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 __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_tree_read_lock(eb_rewin);
1301         __tree_mod_log_rewind(fs_info, eb_rewin, time_seq, tm);
1302         WARN_ON(btrfs_header_nritems(eb_rewin) >
1303                 BTRFS_NODEPTRS_PER_BLOCK(fs_info));
1304
1305         return eb_rewin;
1306 }
1307
1308 /*
1309  * get_old_root() rewinds the state of @root's root node to the given @time_seq
1310  * value. If there are no changes, the current root->root_node is returned. If
1311  * anything changed in between, there's a fresh buffer allocated on which the
1312  * rewind operations are done. In any case, the returned buffer is read locked.
1313  * Returns NULL on error (with no locks held).
1314  */
1315 static inline struct extent_buffer *
1316 get_old_root(struct btrfs_root *root, u64 time_seq)
1317 {
1318         struct btrfs_fs_info *fs_info = root->fs_info;
1319         struct tree_mod_elem *tm;
1320         struct extent_buffer *eb = NULL;
1321         struct extent_buffer *eb_root;
1322         u64 eb_root_owner = 0;
1323         struct extent_buffer *old;
1324         struct tree_mod_root *old_root = NULL;
1325         u64 old_generation = 0;
1326         u64 logical;
1327         int level;
1328
1329         eb_root = btrfs_read_lock_root_node(root);
1330         tm = __tree_mod_log_oldest_root(eb_root, time_seq);
1331         if (!tm)
1332                 return eb_root;
1333
1334         if (tm->op == MOD_LOG_ROOT_REPLACE) {
1335                 old_root = &tm->old_root;
1336                 old_generation = tm->generation;
1337                 logical = old_root->logical;
1338                 level = old_root->level;
1339         } else {
1340                 logical = eb_root->start;
1341                 level = btrfs_header_level(eb_root);
1342         }
1343
1344         tm = tree_mod_log_search(fs_info, logical, time_seq);
1345         if (old_root && tm && tm->op != MOD_LOG_KEY_REMOVE_WHILE_FREEING) {
1346                 btrfs_tree_read_unlock(eb_root);
1347                 free_extent_buffer(eb_root);
1348                 old = read_tree_block(fs_info, logical, 0, level, NULL);
1349                 if (WARN_ON(IS_ERR(old) || !extent_buffer_uptodate(old))) {
1350                         if (!IS_ERR(old))
1351                                 free_extent_buffer(old);
1352                         btrfs_warn(fs_info,
1353                                    "failed to read tree block %llu from get_old_root",
1354                                    logical);
1355                 } else {
1356                         eb = btrfs_clone_extent_buffer(old);
1357                         free_extent_buffer(old);
1358                 }
1359         } else if (old_root) {
1360                 eb_root_owner = btrfs_header_owner(eb_root);
1361                 btrfs_tree_read_unlock(eb_root);
1362                 free_extent_buffer(eb_root);
1363                 eb = alloc_dummy_extent_buffer(fs_info, logical);
1364         } else {
1365                 btrfs_set_lock_blocking_read(eb_root);
1366                 eb = btrfs_clone_extent_buffer(eb_root);
1367                 btrfs_tree_read_unlock_blocking(eb_root);
1368                 free_extent_buffer(eb_root);
1369         }
1370
1371         if (!eb)
1372                 return NULL;
1373         btrfs_tree_read_lock(eb);
1374         if (old_root) {
1375                 btrfs_set_header_bytenr(eb, eb->start);
1376                 btrfs_set_header_backref_rev(eb, BTRFS_MIXED_BACKREF_REV);
1377                 btrfs_set_header_owner(eb, eb_root_owner);
1378                 btrfs_set_header_level(eb, old_root->level);
1379                 btrfs_set_header_generation(eb, old_generation);
1380         }
1381         if (tm)
1382                 __tree_mod_log_rewind(fs_info, eb, time_seq, tm);
1383         else
1384                 WARN_ON(btrfs_header_level(eb) != 0);
1385         WARN_ON(btrfs_header_nritems(eb) > BTRFS_NODEPTRS_PER_BLOCK(fs_info));
1386
1387         return eb;
1388 }
1389
1390 int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq)
1391 {
1392         struct tree_mod_elem *tm;
1393         int level;
1394         struct extent_buffer *eb_root = btrfs_root_node(root);
1395
1396         tm = __tree_mod_log_oldest_root(eb_root, time_seq);
1397         if (tm && tm->op == MOD_LOG_ROOT_REPLACE) {
1398                 level = tm->old_root.level;
1399         } else {
1400                 level = btrfs_header_level(eb_root);
1401         }
1402         free_extent_buffer(eb_root);
1403
1404         return level;
1405 }
1406
1407 static inline int should_cow_block(struct btrfs_trans_handle *trans,
1408                                    struct btrfs_root *root,
1409                                    struct extent_buffer *buf)
1410 {
1411         if (btrfs_is_testing(root->fs_info))
1412                 return 0;
1413
1414         /* Ensure we can see the FORCE_COW bit */
1415         smp_mb__before_atomic();
1416
1417         /*
1418          * We do not need to cow a block if
1419          * 1) this block is not created or changed in this transaction;
1420          * 2) this block does not belong to TREE_RELOC tree;
1421          * 3) the root is not forced COW.
1422          *
1423          * What is forced COW:
1424          *    when we create snapshot during committing the transaction,
1425          *    after we've finished copying src root, we must COW the shared
1426          *    block to ensure the metadata consistency.
1427          */
1428         if (btrfs_header_generation(buf) == trans->transid &&
1429             !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) &&
1430             !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
1431               btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)) &&
1432             !test_bit(BTRFS_ROOT_FORCE_COW, &root->state))
1433                 return 0;
1434         return 1;
1435 }
1436
1437 /*
1438  * cows a single block, see __btrfs_cow_block for the real work.
1439  * This version of it has extra checks so that a block isn't COWed more than
1440  * once per transaction, as long as it hasn't been written yet
1441  */
1442 noinline int btrfs_cow_block(struct btrfs_trans_handle *trans,
1443                     struct btrfs_root *root, struct extent_buffer *buf,
1444                     struct extent_buffer *parent, int parent_slot,
1445                     struct extent_buffer **cow_ret)
1446 {
1447         struct btrfs_fs_info *fs_info = root->fs_info;
1448         u64 search_start;
1449         int ret;
1450
1451         if (test_bit(BTRFS_ROOT_DELETING, &root->state))
1452                 btrfs_err(fs_info,
1453                         "COW'ing blocks on a fs root that's being dropped");
1454
1455         if (trans->transaction != fs_info->running_transaction)
1456                 WARN(1, KERN_CRIT "trans %llu running %llu\n",
1457                        trans->transid,
1458                        fs_info->running_transaction->transid);
1459
1460         if (trans->transid != fs_info->generation)
1461                 WARN(1, KERN_CRIT "trans %llu running %llu\n",
1462                        trans->transid, fs_info->generation);
1463
1464         if (!should_cow_block(trans, root, buf)) {
1465                 trans->dirty = true;
1466                 *cow_ret = buf;
1467                 return 0;
1468         }
1469
1470         search_start = buf->start & ~((u64)SZ_1G - 1);
1471
1472         if (parent)
1473                 btrfs_set_lock_blocking_write(parent);
1474         btrfs_set_lock_blocking_write(buf);
1475
1476         /*
1477          * Before CoWing this block for later modification, check if it's
1478          * the subtree root and do the delayed subtree trace if needed.
1479          *
1480          * Also We don't care about the error, as it's handled internally.
1481          */
1482         btrfs_qgroup_trace_subtree_after_cow(trans, root, buf);
1483         ret = __btrfs_cow_block(trans, root, buf, parent,
1484                                  parent_slot, cow_ret, search_start, 0);
1485
1486         trace_btrfs_cow_block(root, buf, *cow_ret);
1487
1488         return ret;
1489 }
1490
1491 /*
1492  * helper function for defrag to decide if two blocks pointed to by a
1493  * node are actually close by
1494  */
1495 static int close_blocks(u64 blocknr, u64 other, u32 blocksize)
1496 {
1497         if (blocknr < other && other - (blocknr + blocksize) < 32768)
1498                 return 1;
1499         if (blocknr > other && blocknr - (other + blocksize) < 32768)
1500                 return 1;
1501         return 0;
1502 }
1503
1504 /*
1505  * compare two keys in a memcmp fashion
1506  */
1507 static int comp_keys(const struct btrfs_disk_key *disk,
1508                      const struct btrfs_key *k2)
1509 {
1510         struct btrfs_key k1;
1511
1512         btrfs_disk_key_to_cpu(&k1, disk);
1513
1514         return btrfs_comp_cpu_keys(&k1, k2);
1515 }
1516
1517 /*
1518  * same as comp_keys only with two btrfs_key's
1519  */
1520 int __pure btrfs_comp_cpu_keys(const struct btrfs_key *k1, const struct btrfs_key *k2)
1521 {
1522         if (k1->objectid > k2->objectid)
1523                 return 1;
1524         if (k1->objectid < k2->objectid)
1525                 return -1;
1526         if (k1->type > k2->type)
1527                 return 1;
1528         if (k1->type < k2->type)
1529                 return -1;
1530         if (k1->offset > k2->offset)
1531                 return 1;
1532         if (k1->offset < k2->offset)
1533                 return -1;
1534         return 0;
1535 }
1536
1537 /*
1538  * this is used by the defrag code to go through all the
1539  * leaves pointed to by a node and reallocate them so that
1540  * disk order is close to key order
1541  */
1542 int btrfs_realloc_node(struct btrfs_trans_handle *trans,
1543                        struct btrfs_root *root, struct extent_buffer *parent,
1544                        int start_slot, u64 *last_ret,
1545                        struct btrfs_key *progress)
1546 {
1547         struct btrfs_fs_info *fs_info = root->fs_info;
1548         struct extent_buffer *cur;
1549         u64 blocknr;
1550         u64 gen;
1551         u64 search_start = *last_ret;
1552         u64 last_block = 0;
1553         u64 other;
1554         u32 parent_nritems;
1555         int end_slot;
1556         int i;
1557         int err = 0;
1558         int parent_level;
1559         int uptodate;
1560         u32 blocksize;
1561         int progress_passed = 0;
1562         struct btrfs_disk_key disk_key;
1563
1564         parent_level = btrfs_header_level(parent);
1565
1566         WARN_ON(trans->transaction != fs_info->running_transaction);
1567         WARN_ON(trans->transid != fs_info->generation);
1568
1569         parent_nritems = btrfs_header_nritems(parent);
1570         blocksize = fs_info->nodesize;
1571         end_slot = parent_nritems - 1;
1572
1573         if (parent_nritems <= 1)
1574                 return 0;
1575
1576         btrfs_set_lock_blocking_write(parent);
1577
1578         for (i = start_slot; i <= end_slot; i++) {
1579                 struct btrfs_key first_key;
1580                 int close = 1;
1581
1582                 btrfs_node_key(parent, &disk_key, i);
1583                 if (!progress_passed && comp_keys(&disk_key, progress) < 0)
1584                         continue;
1585
1586                 progress_passed = 1;
1587                 blocknr = btrfs_node_blockptr(parent, i);
1588                 gen = btrfs_node_ptr_generation(parent, i);
1589                 btrfs_node_key_to_cpu(parent, &first_key, i);
1590                 if (last_block == 0)
1591                         last_block = blocknr;
1592
1593                 if (i > 0) {
1594                         other = btrfs_node_blockptr(parent, i - 1);
1595                         close = close_blocks(blocknr, other, blocksize);
1596                 }
1597                 if (!close && i < end_slot) {
1598                         other = btrfs_node_blockptr(parent, i + 1);
1599                         close = close_blocks(blocknr, other, blocksize);
1600                 }
1601                 if (close) {
1602                         last_block = blocknr;
1603                         continue;
1604                 }
1605
1606                 cur = find_extent_buffer(fs_info, blocknr);
1607                 if (cur)
1608                         uptodate = btrfs_buffer_uptodate(cur, gen, 0);
1609                 else
1610                         uptodate = 0;
1611                 if (!cur || !uptodate) {
1612                         if (!cur) {
1613                                 cur = read_tree_block(fs_info, blocknr, gen,
1614                                                       parent_level - 1,
1615                                                       &first_key);
1616                                 if (IS_ERR(cur)) {
1617                                         return PTR_ERR(cur);
1618                                 } else if (!extent_buffer_uptodate(cur)) {
1619                                         free_extent_buffer(cur);
1620                                         return -EIO;
1621                                 }
1622                         } else if (!uptodate) {
1623                                 err = btrfs_read_buffer(cur, gen,
1624                                                 parent_level - 1,&first_key);
1625                                 if (err) {
1626                                         free_extent_buffer(cur);
1627                                         return err;
1628                                 }
1629                         }
1630                 }
1631                 if (search_start == 0)
1632                         search_start = last_block;
1633
1634                 btrfs_tree_lock(cur);
1635                 btrfs_set_lock_blocking_write(cur);
1636                 err = __btrfs_cow_block(trans, root, cur, parent, i,
1637                                         &cur, search_start,
1638                                         min(16 * blocksize,
1639                                             (end_slot - i) * blocksize));
1640                 if (err) {
1641                         btrfs_tree_unlock(cur);
1642                         free_extent_buffer(cur);
1643                         break;
1644                 }
1645                 search_start = cur->start;
1646                 last_block = cur->start;
1647                 *last_ret = search_start;
1648                 btrfs_tree_unlock(cur);
1649                 free_extent_buffer(cur);
1650         }
1651         return err;
1652 }
1653
1654 /*
1655  * search for key in the extent_buffer.  The items start at offset p,
1656  * and they are item_size apart.  There are 'max' items in p.
1657  *
1658  * the slot in the array is returned via slot, and it points to
1659  * the place where you would insert key if it is not found in
1660  * the array.
1661  *
1662  * slot may point to max if the key is bigger than all of the keys
1663  */
1664 static noinline int generic_bin_search(struct extent_buffer *eb,
1665                                        unsigned long p, int item_size,
1666                                        const struct btrfs_key *key,
1667                                        int max, int *slot)
1668 {
1669         int low = 0;
1670         int high = max;
1671         int ret;
1672         const int key_size = sizeof(struct btrfs_disk_key);
1673
1674         if (low > high) {
1675                 btrfs_err(eb->fs_info,
1676                  "%s: low (%d) > high (%d) eb %llu owner %llu level %d",
1677                           __func__, low, high, eb->start,
1678                           btrfs_header_owner(eb), btrfs_header_level(eb));
1679                 return -EINVAL;
1680         }
1681
1682         while (low < high) {
1683                 unsigned long oip;
1684                 unsigned long offset;
1685                 struct btrfs_disk_key *tmp;
1686                 struct btrfs_disk_key unaligned;
1687                 int mid;
1688
1689                 mid = (low + high) / 2;
1690                 offset = p + mid * item_size;
1691                 oip = offset_in_page(offset);
1692
1693                 if (oip + key_size <= PAGE_SIZE) {
1694                         const unsigned long idx = offset >> PAGE_SHIFT;
1695                         char *kaddr = page_address(eb->pages[idx]);
1696
1697                         tmp = (struct btrfs_disk_key *)(kaddr + oip);
1698                 } else {
1699                         read_extent_buffer(eb, &unaligned, offset, key_size);
1700                         tmp = &unaligned;
1701                 }
1702
1703                 ret = comp_keys(tmp, key);
1704
1705                 if (ret < 0)
1706                         low = mid + 1;
1707                 else if (ret > 0)
1708                         high = mid;
1709                 else {
1710                         *slot = mid;
1711                         return 0;
1712                 }
1713         }
1714         *slot = low;
1715         return 1;
1716 }
1717
1718 /*
1719  * simple bin_search frontend that does the right thing for
1720  * leaves vs nodes
1721  */
1722 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
1723                      int *slot)
1724 {
1725         if (btrfs_header_level(eb) == 0)
1726                 return generic_bin_search(eb,
1727                                           offsetof(struct btrfs_leaf, items),
1728                                           sizeof(struct btrfs_item),
1729                                           key, btrfs_header_nritems(eb),
1730                                           slot);
1731         else
1732                 return generic_bin_search(eb,
1733                                           offsetof(struct btrfs_node, ptrs),
1734                                           sizeof(struct btrfs_key_ptr),
1735                                           key, btrfs_header_nritems(eb),
1736                                           slot);
1737 }
1738
1739 static void root_add_used(struct btrfs_root *root, u32 size)
1740 {
1741         spin_lock(&root->accounting_lock);
1742         btrfs_set_root_used(&root->root_item,
1743                             btrfs_root_used(&root->root_item) + size);
1744         spin_unlock(&root->accounting_lock);
1745 }
1746
1747 static void root_sub_used(struct btrfs_root *root, u32 size)
1748 {
1749         spin_lock(&root->accounting_lock);
1750         btrfs_set_root_used(&root->root_item,
1751                             btrfs_root_used(&root->root_item) - size);
1752         spin_unlock(&root->accounting_lock);
1753 }
1754
1755 /* given a node and slot number, this reads the blocks it points to.  The
1756  * extent buffer is returned with a reference taken (but unlocked).
1757  */
1758 struct extent_buffer *btrfs_read_node_slot(struct extent_buffer *parent,
1759                                            int slot)
1760 {
1761         int level = btrfs_header_level(parent);
1762         struct extent_buffer *eb;
1763         struct btrfs_key first_key;
1764
1765         if (slot < 0 || slot >= btrfs_header_nritems(parent))
1766                 return ERR_PTR(-ENOENT);
1767
1768         BUG_ON(level == 0);
1769
1770         btrfs_node_key_to_cpu(parent, &first_key, slot);
1771         eb = read_tree_block(parent->fs_info, btrfs_node_blockptr(parent, slot),
1772                              btrfs_node_ptr_generation(parent, slot),
1773                              level - 1, &first_key);
1774         if (!IS_ERR(eb) && !extent_buffer_uptodate(eb)) {
1775                 free_extent_buffer(eb);
1776                 eb = ERR_PTR(-EIO);
1777         }
1778
1779         return eb;
1780 }
1781
1782 /*
1783  * node level balancing, used to make sure nodes are in proper order for
1784  * item deletion.  We balance from the top down, so we have to make sure
1785  * that a deletion won't leave an node completely empty later on.
1786  */
1787 static noinline int balance_level(struct btrfs_trans_handle *trans,
1788                          struct btrfs_root *root,
1789                          struct btrfs_path *path, int level)
1790 {
1791         struct btrfs_fs_info *fs_info = root->fs_info;
1792         struct extent_buffer *right = NULL;
1793         struct extent_buffer *mid;
1794         struct extent_buffer *left = NULL;
1795         struct extent_buffer *parent = NULL;
1796         int ret = 0;
1797         int wret;
1798         int pslot;
1799         int orig_slot = path->slots[level];
1800         u64 orig_ptr;
1801
1802         ASSERT(level > 0);
1803
1804         mid = path->nodes[level];
1805
1806         WARN_ON(path->locks[level] != BTRFS_WRITE_LOCK &&
1807                 path->locks[level] != BTRFS_WRITE_LOCK_BLOCKING);
1808         WARN_ON(btrfs_header_generation(mid) != trans->transid);
1809
1810         orig_ptr = btrfs_node_blockptr(mid, orig_slot);
1811
1812         if (level < BTRFS_MAX_LEVEL - 1) {
1813                 parent = path->nodes[level + 1];
1814                 pslot = path->slots[level + 1];
1815         }
1816
1817         /*
1818          * deal with the case where there is only one pointer in the root
1819          * by promoting the node below to a root
1820          */
1821         if (!parent) {
1822                 struct extent_buffer *child;
1823
1824                 if (btrfs_header_nritems(mid) != 1)
1825                         return 0;
1826
1827                 /* promote the child to a root */
1828                 child = btrfs_read_node_slot(mid, 0);
1829                 if (IS_ERR(child)) {
1830                         ret = PTR_ERR(child);
1831                         btrfs_handle_fs_error(fs_info, ret, NULL);
1832                         goto enospc;
1833                 }
1834
1835                 btrfs_tree_lock(child);
1836                 btrfs_set_lock_blocking_write(child);
1837                 ret = btrfs_cow_block(trans, root, child, mid, 0, &child);
1838                 if (ret) {
1839                         btrfs_tree_unlock(child);
1840                         free_extent_buffer(child);
1841                         goto enospc;
1842                 }
1843
1844                 ret = tree_mod_log_insert_root(root->node, child, 1);
1845                 BUG_ON(ret < 0);
1846                 rcu_assign_pointer(root->node, child);
1847
1848                 add_root_to_dirty_list(root);
1849                 btrfs_tree_unlock(child);
1850
1851                 path->locks[level] = 0;
1852                 path->nodes[level] = NULL;
1853                 btrfs_clean_tree_block(mid);
1854                 btrfs_tree_unlock(mid);
1855                 /* once for the path */
1856                 free_extent_buffer(mid);
1857
1858                 root_sub_used(root, mid->len);
1859                 btrfs_free_tree_block(trans, root, mid, 0, 1);
1860                 /* once for the root ptr */
1861                 free_extent_buffer_stale(mid);
1862                 return 0;
1863         }
1864         if (btrfs_header_nritems(mid) >
1865             BTRFS_NODEPTRS_PER_BLOCK(fs_info) / 4)
1866                 return 0;
1867
1868         left = btrfs_read_node_slot(parent, pslot - 1);
1869         if (IS_ERR(left))
1870                 left = NULL;
1871
1872         if (left) {
1873                 btrfs_tree_lock(left);
1874                 btrfs_set_lock_blocking_write(left);
1875                 wret = btrfs_cow_block(trans, root, left,
1876                                        parent, pslot - 1, &left);
1877                 if (wret) {
1878                         ret = wret;
1879                         goto enospc;
1880                 }
1881         }
1882
1883         right = btrfs_read_node_slot(parent, pslot + 1);
1884         if (IS_ERR(right))
1885                 right = NULL;
1886
1887         if (right) {
1888                 btrfs_tree_lock(right);
1889                 btrfs_set_lock_blocking_write(right);
1890                 wret = btrfs_cow_block(trans, root, right,
1891                                        parent, pslot + 1, &right);
1892                 if (wret) {
1893                         ret = wret;
1894                         goto enospc;
1895                 }
1896         }
1897
1898         /* first, try to make some room in the middle buffer */
1899         if (left) {
1900                 orig_slot += btrfs_header_nritems(left);
1901                 wret = push_node_left(trans, left, mid, 1);
1902                 if (wret < 0)
1903                         ret = wret;
1904         }
1905
1906         /*
1907          * then try to empty the right most buffer into the middle
1908          */
1909         if (right) {
1910                 wret = push_node_left(trans, mid, right, 1);
1911                 if (wret < 0 && wret != -ENOSPC)
1912                         ret = wret;
1913                 if (btrfs_header_nritems(right) == 0) {
1914                         btrfs_clean_tree_block(right);
1915                         btrfs_tree_unlock(right);
1916                         del_ptr(root, path, level + 1, pslot + 1);
1917                         root_sub_used(root, right->len);
1918                         btrfs_free_tree_block(trans, root, right, 0, 1);
1919                         free_extent_buffer_stale(right);
1920                         right = NULL;
1921                 } else {
1922                         struct btrfs_disk_key right_key;
1923                         btrfs_node_key(right, &right_key, 0);
1924                         ret = tree_mod_log_insert_key(parent, pslot + 1,
1925                                         MOD_LOG_KEY_REPLACE, GFP_NOFS);
1926                         BUG_ON(ret < 0);
1927                         btrfs_set_node_key(parent, &right_key, pslot + 1);
1928                         btrfs_mark_buffer_dirty(parent);
1929                 }
1930         }
1931         if (btrfs_header_nritems(mid) == 1) {
1932                 /*
1933                  * we're not allowed to leave a node with one item in the
1934                  * tree during a delete.  A deletion from lower in the tree
1935                  * could try to delete the only pointer in this node.
1936                  * So, pull some keys from the left.
1937                  * There has to be a left pointer at this point because
1938                  * otherwise we would have pulled some pointers from the
1939                  * right
1940                  */
1941                 if (!left) {
1942                         ret = -EROFS;
1943                         btrfs_handle_fs_error(fs_info, ret, NULL);
1944                         goto enospc;
1945                 }
1946                 wret = balance_node_right(trans, mid, left);
1947                 if (wret < 0) {
1948                         ret = wret;
1949                         goto enospc;
1950                 }
1951                 if (wret == 1) {
1952                         wret = push_node_left(trans, left, mid, 1);
1953                         if (wret < 0)
1954                                 ret = wret;
1955                 }
1956                 BUG_ON(wret == 1);
1957         }
1958         if (btrfs_header_nritems(mid) == 0) {
1959                 btrfs_clean_tree_block(mid);
1960                 btrfs_tree_unlock(mid);
1961                 del_ptr(root, path, level + 1, pslot);
1962                 root_sub_used(root, mid->len);
1963                 btrfs_free_tree_block(trans, root, mid, 0, 1);
1964                 free_extent_buffer_stale(mid);
1965                 mid = NULL;
1966         } else {
1967                 /* update the parent key to reflect our changes */
1968                 struct btrfs_disk_key mid_key;
1969                 btrfs_node_key(mid, &mid_key, 0);
1970                 ret = tree_mod_log_insert_key(parent, pslot,
1971                                 MOD_LOG_KEY_REPLACE, GFP_NOFS);
1972                 BUG_ON(ret < 0);
1973                 btrfs_set_node_key(parent, &mid_key, pslot);
1974                 btrfs_mark_buffer_dirty(parent);
1975         }
1976
1977         /* update the path */
1978         if (left) {
1979                 if (btrfs_header_nritems(left) > orig_slot) {
1980                         atomic_inc(&left->refs);
1981                         /* left was locked after cow */
1982                         path->nodes[level] = left;
1983                         path->slots[level + 1] -= 1;
1984                         path->slots[level] = orig_slot;
1985                         if (mid) {
1986                                 btrfs_tree_unlock(mid);
1987                                 free_extent_buffer(mid);
1988                         }
1989                 } else {
1990                         orig_slot -= btrfs_header_nritems(left);
1991                         path->slots[level] = orig_slot;
1992                 }
1993         }
1994         /* double check we haven't messed things up */
1995         if (orig_ptr !=
1996             btrfs_node_blockptr(path->nodes[level], path->slots[level]))
1997                 BUG();
1998 enospc:
1999         if (right) {
2000                 btrfs_tree_unlock(right);
2001                 free_extent_buffer(right);
2002         }
2003         if (left) {
2004                 if (path->nodes[level] != left)
2005                         btrfs_tree_unlock(left);
2006                 free_extent_buffer(left);
2007         }
2008         return ret;
2009 }
2010
2011 /* Node balancing for insertion.  Here we only split or push nodes around
2012  * when they are completely full.  This is also done top down, so we
2013  * have to be pessimistic.
2014  */
2015 static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans,
2016                                           struct btrfs_root *root,
2017                                           struct btrfs_path *path, int level)
2018 {
2019         struct btrfs_fs_info *fs_info = root->fs_info;
2020         struct extent_buffer *right = NULL;
2021         struct extent_buffer *mid;
2022         struct extent_buffer *left = NULL;
2023         struct extent_buffer *parent = NULL;
2024         int ret = 0;
2025         int wret;
2026         int pslot;
2027         int orig_slot = path->slots[level];
2028
2029         if (level == 0)
2030                 return 1;
2031
2032         mid = path->nodes[level];
2033         WARN_ON(btrfs_header_generation(mid) != trans->transid);
2034
2035         if (level < BTRFS_MAX_LEVEL - 1) {
2036                 parent = path->nodes[level + 1];
2037                 pslot = path->slots[level + 1];
2038         }
2039
2040         if (!parent)
2041                 return 1;
2042
2043         left = btrfs_read_node_slot(parent, pslot - 1);
2044         if (IS_ERR(left))
2045                 left = NULL;
2046
2047         /* first, try to make some room in the middle buffer */
2048         if (left) {
2049                 u32 left_nr;
2050
2051                 btrfs_tree_lock(left);
2052                 btrfs_set_lock_blocking_write(left);
2053
2054                 left_nr = btrfs_header_nritems(left);
2055                 if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
2056                         wret = 1;
2057                 } else {
2058                         ret = btrfs_cow_block(trans, root, left, parent,
2059                                               pslot - 1, &left);
2060                         if (ret)
2061                                 wret = 1;
2062                         else {
2063                                 wret = push_node_left(trans, left, mid, 0);
2064                         }
2065                 }
2066                 if (wret < 0)
2067                         ret = wret;
2068                 if (wret == 0) {
2069                         struct btrfs_disk_key disk_key;
2070                         orig_slot += left_nr;
2071                         btrfs_node_key(mid, &disk_key, 0);
2072                         ret = tree_mod_log_insert_key(parent, pslot,
2073                                         MOD_LOG_KEY_REPLACE, GFP_NOFS);
2074                         BUG_ON(ret < 0);
2075                         btrfs_set_node_key(parent, &disk_key, pslot);
2076                         btrfs_mark_buffer_dirty(parent);
2077                         if (btrfs_header_nritems(left) > orig_slot) {
2078                                 path->nodes[level] = left;
2079                                 path->slots[level + 1] -= 1;
2080                                 path->slots[level] = orig_slot;
2081                                 btrfs_tree_unlock(mid);
2082                                 free_extent_buffer(mid);
2083                         } else {
2084                                 orig_slot -=
2085                                         btrfs_header_nritems(left);
2086                                 path->slots[level] = orig_slot;
2087                                 btrfs_tree_unlock(left);
2088                                 free_extent_buffer(left);
2089                         }
2090                         return 0;
2091                 }
2092                 btrfs_tree_unlock(left);
2093                 free_extent_buffer(left);
2094         }
2095         right = btrfs_read_node_slot(parent, pslot + 1);
2096         if (IS_ERR(right))
2097                 right = NULL;
2098
2099         /*
2100          * then try to empty the right most buffer into the middle
2101          */
2102         if (right) {
2103                 u32 right_nr;
2104
2105                 btrfs_tree_lock(right);
2106                 btrfs_set_lock_blocking_write(right);
2107
2108                 right_nr = btrfs_header_nritems(right);
2109                 if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 1) {
2110                         wret = 1;
2111                 } else {
2112                         ret = btrfs_cow_block(trans, root, right,
2113                                               parent, pslot + 1,
2114                                               &right);
2115                         if (ret)
2116                                 wret = 1;
2117                         else {
2118                                 wret = balance_node_right(trans, right, mid);
2119                         }
2120                 }
2121                 if (wret < 0)
2122                         ret = wret;
2123                 if (wret == 0) {
2124                         struct btrfs_disk_key disk_key;
2125
2126                         btrfs_node_key(right, &disk_key, 0);
2127                         ret = tree_mod_log_insert_key(parent, pslot + 1,
2128                                         MOD_LOG_KEY_REPLACE, GFP_NOFS);
2129                         BUG_ON(ret < 0);
2130                         btrfs_set_node_key(parent, &disk_key, pslot + 1);
2131                         btrfs_mark_buffer_dirty(parent);
2132
2133                         if (btrfs_header_nritems(mid) <= orig_slot) {
2134                                 path->nodes[level] = right;
2135                                 path->slots[level + 1] += 1;
2136                                 path->slots[level] = orig_slot -
2137                                         btrfs_header_nritems(mid);
2138                                 btrfs_tree_unlock(mid);
2139                                 free_extent_buffer(mid);
2140                         } else {
2141                                 btrfs_tree_unlock(right);
2142                                 free_extent_buffer(right);
2143                         }
2144                         return 0;
2145                 }
2146                 btrfs_tree_unlock(right);
2147                 free_extent_buffer(right);
2148         }
2149         return 1;
2150 }
2151
2152 /*
2153  * readahead one full node of leaves, finding things that are close
2154  * to the block in 'slot', and triggering ra on them.
2155  */
2156 static void reada_for_search(struct btrfs_fs_info *fs_info,
2157                              struct btrfs_path *path,
2158                              int level, int slot, u64 objectid)
2159 {
2160         struct extent_buffer *node;
2161         struct btrfs_disk_key disk_key;
2162         u32 nritems;
2163         u64 search;
2164         u64 target;
2165         u64 nread = 0;
2166         struct extent_buffer *eb;
2167         u32 nr;
2168         u32 blocksize;
2169         u32 nscan = 0;
2170
2171         if (level != 1)
2172                 return;
2173
2174         if (!path->nodes[level])
2175                 return;
2176
2177         node = path->nodes[level];
2178
2179         search = btrfs_node_blockptr(node, slot);
2180         blocksize = fs_info->nodesize;
2181         eb = find_extent_buffer(fs_info, search);
2182         if (eb) {
2183                 free_extent_buffer(eb);
2184                 return;
2185         }
2186
2187         target = search;
2188
2189         nritems = btrfs_header_nritems(node);
2190         nr = slot;
2191
2192         while (1) {
2193                 if (path->reada == READA_BACK) {
2194                         if (nr == 0)
2195                                 break;
2196                         nr--;
2197                 } else if (path->reada == READA_FORWARD) {
2198                         nr++;
2199                         if (nr >= nritems)
2200                                 break;
2201                 }
2202                 if (path->reada == READA_BACK && objectid) {
2203                         btrfs_node_key(node, &disk_key, nr);
2204                         if (btrfs_disk_key_objectid(&disk_key) != objectid)
2205                                 break;
2206                 }
2207                 search = btrfs_node_blockptr(node, nr);
2208                 if ((search <= target && target - search <= 65536) ||
2209                     (search > target && search - target <= 65536)) {
2210                         readahead_tree_block(fs_info, search);
2211                         nread += blocksize;
2212                 }
2213                 nscan++;
2214                 if ((nread > 65536 || nscan > 32))
2215                         break;
2216         }
2217 }
2218
2219 static noinline void reada_for_balance(struct btrfs_fs_info *fs_info,
2220                                        struct btrfs_path *path, int level)
2221 {
2222         int slot;
2223         int nritems;
2224         struct extent_buffer *parent;
2225         struct extent_buffer *eb;
2226         u64 gen;
2227         u64 block1 = 0;
2228         u64 block2 = 0;
2229
2230         parent = path->nodes[level + 1];
2231         if (!parent)
2232                 return;
2233
2234         nritems = btrfs_header_nritems(parent);
2235         slot = path->slots[level + 1];
2236
2237         if (slot > 0) {
2238                 block1 = btrfs_node_blockptr(parent, slot - 1);
2239                 gen = btrfs_node_ptr_generation(parent, slot - 1);
2240                 eb = find_extent_buffer(fs_info, block1);
2241                 /*
2242                  * if we get -eagain from btrfs_buffer_uptodate, we
2243                  * don't want to return eagain here.  That will loop
2244                  * forever
2245                  */
2246                 if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0)
2247                         block1 = 0;
2248                 free_extent_buffer(eb);
2249         }
2250         if (slot + 1 < nritems) {
2251                 block2 = btrfs_node_blockptr(parent, slot + 1);
2252                 gen = btrfs_node_ptr_generation(parent, slot + 1);
2253                 eb = find_extent_buffer(fs_info, block2);
2254                 if (eb && btrfs_buffer_uptodate(eb, gen, 1) != 0)
2255                         block2 = 0;
2256                 free_extent_buffer(eb);
2257         }
2258
2259         if (block1)
2260                 readahead_tree_block(fs_info, block1);
2261         if (block2)
2262                 readahead_tree_block(fs_info, block2);
2263 }
2264
2265
2266 /*
2267  * when we walk down the tree, it is usually safe to unlock the higher layers
2268  * in the tree.  The exceptions are when our path goes through slot 0, because
2269  * operations on the tree might require changing key pointers higher up in the
2270  * tree.
2271  *
2272  * callers might also have set path->keep_locks, which tells this code to keep
2273  * the lock if the path points to the last slot in the block.  This is part of
2274  * walking through the tree, and selecting the next slot in the higher block.
2275  *
2276  * lowest_unlock sets the lowest level in the tree we're allowed to unlock.  so
2277  * if lowest_unlock is 1, level 0 won't be unlocked
2278  */
2279 static noinline void unlock_up(struct btrfs_path *path, int level,
2280                                int lowest_unlock, int min_write_lock_level,
2281                                int *write_lock_level)
2282 {
2283         int i;
2284         int skip_level = level;
2285         int no_skips = 0;
2286         struct extent_buffer *t;
2287
2288         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
2289                 if (!path->nodes[i])
2290                         break;
2291                 if (!path->locks[i])
2292                         break;
2293                 if (!no_skips && path->slots[i] == 0) {
2294                         skip_level = i + 1;
2295                         continue;
2296                 }
2297                 if (!no_skips && path->keep_locks) {
2298                         u32 nritems;
2299                         t = path->nodes[i];
2300                         nritems = btrfs_header_nritems(t);
2301                         if (nritems < 1 || path->slots[i] >= nritems - 1) {
2302                                 skip_level = i + 1;
2303                                 continue;
2304                         }
2305                 }
2306                 if (skip_level < i && i >= lowest_unlock)
2307                         no_skips = 1;
2308
2309                 t = path->nodes[i];
2310                 if (i >= lowest_unlock && i > skip_level) {
2311                         btrfs_tree_unlock_rw(t, path->locks[i]);
2312                         path->locks[i] = 0;
2313                         if (write_lock_level &&
2314                             i > min_write_lock_level &&
2315                             i <= *write_lock_level) {
2316                                 *write_lock_level = i - 1;
2317                         }
2318                 }
2319         }
2320 }
2321
2322 /*
2323  * helper function for btrfs_search_slot.  The goal is to find a block
2324  * in cache without setting the path to blocking.  If we find the block
2325  * we return zero and the path is unchanged.
2326  *
2327  * If we can't find the block, we set the path blocking and do some
2328  * reada.  -EAGAIN is returned and the search must be repeated.
2329  */
2330 static int
2331 read_block_for_search(struct btrfs_root *root, struct btrfs_path *p,
2332                       struct extent_buffer **eb_ret, int level, int slot,
2333                       const struct btrfs_key *key)
2334 {
2335         struct btrfs_fs_info *fs_info = root->fs_info;
2336         u64 blocknr;
2337         u64 gen;
2338         struct extent_buffer *tmp;
2339         struct btrfs_key first_key;
2340         int ret;
2341         int parent_level;
2342
2343         blocknr = btrfs_node_blockptr(*eb_ret, slot);
2344         gen = btrfs_node_ptr_generation(*eb_ret, slot);
2345         parent_level = btrfs_header_level(*eb_ret);
2346         btrfs_node_key_to_cpu(*eb_ret, &first_key, slot);
2347
2348         tmp = find_extent_buffer(fs_info, blocknr);
2349         if (tmp) {
2350                 /* first we do an atomic uptodate check */
2351                 if (btrfs_buffer_uptodate(tmp, gen, 1) > 0) {
2352                         /*
2353                          * Do extra check for first_key, eb can be stale due to
2354                          * being cached, read from scrub, or have multiple
2355                          * parents (shared tree blocks).
2356                          */
2357                         if (btrfs_verify_level_key(tmp,
2358                                         parent_level - 1, &first_key, gen)) {
2359                                 free_extent_buffer(tmp);
2360                                 return -EUCLEAN;
2361                         }
2362                         *eb_ret = tmp;
2363                         return 0;
2364                 }
2365
2366                 /* the pages were up to date, but we failed
2367                  * the generation number check.  Do a full
2368                  * read for the generation number that is correct.
2369                  * We must do this without dropping locks so
2370                  * we can trust our generation number
2371                  */
2372                 btrfs_set_path_blocking(p);
2373
2374                 /* now we're allowed to do a blocking uptodate check */
2375                 ret = btrfs_read_buffer(tmp, gen, parent_level - 1, &first_key);
2376                 if (!ret) {
2377                         *eb_ret = tmp;
2378                         return 0;
2379                 }
2380                 free_extent_buffer(tmp);
2381                 btrfs_release_path(p);
2382                 return -EIO;
2383         }
2384
2385         /*
2386          * reduce lock contention at high levels
2387          * of the btree by dropping locks before
2388          * we read.  Don't release the lock on the current
2389          * level because we need to walk this node to figure
2390          * out which blocks to read.
2391          */
2392         btrfs_unlock_up_safe(p, level + 1);
2393         btrfs_set_path_blocking(p);
2394
2395         if (p->reada != READA_NONE)
2396                 reada_for_search(fs_info, p, level, slot, key->objectid);
2397
2398         ret = -EAGAIN;
2399         tmp = read_tree_block(fs_info, blocknr, gen, parent_level - 1,
2400                               &first_key);
2401         if (!IS_ERR(tmp)) {
2402                 /*
2403                  * If the read above didn't mark this buffer up to date,
2404                  * it will never end up being up to date.  Set ret to EIO now
2405                  * and give up so that our caller doesn't loop forever
2406                  * on our EAGAINs.
2407                  */
2408                 if (!extent_buffer_uptodate(tmp))
2409                         ret = -EIO;
2410                 free_extent_buffer(tmp);
2411         } else {
2412                 ret = PTR_ERR(tmp);
2413         }
2414
2415         btrfs_release_path(p);
2416         return ret;
2417 }
2418
2419 /*
2420  * helper function for btrfs_search_slot.  This does all of the checks
2421  * for node-level blocks and does any balancing required based on
2422  * the ins_len.
2423  *
2424  * If no extra work was required, zero is returned.  If we had to
2425  * drop the path, -EAGAIN is returned and btrfs_search_slot must
2426  * start over
2427  */
2428 static int
2429 setup_nodes_for_search(struct btrfs_trans_handle *trans,
2430                        struct btrfs_root *root, struct btrfs_path *p,
2431                        struct extent_buffer *b, int level, int ins_len,
2432                        int *write_lock_level)
2433 {
2434         struct btrfs_fs_info *fs_info = root->fs_info;
2435         int ret;
2436
2437         if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >=
2438             BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 3) {
2439                 int sret;
2440
2441                 if (*write_lock_level < level + 1) {
2442                         *write_lock_level = level + 1;
2443                         btrfs_release_path(p);
2444                         goto again;
2445                 }
2446
2447                 btrfs_set_path_blocking(p);
2448                 reada_for_balance(fs_info, p, level);
2449                 sret = split_node(trans, root, p, level);
2450
2451                 BUG_ON(sret > 0);
2452                 if (sret) {
2453                         ret = sret;
2454                         goto done;
2455                 }
2456                 b = p->nodes[level];
2457         } else if (ins_len < 0 && btrfs_header_nritems(b) <
2458                    BTRFS_NODEPTRS_PER_BLOCK(fs_info) / 2) {
2459                 int sret;
2460
2461                 if (*write_lock_level < level + 1) {
2462                         *write_lock_level = level + 1;
2463                         btrfs_release_path(p);
2464                         goto again;
2465                 }
2466
2467                 btrfs_set_path_blocking(p);
2468                 reada_for_balance(fs_info, p, level);
2469                 sret = balance_level(trans, root, p, level);
2470
2471                 if (sret) {
2472                         ret = sret;
2473                         goto done;
2474                 }
2475                 b = p->nodes[level];
2476                 if (!b) {
2477                         btrfs_release_path(p);
2478                         goto again;
2479                 }
2480                 BUG_ON(btrfs_header_nritems(b) == 1);
2481         }
2482         return 0;
2483
2484 again:
2485         ret = -EAGAIN;
2486 done:
2487         return ret;
2488 }
2489
2490 int btrfs_find_item(struct btrfs_root *fs_root, struct btrfs_path *path,
2491                 u64 iobjectid, u64 ioff, u8 key_type,
2492                 struct btrfs_key *found_key)
2493 {
2494         int ret;
2495         struct btrfs_key key;
2496         struct extent_buffer *eb;
2497
2498         ASSERT(path);
2499         ASSERT(found_key);
2500
2501         key.type = key_type;
2502         key.objectid = iobjectid;
2503         key.offset = ioff;
2504
2505         ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
2506         if (ret < 0)
2507                 return ret;
2508
2509         eb = path->nodes[0];
2510         if (ret && path->slots[0] >= btrfs_header_nritems(eb)) {
2511                 ret = btrfs_next_leaf(fs_root, path);
2512                 if (ret)
2513                         return ret;
2514                 eb = path->nodes[0];
2515         }
2516
2517         btrfs_item_key_to_cpu(eb, found_key, path->slots[0]);
2518         if (found_key->type != key.type ||
2519                         found_key->objectid != key.objectid)
2520                 return 1;
2521
2522         return 0;
2523 }
2524
2525 static struct extent_buffer *btrfs_search_slot_get_root(struct btrfs_root *root,
2526                                                         struct btrfs_path *p,
2527                                                         int write_lock_level)
2528 {
2529         struct btrfs_fs_info *fs_info = root->fs_info;
2530         struct extent_buffer *b;
2531         int root_lock;
2532         int level = 0;
2533
2534         /* We try very hard to do read locks on the root */
2535         root_lock = BTRFS_READ_LOCK;
2536
2537         if (p->search_commit_root) {
2538                 /*
2539                  * The commit roots are read only so we always do read locks,
2540                  * and we always must hold the commit_root_sem when doing
2541                  * searches on them, the only exception is send where we don't
2542                  * want to block transaction commits for a long time, so
2543                  * we need to clone the commit root in order to avoid races
2544                  * with transaction commits that create a snapshot of one of
2545                  * the roots used by a send operation.
2546                  */
2547                 if (p->need_commit_sem) {
2548                         down_read(&fs_info->commit_root_sem);
2549                         b = btrfs_clone_extent_buffer(root->commit_root);
2550                         up_read(&fs_info->commit_root_sem);
2551                         if (!b)
2552                                 return ERR_PTR(-ENOMEM);
2553
2554                 } else {
2555                         b = root->commit_root;
2556                         atomic_inc(&b->refs);
2557                 }
2558                 level = btrfs_header_level(b);
2559                 /*
2560                  * Ensure that all callers have set skip_locking when
2561                  * p->search_commit_root = 1.
2562                  */
2563                 ASSERT(p->skip_locking == 1);
2564
2565                 goto out;
2566         }
2567
2568         if (p->skip_locking) {
2569                 b = btrfs_root_node(root);
2570                 level = btrfs_header_level(b);
2571                 goto out;
2572         }
2573
2574         /*
2575          * If the level is set to maximum, we can skip trying to get the read
2576          * lock.
2577          */
2578         if (write_lock_level < BTRFS_MAX_LEVEL) {
2579                 /*
2580                  * We don't know the level of the root node until we actually
2581                  * have it read locked
2582                  */
2583                 b = btrfs_read_lock_root_node(root);
2584                 level = btrfs_header_level(b);
2585                 if (level > write_lock_level)
2586                         goto out;
2587
2588                 /* Whoops, must trade for write lock */
2589                 btrfs_tree_read_unlock(b);
2590                 free_extent_buffer(b);
2591         }
2592
2593         b = btrfs_lock_root_node(root);
2594         root_lock = BTRFS_WRITE_LOCK;
2595
2596         /* The level might have changed, check again */
2597         level = btrfs_header_level(b);
2598
2599 out:
2600         p->nodes[level] = b;
2601         if (!p->skip_locking)
2602                 p->locks[level] = root_lock;
2603         /*
2604          * Callers are responsible for dropping b's references.
2605          */
2606         return b;
2607 }
2608
2609
2610 /*
2611  * btrfs_search_slot - look for a key in a tree and perform necessary
2612  * modifications to preserve tree invariants.
2613  *
2614  * @trans:      Handle of transaction, used when modifying the tree
2615  * @p:          Holds all btree nodes along the search path
2616  * @root:       The root node of the tree
2617  * @key:        The key we are looking for
2618  * @ins_len:    Indicates purpose of search, for inserts it is 1, for
2619  *              deletions it's -1. 0 for plain searches
2620  * @cow:        boolean should CoW operations be performed. Must always be 1
2621  *              when modifying the tree.
2622  *
2623  * If @ins_len > 0, nodes and leaves will be split as we walk down the tree.
2624  * If @ins_len < 0, nodes will be merged as we walk down the tree (if possible)
2625  *
2626  * If @key is found, 0 is returned and you can find the item in the leaf level
2627  * of the path (level 0)
2628  *
2629  * If @key isn't found, 1 is returned and the leaf level of the path (level 0)
2630  * points to the slot where it should be inserted
2631  *
2632  * If an error is encountered while searching the tree a negative error number
2633  * is returned
2634  */
2635 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2636                       const struct btrfs_key *key, struct btrfs_path *p,
2637                       int ins_len, int cow)
2638 {
2639         struct extent_buffer *b;
2640         int slot;
2641         int ret;
2642         int err;
2643         int level;
2644         int lowest_unlock = 1;
2645         /* everything at write_lock_level or lower must be write locked */
2646         int write_lock_level = 0;
2647         u8 lowest_level = 0;
2648         int min_write_lock_level;
2649         int prev_cmp;
2650
2651         lowest_level = p->lowest_level;
2652         WARN_ON(lowest_level && ins_len > 0);
2653         WARN_ON(p->nodes[0] != NULL);
2654         BUG_ON(!cow && ins_len);
2655
2656         if (ins_len < 0) {
2657                 lowest_unlock = 2;
2658
2659                 /* when we are removing items, we might have to go up to level
2660                  * two as we update tree pointers  Make sure we keep write
2661                  * for those levels as well
2662                  */
2663                 write_lock_level = 2;
2664         } else if (ins_len > 0) {
2665                 /*
2666                  * for inserting items, make sure we have a write lock on
2667                  * level 1 so we can update keys
2668                  */
2669                 write_lock_level = 1;
2670         }
2671
2672         if (!cow)
2673                 write_lock_level = -1;
2674
2675         if (cow && (p->keep_locks || p->lowest_level))
2676                 write_lock_level = BTRFS_MAX_LEVEL;
2677
2678         min_write_lock_level = write_lock_level;
2679
2680 again:
2681         prev_cmp = -1;
2682         b = btrfs_search_slot_get_root(root, p, write_lock_level);
2683         if (IS_ERR(b)) {
2684                 ret = PTR_ERR(b);
2685                 goto done;
2686         }
2687
2688         while (b) {
2689                 int dec = 0;
2690
2691                 level = btrfs_header_level(b);
2692
2693                 if (cow) {
2694                         bool last_level = (level == (BTRFS_MAX_LEVEL - 1));
2695
2696                         /*
2697                          * if we don't really need to cow this block
2698                          * then we don't want to set the path blocking,
2699                          * so we test it here
2700                          */
2701                         if (!should_cow_block(trans, root, b)) {
2702                                 trans->dirty = true;
2703                                 goto cow_done;
2704                         }
2705
2706                         /*
2707                          * must have write locks on this node and the
2708                          * parent
2709                          */
2710                         if (level > write_lock_level ||
2711                             (level + 1 > write_lock_level &&
2712                             level + 1 < BTRFS_MAX_LEVEL &&
2713                             p->nodes[level + 1])) {
2714                                 write_lock_level = level + 1;
2715                                 btrfs_release_path(p);
2716                                 goto again;
2717                         }
2718
2719                         btrfs_set_path_blocking(p);
2720                         if (last_level)
2721                                 err = btrfs_cow_block(trans, root, b, NULL, 0,
2722                                                       &b);
2723                         else
2724                                 err = btrfs_cow_block(trans, root, b,
2725                                                       p->nodes[level + 1],
2726                                                       p->slots[level + 1], &b);
2727                         if (err) {
2728                                 ret = err;
2729                                 goto done;
2730                         }
2731                 }
2732 cow_done:
2733                 p->nodes[level] = b;
2734                 /*
2735                  * Leave path with blocking locks to avoid massive
2736                  * lock context switch, this is made on purpose.
2737                  */
2738
2739                 /*
2740                  * we have a lock on b and as long as we aren't changing
2741                  * the tree, there is no way to for the items in b to change.
2742                  * It is safe to drop the lock on our parent before we
2743                  * go through the expensive btree search on b.
2744                  *
2745                  * If we're inserting or deleting (ins_len != 0), then we might
2746                  * be changing slot zero, which may require changing the parent.
2747                  * So, we can't drop the lock until after we know which slot
2748                  * we're operating on.
2749                  */
2750                 if (!ins_len && !p->keep_locks) {
2751                         int u = level + 1;
2752
2753                         if (u < BTRFS_MAX_LEVEL && p->locks[u]) {
2754                                 btrfs_tree_unlock_rw(p->nodes[u], p->locks[u]);
2755                                 p->locks[u] = 0;
2756                         }
2757                 }
2758
2759                 /*
2760                  * If btrfs_bin_search returns an exact match (prev_cmp == 0)
2761                  * we can safely assume the target key will always be in slot 0
2762                  * on lower levels due to the invariants BTRFS' btree provides,
2763                  * namely that a btrfs_key_ptr entry always points to the
2764                  * lowest key in the child node, thus we can skip searching
2765                  * lower levels
2766                  */
2767                 if (prev_cmp == 0) {
2768                         slot = 0;
2769                         ret = 0;
2770                 } else {
2771                         ret = btrfs_bin_search(b, key, &slot);
2772                         prev_cmp = ret;
2773                         if (ret < 0)
2774                                 goto done;
2775                 }
2776
2777                 if (level == 0) {
2778                         p->slots[level] = slot;
2779                         if (ins_len > 0 &&
2780                             btrfs_leaf_free_space(b) < ins_len) {
2781                                 if (write_lock_level < 1) {
2782                                         write_lock_level = 1;
2783                                         btrfs_release_path(p);
2784                                         goto again;
2785                                 }
2786
2787                                 btrfs_set_path_blocking(p);
2788                                 err = split_leaf(trans, root, key,
2789                                                  p, ins_len, ret == 0);
2790
2791                                 BUG_ON(err > 0);
2792                                 if (err) {
2793                                         ret = err;
2794                                         goto done;
2795                                 }
2796                         }
2797                         if (!p->search_for_split)
2798                                 unlock_up(p, level, lowest_unlock,
2799                                           min_write_lock_level, NULL);
2800                         goto done;
2801                 }
2802                 if (ret && slot > 0) {
2803                         dec = 1;
2804                         slot--;
2805                 }
2806                 p->slots[level] = slot;
2807                 err = setup_nodes_for_search(trans, root, p, b, level, ins_len,
2808                                              &write_lock_level);
2809                 if (err == -EAGAIN)
2810                         goto again;
2811                 if (err) {
2812                         ret = err;
2813                         goto done;
2814                 }
2815                 b = p->nodes[level];
2816                 slot = p->slots[level];
2817
2818                 /*
2819                  * Slot 0 is special, if we change the key we have to update
2820                  * the parent pointer which means we must have a write lock on
2821                  * the parent
2822                  */
2823                 if (slot == 0 && ins_len && write_lock_level < level + 1) {
2824                         write_lock_level = level + 1;
2825                         btrfs_release_path(p);
2826                         goto again;
2827                 }
2828
2829                 unlock_up(p, level, lowest_unlock, min_write_lock_level,
2830                           &write_lock_level);
2831
2832                 if (level == lowest_level) {
2833                         if (dec)
2834                                 p->slots[level]++;
2835                         goto done;
2836                 }
2837
2838                 err = read_block_for_search(root, p, &b, level, slot, key);
2839                 if (err == -EAGAIN)
2840                         goto again;
2841                 if (err) {
2842                         ret = err;
2843                         goto done;
2844                 }
2845
2846                 if (!p->skip_locking) {
2847                         level = btrfs_header_level(b);
2848                         if (level <= write_lock_level) {
2849                                 if (!btrfs_try_tree_write_lock(b)) {
2850                                         btrfs_set_path_blocking(p);
2851                                         btrfs_tree_lock(b);
2852                                 }
2853                                 p->locks[level] = BTRFS_WRITE_LOCK;
2854                         } else {
2855                                 if (!btrfs_tree_read_lock_atomic(b)) {
2856                                         btrfs_set_path_blocking(p);
2857                                         btrfs_tree_read_lock(b);
2858                                 }
2859                                 p->locks[level] = BTRFS_READ_LOCK;
2860                         }
2861                         p->nodes[level] = b;
2862                 }
2863         }
2864         ret = 1;
2865 done:
2866         /*
2867          * we don't really know what they plan on doing with the path
2868          * from here on, so for now just mark it as blocking
2869          */
2870         if (!p->leave_spinning)
2871                 btrfs_set_path_blocking(p);
2872         if (ret < 0 && !p->skip_release_on_error)
2873                 btrfs_release_path(p);
2874         return ret;
2875 }
2876
2877 /*
2878  * Like btrfs_search_slot, this looks for a key in the given tree. It uses the
2879  * current state of the tree together with the operations recorded in the tree
2880  * modification log to search for the key in a previous version of this tree, as
2881  * denoted by the time_seq parameter.
2882  *
2883  * Naturally, there is no support for insert, delete or cow operations.
2884  *
2885  * The resulting path and return value will be set up as if we called
2886  * btrfs_search_slot at that point in time with ins_len and cow both set to 0.
2887  */
2888 int btrfs_search_old_slot(struct btrfs_root *root, const struct btrfs_key *key,
2889                           struct btrfs_path *p, u64 time_seq)
2890 {
2891         struct btrfs_fs_info *fs_info = root->fs_info;
2892         struct extent_buffer *b;
2893         int slot;
2894         int ret;
2895         int err;
2896         int level;
2897         int lowest_unlock = 1;
2898         u8 lowest_level = 0;
2899
2900         lowest_level = p->lowest_level;
2901         WARN_ON(p->nodes[0] != NULL);
2902
2903         if (p->search_commit_root) {
2904                 BUG_ON(time_seq);
2905                 return btrfs_search_slot(NULL, root, key, p, 0, 0);
2906         }
2907
2908 again:
2909         b = get_old_root(root, time_seq);
2910         if (!b) {
2911                 ret = -EIO;
2912                 goto done;
2913         }
2914         level = btrfs_header_level(b);
2915         p->locks[level] = BTRFS_READ_LOCK;
2916
2917         while (b) {
2918                 int dec = 0;
2919
2920                 level = btrfs_header_level(b);
2921                 p->nodes[level] = b;
2922
2923                 /*
2924                  * we have a lock on b and as long as we aren't changing
2925                  * the tree, there is no way to for the items in b to change.
2926                  * It is safe to drop the lock on our parent before we
2927                  * go through the expensive btree search on b.
2928                  */
2929                 btrfs_unlock_up_safe(p, level + 1);
2930
2931                 ret = btrfs_bin_search(b, key, &slot);
2932                 if (ret < 0)
2933                         goto done;
2934
2935                 if (level == 0) {
2936                         p->slots[level] = slot;
2937                         unlock_up(p, level, lowest_unlock, 0, NULL);
2938                         goto done;
2939                 }
2940
2941                 if (ret && slot > 0) {
2942                         dec = 1;
2943                         slot--;
2944                 }
2945                 p->slots[level] = slot;
2946                 unlock_up(p, level, lowest_unlock, 0, NULL);
2947
2948                 if (level == lowest_level) {
2949                         if (dec)
2950                                 p->slots[level]++;
2951                         goto done;
2952                 }
2953
2954                 err = read_block_for_search(root, p, &b, level, slot, key);
2955                 if (err == -EAGAIN)
2956                         goto again;
2957                 if (err) {
2958                         ret = err;
2959                         goto done;
2960                 }
2961
2962                 level = btrfs_header_level(b);
2963                 if (!btrfs_tree_read_lock_atomic(b)) {
2964                         btrfs_set_path_blocking(p);
2965                         btrfs_tree_read_lock(b);
2966                 }
2967                 b = tree_mod_log_rewind(fs_info, p, b, time_seq);
2968                 if (!b) {
2969                         ret = -ENOMEM;
2970                         goto done;
2971                 }
2972                 p->locks[level] = BTRFS_READ_LOCK;
2973                 p->nodes[level] = b;
2974         }
2975         ret = 1;
2976 done:
2977         if (!p->leave_spinning)
2978                 btrfs_set_path_blocking(p);
2979         if (ret < 0)
2980                 btrfs_release_path(p);
2981
2982         return ret;
2983 }
2984
2985 /*
2986  * helper to use instead of search slot if no exact match is needed but
2987  * instead the next or previous item should be returned.
2988  * When find_higher is true, the next higher item is returned, the next lower
2989  * otherwise.
2990  * When return_any and find_higher are both true, and no higher item is found,
2991  * return the next lower instead.
2992  * When return_any is true and find_higher is false, and no lower item is found,
2993  * return the next higher instead.
2994  * It returns 0 if any item is found, 1 if none is found (tree empty), and
2995  * < 0 on error
2996  */
2997 int btrfs_search_slot_for_read(struct btrfs_root *root,
2998                                const struct btrfs_key *key,
2999                                struct btrfs_path *p, int find_higher,
3000                                int return_any)
3001 {
3002         int ret;
3003         struct extent_buffer *leaf;
3004
3005 again:
3006         ret = btrfs_search_slot(NULL, root, key, p, 0, 0);
3007         if (ret <= 0)
3008                 return ret;
3009         /*
3010          * a return value of 1 means the path is at the position where the
3011          * item should be inserted. Normally this is the next bigger item,
3012          * but in case the previous item is the last in a leaf, path points
3013          * to the first free slot in the previous leaf, i.e. at an invalid
3014          * item.
3015          */
3016         leaf = p->nodes[0];
3017
3018         if (find_higher) {
3019                 if (p->slots[0] >= btrfs_header_nritems(leaf)) {
3020                         ret = btrfs_next_leaf(root, p);
3021                         if (ret <= 0)
3022                                 return ret;
3023                         if (!return_any)
3024                                 return 1;
3025                         /*
3026                          * no higher item found, return the next
3027                          * lower instead
3028                          */
3029                         return_any = 0;
3030                         find_higher = 0;
3031                         btrfs_release_path(p);
3032                         goto again;
3033                 }
3034         } else {
3035                 if (p->slots[0] == 0) {
3036                         ret = btrfs_prev_leaf(root, p);
3037                         if (ret < 0)
3038                                 return ret;
3039                         if (!ret) {
3040                                 leaf = p->nodes[0];
3041                                 if (p->slots[0] == btrfs_header_nritems(leaf))
3042                                         p->slots[0]--;
3043                                 return 0;
3044                         }
3045                         if (!return_any)
3046                                 return 1;
3047                         /*
3048                          * no lower item found, return the next
3049                          * higher instead
3050                          */
3051                         return_any = 0;
3052                         find_higher = 1;
3053                         btrfs_release_path(p);
3054                         goto again;
3055                 } else {
3056                         --p->slots[0];
3057                 }
3058         }
3059         return 0;
3060 }
3061
3062 /*
3063  * adjust the pointers going up the tree, starting at level
3064  * making sure the right key of each node is points to 'key'.
3065  * This is used after shifting pointers to the left, so it stops
3066  * fixing up pointers when a given leaf/node is not in slot 0 of the
3067  * higher levels
3068  *
3069  */
3070 static void fixup_low_keys(struct btrfs_path *path,
3071                            struct btrfs_disk_key *key, int level)
3072 {
3073         int i;
3074         struct extent_buffer *t;
3075         int ret;
3076
3077         for (i = level; i < BTRFS_MAX_LEVEL; i++) {
3078                 int tslot = path->slots[i];
3079
3080                 if (!path->nodes[i])
3081                         break;
3082                 t = path->nodes[i];
3083                 ret = tree_mod_log_insert_key(t, tslot, MOD_LOG_KEY_REPLACE,
3084                                 GFP_ATOMIC);
3085                 BUG_ON(ret < 0);
3086                 btrfs_set_node_key(t, key, tslot);
3087                 btrfs_mark_buffer_dirty(path->nodes[i]);
3088                 if (tslot != 0)
3089                         break;
3090         }
3091 }
3092
3093 /*
3094  * update item key.
3095  *
3096  * This function isn't completely safe. It's the caller's responsibility
3097  * that the new key won't break the order
3098  */
3099 void btrfs_set_item_key_safe(struct btrfs_fs_info *fs_info,
3100                              struct btrfs_path *path,
3101                              const struct btrfs_key *new_key)
3102 {
3103         struct btrfs_disk_key disk_key;
3104         struct extent_buffer *eb;
3105         int slot;
3106
3107         eb = path->nodes[0];
3108         slot = path->slots[0];
3109         if (slot > 0) {
3110                 btrfs_item_key(eb, &disk_key, slot - 1);
3111                 if (unlikely(comp_keys(&disk_key, new_key) >= 0)) {
3112                         btrfs_crit(fs_info,
3113                 "slot %u key (%llu %u %llu) new key (%llu %u %llu)",
3114                                    slot, btrfs_disk_key_objectid(&disk_key),
3115                                    btrfs_disk_key_type(&disk_key),
3116                                    btrfs_disk_key_offset(&disk_key),
3117                                    new_key->objectid, new_key->type,
3118                                    new_key->offset);
3119                         btrfs_print_leaf(eb);
3120                         BUG();
3121                 }
3122         }
3123         if (slot < btrfs_header_nritems(eb) - 1) {
3124                 btrfs_item_key(eb, &disk_key, slot + 1);
3125                 if (unlikely(comp_keys(&disk_key, new_key) <= 0)) {
3126                         btrfs_crit(fs_info,
3127                 "slot %u key (%llu %u %llu) new key (%llu %u %llu)",
3128                                    slot, btrfs_disk_key_objectid(&disk_key),
3129                                    btrfs_disk_key_type(&disk_key),
3130                                    btrfs_disk_key_offset(&disk_key),
3131                                    new_key->objectid, new_key->type,
3132                                    new_key->offset);
3133                         btrfs_print_leaf(eb);
3134                         BUG();
3135                 }
3136         }
3137
3138         btrfs_cpu_key_to_disk(&disk_key, new_key);
3139         btrfs_set_item_key(eb, &disk_key, slot);
3140         btrfs_mark_buffer_dirty(eb);
3141         if (slot == 0)
3142                 fixup_low_keys(path, &disk_key, 1);
3143 }
3144
3145 /*
3146  * try to push data from one node into the next node left in the
3147  * tree.
3148  *
3149  * returns 0 if some ptrs were pushed left, < 0 if there was some horrible
3150  * error, and > 0 if there was no room in the left hand block.
3151  */
3152 static int push_node_left(struct btrfs_trans_handle *trans,
3153                           struct extent_buffer *dst,
3154                           struct extent_buffer *src, int empty)
3155 {
3156         struct btrfs_fs_info *fs_info = trans->fs_info;
3157         int push_items = 0;
3158         int src_nritems;
3159         int dst_nritems;
3160         int ret = 0;
3161
3162         src_nritems = btrfs_header_nritems(src);
3163         dst_nritems = btrfs_header_nritems(dst);
3164         push_items = BTRFS_NODEPTRS_PER_BLOCK(fs_info) - dst_nritems;
3165         WARN_ON(btrfs_header_generation(src) != trans->transid);
3166         WARN_ON(btrfs_header_generation(dst) != trans->transid);
3167
3168         if (!empty && src_nritems <= 8)
3169                 return 1;
3170
3171         if (push_items <= 0)
3172                 return 1;
3173
3174         if (empty) {
3175                 push_items = min(src_nritems, push_items);
3176                 if (push_items < src_nritems) {
3177                         /* leave at least 8 pointers in the node if
3178                          * we aren't going to empty it
3179                          */
3180                         if (src_nritems - push_items < 8) {
3181                                 if (push_items <= 8)
3182                                         return 1;
3183                                 push_items -= 8;
3184                         }
3185                 }
3186         } else
3187                 push_items = min(src_nritems - 8, push_items);
3188
3189         ret = tree_mod_log_eb_copy(dst, src, dst_nritems, 0, push_items);
3190         if (ret) {
3191                 btrfs_abort_transaction(trans, ret);
3192                 return ret;
3193         }
3194         copy_extent_buffer(dst, src,
3195                            btrfs_node_key_ptr_offset(dst_nritems),
3196                            btrfs_node_key_ptr_offset(0),
3197                            push_items * sizeof(struct btrfs_key_ptr));
3198
3199         if (push_items < src_nritems) {
3200                 /*
3201                  * Don't call tree_mod_log_insert_move here, key removal was
3202                  * already fully logged by tree_mod_log_eb_copy above.
3203                  */
3204                 memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0),
3205                                       btrfs_node_key_ptr_offset(push_items),
3206                                       (src_nritems - push_items) *
3207                                       sizeof(struct btrfs_key_ptr));
3208         }
3209         btrfs_set_header_nritems(src, src_nritems - push_items);
3210         btrfs_set_header_nritems(dst, dst_nritems + push_items);
3211         btrfs_mark_buffer_dirty(src);
3212         btrfs_mark_buffer_dirty(dst);
3213
3214         return ret;
3215 }
3216
3217 /*
3218  * try to push data from one node into the next node right in the
3219  * tree.
3220  *
3221  * returns 0 if some ptrs were pushed, < 0 if there was some horrible
3222  * error, and > 0 if there was no room in the right hand block.
3223  *
3224  * this will  only push up to 1/2 the contents of the left node over
3225  */
3226 static int balance_node_right(struct btrfs_trans_handle *trans,
3227                               struct extent_buffer *dst,
3228                               struct extent_buffer *src)
3229 {
3230         struct btrfs_fs_info *fs_info = trans->fs_info;
3231         int push_items = 0;
3232         int max_push;
3233         int src_nritems;
3234         int dst_nritems;
3235         int ret = 0;
3236
3237         WARN_ON(btrfs_header_generation(src) != trans->transid);
3238         WARN_ON(btrfs_header_generation(dst) != trans->transid);
3239
3240         src_nritems = btrfs_header_nritems(src);
3241         dst_nritems = btrfs_header_nritems(dst);
3242         push_items = BTRFS_NODEPTRS_PER_BLOCK(fs_info) - dst_nritems;
3243         if (push_items <= 0)
3244                 return 1;
3245
3246         if (src_nritems < 4)
3247                 return 1;
3248
3249         max_push = src_nritems / 2 + 1;
3250         /* don't try to empty the node */
3251         if (max_push >= src_nritems)
3252                 return 1;
3253
3254         if (max_push < push_items)
3255                 push_items = max_push;
3256
3257         ret = tree_mod_log_insert_move(dst, push_items, 0, dst_nritems);
3258         BUG_ON(ret < 0);
3259         memmove_extent_buffer(dst, btrfs_node_key_ptr_offset(push_items),
3260                                       btrfs_node_key_ptr_offset(0),
3261                                       (dst_nritems) *
3262                                       sizeof(struct btrfs_key_ptr));
3263
3264         ret = tree_mod_log_eb_copy(dst, src, 0, src_nritems - push_items,
3265                                    push_items);
3266         if (ret) {
3267                 btrfs_abort_transaction(trans, ret);
3268                 return ret;
3269         }
3270         copy_extent_buffer(dst, src,
3271                            btrfs_node_key_ptr_offset(0),
3272                            btrfs_node_key_ptr_offset(src_nritems - push_items),
3273                            push_items * sizeof(struct btrfs_key_ptr));
3274
3275         btrfs_set_header_nritems(src, src_nritems - push_items);
3276         btrfs_set_header_nritems(dst, dst_nritems + push_items);
3277
3278         btrfs_mark_buffer_dirty(src);
3279         btrfs_mark_buffer_dirty(dst);
3280
3281         return ret;
3282 }
3283
3284 /*
3285  * helper function to insert a new root level in the tree.
3286  * A new node is allocated, and a single item is inserted to
3287  * point to the existing root
3288  *
3289  * returns zero on success or < 0 on failure.
3290  */
3291 static noinline int insert_new_root(struct btrfs_trans_handle *trans,
3292                            struct btrfs_root *root,
3293                            struct btrfs_path *path, int level)
3294 {
3295         struct btrfs_fs_info *fs_info = root->fs_info;
3296         u64 lower_gen;
3297         struct extent_buffer *lower;
3298         struct extent_buffer *c;
3299         struct extent_buffer *old;
3300         struct btrfs_disk_key lower_key;
3301         int ret;
3302
3303         BUG_ON(path->nodes[level]);
3304         BUG_ON(path->nodes[level-1] != root->node);
3305
3306         lower = path->nodes[level-1];
3307         if (level == 1)
3308                 btrfs_item_key(lower, &lower_key, 0);
3309         else
3310                 btrfs_node_key(lower, &lower_key, 0);
3311
3312         c = alloc_tree_block_no_bg_flush(trans, root, 0, &lower_key, level,
3313                                          root->node->start, 0);
3314         if (IS_ERR(c))
3315                 return PTR_ERR(c);
3316
3317         root_add_used(root, fs_info->nodesize);
3318
3319         btrfs_set_header_nritems(c, 1);
3320         btrfs_set_node_key(c, &lower_key, 0);
3321         btrfs_set_node_blockptr(c, 0, lower->start);
3322         lower_gen = btrfs_header_generation(lower);
3323         WARN_ON(lower_gen != trans->transid);
3324
3325         btrfs_set_node_ptr_generation(c, 0, lower_gen);
3326
3327         btrfs_mark_buffer_dirty(c);
3328
3329         old = root->node;
3330         ret = tree_mod_log_insert_root(root->node, c, 0);
3331         BUG_ON(ret < 0);
3332         rcu_assign_pointer(root->node, c);
3333
3334         /* the super has an extra ref to root->node */
3335         free_extent_buffer(old);
3336
3337         add_root_to_dirty_list(root);
3338         atomic_inc(&c->refs);
3339         path->nodes[level] = c;
3340         path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
3341         path->slots[level] = 0;
3342         return 0;
3343 }
3344
3345 /*
3346  * worker function to insert a single pointer in a node.
3347  * the node should have enough room for the pointer already
3348  *
3349  * slot and level indicate where you want the key to go, and
3350  * blocknr is the block the key points to.
3351  */
3352 static void insert_ptr(struct btrfs_trans_handle *trans,
3353                        struct btrfs_path *path,
3354                        struct btrfs_disk_key *key, u64 bytenr,
3355                        int slot, int level)
3356 {
3357         struct extent_buffer *lower;
3358         int nritems;
3359         int ret;
3360
3361         BUG_ON(!path->nodes[level]);
3362         btrfs_assert_tree_locked(path->nodes[level]);
3363         lower = path->nodes[level];
3364         nritems = btrfs_header_nritems(lower);
3365         BUG_ON(slot > nritems);
3366         BUG_ON(nritems == BTRFS_NODEPTRS_PER_BLOCK(trans->fs_info));
3367         if (slot != nritems) {
3368                 if (level) {
3369                         ret = tree_mod_log_insert_move(lower, slot + 1, slot,
3370                                         nritems - slot);
3371                         BUG_ON(ret < 0);
3372                 }
3373                 memmove_extent_buffer(lower,
3374                               btrfs_node_key_ptr_offset(slot + 1),
3375                               btrfs_node_key_ptr_offset(slot),
3376                               (nritems - slot) * sizeof(struct btrfs_key_ptr));
3377         }
3378         if (level) {
3379                 ret = tree_mod_log_insert_key(lower, slot, MOD_LOG_KEY_ADD,
3380                                 GFP_NOFS);
3381                 BUG_ON(ret < 0);
3382         }
3383         btrfs_set_node_key(lower, key, slot);
3384         btrfs_set_node_blockptr(lower, slot, bytenr);
3385         WARN_ON(trans->transid == 0);
3386         btrfs_set_node_ptr_generation(lower, slot, trans->transid);
3387         btrfs_set_header_nritems(lower, nritems + 1);
3388         btrfs_mark_buffer_dirty(lower);
3389 }
3390
3391 /*
3392  * split the node at the specified level in path in two.
3393  * The path is corrected to point to the appropriate node after the split
3394  *
3395  * Before splitting this tries to make some room in the node by pushing
3396  * left and right, if either one works, it returns right away.
3397  *
3398  * returns 0 on success and < 0 on failure
3399  */
3400 static noinline int split_node(struct btrfs_trans_handle *trans,
3401                                struct btrfs_root *root,
3402                                struct btrfs_path *path, int level)
3403 {
3404         struct btrfs_fs_info *fs_info = root->fs_info;
3405         struct extent_buffer *c;
3406         struct extent_buffer *split;
3407         struct btrfs_disk_key disk_key;
3408         int mid;
3409         int ret;
3410         u32 c_nritems;
3411
3412         c = path->nodes[level];
3413         WARN_ON(btrfs_header_generation(c) != trans->transid);
3414         if (c == root->node) {
3415                 /*
3416                  * trying to split the root, lets make a new one
3417                  *
3418                  * tree mod log: We don't log_removal old root in
3419                  * insert_new_root, because that root buffer will be kept as a
3420                  * normal node. We are going to log removal of half of the
3421                  * elements below with tree_mod_log_eb_copy. We're holding a
3422                  * tree lock on the buffer, which is why we cannot race with
3423                  * other tree_mod_log users.
3424                  */
3425                 ret = insert_new_root(trans, root, path, level + 1);
3426                 if (ret)
3427                         return ret;
3428         } else {
3429                 ret = push_nodes_for_insert(trans, root, path, level);
3430                 c = path->nodes[level];
3431                 if (!ret && btrfs_header_nritems(c) <
3432                     BTRFS_NODEPTRS_PER_BLOCK(fs_info) - 3)
3433                         return 0;
3434                 if (ret < 0)
3435                         return ret;
3436         }
3437
3438         c_nritems = btrfs_header_nritems(c);
3439         mid = (c_nritems + 1) / 2;
3440         btrfs_node_key(c, &disk_key, mid);
3441
3442         split = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, level,
3443                                              c->start, 0);
3444         if (IS_ERR(split))
3445                 return PTR_ERR(split);
3446
3447         root_add_used(root, fs_info->nodesize);
3448         ASSERT(btrfs_header_level(c) == level);
3449
3450         ret = tree_mod_log_eb_copy(split, c, 0, mid, c_nritems - mid);
3451         if (ret) {
3452                 btrfs_abort_transaction(trans, ret);
3453                 return ret;
3454         }
3455         copy_extent_buffer(split, c,
3456                            btrfs_node_key_ptr_offset(0),
3457                            btrfs_node_key_ptr_offset(mid),
3458                            (c_nritems - mid) * sizeof(struct btrfs_key_ptr));
3459         btrfs_set_header_nritems(split, c_nritems - mid);
3460         btrfs_set_header_nritems(c, mid);
3461         ret = 0;
3462
3463         btrfs_mark_buffer_dirty(c);
3464         btrfs_mark_buffer_dirty(split);
3465
3466         insert_ptr(trans, path, &disk_key, split->start,
3467                    path->slots[level + 1] + 1, level + 1);
3468
3469         if (path->slots[level] >= mid) {
3470                 path->slots[level] -= mid;
3471                 btrfs_tree_unlock(c);
3472                 free_extent_buffer(c);
3473                 path->nodes[level] = split;
3474                 path->slots[level + 1] += 1;
3475         } else {
3476                 btrfs_tree_unlock(split);
3477                 free_extent_buffer(split);
3478         }
3479         return ret;
3480 }
3481
3482 /*
3483  * how many bytes are required to store the items in a leaf.  start
3484  * and nr indicate which items in the leaf to check.  This totals up the
3485  * space used both by the item structs and the item data
3486  */
3487 static int leaf_space_used(struct extent_buffer *l, int start, int nr)
3488 {
3489         struct btrfs_item *start_item;
3490         struct btrfs_item *end_item;
3491         int data_len;
3492         int nritems = btrfs_header_nritems(l);
3493         int end = min(nritems, start + nr) - 1;
3494
3495         if (!nr)
3496                 return 0;
3497         start_item = btrfs_item_nr(start);
3498         end_item = btrfs_item_nr(end);
3499         data_len = btrfs_item_offset(l, start_item) +
3500                    btrfs_item_size(l, start_item);
3501         data_len = data_len - btrfs_item_offset(l, end_item);
3502         data_len += sizeof(struct btrfs_item) * nr;
3503         WARN_ON(data_len < 0);
3504         return data_len;
3505 }
3506
3507 /*
3508  * The space between the end of the leaf items and
3509  * the start of the leaf data.  IOW, how much room
3510  * the leaf has left for both items and data
3511  */
3512 noinline int btrfs_leaf_free_space(struct extent_buffer *leaf)
3513 {
3514         struct btrfs_fs_info *fs_info = leaf->fs_info;
3515         int nritems = btrfs_header_nritems(leaf);
3516         int ret;
3517
3518         ret = BTRFS_LEAF_DATA_SIZE(fs_info) - leaf_space_used(leaf, 0, nritems);
3519         if (ret < 0) {
3520                 btrfs_crit(fs_info,
3521                            "leaf free space ret %d, leaf data size %lu, used %d nritems %d",
3522                            ret,
3523                            (unsigned long) BTRFS_LEAF_DATA_SIZE(fs_info),
3524                            leaf_space_used(leaf, 0, nritems), nritems);
3525         }
3526         return ret;
3527 }
3528
3529 /*
3530  * min slot controls the lowest index we're willing to push to the
3531  * right.  We'll push up to and including min_slot, but no lower
3532  */
3533 static noinline int __push_leaf_right(struct btrfs_path *path,
3534                                       int data_size, int empty,
3535                                       struct extent_buffer *right,
3536                                       int free_space, u32 left_nritems,
3537                                       u32 min_slot)
3538 {
3539         struct btrfs_fs_info *fs_info = right->fs_info;
3540         struct extent_buffer *left = path->nodes[0];
3541         struct extent_buffer *upper = path->nodes[1];
3542         struct btrfs_map_token token;
3543         struct btrfs_disk_key disk_key;
3544         int slot;
3545         u32 i;
3546         int push_space = 0;
3547         int push_items = 0;
3548         struct btrfs_item *item;
3549         u32 nr;
3550         u32 right_nritems;
3551         u32 data_end;
3552         u32 this_item_size;
3553
3554         if (empty)
3555                 nr = 0;
3556         else
3557                 nr = max_t(u32, 1, min_slot);
3558
3559         if (path->slots[0] >= left_nritems)
3560                 push_space += data_size;
3561
3562         slot = path->slots[1];
3563         i = left_nritems - 1;
3564         while (i >= nr) {
3565                 item = btrfs_item_nr(i);
3566
3567                 if (!empty && push_items > 0) {
3568                         if (path->slots[0] > i)
3569                                 break;
3570                         if (path->slots[0] == i) {
3571                                 int space = btrfs_leaf_free_space(left);
3572
3573                                 if (space + push_space * 2 > free_space)
3574                                         break;
3575                         }
3576                 }
3577
3578                 if (path->slots[0] == i)
3579                         push_space += data_size;
3580
3581                 this_item_size = btrfs_item_size(left, item);
3582                 if (this_item_size + sizeof(*item) + push_space > free_space)
3583                         break;
3584
3585                 push_items++;
3586                 push_space += this_item_size + sizeof(*item);
3587                 if (i == 0)
3588                         break;
3589                 i--;
3590         }
3591
3592         if (push_items == 0)
3593                 goto out_unlock;
3594
3595         WARN_ON(!empty && push_items == left_nritems);
3596
3597         /* push left to right */
3598         right_nritems = btrfs_header_nritems(right);
3599
3600         push_space = btrfs_item_end_nr(left, left_nritems - push_items);
3601         push_space -= leaf_data_end(left);
3602
3603         /* make room in the right data area */
3604         data_end = leaf_data_end(right);
3605         memmove_extent_buffer(right,
3606                               BTRFS_LEAF_DATA_OFFSET + data_end - push_space,
3607                               BTRFS_LEAF_DATA_OFFSET + data_end,
3608                               BTRFS_LEAF_DATA_SIZE(fs_info) - data_end);
3609
3610         /* copy from the left data area */
3611         copy_extent_buffer(right, left, BTRFS_LEAF_DATA_OFFSET +
3612                      BTRFS_LEAF_DATA_SIZE(fs_info) - push_space,
3613                      BTRFS_LEAF_DATA_OFFSET + leaf_data_end(left),
3614                      push_space);
3615
3616         memmove_extent_buffer(right, btrfs_item_nr_offset(push_items),
3617                               btrfs_item_nr_offset(0),
3618                               right_nritems * sizeof(struct btrfs_item));
3619
3620         /* copy the items from left to right */
3621         copy_extent_buffer(right, left, btrfs_item_nr_offset(0),
3622                    btrfs_item_nr_offset(left_nritems - push_items),
3623                    push_items * sizeof(struct btrfs_item));
3624
3625         /* update the item pointers */
3626         btrfs_init_map_token(&token, right);
3627         right_nritems += push_items;
3628         btrfs_set_header_nritems(right, right_nritems);
3629         push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
3630         for (i = 0; i < right_nritems; i++) {
3631                 item = btrfs_item_nr(i);
3632                 push_space -= btrfs_token_item_size(&token, item);
3633                 btrfs_set_token_item_offset(&token, item, push_space);
3634         }
3635
3636         left_nritems -= push_items;
3637         btrfs_set_header_nritems(left, left_nritems);
3638
3639         if (left_nritems)
3640                 btrfs_mark_buffer_dirty(left);
3641         else
3642                 btrfs_clean_tree_block(left);
3643
3644         btrfs_mark_buffer_dirty(right);
3645
3646         btrfs_item_key(right, &disk_key, 0);
3647         btrfs_set_node_key(upper, &disk_key, slot + 1);
3648         btrfs_mark_buffer_dirty(upper);
3649
3650         /* then fixup the leaf pointer in the path */
3651         if (path->slots[0] >= left_nritems) {
3652                 path->slots[0] -= left_nritems;
3653                 if (btrfs_header_nritems(path->nodes[0]) == 0)
3654                         btrfs_clean_tree_block(path->nodes[0]);
3655                 btrfs_tree_unlock(path->nodes[0]);
3656                 free_extent_buffer(path->nodes[0]);
3657                 path->nodes[0] = right;
3658                 path->slots[1] += 1;
3659         } else {
3660                 btrfs_tree_unlock(right);
3661                 free_extent_buffer(right);
3662         }
3663         return 0;
3664
3665 out_unlock:
3666         btrfs_tree_unlock(right);
3667         free_extent_buffer(right);
3668         return 1;
3669 }
3670
3671 /*
3672  * push some data in the path leaf to the right, trying to free up at
3673  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3674  *
3675  * returns 1 if the push failed because the other node didn't have enough
3676  * room, 0 if everything worked out and < 0 if there were major errors.
3677  *
3678  * this will push starting from min_slot to the end of the leaf.  It won't
3679  * push any slot lower than min_slot
3680  */
3681 static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
3682                            *root, struct btrfs_path *path,
3683                            int min_data_size, int data_size,
3684                            int empty, u32 min_slot)
3685 {
3686         struct extent_buffer *left = path->nodes[0];
3687         struct extent_buffer *right;
3688         struct extent_buffer *upper;
3689         int slot;
3690         int free_space;
3691         u32 left_nritems;
3692         int ret;
3693
3694         if (!path->nodes[1])
3695                 return 1;
3696
3697         slot = path->slots[1];
3698         upper = path->nodes[1];
3699         if (slot >= btrfs_header_nritems(upper) - 1)
3700                 return 1;
3701
3702         btrfs_assert_tree_locked(path->nodes[1]);
3703
3704         right = btrfs_read_node_slot(upper, slot + 1);
3705         /*
3706          * slot + 1 is not valid or we fail to read the right node,
3707          * no big deal, just return.
3708          */
3709         if (IS_ERR(right))
3710                 return 1;
3711
3712         btrfs_tree_lock(right);
3713         btrfs_set_lock_blocking_write(right);
3714
3715         free_space = btrfs_leaf_free_space(right);
3716         if (free_space < data_size)
3717                 goto out_unlock;
3718
3719         /* cow and double check */
3720         ret = btrfs_cow_block(trans, root, right, upper,
3721                               slot + 1, &right);
3722         if (ret)
3723                 goto out_unlock;
3724
3725         free_space = btrfs_leaf_free_space(right);
3726         if (free_space < data_size)
3727                 goto out_unlock;
3728
3729         left_nritems = btrfs_header_nritems(left);
3730         if (left_nritems == 0)
3731                 goto out_unlock;
3732
3733         if (path->slots[0] == left_nritems && !empty) {
3734                 /* Key greater than all keys in the leaf, right neighbor has
3735                  * enough room for it and we're not emptying our leaf to delete
3736                  * it, therefore use right neighbor to insert the new item and
3737                  * no need to touch/dirty our left leaf. */
3738                 btrfs_tree_unlock(left);
3739                 free_extent_buffer(left);
3740                 path->nodes[0] = right;
3741                 path->slots[0] = 0;
3742                 path->slots[1]++;
3743                 return 0;
3744         }
3745
3746         return __push_leaf_right(path, min_data_size, empty,
3747                                 right, free_space, left_nritems, min_slot);
3748 out_unlock:
3749         btrfs_tree_unlock(right);
3750         free_extent_buffer(right);
3751         return 1;
3752 }
3753
3754 /*
3755  * push some data in the path leaf to the left, trying to free up at
3756  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3757  *
3758  * max_slot can put a limit on how far into the leaf we'll push items.  The
3759  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us do all the
3760  * items
3761  */
3762 static noinline int __push_leaf_left(struct btrfs_path *path, int data_size,
3763                                      int empty, struct extent_buffer *left,
3764                                      int free_space, u32 right_nritems,
3765                                      u32 max_slot)
3766 {
3767         struct btrfs_fs_info *fs_info = left->fs_info;
3768         struct btrfs_disk_key disk_key;
3769         struct extent_buffer *right = path->nodes[0];
3770         int i;
3771         int push_space = 0;
3772         int push_items = 0;
3773         struct btrfs_item *item;
3774         u32 old_left_nritems;
3775         u32 nr;
3776         int ret = 0;
3777         u32 this_item_size;
3778         u32 old_left_item_size;
3779         struct btrfs_map_token token;
3780
3781         if (empty)
3782                 nr = min(right_nritems, max_slot);
3783         else
3784                 nr = min(right_nritems - 1, max_slot);
3785
3786         for (i = 0; i < nr; i++) {
3787                 item = btrfs_item_nr(i);
3788
3789                 if (!empty && push_items > 0) {
3790                         if (path->slots[0] < i)
3791                                 break;
3792                         if (path->slots[0] == i) {
3793                                 int space = btrfs_leaf_free_space(right);
3794
3795                                 if (space + push_space * 2 > free_space)
3796                                         break;
3797                         }
3798                 }
3799
3800                 if (path->slots[0] == i)
3801                         push_space += data_size;
3802
3803                 this_item_size = btrfs_item_size(right, item);
3804                 if (this_item_size + sizeof(*item) + push_space > free_space)
3805                         break;
3806
3807                 push_items++;
3808                 push_space += this_item_size + sizeof(*item);
3809         }
3810
3811         if (push_items == 0) {
3812                 ret = 1;
3813                 goto out;
3814         }
3815         WARN_ON(!empty && push_items == btrfs_header_nritems(right));
3816
3817         /* push data from right to left */
3818         copy_extent_buffer(left, right,
3819                            btrfs_item_nr_offset(btrfs_header_nritems(left)),
3820                            btrfs_item_nr_offset(0),
3821                            push_items * sizeof(struct btrfs_item));
3822
3823         push_space = BTRFS_LEAF_DATA_SIZE(fs_info) -
3824                      btrfs_item_offset_nr(right, push_items - 1);
3825
3826         copy_extent_buffer(left, right, BTRFS_LEAF_DATA_OFFSET +
3827                      leaf_data_end(left) - push_space,
3828                      BTRFS_LEAF_DATA_OFFSET +
3829                      btrfs_item_offset_nr(right, push_items - 1),
3830                      push_space);
3831         old_left_nritems = btrfs_header_nritems(left);
3832         BUG_ON(old_left_nritems <= 0);
3833
3834         btrfs_init_map_token(&token, left);
3835         old_left_item_size = btrfs_item_offset_nr(left, old_left_nritems - 1);
3836         for (i = old_left_nritems; i < old_left_nritems + push_items; i++) {
3837                 u32 ioff;
3838
3839                 item = btrfs_item_nr(i);
3840
3841                 ioff = btrfs_token_item_offset(&token, item);
3842                 btrfs_set_token_item_offset(&token, item,
3843                       ioff - (BTRFS_LEAF_DATA_SIZE(fs_info) - old_left_item_size));
3844         }
3845         btrfs_set_header_nritems(left, old_left_nritems + push_items);
3846
3847         /* fixup right node */
3848         if (push_items > right_nritems)
3849                 WARN(1, KERN_CRIT "push items %d nr %u\n", push_items,
3850                        right_nritems);
3851
3852         if (push_items < right_nritems) {
3853                 push_space = btrfs_item_offset_nr(right, push_items - 1) -
3854                                                   leaf_data_end(right);
3855                 memmove_extent_buffer(right, BTRFS_LEAF_DATA_OFFSET +
3856                                       BTRFS_LEAF_DATA_SIZE(fs_info) - push_space,
3857                                       BTRFS_LEAF_DATA_OFFSET +
3858                                       leaf_data_end(right), push_space);
3859
3860                 memmove_extent_buffer(right, btrfs_item_nr_offset(0),
3861                               btrfs_item_nr_offset(push_items),
3862                              (btrfs_header_nritems(right) - push_items) *
3863                              sizeof(struct btrfs_item));
3864         }
3865
3866         btrfs_init_map_token(&token, right);
3867         right_nritems -= push_items;
3868         btrfs_set_header_nritems(right, right_nritems);
3869         push_space = BTRFS_LEAF_DATA_SIZE(fs_info);
3870         for (i = 0; i < right_nritems; i++) {
3871                 item = btrfs_item_nr(i);
3872
3873                 push_space = push_space - btrfs_token_item_size(&token, item);
3874                 btrfs_set_token_item_offset(&token, item, push_space);
3875         }
3876
3877         btrfs_mark_buffer_dirty(left);
3878         if (right_nritems)
3879                 btrfs_mark_buffer_dirty(right);
3880         else
3881                 btrfs_clean_tree_block(right);
3882
3883         btrfs_item_key(right, &disk_key, 0);
3884         fixup_low_keys(path, &disk_key, 1);
3885
3886         /* then fixup the leaf pointer in the path */
3887         if (path->slots[0] < push_items) {
3888                 path->slots[0] += old_left_nritems;
3889                 btrfs_tree_unlock(path->nodes[0]);
3890                 free_extent_buffer(path->nodes[0]);
3891                 path->nodes[0] = left;
3892                 path->slots[1] -= 1;
3893         } else {
3894                 btrfs_tree_unlock(left);
3895                 free_extent_buffer(left);
3896                 path->slots[0] -= push_items;
3897         }
3898         BUG_ON(path->slots[0] < 0);
3899         return ret;
3900 out:
3901         btrfs_tree_unlock(left);
3902         free_extent_buffer(left);
3903         return ret;
3904 }
3905
3906 /*
3907  * push some data in the path leaf to the left, trying to free up at
3908  * least data_size bytes.  returns zero if the push worked, nonzero otherwise
3909  *
3910  * max_slot can put a limit on how far into the leaf we'll push items.  The
3911  * item at 'max_slot' won't be touched.  Use (u32)-1 to make us push all the
3912  * items
3913  */
3914 static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
3915                           *root, struct btrfs_path *path, int min_data_size,
3916                           int data_size, int empty, u32 max_slot)
3917 {
3918         struct extent_buffer *right = path->nodes[0];
3919         struct extent_buffer *left;
3920         int slot;
3921         int free_space;
3922         u32 right_nritems;
3923         int ret = 0;
3924
3925         slot = path->slots[1];
3926         if (slot == 0)
3927                 return 1;
3928         if (!path->nodes[1])
3929                 return 1;
3930
3931         right_nritems = btrfs_header_nritems(right);
3932         if (right_nritems == 0)
3933                 return 1;
3934
3935         btrfs_assert_tree_locked(path->nodes[1]);
3936
3937         left = btrfs_read_node_slot(path->nodes[1], slot - 1);
3938         /*
3939          * slot - 1 is not valid or we fail to read the left node,
3940          * no big deal, just return.
3941          */
3942         if (IS_ERR(left))
3943                 return 1;
3944
3945         btrfs_tree_lock(left);
3946         btrfs_set_lock_blocking_write(left);
3947
3948         free_space = btrfs_leaf_free_space(left);
3949         if (free_space < data_size) {
3950                 ret = 1;
3951                 goto out;
3952         }
3953
3954         /* cow and double check */
3955         ret = btrfs_cow_block(trans, root, left,
3956                               path->nodes[1], slot - 1, &left);
3957         if (ret) {
3958                 /* we hit -ENOSPC, but it isn't fatal here */
3959                 if (ret == -ENOSPC)
3960                         ret = 1;
3961                 goto out;
3962         }
3963
3964         free_space = btrfs_leaf_free_space(left);
3965         if (free_space < data_size) {
3966                 ret = 1;
3967                 goto out;
3968         }
3969
3970         return __push_leaf_left(path, min_data_size,
3971                                empty, left, free_space, right_nritems,
3972                                max_slot);
3973 out:
3974         btrfs_tree_unlock(left);
3975         free_extent_buffer(left);
3976         return ret;
3977 }
3978
3979 /*
3980  * split the path's leaf in two, making sure there is at least data_size
3981  * available for the resulting leaf level of the path.
3982  */
3983 static noinline void copy_for_split(struct btrfs_trans_handle *trans,
3984                                     struct btrfs_path *path,
3985                                     struct extent_buffer *l,
3986                                     struct extent_buffer *right,
3987                                     int slot, int mid, int nritems)
3988 {
3989         struct btrfs_fs_info *fs_info = trans->fs_info;
3990         int data_copy_size;
3991         int rt_data_off;
3992         int i;
3993         struct btrfs_disk_key disk_key;
3994         struct btrfs_map_token token;
3995
3996         nritems = nritems - mid;
3997         btrfs_set_header_nritems(right, nritems);
3998         data_copy_size = btrfs_item_end_nr(l, mid) - leaf_data_end(l);
3999
4000         copy_extent_buffer(right, l, btrfs_item_nr_offset(0),
4001                            btrfs_item_nr_offset(mid),
4002                            nritems * sizeof(struct btrfs_item));
4003
4004         copy_extent_buffer(right, l,
4005                      BTRFS_LEAF_DATA_OFFSET + BTRFS_LEAF_DATA_SIZE(fs_info) -
4006                      data_copy_size, BTRFS_LEAF_DATA_OFFSET +
4007                      leaf_data_end(l), data_copy_size);
4008
4009         rt_data_off = BTRFS_LEAF_DATA_SIZE(fs_info) - btrfs_item_end_nr(l, mid);
4010
4011         btrfs_init_map_token(&token, right);
4012         for (i = 0; i < nritems; i++) {
4013                 struct btrfs_item *item = btrfs_item_nr(i);
4014                 u32 ioff;
4015
4016                 ioff = btrfs_token_item_offset(&token, item);
4017                 btrfs_set_token_item_offset(&token, item, ioff + rt_data_off);
4018         }
4019
4020         btrfs_set_header_nritems(l, mid);
4021         btrfs_item_key(right, &disk_key, 0);
4022         insert_ptr(trans, path, &disk_key, right->start, path->slots[1] + 1, 1);
4023
4024         btrfs_mark_buffer_dirty(right);
4025         btrfs_mark_buffer_dirty(l);
4026         BUG_ON(path->slots[0] != slot);
4027
4028         if (mid <= slot) {
4029                 btrfs_tree_unlock(path->nodes[0]);
4030                 free_extent_buffer(path->nodes[0]);
4031                 path->nodes[0] = right;
4032                 path->slots[0] -= mid;
4033                 path->slots[1] += 1;
4034         } else {
4035                 btrfs_tree_unlock(right);
4036                 free_extent_buffer(right);
4037         }
4038
4039         BUG_ON(path->slots[0] < 0);
4040 }
4041
4042 /*
4043  * double splits happen when we need to insert a big item in the middle
4044  * of a leaf.  A double split can leave us with 3 mostly empty leaves:
4045  * leaf: [ slots 0 - N] [ our target ] [ N + 1 - total in leaf ]
4046  *          A                 B                 C
4047  *
4048  * We avoid this by trying to push the items on either side of our target
4049  * into the adjacent leaves.  If all goes well we can avoid the double split
4050  * completely.
4051  */
4052 static noinline int push_for_double_split(struct btrfs_trans_handle *trans,
4053                                           struct btrfs_root *root,
4054                                           struct btrfs_path *path,
4055                                           int data_size)
4056 {
4057         int ret;
4058         int progress = 0;
4059         int slot;
4060         u32 nritems;
4061         int space_needed = data_size;
4062
4063         slot = path->slots[0];
4064         if (slot < btrfs_header_nritems(path->nodes[0]))
4065                 space_needed -= btrfs_leaf_free_space(path->nodes[0]);
4066
4067         /*
4068          * try to push all the items after our slot into the
4069          * right leaf
4070          */
4071         ret = push_leaf_right(trans, root, path, 1, space_needed, 0, slot);
4072         if (ret < 0)
4073                 return ret;
4074
4075         if (ret == 0)
4076                 progress++;
4077
4078         nritems = btrfs_header_nritems(path->nodes[0]);
4079         /*
4080          * our goal is to get our slot at the start or end of a leaf.  If
4081          * we've done so we're done
4082          */
4083         if (path->slots[0] == 0 || path->slots[0] == nritems)
4084                 return 0;
4085
4086         if (btrfs_leaf_free_space(path->nodes[0]) >= data_size)
4087                 return 0;
4088
4089         /* try to push all the items before our slot into the next leaf */
4090         slot = path->slots[0];
4091         space_needed = data_size;
4092         if (slot > 0)
4093                 space_needed -= btrfs_leaf_free_space(path->nodes[0]);
4094         ret = push_leaf_left(trans, root, path, 1, space_needed, 0, slot);
4095         if (ret < 0)
4096                 return ret;
4097
4098         if (ret == 0)
4099                 progress++;
4100
4101         if (progress)
4102                 return 0;
4103         return 1;
4104 }
4105
4106 /*
4107  * split the path's leaf in two, making sure there is at least data_size
4108  * available for the resulting leaf level of the path.
4109  *
4110  * returns 0 if all went well and < 0 on failure.
4111  */
4112 static noinline int split_leaf(struct btrfs_trans_handle *trans,
4113                                struct btrfs_root *root,
4114                                const struct btrfs_key *ins_key,
4115                                struct btrfs_path *path, int data_size,
4116                                int extend)
4117 {
4118         struct btrfs_disk_key disk_key;
4119         struct extent_buffer *l;
4120         u32 nritems;
4121         int mid;
4122         int slot;
4123         struct extent_buffer *right;
4124         struct btrfs_fs_info *fs_info = root->fs_info;
4125         int ret = 0;
4126         int wret;
4127         int split;
4128         int num_doubles = 0;
4129         int tried_avoid_double = 0;
4130
4131         l = path->nodes[0];
4132         slot = path->slots[0];
4133         if (extend && data_size + btrfs_item_size_nr(l, slot) +
4134             sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(fs_info))
4135                 return -EOVERFLOW;
4136
4137         /* first try to make some room by pushing left and right */
4138         if (data_size && path->nodes[1]) {
4139                 int space_needed = data_size;
4140
4141                 if (slot < btrfs_header_nritems(l))
4142                         space_needed -= btrfs_leaf_free_space(l);
4143
4144                 wret = push_leaf_right(trans, root, path, space_needed,
4145                                        space_needed, 0, 0);
4146                 if (wret < 0)
4147                         return wret;
4148                 if (wret) {
4149                         space_needed = data_size;
4150                         if (slot > 0)
4151                                 space_needed -= btrfs_leaf_free_space(l);
4152                         wret = push_leaf_left(trans, root, path, space_needed,
4153                                               space_needed, 0, (u32)-1);
4154                         if (wret < 0)
4155                                 return wret;
4156                 }
4157                 l = path->nodes[0];
4158
4159                 /* did the pushes work? */
4160                 if (btrfs_leaf_free_space(l) >= data_size)
4161                         return 0;
4162         }
4163
4164         if (!path->nodes[1]) {
4165                 ret = insert_new_root(trans, root, path, 1);
4166                 if (ret)
4167                         return ret;
4168         }
4169 again:
4170         split = 1;
4171         l = path->nodes[0];
4172         slot = path->slots[0];
4173         nritems = btrfs_header_nritems(l);
4174         mid = (nritems + 1) / 2;
4175
4176         if (mid <= slot) {
4177                 if (nritems == 1 ||
4178                     leaf_space_used(l, mid, nritems - mid) + data_size >
4179                         BTRFS_LEAF_DATA_SIZE(fs_info)) {
4180                         if (slot >= nritems) {
4181                                 split = 0;
4182                         } else {
4183                                 mid = slot;
4184                                 if (mid != nritems &&
4185                                     leaf_space_used(l, mid, nritems - mid) +
4186                                     data_size > BTRFS_LEAF_DATA_SIZE(fs_info)) {
4187                                         if (data_size && !tried_avoid_double)
4188                                                 goto push_for_double;
4189                                         split = 2;
4190                                 }
4191                         }
4192                 }
4193         } else {
4194                 if (leaf_space_used(l, 0, mid) + data_size >
4195                         BTRFS_LEAF_DATA_SIZE(fs_info)) {
4196                         if (!extend && data_size && slot == 0) {
4197                                 split = 0;
4198                         } else if ((extend || !data_size) && slot == 0) {
4199                                 mid = 1;
4200                         } else {
4201                                 mid = slot;
4202                                 if (mid != nritems &&
4203                                     leaf_space_used(l, mid, nritems - mid) +
4204                                     data_size > BTRFS_LEAF_DATA_SIZE(fs_info)) {
4205                                         if (data_size && !tried_avoid_double)
4206                                                 goto push_for_double;
4207                                         split = 2;
4208                                 }
4209                         }
4210                 }
4211         }
4212
4213         if (split == 0)
4214                 btrfs_cpu_key_to_disk(&disk_key, ins_key);
4215         else
4216                 btrfs_item_key(l, &disk_key, mid);
4217
4218         right = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, 0,
4219                                              l->start, 0);
4220         if (IS_ERR(right))
4221                 return PTR_ERR(right);
4222
4223         root_add_used(root, fs_info->nodesize);
4224
4225         if (split == 0) {
4226                 if (mid <= slot) {
4227                         btrfs_set_header_nritems(right, 0);
4228                         insert_ptr(trans, path, &disk_key,
4229                                    right->start, path->slots[1] + 1, 1);
4230                         btrfs_tree_unlock(path->nodes[0]);
4231                         free_extent_buffer(path->nodes[0]);
4232                         path->nodes[0] = right;
4233                         path->slots[0] = 0;
4234                         path->slots[1] += 1;
4235                 } else {
4236                         btrfs_set_header_nritems(right, 0);
4237                         insert_ptr(trans, path, &disk_key,
4238                                    right->start, path->slots[1], 1);
4239                         btrfs_tree_unlock(path->nodes[0]);
4240                         free_extent_buffer(path->nodes[0]);
4241                         path->nodes[0] = right;
4242                         path->slots[0] = 0;
4243                         if (path->slots[1] == 0)
4244                                 fixup_low_keys(path, &disk_key, 1);
4245                 }
4246                 /*
4247                  * We create a new leaf 'right' for the required ins_len and
4248                  * we'll do btrfs_mark_buffer_dirty() on this leaf after copying
4249                  * the content of ins_len to 'right'.
4250                  */
4251                 return ret;
4252         }
4253
4254         copy_for_split(trans, path, l, right, slot, mid, nritems);
4255
4256         if (split == 2) {
4257                 BUG_ON(num_doubles != 0);
4258                 num_doubles++;
4259                 goto again;
4260         }
4261
4262         return 0;
4263
4264 push_for_double:
4265         push_for_double_split(trans, root, path, data_size);
4266         tried_avoid_double = 1;
4267         if (btrfs_leaf_free_space(path->nodes[0]) >= data_size)
4268                 return 0;
4269         goto again;
4270 }
4271
4272 static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans,
4273                                          struct btrfs_root *root,
4274                                          struct btrfs_path *path, int ins_len)
4275 {
4276         struct btrfs_key key;
4277         struct extent_buffer *leaf;
4278         struct btrfs_file_extent_item *fi;
4279         u64 extent_len = 0;
4280         u32 item_size;
4281         int ret;
4282
4283         leaf = path->nodes[0];
4284         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
4285
4286         BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY &&
4287                key.type != BTRFS_EXTENT_CSUM_KEY);
4288
4289         if (btrfs_leaf_free_space(leaf) >= ins_len)
4290                 return 0;
4291
4292         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
4293         if (key.type == BTRFS_EXTENT_DATA_KEY) {
4294                 fi = btrfs_item_ptr(leaf, path->slots[0],
4295                                     struct btrfs_file_extent_item);
4296                 extent_len = btrfs_file_extent_num_bytes(leaf, fi);
4297         }
4298         btrfs_release_path(path);
4299
4300         path->keep_locks = 1;
4301         path->search_for_split = 1;
4302         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
4303         path->search_for_split = 0;
4304         if (ret > 0)
4305                 ret = -EAGAIN;
4306         if (ret < 0)
4307                 goto err;
4308
4309         ret = -EAGAIN;
4310         leaf = path->nodes[0];
4311         /* if our item isn't there, return now */
4312         if (item_size != btrfs_item_size_nr(leaf, path->slots[0]))
4313                 goto err;
4314
4315         /* the leaf has  changed, it now has room.  return now */
4316         if (btrfs_leaf_free_space(path->nodes[0]) >= ins_len)
4317                 goto err;
4318
4319         if (key.type == BTRFS_EXTENT_DATA_KEY) {
4320                 fi = btrfs_item_ptr(leaf, path->slots[0],
4321                                     struct btrfs_file_extent_item);
4322                 if (extent_len != btrfs_file_extent_num_bytes(leaf, fi))
4323                         goto err;
4324         }
4325
4326         btrfs_set_path_blocking(path);
4327         ret = split_leaf(trans, root, &key, path, ins_len, 1);
4328         if (ret)
4329                 goto err;
4330
4331         path->keep_locks = 0;
4332         btrfs_unlock_up_safe(path, 1);
4333         return 0;
4334 err:
4335         path->keep_locks = 0;
4336         return ret;
4337 }
4338
4339 static noinline int split_item(struct btrfs_path *path,
4340                                const struct btrfs_key *new_key,
4341                                unsigned long split_offset)
4342 {
4343         struct extent_buffer *leaf;
4344         struct btrfs_item *item;
4345         struct btrfs_item *new_item;
4346         int slot;
4347         char *buf;
4348         u32 nritems;
4349         u32 item_size;
4350         u32 orig_offset;
4351         struct btrfs_disk_key disk_key;
4352
4353         leaf = path->nodes[0];
4354         BUG_ON(btrfs_leaf_free_space(leaf) < sizeof(struct btrfs_item));
4355
4356         btrfs_set_path_blocking(path);
4357
4358         item = btrfs_item_nr(path->slots[0]);
4359         orig_offset = btrfs_item_offset(leaf, item);
4360         item_size = btrfs_item_size(leaf, item);
4361
4362         buf = kmalloc(item_size, GFP_NOFS);
4363         if (!buf)
4364                 return -ENOMEM;
4365
4366         read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf,
4367                             path->slots[0]), item_size);
4368
4369         slot = path->slots[0] + 1;
4370         nritems = btrfs_header_nritems(leaf);
4371         if (slot != nritems) {
4372                 /* shift the items */
4373                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1),
4374                                 btrfs_item_nr_offset(slot),
4375                                 (nritems - slot) * sizeof(struct btrfs_item));
4376         }
4377
4378         btrfs_cpu_key_to_disk(&disk_key, new_key);
4379         btrfs_set_item_key(leaf, &disk_key, slot);
4380
4381         new_item = btrfs_item_nr(slot);
4382
4383         btrfs_set_item_offset(leaf, new_item, orig_offset);
4384         btrfs_set_item_size(leaf, new_item, item_size - split_offset);
4385
4386         btrfs_set_item_offset(leaf, item,
4387                               orig_offset + item_size - split_offset);
4388         btrfs_set_item_size(leaf, item, split_offset);
4389
4390         btrfs_set_header_nritems(leaf, nritems + 1);
4391
4392         /* write the data for the start of the original item */
4393         write_extent_buffer(leaf, buf,
4394                             btrfs_item_ptr_offset(leaf, path->slots[0]),
4395                             split_offset);
4396
4397         /* write the data for the new item */
4398         write_extent_buffer(leaf, buf + split_offset,
4399                             btrfs_item_ptr_offset(leaf, slot),
4400                             item_size - split_offset);
4401         btrfs_mark_buffer_dirty(leaf);
4402
4403         BUG_ON(btrfs_leaf_free_space(leaf) < 0);
4404         kfree(buf);
4405         return 0;
4406 }
4407
4408 /*
4409  * This function splits a single item into two items,
4410  * giving 'new_key' to the new item and splitting the
4411  * old one at split_offset (from the start of the item).
4412  *
4413  * The path may be released by this operation.  After
4414  * the split, the path is pointing to the old item.  The
4415  * new item is going to be in the same node as the old one.
4416  *
4417  * Note, the item being split must be smaller enough to live alone on
4418  * a tree block with room for one extra struct btrfs_item
4419  *
4420  * This allows us to split the item in place, keeping a lock on the
4421  * leaf the entire time.
4422  */
4423 int btrfs_split_item(struct btrfs_trans_handle *trans,
4424                      struct btrfs_root *root,
4425                      struct btrfs_path *path,
4426                      const struct btrfs_key *new_key,
4427                      unsigned long split_offset)
4428 {
4429         int ret;
4430         ret = setup_leaf_for_split(trans, root, path,
4431                                    sizeof(struct btrfs_item));
4432         if (ret)
4433                 return ret;
4434
4435         ret = split_item(path, new_key, split_offset);
4436         return ret;
4437 }
4438
4439 /*
4440  * This function duplicate a item, giving 'new_key' to the new item.
4441  * It guarantees both items live in the same tree leaf and the new item
4442  * is contiguous with the original item.
4443  *
4444  * This allows us to split file extent in place, keeping a lock on the
4445  * leaf the entire time.
4446  */
4447 int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
4448                          struct btrfs_root *root,
4449                          struct btrfs_path *path,
4450                          const struct btrfs_key *new_key)
4451 {
4452         struct extent_buffer *leaf;
4453         int ret;
4454         u32 item_size;
4455
4456         leaf = path->nodes[0];
4457         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
4458         ret = setup_leaf_for_split(trans, root, path,
4459                                    item_size + sizeof(struct btrfs_item));
4460         if (ret)
4461                 return ret;
4462
4463         path->slots[0]++;
4464         setup_items_for_insert(root, path, new_key, &item_size,
4465                                item_size, item_size +
4466                                sizeof(struct btrfs_item), 1);
4467         leaf = path->nodes[0];
4468         memcpy_extent_buffer(leaf,
4469                              btrfs_item_ptr_offset(leaf, path->slots[0]),
4470                              btrfs_item_ptr_offset(leaf, path->slots[0] - 1),
4471                              item_size);
4472         return 0;
4473 }
4474
4475 /*
4476  * make the item pointed to by the path smaller.  new_size indicates
4477  * how small to make it, and from_end tells us if we just chop bytes
4478  * off the end of the item or if we shift the item to chop bytes off
4479  * the front.
4480  */
4481 void btrfs_truncate_item(struct btrfs_path *path, u32 new_size, int from_end)
4482 {
4483         int slot;
4484         struct extent_buffer *leaf;
4485         struct btrfs_item *item;
4486         u32 nritems;
4487         unsigned int data_end;
4488         unsigned int old_data_start;
4489         unsigned int old_size;
4490         unsigned int size_diff;
4491         int i;
4492         struct btrfs_map_token token;
4493
4494         leaf = path->nodes[0];
4495         slot = path->slots[0];
4496
4497         old_size = btrfs_item_size_nr(leaf, slot);
4498         if (old_size == new_size)
4499                 return;
4500
4501         nritems = btrfs_header_nritems(leaf);
4502         data_end = leaf_data_end(leaf);
4503
4504         old_data_start = btrfs_item_offset_nr(leaf, slot);
4505
4506         size_diff = old_size - new_size;
4507
4508         BUG_ON(slot < 0);
4509         BUG_ON(slot >= nritems);
4510
4511         /*
4512          * item0..itemN ... dataN.offset..dataN.size .. data0.size
4513          */
4514         /* first correct the data pointers */
4515         btrfs_init_map_token(&token, leaf);
4516         for (i = slot; i < nritems; i++) {
4517                 u32 ioff;
4518                 item = btrfs_item_nr(i);
4519
4520                 ioff = btrfs_token_item_offset(&token, item);
4521                 btrfs_set_token_item_offset(&token, item, ioff + size_diff);
4522         }
4523
4524         /* shift the data */
4525         if (from_end) {
4526                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4527                               data_end + size_diff, BTRFS_LEAF_DATA_OFFSET +
4528                               data_end, old_data_start + new_size - data_end);
4529         } else {
4530                 struct btrfs_disk_key disk_key;
4531                 u64 offset;
4532
4533                 btrfs_item_key(leaf, &disk_key, slot);
4534
4535                 if (btrfs_disk_key_type(&disk_key) == BTRFS_EXTENT_DATA_KEY) {
4536                         unsigned long ptr;
4537                         struct btrfs_file_extent_item *fi;
4538
4539                         fi = btrfs_item_ptr(leaf, slot,
4540                                             struct btrfs_file_extent_item);
4541                         fi = (struct btrfs_file_extent_item *)(
4542                              (unsigned long)fi - size_diff);
4543
4544                         if (btrfs_file_extent_type(leaf, fi) ==
4545                             BTRFS_FILE_EXTENT_INLINE) {
4546                                 ptr = btrfs_item_ptr_offset(leaf, slot);
4547                                 memmove_extent_buffer(leaf, ptr,
4548                                       (unsigned long)fi,
4549                                       BTRFS_FILE_EXTENT_INLINE_DATA_START);
4550                         }
4551                 }
4552
4553                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4554                               data_end + size_diff, BTRFS_LEAF_DATA_OFFSET +
4555                               data_end, old_data_start - data_end);
4556
4557                 offset = btrfs_disk_key_offset(&disk_key);
4558                 btrfs_set_disk_key_offset(&disk_key, offset + size_diff);
4559                 btrfs_set_item_key(leaf, &disk_key, slot);
4560                 if (slot == 0)
4561                         fixup_low_keys(path, &disk_key, 1);
4562         }
4563
4564         item = btrfs_item_nr(slot);
4565         btrfs_set_item_size(leaf, item, new_size);
4566         btrfs_mark_buffer_dirty(leaf);
4567
4568         if (btrfs_leaf_free_space(leaf) < 0) {
4569                 btrfs_print_leaf(leaf);
4570                 BUG();
4571         }
4572 }
4573
4574 /*
4575  * make the item pointed to by the path bigger, data_size is the added size.
4576  */
4577 void btrfs_extend_item(struct btrfs_path *path, u32 data_size)
4578 {
4579         int slot;
4580         struct extent_buffer *leaf;
4581         struct btrfs_item *item;
4582         u32 nritems;
4583         unsigned int data_end;
4584         unsigned int old_data;
4585         unsigned int old_size;
4586         int i;
4587         struct btrfs_map_token token;
4588
4589         leaf = path->nodes[0];
4590
4591         nritems = btrfs_header_nritems(leaf);
4592         data_end = leaf_data_end(leaf);
4593
4594         if (btrfs_leaf_free_space(leaf) < data_size) {
4595                 btrfs_print_leaf(leaf);
4596                 BUG();
4597         }
4598         slot = path->slots[0];
4599         old_data = btrfs_item_end_nr(leaf, slot);
4600
4601         BUG_ON(slot < 0);
4602         if (slot >= nritems) {
4603                 btrfs_print_leaf(leaf);
4604                 btrfs_crit(leaf->fs_info, "slot %d too large, nritems %d",
4605                            slot, nritems);
4606                 BUG();
4607         }
4608
4609         /*
4610          * item0..itemN ... dataN.offset..dataN.size .. data0.size
4611          */
4612         /* first correct the data pointers */
4613         btrfs_init_map_token(&token, leaf);
4614         for (i = slot; i < nritems; i++) {
4615                 u32 ioff;
4616                 item = btrfs_item_nr(i);
4617
4618                 ioff = btrfs_token_item_offset(&token, item);
4619                 btrfs_set_token_item_offset(&token, item, ioff - data_size);
4620         }
4621
4622         /* shift the data */
4623         memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4624                       data_end - data_size, BTRFS_LEAF_DATA_OFFSET +
4625                       data_end, old_data - data_end);
4626
4627         data_end = old_data;
4628         old_size = btrfs_item_size_nr(leaf, slot);
4629         item = btrfs_item_nr(slot);
4630         btrfs_set_item_size(leaf, item, old_size + data_size);
4631         btrfs_mark_buffer_dirty(leaf);
4632
4633         if (btrfs_leaf_free_space(leaf) < 0) {
4634                 btrfs_print_leaf(leaf);
4635                 BUG();
4636         }
4637 }
4638
4639 /*
4640  * this is a helper for btrfs_insert_empty_items, the main goal here is
4641  * to save stack depth by doing the bulk of the work in a function
4642  * that doesn't call btrfs_search_slot
4643  */
4644 void setup_items_for_insert(struct btrfs_root *root, struct btrfs_path *path,
4645                             const struct btrfs_key *cpu_key, u32 *data_size,
4646                             u32 total_data, u32 total_size, int nr)
4647 {
4648         struct btrfs_fs_info *fs_info = root->fs_info;
4649         struct btrfs_item *item;
4650         int i;
4651         u32 nritems;
4652         unsigned int data_end;
4653         struct btrfs_disk_key disk_key;
4654         struct extent_buffer *leaf;
4655         int slot;
4656         struct btrfs_map_token token;
4657
4658         if (path->slots[0] == 0) {
4659                 btrfs_cpu_key_to_disk(&disk_key, cpu_key);
4660                 fixup_low_keys(path, &disk_key, 1);
4661         }
4662         btrfs_unlock_up_safe(path, 1);
4663
4664         leaf = path->nodes[0];
4665         slot = path->slots[0];
4666
4667         nritems = btrfs_header_nritems(leaf);
4668         data_end = leaf_data_end(leaf);
4669
4670         if (btrfs_leaf_free_space(leaf) < total_size) {
4671                 btrfs_print_leaf(leaf);
4672                 btrfs_crit(fs_info, "not enough freespace need %u have %d",
4673                            total_size, btrfs_leaf_free_space(leaf));
4674                 BUG();
4675         }
4676
4677         btrfs_init_map_token(&token, leaf);
4678         if (slot != nritems) {
4679                 unsigned int old_data = btrfs_item_end_nr(leaf, slot);
4680
4681                 if (old_data < data_end) {
4682                         btrfs_print_leaf(leaf);
4683                         btrfs_crit(fs_info, "slot %d old_data %d data_end %d",
4684                                    slot, old_data, data_end);
4685                         BUG();
4686                 }
4687                 /*
4688                  * item0..itemN ... dataN.offset..dataN.size .. data0.size
4689                  */
4690                 /* first correct the data pointers */
4691                 for (i = slot; i < nritems; i++) {
4692                         u32 ioff;
4693
4694                         item = btrfs_item_nr(i);
4695                         ioff = btrfs_token_item_offset(&token, item);
4696                         btrfs_set_token_item_offset(&token, item,
4697                                                     ioff - total_data);
4698                 }
4699                 /* shift the items */
4700                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
4701                               btrfs_item_nr_offset(slot),
4702                               (nritems - slot) * sizeof(struct btrfs_item));
4703
4704                 /* shift the data */
4705                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4706                               data_end - total_data, BTRFS_LEAF_DATA_OFFSET +
4707                               data_end, old_data - data_end);
4708                 data_end = old_data;
4709         }
4710
4711         /* setup the item for the new data */
4712         for (i = 0; i < nr; i++) {
4713                 btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
4714                 btrfs_set_item_key(leaf, &disk_key, slot + i);
4715                 item = btrfs_item_nr(slot + i);
4716                 btrfs_set_token_item_offset(&token, item, data_end - data_size[i]);
4717                 data_end -= data_size[i];
4718                 btrfs_set_token_item_size(&token, item, data_size[i]);
4719         }
4720
4721         btrfs_set_header_nritems(leaf, nritems + nr);
4722         btrfs_mark_buffer_dirty(leaf);
4723
4724         if (btrfs_leaf_free_space(leaf) < 0) {
4725                 btrfs_print_leaf(leaf);
4726                 BUG();
4727         }
4728 }
4729
4730 /*
4731  * Given a key and some data, insert items into the tree.
4732  * This does all the path init required, making room in the tree if needed.
4733  */
4734 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
4735                             struct btrfs_root *root,
4736                             struct btrfs_path *path,
4737                             const struct btrfs_key *cpu_key, u32 *data_size,
4738                             int nr)
4739 {
4740         int ret = 0;
4741         int slot;
4742         int i;
4743         u32 total_size = 0;
4744         u32 total_data = 0;
4745
4746         for (i = 0; i < nr; i++)
4747                 total_data += data_size[i];
4748
4749         total_size = total_data + (nr * sizeof(struct btrfs_item));
4750         ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1);
4751         if (ret == 0)
4752                 return -EEXIST;
4753         if (ret < 0)
4754                 return ret;
4755
4756         slot = path->slots[0];
4757         BUG_ON(slot < 0);
4758
4759         setup_items_for_insert(root, path, cpu_key, data_size,
4760                                total_data, total_size, nr);
4761         return 0;
4762 }
4763
4764 /*
4765  * Given a key and some data, insert an item into the tree.
4766  * This does all the path init required, making room in the tree if needed.
4767  */
4768 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4769                       const struct btrfs_key *cpu_key, void *data,
4770                       u32 data_size)
4771 {
4772         int ret = 0;
4773         struct btrfs_path *path;
4774         struct extent_buffer *leaf;
4775         unsigned long ptr;
4776
4777         path = btrfs_alloc_path();
4778         if (!path)
4779                 return -ENOMEM;
4780         ret = btrfs_insert_empty_item(trans, root, path, cpu_key, data_size);
4781         if (!ret) {
4782                 leaf = path->nodes[0];
4783                 ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
4784                 write_extent_buffer(leaf, data, ptr, data_size);
4785                 btrfs_mark_buffer_dirty(leaf);
4786         }
4787         btrfs_free_path(path);
4788         return ret;
4789 }
4790
4791 /*
4792  * delete the pointer from a given node.
4793  *
4794  * the tree should have been previously balanced so the deletion does not
4795  * empty a node.
4796  */
4797 static void del_ptr(struct btrfs_root *root, struct btrfs_path *path,
4798                     int level, int slot)
4799 {
4800         struct extent_buffer *parent = path->nodes[level];
4801         u32 nritems;
4802         int ret;
4803
4804         nritems = btrfs_header_nritems(parent);
4805         if (slot != nritems - 1) {
4806                 if (level) {
4807                         ret = tree_mod_log_insert_move(parent, slot, slot + 1,
4808                                         nritems - slot - 1);
4809                         BUG_ON(ret < 0);
4810                 }
4811                 memmove_extent_buffer(parent,
4812                               btrfs_node_key_ptr_offset(slot),
4813                               btrfs_node_key_ptr_offset(slot + 1),
4814                               sizeof(struct btrfs_key_ptr) *
4815                               (nritems - slot - 1));
4816         } else if (level) {
4817                 ret = tree_mod_log_insert_key(parent, slot, MOD_LOG_KEY_REMOVE,
4818                                 GFP_NOFS);
4819                 BUG_ON(ret < 0);
4820         }
4821
4822         nritems--;
4823         btrfs_set_header_nritems(parent, nritems);
4824         if (nritems == 0 && parent == root->node) {
4825                 BUG_ON(btrfs_header_level(root->node) != 1);
4826                 /* just turn the root into a leaf and break */
4827                 btrfs_set_header_level(root->node, 0);
4828         } else if (slot == 0) {
4829                 struct btrfs_disk_key disk_key;
4830
4831                 btrfs_node_key(parent, &disk_key, 0);
4832                 fixup_low_keys(path, &disk_key, level + 1);
4833         }
4834         btrfs_mark_buffer_dirty(parent);
4835 }
4836
4837 /*
4838  * a helper function to delete the leaf pointed to by path->slots[1] and
4839  * path->nodes[1].
4840  *
4841  * This deletes the pointer in path->nodes[1] and frees the leaf
4842  * block extent.  zero is returned if it all worked out, < 0 otherwise.
4843  *
4844  * The path must have already been setup for deleting the leaf, including
4845  * all the proper balancing.  path->nodes[1] must be locked.
4846  */
4847 static noinline void btrfs_del_leaf(struct btrfs_trans_handle *trans,
4848                                     struct btrfs_root *root,
4849                                     struct btrfs_path *path,
4850                                     struct extent_buffer *leaf)
4851 {
4852         WARN_ON(btrfs_header_generation(leaf) != trans->transid);
4853         del_ptr(root, path, 1, path->slots[1]);
4854
4855         /*
4856          * btrfs_free_extent is expensive, we want to make sure we
4857          * aren't holding any locks when we call it
4858          */
4859         btrfs_unlock_up_safe(path, 0);
4860
4861         root_sub_used(root, leaf->len);
4862
4863         atomic_inc(&leaf->refs);
4864         btrfs_free_tree_block(trans, root, leaf, 0, 1);
4865         free_extent_buffer_stale(leaf);
4866 }
4867 /*
4868  * delete the item at the leaf level in path.  If that empties
4869  * the leaf, remove it from the tree
4870  */
4871 int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4872                     struct btrfs_path *path, int slot, int nr)
4873 {
4874         struct btrfs_fs_info *fs_info = root->fs_info;
4875         struct extent_buffer *leaf;
4876         struct btrfs_item *item;
4877         u32 last_off;
4878         u32 dsize = 0;
4879         int ret = 0;
4880         int wret;
4881         int i;
4882         u32 nritems;
4883
4884         leaf = path->nodes[0];
4885         last_off = btrfs_item_offset_nr(leaf, slot + nr - 1);
4886
4887         for (i = 0; i < nr; i++)
4888                 dsize += btrfs_item_size_nr(leaf, slot + i);
4889
4890         nritems = btrfs_header_nritems(leaf);
4891
4892         if (slot + nr != nritems) {
4893                 int data_end = leaf_data_end(leaf);
4894                 struct btrfs_map_token token;
4895
4896                 memmove_extent_buffer(leaf, BTRFS_LEAF_DATA_OFFSET +
4897                               data_end + dsize,
4898                               BTRFS_LEAF_DATA_OFFSET + data_end,
4899                               last_off - data_end);
4900
4901                 btrfs_init_map_token(&token, leaf);
4902                 for (i = slot + nr; i < nritems; i++) {
4903                         u32 ioff;
4904
4905                         item = btrfs_item_nr(i);
4906                         ioff = btrfs_token_item_offset(&token, item);
4907                         btrfs_set_token_item_offset(&token, item, ioff + dsize);
4908                 }
4909
4910                 memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot),
4911                               btrfs_item_nr_offset(slot + nr),
4912                               sizeof(struct btrfs_item) *
4913                               (nritems - slot - nr));
4914         }
4915         btrfs_set_header_nritems(leaf, nritems - nr);
4916         nritems -= nr;
4917
4918         /* delete the leaf if we've emptied it */
4919         if (nritems == 0) {
4920                 if (leaf == root->node) {
4921                         btrfs_set_header_level(leaf, 0);
4922                 } else {
4923                         btrfs_set_path_blocking(path);
4924                         btrfs_clean_tree_block(leaf);
4925                         btrfs_del_leaf(trans, root, path, leaf);
4926                 }
4927         } else {
4928                 int used = leaf_space_used(leaf, 0, nritems);
4929                 if (slot == 0) {
4930                         struct btrfs_disk_key disk_key;
4931
4932                         btrfs_item_key(leaf, &disk_key, 0);
4933                         fixup_low_keys(path, &disk_key, 1);
4934                 }
4935
4936                 /* delete the leaf if it is mostly empty */
4937                 if (used < BTRFS_LEAF_DATA_SIZE(fs_info) / 3) {
4938                         /* push_leaf_left fixes the path.
4939                          * make sure the path still points to our leaf
4940                          * for possible call to del_ptr below
4941                          */
4942                         slot = path->slots[1];
4943                         atomic_inc(&leaf->refs);
4944
4945                         btrfs_set_path_blocking(path);
4946                         wret = push_leaf_left(trans, root, path, 1, 1,
4947                                               1, (u32)-1);
4948                         if (wret < 0 && wret != -ENOSPC)
4949                                 ret = wret;
4950
4951                         if (path->nodes[0] == leaf &&
4952                             btrfs_header_nritems(leaf)) {
4953                                 wret = push_leaf_right(trans, root, path, 1,
4954                                                        1, 1, 0);
4955                                 if (wret < 0 && wret != -ENOSPC)
4956                                         ret = wret;
4957                         }
4958
4959                         if (btrfs_header_nritems(leaf) == 0) {
4960                                 path->slots[1] = slot;
4961                                 btrfs_del_leaf(trans, root, path, leaf);
4962                                 free_extent_buffer(leaf);
4963                                 ret = 0;
4964                         } else {
4965                                 /* if we're still in the path, make sure
4966                                  * we're dirty.  Otherwise, one of the
4967                                  * push_leaf functions must have already
4968                                  * dirtied this buffer
4969                                  */
4970                                 if (path->nodes[0] == leaf)
4971                                         btrfs_mark_buffer_dirty(leaf);
4972                                 free_extent_buffer(leaf);
4973                         }
4974                 } else {
4975                         btrfs_mark_buffer_dirty(leaf);
4976                 }
4977         }
4978         return ret;
4979 }
4980
4981 /*
4982  * search the tree again to find a leaf with lesser keys
4983  * returns 0 if it found something or 1 if there are no lesser leaves.
4984  * returns < 0 on io errors.
4985  *
4986  * This may release the path, and so you may lose any locks held at the
4987  * time you call it.
4988  */
4989 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path)
4990 {
4991         struct btrfs_key key;
4992         struct btrfs_disk_key found_key;
4993         int ret;
4994
4995         btrfs_item_key_to_cpu(path->nodes[0], &key, 0);
4996
4997         if (key.offset > 0) {
4998                 key.offset--;
4999         } else if (key.type > 0) {
5000                 key.type--;
5001                 key.offset = (u64)-1;
5002         } else if (key.objectid > 0) {
5003                 key.objectid--;
5004                 key.type = (u8)-1;
5005                 key.offset = (u64)-1;
5006         } else {
5007                 return 1;
5008         }
5009
5010         btrfs_release_path(path);
5011         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5012         if (ret < 0)
5013                 return ret;
5014         btrfs_item_key(path->nodes[0], &found_key, 0);
5015         ret = comp_keys(&found_key, &key);
5016         /*
5017          * We might have had an item with the previous key in the tree right
5018          * before we released our path. And after we released our path, that
5019          * item might have been pushed to the first slot (0) of the leaf we
5020          * were holding due to a tree balance. Alternatively, an item with the
5021          * previous key can exist as the only element of a leaf (big fat item).
5022          * Therefore account for these 2 cases, so that our callers (like
5023          * btrfs_previous_item) don't miss an existing item with a key matching
5024          * the previous key we computed above.
5025          */
5026         if (ret <= 0)
5027                 return 0;
5028         return 1;
5029 }
5030
5031 /*
5032  * A helper function to walk down the tree starting at min_key, and looking
5033  * for nodes or leaves that are have a minimum transaction id.
5034  * This is used by the btree defrag code, and tree logging
5035  *
5036  * This does not cow, but it does stuff the starting key it finds back
5037  * into min_key, so you can call btrfs_search_slot with cow=1 on the
5038  * key and get a writable path.
5039  *
5040  * This honors path->lowest_level to prevent descent past a given level
5041  * of the tree.
5042  *
5043  * min_trans indicates the oldest transaction that you are interested
5044  * in walking through.  Any nodes or leaves older than min_trans are
5045  * skipped over (without reading them).
5046  *
5047  * returns zero if something useful was found, < 0 on error and 1 if there
5048  * was nothing in the tree that matched the search criteria.
5049  */
5050 int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key,
5051                          struct btrfs_path *path,
5052                          u64 min_trans)
5053 {
5054         struct extent_buffer *cur;
5055         struct btrfs_key found_key;
5056         int slot;
5057         int sret;
5058         u32 nritems;
5059         int level;
5060         int ret = 1;
5061         int keep_locks = path->keep_locks;
5062
5063         path->keep_locks = 1;
5064 again:
5065         cur = btrfs_read_lock_root_node(root);
5066         level = btrfs_header_level(cur);
5067         WARN_ON(path->nodes[level]);
5068         path->nodes[level] = cur;
5069         path->locks[level] = BTRFS_READ_LOCK;
5070
5071         if (btrfs_header_generation(cur) < min_trans) {
5072                 ret = 1;
5073                 goto out;
5074         }
5075         while (1) {
5076                 nritems = btrfs_header_nritems(cur);
5077                 level = btrfs_header_level(cur);
5078                 sret = btrfs_bin_search(cur, min_key, &slot);
5079                 if (sret < 0) {
5080                         ret = sret;
5081                         goto out;
5082                 }
5083
5084                 /* at the lowest level, we're done, setup the path and exit */
5085                 if (level == path->lowest_level) {
5086                         if (slot >= nritems)
5087                                 goto find_next_key;
5088                         ret = 0;
5089                         path->slots[level] = slot;
5090                         btrfs_item_key_to_cpu(cur, &found_key, slot);
5091                         goto out;
5092                 }
5093                 if (sret && slot > 0)
5094                         slot--;
5095                 /*
5096                  * check this node pointer against the min_trans parameters.
5097                  * If it is too old, old, skip to the next one.
5098                  */
5099                 while (slot < nritems) {
5100                         u64 gen;
5101
5102                         gen = btrfs_node_ptr_generation(cur, slot);
5103                         if (gen < min_trans) {
5104                                 slot++;
5105                                 continue;
5106                         }
5107                         break;
5108                 }
5109 find_next_key:
5110                 /*
5111                  * we didn't find a candidate key in this node, walk forward
5112                  * and find another one
5113                  */
5114                 if (slot >= nritems) {
5115                         path->slots[level] = slot;
5116                         btrfs_set_path_blocking(path);
5117                         sret = btrfs_find_next_key(root, path, min_key, level,
5118                                                   min_trans);
5119                         if (sret == 0) {
5120                                 btrfs_release_path(path);
5121                                 goto again;
5122                         } else {
5123                                 goto out;
5124                         }
5125                 }
5126                 /* save our key for returning back */
5127                 btrfs_node_key_to_cpu(cur, &found_key, slot);
5128                 path->slots[level] = slot;
5129                 if (level == path->lowest_level) {
5130                         ret = 0;
5131                         goto out;
5132                 }
5133                 btrfs_set_path_blocking(path);
5134                 cur = btrfs_read_node_slot(cur, slot);
5135                 if (IS_ERR(cur)) {
5136                         ret = PTR_ERR(cur);
5137                         goto out;
5138                 }
5139
5140                 btrfs_tree_read_lock(cur);
5141
5142                 path->locks[level - 1] = BTRFS_READ_LOCK;
5143                 path->nodes[level - 1] = cur;
5144                 unlock_up(path, level, 1, 0, NULL);
5145         }
5146 out:
5147         path->keep_locks = keep_locks;
5148         if (ret == 0) {
5149                 btrfs_unlock_up_safe(path, path->lowest_level + 1);
5150                 btrfs_set_path_blocking(path);
5151                 memcpy(min_key, &found_key, sizeof(found_key));
5152         }
5153         return ret;
5154 }
5155
5156 /*
5157  * this is similar to btrfs_next_leaf, but does not try to preserve
5158  * and fixup the path.  It looks for and returns the next key in the
5159  * tree based on the current path and the min_trans parameters.
5160  *
5161  * 0 is returned if another key is found, < 0 if there are any errors
5162  * and 1 is returned if there are no higher keys in the tree
5163  *
5164  * path->keep_locks should be set to 1 on the search made before
5165  * calling this function.
5166  */
5167 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
5168                         struct btrfs_key *key, int level, u64 min_trans)
5169 {
5170         int slot;
5171         struct extent_buffer *c;
5172
5173         WARN_ON(!path->keep_locks && !path->skip_locking);
5174         while (level < BTRFS_MAX_LEVEL) {
5175                 if (!path->nodes[level])
5176                         return 1;
5177
5178                 slot = path->slots[level] + 1;
5179                 c = path->nodes[level];
5180 next:
5181                 if (slot >= btrfs_header_nritems(c)) {
5182                         int ret;
5183                         int orig_lowest;
5184                         struct btrfs_key cur_key;
5185                         if (level + 1 >= BTRFS_MAX_LEVEL ||
5186                             !path->nodes[level + 1])
5187                                 return 1;
5188
5189                         if (path->locks[level + 1] || path->skip_locking) {
5190                                 level++;
5191                                 continue;
5192                         }
5193
5194                         slot = btrfs_header_nritems(c) - 1;
5195                         if (level == 0)
5196                                 btrfs_item_key_to_cpu(c, &cur_key, slot);
5197                         else
5198                                 btrfs_node_key_to_cpu(c, &cur_key, slot);
5199
5200                         orig_lowest = path->lowest_level;
5201                         btrfs_release_path(path);
5202                         path->lowest_level = level;
5203                         ret = btrfs_search_slot(NULL, root, &cur_key, path,
5204                                                 0, 0);
5205                         path->lowest_level = orig_lowest;
5206                         if (ret < 0)
5207                                 return ret;
5208
5209                         c = path->nodes[level];
5210                         slot = path->slots[level];
5211                         if (ret == 0)
5212                                 slot++;
5213                         goto next;
5214                 }
5215
5216                 if (level == 0)
5217                         btrfs_item_key_to_cpu(c, key, slot);
5218                 else {
5219                         u64 gen = btrfs_node_ptr_generation(c, slot);
5220
5221                         if (gen < min_trans) {
5222                                 slot++;
5223                                 goto next;
5224                         }
5225                         btrfs_node_key_to_cpu(c, key, slot);
5226                 }
5227                 return 0;
5228         }
5229         return 1;
5230 }
5231
5232 /*
5233  * search the tree again to find a leaf with greater keys
5234  * returns 0 if it found something or 1 if there are no greater leaves.
5235  * returns < 0 on io errors.
5236  */
5237 int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
5238 {
5239         return btrfs_next_old_leaf(root, path, 0);
5240 }
5241
5242 int btrfs_next_old_leaf(struct btrfs_root *root, struct btrfs_path *path,
5243                         u64 time_seq)
5244 {
5245         int slot;
5246         int level;
5247         struct extent_buffer *c;
5248         struct extent_buffer *next;
5249         struct btrfs_key key;
5250         u32 nritems;
5251         int ret;
5252         int old_spinning = path->leave_spinning;
5253         int next_rw_lock = 0;
5254
5255         nritems = btrfs_header_nritems(path->nodes[0]);
5256         if (nritems == 0)
5257                 return 1;
5258
5259         btrfs_item_key_to_cpu(path->nodes[0], &key, nritems - 1);
5260 again:
5261         level = 1;
5262         next = NULL;
5263         next_rw_lock = 0;
5264         btrfs_release_path(path);
5265
5266         path->keep_locks = 1;
5267         path->leave_spinning = 1;
5268
5269         if (time_seq)
5270                 ret = btrfs_search_old_slot(root, &key, path, time_seq);
5271         else
5272                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5273         path->keep_locks = 0;
5274
5275         if (ret < 0)
5276                 return ret;
5277
5278         nritems = btrfs_header_nritems(path->nodes[0]);
5279         /*
5280          * by releasing the path above we dropped all our locks.  A balance
5281          * could have added more items next to the key that used to be
5282          * at the very end of the block.  So, check again here and
5283          * advance the path if there are now more items available.
5284          */
5285         if (nritems > 0 && path->slots[0] < nritems - 1) {
5286                 if (ret == 0)
5287                         path->slots[0]++;
5288                 ret = 0;
5289                 goto done;
5290         }
5291         /*
5292          * So the above check misses one case:
5293          * - after releasing the path above, someone has removed the item that
5294          *   used to be at the very end of the block, and balance between leafs
5295          *   gets another one with bigger key.offset to replace it.
5296          *
5297          * This one should be returned as well, or we can get leaf corruption
5298          * later(esp. in __btrfs_drop_extents()).
5299          *
5300          * And a bit more explanation about this check,
5301          * with ret > 0, the key isn't found, the path points to the slot
5302          * where it should be inserted, so the path->slots[0] item must be the
5303          * bigger one.
5304          */
5305         if (nritems > 0 && ret > 0 && path->slots[0] == nritems - 1) {
5306                 ret = 0;
5307                 goto done;
5308         }
5309
5310         while (level < BTRFS_MAX_LEVEL) {
5311                 if (!path->nodes[level]) {
5312                         ret = 1;
5313                         goto done;
5314                 }
5315
5316                 slot = path->slots[level] + 1;
5317                 c = path->nodes[level];
5318                 if (slot >= btrfs_header_nritems(c)) {
5319                         level++;
5320                         if (level == BTRFS_MAX_LEVEL) {
5321                                 ret = 1;
5322                                 goto done;
5323                         }
5324                         continue;
5325                 }
5326
5327                 if (next) {
5328                         btrfs_tree_unlock_rw(next, next_rw_lock);
5329                         free_extent_buffer(next);
5330                 }
5331
5332                 next = c;
5333                 next_rw_lock = path->locks[level];
5334                 ret = read_block_for_search(root, path, &next, level,
5335                                             slot, &key);
5336                 if (ret == -EAGAIN)
5337                         goto again;
5338
5339                 if (ret < 0) {
5340                         btrfs_release_path(path);
5341                         goto done;
5342                 }
5343
5344                 if (!path->skip_locking) {
5345                         ret = btrfs_try_tree_read_lock(next);
5346                         if (!ret && time_seq) {
5347                                 /*
5348                                  * If we don't get the lock, we may be racing
5349                                  * with push_leaf_left, holding that lock while
5350                                  * itself waiting for the leaf we've currently
5351                                  * locked. To solve this situation, we give up
5352                                  * on our lock and cycle.
5353                                  */
5354                                 free_extent_buffer(next);
5355                                 btrfs_release_path(path);
5356                                 cond_resched();
5357                                 goto again;
5358                         }
5359                         if (!ret) {
5360                                 btrfs_set_path_blocking(path);
5361                                 btrfs_tree_read_lock(next);
5362                         }
5363                         next_rw_lock = BTRFS_READ_LOCK;
5364                 }
5365                 break;
5366         }
5367         path->slots[level] = slot;
5368         while (1) {
5369                 level--;
5370                 c = path->nodes[level];
5371                 if (path->locks[level])
5372                         btrfs_tree_unlock_rw(c, path->locks[level]);
5373
5374                 free_extent_buffer(c);
5375                 path->nodes[level] = next;
5376                 path->slots[level] = 0;
5377                 if (!path->skip_locking)
5378                         path->locks[level] = next_rw_lock;
5379                 if (!level)
5380                         break;
5381
5382                 ret = read_block_for_search(root, path, &next, level,
5383                                             0, &key);
5384                 if (ret == -EAGAIN)
5385                         goto again;
5386
5387                 if (ret < 0) {
5388                         btrfs_release_path(path);
5389                         goto done;
5390                 }
5391
5392                 if (!path->skip_locking) {
5393                         ret = btrfs_try_tree_read_lock(next);
5394                         if (!ret) {
5395                                 btrfs_set_path_blocking(path);
5396                                 btrfs_tree_read_lock(next);
5397                         }
5398                         next_rw_lock = BTRFS_READ_LOCK;
5399                 }
5400         }
5401         ret = 0;
5402 done:
5403         unlock_up(path, 0, 1, 0, NULL);
5404         path->leave_spinning = old_spinning;
5405         if (!old_spinning)
5406                 btrfs_set_path_blocking(path);
5407
5408         return ret;
5409 }
5410
5411 /*
5412  * this uses btrfs_prev_leaf to walk backwards in the tree, and keeps
5413  * searching until it gets past min_objectid or finds an item of 'type'
5414  *
5415  * returns 0 if something is found, 1 if nothing was found and < 0 on error
5416  */
5417 int btrfs_previous_item(struct btrfs_root *root,
5418                         struct btrfs_path *path, u64 min_objectid,
5419                         int type)
5420 {
5421         struct btrfs_key found_key;
5422         struct extent_buffer *leaf;
5423         u32 nritems;
5424         int ret;
5425
5426         while (1) {
5427                 if (path->slots[0] == 0) {
5428                         btrfs_set_path_blocking(path);
5429                         ret = btrfs_prev_leaf(root, path);
5430                         if (ret != 0)
5431                                 return ret;
5432                 } else {
5433                         path->slots[0]--;
5434                 }
5435                 leaf = path->nodes[0];
5436                 nritems = btrfs_header_nritems(leaf);
5437                 if (nritems == 0)
5438                         return 1;
5439                 if (path->slots[0] == nritems)
5440                         path->slots[0]--;
5441
5442                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
5443                 if (found_key.objectid < min_objectid)
5444                         break;
5445                 if (found_key.type == type)
5446                         return 0;
5447                 if (found_key.objectid == min_objectid &&
5448                     found_key.type < type)
5449                         break;
5450         }
5451         return 1;
5452 }
5453
5454 /*
5455  * search in extent tree to find a previous Metadata/Data extent item with
5456  * min objecitd.
5457  *
5458  * returns 0 if something is found, 1 if nothing was found and < 0 on error
5459  */
5460 int btrfs_previous_extent_item(struct btrfs_root *root,
5461                         struct btrfs_path *path, u64 min_objectid)
5462 {
5463         struct btrfs_key found_key;
5464         struct extent_buffer *leaf;
5465         u32 nritems;
5466         int ret;
5467
5468         while (1) {
5469                 if (path->slots[0] == 0) {
5470                         btrfs_set_path_blocking(path);
5471                         ret = btrfs_prev_leaf(root, path);
5472                         if (ret != 0)
5473                                 return ret;
5474                 } else {
5475                         path->slots[0]--;
5476                 }
5477                 leaf = path->nodes[0];
5478                 nritems = btrfs_header_nritems(leaf);
5479                 if (nritems == 0)
5480                         return 1;
5481                 if (path->slots[0] == nritems)
5482                         path->slots[0]--;
5483
5484                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
5485                 if (found_key.objectid < min_objectid)
5486                         break;
5487                 if (found_key.type == BTRFS_EXTENT_ITEM_KEY ||
5488                     found_key.type == BTRFS_METADATA_ITEM_KEY)
5489                         return 0;
5490                 if (found_key.objectid == min_objectid &&
5491                     found_key.type < BTRFS_EXTENT_ITEM_KEY)
5492                         break;
5493         }
5494         return 1;
5495 }