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
, char nbuf
[16])
73 errstr
= "IO failure";
76 errstr
= "Out of memory";
79 errstr
= "Readonly filesystem";
82 errstr
= "Object already exists";
86 if (snprintf(nbuf
, 16, "error %d", -errno
) >= 0)
95 static void __save_error_info(struct btrfs_fs_info
*fs_info
)
98 * today we only save the error info into ram. Long term we'll
99 * also send it down to the disk
101 set_bit(BTRFS_FS_STATE_ERROR
, &fs_info
->fs_state
);
104 static void save_error_info(struct btrfs_fs_info
*fs_info
)
106 __save_error_info(fs_info
);
109 /* btrfs handle error by forcing the filesystem readonly */
110 static void btrfs_handle_error(struct btrfs_fs_info
*fs_info
)
112 struct super_block
*sb
= fs_info
->sb
;
114 if (sb
->s_flags
& MS_RDONLY
)
117 if (test_bit(BTRFS_FS_STATE_ERROR
, &fs_info
->fs_state
)) {
118 sb
->s_flags
|= MS_RDONLY
;
119 printk(KERN_INFO
"btrfs is forced readonly\n");
121 * Note that a running device replace operation is not
122 * canceled here although there is no way to update
123 * the progress. It would add the risk of a deadlock,
124 * therefore the canceling is ommited. The only penalty
125 * is that some I/O remains active until the procedure
126 * completes. The next time when the filesystem is
127 * mounted writeable again, the device replace
128 * operation continues.
136 * __btrfs_std_error decodes expected errors from the caller and
137 * invokes the approciate error response.
139 void __btrfs_std_error(struct btrfs_fs_info
*fs_info
, const char *function
,
140 unsigned int line
, int errno
, const char *fmt
, ...)
142 struct super_block
*sb
= fs_info
->sb
;
147 * Special case: if the error is EROFS, and we're already
148 * under MS_RDONLY, then it is safe here.
150 if (errno
== -EROFS
&& (sb
->s_flags
& MS_RDONLY
))
153 errstr
= btrfs_decode_error(errno
, nbuf
);
155 struct va_format vaf
;
162 printk(KERN_CRIT
"BTRFS error (device %s) in %s:%d: %s (%pV)\n",
163 sb
->s_id
, function
, line
, errstr
, &vaf
);
166 printk(KERN_CRIT
"BTRFS error (device %s) in %s:%d: %s\n",
167 sb
->s_id
, function
, line
, errstr
);
170 /* Don't go through full error handling during mount */
171 if (sb
->s_flags
& MS_BORN
) {
172 save_error_info(fs_info
);
173 btrfs_handle_error(fs_info
);
177 static const char * const logtypes
[] = {
188 void btrfs_printk(struct btrfs_fs_info
*fs_info
, const char *fmt
, ...)
190 struct super_block
*sb
= fs_info
->sb
;
192 struct va_format vaf
;
194 const char *type
= logtypes
[4];
199 kern_level
= printk_get_level(fmt
);
201 size_t size
= printk_skip_level(fmt
) - fmt
;
202 memcpy(lvl
, fmt
, size
);
205 type
= logtypes
[kern_level
- '0'];
212 printk("%sBTRFS %s (device %s): %pV", lvl
, type
, sb
->s_id
, &vaf
);
219 void __btrfs_std_error(struct btrfs_fs_info
*fs_info
, const char *function
,
220 unsigned int line
, int errno
, const char *fmt
, ...)
222 struct super_block
*sb
= fs_info
->sb
;
225 * Special case: if the error is EROFS, and we're already
226 * under MS_RDONLY, then it is safe here.
228 if (errno
== -EROFS
&& (sb
->s_flags
& MS_RDONLY
))
231 /* Don't go through full error handling during mount */
232 if (sb
->s_flags
& MS_BORN
) {
233 save_error_info(fs_info
);
234 btrfs_handle_error(fs_info
);
240 * We only mark the transaction aborted and then set the file system read-only.
241 * This will prevent new transactions from starting or trying to join this
244 * This means that error recovery at the call site is limited to freeing
245 * any local memory allocations and passing the error code up without
246 * further cleanup. The transaction should complete as it normally would
247 * in the call path but will return -EIO.
249 * We'll complete the cleanup in btrfs_end_transaction and
250 * btrfs_commit_transaction.
252 void __btrfs_abort_transaction(struct btrfs_trans_handle
*trans
,
253 struct btrfs_root
*root
, const char *function
,
254 unsigned int line
, int errno
)
256 WARN_ONCE(1, KERN_DEBUG
"btrfs: Transaction aborted\n");
257 trans
->aborted
= errno
;
258 /* Nothing used. The other threads that have joined this
259 * transaction may be able to continue. */
260 if (!trans
->blocks_used
) {
264 errstr
= btrfs_decode_error(errno
, nbuf
);
265 btrfs_printk(root
->fs_info
,
266 "%s:%d: Aborting unused transaction(%s).\n",
267 function
, line
, errstr
);
270 ACCESS_ONCE(trans
->transaction
->aborted
) = errno
;
271 __btrfs_std_error(root
->fs_info
, function
, line
, errno
, NULL
);
274 * __btrfs_panic decodes unexpected, fatal errors from the caller,
275 * issues an alert, and either panics or BUGs, depending on mount options.
277 void __btrfs_panic(struct btrfs_fs_info
*fs_info
, const char *function
,
278 unsigned int line
, int errno
, const char *fmt
, ...)
281 char *s_id
= "<unknown>";
283 struct va_format vaf
= { .fmt
= fmt
};
287 s_id
= fs_info
->sb
->s_id
;
292 errstr
= btrfs_decode_error(errno
, nbuf
);
293 if (fs_info
&& (fs_info
->mount_opt
& BTRFS_MOUNT_PANIC_ON_FATAL_ERROR
))
294 panic(KERN_CRIT
"BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
295 s_id
, function
, line
, &vaf
, errstr
);
297 printk(KERN_CRIT
"BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
298 s_id
, function
, line
, &vaf
, errstr
);
300 /* Caller calls BUG() */
303 static void btrfs_put_super(struct super_block
*sb
)
305 (void)close_ctree(btrfs_sb(sb
)->tree_root
);
306 /* FIXME: need to fix VFS to return error? */
307 /* AV: return it _where_? ->put_super() can be triggered by any number
308 * of async events, up to and including delivery of SIGKILL to the
309 * last process that kept it busy. Or segfault in the aforementioned
310 * process... Whom would you report that to?
315 Opt_degraded
, Opt_subvol
, Opt_subvolid
, Opt_device
, Opt_nodatasum
,
316 Opt_nodatacow
, Opt_max_inline
, Opt_alloc_start
, Opt_nobarrier
, Opt_ssd
,
317 Opt_nossd
, Opt_ssd_spread
, Opt_thread_pool
, Opt_noacl
, Opt_compress
,
318 Opt_compress_type
, Opt_compress_force
, Opt_compress_force_type
,
319 Opt_notreelog
, Opt_ratio
, Opt_flushoncommit
, Opt_discard
,
320 Opt_space_cache
, Opt_clear_cache
, Opt_user_subvol_rm_allowed
,
321 Opt_enospc_debug
, Opt_subvolrootid
, Opt_defrag
, Opt_inode_cache
,
322 Opt_no_space_cache
, Opt_recovery
, Opt_skip_balance
,
323 Opt_check_integrity
, Opt_check_integrity_including_extent_data
,
324 Opt_check_integrity_print_mask
, Opt_fatal_errors
,
328 static match_table_t tokens
= {
329 {Opt_degraded
, "degraded"},
330 {Opt_subvol
, "subvol=%s"},
331 {Opt_subvolid
, "subvolid=%d"},
332 {Opt_device
, "device=%s"},
333 {Opt_nodatasum
, "nodatasum"},
334 {Opt_nodatacow
, "nodatacow"},
335 {Opt_nobarrier
, "nobarrier"},
336 {Opt_max_inline
, "max_inline=%s"},
337 {Opt_alloc_start
, "alloc_start=%s"},
338 {Opt_thread_pool
, "thread_pool=%d"},
339 {Opt_compress
, "compress"},
340 {Opt_compress_type
, "compress=%s"},
341 {Opt_compress_force
, "compress-force"},
342 {Opt_compress_force_type
, "compress-force=%s"},
344 {Opt_ssd_spread
, "ssd_spread"},
345 {Opt_nossd
, "nossd"},
346 {Opt_noacl
, "noacl"},
347 {Opt_notreelog
, "notreelog"},
348 {Opt_flushoncommit
, "flushoncommit"},
349 {Opt_ratio
, "metadata_ratio=%d"},
350 {Opt_discard
, "discard"},
351 {Opt_space_cache
, "space_cache"},
352 {Opt_clear_cache
, "clear_cache"},
353 {Opt_user_subvol_rm_allowed
, "user_subvol_rm_allowed"},
354 {Opt_enospc_debug
, "enospc_debug"},
355 {Opt_subvolrootid
, "subvolrootid=%d"},
356 {Opt_defrag
, "autodefrag"},
357 {Opt_inode_cache
, "inode_cache"},
358 {Opt_no_space_cache
, "nospace_cache"},
359 {Opt_recovery
, "recovery"},
360 {Opt_skip_balance
, "skip_balance"},
361 {Opt_check_integrity
, "check_int"},
362 {Opt_check_integrity_including_extent_data
, "check_int_data"},
363 {Opt_check_integrity_print_mask
, "check_int_print_mask=%d"},
364 {Opt_fatal_errors
, "fatal_errors=%s"},
369 * Regular mount options parser. Everything that is needed only when
370 * reading in a new superblock is parsed here.
371 * XXX JDM: This needs to be cleaned up for remount.
373 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
375 struct btrfs_fs_info
*info
= root
->fs_info
;
376 substring_t args
[MAX_OPT_ARGS
];
377 char *p
, *num
, *orig
= NULL
;
382 bool compress_force
= false;
384 cache_gen
= btrfs_super_cache_generation(root
->fs_info
->super_copy
);
386 btrfs_set_opt(info
->mount_opt
, SPACE_CACHE
);
392 * strsep changes the string, duplicate it because parse_options
395 options
= kstrdup(options
, GFP_NOFS
);
401 while ((p
= strsep(&options
, ",")) != NULL
) {
406 token
= match_token(p
, tokens
, args
);
409 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
410 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
414 case Opt_subvolrootid
:
417 * These are parsed by btrfs_parse_early_options
418 * and can be happily ignored here.
422 printk(KERN_INFO
"btrfs: setting nodatasum\n");
423 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
426 if (!btrfs_test_opt(root
, COMPRESS
) ||
427 !btrfs_test_opt(root
, FORCE_COMPRESS
)) {
428 printk(KERN_INFO
"btrfs: setting nodatacow, compression disabled\n");
430 printk(KERN_INFO
"btrfs: setting nodatacow\n");
432 info
->compress_type
= BTRFS_COMPRESS_NONE
;
433 btrfs_clear_opt(info
->mount_opt
, COMPRESS
);
434 btrfs_clear_opt(info
->mount_opt
, FORCE_COMPRESS
);
435 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
436 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
438 case Opt_compress_force
:
439 case Opt_compress_force_type
:
440 compress_force
= true;
443 case Opt_compress_type
:
444 if (token
== Opt_compress
||
445 token
== Opt_compress_force
||
446 strcmp(args
[0].from
, "zlib") == 0) {
447 compress_type
= "zlib";
448 info
->compress_type
= BTRFS_COMPRESS_ZLIB
;
449 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
450 btrfs_clear_opt(info
->mount_opt
, NODATACOW
);
451 btrfs_clear_opt(info
->mount_opt
, NODATASUM
);
452 } else if (strcmp(args
[0].from
, "lzo") == 0) {
453 compress_type
= "lzo";
454 info
->compress_type
= BTRFS_COMPRESS_LZO
;
455 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
456 btrfs_clear_opt(info
->mount_opt
, NODATACOW
);
457 btrfs_clear_opt(info
->mount_opt
, NODATASUM
);
458 btrfs_set_fs_incompat(info
, COMPRESS_LZO
);
459 } else if (strncmp(args
[0].from
, "no", 2) == 0) {
460 compress_type
= "no";
461 info
->compress_type
= BTRFS_COMPRESS_NONE
;
462 btrfs_clear_opt(info
->mount_opt
, COMPRESS
);
463 btrfs_clear_opt(info
->mount_opt
, FORCE_COMPRESS
);
464 compress_force
= false;
470 if (compress_force
) {
471 btrfs_set_opt(info
->mount_opt
, FORCE_COMPRESS
);
472 pr_info("btrfs: force %s compression\n",
475 pr_info("btrfs: use %s compression\n",
479 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
480 btrfs_set_opt(info
->mount_opt
, SSD
);
483 printk(KERN_INFO
"btrfs: use spread ssd "
484 "allocation scheme\n");
485 btrfs_set_opt(info
->mount_opt
, SSD
);
486 btrfs_set_opt(info
->mount_opt
, SSD_SPREAD
);
489 printk(KERN_INFO
"btrfs: not using ssd allocation "
491 btrfs_set_opt(info
->mount_opt
, NOSSD
);
492 btrfs_clear_opt(info
->mount_opt
, SSD
);
493 btrfs_clear_opt(info
->mount_opt
, SSD_SPREAD
);
496 printk(KERN_INFO
"btrfs: turning off barriers\n");
497 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
499 case Opt_thread_pool
:
501 match_int(&args
[0], &intarg
);
503 info
->thread_pool_size
= intarg
;
506 num
= match_strdup(&args
[0]);
508 info
->max_inline
= memparse(num
, NULL
);
511 if (info
->max_inline
) {
512 info
->max_inline
= max_t(u64
,
516 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
517 (unsigned long long)info
->max_inline
);
520 case Opt_alloc_start
:
521 num
= match_strdup(&args
[0]);
523 mutex_lock(&info
->chunk_mutex
);
524 info
->alloc_start
= memparse(num
, NULL
);
525 mutex_unlock(&info
->chunk_mutex
);
528 "btrfs: allocations start at %llu\n",
529 (unsigned long long)info
->alloc_start
);
533 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
536 printk(KERN_INFO
"btrfs: disabling tree log\n");
537 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
539 case Opt_flushoncommit
:
540 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
541 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
545 match_int(&args
[0], &intarg
);
547 info
->metadata_ratio
= intarg
;
548 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
549 info
->metadata_ratio
);
553 btrfs_set_opt(info
->mount_opt
, DISCARD
);
555 case Opt_space_cache
:
556 btrfs_set_opt(info
->mount_opt
, SPACE_CACHE
);
558 case Opt_no_space_cache
:
559 printk(KERN_INFO
"btrfs: disabling disk space caching\n");
560 btrfs_clear_opt(info
->mount_opt
, SPACE_CACHE
);
562 case Opt_inode_cache
:
563 printk(KERN_INFO
"btrfs: enabling inode map caching\n");
564 btrfs_set_opt(info
->mount_opt
, INODE_MAP_CACHE
);
566 case Opt_clear_cache
:
567 printk(KERN_INFO
"btrfs: force clearing of disk cache\n");
568 btrfs_set_opt(info
->mount_opt
, CLEAR_CACHE
);
570 case Opt_user_subvol_rm_allowed
:
571 btrfs_set_opt(info
->mount_opt
, USER_SUBVOL_RM_ALLOWED
);
573 case Opt_enospc_debug
:
574 btrfs_set_opt(info
->mount_opt
, ENOSPC_DEBUG
);
577 printk(KERN_INFO
"btrfs: enabling auto defrag\n");
578 btrfs_set_opt(info
->mount_opt
, AUTO_DEFRAG
);
581 printk(KERN_INFO
"btrfs: enabling auto recovery\n");
582 btrfs_set_opt(info
->mount_opt
, RECOVERY
);
584 case Opt_skip_balance
:
585 btrfs_set_opt(info
->mount_opt
, SKIP_BALANCE
);
587 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
588 case Opt_check_integrity_including_extent_data
:
589 printk(KERN_INFO
"btrfs: enabling check integrity"
590 " including extent data\n");
591 btrfs_set_opt(info
->mount_opt
,
592 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA
);
593 btrfs_set_opt(info
->mount_opt
, CHECK_INTEGRITY
);
595 case Opt_check_integrity
:
596 printk(KERN_INFO
"btrfs: enabling check integrity\n");
597 btrfs_set_opt(info
->mount_opt
, CHECK_INTEGRITY
);
599 case Opt_check_integrity_print_mask
:
601 match_int(&args
[0], &intarg
);
603 info
->check_integrity_print_mask
= intarg
;
604 printk(KERN_INFO
"btrfs:"
605 " check_integrity_print_mask 0x%x\n",
606 info
->check_integrity_print_mask
);
610 case Opt_check_integrity_including_extent_data
:
611 case Opt_check_integrity
:
612 case Opt_check_integrity_print_mask
:
613 printk(KERN_ERR
"btrfs: support for check_integrity*"
614 " not compiled in!\n");
618 case Opt_fatal_errors
:
619 if (strcmp(args
[0].from
, "panic") == 0)
620 btrfs_set_opt(info
->mount_opt
,
621 PANIC_ON_FATAL_ERROR
);
622 else if (strcmp(args
[0].from
, "bug") == 0)
623 btrfs_clear_opt(info
->mount_opt
,
624 PANIC_ON_FATAL_ERROR
);
631 printk(KERN_INFO
"btrfs: unrecognized mount option "
640 if (!ret
&& btrfs_test_opt(root
, SPACE_CACHE
))
641 printk(KERN_INFO
"btrfs: disk space caching is enabled\n");
647 * Parse mount options that are required early in the mount process.
649 * All other options will be parsed on much later in the mount process and
650 * only when we need to allocate a new super block.
652 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
653 void *holder
, char **subvol_name
, u64
*subvol_objectid
,
654 u64
*subvol_rootid
, struct btrfs_fs_devices
**fs_devices
)
656 substring_t args
[MAX_OPT_ARGS
];
657 char *device_name
, *opts
, *orig
, *p
;
665 * strsep changes the string, duplicate it because parse_options
668 opts
= kstrdup(options
, GFP_KERNEL
);
673 while ((p
= strsep(&opts
, ",")) != NULL
) {
678 token
= match_token(p
, tokens
, args
);
682 *subvol_name
= match_strdup(&args
[0]);
686 error
= match_int(&args
[0], &intarg
);
688 /* we want the original fs_tree */
691 BTRFS_FS_TREE_OBJECTID
;
693 *subvol_objectid
= intarg
;
696 case Opt_subvolrootid
:
698 error
= match_int(&args
[0], &intarg
);
700 /* we want the original fs_tree */
703 BTRFS_FS_TREE_OBJECTID
;
705 *subvol_rootid
= intarg
;
709 device_name
= match_strdup(&args
[0]);
714 error
= btrfs_scan_one_device(device_name
,
715 flags
, holder
, fs_devices
);
730 static struct dentry
*get_default_root(struct super_block
*sb
,
733 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
734 struct btrfs_root
*root
= fs_info
->tree_root
;
735 struct btrfs_root
*new_root
;
736 struct btrfs_dir_item
*di
;
737 struct btrfs_path
*path
;
738 struct btrfs_key location
;
744 * We have a specific subvol we want to mount, just setup location and
745 * go look up the root.
747 if (subvol_objectid
) {
748 location
.objectid
= subvol_objectid
;
749 location
.type
= BTRFS_ROOT_ITEM_KEY
;
750 location
.offset
= (u64
)-1;
754 path
= btrfs_alloc_path();
756 return ERR_PTR(-ENOMEM
);
757 path
->leave_spinning
= 1;
760 * Find the "default" dir item which points to the root item that we
761 * will mount by default if we haven't been given a specific subvolume
764 dir_id
= btrfs_super_root_dir(fs_info
->super_copy
);
765 di
= btrfs_lookup_dir_item(NULL
, root
, path
, dir_id
, "default", 7, 0);
767 btrfs_free_path(path
);
772 * Ok the default dir item isn't there. This is weird since
773 * it's always been there, but don't freak out, just try and
774 * mount to root most subvolume.
776 btrfs_free_path(path
);
777 dir_id
= BTRFS_FIRST_FREE_OBJECTID
;
778 new_root
= fs_info
->fs_root
;
782 btrfs_dir_item_key_to_cpu(path
->nodes
[0], di
, &location
);
783 btrfs_free_path(path
);
786 new_root
= btrfs_read_fs_root_no_name(fs_info
, &location
);
787 if (IS_ERR(new_root
))
788 return ERR_CAST(new_root
);
790 if (btrfs_root_refs(&new_root
->root_item
) == 0)
791 return ERR_PTR(-ENOENT
);
793 dir_id
= btrfs_root_dirid(&new_root
->root_item
);
795 location
.objectid
= dir_id
;
796 location
.type
= BTRFS_INODE_ITEM_KEY
;
799 inode
= btrfs_iget(sb
, &location
, new_root
, &new);
801 return ERR_CAST(inode
);
804 * If we're just mounting the root most subvol put the inode and return
805 * a reference to the dentry. We will have already gotten a reference
806 * to the inode in btrfs_fill_super so we're good to go.
808 if (!new && sb
->s_root
->d_inode
== inode
) {
810 return dget(sb
->s_root
);
813 return d_obtain_alias(inode
);
816 static int btrfs_fill_super(struct super_block
*sb
,
817 struct btrfs_fs_devices
*fs_devices
,
818 void *data
, int silent
)
821 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
822 struct btrfs_key key
;
825 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
826 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
827 sb
->s_op
= &btrfs_super_ops
;
828 sb
->s_d_op
= &btrfs_dentry_operations
;
829 sb
->s_export_op
= &btrfs_export_ops
;
830 sb
->s_xattr
= btrfs_xattr_handlers
;
832 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
833 sb
->s_flags
|= MS_POSIXACL
;
835 sb
->s_flags
|= MS_I_VERSION
;
836 err
= open_ctree(sb
, fs_devices
, (char *)data
);
838 printk("btrfs: open_ctree failed\n");
842 key
.objectid
= BTRFS_FIRST_FREE_OBJECTID
;
843 key
.type
= BTRFS_INODE_ITEM_KEY
;
845 inode
= btrfs_iget(sb
, &key
, fs_info
->fs_root
, NULL
);
847 err
= PTR_ERR(inode
);
851 sb
->s_root
= d_make_root(inode
);
857 save_mount_options(sb
, data
);
858 cleancache_init_fs(sb
);
859 sb
->s_flags
|= MS_ACTIVE
;
863 close_ctree(fs_info
->tree_root
);
867 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
869 struct btrfs_trans_handle
*trans
;
870 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
871 struct btrfs_root
*root
= fs_info
->tree_root
;
873 trace_btrfs_sync_fs(wait
);
876 filemap_flush(fs_info
->btree_inode
->i_mapping
);
880 btrfs_wait_ordered_extents(root
, 0);
882 trans
= btrfs_attach_transaction_barrier(root
);
884 /* no transaction, don't bother */
885 if (PTR_ERR(trans
) == -ENOENT
)
887 return PTR_ERR(trans
);
889 return btrfs_commit_transaction(trans
, root
);
892 static int btrfs_show_options(struct seq_file
*seq
, struct dentry
*dentry
)
894 struct btrfs_fs_info
*info
= btrfs_sb(dentry
->d_sb
);
895 struct btrfs_root
*root
= info
->tree_root
;
898 if (btrfs_test_opt(root
, DEGRADED
))
899 seq_puts(seq
, ",degraded");
900 if (btrfs_test_opt(root
, NODATASUM
))
901 seq_puts(seq
, ",nodatasum");
902 if (btrfs_test_opt(root
, NODATACOW
))
903 seq_puts(seq
, ",nodatacow");
904 if (btrfs_test_opt(root
, NOBARRIER
))
905 seq_puts(seq
, ",nobarrier");
906 if (info
->max_inline
!= 8192 * 1024)
907 seq_printf(seq
, ",max_inline=%llu",
908 (unsigned long long)info
->max_inline
);
909 if (info
->alloc_start
!= 0)
910 seq_printf(seq
, ",alloc_start=%llu",
911 (unsigned long long)info
->alloc_start
);
912 if (info
->thread_pool_size
!= min_t(unsigned long,
913 num_online_cpus() + 2, 8))
914 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
915 if (btrfs_test_opt(root
, COMPRESS
)) {
916 if (info
->compress_type
== BTRFS_COMPRESS_ZLIB
)
917 compress_type
= "zlib";
919 compress_type
= "lzo";
920 if (btrfs_test_opt(root
, FORCE_COMPRESS
))
921 seq_printf(seq
, ",compress-force=%s", compress_type
);
923 seq_printf(seq
, ",compress=%s", compress_type
);
925 if (btrfs_test_opt(root
, NOSSD
))
926 seq_puts(seq
, ",nossd");
927 if (btrfs_test_opt(root
, SSD_SPREAD
))
928 seq_puts(seq
, ",ssd_spread");
929 else if (btrfs_test_opt(root
, SSD
))
930 seq_puts(seq
, ",ssd");
931 if (btrfs_test_opt(root
, NOTREELOG
))
932 seq_puts(seq
, ",notreelog");
933 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
934 seq_puts(seq
, ",flushoncommit");
935 if (btrfs_test_opt(root
, DISCARD
))
936 seq_puts(seq
, ",discard");
937 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
938 seq_puts(seq
, ",noacl");
939 if (btrfs_test_opt(root
, SPACE_CACHE
))
940 seq_puts(seq
, ",space_cache");
942 seq_puts(seq
, ",nospace_cache");
943 if (btrfs_test_opt(root
, CLEAR_CACHE
))
944 seq_puts(seq
, ",clear_cache");
945 if (btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
946 seq_puts(seq
, ",user_subvol_rm_allowed");
947 if (btrfs_test_opt(root
, ENOSPC_DEBUG
))
948 seq_puts(seq
, ",enospc_debug");
949 if (btrfs_test_opt(root
, AUTO_DEFRAG
))
950 seq_puts(seq
, ",autodefrag");
951 if (btrfs_test_opt(root
, INODE_MAP_CACHE
))
952 seq_puts(seq
, ",inode_cache");
953 if (btrfs_test_opt(root
, SKIP_BALANCE
))
954 seq_puts(seq
, ",skip_balance");
955 if (btrfs_test_opt(root
, PANIC_ON_FATAL_ERROR
))
956 seq_puts(seq
, ",fatal_errors=panic");
960 static int btrfs_test_super(struct super_block
*s
, void *data
)
962 struct btrfs_fs_info
*p
= data
;
963 struct btrfs_fs_info
*fs_info
= btrfs_sb(s
);
965 return fs_info
->fs_devices
== p
->fs_devices
;
968 static int btrfs_set_super(struct super_block
*s
, void *data
)
970 int err
= set_anon_super(s
, data
);
977 * subvolumes are identified by ino 256
979 static inline int is_subvolume_inode(struct inode
*inode
)
981 if (inode
&& inode
->i_ino
== BTRFS_FIRST_FREE_OBJECTID
)
987 * This will strip out the subvol=%s argument for an argument string and add
988 * subvolid=0 to make sure we get the actual tree root for path walking to the
991 static char *setup_root_args(char *args
)
993 unsigned len
= strlen(args
) + 2 + 1;
994 char *src
, *dst
, *buf
;
997 * We need the same args as before, but with this substitution:
998 * s!subvol=[^,]+!subvolid=0!
1000 * Since the replacement string is up to 2 bytes longer than the
1001 * original, allocate strlen(args) + 2 + 1 bytes.
1004 src
= strstr(args
, "subvol=");
1005 /* This shouldn't happen, but just in case.. */
1009 buf
= dst
= kmalloc(len
, GFP_NOFS
);
1014 * If the subvol= arg is not at the start of the string,
1015 * copy whatever precedes it into buf.
1020 dst
+= strlen(args
);
1023 strcpy(dst
, "subvolid=0");
1024 dst
+= strlen("subvolid=0");
1027 * If there is a "," after the original subvol=... string,
1028 * copy that suffix into our buffer. Otherwise, we're done.
1030 src
= strchr(src
, ',');
1037 static struct dentry
*mount_subvol(const char *subvol_name
, int flags
,
1038 const char *device_name
, char *data
)
1040 struct dentry
*root
;
1041 struct vfsmount
*mnt
;
1044 newargs
= setup_root_args(data
);
1046 return ERR_PTR(-ENOMEM
);
1047 mnt
= vfs_kern_mount(&btrfs_fs_type
, flags
, device_name
,
1051 return ERR_CAST(mnt
);
1053 root
= mount_subtree(mnt
, subvol_name
);
1055 if (!IS_ERR(root
) && !is_subvolume_inode(root
->d_inode
)) {
1056 struct super_block
*s
= root
->d_sb
;
1058 root
= ERR_PTR(-EINVAL
);
1059 deactivate_locked_super(s
);
1060 printk(KERN_ERR
"btrfs: '%s' is not a valid subvolume\n",
1068 * Find a superblock for the given device / mount point.
1070 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1071 * for multiple device setup. Make sure to keep it in sync.
1073 static struct dentry
*btrfs_mount(struct file_system_type
*fs_type
, int flags
,
1074 const char *device_name
, void *data
)
1076 struct block_device
*bdev
= NULL
;
1077 struct super_block
*s
;
1078 struct dentry
*root
;
1079 struct btrfs_fs_devices
*fs_devices
= NULL
;
1080 struct btrfs_fs_info
*fs_info
= NULL
;
1081 fmode_t mode
= FMODE_READ
;
1082 char *subvol_name
= NULL
;
1083 u64 subvol_objectid
= 0;
1084 u64 subvol_rootid
= 0;
1087 if (!(flags
& MS_RDONLY
))
1088 mode
|= FMODE_WRITE
;
1090 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
1091 &subvol_name
, &subvol_objectid
,
1092 &subvol_rootid
, &fs_devices
);
1095 return ERR_PTR(error
);
1099 root
= mount_subvol(subvol_name
, flags
, device_name
, data
);
1104 error
= btrfs_scan_one_device(device_name
, mode
, fs_type
, &fs_devices
);
1106 return ERR_PTR(error
);
1109 * Setup a dummy root and fs_info for test/set super. This is because
1110 * we don't actually fill this stuff out until open_ctree, but we need
1111 * it for searching for existing supers, so this lets us do that and
1112 * then open_ctree will properly initialize everything later.
1114 fs_info
= kzalloc(sizeof(struct btrfs_fs_info
), GFP_NOFS
);
1116 return ERR_PTR(-ENOMEM
);
1118 fs_info
->fs_devices
= fs_devices
;
1120 fs_info
->super_copy
= kzalloc(BTRFS_SUPER_INFO_SIZE
, GFP_NOFS
);
1121 fs_info
->super_for_commit
= kzalloc(BTRFS_SUPER_INFO_SIZE
, GFP_NOFS
);
1122 if (!fs_info
->super_copy
|| !fs_info
->super_for_commit
) {
1127 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
1131 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
1133 goto error_close_devices
;
1136 bdev
= fs_devices
->latest_bdev
;
1137 s
= sget(fs_type
, btrfs_test_super
, btrfs_set_super
, flags
| MS_NOSEC
,
1141 goto error_close_devices
;
1145 btrfs_close_devices(fs_devices
);
1146 free_fs_info(fs_info
);
1147 if ((flags
^ s
->s_flags
) & MS_RDONLY
)
1150 char b
[BDEVNAME_SIZE
];
1152 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
1153 btrfs_sb(s
)->bdev_holder
= fs_type
;
1154 error
= btrfs_fill_super(s
, fs_devices
, data
,
1155 flags
& MS_SILENT
? 1 : 0);
1158 root
= !error
? get_default_root(s
, subvol_objectid
) : ERR_PTR(error
);
1160 deactivate_locked_super(s
);
1164 error_close_devices
:
1165 btrfs_close_devices(fs_devices
);
1167 free_fs_info(fs_info
);
1168 return ERR_PTR(error
);
1171 static void btrfs_set_max_workers(struct btrfs_workers
*workers
, int new_limit
)
1173 spin_lock_irq(&workers
->lock
);
1174 workers
->max_workers
= new_limit
;
1175 spin_unlock_irq(&workers
->lock
);
1178 static void btrfs_resize_thread_pool(struct btrfs_fs_info
*fs_info
,
1179 int new_pool_size
, int old_pool_size
)
1181 if (new_pool_size
== old_pool_size
)
1184 fs_info
->thread_pool_size
= new_pool_size
;
1186 printk(KERN_INFO
"btrfs: resize thread pool %d -> %d\n",
1187 old_pool_size
, new_pool_size
);
1189 btrfs_set_max_workers(&fs_info
->generic_worker
, new_pool_size
);
1190 btrfs_set_max_workers(&fs_info
->workers
, new_pool_size
);
1191 btrfs_set_max_workers(&fs_info
->delalloc_workers
, new_pool_size
);
1192 btrfs_set_max_workers(&fs_info
->submit_workers
, new_pool_size
);
1193 btrfs_set_max_workers(&fs_info
->caching_workers
, new_pool_size
);
1194 btrfs_set_max_workers(&fs_info
->fixup_workers
, new_pool_size
);
1195 btrfs_set_max_workers(&fs_info
->endio_workers
, new_pool_size
);
1196 btrfs_set_max_workers(&fs_info
->endio_meta_workers
, new_pool_size
);
1197 btrfs_set_max_workers(&fs_info
->endio_meta_write_workers
, new_pool_size
);
1198 btrfs_set_max_workers(&fs_info
->endio_write_workers
, new_pool_size
);
1199 btrfs_set_max_workers(&fs_info
->endio_freespace_worker
, new_pool_size
);
1200 btrfs_set_max_workers(&fs_info
->delayed_workers
, new_pool_size
);
1201 btrfs_set_max_workers(&fs_info
->readahead_workers
, new_pool_size
);
1202 btrfs_set_max_workers(&fs_info
->scrub_wr_completion_workers
,
1206 static inline void btrfs_remount_prepare(struct btrfs_fs_info
*fs_info
,
1207 unsigned long old_opts
, int flags
)
1209 set_bit(BTRFS_FS_STATE_REMOUNTING
, &fs_info
->fs_state
);
1211 if (btrfs_raw_test_opt(old_opts
, AUTO_DEFRAG
) &&
1212 (!btrfs_raw_test_opt(fs_info
->mount_opt
, AUTO_DEFRAG
) ||
1213 (flags
& MS_RDONLY
))) {
1214 /* wait for any defraggers to finish */
1215 wait_event(fs_info
->transaction_wait
,
1216 (atomic_read(&fs_info
->defrag_running
) == 0));
1217 if (flags
& MS_RDONLY
)
1218 sync_filesystem(fs_info
->sb
);
1222 static inline void btrfs_remount_cleanup(struct btrfs_fs_info
*fs_info
,
1223 unsigned long old_opts
)
1226 * We need cleanup all defragable inodes if the autodefragment is
1227 * close or the fs is R/O.
1229 if (btrfs_raw_test_opt(old_opts
, AUTO_DEFRAG
) &&
1230 (!btrfs_raw_test_opt(fs_info
->mount_opt
, AUTO_DEFRAG
) ||
1231 (fs_info
->sb
->s_flags
& MS_RDONLY
))) {
1232 btrfs_cleanup_defrag_inodes(fs_info
);
1235 clear_bit(BTRFS_FS_STATE_REMOUNTING
, &fs_info
->fs_state
);
1238 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
1240 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1241 struct btrfs_root
*root
= fs_info
->tree_root
;
1242 unsigned old_flags
= sb
->s_flags
;
1243 unsigned long old_opts
= fs_info
->mount_opt
;
1244 unsigned long old_compress_type
= fs_info
->compress_type
;
1245 u64 old_max_inline
= fs_info
->max_inline
;
1246 u64 old_alloc_start
= fs_info
->alloc_start
;
1247 int old_thread_pool_size
= fs_info
->thread_pool_size
;
1248 unsigned int old_metadata_ratio
= fs_info
->metadata_ratio
;
1251 btrfs_remount_prepare(fs_info
, old_opts
, *flags
);
1253 ret
= btrfs_parse_options(root
, data
);
1259 btrfs_resize_thread_pool(fs_info
,
1260 fs_info
->thread_pool_size
, old_thread_pool_size
);
1262 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
1265 if (*flags
& MS_RDONLY
) {
1267 * this also happens on 'umount -rf' or on shutdown, when
1268 * the filesystem is busy.
1270 sb
->s_flags
|= MS_RDONLY
;
1272 btrfs_dev_replace_suspend_for_unmount(fs_info
);
1273 btrfs_scrub_cancel(fs_info
);
1275 ret
= btrfs_commit_super(root
);
1279 if (fs_info
->fs_devices
->rw_devices
== 0) {
1284 if (fs_info
->fs_devices
->missing_devices
>
1285 fs_info
->num_tolerated_disk_barrier_failures
&&
1286 !(*flags
& MS_RDONLY
)) {
1288 "Btrfs: too many missing devices, writeable remount is not allowed\n");
1293 if (btrfs_super_log_root(fs_info
->super_copy
) != 0) {
1298 ret
= btrfs_cleanup_fs_roots(fs_info
);
1302 /* recover relocation */
1303 ret
= btrfs_recover_relocation(root
);
1307 ret
= btrfs_resume_balance_async(fs_info
);
1311 ret
= btrfs_resume_dev_replace_async(fs_info
);
1313 pr_warn("btrfs: failed to resume dev_replace\n");
1316 sb
->s_flags
&= ~MS_RDONLY
;
1319 btrfs_remount_cleanup(fs_info
, old_opts
);
1323 /* We've hit an error - don't reset MS_RDONLY */
1324 if (sb
->s_flags
& MS_RDONLY
)
1325 old_flags
|= MS_RDONLY
;
1326 sb
->s_flags
= old_flags
;
1327 fs_info
->mount_opt
= old_opts
;
1328 fs_info
->compress_type
= old_compress_type
;
1329 fs_info
->max_inline
= old_max_inline
;
1330 mutex_lock(&fs_info
->chunk_mutex
);
1331 fs_info
->alloc_start
= old_alloc_start
;
1332 mutex_unlock(&fs_info
->chunk_mutex
);
1333 btrfs_resize_thread_pool(fs_info
,
1334 old_thread_pool_size
, fs_info
->thread_pool_size
);
1335 fs_info
->metadata_ratio
= old_metadata_ratio
;
1336 btrfs_remount_cleanup(fs_info
, old_opts
);
1340 /* Used to sort the devices by max_avail(descending sort) */
1341 static int btrfs_cmp_device_free_bytes(const void *dev_info1
,
1342 const void *dev_info2
)
1344 if (((struct btrfs_device_info
*)dev_info1
)->max_avail
>
1345 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
1347 else if (((struct btrfs_device_info
*)dev_info1
)->max_avail
<
1348 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
1355 * sort the devices by max_avail, in which max free extent size of each device
1356 * is stored.(Descending Sort)
1358 static inline void btrfs_descending_sort_devices(
1359 struct btrfs_device_info
*devices
,
1362 sort(devices
, nr_devices
, sizeof(struct btrfs_device_info
),
1363 btrfs_cmp_device_free_bytes
, NULL
);
1367 * The helper to calc the free space on the devices that can be used to store
1370 static int btrfs_calc_avail_data_space(struct btrfs_root
*root
, u64
*free_bytes
)
1372 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
1373 struct btrfs_device_info
*devices_info
;
1374 struct btrfs_fs_devices
*fs_devices
= fs_info
->fs_devices
;
1375 struct btrfs_device
*device
;
1380 u64 min_stripe_size
;
1381 int min_stripes
= 1, num_stripes
= 1;
1382 int i
= 0, nr_devices
;
1385 nr_devices
= fs_info
->fs_devices
->open_devices
;
1386 BUG_ON(!nr_devices
);
1388 devices_info
= kmalloc(sizeof(*devices_info
) * nr_devices
,
1393 /* calc min stripe number for data space alloction */
1394 type
= btrfs_get_alloc_profile(root
, 1);
1395 if (type
& BTRFS_BLOCK_GROUP_RAID0
) {
1397 num_stripes
= nr_devices
;
1398 } else if (type
& BTRFS_BLOCK_GROUP_RAID1
) {
1401 } else if (type
& BTRFS_BLOCK_GROUP_RAID10
) {
1406 if (type
& BTRFS_BLOCK_GROUP_DUP
)
1407 min_stripe_size
= 2 * BTRFS_STRIPE_LEN
;
1409 min_stripe_size
= BTRFS_STRIPE_LEN
;
1411 list_for_each_entry(device
, &fs_devices
->devices
, dev_list
) {
1412 if (!device
->in_fs_metadata
|| !device
->bdev
||
1413 device
->is_tgtdev_for_dev_replace
)
1416 avail_space
= device
->total_bytes
- device
->bytes_used
;
1418 /* align with stripe_len */
1419 do_div(avail_space
, BTRFS_STRIPE_LEN
);
1420 avail_space
*= BTRFS_STRIPE_LEN
;
1423 * In order to avoid overwritting the superblock on the drive,
1424 * btrfs starts at an offset of at least 1MB when doing chunk
1427 skip_space
= 1024 * 1024;
1429 /* user can set the offset in fs_info->alloc_start. */
1430 if (fs_info
->alloc_start
+ BTRFS_STRIPE_LEN
<=
1431 device
->total_bytes
)
1432 skip_space
= max(fs_info
->alloc_start
, skip_space
);
1435 * btrfs can not use the free space in [0, skip_space - 1],
1436 * we must subtract it from the total. In order to implement
1437 * it, we account the used space in this range first.
1439 ret
= btrfs_account_dev_extents_size(device
, 0, skip_space
- 1,
1442 kfree(devices_info
);
1446 /* calc the free space in [0, skip_space - 1] */
1447 skip_space
-= used_space
;
1450 * we can use the free space in [0, skip_space - 1], subtract
1451 * it from the total.
1453 if (avail_space
&& avail_space
>= skip_space
)
1454 avail_space
-= skip_space
;
1458 if (avail_space
< min_stripe_size
)
1461 devices_info
[i
].dev
= device
;
1462 devices_info
[i
].max_avail
= avail_space
;
1469 btrfs_descending_sort_devices(devices_info
, nr_devices
);
1473 while (nr_devices
>= min_stripes
) {
1474 if (num_stripes
> nr_devices
)
1475 num_stripes
= nr_devices
;
1477 if (devices_info
[i
].max_avail
>= min_stripe_size
) {
1481 avail_space
+= devices_info
[i
].max_avail
* num_stripes
;
1482 alloc_size
= devices_info
[i
].max_avail
;
1483 for (j
= i
+ 1 - num_stripes
; j
<= i
; j
++)
1484 devices_info
[j
].max_avail
-= alloc_size
;
1490 kfree(devices_info
);
1491 *free_bytes
= avail_space
;
1495 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
1497 struct btrfs_fs_info
*fs_info
= btrfs_sb(dentry
->d_sb
);
1498 struct btrfs_super_block
*disk_super
= fs_info
->super_copy
;
1499 struct list_head
*head
= &fs_info
->space_info
;
1500 struct btrfs_space_info
*found
;
1502 u64 total_free_data
= 0;
1503 int bits
= dentry
->d_sb
->s_blocksize_bits
;
1504 __be32
*fsid
= (__be32
*)fs_info
->fsid
;
1507 /* holding chunk_muext to avoid allocating new chunks */
1508 mutex_lock(&fs_info
->chunk_mutex
);
1510 list_for_each_entry_rcu(found
, head
, list
) {
1511 if (found
->flags
& BTRFS_BLOCK_GROUP_DATA
) {
1512 total_free_data
+= found
->disk_total
- found
->disk_used
;
1514 btrfs_account_ro_block_groups_free_space(found
);
1517 total_used
+= found
->disk_used
;
1521 buf
->f_namelen
= BTRFS_NAME_LEN
;
1522 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
1523 buf
->f_bfree
= buf
->f_blocks
- (total_used
>> bits
);
1524 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
1525 buf
->f_type
= BTRFS_SUPER_MAGIC
;
1526 buf
->f_bavail
= total_free_data
;
1527 ret
= btrfs_calc_avail_data_space(fs_info
->tree_root
, &total_free_data
);
1529 mutex_unlock(&fs_info
->chunk_mutex
);
1532 buf
->f_bavail
+= total_free_data
;
1533 buf
->f_bavail
= buf
->f_bavail
>> bits
;
1534 mutex_unlock(&fs_info
->chunk_mutex
);
1536 /* We treat it as constant endianness (it doesn't matter _which_)
1537 because we want the fsid to come out the same whether mounted
1538 on a big-endian or little-endian host */
1539 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
1540 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
1541 /* Mask in the root object ID too, to disambiguate subvols */
1542 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
1543 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
1548 static void btrfs_kill_super(struct super_block
*sb
)
1550 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1551 kill_anon_super(sb
);
1552 free_fs_info(fs_info
);
1555 static struct file_system_type btrfs_fs_type
= {
1556 .owner
= THIS_MODULE
,
1558 .mount
= btrfs_mount
,
1559 .kill_sb
= btrfs_kill_super
,
1560 .fs_flags
= FS_REQUIRES_DEV
,
1562 MODULE_ALIAS_FS("btrfs");
1565 * used by btrfsctl to scan devices when no FS is mounted
1567 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
1570 struct btrfs_ioctl_vol_args
*vol
;
1571 struct btrfs_fs_devices
*fs_devices
;
1574 if (!capable(CAP_SYS_ADMIN
))
1577 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
1579 return PTR_ERR(vol
);
1582 case BTRFS_IOC_SCAN_DEV
:
1583 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
1584 &btrfs_fs_type
, &fs_devices
);
1586 case BTRFS_IOC_DEVICES_READY
:
1587 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
1588 &btrfs_fs_type
, &fs_devices
);
1591 ret
= !(fs_devices
->num_devices
== fs_devices
->total_devices
);
1599 static int btrfs_freeze(struct super_block
*sb
)
1601 struct btrfs_trans_handle
*trans
;
1602 struct btrfs_root
*root
= btrfs_sb(sb
)->tree_root
;
1604 trans
= btrfs_attach_transaction_barrier(root
);
1605 if (IS_ERR(trans
)) {
1606 /* no transaction, don't bother */
1607 if (PTR_ERR(trans
) == -ENOENT
)
1609 return PTR_ERR(trans
);
1611 return btrfs_commit_transaction(trans
, root
);
1614 static int btrfs_unfreeze(struct super_block
*sb
)
1619 static int btrfs_show_devname(struct seq_file
*m
, struct dentry
*root
)
1621 struct btrfs_fs_info
*fs_info
= btrfs_sb(root
->d_sb
);
1622 struct btrfs_fs_devices
*cur_devices
;
1623 struct btrfs_device
*dev
, *first_dev
= NULL
;
1624 struct list_head
*head
;
1625 struct rcu_string
*name
;
1627 mutex_lock(&fs_info
->fs_devices
->device_list_mutex
);
1628 cur_devices
= fs_info
->fs_devices
;
1629 while (cur_devices
) {
1630 head
= &cur_devices
->devices
;
1631 list_for_each_entry(dev
, head
, dev_list
) {
1634 if (!first_dev
|| dev
->devid
< first_dev
->devid
)
1637 cur_devices
= cur_devices
->seed
;
1642 name
= rcu_dereference(first_dev
->name
);
1643 seq_escape(m
, name
->str
, " \t\n\\");
1648 mutex_unlock(&fs_info
->fs_devices
->device_list_mutex
);
1652 static const struct super_operations btrfs_super_ops
= {
1653 .drop_inode
= btrfs_drop_inode
,
1654 .evict_inode
= btrfs_evict_inode
,
1655 .put_super
= btrfs_put_super
,
1656 .sync_fs
= btrfs_sync_fs
,
1657 .show_options
= btrfs_show_options
,
1658 .show_devname
= btrfs_show_devname
,
1659 .write_inode
= btrfs_write_inode
,
1660 .alloc_inode
= btrfs_alloc_inode
,
1661 .destroy_inode
= btrfs_destroy_inode
,
1662 .statfs
= btrfs_statfs
,
1663 .remount_fs
= btrfs_remount
,
1664 .freeze_fs
= btrfs_freeze
,
1665 .unfreeze_fs
= btrfs_unfreeze
,
1668 static const struct file_operations btrfs_ctl_fops
= {
1669 .unlocked_ioctl
= btrfs_control_ioctl
,
1670 .compat_ioctl
= btrfs_control_ioctl
,
1671 .owner
= THIS_MODULE
,
1672 .llseek
= noop_llseek
,
1675 static struct miscdevice btrfs_misc
= {
1676 .minor
= BTRFS_MINOR
,
1677 .name
= "btrfs-control",
1678 .fops
= &btrfs_ctl_fops
1681 MODULE_ALIAS_MISCDEV(BTRFS_MINOR
);
1682 MODULE_ALIAS("devname:btrfs-control");
1684 static int btrfs_interface_init(void)
1686 return misc_register(&btrfs_misc
);
1689 static void btrfs_interface_exit(void)
1691 if (misc_deregister(&btrfs_misc
) < 0)
1692 printk(KERN_INFO
"btrfs: misc_deregister failed for control device\n");
1695 static int __init
init_btrfs_fs(void)
1699 err
= btrfs_init_sysfs();
1703 btrfs_init_compress();
1705 err
= btrfs_init_cachep();
1709 err
= extent_io_init();
1713 err
= extent_map_init();
1715 goto free_extent_io
;
1717 err
= ordered_data_init();
1719 goto free_extent_map
;
1721 err
= btrfs_delayed_inode_init();
1723 goto free_ordered_data
;
1725 err
= btrfs_auto_defrag_init();
1727 goto free_delayed_inode
;
1729 err
= btrfs_delayed_ref_init();
1731 goto free_auto_defrag
;
1733 err
= btrfs_interface_init();
1735 goto free_delayed_ref
;
1737 err
= register_filesystem(&btrfs_fs_type
);
1739 goto unregister_ioctl
;
1741 btrfs_init_lockdep();
1743 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
1744 btrfs_test_free_space_cache();
1747 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
1751 btrfs_interface_exit();
1753 btrfs_delayed_ref_exit();
1755 btrfs_auto_defrag_exit();
1757 btrfs_delayed_inode_exit();
1759 ordered_data_exit();
1765 btrfs_destroy_cachep();
1767 btrfs_exit_compress();
1772 static void __exit
exit_btrfs_fs(void)
1774 btrfs_destroy_cachep();
1775 btrfs_delayed_ref_exit();
1776 btrfs_auto_defrag_exit();
1777 btrfs_delayed_inode_exit();
1778 ordered_data_exit();
1781 btrfs_interface_exit();
1782 unregister_filesystem(&btrfs_fs_type
);
1784 btrfs_cleanup_fs_uuids();
1785 btrfs_exit_compress();
1788 module_init(init_btrfs_fs
)
1789 module_exit(exit_btrfs_fs
)
1791 MODULE_LICENSE("GPL");