Btrfs: check return value of read_tree_block()
[linux-3.10.git] / fs / btrfs / ctree.c
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 <linux/slab.h>
21 #include "ctree.h"
22 #include "disk-io.h"
23 #include "transaction.h"
24 #include "print-tree.h"
25 #include "locking.h"
26
27 static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root
28                       *root, struct btrfs_path *path, int level);
29 static int split_leaf(struct btrfs_trans_handle *trans, struct btrfs_root
30                       *root, struct btrfs_key *ins_key,
31                       struct btrfs_path *path, int data_size, int extend);
32 static int push_node_left(struct btrfs_trans_handle *trans,
33                           struct btrfs_root *root, struct extent_buffer *dst,
34                           struct extent_buffer *src, int empty);
35 static int balance_node_right(struct btrfs_trans_handle *trans,
36                               struct btrfs_root *root,
37                               struct extent_buffer *dst_buf,
38                               struct extent_buffer *src_buf);
39 static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
40                    struct btrfs_path *path, int level, int slot);
41 static int setup_items_for_insert(struct btrfs_trans_handle *trans,
42                         struct btrfs_root *root, struct btrfs_path *path,
43                         struct btrfs_key *cpu_key, u32 *data_size,
44                         u32 total_data, u32 total_size, int nr);
45
46
47 struct btrfs_path *btrfs_alloc_path(void)
48 {
49         struct btrfs_path *path;
50         path = kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS);
51         if (path)
52                 path->reada = 1;
53         return path;
54 }
55
56 /*
57  * set all locked nodes in the path to blocking locks.  This should
58  * be done before scheduling
59  */
60 noinline void btrfs_set_path_blocking(struct btrfs_path *p)
61 {
62         int i;
63         for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
64                 if (p->nodes[i] && p->locks[i])
65                         btrfs_set_lock_blocking(p->nodes[i]);
66         }
67 }
68
69 /*
70  * reset all the locked nodes in the patch to spinning locks.
71  *
72  * held is used to keep lockdep happy, when lockdep is enabled
73  * we set held to a blocking lock before we go around and
74  * retake all the spinlocks in the path.  You can safely use NULL
75  * for held
76  */
77 noinline void btrfs_clear_path_blocking(struct btrfs_path *p,
78                                         struct extent_buffer *held)
79 {
80         int i;
81
82 #ifdef CONFIG_DEBUG_LOCK_ALLOC
83         /* lockdep really cares that we take all of these spinlocks
84          * in the right order.  If any of the locks in the path are not
85          * currently blocking, it is going to complain.  So, make really
86          * really sure by forcing the path to blocking before we clear
87          * the path blocking.
88          */
89         if (held)
90                 btrfs_set_lock_blocking(held);
91         btrfs_set_path_blocking(p);
92 #endif
93
94         for (i = BTRFS_MAX_LEVEL - 1; i >= 0; i--) {
95                 if (p->nodes[i] && p->locks[i])
96                         btrfs_clear_lock_blocking(p->nodes[i]);
97         }
98
99 #ifdef CONFIG_DEBUG_LOCK_ALLOC
100         if (held)
101                 btrfs_clear_lock_blocking(held);
102 #endif
103 }
104
105 /* this also releases the path */
106 void btrfs_free_path(struct btrfs_path *p)
107 {
108         if (!p)
109                 return;
110         btrfs_release_path(NULL, p);
111         kmem_cache_free(btrfs_path_cachep, p);
112 }
113
114 /*
115  * path release drops references on the extent buffers in the path
116  * and it drops any locks held by this path
117  *
118  * It is safe to call this on paths that no locks or extent buffers held.
119  */
120 noinline void btrfs_release_path(struct btrfs_root *root, struct btrfs_path *p)
121 {
122         int i;
123
124         for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
125                 p->slots[i] = 0;
126                 if (!p->nodes[i])
127                         continue;
128                 if (p->locks[i]) {
129                         btrfs_tree_unlock(p->nodes[i]);
130                         p->locks[i] = 0;
131                 }
132                 free_extent_buffer(p->nodes[i]);
133                 p->nodes[i] = NULL;
134         }
135 }
136
137 /*
138  * safely gets a reference on the root node of a tree.  A lock
139  * is not taken, so a concurrent writer may put a different node
140  * at the root of the tree.  See btrfs_lock_root_node for the
141  * looping required.
142  *
143  * The extent buffer returned by this has a reference taken, so
144  * it won't disappear.  It may stop being the root of the tree
145  * at any time because there are no locks held.
146  */
147 struct extent_buffer *btrfs_root_node(struct btrfs_root *root)
148 {
149         struct extent_buffer *eb;
150
151         rcu_read_lock();
152         eb = rcu_dereference(root->node);
153         extent_buffer_get(eb);
154         rcu_read_unlock();
155         return eb;
156 }
157
158 /* loop around taking references on and locking the root node of the
159  * tree until you end up with a lock on the root.  A locked buffer
160  * is returned, with a reference held.
161  */
162 struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root)
163 {
164         struct extent_buffer *eb;
165
166         while (1) {
167                 eb = btrfs_root_node(root);
168                 btrfs_tree_lock(eb);
169                 if (eb == root->node)
170                         break;
171                 btrfs_tree_unlock(eb);
172                 free_extent_buffer(eb);
173         }
174         return eb;
175 }
176
177 /* cowonly root (everything not a reference counted cow subvolume), just get
178  * put onto a simple dirty list.  transaction.c walks this to make sure they
179  * get properly updated on disk.
180  */
181 static void add_root_to_dirty_list(struct btrfs_root *root)
182 {
183         if (root->track_dirty && list_empty(&root->dirty_list)) {
184                 list_add(&root->dirty_list,
185                          &root->fs_info->dirty_cowonly_roots);
186         }
187 }
188
189 /*
190  * used by snapshot creation to make a copy of a root for a tree with
191  * a given objectid.  The buffer with the new root node is returned in
192  * cow_ret, and this func returns zero on success or a negative error code.
193  */
194 int btrfs_copy_root(struct btrfs_trans_handle *trans,
195                       struct btrfs_root *root,
196                       struct extent_buffer *buf,
197                       struct extent_buffer **cow_ret, u64 new_root_objectid)
198 {
199         struct extent_buffer *cow;
200         int ret = 0;
201         int level;
202         struct btrfs_disk_key disk_key;
203
204         WARN_ON(root->ref_cows && trans->transid !=
205                 root->fs_info->running_transaction->transid);
206         WARN_ON(root->ref_cows && trans->transid != root->last_trans);
207
208         level = btrfs_header_level(buf);
209         if (level == 0)
210                 btrfs_item_key(buf, &disk_key, 0);
211         else
212                 btrfs_node_key(buf, &disk_key, 0);
213
214         cow = btrfs_alloc_free_block(trans, root, buf->len, 0,
215                                      new_root_objectid, &disk_key, level,
216                                      buf->start, 0);
217         if (IS_ERR(cow))
218                 return PTR_ERR(cow);
219
220         copy_extent_buffer(cow, buf, 0, 0, cow->len);
221         btrfs_set_header_bytenr(cow, cow->start);
222         btrfs_set_header_generation(cow, trans->transid);
223         btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
224         btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
225                                      BTRFS_HEADER_FLAG_RELOC);
226         if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
227                 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
228         else
229                 btrfs_set_header_owner(cow, new_root_objectid);
230
231         write_extent_buffer(cow, root->fs_info->fsid,
232                             (unsigned long)btrfs_header_fsid(cow),
233                             BTRFS_FSID_SIZE);
234
235         WARN_ON(btrfs_header_generation(buf) > trans->transid);
236         if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
237                 ret = btrfs_inc_ref(trans, root, cow, 1);
238         else
239                 ret = btrfs_inc_ref(trans, root, cow, 0);
240
241         if (ret)
242                 return ret;
243
244         btrfs_mark_buffer_dirty(cow);
245         *cow_ret = cow;
246         return 0;
247 }
248
249 /*
250  * check if the tree block can be shared by multiple trees
251  */
252 int btrfs_block_can_be_shared(struct btrfs_root *root,
253                               struct extent_buffer *buf)
254 {
255         /*
256          * Tree blocks not in refernece counted trees and tree roots
257          * are never shared. If a block was allocated after the last
258          * snapshot and the block was not allocated by tree relocation,
259          * we know the block is not shared.
260          */
261         if (root->ref_cows &&
262             buf != root->node && buf != root->commit_root &&
263             (btrfs_header_generation(buf) <=
264              btrfs_root_last_snapshot(&root->root_item) ||
265              btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)))
266                 return 1;
267 #ifdef BTRFS_COMPAT_EXTENT_TREE_V0
268         if (root->ref_cows &&
269             btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
270                 return 1;
271 #endif
272         return 0;
273 }
274
275 static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
276                                        struct btrfs_root *root,
277                                        struct extent_buffer *buf,
278                                        struct extent_buffer *cow,
279                                        int *last_ref)
280 {
281         u64 refs;
282         u64 owner;
283         u64 flags;
284         u64 new_flags = 0;
285         int ret;
286
287         /*
288          * Backrefs update rules:
289          *
290          * Always use full backrefs for extent pointers in tree block
291          * allocated by tree relocation.
292          *
293          * If a shared tree block is no longer referenced by its owner
294          * tree (btrfs_header_owner(buf) == root->root_key.objectid),
295          * use full backrefs for extent pointers in tree block.
296          *
297          * If a tree block is been relocating
298          * (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID),
299          * use full backrefs for extent pointers in tree block.
300          * The reason for this is some operations (such as drop tree)
301          * are only allowed for blocks use full backrefs.
302          */
303
304         if (btrfs_block_can_be_shared(root, buf)) {
305                 ret = btrfs_lookup_extent_info(trans, root, buf->start,
306                                                buf->len, &refs, &flags);
307                 BUG_ON(ret);
308                 BUG_ON(refs == 0);
309         } else {
310                 refs = 1;
311                 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
312                     btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
313                         flags = BTRFS_BLOCK_FLAG_FULL_BACKREF;
314                 else
315                         flags = 0;
316         }
317
318         owner = btrfs_header_owner(buf);
319         BUG_ON(owner == BTRFS_TREE_RELOC_OBJECTID &&
320                !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
321
322         if (refs > 1) {
323                 if ((owner == root->root_key.objectid ||
324                      root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) &&
325                     !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) {
326                         ret = btrfs_inc_ref(trans, root, buf, 1);
327                         BUG_ON(ret);
328
329                         if (root->root_key.objectid ==
330                             BTRFS_TREE_RELOC_OBJECTID) {
331                                 ret = btrfs_dec_ref(trans, root, buf, 0);
332                                 BUG_ON(ret);
333                                 ret = btrfs_inc_ref(trans, root, cow, 1);
334                                 BUG_ON(ret);
335                         }
336                         new_flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
337                 } else {
338
339                         if (root->root_key.objectid ==
340                             BTRFS_TREE_RELOC_OBJECTID)
341                                 ret = btrfs_inc_ref(trans, root, cow, 1);
342                         else
343                                 ret = btrfs_inc_ref(trans, root, cow, 0);
344                         BUG_ON(ret);
345                 }
346                 if (new_flags != 0) {
347                         ret = btrfs_set_disk_extent_flags(trans, root,
348                                                           buf->start,
349                                                           buf->len,
350                                                           new_flags, 0);
351                         BUG_ON(ret);
352                 }
353         } else {
354                 if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
355                         if (root->root_key.objectid ==
356                             BTRFS_TREE_RELOC_OBJECTID)
357                                 ret = btrfs_inc_ref(trans, root, cow, 1);
358                         else
359                                 ret = btrfs_inc_ref(trans, root, cow, 0);
360                         BUG_ON(ret);
361                         ret = btrfs_dec_ref(trans, root, buf, 1);
362                         BUG_ON(ret);
363                 }
364                 clean_tree_block(trans, root, buf);
365                 *last_ref = 1;
366         }
367         return 0;
368 }
369
370 /*
371  * does the dirty work in cow of a single block.  The parent block (if
372  * supplied) is updated to point to the new cow copy.  The new buffer is marked
373  * dirty and returned locked.  If you modify the block it needs to be marked
374  * dirty again.
375  *
376  * search_start -- an allocation hint for the new block
377  *
378  * empty_size -- a hint that you plan on doing more cow.  This is the size in
379  * bytes the allocator should try to find free next to the block it returns.
380  * This is just a hint and may be ignored by the allocator.
381  */
382 static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
383                              struct btrfs_root *root,
384                              struct extent_buffer *buf,
385                              struct extent_buffer *parent, int parent_slot,
386                              struct extent_buffer **cow_ret,
387                              u64 search_start, u64 empty_size)
388 {
389         struct btrfs_disk_key disk_key;
390         struct extent_buffer *cow;
391         int level;
392         int last_ref = 0;
393         int unlock_orig = 0;
394         u64 parent_start;
395
396         if (*cow_ret == buf)
397                 unlock_orig = 1;
398
399         btrfs_assert_tree_locked(buf);
400
401         WARN_ON(root->ref_cows && trans->transid !=
402                 root->fs_info->running_transaction->transid);
403         WARN_ON(root->ref_cows && trans->transid != root->last_trans);
404
405         level = btrfs_header_level(buf);
406
407         if (level == 0)
408                 btrfs_item_key(buf, &disk_key, 0);
409         else
410                 btrfs_node_key(buf, &disk_key, 0);
411
412         if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) {
413                 if (parent)
414                         parent_start = parent->start;
415                 else
416                         parent_start = 0;
417         } else
418                 parent_start = 0;
419
420         cow = btrfs_alloc_free_block(trans, root, buf->len, parent_start,
421                                      root->root_key.objectid, &disk_key,
422                                      level, search_start, empty_size);
423         if (IS_ERR(cow))
424                 return PTR_ERR(cow);
425
426         /* cow is set to blocking by btrfs_init_new_buffer */
427
428         copy_extent_buffer(cow, buf, 0, 0, cow->len);
429         btrfs_set_header_bytenr(cow, cow->start);
430         btrfs_set_header_generation(cow, trans->transid);
431         btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
432         btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
433                                      BTRFS_HEADER_FLAG_RELOC);
434         if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
435                 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
436         else
437                 btrfs_set_header_owner(cow, root->root_key.objectid);
438
439         write_extent_buffer(cow, root->fs_info->fsid,
440                             (unsigned long)btrfs_header_fsid(cow),
441                             BTRFS_FSID_SIZE);
442
443         update_ref_for_cow(trans, root, buf, cow, &last_ref);
444
445         if (root->ref_cows)
446                 btrfs_reloc_cow_block(trans, root, buf, cow);
447
448         if (buf == root->node) {
449                 WARN_ON(parent && parent != buf);
450                 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
451                     btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
452                         parent_start = buf->start;
453                 else
454                         parent_start = 0;
455
456                 extent_buffer_get(cow);
457                 rcu_assign_pointer(root->node, cow);
458
459                 btrfs_free_tree_block(trans, root, buf, parent_start,
460                                       last_ref);
461                 free_extent_buffer(buf);
462                 add_root_to_dirty_list(root);
463         } else {
464                 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
465                         parent_start = parent->start;
466                 else
467                         parent_start = 0;
468
469                 WARN_ON(trans->transid != btrfs_header_generation(parent));
470                 btrfs_set_node_blockptr(parent, parent_slot,
471                                         cow->start);
472                 btrfs_set_node_ptr_generation(parent, parent_slot,
473                                               trans->transid);
474                 btrfs_mark_buffer_dirty(parent);
475                 btrfs_free_tree_block(trans, root, buf, parent_start,
476                                       last_ref);
477         }
478         if (unlock_orig)
479                 btrfs_tree_unlock(buf);
480         free_extent_buffer(buf);
481         btrfs_mark_buffer_dirty(cow);
482         *cow_ret = cow;
483         return 0;
484 }
485
486 static inline int should_cow_block(struct btrfs_trans_handle *trans,
487                                    struct btrfs_root *root,
488                                    struct extent_buffer *buf)
489 {
490         if (btrfs_header_generation(buf) == trans->transid &&
491             !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) &&
492             !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
493               btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)))
494                 return 0;
495         return 1;
496 }
497
498 /*
499  * cows a single block, see __btrfs_cow_block for the real work.
500  * This version of it has extra checks so that a block isn't cow'd more than
501  * once per transaction, as long as it hasn't been written yet
502  */
503 noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
504                     struct btrfs_root *root, struct extent_buffer *buf,
505                     struct extent_buffer *parent, int parent_slot,
506                     struct extent_buffer **cow_ret)
507 {
508         u64 search_start;
509         int ret;
510
511         if (trans->transaction != root->fs_info->running_transaction) {
512                 printk(KERN_CRIT "trans %llu running %llu\n",
513                        (unsigned long long)trans->transid,
514                        (unsigned long long)
515                        root->fs_info->running_transaction->transid);
516                 WARN_ON(1);
517         }
518         if (trans->transid != root->fs_info->generation) {
519                 printk(KERN_CRIT "trans %llu running %llu\n",
520                        (unsigned long long)trans->transid,
521                        (unsigned long long)root->fs_info->generation);
522                 WARN_ON(1);
523         }
524
525         if (!should_cow_block(trans, root, buf)) {
526                 *cow_ret = buf;
527                 return 0;
528         }
529
530         search_start = buf->start & ~((u64)(1024 * 1024 * 1024) - 1);
531
532         if (parent)
533                 btrfs_set_lock_blocking(parent);
534         btrfs_set_lock_blocking(buf);
535
536         ret = __btrfs_cow_block(trans, root, buf, parent,
537                                  parent_slot, cow_ret, search_start, 0);
538
539         trace_btrfs_cow_block(root, buf, *cow_ret);
540
541         return ret;
542 }
543
544 /*
545  * helper function for defrag to decide if two blocks pointed to by a
546  * node are actually close by
547  */
548 static int close_blocks(u64 blocknr, u64 other, u32 blocksize)
549 {
550         if (blocknr < other && other - (blocknr + blocksize) < 32768)
551                 return 1;
552         if (blocknr > other && blocknr - (other + blocksize) < 32768)
553                 return 1;
554         return 0;
555 }
556
557 /*
558  * compare two keys in a memcmp fashion
559  */
560 static int comp_keys(struct btrfs_disk_key *disk, struct btrfs_key *k2)
561 {
562         struct btrfs_key k1;
563
564         btrfs_disk_key_to_cpu(&k1, disk);
565
566         return btrfs_comp_cpu_keys(&k1, k2);
567 }
568
569 /*
570  * same as comp_keys only with two btrfs_key's
571  */
572 int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2)
573 {
574         if (k1->objectid > k2->objectid)
575                 return 1;
576         if (k1->objectid < k2->objectid)
577                 return -1;
578         if (k1->type > k2->type)
579                 return 1;
580         if (k1->type < k2->type)
581                 return -1;
582         if (k1->offset > k2->offset)
583                 return 1;
584         if (k1->offset < k2->offset)
585                 return -1;
586         return 0;
587 }
588
589 /*
590  * this is used by the defrag code to go through all the
591  * leaves pointed to by a node and reallocate them so that
592  * disk order is close to key order
593  */
594 int btrfs_realloc_node(struct btrfs_trans_handle *trans,
595                        struct btrfs_root *root, struct extent_buffer *parent,
596                        int start_slot, int cache_only, u64 *last_ret,
597                        struct btrfs_key *progress)
598 {
599         struct extent_buffer *cur;
600         u64 blocknr;
601         u64 gen;
602         u64 search_start = *last_ret;
603         u64 last_block = 0;
604         u64 other;
605         u32 parent_nritems;
606         int end_slot;
607         int i;
608         int err = 0;
609         int parent_level;
610         int uptodate;
611         u32 blocksize;
612         int progress_passed = 0;
613         struct btrfs_disk_key disk_key;
614
615         parent_level = btrfs_header_level(parent);
616         if (cache_only && parent_level != 1)
617                 return 0;
618
619         if (trans->transaction != root->fs_info->running_transaction)
620                 WARN_ON(1);
621         if (trans->transid != root->fs_info->generation)
622                 WARN_ON(1);
623
624         parent_nritems = btrfs_header_nritems(parent);
625         blocksize = btrfs_level_size(root, parent_level - 1);
626         end_slot = parent_nritems;
627
628         if (parent_nritems == 1)
629                 return 0;
630
631         btrfs_set_lock_blocking(parent);
632
633         for (i = start_slot; i < end_slot; i++) {
634                 int close = 1;
635
636                 if (!parent->map_token) {
637                         map_extent_buffer(parent,
638                                         btrfs_node_key_ptr_offset(i),
639                                         sizeof(struct btrfs_key_ptr),
640                                         &parent->map_token, &parent->kaddr,
641                                         &parent->map_start, &parent->map_len,
642                                         KM_USER1);
643                 }
644                 btrfs_node_key(parent, &disk_key, i);
645                 if (!progress_passed && comp_keys(&disk_key, progress) < 0)
646                         continue;
647
648                 progress_passed = 1;
649                 blocknr = btrfs_node_blockptr(parent, i);
650                 gen = btrfs_node_ptr_generation(parent, i);
651                 if (last_block == 0)
652                         last_block = blocknr;
653
654                 if (i > 0) {
655                         other = btrfs_node_blockptr(parent, i - 1);
656                         close = close_blocks(blocknr, other, blocksize);
657                 }
658                 if (!close && i < end_slot - 2) {
659                         other = btrfs_node_blockptr(parent, i + 1);
660                         close = close_blocks(blocknr, other, blocksize);
661                 }
662                 if (close) {
663                         last_block = blocknr;
664                         continue;
665                 }
666                 if (parent->map_token) {
667                         unmap_extent_buffer(parent, parent->map_token,
668                                             KM_USER1);
669                         parent->map_token = NULL;
670                 }
671
672                 cur = btrfs_find_tree_block(root, blocknr, blocksize);
673                 if (cur)
674                         uptodate = btrfs_buffer_uptodate(cur, gen);
675                 else
676                         uptodate = 0;
677                 if (!cur || !uptodate) {
678                         if (cache_only) {
679                                 free_extent_buffer(cur);
680                                 continue;
681                         }
682                         if (!cur) {
683                                 cur = read_tree_block(root, blocknr,
684                                                          blocksize, gen);
685                                 if (!cur)
686                                         return -EIO;
687                         } else if (!uptodate) {
688                                 btrfs_read_buffer(cur, gen);
689                         }
690                 }
691                 if (search_start == 0)
692                         search_start = last_block;
693
694                 btrfs_tree_lock(cur);
695                 btrfs_set_lock_blocking(cur);
696                 err = __btrfs_cow_block(trans, root, cur, parent, i,
697                                         &cur, search_start,
698                                         min(16 * blocksize,
699                                             (end_slot - i) * blocksize));
700                 if (err) {
701                         btrfs_tree_unlock(cur);
702                         free_extent_buffer(cur);
703                         break;
704                 }
705                 search_start = cur->start;
706                 last_block = cur->start;
707                 *last_ret = search_start;
708                 btrfs_tree_unlock(cur);
709                 free_extent_buffer(cur);
710         }
711         if (parent->map_token) {
712                 unmap_extent_buffer(parent, parent->map_token,
713                                     KM_USER1);
714                 parent->map_token = NULL;
715         }
716         return err;
717 }
718
719 /*
720  * The leaf data grows from end-to-front in the node.
721  * this returns the address of the start of the last item,
722  * which is the stop of the leaf data stack
723  */
724 static inline unsigned int leaf_data_end(struct btrfs_root *root,
725                                          struct extent_buffer *leaf)
726 {
727         u32 nr = btrfs_header_nritems(leaf);
728         if (nr == 0)
729                 return BTRFS_LEAF_DATA_SIZE(root);
730         return btrfs_item_offset_nr(leaf, nr - 1);
731 }
732
733
734 /*
735  * search for key in the extent_buffer.  The items start at offset p,
736  * and they are item_size apart.  There are 'max' items in p.
737  *
738  * the slot in the array is returned via slot, and it points to
739  * the place where you would insert key if it is not found in
740  * the array.
741  *
742  * slot may point to max if the key is bigger than all of the keys
743  */
744 static noinline int generic_bin_search(struct extent_buffer *eb,
745                                        unsigned long p,
746                                        int item_size, struct btrfs_key *key,
747                                        int max, int *slot)
748 {
749         int low = 0;
750         int high = max;
751         int mid;
752         int ret;
753         struct btrfs_disk_key *tmp = NULL;
754         struct btrfs_disk_key unaligned;
755         unsigned long offset;
756         char *map_token = NULL;
757         char *kaddr = NULL;
758         unsigned long map_start = 0;
759         unsigned long map_len = 0;
760         int err;
761
762         while (low < high) {
763                 mid = (low + high) / 2;
764                 offset = p + mid * item_size;
765
766                 if (!map_token || offset < map_start ||
767                     (offset + sizeof(struct btrfs_disk_key)) >
768                     map_start + map_len) {
769                         if (map_token) {
770                                 unmap_extent_buffer(eb, map_token, KM_USER0);
771                                 map_token = NULL;
772                         }
773
774                         err = map_private_extent_buffer(eb, offset,
775                                                 sizeof(struct btrfs_disk_key),
776                                                 &map_token, &kaddr,
777                                                 &map_start, &map_len, KM_USER0);
778
779                         if (!err) {
780                                 tmp = (struct btrfs_disk_key *)(kaddr + offset -
781                                                         map_start);
782                         } else {
783                                 read_extent_buffer(eb, &unaligned,
784                                                    offset, sizeof(unaligned));
785                                 tmp = &unaligned;
786                         }
787
788                 } else {
789                         tmp = (struct btrfs_disk_key *)(kaddr + offset -
790                                                         map_start);
791                 }
792                 ret = comp_keys(tmp, key);
793
794                 if (ret < 0)
795                         low = mid + 1;
796                 else if (ret > 0)
797                         high = mid;
798                 else {
799                         *slot = mid;
800                         if (map_token)
801                                 unmap_extent_buffer(eb, map_token, KM_USER0);
802                         return 0;
803                 }
804         }
805         *slot = low;
806         if (map_token)
807                 unmap_extent_buffer(eb, map_token, KM_USER0);
808         return 1;
809 }
810
811 /*
812  * simple bin_search frontend that does the right thing for
813  * leaves vs nodes
814  */
815 static int bin_search(struct extent_buffer *eb, struct btrfs_key *key,
816                       int level, int *slot)
817 {
818         if (level == 0) {
819                 return generic_bin_search(eb,
820                                           offsetof(struct btrfs_leaf, items),
821                                           sizeof(struct btrfs_item),
822                                           key, btrfs_header_nritems(eb),
823                                           slot);
824         } else {
825                 return generic_bin_search(eb,
826                                           offsetof(struct btrfs_node, ptrs),
827                                           sizeof(struct btrfs_key_ptr),
828                                           key, btrfs_header_nritems(eb),
829                                           slot);
830         }
831         return -1;
832 }
833
834 int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key,
835                      int level, int *slot)
836 {
837         return bin_search(eb, key, level, slot);
838 }
839
840 static void root_add_used(struct btrfs_root *root, u32 size)
841 {
842         spin_lock(&root->accounting_lock);
843         btrfs_set_root_used(&root->root_item,
844                             btrfs_root_used(&root->root_item) + size);
845         spin_unlock(&root->accounting_lock);
846 }
847
848 static void root_sub_used(struct btrfs_root *root, u32 size)
849 {
850         spin_lock(&root->accounting_lock);
851         btrfs_set_root_used(&root->root_item,
852                             btrfs_root_used(&root->root_item) - size);
853         spin_unlock(&root->accounting_lock);
854 }
855
856 /* given a node and slot number, this reads the blocks it points to.  The
857  * extent buffer is returned with a reference taken (but unlocked).
858  * NULL is returned on error.
859  */
860 static noinline struct extent_buffer *read_node_slot(struct btrfs_root *root,
861                                    struct extent_buffer *parent, int slot)
862 {
863         int level = btrfs_header_level(parent);
864         if (slot < 0)
865                 return NULL;
866         if (slot >= btrfs_header_nritems(parent))
867                 return NULL;
868
869         BUG_ON(level == 0);
870
871         return read_tree_block(root, btrfs_node_blockptr(parent, slot),
872                        btrfs_level_size(root, level - 1),
873                        btrfs_node_ptr_generation(parent, slot));
874 }
875
876 /*
877  * node level balancing, used to make sure nodes are in proper order for
878  * item deletion.  We balance from the top down, so we have to make sure
879  * that a deletion won't leave an node completely empty later on.
880  */
881 static noinline int balance_level(struct btrfs_trans_handle *trans,
882                          struct btrfs_root *root,
883                          struct btrfs_path *path, int level)
884 {
885         struct extent_buffer *right = NULL;
886         struct extent_buffer *mid;
887         struct extent_buffer *left = NULL;
888         struct extent_buffer *parent = NULL;
889         int ret = 0;
890         int wret;
891         int pslot;
892         int orig_slot = path->slots[level];
893         u64 orig_ptr;
894
895         if (level == 0)
896                 return 0;
897
898         mid = path->nodes[level];
899
900         WARN_ON(!path->locks[level]);
901         WARN_ON(btrfs_header_generation(mid) != trans->transid);
902
903         orig_ptr = btrfs_node_blockptr(mid, orig_slot);
904
905         if (level < BTRFS_MAX_LEVEL - 1)
906                 parent = path->nodes[level + 1];
907         pslot = path->slots[level + 1];
908
909         /*
910          * deal with the case where there is only one pointer in the root
911          * by promoting the node below to a root
912          */
913         if (!parent) {
914                 struct extent_buffer *child;
915
916                 if (btrfs_header_nritems(mid) != 1)
917                         return 0;
918
919                 /* promote the child to a root */
920                 child = read_node_slot(root, mid, 0);
921                 BUG_ON(!child);
922                 btrfs_tree_lock(child);
923                 btrfs_set_lock_blocking(child);
924                 ret = btrfs_cow_block(trans, root, child, mid, 0, &child);
925                 if (ret) {
926                         btrfs_tree_unlock(child);
927                         free_extent_buffer(child);
928                         goto enospc;
929                 }
930
931                 rcu_assign_pointer(root->node, child);
932
933                 add_root_to_dirty_list(root);
934                 btrfs_tree_unlock(child);
935
936                 path->locks[level] = 0;
937                 path->nodes[level] = NULL;
938                 clean_tree_block(trans, root, mid);
939                 btrfs_tree_unlock(mid);
940                 /* once for the path */
941                 free_extent_buffer(mid);
942
943                 root_sub_used(root, mid->len);
944                 btrfs_free_tree_block(trans, root, mid, 0, 1);
945                 /* once for the root ptr */
946                 free_extent_buffer(mid);
947                 return 0;
948         }
949         if (btrfs_header_nritems(mid) >
950             BTRFS_NODEPTRS_PER_BLOCK(root) / 4)
951                 return 0;
952
953         btrfs_header_nritems(mid);
954
955         left = read_node_slot(root, parent, pslot - 1);
956         if (left) {
957                 btrfs_tree_lock(left);
958                 btrfs_set_lock_blocking(left);
959                 wret = btrfs_cow_block(trans, root, left,
960                                        parent, pslot - 1, &left);
961                 if (wret) {
962                         ret = wret;
963                         goto enospc;
964                 }
965         }
966         right = read_node_slot(root, parent, pslot + 1);
967         if (right) {
968                 btrfs_tree_lock(right);
969                 btrfs_set_lock_blocking(right);
970                 wret = btrfs_cow_block(trans, root, right,
971                                        parent, pslot + 1, &right);
972                 if (wret) {
973                         ret = wret;
974                         goto enospc;
975                 }
976         }
977
978         /* first, try to make some room in the middle buffer */
979         if (left) {
980                 orig_slot += btrfs_header_nritems(left);
981                 wret = push_node_left(trans, root, left, mid, 1);
982                 if (wret < 0)
983                         ret = wret;
984                 btrfs_header_nritems(mid);
985         }
986
987         /*
988          * then try to empty the right most buffer into the middle
989          */
990         if (right) {
991                 wret = push_node_left(trans, root, mid, right, 1);
992                 if (wret < 0 && wret != -ENOSPC)
993                         ret = wret;
994                 if (btrfs_header_nritems(right) == 0) {
995                         clean_tree_block(trans, root, right);
996                         btrfs_tree_unlock(right);
997                         wret = del_ptr(trans, root, path, level + 1, pslot +
998                                        1);
999                         if (wret)
1000                                 ret = wret;
1001                         root_sub_used(root, right->len);
1002                         btrfs_free_tree_block(trans, root, right, 0, 1);
1003                         free_extent_buffer(right);
1004                         right = NULL;
1005                 } else {
1006                         struct btrfs_disk_key right_key;
1007                         btrfs_node_key(right, &right_key, 0);
1008                         btrfs_set_node_key(parent, &right_key, pslot + 1);
1009                         btrfs_mark_buffer_dirty(parent);
1010                 }
1011         }
1012         if (btrfs_header_nritems(mid) == 1) {
1013                 /*
1014                  * we're not allowed to leave a node with one item in the
1015                  * tree during a delete.  A deletion from lower in the tree
1016                  * could try to delete the only pointer in this node.
1017                  * So, pull some keys from the left.
1018                  * There has to be a left pointer at this point because
1019                  * otherwise we would have pulled some pointers from the
1020                  * right
1021                  */
1022                 BUG_ON(!left);
1023                 wret = balance_node_right(trans, root, mid, left);
1024                 if (wret < 0) {
1025                         ret = wret;
1026                         goto enospc;
1027                 }
1028                 if (wret == 1) {
1029                         wret = push_node_left(trans, root, left, mid, 1);
1030                         if (wret < 0)
1031                                 ret = wret;
1032                 }
1033                 BUG_ON(wret == 1);
1034         }
1035         if (btrfs_header_nritems(mid) == 0) {
1036                 clean_tree_block(trans, root, mid);
1037                 btrfs_tree_unlock(mid);
1038                 wret = del_ptr(trans, root, path, level + 1, pslot);
1039                 if (wret)
1040                         ret = wret;
1041                 root_sub_used(root, mid->len);
1042                 btrfs_free_tree_block(trans, root, mid, 0, 1);
1043                 free_extent_buffer(mid);
1044                 mid = NULL;
1045         } else {
1046                 /* update the parent key to reflect our changes */
1047                 struct btrfs_disk_key mid_key;
1048                 btrfs_node_key(mid, &mid_key, 0);
1049                 btrfs_set_node_key(parent, &mid_key, pslot);
1050                 btrfs_mark_buffer_dirty(parent);
1051         }
1052
1053         /* update the path */
1054         if (left) {
1055                 if (btrfs_header_nritems(left) > orig_slot) {
1056                         extent_buffer_get(left);
1057                         /* left was locked after cow */
1058                         path->nodes[level] = left;
1059                         path->slots[level + 1] -= 1;
1060                         path->slots[level] = orig_slot;
1061                         if (mid) {
1062                                 btrfs_tree_unlock(mid);
1063                                 free_extent_buffer(mid);
1064                         }
1065                 } else {
1066                         orig_slot -= btrfs_header_nritems(left);
1067                         path->slots[level] = orig_slot;
1068                 }
1069         }
1070         /* double check we haven't messed things up */
1071         if (orig_ptr !=
1072             btrfs_node_blockptr(path->nodes[level], path->slots[level]))
1073                 BUG();
1074 enospc:
1075         if (right) {
1076                 btrfs_tree_unlock(right);
1077                 free_extent_buffer(right);
1078         }
1079         if (left) {
1080                 if (path->nodes[level] != left)
1081                         btrfs_tree_unlock(left);
1082                 free_extent_buffer(left);
1083         }
1084         return ret;
1085 }
1086
1087 /* Node balancing for insertion.  Here we only split or push nodes around
1088  * when they are completely full.  This is also done top down, so we
1089  * have to be pessimistic.
1090  */
1091 static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
1092                                           struct btrfs_root *root,
1093                                           struct btrfs_path *path, int level)
1094 {
1095         struct extent_buffer *right = NULL;
1096         struct extent_buffer *mid;
1097         struct extent_buffer *left = NULL;
1098         struct extent_buffer *parent = NULL;
1099         int ret = 0;
1100         int wret;
1101         int pslot;
1102         int orig_slot = path->slots[level];
1103
1104         if (level == 0)
1105                 return 1;
1106
1107         mid = path->nodes[level];
1108         WARN_ON(btrfs_header_generation(mid) != trans->transid);
1109
1110         if (level < BTRFS_MAX_LEVEL - 1)
1111                 parent = path->nodes[level + 1];
1112         pslot = path->slots[level + 1];
1113
1114         if (!parent)
1115                 return 1;
1116
1117         left = read_node_slot(root, parent, pslot - 1);
1118
1119         /* first, try to make some room in the middle buffer */
1120         if (left) {
1121                 u32 left_nr;
1122
1123                 btrfs_tree_lock(left);
1124                 btrfs_set_lock_blocking(left);
1125
1126                 left_nr = btrfs_header_nritems(left);
1127                 if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) {
1128                         wret = 1;
1129                 } else {
1130                         ret = btrfs_cow_block(trans, root, left, parent,
1131                                               pslot - 1, &left);
1132                         if (ret)
1133                                 wret = 1;
1134                         else {
1135                                 wret = push_node_left(trans, root,
1136                                                       left, mid, 0);
1137                         }
1138                 }
1139                 if (wret < 0)
1140                         ret = wret;
1141                 if (wret == 0) {
1142                         struct btrfs_disk_key disk_key;
1143                         orig_slot += left_nr;
1144                         btrfs_node_key(mid, &disk_key, 0);
1145                         btrfs_set_node_key(parent, &disk_key, pslot);
1146                         btrfs_mark_buffer_dirty(parent);
1147                         if (btrfs_header_nritems(left) > orig_slot) {
1148                                 path->nodes[level] = left;
1149                                 path->slots[level + 1] -= 1;
1150                                 path->slots[level] = orig_slot;
1151                                 btrfs_tree_unlock(mid);
1152                                 free_extent_buffer(mid);
1153                         } else {
1154                                 orig_slot -=
1155                                         btrfs_header_nritems(left);
1156                                 path->slots[level] = orig_slot;
1157                                 btrfs_tree_unlock(left);
1158                                 free_extent_buffer(left);
1159                         }
1160                         return 0;
1161                 }
1162                 btrfs_tree_unlock(left);
1163                 free_extent_buffer(left);
1164         }
1165         right = read_node_slot(root, parent, pslot + 1);
1166
1167         /*
1168          * then try to empty the right most buffer into the middle
1169          */
1170         if (right) {
1171                 u32 right_nr;
1172
1173                 btrfs_tree_lock(right);
1174                 btrfs_set_lock_blocking(right);
1175
1176                 right_nr = btrfs_header_nritems(right);
1177                 if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) {
1178                         wret = 1;
1179                 } else {
1180                         ret = btrfs_cow_block(trans, root, right,
1181                                               parent, pslot + 1,
1182                                               &right);
1183                         if (ret)
1184                                 wret = 1;
1185                         else {
1186                                 wret = balance_node_right(trans, root,
1187                                                           right, mid);
1188                         }
1189                 }
1190                 if (wret < 0)
1191                         ret = wret;
1192                 if (wret == 0) {
1193                         struct btrfs_disk_key disk_key;
1194
1195                         btrfs_node_key(right, &disk_key, 0);
1196                         btrfs_set_node_key(parent, &disk_key, pslot + 1);
1197                         btrfs_mark_buffer_dirty(parent);
1198
1199                         if (btrfs_header_nritems(mid) <= orig_slot) {
1200                                 path->nodes[level] = right;
1201                                 path->slots[level + 1] += 1;
1202                                 path->slots[level] = orig_slot -
1203                                         btrfs_header_nritems(mid);
1204                                 btrfs_tree_unlock(mid);
1205                                 free_extent_buffer(mid);
1206                         } else {
1207                                 btrfs_tree_unlock(right);
1208                                 free_extent_buffer(right);
1209                         }
1210                         return 0;
1211                 }
1212                 btrfs_tree_unlock(right);
1213                 free_extent_buffer(right);
1214         }
1215         return 1;
1216 }
1217
1218 /*
1219  * readahead one full node of leaves, finding things that are close
1220  * to the block in 'slot', and triggering ra on them.
1221  */
1222 static void reada_for_search(struct btrfs_root *root,
1223                              struct btrfs_path *path,
1224                              int level, int slot, u64 objectid)
1225 {
1226         struct extent_buffer *node;
1227         struct btrfs_disk_key disk_key;
1228         u32 nritems;
1229         u64 search;
1230         u64 target;
1231         u64 nread = 0;
1232         int direction = path->reada;
1233         struct extent_buffer *eb;
1234         u32 nr;
1235         u32 blocksize;
1236         u32 nscan = 0;
1237
1238         if (level != 1)
1239                 return;
1240
1241         if (!path->nodes[level])
1242                 return;
1243
1244         node = path->nodes[level];
1245
1246         search = btrfs_node_blockptr(node, slot);
1247         blocksize = btrfs_level_size(root, level - 1);
1248         eb = btrfs_find_tree_block(root, search, blocksize);
1249         if (eb) {
1250                 free_extent_buffer(eb);
1251                 return;
1252         }
1253
1254         target = search;
1255
1256         nritems = btrfs_header_nritems(node);
1257         nr = slot;
1258         while (1) {
1259                 if (direction < 0) {
1260                         if (nr == 0)
1261                                 break;
1262                         nr--;
1263                 } else if (direction > 0) {
1264                         nr++;
1265                         if (nr >= nritems)
1266                                 break;
1267                 }
1268                 if (path->reada < 0 && objectid) {
1269                         btrfs_node_key(node, &disk_key, nr);
1270                         if (btrfs_disk_key_objectid(&disk_key) != objectid)
1271                                 break;
1272                 }
1273                 search = btrfs_node_blockptr(node, nr);
1274                 if ((search <= target && target - search <= 65536) ||
1275                     (search > target && search - target <= 65536)) {
1276                         readahead_tree_block(root, search, blocksize,
1277                                      btrfs_node_ptr_generation(node, nr));
1278                         nread += blocksize;
1279                 }
1280                 nscan++;
1281                 if ((nread > 65536 || nscan > 32))
1282                         break;
1283         }
1284 }
1285
1286 /*
1287  * returns -EAGAIN if it had to drop the path, or zero if everything was in
1288  * cache
1289  */
1290 static noinline int reada_for_balance(struct btrfs_root *root,
1291                                       struct btrfs_path *path, int level)
1292 {
1293         int slot;
1294         int nritems;
1295         struct extent_buffer *parent;
1296         struct extent_buffer *eb;
1297         u64 gen;
1298         u64 block1 = 0;
1299         u64 block2 = 0;
1300         int ret = 0;
1301         int blocksize;
1302
1303         parent = path->nodes[level + 1];
1304         if (!parent)
1305                 return 0;
1306
1307         nritems = btrfs_header_nritems(parent);
1308         slot = path->slots[level + 1];
1309         blocksize = btrfs_level_size(root, level);
1310
1311         if (slot > 0) {
1312                 block1 = btrfs_node_blockptr(parent, slot - 1);
1313                 gen = btrfs_node_ptr_generation(parent, slot - 1);
1314                 eb = btrfs_find_tree_block(root, block1, blocksize);
1315                 if (eb && btrfs_buffer_uptodate(eb, gen))
1316                         block1 = 0;
1317                 free_extent_buffer(eb);
1318         }
1319         if (slot + 1 < nritems) {
1320                 block2 = btrfs_node_blockptr(parent, slot + 1);
1321                 gen = btrfs_node_ptr_generation(parent, slot + 1);
1322                 eb = btrfs_find_tree_block(root, block2, blocksize);
1323                 if (eb && btrfs_buffer_uptodate(eb, gen))
1324                         block2 = 0;
1325                 free_extent_buffer(eb);
1326         }
1327         if (block1 || block2) {
1328                 ret = -EAGAIN;
1329
1330                 /* release the whole path */
1331                 btrfs_release_path(root, path);
1332
1333                 /* read the blocks */
1334                 if (block1)
1335                         readahead_tree_block(root, block1, blocksize, 0);
1336                 if (block2)
1337                         readahead_tree_block(root, block2, blocksize, 0);
1338
1339                 if (block1) {
1340                         eb = read_tree_block(root, block1, blocksize, 0);
1341                         free_extent_buffer(eb);
1342                 }
1343                 if (block2) {
1344                         eb = read_tree_block(root, block2, blocksize, 0);
1345                         free_extent_buffer(eb);
1346                 }
1347         }
1348         return ret;
1349 }
1350
1351
1352 /*
1353  * when we walk down the tree, it is usually safe to unlock the higher layers
1354  * in the tree.  The exceptions are when our path goes through slot 0, because
1355  * operations on the tree might require changing key pointers higher up in the
1356  * tree.
1357  *
1358  * callers might also have set path->keep_locks, which tells this code to keep
1359  * the lock if the path points to the last slot in the block.  This is part of
1360  * walking through the tree, and selecting the next slot in the higher block.
1361  *
1362  * lowest_unlock sets the lowest level in the tree we're allowed to unlock.  so
1363  * if lowest_unlock is 1, level 0 won't be unlocked
1364  */
1365 static noinline void unlock_up(struct btrfs_path *path, int level,
1366                                int lowest_unlock)
1367 {
1368         int i;
1369         int skip_level = level;
1370         int no_skips = 0;
1371         struct extent_buffer *t;
1372
1373         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
1374                 if (!path->nodes[i])
1375                         break;
1376                 if (!path->locks[i])
1377                         break;
1378                 if (!no_skips && path->slots[i] == 0) {
1379                         skip_level = i + 1;
1380                         continue;
1381                 }
1382                 if (!no_skips && path->keep_locks) {
1383                         u32 nritems;
1384                         t = path->nodes[i];
1385                         nritems = btrfs_header_nritems(t);
1386                         if (nritems < 1 || path->slots[i] >= nritems - 1) {
1387                                 skip_level = i + 1;
1388                                 continue;
1389                         }
1390                 }
1391                 if (skip_level < i && i >= lowest_unlock)
1392                         no_skips = 1;
1393
1394                 t = path->nodes[i];
1395                 if (i >= lowest_unlock && i > skip_level && path->locks[i]) {
1396                         btrfs_tree_unlock(t);
1397                         path->locks[i] = 0;
1398                 }
1399         }
1400 }
1401
1402 /*
1403  * This releases any locks held in the path starting at level and
1404  * going all the way up to the root.
1405  *
1406  * btrfs_search_slot will keep the lock held on higher nodes in a few
1407  * corner cases, such as COW of the block at slot zero in the node.  This
1408  * ignores those rules, and it should only be called when there are no
1409  * more updates to be done higher up in the tree.
1410  */
1411 noinline void btrfs_unlock_up_safe(struct btrfs_path *path, int level)
1412 {
1413         int i;
1414
1415         if (path->keep_locks)
1416                 return;
1417
1418         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
1419                 if (!path->nodes[i])
1420                         continue;
1421                 if (!path->locks[i])
1422                         continue;
1423                 btrfs_tree_unlock(path->nodes[i]);
1424                 path->locks[i] = 0;
1425         }
1426 }
1427
1428 /*
1429  * helper function for btrfs_search_slot.  The goal is to find a block
1430  * in cache without setting the path to blocking.  If we find the block
1431  * we return zero and the path is unchanged.
1432  *
1433  * If we can't find the block, we set the path blocking and do some
1434  * reada.  -EAGAIN is returned and the search must be repeated.
1435  */
1436 static int
1437 read_block_for_search(struct btrfs_trans_handle *trans,
1438                        struct btrfs_root *root, struct btrfs_path *p,
1439                        struct extent_buffer **eb_ret, int level, int slot,
1440                        struct btrfs_key *key)
1441 {
1442         u64 blocknr;
1443         u64 gen;
1444         u32 blocksize;
1445         struct extent_buffer *b = *eb_ret;
1446         struct extent_buffer *tmp;
1447         int ret;
1448
1449         blocknr = btrfs_node_blockptr(b, slot);
1450         gen = btrfs_node_ptr_generation(b, slot);
1451         blocksize = btrfs_level_size(root, level - 1);
1452
1453         tmp = btrfs_find_tree_block(root, blocknr, blocksize);
1454         if (tmp) {
1455                 if (btrfs_buffer_uptodate(tmp, 0)) {
1456                         if (btrfs_buffer_uptodate(tmp, gen)) {
1457                                 /*
1458                                  * we found an up to date block without
1459                                  * sleeping, return
1460                                  * right away
1461                                  */
1462                                 *eb_ret = tmp;
1463                                 return 0;
1464                         }
1465                         /* the pages were up to date, but we failed
1466                          * the generation number check.  Do a full
1467                          * read for the generation number that is correct.
1468                          * We must do this without dropping locks so
1469                          * we can trust our generation number
1470                          */
1471                         free_extent_buffer(tmp);
1472                         tmp = read_tree_block(root, blocknr, blocksize, gen);
1473                         if (tmp && btrfs_buffer_uptodate(tmp, gen)) {
1474                                 *eb_ret = tmp;
1475                                 return 0;
1476                         }
1477                         free_extent_buffer(tmp);
1478                         btrfs_release_path(NULL, p);
1479                         return -EIO;
1480                 }
1481         }
1482
1483         /*
1484          * reduce lock contention at high levels
1485          * of the btree by dropping locks before
1486          * we read.  Don't release the lock on the current
1487          * level because we need to walk this node to figure
1488          * out which blocks to read.
1489          */
1490         btrfs_unlock_up_safe(p, level + 1);
1491         btrfs_set_path_blocking(p);
1492
1493         free_extent_buffer(tmp);
1494         if (p->reada)
1495                 reada_for_search(root, p, level, slot, key->objectid);
1496
1497         btrfs_release_path(NULL, p);
1498
1499         ret = -EAGAIN;
1500         tmp = read_tree_block(root, blocknr, blocksize, 0);
1501         if (tmp) {
1502                 /*
1503                  * If the read above didn't mark this buffer up to date,
1504                  * it will never end up being up to date.  Set ret to EIO now
1505                  * and give up so that our caller doesn't loop forever
1506                  * on our EAGAINs.
1507                  */
1508                 if (!btrfs_buffer_uptodate(tmp, 0))
1509                         ret = -EIO;
1510                 free_extent_buffer(tmp);
1511         }
1512         return ret;
1513 }
1514
1515 /*
1516  * helper function for btrfs_search_slot.  This does all of the checks
1517  * for node-level blocks and does any balancing required based on
1518  * the ins_len.
1519  *
1520  * If no extra work was required, zero is returned.  If we had to
1521  * drop the path, -EAGAIN is returned and btrfs_search_slot must
1522  * start over
1523  */
1524 static int
1525 setup_nodes_for_search(struct btrfs_trans_handle *trans,
1526                        struct btrfs_root *root, struct btrfs_path *p,
1527                        struct extent_buffer *b, int level, int ins_len)
1528 {
1529         int ret;
1530         if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >=
1531             BTRFS_NODEPTRS_PER_BLOCK(root) - 3) {
1532                 int sret;
1533
1534                 sret = reada_for_balance(root, p, level);
1535                 if (sret)
1536                         goto again;
1537
1538                 btrfs_set_path_blocking(p);
1539                 sret = split_node(trans, root, p, level);
1540                 btrfs_clear_path_blocking(p, NULL);
1541
1542                 BUG_ON(sret > 0);
1543                 if (sret) {
1544                         ret = sret;
1545                         goto done;
1546                 }
1547                 b = p->nodes[level];
1548         } else if (ins_len < 0 && btrfs_header_nritems(b) <
1549                    BTRFS_NODEPTRS_PER_BLOCK(root) / 2) {
1550                 int sret;
1551
1552                 sret = reada_for_balance(root, p, level);
1553                 if (sret)
1554                         goto again;
1555
1556                 btrfs_set_path_blocking(p);
1557                 sret = balance_level(trans, root, p, level);
1558                 btrfs_clear_path_blocking(p, NULL);
1559
1560                 if (sret) {
1561                         ret = sret;
1562                         goto done;
1563                 }
1564                 b = p->nodes[level];
1565                 if (!b) {
1566                         btrfs_release_path(NULL, p);
1567                         goto again;
1568                 }
1569                 BUG_ON(btrfs_header_nritems(b) == 1);
1570         }
1571         return 0;
1572
1573 again:
1574         ret = -EAGAIN;
1575 done:
1576         return ret;
1577 }
1578
1579 /*
1580  * look for key in the tree.  path is filled in with nodes along the way
1581  * if key is found, we return zero and you can find the item in the leaf
1582  * level of the path (level 0)
1583  *
1584  * If the key isn't found, the path points to the slot where it should
1585  * be inserted, and 1 is returned.  If there are other errors during the
1586  * search a negative error number is returned.
1587  *
1588  * if ins_len > 0, nodes and leaves will be split as we walk down the
1589  * tree.  if ins_len < 0, nodes will be merged as we walk down the tree (if
1590  * possible)
1591  */
1592 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
1593                       *root, struct btrfs_key *key, struct btrfs_path *p, int
1594                       ins_len, int cow)
1595 {
1596         struct extent_buffer *b;
1597         int slot;
1598         int ret;
1599         int err;
1600         int level;
1601         int lowest_unlock = 1;
1602         u8 lowest_level = 0;
1603
1604         lowest_level = p->lowest_level;
1605         WARN_ON(lowest_level && ins_len > 0);
1606         WARN_ON(p->nodes[0] != NULL);
1607
1608         if (ins_len < 0)
1609                 lowest_unlock = 2;
1610
1611 again:
1612         if (p->search_commit_root) {
1613                 b = root->commit_root;
1614                 extent_buffer_get(b);
1615                 if (!p->skip_locking)
1616                         btrfs_tree_lock(b);
1617         } else {
1618                 if (p->skip_locking)
1619                         b = btrfs_root_node(root);
1620                 else
1621                         b = btrfs_lock_root_node(root);
1622         }
1623
1624         while (b) {
1625                 level = btrfs_header_level(b);
1626
1627                 /*
1628                  * setup the path here so we can release it under lock
1629                  * contention with the cow code
1630                  */
1631                 p->nodes[level] = b;
1632                 if (!p->skip_locking)
1633                         p->locks[level] = 1;
1634
1635                 if (cow) {
1636                         /*
1637                          * if we don't really need to cow this block
1638                          * then we don't want to set the path blocking,
1639                          * so we test it here
1640                          */
1641                         if (!should_cow_block(trans, root, b))
1642                                 goto cow_done;
1643
1644                         btrfs_set_path_blocking(p);
1645
1646                         err = btrfs_cow_block(trans, root, b,
1647                                               p->nodes[level + 1],
1648                                               p->slots[level + 1], &b);
1649                         if (err) {
1650                                 ret = err;
1651                                 goto done;
1652                         }
1653                 }
1654 cow_done:
1655                 BUG_ON(!cow && ins_len);
1656                 if (level != btrfs_header_level(b))
1657                         WARN_ON(1);
1658                 level = btrfs_header_level(b);
1659
1660                 p->nodes[level] = b;
1661                 if (!p->skip_locking)
1662                         p->locks[level] = 1;
1663
1664                 btrfs_clear_path_blocking(p, NULL);
1665
1666                 /*
1667                  * we have a lock on b and as long as we aren't changing
1668                  * the tree, there is no way to for the items in b to change.
1669                  * It is safe to drop the lock on our parent before we
1670                  * go through the expensive btree search on b.
1671                  *
1672                  * If cow is true, then we might be changing slot zero,
1673                  * which may require changing the parent.  So, we can't
1674                  * drop the lock until after we know which slot we're
1675                  * operating on.
1676                  */
1677                 if (!cow)
1678                         btrfs_unlock_up_safe(p, level + 1);
1679
1680                 ret = bin_search(b, key, level, &slot);
1681
1682                 if (level != 0) {
1683                         int dec = 0;
1684                         if (ret && slot > 0) {
1685                                 dec = 1;
1686                                 slot -= 1;
1687                         }
1688                         p->slots[level] = slot;
1689                         err = setup_nodes_for_search(trans, root, p, b, level,
1690                                                      ins_len);
1691                         if (err == -EAGAIN)
1692                                 goto again;
1693                         if (err) {
1694                                 ret = err;
1695                                 goto done;
1696                         }
1697                         b = p->nodes[level];
1698                         slot = p->slots[level];
1699
1700                         unlock_up(p, level, lowest_unlock);
1701
1702                         if (level == lowest_level) {
1703                                 if (dec)
1704                                         p->slots[level]++;
1705                                 goto done;
1706                         }
1707
1708                         err = read_block_for_search(trans, root, p,
1709                                                     &b, level, slot, key);
1710                         if (err == -EAGAIN)
1711                                 goto again;
1712                         if (err) {
1713                                 ret = err;
1714                                 goto done;
1715                         }
1716
1717                         if (!p->skip_locking) {
1718                                 btrfs_clear_path_blocking(p, NULL);
1719                                 err = btrfs_try_spin_lock(b);
1720
1721                                 if (!err) {
1722                                         btrfs_set_path_blocking(p);
1723                                         btrfs_tree_lock(b);
1724                                         btrfs_clear_path_blocking(p, b);
1725                                 }
1726                         }
1727                 } else {
1728                         p->slots[level] = slot;
1729                         if (ins_len > 0 &&
1730                             btrfs_leaf_free_space(root, b) < ins_len) {
1731                                 btrfs_set_path_blocking(p);
1732                                 err = split_leaf(trans, root, key,
1733                                                  p, ins_len, ret == 0);
1734                                 btrfs_clear_path_blocking(p, NULL);
1735
1736                                 BUG_ON(err > 0);
1737                                 if (err) {
1738                                         ret = err;
1739                                         goto done;
1740                                 }
1741                         }
1742                         if (!p->search_for_split)
1743                                 unlock_up(p, level, lowest_unlock);
1744                         goto done;
1745                 }
1746         }
1747         ret = 1;
1748 done:
1749         /*
1750          * we don't really know what they plan on doing with the path
1751          * from here on, so for now just mark it as blocking
1752          */
1753         if (!p->leave_spinning)
1754                 btrfs_set_path_blocking(p);
1755         if (ret < 0)
1756                 btrfs_release_path(root, p);
1757         return ret;
1758 }
1759
1760 /*
1761  * adjust the pointers going up the tree, starting at level
1762  * making sure the right key of each node is points to 'key'.
1763  * This is used after shifting pointers to the left, so it stops
1764  * fixing up pointers when a given leaf/node is not in slot 0 of the
1765  * higher levels
1766  *
1767  * If this fails to write a tree block, it returns -1, but continues
1768  * fixing up the blocks in ram so the tree is consistent.
1769  */
1770 static int fixup_low_keys(struct btrfs_trans_handle *trans,
1771                           struct btrfs_root *root, struct btrfs_path *path,
1772                           struct btrfs_disk_key *key, int level)
1773 {
1774         int i;
1775         int ret = 0;
1776         struct extent_buffer *t;
1777
1778         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
1779                 int tslot = path->slots[i];
1780                 if (!path->nodes[i])
1781                         break;
1782                 t = path->nodes[i];
1783                 btrfs_set_node_key(t, key, tslot);
1784                 btrfs_mark_buffer_dirty(path->nodes[i]);
1785                 if (tslot != 0)
1786                         break;
1787         }
1788         return ret;
1789 }
1790
1791 /*
1792  * update item key.
1793  *
1794  * This function isn't completely safe. It's the caller's responsibility
1795  * that the new key won't break the order
1796  */
1797 int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
1798                             struct btrfs_root *root, struct btrfs_path *path,
1799                             struct btrfs_key *new_key)
1800 {
1801         struct btrfs_disk_key disk_key;
1802         struct extent_buffer *eb;
1803         int slot;
1804
1805         eb = path->nodes[0];
1806         slot = path->slots[0];
1807         if (slot > 0) {
1808                 btrfs_item_key(eb, &disk_key, slot - 1);
1809                 if (comp_keys(&disk_key, new_key) >= 0)
1810                         return -1;
1811         }
1812         if (slot < btrfs_header_nritems(eb) - 1) {
1813                 btrfs_item_key(eb, &disk_key, slot + 1);
1814                 if (comp_keys(&disk_key, new_key) <= 0)
1815                         return -1;
1816         }
1817
1818         btrfs_cpu_key_to_disk(&disk_key, new_key);
1819         btrfs_set_item_key(eb, &disk_key, slot);
1820         btrfs_mark_buffer_dirty(eb);
1821         if (slot == 0)
1822                 fixup_low_keys(trans, root, path, &disk_key, 1);
1823         return 0;
1824 }
1825
1826 /*
1827  * try to push data from one node into the next node left in the
1828  * tree.
1829  *
1830  * returns 0 if some ptrs were pushed left, < 0 if there was some horrible
1831  * error, and > 0 if there was no room in the left hand block.
1832  */
1833 static int push_node_left(struct btrfs_trans_handle *trans,
1834                           struct btrfs_root *root, struct extent_buffer *dst,
1835                           struct extent_buffer *src, int empty)
1836 {
1837         int push_items = 0;
1838         int src_nritems;
1839         int dst_nritems;
1840         int ret = 0;
1841
1842         src_nritems = btrfs_header_nritems(src);
1843         dst_nritems = btrfs_header_nritems(dst);
1844         push_items = BTRFS_NODEPTRS_PER_BLOCK(root) - dst_nritems;
1845         WARN_ON(btrfs_header_generation(src) != trans->transid);
1846         WARN_ON(btrfs_header_generation(dst) != trans->transid);
1847
1848         if (!empty && src_nritems <= 8)
1849                 return 1;
1850
1851         if (push_items <= 0)
1852                 return 1;
1853
1854         if (empty) {
1855                 push_items = min(src_nritems, push_items);
1856                 if (push_items < src_nritems) {
1857                         /* leave at least 8 pointers in the node if
1858                          * we aren't going to empty it
1859                          */
1860                         if (src_nritems - push_items < 8) {
1861                                 if (push_items <= 8)
1862                                         return 1;
1863                                 push_items -= 8;
1864                         }
1865                 }
1866         } else
1867                 push_items = min(src_nritems - 8, push_items);
1868
1869         copy_extent_buffer(dst, src,
1870                            btrfs_node_key_ptr_offset(dst_nritems),
1871                            btrfs_node_key_ptr_offset(0),
1872                            push_items * sizeof(struct btrfs_key_ptr));
1873
1874         if (push_items < src_nritems) {
1875                 memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0),
1876                                       btrfs_node_key_ptr_offset(push_items),
1877                                       (src_nritems - push_items) *
1878                                       sizeof(struct btrfs_key_ptr));
1879         }
1880         btrfs_set_header_nritems(src, src_nritems - push_items);
1881         btrfs_set_header_nritems(dst, dst_nritems + push_items);
1882         btrfs_mark_buffer_dirty(src);
1883         btrfs_mark_buffer_dirty(dst);
1884
1885         return ret;
1886 }
1887
1888 /*
1889  * try to push data from one node into the next node right in the
1890  * tree.
1891  *
1892  * returns 0 if some ptrs were pushed, < 0 if there was some horrible
1893  * error, and > 0 if there was no room in the right hand block.
1894  *
1895  * this will  only push up to 1/2 the contents of the left node over
1896  */
1897 static int balance_node_right(struct btrfs_trans_handle *trans,
1898                               struct btrfs_root *root,
1899                               struct extent_buffer *dst,
1900                               struct extent_buffer *src)
1901 {
1902         int push_items = 0;
1903         int max_push;
1904         int src_nritems;
1905         int dst_nritems;
1906         int ret = 0;
1907
1908         WARN_ON(btrfs_header_generation(src) != trans->transid);
1909         WARN_ON(btrfs_header_generation(dst) != trans->transid);
1910
1911         src_nritems = btrfs_header_nritems(src);
1912         dst_nritems = btrfs_header_nritems(dst);
1913         push_items = BTRFS_NODEPTRS_PER_BLOCK(root) - dst_nritems;
1914         if (push_items <= 0)
1915                 return 1;
1916
1917         if (src_nritems < 4)
1918                 return 1;
1919
1920         max_push = src_nritems / 2 + 1;
1921         /* don't try to empty the node */
1922         if (max_push >= src_nritems)
1923                 return 1;
1924
1925         if (max_push < push_items)
1926                 push_items = max_push;
1927
1928         memmove_extent_buffer(dst, btrfs_node_key_ptr_offset(push_items),
1929                                       btrfs_node_key_ptr_offset(0),
1930                                       (dst_nritems) *
1931                                       sizeof(struct btrfs_key_ptr));
1932
1933         copy_extent_buffer(dst, src,
1934                            btrfs_node_key_ptr_offset(0),
1935                            btrfs_node_key_ptr_offset(src_nritems - push_items),
1936                            push_items * sizeof(struct btrfs_key_ptr));
1937
1938         btrfs_set_header_nritems(src, src_nritems - push_items);
1939         btrfs_set_header_nritems(dst, dst_nritems + push_items);
1940
1941         btrfs_mark_buffer_dirty(src);
1942         btrfs_mark_buffer_dirty(dst);
1943
1944         return ret;
1945 }
1946
1947 /*
1948  * helper function to insert a new root level in the tree.
1949  * A new node is allocated, and a single item is inserted to
1950  * point to the existing root
1951  *
1952  * returns zero on success or < 0 on failure.
1953  */
1954 static noinline int insert_new_root(struct btrfs_trans_handle *trans,
1955                            struct btrfs_root *root,
1956                            struct btrfs_path *path, int level)
1957 {
1958         u64 lower_gen;
1959         struct extent_buffer *lower;
1960         struct extent_buffer *c;
1961         struct extent_buffer *old;
1962         struct btrfs_disk_key lower_key;
1963
1964         BUG_ON(path->nodes[level]);
1965         BUG_ON(path->nodes[level-1] != root->node);
1966
1967         lower = path->nodes[level-1];
1968         if (level == 1)
1969                 btrfs_item_key(lower, &lower_key, 0);
1970         else
1971                 btrfs_node_key(lower, &lower_key, 0);
1972
1973         c = btrfs_alloc_free_block(trans, root, root->nodesize, 0,
1974                                    root->root_key.objectid, &lower_key,
1975                                    level, root->node->start, 0);
1976         if (IS_ERR(c))
1977                 return PTR_ERR(c);
1978
1979         root_add_used(root, root->nodesize);
1980
1981         memset_extent_buffer(c, 0, 0, sizeof(struct btrfs_header));
1982         btrfs_set_header_nritems(c, 1);
1983         btrfs_set_header_level(c, level);
1984         btrfs_set_header_bytenr(c, c->start);
1985         btrfs_set_header_generation(c, trans->transid);
1986         btrfs_set_header_backref_rev(c, BTRFS_MIXED_BACKREF_REV);
1987         btrfs_set_header_owner(c, root->root_key.objectid);
1988
1989         write_extent_buffer(c, root->fs_info->fsid,
1990                             (unsigned long)btrfs_header_fsid(c),
1991                             BTRFS_FSID_SIZE);
1992
1993         write_extent_buffer(c, root->fs_info->chunk_tree_uuid,
1994                             (unsigned long)btrfs_header_chunk_tree_uuid(c),
1995                             BTRFS_UUID_SIZE);
1996
1997         btrfs_set_node_key(c, &lower_key, 0);
1998         btrfs_set_node_blockptr(c, 0, lower->start);
1999         lower_gen = btrfs_header_generation(lower);
2000         WARN_ON(lower_gen != trans->transid);
2001
2002         btrfs_set_node_ptr_generation(c, 0, lower_gen);
2003
2004         btrfs_mark_buffer_dirty(c);
2005
2006         old = root->node;
2007         rcu_assign_pointer(root->node, c);
2008
2009         /* the super has an extra ref to root->node */
2010         free_extent_buffer(old);
2011
2012         add_root_to_dirty_list(root);
2013         extent_buffer_get(c);
2014         path->nodes[level] = c;
2015         path->locks[level] = 1;
2016         path->slots[level] = 0;
2017         return 0;
2018 }
2019
2020 /*
2021  * worker function to insert a single pointer in a node.
2022  * the node should have enough room for the pointer already
2023  *
2024  * slot and level indicate where you want the key to go, and
2025  * blocknr is the block the key points to.
2026  *
2027  * returns zero on success and < 0 on any error
2028  */
2029 static int insert_ptr(struct btrfs_trans_handle *trans, struct btrfs_root
2030                       *root, struct btrfs_path *path, struct btrfs_disk_key
2031                       *key, u64 bytenr, int slot, int level)
2032 {
2033         struct extent_buffer *lower;
2034         int nritems;
2035
2036         BUG_ON(!path->nodes[level]);
2037         btrfs_assert_tree_locked(path->nodes[level]);
2038         lower = path->nodes[level];
2039         nritems = btrfs_header_nritems(lower);
2040         BUG_ON(slot > nritems);
2041         if (nritems == BTRFS_NODEPTRS_PER_BLOCK(root))
2042                 BUG();
2043         if (slot != nritems) {
2044                 memmove_extent_buffer(lower,
2045                               btrfs_node_key_ptr_offset(slot + 1),
2046                               btrfs_node_key_ptr_offset(slot),
2047                               (nritems - slot) * sizeof(struct btrfs_key_ptr));
2048         }
2049         btrfs_set_node_key(lower, key, slot);
2050         btrfs_set_node_blockptr(lower, slot, bytenr);
2051         WARN_ON(trans->transid == 0);
2052         btrfs_set_node_ptr_generation(lower, slot, trans->transid);
2053         btrfs_set_header_nritems(lower, nritems + 1);
2054         btrfs_mark_buffer_dirty(lower);
2055         return 0;
2056 }
2057
2058 /*
2059  * split the node at the specified level in path in two.
2060  * The path is corrected to point to the appropriate node after the split
2061  *
2062  * Before splitting this tries to make some room in the node by pushing
2063  * left and right, if either one works, it returns right away.
2064  *
2065  * returns 0 on success and < 0 on failure
2066  */
2067 static noinline int split_node(struct btrfs_trans_handle *trans,
2068                                struct btrfs_root *root,
2069                                struct btrfs_path *path, int level)
2070 {
2071         struct extent_buffer *c;
2072         struct extent_buffer *split;
2073         struct btrfs_disk_key disk_key;
2074         int mid;
2075         int ret;
2076         int wret;
2077         u32 c_nritems;
2078
2079         c = path->nodes[level];
2080         WARN_ON(btrfs_header_generation(c) != trans->transid);
2081         if (c == root->node) {
2082                 /* trying to split the root, lets make a new one */
2083                 ret = insert_new_root(trans, root, path, level + 1);
2084                 if (ret)
2085                         return ret;
2086         } else {
2087                 ret = push_nodes_for_insert(trans, root, path, level);
2088                 c = path->nodes[level];
2089                 if (!ret && btrfs_header_nritems(c) <
2090                     BTRFS_NODEPTRS_PER_BLOCK(root) - 3)
2091                         return 0;
2092                 if (ret < 0)
2093                         return ret;
2094         }
2095
2096         c_nritems = btrfs_header_nritems(c);
2097         mid = (c_nritems + 1) / 2;
2098         btrfs_node_key(c, &disk_key, mid);
2099
2100         split = btrfs_alloc_free_block(trans, root, root->nodesize, 0,
2101                                         root->root_key.objectid,
2102                                         &disk_key, level, c->start, 0);
2103         if (IS_ERR(split))
2104                 return PTR_ERR(split);
2105
2106         root_add_used(root, root->nodesize);
2107
2108         memset_extent_buffer(split, 0, 0, sizeof(struct btrfs_header));
2109         btrfs_set_header_level(split, btrfs_header_level(c));
2110         btrfs_set_header_bytenr(split, split->start);
2111         btrfs_set_header_generation(split, trans->transid);
2112         btrfs_set_header_backref_rev(split, BTRFS_MIXED_BACKREF_REV);
2113         btrfs_set_header_owner(split, root->root_key.objectid);
2114         write_extent_buffer(split, root->fs_info->fsid,
2115                             (unsigned long)btrfs_header_fsid(split),
2116                             BTRFS_FSID_SIZE);
2117         write_extent_buffer(split, root->fs_info->chunk_tree_uuid,
2118                             (unsigned long)btrfs_header_chunk_tree_uuid(split),
2119                             BTRFS_UUID_SIZE);
2120
2121
2122         copy_extent_buffer(split, c,
2123                            btrfs_node_key_ptr_offset(0),
2124                            btrfs_node_key_ptr_offset(mid),
2125                            (c_nritems - mid) * sizeof(struct btrfs_key_ptr));
2126         btrfs_set_header_nritems(split, c_nritems - mid);
2127         btrfs_set_header_nritems(c, mid);
2128         ret = 0;
2129
2130         btrfs_mark_buffer_dirty(c);
2131         btrfs_mark_buffer_dirty(split);
2132
2133         wret = insert_ptr(trans, root, path, &disk_key, split->start,
2134                           path->slots[level + 1] + 1,
2135                           level + 1);
2136         if (wret)
2137                 ret = wret;
2138
2139         if (path->slots[level] >= mid) {
2140                 path->slots[level] -= mid;
2141                 btrfs_tree_unlock(c);
2142                 free_extent_buffer(c);
2143                 path->nodes[level] = split;
2144                 path->slots[level + 1] += 1;
2145         } else {
2146                 btrfs_tree_unlock(split);
2147                 free_extent_buffer(split);
2148         }
2149         return ret;
2150 }
2151
2152 /*
2153  * how many bytes are required to store the items in a leaf.  start
2154  * and nr indicate which items in the leaf to check.  This totals up the
2155  * space used both by the item structs and the item data
2156  */
2157 static int leaf_space_used(struct extent_buffer *l, int start, int nr)
2158 {
2159         int data_len;
2160         int nritems = btrfs_header_nritems(l);
2161         int end = min(nritems, start + nr) - 1;
2162
2163         if (!nr)
2164                 return 0;
2165         data_len = btrfs_item_end_nr(l, start);
2166         data_len = data_len - btrfs_item_offset_nr(l, end);
2167         data_len += sizeof(struct btrfs_item) * nr;
2168         WARN_ON(data_len < 0);
2169         return data_len;
2170 }
2171
2172 /*
2173  * The space between the end of the leaf items and
2174  * the start of the leaf data.  IOW, how much room
2175  * the leaf has left for both items and data
2176  */
2177 noinline int btrfs_leaf_free_space(struct btrfs_root *root,
2178                                    struct extent_buffer *leaf)
2179 {
2180         int nritems = btrfs_header_nritems(leaf);
2181         int ret;
2182         ret = BTRFS_LEAF_DATA_SIZE(root) - leaf_space_used(leaf, 0, nritems);
2183         if (ret < 0) {
2184                 printk(KERN_CRIT "leaf free space ret %d, leaf data size %lu, "
2185                        "used %d nritems %d\n",
2186                        ret, (unsigned long) BTRFS_LEAF_DATA_SIZE(root),
2187                        leaf_space_used(leaf, 0, nritems), nritems);
2188         }
2189         return ret;
2190 }
2191
2192 /*
2193  * min slot controls the lowest index we're willing to push to the
2194  * right.  We'll push up to and including min_slot, but no lower
2195  */
2196 static noinline int __push_leaf_right(struct btrfs_trans_handle *trans,
2197                                       struct btrfs_root *root,
2198                                       struct btrfs_path *path,
2199                                       int data_size, int empty,
2200                                       struct extent_buffer *right,
2201                                       int free_space, u32 left_nritems,
2202                                       u32 min_slot)
2203 {
2204         struct extent_buffer *left = path->nodes[0];
2205         struct extent_buffer *upper = path->nodes[1];
2206         struct btrfs_disk_key disk_key;
2207         int slot;
2208         u32 i;
2209         int push_space = 0;
2210         int push_items = 0;
2211         struct btrfs_item *item;
2212         u32 nr;
2213         u32 right_nritems;
2214         u32 data_end;
2215         u32 this_item_size;
2216
2217         if (empty)
2218                 nr = 0;
2219         else
2220                 nr = max_t(u32, 1, min_slot);
2221
2222         if (path->slots[0] >= left_nritems)
2223                 push_space += data_size;
2224
2225         slot = path->slots[1];
2226         i = left_nritems - 1;
2227         while (i >= nr) {
2228                 item = btrfs_item_nr(left, i);
2229
2230                 if (!empty && push_items > 0) {
2231                         if (path->slots[0] > i)
2232                                 break;
2233                         if (path->slots[0] == i) {
2234                                 int space = btrfs_leaf_free_space(root, left);
2235                                 if (space + push_space * 2 > free_space)
2236                                         break;
2237                         }
2238                 }
2239
2240                 if (path->slots[0] == i)
2241                         push_space += data_size;
2242
2243                 if (!left->map_token) {
2244                         map_extent_buffer(left, (unsigned long)item,
2245                                         sizeof(struct btrfs_item),
2246                                         &left->map_token, &left->kaddr,
2247                                         &left->map_start, &left->map_len,
2248                                         KM_USER1);
2249                 }
2250
2251                 this_item_size = btrfs_item_size(left, item);
2252                 if (this_item_size + sizeof(*item) + push_space > free_space)
2253                         break;
2254
2255                 push_items++;
2256                 push_space += this_item_size + sizeof(*item);
2257                 if (i == 0)
2258                         break;
2259                 i--;
2260         }
2261         if (left->map_token) {
2262                 unmap_extent_buffer(left, left->map_token, KM_USER1);
2263                 left->map_token = NULL;
2264         }
2265
2266         if (push_items == 0)
2267                 goto out_unlock;
2268
2269         if (!empty && push_items == left_nritems)
2270                 WARN_ON(1);
2271
2272         /* push left to right */
2273         right_nritems = btrfs_header_nritems(right);
2274
2275         push_space = btrfs_item_end_nr(left, left_nritems - push_items);
2276         push_space -= leaf_data_end(root, left);
2277
2278         /* make room in the right data area */
2279         data_end = leaf_data_end(root, right);
2280         memmove_extent_buffer(right,
2281                               btrfs_leaf_data(right) + data_end - push_space,
2282                               btrfs_leaf_data(right) + data_end,
2283                               BTRFS_LEAF_DATA_SIZE(root) - data_end);
2284
2285         /* copy from the left data area */
2286         copy_extent_buffer(right, left, btrfs_leaf_data(right) +
2287                      BTRFS_LEAF_DATA_SIZE(root) - push_space,
2288                      btrfs_leaf_data(left) + leaf_data_end(root, left),
2289                      push_space);
2290
2291         memmove_extent_buffer(right, btrfs_item_nr_offset(push_items),
2292                               btrfs_item_nr_offset(0),
2293                               right_nritems * sizeof(struct btrfs_item));
2294
2295         /* copy the items from left to right */
2296         copy_extent_buffer(right, left, btrfs_item_nr_offset(0),
2297                    btrfs_item_nr_offset(left_nritems - push_items),
2298                    push_items * sizeof(struct btrfs_item));
2299
2300         /* update the item pointers */
2301         right_nritems += push_items;
2302         btrfs_set_header_nritems(right, right_nritems);
2303         push_space = BTRFS_LEAF_DATA_SIZE(root);
2304         for (i = 0; i < right_nritems; i++) {
2305                 item = btrfs_item_nr(right, i);
2306                 if (!right->map_token) {
2307                         map_extent_buffer(right, (unsigned long)item,
2308                                         sizeof(struct btrfs_item),
2309                                         &right->map_token, &right->kaddr,
2310                                         &right->map_start, &right->map_len,
2311                                         KM_USER1);
2312                 }
2313                 push_space -= btrfs_item_size(right, item);
2314                 btrfs_set_item_offset(right, item, push_space);
2315         }
2316
2317         if (right->map_token) {
2318                 unmap_extent_buffer(right, right->map_token, KM_USER1);
2319                 right->map_token = NULL;
2320         }
2321         left_nritems -= push_items;
2322         btrfs_set_header_nritems(left, left_nritems);
2323
2324         if (left_nritems)
2325                 btrfs_mark_buffer_dirty(left);
2326         else
2327                 clean_tree_block(trans, root, left);
2328
2329         btrfs_mark_buffer_dirty(right);
2330
2331         btrfs_item_key(right, &disk_key, 0);
2332         btrfs_set_node_key(upper, &disk_key, slot + 1);
2333         btrfs_mark_buffer_dirty(upper);
2334
2335         /* then fixup the leaf pointer in the path */
2336         if (path->slots[0] >= left_nritems) {
2337                 path->slots[0] -= left_nritems;
2338                 if (btrfs_header_nritems(path->nodes[0]) == 0)
2339                         clean_tree_block(trans, root, path->nodes[0]);
2340                 btrfs_tree_unlock(path->nodes[0]);
2341                 free_extent_buffer(path->nodes[0]);
2342                 path->nodes[0] = right;
2343                 path->slots[1] += 1;
2344         } else {
2345                 btrfs_tree_unlock(right);
2346                 free_extent_buffer(right);
2347         }
2348         return 0;
2349
2350 out_unlock:
2351         btrfs_tree_unlock(right);
2352         free_extent_buffer(right);
2353         return 1;
2354 }
2355
2356 /*
2357  * push some data in the path leaf to the right, trying to free up at
2358  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
2359  *
2360  * returns 1 if the push failed because the other node didn't have enough
2361  * room, 0 if everything worked out and < 0 if there were major errors.
2362  *
2363  * this will push starting from min_slot to the end of the leaf.  It won't
2364  * push any slot lower than min_slot
2365  */
2366 static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
2367                            *root, struct btrfs_path *path,
2368                            int min_data_size, int data_size,
2369                            int empty, u32 min_slot)
2370 {
2371         struct extent_buffer *left = path->nodes[0];
2372         struct extent_buffer *right;
2373         struct extent_buffer *upper;
2374         int slot;
2375         int free_space;
2376         u32 left_nritems;
2377         int ret;
2378
2379         if (!path->nodes[1])
2380                 return 1;
2381
2382         slot = path->slots[1];
2383         upper = path->nodes[1];
2384         if (slot >= btrfs_header_nritems(upper) - 1)
2385                 return 1;
2386
2387         btrfs_assert_tree_locked(path->nodes[1]);
2388
2389         right = read_node_slot(root, upper, slot + 1);
2390         if (right == NULL)
2391                 return 1;
2392
2393         btrfs_tree_lock(right);
2394         btrfs_set_lock_blocking(right);
2395
2396         free_space = btrfs_leaf_free_space(root, right);
2397         if (free_space < data_size)
2398                 goto out_unlock;
2399
2400         /* cow and double check */
2401         ret = btrfs_cow_block(trans, root, right, upper,
2402                               slot + 1, &right);
2403         if (ret)
2404                 goto out_unlock;
2405
2406         free_space = btrfs_leaf_free_space(root, right);
2407         if (free_space < data_size)
2408                 goto out_unlock;
2409
2410         left_nritems = btrfs_header_nritems(left);
2411         if (left_nritems == 0)
2412                 goto out_unlock;
2413
2414         return __push_leaf_right(trans, root, path, min_data_size, empty,
2415                                 right, free_space, left_nritems, min_slot);
2416 out_unlock:
2417         btrfs_tree_unlock(right);
2418         free_extent_buffer(right);
2419         return 1;
2420 }
2421
2422 /*
2423  * push some data in the path leaf to the left, trying to free up at
2424  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
2425  *
2426  * max_slot can put a limit on how far into the leaf we'll push items.  The
2427  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us do all the
2428  * items
2429  */
2430 static noinline int __push_leaf_left(struct btrfs_trans_handle *trans,
2431                                      struct btrfs_root *root,
2432                                      struct btrfs_path *path, int data_size,
2433                                      int empty, struct extent_buffer *left,
2434                                      int free_space, u32 right_nritems,
2435                                      u32 max_slot)
2436 {
2437         struct btrfs_disk_key disk_key;
2438         struct extent_buffer *right = path->nodes[0];
2439         int i;
2440         int push_space = 0;
2441         int push_items = 0;
2442         struct btrfs_item *item;
2443         u32 old_left_nritems;
2444         u32 nr;
2445         int ret = 0;
2446         int wret;
2447         u32 this_item_size;
2448         u32 old_left_item_size;
2449
2450         if (empty)
2451                 nr = min(right_nritems, max_slot);
2452         else
2453                 nr = min(right_nritems - 1, max_slot);
2454
2455         for (i = 0; i < nr; i++) {
2456                 item = btrfs_item_nr(right, i);
2457                 if (!right->map_token) {
2458                         map_extent_buffer(right, (unsigned long)item,
2459                                         sizeof(struct btrfs_item),
2460                                         &right->map_token, &right->kaddr,
2461                                         &right->map_start, &right->map_len,
2462                                         KM_USER1);
2463                 }
2464
2465                 if (!empty && push_items > 0) {
2466                         if (path->slots[0] < i)
2467                                 break;
2468                         if (path->slots[0] == i) {
2469                                 int space = btrfs_leaf_free_space(root, right);
2470                                 if (space + push_space * 2 > free_space)
2471                                         break;
2472                         }
2473                 }
2474
2475                 if (path->slots[0] == i)
2476                         push_space += data_size;
2477
2478                 this_item_size = btrfs_item_size(right, item);
2479                 if (this_item_size + sizeof(*item) + push_space > free_space)
2480                         break;
2481
2482                 push_items++;
2483                 push_space += this_item_size + sizeof(*item);
2484         }
2485
2486         if (right->map_token) {
2487                 unmap_extent_buffer(right, right->map_token, KM_USER1);
2488                 right->map_token = NULL;
2489         }
2490
2491         if (push_items == 0) {
2492                 ret = 1;
2493                 goto out;
2494         }
2495         if (!empty && push_items == btrfs_header_nritems(right))
2496                 WARN_ON(1);
2497
2498         /* push data from right to left */
2499         copy_extent_buffer(left, right,
2500                            btrfs_item_nr_offset(btrfs_header_nritems(left)),
2501                            btrfs_item_nr_offset(0),
2502                            push_items * sizeof(struct btrfs_item));
2503
2504         push_space = BTRFS_LEAF_DATA_SIZE(root) -
2505                      btrfs_item_offset_nr(right, push_items - 1);
2506
2507         copy_extent_buffer(left, right, btrfs_leaf_data(left) +
2508                      leaf_data_end(root, left) - push_space,
2509                      btrfs_leaf_data(right) +
2510                      btrfs_item_offset_nr(right, push_items - 1),
2511                      push_space);
2512         old_left_nritems = btrfs_header_nritems(left);
2513         BUG_ON(old_left_nritems <= 0);
2514
2515         old_left_item_size = btrfs_item_offset_nr(left, old_left_nritems - 1);
2516         for (i = old_left_nritems; i < old_left_nritems + push_items; i++) {
2517                 u32 ioff;
2518
2519                 item = btrfs_item_nr(left, i);
2520                 if (!left->map_token) {
2521                         map_extent_buffer(left, (unsigned long)item,
2522                                         sizeof(struct btrfs_item),
2523                                         &left->map_token, &left->kaddr,
2524                                         &left->map_start, &left->map_len,
2525                                         KM_USER1);
2526                 }
2527
2528                 ioff = btrfs_item_offset(left, item);
2529                 btrfs_set_item_offset(left, item,
2530                       ioff - (BTRFS_LEAF_DATA_SIZE(root) - old_left_item_size));
2531         }
2532         btrfs_set_header_nritems(left, old_left_nritems + push_items);
2533         if (left->map_token) {
2534                 unmap_extent_buffer(left, left->map_token, KM_USER1);
2535                 left->map_token = NULL;
2536         }
2537
2538         /* fixup right node */
2539         if (push_items > right_nritems) {
2540                 printk(KERN_CRIT "push items %d nr %u\n", push_items,
2541                        right_nritems);
2542                 WARN_ON(1);
2543         }
2544
2545         if (push_items < right_nritems) {
2546                 push_space = btrfs_item_offset_nr(right, push_items - 1) -
2547                                                   leaf_data_end(root, right);
2548                 memmove_extent_buffer(right, btrfs_leaf_data(right) +
2549                                       BTRFS_LEAF_DATA_SIZE(root) - push_space,
2550                                       btrfs_leaf_data(right) +
2551                                       leaf_data_end(root, right), push_space);
2552
2553                 memmove_extent_buffer(right, btrfs_item_nr_offset(0),
2554                               btrfs_item_nr_offset(push_items),
2555                              (btrfs_header_nritems(right) - push_items) *
2556                              sizeof(struct btrfs_item));
2557         }
2558         right_nritems -= push_items;
2559         btrfs_set_header_nritems(right, right_nritems);
2560         push_space = BTRFS_LEAF_DATA_SIZE(root);
2561         for (i = 0; i < right_nritems; i++) {
2562                 item = btrfs_item_nr(right, i);
2563
2564                 if (!right->map_token) {
2565                         map_extent_buffer(right, (unsigned long)item,
2566                                         sizeof(struct btrfs_item),
2567                                         &right->map_token, &right->kaddr,
2568                                         &right->map_start, &right->map_len,
2569                                         KM_USER1);
2570                 }
2571
2572                 push_space = push_space - btrfs_item_size(right, item);
2573                 btrfs_set_item_offset(right, item, push_space);
2574         }
2575         if (right->map_token) {
2576                 unmap_extent_buffer(right, right->map_token, KM_USER1);
2577                 right->map_token = NULL;
2578         }
2579
2580         btrfs_mark_buffer_dirty(left);
2581         if (right_nritems)
2582                 btrfs_mark_buffer_dirty(right);
2583         else
2584                 clean_tree_block(trans, root, right);
2585
2586         btrfs_item_key(right, &disk_key, 0);
2587         wret = fixup_low_keys(trans, root, path, &disk_key, 1);
2588         if (wret)
2589                 ret = wret;
2590
2591         /* then fixup the leaf pointer in the path */
2592         if (path->slots[0] < push_items) {
2593                 path->slots[0] += old_left_nritems;
2594                 btrfs_tree_unlock(path->nodes[0]);
2595                 free_extent_buffer(path->nodes[0]);
2596                 path->nodes[0] = left;
2597                 path->slots[1] -= 1;
2598         } else {
2599                 btrfs_tree_unlock(left);
2600                 free_extent_buffer(left);
2601                 path->slots[0] -= push_items;
2602         }
2603         BUG_ON(path->slots[0] < 0);
2604         return ret;
2605 out:
2606         btrfs_tree_unlock(left);
2607         free_extent_buffer(left);
2608         return ret;
2609 }
2610
2611 /*
2612  * push some data in the path leaf to the left, trying to free up at
2613  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
2614  *
2615  * max_slot can put a limit on how far into the leaf we'll push items.  The
2616  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us push all the
2617  * items
2618  */
2619 static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
2620                           *root, struct btrfs_path *path, int min_data_size,
2621                           int data_size, int empty, u32 max_slot)
2622 {
2623         struct extent_buffer *right = path->nodes[0];
2624         struct extent_buffer *left;
2625         int slot;
2626         int free_space;
2627         u32 right_nritems;
2628         int ret = 0;
2629
2630         slot = path->slots[1];
2631         if (slot == 0)
2632                 return 1;
2633         if (!path->nodes[1])
2634                 return 1;
2635
2636         right_nritems = btrfs_header_nritems(right);
2637         if (right_nritems == 0)
2638                 return 1;
2639
2640         btrfs_assert_tree_locked(path->nodes[1]);
2641
2642         left = read_node_slot(root, path->nodes[1], slot - 1);
2643         if (left == NULL)
2644                 return 1;
2645
2646         btrfs_tree_lock(left);
2647         btrfs_set_lock_blocking(left);
2648
2649         free_space = btrfs_leaf_free_space(root, left);
2650         if (free_space < data_size) {
2651                 ret = 1;
2652                 goto out;
2653         }
2654
2655         /* cow and double check */
2656         ret = btrfs_cow_block(trans, root, left,
2657                               path->nodes[1], slot - 1, &left);
2658         if (ret) {
2659                 /* we hit -ENOSPC, but it isn't fatal here */
2660                 ret = 1;
2661                 goto out;
2662         }
2663
2664         free_space = btrfs_leaf_free_space(root, left);
2665         if (free_space < data_size) {
2666                 ret = 1;
2667                 goto out;
2668         }
2669
2670         return __push_leaf_left(trans, root, path, min_data_size,
2671                                empty, left, free_space, right_nritems,
2672                                max_slot);
2673 out:
2674         btrfs_tree_unlock(left);
2675         free_extent_buffer(left);
2676         return ret;
2677 }
2678
2679 /*
2680  * split the path's leaf in two, making sure there is at least data_size
2681  * available for the resulting leaf level of the path.
2682  *
2683  * returns 0 if all went well and < 0 on failure.
2684  */
2685 static noinline int copy_for_split(struct btrfs_trans_handle *trans,
2686                                struct btrfs_root *root,
2687                                struct btrfs_path *path,
2688                                struct extent_buffer *l,
2689                                struct extent_buffer *right,
2690                                int slot, int mid, int nritems)
2691 {
2692         int data_copy_size;
2693         int rt_data_off;
2694         int i;
2695         int ret = 0;
2696         int wret;
2697         struct btrfs_disk_key disk_key;
2698
2699         nritems = nritems - mid;
2700         btrfs_set_header_nritems(right, nritems);
2701         data_copy_size = btrfs_item_end_nr(l, mid) - leaf_data_end(root, l);
2702
2703         copy_extent_buffer(right, l, btrfs_item_nr_offset(0),
2704                            btrfs_item_nr_offset(mid),
2705                            nritems * sizeof(struct btrfs_item));
2706
2707         copy_extent_buffer(right, l,
2708                      btrfs_leaf_data(right) + BTRFS_LEAF_DATA_SIZE(root) -
2709                      data_copy_size, btrfs_leaf_data(l) +
2710                      leaf_data_end(root, l), data_copy_size);
2711
2712         rt_data_off = BTRFS_LEAF_DATA_SIZE(root) -
2713                       btrfs_item_end_nr(l, mid);
2714
2715         for (i = 0; i < nritems; i++) {
2716                 struct btrfs_item *item = btrfs_item_nr(right, i);
2717                 u32 ioff;
2718
2719                 if (!right->map_token) {
2720                         map_extent_buffer(right, (unsigned long)item,
2721                                         sizeof(struct btrfs_item),
2722                                         &right->map_token, &right->kaddr,
2723                                         &right->map_start, &right->map_len,
2724                                         KM_USER1);
2725                 }
2726
2727                 ioff = btrfs_item_offset(right, item);
2728                 btrfs_set_item_offset(right, item, ioff + rt_data_off);
2729         }
2730
2731         if (right->map_token) {
2732                 unmap_extent_buffer(right, right->map_token, KM_USER1);
2733                 right->map_token = NULL;
2734         }
2735
2736         btrfs_set_header_nritems(l, mid);
2737         ret = 0;
2738         btrfs_item_key(right, &disk_key, 0);
2739         wret = insert_ptr(trans, root, path, &disk_key, right->start,
2740                           path->slots[1] + 1, 1);
2741         if (wret)
2742                 ret = wret;
2743
2744         btrfs_mark_buffer_dirty(right);
2745         btrfs_mark_buffer_dirty(l);
2746         BUG_ON(path->slots[0] != slot);
2747
2748         if (mid <= slot) {
2749                 btrfs_tree_unlock(path->nodes[0]);
2750                 free_extent_buffer(path->nodes[0]);
2751                 path->nodes[0] = right;
2752                 path->slots[0] -= mid;
2753                 path->slots[1] += 1;
2754         } else {
2755                 btrfs_tree_unlock(right);
2756                 free_extent_buffer(right);
2757         }
2758
2759         BUG_ON(path->slots[0] < 0);
2760
2761         return ret;
2762 }
2763
2764 /*
2765  * double splits happen when we need to insert a big item in the middle
2766  * of a leaf.  A double split can leave us with 3 mostly empty leaves:
2767  * leaf: [ slots 0 - N] [ our target ] [ N + 1 - total in leaf ]
2768  *          A                 B                 C
2769  *
2770  * We avoid this by trying to push the items on either side of our target
2771  * into the adjacent leaves.  If all goes well we can avoid the double split
2772  * completely.
2773  */
2774 static noinline int push_for_double_split(struct btrfs_trans_handle *trans,
2775                                           struct btrfs_root *root,
2776                                           struct btrfs_path *path,
2777                                           int data_size)
2778 {
2779         int ret;
2780         int progress = 0;
2781         int slot;
2782         u32 nritems;
2783
2784         slot = path->slots[0];
2785
2786         /*
2787          * try to push all the items after our slot into the
2788          * right leaf
2789          */
2790         ret = push_leaf_right(trans, root, path, 1, data_size, 0, slot);
2791         if (ret < 0)
2792                 return ret;
2793
2794         if (ret == 0)
2795                 progress++;
2796
2797         nritems = btrfs_header_nritems(path->nodes[0]);
2798         /*
2799          * our goal is to get our slot at the start or end of a leaf.  If
2800          * we've done so we're done
2801          */
2802         if (path->slots[0] == 0 || path->slots[0] == nritems)
2803                 return 0;
2804
2805         if (btrfs_leaf_free_space(root, path->nodes[0]) >= data_size)
2806                 return 0;
2807
2808         /* try to push all the items before our slot into the next leaf */
2809         slot = path->slots[0];
2810         ret = push_leaf_left(trans, root, path, 1, data_size, 0, slot);
2811         if (ret < 0)
2812                 return ret;
2813
2814         if (ret == 0)
2815                 progress++;
2816
2817         if (progress)
2818                 return 0;
2819         return 1;
2820 }
2821
2822 /*
2823  * split the path's leaf in two, making sure there is at least data_size
2824  * available for the resulting leaf level of the path.
2825  *
2826  * returns 0 if all went well and < 0 on failure.
2827  */
2828 static noinline int split_leaf(struct btrfs_trans_handle *trans,
2829                                struct btrfs_root *root,
2830                                struct btrfs_key *ins_key,
2831                                struct btrfs_path *path, int data_size,
2832                                int extend)
2833 {
2834         struct btrfs_disk_key disk_key;
2835         struct extent_buffer *l;
2836         u32 nritems;
2837         int mid;
2838         int slot;
2839         struct extent_buffer *right;
2840         int ret = 0;
2841         int wret;
2842         int split;
2843         int num_doubles = 0;
2844         int tried_avoid_double = 0;
2845
2846         l = path->nodes[0];
2847         slot = path->slots[0];
2848         if (extend && data_size + btrfs_item_size_nr(l, slot) +
2849             sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(root))
2850                 return -EOVERFLOW;
2851
2852         /* first try to make some room by pushing left and right */
2853         if (data_size) {
2854                 wret = push_leaf_right(trans, root, path, data_size,
2855                                        data_size, 0, 0);
2856                 if (wret < 0)
2857                         return wret;
2858                 if (wret) {
2859                         wret = push_leaf_left(trans, root, path, data_size,
2860                                               data_size, 0, (u32)-1);
2861                         if (wret < 0)
2862                                 return wret;
2863                 }
2864                 l = path->nodes[0];
2865
2866                 /* did the pushes work? */
2867                 if (btrfs_leaf_free_space(root, l) >= data_size)
2868                         return 0;
2869         }
2870
2871         if (!path->nodes[1]) {
2872                 ret = insert_new_root(trans, root, path, 1);
2873                 if (ret)
2874                         return ret;
2875         }
2876 again:
2877         split = 1;
2878         l = path->nodes[0];
2879         slot = path->slots[0];
2880         nritems = btrfs_header_nritems(l);
2881         mid = (nritems + 1) / 2;
2882
2883         if (mid <= slot) {
2884                 if (nritems == 1 ||
2885                     leaf_space_used(l, mid, nritems - mid) + data_size >
2886                         BTRFS_LEAF_DATA_SIZE(root)) {
2887                         if (slot >= nritems) {
2888                                 split = 0;
2889                         } else {
2890                                 mid = slot;
2891                                 if (mid != nritems &&
2892                                     leaf_space_used(l, mid, nritems - mid) +
2893                                     data_size > BTRFS_LEAF_DATA_SIZE(root)) {
2894                                         if (data_size && !tried_avoid_double)
2895                                                 goto push_for_double;
2896                                         split = 2;
2897                                 }
2898                         }
2899                 }
2900         } else {
2901                 if (leaf_space_used(l, 0, mid) + data_size >
2902                         BTRFS_LEAF_DATA_SIZE(root)) {
2903                         if (!extend && data_size && slot == 0) {
2904                                 split = 0;
2905                         } else if ((extend || !data_size) && slot == 0) {
2906                                 mid = 1;
2907                         } else {
2908                                 mid = slot;
2909                                 if (mid != nritems &&
2910                                     leaf_space_used(l, mid, nritems - mid) +
2911                                     data_size > BTRFS_LEAF_DATA_SIZE(root)) {
2912                                         if (data_size && !tried_avoid_double)
2913                                                 goto push_for_double;
2914                                         split = 2 ;
2915                                 }
2916                         }
2917                 }
2918         }
2919
2920         if (split == 0)
2921                 btrfs_cpu_key_to_disk(&disk_key, ins_key);
2922         else
2923                 btrfs_item_key(l, &disk_key, mid);
2924
2925         right = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
2926                                         root->root_key.objectid,
2927                                         &disk_key, 0, l->start, 0);
2928         if (IS_ERR(right))
2929                 return PTR_ERR(right);
2930
2931         root_add_used(root, root->leafsize);
2932
2933         memset_extent_buffer(right, 0, 0, sizeof(struct btrfs_header));
2934         btrfs_set_header_bytenr(right, right->start);
2935         btrfs_set_header_generation(right, trans->transid);
2936         btrfs_set_header_backref_rev(right, BTRFS_MIXED_BACKREF_REV);
2937         btrfs_set_header_owner(right, root->root_key.objectid);
2938         btrfs_set_header_level(right, 0);
2939         write_extent_buffer(right, root->fs_info->fsid,
2940                             (unsigned long)btrfs_header_fsid(right),
2941                             BTRFS_FSID_SIZE);
2942
2943         write_extent_buffer(right, root->fs_info->chunk_tree_uuid,
2944                             (unsigned long)btrfs_header_chunk_tree_uuid(right),
2945                             BTRFS_UUID_SIZE);
2946
2947         if (split == 0) {
2948                 if (mid <= slot) {
2949                         btrfs_set_header_nritems(right, 0);
2950                         wret = insert_ptr(trans, root, path,
2951                                           &disk_key, right->start,
2952                                           path->slots[1] + 1, 1);
2953                         if (wret)
2954                                 ret = wret;
2955
2956                         btrfs_tree_unlock(path->nodes[0]);
2957                         free_extent_buffer(path->nodes[0]);
2958                         path->nodes[0] = right;
2959                         path->slots[0] = 0;
2960                         path->slots[1] += 1;
2961                 } else {
2962                         btrfs_set_header_nritems(right, 0);
2963                         wret = insert_ptr(trans, root, path,
2964                                           &disk_key,
2965                                           right->start,
2966                                           path->slots[1], 1);
2967                         if (wret)
2968                                 ret = wret;
2969                         btrfs_tree_unlock(path->nodes[0]);
2970                         free_extent_buffer(path->nodes[0]);
2971                         path->nodes[0] = right;
2972                         path->slots[0] = 0;
2973                         if (path->slots[1] == 0) {
2974                                 wret = fixup_low_keys(trans, root,
2975                                                 path, &disk_key, 1);
2976                                 if (wret)
2977                                         ret = wret;
2978                         }
2979                 }
2980                 btrfs_mark_buffer_dirty(right);
2981                 return ret;
2982         }
2983
2984         ret = copy_for_split(trans, root, path, l, right, slot, mid, nritems);
2985         BUG_ON(ret);
2986
2987         if (split == 2) {
2988                 BUG_ON(num_doubles != 0);
2989                 num_doubles++;
2990                 goto again;
2991         }
2992
2993         return ret;
2994
2995 push_for_double:
2996         push_for_double_split(trans, root, path, data_size);
2997         tried_avoid_double = 1;
2998         if (btrfs_leaf_free_space(root, path->nodes[0]) >= data_size)
2999                 return 0;
3000         goto again;
3001 }
3002
3003 static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
3004                                          struct btrfs_root *root,
3005                                          struct btrfs_path *path, int ins_len)
3006 {
3007         struct btrfs_key key;
3008         struct extent_buffer *leaf;
3009         struct btrfs_file_extent_item *fi;
3010         u64 extent_len = 0;
3011         u32 item_size;
3012         int ret;
3013
3014         leaf = path->nodes[0];
3015         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
3016
3017         BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY &&
3018                key.type != BTRFS_EXTENT_CSUM_KEY);
3019
3020         if (btrfs_leaf_free_space(root, leaf) >= ins_len)
3021                 return 0;
3022
3023         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
3024         if (key.type == BTRFS_EXTENT_DATA_KEY) {
3025                 fi = btrfs_item_ptr(leaf, path->slots[0],
3026                                     struct btrfs_file_extent_item);
3027                 extent_len = btrfs_file_extent_num_bytes(leaf, fi);
3028         }
3029         btrfs_release_path(root, path);
3030
3031         path->keep_locks = 1;
3032         path->search_for_split = 1;
3033         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
3034         path->search_for_split = 0;
3035         if (ret < 0)
3036                 goto err;
3037
3038         ret = -EAGAIN;
3039         leaf = path->nodes[0];
3040         /* if our item isn't there or got smaller, return now */
3041         if (ret > 0 || item_size != btrfs_item_size_nr(leaf, path->slots[0]))
3042                 goto err;
3043
3044         /* the leaf has  changed, it now has room.  return now */
3045         if (btrfs_leaf_free_space(root, path->nodes[0]) >= ins_len)
3046                 goto err;
3047
3048         if (key.type == BTRFS_EXTENT_DATA_KEY) {
3049                 fi = btrfs_item_ptr(leaf, path->slots[0],
3050                                     struct btrfs_file_extent_item);
3051                 if (extent_len != btrfs_file_extent_num_bytes(leaf, fi))
3052                         goto err;
3053         }
3054
3055         btrfs_set_path_blocking(path);
3056         ret = split_leaf(trans, root, &key, path, ins_len, 1);
3057         if (ret)
3058                 goto err;
3059
3060         path->keep_locks = 0;
3061         btrfs_unlock_up_safe(path, 1);
3062         return 0;
3063 err:
3064         path->keep_locks = 0;
3065         return ret;
3066 }
3067
3068 static noinline int split_item(struct btrfs_trans_handle *trans,
3069                                struct btrfs_root *root,
3070                                struct btrfs_path *path,
3071                                struct btrfs_key *new_key,
3072                                unsigned long split_offset)
3073 {
3074         struct extent_buffer *leaf;
3075         struct btrfs_item *item;
3076         struct btrfs_item *new_item;
3077         int slot;
3078         char *buf;
3079         u32 nritems;
3080         u32 item_size;
3081         u32 orig_offset;
3082         struct btrfs_disk_key disk_key;
3083
3084         leaf = path->nodes[0];
3085         BUG_ON(btrfs_leaf_free_space(root, leaf) < sizeof(struct btrfs_item));
3086
3087         btrfs_set_path_blocking(path);
3088
3089         item = btrfs_item_nr(leaf, path->slots[0]);
3090         orig_offset = btrfs_item_offset(leaf, item);
3091         item_size = btrfs_item_size(leaf, item);
3092
3093         buf = kmalloc(item_size, GFP_NOFS);
3094         if (!buf)
3095                 return -ENOMEM;
3096
3097         read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf,
3098                             path->slots[0]), item_size);
3099
3100         slot = path->slots[0] + 1;
3101         nritems = btrfs_header_nritems(leaf);
3102         if (slot != nritems) {
3103                 /* shift the items */
3104                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1),
3105                                 btrfs_item_nr_offset(slot),
3106                                 (nritems - slot) * sizeof(struct btrfs_item));
3107         }
3108
3109         btrfs_cpu_key_to_disk(&disk_key, new_key);
3110         btrfs_set_item_key(leaf, &disk_key, slot);
3111
3112         new_item = btrfs_item_nr(leaf, slot);
3113
3114         btrfs_set_item_offset(leaf, new_item, orig_offset);
3115         btrfs_set_item_size(leaf, new_item, item_size - split_offset);
3116
3117         btrfs_set_item_offset(leaf, item,
3118                               orig_offset + item_size - split_offset);
3119         btrfs_set_item_size(leaf, item, split_offset);
3120
3121         btrfs_set_header_nritems(leaf, nritems + 1);
3122
3123         /* write the data for the start of the original item */
3124         write_extent_buffer(leaf, buf,
3125                             btrfs_item_ptr_offset(leaf, path->slots[0]),
3126                             split_offset);
3127
3128         /* write the data for the new item */
3129         write_extent_buffer(leaf, buf + split_offset,
3130                             btrfs_item_ptr_offset(leaf, slot),
3131                             item_size - split_offset);
3132         btrfs_mark_buffer_dirty(leaf);
3133
3134         BUG_ON(btrfs_leaf_free_space(root, leaf) < 0);
3135         kfree(buf);
3136         return 0;
3137 }
3138
3139 /*
3140  * This function splits a single item into two items,
3141  * giving 'new_key' to the new item and splitting the
3142  * old one at split_offset (from the start of the item).
3143  *
3144  * The path may be released by this operation.  After
3145  * the split, the path is pointing to the old item.  The
3146  * new item is going to be in the same node as the old one.
3147  *
3148  * Note, the item being split must be smaller enough to live alone on
3149  * a tree block with room for one extra struct btrfs_item
3150  *
3151  * This allows us to split the item in place, keeping a lock on the
3152  * leaf the entire time.
3153  */
3154 int btrfs_split_item(struct btrfs_trans_handle *trans,
3155                      struct btrfs_root *root,
3156                      struct btrfs_path *path,
3157                      struct btrfs_key *new_key,
3158                      unsigned long split_offset)
3159 {
3160         int ret;
3161         ret = setup_leaf_for_split(trans, root, path,
3162                                    sizeof(struct btrfs_item));
3163         if (ret)
3164                 return ret;
3165
3166         ret = split_item(trans, root, path, new_key, split_offset);
3167         return ret;
3168 }
3169
3170 /*
3171  * This function duplicate a item, giving 'new_key' to the new item.
3172  * It guarantees both items live in the same tree leaf and the new item
3173  * is contiguous with the original item.
3174  *
3175  * This allows us to split file extent in place, keeping a lock on the
3176  * leaf the entire time.
3177  */
3178 int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
3179                          struct btrfs_root *root,
3180                          struct btrfs_path *path,
3181                          struct btrfs_key *new_key)
3182 {
3183         struct extent_buffer *leaf;
3184         int ret;
3185         u32 item_size;
3186
3187         leaf = path->nodes[0];
3188         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
3189         ret = setup_leaf_for_split(trans, root, path,
3190                                    item_size + sizeof(struct btrfs_item));
3191         if (ret)
3192                 return ret;
3193
3194         path->slots[0]++;
3195         ret = setup_items_for_insert(trans, root, path, new_key, &item_size,
3196                                      item_size, item_size +
3197                                      sizeof(struct btrfs_item), 1);
3198         BUG_ON(ret);
3199
3200         leaf = path->nodes[0];
3201         memcpy_extent_buffer(leaf,
3202                              btrfs_item_ptr_offset(leaf, path->slots[0]),
3203                              btrfs_item_ptr_offset(leaf, path->slots[0] - 1),
3204                              item_size);
3205         return 0;
3206 }
3207
3208 /*
3209  * make the item pointed to by the path smaller.  new_size indicates
3210  * how small to make it, and from_end tells us if we just chop bytes
3211  * off the end of the item or if we shift the item to chop bytes off
3212  * the front.
3213  */
3214 int btrfs_truncate_item(struct btrfs_trans_handle *trans,
3215                         struct btrfs_root *root,
3216                         struct btrfs_path *path,
3217                         u32 new_size, int from_end)
3218 {
3219         int ret = 0;
3220         int slot;
3221         struct extent_buffer *leaf;
3222         struct btrfs_item *item;
3223         u32 nritems;
3224         unsigned int data_end;
3225         unsigned int old_data_start;
3226         unsigned int old_size;
3227         unsigned int size_diff;
3228         int i;
3229
3230         leaf = path->nodes[0];
3231         slot = path->slots[0];
3232
3233         old_size = btrfs_item_size_nr(leaf, slot);
3234         if (old_size == new_size)
3235                 return 0;
3236
3237         nritems = btrfs_header_nritems(leaf);
3238         data_end = leaf_data_end(root, leaf);
3239
3240         old_data_start = btrfs_item_offset_nr(leaf, slot);
3241
3242         size_diff = old_size - new_size;
3243
3244         BUG_ON(slot < 0);
3245         BUG_ON(slot >= nritems);
3246
3247         /*
3248          * item0..itemN ... dataN.offset..dataN.size .. data0.size
3249          */
3250         /* first correct the data pointers */
3251         for (i = slot; i < nritems; i++) {
3252                 u32 ioff;
3253                 item = btrfs_item_nr(leaf, i);
3254
3255                 if (!leaf->map_token) {
3256                         map_extent_buffer(leaf, (unsigned long)item,
3257                                         sizeof(struct btrfs_item),
3258                                         &leaf->map_token, &leaf->kaddr,
3259                                         &leaf->map_start, &leaf->map_len,
3260                                         KM_USER1);
3261                 }
3262
3263                 ioff = btrfs_item_offset(leaf, item);
3264                 btrfs_set_item_offset(leaf, item, ioff + size_diff);
3265         }
3266
3267         if (leaf->map_token) {
3268                 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3269                 leaf->map_token = NULL;
3270         }
3271
3272         /* shift the data */
3273         if (from_end) {
3274                 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3275                               data_end + size_diff, btrfs_leaf_data(leaf) +
3276                               data_end, old_data_start + new_size - data_end);
3277         } else {
3278                 struct btrfs_disk_key disk_key;
3279                 u64 offset;
3280
3281                 btrfs_item_key(leaf, &disk_key, slot);
3282
3283                 if (btrfs_disk_key_type(&disk_key) == BTRFS_EXTENT_DATA_KEY) {
3284                         unsigned long ptr;
3285                         struct btrfs_file_extent_item *fi;
3286
3287                         fi = btrfs_item_ptr(leaf, slot,
3288                                             struct btrfs_file_extent_item);
3289                         fi = (struct btrfs_file_extent_item *)(
3290                              (unsigned long)fi - size_diff);
3291
3292                         if (btrfs_file_extent_type(leaf, fi) ==
3293                             BTRFS_FILE_EXTENT_INLINE) {
3294                                 ptr = btrfs_item_ptr_offset(leaf, slot);
3295                                 memmove_extent_buffer(leaf, ptr,
3296                                       (unsigned long)fi,
3297                                       offsetof(struct btrfs_file_extent_item,
3298                                                  disk_bytenr));
3299                         }
3300                 }
3301
3302                 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3303                               data_end + size_diff, btrfs_leaf_data(leaf) +
3304                               data_end, old_data_start - data_end);
3305
3306                 offset = btrfs_disk_key_offset(&disk_key);
3307                 btrfs_set_disk_key_offset(&disk_key, offset + size_diff);
3308                 btrfs_set_item_key(leaf, &disk_key, slot);
3309                 if (slot == 0)
3310                         fixup_low_keys(trans, root, path, &disk_key, 1);
3311         }
3312
3313         item = btrfs_item_nr(leaf, slot);
3314         btrfs_set_item_size(leaf, item, new_size);
3315         btrfs_mark_buffer_dirty(leaf);
3316
3317         ret = 0;
3318         if (btrfs_leaf_free_space(root, leaf) < 0) {
3319                 btrfs_print_leaf(root, leaf);
3320                 BUG();
3321         }
3322         return ret;
3323 }
3324
3325 /*
3326  * make the item pointed to by the path bigger, data_size is the new size.
3327  */
3328 int btrfs_extend_item(struct btrfs_trans_handle *trans,
3329                       struct btrfs_root *root, struct btrfs_path *path,
3330                       u32 data_size)
3331 {
3332         int ret = 0;
3333         int slot;
3334         struct extent_buffer *leaf;
3335         struct btrfs_item *item;
3336         u32 nritems;
3337         unsigned int data_end;
3338         unsigned int old_data;
3339         unsigned int old_size;
3340         int i;
3341
3342         leaf = path->nodes[0];
3343
3344         nritems = btrfs_header_nritems(leaf);
3345         data_end = leaf_data_end(root, leaf);
3346
3347         if (btrfs_leaf_free_space(root, leaf) < data_size) {
3348                 btrfs_print_leaf(root, leaf);
3349                 BUG();
3350         }
3351         slot = path->slots[0];
3352         old_data = btrfs_item_end_nr(leaf, slot);
3353
3354         BUG_ON(slot < 0);
3355         if (slot >= nritems) {
3356                 btrfs_print_leaf(root, leaf);
3357                 printk(KERN_CRIT "slot %d too large, nritems %d\n",
3358                        slot, nritems);
3359                 BUG_ON(1);
3360         }
3361
3362         /*
3363          * item0..itemN ... dataN.offset..dataN.size .. data0.size
3364          */
3365         /* first correct the data pointers */
3366         for (i = slot; i < nritems; i++) {
3367                 u32 ioff;
3368                 item = btrfs_item_nr(leaf, i);
3369
3370                 if (!leaf->map_token) {
3371                         map_extent_buffer(leaf, (unsigned long)item,
3372                                         sizeof(struct btrfs_item),
3373                                         &leaf->map_token, &leaf->kaddr,
3374                                         &leaf->map_start, &leaf->map_len,
3375                                         KM_USER1);
3376                 }
3377                 ioff = btrfs_item_offset(leaf, item);
3378                 btrfs_set_item_offset(leaf, item, ioff - data_size);
3379         }
3380
3381         if (leaf->map_token) {
3382                 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3383                 leaf->map_token = NULL;
3384         }
3385
3386         /* shift the data */
3387         memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3388                       data_end - data_size, btrfs_leaf_data(leaf) +
3389                       data_end, old_data - data_end);
3390
3391         data_end = old_data;
3392         old_size = btrfs_item_size_nr(leaf, slot);
3393         item = btrfs_item_nr(leaf, slot);
3394         btrfs_set_item_size(leaf, item, old_size + data_size);
3395         btrfs_mark_buffer_dirty(leaf);
3396
3397         ret = 0;
3398         if (btrfs_leaf_free_space(root, leaf) < 0) {
3399                 btrfs_print_leaf(root, leaf);
3400                 BUG();
3401         }
3402         return ret;
3403 }
3404
3405 /*
3406  * Given a key and some data, insert items into the tree.
3407  * This does all the path init required, making room in the tree if needed.
3408  * Returns the number of keys that were inserted.
3409  */
3410 int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
3411                             struct btrfs_root *root,
3412                             struct btrfs_path *path,
3413                             struct btrfs_key *cpu_key, u32 *data_size,
3414                             int nr)
3415 {
3416         struct extent_buffer *leaf;
3417         struct btrfs_item *item;
3418         int ret = 0;
3419         int slot;
3420         int i;
3421         u32 nritems;
3422         u32 total_data = 0;
3423         u32 total_size = 0;
3424         unsigned int data_end;
3425         struct btrfs_disk_key disk_key;
3426         struct btrfs_key found_key;
3427
3428         for (i = 0; i < nr; i++) {
3429                 if (total_size + data_size[i] + sizeof(struct btrfs_item) >
3430                     BTRFS_LEAF_DATA_SIZE(root)) {
3431                         break;
3432                         nr = i;
3433                 }
3434                 total_data += data_size[i];
3435                 total_size += data_size[i] + sizeof(struct btrfs_item);
3436         }
3437         BUG_ON(nr == 0);
3438
3439         ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
3440         if (ret == 0)
3441                 return -EEXIST;
3442         if (ret < 0)
3443                 goto out;
3444
3445         leaf = path->nodes[0];
3446
3447         nritems = btrfs_header_nritems(leaf);
3448         data_end = leaf_data_end(root, leaf);
3449
3450         if (btrfs_leaf_free_space(root, leaf) < total_size) {
3451                 for (i = nr; i >= 0; i--) {
3452                         total_data -= data_size[i];
3453                         total_size -= data_size[i] + sizeof(struct btrfs_item);
3454                         if (total_size < btrfs_leaf_free_space(root, leaf))
3455                                 break;
3456                 }
3457                 nr = i;
3458         }
3459
3460         slot = path->slots[0];
3461         BUG_ON(slot < 0);
3462
3463         if (slot != nritems) {
3464                 unsigned int old_data = btrfs_item_end_nr(leaf, slot);
3465
3466                 item = btrfs_item_nr(leaf, slot);
3467                 btrfs_item_key_to_cpu(leaf, &found_key, slot);
3468
3469                 /* figure out how many keys we can insert in here */
3470                 total_data = data_size[0];
3471                 for (i = 1; i < nr; i++) {
3472                         if (btrfs_comp_cpu_keys(&found_key, cpu_key + i) <= 0)
3473                                 break;
3474                         total_data += data_size[i];
3475                 }
3476                 nr = i;
3477
3478                 if (old_data < data_end) {
3479                         btrfs_print_leaf(root, leaf);
3480                         printk(KERN_CRIT "slot %d old_data %d data_end %d\n",
3481                                slot, old_data, data_end);
3482                         BUG_ON(1);
3483                 }
3484                 /*
3485                  * item0..itemN ... dataN.offset..dataN.size .. data0.size
3486                  */
3487                 /* first correct the data pointers */
3488                 WARN_ON(leaf->map_token);
3489                 for (i = slot; i < nritems; i++) {
3490                         u32 ioff;
3491
3492                         item = btrfs_item_nr(leaf, i);
3493                         if (!leaf->map_token) {
3494                                 map_extent_buffer(leaf, (unsigned long)item,
3495                                         sizeof(struct btrfs_item),
3496                                         &leaf->map_token, &leaf->kaddr,
3497                                         &leaf->map_start, &leaf->map_len,
3498                                         KM_USER1);
3499                         }
3500
3501                         ioff = btrfs_item_offset(leaf, item);
3502                         btrfs_set_item_offset(leaf, item, ioff - total_data);
3503                 }
3504                 if (leaf->map_token) {
3505                         unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3506                         leaf->map_token = NULL;
3507                 }
3508
3509                 /* shift the items */
3510                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
3511                               btrfs_item_nr_offset(slot),
3512                               (nritems - slot) * sizeof(struct btrfs_item));
3513
3514                 /* shift the data */
3515                 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3516                               data_end - total_data, btrfs_leaf_data(leaf) +
3517                               data_end, old_data - data_end);
3518                 data_end = old_data;
3519         } else {
3520                 /*
3521                  * this sucks but it has to be done, if we are inserting at
3522                  * the end of the leaf only insert 1 of the items, since we
3523                  * have no way of knowing whats on the next leaf and we'd have
3524                  * to drop our current locks to figure it out
3525                  */
3526                 nr = 1;
3527         }
3528
3529         /* setup the item for the new data */
3530         for (i = 0; i < nr; i++) {
3531                 btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
3532                 btrfs_set_item_key(leaf, &disk_key, slot + i);
3533                 item = btrfs_item_nr(leaf, slot + i);
3534                 btrfs_set_item_offset(leaf, item, data_end - data_size[i]);
3535                 data_end -= data_size[i];
3536                 btrfs_set_item_size(leaf, item, data_size[i]);
3537         }
3538         btrfs_set_header_nritems(leaf, nritems + nr);
3539         btrfs_mark_buffer_dirty(leaf);
3540
3541         ret = 0;
3542         if (slot == 0) {
3543                 btrfs_cpu_key_to_disk(&disk_key, cpu_key);
3544                 ret = fixup_low_keys(trans, root, path, &disk_key, 1);
3545         }
3546
3547         if (btrfs_leaf_free_space(root, leaf) < 0) {
3548                 btrfs_print_leaf(root, leaf);
3549                 BUG();
3550         }
3551 out:
3552         if (!ret)
3553                 ret = nr;
3554         return ret;
3555 }
3556
3557 /*
3558  * this is a helper for btrfs_insert_empty_items, the main goal here is
3559  * to save stack depth by doing the bulk of the work in a function
3560  * that doesn't call btrfs_search_slot
3561  */
3562 static noinline_for_stack int
3563 setup_items_for_insert(struct btrfs_trans_handle *trans,
3564                       struct btrfs_root *root, struct btrfs_path *path,
3565                       struct btrfs_key *cpu_key, u32 *data_size,
3566                       u32 total_data, u32 total_size, int nr)
3567 {
3568         struct btrfs_item *item;
3569         int i;
3570         u32 nritems;
3571         unsigned int data_end;
3572         struct btrfs_disk_key disk_key;
3573         int ret;
3574         struct extent_buffer *leaf;
3575         int slot;
3576
3577         leaf = path->nodes[0];
3578         slot = path->slots[0];
3579
3580         nritems = btrfs_header_nritems(leaf);
3581         data_end = leaf_data_end(root, leaf);
3582
3583         if (btrfs_leaf_free_space(root, leaf) < total_size) {
3584                 btrfs_print_leaf(root, leaf);
3585                 printk(KERN_CRIT "not enough freespace need %u have %d\n",
3586                        total_size, btrfs_leaf_free_space(root, leaf));
3587                 BUG();
3588         }
3589
3590         if (slot != nritems) {
3591                 unsigned int old_data = btrfs_item_end_nr(leaf, slot);
3592
3593                 if (old_data < data_end) {
3594                         btrfs_print_leaf(root, leaf);
3595                         printk(KERN_CRIT "slot %d old_data %d data_end %d\n",
3596                                slot, old_data, data_end);
3597                         BUG_ON(1);
3598                 }
3599                 /*
3600                  * item0..itemN ... dataN.offset..dataN.size .. data0.size
3601                  */
3602                 /* first correct the data pointers */
3603                 WARN_ON(leaf->map_token);
3604                 for (i = slot; i < nritems; i++) {
3605                         u32 ioff;
3606
3607                         item = btrfs_item_nr(leaf, i);
3608                         if (!leaf->map_token) {
3609                                 map_extent_buffer(leaf, (unsigned long)item,
3610                                         sizeof(struct btrfs_item),
3611                                         &leaf->map_token, &leaf->kaddr,
3612                                         &leaf->map_start, &leaf->map_len,
3613                                         KM_USER1);
3614                         }
3615
3616                         ioff = btrfs_item_offset(leaf, item);
3617                         btrfs_set_item_offset(leaf, item, ioff - total_data);
3618                 }
3619                 if (leaf->map_token) {
3620                         unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3621                         leaf->map_token = NULL;
3622                 }
3623
3624                 /* shift the items */
3625                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
3626                               btrfs_item_nr_offset(slot),
3627                               (nritems - slot) * sizeof(struct btrfs_item));
3628
3629                 /* shift the data */
3630                 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3631                               data_end - total_data, btrfs_leaf_data(leaf) +
3632                               data_end, old_data - data_end);
3633                 data_end = old_data;
3634         }
3635
3636         /* setup the item for the new data */
3637         for (i = 0; i < nr; i++) {
3638                 btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
3639                 btrfs_set_item_key(leaf, &disk_key, slot + i);
3640                 item = btrfs_item_nr(leaf, slot + i);
3641                 btrfs_set_item_offset(leaf, item, data_end - data_size[i]);
3642                 data_end -= data_size[i];
3643                 btrfs_set_item_size(leaf, item, data_size[i]);
3644         }
3645
3646         btrfs_set_header_nritems(leaf, nritems + nr);
3647
3648         ret = 0;
3649         if (slot == 0) {
3650                 struct btrfs_disk_key disk_key;
3651                 btrfs_cpu_key_to_disk(&disk_key, cpu_key);
3652                 ret = fixup_low_keys(trans, root, path, &disk_key, 1);
3653         }
3654         btrfs_unlock_up_safe(path, 1);
3655         btrfs_mark_buffer_dirty(leaf);
3656
3657         if (btrfs_leaf_free_space(root, leaf) < 0) {
3658                 btrfs_print_leaf(root, leaf);
3659                 BUG();
3660         }
3661         return ret;
3662 }
3663
3664 /*
3665  * Given a key and some data, insert items into the tree.
3666  * This does all the path init required, making room in the tree if needed.
3667  */
3668 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
3669                             struct btrfs_root *root,
3670                             struct btrfs_path *path,
3671                             struct btrfs_key *cpu_key, u32 *data_size,
3672                             int nr)
3673 {
3674         int ret = 0;
3675         int slot;
3676         int i;
3677         u32 total_size = 0;
3678         u32 total_data = 0;
3679
3680         for (i = 0; i < nr; i++)
3681                 total_data += data_size[i];
3682
3683         total_size = total_data + (nr * sizeof(struct btrfs_item));
3684         ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
3685         if (ret == 0)
3686                 return -EEXIST;
3687         if (ret < 0)
3688                 goto out;
3689
3690         slot = path->slots[0];
3691         BUG_ON(slot < 0);
3692
3693         ret = setup_items_for_insert(trans, root, path, cpu_key, data_size,
3694                                total_data, total_size, nr);
3695
3696 out:
3697         return ret;
3698 }
3699
3700 /*
3701  * Given a key and some data, insert an item into the tree.
3702  * This does all the path init required, making room in the tree if needed.
3703  */
3704 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root
3705                       *root, struct btrfs_key *cpu_key, void *data, u32
3706                       data_size)
3707 {
3708         int ret = 0;
3709         struct btrfs_path *path;
3710         struct extent_buffer *leaf;
3711         unsigned long ptr;
3712
3713         path = btrfs_alloc_path();
3714         if (!path)
3715                 return -ENOMEM;
3716         ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size);
3717         if (!ret) {
3718                 leaf = path->nodes[0];
3719                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
3720                 write_extent_buffer(leaf, data, ptr, data_size);
3721                 btrfs_mark_buffer_dirty(leaf);
3722         }
3723         btrfs_free_path(path);
3724         return ret;
3725 }
3726
3727 /*
3728  * delete the pointer from a given node.
3729  *
3730  * the tree should have been previously balanced so the deletion does not
3731  * empty a node.
3732  */
3733 static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
3734                    struct btrfs_path *path, int level, int slot)
3735 {
3736         struct extent_buffer *parent = path->nodes[level];
3737         u32 nritems;
3738         int ret = 0;
3739         int wret;
3740
3741         nritems = btrfs_header_nritems(parent);
3742         if (slot != nritems - 1) {
3743                 memmove_extent_buffer(parent,
3744                               btrfs_node_key_ptr_offset(slot),
3745                               btrfs_node_key_ptr_offset(slot + 1),
3746                               sizeof(struct btrfs_key_ptr) *
3747                               (nritems - slot - 1));
3748         }
3749         nritems--;
3750         btrfs_set_header_nritems(parent, nritems);
3751         if (nritems == 0 && parent == root->node) {
3752                 BUG_ON(btrfs_header_level(root->node) != 1);
3753                 /* just turn the root into a leaf and break */
3754                 btrfs_set_header_level(root->node, 0);
3755         } else if (slot == 0) {
3756                 struct btrfs_disk_key disk_key;
3757
3758                 btrfs_node_key(parent, &disk_key, 0);
3759                 wret = fixup_low_keys(trans, root, path, &disk_key, level + 1);
3760                 if (wret)
3761                         ret = wret;
3762         }
3763         btrfs_mark_buffer_dirty(parent);
3764         return ret;
3765 }
3766
3767 /*
3768  * a helper function to delete the leaf pointed to by path->slots[1] and
3769  * path->nodes[1].
3770  *
3771  * This deletes the pointer in path->nodes[1] and frees the leaf
3772  * block extent.  zero is returned if it all worked out, < 0 otherwise.
3773  *
3774  * The path must have already been setup for deleting the leaf, including
3775  * all the proper balancing.  path->nodes[1] must be locked.
3776  */
3777 static noinline int btrfs_del_leaf(struct btrfs_trans_handle *trans,
3778                                    struct btrfs_root *root,
3779                                    struct btrfs_path *path,
3780                                    struct extent_buffer *leaf)
3781 {
3782         int ret;
3783
3784         WARN_ON(btrfs_header_generation(leaf) != trans->transid);
3785         ret = del_ptr(trans, root, path, 1, path->slots[1]);
3786         if (ret)
3787                 return ret;
3788
3789         /*
3790          * btrfs_free_extent is expensive, we want to make sure we
3791          * aren't holding any locks when we call it
3792          */
3793         btrfs_unlock_up_safe(path, 0);
3794
3795         root_sub_used(root, leaf->len);
3796
3797         btrfs_free_tree_block(trans, root, leaf, 0, 1);
3798         return 0;
3799 }
3800 /*
3801  * delete the item at the leaf level in path.  If that empties
3802  * the leaf, remove it from the tree
3803  */
3804 int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
3805                     struct btrfs_path *path, int slot, int nr)
3806 {
3807         struct extent_buffer *leaf;
3808         struct btrfs_item *item;
3809         int last_off;
3810         int dsize = 0;
3811         int ret = 0;
3812         int wret;
3813         int i;
3814         u32 nritems;
3815
3816         leaf = path->nodes[0];
3817         last_off = btrfs_item_offset_nr(leaf, slot + nr - 1);
3818
3819         for (i = 0; i < nr; i++)
3820                 dsize += btrfs_item_size_nr(leaf, slot + i);
3821
3822         nritems = btrfs_header_nritems(leaf);
3823
3824         if (slot + nr != nritems) {
3825                 int data_end = leaf_data_end(root, leaf);
3826
3827                 memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) +
3828                               data_end + dsize,
3829                               btrfs_leaf_data(leaf) + data_end,
3830                               last_off - data_end);
3831
3832                 for (i = slot + nr; i < nritems; i++) {
3833                         u32 ioff;
3834
3835                         item = btrfs_item_nr(leaf, i);
3836                         if (!leaf->map_token) {
3837                                 map_extent_buffer(leaf, (unsigned long)item,
3838                                         sizeof(struct btrfs_item),
3839                                         &leaf->map_token, &leaf->kaddr,
3840                                         &leaf->map_start, &leaf->map_len,
3841                                         KM_USER1);
3842                         }
3843                         ioff = btrfs_item_offset(leaf, item);
3844                         btrfs_set_item_offset(leaf, item, ioff + dsize);
3845                 }
3846
3847                 if (leaf->map_token) {
3848                         unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
3849                         leaf->map_token = NULL;
3850                 }
3851
3852                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot),
3853                               btrfs_item_nr_offset(slot + nr),
3854                               sizeof(struct btrfs_item) *
3855                               (nritems - slot - nr));
3856         }
3857         btrfs_set_header_nritems(leaf, nritems - nr);
3858         nritems -= nr;
3859
3860         /* delete the leaf if we've emptied it */
3861         if (nritems == 0) {
3862                 if (leaf == root->node) {
3863                         btrfs_set_header_level(leaf, 0);
3864                 } else {
3865                         btrfs_set_path_blocking(path);
3866                         clean_tree_block(trans, root, leaf);
3867                         ret = btrfs_del_leaf(trans, root, path, leaf);
3868                         BUG_ON(ret);
3869                 }
3870         } else {
3871                 int used = leaf_space_used(leaf, 0, nritems);
3872                 if (slot == 0) {
3873                         struct btrfs_disk_key disk_key;
3874
3875                         btrfs_item_key(leaf, &disk_key, 0);
3876                         wret = fixup_low_keys(trans, root, path,
3877                                               &disk_key, 1);
3878                         if (wret)
3879                                 ret = wret;
3880                 }
3881
3882                 /* delete the leaf if it is mostly empty */
3883                 if (used < BTRFS_LEAF_DATA_SIZE(root) / 3) {
3884                         /* push_leaf_left fixes the path.
3885                          * make sure the path still points to our leaf
3886                          * for possible call to del_ptr below
3887                          */
3888                         slot = path->slots[1];
3889                         extent_buffer_get(leaf);
3890
3891                         btrfs_set_path_blocking(path);
3892                         wret = push_leaf_left(trans, root, path, 1, 1,
3893                                               1, (u32)-1);
3894                         if (wret < 0 && wret != -ENOSPC)
3895                                 ret = wret;
3896
3897                         if (path->nodes[0] == leaf &&
3898                             btrfs_header_nritems(leaf)) {
3899                                 wret = push_leaf_right(trans, root, path, 1,
3900                                                        1, 1, 0);
3901                                 if (wret < 0 && wret != -ENOSPC)
3902                                         ret = wret;
3903                         }
3904
3905                         if (btrfs_header_nritems(leaf) == 0) {
3906                                 path->slots[1] = slot;
3907                                 ret = btrfs_del_leaf(trans, root, path, leaf);
3908                                 BUG_ON(ret);
3909                                 free_extent_buffer(leaf);
3910                         } else {
3911                                 /* if we're still in the path, make sure
3912                                  * we're dirty.  Otherwise, one of the
3913                                  * push_leaf functions must have already
3914                                  * dirtied this buffer
3915                                  */
3916                                 if (path->nodes[0] == leaf)
3917                                         btrfs_mark_buffer_dirty(leaf);
3918                                 free_extent_buffer(leaf);
3919                         }
3920                 } else {
3921                         btrfs_mark_buffer_dirty(leaf);
3922                 }
3923         }
3924         return ret;
3925 }
3926
3927 /*
3928  * search the tree again to find a leaf with lesser keys
3929  * returns 0 if it found something or 1 if there are no lesser leaves.
3930  * returns < 0 on io errors.
3931  *
3932  * This may release the path, and so you may lose any locks held at the
3933  * time you call it.
3934  */
3935 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path)
3936 {
3937         struct btrfs_key key;
3938         struct btrfs_disk_key found_key;
3939         int ret;
3940
3941         btrfs_item_key_to_cpu(path->nodes[0], &key, 0);
3942
3943         if (key.offset > 0)
3944                 key.offset--;
3945         else if (key.type > 0)
3946                 key.type--;
3947         else if (key.objectid > 0)
3948                 key.objectid--;
3949         else
3950                 return 1;
3951
3952         btrfs_release_path(root, path);
3953         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
3954         if (ret < 0)
3955                 return ret;
3956         btrfs_item_key(path->nodes[0], &found_key, 0);
3957         ret = comp_keys(&found_key, &key);
3958         if (ret < 0)
3959                 return 0;
3960         return 1;
3961 }
3962
3963 /*
3964  * A helper function to walk down the tree starting at min_key, and looking
3965  * for nodes or leaves that are either in cache or have a minimum
3966  * transaction id.  This is used by the btree defrag code, and tree logging
3967  *
3968  * This does not cow, but it does stuff the starting key it finds back
3969  * into min_key, so you can call btrfs_search_slot with cow=1 on the
3970  * key and get a writable path.
3971  *
3972  * This does lock as it descends, and path->keep_locks should be set
3973  * to 1 by the caller.
3974  *
3975  * This honors path->lowest_level to prevent descent past a given level
3976  * of the tree.
3977  *
3978  * min_trans indicates the oldest transaction that you are interested
3979  * in walking through.  Any nodes or leaves older than min_trans are
3980  * skipped over (without reading them).
3981  *
3982  * returns zero if something useful was found, < 0 on error and 1 if there
3983  * was nothing in the tree that matched the search criteria.
3984  */
3985 int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
3986                          struct btrfs_key *max_key,
3987                          struct btrfs_path *path, int cache_only,
3988                          u64 min_trans)
3989 {
3990         struct extent_buffer *cur;
3991         struct btrfs_key found_key;
3992         int slot;
3993         int sret;
3994         u32 nritems;
3995         int level;
3996         int ret = 1;
3997
3998         WARN_ON(!path->keep_locks);
3999 again:
4000         cur = btrfs_lock_root_node(root);
4001         level = btrfs_header_level(cur);
4002         WARN_ON(path->nodes[level]);
4003         path->nodes[level] = cur;
4004         path->locks[level] = 1;
4005
4006         if (btrfs_header_generation(cur) < min_trans) {
4007                 ret = 1;
4008                 goto out;
4009         }
4010         while (1) {
4011                 nritems = btrfs_header_nritems(cur);
4012                 level = btrfs_header_level(cur);
4013                 sret = bin_search(cur, min_key, level, &slot);
4014
4015                 /* at the lowest level, we're done, setup the path and exit */
4016                 if (level == path->lowest_level) {
4017                         if (slot >= nritems)
4018                                 goto find_next_key;
4019                         ret = 0;
4020                         path->slots[level] = slot;
4021                         btrfs_item_key_to_cpu(cur, &found_key, slot);
4022                         goto out;
4023                 }
4024                 if (sret && slot > 0)
4025                         slot--;
4026                 /*
4027                  * check this node pointer against the cache_only and
4028                  * min_trans parameters.  If it isn't in cache or is too
4029                  * old, skip to the next one.
4030                  */
4031                 while (slot < nritems) {
4032                         u64 blockptr;
4033                         u64 gen;
4034                         struct extent_buffer *tmp;
4035                         struct btrfs_disk_key disk_key;
4036
4037                         blockptr = btrfs_node_blockptr(cur, slot);
4038                         gen = btrfs_node_ptr_generation(cur, slot);
4039                         if (gen < min_trans) {
4040                                 slot++;
4041                                 continue;
4042                         }
4043                         if (!cache_only)
4044                                 break;
4045
4046                         if (max_key) {
4047                                 btrfs_node_key(cur, &disk_key, slot);
4048                                 if (comp_keys(&disk_key, max_key) >= 0) {
4049                                         ret = 1;
4050                                         goto out;
4051                                 }
4052                         }
4053
4054                         tmp = btrfs_find_tree_block(root, blockptr,
4055                                             btrfs_level_size(root, level - 1));
4056
4057                         if (tmp && btrfs_buffer_uptodate(tmp, gen)) {
4058                                 free_extent_buffer(tmp);
4059                                 break;
4060                         }
4061                         if (tmp)
4062                                 free_extent_buffer(tmp);
4063                         slot++;
4064                 }
4065 find_next_key:
4066                 /*
4067                  * we didn't find a candidate key in this node, walk forward
4068                  * and find another one
4069                  */
4070                 if (slot >= nritems) {
4071                         path->slots[level] = slot;
4072                         btrfs_set_path_blocking(path);
4073                         sret = btrfs_find_next_key(root, path, min_key, level,
4074                                                   cache_only, min_trans);
4075                         if (sret == 0) {
4076                                 btrfs_release_path(root, path);
4077                                 goto again;
4078                         } else {
4079                                 goto out;
4080                         }
4081                 }
4082                 /* save our key for returning back */
4083                 btrfs_node_key_to_cpu(cur, &found_key, slot);
4084                 path->slots[level] = slot;
4085                 if (level == path->lowest_level) {
4086                         ret = 0;
4087                         unlock_up(path, level, 1);
4088                         goto out;
4089                 }
4090                 btrfs_set_path_blocking(path);
4091                 cur = read_node_slot(root, cur, slot);
4092                 BUG_ON(!cur);
4093
4094                 btrfs_tree_lock(cur);
4095
4096                 path->locks[level - 1] = 1;
4097                 path->nodes[level - 1] = cur;
4098                 unlock_up(path, level, 1);
4099                 btrfs_clear_path_blocking(path, NULL);
4100         }
4101 out:
4102         if (ret == 0)
4103                 memcpy(min_key, &found_key, sizeof(found_key));
4104         btrfs_set_path_blocking(path);
4105         return ret;
4106 }
4107
4108 /*
4109  * this is similar to btrfs_next_leaf, but does not try to preserve
4110  * and fixup the path.  It looks for and returns the next key in the
4111  * tree based on the current path and the cache_only and min_trans
4112  * parameters.
4113  *
4114  * 0 is returned if another key is found, < 0 if there are any errors
4115  * and 1 is returned if there are no higher keys in the tree
4116  *
4117  * path->keep_locks should be set to 1 on the search made before
4118  * calling this function.
4119  */
4120 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
4121                         struct btrfs_key *key, int level,
4122                         int cache_only, u64 min_trans)
4123 {
4124         int slot;
4125         struct extent_buffer *c;
4126
4127         WARN_ON(!path->keep_locks);
4128         while (level < BTRFS_MAX_LEVEL) {
4129                 if (!path->nodes[level])
4130                         return 1;
4131
4132                 slot = path->slots[level] + 1;
4133                 c = path->nodes[level];
4134 next:
4135                 if (slot >= btrfs_header_nritems(c)) {
4136                         int ret;
4137                         int orig_lowest;
4138                         struct btrfs_key cur_key;
4139                         if (level + 1 >= BTRFS_MAX_LEVEL ||
4140                             !path->nodes[level + 1])
4141                                 return 1;
4142
4143                         if (path->locks[level + 1]) {
4144                                 level++;
4145                                 continue;
4146                         }
4147
4148                         slot = btrfs_header_nritems(c) - 1;
4149                         if (level == 0)
4150                                 btrfs_item_key_to_cpu(c, &cur_key, slot);
4151                         else
4152                                 btrfs_node_key_to_cpu(c, &cur_key, slot);
4153
4154                         orig_lowest = path->lowest_level;
4155                         btrfs_release_path(root, path);
4156                         path->lowest_level = level;
4157                         ret = btrfs_search_slot(NULL, root, &cur_key, path,
4158                                                 0, 0);
4159                         path->lowest_level = orig_lowest;
4160                         if (ret < 0)
4161                                 return ret;
4162
4163                         c = path->nodes[level];
4164                         slot = path->slots[level];
4165                         if (ret == 0)
4166                                 slot++;
4167                         goto next;
4168                 }
4169
4170                 if (level == 0)
4171                         btrfs_item_key_to_cpu(c, key, slot);
4172                 else {
4173                         u64 blockptr = btrfs_node_blockptr(c, slot);
4174                         u64 gen = btrfs_node_ptr_generation(c, slot);
4175
4176                         if (cache_only) {
4177                                 struct extent_buffer *cur;
4178                                 cur = btrfs_find_tree_block(root, blockptr,
4179                                             btrfs_level_size(root, level - 1));
4180                                 if (!cur || !btrfs_buffer_uptodate(cur, gen)) {
4181                                         slot++;
4182                                         if (cur)
4183                                                 free_extent_buffer(cur);
4184                                         goto next;
4185                                 }
4186                                 free_extent_buffer(cur);
4187                         }
4188                         if (gen < min_trans) {
4189                                 slot++;
4190                                 goto next;
4191                         }
4192                         btrfs_node_key_to_cpu(c, key, slot);
4193                 }
4194                 return 0;
4195         }
4196         return 1;
4197 }
4198
4199 /*
4200  * search the tree again to find a leaf with greater keys
4201  * returns 0 if it found something or 1 if there are no greater leaves.
4202  * returns < 0 on io errors.
4203  */
4204 int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
4205 {
4206         int slot;
4207         int level;
4208         struct extent_buffer *c;
4209         struct extent_buffer *next;
4210         struct btrfs_key key;
4211         u32 nritems;
4212         int ret;
4213         int old_spinning = path->leave_spinning;
4214         int force_blocking = 0;
4215
4216         nritems = btrfs_header_nritems(path->nodes[0]);
4217         if (nritems == 0)
4218                 return 1;
4219
4220         /*
4221          * we take the blocks in an order that upsets lockdep.  Using
4222          * blocking mode is the only way around it.
4223          */
4224 #ifdef CONFIG_DEBUG_LOCK_ALLOC
4225         force_blocking = 1;
4226 #endif
4227
4228         btrfs_item_key_to_cpu(path->nodes[0], &key, nritems - 1);
4229 again:
4230         level = 1;
4231         next = NULL;
4232         btrfs_release_path(root, path);
4233
4234         path->keep_locks = 1;
4235
4236         if (!force_blocking)
4237                 path->leave_spinning = 1;
4238
4239         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
4240         path->keep_locks = 0;
4241
4242         if (ret < 0)
4243                 return ret;
4244
4245         nritems = btrfs_header_nritems(path->nodes[0]);
4246         /*
4247          * by releasing the path above we dropped all our locks.  A balance
4248          * could have added more items next to the key that used to be
4249          * at the very end of the block.  So, check again here and
4250          * advance the path if there are now more items available.
4251          */
4252         if (nritems > 0 && path->slots[0] < nritems - 1) {
4253                 if (ret == 0)
4254                         path->slots[0]++;
4255                 ret = 0;
4256                 goto done;
4257         }
4258
4259         while (level < BTRFS_MAX_LEVEL) {
4260                 if (!path->nodes[level]) {
4261                         ret = 1;
4262                         goto done;
4263                 }
4264
4265                 slot = path->slots[level] + 1;
4266                 c = path->nodes[level];
4267                 if (slot >= btrfs_header_nritems(c)) {
4268                         level++;
4269                         if (level == BTRFS_MAX_LEVEL) {
4270                                 ret = 1;
4271                                 goto done;
4272                         }
4273                         continue;
4274                 }
4275
4276                 if (next) {
4277                         btrfs_tree_unlock(next);
4278                         free_extent_buffer(next);
4279                 }
4280
4281                 next = c;
4282                 ret = read_block_for_search(NULL, root, path, &next, level,
4283                                             slot, &key);
4284                 if (ret == -EAGAIN)
4285                         goto again;
4286
4287                 if (ret < 0) {
4288                         btrfs_release_path(root, path);
4289                         goto done;
4290                 }
4291
4292                 if (!path->skip_locking) {
4293                         ret = btrfs_try_spin_lock(next);
4294                         if (!ret) {
4295                                 btrfs_set_path_blocking(path);
4296                                 btrfs_tree_lock(next);
4297                                 if (!force_blocking)
4298                                         btrfs_clear_path_blocking(path, next);
4299                         }
4300                         if (force_blocking)
4301                                 btrfs_set_lock_blocking(next);
4302                 }
4303                 break;
4304         }
4305         path->slots[level] = slot;
4306         while (1) {
4307                 level--;
4308                 c = path->nodes[level];
4309                 if (path->locks[level])
4310                         btrfs_tree_unlock(c);
4311
4312                 free_extent_buffer(c);
4313                 path->nodes[level] = next;
4314                 path->slots[level] = 0;
4315                 if (!path->skip_locking)
4316                         path->locks[level] = 1;
4317
4318                 if (!level)
4319                         break;
4320
4321                 ret = read_block_for_search(NULL, root, path, &next, level,
4322                                             0, &key);
4323                 if (ret == -EAGAIN)
4324                         goto again;
4325
4326                 if (ret < 0) {
4327                         btrfs_release_path(root, path);
4328                         goto done;
4329                 }
4330
4331                 if (!path->skip_locking) {
4332                         btrfs_assert_tree_locked(path->nodes[level]);
4333                         ret = btrfs_try_spin_lock(next);
4334                         if (!ret) {
4335                                 btrfs_set_path_blocking(path);
4336                                 btrfs_tree_lock(next);
4337                                 if (!force_blocking)
4338                                         btrfs_clear_path_blocking(path, next);
4339                         }
4340                         if (force_blocking)
4341                                 btrfs_set_lock_blocking(next);
4342                 }
4343         }
4344         ret = 0;
4345 done:
4346         unlock_up(path, 0, 1);
4347         path->leave_spinning = old_spinning;
4348         if (!old_spinning)
4349                 btrfs_set_path_blocking(path);
4350
4351         return ret;
4352 }
4353
4354 /*
4355  * this uses btrfs_prev_leaf to walk backwards in the tree, and keeps
4356  * searching until it gets past min_objectid or finds an item of 'type'
4357  *
4358  * returns 0 if something is found, 1 if nothing was found and < 0 on error
4359  */
4360 int btrfs_previous_item(struct btrfs_root *root,
4361                         struct btrfs_path *path, u64 min_objectid,
4362                         int type)
4363 {
4364         struct btrfs_key found_key;
4365         struct extent_buffer *leaf;
4366         u32 nritems;
4367         int ret;
4368
4369         while (1) {
4370                 if (path->slots[0] == 0) {
4371                         btrfs_set_path_blocking(path);
4372                         ret = btrfs_prev_leaf(root, path);
4373                         if (ret != 0)
4374                                 return ret;
4375                 } else {
4376                         path->slots[0]--;
4377                 }
4378                 leaf = path->nodes[0];
4379                 nritems = btrfs_header_nritems(leaf);
4380                 if (nritems == 0)
4381                         return 1;
4382                 if (path->slots[0] == nritems)
4383                         path->slots[0]--;
4384
4385                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4386                 if (found_key.objectid < min_objectid)
4387                         break;
4388                 if (found_key.type == type)
4389                         return 0;
4390                 if (found_key.objectid == min_objectid &&
4391                     found_key.type < type)
4392                         break;
4393         }
4394         return 1;
4395 }