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, Opt_datacow,
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_datacow, "datacow"},
339 {Opt_nobarrier, "nobarrier"},
340 {Opt_barrier, "barrier"},
341 {Opt_max_inline, "max_inline=%s"},
342 {Opt_alloc_start, "alloc_start=%s"},
343 {Opt_thread_pool, "thread_pool=%d"},
344 {Opt_compress, "compress"},
345 {Opt_compress_type, "compress=%s"},
346 {Opt_compress_force, "compress-force"},
347 {Opt_compress_force_type, "compress-force=%s"},
349 {Opt_ssd_spread, "ssd_spread"},
350 {Opt_nossd, "nossd"},
352 {Opt_noacl, "noacl"},
353 {Opt_notreelog, "notreelog"},
354 {Opt_flushoncommit, "flushoncommit"},
355 {Opt_noflushoncommit, "noflushoncommit"},
356 {Opt_ratio, "metadata_ratio=%d"},
357 {Opt_discard, "discard"},
358 {Opt_nodiscard, "nodiscard"},
359 {Opt_space_cache, "space_cache"},
360 {Opt_clear_cache, "clear_cache"},
361 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
362 {Opt_enospc_debug, "enospc_debug"},
363 {Opt_noenospc_debug, "noenospc_debug"},
364 {Opt_subvolrootid, "subvolrootid=%d"},
365 {Opt_defrag, "autodefrag"},
366 {Opt_nodefrag, "noautodefrag"},
367 {Opt_inode_cache, "inode_cache"},
368 {Opt_no_space_cache, "nospace_cache"},
369 {Opt_recovery, "recovery"},
370 {Opt_skip_balance, "skip_balance"},
371 {Opt_check_integrity, "check_int"},
372 {Opt_check_integrity_including_extent_data, "check_int_data"},
373 {Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
374 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
375 {Opt_fatal_errors, "fatal_errors=%s"},
376 {Opt_commit_interval, "commit=%d"},
381 * Regular mount options parser. Everything that is needed only when
382 * reading in a new superblock is parsed here.
383 * XXX JDM: This needs to be cleaned up for remount.
385 int btrfs_parse_options(struct btrfs_root *root, char *options)
387 struct btrfs_fs_info *info = root->fs_info;
388 substring_t args[MAX_OPT_ARGS];
389 char *p, *num, *orig = NULL;
394 bool compress_force = false;
396 cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
398 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
404 * strsep changes the string, duplicate it because parse_options
407 options = kstrdup(options, GFP_NOFS);
413 while ((p = strsep(&options, ",")) != NULL) {
418 token = match_token(p, tokens, args);
421 btrfs_info(root->fs_info, "allowing degraded mounts");
422 btrfs_set_opt(info->mount_opt, DEGRADED);
426 case Opt_subvolrootid:
429 * These are parsed by btrfs_parse_early_options
430 * and can be happily ignored here.
434 btrfs_info(root->fs_info, "setting nodatasum");
435 btrfs_set_opt(info->mount_opt, NODATASUM);
438 if (!btrfs_test_opt(root, COMPRESS) ||
439 !btrfs_test_opt(root, FORCE_COMPRESS)) {
440 btrfs_info(root->fs_info,
441 "setting nodatacow, compression disabled");
443 btrfs_info(root->fs_info, "setting nodatacow");
445 btrfs_clear_opt(info->mount_opt, COMPRESS);
446 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
447 btrfs_set_opt(info->mount_opt, NODATACOW);
448 btrfs_set_opt(info->mount_opt, NODATASUM);
451 if (btrfs_test_opt(root, NODATACOW))
452 btrfs_info(root->fs_info, "setting datacow");
453 btrfs_clear_opt(info->mount_opt, NODATACOW);
455 case Opt_compress_force:
456 case Opt_compress_force_type:
457 compress_force = true;
460 case Opt_compress_type:
461 if (token == Opt_compress ||
462 token == Opt_compress_force ||
463 strcmp(args[0].from, "zlib") == 0) {
464 compress_type = "zlib";
465 info->compress_type = BTRFS_COMPRESS_ZLIB;
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 } else if (strcmp(args[0].from, "lzo") == 0) {
470 compress_type = "lzo";
471 info->compress_type = BTRFS_COMPRESS_LZO;
472 btrfs_set_opt(info->mount_opt, COMPRESS);
473 btrfs_clear_opt(info->mount_opt, NODATACOW);
474 btrfs_clear_opt(info->mount_opt, NODATASUM);
475 btrfs_set_fs_incompat(info, COMPRESS_LZO);
476 } else if (strncmp(args[0].from, "no", 2) == 0) {
477 compress_type = "no";
478 btrfs_clear_opt(info->mount_opt, COMPRESS);
479 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
480 compress_force = false;
486 if (compress_force) {
487 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
488 btrfs_info(root->fs_info, "force %s compression",
490 } else if (btrfs_test_opt(root, COMPRESS)) {
491 pr_info("btrfs: use %s compression\n",
496 btrfs_info(root->fs_info, "use ssd allocation scheme");
497 btrfs_set_opt(info->mount_opt, SSD);
500 btrfs_info(root->fs_info, "use spread ssd allocation scheme");
501 btrfs_set_opt(info->mount_opt, SSD);
502 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
505 btrfs_info(root->fs_info, "not using ssd allocation scheme");
506 btrfs_set_opt(info->mount_opt, NOSSD);
507 btrfs_clear_opt(info->mount_opt, SSD);
508 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
511 if (btrfs_test_opt(root, NOBARRIER))
512 btrfs_info(root->fs_info, "turning on barriers");
513 btrfs_clear_opt(info->mount_opt, NOBARRIER);
516 btrfs_info(root->fs_info, "turning off barriers");
517 btrfs_set_opt(info->mount_opt, NOBARRIER);
519 case Opt_thread_pool:
520 ret = match_int(&args[0], &intarg);
523 } else if (intarg > 0) {
524 info->thread_pool_size = intarg;
531 num = match_strdup(&args[0]);
533 info->max_inline = memparse(num, NULL);
536 if (info->max_inline) {
537 info->max_inline = max_t(u64,
541 btrfs_info(root->fs_info, "max_inline at %llu",
548 case Opt_alloc_start:
549 num = match_strdup(&args[0]);
551 mutex_lock(&info->chunk_mutex);
552 info->alloc_start = memparse(num, NULL);
553 mutex_unlock(&info->chunk_mutex);
555 btrfs_info(root->fs_info, "allocations start at %llu",
563 root->fs_info->sb->s_flags |= MS_POSIXACL;
566 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
569 btrfs_info(root->fs_info, "disabling tree log");
570 btrfs_set_opt(info->mount_opt, NOTREELOG);
572 case Opt_flushoncommit:
573 btrfs_info(root->fs_info, "turning on flush-on-commit");
574 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
576 case Opt_noflushoncommit:
577 if (btrfs_test_opt(root, FLUSHONCOMMIT))
578 btrfs_info(root->fs_info, "turning off flush-on-commit");
579 btrfs_clear_opt(info->mount_opt, FLUSHONCOMMIT);
582 ret = match_int(&args[0], &intarg);
585 } else if (intarg >= 0) {
586 info->metadata_ratio = intarg;
587 btrfs_info(root->fs_info, "metadata ratio %d",
588 info->metadata_ratio);
595 btrfs_set_opt(info->mount_opt, DISCARD);
598 btrfs_clear_opt(info->mount_opt, DISCARD);
600 case Opt_space_cache:
601 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
603 case Opt_rescan_uuid_tree:
604 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
606 case Opt_no_space_cache:
607 btrfs_info(root->fs_info, "disabling disk space caching");
608 btrfs_clear_opt(info->mount_opt, SPACE_CACHE);
610 case Opt_inode_cache:
611 btrfs_info(root->fs_info, "enabling inode map caching");
612 btrfs_set_opt(info->mount_opt, INODE_MAP_CACHE);
614 case Opt_clear_cache:
615 btrfs_info(root->fs_info, "force clearing of disk cache");
616 btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
618 case Opt_user_subvol_rm_allowed:
619 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
621 case Opt_enospc_debug:
622 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
624 case Opt_noenospc_debug:
625 btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG);
628 btrfs_info(root->fs_info, "enabling auto defrag");
629 btrfs_set_opt(info->mount_opt, AUTO_DEFRAG);
632 if (btrfs_test_opt(root, AUTO_DEFRAG))
633 btrfs_info(root->fs_info, "disabling auto defrag");
634 btrfs_clear_opt(info->mount_opt, AUTO_DEFRAG);
637 btrfs_info(root->fs_info, "enabling auto recovery");
638 btrfs_set_opt(info->mount_opt, RECOVERY);
640 case Opt_skip_balance:
641 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
643 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
644 case Opt_check_integrity_including_extent_data:
645 btrfs_info(root->fs_info,
646 "enabling check integrity including extent data");
647 btrfs_set_opt(info->mount_opt,
648 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
649 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
651 case Opt_check_integrity:
652 btrfs_info(root->fs_info, "enabling check integrity");
653 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
655 case Opt_check_integrity_print_mask:
656 ret = match_int(&args[0], &intarg);
659 } else if (intarg >= 0) {
660 info->check_integrity_print_mask = intarg;
661 btrfs_info(root->fs_info, "check_integrity_print_mask 0x%x",
662 info->check_integrity_print_mask);
669 case Opt_check_integrity_including_extent_data:
670 case Opt_check_integrity:
671 case Opt_check_integrity_print_mask:
672 btrfs_err(root->fs_info,
673 "support for check_integrity* not compiled in!");
677 case Opt_fatal_errors:
678 if (strcmp(args[0].from, "panic") == 0)
679 btrfs_set_opt(info->mount_opt,
680 PANIC_ON_FATAL_ERROR);
681 else if (strcmp(args[0].from, "bug") == 0)
682 btrfs_clear_opt(info->mount_opt,
683 PANIC_ON_FATAL_ERROR);
689 case Opt_commit_interval:
691 ret = match_int(&args[0], &intarg);
693 btrfs_err(root->fs_info, "invalid commit interval");
699 btrfs_warn(root->fs_info, "excessive commit interval %d",
702 info->commit_interval = intarg;
704 btrfs_info(root->fs_info, "using default commit interval %ds",
705 BTRFS_DEFAULT_COMMIT_INTERVAL);
706 info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
710 btrfs_info(root->fs_info, "unrecognized mount option '%s'", p);
718 if (!ret && btrfs_test_opt(root, SPACE_CACHE))
719 btrfs_info(root->fs_info, "disk space caching is enabled");
725 * Parse mount options that are required early in the mount process.
727 * All other options will be parsed on much later in the mount process and
728 * only when we need to allocate a new super block.
730 static int btrfs_parse_early_options(const char *options, fmode_t flags,
731 void *holder, char **subvol_name, u64 *subvol_objectid,
732 struct btrfs_fs_devices **fs_devices)
734 substring_t args[MAX_OPT_ARGS];
735 char *device_name, *opts, *orig, *p;
743 * strsep changes the string, duplicate it because parse_options
746 opts = kstrdup(options, GFP_KERNEL);
751 while ((p = strsep(&opts, ",")) != NULL) {
756 token = match_token(p, tokens, args);
760 *subvol_name = match_strdup(&args[0]);
767 num = match_strdup(&args[0]);
769 *subvol_objectid = memparse(num, NULL);
771 /* we want the original fs_tree */
772 if (!*subvol_objectid)
774 BTRFS_FS_TREE_OBJECTID;
780 case Opt_subvolrootid:
782 "BTRFS: 'subvolrootid' mount option is deprecated and has "
786 device_name = match_strdup(&args[0]);
791 error = btrfs_scan_one_device(device_name,
792 flags, holder, fs_devices);
807 static struct dentry *get_default_root(struct super_block *sb,
810 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
811 struct btrfs_root *root = fs_info->tree_root;
812 struct btrfs_root *new_root;
813 struct btrfs_dir_item *di;
814 struct btrfs_path *path;
815 struct btrfs_key location;
821 * We have a specific subvol we want to mount, just setup location and
822 * go look up the root.
824 if (subvol_objectid) {
825 location.objectid = subvol_objectid;
826 location.type = BTRFS_ROOT_ITEM_KEY;
827 location.offset = (u64)-1;
831 path = btrfs_alloc_path();
833 return ERR_PTR(-ENOMEM);
834 path->leave_spinning = 1;
837 * Find the "default" dir item which points to the root item that we
838 * will mount by default if we haven't been given a specific subvolume
841 dir_id = btrfs_super_root_dir(fs_info->super_copy);
842 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
844 btrfs_free_path(path);
849 * Ok the default dir item isn't there. This is weird since
850 * it's always been there, but don't freak out, just try and
851 * mount to root most subvolume.
853 btrfs_free_path(path);
854 dir_id = BTRFS_FIRST_FREE_OBJECTID;
855 new_root = fs_info->fs_root;
859 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
860 btrfs_free_path(path);
863 new_root = btrfs_read_fs_root_no_name(fs_info, &location);
864 if (IS_ERR(new_root))
865 return ERR_CAST(new_root);
867 dir_id = btrfs_root_dirid(&new_root->root_item);
869 location.objectid = dir_id;
870 location.type = BTRFS_INODE_ITEM_KEY;
873 inode = btrfs_iget(sb, &location, new_root, &new);
875 return ERR_CAST(inode);
878 * If we're just mounting the root most subvol put the inode and return
879 * a reference to the dentry. We will have already gotten a reference
880 * to the inode in btrfs_fill_super so we're good to go.
882 if (!new && sb->s_root->d_inode == inode) {
884 return dget(sb->s_root);
887 return d_obtain_alias(inode);
890 static int btrfs_fill_super(struct super_block *sb,
891 struct btrfs_fs_devices *fs_devices,
892 void *data, int silent)
895 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
896 struct btrfs_key key;
899 sb->s_maxbytes = MAX_LFS_FILESIZE;
900 sb->s_magic = BTRFS_SUPER_MAGIC;
901 sb->s_op = &btrfs_super_ops;
902 sb->s_d_op = &btrfs_dentry_operations;
903 sb->s_export_op = &btrfs_export_ops;
904 sb->s_xattr = btrfs_xattr_handlers;
906 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
907 sb->s_flags |= MS_POSIXACL;
909 sb->s_flags |= MS_I_VERSION;
910 err = open_ctree(sb, fs_devices, (char *)data);
912 printk(KERN_ERR "BTRFS: open_ctree failed\n");
916 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
917 key.type = BTRFS_INODE_ITEM_KEY;
919 inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL);
921 err = PTR_ERR(inode);
925 sb->s_root = d_make_root(inode);
931 save_mount_options(sb, data);
932 cleancache_init_fs(sb);
933 sb->s_flags |= MS_ACTIVE;
937 close_ctree(fs_info->tree_root);
941 int btrfs_sync_fs(struct super_block *sb, int wait)
943 struct btrfs_trans_handle *trans;
944 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
945 struct btrfs_root *root = fs_info->tree_root;
947 trace_btrfs_sync_fs(wait);
950 filemap_flush(fs_info->btree_inode->i_mapping);
954 btrfs_wait_ordered_roots(fs_info, -1);
956 trans = btrfs_attach_transaction_barrier(root);
958 /* no transaction, don't bother */
959 if (PTR_ERR(trans) == -ENOENT)
961 return PTR_ERR(trans);
963 return btrfs_commit_transaction(trans, root);
966 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
968 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
969 struct btrfs_root *root = info->tree_root;
972 if (btrfs_test_opt(root, DEGRADED))
973 seq_puts(seq, ",degraded");
974 if (btrfs_test_opt(root, NODATASUM))
975 seq_puts(seq, ",nodatasum");
976 if (btrfs_test_opt(root, NODATACOW))
977 seq_puts(seq, ",nodatacow");
978 if (btrfs_test_opt(root, NOBARRIER))
979 seq_puts(seq, ",nobarrier");
980 if (info->max_inline != 8192 * 1024)
981 seq_printf(seq, ",max_inline=%llu", info->max_inline);
982 if (info->alloc_start != 0)
983 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
984 if (info->thread_pool_size != min_t(unsigned long,
985 num_online_cpus() + 2, 8))
986 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
987 if (btrfs_test_opt(root, COMPRESS)) {
988 if (info->compress_type == BTRFS_COMPRESS_ZLIB)
989 compress_type = "zlib";
991 compress_type = "lzo";
992 if (btrfs_test_opt(root, FORCE_COMPRESS))
993 seq_printf(seq, ",compress-force=%s", compress_type);
995 seq_printf(seq, ",compress=%s", compress_type);
997 if (btrfs_test_opt(root, NOSSD))
998 seq_puts(seq, ",nossd");
999 if (btrfs_test_opt(root, SSD_SPREAD))
1000 seq_puts(seq, ",ssd_spread");
1001 else if (btrfs_test_opt(root, SSD))
1002 seq_puts(seq, ",ssd");
1003 if (btrfs_test_opt(root, NOTREELOG))
1004 seq_puts(seq, ",notreelog");
1005 if (btrfs_test_opt(root, FLUSHONCOMMIT))
1006 seq_puts(seq, ",flushoncommit");
1007 if (btrfs_test_opt(root, DISCARD))
1008 seq_puts(seq, ",discard");
1009 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
1010 seq_puts(seq, ",noacl");
1011 if (btrfs_test_opt(root, SPACE_CACHE))
1012 seq_puts(seq, ",space_cache");
1014 seq_puts(seq, ",nospace_cache");
1015 if (btrfs_test_opt(root, RESCAN_UUID_TREE))
1016 seq_puts(seq, ",rescan_uuid_tree");
1017 if (btrfs_test_opt(root, CLEAR_CACHE))
1018 seq_puts(seq, ",clear_cache");
1019 if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
1020 seq_puts(seq, ",user_subvol_rm_allowed");
1021 if (btrfs_test_opt(root, ENOSPC_DEBUG))
1022 seq_puts(seq, ",enospc_debug");
1023 if (btrfs_test_opt(root, AUTO_DEFRAG))
1024 seq_puts(seq, ",autodefrag");
1025 if (btrfs_test_opt(root, INODE_MAP_CACHE))
1026 seq_puts(seq, ",inode_cache");
1027 if (btrfs_test_opt(root, SKIP_BALANCE))
1028 seq_puts(seq, ",skip_balance");
1029 if (btrfs_test_opt(root, RECOVERY))
1030 seq_puts(seq, ",recovery");
1031 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1032 if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1033 seq_puts(seq, ",check_int_data");
1034 else if (btrfs_test_opt(root, CHECK_INTEGRITY))
1035 seq_puts(seq, ",check_int");
1036 if (info->check_integrity_print_mask)
1037 seq_printf(seq, ",check_int_print_mask=%d",
1038 info->check_integrity_print_mask);
1040 if (info->metadata_ratio)
1041 seq_printf(seq, ",metadata_ratio=%d",
1042 info->metadata_ratio);
1043 if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
1044 seq_puts(seq, ",fatal_errors=panic");
1045 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1046 seq_printf(seq, ",commit=%d", info->commit_interval);
1050 static int btrfs_test_super(struct super_block *s, void *data)
1052 struct btrfs_fs_info *p = data;
1053 struct btrfs_fs_info *fs_info = btrfs_sb(s);
1055 return fs_info->fs_devices == p->fs_devices;
1058 static int btrfs_set_super(struct super_block *s, void *data)
1060 int err = set_anon_super(s, data);
1062 s->s_fs_info = data;
1067 * subvolumes are identified by ino 256
1069 static inline int is_subvolume_inode(struct inode *inode)
1071 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1077 * This will strip out the subvol=%s argument for an argument string and add
1078 * subvolid=0 to make sure we get the actual tree root for path walking to the
1081 static char *setup_root_args(char *args)
1083 unsigned len = strlen(args) + 2 + 1;
1084 char *src, *dst, *buf;
1087 * We need the same args as before, but with this substitution:
1088 * s!subvol=[^,]+!subvolid=0!
1090 * Since the replacement string is up to 2 bytes longer than the
1091 * original, allocate strlen(args) + 2 + 1 bytes.
1094 src = strstr(args, "subvol=");
1095 /* This shouldn't happen, but just in case.. */
1099 buf = dst = kmalloc(len, GFP_NOFS);
1104 * If the subvol= arg is not at the start of the string,
1105 * copy whatever precedes it into buf.
1110 dst += strlen(args);
1113 strcpy(dst, "subvolid=0");
1114 dst += strlen("subvolid=0");
1117 * If there is a "," after the original subvol=... string,
1118 * copy that suffix into our buffer. Otherwise, we're done.
1120 src = strchr(src, ',');
1127 static struct dentry *mount_subvol(const char *subvol_name, int flags,
1128 const char *device_name, char *data)
1130 struct dentry *root;
1131 struct vfsmount *mnt;
1134 newargs = setup_root_args(data);
1136 return ERR_PTR(-ENOMEM);
1137 mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name,
1141 return ERR_CAST(mnt);
1143 root = mount_subtree(mnt, subvol_name);
1145 if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) {
1146 struct super_block *s = root->d_sb;
1148 root = ERR_PTR(-EINVAL);
1149 deactivate_locked_super(s);
1150 printk(KERN_ERR "BTRFS: '%s' is not a valid subvolume\n",
1158 * Find a superblock for the given device / mount point.
1160 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1161 * for multiple device setup. Make sure to keep it in sync.
1163 static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
1164 const char *device_name, void *data)
1166 struct block_device *bdev = NULL;
1167 struct super_block *s;
1168 struct dentry *root;
1169 struct btrfs_fs_devices *fs_devices = NULL;
1170 struct btrfs_fs_info *fs_info = NULL;
1171 fmode_t mode = FMODE_READ;
1172 char *subvol_name = NULL;
1173 u64 subvol_objectid = 0;
1176 if (!(flags & MS_RDONLY))
1177 mode |= FMODE_WRITE;
1179 error = btrfs_parse_early_options(data, mode, fs_type,
1180 &subvol_name, &subvol_objectid,
1184 return ERR_PTR(error);
1188 root = mount_subvol(subvol_name, flags, device_name, data);
1193 error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices);
1195 return ERR_PTR(error);
1198 * Setup a dummy root and fs_info for test/set super. This is because
1199 * we don't actually fill this stuff out until open_ctree, but we need
1200 * it for searching for existing supers, so this lets us do that and
1201 * then open_ctree will properly initialize everything later.
1203 fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
1205 return ERR_PTR(-ENOMEM);
1207 fs_info->fs_devices = fs_devices;
1209 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1210 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1211 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1216 error = btrfs_open_devices(fs_devices, mode, fs_type);
1220 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
1222 goto error_close_devices;
1225 bdev = fs_devices->latest_bdev;
1226 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC,
1230 goto error_close_devices;
1234 btrfs_close_devices(fs_devices);
1235 free_fs_info(fs_info);
1236 if ((flags ^ s->s_flags) & MS_RDONLY)
1239 char b[BDEVNAME_SIZE];
1241 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
1242 btrfs_sb(s)->bdev_holder = fs_type;
1243 error = btrfs_fill_super(s, fs_devices, data,
1244 flags & MS_SILENT ? 1 : 0);
1247 root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error);
1249 deactivate_locked_super(s);
1253 error_close_devices:
1254 btrfs_close_devices(fs_devices);
1256 free_fs_info(fs_info);
1257 return ERR_PTR(error);
1260 static void btrfs_set_max_workers(struct btrfs_workers *workers, int new_limit)
1262 spin_lock_irq(&workers->lock);
1263 workers->max_workers = new_limit;
1264 spin_unlock_irq(&workers->lock);
1267 static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1268 int new_pool_size, int old_pool_size)
1270 if (new_pool_size == old_pool_size)
1273 fs_info->thread_pool_size = new_pool_size;
1275 btrfs_info(fs_info, "resize thread pool %d -> %d",
1276 old_pool_size, new_pool_size);
1278 btrfs_set_max_workers(&fs_info->generic_worker, new_pool_size);
1279 btrfs_set_max_workers(&fs_info->workers, new_pool_size);
1280 btrfs_set_max_workers(&fs_info->delalloc_workers, new_pool_size);
1281 btrfs_set_max_workers(&fs_info->submit_workers, new_pool_size);
1282 btrfs_set_max_workers(&fs_info->caching_workers, new_pool_size);
1283 btrfs_set_max_workers(&fs_info->fixup_workers, new_pool_size);
1284 btrfs_set_max_workers(&fs_info->endio_workers, new_pool_size);
1285 btrfs_set_max_workers(&fs_info->endio_meta_workers, new_pool_size);
1286 btrfs_set_max_workers(&fs_info->endio_meta_write_workers, new_pool_size);
1287 btrfs_set_max_workers(&fs_info->endio_write_workers, new_pool_size);
1288 btrfs_set_max_workers(&fs_info->endio_freespace_worker, new_pool_size);
1289 btrfs_set_max_workers(&fs_info->delayed_workers, new_pool_size);
1290 btrfs_set_max_workers(&fs_info->readahead_workers, new_pool_size);
1291 btrfs_set_max_workers(&fs_info->scrub_wr_completion_workers,
1295 static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
1297 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1300 static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1301 unsigned long old_opts, int flags)
1303 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1304 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1305 (flags & MS_RDONLY))) {
1306 /* wait for any defraggers to finish */
1307 wait_event(fs_info->transaction_wait,
1308 (atomic_read(&fs_info->defrag_running) == 0));
1309 if (flags & MS_RDONLY)
1310 sync_filesystem(fs_info->sb);
1314 static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1315 unsigned long old_opts)
1318 * We need cleanup all defragable inodes if the autodefragment is
1319 * close or the fs is R/O.
1321 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1322 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1323 (fs_info->sb->s_flags & MS_RDONLY))) {
1324 btrfs_cleanup_defrag_inodes(fs_info);
1327 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1330 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1332 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1333 struct btrfs_root *root = fs_info->tree_root;
1334 unsigned old_flags = sb->s_flags;
1335 unsigned long old_opts = fs_info->mount_opt;
1336 unsigned long old_compress_type = fs_info->compress_type;
1337 u64 old_max_inline = fs_info->max_inline;
1338 u64 old_alloc_start = fs_info->alloc_start;
1339 int old_thread_pool_size = fs_info->thread_pool_size;
1340 unsigned int old_metadata_ratio = fs_info->metadata_ratio;
1343 btrfs_remount_prepare(fs_info);
1345 ret = btrfs_parse_options(root, data);
1351 btrfs_remount_begin(fs_info, old_opts, *flags);
1352 btrfs_resize_thread_pool(fs_info,
1353 fs_info->thread_pool_size, old_thread_pool_size);
1355 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
1358 if (*flags & MS_RDONLY) {
1360 * this also happens on 'umount -rf' or on shutdown, when
1361 * the filesystem is busy.
1364 /* wait for the uuid_scan task to finish */
1365 down(&fs_info->uuid_tree_rescan_sem);
1366 /* avoid complains from lockdep et al. */
1367 up(&fs_info->uuid_tree_rescan_sem);
1369 sb->s_flags |= MS_RDONLY;
1371 btrfs_dev_replace_suspend_for_unmount(fs_info);
1372 btrfs_scrub_cancel(fs_info);
1373 btrfs_pause_balance(fs_info);
1375 ret = btrfs_commit_super(root);
1379 if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) {
1381 "Remounting read-write after error is not allowed");
1385 if (fs_info->fs_devices->rw_devices == 0) {
1390 if (fs_info->fs_devices->missing_devices >
1391 fs_info->num_tolerated_disk_barrier_failures &&
1392 !(*flags & MS_RDONLY)) {
1394 "too many missing devices, writeable remount is not allowed");
1399 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
1404 ret = btrfs_cleanup_fs_roots(fs_info);
1408 /* recover relocation */
1409 ret = btrfs_recover_relocation(root);
1413 ret = btrfs_resume_balance_async(fs_info);
1417 ret = btrfs_resume_dev_replace_async(fs_info);
1419 btrfs_warn(fs_info, "failed to resume dev_replace");
1423 if (!fs_info->uuid_root) {
1424 btrfs_info(fs_info, "creating UUID tree");
1425 ret = btrfs_create_uuid_tree(fs_info);
1427 btrfs_warn(fs_info, "failed to create the UUID tree %d", ret);
1431 sb->s_flags &= ~MS_RDONLY;
1434 btrfs_remount_cleanup(fs_info, old_opts);
1438 /* We've hit an error - don't reset MS_RDONLY */
1439 if (sb->s_flags & MS_RDONLY)
1440 old_flags |= MS_RDONLY;
1441 sb->s_flags = old_flags;
1442 fs_info->mount_opt = old_opts;
1443 fs_info->compress_type = old_compress_type;
1444 fs_info->max_inline = old_max_inline;
1445 mutex_lock(&fs_info->chunk_mutex);
1446 fs_info->alloc_start = old_alloc_start;
1447 mutex_unlock(&fs_info->chunk_mutex);
1448 btrfs_resize_thread_pool(fs_info,
1449 old_thread_pool_size, fs_info->thread_pool_size);
1450 fs_info->metadata_ratio = old_metadata_ratio;
1451 btrfs_remount_cleanup(fs_info, old_opts);
1455 /* Used to sort the devices by max_avail(descending sort) */
1456 static int btrfs_cmp_device_free_bytes(const void *dev_info1,
1457 const void *dev_info2)
1459 if (((struct btrfs_device_info *)dev_info1)->max_avail >
1460 ((struct btrfs_device_info *)dev_info2)->max_avail)
1462 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
1463 ((struct btrfs_device_info *)dev_info2)->max_avail)
1470 * sort the devices by max_avail, in which max free extent size of each device
1471 * is stored.(Descending Sort)
1473 static inline void btrfs_descending_sort_devices(
1474 struct btrfs_device_info *devices,
1477 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
1478 btrfs_cmp_device_free_bytes, NULL);
1482 * The helper to calc the free space on the devices that can be used to store
1485 static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
1487 struct btrfs_fs_info *fs_info = root->fs_info;
1488 struct btrfs_device_info *devices_info;
1489 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
1490 struct btrfs_device *device;
1495 u64 min_stripe_size;
1496 int min_stripes = 1, num_stripes = 1;
1497 int i = 0, nr_devices;
1500 nr_devices = fs_info->fs_devices->open_devices;
1501 BUG_ON(!nr_devices);
1503 devices_info = kmalloc_array(nr_devices, sizeof(*devices_info),
1508 /* calc min stripe number for data space alloction */
1509 type = btrfs_get_alloc_profile(root, 1);
1510 if (type & BTRFS_BLOCK_GROUP_RAID0) {
1512 num_stripes = nr_devices;
1513 } else if (type & BTRFS_BLOCK_GROUP_RAID1) {
1516 } else if (type & BTRFS_BLOCK_GROUP_RAID10) {
1521 if (type & BTRFS_BLOCK_GROUP_DUP)
1522 min_stripe_size = 2 * BTRFS_STRIPE_LEN;
1524 min_stripe_size = BTRFS_STRIPE_LEN;
1526 list_for_each_entry(device, &fs_devices->devices, dev_list) {
1527 if (!device->in_fs_metadata || !device->bdev ||
1528 device->is_tgtdev_for_dev_replace)
1531 avail_space = device->total_bytes - device->bytes_used;
1533 /* align with stripe_len */
1534 do_div(avail_space, BTRFS_STRIPE_LEN);
1535 avail_space *= BTRFS_STRIPE_LEN;
1538 * In order to avoid overwritting the superblock on the drive,
1539 * btrfs starts at an offset of at least 1MB when doing chunk
1542 skip_space = 1024 * 1024;
1544 /* user can set the offset in fs_info->alloc_start. */
1545 if (fs_info->alloc_start + BTRFS_STRIPE_LEN <=
1546 device->total_bytes)
1547 skip_space = max(fs_info->alloc_start, skip_space);
1550 * btrfs can not use the free space in [0, skip_space - 1],
1551 * we must subtract it from the total. In order to implement
1552 * it, we account the used space in this range first.
1554 ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1,
1557 kfree(devices_info);
1561 /* calc the free space in [0, skip_space - 1] */
1562 skip_space -= used_space;
1565 * we can use the free space in [0, skip_space - 1], subtract
1566 * it from the total.
1568 if (avail_space && avail_space >= skip_space)
1569 avail_space -= skip_space;
1573 if (avail_space < min_stripe_size)
1576 devices_info[i].dev = device;
1577 devices_info[i].max_avail = avail_space;
1584 btrfs_descending_sort_devices(devices_info, nr_devices);
1588 while (nr_devices >= min_stripes) {
1589 if (num_stripes > nr_devices)
1590 num_stripes = nr_devices;
1592 if (devices_info[i].max_avail >= min_stripe_size) {
1596 avail_space += devices_info[i].max_avail * num_stripes;
1597 alloc_size = devices_info[i].max_avail;
1598 for (j = i + 1 - num_stripes; j <= i; j++)
1599 devices_info[j].max_avail -= alloc_size;
1605 kfree(devices_info);
1606 *free_bytes = avail_space;
1610 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
1612 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
1613 struct btrfs_super_block *disk_super = fs_info->super_copy;
1614 struct list_head *head = &fs_info->space_info;
1615 struct btrfs_space_info *found;
1617 u64 total_free_data = 0;
1618 int bits = dentry->d_sb->s_blocksize_bits;
1619 __be32 *fsid = (__be32 *)fs_info->fsid;
1622 /* holding chunk_muext to avoid allocating new chunks */
1623 mutex_lock(&fs_info->chunk_mutex);
1625 list_for_each_entry_rcu(found, head, list) {
1626 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
1627 total_free_data += found->disk_total - found->disk_used;
1629 btrfs_account_ro_block_groups_free_space(found);
1632 total_used += found->disk_used;
1636 buf->f_namelen = BTRFS_NAME_LEN;
1637 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
1638 buf->f_bfree = buf->f_blocks - (total_used >> bits);
1639 buf->f_bsize = dentry->d_sb->s_blocksize;
1640 buf->f_type = BTRFS_SUPER_MAGIC;
1641 buf->f_bavail = total_free_data;
1642 ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data);
1644 mutex_unlock(&fs_info->chunk_mutex);
1647 buf->f_bavail += total_free_data;
1648 buf->f_bavail = buf->f_bavail >> bits;
1649 mutex_unlock(&fs_info->chunk_mutex);
1651 /* We treat it as constant endianness (it doesn't matter _which_)
1652 because we want the fsid to come out the same whether mounted
1653 on a big-endian or little-endian host */
1654 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
1655 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
1656 /* Mask in the root object ID too, to disambiguate subvols */
1657 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
1658 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
1663 static void btrfs_kill_super(struct super_block *sb)
1665 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1666 kill_anon_super(sb);
1667 free_fs_info(fs_info);
1670 static struct file_system_type btrfs_fs_type = {
1671 .owner = THIS_MODULE,
1673 .mount = btrfs_mount,
1674 .kill_sb = btrfs_kill_super,
1675 .fs_flags = FS_REQUIRES_DEV,
1677 MODULE_ALIAS_FS("btrfs");
1680 * used by btrfsctl to scan devices when no FS is mounted
1682 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
1685 struct btrfs_ioctl_vol_args *vol;
1686 struct btrfs_fs_devices *fs_devices;
1689 if (!capable(CAP_SYS_ADMIN))
1692 vol = memdup_user((void __user *)arg, sizeof(*vol));
1694 return PTR_ERR(vol);
1697 case BTRFS_IOC_SCAN_DEV:
1698 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1699 &btrfs_fs_type, &fs_devices);
1701 case BTRFS_IOC_DEVICES_READY:
1702 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1703 &btrfs_fs_type, &fs_devices);
1706 ret = !(fs_devices->num_devices == fs_devices->total_devices);
1714 static int btrfs_freeze(struct super_block *sb)
1716 struct btrfs_trans_handle *trans;
1717 struct btrfs_root *root = btrfs_sb(sb)->tree_root;
1719 trans = btrfs_attach_transaction_barrier(root);
1720 if (IS_ERR(trans)) {
1721 /* no transaction, don't bother */
1722 if (PTR_ERR(trans) == -ENOENT)
1724 return PTR_ERR(trans);
1726 return btrfs_commit_transaction(trans, root);
1729 static int btrfs_unfreeze(struct super_block *sb)
1734 static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
1736 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
1737 struct btrfs_fs_devices *cur_devices;
1738 struct btrfs_device *dev, *first_dev = NULL;
1739 struct list_head *head;
1740 struct rcu_string *name;
1742 mutex_lock(&fs_info->fs_devices->device_list_mutex);
1743 cur_devices = fs_info->fs_devices;
1744 while (cur_devices) {
1745 head = &cur_devices->devices;
1746 list_for_each_entry(dev, head, dev_list) {
1749 if (!first_dev || dev->devid < first_dev->devid)
1752 cur_devices = cur_devices->seed;
1757 name = rcu_dereference(first_dev->name);
1758 seq_escape(m, name->str, " \t\n\\");
1763 mutex_unlock(&fs_info->fs_devices->device_list_mutex);
1767 static const struct super_operations btrfs_super_ops = {
1768 .drop_inode = btrfs_drop_inode,
1769 .evict_inode = btrfs_evict_inode,
1770 .put_super = btrfs_put_super,
1771 .sync_fs = btrfs_sync_fs,
1772 .show_options = btrfs_show_options,
1773 .show_devname = btrfs_show_devname,
1774 .write_inode = btrfs_write_inode,
1775 .alloc_inode = btrfs_alloc_inode,
1776 .destroy_inode = btrfs_destroy_inode,
1777 .statfs = btrfs_statfs,
1778 .remount_fs = btrfs_remount,
1779 .freeze_fs = btrfs_freeze,
1780 .unfreeze_fs = btrfs_unfreeze,
1783 static const struct file_operations btrfs_ctl_fops = {
1784 .unlocked_ioctl = btrfs_control_ioctl,
1785 .compat_ioctl = btrfs_control_ioctl,
1786 .owner = THIS_MODULE,
1787 .llseek = noop_llseek,
1790 static struct miscdevice btrfs_misc = {
1791 .minor = BTRFS_MINOR,
1792 .name = "btrfs-control",
1793 .fops = &btrfs_ctl_fops
1796 MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
1797 MODULE_ALIAS("devname:btrfs-control");
1799 static int btrfs_interface_init(void)
1801 return misc_register(&btrfs_misc);
1804 static void btrfs_interface_exit(void)
1806 if (misc_deregister(&btrfs_misc) < 0)
1807 printk(KERN_INFO "BTRFS: misc_deregister failed for control device\n");
1810 static void btrfs_print_info(void)
1812 printk(KERN_INFO "Btrfs loaded"
1813 #ifdef CONFIG_BTRFS_DEBUG
1816 #ifdef CONFIG_BTRFS_ASSERT
1819 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1820 ", integrity-checker=on"
1825 static int btrfs_run_sanity_tests(void)
1829 ret = btrfs_init_test_fs();
1833 ret = btrfs_test_free_space_cache();
1836 ret = btrfs_test_extent_buffer_operations();
1839 ret = btrfs_test_extent_io();
1842 ret = btrfs_test_inodes();
1844 btrfs_destroy_test_fs();
1848 static int __init init_btrfs_fs(void)
1852 err = btrfs_init_sysfs();
1856 btrfs_init_compress();
1858 err = btrfs_init_cachep();
1862 err = extent_io_init();
1866 err = extent_map_init();
1868 goto free_extent_io;
1870 err = ordered_data_init();
1872 goto free_extent_map;
1874 err = btrfs_delayed_inode_init();
1876 goto free_ordered_data;
1878 err = btrfs_auto_defrag_init();
1880 goto free_delayed_inode;
1882 err = btrfs_delayed_ref_init();
1884 goto free_auto_defrag;
1886 err = btrfs_prelim_ref_init();
1888 goto free_prelim_ref;
1890 err = btrfs_interface_init();
1892 goto free_delayed_ref;
1894 btrfs_init_lockdep();
1898 err = btrfs_run_sanity_tests();
1900 goto unregister_ioctl;
1902 err = register_filesystem(&btrfs_fs_type);
1904 goto unregister_ioctl;
1909 btrfs_interface_exit();
1911 btrfs_prelim_ref_exit();
1913 btrfs_delayed_ref_exit();
1915 btrfs_auto_defrag_exit();
1917 btrfs_delayed_inode_exit();
1919 ordered_data_exit();
1925 btrfs_destroy_cachep();
1927 btrfs_exit_compress();
1932 static void __exit exit_btrfs_fs(void)
1934 btrfs_destroy_cachep();
1935 btrfs_delayed_ref_exit();
1936 btrfs_auto_defrag_exit();
1937 btrfs_delayed_inode_exit();
1938 btrfs_prelim_ref_exit();
1939 ordered_data_exit();
1942 btrfs_interface_exit();
1943 unregister_filesystem(&btrfs_fs_type);
1945 btrfs_cleanup_fs_uuids();
1946 btrfs_exit_compress();
1949 module_init(init_btrfs_fs)
1950 module_exit(exit_btrfs_fs)
1952 MODULE_LICENSE("GPL");