]>
Commit | Line | Data |
---|---|---|
c1d7c514 | 1 | // SPDX-License-Identifier: GPL-2.0 |
6cbd5570 CM |
2 | /* |
3 | * Copyright (C) 2007 Oracle. All rights reserved. | |
6cbd5570 CM |
4 | */ |
5 | ||
4b82d6e4 | 6 | #include <linux/blkdev.h> |
2e635a27 CM |
7 | #include <linux/module.h> |
8 | #include <linux/fs.h> | |
9 | #include <linux/pagemap.h> | |
10 | #include <linux/highmem.h> | |
11 | #include <linux/time.h> | |
12 | #include <linux/init.h> | |
a9572a15 | 13 | #include <linux/seq_file.h> |
2e635a27 | 14 | #include <linux/string.h> |
2e635a27 | 15 | #include <linux/backing-dev.h> |
4b82d6e4 | 16 | #include <linux/mount.h> |
75dfe396 | 17 | #include <linux/writeback.h> |
8fd17795 | 18 | #include <linux/statfs.h> |
08607c1b | 19 | #include <linux/compat.h> |
95e05289 | 20 | #include <linux/parser.h> |
c59f8951 | 21 | #include <linux/ctype.h> |
6da6abae | 22 | #include <linux/namei.h> |
a9218f6b | 23 | #include <linux/miscdevice.h> |
1bcbf313 | 24 | #include <linux/magic.h> |
5a0e3ad6 | 25 | #include <linux/slab.h> |
90a887c9 | 26 | #include <linux/cleancache.h> |
22c44fe6 | 27 | #include <linux/ratelimit.h> |
9678c543 | 28 | #include <linux/crc32c.h> |
55e301fd | 29 | #include <linux/btrfs.h> |
16cdcec7 | 30 | #include "delayed-inode.h" |
2e635a27 | 31 | #include "ctree.h" |
e20d96d6 | 32 | #include "disk-io.h" |
d5719762 | 33 | #include "transaction.h" |
2c90e5d6 | 34 | #include "btrfs_inode.h" |
3a686375 | 35 | #include "print-tree.h" |
63541927 | 36 | #include "props.h" |
5103e947 | 37 | #include "xattr.h" |
8a4b83cc | 38 | #include "volumes.h" |
be6e8dc0 | 39 | #include "export.h" |
c8b97818 | 40 | #include "compression.h" |
9c5085c1 | 41 | #include "rcu-string.h" |
8dabb742 | 42 | #include "dev-replace.h" |
74255aa0 | 43 | #include "free-space-cache.h" |
b9e9a6cb | 44 | #include "backref.h" |
8719aaae | 45 | #include "space-info.h" |
89439109 | 46 | #include "sysfs.h" |
dc11dd5d | 47 | #include "tests/btrfs-tests.h" |
aac0023c | 48 | #include "block-group.h" |
b0643e59 | 49 | #include "discard.h" |
2e635a27 | 50 | |
d3982100 | 51 | #include "qgroup.h" |
1abe9b8a | 52 | #define CREATE_TRACE_POINTS |
53 | #include <trace/events/btrfs.h> | |
54 | ||
b87221de | 55 | static const struct super_operations btrfs_super_ops; |
72fa39f5 MT |
56 | |
57 | /* | |
58 | * Types for mounting the default subvolume and a subvolume explicitly | |
59 | * requested by subvol=/path. That way the callchain is straightforward and we | |
60 | * don't have to play tricks with the mount options and recursive calls to | |
61 | * btrfs_mount. | |
312c89fb MT |
62 | * |
63 | * The new btrfs_root_fs_type also servers as a tag for the bdev_holder. | |
72fa39f5 | 64 | */ |
830c4adb | 65 | static struct file_system_type btrfs_fs_type; |
72fa39f5 | 66 | static struct file_system_type btrfs_root_fs_type; |
75dfe396 | 67 | |
0723a047 HH |
68 | static int btrfs_remount(struct super_block *sb, int *flags, char *data); |
69 | ||
4143cb8b | 70 | const char * __attribute_const__ btrfs_decode_error(int errno) |
acce952b | 71 | { |
08748810 | 72 | char *errstr = "unknown"; |
acce952b | 73 | |
74 | switch (errno) { | |
d54f8144 DS |
75 | case -ENOENT: /* -2 */ |
76 | errstr = "No such entry"; | |
77 | break; | |
78 | case -EIO: /* -5 */ | |
acce952b | 79 | errstr = "IO failure"; |
80 | break; | |
d54f8144 | 81 | case -ENOMEM: /* -12*/ |
acce952b | 82 | errstr = "Out of memory"; |
83 | break; | |
d54f8144 | 84 | case -EEXIST: /* -17 */ |
8c342930 JM |
85 | errstr = "Object already exists"; |
86 | break; | |
d54f8144 | 87 | case -ENOSPC: /* -28 */ |
94ef7280 DS |
88 | errstr = "No space left"; |
89 | break; | |
d54f8144 DS |
90 | case -EROFS: /* -30 */ |
91 | errstr = "Readonly filesystem"; | |
94ef7280 | 92 | break; |
fb8521ca DS |
93 | case -EOPNOTSUPP: /* -95 */ |
94 | errstr = "Operation not supported"; | |
95 | break; | |
96 | case -EUCLEAN: /* -117 */ | |
97 | errstr = "Filesystem corrupted"; | |
98 | break; | |
99 | case -EDQUOT: /* -122 */ | |
100 | errstr = "Quota exceeded"; | |
101 | break; | |
acce952b | 102 | } |
103 | ||
104 | return errstr; | |
105 | } | |
106 | ||
acce952b | 107 | /* |
34d97007 | 108 | * __btrfs_handle_fs_error decodes expected errors from the caller and |
52042d8e | 109 | * invokes the appropriate error response. |
acce952b | 110 | */ |
c0d19e2b | 111 | __cold |
34d97007 | 112 | void __btrfs_handle_fs_error(struct btrfs_fs_info *fs_info, const char *function, |
4da35113 | 113 | unsigned int line, int errno, const char *fmt, ...) |
acce952b | 114 | { |
115 | struct super_block *sb = fs_info->sb; | |
57d816a1 | 116 | #ifdef CONFIG_PRINTK |
acce952b | 117 | const char *errstr; |
57d816a1 | 118 | #endif |
acce952b | 119 | |
120 | /* | |
121 | * Special case: if the error is EROFS, and we're already | |
1751e8a6 | 122 | * under SB_RDONLY, then it is safe here. |
acce952b | 123 | */ |
bc98a42c | 124 | if (errno == -EROFS && sb_rdonly(sb)) |
4da35113 JM |
125 | return; |
126 | ||
57d816a1 | 127 | #ifdef CONFIG_PRINTK |
08748810 | 128 | errstr = btrfs_decode_error(errno); |
4da35113 | 129 | if (fmt) { |
37252a66 ES |
130 | struct va_format vaf; |
131 | va_list args; | |
132 | ||
133 | va_start(args, fmt); | |
134 | vaf.fmt = fmt; | |
135 | vaf.va = &args; | |
4da35113 | 136 | |
62e85577 | 137 | pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s (%pV)\n", |
08748810 | 138 | sb->s_id, function, line, errno, errstr, &vaf); |
37252a66 | 139 | va_end(args); |
4da35113 | 140 | } else { |
62e85577 | 141 | pr_crit("BTRFS: error (device %s) in %s:%d: errno=%d %s\n", |
08748810 | 142 | sb->s_id, function, line, errno, errstr); |
4da35113 | 143 | } |
57d816a1 | 144 | #endif |
acce952b | 145 | |
0713d90c AJ |
146 | /* |
147 | * Today we only save the error info to memory. Long term we'll | |
148 | * also send it down to the disk | |
149 | */ | |
150 | set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state); | |
151 | ||
4da35113 | 152 | /* Don't go through full error handling during mount */ |
922ea899 AJ |
153 | if (!(sb->s_flags & SB_BORN)) |
154 | return; | |
155 | ||
156 | if (sb_rdonly(sb)) | |
157 | return; | |
158 | ||
b0643e59 DZ |
159 | btrfs_discard_stop(fs_info); |
160 | ||
922ea899 AJ |
161 | /* btrfs handle error by forcing the filesystem readonly */ |
162 | sb->s_flags |= SB_RDONLY; | |
163 | btrfs_info(fs_info, "forced readonly"); | |
164 | /* | |
165 | * Note that a running device replace operation is not canceled here | |
166 | * although there is no way to update the progress. It would add the | |
167 | * risk of a deadlock, therefore the canceling is omitted. The only | |
168 | * penalty is that some I/O remains active until the procedure | |
52042d8e | 169 | * completes. The next time when the filesystem is mounted writable |
922ea899 AJ |
170 | * again, the device replace operation continues. |
171 | */ | |
4da35113 | 172 | } |
acce952b | 173 | |
57d816a1 | 174 | #ifdef CONFIG_PRINTK |
533574c6 | 175 | static const char * const logtypes[] = { |
4da35113 JM |
176 | "emergency", |
177 | "alert", | |
178 | "critical", | |
179 | "error", | |
180 | "warning", | |
181 | "notice", | |
182 | "info", | |
183 | "debug", | |
184 | }; | |
185 | ||
35f4e5e6 NB |
186 | |
187 | /* | |
188 | * Use one ratelimit state per log level so that a flood of less important | |
189 | * messages doesn't cause more important ones to be dropped. | |
190 | */ | |
191 | static struct ratelimit_state printk_limits[] = { | |
192 | RATELIMIT_STATE_INIT(printk_limits[0], DEFAULT_RATELIMIT_INTERVAL, 100), | |
193 | RATELIMIT_STATE_INIT(printk_limits[1], DEFAULT_RATELIMIT_INTERVAL, 100), | |
194 | RATELIMIT_STATE_INIT(printk_limits[2], DEFAULT_RATELIMIT_INTERVAL, 100), | |
195 | RATELIMIT_STATE_INIT(printk_limits[3], DEFAULT_RATELIMIT_INTERVAL, 100), | |
196 | RATELIMIT_STATE_INIT(printk_limits[4], DEFAULT_RATELIMIT_INTERVAL, 100), | |
197 | RATELIMIT_STATE_INIT(printk_limits[5], DEFAULT_RATELIMIT_INTERVAL, 100), | |
198 | RATELIMIT_STATE_INIT(printk_limits[6], DEFAULT_RATELIMIT_INTERVAL, 100), | |
199 | RATELIMIT_STATE_INIT(printk_limits[7], DEFAULT_RATELIMIT_INTERVAL, 100), | |
200 | }; | |
201 | ||
b105e927 | 202 | void __cold btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...) |
4da35113 | 203 | { |
40f7828b | 204 | char lvl[PRINTK_MAX_SINGLE_HEADER_LEN + 1] = "\0"; |
4da35113 JM |
205 | struct va_format vaf; |
206 | va_list args; | |
533574c6 | 207 | int kern_level; |
40f7828b PM |
208 | const char *type = logtypes[4]; |
209 | struct ratelimit_state *ratelimit = &printk_limits[4]; | |
4da35113 JM |
210 | |
211 | va_start(args, fmt); | |
212 | ||
262c5e86 | 213 | while ((kern_level = printk_get_level(fmt)) != 0) { |
533574c6 | 214 | size_t size = printk_skip_level(fmt) - fmt; |
262c5e86 PM |
215 | |
216 | if (kern_level >= '0' && kern_level <= '7') { | |
217 | memcpy(lvl, fmt, size); | |
218 | lvl[size] = '\0'; | |
219 | type = logtypes[kern_level - '0']; | |
220 | ratelimit = &printk_limits[kern_level - '0']; | |
221 | } | |
533574c6 | 222 | fmt += size; |
262c5e86 PM |
223 | } |
224 | ||
4da35113 JM |
225 | vaf.fmt = fmt; |
226 | vaf.va = &args; | |
533574c6 | 227 | |
35f4e5e6 | 228 | if (__ratelimit(ratelimit)) |
3993b112 CIK |
229 | printk("%sBTRFS %s (device %s): %pV\n", lvl, type, |
230 | fs_info ? fs_info->sb->s_id : "<unknown>", &vaf); | |
533574c6 JP |
231 | |
232 | va_end(args); | |
233 | } | |
533574c6 | 234 | #endif |
acce952b | 235 | |
49b25e05 JM |
236 | /* |
237 | * We only mark the transaction aborted and then set the file system read-only. | |
238 | * This will prevent new transactions from starting or trying to join this | |
239 | * one. | |
240 | * | |
241 | * This means that error recovery at the call site is limited to freeing | |
242 | * any local memory allocations and passing the error code up without | |
243 | * further cleanup. The transaction should complete as it normally would | |
244 | * in the call path but will return -EIO. | |
245 | * | |
246 | * We'll complete the cleanup in btrfs_end_transaction and | |
247 | * btrfs_commit_transaction. | |
248 | */ | |
c0d19e2b | 249 | __cold |
49b25e05 | 250 | void __btrfs_abort_transaction(struct btrfs_trans_handle *trans, |
66642832 | 251 | const char *function, |
49b25e05 JM |
252 | unsigned int line, int errno) |
253 | { | |
66642832 JM |
254 | struct btrfs_fs_info *fs_info = trans->fs_info; |
255 | ||
bf31f87f | 256 | WRITE_ONCE(trans->aborted, errno); |
49b25e05 JM |
257 | /* Nothing used. The other threads that have joined this |
258 | * transaction may be able to continue. */ | |
64c12921 | 259 | if (!trans->dirty && list_empty(&trans->new_bgs)) { |
69ce977a MX |
260 | const char *errstr; |
261 | ||
08748810 | 262 | errstr = btrfs_decode_error(errno); |
66642832 | 263 | btrfs_warn(fs_info, |
c2cf52eb SK |
264 | "%s:%d: Aborting unused transaction(%s).", |
265 | function, line, errstr); | |
acce952b | 266 | return; |
49b25e05 | 267 | } |
20c7bcec | 268 | WRITE_ONCE(trans->transaction->aborted, errno); |
501407aa | 269 | /* Wake up anybody who may be waiting on this transaction */ |
66642832 JM |
270 | wake_up(&fs_info->transaction_wait); |
271 | wake_up(&fs_info->transaction_blocked_wait); | |
272 | __btrfs_handle_fs_error(fs_info, function, line, errno, NULL); | |
49b25e05 | 273 | } |
8c342930 JM |
274 | /* |
275 | * __btrfs_panic decodes unexpected, fatal errors from the caller, | |
276 | * issues an alert, and either panics or BUGs, depending on mount options. | |
277 | */ | |
c0d19e2b | 278 | __cold |
8c342930 JM |
279 | void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function, |
280 | unsigned int line, int errno, const char *fmt, ...) | |
281 | { | |
8c342930 JM |
282 | char *s_id = "<unknown>"; |
283 | const char *errstr; | |
284 | struct va_format vaf = { .fmt = fmt }; | |
285 | va_list args; | |
acce952b | 286 | |
8c342930 JM |
287 | if (fs_info) |
288 | s_id = fs_info->sb->s_id; | |
acce952b | 289 | |
8c342930 JM |
290 | va_start(args, fmt); |
291 | vaf.va = &args; | |
292 | ||
08748810 | 293 | errstr = btrfs_decode_error(errno); |
d8953d69 | 294 | if (fs_info && (btrfs_test_opt(fs_info, PANIC_ON_FATAL_ERROR))) |
08748810 DS |
295 | panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (errno=%d %s)\n", |
296 | s_id, function, line, &vaf, errno, errstr); | |
8c342930 | 297 | |
efe120a0 FH |
298 | btrfs_crit(fs_info, "panic in %s:%d: %pV (errno=%d %s)", |
299 | function, line, &vaf, errno, errstr); | |
8c342930 JM |
300 | va_end(args); |
301 | /* Caller calls BUG() */ | |
acce952b | 302 | } |
303 | ||
d397712b | 304 | static void btrfs_put_super(struct super_block *sb) |
b18c6685 | 305 | { |
6bccf3ab | 306 | close_ctree(btrfs_sb(sb)); |
75dfe396 CM |
307 | } |
308 | ||
95e05289 | 309 | enum { |
416a7202 DS |
310 | Opt_acl, Opt_noacl, |
311 | Opt_clear_cache, | |
312 | Opt_commit_interval, | |
313 | Opt_compress, | |
314 | Opt_compress_force, | |
315 | Opt_compress_force_type, | |
316 | Opt_compress_type, | |
317 | Opt_degraded, | |
318 | Opt_device, | |
319 | Opt_fatal_errors, | |
320 | Opt_flushoncommit, Opt_noflushoncommit, | |
321 | Opt_inode_cache, Opt_noinode_cache, | |
322 | Opt_max_inline, | |
323 | Opt_barrier, Opt_nobarrier, | |
324 | Opt_datacow, Opt_nodatacow, | |
325 | Opt_datasum, Opt_nodatasum, | |
326 | Opt_defrag, Opt_nodefrag, | |
327 | Opt_discard, Opt_nodiscard, | |
b0643e59 | 328 | Opt_discard_mode, |
416a7202 DS |
329 | Opt_norecovery, |
330 | Opt_ratio, | |
331 | Opt_rescan_uuid_tree, | |
332 | Opt_skip_balance, | |
333 | Opt_space_cache, Opt_no_space_cache, | |
334 | Opt_space_cache_version, | |
335 | Opt_ssd, Opt_nossd, | |
336 | Opt_ssd_spread, Opt_nossd_spread, | |
337 | Opt_subvol, | |
37becec9 | 338 | Opt_subvol_empty, |
416a7202 DS |
339 | Opt_subvolid, |
340 | Opt_thread_pool, | |
341 | Opt_treelog, Opt_notreelog, | |
416a7202 DS |
342 | Opt_user_subvol_rm_allowed, |
343 | ||
74ef0018 QW |
344 | /* Rescue options */ |
345 | Opt_rescue, | |
346 | Opt_usebackuproot, | |
347 | Opt_nologreplay, | |
348 | ||
416a7202 | 349 | /* Deprecated options */ |
416a7202 | 350 | Opt_recovery, |
416a7202 DS |
351 | |
352 | /* Debugging options */ | |
353 | Opt_check_integrity, | |
70f6d82e | 354 | Opt_check_integrity_including_extent_data, |
416a7202 DS |
355 | Opt_check_integrity_print_mask, |
356 | Opt_enospc_debug, Opt_noenospc_debug, | |
d0bd4560 JB |
357 | #ifdef CONFIG_BTRFS_DEBUG |
358 | Opt_fragment_data, Opt_fragment_metadata, Opt_fragment_all, | |
fb592373 JB |
359 | #endif |
360 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
361 | Opt_ref_verify, | |
d0bd4560 | 362 | #endif |
9555c6c1 | 363 | Opt_err, |
95e05289 CM |
364 | }; |
365 | ||
4d4ab6d6 | 366 | static const match_table_t tokens = { |
416a7202 DS |
367 | {Opt_acl, "acl"}, |
368 | {Opt_noacl, "noacl"}, | |
369 | {Opt_clear_cache, "clear_cache"}, | |
370 | {Opt_commit_interval, "commit=%u"}, | |
c8b97818 | 371 | {Opt_compress, "compress"}, |
261507a0 | 372 | {Opt_compress_type, "compress=%s"}, |
a555f810 | 373 | {Opt_compress_force, "compress-force"}, |
261507a0 | 374 | {Opt_compress_force_type, "compress-force=%s"}, |
416a7202 DS |
375 | {Opt_degraded, "degraded"}, |
376 | {Opt_device, "device=%s"}, | |
377 | {Opt_fatal_errors, "fatal_errors=%s"}, | |
dccae999 | 378 | {Opt_flushoncommit, "flushoncommit"}, |
2c9ee856 | 379 | {Opt_noflushoncommit, "noflushoncommit"}, |
416a7202 DS |
380 | {Opt_inode_cache, "inode_cache"}, |
381 | {Opt_noinode_cache, "noinode_cache"}, | |
382 | {Opt_max_inline, "max_inline=%s"}, | |
383 | {Opt_barrier, "barrier"}, | |
384 | {Opt_nobarrier, "nobarrier"}, | |
385 | {Opt_datacow, "datacow"}, | |
386 | {Opt_nodatacow, "nodatacow"}, | |
387 | {Opt_datasum, "datasum"}, | |
388 | {Opt_nodatasum, "nodatasum"}, | |
389 | {Opt_defrag, "autodefrag"}, | |
390 | {Opt_nodefrag, "noautodefrag"}, | |
e244a0ae | 391 | {Opt_discard, "discard"}, |
b0643e59 | 392 | {Opt_discard_mode, "discard=%s"}, |
e07a2ade | 393 | {Opt_nodiscard, "nodiscard"}, |
416a7202 DS |
394 | {Opt_norecovery, "norecovery"}, |
395 | {Opt_ratio, "metadata_ratio=%u"}, | |
396 | {Opt_rescan_uuid_tree, "rescan_uuid_tree"}, | |
397 | {Opt_skip_balance, "skip_balance"}, | |
0af3d00b | 398 | {Opt_space_cache, "space_cache"}, |
8965593e | 399 | {Opt_no_space_cache, "nospace_cache"}, |
416a7202 DS |
400 | {Opt_space_cache_version, "space_cache=%s"}, |
401 | {Opt_ssd, "ssd"}, | |
402 | {Opt_nossd, "nossd"}, | |
403 | {Opt_ssd_spread, "ssd_spread"}, | |
404 | {Opt_nossd_spread, "nossd_spread"}, | |
405 | {Opt_subvol, "subvol=%s"}, | |
37becec9 | 406 | {Opt_subvol_empty, "subvol="}, |
416a7202 DS |
407 | {Opt_subvolid, "subvolid=%s"}, |
408 | {Opt_thread_pool, "thread_pool=%u"}, | |
409 | {Opt_treelog, "treelog"}, | |
410 | {Opt_notreelog, "notreelog"}, | |
416a7202 DS |
411 | {Opt_user_subvol_rm_allowed, "user_subvol_rm_allowed"}, |
412 | ||
74ef0018 QW |
413 | /* Rescue options */ |
414 | {Opt_rescue, "rescue=%s"}, | |
415 | /* Deprecated, with alias rescue=nologreplay */ | |
416 | {Opt_nologreplay, "nologreplay"}, | |
417 | /* Deprecated, with alias rescue=usebackuproot */ | |
418 | {Opt_usebackuproot, "usebackuproot"}, | |
419 | ||
416a7202 | 420 | /* Deprecated options */ |
416a7202 | 421 | {Opt_recovery, "recovery"}, |
416a7202 DS |
422 | |
423 | /* Debugging options */ | |
21adbd5c SB |
424 | {Opt_check_integrity, "check_int"}, |
425 | {Opt_check_integrity_including_extent_data, "check_int_data"}, | |
02453bde | 426 | {Opt_check_integrity_print_mask, "check_int_print_mask=%u"}, |
416a7202 DS |
427 | {Opt_enospc_debug, "enospc_debug"}, |
428 | {Opt_noenospc_debug, "noenospc_debug"}, | |
d0bd4560 JB |
429 | #ifdef CONFIG_BTRFS_DEBUG |
430 | {Opt_fragment_data, "fragment=data"}, | |
431 | {Opt_fragment_metadata, "fragment=metadata"}, | |
432 | {Opt_fragment_all, "fragment=all"}, | |
fb592373 JB |
433 | #endif |
434 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
435 | {Opt_ref_verify, "ref_verify"}, | |
d0bd4560 | 436 | #endif |
33268eaf | 437 | {Opt_err, NULL}, |
95e05289 CM |
438 | }; |
439 | ||
74ef0018 QW |
440 | static const match_table_t rescue_tokens = { |
441 | {Opt_usebackuproot, "usebackuproot"}, | |
442 | {Opt_nologreplay, "nologreplay"}, | |
443 | {Opt_err, NULL}, | |
444 | }; | |
445 | ||
446 | static int parse_rescue_options(struct btrfs_fs_info *info, const char *options) | |
447 | { | |
448 | char *opts; | |
449 | char *orig; | |
450 | char *p; | |
451 | substring_t args[MAX_OPT_ARGS]; | |
452 | int ret = 0; | |
453 | ||
454 | opts = kstrdup(options, GFP_KERNEL); | |
455 | if (!opts) | |
456 | return -ENOMEM; | |
457 | orig = opts; | |
458 | ||
459 | while ((p = strsep(&opts, ":")) != NULL) { | |
460 | int token; | |
461 | ||
462 | if (!*p) | |
463 | continue; | |
464 | token = match_token(p, rescue_tokens, args); | |
465 | switch (token){ | |
466 | case Opt_usebackuproot: | |
467 | btrfs_info(info, | |
468 | "trying to use backup root at mount time"); | |
469 | btrfs_set_opt(info->mount_opt, USEBACKUPROOT); | |
470 | break; | |
471 | case Opt_nologreplay: | |
472 | btrfs_set_and_info(info, NOLOGREPLAY, | |
473 | "disabling log replay at mount time"); | |
474 | break; | |
475 | case Opt_err: | |
476 | btrfs_info(info, "unrecognized rescue option '%s'", p); | |
477 | ret = -EINVAL; | |
478 | goto out; | |
479 | default: | |
480 | break; | |
481 | } | |
482 | ||
483 | } | |
484 | out: | |
485 | kfree(orig); | |
486 | return ret; | |
487 | } | |
488 | ||
edf24abe CH |
489 | /* |
490 | * Regular mount options parser. Everything that is needed only when | |
491 | * reading in a new superblock is parsed here. | |
49b25e05 | 492 | * XXX JDM: This needs to be cleaned up for remount. |
edf24abe | 493 | */ |
2ff7e61e | 494 | int btrfs_parse_options(struct btrfs_fs_info *info, char *options, |
96da0919 | 495 | unsigned long new_flags) |
95e05289 | 496 | { |
95e05289 | 497 | substring_t args[MAX_OPT_ARGS]; |
e215772c | 498 | char *p, *num; |
73bc1876 | 499 | u64 cache_gen; |
4543df7e | 500 | int intarg; |
a7a3f7ca | 501 | int ret = 0; |
261507a0 LZ |
502 | char *compress_type; |
503 | bool compress_force = false; | |
b7c47bbb TI |
504 | enum btrfs_compression_type saved_compress_type; |
505 | bool saved_compress_force; | |
506 | int no_compress = 0; | |
b6cda9bc | 507 | |
0b246afa JM |
508 | cache_gen = btrfs_super_cache_generation(info->super_copy); |
509 | if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE)) | |
70f6d82e OS |
510 | btrfs_set_opt(info->mount_opt, FREE_SPACE_TREE); |
511 | else if (cache_gen) | |
73bc1876 JB |
512 | btrfs_set_opt(info->mount_opt, SPACE_CACHE); |
513 | ||
96da0919 QW |
514 | /* |
515 | * Even the options are empty, we still need to do extra check | |
516 | * against new flags | |
517 | */ | |
95e05289 | 518 | if (!options) |
96da0919 | 519 | goto check; |
95e05289 | 520 | |
edf24abe | 521 | while ((p = strsep(&options, ",")) != NULL) { |
95e05289 CM |
522 | int token; |
523 | if (!*p) | |
524 | continue; | |
525 | ||
526 | token = match_token(p, tokens, args); | |
527 | switch (token) { | |
dfe25020 | 528 | case Opt_degraded: |
0b246afa | 529 | btrfs_info(info, "allowing degraded mounts"); |
edf24abe | 530 | btrfs_set_opt(info->mount_opt, DEGRADED); |
dfe25020 | 531 | break; |
95e05289 | 532 | case Opt_subvol: |
37becec9 | 533 | case Opt_subvol_empty: |
73f73415 | 534 | case Opt_subvolid: |
43e570b0 | 535 | case Opt_device: |
edf24abe | 536 | /* |
fa59f27c AJ |
537 | * These are parsed by btrfs_parse_subvol_options or |
538 | * btrfs_parse_device_options and can be ignored here. | |
edf24abe | 539 | */ |
b6cda9bc CM |
540 | break; |
541 | case Opt_nodatasum: | |
3cdde224 | 542 | btrfs_set_and_info(info, NODATASUM, |
07802534 | 543 | "setting nodatasum"); |
be20aa9d | 544 | break; |
d399167d | 545 | case Opt_datasum: |
3cdde224 JM |
546 | if (btrfs_test_opt(info, NODATASUM)) { |
547 | if (btrfs_test_opt(info, NODATACOW)) | |
0b246afa | 548 | btrfs_info(info, |
5d163e0e | 549 | "setting datasum, datacow enabled"); |
07802534 | 550 | else |
0b246afa | 551 | btrfs_info(info, "setting datasum"); |
07802534 | 552 | } |
d399167d QW |
553 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
554 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
555 | break; | |
be20aa9d | 556 | case Opt_nodatacow: |
3cdde224 JM |
557 | if (!btrfs_test_opt(info, NODATACOW)) { |
558 | if (!btrfs_test_opt(info, COMPRESS) || | |
559 | !btrfs_test_opt(info, FORCE_COMPRESS)) { | |
0b246afa | 560 | btrfs_info(info, |
07802534 QW |
561 | "setting nodatacow, compression disabled"); |
562 | } else { | |
0b246afa | 563 | btrfs_info(info, "setting nodatacow"); |
07802534 | 564 | } |
bedb2cca | 565 | } |
bedb2cca AP |
566 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
567 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
edf24abe CH |
568 | btrfs_set_opt(info->mount_opt, NODATACOW); |
569 | btrfs_set_opt(info->mount_opt, NODATASUM); | |
95e05289 | 570 | break; |
a258af7a | 571 | case Opt_datacow: |
3cdde224 | 572 | btrfs_clear_and_info(info, NODATACOW, |
07802534 | 573 | "setting datacow"); |
a258af7a | 574 | break; |
a555f810 | 575 | case Opt_compress_force: |
261507a0 LZ |
576 | case Opt_compress_force_type: |
577 | compress_force = true; | |
c730ae0c | 578 | fallthrough; |
261507a0 LZ |
579 | case Opt_compress: |
580 | case Opt_compress_type: | |
3cdde224 JM |
581 | saved_compress_type = btrfs_test_opt(info, |
582 | COMPRESS) ? | |
b7c47bbb TI |
583 | info->compress_type : BTRFS_COMPRESS_NONE; |
584 | saved_compress_force = | |
3cdde224 | 585 | btrfs_test_opt(info, FORCE_COMPRESS); |
261507a0 LZ |
586 | if (token == Opt_compress || |
587 | token == Opt_compress_force || | |
a7164fa4 | 588 | strncmp(args[0].from, "zlib", 4) == 0) { |
261507a0 | 589 | compress_type = "zlib"; |
eae8d825 | 590 | |
261507a0 | 591 | info->compress_type = BTRFS_COMPRESS_ZLIB; |
eae8d825 QW |
592 | info->compress_level = BTRFS_ZLIB_DEFAULT_LEVEL; |
593 | /* | |
594 | * args[0] contains uninitialized data since | |
595 | * for these tokens we don't expect any | |
596 | * parameter. | |
597 | */ | |
598 | if (token != Opt_compress && | |
599 | token != Opt_compress_force) | |
600 | info->compress_level = | |
d0ab62ce DZ |
601 | btrfs_compress_str2level( |
602 | BTRFS_COMPRESS_ZLIB, | |
603 | args[0].from + 4); | |
063849ea | 604 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
605 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
606 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
b7c47bbb | 607 | no_compress = 0; |
a7164fa4 | 608 | } else if (strncmp(args[0].from, "lzo", 3) == 0) { |
a6fa6fae LZ |
609 | compress_type = "lzo"; |
610 | info->compress_type = BTRFS_COMPRESS_LZO; | |
063849ea | 611 | btrfs_set_opt(info->mount_opt, COMPRESS); |
bedb2cca AP |
612 | btrfs_clear_opt(info->mount_opt, NODATACOW); |
613 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
2b0ce2c2 | 614 | btrfs_set_fs_incompat(info, COMPRESS_LZO); |
b7c47bbb | 615 | no_compress = 0; |
3f93aef5 | 616 | } else if (strncmp(args[0].from, "zstd", 4) == 0) { |
5c1aab1d NT |
617 | compress_type = "zstd"; |
618 | info->compress_type = BTRFS_COMPRESS_ZSTD; | |
3f93aef5 DZ |
619 | info->compress_level = |
620 | btrfs_compress_str2level( | |
621 | BTRFS_COMPRESS_ZSTD, | |
622 | args[0].from + 4); | |
5c1aab1d NT |
623 | btrfs_set_opt(info->mount_opt, COMPRESS); |
624 | btrfs_clear_opt(info->mount_opt, NODATACOW); | |
625 | btrfs_clear_opt(info->mount_opt, NODATASUM); | |
626 | btrfs_set_fs_incompat(info, COMPRESS_ZSTD); | |
627 | no_compress = 0; | |
063849ea AH |
628 | } else if (strncmp(args[0].from, "no", 2) == 0) { |
629 | compress_type = "no"; | |
063849ea AH |
630 | btrfs_clear_opt(info->mount_opt, COMPRESS); |
631 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
632 | compress_force = false; | |
b7c47bbb | 633 | no_compress++; |
261507a0 LZ |
634 | } else { |
635 | ret = -EINVAL; | |
636 | goto out; | |
637 | } | |
638 | ||
261507a0 | 639 | if (compress_force) { |
b7c47bbb | 640 | btrfs_set_opt(info->mount_opt, FORCE_COMPRESS); |
143f3636 | 641 | } else { |
4027e0f4 WS |
642 | /* |
643 | * If we remount from compress-force=xxx to | |
644 | * compress=xxx, we need clear FORCE_COMPRESS | |
645 | * flag, otherwise, there is no way for users | |
646 | * to disable forcible compression separately. | |
647 | */ | |
648 | btrfs_clear_opt(info->mount_opt, FORCE_COMPRESS); | |
a7e252af | 649 | } |
3cdde224 | 650 | if ((btrfs_test_opt(info, COMPRESS) && |
b7c47bbb TI |
651 | (info->compress_type != saved_compress_type || |
652 | compress_force != saved_compress_force)) || | |
3cdde224 | 653 | (!btrfs_test_opt(info, COMPRESS) && |
b7c47bbb | 654 | no_compress == 1)) { |
f51d2b59 | 655 | btrfs_info(info, "%s %s compression, level %d", |
b7c47bbb | 656 | (compress_force) ? "force" : "use", |
f51d2b59 | 657 | compress_type, info->compress_level); |
b7c47bbb TI |
658 | } |
659 | compress_force = false; | |
a555f810 | 660 | break; |
e18e4809 | 661 | case Opt_ssd: |
3cdde224 | 662 | btrfs_set_and_info(info, SSD, |
583b7231 | 663 | "enabling ssd optimizations"); |
951e7966 | 664 | btrfs_clear_opt(info->mount_opt, NOSSD); |
e18e4809 | 665 | break; |
451d7585 | 666 | case Opt_ssd_spread: |
583b7231 HK |
667 | btrfs_set_and_info(info, SSD, |
668 | "enabling ssd optimizations"); | |
3cdde224 | 669 | btrfs_set_and_info(info, SSD_SPREAD, |
583b7231 | 670 | "using spread ssd allocation scheme"); |
951e7966 | 671 | btrfs_clear_opt(info->mount_opt, NOSSD); |
451d7585 | 672 | break; |
3b30c22f | 673 | case Opt_nossd: |
583b7231 HK |
674 | btrfs_set_opt(info->mount_opt, NOSSD); |
675 | btrfs_clear_and_info(info, SSD, | |
676 | "not using ssd optimizations"); | |
c730ae0c | 677 | fallthrough; |
62b8e077 | 678 | case Opt_nossd_spread: |
583b7231 HK |
679 | btrfs_clear_and_info(info, SSD_SPREAD, |
680 | "not using spread ssd allocation scheme"); | |
3b30c22f | 681 | break; |
842bef58 | 682 | case Opt_barrier: |
3cdde224 | 683 | btrfs_clear_and_info(info, NOBARRIER, |
07802534 | 684 | "turning on barriers"); |
842bef58 | 685 | break; |
21ad10cf | 686 | case Opt_nobarrier: |
3cdde224 | 687 | btrfs_set_and_info(info, NOBARRIER, |
07802534 | 688 | "turning off barriers"); |
21ad10cf | 689 | break; |
4543df7e | 690 | case Opt_thread_pool: |
2c334e87 WS |
691 | ret = match_int(&args[0], &intarg); |
692 | if (ret) { | |
693 | goto out; | |
f7b885be | 694 | } else if (intarg == 0) { |
2c334e87 WS |
695 | ret = -EINVAL; |
696 | goto out; | |
697 | } | |
f7b885be | 698 | info->thread_pool_size = intarg; |
4543df7e | 699 | break; |
6f568d35 | 700 | case Opt_max_inline: |
edf24abe CH |
701 | num = match_strdup(&args[0]); |
702 | if (num) { | |
91748467 | 703 | info->max_inline = memparse(num, NULL); |
edf24abe CH |
704 | kfree(num); |
705 | ||
15ada040 | 706 | if (info->max_inline) { |
feb5f965 | 707 | info->max_inline = min_t(u64, |
15ada040 | 708 | info->max_inline, |
0b246afa | 709 | info->sectorsize); |
15ada040 | 710 | } |
0b246afa JM |
711 | btrfs_info(info, "max_inline at %llu", |
712 | info->max_inline); | |
2c334e87 WS |
713 | } else { |
714 | ret = -ENOMEM; | |
715 | goto out; | |
6f568d35 CM |
716 | } |
717 | break; | |
bd0330ad | 718 | case Opt_acl: |
45ff35d6 | 719 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
1751e8a6 | 720 | info->sb->s_flags |= SB_POSIXACL; |
bd0330ad | 721 | break; |
45ff35d6 | 722 | #else |
0b246afa | 723 | btrfs_err(info, "support for ACL not compiled in!"); |
45ff35d6 GZ |
724 | ret = -EINVAL; |
725 | goto out; | |
726 | #endif | |
33268eaf | 727 | case Opt_noacl: |
1751e8a6 | 728 | info->sb->s_flags &= ~SB_POSIXACL; |
33268eaf | 729 | break; |
3a5e1404 | 730 | case Opt_notreelog: |
3cdde224 | 731 | btrfs_set_and_info(info, NOTREELOG, |
07802534 | 732 | "disabling tree log"); |
a88998f2 QW |
733 | break; |
734 | case Opt_treelog: | |
3cdde224 | 735 | btrfs_clear_and_info(info, NOTREELOG, |
07802534 | 736 | "enabling tree log"); |
3a5e1404 | 737 | break; |
fed8f166 | 738 | case Opt_norecovery: |
96da0919 | 739 | case Opt_nologreplay: |
74ef0018 QW |
740 | btrfs_warn(info, |
741 | "'nologreplay' is deprecated, use 'rescue=nologreplay' instead"); | |
3cdde224 | 742 | btrfs_set_and_info(info, NOLOGREPLAY, |
96da0919 QW |
743 | "disabling log replay at mount time"); |
744 | break; | |
dccae999 | 745 | case Opt_flushoncommit: |
3cdde224 | 746 | btrfs_set_and_info(info, FLUSHONCOMMIT, |
07802534 | 747 | "turning on flush-on-commit"); |
dccae999 | 748 | break; |
2c9ee856 | 749 | case Opt_noflushoncommit: |
3cdde224 | 750 | btrfs_clear_and_info(info, FLUSHONCOMMIT, |
07802534 | 751 | "turning off flush-on-commit"); |
2c9ee856 | 752 | break; |
97e728d4 | 753 | case Opt_ratio: |
2c334e87 | 754 | ret = match_int(&args[0], &intarg); |
764cb8b4 | 755 | if (ret) |
2c334e87 | 756 | goto out; |
764cb8b4 AJ |
757 | info->metadata_ratio = intarg; |
758 | btrfs_info(info, "metadata ratio %u", | |
759 | info->metadata_ratio); | |
97e728d4 | 760 | break; |
e244a0ae | 761 | case Opt_discard: |
b0643e59 DZ |
762 | case Opt_discard_mode: |
763 | if (token == Opt_discard || | |
764 | strcmp(args[0].from, "sync") == 0) { | |
765 | btrfs_clear_opt(info->mount_opt, DISCARD_ASYNC); | |
766 | btrfs_set_and_info(info, DISCARD_SYNC, | |
767 | "turning on sync discard"); | |
768 | } else if (strcmp(args[0].from, "async") == 0) { | |
769 | btrfs_clear_opt(info->mount_opt, DISCARD_SYNC); | |
770 | btrfs_set_and_info(info, DISCARD_ASYNC, | |
771 | "turning on async discard"); | |
772 | } else { | |
773 | ret = -EINVAL; | |
774 | goto out; | |
775 | } | |
e244a0ae | 776 | break; |
e07a2ade | 777 | case Opt_nodiscard: |
46b27f50 | 778 | btrfs_clear_and_info(info, DISCARD_SYNC, |
07802534 | 779 | "turning off discard"); |
b0643e59 DZ |
780 | btrfs_clear_and_info(info, DISCARD_ASYNC, |
781 | "turning off async discard"); | |
e07a2ade | 782 | break; |
0af3d00b | 783 | case Opt_space_cache: |
70f6d82e OS |
784 | case Opt_space_cache_version: |
785 | if (token == Opt_space_cache || | |
786 | strcmp(args[0].from, "v1") == 0) { | |
0b246afa | 787 | btrfs_clear_opt(info->mount_opt, |
70f6d82e | 788 | FREE_SPACE_TREE); |
3cdde224 | 789 | btrfs_set_and_info(info, SPACE_CACHE, |
0b246afa | 790 | "enabling disk space caching"); |
70f6d82e | 791 | } else if (strcmp(args[0].from, "v2") == 0) { |
0b246afa | 792 | btrfs_clear_opt(info->mount_opt, |
70f6d82e | 793 | SPACE_CACHE); |
0b246afa | 794 | btrfs_set_and_info(info, FREE_SPACE_TREE, |
70f6d82e OS |
795 | "enabling free space tree"); |
796 | } else { | |
797 | ret = -EINVAL; | |
798 | goto out; | |
799 | } | |
0de90876 | 800 | break; |
f420ee1e SB |
801 | case Opt_rescan_uuid_tree: |
802 | btrfs_set_opt(info->mount_opt, RESCAN_UUID_TREE); | |
803 | break; | |
73bc1876 | 804 | case Opt_no_space_cache: |
3cdde224 | 805 | if (btrfs_test_opt(info, SPACE_CACHE)) { |
0b246afa JM |
806 | btrfs_clear_and_info(info, SPACE_CACHE, |
807 | "disabling disk space caching"); | |
70f6d82e | 808 | } |
3cdde224 | 809 | if (btrfs_test_opt(info, FREE_SPACE_TREE)) { |
0b246afa JM |
810 | btrfs_clear_and_info(info, FREE_SPACE_TREE, |
811 | "disabling free space tree"); | |
70f6d82e | 812 | } |
73bc1876 | 813 | break; |
4b9465cb | 814 | case Opt_inode_cache: |
b547a88e DS |
815 | btrfs_warn(info, |
816 | "the 'inode_cache' option is deprecated and will have no effect from 5.11"); | |
7e1876ac | 817 | btrfs_set_pending_and_info(info, INODE_MAP_CACHE, |
07802534 | 818 | "enabling inode map caching"); |
3818aea2 QW |
819 | break; |
820 | case Opt_noinode_cache: | |
7e1876ac | 821 | btrfs_clear_pending_and_info(info, INODE_MAP_CACHE, |
07802534 | 822 | "disabling inode map caching"); |
4b9465cb | 823 | break; |
88c2ba3b | 824 | case Opt_clear_cache: |
3cdde224 | 825 | btrfs_set_and_info(info, CLEAR_CACHE, |
07802534 | 826 | "force clearing of disk cache"); |
0af3d00b | 827 | break; |
4260f7c7 SW |
828 | case Opt_user_subvol_rm_allowed: |
829 | btrfs_set_opt(info->mount_opt, USER_SUBVOL_RM_ALLOWED); | |
830 | break; | |
91435650 CM |
831 | case Opt_enospc_debug: |
832 | btrfs_set_opt(info->mount_opt, ENOSPC_DEBUG); | |
833 | break; | |
53036293 QW |
834 | case Opt_noenospc_debug: |
835 | btrfs_clear_opt(info->mount_opt, ENOSPC_DEBUG); | |
836 | break; | |
4cb5300b | 837 | case Opt_defrag: |
3cdde224 | 838 | btrfs_set_and_info(info, AUTO_DEFRAG, |
07802534 | 839 | "enabling auto defrag"); |
4cb5300b | 840 | break; |
fc0ca9af | 841 | case Opt_nodefrag: |
3cdde224 | 842 | btrfs_clear_and_info(info, AUTO_DEFRAG, |
07802534 | 843 | "disabling auto defrag"); |
fc0ca9af | 844 | break; |
af31f5e5 | 845 | case Opt_recovery: |
8dcddfa0 | 846 | case Opt_usebackuproot: |
74ef0018 QW |
847 | btrfs_warn(info, |
848 | "'%s' is deprecated, use 'rescue=usebackuproot' instead", | |
849 | token == Opt_recovery ? "recovery" : | |
850 | "usebackuproot"); | |
0b246afa | 851 | btrfs_info(info, |
8dcddfa0 QW |
852 | "trying to use backup root at mount time"); |
853 | btrfs_set_opt(info->mount_opt, USEBACKUPROOT); | |
af31f5e5 | 854 | break; |
9555c6c1 ID |
855 | case Opt_skip_balance: |
856 | btrfs_set_opt(info->mount_opt, SKIP_BALANCE); | |
857 | break; | |
21adbd5c SB |
858 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
859 | case Opt_check_integrity_including_extent_data: | |
0b246afa | 860 | btrfs_info(info, |
efe120a0 | 861 | "enabling check integrity including extent data"); |
21adbd5c SB |
862 | btrfs_set_opt(info->mount_opt, |
863 | CHECK_INTEGRITY_INCLUDING_EXTENT_DATA); | |
864 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); | |
865 | break; | |
866 | case Opt_check_integrity: | |
0b246afa | 867 | btrfs_info(info, "enabling check integrity"); |
21adbd5c SB |
868 | btrfs_set_opt(info->mount_opt, CHECK_INTEGRITY); |
869 | break; | |
870 | case Opt_check_integrity_print_mask: | |
2c334e87 | 871 | ret = match_int(&args[0], &intarg); |
02453bde | 872 | if (ret) |
2c334e87 | 873 | goto out; |
02453bde AJ |
874 | info->check_integrity_print_mask = intarg; |
875 | btrfs_info(info, "check_integrity_print_mask 0x%x", | |
876 | info->check_integrity_print_mask); | |
21adbd5c SB |
877 | break; |
878 | #else | |
879 | case Opt_check_integrity_including_extent_data: | |
880 | case Opt_check_integrity: | |
881 | case Opt_check_integrity_print_mask: | |
0b246afa JM |
882 | btrfs_err(info, |
883 | "support for check_integrity* not compiled in!"); | |
21adbd5c SB |
884 | ret = -EINVAL; |
885 | goto out; | |
886 | #endif | |
8c342930 JM |
887 | case Opt_fatal_errors: |
888 | if (strcmp(args[0].from, "panic") == 0) | |
889 | btrfs_set_opt(info->mount_opt, | |
890 | PANIC_ON_FATAL_ERROR); | |
891 | else if (strcmp(args[0].from, "bug") == 0) | |
892 | btrfs_clear_opt(info->mount_opt, | |
893 | PANIC_ON_FATAL_ERROR); | |
894 | else { | |
895 | ret = -EINVAL; | |
896 | goto out; | |
897 | } | |
898 | break; | |
8b87dc17 DS |
899 | case Opt_commit_interval: |
900 | intarg = 0; | |
901 | ret = match_int(&args[0], &intarg); | |
d3740608 | 902 | if (ret) |
8b87dc17 | 903 | goto out; |
d3740608 | 904 | if (intarg == 0) { |
0b246afa | 905 | btrfs_info(info, |
d3740608 | 906 | "using default commit interval %us", |
5d163e0e | 907 | BTRFS_DEFAULT_COMMIT_INTERVAL); |
d3740608 AJ |
908 | intarg = BTRFS_DEFAULT_COMMIT_INTERVAL; |
909 | } else if (intarg > 300) { | |
910 | btrfs_warn(info, "excessive commit interval %d", | |
911 | intarg); | |
8b87dc17 | 912 | } |
d3740608 | 913 | info->commit_interval = intarg; |
8b87dc17 | 914 | break; |
74ef0018 QW |
915 | case Opt_rescue: |
916 | ret = parse_rescue_options(info, args[0].from); | |
917 | if (ret < 0) | |
918 | goto out; | |
919 | break; | |
d0bd4560 JB |
920 | #ifdef CONFIG_BTRFS_DEBUG |
921 | case Opt_fragment_all: | |
0b246afa | 922 | btrfs_info(info, "fragmenting all space"); |
d0bd4560 JB |
923 | btrfs_set_opt(info->mount_opt, FRAGMENT_DATA); |
924 | btrfs_set_opt(info->mount_opt, FRAGMENT_METADATA); | |
925 | break; | |
926 | case Opt_fragment_metadata: | |
0b246afa | 927 | btrfs_info(info, "fragmenting metadata"); |
d0bd4560 JB |
928 | btrfs_set_opt(info->mount_opt, |
929 | FRAGMENT_METADATA); | |
930 | break; | |
931 | case Opt_fragment_data: | |
0b246afa | 932 | btrfs_info(info, "fragmenting data"); |
d0bd4560 JB |
933 | btrfs_set_opt(info->mount_opt, FRAGMENT_DATA); |
934 | break; | |
fb592373 JB |
935 | #endif |
936 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
937 | case Opt_ref_verify: | |
938 | btrfs_info(info, "doing ref verification"); | |
939 | btrfs_set_opt(info->mount_opt, REF_VERIFY); | |
940 | break; | |
d0bd4560 | 941 | #endif |
a7a3f7ca | 942 | case Opt_err: |
7e8f19e5 | 943 | btrfs_err(info, "unrecognized mount option '%s'", p); |
a7a3f7ca SW |
944 | ret = -EINVAL; |
945 | goto out; | |
95e05289 | 946 | default: |
be20aa9d | 947 | break; |
95e05289 CM |
948 | } |
949 | } | |
96da0919 QW |
950 | check: |
951 | /* | |
952 | * Extra check for current option against current flag | |
953 | */ | |
1751e8a6 | 954 | if (btrfs_test_opt(info, NOLOGREPLAY) && !(new_flags & SB_RDONLY)) { |
0b246afa | 955 | btrfs_err(info, |
96da0919 QW |
956 | "nologreplay must be used with ro mount option"); |
957 | ret = -EINVAL; | |
958 | } | |
a7a3f7ca | 959 | out: |
0b246afa | 960 | if (btrfs_fs_compat_ro(info, FREE_SPACE_TREE) && |
3cdde224 JM |
961 | !btrfs_test_opt(info, FREE_SPACE_TREE) && |
962 | !btrfs_test_opt(info, CLEAR_CACHE)) { | |
0b246afa | 963 | btrfs_err(info, "cannot disable free space tree"); |
70f6d82e OS |
964 | ret = -EINVAL; |
965 | ||
966 | } | |
3cdde224 | 967 | if (!ret && btrfs_test_opt(info, SPACE_CACHE)) |
0b246afa | 968 | btrfs_info(info, "disk space caching is enabled"); |
3cdde224 | 969 | if (!ret && btrfs_test_opt(info, FREE_SPACE_TREE)) |
0b246afa | 970 | btrfs_info(info, "using free space tree"); |
a7a3f7ca | 971 | return ret; |
edf24abe CH |
972 | } |
973 | ||
974 | /* | |
975 | * Parse mount options that are required early in the mount process. | |
976 | * | |
977 | * All other options will be parsed on much later in the mount process and | |
978 | * only when we need to allocate a new super block. | |
979 | */ | |
fa59f27c AJ |
980 | static int btrfs_parse_device_options(const char *options, fmode_t flags, |
981 | void *holder) | |
edf24abe CH |
982 | { |
983 | substring_t args[MAX_OPT_ARGS]; | |
83c8c9bd | 984 | char *device_name, *opts, *orig, *p; |
36350e95 | 985 | struct btrfs_device *device = NULL; |
d7407606 MT |
986 | int error = 0; |
987 | ||
5139cff5 DS |
988 | lockdep_assert_held(&uuid_mutex); |
989 | ||
d7407606 MT |
990 | if (!options) |
991 | return 0; | |
992 | ||
993 | /* | |
994 | * strsep changes the string, duplicate it because btrfs_parse_options | |
995 | * gets called later | |
996 | */ | |
997 | opts = kstrdup(options, GFP_KERNEL); | |
998 | if (!opts) | |
999 | return -ENOMEM; | |
1000 | orig = opts; | |
1001 | ||
1002 | while ((p = strsep(&opts, ",")) != NULL) { | |
1003 | int token; | |
1004 | ||
1005 | if (!*p) | |
1006 | continue; | |
1007 | ||
1008 | token = match_token(p, tokens, args); | |
1009 | if (token == Opt_device) { | |
1010 | device_name = match_strdup(&args[0]); | |
1011 | if (!device_name) { | |
1012 | error = -ENOMEM; | |
1013 | goto out; | |
1014 | } | |
36350e95 GJ |
1015 | device = btrfs_scan_one_device(device_name, flags, |
1016 | holder); | |
d7407606 | 1017 | kfree(device_name); |
36350e95 GJ |
1018 | if (IS_ERR(device)) { |
1019 | error = PTR_ERR(device); | |
d7407606 | 1020 | goto out; |
36350e95 | 1021 | } |
d7407606 MT |
1022 | } |
1023 | } | |
1024 | ||
1025 | out: | |
1026 | kfree(orig); | |
1027 | return error; | |
1028 | } | |
1029 | ||
1030 | /* | |
1031 | * Parse mount options that are related to subvolume id | |
1032 | * | |
1033 | * The value is later passed to mount_subvol() | |
1034 | */ | |
93b9bcdf GJ |
1035 | static int btrfs_parse_subvol_options(const char *options, char **subvol_name, |
1036 | u64 *subvol_objectid) | |
d7407606 MT |
1037 | { |
1038 | substring_t args[MAX_OPT_ARGS]; | |
1039 | char *opts, *orig, *p; | |
edf24abe | 1040 | int error = 0; |
ccb0e7d1 | 1041 | u64 subvolid; |
edf24abe CH |
1042 | |
1043 | if (!options) | |
830c4adb | 1044 | return 0; |
edf24abe CH |
1045 | |
1046 | /* | |
d7407606 | 1047 | * strsep changes the string, duplicate it because |
fa59f27c | 1048 | * btrfs_parse_device_options gets called later |
edf24abe CH |
1049 | */ |
1050 | opts = kstrdup(options, GFP_KERNEL); | |
1051 | if (!opts) | |
1052 | return -ENOMEM; | |
3f3d0bc0 | 1053 | orig = opts; |
edf24abe CH |
1054 | |
1055 | while ((p = strsep(&opts, ",")) != NULL) { | |
1056 | int token; | |
1057 | if (!*p) | |
1058 | continue; | |
1059 | ||
1060 | token = match_token(p, tokens, args); | |
1061 | switch (token) { | |
1062 | case Opt_subvol: | |
a90e8b6f | 1063 | kfree(*subvol_name); |
edf24abe | 1064 | *subvol_name = match_strdup(&args[0]); |
2c334e87 WS |
1065 | if (!*subvol_name) { |
1066 | error = -ENOMEM; | |
1067 | goto out; | |
1068 | } | |
edf24abe | 1069 | break; |
73f73415 | 1070 | case Opt_subvolid: |
ccb0e7d1 AJ |
1071 | error = match_u64(&args[0], &subvolid); |
1072 | if (error) | |
2c334e87 | 1073 | goto out; |
ccb0e7d1 AJ |
1074 | |
1075 | /* we want the original fs_tree */ | |
1076 | if (subvolid == 0) | |
1077 | subvolid = BTRFS_FS_TREE_OBJECTID; | |
1078 | ||
1079 | *subvol_objectid = subvolid; | |
73f73415 | 1080 | break; |
edf24abe CH |
1081 | default: |
1082 | break; | |
1083 | } | |
1084 | } | |
1085 | ||
830c4adb | 1086 | out: |
3f3d0bc0 | 1087 | kfree(orig); |
edf24abe | 1088 | return error; |
95e05289 CM |
1089 | } |
1090 | ||
c0c907a4 MPS |
1091 | char *btrfs_get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info, |
1092 | u64 subvol_objectid) | |
73f73415 | 1093 | { |
815745cf | 1094 | struct btrfs_root *root = fs_info->tree_root; |
5168489a | 1095 | struct btrfs_root *fs_root = NULL; |
05dbe683 OS |
1096 | struct btrfs_root_ref *root_ref; |
1097 | struct btrfs_inode_ref *inode_ref; | |
1098 | struct btrfs_key key; | |
1099 | struct btrfs_path *path = NULL; | |
1100 | char *name = NULL, *ptr; | |
1101 | u64 dirid; | |
1102 | int len; | |
1103 | int ret; | |
1104 | ||
1105 | path = btrfs_alloc_path(); | |
1106 | if (!path) { | |
1107 | ret = -ENOMEM; | |
1108 | goto err; | |
1109 | } | |
1110 | path->leave_spinning = 1; | |
1111 | ||
3ec83621 | 1112 | name = kmalloc(PATH_MAX, GFP_KERNEL); |
05dbe683 OS |
1113 | if (!name) { |
1114 | ret = -ENOMEM; | |
1115 | goto err; | |
1116 | } | |
1117 | ptr = name + PATH_MAX - 1; | |
1118 | ptr[0] = '\0'; | |
73f73415 JB |
1119 | |
1120 | /* | |
05dbe683 OS |
1121 | * Walk up the subvolume trees in the tree of tree roots by root |
1122 | * backrefs until we hit the top-level subvolume. | |
73f73415 | 1123 | */ |
05dbe683 OS |
1124 | while (subvol_objectid != BTRFS_FS_TREE_OBJECTID) { |
1125 | key.objectid = subvol_objectid; | |
1126 | key.type = BTRFS_ROOT_BACKREF_KEY; | |
1127 | key.offset = (u64)-1; | |
1128 | ||
1129 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
1130 | if (ret < 0) { | |
1131 | goto err; | |
1132 | } else if (ret > 0) { | |
1133 | ret = btrfs_previous_item(root, path, subvol_objectid, | |
1134 | BTRFS_ROOT_BACKREF_KEY); | |
1135 | if (ret < 0) { | |
1136 | goto err; | |
1137 | } else if (ret > 0) { | |
1138 | ret = -ENOENT; | |
1139 | goto err; | |
1140 | } | |
1141 | } | |
1142 | ||
1143 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1144 | subvol_objectid = key.offset; | |
1145 | ||
1146 | root_ref = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
1147 | struct btrfs_root_ref); | |
1148 | len = btrfs_root_ref_name_len(path->nodes[0], root_ref); | |
1149 | ptr -= len + 1; | |
1150 | if (ptr < name) { | |
1151 | ret = -ENAMETOOLONG; | |
1152 | goto err; | |
1153 | } | |
1154 | read_extent_buffer(path->nodes[0], ptr + 1, | |
1155 | (unsigned long)(root_ref + 1), len); | |
1156 | ptr[0] = '/'; | |
1157 | dirid = btrfs_root_ref_dirid(path->nodes[0], root_ref); | |
1158 | btrfs_release_path(path); | |
1159 | ||
56e9357a | 1160 | fs_root = btrfs_get_fs_root(fs_info, subvol_objectid, true); |
05dbe683 OS |
1161 | if (IS_ERR(fs_root)) { |
1162 | ret = PTR_ERR(fs_root); | |
5168489a JB |
1163 | fs_root = NULL; |
1164 | goto err; | |
1165 | } | |
05dbe683 OS |
1166 | |
1167 | /* | |
1168 | * Walk up the filesystem tree by inode refs until we hit the | |
1169 | * root directory. | |
1170 | */ | |
1171 | while (dirid != BTRFS_FIRST_FREE_OBJECTID) { | |
1172 | key.objectid = dirid; | |
1173 | key.type = BTRFS_INODE_REF_KEY; | |
1174 | key.offset = (u64)-1; | |
1175 | ||
1176 | ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0); | |
1177 | if (ret < 0) { | |
1178 | goto err; | |
1179 | } else if (ret > 0) { | |
1180 | ret = btrfs_previous_item(fs_root, path, dirid, | |
1181 | BTRFS_INODE_REF_KEY); | |
1182 | if (ret < 0) { | |
1183 | goto err; | |
1184 | } else if (ret > 0) { | |
1185 | ret = -ENOENT; | |
1186 | goto err; | |
1187 | } | |
1188 | } | |
1189 | ||
1190 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1191 | dirid = key.offset; | |
1192 | ||
1193 | inode_ref = btrfs_item_ptr(path->nodes[0], | |
1194 | path->slots[0], | |
1195 | struct btrfs_inode_ref); | |
1196 | len = btrfs_inode_ref_name_len(path->nodes[0], | |
1197 | inode_ref); | |
1198 | ptr -= len + 1; | |
1199 | if (ptr < name) { | |
1200 | ret = -ENAMETOOLONG; | |
1201 | goto err; | |
1202 | } | |
1203 | read_extent_buffer(path->nodes[0], ptr + 1, | |
1204 | (unsigned long)(inode_ref + 1), len); | |
1205 | ptr[0] = '/'; | |
1206 | btrfs_release_path(path); | |
1207 | } | |
00246528 | 1208 | btrfs_put_root(fs_root); |
5168489a | 1209 | fs_root = NULL; |
73f73415 JB |
1210 | } |
1211 | ||
05dbe683 OS |
1212 | btrfs_free_path(path); |
1213 | if (ptr == name + PATH_MAX - 1) { | |
1214 | name[0] = '/'; | |
1215 | name[1] = '\0'; | |
1216 | } else { | |
1217 | memmove(name, ptr, name + PATH_MAX - ptr); | |
1218 | } | |
1219 | return name; | |
1220 | ||
1221 | err: | |
00246528 | 1222 | btrfs_put_root(fs_root); |
05dbe683 OS |
1223 | btrfs_free_path(path); |
1224 | kfree(name); | |
1225 | return ERR_PTR(ret); | |
1226 | } | |
1227 | ||
1228 | static int get_default_subvol_objectid(struct btrfs_fs_info *fs_info, u64 *objectid) | |
1229 | { | |
1230 | struct btrfs_root *root = fs_info->tree_root; | |
1231 | struct btrfs_dir_item *di; | |
1232 | struct btrfs_path *path; | |
1233 | struct btrfs_key location; | |
1234 | u64 dir_id; | |
1235 | ||
73f73415 JB |
1236 | path = btrfs_alloc_path(); |
1237 | if (!path) | |
05dbe683 | 1238 | return -ENOMEM; |
73f73415 JB |
1239 | path->leave_spinning = 1; |
1240 | ||
1241 | /* | |
1242 | * Find the "default" dir item which points to the root item that we | |
1243 | * will mount by default if we haven't been given a specific subvolume | |
1244 | * to mount. | |
1245 | */ | |
815745cf | 1246 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
73f73415 | 1247 | di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0); |
b0839166 JL |
1248 | if (IS_ERR(di)) { |
1249 | btrfs_free_path(path); | |
05dbe683 | 1250 | return PTR_ERR(di); |
b0839166 | 1251 | } |
73f73415 JB |
1252 | if (!di) { |
1253 | /* | |
1254 | * Ok the default dir item isn't there. This is weird since | |
1255 | * it's always been there, but don't freak out, just try and | |
05dbe683 | 1256 | * mount the top-level subvolume. |
73f73415 JB |
1257 | */ |
1258 | btrfs_free_path(path); | |
05dbe683 OS |
1259 | *objectid = BTRFS_FS_TREE_OBJECTID; |
1260 | return 0; | |
73f73415 JB |
1261 | } |
1262 | ||
1263 | btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); | |
1264 | btrfs_free_path(path); | |
05dbe683 OS |
1265 | *objectid = location.objectid; |
1266 | return 0; | |
73f73415 JB |
1267 | } |
1268 | ||
d397712b | 1269 | static int btrfs_fill_super(struct super_block *sb, |
8a4b83cc | 1270 | struct btrfs_fs_devices *fs_devices, |
56e033a7 | 1271 | void *data) |
75dfe396 | 1272 | { |
d397712b | 1273 | struct inode *inode; |
815745cf | 1274 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
39279cc3 | 1275 | int err; |
a429e513 | 1276 | |
39279cc3 CM |
1277 | sb->s_maxbytes = MAX_LFS_FILESIZE; |
1278 | sb->s_magic = BTRFS_SUPER_MAGIC; | |
1279 | sb->s_op = &btrfs_super_ops; | |
af53d29a | 1280 | sb->s_d_op = &btrfs_dentry_operations; |
be6e8dc0 | 1281 | sb->s_export_op = &btrfs_export_ops; |
5103e947 | 1282 | sb->s_xattr = btrfs_xattr_handlers; |
39279cc3 | 1283 | sb->s_time_gran = 1; |
0eda294d | 1284 | #ifdef CONFIG_BTRFS_FS_POSIX_ACL |
1751e8a6 | 1285 | sb->s_flags |= SB_POSIXACL; |
49cf6f45 | 1286 | #endif |
357fdad0 | 1287 | sb->s_flags |= SB_I_VERSION; |
da2f0f74 | 1288 | sb->s_iflags |= SB_I_CGROUPWB; |
9e11ceee JK |
1289 | |
1290 | err = super_setup_bdi(sb); | |
1291 | if (err) { | |
1292 | btrfs_err(fs_info, "super_setup_bdi failed"); | |
1293 | return err; | |
1294 | } | |
1295 | ||
ad2b2c80 AV |
1296 | err = open_ctree(sb, fs_devices, (char *)data); |
1297 | if (err) { | |
ab8d0fc4 | 1298 | btrfs_err(fs_info, "open_ctree failed"); |
ad2b2c80 | 1299 | return err; |
a429e513 CM |
1300 | } |
1301 | ||
0202e83f | 1302 | inode = btrfs_iget(sb, BTRFS_FIRST_FREE_OBJECTID, fs_info->fs_root); |
5d4f98a2 YZ |
1303 | if (IS_ERR(inode)) { |
1304 | err = PTR_ERR(inode); | |
39279cc3 | 1305 | goto fail_close; |
f254e52c | 1306 | } |
f254e52c | 1307 | |
48fde701 AV |
1308 | sb->s_root = d_make_root(inode); |
1309 | if (!sb->s_root) { | |
39279cc3 CM |
1310 | err = -ENOMEM; |
1311 | goto fail_close; | |
f254e52c | 1312 | } |
58176a96 | 1313 | |
90a887c9 | 1314 | cleancache_init_fs(sb); |
1751e8a6 | 1315 | sb->s_flags |= SB_ACTIVE; |
2619ba1f | 1316 | return 0; |
39279cc3 CM |
1317 | |
1318 | fail_close: | |
6bccf3ab | 1319 | close_ctree(fs_info); |
39279cc3 | 1320 | return err; |
2619ba1f CM |
1321 | } |
1322 | ||
6bf13c0c | 1323 | int btrfs_sync_fs(struct super_block *sb, int wait) |
c5739bba CM |
1324 | { |
1325 | struct btrfs_trans_handle *trans; | |
815745cf AV |
1326 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
1327 | struct btrfs_root *root = fs_info->tree_root; | |
2619ba1f | 1328 | |
bc074524 | 1329 | trace_btrfs_sync_fs(fs_info, wait); |
1abe9b8a | 1330 | |
39279cc3 | 1331 | if (!wait) { |
815745cf | 1332 | filemap_flush(fs_info->btree_inode->i_mapping); |
39279cc3 CM |
1333 | return 0; |
1334 | } | |
771ed689 | 1335 | |
6374e57a | 1336 | btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); |
771ed689 | 1337 | |
d4edf39b | 1338 | trans = btrfs_attach_transaction_barrier(root); |
60376ce4 | 1339 | if (IS_ERR(trans)) { |
354aa0fb | 1340 | /* no transaction, don't bother */ |
6b5fe46d DS |
1341 | if (PTR_ERR(trans) == -ENOENT) { |
1342 | /* | |
1343 | * Exit unless we have some pending changes | |
1344 | * that need to go through commit | |
1345 | */ | |
1346 | if (fs_info->pending_changes == 0) | |
1347 | return 0; | |
a53f4f8e QW |
1348 | /* |
1349 | * A non-blocking test if the fs is frozen. We must not | |
1350 | * start a new transaction here otherwise a deadlock | |
1351 | * happens. The pending operations are delayed to the | |
1352 | * next commit after thawing. | |
1353 | */ | |
a7e3c5f2 RP |
1354 | if (sb_start_write_trylock(sb)) |
1355 | sb_end_write(sb); | |
a53f4f8e QW |
1356 | else |
1357 | return 0; | |
6b5fe46d | 1358 | trans = btrfs_start_transaction(root, 0); |
6b5fe46d | 1359 | } |
98bd5c54 DS |
1360 | if (IS_ERR(trans)) |
1361 | return PTR_ERR(trans); | |
60376ce4 | 1362 | } |
3a45bb20 | 1363 | return btrfs_commit_transaction(trans); |
2c90e5d6 CM |
1364 | } |
1365 | ||
34c80b1d | 1366 | static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry) |
a9572a15 | 1367 | { |
815745cf | 1368 | struct btrfs_fs_info *info = btrfs_sb(dentry->d_sb); |
0f628c63 | 1369 | const char *compress_type; |
a9572a15 | 1370 | |
3cdde224 | 1371 | if (btrfs_test_opt(info, DEGRADED)) |
a9572a15 | 1372 | seq_puts(seq, ",degraded"); |
3cdde224 | 1373 | if (btrfs_test_opt(info, NODATASUM)) |
a9572a15 | 1374 | seq_puts(seq, ",nodatasum"); |
3cdde224 | 1375 | if (btrfs_test_opt(info, NODATACOW)) |
a9572a15 | 1376 | seq_puts(seq, ",nodatacow"); |
3cdde224 | 1377 | if (btrfs_test_opt(info, NOBARRIER)) |
a9572a15 | 1378 | seq_puts(seq, ",nobarrier"); |
95ac567a | 1379 | if (info->max_inline != BTRFS_DEFAULT_MAX_INLINE) |
c1c9ff7c | 1380 | seq_printf(seq, ",max_inline=%llu", info->max_inline); |
a9572a15 EP |
1381 | if (info->thread_pool_size != min_t(unsigned long, |
1382 | num_online_cpus() + 2, 8)) | |
f7b885be | 1383 | seq_printf(seq, ",thread_pool=%u", info->thread_pool_size); |
3cdde224 | 1384 | if (btrfs_test_opt(info, COMPRESS)) { |
0f628c63 | 1385 | compress_type = btrfs_compress_type2str(info->compress_type); |
3cdde224 | 1386 | if (btrfs_test_opt(info, FORCE_COMPRESS)) |
200da64e TI |
1387 | seq_printf(seq, ",compress-force=%s", compress_type); |
1388 | else | |
1389 | seq_printf(seq, ",compress=%s", compress_type); | |
f51d2b59 | 1390 | if (info->compress_level) |
fa4d885a | 1391 | seq_printf(seq, ":%d", info->compress_level); |
200da64e | 1392 | } |
3cdde224 | 1393 | if (btrfs_test_opt(info, NOSSD)) |
c289811c | 1394 | seq_puts(seq, ",nossd"); |
3cdde224 | 1395 | if (btrfs_test_opt(info, SSD_SPREAD)) |
451d7585 | 1396 | seq_puts(seq, ",ssd_spread"); |
3cdde224 | 1397 | else if (btrfs_test_opt(info, SSD)) |
a9572a15 | 1398 | seq_puts(seq, ",ssd"); |
3cdde224 | 1399 | if (btrfs_test_opt(info, NOTREELOG)) |
6b65c5c6 | 1400 | seq_puts(seq, ",notreelog"); |
3cdde224 | 1401 | if (btrfs_test_opt(info, NOLOGREPLAY)) |
74ef0018 | 1402 | seq_puts(seq, ",rescue=nologreplay"); |
3cdde224 | 1403 | if (btrfs_test_opt(info, FLUSHONCOMMIT)) |
6b65c5c6 | 1404 | seq_puts(seq, ",flushoncommit"); |
46b27f50 | 1405 | if (btrfs_test_opt(info, DISCARD_SYNC)) |
20a5239a | 1406 | seq_puts(seq, ",discard"); |
b0643e59 DZ |
1407 | if (btrfs_test_opt(info, DISCARD_ASYNC)) |
1408 | seq_puts(seq, ",discard=async"); | |
1751e8a6 | 1409 | if (!(info->sb->s_flags & SB_POSIXACL)) |
a9572a15 | 1410 | seq_puts(seq, ",noacl"); |
3cdde224 | 1411 | if (btrfs_test_opt(info, SPACE_CACHE)) |
200da64e | 1412 | seq_puts(seq, ",space_cache"); |
3cdde224 | 1413 | else if (btrfs_test_opt(info, FREE_SPACE_TREE)) |
70f6d82e | 1414 | seq_puts(seq, ",space_cache=v2"); |
73bc1876 | 1415 | else |
8965593e | 1416 | seq_puts(seq, ",nospace_cache"); |
3cdde224 | 1417 | if (btrfs_test_opt(info, RESCAN_UUID_TREE)) |
f420ee1e | 1418 | seq_puts(seq, ",rescan_uuid_tree"); |
3cdde224 | 1419 | if (btrfs_test_opt(info, CLEAR_CACHE)) |
200da64e | 1420 | seq_puts(seq, ",clear_cache"); |
3cdde224 | 1421 | if (btrfs_test_opt(info, USER_SUBVOL_RM_ALLOWED)) |
200da64e | 1422 | seq_puts(seq, ",user_subvol_rm_allowed"); |
3cdde224 | 1423 | if (btrfs_test_opt(info, ENOSPC_DEBUG)) |
0942caa3 | 1424 | seq_puts(seq, ",enospc_debug"); |
3cdde224 | 1425 | if (btrfs_test_opt(info, AUTO_DEFRAG)) |
0942caa3 | 1426 | seq_puts(seq, ",autodefrag"); |
3cdde224 | 1427 | if (btrfs_test_opt(info, INODE_MAP_CACHE)) |
0942caa3 | 1428 | seq_puts(seq, ",inode_cache"); |
3cdde224 | 1429 | if (btrfs_test_opt(info, SKIP_BALANCE)) |
9555c6c1 | 1430 | seq_puts(seq, ",skip_balance"); |
8507d216 | 1431 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
3cdde224 | 1432 | if (btrfs_test_opt(info, CHECK_INTEGRITY_INCLUDING_EXTENT_DATA)) |
8507d216 | 1433 | seq_puts(seq, ",check_int_data"); |
3cdde224 | 1434 | else if (btrfs_test_opt(info, CHECK_INTEGRITY)) |
8507d216 WS |
1435 | seq_puts(seq, ",check_int"); |
1436 | if (info->check_integrity_print_mask) | |
1437 | seq_printf(seq, ",check_int_print_mask=%d", | |
1438 | info->check_integrity_print_mask); | |
1439 | #endif | |
1440 | if (info->metadata_ratio) | |
764cb8b4 | 1441 | seq_printf(seq, ",metadata_ratio=%u", info->metadata_ratio); |
3cdde224 | 1442 | if (btrfs_test_opt(info, PANIC_ON_FATAL_ERROR)) |
8c342930 | 1443 | seq_puts(seq, ",fatal_errors=panic"); |
8b87dc17 | 1444 | if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL) |
d3740608 | 1445 | seq_printf(seq, ",commit=%u", info->commit_interval); |
d0bd4560 | 1446 | #ifdef CONFIG_BTRFS_DEBUG |
3cdde224 | 1447 | if (btrfs_test_opt(info, FRAGMENT_DATA)) |
d0bd4560 | 1448 | seq_puts(seq, ",fragment=data"); |
3cdde224 | 1449 | if (btrfs_test_opt(info, FRAGMENT_METADATA)) |
d0bd4560 JB |
1450 | seq_puts(seq, ",fragment=metadata"); |
1451 | #endif | |
fb592373 JB |
1452 | if (btrfs_test_opt(info, REF_VERIFY)) |
1453 | seq_puts(seq, ",ref_verify"); | |
c8d3fe02 OS |
1454 | seq_printf(seq, ",subvolid=%llu", |
1455 | BTRFS_I(d_inode(dentry))->root->root_key.objectid); | |
1456 | seq_puts(seq, ",subvol="); | |
1457 | seq_dentry(seq, dentry, " \t\n\\"); | |
a9572a15 EP |
1458 | return 0; |
1459 | } | |
1460 | ||
a061fc8d | 1461 | static int btrfs_test_super(struct super_block *s, void *data) |
4b82d6e4 | 1462 | { |
815745cf AV |
1463 | struct btrfs_fs_info *p = data; |
1464 | struct btrfs_fs_info *fs_info = btrfs_sb(s); | |
4b82d6e4 | 1465 | |
815745cf | 1466 | return fs_info->fs_devices == p->fs_devices; |
4b82d6e4 Y |
1467 | } |
1468 | ||
450ba0ea JB |
1469 | static int btrfs_set_super(struct super_block *s, void *data) |
1470 | { | |
6de1d09d AV |
1471 | int err = set_anon_super(s, data); |
1472 | if (!err) | |
1473 | s->s_fs_info = data; | |
1474 | return err; | |
4b82d6e4 Y |
1475 | } |
1476 | ||
f9d9ef62 DS |
1477 | /* |
1478 | * subvolumes are identified by ino 256 | |
1479 | */ | |
1480 | static inline int is_subvolume_inode(struct inode *inode) | |
1481 | { | |
1482 | if (inode && inode->i_ino == BTRFS_FIRST_FREE_OBJECTID) | |
1483 | return 1; | |
1484 | return 0; | |
1485 | } | |
1486 | ||
bb289b7b | 1487 | static struct dentry *mount_subvol(const char *subvol_name, u64 subvol_objectid, |
ae0bc863 | 1488 | struct vfsmount *mnt) |
830c4adb | 1489 | { |
830c4adb | 1490 | struct dentry *root; |
fa330659 | 1491 | int ret; |
830c4adb | 1492 | |
05dbe683 OS |
1493 | if (!subvol_name) { |
1494 | if (!subvol_objectid) { | |
1495 | ret = get_default_subvol_objectid(btrfs_sb(mnt->mnt_sb), | |
1496 | &subvol_objectid); | |
1497 | if (ret) { | |
1498 | root = ERR_PTR(ret); | |
1499 | goto out; | |
1500 | } | |
1501 | } | |
c0c907a4 MPS |
1502 | subvol_name = btrfs_get_subvol_name_from_objectid( |
1503 | btrfs_sb(mnt->mnt_sb), subvol_objectid); | |
05dbe683 OS |
1504 | if (IS_ERR(subvol_name)) { |
1505 | root = ERR_CAST(subvol_name); | |
1506 | subvol_name = NULL; | |
1507 | goto out; | |
1508 | } | |
1509 | ||
1510 | } | |
1511 | ||
ea441d11 | 1512 | root = mount_subtree(mnt, subvol_name); |
fa330659 OS |
1513 | /* mount_subtree() drops our reference on the vfsmount. */ |
1514 | mnt = NULL; | |
830c4adb | 1515 | |
bb289b7b | 1516 | if (!IS_ERR(root)) { |
ea441d11 | 1517 | struct super_block *s = root->d_sb; |
ab8d0fc4 | 1518 | struct btrfs_fs_info *fs_info = btrfs_sb(s); |
bb289b7b OS |
1519 | struct inode *root_inode = d_inode(root); |
1520 | u64 root_objectid = BTRFS_I(root_inode)->root->root_key.objectid; | |
1521 | ||
1522 | ret = 0; | |
1523 | if (!is_subvolume_inode(root_inode)) { | |
ab8d0fc4 | 1524 | btrfs_err(fs_info, "'%s' is not a valid subvolume", |
bb289b7b OS |
1525 | subvol_name); |
1526 | ret = -EINVAL; | |
1527 | } | |
1528 | if (subvol_objectid && root_objectid != subvol_objectid) { | |
05dbe683 OS |
1529 | /* |
1530 | * This will also catch a race condition where a | |
1531 | * subvolume which was passed by ID is renamed and | |
1532 | * another subvolume is renamed over the old location. | |
1533 | */ | |
ab8d0fc4 JM |
1534 | btrfs_err(fs_info, |
1535 | "subvol '%s' does not match subvolid %llu", | |
1536 | subvol_name, subvol_objectid); | |
bb289b7b OS |
1537 | ret = -EINVAL; |
1538 | } | |
1539 | if (ret) { | |
1540 | dput(root); | |
1541 | root = ERR_PTR(ret); | |
1542 | deactivate_locked_super(s); | |
1543 | } | |
f9d9ef62 DS |
1544 | } |
1545 | ||
fa330659 OS |
1546 | out: |
1547 | mntput(mnt); | |
fa330659 | 1548 | kfree(subvol_name); |
830c4adb JB |
1549 | return root; |
1550 | } | |
450ba0ea | 1551 | |
312c89fb MT |
1552 | /* |
1553 | * Find a superblock for the given device / mount point. | |
1554 | * | |
1555 | * Note: This is based on mount_bdev from fs/super.c with a few additions | |
1556 | * for multiple device setup. Make sure to keep it in sync. | |
1557 | */ | |
72fa39f5 MT |
1558 | static struct dentry *btrfs_mount_root(struct file_system_type *fs_type, |
1559 | int flags, const char *device_name, void *data) | |
1560 | { | |
1561 | struct block_device *bdev = NULL; | |
1562 | struct super_block *s; | |
36350e95 | 1563 | struct btrfs_device *device = NULL; |
72fa39f5 MT |
1564 | struct btrfs_fs_devices *fs_devices = NULL; |
1565 | struct btrfs_fs_info *fs_info = NULL; | |
204cc0cc | 1566 | void *new_sec_opts = NULL; |
72fa39f5 | 1567 | fmode_t mode = FMODE_READ; |
72fa39f5 MT |
1568 | int error = 0; |
1569 | ||
1570 | if (!(flags & SB_RDONLY)) | |
1571 | mode |= FMODE_WRITE; | |
1572 | ||
72fa39f5 | 1573 | if (data) { |
a65001e8 | 1574 | error = security_sb_eat_lsm_opts(data, &new_sec_opts); |
72fa39f5 MT |
1575 | if (error) |
1576 | return ERR_PTR(error); | |
1577 | } | |
1578 | ||
72fa39f5 MT |
1579 | /* |
1580 | * Setup a dummy root and fs_info for test/set super. This is because | |
1581 | * we don't actually fill this stuff out until open_ctree, but we need | |
8260edba JB |
1582 | * then open_ctree will properly initialize the file system specific |
1583 | * settings later. btrfs_init_fs_info initializes the static elements | |
1584 | * of the fs_info (locks and such) to make cleanup easier if we find a | |
1585 | * superblock with our given fs_devices later on at sget() time. | |
72fa39f5 | 1586 | */ |
a8fd1f71 | 1587 | fs_info = kvzalloc(sizeof(struct btrfs_fs_info), GFP_KERNEL); |
72fa39f5 MT |
1588 | if (!fs_info) { |
1589 | error = -ENOMEM; | |
1590 | goto error_sec_opts; | |
1591 | } | |
8260edba | 1592 | btrfs_init_fs_info(fs_info); |
72fa39f5 | 1593 | |
72fa39f5 MT |
1594 | fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); |
1595 | fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); | |
72fa39f5 MT |
1596 | if (!fs_info->super_copy || !fs_info->super_for_commit) { |
1597 | error = -ENOMEM; | |
1598 | goto error_fs_info; | |
1599 | } | |
1600 | ||
399f7f4c | 1601 | mutex_lock(&uuid_mutex); |
fa59f27c | 1602 | error = btrfs_parse_device_options(data, mode, fs_type); |
81ffd56b DS |
1603 | if (error) { |
1604 | mutex_unlock(&uuid_mutex); | |
399f7f4c | 1605 | goto error_fs_info; |
81ffd56b | 1606 | } |
399f7f4c | 1607 | |
36350e95 GJ |
1608 | device = btrfs_scan_one_device(device_name, mode, fs_type); |
1609 | if (IS_ERR(device)) { | |
81ffd56b | 1610 | mutex_unlock(&uuid_mutex); |
36350e95 | 1611 | error = PTR_ERR(device); |
399f7f4c | 1612 | goto error_fs_info; |
81ffd56b | 1613 | } |
399f7f4c | 1614 | |
36350e95 | 1615 | fs_devices = device->fs_devices; |
399f7f4c DS |
1616 | fs_info->fs_devices = fs_devices; |
1617 | ||
72fa39f5 | 1618 | error = btrfs_open_devices(fs_devices, mode, fs_type); |
f5194e34 | 1619 | mutex_unlock(&uuid_mutex); |
72fa39f5 MT |
1620 | if (error) |
1621 | goto error_fs_info; | |
1622 | ||
1623 | if (!(flags & SB_RDONLY) && fs_devices->rw_devices == 0) { | |
1624 | error = -EACCES; | |
1625 | goto error_close_devices; | |
1626 | } | |
1627 | ||
1628 | bdev = fs_devices->latest_bdev; | |
1629 | s = sget(fs_type, btrfs_test_super, btrfs_set_super, flags | SB_NOSEC, | |
1630 | fs_info); | |
1631 | if (IS_ERR(s)) { | |
1632 | error = PTR_ERR(s); | |
1633 | goto error_close_devices; | |
1634 | } | |
1635 | ||
1636 | if (s->s_root) { | |
1637 | btrfs_close_devices(fs_devices); | |
0d4b0463 | 1638 | btrfs_free_fs_info(fs_info); |
72fa39f5 MT |
1639 | if ((flags ^ s->s_flags) & SB_RDONLY) |
1640 | error = -EBUSY; | |
1641 | } else { | |
1642 | snprintf(s->s_id, sizeof(s->s_id), "%pg", bdev); | |
1643 | btrfs_sb(s)->bdev_holder = fs_type; | |
9b4e675a DS |
1644 | if (!strstr(crc32c_impl(), "generic")) |
1645 | set_bit(BTRFS_FS_CSUM_IMPL_FAST, &fs_info->flags); | |
72fa39f5 MT |
1646 | error = btrfs_fill_super(s, fs_devices, data); |
1647 | } | |
a65001e8 | 1648 | if (!error) |
204cc0cc | 1649 | error = security_sb_set_mnt_opts(s, new_sec_opts, 0, NULL); |
a65001e8 | 1650 | security_free_mnt_opts(&new_sec_opts); |
72fa39f5 MT |
1651 | if (error) { |
1652 | deactivate_locked_super(s); | |
a65001e8 | 1653 | return ERR_PTR(error); |
72fa39f5 MT |
1654 | } |
1655 | ||
1656 | return dget(s->s_root); | |
1657 | ||
1658 | error_close_devices: | |
1659 | btrfs_close_devices(fs_devices); | |
1660 | error_fs_info: | |
0d4b0463 | 1661 | btrfs_free_fs_info(fs_info); |
72fa39f5 MT |
1662 | error_sec_opts: |
1663 | security_free_mnt_opts(&new_sec_opts); | |
1664 | return ERR_PTR(error); | |
1665 | } | |
312c89fb | 1666 | |
edf24abe | 1667 | /* |
312c89fb | 1668 | * Mount function which is called by VFS layer. |
edf24abe | 1669 | * |
312c89fb MT |
1670 | * In order to allow mounting a subvolume directly, btrfs uses mount_subtree() |
1671 | * which needs vfsmount* of device's root (/). This means device's root has to | |
1672 | * be mounted internally in any case. | |
1673 | * | |
1674 | * Operation flow: | |
1675 | * 1. Parse subvol id related options for later use in mount_subvol(). | |
1676 | * | |
1677 | * 2. Mount device's root (/) by calling vfs_kern_mount(). | |
1678 | * | |
1679 | * NOTE: vfs_kern_mount() is used by VFS to call btrfs_mount() in the | |
1680 | * first place. In order to avoid calling btrfs_mount() again, we use | |
1681 | * different file_system_type which is not registered to VFS by | |
1682 | * register_filesystem() (btrfs_root_fs_type). As a result, | |
1683 | * btrfs_mount_root() is called. The return value will be used by | |
1684 | * mount_subtree() in mount_subvol(). | |
1685 | * | |
1686 | * 3. Call mount_subvol() to get the dentry of subvolume. Since there is | |
1687 | * "btrfs subvolume set-default", mount_subvol() is called always. | |
edf24abe | 1688 | */ |
061dbc6b | 1689 | static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags, |
306e16ce | 1690 | const char *device_name, void *data) |
4b82d6e4 | 1691 | { |
312c89fb MT |
1692 | struct vfsmount *mnt_root; |
1693 | struct dentry *root; | |
73f73415 JB |
1694 | char *subvol_name = NULL; |
1695 | u64 subvol_objectid = 0; | |
4b82d6e4 Y |
1696 | int error = 0; |
1697 | ||
93b9bcdf GJ |
1698 | error = btrfs_parse_subvol_options(data, &subvol_name, |
1699 | &subvol_objectid); | |
f23c8af8 ID |
1700 | if (error) { |
1701 | kfree(subvol_name); | |
061dbc6b | 1702 | return ERR_PTR(error); |
f23c8af8 | 1703 | } |
edf24abe | 1704 | |
312c89fb MT |
1705 | /* mount device's root (/) */ |
1706 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, flags, device_name, data); | |
1707 | if (PTR_ERR_OR_ZERO(mnt_root) == -EBUSY) { | |
1708 | if (flags & SB_RDONLY) { | |
1709 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, | |
1710 | flags & ~SB_RDONLY, device_name, data); | |
1711 | } else { | |
1712 | mnt_root = vfs_kern_mount(&btrfs_root_fs_type, | |
1713 | flags | SB_RDONLY, device_name, data); | |
1714 | if (IS_ERR(mnt_root)) { | |
1715 | root = ERR_CAST(mnt_root); | |
532b618b | 1716 | kfree(subvol_name); |
312c89fb MT |
1717 | goto out; |
1718 | } | |
4b82d6e4 | 1719 | |
312c89fb MT |
1720 | down_write(&mnt_root->mnt_sb->s_umount); |
1721 | error = btrfs_remount(mnt_root->mnt_sb, &flags, NULL); | |
1722 | up_write(&mnt_root->mnt_sb->s_umount); | |
1723 | if (error < 0) { | |
1724 | root = ERR_PTR(error); | |
1725 | mntput(mnt_root); | |
532b618b | 1726 | kfree(subvol_name); |
312c89fb MT |
1727 | goto out; |
1728 | } | |
1729 | } | |
f667aef6 | 1730 | } |
312c89fb MT |
1731 | if (IS_ERR(mnt_root)) { |
1732 | root = ERR_CAST(mnt_root); | |
532b618b | 1733 | kfree(subvol_name); |
312c89fb | 1734 | goto out; |
f667aef6 | 1735 | } |
4b82d6e4 | 1736 | |
312c89fb | 1737 | /* mount_subvol() will free subvol_name and mnt_root */ |
ae0bc863 | 1738 | root = mount_subvol(subvol_name, subvol_objectid, mnt_root); |
4b82d6e4 | 1739 | |
312c89fb MT |
1740 | out: |
1741 | return root; | |
4b82d6e4 | 1742 | } |
2e635a27 | 1743 | |
0d2450ab | 1744 | static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info, |
f7b885be | 1745 | u32 new_pool_size, u32 old_pool_size) |
0d2450ab ST |
1746 | { |
1747 | if (new_pool_size == old_pool_size) | |
1748 | return; | |
1749 | ||
1750 | fs_info->thread_pool_size = new_pool_size; | |
1751 | ||
efe120a0 | 1752 | btrfs_info(fs_info, "resize thread pool %d -> %d", |
0d2450ab ST |
1753 | old_pool_size, new_pool_size); |
1754 | ||
5cdc7ad3 | 1755 | btrfs_workqueue_set_max(fs_info->workers, new_pool_size); |
afe3d242 | 1756 | btrfs_workqueue_set_max(fs_info->delalloc_workers, new_pool_size); |
e66f0bb1 | 1757 | btrfs_workqueue_set_max(fs_info->caching_workers, new_pool_size); |
fccb5d86 QW |
1758 | btrfs_workqueue_set_max(fs_info->endio_workers, new_pool_size); |
1759 | btrfs_workqueue_set_max(fs_info->endio_meta_workers, new_pool_size); | |
1760 | btrfs_workqueue_set_max(fs_info->endio_meta_write_workers, | |
1761 | new_pool_size); | |
1762 | btrfs_workqueue_set_max(fs_info->endio_write_workers, new_pool_size); | |
1763 | btrfs_workqueue_set_max(fs_info->endio_freespace_worker, new_pool_size); | |
5b3bc44e | 1764 | btrfs_workqueue_set_max(fs_info->delayed_workers, new_pool_size); |
736cfa15 | 1765 | btrfs_workqueue_set_max(fs_info->readahead_workers, new_pool_size); |
0339ef2f QW |
1766 | btrfs_workqueue_set_max(fs_info->scrub_wr_completion_workers, |
1767 | new_pool_size); | |
0d2450ab ST |
1768 | } |
1769 | ||
f42a34b2 | 1770 | static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info) |
dc81cdc5 MX |
1771 | { |
1772 | set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); | |
f42a34b2 | 1773 | } |
dc81cdc5 | 1774 | |
f42a34b2 MX |
1775 | static inline void btrfs_remount_begin(struct btrfs_fs_info *fs_info, |
1776 | unsigned long old_opts, int flags) | |
1777 | { | |
dc81cdc5 MX |
1778 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && |
1779 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || | |
1751e8a6 | 1780 | (flags & SB_RDONLY))) { |
dc81cdc5 MX |
1781 | /* wait for any defraggers to finish */ |
1782 | wait_event(fs_info->transaction_wait, | |
1783 | (atomic_read(&fs_info->defrag_running) == 0)); | |
1751e8a6 | 1784 | if (flags & SB_RDONLY) |
dc81cdc5 MX |
1785 | sync_filesystem(fs_info->sb); |
1786 | } | |
1787 | } | |
1788 | ||
1789 | static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info, | |
1790 | unsigned long old_opts) | |
1791 | { | |
1792 | /* | |
180e4d47 LB |
1793 | * We need to cleanup all defragable inodes if the autodefragment is |
1794 | * close or the filesystem is read only. | |
dc81cdc5 MX |
1795 | */ |
1796 | if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && | |
bc98a42c | 1797 | (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || sb_rdonly(fs_info->sb))) { |
dc81cdc5 MX |
1798 | btrfs_cleanup_defrag_inodes(fs_info); |
1799 | } | |
1800 | ||
b0643e59 DZ |
1801 | /* If we toggled discard async */ |
1802 | if (!btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) && | |
1803 | btrfs_test_opt(fs_info, DISCARD_ASYNC)) | |
1804 | btrfs_discard_resume(fs_info); | |
1805 | else if (btrfs_raw_test_opt(old_opts, DISCARD_ASYNC) && | |
1806 | !btrfs_test_opt(fs_info, DISCARD_ASYNC)) | |
1807 | btrfs_discard_cleanup(fs_info); | |
1808 | ||
dc81cdc5 MX |
1809 | clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); |
1810 | } | |
1811 | ||
c146afad YZ |
1812 | static int btrfs_remount(struct super_block *sb, int *flags, char *data) |
1813 | { | |
815745cf AV |
1814 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
1815 | struct btrfs_root *root = fs_info->tree_root; | |
49b25e05 JM |
1816 | unsigned old_flags = sb->s_flags; |
1817 | unsigned long old_opts = fs_info->mount_opt; | |
1818 | unsigned long old_compress_type = fs_info->compress_type; | |
1819 | u64 old_max_inline = fs_info->max_inline; | |
f7b885be | 1820 | u32 old_thread_pool_size = fs_info->thread_pool_size; |
d612ac59 | 1821 | u32 old_metadata_ratio = fs_info->metadata_ratio; |
c146afad YZ |
1822 | int ret; |
1823 | ||
02b9984d | 1824 | sync_filesystem(sb); |
f42a34b2 | 1825 | btrfs_remount_prepare(fs_info); |
dc81cdc5 | 1826 | |
f667aef6 | 1827 | if (data) { |
204cc0cc | 1828 | void *new_sec_opts = NULL; |
f667aef6 | 1829 | |
a65001e8 AV |
1830 | ret = security_sb_eat_lsm_opts(data, &new_sec_opts); |
1831 | if (!ret) | |
204cc0cc | 1832 | ret = security_sb_remount(sb, new_sec_opts); |
a65001e8 | 1833 | security_free_mnt_opts(&new_sec_opts); |
f667aef6 QW |
1834 | if (ret) |
1835 | goto restore; | |
f667aef6 QW |
1836 | } |
1837 | ||
2ff7e61e | 1838 | ret = btrfs_parse_options(fs_info, data, *flags); |
891f41cb | 1839 | if (ret) |
49b25e05 | 1840 | goto restore; |
b288052e | 1841 | |
f42a34b2 | 1842 | btrfs_remount_begin(fs_info, old_opts, *flags); |
0d2450ab ST |
1843 | btrfs_resize_thread_pool(fs_info, |
1844 | fs_info->thread_pool_size, old_thread_pool_size); | |
1845 | ||
1751e8a6 | 1846 | if ((bool)(*flags & SB_RDONLY) == sb_rdonly(sb)) |
dc81cdc5 | 1847 | goto out; |
c146afad | 1848 | |
1751e8a6 | 1849 | if (*flags & SB_RDONLY) { |
8dabb742 SB |
1850 | /* |
1851 | * this also happens on 'umount -rf' or on shutdown, when | |
1852 | * the filesystem is busy. | |
1853 | */ | |
21c7e756 | 1854 | cancel_work_sync(&fs_info->async_reclaim_work); |
361c093d | 1855 | |
b0643e59 DZ |
1856 | btrfs_discard_cleanup(fs_info); |
1857 | ||
361c093d SB |
1858 | /* wait for the uuid_scan task to finish */ |
1859 | down(&fs_info->uuid_tree_rescan_sem); | |
1860 | /* avoid complains from lockdep et al. */ | |
1861 | up(&fs_info->uuid_tree_rescan_sem); | |
1862 | ||
1751e8a6 | 1863 | sb->s_flags |= SB_RDONLY; |
c146afad | 1864 | |
e44163e1 | 1865 | /* |
1751e8a6 | 1866 | * Setting SB_RDONLY will put the cleaner thread to |
e44163e1 JM |
1867 | * sleep at the next loop if it's already active. |
1868 | * If it's already asleep, we'll leave unused block | |
1869 | * groups on disk until we're mounted read-write again | |
1870 | * unless we clean them up here. | |
1871 | */ | |
e44163e1 | 1872 | btrfs_delete_unused_bgs(fs_info); |
e44163e1 | 1873 | |
8dabb742 SB |
1874 | btrfs_dev_replace_suspend_for_unmount(fs_info); |
1875 | btrfs_scrub_cancel(fs_info); | |
061594ef | 1876 | btrfs_pause_balance(fs_info); |
8dabb742 | 1877 | |
6bccf3ab | 1878 | ret = btrfs_commit_super(fs_info); |
49b25e05 JM |
1879 | if (ret) |
1880 | goto restore; | |
c146afad | 1881 | } else { |
0b246afa | 1882 | if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { |
6ef3de9c | 1883 | btrfs_err(fs_info, |
efe120a0 | 1884 | "Remounting read-write after error is not allowed"); |
6ef3de9c DS |
1885 | ret = -EINVAL; |
1886 | goto restore; | |
1887 | } | |
8a3db184 | 1888 | if (fs_info->fs_devices->rw_devices == 0) { |
49b25e05 JM |
1889 | ret = -EACCES; |
1890 | goto restore; | |
8a3db184 | 1891 | } |
2b82032c | 1892 | |
6528b99d | 1893 | if (!btrfs_check_rw_degradable(fs_info, NULL)) { |
efe120a0 | 1894 | btrfs_warn(fs_info, |
52042d8e | 1895 | "too many missing devices, writable remount is not allowed"); |
292fd7fc SB |
1896 | ret = -EACCES; |
1897 | goto restore; | |
1898 | } | |
1899 | ||
8a3db184 | 1900 | if (btrfs_super_log_root(fs_info->super_copy) != 0) { |
10a3a3ed DS |
1901 | btrfs_warn(fs_info, |
1902 | "mount required to replay tree-log, cannot remount read-write"); | |
49b25e05 JM |
1903 | ret = -EINVAL; |
1904 | goto restore; | |
8a3db184 | 1905 | } |
c146afad | 1906 | |
815745cf | 1907 | ret = btrfs_cleanup_fs_roots(fs_info); |
49b25e05 JM |
1908 | if (ret) |
1909 | goto restore; | |
c146afad | 1910 | |
d68fc57b | 1911 | /* recover relocation */ |
5f316481 | 1912 | mutex_lock(&fs_info->cleaner_mutex); |
d68fc57b | 1913 | ret = btrfs_recover_relocation(root); |
5f316481 | 1914 | mutex_unlock(&fs_info->cleaner_mutex); |
49b25e05 JM |
1915 | if (ret) |
1916 | goto restore; | |
c146afad | 1917 | |
2b6ba629 ID |
1918 | ret = btrfs_resume_balance_async(fs_info); |
1919 | if (ret) | |
1920 | goto restore; | |
1921 | ||
8dabb742 SB |
1922 | ret = btrfs_resume_dev_replace_async(fs_info); |
1923 | if (ret) { | |
efe120a0 | 1924 | btrfs_warn(fs_info, "failed to resume dev_replace"); |
8dabb742 SB |
1925 | goto restore; |
1926 | } | |
94aebfb2 | 1927 | |
6c6b5a39 AS |
1928 | btrfs_qgroup_rescan_resume(fs_info); |
1929 | ||
94aebfb2 | 1930 | if (!fs_info->uuid_root) { |
efe120a0 | 1931 | btrfs_info(fs_info, "creating UUID tree"); |
94aebfb2 JB |
1932 | ret = btrfs_create_uuid_tree(fs_info); |
1933 | if (ret) { | |
5d163e0e JM |
1934 | btrfs_warn(fs_info, |
1935 | "failed to create the UUID tree %d", | |
1936 | ret); | |
94aebfb2 JB |
1937 | goto restore; |
1938 | } | |
1939 | } | |
1751e8a6 | 1940 | sb->s_flags &= ~SB_RDONLY; |
90c711ab | 1941 | |
afcdd129 | 1942 | set_bit(BTRFS_FS_OPEN, &fs_info->flags); |
c146afad | 1943 | } |
dc81cdc5 | 1944 | out: |
2c6a92b0 | 1945 | wake_up_process(fs_info->transaction_kthread); |
dc81cdc5 | 1946 | btrfs_remount_cleanup(fs_info, old_opts); |
c146afad | 1947 | return 0; |
49b25e05 JM |
1948 | |
1949 | restore: | |
1751e8a6 | 1950 | /* We've hit an error - don't reset SB_RDONLY */ |
bc98a42c | 1951 | if (sb_rdonly(sb)) |
1751e8a6 | 1952 | old_flags |= SB_RDONLY; |
49b25e05 JM |
1953 | sb->s_flags = old_flags; |
1954 | fs_info->mount_opt = old_opts; | |
1955 | fs_info->compress_type = old_compress_type; | |
1956 | fs_info->max_inline = old_max_inline; | |
0d2450ab ST |
1957 | btrfs_resize_thread_pool(fs_info, |
1958 | old_thread_pool_size, fs_info->thread_pool_size); | |
49b25e05 | 1959 | fs_info->metadata_ratio = old_metadata_ratio; |
dc81cdc5 | 1960 | btrfs_remount_cleanup(fs_info, old_opts); |
49b25e05 | 1961 | return ret; |
c146afad YZ |
1962 | } |
1963 | ||
bcd53741 | 1964 | /* Used to sort the devices by max_avail(descending sort) */ |
7e17916b | 1965 | static inline int btrfs_cmp_device_free_bytes(const void *dev_info1, |
bcd53741 AJ |
1966 | const void *dev_info2) |
1967 | { | |
1968 | if (((struct btrfs_device_info *)dev_info1)->max_avail > | |
1969 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
1970 | return -1; | |
1971 | else if (((struct btrfs_device_info *)dev_info1)->max_avail < | |
1972 | ((struct btrfs_device_info *)dev_info2)->max_avail) | |
1973 | return 1; | |
1974 | else | |
1975 | return 0; | |
1976 | } | |
1977 | ||
1978 | /* | |
1979 | * sort the devices by max_avail, in which max free extent size of each device | |
1980 | * is stored.(Descending Sort) | |
1981 | */ | |
1982 | static inline void btrfs_descending_sort_devices( | |
1983 | struct btrfs_device_info *devices, | |
1984 | size_t nr_devices) | |
1985 | { | |
1986 | sort(devices, nr_devices, sizeof(struct btrfs_device_info), | |
1987 | btrfs_cmp_device_free_bytes, NULL); | |
1988 | } | |
1989 | ||
6d07bcec MX |
1990 | /* |
1991 | * The helper to calc the free space on the devices that can be used to store | |
1992 | * file data. | |
1993 | */ | |
7e17916b AB |
1994 | static inline int btrfs_calc_avail_data_space(struct btrfs_fs_info *fs_info, |
1995 | u64 *free_bytes) | |
6d07bcec | 1996 | { |
6d07bcec MX |
1997 | struct btrfs_device_info *devices_info; |
1998 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; | |
1999 | struct btrfs_device *device; | |
6d07bcec MX |
2000 | u64 type; |
2001 | u64 avail_space; | |
6d07bcec | 2002 | u64 min_stripe_size; |
559ca6ea | 2003 | int num_stripes = 1; |
6d07bcec | 2004 | int i = 0, nr_devices; |
4f080f57 | 2005 | const struct btrfs_raid_attr *rattr; |
6d07bcec | 2006 | |
7e33fd99 | 2007 | /* |
01327610 | 2008 | * We aren't under the device list lock, so this is racy-ish, but good |
7e33fd99 JB |
2009 | * enough for our purposes. |
2010 | */ | |
b772a86e | 2011 | nr_devices = fs_info->fs_devices->open_devices; |
7e33fd99 JB |
2012 | if (!nr_devices) { |
2013 | smp_mb(); | |
2014 | nr_devices = fs_info->fs_devices->open_devices; | |
2015 | ASSERT(nr_devices); | |
2016 | if (!nr_devices) { | |
2017 | *free_bytes = 0; | |
2018 | return 0; | |
2019 | } | |
2020 | } | |
6d07bcec | 2021 | |
d9b0d9ba | 2022 | devices_info = kmalloc_array(nr_devices, sizeof(*devices_info), |
6a44517d | 2023 | GFP_KERNEL); |
6d07bcec MX |
2024 | if (!devices_info) |
2025 | return -ENOMEM; | |
2026 | ||
01327610 | 2027 | /* calc min stripe number for data space allocation */ |
1b86826d | 2028 | type = btrfs_data_alloc_profile(fs_info); |
4f080f57 DS |
2029 | rattr = &btrfs_raid_array[btrfs_bg_flags_to_raid_index(type)]; |
2030 | ||
e1ea2bee | 2031 | if (type & BTRFS_BLOCK_GROUP_RAID0) |
39fb26c3 | 2032 | num_stripes = nr_devices; |
e1ea2bee | 2033 | else if (type & BTRFS_BLOCK_GROUP_RAID1) |
39fb26c3 | 2034 | num_stripes = 2; |
47e6f742 DS |
2035 | else if (type & BTRFS_BLOCK_GROUP_RAID1C3) |
2036 | num_stripes = 3; | |
8d6fac00 DS |
2037 | else if (type & BTRFS_BLOCK_GROUP_RAID1C4) |
2038 | num_stripes = 4; | |
e1ea2bee | 2039 | else if (type & BTRFS_BLOCK_GROUP_RAID10) |
39fb26c3 | 2040 | num_stripes = 4; |
6d07bcec | 2041 | |
4f080f57 DS |
2042 | /* Adjust for more than 1 stripe per device */ |
2043 | min_stripe_size = rattr->dev_stripes * BTRFS_STRIPE_LEN; | |
6d07bcec | 2044 | |
7e33fd99 JB |
2045 | rcu_read_lock(); |
2046 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { | |
e12c9621 AJ |
2047 | if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, |
2048 | &device->dev_state) || | |
401e29c1 AJ |
2049 | !device->bdev || |
2050 | test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) | |
6d07bcec MX |
2051 | continue; |
2052 | ||
7e33fd99 JB |
2053 | if (i >= nr_devices) |
2054 | break; | |
2055 | ||
6d07bcec MX |
2056 | avail_space = device->total_bytes - device->bytes_used; |
2057 | ||
2058 | /* align with stripe_len */ | |
559ca6ea | 2059 | avail_space = rounddown(avail_space, BTRFS_STRIPE_LEN); |
6d07bcec MX |
2060 | |
2061 | /* | |
01327610 | 2062 | * In order to avoid overwriting the superblock on the drive, |
6d07bcec MX |
2063 | * btrfs starts at an offset of at least 1MB when doing chunk |
2064 | * allocation. | |
559ca6ea NB |
2065 | * |
2066 | * This ensures we have at least min_stripe_size free space | |
2067 | * after excluding 1MB. | |
6d07bcec | 2068 | */ |
559ca6ea | 2069 | if (avail_space <= SZ_1M + min_stripe_size) |
6d07bcec MX |
2070 | continue; |
2071 | ||
559ca6ea NB |
2072 | avail_space -= SZ_1M; |
2073 | ||
6d07bcec MX |
2074 | devices_info[i].dev = device; |
2075 | devices_info[i].max_avail = avail_space; | |
2076 | ||
2077 | i++; | |
2078 | } | |
7e33fd99 | 2079 | rcu_read_unlock(); |
6d07bcec MX |
2080 | |
2081 | nr_devices = i; | |
2082 | ||
2083 | btrfs_descending_sort_devices(devices_info, nr_devices); | |
2084 | ||
2085 | i = nr_devices - 1; | |
2086 | avail_space = 0; | |
559ca6ea NB |
2087 | while (nr_devices >= rattr->devs_min) { |
2088 | num_stripes = min(num_stripes, nr_devices); | |
39fb26c3 | 2089 | |
6d07bcec MX |
2090 | if (devices_info[i].max_avail >= min_stripe_size) { |
2091 | int j; | |
2092 | u64 alloc_size; | |
2093 | ||
39fb26c3 | 2094 | avail_space += devices_info[i].max_avail * num_stripes; |
6d07bcec | 2095 | alloc_size = devices_info[i].max_avail; |
39fb26c3 | 2096 | for (j = i + 1 - num_stripes; j <= i; j++) |
6d07bcec MX |
2097 | devices_info[j].max_avail -= alloc_size; |
2098 | } | |
2099 | i--; | |
2100 | nr_devices--; | |
2101 | } | |
2102 | ||
2103 | kfree(devices_info); | |
2104 | *free_bytes = avail_space; | |
2105 | return 0; | |
2106 | } | |
2107 | ||
ba7b6e62 DS |
2108 | /* |
2109 | * Calculate numbers for 'df', pessimistic in case of mixed raid profiles. | |
2110 | * | |
2111 | * If there's a redundant raid level at DATA block groups, use the respective | |
2112 | * multiplier to scale the sizes. | |
2113 | * | |
2114 | * Unused device space usage is based on simulating the chunk allocator | |
0d0c71b3 DS |
2115 | * algorithm that respects the device sizes and order of allocations. This is |
2116 | * a close approximation of the actual use but there are other factors that may | |
2117 | * change the result (like a new metadata chunk). | |
ba7b6e62 | 2118 | * |
ca8a51b3 | 2119 | * If metadata is exhausted, f_bavail will be 0. |
ba7b6e62 | 2120 | */ |
8fd17795 CM |
2121 | static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) |
2122 | { | |
815745cf AV |
2123 | struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb); |
2124 | struct btrfs_super_block *disk_super = fs_info->super_copy; | |
bd4d1088 JB |
2125 | struct btrfs_space_info *found; |
2126 | u64 total_used = 0; | |
6d07bcec | 2127 | u64 total_free_data = 0; |
ca8a51b3 | 2128 | u64 total_free_meta = 0; |
db94535d | 2129 | int bits = dentry->d_sb->s_blocksize_bits; |
de37aa51 | 2130 | __be32 *fsid = (__be32 *)fs_info->fs_devices->fsid; |
ba7b6e62 DS |
2131 | unsigned factor = 1; |
2132 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; | |
6d07bcec | 2133 | int ret; |
ca8a51b3 | 2134 | u64 thresh = 0; |
ae02d1bd | 2135 | int mixed = 0; |
8fd17795 | 2136 | |
bd4d1088 | 2137 | rcu_read_lock(); |
f5389f33 | 2138 | list_for_each_entry_rcu(found, &fs_info->space_info, list) { |
6d07bcec | 2139 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) { |
ba7b6e62 DS |
2140 | int i; |
2141 | ||
6d07bcec MX |
2142 | total_free_data += found->disk_total - found->disk_used; |
2143 | total_free_data -= | |
2144 | btrfs_account_ro_block_groups_free_space(found); | |
ba7b6e62 DS |
2145 | |
2146 | for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) { | |
46df06b8 DS |
2147 | if (!list_empty(&found->block_groups[i])) |
2148 | factor = btrfs_bg_type_to_factor( | |
2149 | btrfs_raid_array[i].bg_flag); | |
ba7b6e62 | 2150 | } |
6d07bcec | 2151 | } |
ae02d1bd LB |
2152 | |
2153 | /* | |
2154 | * Metadata in mixed block goup profiles are accounted in data | |
2155 | */ | |
2156 | if (!mixed && found->flags & BTRFS_BLOCK_GROUP_METADATA) { | |
2157 | if (found->flags & BTRFS_BLOCK_GROUP_DATA) | |
2158 | mixed = 1; | |
2159 | else | |
2160 | total_free_meta += found->disk_total - | |
2161 | found->disk_used; | |
2162 | } | |
6d07bcec | 2163 | |
b742bb82 | 2164 | total_used += found->disk_used; |
89a55897 | 2165 | } |
ba7b6e62 | 2166 | |
bd4d1088 JB |
2167 | rcu_read_unlock(); |
2168 | ||
ba7b6e62 DS |
2169 | buf->f_blocks = div_u64(btrfs_super_total_bytes(disk_super), factor); |
2170 | buf->f_blocks >>= bits; | |
2171 | buf->f_bfree = buf->f_blocks - (div_u64(total_used, factor) >> bits); | |
2172 | ||
2173 | /* Account global block reserve as used, it's in logical size already */ | |
2174 | spin_lock(&block_rsv->lock); | |
41b34acc LB |
2175 | /* Mixed block groups accounting is not byte-accurate, avoid overflow */ |
2176 | if (buf->f_bfree >= block_rsv->size >> bits) | |
2177 | buf->f_bfree -= block_rsv->size >> bits; | |
2178 | else | |
2179 | buf->f_bfree = 0; | |
ba7b6e62 DS |
2180 | spin_unlock(&block_rsv->lock); |
2181 | ||
0d95c1be | 2182 | buf->f_bavail = div_u64(total_free_data, factor); |
6bccf3ab | 2183 | ret = btrfs_calc_avail_data_space(fs_info, &total_free_data); |
7e33fd99 | 2184 | if (ret) |
6d07bcec | 2185 | return ret; |
ba7b6e62 | 2186 | buf->f_bavail += div_u64(total_free_data, factor); |
6d07bcec | 2187 | buf->f_bavail = buf->f_bavail >> bits; |
d397712b | 2188 | |
ca8a51b3 DS |
2189 | /* |
2190 | * We calculate the remaining metadata space minus global reserve. If | |
2191 | * this is (supposedly) smaller than zero, there's no space. But this | |
2192 | * does not hold in practice, the exhausted state happens where's still | |
2193 | * some positive delta. So we apply some guesswork and compare the | |
2194 | * delta to a 4M threshold. (Practically observed delta was ~2M.) | |
2195 | * | |
2196 | * We probably cannot calculate the exact threshold value because this | |
2197 | * depends on the internal reservations requested by various | |
2198 | * operations, so some operations that consume a few metadata will | |
2199 | * succeed even if the Avail is zero. But this is better than the other | |
2200 | * way around. | |
2201 | */ | |
d4417e22 | 2202 | thresh = SZ_4M; |
ca8a51b3 | 2203 | |
d55966c4 JB |
2204 | /* |
2205 | * We only want to claim there's no available space if we can no longer | |
2206 | * allocate chunks for our metadata profile and our global reserve will | |
2207 | * not fit in the free metadata space. If we aren't ->full then we | |
2208 | * still can allocate chunks and thus are fine using the currently | |
2209 | * calculated f_bavail. | |
2210 | */ | |
2211 | if (!mixed && block_rsv->space_info->full && | |
2212 | total_free_meta - thresh < block_rsv->size) | |
ca8a51b3 DS |
2213 | buf->f_bavail = 0; |
2214 | ||
ba7b6e62 DS |
2215 | buf->f_type = BTRFS_SUPER_MAGIC; |
2216 | buf->f_bsize = dentry->d_sb->s_blocksize; | |
2217 | buf->f_namelen = BTRFS_NAME_LEN; | |
2218 | ||
9d03632e | 2219 | /* We treat it as constant endianness (it doesn't matter _which_) |
d397712b | 2220 | because we want the fsid to come out the same whether mounted |
9d03632e DW |
2221 | on a big-endian or little-endian host */ |
2222 | buf->f_fsid.val[0] = be32_to_cpu(fsid[0]) ^ be32_to_cpu(fsid[2]); | |
2223 | buf->f_fsid.val[1] = be32_to_cpu(fsid[1]) ^ be32_to_cpu(fsid[3]); | |
32d48fa1 | 2224 | /* Mask in the root object ID too, to disambiguate subvols */ |
4fd786e6 MT |
2225 | buf->f_fsid.val[0] ^= |
2226 | BTRFS_I(d_inode(dentry))->root->root_key.objectid >> 32; | |
2227 | buf->f_fsid.val[1] ^= | |
2228 | BTRFS_I(d_inode(dentry))->root->root_key.objectid; | |
32d48fa1 | 2229 | |
8fd17795 CM |
2230 | return 0; |
2231 | } | |
b5133862 | 2232 | |
aea52e19 AV |
2233 | static void btrfs_kill_super(struct super_block *sb) |
2234 | { | |
815745cf | 2235 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
aea52e19 | 2236 | kill_anon_super(sb); |
0d4b0463 | 2237 | btrfs_free_fs_info(fs_info); |
aea52e19 AV |
2238 | } |
2239 | ||
2e635a27 CM |
2240 | static struct file_system_type btrfs_fs_type = { |
2241 | .owner = THIS_MODULE, | |
2242 | .name = "btrfs", | |
061dbc6b | 2243 | .mount = btrfs_mount, |
aea52e19 | 2244 | .kill_sb = btrfs_kill_super, |
f667aef6 | 2245 | .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA, |
2e635a27 | 2246 | }; |
72fa39f5 MT |
2247 | |
2248 | static struct file_system_type btrfs_root_fs_type = { | |
2249 | .owner = THIS_MODULE, | |
2250 | .name = "btrfs", | |
2251 | .mount = btrfs_mount_root, | |
2252 | .kill_sb = btrfs_kill_super, | |
2253 | .fs_flags = FS_REQUIRES_DEV | FS_BINARY_MOUNTDATA, | |
2254 | }; | |
2255 | ||
7f78e035 | 2256 | MODULE_ALIAS_FS("btrfs"); |
a9218f6b | 2257 | |
d8620958 TVB |
2258 | static int btrfs_control_open(struct inode *inode, struct file *file) |
2259 | { | |
2260 | /* | |
2261 | * The control file's private_data is used to hold the | |
2262 | * transaction when it is started and is used to keep | |
2263 | * track of whether a transaction is already in progress. | |
2264 | */ | |
2265 | file->private_data = NULL; | |
2266 | return 0; | |
2267 | } | |
2268 | ||
d352ac68 | 2269 | /* |
cfe953c8 | 2270 | * Used by /dev/btrfs-control for devices ioctls. |
d352ac68 | 2271 | */ |
8a4b83cc CM |
2272 | static long btrfs_control_ioctl(struct file *file, unsigned int cmd, |
2273 | unsigned long arg) | |
2274 | { | |
2275 | struct btrfs_ioctl_vol_args *vol; | |
36350e95 | 2276 | struct btrfs_device *device = NULL; |
c071fcfd | 2277 | int ret = -ENOTTY; |
8a4b83cc | 2278 | |
e441d54d CM |
2279 | if (!capable(CAP_SYS_ADMIN)) |
2280 | return -EPERM; | |
2281 | ||
dae7b665 LZ |
2282 | vol = memdup_user((void __user *)arg, sizeof(*vol)); |
2283 | if (IS_ERR(vol)) | |
2284 | return PTR_ERR(vol); | |
f505754f | 2285 | vol->name[BTRFS_PATH_NAME_MAX] = '\0'; |
c071fcfd | 2286 | |
8a4b83cc CM |
2287 | switch (cmd) { |
2288 | case BTRFS_IOC_SCAN_DEV: | |
899f9307 | 2289 | mutex_lock(&uuid_mutex); |
36350e95 GJ |
2290 | device = btrfs_scan_one_device(vol->name, FMODE_READ, |
2291 | &btrfs_root_fs_type); | |
2292 | ret = PTR_ERR_OR_ZERO(device); | |
899f9307 | 2293 | mutex_unlock(&uuid_mutex); |
8a4b83cc | 2294 | break; |
228a73ab AJ |
2295 | case BTRFS_IOC_FORGET_DEV: |
2296 | ret = btrfs_forget_devices(vol->name); | |
2297 | break; | |
02db0844 | 2298 | case BTRFS_IOC_DEVICES_READY: |
899f9307 | 2299 | mutex_lock(&uuid_mutex); |
36350e95 GJ |
2300 | device = btrfs_scan_one_device(vol->name, FMODE_READ, |
2301 | &btrfs_root_fs_type); | |
2302 | if (IS_ERR(device)) { | |
899f9307 | 2303 | mutex_unlock(&uuid_mutex); |
36350e95 | 2304 | ret = PTR_ERR(device); |
02db0844 | 2305 | break; |
899f9307 | 2306 | } |
36350e95 GJ |
2307 | ret = !(device->fs_devices->num_devices == |
2308 | device->fs_devices->total_devices); | |
899f9307 | 2309 | mutex_unlock(&uuid_mutex); |
02db0844 | 2310 | break; |
c5868f83 | 2311 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 2312 | ret = btrfs_ioctl_get_supported_features((void __user*)arg); |
c5868f83 | 2313 | break; |
8a4b83cc | 2314 | } |
dae7b665 | 2315 | |
8a4b83cc | 2316 | kfree(vol); |
f819d837 | 2317 | return ret; |
8a4b83cc CM |
2318 | } |
2319 | ||
0176260f | 2320 | static int btrfs_freeze(struct super_block *sb) |
ed0dab6b | 2321 | { |
354aa0fb | 2322 | struct btrfs_trans_handle *trans; |
0b246afa JM |
2323 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
2324 | struct btrfs_root *root = fs_info->tree_root; | |
354aa0fb | 2325 | |
fac03c8d | 2326 | set_bit(BTRFS_FS_FROZEN, &fs_info->flags); |
9e7cc91a WX |
2327 | /* |
2328 | * We don't need a barrier here, we'll wait for any transaction that | |
2329 | * could be in progress on other threads (and do delayed iputs that | |
2330 | * we want to avoid on a frozen filesystem), or do the commit | |
2331 | * ourselves. | |
2332 | */ | |
d4edf39b | 2333 | trans = btrfs_attach_transaction_barrier(root); |
354aa0fb MX |
2334 | if (IS_ERR(trans)) { |
2335 | /* no transaction, don't bother */ | |
2336 | if (PTR_ERR(trans) == -ENOENT) | |
2337 | return 0; | |
2338 | return PTR_ERR(trans); | |
2339 | } | |
3a45bb20 | 2340 | return btrfs_commit_transaction(trans); |
ed0dab6b Y |
2341 | } |
2342 | ||
9e7cc91a WX |
2343 | static int btrfs_unfreeze(struct super_block *sb) |
2344 | { | |
fac03c8d DS |
2345 | struct btrfs_fs_info *fs_info = btrfs_sb(sb); |
2346 | ||
2347 | clear_bit(BTRFS_FS_FROZEN, &fs_info->flags); | |
9e7cc91a WX |
2348 | return 0; |
2349 | } | |
2350 | ||
9c5085c1 JB |
2351 | static int btrfs_show_devname(struct seq_file *m, struct dentry *root) |
2352 | { | |
2353 | struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb); | |
9c5085c1 | 2354 | struct btrfs_device *dev, *first_dev = NULL; |
9c5085c1 | 2355 | |
88c14590 DS |
2356 | /* |
2357 | * Lightweight locking of the devices. We should not need | |
2358 | * device_list_mutex here as we only read the device data and the list | |
2359 | * is protected by RCU. Even if a device is deleted during the list | |
2360 | * traversals, we'll get valid data, the freeing callback will wait at | |
52042d8e | 2361 | * least until the rcu_read_unlock. |
88c14590 DS |
2362 | */ |
2363 | rcu_read_lock(); | |
4faf55b0 AJ |
2364 | list_for_each_entry_rcu(dev, &fs_info->fs_devices->devices, dev_list) { |
2365 | if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state)) | |
2366 | continue; | |
2367 | if (!dev->name) | |
2368 | continue; | |
2369 | if (!first_dev || dev->devid < first_dev->devid) | |
2370 | first_dev = dev; | |
9c5085c1 JB |
2371 | } |
2372 | ||
672d5990 MT |
2373 | if (first_dev) |
2374 | seq_escape(m, rcu_str_deref(first_dev->name), " \t\n\\"); | |
2375 | else | |
9c5085c1 | 2376 | WARN_ON(1); |
88c14590 | 2377 | rcu_read_unlock(); |
9c5085c1 JB |
2378 | return 0; |
2379 | } | |
2380 | ||
b87221de | 2381 | static const struct super_operations btrfs_super_ops = { |
76dda93c | 2382 | .drop_inode = btrfs_drop_inode, |
bd555975 | 2383 | .evict_inode = btrfs_evict_inode, |
e20d96d6 | 2384 | .put_super = btrfs_put_super, |
d5719762 | 2385 | .sync_fs = btrfs_sync_fs, |
a9572a15 | 2386 | .show_options = btrfs_show_options, |
9c5085c1 | 2387 | .show_devname = btrfs_show_devname, |
2c90e5d6 CM |
2388 | .alloc_inode = btrfs_alloc_inode, |
2389 | .destroy_inode = btrfs_destroy_inode, | |
26602cab | 2390 | .free_inode = btrfs_free_inode, |
8fd17795 | 2391 | .statfs = btrfs_statfs, |
c146afad | 2392 | .remount_fs = btrfs_remount, |
0176260f | 2393 | .freeze_fs = btrfs_freeze, |
9e7cc91a | 2394 | .unfreeze_fs = btrfs_unfreeze, |
e20d96d6 | 2395 | }; |
a9218f6b CM |
2396 | |
2397 | static const struct file_operations btrfs_ctl_fops = { | |
d8620958 | 2398 | .open = btrfs_control_open, |
a9218f6b | 2399 | .unlocked_ioctl = btrfs_control_ioctl, |
1832f2d8 | 2400 | .compat_ioctl = compat_ptr_ioctl, |
a9218f6b | 2401 | .owner = THIS_MODULE, |
6038f373 | 2402 | .llseek = noop_llseek, |
a9218f6b CM |
2403 | }; |
2404 | ||
2405 | static struct miscdevice btrfs_misc = { | |
578454ff | 2406 | .minor = BTRFS_MINOR, |
a9218f6b CM |
2407 | .name = "btrfs-control", |
2408 | .fops = &btrfs_ctl_fops | |
2409 | }; | |
2410 | ||
578454ff KS |
2411 | MODULE_ALIAS_MISCDEV(BTRFS_MINOR); |
2412 | MODULE_ALIAS("devname:btrfs-control"); | |
2413 | ||
f5c29bd9 | 2414 | static int __init btrfs_interface_init(void) |
a9218f6b CM |
2415 | { |
2416 | return misc_register(&btrfs_misc); | |
2417 | } | |
2418 | ||
e67c718b | 2419 | static __cold void btrfs_interface_exit(void) |
a9218f6b | 2420 | { |
f368ed60 | 2421 | misc_deregister(&btrfs_misc); |
a9218f6b CM |
2422 | } |
2423 | ||
f5c29bd9 | 2424 | static void __init btrfs_print_mod_info(void) |
85965600 | 2425 | { |
edf57cbf | 2426 | static const char options[] = "" |
85965600 DS |
2427 | #ifdef CONFIG_BTRFS_DEBUG |
2428 | ", debug=on" | |
2429 | #endif | |
79556c3d SB |
2430 | #ifdef CONFIG_BTRFS_ASSERT |
2431 | ", assert=on" | |
2432 | #endif | |
85965600 DS |
2433 | #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY |
2434 | ", integrity-checker=on" | |
fb592373 JB |
2435 | #endif |
2436 | #ifdef CONFIG_BTRFS_FS_REF_VERIFY | |
2437 | ", ref-verify=on" | |
85965600 | 2438 | #endif |
edf57cbf BVA |
2439 | ; |
2440 | pr_info("Btrfs loaded, crc32c=%s%s\n", crc32c_impl(), options); | |
85965600 DS |
2441 | } |
2442 | ||
2e635a27 CM |
2443 | static int __init init_btrfs_fs(void) |
2444 | { | |
2c90e5d6 | 2445 | int err; |
58176a96 | 2446 | |
63541927 FDBM |
2447 | btrfs_props_init(); |
2448 | ||
58176a96 JB |
2449 | err = btrfs_init_sysfs(); |
2450 | if (err) | |
9678c543 | 2451 | return err; |
58176a96 | 2452 | |
143bede5 | 2453 | btrfs_init_compress(); |
d1310b2e | 2454 | |
261507a0 LZ |
2455 | err = btrfs_init_cachep(); |
2456 | if (err) | |
2457 | goto free_compress; | |
2458 | ||
d1310b2e | 2459 | err = extent_io_init(); |
2f4cbe64 WB |
2460 | if (err) |
2461 | goto free_cachep; | |
2462 | ||
6f0d04f8 | 2463 | err = extent_state_cache_init(); |
d1310b2e CM |
2464 | if (err) |
2465 | goto free_extent_io; | |
2466 | ||
6f0d04f8 JB |
2467 | err = extent_map_init(); |
2468 | if (err) | |
2469 | goto free_extent_state_cache; | |
2470 | ||
6352b91d | 2471 | err = ordered_data_init(); |
2f4cbe64 WB |
2472 | if (err) |
2473 | goto free_extent_map; | |
c8b97818 | 2474 | |
6352b91d MX |
2475 | err = btrfs_delayed_inode_init(); |
2476 | if (err) | |
2477 | goto free_ordered_data; | |
2478 | ||
9247f317 | 2479 | err = btrfs_auto_defrag_init(); |
16cdcec7 MX |
2480 | if (err) |
2481 | goto free_delayed_inode; | |
2482 | ||
78a6184a | 2483 | err = btrfs_delayed_ref_init(); |
9247f317 MX |
2484 | if (err) |
2485 | goto free_auto_defrag; | |
2486 | ||
b9e9a6cb WS |
2487 | err = btrfs_prelim_ref_init(); |
2488 | if (err) | |
af13b492 | 2489 | goto free_delayed_ref; |
b9e9a6cb | 2490 | |
97eb6b69 | 2491 | err = btrfs_end_io_wq_init(); |
78a6184a | 2492 | if (err) |
af13b492 | 2493 | goto free_prelim_ref; |
78a6184a | 2494 | |
97eb6b69 DS |
2495 | err = btrfs_interface_init(); |
2496 | if (err) | |
2497 | goto free_end_io_wq; | |
2498 | ||
e565d4b9 JS |
2499 | btrfs_init_lockdep(); |
2500 | ||
8ae1af3c | 2501 | btrfs_print_mod_info(); |
dc11dd5d JB |
2502 | |
2503 | err = btrfs_run_sanity_tests(); | |
2504 | if (err) | |
2505 | goto unregister_ioctl; | |
2506 | ||
2507 | err = register_filesystem(&btrfs_fs_type); | |
2508 | if (err) | |
2509 | goto unregister_ioctl; | |
74255aa0 | 2510 | |
2f4cbe64 WB |
2511 | return 0; |
2512 | ||
a9218f6b CM |
2513 | unregister_ioctl: |
2514 | btrfs_interface_exit(); | |
97eb6b69 DS |
2515 | free_end_io_wq: |
2516 | btrfs_end_io_wq_exit(); | |
b9e9a6cb WS |
2517 | free_prelim_ref: |
2518 | btrfs_prelim_ref_exit(); | |
78a6184a MX |
2519 | free_delayed_ref: |
2520 | btrfs_delayed_ref_exit(); | |
9247f317 MX |
2521 | free_auto_defrag: |
2522 | btrfs_auto_defrag_exit(); | |
16cdcec7 MX |
2523 | free_delayed_inode: |
2524 | btrfs_delayed_inode_exit(); | |
6352b91d MX |
2525 | free_ordered_data: |
2526 | ordered_data_exit(); | |
2f4cbe64 WB |
2527 | free_extent_map: |
2528 | extent_map_exit(); | |
6f0d04f8 JB |
2529 | free_extent_state_cache: |
2530 | extent_state_cache_exit(); | |
d1310b2e CM |
2531 | free_extent_io: |
2532 | extent_io_exit(); | |
2f4cbe64 WB |
2533 | free_cachep: |
2534 | btrfs_destroy_cachep(); | |
261507a0 LZ |
2535 | free_compress: |
2536 | btrfs_exit_compress(); | |
2f4cbe64 | 2537 | btrfs_exit_sysfs(); |
9678c543 | 2538 | |
2f4cbe64 | 2539 | return err; |
2e635a27 CM |
2540 | } |
2541 | ||
2542 | static void __exit exit_btrfs_fs(void) | |
2543 | { | |
39279cc3 | 2544 | btrfs_destroy_cachep(); |
78a6184a | 2545 | btrfs_delayed_ref_exit(); |
9247f317 | 2546 | btrfs_auto_defrag_exit(); |
16cdcec7 | 2547 | btrfs_delayed_inode_exit(); |
b9e9a6cb | 2548 | btrfs_prelim_ref_exit(); |
6352b91d | 2549 | ordered_data_exit(); |
a52d9a80 | 2550 | extent_map_exit(); |
6f0d04f8 | 2551 | extent_state_cache_exit(); |
d1310b2e | 2552 | extent_io_exit(); |
a9218f6b | 2553 | btrfs_interface_exit(); |
5ed5f588 | 2554 | btrfs_end_io_wq_exit(); |
2e635a27 | 2555 | unregister_filesystem(&btrfs_fs_type); |
58176a96 | 2556 | btrfs_exit_sysfs(); |
8a4b83cc | 2557 | btrfs_cleanup_fs_uuids(); |
261507a0 | 2558 | btrfs_exit_compress(); |
2e635a27 CM |
2559 | } |
2560 | ||
60efa5eb | 2561 | late_initcall(init_btrfs_fs); |
2e635a27 CM |
2562 | module_exit(exit_btrfs_fs) |
2563 | ||
2564 | MODULE_LICENSE("GPL"); | |
d5178578 | 2565 | MODULE_SOFTDEP("pre: crc32c"); |
3951e7f0 | 2566 | MODULE_SOFTDEP("pre: xxhash64"); |
3831bf00 | 2567 | MODULE_SOFTDEP("pre: sha256"); |
352ae07b | 2568 | MODULE_SOFTDEP("pre: blake2b-256"); |