1 /* SPDX-License-Identifier: GPL-2.0 */
2
3 #ifndef BTRFS_FS_H
4 #define BTRFS_FS_H
5
6 #include <linux/blkdev.h>
7 #include <linux/fs.h>
8 #include <linux/btrfs_tree.h>
9 #include <linux/sizes.h>
10 #include "extent-io-tree.h"
11 #include "extent_map.h"
12 #include "async-thread.h"
13 #include "block-rsv.h"
14
15 #define BTRFS_MAX_EXTENT_SIZE SZ_128M
16
17 #define BTRFS_OLDEST_GENERATION 0ULL
18
19 #define BTRFS_EMPTY_DIR_SIZE 0
20
21 #define BTRFS_DIRTY_METADATA_THRESH SZ_32M
22
23 #define BTRFS_SUPER_INFO_OFFSET SZ_64K
24 #define BTRFS_SUPER_INFO_SIZE 4096
25 static_assert(sizeof(struct btrfs_super_block) == BTRFS_SUPER_INFO_SIZE);
26
27 /*
28 * The reserved space at the beginning of each device. It covers the primary
29 * super block and leaves space for potential use by other tools like
30 * bootloaders or to lower potential damage of accidental overwrite.
31 */
32 #define BTRFS_DEVICE_RANGE_RESERVED (SZ_1M)
33 /*
34 * Runtime (in-memory) states of filesystem
35 */
36 enum {
37 /* Global indicator of serious filesystem errors */
38 BTRFS_FS_STATE_ERROR,
39 /*
40 * Filesystem is being remounted, allow to skip some operations, like
41 * defrag
42 */
43 BTRFS_FS_STATE_REMOUNTING,
44 /* Filesystem in RO mode */
45 BTRFS_FS_STATE_RO,
46 /* Track if a transaction abort has been reported on this filesystem */
47 BTRFS_FS_STATE_TRANS_ABORTED,
48 /*
49 * Bio operations should be blocked on this filesystem because a source
50 * or target device is being destroyed as part of a device replace
51 */
52 BTRFS_FS_STATE_DEV_REPLACING,
53 /* The btrfs_fs_info created for self-tests */
54 BTRFS_FS_STATE_DUMMY_FS_INFO,
55
56 BTRFS_FS_STATE_NO_CSUMS,
57
58 /* Indicates there was an error cleaning up a log tree. */
59 BTRFS_FS_STATE_LOG_CLEANUP_ERROR,
60
61 BTRFS_FS_STATE_COUNT
62 };
63
64 enum {
65 BTRFS_FS_CLOSING_START,
66 BTRFS_FS_CLOSING_DONE,
67 BTRFS_FS_LOG_RECOVERING,
68 BTRFS_FS_OPEN,
69 BTRFS_FS_QUOTA_ENABLED,
70 BTRFS_FS_UPDATE_UUID_TREE_GEN,
71 BTRFS_FS_CREATING_FREE_SPACE_TREE,
72 BTRFS_FS_BTREE_ERR,
73 BTRFS_FS_LOG1_ERR,
74 BTRFS_FS_LOG2_ERR,
75 BTRFS_FS_QUOTA_OVERRIDE,
76 /* Used to record internally whether fs has been frozen */
77 BTRFS_FS_FROZEN,
78 /*
79 * Indicate that balance has been set up from the ioctl and is in the
80 * main phase. The fs_info::balance_ctl is initialized.
81 */
82 BTRFS_FS_BALANCE_RUNNING,
83
84 /*
85 * Indicate that relocation of a chunk has started, it's set per chunk
86 * and is toggled between chunks.
87 */
88 BTRFS_FS_RELOC_RUNNING,
89
90 /* Indicate that the cleaner thread is awake and doing something. */
91 BTRFS_FS_CLEANER_RUNNING,
92
93 /*
94 * The checksumming has an optimized version and is considered fast,
95 * so we don't need to offload checksums to workqueues.
96 */
97 BTRFS_FS_CSUM_IMPL_FAST,
98
99 /* Indicate that the discard workqueue can service discards. */
100 BTRFS_FS_DISCARD_RUNNING,
101
102 /* Indicate that we need to cleanup space cache v1 */
103 BTRFS_FS_CLEANUP_SPACE_CACHE_V1,
104
105 /* Indicate that we can't trust the free space tree for caching yet */
106 BTRFS_FS_FREE_SPACE_TREE_UNTRUSTED,
107
108 /* Indicate whether there are any tree modification log users */
109 BTRFS_FS_TREE_MOD_LOG_USERS,
110
111 /* Indicate that we want the transaction kthread to commit right now. */
112 BTRFS_FS_COMMIT_TRANS,
113
114 /* Indicate we have half completed snapshot deletions pending. */
115 BTRFS_FS_UNFINISHED_DROPS,
116
117 /* Indicate we have to finish a zone to do next allocation. */
118 BTRFS_FS_NEED_ZONE_FINISH,
119
120 /* Indicate that we want to commit the transaction. */
121 BTRFS_FS_NEED_TRANS_COMMIT,
122
123 /*
124 * Indicate metadata over-commit is disabled. This is set when active
125 * zone tracking is needed.
126 */
127 BTRFS_FS_NO_OVERCOMMIT,
128
129 /*
130 * Indicate if we have some features changed, this is mostly for
131 * cleaner thread to update the sysfs interface.
132 */
133 BTRFS_FS_FEATURE_CHANGED,
134
135 #if BITS_PER_LONG == 32
136 /* Indicate if we have error/warn message printed on 32bit systems */
137 BTRFS_FS_32BIT_ERROR,
138 BTRFS_FS_32BIT_WARN,
139 #endif
140 };
141
142 /*
143 * Flags for mount options.
144 *
145 * Note: don't forget to add new options to btrfs_show_options()
146 */
147 enum {
148 BTRFS_MOUNT_NODATASUM = (1UL << 0),
149 BTRFS_MOUNT_NODATACOW = (1UL << 1),
150 BTRFS_MOUNT_NOBARRIER = (1UL << 2),
151 BTRFS_MOUNT_SSD = (1UL << 3),
152 BTRFS_MOUNT_DEGRADED = (1UL << 4),
153 BTRFS_MOUNT_COMPRESS = (1UL << 5),
154 BTRFS_MOUNT_NOTREELOG = (1UL << 6),
155 BTRFS_MOUNT_FLUSHONCOMMIT = (1UL << 7),
156 BTRFS_MOUNT_SSD_SPREAD = (1UL << 8),
157 BTRFS_MOUNT_NOSSD = (1UL << 9),
158 BTRFS_MOUNT_DISCARD_SYNC = (1UL << 10),
159 BTRFS_MOUNT_FORCE_COMPRESS = (1UL << 11),
160 BTRFS_MOUNT_SPACE_CACHE = (1UL << 12),
161 BTRFS_MOUNT_CLEAR_CACHE = (1UL << 13),
162 BTRFS_MOUNT_USER_SUBVOL_RM_ALLOWED = (1UL << 14),
163 BTRFS_MOUNT_ENOSPC_DEBUG = (1UL << 15),
164 BTRFS_MOUNT_AUTO_DEFRAG = (1UL << 16),
165 BTRFS_MOUNT_USEBACKUPROOT = (1UL << 17),
166 BTRFS_MOUNT_SKIP_BALANCE = (1UL << 18),
167 BTRFS_MOUNT_CHECK_INTEGRITY = (1UL << 19),
168 BTRFS_MOUNT_CHECK_INTEGRITY_DATA = (1UL << 20),
169 BTRFS_MOUNT_PANIC_ON_FATAL_ERROR = (1UL << 21),
170 BTRFS_MOUNT_RESCAN_UUID_TREE = (1UL << 22),
171 BTRFS_MOUNT_FRAGMENT_DATA = (1UL << 23),
172 BTRFS_MOUNT_FRAGMENT_METADATA = (1UL << 24),
173 BTRFS_MOUNT_FREE_SPACE_TREE = (1UL << 25),
174 BTRFS_MOUNT_NOLOGREPLAY = (1UL << 26),
175 BTRFS_MOUNT_REF_VERIFY = (1UL << 27),
176 BTRFS_MOUNT_DISCARD_ASYNC = (1UL << 28),
177 BTRFS_MOUNT_IGNOREBADROOTS = (1UL << 29),
178 BTRFS_MOUNT_IGNOREDATACSUMS = (1UL << 30),
179 BTRFS_MOUNT_NODISCARD = (1UL << 31),
180 };
181
182 /*
183 * Compat flags that we support. If any incompat flags are set other than the
184 * ones specified below then we will fail to mount
185 */
186 #define BTRFS_FEATURE_COMPAT_SUPP 0ULL
187 #define BTRFS_FEATURE_COMPAT_SAFE_SET 0ULL
188 #define BTRFS_FEATURE_COMPAT_SAFE_CLEAR 0ULL
189
190 #define BTRFS_FEATURE_COMPAT_RO_SUPP \
191 (BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE | \
192 BTRFS_FEATURE_COMPAT_RO_FREE_SPACE_TREE_VALID | \
193 BTRFS_FEATURE_COMPAT_RO_VERITY | \
194 BTRFS_FEATURE_COMPAT_RO_BLOCK_GROUP_TREE)
195
196 #define BTRFS_FEATURE_COMPAT_RO_SAFE_SET 0ULL
197 #define BTRFS_FEATURE_COMPAT_RO_SAFE_CLEAR 0ULL
198
199 #ifdef CONFIG_BTRFS_DEBUG
200 /*
201 * Extent tree v2 supported only with CONFIG_BTRFS_DEBUG
202 */
203 #define BTRFS_FEATURE_INCOMPAT_SUPP \
204 (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \
205 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL | \
206 BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS | \
207 BTRFS_FEATURE_INCOMPAT_BIG_METADATA | \
208 BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO | \
209 BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD | \
210 BTRFS_FEATURE_INCOMPAT_RAID56 | \
211 BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF | \
212 BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA | \
213 BTRFS_FEATURE_INCOMPAT_NO_HOLES | \
214 BTRFS_FEATURE_INCOMPAT_METADATA_UUID | \
215 BTRFS_FEATURE_INCOMPAT_RAID1C34 | \
216 BTRFS_FEATURE_INCOMPAT_ZONED | \
217 BTRFS_FEATURE_INCOMPAT_EXTENT_TREE_V2)
218 #else
219 #define BTRFS_FEATURE_INCOMPAT_SUPP \
220 (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \
221 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL | \
222 BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS | \
223 BTRFS_FEATURE_INCOMPAT_BIG_METADATA | \
224 BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO | \
225 BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD | \
226 BTRFS_FEATURE_INCOMPAT_RAID56 | \
227 BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF | \
228 BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA | \
229 BTRFS_FEATURE_INCOMPAT_NO_HOLES | \
230 BTRFS_FEATURE_INCOMPAT_METADATA_UUID | \
231 BTRFS_FEATURE_INCOMPAT_RAID1C34 | \
232 BTRFS_FEATURE_INCOMPAT_ZONED)
233 #endif
234
235 #define BTRFS_FEATURE_INCOMPAT_SAFE_SET \
236 (BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF)
237 #define BTRFS_FEATURE_INCOMPAT_SAFE_CLEAR 0ULL
238
239 #define BTRFS_DEFAULT_COMMIT_INTERVAL (30)
240 #define BTRFS_DEFAULT_MAX_INLINE (2048)
241
242 struct btrfs_dev_replace {
243 /* See #define above */
244 u64 replace_state;
245 /* Seconds since 1-Jan-1970 */
246 time64_t time_started;
247 /* Seconds since 1-Jan-1970 */
248 time64_t time_stopped;
249 atomic64_t num_write_errors;
250 atomic64_t num_uncorrectable_read_errors;
251
252 u64 cursor_left;
253 u64 committed_cursor_left;
254 u64 cursor_left_last_write_of_item;
255 u64 cursor_right;
256
257 /* See #define above */
258 u64 cont_reading_from_srcdev_mode;
259
260 int is_valid;
261 int item_needs_writeback;
262 struct btrfs_device *srcdev;
263 struct btrfs_device *tgtdev;
264
265 struct mutex lock_finishing_cancel_unmount;
266 struct rw_semaphore rwsem;
267
268 struct btrfs_scrub_progress scrub_progress;
269
270 struct percpu_counter bio_counter;
271 wait_queue_head_t replace_wait;
272 };
273
274 /*
275 * Free clusters are used to claim free space in relatively large chunks,
276 * allowing us to do less seeky writes. They are used for all metadata
277 * allocations. In ssd_spread mode they are also used for data allocations.
278 */
279 struct btrfs_free_cluster {
280 spinlock_t lock;
281 spinlock_t refill_lock;
282 struct rb_root root;
283
284 /* Largest extent in this cluster */
285 u64 max_size;
286
287 /* First extent starting offset */
288 u64 window_start;
289
290 /* We did a full search and couldn't create a cluster */
291 bool fragmented;
292
293 struct btrfs_block_group *block_group;
294 /*
295 * When a cluster is allocated from a block group, we put the cluster
296 * onto a list in the block group so that it can be freed before the
297 * block group is freed.
298 */
299 struct list_head block_group_list;
300 };
301
302 /* Discard control. */
303 /*
304 * Async discard uses multiple lists to differentiate the discard filter
305 * parameters. Index 0 is for completely free block groups where we need to
306 * ensure the entire block group is trimmed without being lossy. Indices
307 * afterwards represent monotonically decreasing discard filter sizes to
308 * prioritize what should be discarded next.
309 */
310 #define BTRFS_NR_DISCARD_LISTS 3
311 #define BTRFS_DISCARD_INDEX_UNUSED 0
312 #define BTRFS_DISCARD_INDEX_START 1
313
314 struct btrfs_discard_ctl {
315 struct workqueue_struct *discard_workers;
316 struct delayed_work work;
317 spinlock_t lock;
318 struct btrfs_block_group *block_group;
319 struct list_head discard_list[BTRFS_NR_DISCARD_LISTS];
320 u64 prev_discard;
321 u64 prev_discard_time;
322 atomic_t discardable_extents;
323 atomic64_t discardable_bytes;
324 u64 max_discard_size;
325 u64 delay_ms;
326 u32 iops_limit;
327 u32 kbps_limit;
328 u64 discard_extent_bytes;
329 u64 discard_bitmap_bytes;
330 atomic64_t discard_bytes_saved;
331 };
332
333 /*
334 * Exclusive operations (device replace, resize, device add/remove, balance)
335 */
336 enum btrfs_exclusive_operation {
337 BTRFS_EXCLOP_NONE,
338 BTRFS_EXCLOP_BALANCE_PAUSED,
339 BTRFS_EXCLOP_BALANCE,
340 BTRFS_EXCLOP_DEV_ADD,
341 BTRFS_EXCLOP_DEV_REMOVE,
342 BTRFS_EXCLOP_DEV_REPLACE,
343 BTRFS_EXCLOP_RESIZE,
344 BTRFS_EXCLOP_SWAP_ACTIVATE,
345 };
346
347 /* Store data about transaction commits, exported via sysfs. */
348 struct btrfs_commit_stats {
349 /* Total number of commits */
350 u64 commit_count;
351 /* The maximum commit duration so far in ns */
352 u64 max_commit_dur;
353 /* The last commit duration in ns */
354 u64 last_commit_dur;
355 /* The total commit duration in ns */
356 u64 total_commit_dur;
357 };
358
359 struct btrfs_fs_info {
360 u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
361 unsigned long flags;
362 struct btrfs_root *tree_root;
363 struct btrfs_root *chunk_root;
364 struct btrfs_root *dev_root;
365 struct btrfs_root *fs_root;
366 struct btrfs_root *quota_root;
367 struct btrfs_root *uuid_root;
368 struct btrfs_root *data_reloc_root;
369 struct btrfs_root *block_group_root;
370
371 /* The log root tree is a directory of all the other log roots */
372 struct btrfs_root *log_root_tree;
373
374 /* The tree that holds the global roots (csum, extent, etc) */
375 rwlock_t global_root_lock;
376 struct rb_root global_root_tree;
377
378 spinlock_t fs_roots_radix_lock;
379 struct radix_tree_root fs_roots_radix;
380
381 /* Block group cache stuff */
382 rwlock_t block_group_cache_lock;
383 struct rb_root_cached block_group_cache_tree;
384
385 /* Keep track of unallocated space */
386 atomic64_t free_chunk_space;
387
388 /* Track ranges which are used by log trees blocks/logged data extents */
389 struct extent_io_tree excluded_extents;
390
391 /* logical->physical extent mapping */
392 struct extent_map_tree mapping_tree;
393
394 /*
395 * Block reservation for extent, checksum, root tree and delayed dir
396 * index item.
397 */
398 struct btrfs_block_rsv global_block_rsv;
399 /* Block reservation for metadata operations */
400 struct btrfs_block_rsv trans_block_rsv;
401 /* Block reservation for chunk tree */
402 struct btrfs_block_rsv chunk_block_rsv;
403 /* Block reservation for delayed operations */
404 struct btrfs_block_rsv delayed_block_rsv;
405 /* Block reservation for delayed refs */
406 struct btrfs_block_rsv delayed_refs_rsv;
407
408 struct btrfs_block_rsv empty_block_rsv;
409
410 u64 generation;
411 u64 last_trans_committed;
412 /*
413 * Generation of the last transaction used for block group relocation
414 * since the filesystem was last mounted (or 0 if none happened yet).
415 * Must be written and read while holding btrfs_fs_info::commit_root_sem.
416 */
417 u64 last_reloc_trans;
418 u64 avg_delayed_ref_runtime;
419
420 /*
421 * This is updated to the current trans every time a full commit is
422 * required instead of the faster short fsync log commits
423 */
424 u64 last_trans_log_full_commit;
425 unsigned long mount_opt;
426
427 unsigned long compress_type:4;
428 unsigned int compress_level;
429 u32 commit_interval;
430 /*
431 * It is a suggestive number, the read side is safe even it gets a
432 * wrong number because we will write out the data into a regular
433 * extent. The write side(mount/remount) is under ->s_umount lock,
434 * so it is also safe.
435 */
436 u64 max_inline;
437
438 struct btrfs_transaction *running_transaction;
439 wait_queue_head_t transaction_throttle;
440 wait_queue_head_t transaction_wait;
441 wait_queue_head_t transaction_blocked_wait;
442 wait_queue_head_t async_submit_wait;
443
444 /*
445 * Used to protect the incompat_flags, compat_flags, compat_ro_flags
446 * when they are updated.
447 *
448 * Because we do not clear the flags for ever, so we needn't use
449 * the lock on the read side.
450 *
451 * We also needn't use the lock when we mount the fs, because
452 * there is no other task which will update the flag.
453 */
454 spinlock_t super_lock;
455 struct btrfs_super_block *super_copy;
456 struct btrfs_super_block *super_for_commit;
457 struct super_block *sb;
458 struct inode *btree_inode;
459 struct mutex tree_log_mutex;
460 struct mutex transaction_kthread_mutex;
461 struct mutex cleaner_mutex;
462 struct mutex chunk_mutex;
463
464 /*
465 * This is taken to make sure we don't set block groups ro after the
466 * free space cache has been allocated on them.
467 */
468 struct mutex ro_block_group_mutex;
469
470 /*
471 * This is used during read/modify/write to make sure no two ios are
472 * trying to mod the same stripe at the same time.
473 */
474 struct btrfs_stripe_hash_table *stripe_hash_table;
475
476 /*
477 * This protects the ordered operations list only while we are
478 * processing all of the entries on it. This way we make sure the
479 * commit code doesn't find the list temporarily empty because another
480 * function happens to be doing non-waiting preflush before jumping
481 * into the main commit.
482 */
483 struct mutex ordered_operations_mutex;
484
485 struct rw_semaphore commit_root_sem;
486
487 struct rw_semaphore cleanup_work_sem;
488
489 struct rw_semaphore subvol_sem;
490
491 spinlock_t trans_lock;
492 /*
493 * The reloc mutex goes with the trans lock, it is taken during commit
494 * to protect us from the relocation code.
495 */
496 struct mutex reloc_mutex;
497
498 struct list_head trans_list;
499 struct list_head dead_roots;
500 struct list_head caching_block_groups;
501
502 spinlock_t delayed_iput_lock;
503 struct list_head delayed_iputs;
504 atomic_t nr_delayed_iputs;
505 wait_queue_head_t delayed_iputs_wait;
506
507 atomic64_t tree_mod_seq;
508
509 /* This protects tree_mod_log and tree_mod_seq_list */
510 rwlock_t tree_mod_log_lock;
511 struct rb_root tree_mod_log;
512 struct list_head tree_mod_seq_list;
513
514 atomic_t async_delalloc_pages;
515
516 /* This is used to protect the following list -- ordered_roots. */
517 spinlock_t ordered_root_lock;
518
519 /*
520 * All fs/file tree roots in which there are data=ordered extents
521 * pending writeback are added into this list.
522 *
523 * These can span multiple transactions and basically include every
524 * dirty data page that isn't from nodatacow.
525 */
526 struct list_head ordered_roots;
527
528 struct mutex delalloc_root_mutex;
529 spinlock_t delalloc_root_lock;
530 /* All fs/file tree roots that have delalloc inodes. */
531 struct list_head delalloc_roots;
532
533 /*
534 * There is a pool of worker threads for checksumming during writes and
535 * a pool for checksumming after reads. This is because readers can
536 * run with FS locks held, and the writers may be waiting for those
537 * locks. We don't want ordering in the pending list to cause
538 * deadlocks, and so the two are serviced separately.
539 *
540 * A third pool does submit_bio to avoid deadlocking with the other two.
541 */
542 struct btrfs_workqueue *workers;
543 struct btrfs_workqueue *hipri_workers;
544 struct btrfs_workqueue *delalloc_workers;
545 struct btrfs_workqueue *flush_workers;
546 struct workqueue_struct *endio_workers;
547 struct workqueue_struct *endio_meta_workers;
548 struct workqueue_struct *rmw_workers;
549 struct workqueue_struct *compressed_write_workers;
550 struct btrfs_workqueue *endio_write_workers;
551 struct btrfs_workqueue *endio_freespace_worker;
552 struct btrfs_workqueue *caching_workers;
553
554 /*
555 * Fixup workers take dirty pages that didn't properly go through the
556 * cow mechanism and make them safe to write. It happens for the
557 * sys_munmap function call path.
558 */
559 struct btrfs_workqueue *fixup_workers;
560 struct btrfs_workqueue *delayed_workers;
561
562 struct task_struct *transaction_kthread;
563 struct task_struct *cleaner_kthread;
564 u32 thread_pool_size;
565
566 struct kobject *space_info_kobj;
567 struct kobject *qgroups_kobj;
568 struct kobject *discard_kobj;
569
570 /* Used to keep from writing metadata until there is a nice batch */
571 struct percpu_counter dirty_metadata_bytes;
572 struct percpu_counter delalloc_bytes;
573 struct percpu_counter ordered_bytes;
574 s32 dirty_metadata_batch;
575 s32 delalloc_batch;
576
577 struct list_head dirty_cowonly_roots;
578
579 struct btrfs_fs_devices *fs_devices;
580
581 /*
582 * The space_info list is effectively read only after initial setup.
583 * It is populated at mount time and cleaned up after all block groups
584 * are removed. RCU is used to protect it.
585 */
586 struct list_head space_info;
587
588 struct btrfs_space_info *data_sinfo;
589
590 struct reloc_control *reloc_ctl;
591
592 /* data_alloc_cluster is only used in ssd_spread mode */
593 struct btrfs_free_cluster data_alloc_cluster;
594
595 /* All metadata allocations go through this cluster. */
596 struct btrfs_free_cluster meta_alloc_cluster;
597
598 /* Auto defrag inodes go here. */
599 spinlock_t defrag_inodes_lock;
600 struct rb_root defrag_inodes;
601 atomic_t defrag_running;
602
603 /* Used to protect avail_{data, metadata, system}_alloc_bits */
604 seqlock_t profiles_lock;
605 /*
606 * These three are in extended format (availability of single chunks is
607 * denoted by BTRFS_AVAIL_ALLOC_BIT_SINGLE bit, other types are denoted
608 * by corresponding BTRFS_BLOCK_GROUP_* bits)
609 */
610 u64 avail_data_alloc_bits;
611 u64 avail_metadata_alloc_bits;
612 u64 avail_system_alloc_bits;
613
614 /* Balance state */
615 spinlock_t balance_lock;
616 struct mutex balance_mutex;
617 atomic_t balance_pause_req;
618 atomic_t balance_cancel_req;
619 struct btrfs_balance_control *balance_ctl;
620 wait_queue_head_t balance_wait_q;
621
622 /* Cancellation requests for chunk relocation */
623 atomic_t reloc_cancel_req;
624
625 u32 data_chunk_allocations;
626 u32 metadata_ratio;
627
628 void *bdev_holder;
629
630 /* Private scrub information */
631 struct mutex scrub_lock;
632 atomic_t scrubs_running;
633 atomic_t scrub_pause_req;
634 atomic_t scrubs_paused;
635 atomic_t scrub_cancel_req;
636 wait_queue_head_t scrub_pause_wait;
637 /*
638 * The worker pointers are NULL iff the refcount is 0, ie. scrub is not
639 * running.
640 */
641 refcount_t scrub_workers_refcnt;
642 struct workqueue_struct *scrub_workers;
643 struct workqueue_struct *scrub_wr_completion_workers;
644 struct workqueue_struct *scrub_parity_workers;
645 struct btrfs_subpage_info *subpage_info;
646
647 struct btrfs_discard_ctl discard_ctl;
648
649 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
650 u32 check_integrity_print_mask;
651 #endif
652 /* Is qgroup tracking in a consistent state? */
653 u64 qgroup_flags;
654
655 /* Holds configuration and tracking. Protected by qgroup_lock. */
656 struct rb_root qgroup_tree;
657 spinlock_t qgroup_lock;
658
659 /*
660 * Used to avoid frequently calling ulist_alloc()/ulist_free()
661 * when doing qgroup accounting, it must be protected by qgroup_lock.
662 */
663 struct ulist *qgroup_ulist;
664
665 /*
666 * Protect user change for quota operations. If a transaction is needed,
667 * it must be started before locking this lock.
668 */
669 struct mutex qgroup_ioctl_lock;
670
671 /* List of dirty qgroups to be written at next commit. */
672 struct list_head dirty_qgroups;
673
674 /* Used by qgroup for an efficient tree traversal. */
675 u64 qgroup_seq;
676
677 /* Qgroup rescan items. */
678 /* Protects the progress item */
679 struct mutex qgroup_rescan_lock;
680 struct btrfs_key qgroup_rescan_progress;
681 struct btrfs_workqueue *qgroup_rescan_workers;
682 struct completion qgroup_rescan_completion;
683 struct btrfs_work qgroup_rescan_work;
684 /* Protected by qgroup_rescan_lock */
685 bool qgroup_rescan_running;
686 u8 qgroup_drop_subtree_thres;
687
688 /* Filesystem state */
689 unsigned long fs_state;
690
691 struct btrfs_delayed_root *delayed_root;
692
693 /* Extent buffer radix tree */
694 spinlock_t buffer_lock;
695 /* Entries are eb->start / sectorsize */
696 struct radix_tree_root buffer_radix;
697
698 /* Next backup root to be overwritten */
699 int backup_root_index;
700
701 /* Device replace state */
702 struct btrfs_dev_replace dev_replace;
703
704 struct semaphore uuid_tree_rescan_sem;
705
706 /* Used to reclaim the metadata space in the background. */
707 struct work_struct async_reclaim_work;
708 struct work_struct async_data_reclaim_work;
709 struct work_struct preempt_reclaim_work;
710
711 /* Reclaim partially filled block groups in the background */
712 struct work_struct reclaim_bgs_work;
713 struct list_head reclaim_bgs;
714 int bg_reclaim_threshold;
715
716 spinlock_t unused_bgs_lock;
717 struct list_head unused_bgs;
718 struct mutex unused_bg_unpin_mutex;
719 /* Protect block groups that are going to be deleted */
720 struct mutex reclaim_bgs_lock;
721
722 /* Cached block sizes */
723 u32 nodesize;
724 u32 sectorsize;
725 /* ilog2 of sectorsize, use to avoid 64bit division */
726 u32 sectorsize_bits;
727 u32 csum_size;
728 u32 csums_per_leaf;
729 u32 stripesize;
730
731 /*
732 * Maximum size of an extent. BTRFS_MAX_EXTENT_SIZE on regular
733 * filesystem, on zoned it depends on the device constraints.
734 */
735 u64 max_extent_size;
736
737 /* Block groups and devices containing active swapfiles. */
738 spinlock_t swapfile_pins_lock;
739 struct rb_root swapfile_pins;
740
741 struct crypto_shash *csum_shash;
742
743 /* Type of exclusive operation running, protected by super_lock */
744 enum btrfs_exclusive_operation exclusive_operation;
745
746 /*
747 * Zone size > 0 when in ZONED mode, otherwise it's used for a check
748 * if the mode is enabled
749 */
750 u64 zone_size;
751
752 /* Constraints for ZONE_APPEND commands: */
753 struct queue_limits limits;
754 u64 max_zone_append_size;
755
756 struct mutex zoned_meta_io_lock;
757 spinlock_t treelog_bg_lock;
758 u64 treelog_bg;
759
760 /*
761 * Start of the dedicated data relocation block group, protected by
762 * relocation_bg_lock.
763 */
764 spinlock_t relocation_bg_lock;
765 u64 data_reloc_bg;
766 struct mutex zoned_data_reloc_io_lock;
767
768 u64 nr_global_roots;
769
770 spinlock_t zone_active_bgs_lock;
771 struct list_head zone_active_bgs;
772
773 /* Updates are not protected by any lock */
774 struct btrfs_commit_stats commit_stats;
775
776 /*
777 * Last generation where we dropped a non-relocation root.
778 * Use btrfs_set_last_root_drop_gen() and btrfs_get_last_root_drop_gen()
779 * to change it and to read it, respectively.
780 */
781 u64 last_root_drop_gen;
782
783 /*
784 * Annotations for transaction events (structures are empty when
785 * compiled without lockdep).
786 */
787 struct lockdep_map btrfs_trans_num_writers_map;
788 struct lockdep_map btrfs_trans_num_extwriters_map;
789 struct lockdep_map btrfs_state_change_map[4];
790 struct lockdep_map btrfs_trans_pending_ordered_map;
791 struct lockdep_map btrfs_ordered_extent_map;
792
793 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
794 spinlock_t ref_verify_lock;
795 struct rb_root block_tree;
796 #endif
797
798 #ifdef CONFIG_BTRFS_DEBUG
799 struct kobject *debug_kobj;
800 struct list_head allocated_roots;
801
802 spinlock_t eb_leak_lock;
803 struct list_head allocated_ebs;
804 #endif
805 };
806
btrfs_set_last_root_drop_gen(struct btrfs_fs_info * fs_info,u64 gen)807 static inline void btrfs_set_last_root_drop_gen(struct btrfs_fs_info *fs_info,
808 u64 gen)
809 {
810 WRITE_ONCE(fs_info->last_root_drop_gen, gen);
811 }
812
btrfs_get_last_root_drop_gen(const struct btrfs_fs_info * fs_info)813 static inline u64 btrfs_get_last_root_drop_gen(const struct btrfs_fs_info *fs_info)
814 {
815 return READ_ONCE(fs_info->last_root_drop_gen);
816 }
817
818 /*
819 * Take the number of bytes to be checksummed and figure out how many leaves
820 * it would require to store the csums for that many bytes.
821 */
btrfs_csum_bytes_to_leaves(const struct btrfs_fs_info * fs_info,u64 csum_bytes)822 static inline u64 btrfs_csum_bytes_to_leaves(
823 const struct btrfs_fs_info *fs_info, u64 csum_bytes)
824 {
825 const u64 num_csums = csum_bytes >> fs_info->sectorsize_bits;
826
827 return DIV_ROUND_UP_ULL(num_csums, fs_info->csums_per_leaf);
828 }
829
830 /*
831 * Use this if we would be adding new items, as we could split nodes as we cow
832 * down the tree.
833 */
btrfs_calc_insert_metadata_size(struct btrfs_fs_info * fs_info,unsigned num_items)834 static inline u64 btrfs_calc_insert_metadata_size(struct btrfs_fs_info *fs_info,
835 unsigned num_items)
836 {
837 return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * 2 * num_items;
838 }
839
840 /*
841 * Doing a truncate or a modification won't result in new nodes or leaves, just
842 * what we need for COW.
843 */
btrfs_calc_metadata_size(struct btrfs_fs_info * fs_info,unsigned num_items)844 static inline u64 btrfs_calc_metadata_size(struct btrfs_fs_info *fs_info,
845 unsigned num_items)
846 {
847 return (u64)fs_info->nodesize * BTRFS_MAX_LEVEL * num_items;
848 }
849
850 #define BTRFS_MAX_EXTENT_ITEM_SIZE(r) ((BTRFS_LEAF_DATA_SIZE(r->fs_info) >> 4) - \
851 sizeof(struct btrfs_item))
852
btrfs_is_zoned(const struct btrfs_fs_info * fs_info)853 static inline bool btrfs_is_zoned(const struct btrfs_fs_info *fs_info)
854 {
855 return fs_info->zone_size > 0;
856 }
857
858 /*
859 * Count how many fs_info->max_extent_size cover the @size
860 */
count_max_extents(struct btrfs_fs_info * fs_info,u64 size)861 static inline u32 count_max_extents(struct btrfs_fs_info *fs_info, u64 size)
862 {
863 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
864 if (!fs_info)
865 return div_u64(size + BTRFS_MAX_EXTENT_SIZE - 1, BTRFS_MAX_EXTENT_SIZE);
866 #endif
867
868 return div_u64(size + fs_info->max_extent_size - 1, fs_info->max_extent_size);
869 }
870
871 bool btrfs_exclop_start(struct btrfs_fs_info *fs_info,
872 enum btrfs_exclusive_operation type);
873 bool btrfs_exclop_start_try_lock(struct btrfs_fs_info *fs_info,
874 enum btrfs_exclusive_operation type);
875 void btrfs_exclop_start_unlock(struct btrfs_fs_info *fs_info);
876 void btrfs_exclop_finish(struct btrfs_fs_info *fs_info);
877 void btrfs_exclop_balance(struct btrfs_fs_info *fs_info,
878 enum btrfs_exclusive_operation op);
879
880 /* Compatibility and incompatibility defines */
881 void __btrfs_set_fs_incompat(struct btrfs_fs_info *fs_info, u64 flag,
882 const char *name);
883 void __btrfs_clear_fs_incompat(struct btrfs_fs_info *fs_info, u64 flag,
884 const char *name);
885 void __btrfs_set_fs_compat_ro(struct btrfs_fs_info *fs_info, u64 flag,
886 const char *name);
887 void __btrfs_clear_fs_compat_ro(struct btrfs_fs_info *fs_info, u64 flag,
888 const char *name);
889
890 #define __btrfs_fs_incompat(fs_info, flags) \
891 (!!(btrfs_super_incompat_flags((fs_info)->super_copy) & (flags)))
892
893 #define __btrfs_fs_compat_ro(fs_info, flags) \
894 (!!(btrfs_super_compat_ro_flags((fs_info)->super_copy) & (flags)))
895
896 #define btrfs_set_fs_incompat(__fs_info, opt) \
897 __btrfs_set_fs_incompat((__fs_info), BTRFS_FEATURE_INCOMPAT_##opt, #opt)
898
899 #define btrfs_clear_fs_incompat(__fs_info, opt) \
900 __btrfs_clear_fs_incompat((__fs_info), BTRFS_FEATURE_INCOMPAT_##opt, #opt)
901
902 #define btrfs_fs_incompat(fs_info, opt) \
903 __btrfs_fs_incompat((fs_info), BTRFS_FEATURE_INCOMPAT_##opt)
904
905 #define btrfs_set_fs_compat_ro(__fs_info, opt) \
906 __btrfs_set_fs_compat_ro((__fs_info), BTRFS_FEATURE_COMPAT_RO_##opt, #opt)
907
908 #define btrfs_clear_fs_compat_ro(__fs_info, opt) \
909 __btrfs_clear_fs_compat_ro((__fs_info), BTRFS_FEATURE_COMPAT_RO_##opt, #opt)
910
911 #define btrfs_fs_compat_ro(fs_info, opt) \
912 __btrfs_fs_compat_ro((fs_info), BTRFS_FEATURE_COMPAT_RO_##opt)
913
914 #define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt)
915 #define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt)
916 #define btrfs_raw_test_opt(o, opt) ((o) & BTRFS_MOUNT_##opt)
917 #define btrfs_test_opt(fs_info, opt) ((fs_info)->mount_opt & \
918 BTRFS_MOUNT_##opt)
919
920 #define btrfs_set_and_info(fs_info, opt, fmt, args...) \
921 do { \
922 if (!btrfs_test_opt(fs_info, opt)) \
923 btrfs_info(fs_info, fmt, ##args); \
924 btrfs_set_opt(fs_info->mount_opt, opt); \
925 } while (0)
926
927 #define btrfs_clear_and_info(fs_info, opt, fmt, args...) \
928 do { \
929 if (btrfs_test_opt(fs_info, opt)) \
930 btrfs_info(fs_info, fmt, ##args); \
931 btrfs_clear_opt(fs_info->mount_opt, opt); \
932 } while (0)
933
btrfs_fs_closing(struct btrfs_fs_info * fs_info)934 static inline int btrfs_fs_closing(struct btrfs_fs_info *fs_info)
935 {
936 /* Do it this way so we only ever do one test_bit in the normal case. */
937 if (test_bit(BTRFS_FS_CLOSING_START, &fs_info->flags)) {
938 if (test_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags))
939 return 2;
940 return 1;
941 }
942 return 0;
943 }
944
945 /*
946 * If we remount the fs to be R/O or umount the fs, the cleaner needn't do
947 * anything except sleeping. This function is used to check the status of
948 * the fs.
949 * We check for BTRFS_FS_STATE_RO to avoid races with a concurrent remount,
950 * since setting and checking for SB_RDONLY in the superblock's flags is not
951 * atomic.
952 */
btrfs_need_cleaner_sleep(struct btrfs_fs_info * fs_info)953 static inline int btrfs_need_cleaner_sleep(struct btrfs_fs_info *fs_info)
954 {
955 return test_bit(BTRFS_FS_STATE_RO, &fs_info->fs_state) ||
956 btrfs_fs_closing(fs_info);
957 }
958
btrfs_wake_unfinished_drop(struct btrfs_fs_info * fs_info)959 static inline void btrfs_wake_unfinished_drop(struct btrfs_fs_info *fs_info)
960 {
961 clear_and_wake_up_bit(BTRFS_FS_UNFINISHED_DROPS, &fs_info->flags);
962 }
963
964 #define BTRFS_FS_ERROR(fs_info) (unlikely(test_bit(BTRFS_FS_STATE_ERROR, \
965 &(fs_info)->fs_state)))
966 #define BTRFS_FS_LOG_CLEANUP_ERROR(fs_info) \
967 (unlikely(test_bit(BTRFS_FS_STATE_LOG_CLEANUP_ERROR, \
968 &(fs_info)->fs_state)))
969
970 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
971
972 #define EXPORT_FOR_TESTS
973
btrfs_is_testing(struct btrfs_fs_info * fs_info)974 static inline int btrfs_is_testing(struct btrfs_fs_info *fs_info)
975 {
976 return test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
977 }
978
979 void btrfs_test_destroy_inode(struct inode *inode);
980
981 #else
982
983 #define EXPORT_FOR_TESTS static
984
btrfs_is_testing(struct btrfs_fs_info * fs_info)985 static inline int btrfs_is_testing(struct btrfs_fs_info *fs_info)
986 {
987 return 0;
988 }
989 #endif
990
991 #endif
992