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 "ctree.h"
  11 #include "disk-io.h"
  12 #include "transaction.h"
  13 #include "print-tree.h"
  14 #include "locking.h"
  15 #include "volumes.h"
  16 #include "qgroup.h"
  17
  18 static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root
  19                       *root, struct btrfs_path *path, int level);
  20 static int split_leaf(struct btrfs_trans_handle *trans, struct btrfs_root *root,
  21                       const struct btrfs_key *ins_key, struct btrfs_path *path,
  22                       int data_size, int extend);
  23 static int push_node_left(struct btrfs_trans_handle *trans,
  24                           struct extent_buffer *dst,
  25                           struct extent_buffer *src, int empty);
  26 static int balance_node_right(struct btrfs_trans_handle *trans,
  27                               struct extent_buffer *dst_buf,
  28                               struct extent_buffer *src_buf);
  29 static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
  30                     int level, int slot);
  31
  32 static const struct btrfs_csums {
  33         u16             size;
  34         const char      name[10];
  35         const char      driver[12];
  36 } btrfs_csums[] = {
  37         [BTRFS_CSUM_TYPE_CRC32] = { .size = 4, .name = "crc32c" },
  38         [BTRFS_CSUM_TYPE_XXHASH] = { .size = 8, .name = "xxhash64" },
  39         [BTRFS_CSUM_TYPE_SHA256] = { .size = 32, .name = "sha256" },
  40         [BTRFS_CSUM_TYPE_BLAKE2] = { .size = 32, .name = "blake2b",
  41                                      .driver = "blake2b-256" },
  42 };
  43
  44 int btrfs_super_csum_size(const struct btrfs_super_block *s)
  45 {
  46         u16 t = btrfs_super_csum_type(s);
  47         /*
  48          * csum type is validated at mount time
  49          */
  50         return btrfs_csums[t].size;
  51 }
  52
  53 const char *btrfs_super_csum_name(u16 csum_type)
  54 {
  55         /* csum type is validated at mount time */
  56         return btrfs_csums[csum_type].name;
  57 }
  58
  59 /*
  60  * Return driver name if defined, otherwise the name that's also a valid driver
  61  * name
  62  */
  63 const char *btrfs_super_csum_driver(u16 csum_type)
  64 {
  65         /* csum type is validated at mount time */
  66         return btrfs_csums[csum_type].driver[0] ?
  67                 btrfs_csums[csum_type].driver :
  68                 btrfs_csums[csum_type].name;
  69 }
  70
  71 size_t __attribute_const__ btrfs_get_num_csums(void)
  72 {
  73         return ARRAY_SIZE(btrfs_csums);
  74 }
  75
  76 struct btrfs_path *btrfs_alloc_path(void)
  77 {
  78         return kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS);
  79 }
  80
  81 /* this also releases the path */
  82 void btrfs_free_path(struct btrfs_path *p)
  83 {
  84         if (!p)
  85                 return;
  86         btrfs_release_path(p);
  87         kmem_cache_free(btrfs_path_cachep, p);
  88 }
  89
  90 /*
  91  * path release drops references on the extent buffers in the path
  92  * and it drops any locks held by this path
  93  *
  94  * It is safe to call this on paths that no locks or extent buffers held.
  95  */
  96 noinline void btrfs_release_path(struct btrfs_path *p)
  97 {
  98         int i;
  99
 100         for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
 101                 p->slots[i] = 0;
 102                 if (!p->nodes[i])
 103                         continue;
 104                 if (p->locks[i]) {
 105                         btrfs_tree_unlock_rw(p->nodes[i], p->locks[i]);
 106                         p->locks[i] = 0;
 107                 }
 108                 free_extent_buffer(p->nodes[i]);
 109                 p->nodes[i] = NULL;
 110         }
 111 }
 112
 113 /*
 114  * safely gets a reference on the root node of a tree.  A lock
 115  * is not taken, so a concurrent writer may put a different node
 116  * at the root of the tree.  See btrfs_lock_root_node for the
 117  * looping required.
 118  *
 119  * The extent buffer returned by this has a reference taken, so
 120  * it won't disappear.  It may stop being the root of the tree
 121  * at any time because there are no locks held.
 122  */
 123 struct extent_buffer *btrfs_root_node(struct btrfs_root *root)
 124 {
 125         struct extent_buffer *eb;
 126
 127         while (1) {
 128                 rcu_read_lock();
 129                 eb = rcu_dereference(root->node);
 130
 131                 /*
 132                  * RCU really hurts here, we could free up the root node because
 133                  * it was COWed but we may not get the new root node yet so do
 134                  * the inc_not_zero dance and if it doesn't work then
 135                  * synchronize_rcu and try again.
 136                  */
 137                 if (atomic_inc_not_zero(&eb->refs)) {
 138                         rcu_read_unlock();
 139                         break;
 140                 }
 141                 rcu_read_unlock();
 142                 synchronize_rcu();
 143         }
 144         return eb;
 145 }
 146
 147 /*
 148  * Cowonly root (not-shareable trees, everything not subvolume or reloc roots),
 149  * just get put onto a simple dirty list.  Transaction walks this list to make
 150  * sure they get properly updated on disk.
 151  */
 152 static void add_root_to_dirty_list(struct btrfs_root *root)
 153 {
 154         struct btrfs_fs_info *fs_info = root->fs_info;
 155
 156         if (test_bit(BTRFS_ROOT_DIRTY, &root->state) ||
 157             !test_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state))
 158                 return;
 159
 160         spin_lock(&fs_info->trans_lock);
 161         if (!test_and_set_bit(BTRFS_ROOT_DIRTY, &root->state)) {
 162                 /* Want the extent tree to be the last on the list */
 163                 if (root->root_key.objectid == BTRFS_EXTENT_TREE_OBJECTID)
 164                         list_move_tail(&root->dirty_list,
 165                                        &fs_info->dirty_cowonly_roots);
 166                 else
 167                         list_move(&root->dirty_list,
 168                                   &fs_info->dirty_cowonly_roots);
 169         }
 170         spin_unlock(&fs_info->trans_lock);
 171 }
 172
 173 /*
 174  * used by snapshot creation to make a copy of a root for a tree with
 175  * a given objectid.  The buffer with the new root node is returned in
 176  * cow_ret, and this func returns zero on success or a negative error code.
 177  */
 178 int btrfs_copy_root(struct btrfs_trans_handle *trans,
 179                       struct btrfs_root *root,
 180                       struct extent_buffer *buf,
 181                       struct extent_buffer **cow_ret, u64 new_root_objectid)
 182 {
 183         struct btrfs_fs_info *fs_info = root->fs_info;
 184         struct extent_buffer *cow;
 185         int ret = 0;
 186         int level;
 187         struct btrfs_disk_key disk_key;
 188
 189         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 190                 trans->transid != fs_info->running_transaction->transid);
 191         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 192                 trans->transid != root->last_trans);
 193
 194         level = btrfs_header_level(buf);
 195         if (level == 0)
 196                 btrfs_item_key(buf, &disk_key, 0);
 197         else
 198                 btrfs_node_key(buf, &disk_key, 0);
 199
 200         cow = btrfs_alloc_tree_block(trans, root, 0, new_root_objectid,
 201                         &disk_key, level, buf->start, 0);
 202         if (IS_ERR(cow))
 203                 return PTR_ERR(cow);
 204
 205         copy_extent_buffer_full(cow, buf);
 206         btrfs_set_header_bytenr(cow, cow->start);
 207         btrfs_set_header_generation(cow, trans->transid);
 208         btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
 209         btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
 210                                      BTRFS_HEADER_FLAG_RELOC);
 211         if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
 212                 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
 213         else
 214                 btrfs_set_header_owner(cow, new_root_objectid);
 215
 216         write_extent_buffer_fsid(cow, fs_info->fs_devices->metadata_uuid);
 217
 218         WARN_ON(btrfs_header_generation(buf) > trans->transid);
 219         if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID)
 220                 ret = btrfs_inc_ref(trans, root, cow, 1);
 221         else
 222                 ret = btrfs_inc_ref(trans, root, cow, 0);
 223
 224         if (ret)
 225                 return ret;
 226
 227         btrfs_mark_buffer_dirty(cow);
 228         *cow_ret = cow;
 229         return 0;
 230 }
 231
 232 enum mod_log_op {
 233         MOD_LOG_KEY_REPLACE,
 234         MOD_LOG_KEY_ADD,
 235         MOD_LOG_KEY_REMOVE,
 236         MOD_LOG_KEY_REMOVE_WHILE_FREEING,
 237         MOD_LOG_KEY_REMOVE_WHILE_MOVING,
 238         MOD_LOG_MOVE_KEYS,
 239         MOD_LOG_ROOT_REPLACE,
 240 };
 241
 242 struct tree_mod_root {
 243         u64 logical;
 244         u8 level;
 245 };
 246
 247 struct tree_mod_elem {
 248         struct rb_node node;
 249         u64 logical;
 250         u64 seq;
 251         enum mod_log_op op;
 252
 253         /* this is used for MOD_LOG_KEY_* and MOD_LOG_MOVE_KEYS operations */
 254         int slot;
 255
 256         /* this is used for MOD_LOG_KEY* and MOD_LOG_ROOT_REPLACE */
 257         u64 generation;
 258
 259         /* those are used for op == MOD_LOG_KEY_{REPLACE,REMOVE} */
 260         struct btrfs_disk_key key;
 261         u64 blockptr;
 262
 263         /* this is used for op == MOD_LOG_MOVE_KEYS */
 264         struct {
 265                 int dst_slot;
 266                 int nr_items;
 267         } move;
 268
 269         /* this is used for op == MOD_LOG_ROOT_REPLACE */
 270         struct tree_mod_root old_root;
 271 };
 272
 273 /*
 274  * Pull a new tree mod seq number for our operation.
 275  */
 276 static inline u64 btrfs_inc_tree_mod_seq(struct btrfs_fs_info *fs_info)
 277 {
 278         return atomic64_inc_return(&fs_info->tree_mod_seq);
 279 }
 280
 281 /*
 282  * This adds a new blocker to the tree mod log's blocker list if the @elem
 283  * passed does not already have a sequence number set. So when a caller expects
 284  * to record tree modifications, it should ensure to set elem->seq to zero
 285  * before calling btrfs_get_tree_mod_seq.
 286  * Returns a fresh, unused tree log modification sequence number, even if no new
 287  * blocker was added.
 288  */
 289 u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info,
 290                            struct seq_list *elem)
 291 {
 292         write_lock(&fs_info->tree_mod_log_lock);
 293         if (!elem->seq) {
 294                 elem->seq = btrfs_inc_tree_mod_seq(fs_info);
 295                 list_add_tail(&elem->list, &fs_info->tree_mod_seq_list);
 296         }
 297         write_unlock(&fs_info->tree_mod_log_lock);
 298
 299         return elem->seq;
 300 }
 301
 302 void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info,
 303                             struct seq_list *elem)
 304 {
 305         struct rb_root *tm_root;
 306         struct rb_node *node;
 307         struct rb_node *next;
 308         struct tree_mod_elem *tm;
 309         u64 min_seq = (u64)-1;
 310         u64 seq_putting = elem->seq;
 311
 312         if (!seq_putting)
 313                 return;
 314
 315         write_lock(&fs_info->tree_mod_log_lock);
 316         list_del(&elem->list);
 317         elem->seq = 0;
 318
 319         if (!list_empty(&fs_info->tree_mod_seq_list)) {
 320                 struct seq_list *first;
 321
 322                 first = list_first_entry(&fs_info->tree_mod_seq_list,
 323                                          struct seq_list, list);
 324                 if (seq_putting > first->seq) {
 325                         /*
 326                          * Blocker with lower sequence number exists, we
 327                          * cannot remove anything from the log.
 328                          */
 329                         write_unlock(&fs_info->tree_mod_log_lock);
 330                         return;
 331                 }
 332                 min_seq = first->seq;
 333         }
 334
 335         /*
 336          * anything that's lower than the lowest existing (read: blocked)
 337          * sequence number can be removed from the tree.
 338          */
 339         tm_root = &fs_info->tree_mod_log;
 340         for (node = rb_first(tm_root); node; node = next) {
 341                 next = rb_next(node);
 342                 tm = rb_entry(node, struct tree_mod_elem, node);
 343                 if (tm->seq >= min_seq)
 344                         continue;
 345                 rb_erase(node, tm_root);
 346                 kfree(tm);
 347         }
 348         write_unlock(&fs_info->tree_mod_log_lock);
 349 }
 350
 351 /*
 352  * key order of the log:
 353  *       node/leaf start address -> sequence
 354  *
 355  * The 'start address' is the logical address of the *new* root node
 356  * for root replace operations, or the logical address of the affected
 357  * block for all other operations.
 358  */
 359 static noinline int
 360 __tree_mod_log_insert(struct btrfs_fs_info *fs_info, struct tree_mod_elem *tm)
 361 {
 362         struct rb_root *tm_root;
 363         struct rb_node **new;
 364         struct rb_node *parent = NULL;
 365         struct tree_mod_elem *cur;
 366
 367         lockdep_assert_held_write(&fs_info->tree_mod_log_lock);
 368
 369         tm->seq = btrfs_inc_tree_mod_seq(fs_info);
 370
 371         tm_root = &fs_info->tree_mod_log;
 372         new = &tm_root->rb_node;
 373         while (*new) {
 374                 cur = rb_entry(*new, struct tree_mod_elem, node);
 375                 parent = *new;
 376                 if (cur->logical < tm->logical)
 377                         new = &((*new)->rb_left);
 378                 else if (cur->logical > tm->logical)
 379                         new = &((*new)->rb_right);
 380                 else if (cur->seq < tm->seq)
 381                         new = &((*new)->rb_left);
 382                 else if (cur->seq > tm->seq)
 383                         new = &((*new)->rb_right);
 384                 else
 385                         return -EEXIST;
 386         }
 387
 388         rb_link_node(&tm->node, parent, new);
 389         rb_insert_color(&tm->node, tm_root);
 390         return 0;
 391 }
 392
 393 /*
 394  * Determines if logging can be omitted. Returns 1 if it can. Otherwise, it
 395  * returns zero with the tree_mod_log_lock acquired. The caller must hold
 396  * this until all tree mod log insertions are recorded in the rb tree and then
 397  * write unlock fs_info::tree_mod_log_lock.
 398  */
 399 static inline int tree_mod_dont_log(struct btrfs_fs_info *fs_info,
 400                                     struct extent_buffer *eb) {
 401         smp_mb();
 402         if (list_empty(&(fs_info)->tree_mod_seq_list))
 403                 return 1;
 404         if (eb && btrfs_header_level(eb) == 0)
 405                 return 1;
 406
 407         write_lock(&fs_info->tree_mod_log_lock);
 408         if (list_empty(&(fs_info)->tree_mod_seq_list)) {
 409                 write_unlock(&fs_info->tree_mod_log_lock);
 410                 return 1;
 411         }
 412
 413         return 0;
 414 }
 415
 416 /* Similar to tree_mod_dont_log, but doesn't acquire any locks. */
 417 static inline int tree_mod_need_log(const struct btrfs_fs_info *fs_info,
 418                                     struct extent_buffer *eb)
 419 {
 420         smp_mb();
 421         if (list_empty(&(fs_info)->tree_mod_seq_list))
 422                 return 0;
 423         if (eb && btrfs_header_level(eb) == 0)
 424                 return 0;
 425
 426         return 1;
 427 }
 428
 429 static struct tree_mod_elem *
 430 alloc_tree_mod_elem(struct extent_buffer *eb, int slot,
 431                     enum mod_log_op op, gfp_t flags)
 432 {
 433         struct tree_mod_elem *tm;
 434
 435         tm = kzalloc(sizeof(*tm), flags);
 436         if (!tm)
 437                 return NULL;
 438
 439         tm->logical = eb->start;
 440         if (op != MOD_LOG_KEY_ADD) {
 441                 btrfs_node_key(eb, &tm->key, slot);
 442                 tm->blockptr = btrfs_node_blockptr(eb, slot);
 443         }
 444         tm->op = op;
 445         tm->slot = slot;
 446         tm->generation = btrfs_node_ptr_generation(eb, slot);
 447         RB_CLEAR_NODE(&tm->node);
 448
 449         return tm;
 450 }
 451
 452 static noinline int tree_mod_log_insert_key(struct extent_buffer *eb, int slot,
 453                 enum mod_log_op op, gfp_t flags)
 454 {
 455         struct tree_mod_elem *tm;
 456         int ret;
 457
 458         if (!tree_mod_need_log(eb->fs_info, eb))
 459                 return 0;
 460
 461         tm = alloc_tree_mod_elem(eb, slot, op, flags);
 462         if (!tm)
 463                 return -ENOMEM;
 464
 465         if (tree_mod_dont_log(eb->fs_info, eb)) {
 466                 kfree(tm);
 467                 return 0;
 468         }
 469
 470         ret = __tree_mod_log_insert(eb->fs_info, tm);
 471         write_unlock(&eb->fs_info->tree_mod_log_lock);
 472         if (ret)
 473                 kfree(tm);
 474
 475         return ret;
 476 }
 477
 478 static noinline int tree_mod_log_insert_move(struct extent_buffer *eb,
 479                 int dst_slot, int src_slot, int nr_items)
 480 {
 481         struct tree_mod_elem *tm = NULL;
 482         struct tree_mod_elem **tm_list = NULL;
 483         int ret = 0;
 484         int i;
 485         int locked = 0;
 486
 487         if (!tree_mod_need_log(eb->fs_info, eb))
 488                 return 0;
 489
 490         tm_list = kcalloc(nr_items, sizeof(struct tree_mod_elem *), GFP_NOFS);
 491         if (!tm_list)
 492                 return -ENOMEM;
 493
 494         tm = kzalloc(sizeof(*tm), GFP_NOFS);
 495         if (!tm) {
 496                 ret = -ENOMEM;
 497                 goto free_tms;
 498         }
 499
 500         tm->logical = eb->start;
 501         tm->slot = src_slot;
 502         tm->move.dst_slot = dst_slot;
 503         tm->move.nr_items = nr_items;
 504         tm->op = MOD_LOG_MOVE_KEYS;
 505
 506         for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) {
 507                 tm_list[i] = alloc_tree_mod_elem(eb, i + dst_slot,
 508                     MOD_LOG_KEY_REMOVE_WHILE_MOVING, GFP_NOFS);
 509                 if (!tm_list[i]) {
 510                         ret = -ENOMEM;
 511                         goto free_tms;
 512                 }
 513         }
 514
 515         if (tree_mod_dont_log(eb->fs_info, eb))
 516                 goto free_tms;
 517         locked = 1;
 518
 519         /*
 520          * When we override something during the move, we log these removals.
 521          * This can only happen when we move towards the beginning of the
 522          * buffer, i.e. dst_slot < src_slot.
 523          */
 524         for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) {
 525                 ret = __tree_mod_log_insert(eb->fs_info, tm_list[i]);
 526                 if (ret)
 527                         goto free_tms;
 528         }
 529
 530         ret = __tree_mod_log_insert(eb->fs_info, tm);
 531         if (ret)
 532                 goto free_tms;
 533         write_unlock(&eb->fs_info->tree_mod_log_lock);
 534         kfree(tm_list);
 535
 536         return 0;
 537 free_tms:
 538         for (i = 0; i < nr_items; i++) {
 539                 if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node))
 540                         rb_erase(&tm_list[i]->node, &eb->fs_info->tree_mod_log);
 541                 kfree(tm_list[i]);
 542         }
 543         if (locked)
 544                 write_unlock(&eb->fs_info->tree_mod_log_lock);
 545         kfree(tm_list);
 546         kfree(tm);
 547
 548         return ret;
 549 }
 550
 551 static inline int
 552 __tree_mod_log_free_eb(struct btrfs_fs_info *fs_info,
 553                        struct tree_mod_elem **tm_list,
 554                        int nritems)
 555 {
 556         int i, j;
 557         int ret;
 558
 559         for (i = nritems - 1; i >= 0; i--) {
 560                 ret = __tree_mod_log_insert(fs_info, tm_list[i]);
 561                 if (ret) {
 562                         for (j = nritems - 1; j > i; j--)
 563                                 rb_erase(&tm_list[j]->node,
 564                                          &fs_info->tree_mod_log);
 565                         return ret;
 566                 }
 567         }
 568
 569         return 0;
 570 }
 571
 572 static noinline int tree_mod_log_insert_root(struct extent_buffer *old_root,
 573                          struct extent_buffer *new_root, int log_removal)
 574 {
 575         struct btrfs_fs_info *fs_info = old_root->fs_info;
 576         struct tree_mod_elem *tm = NULL;
 577         struct tree_mod_elem **tm_list = NULL;
 578         int nritems = 0;
 579         int ret = 0;
 580         int i;
 581
 582         if (!tree_mod_need_log(fs_info, NULL))
 583                 return 0;
 584
 585         if (log_removal && btrfs_header_level(old_root) > 0) {
 586                 nritems = btrfs_header_nritems(old_root);
 587                 tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *),
 588                                   GFP_NOFS);
 589                 if (!tm_list) {
 590                         ret = -ENOMEM;
 591                         goto free_tms;
 592                 }
 593                 for (i = 0; i < nritems; i++) {
 594                         tm_list[i] = alloc_tree_mod_elem(old_root, i,
 595                             MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS);
 596                         if (!tm_list[i]) {
 597                                 ret = -ENOMEM;
 598                                 goto free_tms;
 599                         }
 600                 }
 601         }
 602
 603         tm = kzalloc(sizeof(*tm), GFP_NOFS);
 604         if (!tm) {
 605                 ret = -ENOMEM;
 606                 goto free_tms;
 607         }
 608
 609         tm->logical = new_root->start;
 610         tm->old_root.logical = old_root->start;
 611         tm->old_root.level = btrfs_header_level(old_root);
 612         tm->generation = btrfs_header_generation(old_root);
 613         tm->op = MOD_LOG_ROOT_REPLACE;
 614
 615         if (tree_mod_dont_log(fs_info, NULL))
 616                 goto free_tms;
 617
 618         if (tm_list)
 619                 ret = __tree_mod_log_free_eb(fs_info, tm_list, nritems);
 620         if (!ret)
 621                 ret = __tree_mod_log_insert(fs_info, tm);
 622
 623         write_unlock(&fs_info->tree_mod_log_lock);
 624         if (ret)
 625                 goto free_tms;
 626         kfree(tm_list);
 627
 628         return ret;
 629
 630 free_tms:
 631         if (tm_list) {
 632                 for (i = 0; i < nritems; i++)
 633                         kfree(tm_list[i]);
 634                 kfree(tm_list);
 635         }
 636         kfree(tm);
 637
 638         return ret;
 639 }
 640
 641 static struct tree_mod_elem *
 642 __tree_mod_log_search(struct btrfs_fs_info *fs_info, u64 start, u64 min_seq,
 643                       int smallest)
 644 {
 645         struct rb_root *tm_root;
 646         struct rb_node *node;
 647         struct tree_mod_elem *cur = NULL;
 648         struct tree_mod_elem *found = NULL;
 649
 650         read_lock(&fs_info->tree_mod_log_lock);
 651         tm_root = &fs_info->tree_mod_log;
 652         node = tm_root->rb_node;
 653         while (node) {
 654                 cur = rb_entry(node, struct tree_mod_elem, node);
 655                 if (cur->logical < start) {
 656                         node = node->rb_left;
 657                 } else if (cur->logical > start) {
 658                         node = node->rb_right;
 659                 } else if (cur->seq < min_seq) {
 660                         node = node->rb_left;
 661                 } else if (!smallest) {
 662                         /* we want the node with the highest seq */
 663                         if (found)
 664                                 BUG_ON(found->seq > cur->seq);
 665                         found = cur;
 666                         node = node->rb_left;
 667                 } else if (cur->seq > min_seq) {
 668                         /* we want the node with the smallest seq */
 669                         if (found)
 670                                 BUG_ON(found->seq < cur->seq);
 671                         found = cur;
 672                         node = node->rb_right;
 673                 } else {
 674                         found = cur;
 675                         break;
 676                 }
 677         }
 678         read_unlock(&fs_info->tree_mod_log_lock);
 679
 680         return found;
 681 }
 682
 683 /*
 684  * this returns the element from the log with the smallest time sequence
 685  * value that's in the log (the oldest log item). any element with a time
 686  * sequence lower than min_seq will be ignored.
 687  */
 688 static struct tree_mod_elem *
 689 tree_mod_log_search_oldest(struct btrfs_fs_info *fs_info, u64 start,
 690                            u64 min_seq)
 691 {
 692         return __tree_mod_log_search(fs_info, start, min_seq, 1);
 693 }
 694
 695 /*
 696  * this returns the element from the log with the largest time sequence
 697  * value that's in the log (the most recent log item). any element with
 698  * a time sequence lower than min_seq will be ignored.
 699  */
 700 static struct tree_mod_elem *
 701 tree_mod_log_search(struct btrfs_fs_info *fs_info, u64 start, u64 min_seq)
 702 {
 703         return __tree_mod_log_search(fs_info, start, min_seq, 0);
 704 }
 705
 706 static noinline int tree_mod_log_eb_copy(struct extent_buffer *dst,
 707                      struct extent_buffer *src, unsigned long dst_offset,
 708                      unsigned long src_offset, int nr_items)
 709 {
 710         struct btrfs_fs_info *fs_info = dst->fs_info;
 711         int ret = 0;
 712         struct tree_mod_elem **tm_list = NULL;
 713         struct tree_mod_elem **tm_list_add, **tm_list_rem;
 714         int i;
 715         int locked = 0;
 716
 717         if (!tree_mod_need_log(fs_info, NULL))
 718                 return 0;
 719
 720         if (btrfs_header_level(dst) == 0 && btrfs_header_level(src) == 0)
 721                 return 0;
 722
 723         tm_list = kcalloc(nr_items * 2, sizeof(struct tree_mod_elem *),
 724                           GFP_NOFS);
 725         if (!tm_list)
 726                 return -ENOMEM;
 727
 728         tm_list_add = tm_list;
 729         tm_list_rem = tm_list + nr_items;
 730         for (i = 0; i < nr_items; i++) {
 731                 tm_list_rem[i] = alloc_tree_mod_elem(src, i + src_offset,
 732                     MOD_LOG_KEY_REMOVE, GFP_NOFS);
 733                 if (!tm_list_rem[i]) {
 734                         ret = -ENOMEM;
 735                         goto free_tms;
 736                 }
 737
 738                 tm_list_add[i] = alloc_tree_mod_elem(dst, i + dst_offset,
 739                     MOD_LOG_KEY_ADD, GFP_NOFS);
 740                 if (!tm_list_add[i]) {
 741                         ret = -ENOMEM;
 742                         goto free_tms;
 743                 }
 744         }
 745
 746         if (tree_mod_dont_log(fs_info, NULL))
 747                 goto free_tms;
 748         locked = 1;
 749
 750         for (i = 0; i < nr_items; i++) {
 751                 ret = __tree_mod_log_insert(fs_info, tm_list_rem[i]);
 752                 if (ret)
 753                         goto free_tms;
 754                 ret = __tree_mod_log_insert(fs_info, tm_list_add[i]);
 755                 if (ret)
 756                         goto free_tms;
 757         }
 758
 759         write_unlock(&fs_info->tree_mod_log_lock);
 760         kfree(tm_list);
 761
 762         return 0;
 763
 764 free_tms:
 765         for (i = 0; i < nr_items * 2; i++) {
 766                 if (tm_list[i] && !RB_EMPTY_NODE(&tm_list[i]->node))
 767                         rb_erase(&tm_list[i]->node, &fs_info->tree_mod_log);
 768                 kfree(tm_list[i]);
 769         }
 770         if (locked)
 771                 write_unlock(&fs_info->tree_mod_log_lock);
 772         kfree(tm_list);
 773
 774         return ret;
 775 }
 776
 777 static noinline int tree_mod_log_free_eb(struct extent_buffer *eb)
 778 {
 779         struct tree_mod_elem **tm_list = NULL;
 780         int nritems = 0;
 781         int i;
 782         int ret = 0;
 783
 784         if (btrfs_header_level(eb) == 0)
 785                 return 0;
 786
 787         if (!tree_mod_need_log(eb->fs_info, NULL))
 788                 return 0;
 789
 790         nritems = btrfs_header_nritems(eb);
 791         tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *), GFP_NOFS);
 792         if (!tm_list)
 793                 return -ENOMEM;
 794
 795         for (i = 0; i < nritems; i++) {
 796                 tm_list[i] = alloc_tree_mod_elem(eb, i,
 797                     MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS);
 798                 if (!tm_list[i]) {
 799                         ret = -ENOMEM;
 800                         goto free_tms;
 801                 }
 802         }
 803
 804         if (tree_mod_dont_log(eb->fs_info, eb))
 805                 goto free_tms;
 806
 807         ret = __tree_mod_log_free_eb(eb->fs_info, tm_list, nritems);
 808         write_unlock(&eb->fs_info->tree_mod_log_lock);
 809         if (ret)
 810                 goto free_tms;
 811         kfree(tm_list);
 812
 813         return 0;
 814
 815 free_tms:
 816         for (i = 0; i < nritems; i++)
 817                 kfree(tm_list[i]);
 818         kfree(tm_list);
 819
 820         return ret;
 821 }
 822
 823 /*
 824  * check if the tree block can be shared by multiple trees
 825  */
 826 int btrfs_block_can_be_shared(struct btrfs_root *root,
 827                               struct extent_buffer *buf)
 828 {
 829         /*
 830          * Tree blocks not in shareable trees and tree roots are never shared.
 831          * If a block was allocated after the last snapshot and the block was
 832          * not allocated by tree relocation, we know the block is not shared.
 833          */
 834         if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
 835             buf != root->node && buf != root->commit_root &&
 836             (btrfs_header_generation(buf) <=
 837              btrfs_root_last_snapshot(&root->root_item) ||
 838              btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)))
 839                 return 1;
 840
 841         return 0;
 842 }
 843
 844 static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 845                                        struct btrfs_root *root,
 846                                        struct extent_buffer *buf,
 847                                        struct extent_buffer *cow,
 848                                        int *last_ref)
 849 {
 850         struct btrfs_fs_info *fs_info = root->fs_info;
 851         u64 refs;
 852         u64 owner;
 853         u64 flags;
 854         u64 new_flags = 0;
 855         int ret;
 856
 857         /*
 858          * Backrefs update rules:
 859          *
 860          * Always use full backrefs for extent pointers in tree block
 861          * allocated by tree relocation.
 862          *
 863          * If a shared tree block is no longer referenced by its owner
 864          * tree (btrfs_header_owner(buf) == root->root_key.objectid),
 865          * use full backrefs for extent pointers in tree block.
 866          *
 867          * If a tree block is been relocating
 868          * (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID),
 869          * use full backrefs for extent pointers in tree block.
 870          * The reason for this is some operations (such as drop tree)
 871          * are only allowed for blocks use full backrefs.
 872          */
 873
 874         if (btrfs_block_can_be_shared(root, buf)) {
 875                 ret = btrfs_lookup_extent_info(trans, fs_info, buf->start,
 876                                                btrfs_header_level(buf), 1,
 877                                                &refs, &flags);
 878                 if (ret)
 879                         return ret;
 880                 if (refs == 0) {
 881                         ret = -EROFS;
 882                         btrfs_handle_fs_error(fs_info, ret, NULL);
 883                         return ret;
 884                 }
 885         } else {
 886                 refs = 1;
 887                 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
 888                     btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
 889                         flags = BTRFS_BLOCK_FLAG_FULL_BACKREF;
 890                 else
 891                         flags = 0;
 892         }
 893
 894         owner = btrfs_header_owner(buf);
 895         BUG_ON(owner == BTRFS_TREE_RELOC_OBJECTID &&
 896                !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
 897
 898         if (refs > 1) {
 899                 if ((owner == root->root_key.objectid ||
 900                      root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) &&
 901                     !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) {
 902                         ret = btrfs_inc_ref(trans, root, buf, 1);
 903                         if (ret)
 904                                 return ret;
 905
 906                         if (root->root_key.objectid ==
 907                             BTRFS_TREE_RELOC_OBJECTID) {
 908                                 ret = btrfs_dec_ref(trans, root, buf, 0);
 909                                 if (ret)
 910                                         return ret;
 911                                 ret = btrfs_inc_ref(trans, root, cow, 1);
 912                                 if (ret)
 913                                         return ret;
 914                         }
 915                         new_flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
 916                 } else {
 917
 918                         if (root->root_key.objectid ==
 919                             BTRFS_TREE_RELOC_OBJECTID)
 920                                 ret = btrfs_inc_ref(trans, root, cow, 1);
 921                         else
 922                                 ret = btrfs_inc_ref(trans, root, cow, 0);
 923                         if (ret)
 924                                 return ret;
 925                 }
 926                 if (new_flags != 0) {
 927                         int level = btrfs_header_level(buf);
 928
 929                         ret = btrfs_set_disk_extent_flags(trans, buf,
 930                                                           new_flags, level, 0);
 931                         if (ret)
 932                                 return ret;
 933                 }
 934         } else {
 935                 if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
 936                         if (root->root_key.objectid ==
 937                             BTRFS_TREE_RELOC_OBJECTID)
 938                                 ret = btrfs_inc_ref(trans, root, cow, 1);
 939                         else
 940                                 ret = btrfs_inc_ref(trans, root, cow, 0);
 941                         if (ret)
 942                                 return ret;
 943                         ret = btrfs_dec_ref(trans, root, buf, 1);
 944                         if (ret)
 945                                 return ret;
 946                 }
 947                 btrfs_clean_tree_block(buf);
 948                 *last_ref = 1;
 949         }
 950         return 0;
 951 }
 952
 953 static struct extent_buffer *alloc_tree_block_no_bg_flush(
 954                                           struct btrfs_trans_handle *trans,
 955                                           struct btrfs_root *root,
 956                                           u64 parent_start,
 957                                           const struct btrfs_disk_key *disk_key,
 958                                           int level,
 959                                           u64 hint,
 960                                           u64 empty_size)
 961 {
 962         struct btrfs_fs_info *fs_info = root->fs_info;
 963         struct extent_buffer *ret;
 964
 965         /*
 966          * If we are COWing a node/leaf from the extent, chunk, device or free
 967          * space trees, make sure that we do not finish block group creation of
 968          * pending block groups. We do this to avoid a deadlock.
 969          * COWing can result in allocation of a new chunk, and flushing pending
 970          * block groups (btrfs_create_pending_block_groups()) can be triggered
 971          * when finishing allocation of a new chunk. Creation of a pending block
 972          * group modifies the extent, chunk, device and free space trees,
 973          * therefore we could deadlock with ourselves since we are holding a
 974          * lock on an extent buffer that btrfs_create_pending_block_groups() may
 975          * try to COW later.
 976          * For similar reasons, we also need to delay flushing pending block
 977          * groups when splitting a leaf or node, from one of those trees, since
 978          * we are holding a write lock on it and its parent or when inserting a
 979          * new root node for one of those trees.
 980          */
 981         if (root == fs_info->extent_root ||
 982             root == fs_info->chunk_root ||
 983             root == fs_info->dev_root ||
 984             root == fs_info->free_space_root)
 985                 trans->can_flush_pending_bgs = false;
 986
 987         ret = btrfs_alloc_tree_block(trans, root, parent_start,
 988                                      root->root_key.objectid, disk_key, level,
 989                                      hint, empty_size);
 990         trans->can_flush_pending_bgs = true;
 991
 992         return ret;
 993 }
 994
 995 /*
 996  * does the dirty work in cow of a single block.  The parent block (if
 997  * supplied) is updated to point to the new cow copy.  The new buffer is marked
 998  * dirty and returned locked.  If you modify the block it needs to be marked
 999  * dirty again.
1000  *
1001  * search_start -- an allocation hint for the new block
1002  *
1003  * empty_size -- a hint that you plan on doing more cow.  This is the size in
1004  * bytes the allocator should try to find free next to the block it returns.
1005  * This is just a hint and may be ignored by the allocator.
1006  */
1007 static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
1008                              struct btrfs_root *root,
1009                              struct extent_buffer *buf,
1010                              struct extent_buffer *parent, int parent_slot,
1011                              struct extent_buffer **cow_ret,
1012                              u64 search_start, u64 empty_size)
1013 {
1014         struct btrfs_fs_info *fs_info = root->fs_info;
1015         struct btrfs_disk_key disk_key;
1016         struct extent_buffer *cow;
1017         int level, ret;
1018         int last_ref = 0;
1019         int unlock_orig = 0;
1020         u64 parent_start = 0;
1021
1022         if (*cow_ret == buf)
1023                 unlock_orig = 1;
1024
1025         btrfs_assert_tree_locked(buf);
1026
1027         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
1028                 trans->transid != fs_info->running_transaction->transid);
1029         WARN_ON(test_bit(BTRFS_ROOT_SHAREABLE, &root->state) &&
1030                 trans->transid != root->last_trans);
1031
1032         level = btrfs_header_level(buf);
1033
1034         if (level == 0)
1035                 btrfs_item_key(buf, &disk_key, 0);
1036         else
1037                 btrfs_node_key(buf, &disk_key, 0);
1038
1039         if ((root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) && parent)
1040                 parent_start = parent->start;
1041
1042         cow = alloc_tree_block_no_bg_flush(trans, root, parent_start, &disk_key,
1043                                            level, search_start, empty_size);
1044         if (IS_ERR(cow))
1045                 return PTR_ERR(cow);
1046
1047         /* cow is set to blocking by btrfs_init_new_buffer */
1048
1049         copy_extent_buffer_full(cow, buf);
1050         btrfs_set_header_bytenr(cow, cow->start);
1051         btrfs_set_header_generation(cow, trans->transid);
1052         btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV);
1053         btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN |
1054                                      BTRFS_HEADER_FLAG_RELOC);
1055         if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
1056                 btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC);
1057         else
1058                 btrfs_set_header_owner(cow, root->root_key.objectid);
1059
1060         write_extent_buffer_fsid(cow, fs_info->fs_devices->metadata_uuid);
1061
1062         ret = update_ref_for_cow(trans, root, buf, cow, &last_ref);
1063         if (ret) {
1064                 btrfs_abort_transaction(trans, ret);
1065                 return ret;
1066         }
1067
1068         if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) {
1069                 ret = btrfs_reloc_cow_block(trans, root, buf, cow);
1070                 if (ret) {
1071                         btrfs_abort_transaction(trans, ret);
1072                         return ret;
1073                 }
1074         }
1075
1076         if (buf == root->node) {
1077                 WARN_ON(parent && parent != buf);
1078                 if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
1079                     btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV)
1080                         parent_start = buf->start;
1081
1082                 atomic_inc(&cow->refs);
1083                 ret = tree_mod_log_insert_root(root->node, cow, 1);
1084                 BUG_ON(ret < 0);
1085                 rcu_assign_pointer(root->node, cow);
1086
1087                 btrfs_free_tree_block(trans, root, buf, parent_start,
1088                                       last_ref);
1089                 free_extent_buffer(buf);
1090                 add_root_to_dirty_list(root);
1091         } else {
1092                 WARN_ON(trans->transid != btrfs_header_generation(parent));
1093                 tree_mod_log_insert_key(parent, parent_slot,
1094                                         MOD_LOG_KEY_REPLACE, GFP_NOFS);
1095                 btrfs_set_node_blockptr(parent, parent_slot,
1096                                         cow->start);
1097                 btrfs_set_node_ptr_generation(parent, parent_slot,
1098                                               trans->transid);
1099                 btrfs_mark_buffer_dirty(parent);
1100                 if (last_ref) {
1101                         ret = tree_mod_log_free_eb(buf);
1102                         if (ret) {
1103                                 btrfs_abort_transaction(trans, ret);
1104                                 return ret;
1105                         }
1106                 }
1107                 btrfs_free_tree_block(trans, root, buf, parent_start,
1108                                       last_ref);
1109         }
1110         if (unlock_orig)
1111                 btrfs_tree_unlock(buf);
1112         free_extent_buffer_stale(buf);
1113         btrfs_mark_buffer_dirty(cow);
1114         *cow_ret = cow;
1115         return 0;
1116 }
1117
1118 /*
1119  * returns the logical address of the oldest predecessor of the given root.
1120  * entries older than time_seq are ignored.
1121  */
1122 static struct tree_mod_elem *__tree_mod_log_oldest_root(
1123                 struct extent_buffer *eb_root, u64 time_seq)
1124 {
1125         struct tree_mod_elem *tm;
1126         struct tree_mod_elem *found = NULL;
1127         u64 root_logical = eb_root->start;
1128         int looped = 0;
1129
1130         if (!time_seq)
1131                 return NULL;
1132
1133         /*
1134          * the very last operation that's logged for a root is the
1135          * replacement operation (if it is replaced at all). this has
1136          * the logical address of the *new* root, making it the very
1137          * first operation that's logged for this root.
1138          */
1139         while (1) {
1140                 tm = tree_mod_log_search_oldest(eb_root->fs_info, root_logical,
1141                                                 time_seq);
1142                 if (!looped && !tm)
1143                         return NULL;
1144                 /*
1145                  * if there are no tree operation for the oldest root, we simply
1146                  * return it. this should only happen if that (old) root is at
1147                  * level 0.
1148                  */
1149                 if (!tm)
1150                         break;
1151
1152                 /*
1153                  * if there's an operation that's not a root replacement, we
1154                  * found the oldest version of our root. normally, we'll find a
1155                  * MOD_LOG_KEY_REMOVE_WHILE_FREEING operation here.
1156                  */
1157                 if (tm->op != MOD_LOG_ROOT_REPLACE)
1158                         break;
1159
1160                 found = tm;
1161                 root_logical = tm->old_root.logical;
1162                 looped = 1;
1163         }
1164
1165         /* if there's no old root to return, return what we found instead */
1166         if (!found)
1167                 found = tm;
1168
1169         return found;
1170 }
1171
1172 /*
1173  * tm is a pointer to the first operation to rewind within eb. then, all
1174  * previous operations will be rewound (until we reach something older than
1175  * time_seq).
1176  */
1177 static void
1178 __tree_mod_log_rewind(struct btrfs_fs_info *fs_info, struct extent_buffer *eb,
1179                       u64 time_seq, struct tree_mod_elem *first_tm)
1180 {
1181         u32 n;
1182         struct rb_node *next;
1183         struct tree_mod_elem *tm = first_tm;
1184         unsigned long o_dst;
1185         unsigned long o_src;
1186         unsigned long p_size = sizeof(struct btrfs_key_ptr);
1187
1188         n = btrfs_header_nritems(eb);
1189         read_lock(&fs_info->tree_mod_log_lock);
1190         while (tm && tm->seq >= time_seq) {
1191                 /*
1192                  * all the operations are recorded with the operator used for
1193                  * the modification. as we're going backwards, we do the
1194                  * opposite of each operation here.
1195                  */
1196                 switch (tm->op) {
1197                 case MOD_LOG_KEY_REMOVE_WHILE_FREEING:
1198                         BUG_ON(tm->slot < n);
1199                         fallthrough;
1200                 case MOD_LOG_KEY_REMOVE_WHILE_MOVING:
1201                 case MOD_LOG_KEY_REMOVE:
1202                         btrfs_set_node_key(eb, &tm->key, tm->slot);
1203                         btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr);
1204                         btrfs_set_node_ptr_generation(eb, tm->slot,
1205                                                       tm->generation);
1206                         n++;
1207                         break;
1208                 case MOD_LOG_KEY_REPLACE:
1209                         BUG_ON(tm->slot >= n);
1210                         btrfs_set_node_key(eb, &tm->key, tm->slot);
1211                         btrfs_set_node_blockptr(eb, tm->slot, tm->blockptr);
1212                         btrfs_set_node_ptr_generation(eb, tm->slot,
1213                                                       tm->generation);
1214                         break;
1215                 case MOD_LOG_KEY_ADD:
1216                         /* if a move operation is needed it's in the log */
1217                         n--;
1218                         break;
1219                 case MOD_LOG_MOVE_KEYS:
1220                         o_dst = btrfs_node_key_ptr_offset(tm->slot);
1221                         o_src = btrfs_node_key_ptr_offset(tm->move.dst_slot);
1222                         memmove_extent_buffer(eb, o_dst, o_src,
1223                                               tm->move.nr_items * p_size);
1224                         break;
1225                 case MOD_LOG_ROOT_REPLACE:
1226                         /*
1227                          * this operation is special. for roots, this must be
1228                          * handled explicitly before rewinding.
1229                          * for non-roots, this operation may exist if the node
1230                          * was a root: root A -> child B; then A gets empty and
1231                          * B is promoted to the new root. in the mod log, we'll
1232                          * have a root-replace operation for B, a tree block
1233                          * that is no root. we simply ignore that operation.
1234                          */
1235                         break;
1236                 }
1237                 next = rb_next(&tm->node);
1238                 if (!next)
1239                         break;
1240                 tm = rb_entry(next, struct tree_mod_elem, node);
1241                 if (tm->logical != first_tm->logical)
1242                         break;
1243         }
1244         read_unlock(&fs_info->tree_mod_log_lock);
1245         btrfs_set_header_nritems(eb, n);
1246 }
1247
1248 /*
1249  * Called with eb read locked. If the buffer cannot be rewound, the same buffer
1250  * is returned. If rewind operations happen, a fresh buffer is returned. The
1251  * returned buffer is always read-locked. If the returned buffer is not the
1252  * input buffer, the lock on the input buffer is released and the input buffer
1253  * is freed (its refcount is decremented).
1254  */
1255 static struct extent_buffer *
1256 tree_mod_log_rewind(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
1257                     struct extent_buffer *eb, u64 time_seq)
1258 {
1259         struct extent_buffer *eb_rewin;
1260         struct tree_mod_elem *tm;
1261
1262         if (!time_seq)
1263                 return eb;
1264
1265         if (btrfs_header_level(eb) == 0)
1266                 return eb;
1267
1268         tm = tree_mod_log_search(fs_info, eb->start, time_seq);
1269         if (!tm)
1270                 return eb;
1271
1272         btrfs_set_path_blocking(path);
1273         btrfs_set_lock_blocking_read(eb);
1274
1275         if (tm->op == MOD_LOG_KEY_REMOVE_WHILE_FREEING) {
1276                 BUG_ON(tm->slot != 0);
1277                 eb_rewin = alloc_dummy_extent_buffer(fs_info, eb->start);
1278                 if (!eb_rewin) {
1279                         btrfs_tree_read_unlock_blocking(eb);
1280                         free_extent_buffer(eb);
1281                         return NULL;
1282                 }
1283                 btrfs_set_header_bytenr(eb_rewin, eb->start);
1284                 btrfs_set_header_backref_rev(eb_rewin,
1285                                              btrfs_header_backref_rev(eb));
1286                 btrfs_set_header_owner(eb_rewin, btrfs_header_owner(eb));
1287                 btrfs_set_header_level(eb_rewin, btrfs_header_level(eb));
1288         } else {
1289                 eb_rewin = btrfs_clone_extent_buffer(eb);
1290                 if (!eb_rewin) {
1291                         btrfs_tree_read_unlock_blocking(eb);
1292                         free_extent_buffer(eb);
1293                         return NULL;
1294                 }
1295         }
1296
1297         btrfs_tree_read_unlock_blocking(eb);
1298         free_extent_buffer(eb);
1299
1300         btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb_rewin),
1301                                        eb_rewin, btrfs_header_level(eb_rewin));
1302         btrfs_tree_read_lock(eb_rewin);
1303         __tree_mod_log_rewind(fs_info, eb_rewin, time_seq, tm);
1304         WARN_ON(btrfs_header_nritems(eb_rewin) >
1305                 BTRFS_NODEPTRS_PER_BLOCK(fs_info));
1306
1307         return eb_rewin;
1308 }
1309
1310 /*
1311  * get_old_root() rewinds the state of @root's root node to the given @time_seq
1312  * value. If there are no changes, the current root->root_node is returned. If
1313  * anything changed in between, there's a fresh buffer allocated on which the
1314  * rewind operations are done. In any case, the returned buffer is read locked.
1315  * Returns NULL on error (with no locks held).
1316  */
1317 static inline struct extent_buffer *
1318 get_old_root(struct btrfs_root *root, u64 time_seq)
1319 {
1320         struct btrfs_fs_info *fs_info = root->fs_info;
1321         struct tree_mod_elem *tm;
1322         struct extent_buffer *eb = NULL;
1323         struct extent_buffer *eb_root;
1324         u64 eb_root_owner = 0;
1325         struct extent_buffer *old;
1326         struct tree_mod_root *old_root = NULL;
1327         u64 old_generation = 0;
1328         u64 logical;
1329         int level;
1330
1331         eb_root = btrfs_read_lock_root_node(root);
1332         tm = __tree_mod_log_oldest_root(eb_root, time_seq);
1333         if (!tm)
1334                 return eb_root;
1335
1336         if (tm->op == MOD_LOG_ROOT_REPLACE) {
1337                 old_root = &tm->old_root;
1338                 old_generation = tm->generation;
1339                 logical = old_root->logical;
1340                 level = old_root->level;
1341         } else {
1342                 logical = eb_root->start;
1343                 level = btrfs_header_level(eb_root);
1344         }
1345
1346         tm = tree_mod_log_search(fs_info, logical, time_seq);
1347         if (old_root && tm && tm->op != MOD_LOG_KEY_REMOVE_WHILE_FREEING) {
1348                 btrfs_tree_read_unlock(eb_root);
1349                 free_extent_buffer(eb_root);
1350                 old = read_tree_block(fs_info, logical, 0, level, NULL);
1351                 if (WARN_ON(IS_ERR(old) || !extent_buffer_uptodate(old))) {
1352                         if (!IS_ERR(old))
1353                                 free_extent_buffer(old);
1354                         btrfs_warn(fs_info,
1355                                    "failed to read tree block %llu from get_old_root",
1356                                    logical);
1357                 } else {
1358                         eb = btrfs_clone_extent_buffer(old);
1359                         free_extent_buffer(old);
1360                 }
1361         } else if (old_root) {
1362                 eb_root_owner = btrfs_header_owner(eb_root);
1363                 btrfs_tree_read_unlock(eb_root);
1364                 free_extent_buffer(eb_root);
1365                 eb = alloc_dummy_extent_buffer(fs_info, logical);
1366         } else {
1367                 btrfs_set_lock_blocking_read(eb_root);
1368                 eb = btrfs_clone_extent_buffer(eb_root);
1369                 btrfs_tree_read_unlock_blocking(eb_root);
1370                 free_extent_buffer(eb_root);
1371         }
1372
1373         if (!eb)
1374                 return NULL;
1375         if (old_root) {
1376                 btrfs_set_header_bytenr(eb, eb->start);
1377                 btrfs_set_header_backref_rev(eb, BTRFS_MIXED_BACKREF_REV);
1378                 btrfs_set_header_owner(eb, eb_root_owner);
1379                 btrfs_set_header_level(eb, old_root->level);
1380                 btrfs_set_header_generation(eb, old_generation);
1381         }
1382         btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb), eb,
1383                                        btrfs_header_level(eb));
1384         btrfs_tree_read_lock(eb);
1385         if (tm)
1386                 __tree_mod_log_rewind(fs_info, eb, time_seq, tm);
1387         else
1388                 WARN_ON(btrfs_header_level(eb) != 0);
1389         WARN_ON(btrfs_header_nritems(eb) > BTRFS_NODEPTRS_PER_BLOCK(fs_info));
1390
1391         return eb;
1392 }
1393
1394 int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq)
1395 {
1396         struct tree_mod_elem *tm;
1397         int level;
1398         struct extent_buffer *eb_root = btrfs_root_node(root);
1399
1400         tm = __tree_mod_log_oldest_root(eb_root, time_seq);
1401         if (tm && tm->op == MOD_LOG_ROOT_REPLACE) {
1402                 level = tm->old_root.level;
1403         } else {
1404                 level = btrfs_header_level(eb_root);
1405         }
1406         free_extent_buffer(eb_root);
1407
1408         return level;
1409 }
1410
1411 static inline int should_cow_block(struct btrfs_trans_handle *trans,
1412                                    struct btrfs_root *root,
1413                                    struct extent_buffer *buf)
1414 {
1415         if (btrfs_is_testing(root->fs_info))
1416                 return 0;
1417
1418         /* Ensure we can see the FORCE_COW bit */
1419         smp_mb__before_atomic();
1420
1421         /*
1422          * We do not need to cow a block if
1423          * 1) this block is not created or changed in this transaction;
1424          * 2) this block does not belong to TREE_RELOC tree;
1425          * 3) the root is not forced COW.
1426          *
1427          * What is forced COW:
1428          *    when we create snapshot during committing the transaction,
1429          *    after we've finished copying src root, we must COW the shared
1430          *    block to ensure the metadata consistency.
1431          */
1432         if (btrfs_header_generation(buf) == trans->transid &&
1433             !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) &&
1434             !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
1435               btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)) &&
1436             !test_bit(BTRFS_ROOT_FORCE_COW, &root->state))
1437                 return 0;
1438         return 1;
1439 }
1440
1441 /*
1442  * cows a single block, see __btrfs_cow_block for the real work.
1443  * This version of it has extra checks so that a block isn't COWed more than
1444  * once per transaction, as long as it hasn't been written yet
1445  */
1446 noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
1447                     struct btrfs_root *root, struct extent_buffer *buf,
1448                     struct extent_buffer *parent, int parent_slot,
1449                     struct extent_buffer **cow_ret)
1450 {
1451         struct btrfs_fs_info *fs_info = root->fs_info;
1452         u64 search_start;
1453         int ret;
1454
1455         if (test_bit(BTRFS_ROOT_DELETING, &root->state))
1456                 btrfs_err(fs_info,
1457                         "COW'ing blocks on a fs root that's being dropped");
1458
1459         if (trans->transaction != fs_info->running_transaction)
1460                 WARN(1, KERN_CRIT "trans %llu running %llu\n",
1461                        trans->transid,
1462                        fs_info->running_transaction->transid);
1463
1464         if (trans->transid != fs_info->generation)
1465                 WARN(1, KERN_CRIT "trans %llu running %llu\n",
1466                        trans->transid, fs_info->generation);
1467
1468         if (!should_cow_block(trans, root, buf)) {
1469                 trans->dirty = true;
1470                 *cow_ret = buf;
1471                 return 0;
1472         }
1473
1474         search_start = buf->start & ~((u64)SZ_1G - 1);
1475
1476         if (parent)
1477                 btrfs_set_lock_blocking_write(parent);
1478         btrfs_set_lock_blocking_write(buf);
1479
1480         /*
1481          * Before CoWing this block for later modification, check if it's
1482          * the subtree root and do the delayed subtree trace if needed.
1483          *
1484          * Also We don't care about the error, as it's handled internally.
1485          */
1486         btrfs_qgroup_trace_subtree_after_cow(trans, root, buf);
1487         ret = __btrfs_cow_block(trans, root, buf, parent,
1488                                  parent_slot, cow_ret, search_start, 0);
1489
1490         trace_btrfs_cow_block(root, buf, *cow_ret);
1491
1492         return ret;
1493 }
1494
1495 /*
1496  * helper function for defrag to decide if two blocks pointed to by a
1497  * node are actually close by
1498  */
1499 static int close_blocks(u64 blocknr, u64 other, u32 blocksize)
1500 {
1501         if (blocknr < other && other - (blocknr + blocksize) < 32768)
1502                 return 1;
1503         if (blocknr > other && blocknr - (other + blocksize) < 32768)
1504                 return 1;
1505         return 0;
1506 }
1507
1508 #ifdef __LITTLE_ENDIAN
1509
1510 /*
1511  * Compare two keys, on little-endian the disk order is same as CPU order and
1512  * we can avoid the conversion.
1513  */
1514 static int comp_keys(const struct btrfs_disk_key *disk_key,
1515                      const struct btrfs_key *k2)
1516 {
1517         const struct btrfs_key *k1 = (const struct btrfs_key *)disk_key;
1518
1519         return btrfs_comp_cpu_keys(k1, k2);
1520 }
1521
1522 #else
1523
1524 /*
1525  * compare two keys in a memcmp fashion
1526  */
1527 static int comp_keys(const struct btrfs_disk_key *disk,
1528                      const struct btrfs_key *k2)
1529 {
1530         struct btrfs_key k1;
1531
1532         btrfs_disk_key_to_cpu(&k1, disk);
1533
1534         return btrfs_comp_cpu_keys(&k1, k2);
1535 }
1536 #endif
1537
1538 /*
1539  * same as comp_keys only with two btrfs_key's
1540  */
1541 int __pure btrfs_comp_cpu_keys(const struct btrfs_key *k1, const struct btrfs_key *k2)
1542 {
1543         if (k1->objectid > k2->objectid)
1544                 return 1;
1545         if (k1->objectid < k2->objectid)
1546                 return -1;
1547         if (k1->type > k2->type)
1548                 return 1;
1549         if (k1->type < k2->type)
1550                 return -1;
1551         if (k1->offset > k2->offset)
1552                 return 1;
1553         if (k1->offset < k2->offset)
1554                 return -1;
1555         return 0;
1556 }
1557
1558 /*
1559  * this is used by the defrag code to go through all the
1560  * leaves pointed to by a node and reallocate them so that
1561  * disk order is close to key order
1562  */
1563 int btrfs_realloc_node(struct btrfs_trans_handle *trans,
1564                        struct btrfs_root *root, struct extent_buffer *parent,
1565                        int start_slot, u64 *last_ret,
1566                        struct btrfs_key *progress)
1567 {
1568         struct btrfs_fs_info *fs_info = root->fs_info;
1569         struct extent_buffer *cur;
1570         u64 blocknr;
1571         u64 gen;
1572         u64 search_start = *last_ret;
1573         u64 last_block = 0;
1574         u64 other;
1575         u32 parent_nritems;
1576         int end_slot;
1577         int i;
1578         int err = 0;
1579         int parent_level;
1580         int uptodate;
1581         u32 blocksize;
1582         int progress_passed = 0;
1583         struct btrfs_disk_key disk_key;
1584
1585         parent_level = btrfs_header_level(parent);
1586
1587         WARN_ON(trans->transaction != fs_info->running_transaction);
1588         WARN_ON(trans->transid != fs_info->generation);
1589
1590         parent_nritems = btrfs_header_nritems(parent);
1591         blocksize = fs_info->nodesize;
1592         end_slot = parent_nritems - 1;
1593
1594         if (parent_nritems <= 1)
1595                 return 0;
1596
1597         btrfs_set_lock_blocking_write(parent);
1598
1599         for (i = start_slot; i <= end_slot; i++) {
1600                 struct btrfs_key first_key;
1601                 int close = 1;
1602
1603                 btrfs_node_key(parent, &disk_key, i);
1604                 if (!progress_passed && comp_keys(&disk_key, progress) < 0)
1605                         continue;
1606
1607                 progress_passed = 1;
1608                 blocknr = btrfs_node_blockptr(parent, i);
1609                 gen = btrfs_node_ptr_generation(parent, i);
1610                 btrfs_node_key_to_cpu(parent, &first_key, i);
1611                 if (last_block == 0)
1612                         last_block = blocknr;
1613
1614                 if (i > 0) {
1615                         other = btrfs_node_blockptr(parent, i - 1);
1616                         close = close_blocks(blocknr, other, blocksize);
1617                 }
1618                 if (!close && i < end_slot) {
1619                         other = btrfs_node_blockptr(parent, i + 1);
1620                         close = close_blocks(blocknr, other, blocksize);
1621                 }
1622                 if (close) {
1623                         last_block = blocknr;
1624                         continue;
1625                 }
1626
1627                 cur = find_extent_buffer(fs_info, blocknr);
1628                 if (cur)
1629                         uptodate = btrfs_buffer_uptodate(cur, gen, 0);
1630                 else
1631                         uptodate = 0;
1632                 if (!cur || !uptodate) {
1633                         if (!cur) {
1634                                 cur = read_tree_block(fs_info, blocknr, gen,
1635                                                       parent_level - 1,
1636                                                       &first_key);
1637                                 if (IS_ERR(cur)) {
1638                                         return PTR_ERR(cur);
1639                                 } else if (!extent_buffer_uptodate(cur)) {
1640                                         free_extent_buffer(cur);
1641                                         return -EIO;
1642                                 }
1643                         } else if (!uptodate) {
1644                                 err = btrfs_read_buffer(cur, gen,
1645                                                 parent_level - 1,&first_key);
1646                                 if (err) {
1647                                         free_extent_buffer(cur);
1648                                         return err;
1649                                 }
1650                         }
1651                 }
1652                 if (search_start == 0)
1653                         search_start = last_block;
1654
1655                 btrfs_tree_lock(cur);
1656                 btrfs_set_lock_blocking_write(cur);
1657                 err = __btrfs_cow_block(trans, root, cur, parent, i,
1658                                         &cur, search_start,
1659                                         min(16 * blocksize,
1660                                             (end_slot - i) * blocksize));
1661                 if (err) {
1662                         btrfs_tree_unlock(cur);
1663                         free_extent_buffer(cur);
1664                         break;
1665                 }
1666                 search_start = cur->start;
1667                 last_block = cur->start;
1668                 *last_ret = search_start;
1669                 btrfs_tree_unlock(cur);
1670                 free_extent_buffer(cur);
1671         }
1672         return err;
1673 }
1674
1675 /*
1676  * search for key in the extent_buffer.  The items start at offset p,
1677  * and they are item_size apart.  There are 'max' items in p.
1678  *
1679  * the slot in the array is returned via slot, and it points to
1680  * the place where you would insert key if it is not found in
1681  * the array.
1682  *
1683  * slot may point to max if the key is bigger than all of the keys
1684  */
1685 static noinline int generic_bin_search(struct extent_buffer *eb,
1686                                        unsigned long p, int item_size,
1687                                        const struct btrfs_key *key,
1688                                        int max, int *slot)
1689 {
1690         int low = 0;
1691         int high = max;
1692         int ret;
1693         const int key_size = sizeof(struct btrfs_disk_key);
1694
1695         if (low > high) {
1696                 btrfs_err(eb->fs_info,
1697                  "%s: low (%d) > high (%d) eb %llu owner %llu level %d",
1698                           __func__, low, high, eb->start,
1699                           btrfs_header_owner(eb), btrfs_header_level(eb));
1700                 return -EINVAL;
1701         }
1702
1703         while (low < high) {
1704                 unsigned long oip;
1705                 unsigned long offset;
1706                 struct btrfs_disk_key *tmp;
1707                 struct btrfs_disk_key unaligned;
1708                 int mid;
1709
1710                 mid = (low + high) / 2;
1711                 offset = p + mid * item_size;
1712                 oip = offset_in_page(offset);
1713
1714                 if (oip + key_size <= PAGE_SIZE) {
1715                         const unsigned long idx = offset >> PAGE_SHIFT;
1716                         char *kaddr = page_address(eb->pages[idx]);
1717
1718                         tmp = (struct btrfs_disk_key *)(kaddr + oip);
1719                 } else {
1720                         read_extent_buffer(eb, &unaligned, offset, key_size);
1721                         tmp = &unaligned;
1722                 }
1723
1724                 ret = comp_keys(tmp, key);
1725
1726                 if (ret < 0)
1727                         low = mid + 1;
1728                 else if (ret > 0)
1729                         high = mid;
1730                 else {
1731                         *slot = mid;
1732                         return 0;
1733                 }
1734         }
1735         *slot = low;
1736         return 1;
1737 }
1738
1739 /*
1740  * simple bin_search frontend that does the right thing for
1741  * leaves vs nodes
1742  */
1743 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
1744                      int *slot)
1745 {
1746         if (btrfs_header_level(eb) == 0)
1747                 return generic_bin_search(eb,
1748                                           offsetof(struct btrfs_leaf, items),
1749                                           sizeof(struct btrfs_item),
1750                                           key, btrfs_header_nritems(eb),
1751                                           slot);
1752         else
1753                 return generic_bin_search(eb,
1754                                           offsetof(struct btrfs_node, ptrs),
1755                                           sizeof(struct btrfs_key_ptr),
1756                                           key, btrfs_header_nritems(eb),
1757                                           slot);
1758 }
1759
1760 static void root_add_used(struct btrfs_root *root, u32 size)
1761 {
1762         spin_lock(&root->accounting_lock);
1763         btrfs_set_root_used(&root->root_item,
1764                             btrfs_root_used(&root->root_item) + size);
1765         spin_unlock(&root->accounting_lock);
1766 }
1767
1768 static void root_sub_used(struct btrfs_root *root, u32 size)
1769 {
1770         spin_lock(&root->accounting_lock);
1771         btrfs_set_root_used(&root->root_item,
1772                             btrfs_root_used(&root->root_item) - size);
1773         spin_unlock(&root->accounting_lock);
1774 }
1775
1776 /* given a node and slot number, this reads the blocks it points to.  The
1777  * extent buffer is returned with a reference taken (but unlocked).
1778  */
1779 struct extent_buffer *btrfs_read_node_slot(struct extent_buffer *parent,
1780                                            int slot)
1781 {
1782         int level = btrfs_header_level(parent);
1783         struct extent_buffer *eb;
1784         struct btrfs_key first_key;
1785
1786         if (slot < 0 || slot >= btrfs_header_nritems(parent))
1787                 return ERR_PTR(-ENOENT);
1788
1789         BUG_ON(level == 0);
1790
1791         btrfs_node_key_to_cpu(parent, &first_key, slot);
1792         eb = read_tree_block(parent->fs_info, btrfs_node_blockptr(parent, slot),
1793                              btrfs_node_ptr_generation(parent, slot),
1794                              level - 1, &first_key);
1795         if (!IS_ERR(eb) && !extent_buffer_uptodate(eb)) {
1796                 free_extent_buffer(eb);
1797                 eb = ERR_PTR(-EIO);
1798         }
1799
1800         return eb;
1801 }
1802
1803 /*
1804  * node level balancing, used to make sure nodes are in proper order for
1805  * item deletion.  We balance from the top down, so we have to make sure
1806  * that a deletion won't leave an node completely empty later on.
1807  */
1808 static noinline int balance_level(struct btrfs_trans_handle *trans,
1809                          struct btrfs_root *root,
1810                          struct btrfs_path *path, int level)
1811 {
1812         struct btrfs_fs_info *fs_info = root->fs_info;
1813         struct extent_buffer *right = NULL;
1814         struct extent_buffer *mid;
1815         struct extent_buffer *left = NULL;
1816         struct extent_buffer *parent = NULL;
1817         int ret = 0;
1818         int wret;
1819         int pslot;
1820         int orig_slot = path->slots[level];
1821         u64 orig_ptr;
1822
1823         ASSERT(level > 0);
1824
1825         mid = path->nodes[level];
1826
1827         WARN_ON(path->locks[level] != BTRFS_WRITE_LOCK &&
1828                 path->locks[level] != BTRFS_WRITE_LOCK_BLOCKING);
1829         WARN_ON(btrfs_header_generation(mid) != trans->transid);
1830
1831         orig_ptr = btrfs_node_blockptr(mid, orig_slot);
1832
1833         if (level < BTRFS_MAX_LEVEL - 1) {
1834                 parent = path->nodes[level + 1];
1835                 pslot = path->slots[level + 1];
1836         }
1837
1838         /*
1839          * deal with the case where there is only one pointer in the root
1840          * by promoting the node below to a root
1841          */
1842         if (!parent) {
1843                 struct extent_buffer *child;
1844
1845                 if (btrfs_header_nritems(mid) != 1)
1846                         return 0;
1847
1848                 /* promote the child to a root */
1849                 child = btrfs_read_node_slot(mid, 0);
1850                 if (IS_ERR(child)) {
1851                         ret = PTR_ERR(child);
1852                         btrfs_handle_fs_error(fs_info, ret, NULL);
1853                         goto enospc;
1854                 }
1855
1856                 btrfs_tree_lock(child);
1857                 btrfs_set_lock_blocking_write(child);
1858                 ret = btrfs_cow_block(trans, root, child, mid, 0, &child);
1859                 if (ret) {
1860                         btrfs_tree_unlock(child);
1861                         free_extent_buffer(child);
1862                         goto enospc;
1863                 }
1864
1865                 ret = tree_mod_log_insert_root(root->node, child, 1);
1866                 BUG_ON(ret < 0);
1867                 rcu_assign_pointer(root->node, child);
1868
1869                 add_root_to_dirty_list(root);
1870                 btrfs_tree_unlock(child);
1871
1872                 path->locks[level] = 0;
1873                 path->nodes[level] = NULL;
1874                 btrfs_clean_tree_block(mid);
1875                 btrfs_tree_unlock(mid);
1876                 /* once for the path */
1877                 free_extent_buffer(mid);
1878
1879                 root_sub_used(root, mid->len);
1880                 btrfs_free_tree_block(trans, root, mid, 0, 1);
1881                 /* once for the root ptr */
1882                 free_extent_buffer_stale(mid);
1883                 return 0;
1884         }
1885         if (btrfs_header_nritems(mid) >
1886             BTRFS_NODEPTRS_PER_BLOCK(fs_info) / 4)
1887                 return 0;
1888
1889         left = btrfs_read_node_slot(parent, pslot - 1);
1890         if (IS_ERR(left))
1891                 left = NULL;
1892
1893         if (left) {
1894                 btrfs_tree_lock(left);
1895                 btrfs_set_lock_blocking_write(left);
1896                 wret = btrfs_cow_block(trans, root, left,
1897                                        parent, pslot - 1, &left);
1898                 if (wret) {
1899                         ret = wret;
1900                         goto enospc;
1901                 }
1902         }
1903
1904         right = btrfs_read_node_slot(parent, pslot + 1);
1905         if (IS_ERR(right))
1906                 right = NULL;
1907
1908         if (right) {
1909                 btrfs_tree_lock(right);
1910                 btrfs_set_lock_blocking_write(right);
1911                 wret = btrfs_cow_block(trans, root, right,
1912                                        parent, pslot + 1, &right);
1913                 if (wret) {
1914                         ret = wret;
1915                         goto enospc;
1916                 }
1917         }
1918
1919         /* first, try to make some room in the middle buffer */
1920         if (left) {
1921                 orig_slot += btrfs_header_nritems(left);
1922                 wret = push_node_left(trans, left, mid, 1);
1923                 if (wret < 0)
1924                         ret = wret;
1925         }
1926
1927         /*
1928          * then try to empty the right most buffer into the middle
1929          */
1930         if (right) {
1931                 wret = push_node_left(trans, mid, right, 1);
1932                 if (wret < 0 && wret != -ENOSPC)
1933                         ret = wret;
1934                 if (btrfs_header_nritems(right) == 0) {
1935                         btrfs_clean_tree_block(right);
1936                         btrfs_tree_unlock(right);
1937                         del_ptr(root, path, level + 1, pslot + 1);
1938                         root_sub_used(root, right->len);
1939                         btrfs_free_tree_block(trans, root, right, 0, 1);
1940                         free_extent_buffer_stale(right);
1941                         right = NULL;
1942                 } else {
1943                         struct btrfs_disk_key right_key;
1944                         btrfs_node_key(right, &right_key, 0);
1945                         ret = tree_mod_log_insert_key(parent, pslot + 1,
1946                                         MOD_LOG_KEY_REPLACE, GFP_NOFS);
1947                         BUG_ON(ret < 0);
1948                         btrfs_set_node_key(parent, &right_key, pslot + 1);
1949                         btrfs_mark_buffer_dirty(parent);
1950                 }
1951         }
1952         if (btrfs_header_nritems(mid) == 1) {
1953                 /*
1954                  * we're not allowed to leave a node with one item in the
1955                  * tree during a delete.  A deletion from lower in the tree
1956                  * could try to delete the only pointer in this node.
1957                  * So, pull some keys from the left.
1958                  * There has to be a left pointer at this point because
1959                  * otherwise we would have pulled some pointers from the
1960                  * right
1961                  */
1962                 if (!left) {
1963                         ret = -EROFS;
1964                         btrfs_handle_fs_error(fs_info, ret, NULL);
1965                         goto enospc;
1966                 }
1967                 wret = balance_node_right(trans, mid, left);
1968                 if (wret < 0) {
1969                         ret = wret;
1970                         goto enospc;
1971                 }
1972                 if (wret == 1) {
1973                         wret = push_node_left(trans, left, mid, 1);
1974                         if (wret < 0)
1975                                 ret = wret;
1976                 }
1977                 BUG_ON(wret == 1);
1978         }
1979         if (btrfs_header_nritems(mid) == 0) {
1980                 btrfs_clean_tree_block(mid);
1981                 btrfs_tree_unlock(mid);
1982                 del_ptr(root, path, level + 1, pslot);
1983                 root_sub_used(root, mid->len);
1984                 btrfs_free_tree_block(trans, root, mid, 0, 1);
1985                 free_extent_buffer_stale(mid);
1986                 mid = NULL;
1987         } else {
1988                 /* update the parent key to reflect our changes */
1989                 struct btrfs_disk_key mid_key;
1990                 btrfs_node_key(mid, &mid_key, 0);
1991                 ret = tree_mod_log_insert_key(parent, pslot,
1992                                 MOD_LOG_KEY_REPLACE, GFP_NOFS);
1993                 BUG_ON(ret < 0);
1994                 btrfs_set_node_key(parent, &mid_key, pslot);
1995                 btrfs_mark_buffer_dirty(parent);
1996         }
1997
1998         /* update the path */
1999         if (left) {
2000                 if (btrfs_header_nritems(left) > orig_slot) {
2001                         atomic_inc(&left->refs);
2002                         /* left was locked after cow */
2003                         path->nodes[level] = left;
2004                         path->slots[level + 1] -= 1;
2005                         path->slots[level] = orig_slot;
2006                         if (mid) {
2007                                 btrfs_tree_unlock(mid);
2008                                 free_extent_buffer(mid);
2009                         }
2010                 } else {
2011                         orig_slot -= btrfs_header_nritems(left);
2012                         path->slots[level] = orig_slot;
2013                 }
2014         }
2015         /* double check we haven't messed things up */
2016         if (orig_ptr !=
2017             btrfs_node_blockptr(path->nodes[level], path->slots[level]))
2018                 BUG();
2019 enospc:
2020         if (right) {
2021                 btrfs_tree_unlock(right);
2022                 free_extent_buffer(right);
2023         }
2024         if (left) {
2025                 if (path->nodes[level] != left)
2026                         btrfs_tree_unlock(left);
2027                 free_extent_buffer(left);
2028         }
2029         return ret;
2030 }
2031
2032 /* Node balancing for insertion.  Here we only split or push nodes around
2033  * when they are completely full.  This is also done top down, so we
2034  * have to be pessimistic.
2035  */
2036 static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
2037                                           struct btrfs_root *root,
2038                                           struct btrfs_path *path, int level)
2039 {
2040         struct btrfs_fs_info *fs_info = root->fs_info;
2041         struct extent_buffer *right = NULL;
2042         struct extent_buffer *mid;
2043         struct extent_buffer *left = NULL;
2044         struct extent_buffer *parent = NULL;
2045         int ret = 0;
2046         int wret;
2047         int pslot;
2048         int orig_slot = path->slots[level];
2049
2050         if (level == 0)
2051                 return 1;
2052
2053         mid = path->nodes[level];
2054         WARN_ON(btrfs_header_generation(mid) != trans->transid);
2055
2056         if (level < BTRFS_MAX_LEVEL - 1) {
2057                 parent = path->nodes[level + 1];
2058                 pslot = path->slots[level + 1];
2059         }
2060
2061         if (!parent)
2062                 return 1;
2063
2064         left = btrfs_read_node_slot(parent, pslot - 1);
2065         if (IS_ERR(left))
2066                 left = NULL;
2067
2068         /* first, try to make some room in the middle buffer */
2069         if (left) {
2070                 u32 left_nr;
2071
2072                 btrfs_tree_lock(left);
2073                 btrfs_set_lock_blocking_write(left);
2074
2075                 left_nr = btrfs_header_nritems(left);
2076                 if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
2077                         wret = 1;
2078                 } else {
2079                         ret = btrfs_cow_block(trans, root, left, parent,
2080                                               pslot - 1, &left);
2081                         if (ret)
2082                                 wret = 1;
2083                         else {
2084                                 wret = push_node_left(trans, left, mid, 0);
2085                         }
2086                 }
2087                 if (wret < 0)
2088                         ret = wret;
2089                 if (wret == 0) {
2090                         struct btrfs_disk_key disk_key;
2091                         orig_slot += left_nr;
2092                         btrfs_node_key(mid, &disk_key, 0);
2093                         ret = tree_mod_log_insert_key(parent, pslot,
2094                                         MOD_LOG_KEY_REPLACE, GFP_NOFS);
2095                         BUG_ON(ret < 0);
2096                         btrfs_set_node_key(parent, &disk_key, pslot);
2097                         btrfs_mark_buffer_dirty(parent);
2098                         if (btrfs_header_nritems(left) > orig_slot) {
2099                                 path->nodes[level] = left;
2100                                 path->slots[level + 1] -= 1;
2101                                 path->slots[level] = orig_slot;
2102                                 btrfs_tree_unlock(mid);
2103                                 free_extent_buffer(mid);
2104                         } else {
2105                                 orig_slot -=
2106                                         btrfs_header_nritems(left);
2107                                 path->slots[level] = orig_slot;
2108                                 btrfs_tree_unlock(left);
2109                                 free_extent_buffer(left);
2110                         }
2111                         return 0;
2112                 }
2113                 btrfs_tree_unlock(left);
2114                 free_extent_buffer(left);
2115         }
2116         right = btrfs_read_node_slot(parent, pslot + 1);
2117         if (IS_ERR(right))
2118                 right = NULL;
2119
2120         /*
2121          * then try to empty the right most buffer into the middle
2122          */
2123         if (right) {
2124                 u32 right_nr;
2125
2126                 btrfs_tree_lock(right);
2127                 btrfs_set_lock_blocking_write(right);
2128
2129                 right_nr = btrfs_header_nritems(right);
2130                 if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
2131                         wret = 1;
2132                 } else {
2133                         ret = btrfs_cow_block(trans, root, right,
2134                                               parent, pslot + 1,
2135                                               &right);
2136                         if (ret)
2137                                 wret = 1;
2138                         else {
2139                                 wret = balance_node_right(trans, right, mid);
2140                         }
2141                 }
2142                 if (wret < 0)
2143                         ret = wret;
2144                 if (wret == 0) {
2145                         struct btrfs_disk_key disk_key;
2146
2147                         btrfs_node_key(right, &disk_key, 0);
2148                         ret = tree_mod_log_insert_key(parent, pslot + 1,
2149                                         MOD_LOG_KEY_REPLACE, GFP_NOFS);
2150                         BUG_ON(ret < 0);
2151                         btrfs_set_node_key(parent, &disk_key, pslot + 1);
2152                         btrfs_mark_buffer_dirty(parent);
2153
2154                         if (btrfs_header_nritems(mid) <= orig_slot) {
2155                                 path->nodes[level] = right;
2156                                 path->slots[level + 1] += 1;
2157                                 path->slots[level] = orig_slot -
2158                                         btrfs_header_nritems(mid);
2159                                 btrfs_tree_unlock(mid);
2160                                 free_extent_buffer(mid);
2161                         } else {
2162                                 btrfs_tree_unlock(right);
2163                                 free_extent_buffer(right);
2164                         }
2165                         return 0;
2166                 }
2167                 btrfs_tree_unlock(right);
2168                 free_extent_buffer(right);
2169         }
2170         return 1;
2171 }
2172
2173 /*
2174  * readahead one full node of leaves, finding things that are close
2175  * to the block in 'slot', and triggering ra on them.
2176  */
2177 static void reada_for_search(struct btrfs_fs_info *fs_info,
2178                              struct btrfs_path *path,
2179                              int level, int slot, u64 objectid)
2180 {
2181         struct extent_buffer *node;
2182         struct btrfs_disk_key disk_key;
2183         u32 nritems;
2184         u64 search;
2185         u64 target;
2186         u64 nread = 0;
2187         struct extent_buffer *eb;
2188         u32 nr;
2189         u32 blocksize;
2190         u32 nscan = 0;
2191
2192         if (level != 1)
2193                 return;
2194
2195         if (!path->nodes[level])
2196                 return;
2197
2198         node = path->nodes[level];
2199
2200         search = btrfs_node_blockptr(node, slot);
2201         blocksize = fs_info->nodesize;
2202         eb = find_extent_buffer(fs_info, search);
2203         if (eb) {
2204                 free_extent_buffer(eb);
2205                 return;
2206         }
2207
2208         target = search;
2209
2210         nritems = btrfs_header_nritems(node);
2211         nr = slot;
2212
2213         while (1) {
2214                 if (path->reada == READA_BACK) {
2215                         if (nr == 0)
2216                                 break;
2217                         nr--;
2218                 } else if (path->reada == READA_FORWARD) {
2219                         nr++;
2220                         if (nr >= nritems)
2221                                 break;
2222                 }
2223                 if (path->reada == READA_BACK && objectid) {
2224                         btrfs_node_key(node, &disk_key, nr);
2225                         if (btrfs_disk_key_objectid(&disk_key) != objectid)
2226                                 break;
2227                 }
2228                 search = btrfs_node_blockptr(node, nr);
2229                 if ((search <= target && target - search <= 65536) ||
2230                     (search > target && search - target <= 65536)) {
2231                         readahead_tree_block(fs_info, search);
2232                         nread += blocksize;
2233                 }
2234                 nscan++;
2235                 if ((nread > 65536 || nscan > 32))
2236                         break;
2237         }
2238 }
2239
2240 static noinline void reada_for_balance(struct btrfs_fs_info *fs_info,
2241                                        struct btrfs_path *path, int level)
2242 {
2243         int slot;
2244         int nritems;
2245         struct extent_buffer *parent;
2246         struct extent_buffer *eb;
2247         u64 gen;
2248         u64 block1 = 0;
2249         u64 block2 = 0;
2250
2251         parent = path->nodes[level + 1];
2252         if (!parent)
2253                 return;
2254
2255         nritems = btrfs_header_nritems(parent);
2256         slot = path->slots[level + 1];
2257
2258         if (slot > 0) {
2259                 block1 = btrfs_node_blockptr(parent, slot - 1);
2260                 gen = btrfs_node_ptr_generation(parent, slot - 1);
2261                 eb = find_extent_buffer(fs_info, block1);
2262                 /*
2263                  * if we get -eagain from btrfs_buffer_uptodate, we
2264                  * don't want to return eagain here.  That will loop
2265                  * forever
2266                  */
2267                 if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0)
2268                         block1 = 0;
2269                 free_extent_buffer(eb);
2270         }
2271         if (slot + 1 < nritems) {
2272                 block2 = btrfs_node_blockptr(parent, slot + 1);
2273                 gen = btrfs_node_ptr_generation(parent, slot + 1);
2274                 eb = find_extent_buffer(fs_info, block2);
2275                 if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0)
2276                         block2 = 0;
2277                 free_extent_buffer(eb);
2278         }
2279
2280         if (block1)
2281                 readahead_tree_block(fs_info, block1);
2282         if (block2)
2283                 readahead_tree_block(fs_info, block2);
2284 }
2285
2286
2287 /*
2288  * when we walk down the tree, it is usually safe to unlock the higher layers
2289  * in the tree.  The exceptions are when our path goes through slot 0, because
2290  * operations on the tree might require changing key pointers higher up in the
2291  * tree.
2292  *
2293  * callers might also have set path->keep_locks, which tells this code to keep
2294  * the lock if the path points to the last slot in the block.  This is part of
2295  * walking through the tree, and selecting the next slot in the higher block.
2296  *
2297  * lowest_unlock sets the lowest level in the tree we're allowed to unlock.  so
2298  * if lowest_unlock is 1, level 0 won't be unlocked
2299  */
2300 static noinline void unlock_up(struct btrfs_path *path, int level,
2301                                int lowest_unlock, int min_write_lock_level,
2302                                int *write_lock_level)
2303 {
2304         int i;
2305         int skip_level = level;
2306         int no_skips = 0;
2307         struct extent_buffer *t;
2308
2309         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
2310                 if (!path->nodes[i])
2311                         break;
2312                 if (!path->locks[i])
2313                         break;
2314                 if (!no_skips && path->slots[i] == 0) {
2315                         skip_level = i + 1;
2316                         continue;
2317                 }
2318                 if (!no_skips && path->keep_locks) {
2319                         u32 nritems;
2320                         t = path->nodes[i];
2321                         nritems = btrfs_header_nritems(t);
2322                         if (nritems < 1 || path->slots[i] >= nritems - 1) {
2323                                 skip_level = i + 1;
2324                                 continue;
2325                         }
2326                 }
2327                 if (skip_level < i && i >= lowest_unlock)
2328                         no_skips = 1;
2329
2330                 t = path->nodes[i];
2331                 if (i >= lowest_unlock && i > skip_level) {
2332                         btrfs_tree_unlock_rw(t, path->locks[i]);
2333                         path->locks[i] = 0;
2334                         if (write_lock_level &&
2335                             i > min_write_lock_level &&
2336                             i <= *write_lock_level) {
2337                                 *write_lock_level = i - 1;
2338                         }
2339                 }
2340         }
2341 }
2342
2343 /*
2344  * helper function for btrfs_search_slot.  The goal is to find a block
2345  * in cache without setting the path to blocking.  If we find the block
2346  * we return zero and the path is unchanged.
2347  *
2348  * If we can't find the block, we set the path blocking and do some
2349  * reada.  -EAGAIN is returned and the search must be repeated.
2350  */
2351 static int
2352 read_block_for_search(struct btrfs_root *root, struct btrfs_path *p,
2353                       struct extent_buffer **eb_ret, int level, int slot,
2354                       const struct btrfs_key *key)
2355 {
2356         struct btrfs_fs_info *fs_info = root->fs_info;
2357         u64 blocknr;
2358         u64 gen;
2359         struct extent_buffer *tmp;
2360         struct btrfs_key first_key;
2361         int ret;
2362         int parent_level;
2363
2364         blocknr = btrfs_node_blockptr(*eb_ret, slot);
2365         gen = btrfs_node_ptr_generation(*eb_ret, slot);
2366         parent_level = btrfs_header_level(*eb_ret);
2367         btrfs_node_key_to_cpu(*eb_ret, &first_key, slot);
2368
2369         tmp = find_extent_buffer(fs_info, blocknr);
2370         if (tmp) {
2371                 /* first we do an atomic uptodate check */
2372                 if (btrfs_buffer_uptodate(tmp, gen, 1) > 0) {
2373                         /*
2374                          * Do extra check for first_key, eb can be stale due to
2375                          * being cached, read from scrub, or have multiple
2376                          * parents (shared tree blocks).
2377                          */
2378                         if (btrfs_verify_level_key(tmp,
2379                                         parent_level - 1, &first_key, gen)) {
2380                                 free_extent_buffer(tmp);
2381                                 return -EUCLEAN;
2382                         }
2383                         *eb_ret = tmp;
2384                         return 0;
2385                 }
2386
2387                 /* the pages were up to date, but we failed
2388                  * the generation number check.  Do a full
2389                  * read for the generation number that is correct.
2390                  * We must do this without dropping locks so
2391                  * we can trust our generation number
2392                  */
2393                 btrfs_set_path_blocking(p);
2394
2395                 /* now we're allowed to do a blocking uptodate check */
2396                 ret = btrfs_read_buffer(tmp, gen, parent_level - 1, &first_key);
2397                 if (!ret) {
2398                         *eb_ret = tmp;
2399                         return 0;
2400                 }
2401                 free_extent_buffer(tmp);
2402                 btrfs_release_path(p);
2403                 return -EIO;
2404         }
2405
2406         /*
2407          * reduce lock contention at high levels
2408          * of the btree by dropping locks before
2409          * we read.  Don't release the lock on the current
2410          * level because we need to walk this node to figure
2411          * out which blocks to read.
2412          */
2413         btrfs_unlock_up_safe(p, level + 1);
2414         btrfs_set_path_blocking(p);
2415
2416         if (p->reada != READA_NONE)
2417                 reada_for_search(fs_info, p, level, slot, key->objectid);
2418
2419         ret = -EAGAIN;
2420         tmp = read_tree_block(fs_info, blocknr, gen, parent_level - 1,
2421                               &first_key);
2422         if (!IS_ERR(tmp)) {
2423                 /*
2424                  * If the read above didn't mark this buffer up to date,
2425                  * it will never end up being up to date.  Set ret to EIO now
2426                  * and give up so that our caller doesn't loop forever
2427                  * on our EAGAINs.
2428                  */
2429                 if (!extent_buffer_uptodate(tmp))
2430                         ret = -EIO;
2431                 free_extent_buffer(tmp);
2432         } else {
2433                 ret = PTR_ERR(tmp);
2434         }
2435
2436         btrfs_release_path(p);
2437         return ret;
2438 }
2439
2440 /*
2441  * helper function for btrfs_search_slot.  This does all of the checks
2442  * for node-level blocks and does any balancing required based on
2443  * the ins_len.
2444  *
2445  * If no extra work was required, zero is returned.  If we had to
2446  * drop the path, -EAGAIN is returned and btrfs_search_slot must
2447  * start over
2448  */
2449 static int
2450 setup_nodes_for_search(struct btrfs_trans_handle *trans,
2451                        struct btrfs_root *root, struct btrfs_path *p,
2452                        struct extent_buffer *b, int level, int ins_len,
2453                        int *write_lock_level)
2454 {
2455         struct btrfs_fs_info *fs_info = root->fs_info;
2456         int ret;
2457
2458         if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >=
2459             BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 3) {
2460                 int sret;
2461
2462                 if (*write_lock_level < level + 1) {
2463                         *write_lock_level = level + 1;
2464                         btrfs_release_path(p);
2465                         goto again;
2466                 }
2467
2468                 btrfs_set_path_blocking(p);
2469                 reada_for_balance(fs_info, p, level);
2470                 sret = split_node(trans, root, p, level);
2471
2472                 BUG_ON(sret > 0);
2473                 if (sret) {
2474                         ret = sret;
2475                         goto done;
2476                 }
2477                 b = p->nodes[level];
2478         } else if (ins_len < 0 && btrfs_header_nritems(b) <
2479                    BTRFS_NODEPTRS_PER_BLOCK(fs_info) / 2) {
2480                 int sret;
2481
2482                 if (*write_lock_level < level + 1) {
2483                         *write_lock_level = level + 1;
2484                         btrfs_release_path(p);
2485                         goto again;
2486                 }
2487
2488                 btrfs_set_path_blocking(p);
2489                 reada_for_balance(fs_info, p, level);
2490                 sret = balance_level(trans, root, p, level);
2491
2492                 if (sret) {
2493                         ret = sret;
2494                         goto done;
2495                 }
2496                 b = p->nodes[level];
2497                 if (!b) {
2498                         btrfs_release_path(p);
2499                         goto again;
2500                 }
2501                 BUG_ON(btrfs_header_nritems(b) == 1);
2502         }
2503         return 0;
2504
2505 again:
2506         ret = -EAGAIN;
2507 done:
2508         return ret;
2509 }
2510
2511 int btrfs_find_item(struct btrfs_root *fs_root, struct btrfs_path *path,
2512                 u64 iobjectid, u64 ioff, u8 key_type,
2513                 struct btrfs_key *found_key)
2514 {
2515         int ret;
2516         struct btrfs_key key;
2517         struct extent_buffer *eb;
2518
2519         ASSERT(path);
2520         ASSERT(found_key);
2521
2522         key.type = key_type;
2523         key.objectid = iobjectid;
2524         key.offset = ioff;
2525
2526         ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
2527         if (ret < 0)
2528                 return ret;
2529
2530         eb = path->nodes[0];
2531         if (ret && path->slots[0] >= btrfs_header_nritems(eb)) {
2532                 ret = btrfs_next_leaf(fs_root, path);
2533                 if (ret)
2534                         return ret;
2535                 eb = path->nodes[0];
2536         }
2537
2538         btrfs_item_key_to_cpu(eb, found_key, path->slots[0]);
2539         if (found_key->type != key.type ||
2540                         found_key->objectid != key.objectid)
2541                 return 1;
2542
2543         return 0;
2544 }
2545
2546 static struct extent_buffer *btrfs_search_slot_get_root(struct btrfs_root *root,
2547                                                         struct btrfs_path *p,
2548                                                         int write_lock_level)
2549 {
2550         struct btrfs_fs_info *fs_info = root->fs_info;
2551         struct extent_buffer *b;
2552         int root_lock;
2553         int level = 0;
2554
2555         /* We try very hard to do read locks on the root */
2556         root_lock = BTRFS_READ_LOCK;
2557
2558         if (p->search_commit_root) {
2559                 /*
2560                  * The commit roots are read only so we always do read locks,
2561                  * and we always must hold the commit_root_sem when doing
2562                  * searches on them, the only exception is send where we don't
2563                  * want to block transaction commits for a long time, so
2564                  * we need to clone the commit root in order to avoid races
2565                  * with transaction commits that create a snapshot of one of
2566                  * the roots used by a send operation.
2567                  */
2568                 if (p->need_commit_sem) {
2569                         down_read(&fs_info->commit_root_sem);
2570                         b = btrfs_clone_extent_buffer(root->commit_root);
2571                         up_read(&fs_info->commit_root_sem);
2572                         if (!b)
2573                                 return ERR_PTR(-ENOMEM);
2574
2575                 } else {
2576                         b = root->commit_root;
2577                         atomic_inc(&b->refs);
2578                 }
2579                 level = btrfs_header_level(b);
2580                 /*
2581                  * Ensure that all callers have set skip_locking when
2582                  * p->search_commit_root = 1.
2583                  */
2584                 ASSERT(p->skip_locking == 1);
2585
2586                 goto out;
2587         }
2588
2589         if (p->skip_locking) {
2590                 b = btrfs_root_node(root);
2591                 level = btrfs_header_level(b);
2592                 goto out;
2593         }
2594
2595         /*
2596          * If the level is set to maximum, we can skip trying to get the read
2597          * lock.
2598          */
2599         if (write_lock_level < BTRFS_MAX_LEVEL) {
2600                 /*
2601                  * We don't know the level of the root node until we actually
2602                  * have it read locked
2603                  */
2604                 b = btrfs_read_lock_root_node(root);
2605                 level = btrfs_header_level(b);
2606                 if (level > write_lock_level)
2607                         goto out;
2608
2609                 /* Whoops, must trade for write lock */
2610                 btrfs_tree_read_unlock(b);
2611                 free_extent_buffer(b);
2612         }
2613
2614         b = btrfs_lock_root_node(root);
2615         root_lock = BTRFS_WRITE_LOCK;
2616
2617         /* The level might have changed, check again */
2618         level = btrfs_header_level(b);
2619
2620 out:
2621         p->nodes[level] = b;
2622         if (!p->skip_locking)
2623                 p->locks[level] = root_lock;
2624         /*
2625          * Callers are responsible for dropping b's references.
2626          */
2627         return b;
2628 }
2629
2630
2631 /*
2632  * btrfs_search_slot - look for a key in a tree and perform necessary
2633  * modifications to preserve tree invariants.
2634  *
2635  * @trans:      Handle of transaction, used when modifying the tree
2636  * @p:          Holds all btree nodes along the search path
2637  * @root:       The root node of the tree
2638  * @key:        The key we are looking for
2639  * @ins_len:    Indicates purpose of search, for inserts it is 1, for
2640  *              deletions it's -1. 0 for plain searches
2641  * @cow:        boolean should CoW operations be performed. Must always be 1
2642  *              when modifying the tree.
2643  *
2644  * If @ins_len > 0, nodes and leaves will be split as we walk down the tree.
2645  * If @ins_len < 0, nodes will be merged as we walk down the tree (if possible)
2646  *
2647  * If @key is found, 0 is returned and you can find the item in the leaf level
2648  * of the path (level 0)
2649  *
2650  * If @key isn't found, 1 is returned and the leaf level of the path (level 0)
2651  * points to the slot where it should be inserted
2652  *
2653  * If an error is encountered while searching the tree a negative error number
2654  * is returned
2655  */
2656 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2657                       const struct btrfs_key *key, struct btrfs_path *p,
2658                       int ins_len, int cow)
2659 {
2660         struct extent_buffer *b;
2661         int slot;
2662         int ret;
2663         int err;
2664         int level;
2665         int lowest_unlock = 1;
2666         /* everything at write_lock_level or lower must be write locked */
2667         int write_lock_level = 0;
2668         u8 lowest_level = 0;
2669         int min_write_lock_level;
2670         int prev_cmp;
2671
2672         lowest_level = p->lowest_level;
2673         WARN_ON(lowest_level && ins_len > 0);
2674         WARN_ON(p->nodes[0] != NULL);
2675         BUG_ON(!cow && ins_len);
2676
2677         if (ins_len < 0) {
2678                 lowest_unlock = 2;
2679
2680                 /* when we are removing items, we might have to go up to level
2681                  * two as we update tree pointers  Make sure we keep write
2682                  * for those levels as well
2683                  */
2684                 write_lock_level = 2;
2685         } else if (ins_len > 0) {
2686                 /*
2687                  * for inserting items, make sure we have a write lock on
2688                  * level 1 so we can update keys
2689                  */
2690                 write_lock_level = 1;
2691         }
2692
2693         if (!cow)
2694                 write_lock_level = -1;
2695
2696         if (cow && (p->keep_locks || p->lowest_level))
2697                 write_lock_level = BTRFS_MAX_LEVEL;
2698
2699         min_write_lock_level = write_lock_level;
2700
2701 again:
2702         prev_cmp = -1;
2703         b = btrfs_search_slot_get_root(root, p, write_lock_level);
2704         if (IS_ERR(b)) {
2705                 ret = PTR_ERR(b);
2706                 goto done;
2707         }
2708
2709         while (b) {
2710                 int dec = 0;
2711
2712                 level = btrfs_header_level(b);
2713
2714                 if (cow) {
2715                         bool last_level = (level == (BTRFS_MAX_LEVEL - 1));
2716
2717                         /*
2718                          * if we don't really need to cow this block
2719                          * then we don't want to set the path blocking,
2720                          * so we test it here
2721                          */
2722                         if (!should_cow_block(trans, root, b)) {
2723                                 trans->dirty = true;
2724                                 goto cow_done;
2725                         }
2726
2727                         /*
2728                          * must have write locks on this node and the
2729                          * parent
2730                          */
2731                         if (level > write_lock_level ||
2732                             (level + 1 > write_lock_level &&
2733                             level + 1 < BTRFS_MAX_LEVEL &&
2734                             p->nodes[level + 1])) {
2735                                 write_lock_level = level + 1;
2736                                 btrfs_release_path(p);
2737                                 goto again;
2738                         }
2739
2740                         btrfs_set_path_blocking(p);
2741                         if (last_level)
2742                                 err = btrfs_cow_block(trans, root, b, NULL, 0,
2743                                                       &b);
2744                         else
2745                                 err = btrfs_cow_block(trans, root, b,
2746                                                       p->nodes[level + 1],
2747                                                       p->slots[level + 1], &b);
2748                         if (err) {
2749                                 ret = err;
2750                                 goto done;
2751                         }
2752                 }
2753 cow_done:
2754                 p->nodes[level] = b;
2755                 /*
2756                  * Leave path with blocking locks to avoid massive
2757                  * lock context switch, this is made on purpose.
2758                  */
2759
2760                 /*
2761                  * we have a lock on b and as long as we aren't changing
2762                  * the tree, there is no way to for the items in b to change.
2763                  * It is safe to drop the lock on our parent before we
2764                  * go through the expensive btree search on b.
2765                  *
2766                  * If we're inserting or deleting (ins_len != 0), then we might
2767                  * be changing slot zero, which may require changing the parent.
2768                  * So, we can't drop the lock until after we know which slot
2769                  * we're operating on.
2770                  */
2771                 if (!ins_len && !p->keep_locks) {
2772                         int u = level + 1;
2773
2774                         if (u < BTRFS_MAX_LEVEL && p->locks[u]) {
2775                                 btrfs_tree_unlock_rw(p->nodes[u], p->locks[u]);
2776                                 p->locks[u] = 0;
2777                         }
2778                 }
2779
2780                 /*
2781                  * If btrfs_bin_search returns an exact match (prev_cmp == 0)
2782                  * we can safely assume the target key will always be in slot 0
2783                  * on lower levels due to the invariants BTRFS' btree provides,
2784                  * namely that a btrfs_key_ptr entry always points to the
2785                  * lowest key in the child node, thus we can skip searching
2786                  * lower levels
2787                  */
2788                 if (prev_cmp == 0) {
2789                         slot = 0;
2790                         ret = 0;
2791                 } else {
2792                         ret = btrfs_bin_search(b, key, &slot);
2793                         prev_cmp = ret;
2794                         if (ret < 0)
2795                                 goto done;
2796                 }
2797
2798                 if (level == 0) {
2799                         p->slots[level] = slot;
2800                         if (ins_len > 0 &&
2801                             btrfs_leaf_free_space(b) < ins_len) {
2802                                 if (write_lock_level < 1) {
2803                                         write_lock_level = 1;
2804                                         btrfs_release_path(p);
2805                                         goto again;
2806                                 }
2807
2808                                 btrfs_set_path_blocking(p);
2809                                 err = split_leaf(trans, root, key,
2810                                                  p, ins_len, ret == 0);
2811
2812                                 BUG_ON(err > 0);
2813                                 if (err) {
2814                                         ret = err;
2815                                         goto done;
2816                                 }
2817                         }
2818                         if (!p->search_for_split)
2819                                 unlock_up(p, level, lowest_unlock,
2820                                           min_write_lock_level, NULL);
2821                         goto done;
2822                 }
2823                 if (ret && slot > 0) {
2824                         dec = 1;
2825                         slot--;
2826                 }
2827                 p->slots[level] = slot;
2828                 err = setup_nodes_for_search(trans, root, p, b, level, ins_len,
2829                                              &write_lock_level);
2830                 if (err == -EAGAIN)
2831                         goto again;
2832                 if (err) {
2833                         ret = err;
2834                         goto done;
2835                 }
2836                 b = p->nodes[level];
2837                 slot = p->slots[level];
2838
2839                 /*
2840                  * Slot 0 is special, if we change the key we have to update
2841                  * the parent pointer which means we must have a write lock on
2842                  * the parent
2843                  */
2844                 if (slot == 0 && ins_len && write_lock_level < level + 1) {
2845                         write_lock_level = level + 1;
2846                         btrfs_release_path(p);
2847                         goto again;
2848                 }
2849
2850                 unlock_up(p, level, lowest_unlock, min_write_lock_level,
2851                           &write_lock_level);
2852
2853                 if (level == lowest_level) {
2854                         if (dec)
2855                                 p->slots[level]++;
2856                         goto done;
2857                 }
2858
2859                 err = read_block_for_search(root, p, &b, level, slot, key);
2860                 if (err == -EAGAIN)
2861                         goto again;
2862                 if (err) {
2863                         ret = err;
2864                         goto done;
2865                 }
2866
2867                 if (!p->skip_locking) {
2868                         level = btrfs_header_level(b);
2869                         if (level <= write_lock_level) {
2870                                 if (!btrfs_try_tree_write_lock(b)) {
2871                                         btrfs_set_path_blocking(p);
2872                                         btrfs_tree_lock(b);
2873                                 }
2874                                 p->locks[level] = BTRFS_WRITE_LOCK;
2875                         } else {
2876                                 if (!btrfs_tree_read_lock_atomic(b)) {
2877                                         btrfs_set_path_blocking(p);
2878                                         btrfs_tree_read_lock(b);
2879                                 }
2880                                 p->locks[level] = BTRFS_READ_LOCK;
2881                         }
2882                         p->nodes[level] = b;
2883                 }
2884         }
2885         ret = 1;
2886 done:
2887         /*
2888          * we don't really know what they plan on doing with the path
2889          * from here on, so for now just mark it as blocking
2890          */
2891         if (!p->leave_spinning)
2892                 btrfs_set_path_blocking(p);
2893         if (ret < 0 && !p->skip_release_on_error)
2894                 btrfs_release_path(p);
2895         return ret;
2896 }
2897
2898 /*
2899  * Like btrfs_search_slot, this looks for a key in the given tree. It uses the
2900  * current state of the tree together with the operations recorded in the tree
2901  * modification log to search for the key in a previous version of this tree, as
2902  * denoted by the time_seq parameter.
2903  *
2904  * Naturally, there is no support for insert, delete or cow operations.
2905  *
2906  * The resulting path and return value will be set up as if we called
2907  * btrfs_search_slot at that point in time with ins_len and cow both set to 0.
2908  */
2909 int btrfs_search_old_slot(struct btrfs_root *root, const struct btrfs_key *key,
2910                           struct btrfs_path *p, u64 time_seq)
2911 {
2912         struct btrfs_fs_info *fs_info = root->fs_info;
2913         struct extent_buffer *b;
2914         int slot;
2915         int ret;
2916         int err;
2917         int level;
2918         int lowest_unlock = 1;
2919         u8 lowest_level = 0;
2920
2921         lowest_level = p->lowest_level;
2922         WARN_ON(p->nodes[0] != NULL);
2923
2924         if (p->search_commit_root) {
2925                 BUG_ON(time_seq);
2926                 return btrfs_search_slot(NULL, root, key, p, 0, 0);
2927         }
2928
2929 again:
2930         b = get_old_root(root, time_seq);
2931         if (!b) {
2932                 ret = -EIO;
2933                 goto done;
2934         }
2935         level = btrfs_header_level(b);
2936         p->locks[level] = BTRFS_READ_LOCK;
2937
2938         while (b) {
2939                 int dec = 0;
2940
2941                 level = btrfs_header_level(b);
2942                 p->nodes[level] = b;
2943
2944                 /*
2945                  * we have a lock on b and as long as we aren't changing
2946                  * the tree, there is no way to for the items in b to change.
2947                  * It is safe to drop the lock on our parent before we
2948                  * go through the expensive btree search on b.
2949                  */
2950                 btrfs_unlock_up_safe(p, level + 1);
2951
2952                 ret = btrfs_bin_search(b, key, &slot);
2953                 if (ret < 0)
2954                         goto done;
2955
2956                 if (level == 0) {
2957                         p->slots[level] = slot;
2958                         unlock_up(p, level, lowest_unlock, 0, NULL);
2959                         goto done;
2960                 }
2961
2962                 if (ret && slot > 0) {
2963                         dec = 1;
2964                         slot--;
2965                 }
2966                 p->slots[level] = slot;
2967                 unlock_up(p, level, lowest_unlock, 0, NULL);
2968
2969                 if (level == lowest_level) {
2970                         if (dec)
2971                                 p->slots[level]++;
2972                         goto done;
2973                 }
2974
2975                 err = read_block_for_search(root, p, &b, level, slot, key);
2976                 if (err == -EAGAIN)
2977                         goto again;
2978                 if (err) {
2979                         ret = err;
2980                         goto done;
2981                 }
2982
2983                 level = btrfs_header_level(b);
2984                 if (!btrfs_tree_read_lock_atomic(b)) {
2985                         btrfs_set_path_blocking(p);
2986                         btrfs_tree_read_lock(b);
2987                 }
2988                 b = tree_mod_log_rewind(fs_info, p, b, time_seq);
2989                 if (!b) {
2990                         ret = -ENOMEM;
2991                         goto done;
2992                 }
2993                 p->locks[level] = BTRFS_READ_LOCK;
2994                 p->nodes[level] = b;
2995         }
2996         ret = 1;
2997 done:
2998         if (!p->leave_spinning)
2999                 btrfs_set_path_blocking(p);
3000         if (ret < 0)
3001                 btrfs_release_path(p);
3002
3003         return ret;
3004 }
3005
3006 /*
3007  * helper to use instead of search slot if no exact match is needed but
3008  * instead the next or previous item should be returned.
3009  * When find_higher is true, the next higher item is returned, the next lower
3010  * otherwise.
3011  * When return_any and find_higher are both true, and no higher item is found,
3012  * return the next lower instead.
3013  * When return_any is true and find_higher is false, and no lower item is found,
3014  * return the next higher instead.
3015  * It returns 0 if any item is found, 1 if none is found (tree empty), and
3016  * < 0 on error
3017  */
3018 int btrfs_search_slot_for_read(struct btrfs_root *root,
3019                                const struct btrfs_key *key,
3020                                struct btrfs_path *p, int find_higher,
3021                                int return_any)
3022 {
3023         int ret;
3024         struct extent_buffer *leaf;
3025
3026 again:
3027         ret = btrfs_search_slot(NULL, root, key, p, 0, 0);
3028         if (ret <= 0)
3029                 return ret;
3030         /*
3031          * a return value of 1 means the path is at the position where the
3032          * item should be inserted. Normally this is the next bigger item,
3033          * but in case the previous item is the last in a leaf, path points
3034          * to the first free slot in the previous leaf, i.e. at an invalid
3035          * item.
3036          */
3037         leaf = p->nodes[0];
3038
3039         if (find_higher) {
3040                 if (p->slots[0] >= btrfs_header_nritems(leaf)) {
3041                         ret = btrfs_next_leaf(root, p);
3042                         if (ret <= 0)
3043                                 return ret;
3044                         if (!return_any)
3045                                 return 1;
3046                         /*
3047                          * no higher item found, return the next
3048                          * lower instead
3049                          */
3050                         return_any = 0;
3051                         find_higher = 0;
3052                         btrfs_release_path(p);
3053                         goto again;
3054                 }
3055         } else {
3056                 if (p->slots[0] == 0) {
3057                         ret = btrfs_prev_leaf(root, p);
3058                         if (ret < 0)
3059                                 return ret;
3060                         if (!ret) {
3061                                 leaf = p->nodes[0];
3062                                 if (p->slots[0] == btrfs_header_nritems(leaf))
3063                                         p->slots[0]--;
3064                                 return 0;
3065                         }
3066                         if (!return_any)
3067                                 return 1;
3068                         /*
3069                          * no lower item found, return the next
3070                          * higher instead
3071                          */
3072                         return_any = 0;
3073                         find_higher = 1;
3074                         btrfs_release_path(p);
3075                         goto again;
3076                 } else {
3077                         --p->slots[0];
3078                 }
3079         }
3080         return 0;
3081 }
3082
3083 /*
3084  * adjust the pointers going up the tree, starting at level
3085  * making sure the right key of each node is points to 'key'.
3086  * This is used after shifting pointers to the left, so it stops
3087  * fixing up pointers when a given leaf/node is not in slot 0 of the
3088  * higher levels
3089  *
3090  */
3091 static void fixup_low_keys(struct btrfs_path *path,
3092                            struct btrfs_disk_key *key, int level)
3093 {
3094         int i;
3095         struct extent_buffer *t;
3096         int ret;
3097
3098         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
3099                 int tslot = path->slots[i];
3100
3101                 if (!path->nodes[i])
3102                         break;
3103                 t = path->nodes[i];
3104                 ret = tree_mod_log_insert_key(t, tslot, MOD_LOG_KEY_REPLACE,
3105                                 GFP_ATOMIC);
3106                 BUG_ON(ret < 0);
3107                 btrfs_set_node_key(t, key, tslot);
3108                 btrfs_mark_buffer_dirty(path->nodes[i]);
3109                 if (tslot != 0)
3110                         break;
3111         }
3112 }
3113
3114 /*
3115  * update item key.
3116  *
3117  * This function isn't completely safe. It's the caller's responsibility
3118  * that the new key won't break the order
3119  */
3120 void btrfs_set_item_key_safe(struct btrfs_fs_info *fs_info,
3121                              struct btrfs_path *path,
3122                              const struct btrfs_key *new_key)
3123 {
3124         struct btrfs_disk_key disk_key;
3125         struct extent_buffer *eb;
3126         int slot;
3127
3128         eb = path->nodes[0];
3129         slot = path->slots[0];
3130         if (slot > 0) {
3131                 btrfs_item_key(eb, &disk_key, slot - 1);
3132                 if (unlikely(comp_keys(&disk_key, new_key) >= 0)) {
3133                         btrfs_crit(fs_info,
3134                 "slot %u key (%llu %u %llu) new key (%llu %u %llu)",
3135                                    slot, btrfs_disk_key_objectid(&disk_key),
3136                                    btrfs_disk_key_type(&disk_key),
3137                                    btrfs_disk_key_offset(&disk_key),
3138                                    new_key->objectid, new_key->type,
3139                                    new_key->offset);
3140                         btrfs_print_leaf(eb);
3141                         BUG();
3142                 }
3143         }
3144         if (slot < btrfs_header_nritems(eb) - 1) {
3145                 btrfs_item_key(eb, &disk_key, slot + 1);
3146                 if (unlikely(comp_keys(&disk_key, new_key) <= 0)) {
3147                         btrfs_crit(fs_info,
3148                 "slot %u key (%llu %u %llu) new key (%llu %u %llu)",
3149                                    slot, btrfs_disk_key_objectid(&disk_key),
3150                                    btrfs_disk_key_type(&disk_key),
3151                                    btrfs_disk_key_offset(&disk_key),
3152                                    new_key->objectid, new_key->type,
3153                                    new_key->offset);
3154                         btrfs_print_leaf(eb);
3155                         BUG();
3156                 }
3157         }
3158
3159         btrfs_cpu_key_to_disk(&disk_key, new_key);
3160         btrfs_set_item_key(eb, &disk_key, slot);
3161         btrfs_mark_buffer_dirty(eb);
3162         if (slot == 0)
3163                 fixup_low_keys(path, &disk_key, 1);
3164 }
3165
3166 /*
3167  * try to push data from one node into the next node left in the
3168  * tree.
3169  *
3170  * returns 0 if some ptrs were pushed left, < 0 if there was some horrible
3171  * error, and > 0 if there was no room in the left hand block.
3172  */
3173 static int push_node_left(struct btrfs_trans_handle *trans,
3174                           struct extent_buffer *dst,
3175                           struct extent_buffer *src, int empty)
3176 {
3177         struct btrfs_fs_info *fs_info = trans->fs_info;
3178         int push_items = 0;
3179         int src_nritems;
3180         int dst_nritems;
3181         int ret = 0;
3182
3183         src_nritems = btrfs_header_nritems(src);
3184         dst_nritems = btrfs_header_nritems(dst);
3185         push_items = BTRFS_NODEPTRS_PER_BLOCK(fs_info) - dst_nritems;
3186         WARN_ON(btrfs_header_generation(src) != trans->transid);
3187         WARN_ON(btrfs_header_generation(dst) != trans->transid);
3188
3189         if (!empty && src_nritems <= 8)
3190                 return 1;
3191
3192         if (push_items <= 0)
3193                 return 1;
3194
3195         if (empty) {
3196                 push_items = min(src_nritems, push_items);
3197                 if (push_items < src_nritems) {
3198                         /* leave at least 8 pointers in the node if
3199                          * we aren't going to empty it
3200                          */
3201                         if (src_nritems - push_items < 8) {
3202                                 if (push_items <= 8)
3203                                         return 1;
3204                                 push_items -= 8;
3205                         }
3206                 }
3207         } else
3208                 push_items = min(src_nritems - 8, push_items);
3209
3210         ret = tree_mod_log_eb_copy(dst, src, dst_nritems, 0, push_items);
3211         if (ret) {
3212                 btrfs_abort_transaction(trans, ret);
3213                 return ret;
3214         }
3215         copy_extent_buffer(dst, src,
3216                            btrfs_node_key_ptr_offset(dst_nritems),
3217                            btrfs_node_key_ptr_offset(0),
3218                            push_items * sizeof(struct btrfs_key_ptr));
3219
3220         if (push_items < src_nritems) {
3221                 /*
3222                  * Don't call tree_mod_log_insert_move here, key removal was
3223                  * already fully logged by tree_mod_log_eb_copy above.
3224                  */
3225                 memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0),
3226                                       btrfs_node_key_ptr_offset(push_items),
3227                                       (src_nritems - push_items) *
3228                                       sizeof(struct btrfs_key_ptr));
3229         }
3230         btrfs_set_header_nritems(src, src_nritems - push_items);
3231         btrfs_set_header_nritems(dst, dst_nritems + push_items);
3232         btrfs_mark_buffer_dirty(src);
3233         btrfs_mark_buffer_dirty(dst);
3234
3235         return ret;
3236 }
3237
3238 /*
3239  * try to push data from one node into the next node right in the
3240  * tree.
3241  *
3242  * returns 0 if some ptrs were pushed, < 0 if there was some horrible
3243  * error, and > 0 if there was no room in the right hand block.
3244  *
3245  * this will  only push up to 1/2 the contents of the left node over
3246  */
3247 static int balance_node_right(struct btrfs_trans_handle *trans,
3248                               struct extent_buffer *dst,
3249                               struct extent_buffer *src)
3250 {
3251         struct btrfs_fs_info *fs_info = trans->fs_info;
3252         int push_items = 0;
3253         int max_push;
3254         int src_nritems;
3255         int dst_nritems;
3256         int ret = 0;
3257
3258         WARN_ON(btrfs_header_generation(src) != trans->transid);
3259         WARN_ON(btrfs_header_generation(dst) != trans->transid);
3260
3261         src_nritems = btrfs_header_nritems(src);
3262         dst_nritems = btrfs_header_nritems(dst);
3263         push_items = BTRFS_NODEPTRS_PER_BLOCK(fs_info) - dst_nritems;
3264         if (push_items <= 0)
3265                 return 1;
3266
3267         if (src_nritems < 4)
3268                 return 1;
3269
3270         max_push = src_nritems / 2 + 1;
3271         /* don't try to empty the node */
3272         if (max_push >= src_nritems)
3273                 return 1;
3274
3275         if (max_push < push_items)
3276                 push_items = max_push;
3277
3278         ret = tree_mod_log_insert_move(dst, push_items, 0, dst_nritems);
3279         BUG_ON(ret < 0);
3280         memmove_extent_buffer(dst, btrfs_node_key_ptr_offset(push_items),
3281                                       btrfs_node_key_ptr_offset(0),
3282                                       (dst_nritems) *
3283                                       sizeof(struct btrfs_key_ptr));
3284
3285         ret = tree_mod_log_eb_copy(dst, src, 0, src_nritems - push_items,
3286                                    push_items);
3287         if (ret) {
3288                 btrfs_abort_transaction(trans, ret);
3289                 return ret;
3290         }
3291         copy_extent_buffer(dst, src,
3292                            btrfs_node_key_ptr_offset(0),
3293                            btrfs_node_key_ptr_offset(src_nritems - push_items),
3294                            push_items * sizeof(struct btrfs_key_ptr));
3295
3296         btrfs_set_header_nritems(src, src_nritems - push_items);
3297         btrfs_set_header_nritems(dst, dst_nritems + push_items);
3298
3299         btrfs_mark_buffer_dirty(src);
3300         btrfs_mark_buffer_dirty(dst);
3301
3302         return ret;
3303 }
3304
3305 /*
3306  * helper function to insert a new root level in the tree.
3307  * A new node is allocated, and a single item is inserted to
3308  * point to the existing root
3309  *
3310  * returns zero on success or < 0 on failure.
3311  */
3312 static noinline int insert_new_root(struct btrfs_trans_handle *trans,
3313                            struct btrfs_root *root,
3314                            struct btrfs_path *path, int level)
3315 {
3316         struct btrfs_fs_info *fs_info = root->fs_info;
3317         u64 lower_gen;
3318         struct extent_buffer *lower;
3319         struct extent_buffer *c;
3320         struct extent_buffer *old;
3321         struct btrfs_disk_key lower_key;
3322         int ret;
3323
3324         BUG_ON(path->nodes[level]);
3325         BUG_ON(path->nodes[level-1] != root->node);
3326
3327         lower = path->nodes[level-1];
3328         if (level == 1)
3329                 btrfs_item_key(lower, &lower_key, 0);
3330         else
3331                 btrfs_node_key(lower, &lower_key, 0);
3332
3333         c = alloc_tree_block_no_bg_flush(trans, root, 0, &lower_key, level,
3334                                          root->node->start, 0);
3335         if (IS_ERR(c))
3336                 return PTR_ERR(c);
3337
3338         root_add_used(root, fs_info->nodesize);
3339
3340         btrfs_set_header_nritems(c, 1);
3341         btrfs_set_node_key(c, &lower_key, 0);
3342         btrfs_set_node_blockptr(c, 0, lower->start);
3343         lower_gen = btrfs_header_generation(lower);
3344         WARN_ON(lower_gen != trans->transid);
3345
3346         btrfs_set_node_ptr_generation(c, 0, lower_gen);
3347
3348         btrfs_mark_buffer_dirty(c);
3349
3350         old = root->node;
3351         ret = tree_mod_log_insert_root(root->node, c, 0);
3352         BUG_ON(ret < 0);
3353         rcu_assign_pointer(root->node, c);
3354
3355         /* the super has an extra ref to root->node */
3356         free_extent_buffer(old);
3357
3358         add_root_to_dirty_list(root);
3359         atomic_inc(&c->refs);
3360         path->nodes[level] = c;
3361         path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
3362         path->slots[level] = 0;
3363         return 0;
3364 }
3365
3366 /*
3367  * worker function to insert a single pointer in a node.
3368  * the node should have enough room for the pointer already
3369  *
3370  * slot and level indicate where you want the key to go, and
3371  * blocknr is the block the key points to.
3372  */
3373 static void insert_ptr(struct btrfs_trans_handle *trans,
3374                        struct btrfs_path *path,
3375                        struct btrfs_disk_key *key, u64 bytenr,
3376                        int slot, int level)
3377 {
3378         struct extent_buffer *lower;
3379         int nritems;
3380         int ret;
3381
3382         BUG_ON(!path->nodes[level]);
3383         btrfs_assert_tree_locked(path->nodes[level]);
3384         lower = path->nodes[level];
3385         nritems = btrfs_header_nritems(lower);
3386         BUG_ON(slot > nritems);
3387         BUG_ON(nritems == BTRFS_NODEPTRS_PER_BLOCK(trans->fs_info));
3388         if (slot != nritems) {
3389                 if (level) {
3390                         ret = tree_mod_log_insert_move(lower, slot + 1, slot,
3391                                         nritems - slot);
3392                         BUG_ON(ret < 0);
3393                 }
3394                 memmove_extent_buffer(lower,
3395                               btrfs_node_key_ptr_offset(slot + 1),
3396                               btrfs_node_key_ptr_offset(slot),
3397                               (nritems - slot) * sizeof(struct btrfs_key_ptr));
3398         }
3399         if (level) {
3400                 ret = tree_mod_log_insert_key(lower, slot, MOD_LOG_KEY_ADD,
3401                                 GFP_NOFS);
3402                 BUG_ON(ret < 0);
3403         }
3404         btrfs_set_node_key(lower, key, slot);
3405         btrfs_set_node_blockptr(lower, slot, bytenr);
3406         WARN_ON(trans->transid == 0);
3407         btrfs_set_node_ptr_generation(lower, slot, trans->transid);
3408         btrfs_set_header_nritems(lower, nritems + 1);
3409         btrfs_mark_buffer_dirty(lower);
3410 }
3411
3412 /*
3413  * split the node at the specified level in path in two.
3414  * The path is corrected to point to the appropriate node after the split
3415  *
3416  * Before splitting this tries to make some room in the node by pushing
3417  * left and right, if either one works, it returns right away.
3418  *
3419  * returns 0 on success and < 0 on failure
3420  */
3421 static noinline int split_node(struct btrfs_trans_handle *trans,
3422                                struct btrfs_root *root,
3423                                struct btrfs_path *path, int level)
3424 {
3425         struct btrfs_fs_info *fs_info = root->fs_info;
3426         struct extent_buffer *c;
3427         struct extent_buffer *split;
3428         struct btrfs_disk_key disk_key;
3429         int mid;
3430         int ret;
3431         u32 c_nritems;
3432
3433         c = path->nodes[level];
3434         WARN_ON(btrfs_header_generation(c) != trans->transid);
3435         if (c == root->node) {
3436                 /*
3437                  * trying to split the root, lets make a new one
3438                  *
3439                  * tree mod log: We don't log_removal old root in
3440                  * insert_new_root, because that root buffer will be kept as a
3441                  * normal node. We are going to log removal of half of the
3442                  * elements below with tree_mod_log_eb_copy. We're holding a
3443                  * tree lock on the buffer, which is why we cannot race with
3444                  * other tree_mod_log users.
3445                  */
3446                 ret = insert_new_root(trans, root, path, level + 1);
3447                 if (ret)
3448                         return ret;
3449         } else {
3450                 ret = push_nodes_for_insert(trans, root, path, level);
3451                 c = path->nodes[level];
3452                 if (!ret && btrfs_header_nritems(c) <
3453                     BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 3)
3454                         return 0;
3455                 if (ret < 0)
3456                         return ret;
3457         }
3458
3459         c_nritems = btrfs_header_nritems(c);
3460         mid = (c_nritems + 1) / 2;
3461         btrfs_node_key(c, &disk_key, mid);
3462
3463         split = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, level,
3464                                              c->start, 0);
3465         if (IS_ERR(split))
3466                 return PTR_ERR(split);
3467
3468         root_add_used(root, fs_info->nodesize);
3469         ASSERT(btrfs_header_level(c) == level);
3470
3471         ret = tree_mod_log_eb_copy(split, c, 0, mid, c_nritems - mid);
3472         if (ret) {
3473                 btrfs_abort_transaction(trans, ret);
3474                 return ret;
3475         }
3476         copy_extent_buffer(split, c,
3477                            btrfs_node_key_ptr_offset(0),
3478                            btrfs_node_key_ptr_offset(mid),
3479                            (c_nritems - mid) * sizeof(struct btrfs_key_ptr));
3480         btrfs_set_header_nritems(split, c_nritems - mid);
3481         btrfs_set_header_nritems(c, mid);
3482         ret = 0;
3483
3484         btrfs_mark_buffer_dirty(c);
3485         btrfs_mark_buffer_dirty(split);
3486
3487         insert_ptr(trans, path, &disk_key, split->start,
3488                    path->slots[level + 1] + 1, level + 1);
3489
3490         if (path->slots[level] >= mid) {
3491                 path->slots[level] -= mid;
3492                 btrfs_tree_unlock(c);
3493                 free_extent_buffer(c);
3494                 path->nodes[level] = split;
3495                 path->slots[level + 1] += 1;
3496         } else {
3497                 btrfs_tree_unlock(split);
3498                 free_extent_buffer(split);
3499         }
3500         return ret;
3501 }
3502
3503 /*
3504  * how many bytes are required to store the items in a leaf.  start
3505  * and nr indicate which items in the leaf to check.  This totals up the
3506  * space used both by the item structs and the item data
3507  */
3508 static int leaf_space_used(struct extent_buffer *l, int start, int nr)
3509 {
3510         struct btrfs_item *start_item;
3511         struct btrfs_item *end_item;
3512         int data_len;
3513         int nritems = btrfs_header_nritems(l);
3514         int end = min(nritems, start + nr) - 1;
3515
3516         if (!nr)
3517                 return 0;
3518         start_item = btrfs_item_nr(start);
3519         end_item = btrfs_item_nr(end);
3520         data_len = btrfs_item_offset(l, start_item) +
3521                    btrfs_item_size(l, start_item);
3522         data_len = data_len - btrfs_item_offset(l, end_item);
3523         data_len += sizeof(struct btrfs_item) * nr;
3524         WARN_ON(data_len < 0);
3525         return data_len;
3526 }
3527
3528 /*
3529  * The space between the end of the leaf items and
3530  * the start of the leaf data.  IOW, how much room
3531  * the leaf has left for both items and data
3532  */
3533 noinline int btrfs_leaf_free_space(struct extent_buffer *leaf)
3534 {
3535         struct btrfs_fs_info *fs_info = leaf->fs_info;
3536         int nritems = btrfs_header_nritems(leaf);
3537         int ret;
3538
3539         ret = BTRFS_LEAF_DATA_SIZE(fs_info) - leaf_space_used(leaf, 0, nritems);
3540         if (ret < 0) {
3541                 btrfs_crit(fs_info,
3542                            "leaf free space ret %d, leaf data size %lu, used %d nritems %d",
3543                            ret,
3544                            (unsigned long) BTRFS_LEAF_DATA_SIZE(fs_info),
3545                            leaf_space_used(leaf, 0, nritems), nritems);
3546         }
3547         return ret;
3548 }
3549
3550 /*
3551  * min slot controls the lowest index we're willing to push to the
3552  * right.  We'll push up to and including min_slot, but no lower
3553  */
3554 static noinline int __push_leaf_right(struct btrfs_path *path,
3555                                       int data_size, int empty,
3556                                       struct extent_buffer *right,
3557                                       int free_space, u32 left_nritems,
3558                                       u32 min_slot)
3559 {
3560         struct btrfs_fs_info *fs_info = right->fs_info;
3561         struct extent_buffer *left = path->nodes[0];
3562         struct extent_buffer *upper = path->nodes[1];
3563         struct btrfs_map_token token;
3564         struct btrfs_disk_key disk_key;
3565         int slot;
3566         u32 i;
3567         int push_space = 0;
3568         int push_items = 0;
3569         struct btrfs_item *item;
3570         u32 nr;
3571         u32 right_nritems;
3572         u32 data_end;
3573         u32 this_item_size;
3574
3575         if (empty)
3576                 nr = 0;
3577         else
3578                 nr = max_t(u32, 1, min_slot);
3579
3580         if (path->slots[0] >= left_nritems)
3581                 push_space += data_size;
3582
3583         slot = path->slots[1];
3584         i = left_nritems - 1;
3585         while (i >= nr) {
3586                 item = btrfs_item_nr(i);
3587
3588                 if (!empty && push_items > 0) {
3589                         if (path->slots[0] > i)
3590                                 break;
3591                         if (path->slots[0] == i) {
3592                                 int space = btrfs_leaf_free_space(left);
3593
3594                                 if (space + push_space * 2 > free_space)
3595                                         break;
3596                         }
3597                 }
3598
3599                 if (path->slots[0] == i)
3600                         push_space += data_size;
3601
3602                 this_item_size = btrfs_item_size(left, item);
3603                 if (this_item_size + sizeof(*item) + push_space > free_space)
3604                         break;
3605
3606                 push_items++;
3607                 push_space += this_item_size + sizeof(*item);
3608                 if (i == 0)
3609                         break;
3610                 i--;
3611         }
3612
3613         if (push_items == 0)
3614                 goto out_unlock;
3615
3616         WARN_ON(!empty && push_items == left_nritems);
3617
3618         /* push left to right */
3619         right_nritems = btrfs_header_nritems(right);
3620
3621         push_space = btrfs_item_end_nr(left, left_nritems - push_items);
3622         push_space -= leaf_data_end(left);
3623
3624         /* make room in the right data area */
3625         data_end = leaf_data_end(right);
3626         memmove_extent_buffer(right,
3627                               BTRFS_LEAF_DATA_OFFSET + data_end - push_space,
3628                               BTRFS_LEAF_DATA_OFFSET + data_end,
3629                               BTRFS_LEAF_DATA_SIZE(fs_info) - data_end);
3630
3631         /* copy from the left data area */
3632         copy_extent_buffer(right, left, BTRFS_LEAF_DATA_OFFSET +
3633                      BTRFS_LEAF_DATA_SIZE(fs_info) - push_space,
3634                      BTRFS_LEAF_DATA_OFFSET + leaf_data_end(left),
3635                      push_space);
3636
3637         memmove_extent_buffer(right, btrfs_item_nr_offset(push_items),
3638                               btrfs_item_nr_offset(0),
3639                               right_nritems * sizeof(struct btrfs_item));
3640
3641         /* copy the items from left to right */
3642         copy_extent_buffer(right, left, btrfs_item_nr_offset(0),
3643                    btrfs_item_nr_offset(left_nritems - push_items),
3644                    push_items * sizeof(struct btrfs_item));
3645
3646         /* update the item pointers */
3647         btrfs_init_map_token(&token, right);
3648         right_nritems += push_items;
3649         btrfs_set_header_nritems(right, right_nritems);
3650         push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
3651         for (i = 0; i < right_nritems; i++) {
3652                 item = btrfs_item_nr(i);
3653                 push_space -= btrfs_token_item_size(&token, item);
3654                 btrfs_set_token_item_offset(&token, item, push_space);
3655         }
3656
3657         left_nritems -= push_items;
3658         btrfs_set_header_nritems(left, left_nritems);
3659
3660         if (left_nritems)
3661                 btrfs_mark_buffer_dirty(left);
3662         else
3663                 btrfs_clean_tree_block(left);
3664
3665         btrfs_mark_buffer_dirty(right);
3666
3667         btrfs_item_key(right, &disk_key, 0);
3668         btrfs_set_node_key(upper, &disk_key, slot + 1);
3669         btrfs_mark_buffer_dirty(upper);
3670
3671         /* then fixup the leaf pointer in the path */
3672         if (path->slots[0] >= left_nritems) {
3673                 path->slots[0] -= left_nritems;
3674                 if (btrfs_header_nritems(path->nodes[0]) == 0)
3675                         btrfs_clean_tree_block(path->nodes[0]);
3676                 btrfs_tree_unlock(path->nodes[0]);
3677                 free_extent_buffer(path->nodes[0]);
3678                 path->nodes[0] = right;
3679                 path->slots[1] += 1;
3680         } else {
3681                 btrfs_tree_unlock(right);
3682                 free_extent_buffer(right);
3683         }
3684         return 0;
3685
3686 out_unlock:
3687         btrfs_tree_unlock(right);
3688         free_extent_buffer(right);
3689         return 1;
3690 }
3691
3692 /*
3693  * push some data in the path leaf to the right, trying to free up at
3694  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3695  *
3696  * returns 1 if the push failed because the other node didn't have enough
3697  * room, 0 if everything worked out and < 0 if there were major errors.
3698  *
3699  * this will push starting from min_slot to the end of the leaf.  It won't
3700  * push any slot lower than min_slot
3701  */
3702 static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
3703                            *root, struct btrfs_path *path,
3704                            int min_data_size, int data_size,
3705                            int empty, u32 min_slot)
3706 {
3707         struct extent_buffer *left = path->nodes[0];
3708         struct extent_buffer *right;
3709         struct extent_buffer *upper;
3710         int slot;
3711         int free_space;
3712         u32 left_nritems;
3713         int ret;
3714
3715         if (!path->nodes[1])
3716                 return 1;
3717
3718         slot = path->slots[1];
3719         upper = path->nodes[1];
3720         if (slot >= btrfs_header_nritems(upper) - 1)
3721                 return 1;
3722
3723         btrfs_assert_tree_locked(path->nodes[1]);
3724
3725         right = btrfs_read_node_slot(upper, slot + 1);
3726         /*
3727          * slot + 1 is not valid or we fail to read the right node,
3728          * no big deal, just return.
3729          */
3730         if (IS_ERR(right))
3731                 return 1;
3732
3733         btrfs_tree_lock(right);
3734         btrfs_set_lock_blocking_write(right);
3735
3736         free_space = btrfs_leaf_free_space(right);
3737         if (free_space < data_size)
3738                 goto out_unlock;
3739
3740         /* cow and double check */
3741         ret = btrfs_cow_block(trans, root, right, upper,
3742                               slot + 1, &right);
3743         if (ret)
3744                 goto out_unlock;
3745
3746         free_space = btrfs_leaf_free_space(right);
3747         if (free_space < data_size)
3748                 goto out_unlock;
3749
3750         left_nritems = btrfs_header_nritems(left);
3751         if (left_nritems == 0)
3752                 goto out_unlock;
3753
3754         if (path->slots[0] == left_nritems && !empty) {
3755                 /* Key greater than all keys in the leaf, right neighbor has
3756                  * enough room for it and we're not emptying our leaf to delete
3757                  * it, therefore use right neighbor to insert the new item and
3758                  * no need to touch/dirty our left leaf. */
3759                 btrfs_tree_unlock(left);
3760                 free_extent_buffer(left);
3761                 path->nodes[0] = right;
3762                 path->slots[0] = 0;
3763                 path->slots[1]++;
3764                 return 0;
3765         }
3766
3767         return __push_leaf_right(path, min_data_size, empty,
3768                                 right, free_space, left_nritems, min_slot);
3769 out_unlock:
3770         btrfs_tree_unlock(right);
3771         free_extent_buffer(right);
3772         return 1;
3773 }
3774
3775 /*
3776  * push some data in the path leaf to the left, trying to free up at
3777  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3778  *
3779  * max_slot can put a limit on how far into the leaf we'll push items.  The
3780  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us do all the
3781  * items
3782  */
3783 static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
3784                                      int empty, struct extent_buffer *left,
3785                                      int free_space, u32 right_nritems,
3786                                      u32 max_slot)
3787 {
3788         struct btrfs_fs_info *fs_info = left->fs_info;
3789         struct btrfs_disk_key disk_key;
3790         struct extent_buffer *right = path->nodes[0];
3791         int i;
3792         int push_space = 0;
3793         int push_items = 0;
3794         struct btrfs_item *item;
3795         u32 old_left_nritems;
3796         u32 nr;
3797         int ret = 0;
3798         u32 this_item_size;
3799         u32 old_left_item_size;
3800         struct btrfs_map_token token;
3801
3802         if (empty)
3803                 nr = min(right_nritems, max_slot);
3804         else
3805                 nr = min(right_nritems - 1, max_slot);
3806
3807         for (i = 0; i < nr; i++) {
3808                 item = btrfs_item_nr(i);
3809
3810                 if (!empty && push_items > 0) {
3811                         if (path->slots[0] < i)
3812                                 break;
3813                         if (path->slots[0] == i) {
3814                                 int space = btrfs_leaf_free_space(right);
3815
3816                                 if (space + push_space * 2 > free_space)
3817                                         break;
3818                         }
3819                 }
3820
3821                 if (path->slots[0] == i)
3822                         push_space += data_size;
3823
3824                 this_item_size = btrfs_item_size(right, item);
3825                 if (this_item_size + sizeof(*item) + push_space > free_space)
3826                         break;
3827
3828                 push_items++;
3829                 push_space += this_item_size + sizeof(*item);
3830         }
3831
3832         if (push_items == 0) {
3833                 ret = 1;
3834                 goto out;
3835         }
3836         WARN_ON(!empty && push_items == btrfs_header_nritems(right));
3837
3838         /* push data from right to left */
3839         copy_extent_buffer(left, right,
3840                            btrfs_item_nr_offset(btrfs_header_nritems(left)),
3841                            btrfs_item_nr_offset(0),
3842                            push_items * sizeof(struct btrfs_item));
3843
3844         push_space = BTRFS_LEAF_DATA_SIZE(fs_info) -
3845                      btrfs_item_offset_nr(right, push_items - 1);
3846
3847         copy_extent_buffer(left, right, BTRFS_LEAF_DATA_OFFSET +
3848                      leaf_data_end(left) - push_space,
3849                      BTRFS_LEAF_DATA_OFFSET +
3850                      btrfs_item_offset_nr(right, push_items - 1),
3851                      push_space);
3852         old_left_nritems = btrfs_header_nritems(left);
3853         BUG_ON(old_left_nritems <= 0);
3854
3855         btrfs_init_map_token(&token, left);
3856         old_left_item_size = btrfs_item_offset_nr(left, old_left_nritems - 1);
3857         for (i = old_left_nritems; i < old_left_nritems + push_items; i++) {
3858                 u32 ioff;
3859
3860                 item = btrfs_item_nr(i);
3861
3862                 ioff = btrfs_token_item_offset(&token, item);
3863                 btrfs_set_token_item_offset(&token, item,
3864                       ioff - (BTRFS_LEAF_DATA_SIZE(fs_info) - old_left_item_size));
3865         }
3866         btrfs_set_header_nritems(left, old_left_nritems + push_items);
3867
3868         /* fixup right node */
3869         if (push_items > right_nritems)
3870                 WARN(1, KERN_CRIT "push items %d nr %u\n", push_items,
3871                        right_nritems);
3872
3873         if (push_items < right_nritems) {
3874                 push_space = btrfs_item_offset_nr(right, push_items - 1) -
3875                                                   leaf_data_end(right);
3876                 memmove_extent_buffer(right, BTRFS_LEAF_DATA_OFFSET +
3877                                       BTRFS_LEAF_DATA_SIZE(fs_info) - push_space,
3878                                       BTRFS_LEAF_DATA_OFFSET +
3879                                       leaf_data_end(right), push_space);
3880
3881                 memmove_extent_buffer(right, btrfs_item_nr_offset(0),
3882                               btrfs_item_nr_offset(push_items),
3883                              (btrfs_header_nritems(right) - push_items) *
3884                              sizeof(struct btrfs_item));
3885         }
3886
3887         btrfs_init_map_token(&token, right);
3888         right_nritems -= push_items;
3889         btrfs_set_header_nritems(right, right_nritems);
3890         push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
3891         for (i = 0; i < right_nritems; i++) {
3892                 item = btrfs_item_nr(i);
3893
3894                 push_space = push_space - btrfs_token_item_size(&token, item);
3895                 btrfs_set_token_item_offset(&token, item, push_space);
3896         }
3897
3898         btrfs_mark_buffer_dirty(left);
3899         if (right_nritems)
3900                 btrfs_mark_buffer_dirty(right);
3901         else
3902                 btrfs_clean_tree_block(right);
3903
3904         btrfs_item_key(right, &disk_key, 0);
3905         fixup_low_keys(path, &disk_key, 1);
3906
3907         /* then fixup the leaf pointer in the path */
3908         if (path->slots[0] < push_items) {
3909                 path->slots[0] += old_left_nritems;
3910                 btrfs_tree_unlock(path->nodes[0]);
3911                 free_extent_buffer(path->nodes[0]);
3912                 path->nodes[0] = left;
3913                 path->slots[1] -= 1;
3914         } else {
3915                 btrfs_tree_unlock(left);
3916                 free_extent_buffer(left);
3917                 path->slots[0] -= push_items;
3918         }
3919         BUG_ON(path->slots[0] < 0);
3920         return ret;
3921 out:
3922         btrfs_tree_unlock(left);
3923         free_extent_buffer(left);
3924         return ret;
3925 }
3926
3927 /*
3928  * push some data in the path leaf to the left, trying to free up at
3929  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3930  *
3931  * max_slot can put a limit on how far into the leaf we'll push items.  The
3932  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us push all the
3933  * items
3934  */
3935 static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
3936                           *root, struct btrfs_path *path, int min_data_size,
3937                           int data_size, int empty, u32 max_slot)
3938 {
3939         struct extent_buffer *right = path->nodes[0];
3940         struct extent_buffer *left;
3941         int slot;
3942         int free_space;
3943         u32 right_nritems;
3944         int ret = 0;
3945
3946         slot = path->slots[1];
3947         if (slot == 0)
3948                 return 1;
3949         if (!path->nodes[1])
3950                 return 1;
3951
3952         right_nritems = btrfs_header_nritems(right);
3953         if (right_nritems == 0)
3954                 return 1;
3955
3956         btrfs_assert_tree_locked(path->nodes[1]);
3957
3958         left = btrfs_read_node_slot(path->nodes[1], slot - 1);
3959         /*
3960          * slot - 1 is not valid or we fail to read the left node,
3961          * no big deal, just return.
3962          */
3963         if (IS_ERR(left))
3964                 return 1;
3965
3966         btrfs_tree_lock(left);
3967         btrfs_set_lock_blocking_write(left);
3968
3969         free_space = btrfs_leaf_free_space(left);
3970         if (free_space < data_size) {
3971                 ret = 1;
3972                 goto out;
3973         }
3974
3975         /* cow and double check */
3976         ret = btrfs_cow_block(trans, root, left,
3977                               path->nodes[1], slot - 1, &left);
3978         if (ret) {
3979                 /* we hit -ENOSPC, but it isn't fatal here */
3980                 if (ret == -ENOSPC)
3981                         ret = 1;
3982                 goto out;
3983         }
3984
3985         free_space = btrfs_leaf_free_space(left);
3986         if (free_space < data_size) {
3987                 ret = 1;
3988                 goto out;
3989         }
3990
3991         return __push_leaf_left(path, min_data_size,
3992                                empty, left, free_space, right_nritems,
3993                                max_slot);
3994 out:
3995         btrfs_tree_unlock(left);
3996         free_extent_buffer(left);
3997         return ret;
3998 }
3999
4000 /*
4001  * split the path's leaf in two, making sure there is at least data_size
4002  * available for the resulting leaf level of the path.
4003  */
4004 static noinline void copy_for_split(struct btrfs_trans_handle *trans,
4005                                     struct btrfs_path *path,
4006                                     struct extent_buffer *l,
4007                                     struct extent_buffer *right,
4008                                     int slot, int mid, int nritems)
4009 {
4010         struct btrfs_fs_info *fs_info = trans->fs_info;
4011         int data_copy_size;
4012         int rt_data_off;
4013         int i;
4014         struct btrfs_disk_key disk_key;
4015         struct btrfs_map_token token;
4016
4017         nritems = nritems - mid;
4018         btrfs_set_header_nritems(right, nritems);
4019         data_copy_size = btrfs_item_end_nr(l, mid) - leaf_data_end(l);
4020
4021         copy_extent_buffer(right, l, btrfs_item_nr_offset(0),
4022                            btrfs_item_nr_offset(mid),
4023                            nritems * sizeof(struct btrfs_item));
4024
4025         copy_extent_buffer(right, l,
4026                      BTRFS_LEAF_DATA_OFFSET + BTRFS_LEAF_DATA_SIZE(fs_info) -
4027                      data_copy_size, BTRFS_LEAF_DATA_OFFSET +
4028                      leaf_data_end(l), data_copy_size);
4029
4030         rt_data_off = BTRFS_LEAF_DATA_SIZE(fs_info) - btrfs_item_end_nr(l, mid);
4031
4032         btrfs_init_map_token(&token, right);
4033         for (i = 0; i < nritems; i++) {
4034                 struct btrfs_item *item = btrfs_item_nr(i);
4035                 u32 ioff;
4036
4037                 ioff = btrfs_token_item_offset(&token, item);
4038                 btrfs_set_token_item_offset(&token, item, ioff + rt_data_off);
4039         }
4040
4041         btrfs_set_header_nritems(l, mid);
4042         btrfs_item_key(right, &disk_key, 0);
4043         insert_ptr(trans, path, &disk_key, right->start, path->slots[1] + 1, 1);
4044
4045         btrfs_mark_buffer_dirty(right);
4046         btrfs_mark_buffer_dirty(l);
4047         BUG_ON(path->slots[0] != slot);
4048
4049         if (mid <= slot) {
4050                 btrfs_tree_unlock(path->nodes[0]);
4051                 free_extent_buffer(path->nodes[0]);
4052                 path->nodes[0] = right;
4053                 path->slots[0] -= mid;
4054                 path->slots[1] += 1;
4055         } else {
4056                 btrfs_tree_unlock(right);
4057                 free_extent_buffer(right);
4058         }
4059
4060         BUG_ON(path->slots[0] < 0);
4061 }
4062
4063 /*
4064  * double splits happen when we need to insert a big item in the middle
4065  * of a leaf.  A double split can leave us with 3 mostly empty leaves:
4066  * leaf: [ slots 0 - N] [ our target ] [ N + 1 - total in leaf ]
4067  *          A                 B                 C
4068  *
4069  * We avoid this by trying to push the items on either side of our target
4070  * into the adjacent leaves.  If all goes well we can avoid the double split
4071  * completely.
4072  */
4073 static noinline int push_for_double_split(struct btrfs_trans_handle *trans,
4074                                           struct btrfs_root *root,
4075                                           struct btrfs_path *path,
4076                                           int data_size)
4077 {
4078         int ret;
4079         int progress = 0;
4080         int slot;
4081         u32 nritems;
4082         int space_needed = data_size;
4083
4084         slot = path->slots[0];
4085         if (slot < btrfs_header_nritems(path->nodes[0]))
4086                 space_needed -= btrfs_leaf_free_space(path->nodes[0]);
4087
4088         /*
4089          * try to push all the items after our slot into the
4090          * right leaf
4091          */
4092         ret = push_leaf_right(trans, root, path, 1, space_needed, 0, slot);
4093         if (ret < 0)
4094                 return ret;
4095
4096         if (ret == 0)
4097                 progress++;
4098
4099         nritems = btrfs_header_nritems(path->nodes[0]);
4100         /*
4101          * our goal is to get our slot at the start or end of a leaf.  If
4102          * we've done so we're done
4103          */
4104         if (path->slots[0] == 0 || path->slots[0] == nritems)
4105                 return 0;
4106
4107         if (btrfs_leaf_free_space(path->nodes[0]) >= data_size)
4108                 return 0;
4109
4110         /* try to push all the items before our slot into the next leaf */
4111         slot = path->slots[0];
4112         space_needed = data_size;
4113         if (slot > 0)
4114                 space_needed -= btrfs_leaf_free_space(path->nodes[0]);
4115         ret = push_leaf_left(trans, root, path, 1, space_needed, 0, slot);
4116         if (ret < 0)
4117                 return ret;
4118
4119         if (ret == 0)
4120                 progress++;
4121
4122         if (progress)
4123                 return 0;
4124         return 1;
4125 }
4126
4127 /*
4128  * split the path's leaf in two, making sure there is at least data_size
4129  * available for the resulting leaf level of the path.
4130  *
4131  * returns 0 if all went well and < 0 on failure.
4132  */
4133 static noinline int split_leaf(struct btrfs_trans_handle *trans,
4134                                struct btrfs_root *root,
4135                                const struct btrfs_key *ins_key,
4136                                struct btrfs_path *path, int data_size,
4137                                int extend)
4138 {
4139         struct btrfs_disk_key disk_key;
4140         struct extent_buffer *l;
4141         u32 nritems;
4142         int mid;
4143         int slot;
4144         struct extent_buffer *right;
4145         struct btrfs_fs_info *fs_info = root->fs_info;
4146         int ret = 0;
4147         int wret;
4148         int split;
4149         int num_doubles = 0;
4150         int tried_avoid_double = 0;
4151
4152         l = path->nodes[0];
4153         slot = path->slots[0];
4154         if (extend && data_size + btrfs_item_size_nr(l, slot) +
4155             sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(fs_info))
4156                 return -EOVERFLOW;
4157
4158         /* first try to make some room by pushing left and right */
4159         if (data_size && path->nodes[1]) {
4160                 int space_needed = data_size;
4161
4162                 if (slot < btrfs_header_nritems(l))
4163                         space_needed -= btrfs_leaf_free_space(l);
4164
4165                 wret = push_leaf_right(trans, root, path, space_needed,
4166                                        space_needed, 0, 0);
4167                 if (wret < 0)
4168                         return wret;
4169                 if (wret) {
4170                         space_needed = data_size;
4171                         if (slot > 0)
4172                                 space_needed -= btrfs_leaf_free_space(l);
4173                         wret = push_leaf_left(trans, root, path, space_needed,
4174                                               space_needed, 0, (u32)-1);
4175                         if (wret < 0)
4176                                 return wret;
4177                 }
4178                 l = path->nodes[0];
4179
4180                 /* did the pushes work? */
4181                 if (btrfs_leaf_free_space(l) >= data_size)
4182                         return 0;
4183         }
4184
4185         if (!path->nodes[1]) {
4186                 ret = insert_new_root(trans, root, path, 1);
4187                 if (ret)
4188                         return ret;
4189         }
4190 again:
4191         split = 1;
4192         l = path->nodes[0];
4193         slot = path->slots[0];
4194         nritems = btrfs_header_nritems(l);
4195         mid = (nritems + 1) / 2;
4196
4197         if (mid <= slot) {
4198                 if (nritems == 1 ||
4199                     leaf_space_used(l, mid, nritems - mid) + data_size >
4200                         BTRFS_LEAF_DATA_SIZE(fs_info)) {
4201                         if (slot >= nritems) {
4202                                 split = 0;
4203                         } else {
4204                                 mid = slot;
4205                                 if (mid != nritems &&
4206                                     leaf_space_used(l, mid, nritems - mid) +
4207                                     data_size > BTRFS_LEAF_DATA_SIZE(fs_info)) {
4208                                         if (data_size && !tried_avoid_double)
4209                                                 goto push_for_double;
4210                                         split = 2;
4211                                 }
4212                         }
4213                 }
4214         } else {
4215                 if (leaf_space_used(l, 0, mid) + data_size >
4216                         BTRFS_LEAF_DATA_SIZE(fs_info)) {
4217                         if (!extend && data_size && slot == 0) {
4218                                 split = 0;
4219                         } else if ((extend || !data_size) && slot == 0) {
4220                                 mid = 1;
4221                         } else {
4222                                 mid = slot;
4223                                 if (mid != nritems &&
4224                                     leaf_space_used(l, mid, nritems - mid) +
4225                                     data_size > BTRFS_LEAF_DATA_SIZE(fs_info)) {
4226                                         if (data_size && !tried_avoid_double)
4227                                                 goto push_for_double;
4228                                         split = 2;
4229                                 }
4230                         }
4231                 }
4232         }
4233
4234         if (split == 0)
4235                 btrfs_cpu_key_to_disk(&disk_key, ins_key);
4236         else
4237                 btrfs_item_key(l, &disk_key, mid);
4238
4239         right = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, 0,
4240                                              l->start, 0);
4241         if (IS_ERR(right))
4242                 return PTR_ERR(right);
4243
4244         root_add_used(root, fs_info->nodesize);
4245
4246         if (split == 0) {
4247                 if (mid <= slot) {
4248                         btrfs_set_header_nritems(right, 0);
4249                         insert_ptr(trans, path, &disk_key,
4250                                    right->start, path->slots[1] + 1, 1);
4251                         btrfs_tree_unlock(path->nodes[0]);
4252                         free_extent_buffer(path->nodes[0]);
4253                         path->nodes[0] = right;
4254                         path->slots[0] = 0;
4255                         path->slots[1] += 1;
4256                 } else {
4257                         btrfs_set_header_nritems(right, 0);
4258                         insert_ptr(trans, path, &disk_key,
4259                                    right->start, path->slots[1], 1);
4260                         btrfs_tree_unlock(path->nodes[0]);
4261                         free_extent_buffer(path->nodes[0]);
4262                         path->nodes[0] = right;
4263                         path->slots[0] = 0;
4264                         if (path->slots[1] == 0)
4265                                 fixup_low_keys(path, &disk_key, 1);
4266                 }
4267                 /*
4268                  * We create a new leaf 'right' for the required ins_len and
4269                  * we'll do btrfs_mark_buffer_dirty() on this leaf after copying
4270                  * the content of ins_len to 'right'.
4271                  */
4272                 return ret;
4273         }
4274
4275         copy_for_split(trans, path, l, right, slot, mid, nritems);
4276
4277         if (split == 2) {
4278                 BUG_ON(num_doubles != 0);
4279                 num_doubles++;
4280                 goto again;
4281         }
4282
4283         return 0;
4284
4285 push_for_double:
4286         push_for_double_split(trans, root, path, data_size);
4287         tried_avoid_double = 1;
4288         if (btrfs_leaf_free_space(path->nodes[0]) >= data_size)
4289                 return 0;
4290         goto again;
4291 }
4292
4293 static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
4294                                          struct btrfs_root *root,
4295                                          struct btrfs_path *path, int ins_len)
4296 {
4297         struct btrfs_key key;
4298         struct extent_buffer *leaf;
4299         struct btrfs_file_extent_item *fi;
4300         u64 extent_len = 0;
4301         u32 item_size;
4302         int ret;
4303
4304         leaf = path->nodes[0];
4305         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
4306
4307         BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY &&
4308                key.type != BTRFS_EXTENT_CSUM_KEY);
4309
4310         if (btrfs_leaf_free_space(leaf) >= ins_len)
4311                 return 0;
4312
4313         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
4314         if (key.type == BTRFS_EXTENT_DATA_KEY) {
4315                 fi = btrfs_item_ptr(leaf, path->slots[0],
4316                                     struct btrfs_file_extent_item);
4317                 extent_len = btrfs_file_extent_num_bytes(leaf, fi);
4318         }
4319         btrfs_release_path(path);
4320
4321         path->keep_locks = 1;
4322         path->search_for_split = 1;
4323         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
4324         path->search_for_split = 0;
4325         if (ret > 0)
4326                 ret = -EAGAIN;
4327         if (ret < 0)
4328                 goto err;
4329
4330         ret = -EAGAIN;
4331         leaf = path->nodes[0];
4332         /* if our item isn't there, return now */
4333         if (item_size != btrfs_item_size_nr(leaf, path->slots[0]))
4334                 goto err;
4335
4336         /* the leaf has  changed, it now has room.  return now */
4337         if (btrfs_leaf_free_space(path->nodes[0]) >= ins_len)
4338                 goto err;
4339
4340         if (key.type == BTRFS_EXTENT_DATA_KEY) {
4341                 fi = btrfs_item_ptr(leaf, path->slots[0],
4342                                     struct btrfs_file_extent_item);
4343                 if (extent_len != btrfs_file_extent_num_bytes(leaf, fi))
4344                         goto err;
4345         }
4346
4347         btrfs_set_path_blocking(path);
4348         ret = split_leaf(trans, root, &key, path, ins_len, 1);
4349         if (ret)
4350                 goto err;
4351
4352         path->keep_locks = 0;
4353         btrfs_unlock_up_safe(path, 1);
4354         return 0;
4355 err:
4356         path->keep_locks = 0;
4357         return ret;
4358 }
4359
4360 static noinline int split_item(struct btrfs_path *path,
4361                                const struct btrfs_key *new_key,
4362                                unsigned long split_offset)
4363 {
4364         struct extent_buffer *leaf;
4365         struct btrfs_item *item;
4366         struct btrfs_item *new_item;
4367         int slot;
4368         char *buf;
4369         u32 nritems;
4370         u32 item_size;
4371         u32 orig_offset;
4372         struct btrfs_disk_key disk_key;
4373
4374         leaf = path->nodes[0];
4375         BUG_ON(btrfs_leaf_free_space(leaf) < sizeof(struct btrfs_item));
4376
4377         btrfs_set_path_blocking(path);
4378
4379         item = btrfs_item_nr(path->slots[0]);
4380         orig_offset = btrfs_item_offset(leaf, item);
4381         item_size = btrfs_item_size(leaf, item);
4382
4383         buf = kmalloc(item_size, GFP_NOFS);
4384         if (!buf)
4385                 return -ENOMEM;
4386
4387         read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf,
4388                             path->slots[0]), item_size);
4389
4390         slot = path->slots[0] + 1;
4391         nritems = btrfs_header_nritems(leaf);
4392         if (slot != nritems) {
4393                 /* shift the items */
4394                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1),
4395                                 btrfs_item_nr_offset(slot),
4396                                 (nritems - slot) * sizeof(struct btrfs_item));
4397         }
4398
4399         btrfs_cpu_key_to_disk(&disk_key, new_key);
4400         btrfs_set_item_key(leaf, &disk_key, slot);
4401
4402         new_item = btrfs_item_nr(slot);
4403
4404         btrfs_set_item_offset(leaf, new_item, orig_offset);
4405         btrfs_set_item_size(leaf, new_item, item_size - split_offset);
4406
4407         btrfs_set_item_offset(leaf, item,
4408                               orig_offset + item_size - split_offset);
4409         btrfs_set_item_size(leaf, item, split_offset);
4410
4411         btrfs_set_header_nritems(leaf, nritems + 1);
4412
4413         /* write the data for the start of the original item */
4414         write_extent_buffer(leaf, buf,
4415                             btrfs_item_ptr_offset(leaf, path->slots[0]),
4416                             split_offset);
4417
4418         /* write the data for the new item */
4419         write_extent_buffer(leaf, buf + split_offset,
4420                             btrfs_item_ptr_offset(leaf, slot),
4421                             item_size - split_offset);
4422         btrfs_mark_buffer_dirty(leaf);
4423
4424         BUG_ON(btrfs_leaf_free_space(leaf) < 0);
4425         kfree(buf);
4426         return 0;
4427 }
4428
4429 /*
4430  * This function splits a single item into two items,
4431  * giving 'new_key' to the new item and splitting the
4432  * old one at split_offset (from the start of the item).
4433  *
4434  * The path may be released by this operation.  After
4435  * the split, the path is pointing to the old item.  The
4436  * new item is going to be in the same node as the old one.
4437  *
4438  * Note, the item being split must be smaller enough to live alone on
4439  * a tree block with room for one extra struct btrfs_item
4440  *
4441  * This allows us to split the item in place, keeping a lock on the
4442  * leaf the entire time.
4443  */
4444 int btrfs_split_item(struct btrfs_trans_handle *trans,
4445                      struct btrfs_root *root,
4446                      struct btrfs_path *path,
4447                      const struct btrfs_key *new_key,
4448                      unsigned long split_offset)
4449 {
4450         int ret;
4451         ret = setup_leaf_for_split(trans, root, path,
4452                                    sizeof(struct btrfs_item));
4453         if (ret)
4454                 return ret;
4455
4456         ret = split_item(path, new_key, split_offset);
4457         return ret;
4458 }
4459
4460 /*
4461  * This function duplicate a item, giving 'new_key' to the new item.
4462  * It guarantees both items live in the same tree leaf and the new item
4463  * is contiguous with the original item.
4464  *
4465  * This allows us to split file extent in place, keeping a lock on the
4466  * leaf the entire time.
4467  */
4468 int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
4469                          struct btrfs_root *root,
4470                          struct btrfs_path *path,
4471                          const struct btrfs_key *new_key)
4472 {
4473         struct extent_buffer *leaf;
4474         int ret;
4475         u32 item_size;
4476
4477         leaf = path->nodes[0];
4478         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
4479         ret = setup_leaf_for_split(trans, root, path,
4480                                    item_size + sizeof(struct btrfs_item));
4481         if (ret)
4482                 return ret;
4483
4484         path->slots[0]++;
4485         setup_items_for_insert(root, path, new_key, &item_size,
4486                                item_size, item_size +
4487                                sizeof(struct btrfs_item), 1);
4488         leaf = path->nodes[0];
4489         memcpy_extent_buffer(leaf,
4490                              btrfs_item_ptr_offset(leaf, path->slots[0]),
4491                              btrfs_item_ptr_offset(leaf, path->slots[0] - 1),
4492                              item_size);
4493         return 0;
4494 }
4495
4496 /*
4497  * make the item pointed to by the path smaller.  new_size indicates
4498  * how small to make it, and from_end tells us if we just chop bytes
4499  * off the end of the item or if we shift the item to chop bytes off
4500  * the front.
4501  */
4502 void btrfs_truncate_item(struct btrfs_path *path, u32 new_size, int from_end)
4503 {
4504         int slot;
4505         struct extent_buffer *leaf;
4506         struct btrfs_item *item;
4507         u32 nritems;
4508         unsigned int data_end;
4509         unsigned int old_data_start;
4510         unsigned int old_size;
4511         unsigned int size_diff;
4512         int i;
4513         struct btrfs_map_token token;
4514
4515         leaf = path->nodes[0];
4516         slot = path->slots[0];
4517
4518         old_size = btrfs_item_size_nr(leaf, slot);
4519         if (old_size == new_size)
4520                 return;
4521
4522         nritems = btrfs_header_nritems(leaf);
4523         data_end = leaf_data_end(leaf);
4524
4525         old_data_start = btrfs_item_offset_nr(leaf, slot);
4526
4527         size_diff = old_size - new_size;
4528
4529         BUG_ON(slot < 0);
4530         BUG_ON(slot >= nritems);
4531
4532         /*
4533          * item0..itemN ... dataN.offset..dataN.size .. data0.size
4534          */
4535         /* first correct the data pointers */
4536         btrfs_init_map_token(&token, leaf);
4537         for (i = slot; i < nritems; i++) {
4538                 u32 ioff;
4539                 item = btrfs_item_nr(i);
4540
4541                 ioff = btrfs_token_item_offset(&token, item);
4542                 btrfs_set_token_item_offset(&token, item, ioff + size_diff);
4543         }
4544
4545         /* shift the data */
4546         if (from_end) {
4547                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4548                               data_end + size_diff, BTRFS_LEAF_DATA_OFFSET +
4549                               data_end, old_data_start + new_size - data_end);
4550         } else {
4551                 struct btrfs_disk_key disk_key;
4552                 u64 offset;
4553
4554                 btrfs_item_key(leaf, &disk_key, slot);
4555
4556                 if (btrfs_disk_key_type(&disk_key) == BTRFS_EXTENT_DATA_KEY) {
4557                         unsigned long ptr;
4558                         struct btrfs_file_extent_item *fi;
4559
4560                         fi = btrfs_item_ptr(leaf, slot,
4561                                             struct btrfs_file_extent_item);
4562                         fi = (struct btrfs_file_extent_item *)(
4563                              (unsigned long)fi - size_diff);
4564
4565                         if (btrfs_file_extent_type(leaf, fi) ==
4566                             BTRFS_FILE_EXTENT_INLINE) {
4567                                 ptr = btrfs_item_ptr_offset(leaf, slot);
4568                                 memmove_extent_buffer(leaf, ptr,
4569                                       (unsigned long)fi,
4570                                       BTRFS_FILE_EXTENT_INLINE_DATA_START);
4571                         }
4572                 }
4573
4574                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4575                               data_end + size_diff, BTRFS_LEAF_DATA_OFFSET +
4576                               data_end, old_data_start - data_end);
4577
4578                 offset = btrfs_disk_key_offset(&disk_key);
4579                 btrfs_set_disk_key_offset(&disk_key, offset + size_diff);
4580                 btrfs_set_item_key(leaf, &disk_key, slot);
4581                 if (slot == 0)
4582                         fixup_low_keys(path, &disk_key, 1);
4583         }
4584
4585         item = btrfs_item_nr(slot);
4586         btrfs_set_item_size(leaf, item, new_size);
4587         btrfs_mark_buffer_dirty(leaf);
4588
4589         if (btrfs_leaf_free_space(leaf) < 0) {
4590                 btrfs_print_leaf(leaf);
4591                 BUG();
4592         }
4593 }
4594
4595 /*
4596  * make the item pointed to by the path bigger, data_size is the added size.
4597  */
4598 void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
4599 {
4600         int slot;
4601         struct extent_buffer *leaf;
4602         struct btrfs_item *item;
4603         u32 nritems;
4604         unsigned int data_end;
4605         unsigned int old_data;
4606         unsigned int old_size;
4607         int i;
4608         struct btrfs_map_token token;
4609
4610         leaf = path->nodes[0];
4611
4612         nritems = btrfs_header_nritems(leaf);
4613         data_end = leaf_data_end(leaf);
4614
4615         if (btrfs_leaf_free_space(leaf) < data_size) {
4616                 btrfs_print_leaf(leaf);
4617                 BUG();
4618         }
4619         slot = path->slots[0];
4620         old_data = btrfs_item_end_nr(leaf, slot);
4621
4622         BUG_ON(slot < 0);
4623         if (slot >= nritems) {
4624                 btrfs_print_leaf(leaf);
4625                 btrfs_crit(leaf->fs_info, "slot %d too large, nritems %d",
4626                            slot, nritems);
4627                 BUG();
4628         }
4629
4630         /*
4631          * item0..itemN ... dataN.offset..dataN.size .. data0.size
4632          */
4633         /* first correct the data pointers */
4634         btrfs_init_map_token(&token, leaf);
4635         for (i = slot; i < nritems; i++) {
4636                 u32 ioff;
4637                 item = btrfs_item_nr(i);
4638
4639                 ioff = btrfs_token_item_offset(&token, item);
4640                 btrfs_set_token_item_offset(&token, item, ioff - data_size);
4641         }
4642
4643         /* shift the data */
4644         memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4645                       data_end - data_size, BTRFS_LEAF_DATA_OFFSET +
4646                       data_end, old_data - data_end);
4647
4648         data_end = old_data;
4649         old_size = btrfs_item_size_nr(leaf, slot);
4650         item = btrfs_item_nr(slot);
4651         btrfs_set_item_size(leaf, item, old_size + data_size);
4652         btrfs_mark_buffer_dirty(leaf);
4653
4654         if (btrfs_leaf_free_space(leaf) < 0) {
4655                 btrfs_print_leaf(leaf);
4656                 BUG();
4657         }
4658 }
4659
4660 /*
4661  * this is a helper for btrfs_insert_empty_items, the main goal here is
4662  * to save stack depth by doing the bulk of the work in a function
4663  * that doesn't call btrfs_search_slot
4664  */
4665 void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
4666                             const struct btrfs_key *cpu_key, u32 *data_size,
4667                             u32 total_data, u32 total_size, int nr)
4668 {
4669         struct btrfs_fs_info *fs_info = root->fs_info;
4670         struct btrfs_item *item;
4671         int i;
4672         u32 nritems;
4673         unsigned int data_end;
4674         struct btrfs_disk_key disk_key;
4675         struct extent_buffer *leaf;
4676         int slot;
4677         struct btrfs_map_token token;
4678
4679         if (path->slots[0] == 0) {
4680                 btrfs_cpu_key_to_disk(&disk_key, cpu_key);
4681                 fixup_low_keys(path, &disk_key, 1);
4682         }
4683         btrfs_unlock_up_safe(path, 1);
4684
4685         leaf = path->nodes[0];
4686         slot = path->slots[0];
4687
4688         nritems = btrfs_header_nritems(leaf);
4689         data_end = leaf_data_end(leaf);
4690
4691         if (btrfs_leaf_free_space(leaf) < total_size) {
4692                 btrfs_print_leaf(leaf);
4693                 btrfs_crit(fs_info, "not enough freespace need %u have %d",
4694                            total_size, btrfs_leaf_free_space(leaf));
4695                 BUG();
4696         }
4697
4698         btrfs_init_map_token(&token, leaf);
4699         if (slot != nritems) {
4700                 unsigned int old_data = btrfs_item_end_nr(leaf, slot);
4701
4702                 if (old_data < data_end) {
4703                         btrfs_print_leaf(leaf);
4704                         btrfs_crit(fs_info, "slot %d old_data %d data_end %d",
4705                                    slot, old_data, data_end);
4706                         BUG();
4707                 }
4708                 /*
4709                  * item0..itemN ... dataN.offset..dataN.size .. data0.size
4710                  */
4711                 /* first correct the data pointers */
4712                 for (i = slot; i < nritems; i++) {
4713                         u32 ioff;
4714
4715                         item = btrfs_item_nr(i);
4716                         ioff = btrfs_token_item_offset(&token, item);
4717                         btrfs_set_token_item_offset(&token, item,
4718                                                     ioff - total_data);
4719                 }
4720                 /* shift the items */
4721                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
4722                               btrfs_item_nr_offset(slot),
4723                               (nritems - slot) * sizeof(struct btrfs_item));
4724
4725                 /* shift the data */
4726                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4727                               data_end - total_data, BTRFS_LEAF_DATA_OFFSET +
4728                               data_end, old_data - data_end);
4729                 data_end = old_data;
4730         }
4731
4732         /* setup the item for the new data */
4733         for (i = 0; i < nr; i++) {
4734                 btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
4735                 btrfs_set_item_key(leaf, &disk_key, slot + i);
4736                 item = btrfs_item_nr(slot + i);
4737                 btrfs_set_token_item_offset(&token, item, data_end - data_size[i]);
4738                 data_end -= data_size[i];
4739                 btrfs_set_token_item_size(&token, item, data_size[i]);
4740         }
4741
4742         btrfs_set_header_nritems(leaf, nritems + nr);
4743         btrfs_mark_buffer_dirty(leaf);
4744
4745         if (btrfs_leaf_free_space(leaf) < 0) {
4746                 btrfs_print_leaf(leaf);
4747                 BUG();
4748         }
4749 }
4750
4751 /*
4752  * Given a key and some data, insert items into the tree.
4753  * This does all the path init required, making room in the tree if needed.
4754  */
4755 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
4756                             struct btrfs_root *root,
4757                             struct btrfs_path *path,
4758                             const struct btrfs_key *cpu_key, u32 *data_size,
4759                             int nr)
4760 {
4761         int ret = 0;
4762         int slot;
4763         int i;
4764         u32 total_size = 0;
4765         u32 total_data = 0;
4766
4767         for (i = 0; i < nr; i++)
4768                 total_data += data_size[i];
4769
4770         total_size = total_data + (nr * sizeof(struct btrfs_item));
4771         ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
4772         if (ret == 0)
4773                 return -EEXIST;
4774         if (ret < 0)
4775                 return ret;
4776
4777         slot = path->slots[0];
4778         BUG_ON(slot < 0);
4779
4780         setup_items_for_insert(root, path, cpu_key, data_size,
4781                                total_data, total_size, nr);
4782         return 0;
4783 }
4784
4785 /*
4786  * Given a key and some data, insert an item into the tree.
4787  * This does all the path init required, making room in the tree if needed.
4788  */
4789 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4790                       const struct btrfs_key *cpu_key, void *data,
4791                       u32 data_size)
4792 {
4793         int ret = 0;
4794         struct btrfs_path *path;
4795         struct extent_buffer *leaf;
4796         unsigned long ptr;
4797
4798         path = btrfs_alloc_path();
4799         if (!path)
4800                 return -ENOMEM;
4801         ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size);
4802         if (!ret) {
4803                 leaf = path->nodes[0];
4804                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
4805                 write_extent_buffer(leaf, data, ptr, data_size);
4806                 btrfs_mark_buffer_dirty(leaf);
4807         }
4808         btrfs_free_path(path);
4809         return ret;
4810 }
4811
4812 /*
4813  * delete the pointer from a given node.
4814  *
4815  * the tree should have been previously balanced so the deletion does not
4816  * empty a node.
4817  */
4818 static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
4819                     int level, int slot)
4820 {
4821         struct extent_buffer *parent = path->nodes[level];
4822         u32 nritems;
4823         int ret;
4824
4825         nritems = btrfs_header_nritems(parent);
4826         if (slot != nritems - 1) {
4827                 if (level) {
4828                         ret = tree_mod_log_insert_move(parent, slot, slot + 1,
4829                                         nritems - slot - 1);
4830                         BUG_ON(ret < 0);
4831                 }
4832                 memmove_extent_buffer(parent,
4833                               btrfs_node_key_ptr_offset(slot),
4834                               btrfs_node_key_ptr_offset(slot + 1),
4835                               sizeof(struct btrfs_key_ptr) *
4836                               (nritems - slot - 1));
4837         } else if (level) {
4838                 ret = tree_mod_log_insert_key(parent, slot, MOD_LOG_KEY_REMOVE,
4839                                 GFP_NOFS);
4840                 BUG_ON(ret < 0);
4841         }
4842
4843         nritems--;
4844         btrfs_set_header_nritems(parent, nritems);
4845         if (nritems == 0 && parent == root->node) {
4846                 BUG_ON(btrfs_header_level(root->node) != 1);
4847                 /* just turn the root into a leaf and break */
4848                 btrfs_set_header_level(root->node, 0);
4849         } else if (slot == 0) {
4850                 struct btrfs_disk_key disk_key;
4851
4852                 btrfs_node_key(parent, &disk_key, 0);
4853                 fixup_low_keys(path, &disk_key, level + 1);
4854         }
4855         btrfs_mark_buffer_dirty(parent);
4856 }
4857
4858 /*
4859  * a helper function to delete the leaf pointed to by path->slots[1] and
4860  * path->nodes[1].
4861  *
4862  * This deletes the pointer in path->nodes[1] and frees the leaf
4863  * block extent.  zero is returned if it all worked out, < 0 otherwise.
4864  *
4865  * The path must have already been setup for deleting the leaf, including
4866  * all the proper balancing.  path->nodes[1] must be locked.
4867  */
4868 static noinline void btrfs_del_leaf(struct btrfs_trans_handle *trans,
4869                                     struct btrfs_root *root,
4870                                     struct btrfs_path *path,
4871                                     struct extent_buffer *leaf)
4872 {
4873         WARN_ON(btrfs_header_generation(leaf) != trans->transid);
4874         del_ptr(root, path, 1, path->slots[1]);
4875
4876         /*
4877          * btrfs_free_extent is expensive, we want to make sure we
4878          * aren't holding any locks when we call it
4879          */
4880         btrfs_unlock_up_safe(path, 0);
4881
4882         root_sub_used(root, leaf->len);
4883
4884         atomic_inc(&leaf->refs);
4885         btrfs_free_tree_block(trans, root, leaf, 0, 1);
4886         free_extent_buffer_stale(leaf);
4887 }
4888 /*
4889  * delete the item at the leaf level in path.  If that empties
4890  * the leaf, remove it from the tree
4891  */
4892 int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4893                     struct btrfs_path *path, int slot, int nr)
4894 {
4895         struct btrfs_fs_info *fs_info = root->fs_info;
4896         struct extent_buffer *leaf;
4897         struct btrfs_item *item;
4898         u32 last_off;
4899         u32 dsize = 0;
4900         int ret = 0;
4901         int wret;
4902         int i;
4903         u32 nritems;
4904
4905         leaf = path->nodes[0];
4906         last_off = btrfs_item_offset_nr(leaf, slot + nr - 1);
4907
4908         for (i = 0; i < nr; i++)
4909                 dsize += btrfs_item_size_nr(leaf, slot + i);
4910
4911         nritems = btrfs_header_nritems(leaf);
4912
4913         if (slot + nr != nritems) {
4914                 int data_end = leaf_data_end(leaf);
4915                 struct btrfs_map_token token;
4916
4917                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4918                               data_end + dsize,
4919                               BTRFS_LEAF_DATA_OFFSET + data_end,
4920                               last_off - data_end);
4921
4922                 btrfs_init_map_token(&token, leaf);
4923                 for (i = slot + nr; i < nritems; i++) {
4924                         u32 ioff;
4925
4926                         item = btrfs_item_nr(i);
4927                         ioff = btrfs_token_item_offset(&token, item);
4928                         btrfs_set_token_item_offset(&token, item, ioff + dsize);
4929                 }
4930
4931                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot),
4932                               btrfs_item_nr_offset(slot + nr),
4933                               sizeof(struct btrfs_item) *
4934                               (nritems - slot - nr));
4935         }
4936         btrfs_set_header_nritems(leaf, nritems - nr);
4937         nritems -= nr;
4938
4939         /* delete the leaf if we've emptied it */
4940         if (nritems == 0) {
4941                 if (leaf == root->node) {
4942                         btrfs_set_header_level(leaf, 0);
4943                 } else {
4944                         btrfs_set_path_blocking(path);
4945                         btrfs_clean_tree_block(leaf);
4946                         btrfs_del_leaf(trans, root, path, leaf);
4947                 }
4948         } else {
4949                 int used = leaf_space_used(leaf, 0, nritems);
4950                 if (slot == 0) {
4951                         struct btrfs_disk_key disk_key;
4952
4953                         btrfs_item_key(leaf, &disk_key, 0);
4954                         fixup_low_keys(path, &disk_key, 1);
4955                 }
4956
4957                 /* delete the leaf if it is mostly empty */
4958                 if (used < BTRFS_LEAF_DATA_SIZE(fs_info) / 3) {
4959                         /* push_leaf_left fixes the path.
4960                          * make sure the path still points to our leaf
4961                          * for possible call to del_ptr below
4962                          */
4963                         slot = path->slots[1];
4964                         atomic_inc(&leaf->refs);
4965
4966                         btrfs_set_path_blocking(path);
4967                         wret = push_leaf_left(trans, root, path, 1, 1,
4968                                               1, (u32)-1);
4969                         if (wret < 0 && wret != -ENOSPC)
4970                                 ret = wret;
4971
4972                         if (path->nodes[0] == leaf &&
4973                             btrfs_header_nritems(leaf)) {
4974                                 wret = push_leaf_right(trans, root, path, 1,
4975                                                        1, 1, 0);
4976                                 if (wret < 0 && wret != -ENOSPC)
4977                                         ret = wret;
4978                         }
4979
4980                         if (btrfs_header_nritems(leaf) == 0) {
4981                                 path->slots[1] = slot;
4982                                 btrfs_del_leaf(trans, root, path, leaf);
4983                                 free_extent_buffer(leaf);
4984                                 ret = 0;
4985                         } else {
4986                                 /* if we're still in the path, make sure
4987                                  * we're dirty.  Otherwise, one of the
4988                                  * push_leaf functions must have already
4989                                  * dirtied this buffer
4990                                  */
4991                                 if (path->nodes[0] == leaf)
4992                                         btrfs_mark_buffer_dirty(leaf);
4993                                 free_extent_buffer(leaf);
4994                         }
4995                 } else {
4996                         btrfs_mark_buffer_dirty(leaf);
4997                 }
4998         }
4999         return ret;
5000 }
5001
5002 /*
5003  * search the tree again to find a leaf with lesser keys
5004  * returns 0 if it found something or 1 if there are no lesser leaves.
5005  * returns < 0 on io errors.
5006  *
5007  * This may release the path, and so you may lose any locks held at the
5008  * time you call it.
5009  */
5010 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path)
5011 {
5012         struct btrfs_key key;
5013         struct btrfs_disk_key found_key;
5014         int ret;
5015
5016         btrfs_item_key_to_cpu(path->nodes[0], &key, 0);
5017
5018         if (key.offset > 0) {
5019                 key.offset--;
5020         } else if (key.type > 0) {
5021                 key.type--;
5022                 key.offset = (u64)-1;
5023         } else if (key.objectid > 0) {
5024                 key.objectid--;
5025                 key.type = (u8)-1;
5026                 key.offset = (u64)-1;
5027         } else {
5028                 return 1;
5029         }
5030
5031         btrfs_release_path(path);
5032         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5033         if (ret < 0)
5034                 return ret;
5035         btrfs_item_key(path->nodes[0], &found_key, 0);
5036         ret = comp_keys(&found_key, &key);
5037         /*
5038          * We might have had an item with the previous key in the tree right
5039          * before we released our path. And after we released our path, that
5040          * item might have been pushed to the first slot (0) of the leaf we
5041          * were holding due to a tree balance. Alternatively, an item with the
5042          * previous key can exist as the only element of a leaf (big fat item).
5043          * Therefore account for these 2 cases, so that our callers (like
5044          * btrfs_previous_item) don't miss an existing item with a key matching
5045          * the previous key we computed above.
5046          */
5047         if (ret <= 0)
5048                 return 0;
5049         return 1;
5050 }
5051
5052 /*
5053  * A helper function to walk down the tree starting at min_key, and looking
5054  * for nodes or leaves that are have a minimum transaction id.
5055  * This is used by the btree defrag code, and tree logging
5056  *
5057  * This does not cow, but it does stuff the starting key it finds back
5058  * into min_key, so you can call btrfs_search_slot with cow=1 on the
5059  * key and get a writable path.
5060  *
5061  * This honors path->lowest_level to prevent descent past a given level
5062  * of the tree.
5063  *
5064  * min_trans indicates the oldest transaction that you are interested
5065  * in walking through.  Any nodes or leaves older than min_trans are
5066  * skipped over (without reading them).
5067  *
5068  * returns zero if something useful was found, < 0 on error and 1 if there
5069  * was nothing in the tree that matched the search criteria.
5070  */
5071 int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
5072                          struct btrfs_path *path,
5073                          u64 min_trans)
5074 {
5075         struct extent_buffer *cur;
5076         struct btrfs_key found_key;
5077         int slot;
5078         int sret;
5079         u32 nritems;
5080         int level;
5081         int ret = 1;
5082         int keep_locks = path->keep_locks;
5083
5084         path->keep_locks = 1;
5085 again:
5086         cur = btrfs_read_lock_root_node(root);
5087         level = btrfs_header_level(cur);
5088         WARN_ON(path->nodes[level]);
5089         path->nodes[level] = cur;
5090         path->locks[level] = BTRFS_READ_LOCK;
5091
5092         if (btrfs_header_generation(cur) < min_trans) {
5093                 ret = 1;
5094                 goto out;
5095         }
5096         while (1) {
5097                 nritems = btrfs_header_nritems(cur);
5098                 level = btrfs_header_level(cur);
5099                 sret = btrfs_bin_search(cur, min_key, &slot);
5100                 if (sret < 0) {
5101                         ret = sret;
5102                         goto out;
5103                 }
5104
5105                 /* at the lowest level, we're done, setup the path and exit */
5106                 if (level == path->lowest_level) {
5107                         if (slot >= nritems)
5108                                 goto find_next_key;
5109                         ret = 0;
5110                         path->slots[level] = slot;
5111                         btrfs_item_key_to_cpu(cur, &found_key, slot);
5112                         goto out;
5113                 }
5114                 if (sret && slot > 0)
5115                         slot--;
5116                 /*
5117                  * check this node pointer against the min_trans parameters.
5118                  * If it is too old, old, skip to the next one.
5119                  */
5120                 while (slot < nritems) {
5121                         u64 gen;
5122
5123                         gen = btrfs_node_ptr_generation(cur, slot);
5124                         if (gen < min_trans) {
5125                                 slot++;
5126                                 continue;
5127                         }
5128                         break;
5129                 }
5130 find_next_key:
5131                 /*
5132                  * we didn't find a candidate key in this node, walk forward
5133                  * and find another one
5134                  */
5135                 if (slot >= nritems) {
5136                         path->slots[level] = slot;
5137                         btrfs_set_path_blocking(path);
5138                         sret = btrfs_find_next_key(root, path, min_key, level,
5139                                                   min_trans);
5140                         if (sret == 0) {
5141                                 btrfs_release_path(path);
5142                                 goto again;
5143                         } else {
5144                                 goto out;
5145                         }
5146                 }
5147                 /* save our key for returning back */
5148                 btrfs_node_key_to_cpu(cur, &found_key, slot);
5149                 path->slots[level] = slot;
5150                 if (level == path->lowest_level) {
5151                         ret = 0;
5152                         goto out;
5153                 }
5154                 btrfs_set_path_blocking(path);
5155                 cur = btrfs_read_node_slot(cur, slot);
5156                 if (IS_ERR(cur)) {
5157                         ret = PTR_ERR(cur);
5158                         goto out;
5159                 }
5160
5161                 btrfs_tree_read_lock(cur);
5162
5163                 path->locks[level - 1] = BTRFS_READ_LOCK;
5164                 path->nodes[level - 1] = cur;
5165                 unlock_up(path, level, 1, 0, NULL);
5166         }
5167 out:
5168         path->keep_locks = keep_locks;
5169         if (ret == 0) {
5170                 btrfs_unlock_up_safe(path, path->lowest_level + 1);
5171                 btrfs_set_path_blocking(path);
5172                 memcpy(min_key, &found_key, sizeof(found_key));
5173         }
5174         return ret;
5175 }
5176
5177 /*
5178  * this is similar to btrfs_next_leaf, but does not try to preserve
5179  * and fixup the path.  It looks for and returns the next key in the
5180  * tree based on the current path and the min_trans parameters.
5181  *
5182  * 0 is returned if another key is found, < 0 if there are any errors
5183  * and 1 is returned if there are no higher keys in the tree
5184  *
5185  * path->keep_locks should be set to 1 on the search made before
5186  * calling this function.
5187  */
5188 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
5189                         struct btrfs_key *key, int level, u64 min_trans)
5190 {
5191         int slot;
5192         struct extent_buffer *c;
5193
5194         WARN_ON(!path->keep_locks && !path->skip_locking);
5195         while (level < BTRFS_MAX_LEVEL) {
5196                 if (!path->nodes[level])
5197                         return 1;
5198
5199                 slot = path->slots[level] + 1;
5200                 c = path->nodes[level];
5201 next:
5202                 if (slot >= btrfs_header_nritems(c)) {
5203                         int ret;
5204                         int orig_lowest;
5205                         struct btrfs_key cur_key;
5206                         if (level + 1 >= BTRFS_MAX_LEVEL ||
5207                             !path->nodes[level + 1])
5208                                 return 1;
5209
5210                         if (path->locks[level + 1] || path->skip_locking) {
5211                                 level++;
5212                                 continue;
5213                         }
5214
5215                         slot = btrfs_header_nritems(c) - 1;
5216                         if (level == 0)
5217                                 btrfs_item_key_to_cpu(c, &cur_key, slot);
5218                         else
5219                                 btrfs_node_key_to_cpu(c, &cur_key, slot);
5220
5221                         orig_lowest = path->lowest_level;
5222                         btrfs_release_path(path);
5223                         path->lowest_level = level;
5224                         ret = btrfs_search_slot(NULL, root, &cur_key, path,
5225                                                 0, 0);
5226                         path->lowest_level = orig_lowest;
5227                         if (ret < 0)
5228                                 return ret;
5229
5230                         c = path->nodes[level];
5231                         slot = path->slots[level];
5232                         if (ret == 0)
5233                                 slot++;
5234                         goto next;
5235                 }
5236
5237                 if (level == 0)
5238                         btrfs_item_key_to_cpu(c, key, slot);
5239                 else {
5240                         u64 gen = btrfs_node_ptr_generation(c, slot);
5241
5242                         if (gen < min_trans) {
5243                                 slot++;
5244                                 goto next;
5245                         }
5246                         btrfs_node_key_to_cpu(c, key, slot);
5247                 }
5248                 return 0;
5249         }
5250         return 1;
5251 }
5252
5253 /*
5254  * search the tree again to find a leaf with greater keys
5255  * returns 0 if it found something or 1 if there are no greater leaves.
5256  * returns < 0 on io errors.
5257  */
5258 int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
5259 {
5260         return btrfs_next_old_leaf(root, path, 0);
5261 }
5262
5263 int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path,
5264                         u64 time_seq)
5265 {
5266         int slot;
5267         int level;
5268         struct extent_buffer *c;
5269         struct extent_buffer *next;
5270         struct btrfs_key key;
5271         u32 nritems;
5272         int ret;
5273         int old_spinning = path->leave_spinning;
5274         int next_rw_lock = 0;
5275
5276         nritems = btrfs_header_nritems(path->nodes[0]);
5277         if (nritems == 0)
5278                 return 1;
5279
5280         btrfs_item_key_to_cpu(path->nodes[0], &key, nritems - 1);
5281 again:
5282         level = 1;
5283         next = NULL;
5284         next_rw_lock = 0;
5285         btrfs_release_path(path);
5286
5287         path->keep_locks = 1;
5288         path->leave_spinning = 1;
5289
5290         if (time_seq)
5291                 ret = btrfs_search_old_slot(root, &key, path, time_seq);
5292         else
5293                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5294         path->keep_locks = 0;
5295
5296         if (ret < 0)
5297                 return ret;
5298
5299         nritems = btrfs_header_nritems(path->nodes[0]);
5300         /*
5301          * by releasing the path above we dropped all our locks.  A balance
5302          * could have added more items next to the key that used to be
5303          * at the very end of the block.  So, check again here and
5304          * advance the path if there are now more items available.
5305          */
5306         if (nritems > 0 && path->slots[0] < nritems - 1) {
5307                 if (ret == 0)
5308                         path->slots[0]++;
5309                 ret = 0;
5310                 goto done;
5311         }
5312         /*
5313          * So the above check misses one case:
5314          * - after releasing the path above, someone has removed the item that
5315          *   used to be at the very end of the block, and balance between leafs
5316          *   gets another one with bigger key.offset to replace it.
5317          *
5318          * This one should be returned as well, or we can get leaf corruption
5319          * later(esp. in __btrfs_drop_extents()).
5320          *
5321          * And a bit more explanation about this check,
5322          * with ret > 0, the key isn't found, the path points to the slot
5323          * where it should be inserted, so the path->slots[0] item must be the
5324          * bigger one.
5325          */
5326         if (nritems > 0 && ret > 0 && path->slots[0] == nritems - 1) {
5327                 ret = 0;
5328                 goto done;
5329         }
5330
5331         while (level < BTRFS_MAX_LEVEL) {
5332                 if (!path->nodes[level]) {
5333                         ret = 1;
5334                         goto done;
5335                 }
5336
5337                 slot = path->slots[level] + 1;
5338                 c = path->nodes[level];
5339                 if (slot >= btrfs_header_nritems(c)) {
5340                         level++;
5341                         if (level == BTRFS_MAX_LEVEL) {
5342                                 ret = 1;
5343                                 goto done;
5344                         }
5345                         continue;
5346                 }
5347
5348                 if (next) {
5349                         btrfs_tree_unlock_rw(next, next_rw_lock);
5350                         free_extent_buffer(next);
5351                 }
5352
5353                 next = c;
5354                 next_rw_lock = path->locks[level];
5355                 ret = read_block_for_search(root, path, &next, level,
5356                                             slot, &key);
5357                 if (ret == -EAGAIN)
5358                         goto again;
5359
5360                 if (ret < 0) {
5361                         btrfs_release_path(path);
5362                         goto done;
5363                 }
5364
5365                 if (!path->skip_locking) {
5366                         ret = btrfs_try_tree_read_lock(next);
5367                         if (!ret && time_seq) {
5368                                 /*
5369                                  * If we don't get the lock, we may be racing
5370                                  * with push_leaf_left, holding that lock while
5371                                  * itself waiting for the leaf we've currently
5372                                  * locked. To solve this situation, we give up
5373                                  * on our lock and cycle.
5374                                  */
5375                                 free_extent_buffer(next);
5376                                 btrfs_release_path(path);
5377                                 cond_resched();
5378                                 goto again;
5379                         }
5380                         if (!ret) {
5381                                 btrfs_set_path_blocking(path);
5382                                 btrfs_tree_read_lock(next);
5383                         }
5384                         next_rw_lock = BTRFS_READ_LOCK;
5385                 }
5386                 break;
5387         }
5388         path->slots[level] = slot;
5389         while (1) {
5390                 level--;
5391                 c = path->nodes[level];
5392                 if (path->locks[level])
5393                         btrfs_tree_unlock_rw(c, path->locks[level]);
5394
5395                 free_extent_buffer(c);
5396                 path->nodes[level] = next;
5397                 path->slots[level] = 0;
5398                 if (!path->skip_locking)
5399                         path->locks[level] = next_rw_lock;
5400                 if (!level)
5401                         break;
5402
5403                 ret = read_block_for_search(root, path, &next, level,
5404                                             0, &key);
5405                 if (ret == -EAGAIN)
5406                         goto again;
5407
5408                 if (ret < 0) {
5409                         btrfs_release_path(path);
5410                         goto done;
5411                 }
5412
5413                 if (!path->skip_locking) {
5414                         ret = btrfs_try_tree_read_lock(next);
5415                         if (!ret) {
5416                                 btrfs_set_path_blocking(path);
5417                                 btrfs_tree_read_lock(next);
5418                         }
5419                         next_rw_lock = BTRFS_READ_LOCK;
5420                 }
5421         }
5422         ret = 0;
5423 done:
5424         unlock_up(path, 0, 1, 0, NULL);
5425         path->leave_spinning = old_spinning;
5426         if (!old_spinning)
5427                 btrfs_set_path_blocking(path);
5428
5429         return ret;
5430 }
5431
5432 /*
5433  * this uses btrfs_prev_leaf to walk backwards in the tree, and keeps
5434  * searching until it gets past min_objectid or finds an item of 'type'
5435  *
5436  * returns 0 if something is found, 1 if nothing was found and < 0 on error
5437  */
5438 int btrfs_previous_item(struct btrfs_root *root,
5439                         struct btrfs_path *path, u64 min_objectid,
5440                         int type)
5441 {
5442         struct btrfs_key found_key;
5443         struct extent_buffer *leaf;
5444         u32 nritems;
5445         int ret;
5446
5447         while (1) {
5448                 if (path->slots[0] == 0) {
5449                         btrfs_set_path_blocking(path);
5450                         ret = btrfs_prev_leaf(root, path);
5451                         if (ret != 0)
5452                                 return ret;
5453                 } else {
5454                         path->slots[0]--;
5455                 }
5456                 leaf = path->nodes[0];
5457                 nritems = btrfs_header_nritems(leaf);
5458                 if (nritems == 0)
5459                         return 1;
5460                 if (path->slots[0] == nritems)
5461                         path->slots[0]--;
5462
5463                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
5464                 if (found_key.objectid < min_objectid)
5465                         break;
5466                 if (found_key.type == type)
5467                         return 0;
5468                 if (found_key.objectid == min_objectid &&
5469                     found_key.type < type)
5470                         break;
5471         }
5472         return 1;
5473 }
5474
5475 /*
5476  * search in extent tree to find a previous Metadata/Data extent item with
5477  * min objecitd.
5478  *
5479  * returns 0 if something is found, 1 if nothing was found and < 0 on error
5480  */
5481 int btrfs_previous_extent_item(struct btrfs_root *root,
5482                         struct btrfs_path *path, u64 min_objectid)
5483 {
5484         struct btrfs_key found_key;
5485         struct extent_buffer *leaf;
5486         u32 nritems;
5487         int ret;
5488
5489         while (1) {
5490                 if (path->slots[0] == 0) {
5491                         btrfs_set_path_blocking(path);
5492                         ret = btrfs_prev_leaf(root, path);
5493                         if (ret != 0)
5494                                 return ret;
5495                 } else {
5496                         path->slots[0]--;
5497                 }
5498                 leaf = path->nodes[0];
5499                 nritems = btrfs_header_nritems(leaf);
5500                 if (nritems == 0)
5501                         return 1;
5502                 if (path->slots[0] == nritems)
5503                         path->slots[0]--;
5504
5505                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
5506                 if (found_key.objectid < min_objectid)
5507                         break;
5508                 if (found_key.type == BTRFS_EXTENT_ITEM_KEY ||
5509                     found_key.type == BTRFS_METADATA_ITEM_KEY)
5510                         return 0;
5511                 if (found_key.objectid == min_objectid &&
5512                     found_key.type < BTRFS_EXTENT_ITEM_KEY)
5513                         break;
5514         }
5515         return 1;
5516 }