xref: /linux/init/do_mounts.c (revision 0d456bad36d42d16022be045c8a53ddbb59ee478)
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 
30 #include <linux/nfs_fs.h>
31 #include <linux/nfs_fs_sb.h>
32 #include <linux/nfs_mount.h>
33 
34 #include "do_mounts.h"
35 
36 int __initdata rd_doload;	/* 1 = load RAM disk, 0 = don't load */
37 
38 int root_mountflags = MS_RDONLY | MS_SILENT;
39 static char * __initdata root_device_name;
40 static char __initdata saved_root_name[64];
41 static int root_wait;
42 
43 dev_t ROOT_DEV;
44 
45 static int __init load_ramdisk(char *str)
46 {
47 	rd_doload = simple_strtol(str,NULL,0) & 3;
48 	return 1;
49 }
50 __setup("load_ramdisk=", load_ramdisk);
51 
52 static int __init readonly(char *str)
53 {
54 	if (*str)
55 		return 0;
56 	root_mountflags |= MS_RDONLY;
57 	return 1;
58 }
59 
60 static int __init readwrite(char *str)
61 {
62 	if (*str)
63 		return 0;
64 	root_mountflags &= ~MS_RDONLY;
65 	return 1;
66 }
67 
68 __setup("ro", readonly);
69 __setup("rw", readwrite);
70 
71 #ifdef CONFIG_BLOCK
72 struct uuidcmp {
73 	const char *uuid;
74 	int len;
75 };
76 
77 /**
78  * match_dev_by_uuid - callback for finding a partition using its uuid
79  * @dev:	device passed in by the caller
80  * @data:	opaque pointer to the desired struct uuidcmp to match
81  *
82  * Returns 1 if the device matches, and 0 otherwise.
83  */
84 static int match_dev_by_uuid(struct device *dev, void *data)
85 {
86 	struct uuidcmp *cmp = data;
87 	struct hd_struct *part = dev_to_part(dev);
88 
89 	if (!part->info)
90 		goto no_match;
91 
92 	if (strncasecmp(cmp->uuid, part->info->uuid, cmp->len))
93 		goto no_match;
94 
95 	return 1;
96 no_match:
97 	return 0;
98 }
99 
100 
101 /**
102  * devt_from_partuuid - looks up the dev_t of a partition by its UUID
103  * @uuid:	char array containing ascii UUID
104  *
105  * The function will return the first partition which contains a matching
106  * UUID value in its partition_meta_info struct.  This does not search
107  * by filesystem UUIDs.
108  *
109  * If @uuid is followed by a "/PARTNROFF=%d", then the number will be
110  * extracted and used as an offset from the partition identified by the UUID.
111  *
112  * Returns the matching dev_t on success or 0 on failure.
113  */
114 static dev_t devt_from_partuuid(const char *uuid_str)
115 {
116 	dev_t res = 0;
117 	struct uuidcmp cmp;
118 	struct device *dev = NULL;
119 	struct gendisk *disk;
120 	struct hd_struct *part;
121 	int offset = 0;
122 	bool clear_root_wait = false;
123 	char *slash;
124 
125 	cmp.uuid = uuid_str;
126 
127 	slash = strchr(uuid_str, '/');
128 	/* Check for optional partition number offset attributes. */
129 	if (slash) {
130 		char c = 0;
131 		/* Explicitly fail on poor PARTUUID syntax. */
132 		if (sscanf(slash + 1,
133 			   "PARTNROFF=%d%c", &offset, &c) != 1) {
134 			clear_root_wait = true;
135 			goto done;
136 		}
137 		cmp.len = slash - uuid_str;
138 	} else {
139 		cmp.len = strlen(uuid_str);
140 	}
141 
142 	if (!cmp.len) {
143 		clear_root_wait = true;
144 		goto done;
145 	}
146 
147 	dev = class_find_device(&block_class, NULL, &cmp,
148 				&match_dev_by_uuid);
149 	if (!dev)
150 		goto done;
151 
152 	res = dev->devt;
153 
154 	/* Attempt to find the partition by offset. */
155 	if (!offset)
156 		goto no_offset;
157 
158 	res = 0;
159 	disk = part_to_disk(dev_to_part(dev));
160 	part = disk_get_part(disk, dev_to_part(dev)->partno + offset);
161 	if (part) {
162 		res = part_devt(part);
163 		put_device(part_to_dev(part));
164 	}
165 
166 no_offset:
167 	put_device(dev);
168 done:
169 	if (clear_root_wait) {
170 		pr_err("VFS: PARTUUID= is invalid.\n"
171 		       "Expected PARTUUID=<valid-uuid-id>[/PARTNROFF=%%d]\n");
172 		if (root_wait)
173 			pr_err("Disabling rootwait; root= is invalid.\n");
174 		root_wait = 0;
175 	}
176 	return res;
177 }
178 #endif
179 
180 /*
181  *	Convert a name into device number.  We accept the following variants:
182  *
183  *	1) device number in hexadecimal	represents itself
184  *	2) /dev/nfs represents Root_NFS (0xff)
185  *	3) /dev/<disk_name> represents the device number of disk
186  *	4) /dev/<disk_name><decimal> represents the device number
187  *         of partition - device number of disk plus the partition number
188  *	5) /dev/<disk_name>p<decimal> - same as the above, that form is
189  *	   used when disk name of partitioned disk ends on a digit.
190  *	6) PARTUUID=00112233-4455-6677-8899-AABBCCDDEEFF representing the
191  *	   unique id of a partition if the partition table provides it.
192  *	   The UUID may be either an EFI/GPT UUID, or refer to an MSDOS
193  *	   partition using the format SSSSSSSS-PP, where SSSSSSSS is a zero-
194  *	   filled hex representation of the 32-bit "NT disk signature", and PP
195  *	   is a zero-filled hex representation of the 1-based partition number.
196  *	7) PARTUUID=<UUID>/PARTNROFF=<int> to select a partition in relation to
197  *	   a partition with a known unique id.
198  *
199  *	If name doesn't have fall into the categories above, we return (0,0).
200  *	block_class is used to check if something is a disk name. If the disk
201  *	name contains slashes, the device name has them replaced with
202  *	bangs.
203  */
204 
205 dev_t name_to_dev_t(char *name)
206 {
207 	char s[32];
208 	char *p;
209 	dev_t res = 0;
210 	int part;
211 
212 #ifdef CONFIG_BLOCK
213 	if (strncmp(name, "PARTUUID=", 9) == 0) {
214 		name += 9;
215 		res = devt_from_partuuid(name);
216 		if (!res)
217 			goto fail;
218 		goto done;
219 	}
220 #endif
221 
222 	if (strncmp(name, "/dev/", 5) != 0) {
223 		unsigned maj, min;
224 
225 		if (sscanf(name, "%u:%u", &maj, &min) == 2) {
226 			res = MKDEV(maj, min);
227 			if (maj != MAJOR(res) || min != MINOR(res))
228 				goto fail;
229 		} else {
230 			res = new_decode_dev(simple_strtoul(name, &p, 16));
231 			if (*p)
232 				goto fail;
233 		}
234 		goto done;
235 	}
236 
237 	name += 5;
238 	res = Root_NFS;
239 	if (strcmp(name, "nfs") == 0)
240 		goto done;
241 	res = Root_RAM0;
242 	if (strcmp(name, "ram") == 0)
243 		goto done;
244 
245 	if (strlen(name) > 31)
246 		goto fail;
247 	strcpy(s, name);
248 	for (p = s; *p; p++)
249 		if (*p == '/')
250 			*p = '!';
251 	res = blk_lookup_devt(s, 0);
252 	if (res)
253 		goto done;
254 
255 	/*
256 	 * try non-existent, but valid partition, which may only exist
257 	 * after revalidating the disk, like partitioned md devices
258 	 */
259 	while (p > s && isdigit(p[-1]))
260 		p--;
261 	if (p == s || !*p || *p == '0')
262 		goto fail;
263 
264 	/* try disk name without <part number> */
265 	part = simple_strtoul(p, NULL, 10);
266 	*p = '\0';
267 	res = blk_lookup_devt(s, part);
268 	if (res)
269 		goto done;
270 
271 	/* try disk name without p<part number> */
272 	if (p < s + 2 || !isdigit(p[-2]) || p[-1] != 'p')
273 		goto fail;
274 	p[-1] = '\0';
275 	res = blk_lookup_devt(s, part);
276 	if (res)
277 		goto done;
278 
279 fail:
280 	return 0;
281 done:
282 	return res;
283 }
284 
285 static int __init root_dev_setup(char *line)
286 {
287 	strlcpy(saved_root_name, line, sizeof(saved_root_name));
288 	return 1;
289 }
290 
291 __setup("root=", root_dev_setup);
292 
293 static int __init rootwait_setup(char *str)
294 {
295 	if (*str)
296 		return 0;
297 	root_wait = 1;
298 	return 1;
299 }
300 
301 __setup("rootwait", rootwait_setup);
302 
303 static char * __initdata root_mount_data;
304 static int __init root_data_setup(char *str)
305 {
306 	root_mount_data = str;
307 	return 1;
308 }
309 
310 static char * __initdata root_fs_names;
311 static int __init fs_names_setup(char *str)
312 {
313 	root_fs_names = str;
314 	return 1;
315 }
316 
317 static unsigned int __initdata root_delay;
318 static int __init root_delay_setup(char *str)
319 {
320 	root_delay = simple_strtoul(str, NULL, 0);
321 	return 1;
322 }
323 
324 __setup("rootflags=", root_data_setup);
325 __setup("rootfstype=", fs_names_setup);
326 __setup("rootdelay=", root_delay_setup);
327 
328 static void __init get_fs_names(char *page)
329 {
330 	char *s = page;
331 
332 	if (root_fs_names) {
333 		strcpy(page, root_fs_names);
334 		while (*s++) {
335 			if (s[-1] == ',')
336 				s[-1] = '\0';
337 		}
338 	} else {
339 		int len = get_filesystem_list(page);
340 		char *p, *next;
341 
342 		page[len] = '\0';
343 		for (p = page-1; p; p = next) {
344 			next = strchr(++p, '\n');
345 			if (*p++ != '\t')
346 				continue;
347 			while ((*s++ = *p++) != '\n')
348 				;
349 			s[-1] = '\0';
350 		}
351 	}
352 	*s = '\0';
353 }
354 
355 static int __init do_mount_root(char *name, char *fs, int flags, void *data)
356 {
357 	struct super_block *s;
358 	int err = sys_mount(name, "/root", fs, flags, data);
359 	if (err)
360 		return err;
361 
362 	sys_chdir("/root");
363 	s = current->fs->pwd.dentry->d_sb;
364 	ROOT_DEV = s->s_dev;
365 	printk(KERN_INFO
366 	       "VFS: Mounted root (%s filesystem)%s on device %u:%u.\n",
367 	       s->s_type->name,
368 	       s->s_flags & MS_RDONLY ?  " readonly" : "",
369 	       MAJOR(ROOT_DEV), MINOR(ROOT_DEV));
370 	return 0;
371 }
372 
373 void __init mount_block_root(char *name, int flags)
374 {
375 	struct page *page = alloc_page(GFP_KERNEL |
376 					__GFP_NOTRACK_FALSE_POSITIVE);
377 	char *fs_names = page_address(page);
378 	char *p;
379 #ifdef CONFIG_BLOCK
380 	char b[BDEVNAME_SIZE];
381 #else
382 	const char *b = name;
383 #endif
384 
385 	get_fs_names(fs_names);
386 retry:
387 	for (p = fs_names; *p; p += strlen(p)+1) {
388 		int err = do_mount_root(name, p, flags, root_mount_data);
389 		switch (err) {
390 			case 0:
391 				goto out;
392 			case -EACCES:
393 				flags |= MS_RDONLY;
394 				goto retry;
395 			case -EINVAL:
396 				continue;
397 		}
398 	        /*
399 		 * Allow the user to distinguish between failed sys_open
400 		 * and bad superblock on root device.
401 		 * and give them a list of the available devices
402 		 */
403 #ifdef CONFIG_BLOCK
404 		__bdevname(ROOT_DEV, b);
405 #endif
406 		printk("VFS: Cannot open root device \"%s\" or %s: error %d\n",
407 				root_device_name, b, err);
408 		printk("Please append a correct \"root=\" boot option; here are the available partitions:\n");
409 
410 		printk_all_partitions();
411 #ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT
412 		printk("DEBUG_BLOCK_EXT_DEVT is enabled, you need to specify "
413 		       "explicit textual name for \"root=\" boot option.\n");
414 #endif
415 		panic("VFS: Unable to mount root fs on %s", b);
416 	}
417 
418 	printk("List of all partitions:\n");
419 	printk_all_partitions();
420 	printk("No filesystem could mount root, tried: ");
421 	for (p = fs_names; *p; p += strlen(p)+1)
422 		printk(" %s", p);
423 	printk("\n");
424 #ifdef CONFIG_BLOCK
425 	__bdevname(ROOT_DEV, b);
426 #endif
427 	panic("VFS: Unable to mount root fs on %s", b);
428 out:
429 	put_page(page);
430 }
431 
432 #ifdef CONFIG_ROOT_NFS
433 
434 #define NFSROOT_TIMEOUT_MIN	5
435 #define NFSROOT_TIMEOUT_MAX	30
436 #define NFSROOT_RETRY_MAX	5
437 
438 static int __init mount_nfs_root(void)
439 {
440 	char *root_dev, *root_data;
441 	unsigned int timeout;
442 	int try, err;
443 
444 	err = nfs_root_data(&root_dev, &root_data);
445 	if (err != 0)
446 		return 0;
447 
448 	/*
449 	 * The server or network may not be ready, so try several
450 	 * times.  Stop after a few tries in case the client wants
451 	 * to fall back to other boot methods.
452 	 */
453 	timeout = NFSROOT_TIMEOUT_MIN;
454 	for (try = 1; ; try++) {
455 		err = do_mount_root(root_dev, "nfs",
456 					root_mountflags, root_data);
457 		if (err == 0)
458 			return 1;
459 		if (try > NFSROOT_RETRY_MAX)
460 			break;
461 
462 		/* Wait, in case the server refused us immediately */
463 		ssleep(timeout);
464 		timeout <<= 1;
465 		if (timeout > NFSROOT_TIMEOUT_MAX)
466 			timeout = NFSROOT_TIMEOUT_MAX;
467 	}
468 	return 0;
469 }
470 #endif
471 
472 #if defined(CONFIG_BLK_DEV_RAM) || defined(CONFIG_BLK_DEV_FD)
473 void __init change_floppy(char *fmt, ...)
474 {
475 	struct termios termios;
476 	char buf[80];
477 	char c;
478 	int fd;
479 	va_list args;
480 	va_start(args, fmt);
481 	vsprintf(buf, fmt, args);
482 	va_end(args);
483 	fd = sys_open("/dev/root", O_RDWR | O_NDELAY, 0);
484 	if (fd >= 0) {
485 		sys_ioctl(fd, FDEJECT, 0);
486 		sys_close(fd);
487 	}
488 	printk(KERN_NOTICE "VFS: Insert %s and press ENTER\n", buf);
489 	fd = sys_open("/dev/console", O_RDWR, 0);
490 	if (fd >= 0) {
491 		sys_ioctl(fd, TCGETS, (long)&termios);
492 		termios.c_lflag &= ~ICANON;
493 		sys_ioctl(fd, TCSETSF, (long)&termios);
494 		sys_read(fd, &c, 1);
495 		termios.c_lflag |= ICANON;
496 		sys_ioctl(fd, TCSETSF, (long)&termios);
497 		sys_close(fd);
498 	}
499 }
500 #endif
501 
502 void __init mount_root(void)
503 {
504 #ifdef CONFIG_ROOT_NFS
505 	if (ROOT_DEV == Root_NFS) {
506 		if (mount_nfs_root())
507 			return;
508 
509 		printk(KERN_ERR "VFS: Unable to mount root fs via NFS, trying floppy.\n");
510 		ROOT_DEV = Root_FD0;
511 	}
512 #endif
513 #ifdef CONFIG_BLK_DEV_FD
514 	if (MAJOR(ROOT_DEV) == FLOPPY_MAJOR) {
515 		/* rd_doload is 2 for a dual initrd/ramload setup */
516 		if (rd_doload==2) {
517 			if (rd_load_disk(1)) {
518 				ROOT_DEV = Root_RAM1;
519 				root_device_name = NULL;
520 			}
521 		} else
522 			change_floppy("root floppy");
523 	}
524 #endif
525 #ifdef CONFIG_BLOCK
526 	create_dev("/dev/root", ROOT_DEV);
527 	mount_block_root("/dev/root", root_mountflags);
528 #endif
529 }
530 
531 /*
532  * Prepare the namespace - decide what/where to mount, load ramdisks, etc.
533  */
534 void __init prepare_namespace(void)
535 {
536 	int is_floppy;
537 
538 	if (root_delay) {
539 		printk(KERN_INFO "Waiting %dsec before mounting root device...\n",
540 		       root_delay);
541 		ssleep(root_delay);
542 	}
543 
544 	/*
545 	 * wait for the known devices to complete their probing
546 	 *
547 	 * Note: this is a potential source of long boot delays.
548 	 * For example, it is not atypical to wait 5 seconds here
549 	 * for the touchpad of a laptop to initialize.
550 	 */
551 	wait_for_device_probe();
552 
553 	md_run_setup();
554 
555 	if (saved_root_name[0]) {
556 		root_device_name = saved_root_name;
557 		if (!strncmp(root_device_name, "mtd", 3) ||
558 		    !strncmp(root_device_name, "ubi", 3)) {
559 			mount_block_root(root_device_name, root_mountflags);
560 			goto out;
561 		}
562 		ROOT_DEV = name_to_dev_t(root_device_name);
563 		if (strncmp(root_device_name, "/dev/", 5) == 0)
564 			root_device_name += 5;
565 	}
566 
567 	if (initrd_load())
568 		goto out;
569 
570 	/* wait for any asynchronous scanning to complete */
571 	if ((ROOT_DEV == 0) && root_wait) {
572 		printk(KERN_INFO "Waiting for root device %s...\n",
573 			saved_root_name);
574 		while (driver_probe_done() != 0 ||
575 			(ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
576 			msleep(100);
577 		async_synchronize_full();
578 	}
579 
580 	is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;
581 
582 	if (is_floppy && rd_doload && rd_load_disk(0))
583 		ROOT_DEV = Root_RAM0;
584 
585 	mount_root();
586 out:
587 	devtmpfs_mount("dev");
588 	sys_mount(".", "/", NULL, MS_MOVE, NULL);
589 	sys_chroot(".");
590 }
591