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,
330 static match_table_t tokens = {
331 {Opt_degraded, "degraded"},
332 {Opt_subvol, "subvol=%s"},
333 {Opt_subvolid, "subvolid=%s"},
334 {Opt_device, "device=%s"},
335 {Opt_nodatasum, "nodatasum"},
336 {Opt_nodatacow, "nodatacow"},
337 {Opt_nobarrier, "nobarrier"},
338 {Opt_barrier, "barrier"},
339 {Opt_max_inline, "max_inline=%s"},
340 {Opt_alloc_start, "alloc_start=%s"},
341 {Opt_thread_pool, "thread_pool=%d"},
342 {Opt_compress, "compress"},
343 {Opt_compress_type, "compress=%s"},
344 {Opt_compress_force, "compress-force"},
345 {Opt_compress_force_type, "compress-force=%s"},
347 {Opt_ssd_spread, "ssd_spread"},
348 {Opt_nossd, "nossd"},
349 {Opt_noacl, "noacl"},
350 {Opt_notreelog, "notreelog"},
351 {Opt_flushoncommit, "flushoncommit"},
352 {Opt_ratio, "metadata_ratio=%d"},
353 {Opt_discard, "discard"},
354 {Opt_nodiscard, "nodiscard"},
355 {Opt_space_cache, "space_cache"},
356 {Opt_clear_cache, "clear_cache"},
357 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
358 {Opt_enospc_debug, "enospc_debug"},
359 {Opt_subvolrootid, "subvolrootid=%d"},
360 {Opt_defrag, "autodefrag"},
361 {Opt_nodefrag, "noautodefrag"},
362 {Opt_inode_cache, "inode_cache"},
363 {Opt_no_space_cache, "nospace_cache"},
364 {Opt_recovery, "recovery"},
365 {Opt_skip_balance, "skip_balance"},
366 {Opt_check_integrity, "check_int"},
367 {Opt_check_integrity_including_extent_data, "check_int_data"},
368 {Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
369 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
370 {Opt_fatal_errors, "fatal_errors=%s"},
371 {Opt_commit_interval, "commit=%d"},
376 * Regular mount options parser. Everything that is needed only when
377 * reading in a new superblock is parsed here.
378 * XXX JDM: This needs to be cleaned up for remount.
380 int btrfs_parse_options(struct btrfs_root *root, char *options)
382 struct btrfs_fs_info *info = root->fs_info;
383 substring_t args[MAX_OPT_ARGS];
384 char *p, *num, *orig = NULL;
389 bool compress_force = false;
391 cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
393 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
399 * strsep changes the string, duplicate it because parse_options
402 options = kstrdup(options, GFP_NOFS);
408 while ((p = strsep(&options, ",")) != NULL) {
413 token = match_token(p, tokens, args);
416 btrfs_info(root->fs_info, "allowing degraded mounts");
417 btrfs_set_opt(info->mount_opt, DEGRADED);
421 case Opt_subvolrootid:
424 * These are parsed by btrfs_parse_early_options
425 * and can be happily ignored here.
429 btrfs_info(root->fs_info, "setting nodatasum");
430 btrfs_set_opt(info->mount_opt, NODATASUM);
433 if (!btrfs_test_opt(root, COMPRESS) ||
434 !btrfs_test_opt(root, FORCE_COMPRESS)) {
435 btrfs_info(root->fs_info,
436 "setting nodatacow, compression disabled");
438 btrfs_info(root->fs_info, "setting nodatacow");
440 btrfs_clear_opt(info->mount_opt, COMPRESS);
441 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
442 btrfs_set_opt(info->mount_opt, NODATACOW);
443 btrfs_set_opt(info->mount_opt, NODATASUM);
445 case Opt_compress_force:
446 case Opt_compress_force_type:
447 compress_force = true;
450 case Opt_compress_type:
451 if (token == Opt_compress ||
452 token == Opt_compress_force ||
453 strcmp(args[0].from, "zlib") == 0) {
454 compress_type = "zlib";
455 info->compress_type = BTRFS_COMPRESS_ZLIB;
456 btrfs_set_opt(info->mount_opt, COMPRESS);
457 btrfs_clear_opt(info->mount_opt, NODATACOW);
458 btrfs_clear_opt(info->mount_opt, NODATASUM);
459 } else if (strcmp(args[0].from, "lzo") == 0) {
460 compress_type = "lzo";
461 info->compress_type = BTRFS_COMPRESS_LZO;
462 btrfs_set_opt(info->mount_opt, COMPRESS);
463 btrfs_clear_opt(info->mount_opt, NODATACOW);
464 btrfs_clear_opt(info->mount_opt, NODATASUM);
465 btrfs_set_fs_incompat(info, COMPRESS_LZO);
466 } else if (strncmp(args[0].from, "no", 2) == 0) {
467 compress_type = "no";
468 btrfs_clear_opt(info->mount_opt, COMPRESS);
469 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
470 compress_force = false;
476 if (compress_force) {
477 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
478 btrfs_info(root->fs_info, "force %s compression",
480 } else if (btrfs_test_opt(root, COMPRESS)) {
481 pr_info("btrfs: use %s compression\n",
486 btrfs_info(root->fs_info, "use ssd allocation scheme");
487 btrfs_set_opt(info->mount_opt, SSD);
490 btrfs_info(root->fs_info, "use spread ssd allocation scheme");
491 btrfs_set_opt(info->mount_opt, SSD);
492 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
495 btrfs_info(root->fs_info, "not using ssd allocation scheme");
496 btrfs_set_opt(info->mount_opt, NOSSD);
497 btrfs_clear_opt(info->mount_opt, SSD);
498 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
501 if (btrfs_test_opt(root, NOBARRIER))
502 btrfs_info(root->fs_info, "turning on barriers");
503 btrfs_clear_opt(info->mount_opt, NOBARRIER);
506 btrfs_info(root->fs_info, "turning off barriers");
507 btrfs_set_opt(info->mount_opt, NOBARRIER);
509 case Opt_thread_pool:
510 ret = match_int(&args[0], &intarg);
513 } else if (intarg > 0) {
514 info->thread_pool_size = intarg;
521 num = match_strdup(&args[0]);
523 info->max_inline = memparse(num, NULL);
526 if (info->max_inline) {
527 info->max_inline = max_t(u64,
531 btrfs_info(root->fs_info, "max_inline at %llu",
538 case Opt_alloc_start:
539 num = match_strdup(&args[0]);
541 mutex_lock(&info->chunk_mutex);
542 info->alloc_start = memparse(num, NULL);
543 mutex_unlock(&info->chunk_mutex);
545 btrfs_info(root->fs_info, "allocations start at %llu",
553 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
556 btrfs_info(root->fs_info, "disabling tree log");
557 btrfs_set_opt(info->mount_opt, NOTREELOG);
559 case Opt_flushoncommit:
560 btrfs_info(root->fs_info, "turning on flush-on-commit");
561 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
564 ret = match_int(&args[0], &intarg);
567 } else if (intarg >= 0) {
568 info->metadata_ratio = intarg;
569 btrfs_info(root->fs_info, "metadata ratio %d",
570 info->metadata_ratio);
577 btrfs_set_opt(info->mount_opt, DISCARD);
580 btrfs_clear_opt(info->mount_opt, DISCARD);
582 case Opt_space_cache:
583 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
585 case Opt_rescan_uuid_tree:
586 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
588 case Opt_no_space_cache:
589 btrfs_info(root->fs_info, "disabling disk space caching");
590 btrfs_clear_opt(info->mount_opt, SPACE_CACHE);
592 case Opt_inode_cache:
593 btrfs_info(root->fs_info, "enabling inode map caching");
594 btrfs_set_opt(info->mount_opt, INODE_MAP_CACHE);
596 case Opt_clear_cache:
597 btrfs_info(root->fs_info, "force clearing of disk cache");
598 btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
600 case Opt_user_subvol_rm_allowed:
601 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
603 case Opt_enospc_debug:
604 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
607 btrfs_info(root->fs_info, "enabling auto defrag");
608 btrfs_set_opt(info->mount_opt, AUTO_DEFRAG);
611 if (btrfs_test_opt(root, AUTO_DEFRAG))
612 btrfs_info(root->fs_info, "disabling auto defrag");
613 btrfs_clear_opt(info->mount_opt, AUTO_DEFRAG);
616 btrfs_info(root->fs_info, "enabling auto recovery");
617 btrfs_set_opt(info->mount_opt, RECOVERY);
619 case Opt_skip_balance:
620 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
622 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
623 case Opt_check_integrity_including_extent_data:
624 btrfs_info(root->fs_info,
625 "enabling check integrity including extent data");
626 btrfs_set_opt(info->mount_opt,
627 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
628 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
630 case Opt_check_integrity:
631 btrfs_info(root->fs_info, "enabling check integrity");
632 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
634 case Opt_check_integrity_print_mask:
635 ret = match_int(&args[0], &intarg);
638 } else if (intarg >= 0) {
639 info->check_integrity_print_mask = intarg;
640 btrfs_info(root->fs_info, "check_integrity_print_mask 0x%x",
641 info->check_integrity_print_mask);
648 case Opt_check_integrity_including_extent_data:
649 case Opt_check_integrity:
650 case Opt_check_integrity_print_mask:
651 btrfs_err(root->fs_info,
652 "support for check_integrity* not compiled in!");
656 case Opt_fatal_errors:
657 if (strcmp(args[0].from, "panic") == 0)
658 btrfs_set_opt(info->mount_opt,
659 PANIC_ON_FATAL_ERROR);
660 else if (strcmp(args[0].from, "bug") == 0)
661 btrfs_clear_opt(info->mount_opt,
662 PANIC_ON_FATAL_ERROR);
668 case Opt_commit_interval:
670 ret = match_int(&args[0], &intarg);
672 btrfs_err(root->fs_info, "invalid commit interval");
678 btrfs_warn(root->fs_info, "excessive commit interval %d",
681 info->commit_interval = intarg;
683 btrfs_info(root->fs_info, "using default commit interval %ds",
684 BTRFS_DEFAULT_COMMIT_INTERVAL);
685 info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
689 btrfs_info(root->fs_info, "unrecognized mount option '%s'", p);
697 if (!ret && btrfs_test_opt(root, SPACE_CACHE))
698 btrfs_info(root->fs_info, "disk space caching is enabled");
704 * Parse mount options that are required early in the mount process.
706 * All other options will be parsed on much later in the mount process and
707 * only when we need to allocate a new super block.
709 static int btrfs_parse_early_options(const char *options, fmode_t flags,
710 void *holder, char **subvol_name, u64 *subvol_objectid,
711 struct btrfs_fs_devices **fs_devices)
713 substring_t args[MAX_OPT_ARGS];
714 char *device_name, *opts, *orig, *p;
722 * strsep changes the string, duplicate it because parse_options
725 opts = kstrdup(options, GFP_KERNEL);
730 while ((p = strsep(&opts, ",")) != NULL) {
735 token = match_token(p, tokens, args);
739 *subvol_name = match_strdup(&args[0]);
746 num = match_strdup(&args[0]);
748 *subvol_objectid = memparse(num, NULL);
750 /* we want the original fs_tree */
751 if (!*subvol_objectid)
753 BTRFS_FS_TREE_OBJECTID;
759 case Opt_subvolrootid:
761 "BTRFS: 'subvolrootid' mount option is deprecated and has "
765 device_name = match_strdup(&args[0]);
770 error = btrfs_scan_one_device(device_name,
771 flags, holder, fs_devices);
786 static struct dentry *get_default_root(struct super_block *sb,
789 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
790 struct btrfs_root *root = fs_info->tree_root;
791 struct btrfs_root *new_root;
792 struct btrfs_dir_item *di;
793 struct btrfs_path *path;
794 struct btrfs_key location;
800 * We have a specific subvol we want to mount, just setup location and
801 * go look up the root.
803 if (subvol_objectid) {
804 location.objectid = subvol_objectid;
805 location.type = BTRFS_ROOT_ITEM_KEY;
806 location.offset = (u64)-1;
810 path = btrfs_alloc_path();
812 return ERR_PTR(-ENOMEM);
813 path->leave_spinning = 1;
816 * Find the "default" dir item which points to the root item that we
817 * will mount by default if we haven't been given a specific subvolume
820 dir_id = btrfs_super_root_dir(fs_info->super_copy);
821 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
823 btrfs_free_path(path);
828 * Ok the default dir item isn't there. This is weird since
829 * it's always been there, but don't freak out, just try and
830 * mount to root most subvolume.
832 btrfs_free_path(path);
833 dir_id = BTRFS_FIRST_FREE_OBJECTID;
834 new_root = fs_info->fs_root;
838 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
839 btrfs_free_path(path);
842 new_root = btrfs_read_fs_root_no_name(fs_info, &location);
843 if (IS_ERR(new_root))
844 return ERR_CAST(new_root);
846 dir_id = btrfs_root_dirid(&new_root->root_item);
848 location.objectid = dir_id;
849 location.type = BTRFS_INODE_ITEM_KEY;
852 inode = btrfs_iget(sb, &location, new_root, &new);
854 return ERR_CAST(inode);
857 * If we're just mounting the root most subvol put the inode and return
858 * a reference to the dentry. We will have already gotten a reference
859 * to the inode in btrfs_fill_super so we're good to go.
861 if (!new && sb->s_root->d_inode == inode) {
863 return dget(sb->s_root);
866 return d_obtain_alias(inode);
869 static int btrfs_fill_super(struct super_block *sb,
870 struct btrfs_fs_devices *fs_devices,
871 void *data, int silent)
874 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
875 struct btrfs_key key;
878 sb->s_maxbytes = MAX_LFS_FILESIZE;
879 sb->s_magic = BTRFS_SUPER_MAGIC;
880 sb->s_op = &btrfs_super_ops;
881 sb->s_d_op = &btrfs_dentry_operations;
882 sb->s_export_op = &btrfs_export_ops;
883 sb->s_xattr = btrfs_xattr_handlers;
885 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
886 sb->s_flags |= MS_POSIXACL;
888 sb->s_flags |= MS_I_VERSION;
889 err = open_ctree(sb, fs_devices, (char *)data);
891 printk(KERN_ERR "BTRFS: open_ctree failed\n");
895 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
896 key.type = BTRFS_INODE_ITEM_KEY;
898 inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL);
900 err = PTR_ERR(inode);
904 sb->s_root = d_make_root(inode);
910 save_mount_options(sb, data);
911 cleancache_init_fs(sb);
912 sb->s_flags |= MS_ACTIVE;
916 close_ctree(fs_info->tree_root);
920 int btrfs_sync_fs(struct super_block *sb, int wait)
922 struct btrfs_trans_handle *trans;
923 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
924 struct btrfs_root *root = fs_info->tree_root;
926 trace_btrfs_sync_fs(wait);
929 filemap_flush(fs_info->btree_inode->i_mapping);
933 btrfs_wait_ordered_roots(fs_info, -1);
935 trans = btrfs_attach_transaction_barrier(root);
937 /* no transaction, don't bother */
938 if (PTR_ERR(trans) == -ENOENT)
940 return PTR_ERR(trans);
942 return btrfs_commit_transaction(trans, root);
945 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
947 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
948 struct btrfs_root *root = info->tree_root;
951 if (btrfs_test_opt(root, DEGRADED))
952 seq_puts(seq, ",degraded");
953 if (btrfs_test_opt(root, NODATASUM))
954 seq_puts(seq, ",nodatasum");
955 if (btrfs_test_opt(root, NODATACOW))
956 seq_puts(seq, ",nodatacow");
957 if (btrfs_test_opt(root, NOBARRIER))
958 seq_puts(seq, ",nobarrier");
959 if (info->max_inline != 8192 * 1024)
960 seq_printf(seq, ",max_inline=%llu", info->max_inline);
961 if (info->alloc_start != 0)
962 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
963 if (info->thread_pool_size != min_t(unsigned long,
964 num_online_cpus() + 2, 8))
965 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
966 if (btrfs_test_opt(root, COMPRESS)) {
967 if (info->compress_type == BTRFS_COMPRESS_ZLIB)
968 compress_type = "zlib";
970 compress_type = "lzo";
971 if (btrfs_test_opt(root, FORCE_COMPRESS))
972 seq_printf(seq, ",compress-force=%s", compress_type);
974 seq_printf(seq, ",compress=%s", compress_type);
976 if (btrfs_test_opt(root, NOSSD))
977 seq_puts(seq, ",nossd");
978 if (btrfs_test_opt(root, SSD_SPREAD))
979 seq_puts(seq, ",ssd_spread");
980 else if (btrfs_test_opt(root, SSD))
981 seq_puts(seq, ",ssd");
982 if (btrfs_test_opt(root, NOTREELOG))
983 seq_puts(seq, ",notreelog");
984 if (btrfs_test_opt(root, FLUSHONCOMMIT))
985 seq_puts(seq, ",flushoncommit");
986 if (btrfs_test_opt(root, DISCARD))
987 seq_puts(seq, ",discard");
988 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
989 seq_puts(seq, ",noacl");
990 if (btrfs_test_opt(root, SPACE_CACHE))
991 seq_puts(seq, ",space_cache");
993 seq_puts(seq, ",nospace_cache");
994 if (btrfs_test_opt(root, RESCAN_UUID_TREE))
995 seq_puts(seq, ",rescan_uuid_tree");
996 if (btrfs_test_opt(root, CLEAR_CACHE))
997 seq_puts(seq, ",clear_cache");
998 if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
999 seq_puts(seq, ",user_subvol_rm_allowed");
1000 if (btrfs_test_opt(root, ENOSPC_DEBUG))
1001 seq_puts(seq, ",enospc_debug");
1002 if (btrfs_test_opt(root, AUTO_DEFRAG))
1003 seq_puts(seq, ",autodefrag");
1004 if (btrfs_test_opt(root, INODE_MAP_CACHE))
1005 seq_puts(seq, ",inode_cache");
1006 if (btrfs_test_opt(root, SKIP_BALANCE))
1007 seq_puts(seq, ",skip_balance");
1008 if (btrfs_test_opt(root, RECOVERY))
1009 seq_puts(seq, ",recovery");
1010 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1011 if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1012 seq_puts(seq, ",check_int_data");
1013 else if (btrfs_test_opt(root, CHECK_INTEGRITY))
1014 seq_puts(seq, ",check_int");
1015 if (info->check_integrity_print_mask)
1016 seq_printf(seq, ",check_int_print_mask=%d",
1017 info->check_integrity_print_mask);
1019 if (info->metadata_ratio)
1020 seq_printf(seq, ",metadata_ratio=%d",
1021 info->metadata_ratio);
1022 if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
1023 seq_puts(seq, ",fatal_errors=panic");
1024 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1025 seq_printf(seq, ",commit=%d", info->commit_interval);
1029 static int btrfs_test_super(struct super_block *s, void *data)
1031 struct btrfs_fs_info *p = data;
1032 struct btrfs_fs_info *fs_info = btrfs_sb(s);
1034 return fs_info->fs_devices == p->fs_devices;
1037 static int btrfs_set_super(struct super_block *s, void *data)
1039 int err = set_anon_super(s, data);
1041 s->s_fs_info = data;
1046 * subvolumes are identified by ino 256
1048 static inline int is_subvolume_inode(struct inode *inode)
1050 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1056 * This will strip out the subvol=%s argument for an argument string and add
1057 * subvolid=0 to make sure we get the actual tree root for path walking to the
1060 static char *setup_root_args(char *args)
1062 unsigned len = strlen(args) + 2 + 1;
1063 char *src, *dst, *buf;
1066 * We need the same args as before, but with this substitution:
1067 * s!subvol=[^,]+!subvolid=0!
1069 * Since the replacement string is up to 2 bytes longer than the
1070 * original, allocate strlen(args) + 2 + 1 bytes.
1073 src = strstr(args, "subvol=");
1074 /* This shouldn't happen, but just in case.. */
1078 buf = dst = kmalloc(len, GFP_NOFS);
1083 * If the subvol= arg is not at the start of the string,
1084 * copy whatever precedes it into buf.
1089 dst += strlen(args);
1092 strcpy(dst, "subvolid=0");
1093 dst += strlen("subvolid=0");
1096 * If there is a "," after the original subvol=... string,
1097 * copy that suffix into our buffer. Otherwise, we're done.
1099 src = strchr(src, ',');
1106 static struct dentry *mount_subvol(const char *subvol_name, int flags,
1107 const char *device_name, char *data)
1109 struct dentry *root;
1110 struct vfsmount *mnt;
1113 newargs = setup_root_args(data);
1115 return ERR_PTR(-ENOMEM);
1116 mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name,
1120 return ERR_CAST(mnt);
1122 root = mount_subtree(mnt, subvol_name);
1124 if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) {
1125 struct super_block *s = root->d_sb;
1127 root = ERR_PTR(-EINVAL);
1128 deactivate_locked_super(s);
1129 printk(KERN_ERR "BTRFS: '%s' is not a valid subvolume\n",
1137 * Find a superblock for the given device / mount point.
1139 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1140 * for multiple device setup. Make sure to keep it in sync.
1142 static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
1143 const char *device_name, void *data)
1145 struct block_device *bdev = NULL;
1146 struct super_block *s;
1147 struct dentry *root;
1148 struct btrfs_fs_devices *fs_devices = NULL;
1149 struct btrfs_fs_info *fs_info = NULL;
1150 fmode_t mode = FMODE_READ;
1151 char *subvol_name = NULL;
1152 u64 subvol_objectid = 0;
1155 if (!(flags & MS_RDONLY))
1156 mode |= FMODE_WRITE;
1158 error = btrfs_parse_early_options(data, mode, fs_type,
1159 &subvol_name, &subvol_objectid,
1163 return ERR_PTR(error);
1167 root = mount_subvol(subvol_name, flags, device_name, data);
1172 error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices);
1174 return ERR_PTR(error);
1177 * Setup a dummy root and fs_info for test/set super. This is because
1178 * we don't actually fill this stuff out until open_ctree, but we need
1179 * it for searching for existing supers, so this lets us do that and
1180 * then open_ctree will properly initialize everything later.
1182 fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
1184 return ERR_PTR(-ENOMEM);
1186 fs_info->fs_devices = fs_devices;
1188 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1189 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1190 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1195 error = btrfs_open_devices(fs_devices, mode, fs_type);
1199 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
1201 goto error_close_devices;
1204 bdev = fs_devices->latest_bdev;
1205 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC,
1209 goto error_close_devices;
1213 btrfs_close_devices(fs_devices);
1214 free_fs_info(fs_info);
1215 if ((flags ^ s->s_flags) & MS_RDONLY)
1218 char b[BDEVNAME_SIZE];
1220 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
1221 btrfs_sb(s)->bdev_holder = fs_type;
1222 error = btrfs_fill_super(s, fs_devices, data,
1223 flags & MS_SILENT ? 1 : 0);
1226 root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error);
1228 deactivate_locked_super(s);
1232 error_close_devices:
1233 btrfs_close_devices(fs_devices);
1235 free_fs_info(fs_info);
1236 return ERR_PTR(error);
1239 static void btrfs_set_max_workers(struct btrfs_workers *workers, int new_limit)
1241 spin_lock_irq(&workers->lock);
1242 workers->max_workers = new_limit;
1243 spin_unlock_irq(&workers->lock);
1246 static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1247 int new_pool_size, int old_pool_size)
1249 if (new_pool_size == old_pool_size)
1252 fs_info->thread_pool_size = new_pool_size;
1254 btrfs_info(fs_info, "resize thread pool %d -> %d",
1255 old_pool_size, new_pool_size);
1257 btrfs_set_max_workers(&fs_info->generic_worker, new_pool_size);
1258 btrfs_set_max_workers(&fs_info->workers, new_pool_size);
1259 btrfs_set_max_workers(&fs_info->delalloc_workers, new_pool_size);
1260 btrfs_set_max_workers(&fs_info->submit_workers, new_pool_size);
1261 btrfs_set_max_workers(&fs_info->caching_workers, new_pool_size);
1262 btrfs_set_max_workers(&fs_info->fixup_workers, new_pool_size);
1263 btrfs_set_max_workers(&fs_info->endio_workers, new_pool_size);
1264 btrfs_set_max_workers(&fs_info->endio_meta_workers, new_pool_size);
1265 btrfs_set_max_workers(&fs_info->endio_meta_write_workers, new_pool_size);
1266 btrfs_set_max_workers(&fs_info->endio_write_workers, new_pool_size);
1267 btrfs_set_max_workers(&fs_info->endio_freespace_worker, new_pool_size);
1268 btrfs_set_max_workers(&fs_info->delayed_workers, new_pool_size);
1269 btrfs_set_max_workers(&fs_info->readahead_workers, new_pool_size);
1270 btrfs_set_max_workers(&fs_info->scrub_wr_completion_workers,
1274 static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
1276 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1279 static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1280 unsigned long old_opts, int flags)
1282 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1283 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1284 (flags & MS_RDONLY))) {
1285 /* wait for any defraggers to finish */
1286 wait_event(fs_info->transaction_wait,
1287 (atomic_read(&fs_info->defrag_running) == 0));
1288 if (flags & MS_RDONLY)
1289 sync_filesystem(fs_info->sb);
1293 static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1294 unsigned long old_opts)
1297 * We need cleanup all defragable inodes if the autodefragment is
1298 * close or the fs is R/O.
1300 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1301 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1302 (fs_info->sb->s_flags & MS_RDONLY))) {
1303 btrfs_cleanup_defrag_inodes(fs_info);
1306 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1309 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1311 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1312 struct btrfs_root *root = fs_info->tree_root;
1313 unsigned old_flags = sb->s_flags;
1314 unsigned long old_opts = fs_info->mount_opt;
1315 unsigned long old_compress_type = fs_info->compress_type;
1316 u64 old_max_inline = fs_info->max_inline;
1317 u64 old_alloc_start = fs_info->alloc_start;
1318 int old_thread_pool_size = fs_info->thread_pool_size;
1319 unsigned int old_metadata_ratio = fs_info->metadata_ratio;
1322 btrfs_remount_prepare(fs_info);
1324 ret = btrfs_parse_options(root, data);
1330 btrfs_remount_begin(fs_info, old_opts, *flags);
1331 btrfs_resize_thread_pool(fs_info,
1332 fs_info->thread_pool_size, old_thread_pool_size);
1334 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
1337 if (*flags & MS_RDONLY) {
1339 * this also happens on 'umount -rf' or on shutdown, when
1340 * the filesystem is busy.
1343 /* wait for the uuid_scan task to finish */
1344 down(&fs_info->uuid_tree_rescan_sem);
1345 /* avoid complains from lockdep et al. */
1346 up(&fs_info->uuid_tree_rescan_sem);
1348 sb->s_flags |= MS_RDONLY;
1350 btrfs_dev_replace_suspend_for_unmount(fs_info);
1351 btrfs_scrub_cancel(fs_info);
1352 btrfs_pause_balance(fs_info);
1354 ret = btrfs_commit_super(root);
1358 if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) {
1360 "Remounting read-write after error is not allowed");
1364 if (fs_info->fs_devices->rw_devices == 0) {
1369 if (fs_info->fs_devices->missing_devices >
1370 fs_info->num_tolerated_disk_barrier_failures &&
1371 !(*flags & MS_RDONLY)) {
1373 "too many missing devices, writeable remount is not allowed");
1378 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
1383 ret = btrfs_cleanup_fs_roots(fs_info);
1387 /* recover relocation */
1388 ret = btrfs_recover_relocation(root);
1392 ret = btrfs_resume_balance_async(fs_info);
1396 ret = btrfs_resume_dev_replace_async(fs_info);
1398 btrfs_warn(fs_info, "failed to resume dev_replace");
1402 if (!fs_info->uuid_root) {
1403 btrfs_info(fs_info, "creating UUID tree");
1404 ret = btrfs_create_uuid_tree(fs_info);
1406 btrfs_warn(fs_info, "failed to create the UUID tree %d", ret);
1410 sb->s_flags &= ~MS_RDONLY;
1413 btrfs_remount_cleanup(fs_info, old_opts);
1417 /* We've hit an error - don't reset MS_RDONLY */
1418 if (sb->s_flags & MS_RDONLY)
1419 old_flags |= MS_RDONLY;
1420 sb->s_flags = old_flags;
1421 fs_info->mount_opt = old_opts;
1422 fs_info->compress_type = old_compress_type;
1423 fs_info->max_inline = old_max_inline;
1424 mutex_lock(&fs_info->chunk_mutex);
1425 fs_info->alloc_start = old_alloc_start;
1426 mutex_unlock(&fs_info->chunk_mutex);
1427 btrfs_resize_thread_pool(fs_info,
1428 old_thread_pool_size, fs_info->thread_pool_size);
1429 fs_info->metadata_ratio = old_metadata_ratio;
1430 btrfs_remount_cleanup(fs_info, old_opts);
1434 /* Used to sort the devices by max_avail(descending sort) */
1435 static int btrfs_cmp_device_free_bytes(const void *dev_info1,
1436 const void *dev_info2)
1438 if (((struct btrfs_device_info *)dev_info1)->max_avail >
1439 ((struct btrfs_device_info *)dev_info2)->max_avail)
1441 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
1442 ((struct btrfs_device_info *)dev_info2)->max_avail)
1449 * sort the devices by max_avail, in which max free extent size of each device
1450 * is stored.(Descending Sort)
1452 static inline void btrfs_descending_sort_devices(
1453 struct btrfs_device_info *devices,
1456 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
1457 btrfs_cmp_device_free_bytes, NULL);
1461 * The helper to calc the free space on the devices that can be used to store
1464 static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
1466 struct btrfs_fs_info *fs_info = root->fs_info;
1467 struct btrfs_device_info *devices_info;
1468 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
1469 struct btrfs_device *device;
1474 u64 min_stripe_size;
1475 int min_stripes = 1, num_stripes = 1;
1476 int i = 0, nr_devices;
1479 nr_devices = fs_info->fs_devices->open_devices;
1480 BUG_ON(!nr_devices);
1482 devices_info = kmalloc_array(nr_devices, sizeof(*devices_info),
1487 /* calc min stripe number for data space alloction */
1488 type = btrfs_get_alloc_profile(root, 1);
1489 if (type & BTRFS_BLOCK_GROUP_RAID0) {
1491 num_stripes = nr_devices;
1492 } else if (type & BTRFS_BLOCK_GROUP_RAID1) {
1495 } else if (type & BTRFS_BLOCK_GROUP_RAID10) {
1500 if (type & BTRFS_BLOCK_GROUP_DUP)
1501 min_stripe_size = 2 * BTRFS_STRIPE_LEN;
1503 min_stripe_size = BTRFS_STRIPE_LEN;
1505 list_for_each_entry(device, &fs_devices->devices, dev_list) {
1506 if (!device->in_fs_metadata || !device->bdev ||
1507 device->is_tgtdev_for_dev_replace)
1510 avail_space = device->total_bytes - device->bytes_used;
1512 /* align with stripe_len */
1513 do_div(avail_space, BTRFS_STRIPE_LEN);
1514 avail_space *= BTRFS_STRIPE_LEN;
1517 * In order to avoid overwritting the superblock on the drive,
1518 * btrfs starts at an offset of at least 1MB when doing chunk
1521 skip_space = 1024 * 1024;
1523 /* user can set the offset in fs_info->alloc_start. */
1524 if (fs_info->alloc_start + BTRFS_STRIPE_LEN <=
1525 device->total_bytes)
1526 skip_space = max(fs_info->alloc_start, skip_space);
1529 * btrfs can not use the free space in [0, skip_space - 1],
1530 * we must subtract it from the total. In order to implement
1531 * it, we account the used space in this range first.
1533 ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1,
1536 kfree(devices_info);
1540 /* calc the free space in [0, skip_space - 1] */
1541 skip_space -= used_space;
1544 * we can use the free space in [0, skip_space - 1], subtract
1545 * it from the total.
1547 if (avail_space && avail_space >= skip_space)
1548 avail_space -= skip_space;
1552 if (avail_space < min_stripe_size)
1555 devices_info[i].dev = device;
1556 devices_info[i].max_avail = avail_space;
1563 btrfs_descending_sort_devices(devices_info, nr_devices);
1567 while (nr_devices >= min_stripes) {
1568 if (num_stripes > nr_devices)
1569 num_stripes = nr_devices;
1571 if (devices_info[i].max_avail >= min_stripe_size) {
1575 avail_space += devices_info[i].max_avail * num_stripes;
1576 alloc_size = devices_info[i].max_avail;
1577 for (j = i + 1 - num_stripes; j <= i; j++)
1578 devices_info[j].max_avail -= alloc_size;
1584 kfree(devices_info);
1585 *free_bytes = avail_space;
1589 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
1591 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
1592 struct btrfs_super_block *disk_super = fs_info->super_copy;
1593 struct list_head *head = &fs_info->space_info;
1594 struct btrfs_space_info *found;
1596 u64 total_free_data = 0;
1597 int bits = dentry->d_sb->s_blocksize_bits;
1598 __be32 *fsid = (__be32 *)fs_info->fsid;
1601 /* holding chunk_muext to avoid allocating new chunks */
1602 mutex_lock(&fs_info->chunk_mutex);
1604 list_for_each_entry_rcu(found, head, list) {
1605 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
1606 total_free_data += found->disk_total - found->disk_used;
1608 btrfs_account_ro_block_groups_free_space(found);
1611 total_used += found->disk_used;
1615 buf->f_namelen = BTRFS_NAME_LEN;
1616 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
1617 buf->f_bfree = buf->f_blocks - (total_used >> bits);
1618 buf->f_bsize = dentry->d_sb->s_blocksize;
1619 buf->f_type = BTRFS_SUPER_MAGIC;
1620 buf->f_bavail = total_free_data;
1621 ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data);
1623 mutex_unlock(&fs_info->chunk_mutex);
1626 buf->f_bavail += total_free_data;
1627 buf->f_bavail = buf->f_bavail >> bits;
1628 mutex_unlock(&fs_info->chunk_mutex);
1630 /* We treat it as constant endianness (it doesn't matter _which_)
1631 because we want the fsid to come out the same whether mounted
1632 on a big-endian or little-endian host */
1633 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
1634 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
1635 /* Mask in the root object ID too, to disambiguate subvols */
1636 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
1637 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
1642 static void btrfs_kill_super(struct super_block *sb)
1644 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1645 kill_anon_super(sb);
1646 free_fs_info(fs_info);
1649 static struct file_system_type btrfs_fs_type = {
1650 .owner = THIS_MODULE,
1652 .mount = btrfs_mount,
1653 .kill_sb = btrfs_kill_super,
1654 .fs_flags = FS_REQUIRES_DEV,
1656 MODULE_ALIAS_FS("btrfs");
1659 * used by btrfsctl to scan devices when no FS is mounted
1661 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
1664 struct btrfs_ioctl_vol_args *vol;
1665 struct btrfs_fs_devices *fs_devices;
1668 if (!capable(CAP_SYS_ADMIN))
1671 vol = memdup_user((void __user *)arg, sizeof(*vol));
1673 return PTR_ERR(vol);
1676 case BTRFS_IOC_SCAN_DEV:
1677 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1678 &btrfs_fs_type, &fs_devices);
1680 case BTRFS_IOC_DEVICES_READY:
1681 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1682 &btrfs_fs_type, &fs_devices);
1685 ret = !(fs_devices->num_devices == fs_devices->total_devices);
1693 static int btrfs_freeze(struct super_block *sb)
1695 struct btrfs_trans_handle *trans;
1696 struct btrfs_root *root = btrfs_sb(sb)->tree_root;
1698 trans = btrfs_attach_transaction_barrier(root);
1699 if (IS_ERR(trans)) {
1700 /* no transaction, don't bother */
1701 if (PTR_ERR(trans) == -ENOENT)
1703 return PTR_ERR(trans);
1705 return btrfs_commit_transaction(trans, root);
1708 static int btrfs_unfreeze(struct super_block *sb)
1713 static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
1715 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
1716 struct btrfs_fs_devices *cur_devices;
1717 struct btrfs_device *dev, *first_dev = NULL;
1718 struct list_head *head;
1719 struct rcu_string *name;
1721 mutex_lock(&fs_info->fs_devices->device_list_mutex);
1722 cur_devices = fs_info->fs_devices;
1723 while (cur_devices) {
1724 head = &cur_devices->devices;
1725 list_for_each_entry(dev, head, dev_list) {
1728 if (!first_dev || dev->devid < first_dev->devid)
1731 cur_devices = cur_devices->seed;
1736 name = rcu_dereference(first_dev->name);
1737 seq_escape(m, name->str, " \t\n\\");
1742 mutex_unlock(&fs_info->fs_devices->device_list_mutex);
1746 static const struct super_operations btrfs_super_ops = {
1747 .drop_inode = btrfs_drop_inode,
1748 .evict_inode = btrfs_evict_inode,
1749 .put_super = btrfs_put_super,
1750 .sync_fs = btrfs_sync_fs,
1751 .show_options = btrfs_show_options,
1752 .show_devname = btrfs_show_devname,
1753 .write_inode = btrfs_write_inode,
1754 .alloc_inode = btrfs_alloc_inode,
1755 .destroy_inode = btrfs_destroy_inode,
1756 .statfs = btrfs_statfs,
1757 .remount_fs = btrfs_remount,
1758 .freeze_fs = btrfs_freeze,
1759 .unfreeze_fs = btrfs_unfreeze,
1762 static const struct file_operations btrfs_ctl_fops = {
1763 .unlocked_ioctl = btrfs_control_ioctl,
1764 .compat_ioctl = btrfs_control_ioctl,
1765 .owner = THIS_MODULE,
1766 .llseek = noop_llseek,
1769 static struct miscdevice btrfs_misc = {
1770 .minor = BTRFS_MINOR,
1771 .name = "btrfs-control",
1772 .fops = &btrfs_ctl_fops
1775 MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
1776 MODULE_ALIAS("devname:btrfs-control");
1778 static int btrfs_interface_init(void)
1780 return misc_register(&btrfs_misc);
1783 static void btrfs_interface_exit(void)
1785 if (misc_deregister(&btrfs_misc) < 0)
1786 printk(KERN_INFO "BTRFS: misc_deregister failed for control device\n");
1789 static void btrfs_print_info(void)
1791 printk(KERN_INFO "Btrfs loaded"
1792 #ifdef CONFIG_BTRFS_DEBUG
1795 #ifdef CONFIG_BTRFS_ASSERT
1798 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1799 ", integrity-checker=on"
1804 static int btrfs_run_sanity_tests(void)
1808 ret = btrfs_init_test_fs();
1812 ret = btrfs_test_free_space_cache();
1815 ret = btrfs_test_extent_buffer_operations();
1818 ret = btrfs_test_extent_io();
1821 ret = btrfs_test_inodes();
1823 btrfs_destroy_test_fs();
1827 static int __init init_btrfs_fs(void)
1831 err = btrfs_init_sysfs();
1835 btrfs_init_compress();
1837 err = btrfs_init_cachep();
1841 err = extent_io_init();
1845 err = extent_map_init();
1847 goto free_extent_io;
1849 err = ordered_data_init();
1851 goto free_extent_map;
1853 err = btrfs_delayed_inode_init();
1855 goto free_ordered_data;
1857 err = btrfs_auto_defrag_init();
1859 goto free_delayed_inode;
1861 err = btrfs_delayed_ref_init();
1863 goto free_auto_defrag;
1865 err = btrfs_prelim_ref_init();
1867 goto free_prelim_ref;
1869 err = btrfs_interface_init();
1871 goto free_delayed_ref;
1873 btrfs_init_lockdep();
1877 err = btrfs_run_sanity_tests();
1879 goto unregister_ioctl;
1881 err = register_filesystem(&btrfs_fs_type);
1883 goto unregister_ioctl;
1888 btrfs_interface_exit();
1890 btrfs_prelim_ref_exit();
1892 btrfs_delayed_ref_exit();
1894 btrfs_auto_defrag_exit();
1896 btrfs_delayed_inode_exit();
1898 ordered_data_exit();
1904 btrfs_destroy_cachep();
1906 btrfs_exit_compress();
1911 static void __exit exit_btrfs_fs(void)
1913 btrfs_destroy_cachep();
1914 btrfs_delayed_ref_exit();
1915 btrfs_auto_defrag_exit();
1916 btrfs_delayed_inode_exit();
1917 btrfs_prelim_ref_exit();
1918 ordered_data_exit();
1921 btrfs_interface_exit();
1922 unregister_filesystem(&btrfs_fs_type);
1924 btrfs_cleanup_fs_uuids();
1925 btrfs_exit_compress();
1928 module_init(init_btrfs_fs)
1929 module_exit(exit_btrfs_fs)
1931 MODULE_LICENSE("GPL");