1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * Copyright (C) 2007 Oracle. All rights reserved.
4 */
5
6 #include <linux/sched.h>
7 #include <linux/sched/signal.h>
8 #include <linux/pagemap.h>
9 #include <linux/writeback.h>
10 #include <linux/blkdev.h>
11 #include <linux/sort.h>
12 #include <linux/rcupdate.h>
13 #include <linux/kthread.h>
14 #include <linux/slab.h>
15 #include <linux/ratelimit.h>
16 #include <linux/percpu_counter.h>
17 #include <linux/lockdep.h>
18 #include <linux/crc32c.h>
19 #include "ctree.h"
20 #include "extent-tree.h"
21 #include "transaction.h"
22 #include "disk-io.h"
23 #include "print-tree.h"
24 #include "volumes.h"
25 #include "raid56.h"
26 #include "locking.h"
27 #include "free-space-cache.h"
28 #include "free-space-tree.h"
29 #include "qgroup.h"
30 #include "ref-verify.h"
31 #include "space-info.h"
32 #include "block-rsv.h"
33 #include "discard.h"
34 #include "zoned.h"
35 #include "dev-replace.h"
36 #include "fs.h"
37 #include "accessors.h"
38 #include "root-tree.h"
39 #include "file-item.h"
40 #include "orphan.h"
41 #include "tree-checker.h"
42 #include "raid-stripe-tree.h"
43
44 #undef SCRAMBLE_DELAYED_REFS
45
46
47 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
48 struct btrfs_delayed_ref_head *href,
49 struct btrfs_delayed_ref_node *node,
50 struct btrfs_delayed_extent_op *extra_op);
51 static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
52 struct extent_buffer *leaf,
53 struct btrfs_extent_item *ei);
54 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
55 u64 parent, u64 root_objectid,
56 u64 flags, u64 owner, u64 offset,
57 struct btrfs_key *ins, int ref_mod, u64 oref_root);
58 static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
59 struct btrfs_delayed_ref_node *node,
60 struct btrfs_delayed_extent_op *extent_op);
61 static int find_next_key(struct btrfs_path *path, int level,
62 struct btrfs_key *key);
63
block_group_bits(struct btrfs_block_group * cache,u64 bits)64 static int block_group_bits(struct btrfs_block_group *cache, u64 bits)
65 {
66 return (cache->flags & bits) == bits;
67 }
68
69 /* simple helper to search for an existing data extent at a given offset */
btrfs_lookup_data_extent(struct btrfs_fs_info * fs_info,u64 start,u64 len)70 int btrfs_lookup_data_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len)
71 {
72 struct btrfs_root *root = btrfs_extent_root(fs_info, start);
73 int ret;
74 struct btrfs_key key;
75 struct btrfs_path *path;
76
77 path = btrfs_alloc_path();
78 if (!path)
79 return -ENOMEM;
80
81 key.objectid = start;
82 key.offset = len;
83 key.type = BTRFS_EXTENT_ITEM_KEY;
84 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
85 btrfs_free_path(path);
86 return ret;
87 }
88
89 /*
90 * helper function to lookup reference count and flags of a tree block.
91 *
92 * the head node for delayed ref is used to store the sum of all the
93 * reference count modifications queued up in the rbtree. the head
94 * node may also store the extent flags to set. This way you can check
95 * to see what the reference count and extent flags would be if all of
96 * the delayed refs are not processed.
97 */
btrfs_lookup_extent_info(struct btrfs_trans_handle * trans,struct btrfs_fs_info * fs_info,u64 bytenr,u64 offset,int metadata,u64 * refs,u64 * flags,u64 * owning_root)98 int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
99 struct btrfs_fs_info *fs_info, u64 bytenr,
100 u64 offset, int metadata, u64 *refs, u64 *flags,
101 u64 *owning_root)
102 {
103 struct btrfs_root *extent_root;
104 struct btrfs_delayed_ref_head *head;
105 struct btrfs_delayed_ref_root *delayed_refs;
106 struct btrfs_path *path;
107 struct btrfs_key key;
108 u64 num_refs;
109 u64 extent_flags;
110 u64 owner = 0;
111 int ret;
112
113 /*
114 * If we don't have skinny metadata, don't bother doing anything
115 * different
116 */
117 if (metadata && !btrfs_fs_incompat(fs_info, SKINNY_METADATA)) {
118 offset = fs_info->nodesize;
119 metadata = 0;
120 }
121
122 path = btrfs_alloc_path();
123 if (!path)
124 return -ENOMEM;
125
126 search_again:
127 key.objectid = bytenr;
128 key.offset = offset;
129 if (metadata)
130 key.type = BTRFS_METADATA_ITEM_KEY;
131 else
132 key.type = BTRFS_EXTENT_ITEM_KEY;
133
134 extent_root = btrfs_extent_root(fs_info, bytenr);
135 ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
136 if (ret < 0)
137 goto out_free;
138
139 if (ret > 0 && key.type == BTRFS_METADATA_ITEM_KEY) {
140 if (path->slots[0]) {
141 path->slots[0]--;
142 btrfs_item_key_to_cpu(path->nodes[0], &key,
143 path->slots[0]);
144 if (key.objectid == bytenr &&
145 key.type == BTRFS_EXTENT_ITEM_KEY &&
146 key.offset == fs_info->nodesize)
147 ret = 0;
148 }
149 }
150
151 if (ret == 0) {
152 struct extent_buffer *leaf = path->nodes[0];
153 struct btrfs_extent_item *ei;
154 const u32 item_size = btrfs_item_size(leaf, path->slots[0]);
155
156 if (unlikely(item_size < sizeof(*ei))) {
157 ret = -EUCLEAN;
158 btrfs_err(fs_info,
159 "unexpected extent item size, has %u expect >= %zu",
160 item_size, sizeof(*ei));
161 btrfs_abort_transaction(trans, ret);
162 goto out_free;
163 }
164
165 ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
166 num_refs = btrfs_extent_refs(leaf, ei);
167 if (unlikely(num_refs == 0)) {
168 ret = -EUCLEAN;
169 btrfs_err(fs_info,
170 "unexpected zero reference count for extent item (%llu %u %llu)",
171 key.objectid, key.type, key.offset);
172 btrfs_abort_transaction(trans, ret);
173 goto out_free;
174 }
175 extent_flags = btrfs_extent_flags(leaf, ei);
176 owner = btrfs_get_extent_owner_root(fs_info, leaf, path->slots[0]);
177 } else {
178 num_refs = 0;
179 extent_flags = 0;
180 ret = 0;
181 }
182
183 delayed_refs = &trans->transaction->delayed_refs;
184 spin_lock(&delayed_refs->lock);
185 head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
186 if (head) {
187 if (!mutex_trylock(&head->mutex)) {
188 refcount_inc(&head->refs);
189 spin_unlock(&delayed_refs->lock);
190
191 btrfs_release_path(path);
192
193 /*
194 * Mutex was contended, block until it's released and try
195 * again
196 */
197 mutex_lock(&head->mutex);
198 mutex_unlock(&head->mutex);
199 btrfs_put_delayed_ref_head(head);
200 goto search_again;
201 }
202 spin_lock(&head->lock);
203 if (head->extent_op && head->extent_op->update_flags)
204 extent_flags |= head->extent_op->flags_to_set;
205
206 num_refs += head->ref_mod;
207 spin_unlock(&head->lock);
208 mutex_unlock(&head->mutex);
209 }
210 spin_unlock(&delayed_refs->lock);
211
212 WARN_ON(num_refs == 0);
213 if (refs)
214 *refs = num_refs;
215 if (flags)
216 *flags = extent_flags;
217 if (owning_root)
218 *owning_root = owner;
219 out_free:
220 btrfs_free_path(path);
221 return ret;
222 }
223
224 /*
225 * Back reference rules. Back refs have three main goals:
226 *
227 * 1) differentiate between all holders of references to an extent so that
228 * when a reference is dropped we can make sure it was a valid reference
229 * before freeing the extent.
230 *
231 * 2) Provide enough information to quickly find the holders of an extent
232 * if we notice a given block is corrupted or bad.
233 *
234 * 3) Make it easy to migrate blocks for FS shrinking or storage pool
235 * maintenance. This is actually the same as #2, but with a slightly
236 * different use case.
237 *
238 * There are two kinds of back refs. The implicit back refs is optimized
239 * for pointers in non-shared tree blocks. For a given pointer in a block,
240 * back refs of this kind provide information about the block's owner tree
241 * and the pointer's key. These information allow us to find the block by
242 * b-tree searching. The full back refs is for pointers in tree blocks not
243 * referenced by their owner trees. The location of tree block is recorded
244 * in the back refs. Actually the full back refs is generic, and can be
245 * used in all cases the implicit back refs is used. The major shortcoming
246 * of the full back refs is its overhead. Every time a tree block gets
247 * COWed, we have to update back refs entry for all pointers in it.
248 *
249 * For a newly allocated tree block, we use implicit back refs for
250 * pointers in it. This means most tree related operations only involve
251 * implicit back refs. For a tree block created in old transaction, the
252 * only way to drop a reference to it is COW it. So we can detect the
253 * event that tree block loses its owner tree's reference and do the
254 * back refs conversion.
255 *
256 * When a tree block is COWed through a tree, there are four cases:
257 *
258 * The reference count of the block is one and the tree is the block's
259 * owner tree. Nothing to do in this case.
260 *
261 * The reference count of the block is one and the tree is not the
262 * block's owner tree. In this case, full back refs is used for pointers
263 * in the block. Remove these full back refs, add implicit back refs for
264 * every pointers in the new block.
265 *
266 * The reference count of the block is greater than one and the tree is
267 * the block's owner tree. In this case, implicit back refs is used for
268 * pointers in the block. Add full back refs for every pointers in the
269 * block, increase lower level extents' reference counts. The original
270 * implicit back refs are entailed to the new block.
271 *
272 * The reference count of the block is greater than one and the tree is
273 * not the block's owner tree. Add implicit back refs for every pointer in
274 * the new block, increase lower level extents' reference count.
275 *
276 * Back Reference Key composing:
277 *
278 * The key objectid corresponds to the first byte in the extent,
279 * The key type is used to differentiate between types of back refs.
280 * There are different meanings of the key offset for different types
281 * of back refs.
282 *
283 * File extents can be referenced by:
284 *
285 * - multiple snapshots, subvolumes, or different generations in one subvol
286 * - different files inside a single subvolume
287 * - different offsets inside a file (bookend extents in file.c)
288 *
289 * The extent ref structure for the implicit back refs has fields for:
290 *
291 * - Objectid of the subvolume root
292 * - objectid of the file holding the reference
293 * - original offset in the file
294 * - how many bookend extents
295 *
296 * The key offset for the implicit back refs is hash of the first
297 * three fields.
298 *
299 * The extent ref structure for the full back refs has field for:
300 *
301 * - number of pointers in the tree leaf
302 *
303 * The key offset for the implicit back refs is the first byte of
304 * the tree leaf
305 *
306 * When a file extent is allocated, The implicit back refs is used.
307 * the fields are filled in:
308 *
309 * (root_key.objectid, inode objectid, offset in file, 1)
310 *
311 * When a file extent is removed file truncation, we find the
312 * corresponding implicit back refs and check the following fields:
313 *
314 * (btrfs_header_owner(leaf), inode objectid, offset in file)
315 *
316 * Btree extents can be referenced by:
317 *
318 * - Different subvolumes
319 *
320 * Both the implicit back refs and the full back refs for tree blocks
321 * only consist of key. The key offset for the implicit back refs is
322 * objectid of block's owner tree. The key offset for the full back refs
323 * is the first byte of parent block.
324 *
325 * When implicit back refs is used, information about the lowest key and
326 * level of the tree block are required. These information are stored in
327 * tree block info structure.
328 */
329
330 /*
331 * is_data == BTRFS_REF_TYPE_BLOCK, tree block type is required,
332 * is_data == BTRFS_REF_TYPE_DATA, data type is requiried,
333 * is_data == BTRFS_REF_TYPE_ANY, either type is OK.
334 */
btrfs_get_extent_inline_ref_type(const struct extent_buffer * eb,struct btrfs_extent_inline_ref * iref,enum btrfs_inline_ref_type is_data)335 int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
336 struct btrfs_extent_inline_ref *iref,
337 enum btrfs_inline_ref_type is_data)
338 {
339 struct btrfs_fs_info *fs_info = eb->fs_info;
340 int type = btrfs_extent_inline_ref_type(eb, iref);
341 u64 offset = btrfs_extent_inline_ref_offset(eb, iref);
342
343 if (type == BTRFS_EXTENT_OWNER_REF_KEY) {
344 ASSERT(btrfs_fs_incompat(fs_info, SIMPLE_QUOTA));
345 return type;
346 }
347
348 if (type == BTRFS_TREE_BLOCK_REF_KEY ||
349 type == BTRFS_SHARED_BLOCK_REF_KEY ||
350 type == BTRFS_SHARED_DATA_REF_KEY ||
351 type == BTRFS_EXTENT_DATA_REF_KEY) {
352 if (is_data == BTRFS_REF_TYPE_BLOCK) {
353 if (type == BTRFS_TREE_BLOCK_REF_KEY)
354 return type;
355 if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
356 ASSERT(fs_info);
357 /*
358 * Every shared one has parent tree block,
359 * which must be aligned to sector size.
360 */
361 if (offset && IS_ALIGNED(offset, fs_info->sectorsize))
362 return type;
363 }
364 } else if (is_data == BTRFS_REF_TYPE_DATA) {
365 if (type == BTRFS_EXTENT_DATA_REF_KEY)
366 return type;
367 if (type == BTRFS_SHARED_DATA_REF_KEY) {
368 ASSERT(fs_info);
369 /*
370 * Every shared one has parent tree block,
371 * which must be aligned to sector size.
372 */
373 if (offset &&
374 IS_ALIGNED(offset, fs_info->sectorsize))
375 return type;
376 }
377 } else {
378 ASSERT(is_data == BTRFS_REF_TYPE_ANY);
379 return type;
380 }
381 }
382
383 WARN_ON(1);
384 btrfs_print_leaf(eb);
385 btrfs_err(fs_info,
386 "eb %llu iref 0x%lx invalid extent inline ref type %d",
387 eb->start, (unsigned long)iref, type);
388
389 return BTRFS_REF_TYPE_INVALID;
390 }
391
hash_extent_data_ref(u64 root_objectid,u64 owner,u64 offset)392 u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset)
393 {
394 u32 high_crc = ~(u32)0;
395 u32 low_crc = ~(u32)0;
396 __le64 lenum;
397
398 lenum = cpu_to_le64(root_objectid);
399 high_crc = crc32c(high_crc, &lenum, sizeof(lenum));
400 lenum = cpu_to_le64(owner);
401 low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
402 lenum = cpu_to_le64(offset);
403 low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
404
405 return ((u64)high_crc << 31) ^ (u64)low_crc;
406 }
407
hash_extent_data_ref_item(struct extent_buffer * leaf,struct btrfs_extent_data_ref * ref)408 static u64 hash_extent_data_ref_item(struct extent_buffer *leaf,
409 struct btrfs_extent_data_ref *ref)
410 {
411 return hash_extent_data_ref(btrfs_extent_data_ref_root(leaf, ref),
412 btrfs_extent_data_ref_objectid(leaf, ref),
413 btrfs_extent_data_ref_offset(leaf, ref));
414 }
415
match_extent_data_ref(struct extent_buffer * leaf,struct btrfs_extent_data_ref * ref,u64 root_objectid,u64 owner,u64 offset)416 static int match_extent_data_ref(struct extent_buffer *leaf,
417 struct btrfs_extent_data_ref *ref,
418 u64 root_objectid, u64 owner, u64 offset)
419 {
420 if (btrfs_extent_data_ref_root(leaf, ref) != root_objectid ||
421 btrfs_extent_data_ref_objectid(leaf, ref) != owner ||
422 btrfs_extent_data_ref_offset(leaf, ref) != offset)
423 return 0;
424 return 1;
425 }
426
lookup_extent_data_ref(struct btrfs_trans_handle * trans,struct btrfs_path * path,u64 bytenr,u64 parent,u64 root_objectid,u64 owner,u64 offset)427 static noinline int lookup_extent_data_ref(struct btrfs_trans_handle *trans,
428 struct btrfs_path *path,
429 u64 bytenr, u64 parent,
430 u64 root_objectid,
431 u64 owner, u64 offset)
432 {
433 struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
434 struct btrfs_key key;
435 struct btrfs_extent_data_ref *ref;
436 struct extent_buffer *leaf;
437 u32 nritems;
438 int recow;
439 int ret;
440
441 key.objectid = bytenr;
442 if (parent) {
443 key.type = BTRFS_SHARED_DATA_REF_KEY;
444 key.offset = parent;
445 } else {
446 key.type = BTRFS_EXTENT_DATA_REF_KEY;
447 key.offset = hash_extent_data_ref(root_objectid,
448 owner, offset);
449 }
450 again:
451 recow = 0;
452 ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
453 if (ret < 0)
454 return ret;
455
456 if (parent) {
457 if (ret)
458 return -ENOENT;
459 return 0;
460 }
461
462 ret = -ENOENT;
463 leaf = path->nodes[0];
464 nritems = btrfs_header_nritems(leaf);
465 while (1) {
466 if (path->slots[0] >= nritems) {
467 ret = btrfs_next_leaf(root, path);
468 if (ret) {
469 if (ret > 0)
470 return -ENOENT;
471 return ret;
472 }
473
474 leaf = path->nodes[0];
475 nritems = btrfs_header_nritems(leaf);
476 recow = 1;
477 }
478
479 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
480 if (key.objectid != bytenr ||
481 key.type != BTRFS_EXTENT_DATA_REF_KEY)
482 goto fail;
483
484 ref = btrfs_item_ptr(leaf, path->slots[0],
485 struct btrfs_extent_data_ref);
486
487 if (match_extent_data_ref(leaf, ref, root_objectid,
488 owner, offset)) {
489 if (recow) {
490 btrfs_release_path(path);
491 goto again;
492 }
493 ret = 0;
494 break;
495 }
496 path->slots[0]++;
497 }
498 fail:
499 return ret;
500 }
501
insert_extent_data_ref(struct btrfs_trans_handle * trans,struct btrfs_path * path,struct btrfs_delayed_ref_node * node,u64 bytenr)502 static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans,
503 struct btrfs_path *path,
504 struct btrfs_delayed_ref_node *node,
505 u64 bytenr)
506 {
507 struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
508 struct btrfs_key key;
509 struct extent_buffer *leaf;
510 u64 owner = btrfs_delayed_ref_owner(node);
511 u64 offset = btrfs_delayed_ref_offset(node);
512 u32 size;
513 u32 num_refs;
514 int ret;
515
516 key.objectid = bytenr;
517 if (node->parent) {
518 key.type = BTRFS_SHARED_DATA_REF_KEY;
519 key.offset = node->parent;
520 size = sizeof(struct btrfs_shared_data_ref);
521 } else {
522 key.type = BTRFS_EXTENT_DATA_REF_KEY;
523 key.offset = hash_extent_data_ref(node->ref_root, owner, offset);
524 size = sizeof(struct btrfs_extent_data_ref);
525 }
526
527 ret = btrfs_insert_empty_item(trans, root, path, &key, size);
528 if (ret && ret != -EEXIST)
529 goto fail;
530
531 leaf = path->nodes[0];
532 if (node->parent) {
533 struct btrfs_shared_data_ref *ref;
534 ref = btrfs_item_ptr(leaf, path->slots[0],
535 struct btrfs_shared_data_ref);
536 if (ret == 0) {
537 btrfs_set_shared_data_ref_count(leaf, ref, node->ref_mod);
538 } else {
539 num_refs = btrfs_shared_data_ref_count(leaf, ref);
540 num_refs += node->ref_mod;
541 btrfs_set_shared_data_ref_count(leaf, ref, num_refs);
542 }
543 } else {
544 struct btrfs_extent_data_ref *ref;
545 while (ret == -EEXIST) {
546 ref = btrfs_item_ptr(leaf, path->slots[0],
547 struct btrfs_extent_data_ref);
548 if (match_extent_data_ref(leaf, ref, node->ref_root,
549 owner, offset))
550 break;
551 btrfs_release_path(path);
552 key.offset++;
553 ret = btrfs_insert_empty_item(trans, root, path, &key,
554 size);
555 if (ret && ret != -EEXIST)
556 goto fail;
557
558 leaf = path->nodes[0];
559 }
560 ref = btrfs_item_ptr(leaf, path->slots[0],
561 struct btrfs_extent_data_ref);
562 if (ret == 0) {
563 btrfs_set_extent_data_ref_root(leaf, ref, node->ref_root);
564 btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
565 btrfs_set_extent_data_ref_offset(leaf, ref, offset);
566 btrfs_set_extent_data_ref_count(leaf, ref, node->ref_mod);
567 } else {
568 num_refs = btrfs_extent_data_ref_count(leaf, ref);
569 num_refs += node->ref_mod;
570 btrfs_set_extent_data_ref_count(leaf, ref, num_refs);
571 }
572 }
573 btrfs_mark_buffer_dirty(trans, leaf);
574 ret = 0;
575 fail:
576 btrfs_release_path(path);
577 return ret;
578 }
579
remove_extent_data_ref(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,int refs_to_drop)580 static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
581 struct btrfs_root *root,
582 struct btrfs_path *path,
583 int refs_to_drop)
584 {
585 struct btrfs_key key;
586 struct btrfs_extent_data_ref *ref1 = NULL;
587 struct btrfs_shared_data_ref *ref2 = NULL;
588 struct extent_buffer *leaf;
589 u32 num_refs = 0;
590 int ret = 0;
591
592 leaf = path->nodes[0];
593 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
594
595 if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
596 ref1 = btrfs_item_ptr(leaf, path->slots[0],
597 struct btrfs_extent_data_ref);
598 num_refs = btrfs_extent_data_ref_count(leaf, ref1);
599 } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
600 ref2 = btrfs_item_ptr(leaf, path->slots[0],
601 struct btrfs_shared_data_ref);
602 num_refs = btrfs_shared_data_ref_count(leaf, ref2);
603 } else {
604 btrfs_err(trans->fs_info,
605 "unrecognized backref key (%llu %u %llu)",
606 key.objectid, key.type, key.offset);
607 btrfs_abort_transaction(trans, -EUCLEAN);
608 return -EUCLEAN;
609 }
610
611 BUG_ON(num_refs < refs_to_drop);
612 num_refs -= refs_to_drop;
613
614 if (num_refs == 0) {
615 ret = btrfs_del_item(trans, root, path);
616 } else {
617 if (key.type == BTRFS_EXTENT_DATA_REF_KEY)
618 btrfs_set_extent_data_ref_count(leaf, ref1, num_refs);
619 else if (key.type == BTRFS_SHARED_DATA_REF_KEY)
620 btrfs_set_shared_data_ref_count(leaf, ref2, num_refs);
621 btrfs_mark_buffer_dirty(trans, leaf);
622 }
623 return ret;
624 }
625
extent_data_ref_count(struct btrfs_path * path,struct btrfs_extent_inline_ref * iref)626 static noinline u32 extent_data_ref_count(struct btrfs_path *path,
627 struct btrfs_extent_inline_ref *iref)
628 {
629 struct btrfs_key key;
630 struct extent_buffer *leaf;
631 struct btrfs_extent_data_ref *ref1;
632 struct btrfs_shared_data_ref *ref2;
633 u32 num_refs = 0;
634 int type;
635
636 leaf = path->nodes[0];
637 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
638
639 if (iref) {
640 /*
641 * If type is invalid, we should have bailed out earlier than
642 * this call.
643 */
644 type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
645 ASSERT(type != BTRFS_REF_TYPE_INVALID);
646 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
647 ref1 = (struct btrfs_extent_data_ref *)(&iref->offset);
648 num_refs = btrfs_extent_data_ref_count(leaf, ref1);
649 } else {
650 ref2 = (struct btrfs_shared_data_ref *)(iref + 1);
651 num_refs = btrfs_shared_data_ref_count(leaf, ref2);
652 }
653 } else if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
654 ref1 = btrfs_item_ptr(leaf, path->slots[0],
655 struct btrfs_extent_data_ref);
656 num_refs = btrfs_extent_data_ref_count(leaf, ref1);
657 } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
658 ref2 = btrfs_item_ptr(leaf, path->slots[0],
659 struct btrfs_shared_data_ref);
660 num_refs = btrfs_shared_data_ref_count(leaf, ref2);
661 } else {
662 WARN_ON(1);
663 }
664 return num_refs;
665 }
666
lookup_tree_block_ref(struct btrfs_trans_handle * trans,struct btrfs_path * path,u64 bytenr,u64 parent,u64 root_objectid)667 static noinline int lookup_tree_block_ref(struct btrfs_trans_handle *trans,
668 struct btrfs_path *path,
669 u64 bytenr, u64 parent,
670 u64 root_objectid)
671 {
672 struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
673 struct btrfs_key key;
674 int ret;
675
676 key.objectid = bytenr;
677 if (parent) {
678 key.type = BTRFS_SHARED_BLOCK_REF_KEY;
679 key.offset = parent;
680 } else {
681 key.type = BTRFS_TREE_BLOCK_REF_KEY;
682 key.offset = root_objectid;
683 }
684
685 ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
686 if (ret > 0)
687 ret = -ENOENT;
688 return ret;
689 }
690
insert_tree_block_ref(struct btrfs_trans_handle * trans,struct btrfs_path * path,struct btrfs_delayed_ref_node * node,u64 bytenr)691 static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans,
692 struct btrfs_path *path,
693 struct btrfs_delayed_ref_node *node,
694 u64 bytenr)
695 {
696 struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
697 struct btrfs_key key;
698 int ret;
699
700 key.objectid = bytenr;
701 if (node->parent) {
702 key.type = BTRFS_SHARED_BLOCK_REF_KEY;
703 key.offset = node->parent;
704 } else {
705 key.type = BTRFS_TREE_BLOCK_REF_KEY;
706 key.offset = node->ref_root;
707 }
708
709 ret = btrfs_insert_empty_item(trans, root, path, &key, 0);
710 btrfs_release_path(path);
711 return ret;
712 }
713
extent_ref_type(u64 parent,u64 owner)714 static inline int extent_ref_type(u64 parent, u64 owner)
715 {
716 int type;
717 if (owner < BTRFS_FIRST_FREE_OBJECTID) {
718 if (parent > 0)
719 type = BTRFS_SHARED_BLOCK_REF_KEY;
720 else
721 type = BTRFS_TREE_BLOCK_REF_KEY;
722 } else {
723 if (parent > 0)
724 type = BTRFS_SHARED_DATA_REF_KEY;
725 else
726 type = BTRFS_EXTENT_DATA_REF_KEY;
727 }
728 return type;
729 }
730
find_next_key(struct btrfs_path * path,int level,struct btrfs_key * key)731 static int find_next_key(struct btrfs_path *path, int level,
732 struct btrfs_key *key)
733
734 {
735 for (; level < BTRFS_MAX_LEVEL; level++) {
736 if (!path->nodes[level])
737 break;
738 if (path->slots[level] + 1 >=
739 btrfs_header_nritems(path->nodes[level]))
740 continue;
741 if (level == 0)
742 btrfs_item_key_to_cpu(path->nodes[level], key,
743 path->slots[level] + 1);
744 else
745 btrfs_node_key_to_cpu(path->nodes[level], key,
746 path->slots[level] + 1);
747 return 0;
748 }
749 return 1;
750 }
751
752 /*
753 * look for inline back ref. if back ref is found, *ref_ret is set
754 * to the address of inline back ref, and 0 is returned.
755 *
756 * if back ref isn't found, *ref_ret is set to the address where it
757 * should be inserted, and -ENOENT is returned.
758 *
759 * if insert is true and there are too many inline back refs, the path
760 * points to the extent item, and -EAGAIN is returned.
761 *
762 * NOTE: inline back refs are ordered in the same way that back ref
763 * items in the tree are ordered.
764 */
765 static noinline_for_stack
lookup_inline_extent_backref(struct btrfs_trans_handle * trans,struct btrfs_path * path,struct btrfs_extent_inline_ref ** ref_ret,u64 bytenr,u64 num_bytes,u64 parent,u64 root_objectid,u64 owner,u64 offset,int insert)766 int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
767 struct btrfs_path *path,
768 struct btrfs_extent_inline_ref **ref_ret,
769 u64 bytenr, u64 num_bytes,
770 u64 parent, u64 root_objectid,
771 u64 owner, u64 offset, int insert)
772 {
773 struct btrfs_fs_info *fs_info = trans->fs_info;
774 struct btrfs_root *root = btrfs_extent_root(fs_info, bytenr);
775 struct btrfs_key key;
776 struct extent_buffer *leaf;
777 struct btrfs_extent_item *ei;
778 struct btrfs_extent_inline_ref *iref;
779 u64 flags;
780 u64 item_size;
781 unsigned long ptr;
782 unsigned long end;
783 int extra_size;
784 int type;
785 int want;
786 int ret;
787 bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
788 int needed;
789
790 key.objectid = bytenr;
791 key.type = BTRFS_EXTENT_ITEM_KEY;
792 key.offset = num_bytes;
793
794 want = extent_ref_type(parent, owner);
795 if (insert) {
796 extra_size = btrfs_extent_inline_ref_size(want);
797 path->search_for_extension = 1;
798 path->keep_locks = 1;
799 } else
800 extra_size = -1;
801
802 /*
803 * Owner is our level, so we can just add one to get the level for the
804 * block we are interested in.
805 */
806 if (skinny_metadata && owner < BTRFS_FIRST_FREE_OBJECTID) {
807 key.type = BTRFS_METADATA_ITEM_KEY;
808 key.offset = owner;
809 }
810
811 again:
812 ret = btrfs_search_slot(trans, root, &key, path, extra_size, 1);
813 if (ret < 0)
814 goto out;
815
816 /*
817 * We may be a newly converted file system which still has the old fat
818 * extent entries for metadata, so try and see if we have one of those.
819 */
820 if (ret > 0 && skinny_metadata) {
821 skinny_metadata = false;
822 if (path->slots[0]) {
823 path->slots[0]--;
824 btrfs_item_key_to_cpu(path->nodes[0], &key,
825 path->slots[0]);
826 if (key.objectid == bytenr &&
827 key.type == BTRFS_EXTENT_ITEM_KEY &&
828 key.offset == num_bytes)
829 ret = 0;
830 }
831 if (ret) {
832 key.objectid = bytenr;
833 key.type = BTRFS_EXTENT_ITEM_KEY;
834 key.offset = num_bytes;
835 btrfs_release_path(path);
836 goto again;
837 }
838 }
839
840 if (ret && !insert) {
841 ret = -ENOENT;
842 goto out;
843 } else if (WARN_ON(ret)) {
844 btrfs_print_leaf(path->nodes[0]);
845 btrfs_err(fs_info,
846 "extent item not found for insert, bytenr %llu num_bytes %llu parent %llu root_objectid %llu owner %llu offset %llu",
847 bytenr, num_bytes, parent, root_objectid, owner,
848 offset);
849 ret = -EUCLEAN;
850 goto out;
851 }
852
853 leaf = path->nodes[0];
854 item_size = btrfs_item_size(leaf, path->slots[0]);
855 if (unlikely(item_size < sizeof(*ei))) {
856 ret = -EUCLEAN;
857 btrfs_err(fs_info,
858 "unexpected extent item size, has %llu expect >= %zu",
859 item_size, sizeof(*ei));
860 btrfs_abort_transaction(trans, ret);
861 goto out;
862 }
863
864 ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
865 flags = btrfs_extent_flags(leaf, ei);
866
867 ptr = (unsigned long)(ei + 1);
868 end = (unsigned long)ei + item_size;
869
870 if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK && !skinny_metadata) {
871 ptr += sizeof(struct btrfs_tree_block_info);
872 BUG_ON(ptr > end);
873 }
874
875 if (owner >= BTRFS_FIRST_FREE_OBJECTID)
876 needed = BTRFS_REF_TYPE_DATA;
877 else
878 needed = BTRFS_REF_TYPE_BLOCK;
879
880 ret = -ENOENT;
881 while (ptr < end) {
882 iref = (struct btrfs_extent_inline_ref *)ptr;
883 type = btrfs_get_extent_inline_ref_type(leaf, iref, needed);
884 if (type == BTRFS_EXTENT_OWNER_REF_KEY) {
885 ASSERT(btrfs_fs_incompat(fs_info, SIMPLE_QUOTA));
886 ptr += btrfs_extent_inline_ref_size(type);
887 continue;
888 }
889 if (type == BTRFS_REF_TYPE_INVALID) {
890 ret = -EUCLEAN;
891 goto out;
892 }
893
894 if (want < type)
895 break;
896 if (want > type) {
897 ptr += btrfs_extent_inline_ref_size(type);
898 continue;
899 }
900
901 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
902 struct btrfs_extent_data_ref *dref;
903 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
904 if (match_extent_data_ref(leaf, dref, root_objectid,
905 owner, offset)) {
906 ret = 0;
907 break;
908 }
909 if (hash_extent_data_ref_item(leaf, dref) <
910 hash_extent_data_ref(root_objectid, owner, offset))
911 break;
912 } else {
913 u64 ref_offset;
914 ref_offset = btrfs_extent_inline_ref_offset(leaf, iref);
915 if (parent > 0) {
916 if (parent == ref_offset) {
917 ret = 0;
918 break;
919 }
920 if (ref_offset < parent)
921 break;
922 } else {
923 if (root_objectid == ref_offset) {
924 ret = 0;
925 break;
926 }
927 if (ref_offset < root_objectid)
928 break;
929 }
930 }
931 ptr += btrfs_extent_inline_ref_size(type);
932 }
933
934 if (unlikely(ptr > end)) {
935 ret = -EUCLEAN;
936 btrfs_print_leaf(path->nodes[0]);
937 btrfs_crit(fs_info,
938 "overrun extent record at slot %d while looking for inline extent for root %llu owner %llu offset %llu parent %llu",
939 path->slots[0], root_objectid, owner, offset, parent);
940 goto out;
941 }
942
943 if (ret == -ENOENT && insert) {
944 if (item_size + extra_size >=
945 BTRFS_MAX_EXTENT_ITEM_SIZE(root)) {
946 ret = -EAGAIN;
947 goto out;
948 }
949 /*
950 * To add new inline back ref, we have to make sure
951 * there is no corresponding back ref item.
952 * For simplicity, we just do not add new inline back
953 * ref if there is any kind of item for this block
954 */
955 if (find_next_key(path, 0, &key) == 0 &&
956 key.objectid == bytenr &&
957 key.type < BTRFS_BLOCK_GROUP_ITEM_KEY) {
958 ret = -EAGAIN;
959 goto out;
960 }
961 }
962 *ref_ret = (struct btrfs_extent_inline_ref *)ptr;
963 out:
964 if (insert) {
965 path->keep_locks = 0;
966 path->search_for_extension = 0;
967 btrfs_unlock_up_safe(path, 1);
968 }
969 return ret;
970 }
971
972 /*
973 * helper to add new inline back ref
974 */
975 static noinline_for_stack
setup_inline_extent_backref(struct btrfs_trans_handle * trans,struct btrfs_path * path,struct btrfs_extent_inline_ref * iref,u64 parent,u64 root_objectid,u64 owner,u64 offset,int refs_to_add,struct btrfs_delayed_extent_op * extent_op)976 void setup_inline_extent_backref(struct btrfs_trans_handle *trans,
977 struct btrfs_path *path,
978 struct btrfs_extent_inline_ref *iref,
979 u64 parent, u64 root_objectid,
980 u64 owner, u64 offset, int refs_to_add,
981 struct btrfs_delayed_extent_op *extent_op)
982 {
983 struct extent_buffer *leaf;
984 struct btrfs_extent_item *ei;
985 unsigned long ptr;
986 unsigned long end;
987 unsigned long item_offset;
988 u64 refs;
989 int size;
990 int type;
991
992 leaf = path->nodes[0];
993 ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
994 item_offset = (unsigned long)iref - (unsigned long)ei;
995
996 type = extent_ref_type(parent, owner);
997 size = btrfs_extent_inline_ref_size(type);
998
999 btrfs_extend_item(trans, path, size);
1000
1001 ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1002 refs = btrfs_extent_refs(leaf, ei);
1003 refs += refs_to_add;
1004 btrfs_set_extent_refs(leaf, ei, refs);
1005 if (extent_op)
1006 __run_delayed_extent_op(extent_op, leaf, ei);
1007
1008 ptr = (unsigned long)ei + item_offset;
1009 end = (unsigned long)ei + btrfs_item_size(leaf, path->slots[0]);
1010 if (ptr < end - size)
1011 memmove_extent_buffer(leaf, ptr + size, ptr,
1012 end - size - ptr);
1013
1014 iref = (struct btrfs_extent_inline_ref *)ptr;
1015 btrfs_set_extent_inline_ref_type(leaf, iref, type);
1016 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
1017 struct btrfs_extent_data_ref *dref;
1018 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
1019 btrfs_set_extent_data_ref_root(leaf, dref, root_objectid);
1020 btrfs_set_extent_data_ref_objectid(leaf, dref, owner);
1021 btrfs_set_extent_data_ref_offset(leaf, dref, offset);
1022 btrfs_set_extent_data_ref_count(leaf, dref, refs_to_add);
1023 } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
1024 struct btrfs_shared_data_ref *sref;
1025 sref = (struct btrfs_shared_data_ref *)(iref + 1);
1026 btrfs_set_shared_data_ref_count(leaf, sref, refs_to_add);
1027 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
1028 } else if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
1029 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
1030 } else {
1031 btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid);
1032 }
1033 btrfs_mark_buffer_dirty(trans, leaf);
1034 }
1035
lookup_extent_backref(struct btrfs_trans_handle * trans,struct btrfs_path * path,struct btrfs_extent_inline_ref ** ref_ret,u64 bytenr,u64 num_bytes,u64 parent,u64 root_objectid,u64 owner,u64 offset)1036 static int lookup_extent_backref(struct btrfs_trans_handle *trans,
1037 struct btrfs_path *path,
1038 struct btrfs_extent_inline_ref **ref_ret,
1039 u64 bytenr, u64 num_bytes, u64 parent,
1040 u64 root_objectid, u64 owner, u64 offset)
1041 {
1042 int ret;
1043
1044 ret = lookup_inline_extent_backref(trans, path, ref_ret, bytenr,
1045 num_bytes, parent, root_objectid,
1046 owner, offset, 0);
1047 if (ret != -ENOENT)
1048 return ret;
1049
1050 btrfs_release_path(path);
1051 *ref_ret = NULL;
1052
1053 if (owner < BTRFS_FIRST_FREE_OBJECTID) {
1054 ret = lookup_tree_block_ref(trans, path, bytenr, parent,
1055 root_objectid);
1056 } else {
1057 ret = lookup_extent_data_ref(trans, path, bytenr, parent,
1058 root_objectid, owner, offset);
1059 }
1060 return ret;
1061 }
1062
1063 /*
1064 * helper to update/remove inline back ref
1065 */
update_inline_extent_backref(struct btrfs_trans_handle * trans,struct btrfs_path * path,struct btrfs_extent_inline_ref * iref,int refs_to_mod,struct btrfs_delayed_extent_op * extent_op)1066 static noinline_for_stack int update_inline_extent_backref(
1067 struct btrfs_trans_handle *trans,
1068 struct btrfs_path *path,
1069 struct btrfs_extent_inline_ref *iref,
1070 int refs_to_mod,
1071 struct btrfs_delayed_extent_op *extent_op)
1072 {
1073 struct extent_buffer *leaf = path->nodes[0];
1074 struct btrfs_fs_info *fs_info = leaf->fs_info;
1075 struct btrfs_extent_item *ei;
1076 struct btrfs_extent_data_ref *dref = NULL;
1077 struct btrfs_shared_data_ref *sref = NULL;
1078 unsigned long ptr;
1079 unsigned long end;
1080 u32 item_size;
1081 int size;
1082 int type;
1083 u64 refs;
1084
1085 ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1086 refs = btrfs_extent_refs(leaf, ei);
1087 if (unlikely(refs_to_mod < 0 && refs + refs_to_mod <= 0)) {
1088 struct btrfs_key key;
1089 u32 extent_size;
1090
1091 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1092 if (key.type == BTRFS_METADATA_ITEM_KEY)
1093 extent_size = fs_info->nodesize;
1094 else
1095 extent_size = key.offset;
1096 btrfs_print_leaf(leaf);
1097 btrfs_err(fs_info,
1098 "invalid refs_to_mod for extent %llu num_bytes %u, has %d expect >= -%llu",
1099 key.objectid, extent_size, refs_to_mod, refs);
1100 return -EUCLEAN;
1101 }
1102 refs += refs_to_mod;
1103 btrfs_set_extent_refs(leaf, ei, refs);
1104 if (extent_op)
1105 __run_delayed_extent_op(extent_op, leaf, ei);
1106
1107 type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_ANY);
1108 /*
1109 * Function btrfs_get_extent_inline_ref_type() has already printed
1110 * error messages.
1111 */
1112 if (unlikely(type == BTRFS_REF_TYPE_INVALID))
1113 return -EUCLEAN;
1114
1115 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
1116 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
1117 refs = btrfs_extent_data_ref_count(leaf, dref);
1118 } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
1119 sref = (struct btrfs_shared_data_ref *)(iref + 1);
1120 refs = btrfs_shared_data_ref_count(leaf, sref);
1121 } else {
1122 refs = 1;
1123 /*
1124 * For tree blocks we can only drop one ref for it, and tree
1125 * blocks should not have refs > 1.
1126 *
1127 * Furthermore if we're inserting a new inline backref, we
1128 * won't reach this path either. That would be
1129 * setup_inline_extent_backref().
1130 */
1131 if (unlikely(refs_to_mod != -1)) {
1132 struct btrfs_key key;
1133
1134 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1135
1136 btrfs_print_leaf(leaf);
1137 btrfs_err(fs_info,
1138 "invalid refs_to_mod for tree block %llu, has %d expect -1",
1139 key.objectid, refs_to_mod);
1140 return -EUCLEAN;
1141 }
1142 }
1143
1144 if (unlikely(refs_to_mod < 0 && refs < -refs_to_mod)) {
1145 struct btrfs_key key;
1146 u32 extent_size;
1147
1148 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1149 if (key.type == BTRFS_METADATA_ITEM_KEY)
1150 extent_size = fs_info->nodesize;
1151 else
1152 extent_size = key.offset;
1153 btrfs_print_leaf(leaf);
1154 btrfs_err(fs_info,
1155 "invalid refs_to_mod for backref entry, iref %lu extent %llu num_bytes %u, has %d expect >= -%llu",
1156 (unsigned long)iref, key.objectid, extent_size,
1157 refs_to_mod, refs);
1158 return -EUCLEAN;
1159 }
1160 refs += refs_to_mod;
1161
1162 if (refs > 0) {
1163 if (type == BTRFS_EXTENT_DATA_REF_KEY)
1164 btrfs_set_extent_data_ref_count(leaf, dref, refs);
1165 else
1166 btrfs_set_shared_data_ref_count(leaf, sref, refs);
1167 } else {
1168 size = btrfs_extent_inline_ref_size(type);
1169 item_size = btrfs_item_size(leaf, path->slots[0]);
1170 ptr = (unsigned long)iref;
1171 end = (unsigned long)ei + item_size;
1172 if (ptr + size < end)
1173 memmove_extent_buffer(leaf, ptr, ptr + size,
1174 end - ptr - size);
1175 item_size -= size;
1176 btrfs_truncate_item(trans, path, item_size, 1);
1177 }
1178 btrfs_mark_buffer_dirty(trans, leaf);
1179 return 0;
1180 }
1181
1182 static noinline_for_stack
insert_inline_extent_backref(struct btrfs_trans_handle * trans,struct btrfs_path * path,u64 bytenr,u64 num_bytes,u64 parent,u64 root_objectid,u64 owner,u64 offset,int refs_to_add,struct btrfs_delayed_extent_op * extent_op)1183 int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
1184 struct btrfs_path *path,
1185 u64 bytenr, u64 num_bytes, u64 parent,
1186 u64 root_objectid, u64 owner,
1187 u64 offset, int refs_to_add,
1188 struct btrfs_delayed_extent_op *extent_op)
1189 {
1190 struct btrfs_extent_inline_ref *iref;
1191 int ret;
1192
1193 ret = lookup_inline_extent_backref(trans, path, &iref, bytenr,
1194 num_bytes, parent, root_objectid,
1195 owner, offset, 1);
1196 if (ret == 0) {
1197 /*
1198 * We're adding refs to a tree block we already own, this
1199 * should not happen at all.
1200 */
1201 if (owner < BTRFS_FIRST_FREE_OBJECTID) {
1202 btrfs_print_leaf(path->nodes[0]);
1203 btrfs_crit(trans->fs_info,
1204 "adding refs to an existing tree ref, bytenr %llu num_bytes %llu root_objectid %llu slot %u",
1205 bytenr, num_bytes, root_objectid, path->slots[0]);
1206 return -EUCLEAN;
1207 }
1208 ret = update_inline_extent_backref(trans, path, iref,
1209 refs_to_add, extent_op);
1210 } else if (ret == -ENOENT) {
1211 setup_inline_extent_backref(trans, path, iref, parent,
1212 root_objectid, owner, offset,
1213 refs_to_add, extent_op);
1214 ret = 0;
1215 }
1216 return ret;
1217 }
1218
remove_extent_backref(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct btrfs_extent_inline_ref * iref,int refs_to_drop,int is_data)1219 static int remove_extent_backref(struct btrfs_trans_handle *trans,
1220 struct btrfs_root *root,
1221 struct btrfs_path *path,
1222 struct btrfs_extent_inline_ref *iref,
1223 int refs_to_drop, int is_data)
1224 {
1225 int ret = 0;
1226
1227 BUG_ON(!is_data && refs_to_drop != 1);
1228 if (iref)
1229 ret = update_inline_extent_backref(trans, path, iref,
1230 -refs_to_drop, NULL);
1231 else if (is_data)
1232 ret = remove_extent_data_ref(trans, root, path, refs_to_drop);
1233 else
1234 ret = btrfs_del_item(trans, root, path);
1235 return ret;
1236 }
1237
btrfs_issue_discard(struct block_device * bdev,u64 start,u64 len,u64 * discarded_bytes)1238 static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len,
1239 u64 *discarded_bytes)
1240 {
1241 int j, ret = 0;
1242 u64 bytes_left, end;
1243 u64 aligned_start = ALIGN(start, 1 << SECTOR_SHIFT);
1244
1245 /* Adjust the range to be aligned to 512B sectors if necessary. */
1246 if (start != aligned_start) {
1247 len -= aligned_start - start;
1248 len = round_down(len, 1 << SECTOR_SHIFT);
1249 start = aligned_start;
1250 }
1251
1252 *discarded_bytes = 0;
1253
1254 if (!len)
1255 return 0;
1256
1257 end = start + len;
1258 bytes_left = len;
1259
1260 /* Skip any superblocks on this device. */
1261 for (j = 0; j < BTRFS_SUPER_MIRROR_MAX; j++) {
1262 u64 sb_start = btrfs_sb_offset(j);
1263 u64 sb_end = sb_start + BTRFS_SUPER_INFO_SIZE;
1264 u64 size = sb_start - start;
1265
1266 if (!in_range(sb_start, start, bytes_left) &&
1267 !in_range(sb_end, start, bytes_left) &&
1268 !in_range(start, sb_start, BTRFS_SUPER_INFO_SIZE))
1269 continue;
1270
1271 /*
1272 * Superblock spans beginning of range. Adjust start and
1273 * try again.
1274 */
1275 if (sb_start <= start) {
1276 start += sb_end - start;
1277 if (start > end) {
1278 bytes_left = 0;
1279 break;
1280 }
1281 bytes_left = end - start;
1282 continue;
1283 }
1284
1285 if (size) {
1286 ret = blkdev_issue_discard(bdev, start >> SECTOR_SHIFT,
1287 size >> SECTOR_SHIFT,
1288 GFP_NOFS);
1289 if (!ret)
1290 *discarded_bytes += size;
1291 else if (ret != -EOPNOTSUPP)
1292 return ret;
1293 }
1294
1295 start = sb_end;
1296 if (start > end) {
1297 bytes_left = 0;
1298 break;
1299 }
1300 bytes_left = end - start;
1301 }
1302
1303 while (bytes_left) {
1304 u64 bytes_to_discard = min(BTRFS_MAX_DISCARD_CHUNK_SIZE, bytes_left);
1305
1306 ret = blkdev_issue_discard(bdev, start >> SECTOR_SHIFT,
1307 bytes_to_discard >> SECTOR_SHIFT,
1308 GFP_NOFS);
1309
1310 if (ret) {
1311 if (ret != -EOPNOTSUPP)
1312 break;
1313 continue;
1314 }
1315
1316 start += bytes_to_discard;
1317 bytes_left -= bytes_to_discard;
1318 *discarded_bytes += bytes_to_discard;
1319
1320 if (btrfs_trim_interrupted()) {
1321 ret = -ERESTARTSYS;
1322 break;
1323 }
1324 }
1325
1326 return ret;
1327 }
1328
do_discard_extent(struct btrfs_discard_stripe * stripe,u64 * bytes)1329 static int do_discard_extent(struct btrfs_discard_stripe *stripe, u64 *bytes)
1330 {
1331 struct btrfs_device *dev = stripe->dev;
1332 struct btrfs_fs_info *fs_info = dev->fs_info;
1333 struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
1334 u64 phys = stripe->physical;
1335 u64 len = stripe->length;
1336 u64 discarded = 0;
1337 int ret = 0;
1338
1339 /* Zone reset on a zoned filesystem */
1340 if (btrfs_can_zone_reset(dev, phys, len)) {
1341 u64 src_disc;
1342
1343 ret = btrfs_reset_device_zone(dev, phys, len, &discarded);
1344 if (ret)
1345 goto out;
1346
1347 if (!btrfs_dev_replace_is_ongoing(dev_replace) ||
1348 dev != dev_replace->srcdev)
1349 goto out;
1350
1351 src_disc = discarded;
1352
1353 /* Send to replace target as well */
1354 ret = btrfs_reset_device_zone(dev_replace->tgtdev, phys, len,
1355 &discarded);
1356 discarded += src_disc;
1357 } else if (bdev_max_discard_sectors(stripe->dev->bdev)) {
1358 ret = btrfs_issue_discard(dev->bdev, phys, len, &discarded);
1359 } else {
1360 ret = 0;
1361 *bytes = 0;
1362 }
1363
1364 out:
1365 *bytes = discarded;
1366 return ret;
1367 }
1368
btrfs_discard_extent(struct btrfs_fs_info * fs_info,u64 bytenr,u64 num_bytes,u64 * actual_bytes)1369 int btrfs_discard_extent(struct btrfs_fs_info *fs_info, u64 bytenr,
1370 u64 num_bytes, u64 *actual_bytes)
1371 {
1372 int ret = 0;
1373 u64 discarded_bytes = 0;
1374 u64 end = bytenr + num_bytes;
1375 u64 cur = bytenr;
1376
1377 /*
1378 * Avoid races with device replace and make sure the devices in the
1379 * stripes don't go away while we are discarding.
1380 */
1381 btrfs_bio_counter_inc_blocked(fs_info);
1382 while (cur < end) {
1383 struct btrfs_discard_stripe *stripes;
1384 unsigned int num_stripes;
1385 int i;
1386
1387 num_bytes = end - cur;
1388 stripes = btrfs_map_discard(fs_info, cur, &num_bytes, &num_stripes);
1389 if (IS_ERR(stripes)) {
1390 ret = PTR_ERR(stripes);
1391 if (ret == -EOPNOTSUPP)
1392 ret = 0;
1393 break;
1394 }
1395
1396 for (i = 0; i < num_stripes; i++) {
1397 struct btrfs_discard_stripe *stripe = stripes + i;
1398 u64 bytes;
1399
1400 if (!stripe->dev->bdev) {
1401 ASSERT(btrfs_test_opt(fs_info, DEGRADED));
1402 continue;
1403 }
1404
1405 if (!test_bit(BTRFS_DEV_STATE_WRITEABLE,
1406 &stripe->dev->dev_state))
1407 continue;
1408
1409 ret = do_discard_extent(stripe, &bytes);
1410 if (ret) {
1411 /*
1412 * Keep going if discard is not supported by the
1413 * device.
1414 */
1415 if (ret != -EOPNOTSUPP)
1416 break;
1417 ret = 0;
1418 } else {
1419 discarded_bytes += bytes;
1420 }
1421 }
1422 kfree(stripes);
1423 if (ret)
1424 break;
1425 cur += num_bytes;
1426 }
1427 btrfs_bio_counter_dec(fs_info);
1428 if (actual_bytes)
1429 *actual_bytes = discarded_bytes;
1430 return ret;
1431 }
1432
1433 /* Can return -ENOMEM */
btrfs_inc_extent_ref(struct btrfs_trans_handle * trans,struct btrfs_ref * generic_ref)1434 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1435 struct btrfs_ref *generic_ref)
1436 {
1437 struct btrfs_fs_info *fs_info = trans->fs_info;
1438 int ret;
1439
1440 ASSERT(generic_ref->type != BTRFS_REF_NOT_SET &&
1441 generic_ref->action);
1442 BUG_ON(generic_ref->type == BTRFS_REF_METADATA &&
1443 generic_ref->ref_root == BTRFS_TREE_LOG_OBJECTID);
1444
1445 if (generic_ref->type == BTRFS_REF_METADATA)
1446 ret = btrfs_add_delayed_tree_ref(trans, generic_ref, NULL);
1447 else
1448 ret = btrfs_add_delayed_data_ref(trans, generic_ref, 0);
1449
1450 btrfs_ref_tree_mod(fs_info, generic_ref);
1451
1452 return ret;
1453 }
1454
1455 /*
1456 * Insert backreference for a given extent.
1457 *
1458 * The counterpart is in __btrfs_free_extent(), with examples and more details
1459 * how it works.
1460 *
1461 * @trans: Handle of transaction
1462 *
1463 * @node: The delayed ref node used to get the bytenr/length for
1464 * extent whose references are incremented.
1465 *
1466 * @extent_op Pointer to a structure, holding information necessary when
1467 * updating a tree block's flags
1468 *
1469 */
__btrfs_inc_extent_ref(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_node * node,struct btrfs_delayed_extent_op * extent_op)1470 static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1471 struct btrfs_delayed_ref_node *node,
1472 struct btrfs_delayed_extent_op *extent_op)
1473 {
1474 struct btrfs_path *path;
1475 struct extent_buffer *leaf;
1476 struct btrfs_extent_item *item;
1477 struct btrfs_key key;
1478 u64 bytenr = node->bytenr;
1479 u64 num_bytes = node->num_bytes;
1480 u64 owner = btrfs_delayed_ref_owner(node);
1481 u64 offset = btrfs_delayed_ref_offset(node);
1482 u64 refs;
1483 int refs_to_add = node->ref_mod;
1484 int ret;
1485
1486 path = btrfs_alloc_path();
1487 if (!path)
1488 return -ENOMEM;
1489
1490 /* this will setup the path even if it fails to insert the back ref */
1491 ret = insert_inline_extent_backref(trans, path, bytenr, num_bytes,
1492 node->parent, node->ref_root, owner,
1493 offset, refs_to_add, extent_op);
1494 if ((ret < 0 && ret != -EAGAIN) || !ret)
1495 goto out;
1496
1497 /*
1498 * Ok we had -EAGAIN which means we didn't have space to insert and
1499 * inline extent ref, so just update the reference count and add a
1500 * normal backref.
1501 */
1502 leaf = path->nodes[0];
1503 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1504 item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1505 refs = btrfs_extent_refs(leaf, item);
1506 btrfs_set_extent_refs(leaf, item, refs + refs_to_add);
1507 if (extent_op)
1508 __run_delayed_extent_op(extent_op, leaf, item);
1509
1510 btrfs_mark_buffer_dirty(trans, leaf);
1511 btrfs_release_path(path);
1512
1513 /* now insert the actual backref */
1514 if (owner < BTRFS_FIRST_FREE_OBJECTID)
1515 ret = insert_tree_block_ref(trans, path, node, bytenr);
1516 else
1517 ret = insert_extent_data_ref(trans, path, node, bytenr);
1518
1519 if (ret)
1520 btrfs_abort_transaction(trans, ret);
1521 out:
1522 btrfs_free_path(path);
1523 return ret;
1524 }
1525
free_head_ref_squota_rsv(struct btrfs_fs_info * fs_info,struct btrfs_delayed_ref_head * href)1526 static void free_head_ref_squota_rsv(struct btrfs_fs_info *fs_info,
1527 struct btrfs_delayed_ref_head *href)
1528 {
1529 u64 root = href->owning_root;
1530
1531 /*
1532 * Don't check must_insert_reserved, as this is called from contexts
1533 * where it has already been unset.
1534 */
1535 if (btrfs_qgroup_mode(fs_info) != BTRFS_QGROUP_MODE_SIMPLE ||
1536 !href->is_data || !is_fstree(root))
1537 return;
1538
1539 btrfs_qgroup_free_refroot(fs_info, root, href->reserved_bytes,
1540 BTRFS_QGROUP_RSV_DATA);
1541 }
1542
run_delayed_data_ref(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * href,struct btrfs_delayed_ref_node * node,struct btrfs_delayed_extent_op * extent_op,bool insert_reserved)1543 static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
1544 struct btrfs_delayed_ref_head *href,
1545 struct btrfs_delayed_ref_node *node,
1546 struct btrfs_delayed_extent_op *extent_op,
1547 bool insert_reserved)
1548 {
1549 int ret = 0;
1550 u64 parent = 0;
1551 u64 flags = 0;
1552
1553 trace_run_delayed_data_ref(trans->fs_info, node);
1554
1555 if (node->type == BTRFS_SHARED_DATA_REF_KEY)
1556 parent = node->parent;
1557
1558 if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
1559 struct btrfs_key key;
1560 struct btrfs_squota_delta delta = {
1561 .root = href->owning_root,
1562 .num_bytes = node->num_bytes,
1563 .is_data = true,
1564 .is_inc = true,
1565 .generation = trans->transid,
1566 };
1567 u64 owner = btrfs_delayed_ref_owner(node);
1568 u64 offset = btrfs_delayed_ref_offset(node);
1569
1570 if (extent_op)
1571 flags |= extent_op->flags_to_set;
1572
1573 key.objectid = node->bytenr;
1574 key.type = BTRFS_EXTENT_ITEM_KEY;
1575 key.offset = node->num_bytes;
1576
1577 ret = alloc_reserved_file_extent(trans, parent, node->ref_root,
1578 flags, owner, offset, &key,
1579 node->ref_mod,
1580 href->owning_root);
1581 free_head_ref_squota_rsv(trans->fs_info, href);
1582 if (!ret)
1583 ret = btrfs_record_squota_delta(trans->fs_info, &delta);
1584 } else if (node->action == BTRFS_ADD_DELAYED_REF) {
1585 ret = __btrfs_inc_extent_ref(trans, node, extent_op);
1586 } else if (node->action == BTRFS_DROP_DELAYED_REF) {
1587 ret = __btrfs_free_extent(trans, href, node, extent_op);
1588 } else {
1589 BUG();
1590 }
1591 return ret;
1592 }
1593
__run_delayed_extent_op(struct btrfs_delayed_extent_op * extent_op,struct extent_buffer * leaf,struct btrfs_extent_item * ei)1594 static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
1595 struct extent_buffer *leaf,
1596 struct btrfs_extent_item *ei)
1597 {
1598 u64 flags = btrfs_extent_flags(leaf, ei);
1599 if (extent_op->update_flags) {
1600 flags |= extent_op->flags_to_set;
1601 btrfs_set_extent_flags(leaf, ei, flags);
1602 }
1603
1604 if (extent_op->update_key) {
1605 struct btrfs_tree_block_info *bi;
1606 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_TREE_BLOCK));
1607 bi = (struct btrfs_tree_block_info *)(ei + 1);
1608 btrfs_set_tree_block_key(leaf, bi, &extent_op->key);
1609 }
1610 }
1611
run_delayed_extent_op(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * head,struct btrfs_delayed_extent_op * extent_op)1612 static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
1613 struct btrfs_delayed_ref_head *head,
1614 struct btrfs_delayed_extent_op *extent_op)
1615 {
1616 struct btrfs_fs_info *fs_info = trans->fs_info;
1617 struct btrfs_root *root;
1618 struct btrfs_key key;
1619 struct btrfs_path *path;
1620 struct btrfs_extent_item *ei;
1621 struct extent_buffer *leaf;
1622 u32 item_size;
1623 int ret;
1624 int metadata = 1;
1625
1626 if (TRANS_ABORTED(trans))
1627 return 0;
1628
1629 if (!btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1630 metadata = 0;
1631
1632 path = btrfs_alloc_path();
1633 if (!path)
1634 return -ENOMEM;
1635
1636 key.objectid = head->bytenr;
1637
1638 if (metadata) {
1639 key.type = BTRFS_METADATA_ITEM_KEY;
1640 key.offset = head->level;
1641 } else {
1642 key.type = BTRFS_EXTENT_ITEM_KEY;
1643 key.offset = head->num_bytes;
1644 }
1645
1646 root = btrfs_extent_root(fs_info, key.objectid);
1647 again:
1648 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
1649 if (ret < 0) {
1650 goto out;
1651 } else if (ret > 0) {
1652 if (metadata) {
1653 if (path->slots[0] > 0) {
1654 path->slots[0]--;
1655 btrfs_item_key_to_cpu(path->nodes[0], &key,
1656 path->slots[0]);
1657 if (key.objectid == head->bytenr &&
1658 key.type == BTRFS_EXTENT_ITEM_KEY &&
1659 key.offset == head->num_bytes)
1660 ret = 0;
1661 }
1662 if (ret > 0) {
1663 btrfs_release_path(path);
1664 metadata = 0;
1665
1666 key.objectid = head->bytenr;
1667 key.offset = head->num_bytes;
1668 key.type = BTRFS_EXTENT_ITEM_KEY;
1669 goto again;
1670 }
1671 } else {
1672 ret = -EUCLEAN;
1673 btrfs_err(fs_info,
1674 "missing extent item for extent %llu num_bytes %llu level %d",
1675 head->bytenr, head->num_bytes, head->level);
1676 goto out;
1677 }
1678 }
1679
1680 leaf = path->nodes[0];
1681 item_size = btrfs_item_size(leaf, path->slots[0]);
1682
1683 if (unlikely(item_size < sizeof(*ei))) {
1684 ret = -EUCLEAN;
1685 btrfs_err(fs_info,
1686 "unexpected extent item size, has %u expect >= %zu",
1687 item_size, sizeof(*ei));
1688 btrfs_abort_transaction(trans, ret);
1689 goto out;
1690 }
1691
1692 ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1693 __run_delayed_extent_op(extent_op, leaf, ei);
1694
1695 btrfs_mark_buffer_dirty(trans, leaf);
1696 out:
1697 btrfs_free_path(path);
1698 return ret;
1699 }
1700
run_delayed_tree_ref(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * href,struct btrfs_delayed_ref_node * node,struct btrfs_delayed_extent_op * extent_op,bool insert_reserved)1701 static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
1702 struct btrfs_delayed_ref_head *href,
1703 struct btrfs_delayed_ref_node *node,
1704 struct btrfs_delayed_extent_op *extent_op,
1705 bool insert_reserved)
1706 {
1707 int ret = 0;
1708 struct btrfs_fs_info *fs_info = trans->fs_info;
1709 u64 parent = 0;
1710 u64 ref_root = 0;
1711
1712 trace_run_delayed_tree_ref(trans->fs_info, node);
1713
1714 if (node->type == BTRFS_SHARED_BLOCK_REF_KEY)
1715 parent = node->parent;
1716 ref_root = node->ref_root;
1717
1718 if (unlikely(node->ref_mod != 1)) {
1719 btrfs_err(trans->fs_info,
1720 "btree block %llu has %d references rather than 1: action %d ref_root %llu parent %llu",
1721 node->bytenr, node->ref_mod, node->action, ref_root,
1722 parent);
1723 return -EUCLEAN;
1724 }
1725 if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
1726 struct btrfs_squota_delta delta = {
1727 .root = href->owning_root,
1728 .num_bytes = fs_info->nodesize,
1729 .is_data = false,
1730 .is_inc = true,
1731 .generation = trans->transid,
1732 };
1733
1734 ret = alloc_reserved_tree_block(trans, node, extent_op);
1735 if (!ret)
1736 btrfs_record_squota_delta(fs_info, &delta);
1737 } else if (node->action == BTRFS_ADD_DELAYED_REF) {
1738 ret = __btrfs_inc_extent_ref(trans, node, extent_op);
1739 } else if (node->action == BTRFS_DROP_DELAYED_REF) {
1740 ret = __btrfs_free_extent(trans, href, node, extent_op);
1741 } else {
1742 BUG();
1743 }
1744 return ret;
1745 }
1746
1747 /* helper function to actually process a single delayed ref entry */
run_one_delayed_ref(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * href,struct btrfs_delayed_ref_node * node,struct btrfs_delayed_extent_op * extent_op,bool insert_reserved)1748 static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
1749 struct btrfs_delayed_ref_head *href,
1750 struct btrfs_delayed_ref_node *node,
1751 struct btrfs_delayed_extent_op *extent_op,
1752 bool insert_reserved)
1753 {
1754 int ret = 0;
1755
1756 if (TRANS_ABORTED(trans)) {
1757 if (insert_reserved) {
1758 btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
1759 free_head_ref_squota_rsv(trans->fs_info, href);
1760 }
1761 return 0;
1762 }
1763
1764 if (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
1765 node->type == BTRFS_SHARED_BLOCK_REF_KEY)
1766 ret = run_delayed_tree_ref(trans, href, node, extent_op,
1767 insert_reserved);
1768 else if (node->type == BTRFS_EXTENT_DATA_REF_KEY ||
1769 node->type == BTRFS_SHARED_DATA_REF_KEY)
1770 ret = run_delayed_data_ref(trans, href, node, extent_op,
1771 insert_reserved);
1772 else if (node->type == BTRFS_EXTENT_OWNER_REF_KEY)
1773 ret = 0;
1774 else
1775 BUG();
1776 if (ret && insert_reserved)
1777 btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
1778 if (ret < 0)
1779 btrfs_err(trans->fs_info,
1780 "failed to run delayed ref for logical %llu num_bytes %llu type %u action %u ref_mod %d: %d",
1781 node->bytenr, node->num_bytes, node->type,
1782 node->action, node->ref_mod, ret);
1783 return ret;
1784 }
1785
1786 static inline struct btrfs_delayed_ref_node *
select_delayed_ref(struct btrfs_delayed_ref_head * head)1787 select_delayed_ref(struct btrfs_delayed_ref_head *head)
1788 {
1789 struct btrfs_delayed_ref_node *ref;
1790
1791 if (RB_EMPTY_ROOT(&head->ref_tree.rb_root))
1792 return NULL;
1793
1794 /*
1795 * Select a delayed ref of type BTRFS_ADD_DELAYED_REF first.
1796 * This is to prevent a ref count from going down to zero, which deletes
1797 * the extent item from the extent tree, when there still are references
1798 * to add, which would fail because they would not find the extent item.
1799 */
1800 if (!list_empty(&head->ref_add_list))
1801 return list_first_entry(&head->ref_add_list,
1802 struct btrfs_delayed_ref_node, add_list);
1803
1804 ref = rb_entry(rb_first_cached(&head->ref_tree),
1805 struct btrfs_delayed_ref_node, ref_node);
1806 ASSERT(list_empty(&ref->add_list));
1807 return ref;
1808 }
1809
unselect_delayed_ref_head(struct btrfs_delayed_ref_root * delayed_refs,struct btrfs_delayed_ref_head * head)1810 static void unselect_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs,
1811 struct btrfs_delayed_ref_head *head)
1812 {
1813 spin_lock(&delayed_refs->lock);
1814 head->processing = false;
1815 delayed_refs->num_heads_ready++;
1816 spin_unlock(&delayed_refs->lock);
1817 btrfs_delayed_ref_unlock(head);
1818 }
1819
cleanup_extent_op(struct btrfs_delayed_ref_head * head)1820 static struct btrfs_delayed_extent_op *cleanup_extent_op(
1821 struct btrfs_delayed_ref_head *head)
1822 {
1823 struct btrfs_delayed_extent_op *extent_op = head->extent_op;
1824
1825 if (!extent_op)
1826 return NULL;
1827
1828 if (head->must_insert_reserved) {
1829 head->extent_op = NULL;
1830 btrfs_free_delayed_extent_op(extent_op);
1831 return NULL;
1832 }
1833 return extent_op;
1834 }
1835
run_and_cleanup_extent_op(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * head)1836 static int run_and_cleanup_extent_op(struct btrfs_trans_handle *trans,
1837 struct btrfs_delayed_ref_head *head)
1838 {
1839 struct btrfs_delayed_extent_op *extent_op;
1840 int ret;
1841
1842 extent_op = cleanup_extent_op(head);
1843 if (!extent_op)
1844 return 0;
1845 head->extent_op = NULL;
1846 spin_unlock(&head->lock);
1847 ret = run_delayed_extent_op(trans, head, extent_op);
1848 btrfs_free_delayed_extent_op(extent_op);
1849 return ret ? ret : 1;
1850 }
1851
btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info * fs_info,struct btrfs_delayed_ref_root * delayed_refs,struct btrfs_delayed_ref_head * head)1852 u64 btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
1853 struct btrfs_delayed_ref_root *delayed_refs,
1854 struct btrfs_delayed_ref_head *head)
1855 {
1856 u64 ret = 0;
1857
1858 /*
1859 * We had csum deletions accounted for in our delayed refs rsv, we need
1860 * to drop the csum leaves for this update from our delayed_refs_rsv.
1861 */
1862 if (head->total_ref_mod < 0 && head->is_data) {
1863 int nr_csums;
1864
1865 spin_lock(&delayed_refs->lock);
1866 delayed_refs->pending_csums -= head->num_bytes;
1867 spin_unlock(&delayed_refs->lock);
1868 nr_csums = btrfs_csum_bytes_to_leaves(fs_info, head->num_bytes);
1869
1870 btrfs_delayed_refs_rsv_release(fs_info, 0, nr_csums);
1871
1872 ret = btrfs_calc_delayed_ref_csum_bytes(fs_info, nr_csums);
1873 }
1874 /* must_insert_reserved can be set only if we didn't run the head ref. */
1875 if (head->must_insert_reserved)
1876 free_head_ref_squota_rsv(fs_info, head);
1877
1878 return ret;
1879 }
1880
cleanup_ref_head(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * head,u64 * bytes_released)1881 static int cleanup_ref_head(struct btrfs_trans_handle *trans,
1882 struct btrfs_delayed_ref_head *head,
1883 u64 *bytes_released)
1884 {
1885
1886 struct btrfs_fs_info *fs_info = trans->fs_info;
1887 struct btrfs_delayed_ref_root *delayed_refs;
1888 int ret;
1889
1890 delayed_refs = &trans->transaction->delayed_refs;
1891
1892 ret = run_and_cleanup_extent_op(trans, head);
1893 if (ret < 0) {
1894 unselect_delayed_ref_head(delayed_refs, head);
1895 btrfs_debug(fs_info, "run_delayed_extent_op returned %d", ret);
1896 return ret;
1897 } else if (ret) {
1898 return ret;
1899 }
1900
1901 /*
1902 * Need to drop our head ref lock and re-acquire the delayed ref lock
1903 * and then re-check to make sure nobody got added.
1904 */
1905 spin_unlock(&head->lock);
1906 spin_lock(&delayed_refs->lock);
1907 spin_lock(&head->lock);
1908 if (!RB_EMPTY_ROOT(&head->ref_tree.rb_root) || head->extent_op) {
1909 spin_unlock(&head->lock);
1910 spin_unlock(&delayed_refs->lock);
1911 return 1;
1912 }
1913 btrfs_delete_ref_head(delayed_refs, head);
1914 spin_unlock(&head->lock);
1915 spin_unlock(&delayed_refs->lock);
1916
1917 if (head->must_insert_reserved) {
1918 btrfs_pin_extent(trans, head->bytenr, head->num_bytes, 1);
1919 if (head->is_data) {
1920 struct btrfs_root *csum_root;
1921
1922 csum_root = btrfs_csum_root(fs_info, head->bytenr);
1923 ret = btrfs_del_csums(trans, csum_root, head->bytenr,
1924 head->num_bytes);
1925 }
1926 }
1927
1928 *bytes_released += btrfs_cleanup_ref_head_accounting(fs_info, delayed_refs, head);
1929
1930 trace_run_delayed_ref_head(fs_info, head, 0);
1931 btrfs_delayed_ref_unlock(head);
1932 btrfs_put_delayed_ref_head(head);
1933 return ret;
1934 }
1935
btrfs_obtain_ref_head(struct btrfs_trans_handle * trans)1936 static struct btrfs_delayed_ref_head *btrfs_obtain_ref_head(
1937 struct btrfs_trans_handle *trans)
1938 {
1939 struct btrfs_delayed_ref_root *delayed_refs =
1940 &trans->transaction->delayed_refs;
1941 struct btrfs_delayed_ref_head *head = NULL;
1942 int ret;
1943
1944 spin_lock(&delayed_refs->lock);
1945 head = btrfs_select_ref_head(delayed_refs);
1946 if (!head) {
1947 spin_unlock(&delayed_refs->lock);
1948 return head;
1949 }
1950
1951 /*
1952 * Grab the lock that says we are going to process all the refs for
1953 * this head
1954 */
1955 ret = btrfs_delayed_ref_lock(delayed_refs, head);
1956 spin_unlock(&delayed_refs->lock);
1957
1958 /*
1959 * We may have dropped the spin lock to get the head mutex lock, and
1960 * that might have given someone else time to free the head. If that's
1961 * true, it has been removed from our list and we can move on.
1962 */
1963 if (ret == -EAGAIN)
1964 head = ERR_PTR(-EAGAIN);
1965
1966 return head;
1967 }
1968
btrfs_run_delayed_refs_for_head(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * locked_ref,u64 * bytes_released)1969 static int btrfs_run_delayed_refs_for_head(struct btrfs_trans_handle *trans,
1970 struct btrfs_delayed_ref_head *locked_ref,
1971 u64 *bytes_released)
1972 {
1973 struct btrfs_fs_info *fs_info = trans->fs_info;
1974 struct btrfs_delayed_ref_root *delayed_refs;
1975 struct btrfs_delayed_extent_op *extent_op;
1976 struct btrfs_delayed_ref_node *ref;
1977 bool must_insert_reserved;
1978 int ret;
1979
1980 delayed_refs = &trans->transaction->delayed_refs;
1981
1982 lockdep_assert_held(&locked_ref->mutex);
1983 lockdep_assert_held(&locked_ref->lock);
1984
1985 while ((ref = select_delayed_ref(locked_ref))) {
1986 if (ref->seq &&
1987 btrfs_check_delayed_seq(fs_info, ref->seq)) {
1988 spin_unlock(&locked_ref->lock);
1989 unselect_delayed_ref_head(delayed_refs, locked_ref);
1990 return -EAGAIN;
1991 }
1992
1993 rb_erase_cached(&ref->ref_node, &locked_ref->ref_tree);
1994 RB_CLEAR_NODE(&ref->ref_node);
1995 if (!list_empty(&ref->add_list))
1996 list_del(&ref->add_list);
1997 /*
1998 * When we play the delayed ref, also correct the ref_mod on
1999 * head
2000 */
2001 switch (ref->action) {
2002 case BTRFS_ADD_DELAYED_REF:
2003 case BTRFS_ADD_DELAYED_EXTENT:
2004 locked_ref->ref_mod -= ref->ref_mod;
2005 break;
2006 case BTRFS_DROP_DELAYED_REF:
2007 locked_ref->ref_mod += ref->ref_mod;
2008 break;
2009 default:
2010 WARN_ON(1);
2011 }
2012 atomic_dec(&delayed_refs->num_entries);
2013
2014 /*
2015 * Record the must_insert_reserved flag before we drop the
2016 * spin lock.
2017 */
2018 must_insert_reserved = locked_ref->must_insert_reserved;
2019 /*
2020 * Unsetting this on the head ref relinquishes ownership of
2021 * the rsv_bytes, so it is critical that every possible code
2022 * path from here forward frees all reserves including qgroup
2023 * reserve.
2024 */
2025 locked_ref->must_insert_reserved = false;
2026
2027 extent_op = locked_ref->extent_op;
2028 locked_ref->extent_op = NULL;
2029 spin_unlock(&locked_ref->lock);
2030
2031 ret = run_one_delayed_ref(trans, locked_ref, ref, extent_op,
2032 must_insert_reserved);
2033 btrfs_delayed_refs_rsv_release(fs_info, 1, 0);
2034 *bytes_released += btrfs_calc_delayed_ref_bytes(fs_info, 1);
2035
2036 btrfs_free_delayed_extent_op(extent_op);
2037 if (ret) {
2038 unselect_delayed_ref_head(delayed_refs, locked_ref);
2039 btrfs_put_delayed_ref(ref);
2040 return ret;
2041 }
2042
2043 btrfs_put_delayed_ref(ref);
2044 cond_resched();
2045
2046 spin_lock(&locked_ref->lock);
2047 btrfs_merge_delayed_refs(fs_info, delayed_refs, locked_ref);
2048 }
2049
2050 return 0;
2051 }
2052
2053 /*
2054 * Returns 0 on success or if called with an already aborted transaction.
2055 * Returns -ENOMEM or -EIO on failure and will abort the transaction.
2056 */
__btrfs_run_delayed_refs(struct btrfs_trans_handle * trans,u64 min_bytes)2057 static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
2058 u64 min_bytes)
2059 {
2060 struct btrfs_fs_info *fs_info = trans->fs_info;
2061 struct btrfs_delayed_ref_root *delayed_refs;
2062 struct btrfs_delayed_ref_head *locked_ref = NULL;
2063 int ret;
2064 unsigned long count = 0;
2065 unsigned long max_count = 0;
2066 u64 bytes_processed = 0;
2067
2068 delayed_refs = &trans->transaction->delayed_refs;
2069 if (min_bytes == 0) {
2070 max_count = delayed_refs->num_heads_ready;
2071 min_bytes = U64_MAX;
2072 }
2073
2074 do {
2075 if (!locked_ref) {
2076 locked_ref = btrfs_obtain_ref_head(trans);
2077 if (IS_ERR_OR_NULL(locked_ref)) {
2078 if (PTR_ERR(locked_ref) == -EAGAIN) {
2079 continue;
2080 } else {
2081 break;
2082 }
2083 }
2084 count++;
2085 }
2086 /*
2087 * We need to try and merge add/drops of the same ref since we
2088 * can run into issues with relocate dropping the implicit ref
2089 * and then it being added back again before the drop can
2090 * finish. If we merged anything we need to re-loop so we can
2091 * get a good ref.
2092 * Or we can get node references of the same type that weren't
2093 * merged when created due to bumps in the tree mod seq, and
2094 * we need to merge them to prevent adding an inline extent
2095 * backref before dropping it (triggering a BUG_ON at
2096 * insert_inline_extent_backref()).
2097 */
2098 spin_lock(&locked_ref->lock);
2099 btrfs_merge_delayed_refs(fs_info, delayed_refs, locked_ref);
2100
2101 ret = btrfs_run_delayed_refs_for_head(trans, locked_ref, &bytes_processed);
2102 if (ret < 0 && ret != -EAGAIN) {
2103 /*
2104 * Error, btrfs_run_delayed_refs_for_head already
2105 * unlocked everything so just bail out
2106 */
2107 return ret;
2108 } else if (!ret) {
2109 /*
2110 * Success, perform the usual cleanup of a processed
2111 * head
2112 */
2113 ret = cleanup_ref_head(trans, locked_ref, &bytes_processed);
2114 if (ret > 0 ) {
2115 /* We dropped our lock, we need to loop. */
2116 ret = 0;
2117 continue;
2118 } else if (ret) {
2119 return ret;
2120 }
2121 }
2122
2123 /*
2124 * Either success case or btrfs_run_delayed_refs_for_head
2125 * returned -EAGAIN, meaning we need to select another head
2126 */
2127
2128 locked_ref = NULL;
2129 cond_resched();
2130 } while ((min_bytes != U64_MAX && bytes_processed < min_bytes) ||
2131 (max_count > 0 && count < max_count) ||
2132 locked_ref);
2133
2134 return 0;
2135 }
2136
2137 #ifdef SCRAMBLE_DELAYED_REFS
2138 /*
2139 * Normally delayed refs get processed in ascending bytenr order. This
2140 * correlates in most cases to the order added. To expose dependencies on this
2141 * order, we start to process the tree in the middle instead of the beginning
2142 */
find_middle(struct rb_root * root)2143 static u64 find_middle(struct rb_root *root)
2144 {
2145 struct rb_node *n = root->rb_node;
2146 struct btrfs_delayed_ref_node *entry;
2147 int alt = 1;
2148 u64 middle;
2149 u64 first = 0, last = 0;
2150
2151 n = rb_first(root);
2152 if (n) {
2153 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2154 first = entry->bytenr;
2155 }
2156 n = rb_last(root);
2157 if (n) {
2158 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2159 last = entry->bytenr;
2160 }
2161 n = root->rb_node;
2162
2163 while (n) {
2164 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2165 WARN_ON(!entry->in_tree);
2166
2167 middle = entry->bytenr;
2168
2169 if (alt)
2170 n = n->rb_left;
2171 else
2172 n = n->rb_right;
2173
2174 alt = 1 - alt;
2175 }
2176 return middle;
2177 }
2178 #endif
2179
2180 /*
2181 * Start processing the delayed reference count updates and extent insertions
2182 * we have queued up so far.
2183 *
2184 * @trans: Transaction handle.
2185 * @min_bytes: How many bytes of delayed references to process. After this
2186 * many bytes we stop processing delayed references if there are
2187 * any more. If 0 it means to run all existing delayed references,
2188 * but not new ones added after running all existing ones.
2189 * Use (u64)-1 (U64_MAX) to run all existing delayed references
2190 * plus any new ones that are added.
2191 *
2192 * Returns 0 on success or if called with an aborted transaction
2193 * Returns <0 on error and aborts the transaction
2194 */
btrfs_run_delayed_refs(struct btrfs_trans_handle * trans,u64 min_bytes)2195 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans, u64 min_bytes)
2196 {
2197 struct btrfs_fs_info *fs_info = trans->fs_info;
2198 struct btrfs_delayed_ref_root *delayed_refs;
2199 int ret;
2200
2201 /* We'll clean this up in btrfs_cleanup_transaction */
2202 if (TRANS_ABORTED(trans))
2203 return 0;
2204
2205 if (test_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags))
2206 return 0;
2207
2208 delayed_refs = &trans->transaction->delayed_refs;
2209 again:
2210 #ifdef SCRAMBLE_DELAYED_REFS
2211 delayed_refs->run_delayed_start = find_middle(&delayed_refs->root);
2212 #endif
2213 ret = __btrfs_run_delayed_refs(trans, min_bytes);
2214 if (ret < 0) {
2215 btrfs_abort_transaction(trans, ret);
2216 return ret;
2217 }
2218
2219 if (min_bytes == U64_MAX) {
2220 btrfs_create_pending_block_groups(trans);
2221
2222 spin_lock(&delayed_refs->lock);
2223 if (RB_EMPTY_ROOT(&delayed_refs->href_root.rb_root)) {
2224 spin_unlock(&delayed_refs->lock);
2225 return 0;
2226 }
2227 spin_unlock(&delayed_refs->lock);
2228
2229 cond_resched();
2230 goto again;
2231 }
2232
2233 return 0;
2234 }
2235
btrfs_set_disk_extent_flags(struct btrfs_trans_handle * trans,struct extent_buffer * eb,u64 flags)2236 int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
2237 struct extent_buffer *eb, u64 flags)
2238 {
2239 struct btrfs_delayed_extent_op *extent_op;
2240 int ret;
2241
2242 extent_op = btrfs_alloc_delayed_extent_op();
2243 if (!extent_op)
2244 return -ENOMEM;
2245
2246 extent_op->flags_to_set = flags;
2247 extent_op->update_flags = true;
2248 extent_op->update_key = false;
2249
2250 ret = btrfs_add_delayed_extent_op(trans, eb->start, eb->len,
2251 btrfs_header_level(eb), extent_op);
2252 if (ret)
2253 btrfs_free_delayed_extent_op(extent_op);
2254 return ret;
2255 }
2256
check_delayed_ref(struct btrfs_root * root,struct btrfs_path * path,u64 objectid,u64 offset,u64 bytenr)2257 static noinline int check_delayed_ref(struct btrfs_root *root,
2258 struct btrfs_path *path,
2259 u64 objectid, u64 offset, u64 bytenr)
2260 {
2261 struct btrfs_delayed_ref_head *head;
2262 struct btrfs_delayed_ref_node *ref;
2263 struct btrfs_delayed_ref_root *delayed_refs;
2264 struct btrfs_transaction *cur_trans;
2265 struct rb_node *node;
2266 int ret = 0;
2267
2268 spin_lock(&root->fs_info->trans_lock);
2269 cur_trans = root->fs_info->running_transaction;
2270 if (cur_trans)
2271 refcount_inc(&cur_trans->use_count);
2272 spin_unlock(&root->fs_info->trans_lock);
2273 if (!cur_trans)
2274 return 0;
2275
2276 delayed_refs = &cur_trans->delayed_refs;
2277 spin_lock(&delayed_refs->lock);
2278 head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
2279 if (!head) {
2280 spin_unlock(&delayed_refs->lock);
2281 btrfs_put_transaction(cur_trans);
2282 return 0;
2283 }
2284
2285 if (!mutex_trylock(&head->mutex)) {
2286 if (path->nowait) {
2287 spin_unlock(&delayed_refs->lock);
2288 btrfs_put_transaction(cur_trans);
2289 return -EAGAIN;
2290 }
2291
2292 refcount_inc(&head->refs);
2293 spin_unlock(&delayed_refs->lock);
2294
2295 btrfs_release_path(path);
2296
2297 /*
2298 * Mutex was contended, block until it's released and let
2299 * caller try again
2300 */
2301 mutex_lock(&head->mutex);
2302 mutex_unlock(&head->mutex);
2303 btrfs_put_delayed_ref_head(head);
2304 btrfs_put_transaction(cur_trans);
2305 return -EAGAIN;
2306 }
2307 spin_unlock(&delayed_refs->lock);
2308
2309 spin_lock(&head->lock);
2310 /*
2311 * XXX: We should replace this with a proper search function in the
2312 * future.
2313 */
2314 for (node = rb_first_cached(&head->ref_tree); node;
2315 node = rb_next(node)) {
2316 u64 ref_owner;
2317 u64 ref_offset;
2318
2319 ref = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
2320 /* If it's a shared ref we know a cross reference exists */
2321 if (ref->type != BTRFS_EXTENT_DATA_REF_KEY) {
2322 ret = 1;
2323 break;
2324 }
2325
2326 ref_owner = btrfs_delayed_ref_owner(ref);
2327 ref_offset = btrfs_delayed_ref_offset(ref);
2328
2329 /*
2330 * If our ref doesn't match the one we're currently looking at
2331 * then we have a cross reference.
2332 */
2333 if (ref->ref_root != btrfs_root_id(root) ||
2334 ref_owner != objectid || ref_offset != offset) {
2335 ret = 1;
2336 break;
2337 }
2338 }
2339 spin_unlock(&head->lock);
2340 mutex_unlock(&head->mutex);
2341 btrfs_put_transaction(cur_trans);
2342 return ret;
2343 }
2344
check_committed_ref(struct btrfs_root * root,struct btrfs_path * path,u64 objectid,u64 offset,u64 bytenr,bool strict)2345 static noinline int check_committed_ref(struct btrfs_root *root,
2346 struct btrfs_path *path,
2347 u64 objectid, u64 offset, u64 bytenr,
2348 bool strict)
2349 {
2350 struct btrfs_fs_info *fs_info = root->fs_info;
2351 struct btrfs_root *extent_root = btrfs_extent_root(fs_info, bytenr);
2352 struct extent_buffer *leaf;
2353 struct btrfs_extent_data_ref *ref;
2354 struct btrfs_extent_inline_ref *iref;
2355 struct btrfs_extent_item *ei;
2356 struct btrfs_key key;
2357 u32 item_size;
2358 u32 expected_size;
2359 int type;
2360 int ret;
2361
2362 key.objectid = bytenr;
2363 key.offset = (u64)-1;
2364 key.type = BTRFS_EXTENT_ITEM_KEY;
2365
2366 ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
2367 if (ret < 0)
2368 goto out;
2369 if (ret == 0) {
2370 /*
2371 * Key with offset -1 found, there would have to exist an extent
2372 * item with such offset, but this is out of the valid range.
2373 */
2374 ret = -EUCLEAN;
2375 goto out;
2376 }
2377
2378 ret = -ENOENT;
2379 if (path->slots[0] == 0)
2380 goto out;
2381
2382 path->slots[0]--;
2383 leaf = path->nodes[0];
2384 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
2385
2386 if (key.objectid != bytenr || key.type != BTRFS_EXTENT_ITEM_KEY)
2387 goto out;
2388
2389 ret = 1;
2390 item_size = btrfs_item_size(leaf, path->slots[0]);
2391 ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
2392 expected_size = sizeof(*ei) + btrfs_extent_inline_ref_size(BTRFS_EXTENT_DATA_REF_KEY);
2393
2394 /* No inline refs; we need to bail before checking for owner ref. */
2395 if (item_size == sizeof(*ei))
2396 goto out;
2397
2398 /* Check for an owner ref; skip over it to the real inline refs. */
2399 iref = (struct btrfs_extent_inline_ref *)(ei + 1);
2400 type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
2401 if (btrfs_fs_incompat(fs_info, SIMPLE_QUOTA) && type == BTRFS_EXTENT_OWNER_REF_KEY) {
2402 expected_size += btrfs_extent_inline_ref_size(BTRFS_EXTENT_OWNER_REF_KEY);
2403 iref = (struct btrfs_extent_inline_ref *)(iref + 1);
2404 }
2405
2406 /* If extent item has more than 1 inline ref then it's shared */
2407 if (item_size != expected_size)
2408 goto out;
2409
2410 /*
2411 * If extent created before last snapshot => it's shared unless the
2412 * snapshot has been deleted. Use the heuristic if strict is false.
2413 */
2414 if (!strict &&
2415 (btrfs_extent_generation(leaf, ei) <=
2416 btrfs_root_last_snapshot(&root->root_item)))
2417 goto out;
2418
2419 /* If this extent has SHARED_DATA_REF then it's shared */
2420 type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
2421 if (type != BTRFS_EXTENT_DATA_REF_KEY)
2422 goto out;
2423
2424 ref = (struct btrfs_extent_data_ref *)(&iref->offset);
2425 if (btrfs_extent_refs(leaf, ei) !=
2426 btrfs_extent_data_ref_count(leaf, ref) ||
2427 btrfs_extent_data_ref_root(leaf, ref) != btrfs_root_id(root) ||
2428 btrfs_extent_data_ref_objectid(leaf, ref) != objectid ||
2429 btrfs_extent_data_ref_offset(leaf, ref) != offset)
2430 goto out;
2431
2432 ret = 0;
2433 out:
2434 return ret;
2435 }
2436
btrfs_cross_ref_exist(struct btrfs_root * root,u64 objectid,u64 offset,u64 bytenr,bool strict,struct btrfs_path * path)2437 int btrfs_cross_ref_exist(struct btrfs_root *root, u64 objectid, u64 offset,
2438 u64 bytenr, bool strict, struct btrfs_path *path)
2439 {
2440 int ret;
2441
2442 do {
2443 ret = check_committed_ref(root, path, objectid,
2444 offset, bytenr, strict);
2445 if (ret && ret != -ENOENT)
2446 goto out;
2447
2448 ret = check_delayed_ref(root, path, objectid, offset, bytenr);
2449 } while (ret == -EAGAIN);
2450
2451 out:
2452 btrfs_release_path(path);
2453 if (btrfs_is_data_reloc_root(root))
2454 WARN_ON(ret > 0);
2455 return ret;
2456 }
2457
__btrfs_mod_ref(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct extent_buffer * buf,int full_backref,int inc)2458 static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
2459 struct btrfs_root *root,
2460 struct extent_buffer *buf,
2461 int full_backref, int inc)
2462 {
2463 struct btrfs_fs_info *fs_info = root->fs_info;
2464 u64 parent;
2465 u64 ref_root;
2466 u32 nritems;
2467 struct btrfs_key key;
2468 struct btrfs_file_extent_item *fi;
2469 bool for_reloc = btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC);
2470 int i;
2471 int action;
2472 int level;
2473 int ret = 0;
2474
2475 if (btrfs_is_testing(fs_info))
2476 return 0;
2477
2478 ref_root = btrfs_header_owner(buf);
2479 nritems = btrfs_header_nritems(buf);
2480 level = btrfs_header_level(buf);
2481
2482 if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state) && level == 0)
2483 return 0;
2484
2485 if (full_backref)
2486 parent = buf->start;
2487 else
2488 parent = 0;
2489 if (inc)
2490 action = BTRFS_ADD_DELAYED_REF;
2491 else
2492 action = BTRFS_DROP_DELAYED_REF;
2493
2494 for (i = 0; i < nritems; i++) {
2495 struct btrfs_ref ref = {
2496 .action = action,
2497 .parent = parent,
2498 .ref_root = ref_root,
2499 };
2500
2501 if (level == 0) {
2502 btrfs_item_key_to_cpu(buf, &key, i);
2503 if (key.type != BTRFS_EXTENT_DATA_KEY)
2504 continue;
2505 fi = btrfs_item_ptr(buf, i,
2506 struct btrfs_file_extent_item);
2507 if (btrfs_file_extent_type(buf, fi) ==
2508 BTRFS_FILE_EXTENT_INLINE)
2509 continue;
2510 ref.bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
2511 if (ref.bytenr == 0)
2512 continue;
2513
2514 ref.num_bytes = btrfs_file_extent_disk_num_bytes(buf, fi);
2515 ref.owning_root = ref_root;
2516
2517 key.offset -= btrfs_file_extent_offset(buf, fi);
2518 btrfs_init_data_ref(&ref, key.objectid, key.offset,
2519 btrfs_root_id(root), for_reloc);
2520 if (inc)
2521 ret = btrfs_inc_extent_ref(trans, &ref);
2522 else
2523 ret = btrfs_free_extent(trans, &ref);
2524 if (ret)
2525 goto fail;
2526 } else {
2527 /* We don't know the owning_root, leave as 0. */
2528 ref.bytenr = btrfs_node_blockptr(buf, i);
2529 ref.num_bytes = fs_info->nodesize;
2530
2531 btrfs_init_tree_ref(&ref, level - 1,
2532 btrfs_root_id(root), for_reloc);
2533 if (inc)
2534 ret = btrfs_inc_extent_ref(trans, &ref);
2535 else
2536 ret = btrfs_free_extent(trans, &ref);
2537 if (ret)
2538 goto fail;
2539 }
2540 }
2541 return 0;
2542 fail:
2543 return ret;
2544 }
2545
btrfs_inc_ref(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct extent_buffer * buf,int full_backref)2546 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2547 struct extent_buffer *buf, int full_backref)
2548 {
2549 return __btrfs_mod_ref(trans, root, buf, full_backref, 1);
2550 }
2551
btrfs_dec_ref(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct extent_buffer * buf,int full_backref)2552 int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2553 struct extent_buffer *buf, int full_backref)
2554 {
2555 return __btrfs_mod_ref(trans, root, buf, full_backref, 0);
2556 }
2557
get_alloc_profile_by_root(struct btrfs_root * root,int data)2558 static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
2559 {
2560 struct btrfs_fs_info *fs_info = root->fs_info;
2561 u64 flags;
2562 u64 ret;
2563
2564 if (data)
2565 flags = BTRFS_BLOCK_GROUP_DATA;
2566 else if (root == fs_info->chunk_root)
2567 flags = BTRFS_BLOCK_GROUP_SYSTEM;
2568 else
2569 flags = BTRFS_BLOCK_GROUP_METADATA;
2570
2571 ret = btrfs_get_alloc_profile(fs_info, flags);
2572 return ret;
2573 }
2574
first_logical_byte(struct btrfs_fs_info * fs_info)2575 static u64 first_logical_byte(struct btrfs_fs_info *fs_info)
2576 {
2577 struct rb_node *leftmost;
2578 u64 bytenr = 0;
2579
2580 read_lock(&fs_info->block_group_cache_lock);
2581 /* Get the block group with the lowest logical start address. */
2582 leftmost = rb_first_cached(&fs_info->block_group_cache_tree);
2583 if (leftmost) {
2584 struct btrfs_block_group *bg;
2585
2586 bg = rb_entry(leftmost, struct btrfs_block_group, cache_node);
2587 bytenr = bg->start;
2588 }
2589 read_unlock(&fs_info->block_group_cache_lock);
2590
2591 return bytenr;
2592 }
2593
pin_down_extent(struct btrfs_trans_handle * trans,struct btrfs_block_group * cache,u64 bytenr,u64 num_bytes,int reserved)2594 static int pin_down_extent(struct btrfs_trans_handle *trans,
2595 struct btrfs_block_group *cache,
2596 u64 bytenr, u64 num_bytes, int reserved)
2597 {
2598 struct btrfs_fs_info *fs_info = cache->fs_info;
2599
2600 spin_lock(&cache->space_info->lock);
2601 spin_lock(&cache->lock);
2602 cache->pinned += num_bytes;
2603 btrfs_space_info_update_bytes_pinned(fs_info, cache->space_info,
2604 num_bytes);
2605 if (reserved) {
2606 cache->reserved -= num_bytes;
2607 cache->space_info->bytes_reserved -= num_bytes;
2608 }
2609 spin_unlock(&cache->lock);
2610 spin_unlock(&cache->space_info->lock);
2611
2612 set_extent_bit(&trans->transaction->pinned_extents, bytenr,
2613 bytenr + num_bytes - 1, EXTENT_DIRTY, NULL);
2614 return 0;
2615 }
2616
btrfs_pin_extent(struct btrfs_trans_handle * trans,u64 bytenr,u64 num_bytes,int reserved)2617 int btrfs_pin_extent(struct btrfs_trans_handle *trans,
2618 u64 bytenr, u64 num_bytes, int reserved)
2619 {
2620 struct btrfs_block_group *cache;
2621
2622 cache = btrfs_lookup_block_group(trans->fs_info, bytenr);
2623 BUG_ON(!cache); /* Logic error */
2624
2625 pin_down_extent(trans, cache, bytenr, num_bytes, reserved);
2626
2627 btrfs_put_block_group(cache);
2628 return 0;
2629 }
2630
btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle * trans,const struct extent_buffer * eb)2631 int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans,
2632 const struct extent_buffer *eb)
2633 {
2634 struct btrfs_block_group *cache;
2635 int ret;
2636
2637 cache = btrfs_lookup_block_group(trans->fs_info, eb->start);
2638 if (!cache)
2639 return -EINVAL;
2640
2641 /*
2642 * Fully cache the free space first so that our pin removes the free space
2643 * from the cache.
2644 */
2645 ret = btrfs_cache_block_group(cache, true);
2646 if (ret)
2647 goto out;
2648
2649 pin_down_extent(trans, cache, eb->start, eb->len, 0);
2650
2651 /* remove us from the free space cache (if we're there at all) */
2652 ret = btrfs_remove_free_space(cache, eb->start, eb->len);
2653 out:
2654 btrfs_put_block_group(cache);
2655 return ret;
2656 }
2657
__exclude_logged_extent(struct btrfs_fs_info * fs_info,u64 start,u64 num_bytes)2658 static int __exclude_logged_extent(struct btrfs_fs_info *fs_info,
2659 u64 start, u64 num_bytes)
2660 {
2661 int ret;
2662 struct btrfs_block_group *block_group;
2663
2664 block_group = btrfs_lookup_block_group(fs_info, start);
2665 if (!block_group)
2666 return -EINVAL;
2667
2668 ret = btrfs_cache_block_group(block_group, true);
2669 if (ret)
2670 goto out;
2671
2672 ret = btrfs_remove_free_space(block_group, start, num_bytes);
2673 out:
2674 btrfs_put_block_group(block_group);
2675 return ret;
2676 }
2677
btrfs_exclude_logged_extents(struct extent_buffer * eb)2678 int btrfs_exclude_logged_extents(struct extent_buffer *eb)
2679 {
2680 struct btrfs_fs_info *fs_info = eb->fs_info;
2681 struct btrfs_file_extent_item *item;
2682 struct btrfs_key key;
2683 int found_type;
2684 int i;
2685 int ret = 0;
2686
2687 if (!btrfs_fs_incompat(fs_info, MIXED_GROUPS))
2688 return 0;
2689
2690 for (i = 0; i < btrfs_header_nritems(eb); i++) {
2691 btrfs_item_key_to_cpu(eb, &key, i);
2692 if (key.type != BTRFS_EXTENT_DATA_KEY)
2693 continue;
2694 item = btrfs_item_ptr(eb, i, struct btrfs_file_extent_item);
2695 found_type = btrfs_file_extent_type(eb, item);
2696 if (found_type == BTRFS_FILE_EXTENT_INLINE)
2697 continue;
2698 if (btrfs_file_extent_disk_bytenr(eb, item) == 0)
2699 continue;
2700 key.objectid = btrfs_file_extent_disk_bytenr(eb, item);
2701 key.offset = btrfs_file_extent_disk_num_bytes(eb, item);
2702 ret = __exclude_logged_extent(fs_info, key.objectid, key.offset);
2703 if (ret)
2704 break;
2705 }
2706
2707 return ret;
2708 }
2709
2710 static void
btrfs_inc_block_group_reservations(struct btrfs_block_group * bg)2711 btrfs_inc_block_group_reservations(struct btrfs_block_group *bg)
2712 {
2713 atomic_inc(&bg->reservations);
2714 }
2715
2716 /*
2717 * Returns the free cluster for the given space info and sets empty_cluster to
2718 * what it should be based on the mount options.
2719 */
2720 static struct btrfs_free_cluster *
fetch_cluster_info(struct btrfs_fs_info * fs_info,struct btrfs_space_info * space_info,u64 * empty_cluster)2721 fetch_cluster_info(struct btrfs_fs_info *fs_info,
2722 struct btrfs_space_info *space_info, u64 *empty_cluster)
2723 {
2724 struct btrfs_free_cluster *ret = NULL;
2725
2726 *empty_cluster = 0;
2727 if (btrfs_mixed_space_info(space_info))
2728 return ret;
2729
2730 if (space_info->flags & BTRFS_BLOCK_GROUP_METADATA) {
2731 ret = &fs_info->meta_alloc_cluster;
2732 if (btrfs_test_opt(fs_info, SSD))
2733 *empty_cluster = SZ_2M;
2734 else
2735 *empty_cluster = SZ_64K;
2736 } else if ((space_info->flags & BTRFS_BLOCK_GROUP_DATA) &&
2737 btrfs_test_opt(fs_info, SSD_SPREAD)) {
2738 *empty_cluster = SZ_2M;
2739 ret = &fs_info->data_alloc_cluster;
2740 }
2741
2742 return ret;
2743 }
2744
unpin_extent_range(struct btrfs_fs_info * fs_info,u64 start,u64 end,const bool return_free_space)2745 static int unpin_extent_range(struct btrfs_fs_info *fs_info,
2746 u64 start, u64 end,
2747 const bool return_free_space)
2748 {
2749 struct btrfs_block_group *cache = NULL;
2750 struct btrfs_space_info *space_info;
2751 struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
2752 struct btrfs_free_cluster *cluster = NULL;
2753 u64 len;
2754 u64 total_unpinned = 0;
2755 u64 empty_cluster = 0;
2756 bool readonly;
2757 int ret = 0;
2758
2759 while (start <= end) {
2760 readonly = false;
2761 if (!cache ||
2762 start >= cache->start + cache->length) {
2763 if (cache)
2764 btrfs_put_block_group(cache);
2765 total_unpinned = 0;
2766 cache = btrfs_lookup_block_group(fs_info, start);
2767 if (cache == NULL) {
2768 /* Logic error, something removed the block group. */
2769 ret = -EUCLEAN;
2770 goto out;
2771 }
2772
2773 cluster = fetch_cluster_info(fs_info,
2774 cache->space_info,
2775 &empty_cluster);
2776 empty_cluster <<= 1;
2777 }
2778
2779 len = cache->start + cache->length - start;
2780 len = min(len, end + 1 - start);
2781
2782 if (return_free_space)
2783 btrfs_add_free_space(cache, start, len);
2784
2785 start += len;
2786 total_unpinned += len;
2787 space_info = cache->space_info;
2788
2789 /*
2790 * If this space cluster has been marked as fragmented and we've
2791 * unpinned enough in this block group to potentially allow a
2792 * cluster to be created inside of it go ahead and clear the
2793 * fragmented check.
2794 */
2795 if (cluster && cluster->fragmented &&
2796 total_unpinned > empty_cluster) {
2797 spin_lock(&cluster->lock);
2798 cluster->fragmented = 0;
2799 spin_unlock(&cluster->lock);
2800 }
2801
2802 spin_lock(&space_info->lock);
2803 spin_lock(&cache->lock);
2804 cache->pinned -= len;
2805 btrfs_space_info_update_bytes_pinned(fs_info, space_info, -len);
2806 space_info->max_extent_size = 0;
2807 if (cache->ro) {
2808 space_info->bytes_readonly += len;
2809 readonly = true;
2810 } else if (btrfs_is_zoned(fs_info)) {
2811 /* Need reset before reusing in a zoned block group */
2812 btrfs_space_info_update_bytes_zone_unusable(fs_info, space_info,
2813 len);
2814 readonly = true;
2815 }
2816 spin_unlock(&cache->lock);
2817 if (!readonly && return_free_space &&
2818 global_rsv->space_info == space_info) {
2819 spin_lock(&global_rsv->lock);
2820 if (!global_rsv->full) {
2821 u64 to_add = min(len, global_rsv->size -
2822 global_rsv->reserved);
2823
2824 global_rsv->reserved += to_add;
2825 btrfs_space_info_update_bytes_may_use(fs_info,
2826 space_info, to_add);
2827 if (global_rsv->reserved >= global_rsv->size)
2828 global_rsv->full = 1;
2829 len -= to_add;
2830 }
2831 spin_unlock(&global_rsv->lock);
2832 }
2833 /* Add to any tickets we may have */
2834 if (!readonly && return_free_space && len)
2835 btrfs_try_granting_tickets(fs_info, space_info);
2836 spin_unlock(&space_info->lock);
2837 }
2838
2839 if (cache)
2840 btrfs_put_block_group(cache);
2841 out:
2842 return ret;
2843 }
2844
btrfs_finish_extent_commit(struct btrfs_trans_handle * trans)2845 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
2846 {
2847 struct btrfs_fs_info *fs_info = trans->fs_info;
2848 struct btrfs_block_group *block_group, *tmp;
2849 struct list_head *deleted_bgs;
2850 struct extent_io_tree *unpin;
2851 u64 start;
2852 u64 end;
2853 int ret;
2854
2855 unpin = &trans->transaction->pinned_extents;
2856
2857 while (!TRANS_ABORTED(trans)) {
2858 struct extent_state *cached_state = NULL;
2859
2860 mutex_lock(&fs_info->unused_bg_unpin_mutex);
2861 if (!find_first_extent_bit(unpin, 0, &start, &end,
2862 EXTENT_DIRTY, &cached_state)) {
2863 mutex_unlock(&fs_info->unused_bg_unpin_mutex);
2864 break;
2865 }
2866
2867 if (btrfs_test_opt(fs_info, DISCARD_SYNC))
2868 ret = btrfs_discard_extent(fs_info, start,
2869 end + 1 - start, NULL);
2870
2871 clear_extent_dirty(unpin, start, end, &cached_state);
2872 ret = unpin_extent_range(fs_info, start, end, true);
2873 BUG_ON(ret);
2874 mutex_unlock(&fs_info->unused_bg_unpin_mutex);
2875 free_extent_state(cached_state);
2876 cond_resched();
2877 }
2878
2879 if (btrfs_test_opt(fs_info, DISCARD_ASYNC)) {
2880 btrfs_discard_calc_delay(&fs_info->discard_ctl);
2881 btrfs_discard_schedule_work(&fs_info->discard_ctl, true);
2882 }
2883
2884 /*
2885 * Transaction is finished. We don't need the lock anymore. We
2886 * do need to clean up the block groups in case of a transaction
2887 * abort.
2888 */
2889 deleted_bgs = &trans->transaction->deleted_bgs;
2890 list_for_each_entry_safe(block_group, tmp, deleted_bgs, bg_list) {
2891 u64 trimmed = 0;
2892
2893 ret = -EROFS;
2894 if (!TRANS_ABORTED(trans))
2895 ret = btrfs_discard_extent(fs_info,
2896 block_group->start,
2897 block_group->length,
2898 &trimmed);
2899
2900 list_del_init(&block_group->bg_list);
2901 btrfs_unfreeze_block_group(block_group);
2902 btrfs_put_block_group(block_group);
2903
2904 if (ret) {
2905 const char *errstr = btrfs_decode_error(ret);
2906 btrfs_warn(fs_info,
2907 "discard failed while removing blockgroup: errno=%d %s",
2908 ret, errstr);
2909 }
2910 }
2911
2912 return 0;
2913 }
2914
2915 /*
2916 * Parse an extent item's inline extents looking for a simple quotas owner ref.
2917 *
2918 * @fs_info: the btrfs_fs_info for this mount
2919 * @leaf: a leaf in the extent tree containing the extent item
2920 * @slot: the slot in the leaf where the extent item is found
2921 *
2922 * Returns the objectid of the root that originally allocated the extent item
2923 * if the inline owner ref is expected and present, otherwise 0.
2924 *
2925 * If an extent item has an owner ref item, it will be the first inline ref
2926 * item. Therefore the logic is to check whether there are any inline ref
2927 * items, then check the type of the first one.
2928 */
btrfs_get_extent_owner_root(struct btrfs_fs_info * fs_info,struct extent_buffer * leaf,int slot)2929 u64 btrfs_get_extent_owner_root(struct btrfs_fs_info *fs_info,
2930 struct extent_buffer *leaf, int slot)
2931 {
2932 struct btrfs_extent_item *ei;
2933 struct btrfs_extent_inline_ref *iref;
2934 struct btrfs_extent_owner_ref *oref;
2935 unsigned long ptr;
2936 unsigned long end;
2937 int type;
2938
2939 if (!btrfs_fs_incompat(fs_info, SIMPLE_QUOTA))
2940 return 0;
2941
2942 ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
2943 ptr = (unsigned long)(ei + 1);
2944 end = (unsigned long)ei + btrfs_item_size(leaf, slot);
2945
2946 /* No inline ref items of any kind, can't check type. */
2947 if (ptr == end)
2948 return 0;
2949
2950 iref = (struct btrfs_extent_inline_ref *)ptr;
2951 type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_ANY);
2952
2953 /* We found an owner ref, get the root out of it. */
2954 if (type == BTRFS_EXTENT_OWNER_REF_KEY) {
2955 oref = (struct btrfs_extent_owner_ref *)(&iref->offset);
2956 return btrfs_extent_owner_ref_root_id(leaf, oref);
2957 }
2958
2959 /* We have inline refs, but not an owner ref. */
2960 return 0;
2961 }
2962
do_free_extent_accounting(struct btrfs_trans_handle * trans,u64 bytenr,struct btrfs_squota_delta * delta)2963 static int do_free_extent_accounting(struct btrfs_trans_handle *trans,
2964 u64 bytenr, struct btrfs_squota_delta *delta)
2965 {
2966 int ret;
2967 u64 num_bytes = delta->num_bytes;
2968
2969 if (delta->is_data) {
2970 struct btrfs_root *csum_root;
2971
2972 csum_root = btrfs_csum_root(trans->fs_info, bytenr);
2973 ret = btrfs_del_csums(trans, csum_root, bytenr, num_bytes);
2974 if (ret) {
2975 btrfs_abort_transaction(trans, ret);
2976 return ret;
2977 }
2978
2979 ret = btrfs_delete_raid_extent(trans, bytenr, num_bytes);
2980 if (ret) {
2981 btrfs_abort_transaction(trans, ret);
2982 return ret;
2983 }
2984 }
2985
2986 ret = btrfs_record_squota_delta(trans->fs_info, delta);
2987 if (ret) {
2988 btrfs_abort_transaction(trans, ret);
2989 return ret;
2990 }
2991
2992 ret = add_to_free_space_tree(trans, bytenr, num_bytes);
2993 if (ret) {
2994 btrfs_abort_transaction(trans, ret);
2995 return ret;
2996 }
2997
2998 ret = btrfs_update_block_group(trans, bytenr, num_bytes, false);
2999 if (ret)
3000 btrfs_abort_transaction(trans, ret);
3001
3002 return ret;
3003 }
3004
3005 #define abort_and_dump(trans, path, fmt, args...) \
3006 ({ \
3007 btrfs_abort_transaction(trans, -EUCLEAN); \
3008 btrfs_print_leaf(path->nodes[0]); \
3009 btrfs_crit(trans->fs_info, fmt, ##args); \
3010 })
3011
3012 /*
3013 * Drop one or more refs of @node.
3014 *
3015 * 1. Locate the extent refs.
3016 * It's either inline in EXTENT/METADATA_ITEM or in keyed SHARED_* item.
3017 * Locate it, then reduce the refs number or remove the ref line completely.
3018 *
3019 * 2. Update the refs count in EXTENT/METADATA_ITEM
3020 *
3021 * Inline backref case:
3022 *
3023 * in extent tree we have:
3024 *
3025 * item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
3026 * refs 2 gen 6 flags DATA
3027 * extent data backref root FS_TREE objectid 258 offset 0 count 1
3028 * extent data backref root FS_TREE objectid 257 offset 0 count 1
3029 *
3030 * This function gets called with:
3031 *
3032 * node->bytenr = 13631488
3033 * node->num_bytes = 1048576
3034 * root_objectid = FS_TREE
3035 * owner_objectid = 257
3036 * owner_offset = 0
3037 * refs_to_drop = 1
3038 *
3039 * Then we should get some like:
3040 *
3041 * item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
3042 * refs 1 gen 6 flags DATA
3043 * extent data backref root FS_TREE objectid 258 offset 0 count 1
3044 *
3045 * Keyed backref case:
3046 *
3047 * in extent tree we have:
3048 *
3049 * item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
3050 * refs 754 gen 6 flags DATA
3051 * [...]
3052 * item 2 key (13631488 EXTENT_DATA_REF <HASH>) itemoff 3915 itemsize 28
3053 * extent data backref root FS_TREE objectid 866 offset 0 count 1
3054 *
3055 * This function get called with:
3056 *
3057 * node->bytenr = 13631488
3058 * node->num_bytes = 1048576
3059 * root_objectid = FS_TREE
3060 * owner_objectid = 866
3061 * owner_offset = 0
3062 * refs_to_drop = 1
3063 *
3064 * Then we should get some like:
3065 *
3066 * item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
3067 * refs 753 gen 6 flags DATA
3068 *
3069 * And that (13631488 EXTENT_DATA_REF <HASH>) gets removed.
3070 */
__btrfs_free_extent(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_head * href,struct btrfs_delayed_ref_node * node,struct btrfs_delayed_extent_op * extent_op)3071 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
3072 struct btrfs_delayed_ref_head *href,
3073 struct btrfs_delayed_ref_node *node,
3074 struct btrfs_delayed_extent_op *extent_op)
3075 {
3076 struct btrfs_fs_info *info = trans->fs_info;
3077 struct btrfs_key key;
3078 struct btrfs_path *path;
3079 struct btrfs_root *extent_root;
3080 struct extent_buffer *leaf;
3081 struct btrfs_extent_item *ei;
3082 struct btrfs_extent_inline_ref *iref;
3083 int ret;
3084 int is_data;
3085 int extent_slot = 0;
3086 int found_extent = 0;
3087 int num_to_del = 1;
3088 int refs_to_drop = node->ref_mod;
3089 u32 item_size;
3090 u64 refs;
3091 u64 bytenr = node->bytenr;
3092 u64 num_bytes = node->num_bytes;
3093 u64 owner_objectid = btrfs_delayed_ref_owner(node);
3094 u64 owner_offset = btrfs_delayed_ref_offset(node);
3095 bool skinny_metadata = btrfs_fs_incompat(info, SKINNY_METADATA);
3096 u64 delayed_ref_root = href->owning_root;
3097
3098 extent_root = btrfs_extent_root(info, bytenr);
3099 ASSERT(extent_root);
3100
3101 path = btrfs_alloc_path();
3102 if (!path)
3103 return -ENOMEM;
3104
3105 is_data = owner_objectid >= BTRFS_FIRST_FREE_OBJECTID;
3106
3107 if (!is_data && refs_to_drop != 1) {
3108 btrfs_crit(info,
3109 "invalid refs_to_drop, dropping more than 1 refs for tree block %llu refs_to_drop %u",
3110 node->bytenr, refs_to_drop);
3111 ret = -EINVAL;
3112 btrfs_abort_transaction(trans, ret);
3113 goto out;
3114 }
3115
3116 if (is_data)
3117 skinny_metadata = false;
3118
3119 ret = lookup_extent_backref(trans, path, &iref, bytenr, num_bytes,
3120 node->parent, node->ref_root, owner_objectid,
3121 owner_offset);
3122 if (ret == 0) {
3123 /*
3124 * Either the inline backref or the SHARED_DATA_REF/
3125 * SHARED_BLOCK_REF is found
3126 *
3127 * Here is a quick path to locate EXTENT/METADATA_ITEM.
3128 * It's possible the EXTENT/METADATA_ITEM is near current slot.
3129 */
3130 extent_slot = path->slots[0];
3131 while (extent_slot >= 0) {
3132 btrfs_item_key_to_cpu(path->nodes[0], &key,
3133 extent_slot);
3134 if (key.objectid != bytenr)
3135 break;
3136 if (key.type == BTRFS_EXTENT_ITEM_KEY &&
3137 key.offset == num_bytes) {
3138 found_extent = 1;
3139 break;
3140 }
3141 if (key.type == BTRFS_METADATA_ITEM_KEY &&
3142 key.offset == owner_objectid) {
3143 found_extent = 1;
3144 break;
3145 }
3146
3147 /* Quick path didn't find the EXTEMT/METADATA_ITEM */
3148 if (path->slots[0] - extent_slot > 5)
3149 break;
3150 extent_slot--;
3151 }
3152
3153 if (!found_extent) {
3154 if (iref) {
3155 abort_and_dump(trans, path,
3156 "invalid iref slot %u, no EXTENT/METADATA_ITEM found but has inline extent ref",
3157 path->slots[0]);
3158 ret = -EUCLEAN;
3159 goto out;
3160 }
3161 /* Must be SHARED_* item, remove the backref first */
3162 ret = remove_extent_backref(trans, extent_root, path,
3163 NULL, refs_to_drop, is_data);
3164 if (ret) {
3165 btrfs_abort_transaction(trans, ret);
3166 goto out;
3167 }
3168 btrfs_release_path(path);
3169
3170 /* Slow path to locate EXTENT/METADATA_ITEM */
3171 key.objectid = bytenr;
3172 key.type = BTRFS_EXTENT_ITEM_KEY;
3173 key.offset = num_bytes;
3174
3175 if (!is_data && skinny_metadata) {
3176 key.type = BTRFS_METADATA_ITEM_KEY;
3177 key.offset = owner_objectid;
3178 }
3179
3180 ret = btrfs_search_slot(trans, extent_root,
3181 &key, path, -1, 1);
3182 if (ret > 0 && skinny_metadata && path->slots[0]) {
3183 /*
3184 * Couldn't find our skinny metadata item,
3185 * see if we have ye olde extent item.
3186 */
3187 path->slots[0]--;
3188 btrfs_item_key_to_cpu(path->nodes[0], &key,
3189 path->slots[0]);
3190 if (key.objectid == bytenr &&
3191 key.type == BTRFS_EXTENT_ITEM_KEY &&
3192 key.offset == num_bytes)
3193 ret = 0;
3194 }
3195
3196 if (ret > 0 && skinny_metadata) {
3197 skinny_metadata = false;
3198 key.objectid = bytenr;
3199 key.type = BTRFS_EXTENT_ITEM_KEY;
3200 key.offset = num_bytes;
3201 btrfs_release_path(path);
3202 ret = btrfs_search_slot(trans, extent_root,
3203 &key, path, -1, 1);
3204 }
3205
3206 if (ret) {
3207 if (ret > 0)
3208 btrfs_print_leaf(path->nodes[0]);
3209 btrfs_err(info,
3210 "umm, got %d back from search, was looking for %llu, slot %d",
3211 ret, bytenr, path->slots[0]);
3212 }
3213 if (ret < 0) {
3214 btrfs_abort_transaction(trans, ret);
3215 goto out;
3216 }
3217 extent_slot = path->slots[0];
3218 }
3219 } else if (WARN_ON(ret == -ENOENT)) {
3220 abort_and_dump(trans, path,
3221 "unable to find ref byte nr %llu parent %llu root %llu owner %llu offset %llu slot %d",
3222 bytenr, node->parent, node->ref_root, owner_objectid,
3223 owner_offset, path->slots[0]);
3224 goto out;
3225 } else {
3226 btrfs_abort_transaction(trans, ret);
3227 goto out;
3228 }
3229
3230 leaf = path->nodes[0];
3231 item_size = btrfs_item_size(leaf, extent_slot);
3232 if (unlikely(item_size < sizeof(*ei))) {
3233 ret = -EUCLEAN;
3234 btrfs_err(trans->fs_info,
3235 "unexpected extent item size, has %u expect >= %zu",
3236 item_size, sizeof(*ei));
3237 btrfs_abort_transaction(trans, ret);
3238 goto out;
3239 }
3240 ei = btrfs_item_ptr(leaf, extent_slot,
3241 struct btrfs_extent_item);
3242 if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID &&
3243 key.type == BTRFS_EXTENT_ITEM_KEY) {
3244 struct btrfs_tree_block_info *bi;
3245
3246 if (item_size < sizeof(*ei) + sizeof(*bi)) {
3247 abort_and_dump(trans, path,
3248 "invalid extent item size for key (%llu, %u, %llu) slot %u owner %llu, has %u expect >= %zu",
3249 key.objectid, key.type, key.offset,
3250 path->slots[0], owner_objectid, item_size,
3251 sizeof(*ei) + sizeof(*bi));
3252 ret = -EUCLEAN;
3253 goto out;
3254 }
3255 bi = (struct btrfs_tree_block_info *)(ei + 1);
3256 WARN_ON(owner_objectid != btrfs_tree_block_level(leaf, bi));
3257 }
3258
3259 refs = btrfs_extent_refs(leaf, ei);
3260 if (refs < refs_to_drop) {
3261 abort_and_dump(trans, path,
3262 "trying to drop %d refs but we only have %llu for bytenr %llu slot %u",
3263 refs_to_drop, refs, bytenr, path->slots[0]);
3264 ret = -EUCLEAN;
3265 goto out;
3266 }
3267 refs -= refs_to_drop;
3268
3269 if (refs > 0) {
3270 if (extent_op)
3271 __run_delayed_extent_op(extent_op, leaf, ei);
3272 /*
3273 * In the case of inline back ref, reference count will
3274 * be updated by remove_extent_backref
3275 */
3276 if (iref) {
3277 if (!found_extent) {
3278 abort_and_dump(trans, path,
3279 "invalid iref, got inlined extent ref but no EXTENT/METADATA_ITEM found, slot %u",
3280 path->slots[0]);
3281 ret = -EUCLEAN;
3282 goto out;
3283 }
3284 } else {
3285 btrfs_set_extent_refs(leaf, ei, refs);
3286 btrfs_mark_buffer_dirty(trans, leaf);
3287 }
3288 if (found_extent) {
3289 ret = remove_extent_backref(trans, extent_root, path,
3290 iref, refs_to_drop, is_data);
3291 if (ret) {
3292 btrfs_abort_transaction(trans, ret);
3293 goto out;
3294 }
3295 }
3296 } else {
3297 struct btrfs_squota_delta delta = {
3298 .root = delayed_ref_root,
3299 .num_bytes = num_bytes,
3300 .is_data = is_data,
3301 .is_inc = false,
3302 .generation = btrfs_extent_generation(leaf, ei),
3303 };
3304
3305 /* In this branch refs == 1 */
3306 if (found_extent) {
3307 if (is_data && refs_to_drop !=
3308 extent_data_ref_count(path, iref)) {
3309 abort_and_dump(trans, path,
3310 "invalid refs_to_drop, current refs %u refs_to_drop %u slot %u",
3311 extent_data_ref_count(path, iref),
3312 refs_to_drop, path->slots[0]);
3313 ret = -EUCLEAN;
3314 goto out;
3315 }
3316 if (iref) {
3317 if (path->slots[0] != extent_slot) {
3318 abort_and_dump(trans, path,
3319 "invalid iref, extent item key (%llu %u %llu) slot %u doesn't have wanted iref",
3320 key.objectid, key.type,
3321 key.offset, path->slots[0]);
3322 ret = -EUCLEAN;
3323 goto out;
3324 }
3325 } else {
3326 /*
3327 * No inline ref, we must be at SHARED_* item,
3328 * And it's single ref, it must be:
3329 * | extent_slot ||extent_slot + 1|
3330 * [ EXTENT/METADATA_ITEM ][ SHARED_* ITEM ]
3331 */
3332 if (path->slots[0] != extent_slot + 1) {
3333 abort_and_dump(trans, path,
3334 "invalid SHARED_* item slot %u, previous item is not EXTENT/METADATA_ITEM",
3335 path->slots[0]);
3336 ret = -EUCLEAN;
3337 goto out;
3338 }
3339 path->slots[0] = extent_slot;
3340 num_to_del = 2;
3341 }
3342 }
3343 /*
3344 * We can't infer the data owner from the delayed ref, so we need
3345 * to try to get it from the owning ref item.
3346 *
3347 * If it is not present, then that extent was not written under
3348 * simple quotas mode, so we don't need to account for its deletion.
3349 */
3350 if (is_data)
3351 delta.root = btrfs_get_extent_owner_root(trans->fs_info,
3352 leaf, extent_slot);
3353
3354 ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
3355 num_to_del);
3356 if (ret) {
3357 btrfs_abort_transaction(trans, ret);
3358 goto out;
3359 }
3360 btrfs_release_path(path);
3361
3362 ret = do_free_extent_accounting(trans, bytenr, &delta);
3363 }
3364 btrfs_release_path(path);
3365
3366 out:
3367 btrfs_free_path(path);
3368 return ret;
3369 }
3370
3371 /*
3372 * when we free an block, it is possible (and likely) that we free the last
3373 * delayed ref for that extent as well. This searches the delayed ref tree for
3374 * a given extent, and if there are no other delayed refs to be processed, it
3375 * removes it from the tree.
3376 */
check_ref_cleanup(struct btrfs_trans_handle * trans,u64 bytenr)3377 static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans,
3378 u64 bytenr)
3379 {
3380 struct btrfs_delayed_ref_head *head;
3381 struct btrfs_delayed_ref_root *delayed_refs;
3382 int ret = 0;
3383
3384 delayed_refs = &trans->transaction->delayed_refs;
3385 spin_lock(&delayed_refs->lock);
3386 head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
3387 if (!head)
3388 goto out_delayed_unlock;
3389
3390 spin_lock(&head->lock);
3391 if (!RB_EMPTY_ROOT(&head->ref_tree.rb_root))
3392 goto out;
3393
3394 if (cleanup_extent_op(head) != NULL)
3395 goto out;
3396
3397 /*
3398 * waiting for the lock here would deadlock. If someone else has it
3399 * locked they are already in the process of dropping it anyway
3400 */
3401 if (!mutex_trylock(&head->mutex))
3402 goto out;
3403
3404 btrfs_delete_ref_head(delayed_refs, head);
3405 head->processing = false;
3406
3407 spin_unlock(&head->lock);
3408 spin_unlock(&delayed_refs->lock);
3409
3410 BUG_ON(head->extent_op);
3411 if (head->must_insert_reserved)
3412 ret = 1;
3413
3414 btrfs_cleanup_ref_head_accounting(trans->fs_info, delayed_refs, head);
3415 mutex_unlock(&head->mutex);
3416 btrfs_put_delayed_ref_head(head);
3417 return ret;
3418 out:
3419 spin_unlock(&head->lock);
3420
3421 out_delayed_unlock:
3422 spin_unlock(&delayed_refs->lock);
3423 return 0;
3424 }
3425
btrfs_free_tree_block(struct btrfs_trans_handle * trans,u64 root_id,struct extent_buffer * buf,u64 parent,int last_ref)3426 int btrfs_free_tree_block(struct btrfs_trans_handle *trans,
3427 u64 root_id,
3428 struct extent_buffer *buf,
3429 u64 parent, int last_ref)
3430 {
3431 struct btrfs_fs_info *fs_info = trans->fs_info;
3432 struct btrfs_block_group *bg;
3433 int ret;
3434
3435 if (root_id != BTRFS_TREE_LOG_OBJECTID) {
3436 struct btrfs_ref generic_ref = {
3437 .action = BTRFS_DROP_DELAYED_REF,
3438 .bytenr = buf->start,
3439 .num_bytes = buf->len,
3440 .parent = parent,
3441 .owning_root = btrfs_header_owner(buf),
3442 .ref_root = root_id,
3443 };
3444
3445 /*
3446 * Assert that the extent buffer is not cleared due to
3447 * EXTENT_BUFFER_ZONED_ZEROOUT. Please refer
3448 * btrfs_clear_buffer_dirty() and btree_csum_one_bio() for
3449 * detail.
3450 */
3451 ASSERT(btrfs_header_bytenr(buf) != 0);
3452
3453 btrfs_init_tree_ref(&generic_ref, btrfs_header_level(buf), 0, false);
3454 btrfs_ref_tree_mod(fs_info, &generic_ref);
3455 ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, NULL);
3456 if (ret < 0)
3457 return ret;
3458 }
3459
3460 if (!last_ref)
3461 return 0;
3462
3463 if (btrfs_header_generation(buf) != trans->transid)
3464 goto out;
3465
3466 if (root_id != BTRFS_TREE_LOG_OBJECTID) {
3467 ret = check_ref_cleanup(trans, buf->start);
3468 if (!ret)
3469 goto out;
3470 }
3471
3472 bg = btrfs_lookup_block_group(fs_info, buf->start);
3473
3474 if (btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
3475 pin_down_extent(trans, bg, buf->start, buf->len, 1);
3476 btrfs_put_block_group(bg);
3477 goto out;
3478 }
3479
3480 /*
3481 * If there are tree mod log users we may have recorded mod log
3482 * operations for this node. If we re-allocate this node we
3483 * could replay operations on this node that happened when it
3484 * existed in a completely different root. For example if it
3485 * was part of root A, then was reallocated to root B, and we
3486 * are doing a btrfs_old_search_slot(root b), we could replay
3487 * operations that happened when the block was part of root A,
3488 * giving us an inconsistent view of the btree.
3489 *
3490 * We are safe from races here because at this point no other
3491 * node or root points to this extent buffer, so if after this
3492 * check a new tree mod log user joins we will not have an
3493 * existing log of operations on this node that we have to
3494 * contend with.
3495 */
3496
3497 if (test_bit(BTRFS_FS_TREE_MOD_LOG_USERS, &fs_info->flags)
3498 || btrfs_is_zoned(fs_info)) {
3499 pin_down_extent(trans, bg, buf->start, buf->len, 1);
3500 btrfs_put_block_group(bg);
3501 goto out;
3502 }
3503
3504 WARN_ON(test_bit(EXTENT_BUFFER_DIRTY, &buf->bflags));
3505
3506 btrfs_add_free_space(bg, buf->start, buf->len);
3507 btrfs_free_reserved_bytes(bg, buf->len, 0);
3508 btrfs_put_block_group(bg);
3509 trace_btrfs_reserved_extent_free(fs_info, buf->start, buf->len);
3510
3511 out:
3512
3513 /*
3514 * Deleting the buffer, clear the corrupt flag since it doesn't
3515 * matter anymore.
3516 */
3517 clear_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags);
3518 return 0;
3519 }
3520
3521 /* Can return -ENOMEM */
btrfs_free_extent(struct btrfs_trans_handle * trans,struct btrfs_ref * ref)3522 int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_ref *ref)
3523 {
3524 struct btrfs_fs_info *fs_info = trans->fs_info;
3525 int ret;
3526
3527 if (btrfs_is_testing(fs_info))
3528 return 0;
3529
3530 /*
3531 * tree log blocks never actually go into the extent allocation
3532 * tree, just update pinning info and exit early.
3533 */
3534 if (ref->ref_root == BTRFS_TREE_LOG_OBJECTID) {
3535 btrfs_pin_extent(trans, ref->bytenr, ref->num_bytes, 1);
3536 ret = 0;
3537 } else if (ref->type == BTRFS_REF_METADATA) {
3538 ret = btrfs_add_delayed_tree_ref(trans, ref, NULL);
3539 } else {
3540 ret = btrfs_add_delayed_data_ref(trans, ref, 0);
3541 }
3542
3543 if (ref->ref_root != BTRFS_TREE_LOG_OBJECTID)
3544 btrfs_ref_tree_mod(fs_info, ref);
3545
3546 return ret;
3547 }
3548
3549 enum btrfs_loop_type {
3550 /*
3551 * Start caching block groups but do not wait for progress or for them
3552 * to be done.
3553 */
3554 LOOP_CACHING_NOWAIT,
3555
3556 /*
3557 * Wait for the block group free_space >= the space we're waiting for if
3558 * the block group isn't cached.
3559 */
3560 LOOP_CACHING_WAIT,
3561
3562 /*
3563 * Allow allocations to happen from block groups that do not yet have a
3564 * size classification.
3565 */
3566 LOOP_UNSET_SIZE_CLASS,
3567
3568 /*
3569 * Allocate a chunk and then retry the allocation.
3570 */
3571 LOOP_ALLOC_CHUNK,
3572
3573 /*
3574 * Ignore the size class restrictions for this allocation.
3575 */
3576 LOOP_WRONG_SIZE_CLASS,
3577
3578 /*
3579 * Ignore the empty size, only try to allocate the number of bytes
3580 * needed for this allocation.
3581 */
3582 LOOP_NO_EMPTY_SIZE,
3583 };
3584
3585 static inline void
btrfs_lock_block_group(struct btrfs_block_group * cache,int delalloc)3586 btrfs_lock_block_group(struct btrfs_block_group *cache,
3587 int delalloc)
3588 {
3589 if (delalloc)
3590 down_read(&cache->data_rwsem);
3591 }
3592
btrfs_grab_block_group(struct btrfs_block_group * cache,int delalloc)3593 static inline void btrfs_grab_block_group(struct btrfs_block_group *cache,
3594 int delalloc)
3595 {
3596 btrfs_get_block_group(cache);
3597 if (delalloc)
3598 down_read(&cache->data_rwsem);
3599 }
3600
btrfs_lock_cluster(struct btrfs_block_group * block_group,struct btrfs_free_cluster * cluster,int delalloc)3601 static struct btrfs_block_group *btrfs_lock_cluster(
3602 struct btrfs_block_group *block_group,
3603 struct btrfs_free_cluster *cluster,
3604 int delalloc)
3605 __acquires(&cluster->refill_lock)
3606 {
3607 struct btrfs_block_group *used_bg = NULL;
3608
3609 spin_lock(&cluster->refill_lock);
3610 while (1) {
3611 used_bg = cluster->block_group;
3612 if (!used_bg)
3613 return NULL;
3614
3615 if (used_bg == block_group)
3616 return used_bg;
3617
3618 btrfs_get_block_group(used_bg);
3619
3620 if (!delalloc)
3621 return used_bg;
3622
3623 if (down_read_trylock(&used_bg->data_rwsem))
3624 return used_bg;
3625
3626 spin_unlock(&cluster->refill_lock);
3627
3628 /* We should only have one-level nested. */
3629 down_read_nested(&used_bg->data_rwsem, SINGLE_DEPTH_NESTING);
3630
3631 spin_lock(&cluster->refill_lock);
3632 if (used_bg == cluster->block_group)
3633 return used_bg;
3634
3635 up_read(&used_bg->data_rwsem);
3636 btrfs_put_block_group(used_bg);
3637 }
3638 }
3639
3640 static inline void
btrfs_release_block_group(struct btrfs_block_group * cache,int delalloc)3641 btrfs_release_block_group(struct btrfs_block_group *cache,
3642 int delalloc)
3643 {
3644 if (delalloc)
3645 up_read(&cache->data_rwsem);
3646 btrfs_put_block_group(cache);
3647 }
3648
3649 /*
3650 * Helper function for find_free_extent().
3651 *
3652 * Return -ENOENT to inform caller that we need fallback to unclustered mode.
3653 * Return >0 to inform caller that we find nothing
3654 * Return 0 means we have found a location and set ffe_ctl->found_offset.
3655 */
find_free_extent_clustered(struct btrfs_block_group * bg,struct find_free_extent_ctl * ffe_ctl,struct btrfs_block_group ** cluster_bg_ret)3656 static int find_free_extent_clustered(struct btrfs_block_group *bg,
3657 struct find_free_extent_ctl *ffe_ctl,
3658 struct btrfs_block_group **cluster_bg_ret)
3659 {
3660 struct btrfs_block_group *cluster_bg;
3661 struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
3662 u64 aligned_cluster;
3663 u64 offset;
3664 int ret;
3665
3666 cluster_bg = btrfs_lock_cluster(bg, last_ptr, ffe_ctl->delalloc);
3667 if (!cluster_bg)
3668 goto refill_cluster;
3669 if (cluster_bg != bg && (cluster_bg->ro ||
3670 !block_group_bits(cluster_bg, ffe_ctl->flags)))
3671 goto release_cluster;
3672
3673 offset = btrfs_alloc_from_cluster(cluster_bg, last_ptr,
3674 ffe_ctl->num_bytes, cluster_bg->start,
3675 &ffe_ctl->max_extent_size);
3676 if (offset) {
3677 /* We have a block, we're done */
3678 spin_unlock(&last_ptr->refill_lock);
3679 trace_btrfs_reserve_extent_cluster(cluster_bg, ffe_ctl);
3680 *cluster_bg_ret = cluster_bg;
3681 ffe_ctl->found_offset = offset;
3682 return 0;
3683 }
3684 WARN_ON(last_ptr->block_group != cluster_bg);
3685
3686 release_cluster:
3687 /*
3688 * If we are on LOOP_NO_EMPTY_SIZE, we can't set up a new clusters, so
3689 * lets just skip it and let the allocator find whatever block it can
3690 * find. If we reach this point, we will have tried the cluster
3691 * allocator plenty of times and not have found anything, so we are
3692 * likely way too fragmented for the clustering stuff to find anything.
3693 *
3694 * However, if the cluster is taken from the current block group,
3695 * release the cluster first, so that we stand a better chance of
3696 * succeeding in the unclustered allocation.
3697 */
3698 if (ffe_ctl->loop >= LOOP_NO_EMPTY_SIZE && cluster_bg != bg) {
3699 spin_unlock(&last_ptr->refill_lock);
3700 btrfs_release_block_group(cluster_bg, ffe_ctl->delalloc);
3701 return -ENOENT;
3702 }
3703
3704 /* This cluster didn't work out, free it and start over */
3705 btrfs_return_cluster_to_free_space(NULL, last_ptr);
3706
3707 if (cluster_bg != bg)
3708 btrfs_release_block_group(cluster_bg, ffe_ctl->delalloc);
3709
3710 refill_cluster:
3711 if (ffe_ctl->loop >= LOOP_NO_EMPTY_SIZE) {
3712 spin_unlock(&last_ptr->refill_lock);
3713 return -ENOENT;
3714 }
3715
3716 aligned_cluster = max_t(u64,
3717 ffe_ctl->empty_cluster + ffe_ctl->empty_size,
3718 bg->full_stripe_len);
3719 ret = btrfs_find_space_cluster(bg, last_ptr, ffe_ctl->search_start,
3720 ffe_ctl->num_bytes, aligned_cluster);
3721 if (ret == 0) {
3722 /* Now pull our allocation out of this cluster */
3723 offset = btrfs_alloc_from_cluster(bg, last_ptr,
3724 ffe_ctl->num_bytes, ffe_ctl->search_start,
3725 &ffe_ctl->max_extent_size);
3726 if (offset) {
3727 /* We found one, proceed */
3728 spin_unlock(&last_ptr->refill_lock);
3729 ffe_ctl->found_offset = offset;
3730 trace_btrfs_reserve_extent_cluster(bg, ffe_ctl);
3731 return 0;
3732 }
3733 }
3734 /*
3735 * At this point we either didn't find a cluster or we weren't able to
3736 * allocate a block from our cluster. Free the cluster we've been
3737 * trying to use, and go to the next block group.
3738 */
3739 btrfs_return_cluster_to_free_space(NULL, last_ptr);
3740 spin_unlock(&last_ptr->refill_lock);
3741 return 1;
3742 }
3743
3744 /*
3745 * Return >0 to inform caller that we find nothing
3746 * Return 0 when we found an free extent and set ffe_ctrl->found_offset
3747 */
find_free_extent_unclustered(struct btrfs_block_group * bg,struct find_free_extent_ctl * ffe_ctl)3748 static int find_free_extent_unclustered(struct btrfs_block_group *bg,
3749 struct find_free_extent_ctl *ffe_ctl)
3750 {
3751 struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
3752 u64 offset;
3753
3754 /*
3755 * We are doing an unclustered allocation, set the fragmented flag so
3756 * we don't bother trying to setup a cluster again until we get more
3757 * space.
3758 */
3759 if (unlikely(last_ptr)) {
3760 spin_lock(&last_ptr->lock);
3761 last_ptr->fragmented = 1;
3762 spin_unlock(&last_ptr->lock);
3763 }
3764 if (ffe_ctl->cached) {
3765 struct btrfs_free_space_ctl *free_space_ctl;
3766
3767 free_space_ctl = bg->free_space_ctl;
3768 spin_lock(&free_space_ctl->tree_lock);
3769 if (free_space_ctl->free_space <
3770 ffe_ctl->num_bytes + ffe_ctl->empty_cluster +
3771 ffe_ctl->empty_size) {
3772 ffe_ctl->total_free_space = max_t(u64,
3773 ffe_ctl->total_free_space,
3774 free_space_ctl->free_space);
3775 spin_unlock(&free_space_ctl->tree_lock);
3776 return 1;
3777 }
3778 spin_unlock(&free_space_ctl->tree_lock);
3779 }
3780
3781 offset = btrfs_find_space_for_alloc(bg, ffe_ctl->search_start,
3782 ffe_ctl->num_bytes, ffe_ctl->empty_size,
3783 &ffe_ctl->max_extent_size);
3784 if (!offset)
3785 return 1;
3786 ffe_ctl->found_offset = offset;
3787 return 0;
3788 }
3789
do_allocation_clustered(struct btrfs_block_group * block_group,struct find_free_extent_ctl * ffe_ctl,struct btrfs_block_group ** bg_ret)3790 static int do_allocation_clustered(struct btrfs_block_group *block_group,
3791 struct find_free_extent_ctl *ffe_ctl,
3792 struct btrfs_block_group **bg_ret)
3793 {
3794 int ret;
3795
3796 /* We want to try and use the cluster allocator, so lets look there */
3797 if (ffe_ctl->last_ptr && ffe_ctl->use_cluster) {
3798 ret = find_free_extent_clustered(block_group, ffe_ctl, bg_ret);
3799 if (ret >= 0)
3800 return ret;
3801 /* ret == -ENOENT case falls through */
3802 }
3803
3804 return find_free_extent_unclustered(block_group, ffe_ctl);
3805 }
3806
3807 /*
3808 * Tree-log block group locking
3809 * ============================
3810 *
3811 * fs_info::treelog_bg_lock protects the fs_info::treelog_bg which
3812 * indicates the starting address of a block group, which is reserved only
3813 * for tree-log metadata.
3814 *
3815 * Lock nesting
3816 * ============
3817 *
3818 * space_info::lock
3819 * block_group::lock
3820 * fs_info::treelog_bg_lock
3821 */
3822
3823 /*
3824 * Simple allocator for sequential-only block group. It only allows sequential
3825 * allocation. No need to play with trees. This function also reserves the
3826 * bytes as in btrfs_add_reserved_bytes.
3827 */
do_allocation_zoned(struct btrfs_block_group * block_group,struct find_free_extent_ctl * ffe_ctl,struct btrfs_block_group ** bg_ret)3828 static int do_allocation_zoned(struct btrfs_block_group *block_group,
3829 struct find_free_extent_ctl *ffe_ctl,
3830 struct btrfs_block_group **bg_ret)
3831 {
3832 struct btrfs_fs_info *fs_info = block_group->fs_info;
3833 struct btrfs_space_info *space_info = block_group->space_info;
3834 struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3835 u64 start = block_group->start;
3836 u64 num_bytes = ffe_ctl->num_bytes;
3837 u64 avail;
3838 u64 bytenr = block_group->start;
3839 u64 log_bytenr;
3840 u64 data_reloc_bytenr;
3841 int ret = 0;
3842 bool skip = false;
3843
3844 ASSERT(btrfs_is_zoned(block_group->fs_info));
3845
3846 /*
3847 * Do not allow non-tree-log blocks in the dedicated tree-log block
3848 * group, and vice versa.
3849 */
3850 spin_lock(&fs_info->treelog_bg_lock);
3851 log_bytenr = fs_info->treelog_bg;
3852 if (log_bytenr && ((ffe_ctl->for_treelog && bytenr != log_bytenr) ||
3853 (!ffe_ctl->for_treelog && bytenr == log_bytenr)))
3854 skip = true;
3855 spin_unlock(&fs_info->treelog_bg_lock);
3856 if (skip)
3857 return 1;
3858
3859 /*
3860 * Do not allow non-relocation blocks in the dedicated relocation block
3861 * group, and vice versa.
3862 */
3863 spin_lock(&fs_info->relocation_bg_lock);
3864 data_reloc_bytenr = fs_info->data_reloc_bg;
3865 if (data_reloc_bytenr &&
3866 ((ffe_ctl->for_data_reloc && bytenr != data_reloc_bytenr) ||
3867 (!ffe_ctl->for_data_reloc && bytenr == data_reloc_bytenr)))
3868 skip = true;
3869 spin_unlock(&fs_info->relocation_bg_lock);
3870 if (skip)
3871 return 1;
3872
3873 /* Check RO and no space case before trying to activate it */
3874 spin_lock(&block_group->lock);
3875 if (block_group->ro || btrfs_zoned_bg_is_full(block_group)) {
3876 ret = 1;
3877 /*
3878 * May need to clear fs_info->{treelog,data_reloc}_bg.
3879 * Return the error after taking the locks.
3880 */
3881 }
3882 spin_unlock(&block_group->lock);
3883
3884 /* Metadata block group is activated at write time. */
3885 if (!ret && (block_group->flags & BTRFS_BLOCK_GROUP_DATA) &&
3886 !btrfs_zone_activate(block_group)) {
3887 ret = 1;
3888 /*
3889 * May need to clear fs_info->{treelog,data_reloc}_bg.
3890 * Return the error after taking the locks.
3891 */
3892 }
3893
3894 spin_lock(&space_info->lock);
3895 spin_lock(&block_group->lock);
3896 spin_lock(&fs_info->treelog_bg_lock);
3897 spin_lock(&fs_info->relocation_bg_lock);
3898
3899 if (ret)
3900 goto out;
3901
3902 ASSERT(!ffe_ctl->for_treelog ||
3903 block_group->start == fs_info->treelog_bg ||
3904 fs_info->treelog_bg == 0);
3905 ASSERT(!ffe_ctl->for_data_reloc ||
3906 block_group->start == fs_info->data_reloc_bg ||
3907 fs_info->data_reloc_bg == 0);
3908
3909 if (block_group->ro ||
3910 (!ffe_ctl->for_data_reloc &&
3911 test_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC, &block_group->runtime_flags))) {
3912 ret = 1;
3913 goto out;
3914 }
3915
3916 /*
3917 * Do not allow currently using block group to be tree-log dedicated
3918 * block group.
3919 */
3920 if (ffe_ctl->for_treelog && !fs_info->treelog_bg &&
3921 (block_group->used || block_group->reserved)) {
3922 ret = 1;
3923 goto out;
3924 }
3925
3926 /*
3927 * Do not allow currently used block group to be the data relocation
3928 * dedicated block group.
3929 */
3930 if (ffe_ctl->for_data_reloc && !fs_info->data_reloc_bg &&
3931 (block_group->used || block_group->reserved)) {
3932 ret = 1;
3933 goto out;
3934 }
3935
3936 WARN_ON_ONCE(block_group->alloc_offset > block_group->zone_capacity);
3937 avail = block_group->zone_capacity - block_group->alloc_offset;
3938 if (avail < num_bytes) {
3939 if (ffe_ctl->max_extent_size < avail) {
3940 /*
3941 * With sequential allocator, free space is always
3942 * contiguous
3943 */
3944 ffe_ctl->max_extent_size = avail;
3945 ffe_ctl->total_free_space = avail;
3946 }
3947 ret = 1;
3948 goto out;
3949 }
3950
3951 if (ffe_ctl->for_treelog && !fs_info->treelog_bg)
3952 fs_info->treelog_bg = block_group->start;
3953
3954 if (ffe_ctl->for_data_reloc) {
3955 if (!fs_info->data_reloc_bg)
3956 fs_info->data_reloc_bg = block_group->start;
3957 /*
3958 * Do not allow allocations from this block group, unless it is
3959 * for data relocation. Compared to increasing the ->ro, setting
3960 * the ->zoned_data_reloc_ongoing flag still allows nocow
3961 * writers to come in. See btrfs_inc_nocow_writers().
3962 *
3963 * We need to disable an allocation to avoid an allocation of
3964 * regular (non-relocation data) extent. With mix of relocation
3965 * extents and regular extents, we can dispatch WRITE commands
3966 * (for relocation extents) and ZONE APPEND commands (for
3967 * regular extents) at the same time to the same zone, which
3968 * easily break the write pointer.
3969 *
3970 * Also, this flag avoids this block group to be zone finished.
3971 */
3972 set_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC, &block_group->runtime_flags);
3973 }
3974
3975 ffe_ctl->found_offset = start + block_group->alloc_offset;
3976 block_group->alloc_offset += num_bytes;
3977 spin_lock(&ctl->tree_lock);
3978 ctl->free_space -= num_bytes;
3979 spin_unlock(&ctl->tree_lock);
3980
3981 /*
3982 * We do not check if found_offset is aligned to stripesize. The
3983 * address is anyway rewritten when using zone append writing.
3984 */
3985
3986 ffe_ctl->search_start = ffe_ctl->found_offset;
3987
3988 out:
3989 if (ret && ffe_ctl->for_treelog)
3990 fs_info->treelog_bg = 0;
3991 if (ret && ffe_ctl->for_data_reloc)
3992 fs_info->data_reloc_bg = 0;
3993 spin_unlock(&fs_info->relocation_bg_lock);
3994 spin_unlock(&fs_info->treelog_bg_lock);
3995 spin_unlock(&block_group->lock);
3996 spin_unlock(&space_info->lock);
3997 return ret;
3998 }
3999
do_allocation(struct btrfs_block_group * block_group,struct find_free_extent_ctl * ffe_ctl,struct btrfs_block_group ** bg_ret)4000 static int do_allocation(struct btrfs_block_group *block_group,
4001 struct find_free_extent_ctl *ffe_ctl,
4002 struct btrfs_block_group **bg_ret)
4003 {
4004 switch (ffe_ctl->policy) {
4005 case BTRFS_EXTENT_ALLOC_CLUSTERED:
4006 return do_allocation_clustered(block_group, ffe_ctl, bg_ret);
4007 case BTRFS_EXTENT_ALLOC_ZONED:
4008 return do_allocation_zoned(block_group, ffe_ctl, bg_ret);
4009 default:
4010 BUG();
4011 }
4012 }
4013
release_block_group(struct btrfs_block_group * block_group,struct find_free_extent_ctl * ffe_ctl,int delalloc)4014 static void release_block_group(struct btrfs_block_group *block_group,
4015 struct find_free_extent_ctl *ffe_ctl,
4016 int delalloc)
4017 {
4018 switch (ffe_ctl->policy) {
4019 case BTRFS_EXTENT_ALLOC_CLUSTERED:
4020 ffe_ctl->retry_uncached = false;
4021 break;
4022 case BTRFS_EXTENT_ALLOC_ZONED:
4023 /* Nothing to do */
4024 break;
4025 default:
4026 BUG();
4027 }
4028
4029 BUG_ON(btrfs_bg_flags_to_raid_index(block_group->flags) !=
4030 ffe_ctl->index);
4031 btrfs_release_block_group(block_group, delalloc);
4032 }
4033
found_extent_clustered(struct find_free_extent_ctl * ffe_ctl,struct btrfs_key * ins)4034 static void found_extent_clustered(struct find_free_extent_ctl *ffe_ctl,
4035 struct btrfs_key *ins)
4036 {
4037 struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
4038
4039 if (!ffe_ctl->use_cluster && last_ptr) {
4040 spin_lock(&last_ptr->lock);
4041 last_ptr->window_start = ins->objectid;
4042 spin_unlock(&last_ptr->lock);
4043 }
4044 }
4045
found_extent(struct find_free_extent_ctl * ffe_ctl,struct btrfs_key * ins)4046 static void found_extent(struct find_free_extent_ctl *ffe_ctl,
4047 struct btrfs_key *ins)
4048 {
4049 switch (ffe_ctl->policy) {
4050 case BTRFS_EXTENT_ALLOC_CLUSTERED:
4051 found_extent_clustered(ffe_ctl, ins);
4052 break;
4053 case BTRFS_EXTENT_ALLOC_ZONED:
4054 /* Nothing to do */
4055 break;
4056 default:
4057 BUG();
4058 }
4059 }
4060
can_allocate_chunk_zoned(struct btrfs_fs_info * fs_info,struct find_free_extent_ctl * ffe_ctl)4061 static int can_allocate_chunk_zoned(struct btrfs_fs_info *fs_info,
4062 struct find_free_extent_ctl *ffe_ctl)
4063 {
4064 /* Block group's activeness is not a requirement for METADATA block groups. */
4065 if (!(ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA))
4066 return 0;
4067
4068 /* If we can activate new zone, just allocate a chunk and use it */
4069 if (btrfs_can_activate_zone(fs_info->fs_devices, ffe_ctl->flags))
4070 return 0;
4071
4072 /*
4073 * We already reached the max active zones. Try to finish one block
4074 * group to make a room for a new block group. This is only possible
4075 * for a data block group because btrfs_zone_finish() may need to wait
4076 * for a running transaction which can cause a deadlock for metadata
4077 * allocation.
4078 */
4079 if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA) {
4080 int ret = btrfs_zone_finish_one_bg(fs_info);
4081
4082 if (ret == 1)
4083 return 0;
4084 else if (ret < 0)
4085 return ret;
4086 }
4087
4088 /*
4089 * If we have enough free space left in an already active block group
4090 * and we can't activate any other zone now, do not allow allocating a
4091 * new chunk and let find_free_extent() retry with a smaller size.
4092 */
4093 if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size)
4094 return -ENOSPC;
4095
4096 /*
4097 * Even min_alloc_size is not left in any block groups. Since we cannot
4098 * activate a new block group, allocating it may not help. Let's tell a
4099 * caller to try again and hope it progress something by writing some
4100 * parts of the region. That is only possible for data block groups,
4101 * where a part of the region can be written.
4102 */
4103 if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA)
4104 return -EAGAIN;
4105
4106 /*
4107 * We cannot activate a new block group and no enough space left in any
4108 * block groups. So, allocating a new block group may not help. But,
4109 * there is nothing to do anyway, so let's go with it.
4110 */
4111 return 0;
4112 }
4113
can_allocate_chunk(struct btrfs_fs_info * fs_info,struct find_free_extent_ctl * ffe_ctl)4114 static int can_allocate_chunk(struct btrfs_fs_info *fs_info,
4115 struct find_free_extent_ctl *ffe_ctl)
4116 {
4117 switch (ffe_ctl->policy) {
4118 case BTRFS_EXTENT_ALLOC_CLUSTERED:
4119 return 0;
4120 case BTRFS_EXTENT_ALLOC_ZONED:
4121 return can_allocate_chunk_zoned(fs_info, ffe_ctl);
4122 default:
4123 BUG();
4124 }
4125 }
4126
4127 /*
4128 * Return >0 means caller needs to re-search for free extent
4129 * Return 0 means we have the needed free extent.
4130 * Return <0 means we failed to locate any free extent.
4131 */
find_free_extent_update_loop(struct btrfs_fs_info * fs_info,struct btrfs_key * ins,struct find_free_extent_ctl * ffe_ctl,bool full_search)4132 static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
4133 struct btrfs_key *ins,
4134 struct find_free_extent_ctl *ffe_ctl,
4135 bool full_search)
4136 {
4137 struct btrfs_root *root = fs_info->chunk_root;
4138 int ret;
4139
4140 if ((ffe_ctl->loop == LOOP_CACHING_NOWAIT) &&
4141 ffe_ctl->have_caching_bg && !ffe_ctl->orig_have_caching_bg)
4142 ffe_ctl->orig_have_caching_bg = true;
4143
4144 if (ins->objectid) {
4145 found_extent(ffe_ctl, ins);
4146 return 0;
4147 }
4148
4149 if (ffe_ctl->loop >= LOOP_CACHING_WAIT && ffe_ctl->have_caching_bg)
4150 return 1;
4151
4152 ffe_ctl->index++;
4153 if (ffe_ctl->index < BTRFS_NR_RAID_TYPES)
4154 return 1;
4155
4156 /* See the comments for btrfs_loop_type for an explanation of the phases. */
4157 if (ffe_ctl->loop < LOOP_NO_EMPTY_SIZE) {
4158 ffe_ctl->index = 0;
4159 /*
4160 * We want to skip the LOOP_CACHING_WAIT step if we don't have
4161 * any uncached bgs and we've already done a full search
4162 * through.
4163 */
4164 if (ffe_ctl->loop == LOOP_CACHING_NOWAIT &&
4165 (!ffe_ctl->orig_have_caching_bg && full_search))
4166 ffe_ctl->loop++;
4167 ffe_ctl->loop++;
4168
4169 if (ffe_ctl->loop == LOOP_ALLOC_CHUNK) {
4170 struct btrfs_trans_handle *trans;
4171 int exist = 0;
4172
4173 /* Check if allocation policy allows to create a new chunk */
4174 ret = can_allocate_chunk(fs_info, ffe_ctl);
4175 if (ret)
4176 return ret;
4177
4178 trans = current->journal_info;
4179 if (trans)
4180 exist = 1;
4181 else
4182 trans = btrfs_join_transaction(root);
4183
4184 if (IS_ERR(trans)) {
4185 ret = PTR_ERR(trans);
4186 return ret;
4187 }
4188
4189 ret = btrfs_chunk_alloc(trans, ffe_ctl->flags,
4190 CHUNK_ALLOC_FORCE_FOR_EXTENT);
4191
4192 /* Do not bail out on ENOSPC since we can do more. */
4193 if (ret == -ENOSPC) {
4194 ret = 0;
4195 ffe_ctl->loop++;
4196 }
4197 else if (ret < 0)
4198 btrfs_abort_transaction(trans, ret);
4199 else
4200 ret = 0;
4201 if (!exist)
4202 btrfs_end_transaction(trans);
4203 if (ret)
4204 return ret;
4205 }
4206
4207 if (ffe_ctl->loop == LOOP_NO_EMPTY_SIZE) {
4208 if (ffe_ctl->policy != BTRFS_EXTENT_ALLOC_CLUSTERED)
4209 return -ENOSPC;
4210
4211 /*
4212 * Don't loop again if we already have no empty_size and
4213 * no empty_cluster.
4214 */
4215 if (ffe_ctl->empty_size == 0 &&
4216 ffe_ctl->empty_cluster == 0)
4217 return -ENOSPC;
4218 ffe_ctl->empty_size = 0;
4219 ffe_ctl->empty_cluster = 0;
4220 }
4221 return 1;
4222 }
4223 return -ENOSPC;
4224 }
4225
find_free_extent_check_size_class(struct find_free_extent_ctl * ffe_ctl,struct btrfs_block_group * bg)4226 static bool find_free_extent_check_size_class(struct find_free_extent_ctl *ffe_ctl,
4227 struct btrfs_block_group *bg)
4228 {
4229 if (ffe_ctl->policy == BTRFS_EXTENT_ALLOC_ZONED)
4230 return true;
4231 if (!btrfs_block_group_should_use_size_class(bg))
4232 return true;
4233 if (ffe_ctl->loop >= LOOP_WRONG_SIZE_CLASS)
4234 return true;
4235 if (ffe_ctl->loop >= LOOP_UNSET_SIZE_CLASS &&
4236 bg->size_class == BTRFS_BG_SZ_NONE)
4237 return true;
4238 return ffe_ctl->size_class == bg->size_class;
4239 }
4240
prepare_allocation_clustered(struct btrfs_fs_info * fs_info,struct find_free_extent_ctl * ffe_ctl,struct btrfs_space_info * space_info,struct btrfs_key * ins)4241 static int prepare_allocation_clustered(struct btrfs_fs_info *fs_info,
4242 struct find_free_extent_ctl *ffe_ctl,
4243 struct btrfs_space_info *space_info,
4244 struct btrfs_key *ins)
4245 {
4246 /*
4247 * If our free space is heavily fragmented we may not be able to make
4248 * big contiguous allocations, so instead of doing the expensive search
4249 * for free space, simply return ENOSPC with our max_extent_size so we
4250 * can go ahead and search for a more manageable chunk.
4251 *
4252 * If our max_extent_size is large enough for our allocation simply
4253 * disable clustering since we will likely not be able to find enough
4254 * space to create a cluster and induce latency trying.
4255 */
4256 if (space_info->max_extent_size) {
4257 spin_lock(&space_info->lock);
4258 if (space_info->max_extent_size &&
4259 ffe_ctl->num_bytes > space_info->max_extent_size) {
4260 ins->offset = space_info->max_extent_size;
4261 spin_unlock(&space_info->lock);
4262 return -ENOSPC;
4263 } else if (space_info->max_extent_size) {
4264 ffe_ctl->use_cluster = false;
4265 }
4266 spin_unlock(&space_info->lock);
4267 }
4268
4269 ffe_ctl->last_ptr = fetch_cluster_info(fs_info, space_info,
4270 &ffe_ctl->empty_cluster);
4271 if (ffe_ctl->last_ptr) {
4272 struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
4273
4274 spin_lock(&last_ptr->lock);
4275 if (last_ptr->block_group)
4276 ffe_ctl->hint_byte = last_ptr->window_start;
4277 if (last_ptr->fragmented) {
4278 /*
4279 * We still set window_start so we can keep track of the
4280 * last place we found an allocation to try and save
4281 * some time.
4282 */
4283 ffe_ctl->hint_byte = last_ptr->window_start;
4284 ffe_ctl->use_cluster = false;
4285 }
4286 spin_unlock(&last_ptr->lock);
4287 }
4288
4289 return 0;
4290 }
4291
prepare_allocation_zoned(struct btrfs_fs_info * fs_info,struct find_free_extent_ctl * ffe_ctl)4292 static int prepare_allocation_zoned(struct btrfs_fs_info *fs_info,
4293 struct find_free_extent_ctl *ffe_ctl)
4294 {
4295 if (ffe_ctl->for_treelog) {
4296 spin_lock(&fs_info->treelog_bg_lock);
4297 if (fs_info->treelog_bg)
4298 ffe_ctl->hint_byte = fs_info->treelog_bg;
4299 spin_unlock(&fs_info->treelog_bg_lock);
4300 } else if (ffe_ctl->for_data_reloc) {
4301 spin_lock(&fs_info->relocation_bg_lock);
4302 if (fs_info->data_reloc_bg)
4303 ffe_ctl->hint_byte = fs_info->data_reloc_bg;
4304 spin_unlock(&fs_info->relocation_bg_lock);
4305 } else if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA) {
4306 struct btrfs_block_group *block_group;
4307
4308 spin_lock(&fs_info->zone_active_bgs_lock);
4309 list_for_each_entry(block_group, &fs_info->zone_active_bgs, active_bg_list) {
4310 /*
4311 * No lock is OK here because avail is monotinically
4312 * decreasing, and this is just a hint.
4313 */
4314 u64 avail = block_group->zone_capacity - block_group->alloc_offset;
4315
4316 if (block_group_bits(block_group, ffe_ctl->flags) &&
4317 avail >= ffe_ctl->num_bytes) {
4318 ffe_ctl->hint_byte = block_group->start;
4319 break;
4320 }
4321 }
4322 spin_unlock(&fs_info->zone_active_bgs_lock);
4323 }
4324
4325 return 0;
4326 }
4327
prepare_allocation(struct btrfs_fs_info * fs_info,struct find_free_extent_ctl * ffe_ctl,struct btrfs_space_info * space_info,struct btrfs_key * ins)4328 static int prepare_allocation(struct btrfs_fs_info *fs_info,
4329 struct find_free_extent_ctl *ffe_ctl,
4330 struct btrfs_space_info *space_info,
4331 struct btrfs_key *ins)
4332 {
4333 switch (ffe_ctl->policy) {
4334 case BTRFS_EXTENT_ALLOC_CLUSTERED:
4335 return prepare_allocation_clustered(fs_info, ffe_ctl,
4336 space_info, ins);
4337 case BTRFS_EXTENT_ALLOC_ZONED:
4338 return prepare_allocation_zoned(fs_info, ffe_ctl);
4339 default:
4340 BUG();
4341 }
4342 }
4343
4344 /*
4345 * walks the btree of allocated extents and find a hole of a given size.
4346 * The key ins is changed to record the hole:
4347 * ins->objectid == start position
4348 * ins->flags = BTRFS_EXTENT_ITEM_KEY
4349 * ins->offset == the size of the hole.
4350 * Any available blocks before search_start are skipped.
4351 *
4352 * If there is no suitable free space, we will record the max size of
4353 * the free space extent currently.
4354 *
4355 * The overall logic and call chain:
4356 *
4357 * find_free_extent()
4358 * |- Iterate through all block groups
4359 * | |- Get a valid block group
4360 * | |- Try to do clustered allocation in that block group
4361 * | |- Try to do unclustered allocation in that block group
4362 * | |- Check if the result is valid
4363 * | | |- If valid, then exit
4364 * | |- Jump to next block group
4365 * |
4366 * |- Push harder to find free extents
4367 * |- If not found, re-iterate all block groups
4368 */
find_free_extent(struct btrfs_root * root,struct btrfs_key * ins,struct find_free_extent_ctl * ffe_ctl)4369 static noinline int find_free_extent(struct btrfs_root *root,
4370 struct btrfs_key *ins,
4371 struct find_free_extent_ctl *ffe_ctl)
4372 {
4373 struct btrfs_fs_info *fs_info = root->fs_info;
4374 int ret = 0;
4375 int cache_block_group_error = 0;
4376 struct btrfs_block_group *block_group = NULL;
4377 struct btrfs_space_info *space_info;
4378 bool full_search = false;
4379
4380 WARN_ON(ffe_ctl->num_bytes < fs_info->sectorsize);
4381
4382 ffe_ctl->search_start = 0;
4383 /* For clustered allocation */
4384 ffe_ctl->empty_cluster = 0;
4385 ffe_ctl->last_ptr = NULL;
4386 ffe_ctl->use_cluster = true;
4387 ffe_ctl->have_caching_bg = false;
4388 ffe_ctl->orig_have_caching_bg = false;
4389 ffe_ctl->index = btrfs_bg_flags_to_raid_index(ffe_ctl->flags);
4390 ffe_ctl->loop = 0;
4391 ffe_ctl->retry_uncached = false;
4392 ffe_ctl->cached = 0;
4393 ffe_ctl->max_extent_size = 0;
4394 ffe_ctl->total_free_space = 0;
4395 ffe_ctl->found_offset = 0;
4396 ffe_ctl->policy = BTRFS_EXTENT_ALLOC_CLUSTERED;
4397 ffe_ctl->size_class = btrfs_calc_block_group_size_class(ffe_ctl->num_bytes);
4398
4399 if (btrfs_is_zoned(fs_info))
4400 ffe_ctl->policy = BTRFS_EXTENT_ALLOC_ZONED;
4401
4402 ins->type = BTRFS_EXTENT_ITEM_KEY;
4403 ins->objectid = 0;
4404 ins->offset = 0;
4405
4406 trace_find_free_extent(root, ffe_ctl);
4407
4408 space_info = btrfs_find_space_info(fs_info, ffe_ctl->flags);
4409 if (!space_info) {
4410 btrfs_err(fs_info, "No space info for %llu", ffe_ctl->flags);
4411 return -ENOSPC;
4412 }
4413
4414 ret = prepare_allocation(fs_info, ffe_ctl, space_info, ins);
4415 if (ret < 0)
4416 return ret;
4417
4418 ffe_ctl->search_start = max(ffe_ctl->search_start,
4419 first_logical_byte(fs_info));
4420 ffe_ctl->search_start = max(ffe_ctl->search_start, ffe_ctl->hint_byte);
4421 if (ffe_ctl->search_start == ffe_ctl->hint_byte) {
4422 block_group = btrfs_lookup_block_group(fs_info,
4423 ffe_ctl->search_start);
4424 /*
4425 * we don't want to use the block group if it doesn't match our
4426 * allocation bits, or if its not cached.
4427 *
4428 * However if we are re-searching with an ideal block group
4429 * picked out then we don't care that the block group is cached.
4430 */
4431 if (block_group && block_group_bits(block_group, ffe_ctl->flags) &&
4432 block_group->cached != BTRFS_CACHE_NO) {
4433 down_read(&space_info->groups_sem);
4434 if (list_empty(&block_group->list) ||
4435 block_group->ro) {
4436 /*
4437 * someone is removing this block group,
4438 * we can't jump into the have_block_group
4439 * target because our list pointers are not
4440 * valid
4441 */
4442 btrfs_put_block_group(block_group);
4443 up_read(&space_info->groups_sem);
4444 } else {
4445 ffe_ctl->index = btrfs_bg_flags_to_raid_index(
4446 block_group->flags);
4447 btrfs_lock_block_group(block_group,
4448 ffe_ctl->delalloc);
4449 ffe_ctl->hinted = true;
4450 goto have_block_group;
4451 }
4452 } else if (block_group) {
4453 btrfs_put_block_group(block_group);
4454 }
4455 }
4456 search:
4457 trace_find_free_extent_search_loop(root, ffe_ctl);
4458 ffe_ctl->have_caching_bg = false;
4459 if (ffe_ctl->index == btrfs_bg_flags_to_raid_index(ffe_ctl->flags) ||
4460 ffe_ctl->index == 0)
4461 full_search = true;
4462 down_read(&space_info->groups_sem);
4463 list_for_each_entry(block_group,
4464 &space_info->block_groups[ffe_ctl->index], list) {
4465 struct btrfs_block_group *bg_ret;
4466
4467 ffe_ctl->hinted = false;
4468 /* If the block group is read-only, we can skip it entirely. */
4469 if (unlikely(block_group->ro)) {
4470 if (ffe_ctl->for_treelog)
4471 btrfs_clear_treelog_bg(block_group);
4472 if (ffe_ctl->for_data_reloc)
4473 btrfs_clear_data_reloc_bg(block_group);
4474 continue;
4475 }
4476
4477 btrfs_grab_block_group(block_group, ffe_ctl->delalloc);
4478 ffe_ctl->search_start = block_group->start;
4479
4480 /*
4481 * this can happen if we end up cycling through all the
4482 * raid types, but we want to make sure we only allocate
4483 * for the proper type.
4484 */
4485 if (!block_group_bits(block_group, ffe_ctl->flags)) {
4486 u64 extra = BTRFS_BLOCK_GROUP_DUP |
4487 BTRFS_BLOCK_GROUP_RAID1_MASK |
4488 BTRFS_BLOCK_GROUP_RAID56_MASK |
4489 BTRFS_BLOCK_GROUP_RAID10;
4490
4491 /*
4492 * if they asked for extra copies and this block group
4493 * doesn't provide them, bail. This does allow us to
4494 * fill raid0 from raid1.
4495 */
4496 if ((ffe_ctl->flags & extra) && !(block_group->flags & extra))
4497 goto loop;
4498
4499 /*
4500 * This block group has different flags than we want.
4501 * It's possible that we have MIXED_GROUP flag but no
4502 * block group is mixed. Just skip such block group.
4503 */
4504 btrfs_release_block_group(block_group, ffe_ctl->delalloc);
4505 continue;
4506 }
4507
4508 have_block_group:
4509 trace_find_free_extent_have_block_group(root, ffe_ctl, block_group);
4510 ffe_ctl->cached = btrfs_block_group_done(block_group);
4511 if (unlikely(!ffe_ctl->cached)) {
4512 ffe_ctl->have_caching_bg = true;
4513 ret = btrfs_cache_block_group(block_group, false);
4514
4515 /*
4516 * If we get ENOMEM here or something else we want to
4517 * try other block groups, because it may not be fatal.
4518 * However if we can't find anything else we need to
4519 * save our return here so that we return the actual
4520 * error that caused problems, not ENOSPC.
4521 */
4522 if (ret < 0) {
4523 if (!cache_block_group_error)
4524 cache_block_group_error = ret;
4525 ret = 0;
4526 goto loop;
4527 }
4528 ret = 0;
4529 }
4530
4531 if (unlikely(block_group->cached == BTRFS_CACHE_ERROR)) {
4532 if (!cache_block_group_error)
4533 cache_block_group_error = -EIO;
4534 goto loop;
4535 }
4536
4537 if (!find_free_extent_check_size_class(ffe_ctl, block_group))
4538 goto loop;
4539
4540 bg_ret = NULL;
4541 ret = do_allocation(block_group, ffe_ctl, &bg_ret);
4542 if (ret > 0)
4543 goto loop;
4544
4545 if (bg_ret && bg_ret != block_group) {
4546 btrfs_release_block_group(block_group, ffe_ctl->delalloc);
4547 block_group = bg_ret;
4548 }
4549
4550 /* Checks */
4551 ffe_ctl->search_start = round_up(ffe_ctl->found_offset,
4552 fs_info->stripesize);
4553
4554 /* move on to the next group */
4555 if (ffe_ctl->search_start + ffe_ctl->num_bytes >
4556 block_group->start + block_group->length) {
4557 btrfs_add_free_space_unused(block_group,
4558 ffe_ctl->found_offset,
4559 ffe_ctl->num_bytes);
4560 goto loop;
4561 }
4562
4563 if (ffe_ctl->found_offset < ffe_ctl->search_start)
4564 btrfs_add_free_space_unused(block_group,
4565 ffe_ctl->found_offset,
4566 ffe_ctl->search_start - ffe_ctl->found_offset);
4567
4568 ret = btrfs_add_reserved_bytes(block_group, ffe_ctl->ram_bytes,
4569 ffe_ctl->num_bytes,
4570 ffe_ctl->delalloc,
4571 ffe_ctl->loop >= LOOP_WRONG_SIZE_CLASS);
4572 if (ret == -EAGAIN) {
4573 btrfs_add_free_space_unused(block_group,
4574 ffe_ctl->found_offset,
4575 ffe_ctl->num_bytes);
4576 goto loop;
4577 }
4578 btrfs_inc_block_group_reservations(block_group);
4579
4580 /* we are all good, lets return */
4581 ins->objectid = ffe_ctl->search_start;
4582 ins->offset = ffe_ctl->num_bytes;
4583
4584 trace_btrfs_reserve_extent(block_group, ffe_ctl);
4585 btrfs_release_block_group(block_group, ffe_ctl->delalloc);
4586 break;
4587 loop:
4588 if (!ffe_ctl->cached && ffe_ctl->loop > LOOP_CACHING_NOWAIT &&
4589 !ffe_ctl->retry_uncached) {
4590 ffe_ctl->retry_uncached = true;
4591 btrfs_wait_block_group_cache_progress(block_group,
4592 ffe_ctl->num_bytes +
4593 ffe_ctl->empty_cluster +
4594 ffe_ctl->empty_size);
4595 goto have_block_group;
4596 }
4597 release_block_group(block_group, ffe_ctl, ffe_ctl->delalloc);
4598 cond_resched();
4599 }
4600 up_read(&space_info->groups_sem);
4601
4602 ret = find_free_extent_update_loop(fs_info, ins, ffe_ctl, full_search);
4603 if (ret > 0)
4604 goto search;
4605
4606 if (ret == -ENOSPC && !cache_block_group_error) {
4607 /*
4608 * Use ffe_ctl->total_free_space as fallback if we can't find
4609 * any contiguous hole.
4610 */
4611 if (!ffe_ctl->max_extent_size)
4612 ffe_ctl->max_extent_size = ffe_ctl->total_free_space;
4613 spin_lock(&space_info->lock);
4614 space_info->max_extent_size = ffe_ctl->max_extent_size;
4615 spin_unlock(&space_info->lock);
4616 ins->offset = ffe_ctl->max_extent_size;
4617 } else if (ret == -ENOSPC) {
4618 ret = cache_block_group_error;
4619 }
4620 return ret;
4621 }
4622
4623 /*
4624 * Entry point to the extent allocator. Tries to find a hole that is at least
4625 * as big as @num_bytes.
4626 *
4627 * @root - The root that will contain this extent
4628 *
4629 * @ram_bytes - The amount of space in ram that @num_bytes take. This
4630 * is used for accounting purposes. This value differs
4631 * from @num_bytes only in the case of compressed extents.
4632 *
4633 * @num_bytes - Number of bytes to allocate on-disk.
4634 *
4635 * @min_alloc_size - Indicates the minimum amount of space that the
4636 * allocator should try to satisfy. In some cases
4637 * @num_bytes may be larger than what is required and if
4638 * the filesystem is fragmented then allocation fails.
4639 * However, the presence of @min_alloc_size gives a
4640 * chance to try and satisfy the smaller allocation.
4641 *
4642 * @empty_size - A hint that you plan on doing more COW. This is the
4643 * size in bytes the allocator should try to find free
4644 * next to the block it returns. This is just a hint and
4645 * may be ignored by the allocator.
4646 *
4647 * @hint_byte - Hint to the allocator to start searching above the byte
4648 * address passed. It might be ignored.
4649 *
4650 * @ins - This key is modified to record the found hole. It will
4651 * have the following values:
4652 * ins->objectid == start position
4653 * ins->flags = BTRFS_EXTENT_ITEM_KEY
4654 * ins->offset == the size of the hole.
4655 *
4656 * @is_data - Boolean flag indicating whether an extent is
4657 * allocated for data (true) or metadata (false)
4658 *
4659 * @delalloc - Boolean flag indicating whether this allocation is for
4660 * delalloc or not. If 'true' data_rwsem of block groups
4661 * is going to be acquired.
4662 *
4663 *
4664 * Returns 0 when an allocation succeeded or < 0 when an error occurred. In
4665 * case -ENOSPC is returned then @ins->offset will contain the size of the
4666 * largest available hole the allocator managed to find.
4667 */
btrfs_reserve_extent(struct btrfs_root * root,u64 ram_bytes,u64 num_bytes,u64 min_alloc_size,u64 empty_size,u64 hint_byte,struct btrfs_key * ins,int is_data,int delalloc)4668 int btrfs_reserve_extent(struct btrfs_root *root, u64 ram_bytes,
4669 u64 num_bytes, u64 min_alloc_size,
4670 u64 empty_size, u64 hint_byte,
4671 struct btrfs_key *ins, int is_data, int delalloc)
4672 {
4673 struct btrfs_fs_info *fs_info = root->fs_info;
4674 struct find_free_extent_ctl ffe_ctl = {};
4675 bool final_tried = num_bytes == min_alloc_size;
4676 u64 flags;
4677 int ret;
4678 bool for_treelog = (btrfs_root_id(root) == BTRFS_TREE_LOG_OBJECTID);
4679 bool for_data_reloc = (btrfs_is_data_reloc_root(root) && is_data);
4680
4681 flags = get_alloc_profile_by_root(root, is_data);
4682 again:
4683 WARN_ON(num_bytes < fs_info->sectorsize);
4684
4685 ffe_ctl.ram_bytes = ram_bytes;
4686 ffe_ctl.num_bytes = num_bytes;
4687 ffe_ctl.min_alloc_size = min_alloc_size;
4688 ffe_ctl.empty_size = empty_size;
4689 ffe_ctl.flags = flags;
4690 ffe_ctl.delalloc = delalloc;
4691 ffe_ctl.hint_byte = hint_byte;
4692 ffe_ctl.for_treelog = for_treelog;
4693 ffe_ctl.for_data_reloc = for_data_reloc;
4694
4695 ret = find_free_extent(root, ins, &ffe_ctl);
4696 if (!ret && !is_data) {
4697 btrfs_dec_block_group_reservations(fs_info, ins->objectid);
4698 } else if (ret == -ENOSPC) {
4699 if (!final_tried && ins->offset) {
4700 num_bytes = min(num_bytes >> 1, ins->offset);
4701 num_bytes = round_down(num_bytes,
4702 fs_info->sectorsize);
4703 num_bytes = max(num_bytes, min_alloc_size);
4704 ram_bytes = num_bytes;
4705 if (num_bytes == min_alloc_size)
4706 final_tried = true;
4707 goto again;
4708 } else if (btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
4709 struct btrfs_space_info *sinfo;
4710
4711 sinfo = btrfs_find_space_info(fs_info, flags);
4712 btrfs_err(fs_info,
4713 "allocation failed flags %llu, wanted %llu tree-log %d, relocation: %d",
4714 flags, num_bytes, for_treelog, for_data_reloc);
4715 if (sinfo)
4716 btrfs_dump_space_info(fs_info, sinfo,
4717 num_bytes, 1);
4718 }
4719 }
4720
4721 return ret;
4722 }
4723
btrfs_free_reserved_extent(struct btrfs_fs_info * fs_info,u64 start,u64 len,int delalloc)4724 int btrfs_free_reserved_extent(struct btrfs_fs_info *fs_info,
4725 u64 start, u64 len, int delalloc)
4726 {
4727 struct btrfs_block_group *cache;
4728
4729 cache = btrfs_lookup_block_group(fs_info, start);
4730 if (!cache) {
4731 btrfs_err(fs_info, "Unable to find block group for %llu",
4732 start);
4733 return -ENOSPC;
4734 }
4735
4736 btrfs_add_free_space(cache, start, len);
4737 btrfs_free_reserved_bytes(cache, len, delalloc);
4738 trace_btrfs_reserved_extent_free(fs_info, start, len);
4739
4740 btrfs_put_block_group(cache);
4741 return 0;
4742 }
4743
btrfs_pin_reserved_extent(struct btrfs_trans_handle * trans,const struct extent_buffer * eb)4744 int btrfs_pin_reserved_extent(struct btrfs_trans_handle *trans,
4745 const struct extent_buffer *eb)
4746 {
4747 struct btrfs_block_group *cache;
4748 int ret = 0;
4749
4750 cache = btrfs_lookup_block_group(trans->fs_info, eb->start);
4751 if (!cache) {
4752 btrfs_err(trans->fs_info, "unable to find block group for %llu",
4753 eb->start);
4754 return -ENOSPC;
4755 }
4756
4757 ret = pin_down_extent(trans, cache, eb->start, eb->len, 1);
4758 btrfs_put_block_group(cache);
4759 return ret;
4760 }
4761
alloc_reserved_extent(struct btrfs_trans_handle * trans,u64 bytenr,u64 num_bytes)4762 static int alloc_reserved_extent(struct btrfs_trans_handle *trans, u64 bytenr,
4763 u64 num_bytes)
4764 {
4765 struct btrfs_fs_info *fs_info = trans->fs_info;
4766 int ret;
4767
4768 ret = remove_from_free_space_tree(trans, bytenr, num_bytes);
4769 if (ret)
4770 return ret;
4771
4772 ret = btrfs_update_block_group(trans, bytenr, num_bytes, true);
4773 if (ret) {
4774 ASSERT(!ret);
4775 btrfs_err(fs_info, "update block group failed for %llu %llu",
4776 bytenr, num_bytes);
4777 return ret;
4778 }
4779
4780 trace_btrfs_reserved_extent_alloc(fs_info, bytenr, num_bytes);
4781 return 0;
4782 }
4783
alloc_reserved_file_extent(struct btrfs_trans_handle * trans,u64 parent,u64 root_objectid,u64 flags,u64 owner,u64 offset,struct btrfs_key * ins,int ref_mod,u64 oref_root)4784 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
4785 u64 parent, u64 root_objectid,
4786 u64 flags, u64 owner, u64 offset,
4787 struct btrfs_key *ins, int ref_mod, u64 oref_root)
4788 {
4789 struct btrfs_fs_info *fs_info = trans->fs_info;
4790 struct btrfs_root *extent_root;
4791 int ret;
4792 struct btrfs_extent_item *extent_item;
4793 struct btrfs_extent_owner_ref *oref;
4794 struct btrfs_extent_inline_ref *iref;
4795 struct btrfs_path *path;
4796 struct extent_buffer *leaf;
4797 int type;
4798 u32 size;
4799 const bool simple_quota = (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE);
4800
4801 if (parent > 0)
4802 type = BTRFS_SHARED_DATA_REF_KEY;
4803 else
4804 type = BTRFS_EXTENT_DATA_REF_KEY;
4805
4806 size = sizeof(*extent_item);
4807 if (simple_quota)
4808 size += btrfs_extent_inline_ref_size(BTRFS_EXTENT_OWNER_REF_KEY);
4809 size += btrfs_extent_inline_ref_size(type);
4810
4811 path = btrfs_alloc_path();
4812 if (!path)
4813 return -ENOMEM;
4814
4815 extent_root = btrfs_extent_root(fs_info, ins->objectid);
4816 ret = btrfs_insert_empty_item(trans, extent_root, path, ins, size);
4817 if (ret) {
4818 btrfs_free_path(path);
4819 return ret;
4820 }
4821
4822 leaf = path->nodes[0];
4823 extent_item = btrfs_item_ptr(leaf, path->slots[0],
4824 struct btrfs_extent_item);
4825 btrfs_set_extent_refs(leaf, extent_item, ref_mod);
4826 btrfs_set_extent_generation(leaf, extent_item, trans->transid);
4827 btrfs_set_extent_flags(leaf, extent_item,
4828 flags | BTRFS_EXTENT_FLAG_DATA);
4829
4830 iref = (struct btrfs_extent_inline_ref *)(extent_item + 1);
4831 if (simple_quota) {
4832 btrfs_set_extent_inline_ref_type(leaf, iref, BTRFS_EXTENT_OWNER_REF_KEY);
4833 oref = (struct btrfs_extent_owner_ref *)(&iref->offset);
4834 btrfs_set_extent_owner_ref_root_id(leaf, oref, oref_root);
4835 iref = (struct btrfs_extent_inline_ref *)(oref + 1);
4836 }
4837 btrfs_set_extent_inline_ref_type(leaf, iref, type);
4838
4839 if (parent > 0) {
4840 struct btrfs_shared_data_ref *ref;
4841 ref = (struct btrfs_shared_data_ref *)(iref + 1);
4842 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
4843 btrfs_set_shared_data_ref_count(leaf, ref, ref_mod);
4844 } else {
4845 struct btrfs_extent_data_ref *ref;
4846 ref = (struct btrfs_extent_data_ref *)(&iref->offset);
4847 btrfs_set_extent_data_ref_root(leaf, ref, root_objectid);
4848 btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
4849 btrfs_set_extent_data_ref_offset(leaf, ref, offset);
4850 btrfs_set_extent_data_ref_count(leaf, ref, ref_mod);
4851 }
4852
4853 btrfs_mark_buffer_dirty(trans, path->nodes[0]);
4854 btrfs_free_path(path);
4855
4856 return alloc_reserved_extent(trans, ins->objectid, ins->offset);
4857 }
4858
alloc_reserved_tree_block(struct btrfs_trans_handle * trans,struct btrfs_delayed_ref_node * node,struct btrfs_delayed_extent_op * extent_op)4859 static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
4860 struct btrfs_delayed_ref_node *node,
4861 struct btrfs_delayed_extent_op *extent_op)
4862 {
4863 struct btrfs_fs_info *fs_info = trans->fs_info;
4864 struct btrfs_root *extent_root;
4865 int ret;
4866 struct btrfs_extent_item *extent_item;
4867 struct btrfs_key extent_key;
4868 struct btrfs_tree_block_info *block_info;
4869 struct btrfs_extent_inline_ref *iref;
4870 struct btrfs_path *path;
4871 struct extent_buffer *leaf;
4872 u32 size = sizeof(*extent_item) + sizeof(*iref);
4873 const u64 flags = (extent_op ? extent_op->flags_to_set : 0);
4874 /* The owner of a tree block is the level. */
4875 int level = btrfs_delayed_ref_owner(node);
4876 bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
4877
4878 extent_key.objectid = node->bytenr;
4879 if (skinny_metadata) {
4880 /* The owner of a tree block is the level. */
4881 extent_key.offset = level;
4882 extent_key.type = BTRFS_METADATA_ITEM_KEY;
4883 } else {
4884 extent_key.offset = node->num_bytes;
4885 extent_key.type = BTRFS_EXTENT_ITEM_KEY;
4886 size += sizeof(*block_info);
4887 }
4888
4889 path = btrfs_alloc_path();
4890 if (!path)
4891 return -ENOMEM;
4892
4893 extent_root = btrfs_extent_root(fs_info, extent_key.objectid);
4894 ret = btrfs_insert_empty_item(trans, extent_root, path, &extent_key,
4895 size);
4896 if (ret) {
4897 btrfs_free_path(path);
4898 return ret;
4899 }
4900
4901 leaf = path->nodes[0];
4902 extent_item = btrfs_item_ptr(leaf, path->slots[0],
4903 struct btrfs_extent_item);
4904 btrfs_set_extent_refs(leaf, extent_item, 1);
4905 btrfs_set_extent_generation(leaf, extent_item, trans->transid);
4906 btrfs_set_extent_flags(leaf, extent_item,
4907 flags | BTRFS_EXTENT_FLAG_TREE_BLOCK);
4908
4909 if (skinny_metadata) {
4910 iref = (struct btrfs_extent_inline_ref *)(extent_item + 1);
4911 } else {
4912 block_info = (struct btrfs_tree_block_info *)(extent_item + 1);
4913 btrfs_set_tree_block_key(leaf, block_info, &extent_op->key);
4914 btrfs_set_tree_block_level(leaf, block_info, level);
4915 iref = (struct btrfs_extent_inline_ref *)(block_info + 1);
4916 }
4917
4918 if (node->type == BTRFS_SHARED_BLOCK_REF_KEY) {
4919 btrfs_set_extent_inline_ref_type(leaf, iref,
4920 BTRFS_SHARED_BLOCK_REF_KEY);
4921 btrfs_set_extent_inline_ref_offset(leaf, iref, node->parent);
4922 } else {
4923 btrfs_set_extent_inline_ref_type(leaf, iref,
4924 BTRFS_TREE_BLOCK_REF_KEY);
4925 btrfs_set_extent_inline_ref_offset(leaf, iref, node->ref_root);
4926 }
4927
4928 btrfs_mark_buffer_dirty(trans, leaf);
4929 btrfs_free_path(path);
4930
4931 return alloc_reserved_extent(trans, node->bytenr, fs_info->nodesize);
4932 }
4933
btrfs_alloc_reserved_file_extent(struct btrfs_trans_handle * trans,struct btrfs_root * root,u64 owner,u64 offset,u64 ram_bytes,struct btrfs_key * ins)4934 int btrfs_alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
4935 struct btrfs_root *root, u64 owner,
4936 u64 offset, u64 ram_bytes,
4937 struct btrfs_key *ins)
4938 {
4939 struct btrfs_ref generic_ref = {
4940 .action = BTRFS_ADD_DELAYED_EXTENT,
4941 .bytenr = ins->objectid,
4942 .num_bytes = ins->offset,
4943 .owning_root = btrfs_root_id(root),
4944 .ref_root = btrfs_root_id(root),
4945 };
4946
4947 ASSERT(generic_ref.ref_root != BTRFS_TREE_LOG_OBJECTID);
4948
4949 if (btrfs_is_data_reloc_root(root) && is_fstree(root->relocation_src_root))
4950 generic_ref.owning_root = root->relocation_src_root;
4951
4952 btrfs_init_data_ref(&generic_ref, owner, offset, 0, false);
4953 btrfs_ref_tree_mod(root->fs_info, &generic_ref);
4954
4955 return btrfs_add_delayed_data_ref(trans, &generic_ref, ram_bytes);
4956 }
4957
4958 /*
4959 * this is used by the tree logging recovery code. It records that
4960 * an extent has been allocated and makes sure to clear the free
4961 * space cache bits as well
4962 */
btrfs_alloc_logged_file_extent(struct btrfs_trans_handle * trans,u64 root_objectid,u64 owner,u64 offset,struct btrfs_key * ins)4963 int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
4964 u64 root_objectid, u64 owner, u64 offset,
4965 struct btrfs_key *ins)
4966 {
4967 struct btrfs_fs_info *fs_info = trans->fs_info;
4968 int ret;
4969 struct btrfs_block_group *block_group;
4970 struct btrfs_space_info *space_info;
4971 struct btrfs_squota_delta delta = {
4972 .root = root_objectid,
4973 .num_bytes = ins->offset,
4974 .generation = trans->transid,
4975 .is_data = true,
4976 .is_inc = true,
4977 };
4978
4979 /*
4980 * Mixed block groups will exclude before processing the log so we only
4981 * need to do the exclude dance if this fs isn't mixed.
4982 */
4983 if (!btrfs_fs_incompat(fs_info, MIXED_GROUPS)) {
4984 ret = __exclude_logged_extent(fs_info, ins->objectid,
4985 ins->offset);
4986 if (ret)
4987 return ret;
4988 }
4989
4990 block_group = btrfs_lookup_block_group(fs_info, ins->objectid);
4991 if (!block_group)
4992 return -EINVAL;
4993
4994 space_info = block_group->space_info;
4995 spin_lock(&space_info->lock);
4996 spin_lock(&block_group->lock);
4997 space_info->bytes_reserved += ins->offset;
4998 block_group->reserved += ins->offset;
4999 spin_unlock(&block_group->lock);
5000 spin_unlock(&space_info->lock);
5001
5002 ret = alloc_reserved_file_extent(trans, 0, root_objectid, 0, owner,
5003 offset, ins, 1, root_objectid);
5004 if (ret)
5005 btrfs_pin_extent(trans, ins->objectid, ins->offset, 1);
5006 ret = btrfs_record_squota_delta(fs_info, &delta);
5007 btrfs_put_block_group(block_group);
5008 return ret;
5009 }
5010
5011 #ifdef CONFIG_BTRFS_DEBUG
5012 /*
5013 * Extra safety check in case the extent tree is corrupted and extent allocator
5014 * chooses to use a tree block which is already used and locked.
5015 */
check_eb_lock_owner(const struct extent_buffer * eb)5016 static bool check_eb_lock_owner(const struct extent_buffer *eb)
5017 {
5018 if (eb->lock_owner == current->pid) {
5019 btrfs_err_rl(eb->fs_info,
5020 "tree block %llu owner %llu already locked by pid=%d, extent tree corruption detected",
5021 eb->start, btrfs_header_owner(eb), current->pid);
5022 return true;
5023 }
5024 return false;
5025 }
5026 #else
check_eb_lock_owner(struct extent_buffer * eb)5027 static bool check_eb_lock_owner(struct extent_buffer *eb)
5028 {
5029 return false;
5030 }
5031 #endif
5032
5033 static struct extent_buffer *
btrfs_init_new_buffer(struct btrfs_trans_handle * trans,struct btrfs_root * root,u64 bytenr,int level,u64 owner,enum btrfs_lock_nesting nest)5034 btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct btrfs_root *root,
5035 u64 bytenr, int level, u64 owner,
5036 enum btrfs_lock_nesting nest)
5037 {
5038 struct btrfs_fs_info *fs_info = root->fs_info;
5039 struct extent_buffer *buf;
5040 u64 lockdep_owner = owner;
5041
5042 buf = btrfs_find_create_tree_block(fs_info, bytenr, owner, level);
5043 if (IS_ERR(buf))
5044 return buf;
5045
5046 if (check_eb_lock_owner(buf)) {
5047 free_extent_buffer(buf);
5048 return ERR_PTR(-EUCLEAN);
5049 }
5050
5051 /*
5052 * The reloc trees are just snapshots, so we need them to appear to be
5053 * just like any other fs tree WRT lockdep.
5054 *
5055 * The exception however is in replace_path() in relocation, where we
5056 * hold the lock on the original fs root and then search for the reloc
5057 * root. At that point we need to make sure any reloc root buffers are
5058 * set to the BTRFS_TREE_RELOC_OBJECTID lockdep class in order to make
5059 * lockdep happy.
5060 */
5061 if (lockdep_owner == BTRFS_TREE_RELOC_OBJECTID &&
5062 !test_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &root->state))
5063 lockdep_owner = BTRFS_FS_TREE_OBJECTID;
5064
5065 /* btrfs_clear_buffer_dirty() accesses generation field. */
5066 btrfs_set_header_generation(buf, trans->transid);
5067
5068 /*
5069 * This needs to stay, because we could allocate a freed block from an
5070 * old tree into a new tree, so we need to make sure this new block is
5071 * set to the appropriate level and owner.
5072 */
5073 btrfs_set_buffer_lockdep_class(lockdep_owner, buf, level);
5074
5075 btrfs_tree_lock_nested(buf, nest);
5076 btrfs_clear_buffer_dirty(trans, buf);
5077 clear_bit(EXTENT_BUFFER_STALE, &buf->bflags);
5078 clear_bit(EXTENT_BUFFER_ZONED_ZEROOUT, &buf->bflags);
5079
5080 set_extent_buffer_uptodate(buf);
5081
5082 memzero_extent_buffer(buf, 0, sizeof(struct btrfs_header));
5083 btrfs_set_header_level(buf, level);
5084 btrfs_set_header_bytenr(buf, buf->start);
5085 btrfs_set_header_generation(buf, trans->transid);
5086 btrfs_set_header_backref_rev(buf, BTRFS_MIXED_BACKREF_REV);
5087 btrfs_set_header_owner(buf, owner);
5088 write_extent_buffer_fsid(buf, fs_info->fs_devices->metadata_uuid);
5089 write_extent_buffer_chunk_tree_uuid(buf, fs_info->chunk_tree_uuid);
5090 if (btrfs_root_id(root) == BTRFS_TREE_LOG_OBJECTID) {
5091 buf->log_index = root->log_transid % 2;
5092 /*
5093 * we allow two log transactions at a time, use different
5094 * EXTENT bit to differentiate dirty pages.
5095 */
5096 if (buf->log_index == 0)
5097 set_extent_bit(&root->dirty_log_pages, buf->start,
5098 buf->start + buf->len - 1,
5099 EXTENT_DIRTY, NULL);
5100 else
5101 set_extent_bit(&root->dirty_log_pages, buf->start,
5102 buf->start + buf->len - 1,
5103 EXTENT_NEW, NULL);
5104 } else {
5105 buf->log_index = -1;
5106 set_extent_bit(&trans->transaction->dirty_pages, buf->start,
5107 buf->start + buf->len - 1, EXTENT_DIRTY, NULL);
5108 }
5109 /* this returns a buffer locked for blocking */
5110 return buf;
5111 }
5112
5113 /*
5114 * finds a free extent and does all the dirty work required for allocation
5115 * returns the tree buffer or an ERR_PTR on error.
5116 */
btrfs_alloc_tree_block(struct btrfs_trans_handle * trans,struct btrfs_root * root,u64 parent,u64 root_objectid,const struct btrfs_disk_key * key,int level,u64 hint,u64 empty_size,u64 reloc_src_root,enum btrfs_lock_nesting nest)5117 struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
5118 struct btrfs_root *root,
5119 u64 parent, u64 root_objectid,
5120 const struct btrfs_disk_key *key,
5121 int level, u64 hint,
5122 u64 empty_size,
5123 u64 reloc_src_root,
5124 enum btrfs_lock_nesting nest)
5125 {
5126 struct btrfs_fs_info *fs_info = root->fs_info;
5127 struct btrfs_key ins;
5128 struct btrfs_block_rsv *block_rsv;
5129 struct extent_buffer *buf;
5130 u64 flags = 0;
5131 int ret;
5132 u32 blocksize = fs_info->nodesize;
5133 bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
5134 u64 owning_root;
5135
5136 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
5137 if (btrfs_is_testing(fs_info)) {
5138 buf = btrfs_init_new_buffer(trans, root, root->alloc_bytenr,
5139 level, root_objectid, nest);
5140 if (!IS_ERR(buf))
5141 root->alloc_bytenr += blocksize;
5142 return buf;
5143 }
5144 #endif
5145
5146 block_rsv = btrfs_use_block_rsv(trans, root, blocksize);
5147 if (IS_ERR(block_rsv))
5148 return ERR_CAST(block_rsv);
5149
5150 ret = btrfs_reserve_extent(root, blocksize, blocksize, blocksize,
5151 empty_size, hint, &ins, 0, 0);
5152 if (ret)
5153 goto out_unuse;
5154
5155 buf = btrfs_init_new_buffer(trans, root, ins.objectid, level,
5156 root_objectid, nest);
5157 if (IS_ERR(buf)) {
5158 ret = PTR_ERR(buf);
5159 goto out_free_reserved;
5160 }
5161 owning_root = btrfs_header_owner(buf);
5162
5163 if (root_objectid == BTRFS_TREE_RELOC_OBJECTID) {
5164 if (parent == 0)
5165 parent = ins.objectid;
5166 flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
5167 owning_root = reloc_src_root;
5168 } else
5169 BUG_ON(parent > 0);
5170
5171 if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
5172 struct btrfs_delayed_extent_op *extent_op;
5173 struct btrfs_ref generic_ref = {
5174 .action = BTRFS_ADD_DELAYED_EXTENT,
5175 .bytenr = ins.objectid,
5176 .num_bytes = ins.offset,
5177 .parent = parent,
5178 .owning_root = owning_root,
5179 .ref_root = root_objectid,
5180 };
5181
5182 if (!skinny_metadata || flags != 0) {
5183 extent_op = btrfs_alloc_delayed_extent_op();
5184 if (!extent_op) {
5185 ret = -ENOMEM;
5186 goto out_free_buf;
5187 }
5188 if (key)
5189 memcpy(&extent_op->key, key, sizeof(extent_op->key));
5190 else
5191 memset(&extent_op->key, 0, sizeof(extent_op->key));
5192 extent_op->flags_to_set = flags;
5193 extent_op->update_key = (skinny_metadata ? false : true);
5194 extent_op->update_flags = (flags != 0);
5195 } else {
5196 extent_op = NULL;
5197 }
5198
5199 btrfs_init_tree_ref(&generic_ref, level, btrfs_root_id(root), false);
5200 btrfs_ref_tree_mod(fs_info, &generic_ref);
5201 ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, extent_op);
5202 if (ret) {
5203 btrfs_free_delayed_extent_op(extent_op);
5204 goto out_free_buf;
5205 }
5206 }
5207 return buf;
5208
5209 out_free_buf:
5210 btrfs_tree_unlock(buf);
5211 free_extent_buffer(buf);
5212 out_free_reserved:
5213 btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 0);
5214 out_unuse:
5215 btrfs_unuse_block_rsv(fs_info, block_rsv, blocksize);
5216 return ERR_PTR(ret);
5217 }
5218
5219 struct walk_control {
5220 u64 refs[BTRFS_MAX_LEVEL];
5221 u64 flags[BTRFS_MAX_LEVEL];
5222 struct btrfs_key update_progress;
5223 struct btrfs_key drop_progress;
5224 int drop_level;
5225 int stage;
5226 int level;
5227 int shared_level;
5228 int update_ref;
5229 int keep_locks;
5230 int reada_slot;
5231 int reada_count;
5232 int restarted;
5233 /* Indicate that extent info needs to be looked up when walking the tree. */
5234 int lookup_info;
5235 };
5236
5237 /*
5238 * This is our normal stage. We are traversing blocks the current snapshot owns
5239 * and we are dropping any of our references to any children we are able to, and
5240 * then freeing the block once we've processed all of the children.
5241 */
5242 #define DROP_REFERENCE 1
5243
5244 /*
5245 * We enter this stage when we have to walk into a child block (meaning we can't
5246 * simply drop our reference to it from our current parent node) and there are
5247 * more than one reference on it. If we are the owner of any of the children
5248 * blocks from the current parent node then we have to do the FULL_BACKREF dance
5249 * on them in order to drop our normal ref and add the shared ref.
5250 */
5251 #define UPDATE_BACKREF 2
5252
5253 /*
5254 * Decide if we need to walk down into this node to adjust the references.
5255 *
5256 * @root: the root we are currently deleting
5257 * @wc: the walk control for this deletion
5258 * @eb: the parent eb that we're currently visiting
5259 * @refs: the number of refs for wc->level - 1
5260 * @flags: the flags for wc->level - 1
5261 * @slot: the slot in the eb that we're currently checking
5262 *
5263 * This is meant to be called when we're evaluating if a node we point to at
5264 * wc->level should be read and walked into, or if we can simply delete our
5265 * reference to it. We return true if we should walk into the node, false if we
5266 * can skip it.
5267 *
5268 * We have assertions in here to make sure this is called correctly. We assume
5269 * that sanity checking on the blocks read to this point has been done, so any
5270 * corrupted file systems must have been caught before calling this function.
5271 */
visit_node_for_delete(struct btrfs_root * root,struct walk_control * wc,struct extent_buffer * eb,u64 refs,u64 flags,int slot)5272 static bool visit_node_for_delete(struct btrfs_root *root, struct walk_control *wc,
5273 struct extent_buffer *eb, u64 refs, u64 flags, int slot)
5274 {
5275 struct btrfs_key key;
5276 u64 generation;
5277 int level = wc->level;
5278
5279 ASSERT(level > 0);
5280 ASSERT(wc->refs[level - 1] > 0);
5281
5282 /*
5283 * The update backref stage we only want to skip if we already have
5284 * FULL_BACKREF set, otherwise we need to read.
5285 */
5286 if (wc->stage == UPDATE_BACKREF) {
5287 if (level == 1 && flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5288 return false;
5289 return true;
5290 }
5291
5292 /*
5293 * We're the last ref on this block, we must walk into it and process
5294 * any refs it's pointing at.
5295 */
5296 if (wc->refs[level - 1] == 1)
5297 return true;
5298
5299 /*
5300 * If we're already FULL_BACKREF then we know we can just drop our
5301 * current reference.
5302 */
5303 if (level == 1 && flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5304 return false;
5305
5306 /*
5307 * This block is older than our creation generation, we can drop our
5308 * reference to it.
5309 */
5310 generation = btrfs_node_ptr_generation(eb, slot);
5311 if (!wc->update_ref || generation <= root->root_key.offset)
5312 return false;
5313
5314 /*
5315 * This block was processed from a previous snapshot deletion run, we
5316 * can skip it.
5317 */
5318 btrfs_node_key_to_cpu(eb, &key, slot);
5319 if (btrfs_comp_cpu_keys(&key, &wc->update_progress) < 0)
5320 return false;
5321
5322 /* All other cases we need to wander into the node. */
5323 return true;
5324 }
5325
reada_walk_down(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct walk_control * wc,struct btrfs_path * path)5326 static noinline void reada_walk_down(struct btrfs_trans_handle *trans,
5327 struct btrfs_root *root,
5328 struct walk_control *wc,
5329 struct btrfs_path *path)
5330 {
5331 struct btrfs_fs_info *fs_info = root->fs_info;
5332 u64 bytenr;
5333 u64 generation;
5334 u64 refs;
5335 u64 flags;
5336 u32 nritems;
5337 struct extent_buffer *eb;
5338 int ret;
5339 int slot;
5340 int nread = 0;
5341
5342 if (path->slots[wc->level] < wc->reada_slot) {
5343 wc->reada_count = wc->reada_count * 2 / 3;
5344 wc->reada_count = max(wc->reada_count, 2);
5345 } else {
5346 wc->reada_count = wc->reada_count * 3 / 2;
5347 wc->reada_count = min_t(int, wc->reada_count,
5348 BTRFS_NODEPTRS_PER_BLOCK(fs_info));
5349 }
5350
5351 eb = path->nodes[wc->level];
5352 nritems = btrfs_header_nritems(eb);
5353
5354 for (slot = path->slots[wc->level]; slot < nritems; slot++) {
5355 if (nread >= wc->reada_count)
5356 break;
5357
5358 cond_resched();
5359 bytenr = btrfs_node_blockptr(eb, slot);
5360 generation = btrfs_node_ptr_generation(eb, slot);
5361
5362 if (slot == path->slots[wc->level])
5363 goto reada;
5364
5365 if (wc->stage == UPDATE_BACKREF &&
5366 generation <= root->root_key.offset)
5367 continue;
5368
5369 /* We don't lock the tree block, it's OK to be racy here */
5370 ret = btrfs_lookup_extent_info(trans, fs_info, bytenr,
5371 wc->level - 1, 1, &refs,
5372 &flags, NULL);
5373 /* We don't care about errors in readahead. */
5374 if (ret < 0)
5375 continue;
5376
5377 /*
5378 * This could be racey, it's conceivable that we raced and end
5379 * up with a bogus refs count, if that's the case just skip, if
5380 * we are actually corrupt we will notice when we look up
5381 * everything again with our locks.
5382 */
5383 if (refs == 0)
5384 continue;
5385
5386 /* If we don't need to visit this node don't reada. */
5387 if (!visit_node_for_delete(root, wc, eb, refs, flags, slot))
5388 continue;
5389 reada:
5390 btrfs_readahead_node_child(eb, slot);
5391 nread++;
5392 }
5393 wc->reada_slot = slot;
5394 }
5395
5396 /*
5397 * helper to process tree block while walking down the tree.
5398 *
5399 * when wc->stage == UPDATE_BACKREF, this function updates
5400 * back refs for pointers in the block.
5401 *
5402 * NOTE: return value 1 means we should stop walking down.
5403 */
walk_down_proc(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct walk_control * wc)5404 static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
5405 struct btrfs_root *root,
5406 struct btrfs_path *path,
5407 struct walk_control *wc)
5408 {
5409 struct btrfs_fs_info *fs_info = root->fs_info;
5410 int level = wc->level;
5411 struct extent_buffer *eb = path->nodes[level];
5412 u64 flag = BTRFS_BLOCK_FLAG_FULL_BACKREF;
5413 int ret;
5414
5415 if (wc->stage == UPDATE_BACKREF && btrfs_header_owner(eb) != btrfs_root_id(root))
5416 return 1;
5417
5418 /*
5419 * when reference count of tree block is 1, it won't increase
5420 * again. once full backref flag is set, we never clear it.
5421 */
5422 if (wc->lookup_info &&
5423 ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) ||
5424 (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag)))) {
5425 ASSERT(path->locks[level]);
5426 ret = btrfs_lookup_extent_info(trans, fs_info,
5427 eb->start, level, 1,
5428 &wc->refs[level],
5429 &wc->flags[level],
5430 NULL);
5431 if (ret)
5432 return ret;
5433 if (unlikely(wc->refs[level] == 0)) {
5434 btrfs_err(fs_info, "bytenr %llu has 0 references, expect > 0",
5435 eb->start);
5436 return -EUCLEAN;
5437 }
5438 }
5439
5440 if (wc->stage == DROP_REFERENCE) {
5441 if (wc->refs[level] > 1)
5442 return 1;
5443
5444 if (path->locks[level] && !wc->keep_locks) {
5445 btrfs_tree_unlock_rw(eb, path->locks[level]);
5446 path->locks[level] = 0;
5447 }
5448 return 0;
5449 }
5450
5451 /* wc->stage == UPDATE_BACKREF */
5452 if (!(wc->flags[level] & flag)) {
5453 ASSERT(path->locks[level]);
5454 ret = btrfs_inc_ref(trans, root, eb, 1);
5455 if (ret) {
5456 btrfs_abort_transaction(trans, ret);
5457 return ret;
5458 }
5459 ret = btrfs_dec_ref(trans, root, eb, 0);
5460 if (ret) {
5461 btrfs_abort_transaction(trans, ret);
5462 return ret;
5463 }
5464 ret = btrfs_set_disk_extent_flags(trans, eb, flag);
5465 if (ret) {
5466 btrfs_abort_transaction(trans, ret);
5467 return ret;
5468 }
5469 wc->flags[level] |= flag;
5470 }
5471
5472 /*
5473 * the block is shared by multiple trees, so it's not good to
5474 * keep the tree lock
5475 */
5476 if (path->locks[level] && level > 0) {
5477 btrfs_tree_unlock_rw(eb, path->locks[level]);
5478 path->locks[level] = 0;
5479 }
5480 return 0;
5481 }
5482
5483 /*
5484 * This is used to verify a ref exists for this root to deal with a bug where we
5485 * would have a drop_progress key that hadn't been updated properly.
5486 */
check_ref_exists(struct btrfs_trans_handle * trans,struct btrfs_root * root,u64 bytenr,u64 parent,int level)5487 static int check_ref_exists(struct btrfs_trans_handle *trans,
5488 struct btrfs_root *root, u64 bytenr, u64 parent,
5489 int level)
5490 {
5491 struct btrfs_delayed_ref_root *delayed_refs;
5492 struct btrfs_delayed_ref_head *head;
5493 struct btrfs_path *path;
5494 struct btrfs_extent_inline_ref *iref;
5495 int ret;
5496 bool exists = false;
5497
5498 path = btrfs_alloc_path();
5499 if (!path)
5500 return -ENOMEM;
5501 again:
5502 ret = lookup_extent_backref(trans, path, &iref, bytenr,
5503 root->fs_info->nodesize, parent,
5504 btrfs_root_id(root), level, 0);
5505 if (ret != -ENOENT) {
5506 /*
5507 * If we get 0 then we found our reference, return 1, else
5508 * return the error if it's not -ENOENT;
5509 */
5510 btrfs_free_path(path);
5511 return (ret < 0 ) ? ret : 1;
5512 }
5513
5514 /*
5515 * We could have a delayed ref with this reference, so look it up while
5516 * we're holding the path open to make sure we don't race with the
5517 * delayed ref running.
5518 */
5519 delayed_refs = &trans->transaction->delayed_refs;
5520 spin_lock(&delayed_refs->lock);
5521 head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
5522 if (!head)
5523 goto out;
5524 if (!mutex_trylock(&head->mutex)) {
5525 /*
5526 * We're contended, means that the delayed ref is running, get a
5527 * reference and wait for the ref head to be complete and then
5528 * try again.
5529 */
5530 refcount_inc(&head->refs);
5531 spin_unlock(&delayed_refs->lock);
5532
5533 btrfs_release_path(path);
5534
5535 mutex_lock(&head->mutex);
5536 mutex_unlock(&head->mutex);
5537 btrfs_put_delayed_ref_head(head);
5538 goto again;
5539 }
5540
5541 exists = btrfs_find_delayed_tree_ref(head, root->root_key.objectid, parent);
5542 mutex_unlock(&head->mutex);
5543 out:
5544 spin_unlock(&delayed_refs->lock);
5545 btrfs_free_path(path);
5546 return exists ? 1 : 0;
5547 }
5548
5549 /*
5550 * We may not have an uptodate block, so if we are going to walk down into this
5551 * block we need to drop the lock, read it off of the disk, re-lock it and
5552 * return to continue dropping the snapshot.
5553 */
check_next_block_uptodate(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct walk_control * wc,struct extent_buffer * next)5554 static int check_next_block_uptodate(struct btrfs_trans_handle *trans,
5555 struct btrfs_root *root,
5556 struct btrfs_path *path,
5557 struct walk_control *wc,
5558 struct extent_buffer *next)
5559 {
5560 struct btrfs_tree_parent_check check = { 0 };
5561 u64 generation;
5562 int level = wc->level;
5563 int ret;
5564
5565 btrfs_assert_tree_write_locked(next);
5566
5567 generation = btrfs_node_ptr_generation(path->nodes[level], path->slots[level]);
5568
5569 if (btrfs_buffer_uptodate(next, generation, 0))
5570 return 0;
5571
5572 check.level = level - 1;
5573 check.transid = generation;
5574 check.owner_root = btrfs_root_id(root);
5575 check.has_first_key = true;
5576 btrfs_node_key_to_cpu(path->nodes[level], &check.first_key, path->slots[level]);
5577
5578 btrfs_tree_unlock(next);
5579 if (level == 1)
5580 reada_walk_down(trans, root, wc, path);
5581 ret = btrfs_read_extent_buffer(next, &check);
5582 if (ret) {
5583 free_extent_buffer(next);
5584 return ret;
5585 }
5586 btrfs_tree_lock(next);
5587 wc->lookup_info = 1;
5588 return 0;
5589 }
5590
5591 /*
5592 * If we determine that we don't have to visit wc->level - 1 then we need to
5593 * determine if we can drop our reference.
5594 *
5595 * If we are UPDATE_BACKREF then we will not, we need to update our backrefs.
5596 *
5597 * If we are DROP_REFERENCE this will figure out if we need to drop our current
5598 * reference, skipping it if we dropped it from a previous incompleted drop, or
5599 * dropping it if we still have a reference to it.
5600 */
maybe_drop_reference(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct walk_control * wc,struct extent_buffer * next,u64 owner_root)5601 static int maybe_drop_reference(struct btrfs_trans_handle *trans, struct btrfs_root *root,
5602 struct btrfs_path *path, struct walk_control *wc,
5603 struct extent_buffer *next, u64 owner_root)
5604 {
5605 struct btrfs_ref ref = {
5606 .action = BTRFS_DROP_DELAYED_REF,
5607 .bytenr = next->start,
5608 .num_bytes = root->fs_info->nodesize,
5609 .owning_root = owner_root,
5610 .ref_root = btrfs_root_id(root),
5611 };
5612 int level = wc->level;
5613 int ret;
5614
5615 /* We are UPDATE_BACKREF, we're not dropping anything. */
5616 if (wc->stage == UPDATE_BACKREF)
5617 return 0;
5618
5619 if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
5620 ref.parent = path->nodes[level]->start;
5621 } else {
5622 ASSERT(btrfs_root_id(root) == btrfs_header_owner(path->nodes[level]));
5623 if (btrfs_root_id(root) != btrfs_header_owner(path->nodes[level])) {
5624 btrfs_err(root->fs_info, "mismatched block owner");
5625 return -EIO;
5626 }
5627 }
5628
5629 /*
5630 * If we had a drop_progress we need to verify the refs are set as
5631 * expected. If we find our ref then we know that from here on out
5632 * everything should be correct, and we can clear the
5633 * ->restarted flag.
5634 */
5635 if (wc->restarted) {
5636 ret = check_ref_exists(trans, root, next->start, ref.parent,
5637 level - 1);
5638 if (ret <= 0)
5639 return ret;
5640 ret = 0;
5641 wc->restarted = 0;
5642 }
5643
5644 /*
5645 * Reloc tree doesn't contribute to qgroup numbers, and we have already
5646 * accounted them at merge time (replace_path), thus we could skip
5647 * expensive subtree trace here.
5648 */
5649 if (btrfs_root_id(root) != BTRFS_TREE_RELOC_OBJECTID &&
5650 wc->refs[level - 1] > 1) {
5651 u64 generation = btrfs_node_ptr_generation(path->nodes[level],
5652 path->slots[level]);
5653
5654 ret = btrfs_qgroup_trace_subtree(trans, next, generation, level - 1);
5655 if (ret) {
5656 btrfs_err_rl(root->fs_info,
5657 "error %d accounting shared subtree, quota is out of sync, rescan required",
5658 ret);
5659 }
5660 }
5661
5662 /*
5663 * We need to update the next key in our walk control so we can update
5664 * the drop_progress key accordingly. We don't care if find_next_key
5665 * doesn't find a key because that means we're at the end and are going
5666 * to clean up now.
5667 */
5668 wc->drop_level = level;
5669 find_next_key(path, level, &wc->drop_progress);
5670
5671 btrfs_init_tree_ref(&ref, level - 1, 0, false);
5672 return btrfs_free_extent(trans, &ref);
5673 }
5674
5675 /*
5676 * helper to process tree block pointer.
5677 *
5678 * when wc->stage == DROP_REFERENCE, this function checks
5679 * reference count of the block pointed to. if the block
5680 * is shared and we need update back refs for the subtree
5681 * rooted at the block, this function changes wc->stage to
5682 * UPDATE_BACKREF. if the block is shared and there is no
5683 * need to update back, this function drops the reference
5684 * to the block.
5685 *
5686 * NOTE: return value 1 means we should stop walking down.
5687 */
do_walk_down(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct walk_control * wc)5688 static noinline int do_walk_down(struct btrfs_trans_handle *trans,
5689 struct btrfs_root *root,
5690 struct btrfs_path *path,
5691 struct walk_control *wc)
5692 {
5693 struct btrfs_fs_info *fs_info = root->fs_info;
5694 u64 bytenr;
5695 u64 generation;
5696 u64 owner_root = 0;
5697 struct extent_buffer *next;
5698 int level = wc->level;
5699 int ret = 0;
5700
5701 generation = btrfs_node_ptr_generation(path->nodes[level],
5702 path->slots[level]);
5703 /*
5704 * if the lower level block was created before the snapshot
5705 * was created, we know there is no need to update back refs
5706 * for the subtree
5707 */
5708 if (wc->stage == UPDATE_BACKREF &&
5709 generation <= root->root_key.offset) {
5710 wc->lookup_info = 1;
5711 return 1;
5712 }
5713
5714 bytenr = btrfs_node_blockptr(path->nodes[level], path->slots[level]);
5715
5716 next = btrfs_find_create_tree_block(fs_info, bytenr, btrfs_root_id(root),
5717 level - 1);
5718 if (IS_ERR(next))
5719 return PTR_ERR(next);
5720
5721 btrfs_tree_lock(next);
5722
5723 ret = btrfs_lookup_extent_info(trans, fs_info, bytenr, level - 1, 1,
5724 &wc->refs[level - 1],
5725 &wc->flags[level - 1],
5726 &owner_root);
5727 if (ret < 0)
5728 goto out_unlock;
5729
5730 if (unlikely(wc->refs[level - 1] == 0)) {
5731 btrfs_err(fs_info, "bytenr %llu has 0 references, expect > 0",
5732 bytenr);
5733 ret = -EUCLEAN;
5734 goto out_unlock;
5735 }
5736 wc->lookup_info = 0;
5737
5738 /* If we don't have to walk into this node skip it. */
5739 if (!visit_node_for_delete(root, wc, path->nodes[level],
5740 wc->refs[level - 1], wc->flags[level - 1],
5741 path->slots[level]))
5742 goto skip;
5743
5744 /*
5745 * We have to walk down into this node, and if we're currently at the
5746 * DROP_REFERNCE stage and this block is shared then we need to switch
5747 * to the UPDATE_BACKREF stage in order to convert to FULL_BACKREF.
5748 */
5749 if (wc->stage == DROP_REFERENCE && wc->refs[level - 1] > 1) {
5750 wc->stage = UPDATE_BACKREF;
5751 wc->shared_level = level - 1;
5752 }
5753
5754 ret = check_next_block_uptodate(trans, root, path, wc, next);
5755 if (ret)
5756 return ret;
5757
5758 level--;
5759 ASSERT(level == btrfs_header_level(next));
5760 if (level != btrfs_header_level(next)) {
5761 btrfs_err(root->fs_info, "mismatched level");
5762 ret = -EIO;
5763 goto out_unlock;
5764 }
5765 path->nodes[level] = next;
5766 path->slots[level] = 0;
5767 path->locks[level] = BTRFS_WRITE_LOCK;
5768 wc->level = level;
5769 if (wc->level == 1)
5770 wc->reada_slot = 0;
5771 return 0;
5772 skip:
5773 ret = maybe_drop_reference(trans, root, path, wc, next, owner_root);
5774 if (ret)
5775 goto out_unlock;
5776 wc->refs[level - 1] = 0;
5777 wc->flags[level - 1] = 0;
5778 wc->lookup_info = 1;
5779 ret = 1;
5780
5781 out_unlock:
5782 btrfs_tree_unlock(next);
5783 free_extent_buffer(next);
5784
5785 return ret;
5786 }
5787
5788 /*
5789 * helper to process tree block while walking up the tree.
5790 *
5791 * when wc->stage == DROP_REFERENCE, this function drops
5792 * reference count on the block.
5793 *
5794 * when wc->stage == UPDATE_BACKREF, this function changes
5795 * wc->stage back to DROP_REFERENCE if we changed wc->stage
5796 * to UPDATE_BACKREF previously while processing the block.
5797 *
5798 * NOTE: return value 1 means we should stop walking up.
5799 */
walk_up_proc(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct walk_control * wc)5800 static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
5801 struct btrfs_root *root,
5802 struct btrfs_path *path,
5803 struct walk_control *wc)
5804 {
5805 struct btrfs_fs_info *fs_info = root->fs_info;
5806 int ret = 0;
5807 int level = wc->level;
5808 struct extent_buffer *eb = path->nodes[level];
5809 u64 parent = 0;
5810
5811 if (wc->stage == UPDATE_BACKREF) {
5812 ASSERT(wc->shared_level >= level);
5813 if (level < wc->shared_level)
5814 goto out;
5815
5816 ret = find_next_key(path, level + 1, &wc->update_progress);
5817 if (ret > 0)
5818 wc->update_ref = 0;
5819
5820 wc->stage = DROP_REFERENCE;
5821 wc->shared_level = -1;
5822 path->slots[level] = 0;
5823
5824 /*
5825 * check reference count again if the block isn't locked.
5826 * we should start walking down the tree again if reference
5827 * count is one.
5828 */
5829 if (!path->locks[level]) {
5830 ASSERT(level > 0);
5831 btrfs_tree_lock(eb);
5832 path->locks[level] = BTRFS_WRITE_LOCK;
5833
5834 ret = btrfs_lookup_extent_info(trans, fs_info,
5835 eb->start, level, 1,
5836 &wc->refs[level],
5837 &wc->flags[level],
5838 NULL);
5839 if (ret < 0) {
5840 btrfs_tree_unlock_rw(eb, path->locks[level]);
5841 path->locks[level] = 0;
5842 return ret;
5843 }
5844 if (unlikely(wc->refs[level] == 0)) {
5845 btrfs_tree_unlock_rw(eb, path->locks[level]);
5846 btrfs_err(fs_info, "bytenr %llu has 0 references, expect > 0",
5847 eb->start);
5848 return -EUCLEAN;
5849 }
5850 if (wc->refs[level] == 1) {
5851 btrfs_tree_unlock_rw(eb, path->locks[level]);
5852 path->locks[level] = 0;
5853 return 1;
5854 }
5855 }
5856 }
5857
5858 /* wc->stage == DROP_REFERENCE */
5859 ASSERT(path->locks[level] || wc->refs[level] == 1);
5860
5861 if (wc->refs[level] == 1) {
5862 if (level == 0) {
5863 if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5864 ret = btrfs_dec_ref(trans, root, eb, 1);
5865 else
5866 ret = btrfs_dec_ref(trans, root, eb, 0);
5867 if (ret) {
5868 btrfs_abort_transaction(trans, ret);
5869 return ret;
5870 }
5871 if (is_fstree(btrfs_root_id(root))) {
5872 ret = btrfs_qgroup_trace_leaf_items(trans, eb);
5873 if (ret) {
5874 btrfs_err_rl(fs_info,
5875 "error %d accounting leaf items, quota is out of sync, rescan required",
5876 ret);
5877 }
5878 }
5879 }
5880 /* Make block locked assertion in btrfs_clear_buffer_dirty happy. */
5881 if (!path->locks[level]) {
5882 btrfs_tree_lock(eb);
5883 path->locks[level] = BTRFS_WRITE_LOCK;
5884 }
5885 btrfs_clear_buffer_dirty(trans, eb);
5886 }
5887
5888 if (eb == root->node) {
5889 if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5890 parent = eb->start;
5891 else if (btrfs_root_id(root) != btrfs_header_owner(eb))
5892 goto owner_mismatch;
5893 } else {
5894 if (wc->flags[level + 1] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5895 parent = path->nodes[level + 1]->start;
5896 else if (btrfs_root_id(root) !=
5897 btrfs_header_owner(path->nodes[level + 1]))
5898 goto owner_mismatch;
5899 }
5900
5901 ret = btrfs_free_tree_block(trans, btrfs_root_id(root), eb, parent,
5902 wc->refs[level] == 1);
5903 if (ret < 0)
5904 btrfs_abort_transaction(trans, ret);
5905 out:
5906 wc->refs[level] = 0;
5907 wc->flags[level] = 0;
5908 return ret;
5909
5910 owner_mismatch:
5911 btrfs_err_rl(fs_info, "unexpected tree owner, have %llu expect %llu",
5912 btrfs_header_owner(eb), btrfs_root_id(root));
5913 return -EUCLEAN;
5914 }
5915
5916 /*
5917 * walk_down_tree consists of two steps.
5918 *
5919 * walk_down_proc(). Look up the reference count and reference of our current
5920 * wc->level. At this point path->nodes[wc->level] should be populated and
5921 * uptodate, and in most cases should already be locked. If we are in
5922 * DROP_REFERENCE and our refcount is > 1 then we've entered a shared node and
5923 * we can walk back up the tree. If we are UPDATE_BACKREF we have to set
5924 * FULL_BACKREF on this node if it's not already set, and then do the
5925 * FULL_BACKREF conversion dance, which is to drop the root reference and add
5926 * the shared reference to all of this nodes children.
5927 *
5928 * do_walk_down(). This is where we actually start iterating on the children of
5929 * our current path->nodes[wc->level]. For DROP_REFERENCE that means dropping
5930 * our reference to the children that return false from visit_node_for_delete(),
5931 * which has various conditions where we know we can just drop our reference
5932 * without visiting the node. For UPDATE_BACKREF we will skip any children that
5933 * visit_node_for_delete() returns false for, only walking down when necessary.
5934 * The bulk of the work for UPDATE_BACKREF occurs in the walk_up_tree() part of
5935 * snapshot deletion.
5936 */
walk_down_tree(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct walk_control * wc)5937 static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
5938 struct btrfs_root *root,
5939 struct btrfs_path *path,
5940 struct walk_control *wc)
5941 {
5942 int level = wc->level;
5943 int ret = 0;
5944
5945 wc->lookup_info = 1;
5946 while (level >= 0) {
5947 ret = walk_down_proc(trans, root, path, wc);
5948 if (ret)
5949 break;
5950
5951 if (level == 0)
5952 break;
5953
5954 if (path->slots[level] >=
5955 btrfs_header_nritems(path->nodes[level]))
5956 break;
5957
5958 ret = do_walk_down(trans, root, path, wc);
5959 if (ret > 0) {
5960 path->slots[level]++;
5961 continue;
5962 } else if (ret < 0)
5963 break;
5964 level = wc->level;
5965 }
5966 return (ret == 1) ? 0 : ret;
5967 }
5968
5969 /*
5970 * walk_up_tree() is responsible for making sure we visit every slot on our
5971 * current node, and if we're at the end of that node then we call
5972 * walk_up_proc() on our current node which will do one of a few things based on
5973 * our stage.
5974 *
5975 * UPDATE_BACKREF. If we wc->level is currently less than our wc->shared_level
5976 * then we need to walk back up the tree, and then going back down into the
5977 * other slots via walk_down_tree to update any other children from our original
5978 * wc->shared_level. Once we're at or above our wc->shared_level we can switch
5979 * back to DROP_REFERENCE, lookup the current nodes refs and flags, and carry on.
5980 *
5981 * DROP_REFERENCE. If our refs == 1 then we're going to free this tree block.
5982 * If we're level 0 then we need to btrfs_dec_ref() on all of the data extents
5983 * in our current leaf. After that we call btrfs_free_tree_block() on the
5984 * current node and walk up to the next node to walk down the next slot.
5985 */
walk_up_tree(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct btrfs_path * path,struct walk_control * wc,int max_level)5986 static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
5987 struct btrfs_root *root,
5988 struct btrfs_path *path,
5989 struct walk_control *wc, int max_level)
5990 {
5991 int level = wc->level;
5992 int ret;
5993
5994 path->slots[level] = btrfs_header_nritems(path->nodes[level]);
5995 while (level < max_level && path->nodes[level]) {
5996 wc->level = level;
5997 if (path->slots[level] + 1 <
5998 btrfs_header_nritems(path->nodes[level])) {
5999 path->slots[level]++;
6000 return 0;
6001 } else {
6002 ret = walk_up_proc(trans, root, path, wc);
6003 if (ret > 0)
6004 return 0;
6005 if (ret < 0)
6006 return ret;
6007
6008 if (path->locks[level]) {
6009 btrfs_tree_unlock_rw(path->nodes[level],
6010 path->locks[level]);
6011 path->locks[level] = 0;
6012 }
6013 free_extent_buffer(path->nodes[level]);
6014 path->nodes[level] = NULL;
6015 level++;
6016 }
6017 }
6018 return 1;
6019 }
6020
6021 /*
6022 * drop a subvolume tree.
6023 *
6024 * this function traverses the tree freeing any blocks that only
6025 * referenced by the tree.
6026 *
6027 * when a shared tree block is found. this function decreases its
6028 * reference count by one. if update_ref is true, this function
6029 * also make sure backrefs for the shared block and all lower level
6030 * blocks are properly updated.
6031 *
6032 * If called with for_reloc == 0, may exit early with -EAGAIN
6033 */
btrfs_drop_snapshot(struct btrfs_root * root,int update_ref,int for_reloc)6034 int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref, int for_reloc)
6035 {
6036 const bool is_reloc_root = (btrfs_root_id(root) == BTRFS_TREE_RELOC_OBJECTID);
6037 struct btrfs_fs_info *fs_info = root->fs_info;
6038 struct btrfs_path *path;
6039 struct btrfs_trans_handle *trans;
6040 struct btrfs_root *tree_root = fs_info->tree_root;
6041 struct btrfs_root_item *root_item = &root->root_item;
6042 struct walk_control *wc;
6043 struct btrfs_key key;
6044 const u64 rootid = btrfs_root_id(root);
6045 int ret = 0;
6046 int level;
6047 bool root_dropped = false;
6048 bool unfinished_drop = false;
6049
6050 btrfs_debug(fs_info, "Drop subvolume %llu", btrfs_root_id(root));
6051
6052 path = btrfs_alloc_path();
6053 if (!path) {
6054 ret = -ENOMEM;
6055 goto out;
6056 }
6057
6058 wc = kzalloc(sizeof(*wc), GFP_NOFS);
6059 if (!wc) {
6060 btrfs_free_path(path);
6061 ret = -ENOMEM;
6062 goto out;
6063 }
6064
6065 /*
6066 * Use join to avoid potential EINTR from transaction start. See
6067 * wait_reserve_ticket and the whole reservation callchain.
6068 */
6069 if (for_reloc)
6070 trans = btrfs_join_transaction(tree_root);
6071 else
6072 trans = btrfs_start_transaction(tree_root, 0);
6073 if (IS_ERR(trans)) {
6074 ret = PTR_ERR(trans);
6075 goto out_free;
6076 }
6077
6078 ret = btrfs_run_delayed_items(trans);
6079 if (ret)
6080 goto out_end_trans;
6081
6082 /*
6083 * This will help us catch people modifying the fs tree while we're
6084 * dropping it. It is unsafe to mess with the fs tree while it's being
6085 * dropped as we unlock the root node and parent nodes as we walk down
6086 * the tree, assuming nothing will change. If something does change
6087 * then we'll have stale information and drop references to blocks we've
6088 * already dropped.
6089 */
6090 set_bit(BTRFS_ROOT_DELETING, &root->state);
6091 unfinished_drop = test_bit(BTRFS_ROOT_UNFINISHED_DROP, &root->state);
6092
6093 if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
6094 level = btrfs_header_level(root->node);
6095 path->nodes[level] = btrfs_lock_root_node(root);
6096 path->slots[level] = 0;
6097 path->locks[level] = BTRFS_WRITE_LOCK;
6098 memset(&wc->update_progress, 0,
6099 sizeof(wc->update_progress));
6100 } else {
6101 btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
6102 memcpy(&wc->update_progress, &key,
6103 sizeof(wc->update_progress));
6104
6105 level = btrfs_root_drop_level(root_item);
6106 BUG_ON(level == 0);
6107 path->lowest_level = level;
6108 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
6109 path->lowest_level = 0;
6110 if (ret < 0)
6111 goto out_end_trans;
6112
6113 WARN_ON(ret > 0);
6114 ret = 0;
6115
6116 /*
6117 * unlock our path, this is safe because only this
6118 * function is allowed to delete this snapshot
6119 */
6120 btrfs_unlock_up_safe(path, 0);
6121
6122 level = btrfs_header_level(root->node);
6123 while (1) {
6124 btrfs_tree_lock(path->nodes[level]);
6125 path->locks[level] = BTRFS_WRITE_LOCK;
6126
6127 /*
6128 * btrfs_lookup_extent_info() returns 0 for success,
6129 * or < 0 for error.
6130 */
6131 ret = btrfs_lookup_extent_info(trans, fs_info,
6132 path->nodes[level]->start,
6133 level, 1, &wc->refs[level],
6134 &wc->flags[level], NULL);
6135 if (ret < 0)
6136 goto out_end_trans;
6137
6138 BUG_ON(wc->refs[level] == 0);
6139
6140 if (level == btrfs_root_drop_level(root_item))
6141 break;
6142
6143 btrfs_tree_unlock(path->nodes[level]);
6144 path->locks[level] = 0;
6145 WARN_ON(wc->refs[level] != 1);
6146 level--;
6147 }
6148 }
6149
6150 wc->restarted = test_bit(BTRFS_ROOT_DEAD_TREE, &root->state);
6151 wc->level = level;
6152 wc->shared_level = -1;
6153 wc->stage = DROP_REFERENCE;
6154 wc->update_ref = update_ref;
6155 wc->keep_locks = 0;
6156 wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(fs_info);
6157
6158 while (1) {
6159
6160 ret = walk_down_tree(trans, root, path, wc);
6161 if (ret < 0) {
6162 btrfs_abort_transaction(trans, ret);
6163 break;
6164 }
6165
6166 ret = walk_up_tree(trans, root, path, wc, BTRFS_MAX_LEVEL);
6167 if (ret < 0) {
6168 btrfs_abort_transaction(trans, ret);
6169 break;
6170 }
6171
6172 if (ret > 0) {
6173 BUG_ON(wc->stage != DROP_REFERENCE);
6174 ret = 0;
6175 break;
6176 }
6177
6178 if (wc->stage == DROP_REFERENCE) {
6179 wc->drop_level = wc->level;
6180 btrfs_node_key_to_cpu(path->nodes[wc->drop_level],
6181 &wc->drop_progress,
6182 path->slots[wc->drop_level]);
6183 }
6184 btrfs_cpu_key_to_disk(&root_item->drop_progress,
6185 &wc->drop_progress);
6186 btrfs_set_root_drop_level(root_item, wc->drop_level);
6187
6188 BUG_ON(wc->level == 0);
6189 if (btrfs_should_end_transaction(trans) ||
6190 (!for_reloc && btrfs_need_cleaner_sleep(fs_info))) {
6191 ret = btrfs_update_root(trans, tree_root,
6192 &root->root_key,
6193 root_item);
6194 if (ret) {
6195 btrfs_abort_transaction(trans, ret);
6196 goto out_end_trans;
6197 }
6198
6199 if (!is_reloc_root)
6200 btrfs_set_last_root_drop_gen(fs_info, trans->transid);
6201
6202 btrfs_end_transaction_throttle(trans);
6203 if (!for_reloc && btrfs_need_cleaner_sleep(fs_info)) {
6204 btrfs_debug(fs_info,
6205 "drop snapshot early exit");
6206 ret = -EAGAIN;
6207 goto out_free;
6208 }
6209
6210 /*
6211 * Use join to avoid potential EINTR from transaction
6212 * start. See wait_reserve_ticket and the whole
6213 * reservation callchain.
6214 */
6215 if (for_reloc)
6216 trans = btrfs_join_transaction(tree_root);
6217 else
6218 trans = btrfs_start_transaction(tree_root, 0);
6219 if (IS_ERR(trans)) {
6220 ret = PTR_ERR(trans);
6221 goto out_free;
6222 }
6223 }
6224 }
6225 btrfs_release_path(path);
6226 if (ret)
6227 goto out_end_trans;
6228
6229 ret = btrfs_del_root(trans, &root->root_key);
6230 if (ret) {
6231 btrfs_abort_transaction(trans, ret);
6232 goto out_end_trans;
6233 }
6234
6235 if (!is_reloc_root) {
6236 ret = btrfs_find_root(tree_root, &root->root_key, path,
6237 NULL, NULL);
6238 if (ret < 0) {
6239 btrfs_abort_transaction(trans, ret);
6240 goto out_end_trans;
6241 } else if (ret > 0) {
6242 ret = 0;
6243 /*
6244 * If we fail to delete the orphan item this time
6245 * around, it'll get picked up the next time.
6246 *
6247 * The most common failure here is just -ENOENT.
6248 */
6249 btrfs_del_orphan_item(trans, tree_root, btrfs_root_id(root));
6250 }
6251 }
6252
6253 /*
6254 * This subvolume is going to be completely dropped, and won't be
6255 * recorded as dirty roots, thus pertrans meta rsv will not be freed at
6256 * commit transaction time. So free it here manually.
6257 */
6258 btrfs_qgroup_convert_reserved_meta(root, INT_MAX);
6259 btrfs_qgroup_free_meta_all_pertrans(root);
6260
6261 if (test_bit(BTRFS_ROOT_IN_RADIX, &root->state))
6262 btrfs_add_dropped_root(trans, root);
6263 else
6264 btrfs_put_root(root);
6265 root_dropped = true;
6266 out_end_trans:
6267 if (!is_reloc_root)
6268 btrfs_set_last_root_drop_gen(fs_info, trans->transid);
6269
6270 btrfs_end_transaction_throttle(trans);
6271 out_free:
6272 kfree(wc);
6273 btrfs_free_path(path);
6274 out:
6275 if (!ret && root_dropped) {
6276 ret = btrfs_qgroup_cleanup_dropped_subvolume(fs_info, rootid);
6277 if (ret < 0)
6278 btrfs_warn_rl(fs_info,
6279 "failed to cleanup qgroup 0/%llu: %d",
6280 rootid, ret);
6281 ret = 0;
6282 }
6283 /*
6284 * We were an unfinished drop root, check to see if there are any
6285 * pending, and if not clear and wake up any waiters.
6286 */
6287 if (!ret && unfinished_drop)
6288 btrfs_maybe_wake_unfinished_drop(fs_info);
6289
6290 /*
6291 * So if we need to stop dropping the snapshot for whatever reason we
6292 * need to make sure to add it back to the dead root list so that we
6293 * keep trying to do the work later. This also cleans up roots if we
6294 * don't have it in the radix (like when we recover after a power fail
6295 * or unmount) so we don't leak memory.
6296 */
6297 if (!for_reloc && !root_dropped)
6298 btrfs_add_dead_root(root);
6299 return ret;
6300 }
6301
6302 /*
6303 * drop subtree rooted at tree block 'node'.
6304 *
6305 * NOTE: this function will unlock and release tree block 'node'
6306 * only used by relocation code
6307 */
btrfs_drop_subtree(struct btrfs_trans_handle * trans,struct btrfs_root * root,struct extent_buffer * node,struct extent_buffer * parent)6308 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
6309 struct btrfs_root *root,
6310 struct extent_buffer *node,
6311 struct extent_buffer *parent)
6312 {
6313 struct btrfs_fs_info *fs_info = root->fs_info;
6314 struct btrfs_path *path;
6315 struct walk_control *wc;
6316 int level;
6317 int parent_level;
6318 int ret = 0;
6319
6320 BUG_ON(btrfs_root_id(root) != BTRFS_TREE_RELOC_OBJECTID);
6321
6322 path = btrfs_alloc_path();
6323 if (!path)
6324 return -ENOMEM;
6325
6326 wc = kzalloc(sizeof(*wc), GFP_NOFS);
6327 if (!wc) {
6328 btrfs_free_path(path);
6329 return -ENOMEM;
6330 }
6331
6332 btrfs_assert_tree_write_locked(parent);
6333 parent_level = btrfs_header_level(parent);
6334 atomic_inc(&parent->refs);
6335 path->nodes[parent_level] = parent;
6336 path->slots[parent_level] = btrfs_header_nritems(parent);
6337
6338 btrfs_assert_tree_write_locked(node);
6339 level = btrfs_header_level(node);
6340 path->nodes[level] = node;
6341 path->slots[level] = 0;
6342 path->locks[level] = BTRFS_WRITE_LOCK;
6343
6344 wc->refs[parent_level] = 1;
6345 wc->flags[parent_level] = BTRFS_BLOCK_FLAG_FULL_BACKREF;
6346 wc->level = level;
6347 wc->shared_level = -1;
6348 wc->stage = DROP_REFERENCE;
6349 wc->update_ref = 0;
6350 wc->keep_locks = 1;
6351 wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(fs_info);
6352
6353 while (1) {
6354 ret = walk_down_tree(trans, root, path, wc);
6355 if (ret < 0)
6356 break;
6357
6358 ret = walk_up_tree(trans, root, path, wc, parent_level);
6359 if (ret) {
6360 if (ret > 0)
6361 ret = 0;
6362 break;
6363 }
6364 }
6365
6366 kfree(wc);
6367 btrfs_free_path(path);
6368 return ret;
6369 }
6370
6371 /*
6372 * Unpin the extent range in an error context and don't add the space back.
6373 * Errors are not propagated further.
6374 */
btrfs_error_unpin_extent_range(struct btrfs_fs_info * fs_info,u64 start,u64 end)6375 void btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info, u64 start, u64 end)
6376 {
6377 unpin_extent_range(fs_info, start, end, false);
6378 }
6379
6380 /*
6381 * It used to be that old block groups would be left around forever.
6382 * Iterating over them would be enough to trim unused space. Since we
6383 * now automatically remove them, we also need to iterate over unallocated
6384 * space.
6385 *
6386 * We don't want a transaction for this since the discard may take a
6387 * substantial amount of time. We don't require that a transaction be
6388 * running, but we do need to take a running transaction into account
6389 * to ensure that we're not discarding chunks that were released or
6390 * allocated in the current transaction.
6391 *
6392 * Holding the chunks lock will prevent other threads from allocating
6393 * or releasing chunks, but it won't prevent a running transaction
6394 * from committing and releasing the memory that the pending chunks
6395 * list head uses. For that, we need to take a reference to the
6396 * transaction and hold the commit root sem. We only need to hold
6397 * it while performing the free space search since we have already
6398 * held back allocations.
6399 */
btrfs_trim_free_extents(struct btrfs_device * device,u64 * trimmed)6400 static int btrfs_trim_free_extents(struct btrfs_device *device, u64 *trimmed)
6401 {
6402 u64 start = BTRFS_DEVICE_RANGE_RESERVED, len = 0, end = 0;
6403 int ret;
6404
6405 *trimmed = 0;
6406
6407 /* Discard not supported = nothing to do. */
6408 if (!bdev_max_discard_sectors(device->bdev))
6409 return 0;
6410
6411 /* Not writable = nothing to do. */
6412 if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state))
6413 return 0;
6414
6415 /* No free space = nothing to do. */
6416 if (device->total_bytes <= device->bytes_used)
6417 return 0;
6418
6419 ret = 0;
6420
6421 while (1) {
6422 struct btrfs_fs_info *fs_info = device->fs_info;
6423 u64 bytes;
6424
6425 ret = mutex_lock_interruptible(&fs_info->chunk_mutex);
6426 if (ret)
6427 break;
6428
6429 find_first_clear_extent_bit(&device->alloc_state, start,
6430 &start, &end,
6431 CHUNK_TRIMMED | CHUNK_ALLOCATED);
6432
6433 /* Check if there are any CHUNK_* bits left */
6434 if (start > device->total_bytes) {
6435 WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
6436 btrfs_warn_in_rcu(fs_info,
6437 "ignoring attempt to trim beyond device size: offset %llu length %llu device %s device size %llu",
6438 start, end - start + 1,
6439 btrfs_dev_name(device),
6440 device->total_bytes);
6441 mutex_unlock(&fs_info->chunk_mutex);
6442 ret = 0;
6443 break;
6444 }
6445
6446 /* Ensure we skip the reserved space on each device. */
6447 start = max_t(u64, start, BTRFS_DEVICE_RANGE_RESERVED);
6448
6449 /*
6450 * If find_first_clear_extent_bit find a range that spans the
6451 * end of the device it will set end to -1, in this case it's up
6452 * to the caller to trim the value to the size of the device.
6453 */
6454 end = min(end, device->total_bytes - 1);
6455
6456 len = end - start + 1;
6457
6458 /* We didn't find any extents */
6459 if (!len) {
6460 mutex_unlock(&fs_info->chunk_mutex);
6461 ret = 0;
6462 break;
6463 }
6464
6465 ret = btrfs_issue_discard(device->bdev, start, len,
6466 &bytes);
6467 if (!ret)
6468 set_extent_bit(&device->alloc_state, start,
6469 start + bytes - 1, CHUNK_TRIMMED, NULL);
6470 mutex_unlock(&fs_info->chunk_mutex);
6471
6472 if (ret)
6473 break;
6474
6475 start += len;
6476 *trimmed += bytes;
6477
6478 if (btrfs_trim_interrupted()) {
6479 ret = -ERESTARTSYS;
6480 break;
6481 }
6482
6483 cond_resched();
6484 }
6485
6486 return ret;
6487 }
6488
6489 /*
6490 * Trim the whole filesystem by:
6491 * 1) trimming the free space in each block group
6492 * 2) trimming the unallocated space on each device
6493 *
6494 * This will also continue trimming even if a block group or device encounters
6495 * an error. The return value will be the last error, or 0 if nothing bad
6496 * happens.
6497 */
btrfs_trim_fs(struct btrfs_fs_info * fs_info,struct fstrim_range * range)6498 int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
6499 {
6500 struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
6501 struct btrfs_block_group *cache = NULL;
6502 struct btrfs_device *device;
6503 u64 group_trimmed;
6504 u64 range_end = U64_MAX;
6505 u64 start;
6506 u64 end;
6507 u64 trimmed = 0;
6508 u64 bg_failed = 0;
6509 u64 dev_failed = 0;
6510 int bg_ret = 0;
6511 int dev_ret = 0;
6512 int ret = 0;
6513
6514 if (range->start == U64_MAX)
6515 return -EINVAL;
6516
6517 /*
6518 * Check range overflow if range->len is set.
6519 * The default range->len is U64_MAX.
6520 */
6521 if (range->len != U64_MAX &&
6522 check_add_overflow(range->start, range->len, &range_end))
6523 return -EINVAL;
6524
6525 cache = btrfs_lookup_first_block_group(fs_info, range->start);
6526 for (; cache; cache = btrfs_next_block_group(cache)) {
6527 if (cache->start >= range_end) {
6528 btrfs_put_block_group(cache);
6529 break;
6530 }
6531
6532 start = max(range->start, cache->start);
6533 end = min(range_end, cache->start + cache->length);
6534
6535 if (end - start >= range->minlen) {
6536 if (!btrfs_block_group_done(cache)) {
6537 ret = btrfs_cache_block_group(cache, true);
6538 if (ret) {
6539 bg_failed++;
6540 bg_ret = ret;
6541 continue;
6542 }
6543 }
6544 ret = btrfs_trim_block_group(cache,
6545 &group_trimmed,
6546 start,
6547 end,
6548 range->minlen);
6549
6550 trimmed += group_trimmed;
6551 if (ret) {
6552 bg_failed++;
6553 bg_ret = ret;
6554 continue;
6555 }
6556 }
6557 }
6558
6559 if (bg_failed)
6560 btrfs_warn(fs_info,
6561 "failed to trim %llu block group(s), last error %d",
6562 bg_failed, bg_ret);
6563
6564 mutex_lock(&fs_devices->device_list_mutex);
6565 list_for_each_entry(device, &fs_devices->devices, dev_list) {
6566 if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state))
6567 continue;
6568
6569 ret = btrfs_trim_free_extents(device, &group_trimmed);
6570
6571 trimmed += group_trimmed;
6572 if (ret) {
6573 dev_failed++;
6574 dev_ret = ret;
6575 break;
6576 }
6577 }
6578 mutex_unlock(&fs_devices->device_list_mutex);
6579
6580 if (dev_failed)
6581 btrfs_warn(fs_info,
6582 "failed to trim %llu device(s), last error %d",
6583 dev_failed, dev_ret);
6584 range->len = trimmed;
6585 if (bg_ret)
6586 return bg_ret;
6587 return dev_ret;
6588 }
6589