]>
Commit | Line | Data |
---|---|---|
e02119d5 CM |
1 | /* |
2 | * Copyright (C) 2008 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 | ||
19 | #include <linux/sched.h> | |
5a0e3ad6 | 20 | #include <linux/slab.h> |
5dc562c5 | 21 | #include <linux/list_sort.h> |
e02119d5 CM |
22 | #include "ctree.h" |
23 | #include "transaction.h" | |
24 | #include "disk-io.h" | |
25 | #include "locking.h" | |
26 | #include "print-tree.h" | |
f186373f | 27 | #include "backref.h" |
e02119d5 | 28 | #include "compat.h" |
b2950863 | 29 | #include "tree-log.h" |
f186373f | 30 | #include "hash.h" |
e02119d5 CM |
31 | |
32 | /* magic values for the inode_only field in btrfs_log_inode: | |
33 | * | |
34 | * LOG_INODE_ALL means to log everything | |
35 | * LOG_INODE_EXISTS means to log just enough to recreate the inode | |
36 | * during log replay | |
37 | */ | |
38 | #define LOG_INODE_ALL 0 | |
39 | #define LOG_INODE_EXISTS 1 | |
40 | ||
12fcfd22 CM |
41 | /* |
42 | * directory trouble cases | |
43 | * | |
44 | * 1) on rename or unlink, if the inode being unlinked isn't in the fsync | |
45 | * log, we must force a full commit before doing an fsync of the directory | |
46 | * where the unlink was done. | |
47 | * ---> record transid of last unlink/rename per directory | |
48 | * | |
49 | * mkdir foo/some_dir | |
50 | * normal commit | |
51 | * rename foo/some_dir foo2/some_dir | |
52 | * mkdir foo/some_dir | |
53 | * fsync foo/some_dir/some_file | |
54 | * | |
55 | * The fsync above will unlink the original some_dir without recording | |
56 | * it in its new location (foo2). After a crash, some_dir will be gone | |
57 | * unless the fsync of some_file forces a full commit | |
58 | * | |
59 | * 2) we must log any new names for any file or dir that is in the fsync | |
60 | * log. ---> check inode while renaming/linking. | |
61 | * | |
62 | * 2a) we must log any new names for any file or dir during rename | |
63 | * when the directory they are being removed from was logged. | |
64 | * ---> check inode and old parent dir during rename | |
65 | * | |
66 | * 2a is actually the more important variant. With the extra logging | |
67 | * a crash might unlink the old name without recreating the new one | |
68 | * | |
69 | * 3) after a crash, we must go through any directories with a link count | |
70 | * of zero and redo the rm -rf | |
71 | * | |
72 | * mkdir f1/foo | |
73 | * normal commit | |
74 | * rm -rf f1/foo | |
75 | * fsync(f1) | |
76 | * | |
77 | * The directory f1 was fully removed from the FS, but fsync was never | |
78 | * called on f1, only its parent dir. After a crash the rm -rf must | |
79 | * be replayed. This must be able to recurse down the entire | |
80 | * directory tree. The inode link count fixup code takes care of the | |
81 | * ugly details. | |
82 | */ | |
83 | ||
e02119d5 CM |
84 | /* |
85 | * stages for the tree walking. The first | |
86 | * stage (0) is to only pin down the blocks we find | |
87 | * the second stage (1) is to make sure that all the inodes | |
88 | * we find in the log are created in the subvolume. | |
89 | * | |
90 | * The last stage is to deal with directories and links and extents | |
91 | * and all the other fun semantics | |
92 | */ | |
93 | #define LOG_WALK_PIN_ONLY 0 | |
94 | #define LOG_WALK_REPLAY_INODES 1 | |
95 | #define LOG_WALK_REPLAY_ALL 2 | |
96 | ||
12fcfd22 | 97 | static int btrfs_log_inode(struct btrfs_trans_handle *trans, |
e02119d5 CM |
98 | struct btrfs_root *root, struct inode *inode, |
99 | int inode_only); | |
ec051c0f YZ |
100 | static int link_to_fixup_dir(struct btrfs_trans_handle *trans, |
101 | struct btrfs_root *root, | |
102 | struct btrfs_path *path, u64 objectid); | |
12fcfd22 CM |
103 | static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans, |
104 | struct btrfs_root *root, | |
105 | struct btrfs_root *log, | |
106 | struct btrfs_path *path, | |
107 | u64 dirid, int del_all); | |
e02119d5 CM |
108 | |
109 | /* | |
110 | * tree logging is a special write ahead log used to make sure that | |
111 | * fsyncs and O_SYNCs can happen without doing full tree commits. | |
112 | * | |
113 | * Full tree commits are expensive because they require commonly | |
114 | * modified blocks to be recowed, creating many dirty pages in the | |
115 | * extent tree an 4x-6x higher write load than ext3. | |
116 | * | |
117 | * Instead of doing a tree commit on every fsync, we use the | |
118 | * key ranges and transaction ids to find items for a given file or directory | |
119 | * that have changed in this transaction. Those items are copied into | |
120 | * a special tree (one per subvolume root), that tree is written to disk | |
121 | * and then the fsync is considered complete. | |
122 | * | |
123 | * After a crash, items are copied out of the log-tree back into the | |
124 | * subvolume tree. Any file data extents found are recorded in the extent | |
125 | * allocation tree, and the log-tree freed. | |
126 | * | |
127 | * The log tree is read three times, once to pin down all the extents it is | |
128 | * using in ram and once, once to create all the inodes logged in the tree | |
129 | * and once to do all the other items. | |
130 | */ | |
131 | ||
e02119d5 CM |
132 | /* |
133 | * start a sub transaction and setup the log tree | |
134 | * this increments the log tree writer count to make the people | |
135 | * syncing the tree wait for us to finish | |
136 | */ | |
137 | static int start_log_trans(struct btrfs_trans_handle *trans, | |
138 | struct btrfs_root *root) | |
139 | { | |
140 | int ret; | |
4a500fd1 | 141 | int err = 0; |
7237f183 YZ |
142 | |
143 | mutex_lock(&root->log_mutex); | |
144 | if (root->log_root) { | |
ff782e0a JB |
145 | if (!root->log_start_pid) { |
146 | root->log_start_pid = current->pid; | |
147 | root->log_multiple_pids = false; | |
148 | } else if (root->log_start_pid != current->pid) { | |
149 | root->log_multiple_pids = true; | |
150 | } | |
151 | ||
2ecb7923 | 152 | atomic_inc(&root->log_batch); |
7237f183 YZ |
153 | atomic_inc(&root->log_writers); |
154 | mutex_unlock(&root->log_mutex); | |
155 | return 0; | |
156 | } | |
ff782e0a JB |
157 | root->log_multiple_pids = false; |
158 | root->log_start_pid = current->pid; | |
e02119d5 CM |
159 | mutex_lock(&root->fs_info->tree_log_mutex); |
160 | if (!root->fs_info->log_root_tree) { | |
161 | ret = btrfs_init_log_root_tree(trans, root->fs_info); | |
4a500fd1 YZ |
162 | if (ret) |
163 | err = ret; | |
e02119d5 | 164 | } |
4a500fd1 | 165 | if (err == 0 && !root->log_root) { |
e02119d5 | 166 | ret = btrfs_add_log_tree(trans, root); |
4a500fd1 YZ |
167 | if (ret) |
168 | err = ret; | |
e02119d5 | 169 | } |
e02119d5 | 170 | mutex_unlock(&root->fs_info->tree_log_mutex); |
2ecb7923 | 171 | atomic_inc(&root->log_batch); |
7237f183 YZ |
172 | atomic_inc(&root->log_writers); |
173 | mutex_unlock(&root->log_mutex); | |
4a500fd1 | 174 | return err; |
e02119d5 CM |
175 | } |
176 | ||
177 | /* | |
178 | * returns 0 if there was a log transaction running and we were able | |
179 | * to join, or returns -ENOENT if there were not transactions | |
180 | * in progress | |
181 | */ | |
182 | static int join_running_log_trans(struct btrfs_root *root) | |
183 | { | |
184 | int ret = -ENOENT; | |
185 | ||
186 | smp_mb(); | |
187 | if (!root->log_root) | |
188 | return -ENOENT; | |
189 | ||
7237f183 | 190 | mutex_lock(&root->log_mutex); |
e02119d5 CM |
191 | if (root->log_root) { |
192 | ret = 0; | |
7237f183 | 193 | atomic_inc(&root->log_writers); |
e02119d5 | 194 | } |
7237f183 | 195 | mutex_unlock(&root->log_mutex); |
e02119d5 CM |
196 | return ret; |
197 | } | |
198 | ||
12fcfd22 CM |
199 | /* |
200 | * This either makes the current running log transaction wait | |
201 | * until you call btrfs_end_log_trans() or it makes any future | |
202 | * log transactions wait until you call btrfs_end_log_trans() | |
203 | */ | |
204 | int btrfs_pin_log_trans(struct btrfs_root *root) | |
205 | { | |
206 | int ret = -ENOENT; | |
207 | ||
208 | mutex_lock(&root->log_mutex); | |
209 | atomic_inc(&root->log_writers); | |
210 | mutex_unlock(&root->log_mutex); | |
211 | return ret; | |
212 | } | |
213 | ||
e02119d5 CM |
214 | /* |
215 | * indicate we're done making changes to the log tree | |
216 | * and wake up anyone waiting to do a sync | |
217 | */ | |
143bede5 | 218 | void btrfs_end_log_trans(struct btrfs_root *root) |
e02119d5 | 219 | { |
7237f183 YZ |
220 | if (atomic_dec_and_test(&root->log_writers)) { |
221 | smp_mb(); | |
222 | if (waitqueue_active(&root->log_writer_wait)) | |
223 | wake_up(&root->log_writer_wait); | |
224 | } | |
e02119d5 CM |
225 | } |
226 | ||
227 | ||
228 | /* | |
229 | * the walk control struct is used to pass state down the chain when | |
230 | * processing the log tree. The stage field tells us which part | |
231 | * of the log tree processing we are currently doing. The others | |
232 | * are state fields used for that specific part | |
233 | */ | |
234 | struct walk_control { | |
235 | /* should we free the extent on disk when done? This is used | |
236 | * at transaction commit time while freeing a log tree | |
237 | */ | |
238 | int free; | |
239 | ||
240 | /* should we write out the extent buffer? This is used | |
241 | * while flushing the log tree to disk during a sync | |
242 | */ | |
243 | int write; | |
244 | ||
245 | /* should we wait for the extent buffer io to finish? Also used | |
246 | * while flushing the log tree to disk for a sync | |
247 | */ | |
248 | int wait; | |
249 | ||
250 | /* pin only walk, we record which extents on disk belong to the | |
251 | * log trees | |
252 | */ | |
253 | int pin; | |
254 | ||
255 | /* what stage of the replay code we're currently in */ | |
256 | int stage; | |
257 | ||
258 | /* the root we are currently replaying */ | |
259 | struct btrfs_root *replay_dest; | |
260 | ||
261 | /* the trans handle for the current replay */ | |
262 | struct btrfs_trans_handle *trans; | |
263 | ||
264 | /* the function that gets used to process blocks we find in the | |
265 | * tree. Note the extent_buffer might not be up to date when it is | |
266 | * passed in, and it must be checked or read if you need the data | |
267 | * inside it | |
268 | */ | |
269 | int (*process_func)(struct btrfs_root *log, struct extent_buffer *eb, | |
270 | struct walk_control *wc, u64 gen); | |
271 | }; | |
272 | ||
273 | /* | |
274 | * process_func used to pin down extents, write them or wait on them | |
275 | */ | |
276 | static int process_one_buffer(struct btrfs_root *log, | |
277 | struct extent_buffer *eb, | |
278 | struct walk_control *wc, u64 gen) | |
279 | { | |
b50c6e25 JB |
280 | int ret = 0; |
281 | ||
04018de5 | 282 | if (wc->pin) |
b50c6e25 JB |
283 | ret = btrfs_pin_extent_for_log_replay(log->fs_info->extent_root, |
284 | eb->start, eb->len); | |
e02119d5 | 285 | |
b50c6e25 | 286 | if (!ret && btrfs_buffer_uptodate(eb, gen, 0)) { |
e02119d5 CM |
287 | if (wc->write) |
288 | btrfs_write_tree_block(eb); | |
289 | if (wc->wait) | |
290 | btrfs_wait_tree_block_writeback(eb); | |
291 | } | |
b50c6e25 | 292 | return ret; |
e02119d5 CM |
293 | } |
294 | ||
295 | /* | |
296 | * Item overwrite used by replay and tree logging. eb, slot and key all refer | |
297 | * to the src data we are copying out. | |
298 | * | |
299 | * root is the tree we are copying into, and path is a scratch | |
300 | * path for use in this function (it should be released on entry and | |
301 | * will be released on exit). | |
302 | * | |
303 | * If the key is already in the destination tree the existing item is | |
304 | * overwritten. If the existing item isn't big enough, it is extended. | |
305 | * If it is too large, it is truncated. | |
306 | * | |
307 | * If the key isn't in the destination yet, a new item is inserted. | |
308 | */ | |
309 | static noinline int overwrite_item(struct btrfs_trans_handle *trans, | |
310 | struct btrfs_root *root, | |
311 | struct btrfs_path *path, | |
312 | struct extent_buffer *eb, int slot, | |
313 | struct btrfs_key *key) | |
314 | { | |
315 | int ret; | |
316 | u32 item_size; | |
317 | u64 saved_i_size = 0; | |
318 | int save_old_i_size = 0; | |
319 | unsigned long src_ptr; | |
320 | unsigned long dst_ptr; | |
321 | int overwrite_root = 0; | |
4bc4bee4 | 322 | bool inode_item = key->type == BTRFS_INODE_ITEM_KEY; |
e02119d5 CM |
323 | |
324 | if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) | |
325 | overwrite_root = 1; | |
326 | ||
327 | item_size = btrfs_item_size_nr(eb, slot); | |
328 | src_ptr = btrfs_item_ptr_offset(eb, slot); | |
329 | ||
330 | /* look for the key in the destination tree */ | |
331 | ret = btrfs_search_slot(NULL, root, key, path, 0, 0); | |
4bc4bee4 JB |
332 | if (ret < 0) |
333 | return ret; | |
334 | ||
e02119d5 CM |
335 | if (ret == 0) { |
336 | char *src_copy; | |
337 | char *dst_copy; | |
338 | u32 dst_size = btrfs_item_size_nr(path->nodes[0], | |
339 | path->slots[0]); | |
340 | if (dst_size != item_size) | |
341 | goto insert; | |
342 | ||
343 | if (item_size == 0) { | |
b3b4aa74 | 344 | btrfs_release_path(path); |
e02119d5 CM |
345 | return 0; |
346 | } | |
347 | dst_copy = kmalloc(item_size, GFP_NOFS); | |
348 | src_copy = kmalloc(item_size, GFP_NOFS); | |
2a29edc6 | 349 | if (!dst_copy || !src_copy) { |
b3b4aa74 | 350 | btrfs_release_path(path); |
2a29edc6 | 351 | kfree(dst_copy); |
352 | kfree(src_copy); | |
353 | return -ENOMEM; | |
354 | } | |
e02119d5 CM |
355 | |
356 | read_extent_buffer(eb, src_copy, src_ptr, item_size); | |
357 | ||
358 | dst_ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]); | |
359 | read_extent_buffer(path->nodes[0], dst_copy, dst_ptr, | |
360 | item_size); | |
361 | ret = memcmp(dst_copy, src_copy, item_size); | |
362 | ||
363 | kfree(dst_copy); | |
364 | kfree(src_copy); | |
365 | /* | |
366 | * they have the same contents, just return, this saves | |
367 | * us from cowing blocks in the destination tree and doing | |
368 | * extra writes that may not have been done by a previous | |
369 | * sync | |
370 | */ | |
371 | if (ret == 0) { | |
b3b4aa74 | 372 | btrfs_release_path(path); |
e02119d5 CM |
373 | return 0; |
374 | } | |
375 | ||
4bc4bee4 JB |
376 | /* |
377 | * We need to load the old nbytes into the inode so when we | |
378 | * replay the extents we've logged we get the right nbytes. | |
379 | */ | |
380 | if (inode_item) { | |
381 | struct btrfs_inode_item *item; | |
382 | u64 nbytes; | |
383 | ||
384 | item = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
385 | struct btrfs_inode_item); | |
386 | nbytes = btrfs_inode_nbytes(path->nodes[0], item); | |
387 | item = btrfs_item_ptr(eb, slot, | |
388 | struct btrfs_inode_item); | |
389 | btrfs_set_inode_nbytes(eb, item, nbytes); | |
390 | } | |
391 | } else if (inode_item) { | |
392 | struct btrfs_inode_item *item; | |
393 | ||
394 | /* | |
395 | * New inode, set nbytes to 0 so that the nbytes comes out | |
396 | * properly when we replay the extents. | |
397 | */ | |
398 | item = btrfs_item_ptr(eb, slot, struct btrfs_inode_item); | |
399 | btrfs_set_inode_nbytes(eb, item, 0); | |
e02119d5 CM |
400 | } |
401 | insert: | |
b3b4aa74 | 402 | btrfs_release_path(path); |
e02119d5 CM |
403 | /* try to insert the key into the destination tree */ |
404 | ret = btrfs_insert_empty_item(trans, root, path, | |
405 | key, item_size); | |
406 | ||
407 | /* make sure any existing item is the correct size */ | |
408 | if (ret == -EEXIST) { | |
409 | u32 found_size; | |
410 | found_size = btrfs_item_size_nr(path->nodes[0], | |
411 | path->slots[0]); | |
143bede5 | 412 | if (found_size > item_size) |
afe5fea7 | 413 | btrfs_truncate_item(root, path, item_size, 1); |
143bede5 | 414 | else if (found_size < item_size) |
4b90c680 | 415 | btrfs_extend_item(root, path, |
143bede5 | 416 | item_size - found_size); |
e02119d5 | 417 | } else if (ret) { |
4a500fd1 | 418 | return ret; |
e02119d5 CM |
419 | } |
420 | dst_ptr = btrfs_item_ptr_offset(path->nodes[0], | |
421 | path->slots[0]); | |
422 | ||
423 | /* don't overwrite an existing inode if the generation number | |
424 | * was logged as zero. This is done when the tree logging code | |
425 | * is just logging an inode to make sure it exists after recovery. | |
426 | * | |
427 | * Also, don't overwrite i_size on directories during replay. | |
428 | * log replay inserts and removes directory items based on the | |
429 | * state of the tree found in the subvolume, and i_size is modified | |
430 | * as it goes | |
431 | */ | |
432 | if (key->type == BTRFS_INODE_ITEM_KEY && ret == -EEXIST) { | |
433 | struct btrfs_inode_item *src_item; | |
434 | struct btrfs_inode_item *dst_item; | |
435 | ||
436 | src_item = (struct btrfs_inode_item *)src_ptr; | |
437 | dst_item = (struct btrfs_inode_item *)dst_ptr; | |
438 | ||
439 | if (btrfs_inode_generation(eb, src_item) == 0) | |
440 | goto no_copy; | |
441 | ||
442 | if (overwrite_root && | |
443 | S_ISDIR(btrfs_inode_mode(eb, src_item)) && | |
444 | S_ISDIR(btrfs_inode_mode(path->nodes[0], dst_item))) { | |
445 | save_old_i_size = 1; | |
446 | saved_i_size = btrfs_inode_size(path->nodes[0], | |
447 | dst_item); | |
448 | } | |
449 | } | |
450 | ||
451 | copy_extent_buffer(path->nodes[0], eb, dst_ptr, | |
452 | src_ptr, item_size); | |
453 | ||
454 | if (save_old_i_size) { | |
455 | struct btrfs_inode_item *dst_item; | |
456 | dst_item = (struct btrfs_inode_item *)dst_ptr; | |
457 | btrfs_set_inode_size(path->nodes[0], dst_item, saved_i_size); | |
458 | } | |
459 | ||
460 | /* make sure the generation is filled in */ | |
461 | if (key->type == BTRFS_INODE_ITEM_KEY) { | |
462 | struct btrfs_inode_item *dst_item; | |
463 | dst_item = (struct btrfs_inode_item *)dst_ptr; | |
464 | if (btrfs_inode_generation(path->nodes[0], dst_item) == 0) { | |
465 | btrfs_set_inode_generation(path->nodes[0], dst_item, | |
466 | trans->transid); | |
467 | } | |
468 | } | |
469 | no_copy: | |
470 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
b3b4aa74 | 471 | btrfs_release_path(path); |
e02119d5 CM |
472 | return 0; |
473 | } | |
474 | ||
475 | /* | |
476 | * simple helper to read an inode off the disk from a given root | |
477 | * This can only be called for subvolume roots and not for the log | |
478 | */ | |
479 | static noinline struct inode *read_one_inode(struct btrfs_root *root, | |
480 | u64 objectid) | |
481 | { | |
5d4f98a2 | 482 | struct btrfs_key key; |
e02119d5 | 483 | struct inode *inode; |
e02119d5 | 484 | |
5d4f98a2 YZ |
485 | key.objectid = objectid; |
486 | key.type = BTRFS_INODE_ITEM_KEY; | |
487 | key.offset = 0; | |
73f73415 | 488 | inode = btrfs_iget(root->fs_info->sb, &key, root, NULL); |
5d4f98a2 YZ |
489 | if (IS_ERR(inode)) { |
490 | inode = NULL; | |
491 | } else if (is_bad_inode(inode)) { | |
e02119d5 CM |
492 | iput(inode); |
493 | inode = NULL; | |
494 | } | |
495 | return inode; | |
496 | } | |
497 | ||
498 | /* replays a single extent in 'eb' at 'slot' with 'key' into the | |
499 | * subvolume 'root'. path is released on entry and should be released | |
500 | * on exit. | |
501 | * | |
502 | * extents in the log tree have not been allocated out of the extent | |
503 | * tree yet. So, this completes the allocation, taking a reference | |
504 | * as required if the extent already exists or creating a new extent | |
505 | * if it isn't in the extent allocation tree yet. | |
506 | * | |
507 | * The extent is inserted into the file, dropping any existing extents | |
508 | * from the file that overlap the new one. | |
509 | */ | |
510 | static noinline int replay_one_extent(struct btrfs_trans_handle *trans, | |
511 | struct btrfs_root *root, | |
512 | struct btrfs_path *path, | |
513 | struct extent_buffer *eb, int slot, | |
514 | struct btrfs_key *key) | |
515 | { | |
516 | int found_type; | |
e02119d5 | 517 | u64 extent_end; |
e02119d5 | 518 | u64 start = key->offset; |
4bc4bee4 | 519 | u64 nbytes = 0; |
e02119d5 CM |
520 | struct btrfs_file_extent_item *item; |
521 | struct inode *inode = NULL; | |
522 | unsigned long size; | |
523 | int ret = 0; | |
524 | ||
525 | item = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item); | |
526 | found_type = btrfs_file_extent_type(eb, item); | |
527 | ||
d899e052 | 528 | if (found_type == BTRFS_FILE_EXTENT_REG || |
4bc4bee4 JB |
529 | found_type == BTRFS_FILE_EXTENT_PREALLOC) { |
530 | nbytes = btrfs_file_extent_num_bytes(eb, item); | |
531 | extent_end = start + nbytes; | |
532 | ||
533 | /* | |
534 | * We don't add to the inodes nbytes if we are prealloc or a | |
535 | * hole. | |
536 | */ | |
537 | if (btrfs_file_extent_disk_bytenr(eb, item) == 0) | |
538 | nbytes = 0; | |
539 | } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { | |
c8b97818 | 540 | size = btrfs_file_extent_inline_len(eb, item); |
4bc4bee4 | 541 | nbytes = btrfs_file_extent_ram_bytes(eb, item); |
fda2832f | 542 | extent_end = ALIGN(start + size, root->sectorsize); |
e02119d5 CM |
543 | } else { |
544 | ret = 0; | |
545 | goto out; | |
546 | } | |
547 | ||
548 | inode = read_one_inode(root, key->objectid); | |
549 | if (!inode) { | |
550 | ret = -EIO; | |
551 | goto out; | |
552 | } | |
553 | ||
554 | /* | |
555 | * first check to see if we already have this extent in the | |
556 | * file. This must be done before the btrfs_drop_extents run | |
557 | * so we don't try to drop this extent. | |
558 | */ | |
33345d01 | 559 | ret = btrfs_lookup_file_extent(trans, root, path, btrfs_ino(inode), |
e02119d5 CM |
560 | start, 0); |
561 | ||
d899e052 YZ |
562 | if (ret == 0 && |
563 | (found_type == BTRFS_FILE_EXTENT_REG || | |
564 | found_type == BTRFS_FILE_EXTENT_PREALLOC)) { | |
e02119d5 CM |
565 | struct btrfs_file_extent_item cmp1; |
566 | struct btrfs_file_extent_item cmp2; | |
567 | struct btrfs_file_extent_item *existing; | |
568 | struct extent_buffer *leaf; | |
569 | ||
570 | leaf = path->nodes[0]; | |
571 | existing = btrfs_item_ptr(leaf, path->slots[0], | |
572 | struct btrfs_file_extent_item); | |
573 | ||
574 | read_extent_buffer(eb, &cmp1, (unsigned long)item, | |
575 | sizeof(cmp1)); | |
576 | read_extent_buffer(leaf, &cmp2, (unsigned long)existing, | |
577 | sizeof(cmp2)); | |
578 | ||
579 | /* | |
580 | * we already have a pointer to this exact extent, | |
581 | * we don't have to do anything | |
582 | */ | |
583 | if (memcmp(&cmp1, &cmp2, sizeof(cmp1)) == 0) { | |
b3b4aa74 | 584 | btrfs_release_path(path); |
e02119d5 CM |
585 | goto out; |
586 | } | |
587 | } | |
b3b4aa74 | 588 | btrfs_release_path(path); |
e02119d5 CM |
589 | |
590 | /* drop any overlapping extents */ | |
2671485d | 591 | ret = btrfs_drop_extents(trans, root, inode, start, extent_end, 1); |
3650860b JB |
592 | if (ret) |
593 | goto out; | |
e02119d5 | 594 | |
07d400a6 YZ |
595 | if (found_type == BTRFS_FILE_EXTENT_REG || |
596 | found_type == BTRFS_FILE_EXTENT_PREALLOC) { | |
5d4f98a2 | 597 | u64 offset; |
07d400a6 YZ |
598 | unsigned long dest_offset; |
599 | struct btrfs_key ins; | |
600 | ||
601 | ret = btrfs_insert_empty_item(trans, root, path, key, | |
602 | sizeof(*item)); | |
3650860b JB |
603 | if (ret) |
604 | goto out; | |
07d400a6 YZ |
605 | dest_offset = btrfs_item_ptr_offset(path->nodes[0], |
606 | path->slots[0]); | |
607 | copy_extent_buffer(path->nodes[0], eb, dest_offset, | |
608 | (unsigned long)item, sizeof(*item)); | |
609 | ||
610 | ins.objectid = btrfs_file_extent_disk_bytenr(eb, item); | |
611 | ins.offset = btrfs_file_extent_disk_num_bytes(eb, item); | |
612 | ins.type = BTRFS_EXTENT_ITEM_KEY; | |
5d4f98a2 | 613 | offset = key->offset - btrfs_file_extent_offset(eb, item); |
07d400a6 YZ |
614 | |
615 | if (ins.objectid > 0) { | |
616 | u64 csum_start; | |
617 | u64 csum_end; | |
618 | LIST_HEAD(ordered_sums); | |
619 | /* | |
620 | * is this extent already allocated in the extent | |
621 | * allocation tree? If so, just add a reference | |
622 | */ | |
623 | ret = btrfs_lookup_extent(root, ins.objectid, | |
624 | ins.offset); | |
625 | if (ret == 0) { | |
626 | ret = btrfs_inc_extent_ref(trans, root, | |
627 | ins.objectid, ins.offset, | |
5d4f98a2 | 628 | 0, root->root_key.objectid, |
66d7e7f0 | 629 | key->objectid, offset, 0); |
b50c6e25 JB |
630 | if (ret) |
631 | goto out; | |
07d400a6 YZ |
632 | } else { |
633 | /* | |
634 | * insert the extent pointer in the extent | |
635 | * allocation tree | |
636 | */ | |
5d4f98a2 YZ |
637 | ret = btrfs_alloc_logged_file_extent(trans, |
638 | root, root->root_key.objectid, | |
639 | key->objectid, offset, &ins); | |
b50c6e25 JB |
640 | if (ret) |
641 | goto out; | |
07d400a6 | 642 | } |
b3b4aa74 | 643 | btrfs_release_path(path); |
07d400a6 YZ |
644 | |
645 | if (btrfs_file_extent_compression(eb, item)) { | |
646 | csum_start = ins.objectid; | |
647 | csum_end = csum_start + ins.offset; | |
648 | } else { | |
649 | csum_start = ins.objectid + | |
650 | btrfs_file_extent_offset(eb, item); | |
651 | csum_end = csum_start + | |
652 | btrfs_file_extent_num_bytes(eb, item); | |
653 | } | |
654 | ||
655 | ret = btrfs_lookup_csums_range(root->log_root, | |
656 | csum_start, csum_end - 1, | |
a2de733c | 657 | &ordered_sums, 0); |
3650860b JB |
658 | if (ret) |
659 | goto out; | |
07d400a6 YZ |
660 | while (!list_empty(&ordered_sums)) { |
661 | struct btrfs_ordered_sum *sums; | |
662 | sums = list_entry(ordered_sums.next, | |
663 | struct btrfs_ordered_sum, | |
664 | list); | |
3650860b JB |
665 | if (!ret) |
666 | ret = btrfs_csum_file_blocks(trans, | |
07d400a6 YZ |
667 | root->fs_info->csum_root, |
668 | sums); | |
07d400a6 YZ |
669 | list_del(&sums->list); |
670 | kfree(sums); | |
671 | } | |
3650860b JB |
672 | if (ret) |
673 | goto out; | |
07d400a6 | 674 | } else { |
b3b4aa74 | 675 | btrfs_release_path(path); |
07d400a6 YZ |
676 | } |
677 | } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { | |
678 | /* inline extents are easy, we just overwrite them */ | |
679 | ret = overwrite_item(trans, root, path, eb, slot, key); | |
3650860b JB |
680 | if (ret) |
681 | goto out; | |
07d400a6 | 682 | } |
e02119d5 | 683 | |
4bc4bee4 | 684 | inode_add_bytes(inode, nbytes); |
b9959295 | 685 | ret = btrfs_update_inode(trans, root, inode); |
e02119d5 CM |
686 | out: |
687 | if (inode) | |
688 | iput(inode); | |
689 | return ret; | |
690 | } | |
691 | ||
692 | /* | |
693 | * when cleaning up conflicts between the directory names in the | |
694 | * subvolume, directory names in the log and directory names in the | |
695 | * inode back references, we may have to unlink inodes from directories. | |
696 | * | |
697 | * This is a helper function to do the unlink of a specific directory | |
698 | * item | |
699 | */ | |
700 | static noinline int drop_one_dir_item(struct btrfs_trans_handle *trans, | |
701 | struct btrfs_root *root, | |
702 | struct btrfs_path *path, | |
703 | struct inode *dir, | |
704 | struct btrfs_dir_item *di) | |
705 | { | |
706 | struct inode *inode; | |
707 | char *name; | |
708 | int name_len; | |
709 | struct extent_buffer *leaf; | |
710 | struct btrfs_key location; | |
711 | int ret; | |
712 | ||
713 | leaf = path->nodes[0]; | |
714 | ||
715 | btrfs_dir_item_key_to_cpu(leaf, di, &location); | |
716 | name_len = btrfs_dir_name_len(leaf, di); | |
717 | name = kmalloc(name_len, GFP_NOFS); | |
2a29edc6 | 718 | if (!name) |
719 | return -ENOMEM; | |
720 | ||
e02119d5 | 721 | read_extent_buffer(leaf, name, (unsigned long)(di + 1), name_len); |
b3b4aa74 | 722 | btrfs_release_path(path); |
e02119d5 CM |
723 | |
724 | inode = read_one_inode(root, location.objectid); | |
c00e9493 | 725 | if (!inode) { |
3650860b JB |
726 | ret = -EIO; |
727 | goto out; | |
c00e9493 | 728 | } |
e02119d5 | 729 | |
ec051c0f | 730 | ret = link_to_fixup_dir(trans, root, path, location.objectid); |
3650860b JB |
731 | if (ret) |
732 | goto out; | |
12fcfd22 | 733 | |
e02119d5 | 734 | ret = btrfs_unlink_inode(trans, root, dir, inode, name, name_len); |
3650860b JB |
735 | if (ret) |
736 | goto out; | |
737 | btrfs_run_delayed_items(trans, root); | |
738 | out: | |
e02119d5 | 739 | kfree(name); |
e02119d5 CM |
740 | iput(inode); |
741 | return ret; | |
742 | } | |
743 | ||
744 | /* | |
745 | * helper function to see if a given name and sequence number found | |
746 | * in an inode back reference are already in a directory and correctly | |
747 | * point to this inode | |
748 | */ | |
749 | static noinline int inode_in_dir(struct btrfs_root *root, | |
750 | struct btrfs_path *path, | |
751 | u64 dirid, u64 objectid, u64 index, | |
752 | const char *name, int name_len) | |
753 | { | |
754 | struct btrfs_dir_item *di; | |
755 | struct btrfs_key location; | |
756 | int match = 0; | |
757 | ||
758 | di = btrfs_lookup_dir_index_item(NULL, root, path, dirid, | |
759 | index, name, name_len, 0); | |
760 | if (di && !IS_ERR(di)) { | |
761 | btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); | |
762 | if (location.objectid != objectid) | |
763 | goto out; | |
764 | } else | |
765 | goto out; | |
b3b4aa74 | 766 | btrfs_release_path(path); |
e02119d5 CM |
767 | |
768 | di = btrfs_lookup_dir_item(NULL, root, path, dirid, name, name_len, 0); | |
769 | if (di && !IS_ERR(di)) { | |
770 | btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); | |
771 | if (location.objectid != objectid) | |
772 | goto out; | |
773 | } else | |
774 | goto out; | |
775 | match = 1; | |
776 | out: | |
b3b4aa74 | 777 | btrfs_release_path(path); |
e02119d5 CM |
778 | return match; |
779 | } | |
780 | ||
781 | /* | |
782 | * helper function to check a log tree for a named back reference in | |
783 | * an inode. This is used to decide if a back reference that is | |
784 | * found in the subvolume conflicts with what we find in the log. | |
785 | * | |
786 | * inode backreferences may have multiple refs in a single item, | |
787 | * during replay we process one reference at a time, and we don't | |
788 | * want to delete valid links to a file from the subvolume if that | |
789 | * link is also in the log. | |
790 | */ | |
791 | static noinline int backref_in_log(struct btrfs_root *log, | |
792 | struct btrfs_key *key, | |
f186373f | 793 | u64 ref_objectid, |
e02119d5 CM |
794 | char *name, int namelen) |
795 | { | |
796 | struct btrfs_path *path; | |
797 | struct btrfs_inode_ref *ref; | |
798 | unsigned long ptr; | |
799 | unsigned long ptr_end; | |
800 | unsigned long name_ptr; | |
801 | int found_name_len; | |
802 | int item_size; | |
803 | int ret; | |
804 | int match = 0; | |
805 | ||
806 | path = btrfs_alloc_path(); | |
2a29edc6 | 807 | if (!path) |
808 | return -ENOMEM; | |
809 | ||
e02119d5 CM |
810 | ret = btrfs_search_slot(NULL, log, key, path, 0, 0); |
811 | if (ret != 0) | |
812 | goto out; | |
813 | ||
e02119d5 | 814 | ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]); |
f186373f MF |
815 | |
816 | if (key->type == BTRFS_INODE_EXTREF_KEY) { | |
817 | if (btrfs_find_name_in_ext_backref(path, ref_objectid, | |
818 | name, namelen, NULL)) | |
819 | match = 1; | |
820 | ||
821 | goto out; | |
822 | } | |
823 | ||
824 | item_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]); | |
e02119d5 CM |
825 | ptr_end = ptr + item_size; |
826 | while (ptr < ptr_end) { | |
827 | ref = (struct btrfs_inode_ref *)ptr; | |
828 | found_name_len = btrfs_inode_ref_name_len(path->nodes[0], ref); | |
829 | if (found_name_len == namelen) { | |
830 | name_ptr = (unsigned long)(ref + 1); | |
831 | ret = memcmp_extent_buffer(path->nodes[0], name, | |
832 | name_ptr, namelen); | |
833 | if (ret == 0) { | |
834 | match = 1; | |
835 | goto out; | |
836 | } | |
837 | } | |
838 | ptr = (unsigned long)(ref + 1) + found_name_len; | |
839 | } | |
840 | out: | |
841 | btrfs_free_path(path); | |
842 | return match; | |
843 | } | |
844 | ||
5a1d7843 | 845 | static inline int __add_inode_ref(struct btrfs_trans_handle *trans, |
e02119d5 | 846 | struct btrfs_root *root, |
e02119d5 | 847 | struct btrfs_path *path, |
5a1d7843 JS |
848 | struct btrfs_root *log_root, |
849 | struct inode *dir, struct inode *inode, | |
5a1d7843 | 850 | struct extent_buffer *eb, |
f186373f MF |
851 | u64 inode_objectid, u64 parent_objectid, |
852 | u64 ref_index, char *name, int namelen, | |
853 | int *search_done) | |
e02119d5 | 854 | { |
34f3e4f2 | 855 | int ret; |
f186373f MF |
856 | char *victim_name; |
857 | int victim_name_len; | |
858 | struct extent_buffer *leaf; | |
5a1d7843 | 859 | struct btrfs_dir_item *di; |
f186373f MF |
860 | struct btrfs_key search_key; |
861 | struct btrfs_inode_extref *extref; | |
c622ae60 | 862 | |
f186373f MF |
863 | again: |
864 | /* Search old style refs */ | |
865 | search_key.objectid = inode_objectid; | |
866 | search_key.type = BTRFS_INODE_REF_KEY; | |
867 | search_key.offset = parent_objectid; | |
868 | ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0); | |
e02119d5 | 869 | if (ret == 0) { |
e02119d5 CM |
870 | struct btrfs_inode_ref *victim_ref; |
871 | unsigned long ptr; | |
872 | unsigned long ptr_end; | |
f186373f MF |
873 | |
874 | leaf = path->nodes[0]; | |
e02119d5 CM |
875 | |
876 | /* are we trying to overwrite a back ref for the root directory | |
877 | * if so, just jump out, we're done | |
878 | */ | |
f186373f | 879 | if (search_key.objectid == search_key.offset) |
5a1d7843 | 880 | return 1; |
e02119d5 CM |
881 | |
882 | /* check all the names in this back reference to see | |
883 | * if they are in the log. if so, we allow them to stay | |
884 | * otherwise they must be unlinked as a conflict | |
885 | */ | |
886 | ptr = btrfs_item_ptr_offset(leaf, path->slots[0]); | |
887 | ptr_end = ptr + btrfs_item_size_nr(leaf, path->slots[0]); | |
d397712b | 888 | while (ptr < ptr_end) { |
e02119d5 CM |
889 | victim_ref = (struct btrfs_inode_ref *)ptr; |
890 | victim_name_len = btrfs_inode_ref_name_len(leaf, | |
891 | victim_ref); | |
892 | victim_name = kmalloc(victim_name_len, GFP_NOFS); | |
3650860b JB |
893 | if (!victim_name) |
894 | return -ENOMEM; | |
e02119d5 CM |
895 | |
896 | read_extent_buffer(leaf, victim_name, | |
897 | (unsigned long)(victim_ref + 1), | |
898 | victim_name_len); | |
899 | ||
f186373f MF |
900 | if (!backref_in_log(log_root, &search_key, |
901 | parent_objectid, | |
902 | victim_name, | |
e02119d5 CM |
903 | victim_name_len)) { |
904 | btrfs_inc_nlink(inode); | |
b3b4aa74 | 905 | btrfs_release_path(path); |
12fcfd22 | 906 | |
e02119d5 CM |
907 | ret = btrfs_unlink_inode(trans, root, dir, |
908 | inode, victim_name, | |
909 | victim_name_len); | |
f186373f | 910 | kfree(victim_name); |
3650860b JB |
911 | if (ret) |
912 | return ret; | |
913 | btrfs_run_delayed_items(trans, root); | |
f186373f MF |
914 | *search_done = 1; |
915 | goto again; | |
e02119d5 CM |
916 | } |
917 | kfree(victim_name); | |
f186373f | 918 | |
e02119d5 CM |
919 | ptr = (unsigned long)(victim_ref + 1) + victim_name_len; |
920 | } | |
e02119d5 | 921 | |
c622ae60 | 922 | /* |
923 | * NOTE: we have searched root tree and checked the | |
924 | * coresponding ref, it does not need to check again. | |
925 | */ | |
5a1d7843 | 926 | *search_done = 1; |
e02119d5 | 927 | } |
b3b4aa74 | 928 | btrfs_release_path(path); |
e02119d5 | 929 | |
f186373f MF |
930 | /* Same search but for extended refs */ |
931 | extref = btrfs_lookup_inode_extref(NULL, root, path, name, namelen, | |
932 | inode_objectid, parent_objectid, 0, | |
933 | 0); | |
934 | if (!IS_ERR_OR_NULL(extref)) { | |
935 | u32 item_size; | |
936 | u32 cur_offset = 0; | |
937 | unsigned long base; | |
938 | struct inode *victim_parent; | |
939 | ||
940 | leaf = path->nodes[0]; | |
941 | ||
942 | item_size = btrfs_item_size_nr(leaf, path->slots[0]); | |
943 | base = btrfs_item_ptr_offset(leaf, path->slots[0]); | |
944 | ||
945 | while (cur_offset < item_size) { | |
946 | extref = (struct btrfs_inode_extref *)base + cur_offset; | |
947 | ||
948 | victim_name_len = btrfs_inode_extref_name_len(leaf, extref); | |
949 | ||
950 | if (btrfs_inode_extref_parent(leaf, extref) != parent_objectid) | |
951 | goto next; | |
952 | ||
953 | victim_name = kmalloc(victim_name_len, GFP_NOFS); | |
3650860b JB |
954 | if (!victim_name) |
955 | return -ENOMEM; | |
f186373f MF |
956 | read_extent_buffer(leaf, victim_name, (unsigned long)&extref->name, |
957 | victim_name_len); | |
958 | ||
959 | search_key.objectid = inode_objectid; | |
960 | search_key.type = BTRFS_INODE_EXTREF_KEY; | |
961 | search_key.offset = btrfs_extref_hash(parent_objectid, | |
962 | victim_name, | |
963 | victim_name_len); | |
964 | ret = 0; | |
965 | if (!backref_in_log(log_root, &search_key, | |
966 | parent_objectid, victim_name, | |
967 | victim_name_len)) { | |
968 | ret = -ENOENT; | |
969 | victim_parent = read_one_inode(root, | |
970 | parent_objectid); | |
971 | if (victim_parent) { | |
972 | btrfs_inc_nlink(inode); | |
973 | btrfs_release_path(path); | |
974 | ||
975 | ret = btrfs_unlink_inode(trans, root, | |
976 | victim_parent, | |
977 | inode, | |
978 | victim_name, | |
979 | victim_name_len); | |
980 | btrfs_run_delayed_items(trans, root); | |
981 | } | |
f186373f MF |
982 | iput(victim_parent); |
983 | kfree(victim_name); | |
3650860b JB |
984 | if (ret) |
985 | return ret; | |
f186373f MF |
986 | *search_done = 1; |
987 | goto again; | |
988 | } | |
989 | kfree(victim_name); | |
3650860b JB |
990 | if (ret) |
991 | return ret; | |
f186373f MF |
992 | next: |
993 | cur_offset += victim_name_len + sizeof(*extref); | |
994 | } | |
995 | *search_done = 1; | |
996 | } | |
997 | btrfs_release_path(path); | |
998 | ||
34f3e4f2 | 999 | /* look for a conflicting sequence number */ |
1000 | di = btrfs_lookup_dir_index_item(trans, root, path, btrfs_ino(dir), | |
f186373f | 1001 | ref_index, name, namelen, 0); |
34f3e4f2 | 1002 | if (di && !IS_ERR(di)) { |
1003 | ret = drop_one_dir_item(trans, root, path, dir, di); | |
3650860b JB |
1004 | if (ret) |
1005 | return ret; | |
34f3e4f2 | 1006 | } |
1007 | btrfs_release_path(path); | |
1008 | ||
1009 | /* look for a conflicing name */ | |
1010 | di = btrfs_lookup_dir_item(trans, root, path, btrfs_ino(dir), | |
1011 | name, namelen, 0); | |
1012 | if (di && !IS_ERR(di)) { | |
1013 | ret = drop_one_dir_item(trans, root, path, dir, di); | |
3650860b JB |
1014 | if (ret) |
1015 | return ret; | |
34f3e4f2 | 1016 | } |
1017 | btrfs_release_path(path); | |
1018 | ||
5a1d7843 JS |
1019 | return 0; |
1020 | } | |
e02119d5 | 1021 | |
f186373f MF |
1022 | static int extref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr, |
1023 | u32 *namelen, char **name, u64 *index, | |
1024 | u64 *parent_objectid) | |
1025 | { | |
1026 | struct btrfs_inode_extref *extref; | |
1027 | ||
1028 | extref = (struct btrfs_inode_extref *)ref_ptr; | |
1029 | ||
1030 | *namelen = btrfs_inode_extref_name_len(eb, extref); | |
1031 | *name = kmalloc(*namelen, GFP_NOFS); | |
1032 | if (*name == NULL) | |
1033 | return -ENOMEM; | |
1034 | ||
1035 | read_extent_buffer(eb, *name, (unsigned long)&extref->name, | |
1036 | *namelen); | |
1037 | ||
1038 | *index = btrfs_inode_extref_index(eb, extref); | |
1039 | if (parent_objectid) | |
1040 | *parent_objectid = btrfs_inode_extref_parent(eb, extref); | |
1041 | ||
1042 | return 0; | |
1043 | } | |
1044 | ||
1045 | static int ref_get_fields(struct extent_buffer *eb, unsigned long ref_ptr, | |
1046 | u32 *namelen, char **name, u64 *index) | |
1047 | { | |
1048 | struct btrfs_inode_ref *ref; | |
1049 | ||
1050 | ref = (struct btrfs_inode_ref *)ref_ptr; | |
1051 | ||
1052 | *namelen = btrfs_inode_ref_name_len(eb, ref); | |
1053 | *name = kmalloc(*namelen, GFP_NOFS); | |
1054 | if (*name == NULL) | |
1055 | return -ENOMEM; | |
1056 | ||
1057 | read_extent_buffer(eb, *name, (unsigned long)(ref + 1), *namelen); | |
1058 | ||
1059 | *index = btrfs_inode_ref_index(eb, ref); | |
1060 | ||
1061 | return 0; | |
1062 | } | |
1063 | ||
5a1d7843 JS |
1064 | /* |
1065 | * replay one inode back reference item found in the log tree. | |
1066 | * eb, slot and key refer to the buffer and key found in the log tree. | |
1067 | * root is the destination we are replaying into, and path is for temp | |
1068 | * use by this function. (it should be released on return). | |
1069 | */ | |
1070 | static noinline int add_inode_ref(struct btrfs_trans_handle *trans, | |
1071 | struct btrfs_root *root, | |
1072 | struct btrfs_root *log, | |
1073 | struct btrfs_path *path, | |
1074 | struct extent_buffer *eb, int slot, | |
1075 | struct btrfs_key *key) | |
1076 | { | |
5a1d7843 JS |
1077 | struct inode *dir; |
1078 | struct inode *inode; | |
1079 | unsigned long ref_ptr; | |
1080 | unsigned long ref_end; | |
1081 | char *name; | |
1082 | int namelen; | |
1083 | int ret; | |
1084 | int search_done = 0; | |
f186373f MF |
1085 | int log_ref_ver = 0; |
1086 | u64 parent_objectid; | |
1087 | u64 inode_objectid; | |
f46dbe3d | 1088 | u64 ref_index = 0; |
f186373f MF |
1089 | int ref_struct_size; |
1090 | ||
1091 | ref_ptr = btrfs_item_ptr_offset(eb, slot); | |
1092 | ref_end = ref_ptr + btrfs_item_size_nr(eb, slot); | |
1093 | ||
1094 | if (key->type == BTRFS_INODE_EXTREF_KEY) { | |
1095 | struct btrfs_inode_extref *r; | |
1096 | ||
1097 | ref_struct_size = sizeof(struct btrfs_inode_extref); | |
1098 | log_ref_ver = 1; | |
1099 | r = (struct btrfs_inode_extref *)ref_ptr; | |
1100 | parent_objectid = btrfs_inode_extref_parent(eb, r); | |
1101 | } else { | |
1102 | ref_struct_size = sizeof(struct btrfs_inode_ref); | |
1103 | parent_objectid = key->offset; | |
1104 | } | |
1105 | inode_objectid = key->objectid; | |
e02119d5 | 1106 | |
5a1d7843 JS |
1107 | /* |
1108 | * it is possible that we didn't log all the parent directories | |
1109 | * for a given inode. If we don't find the dir, just don't | |
1110 | * copy the back ref in. The link count fixup code will take | |
1111 | * care of the rest | |
1112 | */ | |
f186373f | 1113 | dir = read_one_inode(root, parent_objectid); |
5a1d7843 JS |
1114 | if (!dir) |
1115 | return -ENOENT; | |
1116 | ||
f186373f | 1117 | inode = read_one_inode(root, inode_objectid); |
5a1d7843 JS |
1118 | if (!inode) { |
1119 | iput(dir); | |
1120 | return -EIO; | |
1121 | } | |
1122 | ||
5a1d7843 | 1123 | while (ref_ptr < ref_end) { |
f186373f MF |
1124 | if (log_ref_ver) { |
1125 | ret = extref_get_fields(eb, ref_ptr, &namelen, &name, | |
1126 | &ref_index, &parent_objectid); | |
1127 | /* | |
1128 | * parent object can change from one array | |
1129 | * item to another. | |
1130 | */ | |
1131 | if (!dir) | |
1132 | dir = read_one_inode(root, parent_objectid); | |
1133 | if (!dir) | |
1134 | return -ENOENT; | |
1135 | } else { | |
1136 | ret = ref_get_fields(eb, ref_ptr, &namelen, &name, | |
1137 | &ref_index); | |
1138 | } | |
1139 | if (ret) | |
1140 | return ret; | |
5a1d7843 JS |
1141 | |
1142 | /* if we already have a perfect match, we're done */ | |
1143 | if (!inode_in_dir(root, path, btrfs_ino(dir), btrfs_ino(inode), | |
f186373f | 1144 | ref_index, name, namelen)) { |
5a1d7843 JS |
1145 | /* |
1146 | * look for a conflicting back reference in the | |
1147 | * metadata. if we find one we have to unlink that name | |
1148 | * of the file before we add our new link. Later on, we | |
1149 | * overwrite any existing back reference, and we don't | |
1150 | * want to create dangling pointers in the directory. | |
1151 | */ | |
1152 | ||
1153 | if (!search_done) { | |
1154 | ret = __add_inode_ref(trans, root, path, log, | |
f186373f MF |
1155 | dir, inode, eb, |
1156 | inode_objectid, | |
1157 | parent_objectid, | |
1158 | ref_index, name, namelen, | |
5a1d7843 | 1159 | &search_done); |
3650860b JB |
1160 | if (ret == 1) { |
1161 | ret = 0; | |
1162 | goto out; | |
1163 | } | |
1164 | if (ret) | |
5a1d7843 | 1165 | goto out; |
5a1d7843 JS |
1166 | } |
1167 | ||
1168 | /* insert our name */ | |
1169 | ret = btrfs_add_link(trans, dir, inode, name, namelen, | |
f186373f | 1170 | 0, ref_index); |
3650860b JB |
1171 | if (ret) |
1172 | goto out; | |
5a1d7843 JS |
1173 | |
1174 | btrfs_update_inode(trans, root, inode); | |
1175 | } | |
1176 | ||
f186373f | 1177 | ref_ptr = (unsigned long)(ref_ptr + ref_struct_size) + namelen; |
5a1d7843 | 1178 | kfree(name); |
f186373f MF |
1179 | if (log_ref_ver) { |
1180 | iput(dir); | |
1181 | dir = NULL; | |
1182 | } | |
5a1d7843 | 1183 | } |
e02119d5 CM |
1184 | |
1185 | /* finally write the back reference in the inode */ | |
1186 | ret = overwrite_item(trans, root, path, eb, slot, key); | |
5a1d7843 | 1187 | out: |
b3b4aa74 | 1188 | btrfs_release_path(path); |
e02119d5 CM |
1189 | iput(dir); |
1190 | iput(inode); | |
3650860b | 1191 | return ret; |
e02119d5 CM |
1192 | } |
1193 | ||
c71bf099 YZ |
1194 | static int insert_orphan_item(struct btrfs_trans_handle *trans, |
1195 | struct btrfs_root *root, u64 offset) | |
1196 | { | |
1197 | int ret; | |
1198 | ret = btrfs_find_orphan_item(root, offset); | |
1199 | if (ret > 0) | |
1200 | ret = btrfs_insert_orphan_item(trans, root, offset); | |
1201 | return ret; | |
1202 | } | |
1203 | ||
f186373f MF |
1204 | static int count_inode_extrefs(struct btrfs_root *root, |
1205 | struct inode *inode, struct btrfs_path *path) | |
1206 | { | |
1207 | int ret = 0; | |
1208 | int name_len; | |
1209 | unsigned int nlink = 0; | |
1210 | u32 item_size; | |
1211 | u32 cur_offset = 0; | |
1212 | u64 inode_objectid = btrfs_ino(inode); | |
1213 | u64 offset = 0; | |
1214 | unsigned long ptr; | |
1215 | struct btrfs_inode_extref *extref; | |
1216 | struct extent_buffer *leaf; | |
1217 | ||
1218 | while (1) { | |
1219 | ret = btrfs_find_one_extref(root, inode_objectid, offset, path, | |
1220 | &extref, &offset); | |
1221 | if (ret) | |
1222 | break; | |
c71bf099 | 1223 | |
f186373f MF |
1224 | leaf = path->nodes[0]; |
1225 | item_size = btrfs_item_size_nr(leaf, path->slots[0]); | |
1226 | ptr = btrfs_item_ptr_offset(leaf, path->slots[0]); | |
1227 | ||
1228 | while (cur_offset < item_size) { | |
1229 | extref = (struct btrfs_inode_extref *) (ptr + cur_offset); | |
1230 | name_len = btrfs_inode_extref_name_len(leaf, extref); | |
1231 | ||
1232 | nlink++; | |
1233 | ||
1234 | cur_offset += name_len + sizeof(*extref); | |
1235 | } | |
1236 | ||
1237 | offset++; | |
1238 | btrfs_release_path(path); | |
1239 | } | |
1240 | btrfs_release_path(path); | |
1241 | ||
1242 | if (ret < 0) | |
1243 | return ret; | |
1244 | return nlink; | |
1245 | } | |
1246 | ||
1247 | static int count_inode_refs(struct btrfs_root *root, | |
1248 | struct inode *inode, struct btrfs_path *path) | |
e02119d5 | 1249 | { |
e02119d5 CM |
1250 | int ret; |
1251 | struct btrfs_key key; | |
f186373f | 1252 | unsigned int nlink = 0; |
e02119d5 CM |
1253 | unsigned long ptr; |
1254 | unsigned long ptr_end; | |
1255 | int name_len; | |
33345d01 | 1256 | u64 ino = btrfs_ino(inode); |
e02119d5 | 1257 | |
33345d01 | 1258 | key.objectid = ino; |
e02119d5 CM |
1259 | key.type = BTRFS_INODE_REF_KEY; |
1260 | key.offset = (u64)-1; | |
1261 | ||
d397712b | 1262 | while (1) { |
e02119d5 CM |
1263 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); |
1264 | if (ret < 0) | |
1265 | break; | |
1266 | if (ret > 0) { | |
1267 | if (path->slots[0] == 0) | |
1268 | break; | |
1269 | path->slots[0]--; | |
1270 | } | |
1271 | btrfs_item_key_to_cpu(path->nodes[0], &key, | |
1272 | path->slots[0]); | |
33345d01 | 1273 | if (key.objectid != ino || |
e02119d5 CM |
1274 | key.type != BTRFS_INODE_REF_KEY) |
1275 | break; | |
1276 | ptr = btrfs_item_ptr_offset(path->nodes[0], path->slots[0]); | |
1277 | ptr_end = ptr + btrfs_item_size_nr(path->nodes[0], | |
1278 | path->slots[0]); | |
d397712b | 1279 | while (ptr < ptr_end) { |
e02119d5 CM |
1280 | struct btrfs_inode_ref *ref; |
1281 | ||
1282 | ref = (struct btrfs_inode_ref *)ptr; | |
1283 | name_len = btrfs_inode_ref_name_len(path->nodes[0], | |
1284 | ref); | |
1285 | ptr = (unsigned long)(ref + 1) + name_len; | |
1286 | nlink++; | |
1287 | } | |
1288 | ||
1289 | if (key.offset == 0) | |
1290 | break; | |
1291 | key.offset--; | |
b3b4aa74 | 1292 | btrfs_release_path(path); |
e02119d5 | 1293 | } |
b3b4aa74 | 1294 | btrfs_release_path(path); |
f186373f MF |
1295 | |
1296 | return nlink; | |
1297 | } | |
1298 | ||
1299 | /* | |
1300 | * There are a few corners where the link count of the file can't | |
1301 | * be properly maintained during replay. So, instead of adding | |
1302 | * lots of complexity to the log code, we just scan the backrefs | |
1303 | * for any file that has been through replay. | |
1304 | * | |
1305 | * The scan will update the link count on the inode to reflect the | |
1306 | * number of back refs found. If it goes down to zero, the iput | |
1307 | * will free the inode. | |
1308 | */ | |
1309 | static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans, | |
1310 | struct btrfs_root *root, | |
1311 | struct inode *inode) | |
1312 | { | |
1313 | struct btrfs_path *path; | |
1314 | int ret; | |
1315 | u64 nlink = 0; | |
1316 | u64 ino = btrfs_ino(inode); | |
1317 | ||
1318 | path = btrfs_alloc_path(); | |
1319 | if (!path) | |
1320 | return -ENOMEM; | |
1321 | ||
1322 | ret = count_inode_refs(root, inode, path); | |
1323 | if (ret < 0) | |
1324 | goto out; | |
1325 | ||
1326 | nlink = ret; | |
1327 | ||
1328 | ret = count_inode_extrefs(root, inode, path); | |
1329 | if (ret == -ENOENT) | |
1330 | ret = 0; | |
1331 | ||
1332 | if (ret < 0) | |
1333 | goto out; | |
1334 | ||
1335 | nlink += ret; | |
1336 | ||
1337 | ret = 0; | |
1338 | ||
e02119d5 | 1339 | if (nlink != inode->i_nlink) { |
bfe86848 | 1340 | set_nlink(inode, nlink); |
e02119d5 CM |
1341 | btrfs_update_inode(trans, root, inode); |
1342 | } | |
8d5bf1cb | 1343 | BTRFS_I(inode)->index_cnt = (u64)-1; |
e02119d5 | 1344 | |
c71bf099 YZ |
1345 | if (inode->i_nlink == 0) { |
1346 | if (S_ISDIR(inode->i_mode)) { | |
1347 | ret = replay_dir_deletes(trans, root, NULL, path, | |
33345d01 | 1348 | ino, 1); |
3650860b JB |
1349 | if (ret) |
1350 | goto out; | |
c71bf099 | 1351 | } |
33345d01 | 1352 | ret = insert_orphan_item(trans, root, ino); |
12fcfd22 | 1353 | } |
12fcfd22 | 1354 | |
f186373f MF |
1355 | out: |
1356 | btrfs_free_path(path); | |
1357 | return ret; | |
e02119d5 CM |
1358 | } |
1359 | ||
1360 | static noinline int fixup_inode_link_counts(struct btrfs_trans_handle *trans, | |
1361 | struct btrfs_root *root, | |
1362 | struct btrfs_path *path) | |
1363 | { | |
1364 | int ret; | |
1365 | struct btrfs_key key; | |
1366 | struct inode *inode; | |
1367 | ||
1368 | key.objectid = BTRFS_TREE_LOG_FIXUP_OBJECTID; | |
1369 | key.type = BTRFS_ORPHAN_ITEM_KEY; | |
1370 | key.offset = (u64)-1; | |
d397712b | 1371 | while (1) { |
e02119d5 CM |
1372 | ret = btrfs_search_slot(trans, root, &key, path, -1, 1); |
1373 | if (ret < 0) | |
1374 | break; | |
1375 | ||
1376 | if (ret == 1) { | |
1377 | if (path->slots[0] == 0) | |
1378 | break; | |
1379 | path->slots[0]--; | |
1380 | } | |
1381 | ||
1382 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1383 | if (key.objectid != BTRFS_TREE_LOG_FIXUP_OBJECTID || | |
1384 | key.type != BTRFS_ORPHAN_ITEM_KEY) | |
1385 | break; | |
1386 | ||
1387 | ret = btrfs_del_item(trans, root, path); | |
65a246c5 TI |
1388 | if (ret) |
1389 | goto out; | |
e02119d5 | 1390 | |
b3b4aa74 | 1391 | btrfs_release_path(path); |
e02119d5 | 1392 | inode = read_one_inode(root, key.offset); |
c00e9493 TI |
1393 | if (!inode) |
1394 | return -EIO; | |
e02119d5 CM |
1395 | |
1396 | ret = fixup_inode_link_count(trans, root, inode); | |
e02119d5 | 1397 | iput(inode); |
3650860b JB |
1398 | if (ret) |
1399 | goto out; | |
e02119d5 | 1400 | |
12fcfd22 CM |
1401 | /* |
1402 | * fixup on a directory may create new entries, | |
1403 | * make sure we always look for the highset possible | |
1404 | * offset | |
1405 | */ | |
1406 | key.offset = (u64)-1; | |
e02119d5 | 1407 | } |
65a246c5 TI |
1408 | ret = 0; |
1409 | out: | |
b3b4aa74 | 1410 | btrfs_release_path(path); |
65a246c5 | 1411 | return ret; |
e02119d5 CM |
1412 | } |
1413 | ||
1414 | ||
1415 | /* | |
1416 | * record a given inode in the fixup dir so we can check its link | |
1417 | * count when replay is done. The link count is incremented here | |
1418 | * so the inode won't go away until we check it | |
1419 | */ | |
1420 | static noinline int link_to_fixup_dir(struct btrfs_trans_handle *trans, | |
1421 | struct btrfs_root *root, | |
1422 | struct btrfs_path *path, | |
1423 | u64 objectid) | |
1424 | { | |
1425 | struct btrfs_key key; | |
1426 | int ret = 0; | |
1427 | struct inode *inode; | |
1428 | ||
1429 | inode = read_one_inode(root, objectid); | |
c00e9493 TI |
1430 | if (!inode) |
1431 | return -EIO; | |
e02119d5 CM |
1432 | |
1433 | key.objectid = BTRFS_TREE_LOG_FIXUP_OBJECTID; | |
1434 | btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY); | |
1435 | key.offset = objectid; | |
1436 | ||
1437 | ret = btrfs_insert_empty_item(trans, root, path, &key, 0); | |
1438 | ||
b3b4aa74 | 1439 | btrfs_release_path(path); |
e02119d5 | 1440 | if (ret == 0) { |
9bf7a489 JB |
1441 | if (!inode->i_nlink) |
1442 | set_nlink(inode, 1); | |
1443 | else | |
1444 | btrfs_inc_nlink(inode); | |
b9959295 | 1445 | ret = btrfs_update_inode(trans, root, inode); |
e02119d5 CM |
1446 | } else if (ret == -EEXIST) { |
1447 | ret = 0; | |
1448 | } else { | |
3650860b | 1449 | BUG(); /* Logic Error */ |
e02119d5 CM |
1450 | } |
1451 | iput(inode); | |
1452 | ||
1453 | return ret; | |
1454 | } | |
1455 | ||
1456 | /* | |
1457 | * when replaying the log for a directory, we only insert names | |
1458 | * for inodes that actually exist. This means an fsync on a directory | |
1459 | * does not implicitly fsync all the new files in it | |
1460 | */ | |
1461 | static noinline int insert_one_name(struct btrfs_trans_handle *trans, | |
1462 | struct btrfs_root *root, | |
1463 | struct btrfs_path *path, | |
1464 | u64 dirid, u64 index, | |
1465 | char *name, int name_len, u8 type, | |
1466 | struct btrfs_key *location) | |
1467 | { | |
1468 | struct inode *inode; | |
1469 | struct inode *dir; | |
1470 | int ret; | |
1471 | ||
1472 | inode = read_one_inode(root, location->objectid); | |
1473 | if (!inode) | |
1474 | return -ENOENT; | |
1475 | ||
1476 | dir = read_one_inode(root, dirid); | |
1477 | if (!dir) { | |
1478 | iput(inode); | |
1479 | return -EIO; | |
1480 | } | |
1481 | ret = btrfs_add_link(trans, dir, inode, name, name_len, 1, index); | |
1482 | ||
1483 | /* FIXME, put inode into FIXUP list */ | |
1484 | ||
1485 | iput(inode); | |
1486 | iput(dir); | |
1487 | return ret; | |
1488 | } | |
1489 | ||
1490 | /* | |
1491 | * take a single entry in a log directory item and replay it into | |
1492 | * the subvolume. | |
1493 | * | |
1494 | * if a conflicting item exists in the subdirectory already, | |
1495 | * the inode it points to is unlinked and put into the link count | |
1496 | * fix up tree. | |
1497 | * | |
1498 | * If a name from the log points to a file or directory that does | |
1499 | * not exist in the FS, it is skipped. fsyncs on directories | |
1500 | * do not force down inodes inside that directory, just changes to the | |
1501 | * names or unlinks in a directory. | |
1502 | */ | |
1503 | static noinline int replay_one_name(struct btrfs_trans_handle *trans, | |
1504 | struct btrfs_root *root, | |
1505 | struct btrfs_path *path, | |
1506 | struct extent_buffer *eb, | |
1507 | struct btrfs_dir_item *di, | |
1508 | struct btrfs_key *key) | |
1509 | { | |
1510 | char *name; | |
1511 | int name_len; | |
1512 | struct btrfs_dir_item *dst_di; | |
1513 | struct btrfs_key found_key; | |
1514 | struct btrfs_key log_key; | |
1515 | struct inode *dir; | |
e02119d5 | 1516 | u8 log_type; |
4bef0848 | 1517 | int exists; |
3650860b | 1518 | int ret = 0; |
e02119d5 CM |
1519 | |
1520 | dir = read_one_inode(root, key->objectid); | |
c00e9493 TI |
1521 | if (!dir) |
1522 | return -EIO; | |
e02119d5 CM |
1523 | |
1524 | name_len = btrfs_dir_name_len(eb, di); | |
1525 | name = kmalloc(name_len, GFP_NOFS); | |
2a29edc6 | 1526 | if (!name) |
1527 | return -ENOMEM; | |
1528 | ||
e02119d5 CM |
1529 | log_type = btrfs_dir_type(eb, di); |
1530 | read_extent_buffer(eb, name, (unsigned long)(di + 1), | |
1531 | name_len); | |
1532 | ||
1533 | btrfs_dir_item_key_to_cpu(eb, di, &log_key); | |
4bef0848 CM |
1534 | exists = btrfs_lookup_inode(trans, root, path, &log_key, 0); |
1535 | if (exists == 0) | |
1536 | exists = 1; | |
1537 | else | |
1538 | exists = 0; | |
b3b4aa74 | 1539 | btrfs_release_path(path); |
4bef0848 | 1540 | |
e02119d5 CM |
1541 | if (key->type == BTRFS_DIR_ITEM_KEY) { |
1542 | dst_di = btrfs_lookup_dir_item(trans, root, path, key->objectid, | |
1543 | name, name_len, 1); | |
d397712b | 1544 | } else if (key->type == BTRFS_DIR_INDEX_KEY) { |
e02119d5 CM |
1545 | dst_di = btrfs_lookup_dir_index_item(trans, root, path, |
1546 | key->objectid, | |
1547 | key->offset, name, | |
1548 | name_len, 1); | |
1549 | } else { | |
3650860b JB |
1550 | /* Corruption */ |
1551 | ret = -EINVAL; | |
1552 | goto out; | |
e02119d5 | 1553 | } |
c704005d | 1554 | if (IS_ERR_OR_NULL(dst_di)) { |
e02119d5 CM |
1555 | /* we need a sequence number to insert, so we only |
1556 | * do inserts for the BTRFS_DIR_INDEX_KEY types | |
1557 | */ | |
1558 | if (key->type != BTRFS_DIR_INDEX_KEY) | |
1559 | goto out; | |
1560 | goto insert; | |
1561 | } | |
1562 | ||
1563 | btrfs_dir_item_key_to_cpu(path->nodes[0], dst_di, &found_key); | |
1564 | /* the existing item matches the logged item */ | |
1565 | if (found_key.objectid == log_key.objectid && | |
1566 | found_key.type == log_key.type && | |
1567 | found_key.offset == log_key.offset && | |
1568 | btrfs_dir_type(path->nodes[0], dst_di) == log_type) { | |
1569 | goto out; | |
1570 | } | |
1571 | ||
1572 | /* | |
1573 | * don't drop the conflicting directory entry if the inode | |
1574 | * for the new entry doesn't exist | |
1575 | */ | |
4bef0848 | 1576 | if (!exists) |
e02119d5 CM |
1577 | goto out; |
1578 | ||
e02119d5 | 1579 | ret = drop_one_dir_item(trans, root, path, dir, dst_di); |
3650860b JB |
1580 | if (ret) |
1581 | goto out; | |
e02119d5 CM |
1582 | |
1583 | if (key->type == BTRFS_DIR_INDEX_KEY) | |
1584 | goto insert; | |
1585 | out: | |
b3b4aa74 | 1586 | btrfs_release_path(path); |
e02119d5 CM |
1587 | kfree(name); |
1588 | iput(dir); | |
3650860b | 1589 | return ret; |
e02119d5 CM |
1590 | |
1591 | insert: | |
b3b4aa74 | 1592 | btrfs_release_path(path); |
e02119d5 CM |
1593 | ret = insert_one_name(trans, root, path, key->objectid, key->offset, |
1594 | name, name_len, log_type, &log_key); | |
3650860b JB |
1595 | if (ret && ret != -ENOENT) |
1596 | goto out; | |
1597 | ret = 0; | |
e02119d5 CM |
1598 | goto out; |
1599 | } | |
1600 | ||
1601 | /* | |
1602 | * find all the names in a directory item and reconcile them into | |
1603 | * the subvolume. Only BTRFS_DIR_ITEM_KEY types will have more than | |
1604 | * one name in a directory item, but the same code gets used for | |
1605 | * both directory index types | |
1606 | */ | |
1607 | static noinline int replay_one_dir_item(struct btrfs_trans_handle *trans, | |
1608 | struct btrfs_root *root, | |
1609 | struct btrfs_path *path, | |
1610 | struct extent_buffer *eb, int slot, | |
1611 | struct btrfs_key *key) | |
1612 | { | |
1613 | int ret; | |
1614 | u32 item_size = btrfs_item_size_nr(eb, slot); | |
1615 | struct btrfs_dir_item *di; | |
1616 | int name_len; | |
1617 | unsigned long ptr; | |
1618 | unsigned long ptr_end; | |
1619 | ||
1620 | ptr = btrfs_item_ptr_offset(eb, slot); | |
1621 | ptr_end = ptr + item_size; | |
d397712b | 1622 | while (ptr < ptr_end) { |
e02119d5 | 1623 | di = (struct btrfs_dir_item *)ptr; |
22a94d44 JB |
1624 | if (verify_dir_item(root, eb, di)) |
1625 | return -EIO; | |
e02119d5 CM |
1626 | name_len = btrfs_dir_name_len(eb, di); |
1627 | ret = replay_one_name(trans, root, path, eb, di, key); | |
3650860b JB |
1628 | if (ret) |
1629 | return ret; | |
e02119d5 CM |
1630 | ptr = (unsigned long)(di + 1); |
1631 | ptr += name_len; | |
1632 | } | |
1633 | return 0; | |
1634 | } | |
1635 | ||
1636 | /* | |
1637 | * directory replay has two parts. There are the standard directory | |
1638 | * items in the log copied from the subvolume, and range items | |
1639 | * created in the log while the subvolume was logged. | |
1640 | * | |
1641 | * The range items tell us which parts of the key space the log | |
1642 | * is authoritative for. During replay, if a key in the subvolume | |
1643 | * directory is in a logged range item, but not actually in the log | |
1644 | * that means it was deleted from the directory before the fsync | |
1645 | * and should be removed. | |
1646 | */ | |
1647 | static noinline int find_dir_range(struct btrfs_root *root, | |
1648 | struct btrfs_path *path, | |
1649 | u64 dirid, int key_type, | |
1650 | u64 *start_ret, u64 *end_ret) | |
1651 | { | |
1652 | struct btrfs_key key; | |
1653 | u64 found_end; | |
1654 | struct btrfs_dir_log_item *item; | |
1655 | int ret; | |
1656 | int nritems; | |
1657 | ||
1658 | if (*start_ret == (u64)-1) | |
1659 | return 1; | |
1660 | ||
1661 | key.objectid = dirid; | |
1662 | key.type = key_type; | |
1663 | key.offset = *start_ret; | |
1664 | ||
1665 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
1666 | if (ret < 0) | |
1667 | goto out; | |
1668 | if (ret > 0) { | |
1669 | if (path->slots[0] == 0) | |
1670 | goto out; | |
1671 | path->slots[0]--; | |
1672 | } | |
1673 | if (ret != 0) | |
1674 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1675 | ||
1676 | if (key.type != key_type || key.objectid != dirid) { | |
1677 | ret = 1; | |
1678 | goto next; | |
1679 | } | |
1680 | item = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
1681 | struct btrfs_dir_log_item); | |
1682 | found_end = btrfs_dir_log_end(path->nodes[0], item); | |
1683 | ||
1684 | if (*start_ret >= key.offset && *start_ret <= found_end) { | |
1685 | ret = 0; | |
1686 | *start_ret = key.offset; | |
1687 | *end_ret = found_end; | |
1688 | goto out; | |
1689 | } | |
1690 | ret = 1; | |
1691 | next: | |
1692 | /* check the next slot in the tree to see if it is a valid item */ | |
1693 | nritems = btrfs_header_nritems(path->nodes[0]); | |
1694 | if (path->slots[0] >= nritems) { | |
1695 | ret = btrfs_next_leaf(root, path); | |
1696 | if (ret) | |
1697 | goto out; | |
1698 | } else { | |
1699 | path->slots[0]++; | |
1700 | } | |
1701 | ||
1702 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1703 | ||
1704 | if (key.type != key_type || key.objectid != dirid) { | |
1705 | ret = 1; | |
1706 | goto out; | |
1707 | } | |
1708 | item = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
1709 | struct btrfs_dir_log_item); | |
1710 | found_end = btrfs_dir_log_end(path->nodes[0], item); | |
1711 | *start_ret = key.offset; | |
1712 | *end_ret = found_end; | |
1713 | ret = 0; | |
1714 | out: | |
b3b4aa74 | 1715 | btrfs_release_path(path); |
e02119d5 CM |
1716 | return ret; |
1717 | } | |
1718 | ||
1719 | /* | |
1720 | * this looks for a given directory item in the log. If the directory | |
1721 | * item is not in the log, the item is removed and the inode it points | |
1722 | * to is unlinked | |
1723 | */ | |
1724 | static noinline int check_item_in_log(struct btrfs_trans_handle *trans, | |
1725 | struct btrfs_root *root, | |
1726 | struct btrfs_root *log, | |
1727 | struct btrfs_path *path, | |
1728 | struct btrfs_path *log_path, | |
1729 | struct inode *dir, | |
1730 | struct btrfs_key *dir_key) | |
1731 | { | |
1732 | int ret; | |
1733 | struct extent_buffer *eb; | |
1734 | int slot; | |
1735 | u32 item_size; | |
1736 | struct btrfs_dir_item *di; | |
1737 | struct btrfs_dir_item *log_di; | |
1738 | int name_len; | |
1739 | unsigned long ptr; | |
1740 | unsigned long ptr_end; | |
1741 | char *name; | |
1742 | struct inode *inode; | |
1743 | struct btrfs_key location; | |
1744 | ||
1745 | again: | |
1746 | eb = path->nodes[0]; | |
1747 | slot = path->slots[0]; | |
1748 | item_size = btrfs_item_size_nr(eb, slot); | |
1749 | ptr = btrfs_item_ptr_offset(eb, slot); | |
1750 | ptr_end = ptr + item_size; | |
d397712b | 1751 | while (ptr < ptr_end) { |
e02119d5 | 1752 | di = (struct btrfs_dir_item *)ptr; |
22a94d44 JB |
1753 | if (verify_dir_item(root, eb, di)) { |
1754 | ret = -EIO; | |
1755 | goto out; | |
1756 | } | |
1757 | ||
e02119d5 CM |
1758 | name_len = btrfs_dir_name_len(eb, di); |
1759 | name = kmalloc(name_len, GFP_NOFS); | |
1760 | if (!name) { | |
1761 | ret = -ENOMEM; | |
1762 | goto out; | |
1763 | } | |
1764 | read_extent_buffer(eb, name, (unsigned long)(di + 1), | |
1765 | name_len); | |
1766 | log_di = NULL; | |
12fcfd22 | 1767 | if (log && dir_key->type == BTRFS_DIR_ITEM_KEY) { |
e02119d5 CM |
1768 | log_di = btrfs_lookup_dir_item(trans, log, log_path, |
1769 | dir_key->objectid, | |
1770 | name, name_len, 0); | |
12fcfd22 | 1771 | } else if (log && dir_key->type == BTRFS_DIR_INDEX_KEY) { |
e02119d5 CM |
1772 | log_di = btrfs_lookup_dir_index_item(trans, log, |
1773 | log_path, | |
1774 | dir_key->objectid, | |
1775 | dir_key->offset, | |
1776 | name, name_len, 0); | |
1777 | } | |
c704005d | 1778 | if (IS_ERR_OR_NULL(log_di)) { |
e02119d5 | 1779 | btrfs_dir_item_key_to_cpu(eb, di, &location); |
b3b4aa74 DS |
1780 | btrfs_release_path(path); |
1781 | btrfs_release_path(log_path); | |
e02119d5 | 1782 | inode = read_one_inode(root, location.objectid); |
c00e9493 TI |
1783 | if (!inode) { |
1784 | kfree(name); | |
1785 | return -EIO; | |
1786 | } | |
e02119d5 CM |
1787 | |
1788 | ret = link_to_fixup_dir(trans, root, | |
1789 | path, location.objectid); | |
3650860b JB |
1790 | if (ret) { |
1791 | kfree(name); | |
1792 | iput(inode); | |
1793 | goto out; | |
1794 | } | |
1795 | ||
e02119d5 CM |
1796 | btrfs_inc_nlink(inode); |
1797 | ret = btrfs_unlink_inode(trans, root, dir, inode, | |
1798 | name, name_len); | |
3650860b JB |
1799 | if (!ret) |
1800 | btrfs_run_delayed_items(trans, root); | |
e02119d5 CM |
1801 | kfree(name); |
1802 | iput(inode); | |
3650860b JB |
1803 | if (ret) |
1804 | goto out; | |
e02119d5 CM |
1805 | |
1806 | /* there might still be more names under this key | |
1807 | * check and repeat if required | |
1808 | */ | |
1809 | ret = btrfs_search_slot(NULL, root, dir_key, path, | |
1810 | 0, 0); | |
1811 | if (ret == 0) | |
1812 | goto again; | |
1813 | ret = 0; | |
1814 | goto out; | |
1815 | } | |
b3b4aa74 | 1816 | btrfs_release_path(log_path); |
e02119d5 CM |
1817 | kfree(name); |
1818 | ||
1819 | ptr = (unsigned long)(di + 1); | |
1820 | ptr += name_len; | |
1821 | } | |
1822 | ret = 0; | |
1823 | out: | |
b3b4aa74 DS |
1824 | btrfs_release_path(path); |
1825 | btrfs_release_path(log_path); | |
e02119d5 CM |
1826 | return ret; |
1827 | } | |
1828 | ||
1829 | /* | |
1830 | * deletion replay happens before we copy any new directory items | |
1831 | * out of the log or out of backreferences from inodes. It | |
1832 | * scans the log to find ranges of keys that log is authoritative for, | |
1833 | * and then scans the directory to find items in those ranges that are | |
1834 | * not present in the log. | |
1835 | * | |
1836 | * Anything we don't find in the log is unlinked and removed from the | |
1837 | * directory. | |
1838 | */ | |
1839 | static noinline int replay_dir_deletes(struct btrfs_trans_handle *trans, | |
1840 | struct btrfs_root *root, | |
1841 | struct btrfs_root *log, | |
1842 | struct btrfs_path *path, | |
12fcfd22 | 1843 | u64 dirid, int del_all) |
e02119d5 CM |
1844 | { |
1845 | u64 range_start; | |
1846 | u64 range_end; | |
1847 | int key_type = BTRFS_DIR_LOG_ITEM_KEY; | |
1848 | int ret = 0; | |
1849 | struct btrfs_key dir_key; | |
1850 | struct btrfs_key found_key; | |
1851 | struct btrfs_path *log_path; | |
1852 | struct inode *dir; | |
1853 | ||
1854 | dir_key.objectid = dirid; | |
1855 | dir_key.type = BTRFS_DIR_ITEM_KEY; | |
1856 | log_path = btrfs_alloc_path(); | |
1857 | if (!log_path) | |
1858 | return -ENOMEM; | |
1859 | ||
1860 | dir = read_one_inode(root, dirid); | |
1861 | /* it isn't an error if the inode isn't there, that can happen | |
1862 | * because we replay the deletes before we copy in the inode item | |
1863 | * from the log | |
1864 | */ | |
1865 | if (!dir) { | |
1866 | btrfs_free_path(log_path); | |
1867 | return 0; | |
1868 | } | |
1869 | again: | |
1870 | range_start = 0; | |
1871 | range_end = 0; | |
d397712b | 1872 | while (1) { |
12fcfd22 CM |
1873 | if (del_all) |
1874 | range_end = (u64)-1; | |
1875 | else { | |
1876 | ret = find_dir_range(log, path, dirid, key_type, | |
1877 | &range_start, &range_end); | |
1878 | if (ret != 0) | |
1879 | break; | |
1880 | } | |
e02119d5 CM |
1881 | |
1882 | dir_key.offset = range_start; | |
d397712b | 1883 | while (1) { |
e02119d5 CM |
1884 | int nritems; |
1885 | ret = btrfs_search_slot(NULL, root, &dir_key, path, | |
1886 | 0, 0); | |
1887 | if (ret < 0) | |
1888 | goto out; | |
1889 | ||
1890 | nritems = btrfs_header_nritems(path->nodes[0]); | |
1891 | if (path->slots[0] >= nritems) { | |
1892 | ret = btrfs_next_leaf(root, path); | |
1893 | if (ret) | |
1894 | break; | |
1895 | } | |
1896 | btrfs_item_key_to_cpu(path->nodes[0], &found_key, | |
1897 | path->slots[0]); | |
1898 | if (found_key.objectid != dirid || | |
1899 | found_key.type != dir_key.type) | |
1900 | goto next_type; | |
1901 | ||
1902 | if (found_key.offset > range_end) | |
1903 | break; | |
1904 | ||
1905 | ret = check_item_in_log(trans, root, log, path, | |
12fcfd22 CM |
1906 | log_path, dir, |
1907 | &found_key); | |
3650860b JB |
1908 | if (ret) |
1909 | goto out; | |
e02119d5 CM |
1910 | if (found_key.offset == (u64)-1) |
1911 | break; | |
1912 | dir_key.offset = found_key.offset + 1; | |
1913 | } | |
b3b4aa74 | 1914 | btrfs_release_path(path); |
e02119d5 CM |
1915 | if (range_end == (u64)-1) |
1916 | break; | |
1917 | range_start = range_end + 1; | |
1918 | } | |
1919 | ||
1920 | next_type: | |
1921 | ret = 0; | |
1922 | if (key_type == BTRFS_DIR_LOG_ITEM_KEY) { | |
1923 | key_type = BTRFS_DIR_LOG_INDEX_KEY; | |
1924 | dir_key.type = BTRFS_DIR_INDEX_KEY; | |
b3b4aa74 | 1925 | btrfs_release_path(path); |
e02119d5 CM |
1926 | goto again; |
1927 | } | |
1928 | out: | |
b3b4aa74 | 1929 | btrfs_release_path(path); |
e02119d5 CM |
1930 | btrfs_free_path(log_path); |
1931 | iput(dir); | |
1932 | return ret; | |
1933 | } | |
1934 | ||
1935 | /* | |
1936 | * the process_func used to replay items from the log tree. This | |
1937 | * gets called in two different stages. The first stage just looks | |
1938 | * for inodes and makes sure they are all copied into the subvolume. | |
1939 | * | |
1940 | * The second stage copies all the other item types from the log into | |
1941 | * the subvolume. The two stage approach is slower, but gets rid of | |
1942 | * lots of complexity around inodes referencing other inodes that exist | |
1943 | * only in the log (references come from either directory items or inode | |
1944 | * back refs). | |
1945 | */ | |
1946 | static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb, | |
1947 | struct walk_control *wc, u64 gen) | |
1948 | { | |
1949 | int nritems; | |
1950 | struct btrfs_path *path; | |
1951 | struct btrfs_root *root = wc->replay_dest; | |
1952 | struct btrfs_key key; | |
e02119d5 CM |
1953 | int level; |
1954 | int i; | |
1955 | int ret; | |
1956 | ||
018642a1 TI |
1957 | ret = btrfs_read_buffer(eb, gen); |
1958 | if (ret) | |
1959 | return ret; | |
e02119d5 CM |
1960 | |
1961 | level = btrfs_header_level(eb); | |
1962 | ||
1963 | if (level != 0) | |
1964 | return 0; | |
1965 | ||
1966 | path = btrfs_alloc_path(); | |
1e5063d0 MF |
1967 | if (!path) |
1968 | return -ENOMEM; | |
e02119d5 CM |
1969 | |
1970 | nritems = btrfs_header_nritems(eb); | |
1971 | for (i = 0; i < nritems; i++) { | |
1972 | btrfs_item_key_to_cpu(eb, &key, i); | |
e02119d5 CM |
1973 | |
1974 | /* inode keys are done during the first stage */ | |
1975 | if (key.type == BTRFS_INODE_ITEM_KEY && | |
1976 | wc->stage == LOG_WALK_REPLAY_INODES) { | |
e02119d5 CM |
1977 | struct btrfs_inode_item *inode_item; |
1978 | u32 mode; | |
1979 | ||
1980 | inode_item = btrfs_item_ptr(eb, i, | |
1981 | struct btrfs_inode_item); | |
1982 | mode = btrfs_inode_mode(eb, inode_item); | |
1983 | if (S_ISDIR(mode)) { | |
1984 | ret = replay_dir_deletes(wc->trans, | |
12fcfd22 | 1985 | root, log, path, key.objectid, 0); |
b50c6e25 JB |
1986 | if (ret) |
1987 | break; | |
e02119d5 CM |
1988 | } |
1989 | ret = overwrite_item(wc->trans, root, path, | |
1990 | eb, i, &key); | |
b50c6e25 JB |
1991 | if (ret) |
1992 | break; | |
e02119d5 | 1993 | |
c71bf099 YZ |
1994 | /* for regular files, make sure corresponding |
1995 | * orhpan item exist. extents past the new EOF | |
1996 | * will be truncated later by orphan cleanup. | |
e02119d5 CM |
1997 | */ |
1998 | if (S_ISREG(mode)) { | |
c71bf099 YZ |
1999 | ret = insert_orphan_item(wc->trans, root, |
2000 | key.objectid); | |
b50c6e25 JB |
2001 | if (ret) |
2002 | break; | |
e02119d5 | 2003 | } |
c71bf099 | 2004 | |
e02119d5 CM |
2005 | ret = link_to_fixup_dir(wc->trans, root, |
2006 | path, key.objectid); | |
b50c6e25 JB |
2007 | if (ret) |
2008 | break; | |
e02119d5 CM |
2009 | } |
2010 | if (wc->stage < LOG_WALK_REPLAY_ALL) | |
2011 | continue; | |
2012 | ||
2013 | /* these keys are simply copied */ | |
2014 | if (key.type == BTRFS_XATTR_ITEM_KEY) { | |
2015 | ret = overwrite_item(wc->trans, root, path, | |
2016 | eb, i, &key); | |
b50c6e25 JB |
2017 | if (ret) |
2018 | break; | |
e02119d5 CM |
2019 | } else if (key.type == BTRFS_INODE_REF_KEY) { |
2020 | ret = add_inode_ref(wc->trans, root, log, path, | |
2021 | eb, i, &key); | |
b50c6e25 JB |
2022 | if (ret && ret != -ENOENT) |
2023 | break; | |
2024 | ret = 0; | |
f186373f MF |
2025 | } else if (key.type == BTRFS_INODE_EXTREF_KEY) { |
2026 | ret = add_inode_ref(wc->trans, root, log, path, | |
2027 | eb, i, &key); | |
b50c6e25 JB |
2028 | if (ret && ret != -ENOENT) |
2029 | break; | |
2030 | ret = 0; | |
e02119d5 CM |
2031 | } else if (key.type == BTRFS_EXTENT_DATA_KEY) { |
2032 | ret = replay_one_extent(wc->trans, root, path, | |
2033 | eb, i, &key); | |
b50c6e25 JB |
2034 | if (ret) |
2035 | break; | |
e02119d5 CM |
2036 | } else if (key.type == BTRFS_DIR_ITEM_KEY || |
2037 | key.type == BTRFS_DIR_INDEX_KEY) { | |
2038 | ret = replay_one_dir_item(wc->trans, root, path, | |
2039 | eb, i, &key); | |
b50c6e25 JB |
2040 | if (ret) |
2041 | break; | |
e02119d5 CM |
2042 | } |
2043 | } | |
2044 | btrfs_free_path(path); | |
b50c6e25 | 2045 | return ret; |
e02119d5 CM |
2046 | } |
2047 | ||
d397712b | 2048 | static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans, |
e02119d5 CM |
2049 | struct btrfs_root *root, |
2050 | struct btrfs_path *path, int *level, | |
2051 | struct walk_control *wc) | |
2052 | { | |
2053 | u64 root_owner; | |
e02119d5 CM |
2054 | u64 bytenr; |
2055 | u64 ptr_gen; | |
2056 | struct extent_buffer *next; | |
2057 | struct extent_buffer *cur; | |
2058 | struct extent_buffer *parent; | |
2059 | u32 blocksize; | |
2060 | int ret = 0; | |
2061 | ||
2062 | WARN_ON(*level < 0); | |
2063 | WARN_ON(*level >= BTRFS_MAX_LEVEL); | |
2064 | ||
d397712b | 2065 | while (*level > 0) { |
e02119d5 CM |
2066 | WARN_ON(*level < 0); |
2067 | WARN_ON(*level >= BTRFS_MAX_LEVEL); | |
2068 | cur = path->nodes[*level]; | |
2069 | ||
2070 | if (btrfs_header_level(cur) != *level) | |
2071 | WARN_ON(1); | |
2072 | ||
2073 | if (path->slots[*level] >= | |
2074 | btrfs_header_nritems(cur)) | |
2075 | break; | |
2076 | ||
2077 | bytenr = btrfs_node_blockptr(cur, path->slots[*level]); | |
2078 | ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]); | |
2079 | blocksize = btrfs_level_size(root, *level - 1); | |
2080 | ||
2081 | parent = path->nodes[*level]; | |
2082 | root_owner = btrfs_header_owner(parent); | |
e02119d5 CM |
2083 | |
2084 | next = btrfs_find_create_tree_block(root, bytenr, blocksize); | |
2a29edc6 | 2085 | if (!next) |
2086 | return -ENOMEM; | |
e02119d5 | 2087 | |
e02119d5 | 2088 | if (*level == 1) { |
1e5063d0 | 2089 | ret = wc->process_func(root, next, wc, ptr_gen); |
b50c6e25 JB |
2090 | if (ret) { |
2091 | free_extent_buffer(next); | |
1e5063d0 | 2092 | return ret; |
b50c6e25 | 2093 | } |
4a500fd1 | 2094 | |
e02119d5 CM |
2095 | path->slots[*level]++; |
2096 | if (wc->free) { | |
018642a1 TI |
2097 | ret = btrfs_read_buffer(next, ptr_gen); |
2098 | if (ret) { | |
2099 | free_extent_buffer(next); | |
2100 | return ret; | |
2101 | } | |
e02119d5 CM |
2102 | |
2103 | btrfs_tree_lock(next); | |
b4ce94de | 2104 | btrfs_set_lock_blocking(next); |
bd681513 | 2105 | clean_tree_block(trans, root, next); |
e02119d5 CM |
2106 | btrfs_wait_tree_block_writeback(next); |
2107 | btrfs_tree_unlock(next); | |
2108 | ||
e02119d5 CM |
2109 | WARN_ON(root_owner != |
2110 | BTRFS_TREE_LOG_OBJECTID); | |
e688b725 | 2111 | ret = btrfs_free_and_pin_reserved_extent(root, |
d00aff00 | 2112 | bytenr, blocksize); |
3650860b JB |
2113 | if (ret) { |
2114 | free_extent_buffer(next); | |
2115 | return ret; | |
2116 | } | |
e02119d5 CM |
2117 | } |
2118 | free_extent_buffer(next); | |
2119 | continue; | |
2120 | } | |
018642a1 TI |
2121 | ret = btrfs_read_buffer(next, ptr_gen); |
2122 | if (ret) { | |
2123 | free_extent_buffer(next); | |
2124 | return ret; | |
2125 | } | |
e02119d5 CM |
2126 | |
2127 | WARN_ON(*level <= 0); | |
2128 | if (path->nodes[*level-1]) | |
2129 | free_extent_buffer(path->nodes[*level-1]); | |
2130 | path->nodes[*level-1] = next; | |
2131 | *level = btrfs_header_level(next); | |
2132 | path->slots[*level] = 0; | |
2133 | cond_resched(); | |
2134 | } | |
2135 | WARN_ON(*level < 0); | |
2136 | WARN_ON(*level >= BTRFS_MAX_LEVEL); | |
2137 | ||
4a500fd1 | 2138 | path->slots[*level] = btrfs_header_nritems(path->nodes[*level]); |
e02119d5 CM |
2139 | |
2140 | cond_resched(); | |
2141 | return 0; | |
2142 | } | |
2143 | ||
d397712b | 2144 | static noinline int walk_up_log_tree(struct btrfs_trans_handle *trans, |
e02119d5 CM |
2145 | struct btrfs_root *root, |
2146 | struct btrfs_path *path, int *level, | |
2147 | struct walk_control *wc) | |
2148 | { | |
2149 | u64 root_owner; | |
e02119d5 CM |
2150 | int i; |
2151 | int slot; | |
2152 | int ret; | |
2153 | ||
d397712b | 2154 | for (i = *level; i < BTRFS_MAX_LEVEL - 1 && path->nodes[i]; i++) { |
e02119d5 | 2155 | slot = path->slots[i]; |
4a500fd1 | 2156 | if (slot + 1 < btrfs_header_nritems(path->nodes[i])) { |
e02119d5 CM |
2157 | path->slots[i]++; |
2158 | *level = i; | |
2159 | WARN_ON(*level == 0); | |
2160 | return 0; | |
2161 | } else { | |
31840ae1 ZY |
2162 | struct extent_buffer *parent; |
2163 | if (path->nodes[*level] == root->node) | |
2164 | parent = path->nodes[*level]; | |
2165 | else | |
2166 | parent = path->nodes[*level + 1]; | |
2167 | ||
2168 | root_owner = btrfs_header_owner(parent); | |
1e5063d0 | 2169 | ret = wc->process_func(root, path->nodes[*level], wc, |
e02119d5 | 2170 | btrfs_header_generation(path->nodes[*level])); |
1e5063d0 MF |
2171 | if (ret) |
2172 | return ret; | |
2173 | ||
e02119d5 CM |
2174 | if (wc->free) { |
2175 | struct extent_buffer *next; | |
2176 | ||
2177 | next = path->nodes[*level]; | |
2178 | ||
2179 | btrfs_tree_lock(next); | |
b4ce94de | 2180 | btrfs_set_lock_blocking(next); |
bd681513 | 2181 | clean_tree_block(trans, root, next); |
e02119d5 CM |
2182 | btrfs_wait_tree_block_writeback(next); |
2183 | btrfs_tree_unlock(next); | |
2184 | ||
e02119d5 | 2185 | WARN_ON(root_owner != BTRFS_TREE_LOG_OBJECTID); |
e688b725 | 2186 | ret = btrfs_free_and_pin_reserved_extent(root, |
e02119d5 | 2187 | path->nodes[*level]->start, |
d00aff00 | 2188 | path->nodes[*level]->len); |
3650860b JB |
2189 | if (ret) |
2190 | return ret; | |
e02119d5 CM |
2191 | } |
2192 | free_extent_buffer(path->nodes[*level]); | |
2193 | path->nodes[*level] = NULL; | |
2194 | *level = i + 1; | |
2195 | } | |
2196 | } | |
2197 | return 1; | |
2198 | } | |
2199 | ||
2200 | /* | |
2201 | * drop the reference count on the tree rooted at 'snap'. This traverses | |
2202 | * the tree freeing any blocks that have a ref count of zero after being | |
2203 | * decremented. | |
2204 | */ | |
2205 | static int walk_log_tree(struct btrfs_trans_handle *trans, | |
2206 | struct btrfs_root *log, struct walk_control *wc) | |
2207 | { | |
2208 | int ret = 0; | |
2209 | int wret; | |
2210 | int level; | |
2211 | struct btrfs_path *path; | |
e02119d5 CM |
2212 | int orig_level; |
2213 | ||
2214 | path = btrfs_alloc_path(); | |
db5b493a TI |
2215 | if (!path) |
2216 | return -ENOMEM; | |
e02119d5 CM |
2217 | |
2218 | level = btrfs_header_level(log->node); | |
2219 | orig_level = level; | |
2220 | path->nodes[level] = log->node; | |
2221 | extent_buffer_get(log->node); | |
2222 | path->slots[level] = 0; | |
2223 | ||
d397712b | 2224 | while (1) { |
e02119d5 CM |
2225 | wret = walk_down_log_tree(trans, log, path, &level, wc); |
2226 | if (wret > 0) | |
2227 | break; | |
79787eaa | 2228 | if (wret < 0) { |
e02119d5 | 2229 | ret = wret; |
79787eaa JM |
2230 | goto out; |
2231 | } | |
e02119d5 CM |
2232 | |
2233 | wret = walk_up_log_tree(trans, log, path, &level, wc); | |
2234 | if (wret > 0) | |
2235 | break; | |
79787eaa | 2236 | if (wret < 0) { |
e02119d5 | 2237 | ret = wret; |
79787eaa JM |
2238 | goto out; |
2239 | } | |
e02119d5 CM |
2240 | } |
2241 | ||
2242 | /* was the root node processed? if not, catch it here */ | |
2243 | if (path->nodes[orig_level]) { | |
79787eaa | 2244 | ret = wc->process_func(log, path->nodes[orig_level], wc, |
e02119d5 | 2245 | btrfs_header_generation(path->nodes[orig_level])); |
79787eaa JM |
2246 | if (ret) |
2247 | goto out; | |
e02119d5 CM |
2248 | if (wc->free) { |
2249 | struct extent_buffer *next; | |
2250 | ||
2251 | next = path->nodes[orig_level]; | |
2252 | ||
2253 | btrfs_tree_lock(next); | |
b4ce94de | 2254 | btrfs_set_lock_blocking(next); |
bd681513 | 2255 | clean_tree_block(trans, log, next); |
e02119d5 CM |
2256 | btrfs_wait_tree_block_writeback(next); |
2257 | btrfs_tree_unlock(next); | |
2258 | ||
e02119d5 CM |
2259 | WARN_ON(log->root_key.objectid != |
2260 | BTRFS_TREE_LOG_OBJECTID); | |
e688b725 | 2261 | ret = btrfs_free_and_pin_reserved_extent(log, next->start, |
d00aff00 | 2262 | next->len); |
3650860b JB |
2263 | if (ret) |
2264 | goto out; | |
e02119d5 CM |
2265 | } |
2266 | } | |
2267 | ||
79787eaa | 2268 | out: |
e02119d5 | 2269 | btrfs_free_path(path); |
e02119d5 CM |
2270 | return ret; |
2271 | } | |
2272 | ||
7237f183 YZ |
2273 | /* |
2274 | * helper function to update the item for a given subvolumes log root | |
2275 | * in the tree of log roots | |
2276 | */ | |
2277 | static int update_log_root(struct btrfs_trans_handle *trans, | |
2278 | struct btrfs_root *log) | |
2279 | { | |
2280 | int ret; | |
2281 | ||
2282 | if (log->log_transid == 1) { | |
2283 | /* insert root item on the first sync */ | |
2284 | ret = btrfs_insert_root(trans, log->fs_info->log_root_tree, | |
2285 | &log->root_key, &log->root_item); | |
2286 | } else { | |
2287 | ret = btrfs_update_root(trans, log->fs_info->log_root_tree, | |
2288 | &log->root_key, &log->root_item); | |
2289 | } | |
2290 | return ret; | |
2291 | } | |
2292 | ||
12fcfd22 CM |
2293 | static int wait_log_commit(struct btrfs_trans_handle *trans, |
2294 | struct btrfs_root *root, unsigned long transid) | |
e02119d5 CM |
2295 | { |
2296 | DEFINE_WAIT(wait); | |
7237f183 | 2297 | int index = transid % 2; |
e02119d5 | 2298 | |
7237f183 YZ |
2299 | /* |
2300 | * we only allow two pending log transactions at a time, | |
2301 | * so we know that if ours is more than 2 older than the | |
2302 | * current transaction, we're done | |
2303 | */ | |
e02119d5 | 2304 | do { |
7237f183 YZ |
2305 | prepare_to_wait(&root->log_commit_wait[index], |
2306 | &wait, TASK_UNINTERRUPTIBLE); | |
2307 | mutex_unlock(&root->log_mutex); | |
12fcfd22 CM |
2308 | |
2309 | if (root->fs_info->last_trans_log_full_commit != | |
2310 | trans->transid && root->log_transid < transid + 2 && | |
7237f183 YZ |
2311 | atomic_read(&root->log_commit[index])) |
2312 | schedule(); | |
12fcfd22 | 2313 | |
7237f183 YZ |
2314 | finish_wait(&root->log_commit_wait[index], &wait); |
2315 | mutex_lock(&root->log_mutex); | |
6dd70ce4 JK |
2316 | } while (root->fs_info->last_trans_log_full_commit != |
2317 | trans->transid && root->log_transid < transid + 2 && | |
7237f183 YZ |
2318 | atomic_read(&root->log_commit[index])); |
2319 | return 0; | |
2320 | } | |
2321 | ||
143bede5 JM |
2322 | static void wait_for_writer(struct btrfs_trans_handle *trans, |
2323 | struct btrfs_root *root) | |
7237f183 YZ |
2324 | { |
2325 | DEFINE_WAIT(wait); | |
6dd70ce4 JK |
2326 | while (root->fs_info->last_trans_log_full_commit != |
2327 | trans->transid && atomic_read(&root->log_writers)) { | |
7237f183 YZ |
2328 | prepare_to_wait(&root->log_writer_wait, |
2329 | &wait, TASK_UNINTERRUPTIBLE); | |
2330 | mutex_unlock(&root->log_mutex); | |
12fcfd22 CM |
2331 | if (root->fs_info->last_trans_log_full_commit != |
2332 | trans->transid && atomic_read(&root->log_writers)) | |
e02119d5 | 2333 | schedule(); |
7237f183 YZ |
2334 | mutex_lock(&root->log_mutex); |
2335 | finish_wait(&root->log_writer_wait, &wait); | |
2336 | } | |
e02119d5 CM |
2337 | } |
2338 | ||
2339 | /* | |
2340 | * btrfs_sync_log does sends a given tree log down to the disk and | |
2341 | * updates the super blocks to record it. When this call is done, | |
12fcfd22 CM |
2342 | * you know that any inodes previously logged are safely on disk only |
2343 | * if it returns 0. | |
2344 | * | |
2345 | * Any other return value means you need to call btrfs_commit_transaction. | |
2346 | * Some of the edge cases for fsyncing directories that have had unlinks | |
2347 | * or renames done in the past mean that sometimes the only safe | |
2348 | * fsync is to commit the whole FS. When btrfs_sync_log returns -EAGAIN, | |
2349 | * that has happened. | |
e02119d5 CM |
2350 | */ |
2351 | int btrfs_sync_log(struct btrfs_trans_handle *trans, | |
2352 | struct btrfs_root *root) | |
2353 | { | |
7237f183 YZ |
2354 | int index1; |
2355 | int index2; | |
8cef4e16 | 2356 | int mark; |
e02119d5 | 2357 | int ret; |
e02119d5 | 2358 | struct btrfs_root *log = root->log_root; |
7237f183 | 2359 | struct btrfs_root *log_root_tree = root->fs_info->log_root_tree; |
8cef4e16 | 2360 | unsigned long log_transid = 0; |
e02119d5 | 2361 | |
7237f183 | 2362 | mutex_lock(&root->log_mutex); |
2ab28f32 | 2363 | log_transid = root->log_transid; |
7237f183 YZ |
2364 | index1 = root->log_transid % 2; |
2365 | if (atomic_read(&root->log_commit[index1])) { | |
12fcfd22 | 2366 | wait_log_commit(trans, root, root->log_transid); |
7237f183 YZ |
2367 | mutex_unlock(&root->log_mutex); |
2368 | return 0; | |
e02119d5 | 2369 | } |
7237f183 YZ |
2370 | atomic_set(&root->log_commit[index1], 1); |
2371 | ||
2372 | /* wait for previous tree log sync to complete */ | |
2373 | if (atomic_read(&root->log_commit[(index1 + 1) % 2])) | |
12fcfd22 | 2374 | wait_log_commit(trans, root, root->log_transid - 1); |
86df7eb9 | 2375 | while (1) { |
2ecb7923 | 2376 | int batch = atomic_read(&root->log_batch); |
cd354ad6 CM |
2377 | /* when we're on an ssd, just kick the log commit out */ |
2378 | if (!btrfs_test_opt(root, SSD) && root->log_multiple_pids) { | |
86df7eb9 YZ |
2379 | mutex_unlock(&root->log_mutex); |
2380 | schedule_timeout_uninterruptible(1); | |
2381 | mutex_lock(&root->log_mutex); | |
2382 | } | |
12fcfd22 | 2383 | wait_for_writer(trans, root); |
2ecb7923 | 2384 | if (batch == atomic_read(&root->log_batch)) |
e02119d5 CM |
2385 | break; |
2386 | } | |
e02119d5 | 2387 | |
12fcfd22 CM |
2388 | /* bail out if we need to do a full commit */ |
2389 | if (root->fs_info->last_trans_log_full_commit == trans->transid) { | |
2390 | ret = -EAGAIN; | |
2ab28f32 | 2391 | btrfs_free_logged_extents(log, log_transid); |
12fcfd22 CM |
2392 | mutex_unlock(&root->log_mutex); |
2393 | goto out; | |
2394 | } | |
2395 | ||
8cef4e16 YZ |
2396 | if (log_transid % 2 == 0) |
2397 | mark = EXTENT_DIRTY; | |
2398 | else | |
2399 | mark = EXTENT_NEW; | |
2400 | ||
690587d1 CM |
2401 | /* we start IO on all the marked extents here, but we don't actually |
2402 | * wait for them until later. | |
2403 | */ | |
8cef4e16 | 2404 | ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark); |
79787eaa JM |
2405 | if (ret) { |
2406 | btrfs_abort_transaction(trans, root, ret); | |
2ab28f32 | 2407 | btrfs_free_logged_extents(log, log_transid); |
79787eaa JM |
2408 | mutex_unlock(&root->log_mutex); |
2409 | goto out; | |
2410 | } | |
7237f183 | 2411 | |
5d4f98a2 | 2412 | btrfs_set_root_node(&log->root_item, log->node); |
7237f183 | 2413 | |
7237f183 YZ |
2414 | root->log_transid++; |
2415 | log->log_transid = root->log_transid; | |
ff782e0a | 2416 | root->log_start_pid = 0; |
7237f183 YZ |
2417 | smp_mb(); |
2418 | /* | |
8cef4e16 YZ |
2419 | * IO has been started, blocks of the log tree have WRITTEN flag set |
2420 | * in their headers. new modifications of the log will be written to | |
2421 | * new positions. so it's safe to allow log writers to go in. | |
7237f183 YZ |
2422 | */ |
2423 | mutex_unlock(&root->log_mutex); | |
2424 | ||
2425 | mutex_lock(&log_root_tree->log_mutex); | |
2ecb7923 | 2426 | atomic_inc(&log_root_tree->log_batch); |
7237f183 YZ |
2427 | atomic_inc(&log_root_tree->log_writers); |
2428 | mutex_unlock(&log_root_tree->log_mutex); | |
2429 | ||
2430 | ret = update_log_root(trans, log); | |
7237f183 YZ |
2431 | |
2432 | mutex_lock(&log_root_tree->log_mutex); | |
2433 | if (atomic_dec_and_test(&log_root_tree->log_writers)) { | |
2434 | smp_mb(); | |
2435 | if (waitqueue_active(&log_root_tree->log_writer_wait)) | |
2436 | wake_up(&log_root_tree->log_writer_wait); | |
2437 | } | |
2438 | ||
4a500fd1 | 2439 | if (ret) { |
79787eaa JM |
2440 | if (ret != -ENOSPC) { |
2441 | btrfs_abort_transaction(trans, root, ret); | |
2442 | mutex_unlock(&log_root_tree->log_mutex); | |
2443 | goto out; | |
2444 | } | |
4a500fd1 YZ |
2445 | root->fs_info->last_trans_log_full_commit = trans->transid; |
2446 | btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); | |
2ab28f32 | 2447 | btrfs_free_logged_extents(log, log_transid); |
4a500fd1 YZ |
2448 | mutex_unlock(&log_root_tree->log_mutex); |
2449 | ret = -EAGAIN; | |
2450 | goto out; | |
2451 | } | |
2452 | ||
7237f183 YZ |
2453 | index2 = log_root_tree->log_transid % 2; |
2454 | if (atomic_read(&log_root_tree->log_commit[index2])) { | |
8cef4e16 | 2455 | btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); |
12fcfd22 CM |
2456 | wait_log_commit(trans, log_root_tree, |
2457 | log_root_tree->log_transid); | |
2ab28f32 | 2458 | btrfs_free_logged_extents(log, log_transid); |
7237f183 | 2459 | mutex_unlock(&log_root_tree->log_mutex); |
b31eabd8 | 2460 | ret = 0; |
7237f183 YZ |
2461 | goto out; |
2462 | } | |
2463 | atomic_set(&log_root_tree->log_commit[index2], 1); | |
2464 | ||
12fcfd22 CM |
2465 | if (atomic_read(&log_root_tree->log_commit[(index2 + 1) % 2])) { |
2466 | wait_log_commit(trans, log_root_tree, | |
2467 | log_root_tree->log_transid - 1); | |
2468 | } | |
2469 | ||
2470 | wait_for_writer(trans, log_root_tree); | |
7237f183 | 2471 | |
12fcfd22 CM |
2472 | /* |
2473 | * now that we've moved on to the tree of log tree roots, | |
2474 | * check the full commit flag again | |
2475 | */ | |
2476 | if (root->fs_info->last_trans_log_full_commit == trans->transid) { | |
8cef4e16 | 2477 | btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); |
2ab28f32 | 2478 | btrfs_free_logged_extents(log, log_transid); |
12fcfd22 CM |
2479 | mutex_unlock(&log_root_tree->log_mutex); |
2480 | ret = -EAGAIN; | |
2481 | goto out_wake_log_root; | |
2482 | } | |
7237f183 YZ |
2483 | |
2484 | ret = btrfs_write_and_wait_marked_extents(log_root_tree, | |
8cef4e16 YZ |
2485 | &log_root_tree->dirty_log_pages, |
2486 | EXTENT_DIRTY | EXTENT_NEW); | |
79787eaa JM |
2487 | if (ret) { |
2488 | btrfs_abort_transaction(trans, root, ret); | |
2ab28f32 | 2489 | btrfs_free_logged_extents(log, log_transid); |
79787eaa JM |
2490 | mutex_unlock(&log_root_tree->log_mutex); |
2491 | goto out_wake_log_root; | |
2492 | } | |
8cef4e16 | 2493 | btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); |
2ab28f32 | 2494 | btrfs_wait_logged_extents(log, log_transid); |
e02119d5 | 2495 | |
6c41761f | 2496 | btrfs_set_super_log_root(root->fs_info->super_for_commit, |
7237f183 | 2497 | log_root_tree->node->start); |
6c41761f | 2498 | btrfs_set_super_log_root_level(root->fs_info->super_for_commit, |
7237f183 | 2499 | btrfs_header_level(log_root_tree->node)); |
e02119d5 | 2500 | |
7237f183 | 2501 | log_root_tree->log_transid++; |
e02119d5 | 2502 | smp_mb(); |
7237f183 YZ |
2503 | |
2504 | mutex_unlock(&log_root_tree->log_mutex); | |
2505 | ||
2506 | /* | |
2507 | * nobody else is going to jump in and write the the ctree | |
2508 | * super here because the log_commit atomic below is protecting | |
2509 | * us. We must be called with a transaction handle pinning | |
2510 | * the running transaction open, so a full commit can't hop | |
2511 | * in and cause problems either. | |
2512 | */ | |
a2de733c | 2513 | btrfs_scrub_pause_super(root); |
5af3e8cc | 2514 | ret = write_ctree_super(trans, root->fs_info->tree_root, 1); |
a2de733c | 2515 | btrfs_scrub_continue_super(root); |
5af3e8cc SB |
2516 | if (ret) { |
2517 | btrfs_abort_transaction(trans, root, ret); | |
2518 | goto out_wake_log_root; | |
2519 | } | |
7237f183 | 2520 | |
257c62e1 CM |
2521 | mutex_lock(&root->log_mutex); |
2522 | if (root->last_log_commit < log_transid) | |
2523 | root->last_log_commit = log_transid; | |
2524 | mutex_unlock(&root->log_mutex); | |
2525 | ||
12fcfd22 | 2526 | out_wake_log_root: |
7237f183 YZ |
2527 | atomic_set(&log_root_tree->log_commit[index2], 0); |
2528 | smp_mb(); | |
2529 | if (waitqueue_active(&log_root_tree->log_commit_wait[index2])) | |
2530 | wake_up(&log_root_tree->log_commit_wait[index2]); | |
e02119d5 | 2531 | out: |
7237f183 YZ |
2532 | atomic_set(&root->log_commit[index1], 0); |
2533 | smp_mb(); | |
2534 | if (waitqueue_active(&root->log_commit_wait[index1])) | |
2535 | wake_up(&root->log_commit_wait[index1]); | |
b31eabd8 | 2536 | return ret; |
e02119d5 CM |
2537 | } |
2538 | ||
4a500fd1 YZ |
2539 | static void free_log_tree(struct btrfs_trans_handle *trans, |
2540 | struct btrfs_root *log) | |
e02119d5 CM |
2541 | { |
2542 | int ret; | |
d0c803c4 CM |
2543 | u64 start; |
2544 | u64 end; | |
e02119d5 CM |
2545 | struct walk_control wc = { |
2546 | .free = 1, | |
2547 | .process_func = process_one_buffer | |
2548 | }; | |
2549 | ||
3321719e LB |
2550 | if (trans) { |
2551 | ret = walk_log_tree(trans, log, &wc); | |
3650860b JB |
2552 | |
2553 | /* I don't think this can happen but just in case */ | |
2554 | if (ret) | |
2555 | btrfs_abort_transaction(trans, log, ret); | |
3321719e | 2556 | } |
e02119d5 | 2557 | |
d397712b | 2558 | while (1) { |
d0c803c4 | 2559 | ret = find_first_extent_bit(&log->dirty_log_pages, |
e6138876 JB |
2560 | 0, &start, &end, EXTENT_DIRTY | EXTENT_NEW, |
2561 | NULL); | |
d0c803c4 CM |
2562 | if (ret) |
2563 | break; | |
2564 | ||
8cef4e16 YZ |
2565 | clear_extent_bits(&log->dirty_log_pages, start, end, |
2566 | EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS); | |
d0c803c4 CM |
2567 | } |
2568 | ||
2ab28f32 JB |
2569 | /* |
2570 | * We may have short-circuited the log tree with the full commit logic | |
2571 | * and left ordered extents on our list, so clear these out to keep us | |
2572 | * from leaking inodes and memory. | |
2573 | */ | |
2574 | btrfs_free_logged_extents(log, 0); | |
2575 | btrfs_free_logged_extents(log, 1); | |
2576 | ||
7237f183 YZ |
2577 | free_extent_buffer(log->node); |
2578 | kfree(log); | |
4a500fd1 YZ |
2579 | } |
2580 | ||
2581 | /* | |
2582 | * free all the extents used by the tree log. This should be called | |
2583 | * at commit time of the full transaction | |
2584 | */ | |
2585 | int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root) | |
2586 | { | |
2587 | if (root->log_root) { | |
2588 | free_log_tree(trans, root->log_root); | |
2589 | root->log_root = NULL; | |
2590 | } | |
2591 | return 0; | |
2592 | } | |
2593 | ||
2594 | int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans, | |
2595 | struct btrfs_fs_info *fs_info) | |
2596 | { | |
2597 | if (fs_info->log_root_tree) { | |
2598 | free_log_tree(trans, fs_info->log_root_tree); | |
2599 | fs_info->log_root_tree = NULL; | |
2600 | } | |
e02119d5 CM |
2601 | return 0; |
2602 | } | |
2603 | ||
e02119d5 CM |
2604 | /* |
2605 | * If both a file and directory are logged, and unlinks or renames are | |
2606 | * mixed in, we have a few interesting corners: | |
2607 | * | |
2608 | * create file X in dir Y | |
2609 | * link file X to X.link in dir Y | |
2610 | * fsync file X | |
2611 | * unlink file X but leave X.link | |
2612 | * fsync dir Y | |
2613 | * | |
2614 | * After a crash we would expect only X.link to exist. But file X | |
2615 | * didn't get fsync'd again so the log has back refs for X and X.link. | |
2616 | * | |
2617 | * We solve this by removing directory entries and inode backrefs from the | |
2618 | * log when a file that was logged in the current transaction is | |
2619 | * unlinked. Any later fsync will include the updated log entries, and | |
2620 | * we'll be able to reconstruct the proper directory items from backrefs. | |
2621 | * | |
2622 | * This optimizations allows us to avoid relogging the entire inode | |
2623 | * or the entire directory. | |
2624 | */ | |
2625 | int btrfs_del_dir_entries_in_log(struct btrfs_trans_handle *trans, | |
2626 | struct btrfs_root *root, | |
2627 | const char *name, int name_len, | |
2628 | struct inode *dir, u64 index) | |
2629 | { | |
2630 | struct btrfs_root *log; | |
2631 | struct btrfs_dir_item *di; | |
2632 | struct btrfs_path *path; | |
2633 | int ret; | |
4a500fd1 | 2634 | int err = 0; |
e02119d5 | 2635 | int bytes_del = 0; |
33345d01 | 2636 | u64 dir_ino = btrfs_ino(dir); |
e02119d5 | 2637 | |
3a5f1d45 CM |
2638 | if (BTRFS_I(dir)->logged_trans < trans->transid) |
2639 | return 0; | |
2640 | ||
e02119d5 CM |
2641 | ret = join_running_log_trans(root); |
2642 | if (ret) | |
2643 | return 0; | |
2644 | ||
2645 | mutex_lock(&BTRFS_I(dir)->log_mutex); | |
2646 | ||
2647 | log = root->log_root; | |
2648 | path = btrfs_alloc_path(); | |
a62f44a5 TI |
2649 | if (!path) { |
2650 | err = -ENOMEM; | |
2651 | goto out_unlock; | |
2652 | } | |
2a29edc6 | 2653 | |
33345d01 | 2654 | di = btrfs_lookup_dir_item(trans, log, path, dir_ino, |
e02119d5 | 2655 | name, name_len, -1); |
4a500fd1 YZ |
2656 | if (IS_ERR(di)) { |
2657 | err = PTR_ERR(di); | |
2658 | goto fail; | |
2659 | } | |
2660 | if (di) { | |
e02119d5 CM |
2661 | ret = btrfs_delete_one_dir_name(trans, log, path, di); |
2662 | bytes_del += name_len; | |
3650860b JB |
2663 | if (ret) { |
2664 | err = ret; | |
2665 | goto fail; | |
2666 | } | |
e02119d5 | 2667 | } |
b3b4aa74 | 2668 | btrfs_release_path(path); |
33345d01 | 2669 | di = btrfs_lookup_dir_index_item(trans, log, path, dir_ino, |
e02119d5 | 2670 | index, name, name_len, -1); |
4a500fd1 YZ |
2671 | if (IS_ERR(di)) { |
2672 | err = PTR_ERR(di); | |
2673 | goto fail; | |
2674 | } | |
2675 | if (di) { | |
e02119d5 CM |
2676 | ret = btrfs_delete_one_dir_name(trans, log, path, di); |
2677 | bytes_del += name_len; | |
3650860b JB |
2678 | if (ret) { |
2679 | err = ret; | |
2680 | goto fail; | |
2681 | } | |
e02119d5 CM |
2682 | } |
2683 | ||
2684 | /* update the directory size in the log to reflect the names | |
2685 | * we have removed | |
2686 | */ | |
2687 | if (bytes_del) { | |
2688 | struct btrfs_key key; | |
2689 | ||
33345d01 | 2690 | key.objectid = dir_ino; |
e02119d5 CM |
2691 | key.offset = 0; |
2692 | key.type = BTRFS_INODE_ITEM_KEY; | |
b3b4aa74 | 2693 | btrfs_release_path(path); |
e02119d5 CM |
2694 | |
2695 | ret = btrfs_search_slot(trans, log, &key, path, 0, 1); | |
4a500fd1 YZ |
2696 | if (ret < 0) { |
2697 | err = ret; | |
2698 | goto fail; | |
2699 | } | |
e02119d5 CM |
2700 | if (ret == 0) { |
2701 | struct btrfs_inode_item *item; | |
2702 | u64 i_size; | |
2703 | ||
2704 | item = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
2705 | struct btrfs_inode_item); | |
2706 | i_size = btrfs_inode_size(path->nodes[0], item); | |
2707 | if (i_size > bytes_del) | |
2708 | i_size -= bytes_del; | |
2709 | else | |
2710 | i_size = 0; | |
2711 | btrfs_set_inode_size(path->nodes[0], item, i_size); | |
2712 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
2713 | } else | |
2714 | ret = 0; | |
b3b4aa74 | 2715 | btrfs_release_path(path); |
e02119d5 | 2716 | } |
4a500fd1 | 2717 | fail: |
e02119d5 | 2718 | btrfs_free_path(path); |
a62f44a5 | 2719 | out_unlock: |
e02119d5 | 2720 | mutex_unlock(&BTRFS_I(dir)->log_mutex); |
4a500fd1 YZ |
2721 | if (ret == -ENOSPC) { |
2722 | root->fs_info->last_trans_log_full_commit = trans->transid; | |
2723 | ret = 0; | |
79787eaa JM |
2724 | } else if (ret < 0) |
2725 | btrfs_abort_transaction(trans, root, ret); | |
2726 | ||
12fcfd22 | 2727 | btrfs_end_log_trans(root); |
e02119d5 | 2728 | |
411fc6bc | 2729 | return err; |
e02119d5 CM |
2730 | } |
2731 | ||
2732 | /* see comments for btrfs_del_dir_entries_in_log */ | |
2733 | int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans, | |
2734 | struct btrfs_root *root, | |
2735 | const char *name, int name_len, | |
2736 | struct inode *inode, u64 dirid) | |
2737 | { | |
2738 | struct btrfs_root *log; | |
2739 | u64 index; | |
2740 | int ret; | |
2741 | ||
3a5f1d45 CM |
2742 | if (BTRFS_I(inode)->logged_trans < trans->transid) |
2743 | return 0; | |
2744 | ||
e02119d5 CM |
2745 | ret = join_running_log_trans(root); |
2746 | if (ret) | |
2747 | return 0; | |
2748 | log = root->log_root; | |
2749 | mutex_lock(&BTRFS_I(inode)->log_mutex); | |
2750 | ||
33345d01 | 2751 | ret = btrfs_del_inode_ref(trans, log, name, name_len, btrfs_ino(inode), |
e02119d5 CM |
2752 | dirid, &index); |
2753 | mutex_unlock(&BTRFS_I(inode)->log_mutex); | |
4a500fd1 YZ |
2754 | if (ret == -ENOSPC) { |
2755 | root->fs_info->last_trans_log_full_commit = trans->transid; | |
2756 | ret = 0; | |
79787eaa JM |
2757 | } else if (ret < 0 && ret != -ENOENT) |
2758 | btrfs_abort_transaction(trans, root, ret); | |
12fcfd22 | 2759 | btrfs_end_log_trans(root); |
e02119d5 | 2760 | |
e02119d5 CM |
2761 | return ret; |
2762 | } | |
2763 | ||
2764 | /* | |
2765 | * creates a range item in the log for 'dirid'. first_offset and | |
2766 | * last_offset tell us which parts of the key space the log should | |
2767 | * be considered authoritative for. | |
2768 | */ | |
2769 | static noinline int insert_dir_log_key(struct btrfs_trans_handle *trans, | |
2770 | struct btrfs_root *log, | |
2771 | struct btrfs_path *path, | |
2772 | int key_type, u64 dirid, | |
2773 | u64 first_offset, u64 last_offset) | |
2774 | { | |
2775 | int ret; | |
2776 | struct btrfs_key key; | |
2777 | struct btrfs_dir_log_item *item; | |
2778 | ||
2779 | key.objectid = dirid; | |
2780 | key.offset = first_offset; | |
2781 | if (key_type == BTRFS_DIR_ITEM_KEY) | |
2782 | key.type = BTRFS_DIR_LOG_ITEM_KEY; | |
2783 | else | |
2784 | key.type = BTRFS_DIR_LOG_INDEX_KEY; | |
2785 | ret = btrfs_insert_empty_item(trans, log, path, &key, sizeof(*item)); | |
4a500fd1 YZ |
2786 | if (ret) |
2787 | return ret; | |
e02119d5 CM |
2788 | |
2789 | item = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
2790 | struct btrfs_dir_log_item); | |
2791 | btrfs_set_dir_log_end(path->nodes[0], item, last_offset); | |
2792 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
b3b4aa74 | 2793 | btrfs_release_path(path); |
e02119d5 CM |
2794 | return 0; |
2795 | } | |
2796 | ||
2797 | /* | |
2798 | * log all the items included in the current transaction for a given | |
2799 | * directory. This also creates the range items in the log tree required | |
2800 | * to replay anything deleted before the fsync | |
2801 | */ | |
2802 | static noinline int log_dir_items(struct btrfs_trans_handle *trans, | |
2803 | struct btrfs_root *root, struct inode *inode, | |
2804 | struct btrfs_path *path, | |
2805 | struct btrfs_path *dst_path, int key_type, | |
2806 | u64 min_offset, u64 *last_offset_ret) | |
2807 | { | |
2808 | struct btrfs_key min_key; | |
2809 | struct btrfs_key max_key; | |
2810 | struct btrfs_root *log = root->log_root; | |
2811 | struct extent_buffer *src; | |
4a500fd1 | 2812 | int err = 0; |
e02119d5 CM |
2813 | int ret; |
2814 | int i; | |
2815 | int nritems; | |
2816 | u64 first_offset = min_offset; | |
2817 | u64 last_offset = (u64)-1; | |
33345d01 | 2818 | u64 ino = btrfs_ino(inode); |
e02119d5 CM |
2819 | |
2820 | log = root->log_root; | |
33345d01 | 2821 | max_key.objectid = ino; |
e02119d5 CM |
2822 | max_key.offset = (u64)-1; |
2823 | max_key.type = key_type; | |
2824 | ||
33345d01 | 2825 | min_key.objectid = ino; |
e02119d5 CM |
2826 | min_key.type = key_type; |
2827 | min_key.offset = min_offset; | |
2828 | ||
2829 | path->keep_locks = 1; | |
2830 | ||
2831 | ret = btrfs_search_forward(root, &min_key, &max_key, | |
de78b51a | 2832 | path, trans->transid); |
e02119d5 CM |
2833 | |
2834 | /* | |
2835 | * we didn't find anything from this transaction, see if there | |
2836 | * is anything at all | |
2837 | */ | |
33345d01 LZ |
2838 | if (ret != 0 || min_key.objectid != ino || min_key.type != key_type) { |
2839 | min_key.objectid = ino; | |
e02119d5 CM |
2840 | min_key.type = key_type; |
2841 | min_key.offset = (u64)-1; | |
b3b4aa74 | 2842 | btrfs_release_path(path); |
e02119d5 CM |
2843 | ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0); |
2844 | if (ret < 0) { | |
b3b4aa74 | 2845 | btrfs_release_path(path); |
e02119d5 CM |
2846 | return ret; |
2847 | } | |
33345d01 | 2848 | ret = btrfs_previous_item(root, path, ino, key_type); |
e02119d5 CM |
2849 | |
2850 | /* if ret == 0 there are items for this type, | |
2851 | * create a range to tell us the last key of this type. | |
2852 | * otherwise, there are no items in this directory after | |
2853 | * *min_offset, and we create a range to indicate that. | |
2854 | */ | |
2855 | if (ret == 0) { | |
2856 | struct btrfs_key tmp; | |
2857 | btrfs_item_key_to_cpu(path->nodes[0], &tmp, | |
2858 | path->slots[0]); | |
d397712b | 2859 | if (key_type == tmp.type) |
e02119d5 | 2860 | first_offset = max(min_offset, tmp.offset) + 1; |
e02119d5 CM |
2861 | } |
2862 | goto done; | |
2863 | } | |
2864 | ||
2865 | /* go backward to find any previous key */ | |
33345d01 | 2866 | ret = btrfs_previous_item(root, path, ino, key_type); |
e02119d5 CM |
2867 | if (ret == 0) { |
2868 | struct btrfs_key tmp; | |
2869 | btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]); | |
2870 | if (key_type == tmp.type) { | |
2871 | first_offset = tmp.offset; | |
2872 | ret = overwrite_item(trans, log, dst_path, | |
2873 | path->nodes[0], path->slots[0], | |
2874 | &tmp); | |
4a500fd1 YZ |
2875 | if (ret) { |
2876 | err = ret; | |
2877 | goto done; | |
2878 | } | |
e02119d5 CM |
2879 | } |
2880 | } | |
b3b4aa74 | 2881 | btrfs_release_path(path); |
e02119d5 CM |
2882 | |
2883 | /* find the first key from this transaction again */ | |
2884 | ret = btrfs_search_slot(NULL, root, &min_key, path, 0, 0); | |
2885 | if (ret != 0) { | |
2886 | WARN_ON(1); | |
2887 | goto done; | |
2888 | } | |
2889 | ||
2890 | /* | |
2891 | * we have a block from this transaction, log every item in it | |
2892 | * from our directory | |
2893 | */ | |
d397712b | 2894 | while (1) { |
e02119d5 CM |
2895 | struct btrfs_key tmp; |
2896 | src = path->nodes[0]; | |
2897 | nritems = btrfs_header_nritems(src); | |
2898 | for (i = path->slots[0]; i < nritems; i++) { | |
2899 | btrfs_item_key_to_cpu(src, &min_key, i); | |
2900 | ||
33345d01 | 2901 | if (min_key.objectid != ino || min_key.type != key_type) |
e02119d5 CM |
2902 | goto done; |
2903 | ret = overwrite_item(trans, log, dst_path, src, i, | |
2904 | &min_key); | |
4a500fd1 YZ |
2905 | if (ret) { |
2906 | err = ret; | |
2907 | goto done; | |
2908 | } | |
e02119d5 CM |
2909 | } |
2910 | path->slots[0] = nritems; | |
2911 | ||
2912 | /* | |
2913 | * look ahead to the next item and see if it is also | |
2914 | * from this directory and from this transaction | |
2915 | */ | |
2916 | ret = btrfs_next_leaf(root, path); | |
2917 | if (ret == 1) { | |
2918 | last_offset = (u64)-1; | |
2919 | goto done; | |
2920 | } | |
2921 | btrfs_item_key_to_cpu(path->nodes[0], &tmp, path->slots[0]); | |
33345d01 | 2922 | if (tmp.objectid != ino || tmp.type != key_type) { |
e02119d5 CM |
2923 | last_offset = (u64)-1; |
2924 | goto done; | |
2925 | } | |
2926 | if (btrfs_header_generation(path->nodes[0]) != trans->transid) { | |
2927 | ret = overwrite_item(trans, log, dst_path, | |
2928 | path->nodes[0], path->slots[0], | |
2929 | &tmp); | |
4a500fd1 YZ |
2930 | if (ret) |
2931 | err = ret; | |
2932 | else | |
2933 | last_offset = tmp.offset; | |
e02119d5 CM |
2934 | goto done; |
2935 | } | |
2936 | } | |
2937 | done: | |
b3b4aa74 DS |
2938 | btrfs_release_path(path); |
2939 | btrfs_release_path(dst_path); | |
e02119d5 | 2940 | |
4a500fd1 YZ |
2941 | if (err == 0) { |
2942 | *last_offset_ret = last_offset; | |
2943 | /* | |
2944 | * insert the log range keys to indicate where the log | |
2945 | * is valid | |
2946 | */ | |
2947 | ret = insert_dir_log_key(trans, log, path, key_type, | |
33345d01 | 2948 | ino, first_offset, last_offset); |
4a500fd1 YZ |
2949 | if (ret) |
2950 | err = ret; | |
2951 | } | |
2952 | return err; | |
e02119d5 CM |
2953 | } |
2954 | ||
2955 | /* | |
2956 | * logging directories is very similar to logging inodes, We find all the items | |
2957 | * from the current transaction and write them to the log. | |
2958 | * | |
2959 | * The recovery code scans the directory in the subvolume, and if it finds a | |
2960 | * key in the range logged that is not present in the log tree, then it means | |
2961 | * that dir entry was unlinked during the transaction. | |
2962 | * | |
2963 | * In order for that scan to work, we must include one key smaller than | |
2964 | * the smallest logged by this transaction and one key larger than the largest | |
2965 | * key logged by this transaction. | |
2966 | */ | |
2967 | static noinline int log_directory_changes(struct btrfs_trans_handle *trans, | |
2968 | struct btrfs_root *root, struct inode *inode, | |
2969 | struct btrfs_path *path, | |
2970 | struct btrfs_path *dst_path) | |
2971 | { | |
2972 | u64 min_key; | |
2973 | u64 max_key; | |
2974 | int ret; | |
2975 | int key_type = BTRFS_DIR_ITEM_KEY; | |
2976 | ||
2977 | again: | |
2978 | min_key = 0; | |
2979 | max_key = 0; | |
d397712b | 2980 | while (1) { |
e02119d5 CM |
2981 | ret = log_dir_items(trans, root, inode, path, |
2982 | dst_path, key_type, min_key, | |
2983 | &max_key); | |
4a500fd1 YZ |
2984 | if (ret) |
2985 | return ret; | |
e02119d5 CM |
2986 | if (max_key == (u64)-1) |
2987 | break; | |
2988 | min_key = max_key + 1; | |
2989 | } | |
2990 | ||
2991 | if (key_type == BTRFS_DIR_ITEM_KEY) { | |
2992 | key_type = BTRFS_DIR_INDEX_KEY; | |
2993 | goto again; | |
2994 | } | |
2995 | return 0; | |
2996 | } | |
2997 | ||
2998 | /* | |
2999 | * a helper function to drop items from the log before we relog an | |
3000 | * inode. max_key_type indicates the highest item type to remove. | |
3001 | * This cannot be run for file data extents because it does not | |
3002 | * free the extents they point to. | |
3003 | */ | |
3004 | static int drop_objectid_items(struct btrfs_trans_handle *trans, | |
3005 | struct btrfs_root *log, | |
3006 | struct btrfs_path *path, | |
3007 | u64 objectid, int max_key_type) | |
3008 | { | |
3009 | int ret; | |
3010 | struct btrfs_key key; | |
3011 | struct btrfs_key found_key; | |
18ec90d6 | 3012 | int start_slot; |
e02119d5 CM |
3013 | |
3014 | key.objectid = objectid; | |
3015 | key.type = max_key_type; | |
3016 | key.offset = (u64)-1; | |
3017 | ||
d397712b | 3018 | while (1) { |
e02119d5 | 3019 | ret = btrfs_search_slot(trans, log, &key, path, -1, 1); |
3650860b | 3020 | BUG_ON(ret == 0); /* Logic error */ |
4a500fd1 | 3021 | if (ret < 0) |
e02119d5 CM |
3022 | break; |
3023 | ||
3024 | if (path->slots[0] == 0) | |
3025 | break; | |
3026 | ||
3027 | path->slots[0]--; | |
3028 | btrfs_item_key_to_cpu(path->nodes[0], &found_key, | |
3029 | path->slots[0]); | |
3030 | ||
3031 | if (found_key.objectid != objectid) | |
3032 | break; | |
3033 | ||
18ec90d6 JB |
3034 | found_key.offset = 0; |
3035 | found_key.type = 0; | |
3036 | ret = btrfs_bin_search(path->nodes[0], &found_key, 0, | |
3037 | &start_slot); | |
3038 | ||
3039 | ret = btrfs_del_items(trans, log, path, start_slot, | |
3040 | path->slots[0] - start_slot + 1); | |
3041 | /* | |
3042 | * If start slot isn't 0 then we don't need to re-search, we've | |
3043 | * found the last guy with the objectid in this tree. | |
3044 | */ | |
3045 | if (ret || start_slot != 0) | |
65a246c5 | 3046 | break; |
b3b4aa74 | 3047 | btrfs_release_path(path); |
e02119d5 | 3048 | } |
b3b4aa74 | 3049 | btrfs_release_path(path); |
5bdbeb21 JB |
3050 | if (ret > 0) |
3051 | ret = 0; | |
4a500fd1 | 3052 | return ret; |
e02119d5 CM |
3053 | } |
3054 | ||
94edf4ae JB |
3055 | static void fill_inode_item(struct btrfs_trans_handle *trans, |
3056 | struct extent_buffer *leaf, | |
3057 | struct btrfs_inode_item *item, | |
3058 | struct inode *inode, int log_inode_only) | |
3059 | { | |
0b1c6cca JB |
3060 | struct btrfs_map_token token; |
3061 | ||
3062 | btrfs_init_map_token(&token); | |
94edf4ae JB |
3063 | |
3064 | if (log_inode_only) { | |
3065 | /* set the generation to zero so the recover code | |
3066 | * can tell the difference between an logging | |
3067 | * just to say 'this inode exists' and a logging | |
3068 | * to say 'update this inode with these values' | |
3069 | */ | |
0b1c6cca JB |
3070 | btrfs_set_token_inode_generation(leaf, item, 0, &token); |
3071 | btrfs_set_token_inode_size(leaf, item, 0, &token); | |
94edf4ae | 3072 | } else { |
0b1c6cca JB |
3073 | btrfs_set_token_inode_generation(leaf, item, |
3074 | BTRFS_I(inode)->generation, | |
3075 | &token); | |
3076 | btrfs_set_token_inode_size(leaf, item, inode->i_size, &token); | |
3077 | } | |
3078 | ||
3079 | btrfs_set_token_inode_uid(leaf, item, i_uid_read(inode), &token); | |
3080 | btrfs_set_token_inode_gid(leaf, item, i_gid_read(inode), &token); | |
3081 | btrfs_set_token_inode_mode(leaf, item, inode->i_mode, &token); | |
3082 | btrfs_set_token_inode_nlink(leaf, item, inode->i_nlink, &token); | |
3083 | ||
3084 | btrfs_set_token_timespec_sec(leaf, btrfs_inode_atime(item), | |
3085 | inode->i_atime.tv_sec, &token); | |
3086 | btrfs_set_token_timespec_nsec(leaf, btrfs_inode_atime(item), | |
3087 | inode->i_atime.tv_nsec, &token); | |
3088 | ||
3089 | btrfs_set_token_timespec_sec(leaf, btrfs_inode_mtime(item), | |
3090 | inode->i_mtime.tv_sec, &token); | |
3091 | btrfs_set_token_timespec_nsec(leaf, btrfs_inode_mtime(item), | |
3092 | inode->i_mtime.tv_nsec, &token); | |
3093 | ||
3094 | btrfs_set_token_timespec_sec(leaf, btrfs_inode_ctime(item), | |
3095 | inode->i_ctime.tv_sec, &token); | |
3096 | btrfs_set_token_timespec_nsec(leaf, btrfs_inode_ctime(item), | |
3097 | inode->i_ctime.tv_nsec, &token); | |
3098 | ||
3099 | btrfs_set_token_inode_nbytes(leaf, item, inode_get_bytes(inode), | |
3100 | &token); | |
3101 | ||
3102 | btrfs_set_token_inode_sequence(leaf, item, inode->i_version, &token); | |
3103 | btrfs_set_token_inode_transid(leaf, item, trans->transid, &token); | |
3104 | btrfs_set_token_inode_rdev(leaf, item, inode->i_rdev, &token); | |
3105 | btrfs_set_token_inode_flags(leaf, item, BTRFS_I(inode)->flags, &token); | |
3106 | btrfs_set_token_inode_block_group(leaf, item, 0, &token); | |
94edf4ae JB |
3107 | } |
3108 | ||
a95249b3 JB |
3109 | static int log_inode_item(struct btrfs_trans_handle *trans, |
3110 | struct btrfs_root *log, struct btrfs_path *path, | |
3111 | struct inode *inode) | |
3112 | { | |
3113 | struct btrfs_inode_item *inode_item; | |
3114 | struct btrfs_key key; | |
3115 | int ret; | |
3116 | ||
3117 | memcpy(&key, &BTRFS_I(inode)->location, sizeof(key)); | |
3118 | ret = btrfs_insert_empty_item(trans, log, path, &key, | |
3119 | sizeof(*inode_item)); | |
3120 | if (ret && ret != -EEXIST) | |
3121 | return ret; | |
3122 | inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0], | |
3123 | struct btrfs_inode_item); | |
3124 | fill_inode_item(trans, path->nodes[0], inode_item, inode, 0); | |
3125 | btrfs_release_path(path); | |
3126 | return 0; | |
3127 | } | |
3128 | ||
31ff1cd2 | 3129 | static noinline int copy_items(struct btrfs_trans_handle *trans, |
d2794405 | 3130 | struct inode *inode, |
31ff1cd2 CM |
3131 | struct btrfs_path *dst_path, |
3132 | struct extent_buffer *src, | |
3133 | int start_slot, int nr, int inode_only) | |
3134 | { | |
3135 | unsigned long src_offset; | |
3136 | unsigned long dst_offset; | |
d2794405 | 3137 | struct btrfs_root *log = BTRFS_I(inode)->root->log_root; |
31ff1cd2 CM |
3138 | struct btrfs_file_extent_item *extent; |
3139 | struct btrfs_inode_item *inode_item; | |
3140 | int ret; | |
3141 | struct btrfs_key *ins_keys; | |
3142 | u32 *ins_sizes; | |
3143 | char *ins_data; | |
3144 | int i; | |
d20f7043 | 3145 | struct list_head ordered_sums; |
d2794405 | 3146 | int skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM; |
d20f7043 CM |
3147 | |
3148 | INIT_LIST_HEAD(&ordered_sums); | |
31ff1cd2 CM |
3149 | |
3150 | ins_data = kmalloc(nr * sizeof(struct btrfs_key) + | |
3151 | nr * sizeof(u32), GFP_NOFS); | |
2a29edc6 | 3152 | if (!ins_data) |
3153 | return -ENOMEM; | |
3154 | ||
31ff1cd2 CM |
3155 | ins_sizes = (u32 *)ins_data; |
3156 | ins_keys = (struct btrfs_key *)(ins_data + nr * sizeof(u32)); | |
3157 | ||
3158 | for (i = 0; i < nr; i++) { | |
3159 | ins_sizes[i] = btrfs_item_size_nr(src, i + start_slot); | |
3160 | btrfs_item_key_to_cpu(src, ins_keys + i, i + start_slot); | |
3161 | } | |
3162 | ret = btrfs_insert_empty_items(trans, log, dst_path, | |
3163 | ins_keys, ins_sizes, nr); | |
4a500fd1 YZ |
3164 | if (ret) { |
3165 | kfree(ins_data); | |
3166 | return ret; | |
3167 | } | |
31ff1cd2 | 3168 | |
5d4f98a2 | 3169 | for (i = 0; i < nr; i++, dst_path->slots[0]++) { |
31ff1cd2 CM |
3170 | dst_offset = btrfs_item_ptr_offset(dst_path->nodes[0], |
3171 | dst_path->slots[0]); | |
3172 | ||
3173 | src_offset = btrfs_item_ptr_offset(src, start_slot + i); | |
3174 | ||
94edf4ae | 3175 | if (ins_keys[i].type == BTRFS_INODE_ITEM_KEY) { |
31ff1cd2 CM |
3176 | inode_item = btrfs_item_ptr(dst_path->nodes[0], |
3177 | dst_path->slots[0], | |
3178 | struct btrfs_inode_item); | |
94edf4ae JB |
3179 | fill_inode_item(trans, dst_path->nodes[0], inode_item, |
3180 | inode, inode_only == LOG_INODE_EXISTS); | |
3181 | } else { | |
3182 | copy_extent_buffer(dst_path->nodes[0], src, dst_offset, | |
3183 | src_offset, ins_sizes[i]); | |
31ff1cd2 | 3184 | } |
94edf4ae | 3185 | |
31ff1cd2 CM |
3186 | /* take a reference on file data extents so that truncates |
3187 | * or deletes of this inode don't have to relog the inode | |
3188 | * again | |
3189 | */ | |
d2794405 LB |
3190 | if (btrfs_key_type(ins_keys + i) == BTRFS_EXTENT_DATA_KEY && |
3191 | !skip_csum) { | |
31ff1cd2 CM |
3192 | int found_type; |
3193 | extent = btrfs_item_ptr(src, start_slot + i, | |
3194 | struct btrfs_file_extent_item); | |
3195 | ||
8e531cdf | 3196 | if (btrfs_file_extent_generation(src, extent) < trans->transid) |
3197 | continue; | |
3198 | ||
31ff1cd2 | 3199 | found_type = btrfs_file_extent_type(src, extent); |
6f1fed77 | 3200 | if (found_type == BTRFS_FILE_EXTENT_REG) { |
5d4f98a2 YZ |
3201 | u64 ds, dl, cs, cl; |
3202 | ds = btrfs_file_extent_disk_bytenr(src, | |
3203 | extent); | |
3204 | /* ds == 0 is a hole */ | |
3205 | if (ds == 0) | |
3206 | continue; | |
3207 | ||
3208 | dl = btrfs_file_extent_disk_num_bytes(src, | |
3209 | extent); | |
3210 | cs = btrfs_file_extent_offset(src, extent); | |
3211 | cl = btrfs_file_extent_num_bytes(src, | |
a419aef8 | 3212 | extent); |
580afd76 CM |
3213 | if (btrfs_file_extent_compression(src, |
3214 | extent)) { | |
3215 | cs = 0; | |
3216 | cl = dl; | |
3217 | } | |
5d4f98a2 YZ |
3218 | |
3219 | ret = btrfs_lookup_csums_range( | |
3220 | log->fs_info->csum_root, | |
3221 | ds + cs, ds + cs + cl - 1, | |
a2de733c | 3222 | &ordered_sums, 0); |
3650860b JB |
3223 | if (ret) { |
3224 | btrfs_release_path(dst_path); | |
3225 | kfree(ins_data); | |
3226 | return ret; | |
3227 | } | |
31ff1cd2 CM |
3228 | } |
3229 | } | |
31ff1cd2 CM |
3230 | } |
3231 | ||
3232 | btrfs_mark_buffer_dirty(dst_path->nodes[0]); | |
b3b4aa74 | 3233 | btrfs_release_path(dst_path); |
31ff1cd2 | 3234 | kfree(ins_data); |
d20f7043 CM |
3235 | |
3236 | /* | |
3237 | * we have to do this after the loop above to avoid changing the | |
3238 | * log tree while trying to change the log tree. | |
3239 | */ | |
4a500fd1 | 3240 | ret = 0; |
d397712b | 3241 | while (!list_empty(&ordered_sums)) { |
d20f7043 CM |
3242 | struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next, |
3243 | struct btrfs_ordered_sum, | |
3244 | list); | |
4a500fd1 YZ |
3245 | if (!ret) |
3246 | ret = btrfs_csum_file_blocks(trans, log, sums); | |
d20f7043 CM |
3247 | list_del(&sums->list); |
3248 | kfree(sums); | |
3249 | } | |
4a500fd1 | 3250 | return ret; |
31ff1cd2 CM |
3251 | } |
3252 | ||
5dc562c5 JB |
3253 | static int extent_cmp(void *priv, struct list_head *a, struct list_head *b) |
3254 | { | |
3255 | struct extent_map *em1, *em2; | |
3256 | ||
3257 | em1 = list_entry(a, struct extent_map, list); | |
3258 | em2 = list_entry(b, struct extent_map, list); | |
3259 | ||
3260 | if (em1->start < em2->start) | |
3261 | return -1; | |
3262 | else if (em1->start > em2->start) | |
3263 | return 1; | |
3264 | return 0; | |
3265 | } | |
3266 | ||
5dc562c5 JB |
3267 | static int log_one_extent(struct btrfs_trans_handle *trans, |
3268 | struct inode *inode, struct btrfs_root *root, | |
70c8a91c | 3269 | struct extent_map *em, struct btrfs_path *path) |
5dc562c5 JB |
3270 | { |
3271 | struct btrfs_root *log = root->log_root; | |
70c8a91c JB |
3272 | struct btrfs_file_extent_item *fi; |
3273 | struct extent_buffer *leaf; | |
2ab28f32 | 3274 | struct btrfs_ordered_extent *ordered; |
70c8a91c | 3275 | struct list_head ordered_sums; |
0b1c6cca | 3276 | struct btrfs_map_token token; |
5dc562c5 | 3277 | struct btrfs_key key; |
2ab28f32 JB |
3278 | u64 mod_start = em->mod_start; |
3279 | u64 mod_len = em->mod_len; | |
3280 | u64 csum_offset; | |
3281 | u64 csum_len; | |
70c8a91c JB |
3282 | u64 extent_offset = em->start - em->orig_start; |
3283 | u64 block_len; | |
5dc562c5 | 3284 | int ret; |
2ab28f32 | 3285 | int index = log->log_transid % 2; |
70c8a91c | 3286 | bool skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM; |
5dc562c5 | 3287 | |
09a2a8f9 JB |
3288 | ret = __btrfs_drop_extents(trans, log, inode, path, em->start, |
3289 | em->start + em->len, NULL, 0); | |
3290 | if (ret) | |
3291 | return ret; | |
3292 | ||
70c8a91c | 3293 | INIT_LIST_HEAD(&ordered_sums); |
0b1c6cca | 3294 | btrfs_init_map_token(&token); |
70c8a91c JB |
3295 | key.objectid = btrfs_ino(inode); |
3296 | key.type = BTRFS_EXTENT_DATA_KEY; | |
3297 | key.offset = em->start; | |
70c8a91c JB |
3298 | |
3299 | ret = btrfs_insert_empty_item(trans, log, path, &key, sizeof(*fi)); | |
09a2a8f9 | 3300 | if (ret) |
70c8a91c | 3301 | return ret; |
70c8a91c JB |
3302 | leaf = path->nodes[0]; |
3303 | fi = btrfs_item_ptr(leaf, path->slots[0], | |
3304 | struct btrfs_file_extent_item); | |
124fe663 | 3305 | |
0b1c6cca JB |
3306 | btrfs_set_token_file_extent_generation(leaf, fi, em->generation, |
3307 | &token); | |
70c8a91c JB |
3308 | if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) { |
3309 | skip_csum = true; | |
0b1c6cca JB |
3310 | btrfs_set_token_file_extent_type(leaf, fi, |
3311 | BTRFS_FILE_EXTENT_PREALLOC, | |
3312 | &token); | |
70c8a91c | 3313 | } else { |
0b1c6cca JB |
3314 | btrfs_set_token_file_extent_type(leaf, fi, |
3315 | BTRFS_FILE_EXTENT_REG, | |
3316 | &token); | |
70c8a91c JB |
3317 | if (em->block_start == 0) |
3318 | skip_csum = true; | |
3319 | } | |
3320 | ||
3321 | block_len = max(em->block_len, em->orig_block_len); | |
3322 | if (em->compress_type != BTRFS_COMPRESS_NONE) { | |
0b1c6cca JB |
3323 | btrfs_set_token_file_extent_disk_bytenr(leaf, fi, |
3324 | em->block_start, | |
3325 | &token); | |
3326 | btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len, | |
3327 | &token); | |
70c8a91c | 3328 | } else if (em->block_start < EXTENT_MAP_LAST_BYTE) { |
0b1c6cca JB |
3329 | btrfs_set_token_file_extent_disk_bytenr(leaf, fi, |
3330 | em->block_start - | |
3331 | extent_offset, &token); | |
3332 | btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, block_len, | |
3333 | &token); | |
70c8a91c | 3334 | } else { |
0b1c6cca JB |
3335 | btrfs_set_token_file_extent_disk_bytenr(leaf, fi, 0, &token); |
3336 | btrfs_set_token_file_extent_disk_num_bytes(leaf, fi, 0, | |
3337 | &token); | |
3338 | } | |
3339 | ||
3340 | btrfs_set_token_file_extent_offset(leaf, fi, | |
3341 | em->start - em->orig_start, | |
3342 | &token); | |
3343 | btrfs_set_token_file_extent_num_bytes(leaf, fi, em->len, &token); | |
cc95bef6 | 3344 | btrfs_set_token_file_extent_ram_bytes(leaf, fi, em->ram_bytes, &token); |
0b1c6cca JB |
3345 | btrfs_set_token_file_extent_compression(leaf, fi, em->compress_type, |
3346 | &token); | |
3347 | btrfs_set_token_file_extent_encryption(leaf, fi, 0, &token); | |
3348 | btrfs_set_token_file_extent_other_encoding(leaf, fi, 0, &token); | |
70c8a91c | 3349 | btrfs_mark_buffer_dirty(leaf); |
0aa4a17d | 3350 | |
70c8a91c | 3351 | btrfs_release_path(path); |
70c8a91c JB |
3352 | if (ret) { |
3353 | return ret; | |
3354 | } | |
0aa4a17d | 3355 | |
70c8a91c JB |
3356 | if (skip_csum) |
3357 | return 0; | |
5dc562c5 | 3358 | |
192000dd LB |
3359 | if (em->compress_type) { |
3360 | csum_offset = 0; | |
3361 | csum_len = block_len; | |
3362 | } | |
3363 | ||
2ab28f32 JB |
3364 | /* |
3365 | * First check and see if our csums are on our outstanding ordered | |
3366 | * extents. | |
3367 | */ | |
3368 | again: | |
3369 | spin_lock_irq(&log->log_extents_lock[index]); | |
3370 | list_for_each_entry(ordered, &log->logged_list[index], log_list) { | |
3371 | struct btrfs_ordered_sum *sum; | |
3372 | ||
3373 | if (!mod_len) | |
3374 | break; | |
3375 | ||
3376 | if (ordered->inode != inode) | |
3377 | continue; | |
3378 | ||
3379 | if (ordered->file_offset + ordered->len <= mod_start || | |
3380 | mod_start + mod_len <= ordered->file_offset) | |
3381 | continue; | |
3382 | ||
3383 | /* | |
3384 | * We are going to copy all the csums on this ordered extent, so | |
3385 | * go ahead and adjust mod_start and mod_len in case this | |
3386 | * ordered extent has already been logged. | |
3387 | */ | |
3388 | if (ordered->file_offset > mod_start) { | |
3389 | if (ordered->file_offset + ordered->len >= | |
3390 | mod_start + mod_len) | |
3391 | mod_len = ordered->file_offset - mod_start; | |
3392 | /* | |
3393 | * If we have this case | |
3394 | * | |
3395 | * |--------- logged extent ---------| | |
3396 | * |----- ordered extent ----| | |
3397 | * | |
3398 | * Just don't mess with mod_start and mod_len, we'll | |
3399 | * just end up logging more csums than we need and it | |
3400 | * will be ok. | |
3401 | */ | |
3402 | } else { | |
3403 | if (ordered->file_offset + ordered->len < | |
3404 | mod_start + mod_len) { | |
3405 | mod_len = (mod_start + mod_len) - | |
3406 | (ordered->file_offset + ordered->len); | |
3407 | mod_start = ordered->file_offset + | |
3408 | ordered->len; | |
3409 | } else { | |
3410 | mod_len = 0; | |
3411 | } | |
3412 | } | |
3413 | ||
3414 | /* | |
3415 | * To keep us from looping for the above case of an ordered | |
3416 | * extent that falls inside of the logged extent. | |
3417 | */ | |
3418 | if (test_and_set_bit(BTRFS_ORDERED_LOGGED_CSUM, | |
3419 | &ordered->flags)) | |
3420 | continue; | |
3421 | atomic_inc(&ordered->refs); | |
3422 | spin_unlock_irq(&log->log_extents_lock[index]); | |
3423 | /* | |
3424 | * we've dropped the lock, we must either break or | |
3425 | * start over after this. | |
3426 | */ | |
3427 | ||
3428 | wait_event(ordered->wait, ordered->csum_bytes_left == 0); | |
3429 | ||
3430 | list_for_each_entry(sum, &ordered->list, list) { | |
3431 | ret = btrfs_csum_file_blocks(trans, log, sum); | |
3432 | if (ret) { | |
3433 | btrfs_put_ordered_extent(ordered); | |
3434 | goto unlocked; | |
3435 | } | |
3436 | } | |
3437 | btrfs_put_ordered_extent(ordered); | |
3438 | goto again; | |
3439 | ||
3440 | } | |
3441 | spin_unlock_irq(&log->log_extents_lock[index]); | |
3442 | unlocked: | |
3443 | ||
3444 | if (!mod_len || ret) | |
3445 | return ret; | |
3446 | ||
3447 | csum_offset = mod_start - em->start; | |
3448 | csum_len = mod_len; | |
3449 | ||
70c8a91c JB |
3450 | /* block start is already adjusted for the file extent offset. */ |
3451 | ret = btrfs_lookup_csums_range(log->fs_info->csum_root, | |
3452 | em->block_start + csum_offset, | |
3453 | em->block_start + csum_offset + | |
3454 | csum_len - 1, &ordered_sums, 0); | |
3455 | if (ret) | |
3456 | return ret; | |
5dc562c5 | 3457 | |
70c8a91c JB |
3458 | while (!list_empty(&ordered_sums)) { |
3459 | struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next, | |
3460 | struct btrfs_ordered_sum, | |
3461 | list); | |
3462 | if (!ret) | |
3463 | ret = btrfs_csum_file_blocks(trans, log, sums); | |
3464 | list_del(&sums->list); | |
3465 | kfree(sums); | |
5dc562c5 JB |
3466 | } |
3467 | ||
70c8a91c | 3468 | return ret; |
5dc562c5 JB |
3469 | } |
3470 | ||
3471 | static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans, | |
3472 | struct btrfs_root *root, | |
3473 | struct inode *inode, | |
70c8a91c | 3474 | struct btrfs_path *path) |
5dc562c5 | 3475 | { |
5dc562c5 JB |
3476 | struct extent_map *em, *n; |
3477 | struct list_head extents; | |
3478 | struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree; | |
3479 | u64 test_gen; | |
3480 | int ret = 0; | |
2ab28f32 | 3481 | int num = 0; |
5dc562c5 JB |
3482 | |
3483 | INIT_LIST_HEAD(&extents); | |
3484 | ||
5dc562c5 JB |
3485 | write_lock(&tree->lock); |
3486 | test_gen = root->fs_info->last_trans_committed; | |
3487 | ||
3488 | list_for_each_entry_safe(em, n, &tree->modified_extents, list) { | |
3489 | list_del_init(&em->list); | |
2ab28f32 JB |
3490 | |
3491 | /* | |
3492 | * Just an arbitrary number, this can be really CPU intensive | |
3493 | * once we start getting a lot of extents, and really once we | |
3494 | * have a bunch of extents we just want to commit since it will | |
3495 | * be faster. | |
3496 | */ | |
3497 | if (++num > 32768) { | |
3498 | list_del_init(&tree->modified_extents); | |
3499 | ret = -EFBIG; | |
3500 | goto process; | |
3501 | } | |
3502 | ||
5dc562c5 JB |
3503 | if (em->generation <= test_gen) |
3504 | continue; | |
ff44c6e3 JB |
3505 | /* Need a ref to keep it from getting evicted from cache */ |
3506 | atomic_inc(&em->refs); | |
3507 | set_bit(EXTENT_FLAG_LOGGING, &em->flags); | |
5dc562c5 | 3508 | list_add_tail(&em->list, &extents); |
2ab28f32 | 3509 | num++; |
5dc562c5 JB |
3510 | } |
3511 | ||
3512 | list_sort(NULL, &extents, extent_cmp); | |
3513 | ||
2ab28f32 | 3514 | process: |
5dc562c5 JB |
3515 | while (!list_empty(&extents)) { |
3516 | em = list_entry(extents.next, struct extent_map, list); | |
3517 | ||
3518 | list_del_init(&em->list); | |
3519 | ||
3520 | /* | |
3521 | * If we had an error we just need to delete everybody from our | |
3522 | * private list. | |
3523 | */ | |
ff44c6e3 | 3524 | if (ret) { |
201a9038 | 3525 | clear_em_logging(tree, em); |
ff44c6e3 | 3526 | free_extent_map(em); |
5dc562c5 | 3527 | continue; |
ff44c6e3 JB |
3528 | } |
3529 | ||
3530 | write_unlock(&tree->lock); | |
5dc562c5 | 3531 | |
70c8a91c | 3532 | ret = log_one_extent(trans, inode, root, em, path); |
ff44c6e3 | 3533 | write_lock(&tree->lock); |
201a9038 JB |
3534 | clear_em_logging(tree, em); |
3535 | free_extent_map(em); | |
5dc562c5 | 3536 | } |
ff44c6e3 JB |
3537 | WARN_ON(!list_empty(&extents)); |
3538 | write_unlock(&tree->lock); | |
5dc562c5 | 3539 | |
5dc562c5 | 3540 | btrfs_release_path(path); |
5dc562c5 JB |
3541 | return ret; |
3542 | } | |
3543 | ||
e02119d5 CM |
3544 | /* log a single inode in the tree log. |
3545 | * At least one parent directory for this inode must exist in the tree | |
3546 | * or be logged already. | |
3547 | * | |
3548 | * Any items from this inode changed by the current transaction are copied | |
3549 | * to the log tree. An extra reference is taken on any extents in this | |
3550 | * file, allowing us to avoid a whole pile of corner cases around logging | |
3551 | * blocks that have been removed from the tree. | |
3552 | * | |
3553 | * See LOG_INODE_ALL and related defines for a description of what inode_only | |
3554 | * does. | |
3555 | * | |
3556 | * This handles both files and directories. | |
3557 | */ | |
12fcfd22 | 3558 | static int btrfs_log_inode(struct btrfs_trans_handle *trans, |
e02119d5 CM |
3559 | struct btrfs_root *root, struct inode *inode, |
3560 | int inode_only) | |
3561 | { | |
3562 | struct btrfs_path *path; | |
3563 | struct btrfs_path *dst_path; | |
3564 | struct btrfs_key min_key; | |
3565 | struct btrfs_key max_key; | |
3566 | struct btrfs_root *log = root->log_root; | |
31ff1cd2 | 3567 | struct extent_buffer *src = NULL; |
4a500fd1 | 3568 | int err = 0; |
e02119d5 | 3569 | int ret; |
3a5f1d45 | 3570 | int nritems; |
31ff1cd2 CM |
3571 | int ins_start_slot = 0; |
3572 | int ins_nr; | |
5dc562c5 | 3573 | bool fast_search = false; |
33345d01 | 3574 | u64 ino = btrfs_ino(inode); |
e02119d5 | 3575 | |
e02119d5 | 3576 | path = btrfs_alloc_path(); |
5df67083 TI |
3577 | if (!path) |
3578 | return -ENOMEM; | |
e02119d5 | 3579 | dst_path = btrfs_alloc_path(); |
5df67083 TI |
3580 | if (!dst_path) { |
3581 | btrfs_free_path(path); | |
3582 | return -ENOMEM; | |
3583 | } | |
e02119d5 | 3584 | |
33345d01 | 3585 | min_key.objectid = ino; |
e02119d5 CM |
3586 | min_key.type = BTRFS_INODE_ITEM_KEY; |
3587 | min_key.offset = 0; | |
3588 | ||
33345d01 | 3589 | max_key.objectid = ino; |
12fcfd22 | 3590 | |
12fcfd22 | 3591 | |
5dc562c5 | 3592 | /* today the code can only do partial logging of directories */ |
5269b67e MX |
3593 | if (S_ISDIR(inode->i_mode) || |
3594 | (!test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, | |
3595 | &BTRFS_I(inode)->runtime_flags) && | |
3596 | inode_only == LOG_INODE_EXISTS)) | |
e02119d5 CM |
3597 | max_key.type = BTRFS_XATTR_ITEM_KEY; |
3598 | else | |
3599 | max_key.type = (u8)-1; | |
3600 | max_key.offset = (u64)-1; | |
3601 | ||
94edf4ae JB |
3602 | /* Only run delayed items if we are a dir or a new file */ |
3603 | if (S_ISDIR(inode->i_mode) || | |
3604 | BTRFS_I(inode)->generation > root->fs_info->last_trans_committed) { | |
3605 | ret = btrfs_commit_inode_delayed_items(trans, inode); | |
3606 | if (ret) { | |
3607 | btrfs_free_path(path); | |
3608 | btrfs_free_path(dst_path); | |
3609 | return ret; | |
3610 | } | |
16cdcec7 MX |
3611 | } |
3612 | ||
e02119d5 CM |
3613 | mutex_lock(&BTRFS_I(inode)->log_mutex); |
3614 | ||
2ab28f32 JB |
3615 | btrfs_get_logged_extents(log, inode); |
3616 | ||
e02119d5 CM |
3617 | /* |
3618 | * a brute force approach to making sure we get the most uptodate | |
3619 | * copies of everything. | |
3620 | */ | |
3621 | if (S_ISDIR(inode->i_mode)) { | |
3622 | int max_key_type = BTRFS_DIR_LOG_INDEX_KEY; | |
3623 | ||
3624 | if (inode_only == LOG_INODE_EXISTS) | |
3625 | max_key_type = BTRFS_XATTR_ITEM_KEY; | |
33345d01 | 3626 | ret = drop_objectid_items(trans, log, path, ino, max_key_type); |
e02119d5 | 3627 | } else { |
5dc562c5 JB |
3628 | if (test_and_clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC, |
3629 | &BTRFS_I(inode)->runtime_flags)) { | |
e9976151 JB |
3630 | clear_bit(BTRFS_INODE_COPY_EVERYTHING, |
3631 | &BTRFS_I(inode)->runtime_flags); | |
5dc562c5 JB |
3632 | ret = btrfs_truncate_inode_items(trans, log, |
3633 | inode, 0, 0); | |
a95249b3 JB |
3634 | } else if (test_and_clear_bit(BTRFS_INODE_COPY_EVERYTHING, |
3635 | &BTRFS_I(inode)->runtime_flags)) { | |
183f37fa LB |
3636 | if (inode_only == LOG_INODE_ALL) |
3637 | fast_search = true; | |
a95249b3 | 3638 | max_key.type = BTRFS_XATTR_ITEM_KEY; |
5dc562c5 | 3639 | ret = drop_objectid_items(trans, log, path, ino, |
e9976151 | 3640 | max_key.type); |
a95249b3 JB |
3641 | } else { |
3642 | if (inode_only == LOG_INODE_ALL) | |
3643 | fast_search = true; | |
3644 | ret = log_inode_item(trans, log, dst_path, inode); | |
3645 | if (ret) { | |
3646 | err = ret; | |
3647 | goto out_unlock; | |
3648 | } | |
3649 | goto log_extents; | |
5dc562c5 | 3650 | } |
a95249b3 | 3651 | |
e02119d5 | 3652 | } |
4a500fd1 YZ |
3653 | if (ret) { |
3654 | err = ret; | |
3655 | goto out_unlock; | |
3656 | } | |
e02119d5 CM |
3657 | path->keep_locks = 1; |
3658 | ||
d397712b | 3659 | while (1) { |
31ff1cd2 | 3660 | ins_nr = 0; |
e02119d5 | 3661 | ret = btrfs_search_forward(root, &min_key, &max_key, |
de78b51a | 3662 | path, trans->transid); |
e02119d5 CM |
3663 | if (ret != 0) |
3664 | break; | |
3a5f1d45 | 3665 | again: |
31ff1cd2 | 3666 | /* note, ins_nr might be > 0 here, cleanup outside the loop */ |
33345d01 | 3667 | if (min_key.objectid != ino) |
e02119d5 CM |
3668 | break; |
3669 | if (min_key.type > max_key.type) | |
3670 | break; | |
31ff1cd2 | 3671 | |
e02119d5 | 3672 | src = path->nodes[0]; |
31ff1cd2 CM |
3673 | if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) { |
3674 | ins_nr++; | |
3675 | goto next_slot; | |
3676 | } else if (!ins_nr) { | |
3677 | ins_start_slot = path->slots[0]; | |
3678 | ins_nr = 1; | |
3679 | goto next_slot; | |
e02119d5 CM |
3680 | } |
3681 | ||
d2794405 | 3682 | ret = copy_items(trans, inode, dst_path, src, ins_start_slot, |
31ff1cd2 | 3683 | ins_nr, inode_only); |
4a500fd1 YZ |
3684 | if (ret) { |
3685 | err = ret; | |
3686 | goto out_unlock; | |
3687 | } | |
31ff1cd2 CM |
3688 | ins_nr = 1; |
3689 | ins_start_slot = path->slots[0]; | |
3690 | next_slot: | |
e02119d5 | 3691 | |
3a5f1d45 CM |
3692 | nritems = btrfs_header_nritems(path->nodes[0]); |
3693 | path->slots[0]++; | |
3694 | if (path->slots[0] < nritems) { | |
3695 | btrfs_item_key_to_cpu(path->nodes[0], &min_key, | |
3696 | path->slots[0]); | |
3697 | goto again; | |
3698 | } | |
31ff1cd2 | 3699 | if (ins_nr) { |
d2794405 | 3700 | ret = copy_items(trans, inode, dst_path, src, |
31ff1cd2 CM |
3701 | ins_start_slot, |
3702 | ins_nr, inode_only); | |
4a500fd1 YZ |
3703 | if (ret) { |
3704 | err = ret; | |
3705 | goto out_unlock; | |
3706 | } | |
31ff1cd2 CM |
3707 | ins_nr = 0; |
3708 | } | |
b3b4aa74 | 3709 | btrfs_release_path(path); |
3a5f1d45 | 3710 | |
e02119d5 CM |
3711 | if (min_key.offset < (u64)-1) |
3712 | min_key.offset++; | |
3713 | else if (min_key.type < (u8)-1) | |
3714 | min_key.type++; | |
3715 | else if (min_key.objectid < (u64)-1) | |
3716 | min_key.objectid++; | |
3717 | else | |
3718 | break; | |
3719 | } | |
31ff1cd2 | 3720 | if (ins_nr) { |
d2794405 | 3721 | ret = copy_items(trans, inode, dst_path, src, ins_start_slot, |
31ff1cd2 | 3722 | ins_nr, inode_only); |
4a500fd1 YZ |
3723 | if (ret) { |
3724 | err = ret; | |
3725 | goto out_unlock; | |
3726 | } | |
31ff1cd2 CM |
3727 | ins_nr = 0; |
3728 | } | |
5dc562c5 | 3729 | |
a95249b3 | 3730 | log_extents: |
5dc562c5 | 3731 | if (fast_search) { |
5dc562c5 | 3732 | btrfs_release_path(dst_path); |
70c8a91c | 3733 | ret = btrfs_log_changed_extents(trans, root, inode, dst_path); |
5dc562c5 JB |
3734 | if (ret) { |
3735 | err = ret; | |
3736 | goto out_unlock; | |
3737 | } | |
06d3d22b LB |
3738 | } else { |
3739 | struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree; | |
3740 | struct extent_map *em, *n; | |
3741 | ||
bbe14267 | 3742 | write_lock(&tree->lock); |
06d3d22b LB |
3743 | list_for_each_entry_safe(em, n, &tree->modified_extents, list) |
3744 | list_del_init(&em->list); | |
bbe14267 | 3745 | write_unlock(&tree->lock); |
5dc562c5 JB |
3746 | } |
3747 | ||
9623f9a3 | 3748 | if (inode_only == LOG_INODE_ALL && S_ISDIR(inode->i_mode)) { |
b3b4aa74 DS |
3749 | btrfs_release_path(path); |
3750 | btrfs_release_path(dst_path); | |
e02119d5 | 3751 | ret = log_directory_changes(trans, root, inode, path, dst_path); |
4a500fd1 YZ |
3752 | if (ret) { |
3753 | err = ret; | |
3754 | goto out_unlock; | |
3755 | } | |
e02119d5 | 3756 | } |
3a5f1d45 | 3757 | BTRFS_I(inode)->logged_trans = trans->transid; |
46d8bc34 | 3758 | BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->last_sub_trans; |
4a500fd1 | 3759 | out_unlock: |
2ab28f32 JB |
3760 | if (err) |
3761 | btrfs_free_logged_extents(log, log->log_transid); | |
e02119d5 CM |
3762 | mutex_unlock(&BTRFS_I(inode)->log_mutex); |
3763 | ||
3764 | btrfs_free_path(path); | |
3765 | btrfs_free_path(dst_path); | |
4a500fd1 | 3766 | return err; |
e02119d5 CM |
3767 | } |
3768 | ||
12fcfd22 CM |
3769 | /* |
3770 | * follow the dentry parent pointers up the chain and see if any | |
3771 | * of the directories in it require a full commit before they can | |
3772 | * be logged. Returns zero if nothing special needs to be done or 1 if | |
3773 | * a full commit is required. | |
3774 | */ | |
3775 | static noinline int check_parent_dirs_for_sync(struct btrfs_trans_handle *trans, | |
3776 | struct inode *inode, | |
3777 | struct dentry *parent, | |
3778 | struct super_block *sb, | |
3779 | u64 last_committed) | |
e02119d5 | 3780 | { |
12fcfd22 CM |
3781 | int ret = 0; |
3782 | struct btrfs_root *root; | |
6a912213 | 3783 | struct dentry *old_parent = NULL; |
e02119d5 | 3784 | |
af4176b4 CM |
3785 | /* |
3786 | * for regular files, if its inode is already on disk, we don't | |
3787 | * have to worry about the parents at all. This is because | |
3788 | * we can use the last_unlink_trans field to record renames | |
3789 | * and other fun in this file. | |
3790 | */ | |
3791 | if (S_ISREG(inode->i_mode) && | |
3792 | BTRFS_I(inode)->generation <= last_committed && | |
3793 | BTRFS_I(inode)->last_unlink_trans <= last_committed) | |
3794 | goto out; | |
3795 | ||
12fcfd22 CM |
3796 | if (!S_ISDIR(inode->i_mode)) { |
3797 | if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb) | |
3798 | goto out; | |
3799 | inode = parent->d_inode; | |
3800 | } | |
3801 | ||
3802 | while (1) { | |
3803 | BTRFS_I(inode)->logged_trans = trans->transid; | |
3804 | smp_mb(); | |
3805 | ||
3806 | if (BTRFS_I(inode)->last_unlink_trans > last_committed) { | |
3807 | root = BTRFS_I(inode)->root; | |
3808 | ||
3809 | /* | |
3810 | * make sure any commits to the log are forced | |
3811 | * to be full commits | |
3812 | */ | |
3813 | root->fs_info->last_trans_log_full_commit = | |
3814 | trans->transid; | |
3815 | ret = 1; | |
3816 | break; | |
3817 | } | |
3818 | ||
3819 | if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb) | |
3820 | break; | |
3821 | ||
76dda93c | 3822 | if (IS_ROOT(parent)) |
12fcfd22 CM |
3823 | break; |
3824 | ||
6a912213 JB |
3825 | parent = dget_parent(parent); |
3826 | dput(old_parent); | |
3827 | old_parent = parent; | |
12fcfd22 CM |
3828 | inode = parent->d_inode; |
3829 | ||
3830 | } | |
6a912213 | 3831 | dput(old_parent); |
12fcfd22 | 3832 | out: |
e02119d5 CM |
3833 | return ret; |
3834 | } | |
3835 | ||
3836 | /* | |
3837 | * helper function around btrfs_log_inode to make sure newly created | |
3838 | * parent directories also end up in the log. A minimal inode and backref | |
3839 | * only logging is done of any parent directories that are older than | |
3840 | * the last committed transaction | |
3841 | */ | |
48a3b636 ES |
3842 | static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans, |
3843 | struct btrfs_root *root, struct inode *inode, | |
3844 | struct dentry *parent, int exists_only) | |
e02119d5 | 3845 | { |
12fcfd22 | 3846 | int inode_only = exists_only ? LOG_INODE_EXISTS : LOG_INODE_ALL; |
e02119d5 | 3847 | struct super_block *sb; |
6a912213 | 3848 | struct dentry *old_parent = NULL; |
12fcfd22 CM |
3849 | int ret = 0; |
3850 | u64 last_committed = root->fs_info->last_trans_committed; | |
3851 | ||
3852 | sb = inode->i_sb; | |
3853 | ||
3a5e1404 SW |
3854 | if (btrfs_test_opt(root, NOTREELOG)) { |
3855 | ret = 1; | |
3856 | goto end_no_trans; | |
3857 | } | |
3858 | ||
12fcfd22 CM |
3859 | if (root->fs_info->last_trans_log_full_commit > |
3860 | root->fs_info->last_trans_committed) { | |
3861 | ret = 1; | |
3862 | goto end_no_trans; | |
3863 | } | |
3864 | ||
76dda93c YZ |
3865 | if (root != BTRFS_I(inode)->root || |
3866 | btrfs_root_refs(&root->root_item) == 0) { | |
3867 | ret = 1; | |
3868 | goto end_no_trans; | |
3869 | } | |
3870 | ||
12fcfd22 CM |
3871 | ret = check_parent_dirs_for_sync(trans, inode, parent, |
3872 | sb, last_committed); | |
3873 | if (ret) | |
3874 | goto end_no_trans; | |
e02119d5 | 3875 | |
22ee6985 | 3876 | if (btrfs_inode_in_log(inode, trans->transid)) { |
257c62e1 CM |
3877 | ret = BTRFS_NO_LOG_SYNC; |
3878 | goto end_no_trans; | |
3879 | } | |
3880 | ||
4a500fd1 YZ |
3881 | ret = start_log_trans(trans, root); |
3882 | if (ret) | |
3883 | goto end_trans; | |
e02119d5 | 3884 | |
12fcfd22 | 3885 | ret = btrfs_log_inode(trans, root, inode, inode_only); |
4a500fd1 YZ |
3886 | if (ret) |
3887 | goto end_trans; | |
12fcfd22 | 3888 | |
af4176b4 CM |
3889 | /* |
3890 | * for regular files, if its inode is already on disk, we don't | |
3891 | * have to worry about the parents at all. This is because | |
3892 | * we can use the last_unlink_trans field to record renames | |
3893 | * and other fun in this file. | |
3894 | */ | |
3895 | if (S_ISREG(inode->i_mode) && | |
3896 | BTRFS_I(inode)->generation <= last_committed && | |
4a500fd1 YZ |
3897 | BTRFS_I(inode)->last_unlink_trans <= last_committed) { |
3898 | ret = 0; | |
3899 | goto end_trans; | |
3900 | } | |
af4176b4 CM |
3901 | |
3902 | inode_only = LOG_INODE_EXISTS; | |
12fcfd22 CM |
3903 | while (1) { |
3904 | if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb) | |
e02119d5 CM |
3905 | break; |
3906 | ||
12fcfd22 | 3907 | inode = parent->d_inode; |
76dda93c YZ |
3908 | if (root != BTRFS_I(inode)->root) |
3909 | break; | |
3910 | ||
12fcfd22 CM |
3911 | if (BTRFS_I(inode)->generation > |
3912 | root->fs_info->last_trans_committed) { | |
3913 | ret = btrfs_log_inode(trans, root, inode, inode_only); | |
4a500fd1 YZ |
3914 | if (ret) |
3915 | goto end_trans; | |
12fcfd22 | 3916 | } |
76dda93c | 3917 | if (IS_ROOT(parent)) |
e02119d5 | 3918 | break; |
12fcfd22 | 3919 | |
6a912213 JB |
3920 | parent = dget_parent(parent); |
3921 | dput(old_parent); | |
3922 | old_parent = parent; | |
e02119d5 | 3923 | } |
12fcfd22 | 3924 | ret = 0; |
4a500fd1 | 3925 | end_trans: |
6a912213 | 3926 | dput(old_parent); |
4a500fd1 | 3927 | if (ret < 0) { |
4a500fd1 YZ |
3928 | root->fs_info->last_trans_log_full_commit = trans->transid; |
3929 | ret = 1; | |
3930 | } | |
12fcfd22 CM |
3931 | btrfs_end_log_trans(root); |
3932 | end_no_trans: | |
3933 | return ret; | |
e02119d5 CM |
3934 | } |
3935 | ||
3936 | /* | |
3937 | * it is not safe to log dentry if the chunk root has added new | |
3938 | * chunks. This returns 0 if the dentry was logged, and 1 otherwise. | |
3939 | * If this returns 1, you must commit the transaction to safely get your | |
3940 | * data on disk. | |
3941 | */ | |
3942 | int btrfs_log_dentry_safe(struct btrfs_trans_handle *trans, | |
3943 | struct btrfs_root *root, struct dentry *dentry) | |
3944 | { | |
6a912213 JB |
3945 | struct dentry *parent = dget_parent(dentry); |
3946 | int ret; | |
3947 | ||
3948 | ret = btrfs_log_inode_parent(trans, root, dentry->d_inode, parent, 0); | |
3949 | dput(parent); | |
3950 | ||
3951 | return ret; | |
e02119d5 CM |
3952 | } |
3953 | ||
3954 | /* | |
3955 | * should be called during mount to recover any replay any log trees | |
3956 | * from the FS | |
3957 | */ | |
3958 | int btrfs_recover_log_trees(struct btrfs_root *log_root_tree) | |
3959 | { | |
3960 | int ret; | |
3961 | struct btrfs_path *path; | |
3962 | struct btrfs_trans_handle *trans; | |
3963 | struct btrfs_key key; | |
3964 | struct btrfs_key found_key; | |
3965 | struct btrfs_key tmp_key; | |
3966 | struct btrfs_root *log; | |
3967 | struct btrfs_fs_info *fs_info = log_root_tree->fs_info; | |
3968 | struct walk_control wc = { | |
3969 | .process_func = process_one_buffer, | |
3970 | .stage = 0, | |
3971 | }; | |
3972 | ||
e02119d5 | 3973 | path = btrfs_alloc_path(); |
db5b493a TI |
3974 | if (!path) |
3975 | return -ENOMEM; | |
3976 | ||
3977 | fs_info->log_root_recovering = 1; | |
e02119d5 | 3978 | |
4a500fd1 | 3979 | trans = btrfs_start_transaction(fs_info->tree_root, 0); |
79787eaa JM |
3980 | if (IS_ERR(trans)) { |
3981 | ret = PTR_ERR(trans); | |
3982 | goto error; | |
3983 | } | |
e02119d5 CM |
3984 | |
3985 | wc.trans = trans; | |
3986 | wc.pin = 1; | |
3987 | ||
db5b493a | 3988 | ret = walk_log_tree(trans, log_root_tree, &wc); |
79787eaa JM |
3989 | if (ret) { |
3990 | btrfs_error(fs_info, ret, "Failed to pin buffers while " | |
3991 | "recovering log root tree."); | |
3992 | goto error; | |
3993 | } | |
e02119d5 CM |
3994 | |
3995 | again: | |
3996 | key.objectid = BTRFS_TREE_LOG_OBJECTID; | |
3997 | key.offset = (u64)-1; | |
3998 | btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); | |
3999 | ||
d397712b | 4000 | while (1) { |
e02119d5 | 4001 | ret = btrfs_search_slot(NULL, log_root_tree, &key, path, 0, 0); |
79787eaa JM |
4002 | |
4003 | if (ret < 0) { | |
4004 | btrfs_error(fs_info, ret, | |
4005 | "Couldn't find tree log root."); | |
4006 | goto error; | |
4007 | } | |
e02119d5 CM |
4008 | if (ret > 0) { |
4009 | if (path->slots[0] == 0) | |
4010 | break; | |
4011 | path->slots[0]--; | |
4012 | } | |
4013 | btrfs_item_key_to_cpu(path->nodes[0], &found_key, | |
4014 | path->slots[0]); | |
b3b4aa74 | 4015 | btrfs_release_path(path); |
e02119d5 CM |
4016 | if (found_key.objectid != BTRFS_TREE_LOG_OBJECTID) |
4017 | break; | |
4018 | ||
cb517eab | 4019 | log = btrfs_read_fs_root(log_root_tree, &found_key); |
79787eaa JM |
4020 | if (IS_ERR(log)) { |
4021 | ret = PTR_ERR(log); | |
4022 | btrfs_error(fs_info, ret, | |
4023 | "Couldn't read tree log root."); | |
4024 | goto error; | |
4025 | } | |
e02119d5 CM |
4026 | |
4027 | tmp_key.objectid = found_key.offset; | |
4028 | tmp_key.type = BTRFS_ROOT_ITEM_KEY; | |
4029 | tmp_key.offset = (u64)-1; | |
4030 | ||
4031 | wc.replay_dest = btrfs_read_fs_root_no_name(fs_info, &tmp_key); | |
79787eaa JM |
4032 | if (IS_ERR(wc.replay_dest)) { |
4033 | ret = PTR_ERR(wc.replay_dest); | |
b50c6e25 JB |
4034 | free_extent_buffer(log->node); |
4035 | free_extent_buffer(log->commit_root); | |
4036 | kfree(log); | |
79787eaa JM |
4037 | btrfs_error(fs_info, ret, "Couldn't read target root " |
4038 | "for tree log recovery."); | |
4039 | goto error; | |
4040 | } | |
e02119d5 | 4041 | |
07d400a6 | 4042 | wc.replay_dest->log_root = log; |
5d4f98a2 | 4043 | btrfs_record_root_in_trans(trans, wc.replay_dest); |
e02119d5 | 4044 | ret = walk_log_tree(trans, log, &wc); |
e02119d5 | 4045 | |
b50c6e25 | 4046 | if (!ret && wc.stage == LOG_WALK_REPLAY_ALL) { |
e02119d5 CM |
4047 | ret = fixup_inode_link_counts(trans, wc.replay_dest, |
4048 | path); | |
e02119d5 CM |
4049 | } |
4050 | ||
4051 | key.offset = found_key.offset - 1; | |
07d400a6 | 4052 | wc.replay_dest->log_root = NULL; |
e02119d5 | 4053 | free_extent_buffer(log->node); |
b263c2c8 | 4054 | free_extent_buffer(log->commit_root); |
e02119d5 CM |
4055 | kfree(log); |
4056 | ||
b50c6e25 JB |
4057 | if (ret) |
4058 | goto error; | |
4059 | ||
e02119d5 CM |
4060 | if (found_key.offset == 0) |
4061 | break; | |
4062 | } | |
b3b4aa74 | 4063 | btrfs_release_path(path); |
e02119d5 CM |
4064 | |
4065 | /* step one is to pin it all, step two is to replay just inodes */ | |
4066 | if (wc.pin) { | |
4067 | wc.pin = 0; | |
4068 | wc.process_func = replay_one_buffer; | |
4069 | wc.stage = LOG_WALK_REPLAY_INODES; | |
4070 | goto again; | |
4071 | } | |
4072 | /* step three is to replay everything */ | |
4073 | if (wc.stage < LOG_WALK_REPLAY_ALL) { | |
4074 | wc.stage++; | |
4075 | goto again; | |
4076 | } | |
4077 | ||
4078 | btrfs_free_path(path); | |
4079 | ||
abefa55a JB |
4080 | /* step 4: commit the transaction, which also unpins the blocks */ |
4081 | ret = btrfs_commit_transaction(trans, fs_info->tree_root); | |
4082 | if (ret) | |
4083 | return ret; | |
4084 | ||
e02119d5 CM |
4085 | free_extent_buffer(log_root_tree->node); |
4086 | log_root_tree->log_root = NULL; | |
4087 | fs_info->log_root_recovering = 0; | |
e02119d5 | 4088 | kfree(log_root_tree); |
79787eaa | 4089 | |
abefa55a | 4090 | return 0; |
79787eaa | 4091 | error: |
b50c6e25 JB |
4092 | if (wc.trans) |
4093 | btrfs_end_transaction(wc.trans, fs_info->tree_root); | |
79787eaa JM |
4094 | btrfs_free_path(path); |
4095 | return ret; | |
e02119d5 | 4096 | } |
12fcfd22 CM |
4097 | |
4098 | /* | |
4099 | * there are some corner cases where we want to force a full | |
4100 | * commit instead of allowing a directory to be logged. | |
4101 | * | |
4102 | * They revolve around files there were unlinked from the directory, and | |
4103 | * this function updates the parent directory so that a full commit is | |
4104 | * properly done if it is fsync'd later after the unlinks are done. | |
4105 | */ | |
4106 | void btrfs_record_unlink_dir(struct btrfs_trans_handle *trans, | |
4107 | struct inode *dir, struct inode *inode, | |
4108 | int for_rename) | |
4109 | { | |
af4176b4 CM |
4110 | /* |
4111 | * when we're logging a file, if it hasn't been renamed | |
4112 | * or unlinked, and its inode is fully committed on disk, | |
4113 | * we don't have to worry about walking up the directory chain | |
4114 | * to log its parents. | |
4115 | * | |
4116 | * So, we use the last_unlink_trans field to put this transid | |
4117 | * into the file. When the file is logged we check it and | |
4118 | * don't log the parents if the file is fully on disk. | |
4119 | */ | |
4120 | if (S_ISREG(inode->i_mode)) | |
4121 | BTRFS_I(inode)->last_unlink_trans = trans->transid; | |
4122 | ||
12fcfd22 CM |
4123 | /* |
4124 | * if this directory was already logged any new | |
4125 | * names for this file/dir will get recorded | |
4126 | */ | |
4127 | smp_mb(); | |
4128 | if (BTRFS_I(dir)->logged_trans == trans->transid) | |
4129 | return; | |
4130 | ||
4131 | /* | |
4132 | * if the inode we're about to unlink was logged, | |
4133 | * the log will be properly updated for any new names | |
4134 | */ | |
4135 | if (BTRFS_I(inode)->logged_trans == trans->transid) | |
4136 | return; | |
4137 | ||
4138 | /* | |
4139 | * when renaming files across directories, if the directory | |
4140 | * there we're unlinking from gets fsync'd later on, there's | |
4141 | * no way to find the destination directory later and fsync it | |
4142 | * properly. So, we have to be conservative and force commits | |
4143 | * so the new name gets discovered. | |
4144 | */ | |
4145 | if (for_rename) | |
4146 | goto record; | |
4147 | ||
4148 | /* we can safely do the unlink without any special recording */ | |
4149 | return; | |
4150 | ||
4151 | record: | |
4152 | BTRFS_I(dir)->last_unlink_trans = trans->transid; | |
4153 | } | |
4154 | ||
4155 | /* | |
4156 | * Call this after adding a new name for a file and it will properly | |
4157 | * update the log to reflect the new name. | |
4158 | * | |
4159 | * It will return zero if all goes well, and it will return 1 if a | |
4160 | * full transaction commit is required. | |
4161 | */ | |
4162 | int btrfs_log_new_name(struct btrfs_trans_handle *trans, | |
4163 | struct inode *inode, struct inode *old_dir, | |
4164 | struct dentry *parent) | |
4165 | { | |
4166 | struct btrfs_root * root = BTRFS_I(inode)->root; | |
4167 | ||
af4176b4 CM |
4168 | /* |
4169 | * this will force the logging code to walk the dentry chain | |
4170 | * up for the file | |
4171 | */ | |
4172 | if (S_ISREG(inode->i_mode)) | |
4173 | BTRFS_I(inode)->last_unlink_trans = trans->transid; | |
4174 | ||
12fcfd22 CM |
4175 | /* |
4176 | * if this inode hasn't been logged and directory we're renaming it | |
4177 | * from hasn't been logged, we don't need to log it | |
4178 | */ | |
4179 | if (BTRFS_I(inode)->logged_trans <= | |
4180 | root->fs_info->last_trans_committed && | |
4181 | (!old_dir || BTRFS_I(old_dir)->logged_trans <= | |
4182 | root->fs_info->last_trans_committed)) | |
4183 | return 0; | |
4184 | ||
4185 | return btrfs_log_inode_parent(trans, root, inode, parent, 1); | |
4186 | } | |
4187 |