]>
Commit | Line | Data |
---|---|---|
1 | #include <linux/bitops.h> | |
2 | #include <linux/slab.h> | |
3 | #include <linux/bio.h> | |
4 | #include <linux/mm.h> | |
5 | #include <linux/gfp.h> | |
6 | #include <linux/pagemap.h> | |
7 | #include <linux/page-flags.h> | |
8 | #include <linux/module.h> | |
9 | #include <linux/spinlock.h> | |
10 | #include <linux/blkdev.h> | |
11 | #include <linux/swap.h> | |
12 | #include <linux/writeback.h> | |
13 | #include <linux/pagevec.h> | |
14 | #include "extent_io.h" | |
15 | #include "extent_map.h" | |
16 | #include "compat.h" | |
17 | #include "ctree.h" | |
18 | #include "btrfs_inode.h" | |
19 | ||
20 | static struct kmem_cache *extent_state_cache; | |
21 | static struct kmem_cache *extent_buffer_cache; | |
22 | ||
23 | static LIST_HEAD(buffers); | |
24 | static LIST_HEAD(states); | |
25 | ||
26 | #define LEAK_DEBUG 0 | |
27 | #if LEAK_DEBUG | |
28 | static DEFINE_SPINLOCK(leak_lock); | |
29 | #endif | |
30 | ||
31 | #define BUFFER_LRU_MAX 64 | |
32 | ||
33 | struct tree_entry { | |
34 | u64 start; | |
35 | u64 end; | |
36 | struct rb_node rb_node; | |
37 | }; | |
38 | ||
39 | struct extent_page_data { | |
40 | struct bio *bio; | |
41 | struct extent_io_tree *tree; | |
42 | get_extent_t *get_extent; | |
43 | ||
44 | /* tells writepage not to lock the state bits for this range | |
45 | * it still does the unlocking | |
46 | */ | |
47 | unsigned int extent_locked:1; | |
48 | ||
49 | /* tells the submit_bio code to use a WRITE_SYNC */ | |
50 | unsigned int sync_io:1; | |
51 | }; | |
52 | ||
53 | int __init extent_io_init(void) | |
54 | { | |
55 | extent_state_cache = kmem_cache_create("extent_state", | |
56 | sizeof(struct extent_state), 0, | |
57 | SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); | |
58 | if (!extent_state_cache) | |
59 | return -ENOMEM; | |
60 | ||
61 | extent_buffer_cache = kmem_cache_create("extent_buffers", | |
62 | sizeof(struct extent_buffer), 0, | |
63 | SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL); | |
64 | if (!extent_buffer_cache) | |
65 | goto free_state_cache; | |
66 | return 0; | |
67 | ||
68 | free_state_cache: | |
69 | kmem_cache_destroy(extent_state_cache); | |
70 | return -ENOMEM; | |
71 | } | |
72 | ||
73 | void extent_io_exit(void) | |
74 | { | |
75 | struct extent_state *state; | |
76 | struct extent_buffer *eb; | |
77 | ||
78 | while (!list_empty(&states)) { | |
79 | state = list_entry(states.next, struct extent_state, leak_list); | |
80 | printk(KERN_ERR "btrfs state leak: start %llu end %llu " | |
81 | "state %lu in tree %p refs %d\n", | |
82 | (unsigned long long)state->start, | |
83 | (unsigned long long)state->end, | |
84 | state->state, state->tree, atomic_read(&state->refs)); | |
85 | list_del(&state->leak_list); | |
86 | kmem_cache_free(extent_state_cache, state); | |
87 | ||
88 | } | |
89 | ||
90 | while (!list_empty(&buffers)) { | |
91 | eb = list_entry(buffers.next, struct extent_buffer, leak_list); | |
92 | printk(KERN_ERR "btrfs buffer leak start %llu len %lu " | |
93 | "refs %d\n", (unsigned long long)eb->start, | |
94 | eb->len, atomic_read(&eb->refs)); | |
95 | list_del(&eb->leak_list); | |
96 | kmem_cache_free(extent_buffer_cache, eb); | |
97 | } | |
98 | if (extent_state_cache) | |
99 | kmem_cache_destroy(extent_state_cache); | |
100 | if (extent_buffer_cache) | |
101 | kmem_cache_destroy(extent_buffer_cache); | |
102 | } | |
103 | ||
104 | void extent_io_tree_init(struct extent_io_tree *tree, | |
105 | struct address_space *mapping, gfp_t mask) | |
106 | { | |
107 | tree->state.rb_node = NULL; | |
108 | tree->buffer.rb_node = NULL; | |
109 | tree->ops = NULL; | |
110 | tree->dirty_bytes = 0; | |
111 | spin_lock_init(&tree->lock); | |
112 | spin_lock_init(&tree->buffer_lock); | |
113 | tree->mapping = mapping; | |
114 | } | |
115 | ||
116 | static struct extent_state *alloc_extent_state(gfp_t mask) | |
117 | { | |
118 | struct extent_state *state; | |
119 | #if LEAK_DEBUG | |
120 | unsigned long flags; | |
121 | #endif | |
122 | ||
123 | state = kmem_cache_alloc(extent_state_cache, mask); | |
124 | if (!state) | |
125 | return state; | |
126 | state->state = 0; | |
127 | state->private = 0; | |
128 | state->tree = NULL; | |
129 | #if LEAK_DEBUG | |
130 | spin_lock_irqsave(&leak_lock, flags); | |
131 | list_add(&state->leak_list, &states); | |
132 | spin_unlock_irqrestore(&leak_lock, flags); | |
133 | #endif | |
134 | atomic_set(&state->refs, 1); | |
135 | init_waitqueue_head(&state->wq); | |
136 | return state; | |
137 | } | |
138 | ||
139 | static void free_extent_state(struct extent_state *state) | |
140 | { | |
141 | if (!state) | |
142 | return; | |
143 | if (atomic_dec_and_test(&state->refs)) { | |
144 | #if LEAK_DEBUG | |
145 | unsigned long flags; | |
146 | #endif | |
147 | WARN_ON(state->tree); | |
148 | #if LEAK_DEBUG | |
149 | spin_lock_irqsave(&leak_lock, flags); | |
150 | list_del(&state->leak_list); | |
151 | spin_unlock_irqrestore(&leak_lock, flags); | |
152 | #endif | |
153 | kmem_cache_free(extent_state_cache, state); | |
154 | } | |
155 | } | |
156 | ||
157 | static struct rb_node *tree_insert(struct rb_root *root, u64 offset, | |
158 | struct rb_node *node) | |
159 | { | |
160 | struct rb_node **p = &root->rb_node; | |
161 | struct rb_node *parent = NULL; | |
162 | struct tree_entry *entry; | |
163 | ||
164 | while (*p) { | |
165 | parent = *p; | |
166 | entry = rb_entry(parent, struct tree_entry, rb_node); | |
167 | ||
168 | if (offset < entry->start) | |
169 | p = &(*p)->rb_left; | |
170 | else if (offset > entry->end) | |
171 | p = &(*p)->rb_right; | |
172 | else | |
173 | return parent; | |
174 | } | |
175 | ||
176 | entry = rb_entry(node, struct tree_entry, rb_node); | |
177 | rb_link_node(node, parent, p); | |
178 | rb_insert_color(node, root); | |
179 | return NULL; | |
180 | } | |
181 | ||
182 | static struct rb_node *__etree_search(struct extent_io_tree *tree, u64 offset, | |
183 | struct rb_node **prev_ret, | |
184 | struct rb_node **next_ret) | |
185 | { | |
186 | struct rb_root *root = &tree->state; | |
187 | struct rb_node *n = root->rb_node; | |
188 | struct rb_node *prev = NULL; | |
189 | struct rb_node *orig_prev = NULL; | |
190 | struct tree_entry *entry; | |
191 | struct tree_entry *prev_entry = NULL; | |
192 | ||
193 | while (n) { | |
194 | entry = rb_entry(n, struct tree_entry, rb_node); | |
195 | prev = n; | |
196 | prev_entry = entry; | |
197 | ||
198 | if (offset < entry->start) | |
199 | n = n->rb_left; | |
200 | else if (offset > entry->end) | |
201 | n = n->rb_right; | |
202 | else | |
203 | return n; | |
204 | } | |
205 | ||
206 | if (prev_ret) { | |
207 | orig_prev = prev; | |
208 | while (prev && offset > prev_entry->end) { | |
209 | prev = rb_next(prev); | |
210 | prev_entry = rb_entry(prev, struct tree_entry, rb_node); | |
211 | } | |
212 | *prev_ret = prev; | |
213 | prev = orig_prev; | |
214 | } | |
215 | ||
216 | if (next_ret) { | |
217 | prev_entry = rb_entry(prev, struct tree_entry, rb_node); | |
218 | while (prev && offset < prev_entry->start) { | |
219 | prev = rb_prev(prev); | |
220 | prev_entry = rb_entry(prev, struct tree_entry, rb_node); | |
221 | } | |
222 | *next_ret = prev; | |
223 | } | |
224 | return NULL; | |
225 | } | |
226 | ||
227 | static inline struct rb_node *tree_search(struct extent_io_tree *tree, | |
228 | u64 offset) | |
229 | { | |
230 | struct rb_node *prev = NULL; | |
231 | struct rb_node *ret; | |
232 | ||
233 | ret = __etree_search(tree, offset, &prev, NULL); | |
234 | if (!ret) | |
235 | return prev; | |
236 | return ret; | |
237 | } | |
238 | ||
239 | static struct extent_buffer *buffer_tree_insert(struct extent_io_tree *tree, | |
240 | u64 offset, struct rb_node *node) | |
241 | { | |
242 | struct rb_root *root = &tree->buffer; | |
243 | struct rb_node **p = &root->rb_node; | |
244 | struct rb_node *parent = NULL; | |
245 | struct extent_buffer *eb; | |
246 | ||
247 | while (*p) { | |
248 | parent = *p; | |
249 | eb = rb_entry(parent, struct extent_buffer, rb_node); | |
250 | ||
251 | if (offset < eb->start) | |
252 | p = &(*p)->rb_left; | |
253 | else if (offset > eb->start) | |
254 | p = &(*p)->rb_right; | |
255 | else | |
256 | return eb; | |
257 | } | |
258 | ||
259 | rb_link_node(node, parent, p); | |
260 | rb_insert_color(node, root); | |
261 | return NULL; | |
262 | } | |
263 | ||
264 | static struct extent_buffer *buffer_search(struct extent_io_tree *tree, | |
265 | u64 offset) | |
266 | { | |
267 | struct rb_root *root = &tree->buffer; | |
268 | struct rb_node *n = root->rb_node; | |
269 | struct extent_buffer *eb; | |
270 | ||
271 | while (n) { | |
272 | eb = rb_entry(n, struct extent_buffer, rb_node); | |
273 | if (offset < eb->start) | |
274 | n = n->rb_left; | |
275 | else if (offset > eb->start) | |
276 | n = n->rb_right; | |
277 | else | |
278 | return eb; | |
279 | } | |
280 | return NULL; | |
281 | } | |
282 | ||
283 | static void merge_cb(struct extent_io_tree *tree, struct extent_state *new, | |
284 | struct extent_state *other) | |
285 | { | |
286 | if (tree->ops && tree->ops->merge_extent_hook) | |
287 | tree->ops->merge_extent_hook(tree->mapping->host, new, | |
288 | other); | |
289 | } | |
290 | ||
291 | /* | |
292 | * utility function to look for merge candidates inside a given range. | |
293 | * Any extents with matching state are merged together into a single | |
294 | * extent in the tree. Extents with EXTENT_IO in their state field | |
295 | * are not merged because the end_io handlers need to be able to do | |
296 | * operations on them without sleeping (or doing allocations/splits). | |
297 | * | |
298 | * This should be called with the tree lock held. | |
299 | */ | |
300 | static int merge_state(struct extent_io_tree *tree, | |
301 | struct extent_state *state) | |
302 | { | |
303 | struct extent_state *other; | |
304 | struct rb_node *other_node; | |
305 | ||
306 | if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY)) | |
307 | return 0; | |
308 | ||
309 | other_node = rb_prev(&state->rb_node); | |
310 | if (other_node) { | |
311 | other = rb_entry(other_node, struct extent_state, rb_node); | |
312 | if (other->end == state->start - 1 && | |
313 | other->state == state->state) { | |
314 | merge_cb(tree, state, other); | |
315 | state->start = other->start; | |
316 | other->tree = NULL; | |
317 | rb_erase(&other->rb_node, &tree->state); | |
318 | free_extent_state(other); | |
319 | } | |
320 | } | |
321 | other_node = rb_next(&state->rb_node); | |
322 | if (other_node) { | |
323 | other = rb_entry(other_node, struct extent_state, rb_node); | |
324 | if (other->start == state->end + 1 && | |
325 | other->state == state->state) { | |
326 | merge_cb(tree, state, other); | |
327 | other->start = state->start; | |
328 | state->tree = NULL; | |
329 | rb_erase(&state->rb_node, &tree->state); | |
330 | free_extent_state(state); | |
331 | state = NULL; | |
332 | } | |
333 | } | |
334 | ||
335 | return 0; | |
336 | } | |
337 | ||
338 | static int set_state_cb(struct extent_io_tree *tree, | |
339 | struct extent_state *state, | |
340 | unsigned long bits) | |
341 | { | |
342 | if (tree->ops && tree->ops->set_bit_hook) { | |
343 | return tree->ops->set_bit_hook(tree->mapping->host, | |
344 | state->start, state->end, | |
345 | state->state, bits); | |
346 | } | |
347 | ||
348 | return 0; | |
349 | } | |
350 | ||
351 | static void clear_state_cb(struct extent_io_tree *tree, | |
352 | struct extent_state *state, | |
353 | unsigned long bits) | |
354 | { | |
355 | if (tree->ops && tree->ops->clear_bit_hook) | |
356 | tree->ops->clear_bit_hook(tree->mapping->host, state, bits); | |
357 | } | |
358 | ||
359 | /* | |
360 | * insert an extent_state struct into the tree. 'bits' are set on the | |
361 | * struct before it is inserted. | |
362 | * | |
363 | * This may return -EEXIST if the extent is already there, in which case the | |
364 | * state struct is freed. | |
365 | * | |
366 | * The tree lock is not taken internally. This is a utility function and | |
367 | * probably isn't what you want to call (see set/clear_extent_bit). | |
368 | */ | |
369 | static int insert_state(struct extent_io_tree *tree, | |
370 | struct extent_state *state, u64 start, u64 end, | |
371 | int bits) | |
372 | { | |
373 | struct rb_node *node; | |
374 | int ret; | |
375 | ||
376 | if (end < start) { | |
377 | printk(KERN_ERR "btrfs end < start %llu %llu\n", | |
378 | (unsigned long long)end, | |
379 | (unsigned long long)start); | |
380 | WARN_ON(1); | |
381 | } | |
382 | state->start = start; | |
383 | state->end = end; | |
384 | ret = set_state_cb(tree, state, bits); | |
385 | if (ret) | |
386 | return ret; | |
387 | ||
388 | if (bits & EXTENT_DIRTY) | |
389 | tree->dirty_bytes += end - start + 1; | |
390 | state->state |= bits; | |
391 | node = tree_insert(&tree->state, end, &state->rb_node); | |
392 | if (node) { | |
393 | struct extent_state *found; | |
394 | found = rb_entry(node, struct extent_state, rb_node); | |
395 | printk(KERN_ERR "btrfs found node %llu %llu on insert of " | |
396 | "%llu %llu\n", (unsigned long long)found->start, | |
397 | (unsigned long long)found->end, | |
398 | (unsigned long long)start, (unsigned long long)end); | |
399 | free_extent_state(state); | |
400 | return -EEXIST; | |
401 | } | |
402 | state->tree = tree; | |
403 | merge_state(tree, state); | |
404 | return 0; | |
405 | } | |
406 | ||
407 | static int split_cb(struct extent_io_tree *tree, struct extent_state *orig, | |
408 | u64 split) | |
409 | { | |
410 | if (tree->ops && tree->ops->split_extent_hook) | |
411 | return tree->ops->split_extent_hook(tree->mapping->host, | |
412 | orig, split); | |
413 | return 0; | |
414 | } | |
415 | ||
416 | /* | |
417 | * split a given extent state struct in two, inserting the preallocated | |
418 | * struct 'prealloc' as the newly created second half. 'split' indicates an | |
419 | * offset inside 'orig' where it should be split. | |
420 | * | |
421 | * Before calling, | |
422 | * the tree has 'orig' at [orig->start, orig->end]. After calling, there | |
423 | * are two extent state structs in the tree: | |
424 | * prealloc: [orig->start, split - 1] | |
425 | * orig: [ split, orig->end ] | |
426 | * | |
427 | * The tree locks are not taken by this function. They need to be held | |
428 | * by the caller. | |
429 | */ | |
430 | static int split_state(struct extent_io_tree *tree, struct extent_state *orig, | |
431 | struct extent_state *prealloc, u64 split) | |
432 | { | |
433 | struct rb_node *node; | |
434 | ||
435 | split_cb(tree, orig, split); | |
436 | ||
437 | prealloc->start = orig->start; | |
438 | prealloc->end = split - 1; | |
439 | prealloc->state = orig->state; | |
440 | orig->start = split; | |
441 | ||
442 | node = tree_insert(&tree->state, prealloc->end, &prealloc->rb_node); | |
443 | if (node) { | |
444 | free_extent_state(prealloc); | |
445 | return -EEXIST; | |
446 | } | |
447 | prealloc->tree = tree; | |
448 | return 0; | |
449 | } | |
450 | ||
451 | /* | |
452 | * utility function to clear some bits in an extent state struct. | |
453 | * it will optionally wake up any one waiting on this state (wake == 1), or | |
454 | * forcibly remove the state from the tree (delete == 1). | |
455 | * | |
456 | * If no bits are set on the state struct after clearing things, the | |
457 | * struct is freed and removed from the tree | |
458 | */ | |
459 | static int clear_state_bit(struct extent_io_tree *tree, | |
460 | struct extent_state *state, int bits, int wake, | |
461 | int delete) | |
462 | { | |
463 | int ret = state->state & bits; | |
464 | ||
465 | if ((bits & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) { | |
466 | u64 range = state->end - state->start + 1; | |
467 | WARN_ON(range > tree->dirty_bytes); | |
468 | tree->dirty_bytes -= range; | |
469 | } | |
470 | clear_state_cb(tree, state, bits); | |
471 | state->state &= ~bits; | |
472 | if (wake) | |
473 | wake_up(&state->wq); | |
474 | if (delete || state->state == 0) { | |
475 | if (state->tree) { | |
476 | clear_state_cb(tree, state, state->state); | |
477 | rb_erase(&state->rb_node, &tree->state); | |
478 | state->tree = NULL; | |
479 | free_extent_state(state); | |
480 | } else { | |
481 | WARN_ON(1); | |
482 | } | |
483 | } else { | |
484 | merge_state(tree, state); | |
485 | } | |
486 | return ret; | |
487 | } | |
488 | ||
489 | /* | |
490 | * clear some bits on a range in the tree. This may require splitting | |
491 | * or inserting elements in the tree, so the gfp mask is used to | |
492 | * indicate which allocations or sleeping are allowed. | |
493 | * | |
494 | * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove | |
495 | * the given range from the tree regardless of state (ie for truncate). | |
496 | * | |
497 | * the range [start, end] is inclusive. | |
498 | * | |
499 | * This takes the tree lock, and returns < 0 on error, > 0 if any of the | |
500 | * bits were already set, or zero if none of the bits were already set. | |
501 | */ | |
502 | int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, | |
503 | int bits, int wake, int delete, | |
504 | struct extent_state **cached_state, | |
505 | gfp_t mask) | |
506 | { | |
507 | struct extent_state *state; | |
508 | struct extent_state *cached; | |
509 | struct extent_state *prealloc = NULL; | |
510 | struct rb_node *next_node; | |
511 | struct rb_node *node; | |
512 | u64 last_end; | |
513 | int err; | |
514 | int set = 0; | |
515 | ||
516 | again: | |
517 | if (!prealloc && (mask & __GFP_WAIT)) { | |
518 | prealloc = alloc_extent_state(mask); | |
519 | if (!prealloc) | |
520 | return -ENOMEM; | |
521 | } | |
522 | ||
523 | spin_lock(&tree->lock); | |
524 | if (cached_state) { | |
525 | cached = *cached_state; | |
526 | *cached_state = NULL; | |
527 | cached_state = NULL; | |
528 | if (cached && cached->tree && cached->start == start) { | |
529 | atomic_dec(&cached->refs); | |
530 | state = cached; | |
531 | goto hit_next; | |
532 | } | |
533 | free_extent_state(cached); | |
534 | } | |
535 | /* | |
536 | * this search will find the extents that end after | |
537 | * our range starts | |
538 | */ | |
539 | node = tree_search(tree, start); | |
540 | if (!node) | |
541 | goto out; | |
542 | state = rb_entry(node, struct extent_state, rb_node); | |
543 | hit_next: | |
544 | if (state->start > end) | |
545 | goto out; | |
546 | WARN_ON(state->end < start); | |
547 | last_end = state->end; | |
548 | ||
549 | /* | |
550 | * | ---- desired range ---- | | |
551 | * | state | or | |
552 | * | ------------- state -------------- | | |
553 | * | |
554 | * We need to split the extent we found, and may flip | |
555 | * bits on second half. | |
556 | * | |
557 | * If the extent we found extends past our range, we | |
558 | * just split and search again. It'll get split again | |
559 | * the next time though. | |
560 | * | |
561 | * If the extent we found is inside our range, we clear | |
562 | * the desired bit on it. | |
563 | */ | |
564 | ||
565 | if (state->start < start) { | |
566 | if (!prealloc) | |
567 | prealloc = alloc_extent_state(GFP_ATOMIC); | |
568 | err = split_state(tree, state, prealloc, start); | |
569 | BUG_ON(err == -EEXIST); | |
570 | prealloc = NULL; | |
571 | if (err) | |
572 | goto out; | |
573 | if (state->end <= end) { | |
574 | set |= clear_state_bit(tree, state, bits, wake, | |
575 | delete); | |
576 | if (last_end == (u64)-1) | |
577 | goto out; | |
578 | start = last_end + 1; | |
579 | } | |
580 | goto search_again; | |
581 | } | |
582 | /* | |
583 | * | ---- desired range ---- | | |
584 | * | state | | |
585 | * We need to split the extent, and clear the bit | |
586 | * on the first half | |
587 | */ | |
588 | if (state->start <= end && state->end > end) { | |
589 | if (!prealloc) | |
590 | prealloc = alloc_extent_state(GFP_ATOMIC); | |
591 | err = split_state(tree, state, prealloc, end + 1); | |
592 | BUG_ON(err == -EEXIST); | |
593 | if (wake) | |
594 | wake_up(&state->wq); | |
595 | ||
596 | set |= clear_state_bit(tree, prealloc, bits, wake, delete); | |
597 | ||
598 | prealloc = NULL; | |
599 | goto out; | |
600 | } | |
601 | ||
602 | if (state->end < end && prealloc && !need_resched()) | |
603 | next_node = rb_next(&state->rb_node); | |
604 | else | |
605 | next_node = NULL; | |
606 | ||
607 | set |= clear_state_bit(tree, state, bits, wake, delete); | |
608 | if (last_end == (u64)-1) | |
609 | goto out; | |
610 | start = last_end + 1; | |
611 | if (start <= end && next_node) { | |
612 | state = rb_entry(next_node, struct extent_state, | |
613 | rb_node); | |
614 | if (state->start == start) | |
615 | goto hit_next; | |
616 | } | |
617 | goto search_again; | |
618 | ||
619 | out: | |
620 | spin_unlock(&tree->lock); | |
621 | if (prealloc) | |
622 | free_extent_state(prealloc); | |
623 | ||
624 | return set; | |
625 | ||
626 | search_again: | |
627 | if (start > end) | |
628 | goto out; | |
629 | spin_unlock(&tree->lock); | |
630 | if (mask & __GFP_WAIT) | |
631 | cond_resched(); | |
632 | goto again; | |
633 | } | |
634 | ||
635 | static int wait_on_state(struct extent_io_tree *tree, | |
636 | struct extent_state *state) | |
637 | __releases(tree->lock) | |
638 | __acquires(tree->lock) | |
639 | { | |
640 | DEFINE_WAIT(wait); | |
641 | prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE); | |
642 | spin_unlock(&tree->lock); | |
643 | schedule(); | |
644 | spin_lock(&tree->lock); | |
645 | finish_wait(&state->wq, &wait); | |
646 | return 0; | |
647 | } | |
648 | ||
649 | /* | |
650 | * waits for one or more bits to clear on a range in the state tree. | |
651 | * The range [start, end] is inclusive. | |
652 | * The tree lock is taken by this function | |
653 | */ | |
654 | int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits) | |
655 | { | |
656 | struct extent_state *state; | |
657 | struct rb_node *node; | |
658 | ||
659 | spin_lock(&tree->lock); | |
660 | again: | |
661 | while (1) { | |
662 | /* | |
663 | * this search will find all the extents that end after | |
664 | * our range starts | |
665 | */ | |
666 | node = tree_search(tree, start); | |
667 | if (!node) | |
668 | break; | |
669 | ||
670 | state = rb_entry(node, struct extent_state, rb_node); | |
671 | ||
672 | if (state->start > end) | |
673 | goto out; | |
674 | ||
675 | if (state->state & bits) { | |
676 | start = state->start; | |
677 | atomic_inc(&state->refs); | |
678 | wait_on_state(tree, state); | |
679 | free_extent_state(state); | |
680 | goto again; | |
681 | } | |
682 | start = state->end + 1; | |
683 | ||
684 | if (start > end) | |
685 | break; | |
686 | ||
687 | if (need_resched()) { | |
688 | spin_unlock(&tree->lock); | |
689 | cond_resched(); | |
690 | spin_lock(&tree->lock); | |
691 | } | |
692 | } | |
693 | out: | |
694 | spin_unlock(&tree->lock); | |
695 | return 0; | |
696 | } | |
697 | ||
698 | static int set_state_bits(struct extent_io_tree *tree, | |
699 | struct extent_state *state, | |
700 | int bits) | |
701 | { | |
702 | int ret; | |
703 | ||
704 | ret = set_state_cb(tree, state, bits); | |
705 | if (ret) | |
706 | return ret; | |
707 | ||
708 | if ((bits & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) { | |
709 | u64 range = state->end - state->start + 1; | |
710 | tree->dirty_bytes += range; | |
711 | } | |
712 | state->state |= bits; | |
713 | ||
714 | return 0; | |
715 | } | |
716 | ||
717 | static void cache_state(struct extent_state *state, | |
718 | struct extent_state **cached_ptr) | |
719 | { | |
720 | if (cached_ptr && !(*cached_ptr)) { | |
721 | if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY)) { | |
722 | *cached_ptr = state; | |
723 | atomic_inc(&state->refs); | |
724 | } | |
725 | } | |
726 | } | |
727 | ||
728 | /* | |
729 | * set some bits on a range in the tree. This may require allocations or | |
730 | * sleeping, so the gfp mask is used to indicate what is allowed. | |
731 | * | |
732 | * If any of the exclusive bits are set, this will fail with -EEXIST if some | |
733 | * part of the range already has the desired bits set. The start of the | |
734 | * existing range is returned in failed_start in this case. | |
735 | * | |
736 | * [start, end] is inclusive This takes the tree lock. | |
737 | */ | |
738 | ||
739 | static int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, | |
740 | int bits, int exclusive_bits, u64 *failed_start, | |
741 | struct extent_state **cached_state, | |
742 | gfp_t mask) | |
743 | { | |
744 | struct extent_state *state; | |
745 | struct extent_state *prealloc = NULL; | |
746 | struct rb_node *node; | |
747 | int err = 0; | |
748 | u64 last_start; | |
749 | u64 last_end; | |
750 | ||
751 | again: | |
752 | if (!prealloc && (mask & __GFP_WAIT)) { | |
753 | prealloc = alloc_extent_state(mask); | |
754 | if (!prealloc) | |
755 | return -ENOMEM; | |
756 | } | |
757 | ||
758 | spin_lock(&tree->lock); | |
759 | if (cached_state && *cached_state) { | |
760 | state = *cached_state; | |
761 | if (state->start == start && state->tree) { | |
762 | node = &state->rb_node; | |
763 | goto hit_next; | |
764 | } | |
765 | } | |
766 | /* | |
767 | * this search will find all the extents that end after | |
768 | * our range starts. | |
769 | */ | |
770 | node = tree_search(tree, start); | |
771 | if (!node) { | |
772 | err = insert_state(tree, prealloc, start, end, bits); | |
773 | prealloc = NULL; | |
774 | BUG_ON(err == -EEXIST); | |
775 | goto out; | |
776 | } | |
777 | state = rb_entry(node, struct extent_state, rb_node); | |
778 | hit_next: | |
779 | last_start = state->start; | |
780 | last_end = state->end; | |
781 | ||
782 | /* | |
783 | * | ---- desired range ---- | | |
784 | * | state | | |
785 | * | |
786 | * Just lock what we found and keep going | |
787 | */ | |
788 | if (state->start == start && state->end <= end) { | |
789 | struct rb_node *next_node; | |
790 | if (state->state & exclusive_bits) { | |
791 | *failed_start = state->start; | |
792 | err = -EEXIST; | |
793 | goto out; | |
794 | } | |
795 | ||
796 | err = set_state_bits(tree, state, bits); | |
797 | if (err) | |
798 | goto out; | |
799 | ||
800 | cache_state(state, cached_state); | |
801 | merge_state(tree, state); | |
802 | if (last_end == (u64)-1) | |
803 | goto out; | |
804 | ||
805 | start = last_end + 1; | |
806 | if (start < end && prealloc && !need_resched()) { | |
807 | next_node = rb_next(node); | |
808 | if (next_node) { | |
809 | state = rb_entry(next_node, struct extent_state, | |
810 | rb_node); | |
811 | if (state->start == start) | |
812 | goto hit_next; | |
813 | } | |
814 | } | |
815 | goto search_again; | |
816 | } | |
817 | ||
818 | /* | |
819 | * | ---- desired range ---- | | |
820 | * | state | | |
821 | * or | |
822 | * | ------------- state -------------- | | |
823 | * | |
824 | * We need to split the extent we found, and may flip bits on | |
825 | * second half. | |
826 | * | |
827 | * If the extent we found extends past our | |
828 | * range, we just split and search again. It'll get split | |
829 | * again the next time though. | |
830 | * | |
831 | * If the extent we found is inside our range, we set the | |
832 | * desired bit on it. | |
833 | */ | |
834 | if (state->start < start) { | |
835 | if (state->state & exclusive_bits) { | |
836 | *failed_start = start; | |
837 | err = -EEXIST; | |
838 | goto out; | |
839 | } | |
840 | err = split_state(tree, state, prealloc, start); | |
841 | BUG_ON(err == -EEXIST); | |
842 | prealloc = NULL; | |
843 | if (err) | |
844 | goto out; | |
845 | if (state->end <= end) { | |
846 | err = set_state_bits(tree, state, bits); | |
847 | if (err) | |
848 | goto out; | |
849 | cache_state(state, cached_state); | |
850 | merge_state(tree, state); | |
851 | if (last_end == (u64)-1) | |
852 | goto out; | |
853 | start = last_end + 1; | |
854 | } | |
855 | goto search_again; | |
856 | } | |
857 | /* | |
858 | * | ---- desired range ---- | | |
859 | * | state | or | state | | |
860 | * | |
861 | * There's a hole, we need to insert something in it and | |
862 | * ignore the extent we found. | |
863 | */ | |
864 | if (state->start > start) { | |
865 | u64 this_end; | |
866 | if (end < last_start) | |
867 | this_end = end; | |
868 | else | |
869 | this_end = last_start - 1; | |
870 | err = insert_state(tree, prealloc, start, this_end, | |
871 | bits); | |
872 | BUG_ON(err == -EEXIST); | |
873 | if (err) { | |
874 | prealloc = NULL; | |
875 | goto out; | |
876 | } | |
877 | cache_state(prealloc, cached_state); | |
878 | prealloc = NULL; | |
879 | start = this_end + 1; | |
880 | goto search_again; | |
881 | } | |
882 | /* | |
883 | * | ---- desired range ---- | | |
884 | * | state | | |
885 | * We need to split the extent, and set the bit | |
886 | * on the first half | |
887 | */ | |
888 | if (state->start <= end && state->end > end) { | |
889 | if (state->state & exclusive_bits) { | |
890 | *failed_start = start; | |
891 | err = -EEXIST; | |
892 | goto out; | |
893 | } | |
894 | err = split_state(tree, state, prealloc, end + 1); | |
895 | BUG_ON(err == -EEXIST); | |
896 | ||
897 | err = set_state_bits(tree, prealloc, bits); | |
898 | if (err) { | |
899 | prealloc = NULL; | |
900 | goto out; | |
901 | } | |
902 | cache_state(prealloc, cached_state); | |
903 | merge_state(tree, prealloc); | |
904 | prealloc = NULL; | |
905 | goto out; | |
906 | } | |
907 | ||
908 | goto search_again; | |
909 | ||
910 | out: | |
911 | spin_unlock(&tree->lock); | |
912 | if (prealloc) | |
913 | free_extent_state(prealloc); | |
914 | ||
915 | return err; | |
916 | ||
917 | search_again: | |
918 | if (start > end) | |
919 | goto out; | |
920 | spin_unlock(&tree->lock); | |
921 | if (mask & __GFP_WAIT) | |
922 | cond_resched(); | |
923 | goto again; | |
924 | } | |
925 | ||
926 | /* wrappers around set/clear extent bit */ | |
927 | int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, | |
928 | gfp_t mask) | |
929 | { | |
930 | return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL, | |
931 | NULL, mask); | |
932 | } | |
933 | ||
934 | int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, | |
935 | int bits, gfp_t mask) | |
936 | { | |
937 | return set_extent_bit(tree, start, end, bits, 0, NULL, | |
938 | NULL, mask); | |
939 | } | |
940 | ||
941 | int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, | |
942 | int bits, gfp_t mask) | |
943 | { | |
944 | return clear_extent_bit(tree, start, end, bits, 0, 0, NULL, mask); | |
945 | } | |
946 | ||
947 | int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end, | |
948 | gfp_t mask) | |
949 | { | |
950 | return set_extent_bit(tree, start, end, | |
951 | EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE, | |
952 | 0, NULL, NULL, mask); | |
953 | } | |
954 | ||
955 | int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, | |
956 | gfp_t mask) | |
957 | { | |
958 | return clear_extent_bit(tree, start, end, | |
959 | EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, | |
960 | NULL, mask); | |
961 | } | |
962 | ||
963 | int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end, | |
964 | gfp_t mask) | |
965 | { | |
966 | return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL, | |
967 | NULL, mask); | |
968 | } | |
969 | ||
970 | static int clear_extent_new(struct extent_io_tree *tree, u64 start, u64 end, | |
971 | gfp_t mask) | |
972 | { | |
973 | return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0, | |
974 | NULL, mask); | |
975 | } | |
976 | ||
977 | int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, | |
978 | gfp_t mask) | |
979 | { | |
980 | return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, NULL, | |
981 | NULL, mask); | |
982 | } | |
983 | ||
984 | static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, | |
985 | u64 end, gfp_t mask) | |
986 | { | |
987 | return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, | |
988 | NULL, mask); | |
989 | } | |
990 | ||
991 | int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end) | |
992 | { | |
993 | return wait_extent_bit(tree, start, end, EXTENT_WRITEBACK); | |
994 | } | |
995 | ||
996 | /* | |
997 | * either insert or lock state struct between start and end use mask to tell | |
998 | * us if waiting is desired. | |
999 | */ | |
1000 | int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, | |
1001 | int bits, struct extent_state **cached_state, gfp_t mask) | |
1002 | { | |
1003 | int err; | |
1004 | u64 failed_start; | |
1005 | while (1) { | |
1006 | err = set_extent_bit(tree, start, end, EXTENT_LOCKED | bits, | |
1007 | EXTENT_LOCKED, &failed_start, | |
1008 | cached_state, mask); | |
1009 | if (err == -EEXIST && (mask & __GFP_WAIT)) { | |
1010 | wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED); | |
1011 | start = failed_start; | |
1012 | } else { | |
1013 | break; | |
1014 | } | |
1015 | WARN_ON(start > end); | |
1016 | } | |
1017 | return err; | |
1018 | } | |
1019 | ||
1020 | int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask) | |
1021 | { | |
1022 | return lock_extent_bits(tree, start, end, 0, NULL, mask); | |
1023 | } | |
1024 | ||
1025 | int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end, | |
1026 | gfp_t mask) | |
1027 | { | |
1028 | int err; | |
1029 | u64 failed_start; | |
1030 | ||
1031 | err = set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED, | |
1032 | &failed_start, NULL, mask); | |
1033 | if (err == -EEXIST) { | |
1034 | if (failed_start > start) | |
1035 | clear_extent_bit(tree, start, failed_start - 1, | |
1036 | EXTENT_LOCKED, 1, 0, NULL, mask); | |
1037 | return 0; | |
1038 | } | |
1039 | return 1; | |
1040 | } | |
1041 | ||
1042 | int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end, | |
1043 | struct extent_state **cached, gfp_t mask) | |
1044 | { | |
1045 | return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, cached, | |
1046 | mask); | |
1047 | } | |
1048 | ||
1049 | int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, | |
1050 | gfp_t mask) | |
1051 | { | |
1052 | return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL, | |
1053 | mask); | |
1054 | } | |
1055 | ||
1056 | /* | |
1057 | * helper function to set pages and extents in the tree dirty | |
1058 | */ | |
1059 | int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end) | |
1060 | { | |
1061 | unsigned long index = start >> PAGE_CACHE_SHIFT; | |
1062 | unsigned long end_index = end >> PAGE_CACHE_SHIFT; | |
1063 | struct page *page; | |
1064 | ||
1065 | while (index <= end_index) { | |
1066 | page = find_get_page(tree->mapping, index); | |
1067 | BUG_ON(!page); | |
1068 | __set_page_dirty_nobuffers(page); | |
1069 | page_cache_release(page); | |
1070 | index++; | |
1071 | } | |
1072 | return 0; | |
1073 | } | |
1074 | ||
1075 | /* | |
1076 | * helper function to set both pages and extents in the tree writeback | |
1077 | */ | |
1078 | static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end) | |
1079 | { | |
1080 | unsigned long index = start >> PAGE_CACHE_SHIFT; | |
1081 | unsigned long end_index = end >> PAGE_CACHE_SHIFT; | |
1082 | struct page *page; | |
1083 | ||
1084 | while (index <= end_index) { | |
1085 | page = find_get_page(tree->mapping, index); | |
1086 | BUG_ON(!page); | |
1087 | set_page_writeback(page); | |
1088 | page_cache_release(page); | |
1089 | index++; | |
1090 | } | |
1091 | return 0; | |
1092 | } | |
1093 | ||
1094 | /* | |
1095 | * find the first offset in the io tree with 'bits' set. zero is | |
1096 | * returned if we find something, and *start_ret and *end_ret are | |
1097 | * set to reflect the state struct that was found. | |
1098 | * | |
1099 | * If nothing was found, 1 is returned, < 0 on error | |
1100 | */ | |
1101 | int find_first_extent_bit(struct extent_io_tree *tree, u64 start, | |
1102 | u64 *start_ret, u64 *end_ret, int bits) | |
1103 | { | |
1104 | struct rb_node *node; | |
1105 | struct extent_state *state; | |
1106 | int ret = 1; | |
1107 | ||
1108 | spin_lock(&tree->lock); | |
1109 | /* | |
1110 | * this search will find all the extents that end after | |
1111 | * our range starts. | |
1112 | */ | |
1113 | node = tree_search(tree, start); | |
1114 | if (!node) | |
1115 | goto out; | |
1116 | ||
1117 | while (1) { | |
1118 | state = rb_entry(node, struct extent_state, rb_node); | |
1119 | if (state->end >= start && (state->state & bits)) { | |
1120 | *start_ret = state->start; | |
1121 | *end_ret = state->end; | |
1122 | ret = 0; | |
1123 | break; | |
1124 | } | |
1125 | node = rb_next(node); | |
1126 | if (!node) | |
1127 | break; | |
1128 | } | |
1129 | out: | |
1130 | spin_unlock(&tree->lock); | |
1131 | return ret; | |
1132 | } | |
1133 | ||
1134 | /* find the first state struct with 'bits' set after 'start', and | |
1135 | * return it. tree->lock must be held. NULL will returned if | |
1136 | * nothing was found after 'start' | |
1137 | */ | |
1138 | struct extent_state *find_first_extent_bit_state(struct extent_io_tree *tree, | |
1139 | u64 start, int bits) | |
1140 | { | |
1141 | struct rb_node *node; | |
1142 | struct extent_state *state; | |
1143 | ||
1144 | /* | |
1145 | * this search will find all the extents that end after | |
1146 | * our range starts. | |
1147 | */ | |
1148 | node = tree_search(tree, start); | |
1149 | if (!node) | |
1150 | goto out; | |
1151 | ||
1152 | while (1) { | |
1153 | state = rb_entry(node, struct extent_state, rb_node); | |
1154 | if (state->end >= start && (state->state & bits)) | |
1155 | return state; | |
1156 | ||
1157 | node = rb_next(node); | |
1158 | if (!node) | |
1159 | break; | |
1160 | } | |
1161 | out: | |
1162 | return NULL; | |
1163 | } | |
1164 | ||
1165 | /* | |
1166 | * find a contiguous range of bytes in the file marked as delalloc, not | |
1167 | * more than 'max_bytes'. start and end are used to return the range, | |
1168 | * | |
1169 | * 1 is returned if we find something, 0 if nothing was in the tree | |
1170 | */ | |
1171 | static noinline u64 find_delalloc_range(struct extent_io_tree *tree, | |
1172 | u64 *start, u64 *end, u64 max_bytes) | |
1173 | { | |
1174 | struct rb_node *node; | |
1175 | struct extent_state *state; | |
1176 | u64 cur_start = *start; | |
1177 | u64 found = 0; | |
1178 | u64 total_bytes = 0; | |
1179 | ||
1180 | spin_lock(&tree->lock); | |
1181 | ||
1182 | /* | |
1183 | * this search will find all the extents that end after | |
1184 | * our range starts. | |
1185 | */ | |
1186 | node = tree_search(tree, cur_start); | |
1187 | if (!node) { | |
1188 | if (!found) | |
1189 | *end = (u64)-1; | |
1190 | goto out; | |
1191 | } | |
1192 | ||
1193 | while (1) { | |
1194 | state = rb_entry(node, struct extent_state, rb_node); | |
1195 | if (found && (state->start != cur_start || | |
1196 | (state->state & EXTENT_BOUNDARY))) { | |
1197 | goto out; | |
1198 | } | |
1199 | if (!(state->state & EXTENT_DELALLOC)) { | |
1200 | if (!found) | |
1201 | *end = state->end; | |
1202 | goto out; | |
1203 | } | |
1204 | if (!found) | |
1205 | *start = state->start; | |
1206 | found++; | |
1207 | *end = state->end; | |
1208 | cur_start = state->end + 1; | |
1209 | node = rb_next(node); | |
1210 | if (!node) | |
1211 | break; | |
1212 | total_bytes += state->end - state->start + 1; | |
1213 | if (total_bytes >= max_bytes) | |
1214 | break; | |
1215 | } | |
1216 | out: | |
1217 | spin_unlock(&tree->lock); | |
1218 | return found; | |
1219 | } | |
1220 | ||
1221 | static noinline int __unlock_for_delalloc(struct inode *inode, | |
1222 | struct page *locked_page, | |
1223 | u64 start, u64 end) | |
1224 | { | |
1225 | int ret; | |
1226 | struct page *pages[16]; | |
1227 | unsigned long index = start >> PAGE_CACHE_SHIFT; | |
1228 | unsigned long end_index = end >> PAGE_CACHE_SHIFT; | |
1229 | unsigned long nr_pages = end_index - index + 1; | |
1230 | int i; | |
1231 | ||
1232 | if (index == locked_page->index && end_index == index) | |
1233 | return 0; | |
1234 | ||
1235 | while (nr_pages > 0) { | |
1236 | ret = find_get_pages_contig(inode->i_mapping, index, | |
1237 | min_t(unsigned long, nr_pages, | |
1238 | ARRAY_SIZE(pages)), pages); | |
1239 | for (i = 0; i < ret; i++) { | |
1240 | if (pages[i] != locked_page) | |
1241 | unlock_page(pages[i]); | |
1242 | page_cache_release(pages[i]); | |
1243 | } | |
1244 | nr_pages -= ret; | |
1245 | index += ret; | |
1246 | cond_resched(); | |
1247 | } | |
1248 | return 0; | |
1249 | } | |
1250 | ||
1251 | static noinline int lock_delalloc_pages(struct inode *inode, | |
1252 | struct page *locked_page, | |
1253 | u64 delalloc_start, | |
1254 | u64 delalloc_end) | |
1255 | { | |
1256 | unsigned long index = delalloc_start >> PAGE_CACHE_SHIFT; | |
1257 | unsigned long start_index = index; | |
1258 | unsigned long end_index = delalloc_end >> PAGE_CACHE_SHIFT; | |
1259 | unsigned long pages_locked = 0; | |
1260 | struct page *pages[16]; | |
1261 | unsigned long nrpages; | |
1262 | int ret; | |
1263 | int i; | |
1264 | ||
1265 | /* the caller is responsible for locking the start index */ | |
1266 | if (index == locked_page->index && index == end_index) | |
1267 | return 0; | |
1268 | ||
1269 | /* skip the page at the start index */ | |
1270 | nrpages = end_index - index + 1; | |
1271 | while (nrpages > 0) { | |
1272 | ret = find_get_pages_contig(inode->i_mapping, index, | |
1273 | min_t(unsigned long, | |
1274 | nrpages, ARRAY_SIZE(pages)), pages); | |
1275 | if (ret == 0) { | |
1276 | ret = -EAGAIN; | |
1277 | goto done; | |
1278 | } | |
1279 | /* now we have an array of pages, lock them all */ | |
1280 | for (i = 0; i < ret; i++) { | |
1281 | /* | |
1282 | * the caller is taking responsibility for | |
1283 | * locked_page | |
1284 | */ | |
1285 | if (pages[i] != locked_page) { | |
1286 | lock_page(pages[i]); | |
1287 | if (!PageDirty(pages[i]) || | |
1288 | pages[i]->mapping != inode->i_mapping) { | |
1289 | ret = -EAGAIN; | |
1290 | unlock_page(pages[i]); | |
1291 | page_cache_release(pages[i]); | |
1292 | goto done; | |
1293 | } | |
1294 | } | |
1295 | page_cache_release(pages[i]); | |
1296 | pages_locked++; | |
1297 | } | |
1298 | nrpages -= ret; | |
1299 | index += ret; | |
1300 | cond_resched(); | |
1301 | } | |
1302 | ret = 0; | |
1303 | done: | |
1304 | if (ret && pages_locked) { | |
1305 | __unlock_for_delalloc(inode, locked_page, | |
1306 | delalloc_start, | |
1307 | ((u64)(start_index + pages_locked - 1)) << | |
1308 | PAGE_CACHE_SHIFT); | |
1309 | } | |
1310 | return ret; | |
1311 | } | |
1312 | ||
1313 | /* | |
1314 | * find a contiguous range of bytes in the file marked as delalloc, not | |
1315 | * more than 'max_bytes'. start and end are used to return the range, | |
1316 | * | |
1317 | * 1 is returned if we find something, 0 if nothing was in the tree | |
1318 | */ | |
1319 | static noinline u64 find_lock_delalloc_range(struct inode *inode, | |
1320 | struct extent_io_tree *tree, | |
1321 | struct page *locked_page, | |
1322 | u64 *start, u64 *end, | |
1323 | u64 max_bytes) | |
1324 | { | |
1325 | u64 delalloc_start; | |
1326 | u64 delalloc_end; | |
1327 | u64 found; | |
1328 | struct extent_state *cached_state = NULL; | |
1329 | int ret; | |
1330 | int loops = 0; | |
1331 | ||
1332 | again: | |
1333 | /* step one, find a bunch of delalloc bytes starting at start */ | |
1334 | delalloc_start = *start; | |
1335 | delalloc_end = 0; | |
1336 | found = find_delalloc_range(tree, &delalloc_start, &delalloc_end, | |
1337 | max_bytes); | |
1338 | if (!found || delalloc_end <= *start) { | |
1339 | *start = delalloc_start; | |
1340 | *end = delalloc_end; | |
1341 | return found; | |
1342 | } | |
1343 | ||
1344 | /* | |
1345 | * start comes from the offset of locked_page. We have to lock | |
1346 | * pages in order, so we can't process delalloc bytes before | |
1347 | * locked_page | |
1348 | */ | |
1349 | if (delalloc_start < *start) | |
1350 | delalloc_start = *start; | |
1351 | ||
1352 | /* | |
1353 | * make sure to limit the number of pages we try to lock down | |
1354 | * if we're looping. | |
1355 | */ | |
1356 | if (delalloc_end + 1 - delalloc_start > max_bytes && loops) | |
1357 | delalloc_end = delalloc_start + PAGE_CACHE_SIZE - 1; | |
1358 | ||
1359 | /* step two, lock all the pages after the page that has start */ | |
1360 | ret = lock_delalloc_pages(inode, locked_page, | |
1361 | delalloc_start, delalloc_end); | |
1362 | if (ret == -EAGAIN) { | |
1363 | /* some of the pages are gone, lets avoid looping by | |
1364 | * shortening the size of the delalloc range we're searching | |
1365 | */ | |
1366 | free_extent_state(cached_state); | |
1367 | if (!loops) { | |
1368 | unsigned long offset = (*start) & (PAGE_CACHE_SIZE - 1); | |
1369 | max_bytes = PAGE_CACHE_SIZE - offset; | |
1370 | loops = 1; | |
1371 | goto again; | |
1372 | } else { | |
1373 | found = 0; | |
1374 | goto out_failed; | |
1375 | } | |
1376 | } | |
1377 | BUG_ON(ret); | |
1378 | ||
1379 | /* step three, lock the state bits for the whole range */ | |
1380 | lock_extent_bits(tree, delalloc_start, delalloc_end, | |
1381 | 0, &cached_state, GFP_NOFS); | |
1382 | ||
1383 | /* then test to make sure it is all still delalloc */ | |
1384 | ret = test_range_bit(tree, delalloc_start, delalloc_end, | |
1385 | EXTENT_DELALLOC, 1, cached_state); | |
1386 | if (!ret) { | |
1387 | unlock_extent_cached(tree, delalloc_start, delalloc_end, | |
1388 | &cached_state, GFP_NOFS); | |
1389 | __unlock_for_delalloc(inode, locked_page, | |
1390 | delalloc_start, delalloc_end); | |
1391 | cond_resched(); | |
1392 | goto again; | |
1393 | } | |
1394 | free_extent_state(cached_state); | |
1395 | *start = delalloc_start; | |
1396 | *end = delalloc_end; | |
1397 | out_failed: | |
1398 | return found; | |
1399 | } | |
1400 | ||
1401 | int extent_clear_unlock_delalloc(struct inode *inode, | |
1402 | struct extent_io_tree *tree, | |
1403 | u64 start, u64 end, struct page *locked_page, | |
1404 | int unlock_pages, | |
1405 | int clear_unlock, | |
1406 | int clear_delalloc, int clear_dirty, | |
1407 | int set_writeback, | |
1408 | int end_writeback, | |
1409 | int set_private2) | |
1410 | { | |
1411 | int ret; | |
1412 | struct page *pages[16]; | |
1413 | unsigned long index = start >> PAGE_CACHE_SHIFT; | |
1414 | unsigned long end_index = end >> PAGE_CACHE_SHIFT; | |
1415 | unsigned long nr_pages = end_index - index + 1; | |
1416 | int i; | |
1417 | int clear_bits = 0; | |
1418 | ||
1419 | if (clear_unlock) | |
1420 | clear_bits |= EXTENT_LOCKED; | |
1421 | if (clear_dirty) | |
1422 | clear_bits |= EXTENT_DIRTY; | |
1423 | ||
1424 | if (clear_delalloc) | |
1425 | clear_bits |= EXTENT_DELALLOC; | |
1426 | ||
1427 | clear_extent_bit(tree, start, end, clear_bits, 1, 0, NULL, GFP_NOFS); | |
1428 | if (!(unlock_pages || clear_dirty || set_writeback || end_writeback || | |
1429 | set_private2)) | |
1430 | return 0; | |
1431 | ||
1432 | while (nr_pages > 0) { | |
1433 | ret = find_get_pages_contig(inode->i_mapping, index, | |
1434 | min_t(unsigned long, | |
1435 | nr_pages, ARRAY_SIZE(pages)), pages); | |
1436 | for (i = 0; i < ret; i++) { | |
1437 | ||
1438 | if (set_private2) | |
1439 | SetPagePrivate2(pages[i]); | |
1440 | ||
1441 | if (pages[i] == locked_page) { | |
1442 | page_cache_release(pages[i]); | |
1443 | continue; | |
1444 | } | |
1445 | if (clear_dirty) | |
1446 | clear_page_dirty_for_io(pages[i]); | |
1447 | if (set_writeback) | |
1448 | set_page_writeback(pages[i]); | |
1449 | if (end_writeback) | |
1450 | end_page_writeback(pages[i]); | |
1451 | if (unlock_pages) | |
1452 | unlock_page(pages[i]); | |
1453 | page_cache_release(pages[i]); | |
1454 | } | |
1455 | nr_pages -= ret; | |
1456 | index += ret; | |
1457 | cond_resched(); | |
1458 | } | |
1459 | return 0; | |
1460 | } | |
1461 | ||
1462 | /* | |
1463 | * count the number of bytes in the tree that have a given bit(s) | |
1464 | * set. This can be fairly slow, except for EXTENT_DIRTY which is | |
1465 | * cached. The total number found is returned. | |
1466 | */ | |
1467 | u64 count_range_bits(struct extent_io_tree *tree, | |
1468 | u64 *start, u64 search_end, u64 max_bytes, | |
1469 | unsigned long bits) | |
1470 | { | |
1471 | struct rb_node *node; | |
1472 | struct extent_state *state; | |
1473 | u64 cur_start = *start; | |
1474 | u64 total_bytes = 0; | |
1475 | int found = 0; | |
1476 | ||
1477 | if (search_end <= cur_start) { | |
1478 | WARN_ON(1); | |
1479 | return 0; | |
1480 | } | |
1481 | ||
1482 | spin_lock(&tree->lock); | |
1483 | if (cur_start == 0 && bits == EXTENT_DIRTY) { | |
1484 | total_bytes = tree->dirty_bytes; | |
1485 | goto out; | |
1486 | } | |
1487 | /* | |
1488 | * this search will find all the extents that end after | |
1489 | * our range starts. | |
1490 | */ | |
1491 | node = tree_search(tree, cur_start); | |
1492 | if (!node) | |
1493 | goto out; | |
1494 | ||
1495 | while (1) { | |
1496 | state = rb_entry(node, struct extent_state, rb_node); | |
1497 | if (state->start > search_end) | |
1498 | break; | |
1499 | if (state->end >= cur_start && (state->state & bits)) { | |
1500 | total_bytes += min(search_end, state->end) + 1 - | |
1501 | max(cur_start, state->start); | |
1502 | if (total_bytes >= max_bytes) | |
1503 | break; | |
1504 | if (!found) { | |
1505 | *start = state->start; | |
1506 | found = 1; | |
1507 | } | |
1508 | } | |
1509 | node = rb_next(node); | |
1510 | if (!node) | |
1511 | break; | |
1512 | } | |
1513 | out: | |
1514 | spin_unlock(&tree->lock); | |
1515 | return total_bytes; | |
1516 | } | |
1517 | ||
1518 | /* | |
1519 | * set the private field for a given byte offset in the tree. If there isn't | |
1520 | * an extent_state there already, this does nothing. | |
1521 | */ | |
1522 | int set_state_private(struct extent_io_tree *tree, u64 start, u64 private) | |
1523 | { | |
1524 | struct rb_node *node; | |
1525 | struct extent_state *state; | |
1526 | int ret = 0; | |
1527 | ||
1528 | spin_lock(&tree->lock); | |
1529 | /* | |
1530 | * this search will find all the extents that end after | |
1531 | * our range starts. | |
1532 | */ | |
1533 | node = tree_search(tree, start); | |
1534 | if (!node) { | |
1535 | ret = -ENOENT; | |
1536 | goto out; | |
1537 | } | |
1538 | state = rb_entry(node, struct extent_state, rb_node); | |
1539 | if (state->start != start) { | |
1540 | ret = -ENOENT; | |
1541 | goto out; | |
1542 | } | |
1543 | state->private = private; | |
1544 | out: | |
1545 | spin_unlock(&tree->lock); | |
1546 | return ret; | |
1547 | } | |
1548 | ||
1549 | int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private) | |
1550 | { | |
1551 | struct rb_node *node; | |
1552 | struct extent_state *state; | |
1553 | int ret = 0; | |
1554 | ||
1555 | spin_lock(&tree->lock); | |
1556 | /* | |
1557 | * this search will find all the extents that end after | |
1558 | * our range starts. | |
1559 | */ | |
1560 | node = tree_search(tree, start); | |
1561 | if (!node) { | |
1562 | ret = -ENOENT; | |
1563 | goto out; | |
1564 | } | |
1565 | state = rb_entry(node, struct extent_state, rb_node); | |
1566 | if (state->start != start) { | |
1567 | ret = -ENOENT; | |
1568 | goto out; | |
1569 | } | |
1570 | *private = state->private; | |
1571 | out: | |
1572 | spin_unlock(&tree->lock); | |
1573 | return ret; | |
1574 | } | |
1575 | ||
1576 | /* | |
1577 | * searches a range in the state tree for a given mask. | |
1578 | * If 'filled' == 1, this returns 1 only if every extent in the tree | |
1579 | * has the bits set. Otherwise, 1 is returned if any bit in the | |
1580 | * range is found set. | |
1581 | */ | |
1582 | int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end, | |
1583 | int bits, int filled, struct extent_state *cached) | |
1584 | { | |
1585 | struct extent_state *state = NULL; | |
1586 | struct rb_node *node; | |
1587 | int bitset = 0; | |
1588 | ||
1589 | spin_lock(&tree->lock); | |
1590 | if (cached && cached->tree && cached->start == start) | |
1591 | node = &cached->rb_node; | |
1592 | else | |
1593 | node = tree_search(tree, start); | |
1594 | while (node && start <= end) { | |
1595 | state = rb_entry(node, struct extent_state, rb_node); | |
1596 | ||
1597 | if (filled && state->start > start) { | |
1598 | bitset = 0; | |
1599 | break; | |
1600 | } | |
1601 | ||
1602 | if (state->start > end) | |
1603 | break; | |
1604 | ||
1605 | if (state->state & bits) { | |
1606 | bitset = 1; | |
1607 | if (!filled) | |
1608 | break; | |
1609 | } else if (filled) { | |
1610 | bitset = 0; | |
1611 | break; | |
1612 | } | |
1613 | ||
1614 | if (state->end == (u64)-1) | |
1615 | break; | |
1616 | ||
1617 | start = state->end + 1; | |
1618 | if (start > end) | |
1619 | break; | |
1620 | node = rb_next(node); | |
1621 | if (!node) { | |
1622 | if (filled) | |
1623 | bitset = 0; | |
1624 | break; | |
1625 | } | |
1626 | } | |
1627 | spin_unlock(&tree->lock); | |
1628 | return bitset; | |
1629 | } | |
1630 | ||
1631 | /* | |
1632 | * helper function to set a given page up to date if all the | |
1633 | * extents in the tree for that page are up to date | |
1634 | */ | |
1635 | static int check_page_uptodate(struct extent_io_tree *tree, | |
1636 | struct page *page) | |
1637 | { | |
1638 | u64 start = (u64)page->index << PAGE_CACHE_SHIFT; | |
1639 | u64 end = start + PAGE_CACHE_SIZE - 1; | |
1640 | if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL)) | |
1641 | SetPageUptodate(page); | |
1642 | return 0; | |
1643 | } | |
1644 | ||
1645 | /* | |
1646 | * helper function to unlock a page if all the extents in the tree | |
1647 | * for that page are unlocked | |
1648 | */ | |
1649 | static int check_page_locked(struct extent_io_tree *tree, | |
1650 | struct page *page) | |
1651 | { | |
1652 | u64 start = (u64)page->index << PAGE_CACHE_SHIFT; | |
1653 | u64 end = start + PAGE_CACHE_SIZE - 1; | |
1654 | if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL)) | |
1655 | unlock_page(page); | |
1656 | return 0; | |
1657 | } | |
1658 | ||
1659 | /* | |
1660 | * helper function to end page writeback if all the extents | |
1661 | * in the tree for that page are done with writeback | |
1662 | */ | |
1663 | static int check_page_writeback(struct extent_io_tree *tree, | |
1664 | struct page *page) | |
1665 | { | |
1666 | end_page_writeback(page); | |
1667 | return 0; | |
1668 | } | |
1669 | ||
1670 | /* lots and lots of room for performance fixes in the end_bio funcs */ | |
1671 | ||
1672 | /* | |
1673 | * after a writepage IO is done, we need to: | |
1674 | * clear the uptodate bits on error | |
1675 | * clear the writeback bits in the extent tree for this IO | |
1676 | * end_page_writeback if the page has no more pending IO | |
1677 | * | |
1678 | * Scheduling is not allowed, so the extent state tree is expected | |
1679 | * to have one and only one object corresponding to this IO. | |
1680 | */ | |
1681 | static void end_bio_extent_writepage(struct bio *bio, int err) | |
1682 | { | |
1683 | int uptodate = err == 0; | |
1684 | struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; | |
1685 | struct extent_io_tree *tree; | |
1686 | u64 start; | |
1687 | u64 end; | |
1688 | int whole_page; | |
1689 | int ret; | |
1690 | ||
1691 | do { | |
1692 | struct page *page = bvec->bv_page; | |
1693 | tree = &BTRFS_I(page->mapping->host)->io_tree; | |
1694 | ||
1695 | start = ((u64)page->index << PAGE_CACHE_SHIFT) + | |
1696 | bvec->bv_offset; | |
1697 | end = start + bvec->bv_len - 1; | |
1698 | ||
1699 | if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE) | |
1700 | whole_page = 1; | |
1701 | else | |
1702 | whole_page = 0; | |
1703 | ||
1704 | if (--bvec >= bio->bi_io_vec) | |
1705 | prefetchw(&bvec->bv_page->flags); | |
1706 | if (tree->ops && tree->ops->writepage_end_io_hook) { | |
1707 | ret = tree->ops->writepage_end_io_hook(page, start, | |
1708 | end, NULL, uptodate); | |
1709 | if (ret) | |
1710 | uptodate = 0; | |
1711 | } | |
1712 | ||
1713 | if (!uptodate && tree->ops && | |
1714 | tree->ops->writepage_io_failed_hook) { | |
1715 | ret = tree->ops->writepage_io_failed_hook(bio, page, | |
1716 | start, end, NULL); | |
1717 | if (ret == 0) { | |
1718 | uptodate = (err == 0); | |
1719 | continue; | |
1720 | } | |
1721 | } | |
1722 | ||
1723 | if (!uptodate) { | |
1724 | clear_extent_uptodate(tree, start, end, GFP_NOFS); | |
1725 | ClearPageUptodate(page); | |
1726 | SetPageError(page); | |
1727 | } | |
1728 | ||
1729 | if (whole_page) | |
1730 | end_page_writeback(page); | |
1731 | else | |
1732 | check_page_writeback(tree, page); | |
1733 | } while (bvec >= bio->bi_io_vec); | |
1734 | ||
1735 | bio_put(bio); | |
1736 | } | |
1737 | ||
1738 | /* | |
1739 | * after a readpage IO is done, we need to: | |
1740 | * clear the uptodate bits on error | |
1741 | * set the uptodate bits if things worked | |
1742 | * set the page up to date if all extents in the tree are uptodate | |
1743 | * clear the lock bit in the extent tree | |
1744 | * unlock the page if there are no other extents locked for it | |
1745 | * | |
1746 | * Scheduling is not allowed, so the extent state tree is expected | |
1747 | * to have one and only one object corresponding to this IO. | |
1748 | */ | |
1749 | static void end_bio_extent_readpage(struct bio *bio, int err) | |
1750 | { | |
1751 | int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); | |
1752 | struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; | |
1753 | struct extent_io_tree *tree; | |
1754 | u64 start; | |
1755 | u64 end; | |
1756 | int whole_page; | |
1757 | int ret; | |
1758 | ||
1759 | if (err) | |
1760 | uptodate = 0; | |
1761 | ||
1762 | do { | |
1763 | struct page *page = bvec->bv_page; | |
1764 | tree = &BTRFS_I(page->mapping->host)->io_tree; | |
1765 | ||
1766 | start = ((u64)page->index << PAGE_CACHE_SHIFT) + | |
1767 | bvec->bv_offset; | |
1768 | end = start + bvec->bv_len - 1; | |
1769 | ||
1770 | if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE) | |
1771 | whole_page = 1; | |
1772 | else | |
1773 | whole_page = 0; | |
1774 | ||
1775 | if (--bvec >= bio->bi_io_vec) | |
1776 | prefetchw(&bvec->bv_page->flags); | |
1777 | ||
1778 | if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) { | |
1779 | ret = tree->ops->readpage_end_io_hook(page, start, end, | |
1780 | NULL); | |
1781 | if (ret) | |
1782 | uptodate = 0; | |
1783 | } | |
1784 | if (!uptodate && tree->ops && | |
1785 | tree->ops->readpage_io_failed_hook) { | |
1786 | ret = tree->ops->readpage_io_failed_hook(bio, page, | |
1787 | start, end, NULL); | |
1788 | if (ret == 0) { | |
1789 | uptodate = | |
1790 | test_bit(BIO_UPTODATE, &bio->bi_flags); | |
1791 | if (err) | |
1792 | uptodate = 0; | |
1793 | continue; | |
1794 | } | |
1795 | } | |
1796 | ||
1797 | if (uptodate) { | |
1798 | set_extent_uptodate(tree, start, end, | |
1799 | GFP_ATOMIC); | |
1800 | } | |
1801 | unlock_extent(tree, start, end, GFP_ATOMIC); | |
1802 | ||
1803 | if (whole_page) { | |
1804 | if (uptodate) { | |
1805 | SetPageUptodate(page); | |
1806 | } else { | |
1807 | ClearPageUptodate(page); | |
1808 | SetPageError(page); | |
1809 | } | |
1810 | unlock_page(page); | |
1811 | } else { | |
1812 | if (uptodate) { | |
1813 | check_page_uptodate(tree, page); | |
1814 | } else { | |
1815 | ClearPageUptodate(page); | |
1816 | SetPageError(page); | |
1817 | } | |
1818 | check_page_locked(tree, page); | |
1819 | } | |
1820 | } while (bvec >= bio->bi_io_vec); | |
1821 | ||
1822 | bio_put(bio); | |
1823 | } | |
1824 | ||
1825 | /* | |
1826 | * IO done from prepare_write is pretty simple, we just unlock | |
1827 | * the structs in the extent tree when done, and set the uptodate bits | |
1828 | * as appropriate. | |
1829 | */ | |
1830 | static void end_bio_extent_preparewrite(struct bio *bio, int err) | |
1831 | { | |
1832 | const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); | |
1833 | struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; | |
1834 | struct extent_io_tree *tree; | |
1835 | u64 start; | |
1836 | u64 end; | |
1837 | ||
1838 | do { | |
1839 | struct page *page = bvec->bv_page; | |
1840 | tree = &BTRFS_I(page->mapping->host)->io_tree; | |
1841 | ||
1842 | start = ((u64)page->index << PAGE_CACHE_SHIFT) + | |
1843 | bvec->bv_offset; | |
1844 | end = start + bvec->bv_len - 1; | |
1845 | ||
1846 | if (--bvec >= bio->bi_io_vec) | |
1847 | prefetchw(&bvec->bv_page->flags); | |
1848 | ||
1849 | if (uptodate) { | |
1850 | set_extent_uptodate(tree, start, end, GFP_ATOMIC); | |
1851 | } else { | |
1852 | ClearPageUptodate(page); | |
1853 | SetPageError(page); | |
1854 | } | |
1855 | ||
1856 | unlock_extent(tree, start, end, GFP_ATOMIC); | |
1857 | ||
1858 | } while (bvec >= bio->bi_io_vec); | |
1859 | ||
1860 | bio_put(bio); | |
1861 | } | |
1862 | ||
1863 | static struct bio * | |
1864 | extent_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs, | |
1865 | gfp_t gfp_flags) | |
1866 | { | |
1867 | struct bio *bio; | |
1868 | ||
1869 | bio = bio_alloc(gfp_flags, nr_vecs); | |
1870 | ||
1871 | if (bio == NULL && (current->flags & PF_MEMALLOC)) { | |
1872 | while (!bio && (nr_vecs /= 2)) | |
1873 | bio = bio_alloc(gfp_flags, nr_vecs); | |
1874 | } | |
1875 | ||
1876 | if (bio) { | |
1877 | bio->bi_size = 0; | |
1878 | bio->bi_bdev = bdev; | |
1879 | bio->bi_sector = first_sector; | |
1880 | } | |
1881 | return bio; | |
1882 | } | |
1883 | ||
1884 | static int submit_one_bio(int rw, struct bio *bio, int mirror_num, | |
1885 | unsigned long bio_flags) | |
1886 | { | |
1887 | int ret = 0; | |
1888 | struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; | |
1889 | struct page *page = bvec->bv_page; | |
1890 | struct extent_io_tree *tree = bio->bi_private; | |
1891 | u64 start; | |
1892 | u64 end; | |
1893 | ||
1894 | start = ((u64)page->index << PAGE_CACHE_SHIFT) + bvec->bv_offset; | |
1895 | end = start + bvec->bv_len - 1; | |
1896 | ||
1897 | bio->bi_private = NULL; | |
1898 | ||
1899 | bio_get(bio); | |
1900 | ||
1901 | if (tree->ops && tree->ops->submit_bio_hook) | |
1902 | tree->ops->submit_bio_hook(page->mapping->host, rw, bio, | |
1903 | mirror_num, bio_flags); | |
1904 | else | |
1905 | submit_bio(rw, bio); | |
1906 | if (bio_flagged(bio, BIO_EOPNOTSUPP)) | |
1907 | ret = -EOPNOTSUPP; | |
1908 | bio_put(bio); | |
1909 | return ret; | |
1910 | } | |
1911 | ||
1912 | static int submit_extent_page(int rw, struct extent_io_tree *tree, | |
1913 | struct page *page, sector_t sector, | |
1914 | size_t size, unsigned long offset, | |
1915 | struct block_device *bdev, | |
1916 | struct bio **bio_ret, | |
1917 | unsigned long max_pages, | |
1918 | bio_end_io_t end_io_func, | |
1919 | int mirror_num, | |
1920 | unsigned long prev_bio_flags, | |
1921 | unsigned long bio_flags) | |
1922 | { | |
1923 | int ret = 0; | |
1924 | struct bio *bio; | |
1925 | int nr; | |
1926 | int contig = 0; | |
1927 | int this_compressed = bio_flags & EXTENT_BIO_COMPRESSED; | |
1928 | int old_compressed = prev_bio_flags & EXTENT_BIO_COMPRESSED; | |
1929 | size_t page_size = min_t(size_t, size, PAGE_CACHE_SIZE); | |
1930 | ||
1931 | if (bio_ret && *bio_ret) { | |
1932 | bio = *bio_ret; | |
1933 | if (old_compressed) | |
1934 | contig = bio->bi_sector == sector; | |
1935 | else | |
1936 | contig = bio->bi_sector + (bio->bi_size >> 9) == | |
1937 | sector; | |
1938 | ||
1939 | if (prev_bio_flags != bio_flags || !contig || | |
1940 | (tree->ops && tree->ops->merge_bio_hook && | |
1941 | tree->ops->merge_bio_hook(page, offset, page_size, bio, | |
1942 | bio_flags)) || | |
1943 | bio_add_page(bio, page, page_size, offset) < page_size) { | |
1944 | ret = submit_one_bio(rw, bio, mirror_num, | |
1945 | prev_bio_flags); | |
1946 | bio = NULL; | |
1947 | } else { | |
1948 | return 0; | |
1949 | } | |
1950 | } | |
1951 | if (this_compressed) | |
1952 | nr = BIO_MAX_PAGES; | |
1953 | else | |
1954 | nr = bio_get_nr_vecs(bdev); | |
1955 | ||
1956 | bio = extent_bio_alloc(bdev, sector, nr, GFP_NOFS | __GFP_HIGH); | |
1957 | ||
1958 | bio_add_page(bio, page, page_size, offset); | |
1959 | bio->bi_end_io = end_io_func; | |
1960 | bio->bi_private = tree; | |
1961 | ||
1962 | if (bio_ret) | |
1963 | *bio_ret = bio; | |
1964 | else | |
1965 | ret = submit_one_bio(rw, bio, mirror_num, bio_flags); | |
1966 | ||
1967 | return ret; | |
1968 | } | |
1969 | ||
1970 | void set_page_extent_mapped(struct page *page) | |
1971 | { | |
1972 | if (!PagePrivate(page)) { | |
1973 | SetPagePrivate(page); | |
1974 | page_cache_get(page); | |
1975 | set_page_private(page, EXTENT_PAGE_PRIVATE); | |
1976 | } | |
1977 | } | |
1978 | ||
1979 | static void set_page_extent_head(struct page *page, unsigned long len) | |
1980 | { | |
1981 | set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2); | |
1982 | } | |
1983 | ||
1984 | /* | |
1985 | * basic readpage implementation. Locked extent state structs are inserted | |
1986 | * into the tree that are removed when the IO is done (by the end_io | |
1987 | * handlers) | |
1988 | */ | |
1989 | static int __extent_read_full_page(struct extent_io_tree *tree, | |
1990 | struct page *page, | |
1991 | get_extent_t *get_extent, | |
1992 | struct bio **bio, int mirror_num, | |
1993 | unsigned long *bio_flags) | |
1994 | { | |
1995 | struct inode *inode = page->mapping->host; | |
1996 | u64 start = (u64)page->index << PAGE_CACHE_SHIFT; | |
1997 | u64 page_end = start + PAGE_CACHE_SIZE - 1; | |
1998 | u64 end; | |
1999 | u64 cur = start; | |
2000 | u64 extent_offset; | |
2001 | u64 last_byte = i_size_read(inode); | |
2002 | u64 block_start; | |
2003 | u64 cur_end; | |
2004 | sector_t sector; | |
2005 | struct extent_map *em; | |
2006 | struct block_device *bdev; | |
2007 | int ret; | |
2008 | int nr = 0; | |
2009 | size_t page_offset = 0; | |
2010 | size_t iosize; | |
2011 | size_t disk_io_size; | |
2012 | size_t blocksize = inode->i_sb->s_blocksize; | |
2013 | unsigned long this_bio_flag = 0; | |
2014 | ||
2015 | set_page_extent_mapped(page); | |
2016 | ||
2017 | end = page_end; | |
2018 | lock_extent(tree, start, end, GFP_NOFS); | |
2019 | ||
2020 | if (page->index == last_byte >> PAGE_CACHE_SHIFT) { | |
2021 | char *userpage; | |
2022 | size_t zero_offset = last_byte & (PAGE_CACHE_SIZE - 1); | |
2023 | ||
2024 | if (zero_offset) { | |
2025 | iosize = PAGE_CACHE_SIZE - zero_offset; | |
2026 | userpage = kmap_atomic(page, KM_USER0); | |
2027 | memset(userpage + zero_offset, 0, iosize); | |
2028 | flush_dcache_page(page); | |
2029 | kunmap_atomic(userpage, KM_USER0); | |
2030 | } | |
2031 | } | |
2032 | while (cur <= end) { | |
2033 | if (cur >= last_byte) { | |
2034 | char *userpage; | |
2035 | iosize = PAGE_CACHE_SIZE - page_offset; | |
2036 | userpage = kmap_atomic(page, KM_USER0); | |
2037 | memset(userpage + page_offset, 0, iosize); | |
2038 | flush_dcache_page(page); | |
2039 | kunmap_atomic(userpage, KM_USER0); | |
2040 | set_extent_uptodate(tree, cur, cur + iosize - 1, | |
2041 | GFP_NOFS); | |
2042 | unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); | |
2043 | break; | |
2044 | } | |
2045 | em = get_extent(inode, page, page_offset, cur, | |
2046 | end - cur + 1, 0); | |
2047 | if (IS_ERR(em) || !em) { | |
2048 | SetPageError(page); | |
2049 | unlock_extent(tree, cur, end, GFP_NOFS); | |
2050 | break; | |
2051 | } | |
2052 | extent_offset = cur - em->start; | |
2053 | BUG_ON(extent_map_end(em) <= cur); | |
2054 | BUG_ON(end < cur); | |
2055 | ||
2056 | if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) | |
2057 | this_bio_flag = EXTENT_BIO_COMPRESSED; | |
2058 | ||
2059 | iosize = min(extent_map_end(em) - cur, end - cur + 1); | |
2060 | cur_end = min(extent_map_end(em) - 1, end); | |
2061 | iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1); | |
2062 | if (this_bio_flag & EXTENT_BIO_COMPRESSED) { | |
2063 | disk_io_size = em->block_len; | |
2064 | sector = em->block_start >> 9; | |
2065 | } else { | |
2066 | sector = (em->block_start + extent_offset) >> 9; | |
2067 | disk_io_size = iosize; | |
2068 | } | |
2069 | bdev = em->bdev; | |
2070 | block_start = em->block_start; | |
2071 | if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) | |
2072 | block_start = EXTENT_MAP_HOLE; | |
2073 | free_extent_map(em); | |
2074 | em = NULL; | |
2075 | ||
2076 | /* we've found a hole, just zero and go on */ | |
2077 | if (block_start == EXTENT_MAP_HOLE) { | |
2078 | char *userpage; | |
2079 | userpage = kmap_atomic(page, KM_USER0); | |
2080 | memset(userpage + page_offset, 0, iosize); | |
2081 | flush_dcache_page(page); | |
2082 | kunmap_atomic(userpage, KM_USER0); | |
2083 | ||
2084 | set_extent_uptodate(tree, cur, cur + iosize - 1, | |
2085 | GFP_NOFS); | |
2086 | unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); | |
2087 | cur = cur + iosize; | |
2088 | page_offset += iosize; | |
2089 | continue; | |
2090 | } | |
2091 | /* the get_extent function already copied into the page */ | |
2092 | if (test_range_bit(tree, cur, cur_end, | |
2093 | EXTENT_UPTODATE, 1, NULL)) { | |
2094 | check_page_uptodate(tree, page); | |
2095 | unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); | |
2096 | cur = cur + iosize; | |
2097 | page_offset += iosize; | |
2098 | continue; | |
2099 | } | |
2100 | /* we have an inline extent but it didn't get marked up | |
2101 | * to date. Error out | |
2102 | */ | |
2103 | if (block_start == EXTENT_MAP_INLINE) { | |
2104 | SetPageError(page); | |
2105 | unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS); | |
2106 | cur = cur + iosize; | |
2107 | page_offset += iosize; | |
2108 | continue; | |
2109 | } | |
2110 | ||
2111 | ret = 0; | |
2112 | if (tree->ops && tree->ops->readpage_io_hook) { | |
2113 | ret = tree->ops->readpage_io_hook(page, cur, | |
2114 | cur + iosize - 1); | |
2115 | } | |
2116 | if (!ret) { | |
2117 | unsigned long pnr = (last_byte >> PAGE_CACHE_SHIFT) + 1; | |
2118 | pnr -= page->index; | |
2119 | ret = submit_extent_page(READ, tree, page, | |
2120 | sector, disk_io_size, page_offset, | |
2121 | bdev, bio, pnr, | |
2122 | end_bio_extent_readpage, mirror_num, | |
2123 | *bio_flags, | |
2124 | this_bio_flag); | |
2125 | nr++; | |
2126 | *bio_flags = this_bio_flag; | |
2127 | } | |
2128 | if (ret) | |
2129 | SetPageError(page); | |
2130 | cur = cur + iosize; | |
2131 | page_offset += iosize; | |
2132 | } | |
2133 | if (!nr) { | |
2134 | if (!PageError(page)) | |
2135 | SetPageUptodate(page); | |
2136 | unlock_page(page); | |
2137 | } | |
2138 | return 0; | |
2139 | } | |
2140 | ||
2141 | int extent_read_full_page(struct extent_io_tree *tree, struct page *page, | |
2142 | get_extent_t *get_extent) | |
2143 | { | |
2144 | struct bio *bio = NULL; | |
2145 | unsigned long bio_flags = 0; | |
2146 | int ret; | |
2147 | ||
2148 | ret = __extent_read_full_page(tree, page, get_extent, &bio, 0, | |
2149 | &bio_flags); | |
2150 | if (bio) | |
2151 | submit_one_bio(READ, bio, 0, bio_flags); | |
2152 | return ret; | |
2153 | } | |
2154 | ||
2155 | static noinline void update_nr_written(struct page *page, | |
2156 | struct writeback_control *wbc, | |
2157 | unsigned long nr_written) | |
2158 | { | |
2159 | wbc->nr_to_write -= nr_written; | |
2160 | if (wbc->range_cyclic || (wbc->nr_to_write > 0 && | |
2161 | wbc->range_start == 0 && wbc->range_end == LLONG_MAX)) | |
2162 | page->mapping->writeback_index = page->index + nr_written; | |
2163 | } | |
2164 | ||
2165 | /* | |
2166 | * the writepage semantics are similar to regular writepage. extent | |
2167 | * records are inserted to lock ranges in the tree, and as dirty areas | |
2168 | * are found, they are marked writeback. Then the lock bits are removed | |
2169 | * and the end_io handler clears the writeback ranges | |
2170 | */ | |
2171 | static int __extent_writepage(struct page *page, struct writeback_control *wbc, | |
2172 | void *data) | |
2173 | { | |
2174 | struct inode *inode = page->mapping->host; | |
2175 | struct extent_page_data *epd = data; | |
2176 | struct extent_io_tree *tree = epd->tree; | |
2177 | u64 start = (u64)page->index << PAGE_CACHE_SHIFT; | |
2178 | u64 delalloc_start; | |
2179 | u64 page_end = start + PAGE_CACHE_SIZE - 1; | |
2180 | u64 end; | |
2181 | u64 cur = start; | |
2182 | u64 extent_offset; | |
2183 | u64 last_byte = i_size_read(inode); | |
2184 | u64 block_start; | |
2185 | u64 iosize; | |
2186 | u64 unlock_start; | |
2187 | sector_t sector; | |
2188 | struct extent_state *cached_state = NULL; | |
2189 | struct extent_map *em; | |
2190 | struct block_device *bdev; | |
2191 | int ret; | |
2192 | int nr = 0; | |
2193 | size_t pg_offset = 0; | |
2194 | size_t blocksize; | |
2195 | loff_t i_size = i_size_read(inode); | |
2196 | unsigned long end_index = i_size >> PAGE_CACHE_SHIFT; | |
2197 | u64 nr_delalloc; | |
2198 | u64 delalloc_end; | |
2199 | int page_started; | |
2200 | int compressed; | |
2201 | int write_flags; | |
2202 | unsigned long nr_written = 0; | |
2203 | ||
2204 | if (wbc->sync_mode == WB_SYNC_ALL) | |
2205 | write_flags = WRITE_SYNC_PLUG; | |
2206 | else | |
2207 | write_flags = WRITE; | |
2208 | ||
2209 | WARN_ON(!PageLocked(page)); | |
2210 | pg_offset = i_size & (PAGE_CACHE_SIZE - 1); | |
2211 | if (page->index > end_index || | |
2212 | (page->index == end_index && !pg_offset)) { | |
2213 | page->mapping->a_ops->invalidatepage(page, 0); | |
2214 | unlock_page(page); | |
2215 | return 0; | |
2216 | } | |
2217 | ||
2218 | if (page->index == end_index) { | |
2219 | char *userpage; | |
2220 | ||
2221 | userpage = kmap_atomic(page, KM_USER0); | |
2222 | memset(userpage + pg_offset, 0, | |
2223 | PAGE_CACHE_SIZE - pg_offset); | |
2224 | kunmap_atomic(userpage, KM_USER0); | |
2225 | flush_dcache_page(page); | |
2226 | } | |
2227 | pg_offset = 0; | |
2228 | ||
2229 | set_page_extent_mapped(page); | |
2230 | ||
2231 | delalloc_start = start; | |
2232 | delalloc_end = 0; | |
2233 | page_started = 0; | |
2234 | if (!epd->extent_locked) { | |
2235 | u64 delalloc_to_write = 0; | |
2236 | /* | |
2237 | * make sure the wbc mapping index is at least updated | |
2238 | * to this page. | |
2239 | */ | |
2240 | update_nr_written(page, wbc, 0); | |
2241 | ||
2242 | while (delalloc_end < page_end) { | |
2243 | nr_delalloc = find_lock_delalloc_range(inode, tree, | |
2244 | page, | |
2245 | &delalloc_start, | |
2246 | &delalloc_end, | |
2247 | 128 * 1024 * 1024); | |
2248 | if (nr_delalloc == 0) { | |
2249 | delalloc_start = delalloc_end + 1; | |
2250 | continue; | |
2251 | } | |
2252 | tree->ops->fill_delalloc(inode, page, delalloc_start, | |
2253 | delalloc_end, &page_started, | |
2254 | &nr_written); | |
2255 | /* | |
2256 | * delalloc_end is already one less than the total | |
2257 | * length, so we don't subtract one from | |
2258 | * PAGE_CACHE_SIZE | |
2259 | */ | |
2260 | delalloc_to_write += (delalloc_end - delalloc_start + | |
2261 | PAGE_CACHE_SIZE) >> | |
2262 | PAGE_CACHE_SHIFT; | |
2263 | delalloc_start = delalloc_end + 1; | |
2264 | } | |
2265 | if (wbc->nr_to_write < delalloc_to_write) { | |
2266 | int thresh = 8192; | |
2267 | ||
2268 | if (delalloc_to_write < thresh * 2) | |
2269 | thresh = delalloc_to_write; | |
2270 | wbc->nr_to_write = min_t(u64, delalloc_to_write, | |
2271 | thresh); | |
2272 | } | |
2273 | ||
2274 | /* did the fill delalloc function already unlock and start | |
2275 | * the IO? | |
2276 | */ | |
2277 | if (page_started) { | |
2278 | ret = 0; | |
2279 | /* | |
2280 | * we've unlocked the page, so we can't update | |
2281 | * the mapping's writeback index, just update | |
2282 | * nr_to_write. | |
2283 | */ | |
2284 | wbc->nr_to_write -= nr_written; | |
2285 | goto done_unlocked; | |
2286 | } | |
2287 | } | |
2288 | if (tree->ops && tree->ops->writepage_start_hook) { | |
2289 | ret = tree->ops->writepage_start_hook(page, start, | |
2290 | page_end); | |
2291 | if (ret == -EAGAIN) { | |
2292 | redirty_page_for_writepage(wbc, page); | |
2293 | update_nr_written(page, wbc, nr_written); | |
2294 | unlock_page(page); | |
2295 | ret = 0; | |
2296 | goto done_unlocked; | |
2297 | } | |
2298 | } | |
2299 | ||
2300 | /* | |
2301 | * we don't want to touch the inode after unlocking the page, | |
2302 | * so we update the mapping writeback index now | |
2303 | */ | |
2304 | update_nr_written(page, wbc, nr_written + 1); | |
2305 | ||
2306 | end = page_end; | |
2307 | if (last_byte <= start) { | |
2308 | if (tree->ops && tree->ops->writepage_end_io_hook) | |
2309 | tree->ops->writepage_end_io_hook(page, start, | |
2310 | page_end, NULL, 1); | |
2311 | unlock_start = page_end + 1; | |
2312 | goto done; | |
2313 | } | |
2314 | ||
2315 | blocksize = inode->i_sb->s_blocksize; | |
2316 | ||
2317 | while (cur <= end) { | |
2318 | if (cur >= last_byte) { | |
2319 | if (tree->ops && tree->ops->writepage_end_io_hook) | |
2320 | tree->ops->writepage_end_io_hook(page, cur, | |
2321 | page_end, NULL, 1); | |
2322 | unlock_start = page_end + 1; | |
2323 | break; | |
2324 | } | |
2325 | em = epd->get_extent(inode, page, pg_offset, cur, | |
2326 | end - cur + 1, 1); | |
2327 | if (IS_ERR(em) || !em) { | |
2328 | SetPageError(page); | |
2329 | break; | |
2330 | } | |
2331 | ||
2332 | extent_offset = cur - em->start; | |
2333 | BUG_ON(extent_map_end(em) <= cur); | |
2334 | BUG_ON(end < cur); | |
2335 | iosize = min(extent_map_end(em) - cur, end - cur + 1); | |
2336 | iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1); | |
2337 | sector = (em->block_start + extent_offset) >> 9; | |
2338 | bdev = em->bdev; | |
2339 | block_start = em->block_start; | |
2340 | compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags); | |
2341 | free_extent_map(em); | |
2342 | em = NULL; | |
2343 | ||
2344 | /* | |
2345 | * compressed and inline extents are written through other | |
2346 | * paths in the FS | |
2347 | */ | |
2348 | if (compressed || block_start == EXTENT_MAP_HOLE || | |
2349 | block_start == EXTENT_MAP_INLINE) { | |
2350 | /* | |
2351 | * end_io notification does not happen here for | |
2352 | * compressed extents | |
2353 | */ | |
2354 | if (!compressed && tree->ops && | |
2355 | tree->ops->writepage_end_io_hook) | |
2356 | tree->ops->writepage_end_io_hook(page, cur, | |
2357 | cur + iosize - 1, | |
2358 | NULL, 1); | |
2359 | else if (compressed) { | |
2360 | /* we don't want to end_page_writeback on | |
2361 | * a compressed extent. this happens | |
2362 | * elsewhere | |
2363 | */ | |
2364 | nr++; | |
2365 | } | |
2366 | ||
2367 | cur += iosize; | |
2368 | pg_offset += iosize; | |
2369 | unlock_start = cur; | |
2370 | continue; | |
2371 | } | |
2372 | /* leave this out until we have a page_mkwrite call */ | |
2373 | if (0 && !test_range_bit(tree, cur, cur + iosize - 1, | |
2374 | EXTENT_DIRTY, 0, NULL)) { | |
2375 | cur = cur + iosize; | |
2376 | pg_offset += iosize; | |
2377 | continue; | |
2378 | } | |
2379 | ||
2380 | if (tree->ops && tree->ops->writepage_io_hook) { | |
2381 | ret = tree->ops->writepage_io_hook(page, cur, | |
2382 | cur + iosize - 1); | |
2383 | } else { | |
2384 | ret = 0; | |
2385 | } | |
2386 | if (ret) { | |
2387 | SetPageError(page); | |
2388 | } else { | |
2389 | unsigned long max_nr = end_index + 1; | |
2390 | ||
2391 | set_range_writeback(tree, cur, cur + iosize - 1); | |
2392 | if (!PageWriteback(page)) { | |
2393 | printk(KERN_ERR "btrfs warning page %lu not " | |
2394 | "writeback, cur %llu end %llu\n", | |
2395 | page->index, (unsigned long long)cur, | |
2396 | (unsigned long long)end); | |
2397 | } | |
2398 | ||
2399 | ret = submit_extent_page(write_flags, tree, page, | |
2400 | sector, iosize, pg_offset, | |
2401 | bdev, &epd->bio, max_nr, | |
2402 | end_bio_extent_writepage, | |
2403 | 0, 0, 0); | |
2404 | if (ret) | |
2405 | SetPageError(page); | |
2406 | } | |
2407 | cur = cur + iosize; | |
2408 | pg_offset += iosize; | |
2409 | nr++; | |
2410 | } | |
2411 | done: | |
2412 | if (nr == 0) { | |
2413 | /* make sure the mapping tag for page dirty gets cleared */ | |
2414 | set_page_writeback(page); | |
2415 | end_page_writeback(page); | |
2416 | } | |
2417 | unlock_page(page); | |
2418 | ||
2419 | done_unlocked: | |
2420 | ||
2421 | /* drop our reference on any cached states */ | |
2422 | free_extent_state(cached_state); | |
2423 | return 0; | |
2424 | } | |
2425 | ||
2426 | /** | |
2427 | * write_cache_pages - walk the list of dirty pages of the given address space and write all of them. | |
2428 | * @mapping: address space structure to write | |
2429 | * @wbc: subtract the number of written pages from *@wbc->nr_to_write | |
2430 | * @writepage: function called for each page | |
2431 | * @data: data passed to writepage function | |
2432 | * | |
2433 | * If a page is already under I/O, write_cache_pages() skips it, even | |
2434 | * if it's dirty. This is desirable behaviour for memory-cleaning writeback, | |
2435 | * but it is INCORRECT for data-integrity system calls such as fsync(). fsync() | |
2436 | * and msync() need to guarantee that all the data which was dirty at the time | |
2437 | * the call was made get new I/O started against them. If wbc->sync_mode is | |
2438 | * WB_SYNC_ALL then we were called for data integrity and we must wait for | |
2439 | * existing IO to complete. | |
2440 | */ | |
2441 | static int extent_write_cache_pages(struct extent_io_tree *tree, | |
2442 | struct address_space *mapping, | |
2443 | struct writeback_control *wbc, | |
2444 | writepage_t writepage, void *data, | |
2445 | void (*flush_fn)(void *)) | |
2446 | { | |
2447 | int ret = 0; | |
2448 | int done = 0; | |
2449 | int nr_to_write_done = 0; | |
2450 | struct pagevec pvec; | |
2451 | int nr_pages; | |
2452 | pgoff_t index; | |
2453 | pgoff_t end; /* Inclusive */ | |
2454 | int scanned = 0; | |
2455 | int range_whole = 0; | |
2456 | ||
2457 | pagevec_init(&pvec, 0); | |
2458 | if (wbc->range_cyclic) { | |
2459 | index = mapping->writeback_index; /* Start from prev offset */ | |
2460 | end = -1; | |
2461 | } else { | |
2462 | index = wbc->range_start >> PAGE_CACHE_SHIFT; | |
2463 | end = wbc->range_end >> PAGE_CACHE_SHIFT; | |
2464 | if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX) | |
2465 | range_whole = 1; | |
2466 | scanned = 1; | |
2467 | } | |
2468 | retry: | |
2469 | while (!done && !nr_to_write_done && (index <= end) && | |
2470 | (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, | |
2471 | PAGECACHE_TAG_DIRTY, min(end - index, | |
2472 | (pgoff_t)PAGEVEC_SIZE-1) + 1))) { | |
2473 | unsigned i; | |
2474 | ||
2475 | scanned = 1; | |
2476 | for (i = 0; i < nr_pages; i++) { | |
2477 | struct page *page = pvec.pages[i]; | |
2478 | ||
2479 | /* | |
2480 | * At this point we hold neither mapping->tree_lock nor | |
2481 | * lock on the page itself: the page may be truncated or | |
2482 | * invalidated (changing page->mapping to NULL), or even | |
2483 | * swizzled back from swapper_space to tmpfs file | |
2484 | * mapping | |
2485 | */ | |
2486 | if (tree->ops && tree->ops->write_cache_pages_lock_hook) | |
2487 | tree->ops->write_cache_pages_lock_hook(page); | |
2488 | else | |
2489 | lock_page(page); | |
2490 | ||
2491 | if (unlikely(page->mapping != mapping)) { | |
2492 | unlock_page(page); | |
2493 | continue; | |
2494 | } | |
2495 | ||
2496 | if (!wbc->range_cyclic && page->index > end) { | |
2497 | done = 1; | |
2498 | unlock_page(page); | |
2499 | continue; | |
2500 | } | |
2501 | ||
2502 | if (wbc->sync_mode != WB_SYNC_NONE) { | |
2503 | if (PageWriteback(page)) | |
2504 | flush_fn(data); | |
2505 | wait_on_page_writeback(page); | |
2506 | } | |
2507 | ||
2508 | if (PageWriteback(page) || | |
2509 | !clear_page_dirty_for_io(page)) { | |
2510 | unlock_page(page); | |
2511 | continue; | |
2512 | } | |
2513 | ||
2514 | ret = (*writepage)(page, wbc, data); | |
2515 | ||
2516 | if (unlikely(ret == AOP_WRITEPAGE_ACTIVATE)) { | |
2517 | unlock_page(page); | |
2518 | ret = 0; | |
2519 | } | |
2520 | if (ret) | |
2521 | done = 1; | |
2522 | ||
2523 | /* | |
2524 | * the filesystem may choose to bump up nr_to_write. | |
2525 | * We have to make sure to honor the new nr_to_write | |
2526 | * at any time | |
2527 | */ | |
2528 | nr_to_write_done = wbc->nr_to_write <= 0; | |
2529 | } | |
2530 | pagevec_release(&pvec); | |
2531 | cond_resched(); | |
2532 | } | |
2533 | if (!scanned && !done) { | |
2534 | /* | |
2535 | * We hit the last page and there is more work to be done: wrap | |
2536 | * back to the start of the file | |
2537 | */ | |
2538 | scanned = 1; | |
2539 | index = 0; | |
2540 | goto retry; | |
2541 | } | |
2542 | return ret; | |
2543 | } | |
2544 | ||
2545 | static void flush_epd_write_bio(struct extent_page_data *epd) | |
2546 | { | |
2547 | if (epd->bio) { | |
2548 | if (epd->sync_io) | |
2549 | submit_one_bio(WRITE_SYNC, epd->bio, 0, 0); | |
2550 | else | |
2551 | submit_one_bio(WRITE, epd->bio, 0, 0); | |
2552 | epd->bio = NULL; | |
2553 | } | |
2554 | } | |
2555 | ||
2556 | static noinline void flush_write_bio(void *data) | |
2557 | { | |
2558 | struct extent_page_data *epd = data; | |
2559 | flush_epd_write_bio(epd); | |
2560 | } | |
2561 | ||
2562 | int extent_write_full_page(struct extent_io_tree *tree, struct page *page, | |
2563 | get_extent_t *get_extent, | |
2564 | struct writeback_control *wbc) | |
2565 | { | |
2566 | int ret; | |
2567 | struct address_space *mapping = page->mapping; | |
2568 | struct extent_page_data epd = { | |
2569 | .bio = NULL, | |
2570 | .tree = tree, | |
2571 | .get_extent = get_extent, | |
2572 | .extent_locked = 0, | |
2573 | .sync_io = wbc->sync_mode == WB_SYNC_ALL, | |
2574 | }; | |
2575 | struct writeback_control wbc_writepages = { | |
2576 | .bdi = wbc->bdi, | |
2577 | .sync_mode = wbc->sync_mode, | |
2578 | .older_than_this = NULL, | |
2579 | .nr_to_write = 64, | |
2580 | .range_start = page_offset(page) + PAGE_CACHE_SIZE, | |
2581 | .range_end = (loff_t)-1, | |
2582 | }; | |
2583 | ||
2584 | ret = __extent_writepage(page, wbc, &epd); | |
2585 | ||
2586 | extent_write_cache_pages(tree, mapping, &wbc_writepages, | |
2587 | __extent_writepage, &epd, flush_write_bio); | |
2588 | flush_epd_write_bio(&epd); | |
2589 | return ret; | |
2590 | } | |
2591 | ||
2592 | int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode, | |
2593 | u64 start, u64 end, get_extent_t *get_extent, | |
2594 | int mode) | |
2595 | { | |
2596 | int ret = 0; | |
2597 | struct address_space *mapping = inode->i_mapping; | |
2598 | struct page *page; | |
2599 | unsigned long nr_pages = (end - start + PAGE_CACHE_SIZE) >> | |
2600 | PAGE_CACHE_SHIFT; | |
2601 | ||
2602 | struct extent_page_data epd = { | |
2603 | .bio = NULL, | |
2604 | .tree = tree, | |
2605 | .get_extent = get_extent, | |
2606 | .extent_locked = 1, | |
2607 | .sync_io = mode == WB_SYNC_ALL, | |
2608 | }; | |
2609 | struct writeback_control wbc_writepages = { | |
2610 | .bdi = inode->i_mapping->backing_dev_info, | |
2611 | .sync_mode = mode, | |
2612 | .older_than_this = NULL, | |
2613 | .nr_to_write = nr_pages * 2, | |
2614 | .range_start = start, | |
2615 | .range_end = end + 1, | |
2616 | }; | |
2617 | ||
2618 | while (start <= end) { | |
2619 | page = find_get_page(mapping, start >> PAGE_CACHE_SHIFT); | |
2620 | if (clear_page_dirty_for_io(page)) | |
2621 | ret = __extent_writepage(page, &wbc_writepages, &epd); | |
2622 | else { | |
2623 | if (tree->ops && tree->ops->writepage_end_io_hook) | |
2624 | tree->ops->writepage_end_io_hook(page, start, | |
2625 | start + PAGE_CACHE_SIZE - 1, | |
2626 | NULL, 1); | |
2627 | unlock_page(page); | |
2628 | } | |
2629 | page_cache_release(page); | |
2630 | start += PAGE_CACHE_SIZE; | |
2631 | } | |
2632 | ||
2633 | flush_epd_write_bio(&epd); | |
2634 | return ret; | |
2635 | } | |
2636 | ||
2637 | int extent_writepages(struct extent_io_tree *tree, | |
2638 | struct address_space *mapping, | |
2639 | get_extent_t *get_extent, | |
2640 | struct writeback_control *wbc) | |
2641 | { | |
2642 | int ret = 0; | |
2643 | struct extent_page_data epd = { | |
2644 | .bio = NULL, | |
2645 | .tree = tree, | |
2646 | .get_extent = get_extent, | |
2647 | .extent_locked = 0, | |
2648 | .sync_io = wbc->sync_mode == WB_SYNC_ALL, | |
2649 | }; | |
2650 | ||
2651 | ret = extent_write_cache_pages(tree, mapping, wbc, | |
2652 | __extent_writepage, &epd, | |
2653 | flush_write_bio); | |
2654 | flush_epd_write_bio(&epd); | |
2655 | return ret; | |
2656 | } | |
2657 | ||
2658 | int extent_readpages(struct extent_io_tree *tree, | |
2659 | struct address_space *mapping, | |
2660 | struct list_head *pages, unsigned nr_pages, | |
2661 | get_extent_t get_extent) | |
2662 | { | |
2663 | struct bio *bio = NULL; | |
2664 | unsigned page_idx; | |
2665 | struct pagevec pvec; | |
2666 | unsigned long bio_flags = 0; | |
2667 | ||
2668 | pagevec_init(&pvec, 0); | |
2669 | for (page_idx = 0; page_idx < nr_pages; page_idx++) { | |
2670 | struct page *page = list_entry(pages->prev, struct page, lru); | |
2671 | ||
2672 | prefetchw(&page->flags); | |
2673 | list_del(&page->lru); | |
2674 | /* | |
2675 | * what we want to do here is call add_to_page_cache_lru, | |
2676 | * but that isn't exported, so we reproduce it here | |
2677 | */ | |
2678 | if (!add_to_page_cache(page, mapping, | |
2679 | page->index, GFP_KERNEL)) { | |
2680 | ||
2681 | /* open coding of lru_cache_add, also not exported */ | |
2682 | page_cache_get(page); | |
2683 | if (!pagevec_add(&pvec, page)) | |
2684 | __pagevec_lru_add_file(&pvec); | |
2685 | __extent_read_full_page(tree, page, get_extent, | |
2686 | &bio, 0, &bio_flags); | |
2687 | } | |
2688 | page_cache_release(page); | |
2689 | } | |
2690 | if (pagevec_count(&pvec)) | |
2691 | __pagevec_lru_add_file(&pvec); | |
2692 | BUG_ON(!list_empty(pages)); | |
2693 | if (bio) | |
2694 | submit_one_bio(READ, bio, 0, bio_flags); | |
2695 | return 0; | |
2696 | } | |
2697 | ||
2698 | /* | |
2699 | * basic invalidatepage code, this waits on any locked or writeback | |
2700 | * ranges corresponding to the page, and then deletes any extent state | |
2701 | * records from the tree | |
2702 | */ | |
2703 | int extent_invalidatepage(struct extent_io_tree *tree, | |
2704 | struct page *page, unsigned long offset) | |
2705 | { | |
2706 | u64 start = ((u64)page->index << PAGE_CACHE_SHIFT); | |
2707 | u64 end = start + PAGE_CACHE_SIZE - 1; | |
2708 | size_t blocksize = page->mapping->host->i_sb->s_blocksize; | |
2709 | ||
2710 | start += (offset + blocksize - 1) & ~(blocksize - 1); | |
2711 | if (start > end) | |
2712 | return 0; | |
2713 | ||
2714 | lock_extent(tree, start, end, GFP_NOFS); | |
2715 | wait_on_page_writeback(page); | |
2716 | clear_extent_bit(tree, start, end, | |
2717 | EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC, | |
2718 | 1, 1, NULL, GFP_NOFS); | |
2719 | return 0; | |
2720 | } | |
2721 | ||
2722 | /* | |
2723 | * simple commit_write call, set_range_dirty is used to mark both | |
2724 | * the pages and the extent records as dirty | |
2725 | */ | |
2726 | int extent_commit_write(struct extent_io_tree *tree, | |
2727 | struct inode *inode, struct page *page, | |
2728 | unsigned from, unsigned to) | |
2729 | { | |
2730 | loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to; | |
2731 | ||
2732 | set_page_extent_mapped(page); | |
2733 | set_page_dirty(page); | |
2734 | ||
2735 | if (pos > inode->i_size) { | |
2736 | i_size_write(inode, pos); | |
2737 | mark_inode_dirty(inode); | |
2738 | } | |
2739 | return 0; | |
2740 | } | |
2741 | ||
2742 | int extent_prepare_write(struct extent_io_tree *tree, | |
2743 | struct inode *inode, struct page *page, | |
2744 | unsigned from, unsigned to, get_extent_t *get_extent) | |
2745 | { | |
2746 | u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT; | |
2747 | u64 page_end = page_start + PAGE_CACHE_SIZE - 1; | |
2748 | u64 block_start; | |
2749 | u64 orig_block_start; | |
2750 | u64 block_end; | |
2751 | u64 cur_end; | |
2752 | struct extent_map *em; | |
2753 | unsigned blocksize = 1 << inode->i_blkbits; | |
2754 | size_t page_offset = 0; | |
2755 | size_t block_off_start; | |
2756 | size_t block_off_end; | |
2757 | int err = 0; | |
2758 | int iocount = 0; | |
2759 | int ret = 0; | |
2760 | int isnew; | |
2761 | ||
2762 | set_page_extent_mapped(page); | |
2763 | ||
2764 | block_start = (page_start + from) & ~((u64)blocksize - 1); | |
2765 | block_end = (page_start + to - 1) | (blocksize - 1); | |
2766 | orig_block_start = block_start; | |
2767 | ||
2768 | lock_extent(tree, page_start, page_end, GFP_NOFS); | |
2769 | while (block_start <= block_end) { | |
2770 | em = get_extent(inode, page, page_offset, block_start, | |
2771 | block_end - block_start + 1, 1); | |
2772 | if (IS_ERR(em) || !em) | |
2773 | goto err; | |
2774 | ||
2775 | cur_end = min(block_end, extent_map_end(em) - 1); | |
2776 | block_off_start = block_start & (PAGE_CACHE_SIZE - 1); | |
2777 | block_off_end = block_off_start + blocksize; | |
2778 | isnew = clear_extent_new(tree, block_start, cur_end, GFP_NOFS); | |
2779 | ||
2780 | if (!PageUptodate(page) && isnew && | |
2781 | (block_off_end > to || block_off_start < from)) { | |
2782 | void *kaddr; | |
2783 | ||
2784 | kaddr = kmap_atomic(page, KM_USER0); | |
2785 | if (block_off_end > to) | |
2786 | memset(kaddr + to, 0, block_off_end - to); | |
2787 | if (block_off_start < from) | |
2788 | memset(kaddr + block_off_start, 0, | |
2789 | from - block_off_start); | |
2790 | flush_dcache_page(page); | |
2791 | kunmap_atomic(kaddr, KM_USER0); | |
2792 | } | |
2793 | if ((em->block_start != EXTENT_MAP_HOLE && | |
2794 | em->block_start != EXTENT_MAP_INLINE) && | |
2795 | !isnew && !PageUptodate(page) && | |
2796 | (block_off_end > to || block_off_start < from) && | |
2797 | !test_range_bit(tree, block_start, cur_end, | |
2798 | EXTENT_UPTODATE, 1, NULL)) { | |
2799 | u64 sector; | |
2800 | u64 extent_offset = block_start - em->start; | |
2801 | size_t iosize; | |
2802 | sector = (em->block_start + extent_offset) >> 9; | |
2803 | iosize = (cur_end - block_start + blocksize) & | |
2804 | ~((u64)blocksize - 1); | |
2805 | /* | |
2806 | * we've already got the extent locked, but we | |
2807 | * need to split the state such that our end_bio | |
2808 | * handler can clear the lock. | |
2809 | */ | |
2810 | set_extent_bit(tree, block_start, | |
2811 | block_start + iosize - 1, | |
2812 | EXTENT_LOCKED, 0, NULL, NULL, GFP_NOFS); | |
2813 | ret = submit_extent_page(READ, tree, page, | |
2814 | sector, iosize, page_offset, em->bdev, | |
2815 | NULL, 1, | |
2816 | end_bio_extent_preparewrite, 0, | |
2817 | 0, 0); | |
2818 | iocount++; | |
2819 | block_start = block_start + iosize; | |
2820 | } else { | |
2821 | set_extent_uptodate(tree, block_start, cur_end, | |
2822 | GFP_NOFS); | |
2823 | unlock_extent(tree, block_start, cur_end, GFP_NOFS); | |
2824 | block_start = cur_end + 1; | |
2825 | } | |
2826 | page_offset = block_start & (PAGE_CACHE_SIZE - 1); | |
2827 | free_extent_map(em); | |
2828 | } | |
2829 | if (iocount) { | |
2830 | wait_extent_bit(tree, orig_block_start, | |
2831 | block_end, EXTENT_LOCKED); | |
2832 | } | |
2833 | check_page_uptodate(tree, page); | |
2834 | err: | |
2835 | /* FIXME, zero out newly allocated blocks on error */ | |
2836 | return err; | |
2837 | } | |
2838 | ||
2839 | /* | |
2840 | * a helper for releasepage, this tests for areas of the page that | |
2841 | * are locked or under IO and drops the related state bits if it is safe | |
2842 | * to drop the page. | |
2843 | */ | |
2844 | int try_release_extent_state(struct extent_map_tree *map, | |
2845 | struct extent_io_tree *tree, struct page *page, | |
2846 | gfp_t mask) | |
2847 | { | |
2848 | u64 start = (u64)page->index << PAGE_CACHE_SHIFT; | |
2849 | u64 end = start + PAGE_CACHE_SIZE - 1; | |
2850 | int ret = 1; | |
2851 | ||
2852 | if (test_range_bit(tree, start, end, | |
2853 | EXTENT_IOBITS, 0, NULL)) | |
2854 | ret = 0; | |
2855 | else { | |
2856 | if ((mask & GFP_NOFS) == GFP_NOFS) | |
2857 | mask = GFP_NOFS; | |
2858 | /* | |
2859 | * at this point we can safely clear everything except the | |
2860 | * locked bit and the nodatasum bit | |
2861 | */ | |
2862 | clear_extent_bit(tree, start, end, | |
2863 | ~(EXTENT_LOCKED | EXTENT_NODATASUM), | |
2864 | 0, 0, NULL, mask); | |
2865 | } | |
2866 | return ret; | |
2867 | } | |
2868 | ||
2869 | /* | |
2870 | * a helper for releasepage. As long as there are no locked extents | |
2871 | * in the range corresponding to the page, both state records and extent | |
2872 | * map records are removed | |
2873 | */ | |
2874 | int try_release_extent_mapping(struct extent_map_tree *map, | |
2875 | struct extent_io_tree *tree, struct page *page, | |
2876 | gfp_t mask) | |
2877 | { | |
2878 | struct extent_map *em; | |
2879 | u64 start = (u64)page->index << PAGE_CACHE_SHIFT; | |
2880 | u64 end = start + PAGE_CACHE_SIZE - 1; | |
2881 | ||
2882 | if ((mask & __GFP_WAIT) && | |
2883 | page->mapping->host->i_size > 16 * 1024 * 1024) { | |
2884 | u64 len; | |
2885 | while (start <= end) { | |
2886 | len = end - start + 1; | |
2887 | write_lock(&map->lock); | |
2888 | em = lookup_extent_mapping(map, start, len); | |
2889 | if (!em || IS_ERR(em)) { | |
2890 | write_unlock(&map->lock); | |
2891 | break; | |
2892 | } | |
2893 | if (test_bit(EXTENT_FLAG_PINNED, &em->flags) || | |
2894 | em->start != start) { | |
2895 | write_unlock(&map->lock); | |
2896 | free_extent_map(em); | |
2897 | break; | |
2898 | } | |
2899 | if (!test_range_bit(tree, em->start, | |
2900 | extent_map_end(em) - 1, | |
2901 | EXTENT_LOCKED | EXTENT_WRITEBACK, | |
2902 | 0, NULL)) { | |
2903 | remove_extent_mapping(map, em); | |
2904 | /* once for the rb tree */ | |
2905 | free_extent_map(em); | |
2906 | } | |
2907 | start = extent_map_end(em); | |
2908 | write_unlock(&map->lock); | |
2909 | ||
2910 | /* once for us */ | |
2911 | free_extent_map(em); | |
2912 | } | |
2913 | } | |
2914 | return try_release_extent_state(map, tree, page, mask); | |
2915 | } | |
2916 | ||
2917 | sector_t extent_bmap(struct address_space *mapping, sector_t iblock, | |
2918 | get_extent_t *get_extent) | |
2919 | { | |
2920 | struct inode *inode = mapping->host; | |
2921 | u64 start = iblock << inode->i_blkbits; | |
2922 | sector_t sector = 0; | |
2923 | size_t blksize = (1 << inode->i_blkbits); | |
2924 | struct extent_map *em; | |
2925 | ||
2926 | lock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, | |
2927 | GFP_NOFS); | |
2928 | em = get_extent(inode, NULL, 0, start, blksize, 0); | |
2929 | unlock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, | |
2930 | GFP_NOFS); | |
2931 | if (!em || IS_ERR(em)) | |
2932 | return 0; | |
2933 | ||
2934 | if (em->block_start > EXTENT_MAP_LAST_BYTE) | |
2935 | goto out; | |
2936 | ||
2937 | sector = (em->block_start + start - em->start) >> inode->i_blkbits; | |
2938 | out: | |
2939 | free_extent_map(em); | |
2940 | return sector; | |
2941 | } | |
2942 | ||
2943 | int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, | |
2944 | __u64 start, __u64 len, get_extent_t *get_extent) | |
2945 | { | |
2946 | int ret; | |
2947 | u64 off = start; | |
2948 | u64 max = start + len; | |
2949 | u32 flags = 0; | |
2950 | u64 disko = 0; | |
2951 | struct extent_map *em = NULL; | |
2952 | int end = 0; | |
2953 | u64 em_start = 0, em_len = 0; | |
2954 | unsigned long emflags; | |
2955 | ret = 0; | |
2956 | ||
2957 | if (len == 0) | |
2958 | return -EINVAL; | |
2959 | ||
2960 | lock_extent(&BTRFS_I(inode)->io_tree, start, start + len, | |
2961 | GFP_NOFS); | |
2962 | em = get_extent(inode, NULL, 0, off, max - off, 0); | |
2963 | if (!em) | |
2964 | goto out; | |
2965 | if (IS_ERR(em)) { | |
2966 | ret = PTR_ERR(em); | |
2967 | goto out; | |
2968 | } | |
2969 | while (!end) { | |
2970 | off = em->start + em->len; | |
2971 | if (off >= max) | |
2972 | end = 1; | |
2973 | ||
2974 | em_start = em->start; | |
2975 | em_len = em->len; | |
2976 | ||
2977 | disko = 0; | |
2978 | flags = 0; | |
2979 | ||
2980 | if (em->block_start == EXTENT_MAP_LAST_BYTE) { | |
2981 | end = 1; | |
2982 | flags |= FIEMAP_EXTENT_LAST; | |
2983 | } else if (em->block_start == EXTENT_MAP_HOLE) { | |
2984 | flags |= FIEMAP_EXTENT_UNWRITTEN; | |
2985 | } else if (em->block_start == EXTENT_MAP_INLINE) { | |
2986 | flags |= (FIEMAP_EXTENT_DATA_INLINE | | |
2987 | FIEMAP_EXTENT_NOT_ALIGNED); | |
2988 | } else if (em->block_start == EXTENT_MAP_DELALLOC) { | |
2989 | flags |= (FIEMAP_EXTENT_DELALLOC | | |
2990 | FIEMAP_EXTENT_UNKNOWN); | |
2991 | } else { | |
2992 | disko = em->block_start; | |
2993 | } | |
2994 | if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) | |
2995 | flags |= FIEMAP_EXTENT_ENCODED; | |
2996 | ||
2997 | emflags = em->flags; | |
2998 | free_extent_map(em); | |
2999 | em = NULL; | |
3000 | ||
3001 | if (!end) { | |
3002 | em = get_extent(inode, NULL, 0, off, max - off, 0); | |
3003 | if (!em) | |
3004 | goto out; | |
3005 | if (IS_ERR(em)) { | |
3006 | ret = PTR_ERR(em); | |
3007 | goto out; | |
3008 | } | |
3009 | emflags = em->flags; | |
3010 | } | |
3011 | if (test_bit(EXTENT_FLAG_VACANCY, &emflags)) { | |
3012 | flags |= FIEMAP_EXTENT_LAST; | |
3013 | end = 1; | |
3014 | } | |
3015 | ||
3016 | ret = fiemap_fill_next_extent(fieinfo, em_start, disko, | |
3017 | em_len, flags); | |
3018 | if (ret) | |
3019 | goto out_free; | |
3020 | } | |
3021 | out_free: | |
3022 | free_extent_map(em); | |
3023 | out: | |
3024 | unlock_extent(&BTRFS_I(inode)->io_tree, start, start + len, | |
3025 | GFP_NOFS); | |
3026 | return ret; | |
3027 | } | |
3028 | ||
3029 | static inline struct page *extent_buffer_page(struct extent_buffer *eb, | |
3030 | unsigned long i) | |
3031 | { | |
3032 | struct page *p; | |
3033 | struct address_space *mapping; | |
3034 | ||
3035 | if (i == 0) | |
3036 | return eb->first_page; | |
3037 | i += eb->start >> PAGE_CACHE_SHIFT; | |
3038 | mapping = eb->first_page->mapping; | |
3039 | if (!mapping) | |
3040 | return NULL; | |
3041 | ||
3042 | /* | |
3043 | * extent_buffer_page is only called after pinning the page | |
3044 | * by increasing the reference count. So we know the page must | |
3045 | * be in the radix tree. | |
3046 | */ | |
3047 | rcu_read_lock(); | |
3048 | p = radix_tree_lookup(&mapping->page_tree, i); | |
3049 | rcu_read_unlock(); | |
3050 | ||
3051 | return p; | |
3052 | } | |
3053 | ||
3054 | static inline unsigned long num_extent_pages(u64 start, u64 len) | |
3055 | { | |
3056 | return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) - | |
3057 | (start >> PAGE_CACHE_SHIFT); | |
3058 | } | |
3059 | ||
3060 | static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree, | |
3061 | u64 start, | |
3062 | unsigned long len, | |
3063 | gfp_t mask) | |
3064 | { | |
3065 | struct extent_buffer *eb = NULL; | |
3066 | #if LEAK_DEBUG | |
3067 | unsigned long flags; | |
3068 | #endif | |
3069 | ||
3070 | eb = kmem_cache_zalloc(extent_buffer_cache, mask); | |
3071 | eb->start = start; | |
3072 | eb->len = len; | |
3073 | spin_lock_init(&eb->lock); | |
3074 | init_waitqueue_head(&eb->lock_wq); | |
3075 | ||
3076 | #if LEAK_DEBUG | |
3077 | spin_lock_irqsave(&leak_lock, flags); | |
3078 | list_add(&eb->leak_list, &buffers); | |
3079 | spin_unlock_irqrestore(&leak_lock, flags); | |
3080 | #endif | |
3081 | atomic_set(&eb->refs, 1); | |
3082 | ||
3083 | return eb; | |
3084 | } | |
3085 | ||
3086 | static void __free_extent_buffer(struct extent_buffer *eb) | |
3087 | { | |
3088 | #if LEAK_DEBUG | |
3089 | unsigned long flags; | |
3090 | spin_lock_irqsave(&leak_lock, flags); | |
3091 | list_del(&eb->leak_list); | |
3092 | spin_unlock_irqrestore(&leak_lock, flags); | |
3093 | #endif | |
3094 | kmem_cache_free(extent_buffer_cache, eb); | |
3095 | } | |
3096 | ||
3097 | struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, | |
3098 | u64 start, unsigned long len, | |
3099 | struct page *page0, | |
3100 | gfp_t mask) | |
3101 | { | |
3102 | unsigned long num_pages = num_extent_pages(start, len); | |
3103 | unsigned long i; | |
3104 | unsigned long index = start >> PAGE_CACHE_SHIFT; | |
3105 | struct extent_buffer *eb; | |
3106 | struct extent_buffer *exists = NULL; | |
3107 | struct page *p; | |
3108 | struct address_space *mapping = tree->mapping; | |
3109 | int uptodate = 1; | |
3110 | ||
3111 | spin_lock(&tree->buffer_lock); | |
3112 | eb = buffer_search(tree, start); | |
3113 | if (eb) { | |
3114 | atomic_inc(&eb->refs); | |
3115 | spin_unlock(&tree->buffer_lock); | |
3116 | mark_page_accessed(eb->first_page); | |
3117 | return eb; | |
3118 | } | |
3119 | spin_unlock(&tree->buffer_lock); | |
3120 | ||
3121 | eb = __alloc_extent_buffer(tree, start, len, mask); | |
3122 | if (!eb) | |
3123 | return NULL; | |
3124 | ||
3125 | if (page0) { | |
3126 | eb->first_page = page0; | |
3127 | i = 1; | |
3128 | index++; | |
3129 | page_cache_get(page0); | |
3130 | mark_page_accessed(page0); | |
3131 | set_page_extent_mapped(page0); | |
3132 | set_page_extent_head(page0, len); | |
3133 | uptodate = PageUptodate(page0); | |
3134 | } else { | |
3135 | i = 0; | |
3136 | } | |
3137 | for (; i < num_pages; i++, index++) { | |
3138 | p = find_or_create_page(mapping, index, mask | __GFP_HIGHMEM); | |
3139 | if (!p) { | |
3140 | WARN_ON(1); | |
3141 | goto free_eb; | |
3142 | } | |
3143 | set_page_extent_mapped(p); | |
3144 | mark_page_accessed(p); | |
3145 | if (i == 0) { | |
3146 | eb->first_page = p; | |
3147 | set_page_extent_head(p, len); | |
3148 | } else { | |
3149 | set_page_private(p, EXTENT_PAGE_PRIVATE); | |
3150 | } | |
3151 | if (!PageUptodate(p)) | |
3152 | uptodate = 0; | |
3153 | unlock_page(p); | |
3154 | } | |
3155 | if (uptodate) | |
3156 | set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags); | |
3157 | ||
3158 | spin_lock(&tree->buffer_lock); | |
3159 | exists = buffer_tree_insert(tree, start, &eb->rb_node); | |
3160 | if (exists) { | |
3161 | /* add one reference for the caller */ | |
3162 | atomic_inc(&exists->refs); | |
3163 | spin_unlock(&tree->buffer_lock); | |
3164 | goto free_eb; | |
3165 | } | |
3166 | spin_unlock(&tree->buffer_lock); | |
3167 | ||
3168 | /* add one reference for the tree */ | |
3169 | atomic_inc(&eb->refs); | |
3170 | return eb; | |
3171 | ||
3172 | free_eb: | |
3173 | if (!atomic_dec_and_test(&eb->refs)) | |
3174 | return exists; | |
3175 | for (index = 1; index < i; index++) | |
3176 | page_cache_release(extent_buffer_page(eb, index)); | |
3177 | page_cache_release(extent_buffer_page(eb, 0)); | |
3178 | __free_extent_buffer(eb); | |
3179 | return exists; | |
3180 | } | |
3181 | ||
3182 | struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree, | |
3183 | u64 start, unsigned long len, | |
3184 | gfp_t mask) | |
3185 | { | |
3186 | struct extent_buffer *eb; | |
3187 | ||
3188 | spin_lock(&tree->buffer_lock); | |
3189 | eb = buffer_search(tree, start); | |
3190 | if (eb) | |
3191 | atomic_inc(&eb->refs); | |
3192 | spin_unlock(&tree->buffer_lock); | |
3193 | ||
3194 | if (eb) | |
3195 | mark_page_accessed(eb->first_page); | |
3196 | ||
3197 | return eb; | |
3198 | } | |
3199 | ||
3200 | void free_extent_buffer(struct extent_buffer *eb) | |
3201 | { | |
3202 | if (!eb) | |
3203 | return; | |
3204 | ||
3205 | if (!atomic_dec_and_test(&eb->refs)) | |
3206 | return; | |
3207 | ||
3208 | WARN_ON(1); | |
3209 | } | |
3210 | ||
3211 | int clear_extent_buffer_dirty(struct extent_io_tree *tree, | |
3212 | struct extent_buffer *eb) | |
3213 | { | |
3214 | unsigned long i; | |
3215 | unsigned long num_pages; | |
3216 | struct page *page; | |
3217 | ||
3218 | num_pages = num_extent_pages(eb->start, eb->len); | |
3219 | ||
3220 | for (i = 0; i < num_pages; i++) { | |
3221 | page = extent_buffer_page(eb, i); | |
3222 | if (!PageDirty(page)) | |
3223 | continue; | |
3224 | ||
3225 | lock_page(page); | |
3226 | if (i == 0) | |
3227 | set_page_extent_head(page, eb->len); | |
3228 | else | |
3229 | set_page_private(page, EXTENT_PAGE_PRIVATE); | |
3230 | ||
3231 | clear_page_dirty_for_io(page); | |
3232 | spin_lock_irq(&page->mapping->tree_lock); | |
3233 | if (!PageDirty(page)) { | |
3234 | radix_tree_tag_clear(&page->mapping->page_tree, | |
3235 | page_index(page), | |
3236 | PAGECACHE_TAG_DIRTY); | |
3237 | } | |
3238 | spin_unlock_irq(&page->mapping->tree_lock); | |
3239 | unlock_page(page); | |
3240 | } | |
3241 | return 0; | |
3242 | } | |
3243 | ||
3244 | int wait_on_extent_buffer_writeback(struct extent_io_tree *tree, | |
3245 | struct extent_buffer *eb) | |
3246 | { | |
3247 | return wait_on_extent_writeback(tree, eb->start, | |
3248 | eb->start + eb->len - 1); | |
3249 | } | |
3250 | ||
3251 | int set_extent_buffer_dirty(struct extent_io_tree *tree, | |
3252 | struct extent_buffer *eb) | |
3253 | { | |
3254 | unsigned long i; | |
3255 | unsigned long num_pages; | |
3256 | int was_dirty = 0; | |
3257 | ||
3258 | was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags); | |
3259 | num_pages = num_extent_pages(eb->start, eb->len); | |
3260 | for (i = 0; i < num_pages; i++) | |
3261 | __set_page_dirty_nobuffers(extent_buffer_page(eb, i)); | |
3262 | return was_dirty; | |
3263 | } | |
3264 | ||
3265 | int clear_extent_buffer_uptodate(struct extent_io_tree *tree, | |
3266 | struct extent_buffer *eb) | |
3267 | { | |
3268 | unsigned long i; | |
3269 | struct page *page; | |
3270 | unsigned long num_pages; | |
3271 | ||
3272 | num_pages = num_extent_pages(eb->start, eb->len); | |
3273 | clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags); | |
3274 | ||
3275 | clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1, | |
3276 | GFP_NOFS); | |
3277 | for (i = 0; i < num_pages; i++) { | |
3278 | page = extent_buffer_page(eb, i); | |
3279 | if (page) | |
3280 | ClearPageUptodate(page); | |
3281 | } | |
3282 | return 0; | |
3283 | } | |
3284 | ||
3285 | int set_extent_buffer_uptodate(struct extent_io_tree *tree, | |
3286 | struct extent_buffer *eb) | |
3287 | { | |
3288 | unsigned long i; | |
3289 | struct page *page; | |
3290 | unsigned long num_pages; | |
3291 | ||
3292 | num_pages = num_extent_pages(eb->start, eb->len); | |
3293 | ||
3294 | set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1, | |
3295 | GFP_NOFS); | |
3296 | for (i = 0; i < num_pages; i++) { | |
3297 | page = extent_buffer_page(eb, i); | |
3298 | if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) || | |
3299 | ((i == num_pages - 1) && | |
3300 | ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) { | |
3301 | check_page_uptodate(tree, page); | |
3302 | continue; | |
3303 | } | |
3304 | SetPageUptodate(page); | |
3305 | } | |
3306 | return 0; | |
3307 | } | |
3308 | ||
3309 | int extent_range_uptodate(struct extent_io_tree *tree, | |
3310 | u64 start, u64 end) | |
3311 | { | |
3312 | struct page *page; | |
3313 | int ret; | |
3314 | int pg_uptodate = 1; | |
3315 | int uptodate; | |
3316 | unsigned long index; | |
3317 | ||
3318 | ret = test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL); | |
3319 | if (ret) | |
3320 | return 1; | |
3321 | while (start <= end) { | |
3322 | index = start >> PAGE_CACHE_SHIFT; | |
3323 | page = find_get_page(tree->mapping, index); | |
3324 | uptodate = PageUptodate(page); | |
3325 | page_cache_release(page); | |
3326 | if (!uptodate) { | |
3327 | pg_uptodate = 0; | |
3328 | break; | |
3329 | } | |
3330 | start += PAGE_CACHE_SIZE; | |
3331 | } | |
3332 | return pg_uptodate; | |
3333 | } | |
3334 | ||
3335 | int extent_buffer_uptodate(struct extent_io_tree *tree, | |
3336 | struct extent_buffer *eb) | |
3337 | { | |
3338 | int ret = 0; | |
3339 | unsigned long num_pages; | |
3340 | unsigned long i; | |
3341 | struct page *page; | |
3342 | int pg_uptodate = 1; | |
3343 | ||
3344 | if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags)) | |
3345 | return 1; | |
3346 | ||
3347 | ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1, | |
3348 | EXTENT_UPTODATE, 1, NULL); | |
3349 | if (ret) | |
3350 | return ret; | |
3351 | ||
3352 | num_pages = num_extent_pages(eb->start, eb->len); | |
3353 | for (i = 0; i < num_pages; i++) { | |
3354 | page = extent_buffer_page(eb, i); | |
3355 | if (!PageUptodate(page)) { | |
3356 | pg_uptodate = 0; | |
3357 | break; | |
3358 | } | |
3359 | } | |
3360 | return pg_uptodate; | |
3361 | } | |
3362 | ||
3363 | int read_extent_buffer_pages(struct extent_io_tree *tree, | |
3364 | struct extent_buffer *eb, | |
3365 | u64 start, int wait, | |
3366 | get_extent_t *get_extent, int mirror_num) | |
3367 | { | |
3368 | unsigned long i; | |
3369 | unsigned long start_i; | |
3370 | struct page *page; | |
3371 | int err; | |
3372 | int ret = 0; | |
3373 | int locked_pages = 0; | |
3374 | int all_uptodate = 1; | |
3375 | int inc_all_pages = 0; | |
3376 | unsigned long num_pages; | |
3377 | struct bio *bio = NULL; | |
3378 | unsigned long bio_flags = 0; | |
3379 | ||
3380 | if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags)) | |
3381 | return 0; | |
3382 | ||
3383 | if (test_range_bit(tree, eb->start, eb->start + eb->len - 1, | |
3384 | EXTENT_UPTODATE, 1, NULL)) { | |
3385 | return 0; | |
3386 | } | |
3387 | ||
3388 | if (start) { | |
3389 | WARN_ON(start < eb->start); | |
3390 | start_i = (start >> PAGE_CACHE_SHIFT) - | |
3391 | (eb->start >> PAGE_CACHE_SHIFT); | |
3392 | } else { | |
3393 | start_i = 0; | |
3394 | } | |
3395 | ||
3396 | num_pages = num_extent_pages(eb->start, eb->len); | |
3397 | for (i = start_i; i < num_pages; i++) { | |
3398 | page = extent_buffer_page(eb, i); | |
3399 | if (!wait) { | |
3400 | if (!trylock_page(page)) | |
3401 | goto unlock_exit; | |
3402 | } else { | |
3403 | lock_page(page); | |
3404 | } | |
3405 | locked_pages++; | |
3406 | if (!PageUptodate(page)) | |
3407 | all_uptodate = 0; | |
3408 | } | |
3409 | if (all_uptodate) { | |
3410 | if (start_i == 0) | |
3411 | set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags); | |
3412 | goto unlock_exit; | |
3413 | } | |
3414 | ||
3415 | for (i = start_i; i < num_pages; i++) { | |
3416 | page = extent_buffer_page(eb, i); | |
3417 | if (inc_all_pages) | |
3418 | page_cache_get(page); | |
3419 | if (!PageUptodate(page)) { | |
3420 | if (start_i == 0) | |
3421 | inc_all_pages = 1; | |
3422 | ClearPageError(page); | |
3423 | err = __extent_read_full_page(tree, page, | |
3424 | get_extent, &bio, | |
3425 | mirror_num, &bio_flags); | |
3426 | if (err) | |
3427 | ret = err; | |
3428 | } else { | |
3429 | unlock_page(page); | |
3430 | } | |
3431 | } | |
3432 | ||
3433 | if (bio) | |
3434 | submit_one_bio(READ, bio, mirror_num, bio_flags); | |
3435 | ||
3436 | if (ret || !wait) | |
3437 | return ret; | |
3438 | ||
3439 | for (i = start_i; i < num_pages; i++) { | |
3440 | page = extent_buffer_page(eb, i); | |
3441 | wait_on_page_locked(page); | |
3442 | if (!PageUptodate(page)) | |
3443 | ret = -EIO; | |
3444 | } | |
3445 | ||
3446 | if (!ret) | |
3447 | set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags); | |
3448 | return ret; | |
3449 | ||
3450 | unlock_exit: | |
3451 | i = start_i; | |
3452 | while (locked_pages > 0) { | |
3453 | page = extent_buffer_page(eb, i); | |
3454 | i++; | |
3455 | unlock_page(page); | |
3456 | locked_pages--; | |
3457 | } | |
3458 | return ret; | |
3459 | } | |
3460 | ||
3461 | void read_extent_buffer(struct extent_buffer *eb, void *dstv, | |
3462 | unsigned long start, | |
3463 | unsigned long len) | |
3464 | { | |
3465 | size_t cur; | |
3466 | size_t offset; | |
3467 | struct page *page; | |
3468 | char *kaddr; | |
3469 | char *dst = (char *)dstv; | |
3470 | size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3471 | unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; | |
3472 | ||
3473 | WARN_ON(start > eb->len); | |
3474 | WARN_ON(start + len > eb->start + eb->len); | |
3475 | ||
3476 | offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3477 | ||
3478 | while (len > 0) { | |
3479 | page = extent_buffer_page(eb, i); | |
3480 | ||
3481 | cur = min(len, (PAGE_CACHE_SIZE - offset)); | |
3482 | kaddr = kmap_atomic(page, KM_USER1); | |
3483 | memcpy(dst, kaddr + offset, cur); | |
3484 | kunmap_atomic(kaddr, KM_USER1); | |
3485 | ||
3486 | dst += cur; | |
3487 | len -= cur; | |
3488 | offset = 0; | |
3489 | i++; | |
3490 | } | |
3491 | } | |
3492 | ||
3493 | int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start, | |
3494 | unsigned long min_len, char **token, char **map, | |
3495 | unsigned long *map_start, | |
3496 | unsigned long *map_len, int km) | |
3497 | { | |
3498 | size_t offset = start & (PAGE_CACHE_SIZE - 1); | |
3499 | char *kaddr; | |
3500 | struct page *p; | |
3501 | size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3502 | unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; | |
3503 | unsigned long end_i = (start_offset + start + min_len - 1) >> | |
3504 | PAGE_CACHE_SHIFT; | |
3505 | ||
3506 | if (i != end_i) | |
3507 | return -EINVAL; | |
3508 | ||
3509 | if (i == 0) { | |
3510 | offset = start_offset; | |
3511 | *map_start = 0; | |
3512 | } else { | |
3513 | offset = 0; | |
3514 | *map_start = ((u64)i << PAGE_CACHE_SHIFT) - start_offset; | |
3515 | } | |
3516 | ||
3517 | if (start + min_len > eb->len) { | |
3518 | printk(KERN_ERR "btrfs bad mapping eb start %llu len %lu, " | |
3519 | "wanted %lu %lu\n", (unsigned long long)eb->start, | |
3520 | eb->len, start, min_len); | |
3521 | WARN_ON(1); | |
3522 | } | |
3523 | ||
3524 | p = extent_buffer_page(eb, i); | |
3525 | kaddr = kmap_atomic(p, km); | |
3526 | *token = kaddr; | |
3527 | *map = kaddr + offset; | |
3528 | *map_len = PAGE_CACHE_SIZE - offset; | |
3529 | return 0; | |
3530 | } | |
3531 | ||
3532 | int map_extent_buffer(struct extent_buffer *eb, unsigned long start, | |
3533 | unsigned long min_len, | |
3534 | char **token, char **map, | |
3535 | unsigned long *map_start, | |
3536 | unsigned long *map_len, int km) | |
3537 | { | |
3538 | int err; | |
3539 | int save = 0; | |
3540 | if (eb->map_token) { | |
3541 | unmap_extent_buffer(eb, eb->map_token, km); | |
3542 | eb->map_token = NULL; | |
3543 | save = 1; | |
3544 | } | |
3545 | err = map_private_extent_buffer(eb, start, min_len, token, map, | |
3546 | map_start, map_len, km); | |
3547 | if (!err && save) { | |
3548 | eb->map_token = *token; | |
3549 | eb->kaddr = *map; | |
3550 | eb->map_start = *map_start; | |
3551 | eb->map_len = *map_len; | |
3552 | } | |
3553 | return err; | |
3554 | } | |
3555 | ||
3556 | void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km) | |
3557 | { | |
3558 | kunmap_atomic(token, km); | |
3559 | } | |
3560 | ||
3561 | int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv, | |
3562 | unsigned long start, | |
3563 | unsigned long len) | |
3564 | { | |
3565 | size_t cur; | |
3566 | size_t offset; | |
3567 | struct page *page; | |
3568 | char *kaddr; | |
3569 | char *ptr = (char *)ptrv; | |
3570 | size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3571 | unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; | |
3572 | int ret = 0; | |
3573 | ||
3574 | WARN_ON(start > eb->len); | |
3575 | WARN_ON(start + len > eb->start + eb->len); | |
3576 | ||
3577 | offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3578 | ||
3579 | while (len > 0) { | |
3580 | page = extent_buffer_page(eb, i); | |
3581 | ||
3582 | cur = min(len, (PAGE_CACHE_SIZE - offset)); | |
3583 | ||
3584 | kaddr = kmap_atomic(page, KM_USER0); | |
3585 | ret = memcmp(ptr, kaddr + offset, cur); | |
3586 | kunmap_atomic(kaddr, KM_USER0); | |
3587 | if (ret) | |
3588 | break; | |
3589 | ||
3590 | ptr += cur; | |
3591 | len -= cur; | |
3592 | offset = 0; | |
3593 | i++; | |
3594 | } | |
3595 | return ret; | |
3596 | } | |
3597 | ||
3598 | void write_extent_buffer(struct extent_buffer *eb, const void *srcv, | |
3599 | unsigned long start, unsigned long len) | |
3600 | { | |
3601 | size_t cur; | |
3602 | size_t offset; | |
3603 | struct page *page; | |
3604 | char *kaddr; | |
3605 | char *src = (char *)srcv; | |
3606 | size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3607 | unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; | |
3608 | ||
3609 | WARN_ON(start > eb->len); | |
3610 | WARN_ON(start + len > eb->start + eb->len); | |
3611 | ||
3612 | offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3613 | ||
3614 | while (len > 0) { | |
3615 | page = extent_buffer_page(eb, i); | |
3616 | WARN_ON(!PageUptodate(page)); | |
3617 | ||
3618 | cur = min(len, PAGE_CACHE_SIZE - offset); | |
3619 | kaddr = kmap_atomic(page, KM_USER1); | |
3620 | memcpy(kaddr + offset, src, cur); | |
3621 | kunmap_atomic(kaddr, KM_USER1); | |
3622 | ||
3623 | src += cur; | |
3624 | len -= cur; | |
3625 | offset = 0; | |
3626 | i++; | |
3627 | } | |
3628 | } | |
3629 | ||
3630 | void memset_extent_buffer(struct extent_buffer *eb, char c, | |
3631 | unsigned long start, unsigned long len) | |
3632 | { | |
3633 | size_t cur; | |
3634 | size_t offset; | |
3635 | struct page *page; | |
3636 | char *kaddr; | |
3637 | size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3638 | unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT; | |
3639 | ||
3640 | WARN_ON(start > eb->len); | |
3641 | WARN_ON(start + len > eb->start + eb->len); | |
3642 | ||
3643 | offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3644 | ||
3645 | while (len > 0) { | |
3646 | page = extent_buffer_page(eb, i); | |
3647 | WARN_ON(!PageUptodate(page)); | |
3648 | ||
3649 | cur = min(len, PAGE_CACHE_SIZE - offset); | |
3650 | kaddr = kmap_atomic(page, KM_USER0); | |
3651 | memset(kaddr + offset, c, cur); | |
3652 | kunmap_atomic(kaddr, KM_USER0); | |
3653 | ||
3654 | len -= cur; | |
3655 | offset = 0; | |
3656 | i++; | |
3657 | } | |
3658 | } | |
3659 | ||
3660 | void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src, | |
3661 | unsigned long dst_offset, unsigned long src_offset, | |
3662 | unsigned long len) | |
3663 | { | |
3664 | u64 dst_len = dst->len; | |
3665 | size_t cur; | |
3666 | size_t offset; | |
3667 | struct page *page; | |
3668 | char *kaddr; | |
3669 | size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3670 | unsigned long i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT; | |
3671 | ||
3672 | WARN_ON(src->len != dst_len); | |
3673 | ||
3674 | offset = (start_offset + dst_offset) & | |
3675 | ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3676 | ||
3677 | while (len > 0) { | |
3678 | page = extent_buffer_page(dst, i); | |
3679 | WARN_ON(!PageUptodate(page)); | |
3680 | ||
3681 | cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset)); | |
3682 | ||
3683 | kaddr = kmap_atomic(page, KM_USER0); | |
3684 | read_extent_buffer(src, kaddr + offset, src_offset, cur); | |
3685 | kunmap_atomic(kaddr, KM_USER0); | |
3686 | ||
3687 | src_offset += cur; | |
3688 | len -= cur; | |
3689 | offset = 0; | |
3690 | i++; | |
3691 | } | |
3692 | } | |
3693 | ||
3694 | static void move_pages(struct page *dst_page, struct page *src_page, | |
3695 | unsigned long dst_off, unsigned long src_off, | |
3696 | unsigned long len) | |
3697 | { | |
3698 | char *dst_kaddr = kmap_atomic(dst_page, KM_USER0); | |
3699 | if (dst_page == src_page) { | |
3700 | memmove(dst_kaddr + dst_off, dst_kaddr + src_off, len); | |
3701 | } else { | |
3702 | char *src_kaddr = kmap_atomic(src_page, KM_USER1); | |
3703 | char *p = dst_kaddr + dst_off + len; | |
3704 | char *s = src_kaddr + src_off + len; | |
3705 | ||
3706 | while (len--) | |
3707 | *--p = *--s; | |
3708 | ||
3709 | kunmap_atomic(src_kaddr, KM_USER1); | |
3710 | } | |
3711 | kunmap_atomic(dst_kaddr, KM_USER0); | |
3712 | } | |
3713 | ||
3714 | static void copy_pages(struct page *dst_page, struct page *src_page, | |
3715 | unsigned long dst_off, unsigned long src_off, | |
3716 | unsigned long len) | |
3717 | { | |
3718 | char *dst_kaddr = kmap_atomic(dst_page, KM_USER0); | |
3719 | char *src_kaddr; | |
3720 | ||
3721 | if (dst_page != src_page) | |
3722 | src_kaddr = kmap_atomic(src_page, KM_USER1); | |
3723 | else | |
3724 | src_kaddr = dst_kaddr; | |
3725 | ||
3726 | memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len); | |
3727 | kunmap_atomic(dst_kaddr, KM_USER0); | |
3728 | if (dst_page != src_page) | |
3729 | kunmap_atomic(src_kaddr, KM_USER1); | |
3730 | } | |
3731 | ||
3732 | void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, | |
3733 | unsigned long src_offset, unsigned long len) | |
3734 | { | |
3735 | size_t cur; | |
3736 | size_t dst_off_in_page; | |
3737 | size_t src_off_in_page; | |
3738 | size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3739 | unsigned long dst_i; | |
3740 | unsigned long src_i; | |
3741 | ||
3742 | if (src_offset + len > dst->len) { | |
3743 | printk(KERN_ERR "btrfs memmove bogus src_offset %lu move " | |
3744 | "len %lu dst len %lu\n", src_offset, len, dst->len); | |
3745 | BUG_ON(1); | |
3746 | } | |
3747 | if (dst_offset + len > dst->len) { | |
3748 | printk(KERN_ERR "btrfs memmove bogus dst_offset %lu move " | |
3749 | "len %lu dst len %lu\n", dst_offset, len, dst->len); | |
3750 | BUG_ON(1); | |
3751 | } | |
3752 | ||
3753 | while (len > 0) { | |
3754 | dst_off_in_page = (start_offset + dst_offset) & | |
3755 | ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3756 | src_off_in_page = (start_offset + src_offset) & | |
3757 | ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3758 | ||
3759 | dst_i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT; | |
3760 | src_i = (start_offset + src_offset) >> PAGE_CACHE_SHIFT; | |
3761 | ||
3762 | cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - | |
3763 | src_off_in_page)); | |
3764 | cur = min_t(unsigned long, cur, | |
3765 | (unsigned long)(PAGE_CACHE_SIZE - dst_off_in_page)); | |
3766 | ||
3767 | copy_pages(extent_buffer_page(dst, dst_i), | |
3768 | extent_buffer_page(dst, src_i), | |
3769 | dst_off_in_page, src_off_in_page, cur); | |
3770 | ||
3771 | src_offset += cur; | |
3772 | dst_offset += cur; | |
3773 | len -= cur; | |
3774 | } | |
3775 | } | |
3776 | ||
3777 | void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset, | |
3778 | unsigned long src_offset, unsigned long len) | |
3779 | { | |
3780 | size_t cur; | |
3781 | size_t dst_off_in_page; | |
3782 | size_t src_off_in_page; | |
3783 | unsigned long dst_end = dst_offset + len - 1; | |
3784 | unsigned long src_end = src_offset + len - 1; | |
3785 | size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1); | |
3786 | unsigned long dst_i; | |
3787 | unsigned long src_i; | |
3788 | ||
3789 | if (src_offset + len > dst->len) { | |
3790 | printk(KERN_ERR "btrfs memmove bogus src_offset %lu move " | |
3791 | "len %lu len %lu\n", src_offset, len, dst->len); | |
3792 | BUG_ON(1); | |
3793 | } | |
3794 | if (dst_offset + len > dst->len) { | |
3795 | printk(KERN_ERR "btrfs memmove bogus dst_offset %lu move " | |
3796 | "len %lu len %lu\n", dst_offset, len, dst->len); | |
3797 | BUG_ON(1); | |
3798 | } | |
3799 | if (dst_offset < src_offset) { | |
3800 | memcpy_extent_buffer(dst, dst_offset, src_offset, len); | |
3801 | return; | |
3802 | } | |
3803 | while (len > 0) { | |
3804 | dst_i = (start_offset + dst_end) >> PAGE_CACHE_SHIFT; | |
3805 | src_i = (start_offset + src_end) >> PAGE_CACHE_SHIFT; | |
3806 | ||
3807 | dst_off_in_page = (start_offset + dst_end) & | |
3808 | ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3809 | src_off_in_page = (start_offset + src_end) & | |
3810 | ((unsigned long)PAGE_CACHE_SIZE - 1); | |
3811 | ||
3812 | cur = min_t(unsigned long, len, src_off_in_page + 1); | |
3813 | cur = min(cur, dst_off_in_page + 1); | |
3814 | move_pages(extent_buffer_page(dst, dst_i), | |
3815 | extent_buffer_page(dst, src_i), | |
3816 | dst_off_in_page - cur + 1, | |
3817 | src_off_in_page - cur + 1, cur); | |
3818 | ||
3819 | dst_end -= cur; | |
3820 | src_end -= cur; | |
3821 | len -= cur; | |
3822 | } | |
3823 | } | |
3824 | ||
3825 | int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page) | |
3826 | { | |
3827 | u64 start = page_offset(page); | |
3828 | struct extent_buffer *eb; | |
3829 | int ret = 1; | |
3830 | unsigned long i; | |
3831 | unsigned long num_pages; | |
3832 | ||
3833 | spin_lock(&tree->buffer_lock); | |
3834 | eb = buffer_search(tree, start); | |
3835 | if (!eb) | |
3836 | goto out; | |
3837 | ||
3838 | if (atomic_read(&eb->refs) > 1) { | |
3839 | ret = 0; | |
3840 | goto out; | |
3841 | } | |
3842 | if (test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) { | |
3843 | ret = 0; | |
3844 | goto out; | |
3845 | } | |
3846 | /* at this point we can safely release the extent buffer */ | |
3847 | num_pages = num_extent_pages(eb->start, eb->len); | |
3848 | for (i = 0; i < num_pages; i++) | |
3849 | page_cache_release(extent_buffer_page(eb, i)); | |
3850 | rb_erase(&eb->rb_node, &tree->buffer); | |
3851 | __free_extent_buffer(eb); | |
3852 | out: | |
3853 | spin_unlock(&tree->buffer_lock); | |
3854 | return ret; | |
3855 | } |