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/sort.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
  34 #define PENDING_EXTENT_INSERT 0
  35 #define PENDING_EXTENT_DELETE 1
  36 #define PENDING_BACKREF_UPDATE 2
  37
  38 struct pending_extent_op {
  39         int type;
  40         u64 bytenr;
  41         u64 num_bytes;
  42         u64 parent;
  43         u64 orig_parent;
  44         u64 generation;
  45         u64 orig_generation;
  46         int level;
  47         struct list_head list;
  48         int del;
  49 };
  50
  51 static int finish_current_insert(struct btrfs_trans_handle *trans,
  52                                  struct btrfs_root *extent_root, int all);
  53 static int del_pending_extents(struct btrfs_trans_handle *trans,
  54                                struct btrfs_root *extent_root, int all);
  55 static int pin_down_bytes(struct btrfs_trans_handle *trans,
  56                           struct btrfs_root *root,
  57                           u64 bytenr, u64 num_bytes, int is_data);
  58 static int update_block_group(struct btrfs_trans_handle *trans,
  59                               struct btrfs_root *root,
  60                               u64 bytenr, u64 num_bytes, int alloc,
  61                               int mark_free);
  62
  63 static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
  64 {
  65         return (cache->flags & bits) == bits;
  66 }
  67
  68 /*
  69  * this adds the block group to the fs_info rb tree for the block group
  70  * cache
  71  */
  72 static int btrfs_add_block_group_cache(struct btrfs_fs_info *info,
  73                                 struct btrfs_block_group_cache *block_group)
  74 {
  75         struct rb_node **p;
  76         struct rb_node *parent = NULL;
  77         struct btrfs_block_group_cache *cache;
  78
  79         spin_lock(&info->block_group_cache_lock);
  80         p = &info->block_group_cache_tree.rb_node;
  81
  82         while (*p) {
  83                 parent = *p;
  84                 cache = rb_entry(parent, struct btrfs_block_group_cache,
  85                                  cache_node);
  86                 if (block_group->key.objectid < cache->key.objectid) {
  87                         p = &(*p)->rb_left;
  88                 } else if (block_group->key.objectid > cache->key.objectid) {
  89                         p = &(*p)->rb_right;
  90                 } else {
  91                         spin_unlock(&info->block_group_cache_lock);
  92                         return -EEXIST;
  93                 }
  94         }
  95
  96         rb_link_node(&block_group->cache_node, parent, p);
  97         rb_insert_color(&block_group->cache_node,
  98                         &info->block_group_cache_tree);
  99         spin_unlock(&info->block_group_cache_lock);
 100
 101         return 0;
 102 }
 103
 104 /*
 105  * This will return the block group at or after bytenr if contains is 0, else
 106  * it will return the block group that contains the bytenr
 107  */
 108 static struct btrfs_block_group_cache *
 109 block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr,
 110                               int contains)
 111 {
 112         struct btrfs_block_group_cache *cache, *ret = NULL;
 113         struct rb_node *n;
 114         u64 end, start;
 115
 116         spin_lock(&info->block_group_cache_lock);
 117         n = info->block_group_cache_tree.rb_node;
 118
 119         while (n) {
 120                 cache = rb_entry(n, struct btrfs_block_group_cache,
 121                                  cache_node);
 122                 end = cache->key.objectid + cache->key.offset - 1;
 123                 start = cache->key.objectid;
 124
 125                 if (bytenr < start) {
 126                         if (!contains && (!ret || start < ret->key.objectid))
 127                                 ret = cache;
 128                         n = n->rb_left;
 129                 } else if (bytenr > start) {
 130                         if (contains && bytenr <= end) {
 131                                 ret = cache;
 132                                 break;
 133                         }
 134                         n = n->rb_right;
 135                 } else {
 136                         ret = cache;
 137                         break;
 138                 }
 139         }
 140         if (ret)
 141                 atomic_inc(&ret->count);
 142         spin_unlock(&info->block_group_cache_lock);
 143
 144         return ret;
 145 }
 146
 147 /*
 148  * this is only called by cache_block_group, since we could have freed extents
 149  * we need to check the pinned_extents for any extents that can't be used yet
 150  * since their free space will be released as soon as the transaction commits.
 151  */
 152 static int add_new_free_space(struct btrfs_block_group_cache *block_group,
 153                               struct btrfs_fs_info *info, u64 start, u64 end)
 154 {
 155         u64 extent_start, extent_end, size;
 156         int ret;
 157
 158         mutex_lock(&info->pinned_mutex);
 159         while (start < end) {
 160                 ret = find_first_extent_bit(&info->pinned_extents, start,
 161                                             &extent_start, &extent_end,
 162                                             EXTENT_DIRTY);
 163                 if (ret)
 164                         break;
 165
 166                 if (extent_start == start) {
 167                         start = extent_end + 1;
 168                 } else if (extent_start > start && extent_start < end) {
 169                         size = extent_start - start;
 170                         ret = btrfs_add_free_space(block_group, start,
 171                                                    size);
 172                         BUG_ON(ret);
 173                         start = extent_end + 1;
 174                 } else {
 175                         break;
 176                 }
 177         }
 178
 179         if (start < end) {
 180                 size = end - start;
 181                 ret = btrfs_add_free_space(block_group, start, size);
 182                 BUG_ON(ret);
 183         }
 184         mutex_unlock(&info->pinned_mutex);
 185
 186         return 0;
 187 }
 188
 189 static int remove_sb_from_cache(struct btrfs_root *root,
 190                                 struct btrfs_block_group_cache *cache)
 191 {
 192         u64 bytenr;
 193         u64 *logical;
 194         int stripe_len;
 195         int i, nr, ret;
 196
 197         for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
 198                 bytenr = btrfs_sb_offset(i);
 199                 ret = btrfs_rmap_block(&root->fs_info->mapping_tree,
 200                                        cache->key.objectid, bytenr, 0,
 201                                        &logical, &nr, &stripe_len);
 202                 BUG_ON(ret);
 203                 while (nr--) {
 204                         btrfs_remove_free_space(cache, logical[nr],
 205                                                 stripe_len);
 206                 }
 207                 kfree(logical);
 208         }
 209         return 0;
 210 }
 211
 212 static int cache_block_group(struct btrfs_root *root,
 213                              struct btrfs_block_group_cache *block_group)
 214 {
 215         struct btrfs_path *path;
 216         int ret = 0;
 217         struct btrfs_key key;
 218         struct extent_buffer *leaf;
 219         int slot;
 220         u64 last;
 221
 222         if (!block_group)
 223                 return 0;
 224
 225         root = root->fs_info->extent_root;
 226
 227         if (block_group->cached)
 228                 return 0;
 229
 230         path = btrfs_alloc_path();
 231         if (!path)
 232                 return -ENOMEM;
 233
 234         path->reada = 2;
 235         /*
 236          * we get into deadlocks with paths held by callers of this function.
 237          * since the alloc_mutex is protecting things right now, just
 238          * skip the locking here
 239          */
 240         path->skip_locking = 1;
 241         last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
 242         key.objectid = last;
 243         key.offset = 0;
 244         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
 245         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 246         if (ret < 0)
 247                 goto err;
 248
 249         while (1) {
 250                 leaf = path->nodes[0];
 251                 slot = path->slots[0];
 252                 if (slot >= btrfs_header_nritems(leaf)) {
 253                         ret = btrfs_next_leaf(root, path);
 254                         if (ret < 0)
 255                                 goto err;
 256                         if (ret == 0)
 257                                 continue;
 258                         else
 259                                 break;
 260                 }
 261                 btrfs_item_key_to_cpu(leaf, &key, slot);
 262                 if (key.objectid < block_group->key.objectid)
 263                         goto next;
 264
 265                 if (key.objectid >= block_group->key.objectid +
 266                     block_group->key.offset)
 267                         break;
 268
 269                 if (btrfs_key_type(&key) == BTRFS_EXTENT_ITEM_KEY) {
 270                         add_new_free_space(block_group, root->fs_info, last,
 271                                            key.objectid);
 272
 273                         last = key.objectid + key.offset;
 274                 }
 275 next:
 276                 path->slots[0]++;
 277         }
 278
 279         add_new_free_space(block_group, root->fs_info, last,
 280                            block_group->key.objectid +
 281                            block_group->key.offset);
 282
 283         remove_sb_from_cache(root, block_group);
 284         block_group->cached = 1;
 285         ret = 0;
 286 err:
 287         btrfs_free_path(path);
 288         return ret;
 289 }
 290
 291 /*
 292  * return the block group that starts at or after bytenr
 293  */
 294 static struct btrfs_block_group_cache *
 295 btrfs_lookup_first_block_group(struct btrfs_fs_info *info, u64 bytenr)
 296 {
 297         struct btrfs_block_group_cache *cache;
 298
 299         cache = block_group_cache_tree_search(info, bytenr, 0);
 300
 301         return cache;
 302 }
 303
 304 /*
 305  * return the block group that contains teh given bytenr
 306  */
 307 struct btrfs_block_group_cache *btrfs_lookup_block_group(
 308                                                  struct btrfs_fs_info *info,
 309                                                  u64 bytenr)
 310 {
 311         struct btrfs_block_group_cache *cache;
 312
 313         cache = block_group_cache_tree_search(info, bytenr, 1);
 314
 315         return cache;
 316 }
 317
 318 static inline void put_block_group(struct btrfs_block_group_cache *cache)
 319 {
 320         if (atomic_dec_and_test(&cache->count))
 321                 kfree(cache);
 322 }
 323
 324 static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
 325                                                   u64 flags)
 326 {
 327         struct list_head *head = &info->space_info;
 328         struct btrfs_space_info *found;
 329         list_for_each_entry(found, head, list) {
 330                 if (found->flags == flags)
 331                         return found;
 332         }
 333         return NULL;
 334 }
 335
 336 static u64 div_factor(u64 num, int factor)
 337 {
 338         if (factor == 10)
 339                 return num;
 340         num *= factor;
 341         do_div(num, 10);
 342         return num;
 343 }
 344
 345 u64 btrfs_find_block_group(struct btrfs_root *root,
 346                            u64 search_start, u64 search_hint, int owner)
 347 {
 348         struct btrfs_block_group_cache *cache;
 349         u64 used;
 350         u64 last = max(search_hint, search_start);
 351         u64 group_start = 0;
 352         int full_search = 0;
 353         int factor = 9;
 354         int wrapped = 0;
 355 again:
 356         while (1) {
 357                 cache = btrfs_lookup_first_block_group(root->fs_info, last);
 358                 if (!cache)
 359                         break;
 360
 361                 spin_lock(&cache->lock);
 362                 last = cache->key.objectid + cache->key.offset;
 363                 used = btrfs_block_group_used(&cache->item);
 364
 365                 if ((full_search || !cache->ro) &&
 366                     block_group_bits(cache, BTRFS_BLOCK_GROUP_METADATA)) {
 367                         if (used + cache->pinned + cache->reserved <
 368                             div_factor(cache->key.offset, factor)) {
 369                                 group_start = cache->key.objectid;
 370                                 spin_unlock(&cache->lock);
 371                                 put_block_group(cache);
 372                                 goto found;
 373                         }
 374                 }
 375                 spin_unlock(&cache->lock);
 376                 put_block_group(cache);
 377                 cond_resched();
 378         }
 379         if (!wrapped) {
 380                 last = search_start;
 381                 wrapped = 1;
 382                 goto again;
 383         }
 384         if (!full_search && factor < 10) {
 385                 last = search_start;
 386                 full_search = 1;
 387                 factor = 10;
 388                 goto again;
 389         }
 390 found:
 391         return group_start;
 392 }
 393
 394 /* simple helper to search for an existing extent at a given offset */
 395 int btrfs_lookup_extent(struct btrfs_root *root, u64 start, u64 len)
 396 {
 397         int ret;
 398         struct btrfs_key key;
 399         struct btrfs_path *path;
 400
 401         path = btrfs_alloc_path();
 402         BUG_ON(!path);
 403         key.objectid = start;
 404         key.offset = len;
 405         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
 406         ret = btrfs_search_slot(NULL, root->fs_info->extent_root, &key, path,
 407                                 0, 0);
 408         btrfs_free_path(path);
 409         return ret;
 410 }
 411
 412 /*
 413  * Back reference rules.  Back refs have three main goals:
 414  *
 415  * 1) differentiate between all holders of references to an extent so that
 416  *    when a reference is dropped we can make sure it was a valid reference
 417  *    before freeing the extent.
 418  *
 419  * 2) Provide enough information to quickly find the holders of an extent
 420  *    if we notice a given block is corrupted or bad.
 421  *
 422  * 3) Make it easy to migrate blocks for FS shrinking or storage pool
 423  *    maintenance.  This is actually the same as #2, but with a slightly
 424  *    different use case.
 425  *
 426  * File extents can be referenced by:
 427  *
 428  * - multiple snapshots, subvolumes, or different generations in one subvol
 429  * - different files inside a single subvolume
 430  * - different offsets inside a file (bookend extents in file.c)
 431  *
 432  * The extent ref structure has fields for:
 433  *
 434  * - Objectid of the subvolume root
 435  * - Generation number of the tree holding the reference
 436  * - objectid of the file holding the reference
 437  * - number of references holding by parent node (alway 1 for tree blocks)
 438  *
 439  * Btree leaf may hold multiple references to a file extent. In most cases,
 440  * these references are from same file and the corresponding offsets inside
 441  * the file are close together.
 442  *
 443  * When a file extent is allocated the fields are filled in:
 444  *     (root_key.objectid, trans->transid, inode objectid, 1)
 445  *
 446  * When a leaf is cow'd new references are added for every file extent found
 447  * in the leaf.  It looks similar to the create case, but trans->transid will
 448  * be different when the block is cow'd.
 449  *
 450  *     (root_key.objectid, trans->transid, inode objectid,
 451  *      number of references in the leaf)
 452  *
 453  * When a file extent is removed either during snapshot deletion or
 454  * file truncation, we find the corresponding back reference and check
 455  * the following fields:
 456  *
 457  *     (btrfs_header_owner(leaf), btrfs_header_generation(leaf),
 458  *      inode objectid)
 459  *
 460  * Btree extents can be referenced by:
 461  *
 462  * - Different subvolumes
 463  * - Different generations of the same subvolume
 464  *
 465  * When a tree block is created, back references are inserted:
 466  *
 467  * (root->root_key.objectid, trans->transid, level, 1)
 468  *
 469  * When a tree block is cow'd, new back references are added for all the
 470  * blocks it points to. If the tree block isn't in reference counted root,
 471  * the old back references are removed. These new back references are of
 472  * the form (trans->transid will have increased since creation):
 473  *
 474  * (root->root_key.objectid, trans->transid, level, 1)
 475  *
 476  * When a backref is in deleting, the following fields are checked:
 477  *
 478  * if backref was for a tree root:
 479  *     (btrfs_header_owner(itself), btrfs_header_generation(itself), level)
 480  * else
 481  *     (btrfs_header_owner(parent), btrfs_header_generation(parent), level)
 482  *
 483  * Back Reference Key composing:
 484  *
 485  * The key objectid corresponds to the first byte in the extent, the key
 486  * type is set to BTRFS_EXTENT_REF_KEY, and the key offset is the first
 487  * byte of parent extent. If a extent is tree root, the key offset is set
 488  * to the key objectid.
 489  */
 490
 491 static noinline int lookup_extent_backref(struct btrfs_trans_handle *trans,
 492                                           struct btrfs_root *root,
 493                                           struct btrfs_path *path,
 494                                           u64 bytenr, u64 parent,
 495                                           u64 ref_root, u64 ref_generation,
 496                                           u64 owner_objectid, int del)
 497 {
 498         struct btrfs_key key;
 499         struct btrfs_extent_ref *ref;
 500         struct extent_buffer *leaf;
 501         u64 ref_objectid;
 502         int ret;
 503
 504         key.objectid = bytenr;
 505         key.type = BTRFS_EXTENT_REF_KEY;
 506         key.offset = parent;
 507
 508         ret = btrfs_search_slot(trans, root, &key, path, del ? -1 : 0, 1);
 509         if (ret < 0)
 510                 goto out;
 511         if (ret > 0) {
 512                 ret = -ENOENT;
 513                 goto out;
 514         }
 515
 516         leaf = path->nodes[0];
 517         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 518         ref_objectid = btrfs_ref_objectid(leaf, ref);
 519         if (btrfs_ref_root(leaf, ref) != ref_root ||
 520             btrfs_ref_generation(leaf, ref) != ref_generation ||
 521             (ref_objectid != owner_objectid &&
 522              ref_objectid != BTRFS_MULTIPLE_OBJECTIDS)) {
 523                 ret = -EIO;
 524                 WARN_ON(1);
 525                 goto out;
 526         }
 527         ret = 0;
 528 out:
 529         return ret;
 530 }
 531
 532 /*
 533  * updates all the backrefs that are pending on update_list for the
 534  * extent_root
 535  */
 536 static noinline int update_backrefs(struct btrfs_trans_handle *trans,
 537                                     struct btrfs_root *extent_root,
 538                                     struct btrfs_path *path,
 539                                     struct list_head *update_list)
 540 {
 541         struct btrfs_key key;
 542         struct btrfs_extent_ref *ref;
 543         struct btrfs_fs_info *info = extent_root->fs_info;
 544         struct pending_extent_op *op;
 545         struct extent_buffer *leaf;
 546         int ret = 0;
 547         struct list_head *cur = update_list->next;
 548         u64 ref_objectid;
 549         u64 ref_root = extent_root->root_key.objectid;
 550
 551         op = list_entry(cur, struct pending_extent_op, list);
 552
 553 search:
 554         key.objectid = op->bytenr;
 555         key.type = BTRFS_EXTENT_REF_KEY;
 556         key.offset = op->orig_parent;
 557
 558         ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 1);
 559         BUG_ON(ret);
 560
 561         leaf = path->nodes[0];
 562
 563 loop:
 564         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 565
 566         ref_objectid = btrfs_ref_objectid(leaf, ref);
 567
 568         if (btrfs_ref_root(leaf, ref) != ref_root ||
 569             btrfs_ref_generation(leaf, ref) != op->orig_generation ||
 570             (ref_objectid != op->level &&
 571              ref_objectid != BTRFS_MULTIPLE_OBJECTIDS)) {
 572                 printk(KERN_ERR "btrfs couldn't find %llu, parent %llu, "
 573                        "root %llu, owner %u\n",
 574                        (unsigned long long)op->bytenr,
 575                        (unsigned long long)op->orig_parent,
 576                        (unsigned long long)ref_root, op->level);
 577                 btrfs_print_leaf(extent_root, leaf);
 578                 BUG();
 579         }
 580
 581         key.objectid = op->bytenr;
 582         key.offset = op->parent;
 583         key.type = BTRFS_EXTENT_REF_KEY;
 584         ret = btrfs_set_item_key_safe(trans, extent_root, path, &key);
 585         BUG_ON(ret);
 586         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 587         btrfs_set_ref_generation(leaf, ref, op->generation);
 588
 589         cur = cur->next;
 590
 591         list_del_init(&op->list);
 592         unlock_extent(&info->extent_ins, op->bytenr,
 593                       op->bytenr + op->num_bytes - 1, GFP_NOFS);
 594         kfree(op);
 595
 596         if (cur == update_list) {
 597                 btrfs_mark_buffer_dirty(path->nodes[0]);
 598                 btrfs_release_path(extent_root, path);
 599                 goto out;
 600         }
 601
 602         op = list_entry(cur, struct pending_extent_op, list);
 603
 604         path->slots[0]++;
 605         while (path->slots[0] < btrfs_header_nritems(leaf)) {
 606                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 607                 if (key.objectid == op->bytenr &&
 608                     key.type == BTRFS_EXTENT_REF_KEY)
 609                         goto loop;
 610                 path->slots[0]++;
 611         }
 612
 613         btrfs_mark_buffer_dirty(path->nodes[0]);
 614         btrfs_release_path(extent_root, path);
 615         goto search;
 616
 617 out:
 618         return 0;
 619 }
 620
 621 static noinline int insert_extents(struct btrfs_trans_handle *trans,
 622                                    struct btrfs_root *extent_root,
 623                                    struct btrfs_path *path,
 624                                    struct list_head *insert_list, int nr)
 625 {
 626         struct btrfs_key *keys;
 627         u32 *data_size;
 628         struct pending_extent_op *op;
 629         struct extent_buffer *leaf;
 630         struct list_head *cur = insert_list->next;
 631         struct btrfs_fs_info *info = extent_root->fs_info;
 632         u64 ref_root = extent_root->root_key.objectid;
 633         int i = 0, last = 0, ret;
 634         int total = nr * 2;
 635
 636         if (!nr)
 637                 return 0;
 638
 639         keys = kzalloc(total * sizeof(struct btrfs_key), GFP_NOFS);
 640         if (!keys)
 641                 return -ENOMEM;
 642
 643         data_size = kzalloc(total * sizeof(u32), GFP_NOFS);
 644         if (!data_size) {
 645                 kfree(keys);
 646                 return -ENOMEM;
 647         }
 648
 649         list_for_each_entry(op, insert_list, list) {
 650                 keys[i].objectid = op->bytenr;
 651                 keys[i].offset = op->num_bytes;
 652                 keys[i].type = BTRFS_EXTENT_ITEM_KEY;
 653                 data_size[i] = sizeof(struct btrfs_extent_item);
 654                 i++;
 655
 656                 keys[i].objectid = op->bytenr;
 657                 keys[i].offset = op->parent;
 658                 keys[i].type = BTRFS_EXTENT_REF_KEY;
 659                 data_size[i] = sizeof(struct btrfs_extent_ref);
 660                 i++;
 661         }
 662
 663         op = list_entry(cur, struct pending_extent_op, list);
 664         i = 0;
 665         while (i < total) {
 666                 int c;
 667                 ret = btrfs_insert_some_items(trans, extent_root, path,
 668                                               keys+i, data_size+i, total-i);
 669                 BUG_ON(ret < 0);
 670
 671                 if (last && ret > 1)
 672                         BUG();
 673
 674                 leaf = path->nodes[0];
 675                 for (c = 0; c < ret; c++) {
 676                         int ref_first = keys[i].type == BTRFS_EXTENT_REF_KEY;
 677
 678                         /*
 679                          * if the first item we inserted was a backref, then
 680                          * the EXTENT_ITEM will be the odd c's, else it will
 681                          * be the even c's
 682                          */
 683                         if ((ref_first && (c % 2)) ||
 684                             (!ref_first && !(c % 2))) {
 685                                 struct btrfs_extent_item *itm;
 686
 687                                 itm = btrfs_item_ptr(leaf, path->slots[0] + c,
 688                                                      struct btrfs_extent_item);
 689                                 btrfs_set_extent_refs(path->nodes[0], itm, 1);
 690                                 op->del++;
 691                         } else {
 692                                 struct btrfs_extent_ref *ref;
 693
 694                                 ref = btrfs_item_ptr(leaf, path->slots[0] + c,
 695                                                      struct btrfs_extent_ref);
 696                                 btrfs_set_ref_root(leaf, ref, ref_root);
 697                                 btrfs_set_ref_generation(leaf, ref,
 698                                                          op->generation);
 699                                 btrfs_set_ref_objectid(leaf, ref, op->level);
 700                                 btrfs_set_ref_num_refs(leaf, ref, 1);
 701                                 op->del++;
 702                         }
 703
 704                         /*
 705                          * using del to see when its ok to free up the
 706                          * pending_extent_op.  In the case where we insert the
 707                          * last item on the list in order to help do batching
 708                          * we need to not free the extent op until we actually
 709                          * insert the extent_item
 710                          */
 711                         if (op->del == 2) {
 712                                 unlock_extent(&info->extent_ins, op->bytenr,
 713                                               op->bytenr + op->num_bytes - 1,
 714                                               GFP_NOFS);
 715                                 cur = cur->next;
 716                                 list_del_init(&op->list);
 717                                 kfree(op);
 718                                 if (cur != insert_list)
 719                                         op = list_entry(cur,
 720                                                 struct pending_extent_op,
 721                                                 list);
 722                         }
 723                 }
 724                 btrfs_mark_buffer_dirty(leaf);
 725                 btrfs_release_path(extent_root, path);
 726
 727                 /*
 728                  * Ok backref's and items usually go right next to eachother,
 729                  * but if we could only insert 1 item that means that we
 730                  * inserted on the end of a leaf, and we have no idea what may
 731                  * be on the next leaf so we just play it safe.  In order to
 732                  * try and help this case we insert the last thing on our
 733                  * insert list so hopefully it will end up being the last
 734                  * thing on the leaf and everything else will be before it,
 735                  * which will let us insert a whole bunch of items at the same
 736                  * time.
 737                  */
 738                 if (ret == 1 && !last && (i + ret < total)) {
 739                         /*
 740                          * last: where we will pick up the next time around
 741                          * i: our current key to insert, will be total - 1
 742                          * cur: the current op we are screwing with
 743                          * op: duh
 744                          */
 745                         last = i + ret;
 746                         i = total - 1;
 747                         cur = insert_list->prev;
 748                         op = list_entry(cur, struct pending_extent_op, list);
 749                 } else if (last) {
 750                         /*
 751                          * ok we successfully inserted the last item on the
 752                          * list, lets reset everything
 753                          *
 754                          * i: our current key to insert, so where we left off
 755                          *    last time
 756                          * last: done with this
 757                          * cur: the op we are messing with
 758                          * op: duh
 759                          * total: since we inserted the last key, we need to
 760                          *        decrement total so we dont overflow
 761                          */
 762                         i = last;
 763                         last = 0;
 764                         total--;
 765                         if (i < total) {
 766                                 cur = insert_list->next;
 767                                 op = list_entry(cur, struct pending_extent_op,
 768                                                 list);
 769                         }
 770                 } else {
 771                         i += ret;
 772                 }
 773
 774                 cond_resched();
 775         }
 776         ret = 0;
 777         kfree(keys);
 778         kfree(data_size);
 779         return ret;
 780 }
 781
 782 static noinline int insert_extent_backref(struct btrfs_trans_handle *trans,
 783                                           struct btrfs_root *root,
 784                                           struct btrfs_path *path,
 785                                           u64 bytenr, u64 parent,
 786                                           u64 ref_root, u64 ref_generation,
 787                                           u64 owner_objectid)
 788 {
 789         struct btrfs_key key;
 790         struct extent_buffer *leaf;
 791         struct btrfs_extent_ref *ref;
 792         u32 num_refs;
 793         int ret;
 794
 795         key.objectid = bytenr;
 796         key.type = BTRFS_EXTENT_REF_KEY;
 797         key.offset = parent;
 798
 799         ret = btrfs_insert_empty_item(trans, root, path, &key, sizeof(*ref));
 800         if (ret == 0) {
 801                 leaf = path->nodes[0];
 802                 ref = btrfs_item_ptr(leaf, path->slots[0],
 803                                      struct btrfs_extent_ref);
 804                 btrfs_set_ref_root(leaf, ref, ref_root);
 805                 btrfs_set_ref_generation(leaf, ref, ref_generation);
 806                 btrfs_set_ref_objectid(leaf, ref, owner_objectid);
 807                 btrfs_set_ref_num_refs(leaf, ref, 1);
 808         } else if (ret == -EEXIST) {
 809                 u64 existing_owner;
 810                 BUG_ON(owner_objectid < BTRFS_FIRST_FREE_OBJECTID);
 811                 leaf = path->nodes[0];
 812                 ref = btrfs_item_ptr(leaf, path->slots[0],
 813                                      struct btrfs_extent_ref);
 814                 if (btrfs_ref_root(leaf, ref) != ref_root ||
 815                     btrfs_ref_generation(leaf, ref) != ref_generation) {
 816                         ret = -EIO;
 817                         WARN_ON(1);
 818                         goto out;
 819                 }
 820
 821                 num_refs = btrfs_ref_num_refs(leaf, ref);
 822                 BUG_ON(num_refs == 0);
 823                 btrfs_set_ref_num_refs(leaf, ref, num_refs + 1);
 824
 825                 existing_owner = btrfs_ref_objectid(leaf, ref);
 826                 if (existing_owner != owner_objectid &&
 827                     existing_owner != BTRFS_MULTIPLE_OBJECTIDS) {
 828                         btrfs_set_ref_objectid(leaf, ref,
 829                                         BTRFS_MULTIPLE_OBJECTIDS);
 830                 }
 831                 ret = 0;
 832         } else {
 833                 goto out;
 834         }
 835         btrfs_mark_buffer_dirty(path->nodes[0]);
 836 out:
 837         btrfs_release_path(root, path);
 838         return ret;
 839 }
 840
 841 static noinline int remove_extent_backref(struct btrfs_trans_handle *trans,
 842                                           struct btrfs_root *root,
 843                                           struct btrfs_path *path)
 844 {
 845         struct extent_buffer *leaf;
 846         struct btrfs_extent_ref *ref;
 847         u32 num_refs;
 848         int ret = 0;
 849
 850         leaf = path->nodes[0];
 851         ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_ref);
 852         num_refs = btrfs_ref_num_refs(leaf, ref);
 853         BUG_ON(num_refs == 0);
 854         num_refs -= 1;
 855         if (num_refs == 0) {
 856                 ret = btrfs_del_item(trans, root, path);
 857         } else {
 858                 btrfs_set_ref_num_refs(leaf, ref, num_refs);
 859                 btrfs_mark_buffer_dirty(leaf);
 860         }
 861         btrfs_release_path(root, path);
 862         return ret;
 863 }
 864
 865 #ifdef BIO_RW_DISCARD
 866 static void btrfs_issue_discard(struct block_device *bdev,
 867                                 u64 start, u64 len)
 868 {
 869         blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_KERNEL);
 870 }
 871 #endif
 872
 873 static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 874                                 u64 num_bytes)
 875 {
 876 #ifdef BIO_RW_DISCARD
 877         int ret;
 878         u64 map_length = num_bytes;
 879         struct btrfs_multi_bio *multi = NULL;
 880
 881         /* Tell the block device(s) that the sectors can be discarded */
 882         ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
 883                               bytenr, &map_length, &multi, 0);
 884         if (!ret) {
 885                 struct btrfs_bio_stripe *stripe = multi->stripes;
 886                 int i;
 887
 888                 if (map_length > num_bytes)
 889                         map_length = num_bytes;
 890
 891                 for (i = 0; i < multi->num_stripes; i++, stripe++) {
 892                         btrfs_issue_discard(stripe->dev->bdev,
 893                                             stripe->physical,
 894                                             map_length);
 895                 }
 896                 kfree(multi);
 897         }
 898
 899         return ret;
 900 #else
 901         return 0;
 902 #endif
 903 }
 904
 905 static noinline int free_extents(struct btrfs_trans_handle *trans,
 906                                  struct btrfs_root *extent_root,
 907                                  struct list_head *del_list)
 908 {
 909         struct btrfs_fs_info *info = extent_root->fs_info;
 910         struct btrfs_path *path;
 911         struct btrfs_key key, found_key;
 912         struct extent_buffer *leaf;
 913         struct list_head *cur;
 914         struct pending_extent_op *op;
 915         struct btrfs_extent_item *ei;
 916         int ret, num_to_del, extent_slot = 0, found_extent = 0;
 917         u32 refs;
 918         u64 bytes_freed = 0;
 919
 920         path = btrfs_alloc_path();
 921         if (!path)
 922                 return -ENOMEM;
 923         path->reada = 1;
 924
 925 search:
 926         /* search for the backref for the current ref we want to delete */
 927         cur = del_list->next;
 928         op = list_entry(cur, struct pending_extent_op, list);
 929         ret = lookup_extent_backref(trans, extent_root, path, op->bytenr,
 930                                     op->orig_parent,
 931                                     extent_root->root_key.objectid,
 932                                     op->orig_generation, op->level, 1);
 933         if (ret) {
 934                 printk(KERN_ERR "btrfs unable to find backref byte nr %llu "
 935                        "root %llu gen %llu owner %u\n",
 936                        (unsigned long long)op->bytenr,
 937                        (unsigned long long)extent_root->root_key.objectid,
 938                        (unsigned long long)op->orig_generation, op->level);
 939                 btrfs_print_leaf(extent_root, path->nodes[0]);
 940                 WARN_ON(1);
 941                 goto out;
 942         }
 943
 944         extent_slot = path->slots[0];
 945         num_to_del = 1;
 946         found_extent = 0;
 947
 948         /*
 949          * if we aren't the first item on the leaf we can move back one and see
 950          * if our ref is right next to our extent item
 951          */
 952         if (likely(extent_slot)) {
 953                 extent_slot--;
 954                 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
 955                                       extent_slot);
 956                 if (found_key.objectid == op->bytenr &&
 957                     found_key.type == BTRFS_EXTENT_ITEM_KEY &&
 958                     found_key.offset == op->num_bytes) {
 959                         num_to_del++;
 960                         found_extent = 1;
 961                 }
 962         }
 963
 964         /*
 965          * if we didn't find the extent we need to delete the backref and then
 966          * search for the extent item key so we can update its ref count
 967          */
 968         if (!found_extent) {
 969                 key.objectid = op->bytenr;
 970                 key.type = BTRFS_EXTENT_ITEM_KEY;
 971                 key.offset = op->num_bytes;
 972
 973                 ret = remove_extent_backref(trans, extent_root, path);
 974                 BUG_ON(ret);
 975                 btrfs_release_path(extent_root, path);
 976                 ret = btrfs_search_slot(trans, extent_root, &key, path, -1, 1);
 977                 BUG_ON(ret);
 978                 extent_slot = path->slots[0];
 979         }
 980
 981         /* this is where we update the ref count for the extent */
 982         leaf = path->nodes[0];
 983         ei = btrfs_item_ptr(leaf, extent_slot, struct btrfs_extent_item);
 984         refs = btrfs_extent_refs(leaf, ei);
 985         BUG_ON(refs == 0);
 986         refs--;
 987         btrfs_set_extent_refs(leaf, ei, refs);
 988
 989         btrfs_mark_buffer_dirty(leaf);
 990
 991         /*
 992          * This extent needs deleting.  The reason cur_slot is extent_slot +
 993          * num_to_del is because extent_slot points to the slot where the extent
 994          * is, and if the backref was not right next to the extent we will be
 995          * deleting at least 1 item, and will want to start searching at the
 996          * slot directly next to extent_slot.  However if we did find the
 997          * backref next to the extent item them we will be deleting at least 2
 998          * items and will want to start searching directly after the ref slot
 999          */
1000         if (!refs) {
1001                 struct list_head *pos, *n, *end;
1002                 int cur_slot = extent_slot+num_to_del;
1003                 u64 super_used;
1004                 u64 root_used;
1005
1006                 path->slots[0] = extent_slot;
1007                 bytes_freed = op->num_bytes;
1008
1009                 mutex_lock(&info->pinned_mutex);
1010                 ret = pin_down_bytes(trans, extent_root, op->bytenr,
1011                                      op->num_bytes, op->level >=
1012                                      BTRFS_FIRST_FREE_OBJECTID);
1013                 mutex_unlock(&info->pinned_mutex);
1014                 BUG_ON(ret < 0);
1015                 op->del = ret;
1016
1017                 /*
1018                  * we need to see if we can delete multiple things at once, so
1019                  * start looping through the list of extents we are wanting to
1020                  * delete and see if their extent/backref's are right next to
1021                  * eachother and the extents only have 1 ref
1022                  */
1023                 for (pos = cur->next; pos != del_list; pos = pos->next) {
1024                         struct pending_extent_op *tmp;
1025
1026                         tmp = list_entry(pos, struct pending_extent_op, list);
1027
1028                         /* we only want to delete extent+ref at this stage */
1029                         if (cur_slot >= btrfs_header_nritems(leaf) - 1)
1030                                 break;
1031
1032                         btrfs_item_key_to_cpu(leaf, &found_key, cur_slot);
1033                         if (found_key.objectid != tmp->bytenr ||
1034                             found_key.type != BTRFS_EXTENT_ITEM_KEY ||
1035                             found_key.offset != tmp->num_bytes)
1036                                 break;
1037
1038                         /* check to make sure this extent only has one ref */
1039                         ei = btrfs_item_ptr(leaf, cur_slot,
1040                                             struct btrfs_extent_item);
1041                         if (btrfs_extent_refs(leaf, ei) != 1)
1042                                 break;
1043
1044                         btrfs_item_key_to_cpu(leaf, &found_key, cur_slot+1);
1045                         if (found_key.objectid != tmp->bytenr ||
1046                             found_key.type != BTRFS_EXTENT_REF_KEY ||
1047                             found_key.offset != tmp->orig_parent)
1048                                 break;
1049
1050                         /*
1051                          * the ref is right next to the extent, we can set the
1052                          * ref count to 0 since we will delete them both now
1053                          */
1054                         btrfs_set_extent_refs(leaf, ei, 0);
1055
1056                         /* pin down the bytes for this extent */
1057                         mutex_lock(&info->pinned_mutex);
1058                         ret = pin_down_bytes(trans, extent_root, tmp->bytenr,
1059                                              tmp->num_bytes, tmp->level >=
1060                                              BTRFS_FIRST_FREE_OBJECTID);
1061                         mutex_unlock(&info->pinned_mutex);
1062                         BUG_ON(ret < 0);
1063
1064                         /*
1065                          * use the del field to tell if we need to go ahead and
1066                          * free up the extent when we delete the item or not.
1067                          */
1068                         tmp->del = ret;
1069                         bytes_freed += tmp->num_bytes;
1070
1071                         num_to_del += 2;
1072                         cur_slot += 2;
1073                 }
1074                 end = pos;
1075
1076                 /* update the free space counters */
1077                 spin_lock(&info->delalloc_lock);
1078                 super_used = btrfs_super_bytes_used(&info->super_copy);
1079                 btrfs_set_super_bytes_used(&info->super_copy,
1080                                            super_used - bytes_freed);
1081
1082                 root_used = btrfs_root_used(&extent_root->root_item);
1083                 btrfs_set_root_used(&extent_root->root_item,
1084                                     root_used - bytes_freed);
1085                 spin_unlock(&info->delalloc_lock);
1086
1087                 /* delete the items */
1088                 ret = btrfs_del_items(trans, extent_root, path,
1089                                       path->slots[0], num_to_del);
1090                 BUG_ON(ret);
1091
1092                 /*
1093                  * loop through the extents we deleted and do the cleanup work
1094                  * on them
1095                  */
1096                 for (pos = cur, n = pos->next; pos != end;
1097                      pos = n, n = pos->next) {
1098                         struct pending_extent_op *tmp;
1099                         tmp = list_entry(pos, struct pending_extent_op, list);
1100
1101                         /*
1102                          * remember tmp->del tells us wether or not we pinned
1103                          * down the extent
1104                          */
1105                         ret = update_block_group(trans, extent_root,
1106                                                  tmp->bytenr, tmp->num_bytes, 0,
1107                                                  tmp->del);
1108                         BUG_ON(ret);
1109
1110                         list_del_init(&tmp->list);
1111                         unlock_extent(&info->extent_ins, tmp->bytenr,
1112                                       tmp->bytenr + tmp->num_bytes - 1,
1113                                       GFP_NOFS);
1114                         kfree(tmp);
1115                 }
1116         } else if (refs && found_extent) {
1117                 /*
1118                  * the ref and extent were right next to eachother, but the
1119                  * extent still has a ref, so just free the backref and keep
1120                  * going
1121                  */
1122                 ret = remove_extent_backref(trans, extent_root, path);
1123                 BUG_ON(ret);
1124
1125                 list_del_init(&op->list);
1126                 unlock_extent(&info->extent_ins, op->bytenr,
1127                               op->bytenr + op->num_bytes - 1, GFP_NOFS);
1128                 kfree(op);
1129         } else {
1130                 /*
1131                  * the extent has multiple refs and the backref we were looking
1132                  * for was not right next to it, so just unlock and go next,
1133                  * we're good to go
1134                  */
1135                 list_del_init(&op->list);
1136                 unlock_extent(&info->extent_ins, op->bytenr,
1137                               op->bytenr + op->num_bytes - 1, GFP_NOFS);
1138                 kfree(op);
1139         }
1140
1141         btrfs_release_path(extent_root, path);
1142         if (!list_empty(del_list))
1143                 goto search;
1144
1145 out:
1146         btrfs_free_path(path);
1147         return ret;
1148 }
1149
1150 static int __btrfs_update_extent_ref(struct btrfs_trans_handle *trans,
1151                                      struct btrfs_root *root, u64 bytenr,
1152                                      u64 orig_parent, u64 parent,
1153                                      u64 orig_root, u64 ref_root,
1154                                      u64 orig_generation, u64 ref_generation,
1155                                      u64 owner_objectid)
1156 {
1157         int ret;
1158         struct btrfs_root *extent_root = root->fs_info->extent_root;
1159         struct btrfs_path *path;
1160
1161         if (root == root->fs_info->extent_root) {
1162                 struct pending_extent_op *extent_op;
1163                 u64 num_bytes;
1164
1165                 BUG_ON(owner_objectid >= BTRFS_MAX_LEVEL);
1166                 num_bytes = btrfs_level_size(root, (int)owner_objectid);
1167                 mutex_lock(&root->fs_info->extent_ins_mutex);
1168                 if (test_range_bit(&root->fs_info->extent_ins, bytenr,
1169                                 bytenr + num_bytes - 1, EXTENT_WRITEBACK, 0)) {
1170                         u64 priv;
1171                         ret = get_state_private(&root->fs_info->extent_ins,
1172                                                 bytenr, &priv);
1173                         BUG_ON(ret);
1174                         extent_op = (struct pending_extent_op *)
1175                                                         (unsigned long)priv;
1176                         BUG_ON(extent_op->parent != orig_parent);
1177                         BUG_ON(extent_op->generation != orig_generation);
1178
1179                         extent_op->parent = parent;
1180                         extent_op->generation = ref_generation;
1181                 } else {
1182                         extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
1183                         BUG_ON(!extent_op);
1184
1185                         extent_op->type = PENDING_BACKREF_UPDATE;
1186                         extent_op->bytenr = bytenr;
1187                         extent_op->num_bytes = num_bytes;
1188                         extent_op->parent = parent;
1189                         extent_op->orig_parent = orig_parent;
1190                         extent_op->generation = ref_generation;
1191                         extent_op->orig_generation = orig_generation;
1192                         extent_op->level = (int)owner_objectid;
1193                         INIT_LIST_HEAD(&extent_op->list);
1194                         extent_op->del = 0;
1195
1196                         set_extent_bits(&root->fs_info->extent_ins,
1197                                         bytenr, bytenr + num_bytes - 1,
1198                                         EXTENT_WRITEBACK, GFP_NOFS);
1199                         set_state_private(&root->fs_info->extent_ins,
1200                                           bytenr, (unsigned long)extent_op);
1201                 }
1202                 mutex_unlock(&root->fs_info->extent_ins_mutex);
1203                 return 0;
1204         }
1205
1206         path = btrfs_alloc_path();
1207         if (!path)
1208                 return -ENOMEM;
1209         ret = lookup_extent_backref(trans, extent_root, path,
1210                                     bytenr, orig_parent, orig_root,
1211                                     orig_generation, owner_objectid, 1);
1212         if (ret)
1213                 goto out;
1214         ret = remove_extent_backref(trans, extent_root, path);
1215         if (ret)
1216                 goto out;
1217         ret = insert_extent_backref(trans, extent_root, path, bytenr,
1218                                     parent, ref_root, ref_generation,
1219                                     owner_objectid);
1220         BUG_ON(ret);
1221         finish_current_insert(trans, extent_root, 0);
1222         del_pending_extents(trans, extent_root, 0);
1223 out:
1224         btrfs_free_path(path);
1225         return ret;
1226 }
1227
1228 int btrfs_update_extent_ref(struct btrfs_trans_handle *trans,
1229                             struct btrfs_root *root, u64 bytenr,
1230                             u64 orig_parent, u64 parent,
1231                             u64 ref_root, u64 ref_generation,
1232                             u64 owner_objectid)
1233 {
1234         int ret;
1235         if (ref_root == BTRFS_TREE_LOG_OBJECTID &&
1236             owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
1237                 return 0;
1238         ret = __btrfs_update_extent_ref(trans, root, bytenr, orig_parent,
1239                                         parent, ref_root, ref_root,
1240                                         ref_generation, ref_generation,
1241                                         owner_objectid);
1242         return ret;
1243 }
1244
1245 static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1246                                   struct btrfs_root *root, u64 bytenr,
1247                                   u64 orig_parent, u64 parent,
1248                                   u64 orig_root, u64 ref_root,
1249                                   u64 orig_generation, u64 ref_generation,
1250                                   u64 owner_objectid)
1251 {
1252         struct btrfs_path *path;
1253         int ret;
1254         struct btrfs_key key;
1255         struct extent_buffer *l;
1256         struct btrfs_extent_item *item;
1257         u32 refs;
1258
1259         path = btrfs_alloc_path();
1260         if (!path)
1261                 return -ENOMEM;
1262
1263         path->reada = 1;
1264         key.objectid = bytenr;
1265         key.type = BTRFS_EXTENT_ITEM_KEY;
1266         key.offset = (u64)-1;
1267
1268         ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
1269                                 0, 1);
1270         if (ret < 0)
1271                 return ret;
1272         BUG_ON(ret == 0 || path->slots[0] == 0);
1273
1274         path->slots[0]--;
1275         l = path->nodes[0];
1276
1277         btrfs_item_key_to_cpu(l, &key, path->slots[0]);
1278         if (key.objectid != bytenr) {
1279                 btrfs_print_leaf(root->fs_info->extent_root, path->nodes[0]);
1280                 printk(KERN_ERR "btrfs wanted %llu found %llu\n",
1281                        (unsigned long long)bytenr,
1282                        (unsigned long long)key.objectid);
1283                 BUG();
1284         }
1285         BUG_ON(key.type != BTRFS_EXTENT_ITEM_KEY);
1286
1287         item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
1288         refs = btrfs_extent_refs(l, item);
1289         btrfs_set_extent_refs(l, item, refs + 1);
1290         btrfs_mark_buffer_dirty(path->nodes[0]);
1291
1292         btrfs_release_path(root->fs_info->extent_root, path);
1293
1294         path->reada = 1;
1295         ret = insert_extent_backref(trans, root->fs_info->extent_root,
1296                                     path, bytenr, parent,
1297                                     ref_root, ref_generation,
1298                                     owner_objectid);
1299         BUG_ON(ret);
1300         finish_current_insert(trans, root->fs_info->extent_root, 0);
1301         del_pending_extents(trans, root->fs_info->extent_root, 0);
1302
1303         btrfs_free_path(path);
1304         return 0;
1305 }
1306
1307 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1308                          struct btrfs_root *root,
1309                          u64 bytenr, u64 num_bytes, u64 parent,
1310                          u64 ref_root, u64 ref_generation,
1311                          u64 owner_objectid)
1312 {
1313         int ret;
1314         if (ref_root == BTRFS_TREE_LOG_OBJECTID &&
1315             owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
1316                 return 0;
1317         ret = __btrfs_inc_extent_ref(trans, root, bytenr, 0, parent,
1318                                      0, ref_root, 0, ref_generation,
1319                                      owner_objectid);
1320         return ret;
1321 }
1322
1323 int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
1324                          struct btrfs_root *root)
1325 {
1326         finish_current_insert(trans, root->fs_info->extent_root, 1);
1327         del_pending_extents(trans, root->fs_info->extent_root, 1);
1328         return 0;
1329 }
1330
1331 int btrfs_lookup_extent_ref(struct btrfs_trans_handle *trans,
1332                             struct btrfs_root *root, u64 bytenr,
1333                             u64 num_bytes, u32 *refs)
1334 {
1335         struct btrfs_path *path;
1336         int ret;
1337         struct btrfs_key key;
1338         struct extent_buffer *l;
1339         struct btrfs_extent_item *item;
1340
1341         WARN_ON(num_bytes < root->sectorsize);
1342         path = btrfs_alloc_path();
1343         path->reada = 1;
1344         key.objectid = bytenr;
1345         key.offset = num_bytes;
1346         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
1347         ret = btrfs_search_slot(trans, root->fs_info->extent_root, &key, path,
1348                                 0, 0);
1349         if (ret < 0)
1350                 goto out;
1351         if (ret != 0) {
1352                 btrfs_print_leaf(root, path->nodes[0]);
1353                 printk(KERN_INFO "btrfs failed to find block number %llu\n",
1354                        (unsigned long long)bytenr);
1355                 BUG();
1356         }
1357         l = path->nodes[0];
1358         item = btrfs_item_ptr(l, path->slots[0], struct btrfs_extent_item);
1359         *refs = btrfs_extent_refs(l, item);
1360 out:
1361         btrfs_free_path(path);
1362         return 0;
1363 }
1364
1365 int btrfs_cross_ref_exist(struct btrfs_trans_handle *trans,
1366                           struct btrfs_root *root, u64 objectid, u64 bytenr)
1367 {
1368         struct btrfs_root *extent_root = root->fs_info->extent_root;
1369         struct btrfs_path *path;
1370         struct extent_buffer *leaf;
1371         struct btrfs_extent_ref *ref_item;
1372         struct btrfs_key key;
1373         struct btrfs_key found_key;
1374         u64 ref_root;
1375         u64 last_snapshot;
1376         u32 nritems;
1377         int ret;
1378
1379         key.objectid = bytenr;
1380         key.offset = (u64)-1;
1381         key.type = BTRFS_EXTENT_ITEM_KEY;
1382
1383         path = btrfs_alloc_path();
1384         ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
1385         if (ret < 0)
1386                 goto out;
1387         BUG_ON(ret == 0);
1388
1389         ret = -ENOENT;
1390         if (path->slots[0] == 0)
1391                 goto out;
1392
1393         path->slots[0]--;
1394         leaf = path->nodes[0];
1395         btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1396
1397         if (found_key.objectid != bytenr ||
1398             found_key.type != BTRFS_EXTENT_ITEM_KEY)
1399                 goto out;
1400
1401         last_snapshot = btrfs_root_last_snapshot(&root->root_item);
1402         while (1) {
1403                 leaf = path->nodes[0];
1404                 nritems = btrfs_header_nritems(leaf);
1405                 if (path->slots[0] >= nritems) {
1406                         ret = btrfs_next_leaf(extent_root, path);
1407                         if (ret < 0)
1408                                 goto out;
1409                         if (ret == 0)
1410                                 continue;
1411                         break;
1412                 }
1413                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1414                 if (found_key.objectid != bytenr)
1415                         break;
1416
1417                 if (found_key.type != BTRFS_EXTENT_REF_KEY) {
1418                         path->slots[0]++;
1419                         continue;
1420                 }
1421
1422                 ref_item = btrfs_item_ptr(leaf, path->slots[0],
1423                                           struct btrfs_extent_ref);
1424                 ref_root = btrfs_ref_root(leaf, ref_item);
1425                 if ((ref_root != root->root_key.objectid &&
1426                      ref_root != BTRFS_TREE_LOG_OBJECTID) ||
1427                      objectid != btrfs_ref_objectid(leaf, ref_item)) {
1428                         ret = 1;
1429                         goto out;
1430                 }
1431                 if (btrfs_ref_generation(leaf, ref_item) <= last_snapshot) {
1432                         ret = 1;
1433                         goto out;
1434                 }
1435
1436                 path->slots[0]++;
1437         }
1438         ret = 0;
1439 out:
1440         btrfs_free_path(path);
1441         return ret;
1442 }
1443
1444 int btrfs_cache_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
1445                     struct extent_buffer *buf, u32 nr_extents)
1446 {
1447         struct btrfs_key key;
1448         struct btrfs_file_extent_item *fi;
1449         u64 root_gen;
1450         u32 nritems;
1451         int i;
1452         int level;
1453         int ret = 0;
1454         int shared = 0;
1455
1456         if (!root->ref_cows)
1457                 return 0;
1458
1459         if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
1460                 shared = 0;
1461                 root_gen = root->root_key.offset;
1462         } else {
1463                 shared = 1;
1464                 root_gen = trans->transid - 1;
1465         }
1466
1467         level = btrfs_header_level(buf);
1468         nritems = btrfs_header_nritems(buf);
1469
1470         if (level == 0) {
1471                 struct btrfs_leaf_ref *ref;
1472                 struct btrfs_extent_info *info;
1473
1474                 ref = btrfs_alloc_leaf_ref(root, nr_extents);
1475                 if (!ref) {
1476                         ret = -ENOMEM;
1477                         goto out;
1478                 }
1479
1480                 ref->root_gen = root_gen;
1481                 ref->bytenr = buf->start;
1482                 ref->owner = btrfs_header_owner(buf);
1483                 ref->generation = btrfs_header_generation(buf);
1484                 ref->nritems = nr_extents;
1485                 info = ref->extents;
1486
1487                 for (i = 0; nr_extents > 0 && i < nritems; i++) {
1488                         u64 disk_bytenr;
1489                         btrfs_item_key_to_cpu(buf, &key, i);
1490                         if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1491                                 continue;
1492                         fi = btrfs_item_ptr(buf, i,
1493                                             struct btrfs_file_extent_item);
1494                         if (btrfs_file_extent_type(buf, fi) ==
1495                             BTRFS_FILE_EXTENT_INLINE)
1496                                 continue;
1497                         disk_bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1498                         if (disk_bytenr == 0)
1499                                 continue;
1500
1501                         info->bytenr = disk_bytenr;
1502                         info->num_bytes =
1503                                 btrfs_file_extent_disk_num_bytes(buf, fi);
1504                         info->objectid = key.objectid;
1505                         info->offset = key.offset;
1506                         info++;
1507                 }
1508
1509                 ret = btrfs_add_leaf_ref(root, ref, shared);
1510                 if (ret == -EEXIST && shared) {
1511                         struct btrfs_leaf_ref *old;
1512                         old = btrfs_lookup_leaf_ref(root, ref->bytenr);
1513                         BUG_ON(!old);
1514                         btrfs_remove_leaf_ref(root, old);
1515                         btrfs_free_leaf_ref(root, old);
1516                         ret = btrfs_add_leaf_ref(root, ref, shared);
1517                 }
1518                 WARN_ON(ret);
1519                 btrfs_free_leaf_ref(root, ref);
1520         }
1521 out:
1522         return ret;
1523 }
1524
1525 /* when a block goes through cow, we update the reference counts of
1526  * everything that block points to.  The internal pointers of the block
1527  * can be in just about any order, and it is likely to have clusters of
1528  * things that are close together and clusters of things that are not.
1529  *
1530  * To help reduce the seeks that come with updating all of these reference
1531  * counts, sort them by byte number before actual updates are done.
1532  *
1533  * struct refsort is used to match byte number to slot in the btree block.
1534  * we sort based on the byte number and then use the slot to actually
1535  * find the item.
1536  *
1537  * struct refsort is smaller than strcut btrfs_item and smaller than
1538  * struct btrfs_key_ptr.  Since we're currently limited to the page size
1539  * for a btree block, there's no way for a kmalloc of refsorts for a
1540  * single node to be bigger than a page.
1541  */
1542 struct refsort {
1543         u64 bytenr;
1544         u32 slot;
1545 };
1546
1547 /*
1548  * for passing into sort()
1549  */
1550 static int refsort_cmp(const void *a_void, const void *b_void)
1551 {
1552         const struct refsort *a = a_void;
1553         const struct refsort *b = b_void;
1554
1555         if (a->bytenr < b->bytenr)
1556                 return -1;
1557         if (a->bytenr > b->bytenr)
1558                 return 1;
1559         return 0;
1560 }
1561
1562
1563 noinline int btrfs_inc_ref(struct btrfs_trans_handle *trans,
1564                            struct btrfs_root *root,
1565                            struct extent_buffer *orig_buf,
1566                            struct extent_buffer *buf, u32 *nr_extents)
1567 {
1568         u64 bytenr;
1569         u64 ref_root;
1570         u64 orig_root;
1571         u64 ref_generation;
1572         u64 orig_generation;
1573         struct refsort *sorted;
1574         u32 nritems;
1575         u32 nr_file_extents = 0;
1576         struct btrfs_key key;
1577         struct btrfs_file_extent_item *fi;
1578         int i;
1579         int level;
1580         int ret = 0;
1581         int faili = 0;
1582         int refi = 0;
1583         int slot;
1584         int (*process_func)(struct btrfs_trans_handle *, struct btrfs_root *,
1585                             u64, u64, u64, u64, u64, u64, u64, u64);
1586
1587         ref_root = btrfs_header_owner(buf);
1588         ref_generation = btrfs_header_generation(buf);
1589         orig_root = btrfs_header_owner(orig_buf);
1590         orig_generation = btrfs_header_generation(orig_buf);
1591
1592         nritems = btrfs_header_nritems(buf);
1593         level = btrfs_header_level(buf);
1594
1595         sorted = kmalloc(sizeof(struct refsort) * nritems, GFP_NOFS);
1596         BUG_ON(!sorted);
1597
1598         if (root->ref_cows) {
1599                 process_func = __btrfs_inc_extent_ref;
1600         } else {
1601                 if (level == 0 &&
1602                     root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
1603                         goto out;
1604                 if (level != 0 &&
1605                     root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID)
1606                         goto out;
1607                 process_func = __btrfs_update_extent_ref;
1608         }
1609
1610         /*
1611          * we make two passes through the items.  In the first pass we
1612          * only record the byte number and slot.  Then we sort based on
1613          * byte number and do the actual work based on the sorted results
1614          */
1615         for (i = 0; i < nritems; i++) {
1616                 cond_resched();
1617                 if (level == 0) {
1618                         btrfs_item_key_to_cpu(buf, &key, i);
1619                         if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1620                                 continue;
1621                         fi = btrfs_item_ptr(buf, i,
1622                                             struct btrfs_file_extent_item);
1623                         if (btrfs_file_extent_type(buf, fi) ==
1624                             BTRFS_FILE_EXTENT_INLINE)
1625                                 continue;
1626                         bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1627                         if (bytenr == 0)
1628                                 continue;
1629
1630                         nr_file_extents++;
1631                         sorted[refi].bytenr = bytenr;
1632                         sorted[refi].slot = i;
1633                         refi++;
1634                 } else {
1635                         bytenr = btrfs_node_blockptr(buf, i);
1636                         sorted[refi].bytenr = bytenr;
1637                         sorted[refi].slot = i;
1638                         refi++;
1639                 }
1640         }
1641         /*
1642          * if refi == 0, we didn't actually put anything into the sorted
1643          * array and we're done
1644          */
1645         if (refi == 0)
1646                 goto out;
1647
1648         sort(sorted, refi, sizeof(struct refsort), refsort_cmp, NULL);
1649
1650         for (i = 0; i < refi; i++) {
1651                 cond_resched();
1652                 slot = sorted[i].slot;
1653                 bytenr = sorted[i].bytenr;
1654
1655                 if (level == 0) {
1656                         btrfs_item_key_to_cpu(buf, &key, slot);
1657
1658                         ret = process_func(trans, root, bytenr,
1659                                            orig_buf->start, buf->start,
1660                                            orig_root, ref_root,
1661                                            orig_generation, ref_generation,
1662                                            key.objectid);
1663
1664                         if (ret) {
1665                                 faili = slot;
1666                                 WARN_ON(1);
1667                                 goto fail;
1668                         }
1669                 } else {
1670                         ret = process_func(trans, root, bytenr,
1671                                            orig_buf->start, buf->start,
1672                                            orig_root, ref_root,
1673                                            orig_generation, ref_generation,
1674                                            level - 1);
1675                         if (ret) {
1676                                 faili = slot;
1677                                 WARN_ON(1);
1678                                 goto fail;
1679                         }
1680                 }
1681         }
1682 out:
1683         kfree(sorted);
1684         if (nr_extents) {
1685                 if (level == 0)
1686                         *nr_extents = nr_file_extents;
1687                 else
1688                         *nr_extents = nritems;
1689         }
1690         return 0;
1691 fail:
1692         kfree(sorted);
1693         WARN_ON(1);
1694         return ret;
1695 }
1696
1697 int btrfs_update_ref(struct btrfs_trans_handle *trans,
1698                      struct btrfs_root *root, struct extent_buffer *orig_buf,
1699                      struct extent_buffer *buf, int start_slot, int nr)
1700
1701 {
1702         u64 bytenr;
1703         u64 ref_root;
1704         u64 orig_root;
1705         u64 ref_generation;
1706         u64 orig_generation;
1707         struct btrfs_key key;
1708         struct btrfs_file_extent_item *fi;
1709         int i;
1710         int ret;
1711         int slot;
1712         int level;
1713
1714         BUG_ON(start_slot < 0);
1715         BUG_ON(start_slot + nr > btrfs_header_nritems(buf));
1716
1717         ref_root = btrfs_header_owner(buf);
1718         ref_generation = btrfs_header_generation(buf);
1719         orig_root = btrfs_header_owner(orig_buf);
1720         orig_generation = btrfs_header_generation(orig_buf);
1721         level = btrfs_header_level(buf);
1722
1723         if (!root->ref_cows) {
1724                 if (level == 0 &&
1725                     root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
1726                         return 0;
1727                 if (level != 0 &&
1728                     root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID)
1729                         return 0;
1730         }
1731
1732         for (i = 0, slot = start_slot; i < nr; i++, slot++) {
1733                 cond_resched();
1734                 if (level == 0) {
1735                         btrfs_item_key_to_cpu(buf, &key, slot);
1736                         if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
1737                                 continue;
1738                         fi = btrfs_item_ptr(buf, slot,
1739                                             struct btrfs_file_extent_item);
1740                         if (btrfs_file_extent_type(buf, fi) ==
1741                             BTRFS_FILE_EXTENT_INLINE)
1742                                 continue;
1743                         bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
1744                         if (bytenr == 0)
1745                                 continue;
1746                         ret = __btrfs_update_extent_ref(trans, root, bytenr,
1747                                             orig_buf->start, buf->start,
1748                                             orig_root, ref_root,
1749                                             orig_generation, ref_generation,
1750                                             key.objectid);
1751                         if (ret)
1752                                 goto fail;
1753                 } else {
1754                         bytenr = btrfs_node_blockptr(buf, slot);
1755                         ret = __btrfs_update_extent_ref(trans, root, bytenr,
1756                                             orig_buf->start, buf->start,
1757                                             orig_root, ref_root,
1758                                             orig_generation, ref_generation,
1759                                             level - 1);
1760                         if (ret)
1761                                 goto fail;
1762                 }
1763         }
1764         return 0;
1765 fail:
1766         WARN_ON(1);
1767         return -1;
1768 }
1769
1770 static int write_one_cache_group(struct btrfs_trans_handle *trans,
1771                                  struct btrfs_root *root,
1772                                  struct btrfs_path *path,
1773                                  struct btrfs_block_group_cache *cache)
1774 {
1775         int ret;
1776         int pending_ret;
1777         struct btrfs_root *extent_root = root->fs_info->extent_root;
1778         unsigned long bi;
1779         struct extent_buffer *leaf;
1780
1781         ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
1782         if (ret < 0)
1783                 goto fail;
1784         BUG_ON(ret);
1785
1786         leaf = path->nodes[0];
1787         bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
1788         write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
1789         btrfs_mark_buffer_dirty(leaf);
1790         btrfs_release_path(extent_root, path);
1791 fail:
1792         finish_current_insert(trans, extent_root, 0);
1793         pending_ret = del_pending_extents(trans, extent_root, 0);
1794         if (ret)
1795                 return ret;
1796         if (pending_ret)
1797                 return pending_ret;
1798         return 0;
1799
1800 }
1801
1802 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
1803                                    struct btrfs_root *root)
1804 {
1805         struct btrfs_block_group_cache *cache, *entry;
1806         struct rb_node *n;
1807         int err = 0;
1808         int werr = 0;
1809         struct btrfs_path *path;
1810         u64 last = 0;
1811
1812         path = btrfs_alloc_path();
1813         if (!path)
1814                 return -ENOMEM;
1815
1816         while (1) {
1817                 cache = NULL;
1818                 spin_lock(&root->fs_info->block_group_cache_lock);
1819                 for (n = rb_first(&root->fs_info->block_group_cache_tree);
1820                      n; n = rb_next(n)) {
1821                         entry = rb_entry(n, struct btrfs_block_group_cache,
1822                                          cache_node);
1823                         if (entry->dirty) {
1824                                 cache = entry;
1825                                 break;
1826                         }
1827                 }
1828                 spin_unlock(&root->fs_info->block_group_cache_lock);
1829
1830                 if (!cache)
1831                         break;
1832
1833                 cache->dirty = 0;
1834                 last += cache->key.offset;
1835
1836                 err = write_one_cache_group(trans, root,
1837                                             path, cache);
1838                 /*
1839                  * if we fail to write the cache group, we want
1840                  * to keep it marked dirty in hopes that a later
1841                  * write will work
1842                  */
1843                 if (err) {
1844                         werr = err;
1845                         continue;
1846                 }
1847         }
1848         btrfs_free_path(path);
1849         return werr;
1850 }
1851
1852 int btrfs_extent_readonly(struct btrfs_root *root, u64 bytenr)
1853 {
1854         struct btrfs_block_group_cache *block_group;
1855         int readonly = 0;
1856
1857         block_group = btrfs_lookup_block_group(root->fs_info, bytenr);
1858         if (!block_group || block_group->ro)
1859                 readonly = 1;
1860         if (block_group)
1861                 put_block_group(block_group);
1862         return readonly;
1863 }
1864
1865 static int update_space_info(struct btrfs_fs_info *info, u64 flags,
1866                              u64 total_bytes, u64 bytes_used,
1867                              struct btrfs_space_info **space_info)
1868 {
1869         struct btrfs_space_info *found;
1870
1871         found = __find_space_info(info, flags);
1872         if (found) {
1873                 spin_lock(&found->lock);
1874                 found->total_bytes += total_bytes;
1875                 found->bytes_used += bytes_used;
1876                 found->full = 0;
1877                 spin_unlock(&found->lock);
1878                 *space_info = found;
1879                 return 0;
1880         }
1881         found = kzalloc(sizeof(*found), GFP_NOFS);
1882         if (!found)
1883                 return -ENOMEM;
1884
1885         list_add(&found->list, &info->space_info);
1886         INIT_LIST_HEAD(&found->block_groups);
1887         init_rwsem(&found->groups_sem);
1888         spin_lock_init(&found->lock);
1889         found->flags = flags;
1890         found->total_bytes = total_bytes;
1891         found->bytes_used = bytes_used;
1892         found->bytes_pinned = 0;
1893         found->bytes_reserved = 0;
1894         found->bytes_readonly = 0;
1895         found->full = 0;
1896         found->force_alloc = 0;
1897         *space_info = found;
1898         return 0;
1899 }
1900
1901 static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
1902 {
1903         u64 extra_flags = flags & (BTRFS_BLOCK_GROUP_RAID0 |
1904                                    BTRFS_BLOCK_GROUP_RAID1 |
1905                                    BTRFS_BLOCK_GROUP_RAID10 |
1906                                    BTRFS_BLOCK_GROUP_DUP);
1907         if (extra_flags) {
1908                 if (flags & BTRFS_BLOCK_GROUP_DATA)
1909                         fs_info->avail_data_alloc_bits |= extra_flags;
1910                 if (flags & BTRFS_BLOCK_GROUP_METADATA)
1911                         fs_info->avail_metadata_alloc_bits |= extra_flags;
1912                 if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
1913                         fs_info->avail_system_alloc_bits |= extra_flags;
1914         }
1915 }
1916
1917 static void set_block_group_readonly(struct btrfs_block_group_cache *cache)
1918 {
1919         spin_lock(&cache->space_info->lock);
1920         spin_lock(&cache->lock);
1921         if (!cache->ro) {
1922                 cache->space_info->bytes_readonly += cache->key.offset -
1923                                         btrfs_block_group_used(&cache->item);
1924                 cache->ro = 1;
1925         }
1926         spin_unlock(&cache->lock);
1927         spin_unlock(&cache->space_info->lock);
1928 }
1929
1930 u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags)
1931 {
1932         u64 num_devices = root->fs_info->fs_devices->rw_devices;
1933
1934         if (num_devices == 1)
1935                 flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0);
1936         if (num_devices < 4)
1937                 flags &= ~BTRFS_BLOCK_GROUP_RAID10;
1938
1939         if ((flags & BTRFS_BLOCK_GROUP_DUP) &&
1940             (flags & (BTRFS_BLOCK_GROUP_RAID1 |
1941                       BTRFS_BLOCK_GROUP_RAID10))) {
1942                 flags &= ~BTRFS_BLOCK_GROUP_DUP;
1943         }
1944
1945         if ((flags & BTRFS_BLOCK_GROUP_RAID1) &&
1946             (flags & BTRFS_BLOCK_GROUP_RAID10)) {
1947                 flags &= ~BTRFS_BLOCK_GROUP_RAID1;
1948         }
1949
1950         if ((flags & BTRFS_BLOCK_GROUP_RAID0) &&
1951             ((flags & BTRFS_BLOCK_GROUP_RAID1) |
1952              (flags & BTRFS_BLOCK_GROUP_RAID10) |
1953              (flags & BTRFS_BLOCK_GROUP_DUP)))
1954                 flags &= ~BTRFS_BLOCK_GROUP_RAID0;
1955         return flags;
1956 }
1957
1958 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
1959                           struct btrfs_root *extent_root, u64 alloc_bytes,
1960                           u64 flags, int force)
1961 {
1962         struct btrfs_space_info *space_info;
1963         u64 thresh;
1964         int ret = 0;
1965
1966         mutex_lock(&extent_root->fs_info->chunk_mutex);
1967
1968         flags = btrfs_reduce_alloc_profile(extent_root, flags);
1969
1970         space_info = __find_space_info(extent_root->fs_info, flags);
1971         if (!space_info) {
1972                 ret = update_space_info(extent_root->fs_info, flags,
1973                                         0, 0, &space_info);
1974                 BUG_ON(ret);
1975         }
1976         BUG_ON(!space_info);
1977
1978         spin_lock(&space_info->lock);
1979         if (space_info->force_alloc) {
1980                 force = 1;
1981                 space_info->force_alloc = 0;
1982         }
1983         if (space_info->full) {
1984                 spin_unlock(&space_info->lock);
1985                 goto out;
1986         }
1987
1988         thresh = space_info->total_bytes - space_info->bytes_readonly;
1989         thresh = div_factor(thresh, 6);
1990         if (!force &&
1991            (space_info->bytes_used + space_info->bytes_pinned +
1992             space_info->bytes_reserved + alloc_bytes) < thresh) {
1993                 spin_unlock(&space_info->lock);
1994                 goto out;
1995         }
1996         spin_unlock(&space_info->lock);
1997
1998         ret = btrfs_alloc_chunk(trans, extent_root, flags);
1999         if (ret)
2000                 space_info->full = 1;
2001 out:
2002         mutex_unlock(&extent_root->fs_info->chunk_mutex);
2003         return ret;
2004 }
2005
2006 static int update_block_group(struct btrfs_trans_handle *trans,
2007                               struct btrfs_root *root,
2008                               u64 bytenr, u64 num_bytes, int alloc,
2009                               int mark_free)
2010 {
2011         struct btrfs_block_group_cache *cache;
2012         struct btrfs_fs_info *info = root->fs_info;
2013         u64 total = num_bytes;
2014         u64 old_val;
2015         u64 byte_in_group;
2016
2017         while (total) {
2018                 cache = btrfs_lookup_block_group(info, bytenr);
2019                 if (!cache)
2020                         return -1;
2021                 byte_in_group = bytenr - cache->key.objectid;
2022                 WARN_ON(byte_in_group > cache->key.offset);
2023
2024                 spin_lock(&cache->space_info->lock);
2025                 spin_lock(&cache->lock);
2026                 cache->dirty = 1;
2027                 old_val = btrfs_block_group_used(&cache->item);
2028                 num_bytes = min(total, cache->key.offset - byte_in_group);
2029                 if (alloc) {
2030                         old_val += num_bytes;
2031                         cache->space_info->bytes_used += num_bytes;
2032                         if (cache->ro)
2033                                 cache->space_info->bytes_readonly -= num_bytes;
2034                         btrfs_set_block_group_used(&cache->item, old_val);
2035                         spin_unlock(&cache->lock);
2036                         spin_unlock(&cache->space_info->lock);
2037                 } else {
2038                         old_val -= num_bytes;
2039                         cache->space_info->bytes_used -= num_bytes;
2040                         if (cache->ro)
2041                                 cache->space_info->bytes_readonly += num_bytes;
2042                         btrfs_set_block_group_used(&cache->item, old_val);
2043                         spin_unlock(&cache->lock);
2044                         spin_unlock(&cache->space_info->lock);
2045                         if (mark_free) {
2046                                 int ret;
2047
2048                                 ret = btrfs_discard_extent(root, bytenr,
2049                                                            num_bytes);
2050                                 WARN_ON(ret);
2051
2052                                 ret = btrfs_add_free_space(cache, bytenr,
2053                                                            num_bytes);
2054                                 WARN_ON(ret);
2055                         }
2056                 }
2057                 put_block_group(cache);
2058                 total -= num_bytes;
2059                 bytenr += num_bytes;
2060         }
2061         return 0;
2062 }
2063
2064 static u64 first_logical_byte(struct btrfs_root *root, u64 search_start)
2065 {
2066         struct btrfs_block_group_cache *cache;
2067         u64 bytenr;
2068
2069         cache = btrfs_lookup_first_block_group(root->fs_info, search_start);
2070         if (!cache)
2071                 return 0;
2072
2073         bytenr = cache->key.objectid;
2074         put_block_group(cache);
2075
2076         return bytenr;
2077 }
2078
2079 int btrfs_update_pinned_extents(struct btrfs_root *root,
2080                                 u64 bytenr, u64 num, int pin)
2081 {
2082         u64 len;
2083         struct btrfs_block_group_cache *cache;
2084         struct btrfs_fs_info *fs_info = root->fs_info;
2085
2086         WARN_ON(!mutex_is_locked(&root->fs_info->pinned_mutex));
2087         if (pin) {
2088                 set_extent_dirty(&fs_info->pinned_extents,
2089                                 bytenr, bytenr + num - 1, GFP_NOFS);
2090         } else {
2091                 clear_extent_dirty(&fs_info->pinned_extents,
2092                                 bytenr, bytenr + num - 1, GFP_NOFS);
2093         }
2094         while (num > 0) {
2095                 cache = btrfs_lookup_block_group(fs_info, bytenr);
2096                 BUG_ON(!cache);
2097                 len = min(num, cache->key.offset -
2098                           (bytenr - cache->key.objectid));
2099                 if (pin) {
2100                         spin_lock(&cache->space_info->lock);
2101                         spin_lock(&cache->lock);
2102                         cache->pinned += len;
2103                         cache->space_info->bytes_pinned += len;
2104                         spin_unlock(&cache->lock);
2105                         spin_unlock(&cache->space_info->lock);
2106                         fs_info->total_pinned += len;
2107                 } else {
2108                         spin_lock(&cache->space_info->lock);
2109                         spin_lock(&cache->lock);
2110                         cache->pinned -= len;
2111                         cache->space_info->bytes_pinned -= len;
2112                         spin_unlock(&cache->lock);
2113                         spin_unlock(&cache->space_info->lock);
2114                         fs_info->total_pinned -= len;
2115                         if (cache->cached)
2116                                 btrfs_add_free_space(cache, bytenr, len);
2117                 }
2118                 put_block_group(cache);
2119                 bytenr += len;
2120                 num -= len;
2121         }
2122         return 0;
2123 }
2124
2125 static int update_reserved_extents(struct btrfs_root *root,
2126                                    u64 bytenr, u64 num, int reserve)
2127 {
2128         u64 len;
2129         struct btrfs_block_group_cache *cache;
2130         struct btrfs_fs_info *fs_info = root->fs_info;
2131
2132         while (num > 0) {
2133                 cache = btrfs_lookup_block_group(fs_info, bytenr);
2134                 BUG_ON(!cache);
2135                 len = min(num, cache->key.offset -
2136                           (bytenr - cache->key.objectid));
2137
2138                 spin_lock(&cache->space_info->lock);
2139                 spin_lock(&cache->lock);
2140                 if (reserve) {
2141                         cache->reserved += len;
2142                         cache->space_info->bytes_reserved += len;
2143                 } else {
2144                         cache->reserved -= len;
2145                         cache->space_info->bytes_reserved -= len;
2146                 }
2147                 spin_unlock(&cache->lock);
2148                 spin_unlock(&cache->space_info->lock);
2149                 put_block_group(cache);
2150                 bytenr += len;
2151                 num -= len;
2152         }
2153         return 0;
2154 }
2155
2156 int btrfs_copy_pinned(struct btrfs_root *root, struct extent_io_tree *copy)
2157 {
2158         u64 last = 0;
2159         u64 start;
2160         u64 end;
2161         struct extent_io_tree *pinned_extents = &root->fs_info->pinned_extents;
2162         int ret;
2163
2164         mutex_lock(&root->fs_info->pinned_mutex);
2165         while (1) {
2166                 ret = find_first_extent_bit(pinned_extents, last,
2167                                             &start, &end, EXTENT_DIRTY);
2168                 if (ret)
2169                         break;
2170                 set_extent_dirty(copy, start, end, GFP_NOFS);
2171                 last = end + 1;
2172         }
2173         mutex_unlock(&root->fs_info->pinned_mutex);
2174         return 0;
2175 }
2176
2177 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
2178                                struct btrfs_root *root,
2179                                struct extent_io_tree *unpin)
2180 {
2181         u64 start;
2182         u64 end;
2183         int ret;
2184
2185         mutex_lock(&root->fs_info->pinned_mutex);
2186         while (1) {
2187                 ret = find_first_extent_bit(unpin, 0, &start, &end,
2188                                             EXTENT_DIRTY);
2189                 if (ret)
2190                         break;
2191
2192                 ret = btrfs_discard_extent(root, start, end + 1 - start);
2193
2194                 btrfs_update_pinned_extents(root, start, end + 1 - start, 0);
2195                 clear_extent_dirty(unpin, start, end, GFP_NOFS);
2196
2197                 if (need_resched()) {
2198                         mutex_unlock(&root->fs_info->pinned_mutex);
2199                         cond_resched();
2200                         mutex_lock(&root->fs_info->pinned_mutex);
2201                 }
2202         }
2203         mutex_unlock(&root->fs_info->pinned_mutex);
2204         return ret;
2205 }
2206
2207 static int finish_current_insert(struct btrfs_trans_handle *trans,
2208                                  struct btrfs_root *extent_root, int all)
2209 {
2210         u64 start;
2211         u64 end;
2212         u64 priv;
2213         u64 search = 0;
2214         u64 skipped = 0;
2215         struct btrfs_fs_info *info = extent_root->fs_info;
2216         struct btrfs_path *path;
2217         struct pending_extent_op *extent_op, *tmp;
2218         struct list_head insert_list, update_list;
2219         int ret;
2220         int num_inserts = 0, max_inserts;
2221
2222         path = btrfs_alloc_path();
2223         INIT_LIST_HEAD(&insert_list);
2224         INIT_LIST_HEAD(&update_list);
2225
2226         max_inserts = extent_root->leafsize /
2227                 (2 * sizeof(struct btrfs_key) + 2 * sizeof(struct btrfs_item) +
2228                  sizeof(struct btrfs_extent_ref) +
2229                  sizeof(struct btrfs_extent_item));
2230 again:
2231         mutex_lock(&info->extent_ins_mutex);
2232         while (1) {
2233                 ret = find_first_extent_bit(&info->extent_ins, search, &start,
2234                                             &end, EXTENT_WRITEBACK);
2235                 if (ret) {
2236                         if (skipped && all && !num_inserts &&
2237                             list_empty(&update_list)) {
2238                                 skipped = 0;
2239                                 search = 0;
2240                                 continue;
2241                         }
2242                         mutex_unlock(&info->extent_ins_mutex);
2243                         break;
2244                 }
2245
2246                 ret = try_lock_extent(&info->extent_ins, start, end, GFP_NOFS);
2247                 if (!ret) {
2248                         skipped = 1;
2249                         search = end + 1;
2250                         if (need_resched()) {
2251                                 mutex_unlock(&info->extent_ins_mutex);
2252                                 cond_resched();
2253                                 mutex_lock(&info->extent_ins_mutex);
2254                         }
2255                         continue;
2256                 }
2257
2258                 ret = get_state_private(&info->extent_ins, start, &priv);
2259                 BUG_ON(ret);
2260                 extent_op = (struct pending_extent_op *)(unsigned long) priv;
2261
2262                 if (extent_op->type == PENDING_EXTENT_INSERT) {
2263                         num_inserts++;
2264                         list_add_tail(&extent_op->list, &insert_list);
2265                         search = end + 1;
2266                         if (num_inserts == max_inserts) {
2267                                 mutex_unlock(&info->extent_ins_mutex);
2268                                 break;
2269                         }
2270                 } else if (extent_op->type == PENDING_BACKREF_UPDATE) {
2271                         list_add_tail(&extent_op->list, &update_list);
2272                         search = end + 1;
2273                 } else {
2274                         BUG();
2275                 }
2276         }
2277
2278         /*
2279          * process the update list, clear the writeback bit for it, and if
2280          * somebody marked this thing for deletion then just unlock it and be
2281          * done, the free_extents will handle it
2282          */
2283         mutex_lock(&info->extent_ins_mutex);
2284         list_for_each_entry_safe(extent_op, tmp, &update_list, list) {
2285                 clear_extent_bits(&info->extent_ins, extent_op->bytenr,
2286                                   extent_op->bytenr + extent_op->num_bytes - 1,
2287                                   EXTENT_WRITEBACK, GFP_NOFS);
2288                 if (extent_op->del) {
2289                         list_del_init(&extent_op->list);
2290                         unlock_extent(&info->extent_ins, extent_op->bytenr,
2291                                       extent_op->bytenr + extent_op->num_bytes
2292                                       - 1, GFP_NOFS);
2293                         kfree(extent_op);
2294                 }
2295         }
2296         mutex_unlock(&info->extent_ins_mutex);
2297
2298         /*
2299          * still have things left on the update list, go ahead an update
2300          * everything
2301          */
2302         if (!list_empty(&update_list)) {
2303                 ret = update_backrefs(trans, extent_root, path, &update_list);
2304                 BUG_ON(ret);
2305         }
2306
2307         /*
2308          * if no inserts need to be done, but we skipped some extents and we
2309          * need to make sure everything is cleaned then reset everything and
2310          * go back to the beginning
2311          */
2312         if (!num_inserts && all && skipped) {
2313                 search = 0;
2314                 skipped = 0;
2315                 INIT_LIST_HEAD(&update_list);
2316                 INIT_LIST_HEAD(&insert_list);
2317                 goto again;
2318         } else if (!num_inserts) {
2319                 goto out;
2320         }
2321
2322         /*
2323          * process the insert extents list.  Again if we are deleting this
2324          * extent, then just unlock it, pin down the bytes if need be, and be
2325          * done with it.  Saves us from having to actually insert the extent
2326          * into the tree and then subsequently come along and delete it
2327          */
2328         mutex_lock(&info->extent_ins_mutex);
2329         list_for_each_entry_safe(extent_op, tmp, &insert_list, list) {
2330                 clear_extent_bits(&info->extent_ins, extent_op->bytenr,
2331                                   extent_op->bytenr + extent_op->num_bytes - 1,
2332                                   EXTENT_WRITEBACK, GFP_NOFS);
2333                 if (extent_op->del) {
2334                         u64 used;
2335                         list_del_init(&extent_op->list);
2336                         unlock_extent(&info->extent_ins, extent_op->bytenr,
2337                                       extent_op->bytenr + extent_op->num_bytes
2338                                       - 1, GFP_NOFS);
2339
2340                         mutex_lock(&extent_root->fs_info->pinned_mutex);
2341                         ret = pin_down_bytes(trans, extent_root,
2342                                              extent_op->bytenr,
2343                                              extent_op->num_bytes, 0);
2344                         mutex_unlock(&extent_root->fs_info->pinned_mutex);
2345
2346                         spin_lock(&info->delalloc_lock);
2347                         used = btrfs_super_bytes_used(&info->super_copy);
2348                         btrfs_set_super_bytes_used(&info->super_copy,
2349                                         used - extent_op->num_bytes);
2350                         used = btrfs_root_used(&extent_root->root_item);
2351                         btrfs_set_root_used(&extent_root->root_item,
2352                                         used - extent_op->num_bytes);
2353                         spin_unlock(&info->delalloc_lock);
2354
2355                         ret = update_block_group(trans, extent_root,
2356                                                  extent_op->bytenr,
2357                                                  extent_op->num_bytes,
2358                                                  0, ret > 0);
2359                         BUG_ON(ret);
2360                         kfree(extent_op);
2361                         num_inserts--;
2362                 }
2363         }
2364         mutex_unlock(&info->extent_ins_mutex);
2365
2366         ret = insert_extents(trans, extent_root, path, &insert_list,
2367                              num_inserts);
2368         BUG_ON(ret);
2369
2370         /*
2371          * if we broke out of the loop in order to insert stuff because we hit
2372          * the maximum number of inserts at a time we can handle, then loop
2373          * back and pick up where we left off
2374          */
2375         if (num_inserts == max_inserts) {
2376                 INIT_LIST_HEAD(&insert_list);
2377                 INIT_LIST_HEAD(&update_list);
2378                 num_inserts = 0;
2379                 goto again;
2380         }
2381
2382         /*
2383          * again, if we need to make absolutely sure there are no more pending
2384          * extent operations left and we know that we skipped some, go back to
2385          * the beginning and do it all again
2386          */
2387         if (all && skipped) {
2388                 INIT_LIST_HEAD(&insert_list);
2389                 INIT_LIST_HEAD(&update_list);
2390                 search = 0;
2391                 skipped = 0;
2392                 num_inserts = 0;
2393                 goto again;
2394         }
2395 out:
2396         btrfs_free_path(path);
2397         return 0;
2398 }
2399
2400 static int pin_down_bytes(struct btrfs_trans_handle *trans,
2401                           struct btrfs_root *root,
2402                           u64 bytenr, u64 num_bytes, int is_data)
2403 {
2404         int err = 0;
2405         struct extent_buffer *buf;
2406
2407         if (is_data)
2408                 goto pinit;
2409
2410         buf = btrfs_find_tree_block(root, bytenr, num_bytes);
2411         if (!buf)
2412                 goto pinit;
2413
2414         /* we can reuse a block if it hasn't been written
2415          * and it is from this transaction.  We can't
2416          * reuse anything from the tree log root because
2417          * it has tiny sub-transactions.
2418          */
2419         if (btrfs_buffer_uptodate(buf, 0) &&
2420             btrfs_try_tree_lock(buf)) {
2421                 u64 header_owner = btrfs_header_owner(buf);
2422                 u64 header_transid = btrfs_header_generation(buf);
2423                 if (header_owner != BTRFS_TREE_LOG_OBJECTID &&
2424                     header_owner != BTRFS_TREE_RELOC_OBJECTID &&
2425                     header_transid == trans->transid &&
2426                     !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
2427                         clean_tree_block(NULL, root, buf);
2428                         btrfs_tree_unlock(buf);
2429                         free_extent_buffer(buf);
2430                         return 1;
2431                 }
2432                 btrfs_tree_unlock(buf);
2433         }
2434         free_extent_buffer(buf);
2435 pinit:
2436         btrfs_update_pinned_extents(root, bytenr, num_bytes, 1);
2437
2438         BUG_ON(err < 0);
2439         return 0;
2440 }
2441
2442 /*
2443  * remove an extent from the root, returns 0 on success
2444  */
2445 static int __free_extent(struct btrfs_trans_handle *trans,
2446                          struct btrfs_root *root,
2447                          u64 bytenr, u64 num_bytes, u64 parent,
2448                          u64 root_objectid, u64 ref_generation,
2449                          u64 owner_objectid, int pin, int mark_free)
2450 {
2451         struct btrfs_path *path;
2452         struct btrfs_key key;
2453         struct btrfs_fs_info *info = root->fs_info;
2454         struct btrfs_root *extent_root = info->extent_root;
2455         struct extent_buffer *leaf;
2456         int ret;
2457         int extent_slot = 0;
2458         int found_extent = 0;
2459         int num_to_del = 1;
2460         struct btrfs_extent_item *ei;
2461         u32 refs;
2462
2463         key.objectid = bytenr;
2464         btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
2465         key.offset = num_bytes;
2466         path = btrfs_alloc_path();
2467         if (!path)
2468                 return -ENOMEM;
2469
2470         path->reada = 1;
2471         ret = lookup_extent_backref(trans, extent_root, path,
2472                                     bytenr, parent, root_objectid,
2473                                     ref_generation, owner_objectid, 1);
2474         if (ret == 0) {
2475                 struct btrfs_key found_key;
2476                 extent_slot = path->slots[0];
2477                 while (extent_slot > 0) {
2478                         extent_slot--;
2479                         btrfs_item_key_to_cpu(path->nodes[0], &found_key,
2480                                               extent_slot);
2481                         if (found_key.objectid != bytenr)
2482                                 break;
2483                         if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
2484                             found_key.offset == num_bytes) {
2485                                 found_extent = 1;
2486                                 break;
2487                         }
2488                         if (path->slots[0] - extent_slot > 5)
2489                                 break;
2490                 }
2491                 if (!found_extent) {
2492                         ret = remove_extent_backref(trans, extent_root, path);
2493                         BUG_ON(ret);
2494                         btrfs_release_path(extent_root, path);
2495                         ret = btrfs_search_slot(trans, extent_root,
2496                                                 &key, path, -1, 1);
2497                         if (ret) {
2498                                 printk(KERN_ERR "umm, got %d back from search"
2499                                        ", was looking for %llu\n", ret,
2500                                        (unsigned long long)bytenr);
2501                                 btrfs_print_leaf(extent_root, path->nodes[0]);
2502                         }
2503                         BUG_ON(ret);
2504                         extent_slot = path->slots[0];
2505                 }
2506         } else {
2507                 btrfs_print_leaf(extent_root, path->nodes[0]);
2508                 WARN_ON(1);
2509                 printk(KERN_ERR "btrfs unable to find ref byte nr %llu "
2510                        "root %llu gen %llu owner %llu\n",
2511                        (unsigned long long)bytenr,
2512                        (unsigned long long)root_objectid,
2513                        (unsigned long long)ref_generation,
2514                        (unsigned long long)owner_objectid);
2515         }
2516
2517         leaf = path->nodes[0];
2518         ei = btrfs_item_ptr(leaf, extent_slot,
2519                             struct btrfs_extent_item);
2520         refs = btrfs_extent_refs(leaf, ei);
2521         BUG_ON(refs == 0);
2522         refs -= 1;
2523         btrfs_set_extent_refs(leaf, ei, refs);
2524
2525         btrfs_mark_buffer_dirty(leaf);
2526
2527         if (refs == 0 && found_extent && path->slots[0] == extent_slot + 1) {
2528                 struct btrfs_extent_ref *ref;
2529                 ref = btrfs_item_ptr(leaf, path->slots[0],
2530                                      struct btrfs_extent_ref);
2531                 BUG_ON(btrfs_ref_num_refs(leaf, ref) != 1);
2532                 /* if the back ref and the extent are next to each other
2533                  * they get deleted below in one shot
2534                  */
2535                 path->slots[0] = extent_slot;
2536                 num_to_del = 2;
2537         } else if (found_extent) {
2538                 /* otherwise delete the extent back ref */
2539                 ret = remove_extent_backref(trans, extent_root, path);
2540                 BUG_ON(ret);
2541                 /* if refs are 0, we need to setup the path for deletion */
2542                 if (refs == 0) {
2543                         btrfs_release_path(extent_root, path);
2544                         ret = btrfs_search_slot(trans, extent_root, &key, path,
2545                                                 -1, 1);
2546                         BUG_ON(ret);
2547                 }
2548         }
2549
2550         if (refs == 0) {
2551                 u64 super_used;
2552                 u64 root_used;
2553
2554                 if (pin) {
2555                         mutex_lock(&root->fs_info->pinned_mutex);
2556                         ret = pin_down_bytes(trans, root, bytenr, num_bytes,
2557                                 owner_objectid >= BTRFS_FIRST_FREE_OBJECTID);
2558                         mutex_unlock(&root->fs_info->pinned_mutex);
2559                         if (ret > 0)
2560                                 mark_free = 1;
2561                         BUG_ON(ret < 0);
2562                 }
2563                 /* block accounting for super block */
2564                 spin_lock(&info->delalloc_lock);
2565                 super_used = btrfs_super_bytes_used(&info->super_copy);
2566                 btrfs_set_super_bytes_used(&info->super_copy,
2567                                            super_used - num_bytes);
2568
2569                 /* block accounting for root item */
2570                 root_used = btrfs_root_used(&root->root_item);
2571                 btrfs_set_root_used(&root->root_item,
2572                                            root_used - num_bytes);
2573                 spin_unlock(&info->delalloc_lock);
2574                 ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
2575                                       num_to_del);
2576                 BUG_ON(ret);
2577                 btrfs_release_path(extent_root, path);
2578
2579                 if (owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
2580                         ret = btrfs_del_csums(trans, root, bytenr, num_bytes);
2581                         BUG_ON(ret);
2582                 }
2583
2584                 ret = update_block_group(trans, root, bytenr, num_bytes, 0,
2585                                          mark_free);
2586                 BUG_ON(ret);
2587         }
2588         btrfs_free_path(path);
2589         finish_current_insert(trans, extent_root, 0);
2590         return ret;
2591 }
2592
2593 /*
2594  * find all the blocks marked as pending in the radix tree and remove
2595  * them from the extent map
2596  */
2597 static int del_pending_extents(struct btrfs_trans_handle *trans,
2598                                struct btrfs_root *extent_root, int all)
2599 {
2600         int ret;
2601         int err = 0;
2602         u64 start;
2603         u64 end;
2604         u64 priv;
2605         u64 search = 0;
2606         int nr = 0, skipped = 0;
2607         struct extent_io_tree *pending_del;
2608         struct extent_io_tree *extent_ins;
2609         struct pending_extent_op *extent_op;
2610         struct btrfs_fs_info *info = extent_root->fs_info;
2611         struct list_head delete_list;
2612
2613         INIT_LIST_HEAD(&delete_list);
2614         extent_ins = &extent_root->fs_info->extent_ins;
2615         pending_del = &extent_root->fs_info->pending_del;
2616
2617 again:
2618         mutex_lock(&info->extent_ins_mutex);
2619         while (1) {
2620                 ret = find_first_extent_bit(pending_del, search, &start, &end,
2621                                             EXTENT_WRITEBACK);
2622                 if (ret) {
2623                         if (all && skipped && !nr) {
2624                                 search = 0;
2625                                 skipped = 0;
2626                                 continue;
2627                         }
2628                         mutex_unlock(&info->extent_ins_mutex);
2629                         break;
2630                 }
2631
2632                 ret = try_lock_extent(extent_ins, start, end, GFP_NOFS);
2633                 if (!ret) {
2634                         search = end+1;
2635                         skipped = 1;
2636
2637                         if (need_resched()) {
2638                                 mutex_unlock(&info->extent_ins_mutex);
2639                                 cond_resched();
2640                                 mutex_lock(&info->extent_ins_mutex);
2641                         }
2642
2643                         continue;
2644                 }
2645                 BUG_ON(ret < 0);
2646
2647                 ret = get_state_private(pending_del, start, &priv);
2648                 BUG_ON(ret);
2649                 extent_op = (struct pending_extent_op *)(unsigned long)priv;
2650
2651                 clear_extent_bits(pending_del, start, end, EXTENT_WRITEBACK,
2652                                   GFP_NOFS);
2653                 if (!test_range_bit(extent_ins, start, end,
2654                                     EXTENT_WRITEBACK, 0)) {
2655                         list_add_tail(&extent_op->list, &delete_list);
2656                         nr++;
2657                 } else {
2658                         kfree(extent_op);
2659
2660                         ret = get_state_private(&info->extent_ins, start,
2661                                                 &priv);
2662                         BUG_ON(ret);
2663                         extent_op = (struct pending_extent_op *)
2664                                                 (unsigned long)priv;
2665
2666                         clear_extent_bits(&info->extent_ins, start, end,
2667                                           EXTENT_WRITEBACK, GFP_NOFS);
2668
2669                         if (extent_op->type == PENDING_BACKREF_UPDATE) {
2670                                 list_add_tail(&extent_op->list, &delete_list);
2671                                 search = end + 1;
2672                                 nr++;
2673                                 continue;
2674                         }
2675
2676                         mutex_lock(&extent_root->fs_info->pinned_mutex);
2677                         ret = pin_down_bytes(trans, extent_root, start,
2678                                              end + 1 - start, 0);
2679                         mutex_unlock(&extent_root->fs_info->pinned_mutex);
2680
2681                         ret = update_block_group(trans, extent_root, start,
2682                                                 end + 1 - start, 0, ret > 0);
2683
2684                         unlock_extent(extent_ins, start, end, GFP_NOFS);
2685                         BUG_ON(ret);
2686                         kfree(extent_op);
2687                 }
2688                 if (ret)
2689                         err = ret;
2690
2691                 search = end + 1;
2692
2693                 if (need_resched()) {
2694                         mutex_unlock(&info->extent_ins_mutex);
2695                         cond_resched();
2696                         mutex_lock(&info->extent_ins_mutex);
2697                 }
2698         }
2699
2700         if (nr) {
2701                 ret = free_extents(trans, extent_root, &delete_list);
2702                 BUG_ON(ret);
2703         }
2704
2705         if (all && skipped) {
2706                 INIT_LIST_HEAD(&delete_list);
2707                 search = 0;
2708                 nr = 0;
2709                 goto again;
2710         }
2711
2712         return err;
2713 }
2714
2715 /*
2716  * remove an extent from the root, returns 0 on success
2717  */
2718 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
2719                                struct btrfs_root *root,
2720                                u64 bytenr, u64 num_bytes, u64 parent,
2721                                u64 root_objectid, u64 ref_generation,
2722                                u64 owner_objectid, int pin)
2723 {
2724         struct btrfs_root *extent_root = root->fs_info->extent_root;
2725         int pending_ret;
2726         int ret;
2727
2728         WARN_ON(num_bytes < root->sectorsize);
2729         if (root == extent_root) {
2730                 struct pending_extent_op *extent_op = NULL;
2731
2732                 mutex_lock(&root->fs_info->extent_ins_mutex);
2733                 if (test_range_bit(&root->fs_info->extent_ins, bytenr,
2734                                 bytenr + num_bytes - 1, EXTENT_WRITEBACK, 0)) {
2735                         u64 priv;
2736                         ret = get_state_private(&root->fs_info->extent_ins,
2737                                                 bytenr, &priv);
2738                         BUG_ON(ret);
2739                         extent_op = (struct pending_extent_op *)
2740                                                 (unsigned long)priv;
2741
2742                         extent_op->del = 1;
2743                         if (extent_op->type == PENDING_EXTENT_INSERT) {
2744                                 mutex_unlock(&root->fs_info->extent_ins_mutex);
2745                                 return 0;
2746                         }
2747                 }
2748
2749                 if (extent_op) {
2750                         ref_generation = extent_op->orig_generation;
2751                         parent = extent_op->orig_parent;
2752                 }
2753
2754                 extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
2755                 BUG_ON(!extent_op);
2756
2757                 extent_op->type = PENDING_EXTENT_DELETE;
2758                 extent_op->bytenr = bytenr;
2759                 extent_op->num_bytes = num_bytes;
2760                 extent_op->parent = parent;
2761                 extent_op->orig_parent = parent;
2762                 extent_op->generation = ref_generation;
2763                 extent_op->orig_generation = ref_generation;
2764                 extent_op->level = (int)owner_objectid;
2765                 INIT_LIST_HEAD(&extent_op->list);
2766                 extent_op->del = 0;
2767
2768                 set_extent_bits(&root->fs_info->pending_del,
2769                                 bytenr, bytenr + num_bytes - 1,
2770                                 EXTENT_WRITEBACK, GFP_NOFS);
2771                 set_state_private(&root->fs_info->pending_del,
2772                                   bytenr, (unsigned long)extent_op);
2773                 mutex_unlock(&root->fs_info->extent_ins_mutex);
2774                 return 0;
2775         }
2776         /* if metadata always pin */
2777         if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID) {
2778                 if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
2779                         mutex_lock(&root->fs_info->pinned_mutex);
2780                         btrfs_update_pinned_extents(root, bytenr, num_bytes, 1);
2781                         mutex_unlock(&root->fs_info->pinned_mutex);
2782                         update_reserved_extents(root, bytenr, num_bytes, 0);
2783                         return 0;
2784                 }
2785                 pin = 1;
2786         }
2787
2788         /* if data pin when any transaction has committed this */
2789         if (ref_generation != trans->transid)
2790                 pin = 1;
2791
2792         ret = __free_extent(trans, root, bytenr, num_bytes, parent,
2793                             root_objectid, ref_generation,
2794                             owner_objectid, pin, pin == 0);
2795
2796         finish_current_insert(trans, root->fs_info->extent_root, 0);
2797         pending_ret = del_pending_extents(trans, root->fs_info->extent_root, 0);
2798         return ret ? ret : pending_ret;
2799 }
2800
2801 int btrfs_free_extent(struct btrfs_trans_handle *trans,
2802                       struct btrfs_root *root,
2803                       u64 bytenr, u64 num_bytes, u64 parent,
2804                       u64 root_objectid, u64 ref_generation,
2805                       u64 owner_objectid, int pin)
2806 {
2807         int ret;
2808
2809         ret = __btrfs_free_extent(trans, root, bytenr, num_bytes, parent,
2810                                   root_objectid, ref_generation,
2811                                   owner_objectid, pin);
2812         return ret;
2813 }
2814
2815 static u64 stripe_align(struct btrfs_root *root, u64 val)
2816 {
2817         u64 mask = ((u64)root->stripesize - 1);
2818         u64 ret = (val + mask) & ~mask;
2819         return ret;
2820 }
2821
2822 /*
2823  * walks the btree of allocated extents and find a hole of a given size.
2824  * The key ins is changed to record the hole:
2825  * ins->objectid == block start
2826  * ins->flags = BTRFS_EXTENT_ITEM_KEY
2827  * ins->offset == number of blocks
2828  * Any available blocks before search_start are skipped.
2829  */
2830 static noinline int find_free_extent(struct btrfs_trans_handle *trans,
2831                                      struct btrfs_root *orig_root,
2832                                      u64 num_bytes, u64 empty_size,
2833                                      u64 search_start, u64 search_end,
2834                                      u64 hint_byte, struct btrfs_key *ins,
2835                                      u64 exclude_start, u64 exclude_nr,
2836                                      int data)
2837 {
2838         int ret = 0;
2839         struct btrfs_root *root = orig_root->fs_info->extent_root;
2840         u64 total_needed = num_bytes;
2841         u64 *last_ptr = NULL;
2842         u64 last_wanted = 0;
2843         struct btrfs_block_group_cache *block_group = NULL;
2844         int chunk_alloc_done = 0;
2845         int empty_cluster = 2 * 1024 * 1024;
2846         int allowed_chunk_alloc = 0;
2847         struct list_head *head = NULL, *cur = NULL;
2848         int loop = 0;
2849         int extra_loop = 0;
2850         struct btrfs_space_info *space_info;
2851
2852         WARN_ON(num_bytes < root->sectorsize);
2853         btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
2854         ins->objectid = 0;
2855         ins->offset = 0;
2856
2857         if (orig_root->ref_cows || empty_size)
2858                 allowed_chunk_alloc = 1;
2859
2860         if (data & BTRFS_BLOCK_GROUP_METADATA) {
2861                 last_ptr = &root->fs_info->last_alloc;
2862                 empty_cluster = 64 * 1024;
2863         }
2864
2865         if ((data & BTRFS_BLOCK_GROUP_DATA) && btrfs_test_opt(root, SSD))
2866                 last_ptr = &root->fs_info->last_data_alloc;
2867
2868         if (last_ptr) {
2869                 if (*last_ptr) {
2870                         hint_byte = *last_ptr;
2871                         last_wanted = *last_ptr;
2872                 } else
2873                         empty_size += empty_cluster;
2874         } else {
2875                 empty_cluster = 0;
2876         }
2877         search_start = max(search_start, first_logical_byte(root, 0));
2878         search_start = max(search_start, hint_byte);
2879
2880         if (last_wanted && search_start != last_wanted) {
2881                 last_wanted = 0;
2882                 empty_size += empty_cluster;
2883         }
2884
2885         total_needed += empty_size;
2886         block_group = btrfs_lookup_block_group(root->fs_info, search_start);
2887         if (!block_group)
2888                 block_group = btrfs_lookup_first_block_group(root->fs_info,
2889                                                              search_start);
2890         space_info = __find_space_info(root->fs_info, data);
2891
2892         down_read(&space_info->groups_sem);
2893         while (1) {
2894                 struct btrfs_free_space *free_space;
2895                 /*
2896                  * the only way this happens if our hint points to a block
2897                  * group thats not of the proper type, while looping this
2898                  * should never happen
2899                  */
2900                 if (empty_size)
2901                         extra_loop = 1;
2902
2903                 if (!block_group)
2904                         goto new_group_no_lock;
2905
2906                 if (unlikely(!block_group->cached)) {
2907                         mutex_lock(&block_group->cache_mutex);
2908                         ret = cache_block_group(root, block_group);
2909                         mutex_unlock(&block_group->cache_mutex);
2910                         if (ret)
2911                                 break;
2912                 }
2913
2914                 mutex_lock(&block_group->alloc_mutex);
2915                 if (unlikely(!block_group_bits(block_group, data)))
2916                         goto new_group;
2917
2918                 if (unlikely(block_group->ro))
2919                         goto new_group;
2920
2921                 free_space = btrfs_find_free_space(block_group, search_start,
2922                                                    total_needed);
2923                 if (free_space) {
2924                         u64 start = block_group->key.objectid;
2925                         u64 end = block_group->key.objectid +
2926                                 block_group->key.offset;
2927
2928                         search_start = stripe_align(root, free_space->offset);
2929
2930                         /* move on to the next group */
2931                         if (search_start + num_bytes >= search_end)
2932                                 goto new_group;
2933
2934                         /* move on to the next group */
2935                         if (search_start + num_bytes > end)
2936                                 goto new_group;
2937
2938                         if (last_wanted && search_start != last_wanted) {
2939                                 total_needed += empty_cluster;
2940                                 empty_size += empty_cluster;
2941                                 last_wanted = 0;
2942                                 /*
2943                                  * if search_start is still in this block group
2944                                  * then we just re-search this block group
2945                                  */
2946                                 if (search_start >= start &&
2947                                     search_start < end) {
2948                                         mutex_unlock(&block_group->alloc_mutex);
2949                                         continue;
2950                                 }
2951
2952                                 /* else we go to the next block group */
2953                                 goto new_group;
2954                         }
2955
2956                         if (exclude_nr > 0 &&
2957                             (search_start + num_bytes > exclude_start &&
2958                              search_start < exclude_start + exclude_nr)) {
2959                                 search_start = exclude_start + exclude_nr;
2960                                 /*
2961                                  * if search_start is still in this block group
2962                                  * then we just re-search this block group
2963                                  */
2964                                 if (search_start >= start &&
2965                                     search_start < end) {
2966                                         mutex_unlock(&block_group->alloc_mutex);
2967                                         last_wanted = 0;
2968                                         continue;
2969                                 }
2970
2971                                 /* else we go to the next block group */
2972                                 goto new_group;
2973                         }
2974
2975                         ins->objectid = search_start;
2976                         ins->offset = num_bytes;
2977
2978                         btrfs_remove_free_space_lock(block_group, search_start,
2979                                                      num_bytes);
2980                         /* we are all good, lets return */
2981                         mutex_unlock(&block_group->alloc_mutex);
2982                         break;
2983                 }
2984 new_group:
2985                 mutex_unlock(&block_group->alloc_mutex);
2986                 put_block_group(block_group);
2987                 block_group = NULL;
2988 new_group_no_lock:
2989                 /* don't try to compare new allocations against the
2990                  * last allocation any more
2991                  */
2992                 last_wanted = 0;
2993
2994                 /*
2995                  * Here's how this works.
2996                  * loop == 0: we were searching a block group via a hint
2997                  *              and didn't find anything, so we start at
2998                  *              the head of the block groups and keep searching
2999                  * loop == 1: we're searching through all of the block groups
3000                  *              if we hit the head again we have searched
3001                  *              all of the block groups for this space and we
3002                  *              need to try and allocate, if we cant error out.
3003                  * loop == 2: we allocated more space and are looping through
3004                  *              all of the block groups again.
3005                  */
3006                 if (loop == 0) {
3007                         head = &space_info->block_groups;
3008                         cur = head->next;
3009                         loop++;
3010                 } else if (loop == 1 && cur == head) {
3011                         int keep_going;
3012
3013                         /* at this point we give up on the empty_size
3014                          * allocations and just try to allocate the min
3015                          * space.
3016                          *
3017                          * The extra_loop field was set if an empty_size
3018                          * allocation was attempted above, and if this
3019                          * is try we need to try the loop again without
3020                          * the additional empty_size.
3021                          */
3022                         total_needed -= empty_size;
3023                         empty_size = 0;
3024                         keep_going = extra_loop;
3025                         loop++;
3026
3027                         if (allowed_chunk_alloc && !chunk_alloc_done) {
3028                                 up_read(&space_info->groups_sem);
3029                                 ret = do_chunk_alloc(trans, root, num_bytes +
3030                                                      2 * 1024 * 1024, data, 1);
3031                                 down_read(&space_info->groups_sem);
3032                                 if (ret < 0)
3033                                         goto loop_check;
3034                                 head = &space_info->block_groups;
3035                                 /*
3036                                  * we've allocated a new chunk, keep
3037                                  * trying
3038                                  */
3039                                 keep_going = 1;
3040                                 chunk_alloc_done = 1;
3041                         } else if (!allowed_chunk_alloc) {
3042                                 space_info->force_alloc = 1;
3043                         }
3044 loop_check:
3045                         if (keep_going) {
3046                                 cur = head->next;
3047                                 extra_loop = 0;
3048                         } else {
3049                                 break;
3050                         }
3051                 } else if (cur == head) {
3052                         break;
3053                 }
3054
3055                 block_group = list_entry(cur, struct btrfs_block_group_cache,
3056                                          list);
3057                 atomic_inc(&block_group->count);
3058
3059                 search_start = block_group->key.objectid;
3060                 cur = cur->next;
3061         }
3062
3063         /* we found what we needed */
3064         if (ins->objectid) {
3065                 if (!(data & BTRFS_BLOCK_GROUP_DATA))
3066                         trans->block_group = block_group->key.objectid;
3067
3068                 if (last_ptr)
3069                         *last_ptr = ins->objectid + ins->offset;
3070                 ret = 0;
3071         } else if (!ret) {
3072                 printk(KERN_ERR "btrfs searching for %llu bytes, "
3073                        "num_bytes %llu, loop %d, allowed_alloc %d\n",
3074                        (unsigned long long)total_needed,
3075                        (unsigned long long)num_bytes,
3076                        loop, allowed_chunk_alloc);
3077                 ret = -ENOSPC;
3078         }
3079         if (block_group)
3080                 put_block_group(block_group);
3081
3082         up_read(&space_info->groups_sem);
3083         return ret;
3084 }
3085
3086 static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
3087 {
3088         struct btrfs_block_group_cache *cache;
3089
3090         printk(KERN_INFO "space_info has %llu free, is %sfull\n",
3091                (unsigned long long)(info->total_bytes - info->bytes_used -
3092                                     info->bytes_pinned - info->bytes_reserved),
3093                (info->full) ? "" : "not ");
3094
3095         down_read(&info->groups_sem);
3096         list_for_each_entry(cache, &info->block_groups, list) {
3097                 spin_lock(&cache->lock);
3098                 printk(KERN_INFO "block group %llu has %llu bytes, %llu used "
3099                        "%llu pinned %llu reserved\n",
3100                        (unsigned long long)cache->key.objectid,
3101                        (unsigned long long)cache->key.offset,
3102                        (unsigned long long)btrfs_block_group_used(&cache->item),
3103                        (unsigned long long)cache->pinned,
3104                        (unsigned long long)cache->reserved);
3105                 btrfs_dump_free_space(cache, bytes);
3106                 spin_unlock(&cache->lock);
3107         }
3108         up_read(&info->groups_sem);
3109 }
3110
3111 static int __btrfs_reserve_extent(struct btrfs_trans_handle *trans,
3112                                   struct btrfs_root *root,
3113                                   u64 num_bytes, u64 min_alloc_size,
3114                                   u64 empty_size, u64 hint_byte,
3115                                   u64 search_end, struct btrfs_key *ins,
3116                                   u64 data)
3117 {
3118         int ret;
3119         u64 search_start = 0;
3120         u64 alloc_profile;
3121         struct btrfs_fs_info *info = root->fs_info;
3122
3123         if (data) {
3124                 alloc_profile = info->avail_data_alloc_bits &
3125                         info->data_alloc_profile;
3126                 data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
3127         } else if (root == root->fs_info->chunk_root) {
3128                 alloc_profile = info->avail_system_alloc_bits &
3129                         info->system_alloc_profile;
3130                 data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
3131         } else {
3132                 alloc_profile = info->avail_metadata_alloc_bits &
3133                         info->metadata_alloc_profile;
3134                 data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
3135         }
3136 again:
3137         data = btrfs_reduce_alloc_profile(root, data);
3138         /*
3139          * the only place that sets empty_size is btrfs_realloc_node, which
3140          * is not called recursively on allocations
3141          */
3142         if (empty_size || root->ref_cows) {
3143                 if (!(data & BTRFS_BLOCK_GROUP_METADATA)) {
3144                         ret = do_chunk_alloc(trans, root->fs_info->extent_root,
3145                                      2 * 1024 * 1024,
3146                                      BTRFS_BLOCK_GROUP_METADATA |
3147                                      (info->metadata_alloc_profile &
3148                                       info->avail_metadata_alloc_bits), 0);
3149                 }
3150                 ret = do_chunk_alloc(trans, root->fs_info->extent_root,
3151                                      num_bytes + 2 * 1024 * 1024, data, 0);
3152         }
3153
3154         WARN_ON(num_bytes < root->sectorsize);
3155         ret = find_free_extent(trans, root, num_bytes, empty_size,
3156                                search_start, search_end, hint_byte, ins,
3157                                trans->alloc_exclude_start,
3158                                trans->alloc_exclude_nr, data);
3159
3160         if (ret == -ENOSPC && num_bytes > min_alloc_size) {
3161                 num_bytes = num_bytes >> 1;
3162                 num_bytes = num_bytes & ~(root->sectorsize - 1);
3163                 num_bytes = max(num_bytes, min_alloc_size);
3164                 do_chunk_alloc(trans, root->fs_info->extent_root,
3165                                num_bytes, data, 1);
3166                 goto again;
3167         }
3168         if (ret) {
3169                 struct btrfs_space_info *sinfo;
3170
3171                 sinfo = __find_space_info(root->fs_info, data);
3172                 printk(KERN_ERR "btrfs allocation failed flags %llu, "
3173                        "wanted %llu\n", (unsigned long long)data,
3174                        (unsigned long long)num_bytes);
3175                 dump_space_info(sinfo, num_bytes);
3176                 BUG();
3177         }
3178
3179         return ret;
3180 }
3181
3182 int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len)
3183 {
3184         struct btrfs_block_group_cache *cache;
3185         int ret = 0;
3186
3187         cache = btrfs_lookup_block_group(root->fs_info, start);
3188         if (!cache) {
3189                 printk(KERN_ERR "Unable to find block group for %llu\n",
3190                        (unsigned long long)start);
3191                 return -ENOSPC;
3192         }
3193
3194         ret = btrfs_discard_extent(root, start, len);
3195
3196         btrfs_add_free_space(cache, start, len);
3197         put_block_group(cache);
3198         update_reserved_extents(root, start, len, 0);
3199
3200         return ret;
3201 }
3202
3203 int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
3204                                   struct btrfs_root *root,
3205                                   u64 num_bytes, u64 min_alloc_size,
3206                                   u64 empty_size, u64 hint_byte,
3207                                   u64 search_end, struct btrfs_key *ins,
3208                                   u64 data)
3209 {
3210         int ret;
3211         ret = __btrfs_reserve_extent(trans, root, num_bytes, min_alloc_size,
3212                                      empty_size, hint_byte, search_end, ins,
3213                                      data);
3214         update_reserved_extents(root, ins->objectid, ins->offset, 1);
3215         return ret;
3216 }
3217
3218 static int __btrfs_alloc_reserved_extent(struct btrfs_trans_handle *trans,
3219                                          struct btrfs_root *root, u64 parent,
3220                                          u64 root_objectid, u64 ref_generation,
3221                                          u64 owner, struct btrfs_key *ins)
3222 {
3223         int ret;
3224         int pending_ret;
3225         u64 super_used;
3226         u64 root_used;
3227         u64 num_bytes = ins->offset;
3228         u32 sizes[2];
3229         struct btrfs_fs_info *info = root->fs_info;
3230         struct btrfs_root *extent_root = info->extent_root;
3231         struct btrfs_extent_item *extent_item;
3232         struct btrfs_extent_ref *ref;
3233         struct btrfs_path *path;
3234         struct btrfs_key keys[2];
3235
3236         if (parent == 0)
3237                 parent = ins->objectid;
3238
3239         /* block accounting for super block */
3240         spin_lock(&info->delalloc_lock);
3241         super_used = btrfs_super_bytes_used(&info->super_copy);
3242         btrfs_set_super_bytes_used(&info->super_copy, super_used + num_bytes);
3243
3244         /* block accounting for root item */
3245         root_used = btrfs_root_used(&root->root_item);
3246         btrfs_set_root_used(&root->root_item, root_used + num_bytes);
3247         spin_unlock(&info->delalloc_lock);
3248
3249         if (root == extent_root) {
3250                 struct pending_extent_op *extent_op;
3251
3252                 extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
3253                 BUG_ON(!extent_op);
3254
3255                 extent_op->type = PENDING_EXTENT_INSERT;
3256                 extent_op->bytenr = ins->objectid;
3257                 extent_op->num_bytes = ins->offset;
3258                 extent_op->parent = parent;
3259                 extent_op->orig_parent = 0;
3260                 extent_op->generation = ref_generation;
3261                 extent_op->orig_generation = 0;
3262                 extent_op->level = (int)owner;
3263                 INIT_LIST_HEAD(&extent_op->list);
3264                 extent_op->del = 0;
3265
3266                 mutex_lock(&root->fs_info->extent_ins_mutex);
3267                 set_extent_bits(&root->fs_info->extent_ins, ins->objectid,
3268                                 ins->objectid + ins->offset - 1,
3269                                 EXTENT_WRITEBACK, GFP_NOFS);
3270                 set_state_private(&root->fs_info->extent_ins,
3271                                   ins->objectid, (unsigned long)extent_op);
3272                 mutex_unlock(&root->fs_info->extent_ins_mutex);
3273                 goto update_block;
3274         }
3275
3276         memcpy(&keys[0], ins, sizeof(*ins));
3277         keys[1].objectid = ins->objectid;
3278         keys[1].type = BTRFS_EXTENT_REF_KEY;
3279         keys[1].offset = parent;
3280         sizes[0] = sizeof(*extent_item);
3281         sizes[1] = sizeof(*ref);
3282
3283         path = btrfs_alloc_path();
3284         BUG_ON(!path);
3285
3286         ret = btrfs_insert_empty_items(trans, extent_root, path, keys,
3287                                        sizes, 2);
3288         BUG_ON(ret);
3289
3290         extent_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
3291                                      struct btrfs_extent_item);
3292         btrfs_set_extent_refs(path->nodes[0], extent_item, 1);
3293         ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
3294                              struct btrfs_extent_ref);
3295
3296         btrfs_set_ref_root(path->nodes[0], ref, root_objectid);
3297         btrfs_set_ref_generation(path->nodes[0], ref, ref_generation);
3298         btrfs_set_ref_objectid(path->nodes[0], ref, owner);
3299         btrfs_set_ref_num_refs(path->nodes[0], ref, 1);
3300
3301         btrfs_mark_buffer_dirty(path->nodes[0]);
3302
3303         trans->alloc_exclude_start = 0;
3304         trans->alloc_exclude_nr = 0;
3305         btrfs_free_path(path);
3306         finish_current_insert(trans, extent_root, 0);
3307         pending_ret = del_pending_extents(trans, extent_root, 0);
3308
3309         if (ret)
3310                 goto out;
3311         if (pending_ret) {
3312                 ret = pending_ret;
3313                 goto out;
3314         }
3315
3316 update_block:
3317         ret = update_block_group(trans, root, ins->objectid,
3318                                  ins->offset, 1, 0);
3319         if (ret) {
3320                 printk(KERN_ERR "btrfs update block group failed for %llu "
3321                        "%llu\n", (unsigned long long)ins->objectid,
3322                        (unsigned long long)ins->offset);
3323                 BUG();
3324         }
3325 out:
3326         return ret;
3327 }
3328
3329 int btrfs_alloc_reserved_extent(struct btrfs_trans_handle *trans,
3330                                 struct btrfs_root *root, u64 parent,
3331                                 u64 root_objectid, u64 ref_generation,
3332                                 u64 owner, struct btrfs_key *ins)
3333 {
3334         int ret;
3335
3336         if (root_objectid == BTRFS_TREE_LOG_OBJECTID)
3337                 return 0;
3338         ret = __btrfs_alloc_reserved_extent(trans, root, parent, root_objectid,
3339                                             ref_generation, owner, ins);
3340         update_reserved_extents(root, ins->objectid, ins->offset, 0);
3341         return ret;
3342 }
3343
3344 /*
3345  * this is used by the tree logging recovery code.  It records that
3346  * an extent has been allocated and makes sure to clear the free
3347  * space cache bits as well
3348  */
3349 int btrfs_alloc_logged_extent(struct btrfs_trans_handle *trans,
3350                                 struct btrfs_root *root, u64 parent,
3351                                 u64 root_objectid, u64 ref_generation,
3352                                 u64 owner, struct btrfs_key *ins)
3353 {
3354         int ret;
3355         struct btrfs_block_group_cache *block_group;
3356
3357         block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid);
3358         mutex_lock(&block_group->cache_mutex);
3359         cache_block_group(root, block_group);
3360         mutex_unlock(&block_group->cache_mutex);
3361
3362         ret = btrfs_remove_free_space(block_group, ins->objectid,
3363                                       ins->offset);
3364         BUG_ON(ret);
3365         put_block_group(block_group);
3366         ret = __btrfs_alloc_reserved_extent(trans, root, parent, root_objectid,
3367                                             ref_generation, owner, ins);
3368         return ret;
3369 }
3370
3371 /*
3372  * finds a free extent and does all the dirty work required for allocation
3373  * returns the key for the extent through ins, and a tree buffer for
3374  * the first block of the extent through buf.
3375  *
3376  * returns 0 if everything worked, non-zero otherwise.
3377  */
3378 int btrfs_alloc_extent(struct btrfs_trans_handle *trans,
3379                        struct btrfs_root *root,
3380                        u64 num_bytes, u64 parent, u64 min_alloc_size,
3381                        u64 root_objectid, u64 ref_generation,
3382                        u64 owner_objectid, u64 empty_size, u64 hint_byte,
3383                        u64 search_end, struct btrfs_key *ins, u64 data)
3384 {
3385         int ret;
3386
3387         ret = __btrfs_reserve_extent(trans, root, num_bytes,
3388                                      min_alloc_size, empty_size, hint_byte,
3389                                      search_end, ins, data);
3390         BUG_ON(ret);
3391         if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
3392                 ret = __btrfs_alloc_reserved_extent(trans, root, parent,
3393                                         root_objectid, ref_generation,
3394                                         owner_objectid, ins);
3395                 BUG_ON(ret);
3396
3397         } else {
3398                 update_reserved_extents(root, ins->objectid, ins->offset, 1);
3399         }
3400         return ret;
3401 }
3402
3403 struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
3404                                             struct btrfs_root *root,
3405                                             u64 bytenr, u32 blocksize)
3406 {
3407         struct extent_buffer *buf;
3408
3409         buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
3410         if (!buf)
3411                 return ERR_PTR(-ENOMEM);
3412         btrfs_set_header_generation(buf, trans->transid);
3413         btrfs_tree_lock(buf);
3414         clean_tree_block(trans, root, buf);
3415
3416         btrfs_set_lock_blocking(buf);
3417         btrfs_set_buffer_uptodate(buf);
3418
3419         if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
3420                 set_extent_dirty(&root->dirty_log_pages, buf->start,
3421                          buf->start + buf->len - 1, GFP_NOFS);
3422         } else {
3423                 set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
3424                          buf->start + buf->len - 1, GFP_NOFS);
3425         }
3426         trans->blocks_used++;
3427         /* this returns a buffer locked for blocking */
3428         return buf;
3429 }
3430
3431 /*
3432  * helper function to allocate a block for a given tree
3433  * returns the tree buffer or NULL.
3434  */
3435 struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
3436                                              struct btrfs_root *root,
3437                                              u32 blocksize, u64 parent,
3438                                              u64 root_objectid,
3439                                              u64 ref_generation,
3440                                              int level,
3441                                              u64 hint,
3442                                              u64 empty_size)
3443 {
3444         struct btrfs_key ins;
3445         int ret;
3446         struct extent_buffer *buf;
3447
3448         ret = btrfs_alloc_extent(trans, root, blocksize, parent, blocksize,
3449                                  root_objectid, ref_generation, level,
3450                                  empty_size, hint, (u64)-1, &ins, 0);
3451         if (ret) {
3452                 BUG_ON(ret > 0);
3453                 return ERR_PTR(ret);
3454         }
3455
3456         buf = btrfs_init_new_buffer(trans, root, ins.objectid, blocksize);
3457         return buf;
3458 }
3459
3460 int btrfs_drop_leaf_ref(struct btrfs_trans_handle *trans,
3461                         struct btrfs_root *root, struct extent_buffer *leaf)
3462 {
3463         u64 leaf_owner;
3464         u64 leaf_generation;
3465         struct refsort *sorted;
3466         struct btrfs_key key;
3467         struct btrfs_file_extent_item *fi;
3468         int i;
3469         int nritems;
3470         int ret;
3471         int refi = 0;
3472         int slot;
3473
3474         BUG_ON(!btrfs_is_leaf(leaf));
3475         nritems = btrfs_header_nritems(leaf);
3476         leaf_owner = btrfs_header_owner(leaf);
3477         leaf_generation = btrfs_header_generation(leaf);
3478
3479         sorted = kmalloc(sizeof(*sorted) * nritems, GFP_NOFS);
3480         /* we do this loop twice.  The first time we build a list
3481          * of the extents we have a reference on, then we sort the list
3482          * by bytenr.  The second time around we actually do the
3483          * extent freeing.
3484          */
3485         for (i = 0; i < nritems; i++) {
3486                 u64 disk_bytenr;
3487                 cond_resched();
3488
3489                 btrfs_item_key_to_cpu(leaf, &key, i);
3490
3491                 /* only extents have references, skip everything else */
3492                 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
3493                         continue;
3494
3495                 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
3496
3497                 /* inline extents live in the btree, they don't have refs */
3498                 if (btrfs_file_extent_type(leaf, fi) ==
3499                     BTRFS_FILE_EXTENT_INLINE)
3500                         continue;
3501
3502                 disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
3503
3504                 /* holes don't have refs */
3505                 if (disk_bytenr == 0)
3506                         continue;
3507
3508                 sorted[refi].bytenr = disk_bytenr;
3509                 sorted[refi].slot = i;
3510                 refi++;
3511         }
3512
3513         if (refi == 0)
3514                 goto out;
3515
3516         sort(sorted, refi, sizeof(struct refsort), refsort_cmp, NULL);
3517
3518         for (i = 0; i < refi; i++) {
3519                 u64 disk_bytenr;
3520
3521                 disk_bytenr = sorted[i].bytenr;
3522                 slot = sorted[i].slot;
3523
3524                 cond_resched();
3525
3526                 btrfs_item_key_to_cpu(leaf, &key, slot);
3527                 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
3528                         continue;
3529
3530                 fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
3531
3532                 ret = __btrfs_free_extent(trans, root, disk_bytenr,
3533                                 btrfs_file_extent_disk_num_bytes(leaf, fi),
3534                                 leaf->start, leaf_owner, leaf_generation,
3535                                 key.objectid, 0);
3536                 BUG_ON(ret);
3537
3538                 atomic_inc(&root->fs_info->throttle_gen);
3539                 wake_up(&root->fs_info->transaction_throttle);
3540                 cond_resched();
3541         }
3542 out:
3543         kfree(sorted);
3544         return 0;
3545 }
3546
3547 static noinline int cache_drop_leaf_ref(struct btrfs_trans_handle *trans,
3548                                         struct btrfs_root *root,
3549                                         struct btrfs_leaf_ref *ref)
3550 {
3551         int i;
3552         int ret;
3553         struct btrfs_extent_info *info;
3554         struct refsort *sorted;
3555
3556         if (ref->nritems == 0)
3557                 return 0;
3558
3559         sorted = kmalloc(sizeof(*sorted) * ref->nritems, GFP_NOFS);
3560         for (i = 0; i < ref->nritems; i++) {
3561                 sorted[i].bytenr = ref->extents[i].bytenr;
3562                 sorted[i].slot = i;
3563         }
3564         sort(sorted, ref->nritems, sizeof(struct refsort), refsort_cmp, NULL);
3565
3566         /*
3567          * the items in the ref were sorted when the ref was inserted
3568          * into the ref cache, so this is already in order
3569          */
3570         for (i = 0; i < ref->nritems; i++) {
3571                 info = ref->extents + sorted[i].slot;
3572                 ret = __btrfs_free_extent(trans, root, info->bytenr,
3573                                           info->num_bytes, ref->bytenr,
3574                                           ref->owner, ref->generation,
3575                                           info->objectid, 0);
3576
3577                 atomic_inc(&root->fs_info->throttle_gen);
3578                 wake_up(&root->fs_info->transaction_throttle);
3579                 cond_resched();
3580
3581                 BUG_ON(ret);
3582                 info++;
3583         }
3584
3585         kfree(sorted);
3586         return 0;
3587 }
3588
3589 static int drop_snap_lookup_refcount(struct btrfs_root *root, u64 start,
3590                                      u64 len, u32 *refs)
3591 {
3592         int ret;
3593
3594         ret = btrfs_lookup_extent_ref(NULL, root, start, len, refs);
3595         BUG_ON(ret);
3596
3597 #if 0 /* some debugging code in case we see problems here */
3598         /* if the refs count is one, it won't get increased again.  But
3599          * if the ref count is > 1, someone may be decreasing it at
3600          * the same time we are.
3601          */
3602         if (*refs != 1) {
3603                 struct extent_buffer *eb = NULL;
3604                 eb = btrfs_find_create_tree_block(root, start, len);
3605                 if (eb)
3606                         btrfs_tree_lock(eb);
3607
3608                 mutex_lock(&root->fs_info->alloc_mutex);
3609                 ret = lookup_extent_ref(NULL, root, start, len, refs);
3610                 BUG_ON(ret);
3611                 mutex_unlock(&root->fs_info->alloc_mutex);
3612
3613                 if (eb) {
3614                         btrfs_tree_unlock(eb);
3615                         free_extent_buffer(eb);
3616                 }
3617                 if (*refs == 1) {
3618                         printk(KERN_ERR "btrfs block %llu went down to one "
3619                                "during drop_snap\n", (unsigned long long)start);
3620                 }
3621
3622         }
3623 #endif
3624
3625         cond_resched();
3626         return ret;
3627 }
3628
3629 /*
3630  * this is used while deleting old snapshots, and it drops the refs
3631  * on a whole subtree starting from a level 1 node.
3632  *
3633  * The idea is to sort all the leaf pointers, and then drop the
3634  * ref on all the leaves in order.  Most of the time the leaves
3635  * will have ref cache entries, so no leaf IOs will be required to
3636  * find the extents they have references on.
3637  *
3638  * For each leaf, any references it has are also dropped in order
3639  *
3640  * This ends up dropping the references in something close to optimal
3641  * order for reading and modifying the extent allocation tree.
3642  */
3643 static noinline int drop_level_one_refs(struct btrfs_trans_handle *trans,
3644                                         struct btrfs_root *root,
3645                                         struct btrfs_path *path)
3646 {
3647         u64 bytenr;
3648         u64 root_owner;
3649         u64 root_gen;
3650         struct extent_buffer *eb = path->nodes[1];
3651         struct extent_buffer *leaf;
3652         struct btrfs_leaf_ref *ref;
3653         struct refsort *sorted = NULL;
3654         int nritems = btrfs_header_nritems(eb);
3655         int ret;
3656         int i;
3657         int refi = 0;
3658         int slot = path->slots[1];
3659         u32 blocksize = btrfs_level_size(root, 0);
3660         u32 refs;
3661
3662         if (nritems == 0)
3663                 goto out;
3664
3665         root_owner = btrfs_header_owner(eb);
3666         root_gen = btrfs_header_generation(eb);
3667         sorted = kmalloc(sizeof(*sorted) * nritems, GFP_NOFS);
3668
3669         /*
3670          * step one, sort all the leaf pointers so we don't scribble
3671          * randomly into the extent allocation tree
3672          */
3673         for (i = slot; i < nritems; i++) {
3674                 sorted[refi].bytenr = btrfs_node_blockptr(eb, i);
3675                 sorted[refi].slot = i;
3676                 refi++;
3677         }
3678
3679         /*
3680          * nritems won't be zero, but if we're picking up drop_snapshot
3681          * after a crash, slot might be > 0, so double check things
3682          * just in case.
3683          */
3684         if (refi == 0)
3685                 goto out;
3686
3687         sort(sorted, refi, sizeof(struct refsort), refsort_cmp, NULL);
3688
3689         /*
3690          * the first loop frees everything the leaves point to
3691          */
3692         for (i = 0; i < refi; i++) {
3693                 u64 ptr_gen;
3694
3695                 bytenr = sorted[i].bytenr;
3696
3697                 /*
3698                  * check the reference count on this leaf.  If it is > 1
3699                  * we just decrement it below and don't update any
3700                  * of the refs the leaf points to.
3701                  */
3702                 ret = drop_snap_lookup_refcount(root, bytenr, blocksize, &refs);
3703                 BUG_ON(ret);
3704                 if (refs != 1)
3705                         continue;
3706
3707                 ptr_gen = btrfs_node_ptr_generation(eb, sorted[i].slot);
3708
3709                 /*
3710                  * the leaf only had one reference, which means the
3711                  * only thing pointing to this leaf is the snapshot
3712                  * we're deleting.  It isn't possible for the reference
3713                  * count to increase again later
3714                  *
3715                  * The reference cache is checked for the leaf,
3716                  * and if found we'll be able to drop any refs held by
3717                  * the leaf without needing to read it in.
3718                  */
3719                 ref = btrfs_lookup_leaf_ref(root, bytenr);
3720                 if (ref && ref->generation != ptr_gen) {
3721                         btrfs_free_leaf_ref(root, ref);
3722                         ref = NULL;
3723                 }
3724                 if (ref) {
3725                         ret = cache_drop_leaf_ref(trans, root, ref);
3726                         BUG_ON(ret);
3727                         btrfs_remove_leaf_ref(root, ref);
3728                         btrfs_free_leaf_ref(root, ref);
3729                 } else {
3730                         /*
3731                          * the leaf wasn't in the reference cache, so
3732                          * we have to read it.
3733                          */
3734                         leaf = read_tree_block(root, bytenr, blocksize,
3735                                                ptr_gen);
3736                         ret = btrfs_drop_leaf_ref(trans, root, leaf);
3737                         BUG_ON(ret);
3738                         free_extent_buffer(leaf);
3739                 }
3740                 atomic_inc(&root->fs_info->throttle_gen);
3741                 wake_up(&root->fs_info->transaction_throttle);
3742                 cond_resched();
3743         }
3744
3745         /*
3746          * run through the loop again to free the refs on the leaves.
3747          * This is faster than doing it in the loop above because
3748          * the leaves are likely to be clustered together.  We end up
3749          * working in nice chunks on the extent allocation tree.
3750          */
3751         for (i = 0; i < refi; i++) {
3752                 bytenr = sorted[i].bytenr;
3753                 ret = __btrfs_free_extent(trans, root, bytenr,
3754                                         blocksize, eb->start,
3755                                         root_owner, root_gen, 0, 1);
3756                 BUG_ON(ret);
3757
3758                 atomic_inc(&root->fs_info->throttle_gen);
3759                 wake_up(&root->fs_info->transaction_throttle);
3760                 cond_resched();
3761         }
3762 out:
3763         kfree(sorted);
3764
3765         /*
3766          * update the path to show we've processed the entire level 1
3767          * node.  This will get saved into the root's drop_snapshot_progress
3768          * field so these drops are not repeated again if this transaction
3769          * commits.
3770          */
3771         path->slots[1] = nritems;
3772         return 0;
3773 }
3774
3775 /*
3776  * helper function for drop_snapshot, this walks down the tree dropping ref
3777  * counts as it goes.
3778  */
3779 static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
3780                                    struct btrfs_root *root,
3781                                    struct btrfs_path *path, int *level)
3782 {
3783         u64 root_owner;
3784         u64 root_gen;
3785         u64 bytenr;
3786         u64 ptr_gen;
3787         struct extent_buffer *next;
3788         struct extent_buffer *cur;
3789         struct extent_buffer *parent;
3790         u32 blocksize;
3791         int ret;
3792         u32 refs;
3793
3794         WARN_ON(*level < 0);
3795         WARN_ON(*level >= BTRFS_MAX_LEVEL);
3796         ret = drop_snap_lookup_refcount(root, path->nodes[*level]->start,
3797                                 path->nodes[*level]->len, &refs);
3798         BUG_ON(ret);
3799         if (refs > 1)
3800                 goto out;
3801
3802         /*
3803          * walk down to the last node level and free all the leaves
3804          */
3805         while (*level >= 0) {
3806                 WARN_ON(*level < 0);
3807                 WARN_ON(*level >= BTRFS_MAX_LEVEL);
3808                 cur = path->nodes[*level];
3809
3810                 if (btrfs_header_level(cur) != *level)
3811                         WARN_ON(1);
3812
3813                 if (path->slots[*level] >=
3814                     btrfs_header_nritems(cur))
3815                         break;
3816
3817                 /* the new code goes down to level 1 and does all the
3818                  * leaves pointed to that node in bulk.  So, this check
3819                  * for level 0 will always be false.
3820                  *
3821                  * But, the disk format allows the drop_snapshot_progress
3822                  * field in the root to leave things in a state where
3823                  * a leaf will need cleaning up here.  If someone crashes
3824                  * with the old code and then boots with the new code,
3825                  * we might find a leaf here.
3826                  */
3827                 if (*level == 0) {
3828                         ret = btrfs_drop_leaf_ref(trans, root, cur);
3829                         BUG_ON(ret);
3830                         break;
3831                 }
3832
3833                 /*
3834                  * once we get to level one, process the whole node
3835                  * at once, including everything below it.
3836                  */
3837                 if (*level == 1) {
3838                         ret = drop_level_one_refs(trans, root, path);
3839                         BUG_ON(ret);
3840                         break;
3841                 }
3842
3843                 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
3844                 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
3845                 blocksize = btrfs_level_size(root, *level - 1);
3846
3847                 ret = drop_snap_lookup_refcount(root, bytenr, blocksize, &refs);
3848                 BUG_ON(ret);
3849
3850                 /*
3851                  * if there is more than one reference, we don't need
3852                  * to read that node to drop any references it has.  We
3853                  * just drop the ref we hold on that node and move on to the
3854                  * next slot in this level.
3855                  */
3856                 if (refs != 1) {
3857                         parent = path->nodes[*level];
3858                         root_owner = btrfs_header_owner(parent);
3859                         root_gen = btrfs_header_generation(parent);
3860                         path->slots[*level]++;
3861
3862                         ret = __btrfs_free_extent(trans, root, bytenr,
3863                                                 blocksize, parent->start,
3864                                                 root_owner, root_gen,
3865                                                 *level - 1, 1);
3866                         BUG_ON(ret);
3867
3868                         atomic_inc(&root->fs_info->throttle_gen);
3869                         wake_up(&root->fs_info->transaction_throttle);
3870                         cond_resched();
3871
3872                         continue;
3873                 }
3874
3875                 /*
3876                  * we need to keep freeing things in the next level down.
3877                  * read the block and loop around to process it
3878                  */
3879                 next = read_tree_block(root, bytenr, blocksize, ptr_gen);
3880                 WARN_ON(*level <= 0);
3881                 if (path->nodes[*level-1])
3882                         free_extent_buffer(path->nodes[*level-1]);
3883                 path->nodes[*level-1] = next;
3884                 *level = btrfs_header_level(next);
3885                 path->slots[*level] = 0;
3886                 cond_resched();
3887         }
3888 out:
3889         WARN_ON(*level < 0);
3890         WARN_ON(*level >= BTRFS_MAX_LEVEL);
3891
3892         if (path->nodes[*level] == root->node) {
3893                 parent = path->nodes[*level];
3894                 bytenr = path->nodes[*level]->start;
3895         } else {
3896                 parent = path->nodes[*level + 1];
3897                 bytenr = btrfs_node_blockptr(parent, path->slots[*level + 1]);
3898         }
3899
3900         blocksize = btrfs_level_size(root, *level);
3901         root_owner = btrfs_header_owner(parent);
3902         root_gen = btrfs_header_generation(parent);
3903
3904         /*
3905          * cleanup and free the reference on the last node
3906          * we processed
3907          */
3908         ret = __btrfs_free_extent(trans, root, bytenr, blocksize,
3909                                   parent->start, root_owner, root_gen,
3910                                   *level, 1);
3911         free_extent_buffer(path->nodes[*level]);
3912         path->nodes[*level] = NULL;
3913
3914         *level += 1;
3915         BUG_ON(ret);
3916
3917         cond_resched();
3918         return 0;
3919 }
3920
3921 /*
3922  * helper function for drop_subtree, this function is similar to
3923  * walk_down_tree. The main difference is that it checks reference
3924  * counts while tree blocks are locked.
3925  */
3926 static noinline int walk_down_subtree(struct btrfs_trans_handle *trans,
3927                                       struct btrfs_root *root,
3928                                       struct btrfs_path *path, int *level)
3929 {
3930         struct extent_buffer *next;
3931         struct extent_buffer *cur;
3932         struct extent_buffer *parent;
3933         u64 bytenr;
3934         u64 ptr_gen;
3935         u32 blocksize;
3936         u32 refs;
3937         int ret;
3938
3939         cur = path->nodes[*level];
3940         ret = btrfs_lookup_extent_ref(trans, root, cur->start, cur->len,
3941                                       &refs);
3942         BUG_ON(ret);
3943         if (refs > 1)
3944                 goto out;
3945
3946         while (*level >= 0) {
3947                 cur = path->nodes[*level];
3948                 if (*level == 0) {
3949                         ret = btrfs_drop_leaf_ref(trans, root, cur);
3950                         BUG_ON(ret);
3951                         clean_tree_block(trans, root, cur);
3952                         break;
3953                 }
3954                 if (path->slots[*level] >= btrfs_header_nritems(cur)) {
3955                         clean_tree_block(trans, root, cur);
3956                         break;
3957                 }
3958
3959                 bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
3960                 blocksize = btrfs_level_size(root, *level - 1);
3961                 ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
3962
3963                 next = read_tree_block(root, bytenr, blocksize, ptr_gen);
3964                 btrfs_tree_lock(next);
3965                 btrfs_set_lock_blocking(next);
3966
3967                 ret = btrfs_lookup_extent_ref(trans, root, bytenr, blocksize,
3968                                               &refs);
3969                 BUG_ON(ret);
3970                 if (refs > 1) {
3971                         parent = path->nodes[*level];
3972                         ret = btrfs_free_extent(trans, root, bytenr,
3973                                         blocksize, parent->start,
3974                                         btrfs_header_owner(parent),
3975                                         btrfs_header_generation(parent),
3976                                         *level - 1, 1);
3977                         BUG_ON(ret);
3978                         path->slots[*level]++;
3979                         btrfs_tree_unlock(next);
3980                         free_extent_buffer(next);
3981                         continue;
3982                 }
3983
3984                 *level = btrfs_header_level(next);
3985                 path->nodes[*level] = next;
3986                 path->slots[*level] = 0;
3987                 path->locks[*level] = 1;
3988                 cond_resched();
3989         }
3990 out:
3991         parent = path->nodes[*level + 1];
3992         bytenr = path->nodes[*level]->start;
3993         blocksize = path->nodes[*level]->len;
3994
3995         ret = btrfs_free_extent(trans, root, bytenr, blocksize,
3996                         parent->start, btrfs_header_owner(parent),
3997                         btrfs_header_generation(parent), *level, 1);
3998         BUG_ON(ret);
3999
4000         if (path->locks[*level]) {
4001                 btrfs_tree_unlock(path->nodes[*level]);
4002                 path->locks[*level] = 0;
4003         }
4004         free_extent_buffer(path->nodes[*level]);
4005         path->nodes[*level] = NULL;
4006         *level += 1;
4007         cond_resched();
4008         return 0;
4009 }
4010
4011 /*
4012  * helper for dropping snapshots.  This walks back up the tree in the path
4013  * to find the first node higher up where we haven't yet gone through
4014  * all the slots
4015  */
4016 static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
4017                                  struct btrfs_root *root,
4018                                  struct btrfs_path *path,
4019                                  int *level, int max_level)
4020 {
4021         u64 root_owner;
4022         u64 root_gen;
4023         struct btrfs_root_item *root_item = &root->root_item;
4024         int i;
4025         int slot;
4026         int ret;
4027
4028         for (i = *level; i < max_level && path->nodes[i]; i++) {
4029                 slot = path->slots[i];
4030                 if (slot < btrfs_header_nritems(path->nodes[i]) - 1) {
4031                         struct extent_buffer *node;
4032                         struct btrfs_disk_key disk_key;
4033
4034                         /*
4035                          * there is more work to do in this level.
4036                          * Update the drop_progress marker to reflect
4037                          * the work we've done so far, and then bump
4038                          * the slot number
4039                          */
4040                         node = path->nodes[i];
4041                         path->slots[i]++;
4042                         *level = i;
4043                         WARN_ON(*level == 0);
4044                         btrfs_node_key(node, &disk_key, path->slots[i]);
4045                         memcpy(&root_item->drop_progress,
4046                                &disk_key, sizeof(disk_key));
4047                         root_item->drop_level = i;
4048                         return 0;
4049                 } else {
4050                         struct extent_buffer *parent;
4051
4052                         /*
4053                          * this whole node is done, free our reference
4054                          * on it and go up one level
4055                          */
4056                         if (path->nodes[*level] == root->node)
4057                                 parent = path->nodes[*level];
4058                         else
4059                                 parent = path->nodes[*level + 1];
4060
4061                         root_owner = btrfs_header_owner(parent);
4062                         root_gen = btrfs_header_generation(parent);
4063
4064                         clean_tree_block(trans, root, path->nodes[*level]);
4065                         ret = btrfs_free_extent(trans, root,
4066                                                 path->nodes[*level]->start,
4067                                                 path->nodes[*level]->len,
4068                                                 parent->start, root_owner,
4069                                                 root_gen, *level, 1);
4070                         BUG_ON(ret);
4071                         if (path->locks[*level]) {
4072                                 btrfs_tree_unlock(path->nodes[*level]);
4073                                 path->locks[*level] = 0;
4074                         }
4075                         free_extent_buffer(path->nodes[*level]);
4076                         path->nodes[*level] = NULL;
4077                         *level = i + 1;
4078                 }
4079         }
4080         return 1;
4081 }
4082
4083 /*
4084  * drop the reference count on the tree rooted at 'snap'.  This traverses
4085  * the tree freeing any blocks that have a ref count of zero after being
4086  * decremented.
4087  */
4088 int btrfs_drop_snapshot(struct btrfs_trans_handle *trans, struct btrfs_root
4089                         *root)
4090 {
4091         int ret = 0;
4092         int wret;
4093         int level;
4094         struct btrfs_path *path;
4095         int i;
4096         int orig_level;
4097         struct btrfs_root_item *root_item = &root->root_item;
4098
4099         WARN_ON(!mutex_is_locked(&root->fs_info->drop_mutex));
4100         path = btrfs_alloc_path();
4101         BUG_ON(!path);
4102
4103         level = btrfs_header_level(root->node);
4104         orig_level = level;
4105         if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
4106                 path->nodes[level] = root->node;
4107                 extent_buffer_get(root->node);
4108                 path->slots[level] = 0;
4109         } else {
4110                 struct btrfs_key key;
4111                 struct btrfs_disk_key found_key;
4112                 struct extent_buffer *node;
4113
4114                 btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
4115                 level = root_item->drop_level;
4116                 path->lowest_level = level;
4117                 wret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
4118                 if (wret < 0) {
4119                         ret = wret;
4120                         goto out;
4121                 }
4122                 node = path->nodes[level];
4123                 btrfs_node_key(node, &found_key, path->slots[level]);
4124                 WARN_ON(memcmp(&found_key, &root_item->drop_progress,
4125                                sizeof(found_key)));
4126                 /*
4127                  * unlock our path, this is safe because only this
4128                  * function is allowed to delete this snapshot
4129                  */
4130                 for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
4131                         if (path->nodes[i] && path->locks[i]) {
4132                                 path->locks[i] = 0;
4133                                 btrfs_tree_unlock(path->nodes[i]);
4134                         }
4135                 }
4136         }
4137         while (1) {
4138                 wret = walk_down_tree(trans, root, path, &level);
4139                 if (wret > 0)
4140                         break;
4141                 if (wret < 0)
4142                         ret = wret;
4143
4144                 wret = walk_up_tree(trans, root, path, &level,
4145                                     BTRFS_MAX_LEVEL);
4146                 if (wret > 0)
4147                         break;
4148                 if (wret < 0)
4149                         ret = wret;
4150                 if (trans->transaction->in_commit) {
4151                         ret = -EAGAIN;
4152                         break;
4153                 }
4154                 atomic_inc(&root->fs_info->throttle_gen);
4155                 wake_up(&root->fs_info->transaction_throttle);
4156         }
4157         for (i = 0; i <= orig_level; i++) {
4158                 if (path->nodes[i]) {
4159                         free_extent_buffer(path->nodes[i]);
4160                         path->nodes[i] = NULL;
4161                 }
4162         }
4163 out:
4164         btrfs_free_path(path);
4165         return ret;
4166 }
4167
4168 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
4169                         struct btrfs_root *root,
4170                         struct extent_buffer *node,
4171                         struct extent_buffer *parent)
4172 {
4173         struct btrfs_path *path;
4174         int level;
4175         int parent_level;
4176         int ret = 0;
4177         int wret;
4178
4179         path = btrfs_alloc_path();
4180         BUG_ON(!path);
4181
4182         BUG_ON(!btrfs_tree_locked(parent));
4183         parent_level = btrfs_header_level(parent);
4184         extent_buffer_get(parent);
4185         path->nodes[parent_level] = parent;
4186         path->slots[parent_level] = btrfs_header_nritems(parent);
4187
4188         BUG_ON(!btrfs_tree_locked(node));
4189         level = btrfs_header_level(node);
4190         extent_buffer_get(node);
4191         path->nodes[level] = node;
4192         path->slots[level] = 0;
4193
4194         while (1) {
4195                 wret = walk_down_subtree(trans, root, path, &level);
4196                 if (wret < 0)
4197                         ret = wret;
4198                 if (wret != 0)
4199                         break;
4200
4201                 wret = walk_up_tree(trans, root, path, &level, parent_level);
4202                 if (wret < 0)
4203                         ret = wret;
4204                 if (wret != 0)
4205                         break;
4206         }
4207
4208         btrfs_free_path(path);
4209         return ret;
4210 }
4211
4212 static unsigned long calc_ra(unsigned long start, unsigned long last,
4213                              unsigned long nr)
4214 {
4215         return min(last, start + nr - 1);
4216 }
4217
4218 static noinline int relocate_inode_pages(struct inode *inode, u64 start,
4219                                          u64 len)
4220 {
4221         u64 page_start;
4222         u64 page_end;
4223         unsigned long first_index;
4224         unsigned long last_index;
4225         unsigned long i;
4226         struct page *page;
4227         struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
4228         struct file_ra_state *ra;
4229         struct btrfs_ordered_extent *ordered;
4230         unsigned int total_read = 0;
4231         unsigned int total_dirty = 0;
4232         int ret = 0;
4233
4234         ra = kzalloc(sizeof(*ra), GFP_NOFS);
4235
4236         mutex_lock(&inode->i_mutex);
4237         first_index = start >> PAGE_CACHE_SHIFT;
4238         last_index = (start + len - 1) >> PAGE_CACHE_SHIFT;
4239
4240         /* make sure the dirty trick played by the caller work */
4241         ret = invalidate_inode_pages2_range(inode->i_mapping,
4242                                             first_index, last_index);
4243         if (ret)
4244                 goto out_unlock;
4245
4246         file_ra_state_init(ra, inode->i_mapping);
4247
4248         for (i = first_index ; i <= last_index; i++) {
4249                 if (total_read % ra->ra_pages == 0) {
4250                         btrfs_force_ra(inode->i_mapping, ra, NULL, i,
4251                                        calc_ra(i, last_index, ra->ra_pages));
4252                 }
4253                 total_read++;
4254 again:
4255                 if (((u64)i << PAGE_CACHE_SHIFT) > i_size_read(inode))
4256                         BUG_ON(1);
4257                 page = grab_cache_page(inode->i_mapping, i);
4258                 if (!page) {
4259                         ret = -ENOMEM;
4260                         goto out_unlock;
4261                 }
4262                 if (!PageUptodate(page)) {
4263                         btrfs_readpage(NULL, page);
4264                         lock_page(page);
4265                         if (!PageUptodate(page)) {
4266                                 unlock_page(page);
4267                                 page_cache_release(page);
4268                                 ret = -EIO;
4269                                 goto out_unlock;
4270                         }
4271                 }
4272                 wait_on_page_writeback(page);
4273
4274                 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
4275                 page_end = page_start + PAGE_CACHE_SIZE - 1;
4276                 lock_extent(io_tree, page_start, page_end, GFP_NOFS);
4277
4278                 ordered = btrfs_lookup_ordered_extent(inode, page_start);
4279                 if (ordered) {
4280                         unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
4281                         unlock_page(page);
4282                         page_cache_release(page);
4283                         btrfs_start_ordered_extent(inode, ordered, 1);
4284                         btrfs_put_ordered_extent(ordered);
4285                         goto again;
4286                 }
4287                 set_page_extent_mapped(page);
4288
4289                 if (i == first_index)
4290                         set_extent_bits(io_tree, page_start, page_end,
4291                                         EXTENT_BOUNDARY, GFP_NOFS);
4292                 btrfs_set_extent_delalloc(inode, page_start, page_end);
4293
4294                 set_page_dirty(page);
4295                 total_dirty++;
4296
4297                 unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
4298                 unlock_page(page);
4299                 page_cache_release(page);
4300         }
4301
4302 out_unlock:
4303         kfree(ra);
4304         mutex_unlock(&inode->i_mutex);
4305         balance_dirty_pages_ratelimited_nr(inode->i_mapping, total_dirty);
4306         return ret;
4307 }
4308
4309 static noinline int relocate_data_extent(struct inode *reloc_inode,
4310                                          struct btrfs_key *extent_key,
4311                                          u64 offset)
4312 {
4313         struct btrfs_root *root = BTRFS_I(reloc_inode)->root;
4314         struct extent_map_tree *em_tree = &BTRFS_I(reloc_inode)->extent_tree;
4315         struct extent_map *em;
4316         u64 start = extent_key->objectid - offset;
4317         u64 end = start + extent_key->offset - 1;
4318
4319         em = alloc_extent_map(GFP_NOFS);
4320         BUG_ON(!em || IS_ERR(em));
4321
4322         em->start = start;
4323         em->len = extent_key->offset;
4324         em->block_len = extent_key->offset;
4325         em->block_start = extent_key->objectid;
4326         em->bdev = root->fs_info->fs_devices->latest_bdev;
4327         set_bit(EXTENT_FLAG_PINNED, &em->flags);
4328
4329         /* setup extent map to cheat btrfs_readpage */
4330         lock_extent(&BTRFS_I(reloc_inode)->io_tree, start, end, GFP_NOFS);
4331         while (1) {
4332                 int ret;
4333                 spin_lock(&em_tree->lock);
4334                 ret = add_extent_mapping(em_tree, em);
4335                 spin_unlock(&em_tree->lock);
4336                 if (ret != -EEXIST) {
4337                         free_extent_map(em);
4338                         break;
4339                 }
4340                 btrfs_drop_extent_cache(reloc_inode, start, end, 0);
4341         }
4342         unlock_extent(&BTRFS_I(reloc_inode)->io_tree, start, end, GFP_NOFS);
4343
4344         return relocate_inode_pages(reloc_inode, start, extent_key->offset);
4345 }
4346
4347 struct btrfs_ref_path {
4348         u64 extent_start;
4349         u64 nodes[BTRFS_MAX_LEVEL];
4350         u64 root_objectid;
4351         u64 root_generation;
4352         u64 owner_objectid;
4353         u32 num_refs;
4354         int lowest_level;
4355         int current_level;
4356         int shared_level;
4357
4358         struct btrfs_key node_keys[BTRFS_MAX_LEVEL];
4359         u64 new_nodes[BTRFS_MAX_LEVEL];
4360 };
4361
4362 struct disk_extent {
4363         u64 ram_bytes;
4364         u64 disk_bytenr;
4365         u64 disk_num_bytes;
4366         u64 offset;
4367         u64 num_bytes;
4368         u8 compression;
4369         u8 encryption;
4370         u16 other_encoding;
4371 };
4372
4373 static int is_cowonly_root(u64 root_objectid)
4374 {
4375         if (root_objectid == BTRFS_ROOT_TREE_OBJECTID ||
4376             root_objectid == BTRFS_EXTENT_TREE_OBJECTID ||
4377             root_objectid == BTRFS_CHUNK_TREE_OBJECTID ||
4378             root_objectid == BTRFS_DEV_TREE_OBJECTID ||
4379             root_objectid == BTRFS_TREE_LOG_OBJECTID ||
4380             root_objectid == BTRFS_CSUM_TREE_OBJECTID)
4381                 return 1;
4382         return 0;
4383 }
4384
4385 static noinline int __next_ref_path(struct btrfs_trans_handle *trans,
4386                                     struct btrfs_root *extent_root,
4387                                     struct btrfs_ref_path *ref_path,
4388                                     int first_time)
4389 {
4390         struct extent_buffer *leaf;
4391         struct btrfs_path *path;
4392         struct btrfs_extent_ref *ref;
4393         struct btrfs_key key;
4394         struct btrfs_key found_key;
4395         u64 bytenr;
4396         u32 nritems;
4397         int level;
4398         int ret = 1;
4399
4400         path = btrfs_alloc_path();
4401         if (!path)
4402                 return -ENOMEM;
4403
4404         if (first_time) {
4405                 ref_path->lowest_level = -1;
4406                 ref_path->current_level = -1;
4407                 ref_path->shared_level = -1;
4408                 goto walk_up;
4409         }
4410 walk_down:
4411         level = ref_path->current_level - 1;
4412         while (level >= -1) {
4413                 u64 parent;
4414                 if (level < ref_path->lowest_level)
4415                         break;
4416
4417                 if (level >= 0)
4418                         bytenr = ref_path->nodes[level];
4419                 else
4420                         bytenr = ref_path->extent_start;
4421                 BUG_ON(bytenr == 0);
4422
4423                 parent = ref_path->nodes[level + 1];
4424                 ref_path->nodes[level + 1] = 0;
4425                 ref_path->current_level = level;
4426                 BUG_ON(parent == 0);
4427
4428                 key.objectid = bytenr;
4429                 key.offset = parent + 1;
4430                 key.type = BTRFS_EXTENT_REF_KEY;
4431
4432                 ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0);
4433                 if (ret < 0)
4434                         goto out;
4435                 BUG_ON(ret == 0);
4436
4437                 leaf = path->nodes[0];
4438                 nritems = btrfs_header_nritems(leaf);
4439                 if (path->slots[0] >= nritems) {
4440                         ret = btrfs_next_leaf(extent_root, path);
4441                         if (ret < 0)
4442                                 goto out;
4443                         if (ret > 0)
4444                                 goto next;
4445                         leaf = path->nodes[0];
4446                 }
4447
4448                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4449                 if (found_key.objectid == bytenr &&
4450                     found_key.type == BTRFS_EXTENT_REF_KEY) {
4451                         if (level < ref_path->shared_level)
4452                                 ref_path->shared_level = level;
4453                         goto found;
4454                 }
4455 next:
4456                 level--;
4457                 btrfs_release_path(extent_root, path);
4458                 cond_resched();
4459         }
4460         /* reached lowest level */
4461         ret = 1;
4462         goto out;
4463 walk_up:
4464         level = ref_path->current_level;
4465         while (level < BTRFS_MAX_LEVEL - 1) {
4466                 u64 ref_objectid;
4467
4468                 if (level >= 0)
4469                         bytenr = ref_path->nodes[level];
4470                 else
4471                         bytenr = ref_path->extent_start;
4472
4473                 BUG_ON(bytenr == 0);
4474
4475                 key.objectid = bytenr;
4476                 key.offset = 0;
4477                 key.type = BTRFS_EXTENT_REF_KEY;
4478
4479                 ret = btrfs_search_slot(trans, extent_root, &key, path, 0, 0);
4480                 if (ret < 0)
4481                         goto out;
4482
4483                 leaf = path->nodes[0];
4484                 nritems = btrfs_header_nritems(leaf);
4485                 if (path->slots[0] >= nritems) {
4486                         ret = btrfs_next_leaf(extent_root, path);
4487                         if (ret < 0)
4488                                 goto out;
4489                         if (ret > 0) {
4490                                 /* the extent was freed by someone */
4491                                 if (ref_path->lowest_level == level)
4492                                         goto out;
4493                                 btrfs_release_path(extent_root, path);
4494                                 goto walk_down;
4495                         }
4496                         leaf = path->nodes[0];
4497                 }
4498
4499                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4500                 if (found_key.objectid != bytenr ||
4501                                 found_key.type != BTRFS_EXTENT_REF_KEY) {
4502                         /* the extent was freed by someone */
4503                         if (ref_path->lowest_level == level) {
4504                                 ret = 1;
4505                                 goto out;
4506                         }
4507                         btrfs_release_path(extent_root, path);
4508                         goto walk_down;
4509                 }
4510 found:
4511                 ref = btrfs_item_ptr(leaf, path->slots[0],
4512                                 struct btrfs_extent_ref);
4513                 ref_objectid = btrfs_ref_objectid(leaf, ref);
4514                 if (ref_objectid < BTRFS_FIRST_FREE_OBJECTID) {
4515                         if (first_time) {
4516                                 level = (int)ref_objectid;
4517                                 BUG_ON(level >= BTRFS_MAX_LEVEL);
4518                                 ref_path->lowest_level = level;
4519                                 ref_path->current_level = level;
4520                                 ref_path->nodes[level] = bytenr;
4521                         } else {
4522                                 WARN_ON(ref_objectid != level);
4523                         }
4524                 } else {
4525                         WARN_ON(level != -1);
4526                 }
4527                 first_time = 0;
4528
4529                 if (ref_path->lowest_level == level) {
4530                         ref_path->owner_objectid = ref_objectid;
4531                         ref_path->num_refs = btrfs_ref_num_refs(leaf, ref);
4532                 }
4533
4534                 /*
4535                  * the block is tree root or the block isn't in reference
4536                  * counted tree.
4537                  */
4538                 if (found_key.objectid == found_key.offset ||
4539                     is_cowonly_root(btrfs_ref_root(leaf, ref))) {
4540                         ref_path->root_objectid = btrfs_ref_root(leaf, ref);
4541                         ref_path->root_generation =
4542                                 btrfs_ref_generation(leaf, ref);
4543                         if (level < 0) {
4544                                 /* special reference from the tree log */
4545                                 ref_path->nodes[0] = found_key.offset;
4546                                 ref_path->current_level = 0;
4547                         }
4548                         ret = 0;
4549                         goto out;
4550                 }
4551
4552                 level++;
4553                 BUG_ON(ref_path->nodes[level] != 0);
4554                 ref_path->nodes[level] = found_key.offset;
4555                 ref_path->current_level = level;
4556
4557                 /*
4558                  * the reference was created in the running transaction,
4559                  * no need to continue walking up.
4560                  */
4561                 if (btrfs_ref_generation(leaf, ref) == trans->transid) {
4562                         ref_path->root_objectid = btrfs_ref_root(leaf, ref);
4563                         ref_path->root_generation =
4564                                 btrfs_ref_generation(leaf, ref);
4565                         ret = 0;
4566                         goto out;
4567                 }
4568
4569                 btrfs_release_path(extent_root, path);
4570                 cond_resched();
4571         }
4572         /* reached max tree level, but no tree root found. */
4573         BUG();
4574 out:
4575         btrfs_free_path(path);
4576         return ret;
4577 }
4578
4579 static int btrfs_first_ref_path(struct btrfs_trans_handle *trans,
4580                                 struct btrfs_root *extent_root,
4581                                 struct btrfs_ref_path *ref_path,
4582                                 u64 extent_start)
4583 {
4584         memset(ref_path, 0, sizeof(*ref_path));
4585         ref_path->extent_start = extent_start;
4586
4587         return __next_ref_path(trans, extent_root, ref_path, 1);
4588 }
4589
4590 static int btrfs_next_ref_path(struct btrfs_trans_handle *trans,
4591                                struct btrfs_root *extent_root,
4592                                struct btrfs_ref_path *ref_path)
4593 {
4594         return __next_ref_path(trans, extent_root, ref_path, 0);
4595 }
4596
4597 static noinline int get_new_locations(struct inode *reloc_inode,
4598                                       struct btrfs_key *extent_key,
4599                                       u64 offset, int no_fragment,
4600                                       struct disk_extent **extents,
4601                                       int *nr_extents)
4602 {
4603         struct btrfs_root *root = BTRFS_I(reloc_inode)->root;
4604         struct btrfs_path *path;
4605         struct btrfs_file_extent_item *fi;
4606         struct extent_buffer *leaf;
4607         struct disk_extent *exts = *extents;
4608         struct btrfs_key found_key;
4609         u64 cur_pos;
4610         u64 last_byte;
4611         u32 nritems;
4612         int nr = 0;
4613         int max = *nr_extents;
4614         int ret;
4615
4616         WARN_ON(!no_fragment && *extents);
4617         if (!exts) {
4618                 max = 1;
4619                 exts = kmalloc(sizeof(*exts) * max, GFP_NOFS);
4620                 if (!exts)
4621                         return -ENOMEM;
4622         }
4623
4624         path = btrfs_alloc_path();
4625         BUG_ON(!path);
4626
4627         cur_pos = extent_key->objectid - offset;
4628         last_byte = extent_key->objectid + extent_key->offset;
4629         ret = btrfs_lookup_file_extent(NULL, root, path, reloc_inode->i_ino,
4630                                        cur_pos, 0);
4631         if (ret < 0)
4632                 goto out;
4633         if (ret > 0) {
4634                 ret = -ENOENT;
4635                 goto out;
4636         }
4637
4638         while (1) {
4639                 leaf = path->nodes[0];
4640                 nritems = btrfs_header_nritems(leaf);
4641                 if (path->slots[0] >= nritems) {
4642                         ret = btrfs_next_leaf(root, path);
4643                         if (ret < 0)
4644                                 goto out;
4645                         if (ret > 0)
4646                                 break;
4647                         leaf = path->nodes[0];
4648                 }
4649
4650                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
4651                 if (found_key.offset != cur_pos ||
4652                     found_key.type != BTRFS_EXTENT_DATA_KEY ||
4653                     found_key.objectid != reloc_inode->i_ino)
4654                         break;
4655
4656                 fi = btrfs_item_ptr(leaf, path->slots[0],
4657                                     struct btrfs_file_extent_item);
4658                 if (btrfs_file_extent_type(leaf, fi) !=
4659                     BTRFS_FILE_EXTENT_REG ||
4660                     btrfs_file_extent_disk_bytenr(leaf, fi) == 0)
4661                         break;
4662
4663                 if (nr == max) {
4664                         struct disk_extent *old = exts;
4665                         max *= 2;
4666                         exts = kzalloc(sizeof(*exts) * max, GFP_NOFS);
4667                         memcpy(exts, old, sizeof(*exts) * nr);
4668                         if (old != *extents)
4669                                 kfree(old);
4670                 }
4671
4672                 exts[nr].disk_bytenr =
4673                         btrfs_file_extent_disk_bytenr(leaf, fi);
4674                 exts[nr].disk_num_bytes =
4675                         btrfs_file_extent_disk_num_bytes(leaf, fi);
4676                 exts[nr].offset = btrfs_file_extent_offset(leaf, fi);
4677                 exts[nr].num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
4678                 exts[nr].ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi);
4679                 exts[nr].compression = btrfs_file_extent_compression(leaf, fi);
4680                 exts[nr].encryption = btrfs_file_extent_encryption(leaf, fi);
4681                 exts[nr].other_encoding = btrfs_file_extent_other_encoding(leaf,
4682                                                                            fi);
4683                 BUG_ON(exts[nr].offset > 0);
4684                 BUG_ON(exts[nr].compression || exts[nr].encryption);
4685                 BUG_ON(exts[nr].num_bytes != exts[nr].disk_num_bytes);
4686
4687                 cur_pos += exts[nr].num_bytes;
4688                 nr++;
4689
4690                 if (cur_pos + offset >= last_byte)
4691                         break;
4692
4693                 if (no_fragment) {
4694                         ret = 1;
4695                         goto out;
4696                 }
4697                 path->slots[0]++;
4698         }
4699
4700         BUG_ON(cur_pos + offset > last_byte);
4701         if (cur_pos + offset < last_byte) {
4702                 ret = -ENOENT;
4703                 goto out;
4704         }
4705         ret = 0;
4706 out:
4707         btrfs_free_path(path);
4708         if (ret) {
4709                 if (exts != *extents)
4710                         kfree(exts);
4711         } else {
4712                 *extents = exts;
4713                 *nr_extents = nr;
4714         }
4715         return ret;
4716 }
4717
4718 static noinline int replace_one_extent(struct btrfs_trans_handle *trans,
4719                                         struct btrfs_root *root,
4720                                         struct btrfs_path *path,
4721                                         struct btrfs_key *extent_key,
4722                                         struct btrfs_key *leaf_key,
4723                                         struct btrfs_ref_path *ref_path,
4724                                         struct disk_extent *new_extents,
4725                                         int nr_extents)
4726 {
4727         struct extent_buffer *leaf;
4728         struct btrfs_file_extent_item *fi;
4729         struct inode *inode = NULL;
4730         struct btrfs_key key;
4731         u64 lock_start = 0;
4732         u64 lock_end = 0;
4733         u64 num_bytes;
4734         u64 ext_offset;
4735         u64 search_end = (u64)-1;
4736         u32 nritems;
4737         int nr_scaned = 0;
4738         int extent_locked = 0;
4739         int extent_type;
4740         int ret;
4741
4742         memcpy(&key, leaf_key, sizeof(key));
4743         if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) {
4744                 if (key.objectid < ref_path->owner_objectid ||
4745                     (key.objectid == ref_path->owner_objectid &&
4746                      key.type < BTRFS_EXTENT_DATA_KEY)) {
4747                         key.objectid = ref_path->owner_objectid;
4748                         key.type = BTRFS_EXTENT_DATA_KEY;
4749                         key.offset = 0;
4750                 }
4751         }
4752
4753         while (1) {
4754                 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
4755                 if (ret < 0)
4756                         goto out;
4757
4758                 leaf = path->nodes[0];
4759                 nritems = btrfs_header_nritems(leaf);
4760 next:
4761                 if (extent_locked && ret > 0) {
4762                         /*
4763                          * the file extent item was modified by someone
4764                          * before the extent got locked.
4765                          */
4766                         unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4767                                       lock_end, GFP_NOFS);
4768                         extent_locked = 0;
4769                 }
4770
4771                 if (path->slots[0] >= nritems) {
4772                         if (++nr_scaned > 2)
4773                                 break;
4774
4775                         BUG_ON(extent_locked);
4776                         ret = btrfs_next_leaf(root, path);
4777                         if (ret < 0)
4778                                 goto out;
4779                         if (ret > 0)
4780                                 break;
4781                         leaf = path->nodes[0];
4782                         nritems = btrfs_header_nritems(leaf);
4783                 }
4784
4785                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
4786
4787                 if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS) {
4788                         if ((key.objectid > ref_path->owner_objectid) ||
4789                             (key.objectid == ref_path->owner_objectid &&
4790                              key.type > BTRFS_EXTENT_DATA_KEY) ||
4791                             key.offset >= search_end)
4792                                 break;
4793                 }
4794
4795                 if (inode && key.objectid != inode->i_ino) {
4796                         BUG_ON(extent_locked);
4797                         btrfs_release_path(root, path);
4798                         mutex_unlock(&inode->i_mutex);
4799                         iput(inode);
4800                         inode = NULL;
4801                         continue;
4802                 }
4803
4804                 if (key.type != BTRFS_EXTENT_DATA_KEY) {
4805                         path->slots[0]++;
4806                         ret = 1;
4807                         goto next;
4808                 }
4809                 fi = btrfs_item_ptr(leaf, path->slots[0],
4810                                     struct btrfs_file_extent_item);
4811                 extent_type = btrfs_file_extent_type(leaf, fi);
4812                 if ((extent_type != BTRFS_FILE_EXTENT_REG &&
4813                      extent_type != BTRFS_FILE_EXTENT_PREALLOC) ||
4814                     (btrfs_file_extent_disk_bytenr(leaf, fi) !=
4815                      extent_key->objectid)) {
4816                         path->slots[0]++;
4817                         ret = 1;
4818                         goto next;
4819                 }
4820
4821                 num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
4822                 ext_offset = btrfs_file_extent_offset(leaf, fi);
4823
4824                 if (search_end == (u64)-1) {
4825                         search_end = key.offset - ext_offset +
4826                                 btrfs_file_extent_ram_bytes(leaf, fi);
4827                 }
4828
4829                 if (!extent_locked) {
4830                         lock_start = key.offset;
4831                         lock_end = lock_start + num_bytes - 1;
4832                 } else {
4833                         if (lock_start > key.offset ||
4834                             lock_end + 1 < key.offset + num_bytes) {
4835                                 unlock_extent(&BTRFS_I(inode)->io_tree,
4836                                               lock_start, lock_end, GFP_NOFS);
4837                                 extent_locked = 0;
4838                         }
4839                 }
4840
4841                 if (!inode) {
4842                         btrfs_release_path(root, path);
4843
4844                         inode = btrfs_iget_locked(root->fs_info->sb,
4845                                                   key.objectid, root);
4846                         if (inode->i_state & I_NEW) {
4847                                 BTRFS_I(inode)->root = root;
4848                                 BTRFS_I(inode)->location.objectid =
4849                                         key.objectid;
4850                                 BTRFS_I(inode)->location.type =
4851                                         BTRFS_INODE_ITEM_KEY;
4852                                 BTRFS_I(inode)->location.offset = 0;
4853                                 btrfs_read_locked_inode(inode);
4854                                 unlock_new_inode(inode);
4855                         }
4856                         /*
4857                          * some code call btrfs_commit_transaction while
4858                          * holding the i_mutex, so we can't use mutex_lock
4859                          * here.
4860                          */
4861                         if (is_bad_inode(inode) ||
4862                             !mutex_trylock(&inode->i_mutex)) {
4863                                 iput(inode);
4864                                 inode = NULL;
4865                                 key.offset = (u64)-1;
4866                                 goto skip;
4867                         }
4868                 }
4869
4870                 if (!extent_locked) {
4871                         struct btrfs_ordered_extent *ordered;
4872
4873                         btrfs_release_path(root, path);
4874
4875                         lock_extent(&BTRFS_I(inode)->io_tree, lock_start,
4876                                     lock_end, GFP_NOFS);
4877                         ordered = btrfs_lookup_first_ordered_extent(inode,
4878                                                                     lock_end);
4879                         if (ordered &&
4880                             ordered->file_offset <= lock_end &&
4881                             ordered->file_offset + ordered->len > lock_start) {
4882                                 unlock_extent(&BTRFS_I(inode)->io_tree,
4883                                               lock_start, lock_end, GFP_NOFS);
4884                                 btrfs_start_ordered_extent(inode, ordered, 1);
4885                                 btrfs_put_ordered_extent(ordered);
4886                                 key.offset += num_bytes;
4887                                 goto skip;
4888                         }
4889                         if (ordered)
4890                                 btrfs_put_ordered_extent(ordered);
4891
4892                         extent_locked = 1;
4893                         continue;
4894                 }
4895
4896                 if (nr_extents == 1) {
4897                         /* update extent pointer in place */
4898                         btrfs_set_file_extent_disk_bytenr(leaf, fi,
4899                                                 new_extents[0].disk_bytenr);
4900                         btrfs_set_file_extent_disk_num_bytes(leaf, fi,
4901                                                 new_extents[0].disk_num_bytes);
4902                         btrfs_mark_buffer_dirty(leaf);
4903
4904                         btrfs_drop_extent_cache(inode, key.offset,
4905                                                 key.offset + num_bytes - 1, 0);
4906
4907                         ret = btrfs_inc_extent_ref(trans, root,
4908                                                 new_extents[0].disk_bytenr,
4909                                                 new_extents[0].disk_num_bytes,
4910                                                 leaf->start,
4911                                                 root->root_key.objectid,
4912                                                 trans->transid,
4913                                                 key.objectid);
4914                         BUG_ON(ret);
4915
4916                         ret = btrfs_free_extent(trans, root,
4917                                                 extent_key->objectid,
4918                                                 extent_key->offset,
4919                                                 leaf->start,
4920                                                 btrfs_header_owner(leaf),
4921                                                 btrfs_header_generation(leaf),
4922                                                 key.objectid, 0);
4923                         BUG_ON(ret);
4924
4925                         btrfs_release_path(root, path);
4926                         key.offset += num_bytes;
4927                 } else {
4928                         BUG_ON(1);
4929 #if 0
4930                         u64 alloc_hint;
4931                         u64 extent_len;
4932                         int i;
4933                         /*
4934                          * drop old extent pointer at first, then insert the
4935                          * new pointers one bye one
4936                          */
4937                         btrfs_release_path(root, path);
4938                         ret = btrfs_drop_extents(trans, root, inode, key.offset,
4939                                                  key.offset + num_bytes,
4940                                                  key.offset, &alloc_hint);
4941                         BUG_ON(ret);
4942
4943                         for (i = 0; i < nr_extents; i++) {
4944                                 if (ext_offset >= new_extents[i].num_bytes) {
4945                                         ext_offset -= new_extents[i].num_bytes;
4946                                         continue;
4947                                 }
4948                                 extent_len = min(new_extents[i].num_bytes -
4949                                                  ext_offset, num_bytes);
4950
4951                                 ret = btrfs_insert_empty_item(trans, root,
4952                                                               path, &key,
4953                                                               sizeof(*fi));
4954                                 BUG_ON(ret);
4955
4956                                 leaf = path->nodes[0];
4957                                 fi = btrfs_item_ptr(leaf, path->slots[0],
4958                                                 struct btrfs_file_extent_item);
4959                                 btrfs_set_file_extent_generation(leaf, fi,
4960                                                         trans->transid);
4961                                 btrfs_set_file_extent_type(leaf, fi,
4962                                                         BTRFS_FILE_EXTENT_REG);
4963                                 btrfs_set_file_extent_disk_bytenr(leaf, fi,
4964                                                 new_extents[i].disk_bytenr);
4965                                 btrfs_set_file_extent_disk_num_bytes(leaf, fi,
4966                                                 new_extents[i].disk_num_bytes);
4967                                 btrfs_set_file_extent_ram_bytes(leaf, fi,
4968                                                 new_extents[i].ram_bytes);
4969
4970                                 btrfs_set_file_extent_compression(leaf, fi,
4971                                                 new_extents[i].compression);
4972                                 btrfs_set_file_extent_encryption(leaf, fi,
4973                                                 new_extents[i].encryption);
4974                                 btrfs_set_file_extent_other_encoding(leaf, fi,
4975                                                 new_extents[i].other_encoding);
4976
4977                                 btrfs_set_file_extent_num_bytes(leaf, fi,
4978                                                         extent_len);
4979                                 ext_offset += new_extents[i].offset;
4980                                 btrfs_set_file_extent_offset(leaf, fi,
4981                                                         ext_offset);
4982                                 btrfs_mark_buffer_dirty(leaf);
4983
4984                                 btrfs_drop_extent_cache(inode, key.offset,
4985                                                 key.offset + extent_len - 1, 0);
4986
4987                                 ret = btrfs_inc_extent_ref(trans, root,
4988                                                 new_extents[i].disk_bytenr,
4989                                                 new_extents[i].disk_num_bytes,
4990                                                 leaf->start,
4991                                                 root->root_key.objectid,
4992                                                 trans->transid, key.objectid);
4993                                 BUG_ON(ret);
4994                                 btrfs_release_path(root, path);
4995
4996                                 inode_add_bytes(inode, extent_len);
4997
4998                                 ext_offset = 0;
4999                                 num_bytes -= extent_len;
5000                                 key.offset += extent_len;
5001
5002                                 if (num_bytes == 0)
5003                                         break;
5004                         }
5005                         BUG_ON(i >= nr_extents);
5006 #endif
5007                 }
5008
5009                 if (extent_locked) {
5010                         unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
5011                                       lock_end, GFP_NOFS);
5012                         extent_locked = 0;
5013                 }
5014 skip:
5015                 if (ref_path->owner_objectid != BTRFS_MULTIPLE_OBJECTIDS &&
5016                     key.offset >= search_end)
5017                         break;
5018
5019                 cond_resched();
5020         }
5021         ret = 0;
5022 out:
5023         btrfs_release_path(root, path);
5024         if (inode) {
5025                 mutex_unlock(&inode->i_mutex);
5026                 if (extent_locked) {
5027                         unlock_extent(&BTRFS_I(inode)->io_tree, lock_start,
5028                                       lock_end, GFP_NOFS);
5029                 }
5030                 iput(inode);
5031         }
5032         return ret;
5033 }
5034
5035 int btrfs_reloc_tree_cache_ref(struct btrfs_trans_handle *trans,
5036                                struct btrfs_root *root,
5037                                struct extent_buffer *buf, u64 orig_start)
5038 {
5039         int level;
5040         int ret;
5041
5042         BUG_ON(btrfs_header_generation(buf) != trans->transid);
5043         BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
5044
5045         level = btrfs_header_level(buf);
5046         if (level == 0) {
5047                 struct btrfs_leaf_ref *ref;
5048                 struct btrfs_leaf_ref *orig_ref;
5049
5050                 orig_ref = btrfs_lookup_leaf_ref(root, orig_start);
5051                 if (!orig_ref)
5052                         return -ENOENT;
5053
5054                 ref = btrfs_alloc_leaf_ref(root, orig_ref->nritems);
5055                 if (!ref) {
5056                         btrfs_free_leaf_ref(root, orig_ref);
5057                         return -ENOMEM;
5058                 }
5059
5060                 ref->nritems = orig_ref->nritems;
5061                 memcpy(ref->extents, orig_ref->extents,
5062                         sizeof(ref->extents[0]) * ref->nritems);
5063
5064                 btrfs_free_leaf_ref(root, orig_ref);
5065
5066                 ref->root_gen = trans->transid;
5067                 ref->bytenr = buf->start;
5068                 ref->owner = btrfs_header_owner(buf);
5069                 ref->generation = btrfs_header_generation(buf);
5070
5071                 ret = btrfs_add_leaf_ref(root, ref, 0);
5072                 WARN_ON(ret);
5073                 btrfs_free_leaf_ref(root, ref);
5074         }
5075         return 0;
5076 }
5077
5078 static noinline int invalidate_extent_cache(struct btrfs_root *root,
5079                                         struct extent_buffer *leaf,
5080                                         struct btrfs_block_group_cache *group,
5081                                         struct btrfs_root *target_root)
5082 {
5083         struct btrfs_key key;
5084         struct inode *inode = NULL;
5085         struct btrfs_file_extent_item *fi;
5086         u64 num_bytes;
5087         u64 skip_objectid = 0;
5088         u32 nritems;
5089         u32 i;
5090
5091         nritems = btrfs_header_nritems(leaf);
5092         for (i = 0; i < nritems; i++) {
5093                 btrfs_item_key_to_cpu(leaf, &key, i);
5094                 if (key.objectid == skip_objectid ||
5095                     key.type != BTRFS_EXTENT_DATA_KEY)
5096                         continue;
5097                 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
5098                 if (btrfs_file_extent_type(leaf, fi) ==
5099                     BTRFS_FILE_EXTENT_INLINE)
5100                         continue;
5101                 if (btrfs_file_extent_disk_bytenr(leaf, fi) == 0)
5102                         continue;
5103                 if (!inode || inode->i_ino != key.objectid) {
5104                         iput(inode);
5105                         inode = btrfs_ilookup(target_root->fs_info->sb,
5106                                               key.objectid, target_root, 1);
5107                 }
5108                 if (!inode) {
5109                         skip_objectid = key.objectid;
5110                         continue;
5111                 }
5112                 num_bytes = btrfs_file_extent_num_bytes(leaf, fi);
5113
5114                 lock_extent(&BTRFS_I(inode)->io_tree, key.offset,
5115                             key.offset + num_bytes - 1, GFP_NOFS);
5116                 btrfs_drop_extent_cache(inode, key.offset,
5117                                         key.offset + num_bytes - 1, 1);
5118                 unlock_extent(&BTRFS_I(inode)->io_tree, key.offset,
5119                               key.offset + num_bytes - 1, GFP_NOFS);
5120                 cond_resched();
5121         }
5122         iput(inode);
5123         return 0;
5124 }
5125
5126 static noinline int replace_extents_in_leaf(struct btrfs_trans_handle *trans,
5127                                         struct btrfs_root *root,
5128                                         struct extent_buffer *leaf,
5129                                         struct btrfs_block_group_cache *group,
5130                                         struct inode *reloc_inode)
5131 {
5132         struct btrfs_key key;
5133         struct btrfs_key extent_key;
5134         struct btrfs_file_extent_item *fi;
5135         struct btrfs_leaf_ref *ref;
5136         struct disk_extent *new_extent;
5137         u64 bytenr;
5138         u64 num_bytes;
5139         u32 nritems;
5140         u32 i;
5141         int ext_index;
5142         int nr_extent;
5143         int ret;
5144
5145         new_extent = kmalloc(sizeof(*new_extent), GFP_NOFS);
5146         BUG_ON(!new_extent);
5147
5148         ref = btrfs_lookup_leaf_ref(root, leaf->start);
5149         BUG_ON(!ref);
5150
5151         ext_index = -1;
5152         nritems = btrfs_header_nritems(leaf);
5153         for (i = 0; i < nritems; i++) {
5154                 btrfs_item_key_to_cpu(leaf, &key, i);
5155                 if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
5156                         continue;
5157                 fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
5158                 if (btrfs_file_extent_type(leaf, fi) ==
5159                     BTRFS_FILE_EXTENT_INLINE)
5160                         continue;
5161                 bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
5162                 num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
5163                 if (bytenr == 0)
5164                         continue;
5165
5166                 ext_index++;
5167                 if (bytenr >= group->key.objectid + group->key.offset ||
5168                     bytenr + num_bytes <= group->key.objectid)
5169                         continue;
5170
5171                 extent_key.objectid = bytenr;
5172                 extent_key.offset = num_bytes;
5173                 extent_key.type = BTRFS_EXTENT_ITEM_KEY;
5174                 nr_extent = 1;
5175                 ret = get_new_locations(reloc_inode, &extent_key,
5176                                         group->key.objectid, 1,
5177                                         &new_extent, &nr_extent);
5178                 if (ret > 0)
5179                         continue;
5180                 BUG_ON(ret < 0);
5181
5182                 BUG_ON(ref->extents[ext_index].bytenr != bytenr);
5183                 BUG_ON(ref->extents[ext_index].num_bytes != num_bytes);
5184                 ref->extents[ext_index].bytenr = new_extent->disk_bytenr;
5185                 ref->extents[ext_index].num_bytes = new_extent->disk_num_bytes;
5186
5187                 btrfs_set_file_extent_disk_bytenr(leaf, fi,
5188                                                 new_extent->disk_bytenr);
5189                 btrfs_set_file_extent_disk_num_bytes(leaf, fi,
5190                                                 new_extent->disk_num_bytes);
5191                 btrfs_mark_buffer_dirty(leaf);
5192
5193                 ret = btrfs_inc_extent_ref(trans, root,
5194                                         new_extent->disk_bytenr,
5195                                         new_extent->disk_num_bytes,
5196                                         leaf->start,
5197                                         root->root_key.objectid,
5198                                         trans->transid, key.objectid);
5199                 BUG_ON(ret);
5200                 ret = btrfs_free_extent(trans, root,
5201                                         bytenr, num_bytes, leaf->start,
5202                                         btrfs_header_owner(leaf),
5203                                         btrfs_header_generation(leaf),
5204                                         key.objectid, 0);
5205                 BUG_ON(ret);
5206                 cond_resched();
5207         }
5208         kfree(new_extent);
5209         BUG_ON(ext_index + 1 != ref->nritems);
5210         btrfs_free_leaf_ref(root, ref);
5211         return 0;
5212 }
5213
5214 int btrfs_free_reloc_root(struct btrfs_trans_handle *trans,
5215                           struct btrfs_root *root)
5216 {
5217         struct btrfs_root *reloc_root;
5218         int ret;
5219
5220         if (root->reloc_root) {
5221                 reloc_root = root->reloc_root;
5222                 root->reloc_root = NULL;
5223                 list_add(&reloc_root->dead_list,
5224                          &root->fs_info->dead_reloc_roots);
5225
5226                 btrfs_set_root_bytenr(&reloc_root->root_item,
5227                                       reloc_root->node->start);
5228                 btrfs_set_root_level(&root->root_item,
5229                                      btrfs_header_level(reloc_root->node));
5230                 memset(&reloc_root->root_item.drop_progress, 0,
5231                         sizeof(struct btrfs_disk_key));
5232                 reloc_root->root_item.drop_level = 0;
5233
5234                 ret = btrfs_update_root(trans, root->fs_info->tree_root,
5235                                         &reloc_root->root_key,
5236                                         &reloc_root->root_item);
5237                 BUG_ON(ret);
5238         }
5239         return 0;
5240 }
5241
5242 int btrfs_drop_dead_reloc_roots(struct btrfs_root *root)
5243 {
5244         struct btrfs_trans_handle *trans;
5245         struct btrfs_root *reloc_root;
5246         struct btrfs_root *prev_root = NULL;
5247         struct list_head dead_roots;
5248         int ret;
5249         unsigned long nr;
5250
5251         INIT_LIST_HEAD(&dead_roots);
5252         list_splice_init(&root->fs_info->dead_reloc_roots, &dead_roots);
5253
5254         while (!list_empty(&dead_roots)) {
5255                 reloc_root = list_entry(dead_roots.prev,
5256                                         struct btrfs_root, dead_list);
5257                 list_del_init(&reloc_root->dead_list);
5258
5259                 BUG_ON(reloc_root->commit_root != NULL);
5260                 while (1) {
5261                         trans = btrfs_join_transaction(root, 1);
5262                         BUG_ON(!trans);
5263
5264                         mutex_lock(&root->fs_info->drop_mutex);
5265                         ret = btrfs_drop_snapshot(trans, reloc_root);
5266                         if (ret != -EAGAIN)
5267                                 break;
5268                         mutex_unlock(&root->fs_info->drop_mutex);
5269
5270                         nr = trans->blocks_used;
5271                         ret = btrfs_end_transaction(trans, root);
5272                         BUG_ON(ret);
5273                         btrfs_btree_balance_dirty(root, nr);
5274                 }
5275
5276                 free_extent_buffer(reloc_root->node);
5277
5278                 ret = btrfs_del_root(trans, root->fs_info->tree_root,
5279                                      &reloc_root->root_key);
5280                 BUG_ON(ret);
5281                 mutex_unlock(&root->fs_info->drop_mutex);
5282
5283                 nr = trans->blocks_used;
5284                 ret = btrfs_end_transaction(trans, root);
5285                 BUG_ON(ret);
5286                 btrfs_btree_balance_dirty(root, nr);
5287
5288                 kfree(prev_root);
5289                 prev_root = reloc_root;
5290         }
5291         if (prev_root) {
5292                 btrfs_remove_leaf_refs(prev_root, (u64)-1, 0);
5293                 kfree(prev_root);
5294         }
5295         return 0;
5296 }
5297
5298 int btrfs_add_dead_reloc_root(struct btrfs_root *root)
5299 {
5300         list_add(&root->dead_list, &root->fs_info->dead_reloc_roots);
5301         return 0;
5302 }
5303
5304 int btrfs_cleanup_reloc_trees(struct btrfs_root *root)
5305 {
5306         struct btrfs_root *reloc_root;
5307         struct btrfs_trans_handle *trans;
5308         struct btrfs_key location;
5309         int found;
5310         int ret;
5311
5312         mutex_lock(&root->fs_info->tree_reloc_mutex);
5313         ret = btrfs_find_dead_roots(root, BTRFS_TREE_RELOC_OBJECTID, NULL);
5314         BUG_ON(ret);
5315         found = !list_empty(&root->fs_info->dead_reloc_roots);
5316         mutex_unlock(&root->fs_info->tree_reloc_mutex);
5317
5318         if (found) {
5319                 trans = btrfs_start_transaction(root, 1);
5320                 BUG_ON(!trans);
5321                 ret = btrfs_commit_transaction(trans, root);
5322                 BUG_ON(ret);
5323         }
5324
5325         location.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID;
5326         location.offset = (u64)-1;
5327         location.type = BTRFS_ROOT_ITEM_KEY;
5328
5329         reloc_root = btrfs_read_fs_root_no_name(root->fs_info, &location);
5330         BUG_ON(!reloc_root);
5331         btrfs_orphan_cleanup(reloc_root);
5332         return 0;
5333 }
5334
5335 static noinline int init_reloc_tree(struct btrfs_trans_handle *trans,
5336                                     struct btrfs_root *root)
5337 {
5338         struct btrfs_root *reloc_root;
5339         struct extent_buffer *eb;
5340         struct btrfs_root_item *root_item;
5341         struct btrfs_key root_key;
5342         int ret;
5343
5344         BUG_ON(!root->ref_cows);
5345         if (root->reloc_root)
5346                 return 0;
5347
5348         root_item = kmalloc(sizeof(*root_item), GFP_NOFS);
5349         BUG_ON(!root_item);
5350
5351         ret = btrfs_copy_root(trans, root, root->commit_root,
5352                               &eb, BTRFS_TREE_RELOC_OBJECTID);
5353         BUG_ON(ret);
5354
5355         root_key.objectid = BTRFS_TREE_RELOC_OBJECTID;
5356         root_key.offset = root->root_key.objectid;
5357         root_key.type = BTRFS_ROOT_ITEM_KEY;
5358
5359         memcpy(root_item, &root->root_item, sizeof(root_item));
5360         btrfs_set_root_refs(root_item, 0);
5361         btrfs_set_root_bytenr(root_item, eb->start);
5362         btrfs_set_root_level(root_item, btrfs_header_level(eb));
5363         btrfs_set_root_generation(root_item, trans->transid);
5364
5365         btrfs_tree_unlock(eb);
5366         free_extent_buffer(eb);
5367
5368         ret = btrfs_insert_root(trans, root->fs_info->tree_root,
5369                                 &root_key, root_item);
5370         BUG_ON(ret);
5371         kfree(root_item);
5372
5373         reloc_root = btrfs_read_fs_root_no_radix(root->fs_info->tree_root,
5374                                                  &root_key);
5375         BUG_ON(!reloc_root);
5376         reloc_root->last_trans = trans->transid;
5377         reloc_root->commit_root = NULL;
5378         reloc_root->ref_tree = &root->fs_info->reloc_ref_tree;
5379
5380         root->reloc_root = reloc_root;
5381         return 0;
5382 }
5383
5384 /*
5385  * Core function of space balance.
5386  *
5387  * The idea is using reloc trees to relocate tree blocks in reference
5388  * counted roots. There is one reloc tree for each subvol, and all
5389  * reloc trees share same root key objectid. Reloc trees are snapshots
5390  * of the latest committed roots of subvols (root->commit_root).
5391  *
5392  * To relocate a tree block referenced by a subvol, there are two steps.
5393  * COW the block through subvol's reloc tree, then update block pointer
5394  * in the subvol to point to the new block. Since all reloc trees share
5395  * same root key objectid, doing special handing for tree blocks owned
5396  * by them is easy. Once a tree block has been COWed in one reloc tree,
5397  * we can use the resulting new block directly when the same block is
5398  * required to COW again through other reloc trees. By this way, relocated
5399  * tree blocks are shared between reloc trees, so they are also shared
5400  * between subvols.
5401  */
5402 static noinline int relocate_one_path(struct btrfs_trans_handle *trans,
5403                                       struct btrfs_root *root,
5404                                       struct btrfs_path *path,
5405                                       struct btrfs_key *first_key,
5406                                       struct btrfs_ref_path *ref_path,
5407                                       struct btrfs_block_group_cache *group,
5408                                       struct inode *reloc_inode)
5409 {
5410         struct btrfs_root *reloc_root;
5411         struct extent_buffer *eb = NULL;
5412         struct btrfs_key *keys;
5413         u64 *nodes;
5414         int level;
5415         int shared_level;
5416         int lowest_level = 0;
5417         int ret;
5418
5419         if (ref_path->owner_objectid < BTRFS_FIRST_FREE_OBJECTID)
5420                 lowest_level = ref_path->owner_objectid;
5421
5422         if (!root->ref_cows) {
5423                 path->lowest_level = lowest_level;
5424                 ret = btrfs_search_slot(trans, root, first_key, path, 0, 1);
5425                 BUG_ON(ret < 0);
5426                 path->lowest_level = 0;
5427                 btrfs_release_path(root, path);
5428                 return 0;
5429         }
5430
5431         mutex_lock(&root->fs_info->tree_reloc_mutex);
5432         ret = init_reloc_tree(trans, root);
5433         BUG_ON(ret);
5434         reloc_root = root->reloc_root;
5435
5436         shared_level = ref_path->shared_level;
5437         ref_path->shared_level = BTRFS_MAX_LEVEL - 1;
5438
5439         keys = ref_path->node_keys;
5440         nodes = ref_path->new_nodes;
5441         memset(&keys[shared_level + 1], 0,
5442                sizeof(*keys) * (BTRFS_MAX_LEVEL - shared_level - 1));
5443         memset(&nodes[shared_level + 1], 0,
5444                sizeof(*nodes) * (BTRFS_MAX_LEVEL - shared_level - 1));
5445
5446         if (nodes[lowest_level] == 0) {
5447                 path->lowest_level = lowest_level;
5448                 ret = btrfs_search_slot(trans, reloc_root, first_key, path,
5449                                         0, 1);
5450                 BUG_ON(ret);
5451                 for (level = lowest_level; level < BTRFS_MAX_LEVEL; level++) {
5452                         eb = path->nodes[level];
5453                         if (!eb || eb == reloc_root->node)
5454                                 break;
5455                         nodes[level] = eb->start;
5456                         if (level == 0)
5457                                 btrfs_item_key_to_cpu(eb, &keys[level], 0);
5458                         else
5459                                 btrfs_node_key_to_cpu(eb, &keys[level], 0);
5460                 }
5461                 if (nodes[0] &&
5462                     ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5463                         eb = path->nodes[0];
5464                         ret = replace_extents_in_leaf(trans, reloc_root, eb,
5465                                                       group, reloc_inode);
5466                         BUG_ON(ret);
5467                 }
5468                 btrfs_release_path(reloc_root, path);
5469         } else {
5470                 ret = btrfs_merge_path(trans, reloc_root, keys, nodes,
5471                                        lowest_level);
5472                 BUG_ON(ret);
5473         }
5474
5475         /*
5476          * replace tree blocks in the fs tree with tree blocks in
5477          * the reloc tree.
5478          */
5479         ret = btrfs_merge_path(trans, root, keys, nodes, lowest_level);
5480         BUG_ON(ret < 0);
5481
5482         if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5483                 ret = btrfs_search_slot(trans, reloc_root, first_key, path,
5484                                         0, 0);
5485                 BUG_ON(ret);
5486                 extent_buffer_get(path->nodes[0]);
5487                 eb = path->nodes[0];
5488                 btrfs_release_path(reloc_root, path);
5489                 ret = invalidate_extent_cache(reloc_root, eb, group, root);
5490                 BUG_ON(ret);
5491                 free_extent_buffer(eb);
5492         }
5493
5494         mutex_unlock(&root->fs_info->tree_reloc_mutex);
5495         path->lowest_level = 0;
5496         return 0;
5497 }
5498
5499 static noinline int relocate_tree_block(struct btrfs_trans_handle *trans,
5500                                         struct btrfs_root *root,
5501                                         struct btrfs_path *path,
5502                                         struct btrfs_key *first_key,
5503                                         struct btrfs_ref_path *ref_path)
5504 {
5505         int ret;
5506
5507         ret = relocate_one_path(trans, root, path, first_key,
5508                                 ref_path, NULL, NULL);
5509         BUG_ON(ret);
5510
5511         if (root == root->fs_info->extent_root)
5512                 btrfs_extent_post_op(trans, root);
5513
5514         return 0;
5515 }
5516
5517 static noinline int del_extent_zero(struct btrfs_trans_handle *trans,
5518                                     struct btrfs_root *extent_root,
5519                                     struct btrfs_path *path,
5520                                     struct btrfs_key *extent_key)
5521 {
5522         int ret;
5523
5524         ret = btrfs_search_slot(trans, extent_root, extent_key, path, -1, 1);
5525         if (ret)
5526                 goto out;
5527         ret = btrfs_del_item(trans, extent_root, path);
5528 out:
5529         btrfs_release_path(extent_root, path);
5530         return ret;
5531 }
5532
5533 static noinline struct btrfs_root *read_ref_root(struct btrfs_fs_info *fs_info,
5534                                                 struct btrfs_ref_path *ref_path)
5535 {
5536         struct btrfs_key root_key;
5537
5538         root_key.objectid = ref_path->root_objectid;
5539         root_key.type = BTRFS_ROOT_ITEM_KEY;
5540         if (is_cowonly_root(ref_path->root_objectid))
5541                 root_key.offset = 0;
5542         else
5543                 root_key.offset = (u64)-1;
5544
5545         return btrfs_read_fs_root_no_name(fs_info, &root_key);
5546 }
5547
5548 static noinline int relocate_one_extent(struct btrfs_root *extent_root,
5549                                         struct btrfs_path *path,
5550                                         struct btrfs_key *extent_key,
5551                                         struct btrfs_block_group_cache *group,
5552                                         struct inode *reloc_inode, int pass)
5553 {
5554         struct btrfs_trans_handle *trans;
5555         struct btrfs_root *found_root;
5556         struct btrfs_ref_path *ref_path = NULL;
5557         struct disk_extent *new_extents = NULL;
5558         int nr_extents = 0;
5559         int loops;
5560         int ret;
5561         int level;
5562         struct btrfs_key first_key;
5563         u64 prev_block = 0;
5564
5565
5566         trans = btrfs_start_transaction(extent_root, 1);
5567         BUG_ON(!trans);
5568
5569         if (extent_key->objectid == 0) {
5570                 ret = del_extent_zero(trans, extent_root, path, extent_key);
5571                 goto out;
5572         }
5573
5574         ref_path = kmalloc(sizeof(*ref_path), GFP_NOFS);
5575         if (!ref_path) {
5576                 ret = -ENOMEM;
5577                 goto out;
5578         }
5579
5580         for (loops = 0; ; loops++) {
5581                 if (loops == 0) {
5582                         ret = btrfs_first_ref_path(trans, extent_root, ref_path,
5583                                                    extent_key->objectid);
5584                 } else {
5585                         ret = btrfs_next_ref_path(trans, extent_root, ref_path);
5586                 }
5587                 if (ret < 0)
5588                         goto out;
5589                 if (ret > 0)
5590                         break;
5591
5592                 if (ref_path->root_objectid == BTRFS_TREE_LOG_OBJECTID ||
5593                     ref_path->root_objectid == BTRFS_TREE_RELOC_OBJECTID)
5594                         continue;
5595
5596                 found_root = read_ref_root(extent_root->fs_info, ref_path);
5597                 BUG_ON(!found_root);
5598                 /*
5599                  * for reference counted tree, only process reference paths
5600                  * rooted at the latest committed root.
5601                  */
5602                 if (found_root->ref_cows &&
5603                     ref_path->root_generation != found_root->root_key.offset)
5604                         continue;
5605
5606                 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5607                         if (pass == 0) {
5608                                 /*
5609                                  * copy data extents to new locations
5610                                  */
5611                                 u64 group_start = group->key.objectid;
5612                                 ret = relocate_data_extent(reloc_inode,
5613                                                            extent_key,
5614                                                            group_start);
5615                                 if (ret < 0)
5616                                         goto out;
5617                                 break;
5618                         }
5619                         level = 0;
5620                 } else {
5621                         level = ref_path->owner_objectid;
5622                 }
5623
5624                 if (prev_block != ref_path->nodes[level]) {
5625                         struct extent_buffer *eb;
5626                         u64 block_start = ref_path->nodes[level];
5627                         u64 block_size = btrfs_level_size(found_root, level);
5628
5629                         eb = read_tree_block(found_root, block_start,
5630                                              block_size, 0);
5631                         btrfs_tree_lock(eb);
5632                         BUG_ON(level != btrfs_header_level(eb));
5633
5634                         if (level == 0)
5635                                 btrfs_item_key_to_cpu(eb, &first_key, 0);
5636                         else
5637                                 btrfs_node_key_to_cpu(eb, &first_key, 0);
5638
5639                         btrfs_tree_unlock(eb);
5640                         free_extent_buffer(eb);
5641                         prev_block = block_start;
5642                 }
5643
5644                 btrfs_record_root_in_trans(found_root);
5645                 if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
5646                         /*
5647                          * try to update data extent references while
5648                          * keeping metadata shared between snapshots.
5649                          */
5650                         if (pass == 1) {
5651                                 ret = relocate_one_path(trans, found_root,
5652                                                 path, &first_key, ref_path,
5653                                                 group, reloc_inode);
5654                                 if (ret < 0)
5655                                         goto out;
5656                                 continue;
5657                         }
5658                         /*
5659                          * use fallback method to process the remaining
5660                          * references.
5661                          */
5662                         if (!new_extents) {
5663                                 u64 group_start = group->key.objectid;
5664                                 new_extents = kmalloc(sizeof(*new_extents),
5665                                                       GFP_NOFS);
5666                                 nr_extents = 1;
5667                                 ret = get_new_locations(reloc_inode,
5668                                                         extent_key,
5669                                                         group_start, 1,
5670                                                         &new_extents,
5671                                                         &nr_extents);
5672                                 if (ret)
5673                                         goto out;
5674                         }
5675                         ret = replace_one_extent(trans, found_root,
5676                                                 path, extent_key,
5677                                                 &first_key, ref_path,
5678                                                 new_extents, nr_extents);
5679                 } else {
5680                         ret = relocate_tree_block(trans, found_root, path,
5681                                                   &first_key, ref_path);
5682                 }
5683                 if (ret < 0)
5684                         goto out;
5685         }
5686         ret = 0;
5687 out:
5688         btrfs_end_transaction(trans, extent_root);
5689         kfree(new_extents);
5690         kfree(ref_path);
5691         return ret;
5692 }
5693
5694 static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
5695 {
5696         u64 num_devices;
5697         u64 stripped = BTRFS_BLOCK_GROUP_RAID0 |
5698                 BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10;
5699
5700         num_devices = root->fs_info->fs_devices->rw_devices;
5701         if (num_devices == 1) {
5702                 stripped |= BTRFS_BLOCK_GROUP_DUP;
5703                 stripped = flags & ~stripped;
5704
5705                 /* turn raid0 into single device chunks */
5706                 if (flags & BTRFS_BLOCK_GROUP_RAID0)
5707                         return stripped;
5708
5709                 /* turn mirroring into duplication */
5710                 if (flags & (BTRFS_BLOCK_GROUP_RAID1 |
5711                              BTRFS_BLOCK_GROUP_RAID10))
5712                         return stripped | BTRFS_BLOCK_GROUP_DUP;
5713                 return flags;
5714         } else {
5715                 /* they already had raid on here, just return */
5716                 if (flags & stripped)
5717                         return flags;
5718
5719                 stripped |= BTRFS_BLOCK_GROUP_DUP;
5720                 stripped = flags & ~stripped;
5721
5722                 /* switch duplicated blocks with raid1 */
5723                 if (flags & BTRFS_BLOCK_GROUP_DUP)
5724                         return stripped | BTRFS_BLOCK_GROUP_RAID1;
5725
5726                 /* turn single device chunks into raid0 */
5727                 return stripped | BTRFS_BLOCK_GROUP_RAID0;
5728         }
5729         return flags;
5730 }
5731
5732 static int __alloc_chunk_for_shrink(struct btrfs_root *root,
5733                      struct btrfs_block_group_cache *shrink_block_group,
5734                      int force)
5735 {
5736         struct btrfs_trans_handle *trans;
5737         u64 new_alloc_flags;
5738         u64 calc;
5739
5740         spin_lock(&shrink_block_group->lock);
5741         if (btrfs_block_group_used(&shrink_block_group->item) > 0) {
5742                 spin_unlock(&shrink_block_group->lock);
5743
5744                 trans = btrfs_start_transaction(root, 1);
5745                 spin_lock(&shrink_block_group->lock);
5746
5747                 new_alloc_flags = update_block_group_flags(root,
5748                                                    shrink_block_group->flags);
5749                 if (new_alloc_flags != shrink_block_group->flags) {
5750                         calc =
5751                              btrfs_block_group_used(&shrink_block_group->item);
5752                 } else {
5753                         calc = shrink_block_group->key.offset;
5754                 }
5755                 spin_unlock(&shrink_block_group->lock);
5756
5757                 do_chunk_alloc(trans, root->fs_info->extent_root,
5758                                calc + 2 * 1024 * 1024, new_alloc_flags, force);
5759
5760                 btrfs_end_transaction(trans, root);
5761         } else
5762                 spin_unlock(&shrink_block_group->lock);
5763         return 0;
5764 }
5765
5766 static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
5767                                  struct btrfs_root *root,
5768                                  u64 objectid, u64 size)
5769 {
5770         struct btrfs_path *path;
5771         struct btrfs_inode_item *item;
5772         struct extent_buffer *leaf;
5773         int ret;
5774
5775         path = btrfs_alloc_path();
5776         if (!path)
5777                 return -ENOMEM;
5778
5779         ret = btrfs_insert_empty_inode(trans, root, path, objectid);
5780         if (ret)
5781                 goto out;
5782
5783         leaf = path->nodes[0];
5784         item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item);
5785         memset_extent_buffer(leaf, 0, (unsigned long)item, sizeof(*item));
5786         btrfs_set_inode_generation(leaf, item, 1);
5787         btrfs_set_inode_size(leaf, item, size);
5788         btrfs_set_inode_mode(leaf, item, S_IFREG | 0600);
5789         btrfs_set_inode_flags(leaf, item, BTRFS_INODE_NOCOMPRESS);
5790         btrfs_mark_buffer_dirty(leaf);
5791         btrfs_release_path(root, path);
5792 out:
5793         btrfs_free_path(path);
5794         return ret;
5795 }
5796
5797 static noinline struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info,
5798                                         struct btrfs_block_group_cache *group)
5799 {
5800         struct inode *inode = NULL;
5801         struct btrfs_trans_handle *trans;
5802         struct btrfs_root *root;
5803         struct btrfs_key root_key;
5804         u64 objectid = BTRFS_FIRST_FREE_OBJECTID;
5805         int err = 0;
5806
5807         root_key.objectid = BTRFS_DATA_RELOC_TREE_OBJECTID;
5808         root_key.type = BTRFS_ROOT_ITEM_KEY;
5809         root_key.offset = (u64)-1;
5810         root = btrfs_read_fs_root_no_name(fs_info, &root_key);
5811         if (IS_ERR(root))
5812                 return ERR_CAST(root);
5813
5814         trans = btrfs_start_transaction(root, 1);
5815         BUG_ON(!trans);
5816
5817         err = btrfs_find_free_objectid(trans, root, objectid, &objectid);
5818         if (err)
5819                 goto out;
5820
5821         err = __insert_orphan_inode(trans, root, objectid, group->key.offset);
5822         BUG_ON(err);
5823
5824         err = btrfs_insert_file_extent(trans, root, objectid, 0, 0, 0,
5825                                        group->key.offset, 0, group->key.offset,
5826                                        0, 0, 0);
5827         BUG_ON(err);
5828
5829         inode = btrfs_iget_locked(root->fs_info->sb, objectid, root);
5830         if (inode->i_state & I_NEW) {
5831                 BTRFS_I(inode)->root = root;
5832                 BTRFS_I(inode)->location.objectid = objectid;
5833                 BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
5834                 BTRFS_I(inode)->location.offset = 0;
5835                 btrfs_read_locked_inode(inode);
5836                 unlock_new_inode(inode);
5837                 BUG_ON(is_bad_inode(inode));
5838         } else {
5839                 BUG_ON(1);
5840         }
5841         BTRFS_I(inode)->index_cnt = group->key.objectid;
5842
5843         err = btrfs_orphan_add(trans, inode);
5844 out:
5845         btrfs_end_transaction(trans, root);
5846         if (err) {
5847                 if (inode)
5848                         iput(inode);
5849                 inode = ERR_PTR(err);
5850         }
5851         return inode;
5852 }
5853
5854 int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len)
5855 {
5856
5857         struct btrfs_ordered_sum *sums;
5858         struct btrfs_sector_sum *sector_sum;
5859         struct btrfs_ordered_extent *ordered;
5860         struct btrfs_root *root = BTRFS_I(inode)->root;
5861         struct list_head list;
5862         size_t offset;
5863         int ret;
5864         u64 disk_bytenr;
5865
5866         INIT_LIST_HEAD(&list);
5867
5868         ordered = btrfs_lookup_ordered_extent(inode, file_pos);
5869         BUG_ON(ordered->file_offset != file_pos || ordered->len != len);
5870
5871         disk_bytenr = file_pos + BTRFS_I(inode)->index_cnt;
5872         ret = btrfs_lookup_csums_range(root->fs_info->csum_root, disk_bytenr,
5873                                        disk_bytenr + len - 1, &list);
5874
5875         while (!list_empty(&list)) {
5876                 sums = list_entry(list.next, struct btrfs_ordered_sum, list);
5877                 list_del_init(&sums->list);
5878
5879                 sector_sum = sums->sums;
5880                 sums->bytenr = ordered->start;
5881
5882                 offset = 0;
5883                 while (offset < sums->len) {
5884                         sector_sum->bytenr += ordered->start - disk_bytenr;
5885                         sector_sum++;
5886                         offset += root->sectorsize;
5887                 }
5888
5889                 btrfs_add_ordered_sum(inode, ordered, sums);
5890         }
5891         btrfs_put_ordered_extent(ordered);
5892         return 0;
5893 }
5894
5895 int btrfs_relocate_block_group(struct btrfs_root *root, u64 group_start)
5896 {
5897         struct btrfs_trans_handle *trans;
5898         struct btrfs_path *path;
5899         struct btrfs_fs_info *info = root->fs_info;
5900         struct extent_buffer *leaf;
5901         struct inode *reloc_inode;
5902         struct btrfs_block_group_cache *block_group;
5903         struct btrfs_key key;
5904         u64 skipped;
5905         u64 cur_byte;
5906         u64 total_found;
5907         u32 nritems;
5908         int ret;
5909         int progress;
5910         int pass = 0;
5911
5912         root = root->fs_info->extent_root;
5913
5914         block_group = btrfs_lookup_block_group(info, group_start);
5915         BUG_ON(!block_group);
5916
5917         printk(KERN_INFO "btrfs relocating block group %llu flags %llu\n",
5918                (unsigned long long)block_group->key.objectid,
5919                (unsigned long long)block_group->flags);
5920
5921         path = btrfs_alloc_path();
5922         BUG_ON(!path);
5923
5924         reloc_inode = create_reloc_inode(info, block_group);
5925         BUG_ON(IS_ERR(reloc_inode));
5926
5927         __alloc_chunk_for_shrink(root, block_group, 1);
5928         set_block_group_readonly(block_group);
5929
5930         btrfs_start_delalloc_inodes(info->tree_root);
5931         btrfs_wait_ordered_extents(info->tree_root, 0);
5932 again:
5933         skipped = 0;
5934         total_found = 0;
5935         progress = 0;
5936         key.objectid = block_group->key.objectid;
5937         key.offset = 0;
5938         key.type = 0;
5939         cur_byte = key.objectid;
5940
5941         trans = btrfs_start_transaction(info->tree_root, 1);
5942         btrfs_commit_transaction(trans, info->tree_root);
5943
5944         mutex_lock(&root->fs_info->cleaner_mutex);
5945         btrfs_clean_old_snapshots(info->tree_root);
5946         btrfs_remove_leaf_refs(info->tree_root, (u64)-1, 1);
5947         mutex_unlock(&root->fs_info->cleaner_mutex);
5948
5949         while (1) {
5950                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5951                 if (ret < 0)
5952                         goto out;
5953 next:
5954                 leaf = path->nodes[0];
5955                 nritems = btrfs_header_nritems(leaf);
5956                 if (path->slots[0] >= nritems) {
5957                         ret = btrfs_next_leaf(root, path);
5958                         if (ret < 0)
5959                                 goto out;
5960                         if (ret == 1) {
5961                                 ret = 0;
5962                                 break;
5963                         }
5964                         leaf = path->nodes[0];
5965                         nritems = btrfs_header_nritems(leaf);
5966                 }
5967
5968                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
5969
5970                 if (key.objectid >= block_group->key.objectid +
5971                     block_group->key.offset)
5972                         break;
5973
5974                 if (progress && need_resched()) {
5975                         btrfs_release_path(root, path);
5976                         cond_resched();
5977                         progress = 0;
5978                         continue;
5979                 }
5980                 progress = 1;
5981
5982                 if (btrfs_key_type(&key) != BTRFS_EXTENT_ITEM_KEY ||
5983                     key.objectid + key.offset <= cur_byte) {
5984                         path->slots[0]++;
5985                         goto next;
5986                 }
5987
5988                 total_found++;
5989                 cur_byte = key.objectid + key.offset;
5990                 btrfs_release_path(root, path);
5991
5992                 __alloc_chunk_for_shrink(root, block_group, 0);
5993                 ret = relocate_one_extent(root, path, &key, block_group,
5994                                           reloc_inode, pass);
5995                 BUG_ON(ret < 0);
5996                 if (ret > 0)
5997                         skipped++;
5998
5999                 key.objectid = cur_byte;
6000                 key.type = 0;
6001                 key.offset = 0;
6002         }
6003
6004         btrfs_release_path(root, path);
6005
6006         if (pass == 0) {
6007                 btrfs_wait_ordered_range(reloc_inode, 0, (u64)-1);
6008                 invalidate_mapping_pages(reloc_inode->i_mapping, 0, -1);
6009         }
6010
6011         if (total_found > 0) {
6012                 printk(KERN_INFO "btrfs found %llu extents in pass %d\n",
6013                        (unsigned long long)total_found, pass);
6014                 pass++;
6015                 if (total_found == skipped && pass > 2) {
6016                         iput(reloc_inode);
6017                         reloc_inode = create_reloc_inode(info, block_group);
6018                         pass = 0;
6019                 }
6020                 goto again;
6021         }
6022
6023         /* delete reloc_inode */
6024         iput(reloc_inode);
6025
6026         /* unpin extents in this range */
6027         trans = btrfs_start_transaction(info->tree_root, 1);
6028         btrfs_commit_transaction(trans, info->tree_root);
6029
6030         spin_lock(&block_group->lock);
6031         WARN_ON(block_group->pinned > 0);
6032         WARN_ON(block_group->reserved > 0);
6033         WARN_ON(btrfs_block_group_used(&block_group->item) > 0);
6034         spin_unlock(&block_group->lock);
6035         put_block_group(block_group);
6036         ret = 0;
6037 out:
6038         btrfs_free_path(path);
6039         return ret;
6040 }
6041
6042 static int find_first_block_group(struct btrfs_root *root,
6043                 struct btrfs_path *path, struct btrfs_key *key)
6044 {
6045         int ret = 0;
6046         struct btrfs_key found_key;
6047         struct extent_buffer *leaf;
6048         int slot;
6049
6050         ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
6051         if (ret < 0)
6052                 goto out;
6053
6054         while (1) {
6055                 slot = path->slots[0];
6056                 leaf = path->nodes[0];
6057                 if (slot >= btrfs_header_nritems(leaf)) {
6058                         ret = btrfs_next_leaf(root, path);
6059                         if (ret == 0)
6060                                 continue;
6061                         if (ret < 0)
6062                                 goto out;
6063                         break;
6064                 }
6065                 btrfs_item_key_to_cpu(leaf, &found_key, slot);
6066
6067                 if (found_key.objectid >= key->objectid &&
6068                     found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) {
6069                         ret = 0;
6070                         goto out;
6071                 }
6072                 path->slots[0]++;
6073         }
6074         ret = -ENOENT;
6075 out:
6076         return ret;
6077 }
6078
6079 int btrfs_free_block_groups(struct btrfs_fs_info *info)
6080 {
6081         struct btrfs_block_group_cache *block_group;
6082         struct rb_node *n;
6083
6084         spin_lock(&info->block_group_cache_lock);
6085         while ((n = rb_last(&info->block_group_cache_tree)) != NULL) {
6086                 block_group = rb_entry(n, struct btrfs_block_group_cache,
6087                                        cache_node);
6088                 rb_erase(&block_group->cache_node,
6089                          &info->block_group_cache_tree);
6090                 spin_unlock(&info->block_group_cache_lock);
6091
6092                 btrfs_remove_free_space_cache(block_group);
6093                 down_write(&block_group->space_info->groups_sem);
6094                 list_del(&block_group->list);
6095                 up_write(&block_group->space_info->groups_sem);
6096
6097                 WARN_ON(atomic_read(&block_group->count) != 1);
6098                 kfree(block_group);
6099
6100                 spin_lock(&info->block_group_cache_lock);
6101         }
6102         spin_unlock(&info->block_group_cache_lock);
6103         return 0;
6104 }
6105
6106 int btrfs_read_block_groups(struct btrfs_root *root)
6107 {
6108         struct btrfs_path *path;
6109         int ret;
6110         struct btrfs_block_group_cache *cache;
6111         struct btrfs_fs_info *info = root->fs_info;
6112         struct btrfs_space_info *space_info;
6113         struct btrfs_key key;
6114         struct btrfs_key found_key;
6115         struct extent_buffer *leaf;
6116
6117         root = info->extent_root;
6118         key.objectid = 0;
6119         key.offset = 0;
6120         btrfs_set_key_type(&key, BTRFS_BLOCK_GROUP_ITEM_KEY);
6121         path = btrfs_alloc_path();
6122         if (!path)
6123                 return -ENOMEM;
6124
6125         while (1) {
6126                 ret = find_first_block_group(root, path, &key);
6127                 if (ret > 0) {
6128                         ret = 0;
6129                         goto error;
6130                 }
6131                 if (ret != 0)
6132                         goto error;
6133
6134                 leaf = path->nodes[0];
6135                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
6136                 cache = kzalloc(sizeof(*cache), GFP_NOFS);
6137                 if (!cache) {
6138                         ret = -ENOMEM;
6139                         break;
6140                 }
6141
6142                 atomic_set(&cache->count, 1);
6143                 spin_lock_init(&cache->lock);
6144                 mutex_init(&cache->alloc_mutex);
6145                 mutex_init(&cache->cache_mutex);
6146                 INIT_LIST_HEAD(&cache->list);
6147                 read_extent_buffer(leaf, &cache->item,
6148                                    btrfs_item_ptr_offset(leaf, path->slots[0]),
6149                                    sizeof(cache->item));
6150                 memcpy(&cache->key, &found_key, sizeof(found_key));
6151
6152                 key.objectid = found_key.objectid + found_key.offset;
6153                 btrfs_release_path(root, path);
6154                 cache->flags = btrfs_block_group_flags(&cache->item);
6155
6156                 ret = update_space_info(info, cache->flags, found_key.offset,
6157                                         btrfs_block_group_used(&cache->item),
6158                                         &space_info);
6159                 BUG_ON(ret);
6160                 cache->space_info = space_info;
6161                 down_write(&space_info->groups_sem);
6162                 list_add_tail(&cache->list, &space_info->block_groups);
6163                 up_write(&space_info->groups_sem);
6164
6165                 ret = btrfs_add_block_group_cache(root->fs_info, cache);
6166                 BUG_ON(ret);
6167
6168                 set_avail_alloc_bits(root->fs_info, cache->flags);
6169                 if (btrfs_chunk_readonly(root, cache->key.objectid))
6170                         set_block_group_readonly(cache);
6171         }
6172         ret = 0;
6173 error:
6174         btrfs_free_path(path);
6175         return ret;
6176 }
6177
6178 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
6179                            struct btrfs_root *root, u64 bytes_used,
6180                            u64 type, u64 chunk_objectid, u64 chunk_offset,
6181                            u64 size)
6182 {
6183         int ret;
6184         struct btrfs_root *extent_root;
6185         struct btrfs_block_group_cache *cache;
6186
6187         extent_root = root->fs_info->extent_root;
6188
6189         root->fs_info->last_trans_new_blockgroup = trans->transid;
6190
6191         cache = kzalloc(sizeof(*cache), GFP_NOFS);
6192         if (!cache)
6193                 return -ENOMEM;
6194
6195         cache->key.objectid = chunk_offset;
6196         cache->key.offset = size;
6197         cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
6198         atomic_set(&cache->count, 1);
6199         spin_lock_init(&cache->lock);
6200         mutex_init(&cache->alloc_mutex);
6201         mutex_init(&cache->cache_mutex);
6202         INIT_LIST_HEAD(&cache->list);
6203
6204         btrfs_set_block_group_used(&cache->item, bytes_used);
6205         btrfs_set_block_group_chunk_objectid(&cache->item, chunk_objectid);
6206         cache->flags = type;
6207         btrfs_set_block_group_flags(&cache->item, type);
6208
6209         ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
6210                                 &cache->space_info);
6211         BUG_ON(ret);
6212         down_write(&cache->space_info->groups_sem);
6213         list_add_tail(&cache->list, &cache->space_info->block_groups);
6214         up_write(&cache->space_info->groups_sem);
6215
6216         ret = btrfs_add_block_group_cache(root->fs_info, cache);
6217         BUG_ON(ret);
6218
6219         ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
6220                                 sizeof(cache->item));
6221         BUG_ON(ret);
6222
6223         finish_current_insert(trans, extent_root, 0);
6224         ret = del_pending_extents(trans, extent_root, 0);
6225         BUG_ON(ret);
6226         set_avail_alloc_bits(extent_root->fs_info, type);
6227
6228         return 0;
6229 }
6230
6231 int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
6232                              struct btrfs_root *root, u64 group_start)
6233 {
6234         struct btrfs_path *path;
6235         struct btrfs_block_group_cache *block_group;
6236         struct btrfs_key key;
6237         int ret;
6238
6239         root = root->fs_info->extent_root;
6240
6241         block_group = btrfs_lookup_block_group(root->fs_info, group_start);
6242         BUG_ON(!block_group);
6243         BUG_ON(!block_group->ro);
6244
6245         memcpy(&key, &block_group->key, sizeof(key));
6246
6247         path = btrfs_alloc_path();
6248         BUG_ON(!path);
6249
6250         spin_lock(&root->fs_info->block_group_cache_lock);
6251         rb_erase(&block_group->cache_node,
6252                  &root->fs_info->block_group_cache_tree);
6253         spin_unlock(&root->fs_info->block_group_cache_lock);
6254         btrfs_remove_free_space_cache(block_group);
6255         down_write(&block_group->space_info->groups_sem);
6256         list_del(&block_group->list);
6257         up_write(&block_group->space_info->groups_sem);
6258
6259         spin_lock(&block_group->space_info->lock);
6260         block_group->space_info->total_bytes -= block_group->key.offset;
6261         block_group->space_info->bytes_readonly -= block_group->key.offset;
6262         spin_unlock(&block_group->space_info->lock);
6263         block_group->space_info->full = 0;
6264
6265         put_block_group(block_group);
6266         put_block_group(block_group);
6267
6268         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
6269         if (ret > 0)
6270                 ret = -EIO;
6271         if (ret < 0)
6272                 goto out;
6273
6274         ret = btrfs_del_item(trans, root, path);
6275 out:
6276         btrfs_free_path(path);
6277         return ret;
6278 }