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.
8 #warning "Sparse checking disabled for this file"
11 #include <linux/module.h>
12 #include <linux/sched.h>
13 #include <linux/ctype.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>
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>
31 #include <linux/nfs_fs.h>
32 #include <linux/nfs_fs_sb.h>
33 #include <linux/nfs_mount.h>
35 #include "do_mounts.h"
37 int __initdata rd_doload; /* 1 = load RAM disk, 0 = don't load */
39 int root_mountflags = MS_RDONLY | MS_SILENT;
40 static char * __initdata root_device_name;
41 static char __initdata saved_root_name[64];
46 static int __init load_ramdisk(char *str)
48 rd_doload = simple_strtol(str,NULL,0) & 3;
51 __setup("load_ramdisk=", load_ramdisk);
53 static int __init readonly(char *str)
57 root_mountflags |= MS_RDONLY;
61 static int __init readwrite(char *str)
65 root_mountflags &= ~MS_RDONLY;
69 __setup("ro", readonly);
70 __setup("rw", readwrite);
79 * match_dev_by_uuid - callback for finding a partition using its uuid
80 * @dev: device passed in by the caller
81 * @data: opaque pointer to the desired struct uuidcmp to match
83 * Returns 1 if the device matches, and 0 otherwise.
85 static int match_dev_by_uuid(struct device *dev, const void *data)
87 const struct uuidcmp *cmp = data;
88 struct hd_struct *part = dev_to_part(dev);
93 if (strncasecmp(cmp->uuid, part->info->uuid, cmp->len))
103 * devt_from_partuuid - looks up the dev_t of a partition by its UUID
104 * @uuid: char array containing ascii UUID
106 * The function will return the first partition which contains a matching
107 * UUID value in its partition_meta_info struct. This does not search
108 * by filesystem UUIDs.
110 * If @uuid is followed by a "/PARTNROFF=%d", then the number will be
111 * extracted and used as an offset from the partition identified by the UUID.
113 * Returns the matching dev_t on success or 0 on failure.
115 static dev_t devt_from_partuuid(const char *uuid_str)
119 struct device *dev = NULL;
120 struct gendisk *disk;
121 struct hd_struct *part;
123 bool clear_root_wait = false;
128 slash = strchr(uuid_str, '/');
129 /* Check for optional partition number offset attributes. */
132 /* Explicitly fail on poor PARTUUID syntax. */
133 if (sscanf(slash + 1,
134 "PARTNROFF=%d%c", &offset, &c) != 1) {
135 clear_root_wait = true;
138 cmp.len = slash - uuid_str;
140 cmp.len = strlen(uuid_str);
144 clear_root_wait = true;
148 dev = class_find_device(&block_class, NULL, &cmp,
155 /* Attempt to find the partition by offset. */
160 disk = part_to_disk(dev_to_part(dev));
161 part = disk_get_part(disk, dev_to_part(dev)->partno + offset);
163 res = part_devt(part);
164 put_device(part_to_dev(part));
170 if (clear_root_wait) {
171 pr_err("VFS: PARTUUID= is invalid.\n"
172 "Expected PARTUUID=<valid-uuid-id>[/PARTNROFF=%%d]\n");
174 pr_err("Disabling rootwait; root= is invalid.\n");
182 * Convert a name into device number. We accept the following variants:
184 * 1) device number in hexadecimal represents itself
185 * 2) /dev/nfs represents Root_NFS (0xff)
186 * 3) /dev/<disk_name> represents the device number of disk
187 * 4) /dev/<disk_name><decimal> represents the device number
188 * of partition - device number of disk plus the partition number
189 * 5) /dev/<disk_name>p<decimal> - same as the above, that form is
190 * used when disk name of partitioned disk ends on a digit.
191 * 6) PARTUUID=00112233-4455-6677-8899-AABBCCDDEEFF representing the
192 * unique id of a partition if the partition table provides it.
193 * The UUID may be either an EFI/GPT UUID, or refer to an MSDOS
194 * partition using the format SSSSSSSS-PP, where SSSSSSSS is a zero-
195 * filled hex representation of the 32-bit "NT disk signature", and PP
196 * is a zero-filled hex representation of the 1-based partition number.
197 * 7) PARTUUID=<UUID>/PARTNROFF=<int> to select a partition in relation to
198 * a partition with a known unique id.
200 * If name doesn't have fall into the categories above, we return (0,0).
201 * block_class is used to check if something is a disk name. If the disk
202 * name contains slashes, the device name has them replaced with
206 dev_t name_to_dev_t(char *name)
214 if (strncmp(name, "PARTUUID=", 9) == 0) {
216 res = devt_from_partuuid(name);
223 if (strncmp(name, "/dev/", 5) != 0) {
226 if (sscanf(name, "%u:%u", &maj, &min) == 2) {
227 res = MKDEV(maj, min);
228 if (maj != MAJOR(res) || min != MINOR(res))
231 res = new_decode_dev(simple_strtoul(name, &p, 16));
240 if (strcmp(name, "nfs") == 0)
243 if (strcmp(name, "ram") == 0)
246 if (strlen(name) > 31)
252 res = blk_lookup_devt(s, 0);
257 * try non-existent, but valid partition, which may only exist
258 * after revalidating the disk, like partitioned md devices
260 while (p > s && isdigit(p[-1]))
262 if (p == s || !*p || *p == '0')
265 /* try disk name without <part number> */
266 part = simple_strtoul(p, NULL, 10);
268 res = blk_lookup_devt(s, part);
272 /* try disk name without p<part number> */
273 if (p < s + 2 || !isdigit(p[-2]) || p[-1] != 'p')
276 res = blk_lookup_devt(s, part);
286 static int __init root_dev_setup(char *line)
288 strlcpy(saved_root_name, line, sizeof(saved_root_name));
292 __setup("root=", root_dev_setup);
294 static int __init rootwait_setup(char *str)
302 __setup("rootwait", rootwait_setup);
304 static char * __initdata root_mount_data;
305 static int __init root_data_setup(char *str)
307 root_mount_data = str;
311 static char * __initdata root_fs_names;
312 static int __init fs_names_setup(char *str)
318 static unsigned int __initdata root_delay;
319 static int __init root_delay_setup(char *str)
321 root_delay = simple_strtoul(str, NULL, 0);
325 __setup("rootflags=", root_data_setup);
326 __setup("rootfstype=", fs_names_setup);
327 __setup("rootdelay=", root_delay_setup);
329 static void __init get_fs_names(char *page)
334 strcpy(page, root_fs_names);
340 int len = get_filesystem_list(page);
344 for (p = page-1; p; p = next) {
345 next = strchr(++p, '\n');
348 while ((*s++ = *p++) != '\n')
356 static int __init do_mount_root(char *name, char *fs, int flags, void *data)
358 struct super_block *s;
359 int err = sys_mount(name, "/root", fs, flags, data);
364 s = current->fs->pwd.dentry->d_sb;
367 "VFS: Mounted root (%s filesystem)%s on device %u:%u.\n",
369 s->s_flags & MS_RDONLY ? " readonly" : "",
370 MAJOR(ROOT_DEV), MINOR(ROOT_DEV));
374 void __init mount_block_root(char *name, int flags)
376 struct page *page = alloc_page(GFP_KERNEL |
377 __GFP_NOTRACK_FALSE_POSITIVE);
378 char *fs_names = page_address(page);
381 char b[BDEVNAME_SIZE];
383 const char *b = name;
386 get_fs_names(fs_names);
388 for (p = fs_names; *p; p += strlen(p)+1) {
389 int err = do_mount_root(name, p, flags, root_mount_data);
400 * Allow the user to distinguish between failed sys_open
401 * and bad superblock on root device.
402 * and give them a list of the available devices
405 __bdevname(ROOT_DEV, b);
407 printk("VFS: Cannot open root device \"%s\" or %s: error %d\n",
408 root_device_name, b, err);
409 printk("Please append a correct \"root=\" boot option; here are the available partitions:\n");
411 printk_all_partitions();
412 #ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT
413 printk("DEBUG_BLOCK_EXT_DEVT is enabled, you need to specify "
414 "explicit textual name for \"root=\" boot option.\n");
416 panic("VFS: Unable to mount root fs on %s", b);
419 printk("List of all partitions:\n");
420 printk_all_partitions();
421 printk("No filesystem could mount root, tried: ");
422 for (p = fs_names; *p; p += strlen(p)+1)
426 __bdevname(ROOT_DEV, b);
428 panic("VFS: Unable to mount root fs on %s", b);
433 #ifdef CONFIG_ROOT_NFS
435 #define NFSROOT_TIMEOUT_MIN 5
436 #define NFSROOT_TIMEOUT_MAX 30
437 #define NFSROOT_RETRY_MAX 5
439 static int __init mount_nfs_root(void)
441 char *root_dev, *root_data;
442 unsigned int timeout;
445 err = nfs_root_data(&root_dev, &root_data);
450 * The server or network may not be ready, so try several
451 * times. Stop after a few tries in case the client wants
452 * to fall back to other boot methods.
454 timeout = NFSROOT_TIMEOUT_MIN;
455 for (try = 1; ; try++) {
456 err = do_mount_root(root_dev, "nfs",
457 root_mountflags, root_data);
460 if (try > NFSROOT_RETRY_MAX)
463 /* Wait, in case the server refused us immediately */
466 if (timeout > NFSROOT_TIMEOUT_MAX)
467 timeout = NFSROOT_TIMEOUT_MAX;
473 #if defined(CONFIG_BLK_DEV_RAM) || defined(CONFIG_BLK_DEV_FD)
474 void __init change_floppy(char *fmt, ...)
476 struct termios termios;
482 vsprintf(buf, fmt, args);
484 fd = sys_open("/dev/root", O_RDWR | O_NDELAY, 0);
486 sys_ioctl(fd, FDEJECT, 0);
489 printk(KERN_NOTICE "VFS: Insert %s and press ENTER\n", buf);
490 fd = sys_open("/dev/console", O_RDWR, 0);
492 sys_ioctl(fd, TCGETS, (long)&termios);
493 termios.c_lflag &= ~ICANON;
494 sys_ioctl(fd, TCSETSF, (long)&termios);
496 termios.c_lflag |= ICANON;
497 sys_ioctl(fd, TCSETSF, (long)&termios);
503 void __init mount_root(void)
505 #ifdef CONFIG_ROOT_NFS
506 if (ROOT_DEV == Root_NFS) {
507 if (mount_nfs_root())
510 printk(KERN_ERR "VFS: Unable to mount root fs via NFS, trying floppy.\n");
514 #ifdef CONFIG_BLK_DEV_FD
515 if (MAJOR(ROOT_DEV) == FLOPPY_MAJOR) {
516 /* rd_doload is 2 for a dual initrd/ramload setup */
518 if (rd_load_disk(1)) {
519 ROOT_DEV = Root_RAM1;
520 root_device_name = NULL;
523 change_floppy("root floppy");
527 create_dev("/dev/root", ROOT_DEV);
528 mount_block_root("/dev/root", root_mountflags);
533 * Prepare the namespace - decide what/where to mount, load ramdisks, etc.
535 void __init prepare_namespace(void)
540 printk(KERN_INFO "Waiting %d sec before mounting root device...\n",
546 * wait for the known devices to complete their probing
548 * Note: this is a potential source of long boot delays.
549 * For example, it is not atypical to wait 5 seconds here
550 * for the touchpad of a laptop to initialize.
552 wait_for_device_probe();
556 if (saved_root_name[0]) {
557 root_device_name = saved_root_name;
558 if (!strncmp(root_device_name, "mtd", 3) ||
559 !strncmp(root_device_name, "ubi", 3)) {
560 mount_block_root(root_device_name, root_mountflags);
563 ROOT_DEV = name_to_dev_t(root_device_name);
564 if (strncmp(root_device_name, "/dev/", 5) == 0)
565 root_device_name += 5;
571 /* wait for any asynchronous scanning to complete */
572 if ((ROOT_DEV == 0) && root_wait) {
573 printk(KERN_INFO "Waiting for root device %s...\n",
575 while (driver_probe_done() != 0 ||
576 (ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
578 async_synchronize_full();
581 is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;
583 if (is_floppy && rd_doload && rd_load_disk(0))
584 ROOT_DEV = Root_RAM0;
588 devtmpfs_mount("dev");
589 sys_mount(".", "/", NULL, MS_MOVE, NULL);
593 static struct dentry *rootfs_mount(struct file_system_type *fs_type,
594 int flags, const char *dev_name, void *data)
596 static unsigned long once;
598 if (test_and_set_bit(0, &once))
599 return ERR_PTR(-ENODEV);
601 return mount_nodev(fs_type, flags, data, ramfs_fill_super);
604 static struct file_system_type rootfs_fs_type = {
606 .mount = rootfs_mount,
607 .kill_sb = kill_litter_super,
610 int __init init_rootfs(void)
612 int err = register_filesystem(&rootfs_fs_type);
617 err = init_ramfs_fs();
619 unregister_filesystem(&rootfs_fs_type);