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>
45 #include "delayed-inode.h"
48 #include "transaction.h"
49 #include "btrfs_inode.h"
51 #include "print-tree.h"
56 #include "compression.h"
58 #define CREATE_TRACE_POINTS
59 #include <trace/events/btrfs.h>
61 static const struct super_operations btrfs_super_ops
;
62 static struct file_system_type btrfs_fs_type
;
64 static const char *btrfs_decode_error(struct btrfs_fs_info
*fs_info
, int errno
,
71 errstr
= "IO failure";
74 errstr
= "Out of memory";
77 errstr
= "Readonly filesystem";
80 errstr
= "Object already exists";
84 if (snprintf(nbuf
, 16, "error %d", -errno
) >= 0)
93 static void __save_error_info(struct btrfs_fs_info
*fs_info
)
96 * today we only save the error info into ram. Long term we'll
97 * also send it down to the disk
99 fs_info
->fs_state
= BTRFS_SUPER_FLAG_ERROR
;
103 * We move write_super stuff at umount in order to avoid deadlock
104 * for umount hold all lock.
106 static void save_error_info(struct btrfs_fs_info
*fs_info
)
108 __save_error_info(fs_info
);
111 /* btrfs handle error by forcing the filesystem readonly */
112 static void btrfs_handle_error(struct btrfs_fs_info
*fs_info
)
114 struct super_block
*sb
= fs_info
->sb
;
116 if (sb
->s_flags
& MS_RDONLY
)
119 if (fs_info
->fs_state
& BTRFS_SUPER_FLAG_ERROR
) {
120 sb
->s_flags
|= MS_RDONLY
;
121 printk(KERN_INFO
"btrfs is forced readonly\n");
126 * __btrfs_std_error decodes expected errors from the caller and
127 * invokes the approciate error response.
129 void __btrfs_std_error(struct btrfs_fs_info
*fs_info
, const char *function
,
130 unsigned int line
, int errno
)
132 struct super_block
*sb
= fs_info
->sb
;
137 * Special case: if the error is EROFS, and we're already
138 * under MS_RDONLY, then it is safe here.
140 if (errno
== -EROFS
&& (sb
->s_flags
& MS_RDONLY
))
143 errstr
= btrfs_decode_error(fs_info
, errno
, nbuf
);
144 printk(KERN_CRIT
"BTRFS error (device %s) in %s:%d: %s\n",
145 sb
->s_id
, function
, line
, errstr
);
146 save_error_info(fs_info
);
148 btrfs_handle_error(fs_info
);
152 * __btrfs_panic decodes unexpected, fatal errors from the caller,
153 * issues an alert, and either panics or BUGs, depending on mount options.
155 void __btrfs_panic(struct btrfs_fs_info
*fs_info
, const char *function
,
156 unsigned int line
, int errno
, const char *fmt
, ...)
159 char *s_id
= "<unknown>";
161 struct va_format vaf
= { .fmt
= fmt
};
165 s_id
= fs_info
->sb
->s_id
;
170 errstr
= btrfs_decode_error(fs_info
, errno
, nbuf
);
171 if (fs_info
->mount_opt
& BTRFS_MOUNT_PANIC_ON_FATAL_ERROR
)
172 panic(KERN_CRIT
"BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
173 s_id
, function
, line
, &vaf
, errstr
);
175 printk(KERN_CRIT
"BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
176 s_id
, function
, line
, &vaf
, errstr
);
178 /* Caller calls BUG() */
181 static void btrfs_put_super(struct super_block
*sb
)
183 (void)close_ctree(btrfs_sb(sb
)->tree_root
);
184 /* FIXME: need to fix VFS to return error? */
185 /* AV: return it _where_? ->put_super() can be triggered by any number
186 * of async events, up to and including delivery of SIGKILL to the
187 * last process that kept it busy. Or segfault in the aforementioned
188 * process... Whom would you report that to?
193 Opt_degraded
, Opt_subvol
, Opt_subvolid
, Opt_device
, Opt_nodatasum
,
194 Opt_nodatacow
, Opt_max_inline
, Opt_alloc_start
, Opt_nobarrier
, Opt_ssd
,
195 Opt_nossd
, Opt_ssd_spread
, Opt_thread_pool
, Opt_noacl
, Opt_compress
,
196 Opt_compress_type
, Opt_compress_force
, Opt_compress_force_type
,
197 Opt_notreelog
, Opt_ratio
, Opt_flushoncommit
, Opt_discard
,
198 Opt_space_cache
, Opt_clear_cache
, Opt_user_subvol_rm_allowed
,
199 Opt_enospc_debug
, Opt_subvolrootid
, Opt_defrag
, Opt_inode_cache
,
200 Opt_no_space_cache
, Opt_recovery
, Opt_skip_balance
,
201 Opt_check_integrity
, Opt_check_integrity_including_extent_data
,
202 Opt_check_integrity_print_mask
, Opt_fatal_errors
,
206 static match_table_t tokens
= {
207 {Opt_degraded
, "degraded"},
208 {Opt_subvol
, "subvol=%s"},
209 {Opt_subvolid
, "subvolid=%d"},
210 {Opt_device
, "device=%s"},
211 {Opt_nodatasum
, "nodatasum"},
212 {Opt_nodatacow
, "nodatacow"},
213 {Opt_nobarrier
, "nobarrier"},
214 {Opt_max_inline
, "max_inline=%s"},
215 {Opt_alloc_start
, "alloc_start=%s"},
216 {Opt_thread_pool
, "thread_pool=%d"},
217 {Opt_compress
, "compress"},
218 {Opt_compress_type
, "compress=%s"},
219 {Opt_compress_force
, "compress-force"},
220 {Opt_compress_force_type
, "compress-force=%s"},
222 {Opt_ssd_spread
, "ssd_spread"},
223 {Opt_nossd
, "nossd"},
224 {Opt_noacl
, "noacl"},
225 {Opt_notreelog
, "notreelog"},
226 {Opt_flushoncommit
, "flushoncommit"},
227 {Opt_ratio
, "metadata_ratio=%d"},
228 {Opt_discard
, "discard"},
229 {Opt_space_cache
, "space_cache"},
230 {Opt_clear_cache
, "clear_cache"},
231 {Opt_user_subvol_rm_allowed
, "user_subvol_rm_allowed"},
232 {Opt_enospc_debug
, "enospc_debug"},
233 {Opt_subvolrootid
, "subvolrootid=%d"},
234 {Opt_defrag
, "autodefrag"},
235 {Opt_inode_cache
, "inode_cache"},
236 {Opt_no_space_cache
, "nospace_cache"},
237 {Opt_recovery
, "recovery"},
238 {Opt_skip_balance
, "skip_balance"},
239 {Opt_check_integrity
, "check_int"},
240 {Opt_check_integrity_including_extent_data
, "check_int_data"},
241 {Opt_check_integrity_print_mask
, "check_int_print_mask=%d"},
242 {Opt_fatal_errors
, "fatal_errors=%s"},
247 * Regular mount options parser. Everything that is needed only when
248 * reading in a new superblock is parsed here.
250 int btrfs_parse_options(struct btrfs_root
*root
, char *options
)
252 struct btrfs_fs_info
*info
= root
->fs_info
;
253 substring_t args
[MAX_OPT_ARGS
];
254 char *p
, *num
, *orig
= NULL
;
259 bool compress_force
= false;
261 cache_gen
= btrfs_super_cache_generation(root
->fs_info
->super_copy
);
263 btrfs_set_opt(info
->mount_opt
, SPACE_CACHE
);
269 * strsep changes the string, duplicate it because parse_options
272 options
= kstrdup(options
, GFP_NOFS
);
278 while ((p
= strsep(&options
, ",")) != NULL
) {
283 token
= match_token(p
, tokens
, args
);
286 printk(KERN_INFO
"btrfs: allowing degraded mounts\n");
287 btrfs_set_opt(info
->mount_opt
, DEGRADED
);
291 case Opt_subvolrootid
:
294 * These are parsed by btrfs_parse_early_options
295 * and can be happily ignored here.
299 printk(KERN_INFO
"btrfs: setting nodatasum\n");
300 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
303 printk(KERN_INFO
"btrfs: setting nodatacow\n");
304 btrfs_set_opt(info
->mount_opt
, NODATACOW
);
305 btrfs_set_opt(info
->mount_opt
, NODATASUM
);
307 case Opt_compress_force
:
308 case Opt_compress_force_type
:
309 compress_force
= true;
311 case Opt_compress_type
:
312 if (token
== Opt_compress
||
313 token
== Opt_compress_force
||
314 strcmp(args
[0].from
, "zlib") == 0) {
315 compress_type
= "zlib";
316 info
->compress_type
= BTRFS_COMPRESS_ZLIB
;
317 } else if (strcmp(args
[0].from
, "lzo") == 0) {
318 compress_type
= "lzo";
319 info
->compress_type
= BTRFS_COMPRESS_LZO
;
325 btrfs_set_opt(info
->mount_opt
, COMPRESS
);
326 if (compress_force
) {
327 btrfs_set_opt(info
->mount_opt
, FORCE_COMPRESS
);
328 pr_info("btrfs: force %s compression\n",
331 pr_info("btrfs: use %s compression\n",
335 printk(KERN_INFO
"btrfs: use ssd allocation scheme\n");
336 btrfs_set_opt(info
->mount_opt
, SSD
);
339 printk(KERN_INFO
"btrfs: use spread ssd "
340 "allocation scheme\n");
341 btrfs_set_opt(info
->mount_opt
, SSD
);
342 btrfs_set_opt(info
->mount_opt
, SSD_SPREAD
);
345 printk(KERN_INFO
"btrfs: not using ssd allocation "
347 btrfs_set_opt(info
->mount_opt
, NOSSD
);
348 btrfs_clear_opt(info
->mount_opt
, SSD
);
349 btrfs_clear_opt(info
->mount_opt
, SSD_SPREAD
);
352 printk(KERN_INFO
"btrfs: turning off barriers\n");
353 btrfs_set_opt(info
->mount_opt
, NOBARRIER
);
355 case Opt_thread_pool
:
357 match_int(&args
[0], &intarg
);
359 info
->thread_pool_size
= intarg
;
360 printk(KERN_INFO
"btrfs: thread pool %d\n",
361 info
->thread_pool_size
);
365 num
= match_strdup(&args
[0]);
367 info
->max_inline
= memparse(num
, NULL
);
370 if (info
->max_inline
) {
371 info
->max_inline
= max_t(u64
,
375 printk(KERN_INFO
"btrfs: max_inline at %llu\n",
376 (unsigned long long)info
->max_inline
);
379 case Opt_alloc_start
:
380 num
= match_strdup(&args
[0]);
382 info
->alloc_start
= memparse(num
, NULL
);
385 "btrfs: allocations start at %llu\n",
386 (unsigned long long)info
->alloc_start
);
390 root
->fs_info
->sb
->s_flags
&= ~MS_POSIXACL
;
393 printk(KERN_INFO
"btrfs: disabling tree log\n");
394 btrfs_set_opt(info
->mount_opt
, NOTREELOG
);
396 case Opt_flushoncommit
:
397 printk(KERN_INFO
"btrfs: turning on flush-on-commit\n");
398 btrfs_set_opt(info
->mount_opt
, FLUSHONCOMMIT
);
402 match_int(&args
[0], &intarg
);
404 info
->metadata_ratio
= intarg
;
405 printk(KERN_INFO
"btrfs: metadata ratio %d\n",
406 info
->metadata_ratio
);
410 btrfs_set_opt(info
->mount_opt
, DISCARD
);
412 case Opt_space_cache
:
413 btrfs_set_opt(info
->mount_opt
, SPACE_CACHE
);
415 case Opt_no_space_cache
:
416 printk(KERN_INFO
"btrfs: disabling disk space caching\n");
417 btrfs_clear_opt(info
->mount_opt
, SPACE_CACHE
);
419 case Opt_inode_cache
:
420 printk(KERN_INFO
"btrfs: enabling inode map caching\n");
421 btrfs_set_opt(info
->mount_opt
, INODE_MAP_CACHE
);
423 case Opt_clear_cache
:
424 printk(KERN_INFO
"btrfs: force clearing of disk cache\n");
425 btrfs_set_opt(info
->mount_opt
, CLEAR_CACHE
);
427 case Opt_user_subvol_rm_allowed
:
428 btrfs_set_opt(info
->mount_opt
, USER_SUBVOL_RM_ALLOWED
);
430 case Opt_enospc_debug
:
431 btrfs_set_opt(info
->mount_opt
, ENOSPC_DEBUG
);
434 printk(KERN_INFO
"btrfs: enabling auto defrag");
435 btrfs_set_opt(info
->mount_opt
, AUTO_DEFRAG
);
438 printk(KERN_INFO
"btrfs: enabling auto recovery");
439 btrfs_set_opt(info
->mount_opt
, RECOVERY
);
441 case Opt_skip_balance
:
442 btrfs_set_opt(info
->mount_opt
, SKIP_BALANCE
);
444 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
445 case Opt_check_integrity_including_extent_data
:
446 printk(KERN_INFO
"btrfs: enabling check integrity"
447 " including extent data\n");
448 btrfs_set_opt(info
->mount_opt
,
449 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA
);
450 btrfs_set_opt(info
->mount_opt
, CHECK_INTEGRITY
);
452 case Opt_check_integrity
:
453 printk(KERN_INFO
"btrfs: enabling check integrity\n");
454 btrfs_set_opt(info
->mount_opt
, CHECK_INTEGRITY
);
456 case Opt_check_integrity_print_mask
:
458 match_int(&args
[0], &intarg
);
460 info
->check_integrity_print_mask
= intarg
;
461 printk(KERN_INFO
"btrfs:"
462 " check_integrity_print_mask 0x%x\n",
463 info
->check_integrity_print_mask
);
467 case Opt_check_integrity_including_extent_data
:
468 case Opt_check_integrity
:
469 case Opt_check_integrity_print_mask
:
470 printk(KERN_ERR
"btrfs: support for check_integrity*"
471 " not compiled in!\n");
475 case Opt_fatal_errors
:
476 if (strcmp(args
[0].from
, "panic") == 0)
477 btrfs_set_opt(info
->mount_opt
,
478 PANIC_ON_FATAL_ERROR
);
479 else if (strcmp(args
[0].from
, "bug") == 0)
480 btrfs_clear_opt(info
->mount_opt
,
481 PANIC_ON_FATAL_ERROR
);
488 printk(KERN_INFO
"btrfs: unrecognized mount option "
497 if (!ret
&& btrfs_test_opt(root
, SPACE_CACHE
))
498 printk(KERN_INFO
"btrfs: disk space caching is enabled\n");
504 * Parse mount options that are required early in the mount process.
506 * All other options will be parsed on much later in the mount process and
507 * only when we need to allocate a new super block.
509 static int btrfs_parse_early_options(const char *options
, fmode_t flags
,
510 void *holder
, char **subvol_name
, u64
*subvol_objectid
,
511 u64
*subvol_rootid
, struct btrfs_fs_devices
**fs_devices
)
513 substring_t args
[MAX_OPT_ARGS
];
514 char *device_name
, *opts
, *orig
, *p
;
522 * strsep changes the string, duplicate it because parse_options
525 opts
= kstrdup(options
, GFP_KERNEL
);
530 while ((p
= strsep(&opts
, ",")) != NULL
) {
535 token
= match_token(p
, tokens
, args
);
539 *subvol_name
= match_strdup(&args
[0]);
543 error
= match_int(&args
[0], &intarg
);
545 /* we want the original fs_tree */
548 BTRFS_FS_TREE_OBJECTID
;
550 *subvol_objectid
= intarg
;
553 case Opt_subvolrootid
:
555 error
= match_int(&args
[0], &intarg
);
557 /* we want the original fs_tree */
560 BTRFS_FS_TREE_OBJECTID
;
562 *subvol_rootid
= intarg
;
566 device_name
= match_strdup(&args
[0]);
571 error
= btrfs_scan_one_device(device_name
,
572 flags
, holder
, fs_devices
);
587 static struct dentry
*get_default_root(struct super_block
*sb
,
590 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
591 struct btrfs_root
*root
= fs_info
->tree_root
;
592 struct btrfs_root
*new_root
;
593 struct btrfs_dir_item
*di
;
594 struct btrfs_path
*path
;
595 struct btrfs_key location
;
601 * We have a specific subvol we want to mount, just setup location and
602 * go look up the root.
604 if (subvol_objectid
) {
605 location
.objectid
= subvol_objectid
;
606 location
.type
= BTRFS_ROOT_ITEM_KEY
;
607 location
.offset
= (u64
)-1;
611 path
= btrfs_alloc_path();
613 return ERR_PTR(-ENOMEM
);
614 path
->leave_spinning
= 1;
617 * Find the "default" dir item which points to the root item that we
618 * will mount by default if we haven't been given a specific subvolume
621 dir_id
= btrfs_super_root_dir(fs_info
->super_copy
);
622 di
= btrfs_lookup_dir_item(NULL
, root
, path
, dir_id
, "default", 7, 0);
624 btrfs_free_path(path
);
629 * Ok the default dir item isn't there. This is weird since
630 * it's always been there, but don't freak out, just try and
631 * mount to root most subvolume.
633 btrfs_free_path(path
);
634 dir_id
= BTRFS_FIRST_FREE_OBJECTID
;
635 new_root
= fs_info
->fs_root
;
639 btrfs_dir_item_key_to_cpu(path
->nodes
[0], di
, &location
);
640 btrfs_free_path(path
);
643 new_root
= btrfs_read_fs_root_no_name(fs_info
, &location
);
644 if (IS_ERR(new_root
))
645 return ERR_CAST(new_root
);
647 if (btrfs_root_refs(&new_root
->root_item
) == 0)
648 return ERR_PTR(-ENOENT
);
650 dir_id
= btrfs_root_dirid(&new_root
->root_item
);
652 location
.objectid
= dir_id
;
653 location
.type
= BTRFS_INODE_ITEM_KEY
;
656 inode
= btrfs_iget(sb
, &location
, new_root
, &new);
658 return ERR_CAST(inode
);
661 * If we're just mounting the root most subvol put the inode and return
662 * a reference to the dentry. We will have already gotten a reference
663 * to the inode in btrfs_fill_super so we're good to go.
665 if (!new && sb
->s_root
->d_inode
== inode
) {
667 return dget(sb
->s_root
);
670 return d_obtain_alias(inode
);
673 static int btrfs_fill_super(struct super_block
*sb
,
674 struct btrfs_fs_devices
*fs_devices
,
675 void *data
, int silent
)
678 struct dentry
*root_dentry
;
679 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
680 struct btrfs_key key
;
683 sb
->s_maxbytes
= MAX_LFS_FILESIZE
;
684 sb
->s_magic
= BTRFS_SUPER_MAGIC
;
685 sb
->s_op
= &btrfs_super_ops
;
686 sb
->s_d_op
= &btrfs_dentry_operations
;
687 sb
->s_export_op
= &btrfs_export_ops
;
688 sb
->s_xattr
= btrfs_xattr_handlers
;
690 #ifdef CONFIG_BTRFS_FS_POSIX_ACL
691 sb
->s_flags
|= MS_POSIXACL
;
694 err
= open_ctree(sb
, fs_devices
, (char *)data
);
696 printk("btrfs: open_ctree failed\n");
700 key
.objectid
= BTRFS_FIRST_FREE_OBJECTID
;
701 key
.type
= BTRFS_INODE_ITEM_KEY
;
703 inode
= btrfs_iget(sb
, &key
, fs_info
->fs_root
, NULL
);
705 err
= PTR_ERR(inode
);
709 root_dentry
= d_alloc_root(inode
);
716 sb
->s_root
= root_dentry
;
718 save_mount_options(sb
, data
);
719 cleancache_init_fs(sb
);
720 sb
->s_flags
|= MS_ACTIVE
;
724 close_ctree(fs_info
->tree_root
);
728 int btrfs_sync_fs(struct super_block
*sb
, int wait
)
730 struct btrfs_trans_handle
*trans
;
731 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
732 struct btrfs_root
*root
= fs_info
->tree_root
;
735 trace_btrfs_sync_fs(wait
);
738 filemap_flush(fs_info
->btree_inode
->i_mapping
);
742 btrfs_start_delalloc_inodes(root
, 0);
743 btrfs_wait_ordered_extents(root
, 0, 0);
745 trans
= btrfs_start_transaction(root
, 0);
747 return PTR_ERR(trans
);
748 ret
= btrfs_commit_transaction(trans
, root
);
752 static int btrfs_show_options(struct seq_file
*seq
, struct dentry
*dentry
)
754 struct btrfs_fs_info
*info
= btrfs_sb(dentry
->d_sb
);
755 struct btrfs_root
*root
= info
->tree_root
;
758 if (btrfs_test_opt(root
, DEGRADED
))
759 seq_puts(seq
, ",degraded");
760 if (btrfs_test_opt(root
, NODATASUM
))
761 seq_puts(seq
, ",nodatasum");
762 if (btrfs_test_opt(root
, NODATACOW
))
763 seq_puts(seq
, ",nodatacow");
764 if (btrfs_test_opt(root
, NOBARRIER
))
765 seq_puts(seq
, ",nobarrier");
766 if (info
->max_inline
!= 8192 * 1024)
767 seq_printf(seq
, ",max_inline=%llu",
768 (unsigned long long)info
->max_inline
);
769 if (info
->alloc_start
!= 0)
770 seq_printf(seq
, ",alloc_start=%llu",
771 (unsigned long long)info
->alloc_start
);
772 if (info
->thread_pool_size
!= min_t(unsigned long,
773 num_online_cpus() + 2, 8))
774 seq_printf(seq
, ",thread_pool=%d", info
->thread_pool_size
);
775 if (btrfs_test_opt(root
, COMPRESS
)) {
776 if (info
->compress_type
== BTRFS_COMPRESS_ZLIB
)
777 compress_type
= "zlib";
779 compress_type
= "lzo";
780 if (btrfs_test_opt(root
, FORCE_COMPRESS
))
781 seq_printf(seq
, ",compress-force=%s", compress_type
);
783 seq_printf(seq
, ",compress=%s", compress_type
);
785 if (btrfs_test_opt(root
, NOSSD
))
786 seq_puts(seq
, ",nossd");
787 if (btrfs_test_opt(root
, SSD_SPREAD
))
788 seq_puts(seq
, ",ssd_spread");
789 else if (btrfs_test_opt(root
, SSD
))
790 seq_puts(seq
, ",ssd");
791 if (btrfs_test_opt(root
, NOTREELOG
))
792 seq_puts(seq
, ",notreelog");
793 if (btrfs_test_opt(root
, FLUSHONCOMMIT
))
794 seq_puts(seq
, ",flushoncommit");
795 if (btrfs_test_opt(root
, DISCARD
))
796 seq_puts(seq
, ",discard");
797 if (!(root
->fs_info
->sb
->s_flags
& MS_POSIXACL
))
798 seq_puts(seq
, ",noacl");
799 if (btrfs_test_opt(root
, SPACE_CACHE
))
800 seq_puts(seq
, ",space_cache");
802 seq_puts(seq
, ",nospace_cache");
803 if (btrfs_test_opt(root
, CLEAR_CACHE
))
804 seq_puts(seq
, ",clear_cache");
805 if (btrfs_test_opt(root
, USER_SUBVOL_RM_ALLOWED
))
806 seq_puts(seq
, ",user_subvol_rm_allowed");
807 if (btrfs_test_opt(root
, ENOSPC_DEBUG
))
808 seq_puts(seq
, ",enospc_debug");
809 if (btrfs_test_opt(root
, AUTO_DEFRAG
))
810 seq_puts(seq
, ",autodefrag");
811 if (btrfs_test_opt(root
, INODE_MAP_CACHE
))
812 seq_puts(seq
, ",inode_cache");
813 if (btrfs_test_opt(root
, SKIP_BALANCE
))
814 seq_puts(seq
, ",skip_balance");
815 if (btrfs_test_opt(root
, PANIC_ON_FATAL_ERROR
))
816 seq_puts(seq
, ",fatal_errors=panic");
820 static int btrfs_test_super(struct super_block
*s
, void *data
)
822 struct btrfs_fs_info
*p
= data
;
823 struct btrfs_fs_info
*fs_info
= btrfs_sb(s
);
825 return fs_info
->fs_devices
== p
->fs_devices
;
828 static int btrfs_set_super(struct super_block
*s
, void *data
)
830 int err
= set_anon_super(s
, data
);
837 * subvolumes are identified by ino 256
839 static inline int is_subvolume_inode(struct inode
*inode
)
841 if (inode
&& inode
->i_ino
== BTRFS_FIRST_FREE_OBJECTID
)
847 * This will strip out the subvol=%s argument for an argument string and add
848 * subvolid=0 to make sure we get the actual tree root for path walking to the
851 static char *setup_root_args(char *args
)
854 unsigned len
= strlen(args
) + 2;
859 * We need the same args as before, but minus
867 * which is a difference of 2 characters, so we allocate strlen(args) +
870 ret
= kzalloc(len
* sizeof(char), GFP_NOFS
);
873 pos
= strstr(args
, "subvol=");
875 /* This shouldn't happen, but just in case.. */
882 * The subvol=<> arg is not at the front of the string, copy everybody
883 * up to that into ret.
888 copied
+= strlen(args
);
892 strncpy(ret
+ copied
, "subvolid=0", len
- copied
);
894 /* Length of subvolid=0 */
898 * If there is no , after the subvol= option then we know there's no
899 * other options and we can just return.
901 pos
= strchr(pos
, ',');
905 /* Copy the rest of the arguments into our buffer */
906 strncpy(ret
+ copied
, pos
, len
- copied
);
907 copied
+= strlen(pos
);
912 static struct dentry
*mount_subvol(const char *subvol_name
, int flags
,
913 const char *device_name
, char *data
)
916 struct vfsmount
*mnt
;
919 newargs
= setup_root_args(data
);
921 return ERR_PTR(-ENOMEM
);
922 mnt
= vfs_kern_mount(&btrfs_fs_type
, flags
, device_name
,
926 return ERR_CAST(mnt
);
928 root
= mount_subtree(mnt
, subvol_name
);
930 if (!IS_ERR(root
) && !is_subvolume_inode(root
->d_inode
)) {
931 struct super_block
*s
= root
->d_sb
;
933 root
= ERR_PTR(-EINVAL
);
934 deactivate_locked_super(s
);
935 printk(KERN_ERR
"btrfs: '%s' is not a valid subvolume\n",
943 * Find a superblock for the given device / mount point.
945 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
946 * for multiple device setup. Make sure to keep it in sync.
948 static struct dentry
*btrfs_mount(struct file_system_type
*fs_type
, int flags
,
949 const char *device_name
, void *data
)
951 struct block_device
*bdev
= NULL
;
952 struct super_block
*s
;
954 struct btrfs_fs_devices
*fs_devices
= NULL
;
955 struct btrfs_fs_info
*fs_info
= NULL
;
956 fmode_t mode
= FMODE_READ
;
957 char *subvol_name
= NULL
;
958 u64 subvol_objectid
= 0;
959 u64 subvol_rootid
= 0;
962 if (!(flags
& MS_RDONLY
))
965 error
= btrfs_parse_early_options(data
, mode
, fs_type
,
966 &subvol_name
, &subvol_objectid
,
967 &subvol_rootid
, &fs_devices
);
970 return ERR_PTR(error
);
974 root
= mount_subvol(subvol_name
, flags
, device_name
, data
);
979 error
= btrfs_scan_one_device(device_name
, mode
, fs_type
, &fs_devices
);
981 return ERR_PTR(error
);
984 * Setup a dummy root and fs_info for test/set super. This is because
985 * we don't actually fill this stuff out until open_ctree, but we need
986 * it for searching for existing supers, so this lets us do that and
987 * then open_ctree will properly initialize everything later.
989 fs_info
= kzalloc(sizeof(struct btrfs_fs_info
), GFP_NOFS
);
991 return ERR_PTR(-ENOMEM
);
993 fs_info
->fs_devices
= fs_devices
;
995 fs_info
->super_copy
= kzalloc(BTRFS_SUPER_INFO_SIZE
, GFP_NOFS
);
996 fs_info
->super_for_commit
= kzalloc(BTRFS_SUPER_INFO_SIZE
, GFP_NOFS
);
997 if (!fs_info
->super_copy
|| !fs_info
->super_for_commit
) {
1002 error
= btrfs_open_devices(fs_devices
, mode
, fs_type
);
1006 if (!(flags
& MS_RDONLY
) && fs_devices
->rw_devices
== 0) {
1008 goto error_close_devices
;
1011 bdev
= fs_devices
->latest_bdev
;
1012 s
= sget(fs_type
, btrfs_test_super
, btrfs_set_super
, fs_info
);
1015 goto error_close_devices
;
1019 btrfs_close_devices(fs_devices
);
1020 free_fs_info(fs_info
);
1021 if ((flags
^ s
->s_flags
) & MS_RDONLY
)
1024 char b
[BDEVNAME_SIZE
];
1026 s
->s_flags
= flags
| MS_NOSEC
;
1027 strlcpy(s
->s_id
, bdevname(bdev
, b
), sizeof(s
->s_id
));
1028 btrfs_sb(s
)->bdev_holder
= fs_type
;
1029 error
= btrfs_fill_super(s
, fs_devices
, data
,
1030 flags
& MS_SILENT
? 1 : 0);
1033 root
= !error
? get_default_root(s
, subvol_objectid
) : ERR_PTR(error
);
1035 deactivate_locked_super(s
);
1039 error_close_devices
:
1040 btrfs_close_devices(fs_devices
);
1042 free_fs_info(fs_info
);
1043 return ERR_PTR(error
);
1046 static int btrfs_remount(struct super_block
*sb
, int *flags
, char *data
)
1048 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1049 struct btrfs_root
*root
= fs_info
->tree_root
;
1052 ret
= btrfs_parse_options(root
, data
);
1056 if ((*flags
& MS_RDONLY
) == (sb
->s_flags
& MS_RDONLY
))
1059 if (*flags
& MS_RDONLY
) {
1060 sb
->s_flags
|= MS_RDONLY
;
1062 ret
= btrfs_commit_super(root
);
1065 if (fs_info
->fs_devices
->rw_devices
== 0)
1068 if (btrfs_super_log_root(fs_info
->super_copy
) != 0)
1071 ret
= btrfs_cleanup_fs_roots(fs_info
);
1074 /* recover relocation */
1075 ret
= btrfs_recover_relocation(root
);
1078 sb
->s_flags
&= ~MS_RDONLY
;
1084 /* Used to sort the devices by max_avail(descending sort) */
1085 static int btrfs_cmp_device_free_bytes(const void *dev_info1
,
1086 const void *dev_info2
)
1088 if (((struct btrfs_device_info
*)dev_info1
)->max_avail
>
1089 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
1091 else if (((struct btrfs_device_info
*)dev_info1
)->max_avail
<
1092 ((struct btrfs_device_info
*)dev_info2
)->max_avail
)
1099 * sort the devices by max_avail, in which max free extent size of each device
1100 * is stored.(Descending Sort)
1102 static inline void btrfs_descending_sort_devices(
1103 struct btrfs_device_info
*devices
,
1106 sort(devices
, nr_devices
, sizeof(struct btrfs_device_info
),
1107 btrfs_cmp_device_free_bytes
, NULL
);
1111 * The helper to calc the free space on the devices that can be used to store
1114 static int btrfs_calc_avail_data_space(struct btrfs_root
*root
, u64
*free_bytes
)
1116 struct btrfs_fs_info
*fs_info
= root
->fs_info
;
1117 struct btrfs_device_info
*devices_info
;
1118 struct btrfs_fs_devices
*fs_devices
= fs_info
->fs_devices
;
1119 struct btrfs_device
*device
;
1124 u64 min_stripe_size
;
1125 int min_stripes
= 1, num_stripes
= 1;
1126 int i
= 0, nr_devices
;
1129 nr_devices
= fs_info
->fs_devices
->open_devices
;
1130 BUG_ON(!nr_devices
);
1132 devices_info
= kmalloc(sizeof(*devices_info
) * nr_devices
,
1137 /* calc min stripe number for data space alloction */
1138 type
= btrfs_get_alloc_profile(root
, 1);
1139 if (type
& BTRFS_BLOCK_GROUP_RAID0
) {
1141 num_stripes
= nr_devices
;
1142 } else if (type
& BTRFS_BLOCK_GROUP_RAID1
) {
1145 } else if (type
& BTRFS_BLOCK_GROUP_RAID10
) {
1150 if (type
& BTRFS_BLOCK_GROUP_DUP
)
1151 min_stripe_size
= 2 * BTRFS_STRIPE_LEN
;
1153 min_stripe_size
= BTRFS_STRIPE_LEN
;
1155 list_for_each_entry(device
, &fs_devices
->devices
, dev_list
) {
1156 if (!device
->in_fs_metadata
|| !device
->bdev
)
1159 avail_space
= device
->total_bytes
- device
->bytes_used
;
1161 /* align with stripe_len */
1162 do_div(avail_space
, BTRFS_STRIPE_LEN
);
1163 avail_space
*= BTRFS_STRIPE_LEN
;
1166 * In order to avoid overwritting the superblock on the drive,
1167 * btrfs starts at an offset of at least 1MB when doing chunk
1170 skip_space
= 1024 * 1024;
1172 /* user can set the offset in fs_info->alloc_start. */
1173 if (fs_info
->alloc_start
+ BTRFS_STRIPE_LEN
<=
1174 device
->total_bytes
)
1175 skip_space
= max(fs_info
->alloc_start
, skip_space
);
1178 * btrfs can not use the free space in [0, skip_space - 1],
1179 * we must subtract it from the total. In order to implement
1180 * it, we account the used space in this range first.
1182 ret
= btrfs_account_dev_extents_size(device
, 0, skip_space
- 1,
1185 kfree(devices_info
);
1189 /* calc the free space in [0, skip_space - 1] */
1190 skip_space
-= used_space
;
1193 * we can use the free space in [0, skip_space - 1], subtract
1194 * it from the total.
1196 if (avail_space
&& avail_space
>= skip_space
)
1197 avail_space
-= skip_space
;
1201 if (avail_space
< min_stripe_size
)
1204 devices_info
[i
].dev
= device
;
1205 devices_info
[i
].max_avail
= avail_space
;
1212 btrfs_descending_sort_devices(devices_info
, nr_devices
);
1216 while (nr_devices
>= min_stripes
) {
1217 if (num_stripes
> nr_devices
)
1218 num_stripes
= nr_devices
;
1220 if (devices_info
[i
].max_avail
>= min_stripe_size
) {
1224 avail_space
+= devices_info
[i
].max_avail
* num_stripes
;
1225 alloc_size
= devices_info
[i
].max_avail
;
1226 for (j
= i
+ 1 - num_stripes
; j
<= i
; j
++)
1227 devices_info
[j
].max_avail
-= alloc_size
;
1233 kfree(devices_info
);
1234 *free_bytes
= avail_space
;
1238 static int btrfs_statfs(struct dentry
*dentry
, struct kstatfs
*buf
)
1240 struct btrfs_fs_info
*fs_info
= btrfs_sb(dentry
->d_sb
);
1241 struct btrfs_super_block
*disk_super
= fs_info
->super_copy
;
1242 struct list_head
*head
= &fs_info
->space_info
;
1243 struct btrfs_space_info
*found
;
1245 u64 total_free_data
= 0;
1246 int bits
= dentry
->d_sb
->s_blocksize_bits
;
1247 __be32
*fsid
= (__be32
*)fs_info
->fsid
;
1250 /* holding chunk_muext to avoid allocating new chunks */
1251 mutex_lock(&fs_info
->chunk_mutex
);
1253 list_for_each_entry_rcu(found
, head
, list
) {
1254 if (found
->flags
& BTRFS_BLOCK_GROUP_DATA
) {
1255 total_free_data
+= found
->disk_total
- found
->disk_used
;
1257 btrfs_account_ro_block_groups_free_space(found
);
1260 total_used
+= found
->disk_used
;
1264 buf
->f_namelen
= BTRFS_NAME_LEN
;
1265 buf
->f_blocks
= btrfs_super_total_bytes(disk_super
) >> bits
;
1266 buf
->f_bfree
= buf
->f_blocks
- (total_used
>> bits
);
1267 buf
->f_bsize
= dentry
->d_sb
->s_blocksize
;
1268 buf
->f_type
= BTRFS_SUPER_MAGIC
;
1269 buf
->f_bavail
= total_free_data
;
1270 ret
= btrfs_calc_avail_data_space(fs_info
->tree_root
, &total_free_data
);
1272 mutex_unlock(&fs_info
->chunk_mutex
);
1275 buf
->f_bavail
+= total_free_data
;
1276 buf
->f_bavail
= buf
->f_bavail
>> bits
;
1277 mutex_unlock(&fs_info
->chunk_mutex
);
1279 /* We treat it as constant endianness (it doesn't matter _which_)
1280 because we want the fsid to come out the same whether mounted
1281 on a big-endian or little-endian host */
1282 buf
->f_fsid
.val
[0] = be32_to_cpu(fsid
[0]) ^ be32_to_cpu(fsid
[2]);
1283 buf
->f_fsid
.val
[1] = be32_to_cpu(fsid
[1]) ^ be32_to_cpu(fsid
[3]);
1284 /* Mask in the root object ID too, to disambiguate subvols */
1285 buf
->f_fsid
.val
[0] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
>> 32;
1286 buf
->f_fsid
.val
[1] ^= BTRFS_I(dentry
->d_inode
)->root
->objectid
;
1291 static void btrfs_kill_super(struct super_block
*sb
)
1293 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1294 kill_anon_super(sb
);
1295 free_fs_info(fs_info
);
1298 static struct file_system_type btrfs_fs_type
= {
1299 .owner
= THIS_MODULE
,
1301 .mount
= btrfs_mount
,
1302 .kill_sb
= btrfs_kill_super
,
1303 .fs_flags
= FS_REQUIRES_DEV
,
1307 * used by btrfsctl to scan devices when no FS is mounted
1309 static long btrfs_control_ioctl(struct file
*file
, unsigned int cmd
,
1312 struct btrfs_ioctl_vol_args
*vol
;
1313 struct btrfs_fs_devices
*fs_devices
;
1316 if (!capable(CAP_SYS_ADMIN
))
1319 vol
= memdup_user((void __user
*)arg
, sizeof(*vol
));
1321 return PTR_ERR(vol
);
1324 case BTRFS_IOC_SCAN_DEV
:
1325 ret
= btrfs_scan_one_device(vol
->name
, FMODE_READ
,
1326 &btrfs_fs_type
, &fs_devices
);
1334 static int btrfs_freeze(struct super_block
*sb
)
1336 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1337 mutex_lock(&fs_info
->transaction_kthread_mutex
);
1338 mutex_lock(&fs_info
->cleaner_mutex
);
1342 static int btrfs_unfreeze(struct super_block
*sb
)
1344 struct btrfs_fs_info
*fs_info
= btrfs_sb(sb
);
1345 mutex_unlock(&fs_info
->cleaner_mutex
);
1346 mutex_unlock(&fs_info
->transaction_kthread_mutex
);
1350 static void btrfs_fs_dirty_inode(struct inode
*inode
, int flags
)
1354 ret
= btrfs_dirty_inode(inode
);
1356 printk_ratelimited(KERN_ERR
"btrfs: fail to dirty inode %Lu "
1357 "error %d\n", btrfs_ino(inode
), ret
);
1360 static const struct super_operations btrfs_super_ops
= {
1361 .drop_inode
= btrfs_drop_inode
,
1362 .evict_inode
= btrfs_evict_inode
,
1363 .put_super
= btrfs_put_super
,
1364 .sync_fs
= btrfs_sync_fs
,
1365 .show_options
= btrfs_show_options
,
1366 .write_inode
= btrfs_write_inode
,
1367 .dirty_inode
= btrfs_fs_dirty_inode
,
1368 .alloc_inode
= btrfs_alloc_inode
,
1369 .destroy_inode
= btrfs_destroy_inode
,
1370 .statfs
= btrfs_statfs
,
1371 .remount_fs
= btrfs_remount
,
1372 .freeze_fs
= btrfs_freeze
,
1373 .unfreeze_fs
= btrfs_unfreeze
,
1376 static const struct file_operations btrfs_ctl_fops
= {
1377 .unlocked_ioctl
= btrfs_control_ioctl
,
1378 .compat_ioctl
= btrfs_control_ioctl
,
1379 .owner
= THIS_MODULE
,
1380 .llseek
= noop_llseek
,
1383 static struct miscdevice btrfs_misc
= {
1384 .minor
= BTRFS_MINOR
,
1385 .name
= "btrfs-control",
1386 .fops
= &btrfs_ctl_fops
1389 MODULE_ALIAS_MISCDEV(BTRFS_MINOR
);
1390 MODULE_ALIAS("devname:btrfs-control");
1392 static int btrfs_interface_init(void)
1394 return misc_register(&btrfs_misc
);
1397 static void btrfs_interface_exit(void)
1399 if (misc_deregister(&btrfs_misc
) < 0)
1400 printk(KERN_INFO
"misc_deregister failed for control device");
1403 static int __init
init_btrfs_fs(void)
1407 err
= btrfs_init_sysfs();
1411 btrfs_init_compress();
1413 err
= btrfs_init_cachep();
1417 err
= extent_io_init();
1421 err
= extent_map_init();
1423 goto free_extent_io
;
1425 err
= btrfs_delayed_inode_init();
1427 goto free_extent_map
;
1429 err
= btrfs_interface_init();
1431 goto free_delayed_inode
;
1433 err
= register_filesystem(&btrfs_fs_type
);
1435 goto unregister_ioctl
;
1437 printk(KERN_INFO
"%s loaded\n", BTRFS_BUILD_VERSION
);
1441 btrfs_interface_exit();
1443 btrfs_delayed_inode_exit();
1449 btrfs_destroy_cachep();
1451 btrfs_exit_compress();
1456 static void __exit
exit_btrfs_fs(void)
1458 btrfs_destroy_cachep();
1459 btrfs_delayed_inode_exit();
1462 btrfs_interface_exit();
1463 unregister_filesystem(&btrfs_fs_type
);
1465 btrfs_cleanup_fs_uuids();
1466 btrfs_exit_compress();
1469 module_init(init_btrfs_fs
)
1470 module_exit(exit_btrfs_fs
)
1472 MODULE_LICENSE("GPL");