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>
46 #include "delayed-inode.h"
49 #include "transaction.h"
50 #include "btrfs_inode.h"
51 #include "print-tree.h"
55 #include "compression.h"
56 #include "rcu-string.h"
57 #include "dev-replace.h"
58 #include "free-space-cache.h"
60 #include "tests/btrfs-tests.h"
62 #define CREATE_TRACE_POINTS
63 #include <trace/events/btrfs.h>
65 static const struct super_operations btrfs_super_ops;
66 static struct file_system_type btrfs_fs_type;
68 static const char *btrfs_decode_error(int errno)
70 char *errstr = "unknown";
74 errstr = "IO failure";
77 errstr = "Out of memory";
80 errstr = "Readonly filesystem";
83 errstr = "Object already exists";
86 errstr = "No space left";
89 errstr = "No such entry";
96 static void save_error_info(struct btrfs_fs_info *fs_info)
99 * today we only save the error info into ram. Long term we'll
100 * also send it down to the disk
102 set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state);
105 /* btrfs handle error by forcing the filesystem readonly */
106 static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
108 struct super_block *sb = fs_info->sb;
110 if (sb->s_flags & MS_RDONLY)
113 if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
114 sb->s_flags |= MS_RDONLY;
115 btrfs_info(fs_info, "forced readonly");
117 * Note that a running device replace operation is not
118 * canceled here although there is no way to update
119 * the progress. It would add the risk of a deadlock,
120 * therefore the canceling is ommited. The only penalty
121 * is that some I/O remains active until the procedure
122 * completes. The next time when the filesystem is
123 * mounted writeable again, the device replace
124 * operation continues.
131 * __btrfs_std_error decodes expected errors from the caller and
132 * invokes the approciate error response.
134 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
135 unsigned int line, int errno, const char *fmt, ...)
137 struct super_block *sb = fs_info->sb;
141 * Special case: if the error is EROFS, and we're already
142 * under MS_RDONLY, then it is safe here.
144 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
147 errstr = btrfs_decode_error(errno);
149 struct va_format vaf;
156 printk(KERN_CRIT "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 printk(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
299 s_id, 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,
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_max_inline, "max_inline=%s"},
339 {Opt_alloc_start, "alloc_start=%s"},
340 {Opt_thread_pool, "thread_pool=%d"},
341 {Opt_compress, "compress"},
342 {Opt_compress_type, "compress=%s"},
343 {Opt_compress_force, "compress-force"},
344 {Opt_compress_force_type, "compress-force=%s"},
346 {Opt_ssd_spread, "ssd_spread"},
347 {Opt_nossd, "nossd"},
348 {Opt_noacl, "noacl"},
349 {Opt_notreelog, "notreelog"},
350 {Opt_flushoncommit, "flushoncommit"},
351 {Opt_ratio, "metadata_ratio=%d"},
352 {Opt_discard, "discard"},
353 {Opt_space_cache, "space_cache"},
354 {Opt_clear_cache, "clear_cache"},
355 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
356 {Opt_enospc_debug, "enospc_debug"},
357 {Opt_subvolrootid, "subvolrootid=%d"},
358 {Opt_defrag, "autodefrag"},
359 {Opt_inode_cache, "inode_cache"},
360 {Opt_no_space_cache, "nospace_cache"},
361 {Opt_recovery, "recovery"},
362 {Opt_skip_balance, "skip_balance"},
363 {Opt_check_integrity, "check_int"},
364 {Opt_check_integrity_including_extent_data, "check_int_data"},
365 {Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
366 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
367 {Opt_fatal_errors, "fatal_errors=%s"},
368 {Opt_commit_interval, "commit=%d"},
373 * Regular mount options parser. Everything that is needed only when
374 * reading in a new superblock is parsed here.
375 * XXX JDM: This needs to be cleaned up for remount.
377 int btrfs_parse_options(struct btrfs_root *root, char *options)
379 struct btrfs_fs_info *info = root->fs_info;
380 substring_t args[MAX_OPT_ARGS];
381 char *p, *num, *orig = NULL;
386 bool compress_force = false;
388 cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
390 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
396 * strsep changes the string, duplicate it because parse_options
399 options = kstrdup(options, GFP_NOFS);
405 while ((p = strsep(&options, ",")) != NULL) {
410 token = match_token(p, tokens, args);
413 printk(KERN_INFO "btrfs: allowing degraded mounts\n");
414 btrfs_set_opt(info->mount_opt, DEGRADED);
418 case Opt_subvolrootid:
421 * These are parsed by btrfs_parse_early_options
422 * and can be happily ignored here.
426 printk(KERN_INFO "btrfs: setting nodatasum\n");
427 btrfs_set_opt(info->mount_opt, NODATASUM);
430 if (!btrfs_test_opt(root, COMPRESS) ||
431 !btrfs_test_opt(root, FORCE_COMPRESS)) {
432 printk(KERN_INFO "btrfs: setting nodatacow, compression disabled\n");
434 printk(KERN_INFO "btrfs: setting nodatacow\n");
436 info->compress_type = BTRFS_COMPRESS_NONE;
437 btrfs_clear_opt(info->mount_opt, COMPRESS);
438 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
439 btrfs_set_opt(info->mount_opt, NODATACOW);
440 btrfs_set_opt(info->mount_opt, NODATASUM);
442 case Opt_compress_force:
443 case Opt_compress_force_type:
444 compress_force = true;
447 case Opt_compress_type:
448 if (token == Opt_compress ||
449 token == Opt_compress_force ||
450 strcmp(args[0].from, "zlib") == 0) {
451 compress_type = "zlib";
452 info->compress_type = BTRFS_COMPRESS_ZLIB;
453 btrfs_set_opt(info->mount_opt, COMPRESS);
454 btrfs_clear_opt(info->mount_opt, NODATACOW);
455 btrfs_clear_opt(info->mount_opt, NODATASUM);
456 } else if (strcmp(args[0].from, "lzo") == 0) {
457 compress_type = "lzo";
458 info->compress_type = BTRFS_COMPRESS_LZO;
459 btrfs_set_opt(info->mount_opt, COMPRESS);
460 btrfs_clear_opt(info->mount_opt, NODATACOW);
461 btrfs_clear_opt(info->mount_opt, NODATASUM);
462 btrfs_set_fs_incompat(info, COMPRESS_LZO);
463 } else if (strncmp(args[0].from, "no", 2) == 0) {
464 compress_type = "no";
465 info->compress_type = BTRFS_COMPRESS_NONE;
466 btrfs_clear_opt(info->mount_opt, COMPRESS);
467 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
468 compress_force = false;
474 if (compress_force) {
475 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
476 pr_info("btrfs: force %s compression\n",
479 pr_info("btrfs: use %s compression\n",
483 printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
484 btrfs_set_opt(info->mount_opt, SSD);
487 printk(KERN_INFO "btrfs: use spread ssd "
488 "allocation scheme\n");
489 btrfs_set_opt(info->mount_opt, SSD);
490 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
493 printk(KERN_INFO "btrfs: not using ssd allocation "
495 btrfs_set_opt(info->mount_opt, NOSSD);
496 btrfs_clear_opt(info->mount_opt, SSD);
497 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
500 printk(KERN_INFO "btrfs: turning off barriers\n");
501 btrfs_set_opt(info->mount_opt, NOBARRIER);
503 case Opt_thread_pool:
504 ret = match_int(&args[0], &intarg);
507 } else if (intarg > 0) {
508 info->thread_pool_size = intarg;
515 num = match_strdup(&args[0]);
517 info->max_inline = memparse(num, NULL);
520 if (info->max_inline) {
521 info->max_inline = max_t(u64,
525 printk(KERN_INFO "btrfs: max_inline at %llu\n",
532 case Opt_alloc_start:
533 num = match_strdup(&args[0]);
535 mutex_lock(&info->chunk_mutex);
536 info->alloc_start = memparse(num, NULL);
537 mutex_unlock(&info->chunk_mutex);
540 "btrfs: allocations start at %llu\n",
548 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
551 printk(KERN_INFO "btrfs: disabling tree log\n");
552 btrfs_set_opt(info->mount_opt, NOTREELOG);
554 case Opt_flushoncommit:
555 printk(KERN_INFO "btrfs: turning on flush-on-commit\n");
556 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
559 ret = match_int(&args[0], &intarg);
562 } else if (intarg >= 0) {
563 info->metadata_ratio = intarg;
564 printk(KERN_INFO "btrfs: metadata ratio %d\n",
565 info->metadata_ratio);
572 btrfs_set_opt(info->mount_opt, DISCARD);
574 case Opt_space_cache:
575 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
577 case Opt_rescan_uuid_tree:
578 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
580 case Opt_no_space_cache:
581 printk(KERN_INFO "btrfs: disabling disk space caching\n");
582 btrfs_clear_opt(info->mount_opt, SPACE_CACHE);
584 case Opt_inode_cache:
585 printk(KERN_INFO "btrfs: enabling inode map caching\n");
586 btrfs_set_opt(info->mount_opt, INODE_MAP_CACHE);
588 case Opt_clear_cache:
589 printk(KERN_INFO "btrfs: force clearing of disk cache\n");
590 btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
592 case Opt_user_subvol_rm_allowed:
593 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
595 case Opt_enospc_debug:
596 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
599 printk(KERN_INFO "btrfs: enabling auto defrag\n");
600 btrfs_set_opt(info->mount_opt, AUTO_DEFRAG);
603 printk(KERN_INFO "btrfs: enabling auto recovery\n");
604 btrfs_set_opt(info->mount_opt, RECOVERY);
606 case Opt_skip_balance:
607 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
609 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
610 case Opt_check_integrity_including_extent_data:
611 printk(KERN_INFO "btrfs: enabling check integrity"
612 " including extent data\n");
613 btrfs_set_opt(info->mount_opt,
614 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
615 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
617 case Opt_check_integrity:
618 printk(KERN_INFO "btrfs: enabling check integrity\n");
619 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
621 case Opt_check_integrity_print_mask:
622 ret = match_int(&args[0], &intarg);
625 } else if (intarg >= 0) {
626 info->check_integrity_print_mask = intarg;
627 printk(KERN_INFO "btrfs:"
628 " check_integrity_print_mask 0x%x\n",
629 info->check_integrity_print_mask);
636 case Opt_check_integrity_including_extent_data:
637 case Opt_check_integrity:
638 case Opt_check_integrity_print_mask:
639 printk(KERN_ERR "btrfs: support for check_integrity*"
640 " not compiled in!\n");
644 case Opt_fatal_errors:
645 if (strcmp(args[0].from, "panic") == 0)
646 btrfs_set_opt(info->mount_opt,
647 PANIC_ON_FATAL_ERROR);
648 else if (strcmp(args[0].from, "bug") == 0)
649 btrfs_clear_opt(info->mount_opt,
650 PANIC_ON_FATAL_ERROR);
656 case Opt_commit_interval:
658 ret = match_int(&args[0], &intarg);
661 "btrfs: invalid commit interval\n");
668 "btrfs: excessive commit interval %d\n",
671 info->commit_interval = intarg;
674 "btrfs: using default commit interval %ds\n",
675 BTRFS_DEFAULT_COMMIT_INTERVAL);
676 info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
680 printk(KERN_INFO "btrfs: unrecognized mount option "
689 if (!ret && btrfs_test_opt(root, SPACE_CACHE))
690 printk(KERN_INFO "btrfs: disk space caching is enabled\n");
696 * Parse mount options that are required early in the mount process.
698 * All other options will be parsed on much later in the mount process and
699 * only when we need to allocate a new super block.
701 static int btrfs_parse_early_options(const char *options, fmode_t flags,
702 void *holder, char **subvol_name, u64 *subvol_objectid,
703 struct btrfs_fs_devices **fs_devices)
705 substring_t args[MAX_OPT_ARGS];
706 char *device_name, *opts, *orig, *p;
714 * strsep changes the string, duplicate it because parse_options
717 opts = kstrdup(options, GFP_KERNEL);
722 while ((p = strsep(&opts, ",")) != NULL) {
727 token = match_token(p, tokens, args);
731 *subvol_name = match_strdup(&args[0]);
738 num = match_strdup(&args[0]);
740 *subvol_objectid = memparse(num, NULL);
742 /* we want the original fs_tree */
743 if (!*subvol_objectid)
745 BTRFS_FS_TREE_OBJECTID;
751 case Opt_subvolrootid:
753 "btrfs: 'subvolrootid' mount option is deprecated and has no effect\n");
756 device_name = match_strdup(&args[0]);
761 error = btrfs_scan_one_device(device_name,
762 flags, holder, fs_devices);
777 static struct dentry *get_default_root(struct super_block *sb,
780 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
781 struct btrfs_root *root = fs_info->tree_root;
782 struct btrfs_root *new_root;
783 struct btrfs_dir_item *di;
784 struct btrfs_path *path;
785 struct btrfs_key location;
791 * We have a specific subvol we want to mount, just setup location and
792 * go look up the root.
794 if (subvol_objectid) {
795 location.objectid = subvol_objectid;
796 location.type = BTRFS_ROOT_ITEM_KEY;
797 location.offset = (u64)-1;
801 path = btrfs_alloc_path();
803 return ERR_PTR(-ENOMEM);
804 path->leave_spinning = 1;
807 * Find the "default" dir item which points to the root item that we
808 * will mount by default if we haven't been given a specific subvolume
811 dir_id = btrfs_super_root_dir(fs_info->super_copy);
812 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
814 btrfs_free_path(path);
819 * Ok the default dir item isn't there. This is weird since
820 * it's always been there, but don't freak out, just try and
821 * mount to root most subvolume.
823 btrfs_free_path(path);
824 dir_id = BTRFS_FIRST_FREE_OBJECTID;
825 new_root = fs_info->fs_root;
829 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
830 btrfs_free_path(path);
833 new_root = btrfs_read_fs_root_no_name(fs_info, &location);
834 if (IS_ERR(new_root))
835 return ERR_CAST(new_root);
837 dir_id = btrfs_root_dirid(&new_root->root_item);
839 location.objectid = dir_id;
840 location.type = BTRFS_INODE_ITEM_KEY;
843 inode = btrfs_iget(sb, &location, new_root, &new);
845 return ERR_CAST(inode);
848 * If we're just mounting the root most subvol put the inode and return
849 * a reference to the dentry. We will have already gotten a reference
850 * to the inode in btrfs_fill_super so we're good to go.
852 if (!new && sb->s_root->d_inode == inode) {
854 return dget(sb->s_root);
857 return d_obtain_alias(inode);
860 static int btrfs_fill_super(struct super_block *sb,
861 struct btrfs_fs_devices *fs_devices,
862 void *data, int silent)
865 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
866 struct btrfs_key key;
869 sb->s_maxbytes = MAX_LFS_FILESIZE;
870 sb->s_magic = BTRFS_SUPER_MAGIC;
871 sb->s_op = &btrfs_super_ops;
872 sb->s_d_op = &btrfs_dentry_operations;
873 sb->s_export_op = &btrfs_export_ops;
874 sb->s_xattr = btrfs_xattr_handlers;
876 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
877 sb->s_flags |= MS_POSIXACL;
879 sb->s_flags |= MS_I_VERSION;
880 err = open_ctree(sb, fs_devices, (char *)data);
882 printk("btrfs: open_ctree failed\n");
886 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
887 key.type = BTRFS_INODE_ITEM_KEY;
889 inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL);
891 err = PTR_ERR(inode);
895 sb->s_root = d_make_root(inode);
901 save_mount_options(sb, data);
902 cleancache_init_fs(sb);
903 sb->s_flags |= MS_ACTIVE;
907 close_ctree(fs_info->tree_root);
911 int btrfs_sync_fs(struct super_block *sb, int wait)
913 struct btrfs_trans_handle *trans;
914 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
915 struct btrfs_root *root = fs_info->tree_root;
917 trace_btrfs_sync_fs(wait);
920 filemap_flush(fs_info->btree_inode->i_mapping);
924 btrfs_wait_all_ordered_extents(fs_info, 1);
926 trans = btrfs_attach_transaction_barrier(root);
928 /* no transaction, don't bother */
929 if (PTR_ERR(trans) == -ENOENT)
931 return PTR_ERR(trans);
933 return btrfs_commit_transaction(trans, root);
936 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
938 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
939 struct btrfs_root *root = info->tree_root;
942 if (btrfs_test_opt(root, DEGRADED))
943 seq_puts(seq, ",degraded");
944 if (btrfs_test_opt(root, NODATASUM))
945 seq_puts(seq, ",nodatasum");
946 if (btrfs_test_opt(root, NODATACOW))
947 seq_puts(seq, ",nodatacow");
948 if (btrfs_test_opt(root, NOBARRIER))
949 seq_puts(seq, ",nobarrier");
950 if (info->max_inline != 8192 * 1024)
951 seq_printf(seq, ",max_inline=%llu", info->max_inline);
952 if (info->alloc_start != 0)
953 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
954 if (info->thread_pool_size != min_t(unsigned long,
955 num_online_cpus() + 2, 8))
956 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
957 if (btrfs_test_opt(root, COMPRESS)) {
958 if (info->compress_type == BTRFS_COMPRESS_ZLIB)
959 compress_type = "zlib";
961 compress_type = "lzo";
962 if (btrfs_test_opt(root, FORCE_COMPRESS))
963 seq_printf(seq, ",compress-force=%s", compress_type);
965 seq_printf(seq, ",compress=%s", compress_type);
967 if (btrfs_test_opt(root, NOSSD))
968 seq_puts(seq, ",nossd");
969 if (btrfs_test_opt(root, SSD_SPREAD))
970 seq_puts(seq, ",ssd_spread");
971 else if (btrfs_test_opt(root, SSD))
972 seq_puts(seq, ",ssd");
973 if (btrfs_test_opt(root, NOTREELOG))
974 seq_puts(seq, ",notreelog");
975 if (btrfs_test_opt(root, FLUSHONCOMMIT))
976 seq_puts(seq, ",flushoncommit");
977 if (btrfs_test_opt(root, DISCARD))
978 seq_puts(seq, ",discard");
979 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
980 seq_puts(seq, ",noacl");
981 if (btrfs_test_opt(root, SPACE_CACHE))
982 seq_puts(seq, ",space_cache");
984 seq_puts(seq, ",nospace_cache");
985 if (btrfs_test_opt(root, RESCAN_UUID_TREE))
986 seq_puts(seq, ",rescan_uuid_tree");
987 if (btrfs_test_opt(root, CLEAR_CACHE))
988 seq_puts(seq, ",clear_cache");
989 if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
990 seq_puts(seq, ",user_subvol_rm_allowed");
991 if (btrfs_test_opt(root, ENOSPC_DEBUG))
992 seq_puts(seq, ",enospc_debug");
993 if (btrfs_test_opt(root, AUTO_DEFRAG))
994 seq_puts(seq, ",autodefrag");
995 if (btrfs_test_opt(root, INODE_MAP_CACHE))
996 seq_puts(seq, ",inode_cache");
997 if (btrfs_test_opt(root, SKIP_BALANCE))
998 seq_puts(seq, ",skip_balance");
999 if (btrfs_test_opt(root, RECOVERY))
1000 seq_puts(seq, ",recovery");
1001 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1002 if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1003 seq_puts(seq, ",check_int_data");
1004 else if (btrfs_test_opt(root, CHECK_INTEGRITY))
1005 seq_puts(seq, ",check_int");
1006 if (info->check_integrity_print_mask)
1007 seq_printf(seq, ",check_int_print_mask=%d",
1008 info->check_integrity_print_mask);
1010 if (info->metadata_ratio)
1011 seq_printf(seq, ",metadata_ratio=%d",
1012 info->metadata_ratio);
1013 if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
1014 seq_puts(seq, ",fatal_errors=panic");
1015 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1016 seq_printf(seq, ",commit=%d", info->commit_interval);
1020 static int btrfs_test_super(struct super_block *s, void *data)
1022 struct btrfs_fs_info *p = data;
1023 struct btrfs_fs_info *fs_info = btrfs_sb(s);
1025 return fs_info->fs_devices == p->fs_devices;
1028 static int btrfs_set_super(struct super_block *s, void *data)
1030 int err = set_anon_super(s, data);
1032 s->s_fs_info = data;
1037 * subvolumes are identified by ino 256
1039 static inline int is_subvolume_inode(struct inode *inode)
1041 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1047 * This will strip out the subvol=%s argument for an argument string and add
1048 * subvolid=0 to make sure we get the actual tree root for path walking to the
1051 static char *setup_root_args(char *args)
1053 unsigned len = strlen(args) + 2 + 1;
1054 char *src, *dst, *buf;
1057 * We need the same args as before, but with this substitution:
1058 * s!subvol=[^,]+!subvolid=0!
1060 * Since the replacement string is up to 2 bytes longer than the
1061 * original, allocate strlen(args) + 2 + 1 bytes.
1064 src = strstr(args, "subvol=");
1065 /* This shouldn't happen, but just in case.. */
1069 buf = dst = kmalloc(len, GFP_NOFS);
1074 * If the subvol= arg is not at the start of the string,
1075 * copy whatever precedes it into buf.
1080 dst += strlen(args);
1083 strcpy(dst, "subvolid=0");
1084 dst += strlen("subvolid=0");
1087 * If there is a "," after the original subvol=... string,
1088 * copy that suffix into our buffer. Otherwise, we're done.
1090 src = strchr(src, ',');
1097 static struct dentry *mount_subvol(const char *subvol_name, int flags,
1098 const char *device_name, char *data)
1100 struct dentry *root;
1101 struct vfsmount *mnt;
1104 newargs = setup_root_args(data);
1106 return ERR_PTR(-ENOMEM);
1107 mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name,
1111 return ERR_CAST(mnt);
1113 root = mount_subtree(mnt, subvol_name);
1115 if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) {
1116 struct super_block *s = root->d_sb;
1118 root = ERR_PTR(-EINVAL);
1119 deactivate_locked_super(s);
1120 printk(KERN_ERR "btrfs: '%s' is not a valid subvolume\n",
1128 * Find a superblock for the given device / mount point.
1130 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1131 * for multiple device setup. Make sure to keep it in sync.
1133 static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
1134 const char *device_name, void *data)
1136 struct block_device *bdev = NULL;
1137 struct super_block *s;
1138 struct dentry *root;
1139 struct btrfs_fs_devices *fs_devices = NULL;
1140 struct btrfs_fs_info *fs_info = NULL;
1141 fmode_t mode = FMODE_READ;
1142 char *subvol_name = NULL;
1143 u64 subvol_objectid = 0;
1146 if (!(flags & MS_RDONLY))
1147 mode |= FMODE_WRITE;
1149 error = btrfs_parse_early_options(data, mode, fs_type,
1150 &subvol_name, &subvol_objectid,
1154 return ERR_PTR(error);
1158 root = mount_subvol(subvol_name, flags, device_name, data);
1163 error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices);
1165 return ERR_PTR(error);
1168 * Setup a dummy root and fs_info for test/set super. This is because
1169 * we don't actually fill this stuff out until open_ctree, but we need
1170 * it for searching for existing supers, so this lets us do that and
1171 * then open_ctree will properly initialize everything later.
1173 fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
1175 return ERR_PTR(-ENOMEM);
1177 fs_info->fs_devices = fs_devices;
1179 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1180 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1181 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1186 error = btrfs_open_devices(fs_devices, mode, fs_type);
1190 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
1192 goto error_close_devices;
1195 bdev = fs_devices->latest_bdev;
1196 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC,
1200 goto error_close_devices;
1204 btrfs_close_devices(fs_devices);
1205 free_fs_info(fs_info);
1206 if ((flags ^ s->s_flags) & MS_RDONLY)
1209 char b[BDEVNAME_SIZE];
1211 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
1212 btrfs_sb(s)->bdev_holder = fs_type;
1213 error = btrfs_fill_super(s, fs_devices, data,
1214 flags & MS_SILENT ? 1 : 0);
1217 root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error);
1219 deactivate_locked_super(s);
1223 error_close_devices:
1224 btrfs_close_devices(fs_devices);
1226 free_fs_info(fs_info);
1227 return ERR_PTR(error);
1230 static void btrfs_set_max_workers(struct btrfs_workers *workers, int new_limit)
1232 spin_lock_irq(&workers->lock);
1233 workers->max_workers = new_limit;
1234 spin_unlock_irq(&workers->lock);
1237 static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1238 int new_pool_size, int old_pool_size)
1240 if (new_pool_size == old_pool_size)
1243 fs_info->thread_pool_size = new_pool_size;
1245 printk(KERN_INFO "btrfs: resize thread pool %d -> %d\n",
1246 old_pool_size, new_pool_size);
1248 btrfs_set_max_workers(&fs_info->generic_worker, new_pool_size);
1249 btrfs_set_max_workers(&fs_info->workers, new_pool_size);
1250 btrfs_set_max_workers(&fs_info->delalloc_workers, new_pool_size);
1251 btrfs_set_max_workers(&fs_info->submit_workers, new_pool_size);
1252 btrfs_set_max_workers(&fs_info->caching_workers, new_pool_size);
1253 btrfs_set_max_workers(&fs_info->fixup_workers, new_pool_size);
1254 btrfs_set_max_workers(&fs_info->endio_workers, new_pool_size);
1255 btrfs_set_max_workers(&fs_info->endio_meta_workers, new_pool_size);
1256 btrfs_set_max_workers(&fs_info->endio_meta_write_workers, new_pool_size);
1257 btrfs_set_max_workers(&fs_info->endio_write_workers, new_pool_size);
1258 btrfs_set_max_workers(&fs_info->endio_freespace_worker, new_pool_size);
1259 btrfs_set_max_workers(&fs_info->delayed_workers, new_pool_size);
1260 btrfs_set_max_workers(&fs_info->readahead_workers, new_pool_size);
1261 btrfs_set_max_workers(&fs_info->scrub_wr_completion_workers,
1265 static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
1267 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1270 static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1271 unsigned long old_opts, int flags)
1273 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1274 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1275 (flags & MS_RDONLY))) {
1276 /* wait for any defraggers to finish */
1277 wait_event(fs_info->transaction_wait,
1278 (atomic_read(&fs_info->defrag_running) == 0));
1279 if (flags & MS_RDONLY)
1280 sync_filesystem(fs_info->sb);
1284 static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1285 unsigned long old_opts)
1288 * We need cleanup all defragable inodes if the autodefragment is
1289 * close or the fs is R/O.
1291 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1292 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1293 (fs_info->sb->s_flags & MS_RDONLY))) {
1294 btrfs_cleanup_defrag_inodes(fs_info);
1297 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1300 static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1302 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1303 struct btrfs_root *root = fs_info->tree_root;
1304 unsigned old_flags = sb->s_flags;
1305 unsigned long old_opts = fs_info->mount_opt;
1306 unsigned long old_compress_type = fs_info->compress_type;
1307 u64 old_max_inline = fs_info->max_inline;
1308 u64 old_alloc_start = fs_info->alloc_start;
1309 int old_thread_pool_size = fs_info->thread_pool_size;
1310 unsigned int old_metadata_ratio = fs_info->metadata_ratio;
1313 btrfs_remount_prepare(fs_info);
1315 ret = btrfs_parse_options(root, data);
1321 btrfs_remount_begin(fs_info, old_opts, *flags);
1322 btrfs_resize_thread_pool(fs_info,
1323 fs_info->thread_pool_size, old_thread_pool_size);
1325 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
1328 if (*flags & MS_RDONLY) {
1330 * this also happens on 'umount -rf' or on shutdown, when
1331 * the filesystem is busy.
1333 sb->s_flags |= MS_RDONLY;
1335 btrfs_dev_replace_suspend_for_unmount(fs_info);
1336 btrfs_scrub_cancel(fs_info);
1337 btrfs_pause_balance(fs_info);
1339 ret = btrfs_commit_super(root);
1343 if (fs_info->fs_devices->rw_devices == 0) {
1348 if (fs_info->fs_devices->missing_devices >
1349 fs_info->num_tolerated_disk_barrier_failures &&
1350 !(*flags & MS_RDONLY)) {
1352 "Btrfs: too many missing devices, writeable remount is not allowed\n");
1357 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
1362 ret = btrfs_cleanup_fs_roots(fs_info);
1366 /* recover relocation */
1367 ret = btrfs_recover_relocation(root);
1371 ret = btrfs_resume_balance_async(fs_info);
1375 ret = btrfs_resume_dev_replace_async(fs_info);
1377 pr_warn("btrfs: failed to resume dev_replace\n");
1380 sb->s_flags &= ~MS_RDONLY;
1383 btrfs_remount_cleanup(fs_info, old_opts);
1387 /* We've hit an error - don't reset MS_RDONLY */
1388 if (sb->s_flags & MS_RDONLY)
1389 old_flags |= MS_RDONLY;
1390 sb->s_flags = old_flags;
1391 fs_info->mount_opt = old_opts;
1392 fs_info->compress_type = old_compress_type;
1393 fs_info->max_inline = old_max_inline;
1394 mutex_lock(&fs_info->chunk_mutex);
1395 fs_info->alloc_start = old_alloc_start;
1396 mutex_unlock(&fs_info->chunk_mutex);
1397 btrfs_resize_thread_pool(fs_info,
1398 old_thread_pool_size, fs_info->thread_pool_size);
1399 fs_info->metadata_ratio = old_metadata_ratio;
1400 btrfs_remount_cleanup(fs_info, old_opts);
1404 /* Used to sort the devices by max_avail(descending sort) */
1405 static int btrfs_cmp_device_free_bytes(const void *dev_info1,
1406 const void *dev_info2)
1408 if (((struct btrfs_device_info *)dev_info1)->max_avail >
1409 ((struct btrfs_device_info *)dev_info2)->max_avail)
1411 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
1412 ((struct btrfs_device_info *)dev_info2)->max_avail)
1419 * sort the devices by max_avail, in which max free extent size of each device
1420 * is stored.(Descending Sort)
1422 static inline void btrfs_descending_sort_devices(
1423 struct btrfs_device_info *devices,
1426 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
1427 btrfs_cmp_device_free_bytes, NULL);
1431 * The helper to calc the free space on the devices that can be used to store
1434 static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
1436 struct btrfs_fs_info *fs_info = root->fs_info;
1437 struct btrfs_device_info *devices_info;
1438 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
1439 struct btrfs_device *device;
1444 u64 min_stripe_size;
1445 int min_stripes = 1, num_stripes = 1;
1446 int i = 0, nr_devices;
1449 nr_devices = fs_info->fs_devices->open_devices;
1450 BUG_ON(!nr_devices);
1452 devices_info = kmalloc(sizeof(*devices_info) * nr_devices,
1457 /* calc min stripe number for data space alloction */
1458 type = btrfs_get_alloc_profile(root, 1);
1459 if (type & BTRFS_BLOCK_GROUP_RAID0) {
1461 num_stripes = nr_devices;
1462 } else if (type & BTRFS_BLOCK_GROUP_RAID1) {
1465 } else if (type & BTRFS_BLOCK_GROUP_RAID10) {
1470 if (type & BTRFS_BLOCK_GROUP_DUP)
1471 min_stripe_size = 2 * BTRFS_STRIPE_LEN;
1473 min_stripe_size = BTRFS_STRIPE_LEN;
1475 list_for_each_entry(device, &fs_devices->devices, dev_list) {
1476 if (!device->in_fs_metadata || !device->bdev ||
1477 device->is_tgtdev_for_dev_replace)
1480 avail_space = device->total_bytes - device->bytes_used;
1482 /* align with stripe_len */
1483 do_div(avail_space, BTRFS_STRIPE_LEN);
1484 avail_space *= BTRFS_STRIPE_LEN;
1487 * In order to avoid overwritting the superblock on the drive,
1488 * btrfs starts at an offset of at least 1MB when doing chunk
1491 skip_space = 1024 * 1024;
1493 /* user can set the offset in fs_info->alloc_start. */
1494 if (fs_info->alloc_start + BTRFS_STRIPE_LEN <=
1495 device->total_bytes)
1496 skip_space = max(fs_info->alloc_start, skip_space);
1499 * btrfs can not use the free space in [0, skip_space - 1],
1500 * we must subtract it from the total. In order to implement
1501 * it, we account the used space in this range first.
1503 ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1,
1506 kfree(devices_info);
1510 /* calc the free space in [0, skip_space - 1] */
1511 skip_space -= used_space;
1514 * we can use the free space in [0, skip_space - 1], subtract
1515 * it from the total.
1517 if (avail_space && avail_space >= skip_space)
1518 avail_space -= skip_space;
1522 if (avail_space < min_stripe_size)
1525 devices_info[i].dev = device;
1526 devices_info[i].max_avail = avail_space;
1533 btrfs_descending_sort_devices(devices_info, nr_devices);
1537 while (nr_devices >= min_stripes) {
1538 if (num_stripes > nr_devices)
1539 num_stripes = nr_devices;
1541 if (devices_info[i].max_avail >= min_stripe_size) {
1545 avail_space += devices_info[i].max_avail * num_stripes;
1546 alloc_size = devices_info[i].max_avail;
1547 for (j = i + 1 - num_stripes; j <= i; j++)
1548 devices_info[j].max_avail -= alloc_size;
1554 kfree(devices_info);
1555 *free_bytes = avail_space;
1559 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
1561 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
1562 struct btrfs_super_block *disk_super = fs_info->super_copy;
1563 struct list_head *head = &fs_info->space_info;
1564 struct btrfs_space_info *found;
1566 u64 total_free_data = 0;
1567 int bits = dentry->d_sb->s_blocksize_bits;
1568 __be32 *fsid = (__be32 *)fs_info->fsid;
1571 /* holding chunk_muext to avoid allocating new chunks */
1572 mutex_lock(&fs_info->chunk_mutex);
1574 list_for_each_entry_rcu(found, head, list) {
1575 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
1576 total_free_data += found->disk_total - found->disk_used;
1578 btrfs_account_ro_block_groups_free_space(found);
1581 total_used += found->disk_used;
1585 buf->f_namelen = BTRFS_NAME_LEN;
1586 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
1587 buf->f_bfree = buf->f_blocks - (total_used >> bits);
1588 buf->f_bsize = dentry->d_sb->s_blocksize;
1589 buf->f_type = BTRFS_SUPER_MAGIC;
1590 buf->f_bavail = total_free_data;
1591 ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data);
1593 mutex_unlock(&fs_info->chunk_mutex);
1596 buf->f_bavail += total_free_data;
1597 buf->f_bavail = buf->f_bavail >> bits;
1598 mutex_unlock(&fs_info->chunk_mutex);
1600 /* We treat it as constant endianness (it doesn't matter _which_)
1601 because we want the fsid to come out the same whether mounted
1602 on a big-endian or little-endian host */
1603 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
1604 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
1605 /* Mask in the root object ID too, to disambiguate subvols */
1606 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
1607 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
1612 static void btrfs_kill_super(struct super_block *sb)
1614 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1615 kill_anon_super(sb);
1616 free_fs_info(fs_info);
1619 static struct file_system_type btrfs_fs_type = {
1620 .owner = THIS_MODULE,
1622 .mount = btrfs_mount,
1623 .kill_sb = btrfs_kill_super,
1624 .fs_flags = FS_REQUIRES_DEV,
1626 MODULE_ALIAS_FS("btrfs");
1629 * used by btrfsctl to scan devices when no FS is mounted
1631 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
1634 struct btrfs_ioctl_vol_args *vol;
1635 struct btrfs_fs_devices *fs_devices;
1638 if (!capable(CAP_SYS_ADMIN))
1641 vol = memdup_user((void __user *)arg, sizeof(*vol));
1643 return PTR_ERR(vol);
1646 case BTRFS_IOC_SCAN_DEV:
1647 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1648 &btrfs_fs_type, &fs_devices);
1650 case BTRFS_IOC_DEVICES_READY:
1651 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1652 &btrfs_fs_type, &fs_devices);
1655 ret = !(fs_devices->num_devices == fs_devices->total_devices);
1663 static int btrfs_freeze(struct super_block *sb)
1665 struct btrfs_trans_handle *trans;
1666 struct btrfs_root *root = btrfs_sb(sb)->tree_root;
1668 trans = btrfs_attach_transaction_barrier(root);
1669 if (IS_ERR(trans)) {
1670 /* no transaction, don't bother */
1671 if (PTR_ERR(trans) == -ENOENT)
1673 return PTR_ERR(trans);
1675 return btrfs_commit_transaction(trans, root);
1678 static int btrfs_unfreeze(struct super_block *sb)
1683 static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
1685 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
1686 struct btrfs_fs_devices *cur_devices;
1687 struct btrfs_device *dev, *first_dev = NULL;
1688 struct list_head *head;
1689 struct rcu_string *name;
1691 mutex_lock(&fs_info->fs_devices->device_list_mutex);
1692 cur_devices = fs_info->fs_devices;
1693 while (cur_devices) {
1694 head = &cur_devices->devices;
1695 list_for_each_entry(dev, head, dev_list) {
1698 if (!first_dev || dev->devid < first_dev->devid)
1701 cur_devices = cur_devices->seed;
1706 name = rcu_dereference(first_dev->name);
1707 seq_escape(m, name->str, " \t\n\\");
1712 mutex_unlock(&fs_info->fs_devices->device_list_mutex);
1716 static const struct super_operations btrfs_super_ops = {
1717 .drop_inode = btrfs_drop_inode,
1718 .evict_inode = btrfs_evict_inode,
1719 .put_super = btrfs_put_super,
1720 .sync_fs = btrfs_sync_fs,
1721 .show_options = btrfs_show_options,
1722 .show_devname = btrfs_show_devname,
1723 .write_inode = btrfs_write_inode,
1724 .alloc_inode = btrfs_alloc_inode,
1725 .destroy_inode = btrfs_destroy_inode,
1726 .statfs = btrfs_statfs,
1727 .remount_fs = btrfs_remount,
1728 .freeze_fs = btrfs_freeze,
1729 .unfreeze_fs = btrfs_unfreeze,
1732 static const struct file_operations btrfs_ctl_fops = {
1733 .unlocked_ioctl = btrfs_control_ioctl,
1734 .compat_ioctl = btrfs_control_ioctl,
1735 .owner = THIS_MODULE,
1736 .llseek = noop_llseek,
1739 static struct miscdevice btrfs_misc = {
1740 .minor = BTRFS_MINOR,
1741 .name = "btrfs-control",
1742 .fops = &btrfs_ctl_fops
1745 MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
1746 MODULE_ALIAS("devname:btrfs-control");
1748 static int btrfs_interface_init(void)
1750 return misc_register(&btrfs_misc);
1753 static void btrfs_interface_exit(void)
1755 if (misc_deregister(&btrfs_misc) < 0)
1756 printk(KERN_INFO "btrfs: misc_deregister failed for control device\n");
1759 static void btrfs_print_info(void)
1761 printk(KERN_INFO "Btrfs loaded"
1762 #ifdef CONFIG_BTRFS_DEBUG
1765 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1766 ", integrity-checker=on"
1771 static int btrfs_run_sanity_tests(void)
1773 return btrfs_test_free_space_cache();
1776 static int __init init_btrfs_fs(void)
1780 err = btrfs_init_sysfs();
1784 btrfs_init_compress();
1786 err = btrfs_init_cachep();
1790 err = extent_io_init();
1794 err = extent_map_init();
1796 goto free_extent_io;
1798 err = ordered_data_init();
1800 goto free_extent_map;
1802 err = btrfs_delayed_inode_init();
1804 goto free_ordered_data;
1806 err = btrfs_auto_defrag_init();
1808 goto free_delayed_inode;
1810 err = btrfs_delayed_ref_init();
1812 goto free_auto_defrag;
1814 err = btrfs_prelim_ref_init();
1816 goto free_prelim_ref;
1818 err = btrfs_interface_init();
1820 goto free_delayed_ref;
1822 btrfs_init_lockdep();
1826 err = btrfs_run_sanity_tests();
1828 goto unregister_ioctl;
1830 err = register_filesystem(&btrfs_fs_type);
1832 goto unregister_ioctl;
1837 btrfs_interface_exit();
1839 btrfs_prelim_ref_exit();
1841 btrfs_delayed_ref_exit();
1843 btrfs_auto_defrag_exit();
1845 btrfs_delayed_inode_exit();
1847 ordered_data_exit();
1853 btrfs_destroy_cachep();
1855 btrfs_exit_compress();
1860 static void __exit exit_btrfs_fs(void)
1862 btrfs_destroy_cachep();
1863 btrfs_delayed_ref_exit();
1864 btrfs_auto_defrag_exit();
1865 btrfs_delayed_inode_exit();
1866 btrfs_prelim_ref_exit();
1867 ordered_data_exit();
1870 btrfs_interface_exit();
1871 unregister_filesystem(&btrfs_fs_type);
1873 btrfs_cleanup_fs_uuids();
1874 btrfs_exit_compress();
1877 module_init(init_btrfs_fs)
1878 module_exit(exit_btrfs_fs)
1880 MODULE_LICENSE("GPL");