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_compress_force
, Opt_notreelog
, Opt_ratio
,
74 static match_table_t tokens
= {
75 {Opt_degraded
, "degraded"},
76 {Opt_subvol
, "subvol=%s"},
77 {Opt_device
, "device=%s"},
78 {Opt_nodatasum
, "nodatasum"},
79 {Opt_nodatacow
, "nodatacow"},
80 {Opt_nobarrier
, "nobarrier"},
81 {Opt_max_extent
, "max_extent=%s"},
82 {Opt_max_inline
, "max_inline=%s"},
83 {Opt_alloc_start
, "alloc_start=%s"},
84 {Opt_thread_pool
, "thread_pool=%d"},
85 {Opt_compress
, "compress"},
86 {Opt_compress_force
, "compress-force"},
88 {Opt_ssd_spread
, "ssd_spread"},
91 {Opt_notreelog
, "notreelog"},
92 {Opt_flushoncommit
, "flushoncommit"},
93 {Opt_ratio
, "metadata_ratio=%d"},
94 {Opt_discard
, "discard"},
98 u64
btrfs_parse_size(char *str
)
105 res
= simple_strtoul(str
, &end
, 10);
109 last
= tolower(last
);
124 * Regular mount options parser. Everything that is needed only when
125 * reading in a new superblock is parsed here.
127 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
129 struct btrfs_fs_info
*info
= root
->fs_info
;
130 substring_t args
[MAX_OPT_ARGS
];
139 * strsep changes the string, duplicate it because parse_options
142 options
= kstrdup(options
, GFP_NOFS
);
147 while ((p
= strsep(&options
, ",")) != NULL
) {
152 token
= match_token(p
, tokens
, args
);
155 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
156 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
161 * These are parsed by btrfs_parse_early_options
162 * and can be happily ignored here.
166 printk(KERN_INFO
"btrfs: setting nodatasum\n");
167 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
170 printk(KERN_INFO
"btrfs: setting nodatacow\n");
171 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
172 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
175 printk(KERN_INFO
"btrfs: use compression\n");
176 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
178 case Opt_compress_force
:
179 printk(KERN_INFO
"btrfs: forcing compression\n");
180 btrfs_set_opt(info
->mount_opt
, FORCE_COMPRESS
);
181 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
184 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
185 btrfs_set_opt(info
->mount_opt
, SSD
);
188 printk(KERN_INFO
"btrfs: use spread ssd "
189 "allocation scheme\n");
190 btrfs_set_opt(info
->mount_opt
, SSD
);
191 btrfs_set_opt(info
->mount_opt
, SSD_SPREAD
);
194 printk(KERN_INFO
"btrfs: not using ssd allocation "
196 btrfs_set_opt(info
->mount_opt
, NOSSD
);
197 btrfs_clear_opt(info
->mount_opt
, SSD
);
198 btrfs_clear_opt(info
->mount_opt
, SSD_SPREAD
);
201 printk(KERN_INFO
"btrfs: turning off barriers\n");
202 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
204 case Opt_thread_pool
:
206 match_int(&args
[0], &intarg
);
208 info
->thread_pool_size
= intarg
;
209 printk(KERN_INFO
"btrfs: thread pool %d\n",
210 info
->thread_pool_size
);
214 num
= match_strdup(&args
[0]);
216 info
->max_extent
= btrfs_parse_size(num
);
219 info
->max_extent
= max_t(u64
,
220 info
->max_extent
, root
->sectorsize
);
221 printk(KERN_INFO
"btrfs: max_extent at %llu\n",
222 (unsigned long long)info
->max_extent
);
226 num
= match_strdup(&args
[0]);
228 info
->max_inline
= btrfs_parse_size(num
);
231 if (info
->max_inline
) {
232 info
->max_inline
= max_t(u64
,
236 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
237 (unsigned long long)info
->max_inline
);
240 case Opt_alloc_start
:
241 num
= match_strdup(&args
[0]);
243 info
->alloc_start
= btrfs_parse_size(num
);
246 "btrfs: allocations start at %llu\n",
247 (unsigned long long)info
->alloc_start
);
251 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
254 printk(KERN_INFO
"btrfs: disabling tree log\n");
255 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
257 case Opt_flushoncommit
:
258 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
259 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
263 match_int(&args
[0], &intarg
);
265 info
->metadata_ratio
= intarg
;
266 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
267 info
->metadata_ratio
);
271 btrfs_set_opt(info
->mount_opt
, DISCARD
);
274 printk(KERN_INFO
"btrfs: unrecognized mount option "
288 * Parse mount options that are required early in the mount process.
290 * All other options will be parsed on much later in the mount process and
291 * only when we need to allocate a new super block.
293 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
294 void *holder
, char **subvol_name
,
295 struct btrfs_fs_devices
**fs_devices
)
297 substring_t args
[MAX_OPT_ARGS
];
305 * strsep changes the string, duplicate it because parse_options
308 opts
= kstrdup(options
, GFP_KERNEL
);
312 while ((p
= strsep(&opts
, ",")) != NULL
) {
317 token
= match_token(p
, tokens
, args
);
320 *subvol_name
= match_strdup(&args
[0]);
323 error
= btrfs_scan_one_device(match_strdup(&args
[0]),
324 flags
, holder
, fs_devices
);
337 * If no subvolume name is specified we use the default one. Allocate
338 * a copy of the string "." here so that code later in the
339 * mount path doesn't care if it's the default volume or another one.
342 *subvol_name
= kstrdup(".", GFP_KERNEL
);
349 static int btrfs_fill_super(struct super_block
*sb
,
350 struct btrfs_fs_devices
*fs_devices
,
351 void *data
, int silent
)
354 struct dentry
*root_dentry
;
355 struct btrfs_super_block
*disk_super
;
356 struct btrfs_root
*tree_root
;
357 struct btrfs_key key
;
360 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
361 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
362 sb
->s_op
= &btrfs_super_ops
;
363 sb
->s_export_op
= &btrfs_export_ops
;
364 sb
->s_xattr
= btrfs_xattr_handlers
;
366 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
367 sb
->s_flags
|= MS_POSIXACL
;
370 tree_root
= open_ctree(sb
, fs_devices
, (char *)data
);
372 if (IS_ERR(tree_root
)) {
373 printk("btrfs: open_ctree failed\n");
374 return PTR_ERR(tree_root
);
376 sb
->s_fs_info
= tree_root
;
377 disk_super
= &tree_root
->fs_info
->super_copy
;
379 key
.objectid
= BTRFS_FIRST_FREE_OBJECTID
;
380 key
.type
= BTRFS_INODE_ITEM_KEY
;
382 inode
= btrfs_iget(sb
, &key
, tree_root
->fs_info
->fs_root
);
384 err
= PTR_ERR(inode
);
388 root_dentry
= d_alloc_root(inode
);
395 /* this does the super kobj at the same time */
396 err
= btrfs_sysfs_add_super(tree_root
->fs_info
);
401 sb
->s_root
= root_dentry
;
403 save_mount_options(sb
, data
);
407 close_ctree(tree_root
);
411 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
413 struct btrfs_trans_handle
*trans
;
414 struct btrfs_root
*root
= btrfs_sb(sb
);
418 filemap_flush(root
->fs_info
->btree_inode
->i_mapping
);
422 btrfs_start_delalloc_inodes(root
, 0);
423 btrfs_wait_ordered_extents(root
, 0, 0);
425 trans
= btrfs_start_transaction(root
, 1);
426 ret
= btrfs_commit_transaction(trans
, root
);
430 static int btrfs_show_options(struct seq_file
*seq
, struct vfsmount
*vfs
)
432 struct btrfs_root
*root
= btrfs_sb(vfs
->mnt_sb
);
433 struct btrfs_fs_info
*info
= root
->fs_info
;
435 if (btrfs_test_opt(root
, DEGRADED
))
436 seq_puts(seq
, ",degraded");
437 if (btrfs_test_opt(root
, NODATASUM
))
438 seq_puts(seq
, ",nodatasum");
439 if (btrfs_test_opt(root
, NODATACOW
))
440 seq_puts(seq
, ",nodatacow");
441 if (btrfs_test_opt(root
, NOBARRIER
))
442 seq_puts(seq
, ",nobarrier");
443 if (info
->max_extent
!= (u64
)-1)
444 seq_printf(seq
, ",max_extent=%llu",
445 (unsigned long long)info
->max_extent
);
446 if (info
->max_inline
!= 8192 * 1024)
447 seq_printf(seq
, ",max_inline=%llu",
448 (unsigned long long)info
->max_inline
);
449 if (info
->alloc_start
!= 0)
450 seq_printf(seq
, ",alloc_start=%llu",
451 (unsigned long long)info
->alloc_start
);
452 if (info
->thread_pool_size
!= min_t(unsigned long,
453 num_online_cpus() + 2, 8))
454 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
455 if (btrfs_test_opt(root
, COMPRESS
))
456 seq_puts(seq
, ",compress");
457 if (btrfs_test_opt(root
, NOSSD
))
458 seq_puts(seq
, ",nossd");
459 if (btrfs_test_opt(root
, SSD_SPREAD
))
460 seq_puts(seq
, ",ssd_spread");
461 else if (btrfs_test_opt(root
, SSD
))
462 seq_puts(seq
, ",ssd");
463 if (btrfs_test_opt(root
, NOTREELOG
))
464 seq_puts(seq
, ",notreelog");
465 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
466 seq_puts(seq
, ",flushoncommit");
467 if (btrfs_test_opt(root
, DISCARD
))
468 seq_puts(seq
, ",discard");
469 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
470 seq_puts(seq
, ",noacl");
474 static int btrfs_test_super(struct super_block
*s
, void *data
)
476 struct btrfs_fs_devices
*test_fs_devices
= data
;
477 struct btrfs_root
*root
= btrfs_sb(s
);
479 return root
->fs_info
->fs_devices
== test_fs_devices
;
483 * Find a superblock for the given device / mount point.
485 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
486 * for multiple device setup. Make sure to keep it in sync.
488 static int btrfs_get_sb(struct file_system_type
*fs_type
, int flags
,
489 const char *dev_name
, void *data
, struct vfsmount
*mnt
)
491 char *subvol_name
= NULL
;
492 struct block_device
*bdev
= NULL
;
493 struct super_block
*s
;
495 struct btrfs_fs_devices
*fs_devices
= NULL
;
496 fmode_t mode
= FMODE_READ
;
499 if (!(flags
& MS_RDONLY
))
502 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
503 &subvol_name
, &fs_devices
);
507 error
= btrfs_scan_one_device(dev_name
, mode
, fs_type
, &fs_devices
);
509 goto error_free_subvol_name
;
511 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
513 goto error_free_subvol_name
;
515 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
517 goto error_close_devices
;
520 bdev
= fs_devices
->latest_bdev
;
521 s
= sget(fs_type
, btrfs_test_super
, set_anon_super
, fs_devices
);
526 if ((flags
^ s
->s_flags
) & MS_RDONLY
) {
527 deactivate_locked_super(s
);
529 goto error_close_devices
;
532 btrfs_close_devices(fs_devices
);
534 char b
[BDEVNAME_SIZE
];
537 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
538 error
= btrfs_fill_super(s
, fs_devices
, data
,
539 flags
& MS_SILENT
? 1 : 0);
541 deactivate_locked_super(s
);
542 goto error_free_subvol_name
;
545 btrfs_sb(s
)->fs_info
->bdev_holder
= fs_type
;
546 s
->s_flags
|= MS_ACTIVE
;
549 if (!strcmp(subvol_name
, "."))
550 root
= dget(s
->s_root
);
552 mutex_lock(&s
->s_root
->d_inode
->i_mutex
);
553 root
= lookup_one_len(subvol_name
, s
->s_root
,
554 strlen(subvol_name
));
555 mutex_unlock(&s
->s_root
->d_inode
->i_mutex
);
558 deactivate_locked_super(s
);
559 error
= PTR_ERR(root
);
560 goto error_free_subvol_name
;
562 if (!root
->d_inode
) {
564 deactivate_locked_super(s
);
566 goto error_free_subvol_name
;
571 mnt
->mnt_root
= root
;
579 btrfs_close_devices(fs_devices
);
580 error_free_subvol_name
:
585 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
587 struct btrfs_root
*root
= btrfs_sb(sb
);
590 ret
= btrfs_parse_options(root
, data
);
594 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
597 if (*flags
& MS_RDONLY
) {
598 sb
->s_flags
|= MS_RDONLY
;
600 ret
= btrfs_commit_super(root
);
603 if (root
->fs_info
->fs_devices
->rw_devices
== 0)
606 if (btrfs_super_log_root(&root
->fs_info
->super_copy
) != 0)
609 /* recover relocation */
610 ret
= btrfs_recover_relocation(root
);
613 ret
= btrfs_cleanup_fs_roots(root
->fs_info
);
616 sb
->s_flags
&= ~MS_RDONLY
;
622 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
624 struct btrfs_root
*root
= btrfs_sb(dentry
->d_sb
);
625 struct btrfs_super_block
*disk_super
= &root
->fs_info
->super_copy
;
626 int bits
= dentry
->d_sb
->s_blocksize_bits
;
627 __be32
*fsid
= (__be32
*)root
->fs_info
->fsid
;
629 buf
->f_namelen
= BTRFS_NAME_LEN
;
630 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
631 buf
->f_bfree
= buf
->f_blocks
-
632 (btrfs_super_bytes_used(disk_super
) >> bits
);
633 buf
->f_bavail
= buf
->f_bfree
;
634 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
635 buf
->f_type
= BTRFS_SUPER_MAGIC
;
637 /* We treat it as constant endianness (it doesn't matter _which_)
638 because we want the fsid to come out the same whether mounted
639 on a big-endian or little-endian host */
640 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
641 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
642 /* Mask in the root object ID too, to disambiguate subvols */
643 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
644 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
649 static struct file_system_type btrfs_fs_type
= {
650 .owner
= THIS_MODULE
,
652 .get_sb
= btrfs_get_sb
,
653 .kill_sb
= kill_anon_super
,
654 .fs_flags
= FS_REQUIRES_DEV
,
658 * used by btrfsctl to scan devices when no FS is mounted
660 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
663 struct btrfs_ioctl_vol_args
*vol
;
664 struct btrfs_fs_devices
*fs_devices
;
667 if (!capable(CAP_SYS_ADMIN
))
670 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
675 case BTRFS_IOC_SCAN_DEV
:
676 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
677 &btrfs_fs_type
, &fs_devices
);
685 static int btrfs_freeze(struct super_block
*sb
)
687 struct btrfs_root
*root
= btrfs_sb(sb
);
688 mutex_lock(&root
->fs_info
->transaction_kthread_mutex
);
689 mutex_lock(&root
->fs_info
->cleaner_mutex
);
693 static int btrfs_unfreeze(struct super_block
*sb
)
695 struct btrfs_root
*root
= btrfs_sb(sb
);
696 mutex_unlock(&root
->fs_info
->cleaner_mutex
);
697 mutex_unlock(&root
->fs_info
->transaction_kthread_mutex
);
701 static const struct super_operations btrfs_super_ops
= {
702 .drop_inode
= btrfs_drop_inode
,
703 .delete_inode
= btrfs_delete_inode
,
704 .put_super
= btrfs_put_super
,
705 .sync_fs
= btrfs_sync_fs
,
706 .show_options
= btrfs_show_options
,
707 .write_inode
= btrfs_write_inode
,
708 .dirty_inode
= btrfs_dirty_inode
,
709 .alloc_inode
= btrfs_alloc_inode
,
710 .destroy_inode
= btrfs_destroy_inode
,
711 .statfs
= btrfs_statfs
,
712 .remount_fs
= btrfs_remount
,
713 .freeze_fs
= btrfs_freeze
,
714 .unfreeze_fs
= btrfs_unfreeze
,
717 static const struct file_operations btrfs_ctl_fops
= {
718 .unlocked_ioctl
= btrfs_control_ioctl
,
719 .compat_ioctl
= btrfs_control_ioctl
,
720 .owner
= THIS_MODULE
,
723 static struct miscdevice btrfs_misc
= {
724 .minor
= MISC_DYNAMIC_MINOR
,
725 .name
= "btrfs-control",
726 .fops
= &btrfs_ctl_fops
729 static int btrfs_interface_init(void)
731 return misc_register(&btrfs_misc
);
734 static void btrfs_interface_exit(void)
736 if (misc_deregister(&btrfs_misc
) < 0)
737 printk(KERN_INFO
"misc_deregister failed for control device");
740 static int __init
init_btrfs_fs(void)
744 err
= btrfs_init_sysfs();
748 err
= btrfs_init_cachep();
752 err
= extent_io_init();
756 err
= extent_map_init();
760 err
= btrfs_interface_init();
762 goto free_extent_map
;
764 err
= register_filesystem(&btrfs_fs_type
);
766 goto unregister_ioctl
;
768 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
772 btrfs_interface_exit();
778 btrfs_destroy_cachep();
784 static void __exit
exit_btrfs_fs(void)
786 btrfs_destroy_cachep();
789 btrfs_interface_exit();
790 unregister_filesystem(&btrfs_fs_type
);
792 btrfs_cleanup_fs_uuids();
796 module_init(init_btrfs_fs
)
797 module_exit(exit_btrfs_fs
)
799 MODULE_LICENSE("GPL");