?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge 4.9-rc5 into char-misc-next
[linux-2.6/btrfs-unstable.git] / fs / btrfs / free-space-cache.c
blobe4b48f377d3a9110542872c56ddbb3933e6d4a75
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 <linux/ratelimit.h>
24 #include "ctree.h"
25 #include "free-space-cache.h"
26 #include "transaction.h"
27 #include "disk-io.h"
28 #include "extent_io.h"
29 #include "inode-map.h"
30 #include "volumes.h"
31
32 #define BITS_PER_BITMAP (PAGE_SIZE * 8UL)
33 #define MAX_CACHE_BYTES_PER_GIG SZ_32K
34
35 struct btrfs_trim_range {
36 u64 start;
37 u64 bytes;
38 struct list_head list;
39 };
40
41 static int link_free_space(struct btrfs_free_space_ctl *ctl,
42 struct btrfs_free_space *info);
43 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
44 struct btrfs_free_space *info);
45
46 static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
47 struct btrfs_path *path,
48 u64 offset)
49 {
50 struct btrfs_key key;
51 struct btrfs_key location;
52 struct btrfs_disk_key disk_key;
53 struct btrfs_free_space_header *header;
54 struct extent_buffer *leaf;
55 struct inode *inode = NULL;
56 int ret;
57
58 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
59 key.offset = offset;
60 key.type = 0;
61
62 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
63 if (ret < 0)
64 return ERR_PTR(ret);
65 if (ret > 0) {
66 btrfs_release_path(path);
67 return ERR_PTR(-ENOENT);
68 }
69
70 leaf = path->nodes[0];
71 header = btrfs_item_ptr(leaf, path->slots[0],
72 struct btrfs_free_space_header);
73 btrfs_free_space_key(leaf, header, &disk_key);
74 btrfs_disk_key_to_cpu(&location, &disk_key);
75 btrfs_release_path(path);
76
77 inode = btrfs_iget(root->fs_info->sb, &location, root, NULL);
78 if (!inode)
79 return ERR_PTR(-ENOENT);
80 if (IS_ERR(inode))
81 return inode;
82 if (is_bad_inode(inode)) {
83 iput(inode);
84 return ERR_PTR(-ENOENT);
85 }
86
87 mapping_set_gfp_mask(inode->i_mapping,
88 mapping_gfp_constraint(inode->i_mapping,
89 ~(__GFP_FS | __GFP_HIGHMEM)));
90
91 return inode;
92 }
93
94 struct inode *lookup_free_space_inode(struct btrfs_root *root,
95 struct btrfs_block_group_cache
96 *block_group, struct btrfs_path *path)
97 {
98 struct inode *inode = NULL;
99 u32 flags = BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
100
101 spin_lock(&block_group->lock);
102 if (block_group->inode)
103 inode = igrab(block_group->inode);
104 spin_unlock(&block_group->lock);
105 if (inode)
106 return inode;
107
108 inode = __lookup_free_space_inode(root, path,
109 block_group->key.objectid);
110 if (IS_ERR(inode))
111 return inode;
112
113 spin_lock(&block_group->lock);
114 if (!((BTRFS_I(inode)->flags & flags) == flags)) {
115 btrfs_info(root->fs_info,
116 "Old style space inode found, converting.");
117 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM |
118 BTRFS_INODE_NODATACOW;
119 block_group->disk_cache_state = BTRFS_DC_CLEAR;
120 }
121
122 if (!block_group->iref) {
123 block_group->inode = igrab(inode);
124 block_group->iref = 1;
125 }
126 spin_unlock(&block_group->lock);
127
128 return inode;
129 }
130
131 static int __create_free_space_inode(struct btrfs_root *root,
132 struct btrfs_trans_handle *trans,
133 struct btrfs_path *path,
134 u64 ino, u64 offset)
135 {
136 struct btrfs_key key;
137 struct btrfs_disk_key disk_key;
138 struct btrfs_free_space_header *header;
139 struct btrfs_inode_item *inode_item;
140 struct extent_buffer *leaf;
141 u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC;
142 int ret;
143
144 ret = btrfs_insert_empty_inode(trans, root, path, ino);
145 if (ret)
146 return ret;
147
148 /* We inline crc's for the free disk space cache */
149 if (ino != BTRFS_FREE_INO_OBJECTID)
150 flags |= BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
151
152 leaf = path->nodes[0];
153 inode_item = btrfs_item_ptr(leaf, path->slots[0],
154 struct btrfs_inode_item);
155 btrfs_item_key(leaf, &disk_key, path->slots[0]);
156 memset_extent_buffer(leaf, 0, (unsigned long)inode_item,
157 sizeof(*inode_item));
158 btrfs_set_inode_generation(leaf, inode_item, trans->transid);
159 btrfs_set_inode_size(leaf, inode_item, 0);
160 btrfs_set_inode_nbytes(leaf, inode_item, 0);
161 btrfs_set_inode_uid(leaf, inode_item, 0);
162 btrfs_set_inode_gid(leaf, inode_item, 0);
163 btrfs_set_inode_mode(leaf, inode_item, S_IFREG | 0600);
164 btrfs_set_inode_flags(leaf, inode_item, flags);
165 btrfs_set_inode_nlink(leaf, inode_item, 1);
166 btrfs_set_inode_transid(leaf, inode_item, trans->transid);
167 btrfs_set_inode_block_group(leaf, inode_item, offset);
168 btrfs_mark_buffer_dirty(leaf);
169 btrfs_release_path(path);
170
171 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
172 key.offset = offset;
173 key.type = 0;
174 ret = btrfs_insert_empty_item(trans, root, path, &key,
175 sizeof(struct btrfs_free_space_header));
176 if (ret < 0) {
177 btrfs_release_path(path);
178 return ret;
179 }
180
181 leaf = path->nodes[0];
182 header = btrfs_item_ptr(leaf, path->slots[0],
183 struct btrfs_free_space_header);
184 memset_extent_buffer(leaf, 0, (unsigned long)header, sizeof(*header));
185 btrfs_set_free_space_key(leaf, header, &disk_key);
186 btrfs_mark_buffer_dirty(leaf);
187 btrfs_release_path(path);
188
189 return 0;
190 }
191
192 int create_free_space_inode(struct btrfs_root *root,
193 struct btrfs_trans_handle *trans,
194 struct btrfs_block_group_cache *block_group,
195 struct btrfs_path *path)
196 {
197 int ret;
198 u64 ino;
199
200 ret = btrfs_find_free_objectid(root, &ino);
201 if (ret < 0)
202 return ret;
203
204 return __create_free_space_inode(root, trans, path, ino,
205 block_group->key.objectid);
206 }
207
208 int btrfs_check_trunc_cache_free_space(struct btrfs_root *root,
209 struct btrfs_block_rsv *rsv)
210 {
211 u64 needed_bytes;
212 int ret;
213
214 /* 1 for slack space, 1 for updating the inode */
215 needed_bytes = btrfs_calc_trunc_metadata_size(root, 1) +
216 btrfs_calc_trans_metadata_size(root, 1);
217
218 spin_lock(&rsv->lock);
219 if (rsv->reserved < needed_bytes)
220 ret = -ENOSPC;
221 else
222 ret = 0;
223 spin_unlock(&rsv->lock);
224 return ret;
225 }
226
227 int btrfs_truncate_free_space_cache(struct btrfs_root *root,
228 struct btrfs_trans_handle *trans,
229 struct btrfs_block_group_cache *block_group,
230 struct inode *inode)
231 {
232 int ret = 0;
233 struct btrfs_path *path = btrfs_alloc_path();
234 bool locked = false;
235
236 if (!path) {
237 ret = -ENOMEM;
238 goto fail;
239 }
240
241 if (block_group) {
242 locked = true;
243 mutex_lock(&trans->transaction->cache_write_mutex);
244 if (!list_empty(&block_group->io_list)) {
245 list_del_init(&block_group->io_list);
246
247 btrfs_wait_cache_io(root, trans, block_group,
248 &block_group->io_ctl, path,
249 block_group->key.objectid);
250 btrfs_put_block_group(block_group);
251 }
252
253 /*
254 * now that we've truncated the cache away, its no longer
255 * setup or written
256 */
257 spin_lock(&block_group->lock);
258 block_group->disk_cache_state = BTRFS_DC_CLEAR;
259 spin_unlock(&block_group->lock);
260 }
261 btrfs_free_path(path);
262
263 btrfs_i_size_write(inode, 0);
264 truncate_pagecache(inode, 0);
265
266 /*
267 * We don't need an orphan item because truncating the free space cache
268 * will never be split across transactions.
269 * We don't need to check for -EAGAIN because we're a free space
270 * cache inode
271 */
272 ret = btrfs_truncate_inode_items(trans, root, inode,
273 0, BTRFS_EXTENT_DATA_KEY);
274 if (ret)
275 goto fail;
276
277 ret = btrfs_update_inode(trans, root, inode);
278
279 fail:
280 if (locked)
281 mutex_unlock(&trans->transaction->cache_write_mutex);
282 if (ret)
283 btrfs_abort_transaction(trans, ret);
284
285 return ret;
286 }
287
288 static int readahead_cache(struct inode *inode)
289 {
290 struct file_ra_state *ra;
291 unsigned long last_index;
292
293 ra = kzalloc(sizeof(*ra), GFP_NOFS);
294 if (!ra)
295 return -ENOMEM;
296
297 file_ra_state_init(ra, inode->i_mapping);
298 last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
299
300 page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
301
302 kfree(ra);
303
304 return 0;
305 }
306
307 static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
308 struct btrfs_root *root, int write)
309 {
310 int num_pages;
311 int check_crcs = 0;
312
313 num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
314
315 if (btrfs_ino(inode) != BTRFS_FREE_INO_OBJECTID)
316 check_crcs = 1;
317
318 /* Make sure we can fit our crcs into the first page */
319 if (write && check_crcs &&
320 (num_pages * sizeof(u32)) >= PAGE_SIZE)
321 return -ENOSPC;
322
323 memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
324
325 io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
326 if (!io_ctl->pages)
327 return -ENOMEM;
328
329 io_ctl->num_pages = num_pages;
330 io_ctl->root = root;
331 io_ctl->check_crcs = check_crcs;
332 io_ctl->inode = inode;
333
334 return 0;
335 }
336
337 static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
338 {
339 kfree(io_ctl->pages);
340 io_ctl->pages = NULL;
341 }
342
343 static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
344 {
345 if (io_ctl->cur) {
346 io_ctl->cur = NULL;
347 io_ctl->orig = NULL;
348 }
349 }
350
351 static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
352 {
353 ASSERT(io_ctl->index < io_ctl->num_pages);
354 io_ctl->page = io_ctl->pages[io_ctl->index++];
355 io_ctl->cur = page_address(io_ctl->page);
356 io_ctl->orig = io_ctl->cur;
357 io_ctl->size = PAGE_SIZE;
358 if (clear)
359 memset(io_ctl->cur, 0, PAGE_SIZE);
360 }
361
362 static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
363 {
364 int i;
365
366 io_ctl_unmap_page(io_ctl);
367
368 for (i = 0; i < io_ctl->num_pages; i++) {
369 if (io_ctl->pages[i]) {
370 ClearPageChecked(io_ctl->pages[i]);
371 unlock_page(io_ctl->pages[i]);
372 put_page(io_ctl->pages[i]);
373 }
374 }
375 }
376
377 static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, struct inode *inode,
378 int uptodate)
379 {
380 struct page *page;
381 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
382 int i;
383
384 for (i = 0; i < io_ctl->num_pages; i++) {
385 page = find_or_create_page(inode->i_mapping, i, mask);
386 if (!page) {
387 io_ctl_drop_pages(io_ctl);
388 return -ENOMEM;
389 }
390 io_ctl->pages[i] = page;
391 if (uptodate && !PageUptodate(page)) {
392 btrfs_readpage(NULL, page);
393 lock_page(page);
394 if (!PageUptodate(page)) {
395 btrfs_err(BTRFS_I(inode)->root->fs_info,
396 "error reading free space cache");
397 io_ctl_drop_pages(io_ctl);
398 return -EIO;
399 }
400 }
401 }
402
403 for (i = 0; i < io_ctl->num_pages; i++) {
404 clear_page_dirty_for_io(io_ctl->pages[i]);
405 set_page_extent_mapped(io_ctl->pages[i]);
406 }
407
408 return 0;
409 }
410
411 static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
412 {
413 __le64 *val;
414
415 io_ctl_map_page(io_ctl, 1);
416
417 /*
418 * Skip the csum areas. If we don't check crcs then we just have a
419 * 64bit chunk at the front of the first page.
420 */
421 if (io_ctl->check_crcs) {
422 io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
423 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
424 } else {
425 io_ctl->cur += sizeof(u64);
426 io_ctl->size -= sizeof(u64) * 2;
427 }
428
429 val = io_ctl->cur;
430 *val = cpu_to_le64(generation);
431 io_ctl->cur += sizeof(u64);
432 }
433
434 static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
435 {
436 __le64 *gen;
437
438 /*
439 * Skip the crc area. If we don't check crcs then we just have a 64bit
440 * chunk at the front of the first page.
441 */
442 if (io_ctl->check_crcs) {
443 io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
444 io_ctl->size -= sizeof(u64) +
445 (sizeof(u32) * io_ctl->num_pages);
446 } else {
447 io_ctl->cur += sizeof(u64);
448 io_ctl->size -= sizeof(u64) * 2;
449 }
450
451 gen = io_ctl->cur;
452 if (le64_to_cpu(*gen) != generation) {
453 btrfs_err_rl(io_ctl->root->fs_info,
454 "space cache generation (%llu) does not match inode (%llu)",
455 *gen, generation);
456 io_ctl_unmap_page(io_ctl);
457 return -EIO;
458 }
459 io_ctl->cur += sizeof(u64);
460 return 0;
461 }
462
463 static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
464 {
465 u32 *tmp;
466 u32 crc = ~(u32)0;
467 unsigned offset = 0;
468
469 if (!io_ctl->check_crcs) {
470 io_ctl_unmap_page(io_ctl);
471 return;
472 }
473
474 if (index == 0)
475 offset = sizeof(u32) * io_ctl->num_pages;
476
477 crc = btrfs_csum_data(io_ctl->orig + offset, crc,
478 PAGE_SIZE - offset);
479 btrfs_csum_final(crc, (char *)&crc);
480 io_ctl_unmap_page(io_ctl);
481 tmp = page_address(io_ctl->pages[0]);
482 tmp += index;
483 *tmp = crc;
484 }
485
486 static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
487 {
488 u32 *tmp, val;
489 u32 crc = ~(u32)0;
490 unsigned offset = 0;
491
492 if (!io_ctl->check_crcs) {
493 io_ctl_map_page(io_ctl, 0);
494 return 0;
495 }
496
497 if (index == 0)
498 offset = sizeof(u32) * io_ctl->num_pages;
499
500 tmp = page_address(io_ctl->pages[0]);
501 tmp += index;
502 val = *tmp;
503
504 io_ctl_map_page(io_ctl, 0);
505 crc = btrfs_csum_data(io_ctl->orig + offset, crc,
506 PAGE_SIZE - offset);
507 btrfs_csum_final(crc, (char *)&crc);
508 if (val != crc) {
509 btrfs_err_rl(io_ctl->root->fs_info,
510 "csum mismatch on free space cache");
511 io_ctl_unmap_page(io_ctl);
512 return -EIO;
513 }
514
515 return 0;
516 }
517
518 static int io_ctl_add_entry(struct btrfs_io_ctl *io_ctl, u64 offset, u64 bytes,
519 void *bitmap)
520 {
521 struct btrfs_free_space_entry *entry;
522
523 if (!io_ctl->cur)
524 return -ENOSPC;
525
526 entry = io_ctl->cur;
527 entry->offset = cpu_to_le64(offset);
528 entry->bytes = cpu_to_le64(bytes);
529 entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
530 BTRFS_FREE_SPACE_EXTENT;
531 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
532 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
533
534 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
535 return 0;
536
537 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
538
539 /* No more pages to map */
540 if (io_ctl->index >= io_ctl->num_pages)
541 return 0;
542
543 /* map the next page */
544 io_ctl_map_page(io_ctl, 1);
545 return 0;
546 }
547
548 static int io_ctl_add_bitmap(struct btrfs_io_ctl *io_ctl, void *bitmap)
549 {
550 if (!io_ctl->cur)
551 return -ENOSPC;
552
553 /*
554 * If we aren't at the start of the current page, unmap this one and
555 * map the next one if there is any left.
556 */
557 if (io_ctl->cur != io_ctl->orig) {
558 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
559 if (io_ctl->index >= io_ctl->num_pages)
560 return -ENOSPC;
561 io_ctl_map_page(io_ctl, 0);
562 }
563
564 memcpy(io_ctl->cur, bitmap, PAGE_SIZE);
565 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
566 if (io_ctl->index < io_ctl->num_pages)
567 io_ctl_map_page(io_ctl, 0);
568 return 0;
569 }
570
571 static void io_ctl_zero_remaining_pages(struct btrfs_io_ctl *io_ctl)
572 {
573 /*
574 * If we're not on the boundary we know we've modified the page and we
575 * need to crc the page.
576 */
577 if (io_ctl->cur != io_ctl->orig)
578 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
579 else
580 io_ctl_unmap_page(io_ctl);
581
582 while (io_ctl->index < io_ctl->num_pages) {
583 io_ctl_map_page(io_ctl, 1);
584 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
585 }
586 }
587
588 static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
589 struct btrfs_free_space *entry, u8 *type)
590 {
591 struct btrfs_free_space_entry *e;
592 int ret;
593
594 if (!io_ctl->cur) {
595 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
596 if (ret)
597 return ret;
598 }
599
600 e = io_ctl->cur;
601 entry->offset = le64_to_cpu(e->offset);
602 entry->bytes = le64_to_cpu(e->bytes);
603 *type = e->type;
604 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
605 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
606
607 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
608 return 0;
609
610 io_ctl_unmap_page(io_ctl);
611
612 return 0;
613 }
614
615 static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
616 struct btrfs_free_space *entry)
617 {
618 int ret;
619
620 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
621 if (ret)
622 return ret;
623
624 memcpy(entry->bitmap, io_ctl->cur, PAGE_SIZE);
625 io_ctl_unmap_page(io_ctl);
626
627 return 0;
628 }
629
630 /*
631 * Since we attach pinned extents after the fact we can have contiguous sections
632 * of free space that are split up in entries. This poses a problem with the
633 * tree logging stuff since it could have allocated across what appears to be 2
634 * entries since we would have merged the entries when adding the pinned extents
635 * back to the free space cache. So run through the space cache that we just
636 * loaded and merge contiguous entries. This will make the log replay stuff not
637 * blow up and it will make for nicer allocator behavior.
638 */
639 static void merge_space_tree(struct btrfs_free_space_ctl *ctl)
640 {
641 struct btrfs_free_space *e, *prev = NULL;
642 struct rb_node *n;
643
644 again:
645 spin_lock(&ctl->tree_lock);
646 for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
647 e = rb_entry(n, struct btrfs_free_space, offset_index);
648 if (!prev)
649 goto next;
650 if (e->bitmap || prev->bitmap)
651 goto next;
652 if (prev->offset + prev->bytes == e->offset) {
653 unlink_free_space(ctl, prev);
654 unlink_free_space(ctl, e);
655 prev->bytes += e->bytes;
656 kmem_cache_free(btrfs_free_space_cachep, e);
657 link_free_space(ctl, prev);
658 prev = NULL;
659 spin_unlock(&ctl->tree_lock);
660 goto again;
661 }
662 next:
663 prev = e;
664 }
665 spin_unlock(&ctl->tree_lock);
666 }
667
668 static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
669 struct btrfs_free_space_ctl *ctl,
670 struct btrfs_path *path, u64 offset)
671 {
672 struct btrfs_free_space_header *header;
673 struct extent_buffer *leaf;
674 struct btrfs_io_ctl io_ctl;
675 struct btrfs_key key;
676 struct btrfs_free_space *e, *n;
677 LIST_HEAD(bitmaps);
678 u64 num_entries;
679 u64 num_bitmaps;
680 u64 generation;
681 u8 type;
682 int ret = 0;
683
684 /* Nothing in the space cache, goodbye */
685 if (!i_size_read(inode))
686 return 0;
687
688 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
689 key.offset = offset;
690 key.type = 0;
691
692 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
693 if (ret < 0)
694 return 0;
695 else if (ret > 0) {
696 btrfs_release_path(path);
697 return 0;
698 }
699
700 ret = -1;
701
702 leaf = path->nodes[0];
703 header = btrfs_item_ptr(leaf, path->slots[0],
704 struct btrfs_free_space_header);
705 num_entries = btrfs_free_space_entries(leaf, header);
706 num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
707 generation = btrfs_free_space_generation(leaf, header);
708 btrfs_release_path(path);
709
710 if (!BTRFS_I(inode)->generation) {
711 btrfs_info(root->fs_info,
712 "The free space cache file (%llu) is invalid. skip it\n",
713 offset);
714 return 0;
715 }
716
717 if (BTRFS_I(inode)->generation != generation) {
718 btrfs_err(root->fs_info,
719 "free space inode generation (%llu) did not match free space cache generation (%llu)",
720 BTRFS_I(inode)->generation, generation);
721 return 0;
722 }
723
724 if (!num_entries)
725 return 0;
726
727 ret = io_ctl_init(&io_ctl, inode, root, 0);
728 if (ret)
729 return ret;
730
731 ret = readahead_cache(inode);
732 if (ret)
733 goto out;
734
735 ret = io_ctl_prepare_pages(&io_ctl, inode, 1);
736 if (ret)
737 goto out;
738
739 ret = io_ctl_check_crc(&io_ctl, 0);
740 if (ret)
741 goto free_cache;
742
743 ret = io_ctl_check_generation(&io_ctl, generation);
744 if (ret)
745 goto free_cache;
746
747 while (num_entries) {
748 e = kmem_cache_zalloc(btrfs_free_space_cachep,
749 GFP_NOFS);
750 if (!e)
751 goto free_cache;
752
753 ret = io_ctl_read_entry(&io_ctl, e, &type);
754 if (ret) {
755 kmem_cache_free(btrfs_free_space_cachep, e);
756 goto free_cache;
757 }
758
759 if (!e->bytes) {
760 kmem_cache_free(btrfs_free_space_cachep, e);
761 goto free_cache;
762 }
763
764 if (type == BTRFS_FREE_SPACE_EXTENT) {
765 spin_lock(&ctl->tree_lock);
766 ret = link_free_space(ctl, e);
767 spin_unlock(&ctl->tree_lock);
768 if (ret) {
769 btrfs_err(root->fs_info,
770 "Duplicate entries in free space cache, dumping");
771 kmem_cache_free(btrfs_free_space_cachep, e);
772 goto free_cache;
773 }
774 } else {
775 ASSERT(num_bitmaps);
776 num_bitmaps--;
777 e->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
778 if (!e->bitmap) {
779 kmem_cache_free(
780 btrfs_free_space_cachep, e);
781 goto free_cache;
782 }
783 spin_lock(&ctl->tree_lock);
784 ret = link_free_space(ctl, e);
785 ctl->total_bitmaps++;
786 ctl->op->recalc_thresholds(ctl);
787 spin_unlock(&ctl->tree_lock);
788 if (ret) {
789 btrfs_err(root->fs_info,
790 "Duplicate entries in free space cache, dumping");
791 kmem_cache_free(btrfs_free_space_cachep, e);
792 goto free_cache;
793 }
794 list_add_tail(&e->list, &bitmaps);
795 }
796
797 num_entries--;
798 }
799
800 io_ctl_unmap_page(&io_ctl);
801
802 /*
803 * We add the bitmaps at the end of the entries in order that
804 * the bitmap entries are added to the cache.
805 */
806 list_for_each_entry_safe(e, n, &bitmaps, list) {
807 list_del_init(&e->list);
808 ret = io_ctl_read_bitmap(&io_ctl, e);
809 if (ret)
810 goto free_cache;
811 }
812
813 io_ctl_drop_pages(&io_ctl);
814 merge_space_tree(ctl);
815 ret = 1;
816 out:
817 io_ctl_free(&io_ctl);
818 return ret;
819 free_cache:
820 io_ctl_drop_pages(&io_ctl);
821 __btrfs_remove_free_space_cache(ctl);
822 goto out;
823 }
824
825 int load_free_space_cache(struct btrfs_fs_info *fs_info,
826 struct btrfs_block_group_cache *block_group)
827 {
828 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
829 struct btrfs_root *root = fs_info->tree_root;
830 struct inode *inode;
831 struct btrfs_path *path;
832 int ret = 0;
833 bool matched;
834 u64 used = btrfs_block_group_used(&block_group->item);
835
836 /*
837 * If this block group has been marked to be cleared for one reason or
838 * another then we can't trust the on disk cache, so just return.
839 */
840 spin_lock(&block_group->lock);
841 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
842 spin_unlock(&block_group->lock);
843 return 0;
844 }
845 spin_unlock(&block_group->lock);
846
847 path = btrfs_alloc_path();
848 if (!path)
849 return 0;
850 path->search_commit_root = 1;
851 path->skip_locking = 1;
852
853 inode = lookup_free_space_inode(root, block_group, path);
854 if (IS_ERR(inode)) {
855 btrfs_free_path(path);
856 return 0;
857 }
858
859 /* We may have converted the inode and made the cache invalid. */
860 spin_lock(&block_group->lock);
861 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
862 spin_unlock(&block_group->lock);
863 btrfs_free_path(path);
864 goto out;
865 }
866 spin_unlock(&block_group->lock);
867
868 ret = __load_free_space_cache(fs_info->tree_root, inode, ctl,
869 path, block_group->key.objectid);
870 btrfs_free_path(path);
871 if (ret <= 0)
872 goto out;
873
874 spin_lock(&ctl->tree_lock);
875 matched = (ctl->free_space == (block_group->key.offset - used -
876 block_group->bytes_super));
877 spin_unlock(&ctl->tree_lock);
878
879 if (!matched) {
880 __btrfs_remove_free_space_cache(ctl);
881 btrfs_warn(fs_info,
882 "block group %llu has wrong amount of free space",
883 block_group->key.objectid);
884 ret = -1;
885 }
886 out:
887 if (ret < 0) {
888 /* This cache is bogus, make sure it gets cleared */
889 spin_lock(&block_group->lock);
890 block_group->disk_cache_state = BTRFS_DC_CLEAR;
891 spin_unlock(&block_group->lock);
892 ret = 0;
893
894 btrfs_warn(fs_info,
895 "failed to load free space cache for block group %llu, rebuilding it now",
896 block_group->key.objectid);
897 }
898
899 iput(inode);
900 return ret;
901 }
902
903 static noinline_for_stack
904 int write_cache_extent_entries(struct btrfs_io_ctl *io_ctl,
905 struct btrfs_free_space_ctl *ctl,
906 struct btrfs_block_group_cache *block_group,
907 int *entries, int *bitmaps,
908 struct list_head *bitmap_list)
909 {
910 int ret;
911 struct btrfs_free_cluster *cluster = NULL;
912 struct btrfs_free_cluster *cluster_locked = NULL;
913 struct rb_node *node = rb_first(&ctl->free_space_offset);
914 struct btrfs_trim_range *trim_entry;
915
916 /* Get the cluster for this block_group if it exists */
917 if (block_group && !list_empty(&block_group->cluster_list)) {
918 cluster = list_entry(block_group->cluster_list.next,
919 struct btrfs_free_cluster,
920 block_group_list);
921 }
922
923 if (!node && cluster) {
924 cluster_locked = cluster;
925 spin_lock(&cluster_locked->lock);
926 node = rb_first(&cluster->root);
927 cluster = NULL;
928 }
929
930 /* Write out the extent entries */
931 while (node) {
932 struct btrfs_free_space *e;
933
934 e = rb_entry(node, struct btrfs_free_space, offset_index);
935 *entries += 1;
936
937 ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes,
938 e->bitmap);
939 if (ret)
940 goto fail;
941
942 if (e->bitmap) {
943 list_add_tail(&e->list, bitmap_list);
944 *bitmaps += 1;
945 }
946 node = rb_next(node);
947 if (!node && cluster) {
948 node = rb_first(&cluster->root);
949 cluster_locked = cluster;
950 spin_lock(&cluster_locked->lock);
951 cluster = NULL;
952 }
953 }
954 if (cluster_locked) {
955 spin_unlock(&cluster_locked->lock);
956 cluster_locked = NULL;
957 }
958
959 /*
960 * Make sure we don't miss any range that was removed from our rbtree
961 * because trimming is running. Otherwise after a umount+mount (or crash
962 * after committing the transaction) we would leak free space and get
963 * an inconsistent free space cache report from fsck.
964 */
965 list_for_each_entry(trim_entry, &ctl->trimming_ranges, list) {
966 ret = io_ctl_add_entry(io_ctl, trim_entry->start,
967 trim_entry->bytes, NULL);
968 if (ret)
969 goto fail;
970 *entries += 1;
971 }
972
973 return 0;
974 fail:
975 if (cluster_locked)
976 spin_unlock(&cluster_locked->lock);
977 return -ENOSPC;
978 }
979
980 static noinline_for_stack int
981 update_cache_item(struct btrfs_trans_handle *trans,
982 struct btrfs_root *root,
983 struct inode *inode,
984 struct btrfs_path *path, u64 offset,
985 int entries, int bitmaps)
986 {
987 struct btrfs_key key;
988 struct btrfs_free_space_header *header;
989 struct extent_buffer *leaf;
990 int ret;
991
992 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
993 key.offset = offset;
994 key.type = 0;
995
996 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
997 if (ret < 0) {
998 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
999 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
1000 GFP_NOFS);
1001 goto fail;
1002 }
1003 leaf = path->nodes[0];
1004 if (ret > 0) {
1005 struct btrfs_key found_key;
1006 ASSERT(path->slots[0]);
1007 path->slots[0]--;
1008 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1009 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
1010 found_key.offset != offset) {
1011 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
1012 inode->i_size - 1,
1013 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0,
1014 NULL, GFP_NOFS);
1015 btrfs_release_path(path);
1016 goto fail;
1017 }
1018 }
1019
1020 BTRFS_I(inode)->generation = trans->transid;
1021 header = btrfs_item_ptr(leaf, path->slots[0],
1022 struct btrfs_free_space_header);
1023 btrfs_set_free_space_entries(leaf, header, entries);
1024 btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1025 btrfs_set_free_space_generation(leaf, header, trans->transid);
1026 btrfs_mark_buffer_dirty(leaf);
1027 btrfs_release_path(path);
1028
1029 return 0;
1030
1031 fail:
1032 return -1;
1033 }
1034
1035 static noinline_for_stack int
1036 write_pinned_extent_entries(struct btrfs_root *root,
1037 struct btrfs_block_group_cache *block_group,
1038 struct btrfs_io_ctl *io_ctl,
1039 int *entries)
1040 {
1041 u64 start, extent_start, extent_end, len;
1042 struct extent_io_tree *unpin = NULL;
1043 int ret;
1044
1045 if (!block_group)
1046 return 0;
1047
1048 /*
1049 * We want to add any pinned extents to our free space cache
1050 * so we don't leak the space
1051 *
1052 * We shouldn't have switched the pinned extents yet so this is the
1053 * right one
1054 */
1055 unpin = root->fs_info->pinned_extents;
1056
1057 start = block_group->key.objectid;
1058
1059 while (start < block_group->key.objectid + block_group->key.offset) {
1060 ret = find_first_extent_bit(unpin, start,
1061 &extent_start, &extent_end,
1062 EXTENT_DIRTY, NULL);
1063 if (ret)
1064 return 0;
1065
1066 /* This pinned extent is out of our range */
1067 if (extent_start >= block_group->key.objectid +
1068 block_group->key.offset)
1069 return 0;
1070
1071 extent_start = max(extent_start, start);
1072 extent_end = min(block_group->key.objectid +
1073 block_group->key.offset, extent_end + 1);
1074 len = extent_end - extent_start;
1075
1076 *entries += 1;
1077 ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
1078 if (ret)
1079 return -ENOSPC;
1080
1081 start = extent_end;
1082 }
1083
1084 return 0;
1085 }
1086
1087 static noinline_for_stack int
1088 write_bitmap_entries(struct btrfs_io_ctl *io_ctl, struct list_head *bitmap_list)
1089 {
1090 struct btrfs_free_space *entry, *next;
1091 int ret;
1092
1093 /* Write out the bitmaps */
1094 list_for_each_entry_safe(entry, next, bitmap_list, list) {
1095 ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
1096 if (ret)
1097 return -ENOSPC;
1098 list_del_init(&entry->list);
1099 }
1100
1101 return 0;
1102 }
1103
1104 static int flush_dirty_cache(struct inode *inode)
1105 {
1106 int ret;
1107
1108 ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
1109 if (ret)
1110 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1111 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
1112 GFP_NOFS);
1113
1114 return ret;
1115 }
1116
1117 static void noinline_for_stack
1118 cleanup_bitmap_list(struct list_head *bitmap_list)
1119 {
1120 struct btrfs_free_space *entry, *next;
1121
1122 list_for_each_entry_safe(entry, next, bitmap_list, list)
1123 list_del_init(&entry->list);
1124 }
1125
1126 static void noinline_for_stack
1127 cleanup_write_cache_enospc(struct inode *inode,
1128 struct btrfs_io_ctl *io_ctl,
1129 struct extent_state **cached_state,
1130 struct list_head *bitmap_list)
1131 {
1132 io_ctl_drop_pages(io_ctl);
1133 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1134 i_size_read(inode) - 1, cached_state,
1135 GFP_NOFS);
1136 }
1137
1138 int btrfs_wait_cache_io(struct btrfs_root *root,
1139 struct btrfs_trans_handle *trans,
1140 struct btrfs_block_group_cache *block_group,
1141 struct btrfs_io_ctl *io_ctl,
1142 struct btrfs_path *path, u64 offset)
1143 {
1144 int ret;
1145 struct inode *inode = io_ctl->inode;
1146
1147 if (!inode)
1148 return 0;
1149
1150 if (block_group)
1151 root = root->fs_info->tree_root;
1152
1153 /* Flush the dirty pages in the cache file. */
1154 ret = flush_dirty_cache(inode);
1155 if (ret)
1156 goto out;
1157
1158 /* Update the cache item to tell everyone this cache file is valid. */
1159 ret = update_cache_item(trans, root, inode, path, offset,
1160 io_ctl->entries, io_ctl->bitmaps);
1161 out:
1162 io_ctl_free(io_ctl);
1163 if (ret) {
1164 invalidate_inode_pages2(inode->i_mapping);
1165 BTRFS_I(inode)->generation = 0;
1166 if (block_group) {
1167 #ifdef DEBUG
1168 btrfs_err(root->fs_info,
1169 "failed to write free space cache for block group %llu",
1170 block_group->key.objectid);
1171 #endif
1172 }
1173 }
1174 btrfs_update_inode(trans, root, inode);
1175
1176 if (block_group) {
1177 /* the dirty list is protected by the dirty_bgs_lock */
1178 spin_lock(&trans->transaction->dirty_bgs_lock);
1179
1180 /* the disk_cache_state is protected by the block group lock */
1181 spin_lock(&block_group->lock);
1182
1183 /*
1184 * only mark this as written if we didn't get put back on
1185 * the dirty list while waiting for IO. Otherwise our
1186 * cache state won't be right, and we won't get written again
1187 */
1188 if (!ret && list_empty(&block_group->dirty_list))
1189 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1190 else if (ret)
1191 block_group->disk_cache_state = BTRFS_DC_ERROR;
1192
1193 spin_unlock(&block_group->lock);
1194 spin_unlock(&trans->transaction->dirty_bgs_lock);
1195 io_ctl->inode = NULL;
1196 iput(inode);
1197 }
1198
1199 return ret;
1200
1201 }
1202
1203 /**
1204 * __btrfs_write_out_cache - write out cached info to an inode
1205 * @root - the root the inode belongs to
1206 * @ctl - the free space cache we are going to write out
1207 * @block_group - the block_group for this cache if it belongs to a block_group
1208 * @trans - the trans handle
1209 * @path - the path to use
1210 * @offset - the offset for the key we'll insert
1211 *
1212 * This function writes out a free space cache struct to disk for quick recovery
1213 * on mount. This will return 0 if it was successful in writing the cache out,
1214 * or an errno if it was not.
1215 */
1216 static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
1217 struct btrfs_free_space_ctl *ctl,
1218 struct btrfs_block_group_cache *block_group,
1219 struct btrfs_io_ctl *io_ctl,
1220 struct btrfs_trans_handle *trans,
1221 struct btrfs_path *path, u64 offset)
1222 {
1223 struct extent_state *cached_state = NULL;
1224 LIST_HEAD(bitmap_list);
1225 int entries = 0;
1226 int bitmaps = 0;
1227 int ret;
1228 int must_iput = 0;
1229
1230 if (!i_size_read(inode))
1231 return -EIO;
1232
1233 WARN_ON(io_ctl->pages);
1234 ret = io_ctl_init(io_ctl, inode, root, 1);
1235 if (ret)
1236 return ret;
1237
1238 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
1239 down_write(&block_group->data_rwsem);
1240 spin_lock(&block_group->lock);
1241 if (block_group->delalloc_bytes) {
1242 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1243 spin_unlock(&block_group->lock);
1244 up_write(&block_group->data_rwsem);
1245 BTRFS_I(inode)->generation = 0;
1246 ret = 0;
1247 must_iput = 1;
1248 goto out;
1249 }
1250 spin_unlock(&block_group->lock);
1251 }
1252
1253 /* Lock all pages first so we can lock the extent safely. */
1254 ret = io_ctl_prepare_pages(io_ctl, inode, 0);
1255 if (ret)
1256 goto out;
1257
1258 lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1259 &cached_state);
1260
1261 io_ctl_set_generation(io_ctl, trans->transid);
1262
1263 mutex_lock(&ctl->cache_writeout_mutex);
1264 /* Write out the extent entries in the free space cache */
1265 spin_lock(&ctl->tree_lock);
1266 ret = write_cache_extent_entries(io_ctl, ctl,
1267 block_group, &entries, &bitmaps,
1268 &bitmap_list);
1269 if (ret)
1270 goto out_nospc_locked;
1271
1272 /*
1273 * Some spaces that are freed in the current transaction are pinned,
1274 * they will be added into free space cache after the transaction is
1275 * committed, we shouldn't lose them.
1276 *
1277 * If this changes while we are working we'll get added back to
1278 * the dirty list and redo it. No locking needed
1279 */
1280 ret = write_pinned_extent_entries(root, block_group, io_ctl, &entries);
1281 if (ret)
1282 goto out_nospc_locked;
1283
1284 /*
1285 * At last, we write out all the bitmaps and keep cache_writeout_mutex
1286 * locked while doing it because a concurrent trim can be manipulating
1287 * or freeing the bitmap.
1288 */
1289 ret = write_bitmap_entries(io_ctl, &bitmap_list);
1290 spin_unlock(&ctl->tree_lock);
1291 mutex_unlock(&ctl->cache_writeout_mutex);
1292 if (ret)
1293 goto out_nospc;
1294
1295 /* Zero out the rest of the pages just to make sure */
1296 io_ctl_zero_remaining_pages(io_ctl);
1297
1298 /* Everything is written out, now we dirty the pages in the file. */
1299 ret = btrfs_dirty_pages(root, inode, io_ctl->pages, io_ctl->num_pages,
1300 0, i_size_read(inode), &cached_state);
1301 if (ret)
1302 goto out_nospc;
1303
1304 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1305 up_write(&block_group->data_rwsem);
1306 /*
1307 * Release the pages and unlock the extent, we will flush
1308 * them out later
1309 */
1310 io_ctl_drop_pages(io_ctl);
1311
1312 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1313 i_size_read(inode) - 1, &cached_state, GFP_NOFS);
1314
1315 /*
1316 * at this point the pages are under IO and we're happy,
1317 * The caller is responsible for waiting on them and updating the
1318 * the cache and the inode
1319 */
1320 io_ctl->entries = entries;
1321 io_ctl->bitmaps = bitmaps;
1322
1323 ret = btrfs_fdatawrite_range(inode, 0, (u64)-1);
1324 if (ret)
1325 goto out;
1326
1327 return 0;
1328
1329 out:
1330 io_ctl->inode = NULL;
1331 io_ctl_free(io_ctl);
1332 if (ret) {
1333 invalidate_inode_pages2(inode->i_mapping);
1334 BTRFS_I(inode)->generation = 0;
1335 }
1336 btrfs_update_inode(trans, root, inode);
1337 if (must_iput)
1338 iput(inode);
1339 return ret;
1340
1341 out_nospc_locked:
1342 cleanup_bitmap_list(&bitmap_list);
1343 spin_unlock(&ctl->tree_lock);
1344 mutex_unlock(&ctl->cache_writeout_mutex);
1345
1346 out_nospc:
1347 cleanup_write_cache_enospc(inode, io_ctl, &cached_state, &bitmap_list);
1348
1349 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1350 up_write(&block_group->data_rwsem);
1351
1352 goto out;
1353 }
1354
1355 int btrfs_write_out_cache(struct btrfs_root *root,
1356 struct btrfs_trans_handle *trans,
1357 struct btrfs_block_group_cache *block_group,
1358 struct btrfs_path *path)
1359 {
1360 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1361 struct inode *inode;
1362 int ret = 0;
1363
1364 root = root->fs_info->tree_root;
1365
1366 spin_lock(&block_group->lock);
1367 if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1368 spin_unlock(&block_group->lock);
1369 return 0;
1370 }
1371 spin_unlock(&block_group->lock);
1372
1373 inode = lookup_free_space_inode(root, block_group, path);
1374 if (IS_ERR(inode))
1375 return 0;
1376
1377 ret = __btrfs_write_out_cache(root, inode, ctl, block_group,
1378 &block_group->io_ctl, trans,
1379 path, block_group->key.objectid);
1380 if (ret) {
1381 #ifdef DEBUG
1382 btrfs_err(root->fs_info,
1383 "failed to write free space cache for block group %llu",
1384 block_group->key.objectid);
1385 #endif
1386 spin_lock(&block_group->lock);
1387 block_group->disk_cache_state = BTRFS_DC_ERROR;
1388 spin_unlock(&block_group->lock);
1389
1390 block_group->io_ctl.inode = NULL;
1391 iput(inode);
1392 }
1393
1394 /*
1395 * if ret == 0 the caller is expected to call btrfs_wait_cache_io
1396 * to wait for IO and put the inode
1397 */
1398
1399 return ret;
1400 }
1401
1402 static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1403 u64 offset)
1404 {
1405 ASSERT(offset >= bitmap_start);
1406 offset -= bitmap_start;
1407 return (unsigned long)(div_u64(offset, unit));
1408 }
1409
1410 static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1411 {
1412 return (unsigned long)(div_u64(bytes, unit));
1413 }
1414
1415 static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1416 u64 offset)
1417 {
1418 u64 bitmap_start;
1419 u64 bytes_per_bitmap;
1420
1421 bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1422 bitmap_start = offset - ctl->start;
1423 bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1424 bitmap_start *= bytes_per_bitmap;
1425 bitmap_start += ctl->start;
1426
1427 return bitmap_start;
1428 }
1429
1430 static int tree_insert_offset(struct rb_root *root, u64 offset,
1431 struct rb_node *node, int bitmap)
1432 {
1433 struct rb_node **p = &root->rb_node;
1434 struct rb_node *parent = NULL;
1435 struct btrfs_free_space *info;
1436
1437 while (*p) {
1438 parent = *p;
1439 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1440
1441 if (offset < info->offset) {
1442 p = &(*p)->rb_left;
1443 } else if (offset > info->offset) {
1444 p = &(*p)->rb_right;
1445 } else {
1446 /*
1447 * we could have a bitmap entry and an extent entry
1448 * share the same offset. If this is the case, we want
1449 * the extent entry to always be found first if we do a
1450 * linear search through the tree, since we want to have
1451 * the quickest allocation time, and allocating from an
1452 * extent is faster than allocating from a bitmap. So
1453 * if we're inserting a bitmap and we find an entry at
1454 * this offset, we want to go right, or after this entry
1455 * logically. If we are inserting an extent and we've
1456 * found a bitmap, we want to go left, or before
1457 * logically.
1458 */
1459 if (bitmap) {
1460 if (info->bitmap) {
1461 WARN_ON_ONCE(1);
1462 return -EEXIST;
1463 }
1464 p = &(*p)->rb_right;
1465 } else {
1466 if (!info->bitmap) {
1467 WARN_ON_ONCE(1);
1468 return -EEXIST;
1469 }
1470 p = &(*p)->rb_left;
1471 }
1472 }
1473 }
1474
1475 rb_link_node(node, parent, p);
1476 rb_insert_color(node, root);
1477
1478 return 0;
1479 }
1480
1481 /*
1482 * searches the tree for the given offset.
1483 *
1484 * fuzzy - If this is set, then we are trying to make an allocation, and we just
1485 * want a section that has at least bytes size and comes at or after the given
1486 * offset.
1487 */
1488 static struct btrfs_free_space *
1489 tree_search_offset(struct btrfs_free_space_ctl *ctl,
1490 u64 offset, int bitmap_only, int fuzzy)
1491 {
1492 struct rb_node *n = ctl->free_space_offset.rb_node;
1493 struct btrfs_free_space *entry, *prev = NULL;
1494
1495 /* find entry that is closest to the 'offset' */
1496 while (1) {
1497 if (!n) {
1498 entry = NULL;
1499 break;
1500 }
1501
1502 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1503 prev = entry;
1504
1505 if (offset < entry->offset)
1506 n = n->rb_left;
1507 else if (offset > entry->offset)
1508 n = n->rb_right;
1509 else
1510 break;
1511 }
1512
1513 if (bitmap_only) {
1514 if (!entry)
1515 return NULL;
1516 if (entry->bitmap)
1517 return entry;
1518
1519 /*
1520 * bitmap entry and extent entry may share same offset,
1521 * in that case, bitmap entry comes after extent entry.
1522 */
1523 n = rb_next(n);
1524 if (!n)
1525 return NULL;
1526 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1527 if (entry->offset != offset)
1528 return NULL;
1529
1530 WARN_ON(!entry->bitmap);
1531 return entry;
1532 } else if (entry) {
1533 if (entry->bitmap) {
1534 /*
1535 * if previous extent entry covers the offset,
1536 * we should return it instead of the bitmap entry
1537 */
1538 n = rb_prev(&entry->offset_index);
1539 if (n) {
1540 prev = rb_entry(n, struct btrfs_free_space,
1541 offset_index);
1542 if (!prev->bitmap &&
1543 prev->offset + prev->bytes > offset)
1544 entry = prev;
1545 }
1546 }
1547 return entry;
1548 }
1549
1550 if (!prev)
1551 return NULL;
1552
1553 /* find last entry before the 'offset' */
1554 entry = prev;
1555 if (entry->offset > offset) {
1556 n = rb_prev(&entry->offset_index);
1557 if (n) {
1558 entry = rb_entry(n, struct btrfs_free_space,
1559 offset_index);
1560 ASSERT(entry->offset <= offset);
1561 } else {
1562 if (fuzzy)
1563 return entry;
1564 else
1565 return NULL;
1566 }
1567 }
1568
1569 if (entry->bitmap) {
1570 n = rb_prev(&entry->offset_index);
1571 if (n) {
1572 prev = rb_entry(n, struct btrfs_free_space,
1573 offset_index);
1574 if (!prev->bitmap &&
1575 prev->offset + prev->bytes > offset)
1576 return prev;
1577 }
1578 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1579 return entry;
1580 } else if (entry->offset + entry->bytes > offset)
1581 return entry;
1582
1583 if (!fuzzy)
1584 return NULL;
1585
1586 while (1) {
1587 if (entry->bitmap) {
1588 if (entry->offset + BITS_PER_BITMAP *
1589 ctl->unit > offset)
1590 break;
1591 } else {
1592 if (entry->offset + entry->bytes > offset)
1593 break;
1594 }
1595
1596 n = rb_next(&entry->offset_index);
1597 if (!n)
1598 return NULL;
1599 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1600 }
1601 return entry;
1602 }
1603
1604 static inline void
1605 __unlink_free_space(struct btrfs_free_space_ctl *ctl,
1606 struct btrfs_free_space *info)
1607 {
1608 rb_erase(&info->offset_index, &ctl->free_space_offset);
1609 ctl->free_extents--;
1610 }
1611
1612 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1613 struct btrfs_free_space *info)
1614 {
1615 __unlink_free_space(ctl, info);
1616 ctl->free_space -= info->bytes;
1617 }
1618
1619 static int link_free_space(struct btrfs_free_space_ctl *ctl,
1620 struct btrfs_free_space *info)
1621 {
1622 int ret = 0;
1623
1624 ASSERT(info->bytes || info->bitmap);
1625 ret = tree_insert_offset(&ctl->free_space_offset, info->offset,
1626 &info->offset_index, (info->bitmap != NULL));
1627 if (ret)
1628 return ret;
1629
1630 ctl->free_space += info->bytes;
1631 ctl->free_extents++;
1632 return ret;
1633 }
1634
1635 static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
1636 {
1637 struct btrfs_block_group_cache *block_group = ctl->private;
1638 u64 max_bytes;
1639 u64 bitmap_bytes;
1640 u64 extent_bytes;
1641 u64 size = block_group->key.offset;
1642 u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
1643 u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
1644
1645 max_bitmaps = max_t(u64, max_bitmaps, 1);
1646
1647 ASSERT(ctl->total_bitmaps <= max_bitmaps);
1648
1649 /*
1650 * The goal is to keep the total amount of memory used per 1gb of space
1651 * at or below 32k, so we need to adjust how much memory we allow to be
1652 * used by extent based free space tracking
1653 */
1654 if (size < SZ_1G)
1655 max_bytes = MAX_CACHE_BYTES_PER_GIG;
1656 else
1657 max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
1658
1659 /*
1660 * we want to account for 1 more bitmap than what we have so we can make
1661 * sure we don't go over our overall goal of MAX_CACHE_BYTES_PER_GIG as
1662 * we add more bitmaps.
1663 */
1664 bitmap_bytes = (ctl->total_bitmaps + 1) * ctl->unit;
1665
1666 if (bitmap_bytes >= max_bytes) {
1667 ctl->extents_thresh = 0;
1668 return;
1669 }
1670
1671 /*
1672 * we want the extent entry threshold to always be at most 1/2 the max
1673 * bytes we can have, or whatever is less than that.
1674 */
1675 extent_bytes = max_bytes - bitmap_bytes;
1676 extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
1677
1678 ctl->extents_thresh =
1679 div_u64(extent_bytes, sizeof(struct btrfs_free_space));
1680 }
1681
1682 static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1683 struct btrfs_free_space *info,
1684 u64 offset, u64 bytes)
1685 {
1686 unsigned long start, count;
1687
1688 start = offset_to_bit(info->offset, ctl->unit, offset);
1689 count = bytes_to_bits(bytes, ctl->unit);
1690 ASSERT(start + count <= BITS_PER_BITMAP);
1691
1692 bitmap_clear(info->bitmap, start, count);
1693
1694 info->bytes -= bytes;
1695 }
1696
1697 static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1698 struct btrfs_free_space *info, u64 offset,
1699 u64 bytes)
1700 {
1701 __bitmap_clear_bits(ctl, info, offset, bytes);
1702 ctl->free_space -= bytes;
1703 }
1704
1705 static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1706 struct btrfs_free_space *info, u64 offset,
1707 u64 bytes)
1708 {
1709 unsigned long start, count;
1710
1711 start = offset_to_bit(info->offset, ctl->unit, offset);
1712 count = bytes_to_bits(bytes, ctl->unit);
1713 ASSERT(start + count <= BITS_PER_BITMAP);
1714
1715 bitmap_set(info->bitmap, start, count);
1716
1717 info->bytes += bytes;
1718 ctl->free_space += bytes;
1719 }
1720
1721 /*
1722 * If we can not find suitable extent, we will use bytes to record
1723 * the size of the max extent.
1724 */
1725 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1726 struct btrfs_free_space *bitmap_info, u64 *offset,
1727 u64 *bytes, bool for_alloc)
1728 {
1729 unsigned long found_bits = 0;
1730 unsigned long max_bits = 0;
1731 unsigned long bits, i;
1732 unsigned long next_zero;
1733 unsigned long extent_bits;
1734
1735 /*
1736 * Skip searching the bitmap if we don't have a contiguous section that
1737 * is large enough for this allocation.
1738 */
1739 if (for_alloc &&
1740 bitmap_info->max_extent_size &&
1741 bitmap_info->max_extent_size < *bytes) {
1742 *bytes = bitmap_info->max_extent_size;
1743 return -1;
1744 }
1745
1746 i = offset_to_bit(bitmap_info->offset, ctl->unit,
1747 max_t(u64, *offset, bitmap_info->offset));
1748 bits = bytes_to_bits(*bytes, ctl->unit);
1749
1750 for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
1751 if (for_alloc && bits == 1) {
1752 found_bits = 1;
1753 break;
1754 }
1755 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1756 BITS_PER_BITMAP, i);
1757 extent_bits = next_zero - i;
1758 if (extent_bits >= bits) {
1759 found_bits = extent_bits;
1760 break;
1761 } else if (extent_bits > max_bits) {
1762 max_bits = extent_bits;
1763 }
1764 i = next_zero;
1765 }
1766
1767 if (found_bits) {
1768 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1769 *bytes = (u64)(found_bits) * ctl->unit;
1770 return 0;
1771 }
1772
1773 *bytes = (u64)(max_bits) * ctl->unit;
1774 bitmap_info->max_extent_size = *bytes;
1775 return -1;
1776 }
1777
1778 /* Cache the size of the max extent in bytes */
1779 static struct btrfs_free_space *
1780 find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
1781 unsigned long align, u64 *max_extent_size)
1782 {
1783 struct btrfs_free_space *entry;
1784 struct rb_node *node;
1785 u64 tmp;
1786 u64 align_off;
1787 int ret;
1788
1789 if (!ctl->free_space_offset.rb_node)
1790 goto out;
1791
1792 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
1793 if (!entry)
1794 goto out;
1795
1796 for (node = &entry->offset_index; node; node = rb_next(node)) {
1797 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1798 if (entry->bytes < *bytes) {
1799 if (entry->bytes > *max_extent_size)
1800 *max_extent_size = entry->bytes;
1801 continue;
1802 }
1803
1804 /* make sure the space returned is big enough
1805 * to match our requested alignment
1806 */
1807 if (*bytes >= align) {
1808 tmp = entry->offset - ctl->start + align - 1;
1809 tmp = div64_u64(tmp, align);
1810 tmp = tmp * align + ctl->start;
1811 align_off = tmp - entry->offset;
1812 } else {
1813 align_off = 0;
1814 tmp = entry->offset;
1815 }
1816
1817 if (entry->bytes < *bytes + align_off) {
1818 if (entry->bytes > *max_extent_size)
1819 *max_extent_size = entry->bytes;
1820 continue;
1821 }
1822
1823 if (entry->bitmap) {
1824 u64 size = *bytes;
1825
1826 ret = search_bitmap(ctl, entry, &tmp, &size, true);
1827 if (!ret) {
1828 *offset = tmp;
1829 *bytes = size;
1830 return entry;
1831 } else if (size > *max_extent_size) {
1832 *max_extent_size = size;
1833 }
1834 continue;
1835 }
1836
1837 *offset = tmp;
1838 *bytes = entry->bytes - align_off;
1839 return entry;
1840 }
1841 out:
1842 return NULL;
1843 }
1844
1845 static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
1846 struct btrfs_free_space *info, u64 offset)
1847 {
1848 info->offset = offset_to_bitmap(ctl, offset);
1849 info->bytes = 0;
1850 INIT_LIST_HEAD(&info->list);
1851 link_free_space(ctl, info);
1852 ctl->total_bitmaps++;
1853
1854 ctl->op->recalc_thresholds(ctl);
1855 }
1856
1857 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
1858 struct btrfs_free_space *bitmap_info)
1859 {
1860 unlink_free_space(ctl, bitmap_info);
1861 kfree(bitmap_info->bitmap);
1862 kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1863 ctl->total_bitmaps--;
1864 ctl->op->recalc_thresholds(ctl);
1865 }
1866
1867 static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
1868 struct btrfs_free_space *bitmap_info,
1869 u64 *offset, u64 *bytes)
1870 {
1871 u64 end;
1872 u64 search_start, search_bytes;
1873 int ret;
1874
1875 again:
1876 end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
1877
1878 /*
1879 * We need to search for bits in this bitmap. We could only cover some
1880 * of the extent in this bitmap thanks to how we add space, so we need
1881 * to search for as much as it as we can and clear that amount, and then
1882 * go searching for the next bit.
1883 */
1884 search_start = *offset;
1885 search_bytes = ctl->unit;
1886 search_bytes = min(search_bytes, end - search_start + 1);
1887 ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
1888 false);
1889 if (ret < 0 || search_start != *offset)
1890 return -EINVAL;
1891
1892 /* We may have found more bits than what we need */
1893 search_bytes = min(search_bytes, *bytes);
1894
1895 /* Cannot clear past the end of the bitmap */
1896 search_bytes = min(search_bytes, end - search_start + 1);
1897
1898 bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes);
1899 *offset += search_bytes;
1900 *bytes -= search_bytes;
1901
1902 if (*bytes) {
1903 struct rb_node *next = rb_next(&bitmap_info->offset_index);
1904 if (!bitmap_info->bytes)
1905 free_bitmap(ctl, bitmap_info);
1906
1907 /*
1908 * no entry after this bitmap, but we still have bytes to
1909 * remove, so something has gone wrong.
1910 */
1911 if (!next)
1912 return -EINVAL;
1913
1914 bitmap_info = rb_entry(next, struct btrfs_free_space,
1915 offset_index);
1916
1917 /*
1918 * if the next entry isn't a bitmap we need to return to let the
1919 * extent stuff do its work.
1920 */
1921 if (!bitmap_info->bitmap)
1922 return -EAGAIN;
1923
1924 /*
1925 * Ok the next item is a bitmap, but it may not actually hold
1926 * the information for the rest of this free space stuff, so
1927 * look for it, and if we don't find it return so we can try
1928 * everything over again.
1929 */
1930 search_start = *offset;
1931 search_bytes = ctl->unit;
1932 ret = search_bitmap(ctl, bitmap_info, &search_start,
1933 &search_bytes, false);
1934 if (ret < 0 || search_start != *offset)
1935 return -EAGAIN;
1936
1937 goto again;
1938 } else if (!bitmap_info->bytes)
1939 free_bitmap(ctl, bitmap_info);
1940
1941 return 0;
1942 }
1943
1944 static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
1945 struct btrfs_free_space *info, u64 offset,
1946 u64 bytes)
1947 {
1948 u64 bytes_to_set = 0;
1949 u64 end;
1950
1951 end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
1952
1953 bytes_to_set = min(end - offset, bytes);
1954
1955 bitmap_set_bits(ctl, info, offset, bytes_to_set);
1956
1957 /*
1958 * We set some bytes, we have no idea what the max extent size is
1959 * anymore.
1960 */
1961 info->max_extent_size = 0;
1962
1963 return bytes_to_set;
1964
1965 }
1966
1967 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
1968 struct btrfs_free_space *info)
1969 {
1970 struct btrfs_block_group_cache *block_group = ctl->private;
1971 bool forced = false;
1972
1973 #ifdef CONFIG_BTRFS_DEBUG
1974 if (btrfs_should_fragment_free_space(block_group->fs_info->extent_root,
1975 block_group))
1976 forced = true;
1977 #endif
1978
1979 /*
1980 * If we are below the extents threshold then we can add this as an
1981 * extent, and don't have to deal with the bitmap
1982 */
1983 if (!forced && ctl->free_extents < ctl->extents_thresh) {
1984 /*
1985 * If this block group has some small extents we don't want to
1986 * use up all of our free slots in the cache with them, we want
1987 * to reserve them to larger extents, however if we have plenty
1988 * of cache left then go ahead an dadd them, no sense in adding
1989 * the overhead of a bitmap if we don't have to.
1990 */
1991 if (info->bytes <= block_group->sectorsize * 4) {
1992 if (ctl->free_extents * 2 <= ctl->extents_thresh)
1993 return false;
1994 } else {
1995 return false;
1996 }
1997 }
1998
1999 /*
2000 * The original block groups from mkfs can be really small, like 8
2001 * megabytes, so don't bother with a bitmap for those entries. However
2002 * some block groups can be smaller than what a bitmap would cover but
2003 * are still large enough that they could overflow the 32k memory limit,
2004 * so allow those block groups to still be allowed to have a bitmap
2005 * entry.
2006 */
2007 if (((BITS_PER_BITMAP * ctl->unit) >> 1) > block_group->key.offset)
2008 return false;
2009
2010 return true;
2011 }
2012
2013 static const struct btrfs_free_space_op free_space_op = {
2014 .recalc_thresholds = recalculate_thresholds,
2015 .use_bitmap = use_bitmap,
2016 };
2017
2018 static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
2019 struct btrfs_free_space *info)
2020 {
2021 struct btrfs_free_space *bitmap_info;
2022 struct btrfs_block_group_cache *block_group = NULL;
2023 int added = 0;
2024 u64 bytes, offset, bytes_added;
2025 int ret;
2026
2027 bytes = info->bytes;
2028 offset = info->offset;
2029
2030 if (!ctl->op->use_bitmap(ctl, info))
2031 return 0;
2032
2033 if (ctl->op == &free_space_op)
2034 block_group = ctl->private;
2035 again:
2036 /*
2037 * Since we link bitmaps right into the cluster we need to see if we
2038 * have a cluster here, and if so and it has our bitmap we need to add
2039 * the free space to that bitmap.
2040 */
2041 if (block_group && !list_empty(&block_group->cluster_list)) {
2042 struct btrfs_free_cluster *cluster;
2043 struct rb_node *node;
2044 struct btrfs_free_space *entry;
2045
2046 cluster = list_entry(block_group->cluster_list.next,
2047 struct btrfs_free_cluster,
2048 block_group_list);
2049 spin_lock(&cluster->lock);
2050 node = rb_first(&cluster->root);
2051 if (!node) {
2052 spin_unlock(&cluster->lock);
2053 goto no_cluster_bitmap;
2054 }
2055
2056 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2057 if (!entry->bitmap) {
2058 spin_unlock(&cluster->lock);
2059 goto no_cluster_bitmap;
2060 }
2061
2062 if (entry->offset == offset_to_bitmap(ctl, offset)) {
2063 bytes_added = add_bytes_to_bitmap(ctl, entry,
2064 offset, bytes);
2065 bytes -= bytes_added;
2066 offset += bytes_added;
2067 }
2068 spin_unlock(&cluster->lock);
2069 if (!bytes) {
2070 ret = 1;
2071 goto out;
2072 }
2073 }
2074
2075 no_cluster_bitmap:
2076 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2077 1, 0);
2078 if (!bitmap_info) {
2079 ASSERT(added == 0);
2080 goto new_bitmap;
2081 }
2082
2083 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
2084 bytes -= bytes_added;
2085 offset += bytes_added;
2086 added = 0;
2087
2088 if (!bytes) {
2089 ret = 1;
2090 goto out;
2091 } else
2092 goto again;
2093
2094 new_bitmap:
2095 if (info && info->bitmap) {
2096 add_new_bitmap(ctl, info, offset);
2097 added = 1;
2098 info = NULL;
2099 goto again;
2100 } else {
2101 spin_unlock(&ctl->tree_lock);
2102
2103 /* no pre-allocated info, allocate a new one */
2104 if (!info) {
2105 info = kmem_cache_zalloc(btrfs_free_space_cachep,
2106 GFP_NOFS);
2107 if (!info) {
2108 spin_lock(&ctl->tree_lock);
2109 ret = -ENOMEM;
2110 goto out;
2111 }
2112 }
2113
2114 /* allocate the bitmap */
2115 info->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
2116 spin_lock(&ctl->tree_lock);
2117 if (!info->bitmap) {
2118 ret = -ENOMEM;
2119 goto out;
2120 }
2121 goto again;
2122 }
2123
2124 out:
2125 if (info) {
2126 if (info->bitmap)
2127 kfree(info->bitmap);
2128 kmem_cache_free(btrfs_free_space_cachep, info);
2129 }
2130
2131 return ret;
2132 }
2133
2134 static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
2135 struct btrfs_free_space *info, bool update_stat)
2136 {
2137 struct btrfs_free_space *left_info;
2138 struct btrfs_free_space *right_info;
2139 bool merged = false;
2140 u64 offset = info->offset;
2141 u64 bytes = info->bytes;
2142
2143 /*
2144 * first we want to see if there is free space adjacent to the range we
2145 * are adding, if there is remove that struct and add a new one to
2146 * cover the entire range
2147 */
2148 right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
2149 if (right_info && rb_prev(&right_info->offset_index))
2150 left_info = rb_entry(rb_prev(&right_info->offset_index),
2151 struct btrfs_free_space, offset_index);
2152 else
2153 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
2154
2155 if (right_info && !right_info->bitmap) {
2156 if (update_stat)
2157 unlink_free_space(ctl, right_info);
2158 else
2159 __unlink_free_space(ctl, right_info);
2160 info->bytes += right_info->bytes;
2161 kmem_cache_free(btrfs_free_space_cachep, right_info);
2162 merged = true;
2163 }
2164
2165 if (left_info && !left_info->bitmap &&
2166 left_info->offset + left_info->bytes == offset) {
2167 if (update_stat)
2168 unlink_free_space(ctl, left_info);
2169 else
2170 __unlink_free_space(ctl, left_info);
2171 info->offset = left_info->offset;
2172 info->bytes += left_info->bytes;
2173 kmem_cache_free(btrfs_free_space_cachep, left_info);
2174 merged = true;
2175 }
2176
2177 return merged;
2178 }
2179
2180 static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
2181 struct btrfs_free_space *info,
2182 bool update_stat)
2183 {
2184 struct btrfs_free_space *bitmap;
2185 unsigned long i;
2186 unsigned long j;
2187 const u64 end = info->offset + info->bytes;
2188 const u64 bitmap_offset = offset_to_bitmap(ctl, end);
2189 u64 bytes;
2190
2191 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2192 if (!bitmap)
2193 return false;
2194
2195 i = offset_to_bit(bitmap->offset, ctl->unit, end);
2196 j = find_next_zero_bit(bitmap->bitmap, BITS_PER_BITMAP, i);
2197 if (j == i)
2198 return false;
2199 bytes = (j - i) * ctl->unit;
2200 info->bytes += bytes;
2201
2202 if (update_stat)
2203 bitmap_clear_bits(ctl, bitmap, end, bytes);
2204 else
2205 __bitmap_clear_bits(ctl, bitmap, end, bytes);
2206
2207 if (!bitmap->bytes)
2208 free_bitmap(ctl, bitmap);
2209
2210 return true;
2211 }
2212
2213 static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
2214 struct btrfs_free_space *info,
2215 bool update_stat)
2216 {
2217 struct btrfs_free_space *bitmap;
2218 u64 bitmap_offset;
2219 unsigned long i;
2220 unsigned long j;
2221 unsigned long prev_j;
2222 u64 bytes;
2223
2224 bitmap_offset = offset_to_bitmap(ctl, info->offset);
2225 /* If we're on a boundary, try the previous logical bitmap. */
2226 if (bitmap_offset == info->offset) {
2227 if (info->offset == 0)
2228 return false;
2229 bitmap_offset = offset_to_bitmap(ctl, info->offset - 1);
2230 }
2231
2232 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2233 if (!bitmap)
2234 return false;
2235
2236 i = offset_to_bit(bitmap->offset, ctl->unit, info->offset) - 1;
2237 j = 0;
2238 prev_j = (unsigned long)-1;
2239 for_each_clear_bit_from(j, bitmap->bitmap, BITS_PER_BITMAP) {
2240 if (j > i)
2241 break;
2242 prev_j = j;
2243 }
2244 if (prev_j == i)
2245 return false;
2246
2247 if (prev_j == (unsigned long)-1)
2248 bytes = (i + 1) * ctl->unit;
2249 else
2250 bytes = (i - prev_j) * ctl->unit;
2251
2252 info->offset -= bytes;
2253 info->bytes += bytes;
2254
2255 if (update_stat)
2256 bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2257 else
2258 __bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2259
2260 if (!bitmap->bytes)
2261 free_bitmap(ctl, bitmap);
2262
2263 return true;
2264 }
2265
2266 /*
2267 * We prefer always to allocate from extent entries, both for clustered and
2268 * non-clustered allocation requests. So when attempting to add a new extent
2269 * entry, try to see if there's adjacent free space in bitmap entries, and if
2270 * there is, migrate that space from the bitmaps to the extent.
2271 * Like this we get better chances of satisfying space allocation requests
2272 * because we attempt to satisfy them based on a single cache entry, and never
2273 * on 2 or more entries - even if the entries represent a contiguous free space
2274 * region (e.g. 1 extent entry + 1 bitmap entry starting where the extent entry
2275 * ends).
2276 */
2277 static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
2278 struct btrfs_free_space *info,
2279 bool update_stat)
2280 {
2281 /*
2282 * Only work with disconnected entries, as we can change their offset,
2283 * and must be extent entries.
2284 */
2285 ASSERT(!info->bitmap);
2286 ASSERT(RB_EMPTY_NODE(&info->offset_index));
2287
2288 if (ctl->total_bitmaps > 0) {
2289 bool stole_end;
2290 bool stole_front = false;
2291
2292 stole_end = steal_from_bitmap_to_end(ctl, info, update_stat);
2293 if (ctl->total_bitmaps > 0)
2294 stole_front = steal_from_bitmap_to_front(ctl, info,
2295 update_stat);
2296
2297 if (stole_end || stole_front)
2298 try_merge_free_space(ctl, info, update_stat);
2299 }
2300 }
2301
2302 int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
2303 struct btrfs_free_space_ctl *ctl,
2304 u64 offset, u64 bytes)
2305 {
2306 struct btrfs_free_space *info;
2307 int ret = 0;
2308
2309 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
2310 if (!info)
2311 return -ENOMEM;
2312
2313 info->offset = offset;
2314 info->bytes = bytes;
2315 RB_CLEAR_NODE(&info->offset_index);
2316
2317 spin_lock(&ctl->tree_lock);
2318
2319 if (try_merge_free_space(ctl, info, true))
2320 goto link;
2321
2322 /*
2323 * There was no extent directly to the left or right of this new
2324 * extent then we know we're going to have to allocate a new extent, so
2325 * before we do that see if we need to drop this into a bitmap
2326 */
2327 ret = insert_into_bitmap(ctl, info);
2328 if (ret < 0) {
2329 goto out;
2330 } else if (ret) {
2331 ret = 0;
2332 goto out;
2333 }
2334 link:
2335 /*
2336 * Only steal free space from adjacent bitmaps if we're sure we're not
2337 * going to add the new free space to existing bitmap entries - because
2338 * that would mean unnecessary work that would be reverted. Therefore
2339 * attempt to steal space from bitmaps if we're adding an extent entry.
2340 */
2341 steal_from_bitmap(ctl, info, true);
2342
2343 ret = link_free_space(ctl, info);
2344 if (ret)
2345 kmem_cache_free(btrfs_free_space_cachep, info);
2346 out:
2347 spin_unlock(&ctl->tree_lock);
2348
2349 if (ret) {
2350 btrfs_crit(fs_info, "unable to add free space :%d", ret);
2351 ASSERT(ret != -EEXIST);
2352 }
2353
2354 return ret;
2355 }
2356
2357 int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
2358 u64 offset, u64 bytes)
2359 {
2360 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2361 struct btrfs_free_space *info;
2362 int ret;
2363 bool re_search = false;
2364
2365 spin_lock(&ctl->tree_lock);
2366
2367 again:
2368 ret = 0;
2369 if (!bytes)
2370 goto out_lock;
2371
2372 info = tree_search_offset(ctl, offset, 0, 0);
2373 if (!info) {
2374 /*
2375 * oops didn't find an extent that matched the space we wanted
2376 * to remove, look for a bitmap instead
2377 */
2378 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2379 1, 0);
2380 if (!info) {
2381 /*
2382 * If we found a partial bit of our free space in a
2383 * bitmap but then couldn't find the other part this may
2384 * be a problem, so WARN about it.
2385 */
2386 WARN_ON(re_search);
2387 goto out_lock;
2388 }
2389 }
2390
2391 re_search = false;
2392 if (!info->bitmap) {
2393 unlink_free_space(ctl, info);
2394 if (offset == info->offset) {
2395 u64 to_free = min(bytes, info->bytes);
2396
2397 info->bytes -= to_free;
2398 info->offset += to_free;
2399 if (info->bytes) {
2400 ret = link_free_space(ctl, info);
2401 WARN_ON(ret);
2402 } else {
2403 kmem_cache_free(btrfs_free_space_cachep, info);
2404 }
2405
2406 offset += to_free;
2407 bytes -= to_free;
2408 goto again;
2409 } else {
2410 u64 old_end = info->bytes + info->offset;
2411
2412 info->bytes = offset - info->offset;
2413 ret = link_free_space(ctl, info);
2414 WARN_ON(ret);
2415 if (ret)
2416 goto out_lock;
2417
2418 /* Not enough bytes in this entry to satisfy us */
2419 if (old_end < offset + bytes) {
2420 bytes -= old_end - offset;
2421 offset = old_end;
2422 goto again;
2423 } else if (old_end == offset + bytes) {
2424 /* all done */
2425 goto out_lock;
2426 }
2427 spin_unlock(&ctl->tree_lock);
2428
2429 ret = btrfs_add_free_space(block_group, offset + bytes,
2430 old_end - (offset + bytes));
2431 WARN_ON(ret);
2432 goto out;
2433 }
2434 }
2435
2436 ret = remove_from_bitmap(ctl, info, &offset, &bytes);
2437 if (ret == -EAGAIN) {
2438 re_search = true;
2439 goto again;
2440 }
2441 out_lock:
2442 spin_unlock(&ctl->tree_lock);
2443 out:
2444 return ret;
2445 }
2446
2447 void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
2448 u64 bytes)
2449 {
2450 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2451 struct btrfs_free_space *info;
2452 struct rb_node *n;
2453 int count = 0;
2454
2455 for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
2456 info = rb_entry(n, struct btrfs_free_space, offset_index);
2457 if (info->bytes >= bytes && !block_group->ro)
2458 count++;
2459 btrfs_crit(block_group->fs_info,
2460 "entry offset %llu, bytes %llu, bitmap %s",
2461 info->offset, info->bytes,
2462 (info->bitmap) ? "yes" : "no");
2463 }
2464 btrfs_info(block_group->fs_info, "block group has cluster?: %s",
2465 list_empty(&block_group->cluster_list) ? "no" : "yes");
2466 btrfs_info(block_group->fs_info,
2467 "%d blocks of free space at or bigger than bytes is", count);
2468 }
2469
2470 void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group)
2471 {
2472 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2473
2474 spin_lock_init(&ctl->tree_lock);
2475 ctl->unit = block_group->sectorsize;
2476 ctl->start = block_group->key.objectid;
2477 ctl->private = block_group;
2478 ctl->op = &free_space_op;
2479 INIT_LIST_HEAD(&ctl->trimming_ranges);
2480 mutex_init(&ctl->cache_writeout_mutex);
2481
2482 /*
2483 * we only want to have 32k of ram per block group for keeping
2484 * track of free space, and if we pass 1/2 of that we want to
2485 * start converting things over to using bitmaps
2486 */
2487 ctl->extents_thresh = (SZ_32K / 2) / sizeof(struct btrfs_free_space);
2488 }
2489
2490 /*
2491 * for a given cluster, put all of its extents back into the free
2492 * space cache. If the block group passed doesn't match the block group
2493 * pointed to by the cluster, someone else raced in and freed the
2494 * cluster already. In that case, we just return without changing anything
2495 */
2496 static int
2497 __btrfs_return_cluster_to_free_space(
2498 struct btrfs_block_group_cache *block_group,
2499 struct btrfs_free_cluster *cluster)
2500 {
2501 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2502 struct btrfs_free_space *entry;
2503 struct rb_node *node;
2504
2505 spin_lock(&cluster->lock);
2506 if (cluster->block_group != block_group)
2507 goto out;
2508
2509 cluster->block_group = NULL;
2510 cluster->window_start = 0;
2511 list_del_init(&cluster->block_group_list);
2512
2513 node = rb_first(&cluster->root);
2514 while (node) {
2515 bool bitmap;
2516
2517 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2518 node = rb_next(&entry->offset_index);
2519 rb_erase(&entry->offset_index, &cluster->root);
2520 RB_CLEAR_NODE(&entry->offset_index);
2521
2522 bitmap = (entry->bitmap != NULL);
2523 if (!bitmap) {
2524 try_merge_free_space(ctl, entry, false);
2525 steal_from_bitmap(ctl, entry, false);
2526 }
2527 tree_insert_offset(&ctl->free_space_offset,
2528 entry->offset, &entry->offset_index, bitmap);
2529 }
2530 cluster->root = RB_ROOT;
2531
2532 out:
2533 spin_unlock(&cluster->lock);
2534 btrfs_put_block_group(block_group);
2535 return 0;
2536 }
2537
2538 static void __btrfs_remove_free_space_cache_locked(
2539 struct btrfs_free_space_ctl *ctl)
2540 {
2541 struct btrfs_free_space *info;
2542 struct rb_node *node;
2543
2544 while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
2545 info = rb_entry(node, struct btrfs_free_space, offset_index);
2546 if (!info->bitmap) {
2547 unlink_free_space(ctl, info);
2548 kmem_cache_free(btrfs_free_space_cachep, info);
2549 } else {
2550 free_bitmap(ctl, info);
2551 }
2552
2553 cond_resched_lock(&ctl->tree_lock);
2554 }
2555 }
2556
2557 void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
2558 {
2559 spin_lock(&ctl->tree_lock);
2560 __btrfs_remove_free_space_cache_locked(ctl);
2561 spin_unlock(&ctl->tree_lock);
2562 }
2563
2564 void btrfs_remove_free_space_cache(struct btrfs_block_group_cache *block_group)
2565 {
2566 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2567 struct btrfs_free_cluster *cluster;
2568 struct list_head *head;
2569
2570 spin_lock(&ctl->tree_lock);
2571 while ((head = block_group->cluster_list.next) !=
2572 &block_group->cluster_list) {
2573 cluster = list_entry(head, struct btrfs_free_cluster,
2574 block_group_list);
2575
2576 WARN_ON(cluster->block_group != block_group);
2577 __btrfs_return_cluster_to_free_space(block_group, cluster);
2578
2579 cond_resched_lock(&ctl->tree_lock);
2580 }
2581 __btrfs_remove_free_space_cache_locked(ctl);
2582 spin_unlock(&ctl->tree_lock);
2583
2584 }
2585
2586 u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group,
2587 u64 offset, u64 bytes, u64 empty_size,
2588 u64 *max_extent_size)
2589 {
2590 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2591 struct btrfs_free_space *entry = NULL;
2592 u64 bytes_search = bytes + empty_size;
2593 u64 ret = 0;
2594 u64 align_gap = 0;
2595 u64 align_gap_len = 0;
2596
2597 spin_lock(&ctl->tree_lock);
2598 entry = find_free_space(ctl, &offset, &bytes_search,
2599 block_group->full_stripe_len, max_extent_size);
2600 if (!entry)
2601 goto out;
2602
2603 ret = offset;
2604 if (entry->bitmap) {
2605 bitmap_clear_bits(ctl, entry, offset, bytes);
2606 if (!entry->bytes)
2607 free_bitmap(ctl, entry);
2608 } else {
2609 unlink_free_space(ctl, entry);
2610 align_gap_len = offset - entry->offset;
2611 align_gap = entry->offset;
2612
2613 entry->offset = offset + bytes;
2614 WARN_ON(entry->bytes < bytes + align_gap_len);
2615
2616 entry->bytes -= bytes + align_gap_len;
2617 if (!entry->bytes)
2618 kmem_cache_free(btrfs_free_space_cachep, entry);
2619 else
2620 link_free_space(ctl, entry);
2621 }
2622 out:
2623 spin_unlock(&ctl->tree_lock);
2624
2625 if (align_gap_len)
2626 __btrfs_add_free_space(block_group->fs_info, ctl,
2627 align_gap, align_gap_len);
2628 return ret;
2629 }
2630
2631 /*
2632 * given a cluster, put all of its extents back into the free space
2633 * cache. If a block group is passed, this function will only free
2634 * a cluster that belongs to the passed block group.
2635 *
2636 * Otherwise, it'll get a reference on the block group pointed to by the
2637 * cluster and remove the cluster from it.
2638 */
2639 int btrfs_return_cluster_to_free_space(
2640 struct btrfs_block_group_cache *block_group,
2641 struct btrfs_free_cluster *cluster)
2642 {
2643 struct btrfs_free_space_ctl *ctl;
2644 int ret;
2645
2646 /* first, get a safe pointer to the block group */
2647 spin_lock(&cluster->lock);
2648 if (!block_group) {
2649 block_group = cluster->block_group;
2650 if (!block_group) {
2651 spin_unlock(&cluster->lock);
2652 return 0;
2653 }
2654 } else if (cluster->block_group != block_group) {
2655 /* someone else has already freed it don't redo their work */
2656 spin_unlock(&cluster->lock);
2657 return 0;
2658 }
2659 atomic_inc(&block_group->count);
2660 spin_unlock(&cluster->lock);
2661
2662 ctl = block_group->free_space_ctl;
2663
2664 /* now return any extents the cluster had on it */
2665 spin_lock(&ctl->tree_lock);
2666 ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
2667 spin_unlock(&ctl->tree_lock);
2668
2669 /* finally drop our ref */
2670 btrfs_put_block_group(block_group);
2671 return ret;
2672 }
2673
2674 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
2675 struct btrfs_free_cluster *cluster,
2676 struct btrfs_free_space *entry,
2677 u64 bytes, u64 min_start,
2678 u64 *max_extent_size)
2679 {
2680 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2681 int err;
2682 u64 search_start = cluster->window_start;
2683 u64 search_bytes = bytes;
2684 u64 ret = 0;
2685
2686 search_start = min_start;
2687 search_bytes = bytes;
2688
2689 err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
2690 if (err) {
2691 if (search_bytes > *max_extent_size)
2692 *max_extent_size = search_bytes;
2693 return 0;
2694 }
2695
2696 ret = search_start;
2697 __bitmap_clear_bits(ctl, entry, ret, bytes);
2698
2699 return ret;
2700 }
2701
2702 /*
2703 * given a cluster, try to allocate 'bytes' from it, returns 0
2704 * if it couldn't find anything suitably large, or a logical disk offset
2705 * if things worked out
2706 */
2707 u64 btrfs_alloc_from_cluster(struct btrfs_block_group_cache *block_group,
2708 struct btrfs_free_cluster *cluster, u64 bytes,
2709 u64 min_start, u64 *max_extent_size)
2710 {
2711 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2712 struct btrfs_free_space *entry = NULL;
2713 struct rb_node *node;
2714 u64 ret = 0;
2715
2716 spin_lock(&cluster->lock);
2717 if (bytes > cluster->max_size)
2718 goto out;
2719
2720 if (cluster->block_group != block_group)
2721 goto out;
2722
2723 node = rb_first(&cluster->root);
2724 if (!node)
2725 goto out;
2726
2727 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2728 while (1) {
2729 if (entry->bytes < bytes && entry->bytes > *max_extent_size)
2730 *max_extent_size = entry->bytes;
2731
2732 if (entry->bytes < bytes ||
2733 (!entry->bitmap && entry->offset < min_start)) {
2734 node = rb_next(&entry->offset_index);
2735 if (!node)
2736 break;
2737 entry = rb_entry(node, struct btrfs_free_space,
2738 offset_index);
2739 continue;
2740 }
2741
2742 if (entry->bitmap) {
2743 ret = btrfs_alloc_from_bitmap(block_group,
2744 cluster, entry, bytes,
2745 cluster->window_start,
2746 max_extent_size);
2747 if (ret == 0) {
2748 node = rb_next(&entry->offset_index);
2749 if (!node)
2750 break;
2751 entry = rb_entry(node, struct btrfs_free_space,
2752 offset_index);
2753 continue;
2754 }
2755 cluster->window_start += bytes;
2756 } else {
2757 ret = entry->offset;
2758
2759 entry->offset += bytes;
2760 entry->bytes -= bytes;
2761 }
2762
2763 if (entry->bytes == 0)
2764 rb_erase(&entry->offset_index, &cluster->root);
2765 break;
2766 }
2767 out:
2768 spin_unlock(&cluster->lock);
2769
2770 if (!ret)
2771 return 0;
2772
2773 spin_lock(&ctl->tree_lock);
2774
2775 ctl->free_space -= bytes;
2776 if (entry->bytes == 0) {
2777 ctl->free_extents--;
2778 if (entry->bitmap) {
2779 kfree(entry->bitmap);
2780 ctl->total_bitmaps--;
2781 ctl->op->recalc_thresholds(ctl);
2782 }
2783 kmem_cache_free(btrfs_free_space_cachep, entry);
2784 }
2785
2786 spin_unlock(&ctl->tree_lock);
2787
2788 return ret;
2789 }
2790
2791 static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
2792 struct btrfs_free_space *entry,
2793 struct btrfs_free_cluster *cluster,
2794 u64 offset, u64 bytes,
2795 u64 cont1_bytes, u64 min_bytes)
2796 {
2797 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2798 unsigned long next_zero;
2799 unsigned long i;
2800 unsigned long want_bits;
2801 unsigned long min_bits;
2802 unsigned long found_bits;
2803 unsigned long max_bits = 0;
2804 unsigned long start = 0;
2805 unsigned long total_found = 0;
2806 int ret;
2807
2808 i = offset_to_bit(entry->offset, ctl->unit,
2809 max_t(u64, offset, entry->offset));
2810 want_bits = bytes_to_bits(bytes, ctl->unit);
2811 min_bits = bytes_to_bits(min_bytes, ctl->unit);
2812
2813 /*
2814 * Don't bother looking for a cluster in this bitmap if it's heavily
2815 * fragmented.
2816 */
2817 if (entry->max_extent_size &&
2818 entry->max_extent_size < cont1_bytes)
2819 return -ENOSPC;
2820 again:
2821 found_bits = 0;
2822 for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
2823 next_zero = find_next_zero_bit(entry->bitmap,
2824 BITS_PER_BITMAP, i);
2825 if (next_zero - i >= min_bits) {
2826 found_bits = next_zero - i;
2827 if (found_bits > max_bits)
2828 max_bits = found_bits;
2829 break;
2830 }
2831 if (next_zero - i > max_bits)
2832 max_bits = next_zero - i;
2833 i = next_zero;
2834 }
2835
2836 if (!found_bits) {
2837 entry->max_extent_size = (u64)max_bits * ctl->unit;
2838 return -ENOSPC;
2839 }
2840
2841 if (!total_found) {
2842 start = i;
2843 cluster->max_size = 0;
2844 }
2845
2846 total_found += found_bits;
2847
2848 if (cluster->max_size < found_bits * ctl->unit)
2849 cluster->max_size = found_bits * ctl->unit;
2850
2851 if (total_found < want_bits || cluster->max_size < cont1_bytes) {
2852 i = next_zero + 1;
2853 goto again;
2854 }
2855
2856 cluster->window_start = start * ctl->unit + entry->offset;
2857 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2858 ret = tree_insert_offset(&cluster->root, entry->offset,
2859 &entry->offset_index, 1);
2860 ASSERT(!ret); /* -EEXIST; Logic error */
2861
2862 trace_btrfs_setup_cluster(block_group, cluster,
2863 total_found * ctl->unit, 1);
2864 return 0;
2865 }
2866
2867 /*
2868 * This searches the block group for just extents to fill the cluster with.
2869 * Try to find a cluster with at least bytes total bytes, at least one
2870 * extent of cont1_bytes, and other clusters of at least min_bytes.
2871 */
2872 static noinline int
2873 setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2874 struct btrfs_free_cluster *cluster,
2875 struct list_head *bitmaps, u64 offset, u64 bytes,
2876 u64 cont1_bytes, u64 min_bytes)
2877 {
2878 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2879 struct btrfs_free_space *first = NULL;
2880 struct btrfs_free_space *entry = NULL;
2881 struct btrfs_free_space *last;
2882 struct rb_node *node;
2883 u64 window_free;
2884 u64 max_extent;
2885 u64 total_size = 0;
2886
2887 entry = tree_search_offset(ctl, offset, 0, 1);
2888 if (!entry)
2889 return -ENOSPC;
2890
2891 /*
2892 * We don't want bitmaps, so just move along until we find a normal
2893 * extent entry.
2894 */
2895 while (entry->bitmap || entry->bytes < min_bytes) {
2896 if (entry->bitmap && list_empty(&entry->list))
2897 list_add_tail(&entry->list, bitmaps);
2898 node = rb_next(&entry->offset_index);
2899 if (!node)
2900 return -ENOSPC;
2901 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2902 }
2903
2904 window_free = entry->bytes;
2905 max_extent = entry->bytes;
2906 first = entry;
2907 last = entry;
2908
2909 for (node = rb_next(&entry->offset_index); node;
2910 node = rb_next(&entry->offset_index)) {
2911 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2912
2913 if (entry->bitmap) {
2914 if (list_empty(&entry->list))
2915 list_add_tail(&entry->list, bitmaps);
2916 continue;
2917 }
2918
2919 if (entry->bytes < min_bytes)
2920 continue;
2921
2922 last = entry;
2923 window_free += entry->bytes;
2924 if (entry->bytes > max_extent)
2925 max_extent = entry->bytes;
2926 }
2927
2928 if (window_free < bytes || max_extent < cont1_bytes)
2929 return -ENOSPC;
2930
2931 cluster->window_start = first->offset;
2932
2933 node = &first->offset_index;
2934
2935 /*
2936 * now we've found our entries, pull them out of the free space
2937 * cache and put them into the cluster rbtree
2938 */
2939 do {
2940 int ret;
2941
2942 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2943 node = rb_next(&entry->offset_index);
2944 if (entry->bitmap || entry->bytes < min_bytes)
2945 continue;
2946
2947 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2948 ret = tree_insert_offset(&cluster->root, entry->offset,
2949 &entry->offset_index, 0);
2950 total_size += entry->bytes;
2951 ASSERT(!ret); /* -EEXIST; Logic error */
2952 } while (node && entry != last);
2953
2954 cluster->max_size = max_extent;
2955 trace_btrfs_setup_cluster(block_group, cluster, total_size, 0);
2956 return 0;
2957 }
2958
2959 /*
2960 * This specifically looks for bitmaps that may work in the cluster, we assume
2961 * that we have already failed to find extents that will work.
2962 */
2963 static noinline int
2964 setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2965 struct btrfs_free_cluster *cluster,
2966 struct list_head *bitmaps, u64 offset, u64 bytes,
2967 u64 cont1_bytes, u64 min_bytes)
2968 {
2969 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2970 struct btrfs_free_space *entry = NULL;
2971 int ret = -ENOSPC;
2972 u64 bitmap_offset = offset_to_bitmap(ctl, offset);
2973
2974 if (ctl->total_bitmaps == 0)
2975 return -ENOSPC;
2976
2977 /*
2978 * The bitmap that covers offset won't be in the list unless offset
2979 * is just its start offset.
2980 */
2981 if (!list_empty(bitmaps))
2982 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
2983
2984 if (!entry || entry->offset != bitmap_offset) {
2985 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
2986 if (entry && list_empty(&entry->list))
2987 list_add(&entry->list, bitmaps);
2988 }
2989
2990 list_for_each_entry(entry, bitmaps, list) {
2991 if (entry->bytes < bytes)
2992 continue;
2993 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
2994 bytes, cont1_bytes, min_bytes);
2995 if (!ret)
2996 return 0;
2997 }
2998
2999 /*
3000 * The bitmaps list has all the bitmaps that record free space
3001 * starting after offset, so no more search is required.
3002 */
3003 return -ENOSPC;
3004 }
3005
3006 /*
3007 * here we try to find a cluster of blocks in a block group. The goal
3008 * is to find at least bytes+empty_size.
3009 * We might not find them all in one contiguous area.
3010 *
3011 * returns zero and sets up cluster if things worked out, otherwise
3012 * it returns -enospc
3013 */
3014 int btrfs_find_space_cluster(struct btrfs_root *root,
3015 struct btrfs_block_group_cache *block_group,
3016 struct btrfs_free_cluster *cluster,
3017 u64 offset, u64 bytes, u64 empty_size)
3018 {
3019 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3020 struct btrfs_free_space *entry, *tmp;
3021 LIST_HEAD(bitmaps);
3022 u64 min_bytes;
3023 u64 cont1_bytes;
3024 int ret;
3025
3026 /*
3027 * Choose the minimum extent size we'll require for this
3028 * cluster. For SSD_SPREAD, don't allow any fragmentation.
3029 * For metadata, allow allocates with smaller extents. For
3030 * data, keep it dense.
3031 */
3032 if (btrfs_test_opt(root->fs_info, SSD_SPREAD)) {
3033 cont1_bytes = min_bytes = bytes + empty_size;
3034 } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
3035 cont1_bytes = bytes;
3036 min_bytes = block_group->sectorsize;
3037 } else {
3038 cont1_bytes = max(bytes, (bytes + empty_size) >> 2);
3039 min_bytes = block_group->sectorsize;
3040 }
3041
3042 spin_lock(&ctl->tree_lock);
3043
3044 /*
3045 * If we know we don't have enough space to make a cluster don't even
3046 * bother doing all the work to try and find one.
3047 */
3048 if (ctl->free_space < bytes) {
3049 spin_unlock(&ctl->tree_lock);
3050 return -ENOSPC;
3051 }
3052
3053 spin_lock(&cluster->lock);
3054
3055 /* someone already found a cluster, hooray */
3056 if (cluster->block_group) {
3057 ret = 0;
3058 goto out;
3059 }
3060
3061 trace_btrfs_find_cluster(block_group, offset, bytes, empty_size,
3062 min_bytes);
3063
3064 ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
3065 bytes + empty_size,
3066 cont1_bytes, min_bytes);
3067 if (ret)
3068 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
3069 offset, bytes + empty_size,
3070 cont1_bytes, min_bytes);
3071
3072 /* Clear our temporary list */
3073 list_for_each_entry_safe(entry, tmp, &bitmaps, list)
3074 list_del_init(&entry->list);
3075
3076 if (!ret) {
3077 atomic_inc(&block_group->count);
3078 list_add_tail(&cluster->block_group_list,
3079 &block_group->cluster_list);
3080 cluster->block_group = block_group;
3081 } else {
3082 trace_btrfs_failed_cluster_setup(block_group);
3083 }
3084 out:
3085 spin_unlock(&cluster->lock);
3086 spin_unlock(&ctl->tree_lock);
3087
3088 return ret;
3089 }
3090
3091 /*
3092 * simple code to zero out a cluster
3093 */
3094 void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
3095 {
3096 spin_lock_init(&cluster->lock);
3097 spin_lock_init(&cluster->refill_lock);
3098 cluster->root = RB_ROOT;
3099 cluster->max_size = 0;
3100 cluster->fragmented = false;
3101 INIT_LIST_HEAD(&cluster->block_group_list);
3102 cluster->block_group = NULL;
3103 }
3104
3105 static int do_trimming(struct btrfs_block_group_cache *block_group,
3106 u64 *total_trimmed, u64 start, u64 bytes,
3107 u64 reserved_start, u64 reserved_bytes,
3108 struct btrfs_trim_range *trim_entry)
3109 {
3110 struct btrfs_space_info *space_info = block_group->space_info;
3111 struct btrfs_fs_info *fs_info = block_group->fs_info;
3112 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3113 int ret;
3114 int update = 0;
3115 u64 trimmed = 0;
3116
3117 spin_lock(&space_info->lock);
3118 spin_lock(&block_group->lock);
3119 if (!block_group->ro) {
3120 block_group->reserved += reserved_bytes;
3121 space_info->bytes_reserved += reserved_bytes;
3122 update = 1;
3123 }
3124 spin_unlock(&block_group->lock);
3125 spin_unlock(&space_info->lock);
3126
3127 ret = btrfs_discard_extent(fs_info->extent_root,
3128 start, bytes, &trimmed);
3129 if (!ret)
3130 *total_trimmed += trimmed;
3131
3132 mutex_lock(&ctl->cache_writeout_mutex);
3133 btrfs_add_free_space(block_group, reserved_start, reserved_bytes);
3134 list_del(&trim_entry->list);
3135 mutex_unlock(&ctl->cache_writeout_mutex);
3136
3137 if (update) {
3138 spin_lock(&space_info->lock);
3139 spin_lock(&block_group->lock);
3140 if (block_group->ro)
3141 space_info->bytes_readonly += reserved_bytes;
3142 block_group->reserved -= reserved_bytes;
3143 space_info->bytes_reserved -= reserved_bytes;
3144 spin_unlock(&space_info->lock);
3145 spin_unlock(&block_group->lock);
3146 }
3147
3148 return ret;
3149 }
3150
3151 static int trim_no_bitmap(struct btrfs_block_group_cache *block_group,
3152 u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3153 {
3154 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3155 struct btrfs_free_space *entry;
3156 struct rb_node *node;
3157 int ret = 0;
3158 u64 extent_start;
3159 u64 extent_bytes;
3160 u64 bytes;
3161
3162 while (start < end) {
3163 struct btrfs_trim_range trim_entry;
3164
3165 mutex_lock(&ctl->cache_writeout_mutex);
3166 spin_lock(&ctl->tree_lock);
3167
3168 if (ctl->free_space < minlen) {
3169 spin_unlock(&ctl->tree_lock);
3170 mutex_unlock(&ctl->cache_writeout_mutex);
3171 break;
3172 }
3173
3174 entry = tree_search_offset(ctl, start, 0, 1);
3175 if (!entry) {
3176 spin_unlock(&ctl->tree_lock);
3177 mutex_unlock(&ctl->cache_writeout_mutex);
3178 break;
3179 }
3180
3181 /* skip bitmaps */
3182 while (entry->bitmap) {
3183 node = rb_next(&entry->offset_index);
3184 if (!node) {
3185 spin_unlock(&ctl->tree_lock);
3186 mutex_unlock(&ctl->cache_writeout_mutex);
3187 goto out;
3188 }
3189 entry = rb_entry(node, struct btrfs_free_space,
3190 offset_index);
3191 }
3192
3193 if (entry->offset >= end) {
3194 spin_unlock(&ctl->tree_lock);
3195 mutex_unlock(&ctl->cache_writeout_mutex);
3196 break;
3197 }
3198
3199 extent_start = entry->offset;
3200 extent_bytes = entry->bytes;
3201 start = max(start, extent_start);
3202 bytes = min(extent_start + extent_bytes, end) - start;
3203 if (bytes < minlen) {
3204 spin_unlock(&ctl->tree_lock);
3205 mutex_unlock(&ctl->cache_writeout_mutex);
3206 goto next;
3207 }
3208
3209 unlink_free_space(ctl, entry);
3210 kmem_cache_free(btrfs_free_space_cachep, entry);
3211
3212 spin_unlock(&ctl->tree_lock);
3213 trim_entry.start = extent_start;
3214 trim_entry.bytes = extent_bytes;
3215 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3216 mutex_unlock(&ctl->cache_writeout_mutex);
3217
3218 ret = do_trimming(block_group, total_trimmed, start, bytes,
3219 extent_start, extent_bytes, &trim_entry);
3220 if (ret)
3221 break;
3222 next:
3223 start += bytes;
3224
3225 if (fatal_signal_pending(current)) {
3226 ret = -ERESTARTSYS;
3227 break;
3228 }
3229
3230 cond_resched();
3231 }
3232 out:
3233 return ret;
3234 }
3235
3236 static int trim_bitmaps(struct btrfs_block_group_cache *block_group,
3237 u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3238 {
3239 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3240 struct btrfs_free_space *entry;
3241 int ret = 0;
3242 int ret2;
3243 u64 bytes;
3244 u64 offset = offset_to_bitmap(ctl, start);
3245
3246 while (offset < end) {
3247 bool next_bitmap = false;
3248 struct btrfs_trim_range trim_entry;
3249
3250 mutex_lock(&ctl->cache_writeout_mutex);
3251 spin_lock(&ctl->tree_lock);
3252
3253 if (ctl->free_space < minlen) {
3254 spin_unlock(&ctl->tree_lock);
3255 mutex_unlock(&ctl->cache_writeout_mutex);
3256 break;
3257 }
3258
3259 entry = tree_search_offset(ctl, offset, 1, 0);
3260 if (!entry) {
3261 spin_unlock(&ctl->tree_lock);
3262 mutex_unlock(&ctl->cache_writeout_mutex);
3263 next_bitmap = true;
3264 goto next;
3265 }
3266
3267 bytes = minlen;
3268 ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
3269 if (ret2 || start >= end) {
3270 spin_unlock(&ctl->tree_lock);
3271 mutex_unlock(&ctl->cache_writeout_mutex);
3272 next_bitmap = true;
3273 goto next;
3274 }
3275
3276 bytes = min(bytes, end - start);
3277 if (bytes < minlen) {
3278 spin_unlock(&ctl->tree_lock);
3279 mutex_unlock(&ctl->cache_writeout_mutex);
3280 goto next;
3281 }
3282
3283 bitmap_clear_bits(ctl, entry, start, bytes);
3284 if (entry->bytes == 0)
3285 free_bitmap(ctl, entry);
3286
3287 spin_unlock(&ctl->tree_lock);
3288 trim_entry.start = start;
3289 trim_entry.bytes = bytes;
3290 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3291 mutex_unlock(&ctl->cache_writeout_mutex);
3292
3293 ret = do_trimming(block_group, total_trimmed, start, bytes,
3294 start, bytes, &trim_entry);
3295 if (ret)
3296 break;
3297 next:
3298 if (next_bitmap) {
3299 offset += BITS_PER_BITMAP * ctl->unit;
3300 } else {
3301 start += bytes;
3302 if (start >= offset + BITS_PER_BITMAP * ctl->unit)
3303 offset += BITS_PER_BITMAP * ctl->unit;
3304 }
3305
3306 if (fatal_signal_pending(current)) {
3307 ret = -ERESTARTSYS;
3308 break;
3309 }
3310
3311 cond_resched();
3312 }
3313
3314 return ret;
3315 }
3316
3317 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache)
3318 {
3319 atomic_inc(&cache->trimming);
3320 }
3321
3322 void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *block_group)
3323 {
3324 struct extent_map_tree *em_tree;
3325 struct extent_map *em;
3326 bool cleanup;
3327
3328 spin_lock(&block_group->lock);
3329 cleanup = (atomic_dec_and_test(&block_group->trimming) &&
3330 block_group->removed);
3331 spin_unlock(&block_group->lock);
3332
3333 if (cleanup) {
3334 lock_chunks(block_group->fs_info->chunk_root);
3335 em_tree = &block_group->fs_info->mapping_tree.map_tree;
3336 write_lock(&em_tree->lock);
3337 em = lookup_extent_mapping(em_tree, block_group->key.objectid,
3338 1);
3339 BUG_ON(!em); /* logic error, can't happen */
3340 /*
3341 * remove_extent_mapping() will delete us from the pinned_chunks
3342 * list, which is protected by the chunk mutex.
3343 */
3344 remove_extent_mapping(em_tree, em);
3345 write_unlock(&em_tree->lock);
3346 unlock_chunks(block_group->fs_info->chunk_root);
3347
3348 /* once for us and once for the tree */
3349 free_extent_map(em);
3350 free_extent_map(em);
3351
3352 /*
3353 * We've left one free space entry and other tasks trimming
3354 * this block group have left 1 entry each one. Free them.
3355 */
3356 __btrfs_remove_free_space_cache(block_group->free_space_ctl);
3357 }
3358 }
3359
3360 int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
3361 u64 *trimmed, u64 start, u64 end, u64 minlen)
3362 {
3363 int ret;
3364
3365 *trimmed = 0;
3366
3367 spin_lock(&block_group->lock);
3368 if (block_group->removed) {
3369 spin_unlock(&block_group->lock);
3370 return 0;
3371 }
3372 btrfs_get_block_group_trimming(block_group);
3373 spin_unlock(&block_group->lock);
3374
3375 ret = trim_no_bitmap(block_group, trimmed, start, end, minlen);
3376 if (ret)
3377 goto out;
3378
3379 ret = trim_bitmaps(block_group, trimmed, start, end, minlen);
3380 out:
3381 btrfs_put_block_group_trimming(block_group);
3382 return ret;
3383 }
3384
3385 /*
3386 * Find the left-most item in the cache tree, and then return the
3387 * smallest inode number in the item.
3388 *
3389 * Note: the returned inode number may not be the smallest one in
3390 * the tree, if the left-most item is a bitmap.
3391 */
3392 u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
3393 {
3394 struct btrfs_free_space_ctl *ctl = fs_root->free_ino_ctl;
3395 struct btrfs_free_space *entry = NULL;
3396 u64 ino = 0;
3397
3398 spin_lock(&ctl->tree_lock);
3399
3400 if (RB_EMPTY_ROOT(&ctl->free_space_offset))
3401 goto out;
3402
3403 entry = rb_entry(rb_first(&ctl->free_space_offset),
3404 struct btrfs_free_space, offset_index);
3405
3406 if (!entry->bitmap) {
3407 ino = entry->offset;
3408
3409 unlink_free_space(ctl, entry);
3410 entry->offset++;
3411 entry->bytes--;
3412 if (!entry->bytes)
3413 kmem_cache_free(btrfs_free_space_cachep, entry);
3414 else
3415 link_free_space(ctl, entry);
3416 } else {
3417 u64 offset = 0;
3418 u64 count = 1;
3419 int ret;
3420
3421 ret = search_bitmap(ctl, entry, &offset, &count, true);
3422 /* Logic error; Should be empty if it can't find anything */
3423 ASSERT(!ret);
3424
3425 ino = offset;
3426 bitmap_clear_bits(ctl, entry, offset, 1);
3427 if (entry->bytes == 0)
3428 free_bitmap(ctl, entry);
3429 }
3430 out:
3431 spin_unlock(&ctl->tree_lock);
3432
3433 return ino;
3434 }
3435
3436 struct inode *lookup_free_ino_inode(struct btrfs_root *root,
3437 struct btrfs_path *path)
3438 {
3439 struct inode *inode = NULL;
3440
3441 spin_lock(&root->ino_cache_lock);
3442 if (root->ino_cache_inode)
3443 inode = igrab(root->ino_cache_inode);
3444 spin_unlock(&root->ino_cache_lock);
3445 if (inode)
3446 return inode;
3447
3448 inode = __lookup_free_space_inode(root, path, 0);
3449 if (IS_ERR(inode))
3450 return inode;
3451
3452 spin_lock(&root->ino_cache_lock);
3453 if (!btrfs_fs_closing(root->fs_info))
3454 root->ino_cache_inode = igrab(inode);
3455 spin_unlock(&root->ino_cache_lock);
3456
3457 return inode;
3458 }
3459
3460 int create_free_ino_inode(struct btrfs_root *root,
3461 struct btrfs_trans_handle *trans,
3462 struct btrfs_path *path)
3463 {
3464 return __create_free_space_inode(root, trans, path,
3465 BTRFS_FREE_INO_OBJECTID, 0);
3466 }
3467
3468 int load_free_ino_cache(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
3469 {
3470 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3471 struct btrfs_path *path;
3472 struct inode *inode;
3473 int ret = 0;
3474 u64 root_gen = btrfs_root_generation(&root->root_item);
3475
3476 if (!btrfs_test_opt(root->fs_info, INODE_MAP_CACHE))
3477 return 0;
3478
3479 /*
3480 * If we're unmounting then just return, since this does a search on the
3481 * normal root and not the commit root and we could deadlock.
3482 */
3483 if (btrfs_fs_closing(fs_info))
3484 return 0;
3485
3486 path = btrfs_alloc_path();
3487 if (!path)
3488 return 0;
3489
3490 inode = lookup_free_ino_inode(root, path);
3491 if (IS_ERR(inode))
3492 goto out;
3493
3494 if (root_gen != BTRFS_I(inode)->generation)
3495 goto out_put;
3496
3497 ret = __load_free_space_cache(root, inode, ctl, path, 0);
3498
3499 if (ret < 0)
3500 btrfs_err(fs_info,
3501 "failed to load free ino cache for root %llu",
3502 root->root_key.objectid);
3503 out_put:
3504 iput(inode);
3505 out:
3506 btrfs_free_path(path);
3507 return ret;
3508 }
3509
3510 int btrfs_write_out_ino_cache(struct btrfs_root *root,
3511 struct btrfs_trans_handle *trans,
3512 struct btrfs_path *path,
3513 struct inode *inode)
3514 {
3515 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3516 int ret;
3517 struct btrfs_io_ctl io_ctl;
3518 bool release_metadata = true;
3519
3520 if (!btrfs_test_opt(root->fs_info, INODE_MAP_CACHE))
3521 return 0;
3522
3523 memset(&io_ctl, 0, sizeof(io_ctl));
3524 ret = __btrfs_write_out_cache(root, inode, ctl, NULL, &io_ctl,
3525 trans, path, 0);
3526 if (!ret) {
3527 /*
3528 * At this point writepages() didn't error out, so our metadata
3529 * reservation is released when the writeback finishes, at
3530 * inode.c:btrfs_finish_ordered_io(), regardless of it finishing
3531 * with or without an error.
3532 */
3533 release_metadata = false;
3534 ret = btrfs_wait_cache_io(root, trans, NULL, &io_ctl, path, 0);
3535 }
3536
3537 if (ret) {
3538 if (release_metadata)
3539 btrfs_delalloc_release_metadata(inode, inode->i_size);
3540 #ifdef DEBUG
3541 btrfs_err(root->fs_info,
3542 "failed to write free ino cache for root %llu",
3543 root->root_key.objectid);
3544 #endif
3545 }
3546
3547 return ret;
3548 }
3549
3550 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
3551 /*
3552 * Use this if you need to make a bitmap or extent entry specifically, it
3553 * doesn't do any of the merging that add_free_space does, this acts a lot like
3554 * how the free space cache loading stuff works, so you can get really weird
3555 * configurations.
3556 */
3557 int test_add_free_space_entry(struct btrfs_block_group_cache *cache,
3558 u64 offset, u64 bytes, bool bitmap)
3559 {
3560 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3561 struct btrfs_free_space *info = NULL, *bitmap_info;
3562 void *map = NULL;
3563 u64 bytes_added;
3564 int ret;
3565
3566 again:
3567 if (!info) {
3568 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
3569 if (!info)
3570 return -ENOMEM;
3571 }
3572
3573 if (!bitmap) {
3574 spin_lock(&ctl->tree_lock);
3575 info->offset = offset;
3576 info->bytes = bytes;
3577 info->max_extent_size = 0;
3578 ret = link_free_space(ctl, info);
3579 spin_unlock(&ctl->tree_lock);
3580 if (ret)
3581 kmem_cache_free(btrfs_free_space_cachep, info);
3582 return ret;
3583 }
3584
3585 if (!map) {
3586 map = kzalloc(PAGE_SIZE, GFP_NOFS);
3587 if (!map) {
3588 kmem_cache_free(btrfs_free_space_cachep, info);
3589 return -ENOMEM;
3590 }
3591 }
3592
3593 spin_lock(&ctl->tree_lock);
3594 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3595 1, 0);
3596 if (!bitmap_info) {
3597 info->bitmap = map;
3598 map = NULL;
3599 add_new_bitmap(ctl, info, offset);
3600 bitmap_info = info;
3601 info = NULL;
3602 }
3603
3604 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
3605
3606 bytes -= bytes_added;
3607 offset += bytes_added;
3608 spin_unlock(&ctl->tree_lock);
3609
3610 if (bytes)
3611 goto again;
3612
3613 if (info)
3614 kmem_cache_free(btrfs_free_space_cachep, info);
3615 if (map)
3616 kfree(map);
3617 return 0;
3618 }
3619
3620 /*
3621 * Checks to see if the given range is in the free space cache. This is really
3622 * just used to check the absence of space, so if there is free space in the
3623 * range at all we will return 1.
3624 */
3625 int test_check_exists(struct btrfs_block_group_cache *cache,
3626 u64 offset, u64 bytes)
3627 {
3628 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3629 struct btrfs_free_space *info;
3630 int ret = 0;
3631
3632 spin_lock(&ctl->tree_lock);
3633 info = tree_search_offset(ctl, offset, 0, 0);
3634 if (!info) {
3635 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3636 1, 0);
3637 if (!info)
3638 goto out;
3639 }
3640
3641 have_info:
3642 if (info->bitmap) {
3643 u64 bit_off, bit_bytes;
3644 struct rb_node *n;
3645 struct btrfs_free_space *tmp;
3646
3647 bit_off = offset;
3648 bit_bytes = ctl->unit;
3649 ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
3650 if (!ret) {
3651 if (bit_off == offset) {
3652 ret = 1;
3653 goto out;
3654 } else if (bit_off > offset &&
3655 offset + bytes > bit_off) {
3656 ret = 1;
3657 goto out;
3658 }
3659 }
3660
3661 n = rb_prev(&info->offset_index);
3662 while (n) {
3663 tmp = rb_entry(n, struct btrfs_free_space,
3664 offset_index);
3665 if (tmp->offset + tmp->bytes < offset)
3666 break;
3667 if (offset + bytes < tmp->offset) {
3668 n = rb_prev(&tmp->offset_index);
3669 continue;
3670 }
3671 info = tmp;
3672 goto have_info;
3673 }
3674
3675 n = rb_next(&info->offset_index);
3676 while (n) {
3677 tmp = rb_entry(n, struct btrfs_free_space,
3678 offset_index);
3679 if (offset + bytes < tmp->offset)
3680 break;
3681 if (tmp->offset + tmp->bytes < offset) {
3682 n = rb_next(&tmp->offset_index);
3683 continue;
3684 }
3685 info = tmp;
3686 goto have_info;
3687 }
3688
3689 ret = 0;
3690 goto out;
3691 }
3692
3693 if (info->offset == offset) {
3694 ret = 1;
3695 goto out;
3696 }
3697
3698 if (offset > info->offset && offset < info->offset + info->bytes)
3699 ret = 1;
3700 out:
3701 spin_unlock(&ctl->tree_lock);
3702 return ret;
3703 }
3704 #endif /* CONFIG_BTRFS_FS_RUN_SANITY_TESTS */
(deprecated) Btrfs devel tree