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>
43 #include <linux/ratelimit.h>
44 #include <linux/btrfs.h>
46 #include "delayed-inode.h"
49 #include "transaction.h"
50 #include "btrfs_inode.h"
51 #include "print-tree.h"
56 #include "compression.h"
57 #include "rcu-string.h"
58 #include "dev-replace.h"
59 #include "free-space-cache.h"
61 #define CREATE_TRACE_POINTS
62 #include <trace/events/btrfs.h>
64 static const struct super_operations btrfs_super_ops
;
65 static struct file_system_type btrfs_fs_type
;
67 static const char *btrfs_decode_error(int errno
)
69 char *errstr
= "unknown";
73 errstr
= "IO failure";
76 errstr
= "Out of memory";
79 errstr
= "Readonly filesystem";
82 errstr
= "Object already exists";
89 static void save_error_info(struct btrfs_fs_info
*fs_info
)
92 * today we only save the error info into ram. Long term we'll
93 * also send it down to the disk
95 set_bit(BTRFS_FS_STATE_ERROR
, &fs_info
->fs_state
);
98 /* btrfs handle error by forcing the filesystem readonly */
99 static void btrfs_handle_error(struct btrfs_fs_info
*fs_info
)
101 struct super_block
*sb
= fs_info
->sb
;
103 if (sb
->s_flags
& MS_RDONLY
)
106 if (test_bit(BTRFS_FS_STATE_ERROR
, &fs_info
->fs_state
)) {
107 sb
->s_flags
|= MS_RDONLY
;
108 btrfs_info(fs_info
, "forced readonly");
110 * Note that a running device replace operation is not
111 * canceled here although there is no way to update
112 * the progress. It would add the risk of a deadlock,
113 * therefore the canceling is ommited. The only penalty
114 * is that some I/O remains active until the procedure
115 * completes. The next time when the filesystem is
116 * mounted writeable again, the device replace
117 * operation continues.
124 * __btrfs_std_error decodes expected errors from the caller and
125 * invokes the approciate error response.
127 void __btrfs_std_error(struct btrfs_fs_info
*fs_info
, const char *function
,
128 unsigned int line
, int errno
, const char *fmt
, ...)
130 struct super_block
*sb
= fs_info
->sb
;
134 * Special case: if the error is EROFS, and we're already
135 * under MS_RDONLY, then it is safe here.
137 if (errno
== -EROFS
&& (sb
->s_flags
& MS_RDONLY
))
140 errstr
= btrfs_decode_error(errno
);
142 struct va_format vaf
;
149 printk(KERN_CRIT
"BTRFS error (device %s) in %s:%d: errno=%d %s (%pV)\n",
150 sb
->s_id
, function
, line
, errno
, errstr
, &vaf
);
153 printk(KERN_CRIT
"BTRFS error (device %s) in %s:%d: errno=%d %s\n",
154 sb
->s_id
, function
, line
, errno
, errstr
);
157 /* Don't go through full error handling during mount */
158 if (sb
->s_flags
& MS_BORN
) {
159 save_error_info(fs_info
);
160 btrfs_handle_error(fs_info
);
164 static const char * const logtypes
[] = {
175 void btrfs_printk(const struct btrfs_fs_info
*fs_info
, const char *fmt
, ...)
177 struct super_block
*sb
= fs_info
->sb
;
179 struct va_format vaf
;
181 const char *type
= logtypes
[4];
186 kern_level
= printk_get_level(fmt
);
188 size_t size
= printk_skip_level(fmt
) - fmt
;
189 memcpy(lvl
, fmt
, size
);
192 type
= logtypes
[kern_level
- '0'];
199 printk("%sBTRFS %s (device %s): %pV\n", lvl
, type
, sb
->s_id
, &vaf
);
206 void __btrfs_std_error(struct btrfs_fs_info
*fs_info
, const char *function
,
207 unsigned int line
, int errno
, const char *fmt
, ...)
209 struct super_block
*sb
= fs_info
->sb
;
212 * Special case: if the error is EROFS, and we're already
213 * under MS_RDONLY, then it is safe here.
215 if (errno
== -EROFS
&& (sb
->s_flags
& MS_RDONLY
))
218 /* Don't go through full error handling during mount */
219 if (sb
->s_flags
& MS_BORN
) {
220 save_error_info(fs_info
);
221 btrfs_handle_error(fs_info
);
227 * We only mark the transaction aborted and then set the file system read-only.
228 * This will prevent new transactions from starting or trying to join this
231 * This means that error recovery at the call site is limited to freeing
232 * any local memory allocations and passing the error code up without
233 * further cleanup. The transaction should complete as it normally would
234 * in the call path but will return -EIO.
236 * We'll complete the cleanup in btrfs_end_transaction and
237 * btrfs_commit_transaction.
239 void __btrfs_abort_transaction(struct btrfs_trans_handle
*trans
,
240 struct btrfs_root
*root
, const char *function
,
241 unsigned int line
, int errno
)
244 * Report first abort since mount
246 if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED
,
247 &root
->fs_info
->fs_state
)) {
248 WARN(1, KERN_DEBUG
"btrfs: Transaction aborted (error %d)\n",
251 trans
->aborted
= errno
;
252 /* Nothing used. The other threads that have joined this
253 * transaction may be able to continue. */
254 if (!trans
->blocks_used
) {
257 errstr
= btrfs_decode_error(errno
);
258 btrfs_warn(root
->fs_info
,
259 "%s:%d: Aborting unused transaction(%s).",
260 function
, line
, errstr
);
263 ACCESS_ONCE(trans
->transaction
->aborted
) = errno
;
264 __btrfs_std_error(root
->fs_info
, function
, line
, errno
, NULL
);
267 * __btrfs_panic decodes unexpected, fatal errors from the caller,
268 * issues an alert, and either panics or BUGs, depending on mount options.
270 void __btrfs_panic(struct btrfs_fs_info
*fs_info
, const char *function
,
271 unsigned int line
, int errno
, const char *fmt
, ...)
273 char *s_id
= "<unknown>";
275 struct va_format vaf
= { .fmt
= fmt
};
279 s_id
= fs_info
->sb
->s_id
;
284 errstr
= btrfs_decode_error(errno
);
285 if (fs_info
&& (fs_info
->mount_opt
& BTRFS_MOUNT_PANIC_ON_FATAL_ERROR
))
286 panic(KERN_CRIT
"BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
287 s_id
, function
, line
, &vaf
, errno
, errstr
);
289 printk(KERN_CRIT
"BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
290 s_id
, function
, line
, &vaf
, errno
, errstr
);
292 /* Caller calls BUG() */
295 static void btrfs_put_super(struct super_block
*sb
)
297 (void)close_ctree(btrfs_sb(sb
)->tree_root
);
298 /* FIXME: need to fix VFS to return error? */
299 /* AV: return it _where_? ->put_super() can be triggered by any number
300 * of async events, up to and including delivery of SIGKILL to the
301 * last process that kept it busy. Or segfault in the aforementioned
302 * process... Whom would you report that to?
307 Opt_degraded
, Opt_subvol
, Opt_subvolid
, Opt_device
, Opt_nodatasum
,
308 Opt_nodatacow
, Opt_max_inline
, Opt_alloc_start
, Opt_nobarrier
, Opt_ssd
,
309 Opt_nossd
, Opt_ssd_spread
, Opt_thread_pool
, Opt_noacl
, Opt_compress
,
310 Opt_compress_type
, Opt_compress_force
, Opt_compress_force_type
,
311 Opt_notreelog
, Opt_ratio
, Opt_flushoncommit
, Opt_discard
,
312 Opt_space_cache
, Opt_clear_cache
, Opt_user_subvol_rm_allowed
,
313 Opt_enospc_debug
, Opt_subvolrootid
, Opt_defrag
, Opt_inode_cache
,
314 Opt_no_space_cache
, Opt_recovery
, Opt_skip_balance
,
315 Opt_check_integrity
, Opt_check_integrity_including_extent_data
,
316 Opt_check_integrity_print_mask
, Opt_fatal_errors
,
320 static match_table_t tokens
= {
321 {Opt_degraded
, "degraded"},
322 {Opt_subvol
, "subvol=%s"},
323 {Opt_subvolid
, "subvolid=%d"},
324 {Opt_device
, "device=%s"},
325 {Opt_nodatasum
, "nodatasum"},
326 {Opt_nodatacow
, "nodatacow"},
327 {Opt_nobarrier
, "nobarrier"},
328 {Opt_max_inline
, "max_inline=%s"},
329 {Opt_alloc_start
, "alloc_start=%s"},
330 {Opt_thread_pool
, "thread_pool=%d"},
331 {Opt_compress
, "compress"},
332 {Opt_compress_type
, "compress=%s"},
333 {Opt_compress_force
, "compress-force"},
334 {Opt_compress_force_type
, "compress-force=%s"},
336 {Opt_ssd_spread
, "ssd_spread"},
337 {Opt_nossd
, "nossd"},
338 {Opt_noacl
, "noacl"},
339 {Opt_notreelog
, "notreelog"},
340 {Opt_flushoncommit
, "flushoncommit"},
341 {Opt_ratio
, "metadata_ratio=%d"},
342 {Opt_discard
, "discard"},
343 {Opt_space_cache
, "space_cache"},
344 {Opt_clear_cache
, "clear_cache"},
345 {Opt_user_subvol_rm_allowed
, "user_subvol_rm_allowed"},
346 {Opt_enospc_debug
, "enospc_debug"},
347 {Opt_subvolrootid
, "subvolrootid=%d"},
348 {Opt_defrag
, "autodefrag"},
349 {Opt_inode_cache
, "inode_cache"},
350 {Opt_no_space_cache
, "nospace_cache"},
351 {Opt_recovery
, "recovery"},
352 {Opt_skip_balance
, "skip_balance"},
353 {Opt_check_integrity
, "check_int"},
354 {Opt_check_integrity_including_extent_data
, "check_int_data"},
355 {Opt_check_integrity_print_mask
, "check_int_print_mask=%d"},
356 {Opt_fatal_errors
, "fatal_errors=%s"},
361 * Regular mount options parser. Everything that is needed only when
362 * reading in a new superblock is parsed here.
363 * XXX JDM: This needs to be cleaned up for remount.
365 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
367 struct btrfs_fs_info
*info
= root
->fs_info
;
368 substring_t args
[MAX_OPT_ARGS
];
369 char *p
, *num
, *orig
= NULL
;
374 bool compress_force
= false;
376 cache_gen
= btrfs_super_cache_generation(root
->fs_info
->super_copy
);
378 btrfs_set_opt(info
->mount_opt
, SPACE_CACHE
);
384 * strsep changes the string, duplicate it because parse_options
387 options
= kstrdup(options
, GFP_NOFS
);
393 while ((p
= strsep(&options
, ",")) != NULL
) {
398 token
= match_token(p
, tokens
, args
);
401 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
402 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
406 case Opt_subvolrootid
:
409 * These are parsed by btrfs_parse_early_options
410 * and can be happily ignored here.
414 printk(KERN_INFO
"btrfs: setting nodatasum\n");
415 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
418 if (!btrfs_test_opt(root
, COMPRESS
) ||
419 !btrfs_test_opt(root
, FORCE_COMPRESS
)) {
420 printk(KERN_INFO
"btrfs: setting nodatacow, compression disabled\n");
422 printk(KERN_INFO
"btrfs: setting nodatacow\n");
424 info
->compress_type
= BTRFS_COMPRESS_NONE
;
425 btrfs_clear_opt(info
->mount_opt
, COMPRESS
);
426 btrfs_clear_opt(info
->mount_opt
, FORCE_COMPRESS
);
427 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
428 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
430 case Opt_compress_force
:
431 case Opt_compress_force_type
:
432 compress_force
= true;
435 case Opt_compress_type
:
436 if (token
== Opt_compress
||
437 token
== Opt_compress_force
||
438 strcmp(args
[0].from
, "zlib") == 0) {
439 compress_type
= "zlib";
440 info
->compress_type
= BTRFS_COMPRESS_ZLIB
;
441 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
442 btrfs_clear_opt(info
->mount_opt
, NODATACOW
);
443 btrfs_clear_opt(info
->mount_opt
, NODATASUM
);
444 } else if (strcmp(args
[0].from
, "lzo") == 0) {
445 compress_type
= "lzo";
446 info
->compress_type
= BTRFS_COMPRESS_LZO
;
447 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
448 btrfs_clear_opt(info
->mount_opt
, NODATACOW
);
449 btrfs_clear_opt(info
->mount_opt
, NODATASUM
);
450 btrfs_set_fs_incompat(info
, COMPRESS_LZO
);
451 } else if (strncmp(args
[0].from
, "no", 2) == 0) {
452 compress_type
= "no";
453 info
->compress_type
= BTRFS_COMPRESS_NONE
;
454 btrfs_clear_opt(info
->mount_opt
, COMPRESS
);
455 btrfs_clear_opt(info
->mount_opt
, FORCE_COMPRESS
);
456 compress_force
= false;
462 if (compress_force
) {
463 btrfs_set_opt(info
->mount_opt
, FORCE_COMPRESS
);
464 pr_info("btrfs: force %s compression\n",
467 pr_info("btrfs: use %s compression\n",
471 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
472 btrfs_set_opt(info
->mount_opt
, SSD
);
475 printk(KERN_INFO
"btrfs: use spread ssd "
476 "allocation scheme\n");
477 btrfs_set_opt(info
->mount_opt
, SSD
);
478 btrfs_set_opt(info
->mount_opt
, SSD_SPREAD
);
481 printk(KERN_INFO
"btrfs: not using ssd allocation "
483 btrfs_set_opt(info
->mount_opt
, NOSSD
);
484 btrfs_clear_opt(info
->mount_opt
, SSD
);
485 btrfs_clear_opt(info
->mount_opt
, SSD_SPREAD
);
488 printk(KERN_INFO
"btrfs: turning off barriers\n");
489 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
491 case Opt_thread_pool
:
493 match_int(&args
[0], &intarg
);
495 info
->thread_pool_size
= intarg
;
498 num
= match_strdup(&args
[0]);
500 info
->max_inline
= memparse(num
, NULL
);
503 if (info
->max_inline
) {
504 info
->max_inline
= max_t(u64
,
508 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
509 (unsigned long long)info
->max_inline
);
512 case Opt_alloc_start
:
513 num
= match_strdup(&args
[0]);
515 mutex_lock(&info
->chunk_mutex
);
516 info
->alloc_start
= memparse(num
, NULL
);
517 mutex_unlock(&info
->chunk_mutex
);
520 "btrfs: allocations start at %llu\n",
521 (unsigned long long)info
->alloc_start
);
525 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
528 printk(KERN_INFO
"btrfs: disabling tree log\n");
529 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
531 case Opt_flushoncommit
:
532 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
533 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
537 match_int(&args
[0], &intarg
);
539 info
->metadata_ratio
= intarg
;
540 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
541 info
->metadata_ratio
);
545 btrfs_set_opt(info
->mount_opt
, DISCARD
);
547 case Opt_space_cache
:
548 btrfs_set_opt(info
->mount_opt
, SPACE_CACHE
);
550 case Opt_no_space_cache
:
551 printk(KERN_INFO
"btrfs: disabling disk space caching\n");
552 btrfs_clear_opt(info
->mount_opt
, SPACE_CACHE
);
554 case Opt_inode_cache
:
555 printk(KERN_INFO
"btrfs: enabling inode map caching\n");
556 btrfs_set_opt(info
->mount_opt
, INODE_MAP_CACHE
);
558 case Opt_clear_cache
:
559 printk(KERN_INFO
"btrfs: force clearing of disk cache\n");
560 btrfs_set_opt(info
->mount_opt
, CLEAR_CACHE
);
562 case Opt_user_subvol_rm_allowed
:
563 btrfs_set_opt(info
->mount_opt
, USER_SUBVOL_RM_ALLOWED
);
565 case Opt_enospc_debug
:
566 btrfs_set_opt(info
->mount_opt
, ENOSPC_DEBUG
);
569 printk(KERN_INFO
"btrfs: enabling auto defrag\n");
570 btrfs_set_opt(info
->mount_opt
, AUTO_DEFRAG
);
573 printk(KERN_INFO
"btrfs: enabling auto recovery\n");
574 btrfs_set_opt(info
->mount_opt
, RECOVERY
);
576 case Opt_skip_balance
:
577 btrfs_set_opt(info
->mount_opt
, SKIP_BALANCE
);
579 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
580 case Opt_check_integrity_including_extent_data
:
581 printk(KERN_INFO
"btrfs: enabling check integrity"
582 " including extent data\n");
583 btrfs_set_opt(info
->mount_opt
,
584 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA
);
585 btrfs_set_opt(info
->mount_opt
, CHECK_INTEGRITY
);
587 case Opt_check_integrity
:
588 printk(KERN_INFO
"btrfs: enabling check integrity\n");
589 btrfs_set_opt(info
->mount_opt
, CHECK_INTEGRITY
);
591 case Opt_check_integrity_print_mask
:
593 match_int(&args
[0], &intarg
);
595 info
->check_integrity_print_mask
= intarg
;
596 printk(KERN_INFO
"btrfs:"
597 " check_integrity_print_mask 0x%x\n",
598 info
->check_integrity_print_mask
);
602 case Opt_check_integrity_including_extent_data
:
603 case Opt_check_integrity
:
604 case Opt_check_integrity_print_mask
:
605 printk(KERN_ERR
"btrfs: support for check_integrity*"
606 " not compiled in!\n");
610 case Opt_fatal_errors
:
611 if (strcmp(args
[0].from
, "panic") == 0)
612 btrfs_set_opt(info
->mount_opt
,
613 PANIC_ON_FATAL_ERROR
);
614 else if (strcmp(args
[0].from
, "bug") == 0)
615 btrfs_clear_opt(info
->mount_opt
,
616 PANIC_ON_FATAL_ERROR
);
623 printk(KERN_INFO
"btrfs: unrecognized mount option "
632 if (!ret
&& btrfs_test_opt(root
, SPACE_CACHE
))
633 printk(KERN_INFO
"btrfs: disk space caching is enabled\n");
639 * Parse mount options that are required early in the mount process.
641 * All other options will be parsed on much later in the mount process and
642 * only when we need to allocate a new super block.
644 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
645 void *holder
, char **subvol_name
, u64
*subvol_objectid
,
646 u64
*subvol_rootid
, struct btrfs_fs_devices
**fs_devices
)
648 substring_t args
[MAX_OPT_ARGS
];
649 char *device_name
, *opts
, *orig
, *p
;
657 * strsep changes the string, duplicate it because parse_options
660 opts
= kstrdup(options
, GFP_KERNEL
);
665 while ((p
= strsep(&opts
, ",")) != NULL
) {
670 token
= match_token(p
, tokens
, args
);
674 *subvol_name
= match_strdup(&args
[0]);
678 error
= match_int(&args
[0], &intarg
);
680 /* we want the original fs_tree */
683 BTRFS_FS_TREE_OBJECTID
;
685 *subvol_objectid
= intarg
;
688 case Opt_subvolrootid
:
690 error
= match_int(&args
[0], &intarg
);
692 /* we want the original fs_tree */
695 BTRFS_FS_TREE_OBJECTID
;
697 *subvol_rootid
= intarg
;
701 device_name
= match_strdup(&args
[0]);
706 error
= btrfs_scan_one_device(device_name
,
707 flags
, holder
, fs_devices
);
722 static struct dentry
*get_default_root(struct super_block
*sb
,
725 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
726 struct btrfs_root
*root
= fs_info
->tree_root
;
727 struct btrfs_root
*new_root
;
728 struct btrfs_dir_item
*di
;
729 struct btrfs_path
*path
;
730 struct btrfs_key location
;
736 * We have a specific subvol we want to mount, just setup location and
737 * go look up the root.
739 if (subvol_objectid
) {
740 location
.objectid
= subvol_objectid
;
741 location
.type
= BTRFS_ROOT_ITEM_KEY
;
742 location
.offset
= (u64
)-1;
746 path
= btrfs_alloc_path();
748 return ERR_PTR(-ENOMEM
);
749 path
->leave_spinning
= 1;
752 * Find the "default" dir item which points to the root item that we
753 * will mount by default if we haven't been given a specific subvolume
756 dir_id
= btrfs_super_root_dir(fs_info
->super_copy
);
757 di
= btrfs_lookup_dir_item(NULL
, root
, path
, dir_id
, "default", 7, 0);
759 btrfs_free_path(path
);
764 * Ok the default dir item isn't there. This is weird since
765 * it's always been there, but don't freak out, just try and
766 * mount to root most subvolume.
768 btrfs_free_path(path
);
769 dir_id
= BTRFS_FIRST_FREE_OBJECTID
;
770 new_root
= fs_info
->fs_root
;
774 btrfs_dir_item_key_to_cpu(path
->nodes
[0], di
, &location
);
775 btrfs_free_path(path
);
778 new_root
= btrfs_read_fs_root_no_name(fs_info
, &location
);
779 if (IS_ERR(new_root
))
780 return ERR_CAST(new_root
);
782 if (btrfs_root_refs(&new_root
->root_item
) == 0)
783 return ERR_PTR(-ENOENT
);
785 dir_id
= btrfs_root_dirid(&new_root
->root_item
);
787 location
.objectid
= dir_id
;
788 location
.type
= BTRFS_INODE_ITEM_KEY
;
791 inode
= btrfs_iget(sb
, &location
, new_root
, &new);
793 return ERR_CAST(inode
);
796 * If we're just mounting the root most subvol put the inode and return
797 * a reference to the dentry. We will have already gotten a reference
798 * to the inode in btrfs_fill_super so we're good to go.
800 if (!new && sb
->s_root
->d_inode
== inode
) {
802 return dget(sb
->s_root
);
805 return d_obtain_alias(inode
);
808 static int btrfs_fill_super(struct super_block
*sb
,
809 struct btrfs_fs_devices
*fs_devices
,
810 void *data
, int silent
)
813 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
814 struct btrfs_key key
;
817 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
818 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
819 sb
->s_op
= &btrfs_super_ops
;
820 sb
->s_d_op
= &btrfs_dentry_operations
;
821 sb
->s_export_op
= &btrfs_export_ops
;
822 sb
->s_xattr
= btrfs_xattr_handlers
;
824 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
825 sb
->s_flags
|= MS_POSIXACL
;
827 sb
->s_flags
|= MS_I_VERSION
;
828 err
= open_ctree(sb
, fs_devices
, (char *)data
);
830 printk("btrfs: open_ctree failed\n");
834 key
.objectid
= BTRFS_FIRST_FREE_OBJECTID
;
835 key
.type
= BTRFS_INODE_ITEM_KEY
;
837 inode
= btrfs_iget(sb
, &key
, fs_info
->fs_root
, NULL
);
839 err
= PTR_ERR(inode
);
843 sb
->s_root
= d_make_root(inode
);
849 save_mount_options(sb
, data
);
850 cleancache_init_fs(sb
);
851 sb
->s_flags
|= MS_ACTIVE
;
855 close_ctree(fs_info
->tree_root
);
859 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
861 struct btrfs_trans_handle
*trans
;
862 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
863 struct btrfs_root
*root
= fs_info
->tree_root
;
865 trace_btrfs_sync_fs(wait
);
868 filemap_flush(fs_info
->btree_inode
->i_mapping
);
872 btrfs_wait_ordered_extents(root
, 0);
874 trans
= btrfs_attach_transaction_barrier(root
);
876 /* no transaction, don't bother */
877 if (PTR_ERR(trans
) == -ENOENT
)
879 return PTR_ERR(trans
);
881 return btrfs_commit_transaction(trans
, root
);
884 static int btrfs_show_options(struct seq_file
*seq
, struct dentry
*dentry
)
886 struct btrfs_fs_info
*info
= btrfs_sb(dentry
->d_sb
);
887 struct btrfs_root
*root
= info
->tree_root
;
890 if (btrfs_test_opt(root
, DEGRADED
))
891 seq_puts(seq
, ",degraded");
892 if (btrfs_test_opt(root
, NODATASUM
))
893 seq_puts(seq
, ",nodatasum");
894 if (btrfs_test_opt(root
, NODATACOW
))
895 seq_puts(seq
, ",nodatacow");
896 if (btrfs_test_opt(root
, NOBARRIER
))
897 seq_puts(seq
, ",nobarrier");
898 if (info
->max_inline
!= 8192 * 1024)
899 seq_printf(seq
, ",max_inline=%llu",
900 (unsigned long long)info
->max_inline
);
901 if (info
->alloc_start
!= 0)
902 seq_printf(seq
, ",alloc_start=%llu",
903 (unsigned long long)info
->alloc_start
);
904 if (info
->thread_pool_size
!= min_t(unsigned long,
905 num_online_cpus() + 2, 8))
906 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
907 if (btrfs_test_opt(root
, COMPRESS
)) {
908 if (info
->compress_type
== BTRFS_COMPRESS_ZLIB
)
909 compress_type
= "zlib";
911 compress_type
= "lzo";
912 if (btrfs_test_opt(root
, FORCE_COMPRESS
))
913 seq_printf(seq
, ",compress-force=%s", compress_type
);
915 seq_printf(seq
, ",compress=%s", compress_type
);
917 if (btrfs_test_opt(root
, NOSSD
))
918 seq_puts(seq
, ",nossd");
919 if (btrfs_test_opt(root
, SSD_SPREAD
))
920 seq_puts(seq
, ",ssd_spread");
921 else if (btrfs_test_opt(root
, SSD
))
922 seq_puts(seq
, ",ssd");
923 if (btrfs_test_opt(root
, NOTREELOG
))
924 seq_puts(seq
, ",notreelog");
925 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
926 seq_puts(seq
, ",flushoncommit");
927 if (btrfs_test_opt(root
, DISCARD
))
928 seq_puts(seq
, ",discard");
929 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
930 seq_puts(seq
, ",noacl");
931 if (btrfs_test_opt(root
, SPACE_CACHE
))
932 seq_puts(seq
, ",space_cache");
934 seq_puts(seq
, ",nospace_cache");
935 if (btrfs_test_opt(root
, CLEAR_CACHE
))
936 seq_puts(seq
, ",clear_cache");
937 if (btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
938 seq_puts(seq
, ",user_subvol_rm_allowed");
939 if (btrfs_test_opt(root
, ENOSPC_DEBUG
))
940 seq_puts(seq
, ",enospc_debug");
941 if (btrfs_test_opt(root
, AUTO_DEFRAG
))
942 seq_puts(seq
, ",autodefrag");
943 if (btrfs_test_opt(root
, INODE_MAP_CACHE
))
944 seq_puts(seq
, ",inode_cache");
945 if (btrfs_test_opt(root
, SKIP_BALANCE
))
946 seq_puts(seq
, ",skip_balance");
947 if (btrfs_test_opt(root
, PANIC_ON_FATAL_ERROR
))
948 seq_puts(seq
, ",fatal_errors=panic");
952 static int btrfs_test_super(struct super_block
*s
, void *data
)
954 struct btrfs_fs_info
*p
= data
;
955 struct btrfs_fs_info
*fs_info
= btrfs_sb(s
);
957 return fs_info
->fs_devices
== p
->fs_devices
;
960 static int btrfs_set_super(struct super_block
*s
, void *data
)
962 int err
= set_anon_super(s
, data
);
969 * subvolumes are identified by ino 256
971 static inline int is_subvolume_inode(struct inode
*inode
)
973 if (inode
&& inode
->i_ino
== BTRFS_FIRST_FREE_OBJECTID
)
979 * This will strip out the subvol=%s argument for an argument string and add
980 * subvolid=0 to make sure we get the actual tree root for path walking to the
983 static char *setup_root_args(char *args
)
985 unsigned len
= strlen(args
) + 2 + 1;
986 char *src
, *dst
, *buf
;
989 * We need the same args as before, but with this substitution:
990 * s!subvol=[^,]+!subvolid=0!
992 * Since the replacement string is up to 2 bytes longer than the
993 * original, allocate strlen(args) + 2 + 1 bytes.
996 src
= strstr(args
, "subvol=");
997 /* This shouldn't happen, but just in case.. */
1001 buf
= dst
= kmalloc(len
, GFP_NOFS
);
1006 * If the subvol= arg is not at the start of the string,
1007 * copy whatever precedes it into buf.
1012 dst
+= strlen(args
);
1015 strcpy(dst
, "subvolid=0");
1016 dst
+= strlen("subvolid=0");
1019 * If there is a "," after the original subvol=... string,
1020 * copy that suffix into our buffer. Otherwise, we're done.
1022 src
= strchr(src
, ',');
1029 static struct dentry
*mount_subvol(const char *subvol_name
, int flags
,
1030 const char *device_name
, char *data
)
1032 struct dentry
*root
;
1033 struct vfsmount
*mnt
;
1036 newargs
= setup_root_args(data
);
1038 return ERR_PTR(-ENOMEM
);
1039 mnt
= vfs_kern_mount(&btrfs_fs_type
, flags
, device_name
,
1043 return ERR_CAST(mnt
);
1045 root
= mount_subtree(mnt
, subvol_name
);
1047 if (!IS_ERR(root
) && !is_subvolume_inode(root
->d_inode
)) {
1048 struct super_block
*s
= root
->d_sb
;
1050 root
= ERR_PTR(-EINVAL
);
1051 deactivate_locked_super(s
);
1052 printk(KERN_ERR
"btrfs: '%s' is not a valid subvolume\n",
1060 * Find a superblock for the given device / mount point.
1062 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1063 * for multiple device setup. Make sure to keep it in sync.
1065 static struct dentry
*btrfs_mount(struct file_system_type
*fs_type
, int flags
,
1066 const char *device_name
, void *data
)
1068 struct block_device
*bdev
= NULL
;
1069 struct super_block
*s
;
1070 struct dentry
*root
;
1071 struct btrfs_fs_devices
*fs_devices
= NULL
;
1072 struct btrfs_fs_info
*fs_info
= NULL
;
1073 fmode_t mode
= FMODE_READ
;
1074 char *subvol_name
= NULL
;
1075 u64 subvol_objectid
= 0;
1076 u64 subvol_rootid
= 0;
1079 if (!(flags
& MS_RDONLY
))
1080 mode
|= FMODE_WRITE
;
1082 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
1083 &subvol_name
, &subvol_objectid
,
1084 &subvol_rootid
, &fs_devices
);
1087 return ERR_PTR(error
);
1091 root
= mount_subvol(subvol_name
, flags
, device_name
, data
);
1096 error
= btrfs_scan_one_device(device_name
, mode
, fs_type
, &fs_devices
);
1098 return ERR_PTR(error
);
1101 * Setup a dummy root and fs_info for test/set super. This is because
1102 * we don't actually fill this stuff out until open_ctree, but we need
1103 * it for searching for existing supers, so this lets us do that and
1104 * then open_ctree will properly initialize everything later.
1106 fs_info
= kzalloc(sizeof(struct btrfs_fs_info
), GFP_NOFS
);
1108 return ERR_PTR(-ENOMEM
);
1110 fs_info
->fs_devices
= fs_devices
;
1112 fs_info
->super_copy
= kzalloc(BTRFS_SUPER_INFO_SIZE
, GFP_NOFS
);
1113 fs_info
->super_for_commit
= kzalloc(BTRFS_SUPER_INFO_SIZE
, GFP_NOFS
);
1114 if (!fs_info
->super_copy
|| !fs_info
->super_for_commit
) {
1119 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
1123 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
1125 goto error_close_devices
;
1128 bdev
= fs_devices
->latest_bdev
;
1129 s
= sget(fs_type
, btrfs_test_super
, btrfs_set_super
, flags
| MS_NOSEC
,
1133 goto error_close_devices
;
1137 btrfs_close_devices(fs_devices
);
1138 free_fs_info(fs_info
);
1139 if ((flags
^ s
->s_flags
) & MS_RDONLY
)
1142 char b
[BDEVNAME_SIZE
];
1144 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
1145 btrfs_sb(s
)->bdev_holder
= fs_type
;
1146 error
= btrfs_fill_super(s
, fs_devices
, data
,
1147 flags
& MS_SILENT
? 1 : 0);
1150 root
= !error
? get_default_root(s
, subvol_objectid
) : ERR_PTR(error
);
1152 deactivate_locked_super(s
);
1156 error_close_devices
:
1157 btrfs_close_devices(fs_devices
);
1159 free_fs_info(fs_info
);
1160 return ERR_PTR(error
);
1163 static void btrfs_set_max_workers(struct btrfs_workers
*workers
, int new_limit
)
1165 spin_lock_irq(&workers
->lock
);
1166 workers
->max_workers
= new_limit
;
1167 spin_unlock_irq(&workers
->lock
);
1170 static void btrfs_resize_thread_pool(struct btrfs_fs_info
*fs_info
,
1171 int new_pool_size
, int old_pool_size
)
1173 if (new_pool_size
== old_pool_size
)
1176 fs_info
->thread_pool_size
= new_pool_size
;
1178 printk(KERN_INFO
"btrfs: resize thread pool %d -> %d\n",
1179 old_pool_size
, new_pool_size
);
1181 btrfs_set_max_workers(&fs_info
->generic_worker
, new_pool_size
);
1182 btrfs_set_max_workers(&fs_info
->workers
, new_pool_size
);
1183 btrfs_set_max_workers(&fs_info
->delalloc_workers
, new_pool_size
);
1184 btrfs_set_max_workers(&fs_info
->submit_workers
, new_pool_size
);
1185 btrfs_set_max_workers(&fs_info
->caching_workers
, new_pool_size
);
1186 btrfs_set_max_workers(&fs_info
->fixup_workers
, new_pool_size
);
1187 btrfs_set_max_workers(&fs_info
->endio_workers
, new_pool_size
);
1188 btrfs_set_max_workers(&fs_info
->endio_meta_workers
, new_pool_size
);
1189 btrfs_set_max_workers(&fs_info
->endio_meta_write_workers
, new_pool_size
);
1190 btrfs_set_max_workers(&fs_info
->endio_write_workers
, new_pool_size
);
1191 btrfs_set_max_workers(&fs_info
->endio_freespace_worker
, new_pool_size
);
1192 btrfs_set_max_workers(&fs_info
->delayed_workers
, new_pool_size
);
1193 btrfs_set_max_workers(&fs_info
->readahead_workers
, new_pool_size
);
1194 btrfs_set_max_workers(&fs_info
->scrub_wr_completion_workers
,
1198 static inline void btrfs_remount_prepare(struct btrfs_fs_info
*fs_info
,
1199 unsigned long old_opts
, int flags
)
1201 set_bit(BTRFS_FS_STATE_REMOUNTING
, &fs_info
->fs_state
);
1203 if (btrfs_raw_test_opt(old_opts
, AUTO_DEFRAG
) &&
1204 (!btrfs_raw_test_opt(fs_info
->mount_opt
, AUTO_DEFRAG
) ||
1205 (flags
& MS_RDONLY
))) {
1206 /* wait for any defraggers to finish */
1207 wait_event(fs_info
->transaction_wait
,
1208 (atomic_read(&fs_info
->defrag_running
) == 0));
1209 if (flags
& MS_RDONLY
)
1210 sync_filesystem(fs_info
->sb
);
1214 static inline void btrfs_remount_cleanup(struct btrfs_fs_info
*fs_info
,
1215 unsigned long old_opts
)
1218 * We need cleanup all defragable inodes if the autodefragment is
1219 * close or the fs is R/O.
1221 if (btrfs_raw_test_opt(old_opts
, AUTO_DEFRAG
) &&
1222 (!btrfs_raw_test_opt(fs_info
->mount_opt
, AUTO_DEFRAG
) ||
1223 (fs_info
->sb
->s_flags
& MS_RDONLY
))) {
1224 btrfs_cleanup_defrag_inodes(fs_info
);
1227 clear_bit(BTRFS_FS_STATE_REMOUNTING
, &fs_info
->fs_state
);
1230 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
1232 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1233 struct btrfs_root
*root
= fs_info
->tree_root
;
1234 unsigned old_flags
= sb
->s_flags
;
1235 unsigned long old_opts
= fs_info
->mount_opt
;
1236 unsigned long old_compress_type
= fs_info
->compress_type
;
1237 u64 old_max_inline
= fs_info
->max_inline
;
1238 u64 old_alloc_start
= fs_info
->alloc_start
;
1239 int old_thread_pool_size
= fs_info
->thread_pool_size
;
1240 unsigned int old_metadata_ratio
= fs_info
->metadata_ratio
;
1243 btrfs_remount_prepare(fs_info
, old_opts
, *flags
);
1245 ret
= btrfs_parse_options(root
, data
);
1251 btrfs_resize_thread_pool(fs_info
,
1252 fs_info
->thread_pool_size
, old_thread_pool_size
);
1254 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
1257 if (*flags
& MS_RDONLY
) {
1259 * this also happens on 'umount -rf' or on shutdown, when
1260 * the filesystem is busy.
1262 sb
->s_flags
|= MS_RDONLY
;
1264 btrfs_dev_replace_suspend_for_unmount(fs_info
);
1265 btrfs_scrub_cancel(fs_info
);
1267 ret
= btrfs_commit_super(root
);
1271 if (fs_info
->fs_devices
->rw_devices
== 0) {
1276 if (fs_info
->fs_devices
->missing_devices
>
1277 fs_info
->num_tolerated_disk_barrier_failures
&&
1278 !(*flags
& MS_RDONLY
)) {
1280 "Btrfs: too many missing devices, writeable remount is not allowed\n");
1285 if (btrfs_super_log_root(fs_info
->super_copy
) != 0) {
1290 ret
= btrfs_cleanup_fs_roots(fs_info
);
1294 /* recover relocation */
1295 ret
= btrfs_recover_relocation(root
);
1299 ret
= btrfs_resume_balance_async(fs_info
);
1303 ret
= btrfs_resume_dev_replace_async(fs_info
);
1305 pr_warn("btrfs: failed to resume dev_replace\n");
1308 sb
->s_flags
&= ~MS_RDONLY
;
1311 btrfs_remount_cleanup(fs_info
, old_opts
);
1315 /* We've hit an error - don't reset MS_RDONLY */
1316 if (sb
->s_flags
& MS_RDONLY
)
1317 old_flags
|= MS_RDONLY
;
1318 sb
->s_flags
= old_flags
;
1319 fs_info
->mount_opt
= old_opts
;
1320 fs_info
->compress_type
= old_compress_type
;
1321 fs_info
->max_inline
= old_max_inline
;
1322 mutex_lock(&fs_info
->chunk_mutex
);
1323 fs_info
->alloc_start
= old_alloc_start
;
1324 mutex_unlock(&fs_info
->chunk_mutex
);
1325 btrfs_resize_thread_pool(fs_info
,
1326 old_thread_pool_size
, fs_info
->thread_pool_size
);
1327 fs_info
->metadata_ratio
= old_metadata_ratio
;
1328 btrfs_remount_cleanup(fs_info
, old_opts
);
1332 /* Used to sort the devices by max_avail(descending sort) */
1333 static int btrfs_cmp_device_free_bytes(const void *dev_info1
,
1334 const void *dev_info2
)
1336 if (((struct btrfs_device_info
*)dev_info1
)->max_avail
>
1337 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
1339 else if (((struct btrfs_device_info
*)dev_info1
)->max_avail
<
1340 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
1347 * sort the devices by max_avail, in which max free extent size of each device
1348 * is stored.(Descending Sort)
1350 static inline void btrfs_descending_sort_devices(
1351 struct btrfs_device_info
*devices
,
1354 sort(devices
, nr_devices
, sizeof(struct btrfs_device_info
),
1355 btrfs_cmp_device_free_bytes
, NULL
);
1359 * The helper to calc the free space on the devices that can be used to store
1362 static int btrfs_calc_avail_data_space(struct btrfs_root
*root
, u64
*free_bytes
)
1364 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
1365 struct btrfs_device_info
*devices_info
;
1366 struct btrfs_fs_devices
*fs_devices
= fs_info
->fs_devices
;
1367 struct btrfs_device
*device
;
1372 u64 min_stripe_size
;
1373 int min_stripes
= 1, num_stripes
= 1;
1374 int i
= 0, nr_devices
;
1377 nr_devices
= fs_info
->fs_devices
->open_devices
;
1378 BUG_ON(!nr_devices
);
1380 devices_info
= kmalloc(sizeof(*devices_info
) * nr_devices
,
1385 /* calc min stripe number for data space alloction */
1386 type
= btrfs_get_alloc_profile(root
, 1);
1387 if (type
& BTRFS_BLOCK_GROUP_RAID0
) {
1389 num_stripes
= nr_devices
;
1390 } else if (type
& BTRFS_BLOCK_GROUP_RAID1
) {
1393 } else if (type
& BTRFS_BLOCK_GROUP_RAID10
) {
1398 if (type
& BTRFS_BLOCK_GROUP_DUP
)
1399 min_stripe_size
= 2 * BTRFS_STRIPE_LEN
;
1401 min_stripe_size
= BTRFS_STRIPE_LEN
;
1403 list_for_each_entry(device
, &fs_devices
->devices
, dev_list
) {
1404 if (!device
->in_fs_metadata
|| !device
->bdev
||
1405 device
->is_tgtdev_for_dev_replace
)
1408 avail_space
= device
->total_bytes
- device
->bytes_used
;
1410 /* align with stripe_len */
1411 do_div(avail_space
, BTRFS_STRIPE_LEN
);
1412 avail_space
*= BTRFS_STRIPE_LEN
;
1415 * In order to avoid overwritting the superblock on the drive,
1416 * btrfs starts at an offset of at least 1MB when doing chunk
1419 skip_space
= 1024 * 1024;
1421 /* user can set the offset in fs_info->alloc_start. */
1422 if (fs_info
->alloc_start
+ BTRFS_STRIPE_LEN
<=
1423 device
->total_bytes
)
1424 skip_space
= max(fs_info
->alloc_start
, skip_space
);
1427 * btrfs can not use the free space in [0, skip_space - 1],
1428 * we must subtract it from the total. In order to implement
1429 * it, we account the used space in this range first.
1431 ret
= btrfs_account_dev_extents_size(device
, 0, skip_space
- 1,
1434 kfree(devices_info
);
1438 /* calc the free space in [0, skip_space - 1] */
1439 skip_space
-= used_space
;
1442 * we can use the free space in [0, skip_space - 1], subtract
1443 * it from the total.
1445 if (avail_space
&& avail_space
>= skip_space
)
1446 avail_space
-= skip_space
;
1450 if (avail_space
< min_stripe_size
)
1453 devices_info
[i
].dev
= device
;
1454 devices_info
[i
].max_avail
= avail_space
;
1461 btrfs_descending_sort_devices(devices_info
, nr_devices
);
1465 while (nr_devices
>= min_stripes
) {
1466 if (num_stripes
> nr_devices
)
1467 num_stripes
= nr_devices
;
1469 if (devices_info
[i
].max_avail
>= min_stripe_size
) {
1473 avail_space
+= devices_info
[i
].max_avail
* num_stripes
;
1474 alloc_size
= devices_info
[i
].max_avail
;
1475 for (j
= i
+ 1 - num_stripes
; j
<= i
; j
++)
1476 devices_info
[j
].max_avail
-= alloc_size
;
1482 kfree(devices_info
);
1483 *free_bytes
= avail_space
;
1487 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
1489 struct btrfs_fs_info
*fs_info
= btrfs_sb(dentry
->d_sb
);
1490 struct btrfs_super_block
*disk_super
= fs_info
->super_copy
;
1491 struct list_head
*head
= &fs_info
->space_info
;
1492 struct btrfs_space_info
*found
;
1494 u64 total_free_data
= 0;
1495 int bits
= dentry
->d_sb
->s_blocksize_bits
;
1496 __be32
*fsid
= (__be32
*)fs_info
->fsid
;
1499 /* holding chunk_muext to avoid allocating new chunks */
1500 mutex_lock(&fs_info
->chunk_mutex
);
1502 list_for_each_entry_rcu(found
, head
, list
) {
1503 if (found
->flags
& BTRFS_BLOCK_GROUP_DATA
) {
1504 total_free_data
+= found
->disk_total
- found
->disk_used
;
1506 btrfs_account_ro_block_groups_free_space(found
);
1509 total_used
+= found
->disk_used
;
1513 buf
->f_namelen
= BTRFS_NAME_LEN
;
1514 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
1515 buf
->f_bfree
= buf
->f_blocks
- (total_used
>> bits
);
1516 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
1517 buf
->f_type
= BTRFS_SUPER_MAGIC
;
1518 buf
->f_bavail
= total_free_data
;
1519 ret
= btrfs_calc_avail_data_space(fs_info
->tree_root
, &total_free_data
);
1521 mutex_unlock(&fs_info
->chunk_mutex
);
1524 buf
->f_bavail
+= total_free_data
;
1525 buf
->f_bavail
= buf
->f_bavail
>> bits
;
1526 mutex_unlock(&fs_info
->chunk_mutex
);
1528 /* We treat it as constant endianness (it doesn't matter _which_)
1529 because we want the fsid to come out the same whether mounted
1530 on a big-endian or little-endian host */
1531 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
1532 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
1533 /* Mask in the root object ID too, to disambiguate subvols */
1534 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
1535 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
1540 static void btrfs_kill_super(struct super_block
*sb
)
1542 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1543 kill_anon_super(sb
);
1544 free_fs_info(fs_info
);
1547 static struct file_system_type btrfs_fs_type
= {
1548 .owner
= THIS_MODULE
,
1550 .mount
= btrfs_mount
,
1551 .kill_sb
= btrfs_kill_super
,
1552 .fs_flags
= FS_REQUIRES_DEV
,
1554 MODULE_ALIAS_FS("btrfs");
1557 * used by btrfsctl to scan devices when no FS is mounted
1559 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
1562 struct btrfs_ioctl_vol_args
*vol
;
1563 struct btrfs_fs_devices
*fs_devices
;
1566 if (!capable(CAP_SYS_ADMIN
))
1569 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
1571 return PTR_ERR(vol
);
1574 case BTRFS_IOC_SCAN_DEV
:
1575 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
1576 &btrfs_fs_type
, &fs_devices
);
1578 case BTRFS_IOC_DEVICES_READY
:
1579 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
1580 &btrfs_fs_type
, &fs_devices
);
1583 ret
= !(fs_devices
->num_devices
== fs_devices
->total_devices
);
1591 static int btrfs_freeze(struct super_block
*sb
)
1593 struct btrfs_trans_handle
*trans
;
1594 struct btrfs_root
*root
= btrfs_sb(sb
)->tree_root
;
1596 trans
= btrfs_attach_transaction_barrier(root
);
1597 if (IS_ERR(trans
)) {
1598 /* no transaction, don't bother */
1599 if (PTR_ERR(trans
) == -ENOENT
)
1601 return PTR_ERR(trans
);
1603 return btrfs_commit_transaction(trans
, root
);
1606 static int btrfs_unfreeze(struct super_block
*sb
)
1611 static int btrfs_show_devname(struct seq_file
*m
, struct dentry
*root
)
1613 struct btrfs_fs_info
*fs_info
= btrfs_sb(root
->d_sb
);
1614 struct btrfs_fs_devices
*cur_devices
;
1615 struct btrfs_device
*dev
, *first_dev
= NULL
;
1616 struct list_head
*head
;
1617 struct rcu_string
*name
;
1619 mutex_lock(&fs_info
->fs_devices
->device_list_mutex
);
1620 cur_devices
= fs_info
->fs_devices
;
1621 while (cur_devices
) {
1622 head
= &cur_devices
->devices
;
1623 list_for_each_entry(dev
, head
, dev_list
) {
1626 if (!first_dev
|| dev
->devid
< first_dev
->devid
)
1629 cur_devices
= cur_devices
->seed
;
1634 name
= rcu_dereference(first_dev
->name
);
1635 seq_escape(m
, name
->str
, " \t\n\\");
1640 mutex_unlock(&fs_info
->fs_devices
->device_list_mutex
);
1644 static const struct super_operations btrfs_super_ops
= {
1645 .drop_inode
= btrfs_drop_inode
,
1646 .evict_inode
= btrfs_evict_inode
,
1647 .put_super
= btrfs_put_super
,
1648 .sync_fs
= btrfs_sync_fs
,
1649 .show_options
= btrfs_show_options
,
1650 .show_devname
= btrfs_show_devname
,
1651 .write_inode
= btrfs_write_inode
,
1652 .alloc_inode
= btrfs_alloc_inode
,
1653 .destroy_inode
= btrfs_destroy_inode
,
1654 .statfs
= btrfs_statfs
,
1655 .remount_fs
= btrfs_remount
,
1656 .freeze_fs
= btrfs_freeze
,
1657 .unfreeze_fs
= btrfs_unfreeze
,
1660 static const struct file_operations btrfs_ctl_fops
= {
1661 .unlocked_ioctl
= btrfs_control_ioctl
,
1662 .compat_ioctl
= btrfs_control_ioctl
,
1663 .owner
= THIS_MODULE
,
1664 .llseek
= noop_llseek
,
1667 static struct miscdevice btrfs_misc
= {
1668 .minor
= BTRFS_MINOR
,
1669 .name
= "btrfs-control",
1670 .fops
= &btrfs_ctl_fops
1673 MODULE_ALIAS_MISCDEV(BTRFS_MINOR
);
1674 MODULE_ALIAS("devname:btrfs-control");
1676 static int btrfs_interface_init(void)
1678 return misc_register(&btrfs_misc
);
1681 static void btrfs_interface_exit(void)
1683 if (misc_deregister(&btrfs_misc
) < 0)
1684 printk(KERN_INFO
"btrfs: misc_deregister failed for control device\n");
1687 static int __init
init_btrfs_fs(void)
1691 err
= btrfs_init_sysfs();
1695 btrfs_init_compress();
1697 err
= btrfs_init_cachep();
1701 err
= extent_io_init();
1705 err
= extent_map_init();
1707 goto free_extent_io
;
1709 err
= ordered_data_init();
1711 goto free_extent_map
;
1713 err
= btrfs_delayed_inode_init();
1715 goto free_ordered_data
;
1717 err
= btrfs_auto_defrag_init();
1719 goto free_delayed_inode
;
1721 err
= btrfs_delayed_ref_init();
1723 goto free_auto_defrag
;
1725 err
= btrfs_interface_init();
1727 goto free_delayed_ref
;
1729 err
= register_filesystem(&btrfs_fs_type
);
1731 goto unregister_ioctl
;
1733 btrfs_init_lockdep();
1735 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
1736 btrfs_test_free_space_cache();
1739 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
1743 btrfs_interface_exit();
1745 btrfs_delayed_ref_exit();
1747 btrfs_auto_defrag_exit();
1749 btrfs_delayed_inode_exit();
1751 ordered_data_exit();
1757 btrfs_destroy_cachep();
1759 btrfs_exit_compress();
1764 static void __exit
exit_btrfs_fs(void)
1766 btrfs_destroy_cachep();
1767 btrfs_delayed_ref_exit();
1768 btrfs_auto_defrag_exit();
1769 btrfs_delayed_inode_exit();
1770 ordered_data_exit();
1773 btrfs_interface_exit();
1774 unregister_filesystem(&btrfs_fs_type
);
1776 btrfs_cleanup_fs_uuids();
1777 btrfs_exit_compress();
1780 module_init(init_btrfs_fs
)
1781 module_exit(exit_btrfs_fs
)
1783 MODULE_LICENSE("GPL");