?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
gpio: rcar: Fix runtime PM imbalance on error
[linux/fpc-iii.git] / fs / btrfs / free-space-cache.c
blob3613da065a737dca94a3375373b5b8699bc0ad8f
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * Copyright (C) 2008 Red Hat. All rights reserved.
4 */
5
6 #include <linux/pagemap.h>
7 #include <linux/sched.h>
8 #include <linux/sched/signal.h>
9 #include <linux/slab.h>
10 #include <linux/math64.h>
11 #include <linux/ratelimit.h>
12 #include <linux/error-injection.h>
13 #include <linux/sched/mm.h>
14 #include "ctree.h"
15 #include "free-space-cache.h"
16 #include "transaction.h"
17 #include "disk-io.h"
18 #include "extent_io.h"
19 #include "inode-map.h"
20 #include "volumes.h"
21 #include "space-info.h"
22 #include "delalloc-space.h"
23 #include "block-group.h"
24 #include "discard.h"
25
26 #define BITS_PER_BITMAP (PAGE_SIZE * 8UL)
27 #define MAX_CACHE_BYTES_PER_GIG SZ_64K
28 #define FORCE_EXTENT_THRESHOLD SZ_1M
29
30 struct btrfs_trim_range {
31 u64 start;
32 u64 bytes;
33 struct list_head list;
34 };
35
36 static int count_bitmap_extents(struct btrfs_free_space_ctl *ctl,
37 struct btrfs_free_space *bitmap_info);
38 static int link_free_space(struct btrfs_free_space_ctl *ctl,
39 struct btrfs_free_space *info);
40 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
41 struct btrfs_free_space *info);
42 static int btrfs_wait_cache_io_root(struct btrfs_root *root,
43 struct btrfs_trans_handle *trans,
44 struct btrfs_io_ctl *io_ctl,
45 struct btrfs_path *path);
46
47 static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
48 struct btrfs_path *path,
49 u64 offset)
50 {
51 struct btrfs_fs_info *fs_info = root->fs_info;
52 struct btrfs_key key;
53 struct btrfs_key location;
54 struct btrfs_disk_key disk_key;
55 struct btrfs_free_space_header *header;
56 struct extent_buffer *leaf;
57 struct inode *inode = NULL;
58 unsigned nofs_flag;
59 int ret;
60
61 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
62 key.offset = offset;
63 key.type = 0;
64
65 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
66 if (ret < 0)
67 return ERR_PTR(ret);
68 if (ret > 0) {
69 btrfs_release_path(path);
70 return ERR_PTR(-ENOENT);
71 }
72
73 leaf = path->nodes[0];
74 header = btrfs_item_ptr(leaf, path->slots[0],
75 struct btrfs_free_space_header);
76 btrfs_free_space_key(leaf, header, &disk_key);
77 btrfs_disk_key_to_cpu(&location, &disk_key);
78 btrfs_release_path(path);
79
80 /*
81 * We are often under a trans handle at this point, so we need to make
82 * sure NOFS is set to keep us from deadlocking.
83 */
84 nofs_flag = memalloc_nofs_save();
85 inode = btrfs_iget_path(fs_info->sb, &location, root, path);
86 btrfs_release_path(path);
87 memalloc_nofs_restore(nofs_flag);
88 if (IS_ERR(inode))
89 return inode;
90
91 mapping_set_gfp_mask(inode->i_mapping,
92 mapping_gfp_constraint(inode->i_mapping,
93 ~(__GFP_FS | __GFP_HIGHMEM)));
94
95 return inode;
96 }
97
98 struct inode *lookup_free_space_inode(struct btrfs_block_group *block_group,
99 struct btrfs_path *path)
100 {
101 struct btrfs_fs_info *fs_info = block_group->fs_info;
102 struct inode *inode = NULL;
103 u32 flags = BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
104
105 spin_lock(&block_group->lock);
106 if (block_group->inode)
107 inode = igrab(block_group->inode);
108 spin_unlock(&block_group->lock);
109 if (inode)
110 return inode;
111
112 inode = __lookup_free_space_inode(fs_info->tree_root, path,
113 block_group->start);
114 if (IS_ERR(inode))
115 return inode;
116
117 spin_lock(&block_group->lock);
118 if (!((BTRFS_I(inode)->flags & flags) == flags)) {
119 btrfs_info(fs_info, "Old style space inode found, converting.");
120 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM |
121 BTRFS_INODE_NODATACOW;
122 block_group->disk_cache_state = BTRFS_DC_CLEAR;
123 }
124
125 if (!block_group->iref) {
126 block_group->inode = igrab(inode);
127 block_group->iref = 1;
128 }
129 spin_unlock(&block_group->lock);
130
131 return inode;
132 }
133
134 static int __create_free_space_inode(struct btrfs_root *root,
135 struct btrfs_trans_handle *trans,
136 struct btrfs_path *path,
137 u64 ino, u64 offset)
138 {
139 struct btrfs_key key;
140 struct btrfs_disk_key disk_key;
141 struct btrfs_free_space_header *header;
142 struct btrfs_inode_item *inode_item;
143 struct extent_buffer *leaf;
144 u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC;
145 int ret;
146
147 ret = btrfs_insert_empty_inode(trans, root, path, ino);
148 if (ret)
149 return ret;
150
151 /* We inline crc's for the free disk space cache */
152 if (ino != BTRFS_FREE_INO_OBJECTID)
153 flags |= BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
154
155 leaf = path->nodes[0];
156 inode_item = btrfs_item_ptr(leaf, path->slots[0],
157 struct btrfs_inode_item);
158 btrfs_item_key(leaf, &disk_key, path->slots[0]);
159 memzero_extent_buffer(leaf, (unsigned long)inode_item,
160 sizeof(*inode_item));
161 btrfs_set_inode_generation(leaf, inode_item, trans->transid);
162 btrfs_set_inode_size(leaf, inode_item, 0);
163 btrfs_set_inode_nbytes(leaf, inode_item, 0);
164 btrfs_set_inode_uid(leaf, inode_item, 0);
165 btrfs_set_inode_gid(leaf, inode_item, 0);
166 btrfs_set_inode_mode(leaf, inode_item, S_IFREG | 0600);
167 btrfs_set_inode_flags(leaf, inode_item, flags);
168 btrfs_set_inode_nlink(leaf, inode_item, 1);
169 btrfs_set_inode_transid(leaf, inode_item, trans->transid);
170 btrfs_set_inode_block_group(leaf, inode_item, offset);
171 btrfs_mark_buffer_dirty(leaf);
172 btrfs_release_path(path);
173
174 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
175 key.offset = offset;
176 key.type = 0;
177 ret = btrfs_insert_empty_item(trans, root, path, &key,
178 sizeof(struct btrfs_free_space_header));
179 if (ret < 0) {
180 btrfs_release_path(path);
181 return ret;
182 }
183
184 leaf = path->nodes[0];
185 header = btrfs_item_ptr(leaf, path->slots[0],
186 struct btrfs_free_space_header);
187 memzero_extent_buffer(leaf, (unsigned long)header, sizeof(*header));
188 btrfs_set_free_space_key(leaf, header, &disk_key);
189 btrfs_mark_buffer_dirty(leaf);
190 btrfs_release_path(path);
191
192 return 0;
193 }
194
195 int create_free_space_inode(struct btrfs_trans_handle *trans,
196 struct btrfs_block_group *block_group,
197 struct btrfs_path *path)
198 {
199 int ret;
200 u64 ino;
201
202 ret = btrfs_find_free_objectid(trans->fs_info->tree_root, &ino);
203 if (ret < 0)
204 return ret;
205
206 return __create_free_space_inode(trans->fs_info->tree_root, trans, path,
207 ino, block_group->start);
208 }
209
210 int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
211 struct btrfs_block_rsv *rsv)
212 {
213 u64 needed_bytes;
214 int ret;
215
216 /* 1 for slack space, 1 for updating the inode */
217 needed_bytes = btrfs_calc_insert_metadata_size(fs_info, 1) +
218 btrfs_calc_metadata_size(fs_info, 1);
219
220 spin_lock(&rsv->lock);
221 if (rsv->reserved < needed_bytes)
222 ret = -ENOSPC;
223 else
224 ret = 0;
225 spin_unlock(&rsv->lock);
226 return ret;
227 }
228
229 int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
230 struct btrfs_block_group *block_group,
231 struct inode *inode)
232 {
233 struct btrfs_root *root = BTRFS_I(inode)->root;
234 int ret = 0;
235 bool locked = false;
236
237 if (block_group) {
238 struct btrfs_path *path = btrfs_alloc_path();
239
240 if (!path) {
241 ret = -ENOMEM;
242 goto fail;
243 }
244 locked = true;
245 mutex_lock(&trans->transaction->cache_write_mutex);
246 if (!list_empty(&block_group->io_list)) {
247 list_del_init(&block_group->io_list);
248
249 btrfs_wait_cache_io(trans, block_group, path);
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 btrfs_free_path(path);
261 }
262
263 btrfs_i_size_write(BTRFS_I(inode), 0);
264 truncate_pagecache(inode, 0);
265
266 /*
267 * We skip the throttling logic for free space cache inodes, so we don't
268 * need to check for -EAGAIN.
269 */
270 ret = btrfs_truncate_inode_items(trans, root, inode,
271 0, BTRFS_EXTENT_DATA_KEY);
272 if (ret)
273 goto fail;
274
275 ret = btrfs_update_inode(trans, root, inode);
276
277 fail:
278 if (locked)
279 mutex_unlock(&trans->transaction->cache_write_mutex);
280 if (ret)
281 btrfs_abort_transaction(trans, ret);
282
283 return ret;
284 }
285
286 static void readahead_cache(struct inode *inode)
287 {
288 struct file_ra_state *ra;
289 unsigned long last_index;
290
291 ra = kzalloc(sizeof(*ra), GFP_NOFS);
292 if (!ra)
293 return;
294
295 file_ra_state_init(ra, inode->i_mapping);
296 last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
297
298 page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
299
300 kfree(ra);
301 }
302
303 static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
304 int write)
305 {
306 int num_pages;
307 int check_crcs = 0;
308
309 num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
310
311 if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FREE_INO_OBJECTID)
312 check_crcs = 1;
313
314 /* Make sure we can fit our crcs and generation into the first page */
315 if (write && check_crcs &&
316 (num_pages * sizeof(u32) + sizeof(u64)) > PAGE_SIZE)
317 return -ENOSPC;
318
319 memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
320
321 io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
322 if (!io_ctl->pages)
323 return -ENOMEM;
324
325 io_ctl->num_pages = num_pages;
326 io_ctl->fs_info = btrfs_sb(inode->i_sb);
327 io_ctl->check_crcs = check_crcs;
328 io_ctl->inode = inode;
329
330 return 0;
331 }
332 ALLOW_ERROR_INJECTION(io_ctl_init, ERRNO);
333
334 static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
335 {
336 kfree(io_ctl->pages);
337 io_ctl->pages = NULL;
338 }
339
340 static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
341 {
342 if (io_ctl->cur) {
343 io_ctl->cur = NULL;
344 io_ctl->orig = NULL;
345 }
346 }
347
348 static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
349 {
350 ASSERT(io_ctl->index < io_ctl->num_pages);
351 io_ctl->page = io_ctl->pages[io_ctl->index++];
352 io_ctl->cur = page_address(io_ctl->page);
353 io_ctl->orig = io_ctl->cur;
354 io_ctl->size = PAGE_SIZE;
355 if (clear)
356 clear_page(io_ctl->cur);
357 }
358
359 static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
360 {
361 int i;
362
363 io_ctl_unmap_page(io_ctl);
364
365 for (i = 0; i < io_ctl->num_pages; i++) {
366 if (io_ctl->pages[i]) {
367 ClearPageChecked(io_ctl->pages[i]);
368 unlock_page(io_ctl->pages[i]);
369 put_page(io_ctl->pages[i]);
370 }
371 }
372 }
373
374 static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, bool uptodate)
375 {
376 struct page *page;
377 struct inode *inode = io_ctl->inode;
378 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
379 int i;
380
381 for (i = 0; i < io_ctl->num_pages; i++) {
382 page = find_or_create_page(inode->i_mapping, i, mask);
383 if (!page) {
384 io_ctl_drop_pages(io_ctl);
385 return -ENOMEM;
386 }
387 io_ctl->pages[i] = page;
388 if (uptodate && !PageUptodate(page)) {
389 btrfs_readpage(NULL, page);
390 lock_page(page);
391 if (page->mapping != inode->i_mapping) {
392 btrfs_err(BTRFS_I(inode)->root->fs_info,
393 "free space cache page truncated");
394 io_ctl_drop_pages(io_ctl);
395 return -EIO;
396 }
397 if (!PageUptodate(page)) {
398 btrfs_err(BTRFS_I(inode)->root->fs_info,
399 "error reading free space cache");
400 io_ctl_drop_pages(io_ctl);
401 return -EIO;
402 }
403 }
404 }
405
406 for (i = 0; i < io_ctl->num_pages; i++) {
407 clear_page_dirty_for_io(io_ctl->pages[i]);
408 set_page_extent_mapped(io_ctl->pages[i]);
409 }
410
411 return 0;
412 }
413
414 static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
415 {
416 __le64 *val;
417
418 io_ctl_map_page(io_ctl, 1);
419
420 /*
421 * Skip the csum areas. If we don't check crcs then we just have a
422 * 64bit chunk at the front of the first page.
423 */
424 if (io_ctl->check_crcs) {
425 io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
426 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
427 } else {
428 io_ctl->cur += sizeof(u64);
429 io_ctl->size -= sizeof(u64) * 2;
430 }
431
432 val = io_ctl->cur;
433 *val = cpu_to_le64(generation);
434 io_ctl->cur += sizeof(u64);
435 }
436
437 static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
438 {
439 __le64 *gen;
440
441 /*
442 * Skip the crc area. If we don't check crcs then we just have a 64bit
443 * chunk at the front of the first page.
444 */
445 if (io_ctl->check_crcs) {
446 io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
447 io_ctl->size -= sizeof(u64) +
448 (sizeof(u32) * io_ctl->num_pages);
449 } else {
450 io_ctl->cur += sizeof(u64);
451 io_ctl->size -= sizeof(u64) * 2;
452 }
453
454 gen = io_ctl->cur;
455 if (le64_to_cpu(*gen) != generation) {
456 btrfs_err_rl(io_ctl->fs_info,
457 "space cache generation (%llu) does not match inode (%llu)",
458 *gen, generation);
459 io_ctl_unmap_page(io_ctl);
460 return -EIO;
461 }
462 io_ctl->cur += sizeof(u64);
463 return 0;
464 }
465
466 static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
467 {
468 u32 *tmp;
469 u32 crc = ~(u32)0;
470 unsigned offset = 0;
471
472 if (!io_ctl->check_crcs) {
473 io_ctl_unmap_page(io_ctl);
474 return;
475 }
476
477 if (index == 0)
478 offset = sizeof(u32) * io_ctl->num_pages;
479
480 crc = btrfs_crc32c(crc, io_ctl->orig + offset, PAGE_SIZE - offset);
481 btrfs_crc32c_final(crc, (u8 *)&crc);
482 io_ctl_unmap_page(io_ctl);
483 tmp = page_address(io_ctl->pages[0]);
484 tmp += index;
485 *tmp = crc;
486 }
487
488 static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
489 {
490 u32 *tmp, val;
491 u32 crc = ~(u32)0;
492 unsigned offset = 0;
493
494 if (!io_ctl->check_crcs) {
495 io_ctl_map_page(io_ctl, 0);
496 return 0;
497 }
498
499 if (index == 0)
500 offset = sizeof(u32) * io_ctl->num_pages;
501
502 tmp = page_address(io_ctl->pages[0]);
503 tmp += index;
504 val = *tmp;
505
506 io_ctl_map_page(io_ctl, 0);
507 crc = btrfs_crc32c(crc, io_ctl->orig + offset, PAGE_SIZE - offset);
508 btrfs_crc32c_final(crc, (u8 *)&crc);
509 if (val != crc) {
510 btrfs_err_rl(io_ctl->fs_info,
511 "csum mismatch on free space cache");
512 io_ctl_unmap_page(io_ctl);
513 return -EIO;
514 }
515
516 return 0;
517 }
518
519 static int io_ctl_add_entry(struct btrfs_io_ctl *io_ctl, u64 offset, u64 bytes,
520 void *bitmap)
521 {
522 struct btrfs_free_space_entry *entry;
523
524 if (!io_ctl->cur)
525 return -ENOSPC;
526
527 entry = io_ctl->cur;
528 entry->offset = cpu_to_le64(offset);
529 entry->bytes = cpu_to_le64(bytes);
530 entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
531 BTRFS_FREE_SPACE_EXTENT;
532 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
533 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
534
535 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
536 return 0;
537
538 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
539
540 /* No more pages to map */
541 if (io_ctl->index >= io_ctl->num_pages)
542 return 0;
543
544 /* map the next page */
545 io_ctl_map_page(io_ctl, 1);
546 return 0;
547 }
548
549 static int io_ctl_add_bitmap(struct btrfs_io_ctl *io_ctl, void *bitmap)
550 {
551 if (!io_ctl->cur)
552 return -ENOSPC;
553
554 /*
555 * If we aren't at the start of the current page, unmap this one and
556 * map the next one if there is any left.
557 */
558 if (io_ctl->cur != io_ctl->orig) {
559 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
560 if (io_ctl->index >= io_ctl->num_pages)
561 return -ENOSPC;
562 io_ctl_map_page(io_ctl, 0);
563 }
564
565 copy_page(io_ctl->cur, bitmap);
566 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
567 if (io_ctl->index < io_ctl->num_pages)
568 io_ctl_map_page(io_ctl, 0);
569 return 0;
570 }
571
572 static void io_ctl_zero_remaining_pages(struct btrfs_io_ctl *io_ctl)
573 {
574 /*
575 * If we're not on the boundary we know we've modified the page and we
576 * need to crc the page.
577 */
578 if (io_ctl->cur != io_ctl->orig)
579 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
580 else
581 io_ctl_unmap_page(io_ctl);
582
583 while (io_ctl->index < io_ctl->num_pages) {
584 io_ctl_map_page(io_ctl, 1);
585 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
586 }
587 }
588
589 static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
590 struct btrfs_free_space *entry, u8 *type)
591 {
592 struct btrfs_free_space_entry *e;
593 int ret;
594
595 if (!io_ctl->cur) {
596 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
597 if (ret)
598 return ret;
599 }
600
601 e = io_ctl->cur;
602 entry->offset = le64_to_cpu(e->offset);
603 entry->bytes = le64_to_cpu(e->bytes);
604 *type = e->type;
605 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
606 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
607
608 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
609 return 0;
610
611 io_ctl_unmap_page(io_ctl);
612
613 return 0;
614 }
615
616 static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
617 struct btrfs_free_space *entry)
618 {
619 int ret;
620
621 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
622 if (ret)
623 return ret;
624
625 copy_page(entry->bitmap, io_ctl->cur);
626 io_ctl_unmap_page(io_ctl);
627
628 return 0;
629 }
630
631 /*
632 * Since we attach pinned extents after the fact we can have contiguous sections
633 * of free space that are split up in entries. This poses a problem with the
634 * tree logging stuff since it could have allocated across what appears to be 2
635 * entries since we would have merged the entries when adding the pinned extents
636 * back to the free space cache. So run through the space cache that we just
637 * loaded and merge contiguous entries. This will make the log replay stuff not
638 * blow up and it will make for nicer allocator behavior.
639 */
640 static void merge_space_tree(struct btrfs_free_space_ctl *ctl)
641 {
642 struct btrfs_free_space *e, *prev = NULL;
643 struct rb_node *n;
644
645 again:
646 spin_lock(&ctl->tree_lock);
647 for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
648 e = rb_entry(n, struct btrfs_free_space, offset_index);
649 if (!prev)
650 goto next;
651 if (e->bitmap || prev->bitmap)
652 goto next;
653 if (prev->offset + prev->bytes == e->offset) {
654 unlink_free_space(ctl, prev);
655 unlink_free_space(ctl, e);
656 prev->bytes += e->bytes;
657 kmem_cache_free(btrfs_free_space_cachep, e);
658 link_free_space(ctl, prev);
659 prev = NULL;
660 spin_unlock(&ctl->tree_lock);
661 goto again;
662 }
663 next:
664 prev = e;
665 }
666 spin_unlock(&ctl->tree_lock);
667 }
668
669 static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
670 struct btrfs_free_space_ctl *ctl,
671 struct btrfs_path *path, u64 offset)
672 {
673 struct btrfs_fs_info *fs_info = root->fs_info;
674 struct btrfs_free_space_header *header;
675 struct extent_buffer *leaf;
676 struct btrfs_io_ctl io_ctl;
677 struct btrfs_key key;
678 struct btrfs_free_space *e, *n;
679 LIST_HEAD(bitmaps);
680 u64 num_entries;
681 u64 num_bitmaps;
682 u64 generation;
683 u8 type;
684 int ret = 0;
685
686 /* Nothing in the space cache, goodbye */
687 if (!i_size_read(inode))
688 return 0;
689
690 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
691 key.offset = offset;
692 key.type = 0;
693
694 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
695 if (ret < 0)
696 return 0;
697 else if (ret > 0) {
698 btrfs_release_path(path);
699 return 0;
700 }
701
702 ret = -1;
703
704 leaf = path->nodes[0];
705 header = btrfs_item_ptr(leaf, path->slots[0],
706 struct btrfs_free_space_header);
707 num_entries = btrfs_free_space_entries(leaf, header);
708 num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
709 generation = btrfs_free_space_generation(leaf, header);
710 btrfs_release_path(path);
711
712 if (!BTRFS_I(inode)->generation) {
713 btrfs_info(fs_info,
714 "the free space cache file (%llu) is invalid, skip it",
715 offset);
716 return 0;
717 }
718
719 if (BTRFS_I(inode)->generation != generation) {
720 btrfs_err(fs_info,
721 "free space inode generation (%llu) did not match free space cache generation (%llu)",
722 BTRFS_I(inode)->generation, generation);
723 return 0;
724 }
725
726 if (!num_entries)
727 return 0;
728
729 ret = io_ctl_init(&io_ctl, inode, 0);
730 if (ret)
731 return ret;
732
733 readahead_cache(inode);
734
735 ret = io_ctl_prepare_pages(&io_ctl, true);
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 /*
760 * Sync discard ensures that the free space cache is always
761 * trimmed. So when reading this in, the state should reflect
762 * that. We also do this for async as a stop gap for lack of
763 * persistence.
764 */
765 if (btrfs_test_opt(fs_info, DISCARD_SYNC) ||
766 btrfs_test_opt(fs_info, DISCARD_ASYNC))
767 e->trim_state = BTRFS_TRIM_STATE_TRIMMED;
768
769 if (!e->bytes) {
770 kmem_cache_free(btrfs_free_space_cachep, e);
771 goto free_cache;
772 }
773
774 if (type == BTRFS_FREE_SPACE_EXTENT) {
775 spin_lock(&ctl->tree_lock);
776 ret = link_free_space(ctl, e);
777 spin_unlock(&ctl->tree_lock);
778 if (ret) {
779 btrfs_err(fs_info,
780 "Duplicate entries in free space cache, dumping");
781 kmem_cache_free(btrfs_free_space_cachep, e);
782 goto free_cache;
783 }
784 } else {
785 ASSERT(num_bitmaps);
786 num_bitmaps--;
787 e->bitmap = kmem_cache_zalloc(
788 btrfs_free_space_bitmap_cachep, GFP_NOFS);
789 if (!e->bitmap) {
790 kmem_cache_free(
791 btrfs_free_space_cachep, e);
792 goto free_cache;
793 }
794 spin_lock(&ctl->tree_lock);
795 ret = link_free_space(ctl, e);
796 ctl->total_bitmaps++;
797 ctl->op->recalc_thresholds(ctl);
798 spin_unlock(&ctl->tree_lock);
799 if (ret) {
800 btrfs_err(fs_info,
801 "Duplicate entries in free space cache, dumping");
802 kmem_cache_free(btrfs_free_space_cachep, e);
803 goto free_cache;
804 }
805 list_add_tail(&e->list, &bitmaps);
806 }
807
808 num_entries--;
809 }
810
811 io_ctl_unmap_page(&io_ctl);
812
813 /*
814 * We add the bitmaps at the end of the entries in order that
815 * the bitmap entries are added to the cache.
816 */
817 list_for_each_entry_safe(e, n, &bitmaps, list) {
818 list_del_init(&e->list);
819 ret = io_ctl_read_bitmap(&io_ctl, e);
820 if (ret)
821 goto free_cache;
822 e->bitmap_extents = count_bitmap_extents(ctl, e);
823 if (!btrfs_free_space_trimmed(e)) {
824 ctl->discardable_extents[BTRFS_STAT_CURR] +=
825 e->bitmap_extents;
826 ctl->discardable_bytes[BTRFS_STAT_CURR] += e->bytes;
827 }
828 }
829
830 io_ctl_drop_pages(&io_ctl);
831 merge_space_tree(ctl);
832 ret = 1;
833 out:
834 btrfs_discard_update_discardable(ctl->private, ctl);
835 io_ctl_free(&io_ctl);
836 return ret;
837 free_cache:
838 io_ctl_drop_pages(&io_ctl);
839 __btrfs_remove_free_space_cache(ctl);
840 goto out;
841 }
842
843 int load_free_space_cache(struct btrfs_block_group *block_group)
844 {
845 struct btrfs_fs_info *fs_info = block_group->fs_info;
846 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
847 struct inode *inode;
848 struct btrfs_path *path;
849 int ret = 0;
850 bool matched;
851 u64 used = block_group->used;
852
853 /*
854 * If this block group has been marked to be cleared for one reason or
855 * another then we can't trust the on disk cache, so just return.
856 */
857 spin_lock(&block_group->lock);
858 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
859 spin_unlock(&block_group->lock);
860 return 0;
861 }
862 spin_unlock(&block_group->lock);
863
864 path = btrfs_alloc_path();
865 if (!path)
866 return 0;
867 path->search_commit_root = 1;
868 path->skip_locking = 1;
869
870 /*
871 * We must pass a path with search_commit_root set to btrfs_iget in
872 * order to avoid a deadlock when allocating extents for the tree root.
873 *
874 * When we are COWing an extent buffer from the tree root, when looking
875 * for a free extent, at extent-tree.c:find_free_extent(), we can find
876 * block group without its free space cache loaded. When we find one
877 * we must load its space cache which requires reading its free space
878 * cache's inode item from the root tree. If this inode item is located
879 * in the same leaf that we started COWing before, then we end up in
880 * deadlock on the extent buffer (trying to read lock it when we
881 * previously write locked it).
882 *
883 * It's safe to read the inode item using the commit root because
884 * block groups, once loaded, stay in memory forever (until they are
885 * removed) as well as their space caches once loaded. New block groups
886 * once created get their ->cached field set to BTRFS_CACHE_FINISHED so
887 * we will never try to read their inode item while the fs is mounted.
888 */
889 inode = lookup_free_space_inode(block_group, path);
890 if (IS_ERR(inode)) {
891 btrfs_free_path(path);
892 return 0;
893 }
894
895 /* We may have converted the inode and made the cache invalid. */
896 spin_lock(&block_group->lock);
897 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
898 spin_unlock(&block_group->lock);
899 btrfs_free_path(path);
900 goto out;
901 }
902 spin_unlock(&block_group->lock);
903
904 ret = __load_free_space_cache(fs_info->tree_root, inode, ctl,
905 path, block_group->start);
906 btrfs_free_path(path);
907 if (ret <= 0)
908 goto out;
909
910 spin_lock(&ctl->tree_lock);
911 matched = (ctl->free_space == (block_group->length - used -
912 block_group->bytes_super));
913 spin_unlock(&ctl->tree_lock);
914
915 if (!matched) {
916 __btrfs_remove_free_space_cache(ctl);
917 btrfs_warn(fs_info,
918 "block group %llu has wrong amount of free space",
919 block_group->start);
920 ret = -1;
921 }
922 out:
923 if (ret < 0) {
924 /* This cache is bogus, make sure it gets cleared */
925 spin_lock(&block_group->lock);
926 block_group->disk_cache_state = BTRFS_DC_CLEAR;
927 spin_unlock(&block_group->lock);
928 ret = 0;
929
930 btrfs_warn(fs_info,
931 "failed to load free space cache for block group %llu, rebuilding it now",
932 block_group->start);
933 }
934
935 iput(inode);
936 return ret;
937 }
938
939 static noinline_for_stack
940 int write_cache_extent_entries(struct btrfs_io_ctl *io_ctl,
941 struct btrfs_free_space_ctl *ctl,
942 struct btrfs_block_group *block_group,
943 int *entries, int *bitmaps,
944 struct list_head *bitmap_list)
945 {
946 int ret;
947 struct btrfs_free_cluster *cluster = NULL;
948 struct btrfs_free_cluster *cluster_locked = NULL;
949 struct rb_node *node = rb_first(&ctl->free_space_offset);
950 struct btrfs_trim_range *trim_entry;
951
952 /* Get the cluster for this block_group if it exists */
953 if (block_group && !list_empty(&block_group->cluster_list)) {
954 cluster = list_entry(block_group->cluster_list.next,
955 struct btrfs_free_cluster,
956 block_group_list);
957 }
958
959 if (!node && cluster) {
960 cluster_locked = cluster;
961 spin_lock(&cluster_locked->lock);
962 node = rb_first(&cluster->root);
963 cluster = NULL;
964 }
965
966 /* Write out the extent entries */
967 while (node) {
968 struct btrfs_free_space *e;
969
970 e = rb_entry(node, struct btrfs_free_space, offset_index);
971 *entries += 1;
972
973 ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes,
974 e->bitmap);
975 if (ret)
976 goto fail;
977
978 if (e->bitmap) {
979 list_add_tail(&e->list, bitmap_list);
980 *bitmaps += 1;
981 }
982 node = rb_next(node);
983 if (!node && cluster) {
984 node = rb_first(&cluster->root);
985 cluster_locked = cluster;
986 spin_lock(&cluster_locked->lock);
987 cluster = NULL;
988 }
989 }
990 if (cluster_locked) {
991 spin_unlock(&cluster_locked->lock);
992 cluster_locked = NULL;
993 }
994
995 /*
996 * Make sure we don't miss any range that was removed from our rbtree
997 * because trimming is running. Otherwise after a umount+mount (or crash
998 * after committing the transaction) we would leak free space and get
999 * an inconsistent free space cache report from fsck.
1000 */
1001 list_for_each_entry(trim_entry, &ctl->trimming_ranges, list) {
1002 ret = io_ctl_add_entry(io_ctl, trim_entry->start,
1003 trim_entry->bytes, NULL);
1004 if (ret)
1005 goto fail;
1006 *entries += 1;
1007 }
1008
1009 return 0;
1010 fail:
1011 if (cluster_locked)
1012 spin_unlock(&cluster_locked->lock);
1013 return -ENOSPC;
1014 }
1015
1016 static noinline_for_stack int
1017 update_cache_item(struct btrfs_trans_handle *trans,
1018 struct btrfs_root *root,
1019 struct inode *inode,
1020 struct btrfs_path *path, u64 offset,
1021 int entries, int bitmaps)
1022 {
1023 struct btrfs_key key;
1024 struct btrfs_free_space_header *header;
1025 struct extent_buffer *leaf;
1026 int ret;
1027
1028 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
1029 key.offset = offset;
1030 key.type = 0;
1031
1032 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
1033 if (ret < 0) {
1034 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1035 EXTENT_DELALLOC, 0, 0, NULL);
1036 goto fail;
1037 }
1038 leaf = path->nodes[0];
1039 if (ret > 0) {
1040 struct btrfs_key found_key;
1041 ASSERT(path->slots[0]);
1042 path->slots[0]--;
1043 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1044 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
1045 found_key.offset != offset) {
1046 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
1047 inode->i_size - 1, EXTENT_DELALLOC, 0,
1048 0, NULL);
1049 btrfs_release_path(path);
1050 goto fail;
1051 }
1052 }
1053
1054 BTRFS_I(inode)->generation = trans->transid;
1055 header = btrfs_item_ptr(leaf, path->slots[0],
1056 struct btrfs_free_space_header);
1057 btrfs_set_free_space_entries(leaf, header, entries);
1058 btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1059 btrfs_set_free_space_generation(leaf, header, trans->transid);
1060 btrfs_mark_buffer_dirty(leaf);
1061 btrfs_release_path(path);
1062
1063 return 0;
1064
1065 fail:
1066 return -1;
1067 }
1068
1069 static noinline_for_stack int write_pinned_extent_entries(
1070 struct btrfs_trans_handle *trans,
1071 struct btrfs_block_group *block_group,
1072 struct btrfs_io_ctl *io_ctl,
1073 int *entries)
1074 {
1075 u64 start, extent_start, extent_end, len;
1076 struct extent_io_tree *unpin = NULL;
1077 int ret;
1078
1079 if (!block_group)
1080 return 0;
1081
1082 /*
1083 * We want to add any pinned extents to our free space cache
1084 * so we don't leak the space
1085 *
1086 * We shouldn't have switched the pinned extents yet so this is the
1087 * right one
1088 */
1089 unpin = &trans->transaction->pinned_extents;
1090
1091 start = block_group->start;
1092
1093 while (start < block_group->start + block_group->length) {
1094 ret = find_first_extent_bit(unpin, start,
1095 &extent_start, &extent_end,
1096 EXTENT_DIRTY, NULL);
1097 if (ret)
1098 return 0;
1099
1100 /* This pinned extent is out of our range */
1101 if (extent_start >= block_group->start + block_group->length)
1102 return 0;
1103
1104 extent_start = max(extent_start, start);
1105 extent_end = min(block_group->start + block_group->length,
1106 extent_end + 1);
1107 len = extent_end - extent_start;
1108
1109 *entries += 1;
1110 ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
1111 if (ret)
1112 return -ENOSPC;
1113
1114 start = extent_end;
1115 }
1116
1117 return 0;
1118 }
1119
1120 static noinline_for_stack int
1121 write_bitmap_entries(struct btrfs_io_ctl *io_ctl, struct list_head *bitmap_list)
1122 {
1123 struct btrfs_free_space *entry, *next;
1124 int ret;
1125
1126 /* Write out the bitmaps */
1127 list_for_each_entry_safe(entry, next, bitmap_list, list) {
1128 ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
1129 if (ret)
1130 return -ENOSPC;
1131 list_del_init(&entry->list);
1132 }
1133
1134 return 0;
1135 }
1136
1137 static int flush_dirty_cache(struct inode *inode)
1138 {
1139 int ret;
1140
1141 ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
1142 if (ret)
1143 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1144 EXTENT_DELALLOC, 0, 0, NULL);
1145
1146 return ret;
1147 }
1148
1149 static void noinline_for_stack
1150 cleanup_bitmap_list(struct list_head *bitmap_list)
1151 {
1152 struct btrfs_free_space *entry, *next;
1153
1154 list_for_each_entry_safe(entry, next, bitmap_list, list)
1155 list_del_init(&entry->list);
1156 }
1157
1158 static void noinline_for_stack
1159 cleanup_write_cache_enospc(struct inode *inode,
1160 struct btrfs_io_ctl *io_ctl,
1161 struct extent_state **cached_state)
1162 {
1163 io_ctl_drop_pages(io_ctl);
1164 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1165 i_size_read(inode) - 1, cached_state);
1166 }
1167
1168 static int __btrfs_wait_cache_io(struct btrfs_root *root,
1169 struct btrfs_trans_handle *trans,
1170 struct btrfs_block_group *block_group,
1171 struct btrfs_io_ctl *io_ctl,
1172 struct btrfs_path *path, u64 offset)
1173 {
1174 int ret;
1175 struct inode *inode = io_ctl->inode;
1176
1177 if (!inode)
1178 return 0;
1179
1180 /* Flush the dirty pages in the cache file. */
1181 ret = flush_dirty_cache(inode);
1182 if (ret)
1183 goto out;
1184
1185 /* Update the cache item to tell everyone this cache file is valid. */
1186 ret = update_cache_item(trans, root, inode, path, offset,
1187 io_ctl->entries, io_ctl->bitmaps);
1188 out:
1189 io_ctl_free(io_ctl);
1190 if (ret) {
1191 invalidate_inode_pages2(inode->i_mapping);
1192 BTRFS_I(inode)->generation = 0;
1193 if (block_group) {
1194 #ifdef CONFIG_BTRFS_DEBUG
1195 btrfs_err(root->fs_info,
1196 "failed to write free space cache for block group %llu",
1197 block_group->start);
1198 #endif
1199 }
1200 }
1201 btrfs_update_inode(trans, root, inode);
1202
1203 if (block_group) {
1204 /* the dirty list is protected by the dirty_bgs_lock */
1205 spin_lock(&trans->transaction->dirty_bgs_lock);
1206
1207 /* the disk_cache_state is protected by the block group lock */
1208 spin_lock(&block_group->lock);
1209
1210 /*
1211 * only mark this as written if we didn't get put back on
1212 * the dirty list while waiting for IO. Otherwise our
1213 * cache state won't be right, and we won't get written again
1214 */
1215 if (!ret && list_empty(&block_group->dirty_list))
1216 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1217 else if (ret)
1218 block_group->disk_cache_state = BTRFS_DC_ERROR;
1219
1220 spin_unlock(&block_group->lock);
1221 spin_unlock(&trans->transaction->dirty_bgs_lock);
1222 io_ctl->inode = NULL;
1223 iput(inode);
1224 }
1225
1226 return ret;
1227
1228 }
1229
1230 static int btrfs_wait_cache_io_root(struct btrfs_root *root,
1231 struct btrfs_trans_handle *trans,
1232 struct btrfs_io_ctl *io_ctl,
1233 struct btrfs_path *path)
1234 {
1235 return __btrfs_wait_cache_io(root, trans, NULL, io_ctl, path, 0);
1236 }
1237
1238 int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
1239 struct btrfs_block_group *block_group,
1240 struct btrfs_path *path)
1241 {
1242 return __btrfs_wait_cache_io(block_group->fs_info->tree_root, trans,
1243 block_group, &block_group->io_ctl,
1244 path, block_group->start);
1245 }
1246
1247 /**
1248 * __btrfs_write_out_cache - write out cached info to an inode
1249 * @root - the root the inode belongs to
1250 * @ctl - the free space cache we are going to write out
1251 * @block_group - the block_group for this cache if it belongs to a block_group
1252 * @trans - the trans handle
1253 *
1254 * This function writes out a free space cache struct to disk for quick recovery
1255 * on mount. This will return 0 if it was successful in writing the cache out,
1256 * or an errno if it was not.
1257 */
1258 static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
1259 struct btrfs_free_space_ctl *ctl,
1260 struct btrfs_block_group *block_group,
1261 struct btrfs_io_ctl *io_ctl,
1262 struct btrfs_trans_handle *trans)
1263 {
1264 struct extent_state *cached_state = NULL;
1265 LIST_HEAD(bitmap_list);
1266 int entries = 0;
1267 int bitmaps = 0;
1268 int ret;
1269 int must_iput = 0;
1270
1271 if (!i_size_read(inode))
1272 return -EIO;
1273
1274 WARN_ON(io_ctl->pages);
1275 ret = io_ctl_init(io_ctl, inode, 1);
1276 if (ret)
1277 return ret;
1278
1279 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
1280 down_write(&block_group->data_rwsem);
1281 spin_lock(&block_group->lock);
1282 if (block_group->delalloc_bytes) {
1283 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1284 spin_unlock(&block_group->lock);
1285 up_write(&block_group->data_rwsem);
1286 BTRFS_I(inode)->generation = 0;
1287 ret = 0;
1288 must_iput = 1;
1289 goto out;
1290 }
1291 spin_unlock(&block_group->lock);
1292 }
1293
1294 /* Lock all pages first so we can lock the extent safely. */
1295 ret = io_ctl_prepare_pages(io_ctl, false);
1296 if (ret)
1297 goto out_unlock;
1298
1299 lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1300 &cached_state);
1301
1302 io_ctl_set_generation(io_ctl, trans->transid);
1303
1304 mutex_lock(&ctl->cache_writeout_mutex);
1305 /* Write out the extent entries in the free space cache */
1306 spin_lock(&ctl->tree_lock);
1307 ret = write_cache_extent_entries(io_ctl, ctl,
1308 block_group, &entries, &bitmaps,
1309 &bitmap_list);
1310 if (ret)
1311 goto out_nospc_locked;
1312
1313 /*
1314 * Some spaces that are freed in the current transaction are pinned,
1315 * they will be added into free space cache after the transaction is
1316 * committed, we shouldn't lose them.
1317 *
1318 * If this changes while we are working we'll get added back to
1319 * the dirty list and redo it. No locking needed
1320 */
1321 ret = write_pinned_extent_entries(trans, block_group, io_ctl, &entries);
1322 if (ret)
1323 goto out_nospc_locked;
1324
1325 /*
1326 * At last, we write out all the bitmaps and keep cache_writeout_mutex
1327 * locked while doing it because a concurrent trim can be manipulating
1328 * or freeing the bitmap.
1329 */
1330 ret = write_bitmap_entries(io_ctl, &bitmap_list);
1331 spin_unlock(&ctl->tree_lock);
1332 mutex_unlock(&ctl->cache_writeout_mutex);
1333 if (ret)
1334 goto out_nospc;
1335
1336 /* Zero out the rest of the pages just to make sure */
1337 io_ctl_zero_remaining_pages(io_ctl);
1338
1339 /* Everything is written out, now we dirty the pages in the file. */
1340 ret = btrfs_dirty_pages(inode, io_ctl->pages, io_ctl->num_pages, 0,
1341 i_size_read(inode), &cached_state);
1342 if (ret)
1343 goto out_nospc;
1344
1345 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1346 up_write(&block_group->data_rwsem);
1347 /*
1348 * Release the pages and unlock the extent, we will flush
1349 * them out later
1350 */
1351 io_ctl_drop_pages(io_ctl);
1352
1353 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1354 i_size_read(inode) - 1, &cached_state);
1355
1356 /*
1357 * at this point the pages are under IO and we're happy,
1358 * The caller is responsible for waiting on them and updating the
1359 * the cache and the inode
1360 */
1361 io_ctl->entries = entries;
1362 io_ctl->bitmaps = bitmaps;
1363
1364 ret = btrfs_fdatawrite_range(inode, 0, (u64)-1);
1365 if (ret)
1366 goto out;
1367
1368 return 0;
1369
1370 out_nospc_locked:
1371 cleanup_bitmap_list(&bitmap_list);
1372 spin_unlock(&ctl->tree_lock);
1373 mutex_unlock(&ctl->cache_writeout_mutex);
1374
1375 out_nospc:
1376 cleanup_write_cache_enospc(inode, io_ctl, &cached_state);
1377
1378 out_unlock:
1379 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1380 up_write(&block_group->data_rwsem);
1381
1382 out:
1383 io_ctl->inode = NULL;
1384 io_ctl_free(io_ctl);
1385 if (ret) {
1386 invalidate_inode_pages2(inode->i_mapping);
1387 BTRFS_I(inode)->generation = 0;
1388 }
1389 btrfs_update_inode(trans, root, inode);
1390 if (must_iput)
1391 iput(inode);
1392 return ret;
1393 }
1394
1395 int btrfs_write_out_cache(struct btrfs_trans_handle *trans,
1396 struct btrfs_block_group *block_group,
1397 struct btrfs_path *path)
1398 {
1399 struct btrfs_fs_info *fs_info = trans->fs_info;
1400 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1401 struct inode *inode;
1402 int ret = 0;
1403
1404 spin_lock(&block_group->lock);
1405 if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1406 spin_unlock(&block_group->lock);
1407 return 0;
1408 }
1409 spin_unlock(&block_group->lock);
1410
1411 inode = lookup_free_space_inode(block_group, path);
1412 if (IS_ERR(inode))
1413 return 0;
1414
1415 ret = __btrfs_write_out_cache(fs_info->tree_root, inode, ctl,
1416 block_group, &block_group->io_ctl, trans);
1417 if (ret) {
1418 #ifdef CONFIG_BTRFS_DEBUG
1419 btrfs_err(fs_info,
1420 "failed to write free space cache for block group %llu",
1421 block_group->start);
1422 #endif
1423 spin_lock(&block_group->lock);
1424 block_group->disk_cache_state = BTRFS_DC_ERROR;
1425 spin_unlock(&block_group->lock);
1426
1427 block_group->io_ctl.inode = NULL;
1428 iput(inode);
1429 }
1430
1431 /*
1432 * if ret == 0 the caller is expected to call btrfs_wait_cache_io
1433 * to wait for IO and put the inode
1434 */
1435
1436 return ret;
1437 }
1438
1439 static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1440 u64 offset)
1441 {
1442 ASSERT(offset >= bitmap_start);
1443 offset -= bitmap_start;
1444 return (unsigned long)(div_u64(offset, unit));
1445 }
1446
1447 static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1448 {
1449 return (unsigned long)(div_u64(bytes, unit));
1450 }
1451
1452 static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1453 u64 offset)
1454 {
1455 u64 bitmap_start;
1456 u64 bytes_per_bitmap;
1457
1458 bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1459 bitmap_start = offset - ctl->start;
1460 bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1461 bitmap_start *= bytes_per_bitmap;
1462 bitmap_start += ctl->start;
1463
1464 return bitmap_start;
1465 }
1466
1467 static int tree_insert_offset(struct rb_root *root, u64 offset,
1468 struct rb_node *node, int bitmap)
1469 {
1470 struct rb_node **p = &root->rb_node;
1471 struct rb_node *parent = NULL;
1472 struct btrfs_free_space *info;
1473
1474 while (*p) {
1475 parent = *p;
1476 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1477
1478 if (offset < info->offset) {
1479 p = &(*p)->rb_left;
1480 } else if (offset > info->offset) {
1481 p = &(*p)->rb_right;
1482 } else {
1483 /*
1484 * we could have a bitmap entry and an extent entry
1485 * share the same offset. If this is the case, we want
1486 * the extent entry to always be found first if we do a
1487 * linear search through the tree, since we want to have
1488 * the quickest allocation time, and allocating from an
1489 * extent is faster than allocating from a bitmap. So
1490 * if we're inserting a bitmap and we find an entry at
1491 * this offset, we want to go right, or after this entry
1492 * logically. If we are inserting an extent and we've
1493 * found a bitmap, we want to go left, or before
1494 * logically.
1495 */
1496 if (bitmap) {
1497 if (info->bitmap) {
1498 WARN_ON_ONCE(1);
1499 return -EEXIST;
1500 }
1501 p = &(*p)->rb_right;
1502 } else {
1503 if (!info->bitmap) {
1504 WARN_ON_ONCE(1);
1505 return -EEXIST;
1506 }
1507 p = &(*p)->rb_left;
1508 }
1509 }
1510 }
1511
1512 rb_link_node(node, parent, p);
1513 rb_insert_color(node, root);
1514
1515 return 0;
1516 }
1517
1518 /*
1519 * searches the tree for the given offset.
1520 *
1521 * fuzzy - If this is set, then we are trying to make an allocation, and we just
1522 * want a section that has at least bytes size and comes at or after the given
1523 * offset.
1524 */
1525 static struct btrfs_free_space *
1526 tree_search_offset(struct btrfs_free_space_ctl *ctl,
1527 u64 offset, int bitmap_only, int fuzzy)
1528 {
1529 struct rb_node *n = ctl->free_space_offset.rb_node;
1530 struct btrfs_free_space *entry, *prev = NULL;
1531
1532 /* find entry that is closest to the 'offset' */
1533 while (1) {
1534 if (!n) {
1535 entry = NULL;
1536 break;
1537 }
1538
1539 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1540 prev = entry;
1541
1542 if (offset < entry->offset)
1543 n = n->rb_left;
1544 else if (offset > entry->offset)
1545 n = n->rb_right;
1546 else
1547 break;
1548 }
1549
1550 if (bitmap_only) {
1551 if (!entry)
1552 return NULL;
1553 if (entry->bitmap)
1554 return entry;
1555
1556 /*
1557 * bitmap entry and extent entry may share same offset,
1558 * in that case, bitmap entry comes after extent entry.
1559 */
1560 n = rb_next(n);
1561 if (!n)
1562 return NULL;
1563 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1564 if (entry->offset != offset)
1565 return NULL;
1566
1567 WARN_ON(!entry->bitmap);
1568 return entry;
1569 } else if (entry) {
1570 if (entry->bitmap) {
1571 /*
1572 * if previous extent entry covers the offset,
1573 * we should return it instead of the bitmap entry
1574 */
1575 n = rb_prev(&entry->offset_index);
1576 if (n) {
1577 prev = rb_entry(n, struct btrfs_free_space,
1578 offset_index);
1579 if (!prev->bitmap &&
1580 prev->offset + prev->bytes > offset)
1581 entry = prev;
1582 }
1583 }
1584 return entry;
1585 }
1586
1587 if (!prev)
1588 return NULL;
1589
1590 /* find last entry before the 'offset' */
1591 entry = prev;
1592 if (entry->offset > offset) {
1593 n = rb_prev(&entry->offset_index);
1594 if (n) {
1595 entry = rb_entry(n, struct btrfs_free_space,
1596 offset_index);
1597 ASSERT(entry->offset <= offset);
1598 } else {
1599 if (fuzzy)
1600 return entry;
1601 else
1602 return NULL;
1603 }
1604 }
1605
1606 if (entry->bitmap) {
1607 n = rb_prev(&entry->offset_index);
1608 if (n) {
1609 prev = rb_entry(n, struct btrfs_free_space,
1610 offset_index);
1611 if (!prev->bitmap &&
1612 prev->offset + prev->bytes > offset)
1613 return prev;
1614 }
1615 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1616 return entry;
1617 } else if (entry->offset + entry->bytes > offset)
1618 return entry;
1619
1620 if (!fuzzy)
1621 return NULL;
1622
1623 while (1) {
1624 if (entry->bitmap) {
1625 if (entry->offset + BITS_PER_BITMAP *
1626 ctl->unit > offset)
1627 break;
1628 } else {
1629 if (entry->offset + entry->bytes > offset)
1630 break;
1631 }
1632
1633 n = rb_next(&entry->offset_index);
1634 if (!n)
1635 return NULL;
1636 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1637 }
1638 return entry;
1639 }
1640
1641 static inline void
1642 __unlink_free_space(struct btrfs_free_space_ctl *ctl,
1643 struct btrfs_free_space *info)
1644 {
1645 rb_erase(&info->offset_index, &ctl->free_space_offset);
1646 ctl->free_extents--;
1647
1648 if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
1649 ctl->discardable_extents[BTRFS_STAT_CURR]--;
1650 ctl->discardable_bytes[BTRFS_STAT_CURR] -= info->bytes;
1651 }
1652 }
1653
1654 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1655 struct btrfs_free_space *info)
1656 {
1657 __unlink_free_space(ctl, info);
1658 ctl->free_space -= info->bytes;
1659 }
1660
1661 static int link_free_space(struct btrfs_free_space_ctl *ctl,
1662 struct btrfs_free_space *info)
1663 {
1664 int ret = 0;
1665
1666 ASSERT(info->bytes || info->bitmap);
1667 ret = tree_insert_offset(&ctl->free_space_offset, info->offset,
1668 &info->offset_index, (info->bitmap != NULL));
1669 if (ret)
1670 return ret;
1671
1672 if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
1673 ctl->discardable_extents[BTRFS_STAT_CURR]++;
1674 ctl->discardable_bytes[BTRFS_STAT_CURR] += info->bytes;
1675 }
1676
1677 ctl->free_space += info->bytes;
1678 ctl->free_extents++;
1679 return ret;
1680 }
1681
1682 static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
1683 {
1684 struct btrfs_block_group *block_group = ctl->private;
1685 u64 max_bytes;
1686 u64 bitmap_bytes;
1687 u64 extent_bytes;
1688 u64 size = block_group->length;
1689 u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
1690 u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
1691
1692 max_bitmaps = max_t(u64, max_bitmaps, 1);
1693
1694 ASSERT(ctl->total_bitmaps <= max_bitmaps);
1695
1696 /*
1697 * We are trying to keep the total amount of memory used per 1GiB of
1698 * space to be MAX_CACHE_BYTES_PER_GIG. However, with a reclamation
1699 * mechanism of pulling extents >= FORCE_EXTENT_THRESHOLD out of
1700 * bitmaps, we may end up using more memory than this.
1701 */
1702 if (size < SZ_1G)
1703 max_bytes = MAX_CACHE_BYTES_PER_GIG;
1704 else
1705 max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
1706
1707 bitmap_bytes = ctl->total_bitmaps * ctl->unit;
1708
1709 /*
1710 * we want the extent entry threshold to always be at most 1/2 the max
1711 * bytes we can have, or whatever is less than that.
1712 */
1713 extent_bytes = max_bytes - bitmap_bytes;
1714 extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
1715
1716 ctl->extents_thresh =
1717 div_u64(extent_bytes, sizeof(struct btrfs_free_space));
1718 }
1719
1720 static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1721 struct btrfs_free_space *info,
1722 u64 offset, u64 bytes)
1723 {
1724 unsigned long start, count, end;
1725 int extent_delta = -1;
1726
1727 start = offset_to_bit(info->offset, ctl->unit, offset);
1728 count = bytes_to_bits(bytes, ctl->unit);
1729 end = start + count;
1730 ASSERT(end <= BITS_PER_BITMAP);
1731
1732 bitmap_clear(info->bitmap, start, count);
1733
1734 info->bytes -= bytes;
1735 if (info->max_extent_size > ctl->unit)
1736 info->max_extent_size = 0;
1737
1738 if (start && test_bit(start - 1, info->bitmap))
1739 extent_delta++;
1740
1741 if (end < BITS_PER_BITMAP && test_bit(end, info->bitmap))
1742 extent_delta++;
1743
1744 info->bitmap_extents += extent_delta;
1745 if (!btrfs_free_space_trimmed(info)) {
1746 ctl->discardable_extents[BTRFS_STAT_CURR] += extent_delta;
1747 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
1748 }
1749 }
1750
1751 static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1752 struct btrfs_free_space *info, u64 offset,
1753 u64 bytes)
1754 {
1755 __bitmap_clear_bits(ctl, info, offset, bytes);
1756 ctl->free_space -= bytes;
1757 }
1758
1759 static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1760 struct btrfs_free_space *info, u64 offset,
1761 u64 bytes)
1762 {
1763 unsigned long start, count, end;
1764 int extent_delta = 1;
1765
1766 start = offset_to_bit(info->offset, ctl->unit, offset);
1767 count = bytes_to_bits(bytes, ctl->unit);
1768 end = start + count;
1769 ASSERT(end <= BITS_PER_BITMAP);
1770
1771 bitmap_set(info->bitmap, start, count);
1772
1773 info->bytes += bytes;
1774 ctl->free_space += bytes;
1775
1776 if (start && test_bit(start - 1, info->bitmap))
1777 extent_delta--;
1778
1779 if (end < BITS_PER_BITMAP && test_bit(end, info->bitmap))
1780 extent_delta--;
1781
1782 info->bitmap_extents += extent_delta;
1783 if (!btrfs_free_space_trimmed(info)) {
1784 ctl->discardable_extents[BTRFS_STAT_CURR] += extent_delta;
1785 ctl->discardable_bytes[BTRFS_STAT_CURR] += bytes;
1786 }
1787 }
1788
1789 /*
1790 * If we can not find suitable extent, we will use bytes to record
1791 * the size of the max extent.
1792 */
1793 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1794 struct btrfs_free_space *bitmap_info, u64 *offset,
1795 u64 *bytes, bool for_alloc)
1796 {
1797 unsigned long found_bits = 0;
1798 unsigned long max_bits = 0;
1799 unsigned long bits, i;
1800 unsigned long next_zero;
1801 unsigned long extent_bits;
1802
1803 /*
1804 * Skip searching the bitmap if we don't have a contiguous section that
1805 * is large enough for this allocation.
1806 */
1807 if (for_alloc &&
1808 bitmap_info->max_extent_size &&
1809 bitmap_info->max_extent_size < *bytes) {
1810 *bytes = bitmap_info->max_extent_size;
1811 return -1;
1812 }
1813
1814 i = offset_to_bit(bitmap_info->offset, ctl->unit,
1815 max_t(u64, *offset, bitmap_info->offset));
1816 bits = bytes_to_bits(*bytes, ctl->unit);
1817
1818 for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
1819 if (for_alloc && bits == 1) {
1820 found_bits = 1;
1821 break;
1822 }
1823 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1824 BITS_PER_BITMAP, i);
1825 extent_bits = next_zero - i;
1826 if (extent_bits >= bits) {
1827 found_bits = extent_bits;
1828 break;
1829 } else if (extent_bits > max_bits) {
1830 max_bits = extent_bits;
1831 }
1832 i = next_zero;
1833 }
1834
1835 if (found_bits) {
1836 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1837 *bytes = (u64)(found_bits) * ctl->unit;
1838 return 0;
1839 }
1840
1841 *bytes = (u64)(max_bits) * ctl->unit;
1842 bitmap_info->max_extent_size = *bytes;
1843 return -1;
1844 }
1845
1846 static inline u64 get_max_extent_size(struct btrfs_free_space *entry)
1847 {
1848 if (entry->bitmap)
1849 return entry->max_extent_size;
1850 return entry->bytes;
1851 }
1852
1853 /* Cache the size of the max extent in bytes */
1854 static struct btrfs_free_space *
1855 find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
1856 unsigned long align, u64 *max_extent_size)
1857 {
1858 struct btrfs_free_space *entry;
1859 struct rb_node *node;
1860 u64 tmp;
1861 u64 align_off;
1862 int ret;
1863
1864 if (!ctl->free_space_offset.rb_node)
1865 goto out;
1866
1867 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
1868 if (!entry)
1869 goto out;
1870
1871 for (node = &entry->offset_index; node; node = rb_next(node)) {
1872 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1873 if (entry->bytes < *bytes) {
1874 *max_extent_size = max(get_max_extent_size(entry),
1875 *max_extent_size);
1876 continue;
1877 }
1878
1879 /* make sure the space returned is big enough
1880 * to match our requested alignment
1881 */
1882 if (*bytes >= align) {
1883 tmp = entry->offset - ctl->start + align - 1;
1884 tmp = div64_u64(tmp, align);
1885 tmp = tmp * align + ctl->start;
1886 align_off = tmp - entry->offset;
1887 } else {
1888 align_off = 0;
1889 tmp = entry->offset;
1890 }
1891
1892 if (entry->bytes < *bytes + align_off) {
1893 *max_extent_size = max(get_max_extent_size(entry),
1894 *max_extent_size);
1895 continue;
1896 }
1897
1898 if (entry->bitmap) {
1899 u64 size = *bytes;
1900
1901 ret = search_bitmap(ctl, entry, &tmp, &size, true);
1902 if (!ret) {
1903 *offset = tmp;
1904 *bytes = size;
1905 return entry;
1906 } else {
1907 *max_extent_size =
1908 max(get_max_extent_size(entry),
1909 *max_extent_size);
1910 }
1911 continue;
1912 }
1913
1914 *offset = tmp;
1915 *bytes = entry->bytes - align_off;
1916 return entry;
1917 }
1918 out:
1919 return NULL;
1920 }
1921
1922 static int count_bitmap_extents(struct btrfs_free_space_ctl *ctl,
1923 struct btrfs_free_space *bitmap_info)
1924 {
1925 struct btrfs_block_group *block_group = ctl->private;
1926 u64 bytes = bitmap_info->bytes;
1927 unsigned int rs, re;
1928 int count = 0;
1929
1930 if (!block_group || !bytes)
1931 return count;
1932
1933 bitmap_for_each_set_region(bitmap_info->bitmap, rs, re, 0,
1934 BITS_PER_BITMAP) {
1935 bytes -= (rs - re) * ctl->unit;
1936 count++;
1937
1938 if (!bytes)
1939 break;
1940 }
1941
1942 return count;
1943 }
1944
1945 static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
1946 struct btrfs_free_space *info, u64 offset)
1947 {
1948 info->offset = offset_to_bitmap(ctl, offset);
1949 info->bytes = 0;
1950 info->bitmap_extents = 0;
1951 INIT_LIST_HEAD(&info->list);
1952 link_free_space(ctl, info);
1953 ctl->total_bitmaps++;
1954
1955 ctl->op->recalc_thresholds(ctl);
1956 }
1957
1958 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
1959 struct btrfs_free_space *bitmap_info)
1960 {
1961 /*
1962 * Normally when this is called, the bitmap is completely empty. However,
1963 * if we are blowing up the free space cache for one reason or another
1964 * via __btrfs_remove_free_space_cache(), then it may not be freed and
1965 * we may leave stats on the table.
1966 */
1967 if (bitmap_info->bytes && !btrfs_free_space_trimmed(bitmap_info)) {
1968 ctl->discardable_extents[BTRFS_STAT_CURR] -=
1969 bitmap_info->bitmap_extents;
1970 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bitmap_info->bytes;
1971
1972 }
1973 unlink_free_space(ctl, bitmap_info);
1974 kmem_cache_free(btrfs_free_space_bitmap_cachep, bitmap_info->bitmap);
1975 kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1976 ctl->total_bitmaps--;
1977 ctl->op->recalc_thresholds(ctl);
1978 }
1979
1980 static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
1981 struct btrfs_free_space *bitmap_info,
1982 u64 *offset, u64 *bytes)
1983 {
1984 u64 end;
1985 u64 search_start, search_bytes;
1986 int ret;
1987
1988 again:
1989 end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
1990
1991 /*
1992 * We need to search for bits in this bitmap. We could only cover some
1993 * of the extent in this bitmap thanks to how we add space, so we need
1994 * to search for as much as it as we can and clear that amount, and then
1995 * go searching for the next bit.
1996 */
1997 search_start = *offset;
1998 search_bytes = ctl->unit;
1999 search_bytes = min(search_bytes, end - search_start + 1);
2000 ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
2001 false);
2002 if (ret < 0 || search_start != *offset)
2003 return -EINVAL;
2004
2005 /* We may have found more bits than what we need */
2006 search_bytes = min(search_bytes, *bytes);
2007
2008 /* Cannot clear past the end of the bitmap */
2009 search_bytes = min(search_bytes, end - search_start + 1);
2010
2011 bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes);
2012 *offset += search_bytes;
2013 *bytes -= search_bytes;
2014
2015 if (*bytes) {
2016 struct rb_node *next = rb_next(&bitmap_info->offset_index);
2017 if (!bitmap_info->bytes)
2018 free_bitmap(ctl, bitmap_info);
2019
2020 /*
2021 * no entry after this bitmap, but we still have bytes to
2022 * remove, so something has gone wrong.
2023 */
2024 if (!next)
2025 return -EINVAL;
2026
2027 bitmap_info = rb_entry(next, struct btrfs_free_space,
2028 offset_index);
2029
2030 /*
2031 * if the next entry isn't a bitmap we need to return to let the
2032 * extent stuff do its work.
2033 */
2034 if (!bitmap_info->bitmap)
2035 return -EAGAIN;
2036
2037 /*
2038 * Ok the next item is a bitmap, but it may not actually hold
2039 * the information for the rest of this free space stuff, so
2040 * look for it, and if we don't find it return so we can try
2041 * everything over again.
2042 */
2043 search_start = *offset;
2044 search_bytes = ctl->unit;
2045 ret = search_bitmap(ctl, bitmap_info, &search_start,
2046 &search_bytes, false);
2047 if (ret < 0 || search_start != *offset)
2048 return -EAGAIN;
2049
2050 goto again;
2051 } else if (!bitmap_info->bytes)
2052 free_bitmap(ctl, bitmap_info);
2053
2054 return 0;
2055 }
2056
2057 static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
2058 struct btrfs_free_space *info, u64 offset,
2059 u64 bytes, enum btrfs_trim_state trim_state)
2060 {
2061 u64 bytes_to_set = 0;
2062 u64 end;
2063
2064 /*
2065 * This is a tradeoff to make bitmap trim state minimal. We mark the
2066 * whole bitmap untrimmed if at any point we add untrimmed regions.
2067 */
2068 if (trim_state == BTRFS_TRIM_STATE_UNTRIMMED) {
2069 if (btrfs_free_space_trimmed(info)) {
2070 ctl->discardable_extents[BTRFS_STAT_CURR] +=
2071 info->bitmap_extents;
2072 ctl->discardable_bytes[BTRFS_STAT_CURR] += info->bytes;
2073 }
2074 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2075 }
2076
2077 end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
2078
2079 bytes_to_set = min(end - offset, bytes);
2080
2081 bitmap_set_bits(ctl, info, offset, bytes_to_set);
2082
2083 /*
2084 * We set some bytes, we have no idea what the max extent size is
2085 * anymore.
2086 */
2087 info->max_extent_size = 0;
2088
2089 return bytes_to_set;
2090
2091 }
2092
2093 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
2094 struct btrfs_free_space *info)
2095 {
2096 struct btrfs_block_group *block_group = ctl->private;
2097 struct btrfs_fs_info *fs_info = block_group->fs_info;
2098 bool forced = false;
2099
2100 #ifdef CONFIG_BTRFS_DEBUG
2101 if (btrfs_should_fragment_free_space(block_group))
2102 forced = true;
2103 #endif
2104
2105 /* This is a way to reclaim large regions from the bitmaps. */
2106 if (!forced && info->bytes >= FORCE_EXTENT_THRESHOLD)
2107 return false;
2108
2109 /*
2110 * If we are below the extents threshold then we can add this as an
2111 * extent, and don't have to deal with the bitmap
2112 */
2113 if (!forced && ctl->free_extents < ctl->extents_thresh) {
2114 /*
2115 * If this block group has some small extents we don't want to
2116 * use up all of our free slots in the cache with them, we want
2117 * to reserve them to larger extents, however if we have plenty
2118 * of cache left then go ahead an dadd them, no sense in adding
2119 * the overhead of a bitmap if we don't have to.
2120 */
2121 if (info->bytes <= fs_info->sectorsize * 8) {
2122 if (ctl->free_extents * 3 <= ctl->extents_thresh)
2123 return false;
2124 } else {
2125 return false;
2126 }
2127 }
2128
2129 /*
2130 * The original block groups from mkfs can be really small, like 8
2131 * megabytes, so don't bother with a bitmap for those entries. However
2132 * some block groups can be smaller than what a bitmap would cover but
2133 * are still large enough that they could overflow the 32k memory limit,
2134 * so allow those block groups to still be allowed to have a bitmap
2135 * entry.
2136 */
2137 if (((BITS_PER_BITMAP * ctl->unit) >> 1) > block_group->length)
2138 return false;
2139
2140 return true;
2141 }
2142
2143 static const struct btrfs_free_space_op free_space_op = {
2144 .recalc_thresholds = recalculate_thresholds,
2145 .use_bitmap = use_bitmap,
2146 };
2147
2148 static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
2149 struct btrfs_free_space *info)
2150 {
2151 struct btrfs_free_space *bitmap_info;
2152 struct btrfs_block_group *block_group = NULL;
2153 int added = 0;
2154 u64 bytes, offset, bytes_added;
2155 enum btrfs_trim_state trim_state;
2156 int ret;
2157
2158 bytes = info->bytes;
2159 offset = info->offset;
2160 trim_state = info->trim_state;
2161
2162 if (!ctl->op->use_bitmap(ctl, info))
2163 return 0;
2164
2165 if (ctl->op == &free_space_op)
2166 block_group = ctl->private;
2167 again:
2168 /*
2169 * Since we link bitmaps right into the cluster we need to see if we
2170 * have a cluster here, and if so and it has our bitmap we need to add
2171 * the free space to that bitmap.
2172 */
2173 if (block_group && !list_empty(&block_group->cluster_list)) {
2174 struct btrfs_free_cluster *cluster;
2175 struct rb_node *node;
2176 struct btrfs_free_space *entry;
2177
2178 cluster = list_entry(block_group->cluster_list.next,
2179 struct btrfs_free_cluster,
2180 block_group_list);
2181 spin_lock(&cluster->lock);
2182 node = rb_first(&cluster->root);
2183 if (!node) {
2184 spin_unlock(&cluster->lock);
2185 goto no_cluster_bitmap;
2186 }
2187
2188 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2189 if (!entry->bitmap) {
2190 spin_unlock(&cluster->lock);
2191 goto no_cluster_bitmap;
2192 }
2193
2194 if (entry->offset == offset_to_bitmap(ctl, offset)) {
2195 bytes_added = add_bytes_to_bitmap(ctl, entry, offset,
2196 bytes, trim_state);
2197 bytes -= bytes_added;
2198 offset += bytes_added;
2199 }
2200 spin_unlock(&cluster->lock);
2201 if (!bytes) {
2202 ret = 1;
2203 goto out;
2204 }
2205 }
2206
2207 no_cluster_bitmap:
2208 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2209 1, 0);
2210 if (!bitmap_info) {
2211 ASSERT(added == 0);
2212 goto new_bitmap;
2213 }
2214
2215 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes,
2216 trim_state);
2217 bytes -= bytes_added;
2218 offset += bytes_added;
2219 added = 0;
2220
2221 if (!bytes) {
2222 ret = 1;
2223 goto out;
2224 } else
2225 goto again;
2226
2227 new_bitmap:
2228 if (info && info->bitmap) {
2229 add_new_bitmap(ctl, info, offset);
2230 added = 1;
2231 info = NULL;
2232 goto again;
2233 } else {
2234 spin_unlock(&ctl->tree_lock);
2235
2236 /* no pre-allocated info, allocate a new one */
2237 if (!info) {
2238 info = kmem_cache_zalloc(btrfs_free_space_cachep,
2239 GFP_NOFS);
2240 if (!info) {
2241 spin_lock(&ctl->tree_lock);
2242 ret = -ENOMEM;
2243 goto out;
2244 }
2245 }
2246
2247 /* allocate the bitmap */
2248 info->bitmap = kmem_cache_zalloc(btrfs_free_space_bitmap_cachep,
2249 GFP_NOFS);
2250 info->trim_state = BTRFS_TRIM_STATE_TRIMMED;
2251 spin_lock(&ctl->tree_lock);
2252 if (!info->bitmap) {
2253 ret = -ENOMEM;
2254 goto out;
2255 }
2256 goto again;
2257 }
2258
2259 out:
2260 if (info) {
2261 if (info->bitmap)
2262 kmem_cache_free(btrfs_free_space_bitmap_cachep,
2263 info->bitmap);
2264 kmem_cache_free(btrfs_free_space_cachep, info);
2265 }
2266
2267 return ret;
2268 }
2269
2270 /*
2271 * Free space merging rules:
2272 * 1) Merge trimmed areas together
2273 * 2) Let untrimmed areas coalesce with trimmed areas
2274 * 3) Always pull neighboring regions from bitmaps
2275 *
2276 * The above rules are for when we merge free space based on btrfs_trim_state.
2277 * Rules 2 and 3 are subtle because they are suboptimal, but are done for the
2278 * same reason: to promote larger extent regions which makes life easier for
2279 * find_free_extent(). Rule 2 enables coalescing based on the common path
2280 * being returning free space from btrfs_finish_extent_commit(). So when free
2281 * space is trimmed, it will prevent aggregating trimmed new region and
2282 * untrimmed regions in the rb_tree. Rule 3 is purely to obtain larger extents
2283 * and provide find_free_extent() with the largest extents possible hoping for
2284 * the reuse path.
2285 */
2286 static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
2287 struct btrfs_free_space *info, bool update_stat)
2288 {
2289 struct btrfs_free_space *left_info;
2290 struct btrfs_free_space *right_info;
2291 bool merged = false;
2292 u64 offset = info->offset;
2293 u64 bytes = info->bytes;
2294 const bool is_trimmed = btrfs_free_space_trimmed(info);
2295
2296 /*
2297 * first we want to see if there is free space adjacent to the range we
2298 * are adding, if there is remove that struct and add a new one to
2299 * cover the entire range
2300 */
2301 right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
2302 if (right_info && rb_prev(&right_info->offset_index))
2303 left_info = rb_entry(rb_prev(&right_info->offset_index),
2304 struct btrfs_free_space, offset_index);
2305 else
2306 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
2307
2308 /* See try_merge_free_space() comment. */
2309 if (right_info && !right_info->bitmap &&
2310 (!is_trimmed || btrfs_free_space_trimmed(right_info))) {
2311 if (update_stat)
2312 unlink_free_space(ctl, right_info);
2313 else
2314 __unlink_free_space(ctl, right_info);
2315 info->bytes += right_info->bytes;
2316 kmem_cache_free(btrfs_free_space_cachep, right_info);
2317 merged = true;
2318 }
2319
2320 /* See try_merge_free_space() comment. */
2321 if (left_info && !left_info->bitmap &&
2322 left_info->offset + left_info->bytes == offset &&
2323 (!is_trimmed || btrfs_free_space_trimmed(left_info))) {
2324 if (update_stat)
2325 unlink_free_space(ctl, left_info);
2326 else
2327 __unlink_free_space(ctl, left_info);
2328 info->offset = left_info->offset;
2329 info->bytes += left_info->bytes;
2330 kmem_cache_free(btrfs_free_space_cachep, left_info);
2331 merged = true;
2332 }
2333
2334 return merged;
2335 }
2336
2337 static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
2338 struct btrfs_free_space *info,
2339 bool update_stat)
2340 {
2341 struct btrfs_free_space *bitmap;
2342 unsigned long i;
2343 unsigned long j;
2344 const u64 end = info->offset + info->bytes;
2345 const u64 bitmap_offset = offset_to_bitmap(ctl, end);
2346 u64 bytes;
2347
2348 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2349 if (!bitmap)
2350 return false;
2351
2352 i = offset_to_bit(bitmap->offset, ctl->unit, end);
2353 j = find_next_zero_bit(bitmap->bitmap, BITS_PER_BITMAP, i);
2354 if (j == i)
2355 return false;
2356 bytes = (j - i) * ctl->unit;
2357 info->bytes += bytes;
2358
2359 /* See try_merge_free_space() comment. */
2360 if (!btrfs_free_space_trimmed(bitmap))
2361 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2362
2363 if (update_stat)
2364 bitmap_clear_bits(ctl, bitmap, end, bytes);
2365 else
2366 __bitmap_clear_bits(ctl, bitmap, end, bytes);
2367
2368 if (!bitmap->bytes)
2369 free_bitmap(ctl, bitmap);
2370
2371 return true;
2372 }
2373
2374 static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
2375 struct btrfs_free_space *info,
2376 bool update_stat)
2377 {
2378 struct btrfs_free_space *bitmap;
2379 u64 bitmap_offset;
2380 unsigned long i;
2381 unsigned long j;
2382 unsigned long prev_j;
2383 u64 bytes;
2384
2385 bitmap_offset = offset_to_bitmap(ctl, info->offset);
2386 /* If we're on a boundary, try the previous logical bitmap. */
2387 if (bitmap_offset == info->offset) {
2388 if (info->offset == 0)
2389 return false;
2390 bitmap_offset = offset_to_bitmap(ctl, info->offset - 1);
2391 }
2392
2393 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2394 if (!bitmap)
2395 return false;
2396
2397 i = offset_to_bit(bitmap->offset, ctl->unit, info->offset) - 1;
2398 j = 0;
2399 prev_j = (unsigned long)-1;
2400 for_each_clear_bit_from(j, bitmap->bitmap, BITS_PER_BITMAP) {
2401 if (j > i)
2402 break;
2403 prev_j = j;
2404 }
2405 if (prev_j == i)
2406 return false;
2407
2408 if (prev_j == (unsigned long)-1)
2409 bytes = (i + 1) * ctl->unit;
2410 else
2411 bytes = (i - prev_j) * ctl->unit;
2412
2413 info->offset -= bytes;
2414 info->bytes += bytes;
2415
2416 /* See try_merge_free_space() comment. */
2417 if (!btrfs_free_space_trimmed(bitmap))
2418 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2419
2420 if (update_stat)
2421 bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2422 else
2423 __bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2424
2425 if (!bitmap->bytes)
2426 free_bitmap(ctl, bitmap);
2427
2428 return true;
2429 }
2430
2431 /*
2432 * We prefer always to allocate from extent entries, both for clustered and
2433 * non-clustered allocation requests. So when attempting to add a new extent
2434 * entry, try to see if there's adjacent free space in bitmap entries, and if
2435 * there is, migrate that space from the bitmaps to the extent.
2436 * Like this we get better chances of satisfying space allocation requests
2437 * because we attempt to satisfy them based on a single cache entry, and never
2438 * on 2 or more entries - even if the entries represent a contiguous free space
2439 * region (e.g. 1 extent entry + 1 bitmap entry starting where the extent entry
2440 * ends).
2441 */
2442 static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
2443 struct btrfs_free_space *info,
2444 bool update_stat)
2445 {
2446 /*
2447 * Only work with disconnected entries, as we can change their offset,
2448 * and must be extent entries.
2449 */
2450 ASSERT(!info->bitmap);
2451 ASSERT(RB_EMPTY_NODE(&info->offset_index));
2452
2453 if (ctl->total_bitmaps > 0) {
2454 bool stole_end;
2455 bool stole_front = false;
2456
2457 stole_end = steal_from_bitmap_to_end(ctl, info, update_stat);
2458 if (ctl->total_bitmaps > 0)
2459 stole_front = steal_from_bitmap_to_front(ctl, info,
2460 update_stat);
2461
2462 if (stole_end || stole_front)
2463 try_merge_free_space(ctl, info, update_stat);
2464 }
2465 }
2466
2467 int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
2468 struct btrfs_free_space_ctl *ctl,
2469 u64 offset, u64 bytes,
2470 enum btrfs_trim_state trim_state)
2471 {
2472 struct btrfs_block_group *block_group = ctl->private;
2473 struct btrfs_free_space *info;
2474 int ret = 0;
2475 u64 filter_bytes = bytes;
2476
2477 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
2478 if (!info)
2479 return -ENOMEM;
2480
2481 info->offset = offset;
2482 info->bytes = bytes;
2483 info->trim_state = trim_state;
2484 RB_CLEAR_NODE(&info->offset_index);
2485
2486 spin_lock(&ctl->tree_lock);
2487
2488 if (try_merge_free_space(ctl, info, true))
2489 goto link;
2490
2491 /*
2492 * There was no extent directly to the left or right of this new
2493 * extent then we know we're going to have to allocate a new extent, so
2494 * before we do that see if we need to drop this into a bitmap
2495 */
2496 ret = insert_into_bitmap(ctl, info);
2497 if (ret < 0) {
2498 goto out;
2499 } else if (ret) {
2500 ret = 0;
2501 goto out;
2502 }
2503 link:
2504 /*
2505 * Only steal free space from adjacent bitmaps if we're sure we're not
2506 * going to add the new free space to existing bitmap entries - because
2507 * that would mean unnecessary work that would be reverted. Therefore
2508 * attempt to steal space from bitmaps if we're adding an extent entry.
2509 */
2510 steal_from_bitmap(ctl, info, true);
2511
2512 filter_bytes = max(filter_bytes, info->bytes);
2513
2514 ret = link_free_space(ctl, info);
2515 if (ret)
2516 kmem_cache_free(btrfs_free_space_cachep, info);
2517 out:
2518 btrfs_discard_update_discardable(block_group, ctl);
2519 spin_unlock(&ctl->tree_lock);
2520
2521 if (ret) {
2522 btrfs_crit(fs_info, "unable to add free space :%d", ret);
2523 ASSERT(ret != -EEXIST);
2524 }
2525
2526 if (trim_state != BTRFS_TRIM_STATE_TRIMMED) {
2527 btrfs_discard_check_filter(block_group, filter_bytes);
2528 btrfs_discard_queue_work(&fs_info->discard_ctl, block_group);
2529 }
2530
2531 return ret;
2532 }
2533
2534 int btrfs_add_free_space(struct btrfs_block_group *block_group,
2535 u64 bytenr, u64 size)
2536 {
2537 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2538
2539 if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC))
2540 trim_state = BTRFS_TRIM_STATE_TRIMMED;
2541
2542 return __btrfs_add_free_space(block_group->fs_info,
2543 block_group->free_space_ctl,
2544 bytenr, size, trim_state);
2545 }
2546
2547 /*
2548 * This is a subtle distinction because when adding free space back in general,
2549 * we want it to be added as untrimmed for async. But in the case where we add
2550 * it on loading of a block group, we want to consider it trimmed.
2551 */
2552 int btrfs_add_free_space_async_trimmed(struct btrfs_block_group *block_group,
2553 u64 bytenr, u64 size)
2554 {
2555 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2556
2557 if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC) ||
2558 btrfs_test_opt(block_group->fs_info, DISCARD_ASYNC))
2559 trim_state = BTRFS_TRIM_STATE_TRIMMED;
2560
2561 return __btrfs_add_free_space(block_group->fs_info,
2562 block_group->free_space_ctl,
2563 bytenr, size, trim_state);
2564 }
2565
2566 int btrfs_remove_free_space(struct btrfs_block_group *block_group,
2567 u64 offset, u64 bytes)
2568 {
2569 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2570 struct btrfs_free_space *info;
2571 int ret;
2572 bool re_search = false;
2573
2574 spin_lock(&ctl->tree_lock);
2575
2576 again:
2577 ret = 0;
2578 if (!bytes)
2579 goto out_lock;
2580
2581 info = tree_search_offset(ctl, offset, 0, 0);
2582 if (!info) {
2583 /*
2584 * oops didn't find an extent that matched the space we wanted
2585 * to remove, look for a bitmap instead
2586 */
2587 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2588 1, 0);
2589 if (!info) {
2590 /*
2591 * If we found a partial bit of our free space in a
2592 * bitmap but then couldn't find the other part this may
2593 * be a problem, so WARN about it.
2594 */
2595 WARN_ON(re_search);
2596 goto out_lock;
2597 }
2598 }
2599
2600 re_search = false;
2601 if (!info->bitmap) {
2602 unlink_free_space(ctl, info);
2603 if (offset == info->offset) {
2604 u64 to_free = min(bytes, info->bytes);
2605
2606 info->bytes -= to_free;
2607 info->offset += to_free;
2608 if (info->bytes) {
2609 ret = link_free_space(ctl, info);
2610 WARN_ON(ret);
2611 } else {
2612 kmem_cache_free(btrfs_free_space_cachep, info);
2613 }
2614
2615 offset += to_free;
2616 bytes -= to_free;
2617 goto again;
2618 } else {
2619 u64 old_end = info->bytes + info->offset;
2620
2621 info->bytes = offset - info->offset;
2622 ret = link_free_space(ctl, info);
2623 WARN_ON(ret);
2624 if (ret)
2625 goto out_lock;
2626
2627 /* Not enough bytes in this entry to satisfy us */
2628 if (old_end < offset + bytes) {
2629 bytes -= old_end - offset;
2630 offset = old_end;
2631 goto again;
2632 } else if (old_end == offset + bytes) {
2633 /* all done */
2634 goto out_lock;
2635 }
2636 spin_unlock(&ctl->tree_lock);
2637
2638 ret = __btrfs_add_free_space(block_group->fs_info, ctl,
2639 offset + bytes,
2640 old_end - (offset + bytes),
2641 info->trim_state);
2642 WARN_ON(ret);
2643 goto out;
2644 }
2645 }
2646
2647 ret = remove_from_bitmap(ctl, info, &offset, &bytes);
2648 if (ret == -EAGAIN) {
2649 re_search = true;
2650 goto again;
2651 }
2652 out_lock:
2653 btrfs_discard_update_discardable(block_group, ctl);
2654 spin_unlock(&ctl->tree_lock);
2655 out:
2656 return ret;
2657 }
2658
2659 void btrfs_dump_free_space(struct btrfs_block_group *block_group,
2660 u64 bytes)
2661 {
2662 struct btrfs_fs_info *fs_info = block_group->fs_info;
2663 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2664 struct btrfs_free_space *info;
2665 struct rb_node *n;
2666 int count = 0;
2667
2668 spin_lock(&ctl->tree_lock);
2669 for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
2670 info = rb_entry(n, struct btrfs_free_space, offset_index);
2671 if (info->bytes >= bytes && !block_group->ro)
2672 count++;
2673 btrfs_crit(fs_info, "entry offset %llu, bytes %llu, bitmap %s",
2674 info->offset, info->bytes,
2675 (info->bitmap) ? "yes" : "no");
2676 }
2677 spin_unlock(&ctl->tree_lock);
2678 btrfs_info(fs_info, "block group has cluster?: %s",
2679 list_empty(&block_group->cluster_list) ? "no" : "yes");
2680 btrfs_info(fs_info,
2681 "%d blocks of free space at or bigger than bytes is", count);
2682 }
2683
2684 void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group)
2685 {
2686 struct btrfs_fs_info *fs_info = block_group->fs_info;
2687 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2688
2689 spin_lock_init(&ctl->tree_lock);
2690 ctl->unit = fs_info->sectorsize;
2691 ctl->start = block_group->start;
2692 ctl->private = block_group;
2693 ctl->op = &free_space_op;
2694 INIT_LIST_HEAD(&ctl->trimming_ranges);
2695 mutex_init(&ctl->cache_writeout_mutex);
2696
2697 /*
2698 * we only want to have 32k of ram per block group for keeping
2699 * track of free space, and if we pass 1/2 of that we want to
2700 * start converting things over to using bitmaps
2701 */
2702 ctl->extents_thresh = (SZ_32K / 2) / sizeof(struct btrfs_free_space);
2703 }
2704
2705 /*
2706 * for a given cluster, put all of its extents back into the free
2707 * space cache. If the block group passed doesn't match the block group
2708 * pointed to by the cluster, someone else raced in and freed the
2709 * cluster already. In that case, we just return without changing anything
2710 */
2711 static int
2712 __btrfs_return_cluster_to_free_space(
2713 struct btrfs_block_group *block_group,
2714 struct btrfs_free_cluster *cluster)
2715 {
2716 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2717 struct btrfs_free_space *entry;
2718 struct rb_node *node;
2719
2720 spin_lock(&cluster->lock);
2721 if (cluster->block_group != block_group)
2722 goto out;
2723
2724 cluster->block_group = NULL;
2725 cluster->window_start = 0;
2726 list_del_init(&cluster->block_group_list);
2727
2728 node = rb_first(&cluster->root);
2729 while (node) {
2730 bool bitmap;
2731
2732 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2733 node = rb_next(&entry->offset_index);
2734 rb_erase(&entry->offset_index, &cluster->root);
2735 RB_CLEAR_NODE(&entry->offset_index);
2736
2737 bitmap = (entry->bitmap != NULL);
2738 if (!bitmap) {
2739 /* Merging treats extents as if they were new */
2740 if (!btrfs_free_space_trimmed(entry)) {
2741 ctl->discardable_extents[BTRFS_STAT_CURR]--;
2742 ctl->discardable_bytes[BTRFS_STAT_CURR] -=
2743 entry->bytes;
2744 }
2745
2746 try_merge_free_space(ctl, entry, false);
2747 steal_from_bitmap(ctl, entry, false);
2748
2749 /* As we insert directly, update these statistics */
2750 if (!btrfs_free_space_trimmed(entry)) {
2751 ctl->discardable_extents[BTRFS_STAT_CURR]++;
2752 ctl->discardable_bytes[BTRFS_STAT_CURR] +=
2753 entry->bytes;
2754 }
2755 }
2756 tree_insert_offset(&ctl->free_space_offset,
2757 entry->offset, &entry->offset_index, bitmap);
2758 }
2759 cluster->root = RB_ROOT;
2760
2761 out:
2762 spin_unlock(&cluster->lock);
2763 btrfs_put_block_group(block_group);
2764 return 0;
2765 }
2766
2767 static void __btrfs_remove_free_space_cache_locked(
2768 struct btrfs_free_space_ctl *ctl)
2769 {
2770 struct btrfs_free_space *info;
2771 struct rb_node *node;
2772
2773 while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
2774 info = rb_entry(node, struct btrfs_free_space, offset_index);
2775 if (!info->bitmap) {
2776 unlink_free_space(ctl, info);
2777 kmem_cache_free(btrfs_free_space_cachep, info);
2778 } else {
2779 free_bitmap(ctl, info);
2780 }
2781
2782 cond_resched_lock(&ctl->tree_lock);
2783 }
2784 }
2785
2786 void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
2787 {
2788 spin_lock(&ctl->tree_lock);
2789 __btrfs_remove_free_space_cache_locked(ctl);
2790 if (ctl->private)
2791 btrfs_discard_update_discardable(ctl->private, ctl);
2792 spin_unlock(&ctl->tree_lock);
2793 }
2794
2795 void btrfs_remove_free_space_cache(struct btrfs_block_group *block_group)
2796 {
2797 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2798 struct btrfs_free_cluster *cluster;
2799 struct list_head *head;
2800
2801 spin_lock(&ctl->tree_lock);
2802 while ((head = block_group->cluster_list.next) !=
2803 &block_group->cluster_list) {
2804 cluster = list_entry(head, struct btrfs_free_cluster,
2805 block_group_list);
2806
2807 WARN_ON(cluster->block_group != block_group);
2808 __btrfs_return_cluster_to_free_space(block_group, cluster);
2809
2810 cond_resched_lock(&ctl->tree_lock);
2811 }
2812 __btrfs_remove_free_space_cache_locked(ctl);
2813 btrfs_discard_update_discardable(block_group, ctl);
2814 spin_unlock(&ctl->tree_lock);
2815
2816 }
2817
2818 /**
2819 * btrfs_is_free_space_trimmed - see if everything is trimmed
2820 * @block_group: block_group of interest
2821 *
2822 * Walk @block_group's free space rb_tree to determine if everything is trimmed.
2823 */
2824 bool btrfs_is_free_space_trimmed(struct btrfs_block_group *block_group)
2825 {
2826 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2827 struct btrfs_free_space *info;
2828 struct rb_node *node;
2829 bool ret = true;
2830
2831 spin_lock(&ctl->tree_lock);
2832 node = rb_first(&ctl->free_space_offset);
2833
2834 while (node) {
2835 info = rb_entry(node, struct btrfs_free_space, offset_index);
2836
2837 if (!btrfs_free_space_trimmed(info)) {
2838 ret = false;
2839 break;
2840 }
2841
2842 node = rb_next(node);
2843 }
2844
2845 spin_unlock(&ctl->tree_lock);
2846 return ret;
2847 }
2848
2849 u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
2850 u64 offset, u64 bytes, u64 empty_size,
2851 u64 *max_extent_size)
2852 {
2853 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2854 struct btrfs_discard_ctl *discard_ctl =
2855 &block_group->fs_info->discard_ctl;
2856 struct btrfs_free_space *entry = NULL;
2857 u64 bytes_search = bytes + empty_size;
2858 u64 ret = 0;
2859 u64 align_gap = 0;
2860 u64 align_gap_len = 0;
2861 enum btrfs_trim_state align_gap_trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2862
2863 spin_lock(&ctl->tree_lock);
2864 entry = find_free_space(ctl, &offset, &bytes_search,
2865 block_group->full_stripe_len, max_extent_size);
2866 if (!entry)
2867 goto out;
2868
2869 ret = offset;
2870 if (entry->bitmap) {
2871 bitmap_clear_bits(ctl, entry, offset, bytes);
2872
2873 if (!btrfs_free_space_trimmed(entry))
2874 atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
2875
2876 if (!entry->bytes)
2877 free_bitmap(ctl, entry);
2878 } else {
2879 unlink_free_space(ctl, entry);
2880 align_gap_len = offset - entry->offset;
2881 align_gap = entry->offset;
2882 align_gap_trim_state = entry->trim_state;
2883
2884 if (!btrfs_free_space_trimmed(entry))
2885 atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
2886
2887 entry->offset = offset + bytes;
2888 WARN_ON(entry->bytes < bytes + align_gap_len);
2889
2890 entry->bytes -= bytes + align_gap_len;
2891 if (!entry->bytes)
2892 kmem_cache_free(btrfs_free_space_cachep, entry);
2893 else
2894 link_free_space(ctl, entry);
2895 }
2896 out:
2897 btrfs_discard_update_discardable(block_group, ctl);
2898 spin_unlock(&ctl->tree_lock);
2899
2900 if (align_gap_len)
2901 __btrfs_add_free_space(block_group->fs_info, ctl,
2902 align_gap, align_gap_len,
2903 align_gap_trim_state);
2904 return ret;
2905 }
2906
2907 /*
2908 * given a cluster, put all of its extents back into the free space
2909 * cache. If a block group is passed, this function will only free
2910 * a cluster that belongs to the passed block group.
2911 *
2912 * Otherwise, it'll get a reference on the block group pointed to by the
2913 * cluster and remove the cluster from it.
2914 */
2915 int btrfs_return_cluster_to_free_space(
2916 struct btrfs_block_group *block_group,
2917 struct btrfs_free_cluster *cluster)
2918 {
2919 struct btrfs_free_space_ctl *ctl;
2920 int ret;
2921
2922 /* first, get a safe pointer to the block group */
2923 spin_lock(&cluster->lock);
2924 if (!block_group) {
2925 block_group = cluster->block_group;
2926 if (!block_group) {
2927 spin_unlock(&cluster->lock);
2928 return 0;
2929 }
2930 } else if (cluster->block_group != block_group) {
2931 /* someone else has already freed it don't redo their work */
2932 spin_unlock(&cluster->lock);
2933 return 0;
2934 }
2935 atomic_inc(&block_group->count);
2936 spin_unlock(&cluster->lock);
2937
2938 ctl = block_group->free_space_ctl;
2939
2940 /* now return any extents the cluster had on it */
2941 spin_lock(&ctl->tree_lock);
2942 ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
2943 spin_unlock(&ctl->tree_lock);
2944
2945 btrfs_discard_queue_work(&block_group->fs_info->discard_ctl, block_group);
2946
2947 /* finally drop our ref */
2948 btrfs_put_block_group(block_group);
2949 return ret;
2950 }
2951
2952 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group *block_group,
2953 struct btrfs_free_cluster *cluster,
2954 struct btrfs_free_space *entry,
2955 u64 bytes, u64 min_start,
2956 u64 *max_extent_size)
2957 {
2958 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2959 int err;
2960 u64 search_start = cluster->window_start;
2961 u64 search_bytes = bytes;
2962 u64 ret = 0;
2963
2964 search_start = min_start;
2965 search_bytes = bytes;
2966
2967 err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
2968 if (err) {
2969 *max_extent_size = max(get_max_extent_size(entry),
2970 *max_extent_size);
2971 return 0;
2972 }
2973
2974 ret = search_start;
2975 __bitmap_clear_bits(ctl, entry, ret, bytes);
2976
2977 return ret;
2978 }
2979
2980 /*
2981 * given a cluster, try to allocate 'bytes' from it, returns 0
2982 * if it couldn't find anything suitably large, or a logical disk offset
2983 * if things worked out
2984 */
2985 u64 btrfs_alloc_from_cluster(struct btrfs_block_group *block_group,
2986 struct btrfs_free_cluster *cluster, u64 bytes,
2987 u64 min_start, u64 *max_extent_size)
2988 {
2989 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2990 struct btrfs_discard_ctl *discard_ctl =
2991 &block_group->fs_info->discard_ctl;
2992 struct btrfs_free_space *entry = NULL;
2993 struct rb_node *node;
2994 u64 ret = 0;
2995
2996 spin_lock(&cluster->lock);
2997 if (bytes > cluster->max_size)
2998 goto out;
2999
3000 if (cluster->block_group != block_group)
3001 goto out;
3002
3003 node = rb_first(&cluster->root);
3004 if (!node)
3005 goto out;
3006
3007 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3008 while (1) {
3009 if (entry->bytes < bytes)
3010 *max_extent_size = max(get_max_extent_size(entry),
3011 *max_extent_size);
3012
3013 if (entry->bytes < bytes ||
3014 (!entry->bitmap && entry->offset < min_start)) {
3015 node = rb_next(&entry->offset_index);
3016 if (!node)
3017 break;
3018 entry = rb_entry(node, struct btrfs_free_space,
3019 offset_index);
3020 continue;
3021 }
3022
3023 if (entry->bitmap) {
3024 ret = btrfs_alloc_from_bitmap(block_group,
3025 cluster, entry, bytes,
3026 cluster->window_start,
3027 max_extent_size);
3028 if (ret == 0) {
3029 node = rb_next(&entry->offset_index);
3030 if (!node)
3031 break;
3032 entry = rb_entry(node, struct btrfs_free_space,
3033 offset_index);
3034 continue;
3035 }
3036 cluster->window_start += bytes;
3037 } else {
3038 ret = entry->offset;
3039
3040 entry->offset += bytes;
3041 entry->bytes -= bytes;
3042 }
3043
3044 if (entry->bytes == 0)
3045 rb_erase(&entry->offset_index, &cluster->root);
3046 break;
3047 }
3048 out:
3049 spin_unlock(&cluster->lock);
3050
3051 if (!ret)
3052 return 0;
3053
3054 spin_lock(&ctl->tree_lock);
3055
3056 if (!btrfs_free_space_trimmed(entry))
3057 atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
3058
3059 ctl->free_space -= bytes;
3060 if (!entry->bitmap && !btrfs_free_space_trimmed(entry))
3061 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
3062 if (entry->bytes == 0) {
3063 ctl->free_extents--;
3064 if (entry->bitmap) {
3065 kmem_cache_free(btrfs_free_space_bitmap_cachep,
3066 entry->bitmap);
3067 ctl->total_bitmaps--;
3068 ctl->op->recalc_thresholds(ctl);
3069 } else if (!btrfs_free_space_trimmed(entry)) {
3070 ctl->discardable_extents[BTRFS_STAT_CURR]--;
3071 }
3072 kmem_cache_free(btrfs_free_space_cachep, entry);
3073 }
3074
3075 spin_unlock(&ctl->tree_lock);
3076
3077 return ret;
3078 }
3079
3080 static int btrfs_bitmap_cluster(struct btrfs_block_group *block_group,
3081 struct btrfs_free_space *entry,
3082 struct btrfs_free_cluster *cluster,
3083 u64 offset, u64 bytes,
3084 u64 cont1_bytes, u64 min_bytes)
3085 {
3086 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3087 unsigned long next_zero;
3088 unsigned long i;
3089 unsigned long want_bits;
3090 unsigned long min_bits;
3091 unsigned long found_bits;
3092 unsigned long max_bits = 0;
3093 unsigned long start = 0;
3094 unsigned long total_found = 0;
3095 int ret;
3096
3097 i = offset_to_bit(entry->offset, ctl->unit,
3098 max_t(u64, offset, entry->offset));
3099 want_bits = bytes_to_bits(bytes, ctl->unit);
3100 min_bits = bytes_to_bits(min_bytes, ctl->unit);
3101
3102 /*
3103 * Don't bother looking for a cluster in this bitmap if it's heavily
3104 * fragmented.
3105 */
3106 if (entry->max_extent_size &&
3107 entry->max_extent_size < cont1_bytes)
3108 return -ENOSPC;
3109 again:
3110 found_bits = 0;
3111 for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
3112 next_zero = find_next_zero_bit(entry->bitmap,
3113 BITS_PER_BITMAP, i);
3114 if (next_zero - i >= min_bits) {
3115 found_bits = next_zero - i;
3116 if (found_bits > max_bits)
3117 max_bits = found_bits;
3118 break;
3119 }
3120 if (next_zero - i > max_bits)
3121 max_bits = next_zero - i;
3122 i = next_zero;
3123 }
3124
3125 if (!found_bits) {
3126 entry->max_extent_size = (u64)max_bits * ctl->unit;
3127 return -ENOSPC;
3128 }
3129
3130 if (!total_found) {
3131 start = i;
3132 cluster->max_size = 0;
3133 }
3134
3135 total_found += found_bits;
3136
3137 if (cluster->max_size < found_bits * ctl->unit)
3138 cluster->max_size = found_bits * ctl->unit;
3139
3140 if (total_found < want_bits || cluster->max_size < cont1_bytes) {
3141 i = next_zero + 1;
3142 goto again;
3143 }
3144
3145 cluster->window_start = start * ctl->unit + entry->offset;
3146 rb_erase(&entry->offset_index, &ctl->free_space_offset);
3147 ret = tree_insert_offset(&cluster->root, entry->offset,
3148 &entry->offset_index, 1);
3149 ASSERT(!ret); /* -EEXIST; Logic error */
3150
3151 trace_btrfs_setup_cluster(block_group, cluster,
3152 total_found * ctl->unit, 1);
3153 return 0;
3154 }
3155
3156 /*
3157 * This searches the block group for just extents to fill the cluster with.
3158 * Try to find a cluster with at least bytes total bytes, at least one
3159 * extent of cont1_bytes, and other clusters of at least min_bytes.
3160 */
3161 static noinline int
3162 setup_cluster_no_bitmap(struct btrfs_block_group *block_group,
3163 struct btrfs_free_cluster *cluster,
3164 struct list_head *bitmaps, u64 offset, u64 bytes,
3165 u64 cont1_bytes, u64 min_bytes)
3166 {
3167 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3168 struct btrfs_free_space *first = NULL;
3169 struct btrfs_free_space *entry = NULL;
3170 struct btrfs_free_space *last;
3171 struct rb_node *node;
3172 u64 window_free;
3173 u64 max_extent;
3174 u64 total_size = 0;
3175
3176 entry = tree_search_offset(ctl, offset, 0, 1);
3177 if (!entry)
3178 return -ENOSPC;
3179
3180 /*
3181 * We don't want bitmaps, so just move along until we find a normal
3182 * extent entry.
3183 */
3184 while (entry->bitmap || entry->bytes < min_bytes) {
3185 if (entry->bitmap && list_empty(&entry->list))
3186 list_add_tail(&entry->list, bitmaps);
3187 node = rb_next(&entry->offset_index);
3188 if (!node)
3189 return -ENOSPC;
3190 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3191 }
3192
3193 window_free = entry->bytes;
3194 max_extent = entry->bytes;
3195 first = entry;
3196 last = entry;
3197
3198 for (node = rb_next(&entry->offset_index); node;
3199 node = rb_next(&entry->offset_index)) {
3200 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3201
3202 if (entry->bitmap) {
3203 if (list_empty(&entry->list))
3204 list_add_tail(&entry->list, bitmaps);
3205 continue;
3206 }
3207
3208 if (entry->bytes < min_bytes)
3209 continue;
3210
3211 last = entry;
3212 window_free += entry->bytes;
3213 if (entry->bytes > max_extent)
3214 max_extent = entry->bytes;
3215 }
3216
3217 if (window_free < bytes || max_extent < cont1_bytes)
3218 return -ENOSPC;
3219
3220 cluster->window_start = first->offset;
3221
3222 node = &first->offset_index;
3223
3224 /*
3225 * now we've found our entries, pull them out of the free space
3226 * cache and put them into the cluster rbtree
3227 */
3228 do {
3229 int ret;
3230
3231 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3232 node = rb_next(&entry->offset_index);
3233 if (entry->bitmap || entry->bytes < min_bytes)
3234 continue;
3235
3236 rb_erase(&entry->offset_index, &ctl->free_space_offset);
3237 ret = tree_insert_offset(&cluster->root, entry->offset,
3238 &entry->offset_index, 0);
3239 total_size += entry->bytes;
3240 ASSERT(!ret); /* -EEXIST; Logic error */
3241 } while (node && entry != last);
3242
3243 cluster->max_size = max_extent;
3244 trace_btrfs_setup_cluster(block_group, cluster, total_size, 0);
3245 return 0;
3246 }
3247
3248 /*
3249 * This specifically looks for bitmaps that may work in the cluster, we assume
3250 * that we have already failed to find extents that will work.
3251 */
3252 static noinline int
3253 setup_cluster_bitmap(struct btrfs_block_group *block_group,
3254 struct btrfs_free_cluster *cluster,
3255 struct list_head *bitmaps, u64 offset, u64 bytes,
3256 u64 cont1_bytes, u64 min_bytes)
3257 {
3258 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3259 struct btrfs_free_space *entry = NULL;
3260 int ret = -ENOSPC;
3261 u64 bitmap_offset = offset_to_bitmap(ctl, offset);
3262
3263 if (ctl->total_bitmaps == 0)
3264 return -ENOSPC;
3265
3266 /*
3267 * The bitmap that covers offset won't be in the list unless offset
3268 * is just its start offset.
3269 */
3270 if (!list_empty(bitmaps))
3271 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
3272
3273 if (!entry || entry->offset != bitmap_offset) {
3274 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
3275 if (entry && list_empty(&entry->list))
3276 list_add(&entry->list, bitmaps);
3277 }
3278
3279 list_for_each_entry(entry, bitmaps, list) {
3280 if (entry->bytes < bytes)
3281 continue;
3282 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
3283 bytes, cont1_bytes, min_bytes);
3284 if (!ret)
3285 return 0;
3286 }
3287
3288 /*
3289 * The bitmaps list has all the bitmaps that record free space
3290 * starting after offset, so no more search is required.
3291 */
3292 return -ENOSPC;
3293 }
3294
3295 /*
3296 * here we try to find a cluster of blocks in a block group. The goal
3297 * is to find at least bytes+empty_size.
3298 * We might not find them all in one contiguous area.
3299 *
3300 * returns zero and sets up cluster if things worked out, otherwise
3301 * it returns -enospc
3302 */
3303 int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
3304 struct btrfs_free_cluster *cluster,
3305 u64 offset, u64 bytes, u64 empty_size)
3306 {
3307 struct btrfs_fs_info *fs_info = block_group->fs_info;
3308 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3309 struct btrfs_free_space *entry, *tmp;
3310 LIST_HEAD(bitmaps);
3311 u64 min_bytes;
3312 u64 cont1_bytes;
3313 int ret;
3314
3315 /*
3316 * Choose the minimum extent size we'll require for this
3317 * cluster. For SSD_SPREAD, don't allow any fragmentation.
3318 * For metadata, allow allocates with smaller extents. For
3319 * data, keep it dense.
3320 */
3321 if (btrfs_test_opt(fs_info, SSD_SPREAD)) {
3322 cont1_bytes = min_bytes = bytes + empty_size;
3323 } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
3324 cont1_bytes = bytes;
3325 min_bytes = fs_info->sectorsize;
3326 } else {
3327 cont1_bytes = max(bytes, (bytes + empty_size) >> 2);
3328 min_bytes = fs_info->sectorsize;
3329 }
3330
3331 spin_lock(&ctl->tree_lock);
3332
3333 /*
3334 * If we know we don't have enough space to make a cluster don't even
3335 * bother doing all the work to try and find one.
3336 */
3337 if (ctl->free_space < bytes) {
3338 spin_unlock(&ctl->tree_lock);
3339 return -ENOSPC;
3340 }
3341
3342 spin_lock(&cluster->lock);
3343
3344 /* someone already found a cluster, hooray */
3345 if (cluster->block_group) {
3346 ret = 0;
3347 goto out;
3348 }
3349
3350 trace_btrfs_find_cluster(block_group, offset, bytes, empty_size,
3351 min_bytes);
3352
3353 ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
3354 bytes + empty_size,
3355 cont1_bytes, min_bytes);
3356 if (ret)
3357 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
3358 offset, bytes + empty_size,
3359 cont1_bytes, min_bytes);
3360
3361 /* Clear our temporary list */
3362 list_for_each_entry_safe(entry, tmp, &bitmaps, list)
3363 list_del_init(&entry->list);
3364
3365 if (!ret) {
3366 atomic_inc(&block_group->count);
3367 list_add_tail(&cluster->block_group_list,
3368 &block_group->cluster_list);
3369 cluster->block_group = block_group;
3370 } else {
3371 trace_btrfs_failed_cluster_setup(block_group);
3372 }
3373 out:
3374 spin_unlock(&cluster->lock);
3375 spin_unlock(&ctl->tree_lock);
3376
3377 return ret;
3378 }
3379
3380 /*
3381 * simple code to zero out a cluster
3382 */
3383 void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
3384 {
3385 spin_lock_init(&cluster->lock);
3386 spin_lock_init(&cluster->refill_lock);
3387 cluster->root = RB_ROOT;
3388 cluster->max_size = 0;
3389 cluster->fragmented = false;
3390 INIT_LIST_HEAD(&cluster->block_group_list);
3391 cluster->block_group = NULL;
3392 }
3393
3394 static int do_trimming(struct btrfs_block_group *block_group,
3395 u64 *total_trimmed, u64 start, u64 bytes,
3396 u64 reserved_start, u64 reserved_bytes,
3397 enum btrfs_trim_state reserved_trim_state,
3398 struct btrfs_trim_range *trim_entry)
3399 {
3400 struct btrfs_space_info *space_info = block_group->space_info;
3401 struct btrfs_fs_info *fs_info = block_group->fs_info;
3402 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3403 int ret;
3404 int update = 0;
3405 const u64 end = start + bytes;
3406 const u64 reserved_end = reserved_start + reserved_bytes;
3407 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3408 u64 trimmed = 0;
3409
3410 spin_lock(&space_info->lock);
3411 spin_lock(&block_group->lock);
3412 if (!block_group->ro) {
3413 block_group->reserved += reserved_bytes;
3414 space_info->bytes_reserved += reserved_bytes;
3415 update = 1;
3416 }
3417 spin_unlock(&block_group->lock);
3418 spin_unlock(&space_info->lock);
3419
3420 ret = btrfs_discard_extent(fs_info, start, bytes, &trimmed);
3421 if (!ret) {
3422 *total_trimmed += trimmed;
3423 trim_state = BTRFS_TRIM_STATE_TRIMMED;
3424 }
3425
3426 mutex_lock(&ctl->cache_writeout_mutex);
3427 if (reserved_start < start)
3428 __btrfs_add_free_space(fs_info, ctl, reserved_start,
3429 start - reserved_start,
3430 reserved_trim_state);
3431 if (start + bytes < reserved_start + reserved_bytes)
3432 __btrfs_add_free_space(fs_info, ctl, end, reserved_end - end,
3433 reserved_trim_state);
3434 __btrfs_add_free_space(fs_info, ctl, start, bytes, trim_state);
3435 list_del(&trim_entry->list);
3436 mutex_unlock(&ctl->cache_writeout_mutex);
3437
3438 if (update) {
3439 spin_lock(&space_info->lock);
3440 spin_lock(&block_group->lock);
3441 if (block_group->ro)
3442 space_info->bytes_readonly += reserved_bytes;
3443 block_group->reserved -= reserved_bytes;
3444 space_info->bytes_reserved -= reserved_bytes;
3445 spin_unlock(&block_group->lock);
3446 spin_unlock(&space_info->lock);
3447 }
3448
3449 return ret;
3450 }
3451
3452 /*
3453 * If @async is set, then we will trim 1 region and return.
3454 */
3455 static int trim_no_bitmap(struct btrfs_block_group *block_group,
3456 u64 *total_trimmed, u64 start, u64 end, u64 minlen,
3457 bool async)
3458 {
3459 struct btrfs_discard_ctl *discard_ctl =
3460 &block_group->fs_info->discard_ctl;
3461 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3462 struct btrfs_free_space *entry;
3463 struct rb_node *node;
3464 int ret = 0;
3465 u64 extent_start;
3466 u64 extent_bytes;
3467 enum btrfs_trim_state extent_trim_state;
3468 u64 bytes;
3469 const u64 max_discard_size = READ_ONCE(discard_ctl->max_discard_size);
3470
3471 while (start < end) {
3472 struct btrfs_trim_range trim_entry;
3473
3474 mutex_lock(&ctl->cache_writeout_mutex);
3475 spin_lock(&ctl->tree_lock);
3476
3477 if (ctl->free_space < minlen)
3478 goto out_unlock;
3479
3480 entry = tree_search_offset(ctl, start, 0, 1);
3481 if (!entry)
3482 goto out_unlock;
3483
3484 /* Skip bitmaps and if async, already trimmed entries */
3485 while (entry->bitmap ||
3486 (async && btrfs_free_space_trimmed(entry))) {
3487 node = rb_next(&entry->offset_index);
3488 if (!node)
3489 goto out_unlock;
3490 entry = rb_entry(node, struct btrfs_free_space,
3491 offset_index);
3492 }
3493
3494 if (entry->offset >= end)
3495 goto out_unlock;
3496
3497 extent_start = entry->offset;
3498 extent_bytes = entry->bytes;
3499 extent_trim_state = entry->trim_state;
3500 if (async) {
3501 start = entry->offset;
3502 bytes = entry->bytes;
3503 if (bytes < minlen) {
3504 spin_unlock(&ctl->tree_lock);
3505 mutex_unlock(&ctl->cache_writeout_mutex);
3506 goto next;
3507 }
3508 unlink_free_space(ctl, entry);
3509 /*
3510 * Let bytes = BTRFS_MAX_DISCARD_SIZE + X.
3511 * If X < BTRFS_ASYNC_DISCARD_MIN_FILTER, we won't trim
3512 * X when we come back around. So trim it now.
3513 */
3514 if (max_discard_size &&
3515 bytes >= (max_discard_size +
3516 BTRFS_ASYNC_DISCARD_MIN_FILTER)) {
3517 bytes = max_discard_size;
3518 extent_bytes = max_discard_size;
3519 entry->offset += max_discard_size;
3520 entry->bytes -= max_discard_size;
3521 link_free_space(ctl, entry);
3522 } else {
3523 kmem_cache_free(btrfs_free_space_cachep, entry);
3524 }
3525 } else {
3526 start = max(start, extent_start);
3527 bytes = min(extent_start + extent_bytes, end) - start;
3528 if (bytes < minlen) {
3529 spin_unlock(&ctl->tree_lock);
3530 mutex_unlock(&ctl->cache_writeout_mutex);
3531 goto next;
3532 }
3533
3534 unlink_free_space(ctl, entry);
3535 kmem_cache_free(btrfs_free_space_cachep, entry);
3536 }
3537
3538 spin_unlock(&ctl->tree_lock);
3539 trim_entry.start = extent_start;
3540 trim_entry.bytes = extent_bytes;
3541 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3542 mutex_unlock(&ctl->cache_writeout_mutex);
3543
3544 ret = do_trimming(block_group, total_trimmed, start, bytes,
3545 extent_start, extent_bytes, extent_trim_state,
3546 &trim_entry);
3547 if (ret) {
3548 block_group->discard_cursor = start + bytes;
3549 break;
3550 }
3551 next:
3552 start += bytes;
3553 block_group->discard_cursor = start;
3554 if (async && *total_trimmed)
3555 break;
3556
3557 if (fatal_signal_pending(current)) {
3558 ret = -ERESTARTSYS;
3559 break;
3560 }
3561
3562 cond_resched();
3563 }
3564
3565 return ret;
3566
3567 out_unlock:
3568 block_group->discard_cursor = btrfs_block_group_end(block_group);
3569 spin_unlock(&ctl->tree_lock);
3570 mutex_unlock(&ctl->cache_writeout_mutex);
3571
3572 return ret;
3573 }
3574
3575 /*
3576 * If we break out of trimming a bitmap prematurely, we should reset the
3577 * trimming bit. In a rather contrieved case, it's possible to race here so
3578 * reset the state to BTRFS_TRIM_STATE_UNTRIMMED.
3579 *
3580 * start = start of bitmap
3581 * end = near end of bitmap
3582 *
3583 * Thread 1: Thread 2:
3584 * trim_bitmaps(start)
3585 * trim_bitmaps(end)
3586 * end_trimming_bitmap()
3587 * reset_trimming_bitmap()
3588 */
3589 static void reset_trimming_bitmap(struct btrfs_free_space_ctl *ctl, u64 offset)
3590 {
3591 struct btrfs_free_space *entry;
3592
3593 spin_lock(&ctl->tree_lock);
3594 entry = tree_search_offset(ctl, offset, 1, 0);
3595 if (entry) {
3596 if (btrfs_free_space_trimmed(entry)) {
3597 ctl->discardable_extents[BTRFS_STAT_CURR] +=
3598 entry->bitmap_extents;
3599 ctl->discardable_bytes[BTRFS_STAT_CURR] += entry->bytes;
3600 }
3601 entry->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3602 }
3603
3604 spin_unlock(&ctl->tree_lock);
3605 }
3606
3607 static void end_trimming_bitmap(struct btrfs_free_space_ctl *ctl,
3608 struct btrfs_free_space *entry)
3609 {
3610 if (btrfs_free_space_trimming_bitmap(entry)) {
3611 entry->trim_state = BTRFS_TRIM_STATE_TRIMMED;
3612 ctl->discardable_extents[BTRFS_STAT_CURR] -=
3613 entry->bitmap_extents;
3614 ctl->discardable_bytes[BTRFS_STAT_CURR] -= entry->bytes;
3615 }
3616 }
3617
3618 /*
3619 * If @async is set, then we will trim 1 region and return.
3620 */
3621 static int trim_bitmaps(struct btrfs_block_group *block_group,
3622 u64 *total_trimmed, u64 start, u64 end, u64 minlen,
3623 u64 maxlen, bool async)
3624 {
3625 struct btrfs_discard_ctl *discard_ctl =
3626 &block_group->fs_info->discard_ctl;
3627 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3628 struct btrfs_free_space *entry;
3629 int ret = 0;
3630 int ret2;
3631 u64 bytes;
3632 u64 offset = offset_to_bitmap(ctl, start);
3633 const u64 max_discard_size = READ_ONCE(discard_ctl->max_discard_size);
3634
3635 while (offset < end) {
3636 bool next_bitmap = false;
3637 struct btrfs_trim_range trim_entry;
3638
3639 mutex_lock(&ctl->cache_writeout_mutex);
3640 spin_lock(&ctl->tree_lock);
3641
3642 if (ctl->free_space < minlen) {
3643 block_group->discard_cursor =
3644 btrfs_block_group_end(block_group);
3645 spin_unlock(&ctl->tree_lock);
3646 mutex_unlock(&ctl->cache_writeout_mutex);
3647 break;
3648 }
3649
3650 entry = tree_search_offset(ctl, offset, 1, 0);
3651 /*
3652 * Bitmaps are marked trimmed lossily now to prevent constant
3653 * discarding of the same bitmap (the reason why we are bound
3654 * by the filters). So, retrim the block group bitmaps when we
3655 * are preparing to punt to the unused_bgs list. This uses
3656 * @minlen to determine if we are in BTRFS_DISCARD_INDEX_UNUSED
3657 * which is the only discard index which sets minlen to 0.
3658 */
3659 if (!entry || (async && minlen && start == offset &&
3660 btrfs_free_space_trimmed(entry))) {
3661 spin_unlock(&ctl->tree_lock);
3662 mutex_unlock(&ctl->cache_writeout_mutex);
3663 next_bitmap = true;
3664 goto next;
3665 }
3666
3667 /*
3668 * Async discard bitmap trimming begins at by setting the start
3669 * to be key.objectid and the offset_to_bitmap() aligns to the
3670 * start of the bitmap. This lets us know we are fully
3671 * scanning the bitmap rather than only some portion of it.
3672 */
3673 if (start == offset)
3674 entry->trim_state = BTRFS_TRIM_STATE_TRIMMING;
3675
3676 bytes = minlen;
3677 ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
3678 if (ret2 || start >= end) {
3679 /*
3680 * We lossily consider a bitmap trimmed if we only skip
3681 * over regions <= BTRFS_ASYNC_DISCARD_MIN_FILTER.
3682 */
3683 if (ret2 && minlen <= BTRFS_ASYNC_DISCARD_MIN_FILTER)
3684 end_trimming_bitmap(ctl, entry);
3685 else
3686 entry->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3687 spin_unlock(&ctl->tree_lock);
3688 mutex_unlock(&ctl->cache_writeout_mutex);
3689 next_bitmap = true;
3690 goto next;
3691 }
3692
3693 /*
3694 * We already trimmed a region, but are using the locking above
3695 * to reset the trim_state.
3696 */
3697 if (async && *total_trimmed) {
3698 spin_unlock(&ctl->tree_lock);
3699 mutex_unlock(&ctl->cache_writeout_mutex);
3700 goto out;
3701 }
3702
3703 bytes = min(bytes, end - start);
3704 if (bytes < minlen || (async && maxlen && bytes > maxlen)) {
3705 spin_unlock(&ctl->tree_lock);
3706 mutex_unlock(&ctl->cache_writeout_mutex);
3707 goto next;
3708 }
3709
3710 /*
3711 * Let bytes = BTRFS_MAX_DISCARD_SIZE + X.
3712 * If X < @minlen, we won't trim X when we come back around.
3713 * So trim it now. We differ here from trimming extents as we
3714 * don't keep individual state per bit.
3715 */
3716 if (async &&
3717 max_discard_size &&
3718 bytes > (max_discard_size + minlen))
3719 bytes = max_discard_size;
3720
3721 bitmap_clear_bits(ctl, entry, start, bytes);
3722 if (entry->bytes == 0)
3723 free_bitmap(ctl, entry);
3724
3725 spin_unlock(&ctl->tree_lock);
3726 trim_entry.start = start;
3727 trim_entry.bytes = bytes;
3728 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3729 mutex_unlock(&ctl->cache_writeout_mutex);
3730
3731 ret = do_trimming(block_group, total_trimmed, start, bytes,
3732 start, bytes, 0, &trim_entry);
3733 if (ret) {
3734 reset_trimming_bitmap(ctl, offset);
3735 block_group->discard_cursor =
3736 btrfs_block_group_end(block_group);
3737 break;
3738 }
3739 next:
3740 if (next_bitmap) {
3741 offset += BITS_PER_BITMAP * ctl->unit;
3742 start = offset;
3743 } else {
3744 start += bytes;
3745 }
3746 block_group->discard_cursor = start;
3747
3748 if (fatal_signal_pending(current)) {
3749 if (start != offset)
3750 reset_trimming_bitmap(ctl, offset);
3751 ret = -ERESTARTSYS;
3752 break;
3753 }
3754
3755 cond_resched();
3756 }
3757
3758 if (offset >= end)
3759 block_group->discard_cursor = end;
3760
3761 out:
3762 return ret;
3763 }
3764
3765 void btrfs_get_block_group_trimming(struct btrfs_block_group *cache)
3766 {
3767 atomic_inc(&cache->trimming);
3768 }
3769
3770 void btrfs_put_block_group_trimming(struct btrfs_block_group *block_group)
3771 {
3772 struct btrfs_fs_info *fs_info = block_group->fs_info;
3773 struct extent_map_tree *em_tree;
3774 struct extent_map *em;
3775 bool cleanup;
3776
3777 spin_lock(&block_group->lock);
3778 cleanup = (atomic_dec_and_test(&block_group->trimming) &&
3779 block_group->removed);
3780 spin_unlock(&block_group->lock);
3781
3782 if (cleanup) {
3783 mutex_lock(&fs_info->chunk_mutex);
3784 em_tree = &fs_info->mapping_tree;
3785 write_lock(&em_tree->lock);
3786 em = lookup_extent_mapping(em_tree, block_group->start,
3787 1);
3788 BUG_ON(!em); /* logic error, can't happen */
3789 remove_extent_mapping(em_tree, em);
3790 write_unlock(&em_tree->lock);
3791 mutex_unlock(&fs_info->chunk_mutex);
3792
3793 /* once for us and once for the tree */
3794 free_extent_map(em);
3795 free_extent_map(em);
3796
3797 /*
3798 * We've left one free space entry and other tasks trimming
3799 * this block group have left 1 entry each one. Free them.
3800 */
3801 __btrfs_remove_free_space_cache(block_group->free_space_ctl);
3802 }
3803 }
3804
3805 int btrfs_trim_block_group(struct btrfs_block_group *block_group,
3806 u64 *trimmed, u64 start, u64 end, u64 minlen)
3807 {
3808 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3809 int ret;
3810 u64 rem = 0;
3811
3812 *trimmed = 0;
3813
3814 spin_lock(&block_group->lock);
3815 if (block_group->removed) {
3816 spin_unlock(&block_group->lock);
3817 return 0;
3818 }
3819 btrfs_get_block_group_trimming(block_group);
3820 spin_unlock(&block_group->lock);
3821
3822 ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, false);
3823 if (ret)
3824 goto out;
3825
3826 ret = trim_bitmaps(block_group, trimmed, start, end, minlen, 0, false);
3827 div64_u64_rem(end, BITS_PER_BITMAP * ctl->unit, &rem);
3828 /* If we ended in the middle of a bitmap, reset the trimming flag */
3829 if (rem)
3830 reset_trimming_bitmap(ctl, offset_to_bitmap(ctl, end));
3831 out:
3832 btrfs_put_block_group_trimming(block_group);
3833 return ret;
3834 }
3835
3836 int btrfs_trim_block_group_extents(struct btrfs_block_group *block_group,
3837 u64 *trimmed, u64 start, u64 end, u64 minlen,
3838 bool async)
3839 {
3840 int ret;
3841
3842 *trimmed = 0;
3843
3844 spin_lock(&block_group->lock);
3845 if (block_group->removed) {
3846 spin_unlock(&block_group->lock);
3847 return 0;
3848 }
3849 btrfs_get_block_group_trimming(block_group);
3850 spin_unlock(&block_group->lock);
3851
3852 ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, async);
3853 btrfs_put_block_group_trimming(block_group);
3854
3855 return ret;
3856 }
3857
3858 int btrfs_trim_block_group_bitmaps(struct btrfs_block_group *block_group,
3859 u64 *trimmed, u64 start, u64 end, u64 minlen,
3860 u64 maxlen, bool async)
3861 {
3862 int ret;
3863
3864 *trimmed = 0;
3865
3866 spin_lock(&block_group->lock);
3867 if (block_group->removed) {
3868 spin_unlock(&block_group->lock);
3869 return 0;
3870 }
3871 btrfs_get_block_group_trimming(block_group);
3872 spin_unlock(&block_group->lock);
3873
3874 ret = trim_bitmaps(block_group, trimmed, start, end, minlen, maxlen,
3875 async);
3876
3877 btrfs_put_block_group_trimming(block_group);
3878
3879 return ret;
3880 }
3881
3882 /*
3883 * Find the left-most item in the cache tree, and then return the
3884 * smallest inode number in the item.
3885 *
3886 * Note: the returned inode number may not be the smallest one in
3887 * the tree, if the left-most item is a bitmap.
3888 */
3889 u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
3890 {
3891 struct btrfs_free_space_ctl *ctl = fs_root->free_ino_ctl;
3892 struct btrfs_free_space *entry = NULL;
3893 u64 ino = 0;
3894
3895 spin_lock(&ctl->tree_lock);
3896
3897 if (RB_EMPTY_ROOT(&ctl->free_space_offset))
3898 goto out;
3899
3900 entry = rb_entry(rb_first(&ctl->free_space_offset),
3901 struct btrfs_free_space, offset_index);
3902
3903 if (!entry->bitmap) {
3904 ino = entry->offset;
3905
3906 unlink_free_space(ctl, entry);
3907 entry->offset++;
3908 entry->bytes--;
3909 if (!entry->bytes)
3910 kmem_cache_free(btrfs_free_space_cachep, entry);
3911 else
3912 link_free_space(ctl, entry);
3913 } else {
3914 u64 offset = 0;
3915 u64 count = 1;
3916 int ret;
3917
3918 ret = search_bitmap(ctl, entry, &offset, &count, true);
3919 /* Logic error; Should be empty if it can't find anything */
3920 ASSERT(!ret);
3921
3922 ino = offset;
3923 bitmap_clear_bits(ctl, entry, offset, 1);
3924 if (entry->bytes == 0)
3925 free_bitmap(ctl, entry);
3926 }
3927 out:
3928 spin_unlock(&ctl->tree_lock);
3929
3930 return ino;
3931 }
3932
3933 struct inode *lookup_free_ino_inode(struct btrfs_root *root,
3934 struct btrfs_path *path)
3935 {
3936 struct inode *inode = NULL;
3937
3938 spin_lock(&root->ino_cache_lock);
3939 if (root->ino_cache_inode)
3940 inode = igrab(root->ino_cache_inode);
3941 spin_unlock(&root->ino_cache_lock);
3942 if (inode)
3943 return inode;
3944
3945 inode = __lookup_free_space_inode(root, path, 0);
3946 if (IS_ERR(inode))
3947 return inode;
3948
3949 spin_lock(&root->ino_cache_lock);
3950 if (!btrfs_fs_closing(root->fs_info))
3951 root->ino_cache_inode = igrab(inode);
3952 spin_unlock(&root->ino_cache_lock);
3953
3954 return inode;
3955 }
3956
3957 int create_free_ino_inode(struct btrfs_root *root,
3958 struct btrfs_trans_handle *trans,
3959 struct btrfs_path *path)
3960 {
3961 return __create_free_space_inode(root, trans, path,
3962 BTRFS_FREE_INO_OBJECTID, 0);
3963 }
3964
3965 int load_free_ino_cache(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
3966 {
3967 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3968 struct btrfs_path *path;
3969 struct inode *inode;
3970 int ret = 0;
3971 u64 root_gen = btrfs_root_generation(&root->root_item);
3972
3973 if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
3974 return 0;
3975
3976 /*
3977 * If we're unmounting then just return, since this does a search on the
3978 * normal root and not the commit root and we could deadlock.
3979 */
3980 if (btrfs_fs_closing(fs_info))
3981 return 0;
3982
3983 path = btrfs_alloc_path();
3984 if (!path)
3985 return 0;
3986
3987 inode = lookup_free_ino_inode(root, path);
3988 if (IS_ERR(inode))
3989 goto out;
3990
3991 if (root_gen != BTRFS_I(inode)->generation)
3992 goto out_put;
3993
3994 ret = __load_free_space_cache(root, inode, ctl, path, 0);
3995
3996 if (ret < 0)
3997 btrfs_err(fs_info,
3998 "failed to load free ino cache for root %llu",
3999 root->root_key.objectid);
4000 out_put:
4001 iput(inode);
4002 out:
4003 btrfs_free_path(path);
4004 return ret;
4005 }
4006
4007 int btrfs_write_out_ino_cache(struct btrfs_root *root,
4008 struct btrfs_trans_handle *trans,
4009 struct btrfs_path *path,
4010 struct inode *inode)
4011 {
4012 struct btrfs_fs_info *fs_info = root->fs_info;
4013 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
4014 int ret;
4015 struct btrfs_io_ctl io_ctl;
4016 bool release_metadata = true;
4017
4018 if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
4019 return 0;
4020
4021 memset(&io_ctl, 0, sizeof(io_ctl));
4022 ret = __btrfs_write_out_cache(root, inode, ctl, NULL, &io_ctl, trans);
4023 if (!ret) {
4024 /*
4025 * At this point writepages() didn't error out, so our metadata
4026 * reservation is released when the writeback finishes, at
4027 * inode.c:btrfs_finish_ordered_io(), regardless of it finishing
4028 * with or without an error.
4029 */
4030 release_metadata = false;
4031 ret = btrfs_wait_cache_io_root(root, trans, &io_ctl, path);
4032 }
4033
4034 if (ret) {
4035 if (release_metadata)
4036 btrfs_delalloc_release_metadata(BTRFS_I(inode),
4037 inode->i_size, true);
4038 #ifdef CONFIG_BTRFS_DEBUG
4039 btrfs_err(fs_info,
4040 "failed to write free ino cache for root %llu",
4041 root->root_key.objectid);
4042 #endif
4043 }
4044
4045 return ret;
4046 }
4047
4048 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
4049 /*
4050 * Use this if you need to make a bitmap or extent entry specifically, it
4051 * doesn't do any of the merging that add_free_space does, this acts a lot like
4052 * how the free space cache loading stuff works, so you can get really weird
4053 * configurations.
4054 */
4055 int test_add_free_space_entry(struct btrfs_block_group *cache,
4056 u64 offset, u64 bytes, bool bitmap)
4057 {
4058 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
4059 struct btrfs_free_space *info = NULL, *bitmap_info;
4060 void *map = NULL;
4061 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_TRIMMED;
4062 u64 bytes_added;
4063 int ret;
4064
4065 again:
4066 if (!info) {
4067 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
4068 if (!info)
4069 return -ENOMEM;
4070 }
4071
4072 if (!bitmap) {
4073 spin_lock(&ctl->tree_lock);
4074 info->offset = offset;
4075 info->bytes = bytes;
4076 info->max_extent_size = 0;
4077 ret = link_free_space(ctl, info);
4078 spin_unlock(&ctl->tree_lock);
4079 if (ret)
4080 kmem_cache_free(btrfs_free_space_cachep, info);
4081 return ret;
4082 }
4083
4084 if (!map) {
4085 map = kmem_cache_zalloc(btrfs_free_space_bitmap_cachep, GFP_NOFS);
4086 if (!map) {
4087 kmem_cache_free(btrfs_free_space_cachep, info);
4088 return -ENOMEM;
4089 }
4090 }
4091
4092 spin_lock(&ctl->tree_lock);
4093 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
4094 1, 0);
4095 if (!bitmap_info) {
4096 info->bitmap = map;
4097 map = NULL;
4098 add_new_bitmap(ctl, info, offset);
4099 bitmap_info = info;
4100 info = NULL;
4101 }
4102
4103 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes,
4104 trim_state);
4105
4106 bytes -= bytes_added;
4107 offset += bytes_added;
4108 spin_unlock(&ctl->tree_lock);
4109
4110 if (bytes)
4111 goto again;
4112
4113 if (info)
4114 kmem_cache_free(btrfs_free_space_cachep, info);
4115 if (map)
4116 kmem_cache_free(btrfs_free_space_bitmap_cachep, map);
4117 return 0;
4118 }
4119
4120 /*
4121 * Checks to see if the given range is in the free space cache. This is really
4122 * just used to check the absence of space, so if there is free space in the
4123 * range at all we will return 1.
4124 */
4125 int test_check_exists(struct btrfs_block_group *cache,
4126 u64 offset, u64 bytes)
4127 {
4128 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
4129 struct btrfs_free_space *info;
4130 int ret = 0;
4131
4132 spin_lock(&ctl->tree_lock);
4133 info = tree_search_offset(ctl, offset, 0, 0);
4134 if (!info) {
4135 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
4136 1, 0);
4137 if (!info)
4138 goto out;
4139 }
4140
4141 have_info:
4142 if (info->bitmap) {
4143 u64 bit_off, bit_bytes;
4144 struct rb_node *n;
4145 struct btrfs_free_space *tmp;
4146
4147 bit_off = offset;
4148 bit_bytes = ctl->unit;
4149 ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
4150 if (!ret) {
4151 if (bit_off == offset) {
4152 ret = 1;
4153 goto out;
4154 } else if (bit_off > offset &&
4155 offset + bytes > bit_off) {
4156 ret = 1;
4157 goto out;
4158 }
4159 }
4160
4161 n = rb_prev(&info->offset_index);
4162 while (n) {
4163 tmp = rb_entry(n, struct btrfs_free_space,
4164 offset_index);
4165 if (tmp->offset + tmp->bytes < offset)
4166 break;
4167 if (offset + bytes < tmp->offset) {
4168 n = rb_prev(&tmp->offset_index);
4169 continue;
4170 }
4171 info = tmp;
4172 goto have_info;
4173 }
4174
4175 n = rb_next(&info->offset_index);
4176 while (n) {
4177 tmp = rb_entry(n, struct btrfs_free_space,
4178 offset_index);
4179 if (offset + bytes < tmp->offset)
4180 break;
4181 if (tmp->offset + tmp->bytes < offset) {
4182 n = rb_next(&tmp->offset_index);
4183 continue;
4184 }
4185 info = tmp;
4186 goto have_info;
4187 }
4188
4189 ret = 0;
4190 goto out;
4191 }
4192
4193 if (info->offset == offset) {
4194 ret = 1;
4195 goto out;
4196 }
4197
4198 if (offset > info->offset && offset < info->offset + info->bytes)
4199 ret = 1;
4200 out:
4201 spin_unlock(&ctl->tree_lock);
4202 return ret;
4203 }
4204 #endif /* CONFIG_BTRFS_FS_RUN_SANITY_TESTS */
Linux with added FPC-III support