]>
Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
9888c340 DS |
2 | |
3 | #ifndef BTRFS_EXTENT_IO_H | |
4 | #define BTRFS_EXTENT_IO_H | |
d1310b2e CM |
5 | |
6 | #include <linux/rbtree.h> | |
b7ac31b7 | 7 | #include <linux/refcount.h> |
10c5db28 | 8 | #include <linux/fiemap.h> |
ac467772 | 9 | #include "ulist.h" |
d1310b2e | 10 | |
261507a0 LZ |
11 | /* |
12 | * flags for bio submission. The high bits indicate the compression | |
13 | * type for this bio | |
14 | */ | |
c8b97818 | 15 | #define EXTENT_BIO_COMPRESSED 1 |
261507a0 | 16 | #define EXTENT_BIO_FLAG_SHIFT 16 |
c8b97818 | 17 | |
80cb3836 DS |
18 | enum { |
19 | EXTENT_BUFFER_UPTODATE, | |
20 | EXTENT_BUFFER_DIRTY, | |
21 | EXTENT_BUFFER_CORRUPT, | |
22 | /* this got triggered by readahead */ | |
23 | EXTENT_BUFFER_READAHEAD, | |
24 | EXTENT_BUFFER_TREE_REF, | |
25 | EXTENT_BUFFER_STALE, | |
26 | EXTENT_BUFFER_WRITEBACK, | |
27 | /* read IO error */ | |
28 | EXTENT_BUFFER_READ_ERR, | |
29 | EXTENT_BUFFER_UNMAPPED, | |
30 | EXTENT_BUFFER_IN_TREE, | |
31 | /* write IO error */ | |
32 | EXTENT_BUFFER_WRITE_ERR, | |
33 | }; | |
b4ce94de | 34 | |
da2c7009 | 35 | /* these are flags for __process_pages_contig */ |
c2790a2e JB |
36 | #define PAGE_UNLOCK (1 << 0) |
37 | #define PAGE_CLEAR_DIRTY (1 << 1) | |
38 | #define PAGE_SET_WRITEBACK (1 << 2) | |
39 | #define PAGE_END_WRITEBACK (1 << 3) | |
40 | #define PAGE_SET_PRIVATE2 (1 << 4) | |
704de49d | 41 | #define PAGE_SET_ERROR (1 << 5) |
da2c7009 | 42 | #define PAGE_LOCK (1 << 6) |
a791e35e | 43 | |
d1310b2e CM |
44 | /* |
45 | * page->private values. Every page that is controlled by the extent | |
46 | * map has page->private set to one. | |
47 | */ | |
48 | #define EXTENT_PAGE_PRIVATE 1 | |
d1310b2e | 49 | |
2fe1d551 OS |
50 | /* |
51 | * The extent buffer bitmap operations are done with byte granularity instead of | |
52 | * word granularity for two reasons: | |
53 | * 1. The bitmaps must be little-endian on disk. | |
54 | * 2. Bitmap items are not guaranteed to be aligned to a word and therefore a | |
55 | * single word in a bitmap may straddle two pages in the extent buffer. | |
56 | */ | |
57 | #define BIT_BYTE(nr) ((nr) / BITS_PER_BYTE) | |
58 | #define BYTE_MASK ((1 << BITS_PER_BYTE) - 1) | |
59 | #define BITMAP_FIRST_BYTE_MASK(start) \ | |
60 | ((BYTE_MASK << ((start) & (BITS_PER_BYTE - 1))) & BYTE_MASK) | |
61 | #define BITMAP_LAST_BYTE_MASK(nbits) \ | |
62 | (BYTE_MASK >> (-(nbits) & (BITS_PER_BYTE - 1))) | |
63 | ||
ea466794 | 64 | struct btrfs_root; |
6fc0ef68 | 65 | struct btrfs_inode; |
facc8a22 | 66 | struct btrfs_io_bio; |
47dc196a | 67 | struct io_failure_record; |
9c7d3a54 | 68 | struct extent_io_tree; |
a758781d | 69 | |
77d5d689 OS |
70 | typedef blk_status_t (submit_bio_hook_t)(struct inode *inode, struct bio *bio, |
71 | int mirror_num, | |
72 | unsigned long bio_flags); | |
73 | ||
a758781d | 74 | typedef blk_status_t (extent_submit_bio_start_t)(void *private_data, |
d0779291 | 75 | struct bio *bio, u64 bio_offset); |
a758781d | 76 | |
d1310b2e | 77 | struct extent_io_ops { |
4d53dddb | 78 | /* |
52042d8e | 79 | * The following callbacks must be always defined, the function |
4d53dddb DS |
80 | * pointer will be called unconditionally. |
81 | */ | |
77d5d689 | 82 | submit_bio_hook_t *submit_bio_hook; |
4d53dddb DS |
83 | int (*readpage_end_io_hook)(struct btrfs_io_bio *io_bio, u64 phy_offset, |
84 | struct page *page, u64 start, u64 end, | |
85 | int mirror); | |
d1310b2e CM |
86 | }; |
87 | ||
d1310b2e | 88 | |
727011e0 | 89 | #define INLINE_EXTENT_BUFFER_PAGES 16 |
09cbfeaf | 90 | #define MAX_INLINE_EXTENT_BUFFER_SIZE (INLINE_EXTENT_BUFFER_PAGES * PAGE_SIZE) |
d1310b2e CM |
91 | struct extent_buffer { |
92 | u64 start; | |
93 | unsigned long len; | |
b4ce94de | 94 | unsigned long bflags; |
f28491e0 | 95 | struct btrfs_fs_info *fs_info; |
3083ee2e | 96 | spinlock_t refs_lock; |
727011e0 | 97 | atomic_t refs; |
0b32f4bb | 98 | atomic_t io_pages; |
5cf1ab56 | 99 | int read_mirror; |
19fe0a8b | 100 | struct rcu_head rcu_head; |
5b25f70f | 101 | pid_t lock_owner; |
b4ce94de | 102 | |
06297d8c | 103 | int blocking_writers; |
bd681513 | 104 | atomic_t blocking_readers; |
ed1b4ed7 | 105 | bool lock_nested; |
656f30db FM |
106 | /* >= 0 if eb belongs to a log tree, -1 otherwise */ |
107 | short log_index; | |
bd681513 CM |
108 | |
109 | /* protects write locks */ | |
110 | rwlock_t lock; | |
111 | ||
112 | /* readers use lock_wq while they wait for the write | |
113 | * lock holders to unlock | |
114 | */ | |
115 | wait_queue_head_t write_lock_wq; | |
b4ce94de | 116 | |
bd681513 CM |
117 | /* writers use read_lock_wq while they wait for readers |
118 | * to unlock | |
b4ce94de | 119 | */ |
bd681513 | 120 | wait_queue_head_t read_lock_wq; |
b8dae313 | 121 | struct page *pages[INLINE_EXTENT_BUFFER_PAGES]; |
6d49ba1b | 122 | #ifdef CONFIG_BTRFS_DEBUG |
f3dc24c5 | 123 | int spinning_writers; |
afd495a8 | 124 | atomic_t spinning_readers; |
5c9c799a | 125 | atomic_t read_locks; |
00801ae4 | 126 | int write_locks; |
6d49ba1b ES |
127 | struct list_head leak_list; |
128 | #endif | |
d1310b2e CM |
129 | }; |
130 | ||
ac467772 QW |
131 | /* |
132 | * Structure to record how many bytes and which ranges are set/cleared | |
133 | */ | |
134 | struct extent_changeset { | |
135 | /* How many bytes are set/cleared in this operation */ | |
7bc329c1 | 136 | unsigned int bytes_changed; |
ac467772 QW |
137 | |
138 | /* Changed ranges */ | |
53d32359 | 139 | struct ulist range_changed; |
ac467772 QW |
140 | }; |
141 | ||
364ecf36 QW |
142 | static inline void extent_changeset_init(struct extent_changeset *changeset) |
143 | { | |
144 | changeset->bytes_changed = 0; | |
145 | ulist_init(&changeset->range_changed); | |
146 | } | |
147 | ||
148 | static inline struct extent_changeset *extent_changeset_alloc(void) | |
149 | { | |
150 | struct extent_changeset *ret; | |
151 | ||
152 | ret = kmalloc(sizeof(*ret), GFP_KERNEL); | |
153 | if (!ret) | |
154 | return NULL; | |
155 | ||
156 | extent_changeset_init(ret); | |
157 | return ret; | |
158 | } | |
159 | ||
160 | static inline void extent_changeset_release(struct extent_changeset *changeset) | |
161 | { | |
162 | if (!changeset) | |
163 | return; | |
164 | changeset->bytes_changed = 0; | |
165 | ulist_release(&changeset->range_changed); | |
166 | } | |
167 | ||
168 | static inline void extent_changeset_free(struct extent_changeset *changeset) | |
169 | { | |
170 | if (!changeset) | |
171 | return; | |
172 | extent_changeset_release(changeset); | |
173 | kfree(changeset); | |
174 | } | |
175 | ||
261507a0 LZ |
176 | static inline void extent_set_compress_type(unsigned long *bio_flags, |
177 | int compress_type) | |
178 | { | |
179 | *bio_flags |= compress_type << EXTENT_BIO_FLAG_SHIFT; | |
180 | } | |
181 | ||
182 | static inline int extent_compress_type(unsigned long bio_flags) | |
183 | { | |
184 | return bio_flags >> EXTENT_BIO_FLAG_SHIFT; | |
185 | } | |
186 | ||
d1310b2e CM |
187 | struct extent_map_tree; |
188 | ||
fc4f21b1 | 189 | typedef struct extent_map *(get_extent_t)(struct btrfs_inode *inode, |
39b07b5d OS |
190 | struct page *page, size_t pg_offset, |
191 | u64 start, u64 len); | |
d1310b2e | 192 | |
477a30ba | 193 | int try_release_extent_mapping(struct page *page, gfp_t mask); |
f7a52a40 | 194 | int try_release_extent_buffer(struct page *page); |
cd716d8f | 195 | |
71ad38b4 DS |
196 | int extent_read_full_page(struct page *page, get_extent_t *get_extent, |
197 | int mirror_num); | |
0a9b0e53 | 198 | int extent_write_full_page(struct page *page, struct writeback_control *wbc); |
5e3ee236 | 199 | int extent_write_locked_range(struct inode *inode, u64 start, u64 end, |
771ed689 | 200 | int mode); |
8ae225a8 | 201 | int extent_writepages(struct address_space *mapping, |
d1310b2e | 202 | struct writeback_control *wbc); |
0b32f4bb JB |
203 | int btree_write_cache_pages(struct address_space *mapping, |
204 | struct writeback_control *wbc); | |
ba206a02 | 205 | void extent_readahead(struct readahead_control *rac); |
1506fcc8 | 206 | int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, |
2135fb9b | 207 | __u64 start, __u64 len); |
d1310b2e CM |
208 | void set_page_extent_mapped(struct page *page); |
209 | ||
f28491e0 | 210 | struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info, |
ce3e6984 | 211 | u64 start); |
0f331229 OS |
212 | struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info, |
213 | u64 start, unsigned long len); | |
3f556f78 | 214 | struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info, |
da17066c | 215 | u64 start); |
2b48966a | 216 | struct extent_buffer *btrfs_clone_extent_buffer(const struct extent_buffer *src); |
f28491e0 | 217 | struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info, |
452c75c3 | 218 | u64 start); |
d1310b2e | 219 | void free_extent_buffer(struct extent_buffer *eb); |
3083ee2e | 220 | void free_extent_buffer_stale(struct extent_buffer *eb); |
bb82ab88 AJ |
221 | #define WAIT_NONE 0 |
222 | #define WAIT_COMPLETE 1 | |
223 | #define WAIT_PAGE_LOCK 2 | |
c2ccfbc6 | 224 | int read_extent_buffer_pages(struct extent_buffer *eb, int wait, |
6af49dbd | 225 | int mirror_num); |
fd8b2b61 | 226 | void wait_on_extent_buffer_writeback(struct extent_buffer *eb); |
479ed9ab | 227 | |
cc5e31a4 | 228 | static inline int num_extent_pages(const struct extent_buffer *eb) |
479ed9ab | 229 | { |
8791d432 DS |
230 | return (round_up(eb->start + eb->len, PAGE_SIZE) >> PAGE_SHIFT) - |
231 | (eb->start >> PAGE_SHIFT); | |
479ed9ab RD |
232 | } |
233 | ||
2b48966a | 234 | static inline int extent_buffer_uptodate(const struct extent_buffer *eb) |
ba020491 AJ |
235 | { |
236 | return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags); | |
237 | } | |
238 | ||
1cbb1f45 JM |
239 | int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv, |
240 | unsigned long start, unsigned long len); | |
241 | void read_extent_buffer(const struct extent_buffer *eb, void *dst, | |
d1310b2e CM |
242 | unsigned long start, |
243 | unsigned long len); | |
1cbb1f45 JM |
244 | int read_extent_buffer_to_user(const struct extent_buffer *eb, |
245 | void __user *dst, unsigned long start, | |
550ac1d8 | 246 | unsigned long len); |
2b48966a DS |
247 | void write_extent_buffer_fsid(const struct extent_buffer *eb, const void *src); |
248 | void write_extent_buffer_chunk_tree_uuid(const struct extent_buffer *eb, | |
f157bf76 | 249 | const void *src); |
2b48966a | 250 | void write_extent_buffer(const struct extent_buffer *eb, const void *src, |
d1310b2e | 251 | unsigned long start, unsigned long len); |
2b48966a DS |
252 | void copy_extent_buffer_full(const struct extent_buffer *dst, |
253 | const struct extent_buffer *src); | |
254 | void copy_extent_buffer(const struct extent_buffer *dst, | |
255 | const struct extent_buffer *src, | |
d1310b2e CM |
256 | unsigned long dst_offset, unsigned long src_offset, |
257 | unsigned long len); | |
2b48966a DS |
258 | void memcpy_extent_buffer(const struct extent_buffer *dst, |
259 | unsigned long dst_offset, unsigned long src_offset, | |
260 | unsigned long len); | |
261 | void memmove_extent_buffer(const struct extent_buffer *dst, | |
262 | unsigned long dst_offset, unsigned long src_offset, | |
b159fa28 | 263 | unsigned long len); |
2b48966a | 264 | void memzero_extent_buffer(const struct extent_buffer *eb, unsigned long start, |
b159fa28 | 265 | unsigned long len); |
2b48966a | 266 | int extent_buffer_test_bit(const struct extent_buffer *eb, unsigned long start, |
3e1e8bb7 | 267 | unsigned long pos); |
2b48966a | 268 | void extent_buffer_bitmap_set(const struct extent_buffer *eb, unsigned long start, |
3e1e8bb7 | 269 | unsigned long pos, unsigned long len); |
2b48966a DS |
270 | void extent_buffer_bitmap_clear(const struct extent_buffer *eb, |
271 | unsigned long start, unsigned long pos, | |
272 | unsigned long len); | |
273 | void clear_extent_buffer_dirty(const struct extent_buffer *eb); | |
abb57ef3 | 274 | bool set_extent_buffer_dirty(struct extent_buffer *eb); |
09c25a8c | 275 | void set_extent_buffer_uptodate(struct extent_buffer *eb); |
69ba3927 | 276 | void clear_extent_buffer_uptodate(struct extent_buffer *eb); |
2b48966a | 277 | int extent_buffer_under_io(const struct extent_buffer *eb); |
bd1fa4f0 | 278 | void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end); |
f6311572 | 279 | void extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end); |
ad7ff17b | 280 | void extent_clear_unlock_delalloc(struct btrfs_inode *inode, u64 start, u64 end, |
74e9194a NB |
281 | struct page *locked_page, |
282 | unsigned bits_to_clear, | |
283 | unsigned long page_ops); | |
e749af44 | 284 | struct bio *btrfs_bio_alloc(u64 first_byte); |
c5e4c3d7 | 285 | struct bio *btrfs_io_bio_alloc(unsigned int nr_iovecs); |
8b6c1d56 | 286 | struct bio *btrfs_bio_clone(struct bio *bio); |
e477094f | 287 | struct bio *btrfs_bio_clone_partial(struct bio *orig, int offset, int size); |
4a54c8c1 | 288 | |
3ec706c8 | 289 | struct btrfs_fs_info; |
9d4f7f8a | 290 | struct btrfs_inode; |
4a54c8c1 | 291 | |
6ec656bc JB |
292 | int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start, |
293 | u64 length, u64 logical, struct page *page, | |
294 | unsigned int pg_offset, int mirror_num); | |
b5227c07 | 295 | void end_extent_writepage(struct page *page, int err, u64 start, u64 end); |
2b48966a | 296 | int btrfs_repair_eb_io_failure(const struct extent_buffer *eb, int mirror_num); |
2fe6303e MX |
297 | |
298 | /* | |
299 | * When IO fails, either with EIO or csum verification fails, we | |
300 | * try other mirrors that might have a good copy of the data. This | |
301 | * io_failure_record is used to record state as we go through all the | |
302 | * mirrors. If another mirror has good data, the page is set up to date | |
303 | * and things continue. If a good mirror can't be found, the original | |
304 | * bio end_io callback is called to indicate things have failed. | |
305 | */ | |
306 | struct io_failure_record { | |
307 | struct page *page; | |
308 | u64 start; | |
309 | u64 len; | |
310 | u64 logical; | |
311 | unsigned long bio_flags; | |
312 | int this_mirror; | |
313 | int failed_mirror; | |
314 | int in_validation; | |
315 | }; | |
316 | ||
4ac1f4ac | 317 | |
77d5d689 OS |
318 | blk_status_t btrfs_submit_read_repair(struct inode *inode, |
319 | struct bio *failed_bio, u64 phy_offset, | |
320 | struct page *page, unsigned int pgoff, | |
321 | u64 start, u64 end, int failed_mirror, | |
322 | submit_bio_hook_t *submit_bio_hook); | |
323 | ||
294e30fe | 324 | #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS |
9978059b | 325 | bool find_lock_delalloc_range(struct inode *inode, |
ce9f967f JT |
326 | struct page *locked_page, u64 *start, |
327 | u64 *end); | |
0d4cf4e6 | 328 | #endif |
faa2dbf0 | 329 | struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info, |
da17066c | 330 | u64 start); |
3fd63727 JB |
331 | |
332 | #ifdef CONFIG_BTRFS_DEBUG | |
333 | void btrfs_extent_buffer_leak_debug_check(struct btrfs_fs_info *fs_info); | |
334 | #else | |
335 | #define btrfs_extent_buffer_leak_debug_check(fs_info) do {} while (0) | |
336 | #endif | |
337 | ||
294e30fe | 338 | #endif |