]>
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> |
22c44fe6 | 26 | #include <linux/ratelimit.h> |
9678c543 | 27 | #include <linux/crc32c.h> |
55e301fd | 28 | #include <linux/btrfs.h> |
16cdcec7 | 29 | #include "delayed-inode.h" |
2e635a27 | 30 | #include "ctree.h" |
e20d96d6 | 31 | #include "disk-io.h" |
d5719762 | 32 | #include "transaction.h" |
2c90e5d6 | 33 | #include "btrfs_inode.h" |
3a686375 | 34 | #include "print-tree.h" |
63541927 | 35 | #include "props.h" |
5103e947 | 36 | #include "xattr.h" |
8a4b83cc | 37 | #include "volumes.h" |
be6e8dc0 | 38 | #include "export.h" |
c8b97818 | 39 | #include "compression.h" |
9c5085c1 | 40 | #include "rcu-string.h" |
8dabb742 | 41 | #include "dev-replace.h" |
74255aa0 | 42 | #include "free-space-cache.h" |
b9e9a6cb | 43 | #include "backref.h" |
8719aaae | 44 | #include "space-info.h" |
89439109 | 45 | #include "sysfs.h" |
b70f5097 | 46 | #include "zoned.h" |
dc11dd5d | 47 | #include "tests/btrfs-tests.h" |
aac0023c | 48 | #include "block-group.h" |
b0643e59 | 49 | #include "discard.h" |
d3982100 | 50 | #include "qgroup.h" |
1abe9b8a | 51 | #define CREATE_TRACE_POINTS |
52 | #include <trace/events/btrfs.h> | |
53 | ||
b87221de | 54 | static const struct super_operations btrfs_super_ops; |
72fa39f5 MT |
55 | |
56 | /* | |
57 | * Types for mounting the default subvolume and a subvolume explicitly | |
58 | * requested by subvol=/path. That way the callchain is straightforward and we | |
59 | * don't have to play tricks with the mount options and recursive calls to | |
60 | * btrfs_mount. | |
312c89fb MT |
61 | * |
62 | * The new btrfs_root_fs_type also servers as a tag for the bdev_holder. | |
72fa39f5 | 63 | */ |
830c4adb | 64 | static struct file_system_type btrfs_fs_type; |
72fa39f5 | 65 | static struct file_system_type btrfs_root_fs_type; |
75dfe396 | 66 | |
0723a047 HH |
67 | static int btrfs_remount(struct super_block *sb, int *flags, char *data); |
68 | ||
c067da87 STD |
69 | #ifdef CONFIG_PRINTK |
70 | ||
71 | #define STATE_STRING_PREFACE ": state " | |
72 | #define STATE_STRING_BUF_LEN (sizeof(STATE_STRING_PREFACE) + BTRFS_FS_STATE_COUNT) | |
73 | ||
74 | /* | |
75 | * Characters to print to indicate error conditions or uncommon filesystem sate. | |
76 | * RO is not an error. | |
77 | */ | |
78 | static const char fs_state_chars[] = { | |
79 | [BTRFS_FS_STATE_ERROR] = 'E', | |
80 | [BTRFS_FS_STATE_REMOUNTING] = 'M', | |
81 | [BTRFS_FS_STATE_RO] = 0, | |
82 | [BTRFS_FS_STATE_TRANS_ABORTED] = 'A', | |
83 | [BTRFS_FS_STATE_DEV_REPLACING] = 'R', | |
84 | [BTRFS_FS_STATE_DUMMY_FS_INFO] = 0, | |
85 | [BTRFS_FS_STATE_NO_CSUMS] = 'C', | |
86 | [BTRFS_FS_STATE_LOG_CLEANUP_ERROR] = 'L', | |
87 | }; | |
88 | ||
89 | static void btrfs_state_to_string(const struct btrfs_fs_info *info, char *buf) | |
90 | { | |
91 | unsigned int bit; | |
92 | bool states_printed = false; | |
93 | unsigned long fs_state = READ_ONCE(info->fs_state); | |
94 | char *curr = buf; | |
95 | ||
96 | memcpy(curr, STATE_STRING_PREFACE, sizeof(STATE_STRING_PREFACE)); | |
97 | curr += sizeof(STATE_STRING_PREFACE) - 1; | |
98 | ||
99 | for_each_set_bit(bit, &fs_state, sizeof(fs_state)) { | |
100 | WARN_ON_ONCE(bit >= BTRFS_FS_STATE_COUNT); | |
101 | if ((bit < BTRFS_FS_STATE_COUNT) && fs_state_chars[bit]) { | |
102 | *curr++ = fs_state_chars[bit]; | |
103 | states_printed = true; | |
104 | } | |
105 | } | |
106 | ||
107 | /* If no states were printed, reset the buffer */ | |
108 | if (!states_printed) | |
109 | curr = buf; | |
110 | ||
111 | *curr++ = 0; | |
112 | } | |
113 | #endif | |
114 | ||
59131393 JB |
115 | /* |
116 | * Generally the error codes correspond to their respective errors, but there | |
117 | * are a few special cases. | |
118 | * | |
119 | * EUCLEAN: Any sort of corruption that we encounter. The tree-checker for | |
120 | * instance will return EUCLEAN if any of the blocks are corrupted in | |
121 | * a way that is problematic. We want to reserve EUCLEAN for these | |
122 | * sort of corruptions. | |
123 | * | |
124 | * EROFS: If we check BTRFS_FS_STATE_ERROR and fail out with a return error, we | |
125 | * need to use EROFS for this case. We will have no idea of the | |
126 | * original failure, that will have been reported at the time we tripped | |
127 | * over the error. Each subsequent error that doesn't have any context | |
128 | * of the original error should use EROFS when handling BTRFS_FS_STATE_ERROR. | |
129 | */ | |
4143cb8b | 130 | const char * __attribute_const__ btrfs_decode_error(int errno) |
acce952b | 131 | { |
08748810 | 132 | char *errstr = "unknown"; |
acce952b | 133 | |
134 | switch (errno) { | |
d54f8144 DS |
135 | case -ENOENT: /* -2 */ |
136 | errstr = "No such entry"; | |
137 | break; | |
138 | case -EIO: /* -5 */ | |
acce952b | 139 | errstr = "IO failure"; |
140 | break; | |
d54f8144 | 141 | case -ENOMEM: /* -12*/ |
acce952b | 142 | errstr = "Out of memory"; |
143 | break; | |
d54f8144 | 144 | case -EEXIST: /* -17 */ |
8c342930 JM |
145 | errstr = "Object already exists"; |
146 | break; | |
d54f8144 | 147 | case -ENOSPC: /* -28 */ |
94ef7280 DS |
148 | errstr = "No space left"; |
149 | break; | |
d54f8144 DS |
150 | case -EROFS: /* -30 */ |
151 | errstr = "Readonly filesystem"; | |
94ef7280 | 152 | break; |
fb8521ca DS |
153 | case -EOPNOTSUPP: /* -95 */ |
154 | errstr = "Operation not supported"; | |
155 | break; | |
156 | case -EUCLEAN: /* -117 */ | |
157 | errstr = "Filesystem corrupted"; | |
158 | break; | |
159 | case -EDQUOT: /* -122 */ | |
160 | errstr = "Quota exceeded"; | |
161 | break; | |
acce952b | 162 | } |
163 | ||
164 | return errstr; | |
165 | } | |
166 | ||
acce952b | 167 | /* |
34d97007 | 168 | * __btrfs_handle_fs_error decodes expected errors from the caller and |
52042d8e | 169 | * invokes the appropriate error response. |
acce952b | 170 | */ |
c0d19e2b | 171 | __cold |
34d97007 | 172 | void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function, |
4da35113 | 173 | unsigned int line, int errno, const char *fmt, ...) |
acce952b | 174 | { |
175 | struct super_block *sb = fs_info->sb; | |
57d816a1 | 176 | #ifdef CONFIG_PRINTK |
c067da87 | 177 | char statestr[STATE_STRING_BUF_LEN]; |
acce952b | 178 | const char *errstr; |
57d816a1 | 179 | #endif |
acce952b | 180 | |
181 | /* | |
182 | * Special case: if the error is EROFS, and we're already | |
1751e8a6 | 183 | * under SB_RDONLY, then it is safe here. |
acce952b | 184 | */ |
bc98a42c | 185 | if (errno == -EROFS && sb_rdonly(sb)) |
4da35113 JM |
186 | return; |
187 | ||
57d816a1 | 188 | #ifdef CONFIG_PRINTK |
08748810 | 189 | errstr = btrfs_decode_error(errno); |
c067da87 | 190 | btrfs_state_to_string(fs_info, statestr); |
4da35113 | 191 | if (fmt) { |
37252a66 ES |
192 | struct va_format vaf; |
193 | va_list args; | |
194 | ||
195 | va_start(args, fmt); | |
196 | vaf.fmt = fmt; | |
197 | vaf.va = &args; | |
4da35113 | 198 | |
c067da87 STD |
199 | pr_crit("BTRFS: error (device %s%s) in %s:%d: errno=%d %s (%pV)\n", |
200 | sb->s_id, statestr, function, line, errno, errstr, &vaf); | |
37252a66 | 201 | va_end(args); |
4da35113 | 202 | } else { |
c067da87 STD |
203 | pr_crit("BTRFS: error (device %s%s) in %s:%d: errno=%d %s\n", |
204 | sb->s_id, statestr, function, line, errno, errstr); | |
4da35113 | 205 | } |
57d816a1 | 206 | #endif |
acce952b | 207 | |
0713d90c AJ |
208 | /* |
209 | * Today we only save the error info to memory. Long term we'll | |
210 | * also send it down to the disk | |
211 | */ | |
212 | set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state); | |
213 | ||
4da35113 | 214 | /* Don't go through full error handling during mount */ |
922ea899 AJ |
215 | if (!(sb->s_flags & SB_BORN)) |
216 | return; | |
217 | ||
218 | if (sb_rdonly(sb)) | |
219 | return; | |
220 | ||
b0643e59 DZ |
221 | btrfs_discard_stop(fs_info); |
222 | ||
922ea899 | 223 | /* btrfs handle error by forcing the filesystem readonly */ |
a0a1db70 | 224 | btrfs_set_sb_rdonly(sb); |
922ea899 AJ |
225 | btrfs_info(fs_info, "forced readonly"); |
226 | /* | |
227 | * Note that a running device replace operation is not canceled here | |
228 | * although there is no way to update the progress. It would add the | |
229 | * risk of a deadlock, therefore the canceling is omitted. The only | |
230 | * penalty is that some I/O remains active until the procedure | |
52042d8e | 231 | * completes. The next time when the filesystem is mounted writable |
922ea899 AJ |
232 | * again, the device replace operation continues. |
233 | */ | |
4da35113 | 234 | } |
acce952b | 235 | |
57d816a1 | 236 | #ifdef CONFIG_PRINTK |
533574c6 | 237 | static const char * const logtypes[] = { |
4da35113 JM |
238 | "emergency", |
239 | "alert", | |
240 | "critical", | |
241 | "error", | |
242 | "warning", | |
243 | "notice", | |
244 | "info", | |
245 | "debug", | |
246 | }; | |
247 | ||
35f4e5e6 NB |
248 | |
249 | /* | |
250 | * Use one ratelimit state per log level so that a flood of less important | |
251 | * messages doesn't cause more important ones to be dropped. | |
252 | */ | |
253 | static struct ratelimit_state printk_limits[] = { | |
254 | RATELIMIT_STATE_INIT(printk_limits[0], DEFAULT_RATELIMIT_INTERVAL, 100), | |
255 | RATELIMIT_STATE_INIT(printk_limits[1], DEFAULT_RATELIMIT_INTERVAL, 100), | |
256 | RATELIMIT_STATE_INIT(printk_limits[2], DEFAULT_RATELIMIT_INTERVAL, 100), | |
257 | RATELIMIT_STATE_INIT(printk_limits[3], DEFAULT_RATELIMIT_INTERVAL, 100), | |
258 | RATELIMIT_STATE_INIT(printk_limits[4], DEFAULT_RATELIMIT_INTERVAL, 100), | |
259 | RATELIMIT_STATE_INIT(printk_limits[5], DEFAULT_RATELIMIT_INTERVAL, 100), | |
260 | RATELIMIT_STATE_INIT(printk_limits[6], DEFAULT_RATELIMIT_INTERVAL, 100), | |
261 | RATELIMIT_STATE_INIT(printk_limits[7], DEFAULT_RATELIMIT_INTERVAL, 100), | |
262 | }; | |
263 | ||
b0a66a31 | 264 | void __cold _btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...) |
4da35113 | 265 | { |
40f7828b | 266 | char lvl[PRINTK_MAX_SINGLE_HEADER_LEN + 1] = "\0"; |
4da35113 JM |
267 | struct va_format vaf; |
268 | va_list args; | |
533574c6 | 269 | int kern_level; |
40f7828b PM |
270 | const char *type = logtypes[4]; |
271 | struct ratelimit_state *ratelimit = &printk_limits[4]; | |
4da35113 JM |
272 | |
273 | va_start(args, fmt); | |
274 | ||
262c5e86 | 275 | while ((kern_level = printk_get_level(fmt)) != 0) { |
533574c6 | 276 | size_t size = printk_skip_level(fmt) - fmt; |
262c5e86 PM |
277 | |
278 | if (kern_level >= '0' && kern_level <= '7') { | |
279 | memcpy(lvl, fmt, size); | |
280 | lvl[size] = '\0'; | |
281 | type = logtypes[kern_level - '0']; | |
282 | ratelimit = &printk_limits[kern_level - '0']; | |
283 | } | |
533574c6 | 284 | fmt += size; |
262c5e86 PM |
285 | } |
286 | ||
4da35113 JM |
287 | vaf.fmt = fmt; |
288 | vaf.va = &args; | |
533574c6 | 289 | |
a0f6d924 | 290 | if (__ratelimit(ratelimit)) { |
c067da87 STD |
291 | if (fs_info) { |
292 | char statestr[STATE_STRING_BUF_LEN]; | |
293 | ||
294 | btrfs_state_to_string(fs_info, statestr); | |
b0a66a31 | 295 | _printk("%sBTRFS %s (device %s%s): %pV\n", lvl, type, |
c067da87 STD |
296 | fs_info->sb->s_id, statestr, &vaf); |
297 | } else { | |
b0a66a31 | 298 | _printk("%sBTRFS %s: %pV\n", lvl, type, &vaf); |
c067da87 | 299 | } |
a0f6d924 | 300 | } |
533574c6 JP |
301 | |
302 | va_end(args); | |
303 | } | |
533574c6 | 304 | #endif |
acce952b | 305 | |
e9306ad4 QW |
306 | #if BITS_PER_LONG == 32 |
307 | void __cold btrfs_warn_32bit_limit(struct btrfs_fs_info *fs_info) | |
308 | { | |
309 | if (!test_and_set_bit(BTRFS_FS_32BIT_WARN, &fs_info->flags)) { | |
310 | btrfs_warn(fs_info, "reaching 32bit limit for logical addresses"); | |
311 | btrfs_warn(fs_info, | |
312 | "due to page cache limit on 32bit systems, btrfs can't access metadata at or beyond %lluT", | |
313 | BTRFS_32BIT_MAX_FILE_SIZE >> 40); | |
314 | btrfs_warn(fs_info, | |
315 | "please consider upgrading to 64bit kernel/hardware"); | |
316 | } | |
317 | } | |
318 | ||
319 | void __cold btrfs_err_32bit_limit(struct btrfs_fs_info *fs_info) | |
320 | { | |
321 | if (!test_and_set_bit(BTRFS_FS_32BIT_ERROR, &fs_info->flags)) { | |
322 | btrfs_err(fs_info, "reached 32bit limit for logical addresses"); | |
323 | btrfs_err(fs_info, | |
324 | "due to page cache limit on 32bit systems, metadata beyond %lluT can't be accessed", | |
325 | BTRFS_32BIT_MAX_FILE_SIZE >> 40); | |
326 | btrfs_err(fs_info, | |
327 | "please consider upgrading to 64bit kernel/hardware"); | |
328 | } | |
329 | } | |
330 | #endif | |
331 | ||
49b25e05 JM |
332 | /* |
333 | * We only mark the transaction aborted and then set the file system read-only. | |
334 | * This will prevent new transactions from starting or trying to join this | |
335 | * one. | |
336 | * | |
337 | * This means that error recovery at the call site is limited to freeing | |
338 | * any local memory allocations and passing the error code up without | |
339 | * further cleanup. The transaction should complete as it normally would | |
340 | * in the call path but will return -EIO. | |
341 | * | |
342 | * We'll complete the cleanup in btrfs_end_transaction and | |
343 | * btrfs_commit_transaction. | |
344 | */ | |
c0d19e2b | 345 | __cold |
49b25e05 | 346 | void __btrfs_abort_transaction(struct btrfs_trans_handle *trans, |
66642832 | 347 | const char *function, |
49b25e05 JM |
348 | unsigned int line, int errno) |
349 | { | |
66642832 JM |
350 | struct btrfs_fs_info *fs_info = trans->fs_info; |
351 | ||
bf31f87f | 352 | WRITE_ONCE(trans->aborted, errno); |
20c7bcec | 353 | WRITE_ONCE(trans->transaction->aborted, errno); |
501407aa | 354 | /* Wake up anybody who may be waiting on this transaction */ |
66642832 JM |
355 | wake_up(&fs_info->transaction_wait); |
356 | wake_up(&fs_info->transaction_blocked_wait); | |
357 | __btrfs_handle_fs_error(fs_info, function, line, errno, NULL); | |
49b25e05 | 358 | } |
8c342930 JM |
359 | /* |
360 | * __btrfs_panic decodes unexpected, fatal errors from the caller, | |
361 | * issues an alert, and either panics or BUGs, depending on mount options. | |
362 | */ | |
c0d19e2b | 363 | __cold |
8c342930 JM |
364 | void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function, |
365 | unsigned int line, int errno, const char *fmt, ...) | |
366 | { | |
8c342930 JM |
367 | char *s_id = "<unknown>"; |
368 | const char *errstr; | |
369 | struct va_format vaf = { .fmt = fmt }; | |
370 | va_list args; | |
acce952b | 371 | |
8c342930 JM |
372 | if (fs_info) |
373 | s_id = fs_info->sb->s_id; | |
acce952b | 374 | |
8c342930 JM |
375 | va_start(args, fmt); |
376 | vaf.va = &args; | |
377 | ||
08748810 | 378 | errstr = btrfs_decode_error(errno); |
d8953d69 | 379 | if (fs_info && (btrfs_test_opt(fs_info, PANIC_ON_FATAL_ERROR))) |
08748810 DS |
380 | panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n", |
381 | s_id, function, line, &vaf, errno, errstr); | |
8c342930 | 382 | |
efe120a0 FH |
383 | btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)", |
384 | function, line, &vaf, errno, errstr); | |
8c342930 JM |
385 | va_end(args); |
386 | /* Caller calls BUG() */ | |
acce952b | 387 | } |
388 | ||
d397712b | 389 | static void btrfs_put_super(struct super_block *sb) |
b18c6685 | 390 | { |
6bccf3ab | 391 | close_ctree(btrfs_sb(sb)); |
75dfe396 CM |
392 | } |
393 | ||
95e05289 | 394 | enum { |
416a7202 DS |
395 | Opt_acl, Opt_noacl, |
396 | Opt_clear_cache, | |
397 | Opt_commit_interval, | |
398 | Opt_compress, | |
399 | Opt_compress_force, | |
400 | Opt_compress_force_type, | |
401 | Opt_compress_type, | |
402 | Opt_degraded, | |
403 | Opt_device, | |
404 | Opt_fatal_errors, | |
405 | Opt_flushoncommit, Opt_noflushoncommit, | |
416a7202 DS |
406 | Opt_max_inline, |
407 | Opt_barrier, Opt_nobarrier, | |
408 | Opt_datacow, Opt_nodatacow, | |
409 | Opt_datasum, Opt_nodatasum, | |
410 | Opt_defrag, Opt_nodefrag, | |
411 | Opt_discard, Opt_nodiscard, | |
b0643e59 | 412 | Opt_discard_mode, |
416a7202 DS |
413 | Opt_norecovery, |
414 | Opt_ratio, | |
415 | Opt_rescan_uuid_tree, | |
416 | Opt_skip_balance, | |
417 | Opt_space_cache, Opt_no_space_cache, | |
418 | Opt_space_cache_version, | |
419 | Opt_ssd, Opt_nossd, | |
420 | Opt_ssd_spread, Opt_nossd_spread, | |
421 | Opt_subvol, | |
37becec9 | 422 | Opt_subvol_empty, |
416a7202 DS |
423 | Opt_subvolid, |
424 | Opt_thread_pool, | |
425 | Opt_treelog, Opt_notreelog, | |
416a7202 DS |
426 | Opt_user_subvol_rm_allowed, |
427 | ||
74ef0018 QW |
428 | /* Rescue options */ |
429 | Opt_rescue, | |
430 | Opt_usebackuproot, | |
431 | Opt_nologreplay, | |
42437a63 | 432 | Opt_ignorebadroots, |
882dbe0c | 433 | Opt_ignoredatacsums, |
9037d3cb | 434 | Opt_rescue_all, |
74ef0018 | 435 | |
416a7202 | 436 | /* Deprecated options */ |
416a7202 | 437 | Opt_recovery, |
5297199a | 438 | Opt_inode_cache, Opt_noinode_cache, |
416a7202 DS |
439 | |
440 | /* Debugging options */ | |
441 | Opt_check_integrity, | |
70f6d82e | 442 | Opt_check_integrity_including_extent_data, |
416a7202 DS |
443 | Opt_check_integrity_print_mask, |
444 | Opt_enospc_debug, Opt_noenospc_debug, | |
d0bd4560 JB |
445 | #ifdef CONFIG_BTRFS_DEBUG |
446 | Opt_fragment_data, Opt_fragment_metadata, Opt_fragment_all, | |
fb592373 JB |
447 | #endif |
448 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
449 | Opt_ref_verify, | |
d0bd4560 | 450 | #endif |
9555c6c1 | 451 | Opt_err, |
95e05289 CM |
452 | }; |
453 | ||
4d4ab6d6 | 454 | static const match_table_t tokens = { |
416a7202 DS |
455 | {Opt_acl, "acl"}, |
456 | {Opt_noacl, "noacl"}, | |
457 | {Opt_clear_cache, "clear_cache"}, | |
458 | {Opt_commit_interval, "commit=%u"}, | |
c8b97818 | 459 | {Opt_compress, "compress"}, |
261507a0 | 460 | {Opt_compress_type, "compress=%s"}, |
a555f810 | 461 | {Opt_compress_force, "compress-force"}, |
261507a0 | 462 | {Opt_compress_force_type, "compress-force=%s"}, |
416a7202 DS |
463 | {Opt_degraded, "degraded"}, |
464 | {Opt_device, "device=%s"}, | |
465 | {Opt_fatal_errors, "fatal_errors=%s"}, | |
dccae999 | 466 | {Opt_flushoncommit, "flushoncommit"}, |
2c9ee856 | 467 | {Opt_noflushoncommit, "noflushoncommit"}, |
416a7202 DS |
468 | {Opt_inode_cache, "inode_cache"}, |
469 | {Opt_noinode_cache, "noinode_cache"}, | |
470 | {Opt_max_inline, "max_inline=%s"}, | |
471 | {Opt_barrier, "barrier"}, | |
472 | {Opt_nobarrier, "nobarrier"}, | |
473 | {Opt_datacow, "datacow"}, | |
474 | {Opt_nodatacow, "nodatacow"}, | |
475 | {Opt_datasum, "datasum"}, | |
476 | {Opt_nodatasum, "nodatasum"}, | |
477 | {Opt_defrag, "autodefrag"}, | |
478 | {Opt_nodefrag, "noautodefrag"}, | |
e244a0ae | 479 | {Opt_discard, "discard"}, |
b0643e59 | 480 | {Opt_discard_mode, "discard=%s"}, |
e07a2ade | 481 | {Opt_nodiscard, "nodiscard"}, |
416a7202 DS |
482 | {Opt_norecovery, "norecovery"}, |
483 | {Opt_ratio, "metadata_ratio=%u"}, | |
484 | {Opt_rescan_uuid_tree, "rescan_uuid_tree"}, | |
485 | {Opt_skip_balance, "skip_balance"}, | |
0af3d00b | 486 | {Opt_space_cache, "space_cache"}, |
8965593e | 487 | {Opt_no_space_cache, "nospace_cache"}, |
416a7202 DS |
488 | {Opt_space_cache_version, "space_cache=%s"}, |
489 | {Opt_ssd, "ssd"}, | |
490 | {Opt_nossd, "nossd"}, | |
491 | {Opt_ssd_spread, "ssd_spread"}, | |
492 | {Opt_nossd_spread, "nossd_spread"}, | |
493 | {Opt_subvol, "subvol=%s"}, | |
37becec9 | 494 | {Opt_subvol_empty, "subvol="}, |
416a7202 DS |
495 | {Opt_subvolid, "subvolid=%s"}, |
496 | {Opt_thread_pool, "thread_pool=%u"}, | |
497 | {Opt_treelog, "treelog"}, | |
498 | {Opt_notreelog, "notreelog"}, | |
416a7202 DS |
499 | {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"}, |
500 | ||
74ef0018 QW |
501 | /* Rescue options */ |
502 | {Opt_rescue, "rescue=%s"}, | |
503 | /* Deprecated, with alias rescue=nologreplay */ | |
504 | {Opt_nologreplay, "nologreplay"}, | |
505 | /* Deprecated, with alias rescue=usebackuproot */ | |
506 | {Opt_usebackuproot, "usebackuproot"}, | |
507 | ||
416a7202 | 508 | /* Deprecated options */ |
416a7202 | 509 | {Opt_recovery, "recovery"}, |
416a7202 DS |
510 | |
511 | /* Debugging options */ | |
21adbd5c SB |
512 | {Opt_check_integrity, "check_int"}, |
513 | {Opt_check_integrity_including_extent_data, "check_int_data"}, | |
02453bde | 514 | {Opt_check_integrity_print_mask, "check_int_print_mask=%u"}, |
416a7202 DS |
515 | {Opt_enospc_debug, "enospc_debug"}, |
516 | {Opt_noenospc_debug, "noenospc_debug"}, | |
d0bd4560 JB |
517 | #ifdef CONFIG_BTRFS_DEBUG |
518 | {Opt_fragment_data, "fragment=data"}, | |
519 | {Opt_fragment_metadata, "fragment=metadata"}, | |
520 | {Opt_fragment_all, "fragment=all"}, | |
fb592373 JB |
521 | #endif |
522 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
523 | {Opt_ref_verify, "ref_verify"}, | |
d0bd4560 | 524 | #endif |
33268eaf | 525 | {Opt_err, NULL}, |
95e05289 CM |
526 | }; |
527 | ||
74ef0018 QW |
528 | static const match_table_t rescue_tokens = { |
529 | {Opt_usebackuproot, "usebackuproot"}, | |
530 | {Opt_nologreplay, "nologreplay"}, | |
42437a63 JB |
531 | {Opt_ignorebadroots, "ignorebadroots"}, |
532 | {Opt_ignorebadroots, "ibadroots"}, | |
882dbe0c JB |
533 | {Opt_ignoredatacsums, "ignoredatacsums"}, |
534 | {Opt_ignoredatacsums, "idatacsums"}, | |
9037d3cb | 535 | {Opt_rescue_all, "all"}, |
74ef0018 QW |
536 | {Opt_err, NULL}, |
537 | }; | |
538 | ||
d70bf748 JB |
539 | static bool check_ro_option(struct btrfs_fs_info *fs_info, unsigned long opt, |
540 | const char *opt_name) | |
541 | { | |
542 | if (fs_info->mount_opt & opt) { | |
543 | btrfs_err(fs_info, "%s must be used with ro mount option", | |
544 | opt_name); | |
545 | return true; | |
546 | } | |
547 | return false; | |
548 | } | |
549 | ||
74ef0018 QW |
550 | static int parse_rescue_options(struct btrfs_fs_info *info, const char *options) |
551 | { | |
552 | char *opts; | |
553 | char *orig; | |
554 | char *p; | |
555 | substring_t args[MAX_OPT_ARGS]; | |
556 | int ret = 0; | |
557 | ||
558 | opts = kstrdup(options, GFP_KERNEL); | |
559 | if (!opts) | |
560 | return -ENOMEM; | |
561 | orig = opts; | |
562 | ||
563 | while ((p = strsep(&opts, ":")) != NULL) { | |
564 | int token; | |
565 | ||
566 | if (!*p) | |
567 | continue; | |
568 | token = match_token(p, rescue_tokens, args); | |
569 | switch (token){ | |
570 | case Opt_usebackuproot: | |
571 | btrfs_info(info, | |
572 | "trying to use backup root at mount time"); | |
573 | btrfs_set_opt(info->mount_opt, USEBACKUPROOT); | |
574 | break; | |
575 | case Opt_nologreplay: | |
576 | btrfs_set_and_info(info, NOLOGREPLAY, | |
577 | "disabling log replay at mount time"); | |
578 | break; | |
42437a63 JB |
579 | case Opt_ignorebadroots: |
580 | btrfs_set_and_info(info, IGNOREBADROOTS, | |
581 | "ignoring bad roots"); | |
582 | break; | |
882dbe0c JB |
583 | case Opt_ignoredatacsums: |
584 | btrfs_set_and_info(info, IGNOREDATACSUMS, | |
585 | "ignoring data csums"); | |
586 | break; | |
9037d3cb JB |
587 | case Opt_rescue_all: |
588 | btrfs_info(info, "enabling all of the rescue options"); | |
589 | btrfs_set_and_info(info, IGNOREDATACSUMS, | |
590 | "ignoring data csums"); | |
591 | btrfs_set_and_info(info, IGNOREBADROOTS, | |
592 | "ignoring bad roots"); | |
593 | btrfs_set_and_info(info, NOLOGREPLAY, | |
594 | "disabling log replay at mount time"); | |
595 | break; | |
74ef0018 QW |
596 | case Opt_err: |
597 | btrfs_info(info, "unrecognized rescue option '%s'", p); | |
598 | ret = -EINVAL; | |
599 | goto out; | |
600 | default: | |
601 | break; | |
602 | } | |
603 | ||
604 | } | |
605 | out: | |
606 | kfree(orig); | |
607 | return ret; | |
608 | } | |
609 | ||
edf24abe CH |
610 | /* |
611 | * Regular mount options parser. Everything that is needed only when | |
612 | * reading in a new superblock is parsed here. | |
49b25e05 | 613 | * XXX JDM: This needs to be cleaned up for remount. |
edf24abe | 614 | */ |
2ff7e61e | 615 | int btrfs_parse_options(struct btrfs_fs_info *info, char *options, |
96da0919 | 616 | unsigned long new_flags) |
95e05289 | 617 | { |
95e05289 | 618 | substring_t args[MAX_OPT_ARGS]; |
e215772c | 619 | char *p, *num; |
4543df7e | 620 | int intarg; |
a7a3f7ca | 621 | int ret = 0; |
261507a0 LZ |
622 | char *compress_type; |
623 | bool compress_force = false; | |
b7c47bbb | 624 | enum btrfs_compression_type saved_compress_type; |
27942c99 | 625 | int saved_compress_level; |
b7c47bbb TI |
626 | bool saved_compress_force; |
627 | int no_compress = 0; | |
b6cda9bc | 628 | |
0b246afa | 629 | if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE)) |
70f6d82e | 630 | btrfs_set_opt(info->mount_opt, FREE_SPACE_TREE); |
94846229 | 631 | else if (btrfs_free_space_cache_v1_active(info)) { |
5d1ab66c NA |
632 | if (btrfs_is_zoned(info)) { |
633 | btrfs_info(info, | |
634 | "zoned: clearing existing space cache"); | |
635 | btrfs_set_super_cache_generation(info->super_copy, 0); | |
636 | } else { | |
637 | btrfs_set_opt(info->mount_opt, SPACE_CACHE); | |
638 | } | |
639 | } | |
73bc1876 | 640 | |
96da0919 QW |
641 | /* |
642 | * Even the options are empty, we still need to do extra check | |
643 | * against new flags | |
644 | */ | |
95e05289 | 645 | if (!options) |
96da0919 | 646 | goto check; |
95e05289 | 647 | |
edf24abe | 648 | while ((p = strsep(&options, ",")) != NULL) { |
95e05289 CM |
649 | int token; |
650 | if (!*p) | |
651 | continue; | |
652 | ||
653 | token = match_token(p, tokens, args); | |
654 | switch (token) { | |
dfe25020 | 655 | case Opt_degraded: |
0b246afa | 656 | btrfs_info(info, "allowing degraded mounts"); |
edf24abe | 657 | btrfs_set_opt(info->mount_opt, DEGRADED); |
dfe25020 | 658 | break; |
95e05289 | 659 | case Opt_subvol: |
37becec9 | 660 | case Opt_subvol_empty: |
73f73415 | 661 | case Opt_subvolid: |
43e570b0 | 662 | case Opt_device: |
edf24abe | 663 | /* |
fa59f27c AJ |
664 | * These are parsed by btrfs_parse_subvol_options or |
665 | * btrfs_parse_device_options and can be ignored here. | |
edf24abe | 666 | */ |
b6cda9bc CM |
667 | break; |
668 | case Opt_nodatasum: | |
3cdde224 | 669 | btrfs_set_and_info(info, NODATASUM, |
07802534 | 670 | "setting nodatasum"); |
be20aa9d | 671 | break; |
d399167d | 672 | case Opt_datasum: |
3cdde224 JM |
673 | if (btrfs_test_opt(info, NODATASUM)) { |
674 | if (btrfs_test_opt(info, NODATACOW)) | |
0b246afa | 675 | btrfs_info(info, |
5d163e0e | 676 | "setting datasum, datacow enabled"); |
07802534 | 677 | else |
0b246afa | 678 | btrfs_info(info, "setting datasum"); |
07802534 | 679 | } |
d399167d QW |
680 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
681 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
682 | break; | |
be20aa9d | 683 | case Opt_nodatacow: |
3cdde224 JM |
684 | if (!btrfs_test_opt(info, NODATACOW)) { |
685 | if (!btrfs_test_opt(info, COMPRESS) || | |
686 | !btrfs_test_opt(info, FORCE_COMPRESS)) { | |
0b246afa | 687 | btrfs_info(info, |
07802534 QW |
688 | "setting nodatacow, compression disabled"); |
689 | } else { | |
0b246afa | 690 | btrfs_info(info, "setting nodatacow"); |
07802534 | 691 | } |
bedb2cca | 692 | } |
bedb2cca AP |
693 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
694 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
edf24abe CH |
695 | btrfs_set_opt(info->mount_opt, NODATACOW); |
696 | btrfs_set_opt(info->mount_opt, NODATASUM); | |
95e05289 | 697 | break; |
a258af7a | 698 | case Opt_datacow: |
3cdde224 | 699 | btrfs_clear_and_info(info, NODATACOW, |
07802534 | 700 | "setting datacow"); |
a258af7a | 701 | break; |
a555f810 | 702 | case Opt_compress_force: |
261507a0 LZ |
703 | case Opt_compress_force_type: |
704 | compress_force = true; | |
c730ae0c | 705 | fallthrough; |
261507a0 LZ |
706 | case Opt_compress: |
707 | case Opt_compress_type: | |
3cdde224 JM |
708 | saved_compress_type = btrfs_test_opt(info, |
709 | COMPRESS) ? | |
b7c47bbb TI |
710 | info->compress_type : BTRFS_COMPRESS_NONE; |
711 | saved_compress_force = | |
3cdde224 | 712 | btrfs_test_opt(info, FORCE_COMPRESS); |
27942c99 | 713 | saved_compress_level = info->compress_level; |
261507a0 LZ |
714 | if (token == Opt_compress || |
715 | token == Opt_compress_force || | |
a7164fa4 | 716 | strncmp(args[0].from, "zlib", 4) == 0) { |
261507a0 | 717 | compress_type = "zlib"; |
eae8d825 | 718 | |
261507a0 | 719 | info->compress_type = BTRFS_COMPRESS_ZLIB; |
eae8d825 QW |
720 | info->compress_level = BTRFS_ZLIB_DEFAULT_LEVEL; |
721 | /* | |
722 | * args[0] contains uninitialized data since | |
723 | * for these tokens we don't expect any | |
724 | * parameter. | |
725 | */ | |
726 | if (token != Opt_compress && | |
727 | token != Opt_compress_force) | |
728 | info->compress_level = | |
d0ab62ce DZ |
729 | btrfs_compress_str2level( |
730 | BTRFS_COMPRESS_ZLIB, | |
731 | args[0].from + 4); | |
063849ea | 732 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
733 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
734 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
b7c47bbb | 735 | no_compress = 0; |
a7164fa4 | 736 | } else if (strncmp(args[0].from, "lzo", 3) == 0) { |
a6fa6fae LZ |
737 | compress_type = "lzo"; |
738 | info->compress_type = BTRFS_COMPRESS_LZO; | |
282dd7d7 | 739 | info->compress_level = 0; |
063849ea | 740 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
741 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
742 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
2b0ce2c2 | 743 | btrfs_set_fs_incompat(info, COMPRESS_LZO); |
b7c47bbb | 744 | no_compress = 0; |
3f93aef5 | 745 | } else if (strncmp(args[0].from, "zstd", 4) == 0) { |
5c1aab1d NT |
746 | compress_type = "zstd"; |
747 | info->compress_type = BTRFS_COMPRESS_ZSTD; | |
3f93aef5 DZ |
748 | info->compress_level = |
749 | btrfs_compress_str2level( | |
750 | BTRFS_COMPRESS_ZSTD, | |
751 | args[0].from + 4); | |
5c1aab1d NT |
752 | btrfs_set_opt(info->mount_opt, COMPRESS); |
753 | btrfs_clear_opt(info->mount_opt, NODATACOW); | |
754 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
755 | btrfs_set_fs_incompat(info, COMPRESS_ZSTD); | |
756 | no_compress = 0; | |
063849ea AH |
757 | } else if (strncmp(args[0].from, "no", 2) == 0) { |
758 | compress_type = "no"; | |
27942c99 DS |
759 | info->compress_level = 0; |
760 | info->compress_type = 0; | |
063849ea AH |
761 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
762 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
763 | compress_force = false; | |
b7c47bbb | 764 | no_compress++; |
261507a0 LZ |
765 | } else { |
766 | ret = -EINVAL; | |
767 | goto out; | |
768 | } | |
769 | ||
261507a0 | 770 | if (compress_force) { |
b7c47bbb | 771 | btrfs_set_opt(info->mount_opt, FORCE_COMPRESS); |
143f3636 | 772 | } else { |
4027e0f4 WS |
773 | /* |
774 | * If we remount from compress-force=xxx to | |
775 | * compress=xxx, we need clear FORCE_COMPRESS | |
776 | * flag, otherwise, there is no way for users | |
777 | * to disable forcible compression separately. | |
778 | */ | |
779 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
a7e252af | 780 | } |
27942c99 DS |
781 | if (no_compress == 1) { |
782 | btrfs_info(info, "use no compression"); | |
783 | } else if ((info->compress_type != saved_compress_type) || | |
784 | (compress_force != saved_compress_force) || | |
785 | (info->compress_level != saved_compress_level)) { | |
f51d2b59 | 786 | btrfs_info(info, "%s %s compression, level %d", |
b7c47bbb | 787 | (compress_force) ? "force" : "use", |
f51d2b59 | 788 | compress_type, info->compress_level); |
b7c47bbb TI |
789 | } |
790 | compress_force = false; | |
a555f810 | 791 | break; |
e18e4809 | 792 | case Opt_ssd: |
3cdde224 | 793 | btrfs_set_and_info(info, SSD, |
583b7231 | 794 | "enabling ssd optimizations"); |
951e7966 | 795 | btrfs_clear_opt(info->mount_opt, NOSSD); |
e18e4809 | 796 | break; |
451d7585 | 797 | case Opt_ssd_spread: |
583b7231 HK |
798 | btrfs_set_and_info(info, SSD, |
799 | "enabling ssd optimizations"); | |
3cdde224 | 800 | btrfs_set_and_info(info, SSD_SPREAD, |
583b7231 | 801 | "using spread ssd allocation scheme"); |
951e7966 | 802 | btrfs_clear_opt(info->mount_opt, NOSSD); |
451d7585 | 803 | break; |
3b30c22f | 804 | case Opt_nossd: |
583b7231 HK |
805 | btrfs_set_opt(info->mount_opt, NOSSD); |
806 | btrfs_clear_and_info(info, SSD, | |
807 | "not using ssd optimizations"); | |
c730ae0c | 808 | fallthrough; |
62b8e077 | 809 | case Opt_nossd_spread: |
583b7231 HK |
810 | btrfs_clear_and_info(info, SSD_SPREAD, |
811 | "not using spread ssd allocation scheme"); | |
3b30c22f | 812 | break; |
842bef58 | 813 | case Opt_barrier: |
3cdde224 | 814 | btrfs_clear_and_info(info, NOBARRIER, |
07802534 | 815 | "turning on barriers"); |
842bef58 | 816 | break; |
21ad10cf | 817 | case Opt_nobarrier: |
3cdde224 | 818 | btrfs_set_and_info(info, NOBARRIER, |
07802534 | 819 | "turning off barriers"); |
21ad10cf | 820 | break; |
4543df7e | 821 | case Opt_thread_pool: |
2c334e87 WS |
822 | ret = match_int(&args[0], &intarg); |
823 | if (ret) { | |
824 | goto out; | |
f7b885be | 825 | } else if (intarg == 0) { |
2c334e87 WS |
826 | ret = -EINVAL; |
827 | goto out; | |
828 | } | |
f7b885be | 829 | info->thread_pool_size = intarg; |
4543df7e | 830 | break; |
6f568d35 | 831 | case Opt_max_inline: |
edf24abe CH |
832 | num = match_strdup(&args[0]); |
833 | if (num) { | |
91748467 | 834 | info->max_inline = memparse(num, NULL); |
edf24abe CH |
835 | kfree(num); |
836 | ||
15ada040 | 837 | if (info->max_inline) { |
feb5f965 | 838 | info->max_inline = min_t(u64, |
15ada040 | 839 | info->max_inline, |
0b246afa | 840 | info->sectorsize); |
15ada040 | 841 | } |
0b246afa JM |
842 | btrfs_info(info, "max_inline at %llu", |
843 | info->max_inline); | |
2c334e87 WS |
844 | } else { |
845 | ret = -ENOMEM; | |
846 | goto out; | |
6f568d35 CM |
847 | } |
848 | break; | |
bd0330ad | 849 | case Opt_acl: |
45ff35d6 | 850 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
1751e8a6 | 851 | info->sb->s_flags |= SB_POSIXACL; |
bd0330ad | 852 | break; |
45ff35d6 | 853 | #else |
0b246afa | 854 | btrfs_err(info, "support for ACL not compiled in!"); |
45ff35d6 GZ |
855 | ret = -EINVAL; |
856 | goto out; | |
857 | #endif | |
33268eaf | 858 | case Opt_noacl: |
1751e8a6 | 859 | info->sb->s_flags &= ~SB_POSIXACL; |
33268eaf | 860 | break; |
3a5e1404 | 861 | case Opt_notreelog: |
3cdde224 | 862 | btrfs_set_and_info(info, NOTREELOG, |
07802534 | 863 | "disabling tree log"); |
a88998f2 QW |
864 | break; |
865 | case Opt_treelog: | |
3cdde224 | 866 | btrfs_clear_and_info(info, NOTREELOG, |
07802534 | 867 | "enabling tree log"); |
3a5e1404 | 868 | break; |
fed8f166 | 869 | case Opt_norecovery: |
96da0919 | 870 | case Opt_nologreplay: |
74ef0018 QW |
871 | btrfs_warn(info, |
872 | "'nologreplay' is deprecated, use 'rescue=nologreplay' instead"); | |
3cdde224 | 873 | btrfs_set_and_info(info, NOLOGREPLAY, |
96da0919 QW |
874 | "disabling log replay at mount time"); |
875 | break; | |
dccae999 | 876 | case Opt_flushoncommit: |
3cdde224 | 877 | btrfs_set_and_info(info, FLUSHONCOMMIT, |
07802534 | 878 | "turning on flush-on-commit"); |
dccae999 | 879 | break; |
2c9ee856 | 880 | case Opt_noflushoncommit: |
3cdde224 | 881 | btrfs_clear_and_info(info, FLUSHONCOMMIT, |
07802534 | 882 | "turning off flush-on-commit"); |
2c9ee856 | 883 | break; |
97e728d4 | 884 | case Opt_ratio: |
2c334e87 | 885 | ret = match_int(&args[0], &intarg); |
764cb8b4 | 886 | if (ret) |
2c334e87 | 887 | goto out; |
764cb8b4 AJ |
888 | info->metadata_ratio = intarg; |
889 | btrfs_info(info, "metadata ratio %u", | |
890 | info->metadata_ratio); | |
97e728d4 | 891 | break; |
e244a0ae | 892 | case Opt_discard: |
b0643e59 DZ |
893 | case Opt_discard_mode: |
894 | if (token == Opt_discard || | |
895 | strcmp(args[0].from, "sync") == 0) { | |
896 | btrfs_clear_opt(info->mount_opt, DISCARD_ASYNC); | |
897 | btrfs_set_and_info(info, DISCARD_SYNC, | |
898 | "turning on sync discard"); | |
899 | } else if (strcmp(args[0].from, "async") == 0) { | |
900 | btrfs_clear_opt(info->mount_opt, DISCARD_SYNC); | |
901 | btrfs_set_and_info(info, DISCARD_ASYNC, | |
902 | "turning on async discard"); | |
903 | } else { | |
904 | ret = -EINVAL; | |
905 | goto out; | |
906 | } | |
e244a0ae | 907 | break; |
e07a2ade | 908 | case Opt_nodiscard: |
46b27f50 | 909 | btrfs_clear_and_info(info, DISCARD_SYNC, |
07802534 | 910 | "turning off discard"); |
b0643e59 DZ |
911 | btrfs_clear_and_info(info, DISCARD_ASYNC, |
912 | "turning off async discard"); | |
e07a2ade | 913 | break; |
0af3d00b | 914 | case Opt_space_cache: |
70f6d82e | 915 | case Opt_space_cache_version: |
63cd070d JB |
916 | /* |
917 | * We already set FREE_SPACE_TREE above because we have | |
918 | * compat_ro(FREE_SPACE_TREE) set, and we aren't going | |
919 | * to allow v1 to be set for extent tree v2, simply | |
920 | * ignore this setting if we're extent tree v2. | |
921 | */ | |
922 | if (btrfs_fs_incompat(info, EXTENT_TREE_V2)) | |
923 | break; | |
70f6d82e OS |
924 | if (token == Opt_space_cache || |
925 | strcmp(args[0].from, "v1") == 0) { | |
0b246afa | 926 | btrfs_clear_opt(info->mount_opt, |
70f6d82e | 927 | FREE_SPACE_TREE); |
3cdde224 | 928 | btrfs_set_and_info(info, SPACE_CACHE, |
0b246afa | 929 | "enabling disk space caching"); |
70f6d82e | 930 | } else if (strcmp(args[0].from, "v2") == 0) { |
0b246afa | 931 | btrfs_clear_opt(info->mount_opt, |
70f6d82e | 932 | SPACE_CACHE); |
0b246afa | 933 | btrfs_set_and_info(info, FREE_SPACE_TREE, |
70f6d82e OS |
934 | "enabling free space tree"); |
935 | } else { | |
936 | ret = -EINVAL; | |
937 | goto out; | |
938 | } | |
0de90876 | 939 | break; |
f420ee1e SB |
940 | case Opt_rescan_uuid_tree: |
941 | btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE); | |
942 | break; | |
73bc1876 | 943 | case Opt_no_space_cache: |
63cd070d JB |
944 | /* |
945 | * We cannot operate without the free space tree with | |
946 | * extent tree v2, ignore this option. | |
947 | */ | |
948 | if (btrfs_fs_incompat(info, EXTENT_TREE_V2)) | |
949 | break; | |
3cdde224 | 950 | if (btrfs_test_opt(info, SPACE_CACHE)) { |
0b246afa JM |
951 | btrfs_clear_and_info(info, SPACE_CACHE, |
952 | "disabling disk space caching"); | |
70f6d82e | 953 | } |
3cdde224 | 954 | if (btrfs_test_opt(info, FREE_SPACE_TREE)) { |
0b246afa JM |
955 | btrfs_clear_and_info(info, FREE_SPACE_TREE, |
956 | "disabling free space tree"); | |
70f6d82e | 957 | } |
73bc1876 | 958 | break; |
4b9465cb | 959 | case Opt_inode_cache: |
3818aea2 | 960 | case Opt_noinode_cache: |
5297199a NB |
961 | btrfs_warn(info, |
962 | "the 'inode_cache' option is deprecated and has no effect since 5.11"); | |
4b9465cb | 963 | break; |
88c2ba3b | 964 | case Opt_clear_cache: |
63cd070d JB |
965 | /* |
966 | * We cannot clear the free space tree with extent tree | |
967 | * v2, ignore this option. | |
968 | */ | |
969 | if (btrfs_fs_incompat(info, EXTENT_TREE_V2)) | |
970 | break; | |
3cdde224 | 971 | btrfs_set_and_info(info, CLEAR_CACHE, |
07802534 | 972 | "force clearing of disk cache"); |
0af3d00b | 973 | break; |
4260f7c7 SW |
974 | case Opt_user_subvol_rm_allowed: |
975 | btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED); | |
976 | break; | |
91435650 CM |
977 | case Opt_enospc_debug: |
978 | btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG); | |
979 | break; | |
53036293 QW |
980 | case Opt_noenospc_debug: |
981 | btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG); | |
982 | break; | |
4cb5300b | 983 | case Opt_defrag: |
3cdde224 | 984 | btrfs_set_and_info(info, AUTO_DEFRAG, |
07802534 | 985 | "enabling auto defrag"); |
4cb5300b | 986 | break; |
fc0ca9af | 987 | case Opt_nodefrag: |
3cdde224 | 988 | btrfs_clear_and_info(info, AUTO_DEFRAG, |
07802534 | 989 | "disabling auto defrag"); |
fc0ca9af | 990 | break; |
af31f5e5 | 991 | case Opt_recovery: |
8dcddfa0 | 992 | case Opt_usebackuproot: |
74ef0018 QW |
993 | btrfs_warn(info, |
994 | "'%s' is deprecated, use 'rescue=usebackuproot' instead", | |
995 | token == Opt_recovery ? "recovery" : | |
996 | "usebackuproot"); | |
0b246afa | 997 | btrfs_info(info, |
8dcddfa0 QW |
998 | "trying to use backup root at mount time"); |
999 | btrfs_set_opt(info->mount_opt, USEBACKUPROOT); | |
af31f5e5 | 1000 | break; |
9555c6c1 ID |
1001 | case Opt_skip_balance: |
1002 | btrfs_set_opt(info->mount_opt, SKIP_BALANCE); | |
1003 | break; | |
21adbd5c SB |
1004 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
1005 | case Opt_check_integrity_including_extent_data: | |
0b246afa | 1006 | btrfs_info(info, |
efe120a0 | 1007 | "enabling check integrity including extent data"); |
cbeaae4f | 1008 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY_DATA); |
21adbd5c SB |
1009 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); |
1010 | break; | |
1011 | case Opt_check_integrity: | |
0b246afa | 1012 | btrfs_info(info, "enabling check integrity"); |
21adbd5c SB |
1013 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); |
1014 | break; | |
1015 | case Opt_check_integrity_print_mask: | |
2c334e87 | 1016 | ret = match_int(&args[0], &intarg); |
02453bde | 1017 | if (ret) |
2c334e87 | 1018 | goto out; |
02453bde AJ |
1019 | info->check_integrity_print_mask = intarg; |
1020 | btrfs_info(info, "check_integrity_print_mask 0x%x", | |
1021 | info->check_integrity_print_mask); | |
21adbd5c SB |
1022 | break; |
1023 | #else | |
1024 | case Opt_check_integrity_including_extent_data: | |
1025 | case Opt_check_integrity: | |
1026 | case Opt_check_integrity_print_mask: | |
0b246afa JM |
1027 | btrfs_err(info, |
1028 | "support for check_integrity* not compiled in!"); | |
21adbd5c SB |
1029 | ret = -EINVAL; |
1030 | goto out; | |
1031 | #endif | |
8c342930 JM |
1032 | case Opt_fatal_errors: |
1033 | if (strcmp(args[0].from, "panic") == 0) | |
1034 | btrfs_set_opt(info->mount_opt, | |
1035 | PANIC_ON_FATAL_ERROR); | |
1036 | else if (strcmp(args[0].from, "bug") == 0) | |
1037 | btrfs_clear_opt(info->mount_opt, | |
1038 | PANIC_ON_FATAL_ERROR); | |
1039 | else { | |
1040 | ret = -EINVAL; | |
1041 | goto out; | |
1042 | } | |
1043 | break; | |
8b87dc17 DS |
1044 | case Opt_commit_interval: |
1045 | intarg = 0; | |
1046 | ret = match_int(&args[0], &intarg); | |
d3740608 | 1047 | if (ret) |
8b87dc17 | 1048 | goto out; |
d3740608 | 1049 | if (intarg == 0) { |
0b246afa | 1050 | btrfs_info(info, |
d3740608 | 1051 | "using default commit interval %us", |
5d163e0e | 1052 | BTRFS_DEFAULT_COMMIT_INTERVAL); |
d3740608 AJ |
1053 | intarg = BTRFS_DEFAULT_COMMIT_INTERVAL; |
1054 | } else if (intarg > 300) { | |
1055 | btrfs_warn(info, "excessive commit interval %d", | |
1056 | intarg); | |
8b87dc17 | 1057 | } |
d3740608 | 1058 | info->commit_interval = intarg; |
8b87dc17 | 1059 | break; |
74ef0018 QW |
1060 | case Opt_rescue: |
1061 | ret = parse_rescue_options(info, args[0].from); | |
1062 | if (ret < 0) | |
1063 | goto out; | |
1064 | break; | |
d0bd4560 JB |
1065 | #ifdef CONFIG_BTRFS_DEBUG |
1066 | case Opt_fragment_all: | |
0b246afa | 1067 | btrfs_info(info, "fragmenting all space"); |
d0bd4560 JB |
1068 | btrfs_set_opt(info->mount_opt, FRAGMENT_DATA); |
1069 | btrfs_set_opt(info->mount_opt, FRAGMENT_METADATA); | |
1070 | break; | |
1071 | case Opt_fragment_metadata: | |
0b246afa | 1072 | btrfs_info(info, "fragmenting metadata"); |
d0bd4560 JB |
1073 | btrfs_set_opt(info->mount_opt, |
1074 | FRAGMENT_METADATA); | |
1075 | break; | |
1076 | case Opt_fragment_data: | |
0b246afa | 1077 | btrfs_info(info, "fragmenting data"); |
d0bd4560 JB |
1078 | btrfs_set_opt(info->mount_opt, FRAGMENT_DATA); |
1079 | break; | |
fb592373 JB |
1080 | #endif |
1081 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
1082 | case Opt_ref_verify: | |
1083 | btrfs_info(info, "doing ref verification"); | |
1084 | btrfs_set_opt(info->mount_opt, REF_VERIFY); | |
1085 | break; | |
d0bd4560 | 1086 | #endif |
a7a3f7ca | 1087 | case Opt_err: |
7e8f19e5 | 1088 | btrfs_err(info, "unrecognized mount option '%s'", p); |
a7a3f7ca SW |
1089 | ret = -EINVAL; |
1090 | goto out; | |
95e05289 | 1091 | default: |
be20aa9d | 1092 | break; |
95e05289 CM |
1093 | } |
1094 | } | |
96da0919 | 1095 | check: |
d70bf748 JB |
1096 | /* We're read-only, don't have to check. */ |
1097 | if (new_flags & SB_RDONLY) | |
1098 | goto out; | |
1099 | ||
42437a63 | 1100 | if (check_ro_option(info, BTRFS_MOUNT_NOLOGREPLAY, "nologreplay") || |
882dbe0c JB |
1101 | check_ro_option(info, BTRFS_MOUNT_IGNOREBADROOTS, "ignorebadroots") || |
1102 | check_ro_option(info, BTRFS_MOUNT_IGNOREDATACSUMS, "ignoredatacsums")) | |
96da0919 | 1103 | ret = -EINVAL; |
a7a3f7ca | 1104 | out: |
0b246afa | 1105 | if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE) && |
3cdde224 JM |
1106 | !btrfs_test_opt(info, FREE_SPACE_TREE) && |
1107 | !btrfs_test_opt(info, CLEAR_CACHE)) { | |
0b246afa | 1108 | btrfs_err(info, "cannot disable free space tree"); |
70f6d82e OS |
1109 | ret = -EINVAL; |
1110 | ||
1111 | } | |
5d1ab66c NA |
1112 | if (!ret) |
1113 | ret = btrfs_check_mountopts_zoned(info); | |
3cdde224 | 1114 | if (!ret && btrfs_test_opt(info, SPACE_CACHE)) |
0b246afa | 1115 | btrfs_info(info, "disk space caching is enabled"); |
3cdde224 | 1116 | if (!ret && btrfs_test_opt(info, FREE_SPACE_TREE)) |
0b246afa | 1117 | btrfs_info(info, "using free space tree"); |
a7a3f7ca | 1118 | return ret; |
edf24abe CH |
1119 | } |
1120 | ||
1121 | /* | |
1122 | * Parse mount options that are required early in the mount process. | |
1123 | * | |
1124 | * All other options will be parsed on much later in the mount process and | |
1125 | * only when we need to allocate a new super block. | |
1126 | */ | |
fa59f27c AJ |
1127 | static int btrfs_parse_device_options(const char *options, fmode_t flags, |
1128 | void *holder) | |
edf24abe CH |
1129 | { |
1130 | substring_t args[MAX_OPT_ARGS]; | |
83c8c9bd | 1131 | char *device_name, *opts, *orig, *p; |
36350e95 | 1132 | struct btrfs_device *device = NULL; |
d7407606 MT |
1133 | int error = 0; |
1134 | ||
5139cff5 DS |
1135 | lockdep_assert_held(&uuid_mutex); |
1136 | ||
d7407606 MT |
1137 | if (!options) |
1138 | return 0; | |
1139 | ||
1140 | /* | |
1141 | * strsep changes the string, duplicate it because btrfs_parse_options | |
1142 | * gets called later | |
1143 | */ | |
1144 | opts = kstrdup(options, GFP_KERNEL); | |
1145 | if (!opts) | |
1146 | return -ENOMEM; | |
1147 | orig = opts; | |
1148 | ||
1149 | while ((p = strsep(&opts, ",")) != NULL) { | |
1150 | int token; | |
1151 | ||
1152 | if (!*p) | |
1153 | continue; | |
1154 | ||
1155 | token = match_token(p, tokens, args); | |
1156 | if (token == Opt_device) { | |
1157 | device_name = match_strdup(&args[0]); | |
1158 | if (!device_name) { | |
1159 | error = -ENOMEM; | |
1160 | goto out; | |
1161 | } | |
36350e95 GJ |
1162 | device = btrfs_scan_one_device(device_name, flags, |
1163 | holder); | |
d7407606 | 1164 | kfree(device_name); |
36350e95 GJ |
1165 | if (IS_ERR(device)) { |
1166 | error = PTR_ERR(device); | |
d7407606 | 1167 | goto out; |
36350e95 | 1168 | } |
d7407606 MT |
1169 | } |
1170 | } | |
1171 | ||
1172 | out: | |
1173 | kfree(orig); | |
1174 | return error; | |
1175 | } | |
1176 | ||
1177 | /* | |
1178 | * Parse mount options that are related to subvolume id | |
1179 | * | |
1180 | * The value is later passed to mount_subvol() | |
1181 | */ | |
93b9bcdf GJ |
1182 | static int btrfs_parse_subvol_options(const char *options, char **subvol_name, |
1183 | u64 *subvol_objectid) | |
d7407606 MT |
1184 | { |
1185 | substring_t args[MAX_OPT_ARGS]; | |
1186 | char *opts, *orig, *p; | |
edf24abe | 1187 | int error = 0; |
ccb0e7d1 | 1188 | u64 subvolid; |
edf24abe CH |
1189 | |
1190 | if (!options) | |
830c4adb | 1191 | return 0; |
edf24abe CH |
1192 | |
1193 | /* | |
d7407606 | 1194 | * strsep changes the string, duplicate it because |
fa59f27c | 1195 | * btrfs_parse_device_options gets called later |
edf24abe CH |
1196 | */ |
1197 | opts = kstrdup(options, GFP_KERNEL); | |
1198 | if (!opts) | |
1199 | return -ENOMEM; | |
3f3d0bc0 | 1200 | orig = opts; |
edf24abe CH |
1201 | |
1202 | while ((p = strsep(&opts, ",")) != NULL) { | |
1203 | int token; | |
1204 | if (!*p) | |
1205 | continue; | |
1206 | ||
1207 | token = match_token(p, tokens, args); | |
1208 | switch (token) { | |
1209 | case Opt_subvol: | |
a90e8b6f | 1210 | kfree(*subvol_name); |
edf24abe | 1211 | *subvol_name = match_strdup(&args[0]); |
2c334e87 WS |
1212 | if (!*subvol_name) { |
1213 | error = -ENOMEM; | |
1214 | goto out; | |
1215 | } | |
edf24abe | 1216 | break; |
73f73415 | 1217 | case Opt_subvolid: |
ccb0e7d1 AJ |
1218 | error = match_u64(&args[0], &subvolid); |
1219 | if (error) | |
2c334e87 | 1220 | goto out; |
ccb0e7d1 AJ |
1221 | |
1222 | /* we want the original fs_tree */ | |
1223 | if (subvolid == 0) | |
1224 | subvolid = BTRFS_FS_TREE_OBJECTID; | |
1225 | ||
1226 | *subvol_objectid = subvolid; | |
73f73415 | 1227 | break; |
edf24abe CH |
1228 | default: |
1229 | break; | |
1230 | } | |
1231 | } | |
1232 | ||
830c4adb | 1233 | out: |
3f3d0bc0 | 1234 | kfree(orig); |
edf24abe | 1235 | return error; |
95e05289 CM |
1236 | } |
1237 | ||
c0c907a4 MPS |
1238 | char *btrfs_get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info, |
1239 | u64 subvol_objectid) | |
73f73415 | 1240 | { |
815745cf | 1241 | struct btrfs_root *root = fs_info->tree_root; |
5168489a | 1242 | struct btrfs_root *fs_root = NULL; |
05dbe683 OS |
1243 | struct btrfs_root_ref *root_ref; |
1244 | struct btrfs_inode_ref *inode_ref; | |
1245 | struct btrfs_key key; | |
1246 | struct btrfs_path *path = NULL; | |
1247 | char *name = NULL, *ptr; | |
1248 | u64 dirid; | |
1249 | int len; | |
1250 | int ret; | |
1251 | ||
1252 | path = btrfs_alloc_path(); | |
1253 | if (!path) { | |
1254 | ret = -ENOMEM; | |
1255 | goto err; | |
1256 | } | |
05dbe683 | 1257 | |
3ec83621 | 1258 | name = kmalloc(PATH_MAX, GFP_KERNEL); |
05dbe683 OS |
1259 | if (!name) { |
1260 | ret = -ENOMEM; | |
1261 | goto err; | |
1262 | } | |
1263 | ptr = name + PATH_MAX - 1; | |
1264 | ptr[0] = '\0'; | |
73f73415 JB |
1265 | |
1266 | /* | |
05dbe683 OS |
1267 | * Walk up the subvolume trees in the tree of tree roots by root |
1268 | * backrefs until we hit the top-level subvolume. | |
73f73415 | 1269 | */ |
05dbe683 OS |
1270 | while (subvol_objectid != BTRFS_FS_TREE_OBJECTID) { |
1271 | key.objectid = subvol_objectid; | |
1272 | key.type = BTRFS_ROOT_BACKREF_KEY; | |
1273 | key.offset = (u64)-1; | |
1274 | ||
0ff40a91 | 1275 | ret = btrfs_search_backwards(root, &key, path); |
05dbe683 OS |
1276 | if (ret < 0) { |
1277 | goto err; | |
1278 | } else if (ret > 0) { | |
0ff40a91 MPS |
1279 | ret = -ENOENT; |
1280 | goto err; | |
05dbe683 OS |
1281 | } |
1282 | ||
05dbe683 OS |
1283 | subvol_objectid = key.offset; |
1284 | ||
1285 | root_ref = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
1286 | struct btrfs_root_ref); | |
1287 | len = btrfs_root_ref_name_len(path->nodes[0], root_ref); | |
1288 | ptr -= len + 1; | |
1289 | if (ptr < name) { | |
1290 | ret = -ENAMETOOLONG; | |
1291 | goto err; | |
1292 | } | |
1293 | read_extent_buffer(path->nodes[0], ptr + 1, | |
1294 | (unsigned long)(root_ref + 1), len); | |
1295 | ptr[0] = '/'; | |
1296 | dirid = btrfs_root_ref_dirid(path->nodes[0], root_ref); | |
1297 | btrfs_release_path(path); | |
1298 | ||
56e9357a | 1299 | fs_root = btrfs_get_fs_root(fs_info, subvol_objectid, true); |
05dbe683 OS |
1300 | if (IS_ERR(fs_root)) { |
1301 | ret = PTR_ERR(fs_root); | |
5168489a JB |
1302 | fs_root = NULL; |
1303 | goto err; | |
1304 | } | |
05dbe683 OS |
1305 | |
1306 | /* | |
1307 | * Walk up the filesystem tree by inode refs until we hit the | |
1308 | * root directory. | |
1309 | */ | |
1310 | while (dirid != BTRFS_FIRST_FREE_OBJECTID) { | |
1311 | key.objectid = dirid; | |
1312 | key.type = BTRFS_INODE_REF_KEY; | |
1313 | key.offset = (u64)-1; | |
1314 | ||
0ff40a91 | 1315 | ret = btrfs_search_backwards(fs_root, &key, path); |
05dbe683 OS |
1316 | if (ret < 0) { |
1317 | goto err; | |
1318 | } else if (ret > 0) { | |
0ff40a91 MPS |
1319 | ret = -ENOENT; |
1320 | goto err; | |
05dbe683 OS |
1321 | } |
1322 | ||
05dbe683 OS |
1323 | dirid = key.offset; |
1324 | ||
1325 | inode_ref = btrfs_item_ptr(path->nodes[0], | |
1326 | path->slots[0], | |
1327 | struct btrfs_inode_ref); | |
1328 | len = btrfs_inode_ref_name_len(path->nodes[0], | |
1329 | inode_ref); | |
1330 | ptr -= len + 1; | |
1331 | if (ptr < name) { | |
1332 | ret = -ENAMETOOLONG; | |
1333 | goto err; | |
1334 | } | |
1335 | read_extent_buffer(path->nodes[0], ptr + 1, | |
1336 | (unsigned long)(inode_ref + 1), len); | |
1337 | ptr[0] = '/'; | |
1338 | btrfs_release_path(path); | |
1339 | } | |
00246528 | 1340 | btrfs_put_root(fs_root); |
5168489a | 1341 | fs_root = NULL; |
73f73415 JB |
1342 | } |
1343 | ||
05dbe683 OS |
1344 | btrfs_free_path(path); |
1345 | if (ptr == name + PATH_MAX - 1) { | |
1346 | name[0] = '/'; | |
1347 | name[1] = '\0'; | |
1348 | } else { | |
1349 | memmove(name, ptr, name + PATH_MAX - ptr); | |
1350 | } | |
1351 | return name; | |
1352 | ||
1353 | err: | |
00246528 | 1354 | btrfs_put_root(fs_root); |
05dbe683 OS |
1355 | btrfs_free_path(path); |
1356 | kfree(name); | |
1357 | return ERR_PTR(ret); | |
1358 | } | |
1359 | ||
1360 | static int get_default_subvol_objectid(struct btrfs_fs_info *fs_info, u64 *objectid) | |
1361 | { | |
1362 | struct btrfs_root *root = fs_info->tree_root; | |
1363 | struct btrfs_dir_item *di; | |
1364 | struct btrfs_path *path; | |
1365 | struct btrfs_key location; | |
1366 | u64 dir_id; | |
1367 | ||
73f73415 JB |
1368 | path = btrfs_alloc_path(); |
1369 | if (!path) | |
05dbe683 | 1370 | return -ENOMEM; |
73f73415 JB |
1371 | |
1372 | /* | |
1373 | * Find the "default" dir item which points to the root item that we | |
1374 | * will mount by default if we haven't been given a specific subvolume | |
1375 | * to mount. | |
1376 | */ | |
815745cf | 1377 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
73f73415 | 1378 | di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0); |
b0839166 JL |
1379 | if (IS_ERR(di)) { |
1380 | btrfs_free_path(path); | |
05dbe683 | 1381 | return PTR_ERR(di); |
b0839166 | 1382 | } |
73f73415 JB |
1383 | if (!di) { |
1384 | /* | |
1385 | * Ok the default dir item isn't there. This is weird since | |
1386 | * it's always been there, but don't freak out, just try and | |
05dbe683 | 1387 | * mount the top-level subvolume. |
73f73415 JB |
1388 | */ |
1389 | btrfs_free_path(path); | |
05dbe683 OS |
1390 | *objectid = BTRFS_FS_TREE_OBJECTID; |
1391 | return 0; | |
73f73415 JB |
1392 | } |
1393 | ||
1394 | btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); | |
1395 | btrfs_free_path(path); | |
05dbe683 OS |
1396 | *objectid = location.objectid; |
1397 | return 0; | |
73f73415 JB |
1398 | } |
1399 | ||
d397712b | 1400 | static int btrfs_fill_super(struct super_block *sb, |
8a4b83cc | 1401 | struct btrfs_fs_devices *fs_devices, |
56e033a7 | 1402 | void *data) |
75dfe396 | 1403 | { |
d397712b | 1404 | struct inode *inode; |
815745cf | 1405 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
39279cc3 | 1406 | int err; |
a429e513 | 1407 | |
39279cc3 CM |
1408 | sb->s_maxbytes = MAX_LFS_FILESIZE; |
1409 | sb->s_magic = BTRFS_SUPER_MAGIC; | |
1410 | sb->s_op = &btrfs_super_ops; | |
af53d29a | 1411 | sb->s_d_op = &btrfs_dentry_operations; |
be6e8dc0 | 1412 | sb->s_export_op = &btrfs_export_ops; |
14605409 BB |
1413 | #ifdef CONFIG_FS_VERITY |
1414 | sb->s_vop = &btrfs_verityops; | |
1415 | #endif | |
5103e947 | 1416 | sb->s_xattr = btrfs_xattr_handlers; |
39279cc3 | 1417 | sb->s_time_gran = 1; |
0eda294d | 1418 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
1751e8a6 | 1419 | sb->s_flags |= SB_POSIXACL; |
49cf6f45 | 1420 | #endif |
357fdad0 | 1421 | sb->s_flags |= SB_I_VERSION; |
da2f0f74 | 1422 | sb->s_iflags |= SB_I_CGROUPWB; |
9e11ceee JK |
1423 | |
1424 | err = super_setup_bdi(sb); | |
1425 | if (err) { | |
1426 | btrfs_err(fs_info, "super_setup_bdi failed"); | |
1427 | return err; | |
1428 | } | |
1429 | ||
ad2b2c80 AV |
1430 | err = open_ctree(sb, fs_devices, (char *)data); |
1431 | if (err) { | |
ab8d0fc4 | 1432 | btrfs_err(fs_info, "open_ctree failed"); |
ad2b2c80 | 1433 | return err; |
a429e513 CM |
1434 | } |
1435 | ||
0202e83f | 1436 | inode = btrfs_iget(sb, BTRFS_FIRST_FREE_OBJECTID, fs_info->fs_root); |
5d4f98a2 YZ |
1437 | if (IS_ERR(inode)) { |
1438 | err = PTR_ERR(inode); | |
39279cc3 | 1439 | goto fail_close; |
f254e52c | 1440 | } |
f254e52c | 1441 | |
48fde701 AV |
1442 | sb->s_root = d_make_root(inode); |
1443 | if (!sb->s_root) { | |
39279cc3 CM |
1444 | err = -ENOMEM; |
1445 | goto fail_close; | |
f254e52c | 1446 | } |
58176a96 | 1447 | |
1751e8a6 | 1448 | sb->s_flags |= SB_ACTIVE; |
2619ba1f | 1449 | return 0; |
39279cc3 CM |
1450 | |
1451 | fail_close: | |
6bccf3ab | 1452 | close_ctree(fs_info); |
39279cc3 | 1453 | return err; |
2619ba1f CM |
1454 | } |
1455 | ||
6bf13c0c | 1456 | int btrfs_sync_fs(struct super_block *sb, int wait) |
c5739bba CM |
1457 | { |
1458 | struct btrfs_trans_handle *trans; | |
815745cf AV |
1459 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
1460 | struct btrfs_root *root = fs_info->tree_root; | |
2619ba1f | 1461 | |
bc074524 | 1462 | trace_btrfs_sync_fs(fs_info, wait); |
1abe9b8a | 1463 | |
39279cc3 | 1464 | if (!wait) { |
815745cf | 1465 | filemap_flush(fs_info->btree_inode->i_mapping); |
39279cc3 CM |
1466 | return 0; |
1467 | } | |
771ed689 | 1468 | |
6374e57a | 1469 | btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); |
771ed689 | 1470 | |
d4edf39b | 1471 | trans = btrfs_attach_transaction_barrier(root); |
60376ce4 | 1472 | if (IS_ERR(trans)) { |
354aa0fb | 1473 | /* no transaction, don't bother */ |
6b5fe46d DS |
1474 | if (PTR_ERR(trans) == -ENOENT) { |
1475 | /* | |
1476 | * Exit unless we have some pending changes | |
1477 | * that need to go through commit | |
1478 | */ | |
1479 | if (fs_info->pending_changes == 0) | |
1480 | return 0; | |
a53f4f8e QW |
1481 | /* |
1482 | * A non-blocking test if the fs is frozen. We must not | |
1483 | * start a new transaction here otherwise a deadlock | |
1484 | * happens. The pending operations are delayed to the | |
1485 | * next commit after thawing. | |
1486 | */ | |
a7e3c5f2 RP |
1487 | if (sb_start_write_trylock(sb)) |
1488 | sb_end_write(sb); | |
a53f4f8e QW |
1489 | else |
1490 | return 0; | |
6b5fe46d | 1491 | trans = btrfs_start_transaction(root, 0); |
6b5fe46d | 1492 | } |
98bd5c54 DS |
1493 | if (IS_ERR(trans)) |
1494 | return PTR_ERR(trans); | |
60376ce4 | 1495 | } |
3a45bb20 | 1496 | return btrfs_commit_transaction(trans); |
2c90e5d6 CM |
1497 | } |
1498 | ||
ab0b4a3e JB |
1499 | static void print_rescue_option(struct seq_file *seq, const char *s, bool *printed) |
1500 | { | |
1501 | seq_printf(seq, "%s%s", (*printed) ? ":" : ",rescue=", s); | |
1502 | *printed = true; | |
1503 | } | |
1504 | ||
34c80b1d | 1505 | static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry) |
a9572a15 | 1506 | { |
815745cf | 1507 | struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb); |
0f628c63 | 1508 | const char *compress_type; |
3ef3959b | 1509 | const char *subvol_name; |
ab0b4a3e | 1510 | bool printed = false; |
a9572a15 | 1511 | |
3cdde224 | 1512 | if (btrfs_test_opt(info, DEGRADED)) |
a9572a15 | 1513 | seq_puts(seq, ",degraded"); |
3cdde224 | 1514 | if (btrfs_test_opt(info, NODATASUM)) |
a9572a15 | 1515 | seq_puts(seq, ",nodatasum"); |
3cdde224 | 1516 | if (btrfs_test_opt(info, NODATACOW)) |
a9572a15 | 1517 | seq_puts(seq, ",nodatacow"); |
3cdde224 | 1518 | if (btrfs_test_opt(info, NOBARRIER)) |
a9572a15 | 1519 | seq_puts(seq, ",nobarrier"); |
95ac567a | 1520 | if (info->max_inline != BTRFS_DEFAULT_MAX_INLINE) |
c1c9ff7c | 1521 | seq_printf(seq, ",max_inline=%llu", info->max_inline); |
a9572a15 EP |
1522 | if (info->thread_pool_size != min_t(unsigned long, |
1523 | num_online_cpus() + 2, 8)) | |
f7b885be | 1524 | seq_printf(seq, ",thread_pool=%u", info->thread_pool_size); |
3cdde224 | 1525 | if (btrfs_test_opt(info, COMPRESS)) { |
0f628c63 | 1526 | compress_type = btrfs_compress_type2str(info->compress_type); |
3cdde224 | 1527 | if (btrfs_test_opt(info, FORCE_COMPRESS)) |
200da64e TI |
1528 | seq_printf(seq, ",compress-force=%s", compress_type); |
1529 | else | |
1530 | seq_printf(seq, ",compress=%s", compress_type); | |
f51d2b59 | 1531 | if (info->compress_level) |
fa4d885a | 1532 | seq_printf(seq, ":%d", info->compress_level); |
200da64e | 1533 | } |
3cdde224 | 1534 | if (btrfs_test_opt(info, NOSSD)) |
c289811c | 1535 | seq_puts(seq, ",nossd"); |
3cdde224 | 1536 | if (btrfs_test_opt(info, SSD_SPREAD)) |
451d7585 | 1537 | seq_puts(seq, ",ssd_spread"); |
3cdde224 | 1538 | else if (btrfs_test_opt(info, SSD)) |
a9572a15 | 1539 | seq_puts(seq, ",ssd"); |
3cdde224 | 1540 | if (btrfs_test_opt(info, NOTREELOG)) |
6b65c5c6 | 1541 | seq_puts(seq, ",notreelog"); |
3cdde224 | 1542 | if (btrfs_test_opt(info, NOLOGREPLAY)) |
ab0b4a3e | 1543 | print_rescue_option(seq, "nologreplay", &printed); |
68319c18 JB |
1544 | if (btrfs_test_opt(info, USEBACKUPROOT)) |
1545 | print_rescue_option(seq, "usebackuproot", &printed); | |
42437a63 JB |
1546 | if (btrfs_test_opt(info, IGNOREBADROOTS)) |
1547 | print_rescue_option(seq, "ignorebadroots", &printed); | |
882dbe0c JB |
1548 | if (btrfs_test_opt(info, IGNOREDATACSUMS)) |
1549 | print_rescue_option(seq, "ignoredatacsums", &printed); | |
3cdde224 | 1550 | if (btrfs_test_opt(info, FLUSHONCOMMIT)) |
6b65c5c6 | 1551 | seq_puts(seq, ",flushoncommit"); |
46b27f50 | 1552 | if (btrfs_test_opt(info, DISCARD_SYNC)) |
20a5239a | 1553 | seq_puts(seq, ",discard"); |
b0643e59 DZ |
1554 | if (btrfs_test_opt(info, DISCARD_ASYNC)) |
1555 | seq_puts(seq, ",discard=async"); | |
1751e8a6 | 1556 | if (!(info->sb->s_flags & SB_POSIXACL)) |
a9572a15 | 1557 | seq_puts(seq, ",noacl"); |
04c41559 | 1558 | if (btrfs_free_space_cache_v1_active(info)) |
200da64e | 1559 | seq_puts(seq, ",space_cache"); |
04c41559 | 1560 | else if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE)) |
70f6d82e | 1561 | seq_puts(seq, ",space_cache=v2"); |
73bc1876 | 1562 | else |
8965593e | 1563 | seq_puts(seq, ",nospace_cache"); |
3cdde224 | 1564 | if (btrfs_test_opt(info, RESCAN_UUID_TREE)) |
f420ee1e | 1565 | seq_puts(seq, ",rescan_uuid_tree"); |
3cdde224 | 1566 | if (btrfs_test_opt(info, CLEAR_CACHE)) |
200da64e | 1567 | seq_puts(seq, ",clear_cache"); |
3cdde224 | 1568 | if (btrfs_test_opt(info, USER_SUBVOL_RM_ALLOWED)) |
200da64e | 1569 | seq_puts(seq, ",user_subvol_rm_allowed"); |
3cdde224 | 1570 | if (btrfs_test_opt(info, ENOSPC_DEBUG)) |
0942caa3 | 1571 | seq_puts(seq, ",enospc_debug"); |
3cdde224 | 1572 | if (btrfs_test_opt(info, AUTO_DEFRAG)) |
0942caa3 | 1573 | seq_puts(seq, ",autodefrag"); |
3cdde224 | 1574 | if (btrfs_test_opt(info, SKIP_BALANCE)) |
9555c6c1 | 1575 | seq_puts(seq, ",skip_balance"); |
8507d216 | 1576 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
cbeaae4f | 1577 | if (btrfs_test_opt(info, CHECK_INTEGRITY_DATA)) |
8507d216 | 1578 | seq_puts(seq, ",check_int_data"); |
3cdde224 | 1579 | else if (btrfs_test_opt(info, CHECK_INTEGRITY)) |
8507d216 WS |
1580 | seq_puts(seq, ",check_int"); |
1581 | if (info->check_integrity_print_mask) | |
1582 | seq_printf(seq, ",check_int_print_mask=%d", | |
1583 | info->check_integrity_print_mask); | |
1584 | #endif | |
1585 | if (info->metadata_ratio) | |
764cb8b4 | 1586 | seq_printf(seq, ",metadata_ratio=%u", info->metadata_ratio); |
3cdde224 | 1587 | if (btrfs_test_opt(info, PANIC_ON_FATAL_ERROR)) |
8c342930 | 1588 | seq_puts(seq, ",fatal_errors=panic"); |
8b87dc17 | 1589 | if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL) |
d3740608 | 1590 | seq_printf(seq, ",commit=%u", info->commit_interval); |
d0bd4560 | 1591 | #ifdef CONFIG_BTRFS_DEBUG |
3cdde224 | 1592 | if (btrfs_test_opt(info, FRAGMENT_DATA)) |
d0bd4560 | 1593 | seq_puts(seq, ",fragment=data"); |
3cdde224 | 1594 | if (btrfs_test_opt(info, FRAGMENT_METADATA)) |
d0bd4560 JB |
1595 | seq_puts(seq, ",fragment=metadata"); |
1596 | #endif | |
fb592373 JB |
1597 | if (btrfs_test_opt(info, REF_VERIFY)) |
1598 | seq_puts(seq, ",ref_verify"); | |
c8d3fe02 OS |
1599 | seq_printf(seq, ",subvolid=%llu", |
1600 | BTRFS_I(d_inode(dentry))->root->root_key.objectid); | |
3ef3959b JB |
1601 | subvol_name = btrfs_get_subvol_name_from_objectid(info, |
1602 | BTRFS_I(d_inode(dentry))->root->root_key.objectid); | |
1603 | if (!IS_ERR(subvol_name)) { | |
1604 | seq_puts(seq, ",subvol="); | |
1605 | seq_escape(seq, subvol_name, " \t\n\\"); | |
1606 | kfree(subvol_name); | |
1607 | } | |
a9572a15 EP |
1608 | return 0; |
1609 | } | |
1610 | ||
a061fc8d | 1611 | static int btrfs_test_super(struct super_block *s, void *data) |
4b82d6e4 | 1612 | { |
815745cf AV |
1613 | struct btrfs_fs_info *p = data; |
1614 | struct btrfs_fs_info *fs_info = btrfs_sb(s); | |
4b82d6e4 | 1615 | |
815745cf | 1616 | return fs_info->fs_devices == p->fs_devices; |
4b82d6e4 Y |
1617 | } |
1618 | ||
450ba0ea JB |
1619 | static int btrfs_set_super(struct super_block *s, void *data) |
1620 | { | |
6de1d09d AV |
1621 | int err = set_anon_super(s, data); |
1622 | if (!err) | |
1623 | s->s_fs_info = data; | |
1624 | return err; | |
4b82d6e4 Y |
1625 | } |
1626 | ||
f9d9ef62 DS |
1627 | /* |
1628 | * subvolumes are identified by ino 256 | |
1629 | */ | |
1630 | static inline int is_subvolume_inode(struct inode *inode) | |
1631 | { | |
1632 | if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) | |
1633 | return 1; | |
1634 | return 0; | |
1635 | } | |
1636 | ||
bb289b7b | 1637 | static struct dentry *mount_subvol(const char *subvol_name, u64 subvol_objectid, |
ae0bc863 | 1638 | struct vfsmount *mnt) |
830c4adb | 1639 | { |
830c4adb | 1640 | struct dentry *root; |
fa330659 | 1641 | int ret; |
830c4adb | 1642 | |
05dbe683 OS |
1643 | if (!subvol_name) { |
1644 | if (!subvol_objectid) { | |
1645 | ret = get_default_subvol_objectid(btrfs_sb(mnt->mnt_sb), | |
1646 | &subvol_objectid); | |
1647 | if (ret) { | |
1648 | root = ERR_PTR(ret); | |
1649 | goto out; | |
1650 | } | |
1651 | } | |
c0c907a4 MPS |
1652 | subvol_name = btrfs_get_subvol_name_from_objectid( |
1653 | btrfs_sb(mnt->mnt_sb), subvol_objectid); | |
05dbe683 OS |
1654 | if (IS_ERR(subvol_name)) { |
1655 | root = ERR_CAST(subvol_name); | |
1656 | subvol_name = NULL; | |
1657 | goto out; | |
1658 | } | |
1659 | ||
1660 | } | |
1661 | ||
ea441d11 | 1662 | root = mount_subtree(mnt, subvol_name); |
fa330659 OS |
1663 | /* mount_subtree() drops our reference on the vfsmount. */ |
1664 | mnt = NULL; | |
830c4adb | 1665 | |
bb289b7b | 1666 | if (!IS_ERR(root)) { |
ea441d11 | 1667 | struct super_block *s = root->d_sb; |
ab8d0fc4 | 1668 | struct btrfs_fs_info *fs_info = btrfs_sb(s); |
bb289b7b OS |
1669 | struct inode *root_inode = d_inode(root); |
1670 | u64 root_objectid = BTRFS_I(root_inode)->root->root_key.objectid; | |
1671 | ||
1672 | ret = 0; | |
1673 | if (!is_subvolume_inode(root_inode)) { | |
ab8d0fc4 | 1674 | btrfs_err(fs_info, "'%s' is not a valid subvolume", |
bb289b7b OS |
1675 | subvol_name); |
1676 | ret = -EINVAL; | |
1677 | } | |
1678 | if (subvol_objectid && root_objectid != subvol_objectid) { | |
05dbe683 OS |
1679 | /* |
1680 | * This will also catch a race condition where a | |
1681 | * subvolume which was passed by ID is renamed and | |
1682 | * another subvolume is renamed over the old location. | |
1683 | */ | |
ab8d0fc4 JM |
1684 | btrfs_err(fs_info, |
1685 | "subvol '%s' does not match subvolid %llu", | |
1686 | subvol_name, subvol_objectid); | |
bb289b7b OS |
1687 | ret = -EINVAL; |
1688 | } | |
1689 | if (ret) { | |
1690 | dput(root); | |
1691 | root = ERR_PTR(ret); | |
1692 | deactivate_locked_super(s); | |
1693 | } | |
f9d9ef62 DS |
1694 | } |
1695 | ||
fa330659 OS |
1696 | out: |
1697 | mntput(mnt); | |
fa330659 | 1698 | kfree(subvol_name); |
830c4adb JB |
1699 | return root; |
1700 | } | |
450ba0ea | 1701 | |
312c89fb MT |
1702 | /* |
1703 | * Find a superblock for the given device / mount point. | |
1704 | * | |
1705 | * Note: This is based on mount_bdev from fs/super.c with a few additions | |
1706 | * for multiple device setup. Make sure to keep it in sync. | |
1707 | */ | |
72fa39f5 MT |
1708 | static struct dentry *btrfs_mount_root(struct file_system_type *fs_type, |
1709 | int flags, const char *device_name, void *data) | |
1710 | { | |
1711 | struct block_device *bdev = NULL; | |
1712 | struct super_block *s; | |
36350e95 | 1713 | struct btrfs_device *device = NULL; |
72fa39f5 MT |
1714 | struct btrfs_fs_devices *fs_devices = NULL; |
1715 | struct btrfs_fs_info *fs_info = NULL; | |
204cc0cc | 1716 | void *new_sec_opts = NULL; |
72fa39f5 | 1717 | fmode_t mode = FMODE_READ; |
72fa39f5 MT |
1718 | int error = 0; |
1719 | ||
1720 | if (!(flags & SB_RDONLY)) | |
1721 | mode |= FMODE_WRITE; | |
1722 | ||
72fa39f5 | 1723 | if (data) { |
a65001e8 | 1724 | error = security_sb_eat_lsm_opts(data, &new_sec_opts); |
72fa39f5 MT |
1725 | if (error) |
1726 | return ERR_PTR(error); | |
1727 | } | |
1728 | ||
72fa39f5 MT |
1729 | /* |
1730 | * Setup a dummy root and fs_info for test/set super. This is because | |
1731 | * we don't actually fill this stuff out until open_ctree, but we need | |
8260edba JB |
1732 | * then open_ctree will properly initialize the file system specific |
1733 | * settings later. btrfs_init_fs_info initializes the static elements | |
1734 | * of the fs_info (locks and such) to make cleanup easier if we find a | |
1735 | * superblock with our given fs_devices later on at sget() time. | |
72fa39f5 | 1736 | */ |
a8fd1f71 | 1737 | fs_info = kvzalloc(sizeof(struct btrfs_fs_info), GFP_KERNEL); |
72fa39f5 MT |
1738 | if (!fs_info) { |
1739 | error = -ENOMEM; | |
1740 | goto error_sec_opts; | |
1741 | } | |
8260edba | 1742 | btrfs_init_fs_info(fs_info); |
72fa39f5 | 1743 | |
72fa39f5 MT |
1744 | fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); |
1745 | fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); | |
72fa39f5 MT |
1746 | if (!fs_info->super_copy || !fs_info->super_for_commit) { |
1747 | error = -ENOMEM; | |
1748 | goto error_fs_info; | |
1749 | } | |
1750 | ||
399f7f4c | 1751 | mutex_lock(&uuid_mutex); |
fa59f27c | 1752 | error = btrfs_parse_device_options(data, mode, fs_type); |
81ffd56b DS |
1753 | if (error) { |
1754 | mutex_unlock(&uuid_mutex); | |
399f7f4c | 1755 | goto error_fs_info; |
81ffd56b | 1756 | } |
399f7f4c | 1757 | |
36350e95 GJ |
1758 | device = btrfs_scan_one_device(device_name, mode, fs_type); |
1759 | if (IS_ERR(device)) { | |
81ffd56b | 1760 | mutex_unlock(&uuid_mutex); |
36350e95 | 1761 | error = PTR_ERR(device); |
399f7f4c | 1762 | goto error_fs_info; |
81ffd56b | 1763 | } |
399f7f4c | 1764 | |
36350e95 | 1765 | fs_devices = device->fs_devices; |
399f7f4c DS |
1766 | fs_info->fs_devices = fs_devices; |
1767 | ||
72fa39f5 | 1768 | error = btrfs_open_devices(fs_devices, mode, fs_type); |
f5194e34 | 1769 | mutex_unlock(&uuid_mutex); |
72fa39f5 MT |
1770 | if (error) |
1771 | goto error_fs_info; | |
1772 | ||
1773 | if (!(flags & SB_RDONLY) && fs_devices->rw_devices == 0) { | |
1774 | error = -EACCES; | |
1775 | goto error_close_devices; | |
1776 | } | |
1777 | ||
d24fa5c1 | 1778 | bdev = fs_devices->latest_dev->bdev; |
72fa39f5 MT |
1779 | s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | SB_NOSEC, |
1780 | fs_info); | |
1781 | if (IS_ERR(s)) { | |
1782 | error = PTR_ERR(s); | |
1783 | goto error_close_devices; | |
1784 | } | |
1785 | ||
1786 | if (s->s_root) { | |
1787 | btrfs_close_devices(fs_devices); | |
0d4b0463 | 1788 | btrfs_free_fs_info(fs_info); |
72fa39f5 MT |
1789 | if ((flags ^ s->s_flags) & SB_RDONLY) |
1790 | error = -EBUSY; | |
1791 | } else { | |
1792 | snprintf(s->s_id, sizeof(s->s_id), "%pg", bdev); | |
1793 | btrfs_sb(s)->bdev_holder = fs_type; | |
9b4e675a DS |
1794 | if (!strstr(crc32c_impl(), "generic")) |
1795 | set_bit(BTRFS_FS_CSUM_IMPL_FAST, &fs_info->flags); | |
72fa39f5 MT |
1796 | error = btrfs_fill_super(s, fs_devices, data); |
1797 | } | |
a65001e8 | 1798 | if (!error) |
204cc0cc | 1799 | error = security_sb_set_mnt_opts(s, new_sec_opts, 0, NULL); |
a65001e8 | 1800 | security_free_mnt_opts(&new_sec_opts); |
72fa39f5 MT |
1801 | if (error) { |
1802 | deactivate_locked_super(s); | |
a65001e8 | 1803 | return ERR_PTR(error); |
72fa39f5 MT |
1804 | } |
1805 | ||
1806 | return dget(s->s_root); | |
1807 | ||
1808 | error_close_devices: | |
1809 | btrfs_close_devices(fs_devices); | |
1810 | error_fs_info: | |
0d4b0463 | 1811 | btrfs_free_fs_info(fs_info); |
72fa39f5 MT |
1812 | error_sec_opts: |
1813 | security_free_mnt_opts(&new_sec_opts); | |
1814 | return ERR_PTR(error); | |
1815 | } | |
312c89fb | 1816 | |
edf24abe | 1817 | /* |
312c89fb | 1818 | * Mount function which is called by VFS layer. |
edf24abe | 1819 | * |
312c89fb MT |
1820 | * In order to allow mounting a subvolume directly, btrfs uses mount_subtree() |
1821 | * which needs vfsmount* of device's root (/). This means device's root has to | |
1822 | * be mounted internally in any case. | |
1823 | * | |
1824 | * Operation flow: | |
1825 | * 1. Parse subvol id related options for later use in mount_subvol(). | |
1826 | * | |
1827 | * 2. Mount device's root (/) by calling vfs_kern_mount(). | |
1828 | * | |
1829 | * NOTE: vfs_kern_mount() is used by VFS to call btrfs_mount() in the | |
1830 | * first place. In order to avoid calling btrfs_mount() again, we use | |
1831 | * different file_system_type which is not registered to VFS by | |
1832 | * register_filesystem() (btrfs_root_fs_type). As a result, | |
1833 | * btrfs_mount_root() is called. The return value will be used by | |
1834 | * mount_subtree() in mount_subvol(). | |
1835 | * | |
1836 | * 3. Call mount_subvol() to get the dentry of subvolume. Since there is | |
1837 | * "btrfs subvolume set-default", mount_subvol() is called always. | |
edf24abe | 1838 | */ |
061dbc6b | 1839 | static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, |
306e16ce | 1840 | const char *device_name, void *data) |
4b82d6e4 | 1841 | { |
312c89fb MT |
1842 | struct vfsmount *mnt_root; |
1843 | struct dentry *root; | |
73f73415 JB |
1844 | char *subvol_name = NULL; |
1845 | u64 subvol_objectid = 0; | |
4b82d6e4 Y |
1846 | int error = 0; |
1847 | ||
93b9bcdf GJ |
1848 | error = btrfs_parse_subvol_options(data, &subvol_name, |
1849 | &subvol_objectid); | |
f23c8af8 ID |
1850 | if (error) { |
1851 | kfree(subvol_name); | |
061dbc6b | 1852 | return ERR_PTR(error); |
f23c8af8 | 1853 | } |
edf24abe | 1854 | |
312c89fb MT |
1855 | /* mount device's root (/) */ |
1856 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, flags, device_name, data); | |
1857 | if (PTR_ERR_OR_ZERO(mnt_root) == -EBUSY) { | |
1858 | if (flags & SB_RDONLY) { | |
1859 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, | |
1860 | flags & ~SB_RDONLY, device_name, data); | |
1861 | } else { | |
1862 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, | |
1863 | flags | SB_RDONLY, device_name, data); | |
1864 | if (IS_ERR(mnt_root)) { | |
1865 | root = ERR_CAST(mnt_root); | |
532b618b | 1866 | kfree(subvol_name); |
312c89fb MT |
1867 | goto out; |
1868 | } | |
4b82d6e4 | 1869 | |
312c89fb MT |
1870 | down_write(&mnt_root->mnt_sb->s_umount); |
1871 | error = btrfs_remount(mnt_root->mnt_sb, &flags, NULL); | |
1872 | up_write(&mnt_root->mnt_sb->s_umount); | |
1873 | if (error < 0) { | |
1874 | root = ERR_PTR(error); | |
1875 | mntput(mnt_root); | |
532b618b | 1876 | kfree(subvol_name); |
312c89fb MT |
1877 | goto out; |
1878 | } | |
1879 | } | |
f667aef6 | 1880 | } |
312c89fb MT |
1881 | if (IS_ERR(mnt_root)) { |
1882 | root = ERR_CAST(mnt_root); | |
532b618b | 1883 | kfree(subvol_name); |
312c89fb | 1884 | goto out; |
f667aef6 | 1885 | } |
4b82d6e4 | 1886 | |
312c89fb | 1887 | /* mount_subvol() will free subvol_name and mnt_root */ |
ae0bc863 | 1888 | root = mount_subvol(subvol_name, subvol_objectid, mnt_root); |
4b82d6e4 | 1889 | |
312c89fb MT |
1890 | out: |
1891 | return root; | |
4b82d6e4 | 1892 | } |
2e635a27 | 1893 | |
0d2450ab | 1894 | static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info, |
f7b885be | 1895 | u32 new_pool_size, u32 old_pool_size) |
0d2450ab ST |
1896 | { |
1897 | if (new_pool_size == old_pool_size) | |
1898 | return; | |
1899 | ||
1900 | fs_info->thread_pool_size = new_pool_size; | |
1901 | ||
efe120a0 | 1902 | btrfs_info(fs_info, "resize thread pool %d -> %d", |
0d2450ab ST |
1903 | old_pool_size, new_pool_size); |
1904 | ||
5cdc7ad3 | 1905 | btrfs_workqueue_set_max(fs_info->workers, new_pool_size); |
a31b4a43 | 1906 | btrfs_workqueue_set_max(fs_info->hipri_workers, new_pool_size); |
afe3d242 | 1907 | btrfs_workqueue_set_max(fs_info->delalloc_workers, new_pool_size); |
e66f0bb1 | 1908 | btrfs_workqueue_set_max(fs_info->caching_workers, new_pool_size); |
fccb5d86 QW |
1909 | btrfs_workqueue_set_max(fs_info->endio_workers, new_pool_size); |
1910 | btrfs_workqueue_set_max(fs_info->endio_meta_workers, new_pool_size); | |
1911 | btrfs_workqueue_set_max(fs_info->endio_meta_write_workers, | |
1912 | new_pool_size); | |
1913 | btrfs_workqueue_set_max(fs_info->endio_write_workers, new_pool_size); | |
1914 | btrfs_workqueue_set_max(fs_info->endio_freespace_worker, new_pool_size); | |
5b3bc44e | 1915 | btrfs_workqueue_set_max(fs_info->delayed_workers, new_pool_size); |
0d2450ab ST |
1916 | } |
1917 | ||
f42a34b2 MX |
1918 | static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info, |
1919 | unsigned long old_opts, int flags) | |
1920 | { | |
dc81cdc5 MX |
1921 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && |
1922 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || | |
1751e8a6 | 1923 | (flags & SB_RDONLY))) { |
dc81cdc5 MX |
1924 | /* wait for any defraggers to finish */ |
1925 | wait_event(fs_info->transaction_wait, | |
1926 | (atomic_read(&fs_info->defrag_running) == 0)); | |
1751e8a6 | 1927 | if (flags & SB_RDONLY) |
dc81cdc5 MX |
1928 | sync_filesystem(fs_info->sb); |
1929 | } | |
1930 | } | |
1931 | ||
1932 | static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info, | |
1933 | unsigned long old_opts) | |
1934 | { | |
94846229 BB |
1935 | const bool cache_opt = btrfs_test_opt(fs_info, SPACE_CACHE); |
1936 | ||
dc81cdc5 | 1937 | /* |
180e4d47 LB |
1938 | * We need to cleanup all defragable inodes if the autodefragment is |
1939 | * close or the filesystem is read only. | |
dc81cdc5 MX |
1940 | */ |
1941 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && | |
bc98a42c | 1942 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || sb_rdonly(fs_info->sb))) { |
dc81cdc5 MX |
1943 | btrfs_cleanup_defrag_inodes(fs_info); |
1944 | } | |
1945 | ||
b0643e59 DZ |
1946 | /* If we toggled discard async */ |
1947 | if (!btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) && | |
1948 | btrfs_test_opt(fs_info, DISCARD_ASYNC)) | |
1949 | btrfs_discard_resume(fs_info); | |
1950 | else if (btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) && | |
1951 | !btrfs_test_opt(fs_info, DISCARD_ASYNC)) | |
1952 | btrfs_discard_cleanup(fs_info); | |
94846229 BB |
1953 | |
1954 | /* If we toggled space cache */ | |
1955 | if (cache_opt != btrfs_free_space_cache_v1_active(fs_info)) | |
1956 | btrfs_set_free_space_cache_v1_active(fs_info, cache_opt); | |
dc81cdc5 MX |
1957 | } |
1958 | ||
c146afad YZ |
1959 | static int btrfs_remount(struct super_block *sb, int *flags, char *data) |
1960 | { | |
815745cf | 1961 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
49b25e05 JM |
1962 | unsigned old_flags = sb->s_flags; |
1963 | unsigned long old_opts = fs_info->mount_opt; | |
1964 | unsigned long old_compress_type = fs_info->compress_type; | |
1965 | u64 old_max_inline = fs_info->max_inline; | |
f7b885be | 1966 | u32 old_thread_pool_size = fs_info->thread_pool_size; |
d612ac59 | 1967 | u32 old_metadata_ratio = fs_info->metadata_ratio; |
c146afad YZ |
1968 | int ret; |
1969 | ||
02b9984d | 1970 | sync_filesystem(sb); |
88c4703f | 1971 | set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); |
dc81cdc5 | 1972 | |
f667aef6 | 1973 | if (data) { |
204cc0cc | 1974 | void *new_sec_opts = NULL; |
f667aef6 | 1975 | |
a65001e8 AV |
1976 | ret = security_sb_eat_lsm_opts(data, &new_sec_opts); |
1977 | if (!ret) | |
204cc0cc | 1978 | ret = security_sb_remount(sb, new_sec_opts); |
a65001e8 | 1979 | security_free_mnt_opts(&new_sec_opts); |
f667aef6 QW |
1980 | if (ret) |
1981 | goto restore; | |
f667aef6 QW |
1982 | } |
1983 | ||
2ff7e61e | 1984 | ret = btrfs_parse_options(fs_info, data, *flags); |
891f41cb | 1985 | if (ret) |
49b25e05 | 1986 | goto restore; |
b288052e | 1987 | |
0591f040 QW |
1988 | /* V1 cache is not supported for subpage mount. */ |
1989 | if (fs_info->sectorsize < PAGE_SIZE && btrfs_test_opt(fs_info, SPACE_CACHE)) { | |
1990 | btrfs_warn(fs_info, | |
1991 | "v1 space cache is not supported for page size %lu with sectorsize %u", | |
1992 | PAGE_SIZE, fs_info->sectorsize); | |
1993 | ret = -EINVAL; | |
1994 | goto restore; | |
1995 | } | |
f42a34b2 | 1996 | btrfs_remount_begin(fs_info, old_opts, *flags); |
0d2450ab ST |
1997 | btrfs_resize_thread_pool(fs_info, |
1998 | fs_info->thread_pool_size, old_thread_pool_size); | |
1999 | ||
c55a4319 BB |
2000 | if ((bool)btrfs_test_opt(fs_info, FREE_SPACE_TREE) != |
2001 | (bool)btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) && | |
2838d255 BB |
2002 | (!sb_rdonly(sb) || (*flags & SB_RDONLY))) { |
2003 | btrfs_warn(fs_info, | |
2004 | "remount supports changing free space tree only from ro to rw"); | |
2005 | /* Make sure free space cache options match the state on disk */ | |
2006 | if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) { | |
2007 | btrfs_set_opt(fs_info->mount_opt, FREE_SPACE_TREE); | |
2008 | btrfs_clear_opt(fs_info->mount_opt, SPACE_CACHE); | |
2009 | } | |
2010 | if (btrfs_free_space_cache_v1_active(fs_info)) { | |
2011 | btrfs_clear_opt(fs_info->mount_opt, FREE_SPACE_TREE); | |
2012 | btrfs_set_opt(fs_info->mount_opt, SPACE_CACHE); | |
2013 | } | |
2014 | } | |
2015 | ||
1751e8a6 | 2016 | if ((bool)(*flags & SB_RDONLY) == sb_rdonly(sb)) |
dc81cdc5 | 2017 | goto out; |
c146afad | 2018 | |
1751e8a6 | 2019 | if (*flags & SB_RDONLY) { |
8dabb742 SB |
2020 | /* |
2021 | * this also happens on 'umount -rf' or on shutdown, when | |
2022 | * the filesystem is busy. | |
2023 | */ | |
21c7e756 | 2024 | cancel_work_sync(&fs_info->async_reclaim_work); |
57056740 | 2025 | cancel_work_sync(&fs_info->async_data_reclaim_work); |
361c093d | 2026 | |
b0643e59 DZ |
2027 | btrfs_discard_cleanup(fs_info); |
2028 | ||
361c093d SB |
2029 | /* wait for the uuid_scan task to finish */ |
2030 | down(&fs_info->uuid_tree_rescan_sem); | |
2031 | /* avoid complains from lockdep et al. */ | |
2032 | up(&fs_info->uuid_tree_rescan_sem); | |
2033 | ||
a0a1db70 | 2034 | btrfs_set_sb_rdonly(sb); |
c146afad | 2035 | |
e44163e1 | 2036 | /* |
1751e8a6 | 2037 | * Setting SB_RDONLY will put the cleaner thread to |
e44163e1 JM |
2038 | * sleep at the next loop if it's already active. |
2039 | * If it's already asleep, we'll leave unused block | |
2040 | * groups on disk until we're mounted read-write again | |
2041 | * unless we clean them up here. | |
2042 | */ | |
e44163e1 | 2043 | btrfs_delete_unused_bgs(fs_info); |
e44163e1 | 2044 | |
a0a1db70 FM |
2045 | /* |
2046 | * The cleaner task could be already running before we set the | |
2047 | * flag BTRFS_FS_STATE_RO (and SB_RDONLY in the superblock). | |
2048 | * We must make sure that after we finish the remount, i.e. after | |
2049 | * we call btrfs_commit_super(), the cleaner can no longer start | |
2050 | * a transaction - either because it was dropping a dead root, | |
2051 | * running delayed iputs or deleting an unused block group (the | |
2052 | * cleaner picked a block group from the list of unused block | |
2053 | * groups before we were able to in the previous call to | |
2054 | * btrfs_delete_unused_bgs()). | |
2055 | */ | |
2056 | wait_on_bit(&fs_info->flags, BTRFS_FS_CLEANER_RUNNING, | |
2057 | TASK_UNINTERRUPTIBLE); | |
2058 | ||
a8cc263e FM |
2059 | /* |
2060 | * We've set the superblock to RO mode, so we might have made | |
2061 | * the cleaner task sleep without running all pending delayed | |
2062 | * iputs. Go through all the delayed iputs here, so that if an | |
2063 | * unmount happens without remounting RW we don't end up at | |
2064 | * finishing close_ctree() with a non-empty list of delayed | |
2065 | * iputs. | |
2066 | */ | |
2067 | btrfs_run_delayed_iputs(fs_info); | |
2068 | ||
8dabb742 SB |
2069 | btrfs_dev_replace_suspend_for_unmount(fs_info); |
2070 | btrfs_scrub_cancel(fs_info); | |
061594ef | 2071 | btrfs_pause_balance(fs_info); |
8dabb742 | 2072 | |
cb13eea3 FM |
2073 | /* |
2074 | * Pause the qgroup rescan worker if it is running. We don't want | |
2075 | * it to be still running after we are in RO mode, as after that, | |
2076 | * by the time we unmount, it might have left a transaction open, | |
2077 | * so we would leak the transaction and/or crash. | |
2078 | */ | |
2079 | btrfs_qgroup_wait_for_completion(fs_info, false); | |
2080 | ||
6bccf3ab | 2081 | ret = btrfs_commit_super(fs_info); |
49b25e05 JM |
2082 | if (ret) |
2083 | goto restore; | |
c146afad | 2084 | } else { |
84961539 | 2085 | if (BTRFS_FS_ERROR(fs_info)) { |
6ef3de9c | 2086 | btrfs_err(fs_info, |
efe120a0 | 2087 | "Remounting read-write after error is not allowed"); |
6ef3de9c DS |
2088 | ret = -EINVAL; |
2089 | goto restore; | |
2090 | } | |
8a3db184 | 2091 | if (fs_info->fs_devices->rw_devices == 0) { |
49b25e05 JM |
2092 | ret = -EACCES; |
2093 | goto restore; | |
8a3db184 | 2094 | } |
2b82032c | 2095 | |
6528b99d | 2096 | if (!btrfs_check_rw_degradable(fs_info, NULL)) { |
efe120a0 | 2097 | btrfs_warn(fs_info, |
52042d8e | 2098 | "too many missing devices, writable remount is not allowed"); |
292fd7fc SB |
2099 | ret = -EACCES; |
2100 | goto restore; | |
2101 | } | |
2102 | ||
8a3db184 | 2103 | if (btrfs_super_log_root(fs_info->super_copy) != 0) { |
10a3a3ed DS |
2104 | btrfs_warn(fs_info, |
2105 | "mount required to replay tree-log, cannot remount read-write"); | |
49b25e05 JM |
2106 | ret = -EINVAL; |
2107 | goto restore; | |
8a3db184 | 2108 | } |
c146afad | 2109 | |
44c0ca21 BB |
2110 | /* |
2111 | * NOTE: when remounting with a change that does writes, don't | |
2112 | * put it anywhere above this point, as we are not sure to be | |
2113 | * safe to write until we pass the above checks. | |
2114 | */ | |
2115 | ret = btrfs_start_pre_rw_mount(fs_info); | |
2b6ba629 ID |
2116 | if (ret) |
2117 | goto restore; | |
2118 | ||
a0a1db70 | 2119 | btrfs_clear_sb_rdonly(sb); |
90c711ab | 2120 | |
afcdd129 | 2121 | set_bit(BTRFS_FS_OPEN, &fs_info->flags); |
c146afad | 2122 | } |
dc81cdc5 | 2123 | out: |
faa00889 JB |
2124 | /* |
2125 | * We need to set SB_I_VERSION here otherwise it'll get cleared by VFS, | |
2126 | * since the absence of the flag means it can be toggled off by remount. | |
2127 | */ | |
2128 | *flags |= SB_I_VERSION; | |
2129 | ||
2c6a92b0 | 2130 | wake_up_process(fs_info->transaction_kthread); |
dc81cdc5 | 2131 | btrfs_remount_cleanup(fs_info, old_opts); |
8cd29088 | 2132 | btrfs_clear_oneshot_options(fs_info); |
88c4703f JT |
2133 | clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); |
2134 | ||
c146afad | 2135 | return 0; |
49b25e05 JM |
2136 | |
2137 | restore: | |
1751e8a6 | 2138 | /* We've hit an error - don't reset SB_RDONLY */ |
bc98a42c | 2139 | if (sb_rdonly(sb)) |
1751e8a6 | 2140 | old_flags |= SB_RDONLY; |
a0a1db70 FM |
2141 | if (!(old_flags & SB_RDONLY)) |
2142 | clear_bit(BTRFS_FS_STATE_RO, &fs_info->fs_state); | |
49b25e05 JM |
2143 | sb->s_flags = old_flags; |
2144 | fs_info->mount_opt = old_opts; | |
2145 | fs_info->compress_type = old_compress_type; | |
2146 | fs_info->max_inline = old_max_inline; | |
0d2450ab ST |
2147 | btrfs_resize_thread_pool(fs_info, |
2148 | old_thread_pool_size, fs_info->thread_pool_size); | |
49b25e05 | 2149 | fs_info->metadata_ratio = old_metadata_ratio; |
dc81cdc5 | 2150 | btrfs_remount_cleanup(fs_info, old_opts); |
88c4703f JT |
2151 | clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); |
2152 | ||
49b25e05 | 2153 | return ret; |
c146afad YZ |
2154 | } |
2155 | ||
bcd53741 | 2156 | /* Used to sort the devices by max_avail(descending sort) */ |
214cc184 | 2157 | static int btrfs_cmp_device_free_bytes(const void *a, const void *b) |
bcd53741 | 2158 | { |
214cc184 DS |
2159 | const struct btrfs_device_info *dev_info1 = a; |
2160 | const struct btrfs_device_info *dev_info2 = b; | |
2161 | ||
2162 | if (dev_info1->max_avail > dev_info2->max_avail) | |
bcd53741 | 2163 | return -1; |
214cc184 | 2164 | else if (dev_info1->max_avail < dev_info2->max_avail) |
bcd53741 | 2165 | return 1; |
bcd53741 AJ |
2166 | return 0; |
2167 | } | |
2168 | ||
2169 | /* | |
2170 | * sort the devices by max_avail, in which max free extent size of each device | |
2171 | * is stored.(Descending Sort) | |
2172 | */ | |
2173 | static inline void btrfs_descending_sort_devices( | |
2174 | struct btrfs_device_info *devices, | |
2175 | size_t nr_devices) | |
2176 | { | |
2177 | sort(devices, nr_devices, sizeof(struct btrfs_device_info), | |
2178 | btrfs_cmp_device_free_bytes, NULL); | |
2179 | } | |
2180 | ||
6d07bcec MX |
2181 | /* |
2182 | * The helper to calc the free space on the devices that can be used to store | |
2183 | * file data. | |
2184 | */ | |
7e17916b AB |
2185 | static inline int btrfs_calc_avail_data_space(struct btrfs_fs_info *fs_info, |
2186 | u64 *free_bytes) | |
6d07bcec | 2187 | { |
6d07bcec MX |
2188 | struct btrfs_device_info *devices_info; |
2189 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; | |
2190 | struct btrfs_device *device; | |
6d07bcec MX |
2191 | u64 type; |
2192 | u64 avail_space; | |
6d07bcec | 2193 | u64 min_stripe_size; |
559ca6ea | 2194 | int num_stripes = 1; |
6d07bcec | 2195 | int i = 0, nr_devices; |
4f080f57 | 2196 | const struct btrfs_raid_attr *rattr; |
6d07bcec | 2197 | |
7e33fd99 | 2198 | /* |
01327610 | 2199 | * We aren't under the device list lock, so this is racy-ish, but good |
7e33fd99 JB |
2200 | * enough for our purposes. |
2201 | */ | |
b772a86e | 2202 | nr_devices = fs_info->fs_devices->open_devices; |
7e33fd99 JB |
2203 | if (!nr_devices) { |
2204 | smp_mb(); | |
2205 | nr_devices = fs_info->fs_devices->open_devices; | |
2206 | ASSERT(nr_devices); | |
2207 | if (!nr_devices) { | |
2208 | *free_bytes = 0; | |
2209 | return 0; | |
2210 | } | |
2211 | } | |
6d07bcec | 2212 | |
d9b0d9ba | 2213 | devices_info = kmalloc_array(nr_devices, sizeof(*devices_info), |
6a44517d | 2214 | GFP_KERNEL); |
6d07bcec MX |
2215 | if (!devices_info) |
2216 | return -ENOMEM; | |
2217 | ||
01327610 | 2218 | /* calc min stripe number for data space allocation */ |
1b86826d | 2219 | type = btrfs_data_alloc_profile(fs_info); |
4f080f57 DS |
2220 | rattr = &btrfs_raid_array[btrfs_bg_flags_to_raid_index(type)]; |
2221 | ||
e1ea2bee | 2222 | if (type & BTRFS_BLOCK_GROUP_RAID0) |
39fb26c3 | 2223 | num_stripes = nr_devices; |
e1ea2bee | 2224 | else if (type & BTRFS_BLOCK_GROUP_RAID1) |
39fb26c3 | 2225 | num_stripes = 2; |
47e6f742 DS |
2226 | else if (type & BTRFS_BLOCK_GROUP_RAID1C3) |
2227 | num_stripes = 3; | |
8d6fac00 DS |
2228 | else if (type & BTRFS_BLOCK_GROUP_RAID1C4) |
2229 | num_stripes = 4; | |
e1ea2bee | 2230 | else if (type & BTRFS_BLOCK_GROUP_RAID10) |
39fb26c3 | 2231 | num_stripes = 4; |
6d07bcec | 2232 | |
4f080f57 DS |
2233 | /* Adjust for more than 1 stripe per device */ |
2234 | min_stripe_size = rattr->dev_stripes * BTRFS_STRIPE_LEN; | |
6d07bcec | 2235 | |
7e33fd99 JB |
2236 | rcu_read_lock(); |
2237 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { | |
e12c9621 AJ |
2238 | if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, |
2239 | &device->dev_state) || | |
401e29c1 AJ |
2240 | !device->bdev || |
2241 | test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) | |
6d07bcec MX |
2242 | continue; |
2243 | ||
7e33fd99 JB |
2244 | if (i >= nr_devices) |
2245 | break; | |
2246 | ||
6d07bcec MX |
2247 | avail_space = device->total_bytes - device->bytes_used; |
2248 | ||
2249 | /* align with stripe_len */ | |
559ca6ea | 2250 | avail_space = rounddown(avail_space, BTRFS_STRIPE_LEN); |
6d07bcec MX |
2251 | |
2252 | /* | |
01327610 | 2253 | * In order to avoid overwriting the superblock on the drive, |
6d07bcec MX |
2254 | * btrfs starts at an offset of at least 1MB when doing chunk |
2255 | * allocation. | |
559ca6ea NB |
2256 | * |
2257 | * This ensures we have at least min_stripe_size free space | |
2258 | * after excluding 1MB. | |
6d07bcec | 2259 | */ |
559ca6ea | 2260 | if (avail_space <= SZ_1M + min_stripe_size) |
6d07bcec MX |
2261 | continue; |
2262 | ||
559ca6ea NB |
2263 | avail_space -= SZ_1M; |
2264 | ||
6d07bcec MX |
2265 | devices_info[i].dev = device; |
2266 | devices_info[i].max_avail = avail_space; | |
2267 | ||
2268 | i++; | |
2269 | } | |
7e33fd99 | 2270 | rcu_read_unlock(); |
6d07bcec MX |
2271 | |
2272 | nr_devices = i; | |
2273 | ||
2274 | btrfs_descending_sort_devices(devices_info, nr_devices); | |
2275 | ||
2276 | i = nr_devices - 1; | |
2277 | avail_space = 0; | |
559ca6ea NB |
2278 | while (nr_devices >= rattr->devs_min) { |
2279 | num_stripes = min(num_stripes, nr_devices); | |
39fb26c3 | 2280 | |
6d07bcec MX |
2281 | if (devices_info[i].max_avail >= min_stripe_size) { |
2282 | int j; | |
2283 | u64 alloc_size; | |
2284 | ||
39fb26c3 | 2285 | avail_space += devices_info[i].max_avail * num_stripes; |
6d07bcec | 2286 | alloc_size = devices_info[i].max_avail; |
39fb26c3 | 2287 | for (j = i + 1 - num_stripes; j <= i; j++) |
6d07bcec MX |
2288 | devices_info[j].max_avail -= alloc_size; |
2289 | } | |
2290 | i--; | |
2291 | nr_devices--; | |
2292 | } | |
2293 | ||
2294 | kfree(devices_info); | |
2295 | *free_bytes = avail_space; | |
2296 | return 0; | |
2297 | } | |
2298 | ||
ba7b6e62 DS |
2299 | /* |
2300 | * Calculate numbers for 'df', pessimistic in case of mixed raid profiles. | |
2301 | * | |
2302 | * If there's a redundant raid level at DATA block groups, use the respective | |
2303 | * multiplier to scale the sizes. | |
2304 | * | |
2305 | * Unused device space usage is based on simulating the chunk allocator | |
0d0c71b3 DS |
2306 | * algorithm that respects the device sizes and order of allocations. This is |
2307 | * a close approximation of the actual use but there are other factors that may | |
2308 | * change the result (like a new metadata chunk). | |
ba7b6e62 | 2309 | * |
ca8a51b3 | 2310 | * If metadata is exhausted, f_bavail will be 0. |
ba7b6e62 | 2311 | */ |
8fd17795 CM |
2312 | static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) |
2313 | { | |
815745cf AV |
2314 | struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb); |
2315 | struct btrfs_super_block *disk_super = fs_info->super_copy; | |
bd4d1088 JB |
2316 | struct btrfs_space_info *found; |
2317 | u64 total_used = 0; | |
6d07bcec | 2318 | u64 total_free_data = 0; |
ca8a51b3 | 2319 | u64 total_free_meta = 0; |
265fdfa6 | 2320 | u32 bits = fs_info->sectorsize_bits; |
de37aa51 | 2321 | __be32 *fsid = (__be32 *)fs_info->fs_devices->fsid; |
ba7b6e62 DS |
2322 | unsigned factor = 1; |
2323 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; | |
6d07bcec | 2324 | int ret; |
ca8a51b3 | 2325 | u64 thresh = 0; |
ae02d1bd | 2326 | int mixed = 0; |
8fd17795 | 2327 | |
72804905 | 2328 | list_for_each_entry(found, &fs_info->space_info, list) { |
6d07bcec | 2329 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) { |
ba7b6e62 DS |
2330 | int i; |
2331 | ||
6d07bcec MX |
2332 | total_free_data += found->disk_total - found->disk_used; |
2333 | total_free_data -= | |
2334 | btrfs_account_ro_block_groups_free_space(found); | |
ba7b6e62 DS |
2335 | |
2336 | for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) { | |
46df06b8 DS |
2337 | if (!list_empty(&found->block_groups[i])) |
2338 | factor = btrfs_bg_type_to_factor( | |
2339 | btrfs_raid_array[i].bg_flag); | |
ba7b6e62 | 2340 | } |
6d07bcec | 2341 | } |
ae02d1bd LB |
2342 | |
2343 | /* | |
2344 | * Metadata in mixed block goup profiles are accounted in data | |
2345 | */ | |
2346 | if (!mixed && found->flags & BTRFS_BLOCK_GROUP_METADATA) { | |
2347 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) | |
2348 | mixed = 1; | |
2349 | else | |
2350 | total_free_meta += found->disk_total - | |
2351 | found->disk_used; | |
2352 | } | |
6d07bcec | 2353 | |
b742bb82 | 2354 | total_used += found->disk_used; |
89a55897 | 2355 | } |
ba7b6e62 | 2356 | |
ba7b6e62 DS |
2357 | buf->f_blocks = div_u64(btrfs_super_total_bytes(disk_super), factor); |
2358 | buf->f_blocks >>= bits; | |
2359 | buf->f_bfree = buf->f_blocks - (div_u64(total_used, factor) >> bits); | |
2360 | ||
2361 | /* Account global block reserve as used, it's in logical size already */ | |
2362 | spin_lock(&block_rsv->lock); | |
41b34acc LB |
2363 | /* Mixed block groups accounting is not byte-accurate, avoid overflow */ |
2364 | if (buf->f_bfree >= block_rsv->size >> bits) | |
2365 | buf->f_bfree -= block_rsv->size >> bits; | |
2366 | else | |
2367 | buf->f_bfree = 0; | |
ba7b6e62 DS |
2368 | spin_unlock(&block_rsv->lock); |
2369 | ||
0d95c1be | 2370 | buf->f_bavail = div_u64(total_free_data, factor); |
6bccf3ab | 2371 | ret = btrfs_calc_avail_data_space(fs_info, &total_free_data); |
7e33fd99 | 2372 | if (ret) |
6d07bcec | 2373 | return ret; |
ba7b6e62 | 2374 | buf->f_bavail += div_u64(total_free_data, factor); |
6d07bcec | 2375 | buf->f_bavail = buf->f_bavail >> bits; |
d397712b | 2376 | |
ca8a51b3 DS |
2377 | /* |
2378 | * We calculate the remaining metadata space minus global reserve. If | |
2379 | * this is (supposedly) smaller than zero, there's no space. But this | |
2380 | * does not hold in practice, the exhausted state happens where's still | |
2381 | * some positive delta. So we apply some guesswork and compare the | |
2382 | * delta to a 4M threshold. (Practically observed delta was ~2M.) | |
2383 | * | |
2384 | * We probably cannot calculate the exact threshold value because this | |
2385 | * depends on the internal reservations requested by various | |
2386 | * operations, so some operations that consume a few metadata will | |
2387 | * succeed even if the Avail is zero. But this is better than the other | |
2388 | * way around. | |
2389 | */ | |
d4417e22 | 2390 | thresh = SZ_4M; |
ca8a51b3 | 2391 | |
d55966c4 JB |
2392 | /* |
2393 | * We only want to claim there's no available space if we can no longer | |
2394 | * allocate chunks for our metadata profile and our global reserve will | |
2395 | * not fit in the free metadata space. If we aren't ->full then we | |
2396 | * still can allocate chunks and thus are fine using the currently | |
2397 | * calculated f_bavail. | |
2398 | */ | |
2399 | if (!mixed && block_rsv->space_info->full && | |
2400 | total_free_meta - thresh < block_rsv->size) | |
ca8a51b3 DS |
2401 | buf->f_bavail = 0; |
2402 | ||
ba7b6e62 DS |
2403 | buf->f_type = BTRFS_SUPER_MAGIC; |
2404 | buf->f_bsize = dentry->d_sb->s_blocksize; | |
2405 | buf->f_namelen = BTRFS_NAME_LEN; | |
2406 | ||
9d03632e | 2407 | /* We treat it as constant endianness (it doesn't matter _which_) |
d397712b | 2408 | because we want the fsid to come out the same whether mounted |
9d03632e DW |
2409 | on a big-endian or little-endian host */ |
2410 | buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]); | |
2411 | buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]); | |
32d48fa1 | 2412 | /* Mask in the root object ID too, to disambiguate subvols */ |
4fd786e6 MT |
2413 | buf->f_fsid.val[0] ^= |
2414 | BTRFS_I(d_inode(dentry))->root->root_key.objectid >> 32; | |
2415 | buf->f_fsid.val[1] ^= | |
2416 | BTRFS_I(d_inode(dentry))->root->root_key.objectid; | |
32d48fa1 | 2417 | |
8fd17795 CM |
2418 | return 0; |
2419 | } | |
b5133862 | 2420 | |
aea52e19 AV |
2421 | static void btrfs_kill_super(struct super_block *sb) |
2422 | { | |
815745cf | 2423 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
aea52e19 | 2424 | kill_anon_super(sb); |
0d4b0463 | 2425 | btrfs_free_fs_info(fs_info); |
aea52e19 AV |
2426 | } |
2427 | ||
2e635a27 CM |
2428 | static struct file_system_type btrfs_fs_type = { |
2429 | .owner = THIS_MODULE, | |
2430 | .name = "btrfs", | |
061dbc6b | 2431 | .mount = btrfs_mount, |
aea52e19 | 2432 | .kill_sb = btrfs_kill_super, |
f667aef6 | 2433 | .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA, |
2e635a27 | 2434 | }; |
72fa39f5 MT |
2435 | |
2436 | static struct file_system_type btrfs_root_fs_type = { | |
2437 | .owner = THIS_MODULE, | |
2438 | .name = "btrfs", | |
2439 | .mount = btrfs_mount_root, | |
2440 | .kill_sb = btrfs_kill_super, | |
5b9b26f5 | 2441 | .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA | FS_ALLOW_IDMAP, |
72fa39f5 MT |
2442 | }; |
2443 | ||
7f78e035 | 2444 | MODULE_ALIAS_FS("btrfs"); |
a9218f6b | 2445 | |
d8620958 TVB |
2446 | static int btrfs_control_open(struct inode *inode, struct file *file) |
2447 | { | |
2448 | /* | |
2449 | * The control file's private_data is used to hold the | |
2450 | * transaction when it is started and is used to keep | |
2451 | * track of whether a transaction is already in progress. | |
2452 | */ | |
2453 | file->private_data = NULL; | |
2454 | return 0; | |
2455 | } | |
2456 | ||
d352ac68 | 2457 | /* |
cfe953c8 | 2458 | * Used by /dev/btrfs-control for devices ioctls. |
d352ac68 | 2459 | */ |
8a4b83cc CM |
2460 | static long btrfs_control_ioctl(struct file *file, unsigned int cmd, |
2461 | unsigned long arg) | |
2462 | { | |
2463 | struct btrfs_ioctl_vol_args *vol; | |
36350e95 | 2464 | struct btrfs_device *device = NULL; |
16cab91a | 2465 | dev_t devt = 0; |
c071fcfd | 2466 | int ret = -ENOTTY; |
8a4b83cc | 2467 | |
e441d54d CM |
2468 | if (!capable(CAP_SYS_ADMIN)) |
2469 | return -EPERM; | |
2470 | ||
dae7b665 LZ |
2471 | vol = memdup_user((void __user *)arg, sizeof(*vol)); |
2472 | if (IS_ERR(vol)) | |
2473 | return PTR_ERR(vol); | |
f505754f | 2474 | vol->name[BTRFS_PATH_NAME_MAX] = '\0'; |
c071fcfd | 2475 | |
8a4b83cc CM |
2476 | switch (cmd) { |
2477 | case BTRFS_IOC_SCAN_DEV: | |
899f9307 | 2478 | mutex_lock(&uuid_mutex); |
36350e95 GJ |
2479 | device = btrfs_scan_one_device(vol->name, FMODE_READ, |
2480 | &btrfs_root_fs_type); | |
2481 | ret = PTR_ERR_OR_ZERO(device); | |
899f9307 | 2482 | mutex_unlock(&uuid_mutex); |
8a4b83cc | 2483 | break; |
228a73ab | 2484 | case BTRFS_IOC_FORGET_DEV: |
16cab91a AJ |
2485 | if (vol->name[0] != 0) { |
2486 | ret = lookup_bdev(vol->name, &devt); | |
2487 | if (ret) | |
2488 | break; | |
2489 | } | |
2490 | ret = btrfs_forget_devices(devt); | |
228a73ab | 2491 | break; |
02db0844 | 2492 | case BTRFS_IOC_DEVICES_READY: |
899f9307 | 2493 | mutex_lock(&uuid_mutex); |
36350e95 GJ |
2494 | device = btrfs_scan_one_device(vol->name, FMODE_READ, |
2495 | &btrfs_root_fs_type); | |
2496 | if (IS_ERR(device)) { | |
899f9307 | 2497 | mutex_unlock(&uuid_mutex); |
36350e95 | 2498 | ret = PTR_ERR(device); |
02db0844 | 2499 | break; |
899f9307 | 2500 | } |
36350e95 GJ |
2501 | ret = !(device->fs_devices->num_devices == |
2502 | device->fs_devices->total_devices); | |
899f9307 | 2503 | mutex_unlock(&uuid_mutex); |
02db0844 | 2504 | break; |
c5868f83 | 2505 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 2506 | ret = btrfs_ioctl_get_supported_features((void __user*)arg); |
c5868f83 | 2507 | break; |
8a4b83cc | 2508 | } |
dae7b665 | 2509 | |
8a4b83cc | 2510 | kfree(vol); |
f819d837 | 2511 | return ret; |
8a4b83cc CM |
2512 | } |
2513 | ||
0176260f | 2514 | static int btrfs_freeze(struct super_block *sb) |
ed0dab6b | 2515 | { |
354aa0fb | 2516 | struct btrfs_trans_handle *trans; |
0b246afa JM |
2517 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
2518 | struct btrfs_root *root = fs_info->tree_root; | |
354aa0fb | 2519 | |
fac03c8d | 2520 | set_bit(BTRFS_FS_FROZEN, &fs_info->flags); |
9e7cc91a WX |
2521 | /* |
2522 | * We don't need a barrier here, we'll wait for any transaction that | |
2523 | * could be in progress on other threads (and do delayed iputs that | |
2524 | * we want to avoid on a frozen filesystem), or do the commit | |
2525 | * ourselves. | |
2526 | */ | |
d4edf39b | 2527 | trans = btrfs_attach_transaction_barrier(root); |
354aa0fb MX |
2528 | if (IS_ERR(trans)) { |
2529 | /* no transaction, don't bother */ | |
2530 | if (PTR_ERR(trans) == -ENOENT) | |
2531 | return 0; | |
2532 | return PTR_ERR(trans); | |
2533 | } | |
3a45bb20 | 2534 | return btrfs_commit_transaction(trans); |
ed0dab6b Y |
2535 | } |
2536 | ||
9e7cc91a WX |
2537 | static int btrfs_unfreeze(struct super_block *sb) |
2538 | { | |
fac03c8d DS |
2539 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
2540 | ||
2541 | clear_bit(BTRFS_FS_FROZEN, &fs_info->flags); | |
9e7cc91a WX |
2542 | return 0; |
2543 | } | |
2544 | ||
9c5085c1 JB |
2545 | static int btrfs_show_devname(struct seq_file *m, struct dentry *root) |
2546 | { | |
2547 | struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb); | |
9c5085c1 | 2548 | |
88c14590 | 2549 | /* |
6605fd2f AJ |
2550 | * There should be always a valid pointer in latest_dev, it may be stale |
2551 | * for a short moment in case it's being deleted but still valid until | |
2552 | * the end of RCU grace period. | |
88c14590 DS |
2553 | */ |
2554 | rcu_read_lock(); | |
6605fd2f | 2555 | seq_escape(m, rcu_str_deref(fs_info->fs_devices->latest_dev->name), " \t\n\\"); |
88c14590 | 2556 | rcu_read_unlock(); |
6605fd2f | 2557 | |
9c5085c1 JB |
2558 | return 0; |
2559 | } | |
2560 | ||
b87221de | 2561 | static const struct super_operations btrfs_super_ops = { |
76dda93c | 2562 | .drop_inode = btrfs_drop_inode, |
bd555975 | 2563 | .evict_inode = btrfs_evict_inode, |
e20d96d6 | 2564 | .put_super = btrfs_put_super, |
d5719762 | 2565 | .sync_fs = btrfs_sync_fs, |
a9572a15 | 2566 | .show_options = btrfs_show_options, |
9c5085c1 | 2567 | .show_devname = btrfs_show_devname, |
2c90e5d6 CM |
2568 | .alloc_inode = btrfs_alloc_inode, |
2569 | .destroy_inode = btrfs_destroy_inode, | |
26602cab | 2570 | .free_inode = btrfs_free_inode, |
8fd17795 | 2571 | .statfs = btrfs_statfs, |
c146afad | 2572 | .remount_fs = btrfs_remount, |
0176260f | 2573 | .freeze_fs = btrfs_freeze, |
9e7cc91a | 2574 | .unfreeze_fs = btrfs_unfreeze, |
e20d96d6 | 2575 | }; |
a9218f6b CM |
2576 | |
2577 | static const struct file_operations btrfs_ctl_fops = { | |
d8620958 | 2578 | .open = btrfs_control_open, |
a9218f6b | 2579 | .unlocked_ioctl = btrfs_control_ioctl, |
1832f2d8 | 2580 | .compat_ioctl = compat_ptr_ioctl, |
a9218f6b | 2581 | .owner = THIS_MODULE, |
6038f373 | 2582 | .llseek = noop_llseek, |
a9218f6b CM |
2583 | }; |
2584 | ||
2585 | static struct miscdevice btrfs_misc = { | |
578454ff | 2586 | .minor = BTRFS_MINOR, |
a9218f6b CM |
2587 | .name = "btrfs-control", |
2588 | .fops = &btrfs_ctl_fops | |
2589 | }; | |
2590 | ||
578454ff KS |
2591 | MODULE_ALIAS_MISCDEV(BTRFS_MINOR); |
2592 | MODULE_ALIAS("devname:btrfs-control"); | |
2593 | ||
f5c29bd9 | 2594 | static int __init btrfs_interface_init(void) |
a9218f6b CM |
2595 | { |
2596 | return misc_register(&btrfs_misc); | |
2597 | } | |
2598 | ||
e67c718b | 2599 | static __cold void btrfs_interface_exit(void) |
a9218f6b | 2600 | { |
f368ed60 | 2601 | misc_deregister(&btrfs_misc); |
a9218f6b CM |
2602 | } |
2603 | ||
f5c29bd9 | 2604 | static void __init btrfs_print_mod_info(void) |
85965600 | 2605 | { |
edf57cbf | 2606 | static const char options[] = "" |
85965600 DS |
2607 | #ifdef CONFIG_BTRFS_DEBUG |
2608 | ", debug=on" | |
2609 | #endif | |
79556c3d SB |
2610 | #ifdef CONFIG_BTRFS_ASSERT |
2611 | ", assert=on" | |
2612 | #endif | |
85965600 DS |
2613 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
2614 | ", integrity-checker=on" | |
fb592373 JB |
2615 | #endif |
2616 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
2617 | ", ref-verify=on" | |
5b316468 NA |
2618 | #endif |
2619 | #ifdef CONFIG_BLK_DEV_ZONED | |
2620 | ", zoned=yes" | |
2621 | #else | |
2622 | ", zoned=no" | |
ea3dc7d2 DS |
2623 | #endif |
2624 | #ifdef CONFIG_FS_VERITY | |
2625 | ", fsverity=yes" | |
2626 | #else | |
2627 | ", fsverity=no" | |
85965600 | 2628 | #endif |
edf57cbf BVA |
2629 | ; |
2630 | pr_info("Btrfs loaded, crc32c=%s%s\n", crc32c_impl(), options); | |
85965600 DS |
2631 | } |
2632 | ||
2e635a27 CM |
2633 | static int __init init_btrfs_fs(void) |
2634 | { | |
2c90e5d6 | 2635 | int err; |
58176a96 | 2636 | |
63541927 FDBM |
2637 | btrfs_props_init(); |
2638 | ||
58176a96 JB |
2639 | err = btrfs_init_sysfs(); |
2640 | if (err) | |
9678c543 | 2641 | return err; |
58176a96 | 2642 | |
143bede5 | 2643 | btrfs_init_compress(); |
d1310b2e | 2644 | |
261507a0 LZ |
2645 | err = btrfs_init_cachep(); |
2646 | if (err) | |
2647 | goto free_compress; | |
2648 | ||
d1310b2e | 2649 | err = extent_io_init(); |
2f4cbe64 WB |
2650 | if (err) |
2651 | goto free_cachep; | |
2652 | ||
6f0d04f8 | 2653 | err = extent_state_cache_init(); |
d1310b2e CM |
2654 | if (err) |
2655 | goto free_extent_io; | |
2656 | ||
6f0d04f8 JB |
2657 | err = extent_map_init(); |
2658 | if (err) | |
2659 | goto free_extent_state_cache; | |
2660 | ||
6352b91d | 2661 | err = ordered_data_init(); |
2f4cbe64 WB |
2662 | if (err) |
2663 | goto free_extent_map; | |
c8b97818 | 2664 | |
6352b91d MX |
2665 | err = btrfs_delayed_inode_init(); |
2666 | if (err) | |
2667 | goto free_ordered_data; | |
2668 | ||
9247f317 | 2669 | err = btrfs_auto_defrag_init(); |
16cdcec7 MX |
2670 | if (err) |
2671 | goto free_delayed_inode; | |
2672 | ||
78a6184a | 2673 | err = btrfs_delayed_ref_init(); |
9247f317 MX |
2674 | if (err) |
2675 | goto free_auto_defrag; | |
2676 | ||
b9e9a6cb WS |
2677 | err = btrfs_prelim_ref_init(); |
2678 | if (err) | |
af13b492 | 2679 | goto free_delayed_ref; |
b9e9a6cb | 2680 | |
97eb6b69 | 2681 | err = btrfs_end_io_wq_init(); |
78a6184a | 2682 | if (err) |
af13b492 | 2683 | goto free_prelim_ref; |
78a6184a | 2684 | |
97eb6b69 DS |
2685 | err = btrfs_interface_init(); |
2686 | if (err) | |
2687 | goto free_end_io_wq; | |
2688 | ||
8ae1af3c | 2689 | btrfs_print_mod_info(); |
dc11dd5d JB |
2690 | |
2691 | err = btrfs_run_sanity_tests(); | |
2692 | if (err) | |
2693 | goto unregister_ioctl; | |
2694 | ||
2695 | err = register_filesystem(&btrfs_fs_type); | |
2696 | if (err) | |
2697 | goto unregister_ioctl; | |
74255aa0 | 2698 | |
2f4cbe64 WB |
2699 | return 0; |
2700 | ||
a9218f6b CM |
2701 | unregister_ioctl: |
2702 | btrfs_interface_exit(); | |
97eb6b69 DS |
2703 | free_end_io_wq: |
2704 | btrfs_end_io_wq_exit(); | |
b9e9a6cb WS |
2705 | free_prelim_ref: |
2706 | btrfs_prelim_ref_exit(); | |
78a6184a MX |
2707 | free_delayed_ref: |
2708 | btrfs_delayed_ref_exit(); | |
9247f317 MX |
2709 | free_auto_defrag: |
2710 | btrfs_auto_defrag_exit(); | |
16cdcec7 MX |
2711 | free_delayed_inode: |
2712 | btrfs_delayed_inode_exit(); | |
6352b91d MX |
2713 | free_ordered_data: |
2714 | ordered_data_exit(); | |
2f4cbe64 WB |
2715 | free_extent_map: |
2716 | extent_map_exit(); | |
6f0d04f8 JB |
2717 | free_extent_state_cache: |
2718 | extent_state_cache_exit(); | |
d1310b2e CM |
2719 | free_extent_io: |
2720 | extent_io_exit(); | |
2f4cbe64 WB |
2721 | free_cachep: |
2722 | btrfs_destroy_cachep(); | |
261507a0 LZ |
2723 | free_compress: |
2724 | btrfs_exit_compress(); | |
2f4cbe64 | 2725 | btrfs_exit_sysfs(); |
9678c543 | 2726 | |
2f4cbe64 | 2727 | return err; |
2e635a27 CM |
2728 | } |
2729 | ||
2730 | static void __exit exit_btrfs_fs(void) | |
2731 | { | |
39279cc3 | 2732 | btrfs_destroy_cachep(); |
78a6184a | 2733 | btrfs_delayed_ref_exit(); |
9247f317 | 2734 | btrfs_auto_defrag_exit(); |
16cdcec7 | 2735 | btrfs_delayed_inode_exit(); |
b9e9a6cb | 2736 | btrfs_prelim_ref_exit(); |
6352b91d | 2737 | ordered_data_exit(); |
a52d9a80 | 2738 | extent_map_exit(); |
6f0d04f8 | 2739 | extent_state_cache_exit(); |
d1310b2e | 2740 | extent_io_exit(); |
a9218f6b | 2741 | btrfs_interface_exit(); |
5ed5f588 | 2742 | btrfs_end_io_wq_exit(); |
2e635a27 | 2743 | unregister_filesystem(&btrfs_fs_type); |
58176a96 | 2744 | btrfs_exit_sysfs(); |
8a4b83cc | 2745 | btrfs_cleanup_fs_uuids(); |
261507a0 | 2746 | btrfs_exit_compress(); |
2e635a27 CM |
2747 | } |
2748 | ||
60efa5eb | 2749 | late_initcall(init_btrfs_fs); |
2e635a27 CM |
2750 | module_exit(exit_btrfs_fs) |
2751 | ||
2752 | MODULE_LICENSE("GPL"); | |
d5178578 | 2753 | MODULE_SOFTDEP("pre: crc32c"); |
3951e7f0 | 2754 | MODULE_SOFTDEP("pre: xxhash64"); |
3831bf00 | 2755 | MODULE_SOFTDEP("pre: sha256"); |
352ae07b | 2756 | MODULE_SOFTDEP("pre: blake2b-256"); |