ipc: always handle a new value of auto_msgmni
[linux-2.6-block.git] / init / do_mounts.c
CommitLineData
c67e5382
HS
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
1da177e4
LT
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>
dd2a345f 20#include <linux/genhd.h>
d53d9f16 21#include <linux/mount.h>
d779249e 22#include <linux/device.h>
46595390 23#include <linux/init.h>
011e3fcd 24#include <linux/fs.h>
82c8253a 25#include <linux/initrd.h>
22a9d645 26#include <linux/async.h>
5ad4e53b 27#include <linux/fs_struct.h>
5a0e3ad6 28#include <linux/slab.h>
57f150a5 29#include <linux/ramfs.h>
16203a7a 30#include <linux/shmem_fs.h>
1da177e4
LT
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
1da177e4
LT
38int __initdata rd_doload; /* 1 = load RAM disk, 0 = don't load */
39
9b04c997 40int root_mountflags = MS_RDONLY | MS_SILENT;
f56f6d30 41static char * __initdata root_device_name;
1da177e4 42static char __initdata saved_root_name[64];
79975f13 43static int root_wait;
1da177e4 44
1da177e4
LT
45dev_t ROOT_DEV;
46
1da177e4
LT
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
6d0aed7a 73#ifdef CONFIG_BLOCK
1ad7e899
SW
74struct uuidcmp {
75 const char *uuid;
76 int len;
77};
78
b5af921e
WD
79/**
80 * match_dev_by_uuid - callback for finding a partition using its uuid
81 * @dev: device passed in by the caller
1ad7e899 82 * @data: opaque pointer to the desired struct uuidcmp to match
b5af921e
WD
83 *
84 * Returns 1 if the device matches, and 0 otherwise.
85 */
9f3b795a 86static int match_dev_by_uuid(struct device *dev, const void *data)
b5af921e 87{
9f3b795a 88 const struct uuidcmp *cmp = data;
b5af921e
WD
89 struct hd_struct *part = dev_to_part(dev);
90
91 if (!part->info)
92 goto no_match;
93
1ad7e899
SW
94 if (strncasecmp(cmp->uuid, part->info->uuid, cmp->len))
95 goto no_match;
b5af921e
WD
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
a68b3108 105 * @uuid_str: char array containing ascii UUID
b5af921e
WD
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 *
a68b3108 111 * If @uuid_str is followed by a "/PARTNROFF=%d", then the number will be
79975f13
WD
112 * extracted and used as an offset from the partition identified by the UUID.
113 *
b5af921e
WD
114 * Returns the matching dev_t on success or 0 on failure.
115 */
1ad7e899 116static dev_t devt_from_partuuid(const char *uuid_str)
b5af921e
WD
117{
118 dev_t res = 0;
1ad7e899 119 struct uuidcmp cmp;
b5af921e 120 struct device *dev = NULL;
79975f13
WD
121 struct gendisk *disk;
122 struct hd_struct *part;
123 int offset = 0;
283f8fc0
SW
124 bool clear_root_wait = false;
125 char *slash;
79975f13 126
1ad7e899 127 cmp.uuid = uuid_str;
1ad7e899 128
283f8fc0 129 slash = strchr(uuid_str, '/');
79975f13 130 /* Check for optional partition number offset attributes. */
283f8fc0 131 if (slash) {
79975f13
WD
132 char c = 0;
133 /* Explicitly fail on poor PARTUUID syntax. */
283f8fc0
SW
134 if (sscanf(slash + 1,
135 "PARTNROFF=%d%c", &offset, &c) != 1) {
136 clear_root_wait = true;
79975f13
WD
137 goto done;
138 }
283f8fc0
SW
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;
79975f13 147 }
b5af921e 148
1ad7e899
SW
149 dev = class_find_device(&block_class, NULL, &cmp,
150 &match_dev_by_uuid);
b5af921e
WD
151 if (!dev)
152 goto done;
153
154 res = dev->devt;
b5af921e 155
79975f13
WD
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);
b5af921e 170done:
283f8fc0
SW
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 }
b5af921e
WD
178 return res;
179}
6d0aed7a 180#endif
b5af921e 181
1da177e4
LT
182/*
183 * Convert a name into device number. We accept the following variants:
184 *
0bf37ae4
PM
185 * 1) <hex_major><hex_minor> device number in hexadecimal represents itself
186 * no leading 0x, for example b302.
1da177e4
LT
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.
b5af921e
WD
193 * 6) PARTUUID=00112233-4455-6677-8899-AABBCCDDEEFF representing the
194 * unique id of a partition if the partition table provides it.
d33b98fc
SW
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.
79975f13
WD
199 * 7) PARTUUID=<UUID>/PARTNROFF=<int> to select a partition in relation to
200 * a partition with a known unique id.
6c251611
SC
201 * 8) <major>:<minor> major and minor number of the device separated by
202 * a colon.
1da177e4 203 *
edfaa7c3
KS
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.
1da177e4
LT
208 */
209
210dev_t name_to_dev_t(char *name)
211{
212 char s[32];
213 char *p;
214 dev_t res = 0;
30f2f0eb 215 int part;
1da177e4 216
6d0aed7a 217#ifdef CONFIG_BLOCK
b5af921e
WD
218 if (strncmp(name, "PARTUUID=", 9) == 0) {
219 name += 9;
b5af921e
WD
220 res = devt_from_partuuid(name);
221 if (!res)
222 goto fail;
223 goto done;
224 }
6d0aed7a 225#endif
b5af921e 226
1da177e4
LT
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 }
edfaa7c3 241
1da177e4
LT
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 = '!';
30f2f0eb
KS
256 res = blk_lookup_devt(s, 0);
257 if (res)
258 goto done;
259
260 /*
25985edc 261 * try non-existent, but valid partition, which may only exist
30f2f0eb
KS
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);
1da177e4
LT
281 if (res)
282 goto done;
283
edfaa7c3
KS
284fail:
285 return 0;
1da177e4 286done:
1da177e4 287 return res;
1da177e4
LT
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
cc1ed754
PO
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
1da177e4
LT
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{
d8c9584e 362 struct super_block *s;
1da177e4
LT
363 int err = sys_mount(name, "/root", fs, flags, data);
364 if (err)
365 return err;
366
c67e5382 367 sys_chdir("/root");
d8c9584e
AV
368 s = current->fs->pwd.dentry->d_sb;
369 ROOT_DEV = s->s_dev;
80cdc6da
MSB
370 printk(KERN_INFO
371 "VFS: Mounted root (%s filesystem)%s on device %u:%u.\n",
d8c9584e
AV
372 s->s_type->name,
373 s->s_flags & MS_RDONLY ? " readonly" : "",
374 MAJOR(ROOT_DEV), MINOR(ROOT_DEV));
1da177e4
LT
375 return 0;
376}
377
378void __init mount_block_root(char *name, int flags)
379{
a608ca21
JL
380 struct page *page = alloc_page(GFP_KERNEL |
381 __GFP_NOTRACK_FALSE_POSITIVE);
382 char *fs_names = page_address(page);
1da177e4 383 char *p;
9361401e 384#ifdef CONFIG_BLOCK
1da177e4 385 char b[BDEVNAME_SIZE];
9361401e
DH
386#else
387 const char *b = name;
388#endif
1da177e4
LT
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 flags |= MS_RDONLY;
399 goto retry;
400 case -EINVAL:
401 continue;
402 }
403 /*
404 * Allow the user to distinguish between failed sys_open
405 * and bad superblock on root device.
dd2a345f 406 * and give them a list of the available devices
1da177e4 407 */
9361401e 408#ifdef CONFIG_BLOCK
1da177e4 409 __bdevname(ROOT_DEV, b);
9361401e 410#endif
0e0cb892
BW
411 printk("VFS: Cannot open root device \"%s\" or %s: error %d\n",
412 root_device_name, b, err);
dd2a345f 413 printk("Please append a correct \"root=\" boot option; here are the available partitions:\n");
1da177e4 414
dd2a345f 415 printk_all_partitions();
55dc7db7
TH
416#ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT
417 printk("DEBUG_BLOCK_EXT_DEVT is enabled, you need to specify "
418 "explicit textual name for \"root=\" boot option.\n");
419#endif
1da177e4
LT
420 panic("VFS: Unable to mount root fs on %s", b);
421 }
be6e028b 422
dd2a345f
DG
423 printk("List of all partitions:\n");
424 printk_all_partitions();
be6e028b
AW
425 printk("No filesystem could mount root, tried: ");
426 for (p = fs_names; *p; p += strlen(p)+1)
427 printk(" %s", p);
428 printk("\n");
9361401e
DH
429#ifdef CONFIG_BLOCK
430 __bdevname(ROOT_DEV, b);
431#endif
432 panic("VFS: Unable to mount root fs on %s", b);
1da177e4 433out:
a608ca21 434 put_page(page);
1da177e4
LT
435}
436
437#ifdef CONFIG_ROOT_NFS
43717c7d
CL
438
439#define NFSROOT_TIMEOUT_MIN 5
440#define NFSROOT_TIMEOUT_MAX 30
441#define NFSROOT_RETRY_MAX 5
442
1da177e4
LT
443static int __init mount_nfs_root(void)
444{
56463e50 445 char *root_dev, *root_data;
43717c7d
CL
446 unsigned int timeout;
447 int try, err;
1da177e4 448
43717c7d
CL
449 err = nfs_root_data(&root_dev, &root_data);
450 if (err != 0)
56463e50 451 return 0;
43717c7d
CL
452
453 /*
454 * The server or network may not be ready, so try several
455 * times. Stop after a few tries in case the client wants
456 * to fall back to other boot methods.
457 */
458 timeout = NFSROOT_TIMEOUT_MIN;
459 for (try = 1; ; try++) {
460 err = do_mount_root(root_dev, "nfs",
461 root_mountflags, root_data);
462 if (err == 0)
463 return 1;
464 if (try > NFSROOT_RETRY_MAX)
465 break;
466
467 /* Wait, in case the server refused us immediately */
468 ssleep(timeout);
469 timeout <<= 1;
470 if (timeout > NFSROOT_TIMEOUT_MAX)
471 timeout = NFSROOT_TIMEOUT_MAX;
472 }
473 return 0;
1da177e4
LT
474}
475#endif
476
477#if defined(CONFIG_BLK_DEV_RAM) || defined(CONFIG_BLK_DEV_FD)
478void __init change_floppy(char *fmt, ...)
479{
480 struct termios termios;
481 char buf[80];
482 char c;
483 int fd;
484 va_list args;
485 va_start(args, fmt);
486 vsprintf(buf, fmt, args);
487 va_end(args);
488 fd = sys_open("/dev/root", O_RDWR | O_NDELAY, 0);
489 if (fd >= 0) {
490 sys_ioctl(fd, FDEJECT, 0);
491 sys_close(fd);
492 }
493 printk(KERN_NOTICE "VFS: Insert %s and press ENTER\n", buf);
494 fd = sys_open("/dev/console", O_RDWR, 0);
495 if (fd >= 0) {
496 sys_ioctl(fd, TCGETS, (long)&termios);
497 termios.c_lflag &= ~ICANON;
498 sys_ioctl(fd, TCSETSF, (long)&termios);
499 sys_read(fd, &c, 1);
500 termios.c_lflag |= ICANON;
501 sys_ioctl(fd, TCSETSF, (long)&termios);
502 sys_close(fd);
503 }
504}
505#endif
506
507void __init mount_root(void)
508{
509#ifdef CONFIG_ROOT_NFS
377485f6 510 if (ROOT_DEV == Root_NFS) {
1da177e4
LT
511 if (mount_nfs_root())
512 return;
513
514 printk(KERN_ERR "VFS: Unable to mount root fs via NFS, trying floppy.\n");
515 ROOT_DEV = Root_FD0;
516 }
517#endif
518#ifdef CONFIG_BLK_DEV_FD
519 if (MAJOR(ROOT_DEV) == FLOPPY_MAJOR) {
520 /* rd_doload is 2 for a dual initrd/ramload setup */
521 if (rd_doload==2) {
522 if (rd_load_disk(1)) {
523 ROOT_DEV = Root_RAM1;
524 root_device_name = NULL;
525 }
526 } else
527 change_floppy("root floppy");
528 }
529#endif
9361401e 530#ifdef CONFIG_BLOCK
bdaf8529 531 create_dev("/dev/root", ROOT_DEV);
1da177e4 532 mount_block_root("/dev/root", root_mountflags);
9361401e 533#endif
1da177e4
LT
534}
535
536/*
537 * Prepare the namespace - decide what/where to mount, load ramdisks, etc.
538 */
539void __init prepare_namespace(void)
540{
541 int is_floppy;
542
1da177e4 543 if (root_delay) {
ca75b4d8 544 printk(KERN_INFO "Waiting %d sec before mounting root device...\n",
1da177e4
LT
545 root_delay);
546 ssleep(root_delay);
547 }
548
216773a7
AV
549 /*
550 * wait for the known devices to complete their probing
551 *
552 * Note: this is a potential source of long boot delays.
553 * For example, it is not atypical to wait 5 seconds here
554 * for the touchpad of a laptop to initialize.
555 */
556 wait_for_device_probe();
d779249e 557
1da177e4
LT
558 md_run_setup();
559
560 if (saved_root_name[0]) {
561 root_device_name = saved_root_name;
2d62f488
AH
562 if (!strncmp(root_device_name, "mtd", 3) ||
563 !strncmp(root_device_name, "ubi", 3)) {
e9482b43
JE
564 mount_block_root(root_device_name, root_mountflags);
565 goto out;
566 }
1da177e4
LT
567 ROOT_DEV = name_to_dev_t(root_device_name);
568 if (strncmp(root_device_name, "/dev/", 5) == 0)
569 root_device_name += 5;
570 }
571
1da177e4
LT
572 if (initrd_load())
573 goto out;
574
cc1ed754
PO
575 /* wait for any asynchronous scanning to complete */
576 if ((ROOT_DEV == 0) && root_wait) {
577 printk(KERN_INFO "Waiting for root device %s...\n",
578 saved_root_name);
579 while (driver_probe_done() != 0 ||
580 (ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
581 msleep(100);
216773a7 582 async_synchronize_full();
cc1ed754
PO
583 }
584
585 is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;
586
1da177e4
LT
587 if (is_floppy && rd_doload && rd_load_disk(0))
588 ROOT_DEV = Root_RAM0;
589
590 mount_root();
591out:
2b2af54a 592 devtmpfs_mount("dev");
1da177e4 593 sys_mount(".", "/", NULL, MS_MOVE, NULL);
c67e5382 594 sys_chroot(".");
1da177e4 595}
57f150a5 596
6e19eded 597static bool is_tmpfs;
57f150a5
RL
598static struct dentry *rootfs_mount(struct file_system_type *fs_type,
599 int flags, const char *dev_name, void *data)
600{
601 static unsigned long once;
6e19eded 602 void *fill = ramfs_fill_super;
57f150a5
RL
603
604 if (test_and_set_bit(0, &once))
605 return ERR_PTR(-ENODEV);
606
6e19eded
RL
607 if (IS_ENABLED(CONFIG_TMPFS) && is_tmpfs)
608 fill = shmem_fill_super;
609
610 return mount_nodev(fs_type, flags, data, fill);
57f150a5
RL
611}
612
613static struct file_system_type rootfs_fs_type = {
614 .name = "rootfs",
615 .mount = rootfs_mount,
616 .kill_sb = kill_litter_super,
617};
618
619int __init init_rootfs(void)
620{
621 int err = register_filesystem(&rootfs_fs_type);
622
623 if (err)
624 return err;
625
6e19eded
RL
626 if (IS_ENABLED(CONFIG_TMPFS) && !saved_root_name[0] &&
627 (!root_fs_names || strstr(root_fs_names, "tmpfs"))) {
16203a7a 628 err = shmem_init();
6e19eded
RL
629 is_tmpfs = true;
630 } else {
16203a7a 631 err = init_ramfs_fs();
6e19eded 632 }
16203a7a 633
57f150a5
RL
634 if (err)
635 unregister_filesystem(&rootfs_fs_type);
636
637 return err;
638}