]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - fs/btrfs/super.c
Btrfs: faster file extent item replace operations
[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"
5103e947 51#include "xattr.h"
8a4b83cc 52#include "volumes.h"
be6e8dc0 53#include "export.h"
c8b97818 54#include "compression.h"
9c5085c1 55#include "rcu-string.h"
8dabb742 56#include "dev-replace.h"
74255aa0 57#include "free-space-cache.h"
b9e9a6cb 58#include "backref.h"
dc11dd5d 59#include "tests/btrfs-tests.h"
2e635a27 60
1abe9b8a 61#define CREATE_TRACE_POINTS
62#include <trace/events/btrfs.h>
63
b87221de 64static const struct super_operations btrfs_super_ops;
830c4adb 65static struct file_system_type btrfs_fs_type;
75dfe396 66
08748810 67static const char *btrfs_decode_error(int errno)
acce952b 68{
08748810 69 char *errstr = "unknown";
acce952b 70
71 switch (errno) {
72 case -EIO:
73 errstr = "IO failure";
74 break;
75 case -ENOMEM:
76 errstr = "Out of memory";
77 break;
78 case -EROFS:
79 errstr = "Readonly filesystem";
80 break;
8c342930
JM
81 case -EEXIST:
82 errstr = "Object already exists";
83 break;
94ef7280
DS
84 case -ENOSPC:
85 errstr = "No space left";
86 break;
87 case -ENOENT:
88 errstr = "No such entry";
89 break;
acce952b 90 }
91
92 return errstr;
93}
94
bbece8a3 95static void save_error_info(struct btrfs_fs_info *fs_info)
acce952b 96{
97 /*
98 * today we only save the error info into ram. Long term we'll
99 * also send it down to the disk
100 */
87533c47 101 set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state);
acce952b 102}
103
acce952b 104/* btrfs handle error by forcing the filesystem readonly */
105static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
106{
107 struct super_block *sb = fs_info->sb;
108
109 if (sb->s_flags & MS_RDONLY)
110 return;
111
87533c47 112 if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
acce952b 113 sb->s_flags |= MS_RDONLY;
c2cf52eb 114 btrfs_info(fs_info, "forced readonly");
1acd6831
SB
115 /*
116 * Note that a running device replace operation is not
117 * canceled here although there is no way to update
118 * the progress. It would add the risk of a deadlock,
119 * therefore the canceling is ommited. The only penalty
120 * is that some I/O remains active until the procedure
121 * completes. The next time when the filesystem is
122 * mounted writeable again, the device replace
123 * operation continues.
124 */
acce952b 125 }
126}
127
533574c6 128#ifdef CONFIG_PRINTK
acce952b 129/*
130 * __btrfs_std_error decodes expected errors from the caller and
131 * invokes the approciate error response.
132 */
133void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
4da35113 134 unsigned int line, int errno, const char *fmt, ...)
acce952b 135{
136 struct super_block *sb = fs_info->sb;
acce952b 137 const char *errstr;
138
139 /*
140 * Special case: if the error is EROFS, and we're already
141 * under MS_RDONLY, then it is safe here.
142 */
143 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
4da35113
JM
144 return;
145
08748810 146 errstr = btrfs_decode_error(errno);
4da35113 147 if (fmt) {
37252a66
ES
148 struct va_format vaf;
149 va_list args;
150
151 va_start(args, fmt);
152 vaf.fmt = fmt;
153 vaf.va = &args;
4da35113 154
efe120a0
FH
155 printk(KERN_CRIT
156 "BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n",
08748810 157 sb->s_id, function, line, errno, errstr, &vaf);
37252a66 158 va_end(args);
4da35113 159 } else {
efe120a0 160 printk(KERN_CRIT "BTRFS: error (device %s) in %s:%d: errno=%d %s\n",
08748810 161 sb->s_id, function, line, errno, errstr);
4da35113 162 }
acce952b 163
4da35113 164 /* Don't go through full error handling during mount */
cf79ffb5
JB
165 save_error_info(fs_info);
166 if (sb->s_flags & MS_BORN)
4da35113 167 btrfs_handle_error(fs_info);
4da35113 168}
acce952b 169
533574c6 170static const char * const logtypes[] = {
4da35113
JM
171 "emergency",
172 "alert",
173 "critical",
174 "error",
175 "warning",
176 "notice",
177 "info",
178 "debug",
179};
180
c2cf52eb 181void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
4da35113
JM
182{
183 struct super_block *sb = fs_info->sb;
184 char lvl[4];
185 struct va_format vaf;
186 va_list args;
187 const char *type = logtypes[4];
533574c6 188 int kern_level;
4da35113
JM
189
190 va_start(args, fmt);
191
533574c6
JP
192 kern_level = printk_get_level(fmt);
193 if (kern_level) {
194 size_t size = printk_skip_level(fmt) - fmt;
195 memcpy(lvl, fmt, size);
196 lvl[size] = '\0';
197 fmt += size;
198 type = logtypes[kern_level - '0'];
4da35113
JM
199 } else
200 *lvl = '\0';
201
202 vaf.fmt = fmt;
203 vaf.va = &args;
533574c6 204
c2cf52eb 205 printk("%sBTRFS %s (device %s): %pV\n", lvl, type, sb->s_id, &vaf);
533574c6
JP
206
207 va_end(args);
208}
209
210#else
211
212void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
213 unsigned int line, int errno, const char *fmt, ...)
214{
215 struct super_block *sb = fs_info->sb;
216
217 /*
218 * Special case: if the error is EROFS, and we're already
219 * under MS_RDONLY, then it is safe here.
220 */
221 if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
222 return;
223
224 /* Don't go through full error handling during mount */
225 if (sb->s_flags & MS_BORN) {
226 save_error_info(fs_info);
227 btrfs_handle_error(fs_info);
228 }
acce952b 229}
533574c6 230#endif
acce952b 231
49b25e05
JM
232/*
233 * We only mark the transaction aborted and then set the file system read-only.
234 * This will prevent new transactions from starting or trying to join this
235 * one.
236 *
237 * This means that error recovery at the call site is limited to freeing
238 * any local memory allocations and passing the error code up without
239 * further cleanup. The transaction should complete as it normally would
240 * in the call path but will return -EIO.
241 *
242 * We'll complete the cleanup in btrfs_end_transaction and
243 * btrfs_commit_transaction.
244 */
245void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
246 struct btrfs_root *root, const char *function,
247 unsigned int line, int errno)
248{
08748810
DS
249 /*
250 * Report first abort since mount
251 */
252 if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED,
253 &root->fs_info->fs_state)) {
efe120a0 254 WARN(1, KERN_DEBUG "BTRFS: Transaction aborted (error %d)\n",
08748810
DS
255 errno);
256 }
49b25e05
JM
257 trans->aborted = errno;
258 /* Nothing used. The other threads that have joined this
259 * transaction may be able to continue. */
260 if (!trans->blocks_used) {
69ce977a
MX
261 const char *errstr;
262
08748810 263 errstr = btrfs_decode_error(errno);
c2cf52eb
SK
264 btrfs_warn(root->fs_info,
265 "%s:%d: Aborting unused transaction(%s).",
266 function, line, errstr);
acce952b 267 return;
49b25e05 268 }
8d25a086 269 ACCESS_ONCE(trans->transaction->aborted) = errno;
501407aa
JB
270 /* Wake up anybody who may be waiting on this transaction */
271 wake_up(&root->fs_info->transaction_wait);
272 wake_up(&root->fs_info->transaction_blocked_wait);
49b25e05
JM
273 __btrfs_std_error(root->fs_info, function, line, errno, NULL);
274}
8c342930
JM
275/*
276 * __btrfs_panic decodes unexpected, fatal errors from the caller,
277 * issues an alert, and either panics or BUGs, depending on mount options.
278 */
279void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
280 unsigned int line, int errno, const char *fmt, ...)
281{
8c342930
JM
282 char *s_id = "<unknown>";
283 const char *errstr;
284 struct va_format vaf = { .fmt = fmt };
285 va_list args;
acce952b 286
8c342930
JM
287 if (fs_info)
288 s_id = fs_info->sb->s_id;
acce952b 289
8c342930
JM
290 va_start(args, fmt);
291 vaf.va = &args;
292
08748810 293 errstr = btrfs_decode_error(errno);
aa43a17c 294 if (fs_info && (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR))
08748810
DS
295 panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n",
296 s_id, function, line, &vaf, errno, errstr);
8c342930 297
efe120a0
FH
298 btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)",
299 function, line, &vaf, errno, errstr);
8c342930
JM
300 va_end(args);
301 /* Caller calls BUG() */
acce952b 302}
303
d397712b 304static void btrfs_put_super(struct super_block *sb)
b18c6685 305{
815745cf 306 (void)close_ctree(btrfs_sb(sb)->tree_root);
aea52e19
AV
307 /* FIXME: need to fix VFS to return error? */
308 /* AV: return it _where_? ->put_super() can be triggered by any number
309 * of async events, up to and including delivery of SIGKILL to the
310 * last process that kept it busy. Or segfault in the aforementioned
311 * process... Whom would you report that to?
312 */
75dfe396
CM
313}
314
95e05289 315enum {
73f73415 316 Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum,
287a0ab9
JB
317 Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd,
318 Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress,
261507a0
LZ
319 Opt_compress_type, Opt_compress_force, Opt_compress_force_type,
320 Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_discard,
91435650 321 Opt_space_cache, Opt_clear_cache, Opt_user_subvol_rm_allowed,
9555c6c1
ID
322 Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache,
323 Opt_no_space_cache, Opt_recovery, Opt_skip_balance,
21adbd5c 324 Opt_check_integrity, Opt_check_integrity_including_extent_data,
f420ee1e 325 Opt_check_integrity_print_mask, Opt_fatal_errors, Opt_rescan_uuid_tree,
e07a2ade 326 Opt_commit_interval, Opt_barrier, Opt_nodefrag, Opt_nodiscard,
a258af7a 327 Opt_noenospc_debug, Opt_noflushoncommit, Opt_acl, Opt_datacow,
a88998f2 328 Opt_datasum, Opt_treelog,
9555c6c1 329 Opt_err,
95e05289
CM
330};
331
332static match_table_t tokens = {
dfe25020 333 {Opt_degraded, "degraded"},
95e05289 334 {Opt_subvol, "subvol=%s"},
1493381f 335 {Opt_subvolid, "subvolid=%s"},
43e570b0 336 {Opt_device, "device=%s"},
b6cda9bc 337 {Opt_nodatasum, "nodatasum"},
d399167d 338 {Opt_datasum, "datasum"},
be20aa9d 339 {Opt_nodatacow, "nodatacow"},
a258af7a 340 {Opt_datacow, "datacow"},
21ad10cf 341 {Opt_nobarrier, "nobarrier"},
842bef58 342 {Opt_barrier, "barrier"},
6f568d35 343 {Opt_max_inline, "max_inline=%s"},
8f662a76 344 {Opt_alloc_start, "alloc_start=%s"},
4543df7e 345 {Opt_thread_pool, "thread_pool=%d"},
c8b97818 346 {Opt_compress, "compress"},
261507a0 347 {Opt_compress_type, "compress=%s"},
a555f810 348 {Opt_compress_force, "compress-force"},
261507a0 349 {Opt_compress_force_type, "compress-force=%s"},
e18e4809 350 {Opt_ssd, "ssd"},
451d7585 351 {Opt_ssd_spread, "ssd_spread"},
3b30c22f 352 {Opt_nossd, "nossd"},
bd0330ad 353 {Opt_acl, "acl"},
33268eaf 354 {Opt_noacl, "noacl"},
3a5e1404 355 {Opt_notreelog, "notreelog"},
a88998f2 356 {Opt_treelog, "treelog"},
dccae999 357 {Opt_flushoncommit, "flushoncommit"},
2c9ee856 358 {Opt_noflushoncommit, "noflushoncommit"},
97e728d4 359 {Opt_ratio, "metadata_ratio=%d"},
e244a0ae 360 {Opt_discard, "discard"},
e07a2ade 361 {Opt_nodiscard, "nodiscard"},
0af3d00b 362 {Opt_space_cache, "space_cache"},
88c2ba3b 363 {Opt_clear_cache, "clear_cache"},
4260f7c7 364 {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"},
91435650 365 {Opt_enospc_debug, "enospc_debug"},
53036293 366 {Opt_noenospc_debug, "noenospc_debug"},
e15d0542 367 {Opt_subvolrootid, "subvolrootid=%d"},
4cb5300b 368 {Opt_defrag, "autodefrag"},
fc0ca9af 369 {Opt_nodefrag, "noautodefrag"},
4b9465cb 370 {Opt_inode_cache, "inode_cache"},
8965593e 371 {Opt_no_space_cache, "nospace_cache"},
af31f5e5 372 {Opt_recovery, "recovery"},
9555c6c1 373 {Opt_skip_balance, "skip_balance"},
21adbd5c
SB
374 {Opt_check_integrity, "check_int"},
375 {Opt_check_integrity_including_extent_data, "check_int_data"},
376 {Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
f420ee1e 377 {Opt_rescan_uuid_tree, "rescan_uuid_tree"},
8c342930 378 {Opt_fatal_errors, "fatal_errors=%s"},
8b87dc17 379 {Opt_commit_interval, "commit=%d"},
33268eaf 380 {Opt_err, NULL},
95e05289
CM
381};
382
edf24abe
CH
383/*
384 * Regular mount options parser. Everything that is needed only when
385 * reading in a new superblock is parsed here.
49b25e05 386 * XXX JDM: This needs to be cleaned up for remount.
edf24abe
CH
387 */
388int btrfs_parse_options(struct btrfs_root *root, char *options)
95e05289 389{
edf24abe 390 struct btrfs_fs_info *info = root->fs_info;
95e05289 391 substring_t args[MAX_OPT_ARGS];
73bc1876
JB
392 char *p, *num, *orig = NULL;
393 u64 cache_gen;
4543df7e 394 int intarg;
a7a3f7ca 395 int ret = 0;
261507a0
LZ
396 char *compress_type;
397 bool compress_force = false;
b6cda9bc 398
6c41761f 399 cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
73bc1876
JB
400 if (cache_gen)
401 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
402
95e05289 403 if (!options)
73bc1876 404 goto out;
95e05289 405
be20aa9d
CM
406 /*
407 * strsep changes the string, duplicate it because parse_options
408 * gets called twice
409 */
410 options = kstrdup(options, GFP_NOFS);
411 if (!options)
412 return -ENOMEM;
413
da495ecc 414 orig = options;
be20aa9d 415
edf24abe 416 while ((p = strsep(&options, ",")) != NULL) {
95e05289
CM
417 int token;
418 if (!*p)
419 continue;
420
421 token = match_token(p, tokens, args);
422 switch (token) {
dfe25020 423 case Opt_degraded:
efe120a0 424 btrfs_info(root->fs_info, "allowing degraded mounts");
edf24abe 425 btrfs_set_opt(info->mount_opt, DEGRADED);
dfe25020 426 break;
95e05289 427 case Opt_subvol:
73f73415 428 case Opt_subvolid:
e15d0542 429 case Opt_subvolrootid:
43e570b0 430 case Opt_device:
edf24abe 431 /*
43e570b0 432 * These are parsed by btrfs_parse_early_options
edf24abe
CH
433 * and can be happily ignored here.
434 */
b6cda9bc
CM
435 break;
436 case Opt_nodatasum:
efe120a0 437 btrfs_info(root->fs_info, "setting nodatasum");
edf24abe 438 btrfs_set_opt(info->mount_opt, NODATASUM);
be20aa9d 439 break;
d399167d
QW
440 case Opt_datasum:
441 if (btrfs_test_opt(root, NODATACOW))
442 btrfs_info(root->fs_info, "setting datasum, datacow enabled");
443 else
444 btrfs_info(root->fs_info, "setting datasum");
445 btrfs_clear_opt(info->mount_opt, NODATACOW);
446 btrfs_clear_opt(info->mount_opt, NODATASUM);
447 break;
be20aa9d 448 case Opt_nodatacow:
bedb2cca
AP
449 if (!btrfs_test_opt(root, COMPRESS) ||
450 !btrfs_test_opt(root, FORCE_COMPRESS)) {
efe120a0
FH
451 btrfs_info(root->fs_info,
452 "setting nodatacow, compression disabled");
bedb2cca 453 } else {
efe120a0 454 btrfs_info(root->fs_info, "setting nodatacow");
bedb2cca 455 }
bedb2cca
AP
456 btrfs_clear_opt(info->mount_opt, COMPRESS);
457 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
edf24abe
CH
458 btrfs_set_opt(info->mount_opt, NODATACOW);
459 btrfs_set_opt(info->mount_opt, NODATASUM);
95e05289 460 break;
a258af7a
QW
461 case Opt_datacow:
462 if (btrfs_test_opt(root, NODATACOW))
463 btrfs_info(root->fs_info, "setting datacow");
464 btrfs_clear_opt(info->mount_opt, NODATACOW);
465 break;
a555f810 466 case Opt_compress_force:
261507a0
LZ
467 case Opt_compress_force_type:
468 compress_force = true;
1c697d4a 469 /* Fallthrough */
261507a0
LZ
470 case Opt_compress:
471 case Opt_compress_type:
472 if (token == Opt_compress ||
473 token == Opt_compress_force ||
474 strcmp(args[0].from, "zlib") == 0) {
475 compress_type = "zlib";
476 info->compress_type = BTRFS_COMPRESS_ZLIB;
063849ea 477 btrfs_set_opt(info->mount_opt, COMPRESS);
bedb2cca
AP
478 btrfs_clear_opt(info->mount_opt, NODATACOW);
479 btrfs_clear_opt(info->mount_opt, NODATASUM);
a6fa6fae
LZ
480 } else if (strcmp(args[0].from, "lzo") == 0) {
481 compress_type = "lzo";
482 info->compress_type = BTRFS_COMPRESS_LZO;
063849ea 483 btrfs_set_opt(info->mount_opt, COMPRESS);
bedb2cca
AP
484 btrfs_clear_opt(info->mount_opt, NODATACOW);
485 btrfs_clear_opt(info->mount_opt, NODATASUM);
2b0ce2c2 486 btrfs_set_fs_incompat(info, COMPRESS_LZO);
063849ea
AH
487 } else if (strncmp(args[0].from, "no", 2) == 0) {
488 compress_type = "no";
063849ea
AH
489 btrfs_clear_opt(info->mount_opt, COMPRESS);
490 btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS);
491 compress_force = false;
261507a0
LZ
492 } else {
493 ret = -EINVAL;
494 goto out;
495 }
496
261507a0
LZ
497 if (compress_force) {
498 btrfs_set_opt(info->mount_opt, FORCE_COMPRESS);
efe120a0 499 btrfs_info(root->fs_info, "force %s compression",
261507a0 500 compress_type);
a7e252af 501 } else if (btrfs_test_opt(root, COMPRESS)) {
261507a0
LZ
502 pr_info("btrfs: use %s compression\n",
503 compress_type);
a7e252af 504 }
a555f810 505 break;
e18e4809 506 case Opt_ssd:
efe120a0 507 btrfs_info(root->fs_info, "use ssd allocation scheme");
edf24abe 508 btrfs_set_opt(info->mount_opt, SSD);
e18e4809 509 break;
451d7585 510 case Opt_ssd_spread:
efe120a0 511 btrfs_info(root->fs_info, "use spread ssd allocation scheme");
451d7585
CM
512 btrfs_set_opt(info->mount_opt, SSD);
513 btrfs_set_opt(info->mount_opt, SSD_SPREAD);
514 break;
3b30c22f 515 case Opt_nossd:
efe120a0 516 btrfs_info(root->fs_info, "not using ssd allocation scheme");
c289811c 517 btrfs_set_opt(info->mount_opt, NOSSD);
3b30c22f 518 btrfs_clear_opt(info->mount_opt, SSD);
451d7585 519 btrfs_clear_opt(info->mount_opt, SSD_SPREAD);
3b30c22f 520 break;
842bef58
QW
521 case Opt_barrier:
522 if (btrfs_test_opt(root, NOBARRIER))
523 btrfs_info(root->fs_info, "turning on barriers");
524 btrfs_clear_opt(info->mount_opt, NOBARRIER);
525 break;
21ad10cf 526 case Opt_nobarrier:
efe120a0 527 btrfs_info(root->fs_info, "turning off barriers");
edf24abe 528 btrfs_set_opt(info->mount_opt, NOBARRIER);
21ad10cf 529 break;
4543df7e 530 case Opt_thread_pool:
2c334e87
WS
531 ret = match_int(&args[0], &intarg);
532 if (ret) {
533 goto out;
534 } else if (intarg > 0) {
4543df7e 535 info->thread_pool_size = intarg;
2c334e87
WS
536 } else {
537 ret = -EINVAL;
538 goto out;
539 }
4543df7e 540 break;
6f568d35 541 case Opt_max_inline:
edf24abe
CH
542 num = match_strdup(&args[0]);
543 if (num) {
91748467 544 info->max_inline = memparse(num, NULL);
edf24abe
CH
545 kfree(num);
546
15ada040
CM
547 if (info->max_inline) {
548 info->max_inline = max_t(u64,
549 info->max_inline,
550 root->sectorsize);
551 }
efe120a0 552 btrfs_info(root->fs_info, "max_inline at %llu",
c1c9ff7c 553 info->max_inline);
2c334e87
WS
554 } else {
555 ret = -ENOMEM;
556 goto out;
6f568d35
CM
557 }
558 break;
8f662a76 559 case Opt_alloc_start:
edf24abe
CH
560 num = match_strdup(&args[0]);
561 if (num) {
c018daec 562 mutex_lock(&info->chunk_mutex);
91748467 563 info->alloc_start = memparse(num, NULL);
c018daec 564 mutex_unlock(&info->chunk_mutex);
edf24abe 565 kfree(num);
efe120a0 566 btrfs_info(root->fs_info, "allocations start at %llu",
c1c9ff7c 567 info->alloc_start);
2c334e87
WS
568 } else {
569 ret = -ENOMEM;
570 goto out;
8f662a76
CM
571 }
572 break;
bd0330ad
QW
573 case Opt_acl:
574 root->fs_info->sb->s_flags |= MS_POSIXACL;
575 break;
33268eaf
JB
576 case Opt_noacl:
577 root->fs_info->sb->s_flags &= ~MS_POSIXACL;
578 break;
3a5e1404 579 case Opt_notreelog:
efe120a0 580 btrfs_info(root->fs_info, "disabling tree log");
3a5e1404 581 btrfs_set_opt(info->mount_opt, NOTREELOG);
a88998f2
QW
582 break;
583 case Opt_treelog:
584 if (btrfs_test_opt(root, NOTREELOG))
585 btrfs_info(root->fs_info, "enabling tree log");
586 btrfs_clear_opt(info->mount_opt, NOTREELOG);
3a5e1404 587 break;
dccae999 588 case Opt_flushoncommit:
efe120a0 589 btrfs_info(root->fs_info, "turning on flush-on-commit");
dccae999
SW
590 btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT);
591 break;
2c9ee856
QW
592 case Opt_noflushoncommit:
593 if (btrfs_test_opt(root, FLUSHONCOMMIT))
594 btrfs_info(root->fs_info, "turning off flush-on-commit");
595 btrfs_clear_opt(info->mount_opt, FLUSHONCOMMIT);
596 break;
97e728d4 597 case Opt_ratio:
2c334e87
WS
598 ret = match_int(&args[0], &intarg);
599 if (ret) {
600 goto out;
601 } else if (intarg >= 0) {
97e728d4 602 info->metadata_ratio = intarg;
efe120a0 603 btrfs_info(root->fs_info, "metadata ratio %d",
97e728d4 604 info->metadata_ratio);
2c334e87
WS
605 } else {
606 ret = -EINVAL;
607 goto out;
97e728d4
JB
608 }
609 break;
e244a0ae
CH
610 case Opt_discard:
611 btrfs_set_opt(info->mount_opt, DISCARD);
612 break;
e07a2ade
QW
613 case Opt_nodiscard:
614 btrfs_clear_opt(info->mount_opt, DISCARD);
615 break;
0af3d00b 616 case Opt_space_cache:
0af3d00b 617 btrfs_set_opt(info->mount_opt, SPACE_CACHE);
0de90876 618 break;
f420ee1e
SB
619 case Opt_rescan_uuid_tree:
620 btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE);
621 break;
73bc1876 622 case Opt_no_space_cache:
efe120a0 623 btrfs_info(root->fs_info, "disabling disk space caching");
73bc1876
JB
624 btrfs_clear_opt(info->mount_opt, SPACE_CACHE);
625 break;
4b9465cb 626 case Opt_inode_cache:
efe120a0 627 btrfs_info(root->fs_info, "enabling inode map caching");
4b9465cb
CM
628 btrfs_set_opt(info->mount_opt, INODE_MAP_CACHE);
629 break;
88c2ba3b 630 case Opt_clear_cache:
efe120a0 631 btrfs_info(root->fs_info, "force clearing of disk cache");
88c2ba3b 632 btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
0af3d00b 633 break;
4260f7c7
SW
634 case Opt_user_subvol_rm_allowed:
635 btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED);
636 break;
91435650
CM
637 case Opt_enospc_debug:
638 btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG);
639 break;
53036293
QW
640 case Opt_noenospc_debug:
641 btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG);
642 break;
4cb5300b 643 case Opt_defrag:
efe120a0 644 btrfs_info(root->fs_info, "enabling auto defrag");
4cb5300b
CM
645 btrfs_set_opt(info->mount_opt, AUTO_DEFRAG);
646 break;
fc0ca9af
QW
647 case Opt_nodefrag:
648 if (btrfs_test_opt(root, AUTO_DEFRAG))
649 btrfs_info(root->fs_info, "disabling auto defrag");
650 btrfs_clear_opt(info->mount_opt, AUTO_DEFRAG);
651 break;
af31f5e5 652 case Opt_recovery:
efe120a0 653 btrfs_info(root->fs_info, "enabling auto recovery");
af31f5e5
CM
654 btrfs_set_opt(info->mount_opt, RECOVERY);
655 break;
9555c6c1
ID
656 case Opt_skip_balance:
657 btrfs_set_opt(info->mount_opt, SKIP_BALANCE);
658 break;
21adbd5c
SB
659#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
660 case Opt_check_integrity_including_extent_data:
efe120a0
FH
661 btrfs_info(root->fs_info,
662 "enabling check integrity including extent data");
21adbd5c
SB
663 btrfs_set_opt(info->mount_opt,
664 CHECK_INTEGRITY_INCLUDING_EXTENT_DATA);
665 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
666 break;
667 case Opt_check_integrity:
efe120a0 668 btrfs_info(root->fs_info, "enabling check integrity");
21adbd5c
SB
669 btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY);
670 break;
671 case Opt_check_integrity_print_mask:
2c334e87
WS
672 ret = match_int(&args[0], &intarg);
673 if (ret) {
674 goto out;
675 } else if (intarg >= 0) {
21adbd5c 676 info->check_integrity_print_mask = intarg;
efe120a0 677 btrfs_info(root->fs_info, "check_integrity_print_mask 0x%x",
21adbd5c 678 info->check_integrity_print_mask);
2c334e87
WS
679 } else {
680 ret = -EINVAL;
681 goto out;
21adbd5c
SB
682 }
683 break;
684#else
685 case Opt_check_integrity_including_extent_data:
686 case Opt_check_integrity:
687 case Opt_check_integrity_print_mask:
efe120a0
FH
688 btrfs_err(root->fs_info,
689 "support for check_integrity* not compiled in!");
21adbd5c
SB
690 ret = -EINVAL;
691 goto out;
692#endif
8c342930
JM
693 case Opt_fatal_errors:
694 if (strcmp(args[0].from, "panic") == 0)
695 btrfs_set_opt(info->mount_opt,
696 PANIC_ON_FATAL_ERROR);
697 else if (strcmp(args[0].from, "bug") == 0)
698 btrfs_clear_opt(info->mount_opt,
699 PANIC_ON_FATAL_ERROR);
700 else {
701 ret = -EINVAL;
702 goto out;
703 }
704 break;
8b87dc17
DS
705 case Opt_commit_interval:
706 intarg = 0;
707 ret = match_int(&args[0], &intarg);
708 if (ret < 0) {
efe120a0 709 btrfs_err(root->fs_info, "invalid commit interval");
8b87dc17
DS
710 ret = -EINVAL;
711 goto out;
712 }
713 if (intarg > 0) {
714 if (intarg > 300) {
efe120a0 715 btrfs_warn(root->fs_info, "excessive commit interval %d",
8b87dc17
DS
716 intarg);
717 }
718 info->commit_interval = intarg;
719 } else {
efe120a0 720 btrfs_info(root->fs_info, "using default commit interval %ds",
8b87dc17
DS
721 BTRFS_DEFAULT_COMMIT_INTERVAL);
722 info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
723 }
724 break;
a7a3f7ca 725 case Opt_err:
efe120a0 726 btrfs_info(root->fs_info, "unrecognized mount option '%s'", p);
a7a3f7ca
SW
727 ret = -EINVAL;
728 goto out;
95e05289 729 default:
be20aa9d 730 break;
95e05289
CM
731 }
732 }
a7a3f7ca 733out:
73bc1876 734 if (!ret && btrfs_test_opt(root, SPACE_CACHE))
efe120a0 735 btrfs_info(root->fs_info, "disk space caching is enabled");
da495ecc 736 kfree(orig);
a7a3f7ca 737 return ret;
edf24abe
CH
738}
739
740/*
741 * Parse mount options that are required early in the mount process.
742 *
743 * All other options will be parsed on much later in the mount process and
744 * only when we need to allocate a new super block.
745 */
97288f2c 746static int btrfs_parse_early_options(const char *options, fmode_t flags,
73f73415 747 void *holder, char **subvol_name, u64 *subvol_objectid,
5e2a4b25 748 struct btrfs_fs_devices **fs_devices)
edf24abe
CH
749{
750 substring_t args[MAX_OPT_ARGS];
83c8c9bd 751 char *device_name, *opts, *orig, *p;
1493381f 752 char *num = NULL;
edf24abe
CH
753 int error = 0;
754
755 if (!options)
830c4adb 756 return 0;
edf24abe
CH
757
758 /*
759 * strsep changes the string, duplicate it because parse_options
760 * gets called twice
761 */
762 opts = kstrdup(options, GFP_KERNEL);
763 if (!opts)
764 return -ENOMEM;
3f3d0bc0 765 orig = opts;
edf24abe
CH
766
767 while ((p = strsep(&opts, ",")) != NULL) {
768 int token;
769 if (!*p)
770 continue;
771
772 token = match_token(p, tokens, args);
773 switch (token) {
774 case Opt_subvol:
a90e8b6f 775 kfree(*subvol_name);
edf24abe 776 *subvol_name = match_strdup(&args[0]);
2c334e87
WS
777 if (!*subvol_name) {
778 error = -ENOMEM;
779 goto out;
780 }
edf24abe 781 break;
73f73415 782 case Opt_subvolid:
1493381f
WS
783 num = match_strdup(&args[0]);
784 if (num) {
785 *subvol_objectid = memparse(num, NULL);
786 kfree(num);
4849f01d 787 /* we want the original fs_tree */
1493381f 788 if (!*subvol_objectid)
4849f01d
JB
789 *subvol_objectid =
790 BTRFS_FS_TREE_OBJECTID;
2c334e87
WS
791 } else {
792 error = -EINVAL;
793 goto out;
4849f01d 794 }
73f73415 795 break;
e15d0542 796 case Opt_subvolrootid:
5e2a4b25 797 printk(KERN_WARNING
efe120a0
FH
798 "BTRFS: 'subvolrootid' mount option is deprecated and has "
799 "no effect\n");
e15d0542 800 break;
43e570b0 801 case Opt_device:
83c8c9bd
JL
802 device_name = match_strdup(&args[0]);
803 if (!device_name) {
804 error = -ENOMEM;
805 goto out;
806 }
807 error = btrfs_scan_one_device(device_name,
43e570b0 808 flags, holder, fs_devices);
83c8c9bd 809 kfree(device_name);
43e570b0 810 if (error)
830c4adb 811 goto out;
43e570b0 812 break;
edf24abe
CH
813 default:
814 break;
815 }
816 }
817
830c4adb 818out:
3f3d0bc0 819 kfree(orig);
edf24abe 820 return error;
95e05289
CM
821}
822
73f73415
JB
823static struct dentry *get_default_root(struct super_block *sb,
824 u64 subvol_objectid)
825{
815745cf
AV
826 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
827 struct btrfs_root *root = fs_info->tree_root;
73f73415
JB
828 struct btrfs_root *new_root;
829 struct btrfs_dir_item *di;
830 struct btrfs_path *path;
831 struct btrfs_key location;
832 struct inode *inode;
73f73415
JB
833 u64 dir_id;
834 int new = 0;
835
836 /*
837 * We have a specific subvol we want to mount, just setup location and
838 * go look up the root.
839 */
840 if (subvol_objectid) {
841 location.objectid = subvol_objectid;
842 location.type = BTRFS_ROOT_ITEM_KEY;
843 location.offset = (u64)-1;
844 goto find_root;
845 }
846
847 path = btrfs_alloc_path();
848 if (!path)
849 return ERR_PTR(-ENOMEM);
850 path->leave_spinning = 1;
851
852 /*
853 * Find the "default" dir item which points to the root item that we
854 * will mount by default if we haven't been given a specific subvolume
855 * to mount.
856 */
815745cf 857 dir_id = btrfs_super_root_dir(fs_info->super_copy);
73f73415 858 di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
b0839166
JL
859 if (IS_ERR(di)) {
860 btrfs_free_path(path);
fb4f6f91 861 return ERR_CAST(di);
b0839166 862 }
73f73415
JB
863 if (!di) {
864 /*
865 * Ok the default dir item isn't there. This is weird since
866 * it's always been there, but don't freak out, just try and
867 * mount to root most subvolume.
868 */
869 btrfs_free_path(path);
870 dir_id = BTRFS_FIRST_FREE_OBJECTID;
815745cf 871 new_root = fs_info->fs_root;
73f73415
JB
872 goto setup_root;
873 }
874
875 btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
876 btrfs_free_path(path);
877
878find_root:
815745cf 879 new_root = btrfs_read_fs_root_no_name(fs_info, &location);
73f73415 880 if (IS_ERR(new_root))
d0b678cb 881 return ERR_CAST(new_root);
73f73415 882
73f73415
JB
883 dir_id = btrfs_root_dirid(&new_root->root_item);
884setup_root:
885 location.objectid = dir_id;
886 location.type = BTRFS_INODE_ITEM_KEY;
887 location.offset = 0;
888
889 inode = btrfs_iget(sb, &location, new_root, &new);
4cbd1149
DC
890 if (IS_ERR(inode))
891 return ERR_CAST(inode);
73f73415
JB
892
893 /*
894 * If we're just mounting the root most subvol put the inode and return
895 * a reference to the dentry. We will have already gotten a reference
896 * to the inode in btrfs_fill_super so we're good to go.
897 */
898 if (!new && sb->s_root->d_inode == inode) {
899 iput(inode);
900 return dget(sb->s_root);
901 }
902
ba5b8958 903 return d_obtain_alias(inode);
73f73415
JB
904}
905
d397712b 906static int btrfs_fill_super(struct super_block *sb,
8a4b83cc 907 struct btrfs_fs_devices *fs_devices,
d397712b 908 void *data, int silent)
75dfe396 909{
d397712b 910 struct inode *inode;
815745cf 911 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
5d4f98a2 912 struct btrfs_key key;
39279cc3 913 int err;
a429e513 914
39279cc3
CM
915 sb->s_maxbytes = MAX_LFS_FILESIZE;
916 sb->s_magic = BTRFS_SUPER_MAGIC;
917 sb->s_op = &btrfs_super_ops;
af53d29a 918 sb->s_d_op = &btrfs_dentry_operations;
be6e8dc0 919 sb->s_export_op = &btrfs_export_ops;
5103e947 920 sb->s_xattr = btrfs_xattr_handlers;
39279cc3 921 sb->s_time_gran = 1;
0eda294d 922#ifdef CONFIG_BTRFS_FS_POSIX_ACL
33268eaf 923 sb->s_flags |= MS_POSIXACL;
49cf6f45 924#endif
0c4d2d95 925 sb->s_flags |= MS_I_VERSION;
ad2b2c80
AV
926 err = open_ctree(sb, fs_devices, (char *)data);
927 if (err) {
efe120a0 928 printk(KERN_ERR "BTRFS: open_ctree failed\n");
ad2b2c80 929 return err;
a429e513
CM
930 }
931
5d4f98a2
YZ
932 key.objectid = BTRFS_FIRST_FREE_OBJECTID;
933 key.type = BTRFS_INODE_ITEM_KEY;
934 key.offset = 0;
98c7089c 935 inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL);
5d4f98a2
YZ
936 if (IS_ERR(inode)) {
937 err = PTR_ERR(inode);
39279cc3 938 goto fail_close;
f254e52c 939 }
f254e52c 940
48fde701
AV
941 sb->s_root = d_make_root(inode);
942 if (!sb->s_root) {
39279cc3
CM
943 err = -ENOMEM;
944 goto fail_close;
f254e52c 945 }
58176a96 946
6885f308 947 save_mount_options(sb, data);
90a887c9 948 cleancache_init_fs(sb);
59553edf 949 sb->s_flags |= MS_ACTIVE;
2619ba1f 950 return 0;
39279cc3
CM
951
952fail_close:
815745cf 953 close_ctree(fs_info->tree_root);
39279cc3 954 return err;
2619ba1f
CM
955}
956
6bf13c0c 957int btrfs_sync_fs(struct super_block *sb, int wait)
c5739bba
CM
958{
959 struct btrfs_trans_handle *trans;
815745cf
AV
960 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
961 struct btrfs_root *root = fs_info->tree_root;
2619ba1f 962
1abe9b8a 963 trace_btrfs_sync_fs(wait);
964
39279cc3 965 if (!wait) {
815745cf 966 filemap_flush(fs_info->btree_inode->i_mapping);
39279cc3
CM
967 return 0;
968 }
771ed689 969
b0244199 970 btrfs_wait_ordered_roots(fs_info, -1);
771ed689 971
d4edf39b 972 trans = btrfs_attach_transaction_barrier(root);
60376ce4 973 if (IS_ERR(trans)) {
354aa0fb
MX
974 /* no transaction, don't bother */
975 if (PTR_ERR(trans) == -ENOENT)
60376ce4 976 return 0;
98d5dc13 977 return PTR_ERR(trans);
60376ce4 978 }
bd7de2c9 979 return btrfs_commit_transaction(trans, root);
2c90e5d6
CM
980}
981
34c80b1d 982static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
a9572a15 983{
815745cf
AV
984 struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb);
985 struct btrfs_root *root = info->tree_root;
200da64e 986 char *compress_type;
a9572a15
EP
987
988 if (btrfs_test_opt(root, DEGRADED))
989 seq_puts(seq, ",degraded");
990 if (btrfs_test_opt(root, NODATASUM))
991 seq_puts(seq, ",nodatasum");
992 if (btrfs_test_opt(root, NODATACOW))
993 seq_puts(seq, ",nodatacow");
994 if (btrfs_test_opt(root, NOBARRIER))
995 seq_puts(seq, ",nobarrier");
a9572a15 996 if (info->max_inline != 8192 * 1024)
c1c9ff7c 997 seq_printf(seq, ",max_inline=%llu", info->max_inline);
a9572a15 998 if (info->alloc_start != 0)
c1c9ff7c 999 seq_printf(seq, ",alloc_start=%llu", info->alloc_start);
a9572a15
EP
1000 if (info->thread_pool_size != min_t(unsigned long,
1001 num_online_cpus() + 2, 8))
1002 seq_printf(seq, ",thread_pool=%d", info->thread_pool_size);
200da64e
TI
1003 if (btrfs_test_opt(root, COMPRESS)) {
1004 if (info->compress_type == BTRFS_COMPRESS_ZLIB)
1005 compress_type = "zlib";
1006 else
1007 compress_type = "lzo";
1008 if (btrfs_test_opt(root, FORCE_COMPRESS))
1009 seq_printf(seq, ",compress-force=%s", compress_type);
1010 else
1011 seq_printf(seq, ",compress=%s", compress_type);
1012 }
c289811c
CM
1013 if (btrfs_test_opt(root, NOSSD))
1014 seq_puts(seq, ",nossd");
451d7585
CM
1015 if (btrfs_test_opt(root, SSD_SPREAD))
1016 seq_puts(seq, ",ssd_spread");
1017 else if (btrfs_test_opt(root, SSD))
a9572a15 1018 seq_puts(seq, ",ssd");
3a5e1404 1019 if (btrfs_test_opt(root, NOTREELOG))
6b65c5c6 1020 seq_puts(seq, ",notreelog");
dccae999 1021 if (btrfs_test_opt(root, FLUSHONCOMMIT))
6b65c5c6 1022 seq_puts(seq, ",flushoncommit");
20a5239a
MW
1023 if (btrfs_test_opt(root, DISCARD))
1024 seq_puts(seq, ",discard");
a9572a15
EP
1025 if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
1026 seq_puts(seq, ",noacl");
200da64e
TI
1027 if (btrfs_test_opt(root, SPACE_CACHE))
1028 seq_puts(seq, ",space_cache");
73bc1876 1029 else
8965593e 1030 seq_puts(seq, ",nospace_cache");
f420ee1e
SB
1031 if (btrfs_test_opt(root, RESCAN_UUID_TREE))
1032 seq_puts(seq, ",rescan_uuid_tree");
200da64e
TI
1033 if (btrfs_test_opt(root, CLEAR_CACHE))
1034 seq_puts(seq, ",clear_cache");
1035 if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED))
1036 seq_puts(seq, ",user_subvol_rm_allowed");
0942caa3
DS
1037 if (btrfs_test_opt(root, ENOSPC_DEBUG))
1038 seq_puts(seq, ",enospc_debug");
1039 if (btrfs_test_opt(root, AUTO_DEFRAG))
1040 seq_puts(seq, ",autodefrag");
1041 if (btrfs_test_opt(root, INODE_MAP_CACHE))
1042 seq_puts(seq, ",inode_cache");
9555c6c1
ID
1043 if (btrfs_test_opt(root, SKIP_BALANCE))
1044 seq_puts(seq, ",skip_balance");
8507d216
WS
1045 if (btrfs_test_opt(root, RECOVERY))
1046 seq_puts(seq, ",recovery");
1047#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1048 if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA))
1049 seq_puts(seq, ",check_int_data");
1050 else if (btrfs_test_opt(root, CHECK_INTEGRITY))
1051 seq_puts(seq, ",check_int");
1052 if (info->check_integrity_print_mask)
1053 seq_printf(seq, ",check_int_print_mask=%d",
1054 info->check_integrity_print_mask);
1055#endif
1056 if (info->metadata_ratio)
1057 seq_printf(seq, ",metadata_ratio=%d",
1058 info->metadata_ratio);
8c342930
JM
1059 if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
1060 seq_puts(seq, ",fatal_errors=panic");
8b87dc17
DS
1061 if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
1062 seq_printf(seq, ",commit=%d", info->commit_interval);
a9572a15
EP
1063 return 0;
1064}
1065
a061fc8d 1066static int btrfs_test_super(struct super_block *s, void *data)
4b82d6e4 1067{
815745cf
AV
1068 struct btrfs_fs_info *p = data;
1069 struct btrfs_fs_info *fs_info = btrfs_sb(s);
4b82d6e4 1070
815745cf 1071 return fs_info->fs_devices == p->fs_devices;
4b82d6e4
Y
1072}
1073
450ba0ea
JB
1074static int btrfs_set_super(struct super_block *s, void *data)
1075{
6de1d09d
AV
1076 int err = set_anon_super(s, data);
1077 if (!err)
1078 s->s_fs_info = data;
1079 return err;
4b82d6e4
Y
1080}
1081
f9d9ef62
DS
1082/*
1083 * subvolumes are identified by ino 256
1084 */
1085static inline int is_subvolume_inode(struct inode *inode)
1086{
1087 if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
1088 return 1;
1089 return 0;
1090}
1091
830c4adb
JB
1092/*
1093 * This will strip out the subvol=%s argument for an argument string and add
1094 * subvolid=0 to make sure we get the actual tree root for path walking to the
1095 * subvol we want.
1096 */
1097static char *setup_root_args(char *args)
1098{
f60d16a8
JM
1099 unsigned len = strlen(args) + 2 + 1;
1100 char *src, *dst, *buf;
830c4adb
JB
1101
1102 /*
f60d16a8
JM
1103 * We need the same args as before, but with this substitution:
1104 * s!subvol=[^,]+!subvolid=0!
830c4adb 1105 *
f60d16a8
JM
1106 * Since the replacement string is up to 2 bytes longer than the
1107 * original, allocate strlen(args) + 2 + 1 bytes.
830c4adb 1108 */
830c4adb 1109
f60d16a8 1110 src = strstr(args, "subvol=");
830c4adb 1111 /* This shouldn't happen, but just in case.. */
f60d16a8
JM
1112 if (!src)
1113 return NULL;
1114
1115 buf = dst = kmalloc(len, GFP_NOFS);
1116 if (!buf)
830c4adb 1117 return NULL;
830c4adb
JB
1118
1119 /*
f60d16a8
JM
1120 * If the subvol= arg is not at the start of the string,
1121 * copy whatever precedes it into buf.
830c4adb 1122 */
f60d16a8
JM
1123 if (src != args) {
1124 *src++ = '\0';
1125 strcpy(buf, args);
1126 dst += strlen(args);
830c4adb
JB
1127 }
1128
f60d16a8
JM
1129 strcpy(dst, "subvolid=0");
1130 dst += strlen("subvolid=0");
830c4adb
JB
1131
1132 /*
f60d16a8
JM
1133 * If there is a "," after the original subvol=... string,
1134 * copy that suffix into our buffer. Otherwise, we're done.
830c4adb 1135 */
f60d16a8
JM
1136 src = strchr(src, ',');
1137 if (src)
1138 strcpy(dst, src);
830c4adb 1139
f60d16a8 1140 return buf;
830c4adb
JB
1141}
1142
1143static struct dentry *mount_subvol(const char *subvol_name, int flags,
1144 const char *device_name, char *data)
1145{
830c4adb
JB
1146 struct dentry *root;
1147 struct vfsmount *mnt;
830c4adb 1148 char *newargs;
830c4adb
JB
1149
1150 newargs = setup_root_args(data);
1151 if (!newargs)
1152 return ERR_PTR(-ENOMEM);
1153 mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name,
1154 newargs);
1155 kfree(newargs);
1156 if (IS_ERR(mnt))
1157 return ERR_CAST(mnt);
1158
ea441d11 1159 root = mount_subtree(mnt, subvol_name);
830c4adb 1160
ea441d11
AV
1161 if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) {
1162 struct super_block *s = root->d_sb;
1163 dput(root);
1164 root = ERR_PTR(-EINVAL);
1165 deactivate_locked_super(s);
efe120a0 1166 printk(KERN_ERR "BTRFS: '%s' is not a valid subvolume\n",
f9d9ef62 1167 subvol_name);
f9d9ef62
DS
1168 }
1169
830c4adb
JB
1170 return root;
1171}
450ba0ea 1172
edf24abe
CH
1173/*
1174 * Find a superblock for the given device / mount point.
1175 *
1176 * Note: This is based on get_sb_bdev from fs/super.c with a few additions
1177 * for multiple device setup. Make sure to keep it in sync.
1178 */
061dbc6b 1179static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
306e16ce 1180 const char *device_name, void *data)
4b82d6e4
Y
1181{
1182 struct block_device *bdev = NULL;
1183 struct super_block *s;
1184 struct dentry *root;
8a4b83cc 1185 struct btrfs_fs_devices *fs_devices = NULL;
450ba0ea 1186 struct btrfs_fs_info *fs_info = NULL;
97288f2c 1187 fmode_t mode = FMODE_READ;
73f73415
JB
1188 char *subvol_name = NULL;
1189 u64 subvol_objectid = 0;
4b82d6e4
Y
1190 int error = 0;
1191
97288f2c
CH
1192 if (!(flags & MS_RDONLY))
1193 mode |= FMODE_WRITE;
1194
1195 error = btrfs_parse_early_options(data, mode, fs_type,
73f73415 1196 &subvol_name, &subvol_objectid,
5e2a4b25 1197 &fs_devices);
f23c8af8
ID
1198 if (error) {
1199 kfree(subvol_name);
061dbc6b 1200 return ERR_PTR(error);
f23c8af8 1201 }
edf24abe 1202
830c4adb
JB
1203 if (subvol_name) {
1204 root = mount_subvol(subvol_name, flags, device_name, data);
1205 kfree(subvol_name);
1206 return root;
1207 }
1208
306e16ce 1209 error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices);
8a4b83cc 1210 if (error)
830c4adb 1211 return ERR_PTR(error);
4b82d6e4 1212
450ba0ea
JB
1213 /*
1214 * Setup a dummy root and fs_info for test/set super. This is because
1215 * we don't actually fill this stuff out until open_ctree, but we need
1216 * it for searching for existing supers, so this lets us do that and
1217 * then open_ctree will properly initialize everything later.
1218 */
1219 fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS);
04d21a24
ID
1220 if (!fs_info)
1221 return ERR_PTR(-ENOMEM);
1222
450ba0ea 1223 fs_info->fs_devices = fs_devices;
450ba0ea 1224
6c41761f
DS
1225 fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1226 fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS);
1227 if (!fs_info->super_copy || !fs_info->super_for_commit) {
1228 error = -ENOMEM;
04d21a24
ID
1229 goto error_fs_info;
1230 }
1231
1232 error = btrfs_open_devices(fs_devices, mode, fs_type);
1233 if (error)
1234 goto error_fs_info;
1235
1236 if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
1237 error = -EACCES;
6c41761f
DS
1238 goto error_close_devices;
1239 }
1240
dfe25020 1241 bdev = fs_devices->latest_bdev;
9249e17f
DH
1242 s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC,
1243 fs_info);
830c4adb
JB
1244 if (IS_ERR(s)) {
1245 error = PTR_ERR(s);
1246 goto error_close_devices;
1247 }
4b82d6e4
Y
1248
1249 if (s->s_root) {
2b82032c 1250 btrfs_close_devices(fs_devices);
6c41761f 1251 free_fs_info(fs_info);
59553edf
AV
1252 if ((flags ^ s->s_flags) & MS_RDONLY)
1253 error = -EBUSY;
4b82d6e4
Y
1254 } else {
1255 char b[BDEVNAME_SIZE];
1256
4b82d6e4 1257 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
815745cf 1258 btrfs_sb(s)->bdev_holder = fs_type;
8a4b83cc
CM
1259 error = btrfs_fill_super(s, fs_devices, data,
1260 flags & MS_SILENT ? 1 : 0);
4b82d6e4
Y
1261 }
1262
59553edf
AV
1263 root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error);
1264 if (IS_ERR(root))
830c4adb 1265 deactivate_locked_super(s);
4b82d6e4 1266
061dbc6b 1267 return root;
4b82d6e4 1268
c146afad 1269error_close_devices:
8a4b83cc 1270 btrfs_close_devices(fs_devices);
04d21a24 1271error_fs_info:
6c41761f 1272 free_fs_info(fs_info);
061dbc6b 1273 return ERR_PTR(error);
4b82d6e4 1274}
2e635a27 1275
0d2450ab
ST
1276static void btrfs_set_max_workers(struct btrfs_workers *workers, int new_limit)
1277{
1278 spin_lock_irq(&workers->lock);
1279 workers->max_workers = new_limit;
1280 spin_unlock_irq(&workers->lock);
1281}
1282
1283static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info,
1284 int new_pool_size, int old_pool_size)
1285{
1286 if (new_pool_size == old_pool_size)
1287 return;
1288
1289 fs_info->thread_pool_size = new_pool_size;
1290
efe120a0 1291 btrfs_info(fs_info, "resize thread pool %d -> %d",
0d2450ab
ST
1292 old_pool_size, new_pool_size);
1293
1294 btrfs_set_max_workers(&fs_info->generic_worker, new_pool_size);
1295 btrfs_set_max_workers(&fs_info->workers, new_pool_size);
1296 btrfs_set_max_workers(&fs_info->delalloc_workers, new_pool_size);
1297 btrfs_set_max_workers(&fs_info->submit_workers, new_pool_size);
1298 btrfs_set_max_workers(&fs_info->caching_workers, new_pool_size);
1299 btrfs_set_max_workers(&fs_info->fixup_workers, new_pool_size);
1300 btrfs_set_max_workers(&fs_info->endio_workers, new_pool_size);
1301 btrfs_set_max_workers(&fs_info->endio_meta_workers, new_pool_size);
1302 btrfs_set_max_workers(&fs_info->endio_meta_write_workers, new_pool_size);
1303 btrfs_set_max_workers(&fs_info->endio_write_workers, new_pool_size);
1304 btrfs_set_max_workers(&fs_info->endio_freespace_worker, new_pool_size);
1305 btrfs_set_max_workers(&fs_info->delayed_workers, new_pool_size);
1306 btrfs_set_max_workers(&fs_info->readahead_workers, new_pool_size);
ff023aac
SB
1307 btrfs_set_max_workers(&fs_info->scrub_wr_completion_workers,
1308 new_pool_size);
0d2450ab
ST
1309}
1310
f42a34b2 1311static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info)
dc81cdc5
MX
1312{
1313 set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
f42a34b2 1314}
dc81cdc5 1315
f42a34b2
MX
1316static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info,
1317 unsigned long old_opts, int flags)
1318{
dc81cdc5
MX
1319 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1320 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1321 (flags & MS_RDONLY))) {
1322 /* wait for any defraggers to finish */
1323 wait_event(fs_info->transaction_wait,
1324 (atomic_read(&fs_info->defrag_running) == 0));
1325 if (flags & MS_RDONLY)
1326 sync_filesystem(fs_info->sb);
1327 }
1328}
1329
1330static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info,
1331 unsigned long old_opts)
1332{
1333 /*
1334 * We need cleanup all defragable inodes if the autodefragment is
1335 * close or the fs is R/O.
1336 */
1337 if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) &&
1338 (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) ||
1339 (fs_info->sb->s_flags & MS_RDONLY))) {
1340 btrfs_cleanup_defrag_inodes(fs_info);
1341 }
1342
1343 clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state);
1344}
1345
c146afad
YZ
1346static int btrfs_remount(struct super_block *sb, int *flags, char *data)
1347{
815745cf
AV
1348 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1349 struct btrfs_root *root = fs_info->tree_root;
49b25e05
JM
1350 unsigned old_flags = sb->s_flags;
1351 unsigned long old_opts = fs_info->mount_opt;
1352 unsigned long old_compress_type = fs_info->compress_type;
1353 u64 old_max_inline = fs_info->max_inline;
1354 u64 old_alloc_start = fs_info->alloc_start;
1355 int old_thread_pool_size = fs_info->thread_pool_size;
1356 unsigned int old_metadata_ratio = fs_info->metadata_ratio;
c146afad
YZ
1357 int ret;
1358
f42a34b2 1359 btrfs_remount_prepare(fs_info);
dc81cdc5 1360
b288052e 1361 ret = btrfs_parse_options(root, data);
49b25e05
JM
1362 if (ret) {
1363 ret = -EINVAL;
1364 goto restore;
1365 }
b288052e 1366
f42a34b2 1367 btrfs_remount_begin(fs_info, old_opts, *flags);
0d2450ab
ST
1368 btrfs_resize_thread_pool(fs_info,
1369 fs_info->thread_pool_size, old_thread_pool_size);
1370
c146afad 1371 if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
dc81cdc5 1372 goto out;
c146afad
YZ
1373
1374 if (*flags & MS_RDONLY) {
8dabb742
SB
1375 /*
1376 * this also happens on 'umount -rf' or on shutdown, when
1377 * the filesystem is busy.
1378 */
361c093d
SB
1379
1380 /* wait for the uuid_scan task to finish */
1381 down(&fs_info->uuid_tree_rescan_sem);
1382 /* avoid complains from lockdep et al. */
1383 up(&fs_info->uuid_tree_rescan_sem);
1384
c146afad
YZ
1385 sb->s_flags |= MS_RDONLY;
1386
8dabb742
SB
1387 btrfs_dev_replace_suspend_for_unmount(fs_info);
1388 btrfs_scrub_cancel(fs_info);
061594ef 1389 btrfs_pause_balance(fs_info);
8dabb742 1390
49b25e05
JM
1391 ret = btrfs_commit_super(root);
1392 if (ret)
1393 goto restore;
c146afad 1394 } else {
6ef3de9c
DS
1395 if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) {
1396 btrfs_err(fs_info,
efe120a0 1397 "Remounting read-write after error is not allowed");
6ef3de9c
DS
1398 ret = -EINVAL;
1399 goto restore;
1400 }
8a3db184 1401 if (fs_info->fs_devices->rw_devices == 0) {
49b25e05
JM
1402 ret = -EACCES;
1403 goto restore;
8a3db184 1404 }
2b82032c 1405
292fd7fc
SB
1406 if (fs_info->fs_devices->missing_devices >
1407 fs_info->num_tolerated_disk_barrier_failures &&
1408 !(*flags & MS_RDONLY)) {
efe120a0
FH
1409 btrfs_warn(fs_info,
1410 "too many missing devices, writeable remount is not allowed");
292fd7fc
SB
1411 ret = -EACCES;
1412 goto restore;
1413 }
1414
8a3db184 1415 if (btrfs_super_log_root(fs_info->super_copy) != 0) {
49b25e05
JM
1416 ret = -EINVAL;
1417 goto restore;
8a3db184 1418 }
c146afad 1419
815745cf 1420 ret = btrfs_cleanup_fs_roots(fs_info);
49b25e05
JM
1421 if (ret)
1422 goto restore;
c146afad 1423
d68fc57b
YZ
1424 /* recover relocation */
1425 ret = btrfs_recover_relocation(root);
49b25e05
JM
1426 if (ret)
1427 goto restore;
c146afad 1428
2b6ba629
ID
1429 ret = btrfs_resume_balance_async(fs_info);
1430 if (ret)
1431 goto restore;
1432
8dabb742
SB
1433 ret = btrfs_resume_dev_replace_async(fs_info);
1434 if (ret) {
efe120a0 1435 btrfs_warn(fs_info, "failed to resume dev_replace");
8dabb742
SB
1436 goto restore;
1437 }
94aebfb2
JB
1438
1439 if (!fs_info->uuid_root) {
efe120a0 1440 btrfs_info(fs_info, "creating UUID tree");
94aebfb2
JB
1441 ret = btrfs_create_uuid_tree(fs_info);
1442 if (ret) {
efe120a0 1443 btrfs_warn(fs_info, "failed to create the UUID tree %d", ret);
94aebfb2
JB
1444 goto restore;
1445 }
1446 }
c146afad
YZ
1447 sb->s_flags &= ~MS_RDONLY;
1448 }
dc81cdc5
MX
1449out:
1450 btrfs_remount_cleanup(fs_info, old_opts);
c146afad 1451 return 0;
49b25e05
JM
1452
1453restore:
1454 /* We've hit an error - don't reset MS_RDONLY */
1455 if (sb->s_flags & MS_RDONLY)
1456 old_flags |= MS_RDONLY;
1457 sb->s_flags = old_flags;
1458 fs_info->mount_opt = old_opts;
1459 fs_info->compress_type = old_compress_type;
1460 fs_info->max_inline = old_max_inline;
c018daec 1461 mutex_lock(&fs_info->chunk_mutex);
49b25e05 1462 fs_info->alloc_start = old_alloc_start;
c018daec 1463 mutex_unlock(&fs_info->chunk_mutex);
0d2450ab
ST
1464 btrfs_resize_thread_pool(fs_info,
1465 old_thread_pool_size, fs_info->thread_pool_size);
49b25e05 1466 fs_info->metadata_ratio = old_metadata_ratio;
dc81cdc5 1467 btrfs_remount_cleanup(fs_info, old_opts);
49b25e05 1468 return ret;
c146afad
YZ
1469}
1470
bcd53741
AJ
1471/* Used to sort the devices by max_avail(descending sort) */
1472static int btrfs_cmp_device_free_bytes(const void *dev_info1,
1473 const void *dev_info2)
1474{
1475 if (((struct btrfs_device_info *)dev_info1)->max_avail >
1476 ((struct btrfs_device_info *)dev_info2)->max_avail)
1477 return -1;
1478 else if (((struct btrfs_device_info *)dev_info1)->max_avail <
1479 ((struct btrfs_device_info *)dev_info2)->max_avail)
1480 return 1;
1481 else
1482 return 0;
1483}
1484
1485/*
1486 * sort the devices by max_avail, in which max free extent size of each device
1487 * is stored.(Descending Sort)
1488 */
1489static inline void btrfs_descending_sort_devices(
1490 struct btrfs_device_info *devices,
1491 size_t nr_devices)
1492{
1493 sort(devices, nr_devices, sizeof(struct btrfs_device_info),
1494 btrfs_cmp_device_free_bytes, NULL);
1495}
1496
6d07bcec
MX
1497/*
1498 * The helper to calc the free space on the devices that can be used to store
1499 * file data.
1500 */
1501static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
1502{
1503 struct btrfs_fs_info *fs_info = root->fs_info;
1504 struct btrfs_device_info *devices_info;
1505 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
1506 struct btrfs_device *device;
1507 u64 skip_space;
1508 u64 type;
1509 u64 avail_space;
1510 u64 used_space;
1511 u64 min_stripe_size;
39fb26c3 1512 int min_stripes = 1, num_stripes = 1;
6d07bcec
MX
1513 int i = 0, nr_devices;
1514 int ret;
1515
b772a86e 1516 nr_devices = fs_info->fs_devices->open_devices;
6d07bcec
MX
1517 BUG_ON(!nr_devices);
1518
d9b0d9ba 1519 devices_info = kmalloc_array(nr_devices, sizeof(*devices_info),
6d07bcec
MX
1520 GFP_NOFS);
1521 if (!devices_info)
1522 return -ENOMEM;
1523
1524 /* calc min stripe number for data space alloction */
1525 type = btrfs_get_alloc_profile(root, 1);
39fb26c3 1526 if (type & BTRFS_BLOCK_GROUP_RAID0) {
6d07bcec 1527 min_stripes = 2;
39fb26c3
MX
1528 num_stripes = nr_devices;
1529 } else if (type & BTRFS_BLOCK_GROUP_RAID1) {
6d07bcec 1530 min_stripes = 2;
39fb26c3
MX
1531 num_stripes = 2;
1532 } else if (type & BTRFS_BLOCK_GROUP_RAID10) {
6d07bcec 1533 min_stripes = 4;
39fb26c3
MX
1534 num_stripes = 4;
1535 }
6d07bcec
MX
1536
1537 if (type & BTRFS_BLOCK_GROUP_DUP)
1538 min_stripe_size = 2 * BTRFS_STRIPE_LEN;
1539 else
1540 min_stripe_size = BTRFS_STRIPE_LEN;
1541
b772a86e 1542 list_for_each_entry(device, &fs_devices->devices, dev_list) {
63a212ab
SB
1543 if (!device->in_fs_metadata || !device->bdev ||
1544 device->is_tgtdev_for_dev_replace)
6d07bcec
MX
1545 continue;
1546
1547 avail_space = device->total_bytes - device->bytes_used;
1548
1549 /* align with stripe_len */
1550 do_div(avail_space, BTRFS_STRIPE_LEN);
1551 avail_space *= BTRFS_STRIPE_LEN;
1552
1553 /*
1554 * In order to avoid overwritting the superblock on the drive,
1555 * btrfs starts at an offset of at least 1MB when doing chunk
1556 * allocation.
1557 */
1558 skip_space = 1024 * 1024;
1559
1560 /* user can set the offset in fs_info->alloc_start. */
1561 if (fs_info->alloc_start + BTRFS_STRIPE_LEN <=
1562 device->total_bytes)
1563 skip_space = max(fs_info->alloc_start, skip_space);
1564
1565 /*
1566 * btrfs can not use the free space in [0, skip_space - 1],
1567 * we must subtract it from the total. In order to implement
1568 * it, we account the used space in this range first.
1569 */
1570 ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1,
1571 &used_space);
1572 if (ret) {
1573 kfree(devices_info);
1574 return ret;
1575 }
1576
1577 /* calc the free space in [0, skip_space - 1] */
1578 skip_space -= used_space;
1579
1580 /*
1581 * we can use the free space in [0, skip_space - 1], subtract
1582 * it from the total.
1583 */
1584 if (avail_space && avail_space >= skip_space)
1585 avail_space -= skip_space;
1586 else
1587 avail_space = 0;
1588
1589 if (avail_space < min_stripe_size)
1590 continue;
1591
1592 devices_info[i].dev = device;
1593 devices_info[i].max_avail = avail_space;
1594
1595 i++;
1596 }
1597
1598 nr_devices = i;
1599
1600 btrfs_descending_sort_devices(devices_info, nr_devices);
1601
1602 i = nr_devices - 1;
1603 avail_space = 0;
1604 while (nr_devices >= min_stripes) {
39fb26c3
MX
1605 if (num_stripes > nr_devices)
1606 num_stripes = nr_devices;
1607
6d07bcec
MX
1608 if (devices_info[i].max_avail >= min_stripe_size) {
1609 int j;
1610 u64 alloc_size;
1611
39fb26c3 1612 avail_space += devices_info[i].max_avail * num_stripes;
6d07bcec 1613 alloc_size = devices_info[i].max_avail;
39fb26c3 1614 for (j = i + 1 - num_stripes; j <= i; j++)
6d07bcec
MX
1615 devices_info[j].max_avail -= alloc_size;
1616 }
1617 i--;
1618 nr_devices--;
1619 }
1620
1621 kfree(devices_info);
1622 *free_bytes = avail_space;
1623 return 0;
1624}
1625
8fd17795
CM
1626static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
1627{
815745cf
AV
1628 struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
1629 struct btrfs_super_block *disk_super = fs_info->super_copy;
1630 struct list_head *head = &fs_info->space_info;
bd4d1088
JB
1631 struct btrfs_space_info *found;
1632 u64 total_used = 0;
6d07bcec 1633 u64 total_free_data = 0;
db94535d 1634 int bits = dentry->d_sb->s_blocksize_bits;
815745cf 1635 __be32 *fsid = (__be32 *)fs_info->fsid;
6d07bcec 1636 int ret;
8fd17795 1637
6d07bcec 1638 /* holding chunk_muext to avoid allocating new chunks */
815745cf 1639 mutex_lock(&fs_info->chunk_mutex);
bd4d1088 1640 rcu_read_lock();
89a55897 1641 list_for_each_entry_rcu(found, head, list) {
6d07bcec
MX
1642 if (found->flags & BTRFS_BLOCK_GROUP_DATA) {
1643 total_free_data += found->disk_total - found->disk_used;
1644 total_free_data -=
1645 btrfs_account_ro_block_groups_free_space(found);
1646 }
1647
b742bb82 1648 total_used += found->disk_used;
89a55897 1649 }
bd4d1088
JB
1650 rcu_read_unlock();
1651
8fd17795 1652 buf->f_namelen = BTRFS_NAME_LEN;
db94535d 1653 buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
bd4d1088 1654 buf->f_bfree = buf->f_blocks - (total_used >> bits);
8fd17795
CM
1655 buf->f_bsize = dentry->d_sb->s_blocksize;
1656 buf->f_type = BTRFS_SUPER_MAGIC;
6d07bcec 1657 buf->f_bavail = total_free_data;
815745cf 1658 ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data);
6d07bcec 1659 if (ret) {
815745cf 1660 mutex_unlock(&fs_info->chunk_mutex);
6d07bcec
MX
1661 return ret;
1662 }
1663 buf->f_bavail += total_free_data;
1664 buf->f_bavail = buf->f_bavail >> bits;
815745cf 1665 mutex_unlock(&fs_info->chunk_mutex);
d397712b 1666
9d03632e 1667 /* We treat it as constant endianness (it doesn't matter _which_)
d397712b 1668 because we want the fsid to come out the same whether mounted
9d03632e
DW
1669 on a big-endian or little-endian host */
1670 buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]);
1671 buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]);
32d48fa1
DW
1672 /* Mask in the root object ID too, to disambiguate subvols */
1673 buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32;
1674 buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid;
1675
8fd17795
CM
1676 return 0;
1677}
b5133862 1678
aea52e19
AV
1679static void btrfs_kill_super(struct super_block *sb)
1680{
815745cf 1681 struct btrfs_fs_info *fs_info = btrfs_sb(sb);
aea52e19 1682 kill_anon_super(sb);
d22ca7de 1683 free_fs_info(fs_info);
aea52e19
AV
1684}
1685
2e635a27
CM
1686static struct file_system_type btrfs_fs_type = {
1687 .owner = THIS_MODULE,
1688 .name = "btrfs",
061dbc6b 1689 .mount = btrfs_mount,
aea52e19 1690 .kill_sb = btrfs_kill_super,
2e635a27
CM
1691 .fs_flags = FS_REQUIRES_DEV,
1692};
7f78e035 1693MODULE_ALIAS_FS("btrfs");
a9218f6b 1694
d352ac68
CM
1695/*
1696 * used by btrfsctl to scan devices when no FS is mounted
1697 */
8a4b83cc
CM
1698static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
1699 unsigned long arg)
1700{
1701 struct btrfs_ioctl_vol_args *vol;
1702 struct btrfs_fs_devices *fs_devices;
c071fcfd 1703 int ret = -ENOTTY;
8a4b83cc 1704
e441d54d
CM
1705 if (!capable(CAP_SYS_ADMIN))
1706 return -EPERM;
1707
dae7b665
LZ
1708 vol = memdup_user((void __user *)arg, sizeof(*vol));
1709 if (IS_ERR(vol))
1710 return PTR_ERR(vol);
c071fcfd 1711
8a4b83cc
CM
1712 switch (cmd) {
1713 case BTRFS_IOC_SCAN_DEV:
97288f2c 1714 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
8a4b83cc
CM
1715 &btrfs_fs_type, &fs_devices);
1716 break;
02db0844
JB
1717 case BTRFS_IOC_DEVICES_READY:
1718 ret = btrfs_scan_one_device(vol->name, FMODE_READ,
1719 &btrfs_fs_type, &fs_devices);
1720 if (ret)
1721 break;
1722 ret = !(fs_devices->num_devices == fs_devices->total_devices);
1723 break;
8a4b83cc 1724 }
dae7b665 1725
8a4b83cc 1726 kfree(vol);
f819d837 1727 return ret;
8a4b83cc
CM
1728}
1729
0176260f 1730static int btrfs_freeze(struct super_block *sb)
ed0dab6b 1731{
354aa0fb
MX
1732 struct btrfs_trans_handle *trans;
1733 struct btrfs_root *root = btrfs_sb(sb)->tree_root;
1734
d4edf39b 1735 trans = btrfs_attach_transaction_barrier(root);
354aa0fb
MX
1736 if (IS_ERR(trans)) {
1737 /* no transaction, don't bother */
1738 if (PTR_ERR(trans) == -ENOENT)
1739 return 0;
1740 return PTR_ERR(trans);
1741 }
1742 return btrfs_commit_transaction(trans, root);
ed0dab6b
Y
1743}
1744
0176260f 1745static int btrfs_unfreeze(struct super_block *sb)
ed0dab6b 1746{
0176260f 1747 return 0;
ed0dab6b 1748}
2e635a27 1749
9c5085c1
JB
1750static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
1751{
1752 struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb);
1753 struct btrfs_fs_devices *cur_devices;
1754 struct btrfs_device *dev, *first_dev = NULL;
1755 struct list_head *head;
1756 struct rcu_string *name;
1757
1758 mutex_lock(&fs_info->fs_devices->device_list_mutex);
1759 cur_devices = fs_info->fs_devices;
1760 while (cur_devices) {
1761 head = &cur_devices->devices;
1762 list_for_each_entry(dev, head, dev_list) {
aa9ddcd4
JB
1763 if (dev->missing)
1764 continue;
9c5085c1
JB
1765 if (!first_dev || dev->devid < first_dev->devid)
1766 first_dev = dev;
1767 }
1768 cur_devices = cur_devices->seed;
1769 }
1770
1771 if (first_dev) {
1772 rcu_read_lock();
1773 name = rcu_dereference(first_dev->name);
1774 seq_escape(m, name->str, " \t\n\\");
1775 rcu_read_unlock();
1776 } else {
1777 WARN_ON(1);
1778 }
1779 mutex_unlock(&fs_info->fs_devices->device_list_mutex);
1780 return 0;
1781}
1782
b87221de 1783static const struct super_operations btrfs_super_ops = {
76dda93c 1784 .drop_inode = btrfs_drop_inode,
bd555975 1785 .evict_inode = btrfs_evict_inode,
e20d96d6 1786 .put_super = btrfs_put_super,
d5719762 1787 .sync_fs = btrfs_sync_fs,
a9572a15 1788 .show_options = btrfs_show_options,
9c5085c1 1789 .show_devname = btrfs_show_devname,
4730a4bc 1790 .write_inode = btrfs_write_inode,
2c90e5d6
CM
1791 .alloc_inode = btrfs_alloc_inode,
1792 .destroy_inode = btrfs_destroy_inode,
8fd17795 1793 .statfs = btrfs_statfs,
c146afad 1794 .remount_fs = btrfs_remount,
0176260f
LT
1795 .freeze_fs = btrfs_freeze,
1796 .unfreeze_fs = btrfs_unfreeze,
e20d96d6 1797};
a9218f6b
CM
1798
1799static const struct file_operations btrfs_ctl_fops = {
1800 .unlocked_ioctl = btrfs_control_ioctl,
1801 .compat_ioctl = btrfs_control_ioctl,
1802 .owner = THIS_MODULE,
6038f373 1803 .llseek = noop_llseek,
a9218f6b
CM
1804};
1805
1806static struct miscdevice btrfs_misc = {
578454ff 1807 .minor = BTRFS_MINOR,
a9218f6b
CM
1808 .name = "btrfs-control",
1809 .fops = &btrfs_ctl_fops
1810};
1811
578454ff
KS
1812MODULE_ALIAS_MISCDEV(BTRFS_MINOR);
1813MODULE_ALIAS("devname:btrfs-control");
1814
a9218f6b
CM
1815static int btrfs_interface_init(void)
1816{
1817 return misc_register(&btrfs_misc);
1818}
1819
b2950863 1820static void btrfs_interface_exit(void)
a9218f6b
CM
1821{
1822 if (misc_deregister(&btrfs_misc) < 0)
efe120a0 1823 printk(KERN_INFO "BTRFS: misc_deregister failed for control device\n");
a9218f6b
CM
1824}
1825
85965600
DS
1826static void btrfs_print_info(void)
1827{
1828 printk(KERN_INFO "Btrfs loaded"
1829#ifdef CONFIG_BTRFS_DEBUG
1830 ", debug=on"
1831#endif
79556c3d
SB
1832#ifdef CONFIG_BTRFS_ASSERT
1833 ", assert=on"
1834#endif
85965600
DS
1835#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
1836 ", integrity-checker=on"
1837#endif
1838 "\n");
1839}
1840
dc11dd5d
JB
1841static int btrfs_run_sanity_tests(void)
1842{
06ea65a3
JB
1843 int ret;
1844
294e30fe 1845 ret = btrfs_init_test_fs();
06ea65a3
JB
1846 if (ret)
1847 return ret;
294e30fe
JB
1848
1849 ret = btrfs_test_free_space_cache();
1850 if (ret)
1851 goto out;
1852 ret = btrfs_test_extent_buffer_operations();
1853 if (ret)
1854 goto out;
1855 ret = btrfs_test_extent_io();
aaedb55b
JB
1856 if (ret)
1857 goto out;
1858 ret = btrfs_test_inodes();
294e30fe
JB
1859out:
1860 btrfs_destroy_test_fs();
1861 return ret;
dc11dd5d
JB
1862}
1863
2e635a27
CM
1864static int __init init_btrfs_fs(void)
1865{
2c90e5d6 1866 int err;
58176a96
JB
1867
1868 err = btrfs_init_sysfs();
1869 if (err)
1870 return err;
1871
143bede5 1872 btrfs_init_compress();
d1310b2e 1873
261507a0
LZ
1874 err = btrfs_init_cachep();
1875 if (err)
1876 goto free_compress;
1877
d1310b2e 1878 err = extent_io_init();
2f4cbe64
WB
1879 if (err)
1880 goto free_cachep;
1881
d1310b2e
CM
1882 err = extent_map_init();
1883 if (err)
1884 goto free_extent_io;
1885
6352b91d 1886 err = ordered_data_init();
2f4cbe64
WB
1887 if (err)
1888 goto free_extent_map;
c8b97818 1889
6352b91d
MX
1890 err = btrfs_delayed_inode_init();
1891 if (err)
1892 goto free_ordered_data;
1893
9247f317 1894 err = btrfs_auto_defrag_init();
16cdcec7
MX
1895 if (err)
1896 goto free_delayed_inode;
1897
78a6184a 1898 err = btrfs_delayed_ref_init();
9247f317
MX
1899 if (err)
1900 goto free_auto_defrag;
1901
b9e9a6cb
WS
1902 err = btrfs_prelim_ref_init();
1903 if (err)
1904 goto free_prelim_ref;
1905
78a6184a
MX
1906 err = btrfs_interface_init();
1907 if (err)
1908 goto free_delayed_ref;
1909
e565d4b9
JS
1910 btrfs_init_lockdep();
1911
85965600 1912 btrfs_print_info();
dc11dd5d
JB
1913
1914 err = btrfs_run_sanity_tests();
1915 if (err)
1916 goto unregister_ioctl;
1917
1918 err = register_filesystem(&btrfs_fs_type);
1919 if (err)
1920 goto unregister_ioctl;
74255aa0 1921
2f4cbe64
WB
1922 return 0;
1923
a9218f6b
CM
1924unregister_ioctl:
1925 btrfs_interface_exit();
b9e9a6cb
WS
1926free_prelim_ref:
1927 btrfs_prelim_ref_exit();
78a6184a
MX
1928free_delayed_ref:
1929 btrfs_delayed_ref_exit();
9247f317
MX
1930free_auto_defrag:
1931 btrfs_auto_defrag_exit();
16cdcec7
MX
1932free_delayed_inode:
1933 btrfs_delayed_inode_exit();
6352b91d
MX
1934free_ordered_data:
1935 ordered_data_exit();
2f4cbe64
WB
1936free_extent_map:
1937 extent_map_exit();
d1310b2e
CM
1938free_extent_io:
1939 extent_io_exit();
2f4cbe64
WB
1940free_cachep:
1941 btrfs_destroy_cachep();
261507a0
LZ
1942free_compress:
1943 btrfs_exit_compress();
2f4cbe64
WB
1944 btrfs_exit_sysfs();
1945 return err;
2e635a27
CM
1946}
1947
1948static void __exit exit_btrfs_fs(void)
1949{
39279cc3 1950 btrfs_destroy_cachep();
78a6184a 1951 btrfs_delayed_ref_exit();
9247f317 1952 btrfs_auto_defrag_exit();
16cdcec7 1953 btrfs_delayed_inode_exit();
b9e9a6cb 1954 btrfs_prelim_ref_exit();
6352b91d 1955 ordered_data_exit();
a52d9a80 1956 extent_map_exit();
d1310b2e 1957 extent_io_exit();
a9218f6b 1958 btrfs_interface_exit();
2e635a27 1959 unregister_filesystem(&btrfs_fs_type);
58176a96 1960 btrfs_exit_sysfs();
8a4b83cc 1961 btrfs_cleanup_fs_uuids();
261507a0 1962 btrfs_exit_compress();
2e635a27
CM
1963}
1964
1965module_init(init_btrfs_fs)
1966module_exit(exit_btrfs_fs)
1967
1968MODULE_LICENSE("GPL");