]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame - fs/btrfs/extent_io.h
btrfs: update num_extent_pages to support subpage sized extent buffer
[mirror_ubuntu-jammy-kernel.git] / fs / btrfs / extent_io.h
CommitLineData
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>
deb67895 9#include <linux/btrfs_tree.h>
ac467772 10#include "ulist.h"
d1310b2e 11
261507a0
LZ
12/*
13 * flags for bio submission. The high bits indicate the compression
14 * type for this bio
15 */
c8b97818 16#define EXTENT_BIO_COMPRESSED 1
261507a0 17#define EXTENT_BIO_FLAG_SHIFT 16
c8b97818 18
80cb3836
DS
19enum {
20 EXTENT_BUFFER_UPTODATE,
21 EXTENT_BUFFER_DIRTY,
22 EXTENT_BUFFER_CORRUPT,
23 /* this got triggered by readahead */
24 EXTENT_BUFFER_READAHEAD,
25 EXTENT_BUFFER_TREE_REF,
26 EXTENT_BUFFER_STALE,
27 EXTENT_BUFFER_WRITEBACK,
28 /* read IO error */
29 EXTENT_BUFFER_READ_ERR,
30 EXTENT_BUFFER_UNMAPPED,
31 EXTENT_BUFFER_IN_TREE,
32 /* write IO error */
33 EXTENT_BUFFER_WRITE_ERR,
34};
b4ce94de 35
da2c7009 36/* these are flags for __process_pages_contig */
c2790a2e
JB
37#define PAGE_UNLOCK (1 << 0)
38#define PAGE_CLEAR_DIRTY (1 << 1)
39#define PAGE_SET_WRITEBACK (1 << 2)
40#define PAGE_END_WRITEBACK (1 << 3)
41#define PAGE_SET_PRIVATE2 (1 << 4)
704de49d 42#define PAGE_SET_ERROR (1 << 5)
da2c7009 43#define PAGE_LOCK (1 << 6)
a791e35e 44
d1310b2e
CM
45/*
46 * page->private values. Every page that is controlled by the extent
47 * map has page->private set to one.
48 */
49#define EXTENT_PAGE_PRIVATE 1
d1310b2e 50
2fe1d551
OS
51/*
52 * The extent buffer bitmap operations are done with byte granularity instead of
53 * word granularity for two reasons:
54 * 1. The bitmaps must be little-endian on disk.
55 * 2. Bitmap items are not guaranteed to be aligned to a word and therefore a
56 * single word in a bitmap may straddle two pages in the extent buffer.
57 */
58#define BIT_BYTE(nr) ((nr) / BITS_PER_BYTE)
59#define BYTE_MASK ((1 << BITS_PER_BYTE) - 1)
60#define BITMAP_FIRST_BYTE_MASK(start) \
61 ((BYTE_MASK << ((start) & (BITS_PER_BYTE - 1))) & BYTE_MASK)
62#define BITMAP_LAST_BYTE_MASK(nbits) \
63 (BYTE_MASK >> (-(nbits) & (BITS_PER_BYTE - 1)))
64
ea466794 65struct btrfs_root;
6fc0ef68 66struct btrfs_inode;
facc8a22 67struct btrfs_io_bio;
47dc196a 68struct io_failure_record;
9c7d3a54 69struct extent_io_tree;
a758781d 70
77d5d689
OS
71typedef blk_status_t (submit_bio_hook_t)(struct inode *inode, struct bio *bio,
72 int mirror_num,
73 unsigned long bio_flags);
74
8896a08d 75typedef blk_status_t (extent_submit_bio_start_t)(struct inode *inode,
1941b64b 76 struct bio *bio, u64 dio_file_offset);
a758781d 77
deb67895 78#define INLINE_EXTENT_BUFFER_PAGES (BTRFS_MAX_METADATA_BLOCKSIZE / PAGE_SIZE)
d1310b2e
CM
79struct extent_buffer {
80 u64 start;
81 unsigned long len;
b4ce94de 82 unsigned long bflags;
f28491e0 83 struct btrfs_fs_info *fs_info;
3083ee2e 84 spinlock_t refs_lock;
727011e0 85 atomic_t refs;
0b32f4bb 86 atomic_t io_pages;
5cf1ab56 87 int read_mirror;
19fe0a8b 88 struct rcu_head rcu_head;
5b25f70f 89 pid_t lock_owner;
656f30db 90 /* >= 0 if eb belongs to a log tree, -1 otherwise */
dc516164
DS
91 s8 log_index;
92
93 struct rw_semaphore lock;
bd681513 94
b8dae313 95 struct page *pages[INLINE_EXTENT_BUFFER_PAGES];
6d49ba1b
ES
96#ifdef CONFIG_BTRFS_DEBUG
97 struct list_head leak_list;
98#endif
d1310b2e
CM
99};
100
ac467772
QW
101/*
102 * Structure to record how many bytes and which ranges are set/cleared
103 */
104struct extent_changeset {
105 /* How many bytes are set/cleared in this operation */
7bc329c1 106 unsigned int bytes_changed;
ac467772
QW
107
108 /* Changed ranges */
53d32359 109 struct ulist range_changed;
ac467772
QW
110};
111
364ecf36
QW
112static inline void extent_changeset_init(struct extent_changeset *changeset)
113{
114 changeset->bytes_changed = 0;
115 ulist_init(&changeset->range_changed);
116}
117
118static inline struct extent_changeset *extent_changeset_alloc(void)
119{
120 struct extent_changeset *ret;
121
122 ret = kmalloc(sizeof(*ret), GFP_KERNEL);
123 if (!ret)
124 return NULL;
125
126 extent_changeset_init(ret);
127 return ret;
128}
129
130static inline void extent_changeset_release(struct extent_changeset *changeset)
131{
132 if (!changeset)
133 return;
134 changeset->bytes_changed = 0;
135 ulist_release(&changeset->range_changed);
136}
137
138static inline void extent_changeset_free(struct extent_changeset *changeset)
139{
140 if (!changeset)
141 return;
142 extent_changeset_release(changeset);
143 kfree(changeset);
144}
145
261507a0
LZ
146static inline void extent_set_compress_type(unsigned long *bio_flags,
147 int compress_type)
148{
149 *bio_flags |= compress_type << EXTENT_BIO_FLAG_SHIFT;
150}
151
152static inline int extent_compress_type(unsigned long bio_flags)
153{
154 return bio_flags >> EXTENT_BIO_FLAG_SHIFT;
155}
156
d1310b2e
CM
157struct extent_map_tree;
158
fc4f21b1 159typedef struct extent_map *(get_extent_t)(struct btrfs_inode *inode,
39b07b5d
OS
160 struct page *page, size_t pg_offset,
161 u64 start, u64 len);
d1310b2e 162
477a30ba 163int try_release_extent_mapping(struct page *page, gfp_t mask);
f7a52a40 164int try_release_extent_buffer(struct page *page);
cd716d8f 165
c1be9c1a
NB
166int __must_check submit_one_bio(struct bio *bio, int mirror_num,
167 unsigned long bio_flags);
0f208812
NB
168int btrfs_do_readpage(struct page *page, struct extent_map **em_cached,
169 struct bio **bio, unsigned long *bio_flags,
170 unsigned int read_flags, u64 *prev_em_start);
0a9b0e53 171int extent_write_full_page(struct page *page, struct writeback_control *wbc);
5e3ee236 172int extent_write_locked_range(struct inode *inode, u64 start, u64 end,
771ed689 173 int mode);
8ae225a8 174int extent_writepages(struct address_space *mapping,
d1310b2e 175 struct writeback_control *wbc);
0b32f4bb
JB
176int btree_write_cache_pages(struct address_space *mapping,
177 struct writeback_control *wbc);
ba206a02 178void extent_readahead(struct readahead_control *rac);
facee0a0 179int extent_fiemap(struct btrfs_inode *inode, struct fiemap_extent_info *fieinfo,
bab16e21 180 u64 start, u64 len);
d1310b2e
CM
181void set_page_extent_mapped(struct page *page);
182
f28491e0 183struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
3fbaf258 184 u64 start, u64 owner_root, int level);
0f331229
OS
185struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
186 u64 start, unsigned long len);
3f556f78 187struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
da17066c 188 u64 start);
2b48966a 189struct extent_buffer *btrfs_clone_extent_buffer(const struct extent_buffer *src);
f28491e0 190struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info,
452c75c3 191 u64 start);
d1310b2e 192void free_extent_buffer(struct extent_buffer *eb);
3083ee2e 193void free_extent_buffer_stale(struct extent_buffer *eb);
bb82ab88
AJ
194#define WAIT_NONE 0
195#define WAIT_COMPLETE 1
196#define WAIT_PAGE_LOCK 2
c2ccfbc6 197int read_extent_buffer_pages(struct extent_buffer *eb, int wait,
6af49dbd 198 int mirror_num);
fd8b2b61 199void wait_on_extent_buffer_writeback(struct extent_buffer *eb);
bfb484d9 200void btrfs_readahead_tree_block(struct btrfs_fs_info *fs_info,
3fbaf258 201 u64 bytenr, u64 owner_root, u64 gen, int level);
bfb484d9 202void btrfs_readahead_node_child(struct extent_buffer *node, int slot);
479ed9ab 203
cc5e31a4 204static inline int num_extent_pages(const struct extent_buffer *eb)
479ed9ab 205{
4a3dc938
QW
206 /*
207 * For sectorsize == PAGE_SIZE case, since nodesize is always aligned to
208 * sectorsize, it's just eb->len >> PAGE_SHIFT.
209 *
210 * For sectorsize < PAGE_SIZE case, we could have nodesize < PAGE_SIZE,
211 * thus have to ensure we get at least one page.
212 */
213 return (eb->len >> PAGE_SHIFT) ?: 1;
479ed9ab
RD
214}
215
2b48966a 216static inline int extent_buffer_uptodate(const struct extent_buffer *eb)
ba020491
AJ
217{
218 return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
219}
220
1cbb1f45
JM
221int memcmp_extent_buffer(const struct extent_buffer *eb, const void *ptrv,
222 unsigned long start, unsigned long len);
223void read_extent_buffer(const struct extent_buffer *eb, void *dst,
d1310b2e
CM
224 unsigned long start,
225 unsigned long len);
a48b73ec
JB
226int read_extent_buffer_to_user_nofault(const struct extent_buffer *eb,
227 void __user *dst, unsigned long start,
228 unsigned long len);
2b48966a
DS
229void write_extent_buffer_fsid(const struct extent_buffer *eb, const void *src);
230void write_extent_buffer_chunk_tree_uuid(const struct extent_buffer *eb,
f157bf76 231 const void *src);
2b48966a 232void write_extent_buffer(const struct extent_buffer *eb, const void *src,
d1310b2e 233 unsigned long start, unsigned long len);
2b48966a
DS
234void copy_extent_buffer_full(const struct extent_buffer *dst,
235 const struct extent_buffer *src);
236void copy_extent_buffer(const struct extent_buffer *dst,
237 const struct extent_buffer *src,
d1310b2e
CM
238 unsigned long dst_offset, unsigned long src_offset,
239 unsigned long len);
2b48966a
DS
240void memcpy_extent_buffer(const struct extent_buffer *dst,
241 unsigned long dst_offset, unsigned long src_offset,
242 unsigned long len);
243void memmove_extent_buffer(const struct extent_buffer *dst,
244 unsigned long dst_offset, unsigned long src_offset,
b159fa28 245 unsigned long len);
2b48966a 246void memzero_extent_buffer(const struct extent_buffer *eb, unsigned long start,
b159fa28 247 unsigned long len);
2b48966a 248int extent_buffer_test_bit(const struct extent_buffer *eb, unsigned long start,
3e1e8bb7 249 unsigned long pos);
2b48966a 250void extent_buffer_bitmap_set(const struct extent_buffer *eb, unsigned long start,
3e1e8bb7 251 unsigned long pos, unsigned long len);
2b48966a
DS
252void extent_buffer_bitmap_clear(const struct extent_buffer *eb,
253 unsigned long start, unsigned long pos,
254 unsigned long len);
255void clear_extent_buffer_dirty(const struct extent_buffer *eb);
abb57ef3 256bool set_extent_buffer_dirty(struct extent_buffer *eb);
09c25a8c 257void set_extent_buffer_uptodate(struct extent_buffer *eb);
69ba3927 258void clear_extent_buffer_uptodate(struct extent_buffer *eb);
2b48966a 259int extent_buffer_under_io(const struct extent_buffer *eb);
bd1fa4f0 260void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end);
f6311572 261void extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end);
ad7ff17b 262void extent_clear_unlock_delalloc(struct btrfs_inode *inode, u64 start, u64 end,
74e9194a 263 struct page *locked_page,
f97e27e9 264 u32 bits_to_clear, unsigned long page_ops);
e749af44 265struct bio *btrfs_bio_alloc(u64 first_byte);
c5e4c3d7 266struct bio *btrfs_io_bio_alloc(unsigned int nr_iovecs);
8b6c1d56 267struct bio *btrfs_bio_clone(struct bio *bio);
e477094f 268struct bio *btrfs_bio_clone_partial(struct bio *orig, int offset, int size);
4a54c8c1 269
3ec706c8 270struct btrfs_fs_info;
9d4f7f8a 271struct btrfs_inode;
4a54c8c1 272
6ec656bc
JB
273int repair_io_failure(struct btrfs_fs_info *fs_info, u64 ino, u64 start,
274 u64 length, u64 logical, struct page *page,
275 unsigned int pg_offset, int mirror_num);
b5227c07 276void end_extent_writepage(struct page *page, int err, u64 start, u64 end);
2b48966a 277int btrfs_repair_eb_io_failure(const struct extent_buffer *eb, int mirror_num);
2fe6303e
MX
278
279/*
280 * When IO fails, either with EIO or csum verification fails, we
281 * try other mirrors that might have a good copy of the data. This
282 * io_failure_record is used to record state as we go through all the
283 * mirrors. If another mirror has good data, the page is set up to date
284 * and things continue. If a good mirror can't be found, the original
285 * bio end_io callback is called to indicate things have failed.
286 */
287struct io_failure_record {
288 struct page *page;
289 u64 start;
290 u64 len;
291 u64 logical;
292 unsigned long bio_flags;
293 int this_mirror;
294 int failed_mirror;
295 int in_validation;
296};
297
4ac1f4ac 298
77d5d689 299blk_status_t btrfs_submit_read_repair(struct inode *inode,
7ffd27e3 300 struct bio *failed_bio, u32 bio_offset,
77d5d689
OS
301 struct page *page, unsigned int pgoff,
302 u64 start, u64 end, int failed_mirror,
303 submit_bio_hook_t *submit_bio_hook);
304
294e30fe 305#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
9978059b 306bool find_lock_delalloc_range(struct inode *inode,
ce9f967f
JT
307 struct page *locked_page, u64 *start,
308 u64 *end);
0d4cf4e6 309#endif
faa2dbf0 310struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info,
da17066c 311 u64 start);
3fd63727
JB
312
313#ifdef CONFIG_BTRFS_DEBUG
314void btrfs_extent_buffer_leak_debug_check(struct btrfs_fs_info *fs_info);
315#else
316#define btrfs_extent_buffer_leak_debug_check(fs_info) do {} while (0)
317#endif
318
294e30fe 319#endif