init/do_mounts.c

   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
  38 int __initdata rd_doload;       /* 1 = load RAM disk, 0 = don't load */
  39
  40 int root_mountflags = MS_RDONLY | MS_SILENT;
  41 static char * __initdata root_device_name;
  42 static char __initdata saved_root_name[64];
  43 static int root_wait;
  44
  45 dev_t ROOT_DEV;
  46
  47 static 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
  54 static int __init readonly(char *str)
  55 {
  56         if (*str)
  57                 return 0;
  58         root_mountflags |= MS_RDONLY;
  59         return 1;
  60 }
  61
  62 static 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
  74 struct 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  */
  86 static 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;
  98 no_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  */
 116 static 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
 168 no_offset:
 169         put_device(dev);
 170 done:
 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) device number in hexadecimal represents itself
 186  *      2) /dev/nfs represents Root_NFS (0xff)
 187  *      3) /dev/<disk_name> represents the device number of disk
 188  *      4) /dev/<disk_name><decimal> represents the device number
 189  *         of partition - device number of disk plus the partition number
 190  *      5) /dev/<disk_name>p<decimal> - same as the above, that form is
 191  *         used when disk name of partitioned disk ends on a digit.
 192  *      6) PARTUUID=00112233-4455-6677-8899-AABBCCDDEEFF representing the
 193  *         unique id of a partition if the partition table provides it.
 194  *         The UUID may be either an EFI/GPT UUID, or refer to an MSDOS
 195  *         partition using the format SSSSSSSS-PP, where SSSSSSSS is a zero-
 196  *         filled hex representation of the 32-bit "NT disk signature", and PP
 197  *         is a zero-filled hex representation of the 1-based partition number.
 198  *      7) PARTUUID=<UUID>/PARTNROFF=<int> to select a partition in relation to
 199  *         a partition with a known unique id.
 200  *      8) <major>:<minor> major and minor number of the device separated by
 201  *         a colon.
 202  *
 203  *      If name doesn't have fall into the categories above, we return (0,0).
 204  *      block_class is used to check if something is a disk name. If the disk
 205  *      name contains slashes, the device name has them replaced with
 206  *      bangs.
 207  */
 208
 209 dev_t name_to_dev_t(char *name)
 210 {
 211         char s[32];
 212         char *p;
 213         dev_t res = 0;
 214         int part;
 215
 216 #ifdef CONFIG_BLOCK
 217         if (strncmp(name, "PARTUUID=", 9) == 0) {
 218                 name += 9;
 219                 res = devt_from_partuuid(name);
 220                 if (!res)
 221                         goto fail;
 222                 goto done;
 223         }
 224 #endif
 225
 226         if (strncmp(name, "/dev/", 5) != 0) {
 227                 unsigned maj, min;
 228
 229                 if (sscanf(name, "%u:%u", &maj, &min) == 2) {
 230                         res = MKDEV(maj, min);
 231                         if (maj != MAJOR(res) || min != MINOR(res))
 232                                 goto fail;
 233                 } else {
 234                         res = new_decode_dev(simple_strtoul(name, &p, 16));
 235                         if (*p)
 236                                 goto fail;
 237                 }
 238                 goto done;
 239         }
 240
 241         name += 5;
 242         res = Root_NFS;
 243         if (strcmp(name, "nfs") == 0)
 244                 goto done;
 245         res = Root_RAM0;
 246         if (strcmp(name, "ram") == 0)
 247                 goto done;
 248
 249         if (strlen(name) > 31)
 250                 goto fail;
 251         strcpy(s, name);
 252         for (p = s; *p; p++)
 253                 if (*p == '/')
 254                         *p = '!';
 255         res = blk_lookup_devt(s, 0);
 256         if (res)
 257                 goto done;
 258
 259         /*
 260          * try non-existent, but valid partition, which may only exist
 261          * after revalidating the disk, like partitioned md devices
 262          */
 263         while (p > s && isdigit(p[-1]))
 264                 p--;
 265         if (p == s || !*p || *p == '0')
 266                 goto fail;
 267
 268         /* try disk name without <part number> */
 269         part = simple_strtoul(p, NULL, 10);
 270         *p = '\0';
 271         res = blk_lookup_devt(s, part);
 272         if (res)
 273                 goto done;
 274
 275         /* try disk name without p<part number> */
 276         if (p < s + 2 || !isdigit(p[-2]) || p[-1] != 'p')
 277                 goto fail;
 278         p[-1] = '\0';
 279         res = blk_lookup_devt(s, part);
 280         if (res)
 281                 goto done;
 282
 283 fail:
 284         return 0;
 285 done:
 286         return res;
 287 }
 288
 289 static int __init root_dev_setup(char *line)
 290 {
 291         strlcpy(saved_root_name, line, sizeof(saved_root_name));
 292         return 1;
 293 }
 294
 295 __setup("root=", root_dev_setup);
 296
 297 static int __init rootwait_setup(char *str)
 298 {
 299         if (*str)
 300                 return 0;
 301         root_wait = 1;
 302         return 1;
 303 }
 304
 305 __setup("rootwait", rootwait_setup);
 306
 307 static char * __initdata root_mount_data;
 308 static int __init root_data_setup(char *str)
 309 {
 310         root_mount_data = str;
 311         return 1;
 312 }
 313
 314 static char * __initdata root_fs_names;
 315 static int __init fs_names_setup(char *str)
 316 {
 317         root_fs_names = str;
 318         return 1;
 319 }
 320
 321 static unsigned int __initdata root_delay;
 322 static int __init root_delay_setup(char *str)
 323 {
 324         root_delay = simple_strtoul(str, NULL, 0);
 325         return 1;
 326 }
 327
 328 __setup("rootflags=", root_data_setup);
 329 __setup("rootfstype=", fs_names_setup);
 330 __setup("rootdelay=", root_delay_setup);
 331
 332 static void __init get_fs_names(char *page)
 333 {
 334         char *s = page;
 335
 336         if (root_fs_names) {
 337                 strcpy(page, root_fs_names);
 338                 while (*s++) {
 339                         if (s[-1] == ',')
 340                                 s[-1] = '\0';
 341                 }
 342         } else {
 343                 int len = get_filesystem_list(page);
 344                 char *p, *next;
 345
 346                 page[len] = '\0';
 347                 for (p = page-1; p; p = next) {
 348                         next = strchr(++p, '\n');
 349                         if (*p++ != '\t')
 350                                 continue;
 351                         while ((*s++ = *p++) != '\n')
 352                                 ;
 353                         s[-1] = '\0';
 354                 }
 355         }
 356         *s = '\0';
 357 }
 358
 359 static int __init do_mount_root(char *name, char *fs, int flags, void *data)
 360 {
 361         struct super_block *s;
 362         int err = sys_mount(name, "/root", fs, flags, data);
 363         if (err)
 364                 return err;
 365
 366         sys_chdir("/root");
 367         s = current->fs->pwd.dentry->d_sb;
 368         ROOT_DEV = s->s_dev;
 369         printk(KERN_INFO
 370                "VFS: Mounted root (%s filesystem)%s on device %u:%u.\n",
 371                s->s_type->name,
 372                s->s_flags & MS_RDONLY ?  " readonly" : "",
 373                MAJOR(ROOT_DEV), MINOR(ROOT_DEV));
 374         return 0;
 375 }
 376
 377 void __init mount_block_root(char *name, int flags)
 378 {
 379         struct page *page = alloc_page(GFP_KERNEL |
 380                                         __GFP_NOTRACK_FALSE_POSITIVE);
 381         char *fs_names = page_address(page);
 382         char *p;
 383 #ifdef CONFIG_BLOCK
 384         char b[BDEVNAME_SIZE];
 385 #else
 386         const char *b = name;
 387 #endif
 388
 389         get_fs_names(fs_names);
 390 retry:
 391         for (p = fs_names; *p; p += strlen(p)+1) {
 392                 int err = do_mount_root(name, p, flags, root_mount_data);
 393                 switch (err) {
 394                         case 0:
 395                                 goto out;
 396                         case -EACCES:
 397                                 flags |= MS_RDONLY;
 398                                 goto retry;
 399                         case -EINVAL:
 400                                 continue;
 401                 }
 402                 /*
 403                  * Allow the user to distinguish between failed sys_open
 404                  * and bad superblock on root device.
 405                  * and give them a list of the available devices
 406                  */
 407 #ifdef CONFIG_BLOCK
 408                 __bdevname(ROOT_DEV, b);
 409 #endif
 410                 printk("VFS: Cannot open root device \"%s\" or %s: error %d\n",
 411                                 root_device_name, b, err);
 412                 printk("Please append a correct \"root=\" boot option; here are the available partitions:\n");
 413
 414                 printk_all_partitions();
 415 #ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT
 416                 printk("DEBUG_BLOCK_EXT_DEVT is enabled, you need to specify "
 417                        "explicit textual name for \"root=\" boot option.\n");
 418 #endif
 419                 panic("VFS: Unable to mount root fs on %s", b);
 420         }
 421
 422         printk("List of all partitions:\n");
 423         printk_all_partitions();
 424         printk("No filesystem could mount root, tried: ");
 425         for (p = fs_names; *p; p += strlen(p)+1)
 426                 printk(" %s", p);
 427         printk("\n");
 428 #ifdef CONFIG_BLOCK
 429         __bdevname(ROOT_DEV, b);
 430 #endif
 431         panic("VFS: Unable to mount root fs on %s", b);
 432 out:
 433         put_page(page);
 434 }
 435
 436 #ifdef CONFIG_ROOT_NFS
 437
 438 #define NFSROOT_TIMEOUT_MIN     5
 439 #define NFSROOT_TIMEOUT_MAX     30
 440 #define NFSROOT_RETRY_MAX       5
 441
 442 static int __init mount_nfs_root(void)
 443 {
 444         char *root_dev, *root_data;
 445         unsigned int timeout;
 446         int try, err;
 447
 448         err = nfs_root_data(&root_dev, &root_data);
 449         if (err != 0)
 450                 return 0;
 451
 452         /*
 453          * The server or network may not be ready, so try several
 454          * times.  Stop after a few tries in case the client wants
 455          * to fall back to other boot methods.
 456          */
 457         timeout = NFSROOT_TIMEOUT_MIN;
 458         for (try = 1; ; try++) {
 459                 err = do_mount_root(root_dev, "nfs",
 460                                         root_mountflags, root_data);
 461                 if (err == 0)
 462                         return 1;
 463                 if (try > NFSROOT_RETRY_MAX)
 464                         break;
 465
 466                 /* Wait, in case the server refused us immediately */
 467                 ssleep(timeout);
 468                 timeout <<= 1;
 469                 if (timeout > NFSROOT_TIMEOUT_MAX)
 470                         timeout = NFSROOT_TIMEOUT_MAX;
 471         }
 472         return 0;
 473 }
 474 #endif
 475
 476 #if defined(CONFIG_BLK_DEV_RAM) || defined(CONFIG_BLK_DEV_FD)
 477 void __init change_floppy(char *fmt, ...)
 478 {
 479         struct termios termios;
 480         char buf[80];
 481         char c;
 482         int fd;
 483         va_list args;
 484         va_start(args, fmt);
 485         vsprintf(buf, fmt, args);
 486         va_end(args);
 487         fd = sys_open("/dev/root", O_RDWR | O_NDELAY, 0);
 488         if (fd >= 0) {
 489                 sys_ioctl(fd, FDEJECT, 0);
 490                 sys_close(fd);
 491         }
 492         printk(KERN_NOTICE "VFS: Insert %s and press ENTER\n", buf);
 493         fd = sys_open("/dev/console", O_RDWR, 0);
 494         if (fd >= 0) {
 495                 sys_ioctl(fd, TCGETS, (long)&termios);
 496                 termios.c_lflag &= ~ICANON;
 497                 sys_ioctl(fd, TCSETSF, (long)&termios);
 498                 sys_read(fd, &c, 1);
 499                 termios.c_lflag |= ICANON;
 500                 sys_ioctl(fd, TCSETSF, (long)&termios);
 501                 sys_close(fd);
 502         }
 503 }
 504 #endif
 505
 506 void __init mount_root(void)
 507 {
 508 #ifdef CONFIG_ROOT_NFS
 509         if (ROOT_DEV == Root_NFS) {
 510                 if (mount_nfs_root())
 511                         return;
 512
 513                 printk(KERN_ERR "VFS: Unable to mount root fs via NFS, trying floppy.\n");
 514                 ROOT_DEV = Root_FD0;
 515         }
 516 #endif
 517 #ifdef CONFIG_BLK_DEV_FD
 518         if (MAJOR(ROOT_DEV) == FLOPPY_MAJOR) {
 519                 /* rd_doload is 2 for a dual initrd/ramload setup */
 520                 if (rd_doload==2) {
 521                         if (rd_load_disk(1)) {
 522                                 ROOT_DEV = Root_RAM1;
 523                                 root_device_name = NULL;
 524                         }
 525                 } else
 526                         change_floppy("root floppy");
 527         }
 528 #endif
 529 #ifdef CONFIG_BLOCK
 530         create_dev("/dev/root", ROOT_DEV);
 531         mount_block_root("/dev/root", root_mountflags);
 532 #endif
 533 }
 534
 535 /*
 536  * Prepare the namespace - decide what/where to mount, load ramdisks, etc.
 537  */
 538 void __init prepare_namespace(void)
 539 {
 540         int is_floppy;
 541
 542         if (root_delay) {
 543                 printk(KERN_INFO "Waiting %d sec before mounting root device...\n",
 544                        root_delay);
 545                 ssleep(root_delay);
 546         }
 547
 548         /*
 549          * wait for the known devices to complete their probing
 550          *
 551          * Note: this is a potential source of long boot delays.
 552          * For example, it is not atypical to wait 5 seconds here
 553          * for the touchpad of a laptop to initialize.
 554          */
 555         wait_for_device_probe();
 556
 557         md_run_setup();
 558
 559         if (saved_root_name[0]) {
 560                 root_device_name = saved_root_name;
 561                 if (!strncmp(root_device_name, "mtd", 3) ||
 562                     !strncmp(root_device_name, "ubi", 3)) {
 563                         mount_block_root(root_device_name, root_mountflags);
 564                         goto out;
 565                 }
 566                 ROOT_DEV = name_to_dev_t(root_device_name);
 567                 if (strncmp(root_device_name, "/dev/", 5) == 0)
 568                         root_device_name += 5;
 569         }
 570
 571         if (initrd_load())
 572                 goto out;
 573
 574         /* wait for any asynchronous scanning to complete */
 575         if ((ROOT_DEV == 0) && root_wait) {
 576                 printk(KERN_INFO "Waiting for root device %s...\n",
 577                         saved_root_name);
 578                 while (driver_probe_done() != 0 ||
 579                         (ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
 580                         msleep(100);
 581                 async_synchronize_full();
 582         }
 583
 584         is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;
 585
 586         if (is_floppy && rd_doload && rd_load_disk(0))
 587                 ROOT_DEV = Root_RAM0;
 588
 589         mount_root();
 590 out:
 591         devtmpfs_mount("dev");
 592         sys_mount(".", "/", NULL, MS_MOVE, NULL);
 593         sys_chroot(".");
 594 }
 595
 596 static bool is_tmpfs;
 597 static struct dentry *rootfs_mount(struct file_system_type *fs_type,
 598         int flags, const char *dev_name, void *data)
 599 {
 600         static unsigned long once;
 601         void *fill = ramfs_fill_super;
 602
 603         if (test_and_set_bit(0, &once))
 604                 return ERR_PTR(-ENODEV);
 605
 606         if (IS_ENABLED(CONFIG_TMPFS) && is_tmpfs)
 607                 fill = shmem_fill_super;
 608
 609         return mount_nodev(fs_type, flags, data, fill);
 610 }
 611
 612 static struct file_system_type rootfs_fs_type = {
 613         .name           = "rootfs",
 614         .mount          = rootfs_mount,
 615         .kill_sb        = kill_litter_super,
 616 };
 617
 618 int __init init_rootfs(void)
 619 {
 620         int err = register_filesystem(&rootfs_fs_type);
 621
 622         if (err)
 623                 return err;
 624
 625         if (IS_ENABLED(CONFIG_TMPFS) && !saved_root_name[0] &&
 626                 (!root_fs_names || strstr(root_fs_names, "tmpfs"))) {
 627                 err = shmem_init();
 628                 is_tmpfs = true;
 629         } else {
 630                 err = init_ramfs_fs();
 631         }
 632
 633         if (err)
 634                 unregister_filesystem(&rootfs_fs_type);
 635
 636         return err;
 637 }