; - } - } else { - n = (trim_low - start); - if (n > to_trim) - n = to_trim; - - if (n) { - pavail[i].phys_addr += n; - pavail[i].reg_size -= n; - to_trim -= n; - } - if (to_trim) { - n = end - trim_high; - if (n > to_trim) - n = to_trim; - if (n) { - pavail[i].reg_size -= n; - to_trim -= n; - } - } + val = mdesc_get_property(md, target, "latency", NULL); + if (!val) + continue; + + if (*val < best_latency) { + candidate = target; + best_latency = *val; } + } + + if (candidate == MDESC_NODE_NULL) + return -ENODEV; + + return scan_pio_for_cfg_handle(md, candidate, cfg_handle); +} + +int of_node_to_nid(struct device_node *dp) +{ + const struct linux_prom64_registers *regs; + struct mdesc_handle *md; + u32 cfg_handle; + int count, nid; + u64 grp; - if (!to_trim) + if (!mlgroups) + return -1; + + regs = of_get_property(dp, "reg", NULL); + if (!regs) + return -1; + + cfg_handle = (regs->phys_addr >> 32UL) & 0x0fffffff; + + md = mdesc_grab(); + + count = 0; + nid = -1; + mdesc_for_each_node_by_name(md, grp, "group") { + if (!scan_arcs_for_cfg_handle(md, grp, cfg_handle)) { + nid = count; break; + } + count++; } - /* Recalculate. */ - *cur_size_p = 0UL; - for (i = 0; i < pavail_ents; i++) { - *end_of_phys_p = pavail[i].phys_addr + - pavail[i].reg_size; - *cur_size_p += pavail[i].reg_size; - } + mdesc_release(md); + + return nid; } -/* About pages_avail, this is the value we will use to calculate - * the zholes_size[] argument given to free_area_init_node(). The - * page allocator uses this to calculate nr_kernel_pages, - * nr_all_pages and zone->present_pages. On NUMA it is used - * to calculate zone->min_unmapped_pages and zone->min_slab_pages. - * - * So this number should really be set to what the page allocator - * actually ends up with. This means: - * 1) It should include bootmem map pages, we'll release those. - * 2) It should not include the kernel image, except for the - * __init sections which we will also release. - * 3) It should include the initrd image, since we'll release - * that too. - */ -static unsigned long __init bootmem_init(unsigned long *pages_avail, - unsigned long phys_base) +static void add_node_ranges(void) { - unsigned long bootmap_size, end_pfn; - unsigned long end_of_phys_memory = 0UL; - unsigned long bootmap_pfn, bytes_avail, size; int i; - bytes_avail = 0UL; - for (i = 0; i < pavail_ents; i++) { - end_of_phys_memory = pavail[i].phys_addr + - pavail[i].reg_size; - bytes_avail += pavail[i].reg_size; + for (i = 0; i < lmb.memory.cnt; i++) { + unsigned long size = lmb_size_bytes(&lmb.memory, i); + unsigned long start, end; + + start = lmb.memory.region[i].base; + end = start + size; + while (start < end) { + unsigned long this_end; + int nid; + + this_end = nid_range(start, end, &nid); + + numadbg("Adding active range nid[%d] " + "start[%lx] end[%lx]\n", + nid, start, this_end); + + add_active_range(nid, + start >> PAGE_SHIFT, + this_end >> PAGE_SHIFT); + + start = this_end; + } } +} - /* Determine the location of the initial ramdisk before trying - * to honor the "mem=xxx" command line argument. We must know - * where the kernel image and the ramdisk image are so that we - * do not trim those two areas from the physical memory map. - */ +static int __init grab_mlgroups(struct mdesc_handle *md) +{ + unsigned long paddr; + int count = 0; + u64 node; + + mdesc_for_each_node_by_name(md, node, "memory-latency-group") + count++; + if (!count) + return -ENOENT; + + paddr = lmb_alloc(count * sizeof(struct mdesc_mlgroup), + SMP_CACHE_BYTES); + if (!paddr) + return -ENOMEM; + + mlgroups = __va(paddr); + num_mlgroups = count; + + count = 0; + mdesc_for_each_node_by_name(md, node, "memory-latency-group") { + struct mdesc_mlgroup *m = &mlgroups[count++]; + const u64 *val; + + m->node = node; + + val = mdesc_get_property(md, node, "latency", NULL); + m->latency = *val; + val = mdesc_get_property(md, node, "address-match", NULL); + m->match = *val; + val = mdesc_get_property(md, node, "address-mask", NULL); + m->mask = *val; + + numadbg("MLGROUP[%d]: node[%lx] latency[%lx] " + "match[%lx] mask[%lx]\n", + count - 1, m->node, m->latency, m->match, m->mask); + } -#ifdef CONFIG_BLK_DEV_INITRD - /* Now have to check initial ramdisk, so that bootmap does not overwrite it */ - if (sparc_ramdisk_image || sparc_ramdisk_image64) { - unsigned long ramdisk_image = sparc_ramdisk_image ? - sparc_ramdisk_image : sparc_ramdisk_image64; - ramdisk_image -= KERNBASE; - initrd_start = ramdisk_image + phys_base; - initrd_end = initrd_start + sparc_ramdisk_size; - if (initrd_end > end_of_phys_memory) { - printk(KERN_CRIT "initrd extends beyond end of memory " - "(0x%016lx > 0x%016lx)\ndisabling initrd\n", - initrd_end, end_of_phys_memory); - initrd_start = 0; - initrd_end = 0; + return 0; +} + +static int __init grab_mblocks(struct mdesc_handle *md) +{ + unsigned long paddr; + int count = 0; + u64 node; + + mdesc_for_each_node_by_name(md, node, "mblock") + count++; + if (!count) + return -ENOENT; + + paddr = lmb_alloc(count * sizeof(struct mdesc_mblock), + SMP_CACHE_BYTES); + if (!paddr) + return -ENOMEM; + + mblocks = __va(paddr); + num_mblocks = count; + + count = 0; + mdesc_for_each_node_by_name(md, node, "mblock") { + struct mdesc_mblock *m = &mblocks[count++]; + const u64 *val; + + val = mdesc_get_property(md, node, "base", NULL); + m->base = *val; + val = mdesc_get_property(md, node, "size", NULL); + m->size = *val; + val = mdesc_get_property(md, node, + "address-congruence-offset", NULL); + m->offset = *val; + + numadbg("MBLOCK[%d]: base[%lx] size[%lx] offset[%lx]\n", + count - 1, m->base, m->size, m->offset); + } + + return 0; +} + +static void __init numa_parse_mdesc_group_cpus(struct mdesc_handle *md, + u64 grp, cpumask_t *mask) +{ + u64 arc; + + cpus_clear(*mask); + + mdesc_for_each_arc(arc, md, grp, MDESC_ARC_TYPE_BACK) { + u64 target = mdesc_arc_target(md, arc); + const char *name = mdesc_node_name(md, target); + const u64 *id; + + if (strcmp(name, "cpu")) + continue; + id = mdesc_get_property(md, target, "id", NULL); + if (*id < NR_CPUS) + cpu_set(*id, *mask); + } +} + +static struct mdesc_mlgroup * __init find_mlgroup(u64 node) +{ + int i; + + for (i = 0; i < num_mlgroups; i++) { + struct mdesc_mlgroup *m = &mlgroups[i]; + if (m->node == node) + return m; + } + return NULL; +} + +static int __init numa_attach_mlgroup(struct mdesc_handle *md, u64 grp, + int index) +{ + struct mdesc_mlgroup *candidate = NULL; + u64 arc, best_latency = ~(u64)0; + struct node_mem_mask *n; + + mdesc_for_each_arc(arc, md, grp, MDESC_ARC_TYPE_FWD) { + u64 target = mdesc_arc_target(md, arc); + struct mdesc_mlgroup *m = find_mlgroup(target); + if (!m) + continue; + if (m->latency < best_latency) { + candidate = m; + best_latency = m->latency; } } -#endif + if (!candidate) + return -ENOENT; + + if (num_node_masks != index) { + printk(KERN_ERR "Inconsistent NUMA state, " + "index[%d] != num_node_masks[%d]\n", + index, num_node_masks); + return -EINVAL; + } - if (cmdline_memory_size && - bytes_avail > cmdline_memory_size) - trim_pavail(&bytes_avail, - &end_of_phys_memory); + n = &node_masks[num_node_masks++]; - *pages_avail = bytes_avail >> PAGE_SHIFT; + n->mask = candidate->mask; + n->val = candidate->match; - end_pfn = end_of_phys_memory >> PAGE_SHIFT; + numadbg("NUMA NODE[%d]: mask[%lx] val[%lx] (latency[%lx])\n", + index, n->mask, n->val, candidate->latency); - /* Initialize the boot-time allocator. */ - max_pfn = max_low_pfn = end_pfn; - min_low_pfn = (phys_base >> PAGE_SHIFT); + return 0; +} + +static int __init numa_parse_mdesc_group(struct mdesc_handle *md, u64 grp, + int index) +{ + cpumask_t mask; + int cpu; - bootmap_pfn = choose_bootmap_pfn(min_low_pfn, end_pfn); + numa_parse_mdesc_group_cpus(md, grp, &mask); - bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap_pfn, - min_low_pfn, end_pfn); + for_each_cpu_mask(cpu, mask) + numa_cpu_lookup_table[cpu] = index; + numa_cpumask_lookup_table[index] = mask; - /* Now register the available physical memory with the - * allocator. - */ - for (i = 0; i < pavail_ents; i++) - free_bootmem(pavail[i].phys_addr, pavail[i].reg_size); + if (numa_debug) { + printk(KERN_INFO "NUMA GROUP[%d]: cpus [ ", index); + for_each_cpu_mask(cpu, mask) + printk("%d ", cpu); + printk("]\n"); + } -#ifdef CONFIG_BLK_DEV_INITRD - if (initrd_start) { - size = initrd_end - initrd_start; + return numa_attach_mlgroup(md, grp, index); +} + +static int __init numa_parse_mdesc(void) +{ + struct mdesc_handle *md = mdesc_grab(); + int i, err, count; + u64 node; + + node = mdesc_node_by_name(md, MDESC_NODE_NULL, "latency-groups"); + if (node == MDESC_NODE_NULL) { + mdesc_release(md); + return -ENOENT; + } + + err = grab_mblocks(md); + if (err < 0) + goto out; + + err = grab_mlgroups(md); + if (err < 0) + goto out; + + count = 0; + mdesc_for_each_node_by_name(md, node, "group") { + err = numa_parse_mdesc_group(md, node, count); + if (err < 0) + break; + count++; + } + + add_node_ranges(); + + for (i = 0; i < num_node_masks; i++) { + allocate_node_data(i); + node_set_online(i); + } + + err = 0; +out: + mdesc_release(md); + return err; +} + +static int __init numa_parse_sun4u(void) +{ + return -1; +} - /* Reserve the initrd image area. */ - reserve_bootmem(initrd_start, size, BOOTMEM_DEFAULT); +static int __init bootmem_init_numa(void) +{ + int err = -1; - initrd_start += PAGE_OFFSET; - initrd_end += PAGE_OFFSET; + numadbg("bootmem_init_numa()\n"); + + if (numa_enabled) { + if (tlb_type == hypervisor) + err = numa_parse_mdesc(); + else + err = numa_parse_sun4u(); } + return err; +} + +#else + +static int bootmem_init_numa(void) +{ + return -1; +} + #endif - /* Reserve the kernel text/data/bss. */ - reserve_bootmem(kern_base, kern_size, BOOTMEM_DEFAULT); - *pages_avail -= PAGE_ALIGN(kern_size) >> PAGE_SHIFT; - - /* Add back in the initmem pages. */ - size = ((unsigned long)(__init_end) & PAGE_MASK) - - PAGE_ALIGN((unsigned long)__init_begin); - *pages_avail += size >> PAGE_SHIFT; - - /* Reserve the bootmem map. We do not account for it - * in pages_avail because we will release that memory - * in free_all_bootmem. - */ - size = bootmap_size; - reserve_bootmem((bootmap_pfn << PAGE_SHIFT), size, BOOTMEM_DEFAULT); - for (i = 0; i < pavail_ents; i++) { +static void __init bootmem_init_nonnuma(void) +{ + unsigned long top_of_ram = lmb_end_of_DRAM(); + unsigned long total_ram = lmb_phys_mem_size(); + unsigned int i; + + numadbg("bootmem_init_nonnuma()\n"); + + printk(KERN_INFO "Top of RAM: 0x%lx, Total RAM: 0x%lx\n", + top_of_ram, total_ram); + printk(KERN_INFO "Memory hole size: %ldMB\n", + (top_of_ram - total_ram) >> 20); + + init_node_masks_nonnuma(); + + for (i = 0; i < lmb.memory.cnt; i++) { + unsigned long size = lmb_size_bytes(&lmb.memory, i); unsigned long start_pfn, end_pfn; - start_pfn = pavail[i].phys_addr >> PAGE_SHIFT; - end_pfn = (start_pfn + (pavail[i].reg_size >> PAGE_SHIFT)); - memory_present(0, start_pfn, end_pfn); + if (!size) + continue; + + start_pfn = lmb.memory.region[i].base >> PAGE_SHIFT; + end_pfn = start_pfn + lmb_size_pages(&lmb.memory, i); + add_active_range(0, start_pfn, end_pfn); + } + + allocate_node_data(0); + + node_set_online(0); +} + +static void __init reserve_range_in_node(int nid, unsigned long start, + unsigned long end) +{ + numadbg(" reserve_range_in_node(nid[%d],start[%lx],end[%lx]\n", + nid, start, end); + while (start < end) { + unsigned long this_end; + int n; + + this_end = nid_range(start, end, &n); + if (n == nid) { + numadbg(" MATCH reserving range [%lx:%lx]\n", + start, this_end); + reserve_bootmem_node(NODE_DATA(nid), start, + (this_end - start), BOOTMEM_DEFAULT); + } else + numadbg(" NO MATCH, advancing start to %lx\n", + this_end); + + start = this_end; + } +} + +static void __init trim_reserved_in_node(int nid) +{ + int i; + + numadbg(" trim_reserved_in_node(%d)\n", nid); + + for (i = 0; i < lmb.reserved.cnt; i++) { + unsigned long start = lmb.reserved.region[i].base; + unsigned long size = lmb_size_bytes(&lmb.reserved, i); + unsigned long end = start + size; + + reserve_range_in_node(nid, start, end); + } +} + +static void __init bootmem_init_one_node(int nid) +{ + struct pglist_data *p; + + numadbg("bootmem_init_one_node(%d)\n", nid); + + p = NODE_DATA(nid); + + if (p->node_spanned_pages) { + unsigned long paddr = node_masks[nid].bootmem_paddr; + unsigned long end_pfn; + + end_pfn = p->node_start_pfn + p->node_spanned_pages; + + numadbg(" init_bootmem_node(%d, %lx, %lx, %lx)\n", + nid, paddr >> PAGE_SHIFT, p->node_start_pfn, end_pfn); + + init_bootmem_node(p, paddr >> PAGE_SHIFT, + p->node_start_pfn, end_pfn); + + numadbg(" free_bootmem_with_active_regions(%d, %lx)\n", + nid, end_pfn); + free_bootmem_with_active_regions(nid, end_pfn); + + trim_reserved_in_node(nid); + + numadbg(" sparse_memory_present_with_active_regions(%d)\n", + nid); + sparse_memory_present_with_active_regions(nid); } +} + +static unsigned long __init bootmem_init(unsigned long phys_base) +{ + unsigned long end_pfn; + int nid; + + end_pfn = lmb_end_of_DRAM() >> PAGE_SHIFT; + max_pfn = max_low_pfn = end_pfn; + min_low_pfn = (phys_base >> PAGE_SHIFT); + + if (bootmem_init_numa() < 0) + bootmem_init_nonnuma(); + + /* XXX cpu notifier XXX */ + + for_each_online_node(nid) + bootmem_init_one_node(nid); sparse_init(); @@ -1289,7 +1688,7 @@ void __init setup_per_cpu_areas(void) void __init paging_init(void) { - unsigned long end_pfn, pages_avail, shift, phys_base; + unsigned long end_pfn, shift, phys_base; unsigned long real_end, i; /* These build time checkes make sure that the dcache_dirty_cpu() @@ -1330,12 +1729,26 @@ void __init paging_init(void) sun4v_ktsb_init(); } + lmb_init(); + /* Find available physical memory... */ read_obp_memory("available", &pavail[0], &pavail_ents); phys_base = 0xffffffffffffffffUL; - for (i = 0; i < pavail_ents; i++) + for (i = 0; i < pavail_ents; i++) { phys_base = min(phys_base, pavail[i].phys_addr); + lmb_add(pavail[i].phys_addr, pavail[i].reg_size); + } + + lmb_reserve(kern_base, kern_size); + + find_ramdisk(phys_base); + + if (cmdline_memory_size) + lmb_enforce_memory_limit(phys_base + cmdline_memory_size); + + lmb_analyze(); + lmb_dump_all(); set_bit(0, mmu_context_bmap); @@ -1371,14 +1784,10 @@ void __init paging_init(void) if (tlb_type == hypervisor) sun4v_ktsb_register(); - /* Setup bootmem... */ - pages_avail = 0; - last_valid_pfn = end_pfn = bootmem_init(&pages_avail, phys_base); - - max_mapnr = last_valid_pfn; - - kernel_physical_mapping_init(); - + /* We must setup the per-cpu areas before we pull in the + * PROM and the MDESC. The code there fills in cpu and + * other information into per-cpu data structures. + */ real_setup_per_cpu_areas(); prom_build_devicetree(); @@ -1386,20 +1795,22 @@ void __init paging_init(void) if (tlb_type == hypervisor) sun4v_mdesc_init(); + /* Setup bootmem... */ + last_valid_pfn = end_pfn = bootmem_init(phys_base); + +#ifndef CONFIG_NEED_MULTIPLE_NODES + max_mapnr = last_valid_pfn; +#endif + kernel_physical_mapping_init(); + { - unsigned long zones_size[MAX_NR_ZONES]; - unsigned long zholes_size[MAX_NR_ZONES]; - int znum; + unsigned long max_zone_pfns[MAX_NR_ZONES]; - for (znum = 0; znum < MAX_NR_ZONES; znum++) - zones_size[znum] = zholes_size[znum] = 0; + memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); - zones_size[ZONE_NORMAL] = end_pfn; - zholes_size[ZONE_NORMAL] = end_pfn - pages_avail; + max_zone_pfns[ZONE_NORMAL] = end_pfn; - free_area_init_node(0, &contig_page_data, zones_size, - __pa(PAGE_OFFSET) >> PAGE_SHIFT, - zholes_size); + free_area_init_nodes(max_zone_pfns); } printk("Booting Linux...\n"); @@ -1408,21 +1819,52 @@ void __init paging_init(void) cpu_probe(); } -static void __init taint_real_pages(void) +int __init page_in_phys_avail(unsigned long paddr) +{ + int i; + + paddr &= PAGE_MASK; + + for (i = 0; i < pavail_ents; i++) { + unsigned long start, end; + + start = pavail[i].phys_addr; + end = start + pavail[i].reg_size; + + if (paddr >= start && paddr < end) + return 1; + } + if (paddr >= kern_base && paddr < (kern_base + kern_size)) + return 1; +#ifdef CONFIG_BLK_DEV_INITRD + if (paddr >= __pa(initrd_start) && + paddr < __pa(PAGE_ALIGN(initrd_end))) + return 1; +#endif + + return 0; +} + +static struct linux_prom64_registers pavail_rescan[MAX_BANKS] __initdata; +static int pavail_rescan_ents __initdata; + +/* Certain OBP calls, such as fetching "available" properties, can + * claim physical memory. So, along with initializing the valid + * address bitmap, what we do here is refetch the physical available + * memory list again, and make sure it provides at least as much + * memory as 'pavail' does. + */ +static void setup_valid_addr_bitmap_from_pavail(void) { int i; read_obp_memory("available", &pavail_rescan[0], &pavail_rescan_ents); - /* Find changes discovered in the physmem available rescan and - * reserve the lost portions in the bootmem maps. - */ for (i = 0; i < pavail_ents; i++) { unsigned long old_start, old_end; old_start = pavail[i].phys_addr; - old_end = old_start + - pavail[i].reg_size; + old_end = old_start + pavail[i].reg_size; while (old_start < old_end) { int n; @@ -1440,7 +1882,16 @@ static void __init taint_real_pages(void) goto do_next_page; } } - reserve_bootmem(old_start, PAGE_SIZE, BOOTMEM_DEFAULT); + + prom_printf("mem_init: Lost memory in pavail\n"); + prom_printf("mem_init: OLD start[%lx] size[%lx]\n", + pavail[i].phys_addr, + pavail[i].reg_size); + prom_printf("mem_init: NEW start[%lx] size[%lx]\n", + pavail_rescan[i].phys_addr, + pavail_rescan[i].reg_size); + prom_printf("mem_init: Cannot continue, aborting.\n"); + prom_halt(); do_next_page: old_start += PAGE_SIZE; @@ -1448,32 +1899,6 @@ static void __init taint_real_pages(void) } } -int __init page_in_phys_avail(unsigned long paddr) -{ - int i; - - paddr &= PAGE_MASK; - - for (i = 0; i < pavail_rescan_ents; i++) { - unsigned long start, end; - - start = pavail_rescan[i].phys_addr; - end = start + pavail_rescan[i].reg_size; - - if (paddr >= start && paddr < end) - return 1; - } - if (paddr >= kern_base && paddr < (kern_base + kern_size)) - return 1; -#ifdef CONFIG_BLK_DEV_INITRD - if (paddr >= __pa(initrd_start) && - paddr < __pa(PAGE_ALIGN(initrd_end))) - return 1; -#endif - - return 0; -} - void __init mem_init(void) { unsigned long codepages, datapages, initpages; @@ -1496,14 +1921,26 @@ void __init mem_init(void) addr += PAGE_SIZE; } - taint_real_pages(); + setup_valid_addr_bitmap_from_pavail(); high_memory = __va(last_valid_pfn << PAGE_SHIFT); +#ifdef CONFIG_NEED_MULTIPLE_NODES + for_each_online_node(i) { + if (NODE_DATA(i)->node_spanned_pages != 0) { + totalram_pages += + free_all_bootmem_node(NODE_DATA(i)); + } + } +#else + totalram_pages = free_all_bootmem(); +#endif + /* We subtract one to account for the mem_map_zero page * allocated below. */ - totalram_pages = num_physpages = free_all_bootmem() - 1; + totalram_pages -= 1; + num_physpages = totalram_pages; /* * Set up the zero page, mark it reserved, so that page count diff --git a/arch/sparc64/mm/tsb.c b/arch/sparc64/mm/tsb.c index a3e6e4b635b..fe70c8a557b 100644 --- a/arch/sparc64/mm/tsb.c +++ b/arch/sparc64/mm/tsb.c @@ -321,7 +321,8 @@ retry_tsb_alloc: if (new_size > (PAGE_SIZE * 2)) gfp_flags = __GFP_NOWARN | __GFP_NORETRY; - new_tsb = kmem_cache_alloc(tsb_caches[new_cache_index], gfp_flags); + new_tsb = kmem_cache_alloc_node(tsb_caches[new_cache_index], + gfp_flags, numa_node_id()); if (unlikely(!new_tsb)) { /* Not being able to fork due to a high-order TSB * allocation failure is very bad behavior. Just back diff --git a/arch/sparc64/mm/ultra.S b/arch/sparc64/mm/ultra.S index 2865c105b6a..e686a67561a 100644 --- a/arch/sparc64/mm/ultra.S +++ b/arch/sparc64/mm/ultra.S @@ -476,7 +476,6 @@ xcall_sync_tick: #endif call smp_synchronize_tick_client nop - clr %l6 b rtrap_xcall ldx [%sp + PTREGS_OFF + PT_V9_TSTATE], %l1 @@ -511,7 +510,6 @@ xcall_report_regs: #endif call __show_regs add %sp, PTREGS_OFF, %o0 - clr %l6 /* Has to be a non-v9 branch due to the large distance. */ b rtrap_xcall ldx [%sp + PTREGS_OFF + PT_V9_TSTATE], %l1 @@ -576,7 +574,7 @@ __hypervisor_tlb_xcall_error: mov %l4, %o0 call hypervisor_tlbop_error_xcall mov %l5, %o1 - ba,a,pt %xcc, rtrap_clr_l6 + ba,a,pt %xcc, rtrap .globl __hypervisor_xcall_flush_tlb_mm __hypervisor_xcall_flush_tlb_mm: /* 21 insns */ |