]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - fs/btrfs/super.c
Btrfs: Initialize btrfs_root->highest_objectid when loading tree root and subvolume...
[mirror_ubuntu-bionic-kernel.git] / fs / btrfs / super.c
CommitLineData
6cbd5570
CM
1/*
2 * Copyright (C) 2007 Oracle. All rights reserved.
3 *
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.
7 *
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.
12 *
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.
17 */
18
4b82d6e4 19#include <linux/blkdev.h>
2e635a27 20#include <linux/module.h>
e20d96d6 21#include <linux/buffer_head.h>
2e635a27
CM
22#include <linux/fs.h>
23#include <linux/pagemap.h>
24#include <linux/highmem.h>
25#include <linux/time.h>
26#include <linux/init.h>
a9572a15 27#include <linux/seq_file.h>
2e635a27 28#include <linux/string.h>
2e635a27 29#include <linux/backing-dev.h>
4b82d6e4 30#include <linux/mount.h>
dee26a9f 31#include <linux/mpage.h>
75dfe396
CM
32#include <linux/swap.h>
33#include <linux/writeback.h>
8fd17795 34#include <linux/statfs.h>
08607c1b 35#include <linux/compat.h>
95e05289 36#include <linux/parser.h>
c59f8951 37#include <linux/ctype.h>
6da6abae 38#include <linux/namei.h>
a9218f6b 39#include <linux/miscdevice.h>
1bcbf313 40#include <linux/magic.h>
5a0e3ad6 41#include <linux/slab.h>
90a887c9 42#include <linux/cleancache.h>
22c44fe6 43#include <linux/ratelimit.h>
55e301fd 44#include <linux/btrfs.h>
16cdcec7 45#include "delayed-inode.h"
2e635a27 46#include "ctree.h"
e20d96d6 47#include "disk-io.h"
d5719762 48#include "transaction.h"
2c90e5d6 49#include "btrfs_inode.h"
3a686375 50#include "print-tree.h"
14a958e6 51#include "hash.h"
63541927 52#include "props.h"
5103e947 53#include "xattr.h"
8a4b83cc 54#include "volumes.h"
be6e8dc0 55#include "export.h"
c8b97818 56#include "compression.h"
9c5085c1 57#include "rcu-string.h"
8dabb742 58#include "dev-replace.h"
74255aa0 59#include "free-space-cache.h"
b9e9a6cb 60#include "backref.h"
dc11dd5d 61#include "tests/btrfs-tests.h"
2e635a27 62
d3982100 63#include "qgroup.h"
1abe9b8a 64#define CREATE_TRACE_POINTS
65#include <trace/events/btrfs.h>
66
b87221de 67static const struct super_operations btrfs_super_ops;
830c4adb 68static struct file_system_type btrfs_fs_type;
75dfe396 69
0723a047
HH
70static int btrfs_remount(struct super_block *sb, int *flags, char *data);
71
e33e17ee 72const char *btrfs_decode_error(int errno)
acce952b 73{
08748810 74 char *errstr = "unknown";
acce952b 75
76 switch (errno) {
77 case -EIO:
78 errstr = "IO failure";
79 break;
80 case -ENOMEM:
81 errstr = "Out of memory";
82 break;
83 case -EROFS:
84 errstr = "Readonly filesystem";
85 break;
8c342930
JM
86 case -EEXIST:
87 errstr = "Object already exists";
88 break;
94ef7280
DS
89 case -ENOSPC:
90 errstr = "No space left";
91 break;
92 case -ENOENT:
93 errstr = "No such entry";
94 break;
acce952b 95 }
96
97 return errstr;
98}
99
bbece8a3 100static void save_error_info(struct btrfs_fs_info *fs_info)
acce952b 101{
102 /*
103 * today we only save the error info into ram. Long term we'll
104 * also send it down to the disk
105 */
87533c47 106 set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state);
acce952b 107}
108
acce952b 109/* btrfs handle error by forcing the filesystem readonly */
110static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
111{
112 struct super_block *sb = fs_info->sb;
113
114 if (sb->s_flags & MS_RDONLY)
115 return;
116
87533c47 117 if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
acce952b 118 sb->s_flags |= MS_RDONLY;
c2cf52eb 119 btrfs_info(fs_info, "forced readonly");
1acd6831
SB
120 /*
121 * Note that a running device replace operation is not
122 * canceled here although there is no way to update
123 * the progress. It would add the risk of a deadlock,
124 * therefore the canceling is ommited. The only penalty
125 * is that some I/O remains active until the procedure
126 * completes. The next time when the filesystem is
127 * mounted writeable again, the device replace
128 * operation continues.
129 */
acce952b 130 }
131}
132
133/*
134 * __btrfs_std_error decodes expected errors from the caller and
135 * invokes the approciate error response.
136 */
c0d19e2b 137__cold
acce952b 138void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
4da35113 139 unsigned int line, int errno, const char *fmt, ...)
acce952b 140{
141 struct super_block *sb = fs_info->sb;
57d816a1 142#ifdef CONFIG_PRINTK
acce952b 143 const char *errstr;
57d816a1 144#endif
acce952b 145
146 /*
147 * Special case: if the error is EROFS, and we're already
148 * under MS_RDONLY, then it is safe here.
149 */
150 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
4da35113
JM
151 return;
152
57d816a1 153#ifdef CONFIG_PRINTK
08748810 154 errstr = btrfs_decode_error(errno);
4da35113 155 if (fmt) {
37252a66
ES
156 struct va_format vaf;
157 va_list args;
158
159 va_start(args, fmt);
160 vaf.fmt = fmt;
161 vaf.va = &args;
4da35113 162
efe120a0
FH
163 printk(KERN_CRIT
164 "BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n",
08748810 165 sb->s_id, function, line, errno, errstr, &vaf);
37252a66 166 va_end(args);
4da35113 167 } else {
efe120a0 168 printk(KERN_CRIT "BTRFS: error (device %s) in %s:%d: errno=%d %s\n",
08748810 169 sb->s_id, function, line, errno, errstr);
4da35113 170 }
57d816a1 171#endif
acce952b 172
4da35113 173 /* Don't go through full error handling during mount */
cf79ffb5
JB
174 save_error_info(fs_info);
175 if (sb->s_flags & MS_BORN)
4da35113 176 btrfs_handle_error(fs_info);
4da35113 177}
acce952b 178
57d816a1 179#ifdef CONFIG_PRINTK
533574c6 180static const char * const logtypes[] = {
4da35113
JM
181 "emergency",
182 "alert",
183 "critical",
184 "error",
185 "warning",
186 "notice",
187 "info",
188 "debug",
189};
190
c2cf52eb 191void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
4da35113
JM
192{
193 struct super_block *sb = fs_info->sb;
194 char lvl[4];
195 struct va_format vaf;
196 va_list args;
197 const char *type = logtypes[4];
533574c6 198 int kern_level;
4da35113
JM
199
200 va_start(args, fmt);
201
533574c6
JP
202 kern_level = printk_get_level(fmt);
203 if (kern_level) {
204 size_t size = printk_skip_level(fmt) - fmt;
205 memcpy(lvl, fmt, size);
206 lvl[size] = '\0';
207 fmt += size;
208 type = logtypes[kern_level - '0'];
4da35113
JM
209 } else
210 *lvl = '\0';
211
212 vaf.fmt = fmt;
213 vaf.va = &args;
533574c6 214
c2cf52eb 215 printk("%sBTRFS %s (device %s): %pV\n", lvl, type, sb->s_id, &vaf);
533574c6
JP
216
217 va_end(args);
218}
533574c6 219#endif
acce952b 220
49b25e05
JM
221/*
222 * We only mark the transaction aborted and then set the file system read-only.
223 * This will prevent new transactions from starting or trying to join this
224 * one.
225 *
226 * This means that error recovery at the call site is limited to freeing
227 * any local memory allocations and passing the error code up without
228 * further cleanup. The transaction should complete as it normally would
229 * in the call path but will return -EIO.
230 *
231 * We'll complete the cleanup in btrfs_end_transaction and
232 * btrfs_commit_transaction.
233 */
c0d19e2b 234__cold
49b25e05
JM
235void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
236 struct btrfs_root *root, const char *function,
237 unsigned int line, int errno)
238{
49b25e05
JM
239 trans->aborted = errno;
240 /* Nothing used. The other threads that have joined this
241 * transaction may be able to continue. */
c92f6be3 242 if (!trans->blocks_used && list_empty(&trans->new_bgs)) {
69ce977a
MX
243 const char *errstr;
244
08748810 245 errstr = btrfs_decode_error(errno);
c2cf52eb
SK
246 btrfs_warn(root->fs_info,
247 "%s:%d: Aborting unused transaction(%s).",
248 function, line, errstr);
acce952b 249 return;
49b25e05 250 }
8d25a086 251 ACCESS_ONCE(trans->transaction->aborted) = errno;
501407aa
JB
252 /* Wake up anybody who may be waiting on this transaction */
253 wake_up(&root->fs_info->transaction_wait);
254 wake_up(&root->fs_info->transaction_blocked_wait);
49b25e05
JM
255 __btrfs_std_error(root->fs_info, function, line, errno, NULL);
256}
8c342930
JM
257/*
258 * __btrfs_panic decodes unexpected, fatal errors from the caller,
259 * issues an alert, and either panics or BUGs, depending on mount options.
260 */
c0d19e2b 261__cold
8c342930
JM
262void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
263 unsigned int line, int errno, const char *fmt, ...)
264{
8c342930
JM
265 char *s_id = "<unknown>";
266 const char *errstr;
267 struct va_format vaf = { .fmt = fmt };
268 va_list args;
acce952b 269
8c342930
JM
270 if (fs_info)
271 s_id = fs_info->sb->s_id;
acce952b 272
8c342930
JM
273 va_start(args, fmt);
274 vaf.va = &args;
275
08748810 276 errstr = btrfs_decode_error(errno);
aa43a17c 277 if (fs_info && (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR))
08748810
DS
278 panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
279 s_id, function, line, &vaf, errno, errstr);
8c342930 280
efe120a0
FH
281 btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)",
282 function, line, &vaf, errno, errstr);
8c342930
JM
283 va_end(args);
284 /* Caller calls BUG() */
acce952b 285}
286
d397712b 287static void btrfs_put_super(struct super_block *sb)
b18c6685 288{
3abdbd78 289 close_ctree(btrfs_sb(sb)->tree_root);
75dfe396
CM
290}
291
95e05289 292enum {
73f73415 293 Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum,
287a0ab9
JB
294 Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd,
295 Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress,
261507a0
LZ
296 Opt_compress_type, Opt_compress_force, Opt_compress_force_type,
297 Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_discard,
91435650 298 Opt_space_cache, Opt_clear_cache, Opt_user_subvol_rm_allowed,
9555c6c1
ID
299 Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache,
300 Opt_no_space_cache, Opt_recovery, Opt_skip_balance,
21adbd5c 301 Opt_check_integrity, Opt_check_integrity_including_extent_data,
f420ee1e 302 Opt_check_integrity_print_mask, Opt_fatal_errors, Opt_rescan_uuid_tree,
e07a2ade 303 Opt_commit_interval, Opt_barrier, Opt_nodefrag, Opt_nodiscard,
a258af7a 304 Opt_noenospc_debug, Opt_noflushoncommit, Opt_acl, Opt_datacow,
3818aea2 305 Opt_datasum, Opt_treelog, Opt_noinode_cache,
d0bd4560
JB
306#ifdef CONFIG_BTRFS_DEBUG
307 Opt_fragment_data, Opt_fragment_metadata, Opt_fragment_all,
308#endif
9555c6c1 309 Opt_err,
95e05289
CM
310};
311
4d4ab6d6 312static const match_table_t tokens = {
dfe25020 313 {Opt_degraded, "degraded"},
95e05289 314 {Opt_subvol, "subvol=%s"},
1493381f 315 {Opt_subvolid, "subvolid=%s"},
43e570b0 316 {Opt_device, "device=%s"},
b6cda9bc 317 {Opt_nodatasum, "nodatasum"},
d399167d 318 {Opt_datasum, "datasum"},
be20aa9d 319 {Opt_nodatacow, "nodatacow"},
a258af7a 320 {Opt_datacow, "datacow"},
21ad10cf 321 {Opt_nobarrier, "nobarrier"},
842bef58 322 {Opt_barrier, "barrier"},
6f568d35 323 {Opt_max_inline, "max_inline=%s"},
8f662a76 324 {Opt_alloc_start, "alloc_start=%s"},
4543df7e 325 {Opt_thread_pool, "thread_pool=%d"},
c8b97818 326 {Opt_compress, "compress"},
261507a0 327 {Opt_compress_type, "compress=%s"},
a555f810 328 {Opt_compress_force, "compress-force"},
261507a0 329 {Opt_compress_force_type, "compress-force=%s"},
e18e4809 330 {Opt_ssd, "ssd"},
451d7585 331 {Opt_ssd_spread, "ssd_spread"},
3b30c22f 332 {Opt_nossd, "nossd"},
bd0330ad 333 {Opt_acl, "acl"},
33268eaf 334 {Opt_noacl, "noacl"},
3a5e1404 335 {Opt_notreelog, "notreelog"},
a88998f2 336 {Opt_treelog, "treelog"},
dccae999 337 {Opt_flushoncommit, "flushoncommit"},
2c9ee856 338 {Opt_noflushoncommit, "noflushoncommit"},
97e728d4 339 {Opt_ratio, "metadata_ratio=%d"},
e244a0ae 340 {Opt_discard, "discard"},
e07a2ade 341 {Opt_nodiscard, "nodiscard"},
0af3d00b 342 {Opt_space_cache, "space_cache"},
88c2ba3b 343 {Opt_clear_cache, "clear_cache"},
4260f7c7 344 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
91435650 345 {Opt_enospc_debug, "enospc_debug"},
53036293 346 {Opt_noenospc_debug, "noenospc_debug"},
e15d0542 347 {Opt_subvolrootid, "subvolrootid=%d"},
4cb5300b 348 {Opt_defrag, "autodefrag"},
fc0ca9af 349 {Opt_nodefrag, "noautodefrag"},
4b9465cb 350 {Opt_inode_cache, "inode_cache"},
3818aea2 351 {Opt_noinode_cache, "noinode_cache"},
8965593e 352 {Opt_no_space_cache, "nospace_cache"},
af31f5e5 353 {Opt_recovery, "recovery"},
9555c6c1 354 {Opt_skip_balance, "skip_balance"},
21adbd5c
SB
355 {Opt_check_integrity, "check_int"},
356 {Opt_check_integrity_including_extent_data, "check_int_data"},
357 {Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
f420ee1e 358 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
8c342930 359 {Opt_fatal_errors, "fatal_errors=%s"},
8b87dc17 360 {Opt_commit_interval, "commit=%d"},
d0bd4560
JB
361#ifdef CONFIG_BTRFS_DEBUG
362 {Opt_fragment_data, "fragment=data"},
363 {Opt_fragment_metadata, "fragment=metadata"},
364 {Opt_fragment_all, "fragment=all"},
365#endif
33268eaf 366 {Opt_err, NULL},
95e05289
CM
367};
368
edf24abe
CH
369/*
370 * Regular mount options parser. Everything that is needed only when
371 * reading in a new superblock is parsed here.
49b25e05 372 * XXX JDM: This needs to be cleaned up for remount.
edf24abe
CH
373 */
374int btrfs_parse_options(struct btrfs_root *root, char *options)
95e05289 375{
edf24abe 376 struct btrfs_fs_info *info = root->fs_info;
95e05289 377 substring_t args[MAX_OPT_ARGS];
73bc1876
JB
378 char *p, *num, *orig = NULL;
379 u64 cache_gen;
4543df7e 380 int intarg;
a7a3f7ca 381 int ret = 0;
261507a0
LZ
382 char *compress_type;
383 bool compress_force = false;
b6cda9bc 384
6c41761f 385 cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
73bc1876
JB
386 if (cache_gen)
387 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
388
95e05289 389 if (!options)
73bc1876 390 goto out;
95e05289 391
be20aa9d
CM
392 /*
393 * strsep changes the string, duplicate it because parse_options
394 * gets called twice
395 */
396 options = kstrdup(options, GFP_NOFS);
397 if (!options)
398 return -ENOMEM;
399
da495ecc 400 orig = options;
be20aa9d 401
edf24abe 402 while ((p = strsep(&options, ",")) != NULL) {
95e05289
CM
403 int token;
404 if (!*p)
405 continue;
406
407 token = match_token(p, tokens, args);
408 switch (token) {
dfe25020 409 case Opt_degraded:
efe120a0 410 btrfs_info(root->fs_info, "allowing degraded mounts");
edf24abe 411 btrfs_set_opt(info->mount_opt, DEGRADED);
dfe25020 412 break;
95e05289 413 case Opt_subvol:
73f73415 414 case Opt_subvolid:
e15d0542 415 case Opt_subvolrootid:
43e570b0 416 case Opt_device:
edf24abe 417 /*
43e570b0 418 * These are parsed by btrfs_parse_early_options
edf24abe
CH
419 * and can be happily ignored here.
420 */
b6cda9bc
CM
421 break;
422 case Opt_nodatasum:
07802534
QW
423 btrfs_set_and_info(root, NODATASUM,
424 "setting nodatasum");
be20aa9d 425 break;
d399167d 426 case Opt_datasum:
07802534
QW
427 if (btrfs_test_opt(root, NODATASUM)) {
428 if (btrfs_test_opt(root, NODATACOW))
429 btrfs_info(root->fs_info, "setting datasum, datacow enabled");
430 else
431 btrfs_info(root->fs_info, "setting datasum");
432 }
d399167d
QW
433 btrfs_clear_opt(info->mount_opt, NODATACOW);
434 btrfs_clear_opt(info->mount_opt, NODATASUM);
435 break;
be20aa9d 436 case Opt_nodatacow:
07802534
QW
437 if (!btrfs_test_opt(root, NODATACOW)) {
438 if (!btrfs_test_opt(root, COMPRESS) ||
439 !btrfs_test_opt(root, FORCE_COMPRESS)) {
efe120a0 440 btrfs_info(root->fs_info,
07802534
QW
441 "setting nodatacow, compression disabled");
442 } else {
443 btrfs_info(root->fs_info, "setting nodatacow");
444 }
bedb2cca 445 }
bedb2cca
AP
446 btrfs_clear_opt(info->mount_opt, COMPRESS);
447 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
edf24abe
CH
448 btrfs_set_opt(info->mount_opt, NODATACOW);
449 btrfs_set_opt(info->mount_opt, NODATASUM);
95e05289 450 break;
a258af7a 451 case Opt_datacow:
07802534
QW
452 btrfs_clear_and_info(root, NODATACOW,
453 "setting datacow");
a258af7a 454 break;
a555f810 455 case Opt_compress_force:
261507a0
LZ
456 case Opt_compress_force_type:
457 compress_force = true;
1c697d4a 458 /* Fallthrough */
261507a0
LZ
459 case Opt_compress:
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;
063849ea 466 btrfs_set_opt(info->mount_opt, COMPRESS);
bedb2cca
AP
467 btrfs_clear_opt(info->mount_opt, NODATACOW);
468 btrfs_clear_opt(info->mount_opt, NODATASUM);
a6fa6fae
LZ
469 } else if (strcmp(args[0].from, "lzo") == 0) {
470 compress_type = "lzo";
471 info->compress_type = BTRFS_COMPRESS_LZO;
063849ea 472 btrfs_set_opt(info->mount_opt, COMPRESS);
bedb2cca
AP
473 btrfs_clear_opt(info->mount_opt, NODATACOW);
474 btrfs_clear_opt(info->mount_opt, NODATASUM);
2b0ce2c2 475 btrfs_set_fs_incompat(info, COMPRESS_LZO);
063849ea
AH
476 } else if (strncmp(args[0].from, "no", 2) == 0) {
477 compress_type = "no";
063849ea
AH
478 btrfs_clear_opt(info->mount_opt, COMPRESS);
479 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
480 compress_force = false;
261507a0
LZ
481 } else {
482 ret = -EINVAL;
483 goto out;
484 }
485
261507a0 486 if (compress_force) {
07802534
QW
487 btrfs_set_and_info(root, FORCE_COMPRESS,
488 "force %s compression",
489 compress_type);
143f3636 490 } else {
07802534
QW
491 if (!btrfs_test_opt(root, COMPRESS))
492 btrfs_info(root->fs_info,
351fd353 493 "btrfs: use %s compression",
07802534 494 compress_type);
4027e0f4
WS
495 /*
496 * If we remount from compress-force=xxx to
497 * compress=xxx, we need clear FORCE_COMPRESS
498 * flag, otherwise, there is no way for users
499 * to disable forcible compression separately.
500 */
501 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
a7e252af 502 }
a555f810 503 break;
e18e4809 504 case Opt_ssd:
07802534
QW
505 btrfs_set_and_info(root, SSD,
506 "use ssd allocation scheme");
e18e4809 507 break;
451d7585 508 case Opt_ssd_spread:
07802534
QW
509 btrfs_set_and_info(root, SSD_SPREAD,
510 "use spread ssd allocation scheme");
2aa06a35 511 btrfs_set_opt(info->mount_opt, SSD);
451d7585 512 break;
3b30c22f 513 case Opt_nossd:
2aa06a35 514 btrfs_set_and_info(root, NOSSD,
07802534 515 "not using ssd allocation scheme");
3b30c22f
CM
516 btrfs_clear_opt(info->mount_opt, SSD);
517 break;
842bef58 518 case Opt_barrier:
07802534
QW
519 btrfs_clear_and_info(root, NOBARRIER,
520 "turning on barriers");
842bef58 521 break;
21ad10cf 522 case Opt_nobarrier:
07802534
QW
523 btrfs_set_and_info(root, NOBARRIER,
524 "turning off barriers");
21ad10cf 525 break;
4543df7e 526 case Opt_thread_pool:
2c334e87
WS
527 ret = match_int(&args[0], &intarg);
528 if (ret) {
529 goto out;
530 } else if (intarg > 0) {
4543df7e 531 info->thread_pool_size = intarg;
2c334e87
WS
532 } else {
533 ret = -EINVAL;
534 goto out;
535 }
4543df7e 536 break;
6f568d35 537 case Opt_max_inline:
edf24abe
CH
538 num = match_strdup(&args[0]);
539 if (num) {
91748467 540 info->max_inline = memparse(num, NULL);
edf24abe
CH
541 kfree(num);
542
15ada040 543 if (info->max_inline) {
feb5f965 544 info->max_inline = min_t(u64,
15ada040
CM
545 info->max_inline,
546 root->sectorsize);
547 }
efe120a0 548 btrfs_info(root->fs_info, "max_inline at %llu",
c1c9ff7c 549 info->max_inline);
2c334e87
WS
550 } else {
551 ret = -ENOMEM;
552 goto out;
6f568d35
CM
553 }
554 break;
8f662a76 555 case Opt_alloc_start:
edf24abe
CH
556 num = match_strdup(&args[0]);
557 if (num) {
c018daec 558 mutex_lock(&info->chunk_mutex);
91748467 559 info->alloc_start = memparse(num, NULL);
c018daec 560 mutex_unlock(&info->chunk_mutex);
edf24abe 561 kfree(num);
efe120a0 562 btrfs_info(root->fs_info, "allocations start at %llu",
c1c9ff7c 563 info->alloc_start);
2c334e87
WS
564 } else {
565 ret = -ENOMEM;
566 goto out;
8f662a76
CM
567 }
568 break;
bd0330ad 569 case Opt_acl:
45ff35d6 570#ifdef CONFIG_BTRFS_FS_POSIX_ACL
bd0330ad
QW
571 root->fs_info->sb->s_flags |= MS_POSIXACL;
572 break;
45ff35d6
GZ
573#else
574 btrfs_err(root->fs_info,
575 "support for ACL not compiled in!");
576 ret = -EINVAL;
577 goto out;
578#endif
33268eaf
JB
579 case Opt_noacl:
580 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
581 break;
3a5e1404 582 case Opt_notreelog:
07802534
QW
583 btrfs_set_and_info(root, NOTREELOG,
584 "disabling tree log");
a88998f2
QW
585 break;
586 case Opt_treelog:
07802534
QW
587 btrfs_clear_and_info(root, NOTREELOG,
588 "enabling tree log");
3a5e1404 589 break;
dccae999 590 case Opt_flushoncommit:
07802534
QW
591 btrfs_set_and_info(root, FLUSHONCOMMIT,
592 "turning on flush-on-commit");
dccae999 593 break;
2c9ee856 594 case Opt_noflushoncommit:
07802534
QW
595 btrfs_clear_and_info(root, FLUSHONCOMMIT,
596 "turning off flush-on-commit");
2c9ee856 597 break;
97e728d4 598 case Opt_ratio:
2c334e87
WS
599 ret = match_int(&args[0], &intarg);
600 if (ret) {
601 goto out;
602 } else if (intarg >= 0) {
97e728d4 603 info->metadata_ratio = intarg;
efe120a0 604 btrfs_info(root->fs_info, "metadata ratio %d",
97e728d4 605 info->metadata_ratio);
2c334e87
WS
606 } else {
607 ret = -EINVAL;
608 goto out;
97e728d4
JB
609 }
610 break;
e244a0ae 611 case Opt_discard:
07802534
QW
612 btrfs_set_and_info(root, DISCARD,
613 "turning on discard");
e244a0ae 614 break;
e07a2ade 615 case Opt_nodiscard:
07802534
QW
616 btrfs_clear_and_info(root, DISCARD,
617 "turning off discard");
e07a2ade 618 break;
0af3d00b 619 case Opt_space_cache:
07802534
QW
620 btrfs_set_and_info(root, SPACE_CACHE,
621 "enabling disk space caching");
0de90876 622 break;
f420ee1e
SB
623 case Opt_rescan_uuid_tree:
624 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
625 break;
73bc1876 626 case Opt_no_space_cache:
07802534
QW
627 btrfs_clear_and_info(root, SPACE_CACHE,
628 "disabling disk space caching");
73bc1876 629 break;
4b9465cb 630 case Opt_inode_cache:
7e1876ac 631 btrfs_set_pending_and_info(info, INODE_MAP_CACHE,
07802534 632 "enabling inode map caching");
3818aea2
QW
633 break;
634 case Opt_noinode_cache:
7e1876ac 635 btrfs_clear_pending_and_info(info, INODE_MAP_CACHE,
07802534 636 "disabling inode map caching");
4b9465cb 637 break;
88c2ba3b 638 case Opt_clear_cache:
07802534
QW
639 btrfs_set_and_info(root, CLEAR_CACHE,
640 "force clearing of disk cache");
0af3d00b 641 break;
4260f7c7
SW
642 case Opt_user_subvol_rm_allowed:
643 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
644 break;
91435650
CM
645 case Opt_enospc_debug:
646 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
647 break;
53036293
QW
648 case Opt_noenospc_debug:
649 btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG);
650 break;
4cb5300b 651 case Opt_defrag:
07802534
QW
652 btrfs_set_and_info(root, AUTO_DEFRAG,
653 "enabling auto defrag");
4cb5300b 654 break;
fc0ca9af 655 case Opt_nodefrag:
07802534
QW
656 btrfs_clear_and_info(root, AUTO_DEFRAG,
657 "disabling auto defrag");
fc0ca9af 658 break;
af31f5e5 659 case Opt_recovery:
efe120a0 660 btrfs_info(root->fs_info, "enabling auto recovery");
af31f5e5
CM
661 btrfs_set_opt(info->mount_opt, RECOVERY);
662 break;
9555c6c1
ID
663 case Opt_skip_balance:
664 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
665 break;
21adbd5c
SB
666#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
667 case Opt_check_integrity_including_extent_data:
efe120a0
FH
668 btrfs_info(root->fs_info,
669 "enabling check integrity including extent data");
21adbd5c
SB
670 btrfs_set_opt(info->mount_opt,
671 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
672 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
673 break;
674 case Opt_check_integrity:
efe120a0 675 btrfs_info(root->fs_info, "enabling check integrity");
21adbd5c
SB
676 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
677 break;
678 case Opt_check_integrity_print_mask:
2c334e87
WS
679 ret = match_int(&args[0], &intarg);
680 if (ret) {
681 goto out;
682 } else if (intarg >= 0) {
21adbd5c 683 info->check_integrity_print_mask = intarg;
efe120a0 684 btrfs_info(root->fs_info, "check_integrity_print_mask 0x%x",
21adbd5c 685 info->check_integrity_print_mask);
2c334e87
WS
686 } else {
687 ret = -EINVAL;
688 goto out;
21adbd5c
SB
689 }
690 break;
691#else
692 case Opt_check_integrity_including_extent_data:
693 case Opt_check_integrity:
694 case Opt_check_integrity_print_mask:
efe120a0
FH
695 btrfs_err(root->fs_info,
696 "support for check_integrity* not compiled in!");
21adbd5c
SB
697 ret = -EINVAL;
698 goto out;
699#endif
8c342930
JM
700 case Opt_fatal_errors:
701 if (strcmp(args[0].from, "panic") == 0)
702 btrfs_set_opt(info->mount_opt,
703 PANIC_ON_FATAL_ERROR);
704 else if (strcmp(args[0].from, "bug") == 0)
705 btrfs_clear_opt(info->mount_opt,
706 PANIC_ON_FATAL_ERROR);
707 else {
708 ret = -EINVAL;
709 goto out;
710 }
711 break;
8b87dc17
DS
712 case Opt_commit_interval:
713 intarg = 0;
714 ret = match_int(&args[0], &intarg);
715 if (ret < 0) {
efe120a0 716 btrfs_err(root->fs_info, "invalid commit interval");
8b87dc17
DS
717 ret = -EINVAL;
718 goto out;
719 }
720 if (intarg > 0) {
721 if (intarg > 300) {
efe120a0 722 btrfs_warn(root->fs_info, "excessive commit interval %d",
8b87dc17
DS
723 intarg);
724 }
725 info->commit_interval = intarg;
726 } else {
efe120a0 727 btrfs_info(root->fs_info, "using default commit interval %ds",
8b87dc17
DS
728 BTRFS_DEFAULT_COMMIT_INTERVAL);
729 info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
730 }
731 break;
d0bd4560
JB
732#ifdef CONFIG_BTRFS_DEBUG
733 case Opt_fragment_all:
734 btrfs_info(root->fs_info, "fragmenting all space");
735 btrfs_set_opt(info->mount_opt, FRAGMENT_DATA);
736 btrfs_set_opt(info->mount_opt, FRAGMENT_METADATA);
737 break;
738 case Opt_fragment_metadata:
739 btrfs_info(root->fs_info, "fragmenting metadata");
740 btrfs_set_opt(info->mount_opt,
741 FRAGMENT_METADATA);
742 break;
743 case Opt_fragment_data:
744 btrfs_info(root->fs_info, "fragmenting data");
745 btrfs_set_opt(info->mount_opt, FRAGMENT_DATA);
746 break;
747#endif
a7a3f7ca 748 case Opt_err:
efe120a0 749 btrfs_info(root->fs_info, "unrecognized mount option '%s'", p);
a7a3f7ca
SW
750 ret = -EINVAL;
751 goto out;
95e05289 752 default:
be20aa9d 753 break;
95e05289
CM
754 }
755 }
a7a3f7ca 756out:
73bc1876 757 if (!ret && btrfs_test_opt(root, SPACE_CACHE))
efe120a0 758 btrfs_info(root->fs_info, "disk space caching is enabled");
da495ecc 759 kfree(orig);
a7a3f7ca 760 return ret;
edf24abe
CH
761}
762
763/*
764 * Parse mount options that are required early in the mount process.
765 *
766 * All other options will be parsed on much later in the mount process and
767 * only when we need to allocate a new super block.
768 */
97288f2c 769static int btrfs_parse_early_options(const char *options, fmode_t flags,
73f73415 770 void *holder, char **subvol_name, u64 *subvol_objectid,
5e2a4b25 771 struct btrfs_fs_devices **fs_devices)
edf24abe
CH
772{
773 substring_t args[MAX_OPT_ARGS];
83c8c9bd 774 char *device_name, *opts, *orig, *p;
1493381f 775 char *num = NULL;
edf24abe
CH
776 int error = 0;
777
778 if (!options)
830c4adb 779 return 0;
edf24abe
CH
780
781 /*
782 * strsep changes the string, duplicate it because parse_options
783 * gets called twice
784 */
785 opts = kstrdup(options, GFP_KERNEL);
786 if (!opts)
787 return -ENOMEM;
3f3d0bc0 788 orig = opts;
edf24abe
CH
789
790 while ((p = strsep(&opts, ",")) != NULL) {
791 int token;
792 if (!*p)
793 continue;
794
795 token = match_token(p, tokens, args);
796 switch (token) {
797 case Opt_subvol:
a90e8b6f 798 kfree(*subvol_name);
edf24abe 799 *subvol_name = match_strdup(&args[0]);
2c334e87
WS
800 if (!*subvol_name) {
801 error = -ENOMEM;
802 goto out;
803 }
edf24abe 804 break;
73f73415 805 case Opt_subvolid:
1493381f
WS
806 num = match_strdup(&args[0]);
807 if (num) {
808 *subvol_objectid = memparse(num, NULL);
809 kfree(num);
4849f01d 810 /* we want the original fs_tree */
1493381f 811 if (!*subvol_objectid)
4849f01d
JB
812 *subvol_objectid =
813 BTRFS_FS_TREE_OBJECTID;
2c334e87
WS
814 } else {
815 error = -EINVAL;
816 goto out;
4849f01d 817 }
73f73415 818 break;
e15d0542 819 case Opt_subvolrootid:
5e2a4b25 820 printk(KERN_WARNING
efe120a0
FH
821 "BTRFS: 'subvolrootid' mount option is deprecated and has "
822 "no effect\n");
e15d0542 823 break;
43e570b0 824 case Opt_device:
83c8c9bd
JL
825 device_name = match_strdup(&args[0]);
826 if (!device_name) {
827 error = -ENOMEM;
828 goto out;
829 }
830 error = btrfs_scan_one_device(device_name,
43e570b0 831 flags, holder, fs_devices);
83c8c9bd 832 kfree(device_name);
43e570b0 833 if (error)
830c4adb 834 goto out;
43e570b0 835 break;
edf24abe
CH
836 default:
837 break;
838 }
839 }
840
830c4adb 841out:
3f3d0bc0 842 kfree(orig);
edf24abe 843 return error;
95e05289
CM
844}
845
05dbe683
OS
846static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
847 u64 subvol_objectid)
73f73415 848{
815745cf 849 struct btrfs_root *root = fs_info->tree_root;
05dbe683
OS
850 struct btrfs_root *fs_root;
851 struct btrfs_root_ref *root_ref;
852 struct btrfs_inode_ref *inode_ref;
853 struct btrfs_key key;
854 struct btrfs_path *path = NULL;
855 char *name = NULL, *ptr;
856 u64 dirid;
857 int len;
858 int ret;
859
860 path = btrfs_alloc_path();
861 if (!path) {
862 ret = -ENOMEM;
863 goto err;
864 }
865 path->leave_spinning = 1;
866
867 name = kmalloc(PATH_MAX, GFP_NOFS);
868 if (!name) {
869 ret = -ENOMEM;
870 goto err;
871 }
872 ptr = name + PATH_MAX - 1;
873 ptr[0] = '\0';
73f73415
JB
874
875 /*
05dbe683
OS
876 * Walk up the subvolume trees in the tree of tree roots by root
877 * backrefs until we hit the top-level subvolume.
73f73415 878 */
05dbe683
OS
879 while (subvol_objectid != BTRFS_FS_TREE_OBJECTID) {
880 key.objectid = subvol_objectid;
881 key.type = BTRFS_ROOT_BACKREF_KEY;
882 key.offset = (u64)-1;
883
884 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
885 if (ret < 0) {
886 goto err;
887 } else if (ret > 0) {
888 ret = btrfs_previous_item(root, path, subvol_objectid,
889 BTRFS_ROOT_BACKREF_KEY);
890 if (ret < 0) {
891 goto err;
892 } else if (ret > 0) {
893 ret = -ENOENT;
894 goto err;
895 }
896 }
897
898 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
899 subvol_objectid = key.offset;
900
901 root_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
902 struct btrfs_root_ref);
903 len = btrfs_root_ref_name_len(path->nodes[0], root_ref);
904 ptr -= len + 1;
905 if (ptr < name) {
906 ret = -ENAMETOOLONG;
907 goto err;
908 }
909 read_extent_buffer(path->nodes[0], ptr + 1,
910 (unsigned long)(root_ref + 1), len);
911 ptr[0] = '/';
912 dirid = btrfs_root_ref_dirid(path->nodes[0], root_ref);
913 btrfs_release_path(path);
914
915 key.objectid = subvol_objectid;
916 key.type = BTRFS_ROOT_ITEM_KEY;
917 key.offset = (u64)-1;
918 fs_root = btrfs_read_fs_root_no_name(fs_info, &key);
919 if (IS_ERR(fs_root)) {
920 ret = PTR_ERR(fs_root);
921 goto err;
922 }
923
924 /*
925 * Walk up the filesystem tree by inode refs until we hit the
926 * root directory.
927 */
928 while (dirid != BTRFS_FIRST_FREE_OBJECTID) {
929 key.objectid = dirid;
930 key.type = BTRFS_INODE_REF_KEY;
931 key.offset = (u64)-1;
932
933 ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
934 if (ret < 0) {
935 goto err;
936 } else if (ret > 0) {
937 ret = btrfs_previous_item(fs_root, path, dirid,
938 BTRFS_INODE_REF_KEY);
939 if (ret < 0) {
940 goto err;
941 } else if (ret > 0) {
942 ret = -ENOENT;
943 goto err;
944 }
945 }
946
947 btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
948 dirid = key.offset;
949
950 inode_ref = btrfs_item_ptr(path->nodes[0],
951 path->slots[0],
952 struct btrfs_inode_ref);
953 len = btrfs_inode_ref_name_len(path->nodes[0],
954 inode_ref);
955 ptr -= len + 1;
956 if (ptr < name) {
957 ret = -ENAMETOOLONG;
958 goto err;
959 }
960 read_extent_buffer(path->nodes[0], ptr + 1,
961 (unsigned long)(inode_ref + 1), len);
962 ptr[0] = '/';
963 btrfs_release_path(path);
964 }
73f73415
JB
965 }
966
05dbe683
OS
967 btrfs_free_path(path);
968 if (ptr == name + PATH_MAX - 1) {
969 name[0] = '/';
970 name[1] = '\0';
971 } else {
972 memmove(name, ptr, name + PATH_MAX - ptr);
973 }
974 return name;
975
976err:
977 btrfs_free_path(path);
978 kfree(name);
979 return ERR_PTR(ret);
980}
981
982static int get_default_subvol_objectid(struct btrfs_fs_info *fs_info, u64 *objectid)
983{
984 struct btrfs_root *root = fs_info->tree_root;
985 struct btrfs_dir_item *di;
986 struct btrfs_path *path;
987 struct btrfs_key location;
988 u64 dir_id;
989
73f73415
JB
990 path = btrfs_alloc_path();
991 if (!path)
05dbe683 992 return -ENOMEM;
73f73415
JB
993 path->leave_spinning = 1;
994
995 /*
996 * Find the "default" dir item which points to the root item that we
997 * will mount by default if we haven't been given a specific subvolume
998 * to mount.
999 */
815745cf 1000 dir_id = btrfs_super_root_dir(fs_info->super_copy);
73f73415 1001 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
b0839166
JL
1002 if (IS_ERR(di)) {
1003 btrfs_free_path(path);
05dbe683 1004 return PTR_ERR(di);
b0839166 1005 }
73f73415
JB
1006 if (!di) {
1007 /*
1008 * Ok the default dir item isn't there. This is weird since
1009 * it's always been there, but don't freak out, just try and
05dbe683 1010 * mount the top-level subvolume.
73f73415
JB
1011 */
1012 btrfs_free_path(path);
05dbe683
OS
1013 *objectid = BTRFS_FS_TREE_OBJECTID;
1014 return 0;
73f73415
JB
1015 }
1016
1017 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
1018 btrfs_free_path(path);
05dbe683
OS
1019 *objectid = location.objectid;
1020 return 0;
73f73415
JB
1021}
1022
d397712b 1023static int btrfs_fill_super(struct super_block *sb,
8a4b83cc 1024 struct btrfs_fs_devices *fs_devices,
d397712b 1025 void *data, int silent)
75dfe396 1026{
d397712b 1027 struct inode *inode;
815745cf 1028 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
5d4f98a2 1029 struct btrfs_key key;
39279cc3 1030 int err;
a429e513 1031
39279cc3
CM
1032 sb->s_maxbytes = MAX_LFS_FILESIZE;
1033 sb->s_magic = BTRFS_SUPER_MAGIC;
1034 sb->s_op = &btrfs_super_ops;
af53d29a 1035 sb->s_d_op = &btrfs_dentry_operations;
be6e8dc0 1036 sb->s_export_op = &btrfs_export_ops;
5103e947 1037 sb->s_xattr = btrfs_xattr_handlers;
39279cc3 1038 sb->s_time_gran = 1;
0eda294d 1039#ifdef CONFIG_BTRFS_FS_POSIX_ACL
33268eaf 1040 sb->s_flags |= MS_POSIXACL;
49cf6f45 1041#endif
0c4d2d95 1042 sb->s_flags |= MS_I_VERSION;
da2f0f74 1043 sb->s_iflags |= SB_I_CGROUPWB;
ad2b2c80
AV
1044 err = open_ctree(sb, fs_devices, (char *)data);
1045 if (err) {
efe120a0 1046 printk(KERN_ERR "BTRFS: open_ctree failed\n");
ad2b2c80 1047 return err;
a429e513
CM
1048 }
1049
5d4f98a2
YZ
1050 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
1051 key.type = BTRFS_INODE_ITEM_KEY;
1052 key.offset = 0;
98c7089c 1053 inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL);
5d4f98a2
YZ
1054 if (IS_ERR(inode)) {
1055 err = PTR_ERR(inode);
39279cc3 1056 goto fail_close;
f254e52c 1057 }
f254e52c 1058
48fde701
AV
1059 sb->s_root = d_make_root(inode);
1060 if (!sb->s_root) {
39279cc3
CM
1061 err = -ENOMEM;
1062 goto fail_close;
f254e52c 1063 }
58176a96 1064
6885f308 1065 save_mount_options(sb, data);
90a887c9 1066 cleancache_init_fs(sb);
59553edf 1067 sb->s_flags |= MS_ACTIVE;
2619ba1f 1068 return 0;
39279cc3
CM
1069
1070fail_close:
815745cf 1071 close_ctree(fs_info->tree_root);
39279cc3 1072 return err;
2619ba1f
CM
1073}
1074
6bf13c0c 1075int btrfs_sync_fs(struct super_block *sb, int wait)
c5739bba
CM
1076{
1077 struct btrfs_trans_handle *trans;
815745cf
AV
1078 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1079 struct btrfs_root *root = fs_info->tree_root;
2619ba1f 1080
1abe9b8a 1081 trace_btrfs_sync_fs(wait);
1082
39279cc3 1083 if (!wait) {
815745cf 1084 filemap_flush(fs_info->btree_inode->i_mapping);
39279cc3
CM
1085 return 0;
1086 }
771ed689 1087
b0244199 1088 btrfs_wait_ordered_roots(fs_info, -1);
771ed689 1089
d4edf39b 1090 trans = btrfs_attach_transaction_barrier(root);
60376ce4 1091 if (IS_ERR(trans)) {
354aa0fb 1092 /* no transaction, don't bother */
6b5fe46d
DS
1093 if (PTR_ERR(trans) == -ENOENT) {
1094 /*
1095 * Exit unless we have some pending changes
1096 * that need to go through commit
1097 */
1098 if (fs_info->pending_changes == 0)
1099 return 0;
a53f4f8e
QW
1100 /*
1101 * A non-blocking test if the fs is frozen. We must not
1102 * start a new transaction here otherwise a deadlock
1103 * happens. The pending operations are delayed to the
1104 * next commit after thawing.
1105 */
1106 if (__sb_start_write(sb, SB_FREEZE_WRITE, false))
1107 __sb_end_write(sb, SB_FREEZE_WRITE);
1108 else
1109 return 0;
6b5fe46d 1110 trans = btrfs_start_transaction(root, 0);
6b5fe46d 1111 }
98bd5c54
DS
1112 if (IS_ERR(trans))
1113 return PTR_ERR(trans);
60376ce4 1114 }
bd7de2c9 1115 return btrfs_commit_transaction(trans, root);
2c90e5d6
CM
1116}
1117
34c80b1d 1118static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
a9572a15 1119{
815745cf
AV
1120 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
1121 struct btrfs_root *root = info->tree_root;
200da64e 1122 char *compress_type;
a9572a15
EP
1123
1124 if (btrfs_test_opt(root, DEGRADED))
1125 seq_puts(seq, ",degraded");
1126 if (btrfs_test_opt(root, NODATASUM))
1127 seq_puts(seq, ",nodatasum");
1128 if (btrfs_test_opt(root, NODATACOW))
1129 seq_puts(seq, ",nodatacow");
1130 if (btrfs_test_opt(root, NOBARRIER))
1131 seq_puts(seq, ",nobarrier");
95ac567a 1132 if (info->max_inline != BTRFS_DEFAULT_MAX_INLINE)
c1c9ff7c 1133 seq_printf(seq, ",max_inline=%llu", info->max_inline);
a9572a15 1134 if (info->alloc_start != 0)
c1c9ff7c 1135 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
a9572a15
EP
1136 if (info->thread_pool_size != min_t(unsigned long,
1137 num_online_cpus() + 2, 8))
1138 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
200da64e
TI
1139 if (btrfs_test_opt(root, COMPRESS)) {
1140 if (info->compress_type == BTRFS_COMPRESS_ZLIB)
1141 compress_type = "zlib";
1142 else
1143 compress_type = "lzo";
1144 if (btrfs_test_opt(root, FORCE_COMPRESS))
1145 seq_printf(seq, ",compress-force=%s", compress_type);
1146 else
1147 seq_printf(seq, ",compress=%s", compress_type);
1148 }
c289811c
CM
1149 if (btrfs_test_opt(root, NOSSD))
1150 seq_puts(seq, ",nossd");
451d7585
CM
1151 if (btrfs_test_opt(root, SSD_SPREAD))
1152 seq_puts(seq, ",ssd_spread");
1153 else if (btrfs_test_opt(root, SSD))
a9572a15 1154 seq_puts(seq, ",ssd");
3a5e1404 1155 if (btrfs_test_opt(root, NOTREELOG))
6b65c5c6 1156 seq_puts(seq, ",notreelog");
dccae999 1157 if (btrfs_test_opt(root, FLUSHONCOMMIT))
6b65c5c6 1158 seq_puts(seq, ",flushoncommit");
20a5239a
MW
1159 if (btrfs_test_opt(root, DISCARD))
1160 seq_puts(seq, ",discard");
a9572a15
EP
1161 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
1162 seq_puts(seq, ",noacl");
200da64e
TI
1163 if (btrfs_test_opt(root, SPACE_CACHE))
1164 seq_puts(seq, ",space_cache");
73bc1876 1165 else
8965593e 1166 seq_puts(seq, ",nospace_cache");
f420ee1e
SB
1167 if (btrfs_test_opt(root, RESCAN_UUID_TREE))
1168 seq_puts(seq, ",rescan_uuid_tree");
200da64e
TI
1169 if (btrfs_test_opt(root, CLEAR_CACHE))
1170 seq_puts(seq, ",clear_cache");
1171 if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
1172 seq_puts(seq, ",user_subvol_rm_allowed");
0942caa3
DS
1173 if (btrfs_test_opt(root, ENOSPC_DEBUG))
1174 seq_puts(seq, ",enospc_debug");
1175 if (btrfs_test_opt(root, AUTO_DEFRAG))
1176 seq_puts(seq, ",autodefrag");
1177 if (btrfs_test_opt(root, INODE_MAP_CACHE))
1178 seq_puts(seq, ",inode_cache");
9555c6c1
ID
1179 if (btrfs_test_opt(root, SKIP_BALANCE))
1180 seq_puts(seq, ",skip_balance");
8507d216
WS
1181 if (btrfs_test_opt(root, RECOVERY))
1182 seq_puts(seq, ",recovery");
1183#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1184 if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1185 seq_puts(seq, ",check_int_data");
1186 else if (btrfs_test_opt(root, CHECK_INTEGRITY))
1187 seq_puts(seq, ",check_int");
1188 if (info->check_integrity_print_mask)
1189 seq_printf(seq, ",check_int_print_mask=%d",
1190 info->check_integrity_print_mask);
1191#endif
1192 if (info->metadata_ratio)
1193 seq_printf(seq, ",metadata_ratio=%d",
1194 info->metadata_ratio);
8c342930
JM
1195 if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
1196 seq_puts(seq, ",fatal_errors=panic");
8b87dc17
DS
1197 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1198 seq_printf(seq, ",commit=%d", info->commit_interval);
d0bd4560
JB
1199#ifdef CONFIG_BTRFS_DEBUG
1200 if (btrfs_test_opt(root, FRAGMENT_DATA))
1201 seq_puts(seq, ",fragment=data");
1202 if (btrfs_test_opt(root, FRAGMENT_METADATA))
1203 seq_puts(seq, ",fragment=metadata");
1204#endif
c8d3fe02
OS
1205 seq_printf(seq, ",subvolid=%llu",
1206 BTRFS_I(d_inode(dentry))->root->root_key.objectid);
1207 seq_puts(seq, ",subvol=");
1208 seq_dentry(seq, dentry, " \t\n\\");
a9572a15
EP
1209 return 0;
1210}
1211
a061fc8d 1212static int btrfs_test_super(struct super_block *s, void *data)
4b82d6e4 1213{
815745cf
AV
1214 struct btrfs_fs_info *p = data;
1215 struct btrfs_fs_info *fs_info = btrfs_sb(s);
4b82d6e4 1216
815745cf 1217 return fs_info->fs_devices == p->fs_devices;
4b82d6e4
Y
1218}
1219
450ba0ea
JB
1220static int btrfs_set_super(struct super_block *s, void *data)
1221{
6de1d09d
AV
1222 int err = set_anon_super(s, data);
1223 if (!err)
1224 s->s_fs_info = data;
1225 return err;
4b82d6e4
Y
1226}
1227
f9d9ef62
DS
1228/*
1229 * subvolumes are identified by ino 256
1230 */
1231static inline int is_subvolume_inode(struct inode *inode)
1232{
1233 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1234 return 1;
1235 return 0;
1236}
1237
830c4adb 1238/*
e6e4dbe8
OS
1239 * This will add subvolid=0 to the argument string while removing any subvol=
1240 * and subvolid= arguments to make sure we get the top-level root for path
1241 * walking to the subvol we want.
830c4adb
JB
1242 */
1243static char *setup_root_args(char *args)
1244{
e6e4dbe8 1245 char *buf, *dst, *sep;
830c4adb 1246
e6e4dbe8
OS
1247 if (!args)
1248 return kstrdup("subvolid=0", GFP_NOFS);
f60d16a8 1249
e6e4dbe8
OS
1250 /* The worst case is that we add ",subvolid=0" to the end. */
1251 buf = dst = kmalloc(strlen(args) + strlen(",subvolid=0") + 1, GFP_NOFS);
f60d16a8 1252 if (!buf)
830c4adb 1253 return NULL;
830c4adb 1254
e6e4dbe8
OS
1255 while (1) {
1256 sep = strchrnul(args, ',');
1257 if (!strstarts(args, "subvol=") &&
1258 !strstarts(args, "subvolid=")) {
1259 memcpy(dst, args, sep - args);
1260 dst += sep - args;
1261 *dst++ = ',';
1262 }
1263 if (*sep)
1264 args = sep + 1;
1265 else
1266 break;
830c4adb 1267 }
f60d16a8 1268 strcpy(dst, "subvolid=0");
830c4adb 1269
f60d16a8 1270 return buf;
830c4adb
JB
1271}
1272
bb289b7b
OS
1273static struct dentry *mount_subvol(const char *subvol_name, u64 subvol_objectid,
1274 int flags, const char *device_name,
1275 char *data)
830c4adb 1276{
830c4adb 1277 struct dentry *root;
fa330659 1278 struct vfsmount *mnt = NULL;
830c4adb 1279 char *newargs;
fa330659 1280 int ret;
830c4adb
JB
1281
1282 newargs = setup_root_args(data);
fa330659
OS
1283 if (!newargs) {
1284 root = ERR_PTR(-ENOMEM);
1285 goto out;
1286 }
0723a047 1287
fa330659
OS
1288 mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name, newargs);
1289 if (PTR_ERR_OR_ZERO(mnt) == -EBUSY) {
0723a047 1290 if (flags & MS_RDONLY) {
fa330659
OS
1291 mnt = vfs_kern_mount(&btrfs_fs_type, flags & ~MS_RDONLY,
1292 device_name, newargs);
0723a047 1293 } else {
fa330659
OS
1294 mnt = vfs_kern_mount(&btrfs_fs_type, flags | MS_RDONLY,
1295 device_name, newargs);
0040e606 1296 if (IS_ERR(mnt)) {
fa330659
OS
1297 root = ERR_CAST(mnt);
1298 mnt = NULL;
1299 goto out;
0040e606 1300 }
0723a047 1301
773cd04e 1302 down_write(&mnt->mnt_sb->s_umount);
fa330659 1303 ret = btrfs_remount(mnt->mnt_sb, &flags, NULL);
773cd04e 1304 up_write(&mnt->mnt_sb->s_umount);
fa330659
OS
1305 if (ret < 0) {
1306 root = ERR_PTR(ret);
1307 goto out;
0723a047
HH
1308 }
1309 }
1310 }
fa330659
OS
1311 if (IS_ERR(mnt)) {
1312 root = ERR_CAST(mnt);
1313 mnt = NULL;
1314 goto out;
1315 }
830c4adb 1316
05dbe683
OS
1317 if (!subvol_name) {
1318 if (!subvol_objectid) {
1319 ret = get_default_subvol_objectid(btrfs_sb(mnt->mnt_sb),
1320 &subvol_objectid);
1321 if (ret) {
1322 root = ERR_PTR(ret);
1323 goto out;
1324 }
1325 }
1326 subvol_name = get_subvol_name_from_objectid(btrfs_sb(mnt->mnt_sb),
1327 subvol_objectid);
1328 if (IS_ERR(subvol_name)) {
1329 root = ERR_CAST(subvol_name);
1330 subvol_name = NULL;
1331 goto out;
1332 }
1333
1334 }
1335
ea441d11 1336 root = mount_subtree(mnt, subvol_name);
fa330659
OS
1337 /* mount_subtree() drops our reference on the vfsmount. */
1338 mnt = NULL;
830c4adb 1339
bb289b7b 1340 if (!IS_ERR(root)) {
ea441d11 1341 struct super_block *s = root->d_sb;
bb289b7b
OS
1342 struct inode *root_inode = d_inode(root);
1343 u64 root_objectid = BTRFS_I(root_inode)->root->root_key.objectid;
1344
1345 ret = 0;
1346 if (!is_subvolume_inode(root_inode)) {
1347 pr_err("BTRFS: '%s' is not a valid subvolume\n",
1348 subvol_name);
1349 ret = -EINVAL;
1350 }
1351 if (subvol_objectid && root_objectid != subvol_objectid) {
05dbe683
OS
1352 /*
1353 * This will also catch a race condition where a
1354 * subvolume which was passed by ID is renamed and
1355 * another subvolume is renamed over the old location.
1356 */
bb289b7b
OS
1357 pr_err("BTRFS: subvol '%s' does not match subvolid %llu\n",
1358 subvol_name, subvol_objectid);
1359 ret = -EINVAL;
1360 }
1361 if (ret) {
1362 dput(root);
1363 root = ERR_PTR(ret);
1364 deactivate_locked_super(s);
1365 }
f9d9ef62
DS
1366 }
1367
fa330659
OS
1368out:
1369 mntput(mnt);
1370 kfree(newargs);
1371 kfree(subvol_name);
830c4adb
JB
1372 return root;
1373}
450ba0ea 1374
f667aef6
QW
1375static int parse_security_options(char *orig_opts,
1376 struct security_mnt_opts *sec_opts)
1377{
1378 char *secdata = NULL;
1379 int ret = 0;
1380
1381 secdata = alloc_secdata();
1382 if (!secdata)
1383 return -ENOMEM;
1384 ret = security_sb_copy_data(orig_opts, secdata);
1385 if (ret) {
1386 free_secdata(secdata);
1387 return ret;
1388 }
1389 ret = security_sb_parse_opts_str(secdata, sec_opts);
1390 free_secdata(secdata);
1391 return ret;
1392}
1393
1394static int setup_security_options(struct btrfs_fs_info *fs_info,
1395 struct super_block *sb,
1396 struct security_mnt_opts *sec_opts)
1397{
1398 int ret = 0;
1399
1400 /*
1401 * Call security_sb_set_mnt_opts() to check whether new sec_opts
1402 * is valid.
1403 */
1404 ret = security_sb_set_mnt_opts(sb, sec_opts, 0, NULL);
1405 if (ret)
1406 return ret;
1407
a43bb39b 1408#ifdef CONFIG_SECURITY
f667aef6
QW
1409 if (!fs_info->security_opts.num_mnt_opts) {
1410 /* first time security setup, copy sec_opts to fs_info */
1411 memcpy(&fs_info->security_opts, sec_opts, sizeof(*sec_opts));
1412 } else {
1413 /*
1414 * Since SELinux(the only one supports security_mnt_opts) does
1415 * NOT support changing context during remount/mount same sb,
1416 * This must be the same or part of the same security options,
1417 * just free it.
1418 */
1419 security_free_mnt_opts(sec_opts);
1420 }
a43bb39b 1421#endif
f667aef6
QW
1422 return ret;
1423}
1424
edf24abe
CH
1425/*
1426 * Find a superblock for the given device / mount point.
1427 *
1428 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1429 * for multiple device setup. Make sure to keep it in sync.
1430 */
061dbc6b 1431static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
306e16ce 1432 const char *device_name, void *data)
4b82d6e4
Y
1433{
1434 struct block_device *bdev = NULL;
1435 struct super_block *s;
8a4b83cc 1436 struct btrfs_fs_devices *fs_devices = NULL;
450ba0ea 1437 struct btrfs_fs_info *fs_info = NULL;
f667aef6 1438 struct security_mnt_opts new_sec_opts;
97288f2c 1439 fmode_t mode = FMODE_READ;
73f73415
JB
1440 char *subvol_name = NULL;
1441 u64 subvol_objectid = 0;
4b82d6e4
Y
1442 int error = 0;
1443
97288f2c
CH
1444 if (!(flags & MS_RDONLY))
1445 mode |= FMODE_WRITE;
1446
1447 error = btrfs_parse_early_options(data, mode, fs_type,
73f73415 1448 &subvol_name, &subvol_objectid,
5e2a4b25 1449 &fs_devices);
f23c8af8
ID
1450 if (error) {
1451 kfree(subvol_name);
061dbc6b 1452 return ERR_PTR(error);
f23c8af8 1453 }
edf24abe 1454
05dbe683 1455 if (subvol_name || subvol_objectid != BTRFS_FS_TREE_OBJECTID) {
fa330659 1456 /* mount_subvol() will free subvol_name. */
bb289b7b
OS
1457 return mount_subvol(subvol_name, subvol_objectid, flags,
1458 device_name, data);
830c4adb
JB
1459 }
1460
f667aef6
QW
1461 security_init_mnt_opts(&new_sec_opts);
1462 if (data) {
1463 error = parse_security_options(data, &new_sec_opts);
1464 if (error)
1465 return ERR_PTR(error);
1466 }
1467
306e16ce 1468 error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices);
8a4b83cc 1469 if (error)
f667aef6 1470 goto error_sec_opts;
4b82d6e4 1471
450ba0ea
JB
1472 /*
1473 * Setup a dummy root and fs_info for test/set super. This is because
1474 * we don't actually fill this stuff out until open_ctree, but we need
1475 * it for searching for existing supers, so this lets us do that and
1476 * then open_ctree will properly initialize everything later.
1477 */
1478 fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
f667aef6
QW
1479 if (!fs_info) {
1480 error = -ENOMEM;
1481 goto error_sec_opts;
1482 }
04d21a24 1483
450ba0ea 1484 fs_info->fs_devices = fs_devices;
450ba0ea 1485
6c41761f
DS
1486 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1487 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
f667aef6 1488 security_init_mnt_opts(&fs_info->security_opts);
6c41761f
DS
1489 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1490 error = -ENOMEM;
04d21a24
ID
1491 goto error_fs_info;
1492 }
1493
1494 error = btrfs_open_devices(fs_devices, mode, fs_type);
1495 if (error)
1496 goto error_fs_info;
1497
1498 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
1499 error = -EACCES;
6c41761f
DS
1500 goto error_close_devices;
1501 }
1502
dfe25020 1503 bdev = fs_devices->latest_bdev;
9249e17f
DH
1504 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC,
1505 fs_info);
830c4adb
JB
1506 if (IS_ERR(s)) {
1507 error = PTR_ERR(s);
1508 goto error_close_devices;
1509 }
4b82d6e4
Y
1510
1511 if (s->s_root) {
2b82032c 1512 btrfs_close_devices(fs_devices);
6c41761f 1513 free_fs_info(fs_info);
59553edf
AV
1514 if ((flags ^ s->s_flags) & MS_RDONLY)
1515 error = -EBUSY;
4b82d6e4
Y
1516 } else {
1517 char b[BDEVNAME_SIZE];
1518
4b82d6e4 1519 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
815745cf 1520 btrfs_sb(s)->bdev_holder = fs_type;
8a4b83cc
CM
1521 error = btrfs_fill_super(s, fs_devices, data,
1522 flags & MS_SILENT ? 1 : 0);
4b82d6e4 1523 }
05dbe683 1524 if (error) {
f667aef6 1525 deactivate_locked_super(s);
f667aef6
QW
1526 goto error_sec_opts;
1527 }
1528
1529 fs_info = btrfs_sb(s);
1530 error = setup_security_options(fs_info, s, &new_sec_opts);
1531 if (error) {
830c4adb 1532 deactivate_locked_super(s);
f667aef6
QW
1533 goto error_sec_opts;
1534 }
4b82d6e4 1535
05dbe683 1536 return dget(s->s_root);
4b82d6e4 1537
c146afad 1538error_close_devices:
8a4b83cc 1539 btrfs_close_devices(fs_devices);
04d21a24 1540error_fs_info:
6c41761f 1541 free_fs_info(fs_info);
f667aef6
QW
1542error_sec_opts:
1543 security_free_mnt_opts(&new_sec_opts);
061dbc6b 1544 return ERR_PTR(error);
4b82d6e4 1545}
2e635a27 1546
0d2450ab
ST
1547static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1548 int new_pool_size, int old_pool_size)
1549{
1550 if (new_pool_size == old_pool_size)
1551 return;
1552
1553 fs_info->thread_pool_size = new_pool_size;
1554
efe120a0 1555 btrfs_info(fs_info, "resize thread pool %d -> %d",
0d2450ab
ST
1556 old_pool_size, new_pool_size);
1557
5cdc7ad3 1558 btrfs_workqueue_set_max(fs_info->workers, new_pool_size);
afe3d242 1559 btrfs_workqueue_set_max(fs_info->delalloc_workers, new_pool_size);
a8c93d4e 1560 btrfs_workqueue_set_max(fs_info->submit_workers, new_pool_size);
e66f0bb1 1561 btrfs_workqueue_set_max(fs_info->caching_workers, new_pool_size);
fccb5d86
QW
1562 btrfs_workqueue_set_max(fs_info->endio_workers, new_pool_size);
1563 btrfs_workqueue_set_max(fs_info->endio_meta_workers, new_pool_size);
1564 btrfs_workqueue_set_max(fs_info->endio_meta_write_workers,
1565 new_pool_size);
1566 btrfs_workqueue_set_max(fs_info->endio_write_workers, new_pool_size);
1567 btrfs_workqueue_set_max(fs_info->endio_freespace_worker, new_pool_size);
5b3bc44e 1568 btrfs_workqueue_set_max(fs_info->delayed_workers, new_pool_size);
736cfa15 1569 btrfs_workqueue_set_max(fs_info->readahead_workers, new_pool_size);
0339ef2f
QW
1570 btrfs_workqueue_set_max(fs_info->scrub_wr_completion_workers,
1571 new_pool_size);
0d2450ab
ST
1572}
1573
f42a34b2 1574static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
dc81cdc5
MX
1575{
1576 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
f42a34b2 1577}
dc81cdc5 1578
f42a34b2
MX
1579static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1580 unsigned long old_opts, int flags)
1581{
dc81cdc5
MX
1582 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1583 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1584 (flags & MS_RDONLY))) {
1585 /* wait for any defraggers to finish */
1586 wait_event(fs_info->transaction_wait,
1587 (atomic_read(&fs_info->defrag_running) == 0));
1588 if (flags & MS_RDONLY)
1589 sync_filesystem(fs_info->sb);
1590 }
1591}
1592
1593static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1594 unsigned long old_opts)
1595{
1596 /*
1597 * We need cleanup all defragable inodes if the autodefragment is
1598 * close or the fs is R/O.
1599 */
1600 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1601 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1602 (fs_info->sb->s_flags & MS_RDONLY))) {
1603 btrfs_cleanup_defrag_inodes(fs_info);
1604 }
1605
1606 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1607}
1608
c146afad
YZ
1609static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1610{
815745cf
AV
1611 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1612 struct btrfs_root *root = fs_info->tree_root;
49b25e05
JM
1613 unsigned old_flags = sb->s_flags;
1614 unsigned long old_opts = fs_info->mount_opt;
1615 unsigned long old_compress_type = fs_info->compress_type;
1616 u64 old_max_inline = fs_info->max_inline;
1617 u64 old_alloc_start = fs_info->alloc_start;
1618 int old_thread_pool_size = fs_info->thread_pool_size;
1619 unsigned int old_metadata_ratio = fs_info->metadata_ratio;
c146afad
YZ
1620 int ret;
1621
02b9984d 1622 sync_filesystem(sb);
f42a34b2 1623 btrfs_remount_prepare(fs_info);
dc81cdc5 1624
f667aef6
QW
1625 if (data) {
1626 struct security_mnt_opts new_sec_opts;
1627
1628 security_init_mnt_opts(&new_sec_opts);
1629 ret = parse_security_options(data, &new_sec_opts);
1630 if (ret)
1631 goto restore;
1632 ret = setup_security_options(fs_info, sb,
1633 &new_sec_opts);
1634 if (ret) {
1635 security_free_mnt_opts(&new_sec_opts);
1636 goto restore;
1637 }
1638 }
1639
b288052e 1640 ret = btrfs_parse_options(root, data);
49b25e05
JM
1641 if (ret) {
1642 ret = -EINVAL;
1643 goto restore;
1644 }
b288052e 1645
f42a34b2 1646 btrfs_remount_begin(fs_info, old_opts, *flags);
0d2450ab
ST
1647 btrfs_resize_thread_pool(fs_info,
1648 fs_info->thread_pool_size, old_thread_pool_size);
1649
c146afad 1650 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
dc81cdc5 1651 goto out;
c146afad
YZ
1652
1653 if (*flags & MS_RDONLY) {
8dabb742
SB
1654 /*
1655 * this also happens on 'umount -rf' or on shutdown, when
1656 * the filesystem is busy.
1657 */
21c7e756 1658 cancel_work_sync(&fs_info->async_reclaim_work);
361c093d
SB
1659
1660 /* wait for the uuid_scan task to finish */
1661 down(&fs_info->uuid_tree_rescan_sem);
1662 /* avoid complains from lockdep et al. */
1663 up(&fs_info->uuid_tree_rescan_sem);
1664
c146afad
YZ
1665 sb->s_flags |= MS_RDONLY;
1666
e44163e1
JM
1667 /*
1668 * Setting MS_RDONLY will put the cleaner thread to
1669 * sleep at the next loop if it's already active.
1670 * If it's already asleep, we'll leave unused block
1671 * groups on disk until we're mounted read-write again
1672 * unless we clean them up here.
1673 */
e44163e1 1674 btrfs_delete_unused_bgs(fs_info);
e44163e1 1675
8dabb742
SB
1676 btrfs_dev_replace_suspend_for_unmount(fs_info);
1677 btrfs_scrub_cancel(fs_info);
061594ef 1678 btrfs_pause_balance(fs_info);
8dabb742 1679
49b25e05
JM
1680 ret = btrfs_commit_super(root);
1681 if (ret)
1682 goto restore;
c146afad 1683 } else {
6ef3de9c
DS
1684 if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) {
1685 btrfs_err(fs_info,
efe120a0 1686 "Remounting read-write after error is not allowed");
6ef3de9c
DS
1687 ret = -EINVAL;
1688 goto restore;
1689 }
8a3db184 1690 if (fs_info->fs_devices->rw_devices == 0) {
49b25e05
JM
1691 ret = -EACCES;
1692 goto restore;
8a3db184 1693 }
2b82032c 1694
292fd7fc
SB
1695 if (fs_info->fs_devices->missing_devices >
1696 fs_info->num_tolerated_disk_barrier_failures &&
1697 !(*flags & MS_RDONLY)) {
efe120a0
FH
1698 btrfs_warn(fs_info,
1699 "too many missing devices, writeable remount is not allowed");
292fd7fc
SB
1700 ret = -EACCES;
1701 goto restore;
1702 }
1703
8a3db184 1704 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
49b25e05
JM
1705 ret = -EINVAL;
1706 goto restore;
8a3db184 1707 }
c146afad 1708
815745cf 1709 ret = btrfs_cleanup_fs_roots(fs_info);
49b25e05
JM
1710 if (ret)
1711 goto restore;
c146afad 1712
d68fc57b 1713 /* recover relocation */
5f316481 1714 mutex_lock(&fs_info->cleaner_mutex);
d68fc57b 1715 ret = btrfs_recover_relocation(root);
5f316481 1716 mutex_unlock(&fs_info->cleaner_mutex);
49b25e05
JM
1717 if (ret)
1718 goto restore;
c146afad 1719
2b6ba629
ID
1720 ret = btrfs_resume_balance_async(fs_info);
1721 if (ret)
1722 goto restore;
1723
8dabb742
SB
1724 ret = btrfs_resume_dev_replace_async(fs_info);
1725 if (ret) {
efe120a0 1726 btrfs_warn(fs_info, "failed to resume dev_replace");
8dabb742
SB
1727 goto restore;
1728 }
94aebfb2
JB
1729
1730 if (!fs_info->uuid_root) {
efe120a0 1731 btrfs_info(fs_info, "creating UUID tree");
94aebfb2
JB
1732 ret = btrfs_create_uuid_tree(fs_info);
1733 if (ret) {
efe120a0 1734 btrfs_warn(fs_info, "failed to create the UUID tree %d", ret);
94aebfb2
JB
1735 goto restore;
1736 }
1737 }
c146afad
YZ
1738 sb->s_flags &= ~MS_RDONLY;
1739 }
dc81cdc5 1740out:
2c6a92b0 1741 wake_up_process(fs_info->transaction_kthread);
dc81cdc5 1742 btrfs_remount_cleanup(fs_info, old_opts);
c146afad 1743 return 0;
49b25e05
JM
1744
1745restore:
1746 /* We've hit an error - don't reset MS_RDONLY */
1747 if (sb->s_flags & MS_RDONLY)
1748 old_flags |= MS_RDONLY;
1749 sb->s_flags = old_flags;
1750 fs_info->mount_opt = old_opts;
1751 fs_info->compress_type = old_compress_type;
1752 fs_info->max_inline = old_max_inline;
c018daec 1753 mutex_lock(&fs_info->chunk_mutex);
49b25e05 1754 fs_info->alloc_start = old_alloc_start;
c018daec 1755 mutex_unlock(&fs_info->chunk_mutex);
0d2450ab
ST
1756 btrfs_resize_thread_pool(fs_info,
1757 old_thread_pool_size, fs_info->thread_pool_size);
49b25e05 1758 fs_info->metadata_ratio = old_metadata_ratio;
dc81cdc5 1759 btrfs_remount_cleanup(fs_info, old_opts);
49b25e05 1760 return ret;
c146afad
YZ
1761}
1762
bcd53741
AJ
1763/* Used to sort the devices by max_avail(descending sort) */
1764static int btrfs_cmp_device_free_bytes(const void *dev_info1,
1765 const void *dev_info2)
1766{
1767 if (((struct btrfs_device_info *)dev_info1)->max_avail >
1768 ((struct btrfs_device_info *)dev_info2)->max_avail)
1769 return -1;
1770 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
1771 ((struct btrfs_device_info *)dev_info2)->max_avail)
1772 return 1;
1773 else
1774 return 0;
1775}
1776
1777/*
1778 * sort the devices by max_avail, in which max free extent size of each device
1779 * is stored.(Descending Sort)
1780 */
1781static inline void btrfs_descending_sort_devices(
1782 struct btrfs_device_info *devices,
1783 size_t nr_devices)
1784{
1785 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
1786 btrfs_cmp_device_free_bytes, NULL);
1787}
1788
6d07bcec
MX
1789/*
1790 * The helper to calc the free space on the devices that can be used to store
1791 * file data.
1792 */
1793static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
1794{
1795 struct btrfs_fs_info *fs_info = root->fs_info;
1796 struct btrfs_device_info *devices_info;
1797 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
1798 struct btrfs_device *device;
1799 u64 skip_space;
1800 u64 type;
1801 u64 avail_space;
1802 u64 used_space;
1803 u64 min_stripe_size;
39fb26c3 1804 int min_stripes = 1, num_stripes = 1;
6d07bcec
MX
1805 int i = 0, nr_devices;
1806 int ret;
1807
7e33fd99
JB
1808 /*
1809 * We aren't under the device list lock, so this is racey-ish, but good
1810 * enough for our purposes.
1811 */
b772a86e 1812 nr_devices = fs_info->fs_devices->open_devices;
7e33fd99
JB
1813 if (!nr_devices) {
1814 smp_mb();
1815 nr_devices = fs_info->fs_devices->open_devices;
1816 ASSERT(nr_devices);
1817 if (!nr_devices) {
1818 *free_bytes = 0;
1819 return 0;
1820 }
1821 }
6d07bcec 1822
d9b0d9ba 1823 devices_info = kmalloc_array(nr_devices, sizeof(*devices_info),
6d07bcec
MX
1824 GFP_NOFS);
1825 if (!devices_info)
1826 return -ENOMEM;
1827
1828 /* calc min stripe number for data space alloction */
1829 type = btrfs_get_alloc_profile(root, 1);
39fb26c3 1830 if (type & BTRFS_BLOCK_GROUP_RAID0) {
6d07bcec 1831 min_stripes = 2;
39fb26c3
MX
1832 num_stripes = nr_devices;
1833 } else if (type & BTRFS_BLOCK_GROUP_RAID1) {
6d07bcec 1834 min_stripes = 2;
39fb26c3
MX
1835 num_stripes = 2;
1836 } else if (type & BTRFS_BLOCK_GROUP_RAID10) {
6d07bcec 1837 min_stripes = 4;
39fb26c3
MX
1838 num_stripes = 4;
1839 }
6d07bcec
MX
1840
1841 if (type & BTRFS_BLOCK_GROUP_DUP)
1842 min_stripe_size = 2 * BTRFS_STRIPE_LEN;
1843 else
1844 min_stripe_size = BTRFS_STRIPE_LEN;
1845
7e33fd99
JB
1846 if (fs_info->alloc_start)
1847 mutex_lock(&fs_devices->device_list_mutex);
1848 rcu_read_lock();
1849 list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) {
63a212ab
SB
1850 if (!device->in_fs_metadata || !device->bdev ||
1851 device->is_tgtdev_for_dev_replace)
6d07bcec
MX
1852 continue;
1853
7e33fd99
JB
1854 if (i >= nr_devices)
1855 break;
1856
6d07bcec
MX
1857 avail_space = device->total_bytes - device->bytes_used;
1858
1859 /* align with stripe_len */
f8c269d7 1860 avail_space = div_u64(avail_space, BTRFS_STRIPE_LEN);
6d07bcec
MX
1861 avail_space *= BTRFS_STRIPE_LEN;
1862
1863 /*
1864 * In order to avoid overwritting the superblock on the drive,
1865 * btrfs starts at an offset of at least 1MB when doing chunk
1866 * allocation.
1867 */
1868 skip_space = 1024 * 1024;
1869
1870 /* user can set the offset in fs_info->alloc_start. */
7e33fd99
JB
1871 if (fs_info->alloc_start &&
1872 fs_info->alloc_start + BTRFS_STRIPE_LEN <=
1873 device->total_bytes) {
1874 rcu_read_unlock();
6d07bcec
MX
1875 skip_space = max(fs_info->alloc_start, skip_space);
1876
7e33fd99
JB
1877 /*
1878 * btrfs can not use the free space in
1879 * [0, skip_space - 1], we must subtract it from the
1880 * total. In order to implement it, we account the used
1881 * space in this range first.
1882 */
1883 ret = btrfs_account_dev_extents_size(device, 0,
1884 skip_space - 1,
1885 &used_space);
1886 if (ret) {
1887 kfree(devices_info);
1888 mutex_unlock(&fs_devices->device_list_mutex);
1889 return ret;
1890 }
1891
1892 rcu_read_lock();
6d07bcec 1893
7e33fd99
JB
1894 /* calc the free space in [0, skip_space - 1] */
1895 skip_space -= used_space;
1896 }
6d07bcec
MX
1897
1898 /*
1899 * we can use the free space in [0, skip_space - 1], subtract
1900 * it from the total.
1901 */
1902 if (avail_space && avail_space >= skip_space)
1903 avail_space -= skip_space;
1904 else
1905 avail_space = 0;
1906
1907 if (avail_space < min_stripe_size)
1908 continue;
1909
1910 devices_info[i].dev = device;
1911 devices_info[i].max_avail = avail_space;
1912
1913 i++;
1914 }
7e33fd99
JB
1915 rcu_read_unlock();
1916 if (fs_info->alloc_start)
1917 mutex_unlock(&fs_devices->device_list_mutex);
6d07bcec
MX
1918
1919 nr_devices = i;
1920
1921 btrfs_descending_sort_devices(devices_info, nr_devices);
1922
1923 i = nr_devices - 1;
1924 avail_space = 0;
1925 while (nr_devices >= min_stripes) {
39fb26c3
MX
1926 if (num_stripes > nr_devices)
1927 num_stripes = nr_devices;
1928
6d07bcec
MX
1929 if (devices_info[i].max_avail >= min_stripe_size) {
1930 int j;
1931 u64 alloc_size;
1932
39fb26c3 1933 avail_space += devices_info[i].max_avail * num_stripes;
6d07bcec 1934 alloc_size = devices_info[i].max_avail;
39fb26c3 1935 for (j = i + 1 - num_stripes; j <= i; j++)
6d07bcec
MX
1936 devices_info[j].max_avail -= alloc_size;
1937 }
1938 i--;
1939 nr_devices--;
1940 }
1941
1942 kfree(devices_info);
1943 *free_bytes = avail_space;
1944 return 0;
1945}
1946
ba7b6e62
DS
1947/*
1948 * Calculate numbers for 'df', pessimistic in case of mixed raid profiles.
1949 *
1950 * If there's a redundant raid level at DATA block groups, use the respective
1951 * multiplier to scale the sizes.
1952 *
1953 * Unused device space usage is based on simulating the chunk allocator
1954 * algorithm that respects the device sizes, order of allocations and the
1955 * 'alloc_start' value, this is a close approximation of the actual use but
1956 * there are other factors that may change the result (like a new metadata
1957 * chunk).
1958 *
ca8a51b3
DS
1959 * If metadata is exhausted, f_bavail will be 0.
1960 *
ba7b6e62
DS
1961 * FIXME: not accurate for mixed block groups, total and free/used are ok,
1962 * available appears slightly larger.
1963 */
8fd17795
CM
1964static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
1965{
815745cf
AV
1966 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
1967 struct btrfs_super_block *disk_super = fs_info->super_copy;
1968 struct list_head *head = &fs_info->space_info;
bd4d1088
JB
1969 struct btrfs_space_info *found;
1970 u64 total_used = 0;
6d07bcec 1971 u64 total_free_data = 0;
ca8a51b3 1972 u64 total_free_meta = 0;
db94535d 1973 int bits = dentry->d_sb->s_blocksize_bits;
815745cf 1974 __be32 *fsid = (__be32 *)fs_info->fsid;
ba7b6e62
DS
1975 unsigned factor = 1;
1976 struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv;
6d07bcec 1977 int ret;
ca8a51b3 1978 u64 thresh = 0;
8fd17795 1979
15484377
MX
1980 /*
1981 * holding chunk_muext to avoid allocating new chunks, holding
1982 * device_list_mutex to avoid the device being removed
1983 */
bd4d1088 1984 rcu_read_lock();
89a55897 1985 list_for_each_entry_rcu(found, head, list) {
6d07bcec 1986 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
ba7b6e62
DS
1987 int i;
1988
6d07bcec
MX
1989 total_free_data += found->disk_total - found->disk_used;
1990 total_free_data -=
1991 btrfs_account_ro_block_groups_free_space(found);
ba7b6e62
DS
1992
1993 for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) {
1994 if (!list_empty(&found->block_groups[i])) {
1995 switch (i) {
1996 case BTRFS_RAID_DUP:
1997 case BTRFS_RAID_RAID1:
1998 case BTRFS_RAID_RAID10:
1999 factor = 2;
2000 }
2001 }
2002 }
6d07bcec 2003 }
ca8a51b3
DS
2004 if (found->flags & BTRFS_BLOCK_GROUP_METADATA)
2005 total_free_meta += found->disk_total - found->disk_used;
6d07bcec 2006
b742bb82 2007 total_used += found->disk_used;
89a55897 2008 }
ba7b6e62 2009
bd4d1088
JB
2010 rcu_read_unlock();
2011
ba7b6e62
DS
2012 buf->f_blocks = div_u64(btrfs_super_total_bytes(disk_super), factor);
2013 buf->f_blocks >>= bits;
2014 buf->f_bfree = buf->f_blocks - (div_u64(total_used, factor) >> bits);
2015
2016 /* Account global block reserve as used, it's in logical size already */
2017 spin_lock(&block_rsv->lock);
2018 buf->f_bfree -= block_rsv->size >> bits;
2019 spin_unlock(&block_rsv->lock);
2020
0d95c1be 2021 buf->f_bavail = div_u64(total_free_data, factor);
815745cf 2022 ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data);
7e33fd99 2023 if (ret)
6d07bcec 2024 return ret;
ba7b6e62 2025 buf->f_bavail += div_u64(total_free_data, factor);
6d07bcec 2026 buf->f_bavail = buf->f_bavail >> bits;
d397712b 2027
ca8a51b3
DS
2028 /*
2029 * We calculate the remaining metadata space minus global reserve. If
2030 * this is (supposedly) smaller than zero, there's no space. But this
2031 * does not hold in practice, the exhausted state happens where's still
2032 * some positive delta. So we apply some guesswork and compare the
2033 * delta to a 4M threshold. (Practically observed delta was ~2M.)
2034 *
2035 * We probably cannot calculate the exact threshold value because this
2036 * depends on the internal reservations requested by various
2037 * operations, so some operations that consume a few metadata will
2038 * succeed even if the Avail is zero. But this is better than the other
2039 * way around.
2040 */
2041 thresh = 4 * 1024 * 1024;
2042
2043 if (total_free_meta - thresh < block_rsv->size)
2044 buf->f_bavail = 0;
2045
ba7b6e62
DS
2046 buf->f_type = BTRFS_SUPER_MAGIC;
2047 buf->f_bsize = dentry->d_sb->s_blocksize;
2048 buf->f_namelen = BTRFS_NAME_LEN;
2049
9d03632e 2050 /* We treat it as constant endianness (it doesn't matter _which_)
d397712b 2051 because we want the fsid to come out the same whether mounted
9d03632e
DW
2052 on a big-endian or little-endian host */
2053 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
2054 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
32d48fa1 2055 /* Mask in the root object ID too, to disambiguate subvols */
2b0143b5
DH
2056 buf->f_fsid.val[0] ^= BTRFS_I(d_inode(dentry))->root->objectid >> 32;
2057 buf->f_fsid.val[1] ^= BTRFS_I(d_inode(dentry))->root->objectid;
32d48fa1 2058
8fd17795
CM
2059 return 0;
2060}
b5133862 2061
aea52e19
AV
2062static void btrfs_kill_super(struct super_block *sb)
2063{
815745cf 2064 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
aea52e19 2065 kill_anon_super(sb);
d22ca7de 2066 free_fs_info(fs_info);
aea52e19
AV
2067}
2068
2e635a27
CM
2069static struct file_system_type btrfs_fs_type = {
2070 .owner = THIS_MODULE,
2071 .name = "btrfs",
061dbc6b 2072 .mount = btrfs_mount,
aea52e19 2073 .kill_sb = btrfs_kill_super,
f667aef6 2074 .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA,
2e635a27 2075};
7f78e035 2076MODULE_ALIAS_FS("btrfs");
a9218f6b 2077
d8620958
TVB
2078static int btrfs_control_open(struct inode *inode, struct file *file)
2079{
2080 /*
2081 * The control file's private_data is used to hold the
2082 * transaction when it is started and is used to keep
2083 * track of whether a transaction is already in progress.
2084 */
2085 file->private_data = NULL;
2086 return 0;
2087}
2088
d352ac68
CM
2089/*
2090 * used by btrfsctl to scan devices when no FS is mounted
2091 */
8a4b83cc
CM
2092static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
2093 unsigned long arg)
2094{
2095 struct btrfs_ioctl_vol_args *vol;
2096 struct btrfs_fs_devices *fs_devices;
c071fcfd 2097 int ret = -ENOTTY;
8a4b83cc 2098
e441d54d
CM
2099 if (!capable(CAP_SYS_ADMIN))
2100 return -EPERM;
2101
dae7b665
LZ
2102 vol = memdup_user((void __user *)arg, sizeof(*vol));
2103 if (IS_ERR(vol))
2104 return PTR_ERR(vol);
c071fcfd 2105
8a4b83cc
CM
2106 switch (cmd) {
2107 case BTRFS_IOC_SCAN_DEV:
97288f2c 2108 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
8a4b83cc
CM
2109 &btrfs_fs_type, &fs_devices);
2110 break;
02db0844
JB
2111 case BTRFS_IOC_DEVICES_READY:
2112 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
2113 &btrfs_fs_type, &fs_devices);
2114 if (ret)
2115 break;
2116 ret = !(fs_devices->num_devices == fs_devices->total_devices);
2117 break;
8a4b83cc 2118 }
dae7b665 2119
8a4b83cc 2120 kfree(vol);
f819d837 2121 return ret;
8a4b83cc
CM
2122}
2123
0176260f 2124static int btrfs_freeze(struct super_block *sb)
ed0dab6b 2125{
354aa0fb
MX
2126 struct btrfs_trans_handle *trans;
2127 struct btrfs_root *root = btrfs_sb(sb)->tree_root;
2128
d4edf39b 2129 trans = btrfs_attach_transaction_barrier(root);
354aa0fb
MX
2130 if (IS_ERR(trans)) {
2131 /* no transaction, don't bother */
2132 if (PTR_ERR(trans) == -ENOENT)
2133 return 0;
2134 return PTR_ERR(trans);
2135 }
2136 return btrfs_commit_transaction(trans, root);
ed0dab6b
Y
2137}
2138
9c5085c1
JB
2139static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
2140{
2141 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
2142 struct btrfs_fs_devices *cur_devices;
2143 struct btrfs_device *dev, *first_dev = NULL;
2144 struct list_head *head;
2145 struct rcu_string *name;
2146
2147 mutex_lock(&fs_info->fs_devices->device_list_mutex);
2148 cur_devices = fs_info->fs_devices;
2149 while (cur_devices) {
2150 head = &cur_devices->devices;
2151 list_for_each_entry(dev, head, dev_list) {
aa9ddcd4
JB
2152 if (dev->missing)
2153 continue;
0aeb8a6e
AJ
2154 if (!dev->name)
2155 continue;
9c5085c1
JB
2156 if (!first_dev || dev->devid < first_dev->devid)
2157 first_dev = dev;
2158 }
2159 cur_devices = cur_devices->seed;
2160 }
2161
2162 if (first_dev) {
2163 rcu_read_lock();
2164 name = rcu_dereference(first_dev->name);
2165 seq_escape(m, name->str, " \t\n\\");
2166 rcu_read_unlock();
2167 } else {
2168 WARN_ON(1);
2169 }
2170 mutex_unlock(&fs_info->fs_devices->device_list_mutex);
2171 return 0;
2172}
2173
b87221de 2174static const struct super_operations btrfs_super_ops = {
76dda93c 2175 .drop_inode = btrfs_drop_inode,
bd555975 2176 .evict_inode = btrfs_evict_inode,
e20d96d6 2177 .put_super = btrfs_put_super,
d5719762 2178 .sync_fs = btrfs_sync_fs,
a9572a15 2179 .show_options = btrfs_show_options,
9c5085c1 2180 .show_devname = btrfs_show_devname,
4730a4bc 2181 .write_inode = btrfs_write_inode,
2c90e5d6
CM
2182 .alloc_inode = btrfs_alloc_inode,
2183 .destroy_inode = btrfs_destroy_inode,
8fd17795 2184 .statfs = btrfs_statfs,
c146afad 2185 .remount_fs = btrfs_remount,
0176260f 2186 .freeze_fs = btrfs_freeze,
e20d96d6 2187};
a9218f6b
CM
2188
2189static const struct file_operations btrfs_ctl_fops = {
d8620958 2190 .open = btrfs_control_open,
a9218f6b
CM
2191 .unlocked_ioctl = btrfs_control_ioctl,
2192 .compat_ioctl = btrfs_control_ioctl,
2193 .owner = THIS_MODULE,
6038f373 2194 .llseek = noop_llseek,
a9218f6b
CM
2195};
2196
2197static struct miscdevice btrfs_misc = {
578454ff 2198 .minor = BTRFS_MINOR,
a9218f6b
CM
2199 .name = "btrfs-control",
2200 .fops = &btrfs_ctl_fops
2201};
2202
578454ff
KS
2203MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
2204MODULE_ALIAS("devname:btrfs-control");
2205
a9218f6b
CM
2206static int btrfs_interface_init(void)
2207{
2208 return misc_register(&btrfs_misc);
2209}
2210
b2950863 2211static void btrfs_interface_exit(void)
a9218f6b 2212{
f368ed60 2213 misc_deregister(&btrfs_misc);
a9218f6b
CM
2214}
2215
85965600
DS
2216static void btrfs_print_info(void)
2217{
2218 printk(KERN_INFO "Btrfs loaded"
2219#ifdef CONFIG_BTRFS_DEBUG
2220 ", debug=on"
2221#endif
79556c3d
SB
2222#ifdef CONFIG_BTRFS_ASSERT
2223 ", assert=on"
2224#endif
85965600
DS
2225#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
2226 ", integrity-checker=on"
2227#endif
2228 "\n");
2229}
2230
dc11dd5d
JB
2231static int btrfs_run_sanity_tests(void)
2232{
06ea65a3
JB
2233 int ret;
2234
294e30fe 2235 ret = btrfs_init_test_fs();
06ea65a3
JB
2236 if (ret)
2237 return ret;
294e30fe
JB
2238
2239 ret = btrfs_test_free_space_cache();
2240 if (ret)
2241 goto out;
2242 ret = btrfs_test_extent_buffer_operations();
2243 if (ret)
2244 goto out;
2245 ret = btrfs_test_extent_io();
aaedb55b
JB
2246 if (ret)
2247 goto out;
2248 ret = btrfs_test_inodes();
faa2dbf0
JB
2249 if (ret)
2250 goto out;
2251 ret = btrfs_test_qgroups();
294e30fe
JB
2252out:
2253 btrfs_destroy_test_fs();
2254 return ret;
dc11dd5d
JB
2255}
2256
2e635a27
CM
2257static int __init init_btrfs_fs(void)
2258{
2c90e5d6 2259 int err;
58176a96 2260
14a958e6
FDBM
2261 err = btrfs_hash_init();
2262 if (err)
2263 return err;
2264
63541927
FDBM
2265 btrfs_props_init();
2266
58176a96
JB
2267 err = btrfs_init_sysfs();
2268 if (err)
14a958e6 2269 goto free_hash;
58176a96 2270
143bede5 2271 btrfs_init_compress();
d1310b2e 2272
261507a0
LZ
2273 err = btrfs_init_cachep();
2274 if (err)
2275 goto free_compress;
2276
d1310b2e 2277 err = extent_io_init();
2f4cbe64
WB
2278 if (err)
2279 goto free_cachep;
2280
d1310b2e
CM
2281 err = extent_map_init();
2282 if (err)
2283 goto free_extent_io;
2284
6352b91d 2285 err = ordered_data_init();
2f4cbe64
WB
2286 if (err)
2287 goto free_extent_map;
c8b97818 2288
6352b91d
MX
2289 err = btrfs_delayed_inode_init();
2290 if (err)
2291 goto free_ordered_data;
2292
9247f317 2293 err = btrfs_auto_defrag_init();
16cdcec7
MX
2294 if (err)
2295 goto free_delayed_inode;
2296
78a6184a 2297 err = btrfs_delayed_ref_init();
9247f317
MX
2298 if (err)
2299 goto free_auto_defrag;
2300
b9e9a6cb
WS
2301 err = btrfs_prelim_ref_init();
2302 if (err)
af13b492 2303 goto free_delayed_ref;
b9e9a6cb 2304
97eb6b69 2305 err = btrfs_end_io_wq_init();
78a6184a 2306 if (err)
af13b492 2307 goto free_prelim_ref;
78a6184a 2308
97eb6b69
DS
2309 err = btrfs_interface_init();
2310 if (err)
2311 goto free_end_io_wq;
2312
e565d4b9
JS
2313 btrfs_init_lockdep();
2314
85965600 2315 btrfs_print_info();
dc11dd5d
JB
2316
2317 err = btrfs_run_sanity_tests();
2318 if (err)
2319 goto unregister_ioctl;
2320
2321 err = register_filesystem(&btrfs_fs_type);
2322 if (err)
2323 goto unregister_ioctl;
74255aa0 2324
2f4cbe64
WB
2325 return 0;
2326
a9218f6b
CM
2327unregister_ioctl:
2328 btrfs_interface_exit();
97eb6b69
DS
2329free_end_io_wq:
2330 btrfs_end_io_wq_exit();
b9e9a6cb
WS
2331free_prelim_ref:
2332 btrfs_prelim_ref_exit();
78a6184a
MX
2333free_delayed_ref:
2334 btrfs_delayed_ref_exit();
9247f317
MX
2335free_auto_defrag:
2336 btrfs_auto_defrag_exit();
16cdcec7
MX
2337free_delayed_inode:
2338 btrfs_delayed_inode_exit();
6352b91d
MX
2339free_ordered_data:
2340 ordered_data_exit();
2f4cbe64
WB
2341free_extent_map:
2342 extent_map_exit();
d1310b2e
CM
2343free_extent_io:
2344 extent_io_exit();
2f4cbe64
WB
2345free_cachep:
2346 btrfs_destroy_cachep();
261507a0
LZ
2347free_compress:
2348 btrfs_exit_compress();
2f4cbe64 2349 btrfs_exit_sysfs();
14a958e6
FDBM
2350free_hash:
2351 btrfs_hash_exit();
2f4cbe64 2352 return err;
2e635a27
CM
2353}
2354
2355static void __exit exit_btrfs_fs(void)
2356{
39279cc3 2357 btrfs_destroy_cachep();
78a6184a 2358 btrfs_delayed_ref_exit();
9247f317 2359 btrfs_auto_defrag_exit();
16cdcec7 2360 btrfs_delayed_inode_exit();
b9e9a6cb 2361 btrfs_prelim_ref_exit();
6352b91d 2362 ordered_data_exit();
a52d9a80 2363 extent_map_exit();
d1310b2e 2364 extent_io_exit();
a9218f6b 2365 btrfs_interface_exit();
5ed5f588 2366 btrfs_end_io_wq_exit();
2e635a27 2367 unregister_filesystem(&btrfs_fs_type);
58176a96 2368 btrfs_exit_sysfs();
8a4b83cc 2369 btrfs_cleanup_fs_uuids();
261507a0 2370 btrfs_exit_compress();
14a958e6 2371 btrfs_hash_exit();
2e635a27
CM
2372}
2373
60efa5eb 2374late_initcall(init_btrfs_fs);
2e635a27
CM
2375module_exit(exit_btrfs_fs)
2376
2377MODULE_LICENSE("GPL");