?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
[linux-2.6/linux-acpi-2.6/ibm-acpi-2.6.git] / fs / btrfs / ctree.c
blob60a45f3a4e916fe9af5a8c644b3a8eb9b5b73f81
1 /*
2 * Copyright (C) 2007,2008 Oracle. All rights reserved.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
17 */
18
19 #include <linux/sched.h>
20 #include "ctree.h"
21 #include "disk-io.h"
22 #include "transaction.h"
23 #include "print-tree.h"
24 #include "locking.h"
25
26 static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root
27 *root, struct btrfs_path *path, int level);
28 static int split_leaf(struct btrfs_trans_handle *trans, struct btrfs_root
29 *root, struct btrfs_key *ins_key,
30 struct btrfs_path *path, int data_size, int extend);
31 static int push_node_left(struct btrfs_trans_handle *trans,
32 struct btrfs_root *root, struct extent_buffer *dst,
33 struct extent_buffer *src, int empty);
34 static int balance_node_right(struct btrfs_trans_handle *trans,
35 struct btrfs_root *root,
36 struct extent_buffer *dst_buf,
37 struct extent_buffer *src_buf);
38 static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
39 struct btrfs_path *path, int level, int slot);
40
41 struct btrfs_path *btrfs_alloc_path(void)
42 {
43 struct btrfs_path *path;
44 path = kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS);
45 if (path)
46 path->reada = 1;
47 return path;
48 }
49
50 /*
51 * set all locked nodes in the path to blocking locks. This should
52 * be done before scheduling
53 */
54 noinline void btrfs_set_path_blocking(struct btrfs_path *p)
55 {
56 int i;
57 for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
58 if (p->nodes[i] && p->locks[i])
59 btrfs_set_lock_blocking(p->nodes[i]);
60 }
61 }
62
63 /*
64 * reset all the locked nodes in the patch to spinning locks.
65 *
66 * held is used to keep lockdep happy, when lockdep is enabled
67 * we set held to a blocking lock before we go around and
68 * retake all the spinlocks in the path. You can safely use NULL
69 * for held
70 */
71 noinline void btrfs_clear_path_blocking(struct btrfs_path *p,
72 struct extent_buffer *held)
73 {
74 int i;
75
76 #ifdef CONFIG_DEBUG_LOCK_ALLOC
77 /* lockdep really cares that we take all of these spinlocks
78 * in the right order. If any of the locks in the path are not
79 * currently blocking, it is going to complain. So, make really
80 * really sure by forcing the path to blocking before we clear
81 * the path blocking.
82 */
83 if (held)
84 btrfs_set_lock_blocking(held);
85 btrfs_set_path_blocking(p);
86 #endif
87
88 for (i = BTRFS_MAX_LEVEL - 1; i >= 0; i--) {
89 if (p->nodes[i] && p->locks[i])
90 btrfs_clear_lock_blocking(p->nodes[i]);
91 }
92
93 #ifdef CONFIG_DEBUG_LOCK_ALLOC
94 if (held)
95 btrfs_clear_lock_blocking(held);
96 #endif
97 }
98
99 /* this also releases the path */
100 void btrfs_free_path(struct btrfs_path *p)
101 {
102 btrfs_release_path(NULL, p);
103 kmem_cache_free(btrfs_path_cachep, p);
104 }
105
106 /*
107 * path release drops references on the extent buffers in the path
108 * and it drops any locks held by this path
109 *
110 * It is safe to call this on paths that no locks or extent buffers held.
111 */
112 noinline void btrfs_release_path(struct btrfs_root *root, struct btrfs_path *p)
113 {
114 int i;
115
116 for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
117 p->slots[i] = 0;
118 if (!p->nodes[i])
119 continue;
120 if (p->locks[i]) {
121 btrfs_tree_unlock(p->nodes[i]);
122 p->locks[i] = 0;
123 }
124 free_extent_buffer(p->nodes[i]);
125 p->nodes[i] = NULL;
126 }
127 }
128
129 /*
130 * safely gets a reference on the root node of a tree. A lock
131 * is not taken, so a concurrent writer may put a different node
132 * at the root of the tree. See btrfs_lock_root_node for the
133 * looping required.
134 *
135 * The extent buffer returned by this has a reference taken, so
136 * it won't disappear. It may stop being the root of the tree
137 * at any time because there are no locks held.
138 */
139 struct extent_buffer *btrfs_root_node(struct btrfs_root *root)
140 {
141 struct extent_buffer *eb;
142 spin_lock(&root->node_lock);
143 eb = root->node;
144 extent_buffer_get(eb);
145 spin_unlock(&root->node_lock);
146 return eb;
147 }
148
149 /* loop around taking references on and locking the root node of the
150 * tree until you end up with a lock on the root. A locked buffer
151 * is returned, with a reference held.
152 */
153 struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root)
154 {
155 struct extent_buffer *eb;
156
157 while (1) {
158 eb = btrfs_root_node(root);
159 btrfs_tree_lock(eb);
160
161 spin_lock(&root->node_lock);
162 if (eb == root->node) {
163 spin_unlock(&root->node_lock);
164 break;
165 }
166 spin_unlock(&root->node_lock);
167
168 btrfs_tree_unlock(eb);
169 free_extent_buffer(eb);
170 }
171 return eb;
172 }
173
174 /* cowonly root (everything not a reference counted cow subvolume), just get
175 * put onto a simple dirty list. transaction.c walks this to make sure they
176 * get properly updated on disk.
177 */
178 static void add_root_to_dirty_list(struct btrfs_root *root)
179 {
180 if (root->track_dirty && list_empty(&root->dirty_list)) {
181 list_add(&root->dirty_list,
182 &root->fs_info->dirty_cowonly_roots);
183 }
184 }
185
186 /*
187 * used by snapshot creation to make a copy of a root for a tree with
188 * a given objectid. The buffer with the new root node is returned in
189 * cow_ret, and this func returns zero on success or a negative error code.
190 */
191 int btrfs_copy_root(struct btrfs_trans_handle *trans,
192 struct btrfs_root *root,
193 struct extent_buffer *buf,
194 struct extent_buffer **cow_ret, u64 new_root_objectid)
195 {
196 struct extent_buffer *cow;
197 u32 nritems;
198 int ret = 0;
199 int level;
200 struct btrfs_disk_key disk_key;
201
202 WARN_ON(root->ref_cows && trans->transid !=
203 root->fs_info->running_transaction->transid);
204 WARN_ON(root->ref_cows && trans->transid != root->last_trans);
205
206 level = btrfs_header_level(buf);
207 nritems = btrfs_header_nritems(buf);
208 if (level == 0)
209 btrfs_item_key(buf, &disk_key, 0);
210 else
211 btrfs_node_key(buf, &disk_key, 0);
212
213 cow = btrfs_alloc_free_block(trans, root, buf->len, 0,
214 new_root_objectid, &disk_key, level,
215 buf->start, 0);
216 if (IS_ERR(cow))
217 return PTR_ERR(cow);
218
219 copy_extent_buffer(cow, buf, 0, 0, cow->len);
220 btrfs_set_header_bytenr(cow, cow->start);
221 btrfs_set_header_generation(cow, trans->transid);
222 btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
223 btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
224 BTRFS_HEADER_FLAG_RELOC);
225 if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
226 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
227 else
228 btrfs_set_header_owner(cow, new_root_objectid);
229
230 write_extent_buffer(cow, root->fs_info->fsid,
231 (unsigned long)btrfs_header_fsid(cow),
232 BTRFS_FSID_SIZE);
233
234 WARN_ON(btrfs_header_generation(buf) > trans->transid);
235 if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
236 ret = btrfs_inc_ref(trans, root, cow, 1);
237 else
238 ret = btrfs_inc_ref(trans, root, cow, 0);
239
240 if (ret)
241 return ret;
242
243 btrfs_mark_buffer_dirty(cow);
244 *cow_ret = cow;
245 return 0;
246 }
247
248 /*
249 * check if the tree block can be shared by multiple trees
250 */
251 int btrfs_block_can_be_shared(struct btrfs_root *root,
252 struct extent_buffer *buf)
253 {
254 /*
255 * Tree blocks not in refernece counted trees and tree roots
256 * are never shared. If a block was allocated after the last
257 * snapshot and the block was not allocated by tree relocation,
258 * we know the block is not shared.
259 */
260 if (root->ref_cows &&
261 buf != root->node && buf != root->commit_root &&
262 (btrfs_header_generation(buf) <=
263 btrfs_root_last_snapshot(&root->root_item) ||
264 btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)))
265 return 1;
266 #ifdef BTRFS_COMPAT_EXTENT_TREE_V0
267 if (root->ref_cows &&
268 btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
269 return 1;
270 #endif
271 return 0;
272 }
273
274 static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
275 struct btrfs_root *root,
276 struct extent_buffer *buf,
277 struct extent_buffer *cow)
278 {
279 u64 refs;
280 u64 owner;
281 u64 flags;
282 u64 new_flags = 0;
283 int ret;
284
285 /*
286 * Backrefs update rules:
287 *
288 * Always use full backrefs for extent pointers in tree block
289 * allocated by tree relocation.
290 *
291 * If a shared tree block is no longer referenced by its owner
292 * tree (btrfs_header_owner(buf) == root->root_key.objectid),
293 * use full backrefs for extent pointers in tree block.
294 *
295 * If a tree block is been relocating
296 * (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID),
297 * use full backrefs for extent pointers in tree block.
298 * The reason for this is some operations (such as drop tree)
299 * are only allowed for blocks use full backrefs.
300 */
301
302 if (btrfs_block_can_be_shared(root, buf)) {
303 ret = btrfs_lookup_extent_info(trans, root, buf->start,
304 buf->len, &refs, &flags);
305 BUG_ON(ret);
306 BUG_ON(refs == 0);
307 } else {
308 refs = 1;
309 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
310 btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
311 flags = BTRFS_BLOCK_FLAG_FULL_BACKREF;
312 else
313 flags = 0;
314 }
315
316 owner = btrfs_header_owner(buf);
317 BUG_ON(owner == BTRFS_TREE_RELOC_OBJECTID &&
318 !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
319
320 if (refs > 1) {
321 if ((owner == root->root_key.objectid ||
322 root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) &&
323 !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) {
324 ret = btrfs_inc_ref(trans, root, buf, 1);
325 BUG_ON(ret);
326
327 if (root->root_key.objectid ==
328 BTRFS_TREE_RELOC_OBJECTID) {
329 ret = btrfs_dec_ref(trans, root, buf, 0);
330 BUG_ON(ret);
331 ret = btrfs_inc_ref(trans, root, cow, 1);
332 BUG_ON(ret);
333 }
334 new_flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
335 } else {
336
337 if (root->root_key.objectid ==
338 BTRFS_TREE_RELOC_OBJECTID)
339 ret = btrfs_inc_ref(trans, root, cow, 1);
340 else
341 ret = btrfs_inc_ref(trans, root, cow, 0);
342 BUG_ON(ret);
343 }
344 if (new_flags != 0) {
345 ret = btrfs_set_disk_extent_flags(trans, root,
346 buf->start,
347 buf->len,
348 new_flags, 0);
349 BUG_ON(ret);
350 }
351 } else {
352 if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
353 if (root->root_key.objectid ==
354 BTRFS_TREE_RELOC_OBJECTID)
355 ret = btrfs_inc_ref(trans, root, cow, 1);
356 else
357 ret = btrfs_inc_ref(trans, root, cow, 0);
358 BUG_ON(ret);
359 ret = btrfs_dec_ref(trans, root, buf, 1);
360 BUG_ON(ret);
361 }
362 clean_tree_block(trans, root, buf);
363 }
364 return 0;
365 }
366
367 /*
368 * does the dirty work in cow of a single block. The parent block (if
369 * supplied) is updated to point to the new cow copy. The new buffer is marked
370 * dirty and returned locked. If you modify the block it needs to be marked
371 * dirty again.
372 *
373 * search_start -- an allocation hint for the new block
374 *
375 * empty_size -- a hint that you plan on doing more cow. This is the size in
376 * bytes the allocator should try to find free next to the block it returns.
377 * This is just a hint and may be ignored by the allocator.
378 */
379 static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
380 struct btrfs_root *root,
381 struct extent_buffer *buf,
382 struct extent_buffer *parent, int parent_slot,
383 struct extent_buffer **cow_ret,
384 u64 search_start, u64 empty_size)
385 {
386 struct btrfs_disk_key disk_key;
387 struct extent_buffer *cow;
388 int level;
389 int unlock_orig = 0;
390 u64 parent_start;
391
392 if (*cow_ret == buf)
393 unlock_orig = 1;
394
395 btrfs_assert_tree_locked(buf);
396
397 WARN_ON(root->ref_cows && trans->transid !=
398 root->fs_info->running_transaction->transid);
399 WARN_ON(root->ref_cows && trans->transid != root->last_trans);
400
401 level = btrfs_header_level(buf);
402
403 if (level == 0)
404 btrfs_item_key(buf, &disk_key, 0);
405 else
406 btrfs_node_key(buf, &disk_key, 0);
407
408 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) {
409 if (parent)
410 parent_start = parent->start;
411 else
412 parent_start = 0;
413 } else
414 parent_start = 0;
415
416 cow = btrfs_alloc_free_block(trans, root, buf->len, parent_start,
417 root->root_key.objectid, &disk_key,
418 level, search_start, empty_size);
419 if (IS_ERR(cow))
420 return PTR_ERR(cow);
421
422 /* cow is set to blocking by btrfs_init_new_buffer */
423
424 copy_extent_buffer(cow, buf, 0, 0, cow->len);
425 btrfs_set_header_bytenr(cow, cow->start);
426 btrfs_set_header_generation(cow, trans->transid);
427 btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
428 btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
429 BTRFS_HEADER_FLAG_RELOC);
430 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
431 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
432 else
433 btrfs_set_header_owner(cow, root->root_key.objectid);
434
435 write_extent_buffer(cow, root->fs_info->fsid,
436 (unsigned long)btrfs_header_fsid(cow),
437 BTRFS_FSID_SIZE);
438
439 update_ref_for_cow(trans, root, buf, cow);
440
441 if (buf == root->node) {
442 WARN_ON(parent && parent != buf);
443 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
444 btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
445 parent_start = buf->start;
446 else
447 parent_start = 0;
448
449 spin_lock(&root->node_lock);
450 root->node = cow;
451 extent_buffer_get(cow);
452 spin_unlock(&root->node_lock);
453
454 btrfs_free_extent(trans, root, buf->start, buf->len,
455 parent_start, root->root_key.objectid,
456 level, 0);
457 free_extent_buffer(buf);
458 add_root_to_dirty_list(root);
459 } else {
460 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
461 parent_start = parent->start;
462 else
463 parent_start = 0;
464
465 WARN_ON(trans->transid != btrfs_header_generation(parent));
466 btrfs_set_node_blockptr(parent, parent_slot,
467 cow->start);
468 btrfs_set_node_ptr_generation(parent, parent_slot,
469 trans->transid);
470 btrfs_mark_buffer_dirty(parent);
471 btrfs_free_extent(trans, root, buf->start, buf->len,
472 parent_start, root->root_key.objectid,
473 level, 0);
474 }
475 if (unlock_orig)
476 btrfs_tree_unlock(buf);
477 free_extent_buffer(buf);
478 btrfs_mark_buffer_dirty(cow);
479 *cow_ret = cow;
480 return 0;
481 }
482
483 static inline int should_cow_block(struct btrfs_trans_handle *trans,
484 struct btrfs_root *root,
485 struct extent_buffer *buf)
486 {
487 if (btrfs_header_generation(buf) == trans->transid &&
488 !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) &&
489 !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
490 btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)))
491 return 0;
492 return 1;
493 }
494
495 /*
496 * cows a single block, see __btrfs_cow_block for the real work.
497 * This version of it has extra checks so that a block isn't cow'd more than
498 * once per transaction, as long as it hasn't been written yet
499 */
500 noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
501 struct btrfs_root *root, struct extent_buffer *buf,
502 struct extent_buffer *parent, int parent_slot,
503 struct extent_buffer **cow_ret)
504 {
505 u64 search_start;
506 int ret;
507
508 if (trans->transaction != root->fs_info->running_transaction) {
509 printk(KERN_CRIT "trans %llu running %llu\n",
510 (unsigned long long)trans->transid,
511 (unsigned long long)
512 root->fs_info->running_transaction->transid);
513 WARN_ON(1);
514 }
515 if (trans->transid != root->fs_info->generation) {
516 printk(KERN_CRIT "trans %llu running %llu\n",
517 (unsigned long long)trans->transid,
518 (unsigned long long)root->fs_info->generation);
519 WARN_ON(1);
520 }
521
522 if (!should_cow_block(trans, root, buf)) {
523 *cow_ret = buf;
524 return 0;
525 }
526
527 search_start = buf->start & ~((u64)(1024 * 1024 * 1024) - 1);
528
529 if (parent)
530 btrfs_set_lock_blocking(parent);
531 btrfs_set_lock_blocking(buf);
532
533 ret = __btrfs_cow_block(trans, root, buf, parent,
534 parent_slot, cow_ret, search_start, 0);
535 return ret;
536 }
537
538 /*
539 * helper function for defrag to decide if two blocks pointed to by a
540 * node are actually close by
541 */
542 static int close_blocks(u64 blocknr, u64 other, u32 blocksize)
543 {
544 if (blocknr < other && other - (blocknr + blocksize) < 32768)
545 return 1;
546 if (blocknr > other && blocknr - (other + blocksize) < 32768)
547 return 1;
548 return 0;
549 }
550
551 /*
552 * compare two keys in a memcmp fashion
553 */
554 static int comp_keys(struct btrfs_disk_key *disk, struct btrfs_key *k2)
555 {
556 struct btrfs_key k1;
557
558 btrfs_disk_key_to_cpu(&k1, disk);
559
560 if (k1.objectid > k2->objectid)
561 return 1;
562 if (k1.objectid < k2->objectid)
563 return -1;
564 if (k1.type > k2->type)
565 return 1;
566 if (k1.type < k2->type)
567 return -1;
568 if (k1.offset > k2->offset)
569 return 1;
570 if (k1.offset < k2->offset)
571 return -1;
572 return 0;
573 }
574
575 /*
576 * same as comp_keys only with two btrfs_key's
577 */
578 int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2)
579 {
580 if (k1->objectid > k2->objectid)
581 return 1;
582 if (k1->objectid < k2->objectid)
583 return -1;
584 if (k1->type > k2->type)
585 return 1;
586 if (k1->type < k2->type)
587 return -1;
588 if (k1->offset > k2->offset)
589 return 1;
590 if (k1->offset < k2->offset)
591 return -1;
592 return 0;
593 }
594
595 /*
596 * this is used by the defrag code to go through all the
597 * leaves pointed to by a node and reallocate them so that
598 * disk order is close to key order
599 */
600 int btrfs_realloc_node(struct btrfs_trans_handle *trans,
601 struct btrfs_root *root, struct extent_buffer *parent,
602 int start_slot, int cache_only, u64 *last_ret,
603 struct btrfs_key *progress)
604 {
605 struct extent_buffer *cur;
606 u64 blocknr;
607 u64 gen;
608 u64 search_start = *last_ret;
609 u64 last_block = 0;
610 u64 other;
611 u32 parent_nritems;
612 int end_slot;
613 int i;
614 int err = 0;
615 int parent_level;
616 int uptodate;
617 u32 blocksize;
618 int progress_passed = 0;
619 struct btrfs_disk_key disk_key;
620
621 parent_level = btrfs_header_level(parent);
622 if (cache_only && parent_level != 1)
623 return 0;
624
625 if (trans->transaction != root->fs_info->running_transaction)
626 WARN_ON(1);
627 if (trans->transid != root->fs_info->generation)
628 WARN_ON(1);
629
630 parent_nritems = btrfs_header_nritems(parent);
631 blocksize = btrfs_level_size(root, parent_level - 1);
632 end_slot = parent_nritems;
633
634 if (parent_nritems == 1)
635 return 0;
636
637 btrfs_set_lock_blocking(parent);
638
639 for (i = start_slot; i < end_slot; i++) {
640 int close = 1;
641
642 if (!parent->map_token) {
643 map_extent_buffer(parent,
644 btrfs_node_key_ptr_offset(i),
645 sizeof(struct btrfs_key_ptr),
646 &parent->map_token, &parent->kaddr,
647 &parent->map_start, &parent->map_len,
648 KM_USER1);
649 }
650 btrfs_node_key(parent, &disk_key, i);
651 if (!progress_passed && comp_keys(&disk_key, progress) < 0)
652 continue;
653
654 progress_passed = 1;
655 blocknr = btrfs_node_blockptr(parent, i);
656 gen = btrfs_node_ptr_generation(parent, i);
657 if (last_block == 0)
658 last_block = blocknr;
659
660 if (i > 0) {
661 other = btrfs_node_blockptr(parent, i - 1);
662 close = close_blocks(blocknr, other, blocksize);
663 }
664 if (!close && i < end_slot - 2) {
665 other = btrfs_node_blockptr(parent, i + 1);
666 close = close_blocks(blocknr, other, blocksize);
667 }
668 if (close) {
669 last_block = blocknr;
670 continue;
671 }
672 if (parent->map_token) {
673 unmap_extent_buffer(parent, parent->map_token,
674 KM_USER1);
675 parent->map_token = NULL;
676 }
677
678 cur = btrfs_find_tree_block(root, blocknr, blocksize);
679 if (cur)
680 uptodate = btrfs_buffer_uptodate(cur, gen);
681 else
682 uptodate = 0;
683 if (!cur || !uptodate) {
684 if (cache_only) {
685 free_extent_buffer(cur);
686 continue;
687 }
688 if (!cur) {
689 cur = read_tree_block(root, blocknr,
690 blocksize, gen);
691 } else if (!uptodate) {
692 btrfs_read_buffer(cur, gen);
693 }
694 }
695 if (search_start == 0)
696 search_start = last_block;
697
698 btrfs_tree_lock(cur);
699 btrfs_set_lock_blocking(cur);
700 err = __btrfs_cow_block(trans, root, cur, parent, i,
701 &cur, search_start,
702 min(16 * blocksize,
703 (end_slot - i) * blocksize));
704 if (err) {
705 btrfs_tree_unlock(cur);
706 free_extent_buffer(cur);
707 break;
708 }
709 search_start = cur->start;
710 last_block = cur->start;
711 *last_ret = search_start;
712 btrfs_tree_unlock(cur);
713 free_extent_buffer(cur);
714 }
715 if (parent->map_token) {
716 unmap_extent_buffer(parent, parent->map_token,
717 KM_USER1);
718 parent->map_token = NULL;
719 }
720 return err;
721 }
722
723 /*
724 * The leaf data grows from end-to-front in the node.
725 * this returns the address of the start of the last item,
726 * which is the stop of the leaf data stack
727 */
728 static inline unsigned int leaf_data_end(struct btrfs_root *root,
729 struct extent_buffer *leaf)
730 {
731 u32 nr = btrfs_header_nritems(leaf);
732 if (nr == 0)
733 return BTRFS_LEAF_DATA_SIZE(root);
734 return btrfs_item_offset_nr(leaf, nr - 1);
735 }
736
737 /*
738 * extra debugging checks to make sure all the items in a key are
739 * well formed and in the proper order
740 */
741 static int check_node(struct btrfs_root *root, struct btrfs_path *path,
742 int level)
743 {
744 struct extent_buffer *parent = NULL;
745 struct extent_buffer *node = path->nodes[level];
746 struct btrfs_disk_key parent_key;
747 struct btrfs_disk_key node_key;
748 int parent_slot;
749 int slot;
750 struct btrfs_key cpukey;
751 u32 nritems = btrfs_header_nritems(node);
752
753 if (path->nodes[level + 1])
754 parent = path->nodes[level + 1];
755
756 slot = path->slots[level];
757 BUG_ON(nritems == 0);
758 if (parent) {
759 parent_slot = path->slots[level + 1];
760 btrfs_node_key(parent, &parent_key, parent_slot);
761 btrfs_node_key(node, &node_key, 0);
762 BUG_ON(memcmp(&parent_key, &node_key,
763 sizeof(struct btrfs_disk_key)));
764 BUG_ON(btrfs_node_blockptr(parent, parent_slot) !=
765 btrfs_header_bytenr(node));
766 }
767 BUG_ON(nritems > BTRFS_NODEPTRS_PER_BLOCK(root));
768 if (slot != 0) {
769 btrfs_node_key_to_cpu(node, &cpukey, slot - 1);
770 btrfs_node_key(node, &node_key, slot);
771 BUG_ON(comp_keys(&node_key, &cpukey) <= 0);
772 }
773 if (slot < nritems - 1) {
774 btrfs_node_key_to_cpu(node, &cpukey, slot + 1);
775 btrfs_node_key(node, &node_key, slot);
776 BUG_ON(comp_keys(&node_key, &cpukey) >= 0);
777 }
778 return 0;
779 }
780
781 /*
782 * extra checking to make sure all the items in a leaf are
783 * well formed and in the proper order
784 */
785 static int check_leaf(struct btrfs_root *root, struct btrfs_path *path,
786 int level)
787 {
788 struct extent_buffer *leaf = path->nodes[level];
789 struct extent_buffer *parent = NULL;
790 int parent_slot;
791 struct btrfs_key cpukey;
792 struct btrfs_disk_key parent_key;
793 struct btrfs_disk_key leaf_key;
794 int slot = path->slots[0];
795
796 u32 nritems = btrfs_header_nritems(leaf);
797
798 if (path->nodes[level + 1])
799 parent = path->nodes[level + 1];
800
801 if (nritems == 0)
802 return 0;
803
804 if (parent) {
805 parent_slot = path->slots[level + 1];
806 btrfs_node_key(parent, &parent_key, parent_slot);
807 btrfs_item_key(leaf, &leaf_key, 0);
808
809 BUG_ON(memcmp(&parent_key, &leaf_key,
810 sizeof(struct btrfs_disk_key)));
811 BUG_ON(btrfs_node_blockptr(parent, parent_slot) !=
812 btrfs_header_bytenr(leaf));
813 }
814 if (slot != 0 && slot < nritems - 1) {
815 btrfs_item_key(leaf, &leaf_key, slot);
816 btrfs_item_key_to_cpu(leaf, &cpukey, slot - 1);
817 if (comp_keys(&leaf_key, &cpukey) <= 0) {
818 btrfs_print_leaf(root, leaf);
819 printk(KERN_CRIT "slot %d offset bad key\n", slot);
820 BUG_ON(1);
821 }
822 if (btrfs_item_offset_nr(leaf, slot - 1) !=
823 btrfs_item_end_nr(leaf, slot)) {
824 btrfs_print_leaf(root, leaf);
825 printk(KERN_CRIT "slot %d offset bad\n", slot);
826 BUG_ON(1);
827 }
828 }
829 if (slot < nritems - 1) {
830 btrfs_item_key(leaf, &leaf_key, slot);
831 btrfs_item_key_to_cpu(leaf, &cpukey, slot + 1);
832 BUG_ON(comp_keys(&leaf_key, &cpukey) >= 0);
833 if (btrfs_item_offset_nr(leaf, slot) !=
834 btrfs_item_end_nr(leaf, slot + 1)) {
835 btrfs_print_leaf(root, leaf);
836 printk(KERN_CRIT "slot %d offset bad\n", slot);
837 BUG_ON(1);
838 }
839 }
840 BUG_ON(btrfs_item_offset_nr(leaf, 0) +
841 btrfs_item_size_nr(leaf, 0) != BTRFS_LEAF_DATA_SIZE(root));
842 return 0;
843 }
844
845 static noinline int check_block(struct btrfs_root *root,
846 struct btrfs_path *path, int level)
847 {
848 return 0;
849 if (level == 0)
850 return check_leaf(root, path, level);
851 return check_node(root, path, level);
852 }
853
854 /*
855 * search for key in the extent_buffer. The items start at offset p,
856 * and they are item_size apart. There are 'max' items in p.
857 *
858 * the slot in the array is returned via slot, and it points to
859 * the place where you would insert key if it is not found in
860 * the array.
861 *
862 * slot may point to max if the key is bigger than all of the keys
863 */
864 static noinline int generic_bin_search(struct extent_buffer *eb,
865 unsigned long p,
866 int item_size, struct btrfs_key *key,
867 int max, int *slot)
868 {
869 int low = 0;
870 int high = max;
871 int mid;
872 int ret;
873 struct btrfs_disk_key *tmp = NULL;
874 struct btrfs_disk_key unaligned;
875 unsigned long offset;
876 char *map_token = NULL;
877 char *kaddr = NULL;
878 unsigned long map_start = 0;
879 unsigned long map_len = 0;
880 int err;
881
882 while (low < high) {
883 mid = (low + high) / 2;
884 offset = p + mid * item_size;
885
886 if (!map_token || offset < map_start ||
887 (offset + sizeof(struct btrfs_disk_key)) >
888 map_start + map_len) {
889 if (map_token) {
890 unmap_extent_buffer(eb, map_token, KM_USER0);
891 map_token = NULL;
892 }
893
894 err = map_private_extent_buffer(eb, offset,
895 sizeof(struct btrfs_disk_key),
896 &map_token, &kaddr,
897 &map_start, &map_len, KM_USER0);
898
899 if (!err) {
900 tmp = (struct btrfs_disk_key *)(kaddr + offset -
901 map_start);
902 } else {
903 read_extent_buffer(eb, &unaligned,
904 offset, sizeof(unaligned));
905 tmp = &unaligned;
906 }
907
908 } else {
909 tmp = (struct btrfs_disk_key *)(kaddr + offset -
910 map_start);
911 }
912 ret = comp_keys(tmp, key);
913
914 if (ret < 0)
915 low = mid + 1;
916 else if (ret > 0)
917 high = mid;
918 else {
919 *slot = mid;
920 if (map_token)
921 unmap_extent_buffer(eb, map_token, KM_USER0);
922 return 0;
923 }
924 }
925 *slot = low;
926 if (map_token)
927 unmap_extent_buffer(eb, map_token, KM_USER0);
928 return 1;
929 }
930
931 /*
932 * simple bin_search frontend that does the right thing for
933 * leaves vs nodes
934 */
935 static int bin_search(struct extent_buffer *eb, struct btrfs_key *key,
936 int level, int *slot)
937 {
938 if (level == 0) {
939 return generic_bin_search(eb,
940 offsetof(struct btrfs_leaf, items),
941 sizeof(struct btrfs_item),
942 key, btrfs_header_nritems(eb),
943 slot);
944 } else {
945 return generic_bin_search(eb,
946 offsetof(struct btrfs_node, ptrs),
947 sizeof(struct btrfs_key_ptr),
948 key, btrfs_header_nritems(eb),
949 slot);
950 }
951 return -1;
952 }
953
954 int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key,
955 int level, int *slot)
956 {
957 return bin_search(eb, key, level, slot);
958 }
959
960 /* given a node and slot number, this reads the blocks it points to. The
961 * extent buffer is returned with a reference taken (but unlocked).
962 * NULL is returned on error.
963 */
964 static noinline struct extent_buffer *read_node_slot(struct btrfs_root *root,
965 struct extent_buffer *parent, int slot)
966 {
967 int level = btrfs_header_level(parent);
968 if (slot < 0)
969 return NULL;
970 if (slot >= btrfs_header_nritems(parent))
971 return NULL;
972
973 BUG_ON(level == 0);
974
975 return read_tree_block(root, btrfs_node_blockptr(parent, slot),
976 btrfs_level_size(root, level - 1),
977 btrfs_node_ptr_generation(parent, slot));
978 }
979
980 /*
981 * node level balancing, used to make sure nodes are in proper order for
982 * item deletion. We balance from the top down, so we have to make sure
983 * that a deletion won't leave an node completely empty later on.
984 */
985 static noinline int balance_level(struct btrfs_trans_handle *trans,
986 struct btrfs_root *root,
987 struct btrfs_path *path, int level)
988 {
989 struct extent_buffer *right = NULL;
990 struct extent_buffer *mid;
991 struct extent_buffer *left = NULL;
992 struct extent_buffer *parent = NULL;
993 int ret = 0;
994 int wret;
995 int pslot;
996 int orig_slot = path->slots[level];
997 int err_on_enospc = 0;
998 u64 orig_ptr;
999
1000 if (level == 0)
1001 return 0;
1002
1003 mid = path->nodes[level];
1004
1005 WARN_ON(!path->locks[level]);
1006 WARN_ON(btrfs_header_generation(mid) != trans->transid);
1007
1008 orig_ptr = btrfs_node_blockptr(mid, orig_slot);
1009
1010 if (level < BTRFS_MAX_LEVEL - 1)
1011 parent = path->nodes[level + 1];
1012 pslot = path->slots[level + 1];
1013
1014 /*
1015 * deal with the case where there is only one pointer in the root
1016 * by promoting the node below to a root
1017 */
1018 if (!parent) {
1019 struct extent_buffer *child;
1020
1021 if (btrfs_header_nritems(mid) != 1)
1022 return 0;
1023
1024 /* promote the child to a root */
1025 child = read_node_slot(root, mid, 0);
1026 BUG_ON(!child);
1027 btrfs_tree_lock(child);
1028 btrfs_set_lock_blocking(child);
1029 ret = btrfs_cow_block(trans, root, child, mid, 0, &child);
1030 BUG_ON(ret);
1031
1032 spin_lock(&root->node_lock);
1033 root->node = child;
1034 spin_unlock(&root->node_lock);
1035
1036 add_root_to_dirty_list(root);
1037 btrfs_tree_unlock(child);
1038
1039 path->locks[level] = 0;
1040 path->nodes[level] = NULL;
1041 clean_tree_block(trans, root, mid);
1042 btrfs_tree_unlock(mid);
1043 /* once for the path */
1044 free_extent_buffer(mid);
1045 ret = btrfs_free_extent(trans, root, mid->start, mid->len,
1046 0, root->root_key.objectid, level, 1);
1047 /* once for the root ptr */
1048 free_extent_buffer(mid);
1049 return ret;
1050 }
1051 if (btrfs_header_nritems(mid) >
1052 BTRFS_NODEPTRS_PER_BLOCK(root) / 4)
1053 return 0;
1054
1055 if (btrfs_header_nritems(mid) > 2)
1056 return 0;
1057
1058 if (btrfs_header_nritems(mid) < 2)
1059 err_on_enospc = 1;
1060
1061 left = read_node_slot(root, parent, pslot - 1);
1062 if (left) {
1063 btrfs_tree_lock(left);
1064 btrfs_set_lock_blocking(left);
1065 wret = btrfs_cow_block(trans, root, left,
1066 parent, pslot - 1, &left);
1067 if (wret) {
1068 ret = wret;
1069 goto enospc;
1070 }
1071 }
1072 right = read_node_slot(root, parent, pslot + 1);
1073 if (right) {
1074 btrfs_tree_lock(right);
1075 btrfs_set_lock_blocking(right);
1076 wret = btrfs_cow_block(trans, root, right,
1077 parent, pslot + 1, &right);
1078 if (wret) {
1079 ret = wret;
1080 goto enospc;
1081 }
1082 }
1083
1084 /* first, try to make some room in the middle buffer */
1085 if (left) {
1086 orig_slot += btrfs_header_nritems(left);
1087 wret = push_node_left(trans, root, left, mid, 1);
1088 if (wret < 0)
1089 ret = wret;
1090 if (btrfs_header_nritems(mid) < 2)
1091 err_on_enospc = 1;
1092 }
1093
1094 /*
1095 * then try to empty the right most buffer into the middle
1096 */
1097 if (right) {
1098 wret = push_node_left(trans, root, mid, right, 1);
1099 if (wret < 0 && wret != -ENOSPC)
1100 ret = wret;
1101 if (btrfs_header_nritems(right) == 0) {
1102 u64 bytenr = right->start;
1103 u32 blocksize = right->len;
1104
1105 clean_tree_block(trans, root, right);
1106 btrfs_tree_unlock(right);
1107 free_extent_buffer(right);
1108 right = NULL;
1109 wret = del_ptr(trans, root, path, level + 1, pslot +
1110 1);
1111 if (wret)
1112 ret = wret;
1113 wret = btrfs_free_extent(trans, root, bytenr,
1114 blocksize, 0,
1115 root->root_key.objectid,
1116 level, 0);
1117 if (wret)
1118 ret = wret;
1119 } else {
1120 struct btrfs_disk_key right_key;
1121 btrfs_node_key(right, &right_key, 0);
1122 btrfs_set_node_key(parent, &right_key, pslot + 1);
1123 btrfs_mark_buffer_dirty(parent);
1124 }
1125 }
1126 if (btrfs_header_nritems(mid) == 1) {
1127 /*
1128 * we're not allowed to leave a node with one item in the
1129 * tree during a delete. A deletion from lower in the tree
1130 * could try to delete the only pointer in this node.
1131 * So, pull some keys from the left.
1132 * There has to be a left pointer at this point because
1133 * otherwise we would have pulled some pointers from the
1134 * right
1135 */
1136 BUG_ON(!left);
1137 wret = balance_node_right(trans, root, mid, left);
1138 if (wret < 0) {
1139 ret = wret;
1140 goto enospc;
1141 }
1142 if (wret == 1) {
1143 wret = push_node_left(trans, root, left, mid, 1);
1144 if (wret < 0)
1145 ret = wret;
1146 }
1147 BUG_ON(wret == 1);
1148 }
1149 if (btrfs_header_nritems(mid) == 0) {
1150 /* we've managed to empty the middle node, drop it */
1151 u64 bytenr = mid->start;
1152 u32 blocksize = mid->len;
1153
1154 clean_tree_block(trans, root, mid);
1155 btrfs_tree_unlock(mid);
1156 free_extent_buffer(mid);
1157 mid = NULL;
1158 wret = del_ptr(trans, root, path, level + 1, pslot);
1159 if (wret)
1160 ret = wret;
1161 wret = btrfs_free_extent(trans, root, bytenr, blocksize,
1162 0, root->root_key.objectid,
1163 level, 0);
1164 if (wret)
1165 ret = wret;
1166 } else {
1167 /* update the parent key to reflect our changes */
1168 struct btrfs_disk_key mid_key;
1169 btrfs_node_key(mid, &mid_key, 0);
1170 btrfs_set_node_key(parent, &mid_key, pslot);
1171 btrfs_mark_buffer_dirty(parent);
1172 }
1173
1174 /* update the path */
1175 if (left) {
1176 if (btrfs_header_nritems(left) > orig_slot) {
1177 extent_buffer_get(left);
1178 /* left was locked after cow */
1179 path->nodes[level] = left;
1180 path->slots[level + 1] -= 1;
1181 path->slots[level] = orig_slot;
1182 if (mid) {
1183 btrfs_tree_unlock(mid);
1184 free_extent_buffer(mid);
1185 }
1186 } else {
1187 orig_slot -= btrfs_header_nritems(left);
1188 path->slots[level] = orig_slot;
1189 }
1190 }
1191 /* double check we haven't messed things up */
1192 check_block(root, path, level);
1193 if (orig_ptr !=
1194 btrfs_node_blockptr(path->nodes[level], path->slots[level]))
1195 BUG();
1196 enospc:
1197 if (right) {
1198 btrfs_tree_unlock(right);
1199 free_extent_buffer(right);
1200 }
1201 if (left) {
1202 if (path->nodes[level] != left)
1203 btrfs_tree_unlock(left);
1204 free_extent_buffer(left);
1205 }
1206 return ret;
1207 }
1208
1209 /* Node balancing for insertion. Here we only split or push nodes around
1210 * when they are completely full. This is also done top down, so we
1211 * have to be pessimistic.
1212 */
1213 static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
1214 struct btrfs_root *root,
1215 struct btrfs_path *path, int level)
1216 {
1217 struct extent_buffer *right = NULL;
1218 struct extent_buffer *mid;
1219 struct extent_buffer *left = NULL;
1220 struct extent_buffer *parent = NULL;
1221 int ret = 0;
1222 int wret;
1223 int pslot;
1224 int orig_slot = path->slots[level];
1225 u64 orig_ptr;
1226
1227 if (level == 0)
1228 return 1;
1229
1230 mid = path->nodes[level];
1231 WARN_ON(btrfs_header_generation(mid) != trans->transid);
1232 orig_ptr = btrfs_node_blockptr(mid, orig_slot);
1233
1234 if (level < BTRFS_MAX_LEVEL - 1)
1235 parent = path->nodes[level + 1];
1236 pslot = path->slots[level + 1];
1237
1238 if (!parent)
1239 return 1;
1240
1241 left = read_node_slot(root, parent, pslot - 1);
1242
1243 /* first, try to make some room in the middle buffer */
1244 if (left) {
1245 u32 left_nr;
1246
1247 btrfs_tree_lock(left);
1248 btrfs_set_lock_blocking(left);
1249
1250 left_nr = btrfs_header_nritems(left);
1251 if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) {
1252 wret = 1;
1253 } else {
1254 ret = btrfs_cow_block(trans, root, left, parent,
1255 pslot - 1, &left);
1256 if (ret)
1257 wret = 1;
1258 else {
1259 wret = push_node_left(trans, root,
1260 left, mid, 0);
1261 }
1262 }
1263 if (wret < 0)
1264 ret = wret;
1265 if (wret == 0) {
1266 struct btrfs_disk_key disk_key;
1267 orig_slot += left_nr;
1268 btrfs_node_key(mid, &disk_key, 0);
1269 btrfs_set_node_key(parent, &disk_key, pslot);
1270 btrfs_mark_buffer_dirty(parent);
1271 if (btrfs_header_nritems(left) > orig_slot) {
1272 path->nodes[level] = left;
1273 path->slots[level + 1] -= 1;
1274 path->slots[level] = orig_slot;
1275 btrfs_tree_unlock(mid);
1276 free_extent_buffer(mid);
1277 } else {
1278 orig_slot -=
1279 btrfs_header_nritems(left);
1280 path->slots[level] = orig_slot;
1281 btrfs_tree_unlock(left);
1282 free_extent_buffer(left);
1283 }
1284 return 0;
1285 }
1286 btrfs_tree_unlock(left);
1287 free_extent_buffer(left);
1288 }
1289 right = read_node_slot(root, parent, pslot + 1);
1290
1291 /*
1292 * then try to empty the right most buffer into the middle
1293 */
1294 if (right) {
1295 u32 right_nr;
1296
1297 btrfs_tree_lock(right);
1298 btrfs_set_lock_blocking(right);
1299
1300 right_nr = btrfs_header_nritems(right);
1301 if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) {
1302 wret = 1;
1303 } else {
1304 ret = btrfs_cow_block(trans, root, right,
1305 parent, pslot + 1,
1306 &right);
1307 if (ret)
1308 wret = 1;
1309 else {
1310 wret = balance_node_right(trans, root,
1311 right, mid);
1312 }
1313 }
1314 if (wret < 0)
1315 ret = wret;
1316 if (wret == 0) {
1317 struct btrfs_disk_key disk_key;
1318
1319 btrfs_node_key(right, &disk_key, 0);
1320 btrfs_set_node_key(parent, &disk_key, pslot + 1);
1321 btrfs_mark_buffer_dirty(parent);
1322
1323 if (btrfs_header_nritems(mid) <= orig_slot) {
1324 path->nodes[level] = right;
1325 path->slots[level + 1] += 1;
1326 path->slots[level] = orig_slot -
1327 btrfs_header_nritems(mid);
1328 btrfs_tree_unlock(mid);
1329 free_extent_buffer(mid);
1330 } else {
1331 btrfs_tree_unlock(right);
1332 free_extent_buffer(right);
1333 }
1334 return 0;
1335 }
1336 btrfs_tree_unlock(right);
1337 free_extent_buffer(right);
1338 }
1339 return 1;
1340 }
1341
1342 /*
1343 * readahead one full node of leaves, finding things that are close
1344 * to the block in 'slot', and triggering ra on them.
1345 */
1346 static void reada_for_search(struct btrfs_root *root,
1347 struct btrfs_path *path,
1348 int level, int slot, u64 objectid)
1349 {
1350 struct extent_buffer *node;
1351 struct btrfs_disk_key disk_key;
1352 u32 nritems;
1353 u64 search;
1354 u64 target;
1355 u64 nread = 0;
1356 int direction = path->reada;
1357 struct extent_buffer *eb;
1358 u32 nr;
1359 u32 blocksize;
1360 u32 nscan = 0;
1361
1362 if (level != 1)
1363 return;
1364
1365 if (!path->nodes[level])
1366 return;
1367
1368 node = path->nodes[level];
1369
1370 search = btrfs_node_blockptr(node, slot);
1371 blocksize = btrfs_level_size(root, level - 1);
1372 eb = btrfs_find_tree_block(root, search, blocksize);
1373 if (eb) {
1374 free_extent_buffer(eb);
1375 return;
1376 }
1377
1378 target = search;
1379
1380 nritems = btrfs_header_nritems(node);
1381 nr = slot;
1382 while (1) {
1383 if (direction < 0) {
1384 if (nr == 0)
1385 break;
1386 nr--;
1387 } else if (direction > 0) {
1388 nr++;
1389 if (nr >= nritems)
1390 break;
1391 }
1392 if (path->reada < 0 && objectid) {
1393 btrfs_node_key(node, &disk_key, nr);
1394 if (btrfs_disk_key_objectid(&disk_key) != objectid)
1395 break;
1396 }
1397 search = btrfs_node_blockptr(node, nr);
1398 if ((search <= target && target - search <= 65536) ||
1399 (search > target && search - target <= 65536)) {
1400 readahead_tree_block(root, search, blocksize,
1401 btrfs_node_ptr_generation(node, nr));
1402 nread += blocksize;
1403 }
1404 nscan++;
1405 if ((nread > 65536 || nscan > 32))
1406 break;
1407 }
1408 }
1409
1410 /*
1411 * returns -EAGAIN if it had to drop the path, or zero if everything was in
1412 * cache
1413 */
1414 static noinline int reada_for_balance(struct btrfs_root *root,
1415 struct btrfs_path *path, int level)
1416 {
1417 int slot;
1418 int nritems;
1419 struct extent_buffer *parent;
1420 struct extent_buffer *eb;
1421 u64 gen;
1422 u64 block1 = 0;
1423 u64 block2 = 0;
1424 int ret = 0;
1425 int blocksize;
1426
1427 parent = path->nodes[level + 1];
1428 if (!parent)
1429 return 0;
1430
1431 nritems = btrfs_header_nritems(parent);
1432 slot = path->slots[level + 1];
1433 blocksize = btrfs_level_size(root, level);
1434
1435 if (slot > 0) {
1436 block1 = btrfs_node_blockptr(parent, slot - 1);
1437 gen = btrfs_node_ptr_generation(parent, slot - 1);
1438 eb = btrfs_find_tree_block(root, block1, blocksize);
1439 if (eb && btrfs_buffer_uptodate(eb, gen))
1440 block1 = 0;
1441 free_extent_buffer(eb);
1442 }
1443 if (slot + 1 < nritems) {
1444 block2 = btrfs_node_blockptr(parent, slot + 1);
1445 gen = btrfs_node_ptr_generation(parent, slot + 1);
1446 eb = btrfs_find_tree_block(root, block2, blocksize);
1447 if (eb && btrfs_buffer_uptodate(eb, gen))
1448 block2 = 0;
1449 free_extent_buffer(eb);
1450 }
1451 if (block1 || block2) {
1452 ret = -EAGAIN;
1453
1454 /* release the whole path */
1455 btrfs_release_path(root, path);
1456
1457 /* read the blocks */
1458 if (block1)
1459 readahead_tree_block(root, block1, blocksize, 0);
1460 if (block2)
1461 readahead_tree_block(root, block2, blocksize, 0);
1462
1463 if (block1) {
1464 eb = read_tree_block(root, block1, blocksize, 0);
1465 free_extent_buffer(eb);
1466 }
1467 if (block2) {
1468 eb = read_tree_block(root, block2, blocksize, 0);
1469 free_extent_buffer(eb);
1470 }
1471 }
1472 return ret;
1473 }
1474
1475
1476 /*
1477 * when we walk down the tree, it is usually safe to unlock the higher layers
1478 * in the tree. The exceptions are when our path goes through slot 0, because
1479 * operations on the tree might require changing key pointers higher up in the
1480 * tree.
1481 *
1482 * callers might also have set path->keep_locks, which tells this code to keep
1483 * the lock if the path points to the last slot in the block. This is part of
1484 * walking through the tree, and selecting the next slot in the higher block.
1485 *
1486 * lowest_unlock sets the lowest level in the tree we're allowed to unlock. so
1487 * if lowest_unlock is 1, level 0 won't be unlocked
1488 */
1489 static noinline void unlock_up(struct btrfs_path *path, int level,
1490 int lowest_unlock)
1491 {
1492 int i;
1493 int skip_level = level;
1494 int no_skips = 0;
1495 struct extent_buffer *t;
1496
1497 for (i = level; i < BTRFS_MAX_LEVEL; i++) {
1498 if (!path->nodes[i])
1499 break;
1500 if (!path->locks[i])
1501 break;
1502 if (!no_skips && path->slots[i] == 0) {
1503 skip_level = i + 1;
1504 continue;
1505 }
1506 if (!no_skips && path->keep_locks) {
1507 u32 nritems;
1508 t = path->nodes[i];
1509 nritems = btrfs_header_nritems(t);
1510 if (nritems < 1 || path->slots[i] >= nritems - 1) {
1511 skip_level = i + 1;
1512 continue;
1513 }
1514 }
1515 if (skip_level < i && i >= lowest_unlock)
1516 no_skips = 1;
1517
1518 t = path->nodes[i];
1519 if (i >= lowest_unlock && i > skip_level && path->locks[i]) {
1520 btrfs_tree_unlock(t);
1521 path->locks[i] = 0;
1522 }
1523 }
1524 }
1525
1526 /*
1527 * This releases any locks held in the path starting at level and
1528 * going all the way up to the root.
1529 *
1530 * btrfs_search_slot will keep the lock held on higher nodes in a few
1531 * corner cases, such as COW of the block at slot zero in the node. This
1532 * ignores those rules, and it should only be called when there are no
1533 * more updates to be done higher up in the tree.
1534 */
1535 noinline void btrfs_unlock_up_safe(struct btrfs_path *path, int level)
1536 {
1537 int i;
1538
1539 if (path->keep_locks)
1540 return;
1541
1542 for (i = level; i < BTRFS_MAX_LEVEL; i++) {
1543 if (!path->nodes[i])
1544 continue;
1545 if (!path->locks[i])
1546 continue;
1547 btrfs_tree_unlock(path->nodes[i]);
1548 path->locks[i] = 0;
1549 }
1550 }
1551
1552 /*
1553 * helper function for btrfs_search_slot. The goal is to find a block
1554 * in cache without setting the path to blocking. If we find the block
1555 * we return zero and the path is unchanged.
1556 *
1557 * If we can't find the block, we set the path blocking and do some
1558 * reada. -EAGAIN is returned and the search must be repeated.
1559 */
1560 static int
1561 read_block_for_search(struct btrfs_trans_handle *trans,
1562 struct btrfs_root *root, struct btrfs_path *p,
1563 struct extent_buffer **eb_ret, int level, int slot,
1564 struct btrfs_key *key)
1565 {
1566 u64 blocknr;
1567 u64 gen;
1568 u32 blocksize;
1569 struct extent_buffer *b = *eb_ret;
1570 struct extent_buffer *tmp;
1571 int ret;
1572
1573 blocknr = btrfs_node_blockptr(b, slot);
1574 gen = btrfs_node_ptr_generation(b, slot);
1575 blocksize = btrfs_level_size(root, level - 1);
1576
1577 tmp = btrfs_find_tree_block(root, blocknr, blocksize);
1578 if (tmp && btrfs_buffer_uptodate(tmp, gen)) {
1579 /*
1580 * we found an up to date block without sleeping, return
1581 * right away
1582 */
1583 *eb_ret = tmp;
1584 return 0;
1585 }
1586
1587 /*
1588 * reduce lock contention at high levels
1589 * of the btree by dropping locks before
1590 * we read. Don't release the lock on the current
1591 * level because we need to walk this node to figure
1592 * out which blocks to read.
1593 */
1594 btrfs_unlock_up_safe(p, level + 1);
1595 btrfs_set_path_blocking(p);
1596
1597 if (tmp)
1598 free_extent_buffer(tmp);
1599 if (p->reada)
1600 reada_for_search(root, p, level, slot, key->objectid);
1601
1602 btrfs_release_path(NULL, p);
1603
1604 ret = -EAGAIN;
1605 tmp = read_tree_block(root, blocknr, blocksize, gen);
1606 if (tmp) {
1607 /*
1608 * If the read above didn't mark this buffer up to date,
1609 * it will never end up being up to date. Set ret to EIO now
1610 * and give up so that our caller doesn't loop forever
1611 * on our EAGAINs.
1612 */
1613 if (!btrfs_buffer_uptodate(tmp, 0))
1614 ret = -EIO;
1615 free_extent_buffer(tmp);
1616 }
1617 return ret;
1618 }
1619
1620 /*
1621 * helper function for btrfs_search_slot. This does all of the checks
1622 * for node-level blocks and does any balancing required based on
1623 * the ins_len.
1624 *
1625 * If no extra work was required, zero is returned. If we had to
1626 * drop the path, -EAGAIN is returned and btrfs_search_slot must
1627 * start over
1628 */
1629 static int
1630 setup_nodes_for_search(struct btrfs_trans_handle *trans,
1631 struct btrfs_root *root, struct btrfs_path *p,
1632 struct extent_buffer *b, int level, int ins_len)
1633 {
1634 int ret;
1635 if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >=
1636 BTRFS_NODEPTRS_PER_BLOCK(root) - 3) {
1637 int sret;
1638
1639 sret = reada_for_balance(root, p, level);
1640 if (sret)
1641 goto again;
1642
1643 btrfs_set_path_blocking(p);
1644 sret = split_node(trans, root, p, level);
1645 btrfs_clear_path_blocking(p, NULL);
1646
1647 BUG_ON(sret > 0);
1648 if (sret) {
1649 ret = sret;
1650 goto done;
1651 }
1652 b = p->nodes[level];
1653 } else if (ins_len < 0 && btrfs_header_nritems(b) <
1654 BTRFS_NODEPTRS_PER_BLOCK(root) / 2) {
1655 int sret;
1656
1657 sret = reada_for_balance(root, p, level);
1658 if (sret)
1659 goto again;
1660
1661 btrfs_set_path_blocking(p);
1662 sret = balance_level(trans, root, p, level);
1663 btrfs_clear_path_blocking(p, NULL);
1664
1665 if (sret) {
1666 ret = sret;
1667 goto done;
1668 }
1669 b = p->nodes[level];
1670 if (!b) {
1671 btrfs_release_path(NULL, p);
1672 goto again;
1673 }
1674 BUG_ON(btrfs_header_nritems(b) == 1);
1675 }
1676 return 0;
1677
1678 again:
1679 ret = -EAGAIN;
1680 done:
1681 return ret;
1682 }
1683
1684 /*
1685 * look for key in the tree. path is filled in with nodes along the way
1686 * if key is found, we return zero and you can find the item in the leaf
1687 * level of the path (level 0)
1688 *
1689 * If the key isn't found, the path points to the slot where it should
1690 * be inserted, and 1 is returned. If there are other errors during the
1691 * search a negative error number is returned.
1692 *
1693 * if ins_len > 0, nodes and leaves will be split as we walk down the
1694 * tree. if ins_len < 0, nodes will be merged as we walk down the tree (if
1695 * possible)
1696 */
1697 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
1698 *root, struct btrfs_key *key, struct btrfs_path *p, int
1699 ins_len, int cow)
1700 {
1701 struct extent_buffer *b;
1702 int slot;
1703 int ret;
1704 int level;
1705 int lowest_unlock = 1;
1706 u8 lowest_level = 0;
1707
1708 lowest_level = p->lowest_level;
1709 WARN_ON(lowest_level && ins_len > 0);
1710 WARN_ON(p->nodes[0] != NULL);
1711
1712 if (ins_len < 0)
1713 lowest_unlock = 2;
1714
1715 again:
1716 if (p->search_commit_root) {
1717 b = root->commit_root;
1718 extent_buffer_get(b);
1719 if (!p->skip_locking)
1720 btrfs_tree_lock(b);
1721 } else {
1722 if (p->skip_locking)
1723 b = btrfs_root_node(root);
1724 else
1725 b = btrfs_lock_root_node(root);
1726 }
1727
1728 while (b) {
1729 level = btrfs_header_level(b);
1730
1731 /*
1732 * setup the path here so we can release it under lock
1733 * contention with the cow code
1734 */
1735 p->nodes[level] = b;
1736 if (!p->skip_locking)
1737 p->locks[level] = 1;
1738
1739 if (cow) {
1740 int wret;
1741
1742 /*
1743 * if we don't really need to cow this block
1744 * then we don't want to set the path blocking,
1745 * so we test it here
1746 */
1747 if (!should_cow_block(trans, root, b))
1748 goto cow_done;
1749
1750 btrfs_set_path_blocking(p);
1751
1752 wret = btrfs_cow_block(trans, root, b,
1753 p->nodes[level + 1],
1754 p->slots[level + 1], &b);
1755 if (wret) {
1756 free_extent_buffer(b);
1757 ret = wret;
1758 goto done;
1759 }
1760 }
1761 cow_done:
1762 BUG_ON(!cow && ins_len);
1763 if (level != btrfs_header_level(b))
1764 WARN_ON(1);
1765 level = btrfs_header_level(b);
1766
1767 p->nodes[level] = b;
1768 if (!p->skip_locking)
1769 p->locks[level] = 1;
1770
1771 btrfs_clear_path_blocking(p, NULL);
1772
1773 /*
1774 * we have a lock on b and as long as we aren't changing
1775 * the tree, there is no way to for the items in b to change.
1776 * It is safe to drop the lock on our parent before we
1777 * go through the expensive btree search on b.
1778 *
1779 * If cow is true, then we might be changing slot zero,
1780 * which may require changing the parent. So, we can't
1781 * drop the lock until after we know which slot we're
1782 * operating on.
1783 */
1784 if (!cow)
1785 btrfs_unlock_up_safe(p, level + 1);
1786
1787 ret = check_block(root, p, level);
1788 if (ret) {
1789 ret = -1;
1790 goto done;
1791 }
1792
1793 ret = bin_search(b, key, level, &slot);
1794
1795 if (level != 0) {
1796 if (ret && slot > 0)
1797 slot -= 1;
1798 p->slots[level] = slot;
1799 ret = setup_nodes_for_search(trans, root, p, b, level,
1800 ins_len);
1801 if (ret == -EAGAIN)
1802 goto again;
1803 else if (ret)
1804 goto done;
1805 b = p->nodes[level];
1806 slot = p->slots[level];
1807
1808 unlock_up(p, level, lowest_unlock);
1809
1810 /* this is only true while dropping a snapshot */
1811 if (level == lowest_level) {
1812 ret = 0;
1813 goto done;
1814 }
1815
1816 ret = read_block_for_search(trans, root, p,
1817 &b, level, slot, key);
1818 if (ret == -EAGAIN)
1819 goto again;
1820
1821 if (ret == -EIO)
1822 goto done;
1823
1824 if (!p->skip_locking) {
1825 int lret;
1826
1827 btrfs_clear_path_blocking(p, NULL);
1828 lret = btrfs_try_spin_lock(b);
1829
1830 if (!lret) {
1831 btrfs_set_path_blocking(p);
1832 btrfs_tree_lock(b);
1833 btrfs_clear_path_blocking(p, b);
1834 }
1835 }
1836 } else {
1837 p->slots[level] = slot;
1838 if (ins_len > 0 &&
1839 btrfs_leaf_free_space(root, b) < ins_len) {
1840 int sret;
1841
1842 btrfs_set_path_blocking(p);
1843 sret = split_leaf(trans, root, key,
1844 p, ins_len, ret == 0);
1845 btrfs_clear_path_blocking(p, NULL);
1846
1847 BUG_ON(sret > 0);
1848 if (sret) {
1849 ret = sret;
1850 goto done;
1851 }
1852 }
1853 if (!p->search_for_split)
1854 unlock_up(p, level, lowest_unlock);
1855 goto done;
1856 }
1857 }
1858 ret = 1;
1859 done:
1860 /*
1861 * we don't really know what they plan on doing with the path
1862 * from here on, so for now just mark it as blocking
1863 */
1864 if (!p->leave_spinning)
1865 btrfs_set_path_blocking(p);
1866 if (ret < 0)
1867 btrfs_release_path(root, p);
1868 return ret;
1869 }
1870
1871 /*
1872 * adjust the pointers going up the tree, starting at level
1873 * making sure the right key of each node is points to 'key'.
1874 * This is used after shifting pointers to the left, so it stops
1875 * fixing up pointers when a given leaf/node is not in slot 0 of the
1876 * higher levels
1877 *
1878 * If this fails to write a tree block, it returns -1, but continues
1879 * fixing up the blocks in ram so the tree is consistent.
1880 */
1881 static int fixup_low_keys(struct btrfs_trans_handle *trans,
1882 struct btrfs_root *root, struct btrfs_path *path,
1883 struct btrfs_disk_key *key, int level)
1884 {
1885 int i;
1886 int ret = 0;
1887 struct extent_buffer *t;
1888
1889 for (i = level; i < BTRFS_MAX_LEVEL; i++) {
1890 int tslot = path->slots[i];
1891 if (!path->nodes[i])
1892 break;
1893 t = path->nodes[i];
1894 btrfs_set_node_key(t, key, tslot);
1895 btrfs_mark_buffer_dirty(path->nodes[i]);
1896 if (tslot != 0)
1897 break;
1898 }
1899 return ret;
1900 }
1901
1902 /*
1903 * update item key.
1904 *
1905 * This function isn't completely safe. It's the caller's responsibility
1906 * that the new key won't break the order
1907 */
1908 int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
1909 struct btrfs_root *root, struct btrfs_path *path,
1910 struct btrfs_key *new_key)
1911 {
1912 struct btrfs_disk_key disk_key;
1913 struct extent_buffer *eb;
1914 int slot;
1915
1916 eb = path->nodes[0];
1917 slot = path->slots[0];
1918 if (slot > 0) {
1919 btrfs_item_key(eb, &disk_key, slot - 1);
1920 if (comp_keys(&disk_key, new_key) >= 0)
1921 return -1;
1922 }
1923 if (slot < btrfs_header_nritems(eb) - 1) {
1924 btrfs_item_key(eb, &disk_key, slot + 1);
1925 if (comp_keys(&disk_key, new_key) <= 0)
1926 return -1;
1927 }
1928
1929 btrfs_cpu_key_to_disk(&disk_key, new_key);
1930 btrfs_set_item_key(eb, &disk_key, slot);
1931 btrfs_mark_buffer_dirty(eb);
1932 if (slot == 0)
1933 fixup_low_keys(trans, root, path, &disk_key, 1);
1934 return 0;
1935 }
1936
1937 /*
1938 * try to push data from one node into the next node left in the
1939 * tree.
1940 *
1941 * returns 0 if some ptrs were pushed left, < 0 if there was some horrible
1942 * error, and > 0 if there was no room in the left hand block.
1943 */
1944 static int push_node_left(struct btrfs_trans_handle *trans,
1945 struct btrfs_root *root, struct extent_buffer *dst,
1946 struct extent_buffer *src, int empty)
1947 {
1948 int push_items = 0;
1949 int src_nritems;
1950 int dst_nritems;
1951 int ret = 0;
1952
1953 src_nritems = btrfs_header_nritems(src);
1954 dst_nritems = btrfs_header_nritems(dst);
1955 push_items = BTRFS_NODEPTRS_PER_BLOCK(root) - dst_nritems;
1956 WARN_ON(btrfs_header_generation(src) != trans->transid);
1957 WARN_ON(btrfs_header_generation(dst) != trans->transid);
1958
1959 if (!empty && src_nritems <= 8)
1960 return 1;
1961
1962 if (push_items <= 0)
1963 return 1;
1964
1965 if (empty) {
1966 push_items = min(src_nritems, push_items);
1967 if (push_items < src_nritems) {
1968 /* leave at least 8 pointers in the node if
1969 * we aren't going to empty it
1970 */
1971 if (src_nritems - push_items < 8) {
1972 if (push_items <= 8)
1973 return 1;
1974 push_items -= 8;
1975 }
1976 }
1977 } else
1978 push_items = min(src_nritems - 8, push_items);
1979
1980 copy_extent_buffer(dst, src,
1981 btrfs_node_key_ptr_offset(dst_nritems),
1982 btrfs_node_key_ptr_offset(0),
1983 push_items * sizeof(struct btrfs_key_ptr));
1984
1985 if (push_items < src_nritems) {
1986 memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0),
1987 btrfs_node_key_ptr_offset(push_items),
1988 (src_nritems - push_items) *
1989 sizeof(struct btrfs_key_ptr));
1990 }
1991 btrfs_set_header_nritems(src, src_nritems - push_items);
1992 btrfs_set_header_nritems(dst, dst_nritems + push_items);
1993 btrfs_mark_buffer_dirty(src);
1994 btrfs_mark_buffer_dirty(dst);
1995
1996 return ret;
1997 }
1998
1999 /*
2000 * try to push data from one node into the next node right in the
2001 * tree.
2002 *
2003 * returns 0 if some ptrs were pushed, < 0 if there was some horrible
2004 * error, and > 0 if there was no room in the right hand block.
2005 *
2006 * this will only push up to 1/2 the contents of the left node over
2007 */
2008 static int balance_node_right(struct btrfs_trans_handle *trans,
2009 struct btrfs_root *root,
2010 struct extent_buffer *dst,
2011 struct extent_buffer *src)
2012 {
2013 int push_items = 0;
2014 int max_push;
2015 int src_nritems;
2016 int dst_nritems;
2017 int ret = 0;
2018
2019 WARN_ON(btrfs_header_generation(src) != trans->transid);
2020 WARN_ON(btrfs_header_generation(dst) != trans->transid);
2021
2022 src_nritems = btrfs_header_nritems(src);
2023 dst_nritems = btrfs_header_nritems(dst);
2024 push_items = BTRFS_NODEPTRS_PER_BLOCK(root) - dst_nritems;
2025 if (push_items <= 0)
2026 return 1;
2027
2028 if (src_nritems < 4)
2029 return 1;
2030
2031 max_push = src_nritems / 2 + 1;
2032 /* don't try to empty the node */
2033 if (max_push >= src_nritems)
2034 return 1;
2035
2036 if (max_push < push_items)
2037 push_items = max_push;
2038
2039 memmove_extent_buffer(dst, btrfs_node_key_ptr_offset(push_items),
2040 btrfs_node_key_ptr_offset(0),
2041 (dst_nritems) *
2042 sizeof(struct btrfs_key_ptr));
2043
2044 copy_extent_buffer(dst, src,
2045 btrfs_node_key_ptr_offset(0),
2046 btrfs_node_key_ptr_offset(src_nritems - push_items),
2047 push_items * sizeof(struct btrfs_key_ptr));
2048
2049 btrfs_set_header_nritems(src, src_nritems - push_items);
2050 btrfs_set_header_nritems(dst, dst_nritems + push_items);
2051
2052 btrfs_mark_buffer_dirty(src);
2053 btrfs_mark_buffer_dirty(dst);
2054
2055 return ret;
2056 }
2057
2058 /*
2059 * helper function to insert a new root level in the tree.
2060 * A new node is allocated, and a single item is inserted to
2061 * point to the existing root
2062 *
2063 * returns zero on success or < 0 on failure.
2064 */
2065 static noinline int insert_new_root(struct btrfs_trans_handle *trans,
2066 struct btrfs_root *root,
2067 struct btrfs_path *path, int level)
2068 {
2069 u64 lower_gen;
2070 struct extent_buffer *lower;
2071 struct extent_buffer *c;
2072 struct extent_buffer *old;
2073 struct btrfs_disk_key lower_key;
2074
2075 BUG_ON(path->nodes[level]);
2076 BUG_ON(path->nodes[level-1] != root->node);
2077
2078 lower = path->nodes[level-1];
2079 if (level == 1)
2080 btrfs_item_key(lower, &lower_key, 0);
2081 else
2082 btrfs_node_key(lower, &lower_key, 0);
2083
2084 c = btrfs_alloc_free_block(trans, root, root->nodesize, 0,
2085 root->root_key.objectid, &lower_key,
2086 level, root->node->start, 0);
2087 if (IS_ERR(c))
2088 return PTR_ERR(c);
2089
2090 memset_extent_buffer(c, 0, 0, sizeof(struct btrfs_header));
2091 btrfs_set_header_nritems(c, 1);
2092 btrfs_set_header_level(c, level);
2093 btrfs_set_header_bytenr(c, c->start);
2094 btrfs_set_header_generation(c, trans->transid);
2095 btrfs_set_header_backref_rev(c, BTRFS_MIXED_BACKREF_REV);
2096 btrfs_set_header_owner(c, root->root_key.objectid);
2097
2098 write_extent_buffer(c, root->fs_info->fsid,
2099 (unsigned long)btrfs_header_fsid(c),
2100 BTRFS_FSID_SIZE);
2101
2102 write_extent_buffer(c, root->fs_info->chunk_tree_uuid,
2103 (unsigned long)btrfs_header_chunk_tree_uuid(c),
2104 BTRFS_UUID_SIZE);
2105
2106 btrfs_set_node_key(c, &lower_key, 0);
2107 btrfs_set_node_blockptr(c, 0, lower->start);
2108 lower_gen = btrfs_header_generation(lower);
2109 WARN_ON(lower_gen != trans->transid);
2110
2111 btrfs_set_node_ptr_generation(c, 0, lower_gen);
2112
2113 btrfs_mark_buffer_dirty(c);
2114
2115 spin_lock(&root->node_lock);
2116 old = root->node;
2117 root->node = c;
2118 spin_unlock(&root->node_lock);
2119
2120 /* the super has an extra ref to root->node */
2121 free_extent_buffer(old);
2122
2123 add_root_to_dirty_list(root);
2124 extent_buffer_get(c);
2125 path->nodes[level] = c;
2126 path->locks[level] = 1;
2127 path->slots[level] = 0;
2128 return 0;
2129 }
2130
2131 /*
2132 * worker function to insert a single pointer in a node.
2133 * the node should have enough room for the pointer already
2134 *
2135 * slot and level indicate where you want the key to go, and
2136 * blocknr is the block the key points to.
2137 *
2138 * returns zero on success and < 0 on any error
2139 */
2140 static int insert_ptr(struct btrfs_trans_handle *trans, struct btrfs_root
2141 *root, struct btrfs_path *path, struct btrfs_disk_key
2142 *key, u64 bytenr, int slot, int level)
2143 {
2144 struct extent_buffer *lower;
2145 int nritems;
2146
2147 BUG_ON(!path->nodes[level]);
2148 lower = path->nodes[level];
2149 nritems = btrfs_header_nritems(lower);
2150 BUG_ON(slot > nritems);
2151 if (nritems == BTRFS_NODEPTRS_PER_BLOCK(root))
2152 BUG();
2153 if (slot != nritems) {
2154 memmove_extent_buffer(lower,
2155 btrfs_node_key_ptr_offset(slot + 1),
2156 btrfs_node_key_ptr_offset(slot),
2157 (nritems - slot) * sizeof(struct btrfs_key_ptr));
2158 }
2159 btrfs_set_node_key(lower, key, slot);
2160 btrfs_set_node_blockptr(lower, slot, bytenr);
2161 WARN_ON(trans->transid == 0);
2162 btrfs_set_node_ptr_generation(lower, slot, trans->transid);
2163 btrfs_set_header_nritems(lower, nritems + 1);
2164 btrfs_mark_buffer_dirty(lower);
2165 return 0;
2166 }
2167
2168 /*
2169 * split the node at the specified level in path in two.
2170 * The path is corrected to point to the appropriate node after the split
2171 *
2172 * Before splitting this tries to make some room in the node by pushing
2173 * left and right, if either one works, it returns right away.
2174 *
2175 * returns 0 on success and < 0 on failure
2176 */
2177 static noinline int split_node(struct btrfs_trans_handle *trans,
2178 struct btrfs_root *root,
2179 struct btrfs_path *path, int level)
2180 {
2181 struct extent_buffer *c;
2182 struct extent_buffer *split;
2183 struct btrfs_disk_key disk_key;
2184 int mid;
2185 int ret;
2186 int wret;
2187 u32 c_nritems;
2188
2189 c = path->nodes[level];
2190 WARN_ON(btrfs_header_generation(c) != trans->transid);
2191 if (c == root->node) {
2192 /* trying to split the root, lets make a new one */
2193 ret = insert_new_root(trans, root, path, level + 1);
2194 if (ret)
2195 return ret;
2196 } else {
2197 ret = push_nodes_for_insert(trans, root, path, level);
2198 c = path->nodes[level];
2199 if (!ret && btrfs_header_nritems(c) <
2200 BTRFS_NODEPTRS_PER_BLOCK(root) - 3)
2201 return 0;
2202 if (ret < 0)
2203 return ret;
2204 }
2205
2206 c_nritems = btrfs_header_nritems(c);
2207 mid = (c_nritems + 1) / 2;
2208 btrfs_node_key(c, &disk_key, mid);
2209
2210 split = btrfs_alloc_free_block(trans, root, root->nodesize, 0,
2211 root->root_key.objectid,
2212 &disk_key, level, c->start, 0);
2213 if (IS_ERR(split))
2214 return PTR_ERR(split);
2215
2216 memset_extent_buffer(split, 0, 0, sizeof(struct btrfs_header));
2217 btrfs_set_header_level(split, btrfs_header_level(c));
2218 btrfs_set_header_bytenr(split, split->start);
2219 btrfs_set_header_generation(split, trans->transid);
2220 btrfs_set_header_backref_rev(split, BTRFS_MIXED_BACKREF_REV);
2221 btrfs_set_header_owner(split, root->root_key.objectid);
2222 write_extent_buffer(split, root->fs_info->fsid,
2223 (unsigned long)btrfs_header_fsid(split),
2224 BTRFS_FSID_SIZE);
2225 write_extent_buffer(split, root->fs_info->chunk_tree_uuid,
2226 (unsigned long)btrfs_header_chunk_tree_uuid(split),
2227 BTRFS_UUID_SIZE);
2228
2229
2230 copy_extent_buffer(split, c,
2231 btrfs_node_key_ptr_offset(0),
2232 btrfs_node_key_ptr_offset(mid),
2233 (c_nritems - mid) * sizeof(struct btrfs_key_ptr));
2234 btrfs_set_header_nritems(split, c_nritems - mid);
2235 btrfs_set_header_nritems(c, mid);
2236 ret = 0;
2237
2238 btrfs_mark_buffer_dirty(c);
2239 btrfs_mark_buffer_dirty(split);
2240
2241 wret = insert_ptr(trans, root, path, &disk_key, split->start,
2242 path->slots[level + 1] + 1,
2243 level + 1);
2244 if (wret)
2245 ret = wret;
2246
2247 if (path->slots[level] >= mid) {
2248 path->slots[level] -= mid;
2249 btrfs_tree_unlock(c);
2250 free_extent_buffer(c);
2251 path->nodes[level] = split;
2252 path->slots[level + 1] += 1;
2253 } else {
2254 btrfs_tree_unlock(split);
2255 free_extent_buffer(split);
2256 }
2257 return ret;
2258 }
2259
2260 /*
2261 * how many bytes are required to store the items in a leaf. start
2262 * and nr indicate which items in the leaf to check. This totals up the
2263 * space used both by the item structs and the item data
2264 */
2265 static int leaf_space_used(struct extent_buffer *l, int start, int nr)
2266 {
2267 int data_len;
2268 int nritems = btrfs_header_nritems(l);
2269 int end = min(nritems, start + nr) - 1;
2270
2271 if (!nr)
2272 return 0;
2273 data_len = btrfs_item_end_nr(l, start);
2274 data_len = data_len - btrfs_item_offset_nr(l, end);
2275 data_len += sizeof(struct btrfs_item) * nr;
2276 WARN_ON(data_len < 0);
2277 return data_len;
2278 }
2279
2280 /*
2281 * The space between the end of the leaf items and
2282 * the start of the leaf data. IOW, how much room
2283 * the leaf has left for both items and data
2284 */
2285 noinline int btrfs_leaf_free_space(struct btrfs_root *root,
2286 struct extent_buffer *leaf)
2287 {
2288 int nritems = btrfs_header_nritems(leaf);
2289 int ret;
2290 ret = BTRFS_LEAF_DATA_SIZE(root) - leaf_space_used(leaf, 0, nritems);
2291 if (ret < 0) {
2292 printk(KERN_CRIT "leaf free space ret %d, leaf data size %lu, "
2293 "used %d nritems %d\n",
2294 ret, (unsigned long) BTRFS_LEAF_DATA_SIZE(root),
2295 leaf_space_used(leaf, 0, nritems), nritems);
2296 }
2297 return ret;
2298 }
2299
2300 static noinline int __push_leaf_right(struct btrfs_trans_handle *trans,
2301 struct btrfs_root *root,
2302 struct btrfs_path *path,
2303 int data_size, int empty,
2304 struct extent_buffer *right,
2305 int free_space, u32 left_nritems)
2306 {
2307 struct extent_buffer *left = path->nodes[0];
2308 struct extent_buffer *upper = path->nodes[1];
2309 struct btrfs_disk_key disk_key;
2310 int slot;
2311 u32 i;
2312 int push_space = 0;
2313 int push_items = 0;
2314 struct btrfs_item *item;
2315 u32 nr;
2316 u32 right_nritems;
2317 u32 data_end;
2318 u32 this_item_size;
2319
2320 if (empty)
2321 nr = 0;
2322 else
2323 nr = 1;
2324
2325 if (path->slots[0] >= left_nritems)
2326 push_space += data_size;
2327
2328 slot = path->slots[1];
2329 i = left_nritems - 1;
2330 while (i >= nr) {
2331 item = btrfs_item_nr(left, i);
2332
2333 if (!empty && push_items > 0) {
2334 if (path->slots[0] > i)
2335 break;
2336 if (path->slots[0] == i) {
2337 int space = btrfs_leaf_free_space(root, left);
2338 if (space + push_space * 2 > free_space)
2339 break;
2340 }
2341 }
2342
2343 if (path->slots[0] == i)
2344 push_space += data_size;
2345
2346 if (!left->map_token) {
2347 map_extent_buffer(left, (unsigned long)item,
2348 sizeof(struct btrfs_item),
2349 &left->map_token, &left->kaddr,
2350 &left->map_start, &left->map_len,
2351 KM_USER1);
2352 }
2353
2354 this_item_size = btrfs_item_size(left, item);
2355 if (this_item_size + sizeof(*item) + push_space > free_space)
2356 break;
2357
2358 push_items++;
2359 push_space += this_item_size + sizeof(*item);
2360 if (i == 0)
2361 break;
2362 i--;
2363 }
2364 if (left->map_token) {
2365 unmap_extent_buffer(left, left->map_token, KM_USER1);
2366 left->map_token = NULL;
2367 }
2368
2369 if (push_items == 0)
2370 goto out_unlock;
2371
2372 if (!empty && push_items == left_nritems)
2373 WARN_ON(1);
2374
2375 /* push left to right */
2376 right_nritems = btrfs_header_nritems(right);
2377
2378 push_space = btrfs_item_end_nr(left, left_nritems - push_items);
2379 push_space -= leaf_data_end(root, left);
2380
2381 /* make room in the right data area */
2382 data_end = leaf_data_end(root, right);
2383 memmove_extent_buffer(right,
2384 btrfs_leaf_data(right) + data_end - push_space,
2385 btrfs_leaf_data(right) + data_end,
2386 BTRFS_LEAF_DATA_SIZE(root) - data_end);
2387
2388 /* copy from the left data area */
2389 copy_extent_buffer(right, left, btrfs_leaf_data(right) +
2390 BTRFS_LEAF_DATA_SIZE(root) - push_space,
2391 btrfs_leaf_data(left) + leaf_data_end(root, left),
2392 push_space);
2393
2394 memmove_extent_buffer(right, btrfs_item_nr_offset(push_items),
2395 btrfs_item_nr_offset(0),
2396 right_nritems * sizeof(struct btrfs_item));
2397
2398 /* copy the items from left to right */
2399 copy_extent_buffer(right, left, btrfs_item_nr_offset(0),
2400 btrfs_item_nr_offset(left_nritems - push_items),
2401 push_items * sizeof(struct btrfs_item));
2402
2403 /* update the item pointers */
2404 right_nritems += push_items;
2405 btrfs_set_header_nritems(right, right_nritems);
2406 push_space = BTRFS_LEAF_DATA_SIZE(root);
2407 for (i = 0; i < right_nritems; i++) {
2408 item = btrfs_item_nr(right, i);
2409 if (!right->map_token) {
2410 map_extent_buffer(right, (unsigned long)item,
2411 sizeof(struct btrfs_item),
2412 &right->map_token, &right->kaddr,
2413 &right->map_start, &right->map_len,
2414 KM_USER1);
2415 }
2416 push_space -= btrfs_item_size(right, item);
2417 btrfs_set_item_offset(right, item, push_space);
2418 }
2419
2420 if (right->map_token) {
2421 unmap_extent_buffer(right, right->map_token, KM_USER1);
2422 right->map_token = NULL;
2423 }
2424 left_nritems -= push_items;
2425 btrfs_set_header_nritems(left, left_nritems);
2426
2427 if (left_nritems)
2428 btrfs_mark_buffer_dirty(left);
2429 btrfs_mark_buffer_dirty(right);
2430
2431 btrfs_item_key(right, &disk_key, 0);
2432 btrfs_set_node_key(upper, &disk_key, slot + 1);
2433 btrfs_mark_buffer_dirty(upper);
2434
2435 /* then fixup the leaf pointer in the path */
2436 if (path->slots[0] >= left_nritems) {
2437 path->slots[0] -= left_nritems;
2438 if (btrfs_header_nritems(path->nodes[0]) == 0)
2439 clean_tree_block(trans, root, path->nodes[0]);
2440 btrfs_tree_unlock(path->nodes[0]);
2441 free_extent_buffer(path->nodes[0]);
2442 path->nodes[0] = right;
2443 path->slots[1] += 1;
2444 } else {
2445 btrfs_tree_unlock(right);
2446 free_extent_buffer(right);
2447 }
2448 return 0;
2449
2450 out_unlock:
2451 btrfs_tree_unlock(right);
2452 free_extent_buffer(right);
2453 return 1;
2454 }
2455
2456 /*
2457 * push some data in the path leaf to the right, trying to free up at
2458 * least data_size bytes. returns zero if the push worked, nonzero otherwise
2459 *
2460 * returns 1 if the push failed because the other node didn't have enough
2461 * room, 0 if everything worked out and < 0 if there were major errors.
2462 */
2463 static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
2464 *root, struct btrfs_path *path, int data_size,
2465 int empty)
2466 {
2467 struct extent_buffer *left = path->nodes[0];
2468 struct extent_buffer *right;
2469 struct extent_buffer *upper;
2470 int slot;
2471 int free_space;
2472 u32 left_nritems;
2473 int ret;
2474
2475 if (!path->nodes[1])
2476 return 1;
2477
2478 slot = path->slots[1];
2479 upper = path->nodes[1];
2480 if (slot >= btrfs_header_nritems(upper) - 1)
2481 return 1;
2482
2483 btrfs_assert_tree_locked(path->nodes[1]);
2484
2485 right = read_node_slot(root, upper, slot + 1);
2486 btrfs_tree_lock(right);
2487 btrfs_set_lock_blocking(right);
2488
2489 free_space = btrfs_leaf_free_space(root, right);
2490 if (free_space < data_size)
2491 goto out_unlock;
2492
2493 /* cow and double check */
2494 ret = btrfs_cow_block(trans, root, right, upper,
2495 slot + 1, &right);
2496 if (ret)
2497 goto out_unlock;
2498
2499 free_space = btrfs_leaf_free_space(root, right);
2500 if (free_space < data_size)
2501 goto out_unlock;
2502
2503 left_nritems = btrfs_header_nritems(left);
2504 if (left_nritems == 0)
2505 goto out_unlock;
2506
2507 return __push_leaf_right(trans, root, path, data_size, empty,
2508 right, free_space, left_nritems);
2509 out_unlock:
2510 btrfs_tree_unlock(right);
2511 free_extent_buffer(right);
2512 return 1;
2513 }
2514
2515 /*
2516 * push some data in the path leaf to the left, trying to free up at
2517 * least data_size bytes. returns zero if the push worked, nonzero otherwise
2518 */
2519 static noinline int __push_leaf_left(struct btrfs_trans_handle *trans,
2520 struct btrfs_root *root,
2521 struct btrfs_path *path, int data_size,
2522 int empty, struct extent_buffer *left,
2523 int free_space, int right_nritems)
2524 {
2525 struct btrfs_disk_key disk_key;
2526 struct extent_buffer *right = path->nodes[0];
2527 int slot;
2528 int i;
2529 int push_space = 0;
2530 int push_items = 0;
2531 struct btrfs_item *item;
2532 u32 old_left_nritems;
2533 u32 nr;
2534 int ret = 0;
2535 int wret;
2536 u32 this_item_size;
2537 u32 old_left_item_size;
2538
2539 slot = path->slots[1];
2540
2541 if (empty)
2542 nr = right_nritems;
2543 else
2544 nr = right_nritems - 1;
2545
2546 for (i = 0; i < nr; i++) {
2547 item = btrfs_item_nr(right, i);
2548 if (!right->map_token) {
2549 map_extent_buffer(right, (unsigned long)item,
2550 sizeof(struct btrfs_item),
2551 &right->map_token, &right->kaddr,
2552 &right->map_start, &right->map_len,
2553 KM_USER1);
2554 }
2555
2556 if (!empty && push_items > 0) {
2557 if (path->slots[0] < i)
2558 break;
2559 if (path->slots[0] == i) {
2560 int space = btrfs_leaf_free_space(root, right);
2561 if (space + push_space * 2 > free_space)
2562 break;
2563 }
2564 }
2565
2566 if (path->slots[0] == i)
2567 push_space += data_size;
2568
2569 this_item_size = btrfs_item_size(right, item);
2570 if (this_item_size + sizeof(*item) + push_space > free_space)
2571 break;
2572
2573 push_items++;
2574 push_space += this_item_size + sizeof(*item);
2575 }
2576
2577 if (right->map_token) {
2578 unmap_extent_buffer(right, right->map_token, KM_USER1);
2579 right->map_token = NULL;
2580 }
2581
2582 if (push_items == 0) {
2583 ret = 1;
2584 goto out;
2585 }
2586 if (!empty && push_items == btrfs_header_nritems(right))
2587 WARN_ON(1);
2588
2589 /* push data from right to left */
2590 copy_extent_buffer(left, right,
2591 btrfs_item_nr_offset(btrfs_header_nritems(left)),
2592 btrfs_item_nr_offset(0),
2593 push_items * sizeof(struct btrfs_item));
2594
2595 push_space = BTRFS_LEAF_DATA_SIZE(root) -
2596 btrfs_item_offset_nr(right, push_items - 1);
2597
2598 copy_extent_buffer(left, right, btrfs_leaf_data(left) +
2599 leaf_data_end(root, left) - push_space,
2600 btrfs_leaf_data(right) +
2601 btrfs_item_offset_nr(right, push_items - 1),
2602 push_space);
2603 old_left_nritems = btrfs_header_nritems(left);
2604 BUG_ON(old_left_nritems <= 0);
2605
2606 old_left_item_size = btrfs_item_offset_nr(left, old_left_nritems - 1);
2607 for (i = old_left_nritems; i < old_left_nritems + push_items; i++) {
2608 u32 ioff;
2609
2610 item = btrfs_item_nr(left, i);
2611 if (!left->map_token) {
2612 map_extent_buffer(left, (unsigned long)item,
2613 sizeof(struct btrfs_item),
2614 &left->map_token, &left->kaddr,
2615 &left->map_start, &left->map_len,
2616 KM_USER1);
2617 }
2618
2619 ioff = btrfs_item_offset(left, item);
2620 btrfs_set_item_offset(left, item,
2621 ioff - (BTRFS_LEAF_DATA_SIZE(root) - old_left_item_size));
2622 }
2623 btrfs_set_header_nritems(left, old_left_nritems + push_items);
2624 if (left->map_token) {
2625 unmap_extent_buffer(left, left->map_token, KM_USER1);
2626 left->map_token = NULL;
2627 }
2628
2629 /* fixup right node */
2630 if (push_items > right_nritems) {
2631 printk(KERN_CRIT "push items %d nr %u\n", push_items,
2632 right_nritems);
2633 WARN_ON(1);
2634 }
2635
2636 if (push_items < right_nritems) {
2637 push_space = btrfs_item_offset_nr(right, push_items - 1) -
2638 leaf_data_end(root, right);
2639 memmove_extent_buffer(right, btrfs_leaf_data(right) +
2640 BTRFS_LEAF_DATA_SIZE(root) - push_space,
2641 btrfs_leaf_data(right) +
2642 leaf_data_end(root, right), push_space);
2643
2644 memmove_extent_buffer(right, btrfs_item_nr_offset(0),
2645 btrfs_item_nr_offset(push_items),
2646 (btrfs_header_nritems(right) - push_items) *
2647 sizeof(struct btrfs_item));
2648 }
2649 right_nritems -= push_items;
2650 btrfs_set_header_nritems(right, right_nritems);
2651 push_space = BTRFS_LEAF_DATA_SIZE(root);
2652 for (i = 0; i < right_nritems; i++) {
2653 item = btrfs_item_nr(right, i);
2654
2655 if (!right->map_token) {
2656 map_extent_buffer(right, (unsigned long)item,
2657 sizeof(struct btrfs_item),
2658 &right->map_token, &right->kaddr,
2659 &right->map_start, &right->map_len,
2660 KM_USER1);
2661 }
2662
2663 push_space = push_space - btrfs_item_size(right, item);
2664 btrfs_set_item_offset(right, item, push_space);
2665 }
2666 if (right->map_token) {
2667 unmap_extent_buffer(right, right->map_token, KM_USER1);
2668 right->map_token = NULL;
2669 }
2670
2671 btrfs_mark_buffer_dirty(left);
2672 if (right_nritems)
2673 btrfs_mark_buffer_dirty(right);
2674
2675 btrfs_item_key(right, &disk_key, 0);
2676 wret = fixup_low_keys(trans, root, path, &disk_key, 1);
2677 if (wret)
2678 ret = wret;
2679
2680 /* then fixup the leaf pointer in the path */
2681 if (path->slots[0] < push_items) {
2682 path->slots[0] += old_left_nritems;
2683 if (btrfs_header_nritems(path->nodes[0]) == 0)
2684 clean_tree_block(trans, root, path->nodes[0]);
2685 btrfs_tree_unlock(path->nodes[0]);
2686 free_extent_buffer(path->nodes[0]);
2687 path->nodes[0] = left;
2688 path->slots[1] -= 1;
2689 } else {
2690 btrfs_tree_unlock(left);
2691 free_extent_buffer(left);
2692 path->slots[0] -= push_items;
2693 }
2694 BUG_ON(path->slots[0] < 0);
2695 return ret;
2696 out:
2697 btrfs_tree_unlock(left);
2698 free_extent_buffer(left);
2699 return ret;
2700 }
2701
2702 /*
2703 * push some data in the path leaf to the left, trying to free up at
2704 * least data_size bytes. returns zero if the push worked, nonzero otherwise
2705 */
2706 static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
2707 *root, struct btrfs_path *path, int data_size,
2708 int empty)
2709 {
2710 struct extent_buffer *right = path->nodes[0];
2711 struct extent_buffer *left;
2712 int slot;
2713 int free_space;
2714 u32 right_nritems;
2715 int ret = 0;
2716
2717 slot = path->slots[1];
2718 if (slot == 0)
2719 return 1;
2720 if (!path->nodes[1])
2721 return 1;
2722
2723 right_nritems = btrfs_header_nritems(right);
2724 if (right_nritems == 0)
2725 return 1;
2726
2727 btrfs_assert_tree_locked(path->nodes[1]);
2728
2729 left = read_node_slot(root, path->nodes[1], slot - 1);
2730 btrfs_tree_lock(left);
2731 btrfs_set_lock_blocking(left);
2732
2733 free_space = btrfs_leaf_free_space(root, left);
2734 if (free_space < data_size) {
2735 ret = 1;
2736 goto out;
2737 }
2738
2739 /* cow and double check */
2740 ret = btrfs_cow_block(trans, root, left,
2741 path->nodes[1], slot - 1, &left);
2742 if (ret) {
2743 /* we hit -ENOSPC, but it isn't fatal here */
2744 ret = 1;
2745 goto out;
2746 }
2747
2748 free_space = btrfs_leaf_free_space(root, left);
2749 if (free_space < data_size) {
2750 ret = 1;
2751 goto out;
2752 }
2753
2754 return __push_leaf_left(trans, root, path, data_size,
2755 empty, left, free_space, right_nritems);
2756 out:
2757 btrfs_tree_unlock(left);
2758 free_extent_buffer(left);
2759 return ret;
2760 }
2761
2762 /*
2763 * split the path's leaf in two, making sure there is at least data_size
2764 * available for the resulting leaf level of the path.
2765 *
2766 * returns 0 if all went well and < 0 on failure.
2767 */
2768 static noinline int copy_for_split(struct btrfs_trans_handle *trans,
2769 struct btrfs_root *root,
2770 struct btrfs_path *path,
2771 struct extent_buffer *l,
2772 struct extent_buffer *right,
2773 int slot, int mid, int nritems)
2774 {
2775 int data_copy_size;
2776 int rt_data_off;
2777 int i;
2778 int ret = 0;
2779 int wret;
2780 struct btrfs_disk_key disk_key;
2781
2782 nritems = nritems - mid;
2783 btrfs_set_header_nritems(right, nritems);
2784 data_copy_size = btrfs_item_end_nr(l, mid) - leaf_data_end(root, l);
2785
2786 copy_extent_buffer(right, l, btrfs_item_nr_offset(0),
2787 btrfs_item_nr_offset(mid),
2788 nritems * sizeof(struct btrfs_item));
2789
2790 copy_extent_buffer(right, l,
2791 btrfs_leaf_data(right) + BTRFS_LEAF_DATA_SIZE(root) -
2792 data_copy_size, btrfs_leaf_data(l) +
2793 leaf_data_end(root, l), data_copy_size);
2794
2795 rt_data_off = BTRFS_LEAF_DATA_SIZE(root) -
2796 btrfs_item_end_nr(l, mid);
2797
2798 for (i = 0; i < nritems; i++) {
2799 struct btrfs_item *item = btrfs_item_nr(right, i);
2800 u32 ioff;
2801
2802 if (!right->map_token) {
2803 map_extent_buffer(right, (unsigned long)item,
2804 sizeof(struct btrfs_item),
2805 &right->map_token, &right->kaddr,
2806 &right->map_start, &right->map_len,
2807 KM_USER1);
2808 }
2809
2810 ioff = btrfs_item_offset(right, item);
2811 btrfs_set_item_offset(right, item, ioff + rt_data_off);
2812 }
2813
2814 if (right->map_token) {
2815 unmap_extent_buffer(right, right->map_token, KM_USER1);
2816 right->map_token = NULL;
2817 }
2818
2819 btrfs_set_header_nritems(l, mid);
2820 ret = 0;
2821 btrfs_item_key(right, &disk_key, 0);
2822 wret = insert_ptr(trans, root, path, &disk_key, right->start,
2823 path->slots[1] + 1, 1);
2824 if (wret)
2825 ret = wret;
2826
2827 btrfs_mark_buffer_dirty(right);
2828 btrfs_mark_buffer_dirty(l);
2829 BUG_ON(path->slots[0] != slot);
2830
2831 if (mid <= slot) {
2832 btrfs_tree_unlock(path->nodes[0]);
2833 free_extent_buffer(path->nodes[0]);
2834 path->nodes[0] = right;
2835 path->slots[0] -= mid;
2836 path->slots[1] += 1;
2837 } else {
2838 btrfs_tree_unlock(right);
2839 free_extent_buffer(right);
2840 }
2841
2842 BUG_ON(path->slots[0] < 0);
2843
2844 return ret;
2845 }
2846
2847 /*
2848 * split the path's leaf in two, making sure there is at least data_size
2849 * available for the resulting leaf level of the path.
2850 *
2851 * returns 0 if all went well and < 0 on failure.
2852 */
2853 static noinline int split_leaf(struct btrfs_trans_handle *trans,
2854 struct btrfs_root *root,
2855 struct btrfs_key *ins_key,
2856 struct btrfs_path *path, int data_size,
2857 int extend)
2858 {
2859 struct btrfs_disk_key disk_key;
2860 struct extent_buffer *l;
2861 u32 nritems;
2862 int mid;
2863 int slot;
2864 struct extent_buffer *right;
2865 int ret = 0;
2866 int wret;
2867 int split;
2868 int num_doubles = 0;
2869
2870 /* first try to make some room by pushing left and right */
2871 if (data_size && ins_key->type != BTRFS_DIR_ITEM_KEY) {
2872 wret = push_leaf_right(trans, root, path, data_size, 0);
2873 if (wret < 0)
2874 return wret;
2875 if (wret) {
2876 wret = push_leaf_left(trans, root, path, data_size, 0);
2877 if (wret < 0)
2878 return wret;
2879 }
2880 l = path->nodes[0];
2881
2882 /* did the pushes work? */
2883 if (btrfs_leaf_free_space(root, l) >= data_size)
2884 return 0;
2885 }
2886
2887 if (!path->nodes[1]) {
2888 ret = insert_new_root(trans, root, path, 1);
2889 if (ret)
2890 return ret;
2891 }
2892 again:
2893 split = 1;
2894 l = path->nodes[0];
2895 slot = path->slots[0];
2896 nritems = btrfs_header_nritems(l);
2897 mid = (nritems + 1) / 2;
2898
2899 if (mid <= slot) {
2900 if (nritems == 1 ||
2901 leaf_space_used(l, mid, nritems - mid) + data_size >
2902 BTRFS_LEAF_DATA_SIZE(root)) {
2903 if (slot >= nritems) {
2904 split = 0;
2905 } else {
2906 mid = slot;
2907 if (mid != nritems &&
2908 leaf_space_used(l, mid, nritems - mid) +
2909 data_size > BTRFS_LEAF_DATA_SIZE(root)) {
2910 split = 2;
2911 }
2912 }
2913 }
2914 } else {
2915 if (leaf_space_used(l, 0, mid) + data_size >
2916 BTRFS_LEAF_DATA_SIZE(root)) {
2917 if (!extend && data_size && slot == 0) {
2918 split = 0;
2919 } else if ((extend || !data_size) && slot == 0) {
2920 mid = 1;
2921 } else {
2922 mid = slot;
2923 if (mid != nritems &&
2924 leaf_space_used(l, mid, nritems - mid) +
2925 data_size > BTRFS_LEAF_DATA_SIZE(root)) {
2926 split = 2 ;
2927 }
2928 }
2929 }
2930 }
2931
2932 if (split == 0)
2933 btrfs_cpu_key_to_disk(&disk_key, ins_key);
2934 else
2935 btrfs_item_key(l, &disk_key, mid);
2936
2937 right = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
2938 root->root_key.objectid,
2939 &disk_key, 0, l->start, 0);
2940 if (IS_ERR(right)) {
2941 BUG_ON(1);
2942 return PTR_ERR(right);
2943 }
2944
2945 memset_extent_buffer(right, 0, 0, sizeof(struct btrfs_header));
2946 btrfs_set_header_bytenr(right, right->start);
2947 btrfs_set_header_generation(right, trans->transid);
2948 btrfs_set_header_backref_rev(right, BTRFS_MIXED_BACKREF_REV);
2949 btrfs_set_header_owner(right, root->root_key.objectid);
2950 btrfs_set_header_level(right, 0);
2951 write_extent_buffer(right, root->fs_info->fsid,
2952 (unsigned long)btrfs_header_fsid(right),
2953 BTRFS_FSID_SIZE);
2954
2955 write_extent_buffer(right, root->fs_info->chunk_tree_uuid,
2956 (unsigned long)btrfs_header_chunk_tree_uuid(right),
2957 BTRFS_UUID_SIZE);
2958
2959 if (split == 0) {
2960 if (mid <= slot) {
2961 btrfs_set_header_nritems(right, 0);
2962 wret = insert_ptr(trans, root, path,
2963 &disk_key, right->start,
2964 path->slots[1] + 1, 1);
2965 if (wret)
2966 ret = wret;
2967
2968 btrfs_tree_unlock(path->nodes[0]);
2969 free_extent_buffer(path->nodes[0]);
2970 path->nodes[0] = right;
2971 path->slots[0] = 0;
2972 path->slots[1] += 1;
2973 } else {
2974 btrfs_set_header_nritems(right, 0);
2975 wret = insert_ptr(trans, root, path,
2976 &disk_key,
2977 right->start,
2978 path->slots[1], 1);
2979 if (wret)
2980 ret = wret;
2981 btrfs_tree_unlock(path->nodes[0]);
2982 free_extent_buffer(path->nodes[0]);
2983 path->nodes[0] = right;
2984 path->slots[0] = 0;
2985 if (path->slots[1] == 0) {
2986 wret = fixup_low_keys(trans, root,
2987 path, &disk_key, 1);
2988 if (wret)
2989 ret = wret;
2990 }
2991 }
2992 btrfs_mark_buffer_dirty(right);
2993 return ret;
2994 }
2995
2996 ret = copy_for_split(trans, root, path, l, right, slot, mid, nritems);
2997 BUG_ON(ret);
2998
2999 if (split == 2) {
3000 BUG_ON(num_doubles != 0);
3001 num_doubles++;
3002 goto again;
3003 }
3004
3005 return ret;
3006 }
3007
3008 /*
3009 * This function splits a single item into two items,
3010 * giving 'new_key' to the new item and splitting the
3011 * old one at split_offset (from the start of the item).
3012 *
3013 * The path may be released by this operation. After
3014 * the split, the path is pointing to the old item. The
3015 * new item is going to be in the same node as the old one.
3016 *
3017 * Note, the item being split must be smaller enough to live alone on
3018 * a tree block with room for one extra struct btrfs_item
3019 *
3020 * This allows us to split the item in place, keeping a lock on the
3021 * leaf the entire time.
3022 */
3023 int btrfs_split_item(struct btrfs_trans_handle *trans,
3024 struct btrfs_root *root,
3025 struct btrfs_path *path,
3026 struct btrfs_key *new_key,
3027 unsigned long split_offset)
3028 {
3029 u32 item_size;
3030 struct extent_buffer *leaf;
3031 struct btrfs_key orig_key;
3032 struct btrfs_item *item;
3033 struct btrfs_item *new_item;
3034 int ret = 0;
3035 int slot;
3036 u32 nritems;
3037 u32 orig_offset;
3038 struct btrfs_disk_key disk_key;
3039 char *buf;
3040
3041 leaf = path->nodes[0];
3042 btrfs_item_key_to_cpu(leaf, &orig_key, path->slots[0]);
3043 if (btrfs_leaf_free_space(root, leaf) >= sizeof(struct btrfs_item))
3044 goto split;
3045
3046 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
3047 btrfs_release_path(root, path);
3048
3049 path->search_for_split = 1;
3050 path->keep_locks = 1;
3051
3052 ret = btrfs_search_slot(trans, root, &orig_key, path, 0, 1);
3053 path->search_for_split = 0;
3054
3055 /* if our item isn't there or got smaller, return now */
3056 if (ret != 0 || item_size != btrfs_item_size_nr(path->nodes[0],
3057 path->slots[0])) {
3058 path->keep_locks = 0;
3059 return -EAGAIN;
3060 }
3061
3062 btrfs_set_path_blocking(path);
3063 ret = split_leaf(trans, root, &orig_key, path,
3064 sizeof(struct btrfs_item), 1);
3065 path->keep_locks = 0;
3066 BUG_ON(ret);
3067
3068 btrfs_unlock_up_safe(path, 1);
3069 leaf = path->nodes[0];
3070 BUG_ON(btrfs_leaf_free_space(root, leaf) < sizeof(struct btrfs_item));
3071
3072 split:
3073 /*
3074 * make sure any changes to the path from split_leaf leave it
3075 * in a blocking state
3076 */
3077 btrfs_set_path_blocking(path);
3078
3079 item = btrfs_item_nr(leaf, path->slots[0]);
3080 orig_offset = btrfs_item_offset(leaf, item);
3081 item_size = btrfs_item_size(leaf, item);
3082
3083 buf = kmalloc(item_size, GFP_NOFS);
3084 read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf,
3085 path->slots[0]), item_size);
3086 slot = path->slots[0] + 1;
3087 leaf = path->nodes[0];
3088
3089 nritems = btrfs_header_nritems(leaf);
3090
3091 if (slot != nritems) {
3092 /* shift the items */
3093 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1),
3094 btrfs_item_nr_offset(slot),
3095 (nritems - slot) * sizeof(struct btrfs_item));
3096
3097 }
3098
3099 btrfs_cpu_key_to_disk(&disk_key, new_key);
3100 btrfs_set_item_key(leaf, &disk_key, slot);
3101
3102 new_item = btrfs_item_nr(leaf, slot);
3103
3104 btrfs_set_item_offset(leaf, new_item, orig_offset);
3105 btrfs_set_item_size(leaf, new_item, item_size - split_offset);
3106
3107 btrfs_set_item_offset(leaf, item,
3108 orig_offset + item_size - split_offset);
3109 btrfs_set_item_size(leaf, item, split_offset);
3110
3111 btrfs_set_header_nritems(leaf, nritems + 1);
3112
3113 /* write the data for the start of the original item */
3114 write_extent_buffer(leaf, buf,
3115 btrfs_item_ptr_offset(leaf, path->slots[0]),
3116 split_offset);
3117
3118 /* write the data for the new item */
3119 write_extent_buffer(leaf, buf + split_offset,
3120 btrfs_item_ptr_offset(leaf, slot),
3121 item_size - split_offset);
3122 btrfs_mark_buffer_dirty(leaf);
3123
3124 ret = 0;
3125 if (btrfs_leaf_free_space(root, leaf) < 0) {
3126 btrfs_print_leaf(root, leaf);
3127 BUG();
3128 }
3129 kfree(buf);
3130 return ret;
3131 }
3132
3133 /*
3134 * make the item pointed to by the path smaller. new_size indicates
3135 * how small to make it, and from_end tells us if we just chop bytes
3136 * off the end of the item or if we shift the item to chop bytes off
3137 * the front.
3138 */
3139 int btrfs_truncate_item(struct btrfs_trans_handle *trans,
3140 struct btrfs_root *root,
3141 struct btrfs_path *path,
3142 u32 new_size, int from_end)
3143 {
3144 int ret = 0;
3145 int slot;
3146 int slot_orig;
3147 struct extent_buffer *leaf;
3148 struct btrfs_item *item;
3149 u32 nritems;
3150 unsigned int data_end;
3151 unsigned int old_data_start;
3152 unsigned int old_size;
3153 unsigned int size_diff;
3154 int i;
3155
3156 slot_orig = path->slots[0];
3157 leaf = path->nodes[0];
3158 slot = path->slots[0];
3159
3160 old_size = btrfs_item_size_nr(leaf, slot);
3161 if (old_size == new_size)
3162 return 0;
3163
3164 nritems = btrfs_header_nritems(leaf);
3165 data_end = leaf_data_end(root, leaf);
3166
3167 old_data_start = btrfs_item_offset_nr(leaf, slot);
3168
3169 size_diff = old_size - new_size;
3170
3171 BUG_ON(slot < 0);
3172 BUG_ON(slot >= nritems);
3173
3174 /*
3175 * item0..itemN ... dataN.offset..dataN.size .. data0.size
3176 */
3177 /* first correct the data pointers */
3178 for (i = slot; i < nritems; i++) {
3179 u32 ioff;
3180 item = btrfs_item_nr(leaf, i);
3181
3182 if (!leaf->map_token) {
3183 map_extent_buffer(leaf, (unsigned long)item,
3184 sizeof(struct btrfs_item),
3185 &leaf->map_token, &leaf->kaddr,
3186 &leaf->map_start, &leaf->map_len,
3187 KM_USER1);
3188 }
3189
3190 ioff = btrfs_item_offset(leaf, item);
3191 btrfs_set_item_offset(leaf, item, ioff + size_diff);
3192 }
3193
3194 if (leaf->map_token) {
3195 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3196 leaf->map_token = NULL;
3197 }
3198
3199 /* shift the data */
3200 if (from_end) {
3201 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3202 data_end + size_diff, btrfs_leaf_data(leaf) +
3203 data_end, old_data_start + new_size - data_end);
3204 } else {
3205 struct btrfs_disk_key disk_key;
3206 u64 offset;
3207
3208 btrfs_item_key(leaf, &disk_key, slot);
3209
3210 if (btrfs_disk_key_type(&disk_key) == BTRFS_EXTENT_DATA_KEY) {
3211 unsigned long ptr;
3212 struct btrfs_file_extent_item *fi;
3213
3214 fi = btrfs_item_ptr(leaf, slot,
3215 struct btrfs_file_extent_item);
3216 fi = (struct btrfs_file_extent_item *)(
3217 (unsigned long)fi - size_diff);
3218
3219 if (btrfs_file_extent_type(leaf, fi) ==
3220 BTRFS_FILE_EXTENT_INLINE) {
3221 ptr = btrfs_item_ptr_offset(leaf, slot);
3222 memmove_extent_buffer(leaf, ptr,
3223 (unsigned long)fi,
3224 offsetof(struct btrfs_file_extent_item,
3225 disk_bytenr));
3226 }
3227 }
3228
3229 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3230 data_end + size_diff, btrfs_leaf_data(leaf) +
3231 data_end, old_data_start - data_end);
3232
3233 offset = btrfs_disk_key_offset(&disk_key);
3234 btrfs_set_disk_key_offset(&disk_key, offset + size_diff);
3235 btrfs_set_item_key(leaf, &disk_key, slot);
3236 if (slot == 0)
3237 fixup_low_keys(trans, root, path, &disk_key, 1);
3238 }
3239
3240 item = btrfs_item_nr(leaf, slot);
3241 btrfs_set_item_size(leaf, item, new_size);
3242 btrfs_mark_buffer_dirty(leaf);
3243
3244 ret = 0;
3245 if (btrfs_leaf_free_space(root, leaf) < 0) {
3246 btrfs_print_leaf(root, leaf);
3247 BUG();
3248 }
3249 return ret;
3250 }
3251
3252 /*
3253 * make the item pointed to by the path bigger, data_size is the new size.
3254 */
3255 int btrfs_extend_item(struct btrfs_trans_handle *trans,
3256 struct btrfs_root *root, struct btrfs_path *path,
3257 u32 data_size)
3258 {
3259 int ret = 0;
3260 int slot;
3261 int slot_orig;
3262 struct extent_buffer *leaf;
3263 struct btrfs_item *item;
3264 u32 nritems;
3265 unsigned int data_end;
3266 unsigned int old_data;
3267 unsigned int old_size;
3268 int i;
3269
3270 slot_orig = path->slots[0];
3271 leaf = path->nodes[0];
3272
3273 nritems = btrfs_header_nritems(leaf);
3274 data_end = leaf_data_end(root, leaf);
3275
3276 if (btrfs_leaf_free_space(root, leaf) < data_size) {
3277 btrfs_print_leaf(root, leaf);
3278 BUG();
3279 }
3280 slot = path->slots[0];
3281 old_data = btrfs_item_end_nr(leaf, slot);
3282
3283 BUG_ON(slot < 0);
3284 if (slot >= nritems) {
3285 btrfs_print_leaf(root, leaf);
3286 printk(KERN_CRIT "slot %d too large, nritems %d\n",
3287 slot, nritems);
3288 BUG_ON(1);
3289 }
3290
3291 /*
3292 * item0..itemN ... dataN.offset..dataN.size .. data0.size
3293 */
3294 /* first correct the data pointers */
3295 for (i = slot; i < nritems; i++) {
3296 u32 ioff;
3297 item = btrfs_item_nr(leaf, i);
3298
3299 if (!leaf->map_token) {
3300 map_extent_buffer(leaf, (unsigned long)item,
3301 sizeof(struct btrfs_item),
3302 &leaf->map_token, &leaf->kaddr,
3303 &leaf->map_start, &leaf->map_len,
3304 KM_USER1);
3305 }
3306 ioff = btrfs_item_offset(leaf, item);
3307 btrfs_set_item_offset(leaf, item, ioff - data_size);
3308 }
3309
3310 if (leaf->map_token) {
3311 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3312 leaf->map_token = NULL;
3313 }
3314
3315 /* shift the data */
3316 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3317 data_end - data_size, btrfs_leaf_data(leaf) +
3318 data_end, old_data - data_end);
3319
3320 data_end = old_data;
3321 old_size = btrfs_item_size_nr(leaf, slot);
3322 item = btrfs_item_nr(leaf, slot);
3323 btrfs_set_item_size(leaf, item, old_size + data_size);
3324 btrfs_mark_buffer_dirty(leaf);
3325
3326 ret = 0;
3327 if (btrfs_leaf_free_space(root, leaf) < 0) {
3328 btrfs_print_leaf(root, leaf);
3329 BUG();
3330 }
3331 return ret;
3332 }
3333
3334 /*
3335 * Given a key and some data, insert items into the tree.
3336 * This does all the path init required, making room in the tree if needed.
3337 * Returns the number of keys that were inserted.
3338 */
3339 int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
3340 struct btrfs_root *root,
3341 struct btrfs_path *path,
3342 struct btrfs_key *cpu_key, u32 *data_size,
3343 int nr)
3344 {
3345 struct extent_buffer *leaf;
3346 struct btrfs_item *item;
3347 int ret = 0;
3348 int slot;
3349 int i;
3350 u32 nritems;
3351 u32 total_data = 0;
3352 u32 total_size = 0;
3353 unsigned int data_end;
3354 struct btrfs_disk_key disk_key;
3355 struct btrfs_key found_key;
3356
3357 for (i = 0; i < nr; i++) {
3358 if (total_size + data_size[i] + sizeof(struct btrfs_item) >
3359 BTRFS_LEAF_DATA_SIZE(root)) {
3360 break;
3361 nr = i;
3362 }
3363 total_data += data_size[i];
3364 total_size += data_size[i] + sizeof(struct btrfs_item);
3365 }
3366 BUG_ON(nr == 0);
3367
3368 ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
3369 if (ret == 0)
3370 return -EEXIST;
3371 if (ret < 0)
3372 goto out;
3373
3374 leaf = path->nodes[0];
3375
3376 nritems = btrfs_header_nritems(leaf);
3377 data_end = leaf_data_end(root, leaf);
3378
3379 if (btrfs_leaf_free_space(root, leaf) < total_size) {
3380 for (i = nr; i >= 0; i--) {
3381 total_data -= data_size[i];
3382 total_size -= data_size[i] + sizeof(struct btrfs_item);
3383 if (total_size < btrfs_leaf_free_space(root, leaf))
3384 break;
3385 }
3386 nr = i;
3387 }
3388
3389 slot = path->slots[0];
3390 BUG_ON(slot < 0);
3391
3392 if (slot != nritems) {
3393 unsigned int old_data = btrfs_item_end_nr(leaf, slot);
3394
3395 item = btrfs_item_nr(leaf, slot);
3396 btrfs_item_key_to_cpu(leaf, &found_key, slot);
3397
3398 /* figure out how many keys we can insert in here */
3399 total_data = data_size[0];
3400 for (i = 1; i < nr; i++) {
3401 if (btrfs_comp_cpu_keys(&found_key, cpu_key + i) <= 0)
3402 break;
3403 total_data += data_size[i];
3404 }
3405 nr = i;
3406
3407 if (old_data < data_end) {
3408 btrfs_print_leaf(root, leaf);
3409 printk(KERN_CRIT "slot %d old_data %d data_end %d\n",
3410 slot, old_data, data_end);
3411 BUG_ON(1);
3412 }
3413 /*
3414 * item0..itemN ... dataN.offset..dataN.size .. data0.size
3415 */
3416 /* first correct the data pointers */
3417 WARN_ON(leaf->map_token);
3418 for (i = slot; i < nritems; i++) {
3419 u32 ioff;
3420
3421 item = btrfs_item_nr(leaf, i);
3422 if (!leaf->map_token) {
3423 map_extent_buffer(leaf, (unsigned long)item,
3424 sizeof(struct btrfs_item),
3425 &leaf->map_token, &leaf->kaddr,
3426 &leaf->map_start, &leaf->map_len,
3427 KM_USER1);
3428 }
3429
3430 ioff = btrfs_item_offset(leaf, item);
3431 btrfs_set_item_offset(leaf, item, ioff - total_data);
3432 }
3433 if (leaf->map_token) {
3434 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3435 leaf->map_token = NULL;
3436 }
3437
3438 /* shift the items */
3439 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
3440 btrfs_item_nr_offset(slot),
3441 (nritems - slot) * sizeof(struct btrfs_item));
3442
3443 /* shift the data */
3444 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3445 data_end - total_data, btrfs_leaf_data(leaf) +
3446 data_end, old_data - data_end);
3447 data_end = old_data;
3448 } else {
3449 /*
3450 * this sucks but it has to be done, if we are inserting at
3451 * the end of the leaf only insert 1 of the items, since we
3452 * have no way of knowing whats on the next leaf and we'd have
3453 * to drop our current locks to figure it out
3454 */
3455 nr = 1;
3456 }
3457
3458 /* setup the item for the new data */
3459 for (i = 0; i < nr; i++) {
3460 btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
3461 btrfs_set_item_key(leaf, &disk_key, slot + i);
3462 item = btrfs_item_nr(leaf, slot + i);
3463 btrfs_set_item_offset(leaf, item, data_end - data_size[i]);
3464 data_end -= data_size[i];
3465 btrfs_set_item_size(leaf, item, data_size[i]);
3466 }
3467 btrfs_set_header_nritems(leaf, nritems + nr);
3468 btrfs_mark_buffer_dirty(leaf);
3469
3470 ret = 0;
3471 if (slot == 0) {
3472 btrfs_cpu_key_to_disk(&disk_key, cpu_key);
3473 ret = fixup_low_keys(trans, root, path, &disk_key, 1);
3474 }
3475
3476 if (btrfs_leaf_free_space(root, leaf) < 0) {
3477 btrfs_print_leaf(root, leaf);
3478 BUG();
3479 }
3480 out:
3481 if (!ret)
3482 ret = nr;
3483 return ret;
3484 }
3485
3486 /*
3487 * this is a helper for btrfs_insert_empty_items, the main goal here is
3488 * to save stack depth by doing the bulk of the work in a function
3489 * that doesn't call btrfs_search_slot
3490 */
3491 static noinline_for_stack int
3492 setup_items_for_insert(struct btrfs_trans_handle *trans,
3493 struct btrfs_root *root, struct btrfs_path *path,
3494 struct btrfs_key *cpu_key, u32 *data_size,
3495 u32 total_data, u32 total_size, int nr)
3496 {
3497 struct btrfs_item *item;
3498 int i;
3499 u32 nritems;
3500 unsigned int data_end;
3501 struct btrfs_disk_key disk_key;
3502 int ret;
3503 struct extent_buffer *leaf;
3504 int slot;
3505
3506 leaf = path->nodes[0];
3507 slot = path->slots[0];
3508
3509 nritems = btrfs_header_nritems(leaf);
3510 data_end = leaf_data_end(root, leaf);
3511
3512 if (btrfs_leaf_free_space(root, leaf) < total_size) {
3513 btrfs_print_leaf(root, leaf);
3514 printk(KERN_CRIT "not enough freespace need %u have %d\n",
3515 total_size, btrfs_leaf_free_space(root, leaf));
3516 BUG();
3517 }
3518
3519 if (slot != nritems) {
3520 unsigned int old_data = btrfs_item_end_nr(leaf, slot);
3521
3522 if (old_data < data_end) {
3523 btrfs_print_leaf(root, leaf);
3524 printk(KERN_CRIT "slot %d old_data %d data_end %d\n",
3525 slot, old_data, data_end);
3526 BUG_ON(1);
3527 }
3528 /*
3529 * item0..itemN ... dataN.offset..dataN.size .. data0.size
3530 */
3531 /* first correct the data pointers */
3532 WARN_ON(leaf->map_token);
3533 for (i = slot; i < nritems; i++) {
3534 u32 ioff;
3535
3536 item = btrfs_item_nr(leaf, i);
3537 if (!leaf->map_token) {
3538 map_extent_buffer(leaf, (unsigned long)item,
3539 sizeof(struct btrfs_item),
3540 &leaf->map_token, &leaf->kaddr,
3541 &leaf->map_start, &leaf->map_len,
3542 KM_USER1);
3543 }
3544
3545 ioff = btrfs_item_offset(leaf, item);
3546 btrfs_set_item_offset(leaf, item, ioff - total_data);
3547 }
3548 if (leaf->map_token) {
3549 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3550 leaf->map_token = NULL;
3551 }
3552
3553 /* shift the items */
3554 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
3555 btrfs_item_nr_offset(slot),
3556 (nritems - slot) * sizeof(struct btrfs_item));
3557
3558 /* shift the data */
3559 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3560 data_end - total_data, btrfs_leaf_data(leaf) +
3561 data_end, old_data - data_end);
3562 data_end = old_data;
3563 }
3564
3565 /* setup the item for the new data */
3566 for (i = 0; i < nr; i++) {
3567 btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
3568 btrfs_set_item_key(leaf, &disk_key, slot + i);
3569 item = btrfs_item_nr(leaf, slot + i);
3570 btrfs_set_item_offset(leaf, item, data_end - data_size[i]);
3571 data_end -= data_size[i];
3572 btrfs_set_item_size(leaf, item, data_size[i]);
3573 }
3574
3575 btrfs_set_header_nritems(leaf, nritems + nr);
3576
3577 ret = 0;
3578 if (slot == 0) {
3579 struct btrfs_disk_key disk_key;
3580 btrfs_cpu_key_to_disk(&disk_key, cpu_key);
3581 ret = fixup_low_keys(trans, root, path, &disk_key, 1);
3582 }
3583 btrfs_unlock_up_safe(path, 1);
3584 btrfs_mark_buffer_dirty(leaf);
3585
3586 if (btrfs_leaf_free_space(root, leaf) < 0) {
3587 btrfs_print_leaf(root, leaf);
3588 BUG();
3589 }
3590 return ret;
3591 }
3592
3593 /*
3594 * Given a key and some data, insert items into the tree.
3595 * This does all the path init required, making room in the tree if needed.
3596 */
3597 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
3598 struct btrfs_root *root,
3599 struct btrfs_path *path,
3600 struct btrfs_key *cpu_key, u32 *data_size,
3601 int nr)
3602 {
3603 struct extent_buffer *leaf;
3604 int ret = 0;
3605 int slot;
3606 int i;
3607 u32 total_size = 0;
3608 u32 total_data = 0;
3609
3610 for (i = 0; i < nr; i++)
3611 total_data += data_size[i];
3612
3613 total_size = total_data + (nr * sizeof(struct btrfs_item));
3614 ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
3615 if (ret == 0)
3616 return -EEXIST;
3617 if (ret < 0)
3618 goto out;
3619
3620 leaf = path->nodes[0];
3621 slot = path->slots[0];
3622 BUG_ON(slot < 0);
3623
3624 ret = setup_items_for_insert(trans, root, path, cpu_key, data_size,
3625 total_data, total_size, nr);
3626
3627 out:
3628 return ret;
3629 }
3630
3631 /*
3632 * Given a key and some data, insert an item into the tree.
3633 * This does all the path init required, making room in the tree if needed.
3634 */
3635 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root
3636 *root, struct btrfs_key *cpu_key, void *data, u32
3637 data_size)
3638 {
3639 int ret = 0;
3640 struct btrfs_path *path;
3641 struct extent_buffer *leaf;
3642 unsigned long ptr;
3643
3644 path = btrfs_alloc_path();
3645 BUG_ON(!path);
3646 ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size);
3647 if (!ret) {
3648 leaf = path->nodes[0];
3649 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
3650 write_extent_buffer(leaf, data, ptr, data_size);
3651 btrfs_mark_buffer_dirty(leaf);
3652 }
3653 btrfs_free_path(path);
3654 return ret;
3655 }
3656
3657 /*
3658 * delete the pointer from a given node.
3659 *
3660 * the tree should have been previously balanced so the deletion does not
3661 * empty a node.
3662 */
3663 static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
3664 struct btrfs_path *path, int level, int slot)
3665 {
3666 struct extent_buffer *parent = path->nodes[level];
3667 u32 nritems;
3668 int ret = 0;
3669 int wret;
3670
3671 nritems = btrfs_header_nritems(parent);
3672 if (slot != nritems - 1) {
3673 memmove_extent_buffer(parent,
3674 btrfs_node_key_ptr_offset(slot),
3675 btrfs_node_key_ptr_offset(slot + 1),
3676 sizeof(struct btrfs_key_ptr) *
3677 (nritems - slot - 1));
3678 }
3679 nritems--;
3680 btrfs_set_header_nritems(parent, nritems);
3681 if (nritems == 0 && parent == root->node) {
3682 BUG_ON(btrfs_header_level(root->node) != 1);
3683 /* just turn the root into a leaf and break */
3684 btrfs_set_header_level(root->node, 0);
3685 } else if (slot == 0) {
3686 struct btrfs_disk_key disk_key;
3687
3688 btrfs_node_key(parent, &disk_key, 0);
3689 wret = fixup_low_keys(trans, root, path, &disk_key, level + 1);
3690 if (wret)
3691 ret = wret;
3692 }
3693 btrfs_mark_buffer_dirty(parent);
3694 return ret;
3695 }
3696
3697 /*
3698 * a helper function to delete the leaf pointed to by path->slots[1] and
3699 * path->nodes[1].
3700 *
3701 * This deletes the pointer in path->nodes[1] and frees the leaf
3702 * block extent. zero is returned if it all worked out, < 0 otherwise.
3703 *
3704 * The path must have already been setup for deleting the leaf, including
3705 * all the proper balancing. path->nodes[1] must be locked.
3706 */
3707 static noinline int btrfs_del_leaf(struct btrfs_trans_handle *trans,
3708 struct btrfs_root *root,
3709 struct btrfs_path *path,
3710 struct extent_buffer *leaf)
3711 {
3712 int ret;
3713
3714 WARN_ON(btrfs_header_generation(leaf) != trans->transid);
3715 ret = del_ptr(trans, root, path, 1, path->slots[1]);
3716 if (ret)
3717 return ret;
3718
3719 /*
3720 * btrfs_free_extent is expensive, we want to make sure we
3721 * aren't holding any locks when we call it
3722 */
3723 btrfs_unlock_up_safe(path, 0);
3724
3725 ret = btrfs_free_extent(trans, root, leaf->start, leaf->len,
3726 0, root->root_key.objectid, 0, 0);
3727 return ret;
3728 }
3729 /*
3730 * delete the item at the leaf level in path. If that empties
3731 * the leaf, remove it from the tree
3732 */
3733 int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
3734 struct btrfs_path *path, int slot, int nr)
3735 {
3736 struct extent_buffer *leaf;
3737 struct btrfs_item *item;
3738 int last_off;
3739 int dsize = 0;
3740 int ret = 0;
3741 int wret;
3742 int i;
3743 u32 nritems;
3744
3745 leaf = path->nodes[0];
3746 last_off = btrfs_item_offset_nr(leaf, slot + nr - 1);
3747
3748 for (i = 0; i < nr; i++)
3749 dsize += btrfs_item_size_nr(leaf, slot + i);
3750
3751 nritems = btrfs_header_nritems(leaf);
3752
3753 if (slot + nr != nritems) {
3754 int data_end = leaf_data_end(root, leaf);
3755
3756 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3757 data_end + dsize,
3758 btrfs_leaf_data(leaf) + data_end,
3759 last_off - data_end);
3760
3761 for (i = slot + nr; i < nritems; i++) {
3762 u32 ioff;
3763
3764 item = btrfs_item_nr(leaf, i);
3765 if (!leaf->map_token) {
3766 map_extent_buffer(leaf, (unsigned long)item,
3767 sizeof(struct btrfs_item),
3768 &leaf->map_token, &leaf->kaddr,
3769 &leaf->map_start, &leaf->map_len,
3770 KM_USER1);
3771 }
3772 ioff = btrfs_item_offset(leaf, item);
3773 btrfs_set_item_offset(leaf, item, ioff + dsize);
3774 }
3775
3776 if (leaf->map_token) {
3777 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3778 leaf->map_token = NULL;
3779 }
3780
3781 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot),
3782 btrfs_item_nr_offset(slot + nr),
3783 sizeof(struct btrfs_item) *
3784 (nritems - slot - nr));
3785 }
3786 btrfs_set_header_nritems(leaf, nritems - nr);
3787 nritems -= nr;
3788
3789 /* delete the leaf if we've emptied it */
3790 if (nritems == 0) {
3791 if (leaf == root->node) {
3792 btrfs_set_header_level(leaf, 0);
3793 } else {
3794 ret = btrfs_del_leaf(trans, root, path, leaf);
3795 BUG_ON(ret);
3796 }
3797 } else {
3798 int used = leaf_space_used(leaf, 0, nritems);
3799 if (slot == 0) {
3800 struct btrfs_disk_key disk_key;
3801
3802 btrfs_item_key(leaf, &disk_key, 0);
3803 wret = fixup_low_keys(trans, root, path,
3804 &disk_key, 1);
3805 if (wret)
3806 ret = wret;
3807 }
3808
3809 /* delete the leaf if it is mostly empty */
3810 if (used < BTRFS_LEAF_DATA_SIZE(root) / 2) {
3811 /* push_leaf_left fixes the path.
3812 * make sure the path still points to our leaf
3813 * for possible call to del_ptr below
3814 */
3815 slot = path->slots[1];
3816 extent_buffer_get(leaf);
3817
3818 btrfs_set_path_blocking(path);
3819 wret = push_leaf_left(trans, root, path, 1, 1);
3820 if (wret < 0 && wret != -ENOSPC)
3821 ret = wret;
3822
3823 if (path->nodes[0] == leaf &&
3824 btrfs_header_nritems(leaf)) {
3825 wret = push_leaf_right(trans, root, path, 1, 1);
3826 if (wret < 0 && wret != -ENOSPC)
3827 ret = wret;
3828 }
3829
3830 if (btrfs_header_nritems(leaf) == 0) {
3831 path->slots[1] = slot;
3832 ret = btrfs_del_leaf(trans, root, path, leaf);
3833 BUG_ON(ret);
3834 free_extent_buffer(leaf);
3835 } else {
3836 /* if we're still in the path, make sure
3837 * we're dirty. Otherwise, one of the
3838 * push_leaf functions must have already
3839 * dirtied this buffer
3840 */
3841 if (path->nodes[0] == leaf)
3842 btrfs_mark_buffer_dirty(leaf);
3843 free_extent_buffer(leaf);
3844 }
3845 } else {
3846 btrfs_mark_buffer_dirty(leaf);
3847 }
3848 }
3849 return ret;
3850 }
3851
3852 /*
3853 * search the tree again to find a leaf with lesser keys
3854 * returns 0 if it found something or 1 if there are no lesser leaves.
3855 * returns < 0 on io errors.
3856 *
3857 * This may release the path, and so you may lose any locks held at the
3858 * time you call it.
3859 */
3860 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path)
3861 {
3862 struct btrfs_key key;
3863 struct btrfs_disk_key found_key;
3864 int ret;
3865
3866 btrfs_item_key_to_cpu(path->nodes[0], &key, 0);
3867
3868 if (key.offset > 0)
3869 key.offset--;
3870 else if (key.type > 0)
3871 key.type--;
3872 else if (key.objectid > 0)
3873 key.objectid--;
3874 else
3875 return 1;
3876
3877 btrfs_release_path(root, path);
3878 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
3879 if (ret < 0)
3880 return ret;
3881 btrfs_item_key(path->nodes[0], &found_key, 0);
3882 ret = comp_keys(&found_key, &key);
3883 if (ret < 0)
3884 return 0;
3885 return 1;
3886 }
3887
3888 /*
3889 * A helper function to walk down the tree starting at min_key, and looking
3890 * for nodes or leaves that are either in cache or have a minimum
3891 * transaction id. This is used by the btree defrag code, and tree logging
3892 *
3893 * This does not cow, but it does stuff the starting key it finds back
3894 * into min_key, so you can call btrfs_search_slot with cow=1 on the
3895 * key and get a writable path.
3896 *
3897 * This does lock as it descends, and path->keep_locks should be set
3898 * to 1 by the caller.
3899 *
3900 * This honors path->lowest_level to prevent descent past a given level
3901 * of the tree.
3902 *
3903 * min_trans indicates the oldest transaction that you are interested
3904 * in walking through. Any nodes or leaves older than min_trans are
3905 * skipped over (without reading them).
3906 *
3907 * returns zero if something useful was found, < 0 on error and 1 if there
3908 * was nothing in the tree that matched the search criteria.
3909 */
3910 int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
3911 struct btrfs_key *max_key,
3912 struct btrfs_path *path, int cache_only,
3913 u64 min_trans)
3914 {
3915 struct extent_buffer *cur;
3916 struct btrfs_key found_key;
3917 int slot;
3918 int sret;
3919 u32 nritems;
3920 int level;
3921 int ret = 1;
3922
3923 WARN_ON(!path->keep_locks);
3924 again:
3925 cur = btrfs_lock_root_node(root);
3926 level = btrfs_header_level(cur);
3927 WARN_ON(path->nodes[level]);
3928 path->nodes[level] = cur;
3929 path->locks[level] = 1;
3930
3931 if (btrfs_header_generation(cur) < min_trans) {
3932 ret = 1;
3933 goto out;
3934 }
3935 while (1) {
3936 nritems = btrfs_header_nritems(cur);
3937 level = btrfs_header_level(cur);
3938 sret = bin_search(cur, min_key, level, &slot);
3939
3940 /* at the lowest level, we're done, setup the path and exit */
3941 if (level == path->lowest_level) {
3942 if (slot >= nritems)
3943 goto find_next_key;
3944 ret = 0;
3945 path->slots[level] = slot;
3946 btrfs_item_key_to_cpu(cur, &found_key, slot);
3947 goto out;
3948 }
3949 if (sret && slot > 0)
3950 slot--;
3951 /*
3952 * check this node pointer against the cache_only and
3953 * min_trans parameters. If it isn't in cache or is too
3954 * old, skip to the next one.
3955 */
3956 while (slot < nritems) {
3957 u64 blockptr;
3958 u64 gen;
3959 struct extent_buffer *tmp;
3960 struct btrfs_disk_key disk_key;
3961
3962 blockptr = btrfs_node_blockptr(cur, slot);
3963 gen = btrfs_node_ptr_generation(cur, slot);
3964 if (gen < min_trans) {
3965 slot++;
3966 continue;
3967 }
3968 if (!cache_only)
3969 break;
3970
3971 if (max_key) {
3972 btrfs_node_key(cur, &disk_key, slot);
3973 if (comp_keys(&disk_key, max_key) >= 0) {
3974 ret = 1;
3975 goto out;
3976 }
3977 }
3978
3979 tmp = btrfs_find_tree_block(root, blockptr,
3980 btrfs_level_size(root, level - 1));
3981
3982 if (tmp && btrfs_buffer_uptodate(tmp, gen)) {
3983 free_extent_buffer(tmp);
3984 break;
3985 }
3986 if (tmp)
3987 free_extent_buffer(tmp);
3988 slot++;
3989 }
3990 find_next_key:
3991 /*
3992 * we didn't find a candidate key in this node, walk forward
3993 * and find another one
3994 */
3995 if (slot >= nritems) {
3996 path->slots[level] = slot;
3997 btrfs_set_path_blocking(path);
3998 sret = btrfs_find_next_key(root, path, min_key, level,
3999 cache_only, min_trans);
4000 if (sret == 0) {
4001 btrfs_release_path(root, path);
4002 goto again;
4003 } else {
4004 goto out;
4005 }
4006 }
4007 /* save our key for returning back */
4008 btrfs_node_key_to_cpu(cur, &found_key, slot);
4009 path->slots[level] = slot;
4010 if (level == path->lowest_level) {
4011 ret = 0;
4012 unlock_up(path, level, 1);
4013 goto out;
4014 }
4015 btrfs_set_path_blocking(path);
4016 cur = read_node_slot(root, cur, slot);
4017
4018 btrfs_tree_lock(cur);
4019
4020 path->locks[level - 1] = 1;
4021 path->nodes[level - 1] = cur;
4022 unlock_up(path, level, 1);
4023 btrfs_clear_path_blocking(path, NULL);
4024 }
4025 out:
4026 if (ret == 0)
4027 memcpy(min_key, &found_key, sizeof(found_key));
4028 btrfs_set_path_blocking(path);
4029 return ret;
4030 }
4031
4032 /*
4033 * this is similar to btrfs_next_leaf, but does not try to preserve
4034 * and fixup the path. It looks for and returns the next key in the
4035 * tree based on the current path and the cache_only and min_trans
4036 * parameters.
4037 *
4038 * 0 is returned if another key is found, < 0 if there are any errors
4039 * and 1 is returned if there are no higher keys in the tree
4040 *
4041 * path->keep_locks should be set to 1 on the search made before
4042 * calling this function.
4043 */
4044 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
4045 struct btrfs_key *key, int lowest_level,
4046 int cache_only, u64 min_trans)
4047 {
4048 int level = lowest_level;
4049 int slot;
4050 struct extent_buffer *c;
4051
4052 WARN_ON(!path->keep_locks);
4053 while (level < BTRFS_MAX_LEVEL) {
4054 if (!path->nodes[level])
4055 return 1;
4056
4057 slot = path->slots[level] + 1;
4058 c = path->nodes[level];
4059 next:
4060 if (slot >= btrfs_header_nritems(c)) {
4061 level++;
4062 if (level == BTRFS_MAX_LEVEL)
4063 return 1;
4064 continue;
4065 }
4066 if (level == 0)
4067 btrfs_item_key_to_cpu(c, key, slot);
4068 else {
4069 u64 blockptr = btrfs_node_blockptr(c, slot);
4070 u64 gen = btrfs_node_ptr_generation(c, slot);
4071
4072 if (cache_only) {
4073 struct extent_buffer *cur;
4074 cur = btrfs_find_tree_block(root, blockptr,
4075 btrfs_level_size(root, level - 1));
4076 if (!cur || !btrfs_buffer_uptodate(cur, gen)) {
4077 slot++;
4078 if (cur)
4079 free_extent_buffer(cur);
4080 goto next;
4081 }
4082 free_extent_buffer(cur);
4083 }
4084 if (gen < min_trans) {
4085 slot++;
4086 goto next;
4087 }
4088 btrfs_node_key_to_cpu(c, key, slot);
4089 }
4090 return 0;
4091 }
4092 return 1;
4093 }
4094
4095 /*
4096 * search the tree again to find a leaf with greater keys
4097 * returns 0 if it found something or 1 if there are no greater leaves.
4098 * returns < 0 on io errors.
4099 */
4100 int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
4101 {
4102 int slot;
4103 int level;
4104 struct extent_buffer *c;
4105 struct extent_buffer *next;
4106 struct btrfs_key key;
4107 u32 nritems;
4108 int ret;
4109 int old_spinning = path->leave_spinning;
4110 int force_blocking = 0;
4111
4112 nritems = btrfs_header_nritems(path->nodes[0]);
4113 if (nritems == 0)
4114 return 1;
4115
4116 /*
4117 * we take the blocks in an order that upsets lockdep. Using
4118 * blocking mode is the only way around it.
4119 */
4120 #ifdef CONFIG_DEBUG_LOCK_ALLOC
4121 force_blocking = 1;
4122 #endif
4123
4124 btrfs_item_key_to_cpu(path->nodes[0], &key, nritems - 1);
4125 again:
4126 level = 1;
4127 next = NULL;
4128 btrfs_release_path(root, path);
4129
4130 path->keep_locks = 1;
4131
4132 if (!force_blocking)
4133 path->leave_spinning = 1;
4134
4135 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
4136 path->keep_locks = 0;
4137
4138 if (ret < 0)
4139 return ret;
4140
4141 nritems = btrfs_header_nritems(path->nodes[0]);
4142 /*
4143 * by releasing the path above we dropped all our locks. A balance
4144 * could have added more items next to the key that used to be
4145 * at the very end of the block. So, check again here and
4146 * advance the path if there are now more items available.
4147 */
4148 if (nritems > 0 && path->slots[0] < nritems - 1) {
4149 path->slots[0]++;
4150 ret = 0;
4151 goto done;
4152 }
4153
4154 while (level < BTRFS_MAX_LEVEL) {
4155 if (!path->nodes[level]) {
4156 ret = 1;
4157 goto done;
4158 }
4159
4160 slot = path->slots[level] + 1;
4161 c = path->nodes[level];
4162 if (slot >= btrfs_header_nritems(c)) {
4163 level++;
4164 if (level == BTRFS_MAX_LEVEL) {
4165 ret = 1;
4166 goto done;
4167 }
4168 continue;
4169 }
4170
4171 if (next) {
4172 btrfs_tree_unlock(next);
4173 free_extent_buffer(next);
4174 }
4175
4176 next = c;
4177 ret = read_block_for_search(NULL, root, path, &next, level,
4178 slot, &key);
4179 if (ret == -EAGAIN)
4180 goto again;
4181
4182 if (ret < 0) {
4183 btrfs_release_path(root, path);
4184 goto done;
4185 }
4186
4187 if (!path->skip_locking) {
4188 ret = btrfs_try_spin_lock(next);
4189 if (!ret) {
4190 btrfs_set_path_blocking(path);
4191 btrfs_tree_lock(next);
4192 if (!force_blocking)
4193 btrfs_clear_path_blocking(path, next);
4194 }
4195 if (force_blocking)
4196 btrfs_set_lock_blocking(next);
4197 }
4198 break;
4199 }
4200 path->slots[level] = slot;
4201 while (1) {
4202 level--;
4203 c = path->nodes[level];
4204 if (path->locks[level])
4205 btrfs_tree_unlock(c);
4206
4207 free_extent_buffer(c);
4208 path->nodes[level] = next;
4209 path->slots[level] = 0;
4210 if (!path->skip_locking)
4211 path->locks[level] = 1;
4212
4213 if (!level)
4214 break;
4215
4216 ret = read_block_for_search(NULL, root, path, &next, level,
4217 0, &key);
4218 if (ret == -EAGAIN)
4219 goto again;
4220
4221 if (ret < 0) {
4222 btrfs_release_path(root, path);
4223 goto done;
4224 }
4225
4226 if (!path->skip_locking) {
4227 btrfs_assert_tree_locked(path->nodes[level]);
4228 ret = btrfs_try_spin_lock(next);
4229 if (!ret) {
4230 btrfs_set_path_blocking(path);
4231 btrfs_tree_lock(next);
4232 if (!force_blocking)
4233 btrfs_clear_path_blocking(path, next);
4234 }
4235 if (force_blocking)
4236 btrfs_set_lock_blocking(next);
4237 }
4238 }
4239 ret = 0;
4240 done:
4241 unlock_up(path, 0, 1);
4242 path->leave_spinning = old_spinning;
4243 if (!old_spinning)
4244 btrfs_set_path_blocking(path);
4245
4246 return ret;
4247 }
4248
4249 /*
4250 * this uses btrfs_prev_leaf to walk backwards in the tree, and keeps
4251 * searching until it gets past min_objectid or finds an item of 'type'
4252 *
4253 * returns 0 if something is found, 1 if nothing was found and < 0 on error
4254 */
4255 int btrfs_previous_item(struct btrfs_root *root,
4256 struct btrfs_path *path, u64 min_objectid,
4257 int type)
4258 {
4259 struct btrfs_key found_key;
4260 struct extent_buffer *leaf;
4261 u32 nritems;
4262 int ret;
4263
4264 while (1) {
4265 if (path->slots[0] == 0) {
4266 btrfs_set_path_blocking(path);
4267 ret = btrfs_prev_leaf(root, path);
4268 if (ret != 0)
4269 return ret;
4270 } else {
4271 path->slots[0]--;
4272 }
4273 leaf = path->nodes[0];
4274 nritems = btrfs_header_nritems(leaf);
4275 if (nritems == 0)
4276 return 1;
4277 if (path->slots[0] == nritems)
4278 path->slots[0]--;
4279
4280 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4281 if (found_key.type == type)
4282 return 0;
4283 if (found_key.objectid < min_objectid)
4284 break;
4285 if (found_key.objectid == min_objectid &&
4286 found_key.type < type)
4287 break;
4288 }
4289 return 1;
4290 }
Linux 2.6 ibm-acpi/thinkpad-acpi driver git tree