diff options
Diffstat (limited to 'arch/um')
43 files changed, 465 insertions, 580 deletions
diff --git a/arch/um/Kconfig_x86_64 b/arch/um/Kconfig_x86_64 index fd8d7e8982b..f162f50f0b1 100644 --- a/arch/um/Kconfig_x86_64 +++ b/arch/um/Kconfig_x86_64 @@ -6,6 +6,10 @@ config 64BIT bool default y +config TOP_ADDR + hex + default 0x80000000 + config 3_LEVEL_PGTABLES bool default y diff --git a/arch/um/drivers/chan_kern.c b/arch/um/drivers/chan_kern.c index 0150038af79..14a12d6b3df 100644 --- a/arch/um/drivers/chan_kern.c +++ b/arch/um/drivers/chan_kern.c @@ -20,9 +20,17 @@ #include "os.h" #ifdef CONFIG_NOCONFIG_CHAN + +/* The printk's here are wrong because we are complaining that there is no + * output device, but printk is printing to that output device. The user will + * never see the error. printf would be better, except it can't run on a + * kernel stack because it will overflow it. + * Use printk for now since that will avoid crashing. + */ + static void *not_configged_init(char *str, int device, struct chan_opts *opts) { - printf(KERN_ERR "Using a channel type which is configured out of " + printk(KERN_ERR "Using a channel type which is configured out of " "UML\n"); return(NULL); } @@ -30,27 +38,27 @@ static void *not_configged_init(char *str, int device, struct chan_opts *opts) static int not_configged_open(int input, int output, int primary, void *data, char **dev_out) { - printf(KERN_ERR "Using a channel type which is configured out of " + printk(KERN_ERR "Using a channel type which is configured out of " "UML\n"); return(-ENODEV); } static void not_configged_close(int fd, void *data) { - printf(KERN_ERR "Using a channel type which is configured out of " + printk(KERN_ERR "Using a channel type which is configured out of " "UML\n"); } static int not_configged_read(int fd, char *c_out, void *data) { - printf(KERN_ERR "Using a channel type which is configured out of " + printk(KERN_ERR "Using a channel type which is configured out of " "UML\n"); return(-EIO); } static int not_configged_write(int fd, const char *buf, int len, void *data) { - printf(KERN_ERR "Using a channel type which is configured out of " + printk(KERN_ERR "Using a channel type which is configured out of " "UML\n"); return(-EIO); } @@ -58,7 +66,7 @@ static int not_configged_write(int fd, const char *buf, int len, void *data) static int not_configged_console_write(int fd, const char *buf, int len, void *data) { - printf(KERN_ERR "Using a channel type which is configured out of " + printk(KERN_ERR "Using a channel type which is configured out of " "UML\n"); return(-EIO); } @@ -66,7 +74,7 @@ static int not_configged_console_write(int fd, const char *buf, int len, static int not_configged_window_size(int fd, void *data, unsigned short *rows, unsigned short *cols) { - printf(KERN_ERR "Using a channel type which is configured out of " + printk(KERN_ERR "Using a channel type which is configured out of " "UML\n"); return(-ENODEV); } diff --git a/arch/um/drivers/mcast_kern.c b/arch/um/drivers/mcast_kern.c index faf714e87b5..217438cdef3 100644 --- a/arch/um/drivers/mcast_kern.c +++ b/arch/um/drivers/mcast_kern.c @@ -73,7 +73,6 @@ int mcast_setup(char *str, char **mac_out, void *data) struct mcast_init *init = data; char *port_str = NULL, *ttl_str = NULL, *remain; char *last; - int n; *init = ((struct mcast_init) { .addr = "239.192.168.1", @@ -89,13 +88,12 @@ int mcast_setup(char *str, char **mac_out, void *data) } if(port_str != NULL){ - n = simple_strtoul(port_str, &last, 10); + init->port = simple_strtoul(port_str, &last, 10); if((*last != '\0') || (last == port_str)){ printk(KERN_ERR "mcast_setup - Bad port : '%s'\n", port_str); return(0); } - init->port = htons(n); } if(ttl_str != NULL){ diff --git a/arch/um/drivers/mcast_user.c b/arch/um/drivers/mcast_user.c index 0fe1d9fa913..7a0d115b29d 100644 --- a/arch/um/drivers/mcast_user.c +++ b/arch/um/drivers/mcast_user.c @@ -38,7 +38,7 @@ static struct sockaddr_in *new_addr(char *addr, unsigned short port) } sin->sin_family = AF_INET; sin->sin_addr.s_addr = in_aton(addr); - sin->sin_port = port; + sin->sin_port = htons(port); return(sin); } @@ -55,28 +55,25 @@ static int mcast_open(void *data) struct mcast_data *pri = data; struct sockaddr_in *sin = pri->mcast_addr; struct ip_mreq mreq; - int fd, yes = 1; + int fd = -EINVAL, yes = 1, err = -EINVAL;; - if ((sin->sin_addr.s_addr == 0) || (sin->sin_port == 0)) { - fd = -EINVAL; + if ((sin->sin_addr.s_addr == 0) || (sin->sin_port == 0)) goto out; - } fd = socket(AF_INET, SOCK_DGRAM, 0); + if (fd < 0){ printk("mcast_open : data socket failed, errno = %d\n", errno); - fd = -ENOMEM; + fd = -errno; goto out; } if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &yes, sizeof(yes)) < 0) { printk("mcast_open: SO_REUSEADDR failed, errno = %d\n", errno); - os_close_file(fd); - fd = -EINVAL; - goto out; + goto out_close; } /* set ttl according to config */ @@ -84,26 +81,20 @@ static int mcast_open(void *data) sizeof(pri->ttl)) < 0) { printk("mcast_open: IP_MULTICAST_TTL failed, error = %d\n", errno); - os_close_file(fd); - fd = -EINVAL; - goto out; + goto out_close; } /* set LOOP, so data does get fed back to local sockets */ if (setsockopt(fd, SOL_IP, IP_MULTICAST_LOOP, &yes, sizeof(yes)) < 0) { printk("mcast_open: IP_MULTICAST_LOOP failed, error = %d\n", errno); - os_close_file(fd); - fd = -EINVAL; - goto out; + goto out_close; } /* bind socket to mcast address */ if (bind(fd, (struct sockaddr *) sin, sizeof(*sin)) < 0) { printk("mcast_open : data bind failed, errno = %d\n", errno); - os_close_file(fd); - fd = -EINVAL; - goto out; + goto out_close; } /* subscribe to the multicast group */ @@ -117,12 +108,15 @@ static int mcast_open(void *data) "interface on the host.\n"); printk("eth0 should be configured in order to use the " "multicast transport.\n"); - os_close_file(fd); - fd = -EINVAL; + goto out_close; } out: - return(fd); + return fd; + + out_close: + os_close_file(fd); + return err; } static void mcast_close(int fd, void *data) @@ -164,14 +158,3 @@ struct net_user_info mcast_user_info = { .delete_address = NULL, .max_packet = MAX_PACKET - ETH_HEADER_OTHER }; - -/* - * Overrides for Emacs so that we follow Linus's tabbing style. - * Emacs will notice this stuff at the end of the file and automatically - * adjust the settings for this buffer only. This must remain at the end - * of the file. - * --------------------------------------------------------------------------- - * Local variables: - * c-file-style: "linux" - * End: - */ diff --git a/arch/um/drivers/ubd_kern.c b/arch/um/drivers/ubd_kern.c index 9a56ff94308..88f956c34fe 100644 --- a/arch/um/drivers/ubd_kern.c +++ b/arch/um/drivers/ubd_kern.c @@ -55,7 +55,7 @@ #include "mem_kern.h" #include "cow.h" -enum ubd_req { UBD_READ, UBD_WRITE, UBD_MMAP }; +enum ubd_req { UBD_READ, UBD_WRITE }; struct io_thread_req { enum ubd_req op; @@ -68,8 +68,6 @@ struct io_thread_req { unsigned long sector_mask; unsigned long long cow_offset; unsigned long bitmap_words[2]; - int map_fd; - unsigned long long map_offset; int error; }; @@ -122,10 +120,6 @@ static int ubd_ioctl(struct inode * inode, struct file * file, #define MAX_DEV (8) -/* Changed in early boot */ -static int ubd_do_mmap = 0; -#define UBD_MMAP_BLOCK_SIZE PAGE_SIZE - static struct block_device_operations ubd_blops = { .owner = THIS_MODULE, .open = ubd_open, @@ -175,12 +169,6 @@ struct ubd { int no_cow; struct cow cow; struct platform_device pdev; - - int map_writes; - int map_reads; - int nomap_writes; - int nomap_reads; - int write_maps; }; #define DEFAULT_COW { \ @@ -200,11 +188,6 @@ struct ubd { .openflags = OPEN_FLAGS, \ .no_cow = 0, \ .cow = DEFAULT_COW, \ - .map_writes = 0, \ - .map_reads = 0, \ - .nomap_writes = 0, \ - .nomap_reads = 0, \ - .write_maps = 0, \ } struct ubd ubd_dev[MAX_DEV] = { [ 0 ... MAX_DEV - 1 ] = DEFAULT_UBD }; @@ -314,13 +297,6 @@ static int ubd_setup_common(char *str, int *index_out) int major; str++; - if(!strcmp(str, "mmap")){ - CHOOSE_MODE(printk("mmap not supported by the ubd " - "driver in tt mode\n"), - ubd_do_mmap = 1); - return(0); - } - if(!strcmp(str, "sync")){ global_openflags = of_sync(global_openflags); return(0); @@ -524,7 +500,7 @@ static void ubd_handler(void) { struct io_thread_req req; struct request *rq = elv_next_request(ubd_queue); - int n, err; + int n; do_ubd = NULL; intr_count++; @@ -538,19 +514,6 @@ static void ubd_handler(void) return; } - if((req.op != UBD_MMAP) && - ((req.offset != ((__u64) (rq->sector)) << 9) || - (req.length != (rq->current_nr_sectors) << 9))) - panic("I/O op mismatch"); - - if(req.map_fd != -1){ - err = physmem_subst_mapping(req.buffer, req.map_fd, - req.map_offset, 1); - if(err) - printk("ubd_handler - physmem_subst_mapping failed, " - "err = %d\n", -err); - } - ubd_finish(rq, req.error); reactivate_fd(thread_fd, UBD_IRQ); do_ubd_request(ubd_queue); @@ -583,14 +546,10 @@ static int ubd_file_size(struct ubd *dev, __u64 *size_out) static void ubd_close(struct ubd *dev) { - if(ubd_do_mmap) - physmem_forget_descriptor(dev->fd); os_close_file(dev->fd); if(dev->cow.file == NULL) return; - if(ubd_do_mmap) - physmem_forget_descriptor(dev->cow.fd); os_close_file(dev->cow.fd); vfree(dev->cow.bitmap); dev->cow.bitmap = NULL; @@ -1010,94 +969,13 @@ static void cowify_req(struct io_thread_req *req, unsigned long *bitmap, req->bitmap_words, bitmap_len); } -static int mmap_fd(struct request *req, struct ubd *dev, __u64 offset) -{ - __u64 sector; - unsigned char *bitmap; - int bit, i; - - /* mmap must have been requested on the command line */ - if(!ubd_do_mmap) - return(-1); - - /* The buffer must be page aligned */ - if(((unsigned long) req->buffer % UBD_MMAP_BLOCK_SIZE) != 0) - return(-1); - - /* The request must be a page long */ - if((req->current_nr_sectors << 9) != PAGE_SIZE) - return(-1); - - if(dev->cow.file == NULL) - return(dev->fd); - - sector = offset >> 9; - bitmap = (unsigned char *) dev->cow.bitmap; - bit = ubd_test_bit(sector, bitmap); - - for(i = 1; i < req->current_nr_sectors; i++){ - if(ubd_test_bit(sector + i, bitmap) != bit) - return(-1); - } - - if(bit || (rq_data_dir(req) == WRITE)) - offset += dev->cow.data_offset; - - /* The data on disk must be page aligned */ - if((offset % UBD_MMAP_BLOCK_SIZE) != 0) - return(-1); - - return(bit ? dev->fd : dev->cow.fd); -} - -static int prepare_mmap_request(struct ubd *dev, int fd, __u64 offset, - struct request *req, - struct io_thread_req *io_req) -{ - int err; - - if(rq_data_dir(req) == WRITE){ - /* Writes are almost no-ops since the new data is already in the - * host page cache - */ - dev->map_writes++; - if(dev->cow.file != NULL) - cowify_bitmap(io_req->offset, io_req->length, - &io_req->sector_mask, &io_req->cow_offset, - dev->cow.bitmap, dev->cow.bitmap_offset, - io_req->bitmap_words, - dev->cow.bitmap_len); - } - else { - int w; - - if((dev->cow.file != NULL) && (fd == dev->cow.fd)) - w = 0; - else w = dev->openflags.w; - - if((dev->cow.file != NULL) && (fd == dev->fd)) - offset += dev->cow.data_offset; - - err = physmem_subst_mapping(req->buffer, fd, offset, w); - if(err){ - printk("physmem_subst_mapping failed, err = %d\n", - -err); - return(1); - } - dev->map_reads++; - } - io_req->op = UBD_MMAP; - io_req->buffer = req->buffer; - return(0); -} - /* Called with ubd_io_lock held */ static int prepare_request(struct request *req, struct io_thread_req *io_req) { struct gendisk *disk = req->rq_disk; struct ubd *dev = disk->private_data; __u64 offset; - int len, fd; + int len; if(req->rq_status == RQ_INACTIVE) return(1); @@ -1114,34 +992,12 @@ static int prepare_request(struct request *req, struct io_thread_req *io_req) io_req->fds[0] = (dev->cow.file != NULL) ? dev->cow.fd : dev->fd; io_req->fds[1] = dev->fd; - io_req->map_fd = -1; io_req->cow_offset = -1; io_req->offset = offset; io_req->length = len; io_req->error = 0; io_req->sector_mask = 0; - fd = mmap_fd(req, dev, io_req->offset); - if(fd > 0){ - /* If mmapping is otherwise OK, but the first access to the - * page is a write, then it's not mapped in yet. So we have - * to write the data to disk first, then we can map the disk - * page in and continue normally from there. - */ - if((rq_data_dir(req) == WRITE) && !is_remapped(req->buffer)){ - io_req->map_fd = dev->fd; - io_req->map_offset = io_req->offset + - dev->cow.data_offset; - dev->write_maps++; - } - else return(prepare_mmap_request(dev, fd, io_req->offset, req, - io_req)); - } - - if(rq_data_dir(req) == READ) - dev->nomap_reads++; - else dev->nomap_writes++; - io_req->op = (rq_data_dir(req) == READ) ? UBD_READ : UBD_WRITE; io_req->offsets[0] = 0; io_req->offsets[1] = dev->cow.data_offset; @@ -1229,143 +1085,6 @@ static int ubd_ioctl(struct inode * inode, struct file * file, return(-EINVAL); } -static int ubd_check_remapped(int fd, unsigned long address, int is_write, - __u64 offset) -{ - __u64 bitmap_offset; - unsigned long new_bitmap[2]; - int i, err, n; - - /* If it's not a write access, we can't do anything about it */ - if(!is_write) - return(0); - - /* We have a write */ - for(i = 0; i < sizeof(ubd_dev) / sizeof(ubd_dev[0]); i++){ - struct ubd *dev = &ubd_dev[i]; - - if((dev->fd != fd) && (dev->cow.fd != fd)) - continue; - - /* It's a write to a ubd device */ - - /* This should be impossible now */ - if(!dev->openflags.w){ - /* It's a write access on a read-only device - probably - * shouldn't happen. If the kernel is trying to change - * something with no intention of writing it back out, - * then this message will clue us in that this needs - * fixing - */ - printk("Write access to mapped page from readonly ubd " - "device %d\n", i); - return(0); - } - - /* It's a write to a writeable ubd device - it must be COWed - * because, otherwise, the page would have been mapped in - * writeable - */ - - if(!dev->cow.file) - panic("Write fault on writeable non-COW ubd device %d", - i); - - /* It should also be an access to the backing file since the - * COW pages should be mapped in read-write - */ - - if(fd == dev->fd) - panic("Write fault on a backing page of ubd " - "device %d\n", i); - - /* So, we do the write, copying the backing data to the COW - * file... - */ - - err = os_seek_file(dev->fd, offset + dev->cow.data_offset); - if(err < 0) - panic("Couldn't seek to %lld in COW file of ubd " - "device %d, err = %d", - offset + dev->cow.data_offset, i, -err); - - n = os_write_file(dev->fd, (void *) address, PAGE_SIZE); - if(n != PAGE_SIZE) - panic("Couldn't copy data to COW file of ubd " - "device %d, err = %d", i, -n); - - /* ... updating the COW bitmap... */ - - cowify_bitmap(offset, PAGE_SIZE, NULL, &bitmap_offset, - dev->cow.bitmap, dev->cow.bitmap_offset, - new_bitmap, dev->cow.bitmap_len); - - err = os_seek_file(dev->fd, bitmap_offset); - if(err < 0) - panic("Couldn't seek to %lld in COW file of ubd " - "device %d, err = %d", bitmap_offset, i, -err); - - n = os_write_file(dev->fd, new_bitmap, sizeof(new_bitmap)); - if(n != sizeof(new_bitmap)) - panic("Couldn't update bitmap of ubd device %d, " - "err = %d", i, -n); - - /* Maybe we can map the COW page in, and maybe we can't. If - * it is a pre-V3 COW file, we can't, since the alignment will - * be wrong. If it is a V3 or later COW file which has been - * moved to a system with a larger page size, then maybe we - * can't, depending on the exact location of the page. - */ - - offset += dev->cow.data_offset; - - /* Remove the remapping, putting the original anonymous page - * back. If the COW file can be mapped in, that is done. - * Otherwise, the COW page is read in. - */ - - if(!physmem_remove_mapping((void *) address)) - panic("Address 0x%lx not remapped by ubd device %d", - address, i); - if((offset % UBD_MMAP_BLOCK_SIZE) == 0) - physmem_subst_mapping((void *) address, dev->fd, - offset, 1); - else { - err = os_seek_file(dev->fd, offset); - if(err < 0) - panic("Couldn't seek to %lld in COW file of " - "ubd device %d, err = %d", offset, i, - -err); - - n = os_read_file(dev->fd, (void *) address, PAGE_SIZE); - if(n != PAGE_SIZE) - panic("Failed to read page from offset %llx of " - "COW file of ubd device %d, err = %d", - offset, i, -n); - } - - return(1); - } - - /* It's not a write on a ubd device */ - return(0); -} - -static struct remapper ubd_remapper = { - .list = LIST_HEAD_INIT(ubd_remapper.list), - .proc = ubd_check_remapped, -}; - -static int ubd_remapper_setup(void) -{ - if(ubd_do_mmap) - register_remapper(&ubd_remapper); - - return(0); -} - -__initcall(ubd_remapper_setup); - static int same_backing_files(char *from_cmdline, char *from_cow, char *cow) { struct uml_stat buf1, buf2; @@ -1568,15 +1287,6 @@ void do_io(struct io_thread_req *req) int err; __u64 off; - if(req->op == UBD_MMAP){ - /* Touch the page to force the host to do any necessary IO to - * get it into memory - */ - n = *((volatile int *) req->buffer); - req->error = update_bitmap(req); - return; - } - nsectors = req->length / req->sectorsize; start = 0; do { diff --git a/arch/um/include/sysdep-i386/ptrace.h b/arch/um/include/sysdep-i386/ptrace.h index 84ec7ff5cf8..6eaeb991998 100644 --- a/arch/um/include/sysdep-i386/ptrace.h +++ b/arch/um/include/sysdep-i386/ptrace.h @@ -31,7 +31,6 @@ extern int sysemu_supported; #ifdef UML_CONFIG_MODE_SKAS #include "skas_ptregs.h" -#include "sysdep/faultinfo.h" #define REGS_IP(r) ((r)[HOST_IP]) #define REGS_SP(r) ((r)[HOST_SP]) @@ -59,6 +58,7 @@ extern int sysemu_supported; #define PTRACE_SYSEMU_SINGLESTEP 32 #endif +#include "sysdep/faultinfo.h" #include "choose-mode.h" union uml_pt_regs { diff --git a/arch/um/include/sysdep-x86_64/checksum.h b/arch/um/include/sysdep-x86_64/checksum.h index 572c6c19be3..ea97005af69 100644 --- a/arch/um/include/sysdep-x86_64/checksum.h +++ b/arch/um/include/sysdep-x86_64/checksum.h @@ -9,8 +9,6 @@ #include "linux/in6.h" #include "asm/uaccess.h" -extern unsigned int csum_partial_copy_from(const unsigned char *src, unsigned char *dst, int len, - int sum, int *err_ptr); extern unsigned csum_partial(const unsigned char *buff, unsigned len, unsigned sum); @@ -31,10 +29,15 @@ unsigned int csum_partial_copy_nocheck(const unsigned char *src, unsigned char * } static __inline__ -unsigned int csum_partial_copy_from_user(const unsigned char *src, unsigned char *dst, - int len, int sum, int *err_ptr) +unsigned int csum_partial_copy_from_user(const unsigned char *src, + unsigned char *dst, int len, int sum, + int *err_ptr) { - return csum_partial_copy_from(src, dst, len, sum, err_ptr); + if(copy_from_user(dst, src, len)){ + *err_ptr = -EFAULT; + return(-1); + } + return csum_partial(dst, len, sum); } /** @@ -137,15 +140,6 @@ static inline unsigned add32_with_carry(unsigned a, unsigned b) return a; } -#endif +extern unsigned short ip_compute_csum(unsigned char * buff, int len); -/* - * Overrides for Emacs so that we follow Linus's tabbing style. - * Emacs will notice this stuff at the end of the file and automatically - * adjust the settings for this buffer only. This must remain at the end - * of the file. - * --------------------------------------------------------------------------- - * Local variables: - * c-file-style: "linux" - * End: - */ +#endif diff --git a/arch/um/include/sysdep-x86_64/ptrace.h b/arch/um/include/sysdep-x86_64/ptrace.h index 348e8fcd513..be8acd5efd9 100644 --- a/arch/um/include/sysdep-x86_64/ptrace.h +++ b/arch/um/include/sysdep-x86_64/ptrace.h @@ -135,6 +135,7 @@ extern int mode_tt; __CHOOSE_MODE(SC_EFLAGS(UPT_SC(r)), REGS_EFLAGS((r)->skas.regs)) #define UPT_SC(r) ((r)->tt.sc) #define UPT_SYSCALL_NR(r) __CHOOSE_MODE((r)->tt.syscall, (r)->skas.syscall) +#define UPT_SYSCALL_RET(r) UPT_RAX(r) extern int user_context(unsigned long sp); @@ -196,32 +197,32 @@ struct syscall_args { #define UPT_SET(regs, reg, val) \ - ({ unsigned long val; \ + ({ unsigned long __upt_val = val; \ switch(reg){ \ - case R8: UPT_R8(regs) = val; break; \ - case R9: UPT_R9(regs) = val; break; \ - case R10: UPT_R10(regs) = val; break; \ - case R11: UPT_R11(regs) = val; break; \ - case R12: UPT_R12(regs) = val; break; \ - case R13: UPT_R13(regs) = val; break; \ - case R14: UPT_R14(regs) = val; break; \ - case R15: UPT_R15(regs) = val; break; \ - case RIP: UPT_IP(regs) = val; break; \ - case RSP: UPT_SP(regs) = val; break; \ - case RAX: UPT_RAX(regs) = val; break; \ - case RBX: UPT_RBX(regs) = val; break; \ - case RCX: UPT_RCX(regs) = val; break; \ - case RDX: UPT_RDX(regs) = val; break; \ - case RSI: UPT_RSI(regs) = val; break; \ - case RDI: UPT_RDI(regs) = val; break; \ - case RBP: UPT_RBP(regs) = val; break; \ - case ORIG_RAX: UPT_ORIG_RAX(regs) = val; break; \ - case CS: UPT_CS(regs) = val; break; \ - case DS: UPT_DS(regs) = val; break; \ - case ES: UPT_ES(regs) = val; break; \ - case FS: UPT_FS(regs) = val; break; \ - case GS: UPT_GS(regs) = val; break; \ - case EFLAGS: UPT_EFLAGS(regs) = val; break; \ + case R8: UPT_R8(regs) = __upt_val; break; \ + case R9: UPT_R9(regs) = __upt_val; break; \ + case R10: UPT_R10(regs) = __upt_val; break; \ + case R11: UPT_R11(regs) = __upt_val; break; \ + case R12: UPT_R12(regs) = __upt_val; break; \ + case R13: UPT_R13(regs) = __upt_val; break; \ + case R14: UPT_R14(regs) = __upt_val; break; \ + case R15: UPT_R15(regs) = __upt_val; break; \ + case RIP: UPT_IP(regs) = __upt_val; break; \ + case RSP: UPT_SP(regs) = __upt_val; break; \ + case RAX: UPT_RAX(regs) = __upt_val; break; \ + case RBX: UPT_RBX(regs) = __upt_val; break; \ + case RCX: UPT_RCX(regs) = __upt_val; break; \ + case RDX: UPT_RDX(regs) = __upt_val; break; \ + case RSI: UPT_RSI(regs) = __upt_val; break; \ + case RDI: UPT_RDI(regs) = __upt_val; break; \ + case RBP: UPT_RBP(regs) = __upt_val; break; \ + case ORIG_RAX: UPT_ORIG_RAX(regs) = __upt_val; break; \ + case CS: UPT_CS(regs) = __upt_val; break; \ + case DS: UPT_DS(regs) = __upt_val; break; \ + case ES: UPT_ES(regs) = __upt_val; break; \ + case FS: UPT_FS(regs) = __upt_val; break; \ + case GS: UPT_GS(regs) = __upt_val; break; \ + case EFLAGS: UPT_EFLAGS(regs) = __upt_val; break; \ default : \ panic("Bad register in UPT_SET : %d\n", reg); \ break; \ @@ -245,14 +246,3 @@ struct syscall_args { CHOOSE_MODE((&(r)->tt.faultinfo), (&(r)->skas.faultinfo)) #endif - -/* - * Overrides for Emacs so that we follow Linus's tabbing style. - * Emacs will notice this stuff at the end of the file and automatically - * adjust the settings for this buffer only. This must remain at the end - * of the file. - * --------------------------------------------------------------------------- - * Local variables: - * c-file-style: "linux" - * End: - */ diff --git a/arch/um/include/user_util.h b/arch/um/include/user_util.h index 103cd320386..b8c5b8a9525 100644 --- a/arch/um/include/user_util.h +++ b/arch/um/include/user_util.h @@ -67,7 +67,6 @@ extern void *um_kmalloc(int size); extern int switcheroo(int fd, int prot, void *from, void *to, int size); extern void setup_machinename(char *machine_out); extern void setup_hostinfo(void); -extern void add_arg(char *arg); extern void init_new_thread_stack(void *sig_stack, void (*usr1_handler)(int)); extern void init_new_thread_signals(int altstack); extern void do_exec(int old_pid, int new_pid); diff --git a/arch/um/kernel/Makefile b/arch/um/kernel/Makefile index dca65426385..a8918e80df9 100644 --- a/arch/um/kernel/Makefile +++ b/arch/um/kernel/Makefile @@ -14,7 +14,7 @@ obj-y = config.o exec_kern.o exitcode.o \ tlb.o trap_kern.o trap_user.o uaccess_user.o um_arch.o umid.o \ user_util.o -obj-$(CONFIG_BLK_DEV_INITRD) += initrd_kern.o initrd_user.o +obj-$(CONFIG_BLK_DEV_INITRD) += initrd.o obj-$(CONFIG_GPROF) += gprof_syms.o obj-$(CONFIG_GCOV) += gmon_syms.o obj-$(CONFIG_TTY_LOG) += tty_log.o @@ -23,14 +23,10 @@ obj-$(CONFIG_SYSCALL_DEBUG) += syscall_user.o obj-$(CONFIG_MODE_TT) += tt/ obj-$(CONFIG_MODE_SKAS) += skas/ -# This needs be compiled with frame pointers regardless of how the rest of the -# kernel is built. -CFLAGS_frame.o := -fno-omit-frame-pointer - user-objs-$(CONFIG_TTY_LOG) += tty_log.o USER_OBJS := $(user-objs-y) config.o helper.o main.o process.o tempfile.o \ - time.o tty_log.o umid.o user_util.o frame.o + time.o tty_log.o umid.o user_util.o include arch/um/scripts/Makefile.rules diff --git a/arch/um/kernel/checksum.c b/arch/um/kernel/checksum.c deleted file mode 100644 index e69de29bb2d..00000000000 --- a/arch/um/kernel/checksum.c +++ /dev/null diff --git a/arch/um/kernel/initrd.c b/arch/um/kernel/initrd.c new file mode 100644 index 00000000000..82ecf904b09 --- /dev/null +++ b/arch/um/kernel/initrd.c @@ -0,0 +1,78 @@ +/* + * Copyright (C) 2000, 2001, 2002 Jeff Dike (jdike@karaya.com) + * Licensed under the GPL + */ + +#include "linux/init.h" +#include "linux/bootmem.h" +#include "linux/initrd.h" +#include "asm/types.h" +#include "user_util.h" +#include "kern_util.h" +#include "initrd.h" +#include "init.h" +#include "os.h" + +/* Changed by uml_initrd_setup, which is a setup */ +static char *initrd __initdata = NULL; + +static int __init read_initrd(void) +{ + void *area; + long long size; + int err; + + if(initrd == NULL) return 0; + err = os_file_size(initrd, &size); + if(err) return 0; + area = alloc_bootmem(size); + if(area == NULL) return 0; + if(load_initrd(initrd, area, size) == -1) return 0; + initrd_start = (unsigned long) area; + initrd_end = initrd_start + size; + return 0; +} + +__uml_postsetup(read_initrd); + +static int __init uml_initrd_setup(char *line, int *add) +{ + initrd = line; + return 0; +} + +__uml_setup("initrd=", uml_initrd_setup, +"initrd=<initrd image>\n" +" This is used to boot UML from an initrd image. The argument is the\n" +" name of the file containing the image.\n\n" +); + +int load_initrd(char *filename, void *buf, int size) +{ + int fd, n; + + fd = os_open_file(filename, of_read(OPENFLAGS()), 0); + if(fd < 0){ + printk("Opening '%s' failed - err = %d\n", filename, -fd); + return(-1); + } + n = os_read_file(fd, buf, size); + if(n != size){ + printk("Read of %d bytes from '%s' failed, err = %d\n", size, + filename, -n); + return(-1); + } + + os_close_file(fd); + return(0); +} +/* + * Overrides for Emacs so that we follow Linus's tabbing style. + * Emacs will notice this stuff at the end of the file and automatically + * adjust the settings for this buffer only. This must remain at the end + * of the file. + * --------------------------------------------------------------------------- + * Local variables: + * c-file-style: "linux" + * End: + */ diff --git a/arch/um/kernel/irq_user.c b/arch/um/kernel/irq_user.c index 6d6f9484b88..b3074cbaa47 100644 --- a/arch/um/kernel/irq_user.c +++ b/arch/um/kernel/irq_user.c @@ -236,9 +236,15 @@ static void free_irq_by_cb(int (*test)(struct irq_fd *, void *), void *arg) (*prev)->fd, pollfds[i].fd); goto out; } - memcpy(&pollfds[i], &pollfds[i + 1], - (pollfds_num - i - 1) * sizeof(pollfds[0])); + pollfds_num--; + + /* This moves the *whole* array after pollfds[i] (though + * it doesn't spot as such)! */ + + memmove(&pollfds[i], &pollfds[i + 1], + (pollfds_num - i) * sizeof(pollfds[0])); + if(last_irq_ptr == &old_fd->next) last_irq_ptr = prev; *prev = (*prev)->next; diff --git a/arch/um/kernel/ksyms.c b/arch/um/kernel/ksyms.c index 78d69dc74b2..99439fa15ef 100644 --- a/arch/um/kernel/ksyms.c +++ b/arch/um/kernel/ksyms.c @@ -57,6 +57,7 @@ EXPORT_SYMBOL(copy_to_user_tt); EXPORT_SYMBOL(strncpy_from_user_skas); EXPORT_SYMBOL(copy_to_user_skas); EXPORT_SYMBOL(copy_from_user_skas); +EXPORT_SYMBOL(clear_user_skas); #endif EXPORT_SYMBOL(uml_strdup); diff --git a/arch/um/kernel/mem.c b/arch/um/kernel/mem.c index f156661781c..c22825f13e4 100644 --- a/arch/um/kernel/mem.c +++ b/arch/um/kernel/mem.c @@ -100,12 +100,37 @@ void mem_init(void) #endif } +/* + * Create a page table and place a pointer to it in a middle page + * directory entry. + */ +static void __init one_page_table_init(pmd_t *pmd) +{ + if (pmd_none(*pmd)) { + pte_t *pte = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); + set_pmd(pmd, __pmd(_KERNPG_TABLE + + (unsigned long) __pa(pte))); + if (pte != pte_offset_kernel(pmd, 0)) + BUG(); + } +} + +static void __init one_md_table_init(pud_t *pud) +{ +#ifdef CONFIG_3_LEVEL_PGTABLES + pmd_t *pmd_table = (pmd_t *) alloc_bootmem_low_pages(PAGE_SIZE); + set_pud(pud, __pud(_KERNPG_TABLE + (unsigned long) __pa(pmd_table))); + if (pmd_table != pmd_offset(pud, 0)) + BUG(); +#endif +} + static void __init fixrange_init(unsigned long start, unsigned long end, pgd_t *pgd_base) { pgd_t *pgd; + pud_t *pud; pmd_t *pmd; - pte_t *pte; int i, j; unsigned long vaddr; @@ -115,15 +140,12 @@ static void __init fixrange_init(unsigned long start, unsigned long end, pgd = pgd_base + i; for ( ; (i < PTRS_PER_PGD) && (vaddr < end); pgd++, i++) { - pmd = (pmd_t *)pgd; + pud = pud_offset(pgd, vaddr); + if (pud_none(*pud)) + one_md_table_init(pud); + pmd = pmd_offset(pud, vaddr); for (; (j < PTRS_PER_PMD) && (vaddr != end); pmd++, j++) { - if (pmd_none(*pmd)) { - pte = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); - set_pmd(pmd, __pmd(_KERNPG_TABLE + - (unsigned long) __pa(pte))); - if (pte != pte_offset_kernel(pmd, 0)) - BUG(); - } + one_page_table_init(pmd); vaddr += PMD_SIZE; } j = 0; diff --git a/arch/um/kernel/process_kern.c b/arch/um/kernel/process_kern.c index 7a943696f95..c1adf7ba3fd 100644 --- a/arch/um/kernel/process_kern.c +++ b/arch/um/kernel/process_kern.c @@ -115,16 +115,6 @@ int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags) return(pid); } -void switch_mm(struct mm_struct *prev, struct mm_struct *next, - struct task_struct *tsk) -{ - int cpu = smp_processor_id(); - - if (prev != next) - cpu_clear(cpu, prev->cpu_vm_mask); - cpu_set(cpu, next->cpu_vm_mask); -} - void set_current(void *t) { struct task_struct *task = t; @@ -152,7 +142,6 @@ void release_thread(struct task_struct *task) void exit_thread(void) { - CHOOSE_MODE(exit_thread_tt(), exit_thread_skas()); unprotect_stack((unsigned long) current_thread); } @@ -476,12 +465,21 @@ int singlestepping(void * t) return 2; } +/* + * Only x86 and x86_64 have an arch_align_stack(). + * All other arches have "#define arch_align_stack(x) (x)" + * in their asm/system.h + * As this is included in UML from asm-um/system-generic.h, + * we can use it to behave as the subarch does. + */ +#ifndef arch_align_stack unsigned long arch_align_stack(unsigned long sp) { if (randomize_va_space) sp -= get_random_int() % 8192; return sp & ~0xf; } +#endif /* diff --git a/arch/um/kernel/ptrace.c b/arch/um/kernel/ptrace.c index 3efa59a941a..2925e15324d 100644 --- a/arch/um/kernel/ptrace.c +++ b/arch/um/kernel/ptrace.c @@ -19,15 +19,30 @@ #include "skas_ptrace.h" #include "sysdep/ptrace.h" +static inline void set_singlestepping(struct task_struct *child, int on) +{ + if (on) + child->ptrace |= PT_DTRACE; + else + child->ptrace &= ~PT_DTRACE; + child->thread.singlestep_syscall = 0; + +#ifdef SUBARCH_SET_SINGLESTEPPING + SUBARCH_SET_SINGLESTEPPING(child, on); +#endif +} + /* * Called by kernel/ptrace.c when detaching.. */ void ptrace_disable(struct task_struct *child) { - child->ptrace &= ~PT_DTRACE; - child->thread.singlestep_syscall = 0; + set_singlestepping(child,0); } +extern int peek_user(struct task_struct * child, long addr, long data); +extern int poke_user(struct task_struct * child, long addr, long data); + long sys_ptrace(long request, long pid, long addr, long data) { struct task_struct *child; @@ -67,6 +82,10 @@ long sys_ptrace(long request, long pid, long addr, long data) goto out_tsk; } +#ifdef SUBACH_PTRACE_SPECIAL + SUBARCH_PTRACE_SPECIAL(child,request,addr,data); +#endif + ret = ptrace_check_attach(child, request == PTRACE_KILL); if (ret < 0) goto out_tsk; @@ -87,28 +106,9 @@ long sys_ptrace(long request, long pid, long addr, long data) } /* read the word at location addr in the USER area. */ - case PTRACE_PEEKUSR: { - unsigned long tmp; - - ret = -EIO; - if ((addr & 3) || addr < 0) - break; - - tmp = 0; /* Default return condition */ - if(addr < MAX_REG_OFFSET){ - tmp = getreg(child, addr); - } -#if defined(CONFIG_UML_X86) && !defined(CONFIG_64BIT) - else if((addr >= offsetof(struct user, u_debugreg[0])) && - (addr <= offsetof(struct user, u_debugreg[7]))){ - addr -= offsetof(struct user, u_debugreg[0]); - addr = addr >> 2; - tmp = child->thread.arch.debugregs[addr]; - } -#endif - ret = put_user(tmp, (unsigned long __user *) data); - break; - } + case PTRACE_PEEKUSR: + ret = peek_user(child, addr, data); + break; /* when I and D space are separate, this will have to be fixed. */ case PTRACE_POKETEXT: /* write the word at location addr. */ @@ -121,26 +121,8 @@ long sys_ptrace(long request, long pid, long addr, long data) break; case PTRACE_POKEUSR: /* write the word at location addr in the USER area */ - ret = -EIO; - if ((addr & 3) || addr < 0) - break; - - if (addr < MAX_REG_OFFSET) { - ret = putreg(child, addr, data); - break; - } -#if defined(CONFIG_UML_X86) && !defined(CONFIG_64BIT) - else if((addr >= offsetof(struct user, u_debugreg[0])) && - (addr <= offsetof(struct user, u_debugreg[7]))){ - addr -= offsetof(struct user, u_debugreg[0]); - addr = addr >> 2; - if((addr == 4) || (addr == 5)) break; - child->thread.arch.debugregs[addr] = data; - ret = 0; - } -#endif - - break; + ret = poke_user(child, addr, data); + break; case PTRACE_SYSCALL: /* continue and stop at next (return from) syscall */ case PTRACE_CONT: { /* restart after signal. */ @@ -148,8 +130,7 @@ long sys_ptrace(long request, long pid, long addr, long data) if (!valid_signal(data)) break; - child->ptrace &= ~PT_DTRACE; - child->thread.singlestep_syscall = 0; + set_singlestepping(child, 0); if (request == PTRACE_SYSCALL) { set_tsk_thread_flag(child, TIF_SYSCALL_TRACE); } @@ -172,8 +153,7 @@ long sys_ptrace(long request, long pid, long addr, long data) if (child->exit_state == EXIT_ZOMBIE) /* already dead */ break; - child->ptrace &= ~PT_DTRACE; - child->thread.singlestep_syscall = 0; + set_singlestepping(child, 0); child->exit_code = SIGKILL; wake_up_process(child); break; @@ -184,8 +164,7 @@ long sys_ptrace(long request, long pid, long addr, long data) if (!valid_signal(data)) break; clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE); - child->ptrace |= PT_DTRACE; - child->thread.singlestep_syscall = 0; + set_singlestepping(child, 1); child->exit_code = data; /* give it a chance to run. */ wake_up_process(child); diff --git a/arch/um/kernel/skas/include/mode_kern-skas.h b/arch/um/kernel/skas/include/mode_kern-skas.h index 94c56496237..e4849002811 100644 --- a/arch/um/kernel/skas/include/mode_kern-skas.h +++ b/arch/um/kernel/skas/include/mode_kern-skas.h @@ -18,7 +18,6 @@ extern int copy_thread_skas(int nr, unsigned long clone_flags, unsigned long sp, unsigned long stack_top, struct task_struct *p, struct pt_regs *regs); extern void release_thread_skas(struct task_struct *task); -extern void exit_thread_skas(void); extern void initial_thread_cb_skas(void (*proc)(void *), void *arg); extern void init_idle_skas(void); extern void flush_tlb_kernel_range_skas(unsigned long start, diff --git a/arch/um/kernel/skas/include/uaccess-skas.h b/arch/um/kernel/skas/include/uaccess-skas.h index c35620385da..cd6c280482c 100644 --- a/arch/um/kernel/skas/include/uaccess-skas.h +++ b/arch/um/kernel/skas/include/uaccess-skas.h @@ -18,8 +18,8 @@ ((unsigned long) (addr) + (size) <= FIXADDR_USER_END) && \ ((unsigned long) (addr) + (size) >= (unsigned long)(addr)))) -static inline int __deprecated verify_area_skas(int type, const void * addr, - unsigned long size) +static inline int verify_area_skas(int type, const void * addr, + unsigned long size) { return(access_ok_skas(type, addr, size) ? 0 : -EFAULT); } diff --git a/arch/um/kernel/skas/process.c b/arch/um/kernel/skas/process.c index 4dc13bc8cfd..773cd2b525f 100644 --- a/arch/um/kernel/skas/process.c +++ b/arch/um/kernel/skas/process.c @@ -201,6 +201,11 @@ void userspace(union uml_pt_regs *regs) } } } +#define INIT_JMP_NEW_THREAD 0 +#define INIT_JMP_REMOVE_SIGSTACK 1 +#define INIT_JMP_CALLBACK 2 +#define INIT_JMP_HALT 3 +#define INIT_JMP_REBOOT 4 void new_thread(void *stack, void **switch_buf_ptr, void **fork_buf_ptr, void (*handler)(int)) @@ -236,7 +241,7 @@ void thread_wait(void *sw, void *fb) *switch_buf = &buf; fork_buf = fb; if(sigsetjmp(buf, 1) == 0) - siglongjmp(*fork_buf, 1); + siglongjmp(*fork_buf, INIT_JMP_REMOVE_SIGSTACK); } void switch_threads(void *me, void *next) @@ -266,21 +271,25 @@ int start_idle_thread(void *stack, void *switch_buf_ptr, void **fork_buf_ptr) *fork_buf_ptr = &initial_jmpbuf; n = sigsetjmp(initial_jmpbuf, 1); - if(n == 0) - new_thread_proc((void *) stack, new_thread_handler); - else if(n == 1) - remove_sigstack(); - else if(n == 2){ + switch(n){ + case INIT_JMP_NEW_THREAD: + new_thread_proc((void *) stack, new_thread_handler); + break; + case INIT_JMP_REMOVE_SIGSTACK: + remove_sigstack(); + break; + case INIT_JMP_CALLBACK: (*cb_proc)(cb_arg); siglongjmp(*cb_back, 1); - } - else if(n == 3){ + break; + case INIT_JMP_HALT: kmalloc_ok = 0; return(0); - } - else if(n == 4){ + case INIT_JMP_REBOOT: kmalloc_ok = 0; return(1); + default: + panic("Bad sigsetjmp return in start_idle_thread - %d\n", n); } siglongjmp(**switch_buf, 1); } @@ -305,7 +314,7 @@ void initial_thread_cb_skas(void (*proc)(void *), void *arg) block_signals(); if(sigsetjmp(here, 1) == 0) - siglongjmp(initial_jmpbuf, 2); + siglongjmp(initial_jmpbuf, INIT_JMP_CALLBACK); unblock_signals(); cb_proc = NULL; @@ -316,13 +325,13 @@ void initial_thread_cb_skas(void (*proc)(void *), void *arg) void halt_skas(void) { block_signals(); - siglongjmp(initial_jmpbuf, 3); + siglongjmp(initial_jmpbuf, INIT_JMP_HALT); } void reboot_skas(void) { block_signals(); - siglongjmp(initial_jmpbuf, 4); + siglongjmp(initial_jmpbuf, INIT_JMP_REBOOT); } void switch_mm_skas(int mm_fd) diff --git a/arch/um/kernel/skas/process_kern.c b/arch/um/kernel/skas/process_kern.c index 5d096ea63b9..ab5d3271da0 100644 --- a/arch/um/kernel/skas/process_kern.c +++ b/arch/um/kernel/skas/process_kern.c @@ -83,10 +83,6 @@ void release_thread_skas(struct task_struct *task) { } -void exit_thread_skas(void) -{ -} - void fork_handler(int sig) { change_sig(SIGUSR1, 1); diff --git a/arch/um/kernel/skas/uaccess.c b/arch/um/kernel/skas/uaccess.c index f7da9d02767..75195281081 100644 --- a/arch/um/kernel/skas/uaccess.c +++ b/arch/um/kernel/skas/uaccess.c @@ -29,9 +29,12 @@ static unsigned long maybe_map(unsigned long virt, int is_write) if(IS_ERR(phys) || (is_write && !pte_write(pte))){ err = handle_page_fault(virt, 0, is_write, 1, &dummy_code); if(err) - return(0); + return(-1UL); phys = um_virt_to_phys(current, virt, NULL); } + if(IS_ERR(phys)) + phys = (void *) -1; + return((unsigned long) phys); } @@ -42,7 +45,7 @@ static int do_op(unsigned long addr, int len, int is_write, int n; addr = maybe_map(addr, is_write); - if(addr == -1) + if(addr == -1UL) return(-1); page = phys_to_page(addr); diff --git a/arch/um/kernel/syscall_kern.c b/arch/um/kernel/syscall_kern.c index 42731e04f50..b7a55251e89 100644 --- a/arch/um/kernel/syscall_kern.c +++ b/arch/um/kernel/syscall_kern.c @@ -17,7 +17,6 @@ #include "linux/utime.h" #include "asm/mman.h" #include "asm/uaccess.h" -#include "asm/ipc.h" #include "kern_util.h" #include "user_util.h" #include "sysdep/syscalls.h" diff --git a/arch/um/kernel/trap_kern.c b/arch/um/kernel/trap_kern.c index 54e2ec33a43..1de22d8a313 100644 --- a/arch/um/kernel/trap_kern.c +++ b/arch/um/kernel/trap_kern.c @@ -48,7 +48,7 @@ int handle_page_fault(unsigned long address, unsigned long ip, goto good_area; else if(!(vma->vm_flags & VM_GROWSDOWN)) goto out; - else if(!ARCH_IS_STACKGROW(address)) + else if(is_user && !ARCH_IS_STACKGROW(address)) goto out; else if(expand_stack(vma, address)) goto out; @@ -57,10 +57,11 @@ int handle_page_fault(unsigned long address, unsigned long ip, *code_out = SEGV_ACCERR; if(is_write && !(vma->vm_flags & VM_WRITE)) goto out; + + if(!(vma->vm_flags & (VM_READ | VM_EXEC))) + goto out; + page = address & PAGE_MASK; - pgd = pgd_offset(mm, page); - pud = pud_offset(pgd, page); - pmd = pmd_offset(pud, page); do { survive: switch (handle_mm_fault(mm, vma, address, is_write)){ @@ -106,33 +107,6 @@ out_of_memory: goto out; } -LIST_HEAD(physmem_remappers); - -void register_remapper(struct remapper *info) -{ - list_add(&info->list, &physmem_remappers); -} - -static int check_remapped_addr(unsigned long address, int is_write) -{ - struct remapper *remapper; - struct list_head *ele; - __u64 offset; - int fd; - - fd = phys_mapping(__pa(address), &offset); - if(fd == -1) - return(0); - - list_for_each(ele, &physmem_remappers){ - remapper = list_entry(ele, struct remapper, list); - if((*remapper->proc)(fd, address, is_write, offset)) - return(1); - } - - return(0); -} - /* * We give a *copy* of the faultinfo in the regs to segv. * This must be done, since nesting SEGVs could overwrite @@ -151,8 +125,6 @@ unsigned long segv(struct faultinfo fi, unsigned long ip, int is_user, void *sc) flush_tlb_kernel_vm(); return(0); } - else if(check_remapped_addr(address & PAGE_MASK, is_write)) - return(0); else if(current->mm == NULL) panic("Segfault with no mm"); err = handle_page_fault(address, ip, is_write, is_user, &si.si_code); diff --git a/arch/um/kernel/tt/include/mode_kern-tt.h b/arch/um/kernel/tt/include/mode_kern-tt.h index 28aaab3448f..e0ca0e0b251 100644 --- a/arch/um/kernel/tt/include/mode_kern-tt.h +++ b/arch/um/kernel/tt/include/mode_kern-tt.h @@ -19,7 +19,6 @@ extern int copy_thread_tt(int nr, unsigned long clone_flags, unsigned long sp, unsigned long stack_top, struct task_struct *p, struct pt_regs *regs); extern void release_thread_tt(struct task_struct *task); -extern void exit_thread_tt(void); extern void initial_thread_cb_tt(void (*proc)(void *), void *arg); extern void init_idle_tt(void); extern void flush_tlb_kernel_range_tt(unsigned long start, unsigned long end); diff --git a/arch/um/kernel/tt/include/uaccess-tt.h b/arch/um/kernel/tt/include/uaccess-tt.h index bb69d6b7d02..3fbb5fe26f4 100644 --- a/arch/um/kernel/tt/include/uaccess-tt.h +++ b/arch/um/kernel/tt/include/uaccess-tt.h @@ -33,8 +33,8 @@ extern unsigned long uml_physmem; (((unsigned long) (addr) <= ((unsigned long) (addr) + (size))) && \ (under_task_size(addr, size) || is_stack(addr, size)))) -static inline int __deprecated verify_area_tt(int type, const void * addr, - unsigned long size) +static inline int verify_area_tt(int type, const void * addr, + unsigned long size) { return(access_ok_tt(type, addr, size) ? 0 : -EFAULT); } diff --git a/arch/um/kernel/tt/ksyms.c b/arch/um/kernel/tt/ksyms.c index 92ec85d67c7..84a9385a8fe 100644 --- a/arch/um/kernel/tt/ksyms.c +++ b/arch/um/kernel/tt/ksyms.c @@ -12,6 +12,7 @@ EXPORT_SYMBOL(__do_copy_to_user); EXPORT_SYMBOL(__do_strncpy_from_user); EXPORT_SYMBOL(__do_strnlen_user); EXPORT_SYMBOL(__do_clear_user); +EXPORT_SYMBOL(clear_user_tt); EXPORT_SYMBOL(tracing_pid); EXPORT_SYMBOL(honeypot); diff --git a/arch/um/kernel/tt/mem.c b/arch/um/kernel/tt/mem.c index 74346a04a2b..bcb8796c3cb 100644 --- a/arch/um/kernel/tt/mem.c +++ b/arch/um/kernel/tt/mem.c @@ -21,14 +21,8 @@ void before_mem_tt(unsigned long brk_start) remap_data(UML_ROUND_DOWN(&__bss_start), UML_ROUND_UP(&_end), 1); } -#ifdef CONFIG_HOST_2G_2G -#define TOP 0x80000000 -#else -#define TOP 0xc0000000 -#endif - #define SIZE ((CONFIG_NEST_LEVEL + CONFIG_KERNEL_HALF_GIGS) * 0x20000000) -#define START (TOP - SIZE) +#define START (CONFIG_TOP_ADDR - SIZE) unsigned long set_task_sizes_tt(int arg, unsigned long *host_size_out, unsigned long *task_size_out) diff --git a/arch/um/kernel/tt/process_kern.c b/arch/um/kernel/tt/process_kern.c index f19f7c18feb..df810ca8fc1 100644 --- a/arch/um/kernel/tt/process_kern.c +++ b/arch/um/kernel/tt/process_kern.c @@ -65,8 +65,7 @@ void *switch_to_tt(void *prev, void *next, void *last) panic("write of switch_pipe failed, err = %d", -err); reading = 1; - if((from->exit_state == EXIT_ZOMBIE) || - (from->exit_state == EXIT_DEAD)) + if(from->thread.mode.tt.switch_pipe[0] == -1) os_kill_process(os_getpid(), 0); err = os_read_file(from->thread.mode.tt.switch_pipe[0], &c, sizeof(c)); @@ -81,8 +80,7 @@ void *switch_to_tt(void *prev, void *next, void *last) * in case it has not already killed itself. */ prev_sched = current->thread.prev_sched; - if((prev_sched->exit_state == EXIT_ZOMBIE) || - (prev_sched->exit_state == EXIT_DEAD)) + if(prev_sched->thread.mode.tt.switch_pipe[0] == -1) os_kill_process(prev_sched->thread.mode.tt.extern_pid, 1); change_sig(SIGVTALRM, vtalrm); @@ -101,14 +99,18 @@ void release_thread_tt(struct task_struct *task) { int pid = task->thread.mode.tt.extern_pid; + /* + * We first have to kill the other process, before + * closing its switch_pipe. Else it might wake up + * and receive "EOF" before we could kill it. + */ if(os_getpid() != pid) os_kill_process(pid, 0); -} -void exit_thread_tt(void) -{ - os_close_file(current->thread.mode.tt.switch_pipe[0]); - os_close_file(current->thread.mode.tt.switch_pipe[1]); + os_close_file(task->thread.mode.tt.switch_pipe[0]); + os_close_file(task->thread.mode.tt.switch_pipe[1]); + /* use switch_pipe as flag: thread is released */ + task->thread.mode.tt.switch_pipe[0] = -1; } void suspend_new_thread(int fd) diff --git a/arch/um/kernel/tt/syscall_user.c b/arch/um/kernel/tt/syscall_user.c index e4e7e9c2224..b218316cfdb 100644 --- a/arch/um/kernel/tt/syscall_user.c +++ b/arch/um/kernel/tt/syscall_user.c @@ -63,6 +63,10 @@ void do_syscall(void *task, int pid, int local_using_sysemu) UPT_SYSCALL_NR(TASK_REGS(task)) = PT_SYSCALL_NR(proc_regs); +#ifdef UPT_ORIGGPR2 + UPT_ORIGGPR2(TASK_REGS(task)) = REGS_ORIGGPR2(proc_regs); +#endif + if(((unsigned long *) PT_IP(proc_regs) >= &_stext) && ((unsigned long *) PT_IP(proc_regs) <= &_etext)) tracer_panic("I'm tracing myself and I can't get out"); diff --git a/arch/um/kernel/um_arch.c b/arch/um/kernel/um_arch.c index 5c49d88eed3..4d10ec372a6 100644 --- a/arch/um/kernel/um_arch.c +++ b/arch/um/kernel/um_arch.c @@ -23,6 +23,7 @@ #include "asm/ptrace.h" #include "asm/elf.h" #include "asm/user.h" +#include "asm/setup.h" #include "ubd_user.h" #include "asm/current.h" #include "asm/setup.h" @@ -42,9 +43,9 @@ #define DEFAULT_COMMAND_LINE "root=98:0" /* Changed in linux_main and setup_arch, which run before SMP is started */ -char command_line[COMMAND_LINE_SIZE] = { 0 }; +static char command_line[COMMAND_LINE_SIZE] = { 0 }; -void add_arg(char *arg) +static void add_arg(char *arg) { if (strlen(command_line) + strlen(arg) + 1 > COMMAND_LINE_SIZE) { printf("add_arg: Too many command line arguments!\n"); @@ -449,7 +450,7 @@ void __init setup_arch(char **cmdline_p) { notifier_chain_register(&panic_notifier_list, &panic_exit_notifier); paging_init(); - strlcpy(saved_command_line, command_line, COMMAND_LINE_SIZE); + strlcpy(saved_command_line, command_line, COMMAND_LINE_SIZE); *cmdline_p = command_line; setup_hostinfo(); } diff --git a/arch/um/kernel/uml.lds.S b/arch/um/kernel/uml.lds.S index 76eadb30918..dd5355500bd 100644 --- a/arch/um/kernel/uml.lds.S +++ b/arch/um/kernel/uml.lds.S @@ -73,6 +73,8 @@ SECTIONS .got : { *(.got.plt) *(.got) } .dynamic : { *(.dynamic) } + .tdata : { *(.tdata .tdata.* .gnu.linkonce.td.*) } + .tbss : { *(.tbss .tbss.* .gnu.linkonce.tb.*) *(.tcommon) } /* We want the small data sections together, so single-instruction offsets can access them all, and initialized data all before uninitialized, so we can shorten the on-disk segment size. */ diff --git a/arch/um/sys-i386/Makefile b/arch/um/sys-i386/Makefile index fcd67c3414e..4351e560550 100644 --- a/arch/um/sys-i386/Makefile +++ b/arch/um/sys-i386/Makefile @@ -9,11 +9,11 @@ USER_OBJS := bugs.o ptrace_user.o sigcontext.o fault.o SYMLINKS = bitops.c semaphore.c highmem.c module.c +include arch/um/scripts/Makefile.rules + bitops.c-dir = lib semaphore.c-dir = kernel highmem.c-dir = mm module.c-dir = kernel subdir- := util - -include arch/um/scripts/Makefile.rules diff --git a/arch/um/sys-i386/delay.c b/arch/um/sys-i386/delay.c index e9892eef51c..2c11b9770e8 100644 --- a/arch/um/sys-i386/delay.c +++ b/arch/um/sys-i386/delay.c @@ -1,5 +1,7 @@ -#include "linux/delay.h" -#include "asm/param.h" +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/delay.h> +#include <asm/param.h> void __delay(unsigned long time) { @@ -20,13 +22,19 @@ void __udelay(unsigned long usecs) int i, n; n = (loops_per_jiffy * HZ * usecs) / MILLION; - for(i=0;i<n;i++) ; + for(i=0;i<n;i++) + cpu_relax(); } +EXPORT_SYMBOL(__udelay); + void __const_udelay(unsigned long usecs) { int i, n; n = (loops_per_jiffy * HZ * usecs) / MILLION; - for(i=0;i<n;i++) ; + for(i=0;i<n;i++) + cpu_relax(); } + +EXPORT_SYMBOL(__const_udelay); diff --git a/arch/um/sys-i386/ptrace.c b/arch/um/sys-i386/ptrace.c index e470d28cdf8..e839ce65ad2 100644 --- a/arch/um/sys-i386/ptrace.c +++ b/arch/um/sys-i386/ptrace.c @@ -73,6 +73,25 @@ int putreg(struct task_struct *child, int regno, unsigned long value) return 0; } +int poke_user(struct task_struct *child, long addr, long data) +{ + if ((addr & 3) || addr < 0) + return -EIO; + + if (addr < MAX_REG_OFFSET) + return putreg(child, addr, data); + + else if((addr >= offsetof(struct user, u_debugreg[0])) && + (addr <= offsetof(struct user, u_debugreg[7]))){ + addr -= offsetof(struct user, u_debugreg[0]); + addr = addr >> 2; + if((addr == 4) || (addr == 5)) return -EIO; + child->thread.arch.debugregs[addr] = data; + return 0; + } + return -EIO; +} + unsigned long getreg(struct task_struct *child, int regno) { unsigned long retval = ~0UL; @@ -93,6 +112,27 @@ unsigned long getreg(struct task_struct *child, int regno) return retval; } +int peek_user(struct task_struct *child, long addr, long data) +{ +/* read the word at location addr in the USER area. */ + unsigned long tmp; + + if ((addr & 3) || addr < 0) + return -EIO; + + tmp = 0; /* Default return condition */ + if(addr < MAX_REG_OFFSET){ + tmp = getreg(child, addr); + } + else if((addr >= offsetof(struct user, u_debugreg[0])) && + (addr <= offsetof(struct user, u_debugreg[7]))){ + addr -= offsetof(struct user, u_debugreg[0]); + addr = addr >> 2; + tmp = child->thread.arch.debugregs[addr]; + } + return put_user(tmp, (unsigned long *) data); +} + struct i387_fxsave_struct { unsigned short cwd; unsigned short swd; diff --git a/arch/um/sys-ppc/ptrace.c b/arch/um/sys-ppc/ptrace.c index a971366d327..8e71b47f2b8 100644 --- a/arch/um/sys-ppc/ptrace.c +++ b/arch/um/sys-ppc/ptrace.c @@ -8,6 +8,25 @@ int putreg(struct task_struct *child, unsigned long regno, return 0; } +int poke_user(struct task_struct *child, long addr, long data) +{ + if ((addr & 3) || addr < 0) + return -EIO; + + if (addr < MAX_REG_OFFSET) + return putreg(child, addr, data); + + else if((addr >= offsetof(struct user, u_debugreg[0])) && + (addr <= offsetof(struct user, u_debugreg[7]))){ + addr -= offsetof(struct user, u_debugreg[0]); + addr = addr >> 2; + if((addr == 4) || (addr == 5)) return -EIO; + child->thread.arch.debugregs[addr] = data; + return 0; + } + return -EIO; +} + unsigned long getreg(struct task_struct *child, unsigned long regno) { unsigned long retval = ~0UL; @@ -16,6 +35,27 @@ unsigned long getreg(struct task_struct *child, unsigned long regno) return retval; } +int peek_user(struct task_struct *child, long addr, long data) +{ + /* read the word at location addr in the USER area. */ + unsigned long tmp; + + if ((addr & 3) || addr < 0) + return -EIO; + + tmp = 0; /* Default return condition */ + if(addr < MAX_REG_OFFSET){ + tmp = getreg(child, addr); + } + else if((addr >= offsetof(struct user, u_debugreg[0])) && + (addr <= offsetof(struct user, u_debugreg[7]))){ + addr -= offsetof(struct user, u_debugreg[0]); + addr = addr >> 2; + tmp = child->thread.arch.debugregs[addr]; + } + return put_user(tmp, (unsigned long *) data); +} + /* * Overrides for Emacs so that we follow Linus's tabbing style. * Emacs will notice this stuff at the end of the file and automatically diff --git a/arch/um/sys-x86_64/Makefile b/arch/um/sys-x86_64/Makefile index 3d7da911cc8..608466ad6b2 100644 --- a/arch/um/sys-x86_64/Makefile +++ b/arch/um/sys-x86_64/Makefile @@ -14,11 +14,11 @@ obj-$(CONFIG_MODULES) += module.o um_module.o USER_OBJS := ptrace_user.o sigcontext.o -include arch/um/scripts/Makefile.rules - SYMLINKS = bitops.c csum-copy.S csum-partial.c csum-wrappers.c memcpy.S \ semaphore.c thunk.S module.c +include arch/um/scripts/Makefile.rules + bitops.c-dir = lib csum-copy.S-dir = lib csum-partial.c-dir = lib @@ -28,6 +28,4 @@ semaphore.c-dir = kernel thunk.S-dir = lib module.c-dir = kernel -CFLAGS_csum-partial.o := -Dcsum_partial=arch_csum_partial - subdir- := util diff --git a/arch/um/sys-x86_64/delay.c b/arch/um/sys-x86_64/delay.c index 651332aeec2..137f4446b43 100644 --- a/arch/um/sys-x86_64/delay.c +++ b/arch/um/sys-x86_64/delay.c @@ -5,40 +5,37 @@ * Licensed under the GPL */ -#include "linux/delay.h" -#include "asm/processor.h" -#include "asm/param.h" +#include <linux/module.h> +#include <linux/delay.h> +#include <asm/processor.h> +#include <asm/param.h> void __delay(unsigned long loops) { unsigned long i; - for(i = 0; i < loops; i++) ; + for(i = 0; i < loops; i++) + cpu_relax(); } void __udelay(unsigned long usecs) { - int i, n; + unsigned long i, n; n = (loops_per_jiffy * HZ * usecs) / MILLION; - for(i=0;i<n;i++) ; + for(i=0;i<n;i++) + cpu_relax(); } +EXPORT_SYMBOL(__udelay); + void __const_udelay(unsigned long usecs) { - int i, n; + unsigned long i, n; n = (loops_per_jiffy * HZ * usecs) / MILLION; - for(i=0;i<n;i++) ; + for(i=0;i<n;i++) + cpu_relax(); } -/* - * Overrides for Emacs so that we follow Linus's tabbing style. - * Emacs will notice this stuff at the end of the file and automatically - * adjust the settings for this buffer only. This must remain at the end - * of the file. - * --------------------------------------------------------------------------- - * Local variables: - * c-file-style: "linux" - * End: - */ +EXPORT_SYMBOL(__const_udelay); diff --git a/arch/um/sys-x86_64/ksyms.c b/arch/um/sys-x86_64/ksyms.c index a27f0ee6a4f..85927380820 100644 --- a/arch/um/sys-x86_64/ksyms.c +++ b/arch/um/sys-x86_64/ksyms.c @@ -16,5 +16,4 @@ EXPORT_SYMBOL(__up_wakeup); EXPORT_SYMBOL(__memcpy); /* Networking helper routines. */ -/*EXPORT_SYMBOL(csum_partial_copy_from); -EXPORT_SYMBOL(csum_partial_copy_to);*/ +EXPORT_SYMBOL(ip_compute_csum); diff --git a/arch/um/sys-x86_64/ptrace.c b/arch/um/sys-x86_64/ptrace.c index 8c146b2a1e0..74eee5c7c6d 100644 --- a/arch/um/sys-x86_64/ptrace.c +++ b/arch/um/sys-x86_64/ptrace.c @@ -5,10 +5,11 @@ */ #define __FRAME_OFFSETS -#include "asm/ptrace.h" -#include "linux/sched.h" -#include "linux/errno.h" -#include "asm/elf.h" +#include <asm/ptrace.h> +#include <linux/sched.h> +#include <linux/errno.h> +#include <asm/uaccess.h> +#include <asm/elf.h> /* XXX x86_64 */ unsigned long not_ss; @@ -62,6 +63,27 @@ int putreg(struct task_struct *child, int regno, unsigned long value) return 0; } +int poke_user(struct task_struct *child, long addr, long data) +{ + if ((addr & 3) || addr < 0) + return -EIO; + + if (addr < MAX_REG_OFFSET) + return putreg(child, addr, data); + +#if 0 /* Need x86_64 debugregs handling */ + else if((addr >= offsetof(struct user, u_debugreg[0])) && + (addr <= offsetof(struct user, u_debugreg[7]))){ + addr -= offsetof(struct user, u_debugreg[0]); + addr = addr >> 2; + if((addr == 4) || (addr == 5)) return -EIO; + child->thread.arch.debugregs[addr] = data; + return 0; + } +#endif + return -EIO; +} + unsigned long getreg(struct task_struct *child, int regno) { unsigned long retval = ~0UL; @@ -84,6 +106,29 @@ unsigned long getreg(struct task_struct *child, int regno) return retval; } +int peek_user(struct task_struct *child, long addr, long data) +{ + /* read the word at location addr in the USER area. */ + unsigned long tmp; + + if ((addr & 3) || addr < 0) + return -EIO; + + tmp = 0; /* Default return condition */ + if(addr < MAX_REG_OFFSET){ + tmp = getreg(child, addr); + } +#if 0 /* Need x86_64 debugregs handling */ + else if((addr >= offsetof(struct user, u_debugreg[0])) && + (addr <= offsetof(struct user, u_debugreg[7]))){ + addr -= offsetof(struct user, u_debugreg[0]); + addr = addr >> 2; + tmp = child->thread.arch.debugregs[addr]; + } +#endif + return put_user(tmp, (unsigned long *) data); +} + void arch_switch(void) { /* XXX diff --git a/arch/um/sys-x86_64/signal.c b/arch/um/sys-x86_64/signal.c index b740177066a..73a7926f737 100644 --- a/arch/um/sys-x86_64/signal.c +++ b/arch/um/sys-x86_64/signal.c @@ -168,7 +168,7 @@ int setup_signal_stack_si(unsigned long stack_top, int sig, frame = (struct rt_sigframe __user *) round_down(stack_top - sizeof(struct rt_sigframe), 16) - 8; - frame -= 128; + ((unsigned char *) frame) -= 128; if (!access_ok(VERIFY_WRITE, fp, sizeof(struct _fpstate))) goto out; diff --git a/arch/um/sys-x86_64/syscalls.c b/arch/um/sys-x86_64/syscalls.c index 2a575ef52bb..d4a59657fb9 100644 --- a/arch/um/sys-x86_64/syscalls.c +++ b/arch/um/sys-x86_64/syscalls.c @@ -15,6 +15,7 @@ #include "asm/unistd.h" #include "asm/prctl.h" /* XXX This should get the constants from libc */ #include "choose-mode.h" +#include "kern.h" asmlinkage long sys_uname64(struct new_utsname __user * name) { @@ -44,6 +45,8 @@ long sys_modify_ldt_tt(int func, void *ptr, unsigned long bytecount) #ifdef CONFIG_MODE_SKAS extern int userspace_pid[]; +#include "skas_ptrace.h" + long sys_modify_ldt_skas(int func, void *ptr, unsigned long bytecount) { struct ptrace_ldt ldt; diff --git a/arch/um/sys-x86_64/user-offsets.c b/arch/um/sys-x86_64/user-offsets.c index 5e14792e483..513d17ceafd 100644 --- a/arch/um/sys-x86_64/user-offsets.c +++ b/arch/um/sys-x86_64/user-offsets.c @@ -3,6 +3,14 @@ #include <signal.h> #define __FRAME_OFFSETS #include <asm/ptrace.h> +#include <asm/types.h> +/* For some reason, x86_64 defines u64 and u32 only in <pci/types.h>, which I + * refuse to include here, even though they're used throughout the headers. + * These are used in asm/user.h, and that include can't be avoided because of + * the sizeof(struct user_regs_struct) below. + */ +typedef __u64 u64; +typedef __u32 u32; #include <asm/user.h> #define DEFINE(sym, val) \ |