Commit | Line | Data |
---|---|---|
c1d7c514 | 1 | // SPDX-License-Identifier: GPL-2.0 |
6cbd5570 CM |
2 | /* |
3 | * Copyright (C) 2007 Oracle. All rights reserved. | |
6cbd5570 CM |
4 | */ |
5 | ||
4b82d6e4 | 6 | #include <linux/blkdev.h> |
2e635a27 CM |
7 | #include <linux/module.h> |
8 | #include <linux/fs.h> | |
9 | #include <linux/pagemap.h> | |
10 | #include <linux/highmem.h> | |
11 | #include <linux/time.h> | |
12 | #include <linux/init.h> | |
a9572a15 | 13 | #include <linux/seq_file.h> |
2e635a27 | 14 | #include <linux/string.h> |
2e635a27 | 15 | #include <linux/backing-dev.h> |
4b82d6e4 | 16 | #include <linux/mount.h> |
75dfe396 | 17 | #include <linux/writeback.h> |
8fd17795 | 18 | #include <linux/statfs.h> |
08607c1b | 19 | #include <linux/compat.h> |
95e05289 | 20 | #include <linux/parser.h> |
c59f8951 | 21 | #include <linux/ctype.h> |
6da6abae | 22 | #include <linux/namei.h> |
a9218f6b | 23 | #include <linux/miscdevice.h> |
1bcbf313 | 24 | #include <linux/magic.h> |
5a0e3ad6 | 25 | #include <linux/slab.h> |
90a887c9 | 26 | #include <linux/cleancache.h> |
22c44fe6 | 27 | #include <linux/ratelimit.h> |
9678c543 | 28 | #include <linux/crc32c.h> |
55e301fd | 29 | #include <linux/btrfs.h> |
16cdcec7 | 30 | #include "delayed-inode.h" |
2e635a27 | 31 | #include "ctree.h" |
e20d96d6 | 32 | #include "disk-io.h" |
d5719762 | 33 | #include "transaction.h" |
2c90e5d6 | 34 | #include "btrfs_inode.h" |
3a686375 | 35 | #include "print-tree.h" |
63541927 | 36 | #include "props.h" |
5103e947 | 37 | #include "xattr.h" |
8a4b83cc | 38 | #include "volumes.h" |
be6e8dc0 | 39 | #include "export.h" |
c8b97818 | 40 | #include "compression.h" |
9c5085c1 | 41 | #include "rcu-string.h" |
8dabb742 | 42 | #include "dev-replace.h" |
74255aa0 | 43 | #include "free-space-cache.h" |
b9e9a6cb | 44 | #include "backref.h" |
8719aaae | 45 | #include "space-info.h" |
89439109 | 46 | #include "sysfs.h" |
dc11dd5d | 47 | #include "tests/btrfs-tests.h" |
aac0023c | 48 | #include "block-group.h" |
b0643e59 | 49 | #include "discard.h" |
2e635a27 | 50 | |
d3982100 | 51 | #include "qgroup.h" |
1abe9b8a | 52 | #define CREATE_TRACE_POINTS |
53 | #include <trace/events/btrfs.h> | |
54 | ||
b87221de | 55 | static const struct super_operations btrfs_super_ops; |
72fa39f5 MT |
56 | |
57 | /* | |
58 | * Types for mounting the default subvolume and a subvolume explicitly | |
59 | * requested by subvol=/path. That way the callchain is straightforward and we | |
60 | * don't have to play tricks with the mount options and recursive calls to | |
61 | * btrfs_mount. | |
312c89fb MT |
62 | * |
63 | * The new btrfs_root_fs_type also servers as a tag for the bdev_holder. | |
72fa39f5 | 64 | */ |
830c4adb | 65 | static struct file_system_type btrfs_fs_type; |
72fa39f5 | 66 | static struct file_system_type btrfs_root_fs_type; |
75dfe396 | 67 | |
0723a047 HH |
68 | static int btrfs_remount(struct super_block *sb, int *flags, char *data); |
69 | ||
59131393 JB |
70 | /* |
71 | * Generally the error codes correspond to their respective errors, but there | |
72 | * are a few special cases. | |
73 | * | |
74 | * EUCLEAN: Any sort of corruption that we encounter. The tree-checker for | |
75 | * instance will return EUCLEAN if any of the blocks are corrupted in | |
76 | * a way that is problematic. We want to reserve EUCLEAN for these | |
77 | * sort of corruptions. | |
78 | * | |
79 | * EROFS: If we check BTRFS_FS_STATE_ERROR and fail out with a return error, we | |
80 | * need to use EROFS for this case. We will have no idea of the | |
81 | * original failure, that will have been reported at the time we tripped | |
82 | * over the error. Each subsequent error that doesn't have any context | |
83 | * of the original error should use EROFS when handling BTRFS_FS_STATE_ERROR. | |
84 | */ | |
4143cb8b | 85 | const char * __attribute_const__ btrfs_decode_error(int errno) |
acce952b | 86 | { |
08748810 | 87 | char *errstr = "unknown"; |
acce952b | 88 | |
89 | switch (errno) { | |
d54f8144 DS |
90 | case -ENOENT: /* -2 */ |
91 | errstr = "No such entry"; | |
92 | break; | |
93 | case -EIO: /* -5 */ | |
acce952b | 94 | errstr = "IO failure"; |
95 | break; | |
d54f8144 | 96 | case -ENOMEM: /* -12*/ |
acce952b | 97 | errstr = "Out of memory"; |
98 | break; | |
d54f8144 | 99 | case -EEXIST: /* -17 */ |
8c342930 JM |
100 | errstr = "Object already exists"; |
101 | break; | |
d54f8144 | 102 | case -ENOSPC: /* -28 */ |
94ef7280 DS |
103 | errstr = "No space left"; |
104 | break; | |
d54f8144 DS |
105 | case -EROFS: /* -30 */ |
106 | errstr = "Readonly filesystem"; | |
94ef7280 | 107 | break; |
fb8521ca DS |
108 | case -EOPNOTSUPP: /* -95 */ |
109 | errstr = "Operation not supported"; | |
110 | break; | |
111 | case -EUCLEAN: /* -117 */ | |
112 | errstr = "Filesystem corrupted"; | |
113 | break; | |
114 | case -EDQUOT: /* -122 */ | |
115 | errstr = "Quota exceeded"; | |
116 | break; | |
acce952b | 117 | } |
118 | ||
119 | return errstr; | |
120 | } | |
121 | ||
acce952b | 122 | /* |
34d97007 | 123 | * __btrfs_handle_fs_error decodes expected errors from the caller and |
52042d8e | 124 | * invokes the appropriate error response. |
acce952b | 125 | */ |
c0d19e2b | 126 | __cold |
34d97007 | 127 | void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function, |
4da35113 | 128 | unsigned int line, int errno, const char *fmt, ...) |
acce952b | 129 | { |
130 | struct super_block *sb = fs_info->sb; | |
57d816a1 | 131 | #ifdef CONFIG_PRINTK |
acce952b | 132 | const char *errstr; |
57d816a1 | 133 | #endif |
acce952b | 134 | |
135 | /* | |
136 | * Special case: if the error is EROFS, and we're already | |
1751e8a6 | 137 | * under SB_RDONLY, then it is safe here. |
acce952b | 138 | */ |
bc98a42c | 139 | if (errno == -EROFS && sb_rdonly(sb)) |
4da35113 JM |
140 | return; |
141 | ||
57d816a1 | 142 | #ifdef CONFIG_PRINTK |
08748810 | 143 | errstr = btrfs_decode_error(errno); |
4da35113 | 144 | if (fmt) { |
37252a66 ES |
145 | struct va_format vaf; |
146 | va_list args; | |
147 | ||
148 | va_start(args, fmt); | |
149 | vaf.fmt = fmt; | |
150 | vaf.va = &args; | |
4da35113 | 151 | |
62e85577 | 152 | pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n", |
08748810 | 153 | sb->s_id, function, line, errno, errstr, &vaf); |
37252a66 | 154 | va_end(args); |
4da35113 | 155 | } else { |
62e85577 | 156 | pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s\n", |
08748810 | 157 | sb->s_id, function, line, errno, errstr); |
4da35113 | 158 | } |
57d816a1 | 159 | #endif |
acce952b | 160 | |
0713d90c AJ |
161 | /* |
162 | * Today we only save the error info to memory. Long term we'll | |
163 | * also send it down to the disk | |
164 | */ | |
165 | set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state); | |
166 | ||
4da35113 | 167 | /* Don't go through full error handling during mount */ |
922ea899 AJ |
168 | if (!(sb->s_flags & SB_BORN)) |
169 | return; | |
170 | ||
171 | if (sb_rdonly(sb)) | |
172 | return; | |
173 | ||
b0643e59 DZ |
174 | btrfs_discard_stop(fs_info); |
175 | ||
922ea899 AJ |
176 | /* btrfs handle error by forcing the filesystem readonly */ |
177 | sb->s_flags |= SB_RDONLY; | |
178 | btrfs_info(fs_info, "forced readonly"); | |
179 | /* | |
180 | * Note that a running device replace operation is not canceled here | |
181 | * although there is no way to update the progress. It would add the | |
182 | * risk of a deadlock, therefore the canceling is omitted. The only | |
183 | * penalty is that some I/O remains active until the procedure | |
52042d8e | 184 | * completes. The next time when the filesystem is mounted writable |
922ea899 AJ |
185 | * again, the device replace operation continues. |
186 | */ | |
4da35113 | 187 | } |
acce952b | 188 | |
57d816a1 | 189 | #ifdef CONFIG_PRINTK |
533574c6 | 190 | static const char * const logtypes[] = { |
4da35113 JM |
191 | "emergency", |
192 | "alert", | |
193 | "critical", | |
194 | "error", | |
195 | "warning", | |
196 | "notice", | |
197 | "info", | |
198 | "debug", | |
199 | }; | |
200 | ||
35f4e5e6 NB |
201 | |
202 | /* | |
203 | * Use one ratelimit state per log level so that a flood of less important | |
204 | * messages doesn't cause more important ones to be dropped. | |
205 | */ | |
206 | static struct ratelimit_state printk_limits[] = { | |
207 | RATELIMIT_STATE_INIT(printk_limits[0], DEFAULT_RATELIMIT_INTERVAL, 100), | |
208 | RATELIMIT_STATE_INIT(printk_limits[1], DEFAULT_RATELIMIT_INTERVAL, 100), | |
209 | RATELIMIT_STATE_INIT(printk_limits[2], DEFAULT_RATELIMIT_INTERVAL, 100), | |
210 | RATELIMIT_STATE_INIT(printk_limits[3], DEFAULT_RATELIMIT_INTERVAL, 100), | |
211 | RATELIMIT_STATE_INIT(printk_limits[4], DEFAULT_RATELIMIT_INTERVAL, 100), | |
212 | RATELIMIT_STATE_INIT(printk_limits[5], DEFAULT_RATELIMIT_INTERVAL, 100), | |
213 | RATELIMIT_STATE_INIT(printk_limits[6], DEFAULT_RATELIMIT_INTERVAL, 100), | |
214 | RATELIMIT_STATE_INIT(printk_limits[7], DEFAULT_RATELIMIT_INTERVAL, 100), | |
215 | }; | |
216 | ||
b105e927 | 217 | void __cold btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...) |
4da35113 | 218 | { |
40f7828b | 219 | char lvl[PRINTK_MAX_SINGLE_HEADER_LEN + 1] = "\0"; |
4da35113 JM |
220 | struct va_format vaf; |
221 | va_list args; | |
533574c6 | 222 | int kern_level; |
40f7828b PM |
223 | const char *type = logtypes[4]; |
224 | struct ratelimit_state *ratelimit = &printk_limits[4]; | |
4da35113 JM |
225 | |
226 | va_start(args, fmt); | |
227 | ||
262c5e86 | 228 | while ((kern_level = printk_get_level(fmt)) != 0) { |
533574c6 | 229 | size_t size = printk_skip_level(fmt) - fmt; |
262c5e86 PM |
230 | |
231 | if (kern_level >= '0' && kern_level <= '7') { | |
232 | memcpy(lvl, fmt, size); | |
233 | lvl[size] = '\0'; | |
234 | type = logtypes[kern_level - '0']; | |
235 | ratelimit = &printk_limits[kern_level - '0']; | |
236 | } | |
533574c6 | 237 | fmt += size; |
262c5e86 PM |
238 | } |
239 | ||
4da35113 JM |
240 | vaf.fmt = fmt; |
241 | vaf.va = &args; | |
533574c6 | 242 | |
35f4e5e6 | 243 | if (__ratelimit(ratelimit)) |
3993b112 CIK |
244 | printk("%sBTRFS %s (device %s): %pV\n", lvl, type, |
245 | fs_info ? fs_info->sb->s_id : "<unknown>", &vaf); | |
533574c6 JP |
246 | |
247 | va_end(args); | |
248 | } | |
533574c6 | 249 | #endif |
acce952b | 250 | |
49b25e05 JM |
251 | /* |
252 | * We only mark the transaction aborted and then set the file system read-only. | |
253 | * This will prevent new transactions from starting or trying to join this | |
254 | * one. | |
255 | * | |
256 | * This means that error recovery at the call site is limited to freeing | |
257 | * any local memory allocations and passing the error code up without | |
258 | * further cleanup. The transaction should complete as it normally would | |
259 | * in the call path but will return -EIO. | |
260 | * | |
261 | * We'll complete the cleanup in btrfs_end_transaction and | |
262 | * btrfs_commit_transaction. | |
263 | */ | |
c0d19e2b | 264 | __cold |
49b25e05 | 265 | void __btrfs_abort_transaction(struct btrfs_trans_handle *trans, |
66642832 | 266 | const char *function, |
49b25e05 JM |
267 | unsigned int line, int errno) |
268 | { | |
66642832 JM |
269 | struct btrfs_fs_info *fs_info = trans->fs_info; |
270 | ||
bf31f87f | 271 | WRITE_ONCE(trans->aborted, errno); |
49b25e05 JM |
272 | /* Nothing used. The other threads that have joined this |
273 | * transaction may be able to continue. */ | |
64c12921 | 274 | if (!trans->dirty && list_empty(&trans->new_bgs)) { |
69ce977a MX |
275 | const char *errstr; |
276 | ||
08748810 | 277 | errstr = btrfs_decode_error(errno); |
66642832 | 278 | btrfs_warn(fs_info, |
c2cf52eb SK |
279 | "%s:%d: Aborting unused transaction(%s).", |
280 | function, line, errstr); | |
acce952b | 281 | return; |
49b25e05 | 282 | } |
20c7bcec | 283 | WRITE_ONCE(trans->transaction->aborted, errno); |
501407aa | 284 | /* Wake up anybody who may be waiting on this transaction */ |
66642832 JM |
285 | wake_up(&fs_info->transaction_wait); |
286 | wake_up(&fs_info->transaction_blocked_wait); | |
287 | __btrfs_handle_fs_error(fs_info, function, line, errno, NULL); | |
49b25e05 | 288 | } |
8c342930 JM |
289 | /* |
290 | * __btrfs_panic decodes unexpected, fatal errors from the caller, | |
291 | * issues an alert, and either panics or BUGs, depending on mount options. | |
292 | */ | |
c0d19e2b | 293 | __cold |
8c342930 JM |
294 | void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function, |
295 | unsigned int line, int errno, const char *fmt, ...) | |
296 | { | |
8c342930 JM |
297 | char *s_id = "<unknown>"; |
298 | const char *errstr; | |
299 | struct va_format vaf = { .fmt = fmt }; | |
300 | va_list args; | |
acce952b | 301 | |
8c342930 JM |
302 | if (fs_info) |
303 | s_id = fs_info->sb->s_id; | |
acce952b | 304 | |
8c342930 JM |
305 | va_start(args, fmt); |
306 | vaf.va = &args; | |
307 | ||
08748810 | 308 | errstr = btrfs_decode_error(errno); |
d8953d69 | 309 | if (fs_info && (btrfs_test_opt(fs_info, PANIC_ON_FATAL_ERROR))) |
08748810 DS |
310 | panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n", |
311 | s_id, function, line, &vaf, errno, errstr); | |
8c342930 | 312 | |
efe120a0 FH |
313 | btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)", |
314 | function, line, &vaf, errno, errstr); | |
8c342930 JM |
315 | va_end(args); |
316 | /* Caller calls BUG() */ | |
acce952b | 317 | } |
318 | ||
d397712b | 319 | static void btrfs_put_super(struct super_block *sb) |
b18c6685 | 320 | { |
6bccf3ab | 321 | close_ctree(btrfs_sb(sb)); |
75dfe396 CM |
322 | } |
323 | ||
95e05289 | 324 | enum { |
416a7202 DS |
325 | Opt_acl, Opt_noacl, |
326 | Opt_clear_cache, | |
327 | Opt_commit_interval, | |
328 | Opt_compress, | |
329 | Opt_compress_force, | |
330 | Opt_compress_force_type, | |
331 | Opt_compress_type, | |
332 | Opt_degraded, | |
333 | Opt_device, | |
334 | Opt_fatal_errors, | |
335 | Opt_flushoncommit, Opt_noflushoncommit, | |
336 | Opt_inode_cache, Opt_noinode_cache, | |
337 | Opt_max_inline, | |
338 | Opt_barrier, Opt_nobarrier, | |
339 | Opt_datacow, Opt_nodatacow, | |
340 | Opt_datasum, Opt_nodatasum, | |
341 | Opt_defrag, Opt_nodefrag, | |
342 | Opt_discard, Opt_nodiscard, | |
b0643e59 | 343 | Opt_discard_mode, |
416a7202 DS |
344 | Opt_norecovery, |
345 | Opt_ratio, | |
346 | Opt_rescan_uuid_tree, | |
347 | Opt_skip_balance, | |
348 | Opt_space_cache, Opt_no_space_cache, | |
349 | Opt_space_cache_version, | |
350 | Opt_ssd, Opt_nossd, | |
351 | Opt_ssd_spread, Opt_nossd_spread, | |
352 | Opt_subvol, | |
37becec9 | 353 | Opt_subvol_empty, |
416a7202 DS |
354 | Opt_subvolid, |
355 | Opt_thread_pool, | |
356 | Opt_treelog, Opt_notreelog, | |
416a7202 DS |
357 | Opt_user_subvol_rm_allowed, |
358 | ||
74ef0018 QW |
359 | /* Rescue options */ |
360 | Opt_rescue, | |
361 | Opt_usebackuproot, | |
362 | Opt_nologreplay, | |
42437a63 | 363 | Opt_ignorebadroots, |
882dbe0c | 364 | Opt_ignoredatacsums, |
9037d3cb | 365 | Opt_rescue_all, |
74ef0018 | 366 | |
416a7202 | 367 | /* Deprecated options */ |
416a7202 | 368 | Opt_recovery, |
416a7202 DS |
369 | |
370 | /* Debugging options */ | |
371 | Opt_check_integrity, | |
70f6d82e | 372 | Opt_check_integrity_including_extent_data, |
416a7202 DS |
373 | Opt_check_integrity_print_mask, |
374 | Opt_enospc_debug, Opt_noenospc_debug, | |
d0bd4560 JB |
375 | #ifdef CONFIG_BTRFS_DEBUG |
376 | Opt_fragment_data, Opt_fragment_metadata, Opt_fragment_all, | |
fb592373 JB |
377 | #endif |
378 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
379 | Opt_ref_verify, | |
d0bd4560 | 380 | #endif |
9555c6c1 | 381 | Opt_err, |
95e05289 CM |
382 | }; |
383 | ||
4d4ab6d6 | 384 | static const match_table_t tokens = { |
416a7202 DS |
385 | {Opt_acl, "acl"}, |
386 | {Opt_noacl, "noacl"}, | |
387 | {Opt_clear_cache, "clear_cache"}, | |
388 | {Opt_commit_interval, "commit=%u"}, | |
c8b97818 | 389 | {Opt_compress, "compress"}, |
261507a0 | 390 | {Opt_compress_type, "compress=%s"}, |
a555f810 | 391 | {Opt_compress_force, "compress-force"}, |
261507a0 | 392 | {Opt_compress_force_type, "compress-force=%s"}, |
416a7202 DS |
393 | {Opt_degraded, "degraded"}, |
394 | {Opt_device, "device=%s"}, | |
395 | {Opt_fatal_errors, "fatal_errors=%s"}, | |
dccae999 | 396 | {Opt_flushoncommit, "flushoncommit"}, |
2c9ee856 | 397 | {Opt_noflushoncommit, "noflushoncommit"}, |
416a7202 DS |
398 | {Opt_inode_cache, "inode_cache"}, |
399 | {Opt_noinode_cache, "noinode_cache"}, | |
400 | {Opt_max_inline, "max_inline=%s"}, | |
401 | {Opt_barrier, "barrier"}, | |
402 | {Opt_nobarrier, "nobarrier"}, | |
403 | {Opt_datacow, "datacow"}, | |
404 | {Opt_nodatacow, "nodatacow"}, | |
405 | {Opt_datasum, "datasum"}, | |
406 | {Opt_nodatasum, "nodatasum"}, | |
407 | {Opt_defrag, "autodefrag"}, | |
408 | {Opt_nodefrag, "noautodefrag"}, | |
e244a0ae | 409 | {Opt_discard, "discard"}, |
b0643e59 | 410 | {Opt_discard_mode, "discard=%s"}, |
e07a2ade | 411 | {Opt_nodiscard, "nodiscard"}, |
416a7202 DS |
412 | {Opt_norecovery, "norecovery"}, |
413 | {Opt_ratio, "metadata_ratio=%u"}, | |
414 | {Opt_rescan_uuid_tree, "rescan_uuid_tree"}, | |
415 | {Opt_skip_balance, "skip_balance"}, | |
0af3d00b | 416 | {Opt_space_cache, "space_cache"}, |
8965593e | 417 | {Opt_no_space_cache, "nospace_cache"}, |
416a7202 DS |
418 | {Opt_space_cache_version, "space_cache=%s"}, |
419 | {Opt_ssd, "ssd"}, | |
420 | {Opt_nossd, "nossd"}, | |
421 | {Opt_ssd_spread, "ssd_spread"}, | |
422 | {Opt_nossd_spread, "nossd_spread"}, | |
423 | {Opt_subvol, "subvol=%s"}, | |
37becec9 | 424 | {Opt_subvol_empty, "subvol="}, |
416a7202 DS |
425 | {Opt_subvolid, "subvolid=%s"}, |
426 | {Opt_thread_pool, "thread_pool=%u"}, | |
427 | {Opt_treelog, "treelog"}, | |
428 | {Opt_notreelog, "notreelog"}, | |
416a7202 DS |
429 | {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"}, |
430 | ||
74ef0018 QW |
431 | /* Rescue options */ |
432 | {Opt_rescue, "rescue=%s"}, | |
433 | /* Deprecated, with alias rescue=nologreplay */ | |
434 | {Opt_nologreplay, "nologreplay"}, | |
435 | /* Deprecated, with alias rescue=usebackuproot */ | |
436 | {Opt_usebackuproot, "usebackuproot"}, | |
437 | ||
416a7202 | 438 | /* Deprecated options */ |
416a7202 | 439 | {Opt_recovery, "recovery"}, |
416a7202 DS |
440 | |
441 | /* Debugging options */ | |
21adbd5c SB |
442 | {Opt_check_integrity, "check_int"}, |
443 | {Opt_check_integrity_including_extent_data, "check_int_data"}, | |
02453bde | 444 | {Opt_check_integrity_print_mask, "check_int_print_mask=%u"}, |
416a7202 DS |
445 | {Opt_enospc_debug, "enospc_debug"}, |
446 | {Opt_noenospc_debug, "noenospc_debug"}, | |
d0bd4560 JB |
447 | #ifdef CONFIG_BTRFS_DEBUG |
448 | {Opt_fragment_data, "fragment=data"}, | |
449 | {Opt_fragment_metadata, "fragment=metadata"}, | |
450 | {Opt_fragment_all, "fragment=all"}, | |
fb592373 JB |
451 | #endif |
452 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
453 | {Opt_ref_verify, "ref_verify"}, | |
d0bd4560 | 454 | #endif |
33268eaf | 455 | {Opt_err, NULL}, |
95e05289 CM |
456 | }; |
457 | ||
74ef0018 QW |
458 | static const match_table_t rescue_tokens = { |
459 | {Opt_usebackuproot, "usebackuproot"}, | |
460 | {Opt_nologreplay, "nologreplay"}, | |
42437a63 JB |
461 | {Opt_ignorebadroots, "ignorebadroots"}, |
462 | {Opt_ignorebadroots, "ibadroots"}, | |
882dbe0c JB |
463 | {Opt_ignoredatacsums, "ignoredatacsums"}, |
464 | {Opt_ignoredatacsums, "idatacsums"}, | |
9037d3cb | 465 | {Opt_rescue_all, "all"}, |
74ef0018 QW |
466 | {Opt_err, NULL}, |
467 | }; | |
468 | ||
d70bf748 JB |
469 | static bool check_ro_option(struct btrfs_fs_info *fs_info, unsigned long opt, |
470 | const char *opt_name) | |
471 | { | |
472 | if (fs_info->mount_opt & opt) { | |
473 | btrfs_err(fs_info, "%s must be used with ro mount option", | |
474 | opt_name); | |
475 | return true; | |
476 | } | |
477 | return false; | |
478 | } | |
479 | ||
74ef0018 QW |
480 | static int parse_rescue_options(struct btrfs_fs_info *info, const char *options) |
481 | { | |
482 | char *opts; | |
483 | char *orig; | |
484 | char *p; | |
485 | substring_t args[MAX_OPT_ARGS]; | |
486 | int ret = 0; | |
487 | ||
488 | opts = kstrdup(options, GFP_KERNEL); | |
489 | if (!opts) | |
490 | return -ENOMEM; | |
491 | orig = opts; | |
492 | ||
493 | while ((p = strsep(&opts, ":")) != NULL) { | |
494 | int token; | |
495 | ||
496 | if (!*p) | |
497 | continue; | |
498 | token = match_token(p, rescue_tokens, args); | |
499 | switch (token){ | |
500 | case Opt_usebackuproot: | |
501 | btrfs_info(info, | |
502 | "trying to use backup root at mount time"); | |
503 | btrfs_set_opt(info->mount_opt, USEBACKUPROOT); | |
504 | break; | |
505 | case Opt_nologreplay: | |
506 | btrfs_set_and_info(info, NOLOGREPLAY, | |
507 | "disabling log replay at mount time"); | |
508 | break; | |
42437a63 JB |
509 | case Opt_ignorebadroots: |
510 | btrfs_set_and_info(info, IGNOREBADROOTS, | |
511 | "ignoring bad roots"); | |
512 | break; | |
882dbe0c JB |
513 | case Opt_ignoredatacsums: |
514 | btrfs_set_and_info(info, IGNOREDATACSUMS, | |
515 | "ignoring data csums"); | |
516 | break; | |
9037d3cb JB |
517 | case Opt_rescue_all: |
518 | btrfs_info(info, "enabling all of the rescue options"); | |
519 | btrfs_set_and_info(info, IGNOREDATACSUMS, | |
520 | "ignoring data csums"); | |
521 | btrfs_set_and_info(info, IGNOREBADROOTS, | |
522 | "ignoring bad roots"); | |
523 | btrfs_set_and_info(info, NOLOGREPLAY, | |
524 | "disabling log replay at mount time"); | |
525 | break; | |
74ef0018 QW |
526 | case Opt_err: |
527 | btrfs_info(info, "unrecognized rescue option '%s'", p); | |
528 | ret = -EINVAL; | |
529 | goto out; | |
530 | default: | |
531 | break; | |
532 | } | |
533 | ||
534 | } | |
535 | out: | |
536 | kfree(orig); | |
537 | return ret; | |
538 | } | |
539 | ||
edf24abe CH |
540 | /* |
541 | * Regular mount options parser. Everything that is needed only when | |
542 | * reading in a new superblock is parsed here. | |
49b25e05 | 543 | * XXX JDM: This needs to be cleaned up for remount. |
edf24abe | 544 | */ |
2ff7e61e | 545 | int btrfs_parse_options(struct btrfs_fs_info *info, char *options, |
96da0919 | 546 | unsigned long new_flags) |
95e05289 | 547 | { |
95e05289 | 548 | substring_t args[MAX_OPT_ARGS]; |
e215772c | 549 | char *p, *num; |
73bc1876 | 550 | u64 cache_gen; |
4543df7e | 551 | int intarg; |
a7a3f7ca | 552 | int ret = 0; |
261507a0 LZ |
553 | char *compress_type; |
554 | bool compress_force = false; | |
b7c47bbb | 555 | enum btrfs_compression_type saved_compress_type; |
27942c99 | 556 | int saved_compress_level; |
b7c47bbb TI |
557 | bool saved_compress_force; |
558 | int no_compress = 0; | |
b6cda9bc | 559 | |
0b246afa JM |
560 | cache_gen = btrfs_super_cache_generation(info->super_copy); |
561 | if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE)) | |
70f6d82e OS |
562 | btrfs_set_opt(info->mount_opt, FREE_SPACE_TREE); |
563 | else if (cache_gen) | |
73bc1876 JB |
564 | btrfs_set_opt(info->mount_opt, SPACE_CACHE); |
565 | ||
96da0919 QW |
566 | /* |
567 | * Even the options are empty, we still need to do extra check | |
568 | * against new flags | |
569 | */ | |
95e05289 | 570 | if (!options) |
96da0919 | 571 | goto check; |
95e05289 | 572 | |
edf24abe | 573 | while ((p = strsep(&options, ",")) != NULL) { |
95e05289 CM |
574 | int token; |
575 | if (!*p) | |
576 | continue; | |
577 | ||
578 | token = match_token(p, tokens, args); | |
579 | switch (token) { | |
dfe25020 | 580 | case Opt_degraded: |
0b246afa | 581 | btrfs_info(info, "allowing degraded mounts"); |
edf24abe | 582 | btrfs_set_opt(info->mount_opt, DEGRADED); |
dfe25020 | 583 | break; |
95e05289 | 584 | case Opt_subvol: |
37becec9 | 585 | case Opt_subvol_empty: |
73f73415 | 586 | case Opt_subvolid: |
43e570b0 | 587 | case Opt_device: |
edf24abe | 588 | /* |
fa59f27c AJ |
589 | * These are parsed by btrfs_parse_subvol_options or |
590 | * btrfs_parse_device_options and can be ignored here. | |
edf24abe | 591 | */ |
b6cda9bc CM |
592 | break; |
593 | case Opt_nodatasum: | |
3cdde224 | 594 | btrfs_set_and_info(info, NODATASUM, |
07802534 | 595 | "setting nodatasum"); |
be20aa9d | 596 | break; |
d399167d | 597 | case Opt_datasum: |
3cdde224 JM |
598 | if (btrfs_test_opt(info, NODATASUM)) { |
599 | if (btrfs_test_opt(info, NODATACOW)) | |
0b246afa | 600 | btrfs_info(info, |
5d163e0e | 601 | "setting datasum, datacow enabled"); |
07802534 | 602 | else |
0b246afa | 603 | btrfs_info(info, "setting datasum"); |
07802534 | 604 | } |
d399167d QW |
605 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
606 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
607 | break; | |
be20aa9d | 608 | case Opt_nodatacow: |
3cdde224 JM |
609 | if (!btrfs_test_opt(info, NODATACOW)) { |
610 | if (!btrfs_test_opt(info, COMPRESS) || | |
611 | !btrfs_test_opt(info, FORCE_COMPRESS)) { | |
0b246afa | 612 | btrfs_info(info, |
07802534 QW |
613 | "setting nodatacow, compression disabled"); |
614 | } else { | |
0b246afa | 615 | btrfs_info(info, "setting nodatacow"); |
07802534 | 616 | } |
bedb2cca | 617 | } |
bedb2cca AP |
618 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
619 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
edf24abe CH |
620 | btrfs_set_opt(info->mount_opt, NODATACOW); |
621 | btrfs_set_opt(info->mount_opt, NODATASUM); | |
95e05289 | 622 | break; |
a258af7a | 623 | case Opt_datacow: |
3cdde224 | 624 | btrfs_clear_and_info(info, NODATACOW, |
07802534 | 625 | "setting datacow"); |
a258af7a | 626 | break; |
a555f810 | 627 | case Opt_compress_force: |
261507a0 LZ |
628 | case Opt_compress_force_type: |
629 | compress_force = true; | |
c730ae0c | 630 | fallthrough; |
261507a0 LZ |
631 | case Opt_compress: |
632 | case Opt_compress_type: | |
3cdde224 JM |
633 | saved_compress_type = btrfs_test_opt(info, |
634 | COMPRESS) ? | |
b7c47bbb TI |
635 | info->compress_type : BTRFS_COMPRESS_NONE; |
636 | saved_compress_force = | |
3cdde224 | 637 | btrfs_test_opt(info, FORCE_COMPRESS); |
27942c99 | 638 | saved_compress_level = info->compress_level; |
261507a0 LZ |
639 | if (token == Opt_compress || |
640 | token == Opt_compress_force || | |
a7164fa4 | 641 | strncmp(args[0].from, "zlib", 4) == 0) { |
261507a0 | 642 | compress_type = "zlib"; |
eae8d825 | 643 | |
261507a0 | 644 | info->compress_type = BTRFS_COMPRESS_ZLIB; |
eae8d825 QW |
645 | info->compress_level = BTRFS_ZLIB_DEFAULT_LEVEL; |
646 | /* | |
647 | * args[0] contains uninitialized data since | |
648 | * for these tokens we don't expect any | |
649 | * parameter. | |
650 | */ | |
651 | if (token != Opt_compress && | |
652 | token != Opt_compress_force) | |
653 | info->compress_level = | |
d0ab62ce DZ |
654 | btrfs_compress_str2level( |
655 | BTRFS_COMPRESS_ZLIB, | |
656 | args[0].from + 4); | |
063849ea | 657 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
658 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
659 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
b7c47bbb | 660 | no_compress = 0; |
a7164fa4 | 661 | } else if (strncmp(args[0].from, "lzo", 3) == 0) { |
a6fa6fae LZ |
662 | compress_type = "lzo"; |
663 | info->compress_type = BTRFS_COMPRESS_LZO; | |
282dd7d7 | 664 | info->compress_level = 0; |
063849ea | 665 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
666 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
667 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
2b0ce2c2 | 668 | btrfs_set_fs_incompat(info, COMPRESS_LZO); |
b7c47bbb | 669 | no_compress = 0; |
3f93aef5 | 670 | } else if (strncmp(args[0].from, "zstd", 4) == 0) { |
5c1aab1d NT |
671 | compress_type = "zstd"; |
672 | info->compress_type = BTRFS_COMPRESS_ZSTD; | |
3f93aef5 DZ |
673 | info->compress_level = |
674 | btrfs_compress_str2level( | |
675 | BTRFS_COMPRESS_ZSTD, | |
676 | args[0].from + 4); | |
5c1aab1d NT |
677 | btrfs_set_opt(info->mount_opt, COMPRESS); |
678 | btrfs_clear_opt(info->mount_opt, NODATACOW); | |
679 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
680 | btrfs_set_fs_incompat(info, COMPRESS_ZSTD); | |
681 | no_compress = 0; | |
063849ea AH |
682 | } else if (strncmp(args[0].from, "no", 2) == 0) { |
683 | compress_type = "no"; | |
27942c99 DS |
684 | info->compress_level = 0; |
685 | info->compress_type = 0; | |
063849ea AH |
686 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
687 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
688 | compress_force = false; | |
b7c47bbb | 689 | no_compress++; |
261507a0 LZ |
690 | } else { |
691 | ret = -EINVAL; | |
692 | goto out; | |
693 | } | |
694 | ||
261507a0 | 695 | if (compress_force) { |
b7c47bbb | 696 | btrfs_set_opt(info->mount_opt, FORCE_COMPRESS); |
143f3636 | 697 | } else { |
4027e0f4 WS |
698 | /* |
699 | * If we remount from compress-force=xxx to | |
700 | * compress=xxx, we need clear FORCE_COMPRESS | |
701 | * flag, otherwise, there is no way for users | |
702 | * to disable forcible compression separately. | |
703 | */ | |
704 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
a7e252af | 705 | } |
27942c99 DS |
706 | if (no_compress == 1) { |
707 | btrfs_info(info, "use no compression"); | |
708 | } else if ((info->compress_type != saved_compress_type) || | |
709 | (compress_force != saved_compress_force) || | |
710 | (info->compress_level != saved_compress_level)) { | |
f51d2b59 | 711 | btrfs_info(info, "%s %s compression, level %d", |
b7c47bbb | 712 | (compress_force) ? "force" : "use", |
f51d2b59 | 713 | compress_type, info->compress_level); |
b7c47bbb TI |
714 | } |
715 | compress_force = false; | |
a555f810 | 716 | break; |
e18e4809 | 717 | case Opt_ssd: |
3cdde224 | 718 | btrfs_set_and_info(info, SSD, |
583b7231 | 719 | "enabling ssd optimizations"); |
951e7966 | 720 | btrfs_clear_opt(info->mount_opt, NOSSD); |
e18e4809 | 721 | break; |
451d7585 | 722 | case Opt_ssd_spread: |
583b7231 HK |
723 | btrfs_set_and_info(info, SSD, |
724 | "enabling ssd optimizations"); | |
3cdde224 | 725 | btrfs_set_and_info(info, SSD_SPREAD, |
583b7231 | 726 | "using spread ssd allocation scheme"); |
951e7966 | 727 | btrfs_clear_opt(info->mount_opt, NOSSD); |
451d7585 | 728 | break; |
3b30c22f | 729 | case Opt_nossd: |
583b7231 HK |
730 | btrfs_set_opt(info->mount_opt, NOSSD); |
731 | btrfs_clear_and_info(info, SSD, | |
732 | "not using ssd optimizations"); | |
c730ae0c | 733 | fallthrough; |
62b8e077 | 734 | case Opt_nossd_spread: |
583b7231 HK |
735 | btrfs_clear_and_info(info, SSD_SPREAD, |
736 | "not using spread ssd allocation scheme"); | |
3b30c22f | 737 | break; |
842bef58 | 738 | case Opt_barrier: |
3cdde224 | 739 | btrfs_clear_and_info(info, NOBARRIER, |
07802534 | 740 | "turning on barriers"); |
842bef58 | 741 | break; |
21ad10cf | 742 | case Opt_nobarrier: |
3cdde224 | 743 | btrfs_set_and_info(info, NOBARRIER, |
07802534 | 744 | "turning off barriers"); |
21ad10cf | 745 | break; |
4543df7e | 746 | case Opt_thread_pool: |
2c334e87 WS |
747 | ret = match_int(&args[0], &intarg); |
748 | if (ret) { | |
749 | goto out; | |
f7b885be | 750 | } else if (intarg == 0) { |
2c334e87 WS |
751 | ret = -EINVAL; |
752 | goto out; | |
753 | } | |
f7b885be | 754 | info->thread_pool_size = intarg; |
4543df7e | 755 | break; |
6f568d35 | 756 | case Opt_max_inline: |
edf24abe CH |
757 | num = match_strdup(&args[0]); |
758 | if (num) { | |
91748467 | 759 | info->max_inline = memparse(num, NULL); |
edf24abe CH |
760 | kfree(num); |
761 | ||
15ada040 | 762 | if (info->max_inline) { |
feb5f965 | 763 | info->max_inline = min_t(u64, |
15ada040 | 764 | info->max_inline, |
0b246afa | 765 | info->sectorsize); |
15ada040 | 766 | } |
0b246afa JM |
767 | btrfs_info(info, "max_inline at %llu", |
768 | info->max_inline); | |
2c334e87 WS |
769 | } else { |
770 | ret = -ENOMEM; | |
771 | goto out; | |
6f568d35 CM |
772 | } |
773 | break; | |
bd0330ad | 774 | case Opt_acl: |
45ff35d6 | 775 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
1751e8a6 | 776 | info->sb->s_flags |= SB_POSIXACL; |
bd0330ad | 777 | break; |
45ff35d6 | 778 | #else |
0b246afa | 779 | btrfs_err(info, "support for ACL not compiled in!"); |
45ff35d6 GZ |
780 | ret = -EINVAL; |
781 | goto out; | |
782 | #endif | |
33268eaf | 783 | case Opt_noacl: |
1751e8a6 | 784 | info->sb->s_flags &= ~SB_POSIXACL; |
33268eaf | 785 | break; |
3a5e1404 | 786 | case Opt_notreelog: |
3cdde224 | 787 | btrfs_set_and_info(info, NOTREELOG, |
07802534 | 788 | "disabling tree log"); |
a88998f2 QW |
789 | break; |
790 | case Opt_treelog: | |
3cdde224 | 791 | btrfs_clear_and_info(info, NOTREELOG, |
07802534 | 792 | "enabling tree log"); |
3a5e1404 | 793 | break; |
fed8f166 | 794 | case Opt_norecovery: |
96da0919 | 795 | case Opt_nologreplay: |
74ef0018 QW |
796 | btrfs_warn(info, |
797 | "'nologreplay' is deprecated, use 'rescue=nologreplay' instead"); | |
3cdde224 | 798 | btrfs_set_and_info(info, NOLOGREPLAY, |
96da0919 QW |
799 | "disabling log replay at mount time"); |
800 | break; | |
dccae999 | 801 | case Opt_flushoncommit: |
3cdde224 | 802 | btrfs_set_and_info(info, FLUSHONCOMMIT, |
07802534 | 803 | "turning on flush-on-commit"); |
dccae999 | 804 | break; |
2c9ee856 | 805 | case Opt_noflushoncommit: |
3cdde224 | 806 | btrfs_clear_and_info(info, FLUSHONCOMMIT, |
07802534 | 807 | "turning off flush-on-commit"); |
2c9ee856 | 808 | break; |
97e728d4 | 809 | case Opt_ratio: |
2c334e87 | 810 | ret = match_int(&args[0], &intarg); |
764cb8b4 | 811 | if (ret) |
2c334e87 | 812 | goto out; |
764cb8b4 AJ |
813 | info->metadata_ratio = intarg; |
814 | btrfs_info(info, "metadata ratio %u", | |
815 | info->metadata_ratio); | |
97e728d4 | 816 | break; |
e244a0ae | 817 | case Opt_discard: |
b0643e59 DZ |
818 | case Opt_discard_mode: |
819 | if (token == Opt_discard || | |
820 | strcmp(args[0].from, "sync") == 0) { | |
821 | btrfs_clear_opt(info->mount_opt, DISCARD_ASYNC); | |
822 | btrfs_set_and_info(info, DISCARD_SYNC, | |
823 | "turning on sync discard"); | |
824 | } else if (strcmp(args[0].from, "async") == 0) { | |
825 | btrfs_clear_opt(info->mount_opt, DISCARD_SYNC); | |
826 | btrfs_set_and_info(info, DISCARD_ASYNC, | |
827 | "turning on async discard"); | |
828 | } else { | |
829 | ret = -EINVAL; | |
830 | goto out; | |
831 | } | |
e244a0ae | 832 | break; |
e07a2ade | 833 | case Opt_nodiscard: |
46b27f50 | 834 | btrfs_clear_and_info(info, DISCARD_SYNC, |
07802534 | 835 | "turning off discard"); |
b0643e59 DZ |
836 | btrfs_clear_and_info(info, DISCARD_ASYNC, |
837 | "turning off async discard"); | |
e07a2ade | 838 | break; |
0af3d00b | 839 | case Opt_space_cache: |
70f6d82e OS |
840 | case Opt_space_cache_version: |
841 | if (token == Opt_space_cache || | |
842 | strcmp(args[0].from, "v1") == 0) { | |
0b246afa | 843 | btrfs_clear_opt(info->mount_opt, |
70f6d82e | 844 | FREE_SPACE_TREE); |
3cdde224 | 845 | btrfs_set_and_info(info, SPACE_CACHE, |
0b246afa | 846 | "enabling disk space caching"); |
70f6d82e | 847 | } else if (strcmp(args[0].from, "v2") == 0) { |
0b246afa | 848 | btrfs_clear_opt(info->mount_opt, |
70f6d82e | 849 | SPACE_CACHE); |
0b246afa | 850 | btrfs_set_and_info(info, FREE_SPACE_TREE, |
70f6d82e OS |
851 | "enabling free space tree"); |
852 | } else { | |
853 | ret = -EINVAL; | |
854 | goto out; | |
855 | } | |
0de90876 | 856 | break; |
f420ee1e SB |
857 | case Opt_rescan_uuid_tree: |
858 | btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE); | |
859 | break; | |
73bc1876 | 860 | case Opt_no_space_cache: |
3cdde224 | 861 | if (btrfs_test_opt(info, SPACE_CACHE)) { |
0b246afa JM |
862 | btrfs_clear_and_info(info, SPACE_CACHE, |
863 | "disabling disk space caching"); | |
70f6d82e | 864 | } |
3cdde224 | 865 | if (btrfs_test_opt(info, FREE_SPACE_TREE)) { |
0b246afa JM |
866 | btrfs_clear_and_info(info, FREE_SPACE_TREE, |
867 | "disabling free space tree"); | |
70f6d82e | 868 | } |
73bc1876 | 869 | break; |
4b9465cb | 870 | case Opt_inode_cache: |
b547a88e DS |
871 | btrfs_warn(info, |
872 | "the 'inode_cache' option is deprecated and will have no effect from 5.11"); | |
7e1876ac | 873 | btrfs_set_pending_and_info(info, INODE_MAP_CACHE, |
07802534 | 874 | "enabling inode map caching"); |
3818aea2 QW |
875 | break; |
876 | case Opt_noinode_cache: | |
7e1876ac | 877 | btrfs_clear_pending_and_info(info, INODE_MAP_CACHE, |
07802534 | 878 | "disabling inode map caching"); |
4b9465cb | 879 | break; |
88c2ba3b | 880 | case Opt_clear_cache: |
3cdde224 | 881 | btrfs_set_and_info(info, CLEAR_CACHE, |
07802534 | 882 | "force clearing of disk cache"); |
0af3d00b | 883 | break; |
4260f7c7 SW |
884 | case Opt_user_subvol_rm_allowed: |
885 | btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED); | |
886 | break; | |
91435650 CM |
887 | case Opt_enospc_debug: |
888 | btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG); | |
889 | break; | |
53036293 QW |
890 | case Opt_noenospc_debug: |
891 | btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG); | |
892 | break; | |
4cb5300b | 893 | case Opt_defrag: |
3cdde224 | 894 | btrfs_set_and_info(info, AUTO_DEFRAG, |
07802534 | 895 | "enabling auto defrag"); |
4cb5300b | 896 | break; |
fc0ca9af | 897 | case Opt_nodefrag: |
3cdde224 | 898 | btrfs_clear_and_info(info, AUTO_DEFRAG, |
07802534 | 899 | "disabling auto defrag"); |
fc0ca9af | 900 | break; |
af31f5e5 | 901 | case Opt_recovery: |
8dcddfa0 | 902 | case Opt_usebackuproot: |
74ef0018 QW |
903 | btrfs_warn(info, |
904 | "'%s' is deprecated, use 'rescue=usebackuproot' instead", | |
905 | token == Opt_recovery ? "recovery" : | |
906 | "usebackuproot"); | |
0b246afa | 907 | btrfs_info(info, |
8dcddfa0 QW |
908 | "trying to use backup root at mount time"); |
909 | btrfs_set_opt(info->mount_opt, USEBACKUPROOT); | |
af31f5e5 | 910 | break; |
9555c6c1 ID |
911 | case Opt_skip_balance: |
912 | btrfs_set_opt(info->mount_opt, SKIP_BALANCE); | |
913 | break; | |
21adbd5c SB |
914 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
915 | case Opt_check_integrity_including_extent_data: | |
0b246afa | 916 | btrfs_info(info, |
efe120a0 | 917 | "enabling check integrity including extent data"); |
21adbd5c SB |
918 | btrfs_set_opt(info->mount_opt, |
919 | CHECK_INTEGRITY_INCLUDING_EXTENT_DATA); | |
920 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); | |
921 | break; | |
922 | case Opt_check_integrity: | |
0b246afa | 923 | btrfs_info(info, "enabling check integrity"); |
21adbd5c SB |
924 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); |
925 | break; | |
926 | case Opt_check_integrity_print_mask: | |
2c334e87 | 927 | ret = match_int(&args[0], &intarg); |
02453bde | 928 | if (ret) |
2c334e87 | 929 | goto out; |
02453bde AJ |
930 | info->check_integrity_print_mask = intarg; |
931 | btrfs_info(info, "check_integrity_print_mask 0x%x", | |
932 | info->check_integrity_print_mask); | |
21adbd5c SB |
933 | break; |
934 | #else | |
935 | case Opt_check_integrity_including_extent_data: | |
936 | case Opt_check_integrity: | |
937 | case Opt_check_integrity_print_mask: | |
0b246afa JM |
938 | btrfs_err(info, |
939 | "support for check_integrity* not compiled in!"); | |
21adbd5c SB |
940 | ret = -EINVAL; |
941 | goto out; | |
942 | #endif | |
8c342930 JM |
943 | case Opt_fatal_errors: |
944 | if (strcmp(args[0].from, "panic") == 0) | |
945 | btrfs_set_opt(info->mount_opt, | |
946 | PANIC_ON_FATAL_ERROR); | |
947 | else if (strcmp(args[0].from, "bug") == 0) | |
948 | btrfs_clear_opt(info->mount_opt, | |
949 | PANIC_ON_FATAL_ERROR); | |
950 | else { | |
951 | ret = -EINVAL; | |
952 | goto out; | |
953 | } | |
954 | break; | |
8b87dc17 DS |
955 | case Opt_commit_interval: |
956 | intarg = 0; | |
957 | ret = match_int(&args[0], &intarg); | |
d3740608 | 958 | if (ret) |
8b87dc17 | 959 | goto out; |
d3740608 | 960 | if (intarg == 0) { |
0b246afa | 961 | btrfs_info(info, |
d3740608 | 962 | "using default commit interval %us", |
5d163e0e | 963 | BTRFS_DEFAULT_COMMIT_INTERVAL); |
d3740608 AJ |
964 | intarg = BTRFS_DEFAULT_COMMIT_INTERVAL; |
965 | } else if (intarg > 300) { | |
966 | btrfs_warn(info, "excessive commit interval %d", | |
967 | intarg); | |
8b87dc17 | 968 | } |
d3740608 | 969 | info->commit_interval = intarg; |
8b87dc17 | 970 | break; |
74ef0018 QW |
971 | case Opt_rescue: |
972 | ret = parse_rescue_options(info, args[0].from); | |
973 | if (ret < 0) | |
974 | goto out; | |
975 | break; | |
d0bd4560 JB |
976 | #ifdef CONFIG_BTRFS_DEBUG |
977 | case Opt_fragment_all: | |
0b246afa | 978 | btrfs_info(info, "fragmenting all space"); |
d0bd4560 JB |
979 | btrfs_set_opt(info->mount_opt, FRAGMENT_DATA); |
980 | btrfs_set_opt(info->mount_opt, FRAGMENT_METADATA); | |
981 | break; | |
982 | case Opt_fragment_metadata: | |
0b246afa | 983 | btrfs_info(info, "fragmenting metadata"); |
d0bd4560 JB |
984 | btrfs_set_opt(info->mount_opt, |
985 | FRAGMENT_METADATA); | |
986 | break; | |
987 | case Opt_fragment_data: | |
0b246afa | 988 | btrfs_info(info, "fragmenting data"); |
d0bd4560 JB |
989 | btrfs_set_opt(info->mount_opt, FRAGMENT_DATA); |
990 | break; | |
fb592373 JB |
991 | #endif |
992 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
993 | case Opt_ref_verify: | |
994 | btrfs_info(info, "doing ref verification"); | |
995 | btrfs_set_opt(info->mount_opt, REF_VERIFY); | |
996 | break; | |
d0bd4560 | 997 | #endif |
a7a3f7ca | 998 | case Opt_err: |
7e8f19e5 | 999 | btrfs_err(info, "unrecognized mount option '%s'", p); |
a7a3f7ca SW |
1000 | ret = -EINVAL; |
1001 | goto out; | |
95e05289 | 1002 | default: |
be20aa9d | 1003 | break; |
95e05289 CM |
1004 | } |
1005 | } | |
96da0919 | 1006 | check: |
d70bf748 JB |
1007 | /* We're read-only, don't have to check. */ |
1008 | if (new_flags & SB_RDONLY) | |
1009 | goto out; | |
1010 | ||
42437a63 | 1011 | if (check_ro_option(info, BTRFS_MOUNT_NOLOGREPLAY, "nologreplay") || |
882dbe0c JB |
1012 | check_ro_option(info, BTRFS_MOUNT_IGNOREBADROOTS, "ignorebadroots") || |
1013 | check_ro_option(info, BTRFS_MOUNT_IGNOREDATACSUMS, "ignoredatacsums")) | |
96da0919 | 1014 | ret = -EINVAL; |
a7a3f7ca | 1015 | out: |
0b246afa | 1016 | if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE) && |
3cdde224 JM |
1017 | !btrfs_test_opt(info, FREE_SPACE_TREE) && |
1018 | !btrfs_test_opt(info, CLEAR_CACHE)) { | |
0b246afa | 1019 | btrfs_err(info, "cannot disable free space tree"); |
70f6d82e OS |
1020 | ret = -EINVAL; |
1021 | ||
1022 | } | |
3cdde224 | 1023 | if (!ret && btrfs_test_opt(info, SPACE_CACHE)) |
0b246afa | 1024 | btrfs_info(info, "disk space caching is enabled"); |
3cdde224 | 1025 | if (!ret && btrfs_test_opt(info, FREE_SPACE_TREE)) |
0b246afa | 1026 | btrfs_info(info, "using free space tree"); |
a7a3f7ca | 1027 | return ret; |
edf24abe CH |
1028 | } |
1029 | ||
1030 | /* | |
1031 | * Parse mount options that are required early in the mount process. | |
1032 | * | |
1033 | * All other options will be parsed on much later in the mount process and | |
1034 | * only when we need to allocate a new super block. | |
1035 | */ | |
fa59f27c AJ |
1036 | static int btrfs_parse_device_options(const char *options, fmode_t flags, |
1037 | void *holder) | |
edf24abe CH |
1038 | { |
1039 | substring_t args[MAX_OPT_ARGS]; | |
83c8c9bd | 1040 | char *device_name, *opts, *orig, *p; |
36350e95 | 1041 | struct btrfs_device *device = NULL; |
d7407606 MT |
1042 | int error = 0; |
1043 | ||
5139cff5 DS |
1044 | lockdep_assert_held(&uuid_mutex); |
1045 | ||
d7407606 MT |
1046 | if (!options) |
1047 | return 0; | |
1048 | ||
1049 | /* | |
1050 | * strsep changes the string, duplicate it because btrfs_parse_options | |
1051 | * gets called later | |
1052 | */ | |
1053 | opts = kstrdup(options, GFP_KERNEL); | |
1054 | if (!opts) | |
1055 | return -ENOMEM; | |
1056 | orig = opts; | |
1057 | ||
1058 | while ((p = strsep(&opts, ",")) != NULL) { | |
1059 | int token; | |
1060 | ||
1061 | if (!*p) | |
1062 | continue; | |
1063 | ||
1064 | token = match_token(p, tokens, args); | |
1065 | if (token == Opt_device) { | |
1066 | device_name = match_strdup(&args[0]); | |
1067 | if (!device_name) { | |
1068 | error = -ENOMEM; | |
1069 | goto out; | |
1070 | } | |
36350e95 GJ |
1071 | device = btrfs_scan_one_device(device_name, flags, |
1072 | holder); | |
d7407606 | 1073 | kfree(device_name); |
36350e95 GJ |
1074 | if (IS_ERR(device)) { |
1075 | error = PTR_ERR(device); | |
d7407606 | 1076 | goto out; |
36350e95 | 1077 | } |
d7407606 MT |
1078 | } |
1079 | } | |
1080 | ||
1081 | out: | |
1082 | kfree(orig); | |
1083 | return error; | |
1084 | } | |
1085 | ||
1086 | /* | |
1087 | * Parse mount options that are related to subvolume id | |
1088 | * | |
1089 | * The value is later passed to mount_subvol() | |
1090 | */ | |
93b9bcdf GJ |
1091 | static int btrfs_parse_subvol_options(const char *options, char **subvol_name, |
1092 | u64 *subvol_objectid) | |
d7407606 MT |
1093 | { |
1094 | substring_t args[MAX_OPT_ARGS]; | |
1095 | char *opts, *orig, *p; | |
edf24abe | 1096 | int error = 0; |
ccb0e7d1 | 1097 | u64 subvolid; |
edf24abe CH |
1098 | |
1099 | if (!options) | |
830c4adb | 1100 | return 0; |
edf24abe CH |
1101 | |
1102 | /* | |
d7407606 | 1103 | * strsep changes the string, duplicate it because |
fa59f27c | 1104 | * btrfs_parse_device_options gets called later |
edf24abe CH |
1105 | */ |
1106 | opts = kstrdup(options, GFP_KERNEL); | |
1107 | if (!opts) | |
1108 | return -ENOMEM; | |
3f3d0bc0 | 1109 | orig = opts; |
edf24abe CH |
1110 | |
1111 | while ((p = strsep(&opts, ",")) != NULL) { | |
1112 | int token; | |
1113 | if (!*p) | |
1114 | continue; | |
1115 | ||
1116 | token = match_token(p, tokens, args); | |
1117 | switch (token) { | |
1118 | case Opt_subvol: | |
a90e8b6f | 1119 | kfree(*subvol_name); |
edf24abe | 1120 | *subvol_name = match_strdup(&args[0]); |
2c334e87 WS |
1121 | if (!*subvol_name) { |
1122 | error = -ENOMEM; | |
1123 | goto out; | |
1124 | } | |
edf24abe | 1125 | break; |
73f73415 | 1126 | case Opt_subvolid: |
ccb0e7d1 AJ |
1127 | error = match_u64(&args[0], &subvolid); |
1128 | if (error) | |
2c334e87 | 1129 | goto out; |
ccb0e7d1 AJ |
1130 | |
1131 | /* we want the original fs_tree */ | |
1132 | if (subvolid == 0) | |
1133 | subvolid = BTRFS_FS_TREE_OBJECTID; | |
1134 | ||
1135 | *subvol_objectid = subvolid; | |
73f73415 | 1136 | break; |
edf24abe CH |
1137 | default: |
1138 | break; | |
1139 | } | |
1140 | } | |
1141 | ||
830c4adb | 1142 | out: |
3f3d0bc0 | 1143 | kfree(orig); |
edf24abe | 1144 | return error; |
95e05289 CM |
1145 | } |
1146 | ||
c0c907a4 MPS |
1147 | char *btrfs_get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info, |
1148 | u64 subvol_objectid) | |
73f73415 | 1149 | { |
815745cf | 1150 | struct btrfs_root *root = fs_info->tree_root; |
5168489a | 1151 | struct btrfs_root *fs_root = NULL; |
05dbe683 OS |
1152 | struct btrfs_root_ref *root_ref; |
1153 | struct btrfs_inode_ref *inode_ref; | |
1154 | struct btrfs_key key; | |
1155 | struct btrfs_path *path = NULL; | |
1156 | char *name = NULL, *ptr; | |
1157 | u64 dirid; | |
1158 | int len; | |
1159 | int ret; | |
1160 | ||
1161 | path = btrfs_alloc_path(); | |
1162 | if (!path) { | |
1163 | ret = -ENOMEM; | |
1164 | goto err; | |
1165 | } | |
1166 | path->leave_spinning = 1; | |
1167 | ||
3ec83621 | 1168 | name = kmalloc(PATH_MAX, GFP_KERNEL); |
05dbe683 OS |
1169 | if (!name) { |
1170 | ret = -ENOMEM; | |
1171 | goto err; | |
1172 | } | |
1173 | ptr = name + PATH_MAX - 1; | |
1174 | ptr[0] = '\0'; | |
73f73415 JB |
1175 | |
1176 | /* | |
05dbe683 OS |
1177 | * Walk up the subvolume trees in the tree of tree roots by root |
1178 | * backrefs until we hit the top-level subvolume. | |
73f73415 | 1179 | */ |
05dbe683 OS |
1180 | while (subvol_objectid != BTRFS_FS_TREE_OBJECTID) { |
1181 | key.objectid = subvol_objectid; | |
1182 | key.type = BTRFS_ROOT_BACKREF_KEY; | |
1183 | key.offset = (u64)-1; | |
1184 | ||
1185 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
1186 | if (ret < 0) { | |
1187 | goto err; | |
1188 | } else if (ret > 0) { | |
1189 | ret = btrfs_previous_item(root, path, subvol_objectid, | |
1190 | BTRFS_ROOT_BACKREF_KEY); | |
1191 | if (ret < 0) { | |
1192 | goto err; | |
1193 | } else if (ret > 0) { | |
1194 | ret = -ENOENT; | |
1195 | goto err; | |
1196 | } | |
1197 | } | |
1198 | ||
1199 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1200 | subvol_objectid = key.offset; | |
1201 | ||
1202 | root_ref = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
1203 | struct btrfs_root_ref); | |
1204 | len = btrfs_root_ref_name_len(path->nodes[0], root_ref); | |
1205 | ptr -= len + 1; | |
1206 | if (ptr < name) { | |
1207 | ret = -ENAMETOOLONG; | |
1208 | goto err; | |
1209 | } | |
1210 | read_extent_buffer(path->nodes[0], ptr + 1, | |
1211 | (unsigned long)(root_ref + 1), len); | |
1212 | ptr[0] = '/'; | |
1213 | dirid = btrfs_root_ref_dirid(path->nodes[0], root_ref); | |
1214 | btrfs_release_path(path); | |
1215 | ||
56e9357a | 1216 | fs_root = btrfs_get_fs_root(fs_info, subvol_objectid, true); |
05dbe683 OS |
1217 | if (IS_ERR(fs_root)) { |
1218 | ret = PTR_ERR(fs_root); | |
5168489a JB |
1219 | fs_root = NULL; |
1220 | goto err; | |
1221 | } | |
05dbe683 OS |
1222 | |
1223 | /* | |
1224 | * Walk up the filesystem tree by inode refs until we hit the | |
1225 | * root directory. | |
1226 | */ | |
1227 | while (dirid != BTRFS_FIRST_FREE_OBJECTID) { | |
1228 | key.objectid = dirid; | |
1229 | key.type = BTRFS_INODE_REF_KEY; | |
1230 | key.offset = (u64)-1; | |
1231 | ||
1232 | ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0); | |
1233 | if (ret < 0) { | |
1234 | goto err; | |
1235 | } else if (ret > 0) { | |
1236 | ret = btrfs_previous_item(fs_root, path, dirid, | |
1237 | BTRFS_INODE_REF_KEY); | |
1238 | if (ret < 0) { | |
1239 | goto err; | |
1240 | } else if (ret > 0) { | |
1241 | ret = -ENOENT; | |
1242 | goto err; | |
1243 | } | |
1244 | } | |
1245 | ||
1246 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1247 | dirid = key.offset; | |
1248 | ||
1249 | inode_ref = btrfs_item_ptr(path->nodes[0], | |
1250 | path->slots[0], | |
1251 | struct btrfs_inode_ref); | |
1252 | len = btrfs_inode_ref_name_len(path->nodes[0], | |
1253 | inode_ref); | |
1254 | ptr -= len + 1; | |
1255 | if (ptr < name) { | |
1256 | ret = -ENAMETOOLONG; | |
1257 | goto err; | |
1258 | } | |
1259 | read_extent_buffer(path->nodes[0], ptr + 1, | |
1260 | (unsigned long)(inode_ref + 1), len); | |
1261 | ptr[0] = '/'; | |
1262 | btrfs_release_path(path); | |
1263 | } | |
00246528 | 1264 | btrfs_put_root(fs_root); |
5168489a | 1265 | fs_root = NULL; |
73f73415 JB |
1266 | } |
1267 | ||
05dbe683 OS |
1268 | btrfs_free_path(path); |
1269 | if (ptr == name + PATH_MAX - 1) { | |
1270 | name[0] = '/'; | |
1271 | name[1] = '\0'; | |
1272 | } else { | |
1273 | memmove(name, ptr, name + PATH_MAX - ptr); | |
1274 | } | |
1275 | return name; | |
1276 | ||
1277 | err: | |
00246528 | 1278 | btrfs_put_root(fs_root); |
05dbe683 OS |
1279 | btrfs_free_path(path); |
1280 | kfree(name); | |
1281 | return ERR_PTR(ret); | |
1282 | } | |
1283 | ||
1284 | static int get_default_subvol_objectid(struct btrfs_fs_info *fs_info, u64 *objectid) | |
1285 | { | |
1286 | struct btrfs_root *root = fs_info->tree_root; | |
1287 | struct btrfs_dir_item *di; | |
1288 | struct btrfs_path *path; | |
1289 | struct btrfs_key location; | |
1290 | u64 dir_id; | |
1291 | ||
73f73415 JB |
1292 | path = btrfs_alloc_path(); |
1293 | if (!path) | |
05dbe683 | 1294 | return -ENOMEM; |
73f73415 JB |
1295 | path->leave_spinning = 1; |
1296 | ||
1297 | /* | |
1298 | * Find the "default" dir item which points to the root item that we | |
1299 | * will mount by default if we haven't been given a specific subvolume | |
1300 | * to mount. | |
1301 | */ | |
815745cf | 1302 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
73f73415 | 1303 | di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0); |
b0839166 JL |
1304 | if (IS_ERR(di)) { |
1305 | btrfs_free_path(path); | |
05dbe683 | 1306 | return PTR_ERR(di); |
b0839166 | 1307 | } |
73f73415 JB |
1308 | if (!di) { |
1309 | /* | |
1310 | * Ok the default dir item isn't there. This is weird since | |
1311 | * it's always been there, but don't freak out, just try and | |
05dbe683 | 1312 | * mount the top-level subvolume. |
73f73415 JB |
1313 | */ |
1314 | btrfs_free_path(path); | |
05dbe683 OS |
1315 | *objectid = BTRFS_FS_TREE_OBJECTID; |
1316 | return 0; | |
73f73415 JB |
1317 | } |
1318 | ||
1319 | btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); | |
1320 | btrfs_free_path(path); | |
05dbe683 OS |
1321 | *objectid = location.objectid; |
1322 | return 0; | |
73f73415 JB |
1323 | } |
1324 | ||
d397712b | 1325 | static int btrfs_fill_super(struct super_block *sb, |
8a4b83cc | 1326 | struct btrfs_fs_devices *fs_devices, |
56e033a7 | 1327 | void *data) |
75dfe396 | 1328 | { |
d397712b | 1329 | struct inode *inode; |
815745cf | 1330 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
39279cc3 | 1331 | int err; |
a429e513 | 1332 | |
39279cc3 CM |
1333 | sb->s_maxbytes = MAX_LFS_FILESIZE; |
1334 | sb->s_magic = BTRFS_SUPER_MAGIC; | |
1335 | sb->s_op = &btrfs_super_ops; | |
af53d29a | 1336 | sb->s_d_op = &btrfs_dentry_operations; |
be6e8dc0 | 1337 | sb->s_export_op = &btrfs_export_ops; |
5103e947 | 1338 | sb->s_xattr = btrfs_xattr_handlers; |
39279cc3 | 1339 | sb->s_time_gran = 1; |
0eda294d | 1340 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
1751e8a6 | 1341 | sb->s_flags |= SB_POSIXACL; |
49cf6f45 | 1342 | #endif |
357fdad0 | 1343 | sb->s_flags |= SB_I_VERSION; |
da2f0f74 | 1344 | sb->s_iflags |= SB_I_CGROUPWB; |
9e11ceee JK |
1345 | |
1346 | err = super_setup_bdi(sb); | |
1347 | if (err) { | |
1348 | btrfs_err(fs_info, "super_setup_bdi failed"); | |
1349 | return err; | |
1350 | } | |
1351 | ||
ad2b2c80 AV |
1352 | err = open_ctree(sb, fs_devices, (char *)data); |
1353 | if (err) { | |
ab8d0fc4 | 1354 | btrfs_err(fs_info, "open_ctree failed"); |
ad2b2c80 | 1355 | return err; |
a429e513 CM |
1356 | } |
1357 | ||
0202e83f | 1358 | inode = btrfs_iget(sb, BTRFS_FIRST_FREE_OBJECTID, fs_info->fs_root); |
5d4f98a2 YZ |
1359 | if (IS_ERR(inode)) { |
1360 | err = PTR_ERR(inode); | |
39279cc3 | 1361 | goto fail_close; |
f254e52c | 1362 | } |
f254e52c | 1363 | |
48fde701 AV |
1364 | sb->s_root = d_make_root(inode); |
1365 | if (!sb->s_root) { | |
39279cc3 CM |
1366 | err = -ENOMEM; |
1367 | goto fail_close; | |
f254e52c | 1368 | } |
58176a96 | 1369 | |
90a887c9 | 1370 | cleancache_init_fs(sb); |
1751e8a6 | 1371 | sb->s_flags |= SB_ACTIVE; |
2619ba1f | 1372 | return 0; |
39279cc3 CM |
1373 | |
1374 | fail_close: | |
6bccf3ab | 1375 | close_ctree(fs_info); |
39279cc3 | 1376 | return err; |
2619ba1f CM |
1377 | } |
1378 | ||
6bf13c0c | 1379 | int btrfs_sync_fs(struct super_block *sb, int wait) |
c5739bba CM |
1380 | { |
1381 | struct btrfs_trans_handle *trans; | |
815745cf AV |
1382 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
1383 | struct btrfs_root *root = fs_info->tree_root; | |
2619ba1f | 1384 | |
bc074524 | 1385 | trace_btrfs_sync_fs(fs_info, wait); |
1abe9b8a | 1386 | |
39279cc3 | 1387 | if (!wait) { |
815745cf | 1388 | filemap_flush(fs_info->btree_inode->i_mapping); |
39279cc3 CM |
1389 | return 0; |
1390 | } | |
771ed689 | 1391 | |
6374e57a | 1392 | btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); |
771ed689 | 1393 | |
d4edf39b | 1394 | trans = btrfs_attach_transaction_barrier(root); |
60376ce4 | 1395 | if (IS_ERR(trans)) { |
354aa0fb | 1396 | /* no transaction, don't bother */ |
6b5fe46d DS |
1397 | if (PTR_ERR(trans) == -ENOENT) { |
1398 | /* | |
1399 | * Exit unless we have some pending changes | |
1400 | * that need to go through commit | |
1401 | */ | |
1402 | if (fs_info->pending_changes == 0) | |
1403 | return 0; | |
a53f4f8e QW |
1404 | /* |
1405 | * A non-blocking test if the fs is frozen. We must not | |
1406 | * start a new transaction here otherwise a deadlock | |
1407 | * happens. The pending operations are delayed to the | |
1408 | * next commit after thawing. | |
1409 | */ | |
a7e3c5f2 RP |
1410 | if (sb_start_write_trylock(sb)) |
1411 | sb_end_write(sb); | |
a53f4f8e QW |
1412 | else |
1413 | return 0; | |
6b5fe46d | 1414 | trans = btrfs_start_transaction(root, 0); |
6b5fe46d | 1415 | } |
98bd5c54 DS |
1416 | if (IS_ERR(trans)) |
1417 | return PTR_ERR(trans); | |
60376ce4 | 1418 | } |
3a45bb20 | 1419 | return btrfs_commit_transaction(trans); |
2c90e5d6 CM |
1420 | } |
1421 | ||
ab0b4a3e JB |
1422 | static void print_rescue_option(struct seq_file *seq, const char *s, bool *printed) |
1423 | { | |
1424 | seq_printf(seq, "%s%s", (*printed) ? ":" : ",rescue=", s); | |
1425 | *printed = true; | |
1426 | } | |
1427 | ||
34c80b1d | 1428 | static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry) |
a9572a15 | 1429 | { |
815745cf | 1430 | struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb); |
0f628c63 | 1431 | const char *compress_type; |
3ef3959b | 1432 | const char *subvol_name; |
ab0b4a3e | 1433 | bool printed = false; |
a9572a15 | 1434 | |
3cdde224 | 1435 | if (btrfs_test_opt(info, DEGRADED)) |
a9572a15 | 1436 | seq_puts(seq, ",degraded"); |
3cdde224 | 1437 | if (btrfs_test_opt(info, NODATASUM)) |
a9572a15 | 1438 | seq_puts(seq, ",nodatasum"); |
3cdde224 | 1439 | if (btrfs_test_opt(info, NODATACOW)) |
a9572a15 | 1440 | seq_puts(seq, ",nodatacow"); |
3cdde224 | 1441 | if (btrfs_test_opt(info, NOBARRIER)) |
a9572a15 | 1442 | seq_puts(seq, ",nobarrier"); |
95ac567a | 1443 | if (info->max_inline != BTRFS_DEFAULT_MAX_INLINE) |
c1c9ff7c | 1444 | seq_printf(seq, ",max_inline=%llu", info->max_inline); |
a9572a15 EP |
1445 | if (info->thread_pool_size != min_t(unsigned long, |
1446 | num_online_cpus() + 2, 8)) | |
f7b885be | 1447 | seq_printf(seq, ",thread_pool=%u", info->thread_pool_size); |
3cdde224 | 1448 | if (btrfs_test_opt(info, COMPRESS)) { |
0f628c63 | 1449 | compress_type = btrfs_compress_type2str(info->compress_type); |
3cdde224 | 1450 | if (btrfs_test_opt(info, FORCE_COMPRESS)) |
200da64e TI |
1451 | seq_printf(seq, ",compress-force=%s", compress_type); |
1452 | else | |
1453 | seq_printf(seq, ",compress=%s", compress_type); | |
f51d2b59 | 1454 | if (info->compress_level) |
fa4d885a | 1455 | seq_printf(seq, ":%d", info->compress_level); |
200da64e | 1456 | } |
3cdde224 | 1457 | if (btrfs_test_opt(info, NOSSD)) |
c289811c | 1458 | seq_puts(seq, ",nossd"); |
3cdde224 | 1459 | if (btrfs_test_opt(info, SSD_SPREAD)) |
451d7585 | 1460 | seq_puts(seq, ",ssd_spread"); |
3cdde224 | 1461 | else if (btrfs_test_opt(info, SSD)) |
a9572a15 | 1462 | seq_puts(seq, ",ssd"); |
3cdde224 | 1463 | if (btrfs_test_opt(info, NOTREELOG)) |
6b65c5c6 | 1464 | seq_puts(seq, ",notreelog"); |
3cdde224 | 1465 | if (btrfs_test_opt(info, NOLOGREPLAY)) |
ab0b4a3e | 1466 | print_rescue_option(seq, "nologreplay", &printed); |
68319c18 JB |
1467 | if (btrfs_test_opt(info, USEBACKUPROOT)) |
1468 | print_rescue_option(seq, "usebackuproot", &printed); | |
42437a63 JB |
1469 | if (btrfs_test_opt(info, IGNOREBADROOTS)) |
1470 | print_rescue_option(seq, "ignorebadroots", &printed); | |
882dbe0c JB |
1471 | if (btrfs_test_opt(info, IGNOREDATACSUMS)) |
1472 | print_rescue_option(seq, "ignoredatacsums", &printed); | |
3cdde224 | 1473 | if (btrfs_test_opt(info, FLUSHONCOMMIT)) |
6b65c5c6 | 1474 | seq_puts(seq, ",flushoncommit"); |
46b27f50 | 1475 | if (btrfs_test_opt(info, DISCARD_SYNC)) |
20a5239a | 1476 | seq_puts(seq, ",discard"); |
b0643e59 DZ |
1477 | if (btrfs_test_opt(info, DISCARD_ASYNC)) |
1478 | seq_puts(seq, ",discard=async"); | |
1751e8a6 | 1479 | if (!(info->sb->s_flags & SB_POSIXACL)) |
a9572a15 | 1480 | seq_puts(seq, ",noacl"); |
3cdde224 | 1481 | if (btrfs_test_opt(info, SPACE_CACHE)) |
200da64e | 1482 | seq_puts(seq, ",space_cache"); |
3cdde224 | 1483 | else if (btrfs_test_opt(info, FREE_SPACE_TREE)) |
70f6d82e | 1484 | seq_puts(seq, ",space_cache=v2"); |
73bc1876 | 1485 | else |
8965593e | 1486 | seq_puts(seq, ",nospace_cache"); |
3cdde224 | 1487 | if (btrfs_test_opt(info, RESCAN_UUID_TREE)) |
f420ee1e | 1488 | seq_puts(seq, ",rescan_uuid_tree"); |
3cdde224 | 1489 | if (btrfs_test_opt(info, CLEAR_CACHE)) |
200da64e | 1490 | seq_puts(seq, ",clear_cache"); |
3cdde224 | 1491 | if (btrfs_test_opt(info, USER_SUBVOL_RM_ALLOWED)) |
200da64e | 1492 | seq_puts(seq, ",user_subvol_rm_allowed"); |
3cdde224 | 1493 | if (btrfs_test_opt(info, ENOSPC_DEBUG)) |
0942caa3 | 1494 | seq_puts(seq, ",enospc_debug"); |
3cdde224 | 1495 | if (btrfs_test_opt(info, AUTO_DEFRAG)) |
0942caa3 | 1496 | seq_puts(seq, ",autodefrag"); |
3cdde224 | 1497 | if (btrfs_test_opt(info, INODE_MAP_CACHE)) |
0942caa3 | 1498 | seq_puts(seq, ",inode_cache"); |
3cdde224 | 1499 | if (btrfs_test_opt(info, SKIP_BALANCE)) |
9555c6c1 | 1500 | seq_puts(seq, ",skip_balance"); |
8507d216 | 1501 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
3cdde224 | 1502 | if (btrfs_test_opt(info, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA)) |
8507d216 | 1503 | seq_puts(seq, ",check_int_data"); |
3cdde224 | 1504 | else if (btrfs_test_opt(info, CHECK_INTEGRITY)) |
8507d216 WS |
1505 | seq_puts(seq, ",check_int"); |
1506 | if (info->check_integrity_print_mask) | |
1507 | seq_printf(seq, ",check_int_print_mask=%d", | |
1508 | info->check_integrity_print_mask); | |
1509 | #endif | |
1510 | if (info->metadata_ratio) | |
764cb8b4 | 1511 | seq_printf(seq, ",metadata_ratio=%u", info->metadata_ratio); |
3cdde224 | 1512 | if (btrfs_test_opt(info, PANIC_ON_FATAL_ERROR)) |
8c342930 | 1513 | seq_puts(seq, ",fatal_errors=panic"); |
8b87dc17 | 1514 | if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL) |
d3740608 | 1515 | seq_printf(seq, ",commit=%u", info->commit_interval); |
d0bd4560 | 1516 | #ifdef CONFIG_BTRFS_DEBUG |
3cdde224 | 1517 | if (btrfs_test_opt(info, FRAGMENT_DATA)) |
d0bd4560 | 1518 | seq_puts(seq, ",fragment=data"); |
3cdde224 | 1519 | if (btrfs_test_opt(info, FRAGMENT_METADATA)) |
d0bd4560 JB |
1520 | seq_puts(seq, ",fragment=metadata"); |
1521 | #endif | |
fb592373 JB |
1522 | if (btrfs_test_opt(info, REF_VERIFY)) |
1523 | seq_puts(seq, ",ref_verify"); | |
c8d3fe02 OS |
1524 | seq_printf(seq, ",subvolid=%llu", |
1525 | BTRFS_I(d_inode(dentry))->root->root_key.objectid); | |
3ef3959b JB |
1526 | subvol_name = btrfs_get_subvol_name_from_objectid(info, |
1527 | BTRFS_I(d_inode(dentry))->root->root_key.objectid); | |
1528 | if (!IS_ERR(subvol_name)) { | |
1529 | seq_puts(seq, ",subvol="); | |
1530 | seq_escape(seq, subvol_name, " \t\n\\"); | |
1531 | kfree(subvol_name); | |
1532 | } | |
a9572a15 EP |
1533 | return 0; |
1534 | } | |
1535 | ||
a061fc8d | 1536 | static int btrfs_test_super(struct super_block *s, void *data) |
4b82d6e4 | 1537 | { |
815745cf AV |
1538 | struct btrfs_fs_info *p = data; |
1539 | struct btrfs_fs_info *fs_info = btrfs_sb(s); | |
4b82d6e4 | 1540 | |
815745cf | 1541 | return fs_info->fs_devices == p->fs_devices; |
4b82d6e4 Y |
1542 | } |
1543 | ||
450ba0ea JB |
1544 | static int btrfs_set_super(struct super_block *s, void *data) |
1545 | { | |
6de1d09d AV |
1546 | int err = set_anon_super(s, data); |
1547 | if (!err) | |
1548 | s->s_fs_info = data; | |
1549 | return err; | |
4b82d6e4 Y |
1550 | } |
1551 | ||
f9d9ef62 DS |
1552 | /* |
1553 | * subvolumes are identified by ino 256 | |
1554 | */ | |
1555 | static inline int is_subvolume_inode(struct inode *inode) | |
1556 | { | |
1557 | if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) | |
1558 | return 1; | |
1559 | return 0; | |
1560 | } | |
1561 | ||
bb289b7b | 1562 | static struct dentry *mount_subvol(const char *subvol_name, u64 subvol_objectid, |
ae0bc863 | 1563 | struct vfsmount *mnt) |
830c4adb | 1564 | { |
830c4adb | 1565 | struct dentry *root; |
fa330659 | 1566 | int ret; |
830c4adb | 1567 | |
05dbe683 OS |
1568 | if (!subvol_name) { |
1569 | if (!subvol_objectid) { | |
1570 | ret = get_default_subvol_objectid(btrfs_sb(mnt->mnt_sb), | |
1571 | &subvol_objectid); | |
1572 | if (ret) { | |
1573 | root = ERR_PTR(ret); | |
1574 | goto out; | |
1575 | } | |
1576 | } | |
c0c907a4 MPS |
1577 | subvol_name = btrfs_get_subvol_name_from_objectid( |
1578 | btrfs_sb(mnt->mnt_sb), subvol_objectid); | |
05dbe683 OS |
1579 | if (IS_ERR(subvol_name)) { |
1580 | root = ERR_CAST(subvol_name); | |
1581 | subvol_name = NULL; | |
1582 | goto out; | |
1583 | } | |
1584 | ||
1585 | } | |
1586 | ||
ea441d11 | 1587 | root = mount_subtree(mnt, subvol_name); |
fa330659 OS |
1588 | /* mount_subtree() drops our reference on the vfsmount. */ |
1589 | mnt = NULL; | |
830c4adb | 1590 | |
bb289b7b | 1591 | if (!IS_ERR(root)) { |
ea441d11 | 1592 | struct super_block *s = root->d_sb; |
ab8d0fc4 | 1593 | struct btrfs_fs_info *fs_info = btrfs_sb(s); |
bb289b7b OS |
1594 | struct inode *root_inode = d_inode(root); |
1595 | u64 root_objectid = BTRFS_I(root_inode)->root->root_key.objectid; | |
1596 | ||
1597 | ret = 0; | |
1598 | if (!is_subvolume_inode(root_inode)) { | |
ab8d0fc4 | 1599 | btrfs_err(fs_info, "'%s' is not a valid subvolume", |
bb289b7b OS |
1600 | subvol_name); |
1601 | ret = -EINVAL; | |
1602 | } | |
1603 | if (subvol_objectid && root_objectid != subvol_objectid) { | |
05dbe683 OS |
1604 | /* |
1605 | * This will also catch a race condition where a | |
1606 | * subvolume which was passed by ID is renamed and | |
1607 | * another subvolume is renamed over the old location. | |
1608 | */ | |
ab8d0fc4 JM |
1609 | btrfs_err(fs_info, |
1610 | "subvol '%s' does not match subvolid %llu", | |
1611 | subvol_name, subvol_objectid); | |
bb289b7b OS |
1612 | ret = -EINVAL; |
1613 | } | |
1614 | if (ret) { | |
1615 | dput(root); | |
1616 | root = ERR_PTR(ret); | |
1617 | deactivate_locked_super(s); | |
1618 | } | |
f9d9ef62 DS |
1619 | } |
1620 | ||
fa330659 OS |
1621 | out: |
1622 | mntput(mnt); | |
fa330659 | 1623 | kfree(subvol_name); |
830c4adb JB |
1624 | return root; |
1625 | } | |
450ba0ea | 1626 | |
312c89fb MT |
1627 | /* |
1628 | * Find a superblock for the given device / mount point. | |
1629 | * | |
1630 | * Note: This is based on mount_bdev from fs/super.c with a few additions | |
1631 | * for multiple device setup. Make sure to keep it in sync. | |
1632 | */ | |
72fa39f5 MT |
1633 | static struct dentry *btrfs_mount_root(struct file_system_type *fs_type, |
1634 | int flags, const char *device_name, void *data) | |
1635 | { | |
1636 | struct block_device *bdev = NULL; | |
1637 | struct super_block *s; | |
36350e95 | 1638 | struct btrfs_device *device = NULL; |
72fa39f5 MT |
1639 | struct btrfs_fs_devices *fs_devices = NULL; |
1640 | struct btrfs_fs_info *fs_info = NULL; | |
204cc0cc | 1641 | void *new_sec_opts = NULL; |
72fa39f5 | 1642 | fmode_t mode = FMODE_READ; |
72fa39f5 MT |
1643 | int error = 0; |
1644 | ||
1645 | if (!(flags & SB_RDONLY)) | |
1646 | mode |= FMODE_WRITE; | |
1647 | ||
72fa39f5 | 1648 | if (data) { |
a65001e8 | 1649 | error = security_sb_eat_lsm_opts(data, &new_sec_opts); |
72fa39f5 MT |
1650 | if (error) |
1651 | return ERR_PTR(error); | |
1652 | } | |
1653 | ||
72fa39f5 MT |
1654 | /* |
1655 | * Setup a dummy root and fs_info for test/set super. This is because | |
1656 | * we don't actually fill this stuff out until open_ctree, but we need | |
8260edba JB |
1657 | * then open_ctree will properly initialize the file system specific |
1658 | * settings later. btrfs_init_fs_info initializes the static elements | |
1659 | * of the fs_info (locks and such) to make cleanup easier if we find a | |
1660 | * superblock with our given fs_devices later on at sget() time. | |
72fa39f5 | 1661 | */ |
a8fd1f71 | 1662 | fs_info = kvzalloc(sizeof(struct btrfs_fs_info), GFP_KERNEL); |
72fa39f5 MT |
1663 | if (!fs_info) { |
1664 | error = -ENOMEM; | |
1665 | goto error_sec_opts; | |
1666 | } | |
8260edba | 1667 | btrfs_init_fs_info(fs_info); |
72fa39f5 | 1668 | |
72fa39f5 MT |
1669 | fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); |
1670 | fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); | |
72fa39f5 MT |
1671 | if (!fs_info->super_copy || !fs_info->super_for_commit) { |
1672 | error = -ENOMEM; | |
1673 | goto error_fs_info; | |
1674 | } | |
1675 | ||
399f7f4c | 1676 | mutex_lock(&uuid_mutex); |
fa59f27c | 1677 | error = btrfs_parse_device_options(data, mode, fs_type); |
81ffd56b DS |
1678 | if (error) { |
1679 | mutex_unlock(&uuid_mutex); | |
399f7f4c | 1680 | goto error_fs_info; |
81ffd56b | 1681 | } |
399f7f4c | 1682 | |
36350e95 GJ |
1683 | device = btrfs_scan_one_device(device_name, mode, fs_type); |
1684 | if (IS_ERR(device)) { | |
81ffd56b | 1685 | mutex_unlock(&uuid_mutex); |
36350e95 | 1686 | error = PTR_ERR(device); |
399f7f4c | 1687 | goto error_fs_info; |
81ffd56b | 1688 | } |
399f7f4c | 1689 | |
36350e95 | 1690 | fs_devices = device->fs_devices; |
399f7f4c DS |
1691 | fs_info->fs_devices = fs_devices; |
1692 | ||
72fa39f5 | 1693 | error = btrfs_open_devices(fs_devices, mode, fs_type); |
f5194e34 | 1694 | mutex_unlock(&uuid_mutex); |
72fa39f5 MT |
1695 | if (error) |
1696 | goto error_fs_info; | |
1697 | ||
1698 | if (!(flags & SB_RDONLY) && fs_devices->rw_devices == 0) { | |
1699 | error = -EACCES; | |
1700 | goto error_close_devices; | |
1701 | } | |
1702 | ||
1703 | bdev = fs_devices->latest_bdev; | |
1704 | s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | SB_NOSEC, | |
1705 | fs_info); | |
1706 | if (IS_ERR(s)) { | |
1707 | error = PTR_ERR(s); | |
1708 | goto error_close_devices; | |
1709 | } | |
1710 | ||
1711 | if (s->s_root) { | |
1712 | btrfs_close_devices(fs_devices); | |
0d4b0463 | 1713 | btrfs_free_fs_info(fs_info); |
72fa39f5 MT |
1714 | if ((flags ^ s->s_flags) & SB_RDONLY) |
1715 | error = -EBUSY; | |
1716 | } else { | |
1717 | snprintf(s->s_id, sizeof(s->s_id), "%pg", bdev); | |
1718 | btrfs_sb(s)->bdev_holder = fs_type; | |
9b4e675a DS |
1719 | if (!strstr(crc32c_impl(), "generic")) |
1720 | set_bit(BTRFS_FS_CSUM_IMPL_FAST, &fs_info->flags); | |
72fa39f5 MT |
1721 | error = btrfs_fill_super(s, fs_devices, data); |
1722 | } | |
a65001e8 | 1723 | if (!error) |
204cc0cc | 1724 | error = security_sb_set_mnt_opts(s, new_sec_opts, 0, NULL); |
a65001e8 | 1725 | security_free_mnt_opts(&new_sec_opts); |
72fa39f5 MT |
1726 | if (error) { |
1727 | deactivate_locked_super(s); | |
a65001e8 | 1728 | return ERR_PTR(error); |
72fa39f5 MT |
1729 | } |
1730 | ||
1731 | return dget(s->s_root); | |
1732 | ||
1733 | error_close_devices: | |
1734 | btrfs_close_devices(fs_devices); | |
1735 | error_fs_info: | |
0d4b0463 | 1736 | btrfs_free_fs_info(fs_info); |
72fa39f5 MT |
1737 | error_sec_opts: |
1738 | security_free_mnt_opts(&new_sec_opts); | |
1739 | return ERR_PTR(error); | |
1740 | } | |
312c89fb | 1741 | |
edf24abe | 1742 | /* |
312c89fb | 1743 | * Mount function which is called by VFS layer. |
edf24abe | 1744 | * |
312c89fb MT |
1745 | * In order to allow mounting a subvolume directly, btrfs uses mount_subtree() |
1746 | * which needs vfsmount* of device's root (/). This means device's root has to | |
1747 | * be mounted internally in any case. | |
1748 | * | |
1749 | * Operation flow: | |
1750 | * 1. Parse subvol id related options for later use in mount_subvol(). | |
1751 | * | |
1752 | * 2. Mount device's root (/) by calling vfs_kern_mount(). | |
1753 | * | |
1754 | * NOTE: vfs_kern_mount() is used by VFS to call btrfs_mount() in the | |
1755 | * first place. In order to avoid calling btrfs_mount() again, we use | |
1756 | * different file_system_type which is not registered to VFS by | |
1757 | * register_filesystem() (btrfs_root_fs_type). As a result, | |
1758 | * btrfs_mount_root() is called. The return value will be used by | |
1759 | * mount_subtree() in mount_subvol(). | |
1760 | * | |
1761 | * 3. Call mount_subvol() to get the dentry of subvolume. Since there is | |
1762 | * "btrfs subvolume set-default", mount_subvol() is called always. | |
edf24abe | 1763 | */ |
061dbc6b | 1764 | static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, |
306e16ce | 1765 | const char *device_name, void *data) |
4b82d6e4 | 1766 | { |
312c89fb MT |
1767 | struct vfsmount *mnt_root; |
1768 | struct dentry *root; | |
73f73415 JB |
1769 | char *subvol_name = NULL; |
1770 | u64 subvol_objectid = 0; | |
4b82d6e4 Y |
1771 | int error = 0; |
1772 | ||
93b9bcdf GJ |
1773 | error = btrfs_parse_subvol_options(data, &subvol_name, |
1774 | &subvol_objectid); | |
f23c8af8 ID |
1775 | if (error) { |
1776 | kfree(subvol_name); | |
061dbc6b | 1777 | return ERR_PTR(error); |
f23c8af8 | 1778 | } |
edf24abe | 1779 | |
312c89fb MT |
1780 | /* mount device's root (/) */ |
1781 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, flags, device_name, data); | |
1782 | if (PTR_ERR_OR_ZERO(mnt_root) == -EBUSY) { | |
1783 | if (flags & SB_RDONLY) { | |
1784 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, | |
1785 | flags & ~SB_RDONLY, device_name, data); | |
1786 | } else { | |
1787 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, | |
1788 | flags | SB_RDONLY, device_name, data); | |
1789 | if (IS_ERR(mnt_root)) { | |
1790 | root = ERR_CAST(mnt_root); | |
532b618b | 1791 | kfree(subvol_name); |
312c89fb MT |
1792 | goto out; |
1793 | } | |
4b82d6e4 | 1794 | |
312c89fb MT |
1795 | down_write(&mnt_root->mnt_sb->s_umount); |
1796 | error = btrfs_remount(mnt_root->mnt_sb, &flags, NULL); | |
1797 | up_write(&mnt_root->mnt_sb->s_umount); | |
1798 | if (error < 0) { | |
1799 | root = ERR_PTR(error); | |
1800 | mntput(mnt_root); | |
532b618b | 1801 | kfree(subvol_name); |
312c89fb MT |
1802 | goto out; |
1803 | } | |
1804 | } | |
f667aef6 | 1805 | } |
312c89fb MT |
1806 | if (IS_ERR(mnt_root)) { |
1807 | root = ERR_CAST(mnt_root); | |
532b618b | 1808 | kfree(subvol_name); |
312c89fb | 1809 | goto out; |
f667aef6 | 1810 | } |
4b82d6e4 | 1811 | |
312c89fb | 1812 | /* mount_subvol() will free subvol_name and mnt_root */ |
ae0bc863 | 1813 | root = mount_subvol(subvol_name, subvol_objectid, mnt_root); |
4b82d6e4 | 1814 | |
312c89fb MT |
1815 | out: |
1816 | return root; | |
4b82d6e4 | 1817 | } |
2e635a27 | 1818 | |
0d2450ab | 1819 | static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info, |
f7b885be | 1820 | u32 new_pool_size, u32 old_pool_size) |
0d2450ab ST |
1821 | { |
1822 | if (new_pool_size == old_pool_size) | |
1823 | return; | |
1824 | ||
1825 | fs_info->thread_pool_size = new_pool_size; | |
1826 | ||
efe120a0 | 1827 | btrfs_info(fs_info, "resize thread pool %d -> %d", |
0d2450ab ST |
1828 | old_pool_size, new_pool_size); |
1829 | ||
5cdc7ad3 | 1830 | btrfs_workqueue_set_max(fs_info->workers, new_pool_size); |
afe3d242 | 1831 | btrfs_workqueue_set_max(fs_info->delalloc_workers, new_pool_size); |
e66f0bb1 | 1832 | btrfs_workqueue_set_max(fs_info->caching_workers, new_pool_size); |
fccb5d86 QW |
1833 | btrfs_workqueue_set_max(fs_info->endio_workers, new_pool_size); |
1834 | btrfs_workqueue_set_max(fs_info->endio_meta_workers, new_pool_size); | |
1835 | btrfs_workqueue_set_max(fs_info->endio_meta_write_workers, | |
1836 | new_pool_size); | |
1837 | btrfs_workqueue_set_max(fs_info->endio_write_workers, new_pool_size); | |
1838 | btrfs_workqueue_set_max(fs_info->endio_freespace_worker, new_pool_size); | |
5b3bc44e | 1839 | btrfs_workqueue_set_max(fs_info->delayed_workers, new_pool_size); |
736cfa15 | 1840 | btrfs_workqueue_set_max(fs_info->readahead_workers, new_pool_size); |
0339ef2f QW |
1841 | btrfs_workqueue_set_max(fs_info->scrub_wr_completion_workers, |
1842 | new_pool_size); | |
0d2450ab ST |
1843 | } |
1844 | ||
f42a34b2 MX |
1845 | static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info, |
1846 | unsigned long old_opts, int flags) | |
1847 | { | |
dc81cdc5 MX |
1848 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && |
1849 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || | |
1751e8a6 | 1850 | (flags & SB_RDONLY))) { |
dc81cdc5 MX |
1851 | /* wait for any defraggers to finish */ |
1852 | wait_event(fs_info->transaction_wait, | |
1853 | (atomic_read(&fs_info->defrag_running) == 0)); | |
1751e8a6 | 1854 | if (flags & SB_RDONLY) |
dc81cdc5 MX |
1855 | sync_filesystem(fs_info->sb); |
1856 | } | |
1857 | } | |
1858 | ||
1859 | static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info, | |
1860 | unsigned long old_opts) | |
1861 | { | |
1862 | /* | |
180e4d47 LB |
1863 | * We need to cleanup all defragable inodes if the autodefragment is |
1864 | * close or the filesystem is read only. | |
dc81cdc5 MX |
1865 | */ |
1866 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && | |
bc98a42c | 1867 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || sb_rdonly(fs_info->sb))) { |
dc81cdc5 MX |
1868 | btrfs_cleanup_defrag_inodes(fs_info); |
1869 | } | |
1870 | ||
b0643e59 DZ |
1871 | /* If we toggled discard async */ |
1872 | if (!btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) && | |
1873 | btrfs_test_opt(fs_info, DISCARD_ASYNC)) | |
1874 | btrfs_discard_resume(fs_info); | |
1875 | else if (btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) && | |
1876 | !btrfs_test_opt(fs_info, DISCARD_ASYNC)) | |
1877 | btrfs_discard_cleanup(fs_info); | |
dc81cdc5 MX |
1878 | } |
1879 | ||
c146afad YZ |
1880 | static int btrfs_remount(struct super_block *sb, int *flags, char *data) |
1881 | { | |
815745cf AV |
1882 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
1883 | struct btrfs_root *root = fs_info->tree_root; | |
49b25e05 JM |
1884 | unsigned old_flags = sb->s_flags; |
1885 | unsigned long old_opts = fs_info->mount_opt; | |
1886 | unsigned long old_compress_type = fs_info->compress_type; | |
1887 | u64 old_max_inline = fs_info->max_inline; | |
f7b885be | 1888 | u32 old_thread_pool_size = fs_info->thread_pool_size; |
d612ac59 | 1889 | u32 old_metadata_ratio = fs_info->metadata_ratio; |
c146afad YZ |
1890 | int ret; |
1891 | ||
02b9984d | 1892 | sync_filesystem(sb); |
88c4703f | 1893 | set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); |
dc81cdc5 | 1894 | |
f667aef6 | 1895 | if (data) { |
204cc0cc | 1896 | void *new_sec_opts = NULL; |
f667aef6 | 1897 | |
a65001e8 AV |
1898 | ret = security_sb_eat_lsm_opts(data, &new_sec_opts); |
1899 | if (!ret) | |
204cc0cc | 1900 | ret = security_sb_remount(sb, new_sec_opts); |
a65001e8 | 1901 | security_free_mnt_opts(&new_sec_opts); |
f667aef6 QW |
1902 | if (ret) |
1903 | goto restore; | |
f667aef6 QW |
1904 | } |
1905 | ||
2ff7e61e | 1906 | ret = btrfs_parse_options(fs_info, data, *flags); |
891f41cb | 1907 | if (ret) |
49b25e05 | 1908 | goto restore; |
b288052e | 1909 | |
f42a34b2 | 1910 | btrfs_remount_begin(fs_info, old_opts, *flags); |
0d2450ab ST |
1911 | btrfs_resize_thread_pool(fs_info, |
1912 | fs_info->thread_pool_size, old_thread_pool_size); | |
1913 | ||
1751e8a6 | 1914 | if ((bool)(*flags & SB_RDONLY) == sb_rdonly(sb)) |
dc81cdc5 | 1915 | goto out; |
c146afad | 1916 | |
1751e8a6 | 1917 | if (*flags & SB_RDONLY) { |
8dabb742 SB |
1918 | /* |
1919 | * this also happens on 'umount -rf' or on shutdown, when | |
1920 | * the filesystem is busy. | |
1921 | */ | |
21c7e756 | 1922 | cancel_work_sync(&fs_info->async_reclaim_work); |
57056740 | 1923 | cancel_work_sync(&fs_info->async_data_reclaim_work); |
361c093d | 1924 | |
b0643e59 DZ |
1925 | btrfs_discard_cleanup(fs_info); |
1926 | ||
361c093d SB |
1927 | /* wait for the uuid_scan task to finish */ |
1928 | down(&fs_info->uuid_tree_rescan_sem); | |
1929 | /* avoid complains from lockdep et al. */ | |
1930 | up(&fs_info->uuid_tree_rescan_sem); | |
1931 | ||
1751e8a6 | 1932 | sb->s_flags |= SB_RDONLY; |
c146afad | 1933 | |
e44163e1 | 1934 | /* |
1751e8a6 | 1935 | * Setting SB_RDONLY will put the cleaner thread to |
e44163e1 JM |
1936 | * sleep at the next loop if it's already active. |
1937 | * If it's already asleep, we'll leave unused block | |
1938 | * groups on disk until we're mounted read-write again | |
1939 | * unless we clean them up here. | |
1940 | */ | |
e44163e1 | 1941 | btrfs_delete_unused_bgs(fs_info); |
e44163e1 | 1942 | |
8dabb742 SB |
1943 | btrfs_dev_replace_suspend_for_unmount(fs_info); |
1944 | btrfs_scrub_cancel(fs_info); | |
061594ef | 1945 | btrfs_pause_balance(fs_info); |
8dabb742 | 1946 | |
6bccf3ab | 1947 | ret = btrfs_commit_super(fs_info); |
49b25e05 JM |
1948 | if (ret) |
1949 | goto restore; | |
c146afad | 1950 | } else { |
0b246afa | 1951 | if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { |
6ef3de9c | 1952 | btrfs_err(fs_info, |
efe120a0 | 1953 | "Remounting read-write after error is not allowed"); |
6ef3de9c DS |
1954 | ret = -EINVAL; |
1955 | goto restore; | |
1956 | } | |
8a3db184 | 1957 | if (fs_info->fs_devices->rw_devices == 0) { |
49b25e05 JM |
1958 | ret = -EACCES; |
1959 | goto restore; | |
8a3db184 | 1960 | } |
2b82032c | 1961 | |
6528b99d | 1962 | if (!btrfs_check_rw_degradable(fs_info, NULL)) { |
efe120a0 | 1963 | btrfs_warn(fs_info, |
52042d8e | 1964 | "too many missing devices, writable remount is not allowed"); |
292fd7fc SB |
1965 | ret = -EACCES; |
1966 | goto restore; | |
1967 | } | |
1968 | ||
8a3db184 | 1969 | if (btrfs_super_log_root(fs_info->super_copy) != 0) { |
10a3a3ed DS |
1970 | btrfs_warn(fs_info, |
1971 | "mount required to replay tree-log, cannot remount read-write"); | |
49b25e05 JM |
1972 | ret = -EINVAL; |
1973 | goto restore; | |
8a3db184 | 1974 | } |
c146afad | 1975 | |
815745cf | 1976 | ret = btrfs_cleanup_fs_roots(fs_info); |
49b25e05 JM |
1977 | if (ret) |
1978 | goto restore; | |
c146afad | 1979 | |
d68fc57b | 1980 | /* recover relocation */ |
5f316481 | 1981 | mutex_lock(&fs_info->cleaner_mutex); |
d68fc57b | 1982 | ret = btrfs_recover_relocation(root); |
5f316481 | 1983 | mutex_unlock(&fs_info->cleaner_mutex); |
49b25e05 JM |
1984 | if (ret) |
1985 | goto restore; | |
c146afad | 1986 | |
2b6ba629 ID |
1987 | ret = btrfs_resume_balance_async(fs_info); |
1988 | if (ret) | |
1989 | goto restore; | |
1990 | ||
8dabb742 SB |
1991 | ret = btrfs_resume_dev_replace_async(fs_info); |
1992 | if (ret) { | |
efe120a0 | 1993 | btrfs_warn(fs_info, "failed to resume dev_replace"); |
8dabb742 SB |
1994 | goto restore; |
1995 | } | |
94aebfb2 | 1996 | |
6c6b5a39 AS |
1997 | btrfs_qgroup_rescan_resume(fs_info); |
1998 | ||
94aebfb2 | 1999 | if (!fs_info->uuid_root) { |
efe120a0 | 2000 | btrfs_info(fs_info, "creating UUID tree"); |
94aebfb2 JB |
2001 | ret = btrfs_create_uuid_tree(fs_info); |
2002 | if (ret) { | |
5d163e0e JM |
2003 | btrfs_warn(fs_info, |
2004 | "failed to create the UUID tree %d", | |
2005 | ret); | |
94aebfb2 JB |
2006 | goto restore; |
2007 | } | |
2008 | } | |
1751e8a6 | 2009 | sb->s_flags &= ~SB_RDONLY; |
90c711ab | 2010 | |
afcdd129 | 2011 | set_bit(BTRFS_FS_OPEN, &fs_info->flags); |
c146afad | 2012 | } |
dc81cdc5 | 2013 | out: |
faa00889 JB |
2014 | /* |
2015 | * We need to set SB_I_VERSION here otherwise it'll get cleared by VFS, | |
2016 | * since the absence of the flag means it can be toggled off by remount. | |
2017 | */ | |
2018 | *flags |= SB_I_VERSION; | |
2019 | ||
2c6a92b0 | 2020 | wake_up_process(fs_info->transaction_kthread); |
dc81cdc5 | 2021 | btrfs_remount_cleanup(fs_info, old_opts); |
88c4703f JT |
2022 | clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); |
2023 | ||
c146afad | 2024 | return 0; |
49b25e05 JM |
2025 | |
2026 | restore: | |
1751e8a6 | 2027 | /* We've hit an error - don't reset SB_RDONLY */ |
bc98a42c | 2028 | if (sb_rdonly(sb)) |
1751e8a6 | 2029 | old_flags |= SB_RDONLY; |
49b25e05 JM |
2030 | sb->s_flags = old_flags; |
2031 | fs_info->mount_opt = old_opts; | |
2032 | fs_info->compress_type = old_compress_type; | |
2033 | fs_info->max_inline = old_max_inline; | |
0d2450ab ST |
2034 | btrfs_resize_thread_pool(fs_info, |
2035 | old_thread_pool_size, fs_info->thread_pool_size); | |
49b25e05 | 2036 | fs_info->metadata_ratio = old_metadata_ratio; |
dc81cdc5 | 2037 | btrfs_remount_cleanup(fs_info, old_opts); |
88c4703f JT |
2038 | clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); |
2039 | ||
49b25e05 | 2040 | return ret; |
c146afad YZ |
2041 | } |
2042 | ||
bcd53741 | 2043 | /* Used to sort the devices by max_avail(descending sort) */ |
7e17916b | 2044 | static inline int btrfs_cmp_device_free_bytes(const void *dev_info1, |
bcd53741 AJ |
2045 | const void *dev_info2) |
2046 | { | |
2047 | if (((struct btrfs_device_info *)dev_info1)->max_avail > | |
2048 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
2049 | return -1; | |
2050 | else if (((struct btrfs_device_info *)dev_info1)->max_avail < | |
2051 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
2052 | return 1; | |
2053 | else | |
2054 | return 0; | |
2055 | } | |
2056 | ||
2057 | /* | |
2058 | * sort the devices by max_avail, in which max free extent size of each device | |
2059 | * is stored.(Descending Sort) | |
2060 | */ | |
2061 | static inline void btrfs_descending_sort_devices( | |
2062 | struct btrfs_device_info *devices, | |
2063 | size_t nr_devices) | |
2064 | { | |
2065 | sort(devices, nr_devices, sizeof(struct btrfs_device_info), | |
2066 | btrfs_cmp_device_free_bytes, NULL); | |
2067 | } | |
2068 | ||
6d07bcec MX |
2069 | /* |
2070 | * The helper to calc the free space on the devices that can be used to store | |
2071 | * file data. | |
2072 | */ | |
7e17916b AB |
2073 | static inline int btrfs_calc_avail_data_space(struct btrfs_fs_info *fs_info, |
2074 | u64 *free_bytes) | |
6d07bcec | 2075 | { |
6d07bcec MX |
2076 | struct btrfs_device_info *devices_info; |
2077 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; | |
2078 | struct btrfs_device *device; | |
6d07bcec MX |
2079 | u64 type; |
2080 | u64 avail_space; | |
6d07bcec | 2081 | u64 min_stripe_size; |
559ca6ea | 2082 | int num_stripes = 1; |
6d07bcec | 2083 | int i = 0, nr_devices; |
4f080f57 | 2084 | const struct btrfs_raid_attr *rattr; |
6d07bcec | 2085 | |
7e33fd99 | 2086 | /* |
01327610 | 2087 | * We aren't under the device list lock, so this is racy-ish, but good |
7e33fd99 JB |
2088 | * enough for our purposes. |
2089 | */ | |
b772a86e | 2090 | nr_devices = fs_info->fs_devices->open_devices; |
7e33fd99 JB |
2091 | if (!nr_devices) { |
2092 | smp_mb(); | |
2093 | nr_devices = fs_info->fs_devices->open_devices; | |
2094 | ASSERT(nr_devices); | |
2095 | if (!nr_devices) { | |
2096 | *free_bytes = 0; | |
2097 | return 0; | |
2098 | } | |
2099 | } | |
6d07bcec | 2100 | |
d9b0d9ba | 2101 | devices_info = kmalloc_array(nr_devices, sizeof(*devices_info), |
6a44517d | 2102 | GFP_KERNEL); |
6d07bcec MX |
2103 | if (!devices_info) |
2104 | return -ENOMEM; | |
2105 | ||
01327610 | 2106 | /* calc min stripe number for data space allocation */ |
1b86826d | 2107 | type = btrfs_data_alloc_profile(fs_info); |
4f080f57 DS |
2108 | rattr = &btrfs_raid_array[btrfs_bg_flags_to_raid_index(type)]; |
2109 | ||
e1ea2bee | 2110 | if (type & BTRFS_BLOCK_GROUP_RAID0) |
39fb26c3 | 2111 | num_stripes = nr_devices; |
e1ea2bee | 2112 | else if (type & BTRFS_BLOCK_GROUP_RAID1) |
39fb26c3 | 2113 | num_stripes = 2; |
47e6f742 DS |
2114 | else if (type & BTRFS_BLOCK_GROUP_RAID1C3) |
2115 | num_stripes = 3; | |
8d6fac00 DS |
2116 | else if (type & BTRFS_BLOCK_GROUP_RAID1C4) |
2117 | num_stripes = 4; | |
e1ea2bee | 2118 | else if (type & BTRFS_BLOCK_GROUP_RAID10) |
39fb26c3 | 2119 | num_stripes = 4; |
6d07bcec | 2120 | |
4f080f57 DS |
2121 | /* Adjust for more than 1 stripe per device */ |
2122 | min_stripe_size = rattr->dev_stripes * BTRFS_STRIPE_LEN; | |
6d07bcec | 2123 | |
7e33fd99 JB |
2124 | rcu_read_lock(); |
2125 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { | |
e12c9621 AJ |
2126 | if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, |
2127 | &device->dev_state) || | |
401e29c1 AJ |
2128 | !device->bdev || |
2129 | test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) | |
6d07bcec MX |
2130 | continue; |
2131 | ||
7e33fd99 JB |
2132 | if (i >= nr_devices) |
2133 | break; | |
2134 | ||
6d07bcec MX |
2135 | avail_space = device->total_bytes - device->bytes_used; |
2136 | ||
2137 | /* align with stripe_len */ | |
559ca6ea | 2138 | avail_space = rounddown(avail_space, BTRFS_STRIPE_LEN); |
6d07bcec MX |
2139 | |
2140 | /* | |
01327610 | 2141 | * In order to avoid overwriting the superblock on the drive, |
6d07bcec MX |
2142 | * btrfs starts at an offset of at least 1MB when doing chunk |
2143 | * allocation. | |
559ca6ea NB |
2144 | * |
2145 | * This ensures we have at least min_stripe_size free space | |
2146 | * after excluding 1MB. | |
6d07bcec | 2147 | */ |
559ca6ea | 2148 | if (avail_space <= SZ_1M + min_stripe_size) |
6d07bcec MX |
2149 | continue; |
2150 | ||
559ca6ea NB |
2151 | avail_space -= SZ_1M; |
2152 | ||
6d07bcec MX |
2153 | devices_info[i].dev = device; |
2154 | devices_info[i].max_avail = avail_space; | |
2155 | ||
2156 | i++; | |
2157 | } | |
7e33fd99 | 2158 | rcu_read_unlock(); |
6d07bcec MX |
2159 | |
2160 | nr_devices = i; | |
2161 | ||
2162 | btrfs_descending_sort_devices(devices_info, nr_devices); | |
2163 | ||
2164 | i = nr_devices - 1; | |
2165 | avail_space = 0; | |
559ca6ea NB |
2166 | while (nr_devices >= rattr->devs_min) { |
2167 | num_stripes = min(num_stripes, nr_devices); | |
39fb26c3 | 2168 | |
6d07bcec MX |
2169 | if (devices_info[i].max_avail >= min_stripe_size) { |
2170 | int j; | |
2171 | u64 alloc_size; | |
2172 | ||
39fb26c3 | 2173 | avail_space += devices_info[i].max_avail * num_stripes; |
6d07bcec | 2174 | alloc_size = devices_info[i].max_avail; |
39fb26c3 | 2175 | for (j = i + 1 - num_stripes; j <= i; j++) |
6d07bcec MX |
2176 | devices_info[j].max_avail -= alloc_size; |
2177 | } | |
2178 | i--; | |
2179 | nr_devices--; | |
2180 | } | |
2181 | ||
2182 | kfree(devices_info); | |
2183 | *free_bytes = avail_space; | |
2184 | return 0; | |
2185 | } | |
2186 | ||
ba7b6e62 DS |
2187 | /* |
2188 | * Calculate numbers for 'df', pessimistic in case of mixed raid profiles. | |
2189 | * | |
2190 | * If there's a redundant raid level at DATA block groups, use the respective | |
2191 | * multiplier to scale the sizes. | |
2192 | * | |
2193 | * Unused device space usage is based on simulating the chunk allocator | |
0d0c71b3 DS |
2194 | * algorithm that respects the device sizes and order of allocations. This is |
2195 | * a close approximation of the actual use but there are other factors that may | |
2196 | * change the result (like a new metadata chunk). | |
ba7b6e62 | 2197 | * |
ca8a51b3 | 2198 | * If metadata is exhausted, f_bavail will be 0. |
ba7b6e62 | 2199 | */ |
8fd17795 CM |
2200 | static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) |
2201 | { | |
815745cf AV |
2202 | struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb); |
2203 | struct btrfs_super_block *disk_super = fs_info->super_copy; | |
bd4d1088 JB |
2204 | struct btrfs_space_info *found; |
2205 | u64 total_used = 0; | |
6d07bcec | 2206 | u64 total_free_data = 0; |
ca8a51b3 | 2207 | u64 total_free_meta = 0; |
db94535d | 2208 | int bits = dentry->d_sb->s_blocksize_bits; |
de37aa51 | 2209 | __be32 *fsid = (__be32 *)fs_info->fs_devices->fsid; |
ba7b6e62 DS |
2210 | unsigned factor = 1; |
2211 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; | |
6d07bcec | 2212 | int ret; |
ca8a51b3 | 2213 | u64 thresh = 0; |
ae02d1bd | 2214 | int mixed = 0; |
8fd17795 | 2215 | |
72804905 | 2216 | list_for_each_entry(found, &fs_info->space_info, list) { |
6d07bcec | 2217 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) { |
ba7b6e62 DS |
2218 | int i; |
2219 | ||
6d07bcec MX |
2220 | total_free_data += found->disk_total - found->disk_used; |
2221 | total_free_data -= | |
2222 | btrfs_account_ro_block_groups_free_space(found); | |
ba7b6e62 DS |
2223 | |
2224 | for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) { | |
46df06b8 DS |
2225 | if (!list_empty(&found->block_groups[i])) |
2226 | factor = btrfs_bg_type_to_factor( | |
2227 | btrfs_raid_array[i].bg_flag); | |
ba7b6e62 | 2228 | } |
6d07bcec | 2229 | } |
ae02d1bd LB |
2230 | |
2231 | /* | |
2232 | * Metadata in mixed block goup profiles are accounted in data | |
2233 | */ | |
2234 | if (!mixed && found->flags & BTRFS_BLOCK_GROUP_METADATA) { | |
2235 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) | |
2236 | mixed = 1; | |
2237 | else | |
2238 | total_free_meta += found->disk_total - | |
2239 | found->disk_used; | |
2240 | } | |
6d07bcec | 2241 | |
b742bb82 | 2242 | total_used += found->disk_used; |
89a55897 | 2243 | } |
ba7b6e62 | 2244 | |
ba7b6e62 DS |
2245 | buf->f_blocks = div_u64(btrfs_super_total_bytes(disk_super), factor); |
2246 | buf->f_blocks >>= bits; | |
2247 | buf->f_bfree = buf->f_blocks - (div_u64(total_used, factor) >> bits); | |
2248 | ||
2249 | /* Account global block reserve as used, it's in logical size already */ | |
2250 | spin_lock(&block_rsv->lock); | |
41b34acc LB |
2251 | /* Mixed block groups accounting is not byte-accurate, avoid overflow */ |
2252 | if (buf->f_bfree >= block_rsv->size >> bits) | |
2253 | buf->f_bfree -= block_rsv->size >> bits; | |
2254 | else | |
2255 | buf->f_bfree = 0; | |
ba7b6e62 DS |
2256 | spin_unlock(&block_rsv->lock); |
2257 | ||
0d95c1be | 2258 | buf->f_bavail = div_u64(total_free_data, factor); |
6bccf3ab | 2259 | ret = btrfs_calc_avail_data_space(fs_info, &total_free_data); |
7e33fd99 | 2260 | if (ret) |
6d07bcec | 2261 | return ret; |
ba7b6e62 | 2262 | buf->f_bavail += div_u64(total_free_data, factor); |
6d07bcec | 2263 | buf->f_bavail = buf->f_bavail >> bits; |
d397712b | 2264 | |
ca8a51b3 DS |
2265 | /* |
2266 | * We calculate the remaining metadata space minus global reserve. If | |
2267 | * this is (supposedly) smaller than zero, there's no space. But this | |
2268 | * does not hold in practice, the exhausted state happens where's still | |
2269 | * some positive delta. So we apply some guesswork and compare the | |
2270 | * delta to a 4M threshold. (Practically observed delta was ~2M.) | |
2271 | * | |
2272 | * We probably cannot calculate the exact threshold value because this | |
2273 | * depends on the internal reservations requested by various | |
2274 | * operations, so some operations that consume a few metadata will | |
2275 | * succeed even if the Avail is zero. But this is better than the other | |
2276 | * way around. | |
2277 | */ | |
d4417e22 | 2278 | thresh = SZ_4M; |
ca8a51b3 | 2279 | |
d55966c4 JB |
2280 | /* |
2281 | * We only want to claim there's no available space if we can no longer | |
2282 | * allocate chunks for our metadata profile and our global reserve will | |
2283 | * not fit in the free metadata space. If we aren't ->full then we | |
2284 | * still can allocate chunks and thus are fine using the currently | |
2285 | * calculated f_bavail. | |
2286 | */ | |
2287 | if (!mixed && block_rsv->space_info->full && | |
2288 | total_free_meta - thresh < block_rsv->size) | |
ca8a51b3 DS |
2289 | buf->f_bavail = 0; |
2290 | ||
ba7b6e62 DS |
2291 | buf->f_type = BTRFS_SUPER_MAGIC; |
2292 | buf->f_bsize = dentry->d_sb->s_blocksize; | |
2293 | buf->f_namelen = BTRFS_NAME_LEN; | |
2294 | ||
9d03632e | 2295 | /* We treat it as constant endianness (it doesn't matter _which_) |
d397712b | 2296 | because we want the fsid to come out the same whether mounted |
9d03632e DW |
2297 | on a big-endian or little-endian host */ |
2298 | buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]); | |
2299 | buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]); | |
32d48fa1 | 2300 | /* Mask in the root object ID too, to disambiguate subvols */ |
4fd786e6 MT |
2301 | buf->f_fsid.val[0] ^= |
2302 | BTRFS_I(d_inode(dentry))->root->root_key.objectid >> 32; | |
2303 | buf->f_fsid.val[1] ^= | |
2304 | BTRFS_I(d_inode(dentry))->root->root_key.objectid; | |
32d48fa1 | 2305 | |
8fd17795 CM |
2306 | return 0; |
2307 | } | |
b5133862 | 2308 | |
aea52e19 AV |
2309 | static void btrfs_kill_super(struct super_block *sb) |
2310 | { | |
815745cf | 2311 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
aea52e19 | 2312 | kill_anon_super(sb); |
0d4b0463 | 2313 | btrfs_free_fs_info(fs_info); |
aea52e19 AV |
2314 | } |
2315 | ||
2e635a27 CM |
2316 | static struct file_system_type btrfs_fs_type = { |
2317 | .owner = THIS_MODULE, | |
2318 | .name = "btrfs", | |
061dbc6b | 2319 | .mount = btrfs_mount, |
aea52e19 | 2320 | .kill_sb = btrfs_kill_super, |
f667aef6 | 2321 | .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA, |
2e635a27 | 2322 | }; |
72fa39f5 MT |
2323 | |
2324 | static struct file_system_type btrfs_root_fs_type = { | |
2325 | .owner = THIS_MODULE, | |
2326 | .name = "btrfs", | |
2327 | .mount = btrfs_mount_root, | |
2328 | .kill_sb = btrfs_kill_super, | |
2329 | .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA, | |
2330 | }; | |
2331 | ||
7f78e035 | 2332 | MODULE_ALIAS_FS("btrfs"); |
a9218f6b | 2333 | |
d8620958 TVB |
2334 | static int btrfs_control_open(struct inode *inode, struct file *file) |
2335 | { | |
2336 | /* | |
2337 | * The control file's private_data is used to hold the | |
2338 | * transaction when it is started and is used to keep | |
2339 | * track of whether a transaction is already in progress. | |
2340 | */ | |
2341 | file->private_data = NULL; | |
2342 | return 0; | |
2343 | } | |
2344 | ||
d352ac68 | 2345 | /* |
cfe953c8 | 2346 | * Used by /dev/btrfs-control for devices ioctls. |
d352ac68 | 2347 | */ |
8a4b83cc CM |
2348 | static long btrfs_control_ioctl(struct file *file, unsigned int cmd, |
2349 | unsigned long arg) | |
2350 | { | |
2351 | struct btrfs_ioctl_vol_args *vol; | |
36350e95 | 2352 | struct btrfs_device *device = NULL; |
c071fcfd | 2353 | int ret = -ENOTTY; |
8a4b83cc | 2354 | |
e441d54d CM |
2355 | if (!capable(CAP_SYS_ADMIN)) |
2356 | return -EPERM; | |
2357 | ||
dae7b665 LZ |
2358 | vol = memdup_user((void __user *)arg, sizeof(*vol)); |
2359 | if (IS_ERR(vol)) | |
2360 | return PTR_ERR(vol); | |
f505754f | 2361 | vol->name[BTRFS_PATH_NAME_MAX] = '\0'; |
c071fcfd | 2362 | |
8a4b83cc CM |
2363 | switch (cmd) { |
2364 | case BTRFS_IOC_SCAN_DEV: | |
899f9307 | 2365 | mutex_lock(&uuid_mutex); |
36350e95 GJ |
2366 | device = btrfs_scan_one_device(vol->name, FMODE_READ, |
2367 | &btrfs_root_fs_type); | |
2368 | ret = PTR_ERR_OR_ZERO(device); | |
899f9307 | 2369 | mutex_unlock(&uuid_mutex); |
8a4b83cc | 2370 | break; |
228a73ab AJ |
2371 | case BTRFS_IOC_FORGET_DEV: |
2372 | ret = btrfs_forget_devices(vol->name); | |
2373 | break; | |
02db0844 | 2374 | case BTRFS_IOC_DEVICES_READY: |
899f9307 | 2375 | mutex_lock(&uuid_mutex); |
36350e95 GJ |
2376 | device = btrfs_scan_one_device(vol->name, FMODE_READ, |
2377 | &btrfs_root_fs_type); | |
2378 | if (IS_ERR(device)) { | |
899f9307 | 2379 | mutex_unlock(&uuid_mutex); |
36350e95 | 2380 | ret = PTR_ERR(device); |
02db0844 | 2381 | break; |
899f9307 | 2382 | } |
36350e95 GJ |
2383 | ret = !(device->fs_devices->num_devices == |
2384 | device->fs_devices->total_devices); | |
899f9307 | 2385 | mutex_unlock(&uuid_mutex); |
02db0844 | 2386 | break; |
c5868f83 | 2387 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 2388 | ret = btrfs_ioctl_get_supported_features((void __user*)arg); |
c5868f83 | 2389 | break; |
8a4b83cc | 2390 | } |
dae7b665 | 2391 | |
8a4b83cc | 2392 | kfree(vol); |
f819d837 | 2393 | return ret; |
8a4b83cc CM |
2394 | } |
2395 | ||
0176260f | 2396 | static int btrfs_freeze(struct super_block *sb) |
ed0dab6b | 2397 | { |
354aa0fb | 2398 | struct btrfs_trans_handle *trans; |
0b246afa JM |
2399 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
2400 | struct btrfs_root *root = fs_info->tree_root; | |
354aa0fb | 2401 | |
fac03c8d | 2402 | set_bit(BTRFS_FS_FROZEN, &fs_info->flags); |
9e7cc91a WX |
2403 | /* |
2404 | * We don't need a barrier here, we'll wait for any transaction that | |
2405 | * could be in progress on other threads (and do delayed iputs that | |
2406 | * we want to avoid on a frozen filesystem), or do the commit | |
2407 | * ourselves. | |
2408 | */ | |
d4edf39b | 2409 | trans = btrfs_attach_transaction_barrier(root); |
354aa0fb MX |
2410 | if (IS_ERR(trans)) { |
2411 | /* no transaction, don't bother */ | |
2412 | if (PTR_ERR(trans) == -ENOENT) | |
2413 | return 0; | |
2414 | return PTR_ERR(trans); | |
2415 | } | |
3a45bb20 | 2416 | return btrfs_commit_transaction(trans); |
ed0dab6b Y |
2417 | } |
2418 | ||
9e7cc91a WX |
2419 | static int btrfs_unfreeze(struct super_block *sb) |
2420 | { | |
fac03c8d DS |
2421 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
2422 | ||
2423 | clear_bit(BTRFS_FS_FROZEN, &fs_info->flags); | |
9e7cc91a WX |
2424 | return 0; |
2425 | } | |
2426 | ||
9c5085c1 JB |
2427 | static int btrfs_show_devname(struct seq_file *m, struct dentry *root) |
2428 | { | |
2429 | struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb); | |
9c5085c1 | 2430 | struct btrfs_device *dev, *first_dev = NULL; |
9c5085c1 | 2431 | |
88c14590 DS |
2432 | /* |
2433 | * Lightweight locking of the devices. We should not need | |
2434 | * device_list_mutex here as we only read the device data and the list | |
2435 | * is protected by RCU. Even if a device is deleted during the list | |
2436 | * traversals, we'll get valid data, the freeing callback will wait at | |
52042d8e | 2437 | * least until the rcu_read_unlock. |
88c14590 DS |
2438 | */ |
2439 | rcu_read_lock(); | |
4faf55b0 AJ |
2440 | list_for_each_entry_rcu(dev, &fs_info->fs_devices->devices, dev_list) { |
2441 | if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state)) | |
2442 | continue; | |
2443 | if (!dev->name) | |
2444 | continue; | |
2445 | if (!first_dev || dev->devid < first_dev->devid) | |
2446 | first_dev = dev; | |
9c5085c1 JB |
2447 | } |
2448 | ||
672d5990 MT |
2449 | if (first_dev) |
2450 | seq_escape(m, rcu_str_deref(first_dev->name), " \t\n\\"); | |
2451 | else | |
9c5085c1 | 2452 | WARN_ON(1); |
88c14590 | 2453 | rcu_read_unlock(); |
9c5085c1 JB |
2454 | return 0; |
2455 | } | |
2456 | ||
b87221de | 2457 | static const struct super_operations btrfs_super_ops = { |
76dda93c | 2458 | .drop_inode = btrfs_drop_inode, |
bd555975 | 2459 | .evict_inode = btrfs_evict_inode, |
e20d96d6 | 2460 | .put_super = btrfs_put_super, |
d5719762 | 2461 | .sync_fs = btrfs_sync_fs, |
a9572a15 | 2462 | .show_options = btrfs_show_options, |
9c5085c1 | 2463 | .show_devname = btrfs_show_devname, |
2c90e5d6 CM |
2464 | .alloc_inode = btrfs_alloc_inode, |
2465 | .destroy_inode = btrfs_destroy_inode, | |
26602cab | 2466 | .free_inode = btrfs_free_inode, |
8fd17795 | 2467 | .statfs = btrfs_statfs, |
c146afad | 2468 | .remount_fs = btrfs_remount, |
0176260f | 2469 | .freeze_fs = btrfs_freeze, |
9e7cc91a | 2470 | .unfreeze_fs = btrfs_unfreeze, |
e20d96d6 | 2471 | }; |
a9218f6b CM |
2472 | |
2473 | static const struct file_operations btrfs_ctl_fops = { | |
d8620958 | 2474 | .open = btrfs_control_open, |
a9218f6b | 2475 | .unlocked_ioctl = btrfs_control_ioctl, |
1832f2d8 | 2476 | .compat_ioctl = compat_ptr_ioctl, |
a9218f6b | 2477 | .owner = THIS_MODULE, |
6038f373 | 2478 | .llseek = noop_llseek, |
a9218f6b CM |
2479 | }; |
2480 | ||
2481 | static struct miscdevice btrfs_misc = { | |
578454ff | 2482 | .minor = BTRFS_MINOR, |
a9218f6b CM |
2483 | .name = "btrfs-control", |
2484 | .fops = &btrfs_ctl_fops | |
2485 | }; | |
2486 | ||
578454ff KS |
2487 | MODULE_ALIAS_MISCDEV(BTRFS_MINOR); |
2488 | MODULE_ALIAS("devname:btrfs-control"); | |
2489 | ||
f5c29bd9 | 2490 | static int __init btrfs_interface_init(void) |
a9218f6b CM |
2491 | { |
2492 | return misc_register(&btrfs_misc); | |
2493 | } | |
2494 | ||
e67c718b | 2495 | static __cold void btrfs_interface_exit(void) |
a9218f6b | 2496 | { |
f368ed60 | 2497 | misc_deregister(&btrfs_misc); |
a9218f6b CM |
2498 | } |
2499 | ||
f5c29bd9 | 2500 | static void __init btrfs_print_mod_info(void) |
85965600 | 2501 | { |
edf57cbf | 2502 | static const char options[] = "" |
85965600 DS |
2503 | #ifdef CONFIG_BTRFS_DEBUG |
2504 | ", debug=on" | |
2505 | #endif | |
79556c3d SB |
2506 | #ifdef CONFIG_BTRFS_ASSERT |
2507 | ", assert=on" | |
2508 | #endif | |
85965600 DS |
2509 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
2510 | ", integrity-checker=on" | |
fb592373 JB |
2511 | #endif |
2512 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
2513 | ", ref-verify=on" | |
85965600 | 2514 | #endif |
edf57cbf BVA |
2515 | ; |
2516 | pr_info("Btrfs loaded, crc32c=%s%s\n", crc32c_impl(), options); | |
85965600 DS |
2517 | } |
2518 | ||
2e635a27 CM |
2519 | static int __init init_btrfs_fs(void) |
2520 | { | |
2c90e5d6 | 2521 | int err; |
58176a96 | 2522 | |
63541927 FDBM |
2523 | btrfs_props_init(); |
2524 | ||
58176a96 JB |
2525 | err = btrfs_init_sysfs(); |
2526 | if (err) | |
9678c543 | 2527 | return err; |
58176a96 | 2528 | |
143bede5 | 2529 | btrfs_init_compress(); |
d1310b2e | 2530 | |
261507a0 LZ |
2531 | err = btrfs_init_cachep(); |
2532 | if (err) | |
2533 | goto free_compress; | |
2534 | ||
d1310b2e | 2535 | err = extent_io_init(); |
2f4cbe64 WB |
2536 | if (err) |
2537 | goto free_cachep; | |
2538 | ||
6f0d04f8 | 2539 | err = extent_state_cache_init(); |
d1310b2e CM |
2540 | if (err) |
2541 | goto free_extent_io; | |
2542 | ||
6f0d04f8 JB |
2543 | err = extent_map_init(); |
2544 | if (err) | |
2545 | goto free_extent_state_cache; | |
2546 | ||
6352b91d | 2547 | err = ordered_data_init(); |
2f4cbe64 WB |
2548 | if (err) |
2549 | goto free_extent_map; | |
c8b97818 | 2550 | |
6352b91d MX |
2551 | err = btrfs_delayed_inode_init(); |
2552 | if (err) | |
2553 | goto free_ordered_data; | |
2554 | ||
9247f317 | 2555 | err = btrfs_auto_defrag_init(); |
16cdcec7 MX |
2556 | if (err) |
2557 | goto free_delayed_inode; | |
2558 | ||
78a6184a | 2559 | err = btrfs_delayed_ref_init(); |
9247f317 MX |
2560 | if (err) |
2561 | goto free_auto_defrag; | |
2562 | ||
b9e9a6cb WS |
2563 | err = btrfs_prelim_ref_init(); |
2564 | if (err) | |
af13b492 | 2565 | goto free_delayed_ref; |
b9e9a6cb | 2566 | |
97eb6b69 | 2567 | err = btrfs_end_io_wq_init(); |
78a6184a | 2568 | if (err) |
af13b492 | 2569 | goto free_prelim_ref; |
78a6184a | 2570 | |
97eb6b69 DS |
2571 | err = btrfs_interface_init(); |
2572 | if (err) | |
2573 | goto free_end_io_wq; | |
2574 | ||
8ae1af3c | 2575 | btrfs_print_mod_info(); |
dc11dd5d JB |
2576 | |
2577 | err = btrfs_run_sanity_tests(); | |
2578 | if (err) | |
2579 | goto unregister_ioctl; | |
2580 | ||
2581 | err = register_filesystem(&btrfs_fs_type); | |
2582 | if (err) | |
2583 | goto unregister_ioctl; | |
74255aa0 | 2584 | |
2f4cbe64 WB |
2585 | return 0; |
2586 | ||
a9218f6b CM |
2587 | unregister_ioctl: |
2588 | btrfs_interface_exit(); | |
97eb6b69 DS |
2589 | free_end_io_wq: |
2590 | btrfs_end_io_wq_exit(); | |
b9e9a6cb WS |
2591 | free_prelim_ref: |
2592 | btrfs_prelim_ref_exit(); | |
78a6184a MX |
2593 | free_delayed_ref: |
2594 | btrfs_delayed_ref_exit(); | |
9247f317 MX |
2595 | free_auto_defrag: |
2596 | btrfs_auto_defrag_exit(); | |
16cdcec7 MX |
2597 | free_delayed_inode: |
2598 | btrfs_delayed_inode_exit(); | |
6352b91d MX |
2599 | free_ordered_data: |
2600 | ordered_data_exit(); | |
2f4cbe64 WB |
2601 | free_extent_map: |
2602 | extent_map_exit(); | |
6f0d04f8 JB |
2603 | free_extent_state_cache: |
2604 | extent_state_cache_exit(); | |
d1310b2e CM |
2605 | free_extent_io: |
2606 | extent_io_exit(); | |
2f4cbe64 WB |
2607 | free_cachep: |
2608 | btrfs_destroy_cachep(); | |
261507a0 LZ |
2609 | free_compress: |
2610 | btrfs_exit_compress(); | |
2f4cbe64 | 2611 | btrfs_exit_sysfs(); |
9678c543 | 2612 | |
2f4cbe64 | 2613 | return err; |
2e635a27 CM |
2614 | } |
2615 | ||
2616 | static void __exit exit_btrfs_fs(void) | |
2617 | { | |
39279cc3 | 2618 | btrfs_destroy_cachep(); |
78a6184a | 2619 | btrfs_delayed_ref_exit(); |
9247f317 | 2620 | btrfs_auto_defrag_exit(); |
16cdcec7 | 2621 | btrfs_delayed_inode_exit(); |
b9e9a6cb | 2622 | btrfs_prelim_ref_exit(); |
6352b91d | 2623 | ordered_data_exit(); |
a52d9a80 | 2624 | extent_map_exit(); |
6f0d04f8 | 2625 | extent_state_cache_exit(); |
d1310b2e | 2626 | extent_io_exit(); |
a9218f6b | 2627 | btrfs_interface_exit(); |
5ed5f588 | 2628 | btrfs_end_io_wq_exit(); |
2e635a27 | 2629 | unregister_filesystem(&btrfs_fs_type); |
58176a96 | 2630 | btrfs_exit_sysfs(); |
8a4b83cc | 2631 | btrfs_cleanup_fs_uuids(); |
261507a0 | 2632 | btrfs_exit_compress(); |
2e635a27 CM |
2633 | } |
2634 | ||
60efa5eb | 2635 | late_initcall(init_btrfs_fs); |
2e635a27 CM |
2636 | module_exit(exit_btrfs_fs) |
2637 | ||
2638 | MODULE_LICENSE("GPL"); | |
d5178578 | 2639 | MODULE_SOFTDEP("pre: crc32c"); |
3951e7f0 | 2640 | MODULE_SOFTDEP("pre: xxhash64"); |
3831bf00 | 2641 | MODULE_SOFTDEP("pre: sha256"); |
352ae07b | 2642 | MODULE_SOFTDEP("pre: blake2b-256"); |