fs/btrfs/extent-tree.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * Copyright (C) 2007 Oracle.  All rights reserved.
   4  */
   5
   6 #include <linux/sched.h>
   7 #include <linux/sched/signal.h>
   8 #include <linux/pagemap.h>
   9 #include <linux/writeback.h>
  10 #include <linux/blkdev.h>
  11 #include <linux/sort.h>
  12 #include <linux/rcupdate.h>
  13 #include <linux/kthread.h>
  14 #include <linux/slab.h>
  15 #include <linux/ratelimit.h>
  16 #include <linux/percpu_counter.h>
  17 #include <linux/lockdep.h>
  18 #include <linux/crc32c.h>
  19 #include "misc.h"
  20 #include "tree-log.h"
  21 #include "disk-io.h"
  22 #include "print-tree.h"
  23 #include "volumes.h"
  24 #include "raid56.h"
  25 #include "locking.h"
  26 #include "free-space-cache.h"
  27 #include "free-space-tree.h"
  28 #include "sysfs.h"
  29 #include "qgroup.h"
  30 #include "ref-verify.h"
  31 #include "space-info.h"
  32 #include "block-rsv.h"
  33 #include "delalloc-space.h"
  34 #include "block-group.h"
  35
  36 #undef SCRAMBLE_DELAYED_REFS
  37
  38
  39 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
  40                                struct btrfs_delayed_ref_node *node, u64 parent,
  41                                u64 root_objectid, u64 owner_objectid,
  42                                u64 owner_offset, int refs_to_drop,
  43                                struct btrfs_delayed_extent_op *extra_op);
  44 static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
  45                                     struct extent_buffer *leaf,
  46                                     struct btrfs_extent_item *ei);
  47 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
  48                                       u64 parent, u64 root_objectid,
  49                                       u64 flags, u64 owner, u64 offset,
  50                                       struct btrfs_key *ins, int ref_mod);
  51 static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
  52                                      struct btrfs_delayed_ref_node *node,
  53                                      struct btrfs_delayed_extent_op *extent_op);
  54 static int find_next_key(struct btrfs_path *path, int level,
  55                          struct btrfs_key *key);
  56
  57 static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
  58 {
  59         return (cache->flags & bits) == bits;
  60 }
  61
  62 int btrfs_add_excluded_extent(struct btrfs_fs_info *fs_info,
  63                               u64 start, u64 num_bytes)
  64 {
  65         u64 end = start + num_bytes - 1;
  66         set_extent_bits(&fs_info->freed_extents[0],
  67                         start, end, EXTENT_UPTODATE);
  68         set_extent_bits(&fs_info->freed_extents[1],
  69                         start, end, EXTENT_UPTODATE);
  70         return 0;
  71 }
  72
  73 void btrfs_free_excluded_extents(struct btrfs_block_group_cache *cache)
  74 {
  75         struct btrfs_fs_info *fs_info = cache->fs_info;
  76         u64 start, end;
  77
  78         start = cache->key.objectid;
  79         end = start + cache->key.offset - 1;
  80
  81         clear_extent_bits(&fs_info->freed_extents[0],
  82                           start, end, EXTENT_UPTODATE);
  83         clear_extent_bits(&fs_info->freed_extents[1],
  84                           start, end, EXTENT_UPTODATE);
  85 }
  86
  87 static u64 generic_ref_to_space_flags(struct btrfs_ref *ref)
  88 {
  89         if (ref->type == BTRFS_REF_METADATA) {
  90                 if (ref->tree_ref.root == BTRFS_CHUNK_TREE_OBJECTID)
  91                         return BTRFS_BLOCK_GROUP_SYSTEM;
  92                 else
  93                         return BTRFS_BLOCK_GROUP_METADATA;
  94         }
  95         return BTRFS_BLOCK_GROUP_DATA;
  96 }
  97
  98 static void add_pinned_bytes(struct btrfs_fs_info *fs_info,
  99                              struct btrfs_ref *ref)
 100 {
 101         struct btrfs_space_info *space_info;
 102         u64 flags = generic_ref_to_space_flags(ref);
 103
 104         space_info = btrfs_find_space_info(fs_info, flags);
 105         ASSERT(space_info);
 106         percpu_counter_add_batch(&space_info->total_bytes_pinned, ref->len,
 107                     BTRFS_TOTAL_BYTES_PINNED_BATCH);
 108 }
 109
 110 static void sub_pinned_bytes(struct btrfs_fs_info *fs_info,
 111                              struct btrfs_ref *ref)
 112 {
 113         struct btrfs_space_info *space_info;
 114         u64 flags = generic_ref_to_space_flags(ref);
 115
 116         space_info = btrfs_find_space_info(fs_info, flags);
 117         ASSERT(space_info);
 118         percpu_counter_add_batch(&space_info->total_bytes_pinned, -ref->len,
 119                     BTRFS_TOTAL_BYTES_PINNED_BATCH);
 120 }
 121
 122 /* simple helper to search for an existing data extent at a given offset */
 123 int btrfs_lookup_data_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len)
 124 {
 125         int ret;
 126         struct btrfs_key key;
 127         struct btrfs_path *path;
 128
 129         path = btrfs_alloc_path();
 130         if (!path)
 131                 return -ENOMEM;
 132
 133         key.objectid = start;
 134         key.offset = len;
 135         key.type = BTRFS_EXTENT_ITEM_KEY;
 136         ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
 137         btrfs_free_path(path);
 138         return ret;
 139 }
 140
 141 /*
 142  * helper function to lookup reference count and flags of a tree block.
 143  *
 144  * the head node for delayed ref is used to store the sum of all the
 145  * reference count modifications queued up in the rbtree. the head
 146  * node may also store the extent flags to set. This way you can check
 147  * to see what the reference count and extent flags would be if all of
 148  * the delayed refs are not processed.
 149  */
 150 int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
 151                              struct btrfs_fs_info *fs_info, u64 bytenr,
 152                              u64 offset, int metadata, u64 *refs, u64 *flags)
 153 {
 154         struct btrfs_delayed_ref_head *head;
 155         struct btrfs_delayed_ref_root *delayed_refs;
 156         struct btrfs_path *path;
 157         struct btrfs_extent_item *ei;
 158         struct extent_buffer *leaf;
 159         struct btrfs_key key;
 160         u32 item_size;
 161         u64 num_refs;
 162         u64 extent_flags;
 163         int ret;
 164
 165         /*
 166          * If we don't have skinny metadata, don't bother doing anything
 167          * different
 168          */
 169         if (metadata && !btrfs_fs_incompat(fs_info, SKINNY_METADATA)) {
 170                 offset = fs_info->nodesize;
 171                 metadata = 0;
 172         }
 173
 174         path = btrfs_alloc_path();
 175         if (!path)
 176                 return -ENOMEM;
 177
 178         if (!trans) {
 179                 path->skip_locking = 1;
 180                 path->search_commit_root = 1;
 181         }
 182
 183 search_again:
 184         key.objectid = bytenr;
 185         key.offset = offset;
 186         if (metadata)
 187                 key.type = BTRFS_METADATA_ITEM_KEY;
 188         else
 189                 key.type = BTRFS_EXTENT_ITEM_KEY;
 190
 191         ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 0);
 192         if (ret < 0)
 193                 goto out_free;
 194
 195         if (ret > 0 && metadata && key.type == BTRFS_METADATA_ITEM_KEY) {
 196                 if (path->slots[0]) {
 197                         path->slots[0]--;
 198                         btrfs_item_key_to_cpu(path->nodes[0], &key,
 199                                               path->slots[0]);
 200                         if (key.objectid == bytenr &&
 201                             key.type == BTRFS_EXTENT_ITEM_KEY &&
 202                             key.offset == fs_info->nodesize)
 203                                 ret = 0;
 204                 }
 205         }
 206
 207         if (ret == 0) {
 208                 leaf = path->nodes[0];
 209                 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
 210                 if (item_size >= sizeof(*ei)) {
 211                         ei = btrfs_item_ptr(leaf, path->slots[0],
 212                                             struct btrfs_extent_item);
 213                         num_refs = btrfs_extent_refs(leaf, ei);
 214                         extent_flags = btrfs_extent_flags(leaf, ei);
 215                 } else {
 216                         ret = -EINVAL;
 217                         btrfs_print_v0_err(fs_info);
 218                         if (trans)
 219                                 btrfs_abort_transaction(trans, ret);
 220                         else
 221                                 btrfs_handle_fs_error(fs_info, ret, NULL);
 222
 223                         goto out_free;
 224                 }
 225
 226                 BUG_ON(num_refs == 0);
 227         } else {
 228                 num_refs = 0;
 229                 extent_flags = 0;
 230                 ret = 0;
 231         }
 232
 233         if (!trans)
 234                 goto out;
 235
 236         delayed_refs = &trans->transaction->delayed_refs;
 237         spin_lock(&delayed_refs->lock);
 238         head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
 239         if (head) {
 240                 if (!mutex_trylock(&head->mutex)) {
 241                         refcount_inc(&head->refs);
 242                         spin_unlock(&delayed_refs->lock);
 243
 244                         btrfs_release_path(path);
 245
 246                         /*
 247                          * Mutex was contended, block until it's released and try
 248                          * again
 249                          */
 250                         mutex_lock(&head->mutex);
 251                         mutex_unlock(&head->mutex);
 252                         btrfs_put_delayed_ref_head(head);
 253                         goto search_again;
 254                 }
 255                 spin_lock(&head->lock);
 256                 if (head->extent_op && head->extent_op->update_flags)
 257                         extent_flags |= head->extent_op->flags_to_set;
 258                 else
 259                         BUG_ON(num_refs == 0);
 260
 261                 num_refs += head->ref_mod;
 262                 spin_unlock(&head->lock);
 263                 mutex_unlock(&head->mutex);
 264         }
 265         spin_unlock(&delayed_refs->lock);
 266 out:
 267         WARN_ON(num_refs == 0);
 268         if (refs)
 269                 *refs = num_refs;
 270         if (flags)
 271                 *flags = extent_flags;
 272 out_free:
 273         btrfs_free_path(path);
 274         return ret;
 275 }
 276
 277 /*
 278  * Back reference rules.  Back refs have three main goals:
 279  *
 280  * 1) differentiate between all holders of references to an extent so that
 281  *    when a reference is dropped we can make sure it was a valid reference
 282  *    before freeing the extent.
 283  *
 284  * 2) Provide enough information to quickly find the holders of an extent
 285  *    if we notice a given block is corrupted or bad.
 286  *
 287  * 3) Make it easy to migrate blocks for FS shrinking or storage pool
 288  *    maintenance.  This is actually the same as #2, but with a slightly
 289  *    different use case.
 290  *
 291  * There are two kinds of back refs. The implicit back refs is optimized
 292  * for pointers in non-shared tree blocks. For a given pointer in a block,
 293  * back refs of this kind provide information about the block's owner tree
 294  * and the pointer's key. These information allow us to find the block by
 295  * b-tree searching. The full back refs is for pointers in tree blocks not
 296  * referenced by their owner trees. The location of tree block is recorded
 297  * in the back refs. Actually the full back refs is generic, and can be
 298  * used in all cases the implicit back refs is used. The major shortcoming
 299  * of the full back refs is its overhead. Every time a tree block gets
 300  * COWed, we have to update back refs entry for all pointers in it.
 301  *
 302  * For a newly allocated tree block, we use implicit back refs for
 303  * pointers in it. This means most tree related operations only involve
 304  * implicit back refs. For a tree block created in old transaction, the
 305  * only way to drop a reference to it is COW it. So we can detect the
 306  * event that tree block loses its owner tree's reference and do the
 307  * back refs conversion.
 308  *
 309  * When a tree block is COWed through a tree, there are four cases:
 310  *
 311  * The reference count of the block is one and the tree is the block's
 312  * owner tree. Nothing to do in this case.
 313  *
 314  * The reference count of the block is one and the tree is not the
 315  * block's owner tree. In this case, full back refs is used for pointers
 316  * in the block. Remove these full back refs, add implicit back refs for
 317  * every pointers in the new block.
 318  *
 319  * The reference count of the block is greater than one and the tree is
 320  * the block's owner tree. In this case, implicit back refs is used for
 321  * pointers in the block. Add full back refs for every pointers in the
 322  * block, increase lower level extents' reference counts. The original
 323  * implicit back refs are entailed to the new block.
 324  *
 325  * The reference count of the block is greater than one and the tree is
 326  * not the block's owner tree. Add implicit back refs for every pointer in
 327  * the new block, increase lower level extents' reference count.
 328  *
 329  * Back Reference Key composing:
 330  *
 331  * The key objectid corresponds to the first byte in the extent,
 332  * The key type is used to differentiate between types of back refs.
 333  * There are different meanings of the key offset for different types
 334  * of back refs.
 335  *
 336  * File extents can be referenced by:
 337  *
 338  * - multiple snapshots, subvolumes, or different generations in one subvol
 339  * - different files inside a single subvolume
 340  * - different offsets inside a file (bookend extents in file.c)
 341  *
 342  * The extent ref structure for the implicit back refs has fields for:
 343  *
 344  * - Objectid of the subvolume root
 345  * - objectid of the file holding the reference
 346  * - original offset in the file
 347  * - how many bookend extents
 348  *
 349  * The key offset for the implicit back refs is hash of the first
 350  * three fields.
 351  *
 352  * The extent ref structure for the full back refs has field for:
 353  *
 354  * - number of pointers in the tree leaf
 355  *
 356  * The key offset for the implicit back refs is the first byte of
 357  * the tree leaf
 358  *
 359  * When a file extent is allocated, The implicit back refs is used.
 360  * the fields are filled in:
 361  *
 362  *     (root_key.objectid, inode objectid, offset in file, 1)
 363  *
 364  * When a file extent is removed file truncation, we find the
 365  * corresponding implicit back refs and check the following fields:
 366  *
 367  *     (btrfs_header_owner(leaf), inode objectid, offset in file)
 368  *
 369  * Btree extents can be referenced by:
 370  *
 371  * - Different subvolumes
 372  *
 373  * Both the implicit back refs and the full back refs for tree blocks
 374  * only consist of key. The key offset for the implicit back refs is
 375  * objectid of block's owner tree. The key offset for the full back refs
 376  * is the first byte of parent block.
 377  *
 378  * When implicit back refs is used, information about the lowest key and
 379  * level of the tree block are required. These information are stored in
 380  * tree block info structure.
 381  */
 382
 383 /*
 384  * is_data == BTRFS_REF_TYPE_BLOCK, tree block type is required,
 385  * is_data == BTRFS_REF_TYPE_DATA, data type is requiried,
 386  * is_data == BTRFS_REF_TYPE_ANY, either type is OK.
 387  */
 388 int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
 389                                      struct btrfs_extent_inline_ref *iref,
 390                                      enum btrfs_inline_ref_type is_data)
 391 {
 392         int type = btrfs_extent_inline_ref_type(eb, iref);
 393         u64 offset = btrfs_extent_inline_ref_offset(eb, iref);
 394
 395         if (type == BTRFS_TREE_BLOCK_REF_KEY ||
 396             type == BTRFS_SHARED_BLOCK_REF_KEY ||
 397             type == BTRFS_SHARED_DATA_REF_KEY ||
 398             type == BTRFS_EXTENT_DATA_REF_KEY) {
 399                 if (is_data == BTRFS_REF_TYPE_BLOCK) {
 400                         if (type == BTRFS_TREE_BLOCK_REF_KEY)
 401                                 return type;
 402                         if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
 403                                 ASSERT(eb->fs_info);
 404                                 /*
 405                                  * Every shared one has parent tree
 406                                  * block, which must be aligned to
 407                                  * nodesize.
 408                                  */
 409                                 if (offset &&
 410                                     IS_ALIGNED(offset, eb->fs_info->nodesize))
 411                                         return type;
 412                         }
 413                 } else if (is_data == BTRFS_REF_TYPE_DATA) {
 414                         if (type == BTRFS_EXTENT_DATA_REF_KEY)
 415                                 return type;
 416                         if (type == BTRFS_SHARED_DATA_REF_KEY) {
 417                                 ASSERT(eb->fs_info);
 418                                 /*
 419                                  * Every shared one has parent tree
 420                                  * block, which must be aligned to
 421                                  * nodesize.
 422                                  */
 423                                 if (offset &&
 424                                     IS_ALIGNED(offset, eb->fs_info->nodesize))
 425                                         return type;
 426                         }
 427                 } else {
 428                         ASSERT(is_data == BTRFS_REF_TYPE_ANY);
 429                         return type;
 430                 }
 431         }
 432
 433         btrfs_print_leaf((struct extent_buffer *)eb);
 434         btrfs_err(eb->fs_info, "eb %llu invalid extent inline ref type %d",
 435                   eb->start, type);
 436         WARN_ON(1);
 437
 438         return BTRFS_REF_TYPE_INVALID;
 439 }
 440
 441 u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset)
 442 {
 443         u32 high_crc = ~(u32)0;
 444         u32 low_crc = ~(u32)0;
 445         __le64 lenum;
 446
 447         lenum = cpu_to_le64(root_objectid);
 448         high_crc = btrfs_crc32c(high_crc, &lenum, sizeof(lenum));
 449         lenum = cpu_to_le64(owner);
 450         low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
 451         lenum = cpu_to_le64(offset);
 452         low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
 453
 454         return ((u64)high_crc << 31) ^ (u64)low_crc;
 455 }
 456
 457 static u64 hash_extent_data_ref_item(struct extent_buffer *leaf,
 458                                      struct btrfs_extent_data_ref *ref)
 459 {
 460         return hash_extent_data_ref(btrfs_extent_data_ref_root(leaf, ref),
 461                                     btrfs_extent_data_ref_objectid(leaf, ref),
 462                                     btrfs_extent_data_ref_offset(leaf, ref));
 463 }
 464
 465 static int match_extent_data_ref(struct extent_buffer *leaf,
 466                                  struct btrfs_extent_data_ref *ref,
 467                                  u64 root_objectid, u64 owner, u64 offset)
 468 {
 469         if (btrfs_extent_data_ref_root(leaf, ref) != root_objectid ||
 470             btrfs_extent_data_ref_objectid(leaf, ref) != owner ||
 471             btrfs_extent_data_ref_offset(leaf, ref) != offset)
 472                 return 0;
 473         return 1;
 474 }
 475
 476 static noinline int lookup_extent_data_ref(struct btrfs_trans_handle *trans,
 477                                            struct btrfs_path *path,
 478                                            u64 bytenr, u64 parent,
 479                                            u64 root_objectid,
 480                                            u64 owner, u64 offset)
 481 {
 482         struct btrfs_root *root = trans->fs_info->extent_root;
 483         struct btrfs_key key;
 484         struct btrfs_extent_data_ref *ref;
 485         struct extent_buffer *leaf;
 486         u32 nritems;
 487         int ret;
 488         int recow;
 489         int err = -ENOENT;
 490
 491         key.objectid = bytenr;
 492         if (parent) {
 493                 key.type = BTRFS_SHARED_DATA_REF_KEY;
 494                 key.offset = parent;
 495         } else {
 496                 key.type = BTRFS_EXTENT_DATA_REF_KEY;
 497                 key.offset = hash_extent_data_ref(root_objectid,
 498                                                   owner, offset);
 499         }
 500 again:
 501         recow = 0;
 502         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 503         if (ret < 0) {
 504                 err = ret;
 505                 goto fail;
 506         }
 507
 508         if (parent) {
 509                 if (!ret)
 510                         return 0;
 511                 goto fail;
 512         }
 513
 514         leaf = path->nodes[0];
 515         nritems = btrfs_header_nritems(leaf);
 516         while (1) {
 517                 if (path->slots[0] >= nritems) {
 518                         ret = btrfs_next_leaf(root, path);
 519                         if (ret < 0)
 520                                 err = ret;
 521                         if (ret)
 522                                 goto fail;
 523
 524                         leaf = path->nodes[0];
 525                         nritems = btrfs_header_nritems(leaf);
 526                         recow = 1;
 527                 }
 528
 529                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 530                 if (key.objectid != bytenr ||
 531                     key.type != BTRFS_EXTENT_DATA_REF_KEY)
 532                         goto fail;
 533
 534                 ref = btrfs_item_ptr(leaf, path->slots[0],
 535                                      struct btrfs_extent_data_ref);
 536
 537                 if (match_extent_data_ref(leaf, ref, root_objectid,
 538                                           owner, offset)) {
 539                         if (recow) {
 540                                 btrfs_release_path(path);
 541                                 goto again;
 542                         }
 543                         err = 0;
 544                         break;
 545                 }
 546                 path->slots[0]++;
 547         }
 548 fail:
 549         return err;
 550 }
 551
 552 static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans,
 553                                            struct btrfs_path *path,
 554                                            u64 bytenr, u64 parent,
 555                                            u64 root_objectid, u64 owner,
 556                                            u64 offset, int refs_to_add)
 557 {
 558         struct btrfs_root *root = trans->fs_info->extent_root;
 559         struct btrfs_key key;
 560         struct extent_buffer *leaf;
 561         u32 size;
 562         u32 num_refs;
 563         int ret;
 564
 565         key.objectid = bytenr;
 566         if (parent) {
 567                 key.type = BTRFS_SHARED_DATA_REF_KEY;
 568                 key.offset = parent;
 569                 size = sizeof(struct btrfs_shared_data_ref);
 570         } else {
 571                 key.type = BTRFS_EXTENT_DATA_REF_KEY;
 572                 key.offset = hash_extent_data_ref(root_objectid,
 573                                                   owner, offset);
 574                 size = sizeof(struct btrfs_extent_data_ref);
 575         }
 576
 577         ret = btrfs_insert_empty_item(trans, root, path, &key, size);
 578         if (ret && ret != -EEXIST)
 579                 goto fail;
 580
 581         leaf = path->nodes[0];
 582         if (parent) {
 583                 struct btrfs_shared_data_ref *ref;
 584                 ref = btrfs_item_ptr(leaf, path->slots[0],
 585                                      struct btrfs_shared_data_ref);
 586                 if (ret == 0) {
 587                         btrfs_set_shared_data_ref_count(leaf, ref, refs_to_add);
 588                 } else {
 589                         num_refs = btrfs_shared_data_ref_count(leaf, ref);
 590                         num_refs += refs_to_add;
 591                         btrfs_set_shared_data_ref_count(leaf, ref, num_refs);
 592                 }
 593         } else {
 594                 struct btrfs_extent_data_ref *ref;
 595                 while (ret == -EEXIST) {
 596                         ref = btrfs_item_ptr(leaf, path->slots[0],
 597                                              struct btrfs_extent_data_ref);
 598                         if (match_extent_data_ref(leaf, ref, root_objectid,
 599                                                   owner, offset))
 600                                 break;
 601                         btrfs_release_path(path);
 602                         key.offset++;
 603                         ret = btrfs_insert_empty_item(trans, root, path, &key,
 604                                                       size);
 605                         if (ret && ret != -EEXIST)
 606                                 goto fail;
 607
 608                         leaf = path->nodes[0];
 609                 }
 610                 ref = btrfs_item_ptr(leaf, path->slots[0],
 611                                      struct btrfs_extent_data_ref);
 612                 if (ret == 0) {
 613                         btrfs_set_extent_data_ref_root(leaf, ref,
 614                                                        root_objectid);
 615                         btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
 616                         btrfs_set_extent_data_ref_offset(leaf, ref, offset);
 617                         btrfs_set_extent_data_ref_count(leaf, ref, refs_to_add);
 618                 } else {
 619                         num_refs = btrfs_extent_data_ref_count(leaf, ref);
 620                         num_refs += refs_to_add;
 621                         btrfs_set_extent_data_ref_count(leaf, ref, num_refs);
 622                 }
 623         }
 624         btrfs_mark_buffer_dirty(leaf);
 625         ret = 0;
 626 fail:
 627         btrfs_release_path(path);
 628         return ret;
 629 }
 630
 631 static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
 632                                            struct btrfs_path *path,
 633                                            int refs_to_drop, int *last_ref)
 634 {
 635         struct btrfs_key key;
 636         struct btrfs_extent_data_ref *ref1 = NULL;
 637         struct btrfs_shared_data_ref *ref2 = NULL;
 638         struct extent_buffer *leaf;
 639         u32 num_refs = 0;
 640         int ret = 0;
 641
 642         leaf = path->nodes[0];
 643         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 644
 645         if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
 646                 ref1 = btrfs_item_ptr(leaf, path->slots[0],
 647                                       struct btrfs_extent_data_ref);
 648                 num_refs = btrfs_extent_data_ref_count(leaf, ref1);
 649         } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
 650                 ref2 = btrfs_item_ptr(leaf, path->slots[0],
 651                                       struct btrfs_shared_data_ref);
 652                 num_refs = btrfs_shared_data_ref_count(leaf, ref2);
 653         } else if (unlikely(key.type == BTRFS_EXTENT_REF_V0_KEY)) {
 654                 btrfs_print_v0_err(trans->fs_info);
 655                 btrfs_abort_transaction(trans, -EINVAL);
 656                 return -EINVAL;
 657         } else {
 658                 BUG();
 659         }
 660
 661         BUG_ON(num_refs < refs_to_drop);
 662         num_refs -= refs_to_drop;
 663
 664         if (num_refs == 0) {
 665                 ret = btrfs_del_item(trans, trans->fs_info->extent_root, path);
 666                 *last_ref = 1;
 667         } else {
 668                 if (key.type == BTRFS_EXTENT_DATA_REF_KEY)
 669                         btrfs_set_extent_data_ref_count(leaf, ref1, num_refs);
 670                 else if (key.type == BTRFS_SHARED_DATA_REF_KEY)
 671                         btrfs_set_shared_data_ref_count(leaf, ref2, num_refs);
 672                 btrfs_mark_buffer_dirty(leaf);
 673         }
 674         return ret;
 675 }
 676
 677 static noinline u32 extent_data_ref_count(struct btrfs_path *path,
 678                                           struct btrfs_extent_inline_ref *iref)
 679 {
 680         struct btrfs_key key;
 681         struct extent_buffer *leaf;
 682         struct btrfs_extent_data_ref *ref1;
 683         struct btrfs_shared_data_ref *ref2;
 684         u32 num_refs = 0;
 685         int type;
 686
 687         leaf = path->nodes[0];
 688         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 689
 690         BUG_ON(key.type == BTRFS_EXTENT_REF_V0_KEY);
 691         if (iref) {
 692                 /*
 693                  * If type is invalid, we should have bailed out earlier than
 694                  * this call.
 695                  */
 696                 type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
 697                 ASSERT(type != BTRFS_REF_TYPE_INVALID);
 698                 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
 699                         ref1 = (struct btrfs_extent_data_ref *)(&iref->offset);
 700                         num_refs = btrfs_extent_data_ref_count(leaf, ref1);
 701                 } else {
 702                         ref2 = (struct btrfs_shared_data_ref *)(iref + 1);
 703                         num_refs = btrfs_shared_data_ref_count(leaf, ref2);
 704                 }
 705         } else if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
 706                 ref1 = btrfs_item_ptr(leaf, path->slots[0],
 707                                       struct btrfs_extent_data_ref);
 708                 num_refs = btrfs_extent_data_ref_count(leaf, ref1);
 709         } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
 710                 ref2 = btrfs_item_ptr(leaf, path->slots[0],
 711                                       struct btrfs_shared_data_ref);
 712                 num_refs = btrfs_shared_data_ref_count(leaf, ref2);
 713         } else {
 714                 WARN_ON(1);
 715         }
 716         return num_refs;
 717 }
 718
 719 static noinline int lookup_tree_block_ref(struct btrfs_trans_handle *trans,
 720                                           struct btrfs_path *path,
 721                                           u64 bytenr, u64 parent,
 722                                           u64 root_objectid)
 723 {
 724         struct btrfs_root *root = trans->fs_info->extent_root;
 725         struct btrfs_key key;
 726         int ret;
 727
 728         key.objectid = bytenr;
 729         if (parent) {
 730                 key.type = BTRFS_SHARED_BLOCK_REF_KEY;
 731                 key.offset = parent;
 732         } else {
 733                 key.type = BTRFS_TREE_BLOCK_REF_KEY;
 734                 key.offset = root_objectid;
 735         }
 736
 737         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 738         if (ret > 0)
 739                 ret = -ENOENT;
 740         return ret;
 741 }
 742
 743 static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans,
 744                                           struct btrfs_path *path,
 745                                           u64 bytenr, u64 parent,
 746                                           u64 root_objectid)
 747 {
 748         struct btrfs_key key;
 749         int ret;
 750
 751         key.objectid = bytenr;
 752         if (parent) {
 753                 key.type = BTRFS_SHARED_BLOCK_REF_KEY;
 754                 key.offset = parent;
 755         } else {
 756                 key.type = BTRFS_TREE_BLOCK_REF_KEY;
 757                 key.offset = root_objectid;
 758         }
 759
 760         ret = btrfs_insert_empty_item(trans, trans->fs_info->extent_root,
 761                                       path, &key, 0);
 762         btrfs_release_path(path);
 763         return ret;
 764 }
 765
 766 static inline int extent_ref_type(u64 parent, u64 owner)
 767 {
 768         int type;
 769         if (owner < BTRFS_FIRST_FREE_OBJECTID) {
 770                 if (parent > 0)
 771                         type = BTRFS_SHARED_BLOCK_REF_KEY;
 772                 else
 773                         type = BTRFS_TREE_BLOCK_REF_KEY;
 774         } else {
 775                 if (parent > 0)
 776                         type = BTRFS_SHARED_DATA_REF_KEY;
 777                 else
 778                         type = BTRFS_EXTENT_DATA_REF_KEY;
 779         }
 780         return type;
 781 }
 782
 783 static int find_next_key(struct btrfs_path *path, int level,
 784                          struct btrfs_key *key)
 785
 786 {
 787         for (; level < BTRFS_MAX_LEVEL; level++) {
 788                 if (!path->nodes[level])
 789                         break;
 790                 if (path->slots[level] + 1 >=
 791                     btrfs_header_nritems(path->nodes[level]))
 792                         continue;
 793                 if (level == 0)
 794                         btrfs_item_key_to_cpu(path->nodes[level], key,
 795                                               path->slots[level] + 1);
 796                 else
 797                         btrfs_node_key_to_cpu(path->nodes[level], key,
 798                                               path->slots[level] + 1);
 799                 return 0;
 800         }
 801         return 1;
 802 }
 803
 804 /*
 805  * look for inline back ref. if back ref is found, *ref_ret is set
 806  * to the address of inline back ref, and 0 is returned.
 807  *
 808  * if back ref isn't found, *ref_ret is set to the address where it
 809  * should be inserted, and -ENOENT is returned.
 810  *
 811  * if insert is true and there are too many inline back refs, the path
 812  * points to the extent item, and -EAGAIN is returned.
 813  *
 814  * NOTE: inline back refs are ordered in the same way that back ref
 815  *       items in the tree are ordered.
 816  */
 817 static noinline_for_stack
 818 int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
 819                                  struct btrfs_path *path,
 820                                  struct btrfs_extent_inline_ref **ref_ret,
 821                                  u64 bytenr, u64 num_bytes,
 822                                  u64 parent, u64 root_objectid,
 823                                  u64 owner, u64 offset, int insert)
 824 {
 825         struct btrfs_fs_info *fs_info = trans->fs_info;
 826         struct btrfs_root *root = fs_info->extent_root;
 827         struct btrfs_key key;
 828         struct extent_buffer *leaf;
 829         struct btrfs_extent_item *ei;
 830         struct btrfs_extent_inline_ref *iref;
 831         u64 flags;
 832         u64 item_size;
 833         unsigned long ptr;
 834         unsigned long end;
 835         int extra_size;
 836         int type;
 837         int want;
 838         int ret;
 839         int err = 0;
 840         bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
 841         int needed;
 842
 843         key.objectid = bytenr;
 844         key.type = BTRFS_EXTENT_ITEM_KEY;
 845         key.offset = num_bytes;
 846
 847         want = extent_ref_type(parent, owner);
 848         if (insert) {
 849                 extra_size = btrfs_extent_inline_ref_size(want);
 850                 path->keep_locks = 1;
 851         } else
 852                 extra_size = -1;
 853
 854         /*
 855          * Owner is our level, so we can just add one to get the level for the
 856          * block we are interested in.
 857          */
 858         if (skinny_metadata && owner < BTRFS_FIRST_FREE_OBJECTID) {
 859                 key.type = BTRFS_METADATA_ITEM_KEY;
 860                 key.offset = owner;
 861         }
 862
 863 again:
 864         ret = btrfs_search_slot(trans, root, &key, path, extra_size, 1);
 865         if (ret < 0) {
 866                 err = ret;
 867                 goto out;
 868         }
 869
 870         /*
 871          * We may be a newly converted file system which still has the old fat
 872          * extent entries for metadata, so try and see if we have one of those.
 873          */
 874         if (ret > 0 && skinny_metadata) {
 875                 skinny_metadata = false;
 876                 if (path->slots[0]) {
 877                         path->slots[0]--;
 878                         btrfs_item_key_to_cpu(path->nodes[0], &key,
 879                                               path->slots[0]);
 880                         if (key.objectid == bytenr &&
 881                             key.type == BTRFS_EXTENT_ITEM_KEY &&
 882                             key.offset == num_bytes)
 883                                 ret = 0;
 884                 }
 885                 if (ret) {
 886                         key.objectid = bytenr;
 887                         key.type = BTRFS_EXTENT_ITEM_KEY;
 888                         key.offset = num_bytes;
 889                         btrfs_release_path(path);
 890                         goto again;
 891                 }
 892         }
 893
 894         if (ret && !insert) {
 895                 err = -ENOENT;
 896                 goto out;
 897         } else if (WARN_ON(ret)) {
 898                 err = -EIO;
 899                 goto out;
 900         }
 901
 902         leaf = path->nodes[0];
 903         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
 904         if (unlikely(item_size < sizeof(*ei))) {
 905                 err = -EINVAL;
 906                 btrfs_print_v0_err(fs_info);
 907                 btrfs_abort_transaction(trans, err);
 908                 goto out;
 909         }
 910
 911         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
 912         flags = btrfs_extent_flags(leaf, ei);
 913
 914         ptr = (unsigned long)(ei + 1);
 915         end = (unsigned long)ei + item_size;
 916
 917         if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK && !skinny_metadata) {
 918                 ptr += sizeof(struct btrfs_tree_block_info);
 919                 BUG_ON(ptr > end);
 920         }
 921
 922         if (owner >= BTRFS_FIRST_FREE_OBJECTID)
 923                 needed = BTRFS_REF_TYPE_DATA;
 924         else
 925                 needed = BTRFS_REF_TYPE_BLOCK;
 926
 927         err = -ENOENT;
 928         while (1) {
 929                 if (ptr >= end) {
 930                         WARN_ON(ptr > end);
 931                         break;
 932                 }
 933                 iref = (struct btrfs_extent_inline_ref *)ptr;
 934                 type = btrfs_get_extent_inline_ref_type(leaf, iref, needed);
 935                 if (type == BTRFS_REF_TYPE_INVALID) {
 936                         err = -EUCLEAN;
 937                         goto out;
 938                 }
 939
 940                 if (want < type)
 941                         break;
 942                 if (want > type) {
 943                         ptr += btrfs_extent_inline_ref_size(type);
 944                         continue;
 945                 }
 946
 947                 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
 948                         struct btrfs_extent_data_ref *dref;
 949                         dref = (struct btrfs_extent_data_ref *)(&iref->offset);
 950                         if (match_extent_data_ref(leaf, dref, root_objectid,
 951                                                   owner, offset)) {
 952                                 err = 0;
 953                                 break;
 954                         }
 955                         if (hash_extent_data_ref_item(leaf, dref) <
 956                             hash_extent_data_ref(root_objectid, owner, offset))
 957                                 break;
 958                 } else {
 959                         u64 ref_offset;
 960                         ref_offset = btrfs_extent_inline_ref_offset(leaf, iref);
 961                         if (parent > 0) {
 962                                 if (parent == ref_offset) {
 963                                         err = 0;
 964                                         break;
 965                                 }
 966                                 if (ref_offset < parent)
 967                                         break;
 968                         } else {
 969                                 if (root_objectid == ref_offset) {
 970                                         err = 0;
 971                                         break;
 972                                 }
 973                                 if (ref_offset < root_objectid)
 974                                         break;
 975                         }
 976                 }
 977                 ptr += btrfs_extent_inline_ref_size(type);
 978         }
 979         if (err == -ENOENT && insert) {
 980                 if (item_size + extra_size >=
 981                     BTRFS_MAX_EXTENT_ITEM_SIZE(root)) {
 982                         err = -EAGAIN;
 983                         goto out;
 984                 }
 985                 /*
 986                  * To add new inline back ref, we have to make sure
 987                  * there is no corresponding back ref item.
 988                  * For simplicity, we just do not add new inline back
 989                  * ref if there is any kind of item for this block
 990                  */
 991                 if (find_next_key(path, 0, &key) == 0 &&
 992                     key.objectid == bytenr &&
 993                     key.type < BTRFS_BLOCK_GROUP_ITEM_KEY) {
 994                         err = -EAGAIN;
 995                         goto out;
 996                 }
 997         }
 998         *ref_ret = (struct btrfs_extent_inline_ref *)ptr;
 999 out:
1000         if (insert) {
1001                 path->keep_locks = 0;
1002                 btrfs_unlock_up_safe(path, 1);
1003         }
1004         return err;
1005 }
1006
1007 /*
1008  * helper to add new inline back ref
1009  */
1010 static noinline_for_stack
1011 void setup_inline_extent_backref(struct btrfs_fs_info *fs_info,
1012                                  struct btrfs_path *path,
1013                                  struct btrfs_extent_inline_ref *iref,
1014                                  u64 parent, u64 root_objectid,
1015                                  u64 owner, u64 offset, int refs_to_add,
1016                                  struct btrfs_delayed_extent_op *extent_op)
1017 {
1018         struct extent_buffer *leaf;
1019         struct btrfs_extent_item *ei;
1020         unsigned long ptr;
1021         unsigned long end;
1022         unsigned long item_offset;
1023         u64 refs;
1024         int size;
1025         int type;
1026
1027         leaf = path->nodes[0];
1028         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1029         item_offset = (unsigned long)iref - (unsigned long)ei;
1030
1031         type = extent_ref_type(parent, owner);
1032         size = btrfs_extent_inline_ref_size(type);
1033
1034         btrfs_extend_item(path, size);
1035
1036         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1037         refs = btrfs_extent_refs(leaf, ei);
1038         refs += refs_to_add;
1039         btrfs_set_extent_refs(leaf, ei, refs);
1040         if (extent_op)
1041                 __run_delayed_extent_op(extent_op, leaf, ei);
1042
1043         ptr = (unsigned long)ei + item_offset;
1044         end = (unsigned long)ei + btrfs_item_size_nr(leaf, path->slots[0]);
1045         if (ptr < end - size)
1046                 memmove_extent_buffer(leaf, ptr + size, ptr,
1047                                       end - size - ptr);
1048
1049         iref = (struct btrfs_extent_inline_ref *)ptr;
1050         btrfs_set_extent_inline_ref_type(leaf, iref, type);
1051         if (type == BTRFS_EXTENT_DATA_REF_KEY) {
1052                 struct btrfs_extent_data_ref *dref;
1053                 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
1054                 btrfs_set_extent_data_ref_root(leaf, dref, root_objectid);
1055                 btrfs_set_extent_data_ref_objectid(leaf, dref, owner);
1056                 btrfs_set_extent_data_ref_offset(leaf, dref, offset);
1057                 btrfs_set_extent_data_ref_count(leaf, dref, refs_to_add);
1058         } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
1059                 struct btrfs_shared_data_ref *sref;
1060                 sref = (struct btrfs_shared_data_ref *)(iref + 1);
1061                 btrfs_set_shared_data_ref_count(leaf, sref, refs_to_add);
1062                 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
1063         } else if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
1064                 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
1065         } else {
1066                 btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid);
1067         }
1068         btrfs_mark_buffer_dirty(leaf);
1069 }
1070
1071 static int lookup_extent_backref(struct btrfs_trans_handle *trans,
1072                                  struct btrfs_path *path,
1073                                  struct btrfs_extent_inline_ref **ref_ret,
1074                                  u64 bytenr, u64 num_bytes, u64 parent,
1075                                  u64 root_objectid, u64 owner, u64 offset)
1076 {
1077         int ret;
1078
1079         ret = lookup_inline_extent_backref(trans, path, ref_ret, bytenr,
1080                                            num_bytes, parent, root_objectid,
1081                                            owner, offset, 0);
1082         if (ret != -ENOENT)
1083                 return ret;
1084
1085         btrfs_release_path(path);
1086         *ref_ret = NULL;
1087
1088         if (owner < BTRFS_FIRST_FREE_OBJECTID) {
1089                 ret = lookup_tree_block_ref(trans, path, bytenr, parent,
1090                                             root_objectid);
1091         } else {
1092                 ret = lookup_extent_data_ref(trans, path, bytenr, parent,
1093                                              root_objectid, owner, offset);
1094         }
1095         return ret;
1096 }
1097
1098 /*
1099  * helper to update/remove inline back ref
1100  */
1101 static noinline_for_stack
1102 void update_inline_extent_backref(struct btrfs_path *path,
1103                                   struct btrfs_extent_inline_ref *iref,
1104                                   int refs_to_mod,
1105                                   struct btrfs_delayed_extent_op *extent_op,
1106                                   int *last_ref)
1107 {
1108         struct extent_buffer *leaf = path->nodes[0];
1109         struct btrfs_extent_item *ei;
1110         struct btrfs_extent_data_ref *dref = NULL;
1111         struct btrfs_shared_data_ref *sref = NULL;
1112         unsigned long ptr;
1113         unsigned long end;
1114         u32 item_size;
1115         int size;
1116         int type;
1117         u64 refs;
1118
1119         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1120         refs = btrfs_extent_refs(leaf, ei);
1121         WARN_ON(refs_to_mod < 0 && refs + refs_to_mod <= 0);
1122         refs += refs_to_mod;
1123         btrfs_set_extent_refs(leaf, ei, refs);
1124         if (extent_op)
1125                 __run_delayed_extent_op(extent_op, leaf, ei);
1126
1127         /*
1128          * If type is invalid, we should have bailed out after
1129          * lookup_inline_extent_backref().
1130          */
1131         type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_ANY);
1132         ASSERT(type != BTRFS_REF_TYPE_INVALID);
1133
1134         if (type == BTRFS_EXTENT_DATA_REF_KEY) {
1135                 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
1136                 refs = btrfs_extent_data_ref_count(leaf, dref);
1137         } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
1138                 sref = (struct btrfs_shared_data_ref *)(iref + 1);
1139                 refs = btrfs_shared_data_ref_count(leaf, sref);
1140         } else {
1141                 refs = 1;
1142                 BUG_ON(refs_to_mod != -1);
1143         }
1144
1145         BUG_ON(refs_to_mod < 0 && refs < -refs_to_mod);
1146         refs += refs_to_mod;
1147
1148         if (refs > 0) {
1149                 if (type == BTRFS_EXTENT_DATA_REF_KEY)
1150                         btrfs_set_extent_data_ref_count(leaf, dref, refs);
1151                 else
1152                         btrfs_set_shared_data_ref_count(leaf, sref, refs);
1153         } else {
1154                 *last_ref = 1;
1155                 size =  btrfs_extent_inline_ref_size(type);
1156                 item_size = btrfs_item_size_nr(leaf, path->slots[0]);
1157                 ptr = (unsigned long)iref;
1158                 end = (unsigned long)ei + item_size;
1159                 if (ptr + size < end)
1160                         memmove_extent_buffer(leaf, ptr, ptr + size,
1161                                               end - ptr - size);
1162                 item_size -= size;
1163                 btrfs_truncate_item(path, item_size, 1);
1164         }
1165         btrfs_mark_buffer_dirty(leaf);
1166 }
1167
1168 static noinline_for_stack
1169 int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
1170                                  struct btrfs_path *path,
1171                                  u64 bytenr, u64 num_bytes, u64 parent,
1172                                  u64 root_objectid, u64 owner,
1173                                  u64 offset, int refs_to_add,
1174                                  struct btrfs_delayed_extent_op *extent_op)
1175 {
1176         struct btrfs_extent_inline_ref *iref;
1177         int ret;
1178
1179         ret = lookup_inline_extent_backref(trans, path, &iref, bytenr,
1180                                            num_bytes, parent, root_objectid,
1181                                            owner, offset, 1);
1182         if (ret == 0) {
1183                 BUG_ON(owner < BTRFS_FIRST_FREE_OBJECTID);
1184                 update_inline_extent_backref(path, iref, refs_to_add,
1185                                              extent_op, NULL);
1186         } else if (ret == -ENOENT) {
1187                 setup_inline_extent_backref(trans->fs_info, path, iref, parent,
1188                                             root_objectid, owner, offset,
1189                                             refs_to_add, extent_op);
1190                 ret = 0;
1191         }
1192         return ret;
1193 }
1194
1195 static int insert_extent_backref(struct btrfs_trans_handle *trans,
1196                                  struct btrfs_path *path,
1197                                  u64 bytenr, u64 parent, u64 root_objectid,
1198                                  u64 owner, u64 offset, int refs_to_add)
1199 {
1200         int ret;
1201         if (owner < BTRFS_FIRST_FREE_OBJECTID) {
1202                 BUG_ON(refs_to_add != 1);
1203                 ret = insert_tree_block_ref(trans, path, bytenr, parent,
1204                                             root_objectid);
1205         } else {
1206                 ret = insert_extent_data_ref(trans, path, bytenr, parent,
1207                                              root_objectid, owner, offset,
1208                                              refs_to_add);
1209         }
1210         return ret;
1211 }
1212
1213 static int remove_extent_backref(struct btrfs_trans_handle *trans,
1214                                  struct btrfs_path *path,
1215                                  struct btrfs_extent_inline_ref *iref,
1216                                  int refs_to_drop, int is_data, int *last_ref)
1217 {
1218         int ret = 0;
1219
1220         BUG_ON(!is_data && refs_to_drop != 1);
1221         if (iref) {
1222                 update_inline_extent_backref(path, iref, -refs_to_drop, NULL,
1223                                              last_ref);
1224         } else if (is_data) {
1225                 ret = remove_extent_data_ref(trans, path, refs_to_drop,
1226                                              last_ref);
1227         } else {
1228                 *last_ref = 1;
1229                 ret = btrfs_del_item(trans, trans->fs_info->extent_root, path);
1230         }
1231         return ret;
1232 }
1233
1234 static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len,
1235                                u64 *discarded_bytes)
1236 {
1237         int j, ret = 0;
1238         u64 bytes_left, end;
1239         u64 aligned_start = ALIGN(start, 1 << 9);
1240
1241         if (WARN_ON(start != aligned_start)) {
1242                 len -= aligned_start - start;
1243                 len = round_down(len, 1 << 9);
1244                 start = aligned_start;
1245         }
1246
1247         *discarded_bytes = 0;
1248
1249         if (!len)
1250                 return 0;
1251
1252         end = start + len;
1253         bytes_left = len;
1254
1255         /* Skip any superblocks on this device. */
1256         for (j = 0; j < BTRFS_SUPER_MIRROR_MAX; j++) {
1257                 u64 sb_start = btrfs_sb_offset(j);
1258                 u64 sb_end = sb_start + BTRFS_SUPER_INFO_SIZE;
1259                 u64 size = sb_start - start;
1260
1261                 if (!in_range(sb_start, start, bytes_left) &&
1262                     !in_range(sb_end, start, bytes_left) &&
1263                     !in_range(start, sb_start, BTRFS_SUPER_INFO_SIZE))
1264                         continue;
1265
1266                 /*
1267                  * Superblock spans beginning of range.  Adjust start and
1268                  * try again.
1269                  */
1270                 if (sb_start <= start) {
1271                         start += sb_end - start;
1272                         if (start > end) {
1273                                 bytes_left = 0;
1274                                 break;
1275                         }
1276                         bytes_left = end - start;
1277                         continue;
1278                 }
1279
1280                 if (size) {
1281                         ret = blkdev_issue_discard(bdev, start >> 9, size >> 9,
1282                                                    GFP_NOFS, 0);
1283                         if (!ret)
1284                                 *discarded_bytes += size;
1285                         else if (ret != -EOPNOTSUPP)
1286                                 return ret;
1287                 }
1288
1289                 start = sb_end;
1290                 if (start > end) {
1291                         bytes_left = 0;
1292                         break;
1293                 }
1294                 bytes_left = end - start;
1295         }
1296
1297         if (bytes_left) {
1298                 ret = blkdev_issue_discard(bdev, start >> 9, bytes_left >> 9,
1299                                            GFP_NOFS, 0);
1300                 if (!ret)
1301                         *discarded_bytes += bytes_left;
1302         }
1303         return ret;
1304 }
1305
1306 int btrfs_discard_extent(struct btrfs_fs_info *fs_info, u64 bytenr,
1307                          u64 num_bytes, u64 *actual_bytes)
1308 {
1309         int ret;
1310         u64 discarded_bytes = 0;
1311         struct btrfs_bio *bbio = NULL;
1312
1313
1314         /*
1315          * Avoid races with device replace and make sure our bbio has devices
1316          * associated to its stripes that don't go away while we are discarding.
1317          */
1318         btrfs_bio_counter_inc_blocked(fs_info);
1319         /* Tell the block device(s) that the sectors can be discarded */
1320         ret = btrfs_map_block(fs_info, BTRFS_MAP_DISCARD, bytenr, &num_bytes,
1321                               &bbio, 0);
1322         /* Error condition is -ENOMEM */
1323         if (!ret) {
1324                 struct btrfs_bio_stripe *stripe = bbio->stripes;
1325                 int i;
1326
1327
1328                 for (i = 0; i < bbio->num_stripes; i++, stripe++) {
1329                         u64 bytes;
1330                         struct request_queue *req_q;
1331
1332                         if (!stripe->dev->bdev) {
1333                                 ASSERT(btrfs_test_opt(fs_info, DEGRADED));
1334                                 continue;
1335                         }
1336                         req_q = bdev_get_queue(stripe->dev->bdev);
1337                         if (!blk_queue_discard(req_q))
1338                                 continue;
1339
1340                         ret = btrfs_issue_discard(stripe->dev->bdev,
1341                                                   stripe->physical,
1342                                                   stripe->length,
1343                                                   &bytes);
1344                         if (!ret)
1345                                 discarded_bytes += bytes;
1346                         else if (ret != -EOPNOTSUPP)
1347                                 break; /* Logic errors or -ENOMEM, or -EIO but I don't know how that could happen JDM */
1348
1349                         /*
1350                          * Just in case we get back EOPNOTSUPP for some reason,
1351                          * just ignore the return value so we don't screw up
1352                          * people calling discard_extent.
1353                          */
1354                         ret = 0;
1355                 }
1356                 btrfs_put_bbio(bbio);
1357         }
1358         btrfs_bio_counter_dec(fs_info);
1359
1360         if (actual_bytes)
1361                 *actual_bytes = discarded_bytes;
1362
1363
1364         if (ret == -EOPNOTSUPP)
1365                 ret = 0;
1366         return ret;
1367 }
1368
1369 /* Can return -ENOMEM */
1370 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1371                          struct btrfs_ref *generic_ref)
1372 {
1373         struct btrfs_fs_info *fs_info = trans->fs_info;
1374         int old_ref_mod, new_ref_mod;
1375         int ret;
1376
1377         ASSERT(generic_ref->type != BTRFS_REF_NOT_SET &&
1378                generic_ref->action);
1379         BUG_ON(generic_ref->type == BTRFS_REF_METADATA &&
1380                generic_ref->tree_ref.root == BTRFS_TREE_LOG_OBJECTID);
1381
1382         if (generic_ref->type == BTRFS_REF_METADATA)
1383                 ret = btrfs_add_delayed_tree_ref(trans, generic_ref,
1384                                 NULL, &old_ref_mod, &new_ref_mod);
1385         else
1386                 ret = btrfs_add_delayed_data_ref(trans, generic_ref, 0,
1387                                                  &old_ref_mod, &new_ref_mod);
1388
1389         btrfs_ref_tree_mod(fs_info, generic_ref);
1390
1391         if (ret == 0 && old_ref_mod < 0 && new_ref_mod >= 0)
1392                 sub_pinned_bytes(fs_info, generic_ref);
1393
1394         return ret;
1395 }
1396
1397 /*
1398  * __btrfs_inc_extent_ref - insert backreference for a given extent
1399  *
1400  * @trans:          Handle of transaction
1401  *
1402  * @node:           The delayed ref node used to get the bytenr/length for
1403  *                  extent whose references are incremented.
1404  *
1405  * @parent:         If this is a shared extent (BTRFS_SHARED_DATA_REF_KEY/
1406  *                  BTRFS_SHARED_BLOCK_REF_KEY) then it holds the logical
1407  *                  bytenr of the parent block. Since new extents are always
1408  *                  created with indirect references, this will only be the case
1409  *                  when relocating a shared extent. In that case, root_objectid
1410  *                  will be BTRFS_TREE_RELOC_OBJECTID. Otheriwse, parent must
1411  *                  be 0
1412  *
1413  * @root_objectid:  The id of the root where this modification has originated,
1414  *                  this can be either one of the well-known metadata trees or
1415  *                  the subvolume id which references this extent.
1416  *
1417  * @owner:          For data extents it is the inode number of the owning file.
1418  *                  For metadata extents this parameter holds the level in the
1419  *                  tree of the extent.
1420  *
1421  * @offset:         For metadata extents the offset is ignored and is currently
1422  *                  always passed as 0. For data extents it is the fileoffset
1423  *                  this extent belongs to.
1424  *
1425  * @refs_to_add     Number of references to add
1426  *
1427  * @extent_op       Pointer to a structure, holding information necessary when
1428  *                  updating a tree block's flags
1429  *
1430  */
1431 static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1432                                   struct btrfs_delayed_ref_node *node,
1433                                   u64 parent, u64 root_objectid,
1434                                   u64 owner, u64 offset, int refs_to_add,
1435                                   struct btrfs_delayed_extent_op *extent_op)
1436 {
1437         struct btrfs_path *path;
1438         struct extent_buffer *leaf;
1439         struct btrfs_extent_item *item;
1440         struct btrfs_key key;
1441         u64 bytenr = node->bytenr;
1442         u64 num_bytes = node->num_bytes;
1443         u64 refs;
1444         int ret;
1445
1446         path = btrfs_alloc_path();
1447         if (!path)
1448                 return -ENOMEM;
1449
1450         path->reada = READA_FORWARD;
1451         path->leave_spinning = 1;
1452         /* this will setup the path even if it fails to insert the back ref */
1453         ret = insert_inline_extent_backref(trans, path, bytenr, num_bytes,
1454                                            parent, root_objectid, owner,
1455                                            offset, refs_to_add, extent_op);
1456         if ((ret < 0 && ret != -EAGAIN) || !ret)
1457                 goto out;
1458
1459         /*
1460          * Ok we had -EAGAIN which means we didn't have space to insert and
1461          * inline extent ref, so just update the reference count and add a
1462          * normal backref.
1463          */
1464         leaf = path->nodes[0];
1465         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1466         item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1467         refs = btrfs_extent_refs(leaf, item);
1468         btrfs_set_extent_refs(leaf, item, refs + refs_to_add);
1469         if (extent_op)
1470                 __run_delayed_extent_op(extent_op, leaf, item);
1471
1472         btrfs_mark_buffer_dirty(leaf);
1473         btrfs_release_path(path);
1474
1475         path->reada = READA_FORWARD;
1476         path->leave_spinning = 1;
1477         /* now insert the actual backref */
1478         ret = insert_extent_backref(trans, path, bytenr, parent, root_objectid,
1479                                     owner, offset, refs_to_add);
1480         if (ret)
1481                 btrfs_abort_transaction(trans, ret);
1482 out:
1483         btrfs_free_path(path);
1484         return ret;
1485 }
1486
1487 static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
1488                                 struct btrfs_delayed_ref_node *node,
1489                                 struct btrfs_delayed_extent_op *extent_op,
1490                                 int insert_reserved)
1491 {
1492         int ret = 0;
1493         struct btrfs_delayed_data_ref *ref;
1494         struct btrfs_key ins;
1495         u64 parent = 0;
1496         u64 ref_root = 0;
1497         u64 flags = 0;
1498
1499         ins.objectid = node->bytenr;
1500         ins.offset = node->num_bytes;
1501         ins.type = BTRFS_EXTENT_ITEM_KEY;
1502
1503         ref = btrfs_delayed_node_to_data_ref(node);
1504         trace_run_delayed_data_ref(trans->fs_info, node, ref, node->action);
1505
1506         if (node->type == BTRFS_SHARED_DATA_REF_KEY)
1507                 parent = ref->parent;
1508         ref_root = ref->root;
1509
1510         if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
1511                 if (extent_op)
1512                         flags |= extent_op->flags_to_set;
1513                 ret = alloc_reserved_file_extent(trans, parent, ref_root,
1514                                                  flags, ref->objectid,
1515                                                  ref->offset, &ins,
1516                                                  node->ref_mod);
1517         } else if (node->action == BTRFS_ADD_DELAYED_REF) {
1518                 ret = __btrfs_inc_extent_ref(trans, node, parent, ref_root,
1519                                              ref->objectid, ref->offset,
1520                                              node->ref_mod, extent_op);
1521         } else if (node->action == BTRFS_DROP_DELAYED_REF) {
1522                 ret = __btrfs_free_extent(trans, node, parent,
1523                                           ref_root, ref->objectid,
1524                                           ref->offset, node->ref_mod,
1525                                           extent_op);
1526         } else {
1527                 BUG();
1528         }
1529         return ret;
1530 }
1531
1532 static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
1533                                     struct extent_buffer *leaf,
1534                                     struct btrfs_extent_item *ei)
1535 {
1536         u64 flags = btrfs_extent_flags(leaf, ei);
1537         if (extent_op->update_flags) {
1538                 flags |= extent_op->flags_to_set;
1539                 btrfs_set_extent_flags(leaf, ei, flags);
1540         }
1541
1542         if (extent_op->update_key) {
1543                 struct btrfs_tree_block_info *bi;
1544                 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_TREE_BLOCK));
1545                 bi = (struct btrfs_tree_block_info *)(ei + 1);
1546                 btrfs_set_tree_block_key(leaf, bi, &extent_op->key);
1547         }
1548 }
1549
1550 static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
1551                                  struct btrfs_delayed_ref_head *head,
1552                                  struct btrfs_delayed_extent_op *extent_op)
1553 {
1554         struct btrfs_fs_info *fs_info = trans->fs_info;
1555         struct btrfs_key key;
1556         struct btrfs_path *path;
1557         struct btrfs_extent_item *ei;
1558         struct extent_buffer *leaf;
1559         u32 item_size;
1560         int ret;
1561         int err = 0;
1562         int metadata = !extent_op->is_data;
1563
1564         if (trans->aborted)
1565                 return 0;
1566
1567         if (metadata && !btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1568                 metadata = 0;
1569
1570         path = btrfs_alloc_path();
1571         if (!path)
1572                 return -ENOMEM;
1573
1574         key.objectid = head->bytenr;
1575
1576         if (metadata) {
1577                 key.type = BTRFS_METADATA_ITEM_KEY;
1578                 key.offset = extent_op->level;
1579         } else {
1580                 key.type = BTRFS_EXTENT_ITEM_KEY;
1581                 key.offset = head->num_bytes;
1582         }
1583
1584 again:
1585         path->reada = READA_FORWARD;
1586         path->leave_spinning = 1;
1587         ret = btrfs_search_slot(trans, fs_info->extent_root, &key, path, 0, 1);
1588         if (ret < 0) {
1589                 err = ret;
1590                 goto out;
1591         }
1592         if (ret > 0) {
1593                 if (metadata) {
1594                         if (path->slots[0] > 0) {
1595                                 path->slots[0]--;
1596                                 btrfs_item_key_to_cpu(path->nodes[0], &key,
1597                                                       path->slots[0]);
1598                                 if (key.objectid == head->bytenr &&
1599                                     key.type == BTRFS_EXTENT_ITEM_KEY &&
1600                                     key.offset == head->num_bytes)
1601                                         ret = 0;
1602                         }
1603                         if (ret > 0) {
1604                                 btrfs_release_path(path);
1605                                 metadata = 0;
1606
1607                                 key.objectid = head->bytenr;
1608                                 key.offset = head->num_bytes;
1609                                 key.type = BTRFS_EXTENT_ITEM_KEY;
1610                                 goto again;
1611                         }
1612                 } else {
1613                         err = -EIO;
1614                         goto out;
1615                 }
1616         }
1617
1618         leaf = path->nodes[0];
1619         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
1620
1621         if (unlikely(item_size < sizeof(*ei))) {
1622                 err = -EINVAL;
1623                 btrfs_print_v0_err(fs_info);
1624                 btrfs_abort_transaction(trans, err);
1625                 goto out;
1626         }
1627
1628         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1629         __run_delayed_extent_op(extent_op, leaf, ei);
1630
1631         btrfs_mark_buffer_dirty(leaf);
1632 out:
1633         btrfs_free_path(path);
1634         return err;
1635 }
1636
1637 static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
1638                                 struct btrfs_delayed_ref_node *node,
1639                                 struct btrfs_delayed_extent_op *extent_op,
1640                                 int insert_reserved)
1641 {
1642         int ret = 0;
1643         struct btrfs_delayed_tree_ref *ref;
1644         u64 parent = 0;
1645         u64 ref_root = 0;
1646
1647         ref = btrfs_delayed_node_to_tree_ref(node);
1648         trace_run_delayed_tree_ref(trans->fs_info, node, ref, node->action);
1649
1650         if (node->type == BTRFS_SHARED_BLOCK_REF_KEY)
1651                 parent = ref->parent;
1652         ref_root = ref->root;
1653
1654         if (node->ref_mod != 1) {
1655                 btrfs_err(trans->fs_info,
1656         "btree block(%llu) has %d references rather than 1: action %d ref_root %llu parent %llu",
1657                           node->bytenr, node->ref_mod, node->action, ref_root,
1658                           parent);
1659                 return -EIO;
1660         }
1661         if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
1662                 BUG_ON(!extent_op || !extent_op->update_flags);
1663                 ret = alloc_reserved_tree_block(trans, node, extent_op);
1664         } else if (node->action == BTRFS_ADD_DELAYED_REF) {
1665                 ret = __btrfs_inc_extent_ref(trans, node, parent, ref_root,
1666                                              ref->level, 0, 1, extent_op);
1667         } else if (node->action == BTRFS_DROP_DELAYED_REF) {
1668                 ret = __btrfs_free_extent(trans, node, parent, ref_root,
1669                                           ref->level, 0, 1, extent_op);
1670         } else {
1671                 BUG();
1672         }
1673         return ret;
1674 }
1675
1676 /* helper function to actually process a single delayed ref entry */
1677 static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
1678                                struct btrfs_delayed_ref_node *node,
1679                                struct btrfs_delayed_extent_op *extent_op,
1680                                int insert_reserved)
1681 {
1682         int ret = 0;
1683
1684         if (trans->aborted) {
1685                 if (insert_reserved)
1686                         btrfs_pin_extent(trans->fs_info, node->bytenr,
1687                                          node->num_bytes, 1);
1688                 return 0;
1689         }
1690
1691         if (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
1692             node->type == BTRFS_SHARED_BLOCK_REF_KEY)
1693                 ret = run_delayed_tree_ref(trans, node, extent_op,
1694                                            insert_reserved);
1695         else if (node->type == BTRFS_EXTENT_DATA_REF_KEY ||
1696                  node->type == BTRFS_SHARED_DATA_REF_KEY)
1697                 ret = run_delayed_data_ref(trans, node, extent_op,
1698                                            insert_reserved);
1699         else
1700                 BUG();
1701         if (ret && insert_reserved)
1702                 btrfs_pin_extent(trans->fs_info, node->bytenr,
1703                                  node->num_bytes, 1);
1704         return ret;
1705 }
1706
1707 static inline struct btrfs_delayed_ref_node *
1708 select_delayed_ref(struct btrfs_delayed_ref_head *head)
1709 {
1710         struct btrfs_delayed_ref_node *ref;
1711
1712         if (RB_EMPTY_ROOT(&head->ref_tree.rb_root))
1713                 return NULL;
1714
1715         /*
1716          * Select a delayed ref of type BTRFS_ADD_DELAYED_REF first.
1717          * This is to prevent a ref count from going down to zero, which deletes
1718          * the extent item from the extent tree, when there still are references
1719          * to add, which would fail because they would not find the extent item.
1720          */
1721         if (!list_empty(&head->ref_add_list))
1722                 return list_first_entry(&head->ref_add_list,
1723                                 struct btrfs_delayed_ref_node, add_list);
1724
1725         ref = rb_entry(rb_first_cached(&head->ref_tree),
1726                        struct btrfs_delayed_ref_node, ref_node);
1727         ASSERT(list_empty(&ref->add_list));
1728         return ref;
1729 }
1730
1731 static void unselect_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs,
1732                                       struct btrfs_delayed_ref_head *head)
1733 {
1734         spin_lock(&delayed_refs->lock);
1735         head->processing = 0;
1736         delayed_refs->num_heads_ready++;
1737         spin_unlock(&delayed_refs->lock);
1738         btrfs_delayed_ref_unlock(head);
1739 }
1740
1741 static struct btrfs_delayed_extent_op *cleanup_extent_op(
1742                                 struct btrfs_delayed_ref_head *head)
1743 {
1744         struct btrfs_delayed_extent_op *extent_op = head->extent_op;
1745
1746         if (!extent_op)
1747                 return NULL;
1748
1749         if (head->must_insert_reserved) {
1750                 head->extent_op = NULL;
1751                 btrfs_free_delayed_extent_op(extent_op);
1752                 return NULL;
1753         }
1754         return extent_op;
1755 }
1756
1757 static int run_and_cleanup_extent_op(struct btrfs_trans_handle *trans,
1758                                      struct btrfs_delayed_ref_head *head)
1759 {
1760         struct btrfs_delayed_extent_op *extent_op;
1761         int ret;
1762
1763         extent_op = cleanup_extent_op(head);
1764         if (!extent_op)
1765                 return 0;
1766         head->extent_op = NULL;
1767         spin_unlock(&head->lock);
1768         ret = run_delayed_extent_op(trans, head, extent_op);
1769         btrfs_free_delayed_extent_op(extent_op);
1770         return ret ? ret : 1;
1771 }
1772
1773 void btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
1774                                   struct btrfs_delayed_ref_root *delayed_refs,
1775                                   struct btrfs_delayed_ref_head *head)
1776 {
1777         int nr_items = 1;       /* Dropping this ref head update. */
1778
1779         if (head->total_ref_mod < 0) {
1780                 struct btrfs_space_info *space_info;
1781                 u64 flags;
1782
1783                 if (head->is_data)
1784                         flags = BTRFS_BLOCK_GROUP_DATA;
1785                 else if (head->is_system)
1786                         flags = BTRFS_BLOCK_GROUP_SYSTEM;
1787                 else
1788                         flags = BTRFS_BLOCK_GROUP_METADATA;
1789                 space_info = btrfs_find_space_info(fs_info, flags);
1790                 ASSERT(space_info);
1791                 percpu_counter_add_batch(&space_info->total_bytes_pinned,
1792                                    -head->num_bytes,
1793                                    BTRFS_TOTAL_BYTES_PINNED_BATCH);
1794
1795                 /*
1796                  * We had csum deletions accounted for in our delayed refs rsv,
1797                  * we need to drop the csum leaves for this update from our
1798                  * delayed_refs_rsv.
1799                  */
1800                 if (head->is_data) {
1801                         spin_lock(&delayed_refs->lock);
1802                         delayed_refs->pending_csums -= head->num_bytes;
1803                         spin_unlock(&delayed_refs->lock);
1804                         nr_items += btrfs_csum_bytes_to_leaves(fs_info,
1805                                 head->num_bytes);
1806                 }
1807         }
1808
1809         btrfs_delayed_refs_rsv_release(fs_info, nr_items);
1810 }
1811
1812 static int cleanup_ref_head(struct btrfs_trans_handle *trans,
1813                             struct btrfs_delayed_ref_head *head)
1814 {
1815
1816         struct btrfs_fs_info *fs_info = trans->fs_info;
1817         struct btrfs_delayed_ref_root *delayed_refs;
1818         int ret;
1819
1820         delayed_refs = &trans->transaction->delayed_refs;
1821
1822         ret = run_and_cleanup_extent_op(trans, head);
1823         if (ret < 0) {
1824                 unselect_delayed_ref_head(delayed_refs, head);
1825                 btrfs_debug(fs_info, "run_delayed_extent_op returned %d", ret);
1826                 return ret;
1827         } else if (ret) {
1828                 return ret;
1829         }
1830
1831         /*
1832          * Need to drop our head ref lock and re-acquire the delayed ref lock
1833          * and then re-check to make sure nobody got added.
1834          */
1835         spin_unlock(&head->lock);
1836         spin_lock(&delayed_refs->lock);
1837         spin_lock(&head->lock);
1838         if (!RB_EMPTY_ROOT(&head->ref_tree.rb_root) || head->extent_op) {
1839                 spin_unlock(&head->lock);
1840                 spin_unlock(&delayed_refs->lock);
1841                 return 1;
1842         }
1843         btrfs_delete_ref_head(delayed_refs, head);
1844         spin_unlock(&head->lock);
1845         spin_unlock(&delayed_refs->lock);
1846
1847         if (head->must_insert_reserved) {
1848                 btrfs_pin_extent(fs_info, head->bytenr,
1849                                  head->num_bytes, 1);
1850                 if (head->is_data) {
1851                         ret = btrfs_del_csums(trans, fs_info, head->bytenr,
1852                                               head->num_bytes);
1853                 }
1854         }
1855
1856         btrfs_cleanup_ref_head_accounting(fs_info, delayed_refs, head);
1857
1858         trace_run_delayed_ref_head(fs_info, head, 0);
1859         btrfs_delayed_ref_unlock(head);
1860         btrfs_put_delayed_ref_head(head);
1861         return 0;
1862 }
1863
1864 static struct btrfs_delayed_ref_head *btrfs_obtain_ref_head(
1865                                         struct btrfs_trans_handle *trans)
1866 {
1867         struct btrfs_delayed_ref_root *delayed_refs =
1868                 &trans->transaction->delayed_refs;
1869         struct btrfs_delayed_ref_head *head = NULL;
1870         int ret;
1871
1872         spin_lock(&delayed_refs->lock);
1873         head = btrfs_select_ref_head(delayed_refs);
1874         if (!head) {
1875                 spin_unlock(&delayed_refs->lock);
1876                 return head;
1877         }
1878
1879         /*
1880          * Grab the lock that says we are going to process all the refs for
1881          * this head
1882          */
1883         ret = btrfs_delayed_ref_lock(delayed_refs, head);
1884         spin_unlock(&delayed_refs->lock);
1885
1886         /*
1887          * We may have dropped the spin lock to get the head mutex lock, and
1888          * that might have given someone else time to free the head.  If that's
1889          * true, it has been removed from our list and we can move on.
1890          */
1891         if (ret == -EAGAIN)
1892                 head = ERR_PTR(-EAGAIN);
1893
1894         return head;
1895 }
1896
1897 static int btrfs_run_delayed_refs_for_head(struct btrfs_trans_handle *trans,
1898                                     struct btrfs_delayed_ref_head *locked_ref,
1899                                     unsigned long *run_refs)
1900 {
1901         struct btrfs_fs_info *fs_info = trans->fs_info;
1902         struct btrfs_delayed_ref_root *delayed_refs;
1903         struct btrfs_delayed_extent_op *extent_op;
1904         struct btrfs_delayed_ref_node *ref;
1905         int must_insert_reserved = 0;
1906         int ret;
1907
1908         delayed_refs = &trans->transaction->delayed_refs;
1909
1910         lockdep_assert_held(&locked_ref->mutex);
1911         lockdep_assert_held(&locked_ref->lock);
1912
1913         while ((ref = select_delayed_ref(locked_ref))) {
1914                 if (ref->seq &&
1915                     btrfs_check_delayed_seq(fs_info, ref->seq)) {
1916                         spin_unlock(&locked_ref->lock);
1917                         unselect_delayed_ref_head(delayed_refs, locked_ref);
1918                         return -EAGAIN;
1919                 }
1920
1921                 (*run_refs)++;
1922                 ref->in_tree = 0;
1923                 rb_erase_cached(&ref->ref_node, &locked_ref->ref_tree);
1924                 RB_CLEAR_NODE(&ref->ref_node);
1925                 if (!list_empty(&ref->add_list))
1926                         list_del(&ref->add_list);
1927                 /*
1928                  * When we play the delayed ref, also correct the ref_mod on
1929                  * head
1930                  */
1931                 switch (ref->action) {
1932                 case BTRFS_ADD_DELAYED_REF:
1933                 case BTRFS_ADD_DELAYED_EXTENT:
1934                         locked_ref->ref_mod -= ref->ref_mod;
1935                         break;
1936                 case BTRFS_DROP_DELAYED_REF:
1937                         locked_ref->ref_mod += ref->ref_mod;
1938                         break;
1939                 default:
1940                         WARN_ON(1);
1941                 }
1942                 atomic_dec(&delayed_refs->num_entries);
1943
1944                 /*
1945                  * Record the must_insert_reserved flag before we drop the
1946                  * spin lock.
1947                  */
1948                 must_insert_reserved = locked_ref->must_insert_reserved;
1949                 locked_ref->must_insert_reserved = 0;
1950
1951                 extent_op = locked_ref->extent_op;
1952                 locked_ref->extent_op = NULL;
1953                 spin_unlock(&locked_ref->lock);
1954
1955                 ret = run_one_delayed_ref(trans, ref, extent_op,
1956                                           must_insert_reserved);
1957
1958                 btrfs_free_delayed_extent_op(extent_op);
1959                 if (ret) {
1960                         unselect_delayed_ref_head(delayed_refs, locked_ref);
1961                         btrfs_put_delayed_ref(ref);
1962                         btrfs_debug(fs_info, "run_one_delayed_ref returned %d",
1963                                     ret);
1964                         return ret;
1965                 }
1966
1967                 btrfs_put_delayed_ref(ref);
1968                 cond_resched();
1969
1970                 spin_lock(&locked_ref->lock);
1971                 btrfs_merge_delayed_refs(trans, delayed_refs, locked_ref);
1972         }
1973
1974         return 0;
1975 }
1976
1977 /*
1978  * Returns 0 on success or if called with an already aborted transaction.
1979  * Returns -ENOMEM or -EIO on failure and will abort the transaction.
1980  */
1981 static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
1982                                              unsigned long nr)
1983 {
1984         struct btrfs_fs_info *fs_info = trans->fs_info;
1985         struct btrfs_delayed_ref_root *delayed_refs;
1986         struct btrfs_delayed_ref_head *locked_ref = NULL;
1987         ktime_t start = ktime_get();
1988         int ret;
1989         unsigned long count = 0;
1990         unsigned long actual_count = 0;
1991
1992         delayed_refs = &trans->transaction->delayed_refs;
1993         do {
1994                 if (!locked_ref) {
1995                         locked_ref = btrfs_obtain_ref_head(trans);
1996                         if (IS_ERR_OR_NULL(locked_ref)) {
1997                                 if (PTR_ERR(locked_ref) == -EAGAIN) {
1998                                         continue;
1999                                 } else {
2000                                         break;
2001                                 }
2002                         }
2003                         count++;
2004                 }
2005                 /*
2006                  * We need to try and merge add/drops of the same ref since we
2007                  * can run into issues with relocate dropping the implicit ref
2008                  * and then it being added back again before the drop can
2009                  * finish.  If we merged anything we need to re-loop so we can
2010                  * get a good ref.
2011                  * Or we can get node references of the same type that weren't
2012                  * merged when created due to bumps in the tree mod seq, and
2013                  * we need to merge them to prevent adding an inline extent
2014                  * backref before dropping it (triggering a BUG_ON at
2015                  * insert_inline_extent_backref()).
2016                  */
2017                 spin_lock(&locked_ref->lock);
2018                 btrfs_merge_delayed_refs(trans, delayed_refs, locked_ref);
2019
2020                 ret = btrfs_run_delayed_refs_for_head(trans, locked_ref,
2021                                                       &actual_count);
2022                 if (ret < 0 && ret != -EAGAIN) {
2023                         /*
2024                          * Error, btrfs_run_delayed_refs_for_head already
2025                          * unlocked everything so just bail out
2026                          */
2027                         return ret;
2028                 } else if (!ret) {
2029                         /*
2030                          * Success, perform the usual cleanup of a processed
2031                          * head
2032                          */
2033                         ret = cleanup_ref_head(trans, locked_ref);
2034                         if (ret > 0 ) {
2035                                 /* We dropped our lock, we need to loop. */
2036                                 ret = 0;
2037                                 continue;
2038                         } else if (ret) {
2039                                 return ret;
2040                         }
2041                 }
2042
2043                 /*
2044                  * Either success case or btrfs_run_delayed_refs_for_head
2045                  * returned -EAGAIN, meaning we need to select another head
2046                  */
2047
2048                 locked_ref = NULL;
2049                 cond_resched();
2050         } while ((nr != -1 && count < nr) || locked_ref);
2051
2052         /*
2053          * We don't want to include ref heads since we can have empty ref heads
2054          * and those will drastically skew our runtime down since we just do
2055          * accounting, no actual extent tree updates.
2056          */
2057         if (actual_count > 0) {
2058                 u64 runtime = ktime_to_ns(ktime_sub(ktime_get(), start));
2059                 u64 avg;
2060
2061                 /*
2062                  * We weigh the current average higher than our current runtime
2063                  * to avoid large swings in the average.
2064                  */
2065                 spin_lock(&delayed_refs->lock);
2066                 avg = fs_info->avg_delayed_ref_runtime * 3 + runtime;
2067                 fs_info->avg_delayed_ref_runtime = avg >> 2;    /* div by 4 */
2068                 spin_unlock(&delayed_refs->lock);
2069         }
2070         return 0;
2071 }
2072
2073 #ifdef SCRAMBLE_DELAYED_REFS
2074 /*
2075  * Normally delayed refs get processed in ascending bytenr order. This
2076  * correlates in most cases to the order added. To expose dependencies on this
2077  * order, we start to process the tree in the middle instead of the beginning
2078  */
2079 static u64 find_middle(struct rb_root *root)
2080 {
2081         struct rb_node *n = root->rb_node;
2082         struct btrfs_delayed_ref_node *entry;
2083         int alt = 1;
2084         u64 middle;
2085         u64 first = 0, last = 0;
2086
2087         n = rb_first(root);
2088         if (n) {
2089                 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2090                 first = entry->bytenr;
2091         }
2092         n = rb_last(root);
2093         if (n) {
2094                 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2095                 last = entry->bytenr;
2096         }
2097         n = root->rb_node;
2098
2099         while (n) {
2100                 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2101                 WARN_ON(!entry->in_tree);
2102
2103                 middle = entry->bytenr;
2104
2105                 if (alt)
2106                         n = n->rb_left;
2107                 else
2108                         n = n->rb_right;
2109
2110                 alt = 1 - alt;
2111         }
2112         return middle;
2113 }
2114 #endif
2115
2116 static inline u64 heads_to_leaves(struct btrfs_fs_info *fs_info, u64 heads)
2117 {
2118         u64 num_bytes;
2119
2120         num_bytes = heads * (sizeof(struct btrfs_extent_item) +
2121                              sizeof(struct btrfs_extent_inline_ref));
2122         if (!btrfs_fs_incompat(fs_info, SKINNY_METADATA))
2123                 num_bytes += heads * sizeof(struct btrfs_tree_block_info);
2124
2125         /*
2126          * We don't ever fill up leaves all the way so multiply by 2 just to be
2127          * closer to what we're really going to want to use.
2128          */
2129         return div_u64(num_bytes, BTRFS_LEAF_DATA_SIZE(fs_info));
2130 }
2131
2132 /*
2133  * Takes the number of bytes to be csumm'ed and figures out how many leaves it
2134  * would require to store the csums for that many bytes.
2135  */
2136 u64 btrfs_csum_bytes_to_leaves(struct btrfs_fs_info *fs_info, u64 csum_bytes)
2137 {
2138         u64 csum_size;
2139         u64 num_csums_per_leaf;
2140         u64 num_csums;
2141
2142         csum_size = BTRFS_MAX_ITEM_SIZE(fs_info);
2143         num_csums_per_leaf = div64_u64(csum_size,
2144                         (u64)btrfs_super_csum_size(fs_info->super_copy));
2145         num_csums = div64_u64(csum_bytes, fs_info->sectorsize);
2146         num_csums += num_csums_per_leaf - 1;
2147         num_csums = div64_u64(num_csums, num_csums_per_leaf);
2148         return num_csums;
2149 }
2150
2151 /*
2152  * this starts processing the delayed reference count updates and
2153  * extent insertions we have queued up so far.  count can be
2154  * 0, which means to process everything in the tree at the start
2155  * of the run (but not newly added entries), or it can be some target
2156  * number you'd like to process.
2157  *
2158  * Returns 0 on success or if called with an aborted transaction
2159  * Returns <0 on error and aborts the transaction
2160  */
2161 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
2162                            unsigned long count)
2163 {
2164         struct btrfs_fs_info *fs_info = trans->fs_info;
2165         struct rb_node *node;
2166         struct btrfs_delayed_ref_root *delayed_refs;
2167         struct btrfs_delayed_ref_head *head;
2168         int ret;
2169         int run_all = count == (unsigned long)-1;
2170
2171         /* We'll clean this up in btrfs_cleanup_transaction */
2172         if (trans->aborted)
2173                 return 0;
2174
2175         if (test_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags))
2176                 return 0;
2177
2178         delayed_refs = &trans->transaction->delayed_refs;
2179         if (count == 0)
2180                 count = atomic_read(&delayed_refs->num_entries) * 2;
2181
2182 again:
2183 #ifdef SCRAMBLE_DELAYED_REFS
2184         delayed_refs->run_delayed_start = find_middle(&delayed_refs->root);
2185 #endif
2186         ret = __btrfs_run_delayed_refs(trans, count);
2187         if (ret < 0) {
2188                 btrfs_abort_transaction(trans, ret);
2189                 return ret;
2190         }
2191
2192         if (run_all) {
2193                 btrfs_create_pending_block_groups(trans);
2194
2195                 spin_lock(&delayed_refs->lock);
2196                 node = rb_first_cached(&delayed_refs->href_root);
2197                 if (!node) {
2198                         spin_unlock(&delayed_refs->lock);
2199                         goto out;
2200                 }
2201                 head = rb_entry(node, struct btrfs_delayed_ref_head,
2202                                 href_node);
2203                 refcount_inc(&head->refs);
2204                 spin_unlock(&delayed_refs->lock);
2205
2206                 /* Mutex was contended, block until it's released and retry. */
2207                 mutex_lock(&head->mutex);
2208                 mutex_unlock(&head->mutex);
2209
2210                 btrfs_put_delayed_ref_head(head);
2211                 cond_resched();
2212                 goto again;
2213         }
2214 out:
2215         return 0;
2216 }
2217
2218 int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
2219                                 u64 bytenr, u64 num_bytes, u64 flags,
2220                                 int level, int is_data)
2221 {
2222         struct btrfs_delayed_extent_op *extent_op;
2223         int ret;
2224
2225         extent_op = btrfs_alloc_delayed_extent_op();
2226         if (!extent_op)
2227                 return -ENOMEM;
2228
2229         extent_op->flags_to_set = flags;
2230         extent_op->update_flags = true;
2231         extent_op->update_key = false;
2232         extent_op->is_data = is_data ? true : false;
2233         extent_op->level = level;
2234
2235         ret = btrfs_add_delayed_extent_op(trans, bytenr, num_bytes, extent_op);
2236         if (ret)
2237                 btrfs_free_delayed_extent_op(extent_op);
2238         return ret;
2239 }
2240
2241 static noinline int check_delayed_ref(struct btrfs_root *root,
2242                                       struct btrfs_path *path,
2243                                       u64 objectid, u64 offset, u64 bytenr)
2244 {
2245         struct btrfs_delayed_ref_head *head;
2246         struct btrfs_delayed_ref_node *ref;
2247         struct btrfs_delayed_data_ref *data_ref;
2248         struct btrfs_delayed_ref_root *delayed_refs;
2249         struct btrfs_transaction *cur_trans;
2250         struct rb_node *node;
2251         int ret = 0;
2252
2253         spin_lock(&root->fs_info->trans_lock);
2254         cur_trans = root->fs_info->running_transaction;
2255         if (cur_trans)
2256                 refcount_inc(&cur_trans->use_count);
2257         spin_unlock(&root->fs_info->trans_lock);
2258         if (!cur_trans)
2259                 return 0;
2260
2261         delayed_refs = &cur_trans->delayed_refs;
2262         spin_lock(&delayed_refs->lock);
2263         head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
2264         if (!head) {
2265                 spin_unlock(&delayed_refs->lock);
2266                 btrfs_put_transaction(cur_trans);
2267                 return 0;
2268         }
2269
2270         if (!mutex_trylock(&head->mutex)) {
2271                 refcount_inc(&head->refs);
2272                 spin_unlock(&delayed_refs->lock);
2273
2274                 btrfs_release_path(path);
2275
2276                 /*
2277                  * Mutex was contended, block until it's released and let
2278                  * caller try again
2279                  */
2280                 mutex_lock(&head->mutex);
2281                 mutex_unlock(&head->mutex);
2282                 btrfs_put_delayed_ref_head(head);
2283                 btrfs_put_transaction(cur_trans);
2284                 return -EAGAIN;
2285         }
2286         spin_unlock(&delayed_refs->lock);
2287
2288         spin_lock(&head->lock);
2289         /*
2290          * XXX: We should replace this with a proper search function in the
2291          * future.
2292          */
2293         for (node = rb_first_cached(&head->ref_tree); node;
2294              node = rb_next(node)) {
2295                 ref = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
2296                 /* If it's a shared ref we know a cross reference exists */
2297                 if (ref->type != BTRFS_EXTENT_DATA_REF_KEY) {
2298                         ret = 1;
2299                         break;
2300                 }
2301
2302                 data_ref = btrfs_delayed_node_to_data_ref(ref);
2303
2304                 /*
2305                  * If our ref doesn't match the one we're currently looking at
2306                  * then we have a cross reference.
2307                  */
2308                 if (data_ref->root != root->root_key.objectid ||
2309                     data_ref->objectid != objectid ||
2310                     data_ref->offset != offset) {
2311                         ret = 1;
2312                         break;
2313                 }
2314         }
2315         spin_unlock(&head->lock);
2316         mutex_unlock(&head->mutex);
2317         btrfs_put_transaction(cur_trans);
2318         return ret;
2319 }
2320
2321 static noinline int check_committed_ref(struct btrfs_root *root,
2322                                         struct btrfs_path *path,
2323                                         u64 objectid, u64 offset, u64 bytenr)
2324 {
2325         struct btrfs_fs_info *fs_info = root->fs_info;
2326         struct btrfs_root *extent_root = fs_info->extent_root;
2327         struct extent_buffer *leaf;
2328         struct btrfs_extent_data_ref *ref;
2329         struct btrfs_extent_inline_ref *iref;
2330         struct btrfs_extent_item *ei;
2331         struct btrfs_key key;
2332         u32 item_size;
2333         int type;
2334         int ret;
2335
2336         key.objectid = bytenr;
2337         key.offset = (u64)-1;
2338         key.type = BTRFS_EXTENT_ITEM_KEY;
2339
2340         ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
2341         if (ret < 0)
2342                 goto out;
2343         BUG_ON(ret == 0); /* Corruption */
2344
2345         ret = -ENOENT;
2346         if (path->slots[0] == 0)
2347                 goto out;
2348
2349         path->slots[0]--;
2350         leaf = path->nodes[0];
2351         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
2352
2353         if (key.objectid != bytenr || key.type != BTRFS_EXTENT_ITEM_KEY)
2354                 goto out;
2355
2356         ret = 1;
2357         item_size = btrfs_item_size_nr(leaf, path->slots[0]);
2358         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
2359
2360         /* If extent item has more than 1 inline ref then it's shared */
2361         if (item_size != sizeof(*ei) +
2362             btrfs_extent_inline_ref_size(BTRFS_EXTENT_DATA_REF_KEY))
2363                 goto out;
2364
2365         /* If extent created before last snapshot => it's definitely shared */
2366         if (btrfs_extent_generation(leaf, ei) <=
2367             btrfs_root_last_snapshot(&root->root_item))
2368                 goto out;
2369
2370         iref = (struct btrfs_extent_inline_ref *)(ei + 1);
2371
2372         /* If this extent has SHARED_DATA_REF then it's shared */
2373         type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
2374         if (type != BTRFS_EXTENT_DATA_REF_KEY)
2375                 goto out;
2376
2377         ref = (struct btrfs_extent_data_ref *)(&iref->offset);
2378         if (btrfs_extent_refs(leaf, ei) !=
2379             btrfs_extent_data_ref_count(leaf, ref) ||
2380             btrfs_extent_data_ref_root(leaf, ref) !=
2381             root->root_key.objectid ||
2382             btrfs_extent_data_ref_objectid(leaf, ref) != objectid ||
2383             btrfs_extent_data_ref_offset(leaf, ref) != offset)
2384                 goto out;
2385
2386         ret = 0;
2387 out:
2388         return ret;
2389 }
2390
2391 int btrfs_cross_ref_exist(struct btrfs_root *root, u64 objectid, u64 offset,
2392                           u64 bytenr)
2393 {
2394         struct btrfs_path *path;
2395         int ret;
2396
2397         path = btrfs_alloc_path();
2398         if (!path)
2399                 return -ENOMEM;
2400
2401         do {
2402                 ret = check_committed_ref(root, path, objectid,
2403                                           offset, bytenr);
2404                 if (ret && ret != -ENOENT)
2405                         goto out;
2406
2407                 ret = check_delayed_ref(root, path, objectid, offset, bytenr);
2408         } while (ret == -EAGAIN);
2409
2410 out:
2411         btrfs_free_path(path);
2412         if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID)
2413                 WARN_ON(ret > 0);
2414         return ret;
2415 }
2416
2417 static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
2418                            struct btrfs_root *root,
2419                            struct extent_buffer *buf,
2420                            int full_backref, int inc)
2421 {
2422         struct btrfs_fs_info *fs_info = root->fs_info;
2423         u64 bytenr;
2424         u64 num_bytes;
2425         u64 parent;
2426         u64 ref_root;
2427         u32 nritems;
2428         struct btrfs_key key;
2429         struct btrfs_file_extent_item *fi;
2430         struct btrfs_ref generic_ref = { 0 };
2431         bool for_reloc = btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC);
2432         int i;
2433         int action;
2434         int level;
2435         int ret = 0;
2436
2437         if (btrfs_is_testing(fs_info))
2438                 return 0;
2439
2440         ref_root = btrfs_header_owner(buf);
2441         nritems = btrfs_header_nritems(buf);
2442         level = btrfs_header_level(buf);
2443
2444         if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state) && level == 0)
2445                 return 0;
2446
2447         if (full_backref)
2448                 parent = buf->start;
2449         else
2450                 parent = 0;
2451         if (inc)
2452                 action = BTRFS_ADD_DELAYED_REF;
2453         else
2454                 action = BTRFS_DROP_DELAYED_REF;
2455
2456         for (i = 0; i < nritems; i++) {
2457                 if (level == 0) {
2458                         btrfs_item_key_to_cpu(buf, &key, i);
2459                         if (key.type != BTRFS_EXTENT_DATA_KEY)
2460                                 continue;
2461                         fi = btrfs_item_ptr(buf, i,
2462                                             struct btrfs_file_extent_item);
2463                         if (btrfs_file_extent_type(buf, fi) ==
2464                             BTRFS_FILE_EXTENT_INLINE)
2465                                 continue;
2466                         bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
2467                         if (bytenr == 0)
2468                                 continue;
2469
2470                         num_bytes = btrfs_file_extent_disk_num_bytes(buf, fi);
2471                         key.offset -= btrfs_file_extent_offset(buf, fi);
2472                         btrfs_init_generic_ref(&generic_ref, action, bytenr,
2473                                                num_bytes, parent);
2474                         generic_ref.real_root = root->root_key.objectid;
2475                         btrfs_init_data_ref(&generic_ref, ref_root, key.objectid,
2476                                             key.offset);
2477                         generic_ref.skip_qgroup = for_reloc;
2478                         if (inc)
2479                                 ret = btrfs_inc_extent_ref(trans, &generic_ref);
2480                         else
2481                                 ret = btrfs_free_extent(trans, &generic_ref);
2482                         if (ret)
2483                                 goto fail;
2484                 } else {
2485                         bytenr = btrfs_node_blockptr(buf, i);
2486                         num_bytes = fs_info->nodesize;
2487                         btrfs_init_generic_ref(&generic_ref, action, bytenr,
2488                                                num_bytes, parent);
2489                         generic_ref.real_root = root->root_key.objectid;
2490                         btrfs_init_tree_ref(&generic_ref, level - 1, ref_root);
2491                         generic_ref.skip_qgroup = for_reloc;
2492                         if (inc)
2493                                 ret = btrfs_inc_extent_ref(trans, &generic_ref);
2494                         else
2495                                 ret = btrfs_free_extent(trans, &generic_ref);
2496                         if (ret)
2497                                 goto fail;
2498                 }
2499         }
2500         return 0;
2501 fail:
2502         return ret;
2503 }
2504
2505 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2506                   struct extent_buffer *buf, int full_backref)
2507 {
2508         return __btrfs_mod_ref(trans, root, buf, full_backref, 1);
2509 }
2510
2511 int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2512                   struct extent_buffer *buf, int full_backref)
2513 {
2514         return __btrfs_mod_ref(trans, root, buf, full_backref, 0);
2515 }
2516
2517 int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr)
2518 {
2519         struct btrfs_block_group_cache *block_group;
2520         int readonly = 0;
2521
2522         block_group = btrfs_lookup_block_group(fs_info, bytenr);
2523         if (!block_group || block_group->ro)
2524                 readonly = 1;
2525         if (block_group)
2526                 btrfs_put_block_group(block_group);
2527         return readonly;
2528 }
2529
2530 static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
2531 {
2532         struct btrfs_fs_info *fs_info = root->fs_info;
2533         u64 flags;
2534         u64 ret;
2535
2536         if (data)
2537                 flags = BTRFS_BLOCK_GROUP_DATA;
2538         else if (root == fs_info->chunk_root)
2539                 flags = BTRFS_BLOCK_GROUP_SYSTEM;
2540         else
2541                 flags = BTRFS_BLOCK_GROUP_METADATA;
2542
2543         ret = btrfs_get_alloc_profile(fs_info, flags);
2544         return ret;
2545 }
2546
2547 static u64 first_logical_byte(struct btrfs_fs_info *fs_info, u64 search_start)
2548 {
2549         struct btrfs_block_group_cache *cache;
2550         u64 bytenr;
2551
2552         spin_lock(&fs_info->block_group_cache_lock);
2553         bytenr = fs_info->first_logical_byte;
2554         spin_unlock(&fs_info->block_group_cache_lock);
2555
2556         if (bytenr < (u64)-1)
2557                 return bytenr;
2558
2559         cache = btrfs_lookup_first_block_group(fs_info, search_start);
2560         if (!cache)
2561                 return 0;
2562
2563         bytenr = cache->key.objectid;
2564         btrfs_put_block_group(cache);
2565
2566         return bytenr;
2567 }
2568
2569 static int pin_down_extent(struct btrfs_block_group_cache *cache,
2570                            u64 bytenr, u64 num_bytes, int reserved)
2571 {
2572         struct btrfs_fs_info *fs_info = cache->fs_info;
2573
2574         spin_lock(&cache->space_info->lock);
2575         spin_lock(&cache->lock);
2576         cache->pinned += num_bytes;
2577         btrfs_space_info_update_bytes_pinned(fs_info, cache->space_info,
2578                                              num_bytes);
2579         if (reserved) {
2580                 cache->reserved -= num_bytes;
2581                 cache->space_info->bytes_reserved -= num_bytes;
2582         }
2583         spin_unlock(&cache->lock);
2584         spin_unlock(&cache->space_info->lock);
2585
2586         percpu_counter_add_batch(&cache->space_info->total_bytes_pinned,
2587                     num_bytes, BTRFS_TOTAL_BYTES_PINNED_BATCH);
2588         set_extent_dirty(fs_info->pinned_extents, bytenr,
2589                          bytenr + num_bytes - 1, GFP_NOFS | __GFP_NOFAIL);
2590         return 0;
2591 }
2592
2593 /*
2594  * this function must be called within transaction
2595  */
2596 int btrfs_pin_extent(struct btrfs_fs_info *fs_info,
2597                      u64 bytenr, u64 num_bytes, int reserved)
2598 {
2599         struct btrfs_block_group_cache *cache;
2600
2601         cache = btrfs_lookup_block_group(fs_info, bytenr);
2602         BUG_ON(!cache); /* Logic error */
2603
2604         pin_down_extent(cache, bytenr, num_bytes, reserved);
2605
2606         btrfs_put_block_group(cache);
2607         return 0;
2608 }
2609
2610 /*
2611  * this function must be called within transaction
2612  */
2613 int btrfs_pin_extent_for_log_replay(struct btrfs_fs_info *fs_info,
2614                                     u64 bytenr, u64 num_bytes)
2615 {
2616         struct btrfs_block_group_cache *cache;
2617         int ret;
2618
2619         cache = btrfs_lookup_block_group(fs_info, bytenr);
2620         if (!cache)
2621                 return -EINVAL;
2622
2623         /*
2624          * pull in the free space cache (if any) so that our pin
2625          * removes the free space from the cache.  We have load_only set
2626          * to one because the slow code to read in the free extents does check
2627          * the pinned extents.
2628          */
2629         btrfs_cache_block_group(cache, 1);
2630
2631         pin_down_extent(cache, bytenr, num_bytes, 0);
2632
2633         /* remove us from the free space cache (if we're there at all) */
2634         ret = btrfs_remove_free_space(cache, bytenr, num_bytes);
2635         btrfs_put_block_group(cache);
2636         return ret;
2637 }
2638
2639 static int __exclude_logged_extent(struct btrfs_fs_info *fs_info,
2640                                    u64 start, u64 num_bytes)
2641 {
2642         int ret;
2643         struct btrfs_block_group_cache *block_group;
2644         struct btrfs_caching_control *caching_ctl;
2645
2646         block_group = btrfs_lookup_block_group(fs_info, start);
2647         if (!block_group)
2648                 return -EINVAL;
2649
2650         btrfs_cache_block_group(block_group, 0);
2651         caching_ctl = btrfs_get_caching_control(block_group);
2652
2653         if (!caching_ctl) {
2654                 /* Logic error */
2655                 BUG_ON(!btrfs_block_group_cache_done(block_group));
2656                 ret = btrfs_remove_free_space(block_group, start, num_bytes);
2657         } else {
2658                 mutex_lock(&caching_ctl->mutex);
2659
2660                 if (start >= caching_ctl->progress) {
2661                         ret = btrfs_add_excluded_extent(fs_info, start,
2662                                                         num_bytes);
2663                 } else if (start + num_bytes <= caching_ctl->progress) {
2664                         ret = btrfs_remove_free_space(block_group,
2665                                                       start, num_bytes);
2666                 } else {
2667                         num_bytes = caching_ctl->progress - start;
2668                         ret = btrfs_remove_free_space(block_group,
2669                                                       start, num_bytes);
2670                         if (ret)
2671                                 goto out_lock;
2672
2673                         num_bytes = (start + num_bytes) -
2674                                 caching_ctl->progress;
2675                         start = caching_ctl->progress;
2676                         ret = btrfs_add_excluded_extent(fs_info, start,
2677                                                         num_bytes);
2678                 }
2679 out_lock:
2680                 mutex_unlock(&caching_ctl->mutex);
2681                 btrfs_put_caching_control(caching_ctl);
2682         }
2683         btrfs_put_block_group(block_group);
2684         return ret;
2685 }
2686
2687 int btrfs_exclude_logged_extents(struct extent_buffer *eb)
2688 {
2689         struct btrfs_fs_info *fs_info = eb->fs_info;
2690         struct btrfs_file_extent_item *item;
2691         struct btrfs_key key;
2692         int found_type;
2693         int i;
2694         int ret = 0;
2695
2696         if (!btrfs_fs_incompat(fs_info, MIXED_GROUPS))
2697                 return 0;
2698
2699         for (i = 0; i < btrfs_header_nritems(eb); i++) {
2700                 btrfs_item_key_to_cpu(eb, &key, i);
2701                 if (key.type != BTRFS_EXTENT_DATA_KEY)
2702                         continue;
2703                 item = btrfs_item_ptr(eb, i, struct btrfs_file_extent_item);
2704                 found_type = btrfs_file_extent_type(eb, item);
2705                 if (found_type == BTRFS_FILE_EXTENT_INLINE)
2706                         continue;
2707                 if (btrfs_file_extent_disk_bytenr(eb, item) == 0)
2708                         continue;
2709                 key.objectid = btrfs_file_extent_disk_bytenr(eb, item);
2710                 key.offset = btrfs_file_extent_disk_num_bytes(eb, item);
2711                 ret = __exclude_logged_extent(fs_info, key.objectid, key.offset);
2712                 if (ret)
2713                         break;
2714         }
2715
2716         return ret;
2717 }
2718
2719 static void
2720 btrfs_inc_block_group_reservations(struct btrfs_block_group_cache *bg)
2721 {
2722         atomic_inc(&bg->reservations);
2723 }
2724
2725 void btrfs_prepare_extent_commit(struct btrfs_fs_info *fs_info)
2726 {
2727         struct btrfs_caching_control *next;
2728         struct btrfs_caching_control *caching_ctl;
2729         struct btrfs_block_group_cache *cache;
2730
2731         down_write(&fs_info->commit_root_sem);
2732
2733         list_for_each_entry_safe(caching_ctl, next,
2734                                  &fs_info->caching_block_groups, list) {
2735                 cache = caching_ctl->block_group;
2736                 if (btrfs_block_group_cache_done(cache)) {
2737                         cache->last_byte_to_unpin = (u64)-1;
2738                         list_del_init(&caching_ctl->list);
2739                         btrfs_put_caching_control(caching_ctl);
2740                 } else {
2741                         cache->last_byte_to_unpin = caching_ctl->progress;
2742                 }
2743         }
2744
2745         if (fs_info->pinned_extents == &fs_info->freed_extents[0])
2746                 fs_info->pinned_extents = &fs_info->freed_extents[1];
2747         else
2748                 fs_info->pinned_extents = &fs_info->freed_extents[0];
2749
2750         up_write(&fs_info->commit_root_sem);
2751
2752         btrfs_update_global_block_rsv(fs_info);
2753 }
2754
2755 /*
2756  * Returns the free cluster for the given space info and sets empty_cluster to
2757  * what it should be based on the mount options.
2758  */
2759 static struct btrfs_free_cluster *
2760 fetch_cluster_info(struct btrfs_fs_info *fs_info,
2761                    struct btrfs_space_info *space_info, u64 *empty_cluster)
2762 {
2763         struct btrfs_free_cluster *ret = NULL;
2764
2765         *empty_cluster = 0;
2766         if (btrfs_mixed_space_info(space_info))
2767                 return ret;
2768
2769         if (space_info->flags & BTRFS_BLOCK_GROUP_METADATA) {
2770                 ret = &fs_info->meta_alloc_cluster;
2771                 if (btrfs_test_opt(fs_info, SSD))
2772                         *empty_cluster = SZ_2M;
2773                 else
2774                         *empty_cluster = SZ_64K;
2775         } else if ((space_info->flags & BTRFS_BLOCK_GROUP_DATA) &&
2776                    btrfs_test_opt(fs_info, SSD_SPREAD)) {
2777                 *empty_cluster = SZ_2M;
2778                 ret = &fs_info->data_alloc_cluster;
2779         }
2780
2781         return ret;
2782 }
2783
2784 static int unpin_extent_range(struct btrfs_fs_info *fs_info,
2785                               u64 start, u64 end,
2786                               const bool return_free_space)
2787 {
2788         struct btrfs_block_group_cache *cache = NULL;
2789         struct btrfs_space_info *space_info;
2790         struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
2791         struct btrfs_free_cluster *cluster = NULL;
2792         u64 len;
2793         u64 total_unpinned = 0;
2794         u64 empty_cluster = 0;
2795         bool readonly;
2796
2797         while (start <= end) {
2798                 readonly = false;
2799                 if (!cache ||
2800                     start >= cache->key.objectid + cache->key.offset) {
2801                         if (cache)
2802                                 btrfs_put_block_group(cache);
2803                         total_unpinned = 0;
2804                         cache = btrfs_lookup_block_group(fs_info, start);
2805                         BUG_ON(!cache); /* Logic error */
2806
2807                         cluster = fetch_cluster_info(fs_info,
2808                                                      cache->space_info,
2809                                                      &empty_cluster);
2810                         empty_cluster <<= 1;
2811                 }
2812
2813                 len = cache->key.objectid + cache->key.offset - start;
2814                 len = min(len, end + 1 - start);
2815
2816                 if (start < cache->last_byte_to_unpin) {
2817                         len = min(len, cache->last_byte_to_unpin - start);
2818                         if (return_free_space)
2819                                 btrfs_add_free_space(cache, start, len);
2820                 }
2821
2822                 start += len;
2823                 total_unpinned += len;
2824                 space_info = cache->space_info;
2825
2826                 /*
2827                  * If this space cluster has been marked as fragmented and we've
2828                  * unpinned enough in this block group to potentially allow a
2829                  * cluster to be created inside of it go ahead and clear the
2830                  * fragmented check.
2831                  */
2832                 if (cluster && cluster->fragmented &&
2833                     total_unpinned > empty_cluster) {
2834                         spin_lock(&cluster->lock);
2835                         cluster->fragmented = 0;
2836                         spin_unlock(&cluster->lock);
2837                 }
2838
2839                 spin_lock(&space_info->lock);
2840                 spin_lock(&cache->lock);
2841                 cache->pinned -= len;
2842                 btrfs_space_info_update_bytes_pinned(fs_info, space_info, -len);
2843                 space_info->max_extent_size = 0;
2844                 percpu_counter_add_batch(&space_info->total_bytes_pinned,
2845                             -len, BTRFS_TOTAL_BYTES_PINNED_BATCH);
2846                 if (cache->ro) {
2847                         space_info->bytes_readonly += len;
2848                         readonly = true;
2849                 }
2850                 spin_unlock(&cache->lock);
2851                 if (!readonly && return_free_space &&
2852                     global_rsv->space_info == space_info) {
2853                         u64 to_add = len;
2854
2855                         spin_lock(&global_rsv->lock);
2856                         if (!global_rsv->full) {
2857                                 to_add = min(len, global_rsv->size -
2858                                              global_rsv->reserved);
2859                                 global_rsv->reserved += to_add;
2860                                 btrfs_space_info_update_bytes_may_use(fs_info,
2861                                                 space_info, to_add);
2862                                 if (global_rsv->reserved >= global_rsv->size)
2863                                         global_rsv->full = 1;
2864                                 len -= to_add;
2865                         }
2866                         spin_unlock(&global_rsv->lock);
2867                         /* Add to any tickets we may have */
2868                         if (len)
2869                                 btrfs_try_granting_tickets(fs_info,
2870                                                            space_info);
2871                 }
2872                 spin_unlock(&space_info->lock);
2873         }
2874
2875         if (cache)
2876                 btrfs_put_block_group(cache);
2877         return 0;
2878 }
2879
2880 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
2881 {
2882         struct btrfs_fs_info *fs_info = trans->fs_info;
2883         struct btrfs_block_group_cache *block_group, *tmp;
2884         struct list_head *deleted_bgs;
2885         struct extent_io_tree *unpin;
2886         u64 start;
2887         u64 end;
2888         int ret;
2889
2890         if (fs_info->pinned_extents == &fs_info->freed_extents[0])
2891                 unpin = &fs_info->freed_extents[1];
2892         else
2893                 unpin = &fs_info->freed_extents[0];
2894
2895         while (!trans->aborted) {
2896                 struct extent_state *cached_state = NULL;
2897
2898                 mutex_lock(&fs_info->unused_bg_unpin_mutex);
2899                 ret = find_first_extent_bit(unpin, 0, &start, &end,
2900                                             EXTENT_DIRTY, &cached_state);
2901                 if (ret) {
2902                         mutex_unlock(&fs_info->unused_bg_unpin_mutex);
2903                         break;
2904                 }
2905
2906                 if (btrfs_test_opt(fs_info, DISCARD))
2907                         ret = btrfs_discard_extent(fs_info, start,
2908                                                    end + 1 - start, NULL);
2909
2910                 clear_extent_dirty(unpin, start, end, &cached_state);
2911                 unpin_extent_range(fs_info, start, end, true);
2912                 mutex_unlock(&fs_info->unused_bg_unpin_mutex);
2913                 free_extent_state(cached_state);
2914                 cond_resched();
2915         }
2916
2917         /*
2918          * Transaction is finished.  We don't need the lock anymore.  We
2919          * do need to clean up the block groups in case of a transaction
2920          * abort.
2921          */
2922         deleted_bgs = &trans->transaction->deleted_bgs;
2923         list_for_each_entry_safe(block_group, tmp, deleted_bgs, bg_list) {
2924                 u64 trimmed = 0;
2925
2926                 ret = -EROFS;
2927                 if (!trans->aborted)
2928                         ret = btrfs_discard_extent(fs_info,
2929                                                    block_group->key.objectid,
2930                                                    block_group->key.offset,
2931                                                    &trimmed);
2932
2933                 list_del_init(&block_group->bg_list);
2934                 btrfs_put_block_group_trimming(block_group);
2935                 btrfs_put_block_group(block_group);
2936
2937                 if (ret) {
2938                         const char *errstr = btrfs_decode_error(ret);
2939                         btrfs_warn(fs_info,
2940                            "discard failed while removing blockgroup: errno=%d %s",
2941                                    ret, errstr);
2942                 }
2943         }
2944
2945         return 0;
2946 }
2947
2948 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
2949                                struct btrfs_delayed_ref_node *node, u64 parent,
2950                                u64 root_objectid, u64 owner_objectid,
2951                                u64 owner_offset, int refs_to_drop,
2952                                struct btrfs_delayed_extent_op *extent_op)
2953 {
2954         struct btrfs_fs_info *info = trans->fs_info;
2955         struct btrfs_key key;
2956         struct btrfs_path *path;
2957         struct btrfs_root *extent_root = info->extent_root;
2958         struct extent_buffer *leaf;
2959         struct btrfs_extent_item *ei;
2960         struct btrfs_extent_inline_ref *iref;
2961         int ret;
2962         int is_data;
2963         int extent_slot = 0;
2964         int found_extent = 0;
2965         int num_to_del = 1;
2966         u32 item_size;
2967         u64 refs;
2968         u64 bytenr = node->bytenr;
2969         u64 num_bytes = node->num_bytes;
2970         int last_ref = 0;
2971         bool skinny_metadata = btrfs_fs_incompat(info, SKINNY_METADATA);
2972
2973         path = btrfs_alloc_path();
2974         if (!path)
2975                 return -ENOMEM;
2976
2977         path->reada = READA_FORWARD;
2978         path->leave_spinning = 1;
2979
2980         is_data = owner_objectid >= BTRFS_FIRST_FREE_OBJECTID;
2981         BUG_ON(!is_data && refs_to_drop != 1);
2982
2983         if (is_data)
2984                 skinny_metadata = false;
2985
2986         ret = lookup_extent_backref(trans, path, &iref, bytenr, num_bytes,
2987                                     parent, root_objectid, owner_objectid,
2988                                     owner_offset);
2989         if (ret == 0) {
2990                 extent_slot = path->slots[0];
2991                 while (extent_slot >= 0) {
2992                         btrfs_item_key_to_cpu(path->nodes[0], &key,
2993                                               extent_slot);
2994                         if (key.objectid != bytenr)
2995                                 break;
2996                         if (key.type == BTRFS_EXTENT_ITEM_KEY &&
2997                             key.offset == num_bytes) {
2998                                 found_extent = 1;
2999                                 break;
3000                         }
3001                         if (key.type == BTRFS_METADATA_ITEM_KEY &&
3002                             key.offset == owner_objectid) {
3003                                 found_extent = 1;
3004                                 break;
3005                         }
3006                         if (path->slots[0] - extent_slot > 5)
3007                                 break;
3008                         extent_slot--;
3009                 }
3010
3011                 if (!found_extent) {
3012                         BUG_ON(iref);
3013                         ret = remove_extent_backref(trans, path, NULL,
3014                                                     refs_to_drop,
3015                                                     is_data, &last_ref);
3016                         if (ret) {
3017                                 btrfs_abort_transaction(trans, ret);
3018                                 goto out;
3019                         }
3020                         btrfs_release_path(path);
3021                         path->leave_spinning = 1;
3022
3023                         key.objectid = bytenr;
3024                         key.type = BTRFS_EXTENT_ITEM_KEY;
3025                         key.offset = num_bytes;
3026
3027                         if (!is_data && skinny_metadata) {
3028                                 key.type = BTRFS_METADATA_ITEM_KEY;
3029                                 key.offset = owner_objectid;
3030                         }
3031
3032                         ret = btrfs_search_slot(trans, extent_root,
3033                                                 &key, path, -1, 1);
3034                         if (ret > 0 && skinny_metadata && path->slots[0]) {
3035                                 /*
3036                                  * Couldn't find our skinny metadata item,
3037                                  * see if we have ye olde extent item.
3038                                  */
3039                                 path->slots[0]--;
3040                                 btrfs_item_key_to_cpu(path->nodes[0], &key,
3041                                                       path->slots[0]);
3042                                 if (key.objectid == bytenr &&
3043                                     key.type == BTRFS_EXTENT_ITEM_KEY &&
3044                                     key.offset == num_bytes)
3045                                         ret = 0;
3046                         }
3047
3048                         if (ret > 0 && skinny_metadata) {
3049                                 skinny_metadata = false;
3050                                 key.objectid = bytenr;
3051                                 key.type = BTRFS_EXTENT_ITEM_KEY;
3052                                 key.offset = num_bytes;
3053                                 btrfs_release_path(path);
3054                                 ret = btrfs_search_slot(trans, extent_root,
3055                                                         &key, path, -1, 1);
3056                         }
3057
3058                         if (ret) {
3059                                 btrfs_err(info,
3060                                           "umm, got %d back from search, was looking for %llu",
3061                                           ret, bytenr);
3062                                 if (ret > 0)
3063                                         btrfs_print_leaf(path->nodes[0]);
3064                         }
3065                         if (ret < 0) {
3066                                 btrfs_abort_transaction(trans, ret);
3067                                 goto out;
3068                         }
3069                         extent_slot = path->slots[0];
3070                 }
3071         } else if (WARN_ON(ret == -ENOENT)) {
3072                 btrfs_print_leaf(path->nodes[0]);
3073                 btrfs_err(info,
3074                         "unable to find ref byte nr %llu parent %llu root %llu  owner %llu offset %llu",
3075                         bytenr, parent, root_objectid, owner_objectid,
3076                         owner_offset);
3077                 btrfs_abort_transaction(trans, ret);
3078                 goto out;
3079         } else {
3080                 btrfs_abort_transaction(trans, ret);
3081                 goto out;
3082         }
3083
3084         leaf = path->nodes[0];
3085         item_size = btrfs_item_size_nr(leaf, extent_slot);
3086         if (unlikely(item_size < sizeof(*ei))) {
3087                 ret = -EINVAL;
3088                 btrfs_print_v0_err(info);
3089                 btrfs_abort_transaction(trans, ret);
3090                 goto out;
3091         }
3092         ei = btrfs_item_ptr(leaf, extent_slot,
3093                             struct btrfs_extent_item);
3094         if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID &&
3095             key.type == BTRFS_EXTENT_ITEM_KEY) {
3096                 struct btrfs_tree_block_info *bi;
3097                 BUG_ON(item_size < sizeof(*ei) + sizeof(*bi));
3098                 bi = (struct btrfs_tree_block_info *)(ei + 1);
3099                 WARN_ON(owner_objectid != btrfs_tree_block_level(leaf, bi));
3100         }
3101
3102         refs = btrfs_extent_refs(leaf, ei);
3103         if (refs < refs_to_drop) {
3104                 btrfs_err(info,
3105                           "trying to drop %d refs but we only have %Lu for bytenr %Lu",
3106                           refs_to_drop, refs, bytenr);
3107                 ret = -EINVAL;
3108                 btrfs_abort_transaction(trans, ret);
3109                 goto out;
3110         }
3111         refs -= refs_to_drop;
3112
3113         if (refs > 0) {
3114                 if (extent_op)
3115                         __run_delayed_extent_op(extent_op, leaf, ei);
3116                 /*
3117                  * In the case of inline back ref, reference count will
3118                  * be updated by remove_extent_backref
3119                  */
3120                 if (iref) {
3121                         BUG_ON(!found_extent);
3122                 } else {
3123                         btrfs_set_extent_refs(leaf, ei, refs);
3124                         btrfs_mark_buffer_dirty(leaf);
3125                 }
3126                 if (found_extent) {
3127                         ret = remove_extent_backref(trans, path, iref,
3128                                                     refs_to_drop, is_data,
3129                                                     &last_ref);
3130                         if (ret) {
3131                                 btrfs_abort_transaction(trans, ret);
3132                                 goto out;
3133                         }
3134                 }
3135         } else {
3136                 if (found_extent) {
3137                         BUG_ON(is_data && refs_to_drop !=
3138                                extent_data_ref_count(path, iref));
3139                         if (iref) {
3140                                 BUG_ON(path->slots[0] != extent_slot);
3141                         } else {
3142                                 BUG_ON(path->slots[0] != extent_slot + 1);
3143                                 path->slots[0] = extent_slot;
3144                                 num_to_del = 2;
3145                         }
3146                 }
3147
3148                 last_ref = 1;
3149                 ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
3150                                       num_to_del);
3151                 if (ret) {
3152                         btrfs_abort_transaction(trans, ret);
3153                         goto out;
3154                 }
3155                 btrfs_release_path(path);
3156
3157                 if (is_data) {
3158                         ret = btrfs_del_csums(trans, info, bytenr, num_bytes);
3159                         if (ret) {
3160                                 btrfs_abort_transaction(trans, ret);
3161                                 goto out;
3162                         }
3163                 }
3164
3165                 ret = add_to_free_space_tree(trans, bytenr, num_bytes);
3166                 if (ret) {
3167                         btrfs_abort_transaction(trans, ret);
3168                         goto out;
3169                 }
3170
3171                 ret = btrfs_update_block_group(trans, bytenr, num_bytes, 0);
3172                 if (ret) {
3173                         btrfs_abort_transaction(trans, ret);
3174                         goto out;
3175                 }
3176         }
3177         btrfs_release_path(path);
3178
3179 out:
3180         btrfs_free_path(path);
3181         return ret;
3182 }
3183
3184 /*
3185  * when we free an block, it is possible (and likely) that we free the last
3186  * delayed ref for that extent as well.  This searches the delayed ref tree for
3187  * a given extent, and if there are no other delayed refs to be processed, it
3188  * removes it from the tree.
3189  */
3190 static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans,
3191                                       u64 bytenr)
3192 {
3193         struct btrfs_delayed_ref_head *head;
3194         struct btrfs_delayed_ref_root *delayed_refs;
3195         int ret = 0;
3196
3197         delayed_refs = &trans->transaction->delayed_refs;
3198         spin_lock(&delayed_refs->lock);
3199         head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
3200         if (!head)
3201                 goto out_delayed_unlock;
3202
3203         spin_lock(&head->lock);
3204         if (!RB_EMPTY_ROOT(&head->ref_tree.rb_root))
3205                 goto out;
3206
3207         if (cleanup_extent_op(head) != NULL)
3208                 goto out;
3209
3210         /*
3211          * waiting for the lock here would deadlock.  If someone else has it
3212          * locked they are already in the process of dropping it anyway
3213          */
3214         if (!mutex_trylock(&head->mutex))
3215                 goto out;
3216
3217         btrfs_delete_ref_head(delayed_refs, head);
3218         head->processing = 0;
3219
3220         spin_unlock(&head->lock);
3221         spin_unlock(&delayed_refs->lock);
3222
3223         BUG_ON(head->extent_op);
3224         if (head->must_insert_reserved)
3225                 ret = 1;
3226
3227         btrfs_cleanup_ref_head_accounting(trans->fs_info, delayed_refs, head);
3228         mutex_unlock(&head->mutex);
3229         btrfs_put_delayed_ref_head(head);
3230         return ret;
3231 out:
3232         spin_unlock(&head->lock);
3233
3234 out_delayed_unlock:
3235         spin_unlock(&delayed_refs->lock);
3236         return 0;
3237 }
3238
3239 void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
3240                            struct btrfs_root *root,
3241                            struct extent_buffer *buf,
3242                            u64 parent, int last_ref)
3243 {
3244         struct btrfs_fs_info *fs_info = root->fs_info;
3245         struct btrfs_ref generic_ref = { 0 };
3246         int pin = 1;
3247         int ret;
3248
3249         btrfs_init_generic_ref(&generic_ref, BTRFS_DROP_DELAYED_REF,
3250                                buf->start, buf->len, parent);
3251         btrfs_init_tree_ref(&generic_ref, btrfs_header_level(buf),
3252                             root->root_key.objectid);
3253
3254         if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
3255                 int old_ref_mod, new_ref_mod;
3256
3257                 btrfs_ref_tree_mod(fs_info, &generic_ref);
3258                 ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, NULL,
3259                                                  &old_ref_mod, &new_ref_mod);
3260                 BUG_ON(ret); /* -ENOMEM */
3261                 pin = old_ref_mod >= 0 && new_ref_mod < 0;
3262         }
3263
3264         if (last_ref && btrfs_header_generation(buf) == trans->transid) {
3265                 struct btrfs_block_group_cache *cache;
3266
3267                 if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
3268                         ret = check_ref_cleanup(trans, buf->start);
3269                         if (!ret)
3270                                 goto out;
3271                 }
3272
3273                 pin = 0;
3274                 cache = btrfs_lookup_block_group(fs_info, buf->start);
3275
3276                 if (btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
3277                         pin_down_extent(cache, buf->start, buf->len, 1);
3278                         btrfs_put_block_group(cache);
3279                         goto out;
3280                 }
3281
3282                 WARN_ON(test_bit(EXTENT_BUFFER_DIRTY, &buf->bflags));
3283
3284                 btrfs_add_free_space(cache, buf->start, buf->len);
3285                 btrfs_free_reserved_bytes(cache, buf->len, 0);
3286                 btrfs_put_block_group(cache);
3287                 trace_btrfs_reserved_extent_free(fs_info, buf->start, buf->len);
3288         }
3289 out:
3290         if (pin)
3291                 add_pinned_bytes(fs_info, &generic_ref);
3292
3293         if (last_ref) {
3294                 /*
3295                  * Deleting the buffer, clear the corrupt flag since it doesn't
3296                  * matter anymore.
3297                  */
3298                 clear_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags);
3299         }
3300 }
3301
3302 /* Can return -ENOMEM */
3303 int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_ref *ref)
3304 {
3305         struct btrfs_fs_info *fs_info = trans->fs_info;
3306         int old_ref_mod, new_ref_mod;
3307         int ret;
3308
3309         if (btrfs_is_testing(fs_info))
3310                 return 0;
3311
3312         /*
3313          * tree log blocks never actually go into the extent allocation
3314          * tree, just update pinning info and exit early.
3315          */
3316         if ((ref->type == BTRFS_REF_METADATA &&
3317              ref->tree_ref.root == BTRFS_TREE_LOG_OBJECTID) ||
3318             (ref->type == BTRFS_REF_DATA &&
3319              ref->data_ref.ref_root == BTRFS_TREE_LOG_OBJECTID)) {
3320                 /* unlocks the pinned mutex */
3321                 btrfs_pin_extent(fs_info, ref->bytenr, ref->len, 1);
3322                 old_ref_mod = new_ref_mod = 0;
3323                 ret = 0;
3324         } else if (ref->type == BTRFS_REF_METADATA) {
3325                 ret = btrfs_add_delayed_tree_ref(trans, ref, NULL,
3326                                                  &old_ref_mod, &new_ref_mod);
3327         } else {
3328                 ret = btrfs_add_delayed_data_ref(trans, ref, 0,
3329                                                  &old_ref_mod, &new_ref_mod);
3330         }
3331
3332         if (!((ref->type == BTRFS_REF_METADATA &&
3333                ref->tree_ref.root == BTRFS_TREE_LOG_OBJECTID) ||
3334               (ref->type == BTRFS_REF_DATA &&
3335                ref->data_ref.ref_root == BTRFS_TREE_LOG_OBJECTID)))
3336                 btrfs_ref_tree_mod(fs_info, ref);
3337
3338         if (ret == 0 && old_ref_mod >= 0 && new_ref_mod < 0)
3339                 add_pinned_bytes(fs_info, ref);
3340
3341         return ret;
3342 }
3343
3344 enum btrfs_loop_type {
3345         LOOP_CACHING_NOWAIT,
3346         LOOP_CACHING_WAIT,
3347         LOOP_ALLOC_CHUNK,
3348         LOOP_NO_EMPTY_SIZE,
3349 };
3350
3351 static inline void
3352 btrfs_lock_block_group(struct btrfs_block_group_cache *cache,
3353                        int delalloc)
3354 {
3355         if (delalloc)
3356                 down_read(&cache->data_rwsem);
3357 }
3358
3359 static inline void
3360 btrfs_grab_block_group(struct btrfs_block_group_cache *cache,
3361                        int delalloc)
3362 {
3363         btrfs_get_block_group(cache);
3364         if (delalloc)
3365                 down_read(&cache->data_rwsem);
3366 }
3367
3368 static struct btrfs_block_group_cache *
3369 btrfs_lock_cluster(struct btrfs_block_group_cache *block_group,
3370                    struct btrfs_free_cluster *cluster,
3371                    int delalloc)
3372 {
3373         struct btrfs_block_group_cache *used_bg = NULL;
3374
3375         spin_lock(&cluster->refill_lock);
3376         while (1) {
3377                 used_bg = cluster->block_group;
3378                 if (!used_bg)
3379                         return NULL;
3380
3381                 if (used_bg == block_group)
3382                         return used_bg;
3383
3384                 btrfs_get_block_group(used_bg);
3385
3386                 if (!delalloc)
3387                         return used_bg;
3388
3389                 if (down_read_trylock(&used_bg->data_rwsem))
3390                         return used_bg;
3391
3392                 spin_unlock(&cluster->refill_lock);
3393
3394                 /* We should only have one-level nested. */
3395                 down_read_nested(&used_bg->data_rwsem, SINGLE_DEPTH_NESTING);
3396
3397                 spin_lock(&cluster->refill_lock);
3398                 if (used_bg == cluster->block_group)
3399                         return used_bg;
3400
3401                 up_read(&used_bg->data_rwsem);
3402                 btrfs_put_block_group(used_bg);
3403         }
3404 }
3405
3406 static inline void
3407 btrfs_release_block_group(struct btrfs_block_group_cache *cache,
3408                          int delalloc)
3409 {
3410         if (delalloc)
3411                 up_read(&cache->data_rwsem);
3412         btrfs_put_block_group(cache);
3413 }
3414
3415 /*
3416  * Structure used internally for find_free_extent() function.  Wraps needed
3417  * parameters.
3418  */
3419 struct find_free_extent_ctl {
3420         /* Basic allocation info */
3421         u64 ram_bytes;
3422         u64 num_bytes;
3423         u64 empty_size;
3424         u64 flags;
3425         int delalloc;
3426
3427         /* Where to start the search inside the bg */
3428         u64 search_start;
3429
3430         /* For clustered allocation */
3431         u64 empty_cluster;
3432
3433         bool have_caching_bg;
3434         bool orig_have_caching_bg;
3435
3436         /* RAID index, converted from flags */
3437         int index;
3438
3439         /*
3440          * Current loop number, check find_free_extent_update_loop() for details
3441          */
3442         int loop;
3443
3444         /*
3445          * Whether we're refilling a cluster, if true we need to re-search
3446          * current block group but don't try to refill the cluster again.
3447          */
3448         bool retry_clustered;
3449
3450         /*
3451          * Whether we're updating free space cache, if true we need to re-search
3452          * current block group but don't try updating free space cache again.
3453          */
3454         bool retry_unclustered;
3455
3456         /* If current block group is cached */
3457         int cached;
3458
3459         /* Max contiguous hole found */
3460         u64 max_extent_size;
3461
3462         /* Total free space from free space cache, not always contiguous */
3463         u64 total_free_space;
3464
3465         /* Found result */
3466         u64 found_offset;
3467 };
3468
3469
3470 /*
3471  * Helper function for find_free_extent().
3472  *
3473  * Return -ENOENT to inform caller that we need fallback to unclustered mode.
3474  * Return -EAGAIN to inform caller that we need to re-search this block group
3475  * Return >0 to inform caller that we find nothing
3476  * Return 0 means we have found a location and set ffe_ctl->found_offset.
3477  */
3478 static int find_free_extent_clustered(struct btrfs_block_group_cache *bg,
3479                 struct btrfs_free_cluster *last_ptr,
3480                 struct find_free_extent_ctl *ffe_ctl,
3481                 struct btrfs_block_group_cache **cluster_bg_ret)
3482 {
3483         struct btrfs_block_group_cache *cluster_bg;
3484         u64 aligned_cluster;
3485         u64 offset;
3486         int ret;
3487
3488         cluster_bg = btrfs_lock_cluster(bg, last_ptr, ffe_ctl->delalloc);
3489         if (!cluster_bg)
3490                 goto refill_cluster;
3491         if (cluster_bg != bg && (cluster_bg->ro ||
3492             !block_group_bits(cluster_bg, ffe_ctl->flags)))
3493                 goto release_cluster;
3494
3495         offset = btrfs_alloc_from_cluster(cluster_bg, last_ptr,
3496                         ffe_ctl->num_bytes, cluster_bg->key.objectid,
3497                         &ffe_ctl->max_extent_size);
3498         if (offset) {
3499                 /* We have a block, we're done */
3500                 spin_unlock(&last_ptr->refill_lock);
3501                 trace_btrfs_reserve_extent_cluster(cluster_bg,
3502                                 ffe_ctl->search_start, ffe_ctl->num_bytes);
3503                 *cluster_bg_ret = cluster_bg;
3504                 ffe_ctl->found_offset = offset;
3505                 return 0;
3506         }
3507         WARN_ON(last_ptr->block_group != cluster_bg);
3508
3509 release_cluster:
3510         /*
3511          * If we are on LOOP_NO_EMPTY_SIZE, we can't set up a new clusters, so
3512          * lets just skip it and let the allocator find whatever block it can
3513          * find. If we reach this point, we will have tried the cluster
3514          * allocator plenty of times and not have found anything, so we are
3515          * likely way too fragmented for the clustering stuff to find anything.
3516          *
3517          * However, if the cluster is taken from the current block group,
3518          * release the cluster first, so that we stand a better chance of
3519          * succeeding in the unclustered allocation.
3520          */
3521         if (ffe_ctl->loop >= LOOP_NO_EMPTY_SIZE && cluster_bg != bg) {
3522                 spin_unlock(&last_ptr->refill_lock);
3523                 btrfs_release_block_group(cluster_bg, ffe_ctl->delalloc);
3524                 return -ENOENT;
3525         }
3526
3527         /* This cluster didn't work out, free it and start over */
3528         btrfs_return_cluster_to_free_space(NULL, last_ptr);
3529
3530         if (cluster_bg != bg)
3531                 btrfs_release_block_group(cluster_bg, ffe_ctl->delalloc);
3532
3533 refill_cluster:
3534         if (ffe_ctl->loop >= LOOP_NO_EMPTY_SIZE) {
3535                 spin_unlock(&last_ptr->refill_lock);
3536                 return -ENOENT;
3537         }
3538
3539         aligned_cluster = max_t(u64,
3540                         ffe_ctl->empty_cluster + ffe_ctl->empty_size,
3541                         bg->full_stripe_len);
3542         ret = btrfs_find_space_cluster(bg, last_ptr, ffe_ctl->search_start,
3543                         ffe_ctl->num_bytes, aligned_cluster);
3544         if (ret == 0) {
3545                 /* Now pull our allocation out of this cluster */
3546                 offset = btrfs_alloc_from_cluster(bg, last_ptr,
3547                                 ffe_ctl->num_bytes, ffe_ctl->search_start,
3548                                 &ffe_ctl->max_extent_size);
3549                 if (offset) {
3550                         /* We found one, proceed */
3551                         spin_unlock(&last_ptr->refill_lock);
3552                         trace_btrfs_reserve_extent_cluster(bg,
3553                                         ffe_ctl->search_start,
3554                                         ffe_ctl->num_bytes);
3555                         ffe_ctl->found_offset = offset;
3556                         return 0;
3557                 }
3558         } else if (!ffe_ctl->cached && ffe_ctl->loop > LOOP_CACHING_NOWAIT &&
3559                    !ffe_ctl->retry_clustered) {
3560                 spin_unlock(&last_ptr->refill_lock);
3561
3562                 ffe_ctl->retry_clustered = true;
3563                 btrfs_wait_block_group_cache_progress(bg, ffe_ctl->num_bytes +
3564                                 ffe_ctl->empty_cluster + ffe_ctl->empty_size);
3565                 return -EAGAIN;
3566         }
3567         /*
3568          * At this point we either didn't find a cluster or we weren't able to
3569          * allocate a block from our cluster.  Free the cluster we've been
3570          * trying to use, and go to the next block group.
3571          */
3572         btrfs_return_cluster_to_free_space(NULL, last_ptr);
3573         spin_unlock(&last_ptr->refill_lock);
3574         return 1;
3575 }
3576
3577 /*
3578  * Return >0 to inform caller that we find nothing
3579  * Return 0 when we found an free extent and set ffe_ctrl->found_offset
3580  * Return -EAGAIN to inform caller that we need to re-search this block group
3581  */
3582 static int find_free_extent_unclustered(struct btrfs_block_group_cache *bg,
3583                 struct btrfs_free_cluster *last_ptr,
3584                 struct find_free_extent_ctl *ffe_ctl)
3585 {
3586         u64 offset;
3587
3588         /*
3589          * We are doing an unclustered allocation, set the fragmented flag so
3590          * we don't bother trying to setup a cluster again until we get more
3591          * space.
3592          */
3593         if (unlikely(last_ptr)) {
3594                 spin_lock(&last_ptr->lock);
3595                 last_ptr->fragmented = 1;
3596                 spin_unlock(&last_ptr->lock);
3597         }
3598         if (ffe_ctl->cached) {
3599                 struct btrfs_free_space_ctl *free_space_ctl;
3600
3601                 free_space_ctl = bg->free_space_ctl;
3602                 spin_lock(&free_space_ctl->tree_lock);
3603                 if (free_space_ctl->free_space <
3604                     ffe_ctl->num_bytes + ffe_ctl->empty_cluster +
3605                     ffe_ctl->empty_size) {
3606                         ffe_ctl->total_free_space = max_t(u64,
3607                                         ffe_ctl->total_free_space,
3608                                         free_space_ctl->free_space);
3609                         spin_unlock(&free_space_ctl->tree_lock);
3610                         return 1;
3611                 }
3612                 spin_unlock(&free_space_ctl->tree_lock);
3613         }
3614
3615         offset = btrfs_find_space_for_alloc(bg, ffe_ctl->search_start,
3616                         ffe_ctl->num_bytes, ffe_ctl->empty_size,
3617                         &ffe_ctl->max_extent_size);
3618
3619         /*
3620          * If we didn't find a chunk, and we haven't failed on this block group
3621          * before, and this block group is in the middle of caching and we are
3622          * ok with waiting, then go ahead and wait for progress to be made, and
3623          * set @retry_unclustered to true.
3624          *
3625          * If @retry_unclustered is true then we've already waited on this
3626          * block group once and should move on to the next block group.
3627          */
3628         if (!offset && !ffe_ctl->retry_unclustered && !ffe_ctl->cached &&
3629             ffe_ctl->loop > LOOP_CACHING_NOWAIT) {
3630                 btrfs_wait_block_group_cache_progress(bg, ffe_ctl->num_bytes +
3631                                                       ffe_ctl->empty_size);
3632                 ffe_ctl->retry_unclustered = true;
3633                 return -EAGAIN;
3634         } else if (!offset) {
3635                 return 1;
3636         }
3637         ffe_ctl->found_offset = offset;
3638         return 0;
3639 }
3640
3641 /*
3642  * Return >0 means caller needs to re-search for free extent
3643  * Return 0 means we have the needed free extent.
3644  * Return <0 means we failed to locate any free extent.
3645  */
3646 static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
3647                                         struct btrfs_free_cluster *last_ptr,
3648                                         struct btrfs_key *ins,
3649                                         struct find_free_extent_ctl *ffe_ctl,
3650                                         int full_search, bool use_cluster)
3651 {
3652         struct btrfs_root *root = fs_info->extent_root;
3653         int ret;
3654
3655         if ((ffe_ctl->loop == LOOP_CACHING_NOWAIT) &&
3656             ffe_ctl->have_caching_bg && !ffe_ctl->orig_have_caching_bg)
3657                 ffe_ctl->orig_have_caching_bg = true;
3658
3659         if (!ins->objectid && ffe_ctl->loop >= LOOP_CACHING_WAIT &&
3660             ffe_ctl->have_caching_bg)
3661                 return 1;
3662
3663         if (!ins->objectid && ++(ffe_ctl->index) < BTRFS_NR_RAID_TYPES)
3664                 return 1;
3665
3666         if (ins->objectid) {
3667                 if (!use_cluster && last_ptr) {
3668                         spin_lock(&last_ptr->lock);
3669                         last_ptr->window_start = ins->objectid;
3670                         spin_unlock(&last_ptr->lock);
3671                 }
3672                 return 0;
3673         }
3674
3675         /*
3676          * LOOP_CACHING_NOWAIT, search partially cached block groups, kicking
3677          *                      caching kthreads as we move along
3678          * LOOP_CACHING_WAIT, search everything, and wait if our bg is caching
3679          * LOOP_ALLOC_CHUNK, force a chunk allocation and try again
3680          * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try
3681          *                     again
3682          */
3683         if (ffe_ctl->loop < LOOP_NO_EMPTY_SIZE) {
3684                 ffe_ctl->index = 0;
3685                 if (ffe_ctl->loop == LOOP_CACHING_NOWAIT) {
3686                         /*
3687                          * We want to skip the LOOP_CACHING_WAIT step if we
3688                          * don't have any uncached bgs and we've already done a
3689                          * full search through.
3690                          */
3691                         if (ffe_ctl->orig_have_caching_bg || !full_search)
3692                                 ffe_ctl->loop = LOOP_CACHING_WAIT;
3693                         else
3694                                 ffe_ctl->loop = LOOP_ALLOC_CHUNK;
3695                 } else {
3696                         ffe_ctl->loop++;
3697                 }
3698
3699                 if (ffe_ctl->loop == LOOP_ALLOC_CHUNK) {
3700                         struct btrfs_trans_handle *trans;
3701                         int exist = 0;
3702
3703                         trans = current->journal_info;
3704                         if (trans)
3705                                 exist = 1;
3706                         else
3707                                 trans = btrfs_join_transaction(root);
3708
3709                         if (IS_ERR(trans)) {
3710                                 ret = PTR_ERR(trans);
3711                                 return ret;
3712                         }
3713
3714                         ret = btrfs_chunk_alloc(trans, ffe_ctl->flags,
3715                                                 CHUNK_ALLOC_FORCE);
3716
3717                         /*
3718                          * If we can't allocate a new chunk we've already looped
3719                          * through at least once, move on to the NO_EMPTY_SIZE
3720                          * case.
3721                          */
3722                         if (ret == -ENOSPC)
3723                                 ffe_ctl->loop = LOOP_NO_EMPTY_SIZE;
3724
3725                         /* Do not bail out on ENOSPC since we can do more. */
3726                         if (ret < 0 && ret != -ENOSPC)
3727                                 btrfs_abort_transaction(trans, ret);
3728                         else
3729                                 ret = 0;
3730                         if (!exist)
3731                                 btrfs_end_transaction(trans);
3732                         if (ret)
3733                                 return ret;
3734                 }
3735
3736                 if (ffe_ctl->loop == LOOP_NO_EMPTY_SIZE) {
3737                         /*
3738                          * Don't loop again if we already have no empty_size and
3739                          * no empty_cluster.
3740                          */
3741                         if (ffe_ctl->empty_size == 0 &&
3742                             ffe_ctl->empty_cluster == 0)
3743                                 return -ENOSPC;
3744                         ffe_ctl->empty_size = 0;
3745                         ffe_ctl->empty_cluster = 0;
3746                 }
3747                 return 1;
3748         }
3749         return -ENOSPC;
3750 }
3751
3752 /*
3753  * walks the btree of allocated extents and find a hole of a given size.
3754  * The key ins is changed to record the hole:
3755  * ins->objectid == start position
3756  * ins->flags = BTRFS_EXTENT_ITEM_KEY
3757  * ins->offset == the size of the hole.
3758  * Any available blocks before search_start are skipped.
3759  *
3760  * If there is no suitable free space, we will record the max size of
3761  * the free space extent currently.
3762  *
3763  * The overall logic and call chain:
3764  *
3765  * find_free_extent()
3766  * |- Iterate through all block groups
3767  * |  |- Get a valid block group
3768  * |  |- Try to do clustered allocation in that block group
3769  * |  |- Try to do unclustered allocation in that block group
3770  * |  |- Check if the result is valid
3771  * |  |  |- If valid, then exit
3772  * |  |- Jump to next block group
3773  * |
3774  * |- Push harder to find free extents
3775  *    |- If not found, re-iterate all block groups
3776  */
3777 static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
3778                                 u64 ram_bytes, u64 num_bytes, u64 empty_size,
3779                                 u64 hint_byte, struct btrfs_key *ins,
3780                                 u64 flags, int delalloc)
3781 {
3782         int ret = 0;
3783         struct btrfs_free_cluster *last_ptr = NULL;
3784         struct btrfs_block_group_cache *block_group = NULL;
3785         struct find_free_extent_ctl ffe_ctl = {0};
3786         struct btrfs_space_info *space_info;
3787         bool use_cluster = true;
3788         bool full_search = false;
3789
3790         WARN_ON(num_bytes < fs_info->sectorsize);
3791
3792         ffe_ctl.ram_bytes = ram_bytes;
3793         ffe_ctl.num_bytes = num_bytes;
3794         ffe_ctl.empty_size = empty_size;
3795         ffe_ctl.flags = flags;
3796         ffe_ctl.search_start = 0;
3797         ffe_ctl.retry_clustered = false;
3798         ffe_ctl.retry_unclustered = false;
3799         ffe_ctl.delalloc = delalloc;
3800         ffe_ctl.index = btrfs_bg_flags_to_raid_index(flags);
3801         ffe_ctl.have_caching_bg = false;
3802         ffe_ctl.orig_have_caching_bg = false;
3803         ffe_ctl.found_offset = 0;
3804
3805         ins->type = BTRFS_EXTENT_ITEM_KEY;
3806         ins->objectid = 0;
3807         ins->offset = 0;
3808
3809         trace_find_free_extent(fs_info, num_bytes, empty_size, flags);
3810
3811         space_info = btrfs_find_space_info(fs_info, flags);
3812         if (!space_info) {
3813                 btrfs_err(fs_info, "No space info for %llu", flags);
3814                 return -ENOSPC;
3815         }
3816
3817         /*
3818          * If our free space is heavily fragmented we may not be able to make
3819          * big contiguous allocations, so instead of doing the expensive search
3820          * for free space, simply return ENOSPC with our max_extent_size so we
3821          * can go ahead and search for a more manageable chunk.
3822          *
3823          * If our max_extent_size is large enough for our allocation simply
3824          * disable clustering since we will likely not be able to find enough
3825          * space to create a cluster and induce latency trying.
3826          */
3827         if (unlikely(space_info->max_extent_size)) {
3828                 spin_lock(&space_info->lock);
3829                 if (space_info->max_extent_size &&
3830                     num_bytes > space_info->max_extent_size) {
3831                         ins->offset = space_info->max_extent_size;
3832                         spin_unlock(&space_info->lock);
3833                         return -ENOSPC;
3834                 } else if (space_info->max_extent_size) {
3835                         use_cluster = false;
3836                 }
3837                 spin_unlock(&space_info->lock);
3838         }
3839
3840         last_ptr = fetch_cluster_info(fs_info, space_info,
3841                                       &ffe_ctl.empty_cluster);
3842         if (last_ptr) {
3843                 spin_lock(&last_ptr->lock);
3844                 if (last_ptr->block_group)
3845                         hint_byte = last_ptr->window_start;
3846                 if (last_ptr->fragmented) {
3847                         /*
3848                          * We still set window_start so we can keep track of the
3849                          * last place we found an allocation to try and save
3850                          * some time.
3851                          */
3852                         hint_byte = last_ptr->window_start;
3853                         use_cluster = false;
3854                 }
3855                 spin_unlock(&last_ptr->lock);
3856         }
3857
3858         ffe_ctl.search_start = max(ffe_ctl.search_start,
3859                                    first_logical_byte(fs_info, 0));
3860         ffe_ctl.search_start = max(ffe_ctl.search_start, hint_byte);
3861         if (ffe_ctl.search_start == hint_byte) {
3862                 block_group = btrfs_lookup_block_group(fs_info,
3863                                                        ffe_ctl.search_start);
3864                 /*
3865                  * we don't want to use the block group if it doesn't match our
3866                  * allocation bits, or if its not cached.
3867                  *
3868                  * However if we are re-searching with an ideal block group
3869                  * picked out then we don't care that the block group is cached.
3870                  */
3871                 if (block_group && block_group_bits(block_group, flags) &&
3872                     block_group->cached != BTRFS_CACHE_NO) {
3873                         down_read(&space_info->groups_sem);
3874                         if (list_empty(&block_group->list) ||
3875                             block_group->ro) {
3876                                 /*
3877                                  * someone is removing this block group,
3878                                  * we can't jump into the have_block_group
3879                                  * target because our list pointers are not
3880                                  * valid
3881                                  */
3882                                 btrfs_put_block_group(block_group);
3883                                 up_read(&space_info->groups_sem);
3884                         } else {
3885                                 ffe_ctl.index = btrfs_bg_flags_to_raid_index(
3886                                                 block_group->flags);
3887                                 btrfs_lock_block_group(block_group, delalloc);
3888                                 goto have_block_group;
3889                         }
3890                 } else if (block_group) {
3891                         btrfs_put_block_group(block_group);
3892                 }
3893         }
3894 search:
3895         ffe_ctl.have_caching_bg = false;
3896         if (ffe_ctl.index == btrfs_bg_flags_to_raid_index(flags) ||
3897             ffe_ctl.index == 0)
3898                 full_search = true;
3899         down_read(&space_info->groups_sem);
3900         list_for_each_entry(block_group,
3901                             &space_info->block_groups[ffe_ctl.index], list) {
3902                 /* If the block group is read-only, we can skip it entirely. */
3903                 if (unlikely(block_group->ro))
3904                         continue;
3905
3906                 btrfs_grab_block_group(block_group, delalloc);
3907                 ffe_ctl.search_start = block_group->key.objectid;
3908
3909                 /*
3910                  * this can happen if we end up cycling through all the
3911                  * raid types, but we want to make sure we only allocate
3912                  * for the proper type.
3913                  */
3914                 if (!block_group_bits(block_group, flags)) {
3915                         u64 extra = BTRFS_BLOCK_GROUP_DUP |
3916                                 BTRFS_BLOCK_GROUP_RAID1_MASK |
3917                                 BTRFS_BLOCK_GROUP_RAID56_MASK |
3918                                 BTRFS_BLOCK_GROUP_RAID10;
3919
3920                         /*
3921                          * if they asked for extra copies and this block group
3922                          * doesn't provide them, bail.  This does allow us to
3923                          * fill raid0 from raid1.
3924                          */
3925                         if ((flags & extra) && !(block_group->flags & extra))
3926                                 goto loop;
3927
3928                         /*
3929                          * This block group has different flags than we want.
3930                          * It's possible that we have MIXED_GROUP flag but no
3931                          * block group is mixed.  Just skip such block group.
3932                          */
3933                         btrfs_release_block_group(block_group, delalloc);
3934                         continue;
3935                 }
3936
3937 have_block_group:
3938                 ffe_ctl.cached = btrfs_block_group_cache_done(block_group);
3939                 if (unlikely(!ffe_ctl.cached)) {
3940                         ffe_ctl.have_caching_bg = true;
3941                         ret = btrfs_cache_block_group(block_group, 0);
3942                         BUG_ON(ret < 0);
3943                         ret = 0;
3944                 }
3945
3946                 if (unlikely(block_group->cached == BTRFS_CACHE_ERROR))
3947                         goto loop;
3948
3949                 /*
3950                  * Ok we want to try and use the cluster allocator, so
3951                  * lets look there
3952                  */
3953                 if (last_ptr && use_cluster) {
3954                         struct btrfs_block_group_cache *cluster_bg = NULL;
3955
3956                         ret = find_free_extent_clustered(block_group, last_ptr,
3957                                                          &ffe_ctl, &cluster_bg);
3958
3959                         if (ret == 0) {
3960                                 if (cluster_bg && cluster_bg != block_group) {
3961                                         btrfs_release_block_group(block_group,
3962                                                                   delalloc);
3963                                         block_group = cluster_bg;
3964                                 }
3965                                 goto checks;
3966                         } else if (ret == -EAGAIN) {
3967                                 goto have_block_group;
3968                         } else if (ret > 0) {
3969                                 goto loop;
3970                         }
3971                         /* ret == -ENOENT case falls through */
3972                 }
3973
3974                 ret = find_free_extent_unclustered(block_group, last_ptr,
3975                                                    &ffe_ctl);
3976                 if (ret == -EAGAIN)
3977                         goto have_block_group;
3978                 else if (ret > 0)
3979                         goto loop;
3980                 /* ret == 0 case falls through */
3981 checks:
3982                 ffe_ctl.search_start = round_up(ffe_ctl.found_offset,
3983                                              fs_info->stripesize);
3984
3985                 /* move on to the next group */
3986                 if (ffe_ctl.search_start + num_bytes >
3987                     block_group->key.objectid + block_group->key.offset) {
3988                         btrfs_add_free_space(block_group, ffe_ctl.found_offset,
3989                                              num_bytes);
3990                         goto loop;
3991                 }
3992
3993                 if (ffe_ctl.found_offset < ffe_ctl.search_start)
3994                         btrfs_add_free_space(block_group, ffe_ctl.found_offset,
3995                                 ffe_ctl.search_start - ffe_ctl.found_offset);
3996
3997                 ret = btrfs_add_reserved_bytes(block_group, ram_bytes,
3998                                 num_bytes, delalloc);
3999                 if (ret == -EAGAIN) {
4000                         btrfs_add_free_space(block_group, ffe_ctl.found_offset,
4001                                              num_bytes);
4002                         goto loop;
4003                 }
4004                 btrfs_inc_block_group_reservations(block_group);
4005
4006                 /* we are all good, lets return */
4007                 ins->objectid = ffe_ctl.search_start;
4008                 ins->offset = num_bytes;
4009
4010                 trace_btrfs_reserve_extent(block_group, ffe_ctl.search_start,
4011                                            num_bytes);
4012                 btrfs_release_block_group(block_group, delalloc);
4013                 break;
4014 loop:
4015                 ffe_ctl.retry_clustered = false;
4016                 ffe_ctl.retry_unclustered = false;
4017                 BUG_ON(btrfs_bg_flags_to_raid_index(block_group->flags) !=
4018                        ffe_ctl.index);
4019                 btrfs_release_block_group(block_group, delalloc);
4020                 cond_resched();
4021         }
4022         up_read(&space_info->groups_sem);
4023
4024         ret = find_free_extent_update_loop(fs_info, last_ptr, ins, &ffe_ctl,
4025                                            full_search, use_cluster);
4026         if (ret > 0)
4027                 goto search;
4028
4029         if (ret == -ENOSPC) {
4030                 /*
4031                  * Use ffe_ctl->total_free_space as fallback if we can't find
4032                  * any contiguous hole.
4033                  */
4034                 if (!ffe_ctl.max_extent_size)
4035                         ffe_ctl.max_extent_size = ffe_ctl.total_free_space;
4036                 spin_lock(&space_info->lock);
4037                 space_info->max_extent_size = ffe_ctl.max_extent_size;
4038                 spin_unlock(&space_info->lock);
4039                 ins->offset = ffe_ctl.max_extent_size;
4040         }
4041         return ret;
4042 }
4043
4044 /*
4045  * btrfs_reserve_extent - entry point to the extent allocator. Tries to find a
4046  *                        hole that is at least as big as @num_bytes.
4047  *
4048  * @root           -    The root that will contain this extent
4049  *
4050  * @ram_bytes      -    The amount of space in ram that @num_bytes take. This
4051  *                      is used for accounting purposes. This value differs
4052  *                      from @num_bytes only in the case of compressed extents.
4053  *
4054  * @num_bytes      -    Number of bytes to allocate on-disk.
4055  *
4056  * @min_alloc_size -    Indicates the minimum amount of space that the
4057  *                      allocator should try to satisfy. In some cases
4058  *                      @num_bytes may be larger than what is required and if
4059  *                      the filesystem is fragmented then allocation fails.
4060  *                      However, the presence of @min_alloc_size gives a
4061  *                      chance to try and satisfy the smaller allocation.
4062  *
4063  * @empty_size     -    A hint that you plan on doing more COW. This is the
4064  *                      size in bytes the allocator should try to find free
4065  *                      next to the block it returns.  This is just a hint and
4066  *                      may be ignored by the allocator.
4067  *
4068  * @hint_byte      -    Hint to the allocator to start searching above the byte
4069  *                      address passed. It might be ignored.
4070  *
4071  * @ins            -    This key is modified to record the found hole. It will
4072  *                      have the following values:
4073  *                      ins->objectid == start position
4074  *                      ins->flags = BTRFS_EXTENT_ITEM_KEY
4075  *                      ins->offset == the size of the hole.
4076  *
4077  * @is_data        -    Boolean flag indicating whether an extent is
4078  *                      allocated for data (true) or metadata (false)
4079  *
4080  * @delalloc       -    Boolean flag indicating whether this allocation is for
4081  *                      delalloc or not. If 'true' data_rwsem of block groups
4082  *                      is going to be acquired.
4083  *
4084  *
4085  * Returns 0 when an allocation succeeded or < 0 when an error occurred. In
4086  * case -ENOSPC is returned then @ins->offset will contain the size of the
4087  * largest available hole the allocator managed to find.
4088  */
4089 int btrfs_reserve_extent(struct btrfs_root *root, u64 ram_bytes,
4090                          u64 num_bytes, u64 min_alloc_size,
4091                          u64 empty_size, u64 hint_byte,
4092                          struct btrfs_key *ins, int is_data, int delalloc)
4093 {
4094         struct btrfs_fs_info *fs_info = root->fs_info;
4095         bool final_tried = num_bytes == min_alloc_size;
4096         u64 flags;
4097         int ret;
4098
4099         flags = get_alloc_profile_by_root(root, is_data);
4100 again:
4101         WARN_ON(num_bytes < fs_info->sectorsize);
4102         ret = find_free_extent(fs_info, ram_bytes, num_bytes, empty_size,
4103                                hint_byte, ins, flags, delalloc);
4104         if (!ret && !is_data) {
4105                 btrfs_dec_block_group_reservations(fs_info, ins->objectid);
4106         } else if (ret == -ENOSPC) {
4107                 if (!final_tried && ins->offset) {
4108                         num_bytes = min(num_bytes >> 1, ins->offset);
4109                         num_bytes = round_down(num_bytes,
4110                                                fs_info->sectorsize);
4111                         num_bytes = max(num_bytes, min_alloc_size);
4112                         ram_bytes = num_bytes;
4113                         if (num_bytes == min_alloc_size)
4114                                 final_tried = true;
4115                         goto again;
4116                 } else if (btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
4117                         struct btrfs_space_info *sinfo;
4118
4119                         sinfo = btrfs_find_space_info(fs_info, flags);
4120                         btrfs_err(fs_info,
4121                                   "allocation failed flags %llu, wanted %llu",
4122                                   flags, num_bytes);
4123                         if (sinfo)
4124                                 btrfs_dump_space_info(fs_info, sinfo,
4125                                                       num_bytes, 1);
4126                 }
4127         }
4128
4129         return ret;
4130 }
4131
4132 static int __btrfs_free_reserved_extent(struct btrfs_fs_info *fs_info,
4133                                         u64 start, u64 len,
4134                                         int pin, int delalloc)
4135 {
4136         struct btrfs_block_group_cache *cache;
4137         int ret = 0;
4138
4139         cache = btrfs_lookup_block_group(fs_info, start);
4140         if (!cache) {
4141                 btrfs_err(fs_info, "Unable to find block group for %llu",
4142                           start);
4143                 return -ENOSPC;
4144         }
4145
4146         if (pin)
4147                 pin_down_extent(cache, start, len, 1);
4148         else {
4149                 if (btrfs_test_opt(fs_info, DISCARD))
4150                         ret = btrfs_discard_extent(fs_info, start, len, NULL);
4151                 btrfs_add_free_space(cache, start, len);
4152                 btrfs_free_reserved_bytes(cache, len, delalloc);
4153                 trace_btrfs_reserved_extent_free(fs_info, start, len);
4154         }
4155
4156         btrfs_put_block_group(cache);
4157         return ret;
4158 }
4159
4160 int btrfs_free_reserved_extent(struct btrfs_fs_info *fs_info,
4161                                u64 start, u64 len, int delalloc)
4162 {
4163         return __btrfs_free_reserved_extent(fs_info, start, len, 0, delalloc);
4164 }
4165
4166 int btrfs_free_and_pin_reserved_extent(struct btrfs_fs_info *fs_info,
4167                                        u64 start, u64 len)
4168 {
4169         return __btrfs_free_reserved_extent(fs_info, start, len, 1, 0);
4170 }
4171
4172 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
4173                                       u64 parent, u64 root_objectid,
4174                                       u64 flags, u64 owner, u64 offset,
4175                                       struct btrfs_key *ins, int ref_mod)
4176 {
4177         struct btrfs_fs_info *fs_info = trans->fs_info;
4178         int ret;
4179         struct btrfs_extent_item *extent_item;
4180         struct btrfs_extent_inline_ref *iref;
4181         struct btrfs_path *path;
4182         struct extent_buffer *leaf;
4183         int type;
4184         u32 size;
4185
4186         if (parent > 0)
4187                 type = BTRFS_SHARED_DATA_REF_KEY;
4188         else
4189                 type = BTRFS_EXTENT_DATA_REF_KEY;
4190
4191         size = sizeof(*extent_item) + btrfs_extent_inline_ref_size(type);
4192
4193         path = btrfs_alloc_path();
4194         if (!path)
4195                 return -ENOMEM;
4196
4197         path->leave_spinning = 1;
4198         ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path,
4199                                       ins, size);
4200         if (ret) {
4201                 btrfs_free_path(path);
4202                 return ret;
4203         }
4204
4205         leaf = path->nodes[0];
4206         extent_item = btrfs_item_ptr(leaf, path->slots[0],
4207                                      struct btrfs_extent_item);
4208         btrfs_set_extent_refs(leaf, extent_item, ref_mod);
4209         btrfs_set_extent_generation(leaf, extent_item, trans->transid);
4210         btrfs_set_extent_flags(leaf, extent_item,
4211                                flags | BTRFS_EXTENT_FLAG_DATA);
4212
4213         iref = (struct btrfs_extent_inline_ref *)(extent_item + 1);
4214         btrfs_set_extent_inline_ref_type(leaf, iref, type);
4215         if (parent > 0) {
4216                 struct btrfs_shared_data_ref *ref;
4217                 ref = (struct btrfs_shared_data_ref *)(iref + 1);
4218                 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
4219                 btrfs_set_shared_data_ref_count(leaf, ref, ref_mod);
4220         } else {
4221                 struct btrfs_extent_data_ref *ref;
4222                 ref = (struct btrfs_extent_data_ref *)(&iref->offset);
4223                 btrfs_set_extent_data_ref_root(leaf, ref, root_objectid);
4224                 btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
4225                 btrfs_set_extent_data_ref_offset(leaf, ref, offset);
4226                 btrfs_set_extent_data_ref_count(leaf, ref, ref_mod);
4227         }
4228
4229         btrfs_mark_buffer_dirty(path->nodes[0]);
4230         btrfs_free_path(path);
4231
4232         ret = remove_from_free_space_tree(trans, ins->objectid, ins->offset);
4233         if (ret)
4234                 return ret;
4235
4236         ret = btrfs_update_block_group(trans, ins->objectid, ins->offset, 1);
4237         if (ret) { /* -ENOENT, logic error */
4238                 btrfs_err(fs_info, "update block group failed for %llu %llu",
4239                         ins->objectid, ins->offset);
4240                 BUG();
4241         }
4242         trace_btrfs_reserved_extent_alloc(fs_info, ins->objectid, ins->offset);
4243         return ret;
4244 }
4245
4246 static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
4247                                      struct btrfs_delayed_ref_node *node,
4248                                      struct btrfs_delayed_extent_op *extent_op)
4249 {
4250         struct btrfs_fs_info *fs_info = trans->fs_info;
4251         int ret;
4252         struct btrfs_extent_item *extent_item;
4253         struct btrfs_key extent_key;
4254         struct btrfs_tree_block_info *block_info;
4255         struct btrfs_extent_inline_ref *iref;
4256         struct btrfs_path *path;
4257         struct extent_buffer *leaf;
4258         struct btrfs_delayed_tree_ref *ref;
4259         u32 size = sizeof(*extent_item) + sizeof(*iref);
4260         u64 num_bytes;
4261         u64 flags = extent_op->flags_to_set;
4262         bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
4263
4264         ref = btrfs_delayed_node_to_tree_ref(node);
4265
4266         extent_key.objectid = node->bytenr;
4267         if (skinny_metadata) {
4268                 extent_key.offset = ref->level;
4269                 extent_key.type = BTRFS_METADATA_ITEM_KEY;
4270                 num_bytes = fs_info->nodesize;
4271         } else {
4272                 extent_key.offset = node->num_bytes;
4273                 extent_key.type = BTRFS_EXTENT_ITEM_KEY;
4274                 size += sizeof(*block_info);
4275                 num_bytes = node->num_bytes;
4276         }
4277
4278         path = btrfs_alloc_path();
4279         if (!path)
4280                 return -ENOMEM;
4281
4282         path->leave_spinning = 1;
4283         ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path,
4284                                       &extent_key, size);
4285         if (ret) {
4286                 btrfs_free_path(path);
4287                 return ret;
4288         }
4289
4290         leaf = path->nodes[0];
4291         extent_item = btrfs_item_ptr(leaf, path->slots[0],
4292                                      struct btrfs_extent_item);
4293         btrfs_set_extent_refs(leaf, extent_item, 1);
4294         btrfs_set_extent_generation(leaf, extent_item, trans->transid);
4295         btrfs_set_extent_flags(leaf, extent_item,
4296                                flags | BTRFS_EXTENT_FLAG_TREE_BLOCK);
4297
4298         if (skinny_metadata) {
4299                 iref = (struct btrfs_extent_inline_ref *)(extent_item + 1);
4300         } else {
4301                 block_info = (struct btrfs_tree_block_info *)(extent_item + 1);
4302                 btrfs_set_tree_block_key(leaf, block_info, &extent_op->key);
4303                 btrfs_set_tree_block_level(leaf, block_info, ref->level);
4304                 iref = (struct btrfs_extent_inline_ref *)(block_info + 1);
4305         }
4306
4307         if (node->type == BTRFS_SHARED_BLOCK_REF_KEY) {
4308                 BUG_ON(!(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF));
4309                 btrfs_set_extent_inline_ref_type(leaf, iref,
4310                                                  BTRFS_SHARED_BLOCK_REF_KEY);
4311                 btrfs_set_extent_inline_ref_offset(leaf, iref, ref->parent);
4312         } else {
4313                 btrfs_set_extent_inline_ref_type(leaf, iref,
4314                                                  BTRFS_TREE_BLOCK_REF_KEY);
4315                 btrfs_set_extent_inline_ref_offset(leaf, iref, ref->root);
4316         }
4317
4318         btrfs_mark_buffer_dirty(leaf);
4319         btrfs_free_path(path);
4320
4321         ret = remove_from_free_space_tree(trans, extent_key.objectid,
4322                                           num_bytes);
4323         if (ret)
4324                 return ret;
4325
4326         ret = btrfs_update_block_group(trans, extent_key.objectid,
4327                                        fs_info->nodesize, 1);
4328         if (ret) { /* -ENOENT, logic error */
4329                 btrfs_err(fs_info, "update block group failed for %llu %llu",
4330                         extent_key.objectid, extent_key.offset);
4331                 BUG();
4332         }
4333
4334         trace_btrfs_reserved_extent_alloc(fs_info, extent_key.objectid,
4335                                           fs_info->nodesize);
4336         return ret;
4337 }
4338
4339 int btrfs_alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
4340                                      struct btrfs_root *root, u64 owner,
4341                                      u64 offset, u64 ram_bytes,
4342                                      struct btrfs_key *ins)
4343 {
4344         struct btrfs_ref generic_ref = { 0 };
4345         int ret;
4346
4347         BUG_ON(root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID);
4348
4349         btrfs_init_generic_ref(&generic_ref, BTRFS_ADD_DELAYED_EXTENT,
4350                                ins->objectid, ins->offset, 0);
4351         btrfs_init_data_ref(&generic_ref, root->root_key.objectid, owner, offset);
4352         btrfs_ref_tree_mod(root->fs_info, &generic_ref);
4353         ret = btrfs_add_delayed_data_ref(trans, &generic_ref,
4354                                          ram_bytes, NULL, NULL);
4355         return ret;
4356 }
4357
4358 /*
4359  * this is used by the tree logging recovery code.  It records that
4360  * an extent has been allocated and makes sure to clear the free
4361  * space cache bits as well
4362  */
4363 int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
4364                                    u64 root_objectid, u64 owner, u64 offset,
4365                                    struct btrfs_key *ins)
4366 {
4367         struct btrfs_fs_info *fs_info = trans->fs_info;
4368         int ret;
4369         struct btrfs_block_group_cache *block_group;
4370         struct btrfs_space_info *space_info;
4371
4372         /*
4373          * Mixed block groups will exclude before processing the log so we only
4374          * need to do the exclude dance if this fs isn't mixed.
4375          */
4376         if (!btrfs_fs_incompat(fs_info, MIXED_GROUPS)) {
4377                 ret = __exclude_logged_extent(fs_info, ins->objectid,
4378                                               ins->offset);
4379                 if (ret)
4380                         return ret;
4381         }
4382
4383         block_group = btrfs_lookup_block_group(fs_info, ins->objectid);
4384         if (!block_group)
4385                 return -EINVAL;
4386
4387         space_info = block_group->space_info;
4388         spin_lock(&space_info->lock);
4389         spin_lock(&block_group->lock);
4390         space_info->bytes_reserved += ins->offset;
4391         block_group->reserved += ins->offset;
4392         spin_unlock(&block_group->lock);
4393         spin_unlock(&space_info->lock);
4394
4395         ret = alloc_reserved_file_extent(trans, 0, root_objectid, 0, owner,
4396                                          offset, ins, 1);
4397         btrfs_put_block_group(block_group);
4398         return ret;
4399 }
4400
4401 static struct extent_buffer *
4402 btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4403                       u64 bytenr, int level, u64 owner)
4404 {
4405         struct btrfs_fs_info *fs_info = root->fs_info;
4406         struct extent_buffer *buf;
4407
4408         buf = btrfs_find_create_tree_block(fs_info, bytenr);
4409         if (IS_ERR(buf))
4410                 return buf;
4411
4412         /*
4413          * Extra safety check in case the extent tree is corrupted and extent
4414          * allocator chooses to use a tree block which is already used and
4415          * locked.
4416          */
4417         if (buf->lock_owner == current->pid) {
4418                 btrfs_err_rl(fs_info,
4419 "tree block %llu owner %llu already locked by pid=%d, extent tree corruption detected",
4420                         buf->start, btrfs_header_owner(buf), current->pid);
4421                 free_extent_buffer(buf);
4422                 return ERR_PTR(-EUCLEAN);
4423         }
4424
4425         btrfs_set_buffer_lockdep_class(root->root_key.objectid, buf, level);
4426         btrfs_tree_lock(buf);
4427         btrfs_clean_tree_block(buf);
4428         clear_bit(EXTENT_BUFFER_STALE, &buf->bflags);
4429
4430         btrfs_set_lock_blocking_write(buf);
4431         set_extent_buffer_uptodate(buf);
4432
4433         memzero_extent_buffer(buf, 0, sizeof(struct btrfs_header));
4434         btrfs_set_header_level(buf, level);
4435         btrfs_set_header_bytenr(buf, buf->start);
4436         btrfs_set_header_generation(buf, trans->transid);
4437         btrfs_set_header_backref_rev(buf, BTRFS_MIXED_BACKREF_REV);
4438         btrfs_set_header_owner(buf, owner);
4439         write_extent_buffer_fsid(buf, fs_info->fs_devices->metadata_uuid);
4440         write_extent_buffer_chunk_tree_uuid(buf, fs_info->chunk_tree_uuid);
4441         if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
4442                 buf->log_index = root->log_transid % 2;
4443                 /*
4444                  * we allow two log transactions at a time, use different
4445                  * EXTENT bit to differentiate dirty pages.
4446                  */
4447                 if (buf->log_index == 0)
4448                         set_extent_dirty(&root->dirty_log_pages, buf->start,
4449                                         buf->start + buf->len - 1, GFP_NOFS);
4450                 else
4451                         set_extent_new(&root->dirty_log_pages, buf->start,
4452                                         buf->start + buf->len - 1);
4453         } else {
4454                 buf->log_index = -1;
4455                 set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
4456                          buf->start + buf->len - 1, GFP_NOFS);
4457         }
4458         trans->dirty = true;
4459         /* this returns a buffer locked for blocking */
4460         return buf;
4461 }
4462
4463 /*
4464  * finds a free extent and does all the dirty work required for allocation
4465  * returns the tree buffer or an ERR_PTR on error.
4466  */
4467 struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
4468                                              struct btrfs_root *root,
4469                                              u64 parent, u64 root_objectid,
4470                                              const struct btrfs_disk_key *key,
4471                                              int level, u64 hint,
4472                                              u64 empty_size)
4473 {
4474         struct btrfs_fs_info *fs_info = root->fs_info;
4475         struct btrfs_key ins;
4476         struct btrfs_block_rsv *block_rsv;
4477         struct extent_buffer *buf;
4478         struct btrfs_delayed_extent_op *extent_op;
4479         struct btrfs_ref generic_ref = { 0 };
4480         u64 flags = 0;
4481         int ret;
4482         u32 blocksize = fs_info->nodesize;
4483         bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
4484
4485 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
4486         if (btrfs_is_testing(fs_info)) {
4487                 buf = btrfs_init_new_buffer(trans, root, root->alloc_bytenr,
4488                                             level, root_objectid);
4489                 if (!IS_ERR(buf))
4490                         root->alloc_bytenr += blocksize;
4491                 return buf;
4492         }
4493 #endif
4494
4495         block_rsv = btrfs_use_block_rsv(trans, root, blocksize);
4496         if (IS_ERR(block_rsv))
4497                 return ERR_CAST(block_rsv);
4498
4499         ret = btrfs_reserve_extent(root, blocksize, blocksize, blocksize,
4500                                    empty_size, hint, &ins, 0, 0);
4501         if (ret)
4502                 goto out_unuse;
4503
4504         buf = btrfs_init_new_buffer(trans, root, ins.objectid, level,
4505                                     root_objectid);
4506         if (IS_ERR(buf)) {
4507                 ret = PTR_ERR(buf);
4508                 goto out_free_reserved;
4509         }
4510
4511         if (root_objectid == BTRFS_TREE_RELOC_OBJECTID) {
4512                 if (parent == 0)
4513                         parent = ins.objectid;
4514                 flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
4515         } else
4516                 BUG_ON(parent > 0);
4517
4518         if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
4519                 extent_op = btrfs_alloc_delayed_extent_op();
4520                 if (!extent_op) {
4521                         ret = -ENOMEM;
4522                         goto out_free_buf;
4523                 }
4524                 if (key)
4525                         memcpy(&extent_op->key, key, sizeof(extent_op->key));
4526                 else
4527                         memset(&extent_op->key, 0, sizeof(extent_op->key));
4528                 extent_op->flags_to_set = flags;
4529                 extent_op->update_key = skinny_metadata ? false : true;
4530                 extent_op->update_flags = true;
4531                 extent_op->is_data = false;
4532                 extent_op->level = level;
4533
4534                 btrfs_init_generic_ref(&generic_ref, BTRFS_ADD_DELAYED_EXTENT,
4535                                        ins.objectid, ins.offset, parent);
4536                 generic_ref.real_root = root->root_key.objectid;
4537                 btrfs_init_tree_ref(&generic_ref, level, root_objectid);
4538                 btrfs_ref_tree_mod(fs_info, &generic_ref);
4539                 ret = btrfs_add_delayed_tree_ref(trans, &generic_ref,
4540                                                  extent_op, NULL, NULL);
4541                 if (ret)
4542                         goto out_free_delayed;
4543         }
4544         return buf;
4545
4546 out_free_delayed:
4547         btrfs_free_delayed_extent_op(extent_op);
4548 out_free_buf:
4549         free_extent_buffer(buf);
4550 out_free_reserved:
4551         btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 0);
4552 out_unuse:
4553         btrfs_unuse_block_rsv(fs_info, block_rsv, blocksize);
4554         return ERR_PTR(ret);
4555 }
4556
4557 struct walk_control {
4558         u64 refs[BTRFS_MAX_LEVEL];
4559         u64 flags[BTRFS_MAX_LEVEL];
4560         struct btrfs_key update_progress;
4561         struct btrfs_key drop_progress;
4562         int drop_level;
4563         int stage;
4564         int level;
4565         int shared_level;
4566         int update_ref;
4567         int keep_locks;
4568         int reada_slot;
4569         int reada_count;
4570         int restarted;
4571 };
4572
4573 #define DROP_REFERENCE  1
4574 #define UPDATE_BACKREF  2
4575
4576 static noinline void reada_walk_down(struct btrfs_trans_handle *trans,
4577                                      struct btrfs_root *root,
4578                                      struct walk_control *wc,
4579                                      struct btrfs_path *path)
4580 {
4581         struct btrfs_fs_info *fs_info = root->fs_info;
4582         u64 bytenr;
4583         u64 generation;
4584         u64 refs;
4585         u64 flags;
4586         u32 nritems;
4587         struct btrfs_key key;
4588         struct extent_buffer *eb;
4589         int ret;
4590         int slot;
4591         int nread = 0;
4592
4593         if (path->slots[wc->level] < wc->reada_slot) {
4594                 wc->reada_count = wc->reada_count * 2 / 3;
4595                 wc->reada_count = max(wc->reada_count, 2);
4596         } else {
4597                 wc->reada_count = wc->reada_count * 3 / 2;
4598                 wc->reada_count = min_t(int, wc->reada_count,
4599                                         BTRFS_NODEPTRS_PER_BLOCK(fs_info));
4600         }
4601
4602         eb = path->nodes[wc->level];
4603         nritems = btrfs_header_nritems(eb);
4604
4605         for (slot = path->slots[wc->level]; slot < nritems; slot++) {
4606                 if (nread >= wc->reada_count)
4607                         break;
4608
4609                 cond_resched();
4610                 bytenr = btrfs_node_blockptr(eb, slot);
4611                 generation = btrfs_node_ptr_generation(eb, slot);
4612
4613                 if (slot == path->slots[wc->level])
4614                         goto reada;
4615
4616                 if (wc->stage == UPDATE_BACKREF &&
4617                     generation <= root->root_key.offset)
4618                         continue;
4619
4620                 /* We don't lock the tree block, it's OK to be racy here */
4621                 ret = btrfs_lookup_extent_info(trans, fs_info, bytenr,
4622                                                wc->level - 1, 1, &refs,
4623                                                &flags);
4624                 /* We don't care about errors in readahead. */
4625                 if (ret < 0)
4626                         continue;
4627                 BUG_ON(refs == 0);
4628
4629                 if (wc->stage == DROP_REFERENCE) {
4630                         if (refs == 1)
4631                                 goto reada;
4632
4633                         if (wc->level == 1 &&
4634                             (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF))
4635                                 continue;
4636                         if (!wc->update_ref ||
4637                             generation <= root->root_key.offset)
4638                                 continue;
4639                         btrfs_node_key_to_cpu(eb, &key, slot);
4640                         ret = btrfs_comp_cpu_keys(&key,
4641                                                   &wc->update_progress);
4642                         if (ret < 0)
4643                                 continue;
4644                 } else {
4645                         if (wc->level == 1 &&
4646                             (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF))
4647                                 continue;
4648                 }
4649 reada:
4650                 readahead_tree_block(fs_info, bytenr);
4651                 nread++;
4652         }
4653         wc->reada_slot = slot;
4654 }
4655
4656 /*
4657  * helper to process tree block while walking down the tree.
4658  *
4659  * when wc->stage == UPDATE_BACKREF, this function updates
4660  * back refs for pointers in the block.
4661  *
4662  * NOTE: return value 1 means we should stop walking down.
4663  */
4664 static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
4665                                    struct btrfs_root *root,
4666                                    struct btrfs_path *path,
4667                                    struct walk_control *wc, int lookup_info)
4668 {
4669         struct btrfs_fs_info *fs_info = root->fs_info;
4670         int level = wc->level;
4671         struct extent_buffer *eb = path->nodes[level];
4672         u64 flag = BTRFS_BLOCK_FLAG_FULL_BACKREF;
4673         int ret;
4674
4675         if (wc->stage == UPDATE_BACKREF &&
4676             btrfs_header_owner(eb) != root->root_key.objectid)
4677                 return 1;
4678
4679         /*
4680          * when reference count of tree block is 1, it won't increase
4681          * again. once full backref flag is set, we never clear it.
4682          */
4683         if (lookup_info &&
4684             ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) ||
4685              (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag)))) {
4686                 BUG_ON(!path->locks[level]);
4687                 ret = btrfs_lookup_extent_info(trans, fs_info,
4688                                                eb->start, level, 1,
4689                                                &wc->refs[level],
4690                                                &wc->flags[level]);
4691                 BUG_ON(ret == -ENOMEM);
4692                 if (ret)
4693                         return ret;
4694                 BUG_ON(wc->refs[level] == 0);
4695         }
4696
4697         if (wc->stage == DROP_REFERENCE) {
4698                 if (wc->refs[level] > 1)
4699                         return 1;
4700
4701                 if (path->locks[level] && !wc->keep_locks) {
4702                         btrfs_tree_unlock_rw(eb, path->locks[level]);
4703                         path->locks[level] = 0;
4704                 }
4705                 return 0;
4706         }
4707
4708         /* wc->stage == UPDATE_BACKREF */
4709         if (!(wc->flags[level] & flag)) {
4710                 BUG_ON(!path->locks[level]);
4711                 ret = btrfs_inc_ref(trans, root, eb, 1);
4712                 BUG_ON(ret); /* -ENOMEM */
4713                 ret = btrfs_dec_ref(trans, root, eb, 0);
4714                 BUG_ON(ret); /* -ENOMEM */
4715                 ret = btrfs_set_disk_extent_flags(trans, eb->start,
4716                                                   eb->len, flag,
4717                                                   btrfs_header_level(eb), 0);
4718                 BUG_ON(ret); /* -ENOMEM */
4719                 wc->flags[level] |= flag;
4720         }
4721
4722         /*
4723          * the block is shared by multiple trees, so it's not good to
4724          * keep the tree lock
4725          */
4726         if (path->locks[level] && level > 0) {
4727                 btrfs_tree_unlock_rw(eb, path->locks[level]);
4728                 path->locks[level] = 0;
4729         }
4730         return 0;
4731 }
4732
4733 /*
4734  * This is used to verify a ref exists for this root to deal with a bug where we
4735  * would have a drop_progress key that hadn't been updated properly.
4736  */
4737 static int check_ref_exists(struct btrfs_trans_handle *trans,
4738                             struct btrfs_root *root, u64 bytenr, u64 parent,
4739                             int level)
4740 {
4741         struct btrfs_path *path;
4742         struct btrfs_extent_inline_ref *iref;
4743         int ret;
4744
4745         path = btrfs_alloc_path();
4746         if (!path)
4747                 return -ENOMEM;
4748
4749         ret = lookup_extent_backref(trans, path, &iref, bytenr,
4750                                     root->fs_info->nodesize, parent,
4751                                     root->root_key.objectid, level, 0);
4752         btrfs_free_path(path);
4753         if (ret == -ENOENT)
4754                 return 0;
4755         if (ret < 0)
4756                 return ret;
4757         return 1;
4758 }
4759
4760 /*
4761  * helper to process tree block pointer.
4762  *
4763  * when wc->stage == DROP_REFERENCE, this function checks
4764  * reference count of the block pointed to. if the block
4765  * is shared and we need update back refs for the subtree
4766  * rooted at the block, this function changes wc->stage to
4767  * UPDATE_BACKREF. if the block is shared and there is no
4768  * need to update back, this function drops the reference
4769  * to the block.
4770  *
4771  * NOTE: return value 1 means we should stop walking down.
4772  */
4773 static noinline int do_walk_down(struct btrfs_trans_handle *trans,
4774                                  struct btrfs_root *root,
4775                                  struct btrfs_path *path,
4776                                  struct walk_control *wc, int *lookup_info)
4777 {
4778         struct btrfs_fs_info *fs_info = root->fs_info;
4779         u64 bytenr;
4780         u64 generation;
4781         u64 parent;
4782         struct btrfs_key key;
4783         struct btrfs_key first_key;
4784         struct btrfs_ref ref = { 0 };
4785         struct extent_buffer *next;
4786         int level = wc->level;
4787         int reada = 0;
4788         int ret = 0;
4789         bool need_account = false;
4790
4791         generation = btrfs_node_ptr_generation(path->nodes[level],
4792                                                path->slots[level]);
4793         /*
4794          * if the lower level block was created before the snapshot
4795          * was created, we know there is no need to update back refs
4796          * for the subtree
4797          */
4798         if (wc->stage == UPDATE_BACKREF &&
4799             generation <= root->root_key.offset) {
4800                 *lookup_info = 1;
4801                 return 1;
4802         }
4803
4804         bytenr = btrfs_node_blockptr(path->nodes[level], path->slots[level]);
4805         btrfs_node_key_to_cpu(path->nodes[level], &first_key,
4806                               path->slots[level]);
4807
4808         next = find_extent_buffer(fs_info, bytenr);
4809         if (!next) {
4810                 next = btrfs_find_create_tree_block(fs_info, bytenr);
4811                 if (IS_ERR(next))
4812                         return PTR_ERR(next);
4813
4814                 btrfs_set_buffer_lockdep_class(root->root_key.objectid, next,
4815                                                level - 1);
4816                 reada = 1;
4817         }
4818         btrfs_tree_lock(next);
4819         btrfs_set_lock_blocking_write(next);
4820
4821         ret = btrfs_lookup_extent_info(trans, fs_info, bytenr, level - 1, 1,
4822                                        &wc->refs[level - 1],
4823                                        &wc->flags[level - 1]);
4824         if (ret < 0)
4825                 goto out_unlock;
4826
4827         if (unlikely(wc->refs[level - 1] == 0)) {
4828                 btrfs_err(fs_info, "Missing references.");
4829                 ret = -EIO;
4830                 goto out_unlock;
4831         }
4832         *lookup_info = 0;
4833
4834         if (wc->stage == DROP_REFERENCE) {
4835                 if (wc->refs[level - 1] > 1) {
4836                         need_account = true;
4837                         if (level == 1 &&
4838                             (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF))
4839                                 goto skip;
4840
4841                         if (!wc->update_ref ||
4842                             generation <= root->root_key.offset)
4843                                 goto skip;
4844
4845                         btrfs_node_key_to_cpu(path->nodes[level], &key,
4846                                               path->slots[level]);
4847                         ret = btrfs_comp_cpu_keys(&key, &wc->update_progress);
4848                         if (ret < 0)
4849                                 goto skip;
4850
4851                         wc->stage = UPDATE_BACKREF;
4852                         wc->shared_level = level - 1;
4853                 }
4854         } else {
4855                 if (level == 1 &&
4856                     (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF))
4857                         goto skip;
4858         }
4859
4860         if (!btrfs_buffer_uptodate(next, generation, 0)) {
4861                 btrfs_tree_unlock(next);
4862                 free_extent_buffer(next);
4863                 next = NULL;
4864                 *lookup_info = 1;
4865         }
4866
4867         if (!next) {
4868                 if (reada && level == 1)
4869                         reada_walk_down(trans, root, wc, path);
4870                 next = read_tree_block(fs_info, bytenr, generation, level - 1,
4871                                        &first_key);
4872                 if (IS_ERR(next)) {
4873                         return PTR_ERR(next);
4874                 } else if (!extent_buffer_uptodate(next)) {
4875                         free_extent_buffer(next);
4876                         return -EIO;
4877                 }
4878                 btrfs_tree_lock(next);
4879                 btrfs_set_lock_blocking_write(next);
4880         }
4881
4882         level--;
4883         ASSERT(level == btrfs_header_level(next));
4884         if (level != btrfs_header_level(next)) {
4885                 btrfs_err(root->fs_info, "mismatched level");
4886                 ret = -EIO;
4887                 goto out_unlock;
4888         }
4889         path->nodes[level] = next;
4890         path->slots[level] = 0;
4891         path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
4892         wc->level = level;
4893         if (wc->level == 1)
4894                 wc->reada_slot = 0;
4895         return 0;
4896 skip:
4897         wc->refs[level - 1] = 0;
4898         wc->flags[level - 1] = 0;
4899         if (wc->stage == DROP_REFERENCE) {
4900                 if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
4901                         parent = path->nodes[level]->start;
4902                 } else {
4903                         ASSERT(root->root_key.objectid ==
4904                                btrfs_header_owner(path->nodes[level]));
4905                         if (root->root_key.objectid !=
4906                             btrfs_header_owner(path->nodes[level])) {
4907                                 btrfs_err(root->fs_info,
4908                                                 "mismatched block owner");
4909                                 ret = -EIO;
4910                                 goto out_unlock;
4911                         }
4912                         parent = 0;
4913                 }
4914
4915                 /*
4916                  * If we had a drop_progress we need to verify the refs are set
4917                  * as expected.  If we find our ref then we know that from here
4918                  * on out everything should be correct, and we can clear the
4919                  * ->restarted flag.
4920                  */
4921                 if (wc->restarted) {
4922                         ret = check_ref_exists(trans, root, bytenr, parent,
4923                                                level - 1);
4924                         if (ret < 0)
4925                                 goto out_unlock;
4926                         if (ret == 0)
4927                                 goto no_delete;
4928                         ret = 0;
4929                         wc->restarted = 0;
4930                 }
4931
4932                 /*
4933                  * Reloc tree doesn't contribute to qgroup numbers, and we have
4934                  * already accounted them at merge time (replace_path),
4935                  * thus we could skip expensive subtree trace here.
4936                  */
4937                 if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
4938                     need_account) {
4939                         ret = btrfs_qgroup_trace_subtree(trans, next,
4940                                                          generation, level - 1);
4941                         if (ret) {
4942                                 btrfs_err_rl(fs_info,
4943                                              "Error %d accounting shared subtree. Quota is out of sync, rescan required.",
4944                                              ret);
4945                         }
4946                 }
4947
4948                 /*
4949                  * We need to update the next key in our walk control so we can
4950                  * update the drop_progress key accordingly.  We don't care if
4951                  * find_next_key doesn't find a key because that means we're at
4952                  * the end and are going to clean up now.
4953                  */
4954                 wc->drop_level = level;
4955                 find_next_key(path, level, &wc->drop_progress);
4956
4957                 btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, bytenr,
4958                                        fs_info->nodesize, parent);
4959                 btrfs_init_tree_ref(&ref, level - 1, root->root_key.objectid);
4960                 ret = btrfs_free_extent(trans, &ref);
4961                 if (ret)
4962                         goto out_unlock;
4963         }
4964 no_delete:
4965         *lookup_info = 1;
4966         ret = 1;
4967
4968 out_unlock:
4969         btrfs_tree_unlock(next);
4970         free_extent_buffer(next);
4971
4972         return ret;
4973 }
4974
4975 /*
4976  * helper to process tree block while walking up the tree.
4977  *
4978  * when wc->stage == DROP_REFERENCE, this function drops
4979  * reference count on the block.
4980  *
4981  * when wc->stage == UPDATE_BACKREF, this function changes
4982  * wc->stage back to DROP_REFERENCE if we changed wc->stage
4983  * to UPDATE_BACKREF previously while processing the block.
4984  *
4985  * NOTE: return value 1 means we should stop walking up.
4986  */
4987 static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
4988                                  struct btrfs_root *root,
4989                                  struct btrfs_path *path,
4990                                  struct walk_control *wc)
4991 {
4992         struct btrfs_fs_info *fs_info = root->fs_info;
4993         int ret;
4994         int level = wc->level;
4995         struct extent_buffer *eb = path->nodes[level];
4996         u64 parent = 0;
4997
4998         if (wc->stage == UPDATE_BACKREF) {
4999                 BUG_ON(wc->shared_level < level);
5000                 if (level < wc->shared_level)
5001                         goto out;
5002
5003                 ret = find_next_key(path, level + 1, &wc->update_progress);
5004                 if (ret > 0)
5005                         wc->update_ref = 0;
5006
5007                 wc->stage = DROP_REFERENCE;
5008                 wc->shared_level = -1;
5009                 path->slots[level] = 0;
5010
5011                 /*
5012                  * check reference count again if the block isn't locked.
5013                  * we should start walking down the tree again if reference
5014                  * count is one.
5015                  */
5016                 if (!path->locks[level]) {
5017                         BUG_ON(level == 0);
5018                         btrfs_tree_lock(eb);
5019                         btrfs_set_lock_blocking_write(eb);
5020                         path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
5021
5022                         ret = btrfs_lookup_extent_info(trans, fs_info,
5023                                                        eb->start, level, 1,
5024                                                        &wc->refs[level],
5025                                                        &wc->flags[level]);
5026                         if (ret < 0) {
5027                                 btrfs_tree_unlock_rw(eb, path->locks[level]);
5028                                 path->locks[level] = 0;
5029                                 return ret;
5030                         }
5031                         BUG_ON(wc->refs[level] == 0);
5032                         if (wc->refs[level] == 1) {
5033                                 btrfs_tree_unlock_rw(eb, path->locks[level]);
5034                                 path->locks[level] = 0;
5035                                 return 1;
5036                         }
5037                 }
5038         }
5039
5040         /* wc->stage == DROP_REFERENCE */
5041         BUG_ON(wc->refs[level] > 1 && !path->locks[level]);
5042
5043         if (wc->refs[level] == 1) {
5044                 if (level == 0) {
5045                         if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5046                                 ret = btrfs_dec_ref(trans, root, eb, 1);
5047                         else
5048                                 ret = btrfs_dec_ref(trans, root, eb, 0);
5049                         BUG_ON(ret); /* -ENOMEM */
5050                         if (is_fstree(root->root_key.objectid)) {
5051                                 ret = btrfs_qgroup_trace_leaf_items(trans, eb);
5052                                 if (ret) {
5053                                         btrfs_err_rl(fs_info,
5054         "error %d accounting leaf items, quota is out of sync, rescan required",
5055                                              ret);
5056                                 }
5057                         }
5058                 }
5059                 /* make block locked assertion in btrfs_clean_tree_block happy */
5060                 if (!path->locks[level] &&
5061                     btrfs_header_generation(eb) == trans->transid) {
5062                         btrfs_tree_lock(eb);
5063                         btrfs_set_lock_blocking_write(eb);
5064                         path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
5065                 }
5066                 btrfs_clean_tree_block(eb);
5067         }
5068
5069         if (eb == root->node) {
5070                 if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5071                         parent = eb->start;
5072                 else if (root->root_key.objectid != btrfs_header_owner(eb))
5073                         goto owner_mismatch;
5074         } else {
5075                 if (wc->flags[level + 1] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5076                         parent = path->nodes[level + 1]->start;
5077                 else if (root->root_key.objectid !=
5078                          btrfs_header_owner(path->nodes[level + 1]))
5079                         goto owner_mismatch;
5080         }
5081
5082         btrfs_free_tree_block(trans, root, eb, parent, wc->refs[level] == 1);
5083 out:
5084         wc->refs[level] = 0;
5085         wc->flags[level] = 0;
5086         return 0;
5087
5088 owner_mismatch:
5089         btrfs_err_rl(fs_info, "unexpected tree owner, have %llu expect %llu",
5090                      btrfs_header_owner(eb), root->root_key.objectid);
5091         return -EUCLEAN;
5092 }
5093
5094 static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
5095                                    struct btrfs_root *root,
5096                                    struct btrfs_path *path,
5097                                    struct walk_control *wc)
5098 {
5099         int level = wc->level;
5100         int lookup_info = 1;
5101         int ret;
5102
5103         while (level >= 0) {
5104                 ret = walk_down_proc(trans, root, path, wc, lookup_info);
5105                 if (ret > 0)
5106                         break;
5107
5108                 if (level == 0)
5109                         break;
5110
5111                 if (path->slots[level] >=
5112                     btrfs_header_nritems(path->nodes[level]))
5113                         break;
5114
5115                 ret = do_walk_down(trans, root, path, wc, &lookup_info);
5116                 if (ret > 0) {
5117                         path->slots[level]++;
5118                         continue;
5119                 } else if (ret < 0)
5120                         return ret;
5121                 level = wc->level;
5122         }
5123         return 0;
5124 }
5125
5126 static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
5127                                  struct btrfs_root *root,
5128                                  struct btrfs_path *path,
5129                                  struct walk_control *wc, int max_level)
5130 {
5131         int level = wc->level;
5132         int ret;
5133
5134         path->slots[level] = btrfs_header_nritems(path->nodes[level]);
5135         while (level < max_level && path->nodes[level]) {
5136                 wc->level = level;
5137                 if (path->slots[level] + 1 <
5138                     btrfs_header_nritems(path->nodes[level])) {
5139                         path->slots[level]++;
5140                         return 0;
5141                 } else {
5142                         ret = walk_up_proc(trans, root, path, wc);
5143                         if (ret > 0)
5144                                 return 0;
5145                         if (ret < 0)
5146                                 return ret;
5147
5148                         if (path->locks[level]) {
5149                                 btrfs_tree_unlock_rw(path->nodes[level],
5150                                                      path->locks[level]);
5151                                 path->locks[level] = 0;
5152                         }
5153                         free_extent_buffer(path->nodes[level]);
5154                         path->nodes[level] = NULL;
5155                         level++;
5156                 }
5157         }
5158         return 1;
5159 }
5160
5161 /*
5162  * drop a subvolume tree.
5163  *
5164  * this function traverses the tree freeing any blocks that only
5165  * referenced by the tree.
5166  *
5167  * when a shared tree block is found. this function decreases its
5168  * reference count by one. if update_ref is true, this function
5169  * also make sure backrefs for the shared block and all lower level
5170  * blocks are properly updated.
5171  *
5172  * If called with for_reloc == 0, may exit early with -EAGAIN
5173  */
5174 int btrfs_drop_snapshot(struct btrfs_root *root,
5175                          struct btrfs_block_rsv *block_rsv, int update_ref,
5176                          int for_reloc)
5177 {
5178         struct btrfs_fs_info *fs_info = root->fs_info;
5179         struct btrfs_path *path;
5180         struct btrfs_trans_handle *trans;
5181         struct btrfs_root *tree_root = fs_info->tree_root;
5182         struct btrfs_root_item *root_item = &root->root_item;
5183         struct walk_control *wc;
5184         struct btrfs_key key;
5185         int err = 0;
5186         int ret;
5187         int level;
5188         bool root_dropped = false;
5189
5190         btrfs_debug(fs_info, "Drop subvolume %llu", root->root_key.objectid);
5191
5192         path = btrfs_alloc_path();
5193         if (!path) {
5194                 err = -ENOMEM;
5195                 goto out;
5196         }
5197
5198         wc = kzalloc(sizeof(*wc), GFP_NOFS);
5199         if (!wc) {
5200                 btrfs_free_path(path);
5201                 err = -ENOMEM;
5202                 goto out;
5203         }
5204
5205         trans = btrfs_start_transaction(tree_root, 0);
5206         if (IS_ERR(trans)) {
5207                 err = PTR_ERR(trans);
5208                 goto out_free;
5209         }
5210
5211         err = btrfs_run_delayed_items(trans);
5212         if (err)
5213                 goto out_end_trans;
5214
5215         if (block_rsv)
5216                 trans->block_rsv = block_rsv;
5217
5218         /*
5219          * This will help us catch people modifying the fs tree while we're
5220          * dropping it.  It is unsafe to mess with the fs tree while it's being
5221          * dropped as we unlock the root node and parent nodes as we walk down
5222          * the tree, assuming nothing will change.  If something does change
5223          * then we'll have stale information and drop references to blocks we've
5224          * already dropped.
5225          */
5226         set_bit(BTRFS_ROOT_DELETING, &root->state);
5227         if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
5228                 level = btrfs_header_level(root->node);
5229                 path->nodes[level] = btrfs_lock_root_node(root);
5230                 btrfs_set_lock_blocking_write(path->nodes[level]);
5231                 path->slots[level] = 0;
5232                 path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
5233                 memset(&wc->update_progress, 0,
5234                        sizeof(wc->update_progress));
5235         } else {
5236                 btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
5237                 memcpy(&wc->update_progress, &key,
5238                        sizeof(wc->update_progress));
5239
5240                 level = root_item->drop_level;
5241                 BUG_ON(level == 0);
5242                 path->lowest_level = level;
5243                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5244                 path->lowest_level = 0;
5245                 if (ret < 0) {
5246                         err = ret;
5247                         goto out_end_trans;
5248                 }
5249                 WARN_ON(ret > 0);
5250
5251                 /*
5252                  * unlock our path, this is safe because only this
5253                  * function is allowed to delete this snapshot
5254                  */
5255                 btrfs_unlock_up_safe(path, 0);
5256
5257                 level = btrfs_header_level(root->node);
5258                 while (1) {
5259                         btrfs_tree_lock(path->nodes[level]);
5260                         btrfs_set_lock_blocking_write(path->nodes[level]);
5261                         path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
5262
5263                         ret = btrfs_lookup_extent_info(trans, fs_info,
5264                                                 path->nodes[level]->start,
5265                                                 level, 1, &wc->refs[level],
5266                                                 &wc->flags[level]);
5267                         if (ret < 0) {
5268                                 err = ret;
5269                                 goto out_end_trans;
5270                         }
5271                         BUG_ON(wc->refs[level] == 0);
5272
5273                         if (level == root_item->drop_level)
5274                                 break;
5275
5276                         btrfs_tree_unlock(path->nodes[level]);
5277                         path->locks[level] = 0;
5278                         WARN_ON(wc->refs[level] != 1);
5279                         level--;
5280                 }
5281         }
5282
5283         wc->restarted = test_bit(BTRFS_ROOT_DEAD_TREE, &root->state);
5284         wc->level = level;
5285         wc->shared_level = -1;
5286         wc->stage = DROP_REFERENCE;
5287         wc->update_ref = update_ref;
5288         wc->keep_locks = 0;
5289         wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(fs_info);
5290
5291         while (1) {
5292
5293                 ret = walk_down_tree(trans, root, path, wc);
5294                 if (ret < 0) {
5295                         err = ret;
5296                         break;
5297                 }
5298
5299                 ret = walk_up_tree(trans, root, path, wc, BTRFS_MAX_LEVEL);
5300                 if (ret < 0) {
5301                         err = ret;
5302                         break;
5303                 }
5304
5305                 if (ret > 0) {
5306                         BUG_ON(wc->stage != DROP_REFERENCE);
5307                         break;
5308                 }
5309
5310                 if (wc->stage == DROP_REFERENCE) {
5311                         wc->drop_level = wc->level;
5312                         btrfs_node_key_to_cpu(path->nodes[wc->drop_level],
5313                                               &wc->drop_progress,
5314                                               path->slots[wc->drop_level]);
5315                 }
5316                 btrfs_cpu_key_to_disk(&root_item->drop_progress,
5317                                       &wc->drop_progress);
5318                 root_item->drop_level = wc->drop_level;
5319
5320                 BUG_ON(wc->level == 0);
5321                 if (btrfs_should_end_transaction(trans) ||
5322                     (!for_reloc && btrfs_need_cleaner_sleep(fs_info))) {
5323                         ret = btrfs_update_root(trans, tree_root,
5324                                                 &root->root_key,
5325                                                 root_item);
5326                         if (ret) {
5327                                 btrfs_abort_transaction(trans, ret);
5328                                 err = ret;
5329                                 goto out_end_trans;
5330                         }
5331
5332                         btrfs_end_transaction_throttle(trans);
5333                         if (!for_reloc && btrfs_need_cleaner_sleep(fs_info)) {
5334                                 btrfs_debug(fs_info,
5335                                             "drop snapshot early exit");
5336                                 err = -EAGAIN;
5337                                 goto out_free;
5338                         }
5339
5340                         trans = btrfs_start_transaction(tree_root, 0);
5341                         if (IS_ERR(trans)) {
5342                                 err = PTR_ERR(trans);
5343                                 goto out_free;
5344                         }
5345                         if (block_rsv)
5346                                 trans->block_rsv = block_rsv;
5347                 }
5348         }
5349         btrfs_release_path(path);
5350         if (err)
5351                 goto out_end_trans;
5352
5353         ret = btrfs_del_root(trans, &root->root_key);
5354         if (ret) {
5355                 btrfs_abort_transaction(trans, ret);
5356                 err = ret;
5357                 goto out_end_trans;
5358         }
5359
5360         if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
5361                 ret = btrfs_find_root(tree_root, &root->root_key, path,
5362                                       NULL, NULL);
5363                 if (ret < 0) {
5364                         btrfs_abort_transaction(trans, ret);
5365                         err = ret;
5366                         goto out_end_trans;
5367                 } else if (ret > 0) {
5368                         /* if we fail to delete the orphan item this time
5369                          * around, it'll get picked up the next time.
5370                          *
5371                          * The most common failure here is just -ENOENT.
5372                          */
5373                         btrfs_del_orphan_item(trans, tree_root,
5374                                               root->root_key.objectid);
5375                 }
5376         }
5377
5378         if (test_bit(BTRFS_ROOT_IN_RADIX, &root->state)) {
5379                 btrfs_add_dropped_root(trans, root);
5380         } else {
5381                 free_extent_buffer(root->node);
5382                 free_extent_buffer(root->commit_root);
5383                 btrfs_put_fs_root(root);
5384         }
5385         root_dropped = true;
5386 out_end_trans:
5387         btrfs_end_transaction_throttle(trans);
5388 out_free:
5389         kfree(wc);
5390         btrfs_free_path(path);
5391 out:
5392         /*
5393          * So if we need to stop dropping the snapshot for whatever reason we
5394          * need to make sure to add it back to the dead root list so that we
5395          * keep trying to do the work later.  This also cleans up roots if we
5396          * don't have it in the radix (like when we recover after a power fail
5397          * or unmount) so we don't leak memory.
5398          */
5399         if (!for_reloc && !root_dropped)
5400                 btrfs_add_dead_root(root);
5401         if (err && err != -EAGAIN)
5402                 btrfs_handle_fs_error(fs_info, err, NULL);
5403         return err;
5404 }
5405
5406 /*
5407  * drop subtree rooted at tree block 'node'.
5408  *
5409  * NOTE: this function will unlock and release tree block 'node'
5410  * only used by relocation code
5411  */
5412 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
5413                         struct btrfs_root *root,
5414                         struct extent_buffer *node,
5415                         struct extent_buffer *parent)
5416 {
5417         struct btrfs_fs_info *fs_info = root->fs_info;
5418         struct btrfs_path *path;
5419         struct walk_control *wc;
5420         int level;
5421         int parent_level;
5422         int ret = 0;
5423         int wret;
5424
5425         BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
5426
5427         path = btrfs_alloc_path();
5428         if (!path)
5429                 return -ENOMEM;
5430
5431         wc = kzalloc(sizeof(*wc), GFP_NOFS);
5432         if (!wc) {
5433                 btrfs_free_path(path);
5434                 return -ENOMEM;
5435         }
5436
5437         btrfs_assert_tree_locked(parent);
5438         parent_level = btrfs_header_level(parent);
5439         extent_buffer_get(parent);
5440         path->nodes[parent_level] = parent;
5441         path->slots[parent_level] = btrfs_header_nritems(parent);
5442
5443         btrfs_assert_tree_locked(node);
5444         level = btrfs_header_level(node);
5445         path->nodes[level] = node;
5446         path->slots[level] = 0;
5447         path->locks[level] = BTRFS_WRITE_LOCK_BLOCKING;
5448
5449         wc->refs[parent_level] = 1;
5450         wc->flags[parent_level] = BTRFS_BLOCK_FLAG_FULL_BACKREF;
5451         wc->level = level;
5452         wc->shared_level = -1;
5453         wc->stage = DROP_REFERENCE;
5454         wc->update_ref = 0;
5455         wc->keep_locks = 1;
5456         wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(fs_info);
5457
5458         while (1) {
5459                 wret = walk_down_tree(trans, root, path, wc);
5460                 if (wret < 0) {
5461                         ret = wret;
5462                         break;
5463                 }
5464
5465                 wret = walk_up_tree(trans, root, path, wc, parent_level);
5466                 if (wret < 0)
5467                         ret = wret;
5468                 if (wret != 0)
5469                         break;
5470         }
5471
5472         kfree(wc);
5473         btrfs_free_path(path);
5474         return ret;
5475 }
5476
5477 /*
5478  * helper to account the unused space of all the readonly block group in the
5479  * space_info. takes mirrors into account.
5480  */
5481 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo)
5482 {
5483         struct btrfs_block_group_cache *block_group;
5484         u64 free_bytes = 0;
5485         int factor;
5486
5487         /* It's df, we don't care if it's racy */
5488         if (list_empty(&sinfo->ro_bgs))
5489                 return 0;
5490
5491         spin_lock(&sinfo->lock);
5492         list_for_each_entry(block_group, &sinfo->ro_bgs, ro_list) {
5493                 spin_lock(&block_group->lock);
5494
5495                 if (!block_group->ro) {
5496                         spin_unlock(&block_group->lock);
5497                         continue;
5498                 }
5499
5500                 factor = btrfs_bg_type_to_factor(block_group->flags);
5501                 free_bytes += (block_group->key.offset -
5502                                btrfs_block_group_used(&block_group->item)) *
5503                                factor;
5504
5505                 spin_unlock(&block_group->lock);
5506         }
5507         spin_unlock(&sinfo->lock);
5508
5509         return free_bytes;
5510 }
5511
5512 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
5513                                    u64 start, u64 end)
5514 {
5515         return unpin_extent_range(fs_info, start, end, false);
5516 }
5517
5518 /*
5519  * It used to be that old block groups would be left around forever.
5520  * Iterating over them would be enough to trim unused space.  Since we
5521  * now automatically remove them, we also need to iterate over unallocated
5522  * space.
5523  *
5524  * We don't want a transaction for this since the discard may take a
5525  * substantial amount of time.  We don't require that a transaction be
5526  * running, but we do need to take a running transaction into account
5527  * to ensure that we're not discarding chunks that were released or
5528  * allocated in the current transaction.
5529  *
5530  * Holding the chunks lock will prevent other threads from allocating
5531  * or releasing chunks, but it won't prevent a running transaction
5532  * from committing and releasing the memory that the pending chunks
5533  * list head uses.  For that, we need to take a reference to the
5534  * transaction and hold the commit root sem.  We only need to hold
5535  * it while performing the free space search since we have already
5536  * held back allocations.
5537  */
5538 static int btrfs_trim_free_extents(struct btrfs_device *device, u64 *trimmed)
5539 {
5540         u64 start = SZ_1M, len = 0, end = 0;
5541         int ret;
5542
5543         *trimmed = 0;
5544
5545         /* Discard not supported = nothing to do. */
5546         if (!blk_queue_discard(bdev_get_queue(device->bdev)))
5547                 return 0;
5548
5549         /* Not writable = nothing to do. */
5550         if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state))
5551                 return 0;
5552
5553         /* No free space = nothing to do. */
5554         if (device->total_bytes <= device->bytes_used)
5555                 return 0;
5556
5557         ret = 0;
5558
5559         while (1) {
5560                 struct btrfs_fs_info *fs_info = device->fs_info;
5561                 u64 bytes;
5562
5563                 ret = mutex_lock_interruptible(&fs_info->chunk_mutex);
5564                 if (ret)
5565                         break;
5566
5567                 find_first_clear_extent_bit(&device->alloc_state, start,
5568                                             &start, &end,
5569                                             CHUNK_TRIMMED | CHUNK_ALLOCATED);
5570
5571                 /* Ensure we skip the reserved area in the first 1M */
5572                 start = max_t(u64, start, SZ_1M);
5573
5574                 /*
5575                  * If find_first_clear_extent_bit find a range that spans the
5576                  * end of the device it will set end to -1, in this case it's up
5577                  * to the caller to trim the value to the size of the device.
5578                  */
5579                 end = min(end, device->total_bytes - 1);
5580
5581                 len = end - start + 1;
5582
5583                 /* We didn't find any extents */
5584                 if (!len) {
5585                         mutex_unlock(&fs_info->chunk_mutex);
5586                         ret = 0;
5587                         break;
5588                 }
5589
5590                 ret = btrfs_issue_discard(device->bdev, start, len,
5591                                           &bytes);
5592                 if (!ret)
5593                         set_extent_bits(&device->alloc_state, start,
5594                                         start + bytes - 1,
5595                                         CHUNK_TRIMMED);
5596                 mutex_unlock(&fs_info->chunk_mutex);
5597
5598                 if (ret)
5599                         break;
5600
5601                 start += len;
5602                 *trimmed += bytes;
5603
5604                 if (fatal_signal_pending(current)) {
5605                         ret = -ERESTARTSYS;
5606                         break;
5607                 }
5608
5609                 cond_resched();
5610         }
5611
5612         return ret;
5613 }
5614
5615 /*
5616  * Trim the whole filesystem by:
5617  * 1) trimming the free space in each block group
5618  * 2) trimming the unallocated space on each device
5619  *
5620  * This will also continue trimming even if a block group or device encounters
5621  * an error.  The return value will be the last error, or 0 if nothing bad
5622  * happens.
5623  */
5624 int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
5625 {
5626         struct btrfs_block_group_cache *cache = NULL;
5627         struct btrfs_device *device;
5628         struct list_head *devices;
5629         u64 group_trimmed;
5630         u64 range_end = U64_MAX;
5631         u64 start;
5632         u64 end;
5633         u64 trimmed = 0;
5634         u64 bg_failed = 0;
5635         u64 dev_failed = 0;
5636         int bg_ret = 0;
5637         int dev_ret = 0;
5638         int ret = 0;
5639
5640         /*
5641          * Check range overflow if range->len is set.
5642          * The default range->len is U64_MAX.
5643          */
5644         if (range->len != U64_MAX &&
5645             check_add_overflow(range->start, range->len, &range_end))
5646                 return -EINVAL;
5647
5648         cache = btrfs_lookup_first_block_group(fs_info, range->start);
5649         for (; cache; cache = btrfs_next_block_group(cache)) {
5650                 if (cache->key.objectid >= range_end) {
5651                         btrfs_put_block_group(cache);
5652                         break;
5653                 }
5654
5655                 start = max(range->start, cache->key.objectid);
5656                 end = min(range_end, cache->key.objectid + cache->key.offset);
5657
5658                 if (end - start >= range->minlen) {
5659                         if (!btrfs_block_group_cache_done(cache)) {
5660                                 ret = btrfs_cache_block_group(cache, 0);
5661                                 if (ret) {
5662                                         bg_failed++;
5663                                         bg_ret = ret;
5664                                         continue;
5665                                 }
5666                                 ret = btrfs_wait_block_group_cache_done(cache);
5667                                 if (ret) {
5668                                         bg_failed++;
5669                                         bg_ret = ret;
5670                                         continue;
5671                                 }
5672                         }
5673                         ret = btrfs_trim_block_group(cache,
5674                                                      &group_trimmed,
5675                                                      start,
5676                                                      end,
5677                                                      range->minlen);
5678
5679                         trimmed += group_trimmed;
5680                         if (ret) {
5681                                 bg_failed++;
5682                                 bg_ret = ret;
5683                                 continue;
5684                         }
5685                 }
5686         }
5687
5688         if (bg_failed)
5689                 btrfs_warn(fs_info,
5690                         "failed to trim %llu block group(s), last error %d",
5691                         bg_failed, bg_ret);
5692         mutex_lock(&fs_info->fs_devices->device_list_mutex);
5693         devices = &fs_info->fs_devices->devices;
5694         list_for_each_entry(device, devices, dev_list) {
5695                 ret = btrfs_trim_free_extents(device, &group_trimmed);
5696                 if (ret) {
5697                         dev_failed++;
5698                         dev_ret = ret;
5699                         break;
5700                 }
5701
5702                 trimmed += group_trimmed;
5703         }
5704         mutex_unlock(&fs_info->fs_devices->device_list_mutex);
5705
5706         if (dev_failed)
5707                 btrfs_warn(fs_info,
5708                         "failed to trim %llu device(s), last error %d",
5709                         dev_failed, dev_ret);
5710         range->len = trimmed;
5711         if (bg_ret)
5712                 return bg_ret;
5713         return dev_ret;
5714 }
5715
5716 /*
5717  * btrfs_{start,end}_write_no_snapshotting() are similar to
5718  * mnt_{want,drop}_write(), they are used to prevent some tasks from writing
5719  * data into the page cache through nocow before the subvolume is snapshoted,
5720  * but flush the data into disk after the snapshot creation, or to prevent
5721  * operations while snapshotting is ongoing and that cause the snapshot to be
5722  * inconsistent (writes followed by expanding truncates for example).
5723  */
5724 void btrfs_end_write_no_snapshotting(struct btrfs_root *root)
5725 {
5726         percpu_counter_dec(&root->subv_writers->counter);
5727         cond_wake_up(&root->subv_writers->wait);
5728 }
5729
5730 int btrfs_start_write_no_snapshotting(struct btrfs_root *root)
5731 {
5732         if (atomic_read(&root->will_be_snapshotted))
5733                 return 0;
5734
5735         percpu_counter_inc(&root->subv_writers->counter);
5736         /*
5737          * Make sure counter is updated before we check for snapshot creation.
5738          */
5739         smp_mb();
5740         if (atomic_read(&root->will_be_snapshotted)) {
5741                 btrfs_end_write_no_snapshotting(root);
5742                 return 0;
5743         }
5744         return 1;
5745 }
5746
5747 void btrfs_wait_for_snapshot_creation(struct btrfs_root *root)
5748 {
5749         while (true) {
5750                 int ret;
5751
5752                 ret = btrfs_start_write_no_snapshotting(root);
5753                 if (ret)
5754                         break;
5755                 wait_var_event(&root->will_be_snapshotted,
5756                                !atomic_read(&root->will_be_snapshotted));
5757         }
5758 }