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"
54 #include "compression.h"
55 #include "rcu-string.h"
56 #include "dev-replace.h"
57 #include "free-space-cache.h"
59 #include "tests/btrfs-tests.h"
61 #define CREATE_TRACE_POINTS
62 #include <trace/events/btrfs.h>
64 static const struct super_operations btrfs_super_ops;
65 static struct file_system_type btrfs_fs_type;
67 static const char *btrfs_decode_error(int errno)
69 char *errstr = "unknown";
73 errstr = "IO failure";
76 errstr = "Out of memory";
79 errstr = "Readonly filesystem";
82 errstr = "Object already exists";
85 errstr = "No space left";
88 errstr = "No such entry";
95 static void save_error_info(struct btrfs_fs_info *fs_info)
98 * today we only save the error info into ram. Long term we'll
99 * also send it down to the disk
101 set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state);
104 /* btrfs handle error by forcing the filesystem readonly */
105 static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
107 struct super_block *sb = fs_info->sb;
109 if (sb->s_flags & MS_RDONLY)
112 if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
113 sb->s_flags |= MS_RDONLY;
114 btrfs_info(fs_info, "forced readonly");
116 * Note that a running device replace operation is not
117 * canceled here although there is no way to update
118 * the progress. It would add the risk of a deadlock,
119 * therefore the canceling is ommited. The only penalty
120 * is that some I/O remains active until the procedure
121 * completes. The next time when the filesystem is
122 * mounted writeable again, the device replace
123 * operation continues.
130 * __btrfs_std_error decodes expected errors from the caller and
131 * invokes the approciate error response.
133 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
134 unsigned int line, int errno, const char *fmt, ...)
136 struct super_block *sb = fs_info->sb;
140 * Special case: if the error is EROFS, and we're already
141 * under MS_RDONLY, then it is safe here.
143 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
146 errstr = btrfs_decode_error(errno);
148 struct va_format vaf;
156 "BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n",
157 sb->s_id, function, line, errno, errstr, &vaf);
160 printk(KERN_CRIT "BTRFS: error (device %s) in %s:%d: errno=%d %s\n",
161 sb->s_id, function, line, errno, errstr);
164 /* Don't go through full error handling during mount */
165 save_error_info(fs_info);
166 if (sb->s_flags & MS_BORN)
167 btrfs_handle_error(fs_info);
170 static const char * const logtypes[] = {
181 void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
183 struct super_block *sb = fs_info->sb;
185 struct va_format vaf;
187 const char *type = logtypes[4];
192 kern_level = printk_get_level(fmt);
194 size_t size = printk_skip_level(fmt) - fmt;
195 memcpy(lvl, fmt, size);
198 type = logtypes[kern_level - '0'];
205 printk("%sBTRFS %s (device %s): %pV\n", lvl, type, sb->s_id, &vaf);
212 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
213 unsigned int line, int errno, const char *fmt, ...)
215 struct super_block *sb = fs_info->sb;
218 * Special case: if the error is EROFS, and we're already
219 * under MS_RDONLY, then it is safe here.
221 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
224 /* Don't go through full error handling during mount */
225 if (sb->s_flags & MS_BORN) {
226 save_error_info(fs_info);
227 btrfs_handle_error(fs_info);
233 * We only mark the transaction aborted and then set the file system read-only.
234 * This will prevent new transactions from starting or trying to join this
237 * This means that error recovery at the call site is limited to freeing
238 * any local memory allocations and passing the error code up without
239 * further cleanup. The transaction should complete as it normally would
240 * in the call path but will return -EIO.
242 * We'll complete the cleanup in btrfs_end_transaction and
243 * btrfs_commit_transaction.
245 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
246 struct btrfs_root *root, const char *function,
247 unsigned int line, int errno)
250 * Report first abort since mount
252 if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED,
253 &root->fs_info->fs_state)) {
254 WARN(1, KERN_DEBUG "BTRFS: Transaction aborted (error %d)\n",
257 trans->aborted = errno;
258 /* Nothing used. The other threads that have joined this
259 * transaction may be able to continue. */
260 if (!trans->blocks_used) {
263 errstr = btrfs_decode_error(errno);
264 btrfs_warn(root->fs_info,
265 "%s:%d: Aborting unused transaction(%s).",
266 function, line, errstr);
269 ACCESS_ONCE(trans->transaction->aborted) = errno;
270 /* Wake up anybody who may be waiting on this transaction */
271 wake_up(&root->fs_info->transaction_wait);
272 wake_up(&root->fs_info->transaction_blocked_wait);
273 __btrfs_std_error(root->fs_info, function, line, errno, NULL);
276 * __btrfs_panic decodes unexpected, fatal errors from the caller,
277 * issues an alert, and either panics or BUGs, depending on mount options.
279 void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
280 unsigned int line, int errno, const char *fmt, ...)
282 char *s_id = "<unknown>";
284 struct va_format vaf = { .fmt = fmt };
288 s_id = fs_info->sb->s_id;
293 errstr = btrfs_decode_error(errno);
294 if (fs_info && (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR))
295 panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
296 s_id, function, line, &vaf, errno, errstr);
298 btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)",
299 function, line, &vaf, errno, errstr);
301 /* Caller calls BUG() */
304 static void btrfs_put_super(struct super_block *sb)
306 (void)close_ctree(btrfs_sb(sb)->tree_root);
307 /* FIXME: need to fix VFS to return error? */
308 /* AV: return it _where_? ->put_super() can be triggered by any number
309 * of async events, up to and including delivery of SIGKILL to the
310 * last process that kept it busy. Or segfault in the aforementioned
311 * process... Whom would you report that to?
316 Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum,
317 Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd,
318 Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress,
319 Opt_compress_type, Opt_compress_force, Opt_compress_force_type,
320 Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_discard,
321 Opt_space_cache, Opt_clear_cache, Opt_user_subvol_rm_allowed,
322 Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache,
323 Opt_no_space_cache, Opt_recovery, Opt_skip_balance,
324 Opt_check_integrity, Opt_check_integrity_including_extent_data,
325 Opt_check_integrity_print_mask, Opt_fatal_errors, Opt_rescan_uuid_tree,
326 Opt_commit_interval, Opt_barrier, Opt_nodefrag, Opt_nodiscard,
327 Opt_noenospc_debug, Opt_noflushoncommit, Opt_acl,
331 static match_table_t tokens = {
332 {Opt_degraded, "degraded"},
333 {Opt_subvol, "subvol=%s"},
334 {Opt_subvolid, "subvolid=%s"},
335 {Opt_device, "device=%s"},
336 {Opt_nodatasum, "nodatasum"},
337 {Opt_nodatacow, "nodatacow"},
338 {Opt_nobarrier, "nobarrier"},
339 {Opt_barrier, "barrier"},
340 {Opt_max_inline, "max_inline=%s"},
341 {Opt_alloc_start, "alloc_start=%s"},
342 {Opt_thread_pool, "thread_pool=%d"},
343 {Opt_compress, "compress"},
344 {Opt_compress_type, "compress=%s"},
345 {Opt_compress_force, "compress-force"},
346 {Opt_compress_force_type, "compress-force=%s"},
348 {Opt_ssd_spread, "ssd_spread"},
349 {Opt_nossd, "nossd"},
351 {Opt_noacl, "noacl"},
352 {Opt_notreelog, "notreelog"},
353 {Opt_flushoncommit, "flushoncommit"},
354 {Opt_noflushoncommit, "noflushoncommit"},
355 {Opt_ratio, "metadata_ratio=%d"},
356 {Opt_discard, "discard"},
357 {Opt_nodiscard, "nodiscard"},
358 {Opt_space_cache, "space_cache"},
359 {Opt_clear_cache, "clear_cache"},
360 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
361 {Opt_enospc_debug, "enospc_debug"},
362 {Opt_noenospc_debug, "noenospc_debug"},
363 {Opt_subvolrootid, "subvolrootid=%d"},
364 {Opt_defrag, "autodefrag"},
365 {Opt_nodefrag, "noautodefrag"},
366 {Opt_inode_cache, "inode_cache"},
367 {Opt_no_space_cache, "nospace_cache"},
368 {Opt_recovery, "recovery"},
369 {Opt_skip_balance, "skip_balance"},
370 {Opt_check_integrity, "check_int"},
371 {Opt_check_integrity_including_extent_data, "check_int_data"},
372 {Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
373 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
374 {Opt_fatal_errors, "fatal_errors=%s"},
375 {Opt_commit_interval, "commit=%d"},
380 * Regular mount options parser. Everything that is needed only when
381 * reading in a new superblock is parsed here.
382 * XXX JDM: This needs to be cleaned up for remount.
384 int btrfs_parse_options(struct btrfs_root *root, char *options)
386 struct btrfs_fs_info *info = root->fs_info;
387 substring_t args[MAX_OPT_ARGS];
388 char *p, *num, *orig = NULL;
393 bool compress_force = false;
395 cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
397 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
403 * strsep changes the string, duplicate it because parse_options
406 options = kstrdup(options, GFP_NOFS);
412 while ((p = strsep(&options, ",")) != NULL) {
417 token = match_token(p, tokens, args);
420 btrfs_info(root->fs_info, "allowing degraded mounts");
421 btrfs_set_opt(info->mount_opt, DEGRADED);
425 case Opt_subvolrootid:
428 * These are parsed by btrfs_parse_early_options
429 * and can be happily ignored here.
433 btrfs_info(root->fs_info, "setting nodatasum");
434 btrfs_set_opt(info->mount_opt, NODATASUM);
437 if (!btrfs_test_opt(root, COMPRESS) ||
438 !btrfs_test_opt(root, FORCE_COMPRESS)) {
439 btrfs_info(root->fs_info,
440 "setting nodatacow, compression disabled");
442 btrfs_info(root->fs_info, "setting nodatacow");
444 btrfs_clear_opt(info->mount_opt, COMPRESS);
445 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
446 btrfs_set_opt(info->mount_opt, NODATACOW);
447 btrfs_set_opt(info->mount_opt, NODATASUM);
449 case Opt_compress_force:
450 case Opt_compress_force_type:
451 compress_force = true;
454 case Opt_compress_type:
455 if (token == Opt_compress ||
456 token == Opt_compress_force ||
457 strcmp(args[0].from, "zlib") == 0) {
458 compress_type = "zlib";
459 info->compress_type = BTRFS_COMPRESS_ZLIB;
460 btrfs_set_opt(info->mount_opt, COMPRESS);
461 btrfs_clear_opt(info->mount_opt, NODATACOW);
462 btrfs_clear_opt(info->mount_opt, NODATASUM);
463 } else if (strcmp(args[0].from, "lzo") == 0) {
464 compress_type = "lzo";
465 info->compress_type = BTRFS_COMPRESS_LZO;
466 btrfs_set_opt(info->mount_opt, COMPRESS);
467 btrfs_clear_opt(info->mount_opt, NODATACOW);
468 btrfs_clear_opt(info->mount_opt, NODATASUM);
469 btrfs_set_fs_incompat(info, COMPRESS_LZO);
470 } else if (strncmp(args[0].from, "no", 2) == 0) {
471 compress_type = "no";
472 btrfs_clear_opt(info->mount_opt, COMPRESS);
473 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
474 compress_force = false;
480 if (compress_force) {
481 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
482 btrfs_info(root->fs_info, "force %s compression",
484 } else if (btrfs_test_opt(root, COMPRESS)) {
485 pr_info("btrfs: use %s compression\n",
490 btrfs_info(root->fs_info, "use ssd allocation scheme");
491 btrfs_set_opt(info->mount_opt, SSD);
494 btrfs_info(root->fs_info, "use spread ssd allocation scheme");
495 btrfs_set_opt(info->mount_opt, SSD);
496 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
499 btrfs_info(root->fs_info, "not using ssd allocation scheme");
500 btrfs_set_opt(info->mount_opt, NOSSD);
501 btrfs_clear_opt(info->mount_opt, SSD);
502 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
505 if (btrfs_test_opt(root, NOBARRIER))
506 btrfs_info(root->fs_info, "turning on barriers");
507 btrfs_clear_opt(info->mount_opt, NOBARRIER);
510 btrfs_info(root->fs_info, "turning off barriers");
511 btrfs_set_opt(info->mount_opt, NOBARRIER);
513 case Opt_thread_pool:
514 ret = match_int(&args[0], &intarg);
517 } else if (intarg > 0) {
518 info->thread_pool_size = intarg;
525 num = match_strdup(&args[0]);
527 info->max_inline = memparse(num, NULL);
530 if (info->max_inline) {
531 info->max_inline = max_t(u64,
535 btrfs_info(root->fs_info, "max_inline at %llu",
542 case Opt_alloc_start:
543 num = match_strdup(&args[0]);
545 mutex_lock(&info->chunk_mutex);
546 info->alloc_start = memparse(num, NULL);
547 mutex_unlock(&info->chunk_mutex);
549 btrfs_info(root->fs_info, "allocations start at %llu",
557 root->fs_info->sb->s_flags |= MS_POSIXACL;
560 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
563 btrfs_info(root->fs_info, "disabling tree log");
564 btrfs_set_opt(info->mount_opt, NOTREELOG);
566 case Opt_flushoncommit:
567 btrfs_info(root->fs_info, "turning on flush-on-commit");
568 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
570 case Opt_noflushoncommit:
571 if (btrfs_test_opt(root, FLUSHONCOMMIT))
572 btrfs_info(root->fs_info, "turning off flush-on-commit");
573 btrfs_clear_opt(info->mount_opt, FLUSHONCOMMIT);
576 ret = match_int(&args[0], &intarg);
579 } else if (intarg >= 0) {
580 info->metadata_ratio = intarg;
581 btrfs_info(root->fs_info, "metadata ratio %d",
582 info->metadata_ratio);
589 btrfs_set_opt(info->mount_opt, DISCARD);
592 btrfs_clear_opt(info->mount_opt, DISCARD);
594 case Opt_space_cache:
595 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
597 case Opt_rescan_uuid_tree:
598 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
600 case Opt_no_space_cache:
601 btrfs_info(root->fs_info, "disabling disk space caching");
602 btrfs_clear_opt(info->mount_opt, SPACE_CACHE);
604 case Opt_inode_cache:
605 btrfs_info(root->fs_info, "enabling inode map caching");
606 btrfs_set_opt(info->mount_opt, INODE_MAP_CACHE);
608 case Opt_clear_cache:
609 btrfs_info(root->fs_info, "force clearing of disk cache");
610 btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
612 case Opt_user_subvol_rm_allowed:
613 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
615 case Opt_enospc_debug:
616 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
618 case Opt_noenospc_debug:
619 btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG);
622 btrfs_info(root->fs_info, "enabling auto defrag");
623 btrfs_set_opt(info->mount_opt, AUTO_DEFRAG);
626 if (btrfs_test_opt(root, AUTO_DEFRAG))
627 btrfs_info(root->fs_info, "disabling auto defrag");
628 btrfs_clear_opt(info->mount_opt, AUTO_DEFRAG);
631 btrfs_info(root->fs_info, "enabling auto recovery");
632 btrfs_set_opt(info->mount_opt, RECOVERY);
634 case Opt_skip_balance:
635 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
637 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
638 case Opt_check_integrity_including_extent_data:
639 btrfs_info(root->fs_info,
640 "enabling check integrity including extent data");
641 btrfs_set_opt(info->mount_opt,
642 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
643 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
645 case Opt_check_integrity:
646 btrfs_info(root->fs_info, "enabling check integrity");
647 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
649 case Opt_check_integrity_print_mask:
650 ret = match_int(&args[0], &intarg);
653 } else if (intarg >= 0) {
654 info->check_integrity_print_mask = intarg;
655 btrfs_info(root->fs_info, "check_integrity_print_mask 0x%x",
656 info->check_integrity_print_mask);
663 case Opt_check_integrity_including_extent_data:
664 case Opt_check_integrity:
665 case Opt_check_integrity_print_mask:
666 btrfs_err(root->fs_info,
667 "support for check_integrity* not compiled in!");
671 case Opt_fatal_errors:
672 if (strcmp(args[0].from, "panic") == 0)
673 btrfs_set_opt(info->mount_opt,
674 PANIC_ON_FATAL_ERROR);
675 else if (strcmp(args[0].from, "bug") == 0)
676 btrfs_clear_opt(info->mount_opt,
677 PANIC_ON_FATAL_ERROR);
683 case Opt_commit_interval:
685 ret = match_int(&args[0], &intarg);
687 btrfs_err(root->fs_info, "invalid commit interval");
693 btrfs_warn(root->fs_info, "excessive commit interval %d",
696 info->commit_interval = intarg;
698 btrfs_info(root->fs_info, "using default commit interval %ds",
699 BTRFS_DEFAULT_COMMIT_INTERVAL);
700 info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
704 btrfs_info(root->fs_info, "unrecognized mount option '%s'", p);
712 if (!ret && btrfs_test_opt(root, SPACE_CACHE))
713 btrfs_info(root->fs_info, "disk space caching is enabled");
719 * Parse mount options that are required early in the mount process.
721 * All other options will be parsed on much later in the mount process and
722 * only when we need to allocate a new super block.
724 static int btrfs_parse_early_options(const char *options, fmode_t flags,
725 void *holder, char **subvol_name, u64 *subvol_objectid,
726 struct btrfs_fs_devices **fs_devices)
728 substring_t args[MAX_OPT_ARGS];
729 char *device_name, *opts, *orig, *p;
737 * strsep changes the string, duplicate it because parse_options
740 opts = kstrdup(options, GFP_KERNEL);
745 while ((p = strsep(&opts, ",")) != NULL) {
750 token = match_token(p, tokens, args);
754 *subvol_name = match_strdup(&args[0]);
761 num = match_strdup(&args[0]);
763 *subvol_objectid = memparse(num, NULL);
765 /* we want the original fs_tree */
766 if (!*subvol_objectid)
768 BTRFS_FS_TREE_OBJECTID;
774 case Opt_subvolrootid:
776 "BTRFS: 'subvolrootid' mount option is deprecated and has "
780 device_name = match_strdup(&args[0]);
785 error = btrfs_scan_one_device(device_name,
786 flags, holder, fs_devices);
801 static struct dentry *get_default_root(struct super_block *sb,
804 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
805 struct btrfs_root *root = fs_info->tree_root;
806 struct btrfs_root *new_root;
807 struct btrfs_dir_item *di;
808 struct btrfs_path *path;
809 struct btrfs_key location;
815 * We have a specific subvol we want to mount, just setup location and
816 * go look up the root.
818 if (subvol_objectid) {
819 location.objectid = subvol_objectid;
820 location.type = BTRFS_ROOT_ITEM_KEY;
821 location.offset = (u64)-1;
825 path = btrfs_alloc_path();
827 return ERR_PTR(-ENOMEM);
828 path->leave_spinning = 1;
831 * Find the "default" dir item which points to the root item that we
832 * will mount by default if we haven't been given a specific subvolume
835 dir_id = btrfs_super_root_dir(fs_info->super_copy);
836 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
838 btrfs_free_path(path);
843 * Ok the default dir item isn't there. This is weird since
844 * it's always been there, but don't freak out, just try and
845 * mount to root most subvolume.
847 btrfs_free_path(path);
848 dir_id = BTRFS_FIRST_FREE_OBJECTID;
849 new_root = fs_info->fs_root;
853 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
854 btrfs_free_path(path);
857 new_root = btrfs_read_fs_root_no_name(fs_info, &location);
858 if (IS_ERR(new_root))
859 return ERR_CAST(new_root);
861 dir_id = btrfs_root_dirid(&new_root->root_item);
863 location.objectid = dir_id;
864 location.type = BTRFS_INODE_ITEM_KEY;
867 inode = btrfs_iget(sb, &location, new_root, &new);
869 return ERR_CAST(inode);
872 * If we're just mounting the root most subvol put the inode and return
873 * a reference to the dentry. We will have already gotten a reference
874 * to the inode in btrfs_fill_super so we're good to go.
876 if (!new && sb->s_root->d_inode == inode) {
878 return dget(sb->s_root);
881 return d_obtain_alias(inode);
884 static int btrfs_fill_super(struct super_block *sb,
885 struct btrfs_fs_devices *fs_devices,
886 void *data, int silent)
889 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
890 struct btrfs_key key;
893 sb->s_maxbytes = MAX_LFS_FILESIZE;
894 sb->s_magic = BTRFS_SUPER_MAGIC;
895 sb->s_op = &btrfs_super_ops;
896 sb->s_d_op = &btrfs_dentry_operations;
897 sb->s_export_op = &btrfs_export_ops;
898 sb->s_xattr = btrfs_xattr_handlers;
900 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
901 sb->s_flags |= MS_POSIXACL;
903 sb->s_flags |= MS_I_VERSION;
904 err = open_ctree(sb, fs_devices, (char *)data);
906 printk(KERN_ERR "BTRFS: open_ctree failed\n");
910 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
911 key.type = BTRFS_INODE_ITEM_KEY;
913 inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL);
915 err = PTR_ERR(inode);
919 sb->s_root = d_make_root(inode);
925 save_mount_options(sb, data);
926 cleancache_init_fs(sb);
927 sb->s_flags |= MS_ACTIVE;
931 close_ctree(fs_info->tree_root);
935 int btrfs_sync_fs(struct super_block *sb, int wait)
937 struct btrfs_trans_handle *trans;
938 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
939 struct btrfs_root *root = fs_info->tree_root;
941 trace_btrfs_sync_fs(wait);
944 filemap_flush(fs_info->btree_inode->i_mapping);
948 btrfs_wait_ordered_roots(fs_info, -1);
950 trans = btrfs_attach_transaction_barrier(root);
952 /* no transaction, don't bother */
953 if (PTR_ERR(trans) == -ENOENT)
955 return PTR_ERR(trans);
957 return btrfs_commit_transaction(trans, root);
960 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
962 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
963 struct btrfs_root *root = info->tree_root;
966 if (btrfs_test_opt(root, DEGRADED))
967 seq_puts(seq, ",degraded");
968 if (btrfs_test_opt(root, NODATASUM))
969 seq_puts(seq, ",nodatasum");
970 if (btrfs_test_opt(root, NODATACOW))
971 seq_puts(seq, ",nodatacow");
972 if (btrfs_test_opt(root, NOBARRIER))
973 seq_puts(seq, ",nobarrier");
974 if (info->max_inline != 8192 * 1024)
975 seq_printf(seq, ",max_inline=%llu", info->max_inline);
976 if (info->alloc_start != 0)
977 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
978 if (info->thread_pool_size != min_t(unsigned long,
979 num_online_cpus() + 2, 8))
980 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
981 if (btrfs_test_opt(root, COMPRESS)) {
982 if (info->compress_type == BTRFS_COMPRESS_ZLIB)
983 compress_type = "zlib";
985 compress_type = "lzo";
986 if (btrfs_test_opt(root, FORCE_COMPRESS))
987 seq_printf(seq, ",compress-force=%s", compress_type);
989 seq_printf(seq, ",compress=%s", compress_type);
991 if (btrfs_test_opt(root, NOSSD))
992 seq_puts(seq, ",nossd");
993 if (btrfs_test_opt(root, SSD_SPREAD))
994 seq_puts(seq, ",ssd_spread");
995 else if (btrfs_test_opt(root, SSD))
996 seq_puts(seq, ",ssd");
997 if (btrfs_test_opt(root, NOTREELOG))
998 seq_puts(seq, ",notreelog");
999 if (btrfs_test_opt(root, FLUSHONCOMMIT))
1000 seq_puts(seq, ",flushoncommit");
1001 if (btrfs_test_opt(root, DISCARD))
1002 seq_puts(seq, ",discard");
1003 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
1004 seq_puts(seq, ",noacl");
1005 if (btrfs_test_opt(root, SPACE_CACHE))
1006 seq_puts(seq, ",space_cache");
1008 seq_puts(seq, ",nospace_cache");
1009 if (btrfs_test_opt(root, RESCAN_UUID_TREE))
1010 seq_puts(seq, ",rescan_uuid_tree");
1011 if (btrfs_test_opt(root, CLEAR_CACHE))
1012 seq_puts(seq, ",clear_cache");
1013 if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
1014 seq_puts(seq, ",user_subvol_rm_allowed");
1015 if (btrfs_test_opt(root, ENOSPC_DEBUG))
1016 seq_puts(seq, ",enospc_debug");
1017 if (btrfs_test_opt(root, AUTO_DEFRAG))
1018 seq_puts(seq, ",autodefrag");
1019 if (btrfs_test_opt(root, INODE_MAP_CACHE))
1020 seq_puts(seq, ",inode_cache");
1021 if (btrfs_test_opt(root, SKIP_BALANCE))
1022 seq_puts(seq, ",skip_balance");
1023 if (btrfs_test_opt(root, RECOVERY))
1024 seq_puts(seq, ",recovery");
1025 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1026 if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1027 seq_puts(seq, ",check_int_data");
1028 else if (btrfs_test_opt(root, CHECK_INTEGRITY))
1029 seq_puts(seq, ",check_int");
1030 if (info->check_integrity_print_mask)
1031 seq_printf(seq, ",check_int_print_mask=%d",
1032 info->check_integrity_print_mask);
1034 if (info->metadata_ratio)
1035 seq_printf(seq, ",metadata_ratio=%d",
1036 info->metadata_ratio);
1037 if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
1038 seq_puts(seq, ",fatal_errors=panic");
1039 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1040 seq_printf(seq, ",commit=%d", info->commit_interval);
1044 static int btrfs_test_super(struct super_block *s, void *data)
1046 struct btrfs_fs_info *p = data;
1047 struct btrfs_fs_info *fs_info = btrfs_sb(s);
1049 return fs_info->fs_devices == p->fs_devices;
1052 static int btrfs_set_super(struct super_block *s, void *data)
1054 int err = set_anon_super(s, data);
1056 s->s_fs_info = data;
1061 * subvolumes are identified by ino 256
1063 static inline int is_subvolume_inode(struct inode *inode)
1065 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1071 * This will strip out the subvol=%s argument for an argument string and add
1072 * subvolid=0 to make sure we get the actual tree root for path walking to the
1075 static char *setup_root_args(char *args)
1077 unsigned len = strlen(args) + 2 + 1;
1078 char *src, *dst, *buf;
1081 * We need the same args as before, but with this substitution:
1082 * s!subvol=[^,]+!subvolid=0!
1084 * Since the replacement string is up to 2 bytes longer than the
1085 * original, allocate strlen(args) + 2 + 1 bytes.
1088 src = strstr(args, "subvol=");
1089 /* This shouldn't happen, but just in case.. */
1093 buf = dst = kmalloc(len, GFP_NOFS);
1098 * If the subvol= arg is not at the start of the string,
1099 * copy whatever precedes it into buf.
1104 dst += strlen(args);
1107 strcpy(dst, "subvolid=0");
1108 dst += strlen("subvolid=0");
1111 * If there is a "," after the original subvol=... string,
1112 * copy that suffix into our buffer. Otherwise, we're done.
1114 src = strchr(src, ',');
1121 static struct dentry *mount_subvol(const char *subvol_name, int flags,
1122 const char *device_name, char *data)
1124 struct dentry *root;
1125 struct vfsmount *mnt;
1128 newargs = setup_root_args(data);
1130 return ERR_PTR(-ENOMEM);
1131 mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name,
1135 return ERR_CAST(mnt);
1137 root = mount_subtree(mnt, subvol_name);
1139 if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) {
1140 struct super_block *s = root->d_sb;
1142 root = ERR_PTR(-EINVAL);
1143 deactivate_locked_super(s);
1144 printk(KERN_ERR "BTRFS: '%s' is not a valid subvolume\n",
1152 * Find a superblock for the given device / mount point.
1154 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1155 * for multiple device setup. Make sure to keep it in sync.
1157 static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
1158 const char *device_name, void *data)
1160 struct block_device *bdev = NULL;
1161 struct super_block *s;
1162 struct dentry *root;
1163 struct btrfs_fs_devices *fs_devices = NULL;
1164 struct btrfs_fs_info *fs_info = NULL;
1165 fmode_t mode = FMODE_READ;
1166 char *subvol_name = NULL;
1167 u64 subvol_objectid = 0;
1170 if (!(flags & MS_RDONLY))
1171 mode |= FMODE_WRITE;
1173 error = btrfs_parse_early_options(data, mode, fs_type,
1174 &subvol_name, &subvol_objectid,
1178 return ERR_PTR(error);
1182 root = mount_subvol(subvol_name, flags, device_name, data);
1187 error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices);
1189 return ERR_PTR(error);
1192 * Setup a dummy root and fs_info for test/set super. This is because
1193 * we don't actually fill this stuff out until open_ctree, but we need
1194 * it for searching for existing supers, so this lets us do that and
1195 * then open_ctree will properly initialize everything later.
1197 fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
1199 return ERR_PTR(-ENOMEM);
1201 fs_info->fs_devices = fs_devices;
1203 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1204 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1205 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1210 error = btrfs_open_devices(fs_devices, mode, fs_type);
1214 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
1216 goto error_close_devices;
1219 bdev = fs_devices->latest_bdev;
1220 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC,
1224 goto error_close_devices;
1228 btrfs_close_devices(fs_devices);
1229 free_fs_info(fs_info);
1230 if ((flags ^ s->s_flags) & MS_RDONLY)
1233 char b[BDEVNAME_SIZE];
1235 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
1236 btrfs_sb(s)->bdev_holder = fs_type;
1237 error = btrfs_fill_super(s, fs_devices, data,
1238 flags & MS_SILENT ? 1 : 0);
1241 root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error);
1243 deactivate_locked_super(s);
1247 error_close_devices:
1248 btrfs_close_devices(fs_devices);
1250 free_fs_info(fs_info);
1251 return ERR_PTR(error);
1254 static void btrfs_set_max_workers(struct btrfs_workers *workers, int new_limit)
1256 spin_lock_irq(&workers->lock);
1257 workers->max_workers = new_limit;
1258 spin_unlock_irq(&workers->lock);
1261 static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1262 int new_pool_size, int old_pool_size)
1264 if (new_pool_size == old_pool_size)
1267 fs_info->thread_pool_size = new_pool_size;
1269 btrfs_info(fs_info, "resize thread pool %d -> %d",
1270 old_pool_size, new_pool_size);
1272 btrfs_set_max_workers(&fs_info->generic_worker, new_pool_size);
1273 btrfs_set_max_workers(&fs_info->workers, new_pool_size);
1274 btrfs_set_max_workers(&fs_info->delalloc_workers, new_pool_size);
1275 btrfs_set_max_workers(&fs_info->submit_workers, new_pool_size);
1276 btrfs_set_max_workers(&fs_info->caching_workers, new_pool_size);
1277 btrfs_set_max_workers(&fs_info->fixup_workers, new_pool_size);
1278 btrfs_set_max_workers(&fs_info->endio_workers, new_pool_size);
1279 btrfs_set_max_workers(&fs_info->endio_meta_workers, new_pool_size);
1280 btrfs_set_max_workers(&fs_info->endio_meta_write_workers, new_pool_size);
1281 btrfs_set_max_workers(&fs_info->endio_write_workers, new_pool_size);
1282 btrfs_set_max_workers(&fs_info->endio_freespace_worker, new_pool_size);
1283 btrfs_set_max_workers(&fs_info->delayed_workers, new_pool_size);
1284 btrfs_set_max_workers(&fs_info->readahead_workers, new_pool_size);
1285 btrfs_set_max_workers(&fs_info->scrub_wr_completion_workers,
1289 static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
1291 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1294 static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1295 unsigned long old_opts, int flags)
1297 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1298 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1299 (flags & MS_RDONLY))) {
1300 /* wait for any defraggers to finish */
1301 wait_event(fs_info->transaction_wait,
1302 (atomic_read(&fs_info->defrag_running) == 0));
1303 if (flags & MS_RDONLY)
1304 sync_filesystem(fs_info->sb);
1308 static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1309 unsigned long old_opts)
1312 * We need cleanup all defragable inodes if the autodefragment is
1313 * close or the fs is R/O.
1315 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1316 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1317 (fs_info->sb->s_flags & MS_RDONLY))) {
1318 btrfs_cleanup_defrag_inodes(fs_info);
1321 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1324 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1326 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1327 struct btrfs_root *root = fs_info->tree_root;
1328 unsigned old_flags = sb->s_flags;
1329 unsigned long old_opts = fs_info->mount_opt;
1330 unsigned long old_compress_type = fs_info->compress_type;
1331 u64 old_max_inline = fs_info->max_inline;
1332 u64 old_alloc_start = fs_info->alloc_start;
1333 int old_thread_pool_size = fs_info->thread_pool_size;
1334 unsigned int old_metadata_ratio = fs_info->metadata_ratio;
1337 btrfs_remount_prepare(fs_info);
1339 ret = btrfs_parse_options(root, data);
1345 btrfs_remount_begin(fs_info, old_opts, *flags);
1346 btrfs_resize_thread_pool(fs_info,
1347 fs_info->thread_pool_size, old_thread_pool_size);
1349 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
1352 if (*flags & MS_RDONLY) {
1354 * this also happens on 'umount -rf' or on shutdown, when
1355 * the filesystem is busy.
1358 /* wait for the uuid_scan task to finish */
1359 down(&fs_info->uuid_tree_rescan_sem);
1360 /* avoid complains from lockdep et al. */
1361 up(&fs_info->uuid_tree_rescan_sem);
1363 sb->s_flags |= MS_RDONLY;
1365 btrfs_dev_replace_suspend_for_unmount(fs_info);
1366 btrfs_scrub_cancel(fs_info);
1367 btrfs_pause_balance(fs_info);
1369 ret = btrfs_commit_super(root);
1373 if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) {
1375 "Remounting read-write after error is not allowed");
1379 if (fs_info->fs_devices->rw_devices == 0) {
1384 if (fs_info->fs_devices->missing_devices >
1385 fs_info->num_tolerated_disk_barrier_failures &&
1386 !(*flags & MS_RDONLY)) {
1388 "too many missing devices, writeable remount is not allowed");
1393 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
1398 ret = btrfs_cleanup_fs_roots(fs_info);
1402 /* recover relocation */
1403 ret = btrfs_recover_relocation(root);
1407 ret = btrfs_resume_balance_async(fs_info);
1411 ret = btrfs_resume_dev_replace_async(fs_info);
1413 btrfs_warn(fs_info, "failed to resume dev_replace");
1417 if (!fs_info->uuid_root) {
1418 btrfs_info(fs_info, "creating UUID tree");
1419 ret = btrfs_create_uuid_tree(fs_info);
1421 btrfs_warn(fs_info, "failed to create the UUID tree %d", ret);
1425 sb->s_flags &= ~MS_RDONLY;
1428 btrfs_remount_cleanup(fs_info, old_opts);
1432 /* We've hit an error - don't reset MS_RDONLY */
1433 if (sb->s_flags & MS_RDONLY)
1434 old_flags |= MS_RDONLY;
1435 sb->s_flags = old_flags;
1436 fs_info->mount_opt = old_opts;
1437 fs_info->compress_type = old_compress_type;
1438 fs_info->max_inline = old_max_inline;
1439 mutex_lock(&fs_info->chunk_mutex);
1440 fs_info->alloc_start = old_alloc_start;
1441 mutex_unlock(&fs_info->chunk_mutex);
1442 btrfs_resize_thread_pool(fs_info,
1443 old_thread_pool_size, fs_info->thread_pool_size);
1444 fs_info->metadata_ratio = old_metadata_ratio;
1445 btrfs_remount_cleanup(fs_info, old_opts);
1449 /* Used to sort the devices by max_avail(descending sort) */
1450 static int btrfs_cmp_device_free_bytes(const void *dev_info1,
1451 const void *dev_info2)
1453 if (((struct btrfs_device_info *)dev_info1)->max_avail >
1454 ((struct btrfs_device_info *)dev_info2)->max_avail)
1456 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
1457 ((struct btrfs_device_info *)dev_info2)->max_avail)
1464 * sort the devices by max_avail, in which max free extent size of each device
1465 * is stored.(Descending Sort)
1467 static inline void btrfs_descending_sort_devices(
1468 struct btrfs_device_info *devices,
1471 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
1472 btrfs_cmp_device_free_bytes, NULL);
1476 * The helper to calc the free space on the devices that can be used to store
1479 static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
1481 struct btrfs_fs_info *fs_info = root->fs_info;
1482 struct btrfs_device_info *devices_info;
1483 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
1484 struct btrfs_device *device;
1489 u64 min_stripe_size;
1490 int min_stripes = 1, num_stripes = 1;
1491 int i = 0, nr_devices;
1494 nr_devices = fs_info->fs_devices->open_devices;
1495 BUG_ON(!nr_devices);
1497 devices_info = kmalloc_array(nr_devices, sizeof(*devices_info),
1502 /* calc min stripe number for data space alloction */
1503 type = btrfs_get_alloc_profile(root, 1);
1504 if (type & BTRFS_BLOCK_GROUP_RAID0) {
1506 num_stripes = nr_devices;
1507 } else if (type & BTRFS_BLOCK_GROUP_RAID1) {
1510 } else if (type & BTRFS_BLOCK_GROUP_RAID10) {
1515 if (type & BTRFS_BLOCK_GROUP_DUP)
1516 min_stripe_size = 2 * BTRFS_STRIPE_LEN;
1518 min_stripe_size = BTRFS_STRIPE_LEN;
1520 list_for_each_entry(device, &fs_devices->devices, dev_list) {
1521 if (!device->in_fs_metadata || !device->bdev ||
1522 device->is_tgtdev_for_dev_replace)
1525 avail_space = device->total_bytes - device->bytes_used;
1527 /* align with stripe_len */
1528 do_div(avail_space, BTRFS_STRIPE_LEN);
1529 avail_space *= BTRFS_STRIPE_LEN;
1532 * In order to avoid overwritting the superblock on the drive,
1533 * btrfs starts at an offset of at least 1MB when doing chunk
1536 skip_space = 1024 * 1024;
1538 /* user can set the offset in fs_info->alloc_start. */
1539 if (fs_info->alloc_start + BTRFS_STRIPE_LEN <=
1540 device->total_bytes)
1541 skip_space = max(fs_info->alloc_start, skip_space);
1544 * btrfs can not use the free space in [0, skip_space - 1],
1545 * we must subtract it from the total. In order to implement
1546 * it, we account the used space in this range first.
1548 ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1,
1551 kfree(devices_info);
1555 /* calc the free space in [0, skip_space - 1] */
1556 skip_space -= used_space;
1559 * we can use the free space in [0, skip_space - 1], subtract
1560 * it from the total.
1562 if (avail_space && avail_space >= skip_space)
1563 avail_space -= skip_space;
1567 if (avail_space < min_stripe_size)
1570 devices_info[i].dev = device;
1571 devices_info[i].max_avail = avail_space;
1578 btrfs_descending_sort_devices(devices_info, nr_devices);
1582 while (nr_devices >= min_stripes) {
1583 if (num_stripes > nr_devices)
1584 num_stripes = nr_devices;
1586 if (devices_info[i].max_avail >= min_stripe_size) {
1590 avail_space += devices_info[i].max_avail * num_stripes;
1591 alloc_size = devices_info[i].max_avail;
1592 for (j = i + 1 - num_stripes; j <= i; j++)
1593 devices_info[j].max_avail -= alloc_size;
1599 kfree(devices_info);
1600 *free_bytes = avail_space;
1604 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
1606 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
1607 struct btrfs_super_block *disk_super = fs_info->super_copy;
1608 struct list_head *head = &fs_info->space_info;
1609 struct btrfs_space_info *found;
1611 u64 total_free_data = 0;
1612 int bits = dentry->d_sb->s_blocksize_bits;
1613 __be32 *fsid = (__be32 *)fs_info->fsid;
1616 /* holding chunk_muext to avoid allocating new chunks */
1617 mutex_lock(&fs_info->chunk_mutex);
1619 list_for_each_entry_rcu(found, head, list) {
1620 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
1621 total_free_data += found->disk_total - found->disk_used;
1623 btrfs_account_ro_block_groups_free_space(found);
1626 total_used += found->disk_used;
1630 buf->f_namelen = BTRFS_NAME_LEN;
1631 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
1632 buf->f_bfree = buf->f_blocks - (total_used >> bits);
1633 buf->f_bsize = dentry->d_sb->s_blocksize;
1634 buf->f_type = BTRFS_SUPER_MAGIC;
1635 buf->f_bavail = total_free_data;
1636 ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data);
1638 mutex_unlock(&fs_info->chunk_mutex);
1641 buf->f_bavail += total_free_data;
1642 buf->f_bavail = buf->f_bavail >> bits;
1643 mutex_unlock(&fs_info->chunk_mutex);
1645 /* We treat it as constant endianness (it doesn't matter _which_)
1646 because we want the fsid to come out the same whether mounted
1647 on a big-endian or little-endian host */
1648 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
1649 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
1650 /* Mask in the root object ID too, to disambiguate subvols */
1651 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
1652 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
1657 static void btrfs_kill_super(struct super_block *sb)
1659 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1660 kill_anon_super(sb);
1661 free_fs_info(fs_info);
1664 static struct file_system_type btrfs_fs_type = {
1665 .owner = THIS_MODULE,
1667 .mount = btrfs_mount,
1668 .kill_sb = btrfs_kill_super,
1669 .fs_flags = FS_REQUIRES_DEV,
1671 MODULE_ALIAS_FS("btrfs");
1674 * used by btrfsctl to scan devices when no FS is mounted
1676 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
1679 struct btrfs_ioctl_vol_args *vol;
1680 struct btrfs_fs_devices *fs_devices;
1683 if (!capable(CAP_SYS_ADMIN))
1686 vol = memdup_user((void __user *)arg, sizeof(*vol));
1688 return PTR_ERR(vol);
1691 case BTRFS_IOC_SCAN_DEV:
1692 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1693 &btrfs_fs_type, &fs_devices);
1695 case BTRFS_IOC_DEVICES_READY:
1696 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1697 &btrfs_fs_type, &fs_devices);
1700 ret = !(fs_devices->num_devices == fs_devices->total_devices);
1708 static int btrfs_freeze(struct super_block *sb)
1710 struct btrfs_trans_handle *trans;
1711 struct btrfs_root *root = btrfs_sb(sb)->tree_root;
1713 trans = btrfs_attach_transaction_barrier(root);
1714 if (IS_ERR(trans)) {
1715 /* no transaction, don't bother */
1716 if (PTR_ERR(trans) == -ENOENT)
1718 return PTR_ERR(trans);
1720 return btrfs_commit_transaction(trans, root);
1723 static int btrfs_unfreeze(struct super_block *sb)
1728 static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
1730 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
1731 struct btrfs_fs_devices *cur_devices;
1732 struct btrfs_device *dev, *first_dev = NULL;
1733 struct list_head *head;
1734 struct rcu_string *name;
1736 mutex_lock(&fs_info->fs_devices->device_list_mutex);
1737 cur_devices = fs_info->fs_devices;
1738 while (cur_devices) {
1739 head = &cur_devices->devices;
1740 list_for_each_entry(dev, head, dev_list) {
1743 if (!first_dev || dev->devid < first_dev->devid)
1746 cur_devices = cur_devices->seed;
1751 name = rcu_dereference(first_dev->name);
1752 seq_escape(m, name->str, " \t\n\\");
1757 mutex_unlock(&fs_info->fs_devices->device_list_mutex);
1761 static const struct super_operations btrfs_super_ops = {
1762 .drop_inode = btrfs_drop_inode,
1763 .evict_inode = btrfs_evict_inode,
1764 .put_super = btrfs_put_super,
1765 .sync_fs = btrfs_sync_fs,
1766 .show_options = btrfs_show_options,
1767 .show_devname = btrfs_show_devname,
1768 .write_inode = btrfs_write_inode,
1769 .alloc_inode = btrfs_alloc_inode,
1770 .destroy_inode = btrfs_destroy_inode,
1771 .statfs = btrfs_statfs,
1772 .remount_fs = btrfs_remount,
1773 .freeze_fs = btrfs_freeze,
1774 .unfreeze_fs = btrfs_unfreeze,
1777 static const struct file_operations btrfs_ctl_fops = {
1778 .unlocked_ioctl = btrfs_control_ioctl,
1779 .compat_ioctl = btrfs_control_ioctl,
1780 .owner = THIS_MODULE,
1781 .llseek = noop_llseek,
1784 static struct miscdevice btrfs_misc = {
1785 .minor = BTRFS_MINOR,
1786 .name = "btrfs-control",
1787 .fops = &btrfs_ctl_fops
1790 MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
1791 MODULE_ALIAS("devname:btrfs-control");
1793 static int btrfs_interface_init(void)
1795 return misc_register(&btrfs_misc);
1798 static void btrfs_interface_exit(void)
1800 if (misc_deregister(&btrfs_misc) < 0)
1801 printk(KERN_INFO "BTRFS: misc_deregister failed for control device\n");
1804 static void btrfs_print_info(void)
1806 printk(KERN_INFO "Btrfs loaded"
1807 #ifdef CONFIG_BTRFS_DEBUG
1810 #ifdef CONFIG_BTRFS_ASSERT
1813 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1814 ", integrity-checker=on"
1819 static int btrfs_run_sanity_tests(void)
1823 ret = btrfs_init_test_fs();
1827 ret = btrfs_test_free_space_cache();
1830 ret = btrfs_test_extent_buffer_operations();
1833 ret = btrfs_test_extent_io();
1836 ret = btrfs_test_inodes();
1838 btrfs_destroy_test_fs();
1842 static int __init init_btrfs_fs(void)
1846 err = btrfs_init_sysfs();
1850 btrfs_init_compress();
1852 err = btrfs_init_cachep();
1856 err = extent_io_init();
1860 err = extent_map_init();
1862 goto free_extent_io;
1864 err = ordered_data_init();
1866 goto free_extent_map;
1868 err = btrfs_delayed_inode_init();
1870 goto free_ordered_data;
1872 err = btrfs_auto_defrag_init();
1874 goto free_delayed_inode;
1876 err = btrfs_delayed_ref_init();
1878 goto free_auto_defrag;
1880 err = btrfs_prelim_ref_init();
1882 goto free_prelim_ref;
1884 err = btrfs_interface_init();
1886 goto free_delayed_ref;
1888 btrfs_init_lockdep();
1892 err = btrfs_run_sanity_tests();
1894 goto unregister_ioctl;
1896 err = register_filesystem(&btrfs_fs_type);
1898 goto unregister_ioctl;
1903 btrfs_interface_exit();
1905 btrfs_prelim_ref_exit();
1907 btrfs_delayed_ref_exit();
1909 btrfs_auto_defrag_exit();
1911 btrfs_delayed_inode_exit();
1913 ordered_data_exit();
1919 btrfs_destroy_cachep();
1921 btrfs_exit_compress();
1926 static void __exit exit_btrfs_fs(void)
1928 btrfs_destroy_cachep();
1929 btrfs_delayed_ref_exit();
1930 btrfs_auto_defrag_exit();
1931 btrfs_delayed_inode_exit();
1932 btrfs_prelim_ref_exit();
1933 ordered_data_exit();
1936 btrfs_interface_exit();
1937 unregister_filesystem(&btrfs_fs_type);
1939 btrfs_cleanup_fs_uuids();
1940 btrfs_exit_compress();
1943 module_init(init_btrfs_fs)
1944 module_exit(exit_btrfs_fs)
1946 MODULE_LICENSE("GPL");