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> |
4b4e25f2 | 45 | #include "compat.h" |
16cdcec7 | 46 | #include "delayed-inode.h" |
2e635a27 | 47 | #include "ctree.h" |
e20d96d6 | 48 | #include "disk-io.h" |
d5719762 | 49 | #include "transaction.h" |
2c90e5d6 | 50 | #include "btrfs_inode.h" |
3a686375 | 51 | #include "print-tree.h" |
5103e947 | 52 | #include "xattr.h" |
8a4b83cc | 53 | #include "volumes.h" |
b3c3da71 | 54 | #include "version.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" |
2e635a27 | 60 | |
1abe9b8a | 61 | #define CREATE_TRACE_POINTS |
62 | #include <trace/events/btrfs.h> | |
63 | ||
b87221de | 64 | static const struct super_operations btrfs_super_ops; |
830c4adb | 65 | static struct file_system_type btrfs_fs_type; |
75dfe396 | 66 | |
08748810 | 67 | static 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 | 95 | static 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 */ |
105 | static 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 | */ | |
133 | void __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 | |
08748810 DS |
155 | printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: errno=%d %s (%pV)\n", |
156 | sb->s_id, function, line, errno, errstr, &vaf); | |
37252a66 | 157 | va_end(args); |
4da35113 | 158 | } else { |
08748810 DS |
159 | printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: errno=%d %s\n", |
160 | sb->s_id, function, line, errno, errstr); | |
4da35113 | 161 | } |
acce952b | 162 | |
4da35113 | 163 | /* Don't go through full error handling during mount */ |
cf79ffb5 JB |
164 | save_error_info(fs_info); |
165 | if (sb->s_flags & MS_BORN) | |
4da35113 | 166 | btrfs_handle_error(fs_info); |
4da35113 | 167 | } |
acce952b | 168 | |
533574c6 | 169 | static const char * const logtypes[] = { |
4da35113 JM |
170 | "emergency", |
171 | "alert", | |
172 | "critical", | |
173 | "error", | |
174 | "warning", | |
175 | "notice", | |
176 | "info", | |
177 | "debug", | |
178 | }; | |
179 | ||
c2cf52eb | 180 | void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...) |
4da35113 JM |
181 | { |
182 | struct super_block *sb = fs_info->sb; | |
183 | char lvl[4]; | |
184 | struct va_format vaf; | |
185 | va_list args; | |
186 | const char *type = logtypes[4]; | |
533574c6 | 187 | int kern_level; |
4da35113 JM |
188 | |
189 | va_start(args, fmt); | |
190 | ||
533574c6 JP |
191 | kern_level = printk_get_level(fmt); |
192 | if (kern_level) { | |
193 | size_t size = printk_skip_level(fmt) - fmt; | |
194 | memcpy(lvl, fmt, size); | |
195 | lvl[size] = '\0'; | |
196 | fmt += size; | |
197 | type = logtypes[kern_level - '0']; | |
4da35113 JM |
198 | } else |
199 | *lvl = '\0'; | |
200 | ||
201 | vaf.fmt = fmt; | |
202 | vaf.va = &args; | |
533574c6 | 203 | |
c2cf52eb | 204 | printk("%sBTRFS %s (device %s): %pV\n", lvl, type, sb->s_id, &vaf); |
533574c6 JP |
205 | |
206 | va_end(args); | |
207 | } | |
208 | ||
209 | #else | |
210 | ||
211 | void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function, | |
212 | unsigned int line, int errno, const char *fmt, ...) | |
213 | { | |
214 | struct super_block *sb = fs_info->sb; | |
215 | ||
216 | /* | |
217 | * Special case: if the error is EROFS, and we're already | |
218 | * under MS_RDONLY, then it is safe here. | |
219 | */ | |
220 | if (errno == -EROFS && (sb->s_flags & MS_RDONLY)) | |
221 | return; | |
222 | ||
223 | /* Don't go through full error handling during mount */ | |
224 | if (sb->s_flags & MS_BORN) { | |
225 | save_error_info(fs_info); | |
226 | btrfs_handle_error(fs_info); | |
227 | } | |
acce952b | 228 | } |
533574c6 | 229 | #endif |
acce952b | 230 | |
49b25e05 JM |
231 | /* |
232 | * We only mark the transaction aborted and then set the file system read-only. | |
233 | * This will prevent new transactions from starting or trying to join this | |
234 | * one. | |
235 | * | |
236 | * This means that error recovery at the call site is limited to freeing | |
237 | * any local memory allocations and passing the error code up without | |
238 | * further cleanup. The transaction should complete as it normally would | |
239 | * in the call path but will return -EIO. | |
240 | * | |
241 | * We'll complete the cleanup in btrfs_end_transaction and | |
242 | * btrfs_commit_transaction. | |
243 | */ | |
244 | void __btrfs_abort_transaction(struct btrfs_trans_handle *trans, | |
245 | struct btrfs_root *root, const char *function, | |
246 | unsigned int line, int errno) | |
247 | { | |
08748810 DS |
248 | /* |
249 | * Report first abort since mount | |
250 | */ | |
251 | if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED, | |
252 | &root->fs_info->fs_state)) { | |
253 | WARN(1, KERN_DEBUG "btrfs: Transaction aborted (error %d)\n", | |
254 | errno); | |
255 | } | |
49b25e05 JM |
256 | trans->aborted = errno; |
257 | /* Nothing used. The other threads that have joined this | |
258 | * transaction may be able to continue. */ | |
259 | if (!trans->blocks_used) { | |
69ce977a MX |
260 | const char *errstr; |
261 | ||
08748810 | 262 | errstr = btrfs_decode_error(errno); |
c2cf52eb SK |
263 | btrfs_warn(root->fs_info, |
264 | "%s:%d: Aborting unused transaction(%s).", | |
265 | function, line, errstr); | |
acce952b | 266 | return; |
49b25e05 | 267 | } |
8d25a086 | 268 | ACCESS_ONCE(trans->transaction->aborted) = errno; |
49b25e05 JM |
269 | __btrfs_std_error(root->fs_info, function, line, errno, NULL); |
270 | } | |
8c342930 JM |
271 | /* |
272 | * __btrfs_panic decodes unexpected, fatal errors from the caller, | |
273 | * issues an alert, and either panics or BUGs, depending on mount options. | |
274 | */ | |
275 | void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function, | |
276 | unsigned int line, int errno, const char *fmt, ...) | |
277 | { | |
8c342930 JM |
278 | char *s_id = "<unknown>"; |
279 | const char *errstr; | |
280 | struct va_format vaf = { .fmt = fmt }; | |
281 | va_list args; | |
acce952b | 282 | |
8c342930 JM |
283 | if (fs_info) |
284 | s_id = fs_info->sb->s_id; | |
acce952b | 285 | |
8c342930 JM |
286 | va_start(args, fmt); |
287 | vaf.va = &args; | |
288 | ||
08748810 | 289 | errstr = btrfs_decode_error(errno); |
aa43a17c | 290 | if (fs_info && (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR)) |
08748810 DS |
291 | panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n", |
292 | s_id, function, line, &vaf, errno, errstr); | |
8c342930 | 293 | |
08748810 DS |
294 | printk(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n", |
295 | s_id, function, line, &vaf, errno, errstr); | |
8c342930 JM |
296 | va_end(args); |
297 | /* Caller calls BUG() */ | |
acce952b | 298 | } |
299 | ||
d397712b | 300 | static void btrfs_put_super(struct super_block *sb) |
b18c6685 | 301 | { |
815745cf | 302 | (void)close_ctree(btrfs_sb(sb)->tree_root); |
aea52e19 AV |
303 | /* FIXME: need to fix VFS to return error? */ |
304 | /* AV: return it _where_? ->put_super() can be triggered by any number | |
305 | * of async events, up to and including delivery of SIGKILL to the | |
306 | * last process that kept it busy. Or segfault in the aforementioned | |
307 | * process... Whom would you report that to? | |
308 | */ | |
75dfe396 CM |
309 | } |
310 | ||
95e05289 | 311 | enum { |
73f73415 | 312 | Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum, |
287a0ab9 JB |
313 | Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd, |
314 | Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress, | |
261507a0 LZ |
315 | Opt_compress_type, Opt_compress_force, Opt_compress_force_type, |
316 | Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_discard, | |
91435650 | 317 | Opt_space_cache, Opt_clear_cache, Opt_user_subvol_rm_allowed, |
9555c6c1 ID |
318 | Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache, |
319 | Opt_no_space_cache, Opt_recovery, Opt_skip_balance, | |
21adbd5c | 320 | Opt_check_integrity, Opt_check_integrity_including_extent_data, |
8c342930 | 321 | Opt_check_integrity_print_mask, Opt_fatal_errors, |
9555c6c1 | 322 | Opt_err, |
95e05289 CM |
323 | }; |
324 | ||
325 | static match_table_t tokens = { | |
dfe25020 | 326 | {Opt_degraded, "degraded"}, |
95e05289 | 327 | {Opt_subvol, "subvol=%s"}, |
73f73415 | 328 | {Opt_subvolid, "subvolid=%d"}, |
43e570b0 | 329 | {Opt_device, "device=%s"}, |
b6cda9bc | 330 | {Opt_nodatasum, "nodatasum"}, |
be20aa9d | 331 | {Opt_nodatacow, "nodatacow"}, |
21ad10cf | 332 | {Opt_nobarrier, "nobarrier"}, |
6f568d35 | 333 | {Opt_max_inline, "max_inline=%s"}, |
8f662a76 | 334 | {Opt_alloc_start, "alloc_start=%s"}, |
4543df7e | 335 | {Opt_thread_pool, "thread_pool=%d"}, |
c8b97818 | 336 | {Opt_compress, "compress"}, |
261507a0 | 337 | {Opt_compress_type, "compress=%s"}, |
a555f810 | 338 | {Opt_compress_force, "compress-force"}, |
261507a0 | 339 | {Opt_compress_force_type, "compress-force=%s"}, |
e18e4809 | 340 | {Opt_ssd, "ssd"}, |
451d7585 | 341 | {Opt_ssd_spread, "ssd_spread"}, |
3b30c22f | 342 | {Opt_nossd, "nossd"}, |
33268eaf | 343 | {Opt_noacl, "noacl"}, |
3a5e1404 | 344 | {Opt_notreelog, "notreelog"}, |
dccae999 | 345 | {Opt_flushoncommit, "flushoncommit"}, |
97e728d4 | 346 | {Opt_ratio, "metadata_ratio=%d"}, |
e244a0ae | 347 | {Opt_discard, "discard"}, |
0af3d00b | 348 | {Opt_space_cache, "space_cache"}, |
88c2ba3b | 349 | {Opt_clear_cache, "clear_cache"}, |
4260f7c7 | 350 | {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"}, |
91435650 | 351 | {Opt_enospc_debug, "enospc_debug"}, |
e15d0542 | 352 | {Opt_subvolrootid, "subvolrootid=%d"}, |
4cb5300b | 353 | {Opt_defrag, "autodefrag"}, |
4b9465cb | 354 | {Opt_inode_cache, "inode_cache"}, |
8965593e | 355 | {Opt_no_space_cache, "nospace_cache"}, |
af31f5e5 | 356 | {Opt_recovery, "recovery"}, |
9555c6c1 | 357 | {Opt_skip_balance, "skip_balance"}, |
21adbd5c SB |
358 | {Opt_check_integrity, "check_int"}, |
359 | {Opt_check_integrity_including_extent_data, "check_int_data"}, | |
360 | {Opt_check_integrity_print_mask, "check_int_print_mask=%d"}, | |
8c342930 | 361 | {Opt_fatal_errors, "fatal_errors=%s"}, |
33268eaf | 362 | {Opt_err, NULL}, |
95e05289 CM |
363 | }; |
364 | ||
edf24abe CH |
365 | /* |
366 | * Regular mount options parser. Everything that is needed only when | |
367 | * reading in a new superblock is parsed here. | |
49b25e05 | 368 | * XXX JDM: This needs to be cleaned up for remount. |
edf24abe CH |
369 | */ |
370 | int btrfs_parse_options(struct btrfs_root *root, char *options) | |
95e05289 | 371 | { |
edf24abe | 372 | struct btrfs_fs_info *info = root->fs_info; |
95e05289 | 373 | substring_t args[MAX_OPT_ARGS]; |
73bc1876 JB |
374 | char *p, *num, *orig = NULL; |
375 | u64 cache_gen; | |
4543df7e | 376 | int intarg; |
a7a3f7ca | 377 | int ret = 0; |
261507a0 LZ |
378 | char *compress_type; |
379 | bool compress_force = false; | |
b6cda9bc | 380 | |
6c41761f | 381 | cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy); |
73bc1876 JB |
382 | if (cache_gen) |
383 | btrfs_set_opt(info->mount_opt, SPACE_CACHE); | |
384 | ||
95e05289 | 385 | if (!options) |
73bc1876 | 386 | goto out; |
95e05289 | 387 | |
be20aa9d CM |
388 | /* |
389 | * strsep changes the string, duplicate it because parse_options | |
390 | * gets called twice | |
391 | */ | |
392 | options = kstrdup(options, GFP_NOFS); | |
393 | if (!options) | |
394 | return -ENOMEM; | |
395 | ||
da495ecc | 396 | orig = options; |
be20aa9d | 397 | |
edf24abe | 398 | while ((p = strsep(&options, ",")) != NULL) { |
95e05289 CM |
399 | int token; |
400 | if (!*p) | |
401 | continue; | |
402 | ||
403 | token = match_token(p, tokens, args); | |
404 | switch (token) { | |
dfe25020 | 405 | case Opt_degraded: |
edf24abe CH |
406 | printk(KERN_INFO "btrfs: allowing degraded mounts\n"); |
407 | btrfs_set_opt(info->mount_opt, DEGRADED); | |
dfe25020 | 408 | break; |
95e05289 | 409 | case Opt_subvol: |
73f73415 | 410 | case Opt_subvolid: |
e15d0542 | 411 | case Opt_subvolrootid: |
43e570b0 | 412 | case Opt_device: |
edf24abe | 413 | /* |
43e570b0 | 414 | * These are parsed by btrfs_parse_early_options |
edf24abe CH |
415 | * and can be happily ignored here. |
416 | */ | |
b6cda9bc CM |
417 | break; |
418 | case Opt_nodatasum: | |
067c28ad | 419 | printk(KERN_INFO "btrfs: setting nodatasum\n"); |
edf24abe | 420 | btrfs_set_opt(info->mount_opt, NODATASUM); |
be20aa9d CM |
421 | break; |
422 | case Opt_nodatacow: | |
bedb2cca AP |
423 | if (!btrfs_test_opt(root, COMPRESS) || |
424 | !btrfs_test_opt(root, FORCE_COMPRESS)) { | |
425 | printk(KERN_INFO "btrfs: setting nodatacow, compression disabled\n"); | |
426 | } else { | |
427 | printk(KERN_INFO "btrfs: setting nodatacow\n"); | |
428 | } | |
429 | info->compress_type = BTRFS_COMPRESS_NONE; | |
430 | btrfs_clear_opt(info->mount_opt, COMPRESS); | |
431 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
edf24abe CH |
432 | btrfs_set_opt(info->mount_opt, NODATACOW); |
433 | btrfs_set_opt(info->mount_opt, NODATASUM); | |
95e05289 | 434 | break; |
a555f810 | 435 | case Opt_compress_force: |
261507a0 LZ |
436 | case Opt_compress_force_type: |
437 | compress_force = true; | |
1c697d4a | 438 | /* Fallthrough */ |
261507a0 LZ |
439 | case Opt_compress: |
440 | case Opt_compress_type: | |
441 | if (token == Opt_compress || | |
442 | token == Opt_compress_force || | |
443 | strcmp(args[0].from, "zlib") == 0) { | |
444 | compress_type = "zlib"; | |
445 | info->compress_type = BTRFS_COMPRESS_ZLIB; | |
063849ea | 446 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
447 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
448 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
a6fa6fae LZ |
449 | } else if (strcmp(args[0].from, "lzo") == 0) { |
450 | compress_type = "lzo"; | |
451 | info->compress_type = BTRFS_COMPRESS_LZO; | |
063849ea | 452 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
453 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
454 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
2b0ce2c2 | 455 | btrfs_set_fs_incompat(info, COMPRESS_LZO); |
063849ea AH |
456 | } else if (strncmp(args[0].from, "no", 2) == 0) { |
457 | compress_type = "no"; | |
458 | info->compress_type = BTRFS_COMPRESS_NONE; | |
459 | btrfs_clear_opt(info->mount_opt, COMPRESS); | |
460 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
461 | compress_force = false; | |
261507a0 LZ |
462 | } else { |
463 | ret = -EINVAL; | |
464 | goto out; | |
465 | } | |
466 | ||
261507a0 LZ |
467 | if (compress_force) { |
468 | btrfs_set_opt(info->mount_opt, FORCE_COMPRESS); | |
469 | pr_info("btrfs: force %s compression\n", | |
470 | compress_type); | |
471 | } else | |
472 | pr_info("btrfs: use %s compression\n", | |
473 | compress_type); | |
a555f810 | 474 | break; |
e18e4809 | 475 | case Opt_ssd: |
edf24abe CH |
476 | printk(KERN_INFO "btrfs: use ssd allocation scheme\n"); |
477 | btrfs_set_opt(info->mount_opt, SSD); | |
e18e4809 | 478 | break; |
451d7585 CM |
479 | case Opt_ssd_spread: |
480 | printk(KERN_INFO "btrfs: use spread ssd " | |
481 | "allocation scheme\n"); | |
482 | btrfs_set_opt(info->mount_opt, SSD); | |
483 | btrfs_set_opt(info->mount_opt, SSD_SPREAD); | |
484 | break; | |
3b30c22f | 485 | case Opt_nossd: |
451d7585 CM |
486 | printk(KERN_INFO "btrfs: not using ssd allocation " |
487 | "scheme\n"); | |
c289811c | 488 | btrfs_set_opt(info->mount_opt, NOSSD); |
3b30c22f | 489 | btrfs_clear_opt(info->mount_opt, SSD); |
451d7585 | 490 | btrfs_clear_opt(info->mount_opt, SSD_SPREAD); |
3b30c22f | 491 | break; |
21ad10cf | 492 | case Opt_nobarrier: |
edf24abe CH |
493 | printk(KERN_INFO "btrfs: turning off barriers\n"); |
494 | btrfs_set_opt(info->mount_opt, NOBARRIER); | |
21ad10cf | 495 | break; |
4543df7e CM |
496 | case Opt_thread_pool: |
497 | intarg = 0; | |
498 | match_int(&args[0], &intarg); | |
0d2450ab | 499 | if (intarg) |
4543df7e | 500 | info->thread_pool_size = intarg; |
4543df7e | 501 | break; |
6f568d35 | 502 | case Opt_max_inline: |
edf24abe CH |
503 | num = match_strdup(&args[0]); |
504 | if (num) { | |
91748467 | 505 | info->max_inline = memparse(num, NULL); |
edf24abe CH |
506 | kfree(num); |
507 | ||
15ada040 CM |
508 | if (info->max_inline) { |
509 | info->max_inline = max_t(u64, | |
510 | info->max_inline, | |
511 | root->sectorsize); | |
512 | } | |
edf24abe | 513 | printk(KERN_INFO "btrfs: max_inline at %llu\n", |
21380931 | 514 | (unsigned long long)info->max_inline); |
6f568d35 CM |
515 | } |
516 | break; | |
8f662a76 | 517 | case Opt_alloc_start: |
edf24abe CH |
518 | num = match_strdup(&args[0]); |
519 | if (num) { | |
c018daec | 520 | mutex_lock(&info->chunk_mutex); |
91748467 | 521 | info->alloc_start = memparse(num, NULL); |
c018daec | 522 | mutex_unlock(&info->chunk_mutex); |
edf24abe CH |
523 | kfree(num); |
524 | printk(KERN_INFO | |
525 | "btrfs: allocations start at %llu\n", | |
21380931 | 526 | (unsigned long long)info->alloc_start); |
8f662a76 CM |
527 | } |
528 | break; | |
33268eaf JB |
529 | case Opt_noacl: |
530 | root->fs_info->sb->s_flags &= ~MS_POSIXACL; | |
531 | break; | |
3a5e1404 SW |
532 | case Opt_notreelog: |
533 | printk(KERN_INFO "btrfs: disabling tree log\n"); | |
534 | btrfs_set_opt(info->mount_opt, NOTREELOG); | |
535 | break; | |
dccae999 SW |
536 | case Opt_flushoncommit: |
537 | printk(KERN_INFO "btrfs: turning on flush-on-commit\n"); | |
538 | btrfs_set_opt(info->mount_opt, FLUSHONCOMMIT); | |
539 | break; | |
97e728d4 JB |
540 | case Opt_ratio: |
541 | intarg = 0; | |
542 | match_int(&args[0], &intarg); | |
543 | if (intarg) { | |
544 | info->metadata_ratio = intarg; | |
545 | printk(KERN_INFO "btrfs: metadata ratio %d\n", | |
546 | info->metadata_ratio); | |
547 | } | |
548 | break; | |
e244a0ae CH |
549 | case Opt_discard: |
550 | btrfs_set_opt(info->mount_opt, DISCARD); | |
551 | break; | |
0af3d00b | 552 | case Opt_space_cache: |
0af3d00b | 553 | btrfs_set_opt(info->mount_opt, SPACE_CACHE); |
0de90876 | 554 | break; |
73bc1876 JB |
555 | case Opt_no_space_cache: |
556 | printk(KERN_INFO "btrfs: disabling disk space caching\n"); | |
557 | btrfs_clear_opt(info->mount_opt, SPACE_CACHE); | |
558 | break; | |
4b9465cb CM |
559 | case Opt_inode_cache: |
560 | printk(KERN_INFO "btrfs: enabling inode map caching\n"); | |
561 | btrfs_set_opt(info->mount_opt, INODE_MAP_CACHE); | |
562 | break; | |
88c2ba3b JB |
563 | case Opt_clear_cache: |
564 | printk(KERN_INFO "btrfs: force clearing of disk cache\n"); | |
565 | btrfs_set_opt(info->mount_opt, CLEAR_CACHE); | |
0af3d00b | 566 | break; |
4260f7c7 SW |
567 | case Opt_user_subvol_rm_allowed: |
568 | btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED); | |
569 | break; | |
91435650 CM |
570 | case Opt_enospc_debug: |
571 | btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG); | |
572 | break; | |
4cb5300b | 573 | case Opt_defrag: |
48940662 | 574 | printk(KERN_INFO "btrfs: enabling auto defrag\n"); |
4cb5300b CM |
575 | btrfs_set_opt(info->mount_opt, AUTO_DEFRAG); |
576 | break; | |
af31f5e5 | 577 | case Opt_recovery: |
48940662 | 578 | printk(KERN_INFO "btrfs: enabling auto recovery\n"); |
af31f5e5 CM |
579 | btrfs_set_opt(info->mount_opt, RECOVERY); |
580 | break; | |
9555c6c1 ID |
581 | case Opt_skip_balance: |
582 | btrfs_set_opt(info->mount_opt, SKIP_BALANCE); | |
583 | break; | |
21adbd5c SB |
584 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
585 | case Opt_check_integrity_including_extent_data: | |
586 | printk(KERN_INFO "btrfs: enabling check integrity" | |
587 | " including extent data\n"); | |
588 | btrfs_set_opt(info->mount_opt, | |
589 | CHECK_INTEGRITY_INCLUDING_EXTENT_DATA); | |
590 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); | |
591 | break; | |
592 | case Opt_check_integrity: | |
593 | printk(KERN_INFO "btrfs: enabling check integrity\n"); | |
594 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); | |
595 | break; | |
596 | case Opt_check_integrity_print_mask: | |
597 | intarg = 0; | |
598 | match_int(&args[0], &intarg); | |
599 | if (intarg) { | |
600 | info->check_integrity_print_mask = intarg; | |
601 | printk(KERN_INFO "btrfs:" | |
602 | " check_integrity_print_mask 0x%x\n", | |
603 | info->check_integrity_print_mask); | |
604 | } | |
605 | break; | |
606 | #else | |
607 | case Opt_check_integrity_including_extent_data: | |
608 | case Opt_check_integrity: | |
609 | case Opt_check_integrity_print_mask: | |
610 | printk(KERN_ERR "btrfs: support for check_integrity*" | |
611 | " not compiled in!\n"); | |
612 | ret = -EINVAL; | |
613 | goto out; | |
614 | #endif | |
8c342930 JM |
615 | case Opt_fatal_errors: |
616 | if (strcmp(args[0].from, "panic") == 0) | |
617 | btrfs_set_opt(info->mount_opt, | |
618 | PANIC_ON_FATAL_ERROR); | |
619 | else if (strcmp(args[0].from, "bug") == 0) | |
620 | btrfs_clear_opt(info->mount_opt, | |
621 | PANIC_ON_FATAL_ERROR); | |
622 | else { | |
623 | ret = -EINVAL; | |
624 | goto out; | |
625 | } | |
626 | break; | |
a7a3f7ca SW |
627 | case Opt_err: |
628 | printk(KERN_INFO "btrfs: unrecognized mount option " | |
629 | "'%s'\n", p); | |
630 | ret = -EINVAL; | |
631 | goto out; | |
95e05289 | 632 | default: |
be20aa9d | 633 | break; |
95e05289 CM |
634 | } |
635 | } | |
a7a3f7ca | 636 | out: |
73bc1876 JB |
637 | if (!ret && btrfs_test_opt(root, SPACE_CACHE)) |
638 | printk(KERN_INFO "btrfs: disk space caching is enabled\n"); | |
da495ecc | 639 | kfree(orig); |
a7a3f7ca | 640 | return ret; |
edf24abe CH |
641 | } |
642 | ||
643 | /* | |
644 | * Parse mount options that are required early in the mount process. | |
645 | * | |
646 | * All other options will be parsed on much later in the mount process and | |
647 | * only when we need to allocate a new super block. | |
648 | */ | |
97288f2c | 649 | static int btrfs_parse_early_options(const char *options, fmode_t flags, |
73f73415 | 650 | void *holder, char **subvol_name, u64 *subvol_objectid, |
5e2a4b25 | 651 | struct btrfs_fs_devices **fs_devices) |
edf24abe CH |
652 | { |
653 | substring_t args[MAX_OPT_ARGS]; | |
83c8c9bd | 654 | char *device_name, *opts, *orig, *p; |
edf24abe | 655 | int error = 0; |
73f73415 | 656 | int intarg; |
edf24abe CH |
657 | |
658 | if (!options) | |
830c4adb | 659 | return 0; |
edf24abe CH |
660 | |
661 | /* | |
662 | * strsep changes the string, duplicate it because parse_options | |
663 | * gets called twice | |
664 | */ | |
665 | opts = kstrdup(options, GFP_KERNEL); | |
666 | if (!opts) | |
667 | return -ENOMEM; | |
3f3d0bc0 | 668 | orig = opts; |
edf24abe CH |
669 | |
670 | while ((p = strsep(&opts, ",")) != NULL) { | |
671 | int token; | |
672 | if (!*p) | |
673 | continue; | |
674 | ||
675 | token = match_token(p, tokens, args); | |
676 | switch (token) { | |
677 | case Opt_subvol: | |
a90e8b6f | 678 | kfree(*subvol_name); |
edf24abe CH |
679 | *subvol_name = match_strdup(&args[0]); |
680 | break; | |
73f73415 JB |
681 | case Opt_subvolid: |
682 | intarg = 0; | |
4849f01d JB |
683 | error = match_int(&args[0], &intarg); |
684 | if (!error) { | |
685 | /* we want the original fs_tree */ | |
686 | if (!intarg) | |
687 | *subvol_objectid = | |
688 | BTRFS_FS_TREE_OBJECTID; | |
689 | else | |
690 | *subvol_objectid = intarg; | |
691 | } | |
73f73415 | 692 | break; |
e15d0542 | 693 | case Opt_subvolrootid: |
5e2a4b25 DS |
694 | printk(KERN_WARNING |
695 | "btrfs: 'subvolrootid' mount option is deprecated and has no effect\n"); | |
e15d0542 | 696 | break; |
43e570b0 | 697 | case Opt_device: |
83c8c9bd JL |
698 | device_name = match_strdup(&args[0]); |
699 | if (!device_name) { | |
700 | error = -ENOMEM; | |
701 | goto out; | |
702 | } | |
703 | error = btrfs_scan_one_device(device_name, | |
43e570b0 | 704 | flags, holder, fs_devices); |
83c8c9bd | 705 | kfree(device_name); |
43e570b0 | 706 | if (error) |
830c4adb | 707 | goto out; |
43e570b0 | 708 | break; |
edf24abe CH |
709 | default: |
710 | break; | |
711 | } | |
712 | } | |
713 | ||
830c4adb | 714 | out: |
3f3d0bc0 | 715 | kfree(orig); |
edf24abe | 716 | return error; |
95e05289 CM |
717 | } |
718 | ||
73f73415 JB |
719 | static struct dentry *get_default_root(struct super_block *sb, |
720 | u64 subvol_objectid) | |
721 | { | |
815745cf AV |
722 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
723 | struct btrfs_root *root = fs_info->tree_root; | |
73f73415 JB |
724 | struct btrfs_root *new_root; |
725 | struct btrfs_dir_item *di; | |
726 | struct btrfs_path *path; | |
727 | struct btrfs_key location; | |
728 | struct inode *inode; | |
73f73415 JB |
729 | u64 dir_id; |
730 | int new = 0; | |
731 | ||
732 | /* | |
733 | * We have a specific subvol we want to mount, just setup location and | |
734 | * go look up the root. | |
735 | */ | |
736 | if (subvol_objectid) { | |
737 | location.objectid = subvol_objectid; | |
738 | location.type = BTRFS_ROOT_ITEM_KEY; | |
739 | location.offset = (u64)-1; | |
740 | goto find_root; | |
741 | } | |
742 | ||
743 | path = btrfs_alloc_path(); | |
744 | if (!path) | |
745 | return ERR_PTR(-ENOMEM); | |
746 | path->leave_spinning = 1; | |
747 | ||
748 | /* | |
749 | * Find the "default" dir item which points to the root item that we | |
750 | * will mount by default if we haven't been given a specific subvolume | |
751 | * to mount. | |
752 | */ | |
815745cf | 753 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
73f73415 | 754 | di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0); |
b0839166 JL |
755 | if (IS_ERR(di)) { |
756 | btrfs_free_path(path); | |
fb4f6f91 | 757 | return ERR_CAST(di); |
b0839166 | 758 | } |
73f73415 JB |
759 | if (!di) { |
760 | /* | |
761 | * Ok the default dir item isn't there. This is weird since | |
762 | * it's always been there, but don't freak out, just try and | |
763 | * mount to root most subvolume. | |
764 | */ | |
765 | btrfs_free_path(path); | |
766 | dir_id = BTRFS_FIRST_FREE_OBJECTID; | |
815745cf | 767 | new_root = fs_info->fs_root; |
73f73415 JB |
768 | goto setup_root; |
769 | } | |
770 | ||
771 | btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); | |
772 | btrfs_free_path(path); | |
773 | ||
774 | find_root: | |
815745cf | 775 | new_root = btrfs_read_fs_root_no_name(fs_info, &location); |
73f73415 | 776 | if (IS_ERR(new_root)) |
d0b678cb | 777 | return ERR_CAST(new_root); |
73f73415 JB |
778 | |
779 | if (btrfs_root_refs(&new_root->root_item) == 0) | |
780 | return ERR_PTR(-ENOENT); | |
781 | ||
782 | dir_id = btrfs_root_dirid(&new_root->root_item); | |
783 | setup_root: | |
784 | location.objectid = dir_id; | |
785 | location.type = BTRFS_INODE_ITEM_KEY; | |
786 | location.offset = 0; | |
787 | ||
788 | inode = btrfs_iget(sb, &location, new_root, &new); | |
4cbd1149 DC |
789 | if (IS_ERR(inode)) |
790 | return ERR_CAST(inode); | |
73f73415 JB |
791 | |
792 | /* | |
793 | * If we're just mounting the root most subvol put the inode and return | |
794 | * a reference to the dentry. We will have already gotten a reference | |
795 | * to the inode in btrfs_fill_super so we're good to go. | |
796 | */ | |
797 | if (!new && sb->s_root->d_inode == inode) { | |
798 | iput(inode); | |
799 | return dget(sb->s_root); | |
800 | } | |
801 | ||
ba5b8958 | 802 | return d_obtain_alias(inode); |
73f73415 JB |
803 | } |
804 | ||
d397712b | 805 | static int btrfs_fill_super(struct super_block *sb, |
8a4b83cc | 806 | struct btrfs_fs_devices *fs_devices, |
d397712b | 807 | void *data, int silent) |
75dfe396 | 808 | { |
d397712b | 809 | struct inode *inode; |
815745cf | 810 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
5d4f98a2 | 811 | struct btrfs_key key; |
39279cc3 | 812 | int err; |
a429e513 | 813 | |
39279cc3 CM |
814 | sb->s_maxbytes = MAX_LFS_FILESIZE; |
815 | sb->s_magic = BTRFS_SUPER_MAGIC; | |
816 | sb->s_op = &btrfs_super_ops; | |
af53d29a | 817 | sb->s_d_op = &btrfs_dentry_operations; |
be6e8dc0 | 818 | sb->s_export_op = &btrfs_export_ops; |
5103e947 | 819 | sb->s_xattr = btrfs_xattr_handlers; |
39279cc3 | 820 | sb->s_time_gran = 1; |
0eda294d | 821 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
33268eaf | 822 | sb->s_flags |= MS_POSIXACL; |
49cf6f45 | 823 | #endif |
0c4d2d95 | 824 | sb->s_flags |= MS_I_VERSION; |
ad2b2c80 AV |
825 | err = open_ctree(sb, fs_devices, (char *)data); |
826 | if (err) { | |
39279cc3 | 827 | printk("btrfs: open_ctree failed\n"); |
ad2b2c80 | 828 | return err; |
a429e513 CM |
829 | } |
830 | ||
5d4f98a2 YZ |
831 | key.objectid = BTRFS_FIRST_FREE_OBJECTID; |
832 | key.type = BTRFS_INODE_ITEM_KEY; | |
833 | key.offset = 0; | |
98c7089c | 834 | inode = btrfs_iget(sb, &key, fs_info->fs_root, NULL); |
5d4f98a2 YZ |
835 | if (IS_ERR(inode)) { |
836 | err = PTR_ERR(inode); | |
39279cc3 | 837 | goto fail_close; |
f254e52c | 838 | } |
f254e52c | 839 | |
48fde701 AV |
840 | sb->s_root = d_make_root(inode); |
841 | if (!sb->s_root) { | |
39279cc3 CM |
842 | err = -ENOMEM; |
843 | goto fail_close; | |
f254e52c | 844 | } |
58176a96 | 845 | |
6885f308 | 846 | save_mount_options(sb, data); |
90a887c9 | 847 | cleancache_init_fs(sb); |
59553edf | 848 | sb->s_flags |= MS_ACTIVE; |
2619ba1f | 849 | return 0; |
39279cc3 CM |
850 | |
851 | fail_close: | |
815745cf | 852 | close_ctree(fs_info->tree_root); |
39279cc3 | 853 | return err; |
2619ba1f CM |
854 | } |
855 | ||
6bf13c0c | 856 | int btrfs_sync_fs(struct super_block *sb, int wait) |
c5739bba CM |
857 | { |
858 | struct btrfs_trans_handle *trans; | |
815745cf AV |
859 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
860 | struct btrfs_root *root = fs_info->tree_root; | |
2619ba1f | 861 | |
1abe9b8a | 862 | trace_btrfs_sync_fs(wait); |
863 | ||
39279cc3 | 864 | if (!wait) { |
815745cf | 865 | filemap_flush(fs_info->btree_inode->i_mapping); |
39279cc3 CM |
866 | return 0; |
867 | } | |
771ed689 | 868 | |
62dbd717 | 869 | btrfs_wait_ordered_extents(root, 1); |
771ed689 | 870 | |
d4edf39b | 871 | trans = btrfs_attach_transaction_barrier(root); |
60376ce4 | 872 | if (IS_ERR(trans)) { |
354aa0fb MX |
873 | /* no transaction, don't bother */ |
874 | if (PTR_ERR(trans) == -ENOENT) | |
60376ce4 | 875 | return 0; |
98d5dc13 | 876 | return PTR_ERR(trans); |
60376ce4 | 877 | } |
bd7de2c9 | 878 | return btrfs_commit_transaction(trans, root); |
2c90e5d6 CM |
879 | } |
880 | ||
34c80b1d | 881 | static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry) |
a9572a15 | 882 | { |
815745cf AV |
883 | struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb); |
884 | struct btrfs_root *root = info->tree_root; | |
200da64e | 885 | char *compress_type; |
a9572a15 EP |
886 | |
887 | if (btrfs_test_opt(root, DEGRADED)) | |
888 | seq_puts(seq, ",degraded"); | |
889 | if (btrfs_test_opt(root, NODATASUM)) | |
890 | seq_puts(seq, ",nodatasum"); | |
891 | if (btrfs_test_opt(root, NODATACOW)) | |
892 | seq_puts(seq, ",nodatacow"); | |
893 | if (btrfs_test_opt(root, NOBARRIER)) | |
894 | seq_puts(seq, ",nobarrier"); | |
a9572a15 | 895 | if (info->max_inline != 8192 * 1024) |
21380931 JB |
896 | seq_printf(seq, ",max_inline=%llu", |
897 | (unsigned long long)info->max_inline); | |
a9572a15 | 898 | if (info->alloc_start != 0) |
21380931 JB |
899 | seq_printf(seq, ",alloc_start=%llu", |
900 | (unsigned long long)info->alloc_start); | |
a9572a15 EP |
901 | if (info->thread_pool_size != min_t(unsigned long, |
902 | num_online_cpus() + 2, 8)) | |
903 | seq_printf(seq, ",thread_pool=%d", info->thread_pool_size); | |
200da64e TI |
904 | if (btrfs_test_opt(root, COMPRESS)) { |
905 | if (info->compress_type == BTRFS_COMPRESS_ZLIB) | |
906 | compress_type = "zlib"; | |
907 | else | |
908 | compress_type = "lzo"; | |
909 | if (btrfs_test_opt(root, FORCE_COMPRESS)) | |
910 | seq_printf(seq, ",compress-force=%s", compress_type); | |
911 | else | |
912 | seq_printf(seq, ",compress=%s", compress_type); | |
913 | } | |
c289811c CM |
914 | if (btrfs_test_opt(root, NOSSD)) |
915 | seq_puts(seq, ",nossd"); | |
451d7585 CM |
916 | if (btrfs_test_opt(root, SSD_SPREAD)) |
917 | seq_puts(seq, ",ssd_spread"); | |
918 | else if (btrfs_test_opt(root, SSD)) | |
a9572a15 | 919 | seq_puts(seq, ",ssd"); |
3a5e1404 | 920 | if (btrfs_test_opt(root, NOTREELOG)) |
6b65c5c6 | 921 | seq_puts(seq, ",notreelog"); |
dccae999 | 922 | if (btrfs_test_opt(root, FLUSHONCOMMIT)) |
6b65c5c6 | 923 | seq_puts(seq, ",flushoncommit"); |
20a5239a MW |
924 | if (btrfs_test_opt(root, DISCARD)) |
925 | seq_puts(seq, ",discard"); | |
a9572a15 EP |
926 | if (!(root->fs_info->sb->s_flags & MS_POSIXACL)) |
927 | seq_puts(seq, ",noacl"); | |
200da64e TI |
928 | if (btrfs_test_opt(root, SPACE_CACHE)) |
929 | seq_puts(seq, ",space_cache"); | |
73bc1876 | 930 | else |
8965593e | 931 | seq_puts(seq, ",nospace_cache"); |
200da64e TI |
932 | if (btrfs_test_opt(root, CLEAR_CACHE)) |
933 | seq_puts(seq, ",clear_cache"); | |
934 | if (btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED)) | |
935 | seq_puts(seq, ",user_subvol_rm_allowed"); | |
0942caa3 DS |
936 | if (btrfs_test_opt(root, ENOSPC_DEBUG)) |
937 | seq_puts(seq, ",enospc_debug"); | |
938 | if (btrfs_test_opt(root, AUTO_DEFRAG)) | |
939 | seq_puts(seq, ",autodefrag"); | |
940 | if (btrfs_test_opt(root, INODE_MAP_CACHE)) | |
941 | seq_puts(seq, ",inode_cache"); | |
9555c6c1 ID |
942 | if (btrfs_test_opt(root, SKIP_BALANCE)) |
943 | seq_puts(seq, ",skip_balance"); | |
8c342930 JM |
944 | if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR)) |
945 | seq_puts(seq, ",fatal_errors=panic"); | |
a9572a15 EP |
946 | return 0; |
947 | } | |
948 | ||
a061fc8d | 949 | static int btrfs_test_super(struct super_block *s, void *data) |
4b82d6e4 | 950 | { |
815745cf AV |
951 | struct btrfs_fs_info *p = data; |
952 | struct btrfs_fs_info *fs_info = btrfs_sb(s); | |
4b82d6e4 | 953 | |
815745cf | 954 | return fs_info->fs_devices == p->fs_devices; |
4b82d6e4 Y |
955 | } |
956 | ||
450ba0ea JB |
957 | static int btrfs_set_super(struct super_block *s, void *data) |
958 | { | |
6de1d09d AV |
959 | int err = set_anon_super(s, data); |
960 | if (!err) | |
961 | s->s_fs_info = data; | |
962 | return err; | |
4b82d6e4 Y |
963 | } |
964 | ||
f9d9ef62 DS |
965 | /* |
966 | * subvolumes are identified by ino 256 | |
967 | */ | |
968 | static inline int is_subvolume_inode(struct inode *inode) | |
969 | { | |
970 | if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) | |
971 | return 1; | |
972 | return 0; | |
973 | } | |
974 | ||
830c4adb JB |
975 | /* |
976 | * This will strip out the subvol=%s argument for an argument string and add | |
977 | * subvolid=0 to make sure we get the actual tree root for path walking to the | |
978 | * subvol we want. | |
979 | */ | |
980 | static char *setup_root_args(char *args) | |
981 | { | |
f60d16a8 JM |
982 | unsigned len = strlen(args) + 2 + 1; |
983 | char *src, *dst, *buf; | |
830c4adb JB |
984 | |
985 | /* | |
f60d16a8 JM |
986 | * We need the same args as before, but with this substitution: |
987 | * s!subvol=[^,]+!subvolid=0! | |
830c4adb | 988 | * |
f60d16a8 JM |
989 | * Since the replacement string is up to 2 bytes longer than the |
990 | * original, allocate strlen(args) + 2 + 1 bytes. | |
830c4adb | 991 | */ |
830c4adb | 992 | |
f60d16a8 | 993 | src = strstr(args, "subvol="); |
830c4adb | 994 | /* This shouldn't happen, but just in case.. */ |
f60d16a8 JM |
995 | if (!src) |
996 | return NULL; | |
997 | ||
998 | buf = dst = kmalloc(len, GFP_NOFS); | |
999 | if (!buf) | |
830c4adb | 1000 | return NULL; |
830c4adb JB |
1001 | |
1002 | /* | |
f60d16a8 JM |
1003 | * If the subvol= arg is not at the start of the string, |
1004 | * copy whatever precedes it into buf. | |
830c4adb | 1005 | */ |
f60d16a8 JM |
1006 | if (src != args) { |
1007 | *src++ = '\0'; | |
1008 | strcpy(buf, args); | |
1009 | dst += strlen(args); | |
830c4adb JB |
1010 | } |
1011 | ||
f60d16a8 JM |
1012 | strcpy(dst, "subvolid=0"); |
1013 | dst += strlen("subvolid=0"); | |
830c4adb JB |
1014 | |
1015 | /* | |
f60d16a8 JM |
1016 | * If there is a "," after the original subvol=... string, |
1017 | * copy that suffix into our buffer. Otherwise, we're done. | |
830c4adb | 1018 | */ |
f60d16a8 JM |
1019 | src = strchr(src, ','); |
1020 | if (src) | |
1021 | strcpy(dst, src); | |
830c4adb | 1022 | |
f60d16a8 | 1023 | return buf; |
830c4adb JB |
1024 | } |
1025 | ||
1026 | static struct dentry *mount_subvol(const char *subvol_name, int flags, | |
1027 | const char *device_name, char *data) | |
1028 | { | |
830c4adb JB |
1029 | struct dentry *root; |
1030 | struct vfsmount *mnt; | |
830c4adb | 1031 | char *newargs; |
830c4adb JB |
1032 | |
1033 | newargs = setup_root_args(data); | |
1034 | if (!newargs) | |
1035 | return ERR_PTR(-ENOMEM); | |
1036 | mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name, | |
1037 | newargs); | |
1038 | kfree(newargs); | |
1039 | if (IS_ERR(mnt)) | |
1040 | return ERR_CAST(mnt); | |
1041 | ||
ea441d11 | 1042 | root = mount_subtree(mnt, subvol_name); |
830c4adb | 1043 | |
ea441d11 AV |
1044 | if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) { |
1045 | struct super_block *s = root->d_sb; | |
1046 | dput(root); | |
1047 | root = ERR_PTR(-EINVAL); | |
1048 | deactivate_locked_super(s); | |
f9d9ef62 DS |
1049 | printk(KERN_ERR "btrfs: '%s' is not a valid subvolume\n", |
1050 | subvol_name); | |
f9d9ef62 DS |
1051 | } |
1052 | ||
830c4adb JB |
1053 | return root; |
1054 | } | |
450ba0ea | 1055 | |
edf24abe CH |
1056 | /* |
1057 | * Find a superblock for the given device / mount point. | |
1058 | * | |
1059 | * Note: This is based on get_sb_bdev from fs/super.c with a few additions | |
1060 | * for multiple device setup. Make sure to keep it in sync. | |
1061 | */ | |
061dbc6b | 1062 | static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, |
306e16ce | 1063 | const char *device_name, void *data) |
4b82d6e4 Y |
1064 | { |
1065 | struct block_device *bdev = NULL; | |
1066 | struct super_block *s; | |
1067 | struct dentry *root; | |
8a4b83cc | 1068 | struct btrfs_fs_devices *fs_devices = NULL; |
450ba0ea | 1069 | struct btrfs_fs_info *fs_info = NULL; |
97288f2c | 1070 | fmode_t mode = FMODE_READ; |
73f73415 JB |
1071 | char *subvol_name = NULL; |
1072 | u64 subvol_objectid = 0; | |
4b82d6e4 Y |
1073 | int error = 0; |
1074 | ||
97288f2c CH |
1075 | if (!(flags & MS_RDONLY)) |
1076 | mode |= FMODE_WRITE; | |
1077 | ||
1078 | error = btrfs_parse_early_options(data, mode, fs_type, | |
73f73415 | 1079 | &subvol_name, &subvol_objectid, |
5e2a4b25 | 1080 | &fs_devices); |
f23c8af8 ID |
1081 | if (error) { |
1082 | kfree(subvol_name); | |
061dbc6b | 1083 | return ERR_PTR(error); |
f23c8af8 | 1084 | } |
edf24abe | 1085 | |
830c4adb JB |
1086 | if (subvol_name) { |
1087 | root = mount_subvol(subvol_name, flags, device_name, data); | |
1088 | kfree(subvol_name); | |
1089 | return root; | |
1090 | } | |
1091 | ||
306e16ce | 1092 | error = btrfs_scan_one_device(device_name, mode, fs_type, &fs_devices); |
8a4b83cc | 1093 | if (error) |
830c4adb | 1094 | return ERR_PTR(error); |
4b82d6e4 | 1095 | |
450ba0ea JB |
1096 | /* |
1097 | * Setup a dummy root and fs_info for test/set super. This is because | |
1098 | * we don't actually fill this stuff out until open_ctree, but we need | |
1099 | * it for searching for existing supers, so this lets us do that and | |
1100 | * then open_ctree will properly initialize everything later. | |
1101 | */ | |
1102 | fs_info = kzalloc(sizeof(struct btrfs_fs_info), GFP_NOFS); | |
04d21a24 ID |
1103 | if (!fs_info) |
1104 | return ERR_PTR(-ENOMEM); | |
1105 | ||
450ba0ea | 1106 | fs_info->fs_devices = fs_devices; |
450ba0ea | 1107 | |
6c41761f DS |
1108 | fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS); |
1109 | fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_NOFS); | |
1110 | if (!fs_info->super_copy || !fs_info->super_for_commit) { | |
1111 | error = -ENOMEM; | |
04d21a24 ID |
1112 | goto error_fs_info; |
1113 | } | |
1114 | ||
1115 | error = btrfs_open_devices(fs_devices, mode, fs_type); | |
1116 | if (error) | |
1117 | goto error_fs_info; | |
1118 | ||
1119 | if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) { | |
1120 | error = -EACCES; | |
6c41761f DS |
1121 | goto error_close_devices; |
1122 | } | |
1123 | ||
dfe25020 | 1124 | bdev = fs_devices->latest_bdev; |
9249e17f DH |
1125 | s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | MS_NOSEC, |
1126 | fs_info); | |
830c4adb JB |
1127 | if (IS_ERR(s)) { |
1128 | error = PTR_ERR(s); | |
1129 | goto error_close_devices; | |
1130 | } | |
4b82d6e4 Y |
1131 | |
1132 | if (s->s_root) { | |
2b82032c | 1133 | btrfs_close_devices(fs_devices); |
6c41761f | 1134 | free_fs_info(fs_info); |
59553edf AV |
1135 | if ((flags ^ s->s_flags) & MS_RDONLY) |
1136 | error = -EBUSY; | |
4b82d6e4 Y |
1137 | } else { |
1138 | char b[BDEVNAME_SIZE]; | |
1139 | ||
4b82d6e4 | 1140 | strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id)); |
815745cf | 1141 | btrfs_sb(s)->bdev_holder = fs_type; |
8a4b83cc CM |
1142 | error = btrfs_fill_super(s, fs_devices, data, |
1143 | flags & MS_SILENT ? 1 : 0); | |
4b82d6e4 Y |
1144 | } |
1145 | ||
59553edf AV |
1146 | root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error); |
1147 | if (IS_ERR(root)) | |
830c4adb | 1148 | deactivate_locked_super(s); |
4b82d6e4 | 1149 | |
061dbc6b | 1150 | return root; |
4b82d6e4 | 1151 | |
c146afad | 1152 | error_close_devices: |
8a4b83cc | 1153 | btrfs_close_devices(fs_devices); |
04d21a24 | 1154 | error_fs_info: |
6c41761f | 1155 | free_fs_info(fs_info); |
061dbc6b | 1156 | return ERR_PTR(error); |
4b82d6e4 | 1157 | } |
2e635a27 | 1158 | |
0d2450ab ST |
1159 | static void btrfs_set_max_workers(struct btrfs_workers *workers, int new_limit) |
1160 | { | |
1161 | spin_lock_irq(&workers->lock); | |
1162 | workers->max_workers = new_limit; | |
1163 | spin_unlock_irq(&workers->lock); | |
1164 | } | |
1165 | ||
1166 | static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info, | |
1167 | int new_pool_size, int old_pool_size) | |
1168 | { | |
1169 | if (new_pool_size == old_pool_size) | |
1170 | return; | |
1171 | ||
1172 | fs_info->thread_pool_size = new_pool_size; | |
1173 | ||
1174 | printk(KERN_INFO "btrfs: resize thread pool %d -> %d\n", | |
1175 | old_pool_size, new_pool_size); | |
1176 | ||
1177 | btrfs_set_max_workers(&fs_info->generic_worker, new_pool_size); | |
1178 | btrfs_set_max_workers(&fs_info->workers, new_pool_size); | |
1179 | btrfs_set_max_workers(&fs_info->delalloc_workers, new_pool_size); | |
1180 | btrfs_set_max_workers(&fs_info->submit_workers, new_pool_size); | |
1181 | btrfs_set_max_workers(&fs_info->caching_workers, new_pool_size); | |
1182 | btrfs_set_max_workers(&fs_info->fixup_workers, new_pool_size); | |
1183 | btrfs_set_max_workers(&fs_info->endio_workers, new_pool_size); | |
1184 | btrfs_set_max_workers(&fs_info->endio_meta_workers, new_pool_size); | |
1185 | btrfs_set_max_workers(&fs_info->endio_meta_write_workers, new_pool_size); | |
1186 | btrfs_set_max_workers(&fs_info->endio_write_workers, new_pool_size); | |
1187 | btrfs_set_max_workers(&fs_info->endio_freespace_worker, new_pool_size); | |
1188 | btrfs_set_max_workers(&fs_info->delayed_workers, new_pool_size); | |
1189 | btrfs_set_max_workers(&fs_info->readahead_workers, new_pool_size); | |
ff023aac SB |
1190 | btrfs_set_max_workers(&fs_info->scrub_wr_completion_workers, |
1191 | new_pool_size); | |
0d2450ab ST |
1192 | } |
1193 | ||
f42a34b2 | 1194 | static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info) |
dc81cdc5 MX |
1195 | { |
1196 | set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); | |
f42a34b2 | 1197 | } |
dc81cdc5 | 1198 | |
f42a34b2 MX |
1199 | static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info, |
1200 | unsigned long old_opts, int flags) | |
1201 | { | |
dc81cdc5 MX |
1202 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && |
1203 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || | |
1204 | (flags & MS_RDONLY))) { | |
1205 | /* wait for any defraggers to finish */ | |
1206 | wait_event(fs_info->transaction_wait, | |
1207 | (atomic_read(&fs_info->defrag_running) == 0)); | |
1208 | if (flags & MS_RDONLY) | |
1209 | sync_filesystem(fs_info->sb); | |
1210 | } | |
1211 | } | |
1212 | ||
1213 | static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info, | |
1214 | unsigned long old_opts) | |
1215 | { | |
1216 | /* | |
1217 | * We need cleanup all defragable inodes if the autodefragment is | |
1218 | * close or the fs is R/O. | |
1219 | */ | |
1220 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && | |
1221 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || | |
1222 | (fs_info->sb->s_flags & MS_RDONLY))) { | |
1223 | btrfs_cleanup_defrag_inodes(fs_info); | |
1224 | } | |
1225 | ||
1226 | clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); | |
1227 | } | |
1228 | ||
c146afad YZ |
1229 | static int btrfs_remount(struct super_block *sb, int *flags, char *data) |
1230 | { | |
815745cf AV |
1231 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
1232 | struct btrfs_root *root = fs_info->tree_root; | |
49b25e05 JM |
1233 | unsigned old_flags = sb->s_flags; |
1234 | unsigned long old_opts = fs_info->mount_opt; | |
1235 | unsigned long old_compress_type = fs_info->compress_type; | |
1236 | u64 old_max_inline = fs_info->max_inline; | |
1237 | u64 old_alloc_start = fs_info->alloc_start; | |
1238 | int old_thread_pool_size = fs_info->thread_pool_size; | |
1239 | unsigned int old_metadata_ratio = fs_info->metadata_ratio; | |
c146afad YZ |
1240 | int ret; |
1241 | ||
f42a34b2 | 1242 | btrfs_remount_prepare(fs_info); |
dc81cdc5 | 1243 | |
b288052e | 1244 | ret = btrfs_parse_options(root, data); |
49b25e05 JM |
1245 | if (ret) { |
1246 | ret = -EINVAL; | |
1247 | goto restore; | |
1248 | } | |
b288052e | 1249 | |
f42a34b2 | 1250 | btrfs_remount_begin(fs_info, old_opts, *flags); |
0d2450ab ST |
1251 | btrfs_resize_thread_pool(fs_info, |
1252 | fs_info->thread_pool_size, old_thread_pool_size); | |
1253 | ||
c146afad | 1254 | if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY)) |
dc81cdc5 | 1255 | goto out; |
c146afad YZ |
1256 | |
1257 | if (*flags & MS_RDONLY) { | |
8dabb742 SB |
1258 | /* |
1259 | * this also happens on 'umount -rf' or on shutdown, when | |
1260 | * the filesystem is busy. | |
1261 | */ | |
c146afad YZ |
1262 | sb->s_flags |= MS_RDONLY; |
1263 | ||
8dabb742 SB |
1264 | btrfs_dev_replace_suspend_for_unmount(fs_info); |
1265 | btrfs_scrub_cancel(fs_info); | |
061594ef | 1266 | btrfs_pause_balance(fs_info); |
8dabb742 | 1267 | |
49b25e05 JM |
1268 | ret = btrfs_commit_super(root); |
1269 | if (ret) | |
1270 | goto restore; | |
c146afad | 1271 | } else { |
8a3db184 | 1272 | if (fs_info->fs_devices->rw_devices == 0) { |
49b25e05 JM |
1273 | ret = -EACCES; |
1274 | goto restore; | |
8a3db184 | 1275 | } |
2b82032c | 1276 | |
292fd7fc SB |
1277 | if (fs_info->fs_devices->missing_devices > |
1278 | fs_info->num_tolerated_disk_barrier_failures && | |
1279 | !(*flags & MS_RDONLY)) { | |
1280 | printk(KERN_WARNING | |
1281 | "Btrfs: too many missing devices, writeable remount is not allowed\n"); | |
1282 | ret = -EACCES; | |
1283 | goto restore; | |
1284 | } | |
1285 | ||
8a3db184 | 1286 | if (btrfs_super_log_root(fs_info->super_copy) != 0) { |
49b25e05 JM |
1287 | ret = -EINVAL; |
1288 | goto restore; | |
8a3db184 | 1289 | } |
c146afad | 1290 | |
815745cf | 1291 | ret = btrfs_cleanup_fs_roots(fs_info); |
49b25e05 JM |
1292 | if (ret) |
1293 | goto restore; | |
c146afad | 1294 | |
d68fc57b YZ |
1295 | /* recover relocation */ |
1296 | ret = btrfs_recover_relocation(root); | |
49b25e05 JM |
1297 | if (ret) |
1298 | goto restore; | |
c146afad | 1299 | |
2b6ba629 ID |
1300 | ret = btrfs_resume_balance_async(fs_info); |
1301 | if (ret) | |
1302 | goto restore; | |
1303 | ||
8dabb742 SB |
1304 | ret = btrfs_resume_dev_replace_async(fs_info); |
1305 | if (ret) { | |
1306 | pr_warn("btrfs: failed to resume dev_replace\n"); | |
1307 | goto restore; | |
1308 | } | |
c146afad YZ |
1309 | sb->s_flags &= ~MS_RDONLY; |
1310 | } | |
dc81cdc5 MX |
1311 | out: |
1312 | btrfs_remount_cleanup(fs_info, old_opts); | |
c146afad | 1313 | return 0; |
49b25e05 JM |
1314 | |
1315 | restore: | |
1316 | /* We've hit an error - don't reset MS_RDONLY */ | |
1317 | if (sb->s_flags & MS_RDONLY) | |
1318 | old_flags |= MS_RDONLY; | |
1319 | sb->s_flags = old_flags; | |
1320 | fs_info->mount_opt = old_opts; | |
1321 | fs_info->compress_type = old_compress_type; | |
1322 | fs_info->max_inline = old_max_inline; | |
c018daec | 1323 | mutex_lock(&fs_info->chunk_mutex); |
49b25e05 | 1324 | fs_info->alloc_start = old_alloc_start; |
c018daec | 1325 | mutex_unlock(&fs_info->chunk_mutex); |
0d2450ab ST |
1326 | btrfs_resize_thread_pool(fs_info, |
1327 | old_thread_pool_size, fs_info->thread_pool_size); | |
49b25e05 | 1328 | fs_info->metadata_ratio = old_metadata_ratio; |
dc81cdc5 | 1329 | btrfs_remount_cleanup(fs_info, old_opts); |
49b25e05 | 1330 | return ret; |
c146afad YZ |
1331 | } |
1332 | ||
bcd53741 AJ |
1333 | /* Used to sort the devices by max_avail(descending sort) */ |
1334 | static int btrfs_cmp_device_free_bytes(const void *dev_info1, | |
1335 | const void *dev_info2) | |
1336 | { | |
1337 | if (((struct btrfs_device_info *)dev_info1)->max_avail > | |
1338 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
1339 | return -1; | |
1340 | else if (((struct btrfs_device_info *)dev_info1)->max_avail < | |
1341 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
1342 | return 1; | |
1343 | else | |
1344 | return 0; | |
1345 | } | |
1346 | ||
1347 | /* | |
1348 | * sort the devices by max_avail, in which max free extent size of each device | |
1349 | * is stored.(Descending Sort) | |
1350 | */ | |
1351 | static inline void btrfs_descending_sort_devices( | |
1352 | struct btrfs_device_info *devices, | |
1353 | size_t nr_devices) | |
1354 | { | |
1355 | sort(devices, nr_devices, sizeof(struct btrfs_device_info), | |
1356 | btrfs_cmp_device_free_bytes, NULL); | |
1357 | } | |
1358 | ||
6d07bcec MX |
1359 | /* |
1360 | * The helper to calc the free space on the devices that can be used to store | |
1361 | * file data. | |
1362 | */ | |
1363 | static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes) | |
1364 | { | |
1365 | struct btrfs_fs_info *fs_info = root->fs_info; | |
1366 | struct btrfs_device_info *devices_info; | |
1367 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; | |
1368 | struct btrfs_device *device; | |
1369 | u64 skip_space; | |
1370 | u64 type; | |
1371 | u64 avail_space; | |
1372 | u64 used_space; | |
1373 | u64 min_stripe_size; | |
39fb26c3 | 1374 | int min_stripes = 1, num_stripes = 1; |
6d07bcec MX |
1375 | int i = 0, nr_devices; |
1376 | int ret; | |
1377 | ||
b772a86e | 1378 | nr_devices = fs_info->fs_devices->open_devices; |
6d07bcec MX |
1379 | BUG_ON(!nr_devices); |
1380 | ||
1381 | devices_info = kmalloc(sizeof(*devices_info) * nr_devices, | |
1382 | GFP_NOFS); | |
1383 | if (!devices_info) | |
1384 | return -ENOMEM; | |
1385 | ||
1386 | /* calc min stripe number for data space alloction */ | |
1387 | type = btrfs_get_alloc_profile(root, 1); | |
39fb26c3 | 1388 | if (type & BTRFS_BLOCK_GROUP_RAID0) { |
6d07bcec | 1389 | min_stripes = 2; |
39fb26c3 MX |
1390 | num_stripes = nr_devices; |
1391 | } else if (type & BTRFS_BLOCK_GROUP_RAID1) { | |
6d07bcec | 1392 | min_stripes = 2; |
39fb26c3 MX |
1393 | num_stripes = 2; |
1394 | } else if (type & BTRFS_BLOCK_GROUP_RAID10) { | |
6d07bcec | 1395 | min_stripes = 4; |
39fb26c3 MX |
1396 | num_stripes = 4; |
1397 | } | |
6d07bcec MX |
1398 | |
1399 | if (type & BTRFS_BLOCK_GROUP_DUP) | |
1400 | min_stripe_size = 2 * BTRFS_STRIPE_LEN; | |
1401 | else | |
1402 | min_stripe_size = BTRFS_STRIPE_LEN; | |
1403 | ||
b772a86e | 1404 | list_for_each_entry(device, &fs_devices->devices, dev_list) { |
63a212ab SB |
1405 | if (!device->in_fs_metadata || !device->bdev || |
1406 | device->is_tgtdev_for_dev_replace) | |
6d07bcec MX |
1407 | continue; |
1408 | ||
1409 | avail_space = device->total_bytes - device->bytes_used; | |
1410 | ||
1411 | /* align with stripe_len */ | |
1412 | do_div(avail_space, BTRFS_STRIPE_LEN); | |
1413 | avail_space *= BTRFS_STRIPE_LEN; | |
1414 | ||
1415 | /* | |
1416 | * In order to avoid overwritting the superblock on the drive, | |
1417 | * btrfs starts at an offset of at least 1MB when doing chunk | |
1418 | * allocation. | |
1419 | */ | |
1420 | skip_space = 1024 * 1024; | |
1421 | ||
1422 | /* user can set the offset in fs_info->alloc_start. */ | |
1423 | if (fs_info->alloc_start + BTRFS_STRIPE_LEN <= | |
1424 | device->total_bytes) | |
1425 | skip_space = max(fs_info->alloc_start, skip_space); | |
1426 | ||
1427 | /* | |
1428 | * btrfs can not use the free space in [0, skip_space - 1], | |
1429 | * we must subtract it from the total. In order to implement | |
1430 | * it, we account the used space in this range first. | |
1431 | */ | |
1432 | ret = btrfs_account_dev_extents_size(device, 0, skip_space - 1, | |
1433 | &used_space); | |
1434 | if (ret) { | |
1435 | kfree(devices_info); | |
1436 | return ret; | |
1437 | } | |
1438 | ||
1439 | /* calc the free space in [0, skip_space - 1] */ | |
1440 | skip_space -= used_space; | |
1441 | ||
1442 | /* | |
1443 | * we can use the free space in [0, skip_space - 1], subtract | |
1444 | * it from the total. | |
1445 | */ | |
1446 | if (avail_space && avail_space >= skip_space) | |
1447 | avail_space -= skip_space; | |
1448 | else | |
1449 | avail_space = 0; | |
1450 | ||
1451 | if (avail_space < min_stripe_size) | |
1452 | continue; | |
1453 | ||
1454 | devices_info[i].dev = device; | |
1455 | devices_info[i].max_avail = avail_space; | |
1456 | ||
1457 | i++; | |
1458 | } | |
1459 | ||
1460 | nr_devices = i; | |
1461 | ||
1462 | btrfs_descending_sort_devices(devices_info, nr_devices); | |
1463 | ||
1464 | i = nr_devices - 1; | |
1465 | avail_space = 0; | |
1466 | while (nr_devices >= min_stripes) { | |
39fb26c3 MX |
1467 | if (num_stripes > nr_devices) |
1468 | num_stripes = nr_devices; | |
1469 | ||
6d07bcec MX |
1470 | if (devices_info[i].max_avail >= min_stripe_size) { |
1471 | int j; | |
1472 | u64 alloc_size; | |
1473 | ||
39fb26c3 | 1474 | avail_space += devices_info[i].max_avail * num_stripes; |
6d07bcec | 1475 | alloc_size = devices_info[i].max_avail; |
39fb26c3 | 1476 | for (j = i + 1 - num_stripes; j <= i; j++) |
6d07bcec MX |
1477 | devices_info[j].max_avail -= alloc_size; |
1478 | } | |
1479 | i--; | |
1480 | nr_devices--; | |
1481 | } | |
1482 | ||
1483 | kfree(devices_info); | |
1484 | *free_bytes = avail_space; | |
1485 | return 0; | |
1486 | } | |
1487 | ||
8fd17795 CM |
1488 | static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) |
1489 | { | |
815745cf AV |
1490 | struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb); |
1491 | struct btrfs_super_block *disk_super = fs_info->super_copy; | |
1492 | struct list_head *head = &fs_info->space_info; | |
bd4d1088 JB |
1493 | struct btrfs_space_info *found; |
1494 | u64 total_used = 0; | |
6d07bcec | 1495 | u64 total_free_data = 0; |
db94535d | 1496 | int bits = dentry->d_sb->s_blocksize_bits; |
815745cf | 1497 | __be32 *fsid = (__be32 *)fs_info->fsid; |
6d07bcec | 1498 | int ret; |
8fd17795 | 1499 | |
6d07bcec | 1500 | /* holding chunk_muext to avoid allocating new chunks */ |
815745cf | 1501 | mutex_lock(&fs_info->chunk_mutex); |
bd4d1088 | 1502 | rcu_read_lock(); |
89a55897 | 1503 | list_for_each_entry_rcu(found, head, list) { |
6d07bcec MX |
1504 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) { |
1505 | total_free_data += found->disk_total - found->disk_used; | |
1506 | total_free_data -= | |
1507 | btrfs_account_ro_block_groups_free_space(found); | |
1508 | } | |
1509 | ||
b742bb82 | 1510 | total_used += found->disk_used; |
89a55897 | 1511 | } |
bd4d1088 JB |
1512 | rcu_read_unlock(); |
1513 | ||
8fd17795 | 1514 | buf->f_namelen = BTRFS_NAME_LEN; |
db94535d | 1515 | buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits; |
bd4d1088 | 1516 | buf->f_bfree = buf->f_blocks - (total_used >> bits); |
8fd17795 CM |
1517 | buf->f_bsize = dentry->d_sb->s_blocksize; |
1518 | buf->f_type = BTRFS_SUPER_MAGIC; | |
6d07bcec | 1519 | buf->f_bavail = total_free_data; |
815745cf | 1520 | ret = btrfs_calc_avail_data_space(fs_info->tree_root, &total_free_data); |
6d07bcec | 1521 | if (ret) { |
815745cf | 1522 | mutex_unlock(&fs_info->chunk_mutex); |
6d07bcec MX |
1523 | return ret; |
1524 | } | |
1525 | buf->f_bavail += total_free_data; | |
1526 | buf->f_bavail = buf->f_bavail >> bits; | |
815745cf | 1527 | mutex_unlock(&fs_info->chunk_mutex); |
d397712b | 1528 | |
9d03632e | 1529 | /* We treat it as constant endianness (it doesn't matter _which_) |
d397712b | 1530 | because we want the fsid to come out the same whether mounted |
9d03632e DW |
1531 | on a big-endian or little-endian host */ |
1532 | buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]); | |
1533 | buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]); | |
32d48fa1 DW |
1534 | /* Mask in the root object ID too, to disambiguate subvols */ |
1535 | buf->f_fsid.val[0] ^= BTRFS_I(dentry->d_inode)->root->objectid >> 32; | |
1536 | buf->f_fsid.val[1] ^= BTRFS_I(dentry->d_inode)->root->objectid; | |
1537 | ||
8fd17795 CM |
1538 | return 0; |
1539 | } | |
b5133862 | 1540 | |
aea52e19 AV |
1541 | static void btrfs_kill_super(struct super_block *sb) |
1542 | { | |
815745cf | 1543 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
aea52e19 | 1544 | kill_anon_super(sb); |
d22ca7de | 1545 | free_fs_info(fs_info); |
aea52e19 AV |
1546 | } |
1547 | ||
2e635a27 CM |
1548 | static struct file_system_type btrfs_fs_type = { |
1549 | .owner = THIS_MODULE, | |
1550 | .name = "btrfs", | |
061dbc6b | 1551 | .mount = btrfs_mount, |
aea52e19 | 1552 | .kill_sb = btrfs_kill_super, |
2e635a27 CM |
1553 | .fs_flags = FS_REQUIRES_DEV, |
1554 | }; | |
7f78e035 | 1555 | MODULE_ALIAS_FS("btrfs"); |
a9218f6b | 1556 | |
d352ac68 CM |
1557 | /* |
1558 | * used by btrfsctl to scan devices when no FS is mounted | |
1559 | */ | |
8a4b83cc CM |
1560 | static long btrfs_control_ioctl(struct file *file, unsigned int cmd, |
1561 | unsigned long arg) | |
1562 | { | |
1563 | struct btrfs_ioctl_vol_args *vol; | |
1564 | struct btrfs_fs_devices *fs_devices; | |
c071fcfd | 1565 | int ret = -ENOTTY; |
8a4b83cc | 1566 | |
e441d54d CM |
1567 | if (!capable(CAP_SYS_ADMIN)) |
1568 | return -EPERM; | |
1569 | ||
dae7b665 LZ |
1570 | vol = memdup_user((void __user *)arg, sizeof(*vol)); |
1571 | if (IS_ERR(vol)) | |
1572 | return PTR_ERR(vol); | |
c071fcfd | 1573 | |
8a4b83cc CM |
1574 | switch (cmd) { |
1575 | case BTRFS_IOC_SCAN_DEV: | |
97288f2c | 1576 | ret = btrfs_scan_one_device(vol->name, FMODE_READ, |
8a4b83cc CM |
1577 | &btrfs_fs_type, &fs_devices); |
1578 | break; | |
02db0844 JB |
1579 | case BTRFS_IOC_DEVICES_READY: |
1580 | ret = btrfs_scan_one_device(vol->name, FMODE_READ, | |
1581 | &btrfs_fs_type, &fs_devices); | |
1582 | if (ret) | |
1583 | break; | |
1584 | ret = !(fs_devices->num_devices == fs_devices->total_devices); | |
1585 | break; | |
8a4b83cc | 1586 | } |
dae7b665 | 1587 | |
8a4b83cc | 1588 | kfree(vol); |
f819d837 | 1589 | return ret; |
8a4b83cc CM |
1590 | } |
1591 | ||
0176260f | 1592 | static int btrfs_freeze(struct super_block *sb) |
ed0dab6b | 1593 | { |
354aa0fb MX |
1594 | struct btrfs_trans_handle *trans; |
1595 | struct btrfs_root *root = btrfs_sb(sb)->tree_root; | |
1596 | ||
d4edf39b | 1597 | trans = btrfs_attach_transaction_barrier(root); |
354aa0fb MX |
1598 | if (IS_ERR(trans)) { |
1599 | /* no transaction, don't bother */ | |
1600 | if (PTR_ERR(trans) == -ENOENT) | |
1601 | return 0; | |
1602 | return PTR_ERR(trans); | |
1603 | } | |
1604 | return btrfs_commit_transaction(trans, root); | |
ed0dab6b Y |
1605 | } |
1606 | ||
0176260f | 1607 | static int btrfs_unfreeze(struct super_block *sb) |
ed0dab6b | 1608 | { |
0176260f | 1609 | return 0; |
ed0dab6b | 1610 | } |
2e635a27 | 1611 | |
9c5085c1 JB |
1612 | static int btrfs_show_devname(struct seq_file *m, struct dentry *root) |
1613 | { | |
1614 | struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb); | |
1615 | struct btrfs_fs_devices *cur_devices; | |
1616 | struct btrfs_device *dev, *first_dev = NULL; | |
1617 | struct list_head *head; | |
1618 | struct rcu_string *name; | |
1619 | ||
1620 | mutex_lock(&fs_info->fs_devices->device_list_mutex); | |
1621 | cur_devices = fs_info->fs_devices; | |
1622 | while (cur_devices) { | |
1623 | head = &cur_devices->devices; | |
1624 | list_for_each_entry(dev, head, dev_list) { | |
aa9ddcd4 JB |
1625 | if (dev->missing) |
1626 | continue; | |
9c5085c1 JB |
1627 | if (!first_dev || dev->devid < first_dev->devid) |
1628 | first_dev = dev; | |
1629 | } | |
1630 | cur_devices = cur_devices->seed; | |
1631 | } | |
1632 | ||
1633 | if (first_dev) { | |
1634 | rcu_read_lock(); | |
1635 | name = rcu_dereference(first_dev->name); | |
1636 | seq_escape(m, name->str, " \t\n\\"); | |
1637 | rcu_read_unlock(); | |
1638 | } else { | |
1639 | WARN_ON(1); | |
1640 | } | |
1641 | mutex_unlock(&fs_info->fs_devices->device_list_mutex); | |
1642 | return 0; | |
1643 | } | |
1644 | ||
b87221de | 1645 | static const struct super_operations btrfs_super_ops = { |
76dda93c | 1646 | .drop_inode = btrfs_drop_inode, |
bd555975 | 1647 | .evict_inode = btrfs_evict_inode, |
e20d96d6 | 1648 | .put_super = btrfs_put_super, |
d5719762 | 1649 | .sync_fs = btrfs_sync_fs, |
a9572a15 | 1650 | .show_options = btrfs_show_options, |
9c5085c1 | 1651 | .show_devname = btrfs_show_devname, |
4730a4bc | 1652 | .write_inode = btrfs_write_inode, |
2c90e5d6 CM |
1653 | .alloc_inode = btrfs_alloc_inode, |
1654 | .destroy_inode = btrfs_destroy_inode, | |
8fd17795 | 1655 | .statfs = btrfs_statfs, |
c146afad | 1656 | .remount_fs = btrfs_remount, |
0176260f LT |
1657 | .freeze_fs = btrfs_freeze, |
1658 | .unfreeze_fs = btrfs_unfreeze, | |
e20d96d6 | 1659 | }; |
a9218f6b CM |
1660 | |
1661 | static const struct file_operations btrfs_ctl_fops = { | |
1662 | .unlocked_ioctl = btrfs_control_ioctl, | |
1663 | .compat_ioctl = btrfs_control_ioctl, | |
1664 | .owner = THIS_MODULE, | |
6038f373 | 1665 | .llseek = noop_llseek, |
a9218f6b CM |
1666 | }; |
1667 | ||
1668 | static struct miscdevice btrfs_misc = { | |
578454ff | 1669 | .minor = BTRFS_MINOR, |
a9218f6b CM |
1670 | .name = "btrfs-control", |
1671 | .fops = &btrfs_ctl_fops | |
1672 | }; | |
1673 | ||
578454ff KS |
1674 | MODULE_ALIAS_MISCDEV(BTRFS_MINOR); |
1675 | MODULE_ALIAS("devname:btrfs-control"); | |
1676 | ||
a9218f6b CM |
1677 | static int btrfs_interface_init(void) |
1678 | { | |
1679 | return misc_register(&btrfs_misc); | |
1680 | } | |
1681 | ||
b2950863 | 1682 | static void btrfs_interface_exit(void) |
a9218f6b CM |
1683 | { |
1684 | if (misc_deregister(&btrfs_misc) < 0) | |
48940662 | 1685 | printk(KERN_INFO "btrfs: misc_deregister failed for control device\n"); |
a9218f6b CM |
1686 | } |
1687 | ||
2e635a27 CM |
1688 | static int __init init_btrfs_fs(void) |
1689 | { | |
2c90e5d6 | 1690 | int err; |
58176a96 JB |
1691 | |
1692 | err = btrfs_init_sysfs(); | |
1693 | if (err) | |
1694 | return err; | |
1695 | ||
143bede5 | 1696 | btrfs_init_compress(); |
d1310b2e | 1697 | |
261507a0 LZ |
1698 | err = btrfs_init_cachep(); |
1699 | if (err) | |
1700 | goto free_compress; | |
1701 | ||
d1310b2e | 1702 | err = extent_io_init(); |
2f4cbe64 WB |
1703 | if (err) |
1704 | goto free_cachep; | |
1705 | ||
d1310b2e CM |
1706 | err = extent_map_init(); |
1707 | if (err) | |
1708 | goto free_extent_io; | |
1709 | ||
6352b91d | 1710 | err = ordered_data_init(); |
2f4cbe64 WB |
1711 | if (err) |
1712 | goto free_extent_map; | |
c8b97818 | 1713 | |
6352b91d MX |
1714 | err = btrfs_delayed_inode_init(); |
1715 | if (err) | |
1716 | goto free_ordered_data; | |
1717 | ||
9247f317 | 1718 | err = btrfs_auto_defrag_init(); |
16cdcec7 MX |
1719 | if (err) |
1720 | goto free_delayed_inode; | |
1721 | ||
78a6184a | 1722 | err = btrfs_delayed_ref_init(); |
9247f317 MX |
1723 | if (err) |
1724 | goto free_auto_defrag; | |
1725 | ||
78a6184a MX |
1726 | err = btrfs_interface_init(); |
1727 | if (err) | |
1728 | goto free_delayed_ref; | |
1729 | ||
a9218f6b CM |
1730 | err = register_filesystem(&btrfs_fs_type); |
1731 | if (err) | |
1732 | goto unregister_ioctl; | |
b3c3da71 | 1733 | |
e565d4b9 JS |
1734 | btrfs_init_lockdep(); |
1735 | ||
74255aa0 JB |
1736 | #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS |
1737 | btrfs_test_free_space_cache(); | |
1738 | #endif | |
1739 | ||
b3c3da71 | 1740 | printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION); |
2f4cbe64 WB |
1741 | return 0; |
1742 | ||
a9218f6b CM |
1743 | unregister_ioctl: |
1744 | btrfs_interface_exit(); | |
78a6184a MX |
1745 | free_delayed_ref: |
1746 | btrfs_delayed_ref_exit(); | |
9247f317 MX |
1747 | free_auto_defrag: |
1748 | btrfs_auto_defrag_exit(); | |
16cdcec7 MX |
1749 | free_delayed_inode: |
1750 | btrfs_delayed_inode_exit(); | |
6352b91d MX |
1751 | free_ordered_data: |
1752 | ordered_data_exit(); | |
2f4cbe64 WB |
1753 | free_extent_map: |
1754 | extent_map_exit(); | |
d1310b2e CM |
1755 | free_extent_io: |
1756 | extent_io_exit(); | |
2f4cbe64 WB |
1757 | free_cachep: |
1758 | btrfs_destroy_cachep(); | |
261507a0 LZ |
1759 | free_compress: |
1760 | btrfs_exit_compress(); | |
2f4cbe64 WB |
1761 | btrfs_exit_sysfs(); |
1762 | return err; | |
2e635a27 CM |
1763 | } |
1764 | ||
1765 | static void __exit exit_btrfs_fs(void) | |
1766 | { | |
39279cc3 | 1767 | btrfs_destroy_cachep(); |
78a6184a | 1768 | btrfs_delayed_ref_exit(); |
9247f317 | 1769 | btrfs_auto_defrag_exit(); |
16cdcec7 | 1770 | btrfs_delayed_inode_exit(); |
6352b91d | 1771 | ordered_data_exit(); |
a52d9a80 | 1772 | extent_map_exit(); |
d1310b2e | 1773 | extent_io_exit(); |
a9218f6b | 1774 | btrfs_interface_exit(); |
2e635a27 | 1775 | unregister_filesystem(&btrfs_fs_type); |
58176a96 | 1776 | btrfs_exit_sysfs(); |
8a4b83cc | 1777 | btrfs_cleanup_fs_uuids(); |
261507a0 | 1778 | btrfs_exit_compress(); |
2e635a27 CM |
1779 | } |
1780 | ||
1781 | module_init(init_btrfs_fs) | |
1782 | module_exit(exit_btrfs_fs) | |
1783 | ||
1784 | MODULE_LICENSE("GPL"); |