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/seq_file.h>
28 #include <linux/string.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>
40 #include <linux/magic.h>
44 #include "transaction.h"
45 #include "btrfs_inode.h"
47 #include "print-tree.h"
52 #include "compression.h"
54 static const struct super_operations btrfs_super_ops;
56 static void btrfs_put_super(struct super_block *sb)
58 struct btrfs_root *root = btrfs_sb(sb);
61 ret = close_ctree(root);
66 Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow,
67 Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
68 Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl,
69 Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit,
73 static match_table_t tokens = {
74 {Opt_degraded, "degraded"},
75 {Opt_subvol, "subvol=%s"},
76 {Opt_device, "device=%s"},
77 {Opt_nodatasum, "nodatasum"},
78 {Opt_nodatacow, "nodatacow"},
79 {Opt_nobarrier, "nobarrier"},
80 {Opt_max_extent, "max_extent=%s"},
81 {Opt_max_inline, "max_inline=%s"},
82 {Opt_alloc_start, "alloc_start=%s"},
83 {Opt_thread_pool, "thread_pool=%d"},
84 {Opt_compress, "compress"},
86 {Opt_ssd_spread, "ssd_spread"},
89 {Opt_notreelog, "notreelog"},
90 {Opt_flushoncommit, "flushoncommit"},
91 {Opt_ratio, "metadata_ratio=%d"},
92 {Opt_discard, "discard"},
96 u64 btrfs_parse_size(char *str)
103 res = simple_strtoul(str, &end, 10);
107 last = tolower(last);
122 * Regular mount options parser. Everything that is needed only when
123 * reading in a new superblock is parsed here.
125 int btrfs_parse_options(struct btrfs_root *root, char *options)
127 struct btrfs_fs_info *info = root->fs_info;
128 substring_t args[MAX_OPT_ARGS];
129 char *p, *num, *orig;
137 * strsep changes the string, duplicate it because parse_options
140 options = kstrdup(options, GFP_NOFS);
146 while ((p = strsep(&options, ",")) != NULL) {
151 token = match_token(p, tokens, args);
154 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
155 btrfs_set_opt(info->mount_opt, DEGRADED);
160 * These are parsed by btrfs_parse_early_options
161 * and can be happily ignored here.
165 printk(KERN_INFO "btrfs: setting nodatasum\n");
166 btrfs_set_opt(info->mount_opt, NODATASUM);
169 printk(KERN_INFO "btrfs: setting nodatacow\n");
170 btrfs_set_opt(info->mount_opt, NODATACOW);
171 btrfs_set_opt(info->mount_opt, NODATASUM);
174 printk(KERN_INFO "btrfs: use compression\n");
175 btrfs_set_opt(info->mount_opt, COMPRESS);
178 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
179 btrfs_set_opt(info->mount_opt, SSD);
182 printk(KERN_INFO "btrfs: use spread ssd "
183 "allocation scheme\n");
184 btrfs_set_opt(info->mount_opt, SSD);
185 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
188 printk(KERN_INFO "btrfs: not using ssd allocation "
190 btrfs_set_opt(info->mount_opt, NOSSD);
191 btrfs_clear_opt(info->mount_opt, SSD);
192 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
195 printk(KERN_INFO "btrfs: turning off barriers\n");
196 btrfs_set_opt(info->mount_opt, NOBARRIER);
198 case Opt_thread_pool:
200 match_int(&args[0], &intarg);
202 info->thread_pool_size = intarg;
203 printk(KERN_INFO "btrfs: thread pool %d\n",
204 info->thread_pool_size);
208 num = match_strdup(&args[0]);
210 info->max_extent = btrfs_parse_size(num);
213 info->max_extent = max_t(u64,
214 info->max_extent, root->sectorsize);
215 printk(KERN_INFO "btrfs: max_extent at %llu\n",
216 (unsigned long long)info->max_extent);
220 num = match_strdup(&args[0]);
222 info->max_inline = btrfs_parse_size(num);
225 if (info->max_inline) {
226 info->max_inline = max_t(u64,
230 printk(KERN_INFO "btrfs: max_inline at %llu\n",
231 (unsigned long long)info->max_inline);
234 case Opt_alloc_start:
235 num = match_strdup(&args[0]);
237 info->alloc_start = btrfs_parse_size(num);
240 "btrfs: allocations start at %llu\n",
241 (unsigned long long)info->alloc_start);
245 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
248 printk(KERN_INFO "btrfs: disabling tree log\n");
249 btrfs_set_opt(info->mount_opt, NOTREELOG);
251 case Opt_flushoncommit:
252 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
253 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
257 match_int(&args[0], &intarg);
259 info->metadata_ratio = intarg;
260 printk(KERN_INFO "btrfs: metadata ratio %d\n",
261 info->metadata_ratio);
265 btrfs_set_opt(info->mount_opt, DISCARD);
268 printk(KERN_INFO "btrfs: unrecognized mount option "
282 * Parse mount options that are required early in the mount process.
284 * All other options will be parsed on much later in the mount process and
285 * only when we need to allocate a new super block.
287 static int btrfs_parse_early_options(const char *options, fmode_t flags,
288 void *holder, char **subvol_name,
289 struct btrfs_fs_devices **fs_devices)
291 substring_t args[MAX_OPT_ARGS];
299 * strsep changes the string, duplicate it because parse_options
302 opts = kstrdup(options, GFP_KERNEL);
306 while ((p = strsep(&opts, ",")) != NULL) {
311 token = match_token(p, tokens, args);
314 *subvol_name = match_strdup(&args[0]);
317 error = btrfs_scan_one_device(match_strdup(&args[0]),
318 flags, holder, fs_devices);
331 * If no subvolume name is specified we use the default one. Allocate
332 * a copy of the string "." here so that code later in the
333 * mount path doesn't care if it's the default volume or another one.
336 *subvol_name = kstrdup(".", GFP_KERNEL);
343 static int btrfs_fill_super(struct super_block *sb,
344 struct btrfs_fs_devices *fs_devices,
345 void *data, int silent)
348 struct dentry *root_dentry;
349 struct btrfs_super_block *disk_super;
350 struct btrfs_root *tree_root;
351 struct btrfs_key key;
354 sb->s_maxbytes = MAX_LFS_FILESIZE;
355 sb->s_magic = BTRFS_SUPER_MAGIC;
356 sb->s_op = &btrfs_super_ops;
357 sb->s_export_op = &btrfs_export_ops;
358 sb->s_xattr = btrfs_xattr_handlers;
360 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
361 sb->s_flags |= MS_POSIXACL;
364 tree_root = open_ctree(sb, fs_devices, (char *)data);
366 if (IS_ERR(tree_root)) {
367 printk("btrfs: open_ctree failed\n");
368 return PTR_ERR(tree_root);
370 sb->s_fs_info = tree_root;
371 disk_super = &tree_root->fs_info->super_copy;
373 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
374 key.type = BTRFS_INODE_ITEM_KEY;
376 inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root);
378 err = PTR_ERR(inode);
382 root_dentry = d_alloc_root(inode);
389 /* this does the super kobj at the same time */
390 err = btrfs_sysfs_add_super(tree_root->fs_info);
395 sb->s_root = root_dentry;
397 save_mount_options(sb, data);
401 close_ctree(tree_root);
405 int btrfs_sync_fs(struct super_block *sb, int wait)
407 struct btrfs_trans_handle *trans;
408 struct btrfs_root *root = btrfs_sb(sb);
412 filemap_flush(root->fs_info->btree_inode->i_mapping);
416 btrfs_start_delalloc_inodes(root, 0);
417 btrfs_wait_ordered_extents(root, 0, 0);
419 trans = btrfs_start_transaction(root, 1);
420 ret = btrfs_commit_transaction(trans, root);
424 static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
426 struct btrfs_root *root = btrfs_sb(vfs->mnt_sb);
427 struct btrfs_fs_info *info = root->fs_info;
429 if (btrfs_test_opt(root, DEGRADED))
430 seq_puts(seq, ",degraded");
431 if (btrfs_test_opt(root, NODATASUM))
432 seq_puts(seq, ",nodatasum");
433 if (btrfs_test_opt(root, NODATACOW))
434 seq_puts(seq, ",nodatacow");
435 if (btrfs_test_opt(root, NOBARRIER))
436 seq_puts(seq, ",nobarrier");
437 if (info->max_extent != (u64)-1)
438 seq_printf(seq, ",max_extent=%llu",
439 (unsigned long long)info->max_extent);
440 if (info->max_inline != 8192 * 1024)
441 seq_printf(seq, ",max_inline=%llu",
442 (unsigned long long)info->max_inline);
443 if (info->alloc_start != 0)
444 seq_printf(seq, ",alloc_start=%llu",
445 (unsigned long long)info->alloc_start);
446 if (info->thread_pool_size != min_t(unsigned long,
447 num_online_cpus() + 2, 8))
448 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
449 if (btrfs_test_opt(root, COMPRESS))
450 seq_puts(seq, ",compress");
451 if (btrfs_test_opt(root, NOSSD))
452 seq_puts(seq, ",nossd");
453 if (btrfs_test_opt(root, SSD_SPREAD))
454 seq_puts(seq, ",ssd_spread");
455 else if (btrfs_test_opt(root, SSD))
456 seq_puts(seq, ",ssd");
457 if (btrfs_test_opt(root, NOTREELOG))
458 seq_puts(seq, ",notreelog");
459 if (btrfs_test_opt(root, FLUSHONCOMMIT))
460 seq_puts(seq, ",flushoncommit");
461 if (btrfs_test_opt(root, DISCARD))
462 seq_puts(seq, ",discard");
463 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
464 seq_puts(seq, ",noacl");
468 static int btrfs_test_super(struct super_block *s, void *data)
470 struct btrfs_fs_devices *test_fs_devices = data;
471 struct btrfs_root *root = btrfs_sb(s);
473 return root->fs_info->fs_devices == test_fs_devices;
477 * Find a superblock for the given device / mount point.
479 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
480 * for multiple device setup. Make sure to keep it in sync.
482 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
483 const char *dev_name, void *data, struct vfsmount *mnt)
485 char *subvol_name = NULL;
486 struct block_device *bdev = NULL;
487 struct super_block *s;
489 struct btrfs_fs_devices *fs_devices = NULL;
490 fmode_t mode = FMODE_READ;
493 if (!(flags & MS_RDONLY))
496 error = btrfs_parse_early_options(data, mode, fs_type,
497 &subvol_name, &fs_devices);
501 error = btrfs_scan_one_device(dev_name, mode, fs_type, &fs_devices);
503 goto error_free_subvol_name;
505 error = btrfs_open_devices(fs_devices, mode, fs_type);
507 goto error_free_subvol_name;
509 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
511 goto error_close_devices;
514 bdev = fs_devices->latest_bdev;
515 s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
520 if ((flags ^ s->s_flags) & MS_RDONLY) {
521 deactivate_locked_super(s);
523 goto error_close_devices;
526 btrfs_close_devices(fs_devices);
528 char b[BDEVNAME_SIZE];
531 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
532 error = btrfs_fill_super(s, fs_devices, data,
533 flags & MS_SILENT ? 1 : 0);
535 deactivate_locked_super(s);
536 goto error_free_subvol_name;
539 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
540 s->s_flags |= MS_ACTIVE;
543 if (!strcmp(subvol_name, "."))
544 root = dget(s->s_root);
546 mutex_lock(&s->s_root->d_inode->i_mutex);
547 root = lookup_one_len(subvol_name, s->s_root,
548 strlen(subvol_name));
549 mutex_unlock(&s->s_root->d_inode->i_mutex);
552 deactivate_locked_super(s);
553 error = PTR_ERR(root);
554 goto error_free_subvol_name;
556 if (!root->d_inode) {
558 deactivate_locked_super(s);
560 goto error_free_subvol_name;
565 mnt->mnt_root = root;
573 btrfs_close_devices(fs_devices);
574 error_free_subvol_name:
579 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
581 struct btrfs_root *root = btrfs_sb(sb);
584 ret = btrfs_parse_options(root, data);
588 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
591 if (*flags & MS_RDONLY) {
592 sb->s_flags |= MS_RDONLY;
594 ret = btrfs_commit_super(root);
597 if (root->fs_info->fs_devices->rw_devices == 0)
600 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
603 /* recover relocation */
604 ret = btrfs_recover_relocation(root);
607 ret = btrfs_cleanup_fs_roots(root->fs_info);
610 sb->s_flags &= ~MS_RDONLY;
616 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
618 struct btrfs_root *root = btrfs_sb(dentry->d_sb);
619 struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
620 int bits = dentry->d_sb->s_blocksize_bits;
621 __be32 *fsid = (__be32 *)root->fs_info->fsid;
623 buf->f_namelen = BTRFS_NAME_LEN;
624 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
625 buf->f_bfree = buf->f_blocks -
626 (btrfs_super_bytes_used(disk_super) >> bits);
627 buf->f_bavail = buf->f_bfree;
628 buf->f_bsize = dentry->d_sb->s_blocksize;
629 buf->f_type = BTRFS_SUPER_MAGIC;
631 /* We treat it as constant endianness (it doesn't matter _which_)
632 because we want the fsid to come out the same whether mounted
633 on a big-endian or little-endian host */
634 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
635 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
636 /* Mask in the root object ID too, to disambiguate subvols */
637 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
638 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
643 static struct file_system_type btrfs_fs_type = {
644 .owner = THIS_MODULE,
646 .get_sb = btrfs_get_sb,
647 .kill_sb = kill_anon_super,
648 .fs_flags = FS_REQUIRES_DEV,
652 * used by btrfsctl to scan devices when no FS is mounted
654 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
657 struct btrfs_ioctl_vol_args *vol;
658 struct btrfs_fs_devices *fs_devices;
661 if (!capable(CAP_SYS_ADMIN))
664 vol = memdup_user((void __user *)arg, sizeof(*vol));
669 case BTRFS_IOC_SCAN_DEV:
670 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
671 &btrfs_fs_type, &fs_devices);
679 static int btrfs_freeze(struct super_block *sb)
681 struct btrfs_root *root = btrfs_sb(sb);
682 mutex_lock(&root->fs_info->transaction_kthread_mutex);
683 mutex_lock(&root->fs_info->cleaner_mutex);
687 static int btrfs_unfreeze(struct super_block *sb)
689 struct btrfs_root *root = btrfs_sb(sb);
690 mutex_unlock(&root->fs_info->cleaner_mutex);
691 mutex_unlock(&root->fs_info->transaction_kthread_mutex);
695 static const struct super_operations btrfs_super_ops = {
696 .drop_inode = btrfs_drop_inode,
697 .delete_inode = btrfs_delete_inode,
698 .put_super = btrfs_put_super,
699 .sync_fs = btrfs_sync_fs,
700 .show_options = btrfs_show_options,
701 .write_inode = btrfs_write_inode,
702 .dirty_inode = btrfs_dirty_inode,
703 .alloc_inode = btrfs_alloc_inode,
704 .destroy_inode = btrfs_destroy_inode,
705 .statfs = btrfs_statfs,
706 .remount_fs = btrfs_remount,
707 .freeze_fs = btrfs_freeze,
708 .unfreeze_fs = btrfs_unfreeze,
711 static const struct file_operations btrfs_ctl_fops = {
712 .unlocked_ioctl = btrfs_control_ioctl,
713 .compat_ioctl = btrfs_control_ioctl,
714 .owner = THIS_MODULE,
717 static struct miscdevice btrfs_misc = {
718 .minor = MISC_DYNAMIC_MINOR,
719 .name = "btrfs-control",
720 .fops = &btrfs_ctl_fops
723 static int btrfs_interface_init(void)
725 return misc_register(&btrfs_misc);
728 static void btrfs_interface_exit(void)
730 if (misc_deregister(&btrfs_misc) < 0)
731 printk(KERN_INFO "misc_deregister failed for control device");
734 static int __init init_btrfs_fs(void)
738 err = btrfs_init_sysfs();
742 err = btrfs_init_cachep();
746 err = extent_io_init();
750 err = extent_map_init();
754 err = btrfs_interface_init();
756 goto free_extent_map;
758 err = register_filesystem(&btrfs_fs_type);
760 goto unregister_ioctl;
762 printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
766 btrfs_interface_exit();
772 btrfs_destroy_cachep();
778 static void __exit exit_btrfs_fs(void)
780 btrfs_destroy_cachep();
783 btrfs_interface_exit();
784 unregister_filesystem(&btrfs_fs_type);
786 btrfs_cleanup_fs_uuids();
790 module_init(init_btrfs_fs)
791 module_exit(exit_btrfs_fs)
793 MODULE_LICENSE("GPL");