2 * Copyright (C) 2007 Oracle. All rights reserved.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
19 #include <linux/blkdev.h>
20 #include <linux/module.h>
21 #include <linux/buffer_head.h>
23 #include <linux/pagemap.h>
24 #include <linux/highmem.h>
25 #include <linux/time.h>
26 #include <linux/init.h>
27 #include <linux/string.h>
28 #include <linux/smp_lock.h>
29 #include <linux/backing-dev.h>
30 #include <linux/mount.h>
31 #include <linux/mpage.h>
32 #include <linux/swap.h>
33 #include <linux/writeback.h>
34 #include <linux/statfs.h>
35 #include <linux/compat.h>
36 #include <linux/parser.h>
37 #include <linux/ctype.h>
38 #include <linux/namei.h>
39 #include <linux/miscdevice.h>
42 #include "transaction.h"
43 #include "btrfs_inode.h"
45 #include "print-tree.h"
50 #include "compression.h"
52 #define BTRFS_SUPER_MAGIC 0x9123683E
54 static struct super_operations btrfs_super_ops;
56 static void btrfs_put_super (struct super_block * sb)
58 struct btrfs_root *root = btrfs_sb(sb);
59 struct btrfs_fs_info *fs = root->fs_info;
62 ret = close_ctree(root);
64 printk("close ctree returns %d\n", ret);
66 btrfs_sysfs_del_super(fs);
71 Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
72 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
73 Opt_ssd, Opt_thread_pool, Opt_noacl, Opt_compress, Opt_err,
76 static match_table_t tokens = {
77 {Opt_degraded, "degraded"},
78 {Opt_subvol, "subvol=%s"},
79 {Opt_device, "device=%s"},
80 {Opt_nodatasum, "nodatasum"},
81 {Opt_nodatacow, "nodatacow"},
82 {Opt_nobarrier, "nobarrier"},
83 {Opt_max_extent, "max_extent=%s"},
84 {Opt_max_inline, "max_inline=%s"},
85 {Opt_alloc_start, "alloc_start=%s"},
86 {Opt_thread_pool, "thread_pool=%d"},
87 {Opt_compress, "compress"},
93 u64 btrfs_parse_size(char *str)
100 res = simple_strtoul(str, &end, 10);
104 last = tolower(last);
119 * Regular mount options parser. Everything that is needed only when
120 * reading in a new superblock is parsed here.
122 int btrfs_parse_options(struct btrfs_root *root, char *options)
124 struct btrfs_fs_info *info = root->fs_info;
125 substring_t args[MAX_OPT_ARGS];
133 * strsep changes the string, duplicate it because parse_options
136 options = kstrdup(options, GFP_NOFS);
141 while ((p = strsep(&options, ",")) != NULL) {
146 token = match_token(p, tokens, args);
149 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
150 btrfs_set_opt(info->mount_opt, DEGRADED);
155 * These are parsed by btrfs_parse_early_options
156 * and can be happily ignored here.
160 printk(KERN_INFO "btrfs: setting nodatacsum\n");
161 btrfs_set_opt(info->mount_opt, NODATASUM);
164 printk(KERN_INFO "btrfs: setting nodatacow\n");
165 btrfs_set_opt(info->mount_opt, NODATACOW);
166 btrfs_set_opt(info->mount_opt, NODATASUM);
169 printk(KERN_INFO "btrfs: use compression\n");
170 btrfs_set_opt(info->mount_opt, COMPRESS);
173 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
174 btrfs_set_opt(info->mount_opt, SSD);
177 printk(KERN_INFO "btrfs: turning off barriers\n");
178 btrfs_set_opt(info->mount_opt, NOBARRIER);
180 case Opt_thread_pool:
182 match_int(&args[0], &intarg);
184 info->thread_pool_size = intarg;
185 printk(KERN_INFO "btrfs: thread pool %d\n",
186 info->thread_pool_size);
190 num = match_strdup(&args[0]);
192 info->max_extent = btrfs_parse_size(num);
195 info->max_extent = max_t(u64,
196 info->max_extent, root->sectorsize);
197 printk(KERN_INFO "btrfs: max_extent at %llu\n",
202 num = match_strdup(&args[0]);
204 info->max_inline = btrfs_parse_size(num);
207 if (info->max_inline) {
208 info->max_inline = max_t(u64,
212 printk(KERN_INFO "btrfs: max_inline at %llu\n",
216 case Opt_alloc_start:
217 num = match_strdup(&args[0]);
219 info->alloc_start = btrfs_parse_size(num);
222 "btrfs: allocations start at %llu\n",
227 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
238 * Parse mount options that are required early in the mount process.
240 * All other options will be parsed on much later in the mount process and
241 * only when we need to allocate a new super block.
243 static int btrfs_parse_early_options(const char *options, int flags,
244 void *holder, char **subvol_name,
245 struct btrfs_fs_devices **fs_devices)
247 substring_t args[MAX_OPT_ARGS];
255 * strsep changes the string, duplicate it because parse_options
258 opts = kstrdup(options, GFP_KERNEL);
262 while ((p = strsep(&opts, ",")) != NULL) {
267 token = match_token(p, tokens, args);
270 *subvol_name = match_strdup(&args[0]);
273 error = btrfs_scan_one_device(match_strdup(&args[0]),
274 flags, holder, fs_devices);
287 * If no subvolume name is specified we use the default one. Allocate
288 * a copy of the string "default" here so that code later in the
289 * mount path doesn't care if it's the default volume or another one.
292 *subvol_name = kstrdup("default", GFP_KERNEL);
299 static int btrfs_fill_super(struct super_block * sb,
300 struct btrfs_fs_devices *fs_devices,
301 void * data, int silent)
303 struct inode * inode;
304 struct dentry * root_dentry;
305 struct btrfs_super_block *disk_super;
306 struct btrfs_root *tree_root;
307 struct btrfs_inode *bi;
310 sb->s_maxbytes = MAX_LFS_FILESIZE;
311 sb->s_magic = BTRFS_SUPER_MAGIC;
312 sb->s_op = &btrfs_super_ops;
313 sb->s_export_op = &btrfs_export_ops;
314 sb->s_xattr = btrfs_xattr_handlers;
316 sb->s_flags |= MS_POSIXACL;
318 tree_root = open_ctree(sb, fs_devices, (char *)data);
320 if (IS_ERR(tree_root)) {
321 printk("btrfs: open_ctree failed\n");
322 return PTR_ERR(tree_root);
324 sb->s_fs_info = tree_root;
325 disk_super = &tree_root->fs_info->super_copy;
326 inode = btrfs_iget_locked(sb, btrfs_super_root_dir(disk_super),
329 bi->location.objectid = inode->i_ino;
330 bi->location.offset = 0;
331 bi->root = tree_root;
333 btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
339 if (inode->i_state & I_NEW) {
340 btrfs_read_locked_inode(inode);
341 unlock_new_inode(inode);
344 root_dentry = d_alloc_root(inode);
351 /* this does the super kobj at the same time */
352 err = btrfs_sysfs_add_super(tree_root->fs_info);
356 sb->s_root = root_dentry;
358 save_mount_options(sb, data);
362 close_ctree(tree_root);
366 int btrfs_sync_fs(struct super_block *sb, int wait)
368 struct btrfs_trans_handle *trans;
369 struct btrfs_root *root;
373 if (sb->s_flags & MS_RDONLY)
378 filemap_flush(root->fs_info->btree_inode->i_mapping);
382 btrfs_start_delalloc_inodes(root);
383 btrfs_wait_ordered_extents(root, 0);
385 btrfs_clean_old_snapshots(root);
386 trans = btrfs_start_transaction(root, 1);
387 ret = btrfs_commit_transaction(trans, root);
392 static void btrfs_write_super(struct super_block *sb)
397 static int btrfs_test_super(struct super_block *s, void *data)
399 struct btrfs_fs_devices *test_fs_devices = data;
400 struct btrfs_root *root = btrfs_sb(s);
402 return root->fs_info->fs_devices == test_fs_devices;
406 * Find a superblock for the given device / mount point.
408 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
409 * for multiple device setup. Make sure to keep it in sync.
411 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
412 const char *dev_name, void *data, struct vfsmount *mnt)
414 char *subvol_name = NULL;
415 struct block_device *bdev = NULL;
416 struct super_block *s;
418 struct btrfs_fs_devices *fs_devices = NULL;
421 error = btrfs_parse_early_options(data, flags, fs_type,
422 &subvol_name, &fs_devices);
426 error = btrfs_scan_one_device(dev_name, flags, fs_type, &fs_devices);
428 goto error_free_subvol_name;
430 error = btrfs_open_devices(fs_devices, flags, fs_type);
432 goto error_free_subvol_name;
434 bdev = fs_devices->latest_bdev;
435 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
440 if ((flags ^ s->s_flags) & MS_RDONLY) {
441 up_write(&s->s_umount);
444 goto error_close_devices;
448 char b[BDEVNAME_SIZE];
451 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
452 error = btrfs_fill_super(s, fs_devices, data,
453 flags & MS_SILENT ? 1 : 0);
455 up_write(&s->s_umount);
460 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
461 s->s_flags |= MS_ACTIVE;
464 if (!strcmp(subvol_name, "."))
465 root = dget(s->s_root);
467 mutex_lock(&s->s_root->d_inode->i_mutex);
468 root = lookup_one_len(subvol_name, s->s_root, strlen(subvol_name));
469 mutex_unlock(&s->s_root->d_inode->i_mutex);
471 up_write(&s->s_umount);
473 error = PTR_ERR(root);
476 if (!root->d_inode) {
478 up_write(&s->s_umount);
486 mnt->mnt_root = root;
494 btrfs_close_devices(fs_devices);
495 error_free_subvol_name:
501 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
503 struct btrfs_root *root = btrfs_sb(sb);
506 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
509 if (*flags & MS_RDONLY) {
510 sb->s_flags |= MS_RDONLY;
512 ret = btrfs_commit_super(root);
515 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
518 ret = btrfs_cleanup_reloc_trees(root);
521 ret = btrfs_cleanup_fs_roots(root->fs_info);
524 sb->s_flags &= ~MS_RDONLY;
530 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
532 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
533 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
534 int bits = dentry->d_sb->s_blocksize_bits;
535 __be32 *fsid = (__be32 *)root->fs_info->fsid;
537 buf->f_namelen = BTRFS_NAME_LEN;
538 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
539 buf->f_bfree = buf->f_blocks -
540 (btrfs_super_bytes_used(disk_super) >> bits);
541 buf->f_bavail = buf->f_bfree;
542 buf->f_bsize = dentry->d_sb->s_blocksize;
543 buf->f_type = BTRFS_SUPER_MAGIC;
544 /* We treat it as constant endianness (it doesn't matter _which_)
545 because we want the fsid to come out the same whether mounted
546 on a big-endian or little-endian host */
547 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
548 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
549 /* Mask in the root object ID too, to disambiguate subvols */
550 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
551 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
556 static struct file_system_type btrfs_fs_type = {
557 .owner = THIS_MODULE,
559 .get_sb = btrfs_get_sb,
560 .kill_sb = kill_anon_super,
561 .fs_flags = FS_REQUIRES_DEV,
565 * used by btrfsctl to scan devices when no FS is mounted
567 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
570 struct btrfs_ioctl_vol_args *vol;
571 struct btrfs_fs_devices *fs_devices;
575 vol = kmalloc(sizeof(*vol), GFP_KERNEL);
576 if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
580 len = strnlen(vol->name, BTRFS_PATH_NAME_MAX);
582 case BTRFS_IOC_SCAN_DEV:
583 ret = btrfs_scan_one_device(vol->name, MS_RDONLY,
584 &btrfs_fs_type, &fs_devices);
592 static void btrfs_write_super_lockfs(struct super_block *sb)
594 struct btrfs_root *root = btrfs_sb(sb);
595 mutex_lock(&root->fs_info->transaction_kthread_mutex);
596 mutex_lock(&root->fs_info->cleaner_mutex);
599 static void btrfs_unlockfs(struct super_block *sb)
601 struct btrfs_root *root = btrfs_sb(sb);
602 mutex_unlock(&root->fs_info->cleaner_mutex);
603 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
606 static struct super_operations btrfs_super_ops = {
607 .delete_inode = btrfs_delete_inode,
608 .put_super = btrfs_put_super,
609 .write_super = btrfs_write_super,
610 .sync_fs = btrfs_sync_fs,
611 .show_options = generic_show_options,
612 .write_inode = btrfs_write_inode,
613 .dirty_inode = btrfs_dirty_inode,
614 .alloc_inode = btrfs_alloc_inode,
615 .destroy_inode = btrfs_destroy_inode,
616 .statfs = btrfs_statfs,
617 .remount_fs = btrfs_remount,
618 .write_super_lockfs = btrfs_write_super_lockfs,
619 .unlockfs = btrfs_unlockfs,
622 static const struct file_operations btrfs_ctl_fops = {
623 .unlocked_ioctl = btrfs_control_ioctl,
624 .compat_ioctl = btrfs_control_ioctl,
625 .owner = THIS_MODULE,
628 static struct miscdevice btrfs_misc = {
629 .minor = MISC_DYNAMIC_MINOR,
630 .name = "btrfs-control",
631 .fops = &btrfs_ctl_fops
634 static int btrfs_interface_init(void)
636 return misc_register(&btrfs_misc);
639 void btrfs_interface_exit(void)
641 if (misc_deregister(&btrfs_misc) < 0)
642 printk("misc_deregister failed for control device");
645 static int __init init_btrfs_fs(void)
649 err = btrfs_init_sysfs();
653 err = btrfs_init_cachep();
657 err = extent_io_init();
661 err = extent_map_init();
665 err = btrfs_interface_init();
667 goto free_extent_map;
669 err = register_filesystem(&btrfs_fs_type);
671 goto unregister_ioctl;
673 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
677 btrfs_interface_exit();
683 btrfs_destroy_cachep();
689 static void __exit exit_btrfs_fs(void)
691 btrfs_destroy_cachep();
694 btrfs_interface_exit();
695 unregister_filesystem(&btrfs_fs_type);
697 btrfs_cleanup_fs_uuids();
701 module_init(init_btrfs_fs)
702 module_exit(exit_btrfs_fs)
704 MODULE_LICENSE("GPL");