1 // SPDX-License-Identifier: GPL-2.0
3 * Copyright (C) 2007 Oracle. All rights reserved.
6 #include <linux/blkdev.h>
7 #include <linux/module.h>
9 #include <linux/pagemap.h>
10 #include <linux/highmem.h>
11 #include <linux/time.h>
12 #include <linux/init.h>
13 #include <linux/seq_file.h>
14 #include <linux/string.h>
15 #include <linux/backing-dev.h>
16 #include <linux/mount.h>
17 #include <linux/writeback.h>
18 #include <linux/statfs.h>
19 #include <linux/compat.h>
20 #include <linux/parser.h>
21 #include <linux/ctype.h>
22 #include <linux/namei.h>
23 #include <linux/miscdevice.h>
24 #include <linux/magic.h>
25 #include <linux/slab.h>
26 #include <linux/cleancache.h>
27 #include <linux/ratelimit.h>
28 #include <linux/crc32c.h>
29 #include <linux/btrfs.h>
30 #include "delayed-inode.h"
33 #include "transaction.h"
34 #include "btrfs_inode.h"
35 #include "print-tree.h"
40 #include "compression.h"
41 #include "rcu-string.h"
42 #include "dev-replace.h"
43 #include "free-space-cache.h"
45 #include "space-info.h"
48 #include "tests/btrfs-tests.h"
49 #include "block-group.h"
53 #define CREATE_TRACE_POINTS
54 #include <trace/events/btrfs.h>
56 static const struct super_operations btrfs_super_ops;
59 * Types for mounting the default subvolume and a subvolume explicitly
60 * requested by subvol=/path. That way the callchain is straightforward and we
61 * don't have to play tricks with the mount options and recursive calls to
64 * The new btrfs_root_fs_type also servers as a tag for the bdev_holder.
66 static struct file_system_type btrfs_fs_type;
67 static struct file_system_type btrfs_root_fs_type;
69 static int btrfs_remount(struct super_block *sb, int *flags, char *data);
72 * Generally the error codes correspond to their respective errors, but there
73 * are a few special cases.
75 * EUCLEAN: Any sort of corruption that we encounter. The tree-checker for
76 * instance will return EUCLEAN if any of the blocks are corrupted in
77 * a way that is problematic. We want to reserve EUCLEAN for these
78 * sort of corruptions.
80 * EROFS: If we check BTRFS_FS_STATE_ERROR and fail out with a return error, we
81 * need to use EROFS for this case. We will have no idea of the
82 * original failure, that will have been reported at the time we tripped
83 * over the error. Each subsequent error that doesn't have any context
84 * of the original error should use EROFS when handling BTRFS_FS_STATE_ERROR.
86 const char * __attribute_const__ btrfs_decode_error(int errno)
88 char *errstr = "unknown";
91 case -ENOENT: /* -2 */
92 errstr = "No such entry";
95 errstr = "IO failure";
97 case -ENOMEM: /* -12*/
98 errstr = "Out of memory";
100 case -EEXIST: /* -17 */
101 errstr = "Object already exists";
103 case -ENOSPC: /* -28 */
104 errstr = "No space left";
106 case -EROFS: /* -30 */
107 errstr = "Readonly filesystem";
109 case -EOPNOTSUPP: /* -95 */
110 errstr = "Operation not supported";
112 case -EUCLEAN: /* -117 */
113 errstr = "Filesystem corrupted";
115 case -EDQUOT: /* -122 */
116 errstr = "Quota exceeded";
124 * __btrfs_handle_fs_error decodes expected errors from the caller and
125 * invokes the appropriate error response.
128 void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function,
129 unsigned int line, int errno, const char *fmt, ...)
131 struct super_block *sb = fs_info->sb;
137 * Special case: if the error is EROFS, and we're already
138 * under SB_RDONLY, then it is safe here.
140 if (errno == -EROFS && sb_rdonly(sb))
144 errstr = btrfs_decode_error(errno);
146 struct va_format vaf;
153 pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n",
154 sb->s_id, function, line, errno, errstr, &vaf);
157 pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s\n",
158 sb->s_id, function, line, errno, errstr);
163 * Today we only save the error info to memory. Long term we'll
164 * also send it down to the disk
166 set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state);
168 /* Don't go through full error handling during mount */
169 if (!(sb->s_flags & SB_BORN))
175 btrfs_discard_stop(fs_info);
177 /* btrfs handle error by forcing the filesystem readonly */
178 sb->s_flags |= SB_RDONLY;
179 btrfs_info(fs_info, "forced readonly");
181 * Note that a running device replace operation is not canceled here
182 * although there is no way to update the progress. It would add the
183 * risk of a deadlock, therefore the canceling is omitted. The only
184 * penalty is that some I/O remains active until the procedure
185 * completes. The next time when the filesystem is mounted writable
186 * again, the device replace operation continues.
191 static const char * const logtypes[] = {
204 * Use one ratelimit state per log level so that a flood of less important
205 * messages doesn't cause more important ones to be dropped.
207 static struct ratelimit_state printk_limits[] = {
208 RATELIMIT_STATE_INIT(printk_limits[0], DEFAULT_RATELIMIT_INTERVAL, 100),
209 RATELIMIT_STATE_INIT(printk_limits[1], DEFAULT_RATELIMIT_INTERVAL, 100),
210 RATELIMIT_STATE_INIT(printk_limits[2], DEFAULT_RATELIMIT_INTERVAL, 100),
211 RATELIMIT_STATE_INIT(printk_limits[3], DEFAULT_RATELIMIT_INTERVAL, 100),
212 RATELIMIT_STATE_INIT(printk_limits[4], DEFAULT_RATELIMIT_INTERVAL, 100),
213 RATELIMIT_STATE_INIT(printk_limits[5], DEFAULT_RATELIMIT_INTERVAL, 100),
214 RATELIMIT_STATE_INIT(printk_limits[6], DEFAULT_RATELIMIT_INTERVAL, 100),
215 RATELIMIT_STATE_INIT(printk_limits[7], DEFAULT_RATELIMIT_INTERVAL, 100),
218 void __cold btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
220 char lvl[PRINTK_MAX_SINGLE_HEADER_LEN + 1] = "\0";
221 struct va_format vaf;
224 const char *type = logtypes[4];
225 struct ratelimit_state *ratelimit = &printk_limits[4];
229 while ((kern_level = printk_get_level(fmt)) != 0) {
230 size_t size = printk_skip_level(fmt) - fmt;
232 if (kern_level >= '0' && kern_level <= '7') {
233 memcpy(lvl, fmt, size);
235 type = logtypes[kern_level - '0'];
236 ratelimit = &printk_limits[kern_level - '0'];
244 if (__ratelimit(ratelimit)) {
246 printk("%sBTRFS %s (device %s): %pV\n", lvl, type,
247 fs_info->sb->s_id, &vaf);
249 printk("%sBTRFS %s: %pV\n", lvl, type, &vaf);
257 * We only mark the transaction aborted and then set the file system read-only.
258 * This will prevent new transactions from starting or trying to join this
261 * This means that error recovery at the call site is limited to freeing
262 * any local memory allocations and passing the error code up without
263 * further cleanup. The transaction should complete as it normally would
264 * in the call path but will return -EIO.
266 * We'll complete the cleanup in btrfs_end_transaction and
267 * btrfs_commit_transaction.
270 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
271 const char *function,
272 unsigned int line, int errno)
274 struct btrfs_fs_info *fs_info = trans->fs_info;
276 WRITE_ONCE(trans->aborted, errno);
277 /* Nothing used. The other threads that have joined this
278 * transaction may be able to continue. */
279 if (!trans->dirty && list_empty(&trans->new_bgs)) {
282 errstr = btrfs_decode_error(errno);
284 "%s:%d: Aborting unused transaction(%s).",
285 function, line, errstr);
288 WRITE_ONCE(trans->transaction->aborted, errno);
289 /* Wake up anybody who may be waiting on this transaction */
290 wake_up(&fs_info->transaction_wait);
291 wake_up(&fs_info->transaction_blocked_wait);
292 __btrfs_handle_fs_error(fs_info, function, line, errno, NULL);
295 * __btrfs_panic decodes unexpected, fatal errors from the caller,
296 * issues an alert, and either panics or BUGs, depending on mount options.
299 void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
300 unsigned int line, int errno, const char *fmt, ...)
302 char *s_id = "<unknown>";
304 struct va_format vaf = { .fmt = fmt };
308 s_id = fs_info->sb->s_id;
313 errstr = btrfs_decode_error(errno);
314 if (fs_info && (btrfs_test_opt(fs_info, PANIC_ON_FATAL_ERROR)))
315 panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
316 s_id, function, line, &vaf, errno, errstr);
318 btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)",
319 function, line, &vaf, errno, errstr);
321 /* Caller calls BUG() */
324 static void btrfs_put_super(struct super_block *sb)
326 close_ctree(btrfs_sb(sb));
335 Opt_compress_force_type,
340 Opt_flushoncommit, Opt_noflushoncommit,
342 Opt_barrier, Opt_nobarrier,
343 Opt_datacow, Opt_nodatacow,
344 Opt_datasum, Opt_nodatasum,
345 Opt_defrag, Opt_nodefrag,
346 Opt_discard, Opt_nodiscard,
350 Opt_rescan_uuid_tree,
352 Opt_space_cache, Opt_no_space_cache,
353 Opt_space_cache_version,
355 Opt_ssd_spread, Opt_nossd_spread,
360 Opt_treelog, Opt_notreelog,
361 Opt_user_subvol_rm_allowed,
371 /* Deprecated options */
373 Opt_inode_cache, Opt_noinode_cache,
375 /* Debugging options */
377 Opt_check_integrity_including_extent_data,
378 Opt_check_integrity_print_mask,
379 Opt_enospc_debug, Opt_noenospc_debug,
380 #ifdef CONFIG_BTRFS_DEBUG
381 Opt_fragment_data, Opt_fragment_metadata, Opt_fragment_all,
383 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
389 static const match_table_t tokens = {
391 {Opt_noacl, "noacl"},
392 {Opt_clear_cache, "clear_cache"},
393 {Opt_commit_interval, "commit=%u"},
394 {Opt_compress, "compress"},
395 {Opt_compress_type, "compress=%s"},
396 {Opt_compress_force, "compress-force"},
397 {Opt_compress_force_type, "compress-force=%s"},
398 {Opt_degraded, "degraded"},
399 {Opt_device, "device=%s"},
400 {Opt_fatal_errors, "fatal_errors=%s"},
401 {Opt_flushoncommit, "flushoncommit"},
402 {Opt_noflushoncommit, "noflushoncommit"},
403 {Opt_inode_cache, "inode_cache"},
404 {Opt_noinode_cache, "noinode_cache"},
405 {Opt_max_inline, "max_inline=%s"},
406 {Opt_barrier, "barrier"},
407 {Opt_nobarrier, "nobarrier"},
408 {Opt_datacow, "datacow"},
409 {Opt_nodatacow, "nodatacow"},
410 {Opt_datasum, "datasum"},
411 {Opt_nodatasum, "nodatasum"},
412 {Opt_defrag, "autodefrag"},
413 {Opt_nodefrag, "noautodefrag"},
414 {Opt_discard, "discard"},
415 {Opt_discard_mode, "discard=%s"},
416 {Opt_nodiscard, "nodiscard"},
417 {Opt_norecovery, "norecovery"},
418 {Opt_ratio, "metadata_ratio=%u"},
419 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
420 {Opt_skip_balance, "skip_balance"},
421 {Opt_space_cache, "space_cache"},
422 {Opt_no_space_cache, "nospace_cache"},
423 {Opt_space_cache_version, "space_cache=%s"},
425 {Opt_nossd, "nossd"},
426 {Opt_ssd_spread, "ssd_spread"},
427 {Opt_nossd_spread, "nossd_spread"},
428 {Opt_subvol, "subvol=%s"},
429 {Opt_subvol_empty, "subvol="},
430 {Opt_subvolid, "subvolid=%s"},
431 {Opt_thread_pool, "thread_pool=%u"},
432 {Opt_treelog, "treelog"},
433 {Opt_notreelog, "notreelog"},
434 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
437 {Opt_rescue, "rescue=%s"},
438 /* Deprecated, with alias rescue=nologreplay */
439 {Opt_nologreplay, "nologreplay"},
440 /* Deprecated, with alias rescue=usebackuproot */
441 {Opt_usebackuproot, "usebackuproot"},
443 /* Deprecated options */
444 {Opt_recovery, "recovery"},
446 /* Debugging options */
447 {Opt_check_integrity, "check_int"},
448 {Opt_check_integrity_including_extent_data, "check_int_data"},
449 {Opt_check_integrity_print_mask, "check_int_print_mask=%u"},
450 {Opt_enospc_debug, "enospc_debug"},
451 {Opt_noenospc_debug, "noenospc_debug"},
452 #ifdef CONFIG_BTRFS_DEBUG
453 {Opt_fragment_data, "fragment=data"},
454 {Opt_fragment_metadata, "fragment=metadata"},
455 {Opt_fragment_all, "fragment=all"},
457 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
458 {Opt_ref_verify, "ref_verify"},
463 static const match_table_t rescue_tokens = {
464 {Opt_usebackuproot, "usebackuproot"},
465 {Opt_nologreplay, "nologreplay"},
466 {Opt_ignorebadroots, "ignorebadroots"},
467 {Opt_ignorebadroots, "ibadroots"},
468 {Opt_ignoredatacsums, "ignoredatacsums"},
469 {Opt_ignoredatacsums, "idatacsums"},
470 {Opt_rescue_all, "all"},
474 static bool check_ro_option(struct btrfs_fs_info *fs_info, unsigned long opt,
475 const char *opt_name)
477 if (fs_info->mount_opt & opt) {
478 btrfs_err(fs_info, "%s must be used with ro mount option",
485 static int parse_rescue_options(struct btrfs_fs_info *info, const char *options)
490 substring_t args[MAX_OPT_ARGS];
493 opts = kstrdup(options, GFP_KERNEL);
498 while ((p = strsep(&opts, ":")) != NULL) {
503 token = match_token(p, rescue_tokens, args);
505 case Opt_usebackuproot:
507 "trying to use backup root at mount time");
508 btrfs_set_opt(info->mount_opt, USEBACKUPROOT);
510 case Opt_nologreplay:
511 btrfs_set_and_info(info, NOLOGREPLAY,
512 "disabling log replay at mount time");
514 case Opt_ignorebadroots:
515 btrfs_set_and_info(info, IGNOREBADROOTS,
516 "ignoring bad roots");
518 case Opt_ignoredatacsums:
519 btrfs_set_and_info(info, IGNOREDATACSUMS,
520 "ignoring data csums");
523 btrfs_info(info, "enabling all of the rescue options");
524 btrfs_set_and_info(info, IGNOREDATACSUMS,
525 "ignoring data csums");
526 btrfs_set_and_info(info, IGNOREBADROOTS,
527 "ignoring bad roots");
528 btrfs_set_and_info(info, NOLOGREPLAY,
529 "disabling log replay at mount time");
532 btrfs_info(info, "unrecognized rescue option '%s'", p);
546 * Regular mount options parser. Everything that is needed only when
547 * reading in a new superblock is parsed here.
548 * XXX JDM: This needs to be cleaned up for remount.
550 int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
551 unsigned long new_flags)
553 substring_t args[MAX_OPT_ARGS];
558 bool compress_force = false;
559 enum btrfs_compression_type saved_compress_type;
560 int saved_compress_level;
561 bool saved_compress_force;
564 if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE))
565 btrfs_set_opt(info->mount_opt, FREE_SPACE_TREE);
566 else if (btrfs_free_space_cache_v1_active(info)) {
567 if (btrfs_is_zoned(info)) {
569 "zoned: clearing existing space cache");
570 btrfs_set_super_cache_generation(info->super_copy, 0);
572 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
577 * Even the options are empty, we still need to do extra check
583 while ((p = strsep(&options, ",")) != NULL) {
588 token = match_token(p, tokens, args);
591 btrfs_info(info, "allowing degraded mounts");
592 btrfs_set_opt(info->mount_opt, DEGRADED);
595 case Opt_subvol_empty:
599 * These are parsed by btrfs_parse_subvol_options or
600 * btrfs_parse_device_options and can be ignored here.
604 btrfs_set_and_info(info, NODATASUM,
605 "setting nodatasum");
608 if (btrfs_test_opt(info, NODATASUM)) {
609 if (btrfs_test_opt(info, NODATACOW))
611 "setting datasum, datacow enabled");
613 btrfs_info(info, "setting datasum");
615 btrfs_clear_opt(info->mount_opt, NODATACOW);
616 btrfs_clear_opt(info->mount_opt, NODATASUM);
619 if (!btrfs_test_opt(info, NODATACOW)) {
620 if (!btrfs_test_opt(info, COMPRESS) ||
621 !btrfs_test_opt(info, FORCE_COMPRESS)) {
623 "setting nodatacow, compression disabled");
625 btrfs_info(info, "setting nodatacow");
628 btrfs_clear_opt(info->mount_opt, COMPRESS);
629 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
630 btrfs_set_opt(info->mount_opt, NODATACOW);
631 btrfs_set_opt(info->mount_opt, NODATASUM);
634 btrfs_clear_and_info(info, NODATACOW,
637 case Opt_compress_force:
638 case Opt_compress_force_type:
639 compress_force = true;
642 case Opt_compress_type:
643 saved_compress_type = btrfs_test_opt(info,
645 info->compress_type : BTRFS_COMPRESS_NONE;
646 saved_compress_force =
647 btrfs_test_opt(info, FORCE_COMPRESS);
648 saved_compress_level = info->compress_level;
649 if (token == Opt_compress ||
650 token == Opt_compress_force ||
651 strncmp(args[0].from, "zlib", 4) == 0) {
652 compress_type = "zlib";
654 info->compress_type = BTRFS_COMPRESS_ZLIB;
655 info->compress_level = BTRFS_ZLIB_DEFAULT_LEVEL;
657 * args[0] contains uninitialized data since
658 * for these tokens we don't expect any
661 if (token != Opt_compress &&
662 token != Opt_compress_force)
663 info->compress_level =
664 btrfs_compress_str2level(
667 btrfs_set_opt(info->mount_opt, COMPRESS);
668 btrfs_clear_opt(info->mount_opt, NODATACOW);
669 btrfs_clear_opt(info->mount_opt, NODATASUM);
671 } else if (strncmp(args[0].from, "lzo", 3) == 0) {
672 compress_type = "lzo";
673 info->compress_type = BTRFS_COMPRESS_LZO;
674 info->compress_level = 0;
675 btrfs_set_opt(info->mount_opt, COMPRESS);
676 btrfs_clear_opt(info->mount_opt, NODATACOW);
677 btrfs_clear_opt(info->mount_opt, NODATASUM);
678 btrfs_set_fs_incompat(info, COMPRESS_LZO);
680 } else if (strncmp(args[0].from, "zstd", 4) == 0) {
681 compress_type = "zstd";
682 info->compress_type = BTRFS_COMPRESS_ZSTD;
683 info->compress_level =
684 btrfs_compress_str2level(
687 btrfs_set_opt(info->mount_opt, COMPRESS);
688 btrfs_clear_opt(info->mount_opt, NODATACOW);
689 btrfs_clear_opt(info->mount_opt, NODATASUM);
690 btrfs_set_fs_incompat(info, COMPRESS_ZSTD);
692 } else if (strncmp(args[0].from, "no", 2) == 0) {
693 compress_type = "no";
694 info->compress_level = 0;
695 info->compress_type = 0;
696 btrfs_clear_opt(info->mount_opt, COMPRESS);
697 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
698 compress_force = false;
705 if (compress_force) {
706 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
709 * If we remount from compress-force=xxx to
710 * compress=xxx, we need clear FORCE_COMPRESS
711 * flag, otherwise, there is no way for users
712 * to disable forcible compression separately.
714 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
716 if (no_compress == 1) {
717 btrfs_info(info, "use no compression");
718 } else if ((info->compress_type != saved_compress_type) ||
719 (compress_force != saved_compress_force) ||
720 (info->compress_level != saved_compress_level)) {
721 btrfs_info(info, "%s %s compression, level %d",
722 (compress_force) ? "force" : "use",
723 compress_type, info->compress_level);
725 compress_force = false;
728 btrfs_set_and_info(info, SSD,
729 "enabling ssd optimizations");
730 btrfs_clear_opt(info->mount_opt, NOSSD);
733 btrfs_set_and_info(info, SSD,
734 "enabling ssd optimizations");
735 btrfs_set_and_info(info, SSD_SPREAD,
736 "using spread ssd allocation scheme");
737 btrfs_clear_opt(info->mount_opt, NOSSD);
740 btrfs_set_opt(info->mount_opt, NOSSD);
741 btrfs_clear_and_info(info, SSD,
742 "not using ssd optimizations");
744 case Opt_nossd_spread:
745 btrfs_clear_and_info(info, SSD_SPREAD,
746 "not using spread ssd allocation scheme");
749 btrfs_clear_and_info(info, NOBARRIER,
750 "turning on barriers");
753 btrfs_set_and_info(info, NOBARRIER,
754 "turning off barriers");
756 case Opt_thread_pool:
757 ret = match_int(&args[0], &intarg);
760 } else if (intarg == 0) {
764 info->thread_pool_size = intarg;
767 num = match_strdup(&args[0]);
769 info->max_inline = memparse(num, NULL);
772 if (info->max_inline) {
773 info->max_inline = min_t(u64,
777 btrfs_info(info, "max_inline at %llu",
785 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
786 info->sb->s_flags |= SB_POSIXACL;
789 btrfs_err(info, "support for ACL not compiled in!");
794 info->sb->s_flags &= ~SB_POSIXACL;
797 btrfs_set_and_info(info, NOTREELOG,
798 "disabling tree log");
801 btrfs_clear_and_info(info, NOTREELOG,
802 "enabling tree log");
805 case Opt_nologreplay:
807 "'nologreplay' is deprecated, use 'rescue=nologreplay' instead");
808 btrfs_set_and_info(info, NOLOGREPLAY,
809 "disabling log replay at mount time");
811 case Opt_flushoncommit:
812 btrfs_set_and_info(info, FLUSHONCOMMIT,
813 "turning on flush-on-commit");
815 case Opt_noflushoncommit:
816 btrfs_clear_and_info(info, FLUSHONCOMMIT,
817 "turning off flush-on-commit");
820 ret = match_int(&args[0], &intarg);
823 info->metadata_ratio = intarg;
824 btrfs_info(info, "metadata ratio %u",
825 info->metadata_ratio);
828 case Opt_discard_mode:
829 if (token == Opt_discard ||
830 strcmp(args[0].from, "sync") == 0) {
831 btrfs_clear_opt(info->mount_opt, DISCARD_ASYNC);
832 btrfs_set_and_info(info, DISCARD_SYNC,
833 "turning on sync discard");
834 } else if (strcmp(args[0].from, "async") == 0) {
835 btrfs_clear_opt(info->mount_opt, DISCARD_SYNC);
836 btrfs_set_and_info(info, DISCARD_ASYNC,
837 "turning on async discard");
844 btrfs_clear_and_info(info, DISCARD_SYNC,
845 "turning off discard");
846 btrfs_clear_and_info(info, DISCARD_ASYNC,
847 "turning off async discard");
849 case Opt_space_cache:
850 case Opt_space_cache_version:
851 if (token == Opt_space_cache ||
852 strcmp(args[0].from, "v1") == 0) {
853 btrfs_clear_opt(info->mount_opt,
855 btrfs_set_and_info(info, SPACE_CACHE,
856 "enabling disk space caching");
857 } else if (strcmp(args[0].from, "v2") == 0) {
858 btrfs_clear_opt(info->mount_opt,
860 btrfs_set_and_info(info, FREE_SPACE_TREE,
861 "enabling free space tree");
867 case Opt_rescan_uuid_tree:
868 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
870 case Opt_no_space_cache:
871 if (btrfs_test_opt(info, SPACE_CACHE)) {
872 btrfs_clear_and_info(info, SPACE_CACHE,
873 "disabling disk space caching");
875 if (btrfs_test_opt(info, FREE_SPACE_TREE)) {
876 btrfs_clear_and_info(info, FREE_SPACE_TREE,
877 "disabling free space tree");
880 case Opt_inode_cache:
881 case Opt_noinode_cache:
883 "the 'inode_cache' option is deprecated and has no effect since 5.11");
885 case Opt_clear_cache:
886 btrfs_set_and_info(info, CLEAR_CACHE,
887 "force clearing of disk cache");
889 case Opt_user_subvol_rm_allowed:
890 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
892 case Opt_enospc_debug:
893 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
895 case Opt_noenospc_debug:
896 btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG);
899 btrfs_set_and_info(info, AUTO_DEFRAG,
900 "enabling auto defrag");
903 btrfs_clear_and_info(info, AUTO_DEFRAG,
904 "disabling auto defrag");
907 case Opt_usebackuproot:
909 "'%s' is deprecated, use 'rescue=usebackuproot' instead",
910 token == Opt_recovery ? "recovery" :
913 "trying to use backup root at mount time");
914 btrfs_set_opt(info->mount_opt, USEBACKUPROOT);
916 case Opt_skip_balance:
917 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
919 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
920 case Opt_check_integrity_including_extent_data:
922 "enabling check integrity including extent data");
923 btrfs_set_opt(info->mount_opt,
924 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
925 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
927 case Opt_check_integrity:
928 btrfs_info(info, "enabling check integrity");
929 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
931 case Opt_check_integrity_print_mask:
932 ret = match_int(&args[0], &intarg);
935 info->check_integrity_print_mask = intarg;
936 btrfs_info(info, "check_integrity_print_mask 0x%x",
937 info->check_integrity_print_mask);
940 case Opt_check_integrity_including_extent_data:
941 case Opt_check_integrity:
942 case Opt_check_integrity_print_mask:
944 "support for check_integrity* not compiled in!");
948 case Opt_fatal_errors:
949 if (strcmp(args[0].from, "panic") == 0)
950 btrfs_set_opt(info->mount_opt,
951 PANIC_ON_FATAL_ERROR);
952 else if (strcmp(args[0].from, "bug") == 0)
953 btrfs_clear_opt(info->mount_opt,
954 PANIC_ON_FATAL_ERROR);
960 case Opt_commit_interval:
962 ret = match_int(&args[0], &intarg);
967 "using default commit interval %us",
968 BTRFS_DEFAULT_COMMIT_INTERVAL);
969 intarg = BTRFS_DEFAULT_COMMIT_INTERVAL;
970 } else if (intarg > 300) {
971 btrfs_warn(info, "excessive commit interval %d",
974 info->commit_interval = intarg;
977 ret = parse_rescue_options(info, args[0].from);
981 #ifdef CONFIG_BTRFS_DEBUG
982 case Opt_fragment_all:
983 btrfs_info(info, "fragmenting all space");
984 btrfs_set_opt(info->mount_opt, FRAGMENT_DATA);
985 btrfs_set_opt(info->mount_opt, FRAGMENT_METADATA);
987 case Opt_fragment_metadata:
988 btrfs_info(info, "fragmenting metadata");
989 btrfs_set_opt(info->mount_opt,
992 case Opt_fragment_data:
993 btrfs_info(info, "fragmenting data");
994 btrfs_set_opt(info->mount_opt, FRAGMENT_DATA);
997 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
999 btrfs_info(info, "doing ref verification");
1000 btrfs_set_opt(info->mount_opt, REF_VERIFY);
1004 btrfs_err(info, "unrecognized mount option '%s'", p);
1012 /* We're read-only, don't have to check. */
1013 if (new_flags & SB_RDONLY)
1016 if (check_ro_option(info, BTRFS_MOUNT_NOLOGREPLAY, "nologreplay") ||
1017 check_ro_option(info, BTRFS_MOUNT_IGNOREBADROOTS, "ignorebadroots") ||
1018 check_ro_option(info, BTRFS_MOUNT_IGNOREDATACSUMS, "ignoredatacsums"))
1021 if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE) &&
1022 !btrfs_test_opt(info, FREE_SPACE_TREE) &&
1023 !btrfs_test_opt(info, CLEAR_CACHE)) {
1024 btrfs_err(info, "cannot disable free space tree");
1029 ret = btrfs_check_mountopts_zoned(info);
1030 if (!ret && btrfs_test_opt(info, SPACE_CACHE))
1031 btrfs_info(info, "disk space caching is enabled");
1032 if (!ret && btrfs_test_opt(info, FREE_SPACE_TREE))
1033 btrfs_info(info, "using free space tree");
1038 * Parse mount options that are required early in the mount process.
1040 * All other options will be parsed on much later in the mount process and
1041 * only when we need to allocate a new super block.
1043 static int btrfs_parse_device_options(const char *options, fmode_t flags,
1046 substring_t args[MAX_OPT_ARGS];
1047 char *device_name, *opts, *orig, *p;
1048 struct btrfs_device *device = NULL;
1051 lockdep_assert_held(&uuid_mutex);
1057 * strsep changes the string, duplicate it because btrfs_parse_options
1060 opts = kstrdup(options, GFP_KERNEL);
1065 while ((p = strsep(&opts, ",")) != NULL) {
1071 token = match_token(p, tokens, args);
1072 if (token == Opt_device) {
1073 device_name = match_strdup(&args[0]);
1078 device = btrfs_scan_one_device(device_name, flags,
1081 if (IS_ERR(device)) {
1082 error = PTR_ERR(device);
1094 * Parse mount options that are related to subvolume id
1096 * The value is later passed to mount_subvol()
1098 static int btrfs_parse_subvol_options(const char *options, char **subvol_name,
1099 u64 *subvol_objectid)
1101 substring_t args[MAX_OPT_ARGS];
1102 char *opts, *orig, *p;
1110 * strsep changes the string, duplicate it because
1111 * btrfs_parse_device_options gets called later
1113 opts = kstrdup(options, GFP_KERNEL);
1118 while ((p = strsep(&opts, ",")) != NULL) {
1123 token = match_token(p, tokens, args);
1126 kfree(*subvol_name);
1127 *subvol_name = match_strdup(&args[0]);
1128 if (!*subvol_name) {
1134 error = match_u64(&args[0], &subvolid);
1138 /* we want the original fs_tree */
1140 subvolid = BTRFS_FS_TREE_OBJECTID;
1142 *subvol_objectid = subvolid;
1154 char *btrfs_get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
1155 u64 subvol_objectid)
1157 struct btrfs_root *root = fs_info->tree_root;
1158 struct btrfs_root *fs_root = NULL;
1159 struct btrfs_root_ref *root_ref;
1160 struct btrfs_inode_ref *inode_ref;
1161 struct btrfs_key key;
1162 struct btrfs_path *path = NULL;
1163 char *name = NULL, *ptr;
1168 path = btrfs_alloc_path();
1174 name = kmalloc(PATH_MAX, GFP_KERNEL);
1179 ptr = name + PATH_MAX - 1;
1183 * Walk up the subvolume trees in the tree of tree roots by root
1184 * backrefs until we hit the top-level subvolume.
1186 while (subvol_objectid != BTRFS_FS_TREE_OBJECTID) {
1187 key.objectid = subvol_objectid;
1188 key.type = BTRFS_ROOT_BACKREF_KEY;
1189 key.offset = (u64)-1;
1191 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1194 } else if (ret > 0) {
1195 ret = btrfs_previous_item(root, path, subvol_objectid,
1196 BTRFS_ROOT_BACKREF_KEY);
1199 } else if (ret > 0) {
1205 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1206 subvol_objectid = key.offset;
1208 root_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
1209 struct btrfs_root_ref);
1210 len = btrfs_root_ref_name_len(path->nodes[0], root_ref);
1213 ret = -ENAMETOOLONG;
1216 read_extent_buffer(path->nodes[0], ptr + 1,
1217 (unsigned long)(root_ref + 1), len);
1219 dirid = btrfs_root_ref_dirid(path->nodes[0], root_ref);
1220 btrfs_release_path(path);
1222 fs_root = btrfs_get_fs_root(fs_info, subvol_objectid, true);
1223 if (IS_ERR(fs_root)) {
1224 ret = PTR_ERR(fs_root);
1230 * Walk up the filesystem tree by inode refs until we hit the
1233 while (dirid != BTRFS_FIRST_FREE_OBJECTID) {
1234 key.objectid = dirid;
1235 key.type = BTRFS_INODE_REF_KEY;
1236 key.offset = (u64)-1;
1238 ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
1241 } else if (ret > 0) {
1242 ret = btrfs_previous_item(fs_root, path, dirid,
1243 BTRFS_INODE_REF_KEY);
1246 } else if (ret > 0) {
1252 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
1255 inode_ref = btrfs_item_ptr(path->nodes[0],
1257 struct btrfs_inode_ref);
1258 len = btrfs_inode_ref_name_len(path->nodes[0],
1262 ret = -ENAMETOOLONG;
1265 read_extent_buffer(path->nodes[0], ptr + 1,
1266 (unsigned long)(inode_ref + 1), len);
1268 btrfs_release_path(path);
1270 btrfs_put_root(fs_root);
1274 btrfs_free_path(path);
1275 if (ptr == name + PATH_MAX - 1) {
1279 memmove(name, ptr, name + PATH_MAX - ptr);
1284 btrfs_put_root(fs_root);
1285 btrfs_free_path(path);
1287 return ERR_PTR(ret);
1290 static int get_default_subvol_objectid(struct btrfs_fs_info *fs_info, u64 *objectid)
1292 struct btrfs_root *root = fs_info->tree_root;
1293 struct btrfs_dir_item *di;
1294 struct btrfs_path *path;
1295 struct btrfs_key location;
1298 path = btrfs_alloc_path();
1303 * Find the "default" dir item which points to the root item that we
1304 * will mount by default if we haven't been given a specific subvolume
1307 dir_id = btrfs_super_root_dir(fs_info->super_copy);
1308 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
1310 btrfs_free_path(path);
1315 * Ok the default dir item isn't there. This is weird since
1316 * it's always been there, but don't freak out, just try and
1317 * mount the top-level subvolume.
1319 btrfs_free_path(path);
1320 *objectid = BTRFS_FS_TREE_OBJECTID;
1324 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
1325 btrfs_free_path(path);
1326 *objectid = location.objectid;
1330 static int btrfs_fill_super(struct super_block *sb,
1331 struct btrfs_fs_devices *fs_devices,
1334 struct inode *inode;
1335 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1338 sb->s_maxbytes = MAX_LFS_FILESIZE;
1339 sb->s_magic = BTRFS_SUPER_MAGIC;
1340 sb->s_op = &btrfs_super_ops;
1341 sb->s_d_op = &btrfs_dentry_operations;
1342 sb->s_export_op = &btrfs_export_ops;
1343 sb->s_xattr = btrfs_xattr_handlers;
1344 sb->s_time_gran = 1;
1345 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
1346 sb->s_flags |= SB_POSIXACL;
1348 sb->s_flags |= SB_I_VERSION;
1349 sb->s_iflags |= SB_I_CGROUPWB;
1351 err = super_setup_bdi(sb);
1353 btrfs_err(fs_info, "super_setup_bdi failed");
1357 err = open_ctree(sb, fs_devices, (char *)data);
1359 btrfs_err(fs_info, "open_ctree failed");
1363 inode = btrfs_iget(sb, BTRFS_FIRST_FREE_OBJECTID, fs_info->fs_root);
1364 if (IS_ERR(inode)) {
1365 err = PTR_ERR(inode);
1369 sb->s_root = d_make_root(inode);
1375 cleancache_init_fs(sb);
1376 sb->s_flags |= SB_ACTIVE;
1380 close_ctree(fs_info);
1384 int btrfs_sync_fs(struct super_block *sb, int wait)
1386 struct btrfs_trans_handle *trans;
1387 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1388 struct btrfs_root *root = fs_info->tree_root;
1390 trace_btrfs_sync_fs(fs_info, wait);
1393 filemap_flush(fs_info->btree_inode->i_mapping);
1397 btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1);
1399 trans = btrfs_attach_transaction_barrier(root);
1400 if (IS_ERR(trans)) {
1401 /* no transaction, don't bother */
1402 if (PTR_ERR(trans) == -ENOENT) {
1404 * Exit unless we have some pending changes
1405 * that need to go through commit
1407 if (fs_info->pending_changes == 0)
1410 * A non-blocking test if the fs is frozen. We must not
1411 * start a new transaction here otherwise a deadlock
1412 * happens. The pending operations are delayed to the
1413 * next commit after thawing.
1415 if (sb_start_write_trylock(sb))
1419 trans = btrfs_start_transaction(root, 0);
1422 return PTR_ERR(trans);
1424 return btrfs_commit_transaction(trans);
1427 static void print_rescue_option(struct seq_file *seq, const char *s, bool *printed)
1429 seq_printf(seq, "%s%s", (*printed) ? ":" : ",rescue=", s);
1433 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
1435 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
1436 const char *compress_type;
1437 const char *subvol_name;
1438 bool printed = false;
1440 if (btrfs_test_opt(info, DEGRADED))
1441 seq_puts(seq, ",degraded");
1442 if (btrfs_test_opt(info, NODATASUM))
1443 seq_puts(seq, ",nodatasum");
1444 if (btrfs_test_opt(info, NODATACOW))
1445 seq_puts(seq, ",nodatacow");
1446 if (btrfs_test_opt(info, NOBARRIER))
1447 seq_puts(seq, ",nobarrier");
1448 if (info->max_inline != BTRFS_DEFAULT_MAX_INLINE)
1449 seq_printf(seq, ",max_inline=%llu", info->max_inline);
1450 if (info->thread_pool_size != min_t(unsigned long,
1451 num_online_cpus() + 2, 8))
1452 seq_printf(seq, ",thread_pool=%u", info->thread_pool_size);
1453 if (btrfs_test_opt(info, COMPRESS)) {
1454 compress_type = btrfs_compress_type2str(info->compress_type);
1455 if (btrfs_test_opt(info, FORCE_COMPRESS))
1456 seq_printf(seq, ",compress-force=%s", compress_type);
1458 seq_printf(seq, ",compress=%s", compress_type);
1459 if (info->compress_level)
1460 seq_printf(seq, ":%d", info->compress_level);
1462 if (btrfs_test_opt(info, NOSSD))
1463 seq_puts(seq, ",nossd");
1464 if (btrfs_test_opt(info, SSD_SPREAD))
1465 seq_puts(seq, ",ssd_spread");
1466 else if (btrfs_test_opt(info, SSD))
1467 seq_puts(seq, ",ssd");
1468 if (btrfs_test_opt(info, NOTREELOG))
1469 seq_puts(seq, ",notreelog");
1470 if (btrfs_test_opt(info, NOLOGREPLAY))
1471 print_rescue_option(seq, "nologreplay", &printed);
1472 if (btrfs_test_opt(info, USEBACKUPROOT))
1473 print_rescue_option(seq, "usebackuproot", &printed);
1474 if (btrfs_test_opt(info, IGNOREBADROOTS))
1475 print_rescue_option(seq, "ignorebadroots", &printed);
1476 if (btrfs_test_opt(info, IGNOREDATACSUMS))
1477 print_rescue_option(seq, "ignoredatacsums", &printed);
1478 if (btrfs_test_opt(info, FLUSHONCOMMIT))
1479 seq_puts(seq, ",flushoncommit");
1480 if (btrfs_test_opt(info, DISCARD_SYNC))
1481 seq_puts(seq, ",discard");
1482 if (btrfs_test_opt(info, DISCARD_ASYNC))
1483 seq_puts(seq, ",discard=async");
1484 if (!(info->sb->s_flags & SB_POSIXACL))
1485 seq_puts(seq, ",noacl");
1486 if (btrfs_free_space_cache_v1_active(info))
1487 seq_puts(seq, ",space_cache");
1488 else if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE))
1489 seq_puts(seq, ",space_cache=v2");
1491 seq_puts(seq, ",nospace_cache");
1492 if (btrfs_test_opt(info, RESCAN_UUID_TREE))
1493 seq_puts(seq, ",rescan_uuid_tree");
1494 if (btrfs_test_opt(info, CLEAR_CACHE))
1495 seq_puts(seq, ",clear_cache");
1496 if (btrfs_test_opt(info, USER_SUBVOL_RM_ALLOWED))
1497 seq_puts(seq, ",user_subvol_rm_allowed");
1498 if (btrfs_test_opt(info, ENOSPC_DEBUG))
1499 seq_puts(seq, ",enospc_debug");
1500 if (btrfs_test_opt(info, AUTO_DEFRAG))
1501 seq_puts(seq, ",autodefrag");
1502 if (btrfs_test_opt(info, SKIP_BALANCE))
1503 seq_puts(seq, ",skip_balance");
1504 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1505 if (btrfs_test_opt(info, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1506 seq_puts(seq, ",check_int_data");
1507 else if (btrfs_test_opt(info, CHECK_INTEGRITY))
1508 seq_puts(seq, ",check_int");
1509 if (info->check_integrity_print_mask)
1510 seq_printf(seq, ",check_int_print_mask=%d",
1511 info->check_integrity_print_mask);
1513 if (info->metadata_ratio)
1514 seq_printf(seq, ",metadata_ratio=%u", info->metadata_ratio);
1515 if (btrfs_test_opt(info, PANIC_ON_FATAL_ERROR))
1516 seq_puts(seq, ",fatal_errors=panic");
1517 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1518 seq_printf(seq, ",commit=%u", info->commit_interval);
1519 #ifdef CONFIG_BTRFS_DEBUG
1520 if (btrfs_test_opt(info, FRAGMENT_DATA))
1521 seq_puts(seq, ",fragment=data");
1522 if (btrfs_test_opt(info, FRAGMENT_METADATA))
1523 seq_puts(seq, ",fragment=metadata");
1525 if (btrfs_test_opt(info, REF_VERIFY))
1526 seq_puts(seq, ",ref_verify");
1527 seq_printf(seq, ",subvolid=%llu",
1528 BTRFS_I(d_inode(dentry))->root->root_key.objectid);
1529 subvol_name = btrfs_get_subvol_name_from_objectid(info,
1530 BTRFS_I(d_inode(dentry))->root->root_key.objectid);
1531 if (!IS_ERR(subvol_name)) {
1532 seq_puts(seq, ",subvol=");
1533 seq_escape(seq, subvol_name, " \t\n\\");
1539 static int btrfs_test_super(struct super_block *s, void *data)
1541 struct btrfs_fs_info *p = data;
1542 struct btrfs_fs_info *fs_info = btrfs_sb(s);
1544 return fs_info->fs_devices == p->fs_devices;
1547 static int btrfs_set_super(struct super_block *s, void *data)
1549 int err = set_anon_super(s, data);
1551 s->s_fs_info = data;
1556 * subvolumes are identified by ino 256
1558 static inline int is_subvolume_inode(struct inode *inode)
1560 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1565 static struct dentry *mount_subvol(const char *subvol_name, u64 subvol_objectid,
1566 struct vfsmount *mnt)
1568 struct dentry *root;
1572 if (!subvol_objectid) {
1573 ret = get_default_subvol_objectid(btrfs_sb(mnt->mnt_sb),
1576 root = ERR_PTR(ret);
1580 subvol_name = btrfs_get_subvol_name_from_objectid(
1581 btrfs_sb(mnt->mnt_sb), subvol_objectid);
1582 if (IS_ERR(subvol_name)) {
1583 root = ERR_CAST(subvol_name);
1590 root = mount_subtree(mnt, subvol_name);
1591 /* mount_subtree() drops our reference on the vfsmount. */
1594 if (!IS_ERR(root)) {
1595 struct super_block *s = root->d_sb;
1596 struct btrfs_fs_info *fs_info = btrfs_sb(s);
1597 struct inode *root_inode = d_inode(root);
1598 u64 root_objectid = BTRFS_I(root_inode)->root->root_key.objectid;
1601 if (!is_subvolume_inode(root_inode)) {
1602 btrfs_err(fs_info, "'%s' is not a valid subvolume",
1606 if (subvol_objectid && root_objectid != subvol_objectid) {
1608 * This will also catch a race condition where a
1609 * subvolume which was passed by ID is renamed and
1610 * another subvolume is renamed over the old location.
1613 "subvol '%s' does not match subvolid %llu",
1614 subvol_name, subvol_objectid);
1619 root = ERR_PTR(ret);
1620 deactivate_locked_super(s);
1631 * Find a superblock for the given device / mount point.
1633 * Note: This is based on mount_bdev from fs/super.c with a few additions
1634 * for multiple device setup. Make sure to keep it in sync.
1636 static struct dentry *btrfs_mount_root(struct file_system_type *fs_type,
1637 int flags, const char *device_name, void *data)
1639 struct block_device *bdev = NULL;
1640 struct super_block *s;
1641 struct btrfs_device *device = NULL;
1642 struct btrfs_fs_devices *fs_devices = NULL;
1643 struct btrfs_fs_info *fs_info = NULL;
1644 void *new_sec_opts = NULL;
1645 fmode_t mode = FMODE_READ;
1648 if (!(flags & SB_RDONLY))
1649 mode |= FMODE_WRITE;
1652 error = security_sb_eat_lsm_opts(data, &new_sec_opts);
1654 return ERR_PTR(error);
1658 * Setup a dummy root and fs_info for test/set super. This is because
1659 * we don't actually fill this stuff out until open_ctree, but we need
1660 * then open_ctree will properly initialize the file system specific
1661 * settings later. btrfs_init_fs_info initializes the static elements
1662 * of the fs_info (locks and such) to make cleanup easier if we find a
1663 * superblock with our given fs_devices later on at sget() time.
1665 fs_info = kvzalloc(sizeof(struct btrfs_fs_info), GFP_KERNEL);
1668 goto error_sec_opts;
1670 btrfs_init_fs_info(fs_info);
1672 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL);
1673 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL);
1674 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1679 mutex_lock(&uuid_mutex);
1680 error = btrfs_parse_device_options(data, mode, fs_type);
1682 mutex_unlock(&uuid_mutex);
1686 device = btrfs_scan_one_device(device_name, mode, fs_type);
1687 if (IS_ERR(device)) {
1688 mutex_unlock(&uuid_mutex);
1689 error = PTR_ERR(device);
1693 fs_devices = device->fs_devices;
1694 fs_info->fs_devices = fs_devices;
1696 error = btrfs_open_devices(fs_devices, mode, fs_type);
1697 mutex_unlock(&uuid_mutex);
1701 if (!(flags & SB_RDONLY) && fs_devices->rw_devices == 0) {
1703 goto error_close_devices;
1706 bdev = fs_devices->latest_bdev;
1707 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | SB_NOSEC,
1711 goto error_close_devices;
1715 btrfs_close_devices(fs_devices);
1716 btrfs_free_fs_info(fs_info);
1717 if ((flags ^ s->s_flags) & SB_RDONLY)
1720 snprintf(s->s_id, sizeof(s->s_id), "%pg", bdev);
1721 btrfs_sb(s)->bdev_holder = fs_type;
1722 if (!strstr(crc32c_impl(), "generic"))
1723 set_bit(BTRFS_FS_CSUM_IMPL_FAST, &fs_info->flags);
1724 error = btrfs_fill_super(s, fs_devices, data);
1727 error = security_sb_set_mnt_opts(s, new_sec_opts, 0, NULL);
1728 security_free_mnt_opts(&new_sec_opts);
1730 deactivate_locked_super(s);
1731 return ERR_PTR(error);
1734 return dget(s->s_root);
1736 error_close_devices:
1737 btrfs_close_devices(fs_devices);
1739 btrfs_free_fs_info(fs_info);
1741 security_free_mnt_opts(&new_sec_opts);
1742 return ERR_PTR(error);
1746 * Mount function which is called by VFS layer.
1748 * In order to allow mounting a subvolume directly, btrfs uses mount_subtree()
1749 * which needs vfsmount* of device's root (/). This means device's root has to
1750 * be mounted internally in any case.
1753 * 1. Parse subvol id related options for later use in mount_subvol().
1755 * 2. Mount device's root (/) by calling vfs_kern_mount().
1757 * NOTE: vfs_kern_mount() is used by VFS to call btrfs_mount() in the
1758 * first place. In order to avoid calling btrfs_mount() again, we use
1759 * different file_system_type which is not registered to VFS by
1760 * register_filesystem() (btrfs_root_fs_type). As a result,
1761 * btrfs_mount_root() is called. The return value will be used by
1762 * mount_subtree() in mount_subvol().
1764 * 3. Call mount_subvol() to get the dentry of subvolume. Since there is
1765 * "btrfs subvolume set-default", mount_subvol() is called always.
1767 static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
1768 const char *device_name, void *data)
1770 struct vfsmount *mnt_root;
1771 struct dentry *root;
1772 char *subvol_name = NULL;
1773 u64 subvol_objectid = 0;
1776 error = btrfs_parse_subvol_options(data, &subvol_name,
1780 return ERR_PTR(error);
1783 /* mount device's root (/) */
1784 mnt_root = vfs_kern_mount(&btrfs_root_fs_type, flags, device_name, data);
1785 if (PTR_ERR_OR_ZERO(mnt_root) == -EBUSY) {
1786 if (flags & SB_RDONLY) {
1787 mnt_root = vfs_kern_mount(&btrfs_root_fs_type,
1788 flags & ~SB_RDONLY, device_name, data);
1790 mnt_root = vfs_kern_mount(&btrfs_root_fs_type,
1791 flags | SB_RDONLY, device_name, data);
1792 if (IS_ERR(mnt_root)) {
1793 root = ERR_CAST(mnt_root);
1798 down_write(&mnt_root->mnt_sb->s_umount);
1799 error = btrfs_remount(mnt_root->mnt_sb, &flags, NULL);
1800 up_write(&mnt_root->mnt_sb->s_umount);
1802 root = ERR_PTR(error);
1809 if (IS_ERR(mnt_root)) {
1810 root = ERR_CAST(mnt_root);
1815 /* mount_subvol() will free subvol_name and mnt_root */
1816 root = mount_subvol(subvol_name, subvol_objectid, mnt_root);
1822 static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1823 u32 new_pool_size, u32 old_pool_size)
1825 if (new_pool_size == old_pool_size)
1828 fs_info->thread_pool_size = new_pool_size;
1830 btrfs_info(fs_info, "resize thread pool %d -> %d",
1831 old_pool_size, new_pool_size);
1833 btrfs_workqueue_set_max(fs_info->workers, new_pool_size);
1834 btrfs_workqueue_set_max(fs_info->delalloc_workers, new_pool_size);
1835 btrfs_workqueue_set_max(fs_info->caching_workers, new_pool_size);
1836 btrfs_workqueue_set_max(fs_info->endio_workers, new_pool_size);
1837 btrfs_workqueue_set_max(fs_info->endio_meta_workers, new_pool_size);
1838 btrfs_workqueue_set_max(fs_info->endio_meta_write_workers,
1840 btrfs_workqueue_set_max(fs_info->endio_write_workers, new_pool_size);
1841 btrfs_workqueue_set_max(fs_info->endio_freespace_worker, new_pool_size);
1842 btrfs_workqueue_set_max(fs_info->delayed_workers, new_pool_size);
1843 btrfs_workqueue_set_max(fs_info->readahead_workers, new_pool_size);
1844 btrfs_workqueue_set_max(fs_info->scrub_wr_completion_workers,
1848 static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1849 unsigned long old_opts, int flags)
1851 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1852 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1853 (flags & SB_RDONLY))) {
1854 /* wait for any defraggers to finish */
1855 wait_event(fs_info->transaction_wait,
1856 (atomic_read(&fs_info->defrag_running) == 0));
1857 if (flags & SB_RDONLY)
1858 sync_filesystem(fs_info->sb);
1862 static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1863 unsigned long old_opts)
1865 const bool cache_opt = btrfs_test_opt(fs_info, SPACE_CACHE);
1868 * We need to cleanup all defragable inodes if the autodefragment is
1869 * close or the filesystem is read only.
1871 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1872 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || sb_rdonly(fs_info->sb))) {
1873 btrfs_cleanup_defrag_inodes(fs_info);
1876 /* If we toggled discard async */
1877 if (!btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) &&
1878 btrfs_test_opt(fs_info, DISCARD_ASYNC))
1879 btrfs_discard_resume(fs_info);
1880 else if (btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) &&
1881 !btrfs_test_opt(fs_info, DISCARD_ASYNC))
1882 btrfs_discard_cleanup(fs_info);
1884 /* If we toggled space cache */
1885 if (cache_opt != btrfs_free_space_cache_v1_active(fs_info))
1886 btrfs_set_free_space_cache_v1_active(fs_info, cache_opt);
1889 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1891 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1892 unsigned old_flags = sb->s_flags;
1893 unsigned long old_opts = fs_info->mount_opt;
1894 unsigned long old_compress_type = fs_info->compress_type;
1895 u64 old_max_inline = fs_info->max_inline;
1896 u32 old_thread_pool_size = fs_info->thread_pool_size;
1897 u32 old_metadata_ratio = fs_info->metadata_ratio;
1900 sync_filesystem(sb);
1901 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1904 void *new_sec_opts = NULL;
1906 ret = security_sb_eat_lsm_opts(data, &new_sec_opts);
1908 ret = security_sb_remount(sb, new_sec_opts);
1909 security_free_mnt_opts(&new_sec_opts);
1914 ret = btrfs_parse_options(fs_info, data, *flags);
1918 btrfs_remount_begin(fs_info, old_opts, *flags);
1919 btrfs_resize_thread_pool(fs_info,
1920 fs_info->thread_pool_size, old_thread_pool_size);
1922 if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) !=
1923 btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) &&
1924 (!sb_rdonly(sb) || (*flags & SB_RDONLY))) {
1926 "remount supports changing free space tree only from ro to rw");
1927 /* Make sure free space cache options match the state on disk */
1928 if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
1929 btrfs_set_opt(fs_info->mount_opt, FREE_SPACE_TREE);
1930 btrfs_clear_opt(fs_info->mount_opt, SPACE_CACHE);
1932 if (btrfs_free_space_cache_v1_active(fs_info)) {
1933 btrfs_clear_opt(fs_info->mount_opt, FREE_SPACE_TREE);
1934 btrfs_set_opt(fs_info->mount_opt, SPACE_CACHE);
1938 if ((bool)(*flags & SB_RDONLY) == sb_rdonly(sb))
1941 if (*flags & SB_RDONLY) {
1943 * this also happens on 'umount -rf' or on shutdown, when
1944 * the filesystem is busy.
1946 cancel_work_sync(&fs_info->async_reclaim_work);
1947 cancel_work_sync(&fs_info->async_data_reclaim_work);
1949 btrfs_discard_cleanup(fs_info);
1951 /* wait for the uuid_scan task to finish */
1952 down(&fs_info->uuid_tree_rescan_sem);
1953 /* avoid complains from lockdep et al. */
1954 up(&fs_info->uuid_tree_rescan_sem);
1956 sb->s_flags |= SB_RDONLY;
1959 * Setting SB_RDONLY will put the cleaner thread to
1960 * sleep at the next loop if it's already active.
1961 * If it's already asleep, we'll leave unused block
1962 * groups on disk until we're mounted read-write again
1963 * unless we clean them up here.
1965 btrfs_delete_unused_bgs(fs_info);
1967 btrfs_dev_replace_suspend_for_unmount(fs_info);
1968 btrfs_scrub_cancel(fs_info);
1969 btrfs_pause_balance(fs_info);
1971 ret = btrfs_commit_super(fs_info);
1975 if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
1977 "Remounting read-write after error is not allowed");
1981 if (fs_info->fs_devices->rw_devices == 0) {
1986 if (!btrfs_check_rw_degradable(fs_info, NULL)) {
1988 "too many missing devices, writable remount is not allowed");
1993 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
1995 "mount required to replay tree-log, cannot remount read-write");
2001 * NOTE: when remounting with a change that does writes, don't
2002 * put it anywhere above this point, as we are not sure to be
2003 * safe to write until we pass the above checks.
2005 ret = btrfs_start_pre_rw_mount(fs_info);
2009 sb->s_flags &= ~SB_RDONLY;
2011 set_bit(BTRFS_FS_OPEN, &fs_info->flags);
2015 * We need to set SB_I_VERSION here otherwise it'll get cleared by VFS,
2016 * since the absence of the flag means it can be toggled off by remount.
2018 *flags |= SB_I_VERSION;
2020 wake_up_process(fs_info->transaction_kthread);
2021 btrfs_remount_cleanup(fs_info, old_opts);
2022 btrfs_clear_oneshot_options(fs_info);
2023 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
2028 /* We've hit an error - don't reset SB_RDONLY */
2030 old_flags |= SB_RDONLY;
2031 sb->s_flags = old_flags;
2032 fs_info->mount_opt = old_opts;
2033 fs_info->compress_type = old_compress_type;
2034 fs_info->max_inline = old_max_inline;
2035 btrfs_resize_thread_pool(fs_info,
2036 old_thread_pool_size, fs_info->thread_pool_size);
2037 fs_info->metadata_ratio = old_metadata_ratio;
2038 btrfs_remount_cleanup(fs_info, old_opts);
2039 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
2044 /* Used to sort the devices by max_avail(descending sort) */
2045 static inline int btrfs_cmp_device_free_bytes(const void *dev_info1,
2046 const void *dev_info2)
2048 if (((struct btrfs_device_info *)dev_info1)->max_avail >
2049 ((struct btrfs_device_info *)dev_info2)->max_avail)
2051 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
2052 ((struct btrfs_device_info *)dev_info2)->max_avail)
2059 * sort the devices by max_avail, in which max free extent size of each device
2060 * is stored.(Descending Sort)
2062 static inline void btrfs_descending_sort_devices(
2063 struct btrfs_device_info *devices,
2066 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
2067 btrfs_cmp_device_free_bytes, NULL);
2071 * The helper to calc the free space on the devices that can be used to store
2074 static inline int btrfs_calc_avail_data_space(struct btrfs_fs_info *fs_info,
2077 struct btrfs_device_info *devices_info;
2078 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
2079 struct btrfs_device *device;
2082 u64 min_stripe_size;
2083 int num_stripes = 1;
2084 int i = 0, nr_devices;
2085 const struct btrfs_raid_attr *rattr;
2088 * We aren't under the device list lock, so this is racy-ish, but good
2089 * enough for our purposes.
2091 nr_devices = fs_info->fs_devices->open_devices;
2094 nr_devices = fs_info->fs_devices->open_devices;
2102 devices_info = kmalloc_array(nr_devices, sizeof(*devices_info),
2107 /* calc min stripe number for data space allocation */
2108 type = btrfs_data_alloc_profile(fs_info);
2109 rattr = &btrfs_raid_array[btrfs_bg_flags_to_raid_index(type)];
2111 if (type & BTRFS_BLOCK_GROUP_RAID0)
2112 num_stripes = nr_devices;
2113 else if (type & BTRFS_BLOCK_GROUP_RAID1)
2115 else if (type & BTRFS_BLOCK_GROUP_RAID1C3)
2117 else if (type & BTRFS_BLOCK_GROUP_RAID1C4)
2119 else if (type & BTRFS_BLOCK_GROUP_RAID10)
2122 /* Adjust for more than 1 stripe per device */
2123 min_stripe_size = rattr->dev_stripes * BTRFS_STRIPE_LEN;
2126 list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) {
2127 if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA,
2128 &device->dev_state) ||
2130 test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state))
2133 if (i >= nr_devices)
2136 avail_space = device->total_bytes - device->bytes_used;
2138 /* align with stripe_len */
2139 avail_space = rounddown(avail_space, BTRFS_STRIPE_LEN);
2142 * In order to avoid overwriting the superblock on the drive,
2143 * btrfs starts at an offset of at least 1MB when doing chunk
2146 * This ensures we have at least min_stripe_size free space
2147 * after excluding 1MB.
2149 if (avail_space <= SZ_1M + min_stripe_size)
2152 avail_space -= SZ_1M;
2154 devices_info[i].dev = device;
2155 devices_info[i].max_avail = avail_space;
2163 btrfs_descending_sort_devices(devices_info, nr_devices);
2167 while (nr_devices >= rattr->devs_min) {
2168 num_stripes = min(num_stripes, nr_devices);
2170 if (devices_info[i].max_avail >= min_stripe_size) {
2174 avail_space += devices_info[i].max_avail * num_stripes;
2175 alloc_size = devices_info[i].max_avail;
2176 for (j = i + 1 - num_stripes; j <= i; j++)
2177 devices_info[j].max_avail -= alloc_size;
2183 kfree(devices_info);
2184 *free_bytes = avail_space;
2189 * Calculate numbers for 'df', pessimistic in case of mixed raid profiles.
2191 * If there's a redundant raid level at DATA block groups, use the respective
2192 * multiplier to scale the sizes.
2194 * Unused device space usage is based on simulating the chunk allocator
2195 * algorithm that respects the device sizes and order of allocations. This is
2196 * a close approximation of the actual use but there are other factors that may
2197 * change the result (like a new metadata chunk).
2199 * If metadata is exhausted, f_bavail will be 0.
2201 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
2203 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
2204 struct btrfs_super_block *disk_super = fs_info->super_copy;
2205 struct btrfs_space_info *found;
2207 u64 total_free_data = 0;
2208 u64 total_free_meta = 0;
2209 u32 bits = fs_info->sectorsize_bits;
2210 __be32 *fsid = (__be32 *)fs_info->fs_devices->fsid;
2211 unsigned factor = 1;
2212 struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv;
2217 list_for_each_entry(found, &fs_info->space_info, list) {
2218 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
2221 total_free_data += found->disk_total - found->disk_used;
2223 btrfs_account_ro_block_groups_free_space(found);
2225 for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) {
2226 if (!list_empty(&found->block_groups[i]))
2227 factor = btrfs_bg_type_to_factor(
2228 btrfs_raid_array[i].bg_flag);
2233 * Metadata in mixed block goup profiles are accounted in data
2235 if (!mixed && found->flags & BTRFS_BLOCK_GROUP_METADATA) {
2236 if (found->flags & BTRFS_BLOCK_GROUP_DATA)
2239 total_free_meta += found->disk_total -
2243 total_used += found->disk_used;
2246 buf->f_blocks = div_u64(btrfs_super_total_bytes(disk_super), factor);
2247 buf->f_blocks >>= bits;
2248 buf->f_bfree = buf->f_blocks - (div_u64(total_used, factor) >> bits);
2250 /* Account global block reserve as used, it's in logical size already */
2251 spin_lock(&block_rsv->lock);
2252 /* Mixed block groups accounting is not byte-accurate, avoid overflow */
2253 if (buf->f_bfree >= block_rsv->size >> bits)
2254 buf->f_bfree -= block_rsv->size >> bits;
2257 spin_unlock(&block_rsv->lock);
2259 buf->f_bavail = div_u64(total_free_data, factor);
2260 ret = btrfs_calc_avail_data_space(fs_info, &total_free_data);
2263 buf->f_bavail += div_u64(total_free_data, factor);
2264 buf->f_bavail = buf->f_bavail >> bits;
2267 * We calculate the remaining metadata space minus global reserve. If
2268 * this is (supposedly) smaller than zero, there's no space. But this
2269 * does not hold in practice, the exhausted state happens where's still
2270 * some positive delta. So we apply some guesswork and compare the
2271 * delta to a 4M threshold. (Practically observed delta was ~2M.)
2273 * We probably cannot calculate the exact threshold value because this
2274 * depends on the internal reservations requested by various
2275 * operations, so some operations that consume a few metadata will
2276 * succeed even if the Avail is zero. But this is better than the other
2282 * We only want to claim there's no available space if we can no longer
2283 * allocate chunks for our metadata profile and our global reserve will
2284 * not fit in the free metadata space. If we aren't ->full then we
2285 * still can allocate chunks and thus are fine using the currently
2286 * calculated f_bavail.
2288 if (!mixed && block_rsv->space_info->full &&
2289 total_free_meta - thresh < block_rsv->size)
2292 buf->f_type = BTRFS_SUPER_MAGIC;
2293 buf->f_bsize = dentry->d_sb->s_blocksize;
2294 buf->f_namelen = BTRFS_NAME_LEN;
2296 /* We treat it as constant endianness (it doesn't matter _which_)
2297 because we want the fsid to come out the same whether mounted
2298 on a big-endian or little-endian host */
2299 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
2300 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
2301 /* Mask in the root object ID too, to disambiguate subvols */
2302 buf->f_fsid.val[0] ^=
2303 BTRFS_I(d_inode(dentry))->root->root_key.objectid >> 32;
2304 buf->f_fsid.val[1] ^=
2305 BTRFS_I(d_inode(dentry))->root->root_key.objectid;
2310 static void btrfs_kill_super(struct super_block *sb)
2312 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2313 kill_anon_super(sb);
2314 btrfs_free_fs_info(fs_info);
2317 static struct file_system_type btrfs_fs_type = {
2318 .owner = THIS_MODULE,
2320 .mount = btrfs_mount,
2321 .kill_sb = btrfs_kill_super,
2322 .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA,
2325 static struct file_system_type btrfs_root_fs_type = {
2326 .owner = THIS_MODULE,
2328 .mount = btrfs_mount_root,
2329 .kill_sb = btrfs_kill_super,
2330 .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA,
2333 MODULE_ALIAS_FS("btrfs");
2335 static int btrfs_control_open(struct inode *inode, struct file *file)
2338 * The control file's private_data is used to hold the
2339 * transaction when it is started and is used to keep
2340 * track of whether a transaction is already in progress.
2342 file->private_data = NULL;
2347 * Used by /dev/btrfs-control for devices ioctls.
2349 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
2352 struct btrfs_ioctl_vol_args *vol;
2353 struct btrfs_device *device = NULL;
2356 if (!capable(CAP_SYS_ADMIN))
2359 vol = memdup_user((void __user *)arg, sizeof(*vol));
2361 return PTR_ERR(vol);
2362 vol->name[BTRFS_PATH_NAME_MAX] = '\0';
2365 case BTRFS_IOC_SCAN_DEV:
2366 mutex_lock(&uuid_mutex);
2367 device = btrfs_scan_one_device(vol->name, FMODE_READ,
2368 &btrfs_root_fs_type);
2369 ret = PTR_ERR_OR_ZERO(device);
2370 mutex_unlock(&uuid_mutex);
2372 case BTRFS_IOC_FORGET_DEV:
2373 ret = btrfs_forget_devices(vol->name);
2375 case BTRFS_IOC_DEVICES_READY:
2376 mutex_lock(&uuid_mutex);
2377 device = btrfs_scan_one_device(vol->name, FMODE_READ,
2378 &btrfs_root_fs_type);
2379 if (IS_ERR(device)) {
2380 mutex_unlock(&uuid_mutex);
2381 ret = PTR_ERR(device);
2384 ret = !(device->fs_devices->num_devices ==
2385 device->fs_devices->total_devices);
2386 mutex_unlock(&uuid_mutex);
2388 case BTRFS_IOC_GET_SUPPORTED_FEATURES:
2389 ret = btrfs_ioctl_get_supported_features((void __user*)arg);
2397 static int btrfs_freeze(struct super_block *sb)
2399 struct btrfs_trans_handle *trans;
2400 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2401 struct btrfs_root *root = fs_info->tree_root;
2403 set_bit(BTRFS_FS_FROZEN, &fs_info->flags);
2405 * We don't need a barrier here, we'll wait for any transaction that
2406 * could be in progress on other threads (and do delayed iputs that
2407 * we want to avoid on a frozen filesystem), or do the commit
2410 trans = btrfs_attach_transaction_barrier(root);
2411 if (IS_ERR(trans)) {
2412 /* no transaction, don't bother */
2413 if (PTR_ERR(trans) == -ENOENT)
2415 return PTR_ERR(trans);
2417 return btrfs_commit_transaction(trans);
2420 static int btrfs_unfreeze(struct super_block *sb)
2422 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2424 clear_bit(BTRFS_FS_FROZEN, &fs_info->flags);
2428 static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
2430 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
2431 struct btrfs_device *dev, *first_dev = NULL;
2434 * Lightweight locking of the devices. We should not need
2435 * device_list_mutex here as we only read the device data and the list
2436 * is protected by RCU. Even if a device is deleted during the list
2437 * traversals, we'll get valid data, the freeing callback will wait at
2438 * least until the rcu_read_unlock.
2441 list_for_each_entry_rcu(dev, &fs_info->fs_devices->devices, dev_list) {
2442 if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state))
2446 if (!first_dev || dev->devid < first_dev->devid)
2451 seq_escape(m, rcu_str_deref(first_dev->name), " \t\n\\");
2458 static const struct super_operations btrfs_super_ops = {
2459 .drop_inode = btrfs_drop_inode,
2460 .evict_inode = btrfs_evict_inode,
2461 .put_super = btrfs_put_super,
2462 .sync_fs = btrfs_sync_fs,
2463 .show_options = btrfs_show_options,
2464 .show_devname = btrfs_show_devname,
2465 .alloc_inode = btrfs_alloc_inode,
2466 .destroy_inode = btrfs_destroy_inode,
2467 .free_inode = btrfs_free_inode,
2468 .statfs = btrfs_statfs,
2469 .remount_fs = btrfs_remount,
2470 .freeze_fs = btrfs_freeze,
2471 .unfreeze_fs = btrfs_unfreeze,
2474 static const struct file_operations btrfs_ctl_fops = {
2475 .open = btrfs_control_open,
2476 .unlocked_ioctl = btrfs_control_ioctl,
2477 .compat_ioctl = compat_ptr_ioctl,
2478 .owner = THIS_MODULE,
2479 .llseek = noop_llseek,
2482 static struct miscdevice btrfs_misc = {
2483 .minor = BTRFS_MINOR,
2484 .name = "btrfs-control",
2485 .fops = &btrfs_ctl_fops
2488 MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
2489 MODULE_ALIAS("devname:btrfs-control");
2491 static int __init btrfs_interface_init(void)
2493 return misc_register(&btrfs_misc);
2496 static __cold void btrfs_interface_exit(void)
2498 misc_deregister(&btrfs_misc);
2501 static void __init btrfs_print_mod_info(void)
2503 static const char options[] = ""
2504 #ifdef CONFIG_BTRFS_DEBUG
2507 #ifdef CONFIG_BTRFS_ASSERT
2510 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
2511 ", integrity-checker=on"
2513 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
2516 #ifdef CONFIG_BLK_DEV_ZONED
2522 pr_info("Btrfs loaded, crc32c=%s%s\n", crc32c_impl(), options);
2525 static int __init init_btrfs_fs(void)
2531 err = btrfs_init_sysfs();
2535 btrfs_init_compress();
2537 err = btrfs_init_cachep();
2541 err = extent_io_init();
2545 err = extent_state_cache_init();
2547 goto free_extent_io;
2549 err = extent_map_init();
2551 goto free_extent_state_cache;
2553 err = ordered_data_init();
2555 goto free_extent_map;
2557 err = btrfs_delayed_inode_init();
2559 goto free_ordered_data;
2561 err = btrfs_auto_defrag_init();
2563 goto free_delayed_inode;
2565 err = btrfs_delayed_ref_init();
2567 goto free_auto_defrag;
2569 err = btrfs_prelim_ref_init();
2571 goto free_delayed_ref;
2573 err = btrfs_end_io_wq_init();
2575 goto free_prelim_ref;
2577 err = btrfs_interface_init();
2579 goto free_end_io_wq;
2581 btrfs_print_mod_info();
2583 err = btrfs_run_sanity_tests();
2585 goto unregister_ioctl;
2587 err = register_filesystem(&btrfs_fs_type);
2589 goto unregister_ioctl;
2594 btrfs_interface_exit();
2596 btrfs_end_io_wq_exit();
2598 btrfs_prelim_ref_exit();
2600 btrfs_delayed_ref_exit();
2602 btrfs_auto_defrag_exit();
2604 btrfs_delayed_inode_exit();
2606 ordered_data_exit();
2609 free_extent_state_cache:
2610 extent_state_cache_exit();
2614 btrfs_destroy_cachep();
2616 btrfs_exit_compress();
2622 static void __exit exit_btrfs_fs(void)
2624 btrfs_destroy_cachep();
2625 btrfs_delayed_ref_exit();
2626 btrfs_auto_defrag_exit();
2627 btrfs_delayed_inode_exit();
2628 btrfs_prelim_ref_exit();
2629 ordered_data_exit();
2631 extent_state_cache_exit();
2633 btrfs_interface_exit();
2634 btrfs_end_io_wq_exit();
2635 unregister_filesystem(&btrfs_fs_type);
2637 btrfs_cleanup_fs_uuids();
2638 btrfs_exit_compress();
2641 late_initcall(init_btrfs_fs);
2642 module_exit(exit_btrfs_fs)
2644 MODULE_LICENSE("GPL");
2645 MODULE_SOFTDEP("pre: crc32c");
2646 MODULE_SOFTDEP("pre: xxhash64");
2647 MODULE_SOFTDEP("pre: sha256");
2648 MODULE_SOFTDEP("pre: blake2b-256");