Loading...
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 "misc.h"
15#include "ctree.h"
16#include "free-space-cache.h"
17#include "transaction.h"
18#include "disk-io.h"
19#include "extent_io.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
30struct btrfs_trim_range {
31 u64 start;
32 u64 bytes;
33 struct list_head list;
34};
35
36static int link_free_space(struct btrfs_free_space_ctl *ctl,
37 struct btrfs_free_space *info);
38static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
39 struct btrfs_free_space *info);
40static int search_bitmap(struct btrfs_free_space_ctl *ctl,
41 struct btrfs_free_space *bitmap_info, u64 *offset,
42 u64 *bytes, bool for_alloc);
43static void free_bitmap(struct btrfs_free_space_ctl *ctl,
44 struct btrfs_free_space *bitmap_info);
45static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
46 struct btrfs_free_space *info, u64 offset,
47 u64 bytes);
48
49static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
50 struct btrfs_path *path,
51 u64 offset)
52{
53 struct btrfs_fs_info *fs_info = root->fs_info;
54 struct btrfs_key key;
55 struct btrfs_key location;
56 struct btrfs_disk_key disk_key;
57 struct btrfs_free_space_header *header;
58 struct extent_buffer *leaf;
59 struct inode *inode = NULL;
60 unsigned nofs_flag;
61 int ret;
62
63 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
64 key.offset = offset;
65 key.type = 0;
66
67 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
68 if (ret < 0)
69 return ERR_PTR(ret);
70 if (ret > 0) {
71 btrfs_release_path(path);
72 return ERR_PTR(-ENOENT);
73 }
74
75 leaf = path->nodes[0];
76 header = btrfs_item_ptr(leaf, path->slots[0],
77 struct btrfs_free_space_header);
78 btrfs_free_space_key(leaf, header, &disk_key);
79 btrfs_disk_key_to_cpu(&location, &disk_key);
80 btrfs_release_path(path);
81
82 /*
83 * We are often under a trans handle at this point, so we need to make
84 * sure NOFS is set to keep us from deadlocking.
85 */
86 nofs_flag = memalloc_nofs_save();
87 inode = btrfs_iget_path(fs_info->sb, location.objectid, root, path);
88 btrfs_release_path(path);
89 memalloc_nofs_restore(nofs_flag);
90 if (IS_ERR(inode))
91 return inode;
92
93 mapping_set_gfp_mask(inode->i_mapping,
94 mapping_gfp_constraint(inode->i_mapping,
95 ~(__GFP_FS | __GFP_HIGHMEM)));
96
97 return inode;
98}
99
100struct inode *lookup_free_space_inode(struct btrfs_block_group *block_group,
101 struct btrfs_path *path)
102{
103 struct btrfs_fs_info *fs_info = block_group->fs_info;
104 struct inode *inode = NULL;
105 u32 flags = BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
106
107 spin_lock(&block_group->lock);
108 if (block_group->inode)
109 inode = igrab(block_group->inode);
110 spin_unlock(&block_group->lock);
111 if (inode)
112 return inode;
113
114 inode = __lookup_free_space_inode(fs_info->tree_root, path,
115 block_group->start);
116 if (IS_ERR(inode))
117 return inode;
118
119 spin_lock(&block_group->lock);
120 if (!((BTRFS_I(inode)->flags & flags) == flags)) {
121 btrfs_info(fs_info, "Old style space inode found, converting.");
122 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM |
123 BTRFS_INODE_NODATACOW;
124 block_group->disk_cache_state = BTRFS_DC_CLEAR;
125 }
126
127 if (!block_group->iref) {
128 block_group->inode = igrab(inode);
129 block_group->iref = 1;
130 }
131 spin_unlock(&block_group->lock);
132
133 return inode;
134}
135
136static int __create_free_space_inode(struct btrfs_root *root,
137 struct btrfs_trans_handle *trans,
138 struct btrfs_path *path,
139 u64 ino, u64 offset)
140{
141 struct btrfs_key key;
142 struct btrfs_disk_key disk_key;
143 struct btrfs_free_space_header *header;
144 struct btrfs_inode_item *inode_item;
145 struct extent_buffer *leaf;
146 /* We inline CRCs for the free disk space cache */
147 const u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC |
148 BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
149 int ret;
150
151 ret = btrfs_insert_empty_inode(trans, root, path, ino);
152 if (ret)
153 return ret;
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
195int 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_get_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/*
211 * inode is an optional sink: if it is NULL, btrfs_remove_free_space_inode
212 * handles lookup, otherwise it takes ownership and iputs the inode.
213 * Don't reuse an inode pointer after passing it into this function.
214 */
215int btrfs_remove_free_space_inode(struct btrfs_trans_handle *trans,
216 struct inode *inode,
217 struct btrfs_block_group *block_group)
218{
219 struct btrfs_path *path;
220 struct btrfs_key key;
221 int ret = 0;
222
223 path = btrfs_alloc_path();
224 if (!path)
225 return -ENOMEM;
226
227 if (!inode)
228 inode = lookup_free_space_inode(block_group, path);
229 if (IS_ERR(inode)) {
230 if (PTR_ERR(inode) != -ENOENT)
231 ret = PTR_ERR(inode);
232 goto out;
233 }
234 ret = btrfs_orphan_add(trans, BTRFS_I(inode));
235 if (ret) {
236 btrfs_add_delayed_iput(inode);
237 goto out;
238 }
239 clear_nlink(inode);
240 /* One for the block groups ref */
241 spin_lock(&block_group->lock);
242 if (block_group->iref) {
243 block_group->iref = 0;
244 block_group->inode = NULL;
245 spin_unlock(&block_group->lock);
246 iput(inode);
247 } else {
248 spin_unlock(&block_group->lock);
249 }
250 /* One for the lookup ref */
251 btrfs_add_delayed_iput(inode);
252
253 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
254 key.type = 0;
255 key.offset = block_group->start;
256 ret = btrfs_search_slot(trans, trans->fs_info->tree_root, &key, path,
257 -1, 1);
258 if (ret) {
259 if (ret > 0)
260 ret = 0;
261 goto out;
262 }
263 ret = btrfs_del_item(trans, trans->fs_info->tree_root, path);
264out:
265 btrfs_free_path(path);
266 return ret;
267}
268
269int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
270 struct btrfs_block_rsv *rsv)
271{
272 u64 needed_bytes;
273 int ret;
274
275 /* 1 for slack space, 1 for updating the inode */
276 needed_bytes = btrfs_calc_insert_metadata_size(fs_info, 1) +
277 btrfs_calc_metadata_size(fs_info, 1);
278
279 spin_lock(&rsv->lock);
280 if (rsv->reserved < needed_bytes)
281 ret = -ENOSPC;
282 else
283 ret = 0;
284 spin_unlock(&rsv->lock);
285 return ret;
286}
287
288int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
289 struct btrfs_block_group *block_group,
290 struct inode *inode)
291{
292 struct btrfs_root *root = BTRFS_I(inode)->root;
293 int ret = 0;
294 bool locked = false;
295
296 if (block_group) {
297 struct btrfs_path *path = btrfs_alloc_path();
298
299 if (!path) {
300 ret = -ENOMEM;
301 goto fail;
302 }
303 locked = true;
304 mutex_lock(&trans->transaction->cache_write_mutex);
305 if (!list_empty(&block_group->io_list)) {
306 list_del_init(&block_group->io_list);
307
308 btrfs_wait_cache_io(trans, block_group, path);
309 btrfs_put_block_group(block_group);
310 }
311
312 /*
313 * now that we've truncated the cache away, its no longer
314 * setup or written
315 */
316 spin_lock(&block_group->lock);
317 block_group->disk_cache_state = BTRFS_DC_CLEAR;
318 spin_unlock(&block_group->lock);
319 btrfs_free_path(path);
320 }
321
322 btrfs_i_size_write(BTRFS_I(inode), 0);
323 truncate_pagecache(inode, 0);
324
325 /*
326 * We skip the throttling logic for free space cache inodes, so we don't
327 * need to check for -EAGAIN.
328 */
329 ret = btrfs_truncate_inode_items(trans, root, BTRFS_I(inode),
330 0, BTRFS_EXTENT_DATA_KEY, NULL);
331 if (ret)
332 goto fail;
333
334 ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
335
336fail:
337 if (locked)
338 mutex_unlock(&trans->transaction->cache_write_mutex);
339 if (ret)
340 btrfs_abort_transaction(trans, ret);
341
342 return ret;
343}
344
345static void readahead_cache(struct inode *inode)
346{
347 struct file_ra_state *ra;
348 unsigned long last_index;
349
350 ra = kzalloc(sizeof(*ra), GFP_NOFS);
351 if (!ra)
352 return;
353
354 file_ra_state_init(ra, inode->i_mapping);
355 last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
356
357 page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
358
359 kfree(ra);
360}
361
362static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
363 int write)
364{
365 int num_pages;
366
367 num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
368
369 /* Make sure we can fit our crcs and generation into the first page */
370 if (write && (num_pages * sizeof(u32) + sizeof(u64)) > PAGE_SIZE)
371 return -ENOSPC;
372
373 memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
374
375 io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
376 if (!io_ctl->pages)
377 return -ENOMEM;
378
379 io_ctl->num_pages = num_pages;
380 io_ctl->fs_info = btrfs_sb(inode->i_sb);
381 io_ctl->inode = inode;
382
383 return 0;
384}
385ALLOW_ERROR_INJECTION(io_ctl_init, ERRNO);
386
387static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
388{
389 kfree(io_ctl->pages);
390 io_ctl->pages = NULL;
391}
392
393static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
394{
395 if (io_ctl->cur) {
396 io_ctl->cur = NULL;
397 io_ctl->orig = NULL;
398 }
399}
400
401static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
402{
403 ASSERT(io_ctl->index < io_ctl->num_pages);
404 io_ctl->page = io_ctl->pages[io_ctl->index++];
405 io_ctl->cur = page_address(io_ctl->page);
406 io_ctl->orig = io_ctl->cur;
407 io_ctl->size = PAGE_SIZE;
408 if (clear)
409 clear_page(io_ctl->cur);
410}
411
412static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
413{
414 int i;
415
416 io_ctl_unmap_page(io_ctl);
417
418 for (i = 0; i < io_ctl->num_pages; i++) {
419 if (io_ctl->pages[i]) {
420 ClearPageChecked(io_ctl->pages[i]);
421 unlock_page(io_ctl->pages[i]);
422 put_page(io_ctl->pages[i]);
423 }
424 }
425}
426
427static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, bool uptodate)
428{
429 struct page *page;
430 struct inode *inode = io_ctl->inode;
431 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
432 int i;
433
434 for (i = 0; i < io_ctl->num_pages; i++) {
435 int ret;
436
437 page = find_or_create_page(inode->i_mapping, i, mask);
438 if (!page) {
439 io_ctl_drop_pages(io_ctl);
440 return -ENOMEM;
441 }
442
443 ret = set_page_extent_mapped(page);
444 if (ret < 0) {
445 unlock_page(page);
446 put_page(page);
447 io_ctl_drop_pages(io_ctl);
448 return ret;
449 }
450
451 io_ctl->pages[i] = page;
452 if (uptodate && !PageUptodate(page)) {
453 btrfs_readpage(NULL, page);
454 lock_page(page);
455 if (page->mapping != inode->i_mapping) {
456 btrfs_err(BTRFS_I(inode)->root->fs_info,
457 "free space cache page truncated");
458 io_ctl_drop_pages(io_ctl);
459 return -EIO;
460 }
461 if (!PageUptodate(page)) {
462 btrfs_err(BTRFS_I(inode)->root->fs_info,
463 "error reading free space cache");
464 io_ctl_drop_pages(io_ctl);
465 return -EIO;
466 }
467 }
468 }
469
470 for (i = 0; i < io_ctl->num_pages; i++)
471 clear_page_dirty_for_io(io_ctl->pages[i]);
472
473 return 0;
474}
475
476static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
477{
478 io_ctl_map_page(io_ctl, 1);
479
480 /*
481 * Skip the csum areas. If we don't check crcs then we just have a
482 * 64bit chunk at the front of the first page.
483 */
484 io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
485 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
486
487 put_unaligned_le64(generation, io_ctl->cur);
488 io_ctl->cur += sizeof(u64);
489}
490
491static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
492{
493 u64 cache_gen;
494
495 /*
496 * Skip the crc area. If we don't check crcs then we just have a 64bit
497 * chunk at the front of the first page.
498 */
499 io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
500 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
501
502 cache_gen = get_unaligned_le64(io_ctl->cur);
503 if (cache_gen != generation) {
504 btrfs_err_rl(io_ctl->fs_info,
505 "space cache generation (%llu) does not match inode (%llu)",
506 cache_gen, generation);
507 io_ctl_unmap_page(io_ctl);
508 return -EIO;
509 }
510 io_ctl->cur += sizeof(u64);
511 return 0;
512}
513
514static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
515{
516 u32 *tmp;
517 u32 crc = ~(u32)0;
518 unsigned offset = 0;
519
520 if (index == 0)
521 offset = sizeof(u32) * io_ctl->num_pages;
522
523 crc = btrfs_crc32c(crc, io_ctl->orig + offset, PAGE_SIZE - offset);
524 btrfs_crc32c_final(crc, (u8 *)&crc);
525 io_ctl_unmap_page(io_ctl);
526 tmp = page_address(io_ctl->pages[0]);
527 tmp += index;
528 *tmp = crc;
529}
530
531static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
532{
533 u32 *tmp, val;
534 u32 crc = ~(u32)0;
535 unsigned offset = 0;
536
537 if (index == 0)
538 offset = sizeof(u32) * io_ctl->num_pages;
539
540 tmp = page_address(io_ctl->pages[0]);
541 tmp += index;
542 val = *tmp;
543
544 io_ctl_map_page(io_ctl, 0);
545 crc = btrfs_crc32c(crc, io_ctl->orig + offset, PAGE_SIZE - offset);
546 btrfs_crc32c_final(crc, (u8 *)&crc);
547 if (val != crc) {
548 btrfs_err_rl(io_ctl->fs_info,
549 "csum mismatch on free space cache");
550 io_ctl_unmap_page(io_ctl);
551 return -EIO;
552 }
553
554 return 0;
555}
556
557static int io_ctl_add_entry(struct btrfs_io_ctl *io_ctl, u64 offset, u64 bytes,
558 void *bitmap)
559{
560 struct btrfs_free_space_entry *entry;
561
562 if (!io_ctl->cur)
563 return -ENOSPC;
564
565 entry = io_ctl->cur;
566 put_unaligned_le64(offset, &entry->offset);
567 put_unaligned_le64(bytes, &entry->bytes);
568 entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
569 BTRFS_FREE_SPACE_EXTENT;
570 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
571 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
572
573 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
574 return 0;
575
576 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
577
578 /* No more pages to map */
579 if (io_ctl->index >= io_ctl->num_pages)
580 return 0;
581
582 /* map the next page */
583 io_ctl_map_page(io_ctl, 1);
584 return 0;
585}
586
587static int io_ctl_add_bitmap(struct btrfs_io_ctl *io_ctl, void *bitmap)
588{
589 if (!io_ctl->cur)
590 return -ENOSPC;
591
592 /*
593 * If we aren't at the start of the current page, unmap this one and
594 * map the next one if there is any left.
595 */
596 if (io_ctl->cur != io_ctl->orig) {
597 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
598 if (io_ctl->index >= io_ctl->num_pages)
599 return -ENOSPC;
600 io_ctl_map_page(io_ctl, 0);
601 }
602
603 copy_page(io_ctl->cur, bitmap);
604 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
605 if (io_ctl->index < io_ctl->num_pages)
606 io_ctl_map_page(io_ctl, 0);
607 return 0;
608}
609
610static void io_ctl_zero_remaining_pages(struct btrfs_io_ctl *io_ctl)
611{
612 /*
613 * If we're not on the boundary we know we've modified the page and we
614 * need to crc the page.
615 */
616 if (io_ctl->cur != io_ctl->orig)
617 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
618 else
619 io_ctl_unmap_page(io_ctl);
620
621 while (io_ctl->index < io_ctl->num_pages) {
622 io_ctl_map_page(io_ctl, 1);
623 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
624 }
625}
626
627static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
628 struct btrfs_free_space *entry, u8 *type)
629{
630 struct btrfs_free_space_entry *e;
631 int ret;
632
633 if (!io_ctl->cur) {
634 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
635 if (ret)
636 return ret;
637 }
638
639 e = io_ctl->cur;
640 entry->offset = get_unaligned_le64(&e->offset);
641 entry->bytes = get_unaligned_le64(&e->bytes);
642 *type = e->type;
643 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
644 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
645
646 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
647 return 0;
648
649 io_ctl_unmap_page(io_ctl);
650
651 return 0;
652}
653
654static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
655 struct btrfs_free_space *entry)
656{
657 int ret;
658
659 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
660 if (ret)
661 return ret;
662
663 copy_page(entry->bitmap, io_ctl->cur);
664 io_ctl_unmap_page(io_ctl);
665
666 return 0;
667}
668
669static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
670{
671 struct btrfs_block_group *block_group = ctl->private;
672 u64 max_bytes;
673 u64 bitmap_bytes;
674 u64 extent_bytes;
675 u64 size = block_group->length;
676 u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
677 u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
678
679 max_bitmaps = max_t(u64, max_bitmaps, 1);
680
681 ASSERT(ctl->total_bitmaps <= max_bitmaps);
682
683 /*
684 * We are trying to keep the total amount of memory used per 1GiB of
685 * space to be MAX_CACHE_BYTES_PER_GIG. However, with a reclamation
686 * mechanism of pulling extents >= FORCE_EXTENT_THRESHOLD out of
687 * bitmaps, we may end up using more memory than this.
688 */
689 if (size < SZ_1G)
690 max_bytes = MAX_CACHE_BYTES_PER_GIG;
691 else
692 max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
693
694 bitmap_bytes = ctl->total_bitmaps * ctl->unit;
695
696 /*
697 * we want the extent entry threshold to always be at most 1/2 the max
698 * bytes we can have, or whatever is less than that.
699 */
700 extent_bytes = max_bytes - bitmap_bytes;
701 extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
702
703 ctl->extents_thresh =
704 div_u64(extent_bytes, sizeof(struct btrfs_free_space));
705}
706
707static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
708 struct btrfs_free_space_ctl *ctl,
709 struct btrfs_path *path, u64 offset)
710{
711 struct btrfs_fs_info *fs_info = root->fs_info;
712 struct btrfs_free_space_header *header;
713 struct extent_buffer *leaf;
714 struct btrfs_io_ctl io_ctl;
715 struct btrfs_key key;
716 struct btrfs_free_space *e, *n;
717 LIST_HEAD(bitmaps);
718 u64 num_entries;
719 u64 num_bitmaps;
720 u64 generation;
721 u8 type;
722 int ret = 0;
723
724 /* Nothing in the space cache, goodbye */
725 if (!i_size_read(inode))
726 return 0;
727
728 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
729 key.offset = offset;
730 key.type = 0;
731
732 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
733 if (ret < 0)
734 return 0;
735 else if (ret > 0) {
736 btrfs_release_path(path);
737 return 0;
738 }
739
740 ret = -1;
741
742 leaf = path->nodes[0];
743 header = btrfs_item_ptr(leaf, path->slots[0],
744 struct btrfs_free_space_header);
745 num_entries = btrfs_free_space_entries(leaf, header);
746 num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
747 generation = btrfs_free_space_generation(leaf, header);
748 btrfs_release_path(path);
749
750 if (!BTRFS_I(inode)->generation) {
751 btrfs_info(fs_info,
752 "the free space cache file (%llu) is invalid, skip it",
753 offset);
754 return 0;
755 }
756
757 if (BTRFS_I(inode)->generation != generation) {
758 btrfs_err(fs_info,
759 "free space inode generation (%llu) did not match free space cache generation (%llu)",
760 BTRFS_I(inode)->generation, generation);
761 return 0;
762 }
763
764 if (!num_entries)
765 return 0;
766
767 ret = io_ctl_init(&io_ctl, inode, 0);
768 if (ret)
769 return ret;
770
771 readahead_cache(inode);
772
773 ret = io_ctl_prepare_pages(&io_ctl, true);
774 if (ret)
775 goto out;
776
777 ret = io_ctl_check_crc(&io_ctl, 0);
778 if (ret)
779 goto free_cache;
780
781 ret = io_ctl_check_generation(&io_ctl, generation);
782 if (ret)
783 goto free_cache;
784
785 while (num_entries) {
786 e = kmem_cache_zalloc(btrfs_free_space_cachep,
787 GFP_NOFS);
788 if (!e) {
789 ret = -ENOMEM;
790 goto free_cache;
791 }
792
793 ret = io_ctl_read_entry(&io_ctl, e, &type);
794 if (ret) {
795 kmem_cache_free(btrfs_free_space_cachep, e);
796 goto free_cache;
797 }
798
799 if (!e->bytes) {
800 ret = -1;
801 kmem_cache_free(btrfs_free_space_cachep, e);
802 goto free_cache;
803 }
804
805 if (type == BTRFS_FREE_SPACE_EXTENT) {
806 spin_lock(&ctl->tree_lock);
807 ret = link_free_space(ctl, e);
808 spin_unlock(&ctl->tree_lock);
809 if (ret) {
810 btrfs_err(fs_info,
811 "Duplicate entries in free space cache, dumping");
812 kmem_cache_free(btrfs_free_space_cachep, e);
813 goto free_cache;
814 }
815 } else {
816 ASSERT(num_bitmaps);
817 num_bitmaps--;
818 e->bitmap = kmem_cache_zalloc(
819 btrfs_free_space_bitmap_cachep, GFP_NOFS);
820 if (!e->bitmap) {
821 ret = -ENOMEM;
822 kmem_cache_free(
823 btrfs_free_space_cachep, e);
824 goto free_cache;
825 }
826 spin_lock(&ctl->tree_lock);
827 ret = link_free_space(ctl, e);
828 ctl->total_bitmaps++;
829 recalculate_thresholds(ctl);
830 spin_unlock(&ctl->tree_lock);
831 if (ret) {
832 btrfs_err(fs_info,
833 "Duplicate entries in free space cache, dumping");
834 kmem_cache_free(btrfs_free_space_cachep, e);
835 goto free_cache;
836 }
837 list_add_tail(&e->list, &bitmaps);
838 }
839
840 num_entries--;
841 }
842
843 io_ctl_unmap_page(&io_ctl);
844
845 /*
846 * We add the bitmaps at the end of the entries in order that
847 * the bitmap entries are added to the cache.
848 */
849 list_for_each_entry_safe(e, n, &bitmaps, list) {
850 list_del_init(&e->list);
851 ret = io_ctl_read_bitmap(&io_ctl, e);
852 if (ret)
853 goto free_cache;
854 }
855
856 io_ctl_drop_pages(&io_ctl);
857 ret = 1;
858out:
859 io_ctl_free(&io_ctl);
860 return ret;
861free_cache:
862 io_ctl_drop_pages(&io_ctl);
863 __btrfs_remove_free_space_cache(ctl);
864 goto out;
865}
866
867static int copy_free_space_cache(struct btrfs_block_group *block_group,
868 struct btrfs_free_space_ctl *ctl)
869{
870 struct btrfs_free_space *info;
871 struct rb_node *n;
872 int ret = 0;
873
874 while (!ret && (n = rb_first(&ctl->free_space_offset)) != NULL) {
875 info = rb_entry(n, struct btrfs_free_space, offset_index);
876 if (!info->bitmap) {
877 unlink_free_space(ctl, info);
878 ret = btrfs_add_free_space(block_group, info->offset,
879 info->bytes);
880 kmem_cache_free(btrfs_free_space_cachep, info);
881 } else {
882 u64 offset = info->offset;
883 u64 bytes = ctl->unit;
884
885 while (search_bitmap(ctl, info, &offset, &bytes,
886 false) == 0) {
887 ret = btrfs_add_free_space(block_group, offset,
888 bytes);
889 if (ret)
890 break;
891 bitmap_clear_bits(ctl, info, offset, bytes);
892 offset = info->offset;
893 bytes = ctl->unit;
894 }
895 free_bitmap(ctl, info);
896 }
897 cond_resched();
898 }
899 return ret;
900}
901
902int load_free_space_cache(struct btrfs_block_group *block_group)
903{
904 struct btrfs_fs_info *fs_info = block_group->fs_info;
905 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
906 struct btrfs_free_space_ctl tmp_ctl = {};
907 struct inode *inode;
908 struct btrfs_path *path;
909 int ret = 0;
910 bool matched;
911 u64 used = block_group->used;
912
913 /*
914 * Because we could potentially discard our loaded free space, we want
915 * to load everything into a temporary structure first, and then if it's
916 * valid copy it all into the actual free space ctl.
917 */
918 btrfs_init_free_space_ctl(block_group, &tmp_ctl);
919
920 /*
921 * If this block group has been marked to be cleared for one reason or
922 * another then we can't trust the on disk cache, so just return.
923 */
924 spin_lock(&block_group->lock);
925 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
926 spin_unlock(&block_group->lock);
927 return 0;
928 }
929 spin_unlock(&block_group->lock);
930
931 path = btrfs_alloc_path();
932 if (!path)
933 return 0;
934 path->search_commit_root = 1;
935 path->skip_locking = 1;
936
937 /*
938 * We must pass a path with search_commit_root set to btrfs_iget in
939 * order to avoid a deadlock when allocating extents for the tree root.
940 *
941 * When we are COWing an extent buffer from the tree root, when looking
942 * for a free extent, at extent-tree.c:find_free_extent(), we can find
943 * block group without its free space cache loaded. When we find one
944 * we must load its space cache which requires reading its free space
945 * cache's inode item from the root tree. If this inode item is located
946 * in the same leaf that we started COWing before, then we end up in
947 * deadlock on the extent buffer (trying to read lock it when we
948 * previously write locked it).
949 *
950 * It's safe to read the inode item using the commit root because
951 * block groups, once loaded, stay in memory forever (until they are
952 * removed) as well as their space caches once loaded. New block groups
953 * once created get their ->cached field set to BTRFS_CACHE_FINISHED so
954 * we will never try to read their inode item while the fs is mounted.
955 */
956 inode = lookup_free_space_inode(block_group, path);
957 if (IS_ERR(inode)) {
958 btrfs_free_path(path);
959 return 0;
960 }
961
962 /* We may have converted the inode and made the cache invalid. */
963 spin_lock(&block_group->lock);
964 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
965 spin_unlock(&block_group->lock);
966 btrfs_free_path(path);
967 goto out;
968 }
969 spin_unlock(&block_group->lock);
970
971 ret = __load_free_space_cache(fs_info->tree_root, inode, &tmp_ctl,
972 path, block_group->start);
973 btrfs_free_path(path);
974 if (ret <= 0)
975 goto out;
976
977 matched = (tmp_ctl.free_space == (block_group->length - used -
978 block_group->bytes_super));
979
980 if (matched) {
981 ret = copy_free_space_cache(block_group, &tmp_ctl);
982 /*
983 * ret == 1 means we successfully loaded the free space cache,
984 * so we need to re-set it here.
985 */
986 if (ret == 0)
987 ret = 1;
988 } else {
989 __btrfs_remove_free_space_cache(&tmp_ctl);
990 btrfs_warn(fs_info,
991 "block group %llu has wrong amount of free space",
992 block_group->start);
993 ret = -1;
994 }
995out:
996 if (ret < 0) {
997 /* This cache is bogus, make sure it gets cleared */
998 spin_lock(&block_group->lock);
999 block_group->disk_cache_state = BTRFS_DC_CLEAR;
1000 spin_unlock(&block_group->lock);
1001 ret = 0;
1002
1003 btrfs_warn(fs_info,
1004 "failed to load free space cache for block group %llu, rebuilding it now",
1005 block_group->start);
1006 }
1007
1008 spin_lock(&ctl->tree_lock);
1009 btrfs_discard_update_discardable(block_group);
1010 spin_unlock(&ctl->tree_lock);
1011 iput(inode);
1012 return ret;
1013}
1014
1015static noinline_for_stack
1016int write_cache_extent_entries(struct btrfs_io_ctl *io_ctl,
1017 struct btrfs_free_space_ctl *ctl,
1018 struct btrfs_block_group *block_group,
1019 int *entries, int *bitmaps,
1020 struct list_head *bitmap_list)
1021{
1022 int ret;
1023 struct btrfs_free_cluster *cluster = NULL;
1024 struct btrfs_free_cluster *cluster_locked = NULL;
1025 struct rb_node *node = rb_first(&ctl->free_space_offset);
1026 struct btrfs_trim_range *trim_entry;
1027
1028 /* Get the cluster for this block_group if it exists */
1029 if (block_group && !list_empty(&block_group->cluster_list)) {
1030 cluster = list_entry(block_group->cluster_list.next,
1031 struct btrfs_free_cluster,
1032 block_group_list);
1033 }
1034
1035 if (!node && cluster) {
1036 cluster_locked = cluster;
1037 spin_lock(&cluster_locked->lock);
1038 node = rb_first(&cluster->root);
1039 cluster = NULL;
1040 }
1041
1042 /* Write out the extent entries */
1043 while (node) {
1044 struct btrfs_free_space *e;
1045
1046 e = rb_entry(node, struct btrfs_free_space, offset_index);
1047 *entries += 1;
1048
1049 ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes,
1050 e->bitmap);
1051 if (ret)
1052 goto fail;
1053
1054 if (e->bitmap) {
1055 list_add_tail(&e->list, bitmap_list);
1056 *bitmaps += 1;
1057 }
1058 node = rb_next(node);
1059 if (!node && cluster) {
1060 node = rb_first(&cluster->root);
1061 cluster_locked = cluster;
1062 spin_lock(&cluster_locked->lock);
1063 cluster = NULL;
1064 }
1065 }
1066 if (cluster_locked) {
1067 spin_unlock(&cluster_locked->lock);
1068 cluster_locked = NULL;
1069 }
1070
1071 /*
1072 * Make sure we don't miss any range that was removed from our rbtree
1073 * because trimming is running. Otherwise after a umount+mount (or crash
1074 * after committing the transaction) we would leak free space and get
1075 * an inconsistent free space cache report from fsck.
1076 */
1077 list_for_each_entry(trim_entry, &ctl->trimming_ranges, list) {
1078 ret = io_ctl_add_entry(io_ctl, trim_entry->start,
1079 trim_entry->bytes, NULL);
1080 if (ret)
1081 goto fail;
1082 *entries += 1;
1083 }
1084
1085 return 0;
1086fail:
1087 if (cluster_locked)
1088 spin_unlock(&cluster_locked->lock);
1089 return -ENOSPC;
1090}
1091
1092static noinline_for_stack int
1093update_cache_item(struct btrfs_trans_handle *trans,
1094 struct btrfs_root *root,
1095 struct inode *inode,
1096 struct btrfs_path *path, u64 offset,
1097 int entries, int bitmaps)
1098{
1099 struct btrfs_key key;
1100 struct btrfs_free_space_header *header;
1101 struct extent_buffer *leaf;
1102 int ret;
1103
1104 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
1105 key.offset = offset;
1106 key.type = 0;
1107
1108 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
1109 if (ret < 0) {
1110 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1111 EXTENT_DELALLOC, 0, 0, NULL);
1112 goto fail;
1113 }
1114 leaf = path->nodes[0];
1115 if (ret > 0) {
1116 struct btrfs_key found_key;
1117 ASSERT(path->slots[0]);
1118 path->slots[0]--;
1119 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1120 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
1121 found_key.offset != offset) {
1122 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
1123 inode->i_size - 1, EXTENT_DELALLOC, 0,
1124 0, NULL);
1125 btrfs_release_path(path);
1126 goto fail;
1127 }
1128 }
1129
1130 BTRFS_I(inode)->generation = trans->transid;
1131 header = btrfs_item_ptr(leaf, path->slots[0],
1132 struct btrfs_free_space_header);
1133 btrfs_set_free_space_entries(leaf, header, entries);
1134 btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1135 btrfs_set_free_space_generation(leaf, header, trans->transid);
1136 btrfs_mark_buffer_dirty(leaf);
1137 btrfs_release_path(path);
1138
1139 return 0;
1140
1141fail:
1142 return -1;
1143}
1144
1145static noinline_for_stack int write_pinned_extent_entries(
1146 struct btrfs_trans_handle *trans,
1147 struct btrfs_block_group *block_group,
1148 struct btrfs_io_ctl *io_ctl,
1149 int *entries)
1150{
1151 u64 start, extent_start, extent_end, len;
1152 struct extent_io_tree *unpin = NULL;
1153 int ret;
1154
1155 if (!block_group)
1156 return 0;
1157
1158 /*
1159 * We want to add any pinned extents to our free space cache
1160 * so we don't leak the space
1161 *
1162 * We shouldn't have switched the pinned extents yet so this is the
1163 * right one
1164 */
1165 unpin = &trans->transaction->pinned_extents;
1166
1167 start = block_group->start;
1168
1169 while (start < block_group->start + block_group->length) {
1170 ret = find_first_extent_bit(unpin, start,
1171 &extent_start, &extent_end,
1172 EXTENT_DIRTY, NULL);
1173 if (ret)
1174 return 0;
1175
1176 /* This pinned extent is out of our range */
1177 if (extent_start >= block_group->start + block_group->length)
1178 return 0;
1179
1180 extent_start = max(extent_start, start);
1181 extent_end = min(block_group->start + block_group->length,
1182 extent_end + 1);
1183 len = extent_end - extent_start;
1184
1185 *entries += 1;
1186 ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
1187 if (ret)
1188 return -ENOSPC;
1189
1190 start = extent_end;
1191 }
1192
1193 return 0;
1194}
1195
1196static noinline_for_stack int
1197write_bitmap_entries(struct btrfs_io_ctl *io_ctl, struct list_head *bitmap_list)
1198{
1199 struct btrfs_free_space *entry, *next;
1200 int ret;
1201
1202 /* Write out the bitmaps */
1203 list_for_each_entry_safe(entry, next, bitmap_list, list) {
1204 ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
1205 if (ret)
1206 return -ENOSPC;
1207 list_del_init(&entry->list);
1208 }
1209
1210 return 0;
1211}
1212
1213static int flush_dirty_cache(struct inode *inode)
1214{
1215 int ret;
1216
1217 ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
1218 if (ret)
1219 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1220 EXTENT_DELALLOC, 0, 0, NULL);
1221
1222 return ret;
1223}
1224
1225static void noinline_for_stack
1226cleanup_bitmap_list(struct list_head *bitmap_list)
1227{
1228 struct btrfs_free_space *entry, *next;
1229
1230 list_for_each_entry_safe(entry, next, bitmap_list, list)
1231 list_del_init(&entry->list);
1232}
1233
1234static void noinline_for_stack
1235cleanup_write_cache_enospc(struct inode *inode,
1236 struct btrfs_io_ctl *io_ctl,
1237 struct extent_state **cached_state)
1238{
1239 io_ctl_drop_pages(io_ctl);
1240 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1241 i_size_read(inode) - 1, cached_state);
1242}
1243
1244static int __btrfs_wait_cache_io(struct btrfs_root *root,
1245 struct btrfs_trans_handle *trans,
1246 struct btrfs_block_group *block_group,
1247 struct btrfs_io_ctl *io_ctl,
1248 struct btrfs_path *path, u64 offset)
1249{
1250 int ret;
1251 struct inode *inode = io_ctl->inode;
1252
1253 if (!inode)
1254 return 0;
1255
1256 /* Flush the dirty pages in the cache file. */
1257 ret = flush_dirty_cache(inode);
1258 if (ret)
1259 goto out;
1260
1261 /* Update the cache item to tell everyone this cache file is valid. */
1262 ret = update_cache_item(trans, root, inode, path, offset,
1263 io_ctl->entries, io_ctl->bitmaps);
1264out:
1265 if (ret) {
1266 invalidate_inode_pages2(inode->i_mapping);
1267 BTRFS_I(inode)->generation = 0;
1268 if (block_group)
1269 btrfs_debug(root->fs_info,
1270 "failed to write free space cache for block group %llu error %d",
1271 block_group->start, ret);
1272 }
1273 btrfs_update_inode(trans, root, BTRFS_I(inode));
1274
1275 if (block_group) {
1276 /* the dirty list is protected by the dirty_bgs_lock */
1277 spin_lock(&trans->transaction->dirty_bgs_lock);
1278
1279 /* the disk_cache_state is protected by the block group lock */
1280 spin_lock(&block_group->lock);
1281
1282 /*
1283 * only mark this as written if we didn't get put back on
1284 * the dirty list while waiting for IO. Otherwise our
1285 * cache state won't be right, and we won't get written again
1286 */
1287 if (!ret && list_empty(&block_group->dirty_list))
1288 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1289 else if (ret)
1290 block_group->disk_cache_state = BTRFS_DC_ERROR;
1291
1292 spin_unlock(&block_group->lock);
1293 spin_unlock(&trans->transaction->dirty_bgs_lock);
1294 io_ctl->inode = NULL;
1295 iput(inode);
1296 }
1297
1298 return ret;
1299
1300}
1301
1302int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
1303 struct btrfs_block_group *block_group,
1304 struct btrfs_path *path)
1305{
1306 return __btrfs_wait_cache_io(block_group->fs_info->tree_root, trans,
1307 block_group, &block_group->io_ctl,
1308 path, block_group->start);
1309}
1310
1311/**
1312 * Write out cached info to an inode
1313 *
1314 * @root: root the inode belongs to
1315 * @inode: freespace inode we are writing out
1316 * @ctl: free space cache we are going to write out
1317 * @block_group: block_group for this cache if it belongs to a block_group
1318 * @io_ctl: holds context for the io
1319 * @trans: the trans handle
1320 *
1321 * This function writes out a free space cache struct to disk for quick recovery
1322 * on mount. This will return 0 if it was successful in writing the cache out,
1323 * or an errno if it was not.
1324 */
1325static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
1326 struct btrfs_free_space_ctl *ctl,
1327 struct btrfs_block_group *block_group,
1328 struct btrfs_io_ctl *io_ctl,
1329 struct btrfs_trans_handle *trans)
1330{
1331 struct extent_state *cached_state = NULL;
1332 LIST_HEAD(bitmap_list);
1333 int entries = 0;
1334 int bitmaps = 0;
1335 int ret;
1336 int must_iput = 0;
1337
1338 if (!i_size_read(inode))
1339 return -EIO;
1340
1341 WARN_ON(io_ctl->pages);
1342 ret = io_ctl_init(io_ctl, inode, 1);
1343 if (ret)
1344 return ret;
1345
1346 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
1347 down_write(&block_group->data_rwsem);
1348 spin_lock(&block_group->lock);
1349 if (block_group->delalloc_bytes) {
1350 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1351 spin_unlock(&block_group->lock);
1352 up_write(&block_group->data_rwsem);
1353 BTRFS_I(inode)->generation = 0;
1354 ret = 0;
1355 must_iput = 1;
1356 goto out;
1357 }
1358 spin_unlock(&block_group->lock);
1359 }
1360
1361 /* Lock all pages first so we can lock the extent safely. */
1362 ret = io_ctl_prepare_pages(io_ctl, false);
1363 if (ret)
1364 goto out_unlock;
1365
1366 lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1367 &cached_state);
1368
1369 io_ctl_set_generation(io_ctl, trans->transid);
1370
1371 mutex_lock(&ctl->cache_writeout_mutex);
1372 /* Write out the extent entries in the free space cache */
1373 spin_lock(&ctl->tree_lock);
1374 ret = write_cache_extent_entries(io_ctl, ctl,
1375 block_group, &entries, &bitmaps,
1376 &bitmap_list);
1377 if (ret)
1378 goto out_nospc_locked;
1379
1380 /*
1381 * Some spaces that are freed in the current transaction are pinned,
1382 * they will be added into free space cache after the transaction is
1383 * committed, we shouldn't lose them.
1384 *
1385 * If this changes while we are working we'll get added back to
1386 * the dirty list and redo it. No locking needed
1387 */
1388 ret = write_pinned_extent_entries(trans, block_group, io_ctl, &entries);
1389 if (ret)
1390 goto out_nospc_locked;
1391
1392 /*
1393 * At last, we write out all the bitmaps and keep cache_writeout_mutex
1394 * locked while doing it because a concurrent trim can be manipulating
1395 * or freeing the bitmap.
1396 */
1397 ret = write_bitmap_entries(io_ctl, &bitmap_list);
1398 spin_unlock(&ctl->tree_lock);
1399 mutex_unlock(&ctl->cache_writeout_mutex);
1400 if (ret)
1401 goto out_nospc;
1402
1403 /* Zero out the rest of the pages just to make sure */
1404 io_ctl_zero_remaining_pages(io_ctl);
1405
1406 /* Everything is written out, now we dirty the pages in the file. */
1407 ret = btrfs_dirty_pages(BTRFS_I(inode), io_ctl->pages,
1408 io_ctl->num_pages, 0, i_size_read(inode),
1409 &cached_state, false);
1410 if (ret)
1411 goto out_nospc;
1412
1413 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1414 up_write(&block_group->data_rwsem);
1415 /*
1416 * Release the pages and unlock the extent, we will flush
1417 * them out later
1418 */
1419 io_ctl_drop_pages(io_ctl);
1420 io_ctl_free(io_ctl);
1421
1422 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1423 i_size_read(inode) - 1, &cached_state);
1424
1425 /*
1426 * at this point the pages are under IO and we're happy,
1427 * The caller is responsible for waiting on them and updating
1428 * the cache and the inode
1429 */
1430 io_ctl->entries = entries;
1431 io_ctl->bitmaps = bitmaps;
1432
1433 ret = btrfs_fdatawrite_range(inode, 0, (u64)-1);
1434 if (ret)
1435 goto out;
1436
1437 return 0;
1438
1439out_nospc_locked:
1440 cleanup_bitmap_list(&bitmap_list);
1441 spin_unlock(&ctl->tree_lock);
1442 mutex_unlock(&ctl->cache_writeout_mutex);
1443
1444out_nospc:
1445 cleanup_write_cache_enospc(inode, io_ctl, &cached_state);
1446
1447out_unlock:
1448 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1449 up_write(&block_group->data_rwsem);
1450
1451out:
1452 io_ctl->inode = NULL;
1453 io_ctl_free(io_ctl);
1454 if (ret) {
1455 invalidate_inode_pages2(inode->i_mapping);
1456 BTRFS_I(inode)->generation = 0;
1457 }
1458 btrfs_update_inode(trans, root, BTRFS_I(inode));
1459 if (must_iput)
1460 iput(inode);
1461 return ret;
1462}
1463
1464int btrfs_write_out_cache(struct btrfs_trans_handle *trans,
1465 struct btrfs_block_group *block_group,
1466 struct btrfs_path *path)
1467{
1468 struct btrfs_fs_info *fs_info = trans->fs_info;
1469 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1470 struct inode *inode;
1471 int ret = 0;
1472
1473 spin_lock(&block_group->lock);
1474 if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1475 spin_unlock(&block_group->lock);
1476 return 0;
1477 }
1478 spin_unlock(&block_group->lock);
1479
1480 inode = lookup_free_space_inode(block_group, path);
1481 if (IS_ERR(inode))
1482 return 0;
1483
1484 ret = __btrfs_write_out_cache(fs_info->tree_root, inode, ctl,
1485 block_group, &block_group->io_ctl, trans);
1486 if (ret) {
1487 btrfs_debug(fs_info,
1488 "failed to write free space cache for block group %llu error %d",
1489 block_group->start, ret);
1490 spin_lock(&block_group->lock);
1491 block_group->disk_cache_state = BTRFS_DC_ERROR;
1492 spin_unlock(&block_group->lock);
1493
1494 block_group->io_ctl.inode = NULL;
1495 iput(inode);
1496 }
1497
1498 /*
1499 * if ret == 0 the caller is expected to call btrfs_wait_cache_io
1500 * to wait for IO and put the inode
1501 */
1502
1503 return ret;
1504}
1505
1506static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1507 u64 offset)
1508{
1509 ASSERT(offset >= bitmap_start);
1510 offset -= bitmap_start;
1511 return (unsigned long)(div_u64(offset, unit));
1512}
1513
1514static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1515{
1516 return (unsigned long)(div_u64(bytes, unit));
1517}
1518
1519static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1520 u64 offset)
1521{
1522 u64 bitmap_start;
1523 u64 bytes_per_bitmap;
1524
1525 bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1526 bitmap_start = offset - ctl->start;
1527 bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1528 bitmap_start *= bytes_per_bitmap;
1529 bitmap_start += ctl->start;
1530
1531 return bitmap_start;
1532}
1533
1534static int tree_insert_offset(struct rb_root *root, u64 offset,
1535 struct rb_node *node, int bitmap)
1536{
1537 struct rb_node **p = &root->rb_node;
1538 struct rb_node *parent = NULL;
1539 struct btrfs_free_space *info;
1540
1541 while (*p) {
1542 parent = *p;
1543 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1544
1545 if (offset < info->offset) {
1546 p = &(*p)->rb_left;
1547 } else if (offset > info->offset) {
1548 p = &(*p)->rb_right;
1549 } else {
1550 /*
1551 * we could have a bitmap entry and an extent entry
1552 * share the same offset. If this is the case, we want
1553 * the extent entry to always be found first if we do a
1554 * linear search through the tree, since we want to have
1555 * the quickest allocation time, and allocating from an
1556 * extent is faster than allocating from a bitmap. So
1557 * if we're inserting a bitmap and we find an entry at
1558 * this offset, we want to go right, or after this entry
1559 * logically. If we are inserting an extent and we've
1560 * found a bitmap, we want to go left, or before
1561 * logically.
1562 */
1563 if (bitmap) {
1564 if (info->bitmap) {
1565 WARN_ON_ONCE(1);
1566 return -EEXIST;
1567 }
1568 p = &(*p)->rb_right;
1569 } else {
1570 if (!info->bitmap) {
1571 WARN_ON_ONCE(1);
1572 return -EEXIST;
1573 }
1574 p = &(*p)->rb_left;
1575 }
1576 }
1577 }
1578
1579 rb_link_node(node, parent, p);
1580 rb_insert_color(node, root);
1581
1582 return 0;
1583}
1584
1585/*
1586 * searches the tree for the given offset.
1587 *
1588 * fuzzy - If this is set, then we are trying to make an allocation, and we just
1589 * want a section that has at least bytes size and comes at or after the given
1590 * offset.
1591 */
1592static struct btrfs_free_space *
1593tree_search_offset(struct btrfs_free_space_ctl *ctl,
1594 u64 offset, int bitmap_only, int fuzzy)
1595{
1596 struct rb_node *n = ctl->free_space_offset.rb_node;
1597 struct btrfs_free_space *entry, *prev = NULL;
1598
1599 /* find entry that is closest to the 'offset' */
1600 while (1) {
1601 if (!n) {
1602 entry = NULL;
1603 break;
1604 }
1605
1606 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1607 prev = entry;
1608
1609 if (offset < entry->offset)
1610 n = n->rb_left;
1611 else if (offset > entry->offset)
1612 n = n->rb_right;
1613 else
1614 break;
1615 }
1616
1617 if (bitmap_only) {
1618 if (!entry)
1619 return NULL;
1620 if (entry->bitmap)
1621 return entry;
1622
1623 /*
1624 * bitmap entry and extent entry may share same offset,
1625 * in that case, bitmap entry comes after extent entry.
1626 */
1627 n = rb_next(n);
1628 if (!n)
1629 return NULL;
1630 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1631 if (entry->offset != offset)
1632 return NULL;
1633
1634 WARN_ON(!entry->bitmap);
1635 return entry;
1636 } else if (entry) {
1637 if (entry->bitmap) {
1638 /*
1639 * if previous extent entry covers the offset,
1640 * we should return it instead of the bitmap entry
1641 */
1642 n = rb_prev(&entry->offset_index);
1643 if (n) {
1644 prev = rb_entry(n, struct btrfs_free_space,
1645 offset_index);
1646 if (!prev->bitmap &&
1647 prev->offset + prev->bytes > offset)
1648 entry = prev;
1649 }
1650 }
1651 return entry;
1652 }
1653
1654 if (!prev)
1655 return NULL;
1656
1657 /* find last entry before the 'offset' */
1658 entry = prev;
1659 if (entry->offset > offset) {
1660 n = rb_prev(&entry->offset_index);
1661 if (n) {
1662 entry = rb_entry(n, struct btrfs_free_space,
1663 offset_index);
1664 ASSERT(entry->offset <= offset);
1665 } else {
1666 if (fuzzy)
1667 return entry;
1668 else
1669 return NULL;
1670 }
1671 }
1672
1673 if (entry->bitmap) {
1674 n = rb_prev(&entry->offset_index);
1675 if (n) {
1676 prev = rb_entry(n, struct btrfs_free_space,
1677 offset_index);
1678 if (!prev->bitmap &&
1679 prev->offset + prev->bytes > offset)
1680 return prev;
1681 }
1682 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1683 return entry;
1684 } else if (entry->offset + entry->bytes > offset)
1685 return entry;
1686
1687 if (!fuzzy)
1688 return NULL;
1689
1690 while (1) {
1691 if (entry->bitmap) {
1692 if (entry->offset + BITS_PER_BITMAP *
1693 ctl->unit > offset)
1694 break;
1695 } else {
1696 if (entry->offset + entry->bytes > offset)
1697 break;
1698 }
1699
1700 n = rb_next(&entry->offset_index);
1701 if (!n)
1702 return NULL;
1703 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1704 }
1705 return entry;
1706}
1707
1708static inline void
1709__unlink_free_space(struct btrfs_free_space_ctl *ctl,
1710 struct btrfs_free_space *info)
1711{
1712 rb_erase(&info->offset_index, &ctl->free_space_offset);
1713 ctl->free_extents--;
1714
1715 if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
1716 ctl->discardable_extents[BTRFS_STAT_CURR]--;
1717 ctl->discardable_bytes[BTRFS_STAT_CURR] -= info->bytes;
1718 }
1719}
1720
1721static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1722 struct btrfs_free_space *info)
1723{
1724 __unlink_free_space(ctl, info);
1725 ctl->free_space -= info->bytes;
1726}
1727
1728static int link_free_space(struct btrfs_free_space_ctl *ctl,
1729 struct btrfs_free_space *info)
1730{
1731 int ret = 0;
1732
1733 ASSERT(info->bytes || info->bitmap);
1734 ret = tree_insert_offset(&ctl->free_space_offset, info->offset,
1735 &info->offset_index, (info->bitmap != NULL));
1736 if (ret)
1737 return ret;
1738
1739 if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
1740 ctl->discardable_extents[BTRFS_STAT_CURR]++;
1741 ctl->discardable_bytes[BTRFS_STAT_CURR] += info->bytes;
1742 }
1743
1744 ctl->free_space += info->bytes;
1745 ctl->free_extents++;
1746 return ret;
1747}
1748
1749static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1750 struct btrfs_free_space *info,
1751 u64 offset, u64 bytes)
1752{
1753 unsigned long start, count, end;
1754 int extent_delta = -1;
1755
1756 start = offset_to_bit(info->offset, ctl->unit, offset);
1757 count = bytes_to_bits(bytes, ctl->unit);
1758 end = start + count;
1759 ASSERT(end <= BITS_PER_BITMAP);
1760
1761 bitmap_clear(info->bitmap, start, count);
1762
1763 info->bytes -= bytes;
1764 if (info->max_extent_size > ctl->unit)
1765 info->max_extent_size = 0;
1766
1767 if (start && test_bit(start - 1, info->bitmap))
1768 extent_delta++;
1769
1770 if (end < BITS_PER_BITMAP && test_bit(end, info->bitmap))
1771 extent_delta++;
1772
1773 info->bitmap_extents += extent_delta;
1774 if (!btrfs_free_space_trimmed(info)) {
1775 ctl->discardable_extents[BTRFS_STAT_CURR] += extent_delta;
1776 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
1777 }
1778}
1779
1780static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1781 struct btrfs_free_space *info, u64 offset,
1782 u64 bytes)
1783{
1784 __bitmap_clear_bits(ctl, info, offset, bytes);
1785 ctl->free_space -= bytes;
1786}
1787
1788static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1789 struct btrfs_free_space *info, u64 offset,
1790 u64 bytes)
1791{
1792 unsigned long start, count, end;
1793 int extent_delta = 1;
1794
1795 start = offset_to_bit(info->offset, ctl->unit, offset);
1796 count = bytes_to_bits(bytes, ctl->unit);
1797 end = start + count;
1798 ASSERT(end <= BITS_PER_BITMAP);
1799
1800 bitmap_set(info->bitmap, start, count);
1801
1802 info->bytes += bytes;
1803 ctl->free_space += bytes;
1804
1805 if (start && test_bit(start - 1, info->bitmap))
1806 extent_delta--;
1807
1808 if (end < BITS_PER_BITMAP && test_bit(end, info->bitmap))
1809 extent_delta--;
1810
1811 info->bitmap_extents += extent_delta;
1812 if (!btrfs_free_space_trimmed(info)) {
1813 ctl->discardable_extents[BTRFS_STAT_CURR] += extent_delta;
1814 ctl->discardable_bytes[BTRFS_STAT_CURR] += bytes;
1815 }
1816}
1817
1818/*
1819 * If we can not find suitable extent, we will use bytes to record
1820 * the size of the max extent.
1821 */
1822static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1823 struct btrfs_free_space *bitmap_info, u64 *offset,
1824 u64 *bytes, bool for_alloc)
1825{
1826 unsigned long found_bits = 0;
1827 unsigned long max_bits = 0;
1828 unsigned long bits, i;
1829 unsigned long next_zero;
1830 unsigned long extent_bits;
1831
1832 /*
1833 * Skip searching the bitmap if we don't have a contiguous section that
1834 * is large enough for this allocation.
1835 */
1836 if (for_alloc &&
1837 bitmap_info->max_extent_size &&
1838 bitmap_info->max_extent_size < *bytes) {
1839 *bytes = bitmap_info->max_extent_size;
1840 return -1;
1841 }
1842
1843 i = offset_to_bit(bitmap_info->offset, ctl->unit,
1844 max_t(u64, *offset, bitmap_info->offset));
1845 bits = bytes_to_bits(*bytes, ctl->unit);
1846
1847 for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
1848 if (for_alloc && bits == 1) {
1849 found_bits = 1;
1850 break;
1851 }
1852 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1853 BITS_PER_BITMAP, i);
1854 extent_bits = next_zero - i;
1855 if (extent_bits >= bits) {
1856 found_bits = extent_bits;
1857 break;
1858 } else if (extent_bits > max_bits) {
1859 max_bits = extent_bits;
1860 }
1861 i = next_zero;
1862 }
1863
1864 if (found_bits) {
1865 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1866 *bytes = (u64)(found_bits) * ctl->unit;
1867 return 0;
1868 }
1869
1870 *bytes = (u64)(max_bits) * ctl->unit;
1871 bitmap_info->max_extent_size = *bytes;
1872 return -1;
1873}
1874
1875static inline u64 get_max_extent_size(struct btrfs_free_space *entry)
1876{
1877 if (entry->bitmap)
1878 return entry->max_extent_size;
1879 return entry->bytes;
1880}
1881
1882/* Cache the size of the max extent in bytes */
1883static struct btrfs_free_space *
1884find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
1885 unsigned long align, u64 *max_extent_size)
1886{
1887 struct btrfs_free_space *entry;
1888 struct rb_node *node;
1889 u64 tmp;
1890 u64 align_off;
1891 int ret;
1892
1893 if (!ctl->free_space_offset.rb_node)
1894 goto out;
1895
1896 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
1897 if (!entry)
1898 goto out;
1899
1900 for (node = &entry->offset_index; node; node = rb_next(node)) {
1901 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1902 if (entry->bytes < *bytes) {
1903 *max_extent_size = max(get_max_extent_size(entry),
1904 *max_extent_size);
1905 continue;
1906 }
1907
1908 /* make sure the space returned is big enough
1909 * to match our requested alignment
1910 */
1911 if (*bytes >= align) {
1912 tmp = entry->offset - ctl->start + align - 1;
1913 tmp = div64_u64(tmp, align);
1914 tmp = tmp * align + ctl->start;
1915 align_off = tmp - entry->offset;
1916 } else {
1917 align_off = 0;
1918 tmp = entry->offset;
1919 }
1920
1921 if (entry->bytes < *bytes + align_off) {
1922 *max_extent_size = max(get_max_extent_size(entry),
1923 *max_extent_size);
1924 continue;
1925 }
1926
1927 if (entry->bitmap) {
1928 u64 size = *bytes;
1929
1930 ret = search_bitmap(ctl, entry, &tmp, &size, true);
1931 if (!ret) {
1932 *offset = tmp;
1933 *bytes = size;
1934 return entry;
1935 } else {
1936 *max_extent_size =
1937 max(get_max_extent_size(entry),
1938 *max_extent_size);
1939 }
1940 continue;
1941 }
1942
1943 *offset = tmp;
1944 *bytes = entry->bytes - align_off;
1945 return entry;
1946 }
1947out:
1948 return NULL;
1949}
1950
1951static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
1952 struct btrfs_free_space *info, u64 offset)
1953{
1954 info->offset = offset_to_bitmap(ctl, offset);
1955 info->bytes = 0;
1956 info->bitmap_extents = 0;
1957 INIT_LIST_HEAD(&info->list);
1958 link_free_space(ctl, info);
1959 ctl->total_bitmaps++;
1960 recalculate_thresholds(ctl);
1961}
1962
1963static void free_bitmap(struct btrfs_free_space_ctl *ctl,
1964 struct btrfs_free_space *bitmap_info)
1965{
1966 /*
1967 * Normally when this is called, the bitmap is completely empty. However,
1968 * if we are blowing up the free space cache for one reason or another
1969 * via __btrfs_remove_free_space_cache(), then it may not be freed and
1970 * we may leave stats on the table.
1971 */
1972 if (bitmap_info->bytes && !btrfs_free_space_trimmed(bitmap_info)) {
1973 ctl->discardable_extents[BTRFS_STAT_CURR] -=
1974 bitmap_info->bitmap_extents;
1975 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bitmap_info->bytes;
1976
1977 }
1978 unlink_free_space(ctl, bitmap_info);
1979 kmem_cache_free(btrfs_free_space_bitmap_cachep, bitmap_info->bitmap);
1980 kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1981 ctl->total_bitmaps--;
1982 recalculate_thresholds(ctl);
1983}
1984
1985static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
1986 struct btrfs_free_space *bitmap_info,
1987 u64 *offset, u64 *bytes)
1988{
1989 u64 end;
1990 u64 search_start, search_bytes;
1991 int ret;
1992
1993again:
1994 end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
1995
1996 /*
1997 * We need to search for bits in this bitmap. We could only cover some
1998 * of the extent in this bitmap thanks to how we add space, so we need
1999 * to search for as much as it as we can and clear that amount, and then
2000 * go searching for the next bit.
2001 */
2002 search_start = *offset;
2003 search_bytes = ctl->unit;
2004 search_bytes = min(search_bytes, end - search_start + 1);
2005 ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
2006 false);
2007 if (ret < 0 || search_start != *offset)
2008 return -EINVAL;
2009
2010 /* We may have found more bits than what we need */
2011 search_bytes = min(search_bytes, *bytes);
2012
2013 /* Cannot clear past the end of the bitmap */
2014 search_bytes = min(search_bytes, end - search_start + 1);
2015
2016 bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes);
2017 *offset += search_bytes;
2018 *bytes -= search_bytes;
2019
2020 if (*bytes) {
2021 struct rb_node *next = rb_next(&bitmap_info->offset_index);
2022 if (!bitmap_info->bytes)
2023 free_bitmap(ctl, bitmap_info);
2024
2025 /*
2026 * no entry after this bitmap, but we still have bytes to
2027 * remove, so something has gone wrong.
2028 */
2029 if (!next)
2030 return -EINVAL;
2031
2032 bitmap_info = rb_entry(next, struct btrfs_free_space,
2033 offset_index);
2034
2035 /*
2036 * if the next entry isn't a bitmap we need to return to let the
2037 * extent stuff do its work.
2038 */
2039 if (!bitmap_info->bitmap)
2040 return -EAGAIN;
2041
2042 /*
2043 * Ok the next item is a bitmap, but it may not actually hold
2044 * the information for the rest of this free space stuff, so
2045 * look for it, and if we don't find it return so we can try
2046 * everything over again.
2047 */
2048 search_start = *offset;
2049 search_bytes = ctl->unit;
2050 ret = search_bitmap(ctl, bitmap_info, &search_start,
2051 &search_bytes, false);
2052 if (ret < 0 || search_start != *offset)
2053 return -EAGAIN;
2054
2055 goto again;
2056 } else if (!bitmap_info->bytes)
2057 free_bitmap(ctl, bitmap_info);
2058
2059 return 0;
2060}
2061
2062static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
2063 struct btrfs_free_space *info, u64 offset,
2064 u64 bytes, enum btrfs_trim_state trim_state)
2065{
2066 u64 bytes_to_set = 0;
2067 u64 end;
2068
2069 /*
2070 * This is a tradeoff to make bitmap trim state minimal. We mark the
2071 * whole bitmap untrimmed if at any point we add untrimmed regions.
2072 */
2073 if (trim_state == BTRFS_TRIM_STATE_UNTRIMMED) {
2074 if (btrfs_free_space_trimmed(info)) {
2075 ctl->discardable_extents[BTRFS_STAT_CURR] +=
2076 info->bitmap_extents;
2077 ctl->discardable_bytes[BTRFS_STAT_CURR] += info->bytes;
2078 }
2079 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2080 }
2081
2082 end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
2083
2084 bytes_to_set = min(end - offset, bytes);
2085
2086 bitmap_set_bits(ctl, info, offset, bytes_to_set);
2087
2088 /*
2089 * We set some bytes, we have no idea what the max extent size is
2090 * anymore.
2091 */
2092 info->max_extent_size = 0;
2093
2094 return bytes_to_set;
2095
2096}
2097
2098static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
2099 struct btrfs_free_space *info)
2100{
2101 struct btrfs_block_group *block_group = ctl->private;
2102 struct btrfs_fs_info *fs_info = block_group->fs_info;
2103 bool forced = false;
2104
2105#ifdef CONFIG_BTRFS_DEBUG
2106 if (btrfs_should_fragment_free_space(block_group))
2107 forced = true;
2108#endif
2109
2110 /* This is a way to reclaim large regions from the bitmaps. */
2111 if (!forced && info->bytes >= FORCE_EXTENT_THRESHOLD)
2112 return false;
2113
2114 /*
2115 * If we are below the extents threshold then we can add this as an
2116 * extent, and don't have to deal with the bitmap
2117 */
2118 if (!forced && ctl->free_extents < ctl->extents_thresh) {
2119 /*
2120 * If this block group has some small extents we don't want to
2121 * use up all of our free slots in the cache with them, we want
2122 * to reserve them to larger extents, however if we have plenty
2123 * of cache left then go ahead an dadd them, no sense in adding
2124 * the overhead of a bitmap if we don't have to.
2125 */
2126 if (info->bytes <= fs_info->sectorsize * 8) {
2127 if (ctl->free_extents * 3 <= ctl->extents_thresh)
2128 return false;
2129 } else {
2130 return false;
2131 }
2132 }
2133
2134 /*
2135 * The original block groups from mkfs can be really small, like 8
2136 * megabytes, so don't bother with a bitmap for those entries. However
2137 * some block groups can be smaller than what a bitmap would cover but
2138 * are still large enough that they could overflow the 32k memory limit,
2139 * so allow those block groups to still be allowed to have a bitmap
2140 * entry.
2141 */
2142 if (((BITS_PER_BITMAP * ctl->unit) >> 1) > block_group->length)
2143 return false;
2144
2145 return true;
2146}
2147
2148static const struct btrfs_free_space_op free_space_op = {
2149 .use_bitmap = use_bitmap,
2150};
2151
2152static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
2153 struct btrfs_free_space *info)
2154{
2155 struct btrfs_free_space *bitmap_info;
2156 struct btrfs_block_group *block_group = NULL;
2157 int added = 0;
2158 u64 bytes, offset, bytes_added;
2159 enum btrfs_trim_state trim_state;
2160 int ret;
2161
2162 bytes = info->bytes;
2163 offset = info->offset;
2164 trim_state = info->trim_state;
2165
2166 if (!ctl->op->use_bitmap(ctl, info))
2167 return 0;
2168
2169 if (ctl->op == &free_space_op)
2170 block_group = ctl->private;
2171again:
2172 /*
2173 * Since we link bitmaps right into the cluster we need to see if we
2174 * have a cluster here, and if so and it has our bitmap we need to add
2175 * the free space to that bitmap.
2176 */
2177 if (block_group && !list_empty(&block_group->cluster_list)) {
2178 struct btrfs_free_cluster *cluster;
2179 struct rb_node *node;
2180 struct btrfs_free_space *entry;
2181
2182 cluster = list_entry(block_group->cluster_list.next,
2183 struct btrfs_free_cluster,
2184 block_group_list);
2185 spin_lock(&cluster->lock);
2186 node = rb_first(&cluster->root);
2187 if (!node) {
2188 spin_unlock(&cluster->lock);
2189 goto no_cluster_bitmap;
2190 }
2191
2192 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2193 if (!entry->bitmap) {
2194 spin_unlock(&cluster->lock);
2195 goto no_cluster_bitmap;
2196 }
2197
2198 if (entry->offset == offset_to_bitmap(ctl, offset)) {
2199 bytes_added = add_bytes_to_bitmap(ctl, entry, offset,
2200 bytes, trim_state);
2201 bytes -= bytes_added;
2202 offset += bytes_added;
2203 }
2204 spin_unlock(&cluster->lock);
2205 if (!bytes) {
2206 ret = 1;
2207 goto out;
2208 }
2209 }
2210
2211no_cluster_bitmap:
2212 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2213 1, 0);
2214 if (!bitmap_info) {
2215 ASSERT(added == 0);
2216 goto new_bitmap;
2217 }
2218
2219 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes,
2220 trim_state);
2221 bytes -= bytes_added;
2222 offset += bytes_added;
2223 added = 0;
2224
2225 if (!bytes) {
2226 ret = 1;
2227 goto out;
2228 } else
2229 goto again;
2230
2231new_bitmap:
2232 if (info && info->bitmap) {
2233 add_new_bitmap(ctl, info, offset);
2234 added = 1;
2235 info = NULL;
2236 goto again;
2237 } else {
2238 spin_unlock(&ctl->tree_lock);
2239
2240 /* no pre-allocated info, allocate a new one */
2241 if (!info) {
2242 info = kmem_cache_zalloc(btrfs_free_space_cachep,
2243 GFP_NOFS);
2244 if (!info) {
2245 spin_lock(&ctl->tree_lock);
2246 ret = -ENOMEM;
2247 goto out;
2248 }
2249 }
2250
2251 /* allocate the bitmap */
2252 info->bitmap = kmem_cache_zalloc(btrfs_free_space_bitmap_cachep,
2253 GFP_NOFS);
2254 info->trim_state = BTRFS_TRIM_STATE_TRIMMED;
2255 spin_lock(&ctl->tree_lock);
2256 if (!info->bitmap) {
2257 ret = -ENOMEM;
2258 goto out;
2259 }
2260 goto again;
2261 }
2262
2263out:
2264 if (info) {
2265 if (info->bitmap)
2266 kmem_cache_free(btrfs_free_space_bitmap_cachep,
2267 info->bitmap);
2268 kmem_cache_free(btrfs_free_space_cachep, info);
2269 }
2270
2271 return ret;
2272}
2273
2274/*
2275 * Free space merging rules:
2276 * 1) Merge trimmed areas together
2277 * 2) Let untrimmed areas coalesce with trimmed areas
2278 * 3) Always pull neighboring regions from bitmaps
2279 *
2280 * The above rules are for when we merge free space based on btrfs_trim_state.
2281 * Rules 2 and 3 are subtle because they are suboptimal, but are done for the
2282 * same reason: to promote larger extent regions which makes life easier for
2283 * find_free_extent(). Rule 2 enables coalescing based on the common path
2284 * being returning free space from btrfs_finish_extent_commit(). So when free
2285 * space is trimmed, it will prevent aggregating trimmed new region and
2286 * untrimmed regions in the rb_tree. Rule 3 is purely to obtain larger extents
2287 * and provide find_free_extent() with the largest extents possible hoping for
2288 * the reuse path.
2289 */
2290static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
2291 struct btrfs_free_space *info, bool update_stat)
2292{
2293 struct btrfs_free_space *left_info = NULL;
2294 struct btrfs_free_space *right_info;
2295 bool merged = false;
2296 u64 offset = info->offset;
2297 u64 bytes = info->bytes;
2298 const bool is_trimmed = btrfs_free_space_trimmed(info);
2299
2300 /*
2301 * first we want to see if there is free space adjacent to the range we
2302 * are adding, if there is remove that struct and add a new one to
2303 * cover the entire range
2304 */
2305 right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
2306 if (right_info && rb_prev(&right_info->offset_index))
2307 left_info = rb_entry(rb_prev(&right_info->offset_index),
2308 struct btrfs_free_space, offset_index);
2309 else if (!right_info)
2310 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
2311
2312 /* See try_merge_free_space() comment. */
2313 if (right_info && !right_info->bitmap &&
2314 (!is_trimmed || btrfs_free_space_trimmed(right_info))) {
2315 if (update_stat)
2316 unlink_free_space(ctl, right_info);
2317 else
2318 __unlink_free_space(ctl, right_info);
2319 info->bytes += right_info->bytes;
2320 kmem_cache_free(btrfs_free_space_cachep, right_info);
2321 merged = true;
2322 }
2323
2324 /* See try_merge_free_space() comment. */
2325 if (left_info && !left_info->bitmap &&
2326 left_info->offset + left_info->bytes == offset &&
2327 (!is_trimmed || btrfs_free_space_trimmed(left_info))) {
2328 if (update_stat)
2329 unlink_free_space(ctl, left_info);
2330 else
2331 __unlink_free_space(ctl, left_info);
2332 info->offset = left_info->offset;
2333 info->bytes += left_info->bytes;
2334 kmem_cache_free(btrfs_free_space_cachep, left_info);
2335 merged = true;
2336 }
2337
2338 return merged;
2339}
2340
2341static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
2342 struct btrfs_free_space *info,
2343 bool update_stat)
2344{
2345 struct btrfs_free_space *bitmap;
2346 unsigned long i;
2347 unsigned long j;
2348 const u64 end = info->offset + info->bytes;
2349 const u64 bitmap_offset = offset_to_bitmap(ctl, end);
2350 u64 bytes;
2351
2352 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2353 if (!bitmap)
2354 return false;
2355
2356 i = offset_to_bit(bitmap->offset, ctl->unit, end);
2357 j = find_next_zero_bit(bitmap->bitmap, BITS_PER_BITMAP, i);
2358 if (j == i)
2359 return false;
2360 bytes = (j - i) * ctl->unit;
2361 info->bytes += bytes;
2362
2363 /* See try_merge_free_space() comment. */
2364 if (!btrfs_free_space_trimmed(bitmap))
2365 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2366
2367 if (update_stat)
2368 bitmap_clear_bits(ctl, bitmap, end, bytes);
2369 else
2370 __bitmap_clear_bits(ctl, bitmap, end, bytes);
2371
2372 if (!bitmap->bytes)
2373 free_bitmap(ctl, bitmap);
2374
2375 return true;
2376}
2377
2378static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
2379 struct btrfs_free_space *info,
2380 bool update_stat)
2381{
2382 struct btrfs_free_space *bitmap;
2383 u64 bitmap_offset;
2384 unsigned long i;
2385 unsigned long j;
2386 unsigned long prev_j;
2387 u64 bytes;
2388
2389 bitmap_offset = offset_to_bitmap(ctl, info->offset);
2390 /* If we're on a boundary, try the previous logical bitmap. */
2391 if (bitmap_offset == info->offset) {
2392 if (info->offset == 0)
2393 return false;
2394 bitmap_offset = offset_to_bitmap(ctl, info->offset - 1);
2395 }
2396
2397 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2398 if (!bitmap)
2399 return false;
2400
2401 i = offset_to_bit(bitmap->offset, ctl->unit, info->offset) - 1;
2402 j = 0;
2403 prev_j = (unsigned long)-1;
2404 for_each_clear_bit_from(j, bitmap->bitmap, BITS_PER_BITMAP) {
2405 if (j > i)
2406 break;
2407 prev_j = j;
2408 }
2409 if (prev_j == i)
2410 return false;
2411
2412 if (prev_j == (unsigned long)-1)
2413 bytes = (i + 1) * ctl->unit;
2414 else
2415 bytes = (i - prev_j) * ctl->unit;
2416
2417 info->offset -= bytes;
2418 info->bytes += bytes;
2419
2420 /* See try_merge_free_space() comment. */
2421 if (!btrfs_free_space_trimmed(bitmap))
2422 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2423
2424 if (update_stat)
2425 bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2426 else
2427 __bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2428
2429 if (!bitmap->bytes)
2430 free_bitmap(ctl, bitmap);
2431
2432 return true;
2433}
2434
2435/*
2436 * We prefer always to allocate from extent entries, both for clustered and
2437 * non-clustered allocation requests. So when attempting to add a new extent
2438 * entry, try to see if there's adjacent free space in bitmap entries, and if
2439 * there is, migrate that space from the bitmaps to the extent.
2440 * Like this we get better chances of satisfying space allocation requests
2441 * because we attempt to satisfy them based on a single cache entry, and never
2442 * on 2 or more entries - even if the entries represent a contiguous free space
2443 * region (e.g. 1 extent entry + 1 bitmap entry starting where the extent entry
2444 * ends).
2445 */
2446static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
2447 struct btrfs_free_space *info,
2448 bool update_stat)
2449{
2450 /*
2451 * Only work with disconnected entries, as we can change their offset,
2452 * and must be extent entries.
2453 */
2454 ASSERT(!info->bitmap);
2455 ASSERT(RB_EMPTY_NODE(&info->offset_index));
2456
2457 if (ctl->total_bitmaps > 0) {
2458 bool stole_end;
2459 bool stole_front = false;
2460
2461 stole_end = steal_from_bitmap_to_end(ctl, info, update_stat);
2462 if (ctl->total_bitmaps > 0)
2463 stole_front = steal_from_bitmap_to_front(ctl, info,
2464 update_stat);
2465
2466 if (stole_end || stole_front)
2467 try_merge_free_space(ctl, info, update_stat);
2468 }
2469}
2470
2471int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
2472 struct btrfs_free_space_ctl *ctl,
2473 u64 offset, u64 bytes,
2474 enum btrfs_trim_state trim_state)
2475{
2476 struct btrfs_block_group *block_group = ctl->private;
2477 struct btrfs_free_space *info;
2478 int ret = 0;
2479 u64 filter_bytes = bytes;
2480
2481 ASSERT(!btrfs_is_zoned(fs_info));
2482
2483 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
2484 if (!info)
2485 return -ENOMEM;
2486
2487 info->offset = offset;
2488 info->bytes = bytes;
2489 info->trim_state = trim_state;
2490 RB_CLEAR_NODE(&info->offset_index);
2491
2492 spin_lock(&ctl->tree_lock);
2493
2494 if (try_merge_free_space(ctl, info, true))
2495 goto link;
2496
2497 /*
2498 * There was no extent directly to the left or right of this new
2499 * extent then we know we're going to have to allocate a new extent, so
2500 * before we do that see if we need to drop this into a bitmap
2501 */
2502 ret = insert_into_bitmap(ctl, info);
2503 if (ret < 0) {
2504 goto out;
2505 } else if (ret) {
2506 ret = 0;
2507 goto out;
2508 }
2509link:
2510 /*
2511 * Only steal free space from adjacent bitmaps if we're sure we're not
2512 * going to add the new free space to existing bitmap entries - because
2513 * that would mean unnecessary work that would be reverted. Therefore
2514 * attempt to steal space from bitmaps if we're adding an extent entry.
2515 */
2516 steal_from_bitmap(ctl, info, true);
2517
2518 filter_bytes = max(filter_bytes, info->bytes);
2519
2520 ret = link_free_space(ctl, info);
2521 if (ret)
2522 kmem_cache_free(btrfs_free_space_cachep, info);
2523out:
2524 btrfs_discard_update_discardable(block_group);
2525 spin_unlock(&ctl->tree_lock);
2526
2527 if (ret) {
2528 btrfs_crit(fs_info, "unable to add free space :%d", ret);
2529 ASSERT(ret != -EEXIST);
2530 }
2531
2532 if (trim_state != BTRFS_TRIM_STATE_TRIMMED) {
2533 btrfs_discard_check_filter(block_group, filter_bytes);
2534 btrfs_discard_queue_work(&fs_info->discard_ctl, block_group);
2535 }
2536
2537 return ret;
2538}
2539
2540static int __btrfs_add_free_space_zoned(struct btrfs_block_group *block_group,
2541 u64 bytenr, u64 size, bool used)
2542{
2543 struct btrfs_fs_info *fs_info = block_group->fs_info;
2544 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2545 u64 offset = bytenr - block_group->start;
2546 u64 to_free, to_unusable;
2547
2548 spin_lock(&ctl->tree_lock);
2549 if (!used)
2550 to_free = size;
2551 else if (offset >= block_group->alloc_offset)
2552 to_free = size;
2553 else if (offset + size <= block_group->alloc_offset)
2554 to_free = 0;
2555 else
2556 to_free = offset + size - block_group->alloc_offset;
2557 to_unusable = size - to_free;
2558
2559 ctl->free_space += to_free;
2560 /*
2561 * If the block group is read-only, we should account freed space into
2562 * bytes_readonly.
2563 */
2564 if (!block_group->ro)
2565 block_group->zone_unusable += to_unusable;
2566 spin_unlock(&ctl->tree_lock);
2567 if (!used) {
2568 spin_lock(&block_group->lock);
2569 block_group->alloc_offset -= size;
2570 spin_unlock(&block_group->lock);
2571 }
2572
2573 /* All the region is now unusable. Mark it as unused and reclaim */
2574 if (block_group->zone_unusable == block_group->length) {
2575 btrfs_mark_bg_unused(block_group);
2576 } else if (block_group->zone_unusable >=
2577 div_factor_fine(block_group->length,
2578 fs_info->bg_reclaim_threshold)) {
2579 btrfs_mark_bg_to_reclaim(block_group);
2580 }
2581
2582 return 0;
2583}
2584
2585int btrfs_add_free_space(struct btrfs_block_group *block_group,
2586 u64 bytenr, u64 size)
2587{
2588 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2589
2590 if (btrfs_is_zoned(block_group->fs_info))
2591 return __btrfs_add_free_space_zoned(block_group, bytenr, size,
2592 true);
2593
2594 if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC))
2595 trim_state = BTRFS_TRIM_STATE_TRIMMED;
2596
2597 return __btrfs_add_free_space(block_group->fs_info,
2598 block_group->free_space_ctl,
2599 bytenr, size, trim_state);
2600}
2601
2602int btrfs_add_free_space_unused(struct btrfs_block_group *block_group,
2603 u64 bytenr, u64 size)
2604{
2605 if (btrfs_is_zoned(block_group->fs_info))
2606 return __btrfs_add_free_space_zoned(block_group, bytenr, size,
2607 false);
2608
2609 return btrfs_add_free_space(block_group, bytenr, size);
2610}
2611
2612/*
2613 * This is a subtle distinction because when adding free space back in general,
2614 * we want it to be added as untrimmed for async. But in the case where we add
2615 * it on loading of a block group, we want to consider it trimmed.
2616 */
2617int btrfs_add_free_space_async_trimmed(struct btrfs_block_group *block_group,
2618 u64 bytenr, u64 size)
2619{
2620 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2621
2622 if (btrfs_is_zoned(block_group->fs_info))
2623 return __btrfs_add_free_space_zoned(block_group, bytenr, size,
2624 true);
2625
2626 if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC) ||
2627 btrfs_test_opt(block_group->fs_info, DISCARD_ASYNC))
2628 trim_state = BTRFS_TRIM_STATE_TRIMMED;
2629
2630 return __btrfs_add_free_space(block_group->fs_info,
2631 block_group->free_space_ctl,
2632 bytenr, size, trim_state);
2633}
2634
2635int btrfs_remove_free_space(struct btrfs_block_group *block_group,
2636 u64 offset, u64 bytes)
2637{
2638 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2639 struct btrfs_free_space *info;
2640 int ret;
2641 bool re_search = false;
2642
2643 if (btrfs_is_zoned(block_group->fs_info)) {
2644 /*
2645 * This can happen with conventional zones when replaying log.
2646 * Since the allocation info of tree-log nodes are not recorded
2647 * to the extent-tree, calculate_alloc_pointer() failed to
2648 * advance the allocation pointer after last allocated tree log
2649 * node blocks.
2650 *
2651 * This function is called from
2652 * btrfs_pin_extent_for_log_replay() when replaying the log.
2653 * Advance the pointer not to overwrite the tree-log nodes.
2654 */
2655 if (block_group->start + block_group->alloc_offset <
2656 offset + bytes) {
2657 block_group->alloc_offset =
2658 offset + bytes - block_group->start;
2659 }
2660 return 0;
2661 }
2662
2663 spin_lock(&ctl->tree_lock);
2664
2665again:
2666 ret = 0;
2667 if (!bytes)
2668 goto out_lock;
2669
2670 info = tree_search_offset(ctl, offset, 0, 0);
2671 if (!info) {
2672 /*
2673 * oops didn't find an extent that matched the space we wanted
2674 * to remove, look for a bitmap instead
2675 */
2676 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2677 1, 0);
2678 if (!info) {
2679 /*
2680 * If we found a partial bit of our free space in a
2681 * bitmap but then couldn't find the other part this may
2682 * be a problem, so WARN about it.
2683 */
2684 WARN_ON(re_search);
2685 goto out_lock;
2686 }
2687 }
2688
2689 re_search = false;
2690 if (!info->bitmap) {
2691 unlink_free_space(ctl, info);
2692 if (offset == info->offset) {
2693 u64 to_free = min(bytes, info->bytes);
2694
2695 info->bytes -= to_free;
2696 info->offset += to_free;
2697 if (info->bytes) {
2698 ret = link_free_space(ctl, info);
2699 WARN_ON(ret);
2700 } else {
2701 kmem_cache_free(btrfs_free_space_cachep, info);
2702 }
2703
2704 offset += to_free;
2705 bytes -= to_free;
2706 goto again;
2707 } else {
2708 u64 old_end = info->bytes + info->offset;
2709
2710 info->bytes = offset - info->offset;
2711 ret = link_free_space(ctl, info);
2712 WARN_ON(ret);
2713 if (ret)
2714 goto out_lock;
2715
2716 /* Not enough bytes in this entry to satisfy us */
2717 if (old_end < offset + bytes) {
2718 bytes -= old_end - offset;
2719 offset = old_end;
2720 goto again;
2721 } else if (old_end == offset + bytes) {
2722 /* all done */
2723 goto out_lock;
2724 }
2725 spin_unlock(&ctl->tree_lock);
2726
2727 ret = __btrfs_add_free_space(block_group->fs_info, ctl,
2728 offset + bytes,
2729 old_end - (offset + bytes),
2730 info->trim_state);
2731 WARN_ON(ret);
2732 goto out;
2733 }
2734 }
2735
2736 ret = remove_from_bitmap(ctl, info, &offset, &bytes);
2737 if (ret == -EAGAIN) {
2738 re_search = true;
2739 goto again;
2740 }
2741out_lock:
2742 btrfs_discard_update_discardable(block_group);
2743 spin_unlock(&ctl->tree_lock);
2744out:
2745 return ret;
2746}
2747
2748void btrfs_dump_free_space(struct btrfs_block_group *block_group,
2749 u64 bytes)
2750{
2751 struct btrfs_fs_info *fs_info = block_group->fs_info;
2752 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2753 struct btrfs_free_space *info;
2754 struct rb_node *n;
2755 int count = 0;
2756
2757 /*
2758 * Zoned btrfs does not use free space tree and cluster. Just print
2759 * out the free space after the allocation offset.
2760 */
2761 if (btrfs_is_zoned(fs_info)) {
2762 btrfs_info(fs_info, "free space %llu",
2763 block_group->length - block_group->alloc_offset);
2764 return;
2765 }
2766
2767 spin_lock(&ctl->tree_lock);
2768 for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
2769 info = rb_entry(n, struct btrfs_free_space, offset_index);
2770 if (info->bytes >= bytes && !block_group->ro)
2771 count++;
2772 btrfs_crit(fs_info, "entry offset %llu, bytes %llu, bitmap %s",
2773 info->offset, info->bytes,
2774 (info->bitmap) ? "yes" : "no");
2775 }
2776 spin_unlock(&ctl->tree_lock);
2777 btrfs_info(fs_info, "block group has cluster?: %s",
2778 list_empty(&block_group->cluster_list) ? "no" : "yes");
2779 btrfs_info(fs_info,
2780 "%d blocks of free space at or bigger than bytes is", count);
2781}
2782
2783void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group,
2784 struct btrfs_free_space_ctl *ctl)
2785{
2786 struct btrfs_fs_info *fs_info = block_group->fs_info;
2787
2788 spin_lock_init(&ctl->tree_lock);
2789 ctl->unit = fs_info->sectorsize;
2790 ctl->start = block_group->start;
2791 ctl->private = block_group;
2792 ctl->op = &free_space_op;
2793 INIT_LIST_HEAD(&ctl->trimming_ranges);
2794 mutex_init(&ctl->cache_writeout_mutex);
2795
2796 /*
2797 * we only want to have 32k of ram per block group for keeping
2798 * track of free space, and if we pass 1/2 of that we want to
2799 * start converting things over to using bitmaps
2800 */
2801 ctl->extents_thresh = (SZ_32K / 2) / sizeof(struct btrfs_free_space);
2802}
2803
2804/*
2805 * for a given cluster, put all of its extents back into the free
2806 * space cache. If the block group passed doesn't match the block group
2807 * pointed to by the cluster, someone else raced in and freed the
2808 * cluster already. In that case, we just return without changing anything
2809 */
2810static void __btrfs_return_cluster_to_free_space(
2811 struct btrfs_block_group *block_group,
2812 struct btrfs_free_cluster *cluster)
2813{
2814 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2815 struct btrfs_free_space *entry;
2816 struct rb_node *node;
2817
2818 spin_lock(&cluster->lock);
2819 if (cluster->block_group != block_group) {
2820 spin_unlock(&cluster->lock);
2821 return;
2822 }
2823
2824 cluster->block_group = NULL;
2825 cluster->window_start = 0;
2826 list_del_init(&cluster->block_group_list);
2827
2828 node = rb_first(&cluster->root);
2829 while (node) {
2830 bool bitmap;
2831
2832 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2833 node = rb_next(&entry->offset_index);
2834 rb_erase(&entry->offset_index, &cluster->root);
2835 RB_CLEAR_NODE(&entry->offset_index);
2836
2837 bitmap = (entry->bitmap != NULL);
2838 if (!bitmap) {
2839 /* Merging treats extents as if they were new */
2840 if (!btrfs_free_space_trimmed(entry)) {
2841 ctl->discardable_extents[BTRFS_STAT_CURR]--;
2842 ctl->discardable_bytes[BTRFS_STAT_CURR] -=
2843 entry->bytes;
2844 }
2845
2846 try_merge_free_space(ctl, entry, false);
2847 steal_from_bitmap(ctl, entry, false);
2848
2849 /* As we insert directly, update these statistics */
2850 if (!btrfs_free_space_trimmed(entry)) {
2851 ctl->discardable_extents[BTRFS_STAT_CURR]++;
2852 ctl->discardable_bytes[BTRFS_STAT_CURR] +=
2853 entry->bytes;
2854 }
2855 }
2856 tree_insert_offset(&ctl->free_space_offset,
2857 entry->offset, &entry->offset_index, bitmap);
2858 }
2859 cluster->root = RB_ROOT;
2860 spin_unlock(&cluster->lock);
2861 btrfs_put_block_group(block_group);
2862}
2863
2864static void __btrfs_remove_free_space_cache_locked(
2865 struct btrfs_free_space_ctl *ctl)
2866{
2867 struct btrfs_free_space *info;
2868 struct rb_node *node;
2869
2870 while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
2871 info = rb_entry(node, struct btrfs_free_space, offset_index);
2872 if (!info->bitmap) {
2873 unlink_free_space(ctl, info);
2874 kmem_cache_free(btrfs_free_space_cachep, info);
2875 } else {
2876 free_bitmap(ctl, info);
2877 }
2878
2879 cond_resched_lock(&ctl->tree_lock);
2880 }
2881}
2882
2883void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
2884{
2885 spin_lock(&ctl->tree_lock);
2886 __btrfs_remove_free_space_cache_locked(ctl);
2887 if (ctl->private)
2888 btrfs_discard_update_discardable(ctl->private);
2889 spin_unlock(&ctl->tree_lock);
2890}
2891
2892void btrfs_remove_free_space_cache(struct btrfs_block_group *block_group)
2893{
2894 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2895 struct btrfs_free_cluster *cluster;
2896 struct list_head *head;
2897
2898 spin_lock(&ctl->tree_lock);
2899 while ((head = block_group->cluster_list.next) !=
2900 &block_group->cluster_list) {
2901 cluster = list_entry(head, struct btrfs_free_cluster,
2902 block_group_list);
2903
2904 WARN_ON(cluster->block_group != block_group);
2905 __btrfs_return_cluster_to_free_space(block_group, cluster);
2906
2907 cond_resched_lock(&ctl->tree_lock);
2908 }
2909 __btrfs_remove_free_space_cache_locked(ctl);
2910 btrfs_discard_update_discardable(block_group);
2911 spin_unlock(&ctl->tree_lock);
2912
2913}
2914
2915/**
2916 * btrfs_is_free_space_trimmed - see if everything is trimmed
2917 * @block_group: block_group of interest
2918 *
2919 * Walk @block_group's free space rb_tree to determine if everything is trimmed.
2920 */
2921bool btrfs_is_free_space_trimmed(struct btrfs_block_group *block_group)
2922{
2923 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2924 struct btrfs_free_space *info;
2925 struct rb_node *node;
2926 bool ret = true;
2927
2928 spin_lock(&ctl->tree_lock);
2929 node = rb_first(&ctl->free_space_offset);
2930
2931 while (node) {
2932 info = rb_entry(node, struct btrfs_free_space, offset_index);
2933
2934 if (!btrfs_free_space_trimmed(info)) {
2935 ret = false;
2936 break;
2937 }
2938
2939 node = rb_next(node);
2940 }
2941
2942 spin_unlock(&ctl->tree_lock);
2943 return ret;
2944}
2945
2946u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
2947 u64 offset, u64 bytes, u64 empty_size,
2948 u64 *max_extent_size)
2949{
2950 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2951 struct btrfs_discard_ctl *discard_ctl =
2952 &block_group->fs_info->discard_ctl;
2953 struct btrfs_free_space *entry = NULL;
2954 u64 bytes_search = bytes + empty_size;
2955 u64 ret = 0;
2956 u64 align_gap = 0;
2957 u64 align_gap_len = 0;
2958 enum btrfs_trim_state align_gap_trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2959
2960 ASSERT(!btrfs_is_zoned(block_group->fs_info));
2961
2962 spin_lock(&ctl->tree_lock);
2963 entry = find_free_space(ctl, &offset, &bytes_search,
2964 block_group->full_stripe_len, max_extent_size);
2965 if (!entry)
2966 goto out;
2967
2968 ret = offset;
2969 if (entry->bitmap) {
2970 bitmap_clear_bits(ctl, entry, offset, bytes);
2971
2972 if (!btrfs_free_space_trimmed(entry))
2973 atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
2974
2975 if (!entry->bytes)
2976 free_bitmap(ctl, entry);
2977 } else {
2978 unlink_free_space(ctl, entry);
2979 align_gap_len = offset - entry->offset;
2980 align_gap = entry->offset;
2981 align_gap_trim_state = entry->trim_state;
2982
2983 if (!btrfs_free_space_trimmed(entry))
2984 atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
2985
2986 entry->offset = offset + bytes;
2987 WARN_ON(entry->bytes < bytes + align_gap_len);
2988
2989 entry->bytes -= bytes + align_gap_len;
2990 if (!entry->bytes)
2991 kmem_cache_free(btrfs_free_space_cachep, entry);
2992 else
2993 link_free_space(ctl, entry);
2994 }
2995out:
2996 btrfs_discard_update_discardable(block_group);
2997 spin_unlock(&ctl->tree_lock);
2998
2999 if (align_gap_len)
3000 __btrfs_add_free_space(block_group->fs_info, ctl,
3001 align_gap, align_gap_len,
3002 align_gap_trim_state);
3003 return ret;
3004}
3005
3006/*
3007 * given a cluster, put all of its extents back into the free space
3008 * cache. If a block group is passed, this function will only free
3009 * a cluster that belongs to the passed block group.
3010 *
3011 * Otherwise, it'll get a reference on the block group pointed to by the
3012 * cluster and remove the cluster from it.
3013 */
3014void btrfs_return_cluster_to_free_space(
3015 struct btrfs_block_group *block_group,
3016 struct btrfs_free_cluster *cluster)
3017{
3018 struct btrfs_free_space_ctl *ctl;
3019
3020 /* first, get a safe pointer to the block group */
3021 spin_lock(&cluster->lock);
3022 if (!block_group) {
3023 block_group = cluster->block_group;
3024 if (!block_group) {
3025 spin_unlock(&cluster->lock);
3026 return;
3027 }
3028 } else if (cluster->block_group != block_group) {
3029 /* someone else has already freed it don't redo their work */
3030 spin_unlock(&cluster->lock);
3031 return;
3032 }
3033 btrfs_get_block_group(block_group);
3034 spin_unlock(&cluster->lock);
3035
3036 ctl = block_group->free_space_ctl;
3037
3038 /* now return any extents the cluster had on it */
3039 spin_lock(&ctl->tree_lock);
3040 __btrfs_return_cluster_to_free_space(block_group, cluster);
3041 spin_unlock(&ctl->tree_lock);
3042
3043 btrfs_discard_queue_work(&block_group->fs_info->discard_ctl, block_group);
3044
3045 /* finally drop our ref */
3046 btrfs_put_block_group(block_group);
3047}
3048
3049static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group *block_group,
3050 struct btrfs_free_cluster *cluster,
3051 struct btrfs_free_space *entry,
3052 u64 bytes, u64 min_start,
3053 u64 *max_extent_size)
3054{
3055 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3056 int err;
3057 u64 search_start = cluster->window_start;
3058 u64 search_bytes = bytes;
3059 u64 ret = 0;
3060
3061 search_start = min_start;
3062 search_bytes = bytes;
3063
3064 err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
3065 if (err) {
3066 *max_extent_size = max(get_max_extent_size(entry),
3067 *max_extent_size);
3068 return 0;
3069 }
3070
3071 ret = search_start;
3072 __bitmap_clear_bits(ctl, entry, ret, bytes);
3073
3074 return ret;
3075}
3076
3077/*
3078 * given a cluster, try to allocate 'bytes' from it, returns 0
3079 * if it couldn't find anything suitably large, or a logical disk offset
3080 * if things worked out
3081 */
3082u64 btrfs_alloc_from_cluster(struct btrfs_block_group *block_group,
3083 struct btrfs_free_cluster *cluster, u64 bytes,
3084 u64 min_start, u64 *max_extent_size)
3085{
3086 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3087 struct btrfs_discard_ctl *discard_ctl =
3088 &block_group->fs_info->discard_ctl;
3089 struct btrfs_free_space *entry = NULL;
3090 struct rb_node *node;
3091 u64 ret = 0;
3092
3093 ASSERT(!btrfs_is_zoned(block_group->fs_info));
3094
3095 spin_lock(&cluster->lock);
3096 if (bytes > cluster->max_size)
3097 goto out;
3098
3099 if (cluster->block_group != block_group)
3100 goto out;
3101
3102 node = rb_first(&cluster->root);
3103 if (!node)
3104 goto out;
3105
3106 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3107 while (1) {
3108 if (entry->bytes < bytes)
3109 *max_extent_size = max(get_max_extent_size(entry),
3110 *max_extent_size);
3111
3112 if (entry->bytes < bytes ||
3113 (!entry->bitmap && entry->offset < min_start)) {
3114 node = rb_next(&entry->offset_index);
3115 if (!node)
3116 break;
3117 entry = rb_entry(node, struct btrfs_free_space,
3118 offset_index);
3119 continue;
3120 }
3121
3122 if (entry->bitmap) {
3123 ret = btrfs_alloc_from_bitmap(block_group,
3124 cluster, entry, bytes,
3125 cluster->window_start,
3126 max_extent_size);
3127 if (ret == 0) {
3128 node = rb_next(&entry->offset_index);
3129 if (!node)
3130 break;
3131 entry = rb_entry(node, struct btrfs_free_space,
3132 offset_index);
3133 continue;
3134 }
3135 cluster->window_start += bytes;
3136 } else {
3137 ret = entry->offset;
3138
3139 entry->offset += bytes;
3140 entry->bytes -= bytes;
3141 }
3142
3143 break;
3144 }
3145out:
3146 spin_unlock(&cluster->lock);
3147
3148 if (!ret)
3149 return 0;
3150
3151 spin_lock(&ctl->tree_lock);
3152
3153 if (!btrfs_free_space_trimmed(entry))
3154 atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
3155
3156 ctl->free_space -= bytes;
3157 if (!entry->bitmap && !btrfs_free_space_trimmed(entry))
3158 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
3159
3160 spin_lock(&cluster->lock);
3161 if (entry->bytes == 0) {
3162 rb_erase(&entry->offset_index, &cluster->root);
3163 ctl->free_extents--;
3164 if (entry->bitmap) {
3165 kmem_cache_free(btrfs_free_space_bitmap_cachep,
3166 entry->bitmap);
3167 ctl->total_bitmaps--;
3168 recalculate_thresholds(ctl);
3169 } else if (!btrfs_free_space_trimmed(entry)) {
3170 ctl->discardable_extents[BTRFS_STAT_CURR]--;
3171 }
3172 kmem_cache_free(btrfs_free_space_cachep, entry);
3173 }
3174
3175 spin_unlock(&cluster->lock);
3176 spin_unlock(&ctl->tree_lock);
3177
3178 return ret;
3179}
3180
3181static int btrfs_bitmap_cluster(struct btrfs_block_group *block_group,
3182 struct btrfs_free_space *entry,
3183 struct btrfs_free_cluster *cluster,
3184 u64 offset, u64 bytes,
3185 u64 cont1_bytes, u64 min_bytes)
3186{
3187 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3188 unsigned long next_zero;
3189 unsigned long i;
3190 unsigned long want_bits;
3191 unsigned long min_bits;
3192 unsigned long found_bits;
3193 unsigned long max_bits = 0;
3194 unsigned long start = 0;
3195 unsigned long total_found = 0;
3196 int ret;
3197
3198 i = offset_to_bit(entry->offset, ctl->unit,
3199 max_t(u64, offset, entry->offset));
3200 want_bits = bytes_to_bits(bytes, ctl->unit);
3201 min_bits = bytes_to_bits(min_bytes, ctl->unit);
3202
3203 /*
3204 * Don't bother looking for a cluster in this bitmap if it's heavily
3205 * fragmented.
3206 */
3207 if (entry->max_extent_size &&
3208 entry->max_extent_size < cont1_bytes)
3209 return -ENOSPC;
3210again:
3211 found_bits = 0;
3212 for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
3213 next_zero = find_next_zero_bit(entry->bitmap,
3214 BITS_PER_BITMAP, i);
3215 if (next_zero - i >= min_bits) {
3216 found_bits = next_zero - i;
3217 if (found_bits > max_bits)
3218 max_bits = found_bits;
3219 break;
3220 }
3221 if (next_zero - i > max_bits)
3222 max_bits = next_zero - i;
3223 i = next_zero;
3224 }
3225
3226 if (!found_bits) {
3227 entry->max_extent_size = (u64)max_bits * ctl->unit;
3228 return -ENOSPC;
3229 }
3230
3231 if (!total_found) {
3232 start = i;
3233 cluster->max_size = 0;
3234 }
3235
3236 total_found += found_bits;
3237
3238 if (cluster->max_size < found_bits * ctl->unit)
3239 cluster->max_size = found_bits * ctl->unit;
3240
3241 if (total_found < want_bits || cluster->max_size < cont1_bytes) {
3242 i = next_zero + 1;
3243 goto again;
3244 }
3245
3246 cluster->window_start = start * ctl->unit + entry->offset;
3247 rb_erase(&entry->offset_index, &ctl->free_space_offset);
3248 ret = tree_insert_offset(&cluster->root, entry->offset,
3249 &entry->offset_index, 1);
3250 ASSERT(!ret); /* -EEXIST; Logic error */
3251
3252 trace_btrfs_setup_cluster(block_group, cluster,
3253 total_found * ctl->unit, 1);
3254 return 0;
3255}
3256
3257/*
3258 * This searches the block group for just extents to fill the cluster with.
3259 * Try to find a cluster with at least bytes total bytes, at least one
3260 * extent of cont1_bytes, and other clusters of at least min_bytes.
3261 */
3262static noinline int
3263setup_cluster_no_bitmap(struct btrfs_block_group *block_group,
3264 struct btrfs_free_cluster *cluster,
3265 struct list_head *bitmaps, u64 offset, u64 bytes,
3266 u64 cont1_bytes, u64 min_bytes)
3267{
3268 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3269 struct btrfs_free_space *first = NULL;
3270 struct btrfs_free_space *entry = NULL;
3271 struct btrfs_free_space *last;
3272 struct rb_node *node;
3273 u64 window_free;
3274 u64 max_extent;
3275 u64 total_size = 0;
3276
3277 entry = tree_search_offset(ctl, offset, 0, 1);
3278 if (!entry)
3279 return -ENOSPC;
3280
3281 /*
3282 * We don't want bitmaps, so just move along until we find a normal
3283 * extent entry.
3284 */
3285 while (entry->bitmap || entry->bytes < min_bytes) {
3286 if (entry->bitmap && list_empty(&entry->list))
3287 list_add_tail(&entry->list, bitmaps);
3288 node = rb_next(&entry->offset_index);
3289 if (!node)
3290 return -ENOSPC;
3291 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3292 }
3293
3294 window_free = entry->bytes;
3295 max_extent = entry->bytes;
3296 first = entry;
3297 last = entry;
3298
3299 for (node = rb_next(&entry->offset_index); node;
3300 node = rb_next(&entry->offset_index)) {
3301 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3302
3303 if (entry->bitmap) {
3304 if (list_empty(&entry->list))
3305 list_add_tail(&entry->list, bitmaps);
3306 continue;
3307 }
3308
3309 if (entry->bytes < min_bytes)
3310 continue;
3311
3312 last = entry;
3313 window_free += entry->bytes;
3314 if (entry->bytes > max_extent)
3315 max_extent = entry->bytes;
3316 }
3317
3318 if (window_free < bytes || max_extent < cont1_bytes)
3319 return -ENOSPC;
3320
3321 cluster->window_start = first->offset;
3322
3323 node = &first->offset_index;
3324
3325 /*
3326 * now we've found our entries, pull them out of the free space
3327 * cache and put them into the cluster rbtree
3328 */
3329 do {
3330 int ret;
3331
3332 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3333 node = rb_next(&entry->offset_index);
3334 if (entry->bitmap || entry->bytes < min_bytes)
3335 continue;
3336
3337 rb_erase(&entry->offset_index, &ctl->free_space_offset);
3338 ret = tree_insert_offset(&cluster->root, entry->offset,
3339 &entry->offset_index, 0);
3340 total_size += entry->bytes;
3341 ASSERT(!ret); /* -EEXIST; Logic error */
3342 } while (node && entry != last);
3343
3344 cluster->max_size = max_extent;
3345 trace_btrfs_setup_cluster(block_group, cluster, total_size, 0);
3346 return 0;
3347}
3348
3349/*
3350 * This specifically looks for bitmaps that may work in the cluster, we assume
3351 * that we have already failed to find extents that will work.
3352 */
3353static noinline int
3354setup_cluster_bitmap(struct btrfs_block_group *block_group,
3355 struct btrfs_free_cluster *cluster,
3356 struct list_head *bitmaps, u64 offset, u64 bytes,
3357 u64 cont1_bytes, u64 min_bytes)
3358{
3359 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3360 struct btrfs_free_space *entry = NULL;
3361 int ret = -ENOSPC;
3362 u64 bitmap_offset = offset_to_bitmap(ctl, offset);
3363
3364 if (ctl->total_bitmaps == 0)
3365 return -ENOSPC;
3366
3367 /*
3368 * The bitmap that covers offset won't be in the list unless offset
3369 * is just its start offset.
3370 */
3371 if (!list_empty(bitmaps))
3372 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
3373
3374 if (!entry || entry->offset != bitmap_offset) {
3375 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
3376 if (entry && list_empty(&entry->list))
3377 list_add(&entry->list, bitmaps);
3378 }
3379
3380 list_for_each_entry(entry, bitmaps, list) {
3381 if (entry->bytes < bytes)
3382 continue;
3383 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
3384 bytes, cont1_bytes, min_bytes);
3385 if (!ret)
3386 return 0;
3387 }
3388
3389 /*
3390 * The bitmaps list has all the bitmaps that record free space
3391 * starting after offset, so no more search is required.
3392 */
3393 return -ENOSPC;
3394}
3395
3396/*
3397 * here we try to find a cluster of blocks in a block group. The goal
3398 * is to find at least bytes+empty_size.
3399 * We might not find them all in one contiguous area.
3400 *
3401 * returns zero and sets up cluster if things worked out, otherwise
3402 * it returns -enospc
3403 */
3404int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
3405 struct btrfs_free_cluster *cluster,
3406 u64 offset, u64 bytes, u64 empty_size)
3407{
3408 struct btrfs_fs_info *fs_info = block_group->fs_info;
3409 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3410 struct btrfs_free_space *entry, *tmp;
3411 LIST_HEAD(bitmaps);
3412 u64 min_bytes;
3413 u64 cont1_bytes;
3414 int ret;
3415
3416 /*
3417 * Choose the minimum extent size we'll require for this
3418 * cluster. For SSD_SPREAD, don't allow any fragmentation.
3419 * For metadata, allow allocates with smaller extents. For
3420 * data, keep it dense.
3421 */
3422 if (btrfs_test_opt(fs_info, SSD_SPREAD)) {
3423 cont1_bytes = min_bytes = bytes + empty_size;
3424 } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
3425 cont1_bytes = bytes;
3426 min_bytes = fs_info->sectorsize;
3427 } else {
3428 cont1_bytes = max(bytes, (bytes + empty_size) >> 2);
3429 min_bytes = fs_info->sectorsize;
3430 }
3431
3432 spin_lock(&ctl->tree_lock);
3433
3434 /*
3435 * If we know we don't have enough space to make a cluster don't even
3436 * bother doing all the work to try and find one.
3437 */
3438 if (ctl->free_space < bytes) {
3439 spin_unlock(&ctl->tree_lock);
3440 return -ENOSPC;
3441 }
3442
3443 spin_lock(&cluster->lock);
3444
3445 /* someone already found a cluster, hooray */
3446 if (cluster->block_group) {
3447 ret = 0;
3448 goto out;
3449 }
3450
3451 trace_btrfs_find_cluster(block_group, offset, bytes, empty_size,
3452 min_bytes);
3453
3454 ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
3455 bytes + empty_size,
3456 cont1_bytes, min_bytes);
3457 if (ret)
3458 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
3459 offset, bytes + empty_size,
3460 cont1_bytes, min_bytes);
3461
3462 /* Clear our temporary list */
3463 list_for_each_entry_safe(entry, tmp, &bitmaps, list)
3464 list_del_init(&entry->list);
3465
3466 if (!ret) {
3467 btrfs_get_block_group(block_group);
3468 list_add_tail(&cluster->block_group_list,
3469 &block_group->cluster_list);
3470 cluster->block_group = block_group;
3471 } else {
3472 trace_btrfs_failed_cluster_setup(block_group);
3473 }
3474out:
3475 spin_unlock(&cluster->lock);
3476 spin_unlock(&ctl->tree_lock);
3477
3478 return ret;
3479}
3480
3481/*
3482 * simple code to zero out a cluster
3483 */
3484void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
3485{
3486 spin_lock_init(&cluster->lock);
3487 spin_lock_init(&cluster->refill_lock);
3488 cluster->root = RB_ROOT;
3489 cluster->max_size = 0;
3490 cluster->fragmented = false;
3491 INIT_LIST_HEAD(&cluster->block_group_list);
3492 cluster->block_group = NULL;
3493}
3494
3495static int do_trimming(struct btrfs_block_group *block_group,
3496 u64 *total_trimmed, u64 start, u64 bytes,
3497 u64 reserved_start, u64 reserved_bytes,
3498 enum btrfs_trim_state reserved_trim_state,
3499 struct btrfs_trim_range *trim_entry)
3500{
3501 struct btrfs_space_info *space_info = block_group->space_info;
3502 struct btrfs_fs_info *fs_info = block_group->fs_info;
3503 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3504 int ret;
3505 int update = 0;
3506 const u64 end = start + bytes;
3507 const u64 reserved_end = reserved_start + reserved_bytes;
3508 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3509 u64 trimmed = 0;
3510
3511 spin_lock(&space_info->lock);
3512 spin_lock(&block_group->lock);
3513 if (!block_group->ro) {
3514 block_group->reserved += reserved_bytes;
3515 space_info->bytes_reserved += reserved_bytes;
3516 update = 1;
3517 }
3518 spin_unlock(&block_group->lock);
3519 spin_unlock(&space_info->lock);
3520
3521 ret = btrfs_discard_extent(fs_info, start, bytes, &trimmed);
3522 if (!ret) {
3523 *total_trimmed += trimmed;
3524 trim_state = BTRFS_TRIM_STATE_TRIMMED;
3525 }
3526
3527 mutex_lock(&ctl->cache_writeout_mutex);
3528 if (reserved_start < start)
3529 __btrfs_add_free_space(fs_info, ctl, reserved_start,
3530 start - reserved_start,
3531 reserved_trim_state);
3532 if (start + bytes < reserved_start + reserved_bytes)
3533 __btrfs_add_free_space(fs_info, ctl, end, reserved_end - end,
3534 reserved_trim_state);
3535 __btrfs_add_free_space(fs_info, ctl, start, bytes, trim_state);
3536 list_del(&trim_entry->list);
3537 mutex_unlock(&ctl->cache_writeout_mutex);
3538
3539 if (update) {
3540 spin_lock(&space_info->lock);
3541 spin_lock(&block_group->lock);
3542 if (block_group->ro)
3543 space_info->bytes_readonly += reserved_bytes;
3544 block_group->reserved -= reserved_bytes;
3545 space_info->bytes_reserved -= reserved_bytes;
3546 spin_unlock(&block_group->lock);
3547 spin_unlock(&space_info->lock);
3548 }
3549
3550 return ret;
3551}
3552
3553/*
3554 * If @async is set, then we will trim 1 region and return.
3555 */
3556static int trim_no_bitmap(struct btrfs_block_group *block_group,
3557 u64 *total_trimmed, u64 start, u64 end, u64 minlen,
3558 bool async)
3559{
3560 struct btrfs_discard_ctl *discard_ctl =
3561 &block_group->fs_info->discard_ctl;
3562 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3563 struct btrfs_free_space *entry;
3564 struct rb_node *node;
3565 int ret = 0;
3566 u64 extent_start;
3567 u64 extent_bytes;
3568 enum btrfs_trim_state extent_trim_state;
3569 u64 bytes;
3570 const u64 max_discard_size = READ_ONCE(discard_ctl->max_discard_size);
3571
3572 while (start < end) {
3573 struct btrfs_trim_range trim_entry;
3574
3575 mutex_lock(&ctl->cache_writeout_mutex);
3576 spin_lock(&ctl->tree_lock);
3577
3578 if (ctl->free_space < minlen)
3579 goto out_unlock;
3580
3581 entry = tree_search_offset(ctl, start, 0, 1);
3582 if (!entry)
3583 goto out_unlock;
3584
3585 /* Skip bitmaps and if async, already trimmed entries */
3586 while (entry->bitmap ||
3587 (async && btrfs_free_space_trimmed(entry))) {
3588 node = rb_next(&entry->offset_index);
3589 if (!node)
3590 goto out_unlock;
3591 entry = rb_entry(node, struct btrfs_free_space,
3592 offset_index);
3593 }
3594
3595 if (entry->offset >= end)
3596 goto out_unlock;
3597
3598 extent_start = entry->offset;
3599 extent_bytes = entry->bytes;
3600 extent_trim_state = entry->trim_state;
3601 if (async) {
3602 start = entry->offset;
3603 bytes = entry->bytes;
3604 if (bytes < minlen) {
3605 spin_unlock(&ctl->tree_lock);
3606 mutex_unlock(&ctl->cache_writeout_mutex);
3607 goto next;
3608 }
3609 unlink_free_space(ctl, entry);
3610 /*
3611 * Let bytes = BTRFS_MAX_DISCARD_SIZE + X.
3612 * If X < BTRFS_ASYNC_DISCARD_MIN_FILTER, we won't trim
3613 * X when we come back around. So trim it now.
3614 */
3615 if (max_discard_size &&
3616 bytes >= (max_discard_size +
3617 BTRFS_ASYNC_DISCARD_MIN_FILTER)) {
3618 bytes = max_discard_size;
3619 extent_bytes = max_discard_size;
3620 entry->offset += max_discard_size;
3621 entry->bytes -= max_discard_size;
3622 link_free_space(ctl, entry);
3623 } else {
3624 kmem_cache_free(btrfs_free_space_cachep, entry);
3625 }
3626 } else {
3627 start = max(start, extent_start);
3628 bytes = min(extent_start + extent_bytes, end) - start;
3629 if (bytes < minlen) {
3630 spin_unlock(&ctl->tree_lock);
3631 mutex_unlock(&ctl->cache_writeout_mutex);
3632 goto next;
3633 }
3634
3635 unlink_free_space(ctl, entry);
3636 kmem_cache_free(btrfs_free_space_cachep, entry);
3637 }
3638
3639 spin_unlock(&ctl->tree_lock);
3640 trim_entry.start = extent_start;
3641 trim_entry.bytes = extent_bytes;
3642 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3643 mutex_unlock(&ctl->cache_writeout_mutex);
3644
3645 ret = do_trimming(block_group, total_trimmed, start, bytes,
3646 extent_start, extent_bytes, extent_trim_state,
3647 &trim_entry);
3648 if (ret) {
3649 block_group->discard_cursor = start + bytes;
3650 break;
3651 }
3652next:
3653 start += bytes;
3654 block_group->discard_cursor = start;
3655 if (async && *total_trimmed)
3656 break;
3657
3658 if (fatal_signal_pending(current)) {
3659 ret = -ERESTARTSYS;
3660 break;
3661 }
3662
3663 cond_resched();
3664 }
3665
3666 return ret;
3667
3668out_unlock:
3669 block_group->discard_cursor = btrfs_block_group_end(block_group);
3670 spin_unlock(&ctl->tree_lock);
3671 mutex_unlock(&ctl->cache_writeout_mutex);
3672
3673 return ret;
3674}
3675
3676/*
3677 * If we break out of trimming a bitmap prematurely, we should reset the
3678 * trimming bit. In a rather contrieved case, it's possible to race here so
3679 * reset the state to BTRFS_TRIM_STATE_UNTRIMMED.
3680 *
3681 * start = start of bitmap
3682 * end = near end of bitmap
3683 *
3684 * Thread 1: Thread 2:
3685 * trim_bitmaps(start)
3686 * trim_bitmaps(end)
3687 * end_trimming_bitmap()
3688 * reset_trimming_bitmap()
3689 */
3690static void reset_trimming_bitmap(struct btrfs_free_space_ctl *ctl, u64 offset)
3691{
3692 struct btrfs_free_space *entry;
3693
3694 spin_lock(&ctl->tree_lock);
3695 entry = tree_search_offset(ctl, offset, 1, 0);
3696 if (entry) {
3697 if (btrfs_free_space_trimmed(entry)) {
3698 ctl->discardable_extents[BTRFS_STAT_CURR] +=
3699 entry->bitmap_extents;
3700 ctl->discardable_bytes[BTRFS_STAT_CURR] += entry->bytes;
3701 }
3702 entry->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3703 }
3704
3705 spin_unlock(&ctl->tree_lock);
3706}
3707
3708static void end_trimming_bitmap(struct btrfs_free_space_ctl *ctl,
3709 struct btrfs_free_space *entry)
3710{
3711 if (btrfs_free_space_trimming_bitmap(entry)) {
3712 entry->trim_state = BTRFS_TRIM_STATE_TRIMMED;
3713 ctl->discardable_extents[BTRFS_STAT_CURR] -=
3714 entry->bitmap_extents;
3715 ctl->discardable_bytes[BTRFS_STAT_CURR] -= entry->bytes;
3716 }
3717}
3718
3719/*
3720 * If @async is set, then we will trim 1 region and return.
3721 */
3722static int trim_bitmaps(struct btrfs_block_group *block_group,
3723 u64 *total_trimmed, u64 start, u64 end, u64 minlen,
3724 u64 maxlen, bool async)
3725{
3726 struct btrfs_discard_ctl *discard_ctl =
3727 &block_group->fs_info->discard_ctl;
3728 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3729 struct btrfs_free_space *entry;
3730 int ret = 0;
3731 int ret2;
3732 u64 bytes;
3733 u64 offset = offset_to_bitmap(ctl, start);
3734 const u64 max_discard_size = READ_ONCE(discard_ctl->max_discard_size);
3735
3736 while (offset < end) {
3737 bool next_bitmap = false;
3738 struct btrfs_trim_range trim_entry;
3739
3740 mutex_lock(&ctl->cache_writeout_mutex);
3741 spin_lock(&ctl->tree_lock);
3742
3743 if (ctl->free_space < minlen) {
3744 block_group->discard_cursor =
3745 btrfs_block_group_end(block_group);
3746 spin_unlock(&ctl->tree_lock);
3747 mutex_unlock(&ctl->cache_writeout_mutex);
3748 break;
3749 }
3750
3751 entry = tree_search_offset(ctl, offset, 1, 0);
3752 /*
3753 * Bitmaps are marked trimmed lossily now to prevent constant
3754 * discarding of the same bitmap (the reason why we are bound
3755 * by the filters). So, retrim the block group bitmaps when we
3756 * are preparing to punt to the unused_bgs list. This uses
3757 * @minlen to determine if we are in BTRFS_DISCARD_INDEX_UNUSED
3758 * which is the only discard index which sets minlen to 0.
3759 */
3760 if (!entry || (async && minlen && start == offset &&
3761 btrfs_free_space_trimmed(entry))) {
3762 spin_unlock(&ctl->tree_lock);
3763 mutex_unlock(&ctl->cache_writeout_mutex);
3764 next_bitmap = true;
3765 goto next;
3766 }
3767
3768 /*
3769 * Async discard bitmap trimming begins at by setting the start
3770 * to be key.objectid and the offset_to_bitmap() aligns to the
3771 * start of the bitmap. This lets us know we are fully
3772 * scanning the bitmap rather than only some portion of it.
3773 */
3774 if (start == offset)
3775 entry->trim_state = BTRFS_TRIM_STATE_TRIMMING;
3776
3777 bytes = minlen;
3778 ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
3779 if (ret2 || start >= end) {
3780 /*
3781 * We lossily consider a bitmap trimmed if we only skip
3782 * over regions <= BTRFS_ASYNC_DISCARD_MIN_FILTER.
3783 */
3784 if (ret2 && minlen <= BTRFS_ASYNC_DISCARD_MIN_FILTER)
3785 end_trimming_bitmap(ctl, entry);
3786 else
3787 entry->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3788 spin_unlock(&ctl->tree_lock);
3789 mutex_unlock(&ctl->cache_writeout_mutex);
3790 next_bitmap = true;
3791 goto next;
3792 }
3793
3794 /*
3795 * We already trimmed a region, but are using the locking above
3796 * to reset the trim_state.
3797 */
3798 if (async && *total_trimmed) {
3799 spin_unlock(&ctl->tree_lock);
3800 mutex_unlock(&ctl->cache_writeout_mutex);
3801 goto out;
3802 }
3803
3804 bytes = min(bytes, end - start);
3805 if (bytes < minlen || (async && maxlen && bytes > maxlen)) {
3806 spin_unlock(&ctl->tree_lock);
3807 mutex_unlock(&ctl->cache_writeout_mutex);
3808 goto next;
3809 }
3810
3811 /*
3812 * Let bytes = BTRFS_MAX_DISCARD_SIZE + X.
3813 * If X < @minlen, we won't trim X when we come back around.
3814 * So trim it now. We differ here from trimming extents as we
3815 * don't keep individual state per bit.
3816 */
3817 if (async &&
3818 max_discard_size &&
3819 bytes > (max_discard_size + minlen))
3820 bytes = max_discard_size;
3821
3822 bitmap_clear_bits(ctl, entry, start, bytes);
3823 if (entry->bytes == 0)
3824 free_bitmap(ctl, entry);
3825
3826 spin_unlock(&ctl->tree_lock);
3827 trim_entry.start = start;
3828 trim_entry.bytes = bytes;
3829 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3830 mutex_unlock(&ctl->cache_writeout_mutex);
3831
3832 ret = do_trimming(block_group, total_trimmed, start, bytes,
3833 start, bytes, 0, &trim_entry);
3834 if (ret) {
3835 reset_trimming_bitmap(ctl, offset);
3836 block_group->discard_cursor =
3837 btrfs_block_group_end(block_group);
3838 break;
3839 }
3840next:
3841 if (next_bitmap) {
3842 offset += BITS_PER_BITMAP * ctl->unit;
3843 start = offset;
3844 } else {
3845 start += bytes;
3846 }
3847 block_group->discard_cursor = start;
3848
3849 if (fatal_signal_pending(current)) {
3850 if (start != offset)
3851 reset_trimming_bitmap(ctl, offset);
3852 ret = -ERESTARTSYS;
3853 break;
3854 }
3855
3856 cond_resched();
3857 }
3858
3859 if (offset >= end)
3860 block_group->discard_cursor = end;
3861
3862out:
3863 return ret;
3864}
3865
3866int btrfs_trim_block_group(struct btrfs_block_group *block_group,
3867 u64 *trimmed, u64 start, u64 end, u64 minlen)
3868{
3869 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3870 int ret;
3871 u64 rem = 0;
3872
3873 ASSERT(!btrfs_is_zoned(block_group->fs_info));
3874
3875 *trimmed = 0;
3876
3877 spin_lock(&block_group->lock);
3878 if (block_group->removed) {
3879 spin_unlock(&block_group->lock);
3880 return 0;
3881 }
3882 btrfs_freeze_block_group(block_group);
3883 spin_unlock(&block_group->lock);
3884
3885 ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, false);
3886 if (ret)
3887 goto out;
3888
3889 ret = trim_bitmaps(block_group, trimmed, start, end, minlen, 0, false);
3890 div64_u64_rem(end, BITS_PER_BITMAP * ctl->unit, &rem);
3891 /* If we ended in the middle of a bitmap, reset the trimming flag */
3892 if (rem)
3893 reset_trimming_bitmap(ctl, offset_to_bitmap(ctl, end));
3894out:
3895 btrfs_unfreeze_block_group(block_group);
3896 return ret;
3897}
3898
3899int btrfs_trim_block_group_extents(struct btrfs_block_group *block_group,
3900 u64 *trimmed, u64 start, u64 end, u64 minlen,
3901 bool async)
3902{
3903 int ret;
3904
3905 *trimmed = 0;
3906
3907 spin_lock(&block_group->lock);
3908 if (block_group->removed) {
3909 spin_unlock(&block_group->lock);
3910 return 0;
3911 }
3912 btrfs_freeze_block_group(block_group);
3913 spin_unlock(&block_group->lock);
3914
3915 ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, async);
3916 btrfs_unfreeze_block_group(block_group);
3917
3918 return ret;
3919}
3920
3921int btrfs_trim_block_group_bitmaps(struct btrfs_block_group *block_group,
3922 u64 *trimmed, u64 start, u64 end, u64 minlen,
3923 u64 maxlen, bool async)
3924{
3925 int ret;
3926
3927 *trimmed = 0;
3928
3929 spin_lock(&block_group->lock);
3930 if (block_group->removed) {
3931 spin_unlock(&block_group->lock);
3932 return 0;
3933 }
3934 btrfs_freeze_block_group(block_group);
3935 spin_unlock(&block_group->lock);
3936
3937 ret = trim_bitmaps(block_group, trimmed, start, end, minlen, maxlen,
3938 async);
3939
3940 btrfs_unfreeze_block_group(block_group);
3941
3942 return ret;
3943}
3944
3945bool btrfs_free_space_cache_v1_active(struct btrfs_fs_info *fs_info)
3946{
3947 return btrfs_super_cache_generation(fs_info->super_copy);
3948}
3949
3950static int cleanup_free_space_cache_v1(struct btrfs_fs_info *fs_info,
3951 struct btrfs_trans_handle *trans)
3952{
3953 struct btrfs_block_group *block_group;
3954 struct rb_node *node;
3955 int ret = 0;
3956
3957 btrfs_info(fs_info, "cleaning free space cache v1");
3958
3959 node = rb_first(&fs_info->block_group_cache_tree);
3960 while (node) {
3961 block_group = rb_entry(node, struct btrfs_block_group, cache_node);
3962 ret = btrfs_remove_free_space_inode(trans, NULL, block_group);
3963 if (ret)
3964 goto out;
3965 node = rb_next(node);
3966 }
3967out:
3968 return ret;
3969}
3970
3971int btrfs_set_free_space_cache_v1_active(struct btrfs_fs_info *fs_info, bool active)
3972{
3973 struct btrfs_trans_handle *trans;
3974 int ret;
3975
3976 /*
3977 * update_super_roots will appropriately set or unset
3978 * super_copy->cache_generation based on SPACE_CACHE and
3979 * BTRFS_FS_CLEANUP_SPACE_CACHE_V1. For this reason, we need a
3980 * transaction commit whether we are enabling space cache v1 and don't
3981 * have any other work to do, or are disabling it and removing free
3982 * space inodes.
3983 */
3984 trans = btrfs_start_transaction(fs_info->tree_root, 0);
3985 if (IS_ERR(trans))
3986 return PTR_ERR(trans);
3987
3988 if (!active) {
3989 set_bit(BTRFS_FS_CLEANUP_SPACE_CACHE_V1, &fs_info->flags);
3990 ret = cleanup_free_space_cache_v1(fs_info, trans);
3991 if (ret) {
3992 btrfs_abort_transaction(trans, ret);
3993 btrfs_end_transaction(trans);
3994 goto out;
3995 }
3996 }
3997
3998 ret = btrfs_commit_transaction(trans);
3999out:
4000 clear_bit(BTRFS_FS_CLEANUP_SPACE_CACHE_V1, &fs_info->flags);
4001
4002 return ret;
4003}
4004
4005#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
4006/*
4007 * Use this if you need to make a bitmap or extent entry specifically, it
4008 * doesn't do any of the merging that add_free_space does, this acts a lot like
4009 * how the free space cache loading stuff works, so you can get really weird
4010 * configurations.
4011 */
4012int test_add_free_space_entry(struct btrfs_block_group *cache,
4013 u64 offset, u64 bytes, bool bitmap)
4014{
4015 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
4016 struct btrfs_free_space *info = NULL, *bitmap_info;
4017 void *map = NULL;
4018 enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_TRIMMED;
4019 u64 bytes_added;
4020 int ret;
4021
4022again:
4023 if (!info) {
4024 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
4025 if (!info)
4026 return -ENOMEM;
4027 }
4028
4029 if (!bitmap) {
4030 spin_lock(&ctl->tree_lock);
4031 info->offset = offset;
4032 info->bytes = bytes;
4033 info->max_extent_size = 0;
4034 ret = link_free_space(ctl, info);
4035 spin_unlock(&ctl->tree_lock);
4036 if (ret)
4037 kmem_cache_free(btrfs_free_space_cachep, info);
4038 return ret;
4039 }
4040
4041 if (!map) {
4042 map = kmem_cache_zalloc(btrfs_free_space_bitmap_cachep, GFP_NOFS);
4043 if (!map) {
4044 kmem_cache_free(btrfs_free_space_cachep, info);
4045 return -ENOMEM;
4046 }
4047 }
4048
4049 spin_lock(&ctl->tree_lock);
4050 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
4051 1, 0);
4052 if (!bitmap_info) {
4053 info->bitmap = map;
4054 map = NULL;
4055 add_new_bitmap(ctl, info, offset);
4056 bitmap_info = info;
4057 info = NULL;
4058 }
4059
4060 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes,
4061 trim_state);
4062
4063 bytes -= bytes_added;
4064 offset += bytes_added;
4065 spin_unlock(&ctl->tree_lock);
4066
4067 if (bytes)
4068 goto again;
4069
4070 if (info)
4071 kmem_cache_free(btrfs_free_space_cachep, info);
4072 if (map)
4073 kmem_cache_free(btrfs_free_space_bitmap_cachep, map);
4074 return 0;
4075}
4076
4077/*
4078 * Checks to see if the given range is in the free space cache. This is really
4079 * just used to check the absence of space, so if there is free space in the
4080 * range at all we will return 1.
4081 */
4082int test_check_exists(struct btrfs_block_group *cache,
4083 u64 offset, u64 bytes)
4084{
4085 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
4086 struct btrfs_free_space *info;
4087 int ret = 0;
4088
4089 spin_lock(&ctl->tree_lock);
4090 info = tree_search_offset(ctl, offset, 0, 0);
4091 if (!info) {
4092 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
4093 1, 0);
4094 if (!info)
4095 goto out;
4096 }
4097
4098have_info:
4099 if (info->bitmap) {
4100 u64 bit_off, bit_bytes;
4101 struct rb_node *n;
4102 struct btrfs_free_space *tmp;
4103
4104 bit_off = offset;
4105 bit_bytes = ctl->unit;
4106 ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
4107 if (!ret) {
4108 if (bit_off == offset) {
4109 ret = 1;
4110 goto out;
4111 } else if (bit_off > offset &&
4112 offset + bytes > bit_off) {
4113 ret = 1;
4114 goto out;
4115 }
4116 }
4117
4118 n = rb_prev(&info->offset_index);
4119 while (n) {
4120 tmp = rb_entry(n, struct btrfs_free_space,
4121 offset_index);
4122 if (tmp->offset + tmp->bytes < offset)
4123 break;
4124 if (offset + bytes < tmp->offset) {
4125 n = rb_prev(&tmp->offset_index);
4126 continue;
4127 }
4128 info = tmp;
4129 goto have_info;
4130 }
4131
4132 n = rb_next(&info->offset_index);
4133 while (n) {
4134 tmp = rb_entry(n, struct btrfs_free_space,
4135 offset_index);
4136 if (offset + bytes < tmp->offset)
4137 break;
4138 if (tmp->offset + tmp->bytes < offset) {
4139 n = rb_next(&tmp->offset_index);
4140 continue;
4141 }
4142 info = tmp;
4143 goto have_info;
4144 }
4145
4146 ret = 0;
4147 goto out;
4148 }
4149
4150 if (info->offset == offset) {
4151 ret = 1;
4152 goto out;
4153 }
4154
4155 if (offset > info->offset && offset < info->offset + info->bytes)
4156 ret = 1;
4157out:
4158 spin_unlock(&ctl->tree_lock);
4159 return ret;
4160}
4161#endif /* CONFIG_BTRFS_FS_RUN_SANITY_TESTS */
1/*
2 * Copyright (C) 2008 Red Hat. All rights reserved.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
17 */
18
19#include <linux/pagemap.h>
20#include <linux/sched.h>
21#include <linux/slab.h>
22#include <linux/math64.h>
23#include <linux/ratelimit.h>
24#include "ctree.h"
25#include "free-space-cache.h"
26#include "transaction.h"
27#include "disk-io.h"
28#include "extent_io.h"
29#include "inode-map.h"
30#include "volumes.h"
31
32#define BITS_PER_BITMAP (PAGE_SIZE * 8UL)
33#define MAX_CACHE_BYTES_PER_GIG SZ_32K
34
35struct btrfs_trim_range {
36 u64 start;
37 u64 bytes;
38 struct list_head list;
39};
40
41static int link_free_space(struct btrfs_free_space_ctl *ctl,
42 struct btrfs_free_space *info);
43static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
44 struct btrfs_free_space *info);
45static int btrfs_wait_cache_io_root(struct btrfs_root *root,
46 struct btrfs_trans_handle *trans,
47 struct btrfs_io_ctl *io_ctl,
48 struct btrfs_path *path);
49
50static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
51 struct btrfs_path *path,
52 u64 offset)
53{
54 struct btrfs_fs_info *fs_info = root->fs_info;
55 struct btrfs_key key;
56 struct btrfs_key location;
57 struct btrfs_disk_key disk_key;
58 struct btrfs_free_space_header *header;
59 struct extent_buffer *leaf;
60 struct inode *inode = NULL;
61 int ret;
62
63 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
64 key.offset = offset;
65 key.type = 0;
66
67 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
68 if (ret < 0)
69 return ERR_PTR(ret);
70 if (ret > 0) {
71 btrfs_release_path(path);
72 return ERR_PTR(-ENOENT);
73 }
74
75 leaf = path->nodes[0];
76 header = btrfs_item_ptr(leaf, path->slots[0],
77 struct btrfs_free_space_header);
78 btrfs_free_space_key(leaf, header, &disk_key);
79 btrfs_disk_key_to_cpu(&location, &disk_key);
80 btrfs_release_path(path);
81
82 inode = btrfs_iget(fs_info->sb, &location, root, NULL);
83 if (IS_ERR(inode))
84 return inode;
85 if (is_bad_inode(inode)) {
86 iput(inode);
87 return ERR_PTR(-ENOENT);
88 }
89
90 mapping_set_gfp_mask(inode->i_mapping,
91 mapping_gfp_constraint(inode->i_mapping,
92 ~(__GFP_FS | __GFP_HIGHMEM)));
93
94 return inode;
95}
96
97struct inode *lookup_free_space_inode(struct btrfs_root *root,
98 struct btrfs_block_group_cache
99 *block_group, struct btrfs_path *path)
100{
101 struct inode *inode = NULL;
102 struct btrfs_fs_info *fs_info = root->fs_info;
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(root, path,
113 block_group->key.objectid);
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
134static 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
195int create_free_space_inode(struct btrfs_root *root,
196 struct btrfs_trans_handle *trans,
197 struct btrfs_block_group_cache *block_group,
198 struct btrfs_path *path)
199{
200 int ret;
201 u64 ino;
202
203 ret = btrfs_find_free_objectid(root, &ino);
204 if (ret < 0)
205 return ret;
206
207 return __create_free_space_inode(root, trans, path, ino,
208 block_group->key.objectid);
209}
210
211int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
212 struct btrfs_block_rsv *rsv)
213{
214 u64 needed_bytes;
215 int ret;
216
217 /* 1 for slack space, 1 for updating the inode */
218 needed_bytes = btrfs_calc_trunc_metadata_size(fs_info, 1) +
219 btrfs_calc_trans_metadata_size(fs_info, 1);
220
221 spin_lock(&rsv->lock);
222 if (rsv->reserved < needed_bytes)
223 ret = -ENOSPC;
224 else
225 ret = 0;
226 spin_unlock(&rsv->lock);
227 return ret;
228}
229
230int btrfs_truncate_free_space_cache(struct btrfs_root *root,
231 struct btrfs_trans_handle *trans,
232 struct btrfs_block_group_cache *block_group,
233 struct inode *inode)
234{
235 int ret = 0;
236 struct btrfs_path *path = btrfs_alloc_path();
237 bool locked = false;
238
239 if (!path) {
240 ret = -ENOMEM;
241 goto fail;
242 }
243
244 if (block_group) {
245 locked = true;
246 mutex_lock(&trans->transaction->cache_write_mutex);
247 if (!list_empty(&block_group->io_list)) {
248 list_del_init(&block_group->io_list);
249
250 btrfs_wait_cache_io(trans, block_group, path);
251 btrfs_put_block_group(block_group);
252 }
253
254 /*
255 * now that we've truncated the cache away, its no longer
256 * setup or written
257 */
258 spin_lock(&block_group->lock);
259 block_group->disk_cache_state = BTRFS_DC_CLEAR;
260 spin_unlock(&block_group->lock);
261 }
262 btrfs_free_path(path);
263
264 btrfs_i_size_write(inode, 0);
265 truncate_pagecache(inode, 0);
266
267 /*
268 * We don't need an orphan item because truncating the free space cache
269 * will never be split across transactions.
270 * We don't need to check for -EAGAIN because we're a free space
271 * cache inode
272 */
273 ret = btrfs_truncate_inode_items(trans, root, inode,
274 0, BTRFS_EXTENT_DATA_KEY);
275 if (ret)
276 goto fail;
277
278 ret = btrfs_update_inode(trans, root, inode);
279
280fail:
281 if (locked)
282 mutex_unlock(&trans->transaction->cache_write_mutex);
283 if (ret)
284 btrfs_abort_transaction(trans, ret);
285
286 return ret;
287}
288
289static int readahead_cache(struct inode *inode)
290{
291 struct file_ra_state *ra;
292 unsigned long last_index;
293
294 ra = kzalloc(sizeof(*ra), GFP_NOFS);
295 if (!ra)
296 return -ENOMEM;
297
298 file_ra_state_init(ra, inode->i_mapping);
299 last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
300
301 page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
302
303 kfree(ra);
304
305 return 0;
306}
307
308static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
309 int write)
310{
311 int num_pages;
312 int check_crcs = 0;
313
314 num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
315
316 if (btrfs_ino(inode) != BTRFS_FREE_INO_OBJECTID)
317 check_crcs = 1;
318
319 /* Make sure we can fit our crcs into the first page */
320 if (write && check_crcs &&
321 (num_pages * sizeof(u32)) >= PAGE_SIZE)
322 return -ENOSPC;
323
324 memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
325
326 io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
327 if (!io_ctl->pages)
328 return -ENOMEM;
329
330 io_ctl->num_pages = num_pages;
331 io_ctl->fs_info = btrfs_sb(inode->i_sb);
332 io_ctl->check_crcs = check_crcs;
333 io_ctl->inode = inode;
334
335 return 0;
336}
337
338static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
339{
340 kfree(io_ctl->pages);
341 io_ctl->pages = NULL;
342}
343
344static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
345{
346 if (io_ctl->cur) {
347 io_ctl->cur = NULL;
348 io_ctl->orig = NULL;
349 }
350}
351
352static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
353{
354 ASSERT(io_ctl->index < io_ctl->num_pages);
355 io_ctl->page = io_ctl->pages[io_ctl->index++];
356 io_ctl->cur = page_address(io_ctl->page);
357 io_ctl->orig = io_ctl->cur;
358 io_ctl->size = PAGE_SIZE;
359 if (clear)
360 memset(io_ctl->cur, 0, PAGE_SIZE);
361}
362
363static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
364{
365 int i;
366
367 io_ctl_unmap_page(io_ctl);
368
369 for (i = 0; i < io_ctl->num_pages; i++) {
370 if (io_ctl->pages[i]) {
371 ClearPageChecked(io_ctl->pages[i]);
372 unlock_page(io_ctl->pages[i]);
373 put_page(io_ctl->pages[i]);
374 }
375 }
376}
377
378static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, struct inode *inode,
379 int uptodate)
380{
381 struct page *page;
382 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
383 int i;
384
385 for (i = 0; i < io_ctl->num_pages; i++) {
386 page = find_or_create_page(inode->i_mapping, i, mask);
387 if (!page) {
388 io_ctl_drop_pages(io_ctl);
389 return -ENOMEM;
390 }
391 io_ctl->pages[i] = page;
392 if (uptodate && !PageUptodate(page)) {
393 btrfs_readpage(NULL, page);
394 lock_page(page);
395 if (!PageUptodate(page)) {
396 btrfs_err(BTRFS_I(inode)->root->fs_info,
397 "error reading free space cache");
398 io_ctl_drop_pages(io_ctl);
399 return -EIO;
400 }
401 }
402 }
403
404 for (i = 0; i < io_ctl->num_pages; i++) {
405 clear_page_dirty_for_io(io_ctl->pages[i]);
406 set_page_extent_mapped(io_ctl->pages[i]);
407 }
408
409 return 0;
410}
411
412static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
413{
414 __le64 *val;
415
416 io_ctl_map_page(io_ctl, 1);
417
418 /*
419 * Skip the csum areas. If we don't check crcs then we just have a
420 * 64bit chunk at the front of the first page.
421 */
422 if (io_ctl->check_crcs) {
423 io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
424 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
425 } else {
426 io_ctl->cur += sizeof(u64);
427 io_ctl->size -= sizeof(u64) * 2;
428 }
429
430 val = io_ctl->cur;
431 *val = cpu_to_le64(generation);
432 io_ctl->cur += sizeof(u64);
433}
434
435static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
436{
437 __le64 *gen;
438
439 /*
440 * Skip the crc area. If we don't check crcs then we just have a 64bit
441 * chunk at the front of the first page.
442 */
443 if (io_ctl->check_crcs) {
444 io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
445 io_ctl->size -= sizeof(u64) +
446 (sizeof(u32) * io_ctl->num_pages);
447 } else {
448 io_ctl->cur += sizeof(u64);
449 io_ctl->size -= sizeof(u64) * 2;
450 }
451
452 gen = io_ctl->cur;
453 if (le64_to_cpu(*gen) != generation) {
454 btrfs_err_rl(io_ctl->fs_info,
455 "space cache generation (%llu) does not match inode (%llu)",
456 *gen, generation);
457 io_ctl_unmap_page(io_ctl);
458 return -EIO;
459 }
460 io_ctl->cur += sizeof(u64);
461 return 0;
462}
463
464static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
465{
466 u32 *tmp;
467 u32 crc = ~(u32)0;
468 unsigned offset = 0;
469
470 if (!io_ctl->check_crcs) {
471 io_ctl_unmap_page(io_ctl);
472 return;
473 }
474
475 if (index == 0)
476 offset = sizeof(u32) * io_ctl->num_pages;
477
478 crc = btrfs_csum_data(io_ctl->orig + offset, crc,
479 PAGE_SIZE - offset);
480 btrfs_csum_final(crc, (u8 *)&crc);
481 io_ctl_unmap_page(io_ctl);
482 tmp = page_address(io_ctl->pages[0]);
483 tmp += index;
484 *tmp = crc;
485}
486
487static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
488{
489 u32 *tmp, val;
490 u32 crc = ~(u32)0;
491 unsigned offset = 0;
492
493 if (!io_ctl->check_crcs) {
494 io_ctl_map_page(io_ctl, 0);
495 return 0;
496 }
497
498 if (index == 0)
499 offset = sizeof(u32) * io_ctl->num_pages;
500
501 tmp = page_address(io_ctl->pages[0]);
502 tmp += index;
503 val = *tmp;
504
505 io_ctl_map_page(io_ctl, 0);
506 crc = btrfs_csum_data(io_ctl->orig + offset, crc,
507 PAGE_SIZE - offset);
508 btrfs_csum_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
519static 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
549static 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 memcpy(io_ctl->cur, bitmap, PAGE_SIZE);
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
572static 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
589static 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
616static 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 memcpy(entry->bitmap, io_ctl->cur, PAGE_SIZE);
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 */
640static 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
645again:
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 }
663next:
664 prev = e;
665 }
666 spin_unlock(&ctl->tree_lock);
667}
668
669static 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 = btrfs_sb(inode->i_sb);
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\n",
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 ret = readahead_cache(inode);
734 if (ret)
735 goto out;
736
737 ret = io_ctl_prepare_pages(&io_ctl, inode, 1);
738 if (ret)
739 goto out;
740
741 ret = io_ctl_check_crc(&io_ctl, 0);
742 if (ret)
743 goto free_cache;
744
745 ret = io_ctl_check_generation(&io_ctl, generation);
746 if (ret)
747 goto free_cache;
748
749 while (num_entries) {
750 e = kmem_cache_zalloc(btrfs_free_space_cachep,
751 GFP_NOFS);
752 if (!e)
753 goto free_cache;
754
755 ret = io_ctl_read_entry(&io_ctl, e, &type);
756 if (ret) {
757 kmem_cache_free(btrfs_free_space_cachep, e);
758 goto free_cache;
759 }
760
761 if (!e->bytes) {
762 kmem_cache_free(btrfs_free_space_cachep, e);
763 goto free_cache;
764 }
765
766 if (type == BTRFS_FREE_SPACE_EXTENT) {
767 spin_lock(&ctl->tree_lock);
768 ret = link_free_space(ctl, e);
769 spin_unlock(&ctl->tree_lock);
770 if (ret) {
771 btrfs_err(fs_info,
772 "Duplicate entries in free space cache, dumping");
773 kmem_cache_free(btrfs_free_space_cachep, e);
774 goto free_cache;
775 }
776 } else {
777 ASSERT(num_bitmaps);
778 num_bitmaps--;
779 e->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
780 if (!e->bitmap) {
781 kmem_cache_free(
782 btrfs_free_space_cachep, e);
783 goto free_cache;
784 }
785 spin_lock(&ctl->tree_lock);
786 ret = link_free_space(ctl, e);
787 ctl->total_bitmaps++;
788 ctl->op->recalc_thresholds(ctl);
789 spin_unlock(&ctl->tree_lock);
790 if (ret) {
791 btrfs_err(fs_info,
792 "Duplicate entries in free space cache, dumping");
793 kmem_cache_free(btrfs_free_space_cachep, e);
794 goto free_cache;
795 }
796 list_add_tail(&e->list, &bitmaps);
797 }
798
799 num_entries--;
800 }
801
802 io_ctl_unmap_page(&io_ctl);
803
804 /*
805 * We add the bitmaps at the end of the entries in order that
806 * the bitmap entries are added to the cache.
807 */
808 list_for_each_entry_safe(e, n, &bitmaps, list) {
809 list_del_init(&e->list);
810 ret = io_ctl_read_bitmap(&io_ctl, e);
811 if (ret)
812 goto free_cache;
813 }
814
815 io_ctl_drop_pages(&io_ctl);
816 merge_space_tree(ctl);
817 ret = 1;
818out:
819 io_ctl_free(&io_ctl);
820 return ret;
821free_cache:
822 io_ctl_drop_pages(&io_ctl);
823 __btrfs_remove_free_space_cache(ctl);
824 goto out;
825}
826
827int load_free_space_cache(struct btrfs_fs_info *fs_info,
828 struct btrfs_block_group_cache *block_group)
829{
830 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
831 struct btrfs_root *root = fs_info->tree_root;
832 struct inode *inode;
833 struct btrfs_path *path;
834 int ret = 0;
835 bool matched;
836 u64 used = btrfs_block_group_used(&block_group->item);
837
838 /*
839 * If this block group has been marked to be cleared for one reason or
840 * another then we can't trust the on disk cache, so just return.
841 */
842 spin_lock(&block_group->lock);
843 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
844 spin_unlock(&block_group->lock);
845 return 0;
846 }
847 spin_unlock(&block_group->lock);
848
849 path = btrfs_alloc_path();
850 if (!path)
851 return 0;
852 path->search_commit_root = 1;
853 path->skip_locking = 1;
854
855 inode = lookup_free_space_inode(root, block_group, path);
856 if (IS_ERR(inode)) {
857 btrfs_free_path(path);
858 return 0;
859 }
860
861 /* We may have converted the inode and made the cache invalid. */
862 spin_lock(&block_group->lock);
863 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
864 spin_unlock(&block_group->lock);
865 btrfs_free_path(path);
866 goto out;
867 }
868 spin_unlock(&block_group->lock);
869
870 ret = __load_free_space_cache(fs_info->tree_root, inode, ctl,
871 path, block_group->key.objectid);
872 btrfs_free_path(path);
873 if (ret <= 0)
874 goto out;
875
876 spin_lock(&ctl->tree_lock);
877 matched = (ctl->free_space == (block_group->key.offset - used -
878 block_group->bytes_super));
879 spin_unlock(&ctl->tree_lock);
880
881 if (!matched) {
882 __btrfs_remove_free_space_cache(ctl);
883 btrfs_warn(fs_info,
884 "block group %llu has wrong amount of free space",
885 block_group->key.objectid);
886 ret = -1;
887 }
888out:
889 if (ret < 0) {
890 /* This cache is bogus, make sure it gets cleared */
891 spin_lock(&block_group->lock);
892 block_group->disk_cache_state = BTRFS_DC_CLEAR;
893 spin_unlock(&block_group->lock);
894 ret = 0;
895
896 btrfs_warn(fs_info,
897 "failed to load free space cache for block group %llu, rebuilding it now",
898 block_group->key.objectid);
899 }
900
901 iput(inode);
902 return ret;
903}
904
905static noinline_for_stack
906int write_cache_extent_entries(struct btrfs_io_ctl *io_ctl,
907 struct btrfs_free_space_ctl *ctl,
908 struct btrfs_block_group_cache *block_group,
909 int *entries, int *bitmaps,
910 struct list_head *bitmap_list)
911{
912 int ret;
913 struct btrfs_free_cluster *cluster = NULL;
914 struct btrfs_free_cluster *cluster_locked = NULL;
915 struct rb_node *node = rb_first(&ctl->free_space_offset);
916 struct btrfs_trim_range *trim_entry;
917
918 /* Get the cluster for this block_group if it exists */
919 if (block_group && !list_empty(&block_group->cluster_list)) {
920 cluster = list_entry(block_group->cluster_list.next,
921 struct btrfs_free_cluster,
922 block_group_list);
923 }
924
925 if (!node && cluster) {
926 cluster_locked = cluster;
927 spin_lock(&cluster_locked->lock);
928 node = rb_first(&cluster->root);
929 cluster = NULL;
930 }
931
932 /* Write out the extent entries */
933 while (node) {
934 struct btrfs_free_space *e;
935
936 e = rb_entry(node, struct btrfs_free_space, offset_index);
937 *entries += 1;
938
939 ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes,
940 e->bitmap);
941 if (ret)
942 goto fail;
943
944 if (e->bitmap) {
945 list_add_tail(&e->list, bitmap_list);
946 *bitmaps += 1;
947 }
948 node = rb_next(node);
949 if (!node && cluster) {
950 node = rb_first(&cluster->root);
951 cluster_locked = cluster;
952 spin_lock(&cluster_locked->lock);
953 cluster = NULL;
954 }
955 }
956 if (cluster_locked) {
957 spin_unlock(&cluster_locked->lock);
958 cluster_locked = NULL;
959 }
960
961 /*
962 * Make sure we don't miss any range that was removed from our rbtree
963 * because trimming is running. Otherwise after a umount+mount (or crash
964 * after committing the transaction) we would leak free space and get
965 * an inconsistent free space cache report from fsck.
966 */
967 list_for_each_entry(trim_entry, &ctl->trimming_ranges, list) {
968 ret = io_ctl_add_entry(io_ctl, trim_entry->start,
969 trim_entry->bytes, NULL);
970 if (ret)
971 goto fail;
972 *entries += 1;
973 }
974
975 return 0;
976fail:
977 if (cluster_locked)
978 spin_unlock(&cluster_locked->lock);
979 return -ENOSPC;
980}
981
982static noinline_for_stack int
983update_cache_item(struct btrfs_trans_handle *trans,
984 struct btrfs_root *root,
985 struct inode *inode,
986 struct btrfs_path *path, u64 offset,
987 int entries, int bitmaps)
988{
989 struct btrfs_key key;
990 struct btrfs_free_space_header *header;
991 struct extent_buffer *leaf;
992 int ret;
993
994 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
995 key.offset = offset;
996 key.type = 0;
997
998 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
999 if (ret < 0) {
1000 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1001 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
1002 GFP_NOFS);
1003 goto fail;
1004 }
1005 leaf = path->nodes[0];
1006 if (ret > 0) {
1007 struct btrfs_key found_key;
1008 ASSERT(path->slots[0]);
1009 path->slots[0]--;
1010 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1011 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
1012 found_key.offset != offset) {
1013 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
1014 inode->i_size - 1,
1015 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0,
1016 NULL, GFP_NOFS);
1017 btrfs_release_path(path);
1018 goto fail;
1019 }
1020 }
1021
1022 BTRFS_I(inode)->generation = trans->transid;
1023 header = btrfs_item_ptr(leaf, path->slots[0],
1024 struct btrfs_free_space_header);
1025 btrfs_set_free_space_entries(leaf, header, entries);
1026 btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1027 btrfs_set_free_space_generation(leaf, header, trans->transid);
1028 btrfs_mark_buffer_dirty(leaf);
1029 btrfs_release_path(path);
1030
1031 return 0;
1032
1033fail:
1034 return -1;
1035}
1036
1037static noinline_for_stack int
1038write_pinned_extent_entries(struct btrfs_fs_info *fs_info,
1039 struct btrfs_block_group_cache *block_group,
1040 struct btrfs_io_ctl *io_ctl,
1041 int *entries)
1042{
1043 u64 start, extent_start, extent_end, len;
1044 struct extent_io_tree *unpin = NULL;
1045 int ret;
1046
1047 if (!block_group)
1048 return 0;
1049
1050 /*
1051 * We want to add any pinned extents to our free space cache
1052 * so we don't leak the space
1053 *
1054 * We shouldn't have switched the pinned extents yet so this is the
1055 * right one
1056 */
1057 unpin = fs_info->pinned_extents;
1058
1059 start = block_group->key.objectid;
1060
1061 while (start < block_group->key.objectid + block_group->key.offset) {
1062 ret = find_first_extent_bit(unpin, start,
1063 &extent_start, &extent_end,
1064 EXTENT_DIRTY, NULL);
1065 if (ret)
1066 return 0;
1067
1068 /* This pinned extent is out of our range */
1069 if (extent_start >= block_group->key.objectid +
1070 block_group->key.offset)
1071 return 0;
1072
1073 extent_start = max(extent_start, start);
1074 extent_end = min(block_group->key.objectid +
1075 block_group->key.offset, extent_end + 1);
1076 len = extent_end - extent_start;
1077
1078 *entries += 1;
1079 ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
1080 if (ret)
1081 return -ENOSPC;
1082
1083 start = extent_end;
1084 }
1085
1086 return 0;
1087}
1088
1089static noinline_for_stack int
1090write_bitmap_entries(struct btrfs_io_ctl *io_ctl, struct list_head *bitmap_list)
1091{
1092 struct btrfs_free_space *entry, *next;
1093 int ret;
1094
1095 /* Write out the bitmaps */
1096 list_for_each_entry_safe(entry, next, bitmap_list, list) {
1097 ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
1098 if (ret)
1099 return -ENOSPC;
1100 list_del_init(&entry->list);
1101 }
1102
1103 return 0;
1104}
1105
1106static int flush_dirty_cache(struct inode *inode)
1107{
1108 int ret;
1109
1110 ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
1111 if (ret)
1112 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1113 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
1114 GFP_NOFS);
1115
1116 return ret;
1117}
1118
1119static void noinline_for_stack
1120cleanup_bitmap_list(struct list_head *bitmap_list)
1121{
1122 struct btrfs_free_space *entry, *next;
1123
1124 list_for_each_entry_safe(entry, next, bitmap_list, list)
1125 list_del_init(&entry->list);
1126}
1127
1128static void noinline_for_stack
1129cleanup_write_cache_enospc(struct inode *inode,
1130 struct btrfs_io_ctl *io_ctl,
1131 struct extent_state **cached_state,
1132 struct list_head *bitmap_list)
1133{
1134 io_ctl_drop_pages(io_ctl);
1135 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1136 i_size_read(inode) - 1, cached_state,
1137 GFP_NOFS);
1138}
1139
1140static int __btrfs_wait_cache_io(struct btrfs_root *root,
1141 struct btrfs_trans_handle *trans,
1142 struct btrfs_block_group_cache *block_group,
1143 struct btrfs_io_ctl *io_ctl,
1144 struct btrfs_path *path, u64 offset)
1145{
1146 int ret;
1147 struct inode *inode = io_ctl->inode;
1148 struct btrfs_fs_info *fs_info;
1149
1150 if (!inode)
1151 return 0;
1152
1153 fs_info = btrfs_sb(inode->i_sb);
1154
1155 /* Flush the dirty pages in the cache file. */
1156 ret = flush_dirty_cache(inode);
1157 if (ret)
1158 goto out;
1159
1160 /* Update the cache item to tell everyone this cache file is valid. */
1161 ret = update_cache_item(trans, root, inode, path, offset,
1162 io_ctl->entries, io_ctl->bitmaps);
1163out:
1164 io_ctl_free(io_ctl);
1165 if (ret) {
1166 invalidate_inode_pages2(inode->i_mapping);
1167 BTRFS_I(inode)->generation = 0;
1168 if (block_group) {
1169#ifdef DEBUG
1170 btrfs_err(fs_info,
1171 "failed to write free space cache for block group %llu",
1172 block_group->key.objectid);
1173#endif
1174 }
1175 }
1176 btrfs_update_inode(trans, root, inode);
1177
1178 if (block_group) {
1179 /* the dirty list is protected by the dirty_bgs_lock */
1180 spin_lock(&trans->transaction->dirty_bgs_lock);
1181
1182 /* the disk_cache_state is protected by the block group lock */
1183 spin_lock(&block_group->lock);
1184
1185 /*
1186 * only mark this as written if we didn't get put back on
1187 * the dirty list while waiting for IO. Otherwise our
1188 * cache state won't be right, and we won't get written again
1189 */
1190 if (!ret && list_empty(&block_group->dirty_list))
1191 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1192 else if (ret)
1193 block_group->disk_cache_state = BTRFS_DC_ERROR;
1194
1195 spin_unlock(&block_group->lock);
1196 spin_unlock(&trans->transaction->dirty_bgs_lock);
1197 io_ctl->inode = NULL;
1198 iput(inode);
1199 }
1200
1201 return ret;
1202
1203}
1204
1205static int btrfs_wait_cache_io_root(struct btrfs_root *root,
1206 struct btrfs_trans_handle *trans,
1207 struct btrfs_io_ctl *io_ctl,
1208 struct btrfs_path *path)
1209{
1210 return __btrfs_wait_cache_io(root, trans, NULL, io_ctl, path, 0);
1211}
1212
1213int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
1214 struct btrfs_block_group_cache *block_group,
1215 struct btrfs_path *path)
1216{
1217 return __btrfs_wait_cache_io(block_group->fs_info->tree_root, trans,
1218 block_group, &block_group->io_ctl,
1219 path, block_group->key.objectid);
1220}
1221
1222/**
1223 * __btrfs_write_out_cache - write out cached info to an inode
1224 * @root - the root the inode belongs to
1225 * @ctl - the free space cache we are going to write out
1226 * @block_group - the block_group for this cache if it belongs to a block_group
1227 * @trans - the trans handle
1228 * @path - the path to use
1229 * @offset - the offset for the key we'll insert
1230 *
1231 * This function writes out a free space cache struct to disk for quick recovery
1232 * on mount. This will return 0 if it was successful in writing the cache out,
1233 * or an errno if it was not.
1234 */
1235static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
1236 struct btrfs_free_space_ctl *ctl,
1237 struct btrfs_block_group_cache *block_group,
1238 struct btrfs_io_ctl *io_ctl,
1239 struct btrfs_trans_handle *trans,
1240 struct btrfs_path *path, u64 offset)
1241{
1242 struct btrfs_fs_info *fs_info = root->fs_info;
1243 struct extent_state *cached_state = NULL;
1244 LIST_HEAD(bitmap_list);
1245 int entries = 0;
1246 int bitmaps = 0;
1247 int ret;
1248 int must_iput = 0;
1249
1250 if (!i_size_read(inode))
1251 return -EIO;
1252
1253 WARN_ON(io_ctl->pages);
1254 ret = io_ctl_init(io_ctl, inode, 1);
1255 if (ret)
1256 return ret;
1257
1258 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
1259 down_write(&block_group->data_rwsem);
1260 spin_lock(&block_group->lock);
1261 if (block_group->delalloc_bytes) {
1262 block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1263 spin_unlock(&block_group->lock);
1264 up_write(&block_group->data_rwsem);
1265 BTRFS_I(inode)->generation = 0;
1266 ret = 0;
1267 must_iput = 1;
1268 goto out;
1269 }
1270 spin_unlock(&block_group->lock);
1271 }
1272
1273 /* Lock all pages first so we can lock the extent safely. */
1274 ret = io_ctl_prepare_pages(io_ctl, inode, 0);
1275 if (ret)
1276 goto out;
1277
1278 lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1279 &cached_state);
1280
1281 io_ctl_set_generation(io_ctl, trans->transid);
1282
1283 mutex_lock(&ctl->cache_writeout_mutex);
1284 /* Write out the extent entries in the free space cache */
1285 spin_lock(&ctl->tree_lock);
1286 ret = write_cache_extent_entries(io_ctl, ctl,
1287 block_group, &entries, &bitmaps,
1288 &bitmap_list);
1289 if (ret)
1290 goto out_nospc_locked;
1291
1292 /*
1293 * Some spaces that are freed in the current transaction are pinned,
1294 * they will be added into free space cache after the transaction is
1295 * committed, we shouldn't lose them.
1296 *
1297 * If this changes while we are working we'll get added back to
1298 * the dirty list and redo it. No locking needed
1299 */
1300 ret = write_pinned_extent_entries(fs_info, block_group,
1301 io_ctl, &entries);
1302 if (ret)
1303 goto out_nospc_locked;
1304
1305 /*
1306 * At last, we write out all the bitmaps and keep cache_writeout_mutex
1307 * locked while doing it because a concurrent trim can be manipulating
1308 * or freeing the bitmap.
1309 */
1310 ret = write_bitmap_entries(io_ctl, &bitmap_list);
1311 spin_unlock(&ctl->tree_lock);
1312 mutex_unlock(&ctl->cache_writeout_mutex);
1313 if (ret)
1314 goto out_nospc;
1315
1316 /* Zero out the rest of the pages just to make sure */
1317 io_ctl_zero_remaining_pages(io_ctl);
1318
1319 /* Everything is written out, now we dirty the pages in the file. */
1320 ret = btrfs_dirty_pages(inode, io_ctl->pages, io_ctl->num_pages, 0,
1321 i_size_read(inode), &cached_state);
1322 if (ret)
1323 goto out_nospc;
1324
1325 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1326 up_write(&block_group->data_rwsem);
1327 /*
1328 * Release the pages and unlock the extent, we will flush
1329 * them out later
1330 */
1331 io_ctl_drop_pages(io_ctl);
1332
1333 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1334 i_size_read(inode) - 1, &cached_state, GFP_NOFS);
1335
1336 /*
1337 * at this point the pages are under IO and we're happy,
1338 * The caller is responsible for waiting on them and updating the
1339 * the cache and the inode
1340 */
1341 io_ctl->entries = entries;
1342 io_ctl->bitmaps = bitmaps;
1343
1344 ret = btrfs_fdatawrite_range(inode, 0, (u64)-1);
1345 if (ret)
1346 goto out;
1347
1348 return 0;
1349
1350out:
1351 io_ctl->inode = NULL;
1352 io_ctl_free(io_ctl);
1353 if (ret) {
1354 invalidate_inode_pages2(inode->i_mapping);
1355 BTRFS_I(inode)->generation = 0;
1356 }
1357 btrfs_update_inode(trans, root, inode);
1358 if (must_iput)
1359 iput(inode);
1360 return ret;
1361
1362out_nospc_locked:
1363 cleanup_bitmap_list(&bitmap_list);
1364 spin_unlock(&ctl->tree_lock);
1365 mutex_unlock(&ctl->cache_writeout_mutex);
1366
1367out_nospc:
1368 cleanup_write_cache_enospc(inode, io_ctl, &cached_state, &bitmap_list);
1369
1370 if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1371 up_write(&block_group->data_rwsem);
1372
1373 goto out;
1374}
1375
1376int btrfs_write_out_cache(struct btrfs_fs_info *fs_info,
1377 struct btrfs_trans_handle *trans,
1378 struct btrfs_block_group_cache *block_group,
1379 struct btrfs_path *path)
1380{
1381 struct btrfs_root *root = fs_info->tree_root;
1382 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1383 struct inode *inode;
1384 int ret = 0;
1385
1386 spin_lock(&block_group->lock);
1387 if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1388 spin_unlock(&block_group->lock);
1389 return 0;
1390 }
1391 spin_unlock(&block_group->lock);
1392
1393 inode = lookup_free_space_inode(root, block_group, path);
1394 if (IS_ERR(inode))
1395 return 0;
1396
1397 ret = __btrfs_write_out_cache(root, inode, ctl, block_group,
1398 &block_group->io_ctl, trans,
1399 path, block_group->key.objectid);
1400 if (ret) {
1401#ifdef DEBUG
1402 btrfs_err(fs_info,
1403 "failed to write free space cache for block group %llu",
1404 block_group->key.objectid);
1405#endif
1406 spin_lock(&block_group->lock);
1407 block_group->disk_cache_state = BTRFS_DC_ERROR;
1408 spin_unlock(&block_group->lock);
1409
1410 block_group->io_ctl.inode = NULL;
1411 iput(inode);
1412 }
1413
1414 /*
1415 * if ret == 0 the caller is expected to call btrfs_wait_cache_io
1416 * to wait for IO and put the inode
1417 */
1418
1419 return ret;
1420}
1421
1422static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1423 u64 offset)
1424{
1425 ASSERT(offset >= bitmap_start);
1426 offset -= bitmap_start;
1427 return (unsigned long)(div_u64(offset, unit));
1428}
1429
1430static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1431{
1432 return (unsigned long)(div_u64(bytes, unit));
1433}
1434
1435static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1436 u64 offset)
1437{
1438 u64 bitmap_start;
1439 u64 bytes_per_bitmap;
1440
1441 bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1442 bitmap_start = offset - ctl->start;
1443 bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1444 bitmap_start *= bytes_per_bitmap;
1445 bitmap_start += ctl->start;
1446
1447 return bitmap_start;
1448}
1449
1450static int tree_insert_offset(struct rb_root *root, u64 offset,
1451 struct rb_node *node, int bitmap)
1452{
1453 struct rb_node **p = &root->rb_node;
1454 struct rb_node *parent = NULL;
1455 struct btrfs_free_space *info;
1456
1457 while (*p) {
1458 parent = *p;
1459 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1460
1461 if (offset < info->offset) {
1462 p = &(*p)->rb_left;
1463 } else if (offset > info->offset) {
1464 p = &(*p)->rb_right;
1465 } else {
1466 /*
1467 * we could have a bitmap entry and an extent entry
1468 * share the same offset. If this is the case, we want
1469 * the extent entry to always be found first if we do a
1470 * linear search through the tree, since we want to have
1471 * the quickest allocation time, and allocating from an
1472 * extent is faster than allocating from a bitmap. So
1473 * if we're inserting a bitmap and we find an entry at
1474 * this offset, we want to go right, or after this entry
1475 * logically. If we are inserting an extent and we've
1476 * found a bitmap, we want to go left, or before
1477 * logically.
1478 */
1479 if (bitmap) {
1480 if (info->bitmap) {
1481 WARN_ON_ONCE(1);
1482 return -EEXIST;
1483 }
1484 p = &(*p)->rb_right;
1485 } else {
1486 if (!info->bitmap) {
1487 WARN_ON_ONCE(1);
1488 return -EEXIST;
1489 }
1490 p = &(*p)->rb_left;
1491 }
1492 }
1493 }
1494
1495 rb_link_node(node, parent, p);
1496 rb_insert_color(node, root);
1497
1498 return 0;
1499}
1500
1501/*
1502 * searches the tree for the given offset.
1503 *
1504 * fuzzy - If this is set, then we are trying to make an allocation, and we just
1505 * want a section that has at least bytes size and comes at or after the given
1506 * offset.
1507 */
1508static struct btrfs_free_space *
1509tree_search_offset(struct btrfs_free_space_ctl *ctl,
1510 u64 offset, int bitmap_only, int fuzzy)
1511{
1512 struct rb_node *n = ctl->free_space_offset.rb_node;
1513 struct btrfs_free_space *entry, *prev = NULL;
1514
1515 /* find entry that is closest to the 'offset' */
1516 while (1) {
1517 if (!n) {
1518 entry = NULL;
1519 break;
1520 }
1521
1522 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1523 prev = entry;
1524
1525 if (offset < entry->offset)
1526 n = n->rb_left;
1527 else if (offset > entry->offset)
1528 n = n->rb_right;
1529 else
1530 break;
1531 }
1532
1533 if (bitmap_only) {
1534 if (!entry)
1535 return NULL;
1536 if (entry->bitmap)
1537 return entry;
1538
1539 /*
1540 * bitmap entry and extent entry may share same offset,
1541 * in that case, bitmap entry comes after extent entry.
1542 */
1543 n = rb_next(n);
1544 if (!n)
1545 return NULL;
1546 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1547 if (entry->offset != offset)
1548 return NULL;
1549
1550 WARN_ON(!entry->bitmap);
1551 return entry;
1552 } else if (entry) {
1553 if (entry->bitmap) {
1554 /*
1555 * if previous extent entry covers the offset,
1556 * we should return it instead of the bitmap entry
1557 */
1558 n = rb_prev(&entry->offset_index);
1559 if (n) {
1560 prev = rb_entry(n, struct btrfs_free_space,
1561 offset_index);
1562 if (!prev->bitmap &&
1563 prev->offset + prev->bytes > offset)
1564 entry = prev;
1565 }
1566 }
1567 return entry;
1568 }
1569
1570 if (!prev)
1571 return NULL;
1572
1573 /* find last entry before the 'offset' */
1574 entry = prev;
1575 if (entry->offset > offset) {
1576 n = rb_prev(&entry->offset_index);
1577 if (n) {
1578 entry = rb_entry(n, struct btrfs_free_space,
1579 offset_index);
1580 ASSERT(entry->offset <= offset);
1581 } else {
1582 if (fuzzy)
1583 return entry;
1584 else
1585 return NULL;
1586 }
1587 }
1588
1589 if (entry->bitmap) {
1590 n = rb_prev(&entry->offset_index);
1591 if (n) {
1592 prev = rb_entry(n, struct btrfs_free_space,
1593 offset_index);
1594 if (!prev->bitmap &&
1595 prev->offset + prev->bytes > offset)
1596 return prev;
1597 }
1598 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1599 return entry;
1600 } else if (entry->offset + entry->bytes > offset)
1601 return entry;
1602
1603 if (!fuzzy)
1604 return NULL;
1605
1606 while (1) {
1607 if (entry->bitmap) {
1608 if (entry->offset + BITS_PER_BITMAP *
1609 ctl->unit > offset)
1610 break;
1611 } else {
1612 if (entry->offset + entry->bytes > offset)
1613 break;
1614 }
1615
1616 n = rb_next(&entry->offset_index);
1617 if (!n)
1618 return NULL;
1619 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1620 }
1621 return entry;
1622}
1623
1624static inline void
1625__unlink_free_space(struct btrfs_free_space_ctl *ctl,
1626 struct btrfs_free_space *info)
1627{
1628 rb_erase(&info->offset_index, &ctl->free_space_offset);
1629 ctl->free_extents--;
1630}
1631
1632static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1633 struct btrfs_free_space *info)
1634{
1635 __unlink_free_space(ctl, info);
1636 ctl->free_space -= info->bytes;
1637}
1638
1639static int link_free_space(struct btrfs_free_space_ctl *ctl,
1640 struct btrfs_free_space *info)
1641{
1642 int ret = 0;
1643
1644 ASSERT(info->bytes || info->bitmap);
1645 ret = tree_insert_offset(&ctl->free_space_offset, info->offset,
1646 &info->offset_index, (info->bitmap != NULL));
1647 if (ret)
1648 return ret;
1649
1650 ctl->free_space += info->bytes;
1651 ctl->free_extents++;
1652 return ret;
1653}
1654
1655static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
1656{
1657 struct btrfs_block_group_cache *block_group = ctl->private;
1658 u64 max_bytes;
1659 u64 bitmap_bytes;
1660 u64 extent_bytes;
1661 u64 size = block_group->key.offset;
1662 u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
1663 u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
1664
1665 max_bitmaps = max_t(u64, max_bitmaps, 1);
1666
1667 ASSERT(ctl->total_bitmaps <= max_bitmaps);
1668
1669 /*
1670 * The goal is to keep the total amount of memory used per 1gb of space
1671 * at or below 32k, so we need to adjust how much memory we allow to be
1672 * used by extent based free space tracking
1673 */
1674 if (size < SZ_1G)
1675 max_bytes = MAX_CACHE_BYTES_PER_GIG;
1676 else
1677 max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
1678
1679 /*
1680 * we want to account for 1 more bitmap than what we have so we can make
1681 * sure we don't go over our overall goal of MAX_CACHE_BYTES_PER_GIG as
1682 * we add more bitmaps.
1683 */
1684 bitmap_bytes = (ctl->total_bitmaps + 1) * ctl->unit;
1685
1686 if (bitmap_bytes >= max_bytes) {
1687 ctl->extents_thresh = 0;
1688 return;
1689 }
1690
1691 /*
1692 * we want the extent entry threshold to always be at most 1/2 the max
1693 * bytes we can have, or whatever is less than that.
1694 */
1695 extent_bytes = max_bytes - bitmap_bytes;
1696 extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
1697
1698 ctl->extents_thresh =
1699 div_u64(extent_bytes, sizeof(struct btrfs_free_space));
1700}
1701
1702static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1703 struct btrfs_free_space *info,
1704 u64 offset, u64 bytes)
1705{
1706 unsigned long start, count;
1707
1708 start = offset_to_bit(info->offset, ctl->unit, offset);
1709 count = bytes_to_bits(bytes, ctl->unit);
1710 ASSERT(start + count <= BITS_PER_BITMAP);
1711
1712 bitmap_clear(info->bitmap, start, count);
1713
1714 info->bytes -= bytes;
1715}
1716
1717static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1718 struct btrfs_free_space *info, u64 offset,
1719 u64 bytes)
1720{
1721 __bitmap_clear_bits(ctl, info, offset, bytes);
1722 ctl->free_space -= bytes;
1723}
1724
1725static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1726 struct btrfs_free_space *info, u64 offset,
1727 u64 bytes)
1728{
1729 unsigned long start, count;
1730
1731 start = offset_to_bit(info->offset, ctl->unit, offset);
1732 count = bytes_to_bits(bytes, ctl->unit);
1733 ASSERT(start + count <= BITS_PER_BITMAP);
1734
1735 bitmap_set(info->bitmap, start, count);
1736
1737 info->bytes += bytes;
1738 ctl->free_space += bytes;
1739}
1740
1741/*
1742 * If we can not find suitable extent, we will use bytes to record
1743 * the size of the max extent.
1744 */
1745static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1746 struct btrfs_free_space *bitmap_info, u64 *offset,
1747 u64 *bytes, bool for_alloc)
1748{
1749 unsigned long found_bits = 0;
1750 unsigned long max_bits = 0;
1751 unsigned long bits, i;
1752 unsigned long next_zero;
1753 unsigned long extent_bits;
1754
1755 /*
1756 * Skip searching the bitmap if we don't have a contiguous section that
1757 * is large enough for this allocation.
1758 */
1759 if (for_alloc &&
1760 bitmap_info->max_extent_size &&
1761 bitmap_info->max_extent_size < *bytes) {
1762 *bytes = bitmap_info->max_extent_size;
1763 return -1;
1764 }
1765
1766 i = offset_to_bit(bitmap_info->offset, ctl->unit,
1767 max_t(u64, *offset, bitmap_info->offset));
1768 bits = bytes_to_bits(*bytes, ctl->unit);
1769
1770 for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
1771 if (for_alloc && bits == 1) {
1772 found_bits = 1;
1773 break;
1774 }
1775 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1776 BITS_PER_BITMAP, i);
1777 extent_bits = next_zero - i;
1778 if (extent_bits >= bits) {
1779 found_bits = extent_bits;
1780 break;
1781 } else if (extent_bits > max_bits) {
1782 max_bits = extent_bits;
1783 }
1784 i = next_zero;
1785 }
1786
1787 if (found_bits) {
1788 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1789 *bytes = (u64)(found_bits) * ctl->unit;
1790 return 0;
1791 }
1792
1793 *bytes = (u64)(max_bits) * ctl->unit;
1794 bitmap_info->max_extent_size = *bytes;
1795 return -1;
1796}
1797
1798/* Cache the size of the max extent in bytes */
1799static struct btrfs_free_space *
1800find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
1801 unsigned long align, u64 *max_extent_size)
1802{
1803 struct btrfs_free_space *entry;
1804 struct rb_node *node;
1805 u64 tmp;
1806 u64 align_off;
1807 int ret;
1808
1809 if (!ctl->free_space_offset.rb_node)
1810 goto out;
1811
1812 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
1813 if (!entry)
1814 goto out;
1815
1816 for (node = &entry->offset_index; node; node = rb_next(node)) {
1817 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1818 if (entry->bytes < *bytes) {
1819 if (entry->bytes > *max_extent_size)
1820 *max_extent_size = entry->bytes;
1821 continue;
1822 }
1823
1824 /* make sure the space returned is big enough
1825 * to match our requested alignment
1826 */
1827 if (*bytes >= align) {
1828 tmp = entry->offset - ctl->start + align - 1;
1829 tmp = div64_u64(tmp, align);
1830 tmp = tmp * align + ctl->start;
1831 align_off = tmp - entry->offset;
1832 } else {
1833 align_off = 0;
1834 tmp = entry->offset;
1835 }
1836
1837 if (entry->bytes < *bytes + align_off) {
1838 if (entry->bytes > *max_extent_size)
1839 *max_extent_size = entry->bytes;
1840 continue;
1841 }
1842
1843 if (entry->bitmap) {
1844 u64 size = *bytes;
1845
1846 ret = search_bitmap(ctl, entry, &tmp, &size, true);
1847 if (!ret) {
1848 *offset = tmp;
1849 *bytes = size;
1850 return entry;
1851 } else if (size > *max_extent_size) {
1852 *max_extent_size = size;
1853 }
1854 continue;
1855 }
1856
1857 *offset = tmp;
1858 *bytes = entry->bytes - align_off;
1859 return entry;
1860 }
1861out:
1862 return NULL;
1863}
1864
1865static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
1866 struct btrfs_free_space *info, u64 offset)
1867{
1868 info->offset = offset_to_bitmap(ctl, offset);
1869 info->bytes = 0;
1870 INIT_LIST_HEAD(&info->list);
1871 link_free_space(ctl, info);
1872 ctl->total_bitmaps++;
1873
1874 ctl->op->recalc_thresholds(ctl);
1875}
1876
1877static void free_bitmap(struct btrfs_free_space_ctl *ctl,
1878 struct btrfs_free_space *bitmap_info)
1879{
1880 unlink_free_space(ctl, bitmap_info);
1881 kfree(bitmap_info->bitmap);
1882 kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1883 ctl->total_bitmaps--;
1884 ctl->op->recalc_thresholds(ctl);
1885}
1886
1887static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
1888 struct btrfs_free_space *bitmap_info,
1889 u64 *offset, u64 *bytes)
1890{
1891 u64 end;
1892 u64 search_start, search_bytes;
1893 int ret;
1894
1895again:
1896 end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
1897
1898 /*
1899 * We need to search for bits in this bitmap. We could only cover some
1900 * of the extent in this bitmap thanks to how we add space, so we need
1901 * to search for as much as it as we can and clear that amount, and then
1902 * go searching for the next bit.
1903 */
1904 search_start = *offset;
1905 search_bytes = ctl->unit;
1906 search_bytes = min(search_bytes, end - search_start + 1);
1907 ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
1908 false);
1909 if (ret < 0 || search_start != *offset)
1910 return -EINVAL;
1911
1912 /* We may have found more bits than what we need */
1913 search_bytes = min(search_bytes, *bytes);
1914
1915 /* Cannot clear past the end of the bitmap */
1916 search_bytes = min(search_bytes, end - search_start + 1);
1917
1918 bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes);
1919 *offset += search_bytes;
1920 *bytes -= search_bytes;
1921
1922 if (*bytes) {
1923 struct rb_node *next = rb_next(&bitmap_info->offset_index);
1924 if (!bitmap_info->bytes)
1925 free_bitmap(ctl, bitmap_info);
1926
1927 /*
1928 * no entry after this bitmap, but we still have bytes to
1929 * remove, so something has gone wrong.
1930 */
1931 if (!next)
1932 return -EINVAL;
1933
1934 bitmap_info = rb_entry(next, struct btrfs_free_space,
1935 offset_index);
1936
1937 /*
1938 * if the next entry isn't a bitmap we need to return to let the
1939 * extent stuff do its work.
1940 */
1941 if (!bitmap_info->bitmap)
1942 return -EAGAIN;
1943
1944 /*
1945 * Ok the next item is a bitmap, but it may not actually hold
1946 * the information for the rest of this free space stuff, so
1947 * look for it, and if we don't find it return so we can try
1948 * everything over again.
1949 */
1950 search_start = *offset;
1951 search_bytes = ctl->unit;
1952 ret = search_bitmap(ctl, bitmap_info, &search_start,
1953 &search_bytes, false);
1954 if (ret < 0 || search_start != *offset)
1955 return -EAGAIN;
1956
1957 goto again;
1958 } else if (!bitmap_info->bytes)
1959 free_bitmap(ctl, bitmap_info);
1960
1961 return 0;
1962}
1963
1964static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
1965 struct btrfs_free_space *info, u64 offset,
1966 u64 bytes)
1967{
1968 u64 bytes_to_set = 0;
1969 u64 end;
1970
1971 end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
1972
1973 bytes_to_set = min(end - offset, bytes);
1974
1975 bitmap_set_bits(ctl, info, offset, bytes_to_set);
1976
1977 /*
1978 * We set some bytes, we have no idea what the max extent size is
1979 * anymore.
1980 */
1981 info->max_extent_size = 0;
1982
1983 return bytes_to_set;
1984
1985}
1986
1987static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
1988 struct btrfs_free_space *info)
1989{
1990 struct btrfs_block_group_cache *block_group = ctl->private;
1991 struct btrfs_fs_info *fs_info = block_group->fs_info;
1992 bool forced = false;
1993
1994#ifdef CONFIG_BTRFS_DEBUG
1995 if (btrfs_should_fragment_free_space(block_group))
1996 forced = true;
1997#endif
1998
1999 /*
2000 * If we are below the extents threshold then we can add this as an
2001 * extent, and don't have to deal with the bitmap
2002 */
2003 if (!forced && ctl->free_extents < ctl->extents_thresh) {
2004 /*
2005 * If this block group has some small extents we don't want to
2006 * use up all of our free slots in the cache with them, we want
2007 * to reserve them to larger extents, however if we have plenty
2008 * of cache left then go ahead an dadd them, no sense in adding
2009 * the overhead of a bitmap if we don't have to.
2010 */
2011 if (info->bytes <= fs_info->sectorsize * 4) {
2012 if (ctl->free_extents * 2 <= ctl->extents_thresh)
2013 return false;
2014 } else {
2015 return false;
2016 }
2017 }
2018
2019 /*
2020 * The original block groups from mkfs can be really small, like 8
2021 * megabytes, so don't bother with a bitmap for those entries. However
2022 * some block groups can be smaller than what a bitmap would cover but
2023 * are still large enough that they could overflow the 32k memory limit,
2024 * so allow those block groups to still be allowed to have a bitmap
2025 * entry.
2026 */
2027 if (((BITS_PER_BITMAP * ctl->unit) >> 1) > block_group->key.offset)
2028 return false;
2029
2030 return true;
2031}
2032
2033static const struct btrfs_free_space_op free_space_op = {
2034 .recalc_thresholds = recalculate_thresholds,
2035 .use_bitmap = use_bitmap,
2036};
2037
2038static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
2039 struct btrfs_free_space *info)
2040{
2041 struct btrfs_free_space *bitmap_info;
2042 struct btrfs_block_group_cache *block_group = NULL;
2043 int added = 0;
2044 u64 bytes, offset, bytes_added;
2045 int ret;
2046
2047 bytes = info->bytes;
2048 offset = info->offset;
2049
2050 if (!ctl->op->use_bitmap(ctl, info))
2051 return 0;
2052
2053 if (ctl->op == &free_space_op)
2054 block_group = ctl->private;
2055again:
2056 /*
2057 * Since we link bitmaps right into the cluster we need to see if we
2058 * have a cluster here, and if so and it has our bitmap we need to add
2059 * the free space to that bitmap.
2060 */
2061 if (block_group && !list_empty(&block_group->cluster_list)) {
2062 struct btrfs_free_cluster *cluster;
2063 struct rb_node *node;
2064 struct btrfs_free_space *entry;
2065
2066 cluster = list_entry(block_group->cluster_list.next,
2067 struct btrfs_free_cluster,
2068 block_group_list);
2069 spin_lock(&cluster->lock);
2070 node = rb_first(&cluster->root);
2071 if (!node) {
2072 spin_unlock(&cluster->lock);
2073 goto no_cluster_bitmap;
2074 }
2075
2076 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2077 if (!entry->bitmap) {
2078 spin_unlock(&cluster->lock);
2079 goto no_cluster_bitmap;
2080 }
2081
2082 if (entry->offset == offset_to_bitmap(ctl, offset)) {
2083 bytes_added = add_bytes_to_bitmap(ctl, entry,
2084 offset, bytes);
2085 bytes -= bytes_added;
2086 offset += bytes_added;
2087 }
2088 spin_unlock(&cluster->lock);
2089 if (!bytes) {
2090 ret = 1;
2091 goto out;
2092 }
2093 }
2094
2095no_cluster_bitmap:
2096 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2097 1, 0);
2098 if (!bitmap_info) {
2099 ASSERT(added == 0);
2100 goto new_bitmap;
2101 }
2102
2103 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
2104 bytes -= bytes_added;
2105 offset += bytes_added;
2106 added = 0;
2107
2108 if (!bytes) {
2109 ret = 1;
2110 goto out;
2111 } else
2112 goto again;
2113
2114new_bitmap:
2115 if (info && info->bitmap) {
2116 add_new_bitmap(ctl, info, offset);
2117 added = 1;
2118 info = NULL;
2119 goto again;
2120 } else {
2121 spin_unlock(&ctl->tree_lock);
2122
2123 /* no pre-allocated info, allocate a new one */
2124 if (!info) {
2125 info = kmem_cache_zalloc(btrfs_free_space_cachep,
2126 GFP_NOFS);
2127 if (!info) {
2128 spin_lock(&ctl->tree_lock);
2129 ret = -ENOMEM;
2130 goto out;
2131 }
2132 }
2133
2134 /* allocate the bitmap */
2135 info->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
2136 spin_lock(&ctl->tree_lock);
2137 if (!info->bitmap) {
2138 ret = -ENOMEM;
2139 goto out;
2140 }
2141 goto again;
2142 }
2143
2144out:
2145 if (info) {
2146 if (info->bitmap)
2147 kfree(info->bitmap);
2148 kmem_cache_free(btrfs_free_space_cachep, info);
2149 }
2150
2151 return ret;
2152}
2153
2154static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
2155 struct btrfs_free_space *info, bool update_stat)
2156{
2157 struct btrfs_free_space *left_info;
2158 struct btrfs_free_space *right_info;
2159 bool merged = false;
2160 u64 offset = info->offset;
2161 u64 bytes = info->bytes;
2162
2163 /*
2164 * first we want to see if there is free space adjacent to the range we
2165 * are adding, if there is remove that struct and add a new one to
2166 * cover the entire range
2167 */
2168 right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
2169 if (right_info && rb_prev(&right_info->offset_index))
2170 left_info = rb_entry(rb_prev(&right_info->offset_index),
2171 struct btrfs_free_space, offset_index);
2172 else
2173 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
2174
2175 if (right_info && !right_info->bitmap) {
2176 if (update_stat)
2177 unlink_free_space(ctl, right_info);
2178 else
2179 __unlink_free_space(ctl, right_info);
2180 info->bytes += right_info->bytes;
2181 kmem_cache_free(btrfs_free_space_cachep, right_info);
2182 merged = true;
2183 }
2184
2185 if (left_info && !left_info->bitmap &&
2186 left_info->offset + left_info->bytes == offset) {
2187 if (update_stat)
2188 unlink_free_space(ctl, left_info);
2189 else
2190 __unlink_free_space(ctl, left_info);
2191 info->offset = left_info->offset;
2192 info->bytes += left_info->bytes;
2193 kmem_cache_free(btrfs_free_space_cachep, left_info);
2194 merged = true;
2195 }
2196
2197 return merged;
2198}
2199
2200static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
2201 struct btrfs_free_space *info,
2202 bool update_stat)
2203{
2204 struct btrfs_free_space *bitmap;
2205 unsigned long i;
2206 unsigned long j;
2207 const u64 end = info->offset + info->bytes;
2208 const u64 bitmap_offset = offset_to_bitmap(ctl, end);
2209 u64 bytes;
2210
2211 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2212 if (!bitmap)
2213 return false;
2214
2215 i = offset_to_bit(bitmap->offset, ctl->unit, end);
2216 j = find_next_zero_bit(bitmap->bitmap, BITS_PER_BITMAP, i);
2217 if (j == i)
2218 return false;
2219 bytes = (j - i) * ctl->unit;
2220 info->bytes += bytes;
2221
2222 if (update_stat)
2223 bitmap_clear_bits(ctl, bitmap, end, bytes);
2224 else
2225 __bitmap_clear_bits(ctl, bitmap, end, bytes);
2226
2227 if (!bitmap->bytes)
2228 free_bitmap(ctl, bitmap);
2229
2230 return true;
2231}
2232
2233static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
2234 struct btrfs_free_space *info,
2235 bool update_stat)
2236{
2237 struct btrfs_free_space *bitmap;
2238 u64 bitmap_offset;
2239 unsigned long i;
2240 unsigned long j;
2241 unsigned long prev_j;
2242 u64 bytes;
2243
2244 bitmap_offset = offset_to_bitmap(ctl, info->offset);
2245 /* If we're on a boundary, try the previous logical bitmap. */
2246 if (bitmap_offset == info->offset) {
2247 if (info->offset == 0)
2248 return false;
2249 bitmap_offset = offset_to_bitmap(ctl, info->offset - 1);
2250 }
2251
2252 bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2253 if (!bitmap)
2254 return false;
2255
2256 i = offset_to_bit(bitmap->offset, ctl->unit, info->offset) - 1;
2257 j = 0;
2258 prev_j = (unsigned long)-1;
2259 for_each_clear_bit_from(j, bitmap->bitmap, BITS_PER_BITMAP) {
2260 if (j > i)
2261 break;
2262 prev_j = j;
2263 }
2264 if (prev_j == i)
2265 return false;
2266
2267 if (prev_j == (unsigned long)-1)
2268 bytes = (i + 1) * ctl->unit;
2269 else
2270 bytes = (i - prev_j) * ctl->unit;
2271
2272 info->offset -= bytes;
2273 info->bytes += bytes;
2274
2275 if (update_stat)
2276 bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2277 else
2278 __bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2279
2280 if (!bitmap->bytes)
2281 free_bitmap(ctl, bitmap);
2282
2283 return true;
2284}
2285
2286/*
2287 * We prefer always to allocate from extent entries, both for clustered and
2288 * non-clustered allocation requests. So when attempting to add a new extent
2289 * entry, try to see if there's adjacent free space in bitmap entries, and if
2290 * there is, migrate that space from the bitmaps to the extent.
2291 * Like this we get better chances of satisfying space allocation requests
2292 * because we attempt to satisfy them based on a single cache entry, and never
2293 * on 2 or more entries - even if the entries represent a contiguous free space
2294 * region (e.g. 1 extent entry + 1 bitmap entry starting where the extent entry
2295 * ends).
2296 */
2297static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
2298 struct btrfs_free_space *info,
2299 bool update_stat)
2300{
2301 /*
2302 * Only work with disconnected entries, as we can change their offset,
2303 * and must be extent entries.
2304 */
2305 ASSERT(!info->bitmap);
2306 ASSERT(RB_EMPTY_NODE(&info->offset_index));
2307
2308 if (ctl->total_bitmaps > 0) {
2309 bool stole_end;
2310 bool stole_front = false;
2311
2312 stole_end = steal_from_bitmap_to_end(ctl, info, update_stat);
2313 if (ctl->total_bitmaps > 0)
2314 stole_front = steal_from_bitmap_to_front(ctl, info,
2315 update_stat);
2316
2317 if (stole_end || stole_front)
2318 try_merge_free_space(ctl, info, update_stat);
2319 }
2320}
2321
2322int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
2323 struct btrfs_free_space_ctl *ctl,
2324 u64 offset, u64 bytes)
2325{
2326 struct btrfs_free_space *info;
2327 int ret = 0;
2328
2329 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
2330 if (!info)
2331 return -ENOMEM;
2332
2333 info->offset = offset;
2334 info->bytes = bytes;
2335 RB_CLEAR_NODE(&info->offset_index);
2336
2337 spin_lock(&ctl->tree_lock);
2338
2339 if (try_merge_free_space(ctl, info, true))
2340 goto link;
2341
2342 /*
2343 * There was no extent directly to the left or right of this new
2344 * extent then we know we're going to have to allocate a new extent, so
2345 * before we do that see if we need to drop this into a bitmap
2346 */
2347 ret = insert_into_bitmap(ctl, info);
2348 if (ret < 0) {
2349 goto out;
2350 } else if (ret) {
2351 ret = 0;
2352 goto out;
2353 }
2354link:
2355 /*
2356 * Only steal free space from adjacent bitmaps if we're sure we're not
2357 * going to add the new free space to existing bitmap entries - because
2358 * that would mean unnecessary work that would be reverted. Therefore
2359 * attempt to steal space from bitmaps if we're adding an extent entry.
2360 */
2361 steal_from_bitmap(ctl, info, true);
2362
2363 ret = link_free_space(ctl, info);
2364 if (ret)
2365 kmem_cache_free(btrfs_free_space_cachep, info);
2366out:
2367 spin_unlock(&ctl->tree_lock);
2368
2369 if (ret) {
2370 btrfs_crit(fs_info, "unable to add free space :%d", ret);
2371 ASSERT(ret != -EEXIST);
2372 }
2373
2374 return ret;
2375}
2376
2377int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
2378 u64 offset, u64 bytes)
2379{
2380 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2381 struct btrfs_free_space *info;
2382 int ret;
2383 bool re_search = false;
2384
2385 spin_lock(&ctl->tree_lock);
2386
2387again:
2388 ret = 0;
2389 if (!bytes)
2390 goto out_lock;
2391
2392 info = tree_search_offset(ctl, offset, 0, 0);
2393 if (!info) {
2394 /*
2395 * oops didn't find an extent that matched the space we wanted
2396 * to remove, look for a bitmap instead
2397 */
2398 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2399 1, 0);
2400 if (!info) {
2401 /*
2402 * If we found a partial bit of our free space in a
2403 * bitmap but then couldn't find the other part this may
2404 * be a problem, so WARN about it.
2405 */
2406 WARN_ON(re_search);
2407 goto out_lock;
2408 }
2409 }
2410
2411 re_search = false;
2412 if (!info->bitmap) {
2413 unlink_free_space(ctl, info);
2414 if (offset == info->offset) {
2415 u64 to_free = min(bytes, info->bytes);
2416
2417 info->bytes -= to_free;
2418 info->offset += to_free;
2419 if (info->bytes) {
2420 ret = link_free_space(ctl, info);
2421 WARN_ON(ret);
2422 } else {
2423 kmem_cache_free(btrfs_free_space_cachep, info);
2424 }
2425
2426 offset += to_free;
2427 bytes -= to_free;
2428 goto again;
2429 } else {
2430 u64 old_end = info->bytes + info->offset;
2431
2432 info->bytes = offset - info->offset;
2433 ret = link_free_space(ctl, info);
2434 WARN_ON(ret);
2435 if (ret)
2436 goto out_lock;
2437
2438 /* Not enough bytes in this entry to satisfy us */
2439 if (old_end < offset + bytes) {
2440 bytes -= old_end - offset;
2441 offset = old_end;
2442 goto again;
2443 } else if (old_end == offset + bytes) {
2444 /* all done */
2445 goto out_lock;
2446 }
2447 spin_unlock(&ctl->tree_lock);
2448
2449 ret = btrfs_add_free_space(block_group, offset + bytes,
2450 old_end - (offset + bytes));
2451 WARN_ON(ret);
2452 goto out;
2453 }
2454 }
2455
2456 ret = remove_from_bitmap(ctl, info, &offset, &bytes);
2457 if (ret == -EAGAIN) {
2458 re_search = true;
2459 goto again;
2460 }
2461out_lock:
2462 spin_unlock(&ctl->tree_lock);
2463out:
2464 return ret;
2465}
2466
2467void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
2468 u64 bytes)
2469{
2470 struct btrfs_fs_info *fs_info = block_group->fs_info;
2471 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2472 struct btrfs_free_space *info;
2473 struct rb_node *n;
2474 int count = 0;
2475
2476 for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
2477 info = rb_entry(n, struct btrfs_free_space, offset_index);
2478 if (info->bytes >= bytes && !block_group->ro)
2479 count++;
2480 btrfs_crit(fs_info, "entry offset %llu, bytes %llu, bitmap %s",
2481 info->offset, info->bytes,
2482 (info->bitmap) ? "yes" : "no");
2483 }
2484 btrfs_info(fs_info, "block group has cluster?: %s",
2485 list_empty(&block_group->cluster_list) ? "no" : "yes");
2486 btrfs_info(fs_info,
2487 "%d blocks of free space at or bigger than bytes is", count);
2488}
2489
2490void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group)
2491{
2492 struct btrfs_fs_info *fs_info = block_group->fs_info;
2493 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2494
2495 spin_lock_init(&ctl->tree_lock);
2496 ctl->unit = fs_info->sectorsize;
2497 ctl->start = block_group->key.objectid;
2498 ctl->private = block_group;
2499 ctl->op = &free_space_op;
2500 INIT_LIST_HEAD(&ctl->trimming_ranges);
2501 mutex_init(&ctl->cache_writeout_mutex);
2502
2503 /*
2504 * we only want to have 32k of ram per block group for keeping
2505 * track of free space, and if we pass 1/2 of that we want to
2506 * start converting things over to using bitmaps
2507 */
2508 ctl->extents_thresh = (SZ_32K / 2) / sizeof(struct btrfs_free_space);
2509}
2510
2511/*
2512 * for a given cluster, put all of its extents back into the free
2513 * space cache. If the block group passed doesn't match the block group
2514 * pointed to by the cluster, someone else raced in and freed the
2515 * cluster already. In that case, we just return without changing anything
2516 */
2517static int
2518__btrfs_return_cluster_to_free_space(
2519 struct btrfs_block_group_cache *block_group,
2520 struct btrfs_free_cluster *cluster)
2521{
2522 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2523 struct btrfs_free_space *entry;
2524 struct rb_node *node;
2525
2526 spin_lock(&cluster->lock);
2527 if (cluster->block_group != block_group)
2528 goto out;
2529
2530 cluster->block_group = NULL;
2531 cluster->window_start = 0;
2532 list_del_init(&cluster->block_group_list);
2533
2534 node = rb_first(&cluster->root);
2535 while (node) {
2536 bool bitmap;
2537
2538 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2539 node = rb_next(&entry->offset_index);
2540 rb_erase(&entry->offset_index, &cluster->root);
2541 RB_CLEAR_NODE(&entry->offset_index);
2542
2543 bitmap = (entry->bitmap != NULL);
2544 if (!bitmap) {
2545 try_merge_free_space(ctl, entry, false);
2546 steal_from_bitmap(ctl, entry, false);
2547 }
2548 tree_insert_offset(&ctl->free_space_offset,
2549 entry->offset, &entry->offset_index, bitmap);
2550 }
2551 cluster->root = RB_ROOT;
2552
2553out:
2554 spin_unlock(&cluster->lock);
2555 btrfs_put_block_group(block_group);
2556 return 0;
2557}
2558
2559static void __btrfs_remove_free_space_cache_locked(
2560 struct btrfs_free_space_ctl *ctl)
2561{
2562 struct btrfs_free_space *info;
2563 struct rb_node *node;
2564
2565 while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
2566 info = rb_entry(node, struct btrfs_free_space, offset_index);
2567 if (!info->bitmap) {
2568 unlink_free_space(ctl, info);
2569 kmem_cache_free(btrfs_free_space_cachep, info);
2570 } else {
2571 free_bitmap(ctl, info);
2572 }
2573
2574 cond_resched_lock(&ctl->tree_lock);
2575 }
2576}
2577
2578void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
2579{
2580 spin_lock(&ctl->tree_lock);
2581 __btrfs_remove_free_space_cache_locked(ctl);
2582 spin_unlock(&ctl->tree_lock);
2583}
2584
2585void btrfs_remove_free_space_cache(struct btrfs_block_group_cache *block_group)
2586{
2587 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2588 struct btrfs_free_cluster *cluster;
2589 struct list_head *head;
2590
2591 spin_lock(&ctl->tree_lock);
2592 while ((head = block_group->cluster_list.next) !=
2593 &block_group->cluster_list) {
2594 cluster = list_entry(head, struct btrfs_free_cluster,
2595 block_group_list);
2596
2597 WARN_ON(cluster->block_group != block_group);
2598 __btrfs_return_cluster_to_free_space(block_group, cluster);
2599
2600 cond_resched_lock(&ctl->tree_lock);
2601 }
2602 __btrfs_remove_free_space_cache_locked(ctl);
2603 spin_unlock(&ctl->tree_lock);
2604
2605}
2606
2607u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group,
2608 u64 offset, u64 bytes, u64 empty_size,
2609 u64 *max_extent_size)
2610{
2611 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2612 struct btrfs_free_space *entry = NULL;
2613 u64 bytes_search = bytes + empty_size;
2614 u64 ret = 0;
2615 u64 align_gap = 0;
2616 u64 align_gap_len = 0;
2617
2618 spin_lock(&ctl->tree_lock);
2619 entry = find_free_space(ctl, &offset, &bytes_search,
2620 block_group->full_stripe_len, max_extent_size);
2621 if (!entry)
2622 goto out;
2623
2624 ret = offset;
2625 if (entry->bitmap) {
2626 bitmap_clear_bits(ctl, entry, offset, bytes);
2627 if (!entry->bytes)
2628 free_bitmap(ctl, entry);
2629 } else {
2630 unlink_free_space(ctl, entry);
2631 align_gap_len = offset - entry->offset;
2632 align_gap = entry->offset;
2633
2634 entry->offset = offset + bytes;
2635 WARN_ON(entry->bytes < bytes + align_gap_len);
2636
2637 entry->bytes -= bytes + align_gap_len;
2638 if (!entry->bytes)
2639 kmem_cache_free(btrfs_free_space_cachep, entry);
2640 else
2641 link_free_space(ctl, entry);
2642 }
2643out:
2644 spin_unlock(&ctl->tree_lock);
2645
2646 if (align_gap_len)
2647 __btrfs_add_free_space(block_group->fs_info, ctl,
2648 align_gap, align_gap_len);
2649 return ret;
2650}
2651
2652/*
2653 * given a cluster, put all of its extents back into the free space
2654 * cache. If a block group is passed, this function will only free
2655 * a cluster that belongs to the passed block group.
2656 *
2657 * Otherwise, it'll get a reference on the block group pointed to by the
2658 * cluster and remove the cluster from it.
2659 */
2660int btrfs_return_cluster_to_free_space(
2661 struct btrfs_block_group_cache *block_group,
2662 struct btrfs_free_cluster *cluster)
2663{
2664 struct btrfs_free_space_ctl *ctl;
2665 int ret;
2666
2667 /* first, get a safe pointer to the block group */
2668 spin_lock(&cluster->lock);
2669 if (!block_group) {
2670 block_group = cluster->block_group;
2671 if (!block_group) {
2672 spin_unlock(&cluster->lock);
2673 return 0;
2674 }
2675 } else if (cluster->block_group != block_group) {
2676 /* someone else has already freed it don't redo their work */
2677 spin_unlock(&cluster->lock);
2678 return 0;
2679 }
2680 atomic_inc(&block_group->count);
2681 spin_unlock(&cluster->lock);
2682
2683 ctl = block_group->free_space_ctl;
2684
2685 /* now return any extents the cluster had on it */
2686 spin_lock(&ctl->tree_lock);
2687 ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
2688 spin_unlock(&ctl->tree_lock);
2689
2690 /* finally drop our ref */
2691 btrfs_put_block_group(block_group);
2692 return ret;
2693}
2694
2695static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
2696 struct btrfs_free_cluster *cluster,
2697 struct btrfs_free_space *entry,
2698 u64 bytes, u64 min_start,
2699 u64 *max_extent_size)
2700{
2701 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2702 int err;
2703 u64 search_start = cluster->window_start;
2704 u64 search_bytes = bytes;
2705 u64 ret = 0;
2706
2707 search_start = min_start;
2708 search_bytes = bytes;
2709
2710 err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
2711 if (err) {
2712 if (search_bytes > *max_extent_size)
2713 *max_extent_size = search_bytes;
2714 return 0;
2715 }
2716
2717 ret = search_start;
2718 __bitmap_clear_bits(ctl, entry, ret, bytes);
2719
2720 return ret;
2721}
2722
2723/*
2724 * given a cluster, try to allocate 'bytes' from it, returns 0
2725 * if it couldn't find anything suitably large, or a logical disk offset
2726 * if things worked out
2727 */
2728u64 btrfs_alloc_from_cluster(struct btrfs_block_group_cache *block_group,
2729 struct btrfs_free_cluster *cluster, u64 bytes,
2730 u64 min_start, u64 *max_extent_size)
2731{
2732 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2733 struct btrfs_free_space *entry = NULL;
2734 struct rb_node *node;
2735 u64 ret = 0;
2736
2737 spin_lock(&cluster->lock);
2738 if (bytes > cluster->max_size)
2739 goto out;
2740
2741 if (cluster->block_group != block_group)
2742 goto out;
2743
2744 node = rb_first(&cluster->root);
2745 if (!node)
2746 goto out;
2747
2748 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2749 while (1) {
2750 if (entry->bytes < bytes && entry->bytes > *max_extent_size)
2751 *max_extent_size = entry->bytes;
2752
2753 if (entry->bytes < bytes ||
2754 (!entry->bitmap && entry->offset < min_start)) {
2755 node = rb_next(&entry->offset_index);
2756 if (!node)
2757 break;
2758 entry = rb_entry(node, struct btrfs_free_space,
2759 offset_index);
2760 continue;
2761 }
2762
2763 if (entry->bitmap) {
2764 ret = btrfs_alloc_from_bitmap(block_group,
2765 cluster, entry, bytes,
2766 cluster->window_start,
2767 max_extent_size);
2768 if (ret == 0) {
2769 node = rb_next(&entry->offset_index);
2770 if (!node)
2771 break;
2772 entry = rb_entry(node, struct btrfs_free_space,
2773 offset_index);
2774 continue;
2775 }
2776 cluster->window_start += bytes;
2777 } else {
2778 ret = entry->offset;
2779
2780 entry->offset += bytes;
2781 entry->bytes -= bytes;
2782 }
2783
2784 if (entry->bytes == 0)
2785 rb_erase(&entry->offset_index, &cluster->root);
2786 break;
2787 }
2788out:
2789 spin_unlock(&cluster->lock);
2790
2791 if (!ret)
2792 return 0;
2793
2794 spin_lock(&ctl->tree_lock);
2795
2796 ctl->free_space -= bytes;
2797 if (entry->bytes == 0) {
2798 ctl->free_extents--;
2799 if (entry->bitmap) {
2800 kfree(entry->bitmap);
2801 ctl->total_bitmaps--;
2802 ctl->op->recalc_thresholds(ctl);
2803 }
2804 kmem_cache_free(btrfs_free_space_cachep, entry);
2805 }
2806
2807 spin_unlock(&ctl->tree_lock);
2808
2809 return ret;
2810}
2811
2812static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
2813 struct btrfs_free_space *entry,
2814 struct btrfs_free_cluster *cluster,
2815 u64 offset, u64 bytes,
2816 u64 cont1_bytes, u64 min_bytes)
2817{
2818 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2819 unsigned long next_zero;
2820 unsigned long i;
2821 unsigned long want_bits;
2822 unsigned long min_bits;
2823 unsigned long found_bits;
2824 unsigned long max_bits = 0;
2825 unsigned long start = 0;
2826 unsigned long total_found = 0;
2827 int ret;
2828
2829 i = offset_to_bit(entry->offset, ctl->unit,
2830 max_t(u64, offset, entry->offset));
2831 want_bits = bytes_to_bits(bytes, ctl->unit);
2832 min_bits = bytes_to_bits(min_bytes, ctl->unit);
2833
2834 /*
2835 * Don't bother looking for a cluster in this bitmap if it's heavily
2836 * fragmented.
2837 */
2838 if (entry->max_extent_size &&
2839 entry->max_extent_size < cont1_bytes)
2840 return -ENOSPC;
2841again:
2842 found_bits = 0;
2843 for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
2844 next_zero = find_next_zero_bit(entry->bitmap,
2845 BITS_PER_BITMAP, i);
2846 if (next_zero - i >= min_bits) {
2847 found_bits = next_zero - i;
2848 if (found_bits > max_bits)
2849 max_bits = found_bits;
2850 break;
2851 }
2852 if (next_zero - i > max_bits)
2853 max_bits = next_zero - i;
2854 i = next_zero;
2855 }
2856
2857 if (!found_bits) {
2858 entry->max_extent_size = (u64)max_bits * ctl->unit;
2859 return -ENOSPC;
2860 }
2861
2862 if (!total_found) {
2863 start = i;
2864 cluster->max_size = 0;
2865 }
2866
2867 total_found += found_bits;
2868
2869 if (cluster->max_size < found_bits * ctl->unit)
2870 cluster->max_size = found_bits * ctl->unit;
2871
2872 if (total_found < want_bits || cluster->max_size < cont1_bytes) {
2873 i = next_zero + 1;
2874 goto again;
2875 }
2876
2877 cluster->window_start = start * ctl->unit + entry->offset;
2878 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2879 ret = tree_insert_offset(&cluster->root, entry->offset,
2880 &entry->offset_index, 1);
2881 ASSERT(!ret); /* -EEXIST; Logic error */
2882
2883 trace_btrfs_setup_cluster(block_group, cluster,
2884 total_found * ctl->unit, 1);
2885 return 0;
2886}
2887
2888/*
2889 * This searches the block group for just extents to fill the cluster with.
2890 * Try to find a cluster with at least bytes total bytes, at least one
2891 * extent of cont1_bytes, and other clusters of at least min_bytes.
2892 */
2893static noinline int
2894setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2895 struct btrfs_free_cluster *cluster,
2896 struct list_head *bitmaps, u64 offset, u64 bytes,
2897 u64 cont1_bytes, u64 min_bytes)
2898{
2899 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2900 struct btrfs_free_space *first = NULL;
2901 struct btrfs_free_space *entry = NULL;
2902 struct btrfs_free_space *last;
2903 struct rb_node *node;
2904 u64 window_free;
2905 u64 max_extent;
2906 u64 total_size = 0;
2907
2908 entry = tree_search_offset(ctl, offset, 0, 1);
2909 if (!entry)
2910 return -ENOSPC;
2911
2912 /*
2913 * We don't want bitmaps, so just move along until we find a normal
2914 * extent entry.
2915 */
2916 while (entry->bitmap || entry->bytes < min_bytes) {
2917 if (entry->bitmap && list_empty(&entry->list))
2918 list_add_tail(&entry->list, bitmaps);
2919 node = rb_next(&entry->offset_index);
2920 if (!node)
2921 return -ENOSPC;
2922 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2923 }
2924
2925 window_free = entry->bytes;
2926 max_extent = entry->bytes;
2927 first = entry;
2928 last = entry;
2929
2930 for (node = rb_next(&entry->offset_index); node;
2931 node = rb_next(&entry->offset_index)) {
2932 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2933
2934 if (entry->bitmap) {
2935 if (list_empty(&entry->list))
2936 list_add_tail(&entry->list, bitmaps);
2937 continue;
2938 }
2939
2940 if (entry->bytes < min_bytes)
2941 continue;
2942
2943 last = entry;
2944 window_free += entry->bytes;
2945 if (entry->bytes > max_extent)
2946 max_extent = entry->bytes;
2947 }
2948
2949 if (window_free < bytes || max_extent < cont1_bytes)
2950 return -ENOSPC;
2951
2952 cluster->window_start = first->offset;
2953
2954 node = &first->offset_index;
2955
2956 /*
2957 * now we've found our entries, pull them out of the free space
2958 * cache and put them into the cluster rbtree
2959 */
2960 do {
2961 int ret;
2962
2963 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2964 node = rb_next(&entry->offset_index);
2965 if (entry->bitmap || entry->bytes < min_bytes)
2966 continue;
2967
2968 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2969 ret = tree_insert_offset(&cluster->root, entry->offset,
2970 &entry->offset_index, 0);
2971 total_size += entry->bytes;
2972 ASSERT(!ret); /* -EEXIST; Logic error */
2973 } while (node && entry != last);
2974
2975 cluster->max_size = max_extent;
2976 trace_btrfs_setup_cluster(block_group, cluster, total_size, 0);
2977 return 0;
2978}
2979
2980/*
2981 * This specifically looks for bitmaps that may work in the cluster, we assume
2982 * that we have already failed to find extents that will work.
2983 */
2984static noinline int
2985setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2986 struct btrfs_free_cluster *cluster,
2987 struct list_head *bitmaps, u64 offset, u64 bytes,
2988 u64 cont1_bytes, u64 min_bytes)
2989{
2990 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2991 struct btrfs_free_space *entry = NULL;
2992 int ret = -ENOSPC;
2993 u64 bitmap_offset = offset_to_bitmap(ctl, offset);
2994
2995 if (ctl->total_bitmaps == 0)
2996 return -ENOSPC;
2997
2998 /*
2999 * The bitmap that covers offset won't be in the list unless offset
3000 * is just its start offset.
3001 */
3002 if (!list_empty(bitmaps))
3003 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
3004
3005 if (!entry || entry->offset != bitmap_offset) {
3006 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
3007 if (entry && list_empty(&entry->list))
3008 list_add(&entry->list, bitmaps);
3009 }
3010
3011 list_for_each_entry(entry, bitmaps, list) {
3012 if (entry->bytes < bytes)
3013 continue;
3014 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
3015 bytes, cont1_bytes, min_bytes);
3016 if (!ret)
3017 return 0;
3018 }
3019
3020 /*
3021 * The bitmaps list has all the bitmaps that record free space
3022 * starting after offset, so no more search is required.
3023 */
3024 return -ENOSPC;
3025}
3026
3027/*
3028 * here we try to find a cluster of blocks in a block group. The goal
3029 * is to find at least bytes+empty_size.
3030 * We might not find them all in one contiguous area.
3031 *
3032 * returns zero and sets up cluster if things worked out, otherwise
3033 * it returns -enospc
3034 */
3035int btrfs_find_space_cluster(struct btrfs_fs_info *fs_info,
3036 struct btrfs_block_group_cache *block_group,
3037 struct btrfs_free_cluster *cluster,
3038 u64 offset, u64 bytes, u64 empty_size)
3039{
3040 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3041 struct btrfs_free_space *entry, *tmp;
3042 LIST_HEAD(bitmaps);
3043 u64 min_bytes;
3044 u64 cont1_bytes;
3045 int ret;
3046
3047 /*
3048 * Choose the minimum extent size we'll require for this
3049 * cluster. For SSD_SPREAD, don't allow any fragmentation.
3050 * For metadata, allow allocates with smaller extents. For
3051 * data, keep it dense.
3052 */
3053 if (btrfs_test_opt(fs_info, SSD_SPREAD)) {
3054 cont1_bytes = min_bytes = bytes + empty_size;
3055 } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
3056 cont1_bytes = bytes;
3057 min_bytes = fs_info->sectorsize;
3058 } else {
3059 cont1_bytes = max(bytes, (bytes + empty_size) >> 2);
3060 min_bytes = fs_info->sectorsize;
3061 }
3062
3063 spin_lock(&ctl->tree_lock);
3064
3065 /*
3066 * If we know we don't have enough space to make a cluster don't even
3067 * bother doing all the work to try and find one.
3068 */
3069 if (ctl->free_space < bytes) {
3070 spin_unlock(&ctl->tree_lock);
3071 return -ENOSPC;
3072 }
3073
3074 spin_lock(&cluster->lock);
3075
3076 /* someone already found a cluster, hooray */
3077 if (cluster->block_group) {
3078 ret = 0;
3079 goto out;
3080 }
3081
3082 trace_btrfs_find_cluster(block_group, offset, bytes, empty_size,
3083 min_bytes);
3084
3085 ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
3086 bytes + empty_size,
3087 cont1_bytes, min_bytes);
3088 if (ret)
3089 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
3090 offset, bytes + empty_size,
3091 cont1_bytes, min_bytes);
3092
3093 /* Clear our temporary list */
3094 list_for_each_entry_safe(entry, tmp, &bitmaps, list)
3095 list_del_init(&entry->list);
3096
3097 if (!ret) {
3098 atomic_inc(&block_group->count);
3099 list_add_tail(&cluster->block_group_list,
3100 &block_group->cluster_list);
3101 cluster->block_group = block_group;
3102 } else {
3103 trace_btrfs_failed_cluster_setup(block_group);
3104 }
3105out:
3106 spin_unlock(&cluster->lock);
3107 spin_unlock(&ctl->tree_lock);
3108
3109 return ret;
3110}
3111
3112/*
3113 * simple code to zero out a cluster
3114 */
3115void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
3116{
3117 spin_lock_init(&cluster->lock);
3118 spin_lock_init(&cluster->refill_lock);
3119 cluster->root = RB_ROOT;
3120 cluster->max_size = 0;
3121 cluster->fragmented = false;
3122 INIT_LIST_HEAD(&cluster->block_group_list);
3123 cluster->block_group = NULL;
3124}
3125
3126static int do_trimming(struct btrfs_block_group_cache *block_group,
3127 u64 *total_trimmed, u64 start, u64 bytes,
3128 u64 reserved_start, u64 reserved_bytes,
3129 struct btrfs_trim_range *trim_entry)
3130{
3131 struct btrfs_space_info *space_info = block_group->space_info;
3132 struct btrfs_fs_info *fs_info = block_group->fs_info;
3133 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3134 int ret;
3135 int update = 0;
3136 u64 trimmed = 0;
3137
3138 spin_lock(&space_info->lock);
3139 spin_lock(&block_group->lock);
3140 if (!block_group->ro) {
3141 block_group->reserved += reserved_bytes;
3142 space_info->bytes_reserved += reserved_bytes;
3143 update = 1;
3144 }
3145 spin_unlock(&block_group->lock);
3146 spin_unlock(&space_info->lock);
3147
3148 ret = btrfs_discard_extent(fs_info, start, bytes, &trimmed);
3149 if (!ret)
3150 *total_trimmed += trimmed;
3151
3152 mutex_lock(&ctl->cache_writeout_mutex);
3153 btrfs_add_free_space(block_group, reserved_start, reserved_bytes);
3154 list_del(&trim_entry->list);
3155 mutex_unlock(&ctl->cache_writeout_mutex);
3156
3157 if (update) {
3158 spin_lock(&space_info->lock);
3159 spin_lock(&block_group->lock);
3160 if (block_group->ro)
3161 space_info->bytes_readonly += reserved_bytes;
3162 block_group->reserved -= reserved_bytes;
3163 space_info->bytes_reserved -= reserved_bytes;
3164 spin_unlock(&space_info->lock);
3165 spin_unlock(&block_group->lock);
3166 }
3167
3168 return ret;
3169}
3170
3171static int trim_no_bitmap(struct btrfs_block_group_cache *block_group,
3172 u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3173{
3174 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3175 struct btrfs_free_space *entry;
3176 struct rb_node *node;
3177 int ret = 0;
3178 u64 extent_start;
3179 u64 extent_bytes;
3180 u64 bytes;
3181
3182 while (start < end) {
3183 struct btrfs_trim_range trim_entry;
3184
3185 mutex_lock(&ctl->cache_writeout_mutex);
3186 spin_lock(&ctl->tree_lock);
3187
3188 if (ctl->free_space < minlen) {
3189 spin_unlock(&ctl->tree_lock);
3190 mutex_unlock(&ctl->cache_writeout_mutex);
3191 break;
3192 }
3193
3194 entry = tree_search_offset(ctl, start, 0, 1);
3195 if (!entry) {
3196 spin_unlock(&ctl->tree_lock);
3197 mutex_unlock(&ctl->cache_writeout_mutex);
3198 break;
3199 }
3200
3201 /* skip bitmaps */
3202 while (entry->bitmap) {
3203 node = rb_next(&entry->offset_index);
3204 if (!node) {
3205 spin_unlock(&ctl->tree_lock);
3206 mutex_unlock(&ctl->cache_writeout_mutex);
3207 goto out;
3208 }
3209 entry = rb_entry(node, struct btrfs_free_space,
3210 offset_index);
3211 }
3212
3213 if (entry->offset >= end) {
3214 spin_unlock(&ctl->tree_lock);
3215 mutex_unlock(&ctl->cache_writeout_mutex);
3216 break;
3217 }
3218
3219 extent_start = entry->offset;
3220 extent_bytes = entry->bytes;
3221 start = max(start, extent_start);
3222 bytes = min(extent_start + extent_bytes, end) - start;
3223 if (bytes < minlen) {
3224 spin_unlock(&ctl->tree_lock);
3225 mutex_unlock(&ctl->cache_writeout_mutex);
3226 goto next;
3227 }
3228
3229 unlink_free_space(ctl, entry);
3230 kmem_cache_free(btrfs_free_space_cachep, entry);
3231
3232 spin_unlock(&ctl->tree_lock);
3233 trim_entry.start = extent_start;
3234 trim_entry.bytes = extent_bytes;
3235 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3236 mutex_unlock(&ctl->cache_writeout_mutex);
3237
3238 ret = do_trimming(block_group, total_trimmed, start, bytes,
3239 extent_start, extent_bytes, &trim_entry);
3240 if (ret)
3241 break;
3242next:
3243 start += bytes;
3244
3245 if (fatal_signal_pending(current)) {
3246 ret = -ERESTARTSYS;
3247 break;
3248 }
3249
3250 cond_resched();
3251 }
3252out:
3253 return ret;
3254}
3255
3256static int trim_bitmaps(struct btrfs_block_group_cache *block_group,
3257 u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3258{
3259 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3260 struct btrfs_free_space *entry;
3261 int ret = 0;
3262 int ret2;
3263 u64 bytes;
3264 u64 offset = offset_to_bitmap(ctl, start);
3265
3266 while (offset < end) {
3267 bool next_bitmap = false;
3268 struct btrfs_trim_range trim_entry;
3269
3270 mutex_lock(&ctl->cache_writeout_mutex);
3271 spin_lock(&ctl->tree_lock);
3272
3273 if (ctl->free_space < minlen) {
3274 spin_unlock(&ctl->tree_lock);
3275 mutex_unlock(&ctl->cache_writeout_mutex);
3276 break;
3277 }
3278
3279 entry = tree_search_offset(ctl, offset, 1, 0);
3280 if (!entry) {
3281 spin_unlock(&ctl->tree_lock);
3282 mutex_unlock(&ctl->cache_writeout_mutex);
3283 next_bitmap = true;
3284 goto next;
3285 }
3286
3287 bytes = minlen;
3288 ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
3289 if (ret2 || start >= end) {
3290 spin_unlock(&ctl->tree_lock);
3291 mutex_unlock(&ctl->cache_writeout_mutex);
3292 next_bitmap = true;
3293 goto next;
3294 }
3295
3296 bytes = min(bytes, end - start);
3297 if (bytes < minlen) {
3298 spin_unlock(&ctl->tree_lock);
3299 mutex_unlock(&ctl->cache_writeout_mutex);
3300 goto next;
3301 }
3302
3303 bitmap_clear_bits(ctl, entry, start, bytes);
3304 if (entry->bytes == 0)
3305 free_bitmap(ctl, entry);
3306
3307 spin_unlock(&ctl->tree_lock);
3308 trim_entry.start = start;
3309 trim_entry.bytes = bytes;
3310 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3311 mutex_unlock(&ctl->cache_writeout_mutex);
3312
3313 ret = do_trimming(block_group, total_trimmed, start, bytes,
3314 start, bytes, &trim_entry);
3315 if (ret)
3316 break;
3317next:
3318 if (next_bitmap) {
3319 offset += BITS_PER_BITMAP * ctl->unit;
3320 } else {
3321 start += bytes;
3322 if (start >= offset + BITS_PER_BITMAP * ctl->unit)
3323 offset += BITS_PER_BITMAP * ctl->unit;
3324 }
3325
3326 if (fatal_signal_pending(current)) {
3327 ret = -ERESTARTSYS;
3328 break;
3329 }
3330
3331 cond_resched();
3332 }
3333
3334 return ret;
3335}
3336
3337void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache)
3338{
3339 atomic_inc(&cache->trimming);
3340}
3341
3342void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *block_group)
3343{
3344 struct btrfs_fs_info *fs_info = block_group->fs_info;
3345 struct extent_map_tree *em_tree;
3346 struct extent_map *em;
3347 bool cleanup;
3348
3349 spin_lock(&block_group->lock);
3350 cleanup = (atomic_dec_and_test(&block_group->trimming) &&
3351 block_group->removed);
3352 spin_unlock(&block_group->lock);
3353
3354 if (cleanup) {
3355 mutex_lock(&fs_info->chunk_mutex);
3356 em_tree = &fs_info->mapping_tree.map_tree;
3357 write_lock(&em_tree->lock);
3358 em = lookup_extent_mapping(em_tree, block_group->key.objectid,
3359 1);
3360 BUG_ON(!em); /* logic error, can't happen */
3361 /*
3362 * remove_extent_mapping() will delete us from the pinned_chunks
3363 * list, which is protected by the chunk mutex.
3364 */
3365 remove_extent_mapping(em_tree, em);
3366 write_unlock(&em_tree->lock);
3367 mutex_unlock(&fs_info->chunk_mutex);
3368
3369 /* once for us and once for the tree */
3370 free_extent_map(em);
3371 free_extent_map(em);
3372
3373 /*
3374 * We've left one free space entry and other tasks trimming
3375 * this block group have left 1 entry each one. Free them.
3376 */
3377 __btrfs_remove_free_space_cache(block_group->free_space_ctl);
3378 }
3379}
3380
3381int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
3382 u64 *trimmed, u64 start, u64 end, u64 minlen)
3383{
3384 int ret;
3385
3386 *trimmed = 0;
3387
3388 spin_lock(&block_group->lock);
3389 if (block_group->removed) {
3390 spin_unlock(&block_group->lock);
3391 return 0;
3392 }
3393 btrfs_get_block_group_trimming(block_group);
3394 spin_unlock(&block_group->lock);
3395
3396 ret = trim_no_bitmap(block_group, trimmed, start, end, minlen);
3397 if (ret)
3398 goto out;
3399
3400 ret = trim_bitmaps(block_group, trimmed, start, end, minlen);
3401out:
3402 btrfs_put_block_group_trimming(block_group);
3403 return ret;
3404}
3405
3406/*
3407 * Find the left-most item in the cache tree, and then return the
3408 * smallest inode number in the item.
3409 *
3410 * Note: the returned inode number may not be the smallest one in
3411 * the tree, if the left-most item is a bitmap.
3412 */
3413u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
3414{
3415 struct btrfs_free_space_ctl *ctl = fs_root->free_ino_ctl;
3416 struct btrfs_free_space *entry = NULL;
3417 u64 ino = 0;
3418
3419 spin_lock(&ctl->tree_lock);
3420
3421 if (RB_EMPTY_ROOT(&ctl->free_space_offset))
3422 goto out;
3423
3424 entry = rb_entry(rb_first(&ctl->free_space_offset),
3425 struct btrfs_free_space, offset_index);
3426
3427 if (!entry->bitmap) {
3428 ino = entry->offset;
3429
3430 unlink_free_space(ctl, entry);
3431 entry->offset++;
3432 entry->bytes--;
3433 if (!entry->bytes)
3434 kmem_cache_free(btrfs_free_space_cachep, entry);
3435 else
3436 link_free_space(ctl, entry);
3437 } else {
3438 u64 offset = 0;
3439 u64 count = 1;
3440 int ret;
3441
3442 ret = search_bitmap(ctl, entry, &offset, &count, true);
3443 /* Logic error; Should be empty if it can't find anything */
3444 ASSERT(!ret);
3445
3446 ino = offset;
3447 bitmap_clear_bits(ctl, entry, offset, 1);
3448 if (entry->bytes == 0)
3449 free_bitmap(ctl, entry);
3450 }
3451out:
3452 spin_unlock(&ctl->tree_lock);
3453
3454 return ino;
3455}
3456
3457struct inode *lookup_free_ino_inode(struct btrfs_root *root,
3458 struct btrfs_path *path)
3459{
3460 struct inode *inode = NULL;
3461
3462 spin_lock(&root->ino_cache_lock);
3463 if (root->ino_cache_inode)
3464 inode = igrab(root->ino_cache_inode);
3465 spin_unlock(&root->ino_cache_lock);
3466 if (inode)
3467 return inode;
3468
3469 inode = __lookup_free_space_inode(root, path, 0);
3470 if (IS_ERR(inode))
3471 return inode;
3472
3473 spin_lock(&root->ino_cache_lock);
3474 if (!btrfs_fs_closing(root->fs_info))
3475 root->ino_cache_inode = igrab(inode);
3476 spin_unlock(&root->ino_cache_lock);
3477
3478 return inode;
3479}
3480
3481int create_free_ino_inode(struct btrfs_root *root,
3482 struct btrfs_trans_handle *trans,
3483 struct btrfs_path *path)
3484{
3485 return __create_free_space_inode(root, trans, path,
3486 BTRFS_FREE_INO_OBJECTID, 0);
3487}
3488
3489int load_free_ino_cache(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
3490{
3491 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3492 struct btrfs_path *path;
3493 struct inode *inode;
3494 int ret = 0;
3495 u64 root_gen = btrfs_root_generation(&root->root_item);
3496
3497 if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
3498 return 0;
3499
3500 /*
3501 * If we're unmounting then just return, since this does a search on the
3502 * normal root and not the commit root and we could deadlock.
3503 */
3504 if (btrfs_fs_closing(fs_info))
3505 return 0;
3506
3507 path = btrfs_alloc_path();
3508 if (!path)
3509 return 0;
3510
3511 inode = lookup_free_ino_inode(root, path);
3512 if (IS_ERR(inode))
3513 goto out;
3514
3515 if (root_gen != BTRFS_I(inode)->generation)
3516 goto out_put;
3517
3518 ret = __load_free_space_cache(root, inode, ctl, path, 0);
3519
3520 if (ret < 0)
3521 btrfs_err(fs_info,
3522 "failed to load free ino cache for root %llu",
3523 root->root_key.objectid);
3524out_put:
3525 iput(inode);
3526out:
3527 btrfs_free_path(path);
3528 return ret;
3529}
3530
3531int btrfs_write_out_ino_cache(struct btrfs_root *root,
3532 struct btrfs_trans_handle *trans,
3533 struct btrfs_path *path,
3534 struct inode *inode)
3535{
3536 struct btrfs_fs_info *fs_info = root->fs_info;
3537 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3538 int ret;
3539 struct btrfs_io_ctl io_ctl;
3540 bool release_metadata = true;
3541
3542 if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
3543 return 0;
3544
3545 memset(&io_ctl, 0, sizeof(io_ctl));
3546 ret = __btrfs_write_out_cache(root, inode, ctl, NULL, &io_ctl,
3547 trans, path, 0);
3548 if (!ret) {
3549 /*
3550 * At this point writepages() didn't error out, so our metadata
3551 * reservation is released when the writeback finishes, at
3552 * inode.c:btrfs_finish_ordered_io(), regardless of it finishing
3553 * with or without an error.
3554 */
3555 release_metadata = false;
3556 ret = btrfs_wait_cache_io_root(root, trans, &io_ctl, path);
3557 }
3558
3559 if (ret) {
3560 if (release_metadata)
3561 btrfs_delalloc_release_metadata(inode, inode->i_size);
3562#ifdef DEBUG
3563 btrfs_err(fs_info,
3564 "failed to write free ino cache for root %llu",
3565 root->root_key.objectid);
3566#endif
3567 }
3568
3569 return ret;
3570}
3571
3572#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
3573/*
3574 * Use this if you need to make a bitmap or extent entry specifically, it
3575 * doesn't do any of the merging that add_free_space does, this acts a lot like
3576 * how the free space cache loading stuff works, so you can get really weird
3577 * configurations.
3578 */
3579int test_add_free_space_entry(struct btrfs_block_group_cache *cache,
3580 u64 offset, u64 bytes, bool bitmap)
3581{
3582 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3583 struct btrfs_free_space *info = NULL, *bitmap_info;
3584 void *map = NULL;
3585 u64 bytes_added;
3586 int ret;
3587
3588again:
3589 if (!info) {
3590 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
3591 if (!info)
3592 return -ENOMEM;
3593 }
3594
3595 if (!bitmap) {
3596 spin_lock(&ctl->tree_lock);
3597 info->offset = offset;
3598 info->bytes = bytes;
3599 info->max_extent_size = 0;
3600 ret = link_free_space(ctl, info);
3601 spin_unlock(&ctl->tree_lock);
3602 if (ret)
3603 kmem_cache_free(btrfs_free_space_cachep, info);
3604 return ret;
3605 }
3606
3607 if (!map) {
3608 map = kzalloc(PAGE_SIZE, GFP_NOFS);
3609 if (!map) {
3610 kmem_cache_free(btrfs_free_space_cachep, info);
3611 return -ENOMEM;
3612 }
3613 }
3614
3615 spin_lock(&ctl->tree_lock);
3616 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3617 1, 0);
3618 if (!bitmap_info) {
3619 info->bitmap = map;
3620 map = NULL;
3621 add_new_bitmap(ctl, info, offset);
3622 bitmap_info = info;
3623 info = NULL;
3624 }
3625
3626 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
3627
3628 bytes -= bytes_added;
3629 offset += bytes_added;
3630 spin_unlock(&ctl->tree_lock);
3631
3632 if (bytes)
3633 goto again;
3634
3635 if (info)
3636 kmem_cache_free(btrfs_free_space_cachep, info);
3637 if (map)
3638 kfree(map);
3639 return 0;
3640}
3641
3642/*
3643 * Checks to see if the given range is in the free space cache. This is really
3644 * just used to check the absence of space, so if there is free space in the
3645 * range at all we will return 1.
3646 */
3647int test_check_exists(struct btrfs_block_group_cache *cache,
3648 u64 offset, u64 bytes)
3649{
3650 struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3651 struct btrfs_free_space *info;
3652 int ret = 0;
3653
3654 spin_lock(&ctl->tree_lock);
3655 info = tree_search_offset(ctl, offset, 0, 0);
3656 if (!info) {
3657 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3658 1, 0);
3659 if (!info)
3660 goto out;
3661 }
3662
3663have_info:
3664 if (info->bitmap) {
3665 u64 bit_off, bit_bytes;
3666 struct rb_node *n;
3667 struct btrfs_free_space *tmp;
3668
3669 bit_off = offset;
3670 bit_bytes = ctl->unit;
3671 ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
3672 if (!ret) {
3673 if (bit_off == offset) {
3674 ret = 1;
3675 goto out;
3676 } else if (bit_off > offset &&
3677 offset + bytes > bit_off) {
3678 ret = 1;
3679 goto out;
3680 }
3681 }
3682
3683 n = rb_prev(&info->offset_index);
3684 while (n) {
3685 tmp = rb_entry(n, struct btrfs_free_space,
3686 offset_index);
3687 if (tmp->offset + tmp->bytes < offset)
3688 break;
3689 if (offset + bytes < tmp->offset) {
3690 n = rb_prev(&tmp->offset_index);
3691 continue;
3692 }
3693 info = tmp;
3694 goto have_info;
3695 }
3696
3697 n = rb_next(&info->offset_index);
3698 while (n) {
3699 tmp = rb_entry(n, struct btrfs_free_space,
3700 offset_index);
3701 if (offset + bytes < tmp->offset)
3702 break;
3703 if (tmp->offset + tmp->bytes < offset) {
3704 n = rb_next(&tmp->offset_index);
3705 continue;
3706 }
3707 info = tmp;
3708 goto have_info;
3709 }
3710
3711 ret = 0;
3712 goto out;
3713 }
3714
3715 if (info->offset == offset) {
3716 ret = 1;
3717 goto out;
3718 }
3719
3720 if (offset > info->offset && offset < info->offset + info->bytes)
3721 ret = 1;
3722out:
3723 spin_unlock(&ctl->tree_lock);
3724 return ret;
3725}
3726#endif /* CONFIG_BTRFS_FS_RUN_SANITY_TESTS */