?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Kconfig: acpi: Fix typo in comment.
[linux-2.6/libata-dev.git] / fs / btrfs / free-space-cache.c
blob75a7b1147764e41869f3bf8e49c7a33f40836fe2
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
31 #define BITS_PER_BITMAP (PAGE_CACHE_SIZE * 8)
32 #define MAX_CACHE_BYTES_PER_GIG (32 * 1024)
33
34 static int link_free_space(struct btrfs_free_space_ctl *ctl,
35 struct btrfs_free_space *info);
36
37 static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
38 struct btrfs_path *path,
39 u64 offset)
40 {
41 struct btrfs_key key;
42 struct btrfs_key location;
43 struct btrfs_disk_key disk_key;
44 struct btrfs_free_space_header *header;
45 struct extent_buffer *leaf;
46 struct inode *inode = NULL;
47 int ret;
48
49 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
50 key.offset = offset;
51 key.type = 0;
52
53 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
54 if (ret < 0)
55 return ERR_PTR(ret);
56 if (ret > 0) {
57 btrfs_release_path(path);
58 return ERR_PTR(-ENOENT);
59 }
60
61 leaf = path->nodes[0];
62 header = btrfs_item_ptr(leaf, path->slots[0],
63 struct btrfs_free_space_header);
64 btrfs_free_space_key(leaf, header, &disk_key);
65 btrfs_disk_key_to_cpu(&location, &disk_key);
66 btrfs_release_path(path);
67
68 inode = btrfs_iget(root->fs_info->sb, &location, root, NULL);
69 if (!inode)
70 return ERR_PTR(-ENOENT);
71 if (IS_ERR(inode))
72 return inode;
73 if (is_bad_inode(inode)) {
74 iput(inode);
75 return ERR_PTR(-ENOENT);
76 }
77
78 inode->i_mapping->flags &= ~__GFP_FS;
79
80 return inode;
81 }
82
83 struct inode *lookup_free_space_inode(struct btrfs_root *root,
84 struct btrfs_block_group_cache
85 *block_group, struct btrfs_path *path)
86 {
87 struct inode *inode = NULL;
88 u32 flags = BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
89
90 spin_lock(&block_group->lock);
91 if (block_group->inode)
92 inode = igrab(block_group->inode);
93 spin_unlock(&block_group->lock);
94 if (inode)
95 return inode;
96
97 inode = __lookup_free_space_inode(root, path,
98 block_group->key.objectid);
99 if (IS_ERR(inode))
100 return inode;
101
102 spin_lock(&block_group->lock);
103 if (!((BTRFS_I(inode)->flags & flags) == flags)) {
104 printk(KERN_INFO "Old style space inode found, converting.\n");
105 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM |
106 BTRFS_INODE_NODATACOW;
107 block_group->disk_cache_state = BTRFS_DC_CLEAR;
108 }
109
110 if (!block_group->iref) {
111 block_group->inode = igrab(inode);
112 block_group->iref = 1;
113 }
114 spin_unlock(&block_group->lock);
115
116 return inode;
117 }
118
119 int __create_free_space_inode(struct btrfs_root *root,
120 struct btrfs_trans_handle *trans,
121 struct btrfs_path *path, u64 ino, u64 offset)
122 {
123 struct btrfs_key key;
124 struct btrfs_disk_key disk_key;
125 struct btrfs_free_space_header *header;
126 struct btrfs_inode_item *inode_item;
127 struct extent_buffer *leaf;
128 u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC;
129 int ret;
130
131 ret = btrfs_insert_empty_inode(trans, root, path, ino);
132 if (ret)
133 return ret;
134
135 /* We inline crc's for the free disk space cache */
136 if (ino != BTRFS_FREE_INO_OBJECTID)
137 flags |= BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
138
139 leaf = path->nodes[0];
140 inode_item = btrfs_item_ptr(leaf, path->slots[0],
141 struct btrfs_inode_item);
142 btrfs_item_key(leaf, &disk_key, path->slots[0]);
143 memset_extent_buffer(leaf, 0, (unsigned long)inode_item,
144 sizeof(*inode_item));
145 btrfs_set_inode_generation(leaf, inode_item, trans->transid);
146 btrfs_set_inode_size(leaf, inode_item, 0);
147 btrfs_set_inode_nbytes(leaf, inode_item, 0);
148 btrfs_set_inode_uid(leaf, inode_item, 0);
149 btrfs_set_inode_gid(leaf, inode_item, 0);
150 btrfs_set_inode_mode(leaf, inode_item, S_IFREG | 0600);
151 btrfs_set_inode_flags(leaf, inode_item, flags);
152 btrfs_set_inode_nlink(leaf, inode_item, 1);
153 btrfs_set_inode_transid(leaf, inode_item, trans->transid);
154 btrfs_set_inode_block_group(leaf, inode_item, offset);
155 btrfs_mark_buffer_dirty(leaf);
156 btrfs_release_path(path);
157
158 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
159 key.offset = offset;
160 key.type = 0;
161
162 ret = btrfs_insert_empty_item(trans, root, path, &key,
163 sizeof(struct btrfs_free_space_header));
164 if (ret < 0) {
165 btrfs_release_path(path);
166 return ret;
167 }
168 leaf = path->nodes[0];
169 header = btrfs_item_ptr(leaf, path->slots[0],
170 struct btrfs_free_space_header);
171 memset_extent_buffer(leaf, 0, (unsigned long)header, sizeof(*header));
172 btrfs_set_free_space_key(leaf, header, &disk_key);
173 btrfs_mark_buffer_dirty(leaf);
174 btrfs_release_path(path);
175
176 return 0;
177 }
178
179 int create_free_space_inode(struct btrfs_root *root,
180 struct btrfs_trans_handle *trans,
181 struct btrfs_block_group_cache *block_group,
182 struct btrfs_path *path)
183 {
184 int ret;
185 u64 ino;
186
187 ret = btrfs_find_free_objectid(root, &ino);
188 if (ret < 0)
189 return ret;
190
191 return __create_free_space_inode(root, trans, path, ino,
192 block_group->key.objectid);
193 }
194
195 int btrfs_truncate_free_space_cache(struct btrfs_root *root,
196 struct btrfs_trans_handle *trans,
197 struct btrfs_path *path,
198 struct inode *inode)
199 {
200 struct btrfs_block_rsv *rsv;
201 u64 needed_bytes;
202 loff_t oldsize;
203 int ret = 0;
204
205 rsv = trans->block_rsv;
206 trans->block_rsv = &root->fs_info->global_block_rsv;
207
208 /* 1 for slack space, 1 for updating the inode */
209 needed_bytes = btrfs_calc_trunc_metadata_size(root, 1) +
210 btrfs_calc_trans_metadata_size(root, 1);
211
212 spin_lock(&trans->block_rsv->lock);
213 if (trans->block_rsv->reserved < needed_bytes) {
214 spin_unlock(&trans->block_rsv->lock);
215 trans->block_rsv = rsv;
216 return -ENOSPC;
217 }
218 spin_unlock(&trans->block_rsv->lock);
219
220 oldsize = i_size_read(inode);
221 btrfs_i_size_write(inode, 0);
222 truncate_pagecache(inode, oldsize, 0);
223
224 /*
225 * We don't need an orphan item because truncating the free space cache
226 * will never be split across transactions.
227 */
228 ret = btrfs_truncate_inode_items(trans, root, inode,
229 0, BTRFS_EXTENT_DATA_KEY);
230
231 if (ret) {
232 trans->block_rsv = rsv;
233 WARN_ON(1);
234 return ret;
235 }
236
237 ret = btrfs_update_inode(trans, root, inode);
238 trans->block_rsv = rsv;
239
240 return ret;
241 }
242
243 static int readahead_cache(struct inode *inode)
244 {
245 struct file_ra_state *ra;
246 unsigned long last_index;
247
248 ra = kzalloc(sizeof(*ra), GFP_NOFS);
249 if (!ra)
250 return -ENOMEM;
251
252 file_ra_state_init(ra, inode->i_mapping);
253 last_index = (i_size_read(inode) - 1) >> PAGE_CACHE_SHIFT;
254
255 page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
256
257 kfree(ra);
258
259 return 0;
260 }
261
262 struct io_ctl {
263 void *cur, *orig;
264 struct page *page;
265 struct page **pages;
266 struct btrfs_root *root;
267 unsigned long size;
268 int index;
269 int num_pages;
270 unsigned check_crcs:1;
271 };
272
273 static int io_ctl_init(struct io_ctl *io_ctl, struct inode *inode,
274 struct btrfs_root *root)
275 {
276 memset(io_ctl, 0, sizeof(struct io_ctl));
277 io_ctl->num_pages = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
278 PAGE_CACHE_SHIFT;
279 io_ctl->pages = kzalloc(sizeof(struct page *) * io_ctl->num_pages,
280 GFP_NOFS);
281 if (!io_ctl->pages)
282 return -ENOMEM;
283 io_ctl->root = root;
284 if (btrfs_ino(inode) != BTRFS_FREE_INO_OBJECTID)
285 io_ctl->check_crcs = 1;
286 return 0;
287 }
288
289 static void io_ctl_free(struct io_ctl *io_ctl)
290 {
291 kfree(io_ctl->pages);
292 }
293
294 static void io_ctl_unmap_page(struct io_ctl *io_ctl)
295 {
296 if (io_ctl->cur) {
297 kunmap(io_ctl->page);
298 io_ctl->cur = NULL;
299 io_ctl->orig = NULL;
300 }
301 }
302
303 static void io_ctl_map_page(struct io_ctl *io_ctl, int clear)
304 {
305 WARN_ON(io_ctl->cur);
306 BUG_ON(io_ctl->index >= io_ctl->num_pages);
307 io_ctl->page = io_ctl->pages[io_ctl->index++];
308 io_ctl->cur = kmap(io_ctl->page);
309 io_ctl->orig = io_ctl->cur;
310 io_ctl->size = PAGE_CACHE_SIZE;
311 if (clear)
312 memset(io_ctl->cur, 0, PAGE_CACHE_SIZE);
313 }
314
315 static void io_ctl_drop_pages(struct io_ctl *io_ctl)
316 {
317 int i;
318
319 io_ctl_unmap_page(io_ctl);
320
321 for (i = 0; i < io_ctl->num_pages; i++) {
322 ClearPageChecked(io_ctl->pages[i]);
323 unlock_page(io_ctl->pages[i]);
324 page_cache_release(io_ctl->pages[i]);
325 }
326 }
327
328 static int io_ctl_prepare_pages(struct io_ctl *io_ctl, struct inode *inode,
329 int uptodate)
330 {
331 struct page *page;
332 gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
333 int i;
334
335 for (i = 0; i < io_ctl->num_pages; i++) {
336 page = find_or_create_page(inode->i_mapping, i, mask);
337 if (!page) {
338 io_ctl_drop_pages(io_ctl);
339 return -ENOMEM;
340 }
341 io_ctl->pages[i] = page;
342 if (uptodate && !PageUptodate(page)) {
343 btrfs_readpage(NULL, page);
344 lock_page(page);
345 if (!PageUptodate(page)) {
346 printk(KERN_ERR "btrfs: error reading free "
347 "space cache\n");
348 io_ctl_drop_pages(io_ctl);
349 return -EIO;
350 }
351 }
352 }
353
354 return 0;
355 }
356
357 static void io_ctl_set_generation(struct io_ctl *io_ctl, u64 generation)
358 {
359 u64 *val;
360
361 io_ctl_map_page(io_ctl, 1);
362
363 /*
364 * Skip the csum areas. If we don't check crcs then we just have a
365 * 64bit chunk at the front of the first page.
366 */
367 if (io_ctl->check_crcs) {
368 io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
369 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
370 } else {
371 io_ctl->cur += sizeof(u64);
372 io_ctl->size -= sizeof(u64) * 2;
373 }
374
375 val = io_ctl->cur;
376 *val = cpu_to_le64(generation);
377 io_ctl->cur += sizeof(u64);
378 }
379
380 static int io_ctl_check_generation(struct io_ctl *io_ctl, u64 generation)
381 {
382 u64 *gen;
383
384 /*
385 * Skip the crc area. If we don't check crcs then we just have a 64bit
386 * chunk at the front of the first page.
387 */
388 if (io_ctl->check_crcs) {
389 io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
390 io_ctl->size -= sizeof(u64) +
391 (sizeof(u32) * io_ctl->num_pages);
392 } else {
393 io_ctl->cur += sizeof(u64);
394 io_ctl->size -= sizeof(u64) * 2;
395 }
396
397 gen = io_ctl->cur;
398 if (le64_to_cpu(*gen) != generation) {
399 printk_ratelimited(KERN_ERR "btrfs: space cache generation "
400 "(%Lu) does not match inode (%Lu)\n", *gen,
401 generation);
402 io_ctl_unmap_page(io_ctl);
403 return -EIO;
404 }
405 io_ctl->cur += sizeof(u64);
406 return 0;
407 }
408
409 static void io_ctl_set_crc(struct io_ctl *io_ctl, int index)
410 {
411 u32 *tmp;
412 u32 crc = ~(u32)0;
413 unsigned offset = 0;
414
415 if (!io_ctl->check_crcs) {
416 io_ctl_unmap_page(io_ctl);
417 return;
418 }
419
420 if (index == 0)
421 offset = sizeof(u32) * io_ctl->num_pages;
422
423 crc = btrfs_csum_data(io_ctl->root, io_ctl->orig + offset, crc,
424 PAGE_CACHE_SIZE - offset);
425 btrfs_csum_final(crc, (char *)&crc);
426 io_ctl_unmap_page(io_ctl);
427 tmp = kmap(io_ctl->pages[0]);
428 tmp += index;
429 *tmp = crc;
430 kunmap(io_ctl->pages[0]);
431 }
432
433 static int io_ctl_check_crc(struct io_ctl *io_ctl, int index)
434 {
435 u32 *tmp, val;
436 u32 crc = ~(u32)0;
437 unsigned offset = 0;
438
439 if (!io_ctl->check_crcs) {
440 io_ctl_map_page(io_ctl, 0);
441 return 0;
442 }
443
444 if (index == 0)
445 offset = sizeof(u32) * io_ctl->num_pages;
446
447 tmp = kmap(io_ctl->pages[0]);
448 tmp += index;
449 val = *tmp;
450 kunmap(io_ctl->pages[0]);
451
452 io_ctl_map_page(io_ctl, 0);
453 crc = btrfs_csum_data(io_ctl->root, io_ctl->orig + offset, crc,
454 PAGE_CACHE_SIZE - offset);
455 btrfs_csum_final(crc, (char *)&crc);
456 if (val != crc) {
457 printk_ratelimited(KERN_ERR "btrfs: csum mismatch on free "
458 "space cache\n");
459 io_ctl_unmap_page(io_ctl);
460 return -EIO;
461 }
462
463 return 0;
464 }
465
466 static int io_ctl_add_entry(struct io_ctl *io_ctl, u64 offset, u64 bytes,
467 void *bitmap)
468 {
469 struct btrfs_free_space_entry *entry;
470
471 if (!io_ctl->cur)
472 return -ENOSPC;
473
474 entry = io_ctl->cur;
475 entry->offset = cpu_to_le64(offset);
476 entry->bytes = cpu_to_le64(bytes);
477 entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
478 BTRFS_FREE_SPACE_EXTENT;
479 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
480 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
481
482 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
483 return 0;
484
485 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
486
487 /* No more pages to map */
488 if (io_ctl->index >= io_ctl->num_pages)
489 return 0;
490
491 /* map the next page */
492 io_ctl_map_page(io_ctl, 1);
493 return 0;
494 }
495
496 static int io_ctl_add_bitmap(struct io_ctl *io_ctl, void *bitmap)
497 {
498 if (!io_ctl->cur)
499 return -ENOSPC;
500
501 /*
502 * If we aren't at the start of the current page, unmap this one and
503 * map the next one if there is any left.
504 */
505 if (io_ctl->cur != io_ctl->orig) {
506 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
507 if (io_ctl->index >= io_ctl->num_pages)
508 return -ENOSPC;
509 io_ctl_map_page(io_ctl, 0);
510 }
511
512 memcpy(io_ctl->cur, bitmap, PAGE_CACHE_SIZE);
513 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
514 if (io_ctl->index < io_ctl->num_pages)
515 io_ctl_map_page(io_ctl, 0);
516 return 0;
517 }
518
519 static void io_ctl_zero_remaining_pages(struct io_ctl *io_ctl)
520 {
521 /*
522 * If we're not on the boundary we know we've modified the page and we
523 * need to crc the page.
524 */
525 if (io_ctl->cur != io_ctl->orig)
526 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
527 else
528 io_ctl_unmap_page(io_ctl);
529
530 while (io_ctl->index < io_ctl->num_pages) {
531 io_ctl_map_page(io_ctl, 1);
532 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
533 }
534 }
535
536 static int io_ctl_read_entry(struct io_ctl *io_ctl,
537 struct btrfs_free_space *entry, u8 *type)
538 {
539 struct btrfs_free_space_entry *e;
540 int ret;
541
542 if (!io_ctl->cur) {
543 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
544 if (ret)
545 return ret;
546 }
547
548 e = io_ctl->cur;
549 entry->offset = le64_to_cpu(e->offset);
550 entry->bytes = le64_to_cpu(e->bytes);
551 *type = e->type;
552 io_ctl->cur += sizeof(struct btrfs_free_space_entry);
553 io_ctl->size -= sizeof(struct btrfs_free_space_entry);
554
555 if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
556 return 0;
557
558 io_ctl_unmap_page(io_ctl);
559
560 return 0;
561 }
562
563 static int io_ctl_read_bitmap(struct io_ctl *io_ctl,
564 struct btrfs_free_space *entry)
565 {
566 int ret;
567
568 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
569 if (ret)
570 return ret;
571
572 memcpy(entry->bitmap, io_ctl->cur, PAGE_CACHE_SIZE);
573 io_ctl_unmap_page(io_ctl);
574
575 return 0;
576 }
577
578 int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
579 struct btrfs_free_space_ctl *ctl,
580 struct btrfs_path *path, u64 offset)
581 {
582 struct btrfs_free_space_header *header;
583 struct extent_buffer *leaf;
584 struct io_ctl io_ctl;
585 struct btrfs_key key;
586 struct btrfs_free_space *e, *n;
587 struct list_head bitmaps;
588 u64 num_entries;
589 u64 num_bitmaps;
590 u64 generation;
591 u8 type;
592 int ret = 0;
593
594 INIT_LIST_HEAD(&bitmaps);
595
596 /* Nothing in the space cache, goodbye */
597 if (!i_size_read(inode))
598 return 0;
599
600 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
601 key.offset = offset;
602 key.type = 0;
603
604 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
605 if (ret < 0)
606 return 0;
607 else if (ret > 0) {
608 btrfs_release_path(path);
609 return 0;
610 }
611
612 ret = -1;
613
614 leaf = path->nodes[0];
615 header = btrfs_item_ptr(leaf, path->slots[0],
616 struct btrfs_free_space_header);
617 num_entries = btrfs_free_space_entries(leaf, header);
618 num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
619 generation = btrfs_free_space_generation(leaf, header);
620 btrfs_release_path(path);
621
622 if (BTRFS_I(inode)->generation != generation) {
623 printk(KERN_ERR "btrfs: free space inode generation (%llu) did"
624 " not match free space cache generation (%llu)\n",
625 (unsigned long long)BTRFS_I(inode)->generation,
626 (unsigned long long)generation);
627 return 0;
628 }
629
630 if (!num_entries)
631 return 0;
632
633 io_ctl_init(&io_ctl, inode, root);
634 ret = readahead_cache(inode);
635 if (ret)
636 goto out;
637
638 ret = io_ctl_prepare_pages(&io_ctl, inode, 1);
639 if (ret)
640 goto out;
641
642 ret = io_ctl_check_crc(&io_ctl, 0);
643 if (ret)
644 goto free_cache;
645
646 ret = io_ctl_check_generation(&io_ctl, generation);
647 if (ret)
648 goto free_cache;
649
650 while (num_entries) {
651 e = kmem_cache_zalloc(btrfs_free_space_cachep,
652 GFP_NOFS);
653 if (!e)
654 goto free_cache;
655
656 ret = io_ctl_read_entry(&io_ctl, e, &type);
657 if (ret) {
658 kmem_cache_free(btrfs_free_space_cachep, e);
659 goto free_cache;
660 }
661
662 if (!e->bytes) {
663 kmem_cache_free(btrfs_free_space_cachep, e);
664 goto free_cache;
665 }
666
667 if (type == BTRFS_FREE_SPACE_EXTENT) {
668 spin_lock(&ctl->tree_lock);
669 ret = link_free_space(ctl, e);
670 spin_unlock(&ctl->tree_lock);
671 if (ret) {
672 printk(KERN_ERR "Duplicate entries in "
673 "free space cache, dumping\n");
674 kmem_cache_free(btrfs_free_space_cachep, e);
675 goto free_cache;
676 }
677 } else {
678 BUG_ON(!num_bitmaps);
679 num_bitmaps--;
680 e->bitmap = kzalloc(PAGE_CACHE_SIZE, GFP_NOFS);
681 if (!e->bitmap) {
682 kmem_cache_free(
683 btrfs_free_space_cachep, e);
684 goto free_cache;
685 }
686 spin_lock(&ctl->tree_lock);
687 ret = link_free_space(ctl, e);
688 ctl->total_bitmaps++;
689 ctl->op->recalc_thresholds(ctl);
690 spin_unlock(&ctl->tree_lock);
691 if (ret) {
692 printk(KERN_ERR "Duplicate entries in "
693 "free space cache, dumping\n");
694 kmem_cache_free(btrfs_free_space_cachep, e);
695 goto free_cache;
696 }
697 list_add_tail(&e->list, &bitmaps);
698 }
699
700 num_entries--;
701 }
702
703 io_ctl_unmap_page(&io_ctl);
704
705 /*
706 * We add the bitmaps at the end of the entries in order that
707 * the bitmap entries are added to the cache.
708 */
709 list_for_each_entry_safe(e, n, &bitmaps, list) {
710 list_del_init(&e->list);
711 ret = io_ctl_read_bitmap(&io_ctl, e);
712 if (ret)
713 goto free_cache;
714 }
715
716 io_ctl_drop_pages(&io_ctl);
717 ret = 1;
718 out:
719 io_ctl_free(&io_ctl);
720 return ret;
721 free_cache:
722 io_ctl_drop_pages(&io_ctl);
723 __btrfs_remove_free_space_cache(ctl);
724 goto out;
725 }
726
727 int load_free_space_cache(struct btrfs_fs_info *fs_info,
728 struct btrfs_block_group_cache *block_group)
729 {
730 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
731 struct btrfs_root *root = fs_info->tree_root;
732 struct inode *inode;
733 struct btrfs_path *path;
734 int ret = 0;
735 bool matched;
736 u64 used = btrfs_block_group_used(&block_group->item);
737
738 /*
739 * If we're unmounting then just return, since this does a search on the
740 * normal root and not the commit root and we could deadlock.
741 */
742 if (btrfs_fs_closing(fs_info))
743 return 0;
744
745 /*
746 * If this block group has been marked to be cleared for one reason or
747 * another then we can't trust the on disk cache, so just return.
748 */
749 spin_lock(&block_group->lock);
750 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
751 spin_unlock(&block_group->lock);
752 return 0;
753 }
754 spin_unlock(&block_group->lock);
755
756 path = btrfs_alloc_path();
757 if (!path)
758 return 0;
759
760 inode = lookup_free_space_inode(root, block_group, path);
761 if (IS_ERR(inode)) {
762 btrfs_free_path(path);
763 return 0;
764 }
765
766 /* We may have converted the inode and made the cache invalid. */
767 spin_lock(&block_group->lock);
768 if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
769 spin_unlock(&block_group->lock);
770 goto out;
771 }
772 spin_unlock(&block_group->lock);
773
774 ret = __load_free_space_cache(fs_info->tree_root, inode, ctl,
775 path, block_group->key.objectid);
776 btrfs_free_path(path);
777 if (ret <= 0)
778 goto out;
779
780 spin_lock(&ctl->tree_lock);
781 matched = (ctl->free_space == (block_group->key.offset - used -
782 block_group->bytes_super));
783 spin_unlock(&ctl->tree_lock);
784
785 if (!matched) {
786 __btrfs_remove_free_space_cache(ctl);
787 printk(KERN_ERR "block group %llu has an wrong amount of free "
788 "space\n", block_group->key.objectid);
789 ret = -1;
790 }
791 out:
792 if (ret < 0) {
793 /* This cache is bogus, make sure it gets cleared */
794 spin_lock(&block_group->lock);
795 block_group->disk_cache_state = BTRFS_DC_CLEAR;
796 spin_unlock(&block_group->lock);
797 ret = 0;
798
799 printk(KERN_ERR "btrfs: failed to load free space cache "
800 "for block group %llu\n", block_group->key.objectid);
801 }
802
803 iput(inode);
804 return ret;
805 }
806
807 /**
808 * __btrfs_write_out_cache - write out cached info to an inode
809 * @root - the root the inode belongs to
810 * @ctl - the free space cache we are going to write out
811 * @block_group - the block_group for this cache if it belongs to a block_group
812 * @trans - the trans handle
813 * @path - the path to use
814 * @offset - the offset for the key we'll insert
815 *
816 * This function writes out a free space cache struct to disk for quick recovery
817 * on mount. This will return 0 if it was successfull in writing the cache out,
818 * and -1 if it was not.
819 */
820 int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
821 struct btrfs_free_space_ctl *ctl,
822 struct btrfs_block_group_cache *block_group,
823 struct btrfs_trans_handle *trans,
824 struct btrfs_path *path, u64 offset)
825 {
826 struct btrfs_free_space_header *header;
827 struct extent_buffer *leaf;
828 struct rb_node *node;
829 struct list_head *pos, *n;
830 struct extent_state *cached_state = NULL;
831 struct btrfs_free_cluster *cluster = NULL;
832 struct extent_io_tree *unpin = NULL;
833 struct io_ctl io_ctl;
834 struct list_head bitmap_list;
835 struct btrfs_key key;
836 u64 start, end, len;
837 int entries = 0;
838 int bitmaps = 0;
839 int ret;
840 int err = -1;
841
842 INIT_LIST_HEAD(&bitmap_list);
843
844 if (!i_size_read(inode))
845 return -1;
846
847 io_ctl_init(&io_ctl, inode, root);
848
849 /* Get the cluster for this block_group if it exists */
850 if (block_group && !list_empty(&block_group->cluster_list))
851 cluster = list_entry(block_group->cluster_list.next,
852 struct btrfs_free_cluster,
853 block_group_list);
854
855 /*
856 * We shouldn't have switched the pinned extents yet so this is the
857 * right one
858 */
859 unpin = root->fs_info->pinned_extents;
860
861 /* Lock all pages first so we can lock the extent safely. */
862 io_ctl_prepare_pages(&io_ctl, inode, 0);
863
864 lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
865 0, &cached_state, GFP_NOFS);
866
867 /*
868 * When searching for pinned extents, we need to start at our start
869 * offset.
870 */
871 if (block_group)
872 start = block_group->key.objectid;
873
874 node = rb_first(&ctl->free_space_offset);
875 if (!node && cluster) {
876 node = rb_first(&cluster->root);
877 cluster = NULL;
878 }
879
880 /* Make sure we can fit our crcs into the first page */
881 if (io_ctl.check_crcs &&
882 (io_ctl.num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE) {
883 WARN_ON(1);
884 goto out_nospc;
885 }
886
887 io_ctl_set_generation(&io_ctl, trans->transid);
888
889 /* Write out the extent entries */
890 while (node) {
891 struct btrfs_free_space *e;
892
893 e = rb_entry(node, struct btrfs_free_space, offset_index);
894 entries++;
895
896 ret = io_ctl_add_entry(&io_ctl, e->offset, e->bytes,
897 e->bitmap);
898 if (ret)
899 goto out_nospc;
900
901 if (e->bitmap) {
902 list_add_tail(&e->list, &bitmap_list);
903 bitmaps++;
904 }
905 node = rb_next(node);
906 if (!node && cluster) {
907 node = rb_first(&cluster->root);
908 cluster = NULL;
909 }
910 }
911
912 /*
913 * We want to add any pinned extents to our free space cache
914 * so we don't leak the space
915 */
916 while (block_group && (start < block_group->key.objectid +
917 block_group->key.offset)) {
918 ret = find_first_extent_bit(unpin, start, &start, &end,
919 EXTENT_DIRTY);
920 if (ret) {
921 ret = 0;
922 break;
923 }
924
925 /* This pinned extent is out of our range */
926 if (start >= block_group->key.objectid +
927 block_group->key.offset)
928 break;
929
930 len = block_group->key.objectid +
931 block_group->key.offset - start;
932 len = min(len, end + 1 - start);
933
934 entries++;
935 ret = io_ctl_add_entry(&io_ctl, start, len, NULL);
936 if (ret)
937 goto out_nospc;
938
939 start = end + 1;
940 }
941
942 /* Write out the bitmaps */
943 list_for_each_safe(pos, n, &bitmap_list) {
944 struct btrfs_free_space *entry =
945 list_entry(pos, struct btrfs_free_space, list);
946
947 ret = io_ctl_add_bitmap(&io_ctl, entry->bitmap);
948 if (ret)
949 goto out_nospc;
950 list_del_init(&entry->list);
951 }
952
953 /* Zero out the rest of the pages just to make sure */
954 io_ctl_zero_remaining_pages(&io_ctl);
955
956 ret = btrfs_dirty_pages(root, inode, io_ctl.pages, io_ctl.num_pages,
957 0, i_size_read(inode), &cached_state);
958 io_ctl_drop_pages(&io_ctl);
959 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
960 i_size_read(inode) - 1, &cached_state, GFP_NOFS);
961
962 if (ret)
963 goto out;
964
965
966 ret = filemap_write_and_wait(inode->i_mapping);
967 if (ret)
968 goto out;
969
970 key.objectid = BTRFS_FREE_SPACE_OBJECTID;
971 key.offset = offset;
972 key.type = 0;
973
974 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
975 if (ret < 0) {
976 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
977 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
978 GFP_NOFS);
979 goto out;
980 }
981 leaf = path->nodes[0];
982 if (ret > 0) {
983 struct btrfs_key found_key;
984 BUG_ON(!path->slots[0]);
985 path->slots[0]--;
986 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
987 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
988 found_key.offset != offset) {
989 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
990 inode->i_size - 1,
991 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0,
992 NULL, GFP_NOFS);
993 btrfs_release_path(path);
994 goto out;
995 }
996 }
997
998 BTRFS_I(inode)->generation = trans->transid;
999 header = btrfs_item_ptr(leaf, path->slots[0],
1000 struct btrfs_free_space_header);
1001 btrfs_set_free_space_entries(leaf, header, entries);
1002 btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1003 btrfs_set_free_space_generation(leaf, header, trans->transid);
1004 btrfs_mark_buffer_dirty(leaf);
1005 btrfs_release_path(path);
1006
1007 err = 0;
1008 out:
1009 io_ctl_free(&io_ctl);
1010 if (err) {
1011 invalidate_inode_pages2(inode->i_mapping);
1012 BTRFS_I(inode)->generation = 0;
1013 }
1014 btrfs_update_inode(trans, root, inode);
1015 return err;
1016
1017 out_nospc:
1018 list_for_each_safe(pos, n, &bitmap_list) {
1019 struct btrfs_free_space *entry =
1020 list_entry(pos, struct btrfs_free_space, list);
1021 list_del_init(&entry->list);
1022 }
1023 io_ctl_drop_pages(&io_ctl);
1024 unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1025 i_size_read(inode) - 1, &cached_state, GFP_NOFS);
1026 goto out;
1027 }
1028
1029 int btrfs_write_out_cache(struct btrfs_root *root,
1030 struct btrfs_trans_handle *trans,
1031 struct btrfs_block_group_cache *block_group,
1032 struct btrfs_path *path)
1033 {
1034 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1035 struct inode *inode;
1036 int ret = 0;
1037
1038 root = root->fs_info->tree_root;
1039
1040 spin_lock(&block_group->lock);
1041 if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1042 spin_unlock(&block_group->lock);
1043 return 0;
1044 }
1045 spin_unlock(&block_group->lock);
1046
1047 inode = lookup_free_space_inode(root, block_group, path);
1048 if (IS_ERR(inode))
1049 return 0;
1050
1051 ret = __btrfs_write_out_cache(root, inode, ctl, block_group, trans,
1052 path, block_group->key.objectid);
1053 if (ret) {
1054 spin_lock(&block_group->lock);
1055 block_group->disk_cache_state = BTRFS_DC_ERROR;
1056 spin_unlock(&block_group->lock);
1057 ret = 0;
1058 #ifdef DEBUG
1059 printk(KERN_ERR "btrfs: failed to write free space cace "
1060 "for block group %llu\n", block_group->key.objectid);
1061 #endif
1062 }
1063
1064 iput(inode);
1065 return ret;
1066 }
1067
1068 static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1069 u64 offset)
1070 {
1071 BUG_ON(offset < bitmap_start);
1072 offset -= bitmap_start;
1073 return (unsigned long)(div_u64(offset, unit));
1074 }
1075
1076 static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1077 {
1078 return (unsigned long)(div_u64(bytes, unit));
1079 }
1080
1081 static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1082 u64 offset)
1083 {
1084 u64 bitmap_start;
1085 u64 bytes_per_bitmap;
1086
1087 bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1088 bitmap_start = offset - ctl->start;
1089 bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1090 bitmap_start *= bytes_per_bitmap;
1091 bitmap_start += ctl->start;
1092
1093 return bitmap_start;
1094 }
1095
1096 static int tree_insert_offset(struct rb_root *root, u64 offset,
1097 struct rb_node *node, int bitmap)
1098 {
1099 struct rb_node **p = &root->rb_node;
1100 struct rb_node *parent = NULL;
1101 struct btrfs_free_space *info;
1102
1103 while (*p) {
1104 parent = *p;
1105 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1106
1107 if (offset < info->offset) {
1108 p = &(*p)->rb_left;
1109 } else if (offset > info->offset) {
1110 p = &(*p)->rb_right;
1111 } else {
1112 /*
1113 * we could have a bitmap entry and an extent entry
1114 * share the same offset. If this is the case, we want
1115 * the extent entry to always be found first if we do a
1116 * linear search through the tree, since we want to have
1117 * the quickest allocation time, and allocating from an
1118 * extent is faster than allocating from a bitmap. So
1119 * if we're inserting a bitmap and we find an entry at
1120 * this offset, we want to go right, or after this entry
1121 * logically. If we are inserting an extent and we've
1122 * found a bitmap, we want to go left, or before
1123 * logically.
1124 */
1125 if (bitmap) {
1126 if (info->bitmap) {
1127 WARN_ON_ONCE(1);
1128 return -EEXIST;
1129 }
1130 p = &(*p)->rb_right;
1131 } else {
1132 if (!info->bitmap) {
1133 WARN_ON_ONCE(1);
1134 return -EEXIST;
1135 }
1136 p = &(*p)->rb_left;
1137 }
1138 }
1139 }
1140
1141 rb_link_node(node, parent, p);
1142 rb_insert_color(node, root);
1143
1144 return 0;
1145 }
1146
1147 /*
1148 * searches the tree for the given offset.
1149 *
1150 * fuzzy - If this is set, then we are trying to make an allocation, and we just
1151 * want a section that has at least bytes size and comes at or after the given
1152 * offset.
1153 */
1154 static struct btrfs_free_space *
1155 tree_search_offset(struct btrfs_free_space_ctl *ctl,
1156 u64 offset, int bitmap_only, int fuzzy)
1157 {
1158 struct rb_node *n = ctl->free_space_offset.rb_node;
1159 struct btrfs_free_space *entry, *prev = NULL;
1160
1161 /* find entry that is closest to the 'offset' */
1162 while (1) {
1163 if (!n) {
1164 entry = NULL;
1165 break;
1166 }
1167
1168 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1169 prev = entry;
1170
1171 if (offset < entry->offset)
1172 n = n->rb_left;
1173 else if (offset > entry->offset)
1174 n = n->rb_right;
1175 else
1176 break;
1177 }
1178
1179 if (bitmap_only) {
1180 if (!entry)
1181 return NULL;
1182 if (entry->bitmap)
1183 return entry;
1184
1185 /*
1186 * bitmap entry and extent entry may share same offset,
1187 * in that case, bitmap entry comes after extent entry.
1188 */
1189 n = rb_next(n);
1190 if (!n)
1191 return NULL;
1192 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1193 if (entry->offset != offset)
1194 return NULL;
1195
1196 WARN_ON(!entry->bitmap);
1197 return entry;
1198 } else if (entry) {
1199 if (entry->bitmap) {
1200 /*
1201 * if previous extent entry covers the offset,
1202 * we should return it instead of the bitmap entry
1203 */
1204 n = &entry->offset_index;
1205 while (1) {
1206 n = rb_prev(n);
1207 if (!n)
1208 break;
1209 prev = rb_entry(n, struct btrfs_free_space,
1210 offset_index);
1211 if (!prev->bitmap) {
1212 if (prev->offset + prev->bytes > offset)
1213 entry = prev;
1214 break;
1215 }
1216 }
1217 }
1218 return entry;
1219 }
1220
1221 if (!prev)
1222 return NULL;
1223
1224 /* find last entry before the 'offset' */
1225 entry = prev;
1226 if (entry->offset > offset) {
1227 n = rb_prev(&entry->offset_index);
1228 if (n) {
1229 entry = rb_entry(n, struct btrfs_free_space,
1230 offset_index);
1231 BUG_ON(entry->offset > offset);
1232 } else {
1233 if (fuzzy)
1234 return entry;
1235 else
1236 return NULL;
1237 }
1238 }
1239
1240 if (entry->bitmap) {
1241 n = &entry->offset_index;
1242 while (1) {
1243 n = rb_prev(n);
1244 if (!n)
1245 break;
1246 prev = rb_entry(n, struct btrfs_free_space,
1247 offset_index);
1248 if (!prev->bitmap) {
1249 if (prev->offset + prev->bytes > offset)
1250 return prev;
1251 break;
1252 }
1253 }
1254 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1255 return entry;
1256 } else if (entry->offset + entry->bytes > offset)
1257 return entry;
1258
1259 if (!fuzzy)
1260 return NULL;
1261
1262 while (1) {
1263 if (entry->bitmap) {
1264 if (entry->offset + BITS_PER_BITMAP *
1265 ctl->unit > offset)
1266 break;
1267 } else {
1268 if (entry->offset + entry->bytes > offset)
1269 break;
1270 }
1271
1272 n = rb_next(&entry->offset_index);
1273 if (!n)
1274 return NULL;
1275 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1276 }
1277 return entry;
1278 }
1279
1280 static inline void
1281 __unlink_free_space(struct btrfs_free_space_ctl *ctl,
1282 struct btrfs_free_space *info)
1283 {
1284 rb_erase(&info->offset_index, &ctl->free_space_offset);
1285 ctl->free_extents--;
1286 }
1287
1288 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1289 struct btrfs_free_space *info)
1290 {
1291 __unlink_free_space(ctl, info);
1292 ctl->free_space -= info->bytes;
1293 }
1294
1295 static int link_free_space(struct btrfs_free_space_ctl *ctl,
1296 struct btrfs_free_space *info)
1297 {
1298 int ret = 0;
1299
1300 BUG_ON(!info->bitmap && !info->bytes);
1301 ret = tree_insert_offset(&ctl->free_space_offset, info->offset,
1302 &info->offset_index, (info->bitmap != NULL));
1303 if (ret)
1304 return ret;
1305
1306 ctl->free_space += info->bytes;
1307 ctl->free_extents++;
1308 return ret;
1309 }
1310
1311 static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
1312 {
1313 struct btrfs_block_group_cache *block_group = ctl->private;
1314 u64 max_bytes;
1315 u64 bitmap_bytes;
1316 u64 extent_bytes;
1317 u64 size = block_group->key.offset;
1318 u64 bytes_per_bg = BITS_PER_BITMAP * block_group->sectorsize;
1319 int max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
1320
1321 BUG_ON(ctl->total_bitmaps > max_bitmaps);
1322
1323 /*
1324 * The goal is to keep the total amount of memory used per 1gb of space
1325 * at or below 32k, so we need to adjust how much memory we allow to be
1326 * used by extent based free space tracking
1327 */
1328 if (size < 1024 * 1024 * 1024)
1329 max_bytes = MAX_CACHE_BYTES_PER_GIG;
1330 else
1331 max_bytes = MAX_CACHE_BYTES_PER_GIG *
1332 div64_u64(size, 1024 * 1024 * 1024);
1333
1334 /*
1335 * we want to account for 1 more bitmap than what we have so we can make
1336 * sure we don't go over our overall goal of MAX_CACHE_BYTES_PER_GIG as
1337 * we add more bitmaps.
1338 */
1339 bitmap_bytes = (ctl->total_bitmaps + 1) * PAGE_CACHE_SIZE;
1340
1341 if (bitmap_bytes >= max_bytes) {
1342 ctl->extents_thresh = 0;
1343 return;
1344 }
1345
1346 /*
1347 * we want the extent entry threshold to always be at most 1/2 the maxw
1348 * bytes we can have, or whatever is less than that.
1349 */
1350 extent_bytes = max_bytes - bitmap_bytes;
1351 extent_bytes = min_t(u64, extent_bytes, div64_u64(max_bytes, 2));
1352
1353 ctl->extents_thresh =
1354 div64_u64(extent_bytes, (sizeof(struct btrfs_free_space)));
1355 }
1356
1357 static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1358 struct btrfs_free_space *info,
1359 u64 offset, u64 bytes)
1360 {
1361 unsigned long start, count;
1362
1363 start = offset_to_bit(info->offset, ctl->unit, offset);
1364 count = bytes_to_bits(bytes, ctl->unit);
1365 BUG_ON(start + count > BITS_PER_BITMAP);
1366
1367 bitmap_clear(info->bitmap, start, count);
1368
1369 info->bytes -= bytes;
1370 }
1371
1372 static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1373 struct btrfs_free_space *info, u64 offset,
1374 u64 bytes)
1375 {
1376 __bitmap_clear_bits(ctl, info, offset, bytes);
1377 ctl->free_space -= bytes;
1378 }
1379
1380 static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1381 struct btrfs_free_space *info, u64 offset,
1382 u64 bytes)
1383 {
1384 unsigned long start, count;
1385
1386 start = offset_to_bit(info->offset, ctl->unit, offset);
1387 count = bytes_to_bits(bytes, ctl->unit);
1388 BUG_ON(start + count > BITS_PER_BITMAP);
1389
1390 bitmap_set(info->bitmap, start, count);
1391
1392 info->bytes += bytes;
1393 ctl->free_space += bytes;
1394 }
1395
1396 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1397 struct btrfs_free_space *bitmap_info, u64 *offset,
1398 u64 *bytes)
1399 {
1400 unsigned long found_bits = 0;
1401 unsigned long bits, i;
1402 unsigned long next_zero;
1403
1404 i = offset_to_bit(bitmap_info->offset, ctl->unit,
1405 max_t(u64, *offset, bitmap_info->offset));
1406 bits = bytes_to_bits(*bytes, ctl->unit);
1407
1408 for (i = find_next_bit(bitmap_info->bitmap, BITS_PER_BITMAP, i);
1409 i < BITS_PER_BITMAP;
1410 i = find_next_bit(bitmap_info->bitmap, BITS_PER_BITMAP, i + 1)) {
1411 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1412 BITS_PER_BITMAP, i);
1413 if ((next_zero - i) >= bits) {
1414 found_bits = next_zero - i;
1415 break;
1416 }
1417 i = next_zero;
1418 }
1419
1420 if (found_bits) {
1421 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1422 *bytes = (u64)(found_bits) * ctl->unit;
1423 return 0;
1424 }
1425
1426 return -1;
1427 }
1428
1429 static struct btrfs_free_space *
1430 find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes)
1431 {
1432 struct btrfs_free_space *entry;
1433 struct rb_node *node;
1434 int ret;
1435
1436 if (!ctl->free_space_offset.rb_node)
1437 return NULL;
1438
1439 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
1440 if (!entry)
1441 return NULL;
1442
1443 for (node = &entry->offset_index; node; node = rb_next(node)) {
1444 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1445 if (entry->bytes < *bytes)
1446 continue;
1447
1448 if (entry->bitmap) {
1449 ret = search_bitmap(ctl, entry, offset, bytes);
1450 if (!ret)
1451 return entry;
1452 continue;
1453 }
1454
1455 *offset = entry->offset;
1456 *bytes = entry->bytes;
1457 return entry;
1458 }
1459
1460 return NULL;
1461 }
1462
1463 static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
1464 struct btrfs_free_space *info, u64 offset)
1465 {
1466 info->offset = offset_to_bitmap(ctl, offset);
1467 info->bytes = 0;
1468 link_free_space(ctl, info);
1469 ctl->total_bitmaps++;
1470
1471 ctl->op->recalc_thresholds(ctl);
1472 }
1473
1474 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
1475 struct btrfs_free_space *bitmap_info)
1476 {
1477 unlink_free_space(ctl, bitmap_info);
1478 kfree(bitmap_info->bitmap);
1479 kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1480 ctl->total_bitmaps--;
1481 ctl->op->recalc_thresholds(ctl);
1482 }
1483
1484 static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
1485 struct btrfs_free_space *bitmap_info,
1486 u64 *offset, u64 *bytes)
1487 {
1488 u64 end;
1489 u64 search_start, search_bytes;
1490 int ret;
1491
1492 again:
1493 end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
1494
1495 /*
1496 * XXX - this can go away after a few releases.
1497 *
1498 * since the only user of btrfs_remove_free_space is the tree logging
1499 * stuff, and the only way to test that is under crash conditions, we
1500 * want to have this debug stuff here just in case somethings not
1501 * working. Search the bitmap for the space we are trying to use to
1502 * make sure its actually there. If its not there then we need to stop
1503 * because something has gone wrong.
1504 */
1505 search_start = *offset;
1506 search_bytes = *bytes;
1507 search_bytes = min(search_bytes, end - search_start + 1);
1508 ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes);
1509 BUG_ON(ret < 0 || search_start != *offset);
1510
1511 if (*offset > bitmap_info->offset && *offset + *bytes > end) {
1512 bitmap_clear_bits(ctl, bitmap_info, *offset, end - *offset + 1);
1513 *bytes -= end - *offset + 1;
1514 *offset = end + 1;
1515 } else if (*offset >= bitmap_info->offset && *offset + *bytes <= end) {
1516 bitmap_clear_bits(ctl, bitmap_info, *offset, *bytes);
1517 *bytes = 0;
1518 }
1519
1520 if (*bytes) {
1521 struct rb_node *next = rb_next(&bitmap_info->offset_index);
1522 if (!bitmap_info->bytes)
1523 free_bitmap(ctl, bitmap_info);
1524
1525 /*
1526 * no entry after this bitmap, but we still have bytes to
1527 * remove, so something has gone wrong.
1528 */
1529 if (!next)
1530 return -EINVAL;
1531
1532 bitmap_info = rb_entry(next, struct btrfs_free_space,
1533 offset_index);
1534
1535 /*
1536 * if the next entry isn't a bitmap we need to return to let the
1537 * extent stuff do its work.
1538 */
1539 if (!bitmap_info->bitmap)
1540 return -EAGAIN;
1541
1542 /*
1543 * Ok the next item is a bitmap, but it may not actually hold
1544 * the information for the rest of this free space stuff, so
1545 * look for it, and if we don't find it return so we can try
1546 * everything over again.
1547 */
1548 search_start = *offset;
1549 search_bytes = *bytes;
1550 ret = search_bitmap(ctl, bitmap_info, &search_start,
1551 &search_bytes);
1552 if (ret < 0 || search_start != *offset)
1553 return -EAGAIN;
1554
1555 goto again;
1556 } else if (!bitmap_info->bytes)
1557 free_bitmap(ctl, bitmap_info);
1558
1559 return 0;
1560 }
1561
1562 static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
1563 struct btrfs_free_space *info, u64 offset,
1564 u64 bytes)
1565 {
1566 u64 bytes_to_set = 0;
1567 u64 end;
1568
1569 end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
1570
1571 bytes_to_set = min(end - offset, bytes);
1572
1573 bitmap_set_bits(ctl, info, offset, bytes_to_set);
1574
1575 return bytes_to_set;
1576
1577 }
1578
1579 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
1580 struct btrfs_free_space *info)
1581 {
1582 struct btrfs_block_group_cache *block_group = ctl->private;
1583
1584 /*
1585 * If we are below the extents threshold then we can add this as an
1586 * extent, and don't have to deal with the bitmap
1587 */
1588 if (ctl->free_extents < ctl->extents_thresh) {
1589 /*
1590 * If this block group has some small extents we don't want to
1591 * use up all of our free slots in the cache with them, we want
1592 * to reserve them to larger extents, however if we have plent
1593 * of cache left then go ahead an dadd them, no sense in adding
1594 * the overhead of a bitmap if we don't have to.
1595 */
1596 if (info->bytes <= block_group->sectorsize * 4) {
1597 if (ctl->free_extents * 2 <= ctl->extents_thresh)
1598 return false;
1599 } else {
1600 return false;
1601 }
1602 }
1603
1604 /*
1605 * some block groups are so tiny they can't be enveloped by a bitmap, so
1606 * don't even bother to create a bitmap for this
1607 */
1608 if (BITS_PER_BITMAP * block_group->sectorsize >
1609 block_group->key.offset)
1610 return false;
1611
1612 return true;
1613 }
1614
1615 static struct btrfs_free_space_op free_space_op = {
1616 .recalc_thresholds = recalculate_thresholds,
1617 .use_bitmap = use_bitmap,
1618 };
1619
1620 static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
1621 struct btrfs_free_space *info)
1622 {
1623 struct btrfs_free_space *bitmap_info;
1624 struct btrfs_block_group_cache *block_group = NULL;
1625 int added = 0;
1626 u64 bytes, offset, bytes_added;
1627 int ret;
1628
1629 bytes = info->bytes;
1630 offset = info->offset;
1631
1632 if (!ctl->op->use_bitmap(ctl, info))
1633 return 0;
1634
1635 if (ctl->op == &free_space_op)
1636 block_group = ctl->private;
1637 again:
1638 /*
1639 * Since we link bitmaps right into the cluster we need to see if we
1640 * have a cluster here, and if so and it has our bitmap we need to add
1641 * the free space to that bitmap.
1642 */
1643 if (block_group && !list_empty(&block_group->cluster_list)) {
1644 struct btrfs_free_cluster *cluster;
1645 struct rb_node *node;
1646 struct btrfs_free_space *entry;
1647
1648 cluster = list_entry(block_group->cluster_list.next,
1649 struct btrfs_free_cluster,
1650 block_group_list);
1651 spin_lock(&cluster->lock);
1652 node = rb_first(&cluster->root);
1653 if (!node) {
1654 spin_unlock(&cluster->lock);
1655 goto no_cluster_bitmap;
1656 }
1657
1658 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1659 if (!entry->bitmap) {
1660 spin_unlock(&cluster->lock);
1661 goto no_cluster_bitmap;
1662 }
1663
1664 if (entry->offset == offset_to_bitmap(ctl, offset)) {
1665 bytes_added = add_bytes_to_bitmap(ctl, entry,
1666 offset, bytes);
1667 bytes -= bytes_added;
1668 offset += bytes_added;
1669 }
1670 spin_unlock(&cluster->lock);
1671 if (!bytes) {
1672 ret = 1;
1673 goto out;
1674 }
1675 }
1676
1677 no_cluster_bitmap:
1678 bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
1679 1, 0);
1680 if (!bitmap_info) {
1681 BUG_ON(added);
1682 goto new_bitmap;
1683 }
1684
1685 bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
1686 bytes -= bytes_added;
1687 offset += bytes_added;
1688 added = 0;
1689
1690 if (!bytes) {
1691 ret = 1;
1692 goto out;
1693 } else
1694 goto again;
1695
1696 new_bitmap:
1697 if (info && info->bitmap) {
1698 add_new_bitmap(ctl, info, offset);
1699 added = 1;
1700 info = NULL;
1701 goto again;
1702 } else {
1703 spin_unlock(&ctl->tree_lock);
1704
1705 /* no pre-allocated info, allocate a new one */
1706 if (!info) {
1707 info = kmem_cache_zalloc(btrfs_free_space_cachep,
1708 GFP_NOFS);
1709 if (!info) {
1710 spin_lock(&ctl->tree_lock);
1711 ret = -ENOMEM;
1712 goto out;
1713 }
1714 }
1715
1716 /* allocate the bitmap */
1717 info->bitmap = kzalloc(PAGE_CACHE_SIZE, GFP_NOFS);
1718 spin_lock(&ctl->tree_lock);
1719 if (!info->bitmap) {
1720 ret = -ENOMEM;
1721 goto out;
1722 }
1723 goto again;
1724 }
1725
1726 out:
1727 if (info) {
1728 if (info->bitmap)
1729 kfree(info->bitmap);
1730 kmem_cache_free(btrfs_free_space_cachep, info);
1731 }
1732
1733 return ret;
1734 }
1735
1736 static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
1737 struct btrfs_free_space *info, bool update_stat)
1738 {
1739 struct btrfs_free_space *left_info;
1740 struct btrfs_free_space *right_info;
1741 bool merged = false;
1742 u64 offset = info->offset;
1743 u64 bytes = info->bytes;
1744
1745 /*
1746 * first we want to see if there is free space adjacent to the range we
1747 * are adding, if there is remove that struct and add a new one to
1748 * cover the entire range
1749 */
1750 right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
1751 if (right_info && rb_prev(&right_info->offset_index))
1752 left_info = rb_entry(rb_prev(&right_info->offset_index),
1753 struct btrfs_free_space, offset_index);
1754 else
1755 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
1756
1757 if (right_info && !right_info->bitmap) {
1758 if (update_stat)
1759 unlink_free_space(ctl, right_info);
1760 else
1761 __unlink_free_space(ctl, right_info);
1762 info->bytes += right_info->bytes;
1763 kmem_cache_free(btrfs_free_space_cachep, right_info);
1764 merged = true;
1765 }
1766
1767 if (left_info && !left_info->bitmap &&
1768 left_info->offset + left_info->bytes == offset) {
1769 if (update_stat)
1770 unlink_free_space(ctl, left_info);
1771 else
1772 __unlink_free_space(ctl, left_info);
1773 info->offset = left_info->offset;
1774 info->bytes += left_info->bytes;
1775 kmem_cache_free(btrfs_free_space_cachep, left_info);
1776 merged = true;
1777 }
1778
1779 return merged;
1780 }
1781
1782 int __btrfs_add_free_space(struct btrfs_free_space_ctl *ctl,
1783 u64 offset, u64 bytes)
1784 {
1785 struct btrfs_free_space *info;
1786 int ret = 0;
1787
1788 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
1789 if (!info)
1790 return -ENOMEM;
1791
1792 info->offset = offset;
1793 info->bytes = bytes;
1794
1795 spin_lock(&ctl->tree_lock);
1796
1797 if (try_merge_free_space(ctl, info, true))
1798 goto link;
1799
1800 /*
1801 * There was no extent directly to the left or right of this new
1802 * extent then we know we're going to have to allocate a new extent, so
1803 * before we do that see if we need to drop this into a bitmap
1804 */
1805 ret = insert_into_bitmap(ctl, info);
1806 if (ret < 0) {
1807 goto out;
1808 } else if (ret) {
1809 ret = 0;
1810 goto out;
1811 }
1812 link:
1813 ret = link_free_space(ctl, info);
1814 if (ret)
1815 kmem_cache_free(btrfs_free_space_cachep, info);
1816 out:
1817 spin_unlock(&ctl->tree_lock);
1818
1819 if (ret) {
1820 printk(KERN_CRIT "btrfs: unable to add free space :%d\n", ret);
1821 BUG_ON(ret == -EEXIST);
1822 }
1823
1824 return ret;
1825 }
1826
1827 int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
1828 u64 offset, u64 bytes)
1829 {
1830 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1831 struct btrfs_free_space *info;
1832 struct btrfs_free_space *next_info = NULL;
1833 int ret = 0;
1834
1835 spin_lock(&ctl->tree_lock);
1836
1837 again:
1838 info = tree_search_offset(ctl, offset, 0, 0);
1839 if (!info) {
1840 /*
1841 * oops didn't find an extent that matched the space we wanted
1842 * to remove, look for a bitmap instead
1843 */
1844 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
1845 1, 0);
1846 if (!info) {
1847 WARN_ON(1);
1848 goto out_lock;
1849 }
1850 }
1851
1852 if (info->bytes < bytes && rb_next(&info->offset_index)) {
1853 u64 end;
1854 next_info = rb_entry(rb_next(&info->offset_index),
1855 struct btrfs_free_space,
1856 offset_index);
1857
1858 if (next_info->bitmap)
1859 end = next_info->offset +
1860 BITS_PER_BITMAP * ctl->unit - 1;
1861 else
1862 end = next_info->offset + next_info->bytes;
1863
1864 if (next_info->bytes < bytes ||
1865 next_info->offset > offset || offset > end) {
1866 printk(KERN_CRIT "Found free space at %llu, size %llu,"
1867 " trying to use %llu\n",
1868 (unsigned long long)info->offset,
1869 (unsigned long long)info->bytes,
1870 (unsigned long long)bytes);
1871 WARN_ON(1);
1872 ret = -EINVAL;
1873 goto out_lock;
1874 }
1875
1876 info = next_info;
1877 }
1878
1879 if (info->bytes == bytes) {
1880 unlink_free_space(ctl, info);
1881 if (info->bitmap) {
1882 kfree(info->bitmap);
1883 ctl->total_bitmaps--;
1884 }
1885 kmem_cache_free(btrfs_free_space_cachep, info);
1886 ret = 0;
1887 goto out_lock;
1888 }
1889
1890 if (!info->bitmap && info->offset == offset) {
1891 unlink_free_space(ctl, info);
1892 info->offset += bytes;
1893 info->bytes -= bytes;
1894 ret = link_free_space(ctl, info);
1895 WARN_ON(ret);
1896 goto out_lock;
1897 }
1898
1899 if (!info->bitmap && info->offset <= offset &&
1900 info->offset + info->bytes >= offset + bytes) {
1901 u64 old_start = info->offset;
1902 /*
1903 * we're freeing space in the middle of the info,
1904 * this can happen during tree log replay
1905 *
1906 * first unlink the old info and then
1907 * insert it again after the hole we're creating
1908 */
1909 unlink_free_space(ctl, info);
1910 if (offset + bytes < info->offset + info->bytes) {
1911 u64 old_end = info->offset + info->bytes;
1912
1913 info->offset = offset + bytes;
1914 info->bytes = old_end - info->offset;
1915 ret = link_free_space(ctl, info);
1916 WARN_ON(ret);
1917 if (ret)
1918 goto out_lock;
1919 } else {
1920 /* the hole we're creating ends at the end
1921 * of the info struct, just free the info
1922 */
1923 kmem_cache_free(btrfs_free_space_cachep, info);
1924 }
1925 spin_unlock(&ctl->tree_lock);
1926
1927 /* step two, insert a new info struct to cover
1928 * anything before the hole
1929 */
1930 ret = btrfs_add_free_space(block_group, old_start,
1931 offset - old_start);
1932 WARN_ON(ret);
1933 goto out;
1934 }
1935
1936 ret = remove_from_bitmap(ctl, info, &offset, &bytes);
1937 if (ret == -EAGAIN)
1938 goto again;
1939 BUG_ON(ret);
1940 out_lock:
1941 spin_unlock(&ctl->tree_lock);
1942 out:
1943 return ret;
1944 }
1945
1946 void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
1947 u64 bytes)
1948 {
1949 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1950 struct btrfs_free_space *info;
1951 struct rb_node *n;
1952 int count = 0;
1953
1954 for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
1955 info = rb_entry(n, struct btrfs_free_space, offset_index);
1956 if (info->bytes >= bytes)
1957 count++;
1958 printk(KERN_CRIT "entry offset %llu, bytes %llu, bitmap %s\n",
1959 (unsigned long long)info->offset,
1960 (unsigned long long)info->bytes,
1961 (info->bitmap) ? "yes" : "no");
1962 }
1963 printk(KERN_INFO "block group has cluster?: %s\n",
1964 list_empty(&block_group->cluster_list) ? "no" : "yes");
1965 printk(KERN_INFO "%d blocks of free space at or bigger than bytes is"
1966 "\n", count);
1967 }
1968
1969 void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group)
1970 {
1971 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1972
1973 spin_lock_init(&ctl->tree_lock);
1974 ctl->unit = block_group->sectorsize;
1975 ctl->start = block_group->key.objectid;
1976 ctl->private = block_group;
1977 ctl->op = &free_space_op;
1978
1979 /*
1980 * we only want to have 32k of ram per block group for keeping
1981 * track of free space, and if we pass 1/2 of that we want to
1982 * start converting things over to using bitmaps
1983 */
1984 ctl->extents_thresh = ((1024 * 32) / 2) /
1985 sizeof(struct btrfs_free_space);
1986 }
1987
1988 /*
1989 * for a given cluster, put all of its extents back into the free
1990 * space cache. If the block group passed doesn't match the block group
1991 * pointed to by the cluster, someone else raced in and freed the
1992 * cluster already. In that case, we just return without changing anything
1993 */
1994 static int
1995 __btrfs_return_cluster_to_free_space(
1996 struct btrfs_block_group_cache *block_group,
1997 struct btrfs_free_cluster *cluster)
1998 {
1999 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2000 struct btrfs_free_space *entry;
2001 struct rb_node *node;
2002
2003 spin_lock(&cluster->lock);
2004 if (cluster->block_group != block_group)
2005 goto out;
2006
2007 cluster->block_group = NULL;
2008 cluster->window_start = 0;
2009 list_del_init(&cluster->block_group_list);
2010
2011 node = rb_first(&cluster->root);
2012 while (node) {
2013 bool bitmap;
2014
2015 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2016 node = rb_next(&entry->offset_index);
2017 rb_erase(&entry->offset_index, &cluster->root);
2018
2019 bitmap = (entry->bitmap != NULL);
2020 if (!bitmap)
2021 try_merge_free_space(ctl, entry, false);
2022 tree_insert_offset(&ctl->free_space_offset,
2023 entry->offset, &entry->offset_index, bitmap);
2024 }
2025 cluster->root = RB_ROOT;
2026
2027 out:
2028 spin_unlock(&cluster->lock);
2029 btrfs_put_block_group(block_group);
2030 return 0;
2031 }
2032
2033 void __btrfs_remove_free_space_cache_locked(struct btrfs_free_space_ctl *ctl)
2034 {
2035 struct btrfs_free_space *info;
2036 struct rb_node *node;
2037
2038 while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
2039 info = rb_entry(node, struct btrfs_free_space, offset_index);
2040 if (!info->bitmap) {
2041 unlink_free_space(ctl, info);
2042 kmem_cache_free(btrfs_free_space_cachep, info);
2043 } else {
2044 free_bitmap(ctl, info);
2045 }
2046 if (need_resched()) {
2047 spin_unlock(&ctl->tree_lock);
2048 cond_resched();
2049 spin_lock(&ctl->tree_lock);
2050 }
2051 }
2052 }
2053
2054 void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
2055 {
2056 spin_lock(&ctl->tree_lock);
2057 __btrfs_remove_free_space_cache_locked(ctl);
2058 spin_unlock(&ctl->tree_lock);
2059 }
2060
2061 void btrfs_remove_free_space_cache(struct btrfs_block_group_cache *block_group)
2062 {
2063 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2064 struct btrfs_free_cluster *cluster;
2065 struct list_head *head;
2066
2067 spin_lock(&ctl->tree_lock);
2068 while ((head = block_group->cluster_list.next) !=
2069 &block_group->cluster_list) {
2070 cluster = list_entry(head, struct btrfs_free_cluster,
2071 block_group_list);
2072
2073 WARN_ON(cluster->block_group != block_group);
2074 __btrfs_return_cluster_to_free_space(block_group, cluster);
2075 if (need_resched()) {
2076 spin_unlock(&ctl->tree_lock);
2077 cond_resched();
2078 spin_lock(&ctl->tree_lock);
2079 }
2080 }
2081 __btrfs_remove_free_space_cache_locked(ctl);
2082 spin_unlock(&ctl->tree_lock);
2083
2084 }
2085
2086 u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group,
2087 u64 offset, u64 bytes, u64 empty_size)
2088 {
2089 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2090 struct btrfs_free_space *entry = NULL;
2091 u64 bytes_search = bytes + empty_size;
2092 u64 ret = 0;
2093
2094 spin_lock(&ctl->tree_lock);
2095 entry = find_free_space(ctl, &offset, &bytes_search);
2096 if (!entry)
2097 goto out;
2098
2099 ret = offset;
2100 if (entry->bitmap) {
2101 bitmap_clear_bits(ctl, entry, offset, bytes);
2102 if (!entry->bytes)
2103 free_bitmap(ctl, entry);
2104 } else {
2105 unlink_free_space(ctl, entry);
2106 entry->offset += bytes;
2107 entry->bytes -= bytes;
2108 if (!entry->bytes)
2109 kmem_cache_free(btrfs_free_space_cachep, entry);
2110 else
2111 link_free_space(ctl, entry);
2112 }
2113
2114 out:
2115 spin_unlock(&ctl->tree_lock);
2116
2117 return ret;
2118 }
2119
2120 /*
2121 * given a cluster, put all of its extents back into the free space
2122 * cache. If a block group is passed, this function will only free
2123 * a cluster that belongs to the passed block group.
2124 *
2125 * Otherwise, it'll get a reference on the block group pointed to by the
2126 * cluster and remove the cluster from it.
2127 */
2128 int btrfs_return_cluster_to_free_space(
2129 struct btrfs_block_group_cache *block_group,
2130 struct btrfs_free_cluster *cluster)
2131 {
2132 struct btrfs_free_space_ctl *ctl;
2133 int ret;
2134
2135 /* first, get a safe pointer to the block group */
2136 spin_lock(&cluster->lock);
2137 if (!block_group) {
2138 block_group = cluster->block_group;
2139 if (!block_group) {
2140 spin_unlock(&cluster->lock);
2141 return 0;
2142 }
2143 } else if (cluster->block_group != block_group) {
2144 /* someone else has already freed it don't redo their work */
2145 spin_unlock(&cluster->lock);
2146 return 0;
2147 }
2148 atomic_inc(&block_group->count);
2149 spin_unlock(&cluster->lock);
2150
2151 ctl = block_group->free_space_ctl;
2152
2153 /* now return any extents the cluster had on it */
2154 spin_lock(&ctl->tree_lock);
2155 ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
2156 spin_unlock(&ctl->tree_lock);
2157
2158 /* finally drop our ref */
2159 btrfs_put_block_group(block_group);
2160 return ret;
2161 }
2162
2163 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
2164 struct btrfs_free_cluster *cluster,
2165 struct btrfs_free_space *entry,
2166 u64 bytes, u64 min_start)
2167 {
2168 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2169 int err;
2170 u64 search_start = cluster->window_start;
2171 u64 search_bytes = bytes;
2172 u64 ret = 0;
2173
2174 search_start = min_start;
2175 search_bytes = bytes;
2176
2177 err = search_bitmap(ctl, entry, &search_start, &search_bytes);
2178 if (err)
2179 return 0;
2180
2181 ret = search_start;
2182 __bitmap_clear_bits(ctl, entry, ret, bytes);
2183
2184 return ret;
2185 }
2186
2187 /*
2188 * given a cluster, try to allocate 'bytes' from it, returns 0
2189 * if it couldn't find anything suitably large, or a logical disk offset
2190 * if things worked out
2191 */
2192 u64 btrfs_alloc_from_cluster(struct btrfs_block_group_cache *block_group,
2193 struct btrfs_free_cluster *cluster, u64 bytes,
2194 u64 min_start)
2195 {
2196 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2197 struct btrfs_free_space *entry = NULL;
2198 struct rb_node *node;
2199 u64 ret = 0;
2200
2201 spin_lock(&cluster->lock);
2202 if (bytes > cluster->max_size)
2203 goto out;
2204
2205 if (cluster->block_group != block_group)
2206 goto out;
2207
2208 node = rb_first(&cluster->root);
2209 if (!node)
2210 goto out;
2211
2212 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2213 while(1) {
2214 if (entry->bytes < bytes ||
2215 (!entry->bitmap && entry->offset < min_start)) {
2216 node = rb_next(&entry->offset_index);
2217 if (!node)
2218 break;
2219 entry = rb_entry(node, struct btrfs_free_space,
2220 offset_index);
2221 continue;
2222 }
2223
2224 if (entry->bitmap) {
2225 ret = btrfs_alloc_from_bitmap(block_group,
2226 cluster, entry, bytes,
2227 min_start);
2228 if (ret == 0) {
2229 node = rb_next(&entry->offset_index);
2230 if (!node)
2231 break;
2232 entry = rb_entry(node, struct btrfs_free_space,
2233 offset_index);
2234 continue;
2235 }
2236 } else {
2237 ret = entry->offset;
2238
2239 entry->offset += bytes;
2240 entry->bytes -= bytes;
2241 }
2242
2243 if (entry->bytes == 0)
2244 rb_erase(&entry->offset_index, &cluster->root);
2245 break;
2246 }
2247 out:
2248 spin_unlock(&cluster->lock);
2249
2250 if (!ret)
2251 return 0;
2252
2253 spin_lock(&ctl->tree_lock);
2254
2255 ctl->free_space -= bytes;
2256 if (entry->bytes == 0) {
2257 ctl->free_extents--;
2258 if (entry->bitmap) {
2259 kfree(entry->bitmap);
2260 ctl->total_bitmaps--;
2261 ctl->op->recalc_thresholds(ctl);
2262 }
2263 kmem_cache_free(btrfs_free_space_cachep, entry);
2264 }
2265
2266 spin_unlock(&ctl->tree_lock);
2267
2268 return ret;
2269 }
2270
2271 static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
2272 struct btrfs_free_space *entry,
2273 struct btrfs_free_cluster *cluster,
2274 u64 offset, u64 bytes, u64 min_bytes)
2275 {
2276 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2277 unsigned long next_zero;
2278 unsigned long i;
2279 unsigned long search_bits;
2280 unsigned long total_bits;
2281 unsigned long found_bits;
2282 unsigned long start = 0;
2283 unsigned long total_found = 0;
2284 int ret;
2285 bool found = false;
2286
2287 i = offset_to_bit(entry->offset, block_group->sectorsize,
2288 max_t(u64, offset, entry->offset));
2289 search_bits = bytes_to_bits(bytes, block_group->sectorsize);
2290 total_bits = bytes_to_bits(min_bytes, block_group->sectorsize);
2291
2292 again:
2293 found_bits = 0;
2294 for (i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, i);
2295 i < BITS_PER_BITMAP;
2296 i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, i + 1)) {
2297 next_zero = find_next_zero_bit(entry->bitmap,
2298 BITS_PER_BITMAP, i);
2299 if (next_zero - i >= search_bits) {
2300 found_bits = next_zero - i;
2301 break;
2302 }
2303 i = next_zero;
2304 }
2305
2306 if (!found_bits)
2307 return -ENOSPC;
2308
2309 if (!found) {
2310 start = i;
2311 found = true;
2312 }
2313
2314 total_found += found_bits;
2315
2316 if (cluster->max_size < found_bits * block_group->sectorsize)
2317 cluster->max_size = found_bits * block_group->sectorsize;
2318
2319 if (total_found < total_bits) {
2320 i = find_next_bit(entry->bitmap, BITS_PER_BITMAP, next_zero);
2321 if (i - start > total_bits * 2) {
2322 total_found = 0;
2323 cluster->max_size = 0;
2324 found = false;
2325 }
2326 goto again;
2327 }
2328
2329 cluster->window_start = start * block_group->sectorsize +
2330 entry->offset;
2331 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2332 ret = tree_insert_offset(&cluster->root, entry->offset,
2333 &entry->offset_index, 1);
2334 BUG_ON(ret);
2335
2336 return 0;
2337 }
2338
2339 /*
2340 * This searches the block group for just extents to fill the cluster with.
2341 */
2342 static noinline int
2343 setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2344 struct btrfs_free_cluster *cluster,
2345 struct list_head *bitmaps, u64 offset, u64 bytes,
2346 u64 min_bytes)
2347 {
2348 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2349 struct btrfs_free_space *first = NULL;
2350 struct btrfs_free_space *entry = NULL;
2351 struct btrfs_free_space *prev = NULL;
2352 struct btrfs_free_space *last;
2353 struct rb_node *node;
2354 u64 window_start;
2355 u64 window_free;
2356 u64 max_extent;
2357 u64 max_gap = 128 * 1024;
2358
2359 entry = tree_search_offset(ctl, offset, 0, 1);
2360 if (!entry)
2361 return -ENOSPC;
2362
2363 /*
2364 * We don't want bitmaps, so just move along until we find a normal
2365 * extent entry.
2366 */
2367 while (entry->bitmap) {
2368 if (list_empty(&entry->list))
2369 list_add_tail(&entry->list, bitmaps);
2370 node = rb_next(&entry->offset_index);
2371 if (!node)
2372 return -ENOSPC;
2373 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2374 }
2375
2376 window_start = entry->offset;
2377 window_free = entry->bytes;
2378 max_extent = entry->bytes;
2379 first = entry;
2380 last = entry;
2381 prev = entry;
2382
2383 while (window_free <= min_bytes) {
2384 node = rb_next(&entry->offset_index);
2385 if (!node)
2386 return -ENOSPC;
2387 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2388
2389 if (entry->bitmap) {
2390 if (list_empty(&entry->list))
2391 list_add_tail(&entry->list, bitmaps);
2392 continue;
2393 }
2394
2395 /*
2396 * we haven't filled the empty size and the window is
2397 * very large. reset and try again
2398 */
2399 if (entry->offset - (prev->offset + prev->bytes) > max_gap ||
2400 entry->offset - window_start > (min_bytes * 2)) {
2401 first = entry;
2402 window_start = entry->offset;
2403 window_free = entry->bytes;
2404 last = entry;
2405 max_extent = entry->bytes;
2406 } else {
2407 last = entry;
2408 window_free += entry->bytes;
2409 if (entry->bytes > max_extent)
2410 max_extent = entry->bytes;
2411 }
2412 prev = entry;
2413 }
2414
2415 cluster->window_start = first->offset;
2416
2417 node = &first->offset_index;
2418
2419 /*
2420 * now we've found our entries, pull them out of the free space
2421 * cache and put them into the cluster rbtree
2422 */
2423 do {
2424 int ret;
2425
2426 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2427 node = rb_next(&entry->offset_index);
2428 if (entry->bitmap)
2429 continue;
2430
2431 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2432 ret = tree_insert_offset(&cluster->root, entry->offset,
2433 &entry->offset_index, 0);
2434 BUG_ON(ret);
2435 } while (node && entry != last);
2436
2437 cluster->max_size = max_extent;
2438
2439 return 0;
2440 }
2441
2442 /*
2443 * This specifically looks for bitmaps that may work in the cluster, we assume
2444 * that we have already failed to find extents that will work.
2445 */
2446 static noinline int
2447 setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2448 struct btrfs_free_cluster *cluster,
2449 struct list_head *bitmaps, u64 offset, u64 bytes,
2450 u64 min_bytes)
2451 {
2452 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2453 struct btrfs_free_space *entry;
2454 struct rb_node *node;
2455 int ret = -ENOSPC;
2456
2457 if (ctl->total_bitmaps == 0)
2458 return -ENOSPC;
2459
2460 /*
2461 * First check our cached list of bitmaps and see if there is an entry
2462 * here that will work.
2463 */
2464 list_for_each_entry(entry, bitmaps, list) {
2465 if (entry->bytes < min_bytes)
2466 continue;
2467 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
2468 bytes, min_bytes);
2469 if (!ret)
2470 return 0;
2471 }
2472
2473 /*
2474 * If we do have entries on our list and we are here then we didn't find
2475 * anything, so go ahead and get the next entry after the last entry in
2476 * this list and start the search from there.
2477 */
2478 if (!list_empty(bitmaps)) {
2479 entry = list_entry(bitmaps->prev, struct btrfs_free_space,
2480 list);
2481 node = rb_next(&entry->offset_index);
2482 if (!node)
2483 return -ENOSPC;
2484 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2485 goto search;
2486 }
2487
2488 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, offset), 0, 1);
2489 if (!entry)
2490 return -ENOSPC;
2491
2492 search:
2493 node = &entry->offset_index;
2494 do {
2495 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2496 node = rb_next(&entry->offset_index);
2497 if (!entry->bitmap)
2498 continue;
2499 if (entry->bytes < min_bytes)
2500 continue;
2501 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
2502 bytes, min_bytes);
2503 } while (ret && node);
2504
2505 return ret;
2506 }
2507
2508 /*
2509 * here we try to find a cluster of blocks in a block group. The goal
2510 * is to find at least bytes free and up to empty_size + bytes free.
2511 * We might not find them all in one contiguous area.
2512 *
2513 * returns zero and sets up cluster if things worked out, otherwise
2514 * it returns -enospc
2515 */
2516 int btrfs_find_space_cluster(struct btrfs_trans_handle *trans,
2517 struct btrfs_root *root,
2518 struct btrfs_block_group_cache *block_group,
2519 struct btrfs_free_cluster *cluster,
2520 u64 offset, u64 bytes, u64 empty_size)
2521 {
2522 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2523 struct list_head bitmaps;
2524 struct btrfs_free_space *entry, *tmp;
2525 u64 min_bytes;
2526 int ret;
2527
2528 /* for metadata, allow allocates with more holes */
2529 if (btrfs_test_opt(root, SSD_SPREAD)) {
2530 min_bytes = bytes + empty_size;
2531 } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
2532 /*
2533 * we want to do larger allocations when we are
2534 * flushing out the delayed refs, it helps prevent
2535 * making more work as we go along.
2536 */
2537 if (trans->transaction->delayed_refs.flushing)
2538 min_bytes = max(bytes, (bytes + empty_size) >> 1);
2539 else
2540 min_bytes = max(bytes, (bytes + empty_size) >> 4);
2541 } else
2542 min_bytes = max(bytes, (bytes + empty_size) >> 2);
2543
2544 spin_lock(&ctl->tree_lock);
2545
2546 /*
2547 * If we know we don't have enough space to make a cluster don't even
2548 * bother doing all the work to try and find one.
2549 */
2550 if (ctl->free_space < min_bytes) {
2551 spin_unlock(&ctl->tree_lock);
2552 return -ENOSPC;
2553 }
2554
2555 spin_lock(&cluster->lock);
2556
2557 /* someone already found a cluster, hooray */
2558 if (cluster->block_group) {
2559 ret = 0;
2560 goto out;
2561 }
2562
2563 INIT_LIST_HEAD(&bitmaps);
2564 ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
2565 bytes, min_bytes);
2566 if (ret)
2567 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
2568 offset, bytes, min_bytes);
2569
2570 /* Clear our temporary list */
2571 list_for_each_entry_safe(entry, tmp, &bitmaps, list)
2572 list_del_init(&entry->list);
2573
2574 if (!ret) {
2575 atomic_inc(&block_group->count);
2576 list_add_tail(&cluster->block_group_list,
2577 &block_group->cluster_list);
2578 cluster->block_group = block_group;
2579 }
2580 out:
2581 spin_unlock(&cluster->lock);
2582 spin_unlock(&ctl->tree_lock);
2583
2584 return ret;
2585 }
2586
2587 /*
2588 * simple code to zero out a cluster
2589 */
2590 void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
2591 {
2592 spin_lock_init(&cluster->lock);
2593 spin_lock_init(&cluster->refill_lock);
2594 cluster->root = RB_ROOT;
2595 cluster->max_size = 0;
2596 INIT_LIST_HEAD(&cluster->block_group_list);
2597 cluster->block_group = NULL;
2598 }
2599
2600 int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
2601 u64 *trimmed, u64 start, u64 end, u64 minlen)
2602 {
2603 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2604 struct btrfs_free_space *entry = NULL;
2605 struct btrfs_fs_info *fs_info = block_group->fs_info;
2606 u64 bytes = 0;
2607 u64 actually_trimmed;
2608 int ret = 0;
2609
2610 *trimmed = 0;
2611
2612 while (start < end) {
2613 spin_lock(&ctl->tree_lock);
2614
2615 if (ctl->free_space < minlen) {
2616 spin_unlock(&ctl->tree_lock);
2617 break;
2618 }
2619
2620 entry = tree_search_offset(ctl, start, 0, 1);
2621 if (!entry)
2622 entry = tree_search_offset(ctl,
2623 offset_to_bitmap(ctl, start),
2624 1, 1);
2625
2626 if (!entry || entry->offset >= end) {
2627 spin_unlock(&ctl->tree_lock);
2628 break;
2629 }
2630
2631 if (entry->bitmap) {
2632 ret = search_bitmap(ctl, entry, &start, &bytes);
2633 if (!ret) {
2634 if (start >= end) {
2635 spin_unlock(&ctl->tree_lock);
2636 break;
2637 }
2638 bytes = min(bytes, end - start);
2639 bitmap_clear_bits(ctl, entry, start, bytes);
2640 if (entry->bytes == 0)
2641 free_bitmap(ctl, entry);
2642 } else {
2643 start = entry->offset + BITS_PER_BITMAP *
2644 block_group->sectorsize;
2645 spin_unlock(&ctl->tree_lock);
2646 ret = 0;
2647 continue;
2648 }
2649 } else {
2650 start = entry->offset;
2651 bytes = min(entry->bytes, end - start);
2652 unlink_free_space(ctl, entry);
2653 kmem_cache_free(btrfs_free_space_cachep, entry);
2654 }
2655
2656 spin_unlock(&ctl->tree_lock);
2657
2658 if (bytes >= minlen) {
2659 struct btrfs_space_info *space_info;
2660 int update = 0;
2661
2662 space_info = block_group->space_info;
2663 spin_lock(&space_info->lock);
2664 spin_lock(&block_group->lock);
2665 if (!block_group->ro) {
2666 block_group->reserved += bytes;
2667 space_info->bytes_reserved += bytes;
2668 update = 1;
2669 }
2670 spin_unlock(&block_group->lock);
2671 spin_unlock(&space_info->lock);
2672
2673 ret = btrfs_error_discard_extent(fs_info->extent_root,
2674 start,
2675 bytes,
2676 &actually_trimmed);
2677
2678 btrfs_add_free_space(block_group, start, bytes);
2679 if (update) {
2680 spin_lock(&space_info->lock);
2681 spin_lock(&block_group->lock);
2682 if (block_group->ro)
2683 space_info->bytes_readonly += bytes;
2684 block_group->reserved -= bytes;
2685 space_info->bytes_reserved -= bytes;
2686 spin_unlock(&space_info->lock);
2687 spin_unlock(&block_group->lock);
2688 }
2689
2690 if (ret)
2691 break;
2692 *trimmed += actually_trimmed;
2693 }
2694 start += bytes;
2695 bytes = 0;
2696
2697 if (fatal_signal_pending(current)) {
2698 ret = -ERESTARTSYS;
2699 break;
2700 }
2701
2702 cond_resched();
2703 }
2704
2705 return ret;
2706 }
2707
2708 /*
2709 * Find the left-most item in the cache tree, and then return the
2710 * smallest inode number in the item.
2711 *
2712 * Note: the returned inode number may not be the smallest one in
2713 * the tree, if the left-most item is a bitmap.
2714 */
2715 u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
2716 {
2717 struct btrfs_free_space_ctl *ctl = fs_root->free_ino_ctl;
2718 struct btrfs_free_space *entry = NULL;
2719 u64 ino = 0;
2720
2721 spin_lock(&ctl->tree_lock);
2722
2723 if (RB_EMPTY_ROOT(&ctl->free_space_offset))
2724 goto out;
2725
2726 entry = rb_entry(rb_first(&ctl->free_space_offset),
2727 struct btrfs_free_space, offset_index);
2728
2729 if (!entry->bitmap) {
2730 ino = entry->offset;
2731
2732 unlink_free_space(ctl, entry);
2733 entry->offset++;
2734 entry->bytes--;
2735 if (!entry->bytes)
2736 kmem_cache_free(btrfs_free_space_cachep, entry);
2737 else
2738 link_free_space(ctl, entry);
2739 } else {
2740 u64 offset = 0;
2741 u64 count = 1;
2742 int ret;
2743
2744 ret = search_bitmap(ctl, entry, &offset, &count);
2745 BUG_ON(ret);
2746
2747 ino = offset;
2748 bitmap_clear_bits(ctl, entry, offset, 1);
2749 if (entry->bytes == 0)
2750 free_bitmap(ctl, entry);
2751 }
2752 out:
2753 spin_unlock(&ctl->tree_lock);
2754
2755 return ino;
2756 }
2757
2758 struct inode *lookup_free_ino_inode(struct btrfs_root *root,
2759 struct btrfs_path *path)
2760 {
2761 struct inode *inode = NULL;
2762
2763 spin_lock(&root->cache_lock);
2764 if (root->cache_inode)
2765 inode = igrab(root->cache_inode);
2766 spin_unlock(&root->cache_lock);
2767 if (inode)
2768 return inode;
2769
2770 inode = __lookup_free_space_inode(root, path, 0);
2771 if (IS_ERR(inode))
2772 return inode;
2773
2774 spin_lock(&root->cache_lock);
2775 if (!btrfs_fs_closing(root->fs_info))
2776 root->cache_inode = igrab(inode);
2777 spin_unlock(&root->cache_lock);
2778
2779 return inode;
2780 }
2781
2782 int create_free_ino_inode(struct btrfs_root *root,
2783 struct btrfs_trans_handle *trans,
2784 struct btrfs_path *path)
2785 {
2786 return __create_free_space_inode(root, trans, path,
2787 BTRFS_FREE_INO_OBJECTID, 0);
2788 }
2789
2790 int load_free_ino_cache(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
2791 {
2792 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
2793 struct btrfs_path *path;
2794 struct inode *inode;
2795 int ret = 0;
2796 u64 root_gen = btrfs_root_generation(&root->root_item);
2797
2798 if (!btrfs_test_opt(root, INODE_MAP_CACHE))
2799 return 0;
2800
2801 /*
2802 * If we're unmounting then just return, since this does a search on the
2803 * normal root and not the commit root and we could deadlock.
2804 */
2805 if (btrfs_fs_closing(fs_info))
2806 return 0;
2807
2808 path = btrfs_alloc_path();
2809 if (!path)
2810 return 0;
2811
2812 inode = lookup_free_ino_inode(root, path);
2813 if (IS_ERR(inode))
2814 goto out;
2815
2816 if (root_gen != BTRFS_I(inode)->generation)
2817 goto out_put;
2818
2819 ret = __load_free_space_cache(root, inode, ctl, path, 0);
2820
2821 if (ret < 0)
2822 printk(KERN_ERR "btrfs: failed to load free ino cache for "
2823 "root %llu\n", root->root_key.objectid);
2824 out_put:
2825 iput(inode);
2826 out:
2827 btrfs_free_path(path);
2828 return ret;
2829 }
2830
2831 int btrfs_write_out_ino_cache(struct btrfs_root *root,
2832 struct btrfs_trans_handle *trans,
2833 struct btrfs_path *path)
2834 {
2835 struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
2836 struct inode *inode;
2837 int ret;
2838
2839 if (!btrfs_test_opt(root, INODE_MAP_CACHE))
2840 return 0;
2841
2842 inode = lookup_free_ino_inode(root, path);
2843 if (IS_ERR(inode))
2844 return 0;
2845
2846 ret = __btrfs_write_out_cache(root, inode, ctl, NULL, trans, path, 0);
2847 if (ret) {
2848 btrfs_delalloc_release_metadata(inode, inode->i_size);
2849 #ifdef DEBUG
2850 printk(KERN_ERR "btrfs: failed to write free ino cache "
2851 "for root %llu\n", root->root_key.objectid);
2852 #endif
2853 }
2854
2855 iput(inode);
2856 return ret;
2857 }
Linux 2.6 libata-dev (mirror)