Commit | Line | Data |
---|---|---|
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 | 67 | static const struct super_operations btrfs_super_ops; |
830c4adb | 68 | static struct file_system_type btrfs_fs_type; |
75dfe396 | 69 | |
0723a047 HH |
70 | static int btrfs_remount(struct super_block *sb, int *flags, char *data); |
71 | ||
08748810 | 72 | static const 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 | 100 | static 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 */ |
110 | static 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 | ||
533574c6 | 133 | #ifdef CONFIG_PRINTK |
acce952b | 134 | /* |
135 | * __btrfs_std_error decodes expected errors from the caller and | |
136 | * invokes the approciate error response. | |
137 | */ | |
138 | void __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; | |
acce952b | 142 | const char *errstr; |
143 | ||
144 | /* | |
145 | * Special case: if the error is EROFS, and we're already | |
146 | * under MS_RDONLY, then it is safe here. | |
147 | */ | |
148 | if (errno == -EROFS && (sb->s_flags & MS_RDONLY)) | |
4da35113 JM |
149 | return; |
150 | ||
08748810 | 151 | errstr = btrfs_decode_error(errno); |
4da35113 | 152 | if (fmt) { |
37252a66 ES |
153 | struct va_format vaf; |
154 | va_list args; | |
155 | ||
156 | va_start(args, fmt); | |
157 | vaf.fmt = fmt; | |
158 | vaf.va = &args; | |
4da35113 | 159 | |
efe120a0 FH |
160 | printk(KERN_CRIT |
161 | "BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n", | |
08748810 | 162 | sb->s_id, function, line, errno, errstr, &vaf); |
37252a66 | 163 | va_end(args); |
4da35113 | 164 | } else { |
efe120a0 | 165 | printk(KERN_CRIT "BTRFS: error (device %s) in %s:%d: errno=%d %s\n", |
08748810 | 166 | sb->s_id, function, line, errno, errstr); |
4da35113 | 167 | } |
acce952b | 168 | |
4da35113 | 169 | /* Don't go through full error handling during mount */ |
cf79ffb5 JB |
170 | save_error_info(fs_info); |
171 | if (sb->s_flags & MS_BORN) | |
4da35113 | 172 | btrfs_handle_error(fs_info); |
4da35113 | 173 | } |
acce952b | 174 | |
533574c6 | 175 | static const char * const logtypes[] = { |
4da35113 JM |
176 | "emergency", |
177 | "alert", | |
178 | "critical", | |
179 | "error", | |
180 | "warning", | |
181 | "notice", | |
182 | "info", | |
183 | "debug", | |
184 | }; | |
185 | ||
c2cf52eb | 186 | void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...) |
4da35113 JM |
187 | { |
188 | struct super_block *sb = fs_info->sb; | |
189 | char lvl[4]; | |
190 | struct va_format vaf; | |
191 | va_list args; | |
192 | const char *type = logtypes[4]; | |
533574c6 | 193 | int kern_level; |
4da35113 JM |
194 | |
195 | va_start(args, fmt); | |
196 | ||
533574c6 JP |
197 | kern_level = printk_get_level(fmt); |
198 | if (kern_level) { | |
199 | size_t size = printk_skip_level(fmt) - fmt; | |
200 | memcpy(lvl, fmt, size); | |
201 | lvl[size] = '\0'; | |
202 | fmt += size; | |
203 | type = logtypes[kern_level - '0']; | |
4da35113 JM |
204 | } else |
205 | *lvl = '\0'; | |
206 | ||
207 | vaf.fmt = fmt; | |
208 | vaf.va = &args; | |
533574c6 | 209 | |
c2cf52eb | 210 | printk("%sBTRFS %s (device %s): %pV\n", lvl, type, sb->s_id, &vaf); |
533574c6 JP |
211 | |
212 | va_end(args); | |
213 | } | |
214 | ||
215 | #else | |
216 | ||
217 | void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function, | |
218 | unsigned int line, int errno, const char *fmt, ...) | |
219 | { | |
220 | struct super_block *sb = fs_info->sb; | |
221 | ||
222 | /* | |
223 | * Special case: if the error is EROFS, and we're already | |
224 | * under MS_RDONLY, then it is safe here. | |
225 | */ | |
226 | if (errno == -EROFS && (sb->s_flags & MS_RDONLY)) | |
227 | return; | |
228 | ||
229 | /* Don't go through full error handling during mount */ | |
230 | if (sb->s_flags & MS_BORN) { | |
231 | save_error_info(fs_info); | |
232 | btrfs_handle_error(fs_info); | |
233 | } | |
acce952b | 234 | } |
533574c6 | 235 | #endif |
acce952b | 236 | |
49b25e05 JM |
237 | /* |
238 | * We only mark the transaction aborted and then set the file system read-only. | |
239 | * This will prevent new transactions from starting or trying to join this | |
240 | * one. | |
241 | * | |
242 | * This means that error recovery at the call site is limited to freeing | |
243 | * any local memory allocations and passing the error code up without | |
244 | * further cleanup. The transaction should complete as it normally would | |
245 | * in the call path but will return -EIO. | |
246 | * | |
247 | * We'll complete the cleanup in btrfs_end_transaction and | |
248 | * btrfs_commit_transaction. | |
249 | */ | |
250 | void __btrfs_abort_transaction(struct btrfs_trans_handle *trans, | |
251 | struct btrfs_root *root, const char *function, | |
252 | unsigned int line, int errno) | |
253 | { | |
08748810 DS |
254 | /* |
255 | * Report first abort since mount | |
256 | */ | |
257 | if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED, | |
258 | &root->fs_info->fs_state)) { | |
efe120a0 | 259 | WARN(1, KERN_DEBUG "BTRFS: Transaction aborted (error %d)\n", |
08748810 DS |
260 | errno); |
261 | } | |
49b25e05 JM |
262 | trans->aborted = errno; |
263 | /* Nothing used. The other threads that have joined this | |
264 | * transaction may be able to continue. */ | |
265 | if (!trans->blocks_used) { | |
69ce977a MX |
266 | const char *errstr; |
267 | ||
08748810 | 268 | errstr = btrfs_decode_error(errno); |
c2cf52eb SK |
269 | btrfs_warn(root->fs_info, |
270 | "%s:%d: Aborting unused transaction(%s).", | |
271 | function, line, errstr); | |
acce952b | 272 | return; |
49b25e05 | 273 | } |
8d25a086 | 274 | ACCESS_ONCE(trans->transaction->aborted) = errno; |
501407aa JB |
275 | /* Wake up anybody who may be waiting on this transaction */ |
276 | wake_up(&root->fs_info->transaction_wait); | |
277 | wake_up(&root->fs_info->transaction_blocked_wait); | |
49b25e05 JM |
278 | __btrfs_std_error(root->fs_info, function, line, errno, NULL); |
279 | } | |
8c342930 JM |
280 | /* |
281 | * __btrfs_panic decodes unexpected, fatal errors from the caller, | |
282 | * issues an alert, and either panics or BUGs, depending on mount options. | |
283 | */ | |
284 | void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function, | |
285 | unsigned int line, int errno, const char *fmt, ...) | |
286 | { | |
8c342930 JM |
287 | char *s_id = "<unknown>"; |
288 | const char *errstr; | |
289 | struct va_format vaf = { .fmt = fmt }; | |
290 | va_list args; | |
acce952b | 291 | |
8c342930 JM |
292 | if (fs_info) |
293 | s_id = fs_info->sb->s_id; | |
acce952b | 294 | |
8c342930 JM |
295 | va_start(args, fmt); |
296 | vaf.va = &args; | |
297 | ||
08748810 | 298 | errstr = btrfs_decode_error(errno); |
aa43a17c | 299 | if (fs_info && (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR)) |
08748810 DS |
300 | panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n", |
301 | s_id, function, line, &vaf, errno, errstr); | |
8c342930 | 302 | |
efe120a0 FH |
303 | btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)", |
304 | function, line, &vaf, errno, errstr); | |
8c342930 JM |
305 | va_end(args); |
306 | /* Caller calls BUG() */ | |
acce952b | 307 | } |
308 | ||
d397712b | 309 | static void btrfs_put_super(struct super_block *sb) |
b18c6685 | 310 | { |
3abdbd78 | 311 | close_ctree(btrfs_sb(sb)->tree_root); |
75dfe396 CM |
312 | } |
313 | ||
95e05289 | 314 | enum { |
73f73415 | 315 | Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum, |
287a0ab9 JB |
316 | Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd, |
317 | Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress, | |
261507a0 LZ |
318 | Opt_compress_type, Opt_compress_force, Opt_compress_force_type, |
319 | Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_discard, | |
91435650 | 320 | Opt_space_cache, Opt_clear_cache, Opt_user_subvol_rm_allowed, |
9555c6c1 ID |
321 | Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache, |
322 | Opt_no_space_cache, Opt_recovery, Opt_skip_balance, | |
21adbd5c | 323 | Opt_check_integrity, Opt_check_integrity_including_extent_data, |
f420ee1e | 324 | Opt_check_integrity_print_mask, Opt_fatal_errors, Opt_rescan_uuid_tree, |
e07a2ade | 325 | Opt_commit_interval, Opt_barrier, Opt_nodefrag, Opt_nodiscard, |
a258af7a | 326 | Opt_noenospc_debug, Opt_noflushoncommit, Opt_acl, Opt_datacow, |
3818aea2 | 327 | Opt_datasum, Opt_treelog, Opt_noinode_cache, |
9555c6c1 | 328 | Opt_err, |
95e05289 CM |
329 | }; |
330 | ||
331 | static match_table_t tokens = { | |
dfe25020 | 332 | {Opt_degraded, "degraded"}, |
95e05289 | 333 | {Opt_subvol, "subvol=%s"}, |
1493381f | 334 | {Opt_subvolid, "subvolid=%s"}, |
43e570b0 | 335 | {Opt_device, "device=%s"}, |
b6cda9bc | 336 | {Opt_nodatasum, "nodatasum"}, |
d399167d | 337 | {Opt_datasum, "datasum"}, |
be20aa9d | 338 | {Opt_nodatacow, "nodatacow"}, |
a258af7a | 339 | {Opt_datacow, "datacow"}, |
21ad10cf | 340 | {Opt_nobarrier, "nobarrier"}, |
842bef58 | 341 | {Opt_barrier, "barrier"}, |
6f568d35 | 342 | {Opt_max_inline, "max_inline=%s"}, |
8f662a76 | 343 | {Opt_alloc_start, "alloc_start=%s"}, |
4543df7e | 344 | {Opt_thread_pool, "thread_pool=%d"}, |
c8b97818 | 345 | {Opt_compress, "compress"}, |
261507a0 | 346 | {Opt_compress_type, "compress=%s"}, |
a555f810 | 347 | {Opt_compress_force, "compress-force"}, |
261507a0 | 348 | {Opt_compress_force_type, "compress-force=%s"}, |
e18e4809 | 349 | {Opt_ssd, "ssd"}, |
451d7585 | 350 | {Opt_ssd_spread, "ssd_spread"}, |
3b30c22f | 351 | {Opt_nossd, "nossd"}, |
bd0330ad | 352 | {Opt_acl, "acl"}, |
33268eaf | 353 | {Opt_noacl, "noacl"}, |
3a5e1404 | 354 | {Opt_notreelog, "notreelog"}, |
a88998f2 | 355 | {Opt_treelog, "treelog"}, |
dccae999 | 356 | {Opt_flushoncommit, "flushoncommit"}, |
2c9ee856 | 357 | {Opt_noflushoncommit, "noflushoncommit"}, |
97e728d4 | 358 | {Opt_ratio, "metadata_ratio=%d"}, |
e244a0ae | 359 | {Opt_discard, "discard"}, |
e07a2ade | 360 | {Opt_nodiscard, "nodiscard"}, |
0af3d00b | 361 | {Opt_space_cache, "space_cache"}, |
88c2ba3b | 362 | {Opt_clear_cache, "clear_cache"}, |
4260f7c7 | 363 | {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"}, |
91435650 | 364 | {Opt_enospc_debug, "enospc_debug"}, |
53036293 | 365 | {Opt_noenospc_debug, "noenospc_debug"}, |
e15d0542 | 366 | {Opt_subvolrootid, "subvolrootid=%d"}, |
4cb5300b | 367 | {Opt_defrag, "autodefrag"}, |
fc0ca9af | 368 | {Opt_nodefrag, "noautodefrag"}, |
4b9465cb | 369 | {Opt_inode_cache, "inode_cache"}, |
3818aea2 | 370 | {Opt_noinode_cache, "noinode_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 | */ |
388 | int 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: | |
07802534 QW |
437 | btrfs_set_and_info(root, NODATASUM, |
438 | "setting nodatasum"); | |
be20aa9d | 439 | break; |
d399167d | 440 | case Opt_datasum: |
07802534 QW |
441 | if (btrfs_test_opt(root, NODATASUM)) { |
442 | if (btrfs_test_opt(root, NODATACOW)) | |
443 | btrfs_info(root->fs_info, "setting datasum, datacow enabled"); | |
444 | else | |
445 | btrfs_info(root->fs_info, "setting datasum"); | |
446 | } | |
d399167d QW |
447 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
448 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
449 | break; | |
be20aa9d | 450 | case Opt_nodatacow: |
07802534 QW |
451 | if (!btrfs_test_opt(root, NODATACOW)) { |
452 | if (!btrfs_test_opt(root, COMPRESS) || | |
453 | !btrfs_test_opt(root, FORCE_COMPRESS)) { | |
efe120a0 | 454 | btrfs_info(root->fs_info, |
07802534 QW |
455 | "setting nodatacow, compression disabled"); |
456 | } else { | |
457 | btrfs_info(root->fs_info, "setting nodatacow"); | |
458 | } | |
bedb2cca | 459 | } |
bedb2cca AP |
460 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
461 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
edf24abe CH |
462 | btrfs_set_opt(info->mount_opt, NODATACOW); |
463 | btrfs_set_opt(info->mount_opt, NODATASUM); | |
95e05289 | 464 | break; |
a258af7a | 465 | case Opt_datacow: |
07802534 QW |
466 | btrfs_clear_and_info(root, NODATACOW, |
467 | "setting datacow"); | |
a258af7a | 468 | break; |
a555f810 | 469 | case Opt_compress_force: |
261507a0 LZ |
470 | case Opt_compress_force_type: |
471 | compress_force = true; | |
1c697d4a | 472 | /* Fallthrough */ |
261507a0 LZ |
473 | case Opt_compress: |
474 | case Opt_compress_type: | |
475 | if (token == Opt_compress || | |
476 | token == Opt_compress_force || | |
477 | strcmp(args[0].from, "zlib") == 0) { | |
478 | compress_type = "zlib"; | |
479 | info->compress_type = BTRFS_COMPRESS_ZLIB; | |
063849ea | 480 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
481 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
482 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
a6fa6fae LZ |
483 | } else if (strcmp(args[0].from, "lzo") == 0) { |
484 | compress_type = "lzo"; | |
485 | info->compress_type = BTRFS_COMPRESS_LZO; | |
063849ea | 486 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
487 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
488 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
2b0ce2c2 | 489 | btrfs_set_fs_incompat(info, COMPRESS_LZO); |
063849ea AH |
490 | } else if (strncmp(args[0].from, "no", 2) == 0) { |
491 | compress_type = "no"; | |
063849ea AH |
492 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
493 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
494 | compress_force = false; | |
261507a0 LZ |
495 | } else { |
496 | ret = -EINVAL; | |
497 | goto out; | |
498 | } | |
499 | ||
261507a0 | 500 | if (compress_force) { |
07802534 QW |
501 | btrfs_set_and_info(root, FORCE_COMPRESS, |
502 | "force %s compression", | |
503 | compress_type); | |
143f3636 | 504 | } else { |
07802534 QW |
505 | if (!btrfs_test_opt(root, COMPRESS)) |
506 | btrfs_info(root->fs_info, | |
351fd353 | 507 | "btrfs: use %s compression", |
07802534 | 508 | compress_type); |
4027e0f4 WS |
509 | /* |
510 | * If we remount from compress-force=xxx to | |
511 | * compress=xxx, we need clear FORCE_COMPRESS | |
512 | * flag, otherwise, there is no way for users | |
513 | * to disable forcible compression separately. | |
514 | */ | |
515 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
a7e252af | 516 | } |
a555f810 | 517 | break; |
e18e4809 | 518 | case Opt_ssd: |
07802534 QW |
519 | btrfs_set_and_info(root, SSD, |
520 | "use ssd allocation scheme"); | |
e18e4809 | 521 | break; |
451d7585 | 522 | case Opt_ssd_spread: |
07802534 QW |
523 | btrfs_set_and_info(root, SSD_SPREAD, |
524 | "use spread ssd allocation scheme"); | |
2aa06a35 | 525 | btrfs_set_opt(info->mount_opt, SSD); |
451d7585 | 526 | break; |
3b30c22f | 527 | case Opt_nossd: |
2aa06a35 | 528 | btrfs_set_and_info(root, NOSSD, |
07802534 | 529 | "not using ssd allocation scheme"); |
3b30c22f CM |
530 | btrfs_clear_opt(info->mount_opt, SSD); |
531 | break; | |
842bef58 | 532 | case Opt_barrier: |
07802534 QW |
533 | btrfs_clear_and_info(root, NOBARRIER, |
534 | "turning on barriers"); | |
842bef58 | 535 | break; |
21ad10cf | 536 | case Opt_nobarrier: |
07802534 QW |
537 | btrfs_set_and_info(root, NOBARRIER, |
538 | "turning off barriers"); | |
21ad10cf | 539 | break; |
4543df7e | 540 | case Opt_thread_pool: |
2c334e87 WS |
541 | ret = match_int(&args[0], &intarg); |
542 | if (ret) { | |
543 | goto out; | |
544 | } else if (intarg > 0) { | |
4543df7e | 545 | info->thread_pool_size = intarg; |
2c334e87 WS |
546 | } else { |
547 | ret = -EINVAL; | |
548 | goto out; | |
549 | } | |
4543df7e | 550 | break; |
6f568d35 | 551 | case Opt_max_inline: |
edf24abe CH |
552 | num = match_strdup(&args[0]); |
553 | if (num) { | |
91748467 | 554 | info->max_inline = memparse(num, NULL); |
edf24abe CH |
555 | kfree(num); |
556 | ||
15ada040 | 557 | if (info->max_inline) { |
feb5f965 | 558 | info->max_inline = min_t(u64, |
15ada040 CM |
559 | info->max_inline, |
560 | root->sectorsize); | |
561 | } | |
efe120a0 | 562 | btrfs_info(root->fs_info, "max_inline at %llu", |
c1c9ff7c | 563 | info->max_inline); |
2c334e87 WS |
564 | } else { |
565 | ret = -ENOMEM; | |
566 | goto out; | |
6f568d35 CM |
567 | } |
568 | break; | |
8f662a76 | 569 | case Opt_alloc_start: |
edf24abe CH |
570 | num = match_strdup(&args[0]); |
571 | if (num) { | |
c018daec | 572 | mutex_lock(&info->chunk_mutex); |
91748467 | 573 | info->alloc_start = memparse(num, NULL); |
c018daec | 574 | mutex_unlock(&info->chunk_mutex); |
edf24abe | 575 | kfree(num); |
efe120a0 | 576 | btrfs_info(root->fs_info, "allocations start at %llu", |
c1c9ff7c | 577 | info->alloc_start); |
2c334e87 WS |
578 | } else { |
579 | ret = -ENOMEM; | |
580 | goto out; | |
8f662a76 CM |
581 | } |
582 | break; | |
bd0330ad | 583 | case Opt_acl: |
45ff35d6 | 584 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
bd0330ad QW |
585 | root->fs_info->sb->s_flags |= MS_POSIXACL; |
586 | break; | |
45ff35d6 GZ |
587 | #else |
588 | btrfs_err(root->fs_info, | |
589 | "support for ACL not compiled in!"); | |
590 | ret = -EINVAL; | |
591 | goto out; | |
592 | #endif | |
33268eaf JB |
593 | case Opt_noacl: |
594 | root->fs_info->sb->s_flags &= ~MS_POSIXACL; | |
595 | break; | |
3a5e1404 | 596 | case Opt_notreelog: |
07802534 QW |
597 | btrfs_set_and_info(root, NOTREELOG, |
598 | "disabling tree log"); | |
a88998f2 QW |
599 | break; |
600 | case Opt_treelog: | |
07802534 QW |
601 | btrfs_clear_and_info(root, NOTREELOG, |
602 | "enabling tree log"); | |
3a5e1404 | 603 | break; |
dccae999 | 604 | case Opt_flushoncommit: |
07802534 QW |
605 | btrfs_set_and_info(root, FLUSHONCOMMIT, |
606 | "turning on flush-on-commit"); | |
dccae999 | 607 | break; |
2c9ee856 | 608 | case Opt_noflushoncommit: |
07802534 QW |
609 | btrfs_clear_and_info(root, FLUSHONCOMMIT, |
610 | "turning off flush-on-commit"); | |
2c9ee856 | 611 | break; |
97e728d4 | 612 | case Opt_ratio: |
2c334e87 WS |
613 | ret = match_int(&args[0], &intarg); |
614 | if (ret) { | |
615 | goto out; | |
616 | } else if (intarg >= 0) { | |
97e728d4 | 617 | info->metadata_ratio = intarg; |
efe120a0 | 618 | btrfs_info(root->fs_info, "metadata ratio %d", |
97e728d4 | 619 | info->metadata_ratio); |
2c334e87 WS |
620 | } else { |
621 | ret = -EINVAL; | |
622 | goto out; | |
97e728d4 JB |
623 | } |
624 | break; | |
e244a0ae | 625 | case Opt_discard: |
07802534 QW |
626 | btrfs_set_and_info(root, DISCARD, |
627 | "turning on discard"); | |
e244a0ae | 628 | break; |
e07a2ade | 629 | case Opt_nodiscard: |
07802534 QW |
630 | btrfs_clear_and_info(root, DISCARD, |
631 | "turning off discard"); | |
e07a2ade | 632 | break; |
0af3d00b | 633 | case Opt_space_cache: |
07802534 QW |
634 | btrfs_set_and_info(root, SPACE_CACHE, |
635 | "enabling disk space caching"); | |
0de90876 | 636 | break; |
f420ee1e SB |
637 | case Opt_rescan_uuid_tree: |
638 | btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE); | |
639 | break; | |
73bc1876 | 640 | case Opt_no_space_cache: |
07802534 QW |
641 | btrfs_clear_and_info(root, SPACE_CACHE, |
642 | "disabling disk space caching"); | |
73bc1876 | 643 | break; |
4b9465cb | 644 | case Opt_inode_cache: |
07802534 QW |
645 | btrfs_set_and_info(root, CHANGE_INODE_CACHE, |
646 | "enabling inode map caching"); | |
3818aea2 QW |
647 | break; |
648 | case Opt_noinode_cache: | |
07802534 QW |
649 | btrfs_clear_and_info(root, CHANGE_INODE_CACHE, |
650 | "disabling inode map caching"); | |
4b9465cb | 651 | break; |
88c2ba3b | 652 | case Opt_clear_cache: |
07802534 QW |
653 | btrfs_set_and_info(root, CLEAR_CACHE, |
654 | "force clearing of disk cache"); | |
0af3d00b | 655 | break; |
4260f7c7 SW |
656 | case Opt_user_subvol_rm_allowed: |
657 | btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED); | |
658 | break; | |
91435650 CM |
659 | case Opt_enospc_debug: |
660 | btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG); | |
661 | break; | |
53036293 QW |
662 | case Opt_noenospc_debug: |
663 | btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG); | |
664 | break; | |
4cb5300b | 665 | case Opt_defrag: |
07802534 QW |
666 | btrfs_set_and_info(root, AUTO_DEFRAG, |
667 | "enabling auto defrag"); | |
4cb5300b | 668 | break; |
fc0ca9af | 669 | case Opt_nodefrag: |
07802534 QW |
670 | btrfs_clear_and_info(root, AUTO_DEFRAG, |
671 | "disabling auto defrag"); | |
fc0ca9af | 672 | break; |
af31f5e5 | 673 | case Opt_recovery: |
efe120a0 | 674 | btrfs_info(root->fs_info, "enabling auto recovery"); |
af31f5e5 CM |
675 | btrfs_set_opt(info->mount_opt, RECOVERY); |
676 | break; | |
9555c6c1 ID |
677 | case Opt_skip_balance: |
678 | btrfs_set_opt(info->mount_opt, SKIP_BALANCE); | |
679 | break; | |
21adbd5c SB |
680 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
681 | case Opt_check_integrity_including_extent_data: | |
efe120a0 FH |
682 | btrfs_info(root->fs_info, |
683 | "enabling check integrity including extent data"); | |
21adbd5c SB |
684 | btrfs_set_opt(info->mount_opt, |
685 | CHECK_INTEGRITY_INCLUDING_EXTENT_DATA); | |
686 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); | |
687 | break; | |
688 | case Opt_check_integrity: | |
efe120a0 | 689 | btrfs_info(root->fs_info, "enabling check integrity"); |
21adbd5c SB |
690 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); |
691 | break; | |
692 | case Opt_check_integrity_print_mask: | |
2c334e87 WS |
693 | ret = match_int(&args[0], &intarg); |
694 | if (ret) { | |
695 | goto out; | |
696 | } else if (intarg >= 0) { | |
21adbd5c | 697 | info->check_integrity_print_mask = intarg; |
efe120a0 | 698 | btrfs_info(root->fs_info, "check_integrity_print_mask 0x%x", |
21adbd5c | 699 | info->check_integrity_print_mask); |
2c334e87 WS |
700 | } else { |
701 | ret = -EINVAL; | |
702 | goto out; | |
21adbd5c SB |
703 | } |
704 | break; | |
705 | #else | |
706 | case Opt_check_integrity_including_extent_data: | |
707 | case Opt_check_integrity: | |
708 | case Opt_check_integrity_print_mask: | |
efe120a0 FH |
709 | btrfs_err(root->fs_info, |
710 | "support for check_integrity* not compiled in!"); | |
21adbd5c SB |
711 | ret = -EINVAL; |
712 | goto out; | |
713 | #endif | |
8c342930 JM |
714 | case Opt_fatal_errors: |
715 | if (strcmp(args[0].from, "panic") == 0) | |
716 | btrfs_set_opt(info->mount_opt, | |
717 | PANIC_ON_FATAL_ERROR); | |
718 | else if (strcmp(args[0].from, "bug") == 0) | |
719 | btrfs_clear_opt(info->mount_opt, | |
720 | PANIC_ON_FATAL_ERROR); | |
721 | else { | |
722 | ret = -EINVAL; | |
723 | goto out; | |
724 | } | |
725 | break; | |
8b87dc17 DS |
726 | case Opt_commit_interval: |
727 | intarg = 0; | |
728 | ret = match_int(&args[0], &intarg); | |
729 | if (ret < 0) { | |
efe120a0 | 730 | btrfs_err(root->fs_info, "invalid commit interval"); |
8b87dc17 DS |
731 | ret = -EINVAL; |
732 | goto out; | |
733 | } | |
734 | if (intarg > 0) { | |
735 | if (intarg > 300) { | |
efe120a0 | 736 | btrfs_warn(root->fs_info, "excessive commit interval %d", |
8b87dc17 DS |
737 | intarg); |
738 | } | |
739 | info->commit_interval = intarg; | |
740 | } else { | |
efe120a0 | 741 | btrfs_info(root->fs_info, "using default commit interval %ds", |
8b87dc17 DS |
742 | BTRFS_DEFAULT_COMMIT_INTERVAL); |
743 | info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL; | |
744 | } | |
745 | break; | |
a7a3f7ca | 746 | case Opt_err: |
efe120a0 | 747 | btrfs_info(root->fs_info, "unrecognized mount option '%s'", p); |
a7a3f7ca SW |
748 | ret = -EINVAL; |
749 | goto out; | |
95e05289 | 750 | default: |
be20aa9d | 751 | break; |
95e05289 CM |
752 | } |
753 | } | |
a7a3f7ca | 754 | out: |
73bc1876 | 755 | if (!ret && btrfs_test_opt(root, SPACE_CACHE)) |
efe120a0 | 756 | btrfs_info(root->fs_info, "disk space caching is enabled"); |
da495ecc | 757 | kfree(orig); |
a7a3f7ca | 758 | return ret; |
edf24abe CH |
759 | } |
760 | ||
761 | /* | |
762 | * Parse mount options that are required early in the mount process. | |
763 | * | |
764 | * All other options will be parsed on much later in the mount process and | |
765 | * only when we need to allocate a new super block. | |
766 | */ | |
97288f2c | 767 | static int btrfs_parse_early_options(const char *options, fmode_t flags, |
73f73415 | 768 | void *holder, char **subvol_name, u64 *subvol_objectid, |
5e2a4b25 | 769 | struct btrfs_fs_devices **fs_devices) |
edf24abe CH |
770 | { |
771 | substring_t args[MAX_OPT_ARGS]; | |
83c8c9bd | 772 | char *device_name, *opts, *orig, *p; |
1493381f | 773 | char *num = NULL; |
edf24abe CH |
774 | int error = 0; |
775 | ||
776 | if (!options) | |
830c4adb | 777 | return 0; |
edf24abe CH |
778 | |
779 | /* | |
780 | * strsep changes the string, duplicate it because parse_options | |
781 | * gets called twice | |
782 | */ | |
783 | opts = kstrdup(options, GFP_KERNEL); | |
784 | if (!opts) | |
785 | return -ENOMEM; | |
3f3d0bc0 | 786 | orig = opts; |
edf24abe CH |
787 | |
788 | while ((p = strsep(&opts, ",")) != NULL) { | |
789 | int token; | |
790 | if (!*p) | |
791 | continue; | |
792 | ||
793 | token = match_token(p, tokens, args); | |
794 | switch (token) { | |
795 | case Opt_subvol: | |
a90e8b6f | 796 | kfree(*subvol_name); |
edf24abe | 797 | *subvol_name = match_strdup(&args[0]); |
2c334e87 WS |
798 | if (!*subvol_name) { |
799 | error = -ENOMEM; | |
800 | goto out; | |
801 | } | |
edf24abe | 802 | break; |
73f73415 | 803 | case Opt_subvolid: |
1493381f WS |
804 | num = match_strdup(&args[0]); |
805 | if (num) { | |
806 | *subvol_objectid = memparse(num, NULL); | |
807 | kfree(num); | |
4849f01d | 808 | /* we want the original fs_tree */ |
1493381f | 809 | if (!*subvol_objectid) |
4849f01d JB |
810 | *subvol_objectid = |
811 | BTRFS_FS_TREE_OBJECTID; | |
2c334e87 WS |
812 | } else { |
813 | error = -EINVAL; | |
814 | goto out; | |
4849f01d | 815 | } |
73f73415 | 816 | break; |
e15d0542 | 817 | case Opt_subvolrootid: |
5e2a4b25 | 818 | printk(KERN_WARNING |
efe120a0 FH |
819 | "BTRFS: 'subvolrootid' mount option is deprecated and has " |
820 | "no effect\n"); | |
e15d0542 | 821 | break; |
43e570b0 | 822 | case Opt_device: |
83c8c9bd JL |
823 | device_name = match_strdup(&args[0]); |
824 | if (!device_name) { | |
825 | error = -ENOMEM; | |
826 | goto out; | |
827 | } | |
828 | error = btrfs_scan_one_device(device_name, | |
43e570b0 | 829 | flags, holder, fs_devices); |
83c8c9bd | 830 | kfree(device_name); |
43e570b0 | 831 | if (error) |
830c4adb | 832 | goto out; |
43e570b0 | 833 | break; |
edf24abe CH |
834 | default: |
835 | break; | |
836 | } | |
837 | } | |
838 | ||
830c4adb | 839 | out: |
3f3d0bc0 | 840 | kfree(orig); |
edf24abe | 841 | return error; |
95e05289 CM |
842 | } |
843 | ||
73f73415 JB |
844 | static struct dentry *get_default_root(struct super_block *sb, |
845 | u64 subvol_objectid) | |
846 | { | |
815745cf AV |
847 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
848 | struct btrfs_root *root = fs_info->tree_root; | |
73f73415 JB |
849 | struct btrfs_root *new_root; |
850 | struct btrfs_dir_item *di; | |
851 | struct btrfs_path *path; | |
852 | struct btrfs_key location; | |
853 | struct inode *inode; | |
73f73415 JB |
854 | u64 dir_id; |
855 | int new = 0; | |
856 | ||
857 | /* | |
858 | * We have a specific subvol we want to mount, just setup location and | |
859 | * go look up the root. | |
860 | */ | |
861 | if (subvol_objectid) { | |
862 | location.objectid = subvol_objectid; | |
863 | location.type = BTRFS_ROOT_ITEM_KEY; | |
864 | location.offset = (u64)-1; | |
865 | goto find_root; | |
866 | } | |
867 | ||
868 | path = btrfs_alloc_path(); | |
869 | if (!path) | |
870 | return ERR_PTR(-ENOMEM); | |
871 | path->leave_spinning = 1; | |
872 | ||
873 | /* | |
874 | * Find the "default" dir item which points to the root item that we | |
875 | * will mount by default if we haven't been given a specific subvolume | |
876 | * to mount. | |
877 | */ | |
815745cf | 878 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
73f73415 | 879 | di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0); |
b0839166 JL |
880 | if (IS_ERR(di)) { |
881 | btrfs_free_path(path); | |
fb4f6f91 | 882 | return ERR_CAST(di); |
b0839166 | 883 | } |
73f73415 JB |
884 | if (!di) { |
885 | /* | |
886 | * Ok the default dir item isn't there. This is weird since | |
887 | * it's always been there, but don't freak out, just try and | |
888 | * mount to root most subvolume. | |
889 | */ | |
890 | btrfs_free_path(path); | |
891 | dir_id = BTRFS_FIRST_FREE_OBJECTID; | |
815745cf | 892 | new_root = fs_info->fs_root; |
73f73415 JB |
893 | goto setup_root; |
894 | } | |
895 | ||
896 | btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); | |
897 | btrfs_free_path(path); | |
898 | ||
899 | find_root: | |
815745cf | 900 | new_root = btrfs_read_fs_root_no_name(fs_info, &location); |
73f73415 | 901 | if (IS_ERR(new_root)) |
d0b678cb | 902 | return ERR_CAST(new_root); |
73f73415 | 903 | |
73f73415 JB |
904 | dir_id = btrfs_root_dirid(&new_root->root_item); |
905 | setup_root: | |
906 | location.objectid = dir_id; | |
907 | location.type = BTRFS_INODE_ITEM_KEY; | |
908 | location.offset = 0; | |
909 | ||
910 | inode = btrfs_iget(sb, &location, new_root, &new); | |
4cbd1149 DC |
911 | if (IS_ERR(inode)) |
912 | return ERR_CAST(inode); | |
73f73415 JB |
913 | |
914 | /* | |
915 | * If we're just mounting the root most subvol put the inode and return | |
916 | * a reference to the dentry. We will have already gotten a reference | |
917 | * to the inode in btrfs_fill_super so we're good to go. | |
918 | */ | |
919 | if (!new && sb->s_root->d_inode == inode) { | |
920 | iput(inode); | |
921 | return dget(sb->s_root); | |
922 | } | |
923 | ||
1a0a397e | 924 | return d_obtain_root(inode); |
73f73415 JB |
925 | } |
926 | ||
d397712b | 927 | static int btrfs_fill_super(struct super_block *sb, |
8a4b83cc | 928 | struct btrfs_fs_devices *fs_devices, |
d397712b | 929 | void *data, int silent) |
75dfe396 | 930 | { |
d397712b | 931 | struct inode *inode; |
815745cf | 932 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
5d4f98a2 | 933 | struct btrfs_key key; |
39279cc3 | 934 | int err; |
a429e513 | 935 | |
39279cc3 CM |
936 | sb->s_maxbytes = MAX_LFS_FILESIZE; |
937 | sb->s_magic = BTRFS_SUPER_MAGIC; | |
938 | sb->s_op = &btrfs_super_ops; | |
af53d29a | 939 | sb->s_d_op = &btrfs_dentry_operations; |
be6e8dc0 | 940 | sb->s_export_op = &btrfs_export_ops; |
5103e947 | 941 | sb->s_xattr = btrfs_xattr_handlers; |
39279cc3 | 942 | sb->s_time_gran = 1; |
0eda294d | 943 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
33268eaf | 944 | sb->s_flags |= MS_POSIXACL; |
49cf6f45 | 945 | #endif |
0c4d2d95 | 946 | sb->s_flags |= MS_I_VERSION; |
ad2b2c80 AV |
947 | err = open_ctree(sb, fs_devices, (char *)data); |
948 | if (err) { | |
efe120a0 | 949 | printk(KERN_ERR "BTRFS: open_ctree failed\n"); |
ad2b2c80 | 950 | return err; |
a429e513 CM |
951 | } |
952 | ||
5d4f98a2 YZ |
953 | key.objectid = BTRFS_FIRST_FREE_OBJECTID; |
954 | key.type = BTRFS_INODE_ITEM_KEY; | |
955 | key.offset = 0; | |
98c7089c | 956 | inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL); |
5d4f98a2 YZ |
957 | if (IS_ERR(inode)) { |
958 | err = PTR_ERR(inode); | |
39279cc3 | 959 | goto fail_close; |
f254e52c | 960 | } |
f254e52c | 961 | |
48fde701 AV |
962 | sb->s_root = d_make_root(inode); |
963 | if (!sb->s_root) { | |
39279cc3 CM |
964 | err = -ENOMEM; |
965 | goto fail_close; | |
f254e52c | 966 | } |
58176a96 | 967 | |
6885f308 | 968 | save_mount_options(sb, data); |
90a887c9 | 969 | cleancache_init_fs(sb); |
59553edf | 970 | sb->s_flags |= MS_ACTIVE; |
2619ba1f | 971 | return 0; |
39279cc3 CM |
972 | |
973 | fail_close: | |
815745cf | 974 | close_ctree(fs_info->tree_root); |
39279cc3 | 975 | return err; |
2619ba1f CM |
976 | } |
977 | ||
6bf13c0c | 978 | int btrfs_sync_fs(struct super_block *sb, int wait) |
c5739bba CM |
979 | { |
980 | struct btrfs_trans_handle *trans; | |
815745cf AV |
981 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
982 | struct btrfs_root *root = fs_info->tree_root; | |
2619ba1f | 983 | |
1abe9b8a | 984 | trace_btrfs_sync_fs(wait); |
985 | ||
39279cc3 | 986 | if (!wait) { |
815745cf | 987 | filemap_flush(fs_info->btree_inode->i_mapping); |
39279cc3 CM |
988 | return 0; |
989 | } | |
771ed689 | 990 | |
b0244199 | 991 | btrfs_wait_ordered_roots(fs_info, -1); |
771ed689 | 992 | |
d4edf39b | 993 | trans = btrfs_attach_transaction_barrier(root); |
60376ce4 | 994 | if (IS_ERR(trans)) { |
354aa0fb MX |
995 | /* no transaction, don't bother */ |
996 | if (PTR_ERR(trans) == -ENOENT) | |
60376ce4 | 997 | return 0; |
98d5dc13 | 998 | return PTR_ERR(trans); |
60376ce4 | 999 | } |
bd7de2c9 | 1000 | return btrfs_commit_transaction(trans, root); |
2c90e5d6 CM |
1001 | } |
1002 | ||
34c80b1d | 1003 | static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry) |
a9572a15 | 1004 | { |
815745cf AV |
1005 | struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb); |
1006 | struct btrfs_root *root = info->tree_root; | |
200da64e | 1007 | char *compress_type; |
a9572a15 EP |
1008 | |
1009 | if (btrfs_test_opt(root, DEGRADED)) | |
1010 | seq_puts(seq, ",degraded"); | |
1011 | if (btrfs_test_opt(root, NODATASUM)) | |
1012 | seq_puts(seq, ",nodatasum"); | |
1013 | if (btrfs_test_opt(root, NODATACOW)) | |
1014 | seq_puts(seq, ",nodatacow"); | |
1015 | if (btrfs_test_opt(root, NOBARRIER)) | |
1016 | seq_puts(seq, ",nobarrier"); | |
95ac567a | 1017 | if (info->max_inline != BTRFS_DEFAULT_MAX_INLINE) |
c1c9ff7c | 1018 | seq_printf(seq, ",max_inline=%llu", info->max_inline); |
a9572a15 | 1019 | if (info->alloc_start != 0) |
c1c9ff7c | 1020 | seq_printf(seq, ",alloc_start=%llu", info->alloc_start); |
a9572a15 EP |
1021 | if (info->thread_pool_size != min_t(unsigned long, |
1022 | num_online_cpus() + 2, 8)) | |
1023 | seq_printf(seq, ",thread_pool=%d", info->thread_pool_size); | |
200da64e TI |
1024 | if (btrfs_test_opt(root, COMPRESS)) { |
1025 | if (info->compress_type == BTRFS_COMPRESS_ZLIB) | |
1026 | compress_type = "zlib"; | |
1027 | else | |
1028 | compress_type = "lzo"; | |
1029 | if (btrfs_test_opt(root, FORCE_COMPRESS)) | |
1030 | seq_printf(seq, ",compress-force=%s", compress_type); | |
1031 | else | |
1032 | seq_printf(seq, ",compress=%s", compress_type); | |
1033 | } | |
c289811c CM |
1034 | if (btrfs_test_opt(root, NOSSD)) |
1035 | seq_puts(seq, ",nossd"); | |
451d7585 CM |
1036 | if (btrfs_test_opt(root, SSD_SPREAD)) |
1037 | seq_puts(seq, ",ssd_spread"); | |
1038 | else if (btrfs_test_opt(root, SSD)) | |
a9572a15 | 1039 | seq_puts(seq, ",ssd"); |
3a5e1404 | 1040 | if (btrfs_test_opt(root, NOTREELOG)) |
6b65c5c6 | 1041 | seq_puts(seq, ",notreelog"); |
dccae999 | 1042 | if (btrfs_test_opt(root, FLUSHONCOMMIT)) |
6b65c5c6 | 1043 | seq_puts(seq, ",flushoncommit"); |
20a5239a MW |
1044 | if (btrfs_test_opt(root, DISCARD)) |
1045 | seq_puts(seq, ",discard"); | |
a9572a15 EP |
1046 | if (!(root->fs_info->sb->s_flags & MS_POSIXACL)) |
1047 | seq_puts(seq, ",noacl"); | |
200da64e TI |
1048 | if (btrfs_test_opt(root, SPACE_CACHE)) |
1049 | seq_puts(seq, ",space_cache"); | |
73bc1876 | 1050 | else |
8965593e | 1051 | seq_puts(seq, ",nospace_cache"); |
f420ee1e SB |
1052 | if (btrfs_test_opt(root, RESCAN_UUID_TREE)) |
1053 | seq_puts(seq, ",rescan_uuid_tree"); | |
200da64e TI |
1054 | if (btrfs_test_opt(root, CLEAR_CACHE)) |
1055 | seq_puts(seq, ",clear_cache"); | |
1056 | if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED)) | |
1057 | seq_puts(seq, ",user_subvol_rm_allowed"); | |
0942caa3 DS |
1058 | if (btrfs_test_opt(root, ENOSPC_DEBUG)) |
1059 | seq_puts(seq, ",enospc_debug"); | |
1060 | if (btrfs_test_opt(root, AUTO_DEFRAG)) | |
1061 | seq_puts(seq, ",autodefrag"); | |
1062 | if (btrfs_test_opt(root, INODE_MAP_CACHE)) | |
1063 | seq_puts(seq, ",inode_cache"); | |
9555c6c1 ID |
1064 | if (btrfs_test_opt(root, SKIP_BALANCE)) |
1065 | seq_puts(seq, ",skip_balance"); | |
8507d216 WS |
1066 | if (btrfs_test_opt(root, RECOVERY)) |
1067 | seq_puts(seq, ",recovery"); | |
1068 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY | |
1069 | if (btrfs_test_opt(root, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA)) | |
1070 | seq_puts(seq, ",check_int_data"); | |
1071 | else if (btrfs_test_opt(root, CHECK_INTEGRITY)) | |
1072 | seq_puts(seq, ",check_int"); | |
1073 | if (info->check_integrity_print_mask) | |
1074 | seq_printf(seq, ",check_int_print_mask=%d", | |
1075 | info->check_integrity_print_mask); | |
1076 | #endif | |
1077 | if (info->metadata_ratio) | |
1078 | seq_printf(seq, ",metadata_ratio=%d", | |
1079 | info->metadata_ratio); | |
8c342930 JM |
1080 | if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR)) |
1081 | seq_puts(seq, ",fatal_errors=panic"); | |
8b87dc17 DS |
1082 | if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL) |
1083 | seq_printf(seq, ",commit=%d", info->commit_interval); | |
a9572a15 EP |
1084 | return 0; |
1085 | } | |
1086 | ||
a061fc8d | 1087 | static int btrfs_test_super(struct super_block *s, void *data) |
4b82d6e4 | 1088 | { |
815745cf AV |
1089 | struct btrfs_fs_info *p = data; |
1090 | struct btrfs_fs_info *fs_info = btrfs_sb(s); | |
4b82d6e4 | 1091 | |
815745cf | 1092 | return fs_info->fs_devices == p->fs_devices; |
4b82d6e4 Y |
1093 | } |
1094 | ||
450ba0ea JB |
1095 | static int btrfs_set_super(struct super_block *s, void *data) |
1096 | { | |
6de1d09d AV |
1097 | int err = set_anon_super(s, data); |
1098 | if (!err) | |
1099 | s->s_fs_info = data; | |
1100 | return err; | |
4b82d6e4 Y |
1101 | } |
1102 | ||
f9d9ef62 DS |
1103 | /* |
1104 | * subvolumes are identified by ino 256 | |
1105 | */ | |
1106 | static inline int is_subvolume_inode(struct inode *inode) | |
1107 | { | |
1108 | if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) | |
1109 | return 1; | |
1110 | return 0; | |
1111 | } | |
1112 | ||
830c4adb JB |
1113 | /* |
1114 | * This will strip out the subvol=%s argument for an argument string and add | |
1115 | * subvolid=0 to make sure we get the actual tree root for path walking to the | |
1116 | * subvol we want. | |
1117 | */ | |
1118 | static char *setup_root_args(char *args) | |
1119 | { | |
f60d16a8 JM |
1120 | unsigned len = strlen(args) + 2 + 1; |
1121 | char *src, *dst, *buf; | |
830c4adb JB |
1122 | |
1123 | /* | |
f60d16a8 JM |
1124 | * We need the same args as before, but with this substitution: |
1125 | * s!subvol=[^,]+!subvolid=0! | |
830c4adb | 1126 | * |
f60d16a8 JM |
1127 | * Since the replacement string is up to 2 bytes longer than the |
1128 | * original, allocate strlen(args) + 2 + 1 bytes. | |
830c4adb | 1129 | */ |
830c4adb | 1130 | |
f60d16a8 | 1131 | src = strstr(args, "subvol="); |
830c4adb | 1132 | /* This shouldn't happen, but just in case.. */ |
f60d16a8 JM |
1133 | if (!src) |
1134 | return NULL; | |
1135 | ||
1136 | buf = dst = kmalloc(len, GFP_NOFS); | |
1137 | if (!buf) | |
830c4adb | 1138 | return NULL; |
830c4adb JB |
1139 | |
1140 | /* | |
f60d16a8 JM |
1141 | * If the subvol= arg is not at the start of the string, |
1142 | * copy whatever precedes it into buf. | |
830c4adb | 1143 | */ |
f60d16a8 JM |
1144 | if (src != args) { |
1145 | *src++ = '\0'; | |
1146 | strcpy(buf, args); | |
1147 | dst += strlen(args); | |
830c4adb JB |
1148 | } |
1149 | ||
f60d16a8 JM |
1150 | strcpy(dst, "subvolid=0"); |
1151 | dst += strlen("subvolid=0"); | |
830c4adb JB |
1152 | |
1153 | /* | |
f60d16a8 JM |
1154 | * If there is a "," after the original subvol=... string, |
1155 | * copy that suffix into our buffer. Otherwise, we're done. | |
830c4adb | 1156 | */ |
f60d16a8 JM |
1157 | src = strchr(src, ','); |
1158 | if (src) | |
1159 | strcpy(dst, src); | |
830c4adb | 1160 | |
f60d16a8 | 1161 | return buf; |
830c4adb JB |
1162 | } |
1163 | ||
1164 | static struct dentry *mount_subvol(const char *subvol_name, int flags, | |
1165 | const char *device_name, char *data) | |
1166 | { | |
830c4adb JB |
1167 | struct dentry *root; |
1168 | struct vfsmount *mnt; | |
830c4adb | 1169 | char *newargs; |
830c4adb JB |
1170 | |
1171 | newargs = setup_root_args(data); | |
1172 | if (!newargs) | |
1173 | return ERR_PTR(-ENOMEM); | |
1174 | mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name, | |
1175 | newargs); | |
0723a047 HH |
1176 | |
1177 | if (PTR_RET(mnt) == -EBUSY) { | |
1178 | if (flags & MS_RDONLY) { | |
1179 | mnt = vfs_kern_mount(&btrfs_fs_type, flags & ~MS_RDONLY, device_name, | |
1180 | newargs); | |
1181 | } else { | |
1182 | int r; | |
1183 | mnt = vfs_kern_mount(&btrfs_fs_type, flags | MS_RDONLY, device_name, | |
1184 | newargs); | |
0040e606 CJ |
1185 | if (IS_ERR(mnt)) { |
1186 | kfree(newargs); | |
0723a047 | 1187 | return ERR_CAST(mnt); |
0040e606 | 1188 | } |
0723a047 HH |
1189 | |
1190 | r = btrfs_remount(mnt->mnt_sb, &flags, NULL); | |
1191 | if (r < 0) { | |
1192 | /* FIXME: release vfsmount mnt ??*/ | |
0040e606 | 1193 | kfree(newargs); |
0723a047 HH |
1194 | return ERR_PTR(r); |
1195 | } | |
1196 | } | |
1197 | } | |
1198 | ||
0040e606 CJ |
1199 | kfree(newargs); |
1200 | ||
830c4adb JB |
1201 | if (IS_ERR(mnt)) |
1202 | return ERR_CAST(mnt); | |
1203 | ||
ea441d11 | 1204 | root = mount_subtree(mnt, subvol_name); |
830c4adb | 1205 | |
ea441d11 AV |
1206 | if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) { |
1207 | struct super_block *s = root->d_sb; | |
1208 | dput(root); | |
1209 | root = ERR_PTR(-EINVAL); | |
1210 | deactivate_locked_super(s); | |
efe120a0 | 1211 | printk(KERN_ERR "BTRFS: '%s' is not a valid subvolume\n", |
f9d9ef62 | 1212 | subvol_name); |
f9d9ef62 DS |
1213 | } |
1214 | ||
830c4adb JB |
1215 | return root; |
1216 | } | |
450ba0ea | 1217 | |
f667aef6 QW |
1218 | static int parse_security_options(char *orig_opts, |
1219 | struct security_mnt_opts *sec_opts) | |
1220 | { | |
1221 | char *secdata = NULL; | |
1222 | int ret = 0; | |
1223 | ||
1224 | secdata = alloc_secdata(); | |
1225 | if (!secdata) | |
1226 | return -ENOMEM; | |
1227 | ret = security_sb_copy_data(orig_opts, secdata); | |
1228 | if (ret) { | |
1229 | free_secdata(secdata); | |
1230 | return ret; | |
1231 | } | |
1232 | ret = security_sb_parse_opts_str(secdata, sec_opts); | |
1233 | free_secdata(secdata); | |
1234 | return ret; | |
1235 | } | |
1236 | ||
1237 | static int setup_security_options(struct btrfs_fs_info *fs_info, | |
1238 | struct super_block *sb, | |
1239 | struct security_mnt_opts *sec_opts) | |
1240 | { | |
1241 | int ret = 0; | |
1242 | ||
1243 | /* | |
1244 | * Call security_sb_set_mnt_opts() to check whether new sec_opts | |
1245 | * is valid. | |
1246 | */ | |
1247 | ret = security_sb_set_mnt_opts(sb, sec_opts, 0, NULL); | |
1248 | if (ret) | |
1249 | return ret; | |
1250 | ||
1251 | if (!fs_info->security_opts.num_mnt_opts) { | |
1252 | /* first time security setup, copy sec_opts to fs_info */ | |
1253 | memcpy(&fs_info->security_opts, sec_opts, sizeof(*sec_opts)); | |
1254 | } else { | |
1255 | /* | |
1256 | * Since SELinux(the only one supports security_mnt_opts) does | |
1257 | * NOT support changing context during remount/mount same sb, | |
1258 | * This must be the same or part of the same security options, | |
1259 | * just free it. | |
1260 | */ | |
1261 | security_free_mnt_opts(sec_opts); | |
1262 | } | |
1263 | return ret; | |
1264 | } | |
1265 | ||
edf24abe CH |
1266 | /* |
1267 | * Find a superblock for the given device / mount point. | |
1268 | * | |
1269 | * Note: This is based on get_sb_bdev from fs/super.c with a few additions | |
1270 | * for multiple device setup. Make sure to keep it in sync. | |
1271 | */ | |
061dbc6b | 1272 | static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, |
306e16ce | 1273 | const char *device_name, void *data) |
4b82d6e4 Y |
1274 | { |
1275 | struct block_device *bdev = NULL; | |
1276 | struct super_block *s; | |
1277 | struct dentry *root; | |
8a4b83cc | 1278 | struct btrfs_fs_devices *fs_devices = NULL; |
450ba0ea | 1279 | struct btrfs_fs_info *fs_info = NULL; |
f667aef6 | 1280 | struct security_mnt_opts new_sec_opts; |
97288f2c | 1281 | fmode_t mode = FMODE_READ; |
73f73415 JB |
1282 | char *subvol_name = NULL; |
1283 | u64 subvol_objectid = 0; | |
4b82d6e4 Y |
1284 | int error = 0; |
1285 | ||
97288f2c CH |
1286 | if (!(flags & MS_RDONLY)) |
1287 | mode |= FMODE_WRITE; | |
1288 | ||
1289 | error = btrfs_parse_early_options(data, mode, fs_type, | |
73f73415 | 1290 | &subvol_name, &subvol_objectid, |
5e2a4b25 | 1291 | &fs_devices); |
f23c8af8 ID |
1292 | if (error) { |
1293 | kfree(subvol_name); | |
061dbc6b | 1294 | return ERR_PTR(error); |
f23c8af8 | 1295 | } |
edf24abe | 1296 | |
830c4adb JB |
1297 | if (subvol_name) { |
1298 | root = mount_subvol(subvol_name, flags, device_name, data); | |
1299 | kfree(subvol_name); | |
1300 | return root; | |
1301 | } | |
1302 | ||
f667aef6 QW |
1303 | security_init_mnt_opts(&new_sec_opts); |
1304 | if (data) { | |
1305 | error = parse_security_options(data, &new_sec_opts); | |
1306 | if (error) | |
1307 | return ERR_PTR(error); | |
1308 | } | |
1309 | ||
306e16ce | 1310 | error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices); |
8a4b83cc | 1311 | if (error) |
f667aef6 | 1312 | goto error_sec_opts; |
4b82d6e4 | 1313 | |
450ba0ea JB |
1314 | /* |
1315 | * Setup a dummy root and fs_info for test/set super. This is because | |
1316 | * we don't actually fill this stuff out until open_ctree, but we need | |
1317 | * it for searching for existing supers, so this lets us do that and | |
1318 | * then open_ctree will properly initialize everything later. | |
1319 | */ | |
1320 | fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS); | |
f667aef6 QW |
1321 | if (!fs_info) { |
1322 | error = -ENOMEM; | |
1323 | goto error_sec_opts; | |
1324 | } | |
04d21a24 | 1325 | |
450ba0ea | 1326 | fs_info->fs_devices = fs_devices; |
450ba0ea | 1327 | |
6c41761f DS |
1328 | fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS); |
1329 | fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS); | |
f667aef6 | 1330 | security_init_mnt_opts(&fs_info->security_opts); |
6c41761f DS |
1331 | if (!fs_info->super_copy || !fs_info->super_for_commit) { |
1332 | error = -ENOMEM; | |
04d21a24 ID |
1333 | goto error_fs_info; |
1334 | } | |
1335 | ||
1336 | error = btrfs_open_devices(fs_devices, mode, fs_type); | |
1337 | if (error) | |
1338 | goto error_fs_info; | |
1339 | ||
1340 | if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) { | |
1341 | error = -EACCES; | |
6c41761f DS |
1342 | goto error_close_devices; |
1343 | } | |
1344 | ||
dfe25020 | 1345 | bdev = fs_devices->latest_bdev; |
9249e17f DH |
1346 | s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC, |
1347 | fs_info); | |
830c4adb JB |
1348 | if (IS_ERR(s)) { |
1349 | error = PTR_ERR(s); | |
1350 | goto error_close_devices; | |
1351 | } | |
4b82d6e4 Y |
1352 | |
1353 | if (s->s_root) { | |
2b82032c | 1354 | btrfs_close_devices(fs_devices); |
6c41761f | 1355 | free_fs_info(fs_info); |
59553edf AV |
1356 | if ((flags ^ s->s_flags) & MS_RDONLY) |
1357 | error = -EBUSY; | |
4b82d6e4 Y |
1358 | } else { |
1359 | char b[BDEVNAME_SIZE]; | |
1360 | ||
4b82d6e4 | 1361 | strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id)); |
815745cf | 1362 | btrfs_sb(s)->bdev_holder = fs_type; |
8a4b83cc CM |
1363 | error = btrfs_fill_super(s, fs_devices, data, |
1364 | flags & MS_SILENT ? 1 : 0); | |
4b82d6e4 Y |
1365 | } |
1366 | ||
59553edf | 1367 | root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error); |
f667aef6 QW |
1368 | if (IS_ERR(root)) { |
1369 | deactivate_locked_super(s); | |
1370 | error = PTR_ERR(root); | |
1371 | goto error_sec_opts; | |
1372 | } | |
1373 | ||
1374 | fs_info = btrfs_sb(s); | |
1375 | error = setup_security_options(fs_info, s, &new_sec_opts); | |
1376 | if (error) { | |
1377 | dput(root); | |
830c4adb | 1378 | deactivate_locked_super(s); |
f667aef6 QW |
1379 | goto error_sec_opts; |
1380 | } | |
4b82d6e4 | 1381 | |
061dbc6b | 1382 | return root; |
4b82d6e4 | 1383 | |
c146afad | 1384 | error_close_devices: |
8a4b83cc | 1385 | btrfs_close_devices(fs_devices); |
04d21a24 | 1386 | error_fs_info: |
6c41761f | 1387 | free_fs_info(fs_info); |
f667aef6 QW |
1388 | error_sec_opts: |
1389 | security_free_mnt_opts(&new_sec_opts); | |
061dbc6b | 1390 | return ERR_PTR(error); |
4b82d6e4 | 1391 | } |
2e635a27 | 1392 | |
0d2450ab ST |
1393 | static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info, |
1394 | int new_pool_size, int old_pool_size) | |
1395 | { | |
1396 | if (new_pool_size == old_pool_size) | |
1397 | return; | |
1398 | ||
1399 | fs_info->thread_pool_size = new_pool_size; | |
1400 | ||
efe120a0 | 1401 | btrfs_info(fs_info, "resize thread pool %d -> %d", |
0d2450ab ST |
1402 | old_pool_size, new_pool_size); |
1403 | ||
5cdc7ad3 | 1404 | btrfs_workqueue_set_max(fs_info->workers, new_pool_size); |
afe3d242 | 1405 | btrfs_workqueue_set_max(fs_info->delalloc_workers, new_pool_size); |
a8c93d4e | 1406 | btrfs_workqueue_set_max(fs_info->submit_workers, new_pool_size); |
e66f0bb1 | 1407 | btrfs_workqueue_set_max(fs_info->caching_workers, new_pool_size); |
fccb5d86 QW |
1408 | btrfs_workqueue_set_max(fs_info->endio_workers, new_pool_size); |
1409 | btrfs_workqueue_set_max(fs_info->endio_meta_workers, new_pool_size); | |
1410 | btrfs_workqueue_set_max(fs_info->endio_meta_write_workers, | |
1411 | new_pool_size); | |
1412 | btrfs_workqueue_set_max(fs_info->endio_write_workers, new_pool_size); | |
1413 | btrfs_workqueue_set_max(fs_info->endio_freespace_worker, new_pool_size); | |
5b3bc44e | 1414 | btrfs_workqueue_set_max(fs_info->delayed_workers, new_pool_size); |
736cfa15 | 1415 | btrfs_workqueue_set_max(fs_info->readahead_workers, new_pool_size); |
0339ef2f QW |
1416 | btrfs_workqueue_set_max(fs_info->scrub_wr_completion_workers, |
1417 | new_pool_size); | |
0d2450ab ST |
1418 | } |
1419 | ||
f42a34b2 | 1420 | static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info) |
dc81cdc5 MX |
1421 | { |
1422 | set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); | |
f42a34b2 | 1423 | } |
dc81cdc5 | 1424 | |
f42a34b2 MX |
1425 | static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info, |
1426 | unsigned long old_opts, int flags) | |
1427 | { | |
dc81cdc5 MX |
1428 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && |
1429 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || | |
1430 | (flags & MS_RDONLY))) { | |
1431 | /* wait for any defraggers to finish */ | |
1432 | wait_event(fs_info->transaction_wait, | |
1433 | (atomic_read(&fs_info->defrag_running) == 0)); | |
1434 | if (flags & MS_RDONLY) | |
1435 | sync_filesystem(fs_info->sb); | |
1436 | } | |
1437 | } | |
1438 | ||
1439 | static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info, | |
1440 | unsigned long old_opts) | |
1441 | { | |
1442 | /* | |
1443 | * We need cleanup all defragable inodes if the autodefragment is | |
1444 | * close or the fs is R/O. | |
1445 | */ | |
1446 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && | |
1447 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || | |
1448 | (fs_info->sb->s_flags & MS_RDONLY))) { | |
1449 | btrfs_cleanup_defrag_inodes(fs_info); | |
1450 | } | |
1451 | ||
1452 | clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); | |
1453 | } | |
1454 | ||
c146afad YZ |
1455 | static int btrfs_remount(struct super_block *sb, int *flags, char *data) |
1456 | { | |
815745cf AV |
1457 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
1458 | struct btrfs_root *root = fs_info->tree_root; | |
49b25e05 JM |
1459 | unsigned old_flags = sb->s_flags; |
1460 | unsigned long old_opts = fs_info->mount_opt; | |
1461 | unsigned long old_compress_type = fs_info->compress_type; | |
1462 | u64 old_max_inline = fs_info->max_inline; | |
1463 | u64 old_alloc_start = fs_info->alloc_start; | |
1464 | int old_thread_pool_size = fs_info->thread_pool_size; | |
1465 | unsigned int old_metadata_ratio = fs_info->metadata_ratio; | |
c146afad YZ |
1466 | int ret; |
1467 | ||
02b9984d | 1468 | sync_filesystem(sb); |
f42a34b2 | 1469 | btrfs_remount_prepare(fs_info); |
dc81cdc5 | 1470 | |
f667aef6 QW |
1471 | if (data) { |
1472 | struct security_mnt_opts new_sec_opts; | |
1473 | ||
1474 | security_init_mnt_opts(&new_sec_opts); | |
1475 | ret = parse_security_options(data, &new_sec_opts); | |
1476 | if (ret) | |
1477 | goto restore; | |
1478 | ret = setup_security_options(fs_info, sb, | |
1479 | &new_sec_opts); | |
1480 | if (ret) { | |
1481 | security_free_mnt_opts(&new_sec_opts); | |
1482 | goto restore; | |
1483 | } | |
1484 | } | |
1485 | ||
b288052e | 1486 | ret = btrfs_parse_options(root, data); |
49b25e05 JM |
1487 | if (ret) { |
1488 | ret = -EINVAL; | |
1489 | goto restore; | |
1490 | } | |
b288052e | 1491 | |
f42a34b2 | 1492 | btrfs_remount_begin(fs_info, old_opts, *flags); |
0d2450ab ST |
1493 | btrfs_resize_thread_pool(fs_info, |
1494 | fs_info->thread_pool_size, old_thread_pool_size); | |
1495 | ||
c146afad | 1496 | if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY)) |
dc81cdc5 | 1497 | goto out; |
c146afad YZ |
1498 | |
1499 | if (*flags & MS_RDONLY) { | |
8dabb742 SB |
1500 | /* |
1501 | * this also happens on 'umount -rf' or on shutdown, when | |
1502 | * the filesystem is busy. | |
1503 | */ | |
21c7e756 | 1504 | cancel_work_sync(&fs_info->async_reclaim_work); |
361c093d SB |
1505 | |
1506 | /* wait for the uuid_scan task to finish */ | |
1507 | down(&fs_info->uuid_tree_rescan_sem); | |
1508 | /* avoid complains from lockdep et al. */ | |
1509 | up(&fs_info->uuid_tree_rescan_sem); | |
1510 | ||
c146afad YZ |
1511 | sb->s_flags |= MS_RDONLY; |
1512 | ||
8dabb742 SB |
1513 | btrfs_dev_replace_suspend_for_unmount(fs_info); |
1514 | btrfs_scrub_cancel(fs_info); | |
061594ef | 1515 | btrfs_pause_balance(fs_info); |
8dabb742 | 1516 | |
49b25e05 JM |
1517 | ret = btrfs_commit_super(root); |
1518 | if (ret) | |
1519 | goto restore; | |
c146afad | 1520 | } else { |
6ef3de9c DS |
1521 | if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) { |
1522 | btrfs_err(fs_info, | |
efe120a0 | 1523 | "Remounting read-write after error is not allowed"); |
6ef3de9c DS |
1524 | ret = -EINVAL; |
1525 | goto restore; | |
1526 | } | |
8a3db184 | 1527 | if (fs_info->fs_devices->rw_devices == 0) { |
49b25e05 JM |
1528 | ret = -EACCES; |
1529 | goto restore; | |
8a3db184 | 1530 | } |
2b82032c | 1531 | |
292fd7fc SB |
1532 | if (fs_info->fs_devices->missing_devices > |
1533 | fs_info->num_tolerated_disk_barrier_failures && | |
1534 | !(*flags & MS_RDONLY)) { | |
efe120a0 FH |
1535 | btrfs_warn(fs_info, |
1536 | "too many missing devices, writeable remount is not allowed"); | |
292fd7fc SB |
1537 | ret = -EACCES; |
1538 | goto restore; | |
1539 | } | |
1540 | ||
8a3db184 | 1541 | if (btrfs_super_log_root(fs_info->super_copy) != 0) { |
49b25e05 JM |
1542 | ret = -EINVAL; |
1543 | goto restore; | |
8a3db184 | 1544 | } |
c146afad | 1545 | |
815745cf | 1546 | ret = btrfs_cleanup_fs_roots(fs_info); |
49b25e05 JM |
1547 | if (ret) |
1548 | goto restore; | |
c146afad | 1549 | |
d68fc57b | 1550 | /* recover relocation */ |
5f316481 | 1551 | mutex_lock(&fs_info->cleaner_mutex); |
d68fc57b | 1552 | ret = btrfs_recover_relocation(root); |
5f316481 | 1553 | mutex_unlock(&fs_info->cleaner_mutex); |
49b25e05 JM |
1554 | if (ret) |
1555 | goto restore; | |
c146afad | 1556 | |
2b6ba629 ID |
1557 | ret = btrfs_resume_balance_async(fs_info); |
1558 | if (ret) | |
1559 | goto restore; | |
1560 | ||
8dabb742 SB |
1561 | ret = btrfs_resume_dev_replace_async(fs_info); |
1562 | if (ret) { | |
efe120a0 | 1563 | btrfs_warn(fs_info, "failed to resume dev_replace"); |
8dabb742 SB |
1564 | goto restore; |
1565 | } | |
94aebfb2 JB |
1566 | |
1567 | if (!fs_info->uuid_root) { | |
efe120a0 | 1568 | btrfs_info(fs_info, "creating UUID tree"); |
94aebfb2 JB |
1569 | ret = btrfs_create_uuid_tree(fs_info); |
1570 | if (ret) { | |
efe120a0 | 1571 | btrfs_warn(fs_info, "failed to create the UUID tree %d", ret); |
94aebfb2 JB |
1572 | goto restore; |
1573 | } | |
1574 | } | |
c146afad YZ |
1575 | sb->s_flags &= ~MS_RDONLY; |
1576 | } | |
dc81cdc5 | 1577 | out: |
2c6a92b0 | 1578 | wake_up_process(fs_info->transaction_kthread); |
dc81cdc5 | 1579 | btrfs_remount_cleanup(fs_info, old_opts); |
c146afad | 1580 | return 0; |
49b25e05 JM |
1581 | |
1582 | restore: | |
1583 | /* We've hit an error - don't reset MS_RDONLY */ | |
1584 | if (sb->s_flags & MS_RDONLY) | |
1585 | old_flags |= MS_RDONLY; | |
1586 | sb->s_flags = old_flags; | |
1587 | fs_info->mount_opt = old_opts; | |
1588 | fs_info->compress_type = old_compress_type; | |
1589 | fs_info->max_inline = old_max_inline; | |
c018daec | 1590 | mutex_lock(&fs_info->chunk_mutex); |
49b25e05 | 1591 | fs_info->alloc_start = old_alloc_start; |
c018daec | 1592 | mutex_unlock(&fs_info->chunk_mutex); |
0d2450ab ST |
1593 | btrfs_resize_thread_pool(fs_info, |
1594 | old_thread_pool_size, fs_info->thread_pool_size); | |
49b25e05 | 1595 | fs_info->metadata_ratio = old_metadata_ratio; |
dc81cdc5 | 1596 | btrfs_remount_cleanup(fs_info, old_opts); |
49b25e05 | 1597 | return ret; |
c146afad YZ |
1598 | } |
1599 | ||
bcd53741 AJ |
1600 | /* Used to sort the devices by max_avail(descending sort) */ |
1601 | static int btrfs_cmp_device_free_bytes(const void *dev_info1, | |
1602 | const void *dev_info2) | |
1603 | { | |
1604 | if (((struct btrfs_device_info *)dev_info1)->max_avail > | |
1605 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
1606 | return -1; | |
1607 | else if (((struct btrfs_device_info *)dev_info1)->max_avail < | |
1608 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
1609 | return 1; | |
1610 | else | |
1611 | return 0; | |
1612 | } | |
1613 | ||
1614 | /* | |
1615 | * sort the devices by max_avail, in which max free extent size of each device | |
1616 | * is stored.(Descending Sort) | |
1617 | */ | |
1618 | static inline void btrfs_descending_sort_devices( | |
1619 | struct btrfs_device_info *devices, | |
1620 | size_t nr_devices) | |
1621 | { | |
1622 | sort(devices, nr_devices, sizeof(struct btrfs_device_info), | |
1623 | btrfs_cmp_device_free_bytes, NULL); | |
1624 | } | |
1625 | ||
6d07bcec MX |
1626 | /* |
1627 | * The helper to calc the free space on the devices that can be used to store | |
1628 | * file data. | |
1629 | */ | |
1630 | static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes) | |
1631 | { | |
1632 | struct btrfs_fs_info *fs_info = root->fs_info; | |
1633 | struct btrfs_device_info *devices_info; | |
1634 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; | |
1635 | struct btrfs_device *device; | |
1636 | u64 skip_space; | |
1637 | u64 type; | |
1638 | u64 avail_space; | |
1639 | u64 used_space; | |
1640 | u64 min_stripe_size; | |
39fb26c3 | 1641 | int min_stripes = 1, num_stripes = 1; |
6d07bcec MX |
1642 | int i = 0, nr_devices; |
1643 | int ret; | |
1644 | ||
b772a86e | 1645 | nr_devices = fs_info->fs_devices->open_devices; |
6d07bcec MX |
1646 | BUG_ON(!nr_devices); |
1647 | ||
d9b0d9ba | 1648 | devices_info = kmalloc_array(nr_devices, sizeof(*devices_info), |
6d07bcec MX |
1649 | GFP_NOFS); |
1650 | if (!devices_info) | |
1651 | return -ENOMEM; | |
1652 | ||
1653 | /* calc min stripe number for data space alloction */ | |
1654 | type = btrfs_get_alloc_profile(root, 1); | |
39fb26c3 | 1655 | if (type & BTRFS_BLOCK_GROUP_RAID0) { |
6d07bcec | 1656 | min_stripes = 2; |
39fb26c3 MX |
1657 | num_stripes = nr_devices; |
1658 | } else if (type & BTRFS_BLOCK_GROUP_RAID1) { | |
6d07bcec | 1659 | min_stripes = 2; |
39fb26c3 MX |
1660 | num_stripes = 2; |
1661 | } else if (type & BTRFS_BLOCK_GROUP_RAID10) { | |
6d07bcec | 1662 | min_stripes = 4; |
39fb26c3 MX |
1663 | num_stripes = 4; |
1664 | } | |
6d07bcec MX |
1665 | |
1666 | if (type & BTRFS_BLOCK_GROUP_DUP) | |
1667 | min_stripe_size = 2 * BTRFS_STRIPE_LEN; | |
1668 | else | |
1669 | min_stripe_size = BTRFS_STRIPE_LEN; | |
1670 | ||
b772a86e | 1671 | list_for_each_entry(device, &fs_devices->devices, dev_list) { |
63a212ab SB |
1672 | if (!device->in_fs_metadata || !device->bdev || |
1673 | device->is_tgtdev_for_dev_replace) | |
6d07bcec MX |
1674 | continue; |
1675 | ||
1676 | avail_space = device->total_bytes - device->bytes_used; | |
1677 | ||
1678 | /* align with stripe_len */ | |
1679 | do_div(avail_space, BTRFS_STRIPE_LEN); | |
1680 | avail_space *= BTRFS_STRIPE_LEN; | |
1681 | ||
1682 | /* | |
1683 | * In order to avoid overwritting the superblock on the drive, | |
1684 | * btrfs starts at an offset of at least 1MB when doing chunk | |
1685 | * allocation. | |
1686 | */ | |
1687 | skip_space = 1024 * 1024; | |
1688 | ||
1689 | /* user can set the offset in fs_info->alloc_start. */ | |
1690 | if (fs_info->alloc_start + BTRFS_STRIPE_LEN <= | |
1691 | device->total_bytes) | |
1692 | skip_space = max(fs_info->alloc_start, skip_space); | |
1693 | ||
1694 | /* | |
1695 | * btrfs can not use the free space in [0, skip_space - 1], | |
1696 | * we must subtract it from the total. In order to implement | |
1697 | * it, we account the used space in this range first. | |
1698 | */ | |
1699 | ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1, | |
1700 | &used_space); | |
1701 | if (ret) { | |
1702 | kfree(devices_info); | |
1703 | return ret; | |
1704 | } | |
1705 | ||
1706 | /* calc the free space in [0, skip_space - 1] */ | |
1707 | skip_space -= used_space; | |
1708 | ||
1709 | /* | |
1710 | * we can use the free space in [0, skip_space - 1], subtract | |
1711 | * it from the total. | |
1712 | */ | |
1713 | if (avail_space && avail_space >= skip_space) | |
1714 | avail_space -= skip_space; | |
1715 | else | |
1716 | avail_space = 0; | |
1717 | ||
1718 | if (avail_space < min_stripe_size) | |
1719 | continue; | |
1720 | ||
1721 | devices_info[i].dev = device; | |
1722 | devices_info[i].max_avail = avail_space; | |
1723 | ||
1724 | i++; | |
1725 | } | |
1726 | ||
1727 | nr_devices = i; | |
1728 | ||
1729 | btrfs_descending_sort_devices(devices_info, nr_devices); | |
1730 | ||
1731 | i = nr_devices - 1; | |
1732 | avail_space = 0; | |
1733 | while (nr_devices >= min_stripes) { | |
39fb26c3 MX |
1734 | if (num_stripes > nr_devices) |
1735 | num_stripes = nr_devices; | |
1736 | ||
6d07bcec MX |
1737 | if (devices_info[i].max_avail >= min_stripe_size) { |
1738 | int j; | |
1739 | u64 alloc_size; | |
1740 | ||
39fb26c3 | 1741 | avail_space += devices_info[i].max_avail * num_stripes; |
6d07bcec | 1742 | alloc_size = devices_info[i].max_avail; |
39fb26c3 | 1743 | for (j = i + 1 - num_stripes; j <= i; j++) |
6d07bcec MX |
1744 | devices_info[j].max_avail -= alloc_size; |
1745 | } | |
1746 | i--; | |
1747 | nr_devices--; | |
1748 | } | |
1749 | ||
1750 | kfree(devices_info); | |
1751 | *free_bytes = avail_space; | |
1752 | return 0; | |
1753 | } | |
1754 | ||
ba7b6e62 DS |
1755 | /* |
1756 | * Calculate numbers for 'df', pessimistic in case of mixed raid profiles. | |
1757 | * | |
1758 | * If there's a redundant raid level at DATA block groups, use the respective | |
1759 | * multiplier to scale the sizes. | |
1760 | * | |
1761 | * Unused device space usage is based on simulating the chunk allocator | |
1762 | * algorithm that respects the device sizes, order of allocations and the | |
1763 | * 'alloc_start' value, this is a close approximation of the actual use but | |
1764 | * there are other factors that may change the result (like a new metadata | |
1765 | * chunk). | |
1766 | * | |
1767 | * FIXME: not accurate for mixed block groups, total and free/used are ok, | |
1768 | * available appears slightly larger. | |
1769 | */ | |
8fd17795 CM |
1770 | static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) |
1771 | { | |
815745cf AV |
1772 | struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb); |
1773 | struct btrfs_super_block *disk_super = fs_info->super_copy; | |
1774 | struct list_head *head = &fs_info->space_info; | |
bd4d1088 JB |
1775 | struct btrfs_space_info *found; |
1776 | u64 total_used = 0; | |
6d07bcec | 1777 | u64 total_free_data = 0; |
db94535d | 1778 | int bits = dentry->d_sb->s_blocksize_bits; |
815745cf | 1779 | __be32 *fsid = (__be32 *)fs_info->fsid; |
ba7b6e62 DS |
1780 | unsigned factor = 1; |
1781 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; | |
6d07bcec | 1782 | int ret; |
8fd17795 | 1783 | |
15484377 MX |
1784 | /* |
1785 | * holding chunk_muext to avoid allocating new chunks, holding | |
1786 | * device_list_mutex to avoid the device being removed | |
1787 | */ | |
1788 | mutex_lock(&fs_info->fs_devices->device_list_mutex); | |
815745cf | 1789 | mutex_lock(&fs_info->chunk_mutex); |
bd4d1088 | 1790 | rcu_read_lock(); |
89a55897 | 1791 | list_for_each_entry_rcu(found, head, list) { |
6d07bcec | 1792 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) { |
ba7b6e62 DS |
1793 | int i; |
1794 | ||
6d07bcec MX |
1795 | total_free_data += found->disk_total - found->disk_used; |
1796 | total_free_data -= | |
1797 | btrfs_account_ro_block_groups_free_space(found); | |
ba7b6e62 DS |
1798 | |
1799 | for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) { | |
1800 | if (!list_empty(&found->block_groups[i])) { | |
1801 | switch (i) { | |
1802 | case BTRFS_RAID_DUP: | |
1803 | case BTRFS_RAID_RAID1: | |
1804 | case BTRFS_RAID_RAID10: | |
1805 | factor = 2; | |
1806 | } | |
1807 | } | |
1808 | } | |
6d07bcec MX |
1809 | } |
1810 | ||
b742bb82 | 1811 | total_used += found->disk_used; |
89a55897 | 1812 | } |
ba7b6e62 | 1813 | |
bd4d1088 JB |
1814 | rcu_read_unlock(); |
1815 | ||
ba7b6e62 DS |
1816 | buf->f_blocks = div_u64(btrfs_super_total_bytes(disk_super), factor); |
1817 | buf->f_blocks >>= bits; | |
1818 | buf->f_bfree = buf->f_blocks - (div_u64(total_used, factor) >> bits); | |
1819 | ||
1820 | /* Account global block reserve as used, it's in logical size already */ | |
1821 | spin_lock(&block_rsv->lock); | |
1822 | buf->f_bfree -= block_rsv->size >> bits; | |
1823 | spin_unlock(&block_rsv->lock); | |
1824 | ||
6d07bcec | 1825 | buf->f_bavail = total_free_data; |
815745cf | 1826 | ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data); |
6d07bcec | 1827 | if (ret) { |
815745cf | 1828 | mutex_unlock(&fs_info->chunk_mutex); |
15484377 | 1829 | mutex_unlock(&fs_info->fs_devices->device_list_mutex); |
6d07bcec MX |
1830 | return ret; |
1831 | } | |
ba7b6e62 | 1832 | buf->f_bavail += div_u64(total_free_data, factor); |
6d07bcec | 1833 | buf->f_bavail = buf->f_bavail >> bits; |
815745cf | 1834 | mutex_unlock(&fs_info->chunk_mutex); |
15484377 | 1835 | mutex_unlock(&fs_info->fs_devices->device_list_mutex); |
d397712b | 1836 | |
ba7b6e62 DS |
1837 | buf->f_type = BTRFS_SUPER_MAGIC; |
1838 | buf->f_bsize = dentry->d_sb->s_blocksize; | |
1839 | buf->f_namelen = BTRFS_NAME_LEN; | |
1840 | ||
9d03632e | 1841 | /* We treat it as constant endianness (it doesn't matter _which_) |
d397712b | 1842 | because we want the fsid to come out the same whether mounted |
9d03632e DW |
1843 | on a big-endian or little-endian host */ |
1844 | buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]); | |
1845 | buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]); | |
32d48fa1 DW |
1846 | /* Mask in the root object ID too, to disambiguate subvols */ |
1847 | buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32; | |
1848 | buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid; | |
1849 | ||
8fd17795 CM |
1850 | return 0; |
1851 | } | |
b5133862 | 1852 | |
aea52e19 AV |
1853 | static void btrfs_kill_super(struct super_block *sb) |
1854 | { | |
815745cf | 1855 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
aea52e19 | 1856 | kill_anon_super(sb); |
d22ca7de | 1857 | free_fs_info(fs_info); |
aea52e19 AV |
1858 | } |
1859 | ||
2e635a27 CM |
1860 | static struct file_system_type btrfs_fs_type = { |
1861 | .owner = THIS_MODULE, | |
1862 | .name = "btrfs", | |
061dbc6b | 1863 | .mount = btrfs_mount, |
aea52e19 | 1864 | .kill_sb = btrfs_kill_super, |
f667aef6 | 1865 | .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA, |
2e635a27 | 1866 | }; |
7f78e035 | 1867 | MODULE_ALIAS_FS("btrfs"); |
a9218f6b | 1868 | |
d352ac68 CM |
1869 | /* |
1870 | * used by btrfsctl to scan devices when no FS is mounted | |
1871 | */ | |
8a4b83cc CM |
1872 | static long btrfs_control_ioctl(struct file *file, unsigned int cmd, |
1873 | unsigned long arg) | |
1874 | { | |
1875 | struct btrfs_ioctl_vol_args *vol; | |
1876 | struct btrfs_fs_devices *fs_devices; | |
c071fcfd | 1877 | int ret = -ENOTTY; |
8a4b83cc | 1878 | |
e441d54d CM |
1879 | if (!capable(CAP_SYS_ADMIN)) |
1880 | return -EPERM; | |
1881 | ||
dae7b665 LZ |
1882 | vol = memdup_user((void __user *)arg, sizeof(*vol)); |
1883 | if (IS_ERR(vol)) | |
1884 | return PTR_ERR(vol); | |
c071fcfd | 1885 | |
8a4b83cc CM |
1886 | switch (cmd) { |
1887 | case BTRFS_IOC_SCAN_DEV: | |
97288f2c | 1888 | ret = btrfs_scan_one_device(vol->name, FMODE_READ, |
8a4b83cc CM |
1889 | &btrfs_fs_type, &fs_devices); |
1890 | break; | |
02db0844 JB |
1891 | case BTRFS_IOC_DEVICES_READY: |
1892 | ret = btrfs_scan_one_device(vol->name, FMODE_READ, | |
1893 | &btrfs_fs_type, &fs_devices); | |
1894 | if (ret) | |
1895 | break; | |
1896 | ret = !(fs_devices->num_devices == fs_devices->total_devices); | |
1897 | break; | |
8a4b83cc | 1898 | } |
dae7b665 | 1899 | |
8a4b83cc | 1900 | kfree(vol); |
f819d837 | 1901 | return ret; |
8a4b83cc CM |
1902 | } |
1903 | ||
0176260f | 1904 | static int btrfs_freeze(struct super_block *sb) |
ed0dab6b | 1905 | { |
354aa0fb MX |
1906 | struct btrfs_trans_handle *trans; |
1907 | struct btrfs_root *root = btrfs_sb(sb)->tree_root; | |
1908 | ||
d4edf39b | 1909 | trans = btrfs_attach_transaction_barrier(root); |
354aa0fb MX |
1910 | if (IS_ERR(trans)) { |
1911 | /* no transaction, don't bother */ | |
1912 | if (PTR_ERR(trans) == -ENOENT) | |
1913 | return 0; | |
1914 | return PTR_ERR(trans); | |
1915 | } | |
1916 | return btrfs_commit_transaction(trans, root); | |
ed0dab6b Y |
1917 | } |
1918 | ||
0176260f | 1919 | static int btrfs_unfreeze(struct super_block *sb) |
ed0dab6b | 1920 | { |
0176260f | 1921 | return 0; |
ed0dab6b | 1922 | } |
2e635a27 | 1923 | |
9c5085c1 JB |
1924 | static int btrfs_show_devname(struct seq_file *m, struct dentry *root) |
1925 | { | |
1926 | struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb); | |
1927 | struct btrfs_fs_devices *cur_devices; | |
1928 | struct btrfs_device *dev, *first_dev = NULL; | |
1929 | struct list_head *head; | |
1930 | struct rcu_string *name; | |
1931 | ||
1932 | mutex_lock(&fs_info->fs_devices->device_list_mutex); | |
1933 | cur_devices = fs_info->fs_devices; | |
1934 | while (cur_devices) { | |
1935 | head = &cur_devices->devices; | |
1936 | list_for_each_entry(dev, head, dev_list) { | |
aa9ddcd4 JB |
1937 | if (dev->missing) |
1938 | continue; | |
0aeb8a6e AJ |
1939 | if (!dev->name) |
1940 | continue; | |
9c5085c1 JB |
1941 | if (!first_dev || dev->devid < first_dev->devid) |
1942 | first_dev = dev; | |
1943 | } | |
1944 | cur_devices = cur_devices->seed; | |
1945 | } | |
1946 | ||
1947 | if (first_dev) { | |
1948 | rcu_read_lock(); | |
1949 | name = rcu_dereference(first_dev->name); | |
1950 | seq_escape(m, name->str, " \t\n\\"); | |
1951 | rcu_read_unlock(); | |
1952 | } else { | |
1953 | WARN_ON(1); | |
1954 | } | |
1955 | mutex_unlock(&fs_info->fs_devices->device_list_mutex); | |
1956 | return 0; | |
1957 | } | |
1958 | ||
b87221de | 1959 | static const struct super_operations btrfs_super_ops = { |
76dda93c | 1960 | .drop_inode = btrfs_drop_inode, |
bd555975 | 1961 | .evict_inode = btrfs_evict_inode, |
e20d96d6 | 1962 | .put_super = btrfs_put_super, |
d5719762 | 1963 | .sync_fs = btrfs_sync_fs, |
a9572a15 | 1964 | .show_options = btrfs_show_options, |
9c5085c1 | 1965 | .show_devname = btrfs_show_devname, |
4730a4bc | 1966 | .write_inode = btrfs_write_inode, |
2c90e5d6 CM |
1967 | .alloc_inode = btrfs_alloc_inode, |
1968 | .destroy_inode = btrfs_destroy_inode, | |
8fd17795 | 1969 | .statfs = btrfs_statfs, |
c146afad | 1970 | .remount_fs = btrfs_remount, |
0176260f LT |
1971 | .freeze_fs = btrfs_freeze, |
1972 | .unfreeze_fs = btrfs_unfreeze, | |
e20d96d6 | 1973 | }; |
a9218f6b CM |
1974 | |
1975 | static const struct file_operations btrfs_ctl_fops = { | |
1976 | .unlocked_ioctl = btrfs_control_ioctl, | |
1977 | .compat_ioctl = btrfs_control_ioctl, | |
1978 | .owner = THIS_MODULE, | |
6038f373 | 1979 | .llseek = noop_llseek, |
a9218f6b CM |
1980 | }; |
1981 | ||
1982 | static struct miscdevice btrfs_misc = { | |
578454ff | 1983 | .minor = BTRFS_MINOR, |
a9218f6b CM |
1984 | .name = "btrfs-control", |
1985 | .fops = &btrfs_ctl_fops | |
1986 | }; | |
1987 | ||
578454ff KS |
1988 | MODULE_ALIAS_MISCDEV(BTRFS_MINOR); |
1989 | MODULE_ALIAS("devname:btrfs-control"); | |
1990 | ||
a9218f6b CM |
1991 | static int btrfs_interface_init(void) |
1992 | { | |
1993 | return misc_register(&btrfs_misc); | |
1994 | } | |
1995 | ||
b2950863 | 1996 | static void btrfs_interface_exit(void) |
a9218f6b CM |
1997 | { |
1998 | if (misc_deregister(&btrfs_misc) < 0) | |
efe120a0 | 1999 | printk(KERN_INFO "BTRFS: misc_deregister failed for control device\n"); |
a9218f6b CM |
2000 | } |
2001 | ||
85965600 DS |
2002 | static void btrfs_print_info(void) |
2003 | { | |
2004 | printk(KERN_INFO "Btrfs loaded" | |
2005 | #ifdef CONFIG_BTRFS_DEBUG | |
2006 | ", debug=on" | |
2007 | #endif | |
79556c3d SB |
2008 | #ifdef CONFIG_BTRFS_ASSERT |
2009 | ", assert=on" | |
2010 | #endif | |
85965600 DS |
2011 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
2012 | ", integrity-checker=on" | |
2013 | #endif | |
2014 | "\n"); | |
2015 | } | |
2016 | ||
dc11dd5d JB |
2017 | static int btrfs_run_sanity_tests(void) |
2018 | { | |
06ea65a3 JB |
2019 | int ret; |
2020 | ||
294e30fe | 2021 | ret = btrfs_init_test_fs(); |
06ea65a3 JB |
2022 | if (ret) |
2023 | return ret; | |
294e30fe JB |
2024 | |
2025 | ret = btrfs_test_free_space_cache(); | |
2026 | if (ret) | |
2027 | goto out; | |
2028 | ret = btrfs_test_extent_buffer_operations(); | |
2029 | if (ret) | |
2030 | goto out; | |
2031 | ret = btrfs_test_extent_io(); | |
aaedb55b JB |
2032 | if (ret) |
2033 | goto out; | |
2034 | ret = btrfs_test_inodes(); | |
faa2dbf0 JB |
2035 | if (ret) |
2036 | goto out; | |
2037 | ret = btrfs_test_qgroups(); | |
294e30fe JB |
2038 | out: |
2039 | btrfs_destroy_test_fs(); | |
2040 | return ret; | |
dc11dd5d JB |
2041 | } |
2042 | ||
2e635a27 CM |
2043 | static int __init init_btrfs_fs(void) |
2044 | { | |
2c90e5d6 | 2045 | int err; |
58176a96 | 2046 | |
14a958e6 FDBM |
2047 | err = btrfs_hash_init(); |
2048 | if (err) | |
2049 | return err; | |
2050 | ||
63541927 FDBM |
2051 | btrfs_props_init(); |
2052 | ||
58176a96 JB |
2053 | err = btrfs_init_sysfs(); |
2054 | if (err) | |
14a958e6 | 2055 | goto free_hash; |
58176a96 | 2056 | |
143bede5 | 2057 | btrfs_init_compress(); |
d1310b2e | 2058 | |
261507a0 LZ |
2059 | err = btrfs_init_cachep(); |
2060 | if (err) | |
2061 | goto free_compress; | |
2062 | ||
d1310b2e | 2063 | err = extent_io_init(); |
2f4cbe64 WB |
2064 | if (err) |
2065 | goto free_cachep; | |
2066 | ||
d1310b2e CM |
2067 | err = extent_map_init(); |
2068 | if (err) | |
2069 | goto free_extent_io; | |
2070 | ||
6352b91d | 2071 | err = ordered_data_init(); |
2f4cbe64 WB |
2072 | if (err) |
2073 | goto free_extent_map; | |
c8b97818 | 2074 | |
6352b91d MX |
2075 | err = btrfs_delayed_inode_init(); |
2076 | if (err) | |
2077 | goto free_ordered_data; | |
2078 | ||
9247f317 | 2079 | err = btrfs_auto_defrag_init(); |
16cdcec7 MX |
2080 | if (err) |
2081 | goto free_delayed_inode; | |
2082 | ||
78a6184a | 2083 | err = btrfs_delayed_ref_init(); |
9247f317 MX |
2084 | if (err) |
2085 | goto free_auto_defrag; | |
2086 | ||
b9e9a6cb WS |
2087 | err = btrfs_prelim_ref_init(); |
2088 | if (err) | |
af13b492 | 2089 | goto free_delayed_ref; |
b9e9a6cb | 2090 | |
97eb6b69 | 2091 | err = btrfs_end_io_wq_init(); |
78a6184a | 2092 | if (err) |
af13b492 | 2093 | goto free_prelim_ref; |
78a6184a | 2094 | |
97eb6b69 DS |
2095 | err = btrfs_interface_init(); |
2096 | if (err) | |
2097 | goto free_end_io_wq; | |
2098 | ||
e565d4b9 JS |
2099 | btrfs_init_lockdep(); |
2100 | ||
85965600 | 2101 | btrfs_print_info(); |
dc11dd5d JB |
2102 | |
2103 | err = btrfs_run_sanity_tests(); | |
2104 | if (err) | |
2105 | goto unregister_ioctl; | |
2106 | ||
2107 | err = register_filesystem(&btrfs_fs_type); | |
2108 | if (err) | |
2109 | goto unregister_ioctl; | |
74255aa0 | 2110 | |
2f4cbe64 WB |
2111 | return 0; |
2112 | ||
a9218f6b CM |
2113 | unregister_ioctl: |
2114 | btrfs_interface_exit(); | |
97eb6b69 DS |
2115 | free_end_io_wq: |
2116 | btrfs_end_io_wq_exit(); | |
b9e9a6cb WS |
2117 | free_prelim_ref: |
2118 | btrfs_prelim_ref_exit(); | |
78a6184a MX |
2119 | free_delayed_ref: |
2120 | btrfs_delayed_ref_exit(); | |
9247f317 MX |
2121 | free_auto_defrag: |
2122 | btrfs_auto_defrag_exit(); | |
16cdcec7 MX |
2123 | free_delayed_inode: |
2124 | btrfs_delayed_inode_exit(); | |
6352b91d MX |
2125 | free_ordered_data: |
2126 | ordered_data_exit(); | |
2f4cbe64 WB |
2127 | free_extent_map: |
2128 | extent_map_exit(); | |
d1310b2e CM |
2129 | free_extent_io: |
2130 | extent_io_exit(); | |
2f4cbe64 WB |
2131 | free_cachep: |
2132 | btrfs_destroy_cachep(); | |
261507a0 LZ |
2133 | free_compress: |
2134 | btrfs_exit_compress(); | |
2f4cbe64 | 2135 | btrfs_exit_sysfs(); |
14a958e6 FDBM |
2136 | free_hash: |
2137 | btrfs_hash_exit(); | |
2f4cbe64 | 2138 | return err; |
2e635a27 CM |
2139 | } |
2140 | ||
2141 | static void __exit exit_btrfs_fs(void) | |
2142 | { | |
39279cc3 | 2143 | btrfs_destroy_cachep(); |
78a6184a | 2144 | btrfs_delayed_ref_exit(); |
9247f317 | 2145 | btrfs_auto_defrag_exit(); |
16cdcec7 | 2146 | btrfs_delayed_inode_exit(); |
b9e9a6cb | 2147 | btrfs_prelim_ref_exit(); |
6352b91d | 2148 | ordered_data_exit(); |
a52d9a80 | 2149 | extent_map_exit(); |
d1310b2e | 2150 | extent_io_exit(); |
a9218f6b | 2151 | btrfs_interface_exit(); |
2e635a27 | 2152 | unregister_filesystem(&btrfs_fs_type); |
58176a96 | 2153 | btrfs_exit_sysfs(); |
8a4b83cc | 2154 | btrfs_cleanup_fs_uuids(); |
261507a0 | 2155 | btrfs_exit_compress(); |
14a958e6 | 2156 | btrfs_hash_exit(); |
2e635a27 CM |
2157 | } |
2158 | ||
60efa5eb | 2159 | late_initcall(init_btrfs_fs); |
2e635a27 CM |
2160 | module_exit(exit_btrfs_fs) |
2161 | ||
2162 | MODULE_LICENSE("GPL"); |