GNU Linux-libre 5.19-rc6-gnu
[releases.git] / fs / btrfs / ctree.c
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Copyright (C) 2007,2008 Oracle.  All rights reserved.
4  */
5
6 #include <linux/sched.h>
7 #include <linux/slab.h>
8 #include <linux/rbtree.h>
9 #include <linux/mm.h>
10 #include <linux/error-injection.h>
11 #include "ctree.h"
12 #include "disk-io.h"
13 #include "transaction.h"
14 #include "print-tree.h"
15 #include "locking.h"
16 #include "volumes.h"
17 #include "qgroup.h"
18 #include "tree-mod-log.h"
19 #include "tree-checker.h"
20
21 static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root
22                       *root, struct btrfs_path *path, int level);
23 static int split_leaf(struct btrfs_trans_handle *trans, struct btrfs_root *root,
24                       const struct btrfs_key *ins_key, struct btrfs_path *path,
25                       int data_size, int extend);
26 static int push_node_left(struct btrfs_trans_handle *trans,
27                           struct extent_buffer *dst,
28                           struct extent_buffer *src, int empty);
29 static int balance_node_right(struct btrfs_trans_handle *trans,
30                               struct extent_buffer *dst_buf,
31                               struct extent_buffer *src_buf);
32 static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
33                     int level, int slot);
34
35 static const struct btrfs_csums {
36         u16             size;
37         const char      name[10];
38         const char      driver[12];
39 } btrfs_csums[] = {
40         [BTRFS_CSUM_TYPE_CRC32] = { .size = 4, .name = "crc32c" },
41         [BTRFS_CSUM_TYPE_XXHASH] = { .size = 8, .name = "xxhash64" },
42         [BTRFS_CSUM_TYPE_SHA256] = { .size = 32, .name = "sha256" },
43         [BTRFS_CSUM_TYPE_BLAKE2] = { .size = 32, .name = "blake2b",
44                                      .driver = "blake2b-256" },
45 };
46
47 int btrfs_super_csum_size(const struct btrfs_super_block *s)
48 {
49         u16 t = btrfs_super_csum_type(s);
50         /*
51          * csum type is validated at mount time
52          */
53         return btrfs_csums[t].size;
54 }
55
56 const char *btrfs_super_csum_name(u16 csum_type)
57 {
58         /* csum type is validated at mount time */
59         return btrfs_csums[csum_type].name;
60 }
61
62 /*
63  * Return driver name if defined, otherwise the name that's also a valid driver
64  * name
65  */
66 const char *btrfs_super_csum_driver(u16 csum_type)
67 {
68         /* csum type is validated at mount time */
69         return btrfs_csums[csum_type].driver[0] ?
70                 btrfs_csums[csum_type].driver :
71                 btrfs_csums[csum_type].name;
72 }
73
74 size_t __attribute_const__ btrfs_get_num_csums(void)
75 {
76         return ARRAY_SIZE(btrfs_csums);
77 }
78
79 struct btrfs_path *btrfs_alloc_path(void)
80 {
81         return kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS);
82 }
83
84 /* this also releases the path */
85 void btrfs_free_path(struct btrfs_path *p)
86 {
87         if (!p)
88                 return;
89         btrfs_release_path(p);
90         kmem_cache_free(btrfs_path_cachep, p);
91 }
92
93 /*
94  * path release drops references on the extent buffers in the path
95  * and it drops any locks held by this path
96  *
97  * It is safe to call this on paths that no locks or extent buffers held.
98  */
99 noinline void btrfs_release_path(struct btrfs_path *p)
100 {
101         int i;
102
103         for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
104                 p->slots[i] = 0;
105                 if (!p->nodes[i])
106                         continue;
107                 if (p->locks[i]) {
108                         btrfs_tree_unlock_rw(p->nodes[i], p->locks[i]);
109                         p->locks[i] = 0;
110                 }
111                 free_extent_buffer(p->nodes[i]);
112                 p->nodes[i] = NULL;
113         }
114 }
115
116 /*
117  * safely gets a reference on the root node of a tree.  A lock
118  * is not taken, so a concurrent writer may put a different node
119  * at the root of the tree.  See btrfs_lock_root_node for the
120  * looping required.
121  *
122  * The extent buffer returned by this has a reference taken, so
123  * it won't disappear.  It may stop being the root of the tree
124  * at any time because there are no locks held.
125  */
126 struct extent_buffer *btrfs_root_node(struct btrfs_root *root)
127 {
128         struct extent_buffer *eb;
129
130         while (1) {
131                 rcu_read_lock();
132                 eb = rcu_dereference(root->node);
133
134                 /*
135                  * RCU really hurts here, we could free up the root node because
136                  * it was COWed but we may not get the new root node yet so do
137                  * the inc_not_zero dance and if it doesn't work then
138                  * synchronize_rcu and try again.
139                  */
140                 if (atomic_inc_not_zero(&eb->refs)) {
141                         rcu_read_unlock();
142                         break;
143                 }
144                 rcu_read_unlock();
145                 synchronize_rcu();
146         }
147         return eb;
148 }
149
150 /*
151  * Cowonly root (not-shareable trees, everything not subvolume or reloc roots),
152  * just get put onto a simple dirty list.  Transaction walks this list to make
153  * sure they get properly updated on disk.
154  */
155 static void add_root_to_dirty_list(struct btrfs_root *root)
156 {
157         struct btrfs_fs_info *fs_info = root->fs_info;
158
159         if (test_bit(BTRFS_ROOT_DIRTY, &root->state) ||
160             !test_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state))
161                 return;
162
163         spin_lock(&fs_info->trans_lock);
164         if (!test_and_set_bit(BTRFS_ROOT_DIRTY, &root->state)) {
165                 /* Want the extent tree to be the last on the list */
166                 if (root->root_key.objectid == BTRFS_EXTENT_TREE_OBJECTID)
167                         list_move_tail(&root->dirty_list,
168                                        &fs_info->dirty_cowonly_roots);
169                 else
170                         list_move(&root->dirty_list,
171                                   &fs_info->dirty_cowonly_roots);
172         }
173         spin_unlock(&fs_info->trans_lock);
174 }
175
176 /*
177  * used by snapshot creation to make a copy of a root for a tree with
178  * a given objectid.  The buffer with the new root node is returned in
179  * cow_ret, and this func returns zero on success or a negative error code.
180  */
181 int btrfs_copy_root(struct btrfs_trans_handle *trans,
182                       struct btrfs_root *root,
183                       struct extent_buffer *buf,
184                       struct extent_buffer **cow_ret, u64 new_root_objectid)
185 {
186         struct btrfs_fs_info *fs_info = root->fs_info;
187         struct extent_buffer *cow;
188         int ret = 0;
189         int level;
190         struct btrfs_disk_key disk_key;
191
192         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
193                 trans->transid != fs_info->running_transaction->transid);
194         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
195                 trans->transid != root->last_trans);
196
197         level = btrfs_header_level(buf);
198         if (level == 0)
199                 btrfs_item_key(buf, &disk_key, 0);
200         else
201                 btrfs_node_key(buf, &disk_key, 0);
202
203         cow = btrfs_alloc_tree_block(trans, root, 0, new_root_objectid,
204                                      &disk_key, level, buf->start, 0,
205                                      BTRFS_NESTING_NEW_ROOT);
206         if (IS_ERR(cow))
207                 return PTR_ERR(cow);
208
209         copy_extent_buffer_full(cow, buf);
210         btrfs_set_header_bytenr(cow, cow->start);
211         btrfs_set_header_generation(cow, trans->transid);
212         btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
213         btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
214                                      BTRFS_HEADER_FLAG_RELOC);
215         if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
216                 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
217         else
218                 btrfs_set_header_owner(cow, new_root_objectid);
219
220         write_extent_buffer_fsid(cow, fs_info->fs_devices->metadata_uuid);
221
222         WARN_ON(btrfs_header_generation(buf) > trans->transid);
223         if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
224                 ret = btrfs_inc_ref(trans, root, cow, 1);
225         else
226                 ret = btrfs_inc_ref(trans, root, cow, 0);
227         if (ret) {
228                 btrfs_tree_unlock(cow);
229                 free_extent_buffer(cow);
230                 btrfs_abort_transaction(trans, ret);
231                 return ret;
232         }
233
234         btrfs_mark_buffer_dirty(cow);
235         *cow_ret = cow;
236         return 0;
237 }
238
239 /*
240  * check if the tree block can be shared by multiple trees
241  */
242 int btrfs_block_can_be_shared(struct btrfs_root *root,
243                               struct extent_buffer *buf)
244 {
245         /*
246          * Tree blocks not in shareable trees and tree roots are never shared.
247          * If a block was allocated after the last snapshot and the block was
248          * not allocated by tree relocation, we know the block is not shared.
249          */
250         if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
251             buf != root->node && buf != root->commit_root &&
252             (btrfs_header_generation(buf) <=
253              btrfs_root_last_snapshot(&root->root_item) ||
254              btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)))
255                 return 1;
256
257         return 0;
258 }
259
260 static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
261                                        struct btrfs_root *root,
262                                        struct extent_buffer *buf,
263                                        struct extent_buffer *cow,
264                                        int *last_ref)
265 {
266         struct btrfs_fs_info *fs_info = root->fs_info;
267         u64 refs;
268         u64 owner;
269         u64 flags;
270         u64 new_flags = 0;
271         int ret;
272
273         /*
274          * Backrefs update rules:
275          *
276          * Always use full backrefs for extent pointers in tree block
277          * allocated by tree relocation.
278          *
279          * If a shared tree block is no longer referenced by its owner
280          * tree (btrfs_header_owner(buf) == root->root_key.objectid),
281          * use full backrefs for extent pointers in tree block.
282          *
283          * If a tree block is been relocating
284          * (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID),
285          * use full backrefs for extent pointers in tree block.
286          * The reason for this is some operations (such as drop tree)
287          * are only allowed for blocks use full backrefs.
288          */
289
290         if (btrfs_block_can_be_shared(root, buf)) {
291                 ret = btrfs_lookup_extent_info(trans, fs_info, buf->start,
292                                                btrfs_header_level(buf), 1,
293                                                &refs, &flags);
294                 if (ret)
295                         return ret;
296                 if (refs == 0) {
297                         ret = -EROFS;
298                         btrfs_handle_fs_error(fs_info, ret, NULL);
299                         return ret;
300                 }
301         } else {
302                 refs = 1;
303                 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
304                     btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
305                         flags = BTRFS_BLOCK_FLAG_FULL_BACKREF;
306                 else
307                         flags = 0;
308         }
309
310         owner = btrfs_header_owner(buf);
311         BUG_ON(owner == BTRFS_TREE_RELOC_OBJECTID &&
312                !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
313
314         if (refs > 1) {
315                 if ((owner == root->root_key.objectid ||
316                      root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) &&
317                     !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) {
318                         ret = btrfs_inc_ref(trans, root, buf, 1);
319                         if (ret)
320                                 return ret;
321
322                         if (root->root_key.objectid ==
323                             BTRFS_TREE_RELOC_OBJECTID) {
324                                 ret = btrfs_dec_ref(trans, root, buf, 0);
325                                 if (ret)
326                                         return ret;
327                                 ret = btrfs_inc_ref(trans, root, cow, 1);
328                                 if (ret)
329                                         return ret;
330                         }
331                         new_flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
332                 } else {
333
334                         if (root->root_key.objectid ==
335                             BTRFS_TREE_RELOC_OBJECTID)
336                                 ret = btrfs_inc_ref(trans, root, cow, 1);
337                         else
338                                 ret = btrfs_inc_ref(trans, root, cow, 0);
339                         if (ret)
340                                 return ret;
341                 }
342                 if (new_flags != 0) {
343                         int level = btrfs_header_level(buf);
344
345                         ret = btrfs_set_disk_extent_flags(trans, buf,
346                                                           new_flags, level);
347                         if (ret)
348                                 return ret;
349                 }
350         } else {
351                 if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
352                         if (root->root_key.objectid ==
353                             BTRFS_TREE_RELOC_OBJECTID)
354                                 ret = btrfs_inc_ref(trans, root, cow, 1);
355                         else
356                                 ret = btrfs_inc_ref(trans, root, cow, 0);
357                         if (ret)
358                                 return ret;
359                         ret = btrfs_dec_ref(trans, root, buf, 1);
360                         if (ret)
361                                 return ret;
362                 }
363                 btrfs_clean_tree_block(buf);
364                 *last_ref = 1;
365         }
366         return 0;
367 }
368
369 /*
370  * does the dirty work in cow of a single block.  The parent block (if
371  * supplied) is updated to point to the new cow copy.  The new buffer is marked
372  * dirty and returned locked.  If you modify the block it needs to be marked
373  * dirty again.
374  *
375  * search_start -- an allocation hint for the new block
376  *
377  * empty_size -- a hint that you plan on doing more cow.  This is the size in
378  * bytes the allocator should try to find free next to the block it returns.
379  * This is just a hint and may be ignored by the allocator.
380  */
381 static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
382                              struct btrfs_root *root,
383                              struct extent_buffer *buf,
384                              struct extent_buffer *parent, int parent_slot,
385                              struct extent_buffer **cow_ret,
386                              u64 search_start, u64 empty_size,
387                              enum btrfs_lock_nesting nest)
388 {
389         struct btrfs_fs_info *fs_info = root->fs_info;
390         struct btrfs_disk_key disk_key;
391         struct extent_buffer *cow;
392         int level, ret;
393         int last_ref = 0;
394         int unlock_orig = 0;
395         u64 parent_start = 0;
396
397         if (*cow_ret == buf)
398                 unlock_orig = 1;
399
400         btrfs_assert_tree_write_locked(buf);
401
402         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
403                 trans->transid != fs_info->running_transaction->transid);
404         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
405                 trans->transid != root->last_trans);
406
407         level = btrfs_header_level(buf);
408
409         if (level == 0)
410                 btrfs_item_key(buf, &disk_key, 0);
411         else
412                 btrfs_node_key(buf, &disk_key, 0);
413
414         if ((root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) && parent)
415                 parent_start = parent->start;
416
417         cow = btrfs_alloc_tree_block(trans, root, parent_start,
418                                      root->root_key.objectid, &disk_key, level,
419                                      search_start, empty_size, nest);
420         if (IS_ERR(cow))
421                 return PTR_ERR(cow);
422
423         /* cow is set to blocking by btrfs_init_new_buffer */
424
425         copy_extent_buffer_full(cow, buf);
426         btrfs_set_header_bytenr(cow, cow->start);
427         btrfs_set_header_generation(cow, trans->transid);
428         btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
429         btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
430                                      BTRFS_HEADER_FLAG_RELOC);
431         if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
432                 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
433         else
434                 btrfs_set_header_owner(cow, root->root_key.objectid);
435
436         write_extent_buffer_fsid(cow, fs_info->fs_devices->metadata_uuid);
437
438         ret = update_ref_for_cow(trans, root, buf, cow, &last_ref);
439         if (ret) {
440                 btrfs_tree_unlock(cow);
441                 free_extent_buffer(cow);
442                 btrfs_abort_transaction(trans, ret);
443                 return ret;
444         }
445
446         if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) {
447                 ret = btrfs_reloc_cow_block(trans, root, buf, cow);
448                 if (ret) {
449                         btrfs_tree_unlock(cow);
450                         free_extent_buffer(cow);
451                         btrfs_abort_transaction(trans, ret);
452                         return ret;
453                 }
454         }
455
456         if (buf == root->node) {
457                 WARN_ON(parent && parent != buf);
458                 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
459                     btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
460                         parent_start = buf->start;
461
462                 atomic_inc(&cow->refs);
463                 ret = btrfs_tree_mod_log_insert_root(root->node, cow, true);
464                 BUG_ON(ret < 0);
465                 rcu_assign_pointer(root->node, cow);
466
467                 btrfs_free_tree_block(trans, btrfs_root_id(root), buf,
468                                       parent_start, last_ref);
469                 free_extent_buffer(buf);
470                 add_root_to_dirty_list(root);
471         } else {
472                 WARN_ON(trans->transid != btrfs_header_generation(parent));
473                 btrfs_tree_mod_log_insert_key(parent, parent_slot,
474                                               BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
475                 btrfs_set_node_blockptr(parent, parent_slot,
476                                         cow->start);
477                 btrfs_set_node_ptr_generation(parent, parent_slot,
478                                               trans->transid);
479                 btrfs_mark_buffer_dirty(parent);
480                 if (last_ref) {
481                         ret = btrfs_tree_mod_log_free_eb(buf);
482                         if (ret) {
483                                 btrfs_tree_unlock(cow);
484                                 free_extent_buffer(cow);
485                                 btrfs_abort_transaction(trans, ret);
486                                 return ret;
487                         }
488                 }
489                 btrfs_free_tree_block(trans, btrfs_root_id(root), buf,
490                                       parent_start, last_ref);
491         }
492         if (unlock_orig)
493                 btrfs_tree_unlock(buf);
494         free_extent_buffer_stale(buf);
495         btrfs_mark_buffer_dirty(cow);
496         *cow_ret = cow;
497         return 0;
498 }
499
500 static inline int should_cow_block(struct btrfs_trans_handle *trans,
501                                    struct btrfs_root *root,
502                                    struct extent_buffer *buf)
503 {
504         if (btrfs_is_testing(root->fs_info))
505                 return 0;
506
507         /* Ensure we can see the FORCE_COW bit */
508         smp_mb__before_atomic();
509
510         /*
511          * We do not need to cow a block if
512          * 1) this block is not created or changed in this transaction;
513          * 2) this block does not belong to TREE_RELOC tree;
514          * 3) the root is not forced COW.
515          *
516          * What is forced COW:
517          *    when we create snapshot during committing the transaction,
518          *    after we've finished copying src root, we must COW the shared
519          *    block to ensure the metadata consistency.
520          */
521         if (btrfs_header_generation(buf) == trans->transid &&
522             !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) &&
523             !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
524               btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)) &&
525             !test_bit(BTRFS_ROOT_FORCE_COW, &root->state))
526                 return 0;
527         return 1;
528 }
529
530 /*
531  * cows a single block, see __btrfs_cow_block for the real work.
532  * This version of it has extra checks so that a block isn't COWed more than
533  * once per transaction, as long as it hasn't been written yet
534  */
535 noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
536                     struct btrfs_root *root, struct extent_buffer *buf,
537                     struct extent_buffer *parent, int parent_slot,
538                     struct extent_buffer **cow_ret,
539                     enum btrfs_lock_nesting nest)
540 {
541         struct btrfs_fs_info *fs_info = root->fs_info;
542         u64 search_start;
543         int ret;
544
545         if (test_bit(BTRFS_ROOT_DELETING, &root->state))
546                 btrfs_err(fs_info,
547                         "COW'ing blocks on a fs root that's being dropped");
548
549         if (trans->transaction != fs_info->running_transaction)
550                 WARN(1, KERN_CRIT "trans %llu running %llu\n",
551                        trans->transid,
552                        fs_info->running_transaction->transid);
553
554         if (trans->transid != fs_info->generation)
555                 WARN(1, KERN_CRIT "trans %llu running %llu\n",
556                        trans->transid, fs_info->generation);
557
558         if (!should_cow_block(trans, root, buf)) {
559                 *cow_ret = buf;
560                 return 0;
561         }
562
563         search_start = buf->start & ~((u64)SZ_1G - 1);
564
565         /*
566          * Before CoWing this block for later modification, check if it's
567          * the subtree root and do the delayed subtree trace if needed.
568          *
569          * Also We don't care about the error, as it's handled internally.
570          */
571         btrfs_qgroup_trace_subtree_after_cow(trans, root, buf);
572         ret = __btrfs_cow_block(trans, root, buf, parent,
573                                  parent_slot, cow_ret, search_start, 0, nest);
574
575         trace_btrfs_cow_block(root, buf, *cow_ret);
576
577         return ret;
578 }
579 ALLOW_ERROR_INJECTION(btrfs_cow_block, ERRNO);
580
581 /*
582  * helper function for defrag to decide if two blocks pointed to by a
583  * node are actually close by
584  */
585 static int close_blocks(u64 blocknr, u64 other, u32 blocksize)
586 {
587         if (blocknr < other && other - (blocknr + blocksize) < 32768)
588                 return 1;
589         if (blocknr > other && blocknr - (other + blocksize) < 32768)
590                 return 1;
591         return 0;
592 }
593
594 #ifdef __LITTLE_ENDIAN
595
596 /*
597  * Compare two keys, on little-endian the disk order is same as CPU order and
598  * we can avoid the conversion.
599  */
600 static int comp_keys(const struct btrfs_disk_key *disk_key,
601                      const struct btrfs_key *k2)
602 {
603         const struct btrfs_key *k1 = (const struct btrfs_key *)disk_key;
604
605         return btrfs_comp_cpu_keys(k1, k2);
606 }
607
608 #else
609
610 /*
611  * compare two keys in a memcmp fashion
612  */
613 static int comp_keys(const struct btrfs_disk_key *disk,
614                      const struct btrfs_key *k2)
615 {
616         struct btrfs_key k1;
617
618         btrfs_disk_key_to_cpu(&k1, disk);
619
620         return btrfs_comp_cpu_keys(&k1, k2);
621 }
622 #endif
623
624 /*
625  * same as comp_keys only with two btrfs_key's
626  */
627 int __pure btrfs_comp_cpu_keys(const struct btrfs_key *k1, const struct btrfs_key *k2)
628 {
629         if (k1->objectid > k2->objectid)
630                 return 1;
631         if (k1->objectid < k2->objectid)
632                 return -1;
633         if (k1->type > k2->type)
634                 return 1;
635         if (k1->type < k2->type)
636                 return -1;
637         if (k1->offset > k2->offset)
638                 return 1;
639         if (k1->offset < k2->offset)
640                 return -1;
641         return 0;
642 }
643
644 /*
645  * this is used by the defrag code to go through all the
646  * leaves pointed to by a node and reallocate them so that
647  * disk order is close to key order
648  */
649 int btrfs_realloc_node(struct btrfs_trans_handle *trans,
650                        struct btrfs_root *root, struct extent_buffer *parent,
651                        int start_slot, u64 *last_ret,
652                        struct btrfs_key *progress)
653 {
654         struct btrfs_fs_info *fs_info = root->fs_info;
655         struct extent_buffer *cur;
656         u64 blocknr;
657         u64 search_start = *last_ret;
658         u64 last_block = 0;
659         u64 other;
660         u32 parent_nritems;
661         int end_slot;
662         int i;
663         int err = 0;
664         u32 blocksize;
665         int progress_passed = 0;
666         struct btrfs_disk_key disk_key;
667
668         WARN_ON(trans->transaction != fs_info->running_transaction);
669         WARN_ON(trans->transid != fs_info->generation);
670
671         parent_nritems = btrfs_header_nritems(parent);
672         blocksize = fs_info->nodesize;
673         end_slot = parent_nritems - 1;
674
675         if (parent_nritems <= 1)
676                 return 0;
677
678         for (i = start_slot; i <= end_slot; i++) {
679                 int close = 1;
680
681                 btrfs_node_key(parent, &disk_key, i);
682                 if (!progress_passed && comp_keys(&disk_key, progress) < 0)
683                         continue;
684
685                 progress_passed = 1;
686                 blocknr = btrfs_node_blockptr(parent, i);
687                 if (last_block == 0)
688                         last_block = blocknr;
689
690                 if (i > 0) {
691                         other = btrfs_node_blockptr(parent, i - 1);
692                         close = close_blocks(blocknr, other, blocksize);
693                 }
694                 if (!close && i < end_slot) {
695                         other = btrfs_node_blockptr(parent, i + 1);
696                         close = close_blocks(blocknr, other, blocksize);
697                 }
698                 if (close) {
699                         last_block = blocknr;
700                         continue;
701                 }
702
703                 cur = btrfs_read_node_slot(parent, i);
704                 if (IS_ERR(cur))
705                         return PTR_ERR(cur);
706                 if (search_start == 0)
707                         search_start = last_block;
708
709                 btrfs_tree_lock(cur);
710                 err = __btrfs_cow_block(trans, root, cur, parent, i,
711                                         &cur, search_start,
712                                         min(16 * blocksize,
713                                             (end_slot - i) * blocksize),
714                                         BTRFS_NESTING_COW);
715                 if (err) {
716                         btrfs_tree_unlock(cur);
717                         free_extent_buffer(cur);
718                         break;
719                 }
720                 search_start = cur->start;
721                 last_block = cur->start;
722                 *last_ret = search_start;
723                 btrfs_tree_unlock(cur);
724                 free_extent_buffer(cur);
725         }
726         return err;
727 }
728
729 /*
730  * Search for a key in the given extent_buffer.
731  *
732  * The lower boundary for the search is specified by the slot number @low. Use a
733  * value of 0 to search over the whole extent buffer.
734  *
735  * The slot in the extent buffer is returned via @slot. If the key exists in the
736  * extent buffer, then @slot will point to the slot where the key is, otherwise
737  * it points to the slot where you would insert the key.
738  *
739  * Slot may point to the total number of items (i.e. one position beyond the last
740  * key) if the key is bigger than the last key in the extent buffer.
741  */
742 static noinline int generic_bin_search(struct extent_buffer *eb, int low,
743                                        const struct btrfs_key *key, int *slot)
744 {
745         unsigned long p;
746         int item_size;
747         int high = btrfs_header_nritems(eb);
748         int ret;
749         const int key_size = sizeof(struct btrfs_disk_key);
750
751         if (low > high) {
752                 btrfs_err(eb->fs_info,
753                  "%s: low (%d) > high (%d) eb %llu owner %llu level %d",
754                           __func__, low, high, eb->start,
755                           btrfs_header_owner(eb), btrfs_header_level(eb));
756                 return -EINVAL;
757         }
758
759         if (btrfs_header_level(eb) == 0) {
760                 p = offsetof(struct btrfs_leaf, items);
761                 item_size = sizeof(struct btrfs_item);
762         } else {
763                 p = offsetof(struct btrfs_node, ptrs);
764                 item_size = sizeof(struct btrfs_key_ptr);
765         }
766
767         while (low < high) {
768                 unsigned long oip;
769                 unsigned long offset;
770                 struct btrfs_disk_key *tmp;
771                 struct btrfs_disk_key unaligned;
772                 int mid;
773
774                 mid = (low + high) / 2;
775                 offset = p + mid * item_size;
776                 oip = offset_in_page(offset);
777
778                 if (oip + key_size <= PAGE_SIZE) {
779                         const unsigned long idx = get_eb_page_index(offset);
780                         char *kaddr = page_address(eb->pages[idx]);
781
782                         oip = get_eb_offset_in_page(eb, offset);
783                         tmp = (struct btrfs_disk_key *)(kaddr + oip);
784                 } else {
785                         read_extent_buffer(eb, &unaligned, offset, key_size);
786                         tmp = &unaligned;
787                 }
788
789                 ret = comp_keys(tmp, key);
790
791                 if (ret < 0)
792                         low = mid + 1;
793                 else if (ret > 0)
794                         high = mid;
795                 else {
796                         *slot = mid;
797                         return 0;
798                 }
799         }
800         *slot = low;
801         return 1;
802 }
803
804 /*
805  * Simple binary search on an extent buffer. Works for both leaves and nodes, and
806  * always searches over the whole range of keys (slot 0 to slot 'nritems - 1').
807  */
808 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
809                      int *slot)
810 {
811         return generic_bin_search(eb, 0, key, slot);
812 }
813
814 static void root_add_used(struct btrfs_root *root, u32 size)
815 {
816         spin_lock(&root->accounting_lock);
817         btrfs_set_root_used(&root->root_item,
818                             btrfs_root_used(&root->root_item) + size);
819         spin_unlock(&root->accounting_lock);
820 }
821
822 static void root_sub_used(struct btrfs_root *root, u32 size)
823 {
824         spin_lock(&root->accounting_lock);
825         btrfs_set_root_used(&root->root_item,
826                             btrfs_root_used(&root->root_item) - size);
827         spin_unlock(&root->accounting_lock);
828 }
829
830 /* given a node and slot number, this reads the blocks it points to.  The
831  * extent buffer is returned with a reference taken (but unlocked).
832  */
833 struct extent_buffer *btrfs_read_node_slot(struct extent_buffer *parent,
834                                            int slot)
835 {
836         int level = btrfs_header_level(parent);
837         struct extent_buffer *eb;
838         struct btrfs_key first_key;
839
840         if (slot < 0 || slot >= btrfs_header_nritems(parent))
841                 return ERR_PTR(-ENOENT);
842
843         BUG_ON(level == 0);
844
845         btrfs_node_key_to_cpu(parent, &first_key, slot);
846         eb = read_tree_block(parent->fs_info, btrfs_node_blockptr(parent, slot),
847                              btrfs_header_owner(parent),
848                              btrfs_node_ptr_generation(parent, slot),
849                              level - 1, &first_key);
850         if (IS_ERR(eb))
851                 return eb;
852         if (!extent_buffer_uptodate(eb)) {
853                 free_extent_buffer(eb);
854                 return ERR_PTR(-EIO);
855         }
856
857         return eb;
858 }
859
860 /*
861  * node level balancing, used to make sure nodes are in proper order for
862  * item deletion.  We balance from the top down, so we have to make sure
863  * that a deletion won't leave an node completely empty later on.
864  */
865 static noinline int balance_level(struct btrfs_trans_handle *trans,
866                          struct btrfs_root *root,
867                          struct btrfs_path *path, int level)
868 {
869         struct btrfs_fs_info *fs_info = root->fs_info;
870         struct extent_buffer *right = NULL;
871         struct extent_buffer *mid;
872         struct extent_buffer *left = NULL;
873         struct extent_buffer *parent = NULL;
874         int ret = 0;
875         int wret;
876         int pslot;
877         int orig_slot = path->slots[level];
878         u64 orig_ptr;
879
880         ASSERT(level > 0);
881
882         mid = path->nodes[level];
883
884         WARN_ON(path->locks[level] != BTRFS_WRITE_LOCK);
885         WARN_ON(btrfs_header_generation(mid) != trans->transid);
886
887         orig_ptr = btrfs_node_blockptr(mid, orig_slot);
888
889         if (level < BTRFS_MAX_LEVEL - 1) {
890                 parent = path->nodes[level + 1];
891                 pslot = path->slots[level + 1];
892         }
893
894         /*
895          * deal with the case where there is only one pointer in the root
896          * by promoting the node below to a root
897          */
898         if (!parent) {
899                 struct extent_buffer *child;
900
901                 if (btrfs_header_nritems(mid) != 1)
902                         return 0;
903
904                 /* promote the child to a root */
905                 child = btrfs_read_node_slot(mid, 0);
906                 if (IS_ERR(child)) {
907                         ret = PTR_ERR(child);
908                         btrfs_handle_fs_error(fs_info, ret, NULL);
909                         goto enospc;
910                 }
911
912                 btrfs_tree_lock(child);
913                 ret = btrfs_cow_block(trans, root, child, mid, 0, &child,
914                                       BTRFS_NESTING_COW);
915                 if (ret) {
916                         btrfs_tree_unlock(child);
917                         free_extent_buffer(child);
918                         goto enospc;
919                 }
920
921                 ret = btrfs_tree_mod_log_insert_root(root->node, child, true);
922                 BUG_ON(ret < 0);
923                 rcu_assign_pointer(root->node, child);
924
925                 add_root_to_dirty_list(root);
926                 btrfs_tree_unlock(child);
927
928                 path->locks[level] = 0;
929                 path->nodes[level] = NULL;
930                 btrfs_clean_tree_block(mid);
931                 btrfs_tree_unlock(mid);
932                 /* once for the path */
933                 free_extent_buffer(mid);
934
935                 root_sub_used(root, mid->len);
936                 btrfs_free_tree_block(trans, btrfs_root_id(root), mid, 0, 1);
937                 /* once for the root ptr */
938                 free_extent_buffer_stale(mid);
939                 return 0;
940         }
941         if (btrfs_header_nritems(mid) >
942             BTRFS_NODEPTRS_PER_BLOCK(fs_info) / 4)
943                 return 0;
944
945         left = btrfs_read_node_slot(parent, pslot - 1);
946         if (IS_ERR(left))
947                 left = NULL;
948
949         if (left) {
950                 __btrfs_tree_lock(left, BTRFS_NESTING_LEFT);
951                 wret = btrfs_cow_block(trans, root, left,
952                                        parent, pslot - 1, &left,
953                                        BTRFS_NESTING_LEFT_COW);
954                 if (wret) {
955                         ret = wret;
956                         goto enospc;
957                 }
958         }
959
960         right = btrfs_read_node_slot(parent, pslot + 1);
961         if (IS_ERR(right))
962                 right = NULL;
963
964         if (right) {
965                 __btrfs_tree_lock(right, BTRFS_NESTING_RIGHT);
966                 wret = btrfs_cow_block(trans, root, right,
967                                        parent, pslot + 1, &right,
968                                        BTRFS_NESTING_RIGHT_COW);
969                 if (wret) {
970                         ret = wret;
971                         goto enospc;
972                 }
973         }
974
975         /* first, try to make some room in the middle buffer */
976         if (left) {
977                 orig_slot += btrfs_header_nritems(left);
978                 wret = push_node_left(trans, left, mid, 1);
979                 if (wret < 0)
980                         ret = wret;
981         }
982
983         /*
984          * then try to empty the right most buffer into the middle
985          */
986         if (right) {
987                 wret = push_node_left(trans, mid, right, 1);
988                 if (wret < 0 && wret != -ENOSPC)
989                         ret = wret;
990                 if (btrfs_header_nritems(right) == 0) {
991                         btrfs_clean_tree_block(right);
992                         btrfs_tree_unlock(right);
993                         del_ptr(root, path, level + 1, pslot + 1);
994                         root_sub_used(root, right->len);
995                         btrfs_free_tree_block(trans, btrfs_root_id(root), right,
996                                               0, 1);
997                         free_extent_buffer_stale(right);
998                         right = NULL;
999                 } else {
1000                         struct btrfs_disk_key right_key;
1001                         btrfs_node_key(right, &right_key, 0);
1002                         ret = btrfs_tree_mod_log_insert_key(parent, pslot + 1,
1003                                         BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
1004                         BUG_ON(ret < 0);
1005                         btrfs_set_node_key(parent, &right_key, pslot + 1);
1006                         btrfs_mark_buffer_dirty(parent);
1007                 }
1008         }
1009         if (btrfs_header_nritems(mid) == 1) {
1010                 /*
1011                  * we're not allowed to leave a node with one item in the
1012                  * tree during a delete.  A deletion from lower in the tree
1013                  * could try to delete the only pointer in this node.
1014                  * So, pull some keys from the left.
1015                  * There has to be a left pointer at this point because
1016                  * otherwise we would have pulled some pointers from the
1017                  * right
1018                  */
1019                 if (!left) {
1020                         ret = -EROFS;
1021                         btrfs_handle_fs_error(fs_info, ret, NULL);
1022                         goto enospc;
1023                 }
1024                 wret = balance_node_right(trans, mid, left);
1025                 if (wret < 0) {
1026                         ret = wret;
1027                         goto enospc;
1028                 }
1029                 if (wret == 1) {
1030                         wret = push_node_left(trans, left, mid, 1);
1031                         if (wret < 0)
1032                                 ret = wret;
1033                 }
1034                 BUG_ON(wret == 1);
1035         }
1036         if (btrfs_header_nritems(mid) == 0) {
1037                 btrfs_clean_tree_block(mid);
1038                 btrfs_tree_unlock(mid);
1039                 del_ptr(root, path, level + 1, pslot);
1040                 root_sub_used(root, mid->len);
1041                 btrfs_free_tree_block(trans, btrfs_root_id(root), mid, 0, 1);
1042                 free_extent_buffer_stale(mid);
1043                 mid = NULL;
1044         } else {
1045                 /* update the parent key to reflect our changes */
1046                 struct btrfs_disk_key mid_key;
1047                 btrfs_node_key(mid, &mid_key, 0);
1048                 ret = btrfs_tree_mod_log_insert_key(parent, pslot,
1049                                 BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
1050                 BUG_ON(ret < 0);
1051                 btrfs_set_node_key(parent, &mid_key, pslot);
1052                 btrfs_mark_buffer_dirty(parent);
1053         }
1054
1055         /* update the path */
1056         if (left) {
1057                 if (btrfs_header_nritems(left) > orig_slot) {
1058                         atomic_inc(&left->refs);
1059                         /* left was locked after cow */
1060                         path->nodes[level] = left;
1061                         path->slots[level + 1] -= 1;
1062                         path->slots[level] = orig_slot;
1063                         if (mid) {
1064                                 btrfs_tree_unlock(mid);
1065                                 free_extent_buffer(mid);
1066                         }
1067                 } else {
1068                         orig_slot -= btrfs_header_nritems(left);
1069                         path->slots[level] = orig_slot;
1070                 }
1071         }
1072         /* double check we haven't messed things up */
1073         if (orig_ptr !=
1074             btrfs_node_blockptr(path->nodes[level], path->slots[level]))
1075                 BUG();
1076 enospc:
1077         if (right) {
1078                 btrfs_tree_unlock(right);
1079                 free_extent_buffer(right);
1080         }
1081         if (left) {
1082                 if (path->nodes[level] != left)
1083                         btrfs_tree_unlock(left);
1084                 free_extent_buffer(left);
1085         }
1086         return ret;
1087 }
1088
1089 /* Node balancing for insertion.  Here we only split or push nodes around
1090  * when they are completely full.  This is also done top down, so we
1091  * have to be pessimistic.
1092  */
1093 static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
1094                                           struct btrfs_root *root,
1095                                           struct btrfs_path *path, int level)
1096 {
1097         struct btrfs_fs_info *fs_info = root->fs_info;
1098         struct extent_buffer *right = NULL;
1099         struct extent_buffer *mid;
1100         struct extent_buffer *left = NULL;
1101         struct extent_buffer *parent = NULL;
1102         int ret = 0;
1103         int wret;
1104         int pslot;
1105         int orig_slot = path->slots[level];
1106
1107         if (level == 0)
1108                 return 1;
1109
1110         mid = path->nodes[level];
1111         WARN_ON(btrfs_header_generation(mid) != trans->transid);
1112
1113         if (level < BTRFS_MAX_LEVEL - 1) {
1114                 parent = path->nodes[level + 1];
1115                 pslot = path->slots[level + 1];
1116         }
1117
1118         if (!parent)
1119                 return 1;
1120
1121         left = btrfs_read_node_slot(parent, pslot - 1);
1122         if (IS_ERR(left))
1123                 left = NULL;
1124
1125         /* first, try to make some room in the middle buffer */
1126         if (left) {
1127                 u32 left_nr;
1128
1129                 __btrfs_tree_lock(left, BTRFS_NESTING_LEFT);
1130
1131                 left_nr = btrfs_header_nritems(left);
1132                 if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
1133                         wret = 1;
1134                 } else {
1135                         ret = btrfs_cow_block(trans, root, left, parent,
1136                                               pslot - 1, &left,
1137                                               BTRFS_NESTING_LEFT_COW);
1138                         if (ret)
1139                                 wret = 1;
1140                         else {
1141                                 wret = push_node_left(trans, left, mid, 0);
1142                         }
1143                 }
1144                 if (wret < 0)
1145                         ret = wret;
1146                 if (wret == 0) {
1147                         struct btrfs_disk_key disk_key;
1148                         orig_slot += left_nr;
1149                         btrfs_node_key(mid, &disk_key, 0);
1150                         ret = btrfs_tree_mod_log_insert_key(parent, pslot,
1151                                         BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
1152                         BUG_ON(ret < 0);
1153                         btrfs_set_node_key(parent, &disk_key, pslot);
1154                         btrfs_mark_buffer_dirty(parent);
1155                         if (btrfs_header_nritems(left) > orig_slot) {
1156                                 path->nodes[level] = left;
1157                                 path->slots[level + 1] -= 1;
1158                                 path->slots[level] = orig_slot;
1159                                 btrfs_tree_unlock(mid);
1160                                 free_extent_buffer(mid);
1161                         } else {
1162                                 orig_slot -=
1163                                         btrfs_header_nritems(left);
1164                                 path->slots[level] = orig_slot;
1165                                 btrfs_tree_unlock(left);
1166                                 free_extent_buffer(left);
1167                         }
1168                         return 0;
1169                 }
1170                 btrfs_tree_unlock(left);
1171                 free_extent_buffer(left);
1172         }
1173         right = btrfs_read_node_slot(parent, pslot + 1);
1174         if (IS_ERR(right))
1175                 right = NULL;
1176
1177         /*
1178          * then try to empty the right most buffer into the middle
1179          */
1180         if (right) {
1181                 u32 right_nr;
1182
1183                 __btrfs_tree_lock(right, BTRFS_NESTING_RIGHT);
1184
1185                 right_nr = btrfs_header_nritems(right);
1186                 if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
1187                         wret = 1;
1188                 } else {
1189                         ret = btrfs_cow_block(trans, root, right,
1190                                               parent, pslot + 1,
1191                                               &right, BTRFS_NESTING_RIGHT_COW);
1192                         if (ret)
1193                                 wret = 1;
1194                         else {
1195                                 wret = balance_node_right(trans, right, mid);
1196                         }
1197                 }
1198                 if (wret < 0)
1199                         ret = wret;
1200                 if (wret == 0) {
1201                         struct btrfs_disk_key disk_key;
1202
1203                         btrfs_node_key(right, &disk_key, 0);
1204                         ret = btrfs_tree_mod_log_insert_key(parent, pslot + 1,
1205                                         BTRFS_MOD_LOG_KEY_REPLACE, GFP_NOFS);
1206                         BUG_ON(ret < 0);
1207                         btrfs_set_node_key(parent, &disk_key, pslot + 1);
1208                         btrfs_mark_buffer_dirty(parent);
1209
1210                         if (btrfs_header_nritems(mid) <= orig_slot) {
1211                                 path->nodes[level] = right;
1212                                 path->slots[level + 1] += 1;
1213                                 path->slots[level] = orig_slot -
1214                                         btrfs_header_nritems(mid);
1215                                 btrfs_tree_unlock(mid);
1216                                 free_extent_buffer(mid);
1217                         } else {
1218                                 btrfs_tree_unlock(right);
1219                                 free_extent_buffer(right);
1220                         }
1221                         return 0;
1222                 }
1223                 btrfs_tree_unlock(right);
1224                 free_extent_buffer(right);
1225         }
1226         return 1;
1227 }
1228
1229 /*
1230  * readahead one full node of leaves, finding things that are close
1231  * to the block in 'slot', and triggering ra on them.
1232  */
1233 static void reada_for_search(struct btrfs_fs_info *fs_info,
1234                              struct btrfs_path *path,
1235                              int level, int slot, u64 objectid)
1236 {
1237         struct extent_buffer *node;
1238         struct btrfs_disk_key disk_key;
1239         u32 nritems;
1240         u64 search;
1241         u64 target;
1242         u64 nread = 0;
1243         u64 nread_max;
1244         u32 nr;
1245         u32 blocksize;
1246         u32 nscan = 0;
1247
1248         if (level != 1 && path->reada != READA_FORWARD_ALWAYS)
1249                 return;
1250
1251         if (!path->nodes[level])
1252                 return;
1253
1254         node = path->nodes[level];
1255
1256         /*
1257          * Since the time between visiting leaves is much shorter than the time
1258          * between visiting nodes, limit read ahead of nodes to 1, to avoid too
1259          * much IO at once (possibly random).
1260          */
1261         if (path->reada == READA_FORWARD_ALWAYS) {
1262                 if (level > 1)
1263                         nread_max = node->fs_info->nodesize;
1264                 else
1265                         nread_max = SZ_128K;
1266         } else {
1267                 nread_max = SZ_64K;
1268         }
1269
1270         search = btrfs_node_blockptr(node, slot);
1271         blocksize = fs_info->nodesize;
1272         if (path->reada != READA_FORWARD_ALWAYS) {
1273                 struct extent_buffer *eb;
1274
1275                 eb = find_extent_buffer(fs_info, search);
1276                 if (eb) {
1277                         free_extent_buffer(eb);
1278                         return;
1279                 }
1280         }
1281
1282         target = search;
1283
1284         nritems = btrfs_header_nritems(node);
1285         nr = slot;
1286
1287         while (1) {
1288                 if (path->reada == READA_BACK) {
1289                         if (nr == 0)
1290                                 break;
1291                         nr--;
1292                 } else if (path->reada == READA_FORWARD ||
1293                            path->reada == READA_FORWARD_ALWAYS) {
1294                         nr++;
1295                         if (nr >= nritems)
1296                                 break;
1297                 }
1298                 if (path->reada == READA_BACK && objectid) {
1299                         btrfs_node_key(node, &disk_key, nr);
1300                         if (btrfs_disk_key_objectid(&disk_key) != objectid)
1301                                 break;
1302                 }
1303                 search = btrfs_node_blockptr(node, nr);
1304                 if (path->reada == READA_FORWARD_ALWAYS ||
1305                     (search <= target && target - search <= 65536) ||
1306                     (search > target && search - target <= 65536)) {
1307                         btrfs_readahead_node_child(node, nr);
1308                         nread += blocksize;
1309                 }
1310                 nscan++;
1311                 if (nread > nread_max || nscan > 32)
1312                         break;
1313         }
1314 }
1315
1316 static noinline void reada_for_balance(struct btrfs_path *path, int level)
1317 {
1318         struct extent_buffer *parent;
1319         int slot;
1320         int nritems;
1321
1322         parent = path->nodes[level + 1];
1323         if (!parent)
1324                 return;
1325
1326         nritems = btrfs_header_nritems(parent);
1327         slot = path->slots[level + 1];
1328
1329         if (slot > 0)
1330                 btrfs_readahead_node_child(parent, slot - 1);
1331         if (slot + 1 < nritems)
1332                 btrfs_readahead_node_child(parent, slot + 1);
1333 }
1334
1335
1336 /*
1337  * when we walk down the tree, it is usually safe to unlock the higher layers
1338  * in the tree.  The exceptions are when our path goes through slot 0, because
1339  * operations on the tree might require changing key pointers higher up in the
1340  * tree.
1341  *
1342  * callers might also have set path->keep_locks, which tells this code to keep
1343  * the lock if the path points to the last slot in the block.  This is part of
1344  * walking through the tree, and selecting the next slot in the higher block.
1345  *
1346  * lowest_unlock sets the lowest level in the tree we're allowed to unlock.  so
1347  * if lowest_unlock is 1, level 0 won't be unlocked
1348  */
1349 static noinline void unlock_up(struct btrfs_path *path, int level,
1350                                int lowest_unlock, int min_write_lock_level,
1351                                int *write_lock_level)
1352 {
1353         int i;
1354         int skip_level = level;
1355         bool check_skip = true;
1356
1357         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
1358                 if (!path->nodes[i])
1359                         break;
1360                 if (!path->locks[i])
1361                         break;
1362
1363                 if (check_skip) {
1364                         if (path->slots[i] == 0) {
1365                                 skip_level = i + 1;
1366                                 continue;
1367                         }
1368
1369                         if (path->keep_locks) {
1370                                 u32 nritems;
1371
1372                                 nritems = btrfs_header_nritems(path->nodes[i]);
1373                                 if (nritems < 1 || path->slots[i] >= nritems - 1) {
1374                                         skip_level = i + 1;
1375                                         continue;
1376                                 }
1377                         }
1378                 }
1379
1380                 if (i >= lowest_unlock && i > skip_level) {
1381                         check_skip = false;
1382                         btrfs_tree_unlock_rw(path->nodes[i], path->locks[i]);
1383                         path->locks[i] = 0;
1384                         if (write_lock_level &&
1385                             i > min_write_lock_level &&
1386                             i <= *write_lock_level) {
1387                                 *write_lock_level = i - 1;
1388                         }
1389                 }
1390         }
1391 }
1392
1393 /*
1394  * Helper function for btrfs_search_slot() and other functions that do a search
1395  * on a btree. The goal is to find a tree block in the cache (the radix tree at
1396  * fs_info->buffer_radix), but if we can't find it, or it's not up to date, read
1397  * its pages from disk.
1398  *
1399  * Returns -EAGAIN, with the path unlocked, if the caller needs to repeat the
1400  * whole btree search, starting again from the current root node.
1401  */
1402 static int
1403 read_block_for_search(struct btrfs_root *root, struct btrfs_path *p,
1404                       struct extent_buffer **eb_ret, int level, int slot,
1405                       const struct btrfs_key *key)
1406 {
1407         struct btrfs_fs_info *fs_info = root->fs_info;
1408         u64 blocknr;
1409         u64 gen;
1410         struct extent_buffer *tmp;
1411         struct btrfs_key first_key;
1412         int ret;
1413         int parent_level;
1414         bool unlock_up;
1415
1416         unlock_up = ((level + 1 < BTRFS_MAX_LEVEL) && p->locks[level + 1]);
1417         blocknr = btrfs_node_blockptr(*eb_ret, slot);
1418         gen = btrfs_node_ptr_generation(*eb_ret, slot);
1419         parent_level = btrfs_header_level(*eb_ret);
1420         btrfs_node_key_to_cpu(*eb_ret, &first_key, slot);
1421
1422         /*
1423          * If we need to read an extent buffer from disk and we are holding locks
1424          * on upper level nodes, we unlock all the upper nodes before reading the
1425          * extent buffer, and then return -EAGAIN to the caller as it needs to
1426          * restart the search. We don't release the lock on the current level
1427          * because we need to walk this node to figure out which blocks to read.
1428          */
1429         tmp = find_extent_buffer(fs_info, blocknr);
1430         if (tmp) {
1431                 if (p->reada == READA_FORWARD_ALWAYS)
1432                         reada_for_search(fs_info, p, level, slot, key->objectid);
1433
1434                 /* first we do an atomic uptodate check */
1435                 if (btrfs_buffer_uptodate(tmp, gen, 1) > 0) {
1436                         /*
1437                          * Do extra check for first_key, eb can be stale due to
1438                          * being cached, read from scrub, or have multiple
1439                          * parents (shared tree blocks).
1440                          */
1441                         if (btrfs_verify_level_key(tmp,
1442                                         parent_level - 1, &first_key, gen)) {
1443                                 free_extent_buffer(tmp);
1444                                 return -EUCLEAN;
1445                         }
1446                         *eb_ret = tmp;
1447                         return 0;
1448                 }
1449
1450                 if (unlock_up)
1451                         btrfs_unlock_up_safe(p, level + 1);
1452
1453                 /* now we're allowed to do a blocking uptodate check */
1454                 ret = btrfs_read_extent_buffer(tmp, gen, parent_level - 1, &first_key);
1455                 if (ret) {
1456                         free_extent_buffer(tmp);
1457                         btrfs_release_path(p);
1458                         return -EIO;
1459                 }
1460                 if (btrfs_check_eb_owner(tmp, root->root_key.objectid)) {
1461                         free_extent_buffer(tmp);
1462                         btrfs_release_path(p);
1463                         return -EUCLEAN;
1464                 }
1465
1466                 if (unlock_up)
1467                         ret = -EAGAIN;
1468
1469                 goto out;
1470         }
1471
1472         if (unlock_up) {
1473                 btrfs_unlock_up_safe(p, level + 1);
1474                 ret = -EAGAIN;
1475         } else {
1476                 ret = 0;
1477         }
1478
1479         if (p->reada != READA_NONE)
1480                 reada_for_search(fs_info, p, level, slot, key->objectid);
1481
1482         tmp = read_tree_block(fs_info, blocknr, root->root_key.objectid,
1483                               gen, parent_level - 1, &first_key);
1484         if (IS_ERR(tmp)) {
1485                 btrfs_release_path(p);
1486                 return PTR_ERR(tmp);
1487         }
1488         /*
1489          * If the read above didn't mark this buffer up to date,
1490          * it will never end up being up to date.  Set ret to EIO now
1491          * and give up so that our caller doesn't loop forever
1492          * on our EAGAINs.
1493          */
1494         if (!extent_buffer_uptodate(tmp))
1495                 ret = -EIO;
1496
1497 out:
1498         if (ret == 0) {
1499                 *eb_ret = tmp;
1500         } else {
1501                 free_extent_buffer(tmp);
1502                 btrfs_release_path(p);
1503         }
1504
1505         return ret;
1506 }
1507
1508 /*
1509  * helper function for btrfs_search_slot.  This does all of the checks
1510  * for node-level blocks and does any balancing required based on
1511  * the ins_len.
1512  *
1513  * If no extra work was required, zero is returned.  If we had to
1514  * drop the path, -EAGAIN is returned and btrfs_search_slot must
1515  * start over
1516  */
1517 static int
1518 setup_nodes_for_search(struct btrfs_trans_handle *trans,
1519                        struct btrfs_root *root, struct btrfs_path *p,
1520                        struct extent_buffer *b, int level, int ins_len,
1521                        int *write_lock_level)
1522 {
1523         struct btrfs_fs_info *fs_info = root->fs_info;
1524         int ret = 0;
1525
1526         if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >=
1527             BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 3) {
1528
1529                 if (*write_lock_level < level + 1) {
1530                         *write_lock_level = level + 1;
1531                         btrfs_release_path(p);
1532                         return -EAGAIN;
1533                 }
1534
1535                 reada_for_balance(p, level);
1536                 ret = split_node(trans, root, p, level);
1537
1538                 b = p->nodes[level];
1539         } else if (ins_len < 0 && btrfs_header_nritems(b) <
1540                    BTRFS_NODEPTRS_PER_BLOCK(fs_info) / 2) {
1541
1542                 if (*write_lock_level < level + 1) {
1543                         *write_lock_level = level + 1;
1544                         btrfs_release_path(p);
1545                         return -EAGAIN;
1546                 }
1547
1548                 reada_for_balance(p, level);
1549                 ret = balance_level(trans, root, p, level);
1550                 if (ret)
1551                         return ret;
1552
1553                 b = p->nodes[level];
1554                 if (!b) {
1555                         btrfs_release_path(p);
1556                         return -EAGAIN;
1557                 }
1558                 BUG_ON(btrfs_header_nritems(b) == 1);
1559         }
1560         return ret;
1561 }
1562
1563 int btrfs_find_item(struct btrfs_root *fs_root, struct btrfs_path *path,
1564                 u64 iobjectid, u64 ioff, u8 key_type,
1565                 struct btrfs_key *found_key)
1566 {
1567         int ret;
1568         struct btrfs_key key;
1569         struct extent_buffer *eb;
1570
1571         ASSERT(path);
1572         ASSERT(found_key);
1573
1574         key.type = key_type;
1575         key.objectid = iobjectid;
1576         key.offset = ioff;
1577
1578         ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
1579         if (ret < 0)
1580                 return ret;
1581
1582         eb = path->nodes[0];
1583         if (ret && path->slots[0] >= btrfs_header_nritems(eb)) {
1584                 ret = btrfs_next_leaf(fs_root, path);
1585                 if (ret)
1586                         return ret;
1587                 eb = path->nodes[0];
1588         }
1589
1590         btrfs_item_key_to_cpu(eb, found_key, path->slots[0]);
1591         if (found_key->type != key.type ||
1592                         found_key->objectid != key.objectid)
1593                 return 1;
1594
1595         return 0;
1596 }
1597
1598 static struct extent_buffer *btrfs_search_slot_get_root(struct btrfs_root *root,
1599                                                         struct btrfs_path *p,
1600                                                         int write_lock_level)
1601 {
1602         struct extent_buffer *b;
1603         int root_lock = 0;
1604         int level = 0;
1605
1606         if (p->search_commit_root) {
1607                 b = root->commit_root;
1608                 atomic_inc(&b->refs);
1609                 level = btrfs_header_level(b);
1610                 /*
1611                  * Ensure that all callers have set skip_locking when
1612                  * p->search_commit_root = 1.
1613                  */
1614                 ASSERT(p->skip_locking == 1);
1615
1616                 goto out;
1617         }
1618
1619         if (p->skip_locking) {
1620                 b = btrfs_root_node(root);
1621                 level = btrfs_header_level(b);
1622                 goto out;
1623         }
1624
1625         /* We try very hard to do read locks on the root */
1626         root_lock = BTRFS_READ_LOCK;
1627
1628         /*
1629          * If the level is set to maximum, we can skip trying to get the read
1630          * lock.
1631          */
1632         if (write_lock_level < BTRFS_MAX_LEVEL) {
1633                 /*
1634                  * We don't know the level of the root node until we actually
1635                  * have it read locked
1636                  */
1637                 b = btrfs_read_lock_root_node(root);
1638                 level = btrfs_header_level(b);
1639                 if (level > write_lock_level)
1640                         goto out;
1641
1642                 /* Whoops, must trade for write lock */
1643                 btrfs_tree_read_unlock(b);
1644                 free_extent_buffer(b);
1645         }
1646
1647         b = btrfs_lock_root_node(root);
1648         root_lock = BTRFS_WRITE_LOCK;
1649
1650         /* The level might have changed, check again */
1651         level = btrfs_header_level(b);
1652
1653 out:
1654         /*
1655          * The root may have failed to write out at some point, and thus is no
1656          * longer valid, return an error in this case.
1657          */
1658         if (!extent_buffer_uptodate(b)) {
1659                 if (root_lock)
1660                         btrfs_tree_unlock_rw(b, root_lock);
1661                 free_extent_buffer(b);
1662                 return ERR_PTR(-EIO);
1663         }
1664
1665         p->nodes[level] = b;
1666         if (!p->skip_locking)
1667                 p->locks[level] = root_lock;
1668         /*
1669          * Callers are responsible for dropping b's references.
1670          */
1671         return b;
1672 }
1673
1674 /*
1675  * Replace the extent buffer at the lowest level of the path with a cloned
1676  * version. The purpose is to be able to use it safely, after releasing the
1677  * commit root semaphore, even if relocation is happening in parallel, the
1678  * transaction used for relocation is committed and the extent buffer is
1679  * reallocated in the next transaction.
1680  *
1681  * This is used in a context where the caller does not prevent transaction
1682  * commits from happening, either by holding a transaction handle or holding
1683  * some lock, while it's doing searches through a commit root.
1684  * At the moment it's only used for send operations.
1685  */
1686 static int finish_need_commit_sem_search(struct btrfs_path *path)
1687 {
1688         const int i = path->lowest_level;
1689         const int slot = path->slots[i];
1690         struct extent_buffer *lowest = path->nodes[i];
1691         struct extent_buffer *clone;
1692
1693         ASSERT(path->need_commit_sem);
1694
1695         if (!lowest)
1696                 return 0;
1697
1698         lockdep_assert_held_read(&lowest->fs_info->commit_root_sem);
1699
1700         clone = btrfs_clone_extent_buffer(lowest);
1701         if (!clone)
1702                 return -ENOMEM;
1703
1704         btrfs_release_path(path);
1705         path->nodes[i] = clone;
1706         path->slots[i] = slot;
1707
1708         return 0;
1709 }
1710
1711 static inline int search_for_key_slot(struct extent_buffer *eb,
1712                                       int search_low_slot,
1713                                       const struct btrfs_key *key,
1714                                       int prev_cmp,
1715                                       int *slot)
1716 {
1717         /*
1718          * If a previous call to btrfs_bin_search() on a parent node returned an
1719          * exact match (prev_cmp == 0), we can safely assume the target key will
1720          * always be at slot 0 on lower levels, since each key pointer
1721          * (struct btrfs_key_ptr) refers to the lowest key accessible from the
1722          * subtree it points to. Thus we can skip searching lower levels.
1723          */
1724         if (prev_cmp == 0) {
1725                 *slot = 0;
1726                 return 0;
1727         }
1728
1729         return generic_bin_search(eb, search_low_slot, key, slot);
1730 }
1731
1732 static int search_leaf(struct btrfs_trans_handle *trans,
1733                        struct btrfs_root *root,
1734                        const struct btrfs_key *key,
1735                        struct btrfs_path *path,
1736                        int ins_len,
1737                        int prev_cmp)
1738 {
1739         struct extent_buffer *leaf = path->nodes[0];
1740         int leaf_free_space = -1;
1741         int search_low_slot = 0;
1742         int ret;
1743         bool do_bin_search = true;
1744
1745         /*
1746          * If we are doing an insertion, the leaf has enough free space and the
1747          * destination slot for the key is not slot 0, then we can unlock our
1748          * write lock on the parent, and any other upper nodes, before doing the
1749          * binary search on the leaf (with search_for_key_slot()), allowing other
1750          * tasks to lock the parent and any other upper nodes.
1751          */
1752         if (ins_len > 0) {
1753                 /*
1754                  * Cache the leaf free space, since we will need it later and it
1755                  * will not change until then.
1756                  */
1757                 leaf_free_space = btrfs_leaf_free_space(leaf);
1758
1759                 /*
1760                  * !path->locks[1] means we have a single node tree, the leaf is
1761                  * the root of the tree.
1762                  */
1763                 if (path->locks[1] && leaf_free_space >= ins_len) {
1764                         struct btrfs_disk_key first_key;
1765
1766                         ASSERT(btrfs_header_nritems(leaf) > 0);
1767                         btrfs_item_key(leaf, &first_key, 0);
1768
1769                         /*
1770                          * Doing the extra comparison with the first key is cheap,
1771                          * taking into account that the first key is very likely
1772                          * already in a cache line because it immediately follows
1773                          * the extent buffer's header and we have recently accessed
1774                          * the header's level field.
1775                          */
1776                         ret = comp_keys(&first_key, key);
1777                         if (ret < 0) {
1778                                 /*
1779                                  * The first key is smaller than the key we want
1780                                  * to insert, so we are safe to unlock all upper
1781                                  * nodes and we have to do the binary search.
1782                                  *
1783                                  * We do use btrfs_unlock_up_safe() and not
1784                                  * unlock_up() because the later does not unlock
1785                                  * nodes with a slot of 0 - we can safely unlock
1786                                  * any node even if its slot is 0 since in this
1787                                  * case the key does not end up at slot 0 of the
1788                                  * leaf and there's no need to split the leaf.
1789                                  */
1790                                 btrfs_unlock_up_safe(path, 1);
1791                                 search_low_slot = 1;
1792                         } else {
1793                                 /*
1794                                  * The first key is >= then the key we want to
1795                                  * insert, so we can skip the binary search as
1796                                  * the target key will be at slot 0.
1797                                  *
1798                                  * We can not unlock upper nodes when the key is
1799                                  * less than the first key, because we will need
1800                                  * to update the key at slot 0 of the parent node
1801                                  * and possibly of other upper nodes too.
1802                                  * If the key matches the first key, then we can
1803                                  * unlock all the upper nodes, using
1804                                  * btrfs_unlock_up_safe() instead of unlock_up()
1805                                  * as stated above.
1806                                  */
1807                                 if (ret == 0)
1808                                         btrfs_unlock_up_safe(path, 1);
1809                                 /*
1810                                  * ret is already 0 or 1, matching the result of
1811                                  * a btrfs_bin_search() call, so there is no need
1812                                  * to adjust it.
1813                                  */
1814                                 do_bin_search = false;
1815                                 path->slots[0] = 0;
1816                         }
1817                 }
1818         }
1819
1820         if (do_bin_search) {
1821                 ret = search_for_key_slot(leaf, search_low_slot, key,
1822                                           prev_cmp, &path->slots[0]);
1823                 if (ret < 0)
1824                         return ret;
1825         }
1826
1827         if (ins_len > 0) {
1828                 /*
1829                  * Item key already exists. In this case, if we are allowed to
1830                  * insert the item (for example, in dir_item case, item key
1831                  * collision is allowed), it will be merged with the original
1832                  * item. Only the item size grows, no new btrfs item will be
1833                  * added. If search_for_extension is not set, ins_len already
1834                  * accounts the size btrfs_item, deduct it here so leaf space
1835                  * check will be correct.
1836                  */
1837                 if (ret == 0 && !path->search_for_extension) {
1838                         ASSERT(ins_len >= sizeof(struct btrfs_item));
1839                         ins_len -= sizeof(struct btrfs_item);
1840                 }
1841
1842                 ASSERT(leaf_free_space >= 0);
1843
1844                 if (leaf_free_space < ins_len) {
1845                         int err;
1846
1847                         err = split_leaf(trans, root, key, path, ins_len,
1848                                          (ret == 0));
1849                         ASSERT(err <= 0);
1850                         if (WARN_ON(err > 0))
1851                                 err = -EUCLEAN;
1852                         if (err)
1853                                 ret = err;
1854                 }
1855         }
1856
1857         return ret;
1858 }
1859
1860 /*
1861  * btrfs_search_slot - look for a key in a tree and perform necessary
1862  * modifications to preserve tree invariants.
1863  *
1864  * @trans:      Handle of transaction, used when modifying the tree
1865  * @p:          Holds all btree nodes along the search path
1866  * @root:       The root node of the tree
1867  * @key:        The key we are looking for
1868  * @ins_len:    Indicates purpose of search:
1869  *              >0  for inserts it's size of item inserted (*)
1870  *              <0  for deletions
1871  *               0  for plain searches, not modifying the tree
1872  *
1873  *              (*) If size of item inserted doesn't include
1874  *              sizeof(struct btrfs_item), then p->search_for_extension must
1875  *              be set.
1876  * @cow:        boolean should CoW operations be performed. Must always be 1
1877  *              when modifying the tree.
1878  *
1879  * If @ins_len > 0, nodes and leaves will be split as we walk down the tree.
1880  * If @ins_len < 0, nodes will be merged as we walk down the tree (if possible)
1881  *
1882  * If @key is found, 0 is returned and you can find the item in the leaf level
1883  * of the path (level 0)
1884  *
1885  * If @key isn't found, 1 is returned and the leaf level of the path (level 0)
1886  * points to the slot where it should be inserted
1887  *
1888  * If an error is encountered while searching the tree a negative error number
1889  * is returned
1890  */
1891 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root *root,
1892                       const struct btrfs_key *key, struct btrfs_path *p,
1893                       int ins_len, int cow)
1894 {
1895         struct btrfs_fs_info *fs_info = root->fs_info;
1896         struct extent_buffer *b;
1897         int slot;
1898         int ret;
1899         int err;
1900         int level;
1901         int lowest_unlock = 1;
1902         /* everything at write_lock_level or lower must be write locked */
1903         int write_lock_level = 0;
1904         u8 lowest_level = 0;
1905         int min_write_lock_level;
1906         int prev_cmp;
1907
1908         lowest_level = p->lowest_level;
1909         WARN_ON(lowest_level && ins_len > 0);
1910         WARN_ON(p->nodes[0] != NULL);
1911         BUG_ON(!cow && ins_len);
1912
1913         if (ins_len < 0) {
1914                 lowest_unlock = 2;
1915
1916                 /* when we are removing items, we might have to go up to level
1917                  * two as we update tree pointers  Make sure we keep write
1918                  * for those levels as well
1919                  */
1920                 write_lock_level = 2;
1921         } else if (ins_len > 0) {
1922                 /*
1923                  * for inserting items, make sure we have a write lock on
1924                  * level 1 so we can update keys
1925                  */
1926                 write_lock_level = 1;
1927         }
1928
1929         if (!cow)
1930                 write_lock_level = -1;
1931
1932         if (cow && (p->keep_locks || p->lowest_level))
1933                 write_lock_level = BTRFS_MAX_LEVEL;
1934
1935         min_write_lock_level = write_lock_level;
1936
1937         if (p->need_commit_sem) {
1938                 ASSERT(p->search_commit_root);
1939                 down_read(&fs_info->commit_root_sem);
1940         }
1941
1942 again:
1943         prev_cmp = -1;
1944         b = btrfs_search_slot_get_root(root, p, write_lock_level);
1945         if (IS_ERR(b)) {
1946                 ret = PTR_ERR(b);
1947                 goto done;
1948         }
1949
1950         while (b) {
1951                 int dec = 0;
1952
1953                 level = btrfs_header_level(b);
1954
1955                 if (cow) {
1956                         bool last_level = (level == (BTRFS_MAX_LEVEL - 1));
1957
1958                         /*
1959                          * if we don't really need to cow this block
1960                          * then we don't want to set the path blocking,
1961                          * so we test it here
1962                          */
1963                         if (!should_cow_block(trans, root, b))
1964                                 goto cow_done;
1965
1966                         /*
1967                          * must have write locks on this node and the
1968                          * parent
1969                          */
1970                         if (level > write_lock_level ||
1971                             (level + 1 > write_lock_level &&
1972                             level + 1 < BTRFS_MAX_LEVEL &&
1973                             p->nodes[level + 1])) {
1974                                 write_lock_level = level + 1;
1975                                 btrfs_release_path(p);
1976                                 goto again;
1977                         }
1978
1979                         if (last_level)
1980                                 err = btrfs_cow_block(trans, root, b, NULL, 0,
1981                                                       &b,
1982                                                       BTRFS_NESTING_COW);
1983                         else
1984                                 err = btrfs_cow_block(trans, root, b,
1985                                                       p->nodes[level + 1],
1986                                                       p->slots[level + 1], &b,
1987                                                       BTRFS_NESTING_COW);
1988                         if (err) {
1989                                 ret = err;
1990                                 goto done;
1991                         }
1992                 }
1993 cow_done:
1994                 p->nodes[level] = b;
1995
1996                 /*
1997                  * we have a lock on b and as long as we aren't changing
1998                  * the tree, there is no way to for the items in b to change.
1999                  * It is safe to drop the lock on our parent before we
2000                  * go through the expensive btree search on b.
2001                  *
2002                  * If we're inserting or deleting (ins_len != 0), then we might
2003                  * be changing slot zero, which may require changing the parent.
2004                  * So, we can't drop the lock until after we know which slot
2005                  * we're operating on.
2006                  */
2007                 if (!ins_len && !p->keep_locks) {
2008                         int u = level + 1;
2009
2010                         if (u < BTRFS_MAX_LEVEL && p->locks[u]) {
2011                                 btrfs_tree_unlock_rw(p->nodes[u], p->locks[u]);
2012                                 p->locks[u] = 0;
2013                         }
2014                 }
2015
2016                 if (level == 0) {
2017                         if (ins_len > 0)
2018                                 ASSERT(write_lock_level >= 1);
2019
2020                         ret = search_leaf(trans, root, key, p, ins_len, prev_cmp);
2021                         if (!p->search_for_split)
2022                                 unlock_up(p, level, lowest_unlock,
2023                                           min_write_lock_level, NULL);
2024                         goto done;
2025                 }
2026
2027                 ret = search_for_key_slot(b, 0, key, prev_cmp, &slot);
2028                 if (ret < 0)
2029                         goto done;
2030                 prev_cmp = ret;
2031
2032                 if (ret && slot > 0) {
2033                         dec = 1;
2034                         slot--;
2035                 }
2036                 p->slots[level] = slot;
2037                 err = setup_nodes_for_search(trans, root, p, b, level, ins_len,
2038                                              &write_lock_level);
2039                 if (err == -EAGAIN)
2040                         goto again;
2041                 if (err) {
2042                         ret = err;
2043                         goto done;
2044                 }
2045                 b = p->nodes[level];
2046                 slot = p->slots[level];
2047
2048                 /*
2049                  * Slot 0 is special, if we change the key we have to update
2050                  * the parent pointer which means we must have a write lock on
2051                  * the parent
2052                  */
2053                 if (slot == 0 && ins_len && write_lock_level < level + 1) {
2054                         write_lock_level = level + 1;
2055                         btrfs_release_path(p);
2056                         goto again;
2057                 }
2058
2059                 unlock_up(p, level, lowest_unlock, min_write_lock_level,
2060                           &write_lock_level);
2061
2062                 if (level == lowest_level) {
2063                         if (dec)
2064                                 p->slots[level]++;
2065                         goto done;
2066                 }
2067
2068                 err = read_block_for_search(root, p, &b, level, slot, key);
2069                 if (err == -EAGAIN)
2070                         goto again;
2071                 if (err) {
2072                         ret = err;
2073                         goto done;
2074                 }
2075
2076                 if (!p->skip_locking) {
2077                         level = btrfs_header_level(b);
2078                         if (level <= write_lock_level) {
2079                                 btrfs_tree_lock(b);
2080                                 p->locks[level] = BTRFS_WRITE_LOCK;
2081                         } else {
2082                                 btrfs_tree_read_lock(b);
2083                                 p->locks[level] = BTRFS_READ_LOCK;
2084                         }
2085                         p->nodes[level] = b;
2086                 }
2087         }
2088         ret = 1;
2089 done:
2090         if (ret < 0 && !p->skip_release_on_error)
2091                 btrfs_release_path(p);
2092
2093         if (p->need_commit_sem) {
2094                 int ret2;
2095
2096                 ret2 = finish_need_commit_sem_search(p);
2097                 up_read(&fs_info->commit_root_sem);
2098                 if (ret2)
2099                         ret = ret2;
2100         }
2101
2102         return ret;
2103 }
2104 ALLOW_ERROR_INJECTION(btrfs_search_slot, ERRNO);
2105
2106 /*
2107  * Like btrfs_search_slot, this looks for a key in the given tree. It uses the
2108  * current state of the tree together with the operations recorded in the tree
2109  * modification log to search for the key in a previous version of this tree, as
2110  * denoted by the time_seq parameter.
2111  *
2112  * Naturally, there is no support for insert, delete or cow operations.
2113  *
2114  * The resulting path and return value will be set up as if we called
2115  * btrfs_search_slot at that point in time with ins_len and cow both set to 0.
2116  */
2117 int btrfs_search_old_slot(struct btrfs_root *root, const struct btrfs_key *key,
2118                           struct btrfs_path *p, u64 time_seq)
2119 {
2120         struct btrfs_fs_info *fs_info = root->fs_info;
2121         struct extent_buffer *b;
2122         int slot;
2123         int ret;
2124         int err;
2125         int level;
2126         int lowest_unlock = 1;
2127         u8 lowest_level = 0;
2128
2129         lowest_level = p->lowest_level;
2130         WARN_ON(p->nodes[0] != NULL);
2131
2132         if (p->search_commit_root) {
2133                 BUG_ON(time_seq);
2134                 return btrfs_search_slot(NULL, root, key, p, 0, 0);
2135         }
2136
2137 again:
2138         b = btrfs_get_old_root(root, time_seq);
2139         if (!b) {
2140                 ret = -EIO;
2141                 goto done;
2142         }
2143         level = btrfs_header_level(b);
2144         p->locks[level] = BTRFS_READ_LOCK;
2145
2146         while (b) {
2147                 int dec = 0;
2148
2149                 level = btrfs_header_level(b);
2150                 p->nodes[level] = b;
2151
2152                 /*
2153                  * we have a lock on b and as long as we aren't changing
2154                  * the tree, there is no way to for the items in b to change.
2155                  * It is safe to drop the lock on our parent before we
2156                  * go through the expensive btree search on b.
2157                  */
2158                 btrfs_unlock_up_safe(p, level + 1);
2159
2160                 ret = btrfs_bin_search(b, key, &slot);
2161                 if (ret < 0)
2162                         goto done;
2163
2164                 if (level == 0) {
2165                         p->slots[level] = slot;
2166                         unlock_up(p, level, lowest_unlock, 0, NULL);
2167                         goto done;
2168                 }
2169
2170                 if (ret && slot > 0) {
2171                         dec = 1;
2172                         slot--;
2173                 }
2174                 p->slots[level] = slot;
2175                 unlock_up(p, level, lowest_unlock, 0, NULL);
2176
2177                 if (level == lowest_level) {
2178                         if (dec)
2179                                 p->slots[level]++;
2180                         goto done;
2181                 }
2182
2183                 err = read_block_for_search(root, p, &b, level, slot, key);
2184                 if (err == -EAGAIN)
2185                         goto again;
2186                 if (err) {
2187                         ret = err;
2188                         goto done;
2189                 }
2190
2191                 level = btrfs_header_level(b);
2192                 btrfs_tree_read_lock(b);
2193                 b = btrfs_tree_mod_log_rewind(fs_info, p, b, time_seq);
2194                 if (!b) {
2195                         ret = -ENOMEM;
2196                         goto done;
2197                 }
2198                 p->locks[level] = BTRFS_READ_LOCK;
2199                 p->nodes[level] = b;
2200         }
2201         ret = 1;
2202 done:
2203         if (ret < 0)
2204                 btrfs_release_path(p);
2205
2206         return ret;
2207 }
2208
2209 /*
2210  * helper to use instead of search slot if no exact match is needed but
2211  * instead the next or previous item should be returned.
2212  * When find_higher is true, the next higher item is returned, the next lower
2213  * otherwise.
2214  * When return_any and find_higher are both true, and no higher item is found,
2215  * return the next lower instead.
2216  * When return_any is true and find_higher is false, and no lower item is found,
2217  * return the next higher instead.
2218  * It returns 0 if any item is found, 1 if none is found (tree empty), and
2219  * < 0 on error
2220  */
2221 int btrfs_search_slot_for_read(struct btrfs_root *root,
2222                                const struct btrfs_key *key,
2223                                struct btrfs_path *p, int find_higher,
2224                                int return_any)
2225 {
2226         int ret;
2227         struct extent_buffer *leaf;
2228
2229 again:
2230         ret = btrfs_search_slot(NULL, root, key, p, 0, 0);
2231         if (ret <= 0)
2232                 return ret;
2233         /*
2234          * a return value of 1 means the path is at the position where the
2235          * item should be inserted. Normally this is the next bigger item,
2236          * but in case the previous item is the last in a leaf, path points
2237          * to the first free slot in the previous leaf, i.e. at an invalid
2238          * item.
2239          */
2240         leaf = p->nodes[0];
2241
2242         if (find_higher) {
2243                 if (p->slots[0] >= btrfs_header_nritems(leaf)) {
2244                         ret = btrfs_next_leaf(root, p);
2245                         if (ret <= 0)
2246                                 return ret;
2247                         if (!return_any)
2248                                 return 1;
2249                         /*
2250                          * no higher item found, return the next
2251                          * lower instead
2252                          */
2253                         return_any = 0;
2254                         find_higher = 0;
2255                         btrfs_release_path(p);
2256                         goto again;
2257                 }
2258         } else {
2259                 if (p->slots[0] == 0) {
2260                         ret = btrfs_prev_leaf(root, p);
2261                         if (ret < 0)
2262                                 return ret;
2263                         if (!ret) {
2264                                 leaf = p->nodes[0];
2265                                 if (p->slots[0] == btrfs_header_nritems(leaf))
2266                                         p->slots[0]--;
2267                                 return 0;
2268                         }
2269                         if (!return_any)
2270                                 return 1;
2271                         /*
2272                          * no lower item found, return the next
2273                          * higher instead
2274                          */
2275                         return_any = 0;
2276                         find_higher = 1;
2277                         btrfs_release_path(p);
2278                         goto again;
2279                 } else {
2280                         --p->slots[0];
2281                 }
2282         }
2283         return 0;
2284 }
2285
2286 /*
2287  * Execute search and call btrfs_previous_item to traverse backwards if the item
2288  * was not found.
2289  *
2290  * Return 0 if found, 1 if not found and < 0 if error.
2291  */
2292 int btrfs_search_backwards(struct btrfs_root *root, struct btrfs_key *key,
2293                            struct btrfs_path *path)
2294 {
2295         int ret;
2296
2297         ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
2298         if (ret > 0)
2299                 ret = btrfs_previous_item(root, path, key->objectid, key->type);
2300
2301         if (ret == 0)
2302                 btrfs_item_key_to_cpu(path->nodes[0], key, path->slots[0]);
2303
2304         return ret;
2305 }
2306
2307 /**
2308  * Search for a valid slot for the given path.
2309  *
2310  * @root:       The root node of the tree.
2311  * @key:        Will contain a valid item if found.
2312  * @path:       The starting point to validate the slot.
2313  *
2314  * Return: 0  if the item is valid
2315  *         1  if not found
2316  *         <0 if error.
2317  */
2318 int btrfs_get_next_valid_item(struct btrfs_root *root, struct btrfs_key *key,
2319                               struct btrfs_path *path)
2320 {
2321         while (1) {
2322                 int ret;
2323                 const int slot = path->slots[0];
2324                 const struct extent_buffer *leaf = path->nodes[0];
2325
2326                 /* This is where we start walking the path. */
2327                 if (slot >= btrfs_header_nritems(leaf)) {
2328                         /*
2329                          * If we've reached the last slot in this leaf we need
2330                          * to go to the next leaf and reset the path.
2331                          */
2332                         ret = btrfs_next_leaf(root, path);
2333                         if (ret)
2334                                 return ret;
2335                         continue;
2336                 }
2337                 /* Store the found, valid item in @key. */
2338                 btrfs_item_key_to_cpu(leaf, key, slot);
2339                 break;
2340         }
2341         return 0;
2342 }
2343
2344 /*
2345  * adjust the pointers going up the tree, starting at level
2346  * making sure the right key of each node is points to 'key'.
2347  * This is used after shifting pointers to the left, so it stops
2348  * fixing up pointers when a given leaf/node is not in slot 0 of the
2349  * higher levels
2350  *
2351  */
2352 static void fixup_low_keys(struct btrfs_path *path,
2353                            struct btrfs_disk_key *key, int level)
2354 {
2355         int i;
2356         struct extent_buffer *t;
2357         int ret;
2358
2359         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
2360                 int tslot = path->slots[i];
2361
2362                 if (!path->nodes[i])
2363                         break;
2364                 t = path->nodes[i];
2365                 ret = btrfs_tree_mod_log_insert_key(t, tslot,
2366                                 BTRFS_MOD_LOG_KEY_REPLACE, GFP_ATOMIC);
2367                 BUG_ON(ret < 0);
2368                 btrfs_set_node_key(t, key, tslot);
2369                 btrfs_mark_buffer_dirty(path->nodes[i]);
2370                 if (tslot != 0)
2371                         break;
2372         }
2373 }
2374
2375 /*
2376  * update item key.
2377  *
2378  * This function isn't completely safe. It's the caller's responsibility
2379  * that the new key won't break the order
2380  */
2381 void btrfs_set_item_key_safe(struct btrfs_fs_info *fs_info,
2382                              struct btrfs_path *path,
2383                              const struct btrfs_key *new_key)
2384 {
2385         struct btrfs_disk_key disk_key;
2386         struct extent_buffer *eb;
2387         int slot;
2388
2389         eb = path->nodes[0];
2390         slot = path->slots[0];
2391         if (slot > 0) {
2392                 btrfs_item_key(eb, &disk_key, slot - 1);
2393                 if (unlikely(comp_keys(&disk_key, new_key) >= 0)) {
2394                         btrfs_crit(fs_info,
2395                 "slot %u key (%llu %u %llu) new key (%llu %u %llu)",
2396                                    slot, btrfs_disk_key_objectid(&disk_key),
2397                                    btrfs_disk_key_type(&disk_key),
2398                                    btrfs_disk_key_offset(&disk_key),
2399                                    new_key->objectid, new_key->type,
2400                                    new_key->offset);
2401                         btrfs_print_leaf(eb);
2402                         BUG();
2403                 }
2404         }
2405         if (slot < btrfs_header_nritems(eb) - 1) {
2406                 btrfs_item_key(eb, &disk_key, slot + 1);
2407                 if (unlikely(comp_keys(&disk_key, new_key) <= 0)) {
2408                         btrfs_crit(fs_info,
2409                 "slot %u key (%llu %u %llu) new key (%llu %u %llu)",
2410                                    slot, btrfs_disk_key_objectid(&disk_key),
2411                                    btrfs_disk_key_type(&disk_key),
2412                                    btrfs_disk_key_offset(&disk_key),
2413                                    new_key->objectid, new_key->type,
2414                                    new_key->offset);
2415                         btrfs_print_leaf(eb);
2416                         BUG();
2417                 }
2418         }
2419
2420         btrfs_cpu_key_to_disk(&disk_key, new_key);
2421         btrfs_set_item_key(eb, &disk_key, slot);
2422         btrfs_mark_buffer_dirty(eb);
2423         if (slot == 0)
2424                 fixup_low_keys(path, &disk_key, 1);
2425 }
2426
2427 /*
2428  * Check key order of two sibling extent buffers.
2429  *
2430  * Return true if something is wrong.
2431  * Return false if everything is fine.
2432  *
2433  * Tree-checker only works inside one tree block, thus the following
2434  * corruption can not be detected by tree-checker:
2435  *
2436  * Leaf @left                   | Leaf @right
2437  * --------------------------------------------------------------
2438  * | 1 | 2 | 3 | 4 | 5 | f6 |   | 7 | 8 |
2439  *
2440  * Key f6 in leaf @left itself is valid, but not valid when the next
2441  * key in leaf @right is 7.
2442  * This can only be checked at tree block merge time.
2443  * And since tree checker has ensured all key order in each tree block
2444  * is correct, we only need to bother the last key of @left and the first
2445  * key of @right.
2446  */
2447 static bool check_sibling_keys(struct extent_buffer *left,
2448                                struct extent_buffer *right)
2449 {
2450         struct btrfs_key left_last;
2451         struct btrfs_key right_first;
2452         int level = btrfs_header_level(left);
2453         int nr_left = btrfs_header_nritems(left);
2454         int nr_right = btrfs_header_nritems(right);
2455
2456         /* No key to check in one of the tree blocks */
2457         if (!nr_left || !nr_right)
2458                 return false;
2459
2460         if (level) {
2461                 btrfs_node_key_to_cpu(left, &left_last, nr_left - 1);
2462                 btrfs_node_key_to_cpu(right, &right_first, 0);
2463         } else {
2464                 btrfs_item_key_to_cpu(left, &left_last, nr_left - 1);
2465                 btrfs_item_key_to_cpu(right, &right_first, 0);
2466         }
2467
2468         if (btrfs_comp_cpu_keys(&left_last, &right_first) >= 0) {
2469                 btrfs_crit(left->fs_info,
2470 "bad key order, sibling blocks, left last (%llu %u %llu) right first (%llu %u %llu)",
2471                            left_last.objectid, left_last.type,
2472                            left_last.offset, right_first.objectid,
2473                            right_first.type, right_first.offset);
2474                 return true;
2475         }
2476         return false;
2477 }
2478
2479 /*
2480  * try to push data from one node into the next node left in the
2481  * tree.
2482  *
2483  * returns 0 if some ptrs were pushed left, < 0 if there was some horrible
2484  * error, and > 0 if there was no room in the left hand block.
2485  */
2486 static int push_node_left(struct btrfs_trans_handle *trans,
2487                           struct extent_buffer *dst,
2488                           struct extent_buffer *src, int empty)
2489 {
2490         struct btrfs_fs_info *fs_info = trans->fs_info;
2491         int push_items = 0;
2492         int src_nritems;
2493         int dst_nritems;
2494         int ret = 0;
2495
2496         src_nritems = btrfs_header_nritems(src);
2497         dst_nritems = btrfs_header_nritems(dst);
2498         push_items = BTRFS_NODEPTRS_PER_BLOCK(fs_info) - dst_nritems;
2499         WARN_ON(btrfs_header_generation(src) != trans->transid);
2500         WARN_ON(btrfs_header_generation(dst) != trans->transid);
2501
2502         if (!empty && src_nritems <= 8)
2503                 return 1;
2504
2505         if (push_items <= 0)
2506                 return 1;
2507
2508         if (empty) {
2509                 push_items = min(src_nritems, push_items);
2510                 if (push_items < src_nritems) {
2511                         /* leave at least 8 pointers in the node if
2512                          * we aren't going to empty it
2513                          */
2514                         if (src_nritems - push_items < 8) {
2515                                 if (push_items <= 8)
2516                                         return 1;
2517                                 push_items -= 8;
2518                         }
2519                 }
2520         } else
2521                 push_items = min(src_nritems - 8, push_items);
2522
2523         /* dst is the left eb, src is the middle eb */
2524         if (check_sibling_keys(dst, src)) {
2525                 ret = -EUCLEAN;
2526                 btrfs_abort_transaction(trans, ret);
2527                 return ret;
2528         }
2529         ret = btrfs_tree_mod_log_eb_copy(dst, src, dst_nritems, 0, push_items);
2530         if (ret) {
2531                 btrfs_abort_transaction(trans, ret);
2532                 return ret;
2533         }
2534         copy_extent_buffer(dst, src,
2535                            btrfs_node_key_ptr_offset(dst_nritems),
2536                            btrfs_node_key_ptr_offset(0),
2537                            push_items * sizeof(struct btrfs_key_ptr));
2538
2539         if (push_items < src_nritems) {
2540                 /*
2541                  * Don't call btrfs_tree_mod_log_insert_move() here, key removal
2542                  * was already fully logged by btrfs_tree_mod_log_eb_copy() above.
2543                  */
2544                 memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0),
2545                                       btrfs_node_key_ptr_offset(push_items),
2546                                       (src_nritems - push_items) *
2547                                       sizeof(struct btrfs_key_ptr));
2548         }
2549         btrfs_set_header_nritems(src, src_nritems - push_items);
2550         btrfs_set_header_nritems(dst, dst_nritems + push_items);
2551         btrfs_mark_buffer_dirty(src);
2552         btrfs_mark_buffer_dirty(dst);
2553
2554         return ret;
2555 }
2556
2557 /*
2558  * try to push data from one node into the next node right in the
2559  * tree.
2560  *
2561  * returns 0 if some ptrs were pushed, < 0 if there was some horrible
2562  * error, and > 0 if there was no room in the right hand block.
2563  *
2564  * this will  only push up to 1/2 the contents of the left node over
2565  */
2566 static int balance_node_right(struct btrfs_trans_handle *trans,
2567                               struct extent_buffer *dst,
2568                               struct extent_buffer *src)
2569 {
2570         struct btrfs_fs_info *fs_info = trans->fs_info;
2571         int push_items = 0;
2572         int max_push;
2573         int src_nritems;
2574         int dst_nritems;
2575         int ret = 0;
2576
2577         WARN_ON(btrfs_header_generation(src) != trans->transid);
2578         WARN_ON(btrfs_header_generation(dst) != trans->transid);
2579
2580         src_nritems = btrfs_header_nritems(src);
2581         dst_nritems = btrfs_header_nritems(dst);
2582         push_items = BTRFS_NODEPTRS_PER_BLOCK(fs_info) - dst_nritems;
2583         if (push_items <= 0)
2584                 return 1;
2585
2586         if (src_nritems < 4)
2587                 return 1;
2588
2589         max_push = src_nritems / 2 + 1;
2590         /* don't try to empty the node */
2591         if (max_push >= src_nritems)
2592                 return 1;
2593
2594         if (max_push < push_items)
2595                 push_items = max_push;
2596
2597         /* dst is the right eb, src is the middle eb */
2598         if (check_sibling_keys(src, dst)) {
2599                 ret = -EUCLEAN;
2600                 btrfs_abort_transaction(trans, ret);
2601                 return ret;
2602         }
2603         ret = btrfs_tree_mod_log_insert_move(dst, push_items, 0, dst_nritems);
2604         BUG_ON(ret < 0);
2605         memmove_extent_buffer(dst, btrfs_node_key_ptr_offset(push_items),
2606                                       btrfs_node_key_ptr_offset(0),
2607                                       (dst_nritems) *
2608                                       sizeof(struct btrfs_key_ptr));
2609
2610         ret = btrfs_tree_mod_log_eb_copy(dst, src, 0, src_nritems - push_items,
2611                                          push_items);
2612         if (ret) {
2613                 btrfs_abort_transaction(trans, ret);
2614                 return ret;
2615         }
2616         copy_extent_buffer(dst, src,
2617                            btrfs_node_key_ptr_offset(0),
2618                            btrfs_node_key_ptr_offset(src_nritems - push_items),
2619                            push_items * sizeof(struct btrfs_key_ptr));
2620
2621         btrfs_set_header_nritems(src, src_nritems - push_items);
2622         btrfs_set_header_nritems(dst, dst_nritems + push_items);
2623
2624         btrfs_mark_buffer_dirty(src);
2625         btrfs_mark_buffer_dirty(dst);
2626
2627         return ret;
2628 }
2629
2630 /*
2631  * helper function to insert a new root level in the tree.
2632  * A new node is allocated, and a single item is inserted to
2633  * point to the existing root
2634  *
2635  * returns zero on success or < 0 on failure.
2636  */
2637 static noinline int insert_new_root(struct btrfs_trans_handle *trans,
2638                            struct btrfs_root *root,
2639                            struct btrfs_path *path, int level)
2640 {
2641         struct btrfs_fs_info *fs_info = root->fs_info;
2642         u64 lower_gen;
2643         struct extent_buffer *lower;
2644         struct extent_buffer *c;
2645         struct extent_buffer *old;
2646         struct btrfs_disk_key lower_key;
2647         int ret;
2648
2649         BUG_ON(path->nodes[level]);
2650         BUG_ON(path->nodes[level-1] != root->node);
2651
2652         lower = path->nodes[level-1];
2653         if (level == 1)
2654                 btrfs_item_key(lower, &lower_key, 0);
2655         else
2656                 btrfs_node_key(lower, &lower_key, 0);
2657
2658         c = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
2659                                    &lower_key, level, root->node->start, 0,
2660                                    BTRFS_NESTING_NEW_ROOT);
2661         if (IS_ERR(c))
2662                 return PTR_ERR(c);
2663
2664         root_add_used(root, fs_info->nodesize);
2665
2666         btrfs_set_header_nritems(c, 1);
2667         btrfs_set_node_key(c, &lower_key, 0);
2668         btrfs_set_node_blockptr(c, 0, lower->start);
2669         lower_gen = btrfs_header_generation(lower);
2670         WARN_ON(lower_gen != trans->transid);
2671
2672         btrfs_set_node_ptr_generation(c, 0, lower_gen);
2673
2674         btrfs_mark_buffer_dirty(c);
2675
2676         old = root->node;
2677         ret = btrfs_tree_mod_log_insert_root(root->node, c, false);
2678         BUG_ON(ret < 0);
2679         rcu_assign_pointer(root->node, c);
2680
2681         /* the super has an extra ref to root->node */
2682         free_extent_buffer(old);
2683
2684         add_root_to_dirty_list(root);
2685         atomic_inc(&c->refs);
2686         path->nodes[level] = c;
2687         path->locks[level] = BTRFS_WRITE_LOCK;
2688         path->slots[level] = 0;
2689         return 0;
2690 }
2691
2692 /*
2693  * worker function to insert a single pointer in a node.
2694  * the node should have enough room for the pointer already
2695  *
2696  * slot and level indicate where you want the key to go, and
2697  * blocknr is the block the key points to.
2698  */
2699 static void insert_ptr(struct btrfs_trans_handle *trans,
2700                        struct btrfs_path *path,
2701                        struct btrfs_disk_key *key, u64 bytenr,
2702                        int slot, int level)
2703 {
2704         struct extent_buffer *lower;
2705         int nritems;
2706         int ret;
2707
2708         BUG_ON(!path->nodes[level]);
2709         btrfs_assert_tree_write_locked(path->nodes[level]);
2710         lower = path->nodes[level];
2711         nritems = btrfs_header_nritems(lower);
2712         BUG_ON(slot > nritems);
2713         BUG_ON(nritems == BTRFS_NODEPTRS_PER_BLOCK(trans->fs_info));
2714         if (slot != nritems) {
2715                 if (level) {
2716                         ret = btrfs_tree_mod_log_insert_move(lower, slot + 1,
2717                                         slot, nritems - slot);
2718                         BUG_ON(ret < 0);
2719                 }
2720                 memmove_extent_buffer(lower,
2721                               btrfs_node_key_ptr_offset(slot + 1),
2722                               btrfs_node_key_ptr_offset(slot),
2723                               (nritems - slot) * sizeof(struct btrfs_key_ptr));
2724         }
2725         if (level) {
2726                 ret = btrfs_tree_mod_log_insert_key(lower, slot,
2727                                             BTRFS_MOD_LOG_KEY_ADD, GFP_NOFS);
2728                 BUG_ON(ret < 0);
2729         }
2730         btrfs_set_node_key(lower, key, slot);
2731         btrfs_set_node_blockptr(lower, slot, bytenr);
2732         WARN_ON(trans->transid == 0);
2733         btrfs_set_node_ptr_generation(lower, slot, trans->transid);
2734         btrfs_set_header_nritems(lower, nritems + 1);
2735         btrfs_mark_buffer_dirty(lower);
2736 }
2737
2738 /*
2739  * split the node at the specified level in path in two.
2740  * The path is corrected to point to the appropriate node after the split
2741  *
2742  * Before splitting this tries to make some room in the node by pushing
2743  * left and right, if either one works, it returns right away.
2744  *
2745  * returns 0 on success and < 0 on failure
2746  */
2747 static noinline int split_node(struct btrfs_trans_handle *trans,
2748                                struct btrfs_root *root,
2749                                struct btrfs_path *path, int level)
2750 {
2751         struct btrfs_fs_info *fs_info = root->fs_info;
2752         struct extent_buffer *c;
2753         struct extent_buffer *split;
2754         struct btrfs_disk_key disk_key;
2755         int mid;
2756         int ret;
2757         u32 c_nritems;
2758
2759         c = path->nodes[level];
2760         WARN_ON(btrfs_header_generation(c) != trans->transid);
2761         if (c == root->node) {
2762                 /*
2763                  * trying to split the root, lets make a new one
2764                  *
2765                  * tree mod log: We don't log_removal old root in
2766                  * insert_new_root, because that root buffer will be kept as a
2767                  * normal node. We are going to log removal of half of the
2768                  * elements below with btrfs_tree_mod_log_eb_copy(). We're
2769                  * holding a tree lock on the buffer, which is why we cannot
2770                  * race with other tree_mod_log users.
2771                  */
2772                 ret = insert_new_root(trans, root, path, level + 1);
2773                 if (ret)
2774                         return ret;
2775         } else {
2776                 ret = push_nodes_for_insert(trans, root, path, level);
2777                 c = path->nodes[level];
2778                 if (!ret && btrfs_header_nritems(c) <
2779                     BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 3)
2780                         return 0;
2781                 if (ret < 0)
2782                         return ret;
2783         }
2784
2785         c_nritems = btrfs_header_nritems(c);
2786         mid = (c_nritems + 1) / 2;
2787         btrfs_node_key(c, &disk_key, mid);
2788
2789         split = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
2790                                        &disk_key, level, c->start, 0,
2791                                        BTRFS_NESTING_SPLIT);
2792         if (IS_ERR(split))
2793                 return PTR_ERR(split);
2794
2795         root_add_used(root, fs_info->nodesize);
2796         ASSERT(btrfs_header_level(c) == level);
2797
2798         ret = btrfs_tree_mod_log_eb_copy(split, c, 0, mid, c_nritems - mid);
2799         if (ret) {
2800                 btrfs_abort_transaction(trans, ret);
2801                 return ret;
2802         }
2803         copy_extent_buffer(split, c,
2804                            btrfs_node_key_ptr_offset(0),
2805                            btrfs_node_key_ptr_offset(mid),
2806                            (c_nritems - mid) * sizeof(struct btrfs_key_ptr));
2807         btrfs_set_header_nritems(split, c_nritems - mid);
2808         btrfs_set_header_nritems(c, mid);
2809
2810         btrfs_mark_buffer_dirty(c);
2811         btrfs_mark_buffer_dirty(split);
2812
2813         insert_ptr(trans, path, &disk_key, split->start,
2814                    path->slots[level + 1] + 1, level + 1);
2815
2816         if (path->slots[level] >= mid) {
2817                 path->slots[level] -= mid;
2818                 btrfs_tree_unlock(c);
2819                 free_extent_buffer(c);
2820                 path->nodes[level] = split;
2821                 path->slots[level + 1] += 1;
2822         } else {
2823                 btrfs_tree_unlock(split);
2824                 free_extent_buffer(split);
2825         }
2826         return 0;
2827 }
2828
2829 /*
2830  * how many bytes are required to store the items in a leaf.  start
2831  * and nr indicate which items in the leaf to check.  This totals up the
2832  * space used both by the item structs and the item data
2833  */
2834 static int leaf_space_used(struct extent_buffer *l, int start, int nr)
2835 {
2836         int data_len;
2837         int nritems = btrfs_header_nritems(l);
2838         int end = min(nritems, start + nr) - 1;
2839
2840         if (!nr)
2841                 return 0;
2842         data_len = btrfs_item_offset(l, start) + btrfs_item_size(l, start);
2843         data_len = data_len - btrfs_item_offset(l, end);
2844         data_len += sizeof(struct btrfs_item) * nr;
2845         WARN_ON(data_len < 0);
2846         return data_len;
2847 }
2848
2849 /*
2850  * The space between the end of the leaf items and
2851  * the start of the leaf data.  IOW, how much room
2852  * the leaf has left for both items and data
2853  */
2854 noinline int btrfs_leaf_free_space(struct extent_buffer *leaf)
2855 {
2856         struct btrfs_fs_info *fs_info = leaf->fs_info;
2857         int nritems = btrfs_header_nritems(leaf);
2858         int ret;
2859
2860         ret = BTRFS_LEAF_DATA_SIZE(fs_info) - leaf_space_used(leaf, 0, nritems);
2861         if (ret < 0) {
2862                 btrfs_crit(fs_info,
2863                            "leaf free space ret %d, leaf data size %lu, used %d nritems %d",
2864                            ret,
2865                            (unsigned long) BTRFS_LEAF_DATA_SIZE(fs_info),
2866                            leaf_space_used(leaf, 0, nritems), nritems);
2867         }
2868         return ret;
2869 }
2870
2871 /*
2872  * min slot controls the lowest index we're willing to push to the
2873  * right.  We'll push up to and including min_slot, but no lower
2874  */
2875 static noinline int __push_leaf_right(struct btrfs_path *path,
2876                                       int data_size, int empty,
2877                                       struct extent_buffer *right,
2878                                       int free_space, u32 left_nritems,
2879                                       u32 min_slot)
2880 {
2881         struct btrfs_fs_info *fs_info = right->fs_info;
2882         struct extent_buffer *left = path->nodes[0];
2883         struct extent_buffer *upper = path->nodes[1];
2884         struct btrfs_map_token token;
2885         struct btrfs_disk_key disk_key;
2886         int slot;
2887         u32 i;
2888         int push_space = 0;
2889         int push_items = 0;
2890         u32 nr;
2891         u32 right_nritems;
2892         u32 data_end;
2893         u32 this_item_size;
2894
2895         if (empty)
2896                 nr = 0;
2897         else
2898                 nr = max_t(u32, 1, min_slot);
2899
2900         if (path->slots[0] >= left_nritems)
2901                 push_space += data_size;
2902
2903         slot = path->slots[1];
2904         i = left_nritems - 1;
2905         while (i >= nr) {
2906                 if (!empty && push_items > 0) {
2907                         if (path->slots[0] > i)
2908                                 break;
2909                         if (path->slots[0] == i) {
2910                                 int space = btrfs_leaf_free_space(left);
2911
2912                                 if (space + push_space * 2 > free_space)
2913                                         break;
2914                         }
2915                 }
2916
2917                 if (path->slots[0] == i)
2918                         push_space += data_size;
2919
2920                 this_item_size = btrfs_item_size(left, i);
2921                 if (this_item_size + sizeof(struct btrfs_item) +
2922                     push_space > free_space)
2923                         break;
2924
2925                 push_items++;
2926                 push_space += this_item_size + sizeof(struct btrfs_item);
2927                 if (i == 0)
2928                         break;
2929                 i--;
2930         }
2931
2932         if (push_items == 0)
2933                 goto out_unlock;
2934
2935         WARN_ON(!empty && push_items == left_nritems);
2936
2937         /* push left to right */
2938         right_nritems = btrfs_header_nritems(right);
2939
2940         push_space = btrfs_item_data_end(left, left_nritems - push_items);
2941         push_space -= leaf_data_end(left);
2942
2943         /* make room in the right data area */
2944         data_end = leaf_data_end(right);
2945         memmove_extent_buffer(right,
2946                               BTRFS_LEAF_DATA_OFFSET + data_end - push_space,
2947                               BTRFS_LEAF_DATA_OFFSET + data_end,
2948                               BTRFS_LEAF_DATA_SIZE(fs_info) - data_end);
2949
2950         /* copy from the left data area */
2951         copy_extent_buffer(right, left, BTRFS_LEAF_DATA_OFFSET +
2952                      BTRFS_LEAF_DATA_SIZE(fs_info) - push_space,
2953                      BTRFS_LEAF_DATA_OFFSET + leaf_data_end(left),
2954                      push_space);
2955
2956         memmove_extent_buffer(right, btrfs_item_nr_offset(push_items),
2957                               btrfs_item_nr_offset(0),
2958                               right_nritems * sizeof(struct btrfs_item));
2959
2960         /* copy the items from left to right */
2961         copy_extent_buffer(right, left, btrfs_item_nr_offset(0),
2962                    btrfs_item_nr_offset(left_nritems - push_items),
2963                    push_items * sizeof(struct btrfs_item));
2964
2965         /* update the item pointers */
2966         btrfs_init_map_token(&token, right);
2967         right_nritems += push_items;
2968         btrfs_set_header_nritems(right, right_nritems);
2969         push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
2970         for (i = 0; i < right_nritems; i++) {
2971                 push_space -= btrfs_token_item_size(&token, i);
2972                 btrfs_set_token_item_offset(&token, i, push_space);
2973         }
2974
2975         left_nritems -= push_items;
2976         btrfs_set_header_nritems(left, left_nritems);
2977
2978         if (left_nritems)
2979                 btrfs_mark_buffer_dirty(left);
2980         else
2981                 btrfs_clean_tree_block(left);
2982
2983         btrfs_mark_buffer_dirty(right);
2984
2985         btrfs_item_key(right, &disk_key, 0);
2986         btrfs_set_node_key(upper, &disk_key, slot + 1);
2987         btrfs_mark_buffer_dirty(upper);
2988
2989         /* then fixup the leaf pointer in the path */
2990         if (path->slots[0] >= left_nritems) {
2991                 path->slots[0] -= left_nritems;
2992                 if (btrfs_header_nritems(path->nodes[0]) == 0)
2993                         btrfs_clean_tree_block(path->nodes[0]);
2994                 btrfs_tree_unlock(path->nodes[0]);
2995                 free_extent_buffer(path->nodes[0]);
2996                 path->nodes[0] = right;
2997                 path->slots[1] += 1;
2998         } else {
2999                 btrfs_tree_unlock(right);
3000                 free_extent_buffer(right);
3001         }
3002         return 0;
3003
3004 out_unlock:
3005         btrfs_tree_unlock(right);
3006         free_extent_buffer(right);
3007         return 1;
3008 }
3009
3010 /*
3011  * push some data in the path leaf to the right, trying to free up at
3012  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3013  *
3014  * returns 1 if the push failed because the other node didn't have enough
3015  * room, 0 if everything worked out and < 0 if there were major errors.
3016  *
3017  * this will push starting from min_slot to the end of the leaf.  It won't
3018  * push any slot lower than min_slot
3019  */
3020 static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
3021                            *root, struct btrfs_path *path,
3022                            int min_data_size, int data_size,
3023                            int empty, u32 min_slot)
3024 {
3025         struct extent_buffer *left = path->nodes[0];
3026         struct extent_buffer *right;
3027         struct extent_buffer *upper;
3028         int slot;
3029         int free_space;
3030         u32 left_nritems;
3031         int ret;
3032
3033         if (!path->nodes[1])
3034                 return 1;
3035
3036         slot = path->slots[1];
3037         upper = path->nodes[1];
3038         if (slot >= btrfs_header_nritems(upper) - 1)
3039                 return 1;
3040
3041         btrfs_assert_tree_write_locked(path->nodes[1]);
3042
3043         right = btrfs_read_node_slot(upper, slot + 1);
3044         /*
3045          * slot + 1 is not valid or we fail to read the right node,
3046          * no big deal, just return.
3047          */
3048         if (IS_ERR(right))
3049                 return 1;
3050
3051         __btrfs_tree_lock(right, BTRFS_NESTING_RIGHT);
3052
3053         free_space = btrfs_leaf_free_space(right);
3054         if (free_space < data_size)
3055                 goto out_unlock;
3056
3057         ret = btrfs_cow_block(trans, root, right, upper,
3058                               slot + 1, &right, BTRFS_NESTING_RIGHT_COW);
3059         if (ret)
3060                 goto out_unlock;
3061
3062         left_nritems = btrfs_header_nritems(left);
3063         if (left_nritems == 0)
3064                 goto out_unlock;
3065
3066         if (check_sibling_keys(left, right)) {
3067                 ret = -EUCLEAN;
3068                 btrfs_tree_unlock(right);
3069                 free_extent_buffer(right);
3070                 return ret;
3071         }
3072         if (path->slots[0] == left_nritems && !empty) {
3073                 /* Key greater than all keys in the leaf, right neighbor has
3074                  * enough room for it and we're not emptying our leaf to delete
3075                  * it, therefore use right neighbor to insert the new item and
3076                  * no need to touch/dirty our left leaf. */
3077                 btrfs_tree_unlock(left);
3078                 free_extent_buffer(left);
3079                 path->nodes[0] = right;
3080                 path->slots[0] = 0;
3081                 path->slots[1]++;
3082                 return 0;
3083         }
3084
3085         return __push_leaf_right(path, min_data_size, empty,
3086                                 right, free_space, left_nritems, min_slot);
3087 out_unlock:
3088         btrfs_tree_unlock(right);
3089         free_extent_buffer(right);
3090         return 1;
3091 }
3092
3093 /*
3094  * push some data in the path leaf to the left, trying to free up at
3095  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3096  *
3097  * max_slot can put a limit on how far into the leaf we'll push items.  The
3098  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us do all the
3099  * items
3100  */
3101 static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
3102                                      int empty, struct extent_buffer *left,
3103                                      int free_space, u32 right_nritems,
3104                                      u32 max_slot)
3105 {
3106         struct btrfs_fs_info *fs_info = left->fs_info;
3107         struct btrfs_disk_key disk_key;
3108         struct extent_buffer *right = path->nodes[0];
3109         int i;
3110         int push_space = 0;
3111         int push_items = 0;
3112         u32 old_left_nritems;
3113         u32 nr;
3114         int ret = 0;
3115         u32 this_item_size;
3116         u32 old_left_item_size;
3117         struct btrfs_map_token token;
3118
3119         if (empty)
3120                 nr = min(right_nritems, max_slot);
3121         else
3122                 nr = min(right_nritems - 1, max_slot);
3123
3124         for (i = 0; i < nr; i++) {
3125                 if (!empty && push_items > 0) {
3126                         if (path->slots[0] < i)
3127                                 break;
3128                         if (path->slots[0] == i) {
3129                                 int space = btrfs_leaf_free_space(right);
3130
3131                                 if (space + push_space * 2 > free_space)
3132                                         break;
3133                         }
3134                 }
3135
3136                 if (path->slots[0] == i)
3137                         push_space += data_size;
3138
3139                 this_item_size = btrfs_item_size(right, i);
3140                 if (this_item_size + sizeof(struct btrfs_item) + push_space >
3141                     free_space)
3142                         break;
3143
3144                 push_items++;
3145                 push_space += this_item_size + sizeof(struct btrfs_item);
3146         }
3147
3148         if (push_items == 0) {
3149                 ret = 1;
3150                 goto out;
3151         }
3152         WARN_ON(!empty && push_items == btrfs_header_nritems(right));
3153
3154         /* push data from right to left */
3155         copy_extent_buffer(left, right,
3156                            btrfs_item_nr_offset(btrfs_header_nritems(left)),
3157                            btrfs_item_nr_offset(0),
3158                            push_items * sizeof(struct btrfs_item));
3159
3160         push_space = BTRFS_LEAF_DATA_SIZE(fs_info) -
3161                      btrfs_item_offset(right, push_items - 1);
3162
3163         copy_extent_buffer(left, right, BTRFS_LEAF_DATA_OFFSET +
3164                      leaf_data_end(left) - push_space,
3165                      BTRFS_LEAF_DATA_OFFSET +
3166                      btrfs_item_offset(right, push_items - 1),
3167                      push_space);
3168         old_left_nritems = btrfs_header_nritems(left);
3169         BUG_ON(old_left_nritems <= 0);
3170
3171         btrfs_init_map_token(&token, left);
3172         old_left_item_size = btrfs_item_offset(left, old_left_nritems - 1);
3173         for (i = old_left_nritems; i < old_left_nritems + push_items; i++) {
3174                 u32 ioff;
3175
3176                 ioff = btrfs_token_item_offset(&token, i);
3177                 btrfs_set_token_item_offset(&token, i,
3178                       ioff - (BTRFS_LEAF_DATA_SIZE(fs_info) - old_left_item_size));
3179         }
3180         btrfs_set_header_nritems(left, old_left_nritems + push_items);
3181
3182         /* fixup right node */
3183         if (push_items > right_nritems)
3184                 WARN(1, KERN_CRIT "push items %d nr %u\n", push_items,
3185                        right_nritems);
3186
3187         if (push_items < right_nritems) {
3188                 push_space = btrfs_item_offset(right, push_items - 1) -
3189                                                   leaf_data_end(right);
3190                 memmove_extent_buffer(right, BTRFS_LEAF_DATA_OFFSET +
3191                                       BTRFS_LEAF_DATA_SIZE(fs_info) - push_space,
3192                                       BTRFS_LEAF_DATA_OFFSET +
3193                                       leaf_data_end(right), push_space);
3194
3195                 memmove_extent_buffer(right, btrfs_item_nr_offset(0),
3196                               btrfs_item_nr_offset(push_items),
3197                              (btrfs_header_nritems(right) - push_items) *
3198                              sizeof(struct btrfs_item));
3199         }
3200
3201         btrfs_init_map_token(&token, right);
3202         right_nritems -= push_items;
3203         btrfs_set_header_nritems(right, right_nritems);
3204         push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
3205         for (i = 0; i < right_nritems; i++) {
3206                 push_space = push_space - btrfs_token_item_size(&token, i);
3207                 btrfs_set_token_item_offset(&token, i, push_space);
3208         }
3209
3210         btrfs_mark_buffer_dirty(left);
3211         if (right_nritems)
3212                 btrfs_mark_buffer_dirty(right);
3213         else
3214                 btrfs_clean_tree_block(right);
3215
3216         btrfs_item_key(right, &disk_key, 0);
3217         fixup_low_keys(path, &disk_key, 1);
3218
3219         /* then fixup the leaf pointer in the path */
3220         if (path->slots[0] < push_items) {
3221                 path->slots[0] += old_left_nritems;
3222                 btrfs_tree_unlock(path->nodes[0]);
3223                 free_extent_buffer(path->nodes[0]);
3224                 path->nodes[0] = left;
3225                 path->slots[1] -= 1;
3226         } else {
3227                 btrfs_tree_unlock(left);
3228                 free_extent_buffer(left);
3229                 path->slots[0] -= push_items;
3230         }
3231         BUG_ON(path->slots[0] < 0);
3232         return ret;
3233 out:
3234         btrfs_tree_unlock(left);
3235         free_extent_buffer(left);
3236         return ret;
3237 }
3238
3239 /*
3240  * push some data in the path leaf to the left, trying to free up at
3241  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3242  *
3243  * max_slot can put a limit on how far into the leaf we'll push items.  The
3244  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us push all the
3245  * items
3246  */
3247 static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
3248                           *root, struct btrfs_path *path, int min_data_size,
3249                           int data_size, int empty, u32 max_slot)
3250 {
3251         struct extent_buffer *right = path->nodes[0];
3252         struct extent_buffer *left;
3253         int slot;
3254         int free_space;
3255         u32 right_nritems;
3256         int ret = 0;
3257
3258         slot = path->slots[1];
3259         if (slot == 0)
3260                 return 1;
3261         if (!path->nodes[1])
3262                 return 1;
3263
3264         right_nritems = btrfs_header_nritems(right);
3265         if (right_nritems == 0)
3266                 return 1;
3267
3268         btrfs_assert_tree_write_locked(path->nodes[1]);
3269
3270         left = btrfs_read_node_slot(path->nodes[1], slot - 1);
3271         /*
3272          * slot - 1 is not valid or we fail to read the left node,
3273          * no big deal, just return.
3274          */
3275         if (IS_ERR(left))
3276                 return 1;
3277
3278         __btrfs_tree_lock(left, BTRFS_NESTING_LEFT);
3279
3280         free_space = btrfs_leaf_free_space(left);
3281         if (free_space < data_size) {
3282                 ret = 1;
3283                 goto out;
3284         }
3285
3286         ret = btrfs_cow_block(trans, root, left,
3287                               path->nodes[1], slot - 1, &left,
3288                               BTRFS_NESTING_LEFT_COW);
3289         if (ret) {
3290                 /* we hit -ENOSPC, but it isn't fatal here */
3291                 if (ret == -ENOSPC)
3292                         ret = 1;
3293                 goto out;
3294         }
3295
3296         if (check_sibling_keys(left, right)) {
3297                 ret = -EUCLEAN;
3298                 goto out;
3299         }
3300         return __push_leaf_left(path, min_data_size,
3301                                empty, left, free_space, right_nritems,
3302                                max_slot);
3303 out:
3304         btrfs_tree_unlock(left);
3305         free_extent_buffer(left);
3306         return ret;
3307 }
3308
3309 /*
3310  * split the path's leaf in two, making sure there is at least data_size
3311  * available for the resulting leaf level of the path.
3312  */
3313 static noinline void copy_for_split(struct btrfs_trans_handle *trans,
3314                                     struct btrfs_path *path,
3315                                     struct extent_buffer *l,
3316                                     struct extent_buffer *right,
3317                                     int slot, int mid, int nritems)
3318 {
3319         struct btrfs_fs_info *fs_info = trans->fs_info;
3320         int data_copy_size;
3321         int rt_data_off;
3322         int i;
3323         struct btrfs_disk_key disk_key;
3324         struct btrfs_map_token token;
3325
3326         nritems = nritems - mid;
3327         btrfs_set_header_nritems(right, nritems);
3328         data_copy_size = btrfs_item_data_end(l, mid) - leaf_data_end(l);
3329
3330         copy_extent_buffer(right, l, btrfs_item_nr_offset(0),
3331                            btrfs_item_nr_offset(mid),
3332                            nritems * sizeof(struct btrfs_item));
3333
3334         copy_extent_buffer(right, l,
3335                      BTRFS_LEAF_DATA_OFFSET + BTRFS_LEAF_DATA_SIZE(fs_info) -
3336                      data_copy_size, BTRFS_LEAF_DATA_OFFSET +
3337                      leaf_data_end(l), data_copy_size);
3338
3339         rt_data_off = BTRFS_LEAF_DATA_SIZE(fs_info) - btrfs_item_data_end(l, mid);
3340
3341         btrfs_init_map_token(&token, right);
3342         for (i = 0; i < nritems; i++) {
3343                 u32 ioff;
3344
3345                 ioff = btrfs_token_item_offset(&token, i);
3346                 btrfs_set_token_item_offset(&token, i, ioff + rt_data_off);
3347         }
3348
3349         btrfs_set_header_nritems(l, mid);
3350         btrfs_item_key(right, &disk_key, 0);
3351         insert_ptr(trans, path, &disk_key, right->start, path->slots[1] + 1, 1);
3352
3353         btrfs_mark_buffer_dirty(right);
3354         btrfs_mark_buffer_dirty(l);
3355         BUG_ON(path->slots[0] != slot);
3356
3357         if (mid <= slot) {
3358                 btrfs_tree_unlock(path->nodes[0]);
3359                 free_extent_buffer(path->nodes[0]);
3360                 path->nodes[0] = right;
3361                 path->slots[0] -= mid;
3362                 path->slots[1] += 1;
3363         } else {
3364                 btrfs_tree_unlock(right);
3365                 free_extent_buffer(right);
3366         }
3367
3368         BUG_ON(path->slots[0] < 0);
3369 }
3370
3371 /*
3372  * double splits happen when we need to insert a big item in the middle
3373  * of a leaf.  A double split can leave us with 3 mostly empty leaves:
3374  * leaf: [ slots 0 - N] [ our target ] [ N + 1 - total in leaf ]
3375  *          A                 B                 C
3376  *
3377  * We avoid this by trying to push the items on either side of our target
3378  * into the adjacent leaves.  If all goes well we can avoid the double split
3379  * completely.
3380  */
3381 static noinline int push_for_double_split(struct btrfs_trans_handle *trans,
3382                                           struct btrfs_root *root,
3383                                           struct btrfs_path *path,
3384                                           int data_size)
3385 {
3386         int ret;
3387         int progress = 0;
3388         int slot;
3389         u32 nritems;
3390         int space_needed = data_size;
3391
3392         slot = path->slots[0];
3393         if (slot < btrfs_header_nritems(path->nodes[0]))
3394                 space_needed -= btrfs_leaf_free_space(path->nodes[0]);
3395
3396         /*
3397          * try to push all the items after our slot into the
3398          * right leaf
3399          */
3400         ret = push_leaf_right(trans, root, path, 1, space_needed, 0, slot);
3401         if (ret < 0)
3402                 return ret;
3403
3404         if (ret == 0)
3405                 progress++;
3406
3407         nritems = btrfs_header_nritems(path->nodes[0]);
3408         /*
3409          * our goal is to get our slot at the start or end of a leaf.  If
3410          * we've done so we're done
3411          */
3412         if (path->slots[0] == 0 || path->slots[0] == nritems)
3413                 return 0;
3414
3415         if (btrfs_leaf_free_space(path->nodes[0]) >= data_size)
3416                 return 0;
3417
3418         /* try to push all the items before our slot into the next leaf */
3419         slot = path->slots[0];
3420         space_needed = data_size;
3421         if (slot > 0)
3422                 space_needed -= btrfs_leaf_free_space(path->nodes[0]);
3423         ret = push_leaf_left(trans, root, path, 1, space_needed, 0, slot);
3424         if (ret < 0)
3425                 return ret;
3426
3427         if (ret == 0)
3428                 progress++;
3429
3430         if (progress)
3431                 return 0;
3432         return 1;
3433 }
3434
3435 /*
3436  * split the path's leaf in two, making sure there is at least data_size
3437  * available for the resulting leaf level of the path.
3438  *
3439  * returns 0 if all went well and < 0 on failure.
3440  */
3441 static noinline int split_leaf(struct btrfs_trans_handle *trans,
3442                                struct btrfs_root *root,
3443                                const struct btrfs_key *ins_key,
3444                                struct btrfs_path *path, int data_size,
3445                                int extend)
3446 {
3447         struct btrfs_disk_key disk_key;
3448         struct extent_buffer *l;
3449         u32 nritems;
3450         int mid;
3451         int slot;
3452         struct extent_buffer *right;
3453         struct btrfs_fs_info *fs_info = root->fs_info;
3454         int ret = 0;
3455         int wret;
3456         int split;
3457         int num_doubles = 0;
3458         int tried_avoid_double = 0;
3459
3460         l = path->nodes[0];
3461         slot = path->slots[0];
3462         if (extend && data_size + btrfs_item_size(l, slot) +
3463             sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(fs_info))
3464                 return -EOVERFLOW;
3465
3466         /* first try to make some room by pushing left and right */
3467         if (data_size && path->nodes[1]) {
3468                 int space_needed = data_size;
3469
3470                 if (slot < btrfs_header_nritems(l))
3471                         space_needed -= btrfs_leaf_free_space(l);
3472
3473                 wret = push_leaf_right(trans, root, path, space_needed,
3474                                        space_needed, 0, 0);
3475                 if (wret < 0)
3476                         return wret;
3477                 if (wret) {
3478                         space_needed = data_size;
3479                         if (slot > 0)
3480                                 space_needed -= btrfs_leaf_free_space(l);
3481                         wret = push_leaf_left(trans, root, path, space_needed,
3482                                               space_needed, 0, (u32)-1);
3483                         if (wret < 0)
3484                                 return wret;
3485                 }
3486                 l = path->nodes[0];
3487
3488                 /* did the pushes work? */
3489                 if (btrfs_leaf_free_space(l) >= data_size)
3490                         return 0;
3491         }
3492
3493         if (!path->nodes[1]) {
3494                 ret = insert_new_root(trans, root, path, 1);
3495                 if (ret)
3496                         return ret;
3497         }
3498 again:
3499         split = 1;
3500         l = path->nodes[0];
3501         slot = path->slots[0];
3502         nritems = btrfs_header_nritems(l);
3503         mid = (nritems + 1) / 2;
3504
3505         if (mid <= slot) {
3506                 if (nritems == 1 ||
3507                     leaf_space_used(l, mid, nritems - mid) + data_size >
3508                         BTRFS_LEAF_DATA_SIZE(fs_info)) {
3509                         if (slot >= nritems) {
3510                                 split = 0;
3511                         } else {
3512                                 mid = slot;
3513                                 if (mid != nritems &&
3514                                     leaf_space_used(l, mid, nritems - mid) +
3515                                     data_size > BTRFS_LEAF_DATA_SIZE(fs_info)) {
3516                                         if (data_size && !tried_avoid_double)
3517                                                 goto push_for_double;
3518                                         split = 2;
3519                                 }
3520                         }
3521                 }
3522         } else {
3523                 if (leaf_space_used(l, 0, mid) + data_size >
3524                         BTRFS_LEAF_DATA_SIZE(fs_info)) {
3525                         if (!extend && data_size && slot == 0) {
3526                                 split = 0;
3527                         } else if ((extend || !data_size) && slot == 0) {
3528                                 mid = 1;
3529                         } else {
3530                                 mid = slot;
3531                                 if (mid != nritems &&
3532                                     leaf_space_used(l, mid, nritems - mid) +
3533                                     data_size > BTRFS_LEAF_DATA_SIZE(fs_info)) {
3534                                         if (data_size && !tried_avoid_double)
3535                                                 goto push_for_double;
3536                                         split = 2;
3537                                 }
3538                         }
3539                 }
3540         }
3541
3542         if (split == 0)
3543                 btrfs_cpu_key_to_disk(&disk_key, ins_key);
3544         else
3545                 btrfs_item_key(l, &disk_key, mid);
3546
3547         /*
3548          * We have to about BTRFS_NESTING_NEW_ROOT here if we've done a double
3549          * split, because we're only allowed to have MAX_LOCKDEP_SUBCLASSES
3550          * subclasses, which is 8 at the time of this patch, and we've maxed it
3551          * out.  In the future we could add a
3552          * BTRFS_NESTING_SPLIT_THE_SPLITTENING if we need to, but for now just
3553          * use BTRFS_NESTING_NEW_ROOT.
3554          */
3555         right = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
3556                                        &disk_key, 0, l->start, 0,
3557                                        num_doubles ? BTRFS_NESTING_NEW_ROOT :
3558                                        BTRFS_NESTING_SPLIT);
3559         if (IS_ERR(right))
3560                 return PTR_ERR(right);
3561
3562         root_add_used(root, fs_info->nodesize);
3563
3564         if (split == 0) {
3565                 if (mid <= slot) {
3566                         btrfs_set_header_nritems(right, 0);
3567                         insert_ptr(trans, path, &disk_key,
3568                                    right->start, path->slots[1] + 1, 1);
3569                         btrfs_tree_unlock(path->nodes[0]);
3570                         free_extent_buffer(path->nodes[0]);
3571                         path->nodes[0] = right;
3572                         path->slots[0] = 0;
3573                         path->slots[1] += 1;
3574                 } else {
3575                         btrfs_set_header_nritems(right, 0);
3576                         insert_ptr(trans, path, &disk_key,
3577                                    right->start, path->slots[1], 1);
3578                         btrfs_tree_unlock(path->nodes[0]);
3579                         free_extent_buffer(path->nodes[0]);
3580                         path->nodes[0] = right;
3581                         path->slots[0] = 0;
3582                         if (path->slots[1] == 0)
3583                                 fixup_low_keys(path, &disk_key, 1);
3584                 }
3585                 /*
3586                  * We create a new leaf 'right' for the required ins_len and
3587                  * we'll do btrfs_mark_buffer_dirty() on this leaf after copying
3588                  * the content of ins_len to 'right'.
3589                  */
3590                 return ret;
3591         }
3592
3593         copy_for_split(trans, path, l, right, slot, mid, nritems);
3594
3595         if (split == 2) {
3596                 BUG_ON(num_doubles != 0);
3597                 num_doubles++;
3598                 goto again;
3599         }
3600
3601         return 0;
3602
3603 push_for_double:
3604         push_for_double_split(trans, root, path, data_size);
3605         tried_avoid_double = 1;
3606         if (btrfs_leaf_free_space(path->nodes[0]) >= data_size)
3607                 return 0;
3608         goto again;
3609 }
3610
3611 static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
3612                                          struct btrfs_root *root,
3613                                          struct btrfs_path *path, int ins_len)
3614 {
3615         struct btrfs_key key;
3616         struct extent_buffer *leaf;
3617         struct btrfs_file_extent_item *fi;
3618         u64 extent_len = 0;
3619         u32 item_size;
3620         int ret;
3621
3622         leaf = path->nodes[0];
3623         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
3624
3625         BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY &&
3626                key.type != BTRFS_EXTENT_CSUM_KEY);
3627
3628         if (btrfs_leaf_free_space(leaf) >= ins_len)
3629                 return 0;
3630
3631         item_size = btrfs_item_size(leaf, path->slots[0]);
3632         if (key.type == BTRFS_EXTENT_DATA_KEY) {
3633                 fi = btrfs_item_ptr(leaf, path->slots[0],
3634                                     struct btrfs_file_extent_item);
3635                 extent_len = btrfs_file_extent_num_bytes(leaf, fi);
3636         }
3637         btrfs_release_path(path);
3638
3639         path->keep_locks = 1;
3640         path->search_for_split = 1;
3641         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
3642         path->search_for_split = 0;
3643         if (ret > 0)
3644                 ret = -EAGAIN;
3645         if (ret < 0)
3646                 goto err;
3647
3648         ret = -EAGAIN;
3649         leaf = path->nodes[0];
3650         /* if our item isn't there, return now */
3651         if (item_size != btrfs_item_size(leaf, path->slots[0]))
3652                 goto err;
3653
3654         /* the leaf has  changed, it now has room.  return now */
3655         if (btrfs_leaf_free_space(path->nodes[0]) >= ins_len)
3656                 goto err;
3657
3658         if (key.type == BTRFS_EXTENT_DATA_KEY) {
3659                 fi = btrfs_item_ptr(leaf, path->slots[0],
3660                                     struct btrfs_file_extent_item);
3661                 if (extent_len != btrfs_file_extent_num_bytes(leaf, fi))
3662                         goto err;
3663         }
3664
3665         ret = split_leaf(trans, root, &key, path, ins_len, 1);
3666         if (ret)
3667                 goto err;
3668
3669         path->keep_locks = 0;
3670         btrfs_unlock_up_safe(path, 1);
3671         return 0;
3672 err:
3673         path->keep_locks = 0;
3674         return ret;
3675 }
3676
3677 static noinline int split_item(struct btrfs_path *path,
3678                                const struct btrfs_key *new_key,
3679                                unsigned long split_offset)
3680 {
3681         struct extent_buffer *leaf;
3682         int orig_slot, slot;
3683         char *buf;
3684         u32 nritems;
3685         u32 item_size;
3686         u32 orig_offset;
3687         struct btrfs_disk_key disk_key;
3688
3689         leaf = path->nodes[0];
3690         BUG_ON(btrfs_leaf_free_space(leaf) < sizeof(struct btrfs_item));
3691
3692         orig_slot = path->slots[0];
3693         orig_offset = btrfs_item_offset(leaf, path->slots[0]);
3694         item_size = btrfs_item_size(leaf, path->slots[0]);
3695
3696         buf = kmalloc(item_size, GFP_NOFS);
3697         if (!buf)
3698                 return -ENOMEM;
3699
3700         read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf,
3701                             path->slots[0]), item_size);
3702
3703         slot = path->slots[0] + 1;
3704         nritems = btrfs_header_nritems(leaf);
3705         if (slot != nritems) {
3706                 /* shift the items */
3707                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1),
3708                                 btrfs_item_nr_offset(slot),
3709                                 (nritems - slot) * sizeof(struct btrfs_item));
3710         }
3711
3712         btrfs_cpu_key_to_disk(&disk_key, new_key);
3713         btrfs_set_item_key(leaf, &disk_key, slot);
3714
3715         btrfs_set_item_offset(leaf, slot, orig_offset);
3716         btrfs_set_item_size(leaf, slot, item_size - split_offset);
3717
3718         btrfs_set_item_offset(leaf, orig_slot,
3719                                  orig_offset + item_size - split_offset);
3720         btrfs_set_item_size(leaf, orig_slot, split_offset);
3721
3722         btrfs_set_header_nritems(leaf, nritems + 1);
3723
3724         /* write the data for the start of the original item */
3725         write_extent_buffer(leaf, buf,
3726                             btrfs_item_ptr_offset(leaf, path->slots[0]),
3727                             split_offset);
3728
3729         /* write the data for the new item */
3730         write_extent_buffer(leaf, buf + split_offset,
3731                             btrfs_item_ptr_offset(leaf, slot),
3732                             item_size - split_offset);
3733         btrfs_mark_buffer_dirty(leaf);
3734
3735         BUG_ON(btrfs_leaf_free_space(leaf) < 0);
3736         kfree(buf);
3737         return 0;
3738 }
3739
3740 /*
3741  * This function splits a single item into two items,
3742  * giving 'new_key' to the new item and splitting the
3743  * old one at split_offset (from the start of the item).
3744  *
3745  * The path may be released by this operation.  After
3746  * the split, the path is pointing to the old item.  The
3747  * new item is going to be in the same node as the old one.
3748  *
3749  * Note, the item being split must be smaller enough to live alone on
3750  * a tree block with room for one extra struct btrfs_item
3751  *
3752  * This allows us to split the item in place, keeping a lock on the
3753  * leaf the entire time.
3754  */
3755 int btrfs_split_item(struct btrfs_trans_handle *trans,
3756                      struct btrfs_root *root,
3757                      struct btrfs_path *path,
3758                      const struct btrfs_key *new_key,
3759                      unsigned long split_offset)
3760 {
3761         int ret;
3762         ret = setup_leaf_for_split(trans, root, path,
3763                                    sizeof(struct btrfs_item));
3764         if (ret)
3765                 return ret;
3766
3767         ret = split_item(path, new_key, split_offset);
3768         return ret;
3769 }
3770
3771 /*
3772  * make the item pointed to by the path smaller.  new_size indicates
3773  * how small to make it, and from_end tells us if we just chop bytes
3774  * off the end of the item or if we shift the item to chop bytes off
3775  * the front.
3776  */
3777 void btrfs_truncate_item(struct btrfs_path *path, u32 new_size, int from_end)
3778 {
3779         int slot;
3780         struct extent_buffer *leaf;
3781         u32 nritems;
3782         unsigned int data_end;
3783         unsigned int old_data_start;
3784         unsigned int old_size;
3785         unsigned int size_diff;
3786         int i;
3787         struct btrfs_map_token token;
3788
3789         leaf = path->nodes[0];
3790         slot = path->slots[0];
3791
3792         old_size = btrfs_item_size(leaf, slot);
3793         if (old_size == new_size)
3794                 return;
3795
3796         nritems = btrfs_header_nritems(leaf);
3797         data_end = leaf_data_end(leaf);
3798
3799         old_data_start = btrfs_item_offset(leaf, slot);
3800
3801         size_diff = old_size - new_size;
3802
3803         BUG_ON(slot < 0);
3804         BUG_ON(slot >= nritems);
3805
3806         /*
3807          * item0..itemN ... dataN.offset..dataN.size .. data0.size
3808          */
3809         /* first correct the data pointers */
3810         btrfs_init_map_token(&token, leaf);
3811         for (i = slot; i < nritems; i++) {
3812                 u32 ioff;
3813
3814                 ioff = btrfs_token_item_offset(&token, i);
3815                 btrfs_set_token_item_offset(&token, i, ioff + size_diff);
3816         }
3817
3818         /* shift the data */
3819         if (from_end) {
3820                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
3821                               data_end + size_diff, BTRFS_LEAF_DATA_OFFSET +
3822                               data_end, old_data_start + new_size - data_end);
3823         } else {
3824                 struct btrfs_disk_key disk_key;
3825                 u64 offset;
3826
3827                 btrfs_item_key(leaf, &disk_key, slot);
3828
3829                 if (btrfs_disk_key_type(&disk_key) == BTRFS_EXTENT_DATA_KEY) {
3830                         unsigned long ptr;
3831                         struct btrfs_file_extent_item *fi;
3832
3833                         fi = btrfs_item_ptr(leaf, slot,
3834                                             struct btrfs_file_extent_item);
3835                         fi = (struct btrfs_file_extent_item *)(
3836                              (unsigned long)fi - size_diff);
3837
3838                         if (btrfs_file_extent_type(leaf, fi) ==
3839                             BTRFS_FILE_EXTENT_INLINE) {
3840                                 ptr = btrfs_item_ptr_offset(leaf, slot);
3841                                 memmove_extent_buffer(leaf, ptr,
3842                                       (unsigned long)fi,
3843                                       BTRFS_FILE_EXTENT_INLINE_DATA_START);
3844                         }
3845                 }
3846
3847                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
3848                               data_end + size_diff, BTRFS_LEAF_DATA_OFFSET +
3849                               data_end, old_data_start - data_end);
3850
3851                 offset = btrfs_disk_key_offset(&disk_key);
3852                 btrfs_set_disk_key_offset(&disk_key, offset + size_diff);
3853                 btrfs_set_item_key(leaf, &disk_key, slot);
3854                 if (slot == 0)
3855                         fixup_low_keys(path, &disk_key, 1);
3856         }
3857
3858         btrfs_set_item_size(leaf, slot, new_size);
3859         btrfs_mark_buffer_dirty(leaf);
3860
3861         if (btrfs_leaf_free_space(leaf) < 0) {
3862                 btrfs_print_leaf(leaf);
3863                 BUG();
3864         }
3865 }
3866
3867 /*
3868  * make the item pointed to by the path bigger, data_size is the added size.
3869  */
3870 void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
3871 {
3872         int slot;
3873         struct extent_buffer *leaf;
3874         u32 nritems;
3875         unsigned int data_end;
3876         unsigned int old_data;
3877         unsigned int old_size;
3878         int i;
3879         struct btrfs_map_token token;
3880
3881         leaf = path->nodes[0];
3882
3883         nritems = btrfs_header_nritems(leaf);
3884         data_end = leaf_data_end(leaf);
3885
3886         if (btrfs_leaf_free_space(leaf) < data_size) {
3887                 btrfs_print_leaf(leaf);
3888                 BUG();
3889         }
3890         slot = path->slots[0];
3891         old_data = btrfs_item_data_end(leaf, slot);
3892
3893         BUG_ON(slot < 0);
3894         if (slot >= nritems) {
3895                 btrfs_print_leaf(leaf);
3896                 btrfs_crit(leaf->fs_info, "slot %d too large, nritems %d",
3897                            slot, nritems);
3898                 BUG();
3899         }
3900
3901         /*
3902          * item0..itemN ... dataN.offset..dataN.size .. data0.size
3903          */
3904         /* first correct the data pointers */
3905         btrfs_init_map_token(&token, leaf);
3906         for (i = slot; i < nritems; i++) {
3907                 u32 ioff;
3908
3909                 ioff = btrfs_token_item_offset(&token, i);
3910                 btrfs_set_token_item_offset(&token, i, ioff - data_size);
3911         }
3912
3913         /* shift the data */
3914         memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
3915                       data_end - data_size, BTRFS_LEAF_DATA_OFFSET +
3916                       data_end, old_data - data_end);
3917
3918         data_end = old_data;
3919         old_size = btrfs_item_size(leaf, slot);
3920         btrfs_set_item_size(leaf, slot, old_size + data_size);
3921         btrfs_mark_buffer_dirty(leaf);
3922
3923         if (btrfs_leaf_free_space(leaf) < 0) {
3924                 btrfs_print_leaf(leaf);
3925                 BUG();
3926         }
3927 }
3928
3929 /**
3930  * setup_items_for_insert - Helper called before inserting one or more items
3931  * to a leaf. Main purpose is to save stack depth by doing the bulk of the work
3932  * in a function that doesn't call btrfs_search_slot
3933  *
3934  * @root:       root we are inserting items to
3935  * @path:       points to the leaf/slot where we are going to insert new items
3936  * @batch:      information about the batch of items to insert
3937  */
3938 static void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
3939                                    const struct btrfs_item_batch *batch)
3940 {
3941         struct btrfs_fs_info *fs_info = root->fs_info;
3942         int i;
3943         u32 nritems;
3944         unsigned int data_end;
3945         struct btrfs_disk_key disk_key;
3946         struct extent_buffer *leaf;
3947         int slot;
3948         struct btrfs_map_token token;
3949         u32 total_size;
3950
3951         /*
3952          * Before anything else, update keys in the parent and other ancestors
3953          * if needed, then release the write locks on them, so that other tasks
3954          * can use them while we modify the leaf.
3955          */
3956         if (path->slots[0] == 0) {
3957                 btrfs_cpu_key_to_disk(&disk_key, &batch->keys[0]);
3958                 fixup_low_keys(path, &disk_key, 1);
3959         }
3960         btrfs_unlock_up_safe(path, 1);
3961
3962         leaf = path->nodes[0];
3963         slot = path->slots[0];
3964
3965         nritems = btrfs_header_nritems(leaf);
3966         data_end = leaf_data_end(leaf);
3967         total_size = batch->total_data_size + (batch->nr * sizeof(struct btrfs_item));
3968
3969         if (btrfs_leaf_free_space(leaf) < total_size) {
3970                 btrfs_print_leaf(leaf);
3971                 btrfs_crit(fs_info, "not enough freespace need %u have %d",
3972                            total_size, btrfs_leaf_free_space(leaf));
3973                 BUG();
3974         }
3975
3976         btrfs_init_map_token(&token, leaf);
3977         if (slot != nritems) {
3978                 unsigned int old_data = btrfs_item_data_end(leaf, slot);
3979
3980                 if (old_data < data_end) {
3981                         btrfs_print_leaf(leaf);
3982                         btrfs_crit(fs_info,
3983                 "item at slot %d with data offset %u beyond data end of leaf %u",
3984                                    slot, old_data, data_end);
3985                         BUG();
3986                 }
3987                 /*
3988                  * item0..itemN ... dataN.offset..dataN.size .. data0.size
3989                  */
3990                 /* first correct the data pointers */
3991                 for (i = slot; i < nritems; i++) {
3992                         u32 ioff;
3993
3994                         ioff = btrfs_token_item_offset(&token, i);
3995                         btrfs_set_token_item_offset(&token, i,
3996                                                        ioff - batch->total_data_size);
3997                 }
3998                 /* shift the items */
3999                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + batch->nr),
4000                               btrfs_item_nr_offset(slot),
4001                               (nritems - slot) * sizeof(struct btrfs_item));
4002
4003                 /* shift the data */
4004                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4005                                       data_end - batch->total_data_size,
4006                                       BTRFS_LEAF_DATA_OFFSET + data_end,
4007                                       old_data - data_end);
4008                 data_end = old_data;
4009         }
4010
4011         /* setup the item for the new data */
4012         for (i = 0; i < batch->nr; i++) {
4013                 btrfs_cpu_key_to_disk(&disk_key, &batch->keys[i]);
4014                 btrfs_set_item_key(leaf, &disk_key, slot + i);
4015                 data_end -= batch->data_sizes[i];
4016                 btrfs_set_token_item_offset(&token, slot + i, data_end);
4017                 btrfs_set_token_item_size(&token, slot + i, batch->data_sizes[i]);
4018         }
4019
4020         btrfs_set_header_nritems(leaf, nritems + batch->nr);
4021         btrfs_mark_buffer_dirty(leaf);
4022
4023         if (btrfs_leaf_free_space(leaf) < 0) {
4024                 btrfs_print_leaf(leaf);
4025                 BUG();
4026         }
4027 }
4028
4029 /*
4030  * Insert a new item into a leaf.
4031  *
4032  * @root:      The root of the btree.
4033  * @path:      A path pointing to the target leaf and slot.
4034  * @key:       The key of the new item.
4035  * @data_size: The size of the data associated with the new key.
4036  */
4037 void btrfs_setup_item_for_insert(struct btrfs_root *root,
4038                                  struct btrfs_path *path,
4039                                  const struct btrfs_key *key,
4040                                  u32 data_size)
4041 {
4042         struct btrfs_item_batch batch;
4043
4044         batch.keys = key;
4045         batch.data_sizes = &data_size;
4046         batch.total_data_size = data_size;
4047         batch.nr = 1;
4048
4049         setup_items_for_insert(root, path, &batch);
4050 }
4051
4052 /*
4053  * Given a key and some data, insert items into the tree.
4054  * This does all the path init required, making room in the tree if needed.
4055  */
4056 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
4057                             struct btrfs_root *root,
4058                             struct btrfs_path *path,
4059                             const struct btrfs_item_batch *batch)
4060 {
4061         int ret = 0;
4062         int slot;
4063         u32 total_size;
4064
4065         total_size = batch->total_data_size + (batch->nr * sizeof(struct btrfs_item));
4066         ret = btrfs_search_slot(trans, root, &batch->keys[0], path, total_size, 1);
4067         if (ret == 0)
4068                 return -EEXIST;
4069         if (ret < 0)
4070                 return ret;
4071
4072         slot = path->slots[0];
4073         BUG_ON(slot < 0);
4074
4075         setup_items_for_insert(root, path, batch);
4076         return 0;
4077 }
4078
4079 /*
4080  * Given a key and some data, insert an item into the tree.
4081  * This does all the path init required, making room in the tree if needed.
4082  */
4083 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4084                       const struct btrfs_key *cpu_key, void *data,
4085                       u32 data_size)
4086 {
4087         int ret = 0;
4088         struct btrfs_path *path;
4089         struct extent_buffer *leaf;
4090         unsigned long ptr;
4091
4092         path = btrfs_alloc_path();
4093         if (!path)
4094                 return -ENOMEM;
4095         ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size);
4096         if (!ret) {
4097                 leaf = path->nodes[0];
4098                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
4099                 write_extent_buffer(leaf, data, ptr, data_size);
4100                 btrfs_mark_buffer_dirty(leaf);
4101         }
4102         btrfs_free_path(path);
4103         return ret;
4104 }
4105
4106 /*
4107  * This function duplicates an item, giving 'new_key' to the new item.
4108  * It guarantees both items live in the same tree leaf and the new item is
4109  * contiguous with the original item.
4110  *
4111  * This allows us to split a file extent in place, keeping a lock on the leaf
4112  * the entire time.
4113  */
4114 int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
4115                          struct btrfs_root *root,
4116                          struct btrfs_path *path,
4117                          const struct btrfs_key *new_key)
4118 {
4119         struct extent_buffer *leaf;
4120         int ret;
4121         u32 item_size;
4122
4123         leaf = path->nodes[0];
4124         item_size = btrfs_item_size(leaf, path->slots[0]);
4125         ret = setup_leaf_for_split(trans, root, path,
4126                                    item_size + sizeof(struct btrfs_item));
4127         if (ret)
4128                 return ret;
4129
4130         path->slots[0]++;
4131         btrfs_setup_item_for_insert(root, path, new_key, item_size);
4132         leaf = path->nodes[0];
4133         memcpy_extent_buffer(leaf,
4134                              btrfs_item_ptr_offset(leaf, path->slots[0]),
4135                              btrfs_item_ptr_offset(leaf, path->slots[0] - 1),
4136                              item_size);
4137         return 0;
4138 }
4139
4140 /*
4141  * delete the pointer from a given node.
4142  *
4143  * the tree should have been previously balanced so the deletion does not
4144  * empty a node.
4145  */
4146 static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
4147                     int level, int slot)
4148 {
4149         struct extent_buffer *parent = path->nodes[level];
4150         u32 nritems;
4151         int ret;
4152
4153         nritems = btrfs_header_nritems(parent);
4154         if (slot != nritems - 1) {
4155                 if (level) {
4156                         ret = btrfs_tree_mod_log_insert_move(parent, slot,
4157                                         slot + 1, nritems - slot - 1);
4158                         BUG_ON(ret < 0);
4159                 }
4160                 memmove_extent_buffer(parent,
4161                               btrfs_node_key_ptr_offset(slot),
4162                               btrfs_node_key_ptr_offset(slot + 1),
4163                               sizeof(struct btrfs_key_ptr) *
4164                               (nritems - slot - 1));
4165         } else if (level) {
4166                 ret = btrfs_tree_mod_log_insert_key(parent, slot,
4167                                 BTRFS_MOD_LOG_KEY_REMOVE, GFP_NOFS);
4168                 BUG_ON(ret < 0);
4169         }
4170
4171         nritems--;
4172         btrfs_set_header_nritems(parent, nritems);
4173         if (nritems == 0 && parent == root->node) {
4174                 BUG_ON(btrfs_header_level(root->node) != 1);
4175                 /* just turn the root into a leaf and break */
4176                 btrfs_set_header_level(root->node, 0);
4177         } else if (slot == 0) {
4178                 struct btrfs_disk_key disk_key;
4179
4180                 btrfs_node_key(parent, &disk_key, 0);
4181                 fixup_low_keys(path, &disk_key, level + 1);
4182         }
4183         btrfs_mark_buffer_dirty(parent);
4184 }
4185
4186 /*
4187  * a helper function to delete the leaf pointed to by path->slots[1] and
4188  * path->nodes[1].
4189  *
4190  * This deletes the pointer in path->nodes[1] and frees the leaf
4191  * block extent.  zero is returned if it all worked out, < 0 otherwise.
4192  *
4193  * The path must have already been setup for deleting the leaf, including
4194  * all the proper balancing.  path->nodes[1] must be locked.
4195  */
4196 static noinline void btrfs_del_leaf(struct btrfs_trans_handle *trans,
4197                                     struct btrfs_root *root,
4198                                     struct btrfs_path *path,
4199                                     struct extent_buffer *leaf)
4200 {
4201         WARN_ON(btrfs_header_generation(leaf) != trans->transid);
4202         del_ptr(root, path, 1, path->slots[1]);
4203
4204         /*
4205          * btrfs_free_extent is expensive, we want to make sure we
4206          * aren't holding any locks when we call it
4207          */
4208         btrfs_unlock_up_safe(path, 0);
4209
4210         root_sub_used(root, leaf->len);
4211
4212         atomic_inc(&leaf->refs);
4213         btrfs_free_tree_block(trans, btrfs_root_id(root), leaf, 0, 1);
4214         free_extent_buffer_stale(leaf);
4215 }
4216 /*
4217  * delete the item at the leaf level in path.  If that empties
4218  * the leaf, remove it from the tree
4219  */
4220 int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4221                     struct btrfs_path *path, int slot, int nr)
4222 {
4223         struct btrfs_fs_info *fs_info = root->fs_info;
4224         struct extent_buffer *leaf;
4225         int ret = 0;
4226         int wret;
4227         u32 nritems;
4228
4229         leaf = path->nodes[0];
4230         nritems = btrfs_header_nritems(leaf);
4231
4232         if (slot + nr != nritems) {
4233                 const u32 last_off = btrfs_item_offset(leaf, slot + nr - 1);
4234                 const int data_end = leaf_data_end(leaf);
4235                 struct btrfs_map_token token;
4236                 u32 dsize = 0;
4237                 int i;
4238
4239                 for (i = 0; i < nr; i++)
4240                         dsize += btrfs_item_size(leaf, slot + i);
4241
4242                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4243                               data_end + dsize,
4244                               BTRFS_LEAF_DATA_OFFSET + data_end,
4245                               last_off - data_end);
4246
4247                 btrfs_init_map_token(&token, leaf);
4248                 for (i = slot + nr; i < nritems; i++) {
4249                         u32 ioff;
4250
4251                         ioff = btrfs_token_item_offset(&token, i);
4252                         btrfs_set_token_item_offset(&token, i, ioff + dsize);
4253                 }
4254
4255                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot),
4256                               btrfs_item_nr_offset(slot + nr),
4257                               sizeof(struct btrfs_item) *
4258                               (nritems - slot - nr));
4259         }
4260         btrfs_set_header_nritems(leaf, nritems - nr);
4261         nritems -= nr;
4262
4263         /* delete the leaf if we've emptied it */
4264         if (nritems == 0) {
4265                 if (leaf == root->node) {
4266                         btrfs_set_header_level(leaf, 0);
4267                 } else {
4268                         btrfs_clean_tree_block(leaf);
4269                         btrfs_del_leaf(trans, root, path, leaf);
4270                 }
4271         } else {
4272                 int used = leaf_space_used(leaf, 0, nritems);
4273                 if (slot == 0) {
4274                         struct btrfs_disk_key disk_key;
4275
4276                         btrfs_item_key(leaf, &disk_key, 0);
4277                         fixup_low_keys(path, &disk_key, 1);
4278                 }
4279
4280                 /*
4281                  * Try to delete the leaf if it is mostly empty. We do this by
4282                  * trying to move all its items into its left and right neighbours.
4283                  * If we can't move all the items, then we don't delete it - it's
4284                  * not ideal, but future insertions might fill the leaf with more
4285                  * items, or items from other leaves might be moved later into our
4286                  * leaf due to deletions on those leaves.
4287                  */
4288                 if (used < BTRFS_LEAF_DATA_SIZE(fs_info) / 3) {
4289                         u32 min_push_space;
4290
4291                         /* push_leaf_left fixes the path.
4292                          * make sure the path still points to our leaf
4293                          * for possible call to del_ptr below
4294                          */
4295                         slot = path->slots[1];
4296                         atomic_inc(&leaf->refs);
4297                         /*
4298                          * We want to be able to at least push one item to the
4299                          * left neighbour leaf, and that's the first item.
4300                          */
4301                         min_push_space = sizeof(struct btrfs_item) +
4302                                 btrfs_item_size(leaf, 0);
4303                         wret = push_leaf_left(trans, root, path, 0,
4304                                               min_push_space, 1, (u32)-1);
4305                         if (wret < 0 && wret != -ENOSPC)
4306                                 ret = wret;
4307
4308                         if (path->nodes[0] == leaf &&
4309                             btrfs_header_nritems(leaf)) {
4310                                 /*
4311                                  * If we were not able to push all items from our
4312                                  * leaf to its left neighbour, then attempt to
4313                                  * either push all the remaining items to the
4314                                  * right neighbour or none. There's no advantage
4315                                  * in pushing only some items, instead of all, as
4316                                  * it's pointless to end up with a leaf having
4317                                  * too few items while the neighbours can be full
4318                                  * or nearly full.
4319                                  */
4320                                 nritems = btrfs_header_nritems(leaf);
4321                                 min_push_space = leaf_space_used(leaf, 0, nritems);
4322                                 wret = push_leaf_right(trans, root, path, 0,
4323                                                        min_push_space, 1, 0);
4324                                 if (wret < 0 && wret != -ENOSPC)
4325                                         ret = wret;
4326                         }
4327
4328                         if (btrfs_header_nritems(leaf) == 0) {
4329                                 path->slots[1] = slot;
4330                                 btrfs_del_leaf(trans, root, path, leaf);
4331                                 free_extent_buffer(leaf);
4332                                 ret = 0;
4333                         } else {
4334                                 /* if we're still in the path, make sure
4335                                  * we're dirty.  Otherwise, one of the
4336                                  * push_leaf functions must have already
4337                                  * dirtied this buffer
4338                                  */
4339                                 if (path->nodes[0] == leaf)
4340                                         btrfs_mark_buffer_dirty(leaf);
4341                                 free_extent_buffer(leaf);
4342                         }
4343                 } else {
4344                         btrfs_mark_buffer_dirty(leaf);
4345                 }
4346         }
4347         return ret;
4348 }
4349
4350 /*
4351  * search the tree again to find a leaf with lesser keys
4352  * returns 0 if it found something or 1 if there are no lesser leaves.
4353  * returns < 0 on io errors.
4354  *
4355  * This may release the path, and so you may lose any locks held at the
4356  * time you call it.
4357  */
4358 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path)
4359 {
4360         struct btrfs_key key;
4361         struct btrfs_disk_key found_key;
4362         int ret;
4363
4364         btrfs_item_key_to_cpu(path->nodes[0], &key, 0);
4365
4366         if (key.offset > 0) {
4367                 key.offset--;
4368         } else if (key.type > 0) {
4369                 key.type--;
4370                 key.offset = (u64)-1;
4371         } else if (key.objectid > 0) {
4372                 key.objectid--;
4373                 key.type = (u8)-1;
4374                 key.offset = (u64)-1;
4375         } else {
4376                 return 1;
4377         }
4378
4379         btrfs_release_path(path);
4380         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
4381         if (ret < 0)
4382                 return ret;
4383         btrfs_item_key(path->nodes[0], &found_key, 0);
4384         ret = comp_keys(&found_key, &key);
4385         /*
4386          * We might have had an item with the previous key in the tree right
4387          * before we released our path. And after we released our path, that
4388          * item might have been pushed to the first slot (0) of the leaf we
4389          * were holding due to a tree balance. Alternatively, an item with the
4390          * previous key can exist as the only element of a leaf (big fat item).
4391          * Therefore account for these 2 cases, so that our callers (like
4392          * btrfs_previous_item) don't miss an existing item with a key matching
4393          * the previous key we computed above.
4394          */
4395         if (ret <= 0)
4396                 return 0;
4397         return 1;
4398 }
4399
4400 /*
4401  * A helper function to walk down the tree starting at min_key, and looking
4402  * for nodes or leaves that are have a minimum transaction id.
4403  * This is used by the btree defrag code, and tree logging
4404  *
4405  * This does not cow, but it does stuff the starting key it finds back
4406  * into min_key, so you can call btrfs_search_slot with cow=1 on the
4407  * key and get a writable path.
4408  *
4409  * This honors path->lowest_level to prevent descent past a given level
4410  * of the tree.
4411  *
4412  * min_trans indicates the oldest transaction that you are interested
4413  * in walking through.  Any nodes or leaves older than min_trans are
4414  * skipped over (without reading them).
4415  *
4416  * returns zero if something useful was found, < 0 on error and 1 if there
4417  * was nothing in the tree that matched the search criteria.
4418  */
4419 int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
4420                          struct btrfs_path *path,
4421                          u64 min_trans)
4422 {
4423         struct extent_buffer *cur;
4424         struct btrfs_key found_key;
4425         int slot;
4426         int sret;
4427         u32 nritems;
4428         int level;
4429         int ret = 1;
4430         int keep_locks = path->keep_locks;
4431
4432         path->keep_locks = 1;
4433 again:
4434         cur = btrfs_read_lock_root_node(root);
4435         level = btrfs_header_level(cur);
4436         WARN_ON(path->nodes[level]);
4437         path->nodes[level] = cur;
4438         path->locks[level] = BTRFS_READ_LOCK;
4439
4440         if (btrfs_header_generation(cur) < min_trans) {
4441                 ret = 1;
4442                 goto out;
4443         }
4444         while (1) {
4445                 nritems = btrfs_header_nritems(cur);
4446                 level = btrfs_header_level(cur);
4447                 sret = btrfs_bin_search(cur, min_key, &slot);
4448                 if (sret < 0) {
4449                         ret = sret;
4450                         goto out;
4451                 }
4452
4453                 /* at the lowest level, we're done, setup the path and exit */
4454                 if (level == path->lowest_level) {
4455                         if (slot >= nritems)
4456                                 goto find_next_key;
4457                         ret = 0;
4458                         path->slots[level] = slot;
4459                         btrfs_item_key_to_cpu(cur, &found_key, slot);
4460                         goto out;
4461                 }
4462                 if (sret && slot > 0)
4463                         slot--;
4464                 /*
4465                  * check this node pointer against the min_trans parameters.
4466                  * If it is too old, skip to the next one.
4467                  */
4468                 while (slot < nritems) {
4469                         u64 gen;
4470
4471                         gen = btrfs_node_ptr_generation(cur, slot);
4472                         if (gen < min_trans) {
4473                                 slot++;
4474                                 continue;
4475                         }
4476                         break;
4477                 }
4478 find_next_key:
4479                 /*
4480                  * we didn't find a candidate key in this node, walk forward
4481                  * and find another one
4482                  */
4483                 if (slot >= nritems) {
4484                         path->slots[level] = slot;
4485                         sret = btrfs_find_next_key(root, path, min_key, level,
4486                                                   min_trans);
4487                         if (sret == 0) {
4488                                 btrfs_release_path(path);
4489                                 goto again;
4490                         } else {
4491                                 goto out;
4492                         }
4493                 }
4494                 /* save our key for returning back */
4495                 btrfs_node_key_to_cpu(cur, &found_key, slot);
4496                 path->slots[level] = slot;
4497                 if (level == path->lowest_level) {
4498                         ret = 0;
4499                         goto out;
4500                 }
4501                 cur = btrfs_read_node_slot(cur, slot);
4502                 if (IS_ERR(cur)) {
4503                         ret = PTR_ERR(cur);
4504                         goto out;
4505                 }
4506
4507                 btrfs_tree_read_lock(cur);
4508
4509                 path->locks[level - 1] = BTRFS_READ_LOCK;
4510                 path->nodes[level - 1] = cur;
4511                 unlock_up(path, level, 1, 0, NULL);
4512         }
4513 out:
4514         path->keep_locks = keep_locks;
4515         if (ret == 0) {
4516                 btrfs_unlock_up_safe(path, path->lowest_level + 1);
4517                 memcpy(min_key, &found_key, sizeof(found_key));
4518         }
4519         return ret;
4520 }
4521
4522 /*
4523  * this is similar to btrfs_next_leaf, but does not try to preserve
4524  * and fixup the path.  It looks for and returns the next key in the
4525  * tree based on the current path and the min_trans parameters.
4526  *
4527  * 0 is returned if another key is found, < 0 if there are any errors
4528  * and 1 is returned if there are no higher keys in the tree
4529  *
4530  * path->keep_locks should be set to 1 on the search made before
4531  * calling this function.
4532  */
4533 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
4534                         struct btrfs_key *key, int level, u64 min_trans)
4535 {
4536         int slot;
4537         struct extent_buffer *c;
4538
4539         WARN_ON(!path->keep_locks && !path->skip_locking);
4540         while (level < BTRFS_MAX_LEVEL) {
4541                 if (!path->nodes[level])
4542                         return 1;
4543
4544                 slot = path->slots[level] + 1;
4545                 c = path->nodes[level];
4546 next:
4547                 if (slot >= btrfs_header_nritems(c)) {
4548                         int ret;
4549                         int orig_lowest;
4550                         struct btrfs_key cur_key;
4551                         if (level + 1 >= BTRFS_MAX_LEVEL ||
4552                             !path->nodes[level + 1])
4553                                 return 1;
4554
4555                         if (path->locks[level + 1] || path->skip_locking) {
4556                                 level++;
4557                                 continue;
4558                         }
4559
4560                         slot = btrfs_header_nritems(c) - 1;
4561                         if (level == 0)
4562                                 btrfs_item_key_to_cpu(c, &cur_key, slot);
4563                         else
4564                                 btrfs_node_key_to_cpu(c, &cur_key, slot);
4565
4566                         orig_lowest = path->lowest_level;
4567                         btrfs_release_path(path);
4568                         path->lowest_level = level;
4569                         ret = btrfs_search_slot(NULL, root, &cur_key, path,
4570                                                 0, 0);
4571                         path->lowest_level = orig_lowest;
4572                         if (ret < 0)
4573                                 return ret;
4574
4575                         c = path->nodes[level];
4576                         slot = path->slots[level];
4577                         if (ret == 0)
4578                                 slot++;
4579                         goto next;
4580                 }
4581
4582                 if (level == 0)
4583                         btrfs_item_key_to_cpu(c, key, slot);
4584                 else {
4585                         u64 gen = btrfs_node_ptr_generation(c, slot);
4586
4587                         if (gen < min_trans) {
4588                                 slot++;
4589                                 goto next;
4590                         }
4591                         btrfs_node_key_to_cpu(c, key, slot);
4592                 }
4593                 return 0;
4594         }
4595         return 1;
4596 }
4597
4598 int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path,
4599                         u64 time_seq)
4600 {
4601         int slot;
4602         int level;
4603         struct extent_buffer *c;
4604         struct extent_buffer *next;
4605         struct btrfs_fs_info *fs_info = root->fs_info;
4606         struct btrfs_key key;
4607         bool need_commit_sem = false;
4608         u32 nritems;
4609         int ret;
4610         int i;
4611
4612         nritems = btrfs_header_nritems(path->nodes[0]);
4613         if (nritems == 0)
4614                 return 1;
4615
4616         btrfs_item_key_to_cpu(path->nodes[0], &key, nritems - 1);
4617 again:
4618         level = 1;
4619         next = NULL;
4620         btrfs_release_path(path);
4621
4622         path->keep_locks = 1;
4623
4624         if (time_seq) {
4625                 ret = btrfs_search_old_slot(root, &key, path, time_seq);
4626         } else {
4627                 if (path->need_commit_sem) {
4628                         path->need_commit_sem = 0;
4629                         need_commit_sem = true;
4630                         down_read(&fs_info->commit_root_sem);
4631                 }
4632                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
4633         }
4634         path->keep_locks = 0;
4635
4636         if (ret < 0)
4637                 goto done;
4638
4639         nritems = btrfs_header_nritems(path->nodes[0]);
4640         /*
4641          * by releasing the path above we dropped all our locks.  A balance
4642          * could have added more items next to the key that used to be
4643          * at the very end of the block.  So, check again here and
4644          * advance the path if there are now more items available.
4645          */
4646         if (nritems > 0 && path->slots[0] < nritems - 1) {
4647                 if (ret == 0)
4648                         path->slots[0]++;
4649                 ret = 0;
4650                 goto done;
4651         }
4652         /*
4653          * So the above check misses one case:
4654          * - after releasing the path above, someone has removed the item that
4655          *   used to be at the very end of the block, and balance between leafs
4656          *   gets another one with bigger key.offset to replace it.
4657          *
4658          * This one should be returned as well, or we can get leaf corruption
4659          * later(esp. in __btrfs_drop_extents()).
4660          *
4661          * And a bit more explanation about this check,
4662          * with ret > 0, the key isn't found, the path points to the slot
4663          * where it should be inserted, so the path->slots[0] item must be the
4664          * bigger one.
4665          */
4666         if (nritems > 0 && ret > 0 && path->slots[0] == nritems - 1) {
4667                 ret = 0;
4668                 goto done;
4669         }
4670
4671         while (level < BTRFS_MAX_LEVEL) {
4672                 if (!path->nodes[level]) {
4673                         ret = 1;
4674                         goto done;
4675                 }
4676
4677                 slot = path->slots[level] + 1;
4678                 c = path->nodes[level];
4679                 if (slot >= btrfs_header_nritems(c)) {
4680                         level++;
4681                         if (level == BTRFS_MAX_LEVEL) {
4682                                 ret = 1;
4683                                 goto done;
4684                         }
4685                         continue;
4686                 }
4687
4688
4689                 /*
4690                  * Our current level is where we're going to start from, and to
4691                  * make sure lockdep doesn't complain we need to drop our locks
4692                  * and nodes from 0 to our current level.
4693                  */
4694                 for (i = 0; i < level; i++) {
4695                         if (path->locks[level]) {
4696                                 btrfs_tree_read_unlock(path->nodes[i]);
4697                                 path->locks[i] = 0;
4698                         }
4699                         free_extent_buffer(path->nodes[i]);
4700                         path->nodes[i] = NULL;
4701                 }
4702
4703                 next = c;
4704                 ret = read_block_for_search(root, path, &next, level,
4705                                             slot, &key);
4706                 if (ret == -EAGAIN)
4707                         goto again;
4708
4709                 if (ret < 0) {
4710                         btrfs_release_path(path);
4711                         goto done;
4712                 }
4713
4714                 if (!path->skip_locking) {
4715                         ret = btrfs_try_tree_read_lock(next);
4716                         if (!ret && time_seq) {
4717                                 /*
4718                                  * If we don't get the lock, we may be racing
4719                                  * with push_leaf_left, holding that lock while
4720                                  * itself waiting for the leaf we've currently
4721                                  * locked. To solve this situation, we give up
4722                                  * on our lock and cycle.
4723                                  */
4724                                 free_extent_buffer(next);
4725                                 btrfs_release_path(path);
4726                                 cond_resched();
4727                                 goto again;
4728                         }
4729                         if (!ret)
4730                                 btrfs_tree_read_lock(next);
4731                 }
4732                 break;
4733         }
4734         path->slots[level] = slot;
4735         while (1) {
4736                 level--;
4737                 path->nodes[level] = next;
4738                 path->slots[level] = 0;
4739                 if (!path->skip_locking)
4740                         path->locks[level] = BTRFS_READ_LOCK;
4741                 if (!level)
4742                         break;
4743
4744                 ret = read_block_for_search(root, path, &next, level,
4745                                             0, &key);
4746                 if (ret == -EAGAIN)
4747                         goto again;
4748
4749                 if (ret < 0) {
4750                         btrfs_release_path(path);
4751                         goto done;
4752                 }
4753
4754                 if (!path->skip_locking)
4755                         btrfs_tree_read_lock(next);
4756         }
4757         ret = 0;
4758 done:
4759         unlock_up(path, 0, 1, 0, NULL);
4760         if (need_commit_sem) {
4761                 int ret2;
4762
4763                 path->need_commit_sem = 1;
4764                 ret2 = finish_need_commit_sem_search(path);
4765                 up_read(&fs_info->commit_root_sem);
4766                 if (ret2)
4767                         ret = ret2;
4768         }
4769
4770         return ret;
4771 }
4772
4773 /*
4774  * this uses btrfs_prev_leaf to walk backwards in the tree, and keeps
4775  * searching until it gets past min_objectid or finds an item of 'type'
4776  *
4777  * returns 0 if something is found, 1 if nothing was found and < 0 on error
4778  */
4779 int btrfs_previous_item(struct btrfs_root *root,
4780                         struct btrfs_path *path, u64 min_objectid,
4781                         int type)
4782 {
4783         struct btrfs_key found_key;
4784         struct extent_buffer *leaf;
4785         u32 nritems;
4786         int ret;
4787
4788         while (1) {
4789                 if (path->slots[0] == 0) {
4790                         ret = btrfs_prev_leaf(root, path);
4791                         if (ret != 0)
4792                                 return ret;
4793                 } else {
4794                         path->slots[0]--;
4795                 }
4796                 leaf = path->nodes[0];
4797                 nritems = btrfs_header_nritems(leaf);
4798                 if (nritems == 0)
4799                         return 1;
4800                 if (path->slots[0] == nritems)
4801                         path->slots[0]--;
4802
4803                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4804                 if (found_key.objectid < min_objectid)
4805                         break;
4806                 if (found_key.type == type)
4807                         return 0;
4808                 if (found_key.objectid == min_objectid &&
4809                     found_key.type < type)
4810                         break;
4811         }
4812         return 1;
4813 }
4814
4815 /*
4816  * search in extent tree to find a previous Metadata/Data extent item with
4817  * min objecitd.
4818  *
4819  * returns 0 if something is found, 1 if nothing was found and < 0 on error
4820  */
4821 int btrfs_previous_extent_item(struct btrfs_root *root,
4822                         struct btrfs_path *path, u64 min_objectid)
4823 {
4824         struct btrfs_key found_key;
4825         struct extent_buffer *leaf;
4826         u32 nritems;
4827         int ret;
4828
4829         while (1) {
4830                 if (path->slots[0] == 0) {
4831                         ret = btrfs_prev_leaf(root, path);
4832                         if (ret != 0)
4833                                 return ret;
4834                 } else {
4835                         path->slots[0]--;
4836                 }
4837                 leaf = path->nodes[0];
4838                 nritems = btrfs_header_nritems(leaf);
4839                 if (nritems == 0)
4840                         return 1;
4841                 if (path->slots[0] == nritems)
4842                         path->slots[0]--;
4843
4844                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4845                 if (found_key.objectid < min_objectid)
4846                         break;
4847                 if (found_key.type == BTRFS_EXTENT_ITEM_KEY ||
4848                     found_key.type == BTRFS_METADATA_ITEM_KEY)
4849                         return 0;
4850                 if (found_key.objectid == min_objectid &&
4851                     found_key.type < BTRFS_EXTENT_ITEM_KEY)
4852                         break;
4853         }
4854         return 1;
4855 }