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>
41 #include <linux/slab.h>
42 #include <linux/cleancache.h>
44 #include "delayed-inode.h"
47 #include "transaction.h"
48 #include "btrfs_inode.h"
50 #include "print-tree.h"
55 #include "compression.h"
57 #define CREATE_TRACE_POINTS
58 #include <trace/events/btrfs.h>
60 static const struct super_operations btrfs_super_ops
;
62 static const char *btrfs_decode_error(struct btrfs_fs_info
*fs_info
, int errno
,
69 errstr
= "IO failure";
72 errstr
= "Out of memory";
75 errstr
= "Readonly filesystem";
79 if (snprintf(nbuf
, 16, "error %d", -errno
) >= 0)
88 static void __save_error_info(struct btrfs_fs_info
*fs_info
)
91 * today we only save the error info into ram. Long term we'll
92 * also send it down to the disk
94 fs_info
->fs_state
= BTRFS_SUPER_FLAG_ERROR
;
98 * We move write_super stuff at umount in order to avoid deadlock
99 * for umount hold all lock.
101 static void save_error_info(struct btrfs_fs_info
*fs_info
)
103 __save_error_info(fs_info
);
106 /* btrfs handle error by forcing the filesystem readonly */
107 static void btrfs_handle_error(struct btrfs_fs_info
*fs_info
)
109 struct super_block
*sb
= fs_info
->sb
;
111 if (sb
->s_flags
& MS_RDONLY
)
114 if (fs_info
->fs_state
& BTRFS_SUPER_FLAG_ERROR
) {
115 sb
->s_flags
|= MS_RDONLY
;
116 printk(KERN_INFO
"btrfs is forced readonly\n");
121 * __btrfs_std_error decodes expected errors from the caller and
122 * invokes the approciate error response.
124 void __btrfs_std_error(struct btrfs_fs_info
*fs_info
, const char *function
,
125 unsigned int line
, int errno
)
127 struct super_block
*sb
= fs_info
->sb
;
132 * Special case: if the error is EROFS, and we're already
133 * under MS_RDONLY, then it is safe here.
135 if (errno
== -EROFS
&& (sb
->s_flags
& MS_RDONLY
))
138 errstr
= btrfs_decode_error(fs_info
, errno
, nbuf
);
139 printk(KERN_CRIT
"BTRFS error (device %s) in %s:%d: %s\n",
140 sb
->s_id
, function
, line
, errstr
);
141 save_error_info(fs_info
);
143 btrfs_handle_error(fs_info
);
146 static void btrfs_put_super(struct super_block
*sb
)
148 struct btrfs_root
*root
= btrfs_sb(sb
);
151 ret
= close_ctree(root
);
152 sb
->s_fs_info
= NULL
;
154 (void)ret
; /* FIXME: need to fix VFS to return error? */
158 Opt_degraded
, Opt_subvol
, Opt_subvolid
, Opt_device
, Opt_nodatasum
,
159 Opt_nodatacow
, Opt_max_inline
, Opt_alloc_start
, Opt_nobarrier
, Opt_ssd
,
160 Opt_nossd
, Opt_ssd_spread
, Opt_thread_pool
, Opt_noacl
, Opt_compress
,
161 Opt_compress_type
, Opt_compress_force
, Opt_compress_force_type
,
162 Opt_notreelog
, Opt_ratio
, Opt_flushoncommit
, Opt_discard
,
163 Opt_space_cache
, Opt_clear_cache
, Opt_user_subvol_rm_allowed
,
164 Opt_enospc_debug
, Opt_subvolrootid
, Opt_defrag
, Opt_err
,
167 static match_table_t tokens
= {
168 {Opt_degraded
, "degraded"},
169 {Opt_subvol
, "subvol=%s"},
170 {Opt_subvolid
, "subvolid=%d"},
171 {Opt_device
, "device=%s"},
172 {Opt_nodatasum
, "nodatasum"},
173 {Opt_nodatacow
, "nodatacow"},
174 {Opt_nobarrier
, "nobarrier"},
175 {Opt_max_inline
, "max_inline=%s"},
176 {Opt_alloc_start
, "alloc_start=%s"},
177 {Opt_thread_pool
, "thread_pool=%d"},
178 {Opt_compress
, "compress"},
179 {Opt_compress_type
, "compress=%s"},
180 {Opt_compress_force
, "compress-force"},
181 {Opt_compress_force_type
, "compress-force=%s"},
183 {Opt_ssd_spread
, "ssd_spread"},
184 {Opt_nossd
, "nossd"},
185 {Opt_noacl
, "noacl"},
186 {Opt_notreelog
, "notreelog"},
187 {Opt_flushoncommit
, "flushoncommit"},
188 {Opt_ratio
, "metadata_ratio=%d"},
189 {Opt_discard
, "discard"},
190 {Opt_space_cache
, "space_cache"},
191 {Opt_clear_cache
, "clear_cache"},
192 {Opt_user_subvol_rm_allowed
, "user_subvol_rm_allowed"},
193 {Opt_enospc_debug
, "enospc_debug"},
194 {Opt_subvolrootid
, "subvolrootid=%d"},
195 {Opt_defrag
, "autodefrag"},
200 * Regular mount options parser. Everything that is needed only when
201 * reading in a new superblock is parsed here.
203 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
205 struct btrfs_fs_info
*info
= root
->fs_info
;
206 substring_t args
[MAX_OPT_ARGS
];
207 char *p
, *num
, *orig
;
211 bool compress_force
= false;
217 * strsep changes the string, duplicate it because parse_options
220 options
= kstrdup(options
, GFP_NOFS
);
226 while ((p
= strsep(&options
, ",")) != NULL
) {
231 token
= match_token(p
, tokens
, args
);
234 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
235 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
239 case Opt_subvolrootid
:
242 * These are parsed by btrfs_parse_early_options
243 * and can be happily ignored here.
247 printk(KERN_INFO
"btrfs: setting nodatasum\n");
248 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
251 printk(KERN_INFO
"btrfs: setting nodatacow\n");
252 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
253 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
255 case Opt_compress_force
:
256 case Opt_compress_force_type
:
257 compress_force
= true;
259 case Opt_compress_type
:
260 if (token
== Opt_compress
||
261 token
== Opt_compress_force
||
262 strcmp(args
[0].from
, "zlib") == 0) {
263 compress_type
= "zlib";
264 info
->compress_type
= BTRFS_COMPRESS_ZLIB
;
265 } else if (strcmp(args
[0].from
, "lzo") == 0) {
266 compress_type
= "lzo";
267 info
->compress_type
= BTRFS_COMPRESS_LZO
;
273 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
274 if (compress_force
) {
275 btrfs_set_opt(info
->mount_opt
, FORCE_COMPRESS
);
276 pr_info("btrfs: force %s compression\n",
279 pr_info("btrfs: use %s compression\n",
283 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
284 btrfs_set_opt(info
->mount_opt
, SSD
);
287 printk(KERN_INFO
"btrfs: use spread ssd "
288 "allocation scheme\n");
289 btrfs_set_opt(info
->mount_opt
, SSD
);
290 btrfs_set_opt(info
->mount_opt
, SSD_SPREAD
);
293 printk(KERN_INFO
"btrfs: not using ssd allocation "
295 btrfs_set_opt(info
->mount_opt
, NOSSD
);
296 btrfs_clear_opt(info
->mount_opt
, SSD
);
297 btrfs_clear_opt(info
->mount_opt
, SSD_SPREAD
);
300 printk(KERN_INFO
"btrfs: turning off barriers\n");
301 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
303 case Opt_thread_pool
:
305 match_int(&args
[0], &intarg
);
307 info
->thread_pool_size
= intarg
;
308 printk(KERN_INFO
"btrfs: thread pool %d\n",
309 info
->thread_pool_size
);
313 num
= match_strdup(&args
[0]);
315 info
->max_inline
= memparse(num
, NULL
);
318 if (info
->max_inline
) {
319 info
->max_inline
= max_t(u64
,
323 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
324 (unsigned long long)info
->max_inline
);
327 case Opt_alloc_start
:
328 num
= match_strdup(&args
[0]);
330 info
->alloc_start
= memparse(num
, NULL
);
333 "btrfs: allocations start at %llu\n",
334 (unsigned long long)info
->alloc_start
);
338 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
341 printk(KERN_INFO
"btrfs: disabling tree log\n");
342 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
344 case Opt_flushoncommit
:
345 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
346 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
350 match_int(&args
[0], &intarg
);
352 info
->metadata_ratio
= intarg
;
353 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
354 info
->metadata_ratio
);
358 btrfs_set_opt(info
->mount_opt
, DISCARD
);
360 case Opt_space_cache
:
361 printk(KERN_INFO
"btrfs: enabling disk space caching\n");
362 btrfs_set_opt(info
->mount_opt
, SPACE_CACHE
);
364 case Opt_clear_cache
:
365 printk(KERN_INFO
"btrfs: force clearing of disk cache\n");
366 btrfs_set_opt(info
->mount_opt
, CLEAR_CACHE
);
368 case Opt_user_subvol_rm_allowed
:
369 btrfs_set_opt(info
->mount_opt
, USER_SUBVOL_RM_ALLOWED
);
371 case Opt_enospc_debug
:
372 btrfs_set_opt(info
->mount_opt
, ENOSPC_DEBUG
);
375 printk(KERN_INFO
"btrfs: enabling auto defrag");
376 btrfs_set_opt(info
->mount_opt
, AUTO_DEFRAG
);
379 printk(KERN_INFO
"btrfs: unrecognized mount option "
393 * Parse mount options that are required early in the mount process.
395 * All other options will be parsed on much later in the mount process and
396 * only when we need to allocate a new super block.
398 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
399 void *holder
, char **subvol_name
, u64
*subvol_objectid
,
400 u64
*subvol_rootid
, struct btrfs_fs_devices
**fs_devices
)
402 substring_t args
[MAX_OPT_ARGS
];
403 char *opts
, *orig
, *p
;
411 * strsep changes the string, duplicate it because parse_options
414 opts
= kstrdup(options
, GFP_KERNEL
);
419 while ((p
= strsep(&opts
, ",")) != NULL
) {
424 token
= match_token(p
, tokens
, args
);
427 *subvol_name
= match_strdup(&args
[0]);
431 error
= match_int(&args
[0], &intarg
);
433 /* we want the original fs_tree */
436 BTRFS_FS_TREE_OBJECTID
;
438 *subvol_objectid
= intarg
;
441 case Opt_subvolrootid
:
443 error
= match_int(&args
[0], &intarg
);
445 /* we want the original fs_tree */
448 BTRFS_FS_TREE_OBJECTID
;
450 *subvol_rootid
= intarg
;
454 error
= btrfs_scan_one_device(match_strdup(&args
[0]),
455 flags
, holder
, fs_devices
);
468 * If no subvolume name is specified we use the default one. Allocate
469 * a copy of the string "." here so that code later in the
470 * mount path doesn't care if it's the default volume or another one.
473 *subvol_name
= kstrdup(".", GFP_KERNEL
);
480 static struct dentry
*get_default_root(struct super_block
*sb
,
483 struct btrfs_root
*root
= sb
->s_fs_info
;
484 struct btrfs_root
*new_root
;
485 struct btrfs_dir_item
*di
;
486 struct btrfs_path
*path
;
487 struct btrfs_key location
;
489 struct dentry
*dentry
;
494 * We have a specific subvol we want to mount, just setup location and
495 * go look up the root.
497 if (subvol_objectid
) {
498 location
.objectid
= subvol_objectid
;
499 location
.type
= BTRFS_ROOT_ITEM_KEY
;
500 location
.offset
= (u64
)-1;
504 path
= btrfs_alloc_path();
506 return ERR_PTR(-ENOMEM
);
507 path
->leave_spinning
= 1;
510 * Find the "default" dir item which points to the root item that we
511 * will mount by default if we haven't been given a specific subvolume
514 dir_id
= btrfs_super_root_dir(&root
->fs_info
->super_copy
);
515 di
= btrfs_lookup_dir_item(NULL
, root
, path
, dir_id
, "default", 7, 0);
517 btrfs_free_path(path
);
522 * Ok the default dir item isn't there. This is weird since
523 * it's always been there, but don't freak out, just try and
524 * mount to root most subvolume.
526 btrfs_free_path(path
);
527 dir_id
= BTRFS_FIRST_FREE_OBJECTID
;
528 new_root
= root
->fs_info
->fs_root
;
532 btrfs_dir_item_key_to_cpu(path
->nodes
[0], di
, &location
);
533 btrfs_free_path(path
);
536 new_root
= btrfs_read_fs_root_no_name(root
->fs_info
, &location
);
537 if (IS_ERR(new_root
))
538 return ERR_CAST(new_root
);
540 if (btrfs_root_refs(&new_root
->root_item
) == 0)
541 return ERR_PTR(-ENOENT
);
543 dir_id
= btrfs_root_dirid(&new_root
->root_item
);
545 location
.objectid
= dir_id
;
546 location
.type
= BTRFS_INODE_ITEM_KEY
;
549 inode
= btrfs_iget(sb
, &location
, new_root
, &new);
551 return ERR_CAST(inode
);
554 * If we're just mounting the root most subvol put the inode and return
555 * a reference to the dentry. We will have already gotten a reference
556 * to the inode in btrfs_fill_super so we're good to go.
558 if (!new && sb
->s_root
->d_inode
== inode
) {
560 return dget(sb
->s_root
);
564 const struct qstr name
= { .name
= "/", .len
= 1 };
567 * New inode, we need to make the dentry a sibling of s_root so
568 * everything gets cleaned up properly on unmount.
570 dentry
= d_alloc(sb
->s_root
, &name
);
573 return ERR_PTR(-ENOMEM
);
575 d_splice_alias(inode
, dentry
);
578 * We found the inode in cache, just find a dentry for it and
579 * put the reference to the inode we just got.
581 dentry
= d_find_alias(inode
);
588 static int btrfs_fill_super(struct super_block
*sb
,
589 struct btrfs_fs_devices
*fs_devices
,
590 void *data
, int silent
)
593 struct dentry
*root_dentry
;
594 struct btrfs_root
*tree_root
;
595 struct btrfs_key key
;
598 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
599 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
600 sb
->s_op
= &btrfs_super_ops
;
601 sb
->s_d_op
= &btrfs_dentry_operations
;
602 sb
->s_export_op
= &btrfs_export_ops
;
603 sb
->s_xattr
= btrfs_xattr_handlers
;
605 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
606 sb
->s_flags
|= MS_POSIXACL
;
609 tree_root
= open_ctree(sb
, fs_devices
, (char *)data
);
611 if (IS_ERR(tree_root
)) {
612 printk("btrfs: open_ctree failed\n");
613 return PTR_ERR(tree_root
);
615 sb
->s_fs_info
= tree_root
;
617 key
.objectid
= BTRFS_FIRST_FREE_OBJECTID
;
618 key
.type
= BTRFS_INODE_ITEM_KEY
;
620 inode
= btrfs_iget(sb
, &key
, tree_root
->fs_info
->fs_root
, NULL
);
622 err
= PTR_ERR(inode
);
626 root_dentry
= d_alloc_root(inode
);
633 sb
->s_root
= root_dentry
;
635 save_mount_options(sb
, data
);
636 cleancache_init_fs(sb
);
640 close_ctree(tree_root
);
644 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
646 struct btrfs_trans_handle
*trans
;
647 struct btrfs_root
*root
= btrfs_sb(sb
);
650 trace_btrfs_sync_fs(wait
);
653 filemap_flush(root
->fs_info
->btree_inode
->i_mapping
);
657 btrfs_start_delalloc_inodes(root
, 0);
658 btrfs_wait_ordered_extents(root
, 0, 0);
660 trans
= btrfs_start_transaction(root
, 0);
662 return PTR_ERR(trans
);
663 ret
= btrfs_commit_transaction(trans
, root
);
667 static int btrfs_show_options(struct seq_file
*seq
, struct vfsmount
*vfs
)
669 struct btrfs_root
*root
= btrfs_sb(vfs
->mnt_sb
);
670 struct btrfs_fs_info
*info
= root
->fs_info
;
673 if (btrfs_test_opt(root
, DEGRADED
))
674 seq_puts(seq
, ",degraded");
675 if (btrfs_test_opt(root
, NODATASUM
))
676 seq_puts(seq
, ",nodatasum");
677 if (btrfs_test_opt(root
, NODATACOW
))
678 seq_puts(seq
, ",nodatacow");
679 if (btrfs_test_opt(root
, NOBARRIER
))
680 seq_puts(seq
, ",nobarrier");
681 if (info
->max_inline
!= 8192 * 1024)
682 seq_printf(seq
, ",max_inline=%llu",
683 (unsigned long long)info
->max_inline
);
684 if (info
->alloc_start
!= 0)
685 seq_printf(seq
, ",alloc_start=%llu",
686 (unsigned long long)info
->alloc_start
);
687 if (info
->thread_pool_size
!= min_t(unsigned long,
688 num_online_cpus() + 2, 8))
689 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
690 if (btrfs_test_opt(root
, COMPRESS
)) {
691 if (info
->compress_type
== BTRFS_COMPRESS_ZLIB
)
692 compress_type
= "zlib";
694 compress_type
= "lzo";
695 if (btrfs_test_opt(root
, FORCE_COMPRESS
))
696 seq_printf(seq
, ",compress-force=%s", compress_type
);
698 seq_printf(seq
, ",compress=%s", compress_type
);
700 if (btrfs_test_opt(root
, NOSSD
))
701 seq_puts(seq
, ",nossd");
702 if (btrfs_test_opt(root
, SSD_SPREAD
))
703 seq_puts(seq
, ",ssd_spread");
704 else if (btrfs_test_opt(root
, SSD
))
705 seq_puts(seq
, ",ssd");
706 if (btrfs_test_opt(root
, NOTREELOG
))
707 seq_puts(seq
, ",notreelog");
708 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
709 seq_puts(seq
, ",flushoncommit");
710 if (btrfs_test_opt(root
, DISCARD
))
711 seq_puts(seq
, ",discard");
712 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
713 seq_puts(seq
, ",noacl");
714 if (btrfs_test_opt(root
, SPACE_CACHE
))
715 seq_puts(seq
, ",space_cache");
716 if (btrfs_test_opt(root
, CLEAR_CACHE
))
717 seq_puts(seq
, ",clear_cache");
718 if (btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
719 seq_puts(seq
, ",user_subvol_rm_allowed");
723 static int btrfs_test_super(struct super_block
*s
, void *data
)
725 struct btrfs_root
*test_root
= data
;
726 struct btrfs_root
*root
= btrfs_sb(s
);
729 * If this super block is going away, return false as it
730 * can't match as an existing super block.
732 if (!atomic_read(&s
->s_active
))
734 return root
->fs_info
->fs_devices
== test_root
->fs_info
->fs_devices
;
737 static int btrfs_set_super(struct super_block
*s
, void *data
)
741 return set_anon_super(s
, data
);
746 * Find a superblock for the given device / mount point.
748 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
749 * for multiple device setup. Make sure to keep it in sync.
751 static struct dentry
*btrfs_mount(struct file_system_type
*fs_type
, int flags
,
752 const char *device_name
, void *data
)
754 struct block_device
*bdev
= NULL
;
755 struct super_block
*s
;
757 struct btrfs_fs_devices
*fs_devices
= NULL
;
758 struct btrfs_root
*tree_root
= NULL
;
759 struct btrfs_fs_info
*fs_info
= NULL
;
760 fmode_t mode
= FMODE_READ
;
761 char *subvol_name
= NULL
;
762 u64 subvol_objectid
= 0;
763 u64 subvol_rootid
= 0;
766 if (!(flags
& MS_RDONLY
))
769 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
770 &subvol_name
, &subvol_objectid
,
771 &subvol_rootid
, &fs_devices
);
773 return ERR_PTR(error
);
775 error
= btrfs_scan_one_device(device_name
, mode
, fs_type
, &fs_devices
);
777 goto error_free_subvol_name
;
779 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
781 goto error_free_subvol_name
;
783 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
785 goto error_close_devices
;
789 * Setup a dummy root and fs_info for test/set super. This is because
790 * we don't actually fill this stuff out until open_ctree, but we need
791 * it for searching for existing supers, so this lets us do that and
792 * then open_ctree will properly initialize everything later.
794 fs_info
= kzalloc(sizeof(struct btrfs_fs_info
), GFP_NOFS
);
795 tree_root
= kzalloc(sizeof(struct btrfs_root
), GFP_NOFS
);
796 if (!fs_info
|| !tree_root
) {
798 goto error_close_devices
;
800 fs_info
->tree_root
= tree_root
;
801 fs_info
->fs_devices
= fs_devices
;
802 tree_root
->fs_info
= fs_info
;
804 bdev
= fs_devices
->latest_bdev
;
805 s
= sget(fs_type
, btrfs_test_super
, btrfs_set_super
, tree_root
);
810 if ((flags
^ s
->s_flags
) & MS_RDONLY
) {
811 deactivate_locked_super(s
);
813 goto error_close_devices
;
816 btrfs_close_devices(fs_devices
);
820 char b
[BDEVNAME_SIZE
];
823 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
824 error
= btrfs_fill_super(s
, fs_devices
, data
,
825 flags
& MS_SILENT
? 1 : 0);
827 deactivate_locked_super(s
);
828 goto error_free_subvol_name
;
831 btrfs_sb(s
)->fs_info
->bdev_holder
= fs_type
;
832 s
->s_flags
|= MS_ACTIVE
;
835 /* if they gave us a subvolume name bind mount into that */
836 if (strcmp(subvol_name
, ".")) {
837 struct dentry
*new_root
;
839 root
= get_default_root(s
, subvol_rootid
);
841 error
= PTR_ERR(root
);
842 deactivate_locked_super(s
);
843 goto error_free_subvol_name
;
846 mutex_lock(&root
->d_inode
->i_mutex
);
847 new_root
= lookup_one_len(subvol_name
, root
,
848 strlen(subvol_name
));
849 mutex_unlock(&root
->d_inode
->i_mutex
);
851 if (IS_ERR(new_root
)) {
853 deactivate_locked_super(s
);
854 error
= PTR_ERR(new_root
);
855 goto error_free_subvol_name
;
857 if (!new_root
->d_inode
) {
860 deactivate_locked_super(s
);
862 goto error_free_subvol_name
;
867 root
= get_default_root(s
, subvol_objectid
);
869 error
= PTR_ERR(root
);
870 deactivate_locked_super(s
);
871 goto error_free_subvol_name
;
881 btrfs_close_devices(fs_devices
);
884 error_free_subvol_name
:
886 return ERR_PTR(error
);
889 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
891 struct btrfs_root
*root
= btrfs_sb(sb
);
894 ret
= btrfs_parse_options(root
, data
);
898 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
901 if (*flags
& MS_RDONLY
) {
902 sb
->s_flags
|= MS_RDONLY
;
904 ret
= btrfs_commit_super(root
);
907 if (root
->fs_info
->fs_devices
->rw_devices
== 0)
910 if (btrfs_super_log_root(&root
->fs_info
->super_copy
) != 0)
913 ret
= btrfs_cleanup_fs_roots(root
->fs_info
);
916 /* recover relocation */
917 ret
= btrfs_recover_relocation(root
);
920 sb
->s_flags
&= ~MS_RDONLY
;
926 /* Used to sort the devices by max_avail(descending sort) */
927 static int btrfs_cmp_device_free_bytes(const void *dev_info1
,
928 const void *dev_info2
)
930 if (((struct btrfs_device_info
*)dev_info1
)->max_avail
>
931 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
933 else if (((struct btrfs_device_info
*)dev_info1
)->max_avail
<
934 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
941 * sort the devices by max_avail, in which max free extent size of each device
942 * is stored.(Descending Sort)
944 static inline void btrfs_descending_sort_devices(
945 struct btrfs_device_info
*devices
,
948 sort(devices
, nr_devices
, sizeof(struct btrfs_device_info
),
949 btrfs_cmp_device_free_bytes
, NULL
);
953 * The helper to calc the free space on the devices that can be used to store
956 static int btrfs_calc_avail_data_space(struct btrfs_root
*root
, u64
*free_bytes
)
958 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
959 struct btrfs_device_info
*devices_info
;
960 struct btrfs_fs_devices
*fs_devices
= fs_info
->fs_devices
;
961 struct btrfs_device
*device
;
968 int i
= 0, nr_devices
;
971 nr_devices
= fs_info
->fs_devices
->rw_devices
;
974 devices_info
= kmalloc(sizeof(*devices_info
) * nr_devices
,
979 /* calc min stripe number for data space alloction */
980 type
= btrfs_get_alloc_profile(root
, 1);
981 if (type
& BTRFS_BLOCK_GROUP_RAID0
)
983 else if (type
& BTRFS_BLOCK_GROUP_RAID1
)
985 else if (type
& BTRFS_BLOCK_GROUP_RAID10
)
988 if (type
& BTRFS_BLOCK_GROUP_DUP
)
989 min_stripe_size
= 2 * BTRFS_STRIPE_LEN
;
991 min_stripe_size
= BTRFS_STRIPE_LEN
;
993 list_for_each_entry(device
, &fs_devices
->alloc_list
, dev_alloc_list
) {
994 if (!device
->in_fs_metadata
)
997 avail_space
= device
->total_bytes
- device
->bytes_used
;
999 /* align with stripe_len */
1000 do_div(avail_space
, BTRFS_STRIPE_LEN
);
1001 avail_space
*= BTRFS_STRIPE_LEN
;
1004 * In order to avoid overwritting the superblock on the drive,
1005 * btrfs starts at an offset of at least 1MB when doing chunk
1008 skip_space
= 1024 * 1024;
1010 /* user can set the offset in fs_info->alloc_start. */
1011 if (fs_info
->alloc_start
+ BTRFS_STRIPE_LEN
<=
1012 device
->total_bytes
)
1013 skip_space
= max(fs_info
->alloc_start
, skip_space
);
1016 * btrfs can not use the free space in [0, skip_space - 1],
1017 * we must subtract it from the total. In order to implement
1018 * it, we account the used space in this range first.
1020 ret
= btrfs_account_dev_extents_size(device
, 0, skip_space
- 1,
1023 kfree(devices_info
);
1027 /* calc the free space in [0, skip_space - 1] */
1028 skip_space
-= used_space
;
1031 * we can use the free space in [0, skip_space - 1], subtract
1032 * it from the total.
1034 if (avail_space
&& avail_space
>= skip_space
)
1035 avail_space
-= skip_space
;
1039 if (avail_space
< min_stripe_size
)
1042 devices_info
[i
].dev
= device
;
1043 devices_info
[i
].max_avail
= avail_space
;
1050 btrfs_descending_sort_devices(devices_info
, nr_devices
);
1054 while (nr_devices
>= min_stripes
) {
1055 if (devices_info
[i
].max_avail
>= min_stripe_size
) {
1059 avail_space
+= devices_info
[i
].max_avail
* min_stripes
;
1060 alloc_size
= devices_info
[i
].max_avail
;
1061 for (j
= i
+ 1 - min_stripes
; j
<= i
; j
++)
1062 devices_info
[j
].max_avail
-= alloc_size
;
1068 kfree(devices_info
);
1069 *free_bytes
= avail_space
;
1073 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
1075 struct btrfs_root
*root
= btrfs_sb(dentry
->d_sb
);
1076 struct btrfs_super_block
*disk_super
= &root
->fs_info
->super_copy
;
1077 struct list_head
*head
= &root
->fs_info
->space_info
;
1078 struct btrfs_space_info
*found
;
1080 u64 total_free_data
= 0;
1081 int bits
= dentry
->d_sb
->s_blocksize_bits
;
1082 __be32
*fsid
= (__be32
*)root
->fs_info
->fsid
;
1085 /* holding chunk_muext to avoid allocating new chunks */
1086 mutex_lock(&root
->fs_info
->chunk_mutex
);
1088 list_for_each_entry_rcu(found
, head
, list
) {
1089 if (found
->flags
& BTRFS_BLOCK_GROUP_DATA
) {
1090 total_free_data
+= found
->disk_total
- found
->disk_used
;
1092 btrfs_account_ro_block_groups_free_space(found
);
1095 total_used
+= found
->disk_used
;
1099 buf
->f_namelen
= BTRFS_NAME_LEN
;
1100 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
1101 buf
->f_bfree
= buf
->f_blocks
- (total_used
>> bits
);
1102 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
1103 buf
->f_type
= BTRFS_SUPER_MAGIC
;
1104 buf
->f_bavail
= total_free_data
;
1105 ret
= btrfs_calc_avail_data_space(root
, &total_free_data
);
1107 mutex_unlock(&root
->fs_info
->chunk_mutex
);
1110 buf
->f_bavail
+= total_free_data
;
1111 buf
->f_bavail
= buf
->f_bavail
>> bits
;
1112 mutex_unlock(&root
->fs_info
->chunk_mutex
);
1114 /* We treat it as constant endianness (it doesn't matter _which_)
1115 because we want the fsid to come out the same whether mounted
1116 on a big-endian or little-endian host */
1117 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
1118 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
1119 /* Mask in the root object ID too, to disambiguate subvols */
1120 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
1121 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
1126 static struct file_system_type btrfs_fs_type
= {
1127 .owner
= THIS_MODULE
,
1129 .mount
= btrfs_mount
,
1130 .kill_sb
= kill_anon_super
,
1131 .fs_flags
= FS_REQUIRES_DEV
,
1135 * used by btrfsctl to scan devices when no FS is mounted
1137 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
1140 struct btrfs_ioctl_vol_args
*vol
;
1141 struct btrfs_fs_devices
*fs_devices
;
1144 if (!capable(CAP_SYS_ADMIN
))
1147 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
1149 return PTR_ERR(vol
);
1152 case BTRFS_IOC_SCAN_DEV
:
1153 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
1154 &btrfs_fs_type
, &fs_devices
);
1162 static int btrfs_freeze(struct super_block
*sb
)
1164 struct btrfs_root
*root
= btrfs_sb(sb
);
1165 mutex_lock(&root
->fs_info
->transaction_kthread_mutex
);
1166 mutex_lock(&root
->fs_info
->cleaner_mutex
);
1170 static int btrfs_unfreeze(struct super_block
*sb
)
1172 struct btrfs_root
*root
= btrfs_sb(sb
);
1173 mutex_unlock(&root
->fs_info
->cleaner_mutex
);
1174 mutex_unlock(&root
->fs_info
->transaction_kthread_mutex
);
1178 static const struct super_operations btrfs_super_ops
= {
1179 .drop_inode
= btrfs_drop_inode
,
1180 .evict_inode
= btrfs_evict_inode
,
1181 .put_super
= btrfs_put_super
,
1182 .sync_fs
= btrfs_sync_fs
,
1183 .show_options
= btrfs_show_options
,
1184 .write_inode
= btrfs_write_inode
,
1185 .dirty_inode
= btrfs_dirty_inode
,
1186 .alloc_inode
= btrfs_alloc_inode
,
1187 .destroy_inode
= btrfs_destroy_inode
,
1188 .statfs
= btrfs_statfs
,
1189 .remount_fs
= btrfs_remount
,
1190 .freeze_fs
= btrfs_freeze
,
1191 .unfreeze_fs
= btrfs_unfreeze
,
1194 static const struct file_operations btrfs_ctl_fops
= {
1195 .unlocked_ioctl
= btrfs_control_ioctl
,
1196 .compat_ioctl
= btrfs_control_ioctl
,
1197 .owner
= THIS_MODULE
,
1198 .llseek
= noop_llseek
,
1201 static struct miscdevice btrfs_misc
= {
1202 .minor
= BTRFS_MINOR
,
1203 .name
= "btrfs-control",
1204 .fops
= &btrfs_ctl_fops
1207 MODULE_ALIAS_MISCDEV(BTRFS_MINOR
);
1208 MODULE_ALIAS("devname:btrfs-control");
1210 static int btrfs_interface_init(void)
1212 return misc_register(&btrfs_misc
);
1215 static void btrfs_interface_exit(void)
1217 if (misc_deregister(&btrfs_misc
) < 0)
1218 printk(KERN_INFO
"misc_deregister failed for control device");
1221 static int __init
init_btrfs_fs(void)
1225 err
= btrfs_init_sysfs();
1229 err
= btrfs_init_compress();
1233 err
= btrfs_init_cachep();
1237 err
= extent_io_init();
1241 err
= extent_map_init();
1243 goto free_extent_io
;
1245 err
= btrfs_delayed_inode_init();
1247 goto free_extent_map
;
1249 err
= btrfs_interface_init();
1251 goto free_delayed_inode
;
1253 err
= register_filesystem(&btrfs_fs_type
);
1255 goto unregister_ioctl
;
1257 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
1261 btrfs_interface_exit();
1263 btrfs_delayed_inode_exit();
1269 btrfs_destroy_cachep();
1271 btrfs_exit_compress();
1277 static void __exit
exit_btrfs_fs(void)
1279 btrfs_destroy_cachep();
1280 btrfs_delayed_inode_exit();
1283 btrfs_interface_exit();
1284 unregister_filesystem(&btrfs_fs_type
);
1286 btrfs_cleanup_fs_uuids();
1287 btrfs_exit_compress();
1290 module_init(init_btrfs_fs
)
1291 module_exit(exit_btrfs_fs
)
1293 MODULE_LICENSE("GPL");