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>
45 #include "delayed-inode.h"
48 #include "transaction.h"
49 #include "btrfs_inode.h"
50 #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 #include "tests/btrfs-tests.h"
63 #define CREATE_TRACE_POINTS
64 #include <trace/events/btrfs.h>
66 static const struct super_operations btrfs_super_ops;
67 static struct file_system_type btrfs_fs_type;
69 static const char *btrfs_decode_error(int errno)
71 char *errstr = "unknown";
75 errstr = "IO failure";
78 errstr = "Out of memory";
81 errstr = "Readonly filesystem";
84 errstr = "Object already exists";
87 errstr = "No space left";
90 errstr = "No such entry";
97 static void save_error_info(struct btrfs_fs_info *fs_info)
100 * today we only save the error info into ram. Long term we'll
101 * also send it down to the disk
103 set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state);
106 /* btrfs handle error by forcing the filesystem readonly */
107 static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
109 struct super_block *sb = fs_info->sb;
111 if (sb->s_flags & MS_RDONLY)
114 if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
115 sb->s_flags |= MS_RDONLY;
116 btrfs_info(fs_info, "forced readonly");
118 * Note that a running device replace operation is not
119 * canceled here although there is no way to update
120 * the progress. It would add the risk of a deadlock,
121 * therefore the canceling is ommited. The only penalty
122 * is that some I/O remains active until the procedure
123 * completes. The next time when the filesystem is
124 * mounted writeable again, the device replace
125 * operation continues.
132 * __btrfs_std_error decodes expected errors from the caller and
133 * invokes the approciate error response.
135 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
136 unsigned int line, int errno, const char *fmt, ...)
138 struct super_block *sb = fs_info->sb;
142 * Special case: if the error is EROFS, and we're already
143 * under MS_RDONLY, then it is safe here.
145 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
148 errstr = btrfs_decode_error(errno);
150 struct va_format vaf;
158 "BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n",
159 sb->s_id, function, line, errno, errstr, &vaf);
162 printk(KERN_CRIT "BTRFS: error (device %s) in %s:%d: errno=%d %s\n",
163 sb->s_id, function, line, errno, errstr);
166 /* Don't go through full error handling during mount */
167 save_error_info(fs_info);
168 if (sb->s_flags & MS_BORN)
169 btrfs_handle_error(fs_info);
172 static const char * const logtypes[] = {
183 void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
185 struct super_block *sb = fs_info->sb;
187 struct va_format vaf;
189 const char *type = logtypes[4];
194 kern_level = printk_get_level(fmt);
196 size_t size = printk_skip_level(fmt) - fmt;
197 memcpy(lvl, fmt, size);
200 type = logtypes[kern_level - '0'];
207 printk("%sBTRFS %s (device %s): %pV\n", lvl, type, sb->s_id, &vaf);
214 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
215 unsigned int line, int errno, const char *fmt, ...)
217 struct super_block *sb = fs_info->sb;
220 * Special case: if the error is EROFS, and we're already
221 * under MS_RDONLY, then it is safe here.
223 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
226 /* Don't go through full error handling during mount */
227 if (sb->s_flags & MS_BORN) {
228 save_error_info(fs_info);
229 btrfs_handle_error(fs_info);
235 * We only mark the transaction aborted and then set the file system read-only.
236 * This will prevent new transactions from starting or trying to join this
239 * This means that error recovery at the call site is limited to freeing
240 * any local memory allocations and passing the error code up without
241 * further cleanup. The transaction should complete as it normally would
242 * in the call path but will return -EIO.
244 * We'll complete the cleanup in btrfs_end_transaction and
245 * btrfs_commit_transaction.
247 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
248 struct btrfs_root *root, const char *function,
249 unsigned int line, int errno)
252 * Report first abort since mount
254 if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED,
255 &root->fs_info->fs_state)) {
256 WARN(1, KERN_DEBUG "BTRFS: Transaction aborted (error %d)\n",
259 trans->aborted = errno;
260 /* Nothing used. The other threads that have joined this
261 * transaction may be able to continue. */
262 if (!trans->blocks_used) {
265 errstr = btrfs_decode_error(errno);
266 btrfs_warn(root->fs_info,
267 "%s:%d: Aborting unused transaction(%s).",
268 function, line, errstr);
271 ACCESS_ONCE(trans->transaction->aborted) = errno;
272 /* Wake up anybody who may be waiting on this transaction */
273 wake_up(&root->fs_info->transaction_wait);
274 wake_up(&root->fs_info->transaction_blocked_wait);
275 __btrfs_std_error(root->fs_info, function, line, errno, NULL);
278 * __btrfs_panic decodes unexpected, fatal errors from the caller,
279 * issues an alert, and either panics or BUGs, depending on mount options.
281 void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
282 unsigned int line, int errno, const char *fmt, ...)
284 char *s_id = "<unknown>";
286 struct va_format vaf = { .fmt = fmt };
290 s_id = fs_info->sb->s_id;
295 errstr = btrfs_decode_error(errno);
296 if (fs_info && (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR))
297 panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
298 s_id, function, line, &vaf, errno, errstr);
300 btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)",
301 function, line, &vaf, errno, errstr);
303 /* Caller calls BUG() */
306 static void btrfs_put_super(struct super_block *sb)
308 (void)close_ctree(btrfs_sb(sb)->tree_root);
309 /* FIXME: need to fix VFS to return error? */
310 /* AV: return it _where_? ->put_super() can be triggered by any number
311 * of async events, up to and including delivery of SIGKILL to the
312 * last process that kept it busy. Or segfault in the aforementioned
313 * process... Whom would you report that to?
318 Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum,
319 Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd,
320 Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress,
321 Opt_compress_type, Opt_compress_force, Opt_compress_force_type,
322 Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_discard,
323 Opt_space_cache, Opt_clear_cache, Opt_user_subvol_rm_allowed,
324 Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache,
325 Opt_no_space_cache, Opt_recovery, Opt_skip_balance,
326 Opt_check_integrity, Opt_check_integrity_including_extent_data,
327 Opt_check_integrity_print_mask, Opt_fatal_errors, Opt_rescan_uuid_tree,
328 Opt_commit_interval, Opt_barrier, Opt_nodefrag, Opt_nodiscard,
329 Opt_noenospc_debug, Opt_noflushoncommit, Opt_acl, Opt_datacow,
330 Opt_datasum, Opt_treelog, Opt_noinode_cache,
334 static match_table_t tokens = {
335 {Opt_degraded, "degraded"},
336 {Opt_subvol, "subvol=%s"},
337 {Opt_subvolid, "subvolid=%s"},
338 {Opt_device, "device=%s"},
339 {Opt_nodatasum, "nodatasum"},
340 {Opt_datasum, "datasum"},
341 {Opt_nodatacow, "nodatacow"},
342 {Opt_datacow, "datacow"},
343 {Opt_nobarrier, "nobarrier"},
344 {Opt_barrier, "barrier"},
345 {Opt_max_inline, "max_inline=%s"},
346 {Opt_alloc_start, "alloc_start=%s"},
347 {Opt_thread_pool, "thread_pool=%d"},
348 {Opt_compress, "compress"},
349 {Opt_compress_type, "compress=%s"},
350 {Opt_compress_force, "compress-force"},
351 {Opt_compress_force_type, "compress-force=%s"},
353 {Opt_ssd_spread, "ssd_spread"},
354 {Opt_nossd, "nossd"},
356 {Opt_noacl, "noacl"},
357 {Opt_notreelog, "notreelog"},
358 {Opt_treelog, "treelog"},
359 {Opt_flushoncommit, "flushoncommit"},
360 {Opt_noflushoncommit, "noflushoncommit"},
361 {Opt_ratio, "metadata_ratio=%d"},
362 {Opt_discard, "discard"},
363 {Opt_nodiscard, "nodiscard"},
364 {Opt_space_cache, "space_cache"},
365 {Opt_clear_cache, "clear_cache"},
366 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
367 {Opt_enospc_debug, "enospc_debug"},
368 {Opt_noenospc_debug, "noenospc_debug"},
369 {Opt_subvolrootid, "subvolrootid=%d"},
370 {Opt_defrag, "autodefrag"},
371 {Opt_nodefrag, "noautodefrag"},
372 {Opt_inode_cache, "inode_cache"},
373 {Opt_noinode_cache, "noinode_cache"},
374 {Opt_no_space_cache, "nospace_cache"},
375 {Opt_recovery, "recovery"},
376 {Opt_skip_balance, "skip_balance"},
377 {Opt_check_integrity, "check_int"},
378 {Opt_check_integrity_including_extent_data, "check_int_data"},
379 {Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
380 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
381 {Opt_fatal_errors, "fatal_errors=%s"},
382 {Opt_commit_interval, "commit=%d"},
387 * Regular mount options parser. Everything that is needed only when
388 * reading in a new superblock is parsed here.
389 * XXX JDM: This needs to be cleaned up for remount.
391 int btrfs_parse_options(struct btrfs_root *root, char *options)
393 struct btrfs_fs_info *info = root->fs_info;
394 substring_t args[MAX_OPT_ARGS];
395 char *p, *num, *orig = NULL;
400 bool compress_force = false;
402 cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
404 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
410 * strsep changes the string, duplicate it because parse_options
413 options = kstrdup(options, GFP_NOFS);
419 while ((p = strsep(&options, ",")) != NULL) {
424 token = match_token(p, tokens, args);
427 btrfs_info(root->fs_info, "allowing degraded mounts");
428 btrfs_set_opt(info->mount_opt, DEGRADED);
432 case Opt_subvolrootid:
435 * These are parsed by btrfs_parse_early_options
436 * and can be happily ignored here.
440 btrfs_info(root->fs_info, "setting nodatasum");
441 btrfs_set_opt(info->mount_opt, NODATASUM);
444 if (btrfs_test_opt(root, NODATACOW))
445 btrfs_info(root->fs_info, "setting datasum, datacow enabled");
447 btrfs_info(root->fs_info, "setting datasum");
448 btrfs_clear_opt(info->mount_opt, NODATACOW);
449 btrfs_clear_opt(info->mount_opt, NODATASUM);
452 if (!btrfs_test_opt(root, COMPRESS) ||
453 !btrfs_test_opt(root, FORCE_COMPRESS)) {
454 btrfs_info(root->fs_info,
455 "setting nodatacow, compression disabled");
457 btrfs_info(root->fs_info, "setting nodatacow");
459 btrfs_clear_opt(info->mount_opt, COMPRESS);
460 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
461 btrfs_set_opt(info->mount_opt, NODATACOW);
462 btrfs_set_opt(info->mount_opt, NODATASUM);
465 if (btrfs_test_opt(root, NODATACOW))
466 btrfs_info(root->fs_info, "setting datacow");
467 btrfs_clear_opt(info->mount_opt, NODATACOW);
469 case Opt_compress_force:
470 case Opt_compress_force_type:
471 compress_force = true;
474 case Opt_compress_type:
475 if (token == Opt_compress ||
476 token == Opt_compress_force ||
477 strcmp(args[0].from, "zlib") == 0) {
478 compress_type = "zlib";
479 info->compress_type = BTRFS_COMPRESS_ZLIB;
480 btrfs_set_opt(info->mount_opt, COMPRESS);
481 btrfs_clear_opt(info->mount_opt, NODATACOW);
482 btrfs_clear_opt(info->mount_opt, NODATASUM);
483 } else if (strcmp(args[0].from, "lzo") == 0) {
484 compress_type = "lzo";
485 info->compress_type = BTRFS_COMPRESS_LZO;
486 btrfs_set_opt(info->mount_opt, COMPRESS);
487 btrfs_clear_opt(info->mount_opt, NODATACOW);
488 btrfs_clear_opt(info->mount_opt, NODATASUM);
489 btrfs_set_fs_incompat(info, COMPRESS_LZO);
490 } else if (strncmp(args[0].from, "no", 2) == 0) {
491 compress_type = "no";
492 btrfs_clear_opt(info->mount_opt, COMPRESS);
493 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
494 compress_force = false;
500 if (compress_force) {
501 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
502 btrfs_info(root->fs_info, "force %s compression",
504 } else if (btrfs_test_opt(root, COMPRESS)) {
505 pr_info("btrfs: use %s compression\n",
510 btrfs_info(root->fs_info, "use ssd allocation scheme");
511 btrfs_set_opt(info->mount_opt, SSD);
514 btrfs_info(root->fs_info, "use spread ssd allocation scheme");
515 btrfs_set_opt(info->mount_opt, SSD);
516 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
519 btrfs_info(root->fs_info, "not using ssd allocation scheme");
520 btrfs_set_opt(info->mount_opt, NOSSD);
521 btrfs_clear_opt(info->mount_opt, SSD);
522 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
525 if (btrfs_test_opt(root, NOBARRIER))
526 btrfs_info(root->fs_info, "turning on barriers");
527 btrfs_clear_opt(info->mount_opt, NOBARRIER);
530 btrfs_info(root->fs_info, "turning off barriers");
531 btrfs_set_opt(info->mount_opt, NOBARRIER);
533 case Opt_thread_pool:
534 ret = match_int(&args[0], &intarg);
537 } else if (intarg > 0) {
538 info->thread_pool_size = intarg;
545 num = match_strdup(&args[0]);
547 info->max_inline = memparse(num, NULL);
550 if (info->max_inline) {
551 info->max_inline = max_t(u64,
555 btrfs_info(root->fs_info, "max_inline at %llu",
562 case Opt_alloc_start:
563 num = match_strdup(&args[0]);
565 mutex_lock(&info->chunk_mutex);
566 info->alloc_start = memparse(num, NULL);
567 mutex_unlock(&info->chunk_mutex);
569 btrfs_info(root->fs_info, "allocations start at %llu",
577 root->fs_info->sb->s_flags |= MS_POSIXACL;
580 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
583 btrfs_info(root->fs_info, "disabling tree log");
584 btrfs_set_opt(info->mount_opt, NOTREELOG);
587 if (btrfs_test_opt(root, NOTREELOG))
588 btrfs_info(root->fs_info, "enabling tree log");
589 btrfs_clear_opt(info->mount_opt, NOTREELOG);
591 case Opt_flushoncommit:
592 btrfs_info(root->fs_info, "turning on flush-on-commit");
593 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
595 case Opt_noflushoncommit:
596 if (btrfs_test_opt(root, FLUSHONCOMMIT))
597 btrfs_info(root->fs_info, "turning off flush-on-commit");
598 btrfs_clear_opt(info->mount_opt, FLUSHONCOMMIT);
601 ret = match_int(&args[0], &intarg);
604 } else if (intarg >= 0) {
605 info->metadata_ratio = intarg;
606 btrfs_info(root->fs_info, "metadata ratio %d",
607 info->metadata_ratio);
614 btrfs_set_opt(info->mount_opt, DISCARD);
617 btrfs_clear_opt(info->mount_opt, DISCARD);
619 case Opt_space_cache:
620 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
622 case Opt_rescan_uuid_tree:
623 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
625 case Opt_no_space_cache:
626 btrfs_info(root->fs_info, "disabling disk space caching");
627 btrfs_clear_opt(info->mount_opt, SPACE_CACHE);
629 case Opt_inode_cache:
630 btrfs_info(root->fs_info, "enabling inode map caching");
631 btrfs_set_opt(info->mount_opt, CHANGE_INODE_CACHE);
633 case Opt_noinode_cache:
634 if (btrfs_test_opt(root, CHANGE_INODE_CACHE))
635 btrfs_info(root->fs_info, "disabling inode map caching");
636 btrfs_clear_opt(info->mount_opt, CHANGE_INODE_CACHE);
638 case Opt_clear_cache:
639 btrfs_info(root->fs_info, "force clearing of disk cache");
640 btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
642 case Opt_user_subvol_rm_allowed:
643 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
645 case Opt_enospc_debug:
646 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
648 case Opt_noenospc_debug:
649 btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG);
652 btrfs_info(root->fs_info, "enabling auto defrag");
653 btrfs_set_opt(info->mount_opt, AUTO_DEFRAG);
656 if (btrfs_test_opt(root, AUTO_DEFRAG))
657 btrfs_info(root->fs_info, "disabling auto defrag");
658 btrfs_clear_opt(info->mount_opt, AUTO_DEFRAG);
661 btrfs_info(root->fs_info, "enabling auto recovery");
662 btrfs_set_opt(info->mount_opt, RECOVERY);
664 case Opt_skip_balance:
665 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
667 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
668 case Opt_check_integrity_including_extent_data:
669 btrfs_info(root->fs_info,
670 "enabling check integrity including extent data");
671 btrfs_set_opt(info->mount_opt,
672 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
673 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
675 case Opt_check_integrity:
676 btrfs_info(root->fs_info, "enabling check integrity");
677 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
679 case Opt_check_integrity_print_mask:
680 ret = match_int(&args[0], &intarg);
683 } else if (intarg >= 0) {
684 info->check_integrity_print_mask = intarg;
685 btrfs_info(root->fs_info, "check_integrity_print_mask 0x%x",
686 info->check_integrity_print_mask);
693 case Opt_check_integrity_including_extent_data:
694 case Opt_check_integrity:
695 case Opt_check_integrity_print_mask:
696 btrfs_err(root->fs_info,
697 "support for check_integrity* not compiled in!");
701 case Opt_fatal_errors:
702 if (strcmp(args[0].from, "panic") == 0)
703 btrfs_set_opt(info->mount_opt,
704 PANIC_ON_FATAL_ERROR);
705 else if (strcmp(args[0].from, "bug") == 0)
706 btrfs_clear_opt(info->mount_opt,
707 PANIC_ON_FATAL_ERROR);
713 case Opt_commit_interval:
715 ret = match_int(&args[0], &intarg);
717 btrfs_err(root->fs_info, "invalid commit interval");
723 btrfs_warn(root->fs_info, "excessive commit interval %d",
726 info->commit_interval = intarg;
728 btrfs_info(root->fs_info, "using default commit interval %ds",
729 BTRFS_DEFAULT_COMMIT_INTERVAL);
730 info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
734 btrfs_info(root->fs_info, "unrecognized mount option '%s'", p);
742 if (!ret && btrfs_test_opt(root, SPACE_CACHE))
743 btrfs_info(root->fs_info, "disk space caching is enabled");
749 * Parse mount options that are required early in the mount process.
751 * All other options will be parsed on much later in the mount process and
752 * only when we need to allocate a new super block.
754 static int btrfs_parse_early_options(const char *options, fmode_t flags,
755 void *holder, char **subvol_name, u64 *subvol_objectid,
756 struct btrfs_fs_devices **fs_devices)
758 substring_t args[MAX_OPT_ARGS];
759 char *device_name, *opts, *orig, *p;
767 * strsep changes the string, duplicate it because parse_options
770 opts = kstrdup(options, GFP_KERNEL);
775 while ((p = strsep(&opts, ",")) != NULL) {
780 token = match_token(p, tokens, args);
784 *subvol_name = match_strdup(&args[0]);
791 num = match_strdup(&args[0]);
793 *subvol_objectid = memparse(num, NULL);
795 /* we want the original fs_tree */
796 if (!*subvol_objectid)
798 BTRFS_FS_TREE_OBJECTID;
804 case Opt_subvolrootid:
806 "BTRFS: 'subvolrootid' mount option is deprecated and has "
810 device_name = match_strdup(&args[0]);
815 error = btrfs_scan_one_device(device_name,
816 flags, holder, fs_devices);
831 static struct dentry *get_default_root(struct super_block *sb,
834 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
835 struct btrfs_root *root = fs_info->tree_root;
836 struct btrfs_root *new_root;
837 struct btrfs_dir_item *di;
838 struct btrfs_path *path;
839 struct btrfs_key location;
845 * We have a specific subvol we want to mount, just setup location and
846 * go look up the root.
848 if (subvol_objectid) {
849 location.objectid = subvol_objectid;
850 location.type = BTRFS_ROOT_ITEM_KEY;
851 location.offset = (u64)-1;
855 path = btrfs_alloc_path();
857 return ERR_PTR(-ENOMEM);
858 path->leave_spinning = 1;
861 * Find the "default" dir item which points to the root item that we
862 * will mount by default if we haven't been given a specific subvolume
865 dir_id = btrfs_super_root_dir(fs_info->super_copy);
866 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
868 btrfs_free_path(path);
873 * Ok the default dir item isn't there. This is weird since
874 * it's always been there, but don't freak out, just try and
875 * mount to root most subvolume.
877 btrfs_free_path(path);
878 dir_id = BTRFS_FIRST_FREE_OBJECTID;
879 new_root = fs_info->fs_root;
883 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
884 btrfs_free_path(path);
887 new_root = btrfs_read_fs_root_no_name(fs_info, &location);
888 if (IS_ERR(new_root))
889 return ERR_CAST(new_root);
891 dir_id = btrfs_root_dirid(&new_root->root_item);
893 location.objectid = dir_id;
894 location.type = BTRFS_INODE_ITEM_KEY;
897 inode = btrfs_iget(sb, &location, new_root, &new);
899 return ERR_CAST(inode);
902 * If we're just mounting the root most subvol put the inode and return
903 * a reference to the dentry. We will have already gotten a reference
904 * to the inode in btrfs_fill_super so we're good to go.
906 if (!new && sb->s_root->d_inode == inode) {
908 return dget(sb->s_root);
911 return d_obtain_alias(inode);
914 static int btrfs_fill_super(struct super_block *sb,
915 struct btrfs_fs_devices *fs_devices,
916 void *data, int silent)
919 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
920 struct btrfs_key key;
923 sb->s_maxbytes = MAX_LFS_FILESIZE;
924 sb->s_magic = BTRFS_SUPER_MAGIC;
925 sb->s_op = &btrfs_super_ops;
926 sb->s_d_op = &btrfs_dentry_operations;
927 sb->s_export_op = &btrfs_export_ops;
928 sb->s_xattr = btrfs_xattr_handlers;
930 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
931 sb->s_flags |= MS_POSIXACL;
933 sb->s_flags |= MS_I_VERSION;
934 err = open_ctree(sb, fs_devices, (char *)data);
936 printk(KERN_ERR "BTRFS: open_ctree failed\n");
940 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
941 key.type = BTRFS_INODE_ITEM_KEY;
943 inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL);
945 err = PTR_ERR(inode);
949 sb->s_root = d_make_root(inode);
955 save_mount_options(sb, data);
956 cleancache_init_fs(sb);
957 sb->s_flags |= MS_ACTIVE;
961 close_ctree(fs_info->tree_root);
965 int btrfs_sync_fs(struct super_block *sb, int wait)
967 struct btrfs_trans_handle *trans;
968 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
969 struct btrfs_root *root = fs_info->tree_root;
971 trace_btrfs_sync_fs(wait);
974 filemap_flush(fs_info->btree_inode->i_mapping);
978 btrfs_wait_ordered_roots(fs_info, -1);
980 trans = btrfs_attach_transaction_barrier(root);
982 /* no transaction, don't bother */
983 if (PTR_ERR(trans) == -ENOENT)
985 return PTR_ERR(trans);
987 return btrfs_commit_transaction(trans, root);
990 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
992 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
993 struct btrfs_root *root = info->tree_root;
996 if (btrfs_test_opt(root, DEGRADED))
997 seq_puts(seq, ",degraded");
998 if (btrfs_test_opt(root, NODATASUM))
999 seq_puts(seq, ",nodatasum");
1000 if (btrfs_test_opt(root, NODATACOW))
1001 seq_puts(seq, ",nodatacow");
1002 if (btrfs_test_opt(root, NOBARRIER))
1003 seq_puts(seq, ",nobarrier");
1004 if (info->max_inline != 8192 * 1024)
1005 seq_printf(seq, ",max_inline=%llu", info->max_inline);
1006 if (info->alloc_start != 0)
1007 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
1008 if (info->thread_pool_size != min_t(unsigned long,
1009 num_online_cpus() + 2, 8))
1010 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
1011 if (btrfs_test_opt(root, COMPRESS)) {
1012 if (info->compress_type == BTRFS_COMPRESS_ZLIB)
1013 compress_type = "zlib";
1015 compress_type = "lzo";
1016 if (btrfs_test_opt(root, FORCE_COMPRESS))
1017 seq_printf(seq, ",compress-force=%s", compress_type);
1019 seq_printf(seq, ",compress=%s", compress_type);
1021 if (btrfs_test_opt(root, NOSSD))
1022 seq_puts(seq, ",nossd");
1023 if (btrfs_test_opt(root, SSD_SPREAD))
1024 seq_puts(seq, ",ssd_spread");
1025 else if (btrfs_test_opt(root, SSD))
1026 seq_puts(seq, ",ssd");
1027 if (btrfs_test_opt(root, NOTREELOG))
1028 seq_puts(seq, ",notreelog");
1029 if (btrfs_test_opt(root, FLUSHONCOMMIT))
1030 seq_puts(seq, ",flushoncommit");
1031 if (btrfs_test_opt(root, DISCARD))
1032 seq_puts(seq, ",discard");
1033 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
1034 seq_puts(seq, ",noacl");
1035 if (btrfs_test_opt(root, SPACE_CACHE))
1036 seq_puts(seq, ",space_cache");
1038 seq_puts(seq, ",nospace_cache");
1039 if (btrfs_test_opt(root, RESCAN_UUID_TREE))
1040 seq_puts(seq, ",rescan_uuid_tree");
1041 if (btrfs_test_opt(root, CLEAR_CACHE))
1042 seq_puts(seq, ",clear_cache");
1043 if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
1044 seq_puts(seq, ",user_subvol_rm_allowed");
1045 if (btrfs_test_opt(root, ENOSPC_DEBUG))
1046 seq_puts(seq, ",enospc_debug");
1047 if (btrfs_test_opt(root, AUTO_DEFRAG))
1048 seq_puts(seq, ",autodefrag");
1049 if (btrfs_test_opt(root, INODE_MAP_CACHE))
1050 seq_puts(seq, ",inode_cache");
1051 if (btrfs_test_opt(root, SKIP_BALANCE))
1052 seq_puts(seq, ",skip_balance");
1053 if (btrfs_test_opt(root, RECOVERY))
1054 seq_puts(seq, ",recovery");
1055 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1056 if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1057 seq_puts(seq, ",check_int_data");
1058 else if (btrfs_test_opt(root, CHECK_INTEGRITY))
1059 seq_puts(seq, ",check_int");
1060 if (info->check_integrity_print_mask)
1061 seq_printf(seq, ",check_int_print_mask=%d",
1062 info->check_integrity_print_mask);
1064 if (info->metadata_ratio)
1065 seq_printf(seq, ",metadata_ratio=%d",
1066 info->metadata_ratio);
1067 if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
1068 seq_puts(seq, ",fatal_errors=panic");
1069 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1070 seq_printf(seq, ",commit=%d", info->commit_interval);
1074 static int btrfs_test_super(struct super_block *s, void *data)
1076 struct btrfs_fs_info *p = data;
1077 struct btrfs_fs_info *fs_info = btrfs_sb(s);
1079 return fs_info->fs_devices == p->fs_devices;
1082 static int btrfs_set_super(struct super_block *s, void *data)
1084 int err = set_anon_super(s, data);
1086 s->s_fs_info = data;
1091 * subvolumes are identified by ino 256
1093 static inline int is_subvolume_inode(struct inode *inode)
1095 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1101 * This will strip out the subvol=%s argument for an argument string and add
1102 * subvolid=0 to make sure we get the actual tree root for path walking to the
1105 static char *setup_root_args(char *args)
1107 unsigned len = strlen(args) + 2 + 1;
1108 char *src, *dst, *buf;
1111 * We need the same args as before, but with this substitution:
1112 * s!subvol=[^,]+!subvolid=0!
1114 * Since the replacement string is up to 2 bytes longer than the
1115 * original, allocate strlen(args) + 2 + 1 bytes.
1118 src = strstr(args, "subvol=");
1119 /* This shouldn't happen, but just in case.. */
1123 buf = dst = kmalloc(len, GFP_NOFS);
1128 * If the subvol= arg is not at the start of the string,
1129 * copy whatever precedes it into buf.
1134 dst += strlen(args);
1137 strcpy(dst, "subvolid=0");
1138 dst += strlen("subvolid=0");
1141 * If there is a "," after the original subvol=... string,
1142 * copy that suffix into our buffer. Otherwise, we're done.
1144 src = strchr(src, ',');
1151 static struct dentry *mount_subvol(const char *subvol_name, int flags,
1152 const char *device_name, char *data)
1154 struct dentry *root;
1155 struct vfsmount *mnt;
1158 newargs = setup_root_args(data);
1160 return ERR_PTR(-ENOMEM);
1161 mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name,
1165 return ERR_CAST(mnt);
1167 root = mount_subtree(mnt, subvol_name);
1169 if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) {
1170 struct super_block *s = root->d_sb;
1172 root = ERR_PTR(-EINVAL);
1173 deactivate_locked_super(s);
1174 printk(KERN_ERR "BTRFS: '%s' is not a valid subvolume\n",
1182 * Find a superblock for the given device / mount point.
1184 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1185 * for multiple device setup. Make sure to keep it in sync.
1187 static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
1188 const char *device_name, void *data)
1190 struct block_device *bdev = NULL;
1191 struct super_block *s;
1192 struct dentry *root;
1193 struct btrfs_fs_devices *fs_devices = NULL;
1194 struct btrfs_fs_info *fs_info = NULL;
1195 fmode_t mode = FMODE_READ;
1196 char *subvol_name = NULL;
1197 u64 subvol_objectid = 0;
1200 if (!(flags & MS_RDONLY))
1201 mode |= FMODE_WRITE;
1203 error = btrfs_parse_early_options(data, mode, fs_type,
1204 &subvol_name, &subvol_objectid,
1208 return ERR_PTR(error);
1212 root = mount_subvol(subvol_name, flags, device_name, data);
1217 error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices);
1219 return ERR_PTR(error);
1222 * Setup a dummy root and fs_info for test/set super. This is because
1223 * we don't actually fill this stuff out until open_ctree, but we need
1224 * it for searching for existing supers, so this lets us do that and
1225 * then open_ctree will properly initialize everything later.
1227 fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
1229 return ERR_PTR(-ENOMEM);
1231 fs_info->fs_devices = fs_devices;
1233 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1234 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1235 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1240 error = btrfs_open_devices(fs_devices, mode, fs_type);
1244 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
1246 goto error_close_devices;
1249 bdev = fs_devices->latest_bdev;
1250 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC,
1254 goto error_close_devices;
1258 btrfs_close_devices(fs_devices);
1259 free_fs_info(fs_info);
1260 if ((flags ^ s->s_flags) & MS_RDONLY)
1263 char b[BDEVNAME_SIZE];
1265 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
1266 btrfs_sb(s)->bdev_holder = fs_type;
1267 error = btrfs_fill_super(s, fs_devices, data,
1268 flags & MS_SILENT ? 1 : 0);
1271 root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error);
1273 deactivate_locked_super(s);
1277 error_close_devices:
1278 btrfs_close_devices(fs_devices);
1280 free_fs_info(fs_info);
1281 return ERR_PTR(error);
1284 static void btrfs_set_max_workers(struct btrfs_workers *workers, int new_limit)
1286 spin_lock_irq(&workers->lock);
1287 workers->max_workers = new_limit;
1288 spin_unlock_irq(&workers->lock);
1291 static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1292 int new_pool_size, int old_pool_size)
1294 if (new_pool_size == old_pool_size)
1297 fs_info->thread_pool_size = new_pool_size;
1299 btrfs_info(fs_info, "resize thread pool %d -> %d",
1300 old_pool_size, new_pool_size);
1302 btrfs_set_max_workers(&fs_info->generic_worker, new_pool_size);
1303 btrfs_set_max_workers(&fs_info->workers, new_pool_size);
1304 btrfs_set_max_workers(&fs_info->delalloc_workers, new_pool_size);
1305 btrfs_set_max_workers(&fs_info->submit_workers, new_pool_size);
1306 btrfs_set_max_workers(&fs_info->caching_workers, new_pool_size);
1307 btrfs_set_max_workers(&fs_info->fixup_workers, new_pool_size);
1308 btrfs_set_max_workers(&fs_info->endio_workers, new_pool_size);
1309 btrfs_set_max_workers(&fs_info->endio_meta_workers, new_pool_size);
1310 btrfs_set_max_workers(&fs_info->endio_meta_write_workers, new_pool_size);
1311 btrfs_set_max_workers(&fs_info->endio_write_workers, new_pool_size);
1312 btrfs_set_max_workers(&fs_info->endio_freespace_worker, new_pool_size);
1313 btrfs_set_max_workers(&fs_info->delayed_workers, new_pool_size);
1314 btrfs_set_max_workers(&fs_info->readahead_workers, new_pool_size);
1315 btrfs_set_max_workers(&fs_info->scrub_wr_completion_workers,
1319 static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
1321 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1324 static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1325 unsigned long old_opts, int flags)
1327 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1328 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1329 (flags & MS_RDONLY))) {
1330 /* wait for any defraggers to finish */
1331 wait_event(fs_info->transaction_wait,
1332 (atomic_read(&fs_info->defrag_running) == 0));
1333 if (flags & MS_RDONLY)
1334 sync_filesystem(fs_info->sb);
1338 static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1339 unsigned long old_opts)
1342 * We need cleanup all defragable inodes if the autodefragment is
1343 * close or the fs is R/O.
1345 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1346 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1347 (fs_info->sb->s_flags & MS_RDONLY))) {
1348 btrfs_cleanup_defrag_inodes(fs_info);
1351 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1354 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1356 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1357 struct btrfs_root *root = fs_info->tree_root;
1358 unsigned old_flags = sb->s_flags;
1359 unsigned long old_opts = fs_info->mount_opt;
1360 unsigned long old_compress_type = fs_info->compress_type;
1361 u64 old_max_inline = fs_info->max_inline;
1362 u64 old_alloc_start = fs_info->alloc_start;
1363 int old_thread_pool_size = fs_info->thread_pool_size;
1364 unsigned int old_metadata_ratio = fs_info->metadata_ratio;
1367 btrfs_remount_prepare(fs_info);
1369 ret = btrfs_parse_options(root, data);
1375 btrfs_remount_begin(fs_info, old_opts, *flags);
1376 btrfs_resize_thread_pool(fs_info,
1377 fs_info->thread_pool_size, old_thread_pool_size);
1379 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
1382 if (*flags & MS_RDONLY) {
1384 * this also happens on 'umount -rf' or on shutdown, when
1385 * the filesystem is busy.
1388 /* wait for the uuid_scan task to finish */
1389 down(&fs_info->uuid_tree_rescan_sem);
1390 /* avoid complains from lockdep et al. */
1391 up(&fs_info->uuid_tree_rescan_sem);
1393 sb->s_flags |= MS_RDONLY;
1395 btrfs_dev_replace_suspend_for_unmount(fs_info);
1396 btrfs_scrub_cancel(fs_info);
1397 btrfs_pause_balance(fs_info);
1399 ret = btrfs_commit_super(root);
1403 if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) {
1405 "Remounting read-write after error is not allowed");
1409 if (fs_info->fs_devices->rw_devices == 0) {
1414 if (fs_info->fs_devices->missing_devices >
1415 fs_info->num_tolerated_disk_barrier_failures &&
1416 !(*flags & MS_RDONLY)) {
1418 "too many missing devices, writeable remount is not allowed");
1423 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
1428 ret = btrfs_cleanup_fs_roots(fs_info);
1432 /* recover relocation */
1433 ret = btrfs_recover_relocation(root);
1437 ret = btrfs_resume_balance_async(fs_info);
1441 ret = btrfs_resume_dev_replace_async(fs_info);
1443 btrfs_warn(fs_info, "failed to resume dev_replace");
1447 if (!fs_info->uuid_root) {
1448 btrfs_info(fs_info, "creating UUID tree");
1449 ret = btrfs_create_uuid_tree(fs_info);
1451 btrfs_warn(fs_info, "failed to create the UUID tree %d", ret);
1455 sb->s_flags &= ~MS_RDONLY;
1458 btrfs_remount_cleanup(fs_info, old_opts);
1462 /* We've hit an error - don't reset MS_RDONLY */
1463 if (sb->s_flags & MS_RDONLY)
1464 old_flags |= MS_RDONLY;
1465 sb->s_flags = old_flags;
1466 fs_info->mount_opt = old_opts;
1467 fs_info->compress_type = old_compress_type;
1468 fs_info->max_inline = old_max_inline;
1469 mutex_lock(&fs_info->chunk_mutex);
1470 fs_info->alloc_start = old_alloc_start;
1471 mutex_unlock(&fs_info->chunk_mutex);
1472 btrfs_resize_thread_pool(fs_info,
1473 old_thread_pool_size, fs_info->thread_pool_size);
1474 fs_info->metadata_ratio = old_metadata_ratio;
1475 btrfs_remount_cleanup(fs_info, old_opts);
1479 /* Used to sort the devices by max_avail(descending sort) */
1480 static int btrfs_cmp_device_free_bytes(const void *dev_info1,
1481 const void *dev_info2)
1483 if (((struct btrfs_device_info *)dev_info1)->max_avail >
1484 ((struct btrfs_device_info *)dev_info2)->max_avail)
1486 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
1487 ((struct btrfs_device_info *)dev_info2)->max_avail)
1494 * sort the devices by max_avail, in which max free extent size of each device
1495 * is stored.(Descending Sort)
1497 static inline void btrfs_descending_sort_devices(
1498 struct btrfs_device_info *devices,
1501 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
1502 btrfs_cmp_device_free_bytes, NULL);
1506 * The helper to calc the free space on the devices that can be used to store
1509 static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
1511 struct btrfs_fs_info *fs_info = root->fs_info;
1512 struct btrfs_device_info *devices_info;
1513 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
1514 struct btrfs_device *device;
1519 u64 min_stripe_size;
1520 int min_stripes = 1, num_stripes = 1;
1521 int i = 0, nr_devices;
1524 nr_devices = fs_info->fs_devices->open_devices;
1525 BUG_ON(!nr_devices);
1527 devices_info = kmalloc_array(nr_devices, sizeof(*devices_info),
1532 /* calc min stripe number for data space alloction */
1533 type = btrfs_get_alloc_profile(root, 1);
1534 if (type & BTRFS_BLOCK_GROUP_RAID0) {
1536 num_stripes = nr_devices;
1537 } else if (type & BTRFS_BLOCK_GROUP_RAID1) {
1540 } else if (type & BTRFS_BLOCK_GROUP_RAID10) {
1545 if (type & BTRFS_BLOCK_GROUP_DUP)
1546 min_stripe_size = 2 * BTRFS_STRIPE_LEN;
1548 min_stripe_size = BTRFS_STRIPE_LEN;
1550 list_for_each_entry(device, &fs_devices->devices, dev_list) {
1551 if (!device->in_fs_metadata || !device->bdev ||
1552 device->is_tgtdev_for_dev_replace)
1555 avail_space = device->total_bytes - device->bytes_used;
1557 /* align with stripe_len */
1558 do_div(avail_space, BTRFS_STRIPE_LEN);
1559 avail_space *= BTRFS_STRIPE_LEN;
1562 * In order to avoid overwritting the superblock on the drive,
1563 * btrfs starts at an offset of at least 1MB when doing chunk
1566 skip_space = 1024 * 1024;
1568 /* user can set the offset in fs_info->alloc_start. */
1569 if (fs_info->alloc_start + BTRFS_STRIPE_LEN <=
1570 device->total_bytes)
1571 skip_space = max(fs_info->alloc_start, skip_space);
1574 * btrfs can not use the free space in [0, skip_space - 1],
1575 * we must subtract it from the total. In order to implement
1576 * it, we account the used space in this range first.
1578 ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1,
1581 kfree(devices_info);
1585 /* calc the free space in [0, skip_space - 1] */
1586 skip_space -= used_space;
1589 * we can use the free space in [0, skip_space - 1], subtract
1590 * it from the total.
1592 if (avail_space && avail_space >= skip_space)
1593 avail_space -= skip_space;
1597 if (avail_space < min_stripe_size)
1600 devices_info[i].dev = device;
1601 devices_info[i].max_avail = avail_space;
1608 btrfs_descending_sort_devices(devices_info, nr_devices);
1612 while (nr_devices >= min_stripes) {
1613 if (num_stripes > nr_devices)
1614 num_stripes = nr_devices;
1616 if (devices_info[i].max_avail >= min_stripe_size) {
1620 avail_space += devices_info[i].max_avail * num_stripes;
1621 alloc_size = devices_info[i].max_avail;
1622 for (j = i + 1 - num_stripes; j <= i; j++)
1623 devices_info[j].max_avail -= alloc_size;
1629 kfree(devices_info);
1630 *free_bytes = avail_space;
1634 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
1636 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
1637 struct btrfs_super_block *disk_super = fs_info->super_copy;
1638 struct list_head *head = &fs_info->space_info;
1639 struct btrfs_space_info *found;
1641 u64 total_free_data = 0;
1642 int bits = dentry->d_sb->s_blocksize_bits;
1643 __be32 *fsid = (__be32 *)fs_info->fsid;
1646 /* holding chunk_muext to avoid allocating new chunks */
1647 mutex_lock(&fs_info->chunk_mutex);
1649 list_for_each_entry_rcu(found, head, list) {
1650 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
1651 total_free_data += found->disk_total - found->disk_used;
1653 btrfs_account_ro_block_groups_free_space(found);
1656 total_used += found->disk_used;
1660 buf->f_namelen = BTRFS_NAME_LEN;
1661 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
1662 buf->f_bfree = buf->f_blocks - (total_used >> bits);
1663 buf->f_bsize = dentry->d_sb->s_blocksize;
1664 buf->f_type = BTRFS_SUPER_MAGIC;
1665 buf->f_bavail = total_free_data;
1666 ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data);
1668 mutex_unlock(&fs_info->chunk_mutex);
1671 buf->f_bavail += total_free_data;
1672 buf->f_bavail = buf->f_bavail >> bits;
1673 mutex_unlock(&fs_info->chunk_mutex);
1675 /* We treat it as constant endianness (it doesn't matter _which_)
1676 because we want the fsid to come out the same whether mounted
1677 on a big-endian or little-endian host */
1678 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
1679 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
1680 /* Mask in the root object ID too, to disambiguate subvols */
1681 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
1682 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
1687 static void btrfs_kill_super(struct super_block *sb)
1689 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1690 kill_anon_super(sb);
1691 free_fs_info(fs_info);
1694 static struct file_system_type btrfs_fs_type = {
1695 .owner = THIS_MODULE,
1697 .mount = btrfs_mount,
1698 .kill_sb = btrfs_kill_super,
1699 .fs_flags = FS_REQUIRES_DEV,
1701 MODULE_ALIAS_FS("btrfs");
1704 * used by btrfsctl to scan devices when no FS is mounted
1706 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
1709 struct btrfs_ioctl_vol_args *vol;
1710 struct btrfs_fs_devices *fs_devices;
1713 if (!capable(CAP_SYS_ADMIN))
1716 vol = memdup_user((void __user *)arg, sizeof(*vol));
1718 return PTR_ERR(vol);
1721 case BTRFS_IOC_SCAN_DEV:
1722 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1723 &btrfs_fs_type, &fs_devices);
1725 case BTRFS_IOC_DEVICES_READY:
1726 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1727 &btrfs_fs_type, &fs_devices);
1730 ret = !(fs_devices->num_devices == fs_devices->total_devices);
1738 static int btrfs_freeze(struct super_block *sb)
1740 struct btrfs_trans_handle *trans;
1741 struct btrfs_root *root = btrfs_sb(sb)->tree_root;
1743 trans = btrfs_attach_transaction_barrier(root);
1744 if (IS_ERR(trans)) {
1745 /* no transaction, don't bother */
1746 if (PTR_ERR(trans) == -ENOENT)
1748 return PTR_ERR(trans);
1750 return btrfs_commit_transaction(trans, root);
1753 static int btrfs_unfreeze(struct super_block *sb)
1758 static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
1760 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
1761 struct btrfs_fs_devices *cur_devices;
1762 struct btrfs_device *dev, *first_dev = NULL;
1763 struct list_head *head;
1764 struct rcu_string *name;
1766 mutex_lock(&fs_info->fs_devices->device_list_mutex);
1767 cur_devices = fs_info->fs_devices;
1768 while (cur_devices) {
1769 head = &cur_devices->devices;
1770 list_for_each_entry(dev, head, dev_list) {
1773 if (!first_dev || dev->devid < first_dev->devid)
1776 cur_devices = cur_devices->seed;
1781 name = rcu_dereference(first_dev->name);
1782 seq_escape(m, name->str, " \t\n\\");
1787 mutex_unlock(&fs_info->fs_devices->device_list_mutex);
1791 static const struct super_operations btrfs_super_ops = {
1792 .drop_inode = btrfs_drop_inode,
1793 .evict_inode = btrfs_evict_inode,
1794 .put_super = btrfs_put_super,
1795 .sync_fs = btrfs_sync_fs,
1796 .show_options = btrfs_show_options,
1797 .show_devname = btrfs_show_devname,
1798 .write_inode = btrfs_write_inode,
1799 .alloc_inode = btrfs_alloc_inode,
1800 .destroy_inode = btrfs_destroy_inode,
1801 .statfs = btrfs_statfs,
1802 .remount_fs = btrfs_remount,
1803 .freeze_fs = btrfs_freeze,
1804 .unfreeze_fs = btrfs_unfreeze,
1807 static const struct file_operations btrfs_ctl_fops = {
1808 .unlocked_ioctl = btrfs_control_ioctl,
1809 .compat_ioctl = btrfs_control_ioctl,
1810 .owner = THIS_MODULE,
1811 .llseek = noop_llseek,
1814 static struct miscdevice btrfs_misc = {
1815 .minor = BTRFS_MINOR,
1816 .name = "btrfs-control",
1817 .fops = &btrfs_ctl_fops
1820 MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
1821 MODULE_ALIAS("devname:btrfs-control");
1823 static int btrfs_interface_init(void)
1825 return misc_register(&btrfs_misc);
1828 static void btrfs_interface_exit(void)
1830 if (misc_deregister(&btrfs_misc) < 0)
1831 printk(KERN_INFO "BTRFS: misc_deregister failed for control device\n");
1834 static void btrfs_print_info(void)
1836 printk(KERN_INFO "Btrfs loaded"
1837 #ifdef CONFIG_BTRFS_DEBUG
1840 #ifdef CONFIG_BTRFS_ASSERT
1843 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1844 ", integrity-checker=on"
1849 static int btrfs_run_sanity_tests(void)
1853 ret = btrfs_init_test_fs();
1857 ret = btrfs_test_free_space_cache();
1860 ret = btrfs_test_extent_buffer_operations();
1863 ret = btrfs_test_extent_io();
1866 ret = btrfs_test_inodes();
1868 btrfs_destroy_test_fs();
1872 static int __init init_btrfs_fs(void)
1876 err = btrfs_hash_init();
1882 err = btrfs_init_sysfs();
1886 btrfs_init_compress();
1888 err = btrfs_init_cachep();
1892 err = extent_io_init();
1896 err = extent_map_init();
1898 goto free_extent_io;
1900 err = ordered_data_init();
1902 goto free_extent_map;
1904 err = btrfs_delayed_inode_init();
1906 goto free_ordered_data;
1908 err = btrfs_auto_defrag_init();
1910 goto free_delayed_inode;
1912 err = btrfs_delayed_ref_init();
1914 goto free_auto_defrag;
1916 err = btrfs_prelim_ref_init();
1918 goto free_prelim_ref;
1920 err = btrfs_interface_init();
1922 goto free_delayed_ref;
1924 btrfs_init_lockdep();
1928 err = btrfs_run_sanity_tests();
1930 goto unregister_ioctl;
1932 err = register_filesystem(&btrfs_fs_type);
1934 goto unregister_ioctl;
1939 btrfs_interface_exit();
1941 btrfs_prelim_ref_exit();
1943 btrfs_delayed_ref_exit();
1945 btrfs_auto_defrag_exit();
1947 btrfs_delayed_inode_exit();
1949 ordered_data_exit();
1955 btrfs_destroy_cachep();
1957 btrfs_exit_compress();
1964 static void __exit exit_btrfs_fs(void)
1966 btrfs_destroy_cachep();
1967 btrfs_delayed_ref_exit();
1968 btrfs_auto_defrag_exit();
1969 btrfs_delayed_inode_exit();
1970 btrfs_prelim_ref_exit();
1971 ordered_data_exit();
1974 btrfs_interface_exit();
1975 unregister_filesystem(&btrfs_fs_type);
1977 btrfs_cleanup_fs_uuids();
1978 btrfs_exit_compress();
1982 module_init(init_btrfs_fs)
1983 module_exit(exit_btrfs_fs)
1985 MODULE_LICENSE("GPL");