fs/btrfs/extent-tree.c

   1 /*
   2  * Copyright (C) 2007 Oracle.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18 #include <linux/sched.h>
  19 #include <linux/pagemap.h>
  20 #include <linux/writeback.h>
  21 #include <linux/blkdev.h>
  22 #include <linux/version.h>
  23 #include "compat.h"
  24 #include "hash.h"
  25 #include "crc32c.h"
  26 #include "ctree.h"
  27 #include "disk-io.h"
  28 #include "print-tree.h"
  29 #include "transaction.h"
  30 #include "volumes.h"
  31 #include "locking.h"
  32 #include "ref-cache.h"
  33 #include "compat.h"
  34
  35 #define PENDING_EXTENT_INSERT 0
  36 #define PENDING_EXTENT_DELETE 1
  37 #define PENDING_BACKREF_UPDATE 2
  38
  39 struct pending_extent_op {
  40         int type;
  41         u64 bytenr;
  42         u64 num_bytes;
  43         u64 parent;
  44         u64 orig_parent;
  45         u64 generation;
  46         u64 orig_generation;
  47         int level;
  48         struct list_head list;
  49         int del;
  50 };
  51
  52 static int finish_current_insert(struct btrfs_trans_handle *trans,
  53                                  struct btrfs_root *extent_root, int all);
  54 static int del_pending_extents(struct btrfs_trans_handle *trans,
  55                                struct btrfs_root *extent_root, int all);
  56 static int pin_down_bytes(struct btrfs_trans_handle *trans,
  57                           struct btrfs_root *root,
  58                           u64 bytenr, u64 num_bytes, int is_data);
  59 static int update_block_group(struct btrfs_trans_handle *trans,
  60                               struct btrfs_root *root,
  61                               u64 bytenr, u64 num_bytes, int alloc,
  62                               int mark_free);
  63
  64 static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
  65 {
  66         return (cache->flags & bits) == bits;
  67 }
  68
  69 /*
  70  * this adds the block group to the fs_info rb tree for the block group
  71  * cache
  72  */
  73 static int btrfs_add_block_group_cache(struct btrfs_fs_info *info,
  74                                 struct btrfs_block_group_cache *block_group)
  75 {
  76         struct rb_node **p;
  77         struct rb_node *parent = NULL;
  78         struct btrfs_block_group_cache *cache;
  79
  80         spin_lock(&info->block_group_cache_lock);
  81         p = &info->block_group_cache_tree.rb_node;
  82
  83         while (*p) {
  84                 parent = *p;
  85                 cache = rb_entry(parent, struct btrfs_block_group_cache,
  86                                  cache_node);
  87                 if (block_group->key.objectid < cache->key.objectid) {
  88                         p = &(*p)->rb_left;
  89                 } else if (block_group->key.objectid > cache->key.objectid) {
  90                         p = &(*p)->rb_right;
  91                 } else {
  92                         spin_unlock(&info->block_group_cache_lock);
  93                         return -EEXIST;
  94                 }
  95         }
  96
  97         rb_link_node(&block_group->cache_node, parent, p);
  98         rb_insert_color(&block_group->cache_node,
  99                         &info->block_group_cache_tree);
 100         spin_unlock(&info->block_group_cache_lock);
 101
 102         return 0;
 103 }
 104
 105 /*
 106  * This will return the block group at or after bytenr if contains is 0, else
 107  * it will return the block group that contains the bytenr
 108  */
 109 static struct btrfs_block_group_cache *
 110 block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr,
 111                               int contains)
 112 {
 113         struct btrfs_block_group_cache *cache, *ret = NULL;
 114         struct rb_node *n;
 115         u64 end, start;
 116
 117         spin_lock(&info->block_group_cache_lock);
 118         n = info->block_group_cache_tree.rb_node;
 119
 120         while (n) {
 121                 cache = rb_entry(n, struct btrfs_block_group_cache,
 122                                  cache_node);
 123                 end = cache->key.objectid + cache->key.offset - 1;
 124                 start = cache->key.objectid;
 125
 126                 if (bytenr < start) {
 127                         if (!contains && (!ret || start < ret->key.objectid))
 128                                 ret = cache;
 129                         n = n->rb_left;
 130                 } else if (bytenr > start) {
 131                         if (contains && bytenr <= end) {
 132                                 ret = cache;
 133                                 break;
 134                         }
 135                         n = n->rb_right;
 136                 } else {
 137                         ret = cache;
 138                         break;
 139                 }
 140         }
 141         if (ret)
 142                 atomic_inc(&ret->count);
 143         spin_unlock(&info->block_group_cache_lock);
 144
 145         return ret;
 146 }
 147
 148 /*
 149  * this is only called by cache_block_group, since we could have freed extents
 150  * we need to check the pinned_extents for any extents that can't be used yet
 151  * since their free space will be released as soon as the transaction commits.
 152  */
 153 static int add_new_free_space(struct btrfs_block_group_cache *block_group,
 154                               struct btrfs_fs_info *info, u64 start, u64 end)
 155 {
 156         u64 extent_start, extent_end, size;
 157         int ret;
 158
 159         mutex_lock(&info->pinned_mutex);
 160         while (start < end) {
 161                 ret = find_first_extent_bit(&info->pinned_extents, start,
 162                                             &extent_start, &extent_end,
 163                                             EXTENT_DIRTY);
 164                 if (ret)
 165                         break;
 166
 167                 if (extent_start == start) {
 168                         start = extent_end + 1;
 169                 } else if (extent_start > start && extent_start < end) {
 170                         size = extent_start - start;
 171                         ret = btrfs_add_free_space(block_group, start,
 172                                                    size);
 173                         BUG_ON(ret);
 174                         start = extent_end + 1;
 175                 } else {
 176                         break;
 177                 }
 178         }
 179
 180         if (start < end) {
 181                 size = end - start;
 182                 ret = btrfs_add_free_space(block_group, start, size);
 183                 BUG_ON(ret);
 184         }
 185         mutex_unlock(&info->pinned_mutex);
 186
 187         return 0;
 188 }
 189
 190 static int remove_sb_from_cache(struct btrfs_root *root,
 191                                 struct btrfs_block_group_cache *cache)
 192 {
 193         u64 bytenr;
 194         u64 *logical;
 195         int stripe_len;
 196         int i, nr, ret;
 197
 198         for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
 199                 bytenr = btrfs_sb_offset(i);
 200                 ret = btrfs_rmap_block(&root->fs_info->mapping_tree,
 201                                        cache->key.objectid, bytenr, 0,
 202                                        &logical, &nr, &stripe_len);
 203                 BUG_ON(ret);
 204                 while (nr--) {
 205                         btrfs_remove_free_space(cache, logical[nr],
 206                                                 stripe_len);
 207                 }
 208                 kfree(logical);
 209         }
 210         return 0;
 211 }
 212
 213 static int cache_block_group(struct btrfs_root *root,
 214                              struct btrfs_block_group_cache *block_group)
 215 {
 216         struct btrfs_path *path;
 217         int ret = 0;
 218         struct btrfs_key key;
 219         struct extent_buffer *leaf;
 220         int slot;
 221         u64 last;
 222
 223         if (!block_group)
 224                 return 0;
 225
 226         root = root->fs_info->extent_root;
 227
 228         if (block_group->cached)
 229                 return 0;
 230
 231         path = btrfs_alloc_path();
 232         if (!path)
 233                 return -ENOMEM;
 234
 235         path->reada = 2;
 236         /*
 237          * we get into deadlocks with paths held by callers of this function.
 238          * since the alloc_mutex is protecting things right now, just
 239          * skip the locking here
 240          */
 241         path->skip_locking = 1;
 242         last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
 243         key.objectid = last;
 244         key.offset = 0;
 245         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
 246         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 247         if (ret < 0)
 248                 goto err;
 249
 250         while (1) {
 251                 leaf = path->nodes[0];
 252                 slot = path->slots[0];
 253                 if (slot >= btrfs_header_nritems(leaf)) {
 254                         ret = btrfs_next_leaf(root, path);
 255                         if (ret < 0)
 256                                 goto err;
 257                         if (ret == 0)
 258                                 continue;
 259                         else
 260                                 break;
 261                 }
 262                 btrfs_item_key_to_cpu(leaf, &key, slot);
 263                 if (key.objectid < block_group->key.objectid)
 264                         goto next;
 265
 266                 if (key.objectid >= block_group->key.objectid +
 267                     block_group->key.offset)
 268                         break;
 269
 270                 if (btrfs_key_type(&key) == BTRFS_EXTENT_ITEM_KEY) {
 271                         add_new_free_space(block_group, root->fs_info, last,
 272                                            key.objectid);
 273
 274                         last = key.objectid + key.offset;
 275                 }
 276 next:
 277                 path->slots[0]++;
 278         }
 279
 280         add_new_free_space(block_group, root->fs_info, last,
 281                            block_group->key.objectid +
 282                            block_group->key.offset);
 283
 284         remove_sb_from_cache(root, block_group);
 285         block_group->cached = 1;
 286         ret = 0;
 287 err:
 288         btrfs_free_path(path);
 289         return ret;
 290 }
 291
 292 /*
 293  * return the block group that starts at or after bytenr
 294  */
 295 static struct btrfs_block_group_cache *
 296 btrfs_lookup_first_block_group(struct btrfs_fs_info *info, u64 bytenr)
 297 {
 298         struct btrfs_block_group_cache *cache;
 299
 300         cache = block_group_cache_tree_search(info, bytenr, 0);
 301
 302         return cache;
 303 }
 304
 305 /*
 306  * return the block group that contains teh given bytenr
 307  */
 308 struct btrfs_block_group_cache *btrfs_lookup_block_group(
 309                                                  struct btrfs_fs_info *info,
 310                                                  u64 bytenr)
 311 {
 312         struct btrfs_block_group_cache *cache;
 313
 314         cache = block_group_cache_tree_search(info, bytenr, 1);
 315
 316         return cache;
 317 }
 318
 319 static inline void put_block_group(struct btrfs_block_group_cache *cache)
 320 {
 321         if (atomic_dec_and_test(&cache->count))
 322                 kfree(cache);
 323 }
 324
 325 static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
 326                                                   u64 flags)
 327 {
 328         struct list_head *head = &info->space_info;
 329         struct list_head *cur;
 330         struct btrfs_space_info *found;
 331         list_for_each(cur, head) {
 332                 found = list_entry(cur, struct btrfs_space_info, list);
 333                 if (found->flags == flags)
 334                         return found;
 335         }
 336         return NULL;
 337 }
 338
 339 static u64 div_factor(u64 num, int factor)
 340 {
 341         if (factor == 10)
 342                 return num;
 343         num *= factor;
 344         do_div(num, 10);
 345         return num;
 346 }
 347
 348 u64 btrfs_find_block_group(struct btrfs_root *root,
 349                            u64 search_start, u64 search_hint, int owner)
 350 {
 351         struct btrfs_block_group_cache *cache;
 352         u64 used;
 353         u64 last = max(search_hint, search_start);
 354         u64 group_start = 0;
 355         int full_search = 0;
 356         int factor = 9;
 357         int wrapped = 0;
 358 again:
 359         while (1) {
 360                 cache = btrfs_lookup_first_block_group(root->fs_info, last);
 361                 if (!cache)
 362                         break;
 363
 364                 spin_lock(&cache->lock);
 365                 last = cache->key.objectid + cache->key.offset;
 366                 used = btrfs_block_group_used(&cache->item);
 367
 368                 if ((full_search || !cache->ro) &&
 369                     block_group_bits(cache, BTRFS_BLOCK_GROUP_METADATA)) {
 370                         if (used + cache->pinned + cache->reserved <
 371                             div_factor(cache->key.offset, factor)) {
 372                                 group_start = cache->key.objectid;
 373                                 spin_unlock(&cache->lock);
 374                                 put_block_group(cache);
 375                                 goto found;
 376                         }
 377                 }
 378                 spin_unlock(&cache->lock);
 379                 put_block_group(cache);
 380                 cond_resched();
 381         }
 382         if (!wrapped) {
 383                 last = search_start;
 384                 wrapped = 1;
 385                 goto again;
 386         }
 387         if (!full_search && factor < 10) {
 388                 last = search_start;
 389                 full_search = 1;
 390                 factor = 10;
 391                 goto again;
 392         }
 393 found:
 394         return group_start;
 395 }
 396
 397 /* simple helper to search for an existing extent at a given offset */
 398 int btrfs_lookup_extent(struct btrfs_root *root, u64 start, u64 len)
 399 {
 400         int ret;
 401         struct btrfs_key key;
 402         struct btrfs_path *path;
 403
 404         path = btrfs_alloc_path();
 405         BUG_ON(!path);
 406         key.objectid = start;
 407         key.offset = len;
 408         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
 409         ret = btrfs_search_slot(NULL, root->fs_info->extent_root, &key, path,
 410                                 0, 0);
 411         btrfs_free_path(path);
 412         return ret;
 413 }
 414
 415 /*
 416  * Back reference rules.  Back refs have three main goals:
 417  *
 418  * 1) differentiate between all holders of references to an extent so that
 419  *    when a reference is dropped we can make sure it was a valid reference
 420  *    before freeing the extent.
 421  *
 422  * 2) Provide enough information to quickly find the holders of an extent
 423  *    if we notice a given block is corrupted or bad.
 424  *
 425  * 3) Make it easy to migrate blocks for FS shrinking or storage pool
 426  *    maintenance.  This is actually the same as #2, but with a slightly
 427  *    different use case.
 428  *
 429  * File extents can be referenced by:
 430  *
 431  * - multiple snapshots, subvolumes, or different generations in one subvol
 432  * - different files inside a single subvolume
 433  * - different offsets inside a file (bookend extents in file.c)
 434  *
 435  * The extent ref structure has fields for:
 436  *
 437  * - Objectid of the subvolume root
 438  * - Generation number of the tree holding the reference
 439  * - objectid of the file holding the reference
 440  * - number of references holding by parent node (alway 1 for tree blocks)
 441  *
 442  * Btree leaf may hold multiple references to a file extent. In most cases,
 443  * these references are from same file and the corresponding offsets inside
 444  * the file are close together.
 445  *
 446  * When a file extent is allocated the fields are filled in:
 447  *     (root_key.objectid, trans->transid, inode objectid, 1)
 448  *
 449  * When a leaf is cow'd new references are added for every file extent found
 450  * in the leaf.  It looks similar to the create case, but trans->transid will
 451  * be different when the block is cow'd.
 452  *
 453  *     (root_key.objectid, trans->transid, inode objectid,
 454  *      number of references in the leaf)
 455  *
 456  * When a file extent is removed either during snapshot deletion or
 457  * file truncation, we find the corresponding back reference and check
 458  * the following fields:
 459  *
 460  *     (btrfs_header_owner(leaf), btrfs_header_generation(leaf),
 461  *      inode objectid)
 462  *
 463  * Btree extents can be referenced by:
 464  *
 465  * - Different subvolumes
 466  * - Different generations of the same subvolume
 467  *
 468  * When a tree block is created, back references are inserted:
 469  *
 470  * (root->root_key.objectid, trans->transid, level, 1)
 471  *
 472  * When a tree block is cow'd, new back references are added for all the
 473  * blocks it points to. If the tree block isn't in reference counted root,
 474  * the old back references are removed. These new back references are of
 475  * the form (trans->transid will have increased since creation):
 476  *
 477  * (root->root_key.objectid, trans->transid, level, 1)
 478  *
 479  * When a backref is in deleting, the following fields are checked:
 480  *
 481  * if backref was for a tree root:
 482  *     (btrfs_header_owner(itself), btrfs_header_generation(itself), level)
 483  * else
 484  *     (btrfs_header_owner(parent), btrfs_header_generation(parent), level)
 485  *
 486  * Back Reference Key composing:
 487  *
 488  * The key objectid corresponds to the first byte in the extent, the key
 489  * type is set to BTRFS_EXTENT_REF_KEY, and the key offset is the first
 490  * byte of parent extent. If a extent is tree root, the key offset is set
 491  * to the key objectid.
 492  */
 493
 494 static noinline int lookup_extent_backref(struct btrfs_trans_handle *trans,
 495                                           struct btrfs_root *root,
 496                                           struct btrfs_path *path,
 497                                           u64 bytenr, u64 parent,
 498                                           u64 ref_root, u64 ref_generation,
 499                                           u64 owner_objectid, int del)
 500 {
 501         struct btrfs_key key;
 502         struct btrfs_extent_ref *ref;
 503         struct extent_buffer *leaf;
 504         u64 ref_objectid;
 505         int ret;
 506
 507         key.objectid = bytenr;
 508         key.type = BTRFS_EXTENT_REF_KEY;
 509         key.offset = parent;
 510
 511         ret = btrfs_search_slot(trans, root, &key, path, del ? -1 : 0, 1);
 512         if (ret < 0)
 513                 goto out;
 514         if (ret > 0) {
 515                 ret = -ENOENT;
 516                 goto out;
 517         }
 518
 519         leaf = path->nodes[0];
 520         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 521         ref_objectid = btrfs_ref_objectid(leaf, ref);
 522         if (btrfs_ref_root(leaf, ref) != ref_root ||
 523             btrfs_ref_generation(leaf, ref) != ref_generation ||
 524             (ref_objectid != owner_objectid &&
 525              ref_objectid != BTRFS_MULTIPLE_OBJECTIDS)) {
 526                 ret = -EIO;
 527                 WARN_ON(1);
 528                 goto out;
 529         }
 530         ret = 0;
 531 out:
 532         return ret;
 533 }
 534
 535 /*
 536  * updates all the backrefs that are pending on update_list for the
 537  * extent_root
 538  */
 539 static noinline int update_backrefs(struct btrfs_trans_handle *trans,
 540                                     struct btrfs_root *extent_root,
 541                                     struct btrfs_path *path,
 542                                     struct list_head *update_list)
 543 {
 544         struct btrfs_key key;
 545         struct btrfs_extent_ref *ref;
 546         struct btrfs_fs_info *info = extent_root->fs_info;
 547         struct pending_extent_op *op;
 548         struct extent_buffer *leaf;
 549         int ret = 0;
 550         struct list_head *cur = update_list->next;
 551         u64 ref_objectid;
 552         u64 ref_root = extent_root->root_key.objectid;
 553
 554         op = list_entry(cur, struct pending_extent_op, list);
 555
 556 search:
 557         key.objectid = op->bytenr;
 558         key.type = BTRFS_EXTENT_REF_KEY;
 559         key.offset = op->orig_parent;
 560
 561         ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 1);
 562         BUG_ON(ret);
 563
 564         leaf = path->nodes[0];
 565
 566 loop:
 567         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 568
 569         ref_objectid = btrfs_ref_objectid(leaf, ref);
 570
 571         if (btrfs_ref_root(leaf, ref) != ref_root ||
 572             btrfs_ref_generation(leaf, ref) != op->orig_generation ||
 573             (ref_objectid != op->level &&
 574              ref_objectid != BTRFS_MULTIPLE_OBJECTIDS)) {
 575                 printk(KERN_ERR "btrfs couldn't find %llu, parent %llu, "
 576                        "root %llu, owner %u\n",
 577                        (unsigned long long)op->bytenr,
 578                        (unsigned long long)op->orig_parent,
 579                        (unsigned long long)ref_root, op->level);
 580                 btrfs_print_leaf(extent_root, leaf);
 581                 BUG();
 582         }
 583
 584         key.objectid = op->bytenr;
 585         key.offset = op->parent;
 586         key.type = BTRFS_EXTENT_REF_KEY;
 587         ret = btrfs_set_item_key_safe(trans, extent_root, path, &key);
 588         BUG_ON(ret);
 589         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 590         btrfs_set_ref_generation(leaf, ref, op->generation);
 591
 592         cur = cur->next;
 593
 594         list_del_init(&op->list);
 595         unlock_extent(&info->extent_ins, op->bytenr,
 596                       op->bytenr + op->num_bytes - 1, GFP_NOFS);
 597         kfree(op);
 598
 599         if (cur == update_list) {
 600                 btrfs_mark_buffer_dirty(path->nodes[0]);
 601                 btrfs_release_path(extent_root, path);
 602                 goto out;
 603         }
 604
 605         op = list_entry(cur, struct pending_extent_op, list);
 606
 607         path->slots[0]++;
 608         while (path->slots[0] < btrfs_header_nritems(leaf)) {
 609                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 610                 if (key.objectid == op->bytenr &&
 611                     key.type == BTRFS_EXTENT_REF_KEY)
 612                         goto loop;
 613                 path->slots[0]++;
 614         }
 615
 616         btrfs_mark_buffer_dirty(path->nodes[0]);
 617         btrfs_release_path(extent_root, path);
 618         goto search;
 619
 620 out:
 621         return 0;
 622 }
 623
 624 static noinline int insert_extents(struct btrfs_trans_handle *trans,
 625                                    struct btrfs_root *extent_root,
 626                                    struct btrfs_path *path,
 627                                    struct list_head *insert_list, int nr)
 628 {
 629         struct btrfs_key *keys;
 630         u32 *data_size;
 631         struct pending_extent_op *op;
 632         struct extent_buffer *leaf;
 633         struct list_head *cur = insert_list->next;
 634         struct btrfs_fs_info *info = extent_root->fs_info;
 635         u64 ref_root = extent_root->root_key.objectid;
 636         int i = 0, last = 0, ret;
 637         int total = nr * 2;
 638
 639         if (!nr)
 640                 return 0;
 641
 642         keys = kzalloc(total * sizeof(struct btrfs_key), GFP_NOFS);
 643         if (!keys)
 644                 return -ENOMEM;
 645
 646         data_size = kzalloc(total * sizeof(u32), GFP_NOFS);
 647         if (!data_size) {
 648                 kfree(keys);
 649                 return -ENOMEM;
 650         }
 651
 652         list_for_each_entry(op, insert_list, list) {
 653                 keys[i].objectid = op->bytenr;
 654                 keys[i].offset = op->num_bytes;
 655                 keys[i].type = BTRFS_EXTENT_ITEM_KEY;
 656                 data_size[i] = sizeof(struct btrfs_extent_item);
 657                 i++;
 658
 659                 keys[i].objectid = op->bytenr;
 660                 keys[i].offset = op->parent;
 661                 keys[i].type = BTRFS_EXTENT_REF_KEY;
 662                 data_size[i] = sizeof(struct btrfs_extent_ref);
 663                 i++;
 664         }
 665
 666         op = list_entry(cur, struct pending_extent_op, list);
 667         i = 0;
 668         while (i < total) {
 669                 int c;
 670                 ret = btrfs_insert_some_items(trans, extent_root, path,
 671                                               keys+i, data_size+i, total-i);
 672                 BUG_ON(ret < 0);
 673
 674                 if (last && ret > 1)
 675                         BUG();
 676
 677                 leaf = path->nodes[0];
 678                 for (c = 0; c < ret; c++) {
 679                         int ref_first = keys[i].type == BTRFS_EXTENT_REF_KEY;
 680
 681                         /*
 682                          * if the first item we inserted was a backref, then
 683                          * the EXTENT_ITEM will be the odd c's, else it will
 684                          * be the even c's
 685                          */
 686                         if ((ref_first && (c % 2)) ||
 687                             (!ref_first && !(c % 2))) {
 688                                 struct btrfs_extent_item *itm;
 689
 690                                 itm = btrfs_item_ptr(leaf, path->slots[0] + c,
 691                                                      struct btrfs_extent_item);
 692                                 btrfs_set_extent_refs(path->nodes[0], itm, 1);
 693                                 op->del++;
 694                         } else {
 695                                 struct btrfs_extent_ref *ref;
 696
 697                                 ref = btrfs_item_ptr(leaf, path->slots[0] + c,
 698                                                      struct btrfs_extent_ref);
 699                                 btrfs_set_ref_root(leaf, ref, ref_root);
 700                                 btrfs_set_ref_generation(leaf, ref,
 701                                                          op->generation);
 702                                 btrfs_set_ref_objectid(leaf, ref, op->level);
 703                                 btrfs_set_ref_num_refs(leaf, ref, 1);
 704                                 op->del++;
 705                         }
 706
 707                         /*
 708                          * using del to see when its ok to free up the
 709                          * pending_extent_op.  In the case where we insert the
 710                          * last item on the list in order to help do batching
 711                          * we need to not free the extent op until we actually
 712                          * insert the extent_item
 713                          */
 714                         if (op->del == 2) {
 715                                 unlock_extent(&info->extent_ins, op->bytenr,
 716                                               op->bytenr + op->num_bytes - 1,
 717                                               GFP_NOFS);
 718                                 cur = cur->next;
 719                                 list_del_init(&op->list);
 720                                 kfree(op);
 721                                 if (cur != insert_list)
 722                                         op = list_entry(cur,
 723                                                 struct pending_extent_op,
 724                                                 list);
 725                         }
 726                 }
 727                 btrfs_mark_buffer_dirty(leaf);
 728                 btrfs_release_path(extent_root, path);
 729
 730                 /*
 731                  * Ok backref's and items usually go right next to eachother,
 732                  * but if we could only insert 1 item that means that we
 733                  * inserted on the end of a leaf, and we have no idea what may
 734                  * be on the next leaf so we just play it safe.  In order to
 735                  * try and help this case we insert the last thing on our
 736                  * insert list so hopefully it will end up being the last
 737                  * thing on the leaf and everything else will be before it,
 738                  * which will let us insert a whole bunch of items at the same
 739                  * time.
 740                  */
 741                 if (ret == 1 && !last && (i + ret < total)) {
 742                         /*
 743                          * last: where we will pick up the next time around
 744                          * i: our current key to insert, will be total - 1
 745                          * cur: the current op we are screwing with
 746                          * op: duh
 747                          */
 748                         last = i + ret;
 749                         i = total - 1;
 750                         cur = insert_list->prev;
 751                         op = list_entry(cur, struct pending_extent_op, list);
 752                 } else if (last) {
 753                         /*
 754                          * ok we successfully inserted the last item on the
 755                          * list, lets reset everything
 756                          *
 757                          * i: our current key to insert, so where we left off
 758                          *    last time
 759                          * last: done with this
 760                          * cur: the op we are messing with
 761                          * op: duh
 762                          * total: since we inserted the last key, we need to
 763                          *        decrement total so we dont overflow
 764                          */
 765                         i = last;
 766                         last = 0;
 767                         total--;
 768                         if (i < total) {
 769                                 cur = insert_list->next;
 770                                 op = list_entry(cur, struct pending_extent_op,
 771                                                 list);
 772                         }
 773                 } else {
 774                         i += ret;
 775                 }
 776
 777                 cond_resched();
 778         }
 779         ret = 0;
 780         kfree(keys);
 781         kfree(data_size);
 782         return ret;
 783 }
 784
 785 static noinline int insert_extent_backref(struct btrfs_trans_handle *trans,
 786                                           struct btrfs_root *root,
 787                                           struct btrfs_path *path,
 788                                           u64 bytenr, u64 parent,
 789                                           u64 ref_root, u64 ref_generation,
 790                                           u64 owner_objectid)
 791 {
 792         struct btrfs_key key;
 793         struct extent_buffer *leaf;
 794         struct btrfs_extent_ref *ref;
 795         u32 num_refs;
 796         int ret;
 797
 798         key.objectid = bytenr;
 799         key.type = BTRFS_EXTENT_REF_KEY;
 800         key.offset = parent;
 801
 802         ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(*ref));
 803         if (ret == 0) {
 804                 leaf = path->nodes[0];
 805                 ref = btrfs_item_ptr(leaf, path->slots[0],
 806                                      struct btrfs_extent_ref);
 807                 btrfs_set_ref_root(leaf, ref, ref_root);
 808                 btrfs_set_ref_generation(leaf, ref, ref_generation);
 809                 btrfs_set_ref_objectid(leaf, ref, owner_objectid);
 810                 btrfs_set_ref_num_refs(leaf, ref, 1);
 811         } else if (ret == -EEXIST) {
 812                 u64 existing_owner;
 813                 BUG_ON(owner_objectid < BTRFS_FIRST_FREE_OBJECTID);
 814                 leaf = path->nodes[0];
 815                 ref = btrfs_item_ptr(leaf, path->slots[0],
 816                                      struct btrfs_extent_ref);
 817                 if (btrfs_ref_root(leaf, ref) != ref_root ||
 818                     btrfs_ref_generation(leaf, ref) != ref_generation) {
 819                         ret = -EIO;
 820                         WARN_ON(1);
 821                         goto out;
 822                 }
 823
 824                 num_refs = btrfs_ref_num_refs(leaf, ref);
 825                 BUG_ON(num_refs == 0);
 826                 btrfs_set_ref_num_refs(leaf, ref, num_refs + 1);
 827
 828                 existing_owner = btrfs_ref_objectid(leaf, ref);
 829                 if (existing_owner != owner_objectid &&
 830                     existing_owner != BTRFS_MULTIPLE_OBJECTIDS) {
 831                         btrfs_set_ref_objectid(leaf, ref,
 832                                         BTRFS_MULTIPLE_OBJECTIDS);
 833                 }
 834                 ret = 0;
 835         } else {
 836                 goto out;
 837         }
 838         btrfs_mark_buffer_dirty(path->nodes[0]);
 839 out:
 840         btrfs_release_path(root, path);
 841         return ret;
 842 }
 843
 844 static noinline int remove_extent_backref(struct btrfs_trans_handle *trans,
 845                                           struct btrfs_root *root,
 846                                           struct btrfs_path *path)
 847 {
 848         struct extent_buffer *leaf;
 849         struct btrfs_extent_ref *ref;
 850         u32 num_refs;
 851         int ret = 0;
 852
 853         leaf = path->nodes[0];
 854         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 855         num_refs = btrfs_ref_num_refs(leaf, ref);
 856         BUG_ON(num_refs == 0);
 857         num_refs -= 1;
 858         if (num_refs == 0) {
 859                 ret = btrfs_del_item(trans, root, path);
 860         } else {
 861                 btrfs_set_ref_num_refs(leaf, ref, num_refs);
 862                 btrfs_mark_buffer_dirty(leaf);
 863         }
 864         btrfs_release_path(root, path);
 865         return ret;
 866 }
 867
 868 #ifdef BIO_RW_DISCARD
 869 static void btrfs_issue_discard(struct block_device *bdev,
 870                                 u64 start, u64 len)
 871 {
 872         blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL);
 873 }
 874 #endif
 875
 876 static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 877                                 u64 num_bytes)
 878 {
 879 #ifdef BIO_RW_DISCARD
 880         int ret;
 881         u64 map_length = num_bytes;
 882         struct btrfs_multi_bio *multi = NULL;
 883
 884         /* Tell the block device(s) that the sectors can be discarded */
 885         ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
 886                               bytenr, &map_length, &multi, 0);
 887         if (!ret) {
 888                 struct btrfs_bio_stripe *stripe = multi->stripes;
 889                 int i;
 890
 891                 if (map_length > num_bytes)
 892                         map_length = num_bytes;
 893
 894                 for (i = 0; i < multi->num_stripes; i++, stripe++) {
 895                         btrfs_issue_discard(stripe->dev->bdev,
 896                                             stripe->physical,
 897                                             map_length);
 898                 }
 899                 kfree(multi);
 900         }
 901
 902         return ret;
 903 #else
 904         return 0;
 905 #endif
 906 }
 907
 908 static noinline int free_extents(struct btrfs_trans_handle *trans,
 909                                  struct btrfs_root *extent_root,
 910                                  struct list_head *del_list)
 911 {
 912         struct btrfs_fs_info *info = extent_root->fs_info;
 913         struct btrfs_path *path;
 914         struct btrfs_key key, found_key;
 915         struct extent_buffer *leaf;
 916         struct list_head *cur;
 917         struct pending_extent_op *op;
 918         struct btrfs_extent_item *ei;
 919         int ret, num_to_del, extent_slot = 0, found_extent = 0;
 920         u32 refs;
 921         u64 bytes_freed = 0;
 922
 923         path = btrfs_alloc_path();
 924         if (!path)
 925                 return -ENOMEM;
 926         path->reada = 1;
 927
 928 search:
 929         /* search for the backref for the current ref we want to delete */
 930         cur = del_list->next;
 931         op = list_entry(cur, struct pending_extent_op, list);
 932         ret = lookup_extent_backref(trans, extent_root, path, op->bytenr,
 933                                     op->orig_parent,
 934                                     extent_root->root_key.objectid,
 935                                     op->orig_generation, op->level, 1);
 936         if (ret) {
 937                 printk(KERN_ERR "btrfs unable to find backref byte nr %llu "
 938                        "root %llu gen %llu owner %u\n",
 939                        (unsigned long long)op->bytenr,
 940                        (unsigned long long)extent_root->root_key.objectid,
 941                        (unsigned long long)op->orig_generation, op->level);
 942                 btrfs_print_leaf(extent_root, path->nodes[0]);
 943                 WARN_ON(1);
 944                 goto out;
 945         }
 946
 947         extent_slot = path->slots[0];
 948         num_to_del = 1;
 949         found_extent = 0;
 950
 951         /*
 952          * if we aren't the first item on the leaf we can move back one and see
 953          * if our ref is right next to our extent item
 954          */
 955         if (likely(extent_slot)) {
 956                 extent_slot--;
 957                 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
 958                                       extent_slot);
 959                 if (found_key.objectid == op->bytenr &&
 960                     found_key.type == BTRFS_EXTENT_ITEM_KEY &&
 961                     found_key.offset == op->num_bytes) {
 962                         num_to_del++;
 963                         found_extent = 1;
 964                 }
 965         }
 966
 967         /*
 968          * if we didn't find the extent we need to delete the backref and then
 969          * search for the extent item key so we can update its ref count
 970          */
 971         if (!found_extent) {
 972                 key.objectid = op->bytenr;
 973                 key.type = BTRFS_EXTENT_ITEM_KEY;
 974                 key.offset = op->num_bytes;
 975
 976                 ret = remove_extent_backref(trans, extent_root, path);
 977                 BUG_ON(ret);
 978                 btrfs_release_path(extent_root, path);
 979                 ret = btrfs_search_slot(trans, extent_root, &key, path, -1, 1);
 980                 BUG_ON(ret);
 981                 extent_slot = path->slots[0];
 982         }
 983
 984         /* this is where we update the ref count for the extent */
 985         leaf = path->nodes[0];
 986         ei = btrfs_item_ptr(leaf, extent_slot, struct btrfs_extent_item);
 987         refs = btrfs_extent_refs(leaf, ei);
 988         BUG_ON(refs == 0);
 989         refs--;
 990         btrfs_set_extent_refs(leaf, ei, refs);
 991
 992         btrfs_mark_buffer_dirty(leaf);
 993
 994         /*
 995          * This extent needs deleting.  The reason cur_slot is extent_slot +
 996          * num_to_del is because extent_slot points to the slot where the extent
 997          * is, and if the backref was not right next to the extent we will be
 998          * deleting at least 1 item, and will want to start searching at the
 999          * slot directly next to extent_slot.  However if we did find the
1000          * backref next to the extent item them we will be deleting at least 2
1001          * items and will want to start searching directly after the ref slot
1002          */
1003         if (!refs) {
1004                 struct list_head *pos, *n, *end;
1005                 int cur_slot = extent_slot+num_to_del;
1006                 u64 super_used;
1007                 u64 root_used;
1008
1009                 path->slots[0] = extent_slot;
1010                 bytes_freed = op->num_bytes;
1011
1012                 mutex_lock(&info->pinned_mutex);
1013                 ret = pin_down_bytes(trans, extent_root, op->bytenr,
1014                                      op->num_bytes, op->level >=
1015                                      BTRFS_FIRST_FREE_OBJECTID);
1016                 mutex_unlock(&info->pinned_mutex);
1017                 BUG_ON(ret < 0);
1018                 op->del = ret;
1019
1020                 /*
1021                  * we need to see if we can delete multiple things at once, so
1022                  * start looping through the list of extents we are wanting to
1023                  * delete and see if their extent/backref's are right next to
1024                  * eachother and the extents only have 1 ref
1025                  */
1026                 for (pos = cur->next; pos != del_list; pos = pos->next) {
1027                         struct pending_extent_op *tmp;
1028
1029                         tmp = list_entry(pos, struct pending_extent_op, list);
1030
1031                         /* we only want to delete extent+ref at this stage */
1032                         if (cur_slot >= btrfs_header_nritems(leaf) - 1)
1033                                 break;
1034
1035                         btrfs_item_key_to_cpu(leaf, &found_key, cur_slot);
1036                         if (found_key.objectid != tmp->bytenr ||
1037                             found_key.type != BTRFS_EXTENT_ITEM_KEY ||
1038                             found_key.offset != tmp->num_bytes)
1039                                 break;
1040
1041                         /* check to make sure this extent only has one ref */
1042                         ei = btrfs_item_ptr(leaf, cur_slot,
1043                                             struct btrfs_extent_item);
1044                         if (btrfs_extent_refs(leaf, ei) != 1)
1045                                 break;
1046
1047                         btrfs_item_key_to_cpu(leaf, &found_key, cur_slot+1);
1048                         if (found_key.objectid != tmp->bytenr ||
1049                             found_key.type != BTRFS_EXTENT_REF_KEY ||
1050                             found_key.offset != tmp->orig_parent)
1051                                 break;
1052
1053                         /*
1054                          * the ref is right next to the extent, we can set the
1055                          * ref count to 0 since we will delete them both now
1056                          */
1057                         btrfs_set_extent_refs(leaf, ei, 0);
1058
1059                         /* pin down the bytes for this extent */
1060                         mutex_lock(&info->pinned_mutex);
1061                         ret = pin_down_bytes(trans, extent_root, tmp->bytenr,
1062                                              tmp->num_bytes, tmp->level >=
1063                                              BTRFS_FIRST_FREE_OBJECTID);
1064                         mutex_unlock(&info->pinned_mutex);
1065                         BUG_ON(ret < 0);
1066
1067                         /*
1068                          * use the del field to tell if we need to go ahead and
1069                          * free up the extent when we delete the item or not.
1070                          */
1071                         tmp->del = ret;
1072                         bytes_freed += tmp->num_bytes;
1073
1074                         num_to_del += 2;
1075                         cur_slot += 2;
1076                 }
1077                 end = pos;
1078
1079                 /* update the free space counters */
1080                 spin_lock(&info->delalloc_lock);
1081                 super_used = btrfs_super_bytes_used(&info->super_copy);
1082                 btrfs_set_super_bytes_used(&info->super_copy,
1083                                            super_used - bytes_freed);
1084
1085                 root_used = btrfs_root_used(&extent_root->root_item);
1086                 btrfs_set_root_used(&extent_root->root_item,
1087                                     root_used - bytes_freed);
1088                 spin_unlock(&info->delalloc_lock);
1089
1090                 /* delete the items */
1091                 ret = btrfs_del_items(trans, extent_root, path,
1092                                       path->slots[0], num_to_del);
1093                 BUG_ON(ret);
1094
1095                 /*
1096                  * loop through the extents we deleted and do the cleanup work
1097                  * on them
1098                  */
1099                 for (pos = cur, n = pos->next; pos != end;
1100                      pos = n, n = pos->next) {
1101                         struct pending_extent_op *tmp;
1102                         tmp = list_entry(pos, struct pending_extent_op, list);
1103
1104                         /*
1105                          * remember tmp->del tells us wether or not we pinned
1106                          * down the extent
1107                          */
1108                         ret = update_block_group(trans, extent_root,
1109                                                  tmp->bytenr, tmp->num_bytes, 0,
1110                                                  tmp->del);
1111                         BUG_ON(ret);
1112
1113                         list_del_init(&tmp->list);
1114                         unlock_extent(&info->extent_ins, tmp->bytenr,
1115                                       tmp->bytenr + tmp->num_bytes - 1,
1116                                       GFP_NOFS);
1117                         kfree(tmp);
1118                 }
1119         } else if (refs && found_extent) {
1120                 /*
1121                  * the ref and extent were right next to eachother, but the
1122                  * extent still has a ref, so just free the backref and keep
1123                  * going
1124                  */
1125                 ret = remove_extent_backref(trans, extent_root, path);
1126                 BUG_ON(ret);
1127
1128                 list_del_init(&op->list);
1129                 unlock_extent(&info->extent_ins, op->bytenr,
1130                               op->bytenr + op->num_bytes - 1, GFP_NOFS);
1131                 kfree(op);
1132         } else {
1133                 /*
1134                  * the extent has multiple refs and the backref we were looking
1135                  * for was not right next to it, so just unlock and go next,
1136                  * we're good to go
1137                  */
1138                 list_del_init(&op->list);
1139                 unlock_extent(&info->extent_ins, op->bytenr,
1140                               op->bytenr + op->num_bytes - 1, GFP_NOFS);
1141                 kfree(op);
1142         }
1143
1144         btrfs_release_path(extent_root, path);
1145         if (!list_empty(del_list))
1146                 goto search;
1147
1148 out:
1149         btrfs_free_path(path);
1150         return ret;
1151 }
1152
1153 static int __btrfs_update_extent_ref(struct btrfs_trans_handle *trans,
1154                                      struct btrfs_root *root, u64 bytenr,
1155                                      u64 orig_parent, u64 parent,
1156                                      u64 orig_root, u64 ref_root,
1157                                      u64 orig_generation, u64 ref_generation,
1158                                      u64 owner_objectid)
1159 {
1160         int ret;
1161         struct btrfs_root *extent_root = root->fs_info->extent_root;
1162         struct btrfs_path *path;
1163
1164         if (root == root->fs_info->extent_root) {
1165                 struct pending_extent_op *extent_op;
1166                 u64 num_bytes;
1167
1168                 BUG_ON(owner_objectid >= BTRFS_MAX_LEVEL);
1169                 num_bytes = btrfs_level_size(root, (int)owner_objectid);
1170                 mutex_lock(&root->fs_info->extent_ins_mutex);
1171                 if (test_range_bit(&root->fs_info->extent_ins, bytenr,
1172                                 bytenr + num_bytes - 1, EXTENT_WRITEBACK, 0)) {
1173                         u64 priv;
1174                         ret = get_state_private(&root->fs_info->extent_ins,
1175                                                 bytenr, &priv);
1176                         BUG_ON(ret);
1177                         extent_op = (struct pending_extent_op *)
1178                                                         (unsigned long)priv;
1179                         BUG_ON(extent_op->parent != orig_parent);
1180                         BUG_ON(extent_op->generation != orig_generation);
1181
1182                         extent_op->parent = parent;
1183                         extent_op->generation = ref_generation;
1184                 } else {
1185                         extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
1186                         BUG_ON(!extent_op);
1187
1188                         extent_op->type = PENDING_BACKREF_UPDATE;
1189                         extent_op->bytenr = bytenr;
1190                         extent_op->num_bytes = num_bytes;
1191                         extent_op->parent = parent;
1192                         extent_op->orig_parent = orig_parent;
1193                         extent_op->generation = ref_generation;
1194                         extent_op->orig_generation = orig_generation;
1195                         extent_op->level = (int)owner_objectid;
1196                         INIT_LIST_HEAD(&extent_op->list);
1197                         extent_op->del = 0;
1198
1199                         set_extent_bits(&root->fs_info->extent_ins,
1200                                         bytenr, bytenr + num_bytes - 1,
1201                                         EXTENT_WRITEBACK, GFP_NOFS);
1202                         set_state_private(&root->fs_info->extent_ins,
1203                                           bytenr, (unsigned long)extent_op);
1204                 }
1205                 mutex_unlock(&root->fs_info->extent_ins_mutex);
1206                 return 0;
1207         }
1208
1209         path = btrfs_alloc_path();
1210         if (!path)
1211                 return -ENOMEM;
1212         ret = lookup_extent_backref(trans, extent_root, path,
1213                                     bytenr, orig_parent, orig_root,
1214                                     orig_generation, owner_objectid, 1);
1215         if (ret)
1216                 goto out;
1217         ret = remove_extent_backref(trans, extent_root, path);
1218         if (ret)
1219                 goto out;
1220         ret = insert_extent_backref(trans, extent_root, path, bytenr,
1221                                     parent, ref_root, ref_generation,
1222                                     owner_objectid);
1223         BUG_ON(ret);
1224         finish_current_insert(trans, extent_root, 0);
1225         del_pending_extents(trans, extent_root, 0);
1226 out:
1227         btrfs_free_path(path);
1228         return ret;
1229 }
1230
1231 int btrfs_update_extent_ref(struct btrfs_trans_handle *trans,
1232                             struct btrfs_root *root, u64 bytenr,
1233                             u64 orig_parent, u64 parent,
1234                             u64 ref_root, u64 ref_generation,
1235                             u64 owner_objectid)
1236 {
1237         int ret;
1238         if (ref_root == BTRFS_TREE_LOG_OBJECTID &&
1239             owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
1240                 return 0;
1241         ret = __btrfs_update_extent_ref(trans, root, bytenr, orig_parent,
1242                                         parent, ref_root, ref_root,
1243                                         ref_generation, ref_generation,
1244                                         owner_objectid);
1245         return ret;
1246 }
1247
1248 static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1249                                   struct btrfs_root *root, u64 bytenr,
1250                                   u64 orig_parent, u64 parent,
1251                                   u64 orig_root, u64 ref_root,
1252                                   u64 orig_generation, u64 ref_generation,
1253                                   u64 owner_objectid)
1254 {
1255         struct btrfs_path *path;
1256         int ret;
1257         struct btrfs_key key;
1258         struct extent_buffer *l;
1259         struct btrfs_extent_item *item;
1260         u32 refs;
1261
1262         path = btrfs_alloc_path();
1263         if (!path)
1264                 return -ENOMEM;
1265
1266         path->reada = 1;
1267         key.objectid = bytenr;
1268         key.type = BTRFS_EXTENT_ITEM_KEY;
1269         key.offset = (u64)-1;
1270
1271         ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
1272                                 0, 1);
1273         if (ret < 0)
1274                 return ret;
1275         BUG_ON(ret == 0 || path->slots[0] == 0);
1276
1277         path->slots[0]--;
1278         l = path->nodes[0];
1279
1280         btrfs_item_key_to_cpu(l, &key, path->slots[0]);
1281         if (key.objectid != bytenr) {
1282                 btrfs_print_leaf(root->fs_info->extent_root, path->nodes[0]);
1283                 printk(KERN_ERR "btrfs wanted %llu found %llu\n",
1284                        (unsigned long long)bytenr,
1285                        (unsigned long long)key.objectid);
1286                 BUG();
1287         }
1288         BUG_ON(key.type != BTRFS_EXTENT_ITEM_KEY);
1289
1290         item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
1291         refs = btrfs_extent_refs(l, item);
1292         btrfs_set_extent_refs(l, item, refs + 1);
1293         btrfs_mark_buffer_dirty(path->nodes[0]);
1294
1295         btrfs_release_path(root->fs_info->extent_root, path);
1296
1297         path->reada = 1;
1298         ret = insert_extent_backref(trans, root->fs_info->extent_root,
1299                                     path, bytenr, parent,
1300                                     ref_root, ref_generation,
1301                                     owner_objectid);
1302         BUG_ON(ret);
1303         finish_current_insert(trans, root->fs_info->extent_root, 0);
1304         del_pending_extents(trans, root->fs_info->extent_root, 0);
1305
1306         btrfs_free_path(path);
1307         return 0;
1308 }
1309
1310 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1311                          struct btrfs_root *root,
1312                          u64 bytenr, u64 num_bytes, u64 parent,
1313                          u64 ref_root, u64 ref_generation,
1314                          u64 owner_objectid)
1315 {
1316         int ret;
1317         if (ref_root == BTRFS_TREE_LOG_OBJECTID &&
1318             owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
1319                 return 0;
1320         ret = __btrfs_inc_extent_ref(trans, root, bytenr, 0, parent,
1321                                      0, ref_root, 0, ref_generation,
1322                                      owner_objectid);
1323         return ret;
1324 }
1325
1326 int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
1327                          struct btrfs_root *root)
1328 {
1329         finish_current_insert(trans, root->fs_info->extent_root, 1);
1330         del_pending_extents(trans, root->fs_info->extent_root, 1);
1331         return 0;
1332 }
1333
1334 int btrfs_lookup_extent_ref(struct btrfs_trans_handle *trans,
1335                             struct btrfs_root *root, u64 bytenr,
1336                             u64 num_bytes, u32 *refs)
1337 {
1338         struct btrfs_path *path;
1339         int ret;
1340         struct btrfs_key key;
1341         struct extent_buffer *l;
1342         struct btrfs_extent_item *item;
1343
1344         WARN_ON(num_bytes < root->sectorsize);
1345         path = btrfs_alloc_path();
1346         path->reada = 1;
1347         key.objectid = bytenr;
1348         key.offset = num_bytes;
1349         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
1350         ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
1351                                 0, 0);
1352         if (ret < 0)
1353                 goto out;
1354         if (ret != 0) {
1355                 btrfs_print_leaf(root, path->nodes[0]);
1356                 printk(KERN_INFO "btrfs failed to find block number %llu\n",
1357                        (unsigned long long)bytenr);
1358                 BUG();
1359         }
1360         l = path->nodes[0];
1361         item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
1362         *refs = btrfs_extent_refs(l, item);
1363 out:
1364         btrfs_free_path(path);
1365         return 0;
1366 }
1367
1368 int btrfs_cross_ref_exist(struct btrfs_trans_handle *trans,
1369                           struct btrfs_root *root, u64 objectid, u64 bytenr)
1370 {
1371         struct btrfs_root *extent_root = root->fs_info->extent_root;
1372         struct btrfs_path *path;
1373         struct extent_buffer *leaf;
1374         struct btrfs_extent_ref *ref_item;
1375         struct btrfs_key key;
1376         struct btrfs_key found_key;
1377         u64 ref_root;
1378         u64 last_snapshot;
1379         u32 nritems;
1380         int ret;
1381
1382         key.objectid = bytenr;
1383         key.offset = (u64)-1;
1384         key.type = BTRFS_EXTENT_ITEM_KEY;
1385
1386         path = btrfs_alloc_path();
1387         ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
1388         if (ret < 0)
1389                 goto out;
1390         BUG_ON(ret == 0);
1391
1392         ret = -ENOENT;
1393         if (path->slots[0] == 0)
1394                 goto out;
1395
1396         path->slots[0]--;
1397         leaf = path->nodes[0];
1398         btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1399
1400         if (found_key.objectid != bytenr ||
1401             found_key.type != BTRFS_EXTENT_ITEM_KEY)
1402                 goto out;
1403
1404         last_snapshot = btrfs_root_last_snapshot(&root->root_item);
1405         while (1) {
1406                 leaf = path->nodes[0];
1407                 nritems = btrfs_header_nritems(leaf);
1408                 if (path->slots[0] >= nritems) {
1409                         ret = btrfs_next_leaf(extent_root, path);
1410                         if (ret < 0)
1411                                 goto out;
1412                         if (ret == 0)
1413                                 continue;
1414                         break;
1415                 }
1416                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1417                 if (found_key.objectid != bytenr)
1418                         break;
1419
1420                 if (found_key.type != BTRFS_EXTENT_REF_KEY) {
1421                         path->slots[0]++;
1422                         continue;
1423                 }
1424
1425                 ref_item = btrfs_item_ptr(leaf, path->slots[0],
1426                                           struct btrfs_extent_ref);
1427                 ref_root = btrfs_ref_root(leaf, ref_item);
1428                 if ((ref_root != root->root_key.objectid &&
1429                      ref_root != BTRFS_TREE_LOG_OBJECTID) ||
1430                      objectid != btrfs_ref_objectid(leaf, ref_item)) {
1431                         ret = 1;
1432                         goto out;
1433                 }
1434                 if (btrfs_ref_generation(leaf, ref_item) <= last_snapshot) {
1435                         ret = 1;
1436                         goto out;
1437                 }
1438
1439                 path->slots[0]++;
1440         }
1441         ret = 0;
1442 out:
1443         btrfs_free_path(path);
1444         return ret;
1445 }
1446
1447 int btrfs_cache_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
1448                     struct extent_buffer *buf, u32 nr_extents)
1449 {
1450         struct btrfs_key key;
1451         struct btrfs_file_extent_item *fi;
1452         u64 root_gen;
1453         u32 nritems;
1454         int i;
1455         int level;
1456         int ret = 0;
1457         int shared = 0;
1458
1459         if (!root->ref_cows)
1460                 return 0;
1461
1462         if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
1463                 shared = 0;
1464                 root_gen = root->root_key.offset;
1465         } else {
1466                 shared = 1;
1467                 root_gen = trans->transid - 1;
1468         }
1469
1470         level = btrfs_header_level(buf);
1471         nritems = btrfs_header_nritems(buf);
1472
1473         if (level == 0) {
1474                 struct btrfs_leaf_ref *ref;
1475                 struct btrfs_extent_info *info;
1476
1477                 ref = btrfs_alloc_leaf_ref(root, nr_extents);
1478                 if (!ref) {
1479                         ret = -ENOMEM;
1480                         goto out;
1481                 }
1482
1483                 ref->root_gen = root_gen;
1484                 ref->bytenr = buf->start;
1485                 ref->owner = btrfs_header_owner(buf);
1486                 ref->generation = btrfs_header_generation(buf);
1487                 ref->nritems = nr_extents;
1488                 info = ref->extents;
1489
1490                 for (i = 0; nr_extents > 0 && i < nritems; i++) {
1491                         u64 disk_bytenr;
1492                         btrfs_item_key_to_cpu(buf, &key, i);
1493                         if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1494                                 continue;
1495                         fi = btrfs_item_ptr(buf, i,
1496                                             struct btrfs_file_extent_item);
1497                         if (btrfs_file_extent_type(buf, fi) ==
1498                             BTRFS_FILE_EXTENT_INLINE)
1499                                 continue;
1500                         disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1501                         if (disk_bytenr == 0)
1502                                 continue;
1503
1504                         info->bytenr = disk_bytenr;
1505                         info->num_bytes =
1506                                 btrfs_file_extent_disk_num_bytes(buf, fi);
1507                         info->objectid = key.objectid;
1508                         info->offset = key.offset;
1509                         info++;
1510                 }
1511
1512                 ret = btrfs_add_leaf_ref(root, ref, shared);
1513                 if (ret == -EEXIST && shared) {
1514                         struct btrfs_leaf_ref *old;
1515                         old = btrfs_lookup_leaf_ref(root, ref->bytenr);
1516                         BUG_ON(!old);
1517                         btrfs_remove_leaf_ref(root, old);
1518                         btrfs_free_leaf_ref(root, old);
1519                         ret = btrfs_add_leaf_ref(root, ref, shared);
1520                 }
1521                 WARN_ON(ret);
1522                 btrfs_free_leaf_ref(root, ref);
1523         }
1524 out:
1525         return ret;
1526 }
1527
1528 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
1529                   struct extent_buffer *orig_buf, struct extent_buffer *buf,
1530                   u32 *nr_extents)
1531 {
1532         u64 bytenr;
1533         u64 ref_root;
1534         u64 orig_root;
1535         u64 ref_generation;
1536         u64 orig_generation;
1537         u32 nritems;
1538         u32 nr_file_extents = 0;
1539         struct btrfs_key key;
1540         struct btrfs_file_extent_item *fi;
1541         int i;
1542         int level;
1543         int ret = 0;
1544         int faili = 0;
1545         int (*process_func)(struct btrfs_trans_handle *, struct btrfs_root *,
1546                             u64, u64, u64, u64, u64, u64, u64, u64);
1547
1548         ref_root = btrfs_header_owner(buf);
1549         ref_generation = btrfs_header_generation(buf);
1550         orig_root = btrfs_header_owner(orig_buf);
1551         orig_generation = btrfs_header_generation(orig_buf);
1552
1553         nritems = btrfs_header_nritems(buf);
1554         level = btrfs_header_level(buf);
1555
1556         if (root->ref_cows) {
1557                 process_func = __btrfs_inc_extent_ref;
1558         } else {
1559                 if (level == 0 &&
1560                     root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
1561                         goto out;
1562                 if (level != 0 &&
1563                     root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID)
1564                         goto out;
1565                 process_func = __btrfs_update_extent_ref;
1566         }
1567
1568         for (i = 0; i < nritems; i++) {
1569                 cond_resched();
1570                 if (level == 0) {
1571                         btrfs_item_key_to_cpu(buf, &key, i);
1572                         if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1573                                 continue;
1574                         fi = btrfs_item_ptr(buf, i,
1575                                             struct btrfs_file_extent_item);
1576                         if (btrfs_file_extent_type(buf, fi) ==
1577                             BTRFS_FILE_EXTENT_INLINE)
1578                                 continue;
1579                         bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1580                         if (bytenr == 0)
1581                                 continue;
1582
1583                         nr_file_extents++;
1584
1585                         ret = process_func(trans, root, bytenr,
1586                                            orig_buf->start, buf->start,
1587                                            orig_root, ref_root,
1588                                            orig_generation, ref_generation,
1589                                            key.objectid);
1590
1591                         if (ret) {
1592                                 faili = i;
1593                                 WARN_ON(1);
1594                                 goto fail;
1595                         }
1596                 } else {
1597                         bytenr = btrfs_node_blockptr(buf, i);
1598                         ret = process_func(trans, root, bytenr,
1599                                            orig_buf->start, buf->start,
1600                                            orig_root, ref_root,
1601                                            orig_generation, ref_generation,
1602                                            level - 1);
1603                         if (ret) {
1604                                 faili = i;
1605                                 WARN_ON(1);
1606                                 goto fail;
1607                         }
1608                 }
1609         }
1610 out:
1611         if (nr_extents) {
1612                 if (level == 0)
1613                         *nr_extents = nr_file_extents;
1614                 else
1615                         *nr_extents = nritems;
1616         }
1617         return 0;
1618 fail:
1619         WARN_ON(1);
1620         return ret;
1621 }
1622
1623 int btrfs_update_ref(struct btrfs_trans_handle *trans,
1624                      struct btrfs_root *root, struct extent_buffer *orig_buf,
1625                      struct extent_buffer *buf, int start_slot, int nr)
1626
1627 {
1628         u64 bytenr;
1629         u64 ref_root;
1630         u64 orig_root;
1631         u64 ref_generation;
1632         u64 orig_generation;
1633         struct btrfs_key key;
1634         struct btrfs_file_extent_item *fi;
1635         int i;
1636         int ret;
1637         int slot;
1638         int level;
1639
1640         BUG_ON(start_slot < 0);
1641         BUG_ON(start_slot + nr > btrfs_header_nritems(buf));
1642
1643         ref_root = btrfs_header_owner(buf);
1644         ref_generation = btrfs_header_generation(buf);
1645         orig_root = btrfs_header_owner(orig_buf);
1646         orig_generation = btrfs_header_generation(orig_buf);
1647         level = btrfs_header_level(buf);
1648
1649         if (!root->ref_cows) {
1650                 if (level == 0 &&
1651                     root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
1652                         return 0;
1653                 if (level != 0 &&
1654                     root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID)
1655                         return 0;
1656         }
1657
1658         for (i = 0, slot = start_slot; i < nr; i++, slot++) {
1659                 cond_resched();
1660                 if (level == 0) {
1661                         btrfs_item_key_to_cpu(buf, &key, slot);
1662                         if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1663                                 continue;
1664                         fi = btrfs_item_ptr(buf, slot,
1665                                             struct btrfs_file_extent_item);
1666                         if (btrfs_file_extent_type(buf, fi) ==
1667                             BTRFS_FILE_EXTENT_INLINE)
1668                                 continue;
1669                         bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1670                         if (bytenr == 0)
1671                                 continue;
1672                         ret = __btrfs_update_extent_ref(trans, root, bytenr,
1673                                             orig_buf->start, buf->start,
1674                                             orig_root, ref_root,
1675                                             orig_generation, ref_generation,
1676                                             key.objectid);
1677                         if (ret)
1678                                 goto fail;
1679                 } else {
1680                         bytenr = btrfs_node_blockptr(buf, slot);
1681                         ret = __btrfs_update_extent_ref(trans, root, bytenr,
1682                                             orig_buf->start, buf->start,
1683                                             orig_root, ref_root,
1684                                             orig_generation, ref_generation,
1685                                             level - 1);
1686                         if (ret)
1687                                 goto fail;
1688                 }
1689         }
1690         return 0;
1691 fail:
1692         WARN_ON(1);
1693         return -1;
1694 }
1695
1696 static int write_one_cache_group(struct btrfs_trans_handle *trans,
1697                                  struct btrfs_root *root,
1698                                  struct btrfs_path *path,
1699                                  struct btrfs_block_group_cache *cache)
1700 {
1701         int ret;
1702         int pending_ret;
1703         struct btrfs_root *extent_root = root->fs_info->extent_root;
1704         unsigned long bi;
1705         struct extent_buffer *leaf;
1706
1707         ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
1708         if (ret < 0)
1709                 goto fail;
1710         BUG_ON(ret);
1711
1712         leaf = path->nodes[0];
1713         bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
1714         write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
1715         btrfs_mark_buffer_dirty(leaf);
1716         btrfs_release_path(extent_root, path);
1717 fail:
1718         finish_current_insert(trans, extent_root, 0);
1719         pending_ret = del_pending_extents(trans, extent_root, 0);
1720         if (ret)
1721                 return ret;
1722         if (pending_ret)
1723                 return pending_ret;
1724         return 0;
1725
1726 }
1727
1728 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
1729                                    struct btrfs_root *root)
1730 {
1731         struct btrfs_block_group_cache *cache, *entry;
1732         struct rb_node *n;
1733         int err = 0;
1734         int werr = 0;
1735         struct btrfs_path *path;
1736         u64 last = 0;
1737
1738         path = btrfs_alloc_path();
1739         if (!path)
1740                 return -ENOMEM;
1741
1742         while (1) {
1743                 cache = NULL;
1744                 spin_lock(&root->fs_info->block_group_cache_lock);
1745                 for (n = rb_first(&root->fs_info->block_group_cache_tree);
1746                      n; n = rb_next(n)) {
1747                         entry = rb_entry(n, struct btrfs_block_group_cache,
1748                                          cache_node);
1749                         if (entry->dirty) {
1750                                 cache = entry;
1751                                 break;
1752                         }
1753                 }
1754                 spin_unlock(&root->fs_info->block_group_cache_lock);
1755
1756                 if (!cache)
1757                         break;
1758
1759                 cache->dirty = 0;
1760                 last += cache->key.offset;
1761
1762                 err = write_one_cache_group(trans, root,
1763                                             path, cache);
1764                 /*
1765                  * if we fail to write the cache group, we want
1766                  * to keep it marked dirty in hopes that a later
1767                  * write will work
1768                  */
1769                 if (err) {
1770                         werr = err;
1771                         continue;
1772                 }
1773         }
1774         btrfs_free_path(path);
1775         return werr;
1776 }
1777
1778 int btrfs_extent_readonly(struct btrfs_root *root, u64 bytenr)
1779 {
1780         struct btrfs_block_group_cache *block_group;
1781         int readonly = 0;
1782
1783         block_group = btrfs_lookup_block_group(root->fs_info, bytenr);
1784         if (!block_group || block_group->ro)
1785                 readonly = 1;
1786         if (block_group)
1787                 put_block_group(block_group);
1788         return readonly;
1789 }
1790
1791 static int update_space_info(struct btrfs_fs_info *info, u64 flags,
1792                              u64 total_bytes, u64 bytes_used,
1793                              struct btrfs_space_info **space_info)
1794 {
1795         struct btrfs_space_info *found;
1796
1797         found = __find_space_info(info, flags);
1798         if (found) {
1799                 spin_lock(&found->lock);
1800                 found->total_bytes += total_bytes;
1801                 found->bytes_used += bytes_used;
1802                 found->full = 0;
1803                 spin_unlock(&found->lock);
1804                 *space_info = found;
1805                 return 0;
1806         }
1807         found = kzalloc(sizeof(*found), GFP_NOFS);
1808         if (!found)
1809                 return -ENOMEM;
1810
1811         list_add(&found->list, &info->space_info);
1812         INIT_LIST_HEAD(&found->block_groups);
1813         init_rwsem(&found->groups_sem);
1814         spin_lock_init(&found->lock);
1815         found->flags = flags;
1816         found->total_bytes = total_bytes;
1817         found->bytes_used = bytes_used;
1818         found->bytes_pinned = 0;
1819         found->bytes_reserved = 0;
1820         found->bytes_readonly = 0;
1821         found->full = 0;
1822         found->force_alloc = 0;
1823         *space_info = found;
1824         return 0;
1825 }
1826
1827 static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
1828 {
1829         u64 extra_flags = flags & (BTRFS_BLOCK_GROUP_RAID0 |
1830                                    BTRFS_BLOCK_GROUP_RAID1 |
1831                                    BTRFS_BLOCK_GROUP_RAID10 |
1832                                    BTRFS_BLOCK_GROUP_DUP);
1833         if (extra_flags) {
1834                 if (flags & BTRFS_BLOCK_GROUP_DATA)
1835                         fs_info->avail_data_alloc_bits |= extra_flags;
1836                 if (flags & BTRFS_BLOCK_GROUP_METADATA)
1837                         fs_info->avail_metadata_alloc_bits |= extra_flags;
1838                 if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
1839                         fs_info->avail_system_alloc_bits |= extra_flags;
1840         }
1841 }
1842
1843 static void set_block_group_readonly(struct btrfs_block_group_cache *cache)
1844 {
1845         spin_lock(&cache->space_info->lock);
1846         spin_lock(&cache->lock);
1847         if (!cache->ro) {
1848                 cache->space_info->bytes_readonly += cache->key.offset -
1849                                         btrfs_block_group_used(&cache->item);
1850                 cache->ro = 1;
1851         }
1852         spin_unlock(&cache->lock);
1853         spin_unlock(&cache->space_info->lock);
1854 }
1855
1856 u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags)
1857 {
1858         u64 num_devices = root->fs_info->fs_devices->rw_devices;
1859
1860         if (num_devices == 1)
1861                 flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0);
1862         if (num_devices < 4)
1863                 flags &= ~BTRFS_BLOCK_GROUP_RAID10;
1864
1865         if ((flags & BTRFS_BLOCK_GROUP_DUP) &&
1866             (flags & (BTRFS_BLOCK_GROUP_RAID1 |
1867                       BTRFS_BLOCK_GROUP_RAID10))) {
1868                 flags &= ~BTRFS_BLOCK_GROUP_DUP;
1869         }
1870
1871         if ((flags & BTRFS_BLOCK_GROUP_RAID1) &&
1872             (flags & BTRFS_BLOCK_GROUP_RAID10)) {
1873                 flags &= ~BTRFS_BLOCK_GROUP_RAID1;
1874         }
1875
1876         if ((flags & BTRFS_BLOCK_GROUP_RAID0) &&
1877             ((flags & BTRFS_BLOCK_GROUP_RAID1) |
1878              (flags & BTRFS_BLOCK_GROUP_RAID10) |
1879              (flags & BTRFS_BLOCK_GROUP_DUP)))
1880                 flags &= ~BTRFS_BLOCK_GROUP_RAID0;
1881         return flags;
1882 }
1883
1884 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
1885                           struct btrfs_root *extent_root, u64 alloc_bytes,
1886                           u64 flags, int force)
1887 {
1888         struct btrfs_space_info *space_info;
1889         u64 thresh;
1890         int ret = 0;
1891
1892         mutex_lock(&extent_root->fs_info->chunk_mutex);
1893
1894         flags = btrfs_reduce_alloc_profile(extent_root, flags);
1895
1896         space_info = __find_space_info(extent_root->fs_info, flags);
1897         if (!space_info) {
1898                 ret = update_space_info(extent_root->fs_info, flags,
1899                                         0, 0, &space_info);
1900                 BUG_ON(ret);
1901         }
1902         BUG_ON(!space_info);
1903
1904         spin_lock(&space_info->lock);
1905         if (space_info->force_alloc) {
1906                 force = 1;
1907                 space_info->force_alloc = 0;
1908         }
1909         if (space_info->full) {
1910                 spin_unlock(&space_info->lock);
1911                 goto out;
1912         }
1913
1914         thresh = space_info->total_bytes - space_info->bytes_readonly;
1915         thresh = div_factor(thresh, 6);
1916         if (!force &&
1917            (space_info->bytes_used + space_info->bytes_pinned +
1918             space_info->bytes_reserved + alloc_bytes) < thresh) {
1919                 spin_unlock(&space_info->lock);
1920                 goto out;
1921         }
1922         spin_unlock(&space_info->lock);
1923
1924         ret = btrfs_alloc_chunk(trans, extent_root, flags);
1925         if (ret)
1926                 space_info->full = 1;
1927 out:
1928         mutex_unlock(&extent_root->fs_info->chunk_mutex);
1929         return ret;
1930 }
1931
1932 static int update_block_group(struct btrfs_trans_handle *trans,
1933                               struct btrfs_root *root,
1934                               u64 bytenr, u64 num_bytes, int alloc,
1935                               int mark_free)
1936 {
1937         struct btrfs_block_group_cache *cache;
1938         struct btrfs_fs_info *info = root->fs_info;
1939         u64 total = num_bytes;
1940         u64 old_val;
1941         u64 byte_in_group;
1942
1943         while (total) {
1944                 cache = btrfs_lookup_block_group(info, bytenr);
1945                 if (!cache)
1946                         return -1;
1947                 byte_in_group = bytenr - cache->key.objectid;
1948                 WARN_ON(byte_in_group > cache->key.offset);
1949
1950                 spin_lock(&cache->space_info->lock);
1951                 spin_lock(&cache->lock);
1952                 cache->dirty = 1;
1953                 old_val = btrfs_block_group_used(&cache->item);
1954                 num_bytes = min(total, cache->key.offset - byte_in_group);
1955                 if (alloc) {
1956                         old_val += num_bytes;
1957                         cache->space_info->bytes_used += num_bytes;
1958                         if (cache->ro)
1959                                 cache->space_info->bytes_readonly -= num_bytes;
1960                         btrfs_set_block_group_used(&cache->item, old_val);
1961                         spin_unlock(&cache->lock);
1962                         spin_unlock(&cache->space_info->lock);
1963                 } else {
1964                         old_val -= num_bytes;
1965                         cache->space_info->bytes_used -= num_bytes;
1966                         if (cache->ro)
1967                                 cache->space_info->bytes_readonly += num_bytes;
1968                         btrfs_set_block_group_used(&cache->item, old_val);
1969                         spin_unlock(&cache->lock);
1970                         spin_unlock(&cache->space_info->lock);
1971                         if (mark_free) {
1972                                 int ret;
1973
1974                                 ret = btrfs_discard_extent(root, bytenr,
1975                                                            num_bytes);
1976                                 WARN_ON(ret);
1977
1978                                 ret = btrfs_add_free_space(cache, bytenr,
1979                                                            num_bytes);
1980                                 WARN_ON(ret);
1981                         }
1982                 }
1983                 put_block_group(cache);
1984                 total -= num_bytes;
1985                 bytenr += num_bytes;
1986         }
1987         return 0;
1988 }
1989
1990 static u64 first_logical_byte(struct btrfs_root *root, u64 search_start)
1991 {
1992         struct btrfs_block_group_cache *cache;
1993         u64 bytenr;
1994
1995         cache = btrfs_lookup_first_block_group(root->fs_info, search_start);
1996         if (!cache)
1997                 return 0;
1998
1999         bytenr = cache->key.objectid;
2000         put_block_group(cache);
2001
2002         return bytenr;
2003 }
2004
2005 int btrfs_update_pinned_extents(struct btrfs_root *root,
2006                                 u64 bytenr, u64 num, int pin)
2007 {
2008         u64 len;
2009         struct btrfs_block_group_cache *cache;
2010         struct btrfs_fs_info *fs_info = root->fs_info;
2011
2012         WARN_ON(!mutex_is_locked(&root->fs_info->pinned_mutex));
2013         if (pin) {
2014                 set_extent_dirty(&fs_info->pinned_extents,
2015                                 bytenr, bytenr + num - 1, GFP_NOFS);
2016         } else {
2017                 clear_extent_dirty(&fs_info->pinned_extents,
2018                                 bytenr, bytenr + num - 1, GFP_NOFS);
2019         }
2020         while (num > 0) {
2021                 cache = btrfs_lookup_block_group(fs_info, bytenr);
2022                 BUG_ON(!cache);
2023                 len = min(num, cache->key.offset -
2024                           (bytenr - cache->key.objectid));
2025                 if (pin) {
2026                         spin_lock(&cache->space_info->lock);
2027                         spin_lock(&cache->lock);
2028                         cache->pinned += len;
2029                         cache->space_info->bytes_pinned += len;
2030                         spin_unlock(&cache->lock);
2031                         spin_unlock(&cache->space_info->lock);
2032                         fs_info->total_pinned += len;
2033                 } else {
2034                         spin_lock(&cache->space_info->lock);
2035                         spin_lock(&cache->lock);
2036                         cache->pinned -= len;
2037                         cache->space_info->bytes_pinned -= len;
2038                         spin_unlock(&cache->lock);
2039                         spin_unlock(&cache->space_info->lock);
2040                         fs_info->total_pinned -= len;
2041                         if (cache->cached)
2042                                 btrfs_add_free_space(cache, bytenr, len);
2043                 }
2044                 put_block_group(cache);
2045                 bytenr += len;
2046                 num -= len;
2047         }
2048         return 0;
2049 }
2050
2051 static int update_reserved_extents(struct btrfs_root *root,
2052                                    u64 bytenr, u64 num, int reserve)
2053 {
2054         u64 len;
2055         struct btrfs_block_group_cache *cache;
2056         struct btrfs_fs_info *fs_info = root->fs_info;
2057
2058         while (num > 0) {
2059                 cache = btrfs_lookup_block_group(fs_info, bytenr);
2060                 BUG_ON(!cache);
2061                 len = min(num, cache->key.offset -
2062                           (bytenr - cache->key.objectid));
2063
2064                 spin_lock(&cache->space_info->lock);
2065                 spin_lock(&cache->lock);
2066                 if (reserve) {
2067                         cache->reserved += len;
2068                         cache->space_info->bytes_reserved += len;
2069                 } else {
2070                         cache->reserved -= len;
2071                         cache->space_info->bytes_reserved -= len;
2072                 }
2073                 spin_unlock(&cache->lock);
2074                 spin_unlock(&cache->space_info->lock);
2075                 put_block_group(cache);
2076                 bytenr += len;
2077                 num -= len;
2078         }
2079         return 0;
2080 }
2081
2082 int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy)
2083 {
2084         u64 last = 0;
2085         u64 start;
2086         u64 end;
2087         struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents;
2088         int ret;
2089
2090         mutex_lock(&root->fs_info->pinned_mutex);
2091         while (1) {
2092                 ret = find_first_extent_bit(pinned_extents, last,
2093                                             &start, &end, EXTENT_DIRTY);
2094                 if (ret)
2095                         break;
2096                 set_extent_dirty(copy, start, end, GFP_NOFS);
2097                 last = end + 1;
2098         }
2099         mutex_unlock(&root->fs_info->pinned_mutex);
2100         return 0;
2101 }
2102
2103 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
2104                                struct btrfs_root *root,
2105                                struct extent_io_tree *unpin)
2106 {
2107         u64 start;
2108         u64 end;
2109         int ret;
2110
2111         mutex_lock(&root->fs_info->pinned_mutex);
2112         while (1) {
2113                 ret = find_first_extent_bit(unpin, 0, &start, &end,
2114                                             EXTENT_DIRTY);
2115                 if (ret)
2116                         break;
2117
2118                 ret = btrfs_discard_extent(root, start, end + 1 - start);
2119
2120                 btrfs_update_pinned_extents(root, start, end + 1 - start, 0);
2121                 clear_extent_dirty(unpin, start, end, GFP_NOFS);
2122
2123                 if (need_resched()) {
2124                         mutex_unlock(&root->fs_info->pinned_mutex);
2125                         cond_resched();
2126                         mutex_lock(&root->fs_info->pinned_mutex);
2127                 }
2128         }
2129         mutex_unlock(&root->fs_info->pinned_mutex);
2130         return ret;
2131 }
2132
2133 static int finish_current_insert(struct btrfs_trans_handle *trans,
2134                                  struct btrfs_root *extent_root, int all)
2135 {
2136         u64 start;
2137         u64 end;
2138         u64 priv;
2139         u64 search = 0;
2140         u64 skipped = 0;
2141         struct btrfs_fs_info *info = extent_root->fs_info;
2142         struct btrfs_path *path;
2143         struct pending_extent_op *extent_op, *tmp;
2144         struct list_head insert_list, update_list;
2145         int ret;
2146         int num_inserts = 0, max_inserts;
2147
2148         path = btrfs_alloc_path();
2149         INIT_LIST_HEAD(&insert_list);
2150         INIT_LIST_HEAD(&update_list);
2151
2152         max_inserts = extent_root->leafsize /
2153                 (2 * sizeof(struct btrfs_key) + 2 * sizeof(struct btrfs_item) +
2154                  sizeof(struct btrfs_extent_ref) +
2155                  sizeof(struct btrfs_extent_item));
2156 again:
2157         mutex_lock(&info->extent_ins_mutex);
2158         while (1) {
2159                 ret = find_first_extent_bit(&info->extent_ins, search, &start,
2160                                             &end, EXTENT_WRITEBACK);
2161                 if (ret) {
2162                         if (skipped && all && !num_inserts) {
2163                                 skipped = 0;
2164                                 search = 0;
2165                                 continue;
2166                         }
2167                         mutex_unlock(&info->extent_ins_mutex);
2168                         break;
2169                 }
2170
2171                 ret = try_lock_extent(&info->extent_ins, start, end, GFP_NOFS);
2172                 if (!ret) {
2173                         skipped = 1;
2174                         search = end + 1;
2175                         if (need_resched()) {
2176                                 mutex_unlock(&info->extent_ins_mutex);
2177                                 cond_resched();
2178                                 mutex_lock(&info->extent_ins_mutex);
2179                         }
2180                         continue;
2181                 }
2182
2183                 ret = get_state_private(&info->extent_ins, start, &priv);
2184                 BUG_ON(ret);
2185                 extent_op = (struct pending_extent_op *)(unsigned long) priv;
2186
2187                 if (extent_op->type == PENDING_EXTENT_INSERT) {
2188                         num_inserts++;
2189                         list_add_tail(&extent_op->list, &insert_list);
2190                         search = end + 1;
2191                         if (num_inserts == max_inserts) {
2192                                 mutex_unlock(&info->extent_ins_mutex);
2193                                 break;
2194                         }
2195                 } else if (extent_op->type == PENDING_BACKREF_UPDATE) {
2196                         list_add_tail(&extent_op->list, &update_list);
2197                         search = end + 1;
2198                 } else {
2199                         BUG();
2200                 }
2201         }
2202
2203         /*
2204          * process the update list, clear the writeback bit for it, and if
2205          * somebody marked this thing for deletion then just unlock it and be
2206          * done, the free_extents will handle it
2207          */
2208         mutex_lock(&info->extent_ins_mutex);
2209         list_for_each_entry_safe(extent_op, tmp, &update_list, list) {
2210                 clear_extent_bits(&info->extent_ins, extent_op->bytenr,
2211                                   extent_op->bytenr + extent_op->num_bytes - 1,
2212                                   EXTENT_WRITEBACK, GFP_NOFS);
2213                 if (extent_op->del) {
2214                         list_del_init(&extent_op->list);
2215                         unlock_extent(&info->extent_ins, extent_op->bytenr,
2216                                       extent_op->bytenr + extent_op->num_bytes
2217                                       - 1, GFP_NOFS);
2218                         kfree(extent_op);
2219                 }
2220         }
2221         mutex_unlock(&info->extent_ins_mutex);
2222
2223         /*
2224          * still have things left on the update list, go ahead an update
2225          * everything
2226          */
2227         if (!list_empty(&update_list)) {
2228                 ret = update_backrefs(trans, extent_root, path, &update_list);
2229                 BUG_ON(ret);
2230         }
2231
2232         /*
2233          * if no inserts need to be done, but we skipped some extents and we
2234          * need to make sure everything is cleaned then reset everything and
2235          * go back to the beginning
2236          */
2237         if (!num_inserts && all && skipped) {
2238                 search = 0;
2239                 skipped = 0;
2240                 INIT_LIST_HEAD(&update_list);
2241                 INIT_LIST_HEAD(&insert_list);
2242                 goto again;
2243         } else if (!num_inserts) {
2244                 goto out;
2245         }
2246
2247         /*
2248          * process the insert extents list.  Again if we are deleting this
2249          * extent, then just unlock it, pin down the bytes if need be, and be
2250          * done with it.  Saves us from having to actually insert the extent
2251          * into the tree and then subsequently come along and delete it
2252          */
2253         mutex_lock(&info->extent_ins_mutex);
2254         list_for_each_entry_safe(extent_op, tmp, &insert_list, list) {
2255                 clear_extent_bits(&info->extent_ins, extent_op->bytenr,
2256                                   extent_op->bytenr + extent_op->num_bytes - 1,
2257                                   EXTENT_WRITEBACK, GFP_NOFS);
2258                 if (extent_op->del) {
2259                         u64 used;
2260                         list_del_init(&extent_op->list);
2261                         unlock_extent(&info->extent_ins, extent_op->bytenr,
2262                                       extent_op->bytenr + extent_op->num_bytes
2263                                       - 1, GFP_NOFS);
2264
2265                         mutex_lock(&extent_root->fs_info->pinned_mutex);
2266                         ret = pin_down_bytes(trans, extent_root,
2267                                              extent_op->bytenr,
2268                                              extent_op->num_bytes, 0);
2269                         mutex_unlock(&extent_root->fs_info->pinned_mutex);
2270
2271                         spin_lock(&info->delalloc_lock);
2272                         used = btrfs_super_bytes_used(&info->super_copy);
2273                         btrfs_set_super_bytes_used(&info->super_copy,
2274                                         used - extent_op->num_bytes);
2275                         used = btrfs_root_used(&extent_root->root_item);
2276                         btrfs_set_root_used(&extent_root->root_item,
2277                                         used - extent_op->num_bytes);
2278                         spin_unlock(&info->delalloc_lock);
2279
2280                         ret = update_block_group(trans, extent_root,
2281                                                  extent_op->bytenr,
2282                                                  extent_op->num_bytes,
2283                                                  0, ret > 0);
2284                         BUG_ON(ret);
2285                         kfree(extent_op);
2286                         num_inserts--;
2287                 }
2288         }
2289         mutex_unlock(&info->extent_ins_mutex);
2290
2291         ret = insert_extents(trans, extent_root, path, &insert_list,
2292                              num_inserts);
2293         BUG_ON(ret);
2294
2295         /*
2296          * if we broke out of the loop in order to insert stuff because we hit
2297          * the maximum number of inserts at a time we can handle, then loop
2298          * back and pick up where we left off
2299          */
2300         if (num_inserts == max_inserts) {
2301                 INIT_LIST_HEAD(&insert_list);
2302                 INIT_LIST_HEAD(&update_list);
2303                 num_inserts = 0;
2304                 goto again;
2305         }
2306
2307         /*
2308          * again, if we need to make absolutely sure there are no more pending
2309          * extent operations left and we know that we skipped some, go back to
2310          * the beginning and do it all again
2311          */
2312         if (all && skipped) {
2313                 INIT_LIST_HEAD(&insert_list);
2314                 INIT_LIST_HEAD(&update_list);
2315                 search = 0;
2316                 skipped = 0;
2317                 num_inserts = 0;
2318                 goto again;
2319         }
2320 out:
2321         btrfs_free_path(path);
2322         return 0;
2323 }
2324
2325 static int pin_down_bytes(struct btrfs_trans_handle *trans,
2326                           struct btrfs_root *root,
2327                           u64 bytenr, u64 num_bytes, int is_data)
2328 {
2329         int err = 0;
2330         struct extent_buffer *buf;
2331
2332         if (is_data)
2333                 goto pinit;
2334
2335         buf = btrfs_find_tree_block(root, bytenr, num_bytes);
2336         if (!buf)
2337                 goto pinit;
2338
2339         /* we can reuse a block if it hasn't been written
2340          * and it is from this transaction.  We can't
2341          * reuse anything from the tree log root because
2342          * it has tiny sub-transactions.
2343          */
2344         if (btrfs_buffer_uptodate(buf, 0) &&
2345             btrfs_try_tree_lock(buf)) {
2346                 u64 header_owner = btrfs_header_owner(buf);
2347                 u64 header_transid = btrfs_header_generation(buf);
2348                 if (header_owner != BTRFS_TREE_LOG_OBJECTID &&
2349                     header_owner != BTRFS_TREE_RELOC_OBJECTID &&
2350                     header_transid == trans->transid &&
2351                     !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
2352                         clean_tree_block(NULL, root, buf);
2353                         btrfs_tree_unlock(buf);
2354                         free_extent_buffer(buf);
2355                         return 1;
2356                 }
2357                 btrfs_tree_unlock(buf);
2358         }
2359         free_extent_buffer(buf);
2360 pinit:
2361         btrfs_update_pinned_extents(root, bytenr, num_bytes, 1);
2362
2363         BUG_ON(err < 0);
2364         return 0;
2365 }
2366
2367 /*
2368  * remove an extent from the root, returns 0 on success
2369  */
2370 static int __free_extent(struct btrfs_trans_handle *trans,
2371                          struct btrfs_root *root,
2372                          u64 bytenr, u64 num_bytes, u64 parent,
2373                          u64 root_objectid, u64 ref_generation,
2374                          u64 owner_objectid, int pin, int mark_free)
2375 {
2376         struct btrfs_path *path;
2377         struct btrfs_key key;
2378         struct btrfs_fs_info *info = root->fs_info;
2379         struct btrfs_root *extent_root = info->extent_root;
2380         struct extent_buffer *leaf;
2381         int ret;
2382         int extent_slot = 0;
2383         int found_extent = 0;
2384         int num_to_del = 1;
2385         struct btrfs_extent_item *ei;
2386         u32 refs;
2387
2388         key.objectid = bytenr;
2389         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
2390         key.offset = num_bytes;
2391         path = btrfs_alloc_path();
2392         if (!path)
2393                 return -ENOMEM;
2394
2395         path->reada = 1;
2396         ret = lookup_extent_backref(trans, extent_root, path,
2397                                     bytenr, parent, root_objectid,
2398                                     ref_generation, owner_objectid, 1);
2399         if (ret == 0) {
2400                 struct btrfs_key found_key;
2401                 extent_slot = path->slots[0];
2402                 while (extent_slot > 0) {
2403                         extent_slot--;
2404                         btrfs_item_key_to_cpu(path->nodes[0], &found_key,
2405                                               extent_slot);
2406                         if (found_key.objectid != bytenr)
2407                                 break;
2408                         if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
2409                             found_key.offset == num_bytes) {
2410                                 found_extent = 1;
2411                                 break;
2412                         }
2413                         if (path->slots[0] - extent_slot > 5)
2414                                 break;
2415                 }
2416                 if (!found_extent) {
2417                         ret = remove_extent_backref(trans, extent_root, path);
2418                         BUG_ON(ret);
2419                         btrfs_release_path(extent_root, path);
2420                         ret = btrfs_search_slot(trans, extent_root,
2421                                                 &key, path, -1, 1);
2422                         if (ret) {
2423                                 printk(KERN_ERR "umm, got %d back from search"
2424                                        ", was looking for %llu\n", ret,
2425                                        (unsigned long long)bytenr);
2426                                 btrfs_print_leaf(extent_root, path->nodes[0]);
2427                         }
2428                         BUG_ON(ret);
2429                         extent_slot = path->slots[0];
2430                 }
2431         } else {
2432                 btrfs_print_leaf(extent_root, path->nodes[0]);
2433                 WARN_ON(1);
2434                 printk(KERN_ERR "btrfs unable to find ref byte nr %llu "
2435                        "root %llu gen %llu owner %llu\n",
2436                        (unsigned long long)bytenr,
2437                        (unsigned long long)root_objectid,
2438                        (unsigned long long)ref_generation,
2439                        (unsigned long long)owner_objectid);
2440         }
2441
2442         leaf = path->nodes[0];
2443         ei = btrfs_item_ptr(leaf, extent_slot,
2444                             struct btrfs_extent_item);
2445         refs = btrfs_extent_refs(leaf, ei);
2446         BUG_ON(refs == 0);
2447         refs -= 1;
2448         btrfs_set_extent_refs(leaf, ei, refs);
2449
2450         btrfs_mark_buffer_dirty(leaf);
2451
2452         if (refs == 0 && found_extent && path->slots[0] == extent_slot + 1) {
2453                 struct btrfs_extent_ref *ref;
2454                 ref = btrfs_item_ptr(leaf, path->slots[0],
2455                                      struct btrfs_extent_ref);
2456                 BUG_ON(btrfs_ref_num_refs(leaf, ref) != 1);
2457                 /* if the back ref and the extent are next to each other
2458                  * they get deleted below in one shot
2459                  */
2460                 path->slots[0] = extent_slot;
2461                 num_to_del = 2;
2462         } else if (found_extent) {
2463                 /* otherwise delete the extent back ref */
2464                 ret = remove_extent_backref(trans, extent_root, path);
2465                 BUG_ON(ret);
2466                 /* if refs are 0, we need to setup the path for deletion */
2467                 if (refs == 0) {
2468                         btrfs_release_path(extent_root, path);
2469                         ret = btrfs_search_slot(trans, extent_root, &key, path,
2470                                                 -1, 1);
2471                         BUG_ON(ret);
2472                 }
2473         }
2474
2475         if (refs == 0) {
2476                 u64 super_used;
2477                 u64 root_used;
2478
2479                 if (pin) {
2480                         mutex_lock(&root->fs_info->pinned_mutex);
2481                         ret = pin_down_bytes(trans, root, bytenr, num_bytes,
2482                                 owner_objectid >= BTRFS_FIRST_FREE_OBJECTID);
2483                         mutex_unlock(&root->fs_info->pinned_mutex);
2484                         if (ret > 0)
2485                                 mark_free = 1;
2486                         BUG_ON(ret < 0);
2487                 }
2488                 /* block accounting for super block */
2489                 spin_lock(&info->delalloc_lock);
2490                 super_used = btrfs_super_bytes_used(&info->super_copy);
2491                 btrfs_set_super_bytes_used(&info->super_copy,
2492                                            super_used - num_bytes);
2493
2494                 /* block accounting for root item */
2495                 root_used = btrfs_root_used(&root->root_item);
2496                 btrfs_set_root_used(&root->root_item,
2497                                            root_used - num_bytes);
2498                 spin_unlock(&info->delalloc_lock);
2499                 ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
2500                                       num_to_del);
2501                 BUG_ON(ret);
2502                 btrfs_release_path(extent_root, path);
2503
2504                 if (owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
2505                         ret = btrfs_del_csums(trans, root, bytenr, num_bytes);
2506                         BUG_ON(ret);
2507                 }
2508
2509                 ret = update_block_group(trans, root, bytenr, num_bytes, 0,
2510                                          mark_free);
2511                 BUG_ON(ret);
2512         }
2513         btrfs_free_path(path);
2514         finish_current_insert(trans, extent_root, 0);
2515         return ret;
2516 }
2517
2518 /*
2519  * find all the blocks marked as pending in the radix tree and remove
2520  * them from the extent map
2521  */
2522 static int del_pending_extents(struct btrfs_trans_handle *trans,
2523                                struct btrfs_root *extent_root, int all)
2524 {
2525         int ret;
2526         int err = 0;
2527         u64 start;
2528         u64 end;
2529         u64 priv;
2530         u64 search = 0;
2531         int nr = 0, skipped = 0;
2532         struct extent_io_tree *pending_del;
2533         struct extent_io_tree *extent_ins;
2534         struct pending_extent_op *extent_op;
2535         struct btrfs_fs_info *info = extent_root->fs_info;
2536         struct list_head delete_list;
2537
2538         INIT_LIST_HEAD(&delete_list);
2539         extent_ins = &extent_root->fs_info->extent_ins;
2540         pending_del = &extent_root->fs_info->pending_del;
2541
2542 again:
2543         mutex_lock(&info->extent_ins_mutex);
2544         while (1) {
2545                 ret = find_first_extent_bit(pending_del, search, &start, &end,
2546                                             EXTENT_WRITEBACK);
2547                 if (ret) {
2548                         if (all && skipped && !nr) {
2549                                 search = 0;
2550                                 continue;
2551                         }
2552                         mutex_unlock(&info->extent_ins_mutex);
2553                         break;
2554                 }
2555
2556                 ret = try_lock_extent(extent_ins, start, end, GFP_NOFS);
2557                 if (!ret) {
2558                         search = end+1;
2559                         skipped = 1;
2560
2561                         if (need_resched()) {
2562                                 mutex_unlock(&info->extent_ins_mutex);
2563                                 cond_resched();
2564                                 mutex_lock(&info->extent_ins_mutex);
2565                         }
2566
2567                         continue;
2568                 }
2569                 BUG_ON(ret < 0);
2570
2571                 ret = get_state_private(pending_del, start, &priv);
2572                 BUG_ON(ret);
2573                 extent_op = (struct pending_extent_op *)(unsigned long)priv;
2574
2575                 clear_extent_bits(pending_del, start, end, EXTENT_WRITEBACK,
2576                                   GFP_NOFS);
2577                 if (!test_range_bit(extent_ins, start, end,
2578                                     EXTENT_WRITEBACK, 0)) {
2579                         list_add_tail(&extent_op->list, &delete_list);
2580                         nr++;
2581                 } else {
2582                         kfree(extent_op);
2583
2584                         ret = get_state_private(&info->extent_ins, start,
2585                                                 &priv);
2586                         BUG_ON(ret);
2587                         extent_op = (struct pending_extent_op *)
2588                                                 (unsigned long)priv;
2589
2590                         clear_extent_bits(&info->extent_ins, start, end,
2591                                           EXTENT_WRITEBACK, GFP_NOFS);
2592
2593                         if (extent_op->type == PENDING_BACKREF_UPDATE) {
2594                                 list_add_tail(&extent_op->list, &delete_list);
2595                                 search = end + 1;
2596                                 nr++;
2597                                 continue;
2598                         }
2599
2600                         mutex_lock(&extent_root->fs_info->pinned_mutex);
2601                         ret = pin_down_bytes(trans, extent_root, start,
2602                                              end + 1 - start, 0);
2603                         mutex_unlock(&extent_root->fs_info->pinned_mutex);
2604
2605                         ret = update_block_group(trans, extent_root, start,
2606                                                 end + 1 - start, 0, ret > 0);
2607
2608                         unlock_extent(extent_ins, start, end, GFP_NOFS);
2609                         BUG_ON(ret);
2610                         kfree(extent_op);
2611                 }
2612                 if (ret)
2613                         err = ret;
2614
2615                 search = end + 1;
2616
2617                 if (need_resched()) {
2618                         mutex_unlock(&info->extent_ins_mutex);
2619                         cond_resched();
2620                         mutex_lock(&info->extent_ins_mutex);
2621                 }
2622         }
2623
2624         if (nr) {
2625                 ret = free_extents(trans, extent_root, &delete_list);
2626                 BUG_ON(ret);
2627         }
2628
2629         if (all && skipped) {
2630                 INIT_LIST_HEAD(&delete_list);
2631                 search = 0;
2632                 nr = 0;
2633                 goto again;
2634         }
2635
2636         return err;
2637 }
2638
2639 /*
2640  * remove an extent from the root, returns 0 on success
2641  */
2642 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
2643                                struct btrfs_root *root,
2644                                u64 bytenr, u64 num_bytes, u64 parent,
2645                                u64 root_objectid, u64 ref_generation,
2646                                u64 owner_objectid, int pin)
2647 {
2648         struct btrfs_root *extent_root = root->fs_info->extent_root;
2649         int pending_ret;
2650         int ret;
2651
2652         WARN_ON(num_bytes < root->sectorsize);
2653         if (root == extent_root) {
2654                 struct pending_extent_op *extent_op = NULL;
2655
2656                 mutex_lock(&root->fs_info->extent_ins_mutex);
2657                 if (test_range_bit(&root->fs_info->extent_ins, bytenr,
2658                                 bytenr + num_bytes - 1, EXTENT_WRITEBACK, 0)) {
2659                         u64 priv;
2660                         ret = get_state_private(&root->fs_info->extent_ins,
2661                                                 bytenr, &priv);
2662                         BUG_ON(ret);
2663                         extent_op = (struct pending_extent_op *)
2664                                                 (unsigned long)priv;
2665
2666                         extent_op->del = 1;
2667                         if (extent_op->type == PENDING_EXTENT_INSERT) {
2668                                 mutex_unlock(&root->fs_info->extent_ins_mutex);
2669                                 return 0;
2670                         }
2671                 }
2672
2673                 if (extent_op) {
2674                         ref_generation = extent_op->orig_generation;
2675                         parent = extent_op->orig_parent;
2676                 }
2677
2678                 extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
2679                 BUG_ON(!extent_op);
2680
2681                 extent_op->type = PENDING_EXTENT_DELETE;
2682                 extent_op->bytenr = bytenr;
2683                 extent_op->num_bytes = num_bytes;
2684                 extent_op->parent = parent;
2685                 extent_op->orig_parent = parent;
2686                 extent_op->generation = ref_generation;
2687                 extent_op->orig_generation = ref_generation;
2688                 extent_op->level = (int)owner_objectid;
2689                 INIT_LIST_HEAD(&extent_op->list);
2690                 extent_op->del = 0;
2691
2692                 set_extent_bits(&root->fs_info->pending_del,
2693                                 bytenr, bytenr + num_bytes - 1,
2694                                 EXTENT_WRITEBACK, GFP_NOFS);
2695                 set_state_private(&root->fs_info->pending_del,
2696                                   bytenr, (unsigned long)extent_op);
2697                 mutex_unlock(&root->fs_info->extent_ins_mutex);
2698                 return 0;
2699         }
2700         /* if metadata always pin */
2701         if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID) {
2702                 if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
2703                         struct btrfs_block_group_cache *cache;
2704
2705                         /* btrfs_free_reserved_extent */
2706                         cache = btrfs_lookup_block_group(root->fs_info, bytenr);
2707                         BUG_ON(!cache);
2708                         btrfs_add_free_space(cache, bytenr, num_bytes);
2709                         put_block_group(cache);
2710                         update_reserved_extents(root, bytenr, num_bytes, 0);
2711                         return 0;
2712                 }
2713                 pin = 1;
2714         }
2715
2716         /* if data pin when any transaction has committed this */
2717         if (ref_generation != trans->transid)
2718                 pin = 1;
2719
2720         ret = __free_extent(trans, root, bytenr, num_bytes, parent,
2721                             root_objectid, ref_generation,
2722                             owner_objectid, pin, pin == 0);
2723
2724         finish_current_insert(trans, root->fs_info->extent_root, 0);
2725         pending_ret = del_pending_extents(trans, root->fs_info->extent_root, 0);
2726         return ret ? ret : pending_ret;
2727 }
2728
2729 int btrfs_free_extent(struct btrfs_trans_handle *trans,
2730                       struct btrfs_root *root,
2731                       u64 bytenr, u64 num_bytes, u64 parent,
2732                       u64 root_objectid, u64 ref_generation,
2733                       u64 owner_objectid, int pin)
2734 {
2735         int ret;
2736
2737         ret = __btrfs_free_extent(trans, root, bytenr, num_bytes, parent,
2738                                   root_objectid, ref_generation,
2739                                   owner_objectid, pin);
2740         return ret;
2741 }
2742
2743 static u64 stripe_align(struct btrfs_root *root, u64 val)
2744 {
2745         u64 mask = ((u64)root->stripesize - 1);
2746         u64 ret = (val + mask) & ~mask;
2747         return ret;
2748 }
2749
2750 /*
2751  * walks the btree of allocated extents and find a hole of a given size.
2752  * The key ins is changed to record the hole:
2753  * ins->objectid == block start
2754  * ins->flags = BTRFS_EXTENT_ITEM_KEY
2755  * ins->offset == number of blocks
2756  * Any available blocks before search_start are skipped.
2757  */
2758 static noinline int find_free_extent(struct btrfs_trans_handle *trans,
2759                                      struct btrfs_root *orig_root,
2760                                      u64 num_bytes, u64 empty_size,
2761                                      u64 search_start, u64 search_end,
2762                                      u64 hint_byte, struct btrfs_key *ins,
2763                                      u64 exclude_start, u64 exclude_nr,
2764                                      int data)
2765 {
2766         int ret = 0;
2767         struct btrfs_root *root = orig_root->fs_info->extent_root;
2768         u64 total_needed = num_bytes;
2769         u64 *last_ptr = NULL;
2770         u64 last_wanted = 0;
2771         struct btrfs_block_group_cache *block_group = NULL;
2772         int chunk_alloc_done = 0;
2773         int empty_cluster = 2 * 1024 * 1024;
2774         int allowed_chunk_alloc = 0;
2775         struct list_head *head = NULL, *cur = NULL;
2776         int loop = 0;
2777         int extra_loop = 0;
2778         struct btrfs_space_info *space_info;
2779
2780         WARN_ON(num_bytes < root->sectorsize);
2781         btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
2782         ins->objectid = 0;
2783         ins->offset = 0;
2784
2785         if (orig_root->ref_cows || empty_size)
2786                 allowed_chunk_alloc = 1;
2787
2788         if (data & BTRFS_BLOCK_GROUP_METADATA) {
2789                 last_ptr = &root->fs_info->last_alloc;
2790                 empty_cluster = 64 * 1024;
2791         }
2792
2793         if ((data & BTRFS_BLOCK_GROUP_DATA) && btrfs_test_opt(root, SSD))
2794                 last_ptr = &root->fs_info->last_data_alloc;
2795
2796         if (last_ptr) {
2797                 if (*last_ptr) {
2798                         hint_byte = *last_ptr;
2799                         last_wanted = *last_ptr;
2800                 } else
2801                         empty_size += empty_cluster;
2802         } else {
2803                 empty_cluster = 0;
2804         }
2805         search_start = max(search_start, first_logical_byte(root, 0));
2806         search_start = max(search_start, hint_byte);
2807
2808         if (last_wanted && search_start != last_wanted) {
2809                 last_wanted = 0;
2810                 empty_size += empty_cluster;
2811         }
2812
2813         total_needed += empty_size;
2814         block_group = btrfs_lookup_block_group(root->fs_info, search_start);
2815         if (!block_group)
2816                 block_group = btrfs_lookup_first_block_group(root->fs_info,
2817                                                              search_start);
2818         space_info = __find_space_info(root->fs_info, data);
2819
2820         down_read(&space_info->groups_sem);
2821         while (1) {
2822                 struct btrfs_free_space *free_space;
2823                 /*
2824                  * the only way this happens if our hint points to a block
2825                  * group thats not of the proper type, while looping this
2826                  * should never happen
2827                  */
2828                 if (empty_size)
2829                         extra_loop = 1;
2830
2831                 if (!block_group)
2832                         goto new_group_no_lock;
2833
2834                 if (unlikely(!block_group->cached)) {
2835                         mutex_lock(&block_group->cache_mutex);
2836                         ret = cache_block_group(root, block_group);
2837                         mutex_unlock(&block_group->cache_mutex);
2838                         if (ret)
2839                                 break;
2840                 }
2841
2842                 mutex_lock(&block_group->alloc_mutex);
2843                 if (unlikely(!block_group_bits(block_group, data)))
2844                         goto new_group;
2845
2846                 if (unlikely(block_group->ro))
2847                         goto new_group;
2848
2849                 free_space = btrfs_find_free_space(block_group, search_start,
2850                                                    total_needed);
2851                 if (free_space) {
2852                         u64 start = block_group->key.objectid;
2853                         u64 end = block_group->key.objectid +
2854                                 block_group->key.offset;
2855
2856                         search_start = stripe_align(root, free_space->offset);
2857
2858                         /* move on to the next group */
2859                         if (search_start + num_bytes >= search_end)
2860                                 goto new_group;
2861
2862                         /* move on to the next group */
2863                         if (search_start + num_bytes > end)
2864                                 goto new_group;
2865
2866                         if (last_wanted && search_start != last_wanted) {
2867                                 total_needed += empty_cluster;
2868                                 empty_size += empty_cluster;
2869                                 last_wanted = 0;
2870                                 /*
2871                                  * if search_start is still in this block group
2872                                  * then we just re-search this block group
2873                                  */
2874                                 if (search_start >= start &&
2875                                     search_start < end) {
2876                                         mutex_unlock(&block_group->alloc_mutex);
2877                                         continue;
2878                                 }
2879
2880                                 /* else we go to the next block group */
2881                                 goto new_group;
2882                         }
2883
2884                         if (exclude_nr > 0 &&
2885                             (search_start + num_bytes > exclude_start &&
2886                              search_start < exclude_start + exclude_nr)) {
2887                                 search_start = exclude_start + exclude_nr;
2888                                 /*
2889                                  * if search_start is still in this block group
2890                                  * then we just re-search this block group
2891                                  */
2892                                 if (search_start >= start &&
2893                                     search_start < end) {
2894                                         mutex_unlock(&block_group->alloc_mutex);
2895                                         last_wanted = 0;
2896                                         continue;
2897                                 }
2898
2899                                 /* else we go to the next block group */
2900                                 goto new_group;
2901                         }
2902
2903                         ins->objectid = search_start;
2904                         ins->offset = num_bytes;
2905
2906                         btrfs_remove_free_space_lock(block_group, search_start,
2907                                                      num_bytes);
2908                         /* we are all good, lets return */
2909                         mutex_unlock(&block_group->alloc_mutex);
2910                         break;
2911                 }
2912 new_group:
2913                 mutex_unlock(&block_group->alloc_mutex);
2914                 put_block_group(block_group);
2915                 block_group = NULL;
2916 new_group_no_lock:
2917                 /* don't try to compare new allocations against the
2918                  * last allocation any more
2919                  */
2920                 last_wanted = 0;
2921
2922                 /*
2923                  * Here's how this works.
2924                  * loop == 0: we were searching a block group via a hint
2925                  *              and didn't find anything, so we start at
2926                  *              the head of the block groups and keep searching
2927                  * loop == 1: we're searching through all of the block groups
2928                  *              if we hit the head again we have searched
2929                  *              all of the block groups for this space and we
2930                  *              need to try and allocate, if we cant error out.
2931                  * loop == 2: we allocated more space and are looping through
2932                  *              all of the block groups again.
2933                  */
2934                 if (loop == 0) {
2935                         head = &space_info->block_groups;
2936                         cur = head->next;
2937                         loop++;
2938                 } else if (loop == 1 && cur == head) {
2939                         int keep_going;
2940
2941                         /* at this point we give up on the empty_size
2942                          * allocations and just try to allocate the min
2943                          * space.
2944                          *
2945                          * The extra_loop field was set if an empty_size
2946                          * allocation was attempted above, and if this
2947                          * is try we need to try the loop again without
2948                          * the additional empty_size.
2949                          */
2950                         total_needed -= empty_size;
2951                         empty_size = 0;
2952                         keep_going = extra_loop;
2953                         loop++;
2954
2955                         if (allowed_chunk_alloc && !chunk_alloc_done) {
2956                                 up_read(&space_info->groups_sem);
2957                                 ret = do_chunk_alloc(trans, root, num_bytes +
2958                                                      2 * 1024 * 1024, data, 1);
2959                                 down_read(&space_info->groups_sem);
2960                                 if (ret < 0)
2961                                         goto loop_check;
2962                                 head = &space_info->block_groups;
2963                                 /*
2964                                  * we've allocated a new chunk, keep
2965                                  * trying
2966                                  */
2967                                 keep_going = 1;
2968                                 chunk_alloc_done = 1;
2969                         } else if (!allowed_chunk_alloc) {
2970                                 space_info->force_alloc = 1;
2971                         }
2972 loop_check:
2973                         if (keep_going) {
2974                                 cur = head->next;
2975                                 extra_loop = 0;
2976                         } else {
2977                                 break;
2978                         }
2979                 } else if (cur == head) {
2980                         break;
2981                 }
2982
2983                 block_group = list_entry(cur, struct btrfs_block_group_cache,
2984                                          list);
2985                 atomic_inc(&block_group->count);
2986
2987                 search_start = block_group->key.objectid;
2988                 cur = cur->next;
2989         }
2990
2991         /* we found what we needed */
2992         if (ins->objectid) {
2993                 if (!(data & BTRFS_BLOCK_GROUP_DATA))
2994                         trans->block_group = block_group->key.objectid;
2995
2996                 if (last_ptr)
2997                         *last_ptr = ins->objectid + ins->offset;
2998                 ret = 0;
2999         } else if (!ret) {
3000                 printk(KERN_ERR "btrfs searching for %llu bytes, "
3001                        "num_bytes %llu, loop %d, allowed_alloc %d\n",
3002                        (unsigned long long)total_needed,
3003                        (unsigned long long)num_bytes,
3004                        loop, allowed_chunk_alloc);
3005                 ret = -ENOSPC;
3006         }
3007         if (block_group)
3008                 put_block_group(block_group);
3009
3010         up_read(&space_info->groups_sem);
3011         return ret;
3012 }
3013
3014 static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
3015 {
3016         struct btrfs_block_group_cache *cache;
3017         struct list_head *l;
3018
3019         printk(KERN_INFO "space_info has %llu free, is %sfull\n",
3020                (unsigned long long)(info->total_bytes - info->bytes_used -
3021                                     info->bytes_pinned - info->bytes_reserved),
3022                (info->full) ? "" : "not ");
3023
3024         down_read(&info->groups_sem);
3025         list_for_each(l, &info->block_groups) {
3026                 cache = list_entry(l, struct btrfs_block_group_cache, list);
3027                 spin_lock(&cache->lock);
3028                 printk(KERN_INFO "block group %llu has %llu bytes, %llu used "
3029                        "%llu pinned %llu reserved\n",
3030                        (unsigned long long)cache->key.objectid,
3031                        (unsigned long long)cache->key.offset,
3032                        (unsigned long long)btrfs_block_group_used(&cache->item),
3033                        (unsigned long long)cache->pinned,
3034                        (unsigned long long)cache->reserved);
3035                 btrfs_dump_free_space(cache, bytes);
3036                 spin_unlock(&cache->lock);
3037         }
3038         up_read(&info->groups_sem);
3039 }
3040
3041 static int __btrfs_reserve_extent(struct btrfs_trans_handle *trans,
3042                                   struct btrfs_root *root,
3043                                   u64 num_bytes, u64 min_alloc_size,
3044                                   u64 empty_size, u64 hint_byte,
3045                                   u64 search_end, struct btrfs_key *ins,
3046                                   u64 data)
3047 {
3048         int ret;
3049         u64 search_start = 0;
3050         u64 alloc_profile;
3051         struct btrfs_fs_info *info = root->fs_info;
3052
3053         if (data) {
3054                 alloc_profile = info->avail_data_alloc_bits &
3055                         info->data_alloc_profile;
3056                 data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
3057         } else if (root == root->fs_info->chunk_root) {
3058                 alloc_profile = info->avail_system_alloc_bits &
3059                         info->system_alloc_profile;
3060                 data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
3061         } else {
3062                 alloc_profile = info->avail_metadata_alloc_bits &
3063                         info->metadata_alloc_profile;
3064                 data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
3065         }
3066 again:
3067         data = btrfs_reduce_alloc_profile(root, data);
3068         /*
3069          * the only place that sets empty_size is btrfs_realloc_node, which
3070          * is not called recursively on allocations
3071          */
3072         if (empty_size || root->ref_cows) {
3073                 if (!(data & BTRFS_BLOCK_GROUP_METADATA)) {
3074                         ret = do_chunk_alloc(trans, root->fs_info->extent_root,
3075                                      2 * 1024 * 1024,
3076                                      BTRFS_BLOCK_GROUP_METADATA |
3077                                      (info->metadata_alloc_profile &
3078                                       info->avail_metadata_alloc_bits), 0);
3079                 }
3080                 ret = do_chunk_alloc(trans, root->fs_info->extent_root,
3081                                      num_bytes + 2 * 1024 * 1024, data, 0);
3082         }
3083
3084         WARN_ON(num_bytes < root->sectorsize);
3085         ret = find_free_extent(trans, root, num_bytes, empty_size,
3086                                search_start, search_end, hint_byte, ins,
3087                                trans->alloc_exclude_start,
3088                                trans->alloc_exclude_nr, data);
3089
3090         if (ret == -ENOSPC && num_bytes > min_alloc_size) {
3091                 num_bytes = num_bytes >> 1;
3092                 num_bytes = num_bytes & ~(root->sectorsize - 1);
3093                 num_bytes = max(num_bytes, min_alloc_size);
3094                 do_chunk_alloc(trans, root->fs_info->extent_root,
3095                                num_bytes, data, 1);
3096                 goto again;
3097         }
3098         if (ret) {
3099                 struct btrfs_space_info *sinfo;
3100
3101                 sinfo = __find_space_info(root->fs_info, data);
3102                 printk(KERN_ERR "btrfs allocation failed flags %llu, "
3103                        "wanted %llu\n", (unsigned long long)data,
3104                        (unsigned long long)num_bytes);
3105                 dump_space_info(sinfo, num_bytes);
3106                 BUG();
3107         }
3108
3109         return ret;
3110 }
3111
3112 int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len)
3113 {
3114         struct btrfs_block_group_cache *cache;
3115         int ret = 0;
3116
3117         cache = btrfs_lookup_block_group(root->fs_info, start);
3118         if (!cache) {
3119                 printk(KERN_ERR "Unable to find block group for %llu\n",
3120                        (unsigned long long)start);
3121                 return -ENOSPC;
3122         }
3123
3124         ret = btrfs_discard_extent(root, start, len);
3125
3126         btrfs_add_free_space(cache, start, len);
3127         put_block_group(cache);
3128         update_reserved_extents(root, start, len, 0);
3129
3130         return ret;
3131 }
3132
3133 int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
3134                                   struct btrfs_root *root,
3135                                   u64 num_bytes, u64 min_alloc_size,
3136                                   u64 empty_size, u64 hint_byte,
3137                                   u64 search_end, struct btrfs_key *ins,
3138                                   u64 data)
3139 {
3140         int ret;
3141         ret = __btrfs_reserve_extent(trans, root, num_bytes, min_alloc_size,
3142                                      empty_size, hint_byte, search_end, ins,
3143                                      data);
3144         update_reserved_extents(root, ins->objectid, ins->offset, 1);
3145         return ret;
3146 }
3147
3148 static int __btrfs_alloc_reserved_extent(struct btrfs_trans_handle *trans,
3149                                          struct btrfs_root *root, u64 parent,
3150                                          u64 root_objectid, u64 ref_generation,
3151                                          u64 owner, struct btrfs_key *ins)
3152 {
3153         int ret;
3154         int pending_ret;
3155         u64 super_used;
3156         u64 root_used;
3157         u64 num_bytes = ins->offset;
3158         u32 sizes[2];
3159         struct btrfs_fs_info *info = root->fs_info;
3160         struct btrfs_root *extent_root = info->extent_root;
3161         struct btrfs_extent_item *extent_item;
3162         struct btrfs_extent_ref *ref;
3163         struct btrfs_path *path;
3164         struct btrfs_key keys[2];
3165
3166         if (parent == 0)
3167                 parent = ins->objectid;
3168
3169         /* block accounting for super block */
3170         spin_lock(&info->delalloc_lock);
3171         super_used = btrfs_super_bytes_used(&info->super_copy);
3172         btrfs_set_super_bytes_used(&info->super_copy, super_used + num_bytes);
3173
3174         /* block accounting for root item */
3175         root_used = btrfs_root_used(&root->root_item);
3176         btrfs_set_root_used(&root->root_item, root_used + num_bytes);
3177         spin_unlock(&info->delalloc_lock);
3178
3179         if (root == extent_root) {
3180                 struct pending_extent_op *extent_op;
3181
3182                 extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
3183                 BUG_ON(!extent_op);
3184
3185                 extent_op->type = PENDING_EXTENT_INSERT;
3186                 extent_op->bytenr = ins->objectid;
3187                 extent_op->num_bytes = ins->offset;
3188                 extent_op->parent = parent;
3189                 extent_op->orig_parent = 0;
3190                 extent_op->generation = ref_generation;
3191                 extent_op->orig_generation = 0;
3192                 extent_op->level = (int)owner;
3193                 INIT_LIST_HEAD(&extent_op->list);
3194                 extent_op->del = 0;
3195
3196                 mutex_lock(&root->fs_info->extent_ins_mutex);
3197                 set_extent_bits(&root->fs_info->extent_ins, ins->objectid,
3198                                 ins->objectid + ins->offset - 1,
3199                                 EXTENT_WRITEBACK, GFP_NOFS);
3200                 set_state_private(&root->fs_info->extent_ins,
3201                                   ins->objectid, (unsigned long)extent_op);
3202                 mutex_unlock(&root->fs_info->extent_ins_mutex);
3203                 goto update_block;
3204         }
3205
3206         memcpy(&keys[0], ins, sizeof(*ins));
3207         keys[1].objectid = ins->objectid;
3208         keys[1].type = BTRFS_EXTENT_REF_KEY;
3209         keys[1].offset = parent;
3210         sizes[0] = sizeof(*extent_item);
3211         sizes[1] = sizeof(*ref);
3212
3213         path = btrfs_alloc_path();
3214         BUG_ON(!path);
3215
3216         ret = btrfs_insert_empty_items(trans, extent_root, path, keys,
3217                                        sizes, 2);
3218         BUG_ON(ret);
3219
3220         extent_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
3221                                      struct btrfs_extent_item);
3222         btrfs_set_extent_refs(path->nodes[0], extent_item, 1);
3223         ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
3224                              struct btrfs_extent_ref);
3225
3226         btrfs_set_ref_root(path->nodes[0], ref, root_objectid);
3227         btrfs_set_ref_generation(path->nodes[0], ref, ref_generation);
3228         btrfs_set_ref_objectid(path->nodes[0], ref, owner);
3229         btrfs_set_ref_num_refs(path->nodes[0], ref, 1);
3230
3231         btrfs_mark_buffer_dirty(path->nodes[0]);
3232
3233         trans->alloc_exclude_start = 0;
3234         trans->alloc_exclude_nr = 0;
3235         btrfs_free_path(path);
3236         finish_current_insert(trans, extent_root, 0);
3237         pending_ret = del_pending_extents(trans, extent_root, 0);
3238
3239         if (ret)
3240                 goto out;
3241         if (pending_ret) {
3242                 ret = pending_ret;
3243                 goto out;
3244         }
3245
3246 update_block:
3247         ret = update_block_group(trans, root, ins->objectid,
3248                                  ins->offset, 1, 0);
3249         if (ret) {
3250                 printk(KERN_ERR "btrfs update block group failed for %llu "
3251                        "%llu\n", (unsigned long long)ins->objectid,
3252                        (unsigned long long)ins->offset);
3253                 BUG();
3254         }
3255 out:
3256         return ret;
3257 }
3258
3259 int btrfs_alloc_reserved_extent(struct btrfs_trans_handle *trans,
3260                                 struct btrfs_root *root, u64 parent,
3261                                 u64 root_objectid, u64 ref_generation,
3262                                 u64 owner, struct btrfs_key *ins)
3263 {
3264         int ret;
3265
3266         if (root_objectid == BTRFS_TREE_LOG_OBJECTID)
3267                 return 0;
3268         ret = __btrfs_alloc_reserved_extent(trans, root, parent, root_objectid,
3269                                             ref_generation, owner, ins);
3270         update_reserved_extents(root, ins->objectid, ins->offset, 0);
3271         return ret;
3272 }
3273
3274 /*
3275  * this is used by the tree logging recovery code.  It records that
3276  * an extent has been allocated and makes sure to clear the free
3277  * space cache bits as well
3278  */
3279 int btrfs_alloc_logged_extent(struct btrfs_trans_handle *trans,
3280                                 struct btrfs_root *root, u64 parent,
3281                                 u64 root_objectid, u64 ref_generation,
3282                                 u64 owner, struct btrfs_key *ins)
3283 {
3284         int ret;
3285         struct btrfs_block_group_cache *block_group;
3286
3287         block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid);
3288         mutex_lock(&block_group->cache_mutex);
3289         cache_block_group(root, block_group);
3290         mutex_unlock(&block_group->cache_mutex);
3291
3292         ret = btrfs_remove_free_space(block_group, ins->objectid,
3293                                       ins->offset);
3294         BUG_ON(ret);
3295         put_block_group(block_group);
3296         ret = __btrfs_alloc_reserved_extent(trans, root, parent, root_objectid,
3297                                             ref_generation, owner, ins);
3298         return ret;
3299 }
3300
3301 /*
3302  * finds a free extent and does all the dirty work required for allocation
3303  * returns the key for the extent through ins, and a tree buffer for
3304  * the first block of the extent through buf.
3305  *
3306  * returns 0 if everything worked, non-zero otherwise.
3307  */
3308 int btrfs_alloc_extent(struct btrfs_trans_handle *trans,
3309                        struct btrfs_root *root,
3310                        u64 num_bytes, u64 parent, u64 min_alloc_size,
3311                        u64 root_objectid, u64 ref_generation,
3312                        u64 owner_objectid, u64 empty_size, u64 hint_byte,
3313                        u64 search_end, struct btrfs_key *ins, u64 data)
3314 {
3315         int ret;
3316
3317         ret = __btrfs_reserve_extent(trans, root, num_bytes,
3318                                      min_alloc_size, empty_size, hint_byte,
3319                                      search_end, ins, data);
3320         BUG_ON(ret);
3321         if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
3322                 ret = __btrfs_alloc_reserved_extent(trans, root, parent,
3323                                         root_objectid, ref_generation,
3324                                         owner_objectid, ins);
3325                 BUG_ON(ret);
3326
3327         } else {
3328                 update_reserved_extents(root, ins->objectid, ins->offset, 1);
3329         }
3330         return ret;
3331 }
3332
3333 struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
3334                                             struct btrfs_root *root,
3335                                             u64 bytenr, u32 blocksize)
3336 {
3337         struct extent_buffer *buf;
3338
3339         buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
3340         if (!buf)
3341                 return ERR_PTR(-ENOMEM);
3342         btrfs_set_header_generation(buf, trans->transid);
3343         btrfs_tree_lock(buf);
3344         clean_tree_block(trans, root, buf);
3345         btrfs_set_buffer_uptodate(buf);
3346         if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
3347                 set_extent_dirty(&root->dirty_log_pages, buf->start,
3348                          buf->start + buf->len - 1, GFP_NOFS);
3349         } else {
3350                 set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
3351                          buf->start + buf->len - 1, GFP_NOFS);
3352         }
3353         trans->blocks_used++;
3354         return buf;
3355 }
3356
3357 /*
3358  * helper function to allocate a block for a given tree
3359  * returns the tree buffer or NULL.
3360  */
3361 struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
3362                                              struct btrfs_root *root,
3363                                              u32 blocksize, u64 parent,
3364                                              u64 root_objectid,
3365                                              u64 ref_generation,
3366                                              int level,
3367                                              u64 hint,
3368                                              u64 empty_size)
3369 {
3370         struct btrfs_key ins;
3371         int ret;
3372         struct extent_buffer *buf;
3373
3374         ret = btrfs_alloc_extent(trans, root, blocksize, parent, blocksize,
3375                                  root_objectid, ref_generation, level,
3376                                  empty_size, hint, (u64)-1, &ins, 0);
3377         if (ret) {
3378                 BUG_ON(ret > 0);
3379                 return ERR_PTR(ret);
3380         }
3381
3382         buf = btrfs_init_new_buffer(trans, root, ins.objectid, blocksize);
3383         return buf;
3384 }
3385
3386 int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
3387                         struct btrfs_root *root, struct extent_buffer *leaf)
3388 {
3389         u64 leaf_owner;
3390         u64 leaf_generation;
3391         struct btrfs_key key;
3392         struct btrfs_file_extent_item *fi;
3393         int i;
3394         int nritems;
3395         int ret;
3396
3397         BUG_ON(!btrfs_is_leaf(leaf));
3398         nritems = btrfs_header_nritems(leaf);
3399         leaf_owner = btrfs_header_owner(leaf);
3400         leaf_generation = btrfs_header_generation(leaf);
3401
3402         for (i = 0; i < nritems; i++) {
3403                 u64 disk_bytenr;
3404                 cond_resched();
3405
3406                 btrfs_item_key_to_cpu(leaf, &key, i);
3407                 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
3408                         continue;
3409                 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
3410                 if (btrfs_file_extent_type(leaf, fi) ==
3411                     BTRFS_FILE_EXTENT_INLINE)
3412                         continue;
3413                 /*
3414                  * FIXME make sure to insert a trans record that
3415                  * repeats the snapshot del on crash
3416                  */
3417                 disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
3418                 if (disk_bytenr == 0)
3419                         continue;
3420
3421                 ret = __btrfs_free_extent(trans, root, disk_bytenr,
3422                                 btrfs_file_extent_disk_num_bytes(leaf, fi),
3423                                 leaf->start, leaf_owner, leaf_generation,
3424                                 key.objectid, 0);
3425                 BUG_ON(ret);
3426
3427                 atomic_inc(&root->fs_info->throttle_gen);
3428                 wake_up(&root->fs_info->transaction_throttle);
3429                 cond_resched();
3430         }
3431         return 0;
3432 }
3433
3434 static noinline int cache_drop_leaf_ref(struct btrfs_trans_handle *trans,
3435                                         struct btrfs_root *root,
3436                                         struct btrfs_leaf_ref *ref)
3437 {
3438         int i;
3439         int ret;
3440         struct btrfs_extent_info *info = ref->extents;
3441
3442         for (i = 0; i < ref->nritems; i++) {
3443                 ret = __btrfs_free_extent(trans, root, info->bytenr,
3444                                           info->num_bytes, ref->bytenr,
3445                                           ref->owner, ref->generation,
3446                                           info->objectid, 0);
3447
3448                 atomic_inc(&root->fs_info->throttle_gen);
3449                 wake_up(&root->fs_info->transaction_throttle);
3450                 cond_resched();
3451
3452                 BUG_ON(ret);
3453                 info++;
3454         }
3455
3456         return 0;
3457 }
3458
3459 static int drop_snap_lookup_refcount(struct btrfs_root *root, u64 start,
3460                                      u64 len, u32 *refs)
3461 {
3462         int ret;
3463
3464         ret = btrfs_lookup_extent_ref(NULL, root, start, len, refs);
3465         BUG_ON(ret);
3466
3467 #if 0 /* some debugging code in case we see problems here */
3468         /* if the refs count is one, it won't get increased again.  But
3469          * if the ref count is > 1, someone may be decreasing it at
3470          * the same time we are.
3471          */
3472         if (*refs != 1) {
3473                 struct extent_buffer *eb = NULL;
3474                 eb = btrfs_find_create_tree_block(root, start, len);
3475                 if (eb)
3476                         btrfs_tree_lock(eb);
3477
3478                 mutex_lock(&root->fs_info->alloc_mutex);
3479                 ret = lookup_extent_ref(NULL, root, start, len, refs);
3480                 BUG_ON(ret);
3481                 mutex_unlock(&root->fs_info->alloc_mutex);
3482
3483                 if (eb) {
3484                         btrfs_tree_unlock(eb);
3485                         free_extent_buffer(eb);
3486                 }
3487                 if (*refs == 1) {
3488                         printk(KERN_ERR "btrfs block %llu went down to one "
3489                                "during drop_snap\n", (unsigned long long)start);
3490                 }
3491
3492         }
3493 #endif
3494
3495         cond_resched();
3496         return ret;
3497 }
3498
3499 /*
3500  * helper function for drop_snapshot, this walks down the tree dropping ref
3501  * counts as it goes.
3502  */
3503 static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
3504                                    struct btrfs_root *root,
3505                                    struct btrfs_path *path, int *level)
3506 {
3507         u64 root_owner;
3508         u64 root_gen;
3509         u64 bytenr;
3510         u64 ptr_gen;
3511         struct extent_buffer *next;
3512         struct extent_buffer *cur;
3513         struct extent_buffer *parent;
3514         struct btrfs_leaf_ref *ref;
3515         u32 blocksize;
3516         int ret;
3517         u32 refs;
3518
3519         WARN_ON(*level < 0);
3520         WARN_ON(*level >= BTRFS_MAX_LEVEL);
3521         ret = drop_snap_lookup_refcount(root, path->nodes[*level]->start,
3522                                 path->nodes[*level]->len, &refs);
3523         BUG_ON(ret);
3524         if (refs > 1)
3525                 goto out;
3526
3527         /*
3528          * walk down to the last node level and free all the leaves
3529          */
3530         while (*level >= 0) {
3531                 WARN_ON(*level < 0);
3532                 WARN_ON(*level >= BTRFS_MAX_LEVEL);
3533                 cur = path->nodes[*level];
3534
3535                 if (btrfs_header_level(cur) != *level)
3536                         WARN_ON(1);
3537
3538                 if (path->slots[*level] >=
3539                     btrfs_header_nritems(cur))
3540                         break;
3541                 if (*level == 0) {
3542                         ret = btrfs_drop_leaf_ref(trans, root, cur);
3543                         BUG_ON(ret);
3544                         break;
3545                 }
3546                 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
3547                 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
3548                 blocksize = btrfs_level_size(root, *level - 1);
3549
3550                 ret = drop_snap_lookup_refcount(root, bytenr, blocksize, &refs);
3551                 BUG_ON(ret);
3552                 if (refs != 1) {
3553                         parent = path->nodes[*level];
3554                         root_owner = btrfs_header_owner(parent);
3555                         root_gen = btrfs_header_generation(parent);
3556                         path->slots[*level]++;
3557
3558                         ret = __btrfs_free_extent(trans, root, bytenr,
3559                                                 blocksize, parent->start,
3560                                                 root_owner, root_gen,
3561                                                 *level - 1, 1);
3562                         BUG_ON(ret);
3563
3564                         atomic_inc(&root->fs_info->throttle_gen);
3565                         wake_up(&root->fs_info->transaction_throttle);
3566                         cond_resched();
3567
3568                         continue;
3569                 }
3570                 /*
3571                  * at this point, we have a single ref, and since the
3572                  * only place referencing this extent is a dead root
3573                  * the reference count should never go higher.
3574                  * So, we don't need to check it again
3575                  */
3576                 if (*level == 1) {
3577                         ref = btrfs_lookup_leaf_ref(root, bytenr);
3578                         if (ref && ref->generation != ptr_gen) {
3579                                 btrfs_free_leaf_ref(root, ref);
3580                                 ref = NULL;
3581                         }
3582                         if (ref) {
3583                                 ret = cache_drop_leaf_ref(trans, root, ref);
3584                                 BUG_ON(ret);
3585                                 btrfs_remove_leaf_ref(root, ref);
3586                                 btrfs_free_leaf_ref(root, ref);
3587                                 *level = 0;
3588                                 break;
3589                         }
3590                 }
3591                 next = btrfs_find_tree_block(root, bytenr, blocksize);
3592                 if (!next || !btrfs_buffer_uptodate(next, ptr_gen)) {
3593                         free_extent_buffer(next);
3594
3595                         next = read_tree_block(root, bytenr, blocksize,
3596                                                ptr_gen);
3597                         cond_resched();
3598 #if 0
3599                         /*
3600                          * this is a debugging check and can go away
3601                          * the ref should never go all the way down to 1
3602                          * at this point
3603                          */
3604                         ret = lookup_extent_ref(NULL, root, bytenr, blocksize,
3605                                                 &refs);
3606                         BUG_ON(ret);
3607                         WARN_ON(refs != 1);
3608 #endif
3609                 }
3610                 WARN_ON(*level <= 0);
3611                 if (path->nodes[*level-1])
3612                         free_extent_buffer(path->nodes[*level-1]);
3613                 path->nodes[*level-1] = next;
3614                 *level = btrfs_header_level(next);
3615                 path->slots[*level] = 0;
3616                 cond_resched();
3617         }
3618 out:
3619         WARN_ON(*level < 0);
3620         WARN_ON(*level >= BTRFS_MAX_LEVEL);
3621
3622         if (path->nodes[*level] == root->node) {
3623                 parent = path->nodes[*level];
3624                 bytenr = path->nodes[*level]->start;
3625         } else {
3626                 parent = path->nodes[*level + 1];
3627                 bytenr = btrfs_node_blockptr(parent, path->slots[*level + 1]);
3628         }
3629
3630         blocksize = btrfs_level_size(root, *level);
3631         root_owner = btrfs_header_owner(parent);
3632         root_gen = btrfs_header_generation(parent);
3633
3634         ret = __btrfs_free_extent(trans, root, bytenr, blocksize,
3635                                   parent->start, root_owner, root_gen,
3636                                   *level, 1);
3637         free_extent_buffer(path->nodes[*level]);
3638         path->nodes[*level] = NULL;
3639         *level += 1;
3640         BUG_ON(ret);
3641
3642         cond_resched();
3643         return 0;
3644 }
3645
3646 /*
3647  * helper function for drop_subtree, this function is similar to
3648  * walk_down_tree. The main difference is that it checks reference
3649  * counts while tree blocks are locked.
3650  */
3651 static noinline int walk_down_subtree(struct btrfs_trans_handle *trans,
3652                                       struct btrfs_root *root,
3653                                       struct btrfs_path *path, int *level)
3654 {
3655         struct extent_buffer *next;
3656         struct extent_buffer *cur;
3657         struct extent_buffer *parent;
3658         u64 bytenr;
3659         u64 ptr_gen;
3660         u32 blocksize;
3661         u32 refs;
3662         int ret;
3663
3664         cur = path->nodes[*level];
3665         ret = btrfs_lookup_extent_ref(trans, root, cur->start, cur->len,
3666                                       &refs);
3667         BUG_ON(ret);
3668         if (refs > 1)
3669                 goto out;
3670
3671         while (*level >= 0) {
3672                 cur = path->nodes[*level];
3673                 if (*level == 0) {
3674                         ret = btrfs_drop_leaf_ref(trans, root, cur);
3675                         BUG_ON(ret);
3676                         clean_tree_block(trans, root, cur);
3677                         break;
3678                 }
3679                 if (path->slots[*level] >= btrfs_header_nritems(cur)) {
3680                         clean_tree_block(trans, root, cur);
3681                         break;
3682                 }
3683
3684                 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
3685                 blocksize = btrfs_level_size(root, *level - 1);
3686                 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
3687
3688                 next = read_tree_block(root, bytenr, blocksize, ptr_gen);
3689                 btrfs_tree_lock(next);
3690
3691                 ret = btrfs_lookup_extent_ref(trans, root, bytenr, blocksize,
3692                                               &refs);
3693                 BUG_ON(ret);
3694                 if (refs > 1) {
3695                         parent = path->nodes[*level];
3696                         ret = btrfs_free_extent(trans, root, bytenr,
3697                                         blocksize, parent->start,
3698                                         btrfs_header_owner(parent),
3699                                         btrfs_header_generation(parent),
3700                                         *level - 1, 1);
3701                         BUG_ON(ret);
3702                         path->slots[*level]++;
3703                         btrfs_tree_unlock(next);
3704                         free_extent_buffer(next);
3705                         continue;
3706                 }
3707
3708                 *level = btrfs_header_level(next);
3709                 path->nodes[*level] = next;
3710                 path->slots[*level] = 0;
3711                 path->locks[*level] = 1;
3712                 cond_resched();
3713         }
3714 out:
3715         parent = path->nodes[*level + 1];
3716         bytenr = path->nodes[*level]->start;
3717         blocksize = path->nodes[*level]->len;
3718
3719         ret = btrfs_free_extent(trans, root, bytenr, blocksize,
3720                         parent->start, btrfs_header_owner(parent),
3721                         btrfs_header_generation(parent), *level, 1);
3722         BUG_ON(ret);
3723
3724         if (path->locks[*level]) {
3725                 btrfs_tree_unlock(path->nodes[*level]);
3726                 path->locks[*level] = 0;
3727         }
3728         free_extent_buffer(path->nodes[*level]);
3729         path->nodes[*level] = NULL;
3730         *level += 1;
3731         cond_resched();
3732         return 0;
3733 }
3734
3735 /*
3736  * helper for dropping snapshots.  This walks back up the tree in the path
3737  * to find the first node higher up where we haven't yet gone through
3738  * all the slots
3739  */
3740 static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
3741                                  struct btrfs_root *root,
3742                                  struct btrfs_path *path,
3743                                  int *level, int max_level)
3744 {
3745         u64 root_owner;
3746         u64 root_gen;
3747         struct btrfs_root_item *root_item = &root->root_item;
3748         int i;
3749         int slot;
3750         int ret;
3751
3752         for (i = *level; i < max_level && path->nodes[i]; i++) {
3753                 slot = path->slots[i];
3754                 if (slot < btrfs_header_nritems(path->nodes[i]) - 1) {
3755                         struct extent_buffer *node;
3756                         struct btrfs_disk_key disk_key;
3757                         node = path->nodes[i];
3758                         path->slots[i]++;
3759                         *level = i;
3760                         WARN_ON(*level == 0);
3761                         btrfs_node_key(node, &disk_key, path->slots[i]);
3762                         memcpy(&root_item->drop_progress,
3763                                &disk_key, sizeof(disk_key));
3764                         root_item->drop_level = i;
3765                         return 0;
3766                 } else {
3767                         struct extent_buffer *parent;
3768                         if (path->nodes[*level] == root->node)
3769                                 parent = path->nodes[*level];
3770                         else
3771                                 parent = path->nodes[*level + 1];
3772
3773                         root_owner = btrfs_header_owner(parent);
3774                         root_gen = btrfs_header_generation(parent);
3775
3776                         clean_tree_block(trans, root, path->nodes[*level]);
3777                         ret = btrfs_free_extent(trans, root,
3778                                                 path->nodes[*level]->start,
3779                                                 path->nodes[*level]->len,
3780                                                 parent->start, root_owner,
3781                                                 root_gen, *level, 1);
3782                         BUG_ON(ret);
3783                         if (path->locks[*level]) {
3784                                 btrfs_tree_unlock(path->nodes[*level]);
3785                                 path->locks[*level] = 0;
3786                         }
3787                         free_extent_buffer(path->nodes[*level]);
3788                         path->nodes[*level] = NULL;
3789                         *level = i + 1;
3790                 }
3791         }
3792         return 1;
3793 }
3794
3795 /*
3796  * drop the reference count on the tree rooted at 'snap'.  This traverses
3797  * the tree freeing any blocks that have a ref count of zero after being
3798  * decremented.
3799  */
3800 int btrfs_drop_snapshot(struct btrfs_trans_handle *trans, struct btrfs_root
3801                         *root)
3802 {
3803         int ret = 0;
3804         int wret;
3805         int level;
3806         struct btrfs_path *path;
3807         int i;
3808         int orig_level;
3809         struct btrfs_root_item *root_item = &root->root_item;
3810
3811         WARN_ON(!mutex_is_locked(&root->fs_info->drop_mutex));
3812         path = btrfs_alloc_path();
3813         BUG_ON(!path);
3814
3815         level = btrfs_header_level(root->node);
3816         orig_level = level;
3817         if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
3818                 path->nodes[level] = root->node;
3819                 extent_buffer_get(root->node);
3820                 path->slots[level] = 0;
3821         } else {
3822                 struct btrfs_key key;
3823                 struct btrfs_disk_key found_key;
3824                 struct extent_buffer *node;
3825
3826                 btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
3827                 level = root_item->drop_level;
3828                 path->lowest_level = level;
3829                 wret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
3830                 if (wret < 0) {
3831                         ret = wret;
3832                         goto out;
3833                 }
3834                 node = path->nodes[level];
3835                 btrfs_node_key(node, &found_key, path->slots[level]);
3836                 WARN_ON(memcmp(&found_key, &root_item->drop_progress,
3837                                sizeof(found_key)));
3838                 /*
3839                  * unlock our path, this is safe because only this
3840                  * function is allowed to delete this snapshot
3841                  */
3842                 for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
3843                         if (path->nodes[i] && path->locks[i]) {
3844                                 path->locks[i] = 0;
3845                                 btrfs_tree_unlock(path->nodes[i]);
3846                         }
3847                 }
3848         }
3849         while (1) {
3850                 wret = walk_down_tree(trans, root, path, &level);
3851                 if (wret > 0)
3852                         break;
3853                 if (wret < 0)
3854                         ret = wret;
3855
3856                 wret = walk_up_tree(trans, root, path, &level,
3857                                     BTRFS_MAX_LEVEL);
3858                 if (wret > 0)
3859                         break;
3860                 if (wret < 0)
3861                         ret = wret;
3862                 if (trans->transaction->in_commit) {
3863                         ret = -EAGAIN;
3864                         break;
3865                 }
3866                 atomic_inc(&root->fs_info->throttle_gen);
3867                 wake_up(&root->fs_info->transaction_throttle);
3868         }
3869         for (i = 0; i <= orig_level; i++) {
3870                 if (path->nodes[i]) {
3871                         free_extent_buffer(path->nodes[i]);
3872                         path->nodes[i] = NULL;
3873                 }
3874         }
3875 out:
3876         btrfs_free_path(path);
3877         return ret;
3878 }
3879
3880 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
3881                         struct btrfs_root *root,
3882                         struct extent_buffer *node,
3883                         struct extent_buffer *parent)
3884 {
3885         struct btrfs_path *path;
3886         int level;
3887         int parent_level;
3888         int ret = 0;
3889         int wret;
3890
3891         path = btrfs_alloc_path();
3892         BUG_ON(!path);
3893
3894         BUG_ON(!btrfs_tree_locked(parent));
3895         parent_level = btrfs_header_level(parent);
3896         extent_buffer_get(parent);
3897         path->nodes[parent_level] = parent;
3898         path->slots[parent_level] = btrfs_header_nritems(parent);
3899
3900         BUG_ON(!btrfs_tree_locked(node));
3901         level = btrfs_header_level(node);
3902         extent_buffer_get(node);
3903         path->nodes[level] = node;
3904         path->slots[level] = 0;
3905
3906         while (1) {
3907                 wret = walk_down_subtree(trans, root, path, &level);
3908                 if (wret < 0)
3909                         ret = wret;
3910                 if (wret != 0)
3911                         break;
3912
3913                 wret = walk_up_tree(trans, root, path, &level, parent_level);
3914                 if (wret < 0)
3915                         ret = wret;
3916                 if (wret != 0)
3917                         break;
3918         }
3919
3920         btrfs_free_path(path);
3921         return ret;
3922 }
3923
3924 static unsigned long calc_ra(unsigned long start, unsigned long last,
3925                              unsigned long nr)
3926 {
3927         return min(last, start + nr - 1);
3928 }
3929
3930 static noinline int relocate_inode_pages(struct inode *inode, u64 start,
3931                                          u64 len)
3932 {
3933         u64 page_start;
3934         u64 page_end;
3935         unsigned long first_index;
3936         unsigned long last_index;
3937         unsigned long i;
3938         struct page *page;
3939         struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
3940         struct file_ra_state *ra;
3941         struct btrfs_ordered_extent *ordered;
3942         unsigned int total_read = 0;
3943         unsigned int total_dirty = 0;
3944         int ret = 0;
3945
3946         ra = kzalloc(sizeof(*ra), GFP_NOFS);
3947
3948         mutex_lock(&inode->i_mutex);
3949         first_index = start >> PAGE_CACHE_SHIFT;
3950         last_index = (start + len - 1) >> PAGE_CACHE_SHIFT;
3951
3952         /* make sure the dirty trick played by the caller work */
3953         ret = invalidate_inode_pages2_range(inode->i_mapping,
3954                                             first_index, last_index);
3955         if (ret)
3956                 goto out_unlock;
3957
3958         file_ra_state_init(ra, inode->i_mapping);
3959
3960         for (i = first_index ; i <= last_index; i++) {
3961                 if (total_read % ra->ra_pages == 0) {
3962                         btrfs_force_ra(inode->i_mapping, ra, NULL, i,
3963                                        calc_ra(i, last_index, ra->ra_pages));
3964                 }
3965                 total_read++;
3966 again:
3967                 if (((u64)i << PAGE_CACHE_SHIFT) > i_size_read(inode))
3968                         BUG_ON(1);
3969                 page = grab_cache_page(inode->i_mapping, i);
3970                 if (!page) {
3971                         ret = -ENOMEM;
3972                         goto out_unlock;
3973                 }
3974                 if (!PageUptodate(page)) {
3975                         btrfs_readpage(NULL, page);
3976                         lock_page(page);
3977                         if (!PageUptodate(page)) {
3978                                 unlock_page(page);
3979                                 page_cache_release(page);
3980                                 ret = -EIO;
3981                                 goto out_unlock;
3982                         }
3983                 }
3984                 wait_on_page_writeback(page);
3985
3986                 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
3987                 page_end = page_start + PAGE_CACHE_SIZE - 1;
3988                 lock_extent(io_tree, page_start, page_end, GFP_NOFS);
3989
3990                 ordered = btrfs_lookup_ordered_extent(inode, page_start);
3991                 if (ordered) {
3992                         unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
3993                         unlock_page(page);
3994                         page_cache_release(page);
3995                         btrfs_start_ordered_extent(inode, ordered, 1);
3996                         btrfs_put_ordered_extent(ordered);
3997                         goto again;
3998                 }
3999                 set_page_extent_mapped(page);
4000
4001                 if (i == first_index)
4002                         set_extent_bits(io_tree, page_start, page_end,
4003                                         EXTENT_BOUNDARY, GFP_NOFS);
4004                 btrfs_set_extent_delalloc(inode, page_start, page_end);
4005
4006                 set_page_dirty(page);
4007                 total_dirty++;
4008
4009                 unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
4010                 unlock_page(page);
4011                 page_cache_release(page);
4012         }
4013
4014 out_unlock:
4015         kfree(ra);
4016         mutex_unlock(&inode->i_mutex);
4017         balance_dirty_pages_ratelimited_nr(inode->i_mapping, total_dirty);
4018         return ret;
4019 }
4020
4021 static noinline int relocate_data_extent(struct inode *reloc_inode,
4022                                          struct btrfs_key *extent_key,
4023                                          u64 offset)
4024 {
4025         struct btrfs_root *root = BTRFS_I(reloc_inode)->root;
4026         struct extent_map_tree *em_tree = &BTRFS_I(reloc_inode)->extent_tree;
4027         struct extent_map *em;
4028         u64 start = extent_key->objectid - offset;
4029         u64 end = start + extent_key->offset - 1;
4030
4031         em = alloc_extent_map(GFP_NOFS);
4032         BUG_ON(!em || IS_ERR(em));
4033
4034         em->start = start;
4035         em->len = extent_key->offset;
4036         em->block_len = extent_key->offset;
4037         em->block_start = extent_key->objectid;
4038         em->bdev = root->fs_info->fs_devices->latest_bdev;
4039         set_bit(EXTENT_FLAG_PINNED, &em->flags);
4040
4041         /* setup extent map to cheat btrfs_readpage */
4042         lock_extent(&BTRFS_I(reloc_inode)->io_tree, start, end, GFP_NOFS);
4043         while (1) {
4044                 int ret;
4045                 spin_lock(&em_tree->lock);
4046                 ret = add_extent_mapping(em_tree, em);
4047                 spin_unlock(&em_tree->lock);
4048                 if (ret != -EEXIST) {
4049                         free_extent_map(em);
4050                         break;
4051                 }
4052                 btrfs_drop_extent_cache(reloc_inode, start, end, 0);
4053         }
4054         unlock_extent(&BTRFS_I(reloc_inode)->io_tree, start, end, GFP_NOFS);
4055
4056         return relocate_inode_pages(reloc_inode, start, extent_key->offset);
4057 }
4058
4059 struct btrfs_ref_path {
4060         u64 extent_start;
4061         u64 nodes[BTRFS_MAX_LEVEL];
4062         u64 root_objectid;
4063         u64 root_generation;
4064         u64 owner_objectid;
4065         u32 num_refs;
4066         int lowest_level;
4067         int current_level;
4068         int shared_level;
4069
4070         struct btrfs_key node_keys[BTRFS_MAX_LEVEL];
4071         u64 new_nodes[BTRFS_MAX_LEVEL];
4072 };
4073
4074 struct disk_extent {
4075         u64 ram_bytes;
4076         u64 disk_bytenr;
4077         u64 disk_num_bytes;
4078         u64 offset;
4079         u64 num_bytes;
4080         u8 compression;
4081         u8 encryption;
4082         u16 other_encoding;
4083 };
4084
4085 static int is_cowonly_root(u64 root_objectid)
4086 {
4087         if (root_objectid == BTRFS_ROOT_TREE_OBJECTID ||
4088             root_objectid == BTRFS_EXTENT_TREE_OBJECTID ||
4089             root_objectid == BTRFS_CHUNK_TREE_OBJECTID ||
4090             root_objectid == BTRFS_DEV_TREE_OBJECTID ||
4091             root_objectid == BTRFS_TREE_LOG_OBJECTID ||
4092             root_objectid == BTRFS_CSUM_TREE_OBJECTID)
4093                 return 1;
4094         return 0;
4095 }
4096
4097 static noinline int __next_ref_path(struct btrfs_trans_handle *trans,
4098                                     struct btrfs_root *extent_root,
4099                                     struct btrfs_ref_path *ref_path,
4100                                     int first_time)
4101 {
4102         struct extent_buffer *leaf;
4103         struct btrfs_path *path;
4104         struct btrfs_extent_ref *ref;
4105         struct btrfs_key key;
4106         struct btrfs_key found_key;
4107         u64 bytenr;
4108         u32 nritems;
4109         int level;
4110         int ret = 1;
4111
4112         path = btrfs_alloc_path();
4113         if (!path)
4114                 return -ENOMEM;
4115
4116         if (first_time) {
4117                 ref_path->lowest_level = -1;
4118                 ref_path->current_level = -1;
4119                 ref_path->shared_level = -1;
4120                 goto walk_up;
4121         }
4122 walk_down:
4123         level = ref_path->current_level - 1;
4124         while (level >= -1) {
4125                 u64 parent;
4126                 if (level < ref_path->lowest_level)
4127                         break;
4128
4129                 if (level >= 0)
4130                         bytenr = ref_path->nodes[level];
4131                 else
4132                         bytenr = ref_path->extent_start;
4133                 BUG_ON(bytenr == 0);
4134
4135                 parent = ref_path->nodes[level + 1];
4136                 ref_path->nodes[level + 1] = 0;
4137                 ref_path->current_level = level;
4138                 BUG_ON(parent == 0);
4139
4140                 key.objectid = bytenr;
4141                 key.offset = parent + 1;
4142                 key.type = BTRFS_EXTENT_REF_KEY;
4143
4144                 ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0);
4145                 if (ret < 0)
4146                         goto out;
4147                 BUG_ON(ret == 0);
4148
4149                 leaf = path->nodes[0];
4150                 nritems = btrfs_header_nritems(leaf);
4151                 if (path->slots[0] >= nritems) {
4152                         ret = btrfs_next_leaf(extent_root, path);
4153                         if (ret < 0)
4154                                 goto out;
4155                         if (ret > 0)
4156                                 goto next;
4157                         leaf = path->nodes[0];
4158                 }
4159
4160                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4161                 if (found_key.objectid == bytenr &&
4162                     found_key.type == BTRFS_EXTENT_REF_KEY) {
4163                         if (level < ref_path->shared_level)
4164                                 ref_path->shared_level = level;
4165                         goto found;
4166                 }
4167 next:
4168                 level--;
4169                 btrfs_release_path(extent_root, path);
4170                 cond_resched();
4171         }
4172         /* reached lowest level */
4173         ret = 1;
4174         goto out;
4175 walk_up:
4176         level = ref_path->current_level;
4177         while (level < BTRFS_MAX_LEVEL - 1) {
4178                 u64 ref_objectid;
4179
4180                 if (level >= 0)
4181                         bytenr = ref_path->nodes[level];
4182                 else
4183                         bytenr = ref_path->extent_start;
4184
4185                 BUG_ON(bytenr == 0);
4186
4187                 key.objectid = bytenr;
4188                 key.offset = 0;
4189                 key.type = BTRFS_EXTENT_REF_KEY;
4190
4191                 ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0);
4192                 if (ret < 0)
4193                         goto out;
4194
4195                 leaf = path->nodes[0];
4196                 nritems = btrfs_header_nritems(leaf);
4197                 if (path->slots[0] >= nritems) {
4198                         ret = btrfs_next_leaf(extent_root, path);
4199                         if (ret < 0)
4200                                 goto out;
4201                         if (ret > 0) {
4202                                 /* the extent was freed by someone */
4203                                 if (ref_path->lowest_level == level)
4204                                         goto out;
4205                                 btrfs_release_path(extent_root, path);
4206                                 goto walk_down;
4207                         }
4208                         leaf = path->nodes[0];
4209                 }
4210
4211                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4212                 if (found_key.objectid != bytenr ||
4213                                 found_key.type != BTRFS_EXTENT_REF_KEY) {
4214                         /* the extent was freed by someone */
4215                         if (ref_path->lowest_level == level) {
4216                                 ret = 1;
4217                                 goto out;
4218                         }
4219                         btrfs_release_path(extent_root, path);
4220                         goto walk_down;
4221                 }
4222 found:
4223                 ref = btrfs_item_ptr(leaf, path->slots[0],
4224                                 struct btrfs_extent_ref);
4225                 ref_objectid = btrfs_ref_objectid(leaf, ref);
4226                 if (ref_objectid < BTRFS_FIRST_FREE_OBJECTID) {
4227                         if (first_time) {
4228                                 level = (int)ref_objectid;
4229                                 BUG_ON(level >= BTRFS_MAX_LEVEL);
4230                                 ref_path->lowest_level = level;
4231                                 ref_path->current_level = level;
4232                                 ref_path->nodes[level] = bytenr;
4233                         } else {
4234                                 WARN_ON(ref_objectid != level);
4235                         }
4236                 } else {
4237                         WARN_ON(level != -1);
4238                 }
4239                 first_time = 0;
4240
4241                 if (ref_path->lowest_level == level) {
4242                         ref_path->owner_objectid = ref_objectid;
4243                         ref_path->num_refs = btrfs_ref_num_refs(leaf, ref);
4244                 }
4245
4246                 /*
4247                  * the block is tree root or the block isn't in reference
4248                  * counted tree.
4249                  */
4250                 if (found_key.objectid == found_key.offset ||
4251                     is_cowonly_root(btrfs_ref_root(leaf, ref))) {
4252                         ref_path->root_objectid = btrfs_ref_root(leaf, ref);
4253                         ref_path->root_generation =
4254                                 btrfs_ref_generation(leaf, ref);
4255                         if (level < 0) {
4256                                 /* special reference from the tree log */
4257                                 ref_path->nodes[0] = found_key.offset;
4258                                 ref_path->current_level = 0;
4259                         }
4260                         ret = 0;
4261                         goto out;
4262                 }
4263
4264                 level++;
4265                 BUG_ON(ref_path->nodes[level] != 0);
4266                 ref_path->nodes[level] = found_key.offset;
4267                 ref_path->current_level = level;
4268
4269                 /*
4270                  * the reference was created in the running transaction,
4271                  * no need to continue walking up.
4272                  */
4273                 if (btrfs_ref_generation(leaf, ref) == trans->transid) {
4274                         ref_path->root_objectid = btrfs_ref_root(leaf, ref);
4275                         ref_path->root_generation =
4276                                 btrfs_ref_generation(leaf, ref);
4277                         ret = 0;
4278                         goto out;
4279                 }
4280
4281                 btrfs_release_path(extent_root, path);
4282                 cond_resched();
4283         }
4284         /* reached max tree level, but no tree root found. */
4285         BUG();
4286 out:
4287         btrfs_free_path(path);
4288         return ret;
4289 }
4290
4291 static int btrfs_first_ref_path(struct btrfs_trans_handle *trans,
4292                                 struct btrfs_root *extent_root,
4293                                 struct btrfs_ref_path *ref_path,
4294                                 u64 extent_start)
4295 {
4296         memset(ref_path, 0, sizeof(*ref_path));
4297         ref_path->extent_start = extent_start;
4298
4299         return __next_ref_path(trans, extent_root, ref_path, 1);
4300 }
4301
4302 static int btrfs_next_ref_path(struct btrfs_trans_handle *trans,
4303                                struct btrfs_root *extent_root,
4304                                struct btrfs_ref_path *ref_path)
4305 {
4306         return __next_ref_path(trans, extent_root, ref_path, 0);
4307 }
4308
4309 static noinline int get_new_locations(struct inode *reloc_inode,
4310                                       struct btrfs_key *extent_key,
4311                                       u64 offset, int no_fragment,
4312                                       struct disk_extent **extents,
4313                                       int *nr_extents)
4314 {
4315         struct btrfs_root *root = BTRFS_I(reloc_inode)->root;
4316         struct btrfs_path *path;
4317         struct btrfs_file_extent_item *fi;
4318         struct extent_buffer *leaf;
4319         struct disk_extent *exts = *extents;
4320         struct btrfs_key found_key;
4321         u64 cur_pos;
4322         u64 last_byte;
4323         u32 nritems;
4324         int nr = 0;
4325         int max = *nr_extents;
4326         int ret;
4327
4328         WARN_ON(!no_fragment && *extents);
4329         if (!exts) {
4330                 max = 1;
4331                 exts = kmalloc(sizeof(*exts) * max, GFP_NOFS);
4332                 if (!exts)
4333                         return -ENOMEM;
4334         }
4335
4336         path = btrfs_alloc_path();
4337         BUG_ON(!path);
4338
4339         cur_pos = extent_key->objectid - offset;
4340         last_byte = extent_key->objectid + extent_key->offset;
4341         ret = btrfs_lookup_file_extent(NULL, root, path, reloc_inode->i_ino,
4342                                        cur_pos, 0);
4343         if (ret < 0)
4344                 goto out;
4345         if (ret > 0) {
4346                 ret = -ENOENT;
4347                 goto out;
4348         }
4349
4350         while (1) {
4351                 leaf = path->nodes[0];
4352                 nritems = btrfs_header_nritems(leaf);
4353                 if (path->slots[0] >= nritems) {
4354                         ret = btrfs_next_leaf(root, path);
4355                         if (ret < 0)
4356                                 goto out;
4357                         if (ret > 0)
4358                                 break;
4359                         leaf = path->nodes[0];
4360                 }
4361
4362                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4363                 if (found_key.offset != cur_pos ||
4364                     found_key.type != BTRFS_EXTENT_DATA_KEY ||
4365                     found_key.objectid != reloc_inode->i_ino)
4366                         break;
4367
4368                 fi = btrfs_item_ptr(leaf, path->slots[0],
4369                                     struct btrfs_file_extent_item);
4370                 if (btrfs_file_extent_type(leaf, fi) !=
4371                     BTRFS_FILE_EXTENT_REG ||
4372                     btrfs_file_extent_disk_bytenr(leaf, fi) == 0)
4373                         break;
4374
4375                 if (nr == max) {
4376                         struct disk_extent *old = exts;
4377                         max *= 2;
4378                         exts = kzalloc(sizeof(*exts) * max, GFP_NOFS);
4379                         memcpy(exts, old, sizeof(*exts) * nr);
4380                         if (old != *extents)
4381                                 kfree(old);
4382                 }
4383
4384                 exts[nr].disk_bytenr =
4385                         btrfs_file_extent_disk_bytenr(leaf, fi);
4386                 exts[nr].disk_num_bytes =
4387                         btrfs_file_extent_disk_num_bytes(leaf, fi);
4388                 exts[nr].offset = btrfs_file_extent_offset(leaf, fi);
4389                 exts[nr].num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
4390                 exts[nr].ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi);
4391                 exts[nr].compression = btrfs_file_extent_compression(leaf, fi);
4392                 exts[nr].encryption = btrfs_file_extent_encryption(leaf, fi);
4393                 exts[nr].other_encoding = btrfs_file_extent_other_encoding(leaf,
4394                                                                            fi);
4395                 BUG_ON(exts[nr].offset > 0);
4396                 BUG_ON(exts[nr].compression || exts[nr].encryption);
4397                 BUG_ON(exts[nr].num_bytes != exts[nr].disk_num_bytes);
4398
4399                 cur_pos += exts[nr].num_bytes;
4400                 nr++;
4401
4402                 if (cur_pos + offset >= last_byte)
4403                         break;
4404
4405                 if (no_fragment) {
4406                         ret = 1;
4407                         goto out;
4408                 }
4409                 path->slots[0]++;
4410         }
4411
4412         BUG_ON(cur_pos + offset > last_byte);
4413         if (cur_pos + offset < last_byte) {
4414                 ret = -ENOENT;
4415                 goto out;
4416         }
4417         ret = 0;
4418 out:
4419         btrfs_free_path(path);
4420         if (ret) {
4421                 if (exts != *extents)
4422                         kfree(exts);
4423         } else {
4424                 *extents = exts;
4425                 *nr_extents = nr;
4426         }
4427         return ret;
4428 }
4429
4430 static noinline int replace_one_extent(struct btrfs_trans_handle *trans,
4431                                         struct btrfs_root *root,
4432                                         struct btrfs_path *path,
4433                                         struct btrfs_key *extent_key,
4434                                         struct btrfs_key *leaf_key,
4435                                         struct btrfs_ref_path *ref_path,
4436                                         struct disk_extent *new_extents,
4437                                         int nr_extents)
4438 {
4439         struct extent_buffer *leaf;
4440         struct btrfs_file_extent_item *fi;
4441         struct inode *inode = NULL;
4442         struct btrfs_key key;
4443         u64 lock_start = 0;
4444         u64 lock_end = 0;
4445         u64 num_bytes;
4446         u64 ext_offset;
4447         u64 first_pos;
4448         u32 nritems;
4449         int nr_scaned = 0;
4450         int extent_locked = 0;
4451         int extent_type;
4452         int ret;
4453
4454         memcpy(&key, leaf_key, sizeof(key));
4455         first_pos = INT_LIMIT(loff_t) - extent_key->offset;
4456         if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) {
4457                 if (key.objectid < ref_path->owner_objectid ||
4458                     (key.objectid == ref_path->owner_objectid &&
4459                      key.type < BTRFS_EXTENT_DATA_KEY)) {
4460                         key.objectid = ref_path->owner_objectid;
4461                         key.type = BTRFS_EXTENT_DATA_KEY;
4462                         key.offset = 0;
4463                 }
4464         }
4465
4466         while (1) {
4467                 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
4468                 if (ret < 0)
4469                         goto out;
4470
4471                 leaf = path->nodes[0];
4472                 nritems = btrfs_header_nritems(leaf);
4473 next:
4474                 if (extent_locked && ret > 0) {
4475                         /*
4476                          * the file extent item was modified by someone
4477                          * before the extent got locked.
4478                          */
4479                         unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4480                                       lock_end, GFP_NOFS);
4481                         extent_locked = 0;
4482                 }
4483
4484                 if (path->slots[0] >= nritems) {
4485                         if (++nr_scaned > 2)
4486                                 break;
4487
4488                         BUG_ON(extent_locked);
4489                         ret = btrfs_next_leaf(root, path);
4490                         if (ret < 0)
4491                                 goto out;
4492                         if (ret > 0)
4493                                 break;
4494                         leaf = path->nodes[0];
4495                         nritems = btrfs_header_nritems(leaf);
4496                 }
4497
4498                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
4499
4500                 if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) {
4501                         if ((key.objectid > ref_path->owner_objectid) ||
4502                             (key.objectid == ref_path->owner_objectid &&
4503                              key.type > BTRFS_EXTENT_DATA_KEY) ||
4504                             (key.offset >= first_pos + extent_key->offset))
4505                                 break;
4506                 }
4507
4508                 if (inode && key.objectid != inode->i_ino) {
4509                         BUG_ON(extent_locked);
4510                         btrfs_release_path(root, path);
4511                         mutex_unlock(&inode->i_mutex);
4512                         iput(inode);
4513                         inode = NULL;
4514                         continue;
4515                 }
4516
4517                 if (key.type != BTRFS_EXTENT_DATA_KEY) {
4518                         path->slots[0]++;
4519                         ret = 1;
4520                         goto next;
4521                 }
4522                 fi = btrfs_item_ptr(leaf, path->slots[0],
4523                                     struct btrfs_file_extent_item);
4524                 extent_type = btrfs_file_extent_type(leaf, fi);
4525                 if ((extent_type != BTRFS_FILE_EXTENT_REG &&
4526                      extent_type != BTRFS_FILE_EXTENT_PREALLOC) ||
4527                     (btrfs_file_extent_disk_bytenr(leaf, fi) !=
4528                      extent_key->objectid)) {
4529                         path->slots[0]++;
4530                         ret = 1;
4531                         goto next;
4532                 }
4533
4534                 num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
4535                 ext_offset = btrfs_file_extent_offset(leaf, fi);
4536
4537                 if (first_pos > key.offset - ext_offset)
4538                         first_pos = key.offset - ext_offset;
4539
4540                 if (!extent_locked) {
4541                         lock_start = key.offset;
4542                         lock_end = lock_start + num_bytes - 1;
4543                 } else {
4544                         if (lock_start > key.offset ||
4545                             lock_end + 1 < key.offset + num_bytes) {
4546                                 unlock_extent(&BTRFS_I(inode)->io_tree,
4547                                               lock_start, lock_end, GFP_NOFS);
4548                                 extent_locked = 0;
4549                         }
4550                 }
4551
4552                 if (!inode) {
4553                         btrfs_release_path(root, path);
4554
4555                         inode = btrfs_iget_locked(root->fs_info->sb,
4556                                                   key.objectid, root);
4557                         if (inode->i_state & I_NEW) {
4558                                 BTRFS_I(inode)->root = root;
4559                                 BTRFS_I(inode)->location.objectid =
4560                                         key.objectid;
4561                                 BTRFS_I(inode)->location.type =
4562                                         BTRFS_INODE_ITEM_KEY;
4563                                 BTRFS_I(inode)->location.offset = 0;
4564                                 btrfs_read_locked_inode(inode);
4565                                 unlock_new_inode(inode);
4566                         }
4567                         /*
4568                          * some code call btrfs_commit_transaction while
4569                          * holding the i_mutex, so we can't use mutex_lock
4570                          * here.
4571                          */
4572                         if (is_bad_inode(inode) ||
4573                             !mutex_trylock(&inode->i_mutex)) {
4574                                 iput(inode);
4575                                 inode = NULL;
4576                                 key.offset = (u64)-1;
4577                                 goto skip;
4578                         }
4579                 }
4580
4581                 if (!extent_locked) {
4582                         struct btrfs_ordered_extent *ordered;
4583
4584                         btrfs_release_path(root, path);
4585
4586                         lock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4587                                     lock_end, GFP_NOFS);
4588                         ordered = btrfs_lookup_first_ordered_extent(inode,
4589                                                                     lock_end);
4590                         if (ordered &&
4591                             ordered->file_offset <= lock_end &&
4592                             ordered->file_offset + ordered->len > lock_start) {
4593                                 unlock_extent(&BTRFS_I(inode)->io_tree,
4594                                               lock_start, lock_end, GFP_NOFS);
4595                                 btrfs_start_ordered_extent(inode, ordered, 1);
4596                                 btrfs_put_ordered_extent(ordered);
4597                                 key.offset += num_bytes;
4598                                 goto skip;
4599                         }
4600                         if (ordered)
4601                                 btrfs_put_ordered_extent(ordered);
4602
4603                         extent_locked = 1;
4604                         continue;
4605                 }
4606
4607                 if (nr_extents == 1) {
4608                         /* update extent pointer in place */
4609                         btrfs_set_file_extent_disk_bytenr(leaf, fi,
4610                                                 new_extents[0].disk_bytenr);
4611                         btrfs_set_file_extent_disk_num_bytes(leaf, fi,
4612                                                 new_extents[0].disk_num_bytes);
4613                         btrfs_mark_buffer_dirty(leaf);
4614
4615                         btrfs_drop_extent_cache(inode, key.offset,
4616                                                 key.offset + num_bytes - 1, 0);
4617
4618                         ret = btrfs_inc_extent_ref(trans, root,
4619                                                 new_extents[0].disk_bytenr,
4620                                                 new_extents[0].disk_num_bytes,
4621                                                 leaf->start,
4622                                                 root->root_key.objectid,
4623                                                 trans->transid,
4624                                                 key.objectid);
4625                         BUG_ON(ret);
4626
4627                         ret = btrfs_free_extent(trans, root,
4628                                                 extent_key->objectid,
4629                                                 extent_key->offset,
4630                                                 leaf->start,
4631                                                 btrfs_header_owner(leaf),
4632                                                 btrfs_header_generation(leaf),
4633                                                 key.objectid, 0);
4634                         BUG_ON(ret);
4635
4636                         btrfs_release_path(root, path);
4637                         key.offset += num_bytes;
4638                 } else {
4639                         BUG_ON(1);
4640 #if 0
4641                         u64 alloc_hint;
4642                         u64 extent_len;
4643                         int i;
4644                         /*
4645                          * drop old extent pointer at first, then insert the
4646                          * new pointers one bye one
4647                          */
4648                         btrfs_release_path(root, path);
4649                         ret = btrfs_drop_extents(trans, root, inode, key.offset,
4650                                                  key.offset + num_bytes,
4651                                                  key.offset, &alloc_hint);
4652                         BUG_ON(ret);
4653
4654                         for (i = 0; i < nr_extents; i++) {
4655                                 if (ext_offset >= new_extents[i].num_bytes) {
4656                                         ext_offset -= new_extents[i].num_bytes;
4657                                         continue;
4658                                 }
4659                                 extent_len = min(new_extents[i].num_bytes -
4660                                                  ext_offset, num_bytes);
4661
4662                                 ret = btrfs_insert_empty_item(trans, root,
4663                                                               path, &key,
4664                                                               sizeof(*fi));
4665                                 BUG_ON(ret);
4666
4667                                 leaf = path->nodes[0];
4668                                 fi = btrfs_item_ptr(leaf, path->slots[0],
4669                                                 struct btrfs_file_extent_item);
4670                                 btrfs_set_file_extent_generation(leaf, fi,
4671                                                         trans->transid);
4672                                 btrfs_set_file_extent_type(leaf, fi,
4673                                                         BTRFS_FILE_EXTENT_REG);
4674                                 btrfs_set_file_extent_disk_bytenr(leaf, fi,
4675                                                 new_extents[i].disk_bytenr);
4676                                 btrfs_set_file_extent_disk_num_bytes(leaf, fi,
4677                                                 new_extents[i].disk_num_bytes);
4678                                 btrfs_set_file_extent_ram_bytes(leaf, fi,
4679                                                 new_extents[i].ram_bytes);
4680
4681                                 btrfs_set_file_extent_compression(leaf, fi,
4682                                                 new_extents[i].compression);
4683                                 btrfs_set_file_extent_encryption(leaf, fi,
4684                                                 new_extents[i].encryption);
4685                                 btrfs_set_file_extent_other_encoding(leaf, fi,
4686                                                 new_extents[i].other_encoding);
4687
4688                                 btrfs_set_file_extent_num_bytes(leaf, fi,
4689                                                         extent_len);
4690                                 ext_offset += new_extents[i].offset;
4691                                 btrfs_set_file_extent_offset(leaf, fi,
4692                                                         ext_offset);
4693                                 btrfs_mark_buffer_dirty(leaf);
4694
4695                                 btrfs_drop_extent_cache(inode, key.offset,
4696                                                 key.offset + extent_len - 1, 0);
4697
4698                                 ret = btrfs_inc_extent_ref(trans, root,
4699                                                 new_extents[i].disk_bytenr,
4700                                                 new_extents[i].disk_num_bytes,
4701                                                 leaf->start,
4702                                                 root->root_key.objectid,
4703                                                 trans->transid, key.objectid);
4704                                 BUG_ON(ret);
4705                                 btrfs_release_path(root, path);
4706
4707                                 inode_add_bytes(inode, extent_len);
4708
4709                                 ext_offset = 0;
4710                                 num_bytes -= extent_len;
4711                                 key.offset += extent_len;
4712
4713                                 if (num_bytes == 0)
4714                                         break;
4715                         }
4716                         BUG_ON(i >= nr_extents);
4717 #endif
4718                 }
4719
4720                 if (extent_locked) {
4721                         unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4722                                       lock_end, GFP_NOFS);
4723                         extent_locked = 0;
4724                 }
4725 skip:
4726                 if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS &&
4727                     key.offset >= first_pos + extent_key->offset)
4728                         break;
4729
4730                 cond_resched();
4731         }
4732         ret = 0;
4733 out:
4734         btrfs_release_path(root, path);
4735         if (inode) {
4736                 mutex_unlock(&inode->i_mutex);
4737                 if (extent_locked) {
4738                         unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4739                                       lock_end, GFP_NOFS);
4740                 }
4741                 iput(inode);
4742         }
4743         return ret;
4744 }
4745
4746 int btrfs_reloc_tree_cache_ref(struct btrfs_trans_handle *trans,
4747                                struct btrfs_root *root,
4748                                struct extent_buffer *buf, u64 orig_start)
4749 {
4750         int level;
4751         int ret;
4752
4753         BUG_ON(btrfs_header_generation(buf) != trans->transid);
4754         BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
4755
4756         level = btrfs_header_level(buf);
4757         if (level == 0) {
4758                 struct btrfs_leaf_ref *ref;
4759                 struct btrfs_leaf_ref *orig_ref;
4760
4761                 orig_ref = btrfs_lookup_leaf_ref(root, orig_start);
4762                 if (!orig_ref)
4763                         return -ENOENT;
4764
4765                 ref = btrfs_alloc_leaf_ref(root, orig_ref->nritems);
4766                 if (!ref) {
4767                         btrfs_free_leaf_ref(root, orig_ref);
4768                         return -ENOMEM;
4769                 }
4770
4771                 ref->nritems = orig_ref->nritems;
4772                 memcpy(ref->extents, orig_ref->extents,
4773                         sizeof(ref->extents[0]) * ref->nritems);
4774
4775                 btrfs_free_leaf_ref(root, orig_ref);
4776
4777                 ref->root_gen = trans->transid;
4778                 ref->bytenr = buf->start;
4779                 ref->owner = btrfs_header_owner(buf);
4780                 ref->generation = btrfs_header_generation(buf);
4781                 ret = btrfs_add_leaf_ref(root, ref, 0);
4782                 WARN_ON(ret);
4783                 btrfs_free_leaf_ref(root, ref);
4784         }
4785         return 0;
4786 }
4787
4788 static noinline int invalidate_extent_cache(struct btrfs_root *root,
4789                                         struct extent_buffer *leaf,
4790                                         struct btrfs_block_group_cache *group,
4791                                         struct btrfs_root *target_root)
4792 {
4793         struct btrfs_key key;
4794         struct inode *inode = NULL;
4795         struct btrfs_file_extent_item *fi;
4796         u64 num_bytes;
4797         u64 skip_objectid = 0;
4798         u32 nritems;
4799         u32 i;
4800
4801         nritems = btrfs_header_nritems(leaf);
4802         for (i = 0; i < nritems; i++) {
4803                 btrfs_item_key_to_cpu(leaf, &key, i);
4804                 if (key.objectid == skip_objectid ||
4805                     key.type != BTRFS_EXTENT_DATA_KEY)
4806                         continue;
4807                 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
4808                 if (btrfs_file_extent_type(leaf, fi) ==
4809                     BTRFS_FILE_EXTENT_INLINE)
4810                         continue;
4811                 if (btrfs_file_extent_disk_bytenr(leaf, fi) == 0)
4812                         continue;
4813                 if (!inode || inode->i_ino != key.objectid) {
4814                         iput(inode);
4815                         inode = btrfs_ilookup(target_root->fs_info->sb,
4816                                               key.objectid, target_root, 1);
4817                 }
4818                 if (!inode) {
4819                         skip_objectid = key.objectid;
4820                         continue;
4821                 }
4822                 num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
4823
4824                 lock_extent(&BTRFS_I(inode)->io_tree, key.offset,
4825                             key.offset + num_bytes - 1, GFP_NOFS);
4826                 btrfs_drop_extent_cache(inode, key.offset,
4827                                         key.offset + num_bytes - 1, 1);
4828                 unlock_extent(&BTRFS_I(inode)->io_tree, key.offset,
4829                               key.offset + num_bytes - 1, GFP_NOFS);
4830                 cond_resched();
4831         }
4832         iput(inode);
4833         return 0;
4834 }
4835
4836 static noinline int replace_extents_in_leaf(struct btrfs_trans_handle *trans,
4837                                         struct btrfs_root *root,
4838                                         struct extent_buffer *leaf,
4839                                         struct btrfs_block_group_cache *group,
4840                                         struct inode *reloc_inode)
4841 {
4842         struct btrfs_key key;
4843         struct btrfs_key extent_key;
4844         struct btrfs_file_extent_item *fi;
4845         struct btrfs_leaf_ref *ref;
4846         struct disk_extent *new_extent;
4847         u64 bytenr;
4848         u64 num_bytes;
4849         u32 nritems;
4850         u32 i;
4851         int ext_index;
4852         int nr_extent;
4853         int ret;
4854
4855         new_extent = kmalloc(sizeof(*new_extent), GFP_NOFS);
4856         BUG_ON(!new_extent);
4857
4858         ref = btrfs_lookup_leaf_ref(root, leaf->start);
4859         BUG_ON(!ref);
4860
4861         ext_index = -1;
4862         nritems = btrfs_header_nritems(leaf);
4863         for (i = 0; i < nritems; i++) {
4864                 btrfs_item_key_to_cpu(leaf, &key, i);
4865                 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
4866                         continue;
4867                 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
4868                 if (btrfs_file_extent_type(leaf, fi) ==
4869                     BTRFS_FILE_EXTENT_INLINE)
4870                         continue;
4871                 bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
4872                 num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
4873                 if (bytenr == 0)
4874                         continue;
4875
4876                 ext_index++;
4877                 if (bytenr >= group->key.objectid + group->key.offset ||
4878                     bytenr + num_bytes <= group->key.objectid)
4879                         continue;
4880
4881                 extent_key.objectid = bytenr;
4882                 extent_key.offset = num_bytes;
4883                 extent_key.type = BTRFS_EXTENT_ITEM_KEY;
4884                 nr_extent = 1;
4885                 ret = get_new_locations(reloc_inode, &extent_key,
4886                                         group->key.objectid, 1,
4887                                         &new_extent, &nr_extent);
4888                 if (ret > 0)
4889                         continue;
4890                 BUG_ON(ret < 0);
4891
4892                 BUG_ON(ref->extents[ext_index].bytenr != bytenr);
4893                 BUG_ON(ref->extents[ext_index].num_bytes != num_bytes);
4894                 ref->extents[ext_index].bytenr = new_extent->disk_bytenr;
4895                 ref->extents[ext_index].num_bytes = new_extent->disk_num_bytes;
4896
4897                 btrfs_set_file_extent_disk_bytenr(leaf, fi,
4898                                                 new_extent->disk_bytenr);
4899                 btrfs_set_file_extent_disk_num_bytes(leaf, fi,
4900                                                 new_extent->disk_num_bytes);
4901                 btrfs_mark_buffer_dirty(leaf);
4902
4903                 ret = btrfs_inc_extent_ref(trans, root,
4904                                         new_extent->disk_bytenr,
4905                                         new_extent->disk_num_bytes,
4906                                         leaf->start,
4907                                         root->root_key.objectid,
4908                                         trans->transid, key.objectid);
4909                 BUG_ON(ret);
4910                 ret = btrfs_free_extent(trans, root,
4911                                         bytenr, num_bytes, leaf->start,
4912                                         btrfs_header_owner(leaf),
4913                                         btrfs_header_generation(leaf),
4914                                         key.objectid, 0);
4915                 BUG_ON(ret);
4916                 cond_resched();
4917         }
4918         kfree(new_extent);
4919         BUG_ON(ext_index + 1 != ref->nritems);
4920         btrfs_free_leaf_ref(root, ref);
4921         return 0;
4922 }
4923
4924 int btrfs_free_reloc_root(struct btrfs_trans_handle *trans,
4925                           struct btrfs_root *root)
4926 {
4927         struct btrfs_root *reloc_root;
4928         int ret;
4929
4930         if (root->reloc_root) {
4931                 reloc_root = root->reloc_root;
4932                 root->reloc_root = NULL;
4933                 list_add(&reloc_root->dead_list,
4934                          &root->fs_info->dead_reloc_roots);
4935
4936                 btrfs_set_root_bytenr(&reloc_root->root_item,
4937                                       reloc_root->node->start);
4938                 btrfs_set_root_level(&root->root_item,
4939                                      btrfs_header_level(reloc_root->node));
4940                 memset(&reloc_root->root_item.drop_progress, 0,
4941                         sizeof(struct btrfs_disk_key));
4942                 reloc_root->root_item.drop_level = 0;
4943
4944                 ret = btrfs_update_root(trans, root->fs_info->tree_root,
4945                                         &reloc_root->root_key,
4946                                         &reloc_root->root_item);
4947                 BUG_ON(ret);
4948         }
4949         return 0;
4950 }
4951
4952 int btrfs_drop_dead_reloc_roots(struct btrfs_root *root)
4953 {
4954         struct btrfs_trans_handle *trans;
4955         struct btrfs_root *reloc_root;
4956         struct btrfs_root *prev_root = NULL;
4957         struct list_head dead_roots;
4958         int ret;
4959         unsigned long nr;
4960
4961         INIT_LIST_HEAD(&dead_roots);
4962         list_splice_init(&root->fs_info->dead_reloc_roots, &dead_roots);
4963
4964         while (!list_empty(&dead_roots)) {
4965                 reloc_root = list_entry(dead_roots.prev,
4966                                         struct btrfs_root, dead_list);
4967                 list_del_init(&reloc_root->dead_list);
4968
4969                 BUG_ON(reloc_root->commit_root != NULL);
4970                 while (1) {
4971                         trans = btrfs_join_transaction(root, 1);
4972                         BUG_ON(!trans);
4973
4974                         mutex_lock(&root->fs_info->drop_mutex);
4975                         ret = btrfs_drop_snapshot(trans, reloc_root);
4976                         if (ret != -EAGAIN)
4977                                 break;
4978                         mutex_unlock(&root->fs_info->drop_mutex);
4979
4980                         nr = trans->blocks_used;
4981                         ret = btrfs_end_transaction(trans, root);
4982                         BUG_ON(ret);
4983                         btrfs_btree_balance_dirty(root, nr);
4984                 }
4985
4986                 free_extent_buffer(reloc_root->node);
4987
4988                 ret = btrfs_del_root(trans, root->fs_info->tree_root,
4989                                      &reloc_root->root_key);
4990                 BUG_ON(ret);
4991                 mutex_unlock(&root->fs_info->drop_mutex);
4992
4993                 nr = trans->blocks_used;
4994                 ret = btrfs_end_transaction(trans, root);
4995                 BUG_ON(ret);
4996                 btrfs_btree_balance_dirty(root, nr);
4997
4998                 kfree(prev_root);
4999                 prev_root = reloc_root;
5000         }
5001         if (prev_root) {
5002                 btrfs_remove_leaf_refs(prev_root, (u64)-1, 0);
5003                 kfree(prev_root);
5004         }
5005         return 0;
5006 }
5007
5008 int btrfs_add_dead_reloc_root(struct btrfs_root *root)
5009 {
5010         list_add(&root->dead_list, &root->fs_info->dead_reloc_roots);
5011         return 0;
5012 }
5013
5014 int btrfs_cleanup_reloc_trees(struct btrfs_root *root)
5015 {
5016         struct btrfs_root *reloc_root;
5017         struct btrfs_trans_handle *trans;
5018         struct btrfs_key location;
5019         int found;
5020         int ret;
5021
5022         mutex_lock(&root->fs_info->tree_reloc_mutex);
5023         ret = btrfs_find_dead_roots(root, BTRFS_TREE_RELOC_OBJECTID, NULL);
5024         BUG_ON(ret);
5025         found = !list_empty(&root->fs_info->dead_reloc_roots);
5026         mutex_unlock(&root->fs_info->tree_reloc_mutex);
5027
5028         if (found) {
5029                 trans = btrfs_start_transaction(root, 1);
5030                 BUG_ON(!trans);
5031                 ret = btrfs_commit_transaction(trans, root);
5032                 BUG_ON(ret);
5033         }
5034
5035         location.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID;
5036         location.offset = (u64)-1;
5037         location.type = BTRFS_ROOT_ITEM_KEY;
5038
5039         reloc_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
5040         BUG_ON(!reloc_root);
5041         btrfs_orphan_cleanup(reloc_root);
5042         return 0;
5043 }
5044
5045 static noinline int init_reloc_tree(struct btrfs_trans_handle *trans,
5046                                     struct btrfs_root *root)
5047 {
5048         struct btrfs_root *reloc_root;
5049         struct extent_buffer *eb;
5050         struct btrfs_root_item *root_item;
5051         struct btrfs_key root_key;
5052         int ret;
5053
5054         BUG_ON(!root->ref_cows);
5055         if (root->reloc_root)
5056                 return 0;
5057
5058         root_item = kmalloc(sizeof(*root_item), GFP_NOFS);
5059         BUG_ON(!root_item);
5060
5061         ret = btrfs_copy_root(trans, root, root->commit_root,
5062                               &eb, BTRFS_TREE_RELOC_OBJECTID);
5063         BUG_ON(ret);
5064
5065         root_key.objectid = BTRFS_TREE_RELOC_OBJECTID;
5066         root_key.offset = root->root_key.objectid;
5067         root_key.type = BTRFS_ROOT_ITEM_KEY;
5068
5069         memcpy(root_item, &root->root_item, sizeof(root_item));
5070         btrfs_set_root_refs(root_item, 0);
5071         btrfs_set_root_bytenr(root_item, eb->start);
5072         btrfs_set_root_level(root_item, btrfs_header_level(eb));
5073         btrfs_set_root_generation(root_item, trans->transid);
5074
5075         btrfs_tree_unlock(eb);
5076         free_extent_buffer(eb);
5077
5078         ret = btrfs_insert_root(trans, root->fs_info->tree_root,
5079                                 &root_key, root_item);
5080         BUG_ON(ret);
5081         kfree(root_item);
5082
5083         reloc_root = btrfs_read_fs_root_no_radix(root->fs_info->tree_root,
5084                                                  &root_key);
5085         BUG_ON(!reloc_root);
5086         reloc_root->last_trans = trans->transid;
5087         reloc_root->commit_root = NULL;
5088         reloc_root->ref_tree = &root->fs_info->reloc_ref_tree;
5089
5090         root->reloc_root = reloc_root;
5091         return 0;
5092 }
5093
5094 /*
5095  * Core function of space balance.
5096  *
5097  * The idea is using reloc trees to relocate tree blocks in reference
5098  * counted roots. There is one reloc tree for each subvol, and all
5099  * reloc trees share same root key objectid. Reloc trees are snapshots
5100  * of the latest committed roots of subvols (root->commit_root).
5101  *
5102  * To relocate a tree block referenced by a subvol, there are two steps.
5103  * COW the block through subvol's reloc tree, then update block pointer
5104  * in the subvol to point to the new block. Since all reloc trees share
5105  * same root key objectid, doing special handing for tree blocks owned
5106  * by them is easy. Once a tree block has been COWed in one reloc tree,
5107  * we can use the resulting new block directly when the same block is
5108  * required to COW again through other reloc trees. By this way, relocated
5109  * tree blocks are shared between reloc trees, so they are also shared
5110  * between subvols.
5111  */
5112 static noinline int relocate_one_path(struct btrfs_trans_handle *trans,
5113                                       struct btrfs_root *root,
5114                                       struct btrfs_path *path,
5115                                       struct btrfs_key *first_key,
5116                                       struct btrfs_ref_path *ref_path,
5117                                       struct btrfs_block_group_cache *group,
5118                                       struct inode *reloc_inode)
5119 {
5120         struct btrfs_root *reloc_root;
5121         struct extent_buffer *eb = NULL;
5122         struct btrfs_key *keys;
5123         u64 *nodes;
5124         int level;
5125         int shared_level;
5126         int lowest_level = 0;
5127         int ret;
5128
5129         if (ref_path->owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
5130                 lowest_level = ref_path->owner_objectid;
5131
5132         if (!root->ref_cows) {
5133                 path->lowest_level = lowest_level;
5134                 ret = btrfs_search_slot(trans, root, first_key, path, 0, 1);
5135                 BUG_ON(ret < 0);
5136                 path->lowest_level = 0;
5137                 btrfs_release_path(root, path);
5138                 return 0;
5139         }
5140
5141         mutex_lock(&root->fs_info->tree_reloc_mutex);
5142         ret = init_reloc_tree(trans, root);
5143         BUG_ON(ret);
5144         reloc_root = root->reloc_root;
5145
5146         shared_level = ref_path->shared_level;
5147         ref_path->shared_level = BTRFS_MAX_LEVEL - 1;
5148
5149         keys = ref_path->node_keys;
5150         nodes = ref_path->new_nodes;
5151         memset(&keys[shared_level + 1], 0,
5152                sizeof(*keys) * (BTRFS_MAX_LEVEL - shared_level - 1));
5153         memset(&nodes[shared_level + 1], 0,
5154                sizeof(*nodes) * (BTRFS_MAX_LEVEL - shared_level - 1));
5155
5156         if (nodes[lowest_level] == 0) {
5157                 path->lowest_level = lowest_level;
5158                 ret = btrfs_search_slot(trans, reloc_root, first_key, path,
5159                                         0, 1);
5160                 BUG_ON(ret);
5161                 for (level = lowest_level; level < BTRFS_MAX_LEVEL; level++) {
5162                         eb = path->nodes[level];
5163                         if (!eb || eb == reloc_root->node)
5164                                 break;
5165                         nodes[level] = eb->start;
5166                         if (level == 0)
5167                                 btrfs_item_key_to_cpu(eb, &keys[level], 0);
5168                         else
5169                                 btrfs_node_key_to_cpu(eb, &keys[level], 0);
5170                 }
5171                 if (nodes[0] &&
5172                     ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5173                         eb = path->nodes[0];
5174                         ret = replace_extents_in_leaf(trans, reloc_root, eb,
5175                                                       group, reloc_inode);
5176                         BUG_ON(ret);
5177                 }
5178                 btrfs_release_path(reloc_root, path);
5179         } else {
5180                 ret = btrfs_merge_path(trans, reloc_root, keys, nodes,
5181                                        lowest_level);
5182                 BUG_ON(ret);
5183         }
5184
5185         /*
5186          * replace tree blocks in the fs tree with tree blocks in
5187          * the reloc tree.
5188          */
5189         ret = btrfs_merge_path(trans, root, keys, nodes, lowest_level);
5190         BUG_ON(ret < 0);
5191
5192         if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5193                 ret = btrfs_search_slot(trans, reloc_root, first_key, path,
5194                                         0, 0);
5195                 BUG_ON(ret);
5196                 extent_buffer_get(path->nodes[0]);
5197                 eb = path->nodes[0];
5198                 btrfs_release_path(reloc_root, path);
5199                 ret = invalidate_extent_cache(reloc_root, eb, group, root);
5200                 BUG_ON(ret);
5201                 free_extent_buffer(eb);
5202         }
5203
5204         mutex_unlock(&root->fs_info->tree_reloc_mutex);
5205         path->lowest_level = 0;
5206         return 0;
5207 }
5208
5209 static noinline int relocate_tree_block(struct btrfs_trans_handle *trans,
5210                                         struct btrfs_root *root,
5211                                         struct btrfs_path *path,
5212                                         struct btrfs_key *first_key,
5213                                         struct btrfs_ref_path *ref_path)
5214 {
5215         int ret;
5216
5217         ret = relocate_one_path(trans, root, path, first_key,
5218                                 ref_path, NULL, NULL);
5219         BUG_ON(ret);
5220
5221         if (root == root->fs_info->extent_root)
5222                 btrfs_extent_post_op(trans, root);
5223
5224         return 0;
5225 }
5226
5227 static noinline int del_extent_zero(struct btrfs_trans_handle *trans,
5228                                     struct btrfs_root *extent_root,
5229                                     struct btrfs_path *path,
5230                                     struct btrfs_key *extent_key)
5231 {
5232         int ret;
5233
5234         ret = btrfs_search_slot(trans, extent_root, extent_key, path, -1, 1);
5235         if (ret)
5236                 goto out;
5237         ret = btrfs_del_item(trans, extent_root, path);
5238 out:
5239         btrfs_release_path(extent_root, path);
5240         return ret;
5241 }
5242
5243 static noinline struct btrfs_root *read_ref_root(struct btrfs_fs_info *fs_info,
5244                                                 struct btrfs_ref_path *ref_path)
5245 {
5246         struct btrfs_key root_key;
5247
5248         root_key.objectid = ref_path->root_objectid;
5249         root_key.type = BTRFS_ROOT_ITEM_KEY;
5250         if (is_cowonly_root(ref_path->root_objectid))
5251                 root_key.offset = 0;
5252         else
5253                 root_key.offset = (u64)-1;
5254
5255         return btrfs_read_fs_root_no_name(fs_info, &root_key);
5256 }
5257
5258 static noinline int relocate_one_extent(struct btrfs_root *extent_root,
5259                                         struct btrfs_path *path,
5260                                         struct btrfs_key *extent_key,
5261                                         struct btrfs_block_group_cache *group,
5262                                         struct inode *reloc_inode, int pass)
5263 {
5264         struct btrfs_trans_handle *trans;
5265         struct btrfs_root *found_root;
5266         struct btrfs_ref_path *ref_path = NULL;
5267         struct disk_extent *new_extents = NULL;
5268         int nr_extents = 0;
5269         int loops;
5270         int ret;
5271         int level;
5272         struct btrfs_key first_key;
5273         u64 prev_block = 0;
5274
5275
5276         trans = btrfs_start_transaction(extent_root, 1);
5277         BUG_ON(!trans);
5278
5279         if (extent_key->objectid == 0) {
5280                 ret = del_extent_zero(trans, extent_root, path, extent_key);
5281                 goto out;
5282         }
5283
5284         ref_path = kmalloc(sizeof(*ref_path), GFP_NOFS);
5285         if (!ref_path) {
5286                 ret = -ENOMEM;
5287                 goto out;
5288         }
5289
5290         for (loops = 0; ; loops++) {
5291                 if (loops == 0) {
5292                         ret = btrfs_first_ref_path(trans, extent_root, ref_path,
5293                                                    extent_key->objectid);
5294                 } else {
5295                         ret = btrfs_next_ref_path(trans, extent_root, ref_path);
5296                 }
5297                 if (ret < 0)
5298                         goto out;
5299                 if (ret > 0)
5300                         break;
5301
5302                 if (ref_path->root_objectid == BTRFS_TREE_LOG_OBJECTID ||
5303                     ref_path->root_objectid == BTRFS_TREE_RELOC_OBJECTID)
5304                         continue;
5305
5306                 found_root = read_ref_root(extent_root->fs_info, ref_path);
5307                 BUG_ON(!found_root);
5308                 /*
5309                  * for reference counted tree, only process reference paths
5310                  * rooted at the latest committed root.
5311                  */
5312                 if (found_root->ref_cows &&
5313                     ref_path->root_generation != found_root->root_key.offset)
5314                         continue;
5315
5316                 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5317                         if (pass == 0) {
5318                                 /*
5319                                  * copy data extents to new locations
5320                                  */
5321                                 u64 group_start = group->key.objectid;
5322                                 ret = relocate_data_extent(reloc_inode,
5323                                                            extent_key,
5324                                                            group_start);
5325                                 if (ret < 0)
5326                                         goto out;
5327                                 break;
5328                         }
5329                         level = 0;
5330                 } else {
5331                         level = ref_path->owner_objectid;
5332                 }
5333
5334                 if (prev_block != ref_path->nodes[level]) {
5335                         struct extent_buffer *eb;
5336                         u64 block_start = ref_path->nodes[level];
5337                         u64 block_size = btrfs_level_size(found_root, level);
5338
5339                         eb = read_tree_block(found_root, block_start,
5340                                              block_size, 0);
5341                         btrfs_tree_lock(eb);
5342                         BUG_ON(level != btrfs_header_level(eb));
5343
5344                         if (level == 0)
5345                                 btrfs_item_key_to_cpu(eb, &first_key, 0);
5346                         else
5347                                 btrfs_node_key_to_cpu(eb, &first_key, 0);
5348
5349                         btrfs_tree_unlock(eb);
5350                         free_extent_buffer(eb);
5351                         prev_block = block_start;
5352                 }
5353
5354                 btrfs_record_root_in_trans(found_root);
5355                 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5356                         /*
5357                          * try to update data extent references while
5358                          * keeping metadata shared between snapshots.
5359                          */
5360                         if (pass == 1) {
5361                                 ret = relocate_one_path(trans, found_root,
5362                                                 path, &first_key, ref_path,
5363                                                 group, reloc_inode);
5364                                 if (ret < 0)
5365                                         goto out;
5366                                 continue;
5367                         }
5368                         /*
5369                          * use fallback method to process the remaining
5370                          * references.
5371                          */
5372                         if (!new_extents) {
5373                                 u64 group_start = group->key.objectid;
5374                                 new_extents = kmalloc(sizeof(*new_extents),
5375                                                       GFP_NOFS);
5376                                 nr_extents = 1;
5377                                 ret = get_new_locations(reloc_inode,
5378                                                         extent_key,
5379                                                         group_start, 1,
5380                                                         &new_extents,
5381                                                         &nr_extents);
5382                                 if (ret)
5383                                         goto out;
5384                         }
5385                         ret = replace_one_extent(trans, found_root,
5386                                                 path, extent_key,
5387                                                 &first_key, ref_path,
5388                                                 new_extents, nr_extents);
5389                 } else {
5390                         ret = relocate_tree_block(trans, found_root, path,
5391                                                   &first_key, ref_path);
5392                 }
5393                 if (ret < 0)
5394                         goto out;
5395         }
5396         ret = 0;
5397 out:
5398         btrfs_end_transaction(trans, extent_root);
5399         kfree(new_extents);
5400         kfree(ref_path);
5401         return ret;
5402 }
5403
5404 static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
5405 {
5406         u64 num_devices;
5407         u64 stripped = BTRFS_BLOCK_GROUP_RAID0 |
5408                 BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10;
5409
5410         num_devices = root->fs_info->fs_devices->rw_devices;
5411         if (num_devices == 1) {
5412                 stripped |= BTRFS_BLOCK_GROUP_DUP;
5413                 stripped = flags & ~stripped;
5414
5415                 /* turn raid0 into single device chunks */
5416                 if (flags & BTRFS_BLOCK_GROUP_RAID0)
5417                         return stripped;
5418
5419                 /* turn mirroring into duplication */
5420                 if (flags & (BTRFS_BLOCK_GROUP_RAID1 |
5421                              BTRFS_BLOCK_GROUP_RAID10))
5422                         return stripped | BTRFS_BLOCK_GROUP_DUP;
5423                 return flags;
5424         } else {
5425                 /* they already had raid on here, just return */
5426                 if (flags & stripped)
5427                         return flags;
5428
5429                 stripped |= BTRFS_BLOCK_GROUP_DUP;
5430                 stripped = flags & ~stripped;
5431
5432                 /* switch duplicated blocks with raid1 */
5433                 if (flags & BTRFS_BLOCK_GROUP_DUP)
5434                         return stripped | BTRFS_BLOCK_GROUP_RAID1;
5435
5436                 /* turn single device chunks into raid0 */
5437                 return stripped | BTRFS_BLOCK_GROUP_RAID0;
5438         }
5439         return flags;
5440 }
5441
5442 static int __alloc_chunk_for_shrink(struct btrfs_root *root,
5443                      struct btrfs_block_group_cache *shrink_block_group,
5444                      int force)
5445 {
5446         struct btrfs_trans_handle *trans;
5447         u64 new_alloc_flags;
5448         u64 calc;
5449
5450         spin_lock(&shrink_block_group->lock);
5451         if (btrfs_block_group_used(&shrink_block_group->item) > 0) {
5452                 spin_unlock(&shrink_block_group->lock);
5453
5454                 trans = btrfs_start_transaction(root, 1);
5455                 spin_lock(&shrink_block_group->lock);
5456
5457                 new_alloc_flags = update_block_group_flags(root,
5458                                                    shrink_block_group->flags);
5459                 if (new_alloc_flags != shrink_block_group->flags) {
5460                         calc =
5461                              btrfs_block_group_used(&shrink_block_group->item);
5462                 } else {
5463                         calc = shrink_block_group->key.offset;
5464                 }
5465                 spin_unlock(&shrink_block_group->lock);
5466
5467                 do_chunk_alloc(trans, root->fs_info->extent_root,
5468                                calc + 2 * 1024 * 1024, new_alloc_flags, force);
5469
5470                 btrfs_end_transaction(trans, root);
5471         } else
5472                 spin_unlock(&shrink_block_group->lock);
5473         return 0;
5474 }
5475
5476 static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
5477                                  struct btrfs_root *root,
5478                                  u64 objectid, u64 size)
5479 {
5480         struct btrfs_path *path;
5481         struct btrfs_inode_item *item;
5482         struct extent_buffer *leaf;
5483         int ret;
5484
5485         path = btrfs_alloc_path();
5486         if (!path)
5487                 return -ENOMEM;
5488
5489         ret = btrfs_insert_empty_inode(trans, root, path, objectid);
5490         if (ret)
5491                 goto out;
5492
5493         leaf = path->nodes[0];
5494         item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item);
5495         memset_extent_buffer(leaf, 0, (unsigned long)item, sizeof(*item));
5496         btrfs_set_inode_generation(leaf, item, 1);
5497         btrfs_set_inode_size(leaf, item, size);
5498         btrfs_set_inode_mode(leaf, item, S_IFREG | 0600);
5499         btrfs_set_inode_flags(leaf, item, BTRFS_INODE_NOCOMPRESS);
5500         btrfs_mark_buffer_dirty(leaf);
5501         btrfs_release_path(root, path);
5502 out:
5503         btrfs_free_path(path);
5504         return ret;
5505 }
5506
5507 static noinline struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info,
5508                                         struct btrfs_block_group_cache *group)
5509 {
5510         struct inode *inode = NULL;
5511         struct btrfs_trans_handle *trans;
5512         struct btrfs_root *root;
5513         struct btrfs_key root_key;
5514         u64 objectid = BTRFS_FIRST_FREE_OBJECTID;
5515         int err = 0;
5516
5517         root_key.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID;
5518         root_key.type = BTRFS_ROOT_ITEM_KEY;
5519         root_key.offset = (u64)-1;
5520         root = btrfs_read_fs_root_no_name(fs_info, &root_key);
5521         if (IS_ERR(root))
5522                 return ERR_CAST(root);
5523
5524         trans = btrfs_start_transaction(root, 1);
5525         BUG_ON(!trans);
5526
5527         err = btrfs_find_free_objectid(trans, root, objectid, &objectid);
5528         if (err)
5529                 goto out;
5530
5531         err = __insert_orphan_inode(trans, root, objectid, group->key.offset);
5532         BUG_ON(err);
5533
5534         err = btrfs_insert_file_extent(trans, root, objectid, 0, 0, 0,
5535                                        group->key.offset, 0, group->key.offset,
5536                                        0, 0, 0);
5537         BUG_ON(err);
5538
5539         inode = btrfs_iget_locked(root->fs_info->sb, objectid, root);
5540         if (inode->i_state & I_NEW) {
5541                 BTRFS_I(inode)->root = root;
5542                 BTRFS_I(inode)->location.objectid = objectid;
5543                 BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
5544                 BTRFS_I(inode)->location.offset = 0;
5545                 btrfs_read_locked_inode(inode);
5546                 unlock_new_inode(inode);
5547                 BUG_ON(is_bad_inode(inode));
5548         } else {
5549                 BUG_ON(1);
5550         }
5551         BTRFS_I(inode)->index_cnt = group->key.objectid;
5552
5553         err = btrfs_orphan_add(trans, inode);
5554 out:
5555         btrfs_end_transaction(trans, root);
5556         if (err) {
5557                 if (inode)
5558                         iput(inode);
5559                 inode = ERR_PTR(err);
5560         }
5561         return inode;
5562 }
5563
5564 int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len)
5565 {
5566
5567         struct btrfs_ordered_sum *sums;
5568         struct btrfs_sector_sum *sector_sum;
5569         struct btrfs_ordered_extent *ordered;
5570         struct btrfs_root *root = BTRFS_I(inode)->root;
5571         struct list_head list;
5572         size_t offset;
5573         int ret;
5574         u64 disk_bytenr;
5575
5576         INIT_LIST_HEAD(&list);
5577
5578         ordered = btrfs_lookup_ordered_extent(inode, file_pos);
5579         BUG_ON(ordered->file_offset != file_pos || ordered->len != len);
5580
5581         disk_bytenr = file_pos + BTRFS_I(inode)->index_cnt;
5582         ret = btrfs_lookup_csums_range(root->fs_info->csum_root, disk_bytenr,
5583                                        disk_bytenr + len - 1, &list);
5584
5585         while (!list_empty(&list)) {
5586                 sums = list_entry(list.next, struct btrfs_ordered_sum, list);
5587                 list_del_init(&sums->list);
5588
5589                 sector_sum = sums->sums;
5590                 sums->bytenr = ordered->start;
5591
5592                 offset = 0;
5593                 while (offset < sums->len) {
5594                         sector_sum->bytenr += ordered->start - disk_bytenr;
5595                         sector_sum++;
5596                         offset += root->sectorsize;
5597                 }
5598
5599                 btrfs_add_ordered_sum(inode, ordered, sums);
5600         }
5601         btrfs_put_ordered_extent(ordered);
5602         return 0;
5603 }
5604
5605 int btrfs_relocate_block_group(struct btrfs_root *root, u64 group_start)
5606 {
5607         struct btrfs_trans_handle *trans;
5608         struct btrfs_path *path;
5609         struct btrfs_fs_info *info = root->fs_info;
5610         struct extent_buffer *leaf;
5611         struct inode *reloc_inode;
5612         struct btrfs_block_group_cache *block_group;
5613         struct btrfs_key key;
5614         u64 skipped;
5615         u64 cur_byte;
5616         u64 total_found;
5617         u32 nritems;
5618         int ret;
5619         int progress;
5620         int pass = 0;
5621
5622         root = root->fs_info->extent_root;
5623
5624         block_group = btrfs_lookup_block_group(info, group_start);
5625         BUG_ON(!block_group);
5626
5627         printk(KERN_INFO "btrfs relocating block group %llu flags %llu\n",
5628                (unsigned long long)block_group->key.objectid,
5629                (unsigned long long)block_group->flags);
5630
5631         path = btrfs_alloc_path();
5632         BUG_ON(!path);
5633
5634         reloc_inode = create_reloc_inode(info, block_group);
5635         BUG_ON(IS_ERR(reloc_inode));
5636
5637         __alloc_chunk_for_shrink(root, block_group, 1);
5638         set_block_group_readonly(block_group);
5639
5640         btrfs_start_delalloc_inodes(info->tree_root);
5641         btrfs_wait_ordered_extents(info->tree_root, 0);
5642 again:
5643         skipped = 0;
5644         total_found = 0;
5645         progress = 0;
5646         key.objectid = block_group->key.objectid;
5647         key.offset = 0;
5648         key.type = 0;
5649         cur_byte = key.objectid;
5650
5651         trans = btrfs_start_transaction(info->tree_root, 1);
5652         btrfs_commit_transaction(trans, info->tree_root);
5653
5654         mutex_lock(&root->fs_info->cleaner_mutex);
5655         btrfs_clean_old_snapshots(info->tree_root);
5656         btrfs_remove_leaf_refs(info->tree_root, (u64)-1, 1);
5657         mutex_unlock(&root->fs_info->cleaner_mutex);
5658
5659         while (1) {
5660                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5661                 if (ret < 0)
5662                         goto out;
5663 next:
5664                 leaf = path->nodes[0];
5665                 nritems = btrfs_header_nritems(leaf);
5666                 if (path->slots[0] >= nritems) {
5667                         ret = btrfs_next_leaf(root, path);
5668                         if (ret < 0)
5669                                 goto out;
5670                         if (ret == 1) {
5671                                 ret = 0;
5672                                 break;
5673                         }
5674                         leaf = path->nodes[0];
5675                         nritems = btrfs_header_nritems(leaf);
5676                 }
5677
5678                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
5679
5680                 if (key.objectid >= block_group->key.objectid +
5681                     block_group->key.offset)
5682                         break;
5683
5684                 if (progress && need_resched()) {
5685                         btrfs_release_path(root, path);
5686                         cond_resched();
5687                         progress = 0;
5688                         continue;
5689                 }
5690                 progress = 1;
5691
5692                 if (btrfs_key_type(&key) != BTRFS_EXTENT_ITEM_KEY ||
5693                     key.objectid + key.offset <= cur_byte) {
5694                         path->slots[0]++;
5695                         goto next;
5696                 }
5697
5698                 total_found++;
5699                 cur_byte = key.objectid + key.offset;
5700                 btrfs_release_path(root, path);
5701
5702                 __alloc_chunk_for_shrink(root, block_group, 0);
5703                 ret = relocate_one_extent(root, path, &key, block_group,
5704                                           reloc_inode, pass);
5705                 BUG_ON(ret < 0);
5706                 if (ret > 0)
5707                         skipped++;
5708
5709                 key.objectid = cur_byte;
5710                 key.type = 0;
5711                 key.offset = 0;
5712         }
5713
5714         btrfs_release_path(root, path);
5715
5716         if (pass == 0) {
5717                 btrfs_wait_ordered_range(reloc_inode, 0, (u64)-1);
5718                 invalidate_mapping_pages(reloc_inode->i_mapping, 0, -1);
5719         }
5720
5721         if (total_found > 0) {
5722                 printk(KERN_INFO "btrfs found %llu extents in pass %d\n",
5723                        (unsigned long long)total_found, pass);
5724                 pass++;
5725                 if (total_found == skipped && pass > 2) {
5726                         iput(reloc_inode);
5727                         reloc_inode = create_reloc_inode(info, block_group);
5728                         pass = 0;
5729                 }
5730                 goto again;
5731         }
5732
5733         /* delete reloc_inode */
5734         iput(reloc_inode);
5735
5736         /* unpin extents in this range */
5737         trans = btrfs_start_transaction(info->tree_root, 1);
5738         btrfs_commit_transaction(trans, info->tree_root);
5739
5740         spin_lock(&block_group->lock);
5741         WARN_ON(block_group->pinned > 0);
5742         WARN_ON(block_group->reserved > 0);
5743         WARN_ON(btrfs_block_group_used(&block_group->item) > 0);
5744         spin_unlock(&block_group->lock);
5745         put_block_group(block_group);
5746         ret = 0;
5747 out:
5748         btrfs_free_path(path);
5749         return ret;
5750 }
5751
5752 static int find_first_block_group(struct btrfs_root *root,
5753                 struct btrfs_path *path, struct btrfs_key *key)
5754 {
5755         int ret = 0;
5756         struct btrfs_key found_key;
5757         struct extent_buffer *leaf;
5758         int slot;
5759
5760         ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
5761         if (ret < 0)
5762                 goto out;
5763
5764         while (1) {
5765                 slot = path->slots[0];
5766                 leaf = path->nodes[0];
5767                 if (slot >= btrfs_header_nritems(leaf)) {
5768                         ret = btrfs_next_leaf(root, path);
5769                         if (ret == 0)
5770                                 continue;
5771                         if (ret < 0)
5772                                 goto out;
5773                         break;
5774                 }
5775                 btrfs_item_key_to_cpu(leaf, &found_key, slot);
5776
5777                 if (found_key.objectid >= key->objectid &&
5778                     found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) {
5779                         ret = 0;
5780                         goto out;
5781                 }
5782                 path->slots[0]++;
5783         }
5784         ret = -ENOENT;
5785 out:
5786         return ret;
5787 }
5788
5789 int btrfs_free_block_groups(struct btrfs_fs_info *info)
5790 {
5791         struct btrfs_block_group_cache *block_group;
5792         struct rb_node *n;
5793
5794         spin_lock(&info->block_group_cache_lock);
5795         while ((n = rb_last(&info->block_group_cache_tree)) != NULL) {
5796                 block_group = rb_entry(n, struct btrfs_block_group_cache,
5797                                        cache_node);
5798                 rb_erase(&block_group->cache_node,
5799                          &info->block_group_cache_tree);
5800                 spin_unlock(&info->block_group_cache_lock);
5801
5802                 btrfs_remove_free_space_cache(block_group);
5803                 down_write(&block_group->space_info->groups_sem);
5804                 list_del(&block_group->list);
5805                 up_write(&block_group->space_info->groups_sem);
5806
5807                 WARN_ON(atomic_read(&block_group->count) != 1);
5808                 kfree(block_group);
5809
5810                 spin_lock(&info->block_group_cache_lock);
5811         }
5812         spin_unlock(&info->block_group_cache_lock);
5813         return 0;
5814 }
5815
5816 int btrfs_read_block_groups(struct btrfs_root *root)
5817 {
5818         struct btrfs_path *path;
5819         int ret;
5820         struct btrfs_block_group_cache *cache;
5821         struct btrfs_fs_info *info = root->fs_info;
5822         struct btrfs_space_info *space_info;
5823         struct btrfs_key key;
5824         struct btrfs_key found_key;
5825         struct extent_buffer *leaf;
5826
5827         root = info->extent_root;
5828         key.objectid = 0;
5829         key.offset = 0;
5830         btrfs_set_key_type(&key, BTRFS_BLOCK_GROUP_ITEM_KEY);
5831         path = btrfs_alloc_path();
5832         if (!path)
5833                 return -ENOMEM;
5834
5835         while (1) {
5836                 ret = find_first_block_group(root, path, &key);
5837                 if (ret > 0) {
5838                         ret = 0;
5839                         goto error;
5840                 }
5841                 if (ret != 0)
5842                         goto error;
5843
5844                 leaf = path->nodes[0];
5845                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
5846                 cache = kzalloc(sizeof(*cache), GFP_NOFS);
5847                 if (!cache) {
5848                         ret = -ENOMEM;
5849                         break;
5850                 }
5851
5852                 atomic_set(&cache->count, 1);
5853                 spin_lock_init(&cache->lock);
5854                 mutex_init(&cache->alloc_mutex);
5855                 mutex_init(&cache->cache_mutex);
5856                 INIT_LIST_HEAD(&cache->list);
5857                 read_extent_buffer(leaf, &cache->item,
5858                                    btrfs_item_ptr_offset(leaf, path->slots[0]),
5859                                    sizeof(cache->item));
5860                 memcpy(&cache->key, &found_key, sizeof(found_key));
5861
5862                 key.objectid = found_key.objectid + found_key.offset;
5863                 btrfs_release_path(root, path);
5864                 cache->flags = btrfs_block_group_flags(&cache->item);
5865
5866                 ret = update_space_info(info, cache->flags, found_key.offset,
5867                                         btrfs_block_group_used(&cache->item),
5868                                         &space_info);
5869                 BUG_ON(ret);
5870                 cache->space_info = space_info;
5871                 down_write(&space_info->groups_sem);
5872                 list_add_tail(&cache->list, &space_info->block_groups);
5873                 up_write(&space_info->groups_sem);
5874
5875                 ret = btrfs_add_block_group_cache(root->fs_info, cache);
5876                 BUG_ON(ret);
5877
5878                 set_avail_alloc_bits(root->fs_info, cache->flags);
5879                 if (btrfs_chunk_readonly(root, cache->key.objectid))
5880                         set_block_group_readonly(cache);
5881         }
5882         ret = 0;
5883 error:
5884         btrfs_free_path(path);
5885         return ret;
5886 }
5887
5888 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
5889                            struct btrfs_root *root, u64 bytes_used,
5890                            u64 type, u64 chunk_objectid, u64 chunk_offset,
5891                            u64 size)
5892 {
5893         int ret;
5894         struct btrfs_root *extent_root;
5895         struct btrfs_block_group_cache *cache;
5896
5897         extent_root = root->fs_info->extent_root;
5898
5899         root->fs_info->last_trans_new_blockgroup = trans->transid;
5900
5901         cache = kzalloc(sizeof(*cache), GFP_NOFS);
5902         if (!cache)
5903                 return -ENOMEM;
5904
5905         cache->key.objectid = chunk_offset;
5906         cache->key.offset = size;
5907         cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
5908         atomic_set(&cache->count, 1);
5909         spin_lock_init(&cache->lock);
5910         mutex_init(&cache->alloc_mutex);
5911         mutex_init(&cache->cache_mutex);
5912         INIT_LIST_HEAD(&cache->list);
5913
5914         btrfs_set_block_group_used(&cache->item, bytes_used);
5915         btrfs_set_block_group_chunk_objectid(&cache->item, chunk_objectid);
5916         cache->flags = type;
5917         btrfs_set_block_group_flags(&cache->item, type);
5918
5919         ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
5920                                 &cache->space_info);
5921         BUG_ON(ret);
5922         down_write(&cache->space_info->groups_sem);
5923         list_add_tail(&cache->list, &cache->space_info->block_groups);
5924         up_write(&cache->space_info->groups_sem);
5925
5926         ret = btrfs_add_block_group_cache(root->fs_info, cache);
5927         BUG_ON(ret);
5928
5929         ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
5930                                 sizeof(cache->item));
5931         BUG_ON(ret);
5932
5933         finish_current_insert(trans, extent_root, 0);
5934         ret = del_pending_extents(trans, extent_root, 0);
5935         BUG_ON(ret);
5936         set_avail_alloc_bits(extent_root->fs_info, type);
5937
5938         return 0;
5939 }
5940
5941 int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
5942                              struct btrfs_root *root, u64 group_start)
5943 {
5944         struct btrfs_path *path;
5945         struct btrfs_block_group_cache *block_group;
5946         struct btrfs_key key;
5947         int ret;
5948
5949         root = root->fs_info->extent_root;
5950
5951         block_group = btrfs_lookup_block_group(root->fs_info, group_start);
5952         BUG_ON(!block_group);
5953         BUG_ON(!block_group->ro);
5954
5955         memcpy(&key, &block_group->key, sizeof(key));
5956
5957         path = btrfs_alloc_path();
5958         BUG_ON(!path);
5959
5960         btrfs_remove_free_space_cache(block_group);
5961         rb_erase(&block_group->cache_node,
5962                  &root->fs_info->block_group_cache_tree);
5963         down_write(&block_group->space_info->groups_sem);
5964         list_del(&block_group->list);
5965         up_write(&block_group->space_info->groups_sem);
5966
5967         spin_lock(&block_group->space_info->lock);
5968         block_group->space_info->total_bytes -= block_group->key.offset;
5969         block_group->space_info->bytes_readonly -= block_group->key.offset;
5970         spin_unlock(&block_group->space_info->lock);
5971         block_group->space_info->full = 0;
5972
5973         put_block_group(block_group);
5974         put_block_group(block_group);
5975
5976         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
5977         if (ret > 0)
5978                 ret = -EIO;
5979         if (ret < 0)
5980                 goto out;
5981
5982         ret = btrfs_del_item(trans, root, path);
5983 out:
5984         btrfs_free_path(path);
5985         return ret;
5986 }