Commit | Line | Data |
---|---|---|
6cbd5570 CM |
1 | /* |
2 | * Copyright (C) 2007 Oracle. All rights reserved. | |
3 | * | |
4 | * This program is free software; you can redistribute it and/or | |
5 | * modify it under the terms of the GNU General Public | |
6 | * License v2 as published by the Free Software Foundation. | |
7 | * | |
8 | * This program is distributed in the hope that it will be useful, | |
9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
11 | * General Public License for more details. | |
12 | * | |
13 | * You should have received a copy of the GNU General Public | |
14 | * License along with this program; if not, write to the | |
15 | * Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
16 | * Boston, MA 021110-1307, USA. | |
17 | */ | |
18 | ||
2c90e5d6 CM |
19 | #ifndef __BTRFS_I__ |
20 | #define __BTRFS_I__ | |
21 | ||
778ba82b | 22 | #include <linux/hash.h> |
a52d9a80 | 23 | #include "extent_map.h" |
d1310b2e | 24 | #include "extent_io.h" |
e6dcd2dc | 25 | #include "ordered-data.h" |
16cdcec7 | 26 | #include "delayed-inode.h" |
a52d9a80 | 27 | |
72ac3c0d JB |
28 | /* |
29 | * ordered_data_close is set by truncate when a file that used | |
30 | * to have good data has been truncated to zero. When it is set | |
31 | * the btrfs file release call will add this inode to the | |
32 | * ordered operations list so that we make sure to flush out any | |
33 | * new data the application may have written before commit. | |
34 | */ | |
35 | #define BTRFS_INODE_ORDERED_DATA_CLOSE 0 | |
36 | #define BTRFS_INODE_ORPHAN_META_RESERVED 1 | |
37 | #define BTRFS_INODE_DUMMY 2 | |
38 | #define BTRFS_INODE_IN_DEFRAG 3 | |
69fe2d75 JB |
39 | #define BTRFS_INODE_HAS_ORPHAN_ITEM 4 |
40 | #define BTRFS_INODE_HAS_ASYNC_EXTENT 5 | |
41 | #define BTRFS_INODE_NEEDS_FULL_SYNC 6 | |
42 | #define BTRFS_INODE_COPY_EVERYTHING 7 | |
43 | #define BTRFS_INODE_IN_DELALLOC_LIST 8 | |
44 | #define BTRFS_INODE_READDIO_NEED_LOCK 9 | |
45 | #define BTRFS_INODE_HAS_PROPS 10 | |
72ac3c0d | 46 | |
f1ace244 | 47 | /* in memory btrfs inode */ |
2c90e5d6 | 48 | struct btrfs_inode { |
d352ac68 | 49 | /* which subvolume this inode belongs to */ |
d6e4a428 | 50 | struct btrfs_root *root; |
d352ac68 | 51 | |
d352ac68 CM |
52 | /* key used to find this inode on disk. This is used by the code |
53 | * to read in roots of subvolumes | |
54 | */ | |
d6e4a428 | 55 | struct btrfs_key location; |
d352ac68 | 56 | |
2f2ff0ee FM |
57 | /* |
58 | * Lock for counters and all fields used to determine if the inode is in | |
59 | * the log or not (last_trans, last_sub_trans, last_log_commit, | |
60 | * logged_trans). | |
61 | */ | |
9e0baf60 JB |
62 | spinlock_t lock; |
63 | ||
d352ac68 | 64 | /* the extent_tree has caches of all the extent mappings to disk */ |
a52d9a80 | 65 | struct extent_map_tree extent_tree; |
d352ac68 CM |
66 | |
67 | /* the io_tree does range state (DIRTY, LOCKED etc) */ | |
d1310b2e | 68 | struct extent_io_tree io_tree; |
d352ac68 CM |
69 | |
70 | /* special utility tree used to record which mirrors have already been | |
71 | * tried when checksums fail for a given block | |
72 | */ | |
7e38326f | 73 | struct extent_io_tree io_failure_tree; |
d352ac68 | 74 | |
d352ac68 | 75 | /* held while logging the inode in tree-log.c */ |
e02119d5 | 76 | struct mutex log_mutex; |
d352ac68 | 77 | |
f248679e JB |
78 | /* held while doing delalloc reservations */ |
79 | struct mutex delalloc_mutex; | |
80 | ||
d352ac68 | 81 | /* used to order data wrt metadata */ |
e6dcd2dc | 82 | struct btrfs_ordered_inode_tree ordered_tree; |
15ee9bc7 | 83 | |
d352ac68 CM |
84 | /* list of all the delalloc inodes in the FS. There are times we need |
85 | * to write all the delalloc pages to disk, and this list is used | |
86 | * to walk them all. | |
87 | */ | |
ea8c2819 CM |
88 | struct list_head delalloc_inodes; |
89 | ||
5d4f98a2 YZ |
90 | /* node for the red-black tree that links inodes in subvolume root */ |
91 | struct rb_node rb_node; | |
92 | ||
72ac3c0d JB |
93 | unsigned long runtime_flags; |
94 | ||
9c931c5a | 95 | /* Keep track of who's O_SYNC/fsyncing currently */ |
b812ce28 JB |
96 | atomic_t sync_writers; |
97 | ||
d352ac68 CM |
98 | /* full 64 bit generation number, struct vfs_inode doesn't have a big |
99 | * enough field for this. | |
100 | */ | |
e02119d5 CM |
101 | u64 generation; |
102 | ||
15ee9bc7 JB |
103 | /* |
104 | * transid of the trans_handle that last modified this inode | |
105 | */ | |
106 | u64 last_trans; | |
257c62e1 CM |
107 | |
108 | /* | |
bb14a59b | 109 | * transid that last logged this inode |
257c62e1 | 110 | */ |
bb14a59b | 111 | u64 logged_trans; |
257c62e1 | 112 | |
e02119d5 | 113 | /* |
bb14a59b | 114 | * log transid when this inode was last modified |
e02119d5 | 115 | */ |
bb14a59b MX |
116 | int last_sub_trans; |
117 | ||
118 | /* a local copy of root's last_log_commit */ | |
119 | int last_log_commit; | |
49eb7e46 | 120 | |
d352ac68 CM |
121 | /* total number of bytes pending delalloc, used by stat to calc the |
122 | * real block usage of the file | |
123 | */ | |
9069218d | 124 | u64 delalloc_bytes; |
d352ac68 | 125 | |
a7e3b975 FM |
126 | /* |
127 | * Total number of bytes pending delalloc that fall within a file | |
128 | * range that is either a hole or beyond EOF (and no prealloc extent | |
129 | * exists in the range). This is always <= delalloc_bytes. | |
130 | */ | |
131 | u64 new_delalloc_bytes; | |
132 | ||
47059d93 WS |
133 | /* |
134 | * total number of bytes pending defrag, used by stat to check whether | |
135 | * it needs COW. | |
136 | */ | |
137 | u64 defrag_bytes; | |
138 | ||
d352ac68 CM |
139 | /* |
140 | * the size of the file stored in the metadata on disk. data=ordered | |
141 | * means the in-memory i_size might be larger than the size on disk | |
142 | * because not all the blocks are written yet. | |
143 | */ | |
dbe674a9 | 144 | u64 disk_i_size; |
d352ac68 | 145 | |
aec7477b JB |
146 | /* |
147 | * if this is a directory then index_cnt is the counter for the index | |
148 | * number for new files that are created | |
149 | */ | |
150 | u64 index_cnt; | |
d352ac68 | 151 | |
67de1176 MX |
152 | /* Cache the directory index number to speed the dir/file remove */ |
153 | u64 dir_index; | |
154 | ||
12fcfd22 CM |
155 | /* the fsync log has some corner cases that mean we have to check |
156 | * directories to see if any unlinks have been done before | |
157 | * the directory was logged. See tree-log.c for all the | |
158 | * details | |
159 | */ | |
160 | u64 last_unlink_trans; | |
161 | ||
7709cde3 JB |
162 | /* |
163 | * Number of bytes outstanding that are going to need csums. This is | |
164 | * used in ENOSPC accounting. | |
165 | */ | |
166 | u64 csum_bytes; | |
167 | ||
f1bdcc0a JB |
168 | /* flags field from the on disk inode */ |
169 | u32 flags; | |
170 | ||
9ed74f2d | 171 | /* |
32c00aff JB |
172 | * Counters to keep track of the number of extent item's we may use due |
173 | * to delalloc and such. outstanding_extents is the number of extent | |
174 | * items we think we'll end up using, and reserved_extents is the number | |
175 | * of extent items we've reserved metadata for. | |
9ed74f2d | 176 | */ |
9e0baf60 | 177 | unsigned outstanding_extents; |
69fe2d75 JB |
178 | |
179 | struct btrfs_block_rsv block_rsv; | |
9ed74f2d | 180 | |
1e701a32 | 181 | /* |
b52aa8c9 | 182 | * Cached values of inode properties |
1e701a32 | 183 | */ |
b52aa8c9 | 184 | unsigned prop_compress; /* per-file compression algorithm */ |
eec63c65 DS |
185 | /* |
186 | * Force compression on the file using the defrag ioctl, could be | |
187 | * different from prop_compress and takes precedence if set | |
188 | */ | |
189 | unsigned defrag_compress; | |
1e701a32 | 190 | |
16cdcec7 MX |
191 | struct btrfs_delayed_node *delayed_node; |
192 | ||
9cc97d64 | 193 | /* File creation time. */ |
194 | struct timespec i_otime; | |
195 | ||
8089fe62 DS |
196 | /* Hook into fs_info->delayed_iputs */ |
197 | struct list_head delayed_iput; | |
8089fe62 | 198 | |
5f9a8a51 FM |
199 | /* |
200 | * To avoid races between lockless (i_mutex not held) direct IO writes | |
201 | * and concurrent fsync requests. Direct IO writes must acquire read | |
202 | * access on this semaphore for creating an extent map and its | |
203 | * corresponding ordered extent. The fast fsync path must acquire write | |
204 | * access on this semaphore before it collects ordered extents and | |
205 | * extent maps. | |
206 | */ | |
207 | struct rw_semaphore dio_sem; | |
208 | ||
d352ac68 | 209 | struct inode vfs_inode; |
2c90e5d6 | 210 | }; |
dbe674a9 | 211 | |
16cdcec7 MX |
212 | extern unsigned char btrfs_filetype_table[]; |
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 | { | |
222 | u64 h = objectid ^ (root->objectid * GOLDEN_RATIO_PRIME); | |
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 | ||
8b62f87b JB |
269 | static inline void btrfs_mod_outstanding_extents(struct btrfs_inode *inode, |
270 | int mod) | |
271 | { | |
272 | lockdep_assert_held(&inode->lock); | |
273 | inode->outstanding_extents += mod; | |
274 | if (btrfs_is_free_space_inode(inode)) | |
275 | return; | |
dd48d407 JB |
276 | trace_btrfs_inode_mod_outstanding_extents(inode->root, btrfs_ino(inode), |
277 | mod); | |
8b62f87b JB |
278 | } |
279 | ||
0f8939b8 | 280 | static inline int btrfs_inode_in_log(struct btrfs_inode *inode, u64 generation) |
22ee6985 | 281 | { |
2f2ff0ee FM |
282 | int ret = 0; |
283 | ||
0f8939b8 NB |
284 | spin_lock(&inode->lock); |
285 | if (inode->logged_trans == generation && | |
286 | inode->last_sub_trans <= inode->last_log_commit && | |
287 | inode->last_sub_trans <= inode->root->last_log_commit) { | |
125c4cf9 FM |
288 | /* |
289 | * After a ranged fsync we might have left some extent maps | |
290 | * (that fall outside the fsync's range). So return false | |
291 | * here if the list isn't empty, to make sure btrfs_log_inode() | |
292 | * will be called and process those extent maps. | |
293 | */ | |
294 | smp_mb(); | |
0f8939b8 | 295 | if (list_empty(&inode->extent_tree.modified_extents)) |
2f2ff0ee | 296 | ret = 1; |
125c4cf9 | 297 | } |
0f8939b8 | 298 | spin_unlock(&inode->lock); |
2f2ff0ee | 299 | return ret; |
22ee6985 JB |
300 | } |
301 | ||
c1dc0896 MX |
302 | #define BTRFS_DIO_ORIG_BIO_SUBMITTED 0x1 |
303 | ||
facc8a22 MX |
304 | struct btrfs_dio_private { |
305 | struct inode *inode; | |
c1dc0896 | 306 | unsigned long flags; |
facc8a22 MX |
307 | u64 logical_offset; |
308 | u64 disk_bytenr; | |
309 | u64 bytes; | |
310 | void *private; | |
311 | ||
312 | /* number of bios pending for this dio */ | |
313 | atomic_t pending_bios; | |
314 | ||
315 | /* IO errors */ | |
316 | int errors; | |
317 | ||
318 | /* orig_bio is our btrfs_io_bio */ | |
319 | struct bio *orig_bio; | |
320 | ||
321 | /* dio_bio came from fs/direct-io.c */ | |
322 | struct bio *dio_bio; | |
c1dc0896 MX |
323 | |
324 | /* | |
01327610 | 325 | * The original bio may be split to several sub-bios, this is |
c1dc0896 MX |
326 | * done during endio of sub-bios |
327 | */ | |
4e4cbee9 CH |
328 | blk_status_t (*subio_endio)(struct inode *, struct btrfs_io_bio *, |
329 | blk_status_t); | |
facc8a22 MX |
330 | }; |
331 | ||
2e60a51e MX |
332 | /* |
333 | * Disable DIO read nolock optimization, so new dio readers will be forced | |
334 | * to grab i_mutex. It is used to avoid the endless truncate due to | |
335 | * nonlocked dio read. | |
336 | */ | |
abcefb1e | 337 | static inline void btrfs_inode_block_unlocked_dio(struct btrfs_inode *inode) |
2e60a51e | 338 | { |
abcefb1e | 339 | set_bit(BTRFS_INODE_READDIO_NEED_LOCK, &inode->runtime_flags); |
2e60a51e MX |
340 | smp_mb(); |
341 | } | |
342 | ||
0b581701 | 343 | static inline void btrfs_inode_resume_unlocked_dio(struct btrfs_inode *inode) |
2e60a51e | 344 | { |
4e857c58 | 345 | smp_mb__before_atomic(); |
0b581701 | 346 | clear_bit(BTRFS_INODE_READDIO_NEED_LOCK, &inode->runtime_flags); |
2e60a51e MX |
347 | } |
348 | ||
0970a22e | 349 | static inline void btrfs_print_data_csum_error(struct btrfs_inode *inode, |
6f6b643e QW |
350 | u64 logical_start, u32 csum, u32 csum_expected, int mirror_num) |
351 | { | |
0970a22e | 352 | struct btrfs_root *root = inode->root; |
6f6b643e QW |
353 | |
354 | /* Output minus objectid, which is more meaningful */ | |
355 | if (root->objectid >= BTRFS_LAST_FREE_OBJECTID) | |
356 | btrfs_warn_rl(root->fs_info, | |
357 | "csum failed root %lld ino %lld off %llu csum 0x%08x expected csum 0x%08x mirror %d", | |
0970a22e | 358 | root->objectid, btrfs_ino(inode), |
6f6b643e QW |
359 | logical_start, csum, csum_expected, mirror_num); |
360 | else | |
361 | btrfs_warn_rl(root->fs_info, | |
362 | "csum failed root %llu ino %llu off %llu csum 0x%08x expected csum 0x%08x mirror %d", | |
0970a22e | 363 | root->objectid, btrfs_ino(inode), |
6f6b643e QW |
364 | logical_start, csum, csum_expected, mirror_num); |
365 | } | |
366 | ||
2c90e5d6 | 367 | #endif |