Commit | Line | Data |
---|---|---|
9888c340 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
6cbd5570 CM |
2 | /* |
3 | * Copyright (C) 2007 Oracle. All rights reserved. | |
6cbd5570 CM |
4 | */ |
5 | ||
9888c340 DS |
6 | #ifndef BTRFS_INODE_H |
7 | #define BTRFS_INODE_H | |
2c90e5d6 | 8 | |
778ba82b | 9 | #include <linux/hash.h> |
e3b318d1 | 10 | #include <linux/refcount.h> |
a52d9a80 | 11 | #include "extent_map.h" |
d1310b2e | 12 | #include "extent_io.h" |
e6dcd2dc | 13 | #include "ordered-data.h" |
16cdcec7 | 14 | #include "delayed-inode.h" |
a52d9a80 | 15 | |
72ac3c0d JB |
16 | /* |
17 | * ordered_data_close is set by truncate when a file that used | |
18 | * to have good data has been truncated to zero. When it is set | |
19 | * the btrfs file release call will add this inode to the | |
20 | * ordered operations list so that we make sure to flush out any | |
21 | * new data the application may have written before commit. | |
22 | */ | |
7efc3e34 | 23 | enum { |
1fd4033d | 24 | BTRFS_INODE_FLUSH_ON_CLOSE, |
7efc3e34 OS |
25 | BTRFS_INODE_DUMMY, |
26 | BTRFS_INODE_IN_DEFRAG, | |
27 | BTRFS_INODE_HAS_ASYNC_EXTENT, | |
48778179 FM |
28 | /* |
29 | * Always set under the VFS' inode lock, otherwise it can cause races | |
30 | * during fsync (we start as a fast fsync and then end up in a full | |
31 | * fsync racing with ordered extent completion). | |
32 | */ | |
7efc3e34 OS |
33 | BTRFS_INODE_NEEDS_FULL_SYNC, |
34 | BTRFS_INODE_COPY_EVERYTHING, | |
35 | BTRFS_INODE_IN_DELALLOC_LIST, | |
7efc3e34 | 36 | BTRFS_INODE_HAS_PROPS, |
3cd24c69 | 37 | BTRFS_INODE_SNAPSHOT_FLUSH, |
7efc3e34 | 38 | }; |
72ac3c0d | 39 | |
f1ace244 | 40 | /* in memory btrfs inode */ |
2c90e5d6 | 41 | struct btrfs_inode { |
d352ac68 | 42 | /* which subvolume this inode belongs to */ |
d6e4a428 | 43 | struct btrfs_root *root; |
d352ac68 | 44 | |
d352ac68 CM |
45 | /* key used to find this inode on disk. This is used by the code |
46 | * to read in roots of subvolumes | |
47 | */ | |
d6e4a428 | 48 | struct btrfs_key location; |
d352ac68 | 49 | |
2f2ff0ee FM |
50 | /* |
51 | * Lock for counters and all fields used to determine if the inode is in | |
52 | * the log or not (last_trans, last_sub_trans, last_log_commit, | |
53 | * logged_trans). | |
54 | */ | |
9e0baf60 JB |
55 | spinlock_t lock; |
56 | ||
d352ac68 | 57 | /* the extent_tree has caches of all the extent mappings to disk */ |
a52d9a80 | 58 | struct extent_map_tree extent_tree; |
d352ac68 CM |
59 | |
60 | /* the io_tree does range state (DIRTY, LOCKED etc) */ | |
d1310b2e | 61 | struct extent_io_tree io_tree; |
d352ac68 CM |
62 | |
63 | /* special utility tree used to record which mirrors have already been | |
64 | * tried when checksums fail for a given block | |
65 | */ | |
7e38326f | 66 | struct extent_io_tree io_failure_tree; |
d352ac68 | 67 | |
41a2ee75 JB |
68 | /* |
69 | * Keep track of where the inode has extent items mapped in order to | |
70 | * make sure the i_size adjustments are accurate | |
71 | */ | |
72 | struct extent_io_tree file_extent_tree; | |
73 | ||
d352ac68 | 74 | /* held while logging the inode in tree-log.c */ |
e02119d5 | 75 | struct mutex log_mutex; |
d352ac68 CM |
76 | |
77 | /* used to order data wrt metadata */ | |
e6dcd2dc | 78 | struct btrfs_ordered_inode_tree ordered_tree; |
15ee9bc7 | 79 | |
d352ac68 CM |
80 | /* list of all the delalloc inodes in the FS. There are times we need |
81 | * to write all the delalloc pages to disk, and this list is used | |
82 | * to walk them all. | |
83 | */ | |
ea8c2819 CM |
84 | struct list_head delalloc_inodes; |
85 | ||
5d4f98a2 YZ |
86 | /* node for the red-black tree that links inodes in subvolume root */ |
87 | struct rb_node rb_node; | |
88 | ||
72ac3c0d JB |
89 | unsigned long runtime_flags; |
90 | ||
9c931c5a | 91 | /* Keep track of who's O_SYNC/fsyncing currently */ |
b812ce28 JB |
92 | atomic_t sync_writers; |
93 | ||
d352ac68 CM |
94 | /* full 64 bit generation number, struct vfs_inode doesn't have a big |
95 | * enough field for this. | |
96 | */ | |
e02119d5 CM |
97 | u64 generation; |
98 | ||
15ee9bc7 JB |
99 | /* |
100 | * transid of the trans_handle that last modified this inode | |
101 | */ | |
102 | u64 last_trans; | |
257c62e1 CM |
103 | |
104 | /* | |
bb14a59b | 105 | * transid that last logged this inode |
257c62e1 | 106 | */ |
bb14a59b | 107 | u64 logged_trans; |
257c62e1 | 108 | |
e02119d5 | 109 | /* |
bb14a59b | 110 | * log transid when this inode was last modified |
e02119d5 | 111 | */ |
bb14a59b MX |
112 | int last_sub_trans; |
113 | ||
114 | /* a local copy of root's last_log_commit */ | |
115 | int last_log_commit; | |
49eb7e46 | 116 | |
d352ac68 CM |
117 | /* total number of bytes pending delalloc, used by stat to calc the |
118 | * real block usage of the file | |
119 | */ | |
9069218d | 120 | u64 delalloc_bytes; |
d352ac68 | 121 | |
a7e3b975 FM |
122 | /* |
123 | * Total number of bytes pending delalloc that fall within a file | |
124 | * range that is either a hole or beyond EOF (and no prealloc extent | |
125 | * exists in the range). This is always <= delalloc_bytes. | |
126 | */ | |
127 | u64 new_delalloc_bytes; | |
128 | ||
47059d93 WS |
129 | /* |
130 | * total number of bytes pending defrag, used by stat to check whether | |
131 | * it needs COW. | |
132 | */ | |
133 | u64 defrag_bytes; | |
134 | ||
d352ac68 CM |
135 | /* |
136 | * the size of the file stored in the metadata on disk. data=ordered | |
137 | * means the in-memory i_size might be larger than the size on disk | |
138 | * because not all the blocks are written yet. | |
139 | */ | |
dbe674a9 | 140 | u64 disk_i_size; |
d352ac68 | 141 | |
aec7477b JB |
142 | /* |
143 | * if this is a directory then index_cnt is the counter for the index | |
144 | * number for new files that are created | |
145 | */ | |
146 | u64 index_cnt; | |
d352ac68 | 147 | |
67de1176 MX |
148 | /* Cache the directory index number to speed the dir/file remove */ |
149 | u64 dir_index; | |
150 | ||
12fcfd22 CM |
151 | /* the fsync log has some corner cases that mean we have to check |
152 | * directories to see if any unlinks have been done before | |
153 | * the directory was logged. See tree-log.c for all the | |
154 | * details | |
155 | */ | |
156 | u64 last_unlink_trans; | |
157 | ||
3ebac17c FM |
158 | /* |
159 | * The id/generation of the last transaction where this inode was | |
160 | * either the source or the destination of a clone/dedupe operation. | |
161 | * Used when logging an inode to know if there are shared extents that | |
162 | * need special care when logging checksum items, to avoid duplicate | |
163 | * checksum items in a log (which can lead to a corruption where we end | |
164 | * up with missing checksum ranges after log replay). | |
165 | * Protected by the vfs inode lock. | |
166 | */ | |
167 | u64 last_reflink_trans; | |
168 | ||
7709cde3 JB |
169 | /* |
170 | * Number of bytes outstanding that are going to need csums. This is | |
171 | * used in ENOSPC accounting. | |
172 | */ | |
173 | u64 csum_bytes; | |
174 | ||
f1bdcc0a JB |
175 | /* flags field from the on disk inode */ |
176 | u32 flags; | |
177 | ||
9ed74f2d | 178 | /* |
32c00aff JB |
179 | * Counters to keep track of the number of extent item's we may use due |
180 | * to delalloc and such. outstanding_extents is the number of extent | |
181 | * items we think we'll end up using, and reserved_extents is the number | |
182 | * of extent items we've reserved metadata for. | |
9ed74f2d | 183 | */ |
9e0baf60 | 184 | unsigned outstanding_extents; |
69fe2d75 JB |
185 | |
186 | struct btrfs_block_rsv block_rsv; | |
9ed74f2d | 187 | |
1e701a32 | 188 | /* |
b52aa8c9 | 189 | * Cached values of inode properties |
1e701a32 | 190 | */ |
b52aa8c9 | 191 | unsigned prop_compress; /* per-file compression algorithm */ |
eec63c65 DS |
192 | /* |
193 | * Force compression on the file using the defrag ioctl, could be | |
194 | * different from prop_compress and takes precedence if set | |
195 | */ | |
196 | unsigned defrag_compress; | |
1e701a32 | 197 | |
16cdcec7 MX |
198 | struct btrfs_delayed_node *delayed_node; |
199 | ||
9cc97d64 | 200 | /* File creation time. */ |
d3c6be6f | 201 | struct timespec64 i_otime; |
9cc97d64 | 202 | |
8089fe62 DS |
203 | /* Hook into fs_info->delayed_iputs */ |
204 | struct list_head delayed_iput; | |
8089fe62 | 205 | |
d352ac68 | 206 | struct inode vfs_inode; |
2c90e5d6 | 207 | }; |
dbe674a9 | 208 | |
6fee248d NB |
209 | static inline u32 btrfs_inode_sectorsize(const struct btrfs_inode *inode) |
210 | { | |
211 | return inode->root->fs_info->sectorsize; | |
212 | } | |
213 | ||
9a35b637 | 214 | static inline struct btrfs_inode *BTRFS_I(const struct inode *inode) |
2c90e5d6 CM |
215 | { |
216 | return container_of(inode, struct btrfs_inode, vfs_inode); | |
217 | } | |
218 | ||
778ba82b FDBM |
219 | static inline unsigned long btrfs_inode_hash(u64 objectid, |
220 | const struct btrfs_root *root) | |
221 | { | |
4fd786e6 | 222 | u64 h = objectid ^ (root->root_key.objectid * GOLDEN_RATIO_PRIME); |
778ba82b FDBM |
223 | |
224 | #if BITS_PER_LONG == 32 | |
225 | h = (h >> 32) ^ (h & 0xffffffff); | |
226 | #endif | |
227 | ||
228 | return (unsigned long)h; | |
229 | } | |
230 | ||
231 | static inline void btrfs_insert_inode_hash(struct inode *inode) | |
232 | { | |
233 | unsigned long h = btrfs_inode_hash(inode->i_ino, BTRFS_I(inode)->root); | |
234 | ||
235 | __insert_inode_hash(inode, h); | |
236 | } | |
237 | ||
9a35b637 | 238 | static inline u64 btrfs_ino(const struct btrfs_inode *inode) |
33345d01 | 239 | { |
4a0cc7ca | 240 | u64 ino = inode->location.objectid; |
33345d01 | 241 | |
14c7cca7 LB |
242 | /* |
243 | * !ino: btree_inode | |
244 | * type == BTRFS_ROOT_ITEM_KEY: subvol dir | |
245 | */ | |
4a0cc7ca NB |
246 | if (!ino || inode->location.type == BTRFS_ROOT_ITEM_KEY) |
247 | ino = inode->vfs_inode.i_ino; | |
33345d01 LZ |
248 | return ino; |
249 | } | |
250 | ||
6ef06d27 | 251 | static inline void btrfs_i_size_write(struct btrfs_inode *inode, u64 size) |
dbe674a9 | 252 | { |
6ef06d27 NB |
253 | i_size_write(&inode->vfs_inode, size); |
254 | inode->disk_i_size = size; | |
dbe674a9 CM |
255 | } |
256 | ||
70ddc553 | 257 | static inline bool btrfs_is_free_space_inode(struct btrfs_inode *inode) |
2cf8572d | 258 | { |
70ddc553 | 259 | struct btrfs_root *root = inode->root; |
83eea1f1 | 260 | |
51a8cf9d | 261 | if (root == root->fs_info->tree_root && |
70ddc553 | 262 | btrfs_ino(inode) != BTRFS_BTREE_INODE_OBJECTID) |
51a8cf9d | 263 | return true; |
70ddc553 | 264 | if (inode->location.objectid == BTRFS_FREE_INO_OBJECTID) |
2cf8572d CM |
265 | return true; |
266 | return false; | |
267 | } | |
268 | ||
06f2548f NB |
269 | static inline bool is_data_inode(struct inode *inode) |
270 | { | |
271 | return btrfs_ino(BTRFS_I(inode)) != BTRFS_BTREE_INODE_OBJECTID; | |
272 | } | |
273 | ||
8b62f87b JB |
274 | static inline void btrfs_mod_outstanding_extents(struct btrfs_inode *inode, |
275 | int mod) | |
276 | { | |
277 | lockdep_assert_held(&inode->lock); | |
278 | inode->outstanding_extents += mod; | |
279 | if (btrfs_is_free_space_inode(inode)) | |
280 | return; | |
dd48d407 JB |
281 | trace_btrfs_inode_mod_outstanding_extents(inode->root, btrfs_ino(inode), |
282 | mod); | |
8b62f87b JB |
283 | } |
284 | ||
0f8939b8 | 285 | static inline int btrfs_inode_in_log(struct btrfs_inode *inode, u64 generation) |
22ee6985 | 286 | { |
2f2ff0ee FM |
287 | int ret = 0; |
288 | ||
0f8939b8 NB |
289 | spin_lock(&inode->lock); |
290 | if (inode->logged_trans == generation && | |
291 | inode->last_sub_trans <= inode->last_log_commit && | |
292 | inode->last_sub_trans <= inode->root->last_log_commit) { | |
125c4cf9 FM |
293 | /* |
294 | * After a ranged fsync we might have left some extent maps | |
295 | * (that fall outside the fsync's range). So return false | |
296 | * here if the list isn't empty, to make sure btrfs_log_inode() | |
297 | * will be called and process those extent maps. | |
298 | */ | |
299 | smp_mb(); | |
0f8939b8 | 300 | if (list_empty(&inode->extent_tree.modified_extents)) |
2f2ff0ee | 301 | ret = 1; |
125c4cf9 | 302 | } |
0f8939b8 | 303 | spin_unlock(&inode->lock); |
2f2ff0ee | 304 | return ret; |
22ee6985 JB |
305 | } |
306 | ||
facc8a22 MX |
307 | struct btrfs_dio_private { |
308 | struct inode *inode; | |
309 | u64 logical_offset; | |
310 | u64 disk_bytenr; | |
311 | u64 bytes; | |
facc8a22 | 312 | |
e3b318d1 OS |
313 | /* |
314 | * References to this structure. There is one reference per in-flight | |
315 | * bio plus one while we're still setting up. | |
316 | */ | |
317 | refcount_t refs; | |
facc8a22 | 318 | |
facc8a22 MX |
319 | /* dio_bio came from fs/direct-io.c */ |
320 | struct bio *dio_bio; | |
c1dc0896 | 321 | |
85879573 OS |
322 | /* Array of checksums */ |
323 | u8 csums[]; | |
facc8a22 MX |
324 | }; |
325 | ||
7ebc7e5f JT |
326 | /* Array of bytes with variable length, hexadecimal format 0x1234 */ |
327 | #define CSUM_FMT "0x%*phN" | |
328 | #define CSUM_FMT_VALUE(size, bytes) size, bytes | |
329 | ||
0970a22e | 330 | static inline void btrfs_print_data_csum_error(struct btrfs_inode *inode, |
ea41d6b2 | 331 | u64 logical_start, u8 *csum, u8 *csum_expected, int mirror_num) |
6f6b643e | 332 | { |
0970a22e | 333 | struct btrfs_root *root = inode->root; |
223486c2 | 334 | const u32 csum_size = root->fs_info->csum_size; |
6f6b643e QW |
335 | |
336 | /* Output minus objectid, which is more meaningful */ | |
4fd786e6 | 337 | if (root->root_key.objectid >= BTRFS_LAST_FREE_OBJECTID) |
6f6b643e | 338 | btrfs_warn_rl(root->fs_info, |
7ebc7e5f | 339 | "csum failed root %lld ino %lld off %llu csum " CSUM_FMT " expected csum " CSUM_FMT " mirror %d", |
4fd786e6 | 340 | root->root_key.objectid, btrfs_ino(inode), |
7ebc7e5f | 341 | logical_start, |
ea41d6b2 JT |
342 | CSUM_FMT_VALUE(csum_size, csum), |
343 | CSUM_FMT_VALUE(csum_size, csum_expected), | |
7ebc7e5f | 344 | mirror_num); |
6f6b643e QW |
345 | else |
346 | btrfs_warn_rl(root->fs_info, | |
7ebc7e5f | 347 | "csum failed root %llu ino %llu off %llu csum " CSUM_FMT " expected csum " CSUM_FMT " mirror %d", |
4fd786e6 | 348 | root->root_key.objectid, btrfs_ino(inode), |
7ebc7e5f | 349 | logical_start, |
ea41d6b2 JT |
350 | CSUM_FMT_VALUE(csum_size, csum), |
351 | CSUM_FMT_VALUE(csum_size, csum_expected), | |
7ebc7e5f | 352 | mirror_num); |
6f6b643e QW |
353 | } |
354 | ||
2c90e5d6 | 355 | #endif |