1 // SPDX-License-Identifier: GPL-2.0-only
2 #include <linux/module.h>
3 #include <linux/sched.h>
4 #include <linux/ctype.h>
7 #include <linux/suspend.h>
8 #include <linux/root_dev.h>
9 #include <linux/security.h>
10 #include <linux/delay.h>
11 #include <linux/genhd.h>
12 #include <linux/mount.h>
13 #include <linux/device.h>
14 #include <linux/init.h>
16 #include <linux/initrd.h>
17 #include <linux/async.h>
18 #include <linux/fs_struct.h>
19 #include <linux/slab.h>
20 #include <linux/ramfs.h>
21 #include <linux/shmem_fs.h>
23 #include <linux/nfs_fs.h>
24 #include <linux/nfs_fs_sb.h>
25 #include <linux/nfs_mount.h>
26 #include <uapi/linux/mount.h>
28 #include "do_mounts.h"
30 int __initdata rd_doload; /* 1 = load RAM disk, 0 = don't load */
32 int root_mountflags = MS_RDONLY | MS_SILENT;
33 static char * __initdata root_device_name;
34 static char __initdata saved_root_name[64];
39 static int __init load_ramdisk(char *str)
41 rd_doload = simple_strtol(str,NULL,0) & 3;
44 __setup("load_ramdisk=", load_ramdisk);
46 static int __init readonly(char *str)
50 root_mountflags |= MS_RDONLY;
54 static int __init readwrite(char *str)
58 root_mountflags &= ~MS_RDONLY;
62 __setup("ro", readonly);
63 __setup("rw", readwrite);
72 * match_dev_by_uuid - callback for finding a partition using its uuid
73 * @dev: device passed in by the caller
74 * @data: opaque pointer to the desired struct uuidcmp to match
76 * Returns 1 if the device matches, and 0 otherwise.
78 static int match_dev_by_uuid(struct device *dev, const void *data)
80 const struct uuidcmp *cmp = data;
81 struct hd_struct *part = dev_to_part(dev);
86 if (strncasecmp(cmp->uuid, part->info->uuid, cmp->len))
96 * devt_from_partuuid - looks up the dev_t of a partition by its UUID
97 * @uuid_str: char array containing ascii UUID
99 * The function will return the first partition which contains a matching
100 * UUID value in its partition_meta_info struct. This does not search
101 * by filesystem UUIDs.
103 * If @uuid_str is followed by a "/PARTNROFF=%d", then the number will be
104 * extracted and used as an offset from the partition identified by the UUID.
106 * Returns the matching dev_t on success or 0 on failure.
108 static dev_t devt_from_partuuid(const char *uuid_str)
112 struct device *dev = NULL;
113 struct gendisk *disk;
114 struct hd_struct *part;
116 bool clear_root_wait = false;
121 slash = strchr(uuid_str, '/');
122 /* Check for optional partition number offset attributes. */
125 /* Explicitly fail on poor PARTUUID syntax. */
126 if (sscanf(slash + 1,
127 "PARTNROFF=%d%c", &offset, &c) != 1) {
128 clear_root_wait = true;
131 cmp.len = slash - uuid_str;
133 cmp.len = strlen(uuid_str);
137 clear_root_wait = true;
141 dev = class_find_device(&block_class, NULL, &cmp,
148 /* Attempt to find the partition by offset. */
153 disk = part_to_disk(dev_to_part(dev));
154 part = disk_get_part(disk, dev_to_part(dev)->partno + offset);
156 res = part_devt(part);
157 put_device(part_to_dev(part));
163 if (clear_root_wait) {
164 pr_err("VFS: PARTUUID= is invalid.\n"
165 "Expected PARTUUID=<valid-uuid-id>[/PARTNROFF=%%d]\n");
167 pr_err("Disabling rootwait; root= is invalid.\n");
174 * match_dev_by_label - callback for finding a partition using its label
175 * @dev: device passed in by the caller
176 * @data: opaque pointer to the label to match
178 * Returns 1 if the device matches, and 0 otherwise.
180 static int match_dev_by_label(struct device *dev, const void *data)
182 const char *label = data;
183 struct hd_struct *part = dev_to_part(dev);
185 if (part->info && !strcmp(label, part->info->volname))
193 * Convert a name into device number. We accept the following variants:
195 * 1) <hex_major><hex_minor> device number in hexadecimal represents itself
196 * no leading 0x, for example b302.
197 * 2) /dev/nfs represents Root_NFS (0xff)
198 * 3) /dev/<disk_name> represents the device number of disk
199 * 4) /dev/<disk_name><decimal> represents the device number
200 * of partition - device number of disk plus the partition number
201 * 5) /dev/<disk_name>p<decimal> - same as the above, that form is
202 * used when disk name of partitioned disk ends on a digit.
203 * 6) PARTUUID=00112233-4455-6677-8899-AABBCCDDEEFF representing the
204 * unique id of a partition if the partition table provides it.
205 * The UUID may be either an EFI/GPT UUID, or refer to an MSDOS
206 * partition using the format SSSSSSSS-PP, where SSSSSSSS is a zero-
207 * filled hex representation of the 32-bit "NT disk signature", and PP
208 * is a zero-filled hex representation of the 1-based partition number.
209 * 7) PARTUUID=<UUID>/PARTNROFF=<int> to select a partition in relation to
210 * a partition with a known unique id.
211 * 8) <major>:<minor> major and minor number of the device separated by
213 * 9) PARTLABEL=<name> with name being the GPT partition label.
214 * MSDOS partitions do not support labels!
215 * 10) /dev/cifs represents Root_CIFS (0xfe)
217 * If name doesn't have fall into the categories above, we return (0,0).
218 * block_class is used to check if something is a disk name. If the disk
219 * name contains slashes, the device name has them replaced with
223 dev_t name_to_dev_t(const char *name)
231 if (strncmp(name, "PARTUUID=", 9) == 0) {
233 res = devt_from_partuuid(name);
237 } else if (strncmp(name, "PARTLABEL=", 10) == 0) {
240 dev = class_find_device(&block_class, NULL, name + 10,
241 &match_dev_by_label);
251 if (strncmp(name, "/dev/", 5) != 0) {
252 unsigned maj, min, offset;
255 if ((sscanf(name, "%u:%u%c", &maj, &min, &dummy) == 2) ||
256 (sscanf(name, "%u:%u:%u:%c", &maj, &min, &offset, &dummy) == 3)) {
257 res = MKDEV(maj, min);
258 if (maj != MAJOR(res) || min != MINOR(res))
261 res = new_decode_dev(simple_strtoul(name, &p, 16));
270 if (strcmp(name, "nfs") == 0)
273 if (strcmp(name, "cifs") == 0)
276 if (strcmp(name, "ram") == 0)
279 if (strlen(name) > 31)
285 res = blk_lookup_devt(s, 0);
290 * try non-existent, but valid partition, which may only exist
291 * after revalidating the disk, like partitioned md devices
293 while (p > s && isdigit(p[-1]))
295 if (p == s || !*p || *p == '0')
298 /* try disk name without <part number> */
299 part = simple_strtoul(p, NULL, 10);
301 res = blk_lookup_devt(s, part);
305 /* try disk name without p<part number> */
306 if (p < s + 2 || !isdigit(p[-2]) || p[-1] != 'p')
309 res = blk_lookup_devt(s, part);
318 EXPORT_SYMBOL_GPL(name_to_dev_t);
320 static int __init root_dev_setup(char *line)
322 strlcpy(saved_root_name, line, sizeof(saved_root_name));
326 __setup("root=", root_dev_setup);
328 static int __init rootwait_setup(char *str)
336 __setup("rootwait", rootwait_setup);
338 static char * __initdata root_mount_data;
339 static int __init root_data_setup(char *str)
341 root_mount_data = str;
345 static char * __initdata root_fs_names;
346 static int __init fs_names_setup(char *str)
352 static unsigned int __initdata root_delay;
353 static int __init root_delay_setup(char *str)
355 root_delay = simple_strtoul(str, NULL, 0);
359 __setup("rootflags=", root_data_setup);
360 __setup("rootfstype=", fs_names_setup);
361 __setup("rootdelay=", root_delay_setup);
363 static void __init get_fs_names(char *page)
368 strcpy(page, root_fs_names);
374 int len = get_filesystem_list(page);
378 for (p = page-1; p; p = next) {
379 next = strchr(++p, '\n');
382 while ((*s++ = *p++) != '\n')
390 static int __init do_mount_root(const char *name, const char *fs,
391 const int flags, const void *data)
393 struct super_block *s;
394 struct page *p = NULL;
395 char *data_page = NULL;
399 /* do_mount() requires a full page as fifth argument */
400 p = alloc_page(GFP_KERNEL);
403 data_page = page_address(p);
404 /* zero-pad. do_mount() will make sure it's terminated */
405 strncpy(data_page, data, PAGE_SIZE);
408 ret = do_mount(name, "/root", fs, flags, data_page);
413 s = current->fs->pwd.dentry->d_sb;
416 "VFS: Mounted root (%s filesystem)%s on device %u:%u.\n",
418 sb_rdonly(s) ? " readonly" : "",
419 MAJOR(ROOT_DEV), MINOR(ROOT_DEV));
427 void __init mount_block_root(char *name, int flags)
429 struct page *page = alloc_page(GFP_KERNEL);
430 char *fs_names = page_address(page);
432 char b[BDEVNAME_SIZE];
434 scnprintf(b, BDEVNAME_SIZE, "unknown-block(%u,%u)",
435 MAJOR(ROOT_DEV), MINOR(ROOT_DEV));
436 get_fs_names(fs_names);
438 for (p = fs_names; *p; p += strlen(p)+1) {
439 int err = do_mount_root(name, p, flags, root_mount_data);
448 * Allow the user to distinguish between failed sys_open
449 * and bad superblock on root device.
450 * and give them a list of the available devices
452 printk("VFS: Cannot open root device \"%s\" or %s: error %d\n",
453 root_device_name, b, err);
454 printk("Please append a correct \"root=\" boot option; here are the available partitions:\n");
456 printk_all_partitions();
457 #ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT
458 printk("DEBUG_BLOCK_EXT_DEVT is enabled, you need to specify "
459 "explicit textual name for \"root=\" boot option.\n");
461 panic("VFS: Unable to mount root fs on %s", b);
463 if (!(flags & SB_RDONLY)) {
468 printk("List of all partitions:\n");
469 printk_all_partitions();
470 printk("No filesystem could mount root, tried: ");
471 for (p = fs_names; *p; p += strlen(p)+1)
474 panic("VFS: Unable to mount root fs on %s", b);
479 #ifdef CONFIG_ROOT_NFS
481 #define NFSROOT_TIMEOUT_MIN 5
482 #define NFSROOT_TIMEOUT_MAX 30
483 #define NFSROOT_RETRY_MAX 5
485 static int __init mount_nfs_root(void)
487 char *root_dev, *root_data;
488 unsigned int timeout;
491 err = nfs_root_data(&root_dev, &root_data);
496 * The server or network may not be ready, so try several
497 * times. Stop after a few tries in case the client wants
498 * to fall back to other boot methods.
500 timeout = NFSROOT_TIMEOUT_MIN;
501 for (try = 1; ; try++) {
502 err = do_mount_root(root_dev, "nfs",
503 root_mountflags, root_data);
506 if (try > NFSROOT_RETRY_MAX)
509 /* Wait, in case the server refused us immediately */
512 if (timeout > NFSROOT_TIMEOUT_MAX)
513 timeout = NFSROOT_TIMEOUT_MAX;
519 #ifdef CONFIG_CIFS_ROOT
521 extern int cifs_root_data(char **dev, char **opts);
523 #define CIFSROOT_TIMEOUT_MIN 5
524 #define CIFSROOT_TIMEOUT_MAX 30
525 #define CIFSROOT_RETRY_MAX 5
527 static int __init mount_cifs_root(void)
529 char *root_dev, *root_data;
530 unsigned int timeout;
533 err = cifs_root_data(&root_dev, &root_data);
537 timeout = CIFSROOT_TIMEOUT_MIN;
538 for (try = 1; ; try++) {
539 err = do_mount_root(root_dev, "cifs", root_mountflags,
543 if (try > CIFSROOT_RETRY_MAX)
548 if (timeout > CIFSROOT_TIMEOUT_MAX)
549 timeout = CIFSROOT_TIMEOUT_MAX;
555 #if defined(CONFIG_BLK_DEV_RAM) || defined(CONFIG_BLK_DEV_FD)
556 void __init change_floppy(char *fmt, ...)
558 struct termios termios;
564 vsprintf(buf, fmt, args);
566 fd = ksys_open("/dev/root", O_RDWR | O_NDELAY, 0);
568 ksys_ioctl(fd, FDEJECT, 0);
571 printk(KERN_NOTICE "VFS: Insert %s and press ENTER\n", buf);
572 fd = ksys_open("/dev/console", O_RDWR, 0);
574 ksys_ioctl(fd, TCGETS, (long)&termios);
575 termios.c_lflag &= ~ICANON;
576 ksys_ioctl(fd, TCSETSF, (long)&termios);
577 ksys_read(fd, &c, 1);
578 termios.c_lflag |= ICANON;
579 ksys_ioctl(fd, TCSETSF, (long)&termios);
585 void __init mount_root(void)
587 #ifdef CONFIG_ROOT_NFS
588 if (ROOT_DEV == Root_NFS) {
589 if (mount_nfs_root())
592 printk(KERN_ERR "VFS: Unable to mount root fs via NFS, trying floppy.\n");
596 #ifdef CONFIG_CIFS_ROOT
597 if (ROOT_DEV == Root_CIFS) {
598 if (mount_cifs_root())
601 printk(KERN_ERR "VFS: Unable to mount root fs via SMB, trying floppy.\n");
605 #ifdef CONFIG_BLK_DEV_FD
606 if (MAJOR(ROOT_DEV) == FLOPPY_MAJOR) {
607 /* rd_doload is 2 for a dual initrd/ramload setup */
609 if (rd_load_disk(1)) {
610 ROOT_DEV = Root_RAM1;
611 root_device_name = NULL;
614 change_floppy("root floppy");
619 int err = create_dev("/dev/root", ROOT_DEV);
622 pr_emerg("Failed to create /dev/root: %d\n", err);
623 mount_block_root("/dev/root", root_mountflags);
629 * Prepare the namespace - decide what/where to mount, load ramdisks, etc.
631 void __init prepare_namespace(void)
636 printk(KERN_INFO "Waiting %d sec before mounting root device...\n",
642 * wait for the known devices to complete their probing
644 * Note: this is a potential source of long boot delays.
645 * For example, it is not atypical to wait 5 seconds here
646 * for the touchpad of a laptop to initialize.
648 wait_for_device_probe();
652 if (saved_root_name[0]) {
653 root_device_name = saved_root_name;
654 if (!strncmp(root_device_name, "mtd", 3) ||
655 !strncmp(root_device_name, "ubi", 3)) {
656 mount_block_root(root_device_name, root_mountflags);
659 ROOT_DEV = name_to_dev_t(root_device_name);
660 if (strncmp(root_device_name, "/dev/", 5) == 0)
661 root_device_name += 5;
667 /* wait for any asynchronous scanning to complete */
668 if ((ROOT_DEV == 0) && root_wait) {
669 printk(KERN_INFO "Waiting for root device %s...\n",
671 while (driver_probe_done() != 0 ||
672 (ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
674 async_synchronize_full();
677 is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;
679 if (is_floppy && rd_doload && rd_load_disk(0))
680 ROOT_DEV = Root_RAM0;
685 do_mount(".", "/", NULL, MS_MOVE, NULL);
689 static bool is_tmpfs;
690 static int rootfs_init_fs_context(struct fs_context *fc)
692 if (IS_ENABLED(CONFIG_TMPFS) && is_tmpfs)
693 return shmem_init_fs_context(fc);
695 return ramfs_init_fs_context(fc);
698 struct file_system_type rootfs_fs_type = {
700 .init_fs_context = rootfs_init_fs_context,
701 .kill_sb = kill_litter_super,
704 void __init init_rootfs(void)
706 if (IS_ENABLED(CONFIG_TMPFS) && !saved_root_name[0] &&
707 (!root_fs_names || strstr(root_fs_names, "tmpfs")))