1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/rbtree.h>
6 #include <linux/refcount.h>
9 /* bits for the extent state */
10 #define EXTENT_DIRTY (1U << 0)
11 #define EXTENT_WRITEBACK (1U << 1)
12 #define EXTENT_UPTODATE (1U << 2)
13 #define EXTENT_LOCKED (1U << 3)
14 #define EXTENT_NEW (1U << 4)
15 #define EXTENT_DELALLOC (1U << 5)
16 #define EXTENT_DEFRAG (1U << 6)
17 #define EXTENT_BOUNDARY (1U << 9)
18 #define EXTENT_NODATASUM (1U << 10)
19 #define EXTENT_CLEAR_META_RESV (1U << 11)
20 #define EXTENT_FIRST_DELALLOC (1U << 12)
21 #define EXTENT_NEED_WAIT (1U << 13)
22 #define EXTENT_DAMAGED (1U << 14)
23 #define EXTENT_NORESERVE (1U << 15)
24 #define EXTENT_QGROUP_RESERVED (1U << 16)
25 #define EXTENT_CLEAR_DATA_RESV (1U << 17)
26 #define EXTENT_DELALLOC_NEW (1U << 18)
27 #define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK)
28 #define EXTENT_DO_ACCOUNTING (EXTENT_CLEAR_META_RESV | \
29 EXTENT_CLEAR_DATA_RESV)
30 #define EXTENT_CTLBITS (EXTENT_DO_ACCOUNTING | EXTENT_FIRST_DELALLOC)
33 * flags for bio submission. The high bits indicate the compression
36 #define EXTENT_BIO_COMPRESSED 1
37 #define EXTENT_BIO_FLAG_SHIFT 16
39 /* these are bit numbers for test/set bit */
40 #define EXTENT_BUFFER_UPTODATE 0
41 #define EXTENT_BUFFER_DIRTY 2
42 #define EXTENT_BUFFER_CORRUPT 3
43 #define EXTENT_BUFFER_READAHEAD 4 /* this got triggered by readahead */
44 #define EXTENT_BUFFER_TREE_REF 5
45 #define EXTENT_BUFFER_STALE 6
46 #define EXTENT_BUFFER_WRITEBACK 7
47 #define EXTENT_BUFFER_READ_ERR 8 /* read IO error */
48 #define EXTENT_BUFFER_DUMMY 9
49 #define EXTENT_BUFFER_IN_TREE 10
50 #define EXTENT_BUFFER_WRITE_ERR 11 /* write IO error */
52 /* these are flags for __process_pages_contig */
53 #define PAGE_UNLOCK (1 << 0)
54 #define PAGE_CLEAR_DIRTY (1 << 1)
55 #define PAGE_SET_WRITEBACK (1 << 2)
56 #define PAGE_END_WRITEBACK (1 << 3)
57 #define PAGE_SET_PRIVATE2 (1 << 4)
58 #define PAGE_SET_ERROR (1 << 5)
59 #define PAGE_LOCK (1 << 6)
62 * page->private values. Every page that is controlled by the extent
63 * map has page->private set to one.
65 #define EXTENT_PAGE_PRIVATE 1
68 * The extent buffer bitmap operations are done with byte granularity instead of
69 * word granularity for two reasons:
70 * 1. The bitmaps must be little-endian on disk.
71 * 2. Bitmap items are not guaranteed to be aligned to a word and therefore a
72 * single word in a bitmap may straddle two pages in the extent buffer.
74 #define BIT_BYTE(nr) ((nr) / BITS_PER_BYTE)
75 #define BYTE_MASK ((1 << BITS_PER_BYTE) - 1)
76 #define BITMAP_FIRST_BYTE_MASK(start) \
77 ((BYTE_MASK << ((start) & (BITS_PER_BYTE - 1))) & BYTE_MASK)
78 #define BITMAP_LAST_BYTE_MASK(nbits) \
79 (BYTE_MASK >> (-(nbits) & (BITS_PER_BYTE - 1)))
81 static inline int le_test_bit(int nr, const u8 *addr)
83 return 1U & (addr[BIT_BYTE(nr)] >> (nr & (BITS_PER_BYTE-1)));
86 extern void le_bitmap_set(u8 *map, unsigned int start, int len);
87 extern void le_bitmap_clear(u8 *map, unsigned int start, int len);
93 struct io_failure_record;
95 typedef blk_status_t (extent_submit_bio_hook_t)(void *private_data, struct bio *bio,
96 int mirror_num, unsigned long bio_flags,
98 struct extent_io_ops {
100 * The following callbacks must be allways defined, the function
101 * pointer will be called unconditionally.
103 extent_submit_bio_hook_t *submit_bio_hook;
104 int (*readpage_end_io_hook)(struct btrfs_io_bio *io_bio, u64 phy_offset,
105 struct page *page, u64 start, u64 end,
107 int (*merge_bio_hook)(struct page *page, unsigned long offset,
108 size_t size, struct bio *bio,
109 unsigned long bio_flags);
110 int (*readpage_io_failed_hook)(struct page *page, int failed_mirror);
111 struct btrfs_fs_info *(*tree_fs_info)(void *private_data);
112 void (*set_range_writeback)(void *private_data, u64 start, u64 end);
115 * Optional hooks, called if the pointer is not NULL
117 int (*fill_delalloc)(void *private_data, struct page *locked_page,
118 u64 start, u64 end, int *page_started,
119 unsigned long *nr_written,
120 struct writeback_control *wbc);
122 int (*writepage_start_hook)(struct page *page, u64 start, u64 end);
123 void (*writepage_end_io_hook)(struct page *page, u64 start, u64 end,
124 struct extent_state *state, int uptodate);
125 void (*set_bit_hook)(void *private_data, struct extent_state *state,
127 void (*clear_bit_hook)(void *private_data,
128 struct extent_state *state,
130 void (*merge_extent_hook)(void *private_data,
131 struct extent_state *new,
132 struct extent_state *other);
133 void (*split_extent_hook)(void *private_data,
134 struct extent_state *orig, u64 split);
135 void (*check_extent_io_range)(void *private_data, const char *caller,
139 struct extent_io_tree {
140 struct rb_root state;
145 const struct extent_io_ops *ops;
148 struct extent_state {
150 u64 end; /* inclusive */
151 struct rb_node rb_node;
153 /* ADD NEW ELEMENTS AFTER THIS */
154 wait_queue_head_t wq;
158 struct io_failure_record *failrec;
160 #ifdef CONFIG_BTRFS_DEBUG
161 struct list_head leak_list;
165 #define INLINE_EXTENT_BUFFER_PAGES 16
166 #define MAX_INLINE_EXTENT_BUFFER_SIZE (INLINE_EXTENT_BUFFER_PAGES * PAGE_SIZE)
167 struct extent_buffer {
170 unsigned long bflags;
171 struct btrfs_fs_info *fs_info;
172 spinlock_t refs_lock;
176 struct rcu_head rcu_head;
179 /* count of read lock holders on the extent buffer */
180 atomic_t write_locks;
182 atomic_t blocking_writers;
183 atomic_t blocking_readers;
184 atomic_t spinning_readers;
185 atomic_t spinning_writers;
187 /* >= 0 if eb belongs to a log tree, -1 otherwise */
190 /* protects write locks */
193 /* readers use lock_wq while they wait for the write
194 * lock holders to unlock
196 wait_queue_head_t write_lock_wq;
198 /* writers use read_lock_wq while they wait for readers
201 wait_queue_head_t read_lock_wq;
202 struct page *pages[INLINE_EXTENT_BUFFER_PAGES];
203 #ifdef CONFIG_BTRFS_DEBUG
204 struct list_head leak_list;
209 * Structure to record how many bytes and which ranges are set/cleared
211 struct extent_changeset {
212 /* How many bytes are set/cleared in this operation */
213 unsigned int bytes_changed;
216 struct ulist range_changed;
219 static inline void extent_changeset_init(struct extent_changeset *changeset)
221 changeset->bytes_changed = 0;
222 ulist_init(&changeset->range_changed);
225 static inline struct extent_changeset *extent_changeset_alloc(void)
227 struct extent_changeset *ret;
229 ret = kmalloc(sizeof(*ret), GFP_KERNEL);
233 extent_changeset_init(ret);
237 static inline void extent_changeset_release(struct extent_changeset *changeset)
241 changeset->bytes_changed = 0;
242 ulist_release(&changeset->range_changed);
245 static inline void extent_changeset_free(struct extent_changeset *changeset)
249 extent_changeset_release(changeset);
253 static inline void extent_set_compress_type(unsigned long *bio_flags,
256 *bio_flags |= compress_type << EXTENT_BIO_FLAG_SHIFT;
259 static inline int extent_compress_type(unsigned long bio_flags)
261 return bio_flags >> EXTENT_BIO_FLAG_SHIFT;
264 struct extent_map_tree;
266 typedef struct extent_map *(get_extent_t)(struct btrfs_inode *inode,
272 void extent_io_tree_init(struct extent_io_tree *tree, void *private_data);
273 int try_release_extent_mapping(struct extent_map_tree *map,
274 struct extent_io_tree *tree, struct page *page,
276 int try_release_extent_buffer(struct page *page);
277 int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
278 struct extent_state **cached);
280 static inline int lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
282 return lock_extent_bits(tree, start, end, NULL);
285 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end);
286 int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
287 get_extent_t *get_extent, int mirror_num);
288 int __init extent_io_init(void);
289 void extent_io_exit(void);
291 u64 count_range_bits(struct extent_io_tree *tree,
292 u64 *start, u64 search_end,
293 u64 max_bytes, unsigned bits, int contig);
295 void free_extent_state(struct extent_state *state);
296 int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
297 unsigned bits, int filled,
298 struct extent_state *cached_state);
299 int clear_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
300 unsigned bits, struct extent_changeset *changeset);
301 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
302 unsigned bits, int wake, int delete,
303 struct extent_state **cached);
304 int __clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
305 unsigned bits, int wake, int delete,
306 struct extent_state **cached, gfp_t mask,
307 struct extent_changeset *changeset);
309 static inline int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end)
311 return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL);
314 static inline int unlock_extent_cached(struct extent_io_tree *tree, u64 start,
315 u64 end, struct extent_state **cached, gfp_t mask)
317 return __clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, cached,
321 static inline int clear_extent_bits(struct extent_io_tree *tree, u64 start,
322 u64 end, unsigned bits)
326 if (bits & EXTENT_LOCKED)
329 return clear_extent_bit(tree, start, end, bits, wake, 0, NULL);
332 int set_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
333 unsigned bits, struct extent_changeset *changeset);
334 int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
335 unsigned bits, u64 *failed_start,
336 struct extent_state **cached_state, gfp_t mask);
338 static inline int set_extent_bits(struct extent_io_tree *tree, u64 start,
339 u64 end, unsigned bits)
341 return set_extent_bit(tree, start, end, bits, NULL, NULL, GFP_NOFS);
344 static inline int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
345 u64 end, struct extent_state **cached_state)
347 return __clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0,
348 cached_state, GFP_NOFS, NULL);
351 static inline int set_extent_dirty(struct extent_io_tree *tree, u64 start,
354 return set_extent_bit(tree, start, end, EXTENT_DIRTY, NULL,
358 static inline int clear_extent_dirty(struct extent_io_tree *tree, u64 start,
361 return clear_extent_bit(tree, start, end,
362 EXTENT_DIRTY | EXTENT_DELALLOC |
363 EXTENT_DO_ACCOUNTING, 0, 0, NULL);
366 int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
367 unsigned bits, unsigned clear_bits,
368 struct extent_state **cached_state);
370 static inline int set_extent_delalloc(struct extent_io_tree *tree, u64 start,
371 u64 end, unsigned int extra_bits,
372 struct extent_state **cached_state)
374 return set_extent_bit(tree, start, end,
375 EXTENT_DELALLOC | EXTENT_UPTODATE | extra_bits,
376 NULL, cached_state, GFP_NOFS);
379 static inline int set_extent_defrag(struct extent_io_tree *tree, u64 start,
380 u64 end, struct extent_state **cached_state)
382 return set_extent_bit(tree, start, end,
383 EXTENT_DELALLOC | EXTENT_UPTODATE | EXTENT_DEFRAG,
384 NULL, cached_state, GFP_NOFS);
387 static inline int set_extent_new(struct extent_io_tree *tree, u64 start,
390 return set_extent_bit(tree, start, end, EXTENT_NEW, NULL, NULL,
394 static inline int set_extent_uptodate(struct extent_io_tree *tree, u64 start,
395 u64 end, struct extent_state **cached_state, gfp_t mask)
397 return set_extent_bit(tree, start, end, EXTENT_UPTODATE, NULL,
401 int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
402 u64 *start_ret, u64 *end_ret, unsigned bits,
403 struct extent_state **cached_state);
404 int extent_invalidatepage(struct extent_io_tree *tree,
405 struct page *page, unsigned long offset);
406 int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
407 struct writeback_control *wbc);
408 int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
409 u64 start, u64 end, int mode);
410 int extent_writepages(struct extent_io_tree *tree,
411 struct address_space *mapping,
412 struct writeback_control *wbc);
413 int btree_write_cache_pages(struct address_space *mapping,
414 struct writeback_control *wbc);
415 int extent_readpages(struct extent_io_tree *tree,
416 struct address_space *mapping,
417 struct list_head *pages, unsigned nr_pages,
418 get_extent_t get_extent);
419 int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
420 __u64 start, __u64 len);
421 void set_page_extent_mapped(struct page *page);
423 struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
425 struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
426 u64 start, unsigned long len);
427 struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
429 struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src);
430 struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info,
432 void free_extent_buffer(struct extent_buffer *eb);
433 void free_extent_buffer_stale(struct extent_buffer *eb);
435 #define WAIT_COMPLETE 1
436 #define WAIT_PAGE_LOCK 2
437 int read_extent_buffer_pages(struct extent_io_tree *tree,
438 struct extent_buffer *eb, int wait,
439 get_extent_t *get_extent, int mirror_num);
440 void wait_on_extent_buffer_writeback(struct extent_buffer *eb);
442 static inline unsigned long num_extent_pages(u64 start, u64 len)
444 return ((start + len + PAGE_SIZE - 1) >> PAGE_SHIFT) -
445 (start >> PAGE_SHIFT);
448 static inline void extent_buffer_get(struct extent_buffer *eb)
450 atomic_inc(&eb->refs);
453 int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv,
454 unsigned long start, unsigned long len);
455 void read_extent_buffer(const struct extent_buffer *eb, void *dst,
458 int read_extent_buffer_to_user(const struct extent_buffer *eb,
459 void __user *dst, unsigned long start,
461 void write_extent_buffer_fsid(struct extent_buffer *eb, const void *src);
462 void write_extent_buffer_chunk_tree_uuid(struct extent_buffer *eb,
464 void write_extent_buffer(struct extent_buffer *eb, const void *src,
465 unsigned long start, unsigned long len);
466 void copy_extent_buffer_full(struct extent_buffer *dst,
467 struct extent_buffer *src);
468 void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
469 unsigned long dst_offset, unsigned long src_offset,
471 void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
472 unsigned long src_offset, unsigned long len);
473 void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
474 unsigned long src_offset, unsigned long len);
475 void memzero_extent_buffer(struct extent_buffer *eb, unsigned long start,
477 int extent_buffer_test_bit(struct extent_buffer *eb, unsigned long start,
479 void extent_buffer_bitmap_set(struct extent_buffer *eb, unsigned long start,
480 unsigned long pos, unsigned long len);
481 void extent_buffer_bitmap_clear(struct extent_buffer *eb, unsigned long start,
482 unsigned long pos, unsigned long len);
483 void clear_extent_buffer_dirty(struct extent_buffer *eb);
484 int set_extent_buffer_dirty(struct extent_buffer *eb);
485 void set_extent_buffer_uptodate(struct extent_buffer *eb);
486 void clear_extent_buffer_uptodate(struct extent_buffer *eb);
487 int extent_buffer_uptodate(struct extent_buffer *eb);
488 int extent_buffer_under_io(struct extent_buffer *eb);
489 int map_private_extent_buffer(const struct extent_buffer *eb,
490 unsigned long offset, unsigned long min_len,
491 char **map, unsigned long *map_start,
492 unsigned long *map_len);
493 void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end);
494 void extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end);
495 void extent_clear_unlock_delalloc(struct inode *inode, u64 start, u64 end,
496 u64 delalloc_end, struct page *locked_page,
497 unsigned bits_to_clear,
498 unsigned long page_ops);
499 struct bio *btrfs_bio_alloc(struct block_device *bdev, u64 first_byte);
500 struct bio *btrfs_io_bio_alloc(unsigned int nr_iovecs);
501 struct bio *btrfs_bio_clone(struct bio *bio);
502 struct bio *btrfs_bio_clone_partial(struct bio *orig, int offset, int size);
504 struct btrfs_fs_info;
507 int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start,
508 u64 length, u64 logical, struct page *page,
509 unsigned int pg_offset, int mirror_num);
510 int clean_io_failure(struct btrfs_fs_info *fs_info,
511 struct extent_io_tree *failure_tree,
512 struct extent_io_tree *io_tree, u64 start,
513 struct page *page, u64 ino, unsigned int pg_offset);
514 void end_extent_writepage(struct page *page, int err, u64 start, u64 end);
515 int repair_eb_io_failure(struct btrfs_fs_info *fs_info,
516 struct extent_buffer *eb, int mirror_num);
519 * When IO fails, either with EIO or csum verification fails, we
520 * try other mirrors that might have a good copy of the data. This
521 * io_failure_record is used to record state as we go through all the
522 * mirrors. If another mirror has good data, the page is set up to date
523 * and things continue. If a good mirror can't be found, the original
524 * bio end_io callback is called to indicate things have failed.
526 struct io_failure_record {
531 unsigned long bio_flags;
538 void btrfs_free_io_failure_record(struct btrfs_inode *inode, u64 start,
540 int btrfs_get_io_failure_record(struct inode *inode, u64 start, u64 end,
541 struct io_failure_record **failrec_ret);
542 bool btrfs_check_repairable(struct inode *inode, struct bio *failed_bio,
543 struct io_failure_record *failrec, int fail_mirror);
544 struct bio *btrfs_create_repair_bio(struct inode *inode, struct bio *failed_bio,
545 struct io_failure_record *failrec,
546 struct page *page, int pg_offset, int icsum,
547 bio_end_io_t *endio_func, void *data);
548 int free_io_failure(struct extent_io_tree *failure_tree,
549 struct extent_io_tree *io_tree,
550 struct io_failure_record *rec);
551 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
552 noinline u64 find_lock_delalloc_range(struct inode *inode,
553 struct extent_io_tree *tree,
554 struct page *locked_page, u64 *start,
555 u64 *end, u64 max_bytes);
557 struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info,