?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Add linux-next specific files for 20110421
[linux-2.6/next.git] / fs / btrfs / free-space-cache.c
blob11d2e9cea09e95b329cfe4e37aec5b50bcc46135
1 /*
2 * Copyright (C) 2008 Red Hat. All rights reserved.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
17 */
18
19 #include <linux/pagemap.h>
20 #include <linux/sched.h>
21 #include <linux/slab.h>
22 #include <linux/math64.h>
23 #include "ctree.h"
24 #include "free-space-cache.h"
25 #include "transaction.h"
26 #include "disk-io.h"
27 #include "extent_io.h"
28
29 #define BITS_PER_BITMAP (PAGE_CACHE_SIZE * 8)
30 #define MAX_CACHE_BYTES_PER_GIG (32 * 1024)
31
32 static void recalculate_thresholds(struct btrfs_block_group_cache
33 *block_group);
34 static int link_free_space(struct btrfs_block_group_cache *block_group,
35 struct btrfs_free_space *info);
36
37 struct inode *lookup_free_space_inode(struct btrfs_root *root,
38 struct btrfs_block_group_cache
39 *block_group, struct btrfs_path *path)
40 {
41 struct btrfs_key key;
42 struct btrfs_key location;
43 struct btrfs_disk_key disk_key;
44 struct btrfs_free_space_header *header;
45 struct extent_buffer *leaf;
46 struct inode *inode = NULL;
47 int ret;
48
49 spin_lock(&block_group->lock);
50 if (block_group->inode)
51 inode = igrab(block_group->inode);
52 spin_unlock(&block_group->lock);
53 if (inode)
54 return inode;
55
56 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
57 key.offset = block_group->key.objectid;
58 key.type = 0;
59
60 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
61 if (ret < 0)
62 return ERR_PTR(ret);
63 if (ret > 0) {
64 btrfs_release_path(root, path);
65 return ERR_PTR(-ENOENT);
66 }
67
68 leaf = path->nodes[0];
69 header = btrfs_item_ptr(leaf, path->slots[0],
70 struct btrfs_free_space_header);
71 btrfs_free_space_key(leaf, header, &disk_key);
72 btrfs_disk_key_to_cpu(&location, &disk_key);
73 btrfs_release_path(root, path);
74
75 inode = btrfs_iget(root->fs_info->sb, &location, root, NULL);
76 if (!inode)
77 return ERR_PTR(-ENOENT);
78 if (IS_ERR(inode))
79 return inode;
80 if (is_bad_inode(inode)) {
81 iput(inode);
82 return ERR_PTR(-ENOENT);
83 }
84
85 inode->i_mapping->flags &= ~__GFP_FS;
86
87 spin_lock(&block_group->lock);
88 if (!root->fs_info->closing) {
89 block_group->inode = igrab(inode);
90 block_group->iref = 1;
91 }
92 spin_unlock(&block_group->lock);
93
94 return inode;
95 }
96
97 int create_free_space_inode(struct btrfs_root *root,
98 struct btrfs_trans_handle *trans,
99 struct btrfs_block_group_cache *block_group,
100 struct btrfs_path *path)
101 {
102 struct btrfs_key key;
103 struct btrfs_disk_key disk_key;
104 struct btrfs_free_space_header *header;
105 struct btrfs_inode_item *inode_item;
106 struct extent_buffer *leaf;
107 u64 objectid;
108 int ret;
109
110 ret = btrfs_find_free_objectid(trans, root, 0, &objectid);
111 if (ret < 0)
112 return ret;
113
114 ret = btrfs_insert_empty_inode(trans, root, path, objectid);
115 if (ret)
116 return ret;
117
118 leaf = path->nodes[0];
119 inode_item = btrfs_item_ptr(leaf, path->slots[0],
120 struct btrfs_inode_item);
121 btrfs_item_key(leaf, &disk_key, path->slots[0]);
122 memset_extent_buffer(leaf, 0, (unsigned long)inode_item,
123 sizeof(*inode_item));
124 btrfs_set_inode_generation(leaf, inode_item, trans->transid);
125 btrfs_set_inode_size(leaf, inode_item, 0);
126 btrfs_set_inode_nbytes(leaf, inode_item, 0);
127 btrfs_set_inode_uid(leaf, inode_item, 0);
128 btrfs_set_inode_gid(leaf, inode_item, 0);
129 btrfs_set_inode_mode(leaf, inode_item, S_IFREG | 0600);
130 btrfs_set_inode_flags(leaf, inode_item, BTRFS_INODE_NOCOMPRESS |
131 BTRFS_INODE_PREALLOC | BTRFS_INODE_NODATASUM);
132 btrfs_set_inode_nlink(leaf, inode_item, 1);
133 btrfs_set_inode_transid(leaf, inode_item, trans->transid);
134 btrfs_set_inode_block_group(leaf, inode_item,
135 block_group->key.objectid);
136 btrfs_mark_buffer_dirty(leaf);
137 btrfs_release_path(root, path);
138
139 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
140 key.offset = block_group->key.objectid;
141 key.type = 0;
142
143 ret = btrfs_insert_empty_item(trans, root, path, &key,
144 sizeof(struct btrfs_free_space_header));
145 if (ret < 0) {
146 btrfs_release_path(root, path);
147 return ret;
148 }
149 leaf = path->nodes[0];
150 header = btrfs_item_ptr(leaf, path->slots[0],
151 struct btrfs_free_space_header);
152 memset_extent_buffer(leaf, 0, (unsigned long)header, sizeof(*header));
153 btrfs_set_free_space_key(leaf, header, &disk_key);
154 btrfs_mark_buffer_dirty(leaf);
155 btrfs_release_path(root, path);
156
157 return 0;
158 }
159
160 int btrfs_truncate_free_space_cache(struct btrfs_root *root,
161 struct btrfs_trans_handle *trans,
162 struct btrfs_path *path,
163 struct inode *inode)
164 {
165 loff_t oldsize;
166 int ret = 0;
167
168 trans->block_rsv = root->orphan_block_rsv;
169 ret = btrfs_block_rsv_check(trans, root,
170 root->orphan_block_rsv,
171 0, 5);
172 if (ret)
173 return ret;
174
175 oldsize = i_size_read(inode);
176 btrfs_i_size_write(inode, 0);
177 truncate_pagecache(inode, oldsize, 0);
178
179 /*
180 * We don't need an orphan item because truncating the free space cache
181 * will never be split across transactions.
182 */
183 ret = btrfs_truncate_inode_items(trans, root, inode,
184 0, BTRFS_EXTENT_DATA_KEY);
185 if (ret) {
186 WARN_ON(1);
187 return ret;
188 }
189
190 return btrfs_update_inode(trans, root, inode);
191 }
192
193 static int readahead_cache(struct inode *inode)
194 {
195 struct file_ra_state *ra;
196 unsigned long last_index;
197
198 ra = kzalloc(sizeof(*ra), GFP_NOFS);
199 if (!ra)
200 return -ENOMEM;
201
202 file_ra_state_init(ra, inode->i_mapping);
203 last_index = (i_size_read(inode) - 1) >> PAGE_CACHE_SHIFT;
204
205 page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
206
207 kfree(ra);
208
209 return 0;
210 }
211
212 int load_free_space_cache(struct btrfs_fs_info *fs_info,
213 struct btrfs_block_group_cache *block_group)
214 {
215 struct btrfs_root *root = fs_info->tree_root;
216 struct inode *inode;
217 struct btrfs_free_space_header *header;
218 struct extent_buffer *leaf;
219 struct page *page;
220 struct btrfs_path *path;
221 u32 *checksums = NULL, *crc;
222 char *disk_crcs = NULL;
223 struct btrfs_key key;
224 struct list_head bitmaps;
225 u64 num_entries;
226 u64 num_bitmaps;
227 u64 generation;
228 u64 used = btrfs_block_group_used(&block_group->item);
229 u32 cur_crc = ~(u32)0;
230 pgoff_t index = 0;
231 unsigned long first_page_offset;
232 int num_checksums;
233 int ret = 0;
234
235 /*
236 * If we're unmounting then just return, since this does a search on the
237 * normal root and not the commit root and we could deadlock.
238 */
239 smp_mb();
240 if (fs_info->closing)
241 return 0;
242
243 /*
244 * If this block group has been marked to be cleared for one reason or
245 * another then we can't trust the on disk cache, so just return.
246 */
247 spin_lock(&block_group->lock);
248 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
249 spin_unlock(&block_group->lock);
250 return 0;
251 }
252 spin_unlock(&block_group->lock);
253
254 INIT_LIST_HEAD(&bitmaps);
255
256 path = btrfs_alloc_path();
257 if (!path)
258 return 0;
259
260 inode = lookup_free_space_inode(root, block_group, path);
261 if (IS_ERR(inode)) {
262 btrfs_free_path(path);
263 return 0;
264 }
265
266 /* Nothing in the space cache, goodbye */
267 if (!i_size_read(inode)) {
268 btrfs_free_path(path);
269 goto out;
270 }
271
272 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
273 key.offset = block_group->key.objectid;
274 key.type = 0;
275
276 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
277 if (ret) {
278 btrfs_free_path(path);
279 goto out;
280 }
281
282 leaf = path->nodes[0];
283 header = btrfs_item_ptr(leaf, path->slots[0],
284 struct btrfs_free_space_header);
285 num_entries = btrfs_free_space_entries(leaf, header);
286 num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
287 generation = btrfs_free_space_generation(leaf, header);
288 btrfs_free_path(path);
289
290 if (BTRFS_I(inode)->generation != generation) {
291 printk(KERN_ERR "btrfs: free space inode generation (%llu) did"
292 " not match free space cache generation (%llu) for "
293 "block group %llu\n",
294 (unsigned long long)BTRFS_I(inode)->generation,
295 (unsigned long long)generation,
296 (unsigned long long)block_group->key.objectid);
297 goto free_cache;
298 }
299
300 if (!num_entries)
301 goto out;
302
303 /* Setup everything for doing checksumming */
304 num_checksums = i_size_read(inode) / PAGE_CACHE_SIZE;
305 checksums = crc = kzalloc(sizeof(u32) * num_checksums, GFP_NOFS);
306 if (!checksums)
307 goto out;
308 first_page_offset = (sizeof(u32) * num_checksums) + sizeof(u64);
309 disk_crcs = kzalloc(first_page_offset, GFP_NOFS);
310 if (!disk_crcs)
311 goto out;
312
313 ret = readahead_cache(inode);
314 if (ret) {
315 ret = 0;
316 goto out;
317 }
318
319 while (1) {
320 struct btrfs_free_space_entry *entry;
321 struct btrfs_free_space *e;
322 void *addr;
323 unsigned long offset = 0;
324 unsigned long start_offset = 0;
325 int need_loop = 0;
326
327 if (!num_entries && !num_bitmaps)
328 break;
329
330 if (index == 0) {
331 start_offset = first_page_offset;
332 offset = start_offset;
333 }
334
335 page = grab_cache_page(inode->i_mapping, index);
336 if (!page) {
337 ret = 0;
338 goto free_cache;
339 }
340
341 if (!PageUptodate(page)) {
342 btrfs_readpage(NULL, page);
343 lock_page(page);
344 if (!PageUptodate(page)) {
345 unlock_page(page);
346 page_cache_release(page);
347 printk(KERN_ERR "btrfs: error reading free "
348 "space cache: %llu\n",
349 (unsigned long long)
350 block_group->key.objectid);
351 goto free_cache;
352 }
353 }
354 addr = kmap(page);
355
356 if (index == 0) {
357 u64 *gen;
358
359 memcpy(disk_crcs, addr, first_page_offset);
360 gen = addr + (sizeof(u32) * num_checksums);
361 if (*gen != BTRFS_I(inode)->generation) {
362 printk(KERN_ERR "btrfs: space cache generation"
363 " (%llu) does not match inode (%llu) "
364 "for block group %llu\n",
365 (unsigned long long)*gen,
366 (unsigned long long)
367 BTRFS_I(inode)->generation,
368 (unsigned long long)
369 block_group->key.objectid);
370 kunmap(page);
371 unlock_page(page);
372 page_cache_release(page);
373 goto free_cache;
374 }
375 crc = (u32 *)disk_crcs;
376 }
377 entry = addr + start_offset;
378
379 /* First lets check our crc before we do anything fun */
380 cur_crc = ~(u32)0;
381 cur_crc = btrfs_csum_data(root, addr + start_offset, cur_crc,
382 PAGE_CACHE_SIZE - start_offset);
383 btrfs_csum_final(cur_crc, (char *)&cur_crc);
384 if (cur_crc != *crc) {
385 printk(KERN_ERR "btrfs: crc mismatch for page %lu in "
386 "block group %llu\n", index,
387 (unsigned long long)block_group->key.objectid);
388 kunmap(page);
389 unlock_page(page);
390 page_cache_release(page);
391 goto free_cache;
392 }
393 crc++;
394
395 while (1) {
396 if (!num_entries)
397 break;
398
399 need_loop = 1;
400 e = kmem_cache_zalloc(btrfs_free_space_cachep,
401 GFP_NOFS);
402 if (!e) {
403 kunmap(page);
404 unlock_page(page);
405 page_cache_release(page);
406 goto free_cache;
407 }
408
409 e->offset = le64_to_cpu(entry->offset);
410 e->bytes = le64_to_cpu(entry->bytes);
411 if (!e->bytes) {
412 kunmap(page);
413 kmem_cache_free(btrfs_free_space_cachep, e);
414 unlock_page(page);
415 page_cache_release(page);
416 goto free_cache;
417 }
418
419 if (entry->type == BTRFS_FREE_SPACE_EXTENT) {
420 spin_lock(&block_group->tree_lock);
421 ret = link_free_space(block_group, e);
422 spin_unlock(&block_group->tree_lock);
423 BUG_ON(ret);
424 } else {
425 e->bitmap = kzalloc(PAGE_CACHE_SIZE, GFP_NOFS);
426 if (!e->bitmap) {
427 kunmap(page);
428 kmem_cache_free(
429 btrfs_free_space_cachep, e);
430 unlock_page(page);
431 page_cache_release(page);
432 goto free_cache;
433 }
434 spin_lock(&block_group->tree_lock);
435 ret = link_free_space(block_group, e);
436 block_group->total_bitmaps++;
437 recalculate_thresholds(block_group);
438 spin_unlock(&block_group->tree_lock);
439 list_add_tail(&e->list, &bitmaps);
440 }
441
442 num_entries--;
443 offset += sizeof(struct btrfs_free_space_entry);
444 if (offset + sizeof(struct btrfs_free_space_entry) >=
445 PAGE_CACHE_SIZE)
446 break;
447 entry++;
448 }
449
450 /*
451 * We read an entry out of this page, we need to move on to the
452 * next page.
453 */
454 if (need_loop) {
455 kunmap(page);
456 goto next;
457 }
458
459 /*
460 * We add the bitmaps at the end of the entries in order that
461 * the bitmap entries are added to the cache.
462 */
463 e = list_entry(bitmaps.next, struct btrfs_free_space, list);
464 list_del_init(&e->list);
465 memcpy(e->bitmap, addr, PAGE_CACHE_SIZE);
466 kunmap(page);
467 num_bitmaps--;
468 next:
469 unlock_page(page);
470 page_cache_release(page);
471 index++;
472 }
473
474 spin_lock(&block_group->tree_lock);
475 if (block_group->free_space != (block_group->key.offset - used -
476 block_group->bytes_super)) {
477 spin_unlock(&block_group->tree_lock);
478 printk(KERN_ERR "block group %llu has an wrong amount of free "
479 "space\n", block_group->key.objectid);
480 ret = 0;
481 goto free_cache;
482 }
483 spin_unlock(&block_group->tree_lock);
484
485 ret = 1;
486 out:
487 kfree(checksums);
488 kfree(disk_crcs);
489 iput(inode);
490 return ret;
491
492 free_cache:
493 /* This cache is bogus, make sure it gets cleared */
494 spin_lock(&block_group->lock);
495 block_group->disk_cache_state = BTRFS_DC_CLEAR;
496 spin_unlock(&block_group->lock);
497 btrfs_remove_free_space_cache(block_group);
498 goto out;
499 }
500
501 int btrfs_write_out_cache(struct btrfs_root *root,
502 struct btrfs_trans_handle *trans,
503 struct btrfs_block_group_cache *block_group,
504 struct btrfs_path *path)
505 {
506 struct btrfs_free_space_header *header;
507 struct extent_buffer *leaf;
508 struct inode *inode;
509 struct rb_node *node;
510 struct list_head *pos, *n;
511 struct page **pages;
512 struct page *page;
513 struct extent_state *cached_state = NULL;
514 struct btrfs_free_cluster *cluster = NULL;
515 struct extent_io_tree *unpin = NULL;
516 struct list_head bitmap_list;
517 struct btrfs_key key;
518 u64 start, end, len;
519 u64 bytes = 0;
520 u32 *crc, *checksums;
521 unsigned long first_page_offset;
522 int index = 0, num_pages = 0;
523 int entries = 0;
524 int bitmaps = 0;
525 int ret = 0;
526 bool next_page = false;
527 bool out_of_space = false;
528
529 root = root->fs_info->tree_root;
530
531 INIT_LIST_HEAD(&bitmap_list);
532
533 spin_lock(&block_group->lock);
534 if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
535 spin_unlock(&block_group->lock);
536 return 0;
537 }
538 spin_unlock(&block_group->lock);
539
540 inode = lookup_free_space_inode(root, block_group, path);
541 if (IS_ERR(inode))
542 return 0;
543
544 if (!i_size_read(inode)) {
545 iput(inode);
546 return 0;
547 }
548
549 node = rb_first(&block_group->free_space_offset);
550 if (!node) {
551 iput(inode);
552 return 0;
553 }
554
555 num_pages = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
556 PAGE_CACHE_SHIFT;
557 filemap_write_and_wait(inode->i_mapping);
558 btrfs_wait_ordered_range(inode, inode->i_size &
559 ~(root->sectorsize - 1), (u64)-1);
560
561 /* We need a checksum per page. */
562 crc = checksums = kzalloc(sizeof(u32) * num_pages, GFP_NOFS);
563 if (!crc) {
564 iput(inode);
565 return 0;
566 }
567
568 pages = kzalloc(sizeof(struct page *) * num_pages, GFP_NOFS);
569 if (!pages) {
570 kfree(crc);
571 iput(inode);
572 return 0;
573 }
574
575 /* Since the first page has all of our checksums and our generation we
576 * need to calculate the offset into the page that we can start writing
577 * our entries.
578 */
579 first_page_offset = (sizeof(u32) * num_pages) + sizeof(u64);
580
581 /* Get the cluster for this block_group if it exists */
582 if (!list_empty(&block_group->cluster_list))
583 cluster = list_entry(block_group->cluster_list.next,
584 struct btrfs_free_cluster,
585 block_group_list);
586
587 /*
588 * We shouldn't have switched the pinned extents yet so this is the
589 * right one
590 */
591 unpin = root->fs_info->pinned_extents;
592
593 /*
594 * Lock all pages first so we can lock the extent safely.
595 *
596 * NOTE: Because we hold the ref the entire time we're going to write to
597 * the page find_get_page should never fail, so we don't do a check
598 * after find_get_page at this point. Just putting this here so people
599 * know and don't freak out.
600 */
601 while (index < num_pages) {
602 page = grab_cache_page(inode->i_mapping, index);
603 if (!page) {
604 int i;
605
606 for (i = 0; i < num_pages; i++) {
607 unlock_page(pages[i]);
608 page_cache_release(pages[i]);
609 }
610 goto out_free;
611 }
612 pages[index] = page;
613 index++;
614 }
615
616 index = 0;
617 lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
618 0, &cached_state, GFP_NOFS);
619
620 /*
621 * When searching for pinned extents, we need to start at our start
622 * offset.
623 */
624 start = block_group->key.objectid;
625
626 /* Write out the extent entries */
627 do {
628 struct btrfs_free_space_entry *entry;
629 void *addr;
630 unsigned long offset = 0;
631 unsigned long start_offset = 0;
632
633 next_page = false;
634
635 if (index == 0) {
636 start_offset = first_page_offset;
637 offset = start_offset;
638 }
639
640 if (index >= num_pages) {
641 out_of_space = true;
642 break;
643 }
644
645 page = pages[index];
646
647 addr = kmap(page);
648 entry = addr + start_offset;
649
650 memset(addr, 0, PAGE_CACHE_SIZE);
651 while (node && !next_page) {
652 struct btrfs_free_space *e;
653
654 e = rb_entry(node, struct btrfs_free_space, offset_index);
655 entries++;
656
657 entry->offset = cpu_to_le64(e->offset);
658 entry->bytes = cpu_to_le64(e->bytes);
659 if (e->bitmap) {
660 entry->type = BTRFS_FREE_SPACE_BITMAP;
661 list_add_tail(&e->list, &bitmap_list);
662 bitmaps++;
663 } else {
664 entry->type = BTRFS_FREE_SPACE_EXTENT;
665 }
666 node = rb_next(node);
667 if (!node && cluster) {
668 node = rb_first(&cluster->root);
669 cluster = NULL;
670 }
671 offset += sizeof(struct btrfs_free_space_entry);
672 if (offset + sizeof(struct btrfs_free_space_entry) >=
673 PAGE_CACHE_SIZE)
674 next_page = true;
675 entry++;
676 }
677
678 /*
679 * We want to add any pinned extents to our free space cache
680 * so we don't leak the space
681 */
682 while (!next_page && (start < block_group->key.objectid +
683 block_group->key.offset)) {
684 ret = find_first_extent_bit(unpin, start, &start, &end,
685 EXTENT_DIRTY);
686 if (ret) {
687 ret = 0;
688 break;
689 }
690
691 /* This pinned extent is out of our range */
692 if (start >= block_group->key.objectid +
693 block_group->key.offset)
694 break;
695
696 len = block_group->key.objectid +
697 block_group->key.offset - start;
698 len = min(len, end + 1 - start);
699
700 entries++;
701 entry->offset = cpu_to_le64(start);
702 entry->bytes = cpu_to_le64(len);
703 entry->type = BTRFS_FREE_SPACE_EXTENT;
704
705 start = end + 1;
706 offset += sizeof(struct btrfs_free_space_entry);
707 if (offset + sizeof(struct btrfs_free_space_entry) >=
708 PAGE_CACHE_SIZE)
709 next_page = true;
710 entry++;
711 }
712 *crc = ~(u32)0;
713 *crc = btrfs_csum_data(root, addr + start_offset, *crc,
714 PAGE_CACHE_SIZE - start_offset);
715 kunmap(page);
716
717 btrfs_csum_final(*crc, (char *)crc);
718 crc++;
719
720 bytes += PAGE_CACHE_SIZE;
721
722 index++;
723 } while (node || next_page);
724
725 /* Write out the bitmaps */
726 list_for_each_safe(pos, n, &bitmap_list) {
727 void *addr;
728 struct btrfs_free_space *entry =
729 list_entry(pos, struct btrfs_free_space, list);
730
731 if (index >= num_pages) {
732 out_of_space = true;
733 break;
734 }
735 page = pages[index];
736
737 addr = kmap(page);
738 memcpy(addr, entry->bitmap, PAGE_CACHE_SIZE);
739 *crc = ~(u32)0;
740 *crc = btrfs_csum_data(root, addr, *crc, PAGE_CACHE_SIZE);
741 kunmap(page);
742 btrfs_csum_final(*crc, (char *)crc);
743 crc++;
744 bytes += PAGE_CACHE_SIZE;
745
746 list_del_init(&entry->list);
747 index++;
748 }
749
750 if (out_of_space) {
751 btrfs_drop_pages(pages, num_pages);
752 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
753 i_size_read(inode) - 1, &cached_state,
754 GFP_NOFS);
755 ret = 0;
756 goto out_free;
757 }
758
759 /* Zero out the rest of the pages just to make sure */
760 while (index < num_pages) {
761 void *addr;
762
763 page = pages[index];
764 addr = kmap(page);
765 memset(addr, 0, PAGE_CACHE_SIZE);
766 kunmap(page);
767 bytes += PAGE_CACHE_SIZE;
768 index++;
769 }
770
771 /* Write the checksums and trans id to the first page */
772 {
773 void *addr;
774 u64 *gen;
775
776 page = pages[0];
777
778 addr = kmap(page);
779 memcpy(addr, checksums, sizeof(u32) * num_pages);
780 gen = addr + (sizeof(u32) * num_pages);
781 *gen = trans->transid;
782 kunmap(page);
783 }
784
785 ret = btrfs_dirty_pages(root, inode, pages, num_pages, 0,
786 bytes, &cached_state);
787 btrfs_drop_pages(pages, num_pages);
788 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
789 i_size_read(inode) - 1, &cached_state, GFP_NOFS);
790
791 if (ret) {
792 ret = 0;
793 goto out_free;
794 }
795
796 BTRFS_I(inode)->generation = trans->transid;
797
798 filemap_write_and_wait(inode->i_mapping);
799
800 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
801 key.offset = block_group->key.objectid;
802 key.type = 0;
803
804 ret = btrfs_search_slot(trans, root, &key, path, 1, 1);
805 if (ret < 0) {
806 ret = 0;
807 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, bytes - 1,
808 EXTENT_DIRTY | EXTENT_DELALLOC |
809 EXTENT_DO_ACCOUNTING, 0, 0, NULL, GFP_NOFS);
810 goto out_free;
811 }
812 leaf = path->nodes[0];
813 if (ret > 0) {
814 struct btrfs_key found_key;
815 BUG_ON(!path->slots[0]);
816 path->slots[0]--;
817 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
818 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
819 found_key.offset != block_group->key.objectid) {
820 ret = 0;
821 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, bytes - 1,
822 EXTENT_DIRTY | EXTENT_DELALLOC |
823 EXTENT_DO_ACCOUNTING, 0, 0, NULL,
824 GFP_NOFS);
825 btrfs_release_path(root, path);
826 goto out_free;
827 }
828 }
829 header = btrfs_item_ptr(leaf, path->slots[0],
830 struct btrfs_free_space_header);
831 btrfs_set_free_space_entries(leaf, header, entries);
832 btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
833 btrfs_set_free_space_generation(leaf, header, trans->transid);
834 btrfs_mark_buffer_dirty(leaf);
835 btrfs_release_path(root, path);
836
837 ret = 1;
838
839 out_free:
840 if (ret == 0) {
841 invalidate_inode_pages2_range(inode->i_mapping, 0, index);
842 spin_lock(&block_group->lock);
843 block_group->disk_cache_state = BTRFS_DC_ERROR;
844 spin_unlock(&block_group->lock);
845 BTRFS_I(inode)->generation = 0;
846 }
847 kfree(checksums);
848 kfree(pages);
849 btrfs_update_inode(trans, root, inode);
850 iput(inode);
851 return ret;
852 }
853
854 static inline unsigned long offset_to_bit(u64 bitmap_start, u64 sectorsize,
855 u64 offset)
856 {
857 BUG_ON(offset < bitmap_start);
858 offset -= bitmap_start;
859 return (unsigned long)(div64_u64(offset, sectorsize));
860 }
861
862 static inline unsigned long bytes_to_bits(u64 bytes, u64 sectorsize)
863 {
864 return (unsigned long)(div64_u64(bytes, sectorsize));
865 }
866
867 static inline u64 offset_to_bitmap(struct btrfs_block_group_cache *block_group,
868 u64 offset)
869 {
870 u64 bitmap_start;
871 u64 bytes_per_bitmap;
872
873 bytes_per_bitmap = BITS_PER_BITMAP * block_group->sectorsize;
874 bitmap_start = offset - block_group->key.objectid;
875 bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
876 bitmap_start *= bytes_per_bitmap;
877 bitmap_start += block_group->key.objectid;
878
879 return bitmap_start;
880 }
881
882 static int tree_insert_offset(struct rb_root *root, u64 offset,
883 struct rb_node *node, int bitmap)
884 {
885 struct rb_node **p = &root->rb_node;
886 struct rb_node *parent = NULL;
887 struct btrfs_free_space *info;
888
889 while (*p) {
890 parent = *p;
891 info = rb_entry(parent, struct btrfs_free_space, offset_index);
892
893 if (offset < info->offset) {
894 p = &(*p)->rb_left;
895 } else if (offset > info->offset) {
896 p = &(*p)->rb_right;
897 } else {
898 /*
899 * we could have a bitmap entry and an extent entry
900 * share the same offset. If this is the case, we want
901 * the extent entry to always be found first if we do a
902 * linear search through the tree, since we want to have
903 * the quickest allocation time, and allocating from an
904 * extent is faster than allocating from a bitmap. So
905 * if we're inserting a bitmap and we find an entry at
906 * this offset, we want to go right, or after this entry
907 * logically. If we are inserting an extent and we've
908 * found a bitmap, we want to go left, or before
909 * logically.
910 */
911 if (bitmap) {
912 WARN_ON(info->bitmap);
913 p = &(*p)->rb_right;
914 } else {
915 WARN_ON(!info->bitmap);
916 p = &(*p)->rb_left;
917 }
918 }
919 }
920
921 rb_link_node(node, parent, p);
922 rb_insert_color(node, root);
923
924 return 0;
925 }
926
927 /*
928 * searches the tree for the given offset.
929 *
930 * fuzzy - If this is set, then we are trying to make an allocation, and we just
931 * want a section that has at least bytes size and comes at or after the given
932 * offset.
933 */
934 static struct btrfs_free_space *
935 tree_search_offset(struct btrfs_block_group_cache *block_group,
936 u64 offset, int bitmap_only, int fuzzy)
937 {
938 struct rb_node *n = block_group->free_space_offset.rb_node;
939 struct btrfs_free_space *entry, *prev = NULL;
940
941 /* find entry that is closest to the 'offset' */
942 while (1) {
943 if (!n) {
944 entry = NULL;
945 break;
946 }
947
948 entry = rb_entry(n, struct btrfs_free_space, offset_index);
949 prev = entry;
950
951 if (offset < entry->offset)
952 n = n->rb_left;
953 else if (offset > entry->offset)
954 n = n->rb_right;
955 else
956 break;
957 }
958
959 if (bitmap_only) {
960 if (!entry)
961 return NULL;
962 if (entry->bitmap)
963 return entry;
964
965 /*
966 * bitmap entry and extent entry may share same offset,
967 * in that case, bitmap entry comes after extent entry.
968 */
969 n = rb_next(n);
970 if (!n)
971 return NULL;
972 entry = rb_entry(n, struct btrfs_free_space, offset_index);
973 if (entry->offset != offset)
974 return NULL;
975
976 WARN_ON(!entry->bitmap);
977 return entry;
978 } else if (entry) {
979 if (entry->bitmap) {
980 /*
981 * if previous extent entry covers the offset,
982 * we should return it instead of the bitmap entry
983 */
984 n = &entry->offset_index;
985 while (1) {
986 n = rb_prev(n);
987 if (!n)
988 break;
989 prev = rb_entry(n, struct btrfs_free_space,
990 offset_index);
991 if (!prev->bitmap) {
992 if (prev->offset + prev->bytes > offset)
993 entry = prev;
994 break;
995 }
996 }
997 }
998 return entry;
999 }
1000
1001 if (!prev)
1002 return NULL;
1003
1004 /* find last entry before the 'offset' */
1005 entry = prev;
1006 if (entry->offset > offset) {
1007 n = rb_prev(&entry->offset_index);
1008 if (n) {
1009 entry = rb_entry(n, struct btrfs_free_space,
1010 offset_index);
1011 BUG_ON(entry->offset > offset);
1012 } else {
1013 if (fuzzy)
1014 return entry;
1015 else
1016 return NULL;
1017 }
1018 }
1019
1020 if (entry->bitmap) {
1021 n = &entry->offset_index;
1022 while (1) {
1023 n = rb_prev(n);
1024 if (!n)
1025 break;
1026 prev = rb_entry(n, struct btrfs_free_space,
1027 offset_index);
1028 if (!prev->bitmap) {
1029 if (prev->offset + prev->bytes > offset)
1030 return prev;
1031 break;
1032 }
1033 }
1034 if (entry->offset + BITS_PER_BITMAP *
1035 block_group->sectorsize > offset)
1036 return entry;
1037 } else if (entry->offset + entry->bytes > offset)
1038 return entry;
1039
1040 if (!fuzzy)
1041 return NULL;
1042
1043 while (1) {
1044 if (entry->bitmap) {
1045 if (entry->offset + BITS_PER_BITMAP *
1046 block_group->sectorsize > offset)
1047 break;
1048 } else {
1049 if (entry->offset + entry->bytes > offset)
1050 break;
1051 }
1052
1053 n = rb_next(&entry->offset_index);
1054 if (!n)
1055 return NULL;
1056 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1057 }
1058 return entry;
1059 }
1060
1061 static inline void
1062 __unlink_free_space(struct btrfs_block_group_cache *block_group,
1063 struct btrfs_free_space *info)
1064 {
1065 rb_erase(&info->offset_index, &block_group->free_space_offset);
1066 block_group->free_extents--;
1067 }
1068
1069 static void unlink_free_space(struct btrfs_block_group_cache *block_group,
1070 struct btrfs_free_space *info)
1071 {
1072 __unlink_free_space(block_group, info);
1073 block_group->free_space -= info->bytes;
1074 }
1075
1076 static int link_free_space(struct btrfs_block_group_cache *block_group,
1077 struct btrfs_free_space *info)
1078 {
1079 int ret = 0;
1080
1081 BUG_ON(!info->bitmap && !info->bytes);
1082 ret = tree_insert_offset(&block_group->free_space_offset, info->offset,
1083 &info->offset_index, (info->bitmap != NULL));
1084 if (ret)
1085 return ret;
1086
1087 block_group->free_space += info->bytes;
1088 block_group->free_extents++;
1089 return ret;
1090 }
1091
1092 static void recalculate_thresholds(struct btrfs_block_group_cache *block_group)
1093 {
1094 u64 max_bytes;
1095 u64 bitmap_bytes;
1096 u64 extent_bytes;
1097 u64 size = block_group->key.offset;
1098
1099 /*
1100 * The goal is to keep the total amount of memory used per 1gb of space
1101 * at or below 32k, so we need to adjust how much memory we allow to be
1102 * used by extent based free space tracking
1103 */
1104 if (size < 1024 * 1024 * 1024)
1105 max_bytes = MAX_CACHE_BYTES_PER_GIG;
1106 else
1107 max_bytes = MAX_CACHE_BYTES_PER_GIG *
1108 div64_u64(size, 1024 * 1024 * 1024);
1109
1110 /*
1111 * we want to account for 1 more bitmap than what we have so we can make
1112 * sure we don't go over our overall goal of MAX_CACHE_BYTES_PER_GIG as
1113 * we add more bitmaps.
1114 */
1115 bitmap_bytes = (block_group->total_bitmaps + 1) * PAGE_CACHE_SIZE;
1116
1117 if (bitmap_bytes >= max_bytes) {
1118 block_group->extents_thresh = 0;
1119 return;
1120 }
1121
1122 /*
1123 * we want the extent entry threshold to always be at most 1/2 the maxw
1124 * bytes we can have, or whatever is less than that.
1125 */
1126 extent_bytes = max_bytes - bitmap_bytes;
1127 extent_bytes = min_t(u64, extent_bytes, div64_u64(max_bytes, 2));
1128
1129 block_group->extents_thresh =
1130 div64_u64(extent_bytes, (sizeof(struct btrfs_free_space)));
1131 }
1132
1133 static void bitmap_clear_bits(struct btrfs_block_group_cache *block_group,
1134 struct btrfs_free_space *info, u64 offset,
1135 u64 bytes)
1136 {
1137 unsigned long start, end;
1138 unsigned long i;
1139
1140 start = offset_to_bit(info->offset, block_group->sectorsize, offset);
1141 end = start + bytes_to_bits(bytes, block_group->sectorsize);
1142 BUG_ON(end > BITS_PER_BITMAP);
1143
1144 for (i = start; i < end; i++)
1145 clear_bit(i, info->bitmap);
1146
1147 info->bytes -= bytes;
1148 block_group->free_space -= bytes;
1149 }
1150
1151 static void bitmap_set_bits(struct btrfs_block_group_cache *block_group,
1152 struct btrfs_free_space *info, u64 offset,
1153 u64 bytes)
1154 {
1155 unsigned long start, end;
1156 unsigned long i;
1157
1158 start = offset_to_bit(info->offset, block_group->sectorsize, offset);
1159 end = start + bytes_to_bits(bytes, block_group->sectorsize);
1160 BUG_ON(end > BITS_PER_BITMAP);
1161
1162 for (i = start; i < end; i++)
1163 set_bit(i, info->bitmap);
1164
1165 info->bytes += bytes;
1166 block_group->free_space += bytes;
1167 }
1168
1169 static int search_bitmap(struct btrfs_block_group_cache *block_group,
1170 struct btrfs_free_space *bitmap_info, u64 *offset,
1171 u64 *bytes)
1172 {
1173 unsigned long found_bits = 0;
1174 unsigned long bits, i;
1175 unsigned long next_zero;
1176
1177 i = offset_to_bit(bitmap_info->offset, block_group->sectorsize,
1178 max_t(u64, *offset, bitmap_info->offset));
1179 bits = bytes_to_bits(*bytes, block_group->sectorsize);
1180
1181 for (i = find_next_bit(bitmap_info->bitmap, BITS_PER_BITMAP, i);
1182 i < BITS_PER_BITMAP;
1183 i = find_next_bit(bitmap_info->bitmap, BITS_PER_BITMAP, i + 1)) {
1184 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1185 BITS_PER_BITMAP, i);
1186 if ((next_zero - i) >= bits) {
1187 found_bits = next_zero - i;
1188 break;
1189 }
1190 i = next_zero;
1191 }
1192
1193 if (found_bits) {
1194 *offset = (u64)(i * block_group->sectorsize) +
1195 bitmap_info->offset;
1196 *bytes = (u64)(found_bits) * block_group->sectorsize;
1197 return 0;
1198 }
1199
1200 return -1;
1201 }
1202
1203 static struct btrfs_free_space *find_free_space(struct btrfs_block_group_cache
1204 *block_group, u64 *offset,
1205 u64 *bytes, int debug)
1206 {
1207 struct btrfs_free_space *entry;
1208 struct rb_node *node;
1209 int ret;
1210
1211 if (!block_group->free_space_offset.rb_node)
1212 return NULL;
1213
1214 entry = tree_search_offset(block_group,
1215 offset_to_bitmap(block_group, *offset),
1216 0, 1);
1217 if (!entry)
1218 return NULL;
1219
1220 for (node = &entry->offset_index; node; node = rb_next(node)) {
1221 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1222 if (entry->bytes < *bytes)
1223 continue;
1224
1225 if (entry->bitmap) {
1226 ret = search_bitmap(block_group, entry, offset, bytes);
1227 if (!ret)
1228 return entry;
1229 continue;
1230 }
1231
1232 *offset = entry->offset;
1233 *bytes = entry->bytes;
1234 return entry;
1235 }
1236
1237 return NULL;
1238 }
1239
1240 static void add_new_bitmap(struct btrfs_block_group_cache *block_group,
1241 struct btrfs_free_space *info, u64 offset)
1242 {
1243 u64 bytes_per_bg = BITS_PER_BITMAP * block_group->sectorsize;
1244 int max_bitmaps = (int)div64_u64(block_group->key.offset +
1245 bytes_per_bg - 1, bytes_per_bg);
1246 BUG_ON(block_group->total_bitmaps >= max_bitmaps);
1247
1248 info->offset = offset_to_bitmap(block_group, offset);
1249 info->bytes = 0;
1250 link_free_space(block_group, info);
1251 block_group->total_bitmaps++;
1252
1253 recalculate_thresholds(block_group);
1254 }
1255
1256 static void free_bitmap(struct btrfs_block_group_cache *block_group,
1257 struct btrfs_free_space *bitmap_info)
1258 {
1259 unlink_free_space(block_group, bitmap_info);
1260 kfree(bitmap_info->bitmap);
1261 kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1262 block_group->total_bitmaps--;
1263 recalculate_thresholds(block_group);
1264 }
1265
1266 static noinline int remove_from_bitmap(struct btrfs_block_group_cache *block_group,
1267 struct btrfs_free_space *bitmap_info,
1268 u64 *offset, u64 *bytes)
1269 {
1270 u64 end;
1271 u64 search_start, search_bytes;
1272 int ret;
1273
1274 again:
1275 end = bitmap_info->offset +
1276 (u64)(BITS_PER_BITMAP * block_group->sectorsize) - 1;
1277
1278 /*
1279 * XXX - this can go away after a few releases.
1280 *
1281 * since the only user of btrfs_remove_free_space is the tree logging
1282 * stuff, and the only way to test that is under crash conditions, we
1283 * want to have this debug stuff here just in case somethings not
1284 * working. Search the bitmap for the space we are trying to use to
1285 * make sure its actually there. If its not there then we need to stop
1286 * because something has gone wrong.
1287 */
1288 search_start = *offset;
1289 search_bytes = *bytes;
1290 search_bytes = min(search_bytes, end - search_start + 1);
1291 ret = search_bitmap(block_group, bitmap_info, &search_start,
1292 &search_bytes);
1293 BUG_ON(ret < 0 || search_start != *offset);
1294
1295 if (*offset > bitmap_info->offset && *offset + *bytes > end) {
1296 bitmap_clear_bits(block_group, bitmap_info, *offset,
1297 end - *offset + 1);
1298 *bytes -= end - *offset + 1;
1299 *offset = end + 1;
1300 } else if (*offset >= bitmap_info->offset && *offset + *bytes <= end) {
1301 bitmap_clear_bits(block_group, bitmap_info, *offset, *bytes);
1302 *bytes = 0;
1303 }
1304
1305 if (*bytes) {
1306 struct rb_node *next = rb_next(&bitmap_info->offset_index);
1307 if (!bitmap_info->bytes)
1308 free_bitmap(block_group, bitmap_info);
1309
1310 /*
1311 * no entry after this bitmap, but we still have bytes to
1312 * remove, so something has gone wrong.
1313 */
1314 if (!next)
1315 return -EINVAL;
1316
1317 bitmap_info = rb_entry(next, struct btrfs_free_space,
1318 offset_index);
1319
1320 /*
1321 * if the next entry isn't a bitmap we need to return to let the
1322 * extent stuff do its work.
1323 */
1324 if (!bitmap_info->bitmap)
1325 return -EAGAIN;
1326
1327 /*
1328 * Ok the next item is a bitmap, but it may not actually hold
1329 * the information for the rest of this free space stuff, so
1330 * look for it, and if we don't find it return so we can try
1331 * everything over again.
1332 */
1333 search_start = *offset;
1334 search_bytes = *bytes;
1335 ret = search_bitmap(block_group, bitmap_info, &search_start,
1336 &search_bytes);
1337 if (ret < 0 || search_start != *offset)
1338 return -EAGAIN;
1339
1340 goto again;
1341 } else if (!bitmap_info->bytes)
1342 free_bitmap(block_group, bitmap_info);
1343
1344 return 0;
1345 }
1346
1347 static int insert_into_bitmap(struct btrfs_block_group_cache *block_group,
1348 struct btrfs_free_space *info)
1349 {
1350 struct btrfs_free_space *bitmap_info;
1351 int added = 0;
1352 u64 bytes, offset, end;
1353 int ret;
1354
1355 /*
1356 * If we are below the extents threshold then we can add this as an
1357 * extent, and don't have to deal with the bitmap
1358 */
1359 if (block_group->free_extents < block_group->extents_thresh) {
1360 /*
1361 * If this block group has some small extents we don't want to
1362 * use up all of our free slots in the cache with them, we want
1363 * to reserve them to larger extents, however if we have plent
1364 * of cache left then go ahead an dadd them, no sense in adding
1365 * the overhead of a bitmap if we don't have to.
1366 */
1367 if (info->bytes <= block_group->sectorsize * 4) {
1368 if (block_group->free_extents * 2 <=
1369 block_group->extents_thresh)
1370 return 0;
1371 } else {
1372 return 0;
1373 }
1374 }
1375
1376 /*
1377 * some block groups are so tiny they can't be enveloped by a bitmap, so
1378 * don't even bother to create a bitmap for this
1379 */
1380 if (BITS_PER_BITMAP * block_group->sectorsize >
1381 block_group->key.offset)
1382 return 0;
1383
1384 bytes = info->bytes;
1385 offset = info->offset;
1386
1387 again:
1388 bitmap_info = tree_search_offset(block_group,
1389 offset_to_bitmap(block_group, offset),
1390 1, 0);
1391 if (!bitmap_info) {
1392 BUG_ON(added);
1393 goto new_bitmap;
1394 }
1395
1396 end = bitmap_info->offset +
1397 (u64)(BITS_PER_BITMAP * block_group->sectorsize);
1398
1399 if (offset >= bitmap_info->offset && offset + bytes > end) {
1400 bitmap_set_bits(block_group, bitmap_info, offset,
1401 end - offset);
1402 bytes -= end - offset;
1403 offset = end;
1404 added = 0;
1405 } else if (offset >= bitmap_info->offset && offset + bytes <= end) {
1406 bitmap_set_bits(block_group, bitmap_info, offset, bytes);
1407 bytes = 0;
1408 } else {
1409 BUG();
1410 }
1411
1412 if (!bytes) {
1413 ret = 1;
1414 goto out;
1415 } else
1416 goto again;
1417
1418 new_bitmap:
1419 if (info && info->bitmap) {
1420 add_new_bitmap(block_group, info, offset);
1421 added = 1;
1422 info = NULL;
1423 goto again;
1424 } else {
1425 spin_unlock(&block_group->tree_lock);
1426
1427 /* no pre-allocated info, allocate a new one */
1428 if (!info) {
1429 info = kmem_cache_zalloc(btrfs_free_space_cachep,
1430 GFP_NOFS);
1431 if (!info) {
1432 spin_lock(&block_group->tree_lock);
1433 ret = -ENOMEM;
1434 goto out;
1435 }
1436 }
1437
1438 /* allocate the bitmap */
1439 info->bitmap = kzalloc(PAGE_CACHE_SIZE, GFP_NOFS);
1440 spin_lock(&block_group->tree_lock);
1441 if (!info->bitmap) {
1442 ret = -ENOMEM;
1443 goto out;
1444 }
1445 goto again;
1446 }
1447
1448 out:
1449 if (info) {
1450 if (info->bitmap)
1451 kfree(info->bitmap);
1452 kmem_cache_free(btrfs_free_space_cachep, info);
1453 }
1454
1455 return ret;
1456 }
1457
1458 bool try_merge_free_space(struct btrfs_block_group_cache *block_group,
1459 struct btrfs_free_space *info, bool update_stat)
1460 {
1461 struct btrfs_free_space *left_info;
1462 struct btrfs_free_space *right_info;
1463 bool merged = false;
1464 u64 offset = info->offset;
1465 u64 bytes = info->bytes;
1466
1467 /*
1468 * first we want to see if there is free space adjacent to the range we
1469 * are adding, if there is remove that struct and add a new one to
1470 * cover the entire range
1471 */
1472 right_info = tree_search_offset(block_group, offset + bytes, 0, 0);
1473 if (right_info && rb_prev(&right_info->offset_index))
1474 left_info = rb_entry(rb_prev(&right_info->offset_index),
1475 struct btrfs_free_space, offset_index);
1476 else
1477 left_info = tree_search_offset(block_group, offset - 1, 0, 0);
1478
1479 if (right_info && !right_info->bitmap) {
1480 if (update_stat)
1481 unlink_free_space(block_group, right_info);
1482 else
1483 __unlink_free_space(block_group, right_info);
1484 info->bytes += right_info->bytes;
1485 kmem_cache_free(btrfs_free_space_cachep, right_info);
1486 merged = true;
1487 }
1488
1489 if (left_info && !left_info->bitmap &&
1490 left_info->offset + left_info->bytes == offset) {
1491 if (update_stat)
1492 unlink_free_space(block_group, left_info);
1493 else
1494 __unlink_free_space(block_group, left_info);
1495 info->offset = left_info->offset;
1496 info->bytes += left_info->bytes;
1497 kmem_cache_free(btrfs_free_space_cachep, left_info);
1498 merged = true;
1499 }
1500
1501 return merged;
1502 }
1503
1504 int btrfs_add_free_space(struct btrfs_block_group_cache *block_group,
1505 u64 offset, u64 bytes)
1506 {
1507 struct btrfs_free_space *info;
1508 int ret = 0;
1509
1510 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
1511 if (!info)
1512 return -ENOMEM;
1513
1514 info->offset = offset;
1515 info->bytes = bytes;
1516
1517 spin_lock(&block_group->tree_lock);
1518
1519 if (try_merge_free_space(block_group, info, true))
1520 goto link;
1521
1522 /*
1523 * There was no extent directly to the left or right of this new
1524 * extent then we know we're going to have to allocate a new extent, so
1525 * before we do that see if we need to drop this into a bitmap
1526 */
1527 ret = insert_into_bitmap(block_group, info);
1528 if (ret < 0) {
1529 goto out;
1530 } else if (ret) {
1531 ret = 0;
1532 goto out;
1533 }
1534 link:
1535 ret = link_free_space(block_group, info);
1536 if (ret)
1537 kmem_cache_free(btrfs_free_space_cachep, info);
1538 out:
1539 spin_unlock(&block_group->tree_lock);
1540
1541 if (ret) {
1542 printk(KERN_CRIT "btrfs: unable to add free space :%d\n", ret);
1543 BUG_ON(ret == -EEXIST);
1544 }
1545
1546 return ret;
1547 }
1548
1549 int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
1550 u64 offset, u64 bytes)
1551 {
1552 struct btrfs_free_space *info;
1553 struct btrfs_free_space *next_info = NULL;
1554 int ret = 0;
1555
1556 spin_lock(&block_group->tree_lock);
1557
1558 again:
1559 info = tree_search_offset(block_group, offset, 0, 0);
1560 if (!info) {
1561 /*
1562 * oops didn't find an extent that matched the space we wanted
1563 * to remove, look for a bitmap instead
1564 */
1565 info = tree_search_offset(block_group,
1566 offset_to_bitmap(block_group, offset),
1567 1, 0);
1568 if (!info) {
1569 WARN_ON(1);
1570 goto out_lock;
1571 }
1572 }
1573
1574 if (info->bytes < bytes && rb_next(&info->offset_index)) {
1575 u64 end;
1576 next_info = rb_entry(rb_next(&info->offset_index),
1577 struct btrfs_free_space,
1578 offset_index);
1579
1580 if (next_info->bitmap)
1581 end = next_info->offset + BITS_PER_BITMAP *
1582 block_group->sectorsize - 1;
1583 else
1584 end = next_info->offset + next_info->bytes;
1585
1586 if (next_info->bytes < bytes ||
1587 next_info->offset > offset || offset > end) {
1588 printk(KERN_CRIT "Found free space at %llu, size %llu,"
1589 " trying to use %llu\n",
1590 (unsigned long long)info->offset,
1591 (unsigned long long)info->bytes,
1592 (unsigned long long)bytes);
1593 WARN_ON(1);
1594 ret = -EINVAL;
1595 goto out_lock;
1596 }
1597
1598 info = next_info;
1599 }
1600
1601 if (info->bytes == bytes) {
1602 unlink_free_space(block_group, info);
1603 if (info->bitmap) {
1604 kfree(info->bitmap);
1605 block_group->total_bitmaps--;
1606 }
1607 kmem_cache_free(btrfs_free_space_cachep, info);
1608 goto out_lock;
1609 }
1610
1611 if (!info->bitmap && info->offset == offset) {
1612 unlink_free_space(block_group, info);
1613 info->offset += bytes;
1614 info->bytes -= bytes;
1615 link_free_space(block_group, info);
1616 goto out_lock;
1617 }
1618
1619 if (!info->bitmap && info->offset <= offset &&
1620 info->offset + info->bytes >= offset + bytes) {
1621 u64 old_start = info->offset;
1622 /*
1623 * we're freeing space in the middle of the info,
1624 * this can happen during tree log replay
1625 *
1626 * first unlink the old info and then
1627 * insert it again after the hole we're creating
1628 */
1629 unlink_free_space(block_group, info);
1630 if (offset + bytes < info->offset + info->bytes) {
1631 u64 old_end = info->offset + info->bytes;
1632
1633 info->offset = offset + bytes;
1634 info->bytes = old_end - info->offset;
1635 ret = link_free_space(block_group, info);
1636 WARN_ON(ret);
1637 if (ret)
1638 goto out_lock;
1639 } else {
1640 /* the hole we're creating ends at the end
1641 * of the info struct, just free the info
1642 */
1643 kmem_cache_free(btrfs_free_space_cachep, info);
1644 }
1645 spin_unlock(&block_group->tree_lock);
1646
1647 /* step two, insert a new info struct to cover
1648 * anything before the hole
1649 */
1650 ret = btrfs_add_free_space(block_group, old_start,
1651 offset - old_start);
1652 WARN_ON(ret);
1653 goto out;
1654 }
1655
1656 ret = remove_from_bitmap(block_group, info, &offset, &bytes);
1657 if (ret == -EAGAIN)
1658 goto again;
1659 BUG_ON(ret);
1660 out_lock:
1661 spin_unlock(&block_group->tree_lock);
1662 out:
1663 return ret;
1664 }
1665
1666 void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
1667 u64 bytes)
1668 {
1669 struct btrfs_free_space *info;
1670 struct rb_node *n;
1671 int count = 0;
1672
1673 for (n = rb_first(&block_group->free_space_offset); n; n = rb_next(n)) {
1674 info = rb_entry(n, struct btrfs_free_space, offset_index);
1675 if (info->bytes >= bytes)
1676 count++;
1677 printk(KERN_CRIT "entry offset %llu, bytes %llu, bitmap %s\n",
1678 (unsigned long long)info->offset,
1679 (unsigned long long)info->bytes,
1680 (info->bitmap) ? "yes" : "no");
1681 }
1682 printk(KERN_INFO "block group has cluster?: %s\n",
1683 list_empty(&block_group->cluster_list) ? "no" : "yes");
1684 printk(KERN_INFO "%d blocks of free space at or bigger than bytes is"
1685 "\n", count);
1686 }
1687
1688 u64 btrfs_block_group_free_space(struct btrfs_block_group_cache *block_group)
1689 {
1690 struct btrfs_free_space *info;
1691 struct rb_node *n;
1692 u64 ret = 0;
1693
1694 for (n = rb_first(&block_group->free_space_offset); n;
1695 n = rb_next(n)) {
1696 info = rb_entry(n, struct btrfs_free_space, offset_index);
1697 ret += info->bytes;
1698 }
1699
1700 return ret;
1701 }
1702
1703 /*
1704 * for a given cluster, put all of its extents back into the free
1705 * space cache. If the block group passed doesn't match the block group
1706 * pointed to by the cluster, someone else raced in and freed the
1707 * cluster already. In that case, we just return without changing anything
1708 */
1709 static int
1710 __btrfs_return_cluster_to_free_space(
1711 struct btrfs_block_group_cache *block_group,
1712 struct btrfs_free_cluster *cluster)
1713 {
1714 struct btrfs_free_space *entry;
1715 struct rb_node *node;
1716
1717 spin_lock(&cluster->lock);
1718 if (cluster->block_group != block_group)
1719 goto out;
1720
1721 cluster->block_group = NULL;
1722 cluster->window_start = 0;
1723 list_del_init(&cluster->block_group_list);
1724
1725 node = rb_first(&cluster->root);
1726 while (node) {
1727 bool bitmap;
1728
1729 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1730 node = rb_next(&entry->offset_index);
1731 rb_erase(&entry->offset_index, &cluster->root);
1732
1733 bitmap = (entry->bitmap != NULL);
1734 if (!bitmap)
1735 try_merge_free_space(block_group, entry, false);
1736 tree_insert_offset(&block_group->free_space_offset,
1737 entry->offset, &entry->offset_index, bitmap);
1738 }
1739 cluster->root = RB_ROOT;
1740
1741 out:
1742 spin_unlock(&cluster->lock);
1743 btrfs_put_block_group(block_group);
1744 return 0;
1745 }
1746
1747 void btrfs_remove_free_space_cache(struct btrfs_block_group_cache *block_group)
1748 {
1749 struct btrfs_free_space *info;
1750 struct rb_node *node;
1751 struct btrfs_free_cluster *cluster;
1752 struct list_head *head;
1753
1754 spin_lock(&block_group->tree_lock);
1755 while ((head = block_group->cluster_list.next) !=
1756 &block_group->cluster_list) {
1757 cluster = list_entry(head, struct btrfs_free_cluster,
1758 block_group_list);
1759
1760 WARN_ON(cluster->block_group != block_group);
1761 __btrfs_return_cluster_to_free_space(block_group, cluster);
1762 if (need_resched()) {
1763 spin_unlock(&block_group->tree_lock);
1764 cond_resched();
1765 spin_lock(&block_group->tree_lock);
1766 }
1767 }
1768
1769 while ((node = rb_last(&block_group->free_space_offset)) != NULL) {
1770 info = rb_entry(node, struct btrfs_free_space, offset_index);
1771 unlink_free_space(block_group, info);
1772 if (info->bitmap)
1773 kfree(info->bitmap);
1774 kmem_cache_free(btrfs_free_space_cachep, info);
1775 if (need_resched()) {
1776 spin_unlock(&block_group->tree_lock);
1777 cond_resched();
1778 spin_lock(&block_group->tree_lock);
1779 }
1780 }
1781
1782 spin_unlock(&block_group->tree_lock);
1783 }
1784
1785 u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group,
1786 u64 offset, u64 bytes, u64 empty_size)
1787 {
1788 struct btrfs_free_space *entry = NULL;
1789 u64 bytes_search = bytes + empty_size;
1790 u64 ret = 0;
1791
1792 spin_lock(&block_group->tree_lock);
1793 entry = find_free_space(block_group, &offset, &bytes_search, 0);
1794 if (!entry)
1795 goto out;
1796
1797 ret = offset;
1798 if (entry->bitmap) {
1799 bitmap_clear_bits(block_group, entry, offset, bytes);
1800 if (!entry->bytes)
1801 free_bitmap(block_group, entry);
1802 } else {
1803 unlink_free_space(block_group, entry);
1804 entry->offset += bytes;
1805 entry->bytes -= bytes;
1806 if (!entry->bytes)
1807 kmem_cache_free(btrfs_free_space_cachep, entry);
1808 else
1809 link_free_space(block_group, entry);
1810 }
1811
1812 out:
1813 spin_unlock(&block_group->tree_lock);
1814
1815 return ret;
1816 }
1817
1818 /*
1819 * given a cluster, put all of its extents back into the free space
1820 * cache. If a block group is passed, this function will only free
1821 * a cluster that belongs to the passed block group.
1822 *
1823 * Otherwise, it'll get a reference on the block group pointed to by the
1824 * cluster and remove the cluster from it.
1825 */
1826 int btrfs_return_cluster_to_free_space(
1827 struct btrfs_block_group_cache *block_group,
1828 struct btrfs_free_cluster *cluster)
1829 {
1830 int ret;
1831
1832 /* first, get a safe pointer to the block group */
1833 spin_lock(&cluster->lock);
1834 if (!block_group) {
1835 block_group = cluster->block_group;
1836 if (!block_group) {
1837 spin_unlock(&cluster->lock);
1838 return 0;
1839 }
1840 } else if (cluster->block_group != block_group) {
1841 /* someone else has already freed it don't redo their work */
1842 spin_unlock(&cluster->lock);
1843 return 0;
1844 }
1845 atomic_inc(&block_group->count);
1846 spin_unlock(&cluster->lock);
1847
1848 /* now return any extents the cluster had on it */
1849 spin_lock(&block_group->tree_lock);
1850 ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
1851 spin_unlock(&block_group->tree_lock);
1852
1853 /* finally drop our ref */
1854 btrfs_put_block_group(block_group);
1855 return ret;
1856 }
1857
1858 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
1859 struct btrfs_free_cluster *cluster,
1860 struct btrfs_free_space *entry,
1861 u64 bytes, u64 min_start)
1862 {
1863 int err;
1864 u64 search_start = cluster->window_start;
1865 u64 search_bytes = bytes;
1866 u64 ret = 0;
1867
1868 search_start = min_start;
1869 search_bytes = bytes;
1870
1871 err = search_bitmap(block_group, entry, &search_start,
1872 &search_bytes);
1873 if (err)
1874 return 0;
1875
1876 ret = search_start;
1877 bitmap_clear_bits(block_group, entry, ret, bytes);
1878
1879 return ret;
1880 }
1881
1882 /*
1883 * given a cluster, try to allocate 'bytes' from it, returns 0
1884 * if it couldn't find anything suitably large, or a logical disk offset
1885 * if things worked out
1886 */
1887 u64 btrfs_alloc_from_cluster(struct btrfs_block_group_cache *block_group,
1888 struct btrfs_free_cluster *cluster, u64 bytes,
1889 u64 min_start)
1890 {
1891 struct btrfs_free_space *entry = NULL;
1892 struct rb_node *node;
1893 u64 ret = 0;
1894
1895 spin_lock(&cluster->lock);
1896 if (bytes > cluster->max_size)
1897 goto out;
1898
1899 if (cluster->block_group != block_group)
1900 goto out;
1901
1902 node = rb_first(&cluster->root);
1903 if (!node)
1904 goto out;
1905
1906 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1907 while(1) {
1908 if (entry->bytes < bytes ||
1909 (!entry->bitmap && entry->offset < min_start)) {
1910 struct rb_node *node;
1911
1912 node = rb_next(&entry->offset_index);
1913 if (!node)
1914 break;
1915 entry = rb_entry(node, struct btrfs_free_space,
1916 offset_index);
1917 continue;
1918 }
1919
1920 if (entry->bitmap) {
1921 ret = btrfs_alloc_from_bitmap(block_group,
1922 cluster, entry, bytes,
1923 min_start);
1924 if (ret == 0) {
1925 struct rb_node *node;
1926 node = rb_next(&entry->offset_index);
1927 if (!node)
1928 break;
1929 entry = rb_entry(node, struct btrfs_free_space,
1930 offset_index);
1931 continue;
1932 }
1933 } else {
1934
1935 ret = entry->offset;
1936
1937 entry->offset += bytes;
1938 entry->bytes -= bytes;
1939 }
1940
1941 if (entry->bytes == 0)
1942 rb_erase(&entry->offset_index, &cluster->root);
1943 break;
1944 }
1945 out:
1946 spin_unlock(&cluster->lock);
1947
1948 if (!ret)
1949 return 0;
1950
1951 spin_lock(&block_group->tree_lock);
1952
1953 block_group->free_space -= bytes;
1954 if (entry->bytes == 0) {
1955 block_group->free_extents--;
1956 if (entry->bitmap) {
1957 kfree(entry->bitmap);
1958 block_group->total_bitmaps--;
1959 recalculate_thresholds(block_group);
1960 }
1961 kmem_cache_free(btrfs_free_space_cachep, entry);
1962 }
1963
1964 spin_unlock(&block_group->tree_lock);
1965
1966 return ret;
1967 }
1968
1969 static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
1970 struct btrfs_free_space *entry,
1971 struct btrfs_free_cluster *cluster,
1972 u64 offset, u64 bytes, u64 min_bytes)
1973 {
1974 unsigned long next_zero;
1975 unsigned long i;
1976 unsigned long search_bits;
1977 unsigned long total_bits;
1978 unsigned long found_bits;
1979 unsigned long start = 0;
1980 unsigned long total_found = 0;
1981 int ret;
1982 bool found = false;
1983
1984 i = offset_to_bit(entry->offset, block_group->sectorsize,
1985 max_t(u64, offset, entry->offset));
1986 search_bits = bytes_to_bits(bytes, block_group->sectorsize);
1987 total_bits = bytes_to_bits(min_bytes, block_group->sectorsize);
1988
1989 again:
1990 found_bits = 0;
1991 for (i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, i);
1992 i < BITS_PER_BITMAP;
1993 i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, i + 1)) {
1994 next_zero = find_next_zero_bit(entry->bitmap,
1995 BITS_PER_BITMAP, i);
1996 if (next_zero - i >= search_bits) {
1997 found_bits = next_zero - i;
1998 break;
1999 }
2000 i = next_zero;
2001 }
2002
2003 if (!found_bits)
2004 return -ENOSPC;
2005
2006 if (!found) {
2007 start = i;
2008 found = true;
2009 }
2010
2011 total_found += found_bits;
2012
2013 if (cluster->max_size < found_bits * block_group->sectorsize)
2014 cluster->max_size = found_bits * block_group->sectorsize;
2015
2016 if (total_found < total_bits) {
2017 i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, next_zero);
2018 if (i - start > total_bits * 2) {
2019 total_found = 0;
2020 cluster->max_size = 0;
2021 found = false;
2022 }
2023 goto again;
2024 }
2025
2026 cluster->window_start = start * block_group->sectorsize +
2027 entry->offset;
2028 rb_erase(&entry->offset_index, &block_group->free_space_offset);
2029 ret = tree_insert_offset(&cluster->root, entry->offset,
2030 &entry->offset_index, 1);
2031 BUG_ON(ret);
2032
2033 return 0;
2034 }
2035
2036 /*
2037 * This searches the block group for just extents to fill the cluster with.
2038 */
2039 static int setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2040 struct btrfs_free_cluster *cluster,
2041 u64 offset, u64 bytes, u64 min_bytes)
2042 {
2043 struct btrfs_free_space *first = NULL;
2044 struct btrfs_free_space *entry = NULL;
2045 struct btrfs_free_space *prev = NULL;
2046 struct btrfs_free_space *last;
2047 struct rb_node *node;
2048 u64 window_start;
2049 u64 window_free;
2050 u64 max_extent;
2051 u64 max_gap = 128 * 1024;
2052
2053 entry = tree_search_offset(block_group, offset, 0, 1);
2054 if (!entry)
2055 return -ENOSPC;
2056
2057 /*
2058 * We don't want bitmaps, so just move along until we find a normal
2059 * extent entry.
2060 */
2061 while (entry->bitmap) {
2062 node = rb_next(&entry->offset_index);
2063 if (!node)
2064 return -ENOSPC;
2065 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2066 }
2067
2068 window_start = entry->offset;
2069 window_free = entry->bytes;
2070 max_extent = entry->bytes;
2071 first = entry;
2072 last = entry;
2073 prev = entry;
2074
2075 while (window_free <= min_bytes) {
2076 node = rb_next(&entry->offset_index);
2077 if (!node)
2078 return -ENOSPC;
2079 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2080
2081 if (entry->bitmap)
2082 continue;
2083 /*
2084 * we haven't filled the empty size and the window is
2085 * very large. reset and try again
2086 */
2087 if (entry->offset - (prev->offset + prev->bytes) > max_gap ||
2088 entry->offset - window_start > (min_bytes * 2)) {
2089 first = entry;
2090 window_start = entry->offset;
2091 window_free = entry->bytes;
2092 last = entry;
2093 max_extent = entry->bytes;
2094 } else {
2095 last = entry;
2096 window_free += entry->bytes;
2097 if (entry->bytes > max_extent)
2098 max_extent = entry->bytes;
2099 }
2100 prev = entry;
2101 }
2102
2103 cluster->window_start = first->offset;
2104
2105 node = &first->offset_index;
2106
2107 /*
2108 * now we've found our entries, pull them out of the free space
2109 * cache and put them into the cluster rbtree
2110 */
2111 do {
2112 int ret;
2113
2114 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2115 node = rb_next(&entry->offset_index);
2116 if (entry->bitmap)
2117 continue;
2118
2119 rb_erase(&entry->offset_index, &block_group->free_space_offset);
2120 ret = tree_insert_offset(&cluster->root, entry->offset,
2121 &entry->offset_index, 0);
2122 BUG_ON(ret);
2123 } while (node && entry != last);
2124
2125 cluster->max_size = max_extent;
2126
2127 return 0;
2128 }
2129
2130 /*
2131 * This specifically looks for bitmaps that may work in the cluster, we assume
2132 * that we have already failed to find extents that will work.
2133 */
2134 static int setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2135 struct btrfs_free_cluster *cluster,
2136 u64 offset, u64 bytes, u64 min_bytes)
2137 {
2138 struct btrfs_free_space *entry;
2139 struct rb_node *node;
2140 int ret = -ENOSPC;
2141
2142 if (block_group->total_bitmaps == 0)
2143 return -ENOSPC;
2144
2145 entry = tree_search_offset(block_group,
2146 offset_to_bitmap(block_group, offset),
2147 0, 1);
2148 if (!entry)
2149 return -ENOSPC;
2150
2151 node = &entry->offset_index;
2152 do {
2153 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2154 node = rb_next(&entry->offset_index);
2155 if (!entry->bitmap)
2156 continue;
2157 if (entry->bytes < min_bytes)
2158 continue;
2159 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
2160 bytes, min_bytes);
2161 } while (ret && node);
2162
2163 return ret;
2164 }
2165
2166 /*
2167 * here we try to find a cluster of blocks in a block group. The goal
2168 * is to find at least bytes free and up to empty_size + bytes free.
2169 * We might not find them all in one contiguous area.
2170 *
2171 * returns zero and sets up cluster if things worked out, otherwise
2172 * it returns -enospc
2173 */
2174 int btrfs_find_space_cluster(struct btrfs_trans_handle *trans,
2175 struct btrfs_root *root,
2176 struct btrfs_block_group_cache *block_group,
2177 struct btrfs_free_cluster *cluster,
2178 u64 offset, u64 bytes, u64 empty_size)
2179 {
2180 u64 min_bytes;
2181 int ret;
2182
2183 /* for metadata, allow allocates with more holes */
2184 if (btrfs_test_opt(root, SSD_SPREAD)) {
2185 min_bytes = bytes + empty_size;
2186 } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
2187 /*
2188 * we want to do larger allocations when we are
2189 * flushing out the delayed refs, it helps prevent
2190 * making more work as we go along.
2191 */
2192 if (trans->transaction->delayed_refs.flushing)
2193 min_bytes = max(bytes, (bytes + empty_size) >> 1);
2194 else
2195 min_bytes = max(bytes, (bytes + empty_size) >> 4);
2196 } else
2197 min_bytes = max(bytes, (bytes + empty_size) >> 2);
2198
2199 spin_lock(&block_group->tree_lock);
2200
2201 /*
2202 * If we know we don't have enough space to make a cluster don't even
2203 * bother doing all the work to try and find one.
2204 */
2205 if (block_group->free_space < min_bytes) {
2206 spin_unlock(&block_group->tree_lock);
2207 return -ENOSPC;
2208 }
2209
2210 spin_lock(&cluster->lock);
2211
2212 /* someone already found a cluster, hooray */
2213 if (cluster->block_group) {
2214 ret = 0;
2215 goto out;
2216 }
2217
2218 ret = setup_cluster_no_bitmap(block_group, cluster, offset, bytes,
2219 min_bytes);
2220 if (ret)
2221 ret = setup_cluster_bitmap(block_group, cluster, offset,
2222 bytes, min_bytes);
2223
2224 if (!ret) {
2225 atomic_inc(&block_group->count);
2226 list_add_tail(&cluster->block_group_list,
2227 &block_group->cluster_list);
2228 cluster->block_group = block_group;
2229 }
2230 out:
2231 spin_unlock(&cluster->lock);
2232 spin_unlock(&block_group->tree_lock);
2233
2234 return ret;
2235 }
2236
2237 /*
2238 * simple code to zero out a cluster
2239 */
2240 void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
2241 {
2242 spin_lock_init(&cluster->lock);
2243 spin_lock_init(&cluster->refill_lock);
2244 cluster->root = RB_ROOT;
2245 cluster->max_size = 0;
2246 INIT_LIST_HEAD(&cluster->block_group_list);
2247 cluster->block_group = NULL;
2248 }
2249
2250 int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
2251 u64 *trimmed, u64 start, u64 end, u64 minlen)
2252 {
2253 struct btrfs_free_space *entry = NULL;
2254 struct btrfs_fs_info *fs_info = block_group->fs_info;
2255 u64 bytes = 0;
2256 u64 actually_trimmed;
2257 int ret = 0;
2258
2259 *trimmed = 0;
2260
2261 while (start < end) {
2262 spin_lock(&block_group->tree_lock);
2263
2264 if (block_group->free_space < minlen) {
2265 spin_unlock(&block_group->tree_lock);
2266 break;
2267 }
2268
2269 entry = tree_search_offset(block_group, start, 0, 1);
2270 if (!entry)
2271 entry = tree_search_offset(block_group,
2272 offset_to_bitmap(block_group,
2273 start),
2274 1, 1);
2275
2276 if (!entry || entry->offset >= end) {
2277 spin_unlock(&block_group->tree_lock);
2278 break;
2279 }
2280
2281 if (entry->bitmap) {
2282 ret = search_bitmap(block_group, entry, &start, &bytes);
2283 if (!ret) {
2284 if (start >= end) {
2285 spin_unlock(&block_group->tree_lock);
2286 break;
2287 }
2288 bytes = min(bytes, end - start);
2289 bitmap_clear_bits(block_group, entry,
2290 start, bytes);
2291 if (entry->bytes == 0)
2292 free_bitmap(block_group, entry);
2293 } else {
2294 start = entry->offset + BITS_PER_BITMAP *
2295 block_group->sectorsize;
2296 spin_unlock(&block_group->tree_lock);
2297 ret = 0;
2298 continue;
2299 }
2300 } else {
2301 start = entry->offset;
2302 bytes = min(entry->bytes, end - start);
2303 unlink_free_space(block_group, entry);
2304 kfree(entry);
2305 }
2306
2307 spin_unlock(&block_group->tree_lock);
2308
2309 if (bytes >= minlen) {
2310 int update_ret;
2311 update_ret = btrfs_update_reserved_bytes(block_group,
2312 bytes, 1, 1);
2313
2314 ret = btrfs_error_discard_extent(fs_info->extent_root,
2315 start,
2316 bytes,
2317 &actually_trimmed);
2318
2319 btrfs_add_free_space(block_group,
2320 start, bytes);
2321 if (!update_ret)
2322 btrfs_update_reserved_bytes(block_group,
2323 bytes, 0, 1);
2324
2325 if (ret)
2326 break;
2327 *trimmed += actually_trimmed;
2328 }
2329 start += bytes;
2330 bytes = 0;
2331
2332 if (fatal_signal_pending(current)) {
2333 ret = -ERESTARTSYS;
2334 break;
2335 }
2336
2337 cond_resched();
2338 }
2339
2340 return ret;
2341 }
linux-next