1 /* SPDX-License-Identifier: GPL-2.0 */
2
3 #ifndef BTRFS_FS_H
4 #define BTRFS_FS_H
5
6 #include <crypto/blake2b.h>
7 #include <crypto/sha2.h>
8 #include <linux/blkdev.h>
9 #include <linux/sizes.h>
10 #include <linux/time64.h>
11 #include <linux/compiler.h>
12 #include <linux/math.h>
13 #include <linux/atomic.h>
14 #include <linux/percpu_counter.h>
15 #include <linux/completion.h>
16 #include <linux/lockdep.h>
17 #include <linux/spinlock.h>
18 #include <linux/mutex.h>
19 #include <linux/rwsem.h>
20 #include <linux/semaphore.h>
21 #include <linux/list.h>
22 #include <linux/pagemap.h>
23 #include <linux/radix-tree.h>
24 #include <linux/workqueue.h>
25 #include <linux/wait.h>
26 #include <linux/wait_bit.h>
27 #include <linux/sched.h>
28 #include <linux/rbtree.h>
29 #include <linux/xxhash.h>
30 #include <uapi/linux/btrfs.h>
31 #include <uapi/linux/btrfs_tree.h>
32 #include "extent-io-tree.h"
33 #include "async-thread.h"
34 #include "block-rsv.h"
35 #include "messages.h"
36
37 struct inode;
38 struct super_block;
39 struct kobject;
40 struct reloc_control;
41 struct ulist;
42 struct btrfs_device;
43 struct btrfs_block_group;
44 struct btrfs_root;
45 struct btrfs_fs_devices;
46 struct btrfs_transaction;
47 struct btrfs_balance_control;
48 struct btrfs_subpage_info;
49 struct btrfs_stripe_hash_table;
50 struct btrfs_space_info;
51
52 /*
53 * Minimum data and metadata block size.
54 *
55 * Normally it's 4K, but for testing subpage block size on 4K page systems, we
56 * allow DEBUG builds to accept 2K page size.
57 */
58 #ifdef CONFIG_BTRFS_DEBUG
59 #define BTRFS_MIN_BLOCKSIZE (SZ_2K)
60 #else
61 #define BTRFS_MIN_BLOCKSIZE (SZ_4K)
62 #endif
63
64 #define BTRFS_MAX_BLOCKSIZE (SZ_64K)
65
66 #define BTRFS_MAX_EXTENT_SIZE SZ_128M
67
68 /*
69 * Maximum length to trim in a single iteration to avoid holding device list
70 * mutex for too long.
71 */
72 #define BTRFS_MAX_TRIM_LENGTH SZ_2G
73
74 #define BTRFS_OLDEST_GENERATION 0ULL
75
76 #define BTRFS_EMPTY_DIR_SIZE 0
77
78 #define BTRFS_DIRTY_METADATA_THRESH SZ_32M
79
80 #define BTRFS_SUPER_INFO_OFFSET SZ_64K
81 #define BTRFS_SUPER_INFO_SIZE 4096
82 static_assert(sizeof(struct btrfs_super_block) == BTRFS_SUPER_INFO_SIZE);
83
84 /* Array of bytes with variable length, hexadecimal format 0x1234 */
85 #define BTRFS_CSUM_FMT "0x%*phN"
86 #define BTRFS_CSUM_FMT_VALUE(size, bytes) size, bytes
87
88 #define BTRFS_KEY_FMT "(%llu %u %llu)"
89 #define BTRFS_KEY_FMT_VALUE(key) (key)->objectid, (key)->type, (key)->offset
90
91 /*
92 * Number of metadata items necessary for an unlink operation:
93 *
94 * 1 for the possible orphan item
95 * 1 for the dir item
96 * 1 for the dir index
97 * 1 for the inode ref
98 * 1 for the inode
99 * 1 for the parent inode
100 */
101 #define BTRFS_UNLINK_METADATA_UNITS 6
102
103 /*
104 * The reserved space at the beginning of each device. It covers the primary
105 * super block and leaves space for potential use by other tools like
106 * bootloaders or to lower potential damage of accidental overwrite.
107 */
108 #define BTRFS_DEVICE_RANGE_RESERVED (SZ_1M)
109 /*
110 * Runtime (in-memory) states of filesystem
111 */
112 enum {
113 /*
114 * Filesystem is being remounted, allow to skip some operations, like
115 * defrag
116 */
117 BTRFS_FS_STATE_REMOUNTING,
118 /* Filesystem in RO mode */
119 BTRFS_FS_STATE_RO,
120 /* Track if a transaction abort has been reported on this filesystem */
121 BTRFS_FS_STATE_TRANS_ABORTED,
122 /* Track if log replay has failed. */
123 BTRFS_FS_STATE_LOG_REPLAY_ABORTED,
124 /*
125 * Bio operations should be blocked on this filesystem because a source
126 * or target device is being destroyed as part of a device replace
127 */
128 BTRFS_FS_STATE_DEV_REPLACING,
129 /* The btrfs_fs_info created for self-tests */
130 BTRFS_FS_STATE_DUMMY_FS_INFO,
131
132 /* Checksum errors are ignored. */
133 BTRFS_FS_STATE_NO_DATA_CSUMS,
134 BTRFS_FS_STATE_SKIP_META_CSUMS,
135
136 /* Indicates there was an error cleaning up a log tree. */
137 BTRFS_FS_STATE_LOG_CLEANUP_ERROR,
138
139 /* No more delayed iput can be queued. */
140 BTRFS_FS_STATE_NO_DELAYED_IPUT,
141
142 /*
143 * Emergency shutdown, a step further than transaction aborted by
144 * rejecting all operations.
145 */
146 BTRFS_FS_STATE_EMERGENCY_SHUTDOWN,
147
148 BTRFS_FS_STATE_COUNT
149 };
150
151 enum {
152 BTRFS_FS_CLOSING_START,
153 BTRFS_FS_CLOSING_DONE,
154 BTRFS_FS_LOG_RECOVERING,
155 BTRFS_FS_OPEN,
156 BTRFS_FS_QUOTA_ENABLED,
157 BTRFS_FS_UPDATE_UUID_TREE_GEN,
158 BTRFS_FS_CREATING_FREE_SPACE_TREE,
159 BTRFS_FS_BTREE_ERR,
160 BTRFS_FS_LOG1_ERR,
161 BTRFS_FS_LOG2_ERR,
162 BTRFS_FS_QUOTA_OVERRIDE,
163 /* Used to record internally whether fs has been frozen */
164 BTRFS_FS_FROZEN,
165 /*
166 * Indicate that balance has been set up from the ioctl and is in the
167 * main phase. The fs_info::balance_ctl is initialized.
168 */
169 BTRFS_FS_BALANCE_RUNNING,
170
171 /*
172 * Indicate that relocation of a chunk has started, it's set per chunk
173 * and is toggled between chunks.
174 */
175 BTRFS_FS_RELOC_RUNNING,
176
177 /* Indicate that the cleaner thread is awake and doing something. */
178 BTRFS_FS_CLEANER_RUNNING,
179
180 /*
181 * The checksumming has an optimized version and is considered fast,
182 * so we don't need to offload checksums to workqueues.
183 */
184 BTRFS_FS_CSUM_IMPL_FAST,
185
186 /* Indicate that the discard workqueue can service discards. */
187 BTRFS_FS_DISCARD_RUNNING,
188
189 /* Indicate that we need to cleanup space cache v1 */
190 BTRFS_FS_CLEANUP_SPACE_CACHE_V1,
191
192 /* Indicate that we can't trust the free space tree for caching yet */
193 BTRFS_FS_FREE_SPACE_TREE_UNTRUSTED,
194
195 /* Indicate whether there are any tree modification log users */
196 BTRFS_FS_TREE_MOD_LOG_USERS,
197
198 /* Indicate that we want the transaction kthread to commit right now. */
199 BTRFS_FS_COMMIT_TRANS,
200
201 /* Indicate we have half completed snapshot deletions pending. */
202 BTRFS_FS_UNFINISHED_DROPS,
203
204 /* Indicate we have to finish a zone to do next allocation. */
205 BTRFS_FS_NEED_ZONE_FINISH,
206
207 /* Indicate that we want to commit the transaction. */
208 BTRFS_FS_NEED_TRANS_COMMIT,
209
210 /* This is set when active zone tracking is needed. */
211 BTRFS_FS_ACTIVE_ZONE_TRACKING,
212
213 /*
214 * Indicate if we have some features changed, this is mostly for
215 * cleaner thread to update the sysfs interface.
216 */
217 BTRFS_FS_FEATURE_CHANGED,
218
219 /*
220 * Indicate that we have found a tree block which is only aligned to
221 * sectorsize, but not to nodesize. This should be rare nowadays.
222 */
223 BTRFS_FS_UNALIGNED_TREE_BLOCK,
224
225 #if BITS_PER_LONG == 32
226 /* Indicate if we have error/warn message printed on 32bit systems */
227 BTRFS_FS_32BIT_ERROR,
228 BTRFS_FS_32BIT_WARN,
229 #endif
230 };
231
232 /*
233 * Flags for mount options.
234 *
235 * Note: don't forget to add new options to btrfs_show_options()
236 */
237 enum {
238 BTRFS_MOUNT_NODATASUM = (1ULL << 0),
239 BTRFS_MOUNT_NODATACOW = (1ULL << 1),
240 BTRFS_MOUNT_NOBARRIER = (1ULL << 2),
241 BTRFS_MOUNT_SSD = (1ULL << 3),
242 BTRFS_MOUNT_DEGRADED = (1ULL << 4),
243 BTRFS_MOUNT_COMPRESS = (1ULL << 5),
244 BTRFS_MOUNT_NOTREELOG = (1ULL << 6),
245 BTRFS_MOUNT_FLUSHONCOMMIT = (1ULL << 7),
246 BTRFS_MOUNT_SSD_SPREAD = (1ULL << 8),
247 BTRFS_MOUNT_NOSSD = (1ULL << 9),
248 BTRFS_MOUNT_DISCARD_SYNC = (1ULL << 10),
249 BTRFS_MOUNT_FORCE_COMPRESS = (1ULL << 11),
250 BTRFS_MOUNT_SPACE_CACHE = (1ULL << 12),
251 BTRFS_MOUNT_CLEAR_CACHE = (1ULL << 13),
252 BTRFS_MOUNT_USER_SUBVOL_RM_ALLOWED = (1ULL << 14),
253 BTRFS_MOUNT_ENOSPC_DEBUG = (1ULL << 15),
254 BTRFS_MOUNT_AUTO_DEFRAG = (1ULL << 16),
255 BTRFS_MOUNT_USEBACKUPROOT = (1ULL << 17),
256 BTRFS_MOUNT_SKIP_BALANCE = (1ULL << 18),
257 BTRFS_MOUNT_PANIC_ON_FATAL_ERROR = (1ULL << 19),
258 BTRFS_MOUNT_RESCAN_UUID_TREE = (1ULL << 20),
259 BTRFS_MOUNT_FRAGMENT_DATA = (1ULL << 21),
260 BTRFS_MOUNT_FRAGMENT_METADATA = (1ULL << 22),
261 BTRFS_MOUNT_FREE_SPACE_TREE = (1ULL << 23),
262 BTRFS_MOUNT_NOLOGREPLAY = (1ULL << 24),
263 BTRFS_MOUNT_REF_VERIFY = (1ULL << 25),
264 BTRFS_MOUNT_DISCARD_ASYNC = (1ULL << 26),
265 BTRFS_MOUNT_IGNOREBADROOTS = (1ULL << 27),
266 BTRFS_MOUNT_IGNOREDATACSUMS = (1ULL << 28),
267 BTRFS_MOUNT_NODISCARD = (1ULL << 29),
268 BTRFS_MOUNT_NOSPACECACHE = (1ULL << 30),
269 BTRFS_MOUNT_IGNOREMETACSUMS = (1ULL << 31),
270 BTRFS_MOUNT_IGNORESUPERFLAGS = (1ULL << 32),
271 BTRFS_MOUNT_REF_TRACKER = (1ULL << 33),
272 };
273
274 /* These mount options require a full read-only fs, no new transaction is allowed. */
275 #define BTRFS_MOUNT_FULL_RO_MASK \
276 (BTRFS_MOUNT_NOLOGREPLAY | \
277 BTRFS_MOUNT_IGNOREBADROOTS | \
278 BTRFS_MOUNT_IGNOREDATACSUMS | \
279 BTRFS_MOUNT_IGNOREMETACSUMS | \
280 BTRFS_MOUNT_IGNORESUPERFLAGS)
281
282 /*
283 * Compat flags that we support. If any incompat flags are set other than the
284 * ones specified below then we will fail to mount
285 */
286 #define BTRFS_FEATURE_COMPAT_SUPP 0ULL
287 #define BTRFS_FEATURE_COMPAT_SAFE_SET 0ULL
288 #define BTRFS_FEATURE_COMPAT_SAFE_CLEAR 0ULL
289
290 #define BTRFS_FEATURE_COMPAT_RO_SUPP \
291 (BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE | \
292 BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE_VALID | \
293 BTRFS_FEATURE_COMPAT_RO_VERITY | \
294 BTRFS_FEATURE_COMPAT_RO_BLOCK_GROUP_TREE)
295
296 #define BTRFS_FEATURE_COMPAT_RO_SAFE_SET 0ULL
297 #define BTRFS_FEATURE_COMPAT_RO_SAFE_CLEAR 0ULL
298
299 #define BTRFS_FEATURE_INCOMPAT_SUPP_STABLE \
300 (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \
301 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL | \
302 BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS | \
303 BTRFS_FEATURE_INCOMPAT_BIG_METADATA | \
304 BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO | \
305 BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD | \
306 BTRFS_FEATURE_INCOMPAT_RAID56 | \
307 BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF | \
308 BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA | \
309 BTRFS_FEATURE_INCOMPAT_NO_HOLES | \
310 BTRFS_FEATURE_INCOMPAT_METADATA_UUID | \
311 BTRFS_FEATURE_INCOMPAT_RAID1C34 | \
312 BTRFS_FEATURE_INCOMPAT_ZONED | \
313 BTRFS_FEATURE_INCOMPAT_SIMPLE_QUOTA)
314
315 #ifdef CONFIG_BTRFS_EXPERIMENTAL
316 /*
317 * Features under development like Extent tree v2 support is enabled
318 * only under CONFIG_BTRFS_EXPERIMENTAL
319 */
320 #define BTRFS_FEATURE_INCOMPAT_SUPP \
321 (BTRFS_FEATURE_INCOMPAT_SUPP_STABLE | \
322 BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE | \
323 BTRFS_FEATURE_INCOMPAT_EXTENT_TREE_V2 | \
324 BTRFS_FEATURE_INCOMPAT_REMAP_TREE)
325
326 #else
327
328 #define BTRFS_FEATURE_INCOMPAT_SUPP \
329 (BTRFS_FEATURE_INCOMPAT_SUPP_STABLE)
330
331 #endif
332
333 #define BTRFS_FEATURE_INCOMPAT_SAFE_SET \
334 (BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF)
335 #define BTRFS_FEATURE_INCOMPAT_SAFE_CLEAR 0ULL
336
337 #define BTRFS_DEFAULT_COMMIT_INTERVAL (30)
338 #define BTRFS_WARNING_COMMIT_INTERVAL (300)
339 #define BTRFS_DEFAULT_MAX_INLINE (2048)
340
341 enum btrfs_compression_type {
342 BTRFS_COMPRESS_NONE = 0,
343 BTRFS_COMPRESS_ZLIB = 1,
344 BTRFS_COMPRESS_LZO = 2,
345 BTRFS_COMPRESS_ZSTD = 3,
346 BTRFS_NR_COMPRESS_TYPES = 4,
347
348 BTRFS_DEFRAG_DONT_COMPRESS,
349 };
350
351 struct btrfs_dev_replace {
352 /* See #define above */
353 u64 replace_state;
354 /* Seconds since 1-Jan-1970 */
355 time64_t time_started;
356 /* Seconds since 1-Jan-1970 */
357 time64_t time_stopped;
358 atomic64_t num_write_errors;
359 atomic64_t num_uncorrectable_read_errors;
360
361 u64 cursor_left;
362 u64 committed_cursor_left;
363 u64 cursor_left_last_write_of_item;
364 u64 cursor_right;
365
366 /* See #define above */
367 u64 cont_reading_from_srcdev_mode;
368
369 int is_valid;
370 int item_needs_writeback;
371 struct btrfs_device *srcdev;
372 struct btrfs_device *tgtdev;
373
374 struct mutex lock_finishing_cancel_unmount;
375 struct rw_semaphore rwsem;
376
377 struct btrfs_scrub_progress scrub_progress;
378
379 struct percpu_counter bio_counter;
380 wait_queue_head_t replace_wait;
381
382 struct task_struct *replace_task;
383 };
384
385 /*
386 * Free clusters are used to claim free space in relatively large chunks,
387 * allowing us to do less seeky writes. They are used for all metadata
388 * allocations. In ssd_spread mode they are also used for data allocations.
389 */
390 struct btrfs_free_cluster {
391 spinlock_t lock;
392 spinlock_t refill_lock;
393 struct rb_root root;
394
395 /* Largest extent in this cluster */
396 u64 max_size;
397
398 /* First extent starting offset */
399 u64 window_start;
400
401 /* We did a full search and couldn't create a cluster */
402 bool fragmented;
403
404 struct btrfs_block_group *block_group;
405 /*
406 * When a cluster is allocated from a block group, we put the cluster
407 * onto a list in the block group so that it can be freed before the
408 * block group is freed.
409 */
410 struct list_head block_group_list;
411 };
412
413 /* Discard control. */
414 /*
415 * Async discard uses multiple lists to differentiate the discard filter
416 * parameters. Index 0 is for completely free block groups where we need to
417 * ensure the entire block group is trimmed without being lossy. Indices
418 * afterwards represent monotonically decreasing discard filter sizes to
419 * prioritize what should be discarded next.
420 */
421 #define BTRFS_NR_DISCARD_LISTS 3
422 #define BTRFS_DISCARD_INDEX_UNUSED 0
423 #define BTRFS_DISCARD_INDEX_START 1
424
425 struct btrfs_discard_ctl {
426 struct workqueue_struct *discard_workers;
427 struct delayed_work work;
428 spinlock_t lock;
429 struct btrfs_block_group *block_group;
430 struct list_head discard_list[BTRFS_NR_DISCARD_LISTS];
431 u64 prev_discard;
432 u64 prev_discard_time;
433 atomic_t discardable_extents;
434 atomic64_t discardable_bytes;
435 u64 max_discard_size;
436 u64 delay_ms;
437 u32 iops_limit;
438 u32 kbps_limit;
439 u64 discard_extent_bytes;
440 u64 discard_bitmap_bytes;
441 atomic64_t discard_bytes_saved;
442 };
443
444 /*
445 * Exclusive operations (device replace, resize, device add/remove, balance)
446 */
447 enum btrfs_exclusive_operation {
448 BTRFS_EXCLOP_NONE,
449 BTRFS_EXCLOP_BALANCE_PAUSED,
450 BTRFS_EXCLOP_BALANCE,
451 BTRFS_EXCLOP_DEV_ADD,
452 BTRFS_EXCLOP_DEV_REMOVE,
453 BTRFS_EXCLOP_DEV_REPLACE,
454 BTRFS_EXCLOP_RESIZE,
455 BTRFS_EXCLOP_SWAP_ACTIVATE,
456 };
457
458 /* Store data about transaction commits, exported via sysfs. */
459 struct btrfs_commit_stats {
460 /* Total number of commits */
461 u64 commit_count;
462 /* The maximum commit duration so far in ns */
463 u64 max_commit_dur;
464 /* The last commit duration in ns */
465 u64 last_commit_dur;
466 /* The total commit duration in ns */
467 u64 total_commit_dur;
468 /* Start of the last critical section in ns. */
469 u64 critical_section_start_time;
470 };
471
472 struct btrfs_delayed_root {
473 spinlock_t lock;
474 int nodes; /* for delayed nodes */
475 struct list_head node_list;
476 /*
477 * Used for delayed nodes which is waiting to be dealt with by the
478 * worker. If the delayed node is inserted into the work queue, we
479 * drop it from this list.
480 */
481 struct list_head prepare_list;
482 atomic_t items; /* for delayed items */
483 atomic_t items_seq; /* for delayed items */
484 wait_queue_head_t wait;
485 };
486
487 struct btrfs_fs_info {
488 u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
489 unsigned long flags;
490 struct btrfs_root *tree_root;
491 struct btrfs_root *chunk_root;
492 struct btrfs_root *dev_root;
493 struct btrfs_root *fs_root;
494 struct btrfs_root *quota_root;
495 struct btrfs_root *uuid_root;
496 struct btrfs_root *data_reloc_root;
497 struct btrfs_root *block_group_root;
498 struct btrfs_root *stripe_root;
499 struct btrfs_root *remap_root;
500
501 /* The log root tree is a directory of all the other log roots */
502 struct btrfs_root *log_root_tree;
503
504 /* The tree that holds the global roots (csum, extent, etc) */
505 rwlock_t global_root_lock;
506 struct rb_root global_root_tree;
507
508 spinlock_t fs_roots_radix_lock;
509 struct radix_tree_root fs_roots_radix;
510
511 /* Block group cache stuff */
512 rwlock_t block_group_cache_lock;
513 struct rb_root_cached block_group_cache_tree;
514
515 /* Keep track of unallocated space */
516 atomic64_t free_chunk_space;
517
518 /* Track ranges which are used by log trees blocks/logged data extents */
519 struct extent_io_tree excluded_extents;
520
521 /* logical->physical extent mapping */
522 struct rb_root_cached mapping_tree;
523 rwlock_t mapping_tree_lock;
524
525 /*
526 * Block reservation for extent, checksum, root tree and delayed dir
527 * index item.
528 */
529 struct btrfs_block_rsv global_block_rsv;
530 /* Block reservation for metadata operations */
531 struct btrfs_block_rsv trans_block_rsv;
532 /* Block reservation for chunk tree */
533 struct btrfs_block_rsv chunk_block_rsv;
534 /* Block reservation for remap tree. */
535 struct btrfs_block_rsv remap_block_rsv;
536 /* Block reservation for delayed operations */
537 struct btrfs_block_rsv delayed_block_rsv;
538 /* Block reservation for delayed refs */
539 struct btrfs_block_rsv delayed_refs_rsv;
540 /* Block reservation for treelog tree */
541 struct btrfs_block_rsv treelog_rsv;
542
543 struct btrfs_block_rsv empty_block_rsv;
544
545 /*
546 * Updated while holding the lock 'trans_lock'. Due to the life cycle of
547 * a transaction, it can be directly read while holding a transaction
548 * handle, everywhere else must be read with btrfs_get_fs_generation().
549 * Should always be updated using btrfs_set_fs_generation().
550 */
551 u64 generation;
552 /*
553 * Always use btrfs_get_last_trans_committed() and
554 * btrfs_set_last_trans_committed() to read and update this field.
555 */
556 u64 last_trans_committed;
557 /*
558 * Generation of the last transaction used for block group relocation
559 * since the filesystem was last mounted (or 0 if none happened yet).
560 * Must be written and read while holding btrfs_fs_info::commit_root_sem.
561 */
562 u64 last_reloc_trans;
563
564 /*
565 * This is updated to the current trans every time a full commit is
566 * required instead of the faster short fsync log commits
567 */
568 u64 last_trans_log_full_commit;
569 unsigned long long mount_opt;
570
571 /* Compress related structures. */
572 void *compr_wsm[BTRFS_NR_COMPRESS_TYPES];
573
574 int compress_type;
575 int compress_level;
576 u32 commit_interval;
577 /*
578 * It is a suggestive number, the read side is safe even it gets a
579 * wrong number because we will write out the data into a regular
580 * extent. The write side(mount/remount) is under ->s_umount lock,
581 * so it is also safe.
582 */
583 u64 max_inline;
584
585 struct btrfs_transaction *running_transaction;
586 wait_queue_head_t transaction_throttle;
587 wait_queue_head_t transaction_wait;
588 wait_queue_head_t transaction_blocked_wait;
589 wait_queue_head_t async_submit_wait;
590
591 /*
592 * Used to protect the incompat_flags, compat_flags, compat_ro_flags
593 * when they are updated.
594 *
595 * Because we do not clear the flags for ever, so we needn't use
596 * the lock on the read side.
597 *
598 * We also needn't use the lock when we mount the fs, because
599 * there is no other task which will update the flag.
600 */
601 spinlock_t super_lock;
602 struct btrfs_super_block *super_copy;
603 struct btrfs_super_block *super_for_commit;
604 struct super_block *sb;
605 struct inode *btree_inode;
606 struct mutex tree_log_mutex;
607 struct mutex transaction_kthread_mutex;
608 struct mutex cleaner_mutex;
609 struct mutex chunk_mutex;
610 struct mutex remap_mutex;
611
612 /*
613 * This is taken to make sure we don't set block groups ro after the
614 * free space cache has been allocated on them.
615 */
616 struct mutex ro_block_group_mutex;
617
618 /*
619 * This is used during read/modify/write to make sure no two ios are
620 * trying to mod the same stripe at the same time.
621 */
622 struct btrfs_stripe_hash_table *stripe_hash_table;
623
624 /*
625 * This protects the ordered operations list only while we are
626 * processing all of the entries on it. This way we make sure the
627 * commit code doesn't find the list temporarily empty because another
628 * function happens to be doing non-waiting preflush before jumping
629 * into the main commit.
630 */
631 struct mutex ordered_operations_mutex;
632
633 struct rw_semaphore commit_root_sem;
634
635 struct rw_semaphore cleanup_work_sem;
636
637 struct rw_semaphore subvol_sem;
638
639 spinlock_t trans_lock;
640 /*
641 * The reloc mutex goes with the trans lock, it is taken during commit
642 * to protect us from the relocation code.
643 */
644 struct mutex reloc_mutex;
645
646 struct list_head trans_list;
647 struct list_head dead_roots;
648 struct list_head caching_block_groups;
649
650 spinlock_t delayed_iput_lock;
651 struct list_head delayed_iputs;
652 atomic_t nr_delayed_iputs;
653 wait_queue_head_t delayed_iputs_wait;
654
655 atomic64_t tree_mod_seq;
656
657 /* This protects tree_mod_log and tree_mod_seq_list */
658 rwlock_t tree_mod_log_lock;
659 struct rb_root tree_mod_log;
660 struct list_head tree_mod_seq_list;
661
662 atomic_t async_delalloc_pages;
663
664 /* This is used to protect the following list -- ordered_roots. */
665 spinlock_t ordered_root_lock;
666
667 /*
668 * All fs/file tree roots in which there are data=ordered extents
669 * pending writeback are added into this list.
670 *
671 * These can span multiple transactions and basically include every
672 * dirty data page that isn't from nodatacow.
673 */
674 struct list_head ordered_roots;
675
676 struct mutex delalloc_root_mutex;
677 spinlock_t delalloc_root_lock;
678 /* All fs/file tree roots that have delalloc inodes. */
679 struct list_head delalloc_roots;
680
681 /*
682 * There is a pool of worker threads for checksumming during writes and
683 * a pool for checksumming after reads. This is because readers can
684 * run with FS locks held, and the writers may be waiting for those
685 * locks. We don't want ordering in the pending list to cause
686 * deadlocks, and so the two are serviced separately.
687 *
688 * A third pool does submit_bio to avoid deadlocking with the other two.
689 */
690 struct btrfs_workqueue *workers;
691 struct btrfs_workqueue *delalloc_workers;
692 struct btrfs_workqueue *flush_workers;
693 struct workqueue_struct *endio_workers;
694 struct workqueue_struct *endio_meta_workers;
695 struct workqueue_struct *rmw_workers;
696 struct btrfs_workqueue *endio_write_workers;
697 struct btrfs_workqueue *endio_freespace_worker;
698 struct btrfs_workqueue *caching_workers;
699
700 /*
701 * Fixup workers take dirty pages that didn't properly go through the
702 * cow mechanism and make them safe to write. It happens for the
703 * sys_munmap function call path.
704 */
705 struct btrfs_workqueue *fixup_workers;
706 struct btrfs_workqueue *delayed_workers;
707
708 struct task_struct *transaction_kthread;
709 struct task_struct *cleaner_kthread;
710 u32 thread_pool_size;
711
712 struct kobject *space_info_kobj;
713 struct kobject *qgroups_kobj;
714 struct kobject *discard_kobj;
715
716 /* Track the number of blocks (sectors) read by the filesystem. */
717 struct percpu_counter stats_read_blocks;
718
719 /* Used to keep from writing metadata until there is a nice batch */
720 struct percpu_counter dirty_metadata_bytes;
721 struct percpu_counter delalloc_bytes;
722 struct percpu_counter ordered_bytes;
723 s32 dirty_metadata_batch;
724 s32 delalloc_batch;
725
726 struct percpu_counter evictable_extent_maps;
727 u64 em_shrinker_last_root;
728 u64 em_shrinker_last_ino;
729 atomic64_t em_shrinker_nr_to_scan;
730 struct work_struct em_shrinker_work;
731
732 /* Protected by 'trans_lock'. */
733 struct list_head dirty_cowonly_roots;
734
735 struct btrfs_fs_devices *fs_devices;
736
737 /*
738 * The space_info list is effectively read only after initial setup.
739 * It is populated at mount time and cleaned up after all block groups
740 * are removed. RCU is used to protect it.
741 */
742 struct list_head space_info;
743
744 struct btrfs_space_info *data_sinfo;
745
746 struct reloc_control *reloc_ctl;
747
748 /* data_alloc_cluster is only used in ssd_spread mode */
749 struct btrfs_free_cluster data_alloc_cluster;
750
751 /* All metadata allocations go through this cluster. */
752 struct btrfs_free_cluster meta_alloc_cluster;
753
754 /* Auto defrag inodes go here. */
755 spinlock_t defrag_inodes_lock;
756 struct rb_root defrag_inodes;
757 atomic_t defrag_running;
758
759 /* Used to protect avail_{data, metadata, system}_alloc_bits */
760 seqlock_t profiles_lock;
761 /*
762 * These three are in extended format (availability of single chunks is
763 * denoted by BTRFS_AVAIL_ALLOC_BIT_SINGLE bit, other types are denoted
764 * by corresponding BTRFS_BLOCK_GROUP_* bits)
765 */
766 u64 avail_data_alloc_bits;
767 u64 avail_metadata_alloc_bits;
768 u64 avail_system_alloc_bits;
769
770 /* Balance state */
771 spinlock_t balance_lock;
772 struct mutex balance_mutex;
773 atomic_t balance_pause_req;
774 atomic_t balance_cancel_req;
775 struct btrfs_balance_control *balance_ctl;
776 wait_queue_head_t balance_wait_q;
777
778 /* Cancellation requests for chunk relocation */
779 atomic_t reloc_cancel_req;
780
781 u32 data_chunk_allocations;
782 u32 metadata_ratio;
783
784 /* Private scrub information */
785 struct mutex scrub_lock;
786 atomic_t scrubs_running;
787 atomic_t scrub_pause_req;
788 atomic_t scrubs_paused;
789 atomic_t scrub_cancel_req;
790 wait_queue_head_t scrub_pause_wait;
791 /*
792 * The worker pointers are NULL iff the refcount is 0, ie. scrub is not
793 * running.
794 */
795 refcount_t scrub_workers_refcnt;
796 struct workqueue_struct *scrub_workers;
797
798 struct btrfs_discard_ctl discard_ctl;
799
800 /* Is qgroup tracking in a consistent state? */
801 u64 qgroup_flags;
802
803 /* Holds configuration and tracking. Protected by qgroup_lock. */
804 struct rb_root qgroup_tree;
805 spinlock_t qgroup_lock;
806
807 /*
808 * Protect user change for quota operations. If a transaction is needed,
809 * it must be started before locking this lock.
810 */
811 struct mutex qgroup_ioctl_lock;
812
813 /* List of dirty qgroups to be written at next commit. */
814 struct list_head dirty_qgroups;
815
816 /* Used by qgroup for an efficient tree traversal. */
817 u64 qgroup_seq;
818
819 /* Qgroup rescan items. */
820 /* Protects the progress item */
821 struct mutex qgroup_rescan_lock;
822 struct btrfs_key qgroup_rescan_progress;
823 struct btrfs_workqueue *qgroup_rescan_workers;
824 struct completion qgroup_rescan_completion;
825 struct btrfs_work qgroup_rescan_work;
826 /* Protected by qgroup_rescan_lock */
827 bool qgroup_rescan_running;
828 u8 qgroup_drop_subtree_thres;
829 u64 qgroup_enable_gen;
830
831 /*
832 * If this is not 0, then it indicates a serious filesystem error has
833 * happened and it contains that error (negative errno value).
834 */
835 int fs_error;
836
837 /* Filesystem state */
838 unsigned long fs_state;
839
840 struct btrfs_delayed_root delayed_root;
841
842 /* Entries are eb->start >> nodesize_bits */
843 struct xarray buffer_tree;
844
845 /* Next backup root to be overwritten */
846 int backup_root_index;
847
848 /* Device replace state */
849 struct btrfs_dev_replace dev_replace;
850
851 struct semaphore uuid_tree_rescan_sem;
852
853 /* Used to reclaim the metadata space in the background. */
854 struct work_struct async_reclaim_work;
855 struct work_struct async_data_reclaim_work;
856 struct work_struct preempt_reclaim_work;
857
858 /* Reclaim partially filled block groups in the background */
859 struct work_struct reclaim_bgs_work;
860 /* Protected by unused_bgs_lock. */
861 struct list_head reclaim_bgs;
862 int bg_reclaim_threshold;
863
864 /* Protects the lists unused_bgs, reclaim_bgs, and fully_remapped_bgs. */
865 spinlock_t unused_bgs_lock;
866 /* Protected by unused_bgs_lock. */
867 struct list_head unused_bgs;
868 struct list_head fully_remapped_bgs;
869 struct mutex unused_bg_unpin_mutex;
870 /* Protect block groups that are going to be deleted */
871 struct mutex reclaim_bgs_lock;
872
873 /* Cached block sizes */
874 u32 nodesize;
875 u32 nodesize_bits;
876 u32 sectorsize;
877 /* ilog2 of sectorsize, use to avoid 64bit division */
878 u32 sectorsize_bits;
879 u32 block_min_order;
880 u32 block_max_order;
881 u32 stripesize;
882 u32 csum_size;
883 u32 csums_per_leaf;
884 u32 csum_type;
885
886 /*
887 * Maximum size of an extent. BTRFS_MAX_EXTENT_SIZE on regular
888 * filesystem, on zoned it depends on the device constraints.
889 */
890 u64 max_extent_size;
891
892 /* Block groups and devices containing active swapfiles. */
893 spinlock_t swapfile_pins_lock;
894 struct rb_root swapfile_pins;
895
896 /* Type of exclusive operation running, protected by super_lock */
897 enum btrfs_exclusive_operation exclusive_operation;
898
899 /*
900 * Zone size > 0 when in ZONED mode, otherwise it's used for a check
901 * if the mode is enabled
902 */
903 u64 zone_size;
904
905 /* Constraints for ZONE_APPEND commands: */
906 struct queue_limits limits;
907 u64 max_zone_append_size;
908
909 struct mutex zoned_meta_io_lock;
910 spinlock_t treelog_bg_lock;
911 u64 treelog_bg;
912
913 /*
914 * Start of the dedicated data relocation block group, protected by
915 * relocation_bg_lock.
916 */
917 spinlock_t relocation_bg_lock;
918 u64 data_reloc_bg;
919 struct mutex zoned_data_reloc_io_lock;
920
921 struct btrfs_block_group *active_meta_bg;
922 struct btrfs_block_group *active_system_bg;
923
924 u64 nr_global_roots;
925
926 spinlock_t zone_active_bgs_lock;
927 struct list_head zone_active_bgs;
928
929 /* Updates are not protected by any lock */
930 struct btrfs_commit_stats commit_stats;
931
932 /*
933 * Last generation where we dropped a non-relocation root.
934 * Use btrfs_set_last_root_drop_gen() and btrfs_get_last_root_drop_gen()
935 * to change it and to read it, respectively.
936 */
937 u64 last_root_drop_gen;
938
939 /*
940 * Annotations for transaction events (structures are empty when
941 * compiled without lockdep).
942 */
943 struct lockdep_map btrfs_trans_num_writers_map;
944 struct lockdep_map btrfs_trans_num_extwriters_map;
945 struct lockdep_map btrfs_state_change_map[4];
946 struct lockdep_map btrfs_trans_pending_ordered_map;
947 struct lockdep_map btrfs_ordered_extent_map;
948
949 #ifdef CONFIG_BTRFS_DEBUG
950 spinlock_t ref_verify_lock;
951 struct rb_root block_tree;
952
953 struct kobject *debug_kobj;
954 struct list_head allocated_roots;
955
956 spinlock_t eb_leak_lock;
957 struct list_head allocated_ebs;
958 #endif
959 };
960
961 #define folio_to_inode(_folio) (BTRFS_I(_Generic((_folio), \
962 struct folio *: (_folio))->mapping->host))
963
964 #define folio_to_fs_info(_folio) (folio_to_inode(_folio)->root->fs_info)
965
966 #define inode_to_fs_info(_inode) (BTRFS_I(_Generic((_inode), \
967 struct inode *: (_inode)))->root->fs_info)
968
btrfs_alloc_write_mask(struct address_space * mapping)969 static inline gfp_t btrfs_alloc_write_mask(struct address_space *mapping)
970 {
971 return mapping_gfp_constraint(mapping, ~__GFP_FS);
972 }
973
974 /* Return the minimal folio size of the fs. */
btrfs_min_folio_size(struct btrfs_fs_info * fs_info)975 static inline unsigned int btrfs_min_folio_size(struct btrfs_fs_info *fs_info)
976 {
977 return 1U << (PAGE_SHIFT + fs_info->block_min_order);
978 }
979
btrfs_get_fs_generation(const struct btrfs_fs_info * fs_info)980 static inline u64 btrfs_get_fs_generation(const struct btrfs_fs_info *fs_info)
981 {
982 return READ_ONCE(fs_info->generation);
983 }
984
btrfs_set_fs_generation(struct btrfs_fs_info * fs_info,u64 gen)985 static inline void btrfs_set_fs_generation(struct btrfs_fs_info *fs_info, u64 gen)
986 {
987 WRITE_ONCE(fs_info->generation, gen);
988 }
989
btrfs_get_last_trans_committed(const struct btrfs_fs_info * fs_info)990 static inline u64 btrfs_get_last_trans_committed(const struct btrfs_fs_info *fs_info)
991 {
992 return READ_ONCE(fs_info->last_trans_committed);
993 }
994
btrfs_set_last_trans_committed(struct btrfs_fs_info * fs_info,u64 gen)995 static inline void btrfs_set_last_trans_committed(struct btrfs_fs_info *fs_info, u64 gen)
996 {
997 WRITE_ONCE(fs_info->last_trans_committed, gen);
998 }
999
btrfs_set_last_root_drop_gen(struct btrfs_fs_info * fs_info,u64 gen)1000 static inline void btrfs_set_last_root_drop_gen(struct btrfs_fs_info *fs_info,
1001 u64 gen)
1002 {
1003 WRITE_ONCE(fs_info->last_root_drop_gen, gen);
1004 }
1005
btrfs_get_last_root_drop_gen(const struct btrfs_fs_info * fs_info)1006 static inline u64 btrfs_get_last_root_drop_gen(const struct btrfs_fs_info *fs_info)
1007 {
1008 return READ_ONCE(fs_info->last_root_drop_gen);
1009 }
1010
1011 /*
1012 * Take the number of bytes to be checksummed and figure out how many leaves
1013 * it would require to store the csums for that many bytes.
1014 */
btrfs_csum_bytes_to_leaves(const struct btrfs_fs_info * fs_info,u64 csum_bytes)1015 static inline u64 btrfs_csum_bytes_to_leaves(
1016 const struct btrfs_fs_info *fs_info, u64 csum_bytes)
1017 {
1018 const u64 num_csums = csum_bytes >> fs_info->sectorsize_bits;
1019
1020 return DIV_ROUND_UP_ULL(num_csums, fs_info->csums_per_leaf);
1021 }
1022
1023 /*
1024 * Use this if we would be adding new items, as we could split nodes as we cow
1025 * down the tree.
1026 */
btrfs_calc_insert_metadata_size(const struct btrfs_fs_info * fs_info,unsigned num_items)1027 static inline u64 btrfs_calc_insert_metadata_size(const struct btrfs_fs_info *fs_info,
1028 unsigned num_items)
1029 {
1030 return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * 2 * num_items;
1031 }
1032
1033 /*
1034 * Doing a truncate or a modification won't result in new nodes or leaves, just
1035 * what we need for COW.
1036 */
btrfs_calc_metadata_size(const struct btrfs_fs_info * fs_info,unsigned num_items)1037 static inline u64 btrfs_calc_metadata_size(const struct btrfs_fs_info *fs_info,
1038 unsigned num_items)
1039 {
1040 return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * num_items;
1041 }
1042
1043 #define BTRFS_MAX_EXTENT_ITEM_SIZE(r) ((BTRFS_LEAF_DATA_SIZE(r->fs_info) >> 4) - \
1044 sizeof(struct btrfs_item))
1045
1046 #define BTRFS_BYTES_TO_BLKS(fs_info, bytes) ((bytes) >> (fs_info)->sectorsize_bits)
1047
btrfs_is_zoned(const struct btrfs_fs_info * fs_info)1048 static inline bool btrfs_is_zoned(const struct btrfs_fs_info *fs_info)
1049 {
1050 return IS_ENABLED(CONFIG_BLK_DEV_ZONED) && fs_info->zone_size > 0;
1051 }
1052
1053 /*
1054 * Count how many fs_info->max_extent_size cover the @size
1055 */
count_max_extents(const struct btrfs_fs_info * fs_info,u64 size)1056 static inline u32 count_max_extents(const struct btrfs_fs_info *fs_info, u64 size)
1057 {
1058 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
1059 if (!fs_info)
1060 return div_u64(size + BTRFS_MAX_EXTENT_SIZE - 1, BTRFS_MAX_EXTENT_SIZE);
1061 #endif
1062
1063 return div_u64(size + fs_info->max_extent_size - 1, fs_info->max_extent_size);
1064 }
1065
btrfs_blocks_per_folio(const struct btrfs_fs_info * fs_info,const struct folio * folio)1066 static inline unsigned int btrfs_blocks_per_folio(const struct btrfs_fs_info *fs_info,
1067 const struct folio *folio)
1068 {
1069 return folio_size(folio) >> fs_info->sectorsize_bits;
1070 }
1071
1072 bool __attribute_const__ btrfs_supported_blocksize(u32 blocksize);
1073 bool btrfs_exclop_start(struct btrfs_fs_info *fs_info,
1074 enum btrfs_exclusive_operation type);
1075 bool btrfs_exclop_start_try_lock(struct btrfs_fs_info *fs_info,
1076 enum btrfs_exclusive_operation type);
1077 void btrfs_exclop_start_unlock(struct btrfs_fs_info *fs_info);
1078 void btrfs_exclop_finish(struct btrfs_fs_info *fs_info);
1079 void btrfs_exclop_balance(struct btrfs_fs_info *fs_info,
1080 enum btrfs_exclusive_operation op);
1081
1082 int btrfs_check_ioctl_vol_args_path(const struct btrfs_ioctl_vol_args *vol_args);
1083
1084 u16 btrfs_csum_type_size(u16 type);
1085 int btrfs_super_csum_size(const struct btrfs_super_block *s);
1086 const char *btrfs_super_csum_name(u16 csum_type);
1087 size_t __attribute_const__ btrfs_get_num_csums(void);
1088 struct btrfs_csum_ctx {
1089 u16 csum_type;
1090 union {
1091 u32 crc32;
1092 struct xxh64_state xxh64;
1093 struct sha256_ctx sha256;
1094 struct blake2b_ctx blake2b;
1095 };
1096 };
1097 void btrfs_csum(u16 csum_type, const u8 *data, size_t len, u8 *out);
1098 void btrfs_csum_init(struct btrfs_csum_ctx *ctx, u16 csum_type);
1099 void btrfs_csum_update(struct btrfs_csum_ctx *ctx, const u8 *data, size_t len);
1100 void btrfs_csum_final(struct btrfs_csum_ctx *ctx, u8 *out);
1101
btrfs_is_empty_uuid(const u8 * uuid)1102 static inline bool btrfs_is_empty_uuid(const u8 *uuid)
1103 {
1104 return uuid_is_null((const uuid_t *)uuid);
1105 }
1106
1107 /* Compatibility and incompatibility defines */
1108 void __btrfs_set_fs_incompat(struct btrfs_fs_info *fs_info, u64 flag,
1109 const char *name);
1110 void __btrfs_clear_fs_incompat(struct btrfs_fs_info *fs_info, u64 flag,
1111 const char *name);
1112 void __btrfs_set_fs_compat_ro(struct btrfs_fs_info *fs_info, u64 flag,
1113 const char *name);
1114 void __btrfs_clear_fs_compat_ro(struct btrfs_fs_info *fs_info, u64 flag,
1115 const char *name);
1116
1117 #define __btrfs_fs_incompat(fs_info, flags) \
1118 (!!(btrfs_super_incompat_flags((fs_info)->super_copy) & (flags)))
1119
1120 #define __btrfs_fs_compat_ro(fs_info, flags) \
1121 (!!(btrfs_super_compat_ro_flags((fs_info)->super_copy) & (flags)))
1122
1123 #define btrfs_set_fs_incompat(__fs_info, opt) \
1124 __btrfs_set_fs_incompat((__fs_info), BTRFS_FEATURE_INCOMPAT_##opt, #opt)
1125
1126 #define btrfs_clear_fs_incompat(__fs_info, opt) \
1127 __btrfs_clear_fs_incompat((__fs_info), BTRFS_FEATURE_INCOMPAT_##opt, #opt)
1128
1129 #define btrfs_fs_incompat(fs_info, opt) \
1130 __btrfs_fs_incompat((fs_info), BTRFS_FEATURE_INCOMPAT_##opt)
1131
1132 #define btrfs_set_fs_compat_ro(__fs_info, opt) \
1133 __btrfs_set_fs_compat_ro((__fs_info), BTRFS_FEATURE_COMPAT_RO_##opt, #opt)
1134
1135 #define btrfs_clear_fs_compat_ro(__fs_info, opt) \
1136 __btrfs_clear_fs_compat_ro((__fs_info), BTRFS_FEATURE_COMPAT_RO_##opt, #opt)
1137
1138 #define btrfs_fs_compat_ro(fs_info, opt) \
1139 __btrfs_fs_compat_ro((fs_info), BTRFS_FEATURE_COMPAT_RO_##opt)
1140
1141 #define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt)
1142 #define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt)
1143 #define btrfs_raw_test_opt(o, opt) ((o) & BTRFS_MOUNT_##opt)
1144 #define btrfs_test_opt(fs_info, opt) ((fs_info)->mount_opt & \
1145 BTRFS_MOUNT_##opt)
1146
btrfs_fs_closing(const struct btrfs_fs_info * fs_info)1147 static inline bool btrfs_fs_closing(const struct btrfs_fs_info *fs_info)
1148 {
1149 return unlikely(test_bit(BTRFS_FS_CLOSING_START, &fs_info->flags));
1150 }
1151
btrfs_fs_closing_done(const struct btrfs_fs_info * fs_info)1152 static inline bool btrfs_fs_closing_done(const struct btrfs_fs_info *fs_info)
1153 {
1154 if (btrfs_fs_closing(fs_info) && test_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags))
1155 return true;
1156
1157 return false;
1158 }
1159
1160 /*
1161 * If we remount the fs to be R/O or umount the fs, the cleaner needn't do
1162 * anything except sleeping. This function is used to check the status of
1163 * the fs.
1164 * We check for BTRFS_FS_STATE_RO to avoid races with a concurrent remount,
1165 * since setting and checking for SB_RDONLY in the superblock's flags is not
1166 * atomic.
1167 */
btrfs_need_cleaner_sleep(const struct btrfs_fs_info * fs_info)1168 static inline int btrfs_need_cleaner_sleep(const struct btrfs_fs_info *fs_info)
1169 {
1170 return test_bit(BTRFS_FS_STATE_RO, &fs_info->fs_state) ||
1171 btrfs_fs_closing(fs_info);
1172 }
1173
btrfs_wake_unfinished_drop(struct btrfs_fs_info * fs_info)1174 static inline void btrfs_wake_unfinished_drop(struct btrfs_fs_info *fs_info)
1175 {
1176 clear_and_wake_up_bit(BTRFS_FS_UNFINISHED_DROPS, &fs_info->flags);
1177 }
1178
1179 #define BTRFS_FS_ERROR(fs_info) (READ_ONCE((fs_info)->fs_error))
1180
1181 #define BTRFS_FS_LOG_CLEANUP_ERROR(fs_info) \
1182 (unlikely(test_bit(BTRFS_FS_STATE_LOG_CLEANUP_ERROR, \
1183 &(fs_info)->fs_state)))
1184
btrfs_is_shutdown(const struct btrfs_fs_info * fs_info)1185 static inline bool btrfs_is_shutdown(const struct btrfs_fs_info *fs_info)
1186 {
1187 return unlikely(test_bit(BTRFS_FS_STATE_EMERGENCY_SHUTDOWN, &fs_info->fs_state));
1188 }
1189
btrfs_force_shutdown(struct btrfs_fs_info * fs_info)1190 static inline void btrfs_force_shutdown(struct btrfs_fs_info *fs_info)
1191 {
1192 /*
1193 * Here we do not want to use handle_fs_error(), which will mark the fs
1194 * read-only.
1195 * Some call sites like shutdown ioctl will mark the fs shutdown when
1196 * the fs is frozen. But thaw path will handle RO and RW fs
1197 * differently.
1198 *
1199 * So here we only mark the fs error without flipping it RO.
1200 */
1201 WRITE_ONCE(fs_info->fs_error, -EIO);
1202 if (!test_and_set_bit(BTRFS_FS_STATE_EMERGENCY_SHUTDOWN, &fs_info->fs_state))
1203 btrfs_crit(fs_info, "emergency shutdown");
1204 }
1205
1206 /*
1207 * We use folio flag owner_2 to indicate there is an ordered extent with
1208 * unfinished IO.
1209 */
1210 #define folio_test_ordered(folio) folio_test_owner_2(folio)
1211 #define folio_set_ordered(folio) folio_set_owner_2(folio)
1212 #define folio_clear_ordered(folio) folio_clear_owner_2(folio)
1213
1214 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
1215
1216 #define EXPORT_FOR_TESTS
1217
btrfs_is_testing(const struct btrfs_fs_info * fs_info)1218 static inline bool btrfs_is_testing(const struct btrfs_fs_info *fs_info)
1219 {
1220 return unlikely(test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state));
1221 }
1222
1223 void btrfs_test_destroy_inode(struct inode *inode);
1224
1225 #else
1226
1227 #define EXPORT_FOR_TESTS static
1228
btrfs_is_testing(const struct btrfs_fs_info * fs_info)1229 static inline bool btrfs_is_testing(const struct btrfs_fs_info *fs_info)
1230 {
1231 return false;
1232 }
1233 #endif
1234
1235 #endif
1236