Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

at v4.0-rc4 640 lines 15 kB view raw
1/* 2 * Many of the syscalls used in this file expect some of the arguments 3 * to be __user pointers not __kernel pointers. To limit the sparse 4 * noise, turn off sparse checking for this file. 5 */ 6#ifdef __CHECKER__ 7#undef __CHECKER__ 8#warning "Sparse checking disabled for this file" 9#endif 10 11#include <linux/module.h> 12#include <linux/sched.h> 13#include <linux/ctype.h> 14#include <linux/fd.h> 15#include <linux/tty.h> 16#include <linux/suspend.h> 17#include <linux/root_dev.h> 18#include <linux/security.h> 19#include <linux/delay.h> 20#include <linux/genhd.h> 21#include <linux/mount.h> 22#include <linux/device.h> 23#include <linux/init.h> 24#include <linux/fs.h> 25#include <linux/initrd.h> 26#include <linux/async.h> 27#include <linux/fs_struct.h> 28#include <linux/slab.h> 29#include <linux/ramfs.h> 30#include <linux/shmem_fs.h> 31 32#include <linux/nfs_fs.h> 33#include <linux/nfs_fs_sb.h> 34#include <linux/nfs_mount.h> 35 36#include "do_mounts.h" 37 38int __initdata rd_doload; /* 1 = load RAM disk, 0 = don't load */ 39 40int root_mountflags = MS_RDONLY | MS_SILENT; 41static char * __initdata root_device_name; 42static char __initdata saved_root_name[64]; 43static int root_wait; 44 45dev_t ROOT_DEV; 46 47static int __init load_ramdisk(char *str) 48{ 49 rd_doload = simple_strtol(str,NULL,0) & 3; 50 return 1; 51} 52__setup("load_ramdisk=", load_ramdisk); 53 54static int __init readonly(char *str) 55{ 56 if (*str) 57 return 0; 58 root_mountflags |= MS_RDONLY; 59 return 1; 60} 61 62static int __init readwrite(char *str) 63{ 64 if (*str) 65 return 0; 66 root_mountflags &= ~MS_RDONLY; 67 return 1; 68} 69 70__setup("ro", readonly); 71__setup("rw", readwrite); 72 73#ifdef CONFIG_BLOCK 74struct uuidcmp { 75 const char *uuid; 76 int len; 77}; 78 79/** 80 * match_dev_by_uuid - callback for finding a partition using its uuid 81 * @dev: device passed in by the caller 82 * @data: opaque pointer to the desired struct uuidcmp to match 83 * 84 * Returns 1 if the device matches, and 0 otherwise. 85 */ 86static int match_dev_by_uuid(struct device *dev, const void *data) 87{ 88 const struct uuidcmp *cmp = data; 89 struct hd_struct *part = dev_to_part(dev); 90 91 if (!part->info) 92 goto no_match; 93 94 if (strncasecmp(cmp->uuid, part->info->uuid, cmp->len)) 95 goto no_match; 96 97 return 1; 98no_match: 99 return 0; 100} 101 102 103/** 104 * devt_from_partuuid - looks up the dev_t of a partition by its UUID 105 * @uuid_str: char array containing ascii UUID 106 * 107 * The function will return the first partition which contains a matching 108 * UUID value in its partition_meta_info struct. This does not search 109 * by filesystem UUIDs. 110 * 111 * If @uuid_str is followed by a "/PARTNROFF=%d", then the number will be 112 * extracted and used as an offset from the partition identified by the UUID. 113 * 114 * Returns the matching dev_t on success or 0 on failure. 115 */ 116static dev_t devt_from_partuuid(const char *uuid_str) 117{ 118 dev_t res = 0; 119 struct uuidcmp cmp; 120 struct device *dev = NULL; 121 struct gendisk *disk; 122 struct hd_struct *part; 123 int offset = 0; 124 bool clear_root_wait = false; 125 char *slash; 126 127 cmp.uuid = uuid_str; 128 129 slash = strchr(uuid_str, '/'); 130 /* Check for optional partition number offset attributes. */ 131 if (slash) { 132 char c = 0; 133 /* Explicitly fail on poor PARTUUID syntax. */ 134 if (sscanf(slash + 1, 135 "PARTNROFF=%d%c", &offset, &c) != 1) { 136 clear_root_wait = true; 137 goto done; 138 } 139 cmp.len = slash - uuid_str; 140 } else { 141 cmp.len = strlen(uuid_str); 142 } 143 144 if (!cmp.len) { 145 clear_root_wait = true; 146 goto done; 147 } 148 149 dev = class_find_device(&block_class, NULL, &cmp, 150 &match_dev_by_uuid); 151 if (!dev) 152 goto done; 153 154 res = dev->devt; 155 156 /* Attempt to find the partition by offset. */ 157 if (!offset) 158 goto no_offset; 159 160 res = 0; 161 disk = part_to_disk(dev_to_part(dev)); 162 part = disk_get_part(disk, dev_to_part(dev)->partno + offset); 163 if (part) { 164 res = part_devt(part); 165 put_device(part_to_dev(part)); 166 } 167 168no_offset: 169 put_device(dev); 170done: 171 if (clear_root_wait) { 172 pr_err("VFS: PARTUUID= is invalid.\n" 173 "Expected PARTUUID=<valid-uuid-id>[/PARTNROFF=%%d]\n"); 174 if (root_wait) 175 pr_err("Disabling rootwait; root= is invalid.\n"); 176 root_wait = 0; 177 } 178 return res; 179} 180#endif 181 182/* 183 * Convert a name into device number. We accept the following variants: 184 * 185 * 1) <hex_major><hex_minor> device number in hexadecimal represents itself 186 * no leading 0x, for example b302. 187 * 2) /dev/nfs represents Root_NFS (0xff) 188 * 3) /dev/<disk_name> represents the device number of disk 189 * 4) /dev/<disk_name><decimal> represents the device number 190 * of partition - device number of disk plus the partition number 191 * 5) /dev/<disk_name>p<decimal> - same as the above, that form is 192 * used when disk name of partitioned disk ends on a digit. 193 * 6) PARTUUID=00112233-4455-6677-8899-AABBCCDDEEFF representing the 194 * unique id of a partition if the partition table provides it. 195 * The UUID may be either an EFI/GPT UUID, or refer to an MSDOS 196 * partition using the format SSSSSSSS-PP, where SSSSSSSS is a zero- 197 * filled hex representation of the 32-bit "NT disk signature", and PP 198 * is a zero-filled hex representation of the 1-based partition number. 199 * 7) PARTUUID=<UUID>/PARTNROFF=<int> to select a partition in relation to 200 * a partition with a known unique id. 201 * 8) <major>:<minor> major and minor number of the device separated by 202 * a colon. 203 * 204 * If name doesn't have fall into the categories above, we return (0,0). 205 * block_class is used to check if something is a disk name. If the disk 206 * name contains slashes, the device name has them replaced with 207 * bangs. 208 */ 209 210dev_t name_to_dev_t(char *name) 211{ 212 char s[32]; 213 char *p; 214 dev_t res = 0; 215 int part; 216 217#ifdef CONFIG_BLOCK 218 if (strncmp(name, "PARTUUID=", 9) == 0) { 219 name += 9; 220 res = devt_from_partuuid(name); 221 if (!res) 222 goto fail; 223 goto done; 224 } 225#endif 226 227 if (strncmp(name, "/dev/", 5) != 0) { 228 unsigned maj, min; 229 230 if (sscanf(name, "%u:%u", &maj, &min) == 2) { 231 res = MKDEV(maj, min); 232 if (maj != MAJOR(res) || min != MINOR(res)) 233 goto fail; 234 } else { 235 res = new_decode_dev(simple_strtoul(name, &p, 16)); 236 if (*p) 237 goto fail; 238 } 239 goto done; 240 } 241 242 name += 5; 243 res = Root_NFS; 244 if (strcmp(name, "nfs") == 0) 245 goto done; 246 res = Root_RAM0; 247 if (strcmp(name, "ram") == 0) 248 goto done; 249 250 if (strlen(name) > 31) 251 goto fail; 252 strcpy(s, name); 253 for (p = s; *p; p++) 254 if (*p == '/') 255 *p = '!'; 256 res = blk_lookup_devt(s, 0); 257 if (res) 258 goto done; 259 260 /* 261 * try non-existent, but valid partition, which may only exist 262 * after revalidating the disk, like partitioned md devices 263 */ 264 while (p > s && isdigit(p[-1])) 265 p--; 266 if (p == s || !*p || *p == '0') 267 goto fail; 268 269 /* try disk name without <part number> */ 270 part = simple_strtoul(p, NULL, 10); 271 *p = '\0'; 272 res = blk_lookup_devt(s, part); 273 if (res) 274 goto done; 275 276 /* try disk name without p<part number> */ 277 if (p < s + 2 || !isdigit(p[-2]) || p[-1] != 'p') 278 goto fail; 279 p[-1] = '\0'; 280 res = blk_lookup_devt(s, part); 281 if (res) 282 goto done; 283 284fail: 285 return 0; 286done: 287 return res; 288} 289 290static int __init root_dev_setup(char *line) 291{ 292 strlcpy(saved_root_name, line, sizeof(saved_root_name)); 293 return 1; 294} 295 296__setup("root=", root_dev_setup); 297 298static int __init rootwait_setup(char *str) 299{ 300 if (*str) 301 return 0; 302 root_wait = 1; 303 return 1; 304} 305 306__setup("rootwait", rootwait_setup); 307 308static char * __initdata root_mount_data; 309static int __init root_data_setup(char *str) 310{ 311 root_mount_data = str; 312 return 1; 313} 314 315static char * __initdata root_fs_names; 316static int __init fs_names_setup(char *str) 317{ 318 root_fs_names = str; 319 return 1; 320} 321 322static unsigned int __initdata root_delay; 323static int __init root_delay_setup(char *str) 324{ 325 root_delay = simple_strtoul(str, NULL, 0); 326 return 1; 327} 328 329__setup("rootflags=", root_data_setup); 330__setup("rootfstype=", fs_names_setup); 331__setup("rootdelay=", root_delay_setup); 332 333static void __init get_fs_names(char *page) 334{ 335 char *s = page; 336 337 if (root_fs_names) { 338 strcpy(page, root_fs_names); 339 while (*s++) { 340 if (s[-1] == ',') 341 s[-1] = '\0'; 342 } 343 } else { 344 int len = get_filesystem_list(page); 345 char *p, *next; 346 347 page[len] = '\0'; 348 for (p = page-1; p; p = next) { 349 next = strchr(++p, '\n'); 350 if (*p++ != '\t') 351 continue; 352 while ((*s++ = *p++) != '\n') 353 ; 354 s[-1] = '\0'; 355 } 356 } 357 *s = '\0'; 358} 359 360static int __init do_mount_root(char *name, char *fs, int flags, void *data) 361{ 362 struct super_block *s; 363 int err = sys_mount(name, "/root", fs, flags, data); 364 if (err) 365 return err; 366 367 sys_chdir("/root"); 368 s = current->fs->pwd.dentry->d_sb; 369 ROOT_DEV = s->s_dev; 370 printk(KERN_INFO 371 "VFS: Mounted root (%s filesystem)%s on device %u:%u.\n", 372 s->s_type->name, 373 s->s_flags & MS_RDONLY ? " readonly" : "", 374 MAJOR(ROOT_DEV), MINOR(ROOT_DEV)); 375 return 0; 376} 377 378void __init mount_block_root(char *name, int flags) 379{ 380 struct page *page = alloc_page(GFP_KERNEL | 381 __GFP_NOTRACK_FALSE_POSITIVE); 382 char *fs_names = page_address(page); 383 char *p; 384#ifdef CONFIG_BLOCK 385 char b[BDEVNAME_SIZE]; 386#else 387 const char *b = name; 388#endif 389 390 get_fs_names(fs_names); 391retry: 392 for (p = fs_names; *p; p += strlen(p)+1) { 393 int err = do_mount_root(name, p, flags, root_mount_data); 394 switch (err) { 395 case 0: 396 goto out; 397 case -EACCES: 398 case -EINVAL: 399 continue; 400 } 401 /* 402 * Allow the user to distinguish between failed sys_open 403 * and bad superblock on root device. 404 * and give them a list of the available devices 405 */ 406#ifdef CONFIG_BLOCK 407 __bdevname(ROOT_DEV, b); 408#endif 409 printk("VFS: Cannot open root device \"%s\" or %s: error %d\n", 410 root_device_name, b, err); 411 printk("Please append a correct \"root=\" boot option; here are the available partitions:\n"); 412 413 printk_all_partitions(); 414#ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT 415 printk("DEBUG_BLOCK_EXT_DEVT is enabled, you need to specify " 416 "explicit textual name for \"root=\" boot option.\n"); 417#endif 418 panic("VFS: Unable to mount root fs on %s", b); 419 } 420 if (!(flags & MS_RDONLY)) { 421 flags |= MS_RDONLY; 422 goto retry; 423 } 424 425 printk("List of all partitions:\n"); 426 printk_all_partitions(); 427 printk("No filesystem could mount root, tried: "); 428 for (p = fs_names; *p; p += strlen(p)+1) 429 printk(" %s", p); 430 printk("\n"); 431#ifdef CONFIG_BLOCK 432 __bdevname(ROOT_DEV, b); 433#endif 434 panic("VFS: Unable to mount root fs on %s", b); 435out: 436 put_page(page); 437} 438 439#ifdef CONFIG_ROOT_NFS 440 441#define NFSROOT_TIMEOUT_MIN 5 442#define NFSROOT_TIMEOUT_MAX 30 443#define NFSROOT_RETRY_MAX 5 444 445static int __init mount_nfs_root(void) 446{ 447 char *root_dev, *root_data; 448 unsigned int timeout; 449 int try, err; 450 451 err = nfs_root_data(&root_dev, &root_data); 452 if (err != 0) 453 return 0; 454 455 /* 456 * The server or network may not be ready, so try several 457 * times. Stop after a few tries in case the client wants 458 * to fall back to other boot methods. 459 */ 460 timeout = NFSROOT_TIMEOUT_MIN; 461 for (try = 1; ; try++) { 462 err = do_mount_root(root_dev, "nfs", 463 root_mountflags, root_data); 464 if (err == 0) 465 return 1; 466 if (try > NFSROOT_RETRY_MAX) 467 break; 468 469 /* Wait, in case the server refused us immediately */ 470 ssleep(timeout); 471 timeout <<= 1; 472 if (timeout > NFSROOT_TIMEOUT_MAX) 473 timeout = NFSROOT_TIMEOUT_MAX; 474 } 475 return 0; 476} 477#endif 478 479#if defined(CONFIG_BLK_DEV_RAM) || defined(CONFIG_BLK_DEV_FD) 480void __init change_floppy(char *fmt, ...) 481{ 482 struct termios termios; 483 char buf[80]; 484 char c; 485 int fd; 486 va_list args; 487 va_start(args, fmt); 488 vsprintf(buf, fmt, args); 489 va_end(args); 490 fd = sys_open("/dev/root", O_RDWR | O_NDELAY, 0); 491 if (fd >= 0) { 492 sys_ioctl(fd, FDEJECT, 0); 493 sys_close(fd); 494 } 495 printk(KERN_NOTICE "VFS: Insert %s and press ENTER\n", buf); 496 fd = sys_open("/dev/console", O_RDWR, 0); 497 if (fd >= 0) { 498 sys_ioctl(fd, TCGETS, (long)&termios); 499 termios.c_lflag &= ~ICANON; 500 sys_ioctl(fd, TCSETSF, (long)&termios); 501 sys_read(fd, &c, 1); 502 termios.c_lflag |= ICANON; 503 sys_ioctl(fd, TCSETSF, (long)&termios); 504 sys_close(fd); 505 } 506} 507#endif 508 509void __init mount_root(void) 510{ 511#ifdef CONFIG_ROOT_NFS 512 if (ROOT_DEV == Root_NFS) { 513 if (mount_nfs_root()) 514 return; 515 516 printk(KERN_ERR "VFS: Unable to mount root fs via NFS, trying floppy.\n"); 517 ROOT_DEV = Root_FD0; 518 } 519#endif 520#ifdef CONFIG_BLK_DEV_FD 521 if (MAJOR(ROOT_DEV) == FLOPPY_MAJOR) { 522 /* rd_doload is 2 for a dual initrd/ramload setup */ 523 if (rd_doload==2) { 524 if (rd_load_disk(1)) { 525 ROOT_DEV = Root_RAM1; 526 root_device_name = NULL; 527 } 528 } else 529 change_floppy("root floppy"); 530 } 531#endif 532#ifdef CONFIG_BLOCK 533 create_dev("/dev/root", ROOT_DEV); 534 mount_block_root("/dev/root", root_mountflags); 535#endif 536} 537 538/* 539 * Prepare the namespace - decide what/where to mount, load ramdisks, etc. 540 */ 541void __init prepare_namespace(void) 542{ 543 int is_floppy; 544 545 if (root_delay) { 546 printk(KERN_INFO "Waiting %d sec before mounting root device...\n", 547 root_delay); 548 ssleep(root_delay); 549 } 550 551 /* 552 * wait for the known devices to complete their probing 553 * 554 * Note: this is a potential source of long boot delays. 555 * For example, it is not atypical to wait 5 seconds here 556 * for the touchpad of a laptop to initialize. 557 */ 558 wait_for_device_probe(); 559 560 md_run_setup(); 561 562 if (saved_root_name[0]) { 563 root_device_name = saved_root_name; 564 if (!strncmp(root_device_name, "mtd", 3) || 565 !strncmp(root_device_name, "ubi", 3)) { 566 mount_block_root(root_device_name, root_mountflags); 567 goto out; 568 } 569 ROOT_DEV = name_to_dev_t(root_device_name); 570 if (strncmp(root_device_name, "/dev/", 5) == 0) 571 root_device_name += 5; 572 } 573 574 if (initrd_load()) 575 goto out; 576 577 /* wait for any asynchronous scanning to complete */ 578 if ((ROOT_DEV == 0) && root_wait) { 579 printk(KERN_INFO "Waiting for root device %s...\n", 580 saved_root_name); 581 while (driver_probe_done() != 0 || 582 (ROOT_DEV = name_to_dev_t(saved_root_name)) == 0) 583 msleep(100); 584 async_synchronize_full(); 585 } 586 587 is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR; 588 589 if (is_floppy && rd_doload && rd_load_disk(0)) 590 ROOT_DEV = Root_RAM0; 591 592 mount_root(); 593out: 594 devtmpfs_mount("dev"); 595 sys_mount(".", "/", NULL, MS_MOVE, NULL); 596 sys_chroot("."); 597} 598 599static bool is_tmpfs; 600static struct dentry *rootfs_mount(struct file_system_type *fs_type, 601 int flags, const char *dev_name, void *data) 602{ 603 static unsigned long once; 604 void *fill = ramfs_fill_super; 605 606 if (test_and_set_bit(0, &once)) 607 return ERR_PTR(-ENODEV); 608 609 if (IS_ENABLED(CONFIG_TMPFS) && is_tmpfs) 610 fill = shmem_fill_super; 611 612 return mount_nodev(fs_type, flags, data, fill); 613} 614 615static struct file_system_type rootfs_fs_type = { 616 .name = "rootfs", 617 .mount = rootfs_mount, 618 .kill_sb = kill_litter_super, 619}; 620 621int __init init_rootfs(void) 622{ 623 int err = register_filesystem(&rootfs_fs_type); 624 625 if (err) 626 return err; 627 628 if (IS_ENABLED(CONFIG_TMPFS) && !saved_root_name[0] && 629 (!root_fs_names || strstr(root_fs_names, "tmpfs"))) { 630 err = shmem_init(); 631 is_tmpfs = true; 632 } else { 633 err = init_ramfs_fs(); 634 } 635 636 if (err) 637 unregister_filesystem(&rootfs_fs_type); 638 639 return err; 640}