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 "ctree.h"
  20 #include "extent-tree.h"
  21 #include "tree-log.h"
  22 #include "disk-io.h"
  23 #include "print-tree.h"
  24 #include "volumes.h"
  25 #include "raid56.h"
  26 #include "locking.h"
  27 #include "free-space-cache.h"
  28 #include "free-space-tree.h"
  29 #include "sysfs.h"
  30 #include "qgroup.h"
  31 #include "ref-verify.h"
  32 #include "space-info.h"
  33 #include "block-rsv.h"
  34 #include "delalloc-space.h"
  35 #include "discard.h"
  36 #include "rcu-string.h"
  37 #include "zoned.h"
  38 #include "dev-replace.h"
  39 #include "fs.h"
  40 #include "accessors.h"
  41 #include "root-tree.h"
  42 #include "file-item.h"
  43 #include "orphan.h"
  44 #include "tree-checker.h"
  45
  46 #undef SCRAMBLE_DELAYED_REFS
  47
  48
  49 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
  50                                struct btrfs_delayed_ref_node *node, u64 parent,
  51                                u64 root_objectid, u64 owner_objectid,
  52                                u64 owner_offset, int refs_to_drop,
  53                                struct btrfs_delayed_extent_op *extra_op);
  54 static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
  55                                     struct extent_buffer *leaf,
  56                                     struct btrfs_extent_item *ei);
  57 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
  58                                       u64 parent, u64 root_objectid,
  59                                       u64 flags, u64 owner, u64 offset,
  60                                       struct btrfs_key *ins, int ref_mod);
  61 static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
  62                                      struct btrfs_delayed_ref_node *node,
  63                                      struct btrfs_delayed_extent_op *extent_op);
  64 static int find_next_key(struct btrfs_path *path, int level,
  65                          struct btrfs_key *key);
  66
  67 static int block_group_bits(struct btrfs_block_group *cache, u64 bits)
  68 {
  69         return (cache->flags & bits) == bits;
  70 }
  71
  72 /* simple helper to search for an existing data extent at a given offset */
  73 int btrfs_lookup_data_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len)
  74 {
  75         struct btrfs_root *root = btrfs_extent_root(fs_info, start);
  76         int ret;
  77         struct btrfs_key key;
  78         struct btrfs_path *path;
  79
  80         path = btrfs_alloc_path();
  81         if (!path)
  82                 return -ENOMEM;
  83
  84         key.objectid = start;
  85         key.offset = len;
  86         key.type = BTRFS_EXTENT_ITEM_KEY;
  87         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
  88         btrfs_free_path(path);
  89         return ret;
  90 }
  91
  92 /*
  93  * helper function to lookup reference count and flags of a tree block.
  94  *
  95  * the head node for delayed ref is used to store the sum of all the
  96  * reference count modifications queued up in the rbtree. the head
  97  * node may also store the extent flags to set. This way you can check
  98  * to see what the reference count and extent flags would be if all of
  99  * the delayed refs are not processed.
 100  */
 101 int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
 102                              struct btrfs_fs_info *fs_info, u64 bytenr,
 103                              u64 offset, int metadata, u64 *refs, u64 *flags)
 104 {
 105         struct btrfs_root *extent_root;
 106         struct btrfs_delayed_ref_head *head;
 107         struct btrfs_delayed_ref_root *delayed_refs;
 108         struct btrfs_path *path;
 109         struct btrfs_extent_item *ei;
 110         struct extent_buffer *leaf;
 111         struct btrfs_key key;
 112         u32 item_size;
 113         u64 num_refs;
 114         u64 extent_flags;
 115         int ret;
 116
 117         /*
 118          * If we don't have skinny metadata, don't bother doing anything
 119          * different
 120          */
 121         if (metadata && !btrfs_fs_incompat(fs_info, SKINNY_METADATA)) {
 122                 offset = fs_info->nodesize;
 123                 metadata = 0;
 124         }
 125
 126         path = btrfs_alloc_path();
 127         if (!path)
 128                 return -ENOMEM;
 129
 130         if (!trans) {
 131                 path->skip_locking = 1;
 132                 path->search_commit_root = 1;
 133         }
 134
 135 search_again:
 136         key.objectid = bytenr;
 137         key.offset = offset;
 138         if (metadata)
 139                 key.type = BTRFS_METADATA_ITEM_KEY;
 140         else
 141                 key.type = BTRFS_EXTENT_ITEM_KEY;
 142
 143         extent_root = btrfs_extent_root(fs_info, bytenr);
 144         ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
 145         if (ret < 0)
 146                 goto out_free;
 147
 148         if (ret > 0 && metadata && key.type == BTRFS_METADATA_ITEM_KEY) {
 149                 if (path->slots[0]) {
 150                         path->slots[0]--;
 151                         btrfs_item_key_to_cpu(path->nodes[0], &key,
 152                                               path->slots[0]);
 153                         if (key.objectid == bytenr &&
 154                             key.type == BTRFS_EXTENT_ITEM_KEY &&
 155                             key.offset == fs_info->nodesize)
 156                                 ret = 0;
 157                 }
 158         }
 159
 160         if (ret == 0) {
 161                 leaf = path->nodes[0];
 162                 item_size = btrfs_item_size(leaf, path->slots[0]);
 163                 if (item_size >= sizeof(*ei)) {
 164                         ei = btrfs_item_ptr(leaf, path->slots[0],
 165                                             struct btrfs_extent_item);
 166                         num_refs = btrfs_extent_refs(leaf, ei);
 167                         extent_flags = btrfs_extent_flags(leaf, ei);
 168                 } else {
 169                         ret = -EUCLEAN;
 170                         btrfs_err(fs_info,
 171                         "unexpected extent item size, has %u expect >= %zu",
 172                                   item_size, sizeof(*ei));
 173                         if (trans)
 174                                 btrfs_abort_transaction(trans, ret);
 175                         else
 176                                 btrfs_handle_fs_error(fs_info, ret, NULL);
 177
 178                         goto out_free;
 179                 }
 180
 181                 BUG_ON(num_refs == 0);
 182         } else {
 183                 num_refs = 0;
 184                 extent_flags = 0;
 185                 ret = 0;
 186         }
 187
 188         if (!trans)
 189                 goto out;
 190
 191         delayed_refs = &trans->transaction->delayed_refs;
 192         spin_lock(&delayed_refs->lock);
 193         head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
 194         if (head) {
 195                 if (!mutex_trylock(&head->mutex)) {
 196                         refcount_inc(&head->refs);
 197                         spin_unlock(&delayed_refs->lock);
 198
 199                         btrfs_release_path(path);
 200
 201                         /*
 202                          * Mutex was contended, block until it's released and try
 203                          * again
 204                          */
 205                         mutex_lock(&head->mutex);
 206                         mutex_unlock(&head->mutex);
 207                         btrfs_put_delayed_ref_head(head);
 208                         goto search_again;
 209                 }
 210                 spin_lock(&head->lock);
 211                 if (head->extent_op && head->extent_op->update_flags)
 212                         extent_flags |= head->extent_op->flags_to_set;
 213                 else
 214                         BUG_ON(num_refs == 0);
 215
 216                 num_refs += head->ref_mod;
 217                 spin_unlock(&head->lock);
 218                 mutex_unlock(&head->mutex);
 219         }
 220         spin_unlock(&delayed_refs->lock);
 221 out:
 222         WARN_ON(num_refs == 0);
 223         if (refs)
 224                 *refs = num_refs;
 225         if (flags)
 226                 *flags = extent_flags;
 227 out_free:
 228         btrfs_free_path(path);
 229         return ret;
 230 }
 231
 232 /*
 233  * Back reference rules.  Back refs have three main goals:
 234  *
 235  * 1) differentiate between all holders of references to an extent so that
 236  *    when a reference is dropped we can make sure it was a valid reference
 237  *    before freeing the extent.
 238  *
 239  * 2) Provide enough information to quickly find the holders of an extent
 240  *    if we notice a given block is corrupted or bad.
 241  *
 242  * 3) Make it easy to migrate blocks for FS shrinking or storage pool
 243  *    maintenance.  This is actually the same as #2, but with a slightly
 244  *    different use case.
 245  *
 246  * There are two kinds of back refs. The implicit back refs is optimized
 247  * for pointers in non-shared tree blocks. For a given pointer in a block,
 248  * back refs of this kind provide information about the block's owner tree
 249  * and the pointer's key. These information allow us to find the block by
 250  * b-tree searching. The full back refs is for pointers in tree blocks not
 251  * referenced by their owner trees. The location of tree block is recorded
 252  * in the back refs. Actually the full back refs is generic, and can be
 253  * used in all cases the implicit back refs is used. The major shortcoming
 254  * of the full back refs is its overhead. Every time a tree block gets
 255  * COWed, we have to update back refs entry for all pointers in it.
 256  *
 257  * For a newly allocated tree block, we use implicit back refs for
 258  * pointers in it. This means most tree related operations only involve
 259  * implicit back refs. For a tree block created in old transaction, the
 260  * only way to drop a reference to it is COW it. So we can detect the
 261  * event that tree block loses its owner tree's reference and do the
 262  * back refs conversion.
 263  *
 264  * When a tree block is COWed through a tree, there are four cases:
 265  *
 266  * The reference count of the block is one and the tree is the block's
 267  * owner tree. Nothing to do in this case.
 268  *
 269  * The reference count of the block is one and the tree is not the
 270  * block's owner tree. In this case, full back refs is used for pointers
 271  * in the block. Remove these full back refs, add implicit back refs for
 272  * every pointers in the new block.
 273  *
 274  * The reference count of the block is greater than one and the tree is
 275  * the block's owner tree. In this case, implicit back refs is used for
 276  * pointers in the block. Add full back refs for every pointers in the
 277  * block, increase lower level extents' reference counts. The original
 278  * implicit back refs are entailed to the new block.
 279  *
 280  * The reference count of the block is greater than one and the tree is
 281  * not the block's owner tree. Add implicit back refs for every pointer in
 282  * the new block, increase lower level extents' reference count.
 283  *
 284  * Back Reference Key composing:
 285  *
 286  * The key objectid corresponds to the first byte in the extent,
 287  * The key type is used to differentiate between types of back refs.
 288  * There are different meanings of the key offset for different types
 289  * of back refs.
 290  *
 291  * File extents can be referenced by:
 292  *
 293  * - multiple snapshots, subvolumes, or different generations in one subvol
 294  * - different files inside a single subvolume
 295  * - different offsets inside a file (bookend extents in file.c)
 296  *
 297  * The extent ref structure for the implicit back refs has fields for:
 298  *
 299  * - Objectid of the subvolume root
 300  * - objectid of the file holding the reference
 301  * - original offset in the file
 302  * - how many bookend extents
 303  *
 304  * The key offset for the implicit back refs is hash of the first
 305  * three fields.
 306  *
 307  * The extent ref structure for the full back refs has field for:
 308  *
 309  * - number of pointers in the tree leaf
 310  *
 311  * The key offset for the implicit back refs is the first byte of
 312  * the tree leaf
 313  *
 314  * When a file extent is allocated, The implicit back refs is used.
 315  * the fields are filled in:
 316  *
 317  *     (root_key.objectid, inode objectid, offset in file, 1)
 318  *
 319  * When a file extent is removed file truncation, we find the
 320  * corresponding implicit back refs and check the following fields:
 321  *
 322  *     (btrfs_header_owner(leaf), inode objectid, offset in file)
 323  *
 324  * Btree extents can be referenced by:
 325  *
 326  * - Different subvolumes
 327  *
 328  * Both the implicit back refs and the full back refs for tree blocks
 329  * only consist of key. The key offset for the implicit back refs is
 330  * objectid of block's owner tree. The key offset for the full back refs
 331  * is the first byte of parent block.
 332  *
 333  * When implicit back refs is used, information about the lowest key and
 334  * level of the tree block are required. These information are stored in
 335  * tree block info structure.
 336  */
 337
 338 /*
 339  * is_data == BTRFS_REF_TYPE_BLOCK, tree block type is required,
 340  * is_data == BTRFS_REF_TYPE_DATA, data type is requiried,
 341  * is_data == BTRFS_REF_TYPE_ANY, either type is OK.
 342  */
 343 int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
 344                                      struct btrfs_extent_inline_ref *iref,
 345                                      enum btrfs_inline_ref_type is_data)
 346 {
 347         int type = btrfs_extent_inline_ref_type(eb, iref);
 348         u64 offset = btrfs_extent_inline_ref_offset(eb, iref);
 349
 350         if (type == BTRFS_TREE_BLOCK_REF_KEY ||
 351             type == BTRFS_SHARED_BLOCK_REF_KEY ||
 352             type == BTRFS_SHARED_DATA_REF_KEY ||
 353             type == BTRFS_EXTENT_DATA_REF_KEY) {
 354                 if (is_data == BTRFS_REF_TYPE_BLOCK) {
 355                         if (type == BTRFS_TREE_BLOCK_REF_KEY)
 356                                 return type;
 357                         if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
 358                                 ASSERT(eb->fs_info);
 359                                 /*
 360                                  * Every shared one has parent tree block,
 361                                  * which must be aligned to sector size.
 362                                  */
 363                                 if (offset &&
 364                                     IS_ALIGNED(offset, eb->fs_info->sectorsize))
 365                                         return type;
 366                         }
 367                 } else if (is_data == BTRFS_REF_TYPE_DATA) {
 368                         if (type == BTRFS_EXTENT_DATA_REF_KEY)
 369                                 return type;
 370                         if (type == BTRFS_SHARED_DATA_REF_KEY) {
 371                                 ASSERT(eb->fs_info);
 372                                 /*
 373                                  * Every shared one has parent tree block,
 374                                  * which must be aligned to sector size.
 375                                  */
 376                                 if (offset &&
 377                                     IS_ALIGNED(offset, eb->fs_info->sectorsize))
 378                                         return type;
 379                         }
 380                 } else {
 381                         ASSERT(is_data == BTRFS_REF_TYPE_ANY);
 382                         return type;
 383                 }
 384         }
 385
 386         WARN_ON(1);
 387         btrfs_print_leaf(eb);
 388         btrfs_err(eb->fs_info,
 389                   "eb %llu iref 0x%lx invalid extent inline ref type %d",
 390                   eb->start, (unsigned long)iref, type);
 391
 392         return BTRFS_REF_TYPE_INVALID;
 393 }
 394
 395 u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset)
 396 {
 397         u32 high_crc = ~(u32)0;
 398         u32 low_crc = ~(u32)0;
 399         __le64 lenum;
 400
 401         lenum = cpu_to_le64(root_objectid);
 402         high_crc = btrfs_crc32c(high_crc, &lenum, sizeof(lenum));
 403         lenum = cpu_to_le64(owner);
 404         low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
 405         lenum = cpu_to_le64(offset);
 406         low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
 407
 408         return ((u64)high_crc << 31) ^ (u64)low_crc;
 409 }
 410
 411 static u64 hash_extent_data_ref_item(struct extent_buffer *leaf,
 412                                      struct btrfs_extent_data_ref *ref)
 413 {
 414         return hash_extent_data_ref(btrfs_extent_data_ref_root(leaf, ref),
 415                                     btrfs_extent_data_ref_objectid(leaf, ref),
 416                                     btrfs_extent_data_ref_offset(leaf, ref));
 417 }
 418
 419 static int match_extent_data_ref(struct extent_buffer *leaf,
 420                                  struct btrfs_extent_data_ref *ref,
 421                                  u64 root_objectid, u64 owner, u64 offset)
 422 {
 423         if (btrfs_extent_data_ref_root(leaf, ref) != root_objectid ||
 424             btrfs_extent_data_ref_objectid(leaf, ref) != owner ||
 425             btrfs_extent_data_ref_offset(leaf, ref) != offset)
 426                 return 0;
 427         return 1;
 428 }
 429
 430 static noinline int lookup_extent_data_ref(struct btrfs_trans_handle *trans,
 431                                            struct btrfs_path *path,
 432                                            u64 bytenr, u64 parent,
 433                                            u64 root_objectid,
 434                                            u64 owner, u64 offset)
 435 {
 436         struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 437         struct btrfs_key key;
 438         struct btrfs_extent_data_ref *ref;
 439         struct extent_buffer *leaf;
 440         u32 nritems;
 441         int ret;
 442         int recow;
 443         int err = -ENOENT;
 444
 445         key.objectid = bytenr;
 446         if (parent) {
 447                 key.type = BTRFS_SHARED_DATA_REF_KEY;
 448                 key.offset = parent;
 449         } else {
 450                 key.type = BTRFS_EXTENT_DATA_REF_KEY;
 451                 key.offset = hash_extent_data_ref(root_objectid,
 452                                                   owner, offset);
 453         }
 454 again:
 455         recow = 0;
 456         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 457         if (ret < 0) {
 458                 err = ret;
 459                 goto fail;
 460         }
 461
 462         if (parent) {
 463                 if (!ret)
 464                         return 0;
 465                 goto fail;
 466         }
 467
 468         leaf = path->nodes[0];
 469         nritems = btrfs_header_nritems(leaf);
 470         while (1) {
 471                 if (path->slots[0] >= nritems) {
 472                         ret = btrfs_next_leaf(root, path);
 473                         if (ret < 0)
 474                                 err = ret;
 475                         if (ret)
 476                                 goto fail;
 477
 478                         leaf = path->nodes[0];
 479                         nritems = btrfs_header_nritems(leaf);
 480                         recow = 1;
 481                 }
 482
 483                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 484                 if (key.objectid != bytenr ||
 485                     key.type != BTRFS_EXTENT_DATA_REF_KEY)
 486                         goto fail;
 487
 488                 ref = btrfs_item_ptr(leaf, path->slots[0],
 489                                      struct btrfs_extent_data_ref);
 490
 491                 if (match_extent_data_ref(leaf, ref, root_objectid,
 492                                           owner, offset)) {
 493                         if (recow) {
 494                                 btrfs_release_path(path);
 495                                 goto again;
 496                         }
 497                         err = 0;
 498                         break;
 499                 }
 500                 path->slots[0]++;
 501         }
 502 fail:
 503         return err;
 504 }
 505
 506 static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans,
 507                                            struct btrfs_path *path,
 508                                            u64 bytenr, u64 parent,
 509                                            u64 root_objectid, u64 owner,
 510                                            u64 offset, int refs_to_add)
 511 {
 512         struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 513         struct btrfs_key key;
 514         struct extent_buffer *leaf;
 515         u32 size;
 516         u32 num_refs;
 517         int ret;
 518
 519         key.objectid = bytenr;
 520         if (parent) {
 521                 key.type = BTRFS_SHARED_DATA_REF_KEY;
 522                 key.offset = parent;
 523                 size = sizeof(struct btrfs_shared_data_ref);
 524         } else {
 525                 key.type = BTRFS_EXTENT_DATA_REF_KEY;
 526                 key.offset = hash_extent_data_ref(root_objectid,
 527                                                   owner, offset);
 528                 size = sizeof(struct btrfs_extent_data_ref);
 529         }
 530
 531         ret = btrfs_insert_empty_item(trans, root, path, &key, size);
 532         if (ret && ret != -EEXIST)
 533                 goto fail;
 534
 535         leaf = path->nodes[0];
 536         if (parent) {
 537                 struct btrfs_shared_data_ref *ref;
 538                 ref = btrfs_item_ptr(leaf, path->slots[0],
 539                                      struct btrfs_shared_data_ref);
 540                 if (ret == 0) {
 541                         btrfs_set_shared_data_ref_count(leaf, ref, refs_to_add);
 542                 } else {
 543                         num_refs = btrfs_shared_data_ref_count(leaf, ref);
 544                         num_refs += refs_to_add;
 545                         btrfs_set_shared_data_ref_count(leaf, ref, num_refs);
 546                 }
 547         } else {
 548                 struct btrfs_extent_data_ref *ref;
 549                 while (ret == -EEXIST) {
 550                         ref = btrfs_item_ptr(leaf, path->slots[0],
 551                                              struct btrfs_extent_data_ref);
 552                         if (match_extent_data_ref(leaf, ref, root_objectid,
 553                                                   owner, offset))
 554                                 break;
 555                         btrfs_release_path(path);
 556                         key.offset++;
 557                         ret = btrfs_insert_empty_item(trans, root, path, &key,
 558                                                       size);
 559                         if (ret && ret != -EEXIST)
 560                                 goto fail;
 561
 562                         leaf = path->nodes[0];
 563                 }
 564                 ref = btrfs_item_ptr(leaf, path->slots[0],
 565                                      struct btrfs_extent_data_ref);
 566                 if (ret == 0) {
 567                         btrfs_set_extent_data_ref_root(leaf, ref,
 568                                                        root_objectid);
 569                         btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
 570                         btrfs_set_extent_data_ref_offset(leaf, ref, offset);
 571                         btrfs_set_extent_data_ref_count(leaf, ref, refs_to_add);
 572                 } else {
 573                         num_refs = btrfs_extent_data_ref_count(leaf, ref);
 574                         num_refs += refs_to_add;
 575                         btrfs_set_extent_data_ref_count(leaf, ref, num_refs);
 576                 }
 577         }
 578         btrfs_mark_buffer_dirty(leaf);
 579         ret = 0;
 580 fail:
 581         btrfs_release_path(path);
 582         return ret;
 583 }
 584
 585 static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
 586                                            struct btrfs_root *root,
 587                                            struct btrfs_path *path,
 588                                            int refs_to_drop)
 589 {
 590         struct btrfs_key key;
 591         struct btrfs_extent_data_ref *ref1 = NULL;
 592         struct btrfs_shared_data_ref *ref2 = NULL;
 593         struct extent_buffer *leaf;
 594         u32 num_refs = 0;
 595         int ret = 0;
 596
 597         leaf = path->nodes[0];
 598         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 599
 600         if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
 601                 ref1 = btrfs_item_ptr(leaf, path->slots[0],
 602                                       struct btrfs_extent_data_ref);
 603                 num_refs = btrfs_extent_data_ref_count(leaf, ref1);
 604         } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
 605                 ref2 = btrfs_item_ptr(leaf, path->slots[0],
 606                                       struct btrfs_shared_data_ref);
 607                 num_refs = btrfs_shared_data_ref_count(leaf, ref2);
 608         } else {
 609                 btrfs_err(trans->fs_info,
 610                           "unrecognized backref key (%llu %u %llu)",
 611                           key.objectid, key.type, key.offset);
 612                 btrfs_abort_transaction(trans, -EUCLEAN);
 613                 return -EUCLEAN;
 614         }
 615
 616         BUG_ON(num_refs < refs_to_drop);
 617         num_refs -= refs_to_drop;
 618
 619         if (num_refs == 0) {
 620                 ret = btrfs_del_item(trans, root, path);
 621         } else {
 622                 if (key.type == BTRFS_EXTENT_DATA_REF_KEY)
 623                         btrfs_set_extent_data_ref_count(leaf, ref1, num_refs);
 624                 else if (key.type == BTRFS_SHARED_DATA_REF_KEY)
 625                         btrfs_set_shared_data_ref_count(leaf, ref2, num_refs);
 626                 btrfs_mark_buffer_dirty(leaf);
 627         }
 628         return ret;
 629 }
 630
 631 static noinline u32 extent_data_ref_count(struct btrfs_path *path,
 632                                           struct btrfs_extent_inline_ref *iref)
 633 {
 634         struct btrfs_key key;
 635         struct extent_buffer *leaf;
 636         struct btrfs_extent_data_ref *ref1;
 637         struct btrfs_shared_data_ref *ref2;
 638         u32 num_refs = 0;
 639         int type;
 640
 641         leaf = path->nodes[0];
 642         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 643
 644         if (iref) {
 645                 /*
 646                  * If type is invalid, we should have bailed out earlier than
 647                  * this call.
 648                  */
 649                 type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
 650                 ASSERT(type != BTRFS_REF_TYPE_INVALID);
 651                 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
 652                         ref1 = (struct btrfs_extent_data_ref *)(&iref->offset);
 653                         num_refs = btrfs_extent_data_ref_count(leaf, ref1);
 654                 } else {
 655                         ref2 = (struct btrfs_shared_data_ref *)(iref + 1);
 656                         num_refs = btrfs_shared_data_ref_count(leaf, ref2);
 657                 }
 658         } else if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
 659                 ref1 = btrfs_item_ptr(leaf, path->slots[0],
 660                                       struct btrfs_extent_data_ref);
 661                 num_refs = btrfs_extent_data_ref_count(leaf, ref1);
 662         } else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
 663                 ref2 = btrfs_item_ptr(leaf, path->slots[0],
 664                                       struct btrfs_shared_data_ref);
 665                 num_refs = btrfs_shared_data_ref_count(leaf, ref2);
 666         } else {
 667                 WARN_ON(1);
 668         }
 669         return num_refs;
 670 }
 671
 672 static noinline int lookup_tree_block_ref(struct btrfs_trans_handle *trans,
 673                                           struct btrfs_path *path,
 674                                           u64 bytenr, u64 parent,
 675                                           u64 root_objectid)
 676 {
 677         struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 678         struct btrfs_key key;
 679         int ret;
 680
 681         key.objectid = bytenr;
 682         if (parent) {
 683                 key.type = BTRFS_SHARED_BLOCK_REF_KEY;
 684                 key.offset = parent;
 685         } else {
 686                 key.type = BTRFS_TREE_BLOCK_REF_KEY;
 687                 key.offset = root_objectid;
 688         }
 689
 690         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 691         if (ret > 0)
 692                 ret = -ENOENT;
 693         return ret;
 694 }
 695
 696 static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans,
 697                                           struct btrfs_path *path,
 698                                           u64 bytenr, u64 parent,
 699                                           u64 root_objectid)
 700 {
 701         struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
 702         struct btrfs_key key;
 703         int ret;
 704
 705         key.objectid = bytenr;
 706         if (parent) {
 707                 key.type = BTRFS_SHARED_BLOCK_REF_KEY;
 708                 key.offset = parent;
 709         } else {
 710                 key.type = BTRFS_TREE_BLOCK_REF_KEY;
 711                 key.offset = root_objectid;
 712         }
 713
 714         ret = btrfs_insert_empty_item(trans, root, path, &key, 0);
 715         btrfs_release_path(path);
 716         return ret;
 717 }
 718
 719 static inline int extent_ref_type(u64 parent, u64 owner)
 720 {
 721         int type;
 722         if (owner < BTRFS_FIRST_FREE_OBJECTID) {
 723                 if (parent > 0)
 724                         type = BTRFS_SHARED_BLOCK_REF_KEY;
 725                 else
 726                         type = BTRFS_TREE_BLOCK_REF_KEY;
 727         } else {
 728                 if (parent > 0)
 729                         type = BTRFS_SHARED_DATA_REF_KEY;
 730                 else
 731                         type = BTRFS_EXTENT_DATA_REF_KEY;
 732         }
 733         return type;
 734 }
 735
 736 static int find_next_key(struct btrfs_path *path, int level,
 737                          struct btrfs_key *key)
 738
 739 {
 740         for (; level < BTRFS_MAX_LEVEL; level++) {
 741                 if (!path->nodes[level])
 742                         break;
 743                 if (path->slots[level] + 1 >=
 744                     btrfs_header_nritems(path->nodes[level]))
 745                         continue;
 746                 if (level == 0)
 747                         btrfs_item_key_to_cpu(path->nodes[level], key,
 748                                               path->slots[level] + 1);
 749                 else
 750                         btrfs_node_key_to_cpu(path->nodes[level], key,
 751                                               path->slots[level] + 1);
 752                 return 0;
 753         }
 754         return 1;
 755 }
 756
 757 /*
 758  * look for inline back ref. if back ref is found, *ref_ret is set
 759  * to the address of inline back ref, and 0 is returned.
 760  *
 761  * if back ref isn't found, *ref_ret is set to the address where it
 762  * should be inserted, and -ENOENT is returned.
 763  *
 764  * if insert is true and there are too many inline back refs, the path
 765  * points to the extent item, and -EAGAIN is returned.
 766  *
 767  * NOTE: inline back refs are ordered in the same way that back ref
 768  *       items in the tree are ordered.
 769  */
 770 static noinline_for_stack
 771 int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
 772                                  struct btrfs_path *path,
 773                                  struct btrfs_extent_inline_ref **ref_ret,
 774                                  u64 bytenr, u64 num_bytes,
 775                                  u64 parent, u64 root_objectid,
 776                                  u64 owner, u64 offset, int insert)
 777 {
 778         struct btrfs_fs_info *fs_info = trans->fs_info;
 779         struct btrfs_root *root = btrfs_extent_root(fs_info, bytenr);
 780         struct btrfs_key key;
 781         struct extent_buffer *leaf;
 782         struct btrfs_extent_item *ei;
 783         struct btrfs_extent_inline_ref *iref;
 784         u64 flags;
 785         u64 item_size;
 786         unsigned long ptr;
 787         unsigned long end;
 788         int extra_size;
 789         int type;
 790         int want;
 791         int ret;
 792         int err = 0;
 793         bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
 794         int needed;
 795
 796         key.objectid = bytenr;
 797         key.type = BTRFS_EXTENT_ITEM_KEY;
 798         key.offset = num_bytes;
 799
 800         want = extent_ref_type(parent, owner);
 801         if (insert) {
 802                 extra_size = btrfs_extent_inline_ref_size(want);
 803                 path->search_for_extension = 1;
 804                 path->keep_locks = 1;
 805         } else
 806                 extra_size = -1;
 807
 808         /*
 809          * Owner is our level, so we can just add one to get the level for the
 810          * block we are interested in.
 811          */
 812         if (skinny_metadata && owner < BTRFS_FIRST_FREE_OBJECTID) {
 813                 key.type = BTRFS_METADATA_ITEM_KEY;
 814                 key.offset = owner;
 815         }
 816
 817 again:
 818         ret = btrfs_search_slot(trans, root, &key, path, extra_size, 1);
 819         if (ret < 0) {
 820                 err = ret;
 821                 goto out;
 822         }
 823
 824         /*
 825          * We may be a newly converted file system which still has the old fat
 826          * extent entries for metadata, so try and see if we have one of those.
 827          */
 828         if (ret > 0 && skinny_metadata) {
 829                 skinny_metadata = false;
 830                 if (path->slots[0]) {
 831                         path->slots[0]--;
 832                         btrfs_item_key_to_cpu(path->nodes[0], &key,
 833                                               path->slots[0]);
 834                         if (key.objectid == bytenr &&
 835                             key.type == BTRFS_EXTENT_ITEM_KEY &&
 836                             key.offset == num_bytes)
 837                                 ret = 0;
 838                 }
 839                 if (ret) {
 840                         key.objectid = bytenr;
 841                         key.type = BTRFS_EXTENT_ITEM_KEY;
 842                         key.offset = num_bytes;
 843                         btrfs_release_path(path);
 844                         goto again;
 845                 }
 846         }
 847
 848         if (ret && !insert) {
 849                 err = -ENOENT;
 850                 goto out;
 851         } else if (WARN_ON(ret)) {
 852                 btrfs_print_leaf(path->nodes[0]);
 853                 btrfs_err(fs_info,
 854 "extent item not found for insert, bytenr %llu num_bytes %llu parent %llu root_objectid %llu owner %llu offset %llu",
 855                           bytenr, num_bytes, parent, root_objectid, owner,
 856                           offset);
 857                 err = -EIO;
 858                 goto out;
 859         }
 860
 861         leaf = path->nodes[0];
 862         item_size = btrfs_item_size(leaf, path->slots[0]);
 863         if (unlikely(item_size < sizeof(*ei))) {
 864                 err = -EUCLEAN;
 865                 btrfs_err(fs_info,
 866                           "unexpected extent item size, has %llu expect >= %zu",
 867                           item_size, sizeof(*ei));
 868                 btrfs_abort_transaction(trans, err);
 869                 goto out;
 870         }
 871
 872         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
 873         flags = btrfs_extent_flags(leaf, ei);
 874
 875         ptr = (unsigned long)(ei + 1);
 876         end = (unsigned long)ei + item_size;
 877
 878         if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK && !skinny_metadata) {
 879                 ptr += sizeof(struct btrfs_tree_block_info);
 880                 BUG_ON(ptr > end);
 881         }
 882
 883         if (owner >= BTRFS_FIRST_FREE_OBJECTID)
 884                 needed = BTRFS_REF_TYPE_DATA;
 885         else
 886                 needed = BTRFS_REF_TYPE_BLOCK;
 887
 888         err = -ENOENT;
 889         while (1) {
 890                 if (ptr >= end) {
 891                         if (ptr > end) {
 892                                 err = -EUCLEAN;
 893                                 btrfs_print_leaf(path->nodes[0]);
 894                                 btrfs_crit(fs_info,
 895 "overrun extent record at slot %d while looking for inline extent for root %llu owner %llu offset %llu parent %llu",
 896                                         path->slots[0], root_objectid, owner, offset, parent);
 897                         }
 898                         break;
 899                 }
 900                 iref = (struct btrfs_extent_inline_ref *)ptr;
 901                 type = btrfs_get_extent_inline_ref_type(leaf, iref, needed);
 902                 if (type == BTRFS_REF_TYPE_INVALID) {
 903                         err = -EUCLEAN;
 904                         goto out;
 905                 }
 906
 907                 if (want < type)
 908                         break;
 909                 if (want > type) {
 910                         ptr += btrfs_extent_inline_ref_size(type);
 911                         continue;
 912                 }
 913
 914                 if (type == BTRFS_EXTENT_DATA_REF_KEY) {
 915                         struct btrfs_extent_data_ref *dref;
 916                         dref = (struct btrfs_extent_data_ref *)(&iref->offset);
 917                         if (match_extent_data_ref(leaf, dref, root_objectid,
 918                                                   owner, offset)) {
 919                                 err = 0;
 920                                 break;
 921                         }
 922                         if (hash_extent_data_ref_item(leaf, dref) <
 923                             hash_extent_data_ref(root_objectid, owner, offset))
 924                                 break;
 925                 } else {
 926                         u64 ref_offset;
 927                         ref_offset = btrfs_extent_inline_ref_offset(leaf, iref);
 928                         if (parent > 0) {
 929                                 if (parent == ref_offset) {
 930                                         err = 0;
 931                                         break;
 932                                 }
 933                                 if (ref_offset < parent)
 934                                         break;
 935                         } else {
 936                                 if (root_objectid == ref_offset) {
 937                                         err = 0;
 938                                         break;
 939                                 }
 940                                 if (ref_offset < root_objectid)
 941                                         break;
 942                         }
 943                 }
 944                 ptr += btrfs_extent_inline_ref_size(type);
 945         }
 946         if (err == -ENOENT && insert) {
 947                 if (item_size + extra_size >=
 948                     BTRFS_MAX_EXTENT_ITEM_SIZE(root)) {
 949                         err = -EAGAIN;
 950                         goto out;
 951                 }
 952                 /*
 953                  * To add new inline back ref, we have to make sure
 954                  * there is no corresponding back ref item.
 955                  * For simplicity, we just do not add new inline back
 956                  * ref if there is any kind of item for this block
 957                  */
 958                 if (find_next_key(path, 0, &key) == 0 &&
 959                     key.objectid == bytenr &&
 960                     key.type < BTRFS_BLOCK_GROUP_ITEM_KEY) {
 961                         err = -EAGAIN;
 962                         goto out;
 963                 }
 964         }
 965         *ref_ret = (struct btrfs_extent_inline_ref *)ptr;
 966 out:
 967         if (insert) {
 968                 path->keep_locks = 0;
 969                 path->search_for_extension = 0;
 970                 btrfs_unlock_up_safe(path, 1);
 971         }
 972         return err;
 973 }
 974
 975 /*
 976  * helper to add new inline back ref
 977  */
 978 static noinline_for_stack
 979 void setup_inline_extent_backref(struct btrfs_fs_info *fs_info,
 980                                  struct btrfs_path *path,
 981                                  struct btrfs_extent_inline_ref *iref,
 982                                  u64 parent, u64 root_objectid,
 983                                  u64 owner, u64 offset, int refs_to_add,
 984                                  struct btrfs_delayed_extent_op *extent_op)
 985 {
 986         struct extent_buffer *leaf;
 987         struct btrfs_extent_item *ei;
 988         unsigned long ptr;
 989         unsigned long end;
 990         unsigned long item_offset;
 991         u64 refs;
 992         int size;
 993         int type;
 994
 995         leaf = path->nodes[0];
 996         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
 997         item_offset = (unsigned long)iref - (unsigned long)ei;
 998
 999         type = extent_ref_type(parent, owner);
1000         size = btrfs_extent_inline_ref_size(type);
1001
1002         btrfs_extend_item(path, size);
1003
1004         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1005         refs = btrfs_extent_refs(leaf, ei);
1006         refs += refs_to_add;
1007         btrfs_set_extent_refs(leaf, ei, refs);
1008         if (extent_op)
1009                 __run_delayed_extent_op(extent_op, leaf, ei);
1010
1011         ptr = (unsigned long)ei + item_offset;
1012         end = (unsigned long)ei + btrfs_item_size(leaf, path->slots[0]);
1013         if (ptr < end - size)
1014                 memmove_extent_buffer(leaf, ptr + size, ptr,
1015                                       end - size - ptr);
1016
1017         iref = (struct btrfs_extent_inline_ref *)ptr;
1018         btrfs_set_extent_inline_ref_type(leaf, iref, type);
1019         if (type == BTRFS_EXTENT_DATA_REF_KEY) {
1020                 struct btrfs_extent_data_ref *dref;
1021                 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
1022                 btrfs_set_extent_data_ref_root(leaf, dref, root_objectid);
1023                 btrfs_set_extent_data_ref_objectid(leaf, dref, owner);
1024                 btrfs_set_extent_data_ref_offset(leaf, dref, offset);
1025                 btrfs_set_extent_data_ref_count(leaf, dref, refs_to_add);
1026         } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
1027                 struct btrfs_shared_data_ref *sref;
1028                 sref = (struct btrfs_shared_data_ref *)(iref + 1);
1029                 btrfs_set_shared_data_ref_count(leaf, sref, refs_to_add);
1030                 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
1031         } else if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
1032                 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
1033         } else {
1034                 btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid);
1035         }
1036         btrfs_mark_buffer_dirty(leaf);
1037 }
1038
1039 static int lookup_extent_backref(struct btrfs_trans_handle *trans,
1040                                  struct btrfs_path *path,
1041                                  struct btrfs_extent_inline_ref **ref_ret,
1042                                  u64 bytenr, u64 num_bytes, u64 parent,
1043                                  u64 root_objectid, u64 owner, u64 offset)
1044 {
1045         int ret;
1046
1047         ret = lookup_inline_extent_backref(trans, path, ref_ret, bytenr,
1048                                            num_bytes, parent, root_objectid,
1049                                            owner, offset, 0);
1050         if (ret != -ENOENT)
1051                 return ret;
1052
1053         btrfs_release_path(path);
1054         *ref_ret = NULL;
1055
1056         if (owner < BTRFS_FIRST_FREE_OBJECTID) {
1057                 ret = lookup_tree_block_ref(trans, path, bytenr, parent,
1058                                             root_objectid);
1059         } else {
1060                 ret = lookup_extent_data_ref(trans, path, bytenr, parent,
1061                                              root_objectid, owner, offset);
1062         }
1063         return ret;
1064 }
1065
1066 /*
1067  * helper to update/remove inline back ref
1068  */
1069 static noinline_for_stack int update_inline_extent_backref(struct btrfs_path *path,
1070                                   struct btrfs_extent_inline_ref *iref,
1071                                   int refs_to_mod,
1072                                   struct btrfs_delayed_extent_op *extent_op)
1073 {
1074         struct extent_buffer *leaf = path->nodes[0];
1075         struct btrfs_fs_info *fs_info = leaf->fs_info;
1076         struct btrfs_extent_item *ei;
1077         struct btrfs_extent_data_ref *dref = NULL;
1078         struct btrfs_shared_data_ref *sref = NULL;
1079         unsigned long ptr;
1080         unsigned long end;
1081         u32 item_size;
1082         int size;
1083         int type;
1084         u64 refs;
1085
1086         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1087         refs = btrfs_extent_refs(leaf, ei);
1088         if (unlikely(refs_to_mod < 0 && refs + refs_to_mod <= 0)) {
1089                 struct btrfs_key key;
1090                 u32 extent_size;
1091
1092                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1093                 if (key.type == BTRFS_METADATA_ITEM_KEY)
1094                         extent_size = fs_info->nodesize;
1095                 else
1096                         extent_size = key.offset;
1097                 btrfs_print_leaf(leaf);
1098                 btrfs_err(fs_info,
1099         "invalid refs_to_mod for extent %llu num_bytes %u, has %d expect >= -%llu",
1100                           key.objectid, extent_size, refs_to_mod, refs);
1101                 return -EUCLEAN;
1102         }
1103         refs += refs_to_mod;
1104         btrfs_set_extent_refs(leaf, ei, refs);
1105         if (extent_op)
1106                 __run_delayed_extent_op(extent_op, leaf, ei);
1107
1108         type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_ANY);
1109         /*
1110          * Function btrfs_get_extent_inline_ref_type() has already printed
1111          * error messages.
1112          */
1113         if (unlikely(type == BTRFS_REF_TYPE_INVALID))
1114                 return -EUCLEAN;
1115
1116         if (type == BTRFS_EXTENT_DATA_REF_KEY) {
1117                 dref = (struct btrfs_extent_data_ref *)(&iref->offset);
1118                 refs = btrfs_extent_data_ref_count(leaf, dref);
1119         } else if (type == BTRFS_SHARED_DATA_REF_KEY) {
1120                 sref = (struct btrfs_shared_data_ref *)(iref + 1);
1121                 refs = btrfs_shared_data_ref_count(leaf, sref);
1122         } else {
1123                 refs = 1;
1124                 /*
1125                  * For tree blocks we can only drop one ref for it, and tree
1126                  * blocks should not have refs > 1.
1127                  *
1128                  * Furthermore if we're inserting a new inline backref, we
1129                  * won't reach this path either. That would be
1130                  * setup_inline_extent_backref().
1131                  */
1132                 if (unlikely(refs_to_mod != -1)) {
1133                         struct btrfs_key key;
1134
1135                         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1136
1137                         btrfs_print_leaf(leaf);
1138                         btrfs_err(fs_info,
1139                         "invalid refs_to_mod for tree block %llu, has %d expect -1",
1140                                   key.objectid, refs_to_mod);
1141                         return -EUCLEAN;
1142                 }
1143         }
1144
1145         if (unlikely(refs_to_mod < 0 && refs < -refs_to_mod)) {
1146                 struct btrfs_key key;
1147                 u32 extent_size;
1148
1149                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1150                 if (key.type == BTRFS_METADATA_ITEM_KEY)
1151                         extent_size = fs_info->nodesize;
1152                 else
1153                         extent_size = key.offset;
1154                 btrfs_print_leaf(leaf);
1155                 btrfs_err(fs_info,
1156 "invalid refs_to_mod for backref entry, iref %lu extent %llu num_bytes %u, has %d expect >= -%llu",
1157                           (unsigned long)iref, key.objectid, extent_size,
1158                           refs_to_mod, refs);
1159                 return -EUCLEAN;
1160         }
1161         refs += refs_to_mod;
1162
1163         if (refs > 0) {
1164                 if (type == BTRFS_EXTENT_DATA_REF_KEY)
1165                         btrfs_set_extent_data_ref_count(leaf, dref, refs);
1166                 else
1167                         btrfs_set_shared_data_ref_count(leaf, sref, refs);
1168         } else {
1169                 size =  btrfs_extent_inline_ref_size(type);
1170                 item_size = btrfs_item_size(leaf, path->slots[0]);
1171                 ptr = (unsigned long)iref;
1172                 end = (unsigned long)ei + item_size;
1173                 if (ptr + size < end)
1174                         memmove_extent_buffer(leaf, ptr, ptr + size,
1175                                               end - ptr - size);
1176                 item_size -= size;
1177                 btrfs_truncate_item(path, item_size, 1);
1178         }
1179         btrfs_mark_buffer_dirty(leaf);
1180         return 0;
1181 }
1182
1183 static noinline_for_stack
1184 int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
1185                                  struct btrfs_path *path,
1186                                  u64 bytenr, u64 num_bytes, u64 parent,
1187                                  u64 root_objectid, u64 owner,
1188                                  u64 offset, int refs_to_add,
1189                                  struct btrfs_delayed_extent_op *extent_op)
1190 {
1191         struct btrfs_extent_inline_ref *iref;
1192         int ret;
1193
1194         ret = lookup_inline_extent_backref(trans, path, &iref, bytenr,
1195                                            num_bytes, parent, root_objectid,
1196                                            owner, offset, 1);
1197         if (ret == 0) {
1198                 /*
1199                  * We're adding refs to a tree block we already own, this
1200                  * should not happen at all.
1201                  */
1202                 if (owner < BTRFS_FIRST_FREE_OBJECTID) {
1203                         btrfs_print_leaf(path->nodes[0]);
1204                         btrfs_crit(trans->fs_info,
1205 "adding refs to an existing tree ref, bytenr %llu num_bytes %llu root_objectid %llu slot %u",
1206                                    bytenr, num_bytes, root_objectid, path->slots[0]);
1207                         return -EUCLEAN;
1208                 }
1209                 ret = update_inline_extent_backref(path, iref, refs_to_add, extent_op);
1210         } else if (ret == -ENOENT) {
1211                 setup_inline_extent_backref(trans->fs_info, path, iref, parent,
1212                                             root_objectid, owner, offset,
1213                                             refs_to_add, extent_op);
1214                 ret = 0;
1215         }
1216         return ret;
1217 }
1218
1219 static int remove_extent_backref(struct btrfs_trans_handle *trans,
1220                                  struct btrfs_root *root,
1221                                  struct btrfs_path *path,
1222                                  struct btrfs_extent_inline_ref *iref,
1223                                  int refs_to_drop, int is_data)
1224 {
1225         int ret = 0;
1226
1227         BUG_ON(!is_data && refs_to_drop != 1);
1228         if (iref)
1229                 ret = update_inline_extent_backref(path, iref, -refs_to_drop, NULL);
1230         else if (is_data)
1231                 ret = remove_extent_data_ref(trans, root, path, refs_to_drop);
1232         else
1233                 ret = btrfs_del_item(trans, root, path);
1234         return ret;
1235 }
1236
1237 static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len,
1238                                u64 *discarded_bytes)
1239 {
1240         int j, ret = 0;
1241         u64 bytes_left, end;
1242         u64 aligned_start = ALIGN(start, 1 << SECTOR_SHIFT);
1243
1244         if (WARN_ON(start != aligned_start)) {
1245                 len -= aligned_start - start;
1246                 len = round_down(len, 1 << SECTOR_SHIFT);
1247                 start = aligned_start;
1248         }
1249
1250         *discarded_bytes = 0;
1251
1252         if (!len)
1253                 return 0;
1254
1255         end = start + len;
1256         bytes_left = len;
1257
1258         /* Skip any superblocks on this device. */
1259         for (j = 0; j < BTRFS_SUPER_MIRROR_MAX; j++) {
1260                 u64 sb_start = btrfs_sb_offset(j);
1261                 u64 sb_end = sb_start + BTRFS_SUPER_INFO_SIZE;
1262                 u64 size = sb_start - start;
1263
1264                 if (!in_range(sb_start, start, bytes_left) &&
1265                     !in_range(sb_end, start, bytes_left) &&
1266                     !in_range(start, sb_start, BTRFS_SUPER_INFO_SIZE))
1267                         continue;
1268
1269                 /*
1270                  * Superblock spans beginning of range.  Adjust start and
1271                  * try again.
1272                  */
1273                 if (sb_start <= start) {
1274                         start += sb_end - start;
1275                         if (start > end) {
1276                                 bytes_left = 0;
1277                                 break;
1278                         }
1279                         bytes_left = end - start;
1280                         continue;
1281                 }
1282
1283                 if (size) {
1284                         ret = blkdev_issue_discard(bdev, start >> SECTOR_SHIFT,
1285                                                    size >> SECTOR_SHIFT,
1286                                                    GFP_NOFS);
1287                         if (!ret)
1288                                 *discarded_bytes += size;
1289                         else if (ret != -EOPNOTSUPP)
1290                                 return ret;
1291                 }
1292
1293                 start = sb_end;
1294                 if (start > end) {
1295                         bytes_left = 0;
1296                         break;
1297                 }
1298                 bytes_left = end - start;
1299         }
1300
1301         if (bytes_left) {
1302                 ret = blkdev_issue_discard(bdev, start >> SECTOR_SHIFT,
1303                                            bytes_left >> SECTOR_SHIFT,
1304                                            GFP_NOFS);
1305                 if (!ret)
1306                         *discarded_bytes += bytes_left;
1307         }
1308         return ret;
1309 }
1310
1311 static int do_discard_extent(struct btrfs_discard_stripe *stripe, u64 *bytes)
1312 {
1313         struct btrfs_device *dev = stripe->dev;
1314         struct btrfs_fs_info *fs_info = dev->fs_info;
1315         struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
1316         u64 phys = stripe->physical;
1317         u64 len = stripe->length;
1318         u64 discarded = 0;
1319         int ret = 0;
1320
1321         /* Zone reset on a zoned filesystem */
1322         if (btrfs_can_zone_reset(dev, phys, len)) {
1323                 u64 src_disc;
1324
1325                 ret = btrfs_reset_device_zone(dev, phys, len, &discarded);
1326                 if (ret)
1327                         goto out;
1328
1329                 if (!btrfs_dev_replace_is_ongoing(dev_replace) ||
1330                     dev != dev_replace->srcdev)
1331                         goto out;
1332
1333                 src_disc = discarded;
1334
1335                 /* Send to replace target as well */
1336                 ret = btrfs_reset_device_zone(dev_replace->tgtdev, phys, len,
1337                                               &discarded);
1338                 discarded += src_disc;
1339         } else if (bdev_max_discard_sectors(stripe->dev->bdev)) {
1340                 ret = btrfs_issue_discard(dev->bdev, phys, len, &discarded);
1341         } else {
1342                 ret = 0;
1343                 *bytes = 0;
1344         }
1345
1346 out:
1347         *bytes = discarded;
1348         return ret;
1349 }
1350
1351 int btrfs_discard_extent(struct btrfs_fs_info *fs_info, u64 bytenr,
1352                          u64 num_bytes, u64 *actual_bytes)
1353 {
1354         int ret = 0;
1355         u64 discarded_bytes = 0;
1356         u64 end = bytenr + num_bytes;
1357         u64 cur = bytenr;
1358
1359         /*
1360          * Avoid races with device replace and make sure the devices in the
1361          * stripes don't go away while we are discarding.
1362          */
1363         btrfs_bio_counter_inc_blocked(fs_info);
1364         while (cur < end) {
1365                 struct btrfs_discard_stripe *stripes;
1366                 unsigned int num_stripes;
1367                 int i;
1368
1369                 num_bytes = end - cur;
1370                 stripes = btrfs_map_discard(fs_info, cur, &num_bytes, &num_stripes);
1371                 if (IS_ERR(stripes)) {
1372                         ret = PTR_ERR(stripes);
1373                         if (ret == -EOPNOTSUPP)
1374                                 ret = 0;
1375                         break;
1376                 }
1377
1378                 for (i = 0; i < num_stripes; i++) {
1379                         struct btrfs_discard_stripe *stripe = stripes + i;
1380                         u64 bytes;
1381
1382                         if (!stripe->dev->bdev) {
1383                                 ASSERT(btrfs_test_opt(fs_info, DEGRADED));
1384                                 continue;
1385                         }
1386
1387                         if (!test_bit(BTRFS_DEV_STATE_WRITEABLE,
1388                                         &stripe->dev->dev_state))
1389                                 continue;
1390
1391                         ret = do_discard_extent(stripe, &bytes);
1392                         if (ret) {
1393                                 /*
1394                                  * Keep going if discard is not supported by the
1395                                  * device.
1396                                  */
1397                                 if (ret != -EOPNOTSUPP)
1398                                         break;
1399                                 ret = 0;
1400                         } else {
1401                                 discarded_bytes += bytes;
1402                         }
1403                 }
1404                 kfree(stripes);
1405                 if (ret)
1406                         break;
1407                 cur += num_bytes;
1408         }
1409         btrfs_bio_counter_dec(fs_info);
1410         if (actual_bytes)
1411                 *actual_bytes = discarded_bytes;
1412         return ret;
1413 }
1414
1415 /* Can return -ENOMEM */
1416 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1417                          struct btrfs_ref *generic_ref)
1418 {
1419         struct btrfs_fs_info *fs_info = trans->fs_info;
1420         int ret;
1421
1422         ASSERT(generic_ref->type != BTRFS_REF_NOT_SET &&
1423                generic_ref->action);
1424         BUG_ON(generic_ref->type == BTRFS_REF_METADATA &&
1425                generic_ref->tree_ref.owning_root == BTRFS_TREE_LOG_OBJECTID);
1426
1427         if (generic_ref->type == BTRFS_REF_METADATA)
1428                 ret = btrfs_add_delayed_tree_ref(trans, generic_ref, NULL);
1429         else
1430                 ret = btrfs_add_delayed_data_ref(trans, generic_ref, 0);
1431
1432         btrfs_ref_tree_mod(fs_info, generic_ref);
1433
1434         return ret;
1435 }
1436
1437 /*
1438  * __btrfs_inc_extent_ref - insert backreference for a given extent
1439  *
1440  * The counterpart is in __btrfs_free_extent(), with examples and more details
1441  * how it works.
1442  *
1443  * @trans:          Handle of transaction
1444  *
1445  * @node:           The delayed ref node used to get the bytenr/length for
1446  *                  extent whose references are incremented.
1447  *
1448  * @parent:         If this is a shared extent (BTRFS_SHARED_DATA_REF_KEY/
1449  *                  BTRFS_SHARED_BLOCK_REF_KEY) then it holds the logical
1450  *                  bytenr of the parent block. Since new extents are always
1451  *                  created with indirect references, this will only be the case
1452  *                  when relocating a shared extent. In that case, root_objectid
1453  *                  will be BTRFS_TREE_RELOC_OBJECTID. Otherwise, parent must
1454  *                  be 0
1455  *
1456  * @root_objectid:  The id of the root where this modification has originated,
1457  *                  this can be either one of the well-known metadata trees or
1458  *                  the subvolume id which references this extent.
1459  *
1460  * @owner:          For data extents it is the inode number of the owning file.
1461  *                  For metadata extents this parameter holds the level in the
1462  *                  tree of the extent.
1463  *
1464  * @offset:         For metadata extents the offset is ignored and is currently
1465  *                  always passed as 0. For data extents it is the fileoffset
1466  *                  this extent belongs to.
1467  *
1468  * @refs_to_add     Number of references to add
1469  *
1470  * @extent_op       Pointer to a structure, holding information necessary when
1471  *                  updating a tree block's flags
1472  *
1473  */
1474 static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
1475                                   struct btrfs_delayed_ref_node *node,
1476                                   u64 parent, u64 root_objectid,
1477                                   u64 owner, u64 offset, int refs_to_add,
1478                                   struct btrfs_delayed_extent_op *extent_op)
1479 {
1480         struct btrfs_path *path;
1481         struct extent_buffer *leaf;
1482         struct btrfs_extent_item *item;
1483         struct btrfs_key key;
1484         u64 bytenr = node->bytenr;
1485         u64 num_bytes = node->num_bytes;
1486         u64 refs;
1487         int ret;
1488
1489         path = btrfs_alloc_path();
1490         if (!path)
1491                 return -ENOMEM;
1492
1493         /* this will setup the path even if it fails to insert the back ref */
1494         ret = insert_inline_extent_backref(trans, path, bytenr, num_bytes,
1495                                            parent, root_objectid, owner,
1496                                            offset, refs_to_add, extent_op);
1497         if ((ret < 0 && ret != -EAGAIN) || !ret)
1498                 goto out;
1499
1500         /*
1501          * Ok we had -EAGAIN which means we didn't have space to insert and
1502          * inline extent ref, so just update the reference count and add a
1503          * normal backref.
1504          */
1505         leaf = path->nodes[0];
1506         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1507         item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1508         refs = btrfs_extent_refs(leaf, item);
1509         btrfs_set_extent_refs(leaf, item, refs + refs_to_add);
1510         if (extent_op)
1511                 __run_delayed_extent_op(extent_op, leaf, item);
1512
1513         btrfs_mark_buffer_dirty(leaf);
1514         btrfs_release_path(path);
1515
1516         /* now insert the actual backref */
1517         if (owner < BTRFS_FIRST_FREE_OBJECTID)
1518                 ret = insert_tree_block_ref(trans, path, bytenr, parent,
1519                                             root_objectid);
1520         else
1521                 ret = insert_extent_data_ref(trans, path, bytenr, parent,
1522                                              root_objectid, owner, offset,
1523                                              refs_to_add);
1524
1525         if (ret)
1526                 btrfs_abort_transaction(trans, ret);
1527 out:
1528         btrfs_free_path(path);
1529         return ret;
1530 }
1531
1532 static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
1533                                 struct btrfs_delayed_ref_node *node,
1534                                 struct btrfs_delayed_extent_op *extent_op,
1535                                 bool insert_reserved)
1536 {
1537         int ret = 0;
1538         struct btrfs_delayed_data_ref *ref;
1539         struct btrfs_key ins;
1540         u64 parent = 0;
1541         u64 ref_root = 0;
1542         u64 flags = 0;
1543
1544         ins.objectid = node->bytenr;
1545         ins.offset = node->num_bytes;
1546         ins.type = BTRFS_EXTENT_ITEM_KEY;
1547
1548         ref = btrfs_delayed_node_to_data_ref(node);
1549         trace_run_delayed_data_ref(trans->fs_info, node, ref, node->action);
1550
1551         if (node->type == BTRFS_SHARED_DATA_REF_KEY)
1552                 parent = ref->parent;
1553         ref_root = ref->root;
1554
1555         if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
1556                 if (extent_op)
1557                         flags |= extent_op->flags_to_set;
1558                 ret = alloc_reserved_file_extent(trans, parent, ref_root,
1559                                                  flags, ref->objectid,
1560                                                  ref->offset, &ins,
1561                                                  node->ref_mod);
1562         } else if (node->action == BTRFS_ADD_DELAYED_REF) {
1563                 ret = __btrfs_inc_extent_ref(trans, node, parent, ref_root,
1564                                              ref->objectid, ref->offset,
1565                                              node->ref_mod, extent_op);
1566         } else if (node->action == BTRFS_DROP_DELAYED_REF) {
1567                 ret = __btrfs_free_extent(trans, node, parent,
1568                                           ref_root, ref->objectid,
1569                                           ref->offset, node->ref_mod,
1570                                           extent_op);
1571         } else {
1572                 BUG();
1573         }
1574         return ret;
1575 }
1576
1577 static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
1578                                     struct extent_buffer *leaf,
1579                                     struct btrfs_extent_item *ei)
1580 {
1581         u64 flags = btrfs_extent_flags(leaf, ei);
1582         if (extent_op->update_flags) {
1583                 flags |= extent_op->flags_to_set;
1584                 btrfs_set_extent_flags(leaf, ei, flags);
1585         }
1586
1587         if (extent_op->update_key) {
1588                 struct btrfs_tree_block_info *bi;
1589                 BUG_ON(!(flags & BTRFS_EXTENT_FLAG_TREE_BLOCK));
1590                 bi = (struct btrfs_tree_block_info *)(ei + 1);
1591                 btrfs_set_tree_block_key(leaf, bi, &extent_op->key);
1592         }
1593 }
1594
1595 static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
1596                                  struct btrfs_delayed_ref_head *head,
1597                                  struct btrfs_delayed_extent_op *extent_op)
1598 {
1599         struct btrfs_fs_info *fs_info = trans->fs_info;
1600         struct btrfs_root *root;
1601         struct btrfs_key key;
1602         struct btrfs_path *path;
1603         struct btrfs_extent_item *ei;
1604         struct extent_buffer *leaf;
1605         u32 item_size;
1606         int ret;
1607         int err = 0;
1608         int metadata = 1;
1609
1610         if (TRANS_ABORTED(trans))
1611                 return 0;
1612
1613         if (!btrfs_fs_incompat(fs_info, SKINNY_METADATA))
1614                 metadata = 0;
1615
1616         path = btrfs_alloc_path();
1617         if (!path)
1618                 return -ENOMEM;
1619
1620         key.objectid = head->bytenr;
1621
1622         if (metadata) {
1623                 key.type = BTRFS_METADATA_ITEM_KEY;
1624                 key.offset = extent_op->level;
1625         } else {
1626                 key.type = BTRFS_EXTENT_ITEM_KEY;
1627                 key.offset = head->num_bytes;
1628         }
1629
1630         root = btrfs_extent_root(fs_info, key.objectid);
1631 again:
1632         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
1633         if (ret < 0) {
1634                 err = ret;
1635                 goto out;
1636         }
1637         if (ret > 0) {
1638                 if (metadata) {
1639                         if (path->slots[0] > 0) {
1640                                 path->slots[0]--;
1641                                 btrfs_item_key_to_cpu(path->nodes[0], &key,
1642                                                       path->slots[0]);
1643                                 if (key.objectid == head->bytenr &&
1644                                     key.type == BTRFS_EXTENT_ITEM_KEY &&
1645                                     key.offset == head->num_bytes)
1646                                         ret = 0;
1647                         }
1648                         if (ret > 0) {
1649                                 btrfs_release_path(path);
1650                                 metadata = 0;
1651
1652                                 key.objectid = head->bytenr;
1653                                 key.offset = head->num_bytes;
1654                                 key.type = BTRFS_EXTENT_ITEM_KEY;
1655                                 goto again;
1656                         }
1657                 } else {
1658                         err = -EUCLEAN;
1659                         btrfs_err(fs_info,
1660                   "missing extent item for extent %llu num_bytes %llu level %d",
1661                                   head->bytenr, head->num_bytes, extent_op->level);
1662                         goto out;
1663                 }
1664         }
1665
1666         leaf = path->nodes[0];
1667         item_size = btrfs_item_size(leaf, path->slots[0]);
1668
1669         if (unlikely(item_size < sizeof(*ei))) {
1670                 err = -EUCLEAN;
1671                 btrfs_err(fs_info,
1672                           "unexpected extent item size, has %u expect >= %zu",
1673                           item_size, sizeof(*ei));
1674                 btrfs_abort_transaction(trans, err);
1675                 goto out;
1676         }
1677
1678         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
1679         __run_delayed_extent_op(extent_op, leaf, ei);
1680
1681         btrfs_mark_buffer_dirty(leaf);
1682 out:
1683         btrfs_free_path(path);
1684         return err;
1685 }
1686
1687 static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
1688                                 struct btrfs_delayed_ref_node *node,
1689                                 struct btrfs_delayed_extent_op *extent_op,
1690                                 bool insert_reserved)
1691 {
1692         int ret = 0;
1693         struct btrfs_delayed_tree_ref *ref;
1694         u64 parent = 0;
1695         u64 ref_root = 0;
1696
1697         ref = btrfs_delayed_node_to_tree_ref(node);
1698         trace_run_delayed_tree_ref(trans->fs_info, node, ref, node->action);
1699
1700         if (node->type == BTRFS_SHARED_BLOCK_REF_KEY)
1701                 parent = ref->parent;
1702         ref_root = ref->root;
1703
1704         if (unlikely(node->ref_mod != 1)) {
1705                 btrfs_err(trans->fs_info,
1706         "btree block %llu has %d references rather than 1: action %d ref_root %llu parent %llu",
1707                           node->bytenr, node->ref_mod, node->action, ref_root,
1708                           parent);
1709                 return -EUCLEAN;
1710         }
1711         if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
1712                 BUG_ON(!extent_op || !extent_op->update_flags);
1713                 ret = alloc_reserved_tree_block(trans, node, extent_op);
1714         } else if (node->action == BTRFS_ADD_DELAYED_REF) {
1715                 ret = __btrfs_inc_extent_ref(trans, node, parent, ref_root,
1716                                              ref->level, 0, 1, extent_op);
1717         } else if (node->action == BTRFS_DROP_DELAYED_REF) {
1718                 ret = __btrfs_free_extent(trans, node, parent, ref_root,
1719                                           ref->level, 0, 1, extent_op);
1720         } else {
1721                 BUG();
1722         }
1723         return ret;
1724 }
1725
1726 /* helper function to actually process a single delayed ref entry */
1727 static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
1728                                struct btrfs_delayed_ref_node *node,
1729                                struct btrfs_delayed_extent_op *extent_op,
1730                                bool insert_reserved)
1731 {
1732         int ret = 0;
1733
1734         if (TRANS_ABORTED(trans)) {
1735                 if (insert_reserved)
1736                         btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
1737                 return 0;
1738         }
1739
1740         if (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
1741             node->type == BTRFS_SHARED_BLOCK_REF_KEY)
1742                 ret = run_delayed_tree_ref(trans, node, extent_op,
1743                                            insert_reserved);
1744         else if (node->type == BTRFS_EXTENT_DATA_REF_KEY ||
1745                  node->type == BTRFS_SHARED_DATA_REF_KEY)
1746                 ret = run_delayed_data_ref(trans, node, extent_op,
1747                                            insert_reserved);
1748         else
1749                 BUG();
1750         if (ret && insert_reserved)
1751                 btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
1752         if (ret < 0)
1753                 btrfs_err(trans->fs_info,
1754 "failed to run delayed ref for logical %llu num_bytes %llu type %u action %u ref_mod %d: %d",
1755                           node->bytenr, node->num_bytes, node->type,
1756                           node->action, node->ref_mod, ret);
1757         return ret;
1758 }
1759
1760 static inline struct btrfs_delayed_ref_node *
1761 select_delayed_ref(struct btrfs_delayed_ref_head *head)
1762 {
1763         struct btrfs_delayed_ref_node *ref;
1764
1765         if (RB_EMPTY_ROOT(&head->ref_tree.rb_root))
1766                 return NULL;
1767
1768         /*
1769          * Select a delayed ref of type BTRFS_ADD_DELAYED_REF first.
1770          * This is to prevent a ref count from going down to zero, which deletes
1771          * the extent item from the extent tree, when there still are references
1772          * to add, which would fail because they would not find the extent item.
1773          */
1774         if (!list_empty(&head->ref_add_list))
1775                 return list_first_entry(&head->ref_add_list,
1776                                 struct btrfs_delayed_ref_node, add_list);
1777
1778         ref = rb_entry(rb_first_cached(&head->ref_tree),
1779                        struct btrfs_delayed_ref_node, ref_node);
1780         ASSERT(list_empty(&ref->add_list));
1781         return ref;
1782 }
1783
1784 static void unselect_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs,
1785                                       struct btrfs_delayed_ref_head *head)
1786 {
1787         spin_lock(&delayed_refs->lock);
1788         head->processing = false;
1789         delayed_refs->num_heads_ready++;
1790         spin_unlock(&delayed_refs->lock);
1791         btrfs_delayed_ref_unlock(head);
1792 }
1793
1794 static struct btrfs_delayed_extent_op *cleanup_extent_op(
1795                                 struct btrfs_delayed_ref_head *head)
1796 {
1797         struct btrfs_delayed_extent_op *extent_op = head->extent_op;
1798
1799         if (!extent_op)
1800                 return NULL;
1801
1802         if (head->must_insert_reserved) {
1803                 head->extent_op = NULL;
1804                 btrfs_free_delayed_extent_op(extent_op);
1805                 return NULL;
1806         }
1807         return extent_op;
1808 }
1809
1810 static int run_and_cleanup_extent_op(struct btrfs_trans_handle *trans,
1811                                      struct btrfs_delayed_ref_head *head)
1812 {
1813         struct btrfs_delayed_extent_op *extent_op;
1814         int ret;
1815
1816         extent_op = cleanup_extent_op(head);
1817         if (!extent_op)
1818                 return 0;
1819         head->extent_op = NULL;
1820         spin_unlock(&head->lock);
1821         ret = run_delayed_extent_op(trans, head, extent_op);
1822         btrfs_free_delayed_extent_op(extent_op);
1823         return ret ? ret : 1;
1824 }
1825
1826 void btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
1827                                   struct btrfs_delayed_ref_root *delayed_refs,
1828                                   struct btrfs_delayed_ref_head *head)
1829 {
1830         int nr_items = 1;       /* Dropping this ref head update. */
1831
1832         /*
1833          * We had csum deletions accounted for in our delayed refs rsv, we need
1834          * to drop the csum leaves for this update from our delayed_refs_rsv.
1835          */
1836         if (head->total_ref_mod < 0 && head->is_data) {
1837                 spin_lock(&delayed_refs->lock);
1838                 delayed_refs->pending_csums -= head->num_bytes;
1839                 spin_unlock(&delayed_refs->lock);
1840                 nr_items += btrfs_csum_bytes_to_leaves(fs_info, head->num_bytes);
1841         }
1842
1843         btrfs_delayed_refs_rsv_release(fs_info, nr_items);
1844 }
1845
1846 static int cleanup_ref_head(struct btrfs_trans_handle *trans,
1847                             struct btrfs_delayed_ref_head *head)
1848 {
1849
1850         struct btrfs_fs_info *fs_info = trans->fs_info;
1851         struct btrfs_delayed_ref_root *delayed_refs;
1852         int ret;
1853
1854         delayed_refs = &trans->transaction->delayed_refs;
1855
1856         ret = run_and_cleanup_extent_op(trans, head);
1857         if (ret < 0) {
1858                 unselect_delayed_ref_head(delayed_refs, head);
1859                 btrfs_debug(fs_info, "run_delayed_extent_op returned %d", ret);
1860                 return ret;
1861         } else if (ret) {
1862                 return ret;
1863         }
1864
1865         /*
1866          * Need to drop our head ref lock and re-acquire the delayed ref lock
1867          * and then re-check to make sure nobody got added.
1868          */
1869         spin_unlock(&head->lock);
1870         spin_lock(&delayed_refs->lock);
1871         spin_lock(&head->lock);
1872         if (!RB_EMPTY_ROOT(&head->ref_tree.rb_root) || head->extent_op) {
1873                 spin_unlock(&head->lock);
1874                 spin_unlock(&delayed_refs->lock);
1875                 return 1;
1876         }
1877         btrfs_delete_ref_head(delayed_refs, head);
1878         spin_unlock(&head->lock);
1879         spin_unlock(&delayed_refs->lock);
1880
1881         if (head->must_insert_reserved) {
1882                 btrfs_pin_extent(trans, head->bytenr, head->num_bytes, 1);
1883                 if (head->is_data) {
1884                         struct btrfs_root *csum_root;
1885
1886                         csum_root = btrfs_csum_root(fs_info, head->bytenr);
1887                         ret = btrfs_del_csums(trans, csum_root, head->bytenr,
1888                                               head->num_bytes);
1889                 }
1890         }
1891
1892         btrfs_cleanup_ref_head_accounting(fs_info, delayed_refs, head);
1893
1894         trace_run_delayed_ref_head(fs_info, head, 0);
1895         btrfs_delayed_ref_unlock(head);
1896         btrfs_put_delayed_ref_head(head);
1897         return ret;
1898 }
1899
1900 static struct btrfs_delayed_ref_head *btrfs_obtain_ref_head(
1901                                         struct btrfs_trans_handle *trans)
1902 {
1903         struct btrfs_delayed_ref_root *delayed_refs =
1904                 &trans->transaction->delayed_refs;
1905         struct btrfs_delayed_ref_head *head = NULL;
1906         int ret;
1907
1908         spin_lock(&delayed_refs->lock);
1909         head = btrfs_select_ref_head(delayed_refs);
1910         if (!head) {
1911                 spin_unlock(&delayed_refs->lock);
1912                 return head;
1913         }
1914
1915         /*
1916          * Grab the lock that says we are going to process all the refs for
1917          * this head
1918          */
1919         ret = btrfs_delayed_ref_lock(delayed_refs, head);
1920         spin_unlock(&delayed_refs->lock);
1921
1922         /*
1923          * We may have dropped the spin lock to get the head mutex lock, and
1924          * that might have given someone else time to free the head.  If that's
1925          * true, it has been removed from our list and we can move on.
1926          */
1927         if (ret == -EAGAIN)
1928                 head = ERR_PTR(-EAGAIN);
1929
1930         return head;
1931 }
1932
1933 static int btrfs_run_delayed_refs_for_head(struct btrfs_trans_handle *trans,
1934                                            struct btrfs_delayed_ref_head *locked_ref)
1935 {
1936         struct btrfs_fs_info *fs_info = trans->fs_info;
1937         struct btrfs_delayed_ref_root *delayed_refs;
1938         struct btrfs_delayed_extent_op *extent_op;
1939         struct btrfs_delayed_ref_node *ref;
1940         bool must_insert_reserved;
1941         int ret;
1942
1943         delayed_refs = &trans->transaction->delayed_refs;
1944
1945         lockdep_assert_held(&locked_ref->mutex);
1946         lockdep_assert_held(&locked_ref->lock);
1947
1948         while ((ref = select_delayed_ref(locked_ref))) {
1949                 if (ref->seq &&
1950                     btrfs_check_delayed_seq(fs_info, ref->seq)) {
1951                         spin_unlock(&locked_ref->lock);
1952                         unselect_delayed_ref_head(delayed_refs, locked_ref);
1953                         return -EAGAIN;
1954                 }
1955
1956                 rb_erase_cached(&ref->ref_node, &locked_ref->ref_tree);
1957                 RB_CLEAR_NODE(&ref->ref_node);
1958                 if (!list_empty(&ref->add_list))
1959                         list_del(&ref->add_list);
1960                 /*
1961                  * When we play the delayed ref, also correct the ref_mod on
1962                  * head
1963                  */
1964                 switch (ref->action) {
1965                 case BTRFS_ADD_DELAYED_REF:
1966                 case BTRFS_ADD_DELAYED_EXTENT:
1967                         locked_ref->ref_mod -= ref->ref_mod;
1968                         break;
1969                 case BTRFS_DROP_DELAYED_REF:
1970                         locked_ref->ref_mod += ref->ref_mod;
1971                         break;
1972                 default:
1973                         WARN_ON(1);
1974                 }
1975                 atomic_dec(&delayed_refs->num_entries);
1976
1977                 /*
1978                  * Record the must_insert_reserved flag before we drop the
1979                  * spin lock.
1980                  */
1981                 must_insert_reserved = locked_ref->must_insert_reserved;
1982                 locked_ref->must_insert_reserved = false;
1983
1984                 extent_op = locked_ref->extent_op;
1985                 locked_ref->extent_op = NULL;
1986                 spin_unlock(&locked_ref->lock);
1987
1988                 ret = run_one_delayed_ref(trans, ref, extent_op,
1989                                           must_insert_reserved);
1990
1991                 btrfs_free_delayed_extent_op(extent_op);
1992                 if (ret) {
1993                         unselect_delayed_ref_head(delayed_refs, locked_ref);
1994                         btrfs_put_delayed_ref(ref);
1995                         return ret;
1996                 }
1997
1998                 btrfs_put_delayed_ref(ref);
1999                 cond_resched();
2000
2001                 spin_lock(&locked_ref->lock);
2002                 btrfs_merge_delayed_refs(fs_info, delayed_refs, locked_ref);
2003         }
2004
2005         return 0;
2006 }
2007
2008 /*
2009  * Returns 0 on success or if called with an already aborted transaction.
2010  * Returns -ENOMEM or -EIO on failure and will abort the transaction.
2011  */
2012 static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
2013                                              unsigned long nr)
2014 {
2015         struct btrfs_fs_info *fs_info = trans->fs_info;
2016         struct btrfs_delayed_ref_root *delayed_refs;
2017         struct btrfs_delayed_ref_head *locked_ref = NULL;
2018         int ret;
2019         unsigned long count = 0;
2020
2021         delayed_refs = &trans->transaction->delayed_refs;
2022         do {
2023                 if (!locked_ref) {
2024                         locked_ref = btrfs_obtain_ref_head(trans);
2025                         if (IS_ERR_OR_NULL(locked_ref)) {
2026                                 if (PTR_ERR(locked_ref) == -EAGAIN) {
2027                                         continue;
2028                                 } else {
2029                                         break;
2030                                 }
2031                         }
2032                         count++;
2033                 }
2034                 /*
2035                  * We need to try and merge add/drops of the same ref since we
2036                  * can run into issues with relocate dropping the implicit ref
2037                  * and then it being added back again before the drop can
2038                  * finish.  If we merged anything we need to re-loop so we can
2039                  * get a good ref.
2040                  * Or we can get node references of the same type that weren't
2041                  * merged when created due to bumps in the tree mod seq, and
2042                  * we need to merge them to prevent adding an inline extent
2043                  * backref before dropping it (triggering a BUG_ON at
2044                  * insert_inline_extent_backref()).
2045                  */
2046                 spin_lock(&locked_ref->lock);
2047                 btrfs_merge_delayed_refs(fs_info, delayed_refs, locked_ref);
2048
2049                 ret = btrfs_run_delayed_refs_for_head(trans, locked_ref);
2050                 if (ret < 0 && ret != -EAGAIN) {
2051                         /*
2052                          * Error, btrfs_run_delayed_refs_for_head already
2053                          * unlocked everything so just bail out
2054                          */
2055                         return ret;
2056                 } else if (!ret) {
2057                         /*
2058                          * Success, perform the usual cleanup of a processed
2059                          * head
2060                          */
2061                         ret = cleanup_ref_head(trans, locked_ref);
2062                         if (ret > 0 ) {
2063                                 /* We dropped our lock, we need to loop. */
2064                                 ret = 0;
2065                                 continue;
2066                         } else if (ret) {
2067                                 return ret;
2068                         }
2069                 }
2070
2071                 /*
2072                  * Either success case or btrfs_run_delayed_refs_for_head
2073                  * returned -EAGAIN, meaning we need to select another head
2074                  */
2075
2076                 locked_ref = NULL;
2077                 cond_resched();
2078         } while ((nr != -1 && count < nr) || locked_ref);
2079
2080         return 0;
2081 }
2082
2083 #ifdef SCRAMBLE_DELAYED_REFS
2084 /*
2085  * Normally delayed refs get processed in ascending bytenr order. This
2086  * correlates in most cases to the order added. To expose dependencies on this
2087  * order, we start to process the tree in the middle instead of the beginning
2088  */
2089 static u64 find_middle(struct rb_root *root)
2090 {
2091         struct rb_node *n = root->rb_node;
2092         struct btrfs_delayed_ref_node *entry;
2093         int alt = 1;
2094         u64 middle;
2095         u64 first = 0, last = 0;
2096
2097         n = rb_first(root);
2098         if (n) {
2099                 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2100                 first = entry->bytenr;
2101         }
2102         n = rb_last(root);
2103         if (n) {
2104                 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2105                 last = entry->bytenr;
2106         }
2107         n = root->rb_node;
2108
2109         while (n) {
2110                 entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
2111                 WARN_ON(!entry->in_tree);
2112
2113                 middle = entry->bytenr;
2114
2115                 if (alt)
2116                         n = n->rb_left;
2117                 else
2118                         n = n->rb_right;
2119
2120                 alt = 1 - alt;
2121         }
2122         return middle;
2123 }
2124 #endif
2125
2126 /*
2127  * this starts processing the delayed reference count updates and
2128  * extent insertions we have queued up so far.  count can be
2129  * 0, which means to process everything in the tree at the start
2130  * of the run (but not newly added entries), or it can be some target
2131  * number you'd like to process.
2132  *
2133  * Returns 0 on success or if called with an aborted transaction
2134  * Returns <0 on error and aborts the transaction
2135  */
2136 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
2137                            unsigned long count)
2138 {
2139         struct btrfs_fs_info *fs_info = trans->fs_info;
2140         struct rb_node *node;
2141         struct btrfs_delayed_ref_root *delayed_refs;
2142         struct btrfs_delayed_ref_head *head;
2143         int ret;
2144         int run_all = count == (unsigned long)-1;
2145
2146         /* We'll clean this up in btrfs_cleanup_transaction */
2147         if (TRANS_ABORTED(trans))
2148                 return 0;
2149
2150         if (test_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags))
2151                 return 0;
2152
2153         delayed_refs = &trans->transaction->delayed_refs;
2154         if (count == 0)
2155                 count = delayed_refs->num_heads_ready;
2156
2157 again:
2158 #ifdef SCRAMBLE_DELAYED_REFS
2159         delayed_refs->run_delayed_start = find_middle(&delayed_refs->root);
2160 #endif
2161         ret = __btrfs_run_delayed_refs(trans, count);
2162         if (ret < 0) {
2163                 btrfs_abort_transaction(trans, ret);
2164                 return ret;
2165         }
2166
2167         if (run_all) {
2168                 btrfs_create_pending_block_groups(trans);
2169
2170                 spin_lock(&delayed_refs->lock);
2171                 node = rb_first_cached(&delayed_refs->href_root);
2172                 if (!node) {
2173                         spin_unlock(&delayed_refs->lock);
2174                         goto out;
2175                 }
2176                 head = rb_entry(node, struct btrfs_delayed_ref_head,
2177                                 href_node);
2178                 refcount_inc(&head->refs);
2179                 spin_unlock(&delayed_refs->lock);
2180
2181                 /* Mutex was contended, block until it's released and retry. */
2182                 mutex_lock(&head->mutex);
2183                 mutex_unlock(&head->mutex);
2184
2185                 btrfs_put_delayed_ref_head(head);
2186                 cond_resched();
2187                 goto again;
2188         }
2189 out:
2190         return 0;
2191 }
2192
2193 int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
2194                                 struct extent_buffer *eb, u64 flags)
2195 {
2196         struct btrfs_delayed_extent_op *extent_op;
2197         int level = btrfs_header_level(eb);
2198         int ret;
2199
2200         extent_op = btrfs_alloc_delayed_extent_op();
2201         if (!extent_op)
2202                 return -ENOMEM;
2203
2204         extent_op->flags_to_set = flags;
2205         extent_op->update_flags = true;
2206         extent_op->update_key = false;
2207         extent_op->level = level;
2208
2209         ret = btrfs_add_delayed_extent_op(trans, eb->start, eb->len, extent_op);
2210         if (ret)
2211                 btrfs_free_delayed_extent_op(extent_op);
2212         return ret;
2213 }
2214
2215 static noinline int check_delayed_ref(struct btrfs_root *root,
2216                                       struct btrfs_path *path,
2217                                       u64 objectid, u64 offset, u64 bytenr)
2218 {
2219         struct btrfs_delayed_ref_head *head;
2220         struct btrfs_delayed_ref_node *ref;
2221         struct btrfs_delayed_data_ref *data_ref;
2222         struct btrfs_delayed_ref_root *delayed_refs;
2223         struct btrfs_transaction *cur_trans;
2224         struct rb_node *node;
2225         int ret = 0;
2226
2227         spin_lock(&root->fs_info->trans_lock);
2228         cur_trans = root->fs_info->running_transaction;
2229         if (cur_trans)
2230                 refcount_inc(&cur_trans->use_count);
2231         spin_unlock(&root->fs_info->trans_lock);
2232         if (!cur_trans)
2233                 return 0;
2234
2235         delayed_refs = &cur_trans->delayed_refs;
2236         spin_lock(&delayed_refs->lock);
2237         head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
2238         if (!head) {
2239                 spin_unlock(&delayed_refs->lock);
2240                 btrfs_put_transaction(cur_trans);
2241                 return 0;
2242         }
2243
2244         if (!mutex_trylock(&head->mutex)) {
2245                 if (path->nowait) {
2246                         spin_unlock(&delayed_refs->lock);
2247                         btrfs_put_transaction(cur_trans);
2248                         return -EAGAIN;
2249                 }
2250
2251                 refcount_inc(&head->refs);
2252                 spin_unlock(&delayed_refs->lock);
2253
2254                 btrfs_release_path(path);
2255
2256                 /*
2257                  * Mutex was contended, block until it's released and let
2258                  * caller try again
2259                  */
2260                 mutex_lock(&head->mutex);
2261                 mutex_unlock(&head->mutex);
2262                 btrfs_put_delayed_ref_head(head);
2263                 btrfs_put_transaction(cur_trans);
2264                 return -EAGAIN;
2265         }
2266         spin_unlock(&delayed_refs->lock);
2267
2268         spin_lock(&head->lock);
2269         /*
2270          * XXX: We should replace this with a proper search function in the
2271          * future.
2272          */
2273         for (node = rb_first_cached(&head->ref_tree); node;
2274              node = rb_next(node)) {
2275                 ref = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
2276                 /* If it's a shared ref we know a cross reference exists */
2277                 if (ref->type != BTRFS_EXTENT_DATA_REF_KEY) {
2278                         ret = 1;
2279                         break;
2280                 }
2281
2282                 data_ref = btrfs_delayed_node_to_data_ref(ref);
2283
2284                 /*
2285                  * If our ref doesn't match the one we're currently looking at
2286                  * then we have a cross reference.
2287                  */
2288                 if (data_ref->root != root->root_key.objectid ||
2289                     data_ref->objectid != objectid ||
2290                     data_ref->offset != offset) {
2291                         ret = 1;
2292                         break;
2293                 }
2294         }
2295         spin_unlock(&head->lock);
2296         mutex_unlock(&head->mutex);
2297         btrfs_put_transaction(cur_trans);
2298         return ret;
2299 }
2300
2301 static noinline int check_committed_ref(struct btrfs_root *root,
2302                                         struct btrfs_path *path,
2303                                         u64 objectid, u64 offset, u64 bytenr,
2304                                         bool strict)
2305 {
2306         struct btrfs_fs_info *fs_info = root->fs_info;
2307         struct btrfs_root *extent_root = btrfs_extent_root(fs_info, bytenr);
2308         struct extent_buffer *leaf;
2309         struct btrfs_extent_data_ref *ref;
2310         struct btrfs_extent_inline_ref *iref;
2311         struct btrfs_extent_item *ei;
2312         struct btrfs_key key;
2313         u32 item_size;
2314         int type;
2315         int ret;
2316
2317         key.objectid = bytenr;
2318         key.offset = (u64)-1;
2319         key.type = BTRFS_EXTENT_ITEM_KEY;
2320
2321         ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
2322         if (ret < 0)
2323                 goto out;
2324         BUG_ON(ret == 0); /* Corruption */
2325
2326         ret = -ENOENT;
2327         if (path->slots[0] == 0)
2328                 goto out;
2329
2330         path->slots[0]--;
2331         leaf = path->nodes[0];
2332         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
2333
2334         if (key.objectid != bytenr || key.type != BTRFS_EXTENT_ITEM_KEY)
2335                 goto out;
2336
2337         ret = 1;
2338         item_size = btrfs_item_size(leaf, path->slots[0]);
2339         ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
2340
2341         /* If extent item has more than 1 inline ref then it's shared */
2342         if (item_size != sizeof(*ei) +
2343             btrfs_extent_inline_ref_size(BTRFS_EXTENT_DATA_REF_KEY))
2344                 goto out;
2345
2346         /*
2347          * If extent created before last snapshot => it's shared unless the
2348          * snapshot has been deleted. Use the heuristic if strict is false.
2349          */
2350         if (!strict &&
2351             (btrfs_extent_generation(leaf, ei) <=
2352              btrfs_root_last_snapshot(&root->root_item)))
2353                 goto out;
2354
2355         iref = (struct btrfs_extent_inline_ref *)(ei + 1);
2356
2357         /* If this extent has SHARED_DATA_REF then it's shared */
2358         type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
2359         if (type != BTRFS_EXTENT_DATA_REF_KEY)
2360                 goto out;
2361
2362         ref = (struct btrfs_extent_data_ref *)(&iref->offset);
2363         if (btrfs_extent_refs(leaf, ei) !=
2364             btrfs_extent_data_ref_count(leaf, ref) ||
2365             btrfs_extent_data_ref_root(leaf, ref) !=
2366             root->root_key.objectid ||
2367             btrfs_extent_data_ref_objectid(leaf, ref) != objectid ||
2368             btrfs_extent_data_ref_offset(leaf, ref) != offset)
2369                 goto out;
2370
2371         ret = 0;
2372 out:
2373         return ret;
2374 }
2375
2376 int btrfs_cross_ref_exist(struct btrfs_root *root, u64 objectid, u64 offset,
2377                           u64 bytenr, bool strict, struct btrfs_path *path)
2378 {
2379         int ret;
2380
2381         do {
2382                 ret = check_committed_ref(root, path, objectid,
2383                                           offset, bytenr, strict);
2384                 if (ret && ret != -ENOENT)
2385                         goto out;
2386
2387                 ret = check_delayed_ref(root, path, objectid, offset, bytenr);
2388         } while (ret == -EAGAIN);
2389
2390 out:
2391         btrfs_release_path(path);
2392         if (btrfs_is_data_reloc_root(root))
2393                 WARN_ON(ret > 0);
2394         return ret;
2395 }
2396
2397 static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
2398                            struct btrfs_root *root,
2399                            struct extent_buffer *buf,
2400                            int full_backref, int inc)
2401 {
2402         struct btrfs_fs_info *fs_info = root->fs_info;
2403         u64 bytenr;
2404         u64 num_bytes;
2405         u64 parent;
2406         u64 ref_root;
2407         u32 nritems;
2408         struct btrfs_key key;
2409         struct btrfs_file_extent_item *fi;
2410         struct btrfs_ref generic_ref = { 0 };
2411         bool for_reloc = btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC);
2412         int i;
2413         int action;
2414         int level;
2415         int ret = 0;
2416
2417         if (btrfs_is_testing(fs_info))
2418                 return 0;
2419
2420         ref_root = btrfs_header_owner(buf);
2421         nritems = btrfs_header_nritems(buf);
2422         level = btrfs_header_level(buf);
2423
2424         if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state) && level == 0)
2425                 return 0;
2426
2427         if (full_backref)
2428                 parent = buf->start;
2429         else
2430                 parent = 0;
2431         if (inc)
2432                 action = BTRFS_ADD_DELAYED_REF;
2433         else
2434                 action = BTRFS_DROP_DELAYED_REF;
2435
2436         for (i = 0; i < nritems; i++) {
2437                 if (level == 0) {
2438                         btrfs_item_key_to_cpu(buf, &key, i);
2439                         if (key.type != BTRFS_EXTENT_DATA_KEY)
2440                                 continue;
2441                         fi = btrfs_item_ptr(buf, i,
2442                                             struct btrfs_file_extent_item);
2443                         if (btrfs_file_extent_type(buf, fi) ==
2444                             BTRFS_FILE_EXTENT_INLINE)
2445                                 continue;
2446                         bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
2447                         if (bytenr == 0)
2448                                 continue;
2449
2450                         num_bytes = btrfs_file_extent_disk_num_bytes(buf, fi);
2451                         key.offset -= btrfs_file_extent_offset(buf, fi);
2452                         btrfs_init_generic_ref(&generic_ref, action, bytenr,
2453                                                num_bytes, parent);
2454                         btrfs_init_data_ref(&generic_ref, ref_root, key.objectid,
2455                                             key.offset, root->root_key.objectid,
2456                                             for_reloc);
2457                         if (inc)
2458                                 ret = btrfs_inc_extent_ref(trans, &generic_ref);
2459                         else
2460                                 ret = btrfs_free_extent(trans, &generic_ref);
2461                         if (ret)
2462                                 goto fail;
2463                 } else {
2464                         bytenr = btrfs_node_blockptr(buf, i);
2465                         num_bytes = fs_info->nodesize;
2466                         btrfs_init_generic_ref(&generic_ref, action, bytenr,
2467                                                num_bytes, parent);
2468                         btrfs_init_tree_ref(&generic_ref, level - 1, ref_root,
2469                                             root->root_key.objectid, for_reloc);
2470                         if (inc)
2471                                 ret = btrfs_inc_extent_ref(trans, &generic_ref);
2472                         else
2473                                 ret = btrfs_free_extent(trans, &generic_ref);
2474                         if (ret)
2475                                 goto fail;
2476                 }
2477         }
2478         return 0;
2479 fail:
2480         return ret;
2481 }
2482
2483 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2484                   struct extent_buffer *buf, int full_backref)
2485 {
2486         return __btrfs_mod_ref(trans, root, buf, full_backref, 1);
2487 }
2488
2489 int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
2490                   struct extent_buffer *buf, int full_backref)
2491 {
2492         return __btrfs_mod_ref(trans, root, buf, full_backref, 0);
2493 }
2494
2495 static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
2496 {
2497         struct btrfs_fs_info *fs_info = root->fs_info;
2498         u64 flags;
2499         u64 ret;
2500
2501         if (data)
2502                 flags = BTRFS_BLOCK_GROUP_DATA;
2503         else if (root == fs_info->chunk_root)
2504                 flags = BTRFS_BLOCK_GROUP_SYSTEM;
2505         else
2506                 flags = BTRFS_BLOCK_GROUP_METADATA;
2507
2508         ret = btrfs_get_alloc_profile(fs_info, flags);
2509         return ret;
2510 }
2511
2512 static u64 first_logical_byte(struct btrfs_fs_info *fs_info)
2513 {
2514         struct rb_node *leftmost;
2515         u64 bytenr = 0;
2516
2517         read_lock(&fs_info->block_group_cache_lock);
2518         /* Get the block group with the lowest logical start address. */
2519         leftmost = rb_first_cached(&fs_info->block_group_cache_tree);
2520         if (leftmost) {
2521                 struct btrfs_block_group *bg;
2522
2523                 bg = rb_entry(leftmost, struct btrfs_block_group, cache_node);
2524                 bytenr = bg->start;
2525         }
2526         read_unlock(&fs_info->block_group_cache_lock);
2527
2528         return bytenr;
2529 }
2530
2531 static int pin_down_extent(struct btrfs_trans_handle *trans,
2532                            struct btrfs_block_group *cache,
2533                            u64 bytenr, u64 num_bytes, int reserved)
2534 {
2535         struct btrfs_fs_info *fs_info = cache->fs_info;
2536
2537         spin_lock(&cache->space_info->lock);
2538         spin_lock(&cache->lock);
2539         cache->pinned += num_bytes;
2540         btrfs_space_info_update_bytes_pinned(fs_info, cache->space_info,
2541                                              num_bytes);
2542         if (reserved) {
2543                 cache->reserved -= num_bytes;
2544                 cache->space_info->bytes_reserved -= num_bytes;
2545         }
2546         spin_unlock(&cache->lock);
2547         spin_unlock(&cache->space_info->lock);
2548
2549         set_extent_bit(&trans->transaction->pinned_extents, bytenr,
2550                        bytenr + num_bytes - 1, EXTENT_DIRTY, NULL);
2551         return 0;
2552 }
2553
2554 int btrfs_pin_extent(struct btrfs_trans_handle *trans,
2555                      u64 bytenr, u64 num_bytes, int reserved)
2556 {
2557         struct btrfs_block_group *cache;
2558
2559         cache = btrfs_lookup_block_group(trans->fs_info, bytenr);
2560         BUG_ON(!cache); /* Logic error */
2561
2562         pin_down_extent(trans, cache, bytenr, num_bytes, reserved);
2563
2564         btrfs_put_block_group(cache);
2565         return 0;
2566 }
2567
2568 /*
2569  * this function must be called within transaction
2570  */
2571 int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans,
2572                                     u64 bytenr, u64 num_bytes)
2573 {
2574         struct btrfs_block_group *cache;
2575         int ret;
2576
2577         cache = btrfs_lookup_block_group(trans->fs_info, bytenr);
2578         if (!cache)
2579                 return -EINVAL;
2580
2581         /*
2582          * Fully cache the free space first so that our pin removes the free space
2583          * from the cache.
2584          */
2585         ret = btrfs_cache_block_group(cache, true);
2586         if (ret)
2587                 goto out;
2588
2589         pin_down_extent(trans, cache, bytenr, num_bytes, 0);
2590
2591         /* remove us from the free space cache (if we're there at all) */
2592         ret = btrfs_remove_free_space(cache, bytenr, num_bytes);
2593 out:
2594         btrfs_put_block_group(cache);
2595         return ret;
2596 }
2597
2598 static int __exclude_logged_extent(struct btrfs_fs_info *fs_info,
2599                                    u64 start, u64 num_bytes)
2600 {
2601         int ret;
2602         struct btrfs_block_group *block_group;
2603
2604         block_group = btrfs_lookup_block_group(fs_info, start);
2605         if (!block_group)
2606                 return -EINVAL;
2607
2608         ret = btrfs_cache_block_group(block_group, true);
2609         if (ret)
2610                 goto out;
2611
2612         ret = btrfs_remove_free_space(block_group, start, num_bytes);
2613 out:
2614         btrfs_put_block_group(block_group);
2615         return ret;
2616 }
2617
2618 int btrfs_exclude_logged_extents(struct extent_buffer *eb)
2619 {
2620         struct btrfs_fs_info *fs_info = eb->fs_info;
2621         struct btrfs_file_extent_item *item;
2622         struct btrfs_key key;
2623         int found_type;
2624         int i;
2625         int ret = 0;
2626
2627         if (!btrfs_fs_incompat(fs_info, MIXED_GROUPS))
2628                 return 0;
2629
2630         for (i = 0; i < btrfs_header_nritems(eb); i++) {
2631                 btrfs_item_key_to_cpu(eb, &key, i);
2632                 if (key.type != BTRFS_EXTENT_DATA_KEY)
2633                         continue;
2634                 item = btrfs_item_ptr(eb, i, struct btrfs_file_extent_item);
2635                 found_type = btrfs_file_extent_type(eb, item);
2636                 if (found_type == BTRFS_FILE_EXTENT_INLINE)
2637                         continue;
2638                 if (btrfs_file_extent_disk_bytenr(eb, item) == 0)
2639                         continue;
2640                 key.objectid = btrfs_file_extent_disk_bytenr(eb, item);
2641                 key.offset = btrfs_file_extent_disk_num_bytes(eb, item);
2642                 ret = __exclude_logged_extent(fs_info, key.objectid, key.offset);
2643                 if (ret)
2644                         break;
2645         }
2646
2647         return ret;
2648 }
2649
2650 static void
2651 btrfs_inc_block_group_reservations(struct btrfs_block_group *bg)
2652 {
2653         atomic_inc(&bg->reservations);
2654 }
2655
2656 /*
2657  * Returns the free cluster for the given space info and sets empty_cluster to
2658  * what it should be based on the mount options.
2659  */
2660 static struct btrfs_free_cluster *
2661 fetch_cluster_info(struct btrfs_fs_info *fs_info,
2662                    struct btrfs_space_info *space_info, u64 *empty_cluster)
2663 {
2664         struct btrfs_free_cluster *ret = NULL;
2665
2666         *empty_cluster = 0;
2667         if (btrfs_mixed_space_info(space_info))
2668                 return ret;
2669
2670         if (space_info->flags & BTRFS_BLOCK_GROUP_METADATA) {
2671                 ret = &fs_info->meta_alloc_cluster;
2672                 if (btrfs_test_opt(fs_info, SSD))
2673                         *empty_cluster = SZ_2M;
2674                 else
2675                         *empty_cluster = SZ_64K;
2676         } else if ((space_info->flags & BTRFS_BLOCK_GROUP_DATA) &&
2677                    btrfs_test_opt(fs_info, SSD_SPREAD)) {
2678                 *empty_cluster = SZ_2M;
2679                 ret = &fs_info->data_alloc_cluster;
2680         }
2681
2682         return ret;
2683 }
2684
2685 static int unpin_extent_range(struct btrfs_fs_info *fs_info,
2686                               u64 start, u64 end,
2687                               const bool return_free_space)
2688 {
2689         struct btrfs_block_group *cache = NULL;
2690         struct btrfs_space_info *space_info;
2691         struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
2692         struct btrfs_free_cluster *cluster = NULL;
2693         u64 len;
2694         u64 total_unpinned = 0;
2695         u64 empty_cluster = 0;
2696         bool readonly;
2697
2698         while (start <= end) {
2699                 readonly = false;
2700                 if (!cache ||
2701                     start >= cache->start + cache->length) {
2702                         if (cache)
2703                                 btrfs_put_block_group(cache);
2704                         total_unpinned = 0;
2705                         cache = btrfs_lookup_block_group(fs_info, start);
2706                         BUG_ON(!cache); /* Logic error */
2707
2708                         cluster = fetch_cluster_info(fs_info,
2709                                                      cache->space_info,
2710                                                      &empty_cluster);
2711                         empty_cluster <<= 1;
2712                 }
2713
2714                 len = cache->start + cache->length - start;
2715                 len = min(len, end + 1 - start);
2716
2717                 if (return_free_space)
2718                         btrfs_add_free_space(cache, start, len);
2719
2720                 start += len;
2721                 total_unpinned += len;
2722                 space_info = cache->space_info;
2723
2724                 /*
2725                  * If this space cluster has been marked as fragmented and we've
2726                  * unpinned enough in this block group to potentially allow a
2727                  * cluster to be created inside of it go ahead and clear the
2728                  * fragmented check.
2729                  */
2730                 if (cluster && cluster->fragmented &&
2731                     total_unpinned > empty_cluster) {
2732                         spin_lock(&cluster->lock);
2733                         cluster->fragmented = 0;
2734                         spin_unlock(&cluster->lock);
2735                 }
2736
2737                 spin_lock(&space_info->lock);
2738                 spin_lock(&cache->lock);
2739                 cache->pinned -= len;
2740                 btrfs_space_info_update_bytes_pinned(fs_info, space_info, -len);
2741                 space_info->max_extent_size = 0;
2742                 if (cache->ro) {
2743                         space_info->bytes_readonly += len;
2744                         readonly = true;
2745                 } else if (btrfs_is_zoned(fs_info)) {
2746                         /* Need reset before reusing in a zoned block group */
2747                         space_info->bytes_zone_unusable += len;
2748                         readonly = true;
2749                 }
2750                 spin_unlock(&cache->lock);
2751                 if (!readonly && return_free_space &&
2752                     global_rsv->space_info == space_info) {
2753                         spin_lock(&global_rsv->lock);
2754                         if (!global_rsv->full) {
2755                                 u64 to_add = min(len, global_rsv->size -
2756                                                       global_rsv->reserved);
2757
2758                                 global_rsv->reserved += to_add;
2759                                 btrfs_space_info_update_bytes_may_use(fs_info,
2760                                                 space_info, to_add);
2761                                 if (global_rsv->reserved >= global_rsv->size)
2762                                         global_rsv->full = 1;
2763                                 len -= to_add;
2764                         }
2765                         spin_unlock(&global_rsv->lock);
2766                 }
2767                 /* Add to any tickets we may have */
2768                 if (!readonly && return_free_space && len)
2769                         btrfs_try_granting_tickets(fs_info, space_info);
2770                 spin_unlock(&space_info->lock);
2771         }
2772
2773         if (cache)
2774                 btrfs_put_block_group(cache);
2775         return 0;
2776 }
2777
2778 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
2779 {
2780         struct btrfs_fs_info *fs_info = trans->fs_info;
2781         struct btrfs_block_group *block_group, *tmp;
2782         struct list_head *deleted_bgs;
2783         struct extent_io_tree *unpin;
2784         u64 start;
2785         u64 end;
2786         int ret;
2787
2788         unpin = &trans->transaction->pinned_extents;
2789
2790         while (!TRANS_ABORTED(trans)) {
2791                 struct extent_state *cached_state = NULL;
2792
2793                 mutex_lock(&fs_info->unused_bg_unpin_mutex);
2794                 if (!find_first_extent_bit(unpin, 0, &start, &end,
2795                                            EXTENT_DIRTY, &cached_state)) {
2796                         mutex_unlock(&fs_info->unused_bg_unpin_mutex);
2797                         break;
2798                 }
2799
2800                 if (btrfs_test_opt(fs_info, DISCARD_SYNC))
2801                         ret = btrfs_discard_extent(fs_info, start,
2802                                                    end + 1 - start, NULL);
2803
2804                 clear_extent_dirty(unpin, start, end, &cached_state);
2805                 unpin_extent_range(fs_info, start, end, true);
2806                 mutex_unlock(&fs_info->unused_bg_unpin_mutex);
2807                 free_extent_state(cached_state);
2808                 cond_resched();
2809         }
2810
2811         if (btrfs_test_opt(fs_info, DISCARD_ASYNC)) {
2812                 btrfs_discard_calc_delay(&fs_info->discard_ctl);
2813                 btrfs_discard_schedule_work(&fs_info->discard_ctl, true);
2814         }
2815
2816         /*
2817          * Transaction is finished.  We don't need the lock anymore.  We
2818          * do need to clean up the block groups in case of a transaction
2819          * abort.
2820          */
2821         deleted_bgs = &trans->transaction->deleted_bgs;
2822         list_for_each_entry_safe(block_group, tmp, deleted_bgs, bg_list) {
2823                 u64 trimmed = 0;
2824
2825                 ret = -EROFS;
2826                 if (!TRANS_ABORTED(trans))
2827                         ret = btrfs_discard_extent(fs_info,
2828                                                    block_group->start,
2829                                                    block_group->length,
2830                                                    &trimmed);
2831
2832                 list_del_init(&block_group->bg_list);
2833                 btrfs_unfreeze_block_group(block_group);
2834                 btrfs_put_block_group(block_group);
2835
2836                 if (ret) {
2837                         const char *errstr = btrfs_decode_error(ret);
2838                         btrfs_warn(fs_info,
2839                            "discard failed while removing blockgroup: errno=%d %s",
2840                                    ret, errstr);
2841                 }
2842         }
2843
2844         return 0;
2845 }
2846
2847 static int do_free_extent_accounting(struct btrfs_trans_handle *trans,
2848                                      u64 bytenr, u64 num_bytes, bool is_data)
2849 {
2850         int ret;
2851
2852         if (is_data) {
2853                 struct btrfs_root *csum_root;
2854
2855                 csum_root = btrfs_csum_root(trans->fs_info, bytenr);
2856                 ret = btrfs_del_csums(trans, csum_root, bytenr, num_bytes);
2857                 if (ret) {
2858                         btrfs_abort_transaction(trans, ret);
2859                         return ret;
2860                 }
2861         }
2862
2863         ret = add_to_free_space_tree(trans, bytenr, num_bytes);
2864         if (ret) {
2865                 btrfs_abort_transaction(trans, ret);
2866                 return ret;
2867         }
2868
2869         ret = btrfs_update_block_group(trans, bytenr, num_bytes, false);
2870         if (ret)
2871                 btrfs_abort_transaction(trans, ret);
2872
2873         return ret;
2874 }
2875
2876 #define abort_and_dump(trans, path, fmt, args...)       \
2877 ({                                                      \
2878         btrfs_abort_transaction(trans, -EUCLEAN);       \
2879         btrfs_print_leaf(path->nodes[0]);               \
2880         btrfs_crit(trans->fs_info, fmt, ##args);        \
2881 })
2882
2883 /*
2884  * Drop one or more refs of @node.
2885  *
2886  * 1. Locate the extent refs.
2887  *    It's either inline in EXTENT/METADATA_ITEM or in keyed SHARED_* item.
2888  *    Locate it, then reduce the refs number or remove the ref line completely.
2889  *
2890  * 2. Update the refs count in EXTENT/METADATA_ITEM
2891  *
2892  * Inline backref case:
2893  *
2894  * in extent tree we have:
2895  *
2896  *      item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
2897  *              refs 2 gen 6 flags DATA
2898  *              extent data backref root FS_TREE objectid 258 offset 0 count 1
2899  *              extent data backref root FS_TREE objectid 257 offset 0 count 1
2900  *
2901  * This function gets called with:
2902  *
2903  *    node->bytenr = 13631488
2904  *    node->num_bytes = 1048576
2905  *    root_objectid = FS_TREE
2906  *    owner_objectid = 257
2907  *    owner_offset = 0
2908  *    refs_to_drop = 1
2909  *
2910  * Then we should get some like:
2911  *
2912  *      item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
2913  *              refs 1 gen 6 flags DATA
2914  *              extent data backref root FS_TREE objectid 258 offset 0 count 1
2915  *
2916  * Keyed backref case:
2917  *
2918  * in extent tree we have:
2919  *
2920  *      item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
2921  *              refs 754 gen 6 flags DATA
2922  *      [...]
2923  *      item 2 key (13631488 EXTENT_DATA_REF <HASH>) itemoff 3915 itemsize 28
2924  *              extent data backref root FS_TREE objectid 866 offset 0 count 1
2925  *
2926  * This function get called with:
2927  *
2928  *    node->bytenr = 13631488
2929  *    node->num_bytes = 1048576
2930  *    root_objectid = FS_TREE
2931  *    owner_objectid = 866
2932  *    owner_offset = 0
2933  *    refs_to_drop = 1
2934  *
2935  * Then we should get some like:
2936  *
2937  *      item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
2938  *              refs 753 gen 6 flags DATA
2939  *
2940  * And that (13631488 EXTENT_DATA_REF <HASH>) gets removed.
2941  */
2942 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
2943                                struct btrfs_delayed_ref_node *node, u64 parent,
2944                                u64 root_objectid, u64 owner_objectid,
2945                                u64 owner_offset, int refs_to_drop,
2946                                struct btrfs_delayed_extent_op *extent_op)
2947 {
2948         struct btrfs_fs_info *info = trans->fs_info;
2949         struct btrfs_key key;
2950         struct btrfs_path *path;
2951         struct btrfs_root *extent_root;
2952         struct extent_buffer *leaf;
2953         struct btrfs_extent_item *ei;
2954         struct btrfs_extent_inline_ref *iref;
2955         int ret;
2956         int is_data;
2957         int extent_slot = 0;
2958         int found_extent = 0;
2959         int num_to_del = 1;
2960         u32 item_size;
2961         u64 refs;
2962         u64 bytenr = node->bytenr;
2963         u64 num_bytes = node->num_bytes;
2964         bool skinny_metadata = btrfs_fs_incompat(info, SKINNY_METADATA);
2965
2966         extent_root = btrfs_extent_root(info, bytenr);
2967         ASSERT(extent_root);
2968
2969         path = btrfs_alloc_path();
2970         if (!path)
2971                 return -ENOMEM;
2972
2973         is_data = owner_objectid >= BTRFS_FIRST_FREE_OBJECTID;
2974
2975         if (!is_data && refs_to_drop != 1) {
2976                 btrfs_crit(info,
2977 "invalid refs_to_drop, dropping more than 1 refs for tree block %llu refs_to_drop %u",
2978                            node->bytenr, refs_to_drop);
2979                 ret = -EINVAL;
2980                 btrfs_abort_transaction(trans, ret);
2981                 goto out;
2982         }
2983
2984         if (is_data)
2985                 skinny_metadata = false;
2986
2987         ret = lookup_extent_backref(trans, path, &iref, bytenr, num_bytes,
2988                                     parent, root_objectid, owner_objectid,
2989                                     owner_offset);
2990         if (ret == 0) {
2991                 /*
2992                  * Either the inline backref or the SHARED_DATA_REF/
2993                  * SHARED_BLOCK_REF is found
2994                  *
2995                  * Here is a quick path to locate EXTENT/METADATA_ITEM.
2996                  * It's possible the EXTENT/METADATA_ITEM is near current slot.
2997                  */
2998                 extent_slot = path->slots[0];
2999                 while (extent_slot >= 0) {
3000                         btrfs_item_key_to_cpu(path->nodes[0], &key,
3001                                               extent_slot);
3002                         if (key.objectid != bytenr)
3003                                 break;
3004                         if (key.type == BTRFS_EXTENT_ITEM_KEY &&
3005                             key.offset == num_bytes) {
3006                                 found_extent = 1;
3007                                 break;
3008                         }
3009                         if (key.type == BTRFS_METADATA_ITEM_KEY &&
3010                             key.offset == owner_objectid) {
3011                                 found_extent = 1;
3012                                 break;
3013                         }
3014
3015                         /* Quick path didn't find the EXTEMT/METADATA_ITEM */
3016                         if (path->slots[0] - extent_slot > 5)
3017                                 break;
3018                         extent_slot--;
3019                 }
3020
3021                 if (!found_extent) {
3022                         if (iref) {
3023                                 abort_and_dump(trans, path,
3024 "invalid iref slot %u, no EXTENT/METADATA_ITEM found but has inline extent ref",
3025                                            path->slots[0]);
3026                                 ret = -EUCLEAN;
3027                                 goto out;
3028                         }
3029                         /* Must be SHARED_* item, remove the backref first */
3030                         ret = remove_extent_backref(trans, extent_root, path,
3031                                                     NULL, refs_to_drop, is_data);
3032                         if (ret) {
3033                                 btrfs_abort_transaction(trans, ret);
3034                                 goto out;
3035                         }
3036                         btrfs_release_path(path);
3037
3038                         /* Slow path to locate EXTENT/METADATA_ITEM */
3039                         key.objectid = bytenr;
3040                         key.type = BTRFS_EXTENT_ITEM_KEY;
3041                         key.offset = num_bytes;
3042
3043                         if (!is_data && skinny_metadata) {
3044                                 key.type = BTRFS_METADATA_ITEM_KEY;
3045                                 key.offset = owner_objectid;
3046                         }
3047
3048                         ret = btrfs_search_slot(trans, extent_root,
3049                                                 &key, path, -1, 1);
3050                         if (ret > 0 && skinny_metadata && path->slots[0]) {
3051                                 /*
3052                                  * Couldn't find our skinny metadata item,
3053                                  * see if we have ye olde extent item.
3054                                  */
3055                                 path->slots[0]--;
3056                                 btrfs_item_key_to_cpu(path->nodes[0], &key,
3057                                                       path->slots[0]);
3058                                 if (key.objectid == bytenr &&
3059                                     key.type == BTRFS_EXTENT_ITEM_KEY &&
3060                                     key.offset == num_bytes)
3061                                         ret = 0;
3062                         }
3063
3064                         if (ret > 0 && skinny_metadata) {
3065                                 skinny_metadata = false;
3066                                 key.objectid = bytenr;
3067                                 key.type = BTRFS_EXTENT_ITEM_KEY;
3068                                 key.offset = num_bytes;
3069                                 btrfs_release_path(path);
3070                                 ret = btrfs_search_slot(trans, extent_root,
3071                                                         &key, path, -1, 1);
3072                         }
3073
3074                         if (ret) {
3075                                 if (ret > 0)
3076                                         btrfs_print_leaf(path->nodes[0]);
3077                                 btrfs_err(info,
3078                         "umm, got %d back from search, was looking for %llu, slot %d",
3079                                           ret, bytenr, path->slots[0]);
3080                         }
3081                         if (ret < 0) {
3082                                 btrfs_abort_transaction(trans, ret);
3083                                 goto out;
3084                         }
3085                         extent_slot = path->slots[0];
3086                 }
3087         } else if (WARN_ON(ret == -ENOENT)) {
3088                 abort_and_dump(trans, path,
3089 "unable to find ref byte nr %llu parent %llu root %llu owner %llu offset %llu slot %d",
3090                                bytenr, parent, root_objectid, owner_objectid,
3091                                owner_offset, path->slots[0]);
3092                 goto out;
3093         } else {
3094                 btrfs_abort_transaction(trans, ret);
3095                 goto out;
3096         }
3097
3098         leaf = path->nodes[0];
3099         item_size = btrfs_item_size(leaf, extent_slot);
3100         if (unlikely(item_size < sizeof(*ei))) {
3101                 ret = -EUCLEAN;
3102                 btrfs_err(trans->fs_info,
3103                           "unexpected extent item size, has %u expect >= %zu",
3104                           item_size, sizeof(*ei));
3105                 btrfs_abort_transaction(trans, ret);
3106                 goto out;
3107         }
3108         ei = btrfs_item_ptr(leaf, extent_slot,
3109                             struct btrfs_extent_item);
3110         if (owner_objectid < BTRFS_FIRST_FREE_OBJECTID &&
3111             key.type == BTRFS_EXTENT_ITEM_KEY) {
3112                 struct btrfs_tree_block_info *bi;
3113
3114                 if (item_size < sizeof(*ei) + sizeof(*bi)) {
3115                         abort_and_dump(trans, path,
3116 "invalid extent item size for key (%llu, %u, %llu) slot %u owner %llu, has %u expect >= %zu",
3117                                        key.objectid, key.type, key.offset,
3118                                        path->slots[0], owner_objectid, item_size,
3119                                        sizeof(*ei) + sizeof(*bi));
3120                         ret = -EUCLEAN;
3121                         goto out;
3122                 }
3123                 bi = (struct btrfs_tree_block_info *)(ei + 1);
3124                 WARN_ON(owner_objectid != btrfs_tree_block_level(leaf, bi));
3125         }
3126
3127         refs = btrfs_extent_refs(leaf, ei);
3128         if (refs < refs_to_drop) {
3129                 abort_and_dump(trans, path,
3130                 "trying to drop %d refs but we only have %llu for bytenr %llu slot %u",
3131                                refs_to_drop, refs, bytenr, path->slots[0]);
3132                 ret = -EUCLEAN;
3133                 goto out;
3134         }
3135         refs -= refs_to_drop;
3136
3137         if (refs > 0) {
3138                 if (extent_op)
3139                         __run_delayed_extent_op(extent_op, leaf, ei);
3140                 /*
3141                  * In the case of inline back ref, reference count will
3142                  * be updated by remove_extent_backref
3143                  */
3144                 if (iref) {
3145                         if (!found_extent) {
3146                                 abort_and_dump(trans, path,
3147 "invalid iref, got inlined extent ref but no EXTENT/METADATA_ITEM found, slot %u",
3148                                                path->slots[0]);
3149                                 ret = -EUCLEAN;
3150                                 goto out;
3151                         }
3152                 } else {
3153                         btrfs_set_extent_refs(leaf, ei, refs);
3154                         btrfs_mark_buffer_dirty(leaf);
3155                 }
3156                 if (found_extent) {
3157                         ret = remove_extent_backref(trans, extent_root, path,
3158                                                     iref, refs_to_drop, is_data);
3159                         if (ret) {
3160                                 btrfs_abort_transaction(trans, ret);
3161                                 goto out;
3162                         }
3163                 }
3164         } else {
3165                 /* In this branch refs == 1 */
3166                 if (found_extent) {
3167                         if (is_data && refs_to_drop !=
3168                             extent_data_ref_count(path, iref)) {
3169                                 abort_and_dump(trans, path,
3170                 "invalid refs_to_drop, current refs %u refs_to_drop %u slot %u",
3171                                                extent_data_ref_count(path, iref),
3172                                                refs_to_drop, path->slots[0]);
3173                                 ret = -EUCLEAN;
3174                                 goto out;
3175                         }
3176                         if (iref) {
3177                                 if (path->slots[0] != extent_slot) {
3178                                         abort_and_dump(trans, path,
3179 "invalid iref, extent item key (%llu %u %llu) slot %u doesn't have wanted iref",
3180                                                        key.objectid, key.type,
3181                                                        key.offset, path->slots[0]);
3182                                         ret = -EUCLEAN;
3183                                         goto out;
3184                                 }
3185                         } else {
3186                                 /*
3187                                  * No inline ref, we must be at SHARED_* item,
3188                                  * And it's single ref, it must be:
3189                                  * |    extent_slot       ||extent_slot + 1|
3190                                  * [ EXTENT/METADATA_ITEM ][ SHARED_* ITEM ]
3191                                  */
3192                                 if (path->slots[0] != extent_slot + 1) {
3193                                         abort_and_dump(trans, path,
3194         "invalid SHARED_* item slot %u, previous item is not EXTENT/METADATA_ITEM",
3195                                                        path->slots[0]);
3196                                         ret = -EUCLEAN;
3197                                         goto out;
3198                                 }
3199                                 path->slots[0] = extent_slot;
3200                                 num_to_del = 2;
3201                         }
3202                 }
3203
3204                 ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
3205                                       num_to_del);
3206                 if (ret) {
3207                         btrfs_abort_transaction(trans, ret);
3208                         goto out;
3209                 }
3210                 btrfs_release_path(path);
3211
3212                 ret = do_free_extent_accounting(trans, bytenr, num_bytes, is_data);
3213         }
3214         btrfs_release_path(path);
3215
3216 out:
3217         btrfs_free_path(path);
3218         return ret;
3219 }
3220
3221 /*
3222  * when we free an block, it is possible (and likely) that we free the last
3223  * delayed ref for that extent as well.  This searches the delayed ref tree for
3224  * a given extent, and if there are no other delayed refs to be processed, it
3225  * removes it from the tree.
3226  */
3227 static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans,
3228                                       u64 bytenr)
3229 {
3230         struct btrfs_delayed_ref_head *head;
3231         struct btrfs_delayed_ref_root *delayed_refs;
3232         int ret = 0;
3233
3234         delayed_refs = &trans->transaction->delayed_refs;
3235         spin_lock(&delayed_refs->lock);
3236         head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
3237         if (!head)
3238                 goto out_delayed_unlock;
3239
3240         spin_lock(&head->lock);
3241         if (!RB_EMPTY_ROOT(&head->ref_tree.rb_root))
3242                 goto out;
3243
3244         if (cleanup_extent_op(head) != NULL)
3245                 goto out;
3246
3247         /*
3248          * waiting for the lock here would deadlock.  If someone else has it
3249          * locked they are already in the process of dropping it anyway
3250          */
3251         if (!mutex_trylock(&head->mutex))
3252                 goto out;
3253
3254         btrfs_delete_ref_head(delayed_refs, head);
3255         head->processing = false;
3256
3257         spin_unlock(&head->lock);
3258         spin_unlock(&delayed_refs->lock);
3259
3260         BUG_ON(head->extent_op);
3261         if (head->must_insert_reserved)
3262                 ret = 1;
3263
3264         btrfs_cleanup_ref_head_accounting(trans->fs_info, delayed_refs, head);
3265         mutex_unlock(&head->mutex);
3266         btrfs_put_delayed_ref_head(head);
3267         return ret;
3268 out:
3269         spin_unlock(&head->lock);
3270
3271 out_delayed_unlock:
3272         spin_unlock(&delayed_refs->lock);
3273         return 0;
3274 }
3275
3276 void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
3277                            u64 root_id,
3278                            struct extent_buffer *buf,
3279                            u64 parent, int last_ref)
3280 {
3281         struct btrfs_fs_info *fs_info = trans->fs_info;
3282         struct btrfs_ref generic_ref = { 0 };
3283         int ret;
3284
3285         btrfs_init_generic_ref(&generic_ref, BTRFS_DROP_DELAYED_REF,
3286                                buf->start, buf->len, parent);
3287         btrfs_init_tree_ref(&generic_ref, btrfs_header_level(buf),
3288                             root_id, 0, false);
3289
3290         if (root_id != BTRFS_TREE_LOG_OBJECTID) {
3291                 btrfs_ref_tree_mod(fs_info, &generic_ref);
3292                 ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, NULL);
3293                 BUG_ON(ret); /* -ENOMEM */
3294         }
3295
3296         if (last_ref && btrfs_header_generation(buf) == trans->transid) {
3297                 struct btrfs_block_group *cache;
3298                 bool must_pin = false;
3299
3300                 if (root_id != BTRFS_TREE_LOG_OBJECTID) {
3301                         ret = check_ref_cleanup(trans, buf->start);
3302                         if (!ret) {
3303                                 btrfs_redirty_list_add(trans->transaction, buf);
3304                                 goto out;
3305                         }
3306                 }
3307
3308                 cache = btrfs_lookup_block_group(fs_info, buf->start);
3309
3310                 if (btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
3311                         pin_down_extent(trans, cache, buf->start, buf->len, 1);
3312                         btrfs_put_block_group(cache);
3313                         goto out;
3314                 }
3315
3316                 /*
3317                  * If there are tree mod log users we may have recorded mod log
3318                  * operations for this node.  If we re-allocate this node we
3319                  * could replay operations on this node that happened when it
3320                  * existed in a completely different root.  For example if it
3321                  * was part of root A, then was reallocated to root B, and we
3322                  * are doing a btrfs_old_search_slot(root b), we could replay
3323                  * operations that happened when the block was part of root A,
3324                  * giving us an inconsistent view of the btree.
3325                  *
3326                  * We are safe from races here because at this point no other
3327                  * node or root points to this extent buffer, so if after this
3328                  * check a new tree mod log user joins we will not have an
3329                  * existing log of operations on this node that we have to
3330                  * contend with.
3331                  */
3332                 if (test_bit(BTRFS_FS_TREE_MOD_LOG_USERS, &fs_info->flags))
3333                         must_pin = true;
3334
3335                 if (must_pin || btrfs_is_zoned(fs_info)) {
3336                         btrfs_redirty_list_add(trans->transaction, buf);
3337                         pin_down_extent(trans, cache, buf->start, buf->len, 1);
3338                         btrfs_put_block_group(cache);
3339                         goto out;
3340                 }
3341
3342                 WARN_ON(test_bit(EXTENT_BUFFER_DIRTY, &buf->bflags));
3343
3344                 btrfs_add_free_space(cache, buf->start, buf->len);
3345                 btrfs_free_reserved_bytes(cache, buf->len, 0);
3346                 btrfs_put_block_group(cache);
3347                 trace_btrfs_reserved_extent_free(fs_info, buf->start, buf->len);
3348         }
3349 out:
3350         if (last_ref) {
3351                 /*
3352                  * Deleting the buffer, clear the corrupt flag since it doesn't
3353                  * matter anymore.
3354                  */
3355                 clear_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags);
3356         }
3357 }
3358
3359 /* Can return -ENOMEM */
3360 int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_ref *ref)
3361 {
3362         struct btrfs_fs_info *fs_info = trans->fs_info;
3363         int ret;
3364
3365         if (btrfs_is_testing(fs_info))
3366                 return 0;
3367
3368         /*
3369          * tree log blocks never actually go into the extent allocation
3370          * tree, just update pinning info and exit early.
3371          */
3372         if ((ref->type == BTRFS_REF_METADATA &&
3373              ref->tree_ref.owning_root == BTRFS_TREE_LOG_OBJECTID) ||
3374             (ref->type == BTRFS_REF_DATA &&
3375              ref->data_ref.owning_root == BTRFS_TREE_LOG_OBJECTID)) {
3376                 /* unlocks the pinned mutex */
3377                 btrfs_pin_extent(trans, ref->bytenr, ref->len, 1);
3378                 ret = 0;
3379         } else if (ref->type == BTRFS_REF_METADATA) {
3380                 ret = btrfs_add_delayed_tree_ref(trans, ref, NULL);
3381         } else {
3382                 ret = btrfs_add_delayed_data_ref(trans, ref, 0);
3383         }
3384
3385         if (!((ref->type == BTRFS_REF_METADATA &&
3386                ref->tree_ref.owning_root == BTRFS_TREE_LOG_OBJECTID) ||
3387               (ref->type == BTRFS_REF_DATA &&
3388                ref->data_ref.owning_root == BTRFS_TREE_LOG_OBJECTID)))
3389                 btrfs_ref_tree_mod(fs_info, ref);
3390
3391         return ret;
3392 }
3393
3394 enum btrfs_loop_type {
3395         /*
3396          * Start caching block groups but do not wait for progress or for them
3397          * to be done.
3398          */
3399         LOOP_CACHING_NOWAIT,
3400
3401         /*
3402          * Wait for the block group free_space >= the space we're waiting for if
3403          * the block group isn't cached.
3404          */
3405         LOOP_CACHING_WAIT,
3406
3407         /*
3408          * Allow allocations to happen from block groups that do not yet have a
3409          * size classification.
3410          */
3411         LOOP_UNSET_SIZE_CLASS,
3412
3413         /*
3414          * Allocate a chunk and then retry the allocation.
3415          */
3416         LOOP_ALLOC_CHUNK,
3417
3418         /*
3419          * Ignore the size class restrictions for this allocation.
3420          */
3421         LOOP_WRONG_SIZE_CLASS,
3422
3423         /*
3424          * Ignore the empty size, only try to allocate the number of bytes
3425          * needed for this allocation.
3426          */
3427         LOOP_NO_EMPTY_SIZE,
3428 };
3429
3430 static inline void
3431 btrfs_lock_block_group(struct btrfs_block_group *cache,
3432                        int delalloc)
3433 {
3434         if (delalloc)
3435                 down_read(&cache->data_rwsem);
3436 }
3437
3438 static inline void btrfs_grab_block_group(struct btrfs_block_group *cache,
3439                        int delalloc)
3440 {
3441         btrfs_get_block_group(cache);
3442         if (delalloc)
3443                 down_read(&cache->data_rwsem);
3444 }
3445
3446 static struct btrfs_block_group *btrfs_lock_cluster(
3447                    struct btrfs_block_group *block_group,
3448                    struct btrfs_free_cluster *cluster,
3449                    int delalloc)
3450         __acquires(&cluster->refill_lock)
3451 {
3452         struct btrfs_block_group *used_bg = NULL;
3453
3454         spin_lock(&cluster->refill_lock);
3455         while (1) {
3456                 used_bg = cluster->block_group;
3457                 if (!used_bg)
3458                         return NULL;
3459
3460                 if (used_bg == block_group)
3461                         return used_bg;
3462
3463                 btrfs_get_block_group(used_bg);
3464
3465                 if (!delalloc)
3466                         return used_bg;
3467
3468                 if (down_read_trylock(&used_bg->data_rwsem))
3469                         return used_bg;
3470
3471                 spin_unlock(&cluster->refill_lock);
3472
3473                 /* We should only have one-level nested. */
3474                 down_read_nested(&used_bg->data_rwsem, SINGLE_DEPTH_NESTING);
3475
3476                 spin_lock(&cluster->refill_lock);
3477                 if (used_bg == cluster->block_group)
3478                         return used_bg;
3479
3480                 up_read(&used_bg->data_rwsem);
3481                 btrfs_put_block_group(used_bg);
3482         }
3483 }
3484
3485 static inline void
3486 btrfs_release_block_group(struct btrfs_block_group *cache,
3487                          int delalloc)
3488 {
3489         if (delalloc)
3490                 up_read(&cache->data_rwsem);
3491         btrfs_put_block_group(cache);
3492 }
3493
3494 /*
3495  * Helper function for find_free_extent().
3496  *
3497  * Return -ENOENT to inform caller that we need fallback to unclustered mode.
3498  * Return >0 to inform caller that we find nothing
3499  * Return 0 means we have found a location and set ffe_ctl->found_offset.
3500  */
3501 static int find_free_extent_clustered(struct btrfs_block_group *bg,
3502                                       struct find_free_extent_ctl *ffe_ctl,
3503                                       struct btrfs_block_group **cluster_bg_ret)
3504 {
3505         struct btrfs_block_group *cluster_bg;
3506         struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
3507         u64 aligned_cluster;
3508         u64 offset;
3509         int ret;
3510
3511         cluster_bg = btrfs_lock_cluster(bg, last_ptr, ffe_ctl->delalloc);
3512         if (!cluster_bg)
3513                 goto refill_cluster;
3514         if (cluster_bg != bg && (cluster_bg->ro ||
3515             !block_group_bits(cluster_bg, ffe_ctl->flags)))
3516                 goto release_cluster;
3517
3518         offset = btrfs_alloc_from_cluster(cluster_bg, last_ptr,
3519                         ffe_ctl->num_bytes, cluster_bg->start,
3520                         &ffe_ctl->max_extent_size);
3521         if (offset) {
3522                 /* We have a block, we're done */
3523                 spin_unlock(&last_ptr->refill_lock);
3524                 trace_btrfs_reserve_extent_cluster(cluster_bg, ffe_ctl);
3525                 *cluster_bg_ret = cluster_bg;
3526                 ffe_ctl->found_offset = offset;
3527                 return 0;
3528         }
3529         WARN_ON(last_ptr->block_group != cluster_bg);
3530
3531 release_cluster:
3532         /*
3533          * If we are on LOOP_NO_EMPTY_SIZE, we can't set up a new clusters, so
3534          * lets just skip it and let the allocator find whatever block it can
3535          * find. If we reach this point, we will have tried the cluster
3536          * allocator plenty of times and not have found anything, so we are
3537          * likely way too fragmented for the clustering stuff to find anything.
3538          *
3539          * However, if the cluster is taken from the current block group,
3540          * release the cluster first, so that we stand a better chance of
3541          * succeeding in the unclustered allocation.
3542          */
3543         if (ffe_ctl->loop >= LOOP_NO_EMPTY_SIZE && cluster_bg != bg) {
3544                 spin_unlock(&last_ptr->refill_lock);
3545                 btrfs_release_block_group(cluster_bg, ffe_ctl->delalloc);
3546                 return -ENOENT;
3547         }
3548
3549         /* This cluster didn't work out, free it and start over */
3550         btrfs_return_cluster_to_free_space(NULL, last_ptr);
3551
3552         if (cluster_bg != bg)
3553                 btrfs_release_block_group(cluster_bg, ffe_ctl->delalloc);
3554
3555 refill_cluster:
3556         if (ffe_ctl->loop >= LOOP_NO_EMPTY_SIZE) {
3557                 spin_unlock(&last_ptr->refill_lock);
3558                 return -ENOENT;
3559         }
3560
3561         aligned_cluster = max_t(u64,
3562                         ffe_ctl->empty_cluster + ffe_ctl->empty_size,
3563                         bg->full_stripe_len);
3564         ret = btrfs_find_space_cluster(bg, last_ptr, ffe_ctl->search_start,
3565                         ffe_ctl->num_bytes, aligned_cluster);
3566         if (ret == 0) {
3567                 /* Now pull our allocation out of this cluster */
3568                 offset = btrfs_alloc_from_cluster(bg, last_ptr,
3569                                 ffe_ctl->num_bytes, ffe_ctl->search_start,
3570                                 &ffe_ctl->max_extent_size);
3571                 if (offset) {
3572                         /* We found one, proceed */
3573                         spin_unlock(&last_ptr->refill_lock);
3574                         ffe_ctl->found_offset = offset;
3575                         trace_btrfs_reserve_extent_cluster(bg, ffe_ctl);
3576                         return 0;
3577                 }
3578         }
3579         /*
3580          * At this point we either didn't find a cluster or we weren't able to
3581          * allocate a block from our cluster.  Free the cluster we've been
3582          * trying to use, and go to the next block group.
3583          */
3584         btrfs_return_cluster_to_free_space(NULL, last_ptr);
3585         spin_unlock(&last_ptr->refill_lock);
3586         return 1;
3587 }
3588
3589 /*
3590  * Return >0 to inform caller that we find nothing
3591  * Return 0 when we found an free extent and set ffe_ctrl->found_offset
3592  */
3593 static int find_free_extent_unclustered(struct btrfs_block_group *bg,
3594                                         struct find_free_extent_ctl *ffe_ctl)
3595 {
3596         struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
3597         u64 offset;
3598
3599         /*
3600          * We are doing an unclustered allocation, set the fragmented flag so
3601          * we don't bother trying to setup a cluster again until we get more
3602          * space.
3603          */
3604         if (unlikely(last_ptr)) {
3605                 spin_lock(&last_ptr->lock);
3606                 last_ptr->fragmented = 1;
3607                 spin_unlock(&last_ptr->lock);
3608         }
3609         if (ffe_ctl->cached) {
3610                 struct btrfs_free_space_ctl *free_space_ctl;
3611
3612                 free_space_ctl = bg->free_space_ctl;
3613                 spin_lock(&free_space_ctl->tree_lock);
3614                 if (free_space_ctl->free_space <
3615                     ffe_ctl->num_bytes + ffe_ctl->empty_cluster +
3616                     ffe_ctl->empty_size) {
3617                         ffe_ctl->total_free_space = max_t(u64,
3618                                         ffe_ctl->total_free_space,
3619                                         free_space_ctl->free_space);
3620                         spin_unlock(&free_space_ctl->tree_lock);
3621                         return 1;
3622                 }
3623                 spin_unlock(&free_space_ctl->tree_lock);
3624         }
3625
3626         offset = btrfs_find_space_for_alloc(bg, ffe_ctl->search_start,
3627                         ffe_ctl->num_bytes, ffe_ctl->empty_size,
3628                         &ffe_ctl->max_extent_size);
3629         if (!offset)
3630                 return 1;
3631         ffe_ctl->found_offset = offset;
3632         return 0;
3633 }
3634
3635 static int do_allocation_clustered(struct btrfs_block_group *block_group,
3636                                    struct find_free_extent_ctl *ffe_ctl,
3637                                    struct btrfs_block_group **bg_ret)
3638 {
3639         int ret;
3640
3641         /* We want to try and use the cluster allocator, so lets look there */
3642         if (ffe_ctl->last_ptr && ffe_ctl->use_cluster) {
3643                 ret = find_free_extent_clustered(block_group, ffe_ctl, bg_ret);
3644                 if (ret >= 0)
3645                         return ret;
3646                 /* ret == -ENOENT case falls through */
3647         }
3648
3649         return find_free_extent_unclustered(block_group, ffe_ctl);
3650 }
3651
3652 /*
3653  * Tree-log block group locking
3654  * ============================
3655  *
3656  * fs_info::treelog_bg_lock protects the fs_info::treelog_bg which
3657  * indicates the starting address of a block group, which is reserved only
3658  * for tree-log metadata.
3659  *
3660  * Lock nesting
3661  * ============
3662  *
3663  * space_info::lock
3664  *   block_group::lock
3665  *     fs_info::treelog_bg_lock
3666  */
3667
3668 /*
3669  * Simple allocator for sequential-only block group. It only allows sequential
3670  * allocation. No need to play with trees. This function also reserves the
3671  * bytes as in btrfs_add_reserved_bytes.
3672  */
3673 static int do_allocation_zoned(struct btrfs_block_group *block_group,
3674                                struct find_free_extent_ctl *ffe_ctl,
3675                                struct btrfs_block_group **bg_ret)
3676 {
3677         struct btrfs_fs_info *fs_info = block_group->fs_info;
3678         struct btrfs_space_info *space_info = block_group->space_info;
3679         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3680         u64 start = block_group->start;
3681         u64 num_bytes = ffe_ctl->num_bytes;
3682         u64 avail;
3683         u64 bytenr = block_group->start;
3684         u64 log_bytenr;
3685         u64 data_reloc_bytenr;
3686         int ret = 0;
3687         bool skip = false;
3688
3689         ASSERT(btrfs_is_zoned(block_group->fs_info));
3690
3691         /*
3692          * Do not allow non-tree-log blocks in the dedicated tree-log block
3693          * group, and vice versa.
3694          */
3695         spin_lock(&fs_info->treelog_bg_lock);
3696         log_bytenr = fs_info->treelog_bg;
3697         if (log_bytenr && ((ffe_ctl->for_treelog && bytenr != log_bytenr) ||
3698                            (!ffe_ctl->for_treelog && bytenr == log_bytenr)))
3699                 skip = true;
3700         spin_unlock(&fs_info->treelog_bg_lock);
3701         if (skip)
3702                 return 1;
3703
3704         /*
3705          * Do not allow non-relocation blocks in the dedicated relocation block
3706          * group, and vice versa.
3707          */
3708         spin_lock(&fs_info->relocation_bg_lock);
3709         data_reloc_bytenr = fs_info->data_reloc_bg;
3710         if (data_reloc_bytenr &&
3711             ((ffe_ctl->for_data_reloc && bytenr != data_reloc_bytenr) ||
3712              (!ffe_ctl->for_data_reloc && bytenr == data_reloc_bytenr)))
3713                 skip = true;
3714         spin_unlock(&fs_info->relocation_bg_lock);
3715         if (skip)
3716                 return 1;
3717
3718         /* Check RO and no space case before trying to activate it */
3719         spin_lock(&block_group->lock);
3720         if (block_group->ro || btrfs_zoned_bg_is_full(block_group)) {
3721                 ret = 1;
3722                 /*
3723                  * May need to clear fs_info->{treelog,data_reloc}_bg.
3724                  * Return the error after taking the locks.
3725                  */
3726         }
3727         spin_unlock(&block_group->lock);
3728
3729         /* Metadata block group is activated at write time. */
3730         if (!ret && (block_group->flags & BTRFS_BLOCK_GROUP_DATA) &&
3731             !btrfs_zone_activate(block_group)) {
3732                 ret = 1;
3733                 /*
3734                  * May need to clear fs_info->{treelog,data_reloc}_bg.
3735                  * Return the error after taking the locks.
3736                  */
3737         }
3738
3739         spin_lock(&space_info->lock);
3740         spin_lock(&block_group->lock);
3741         spin_lock(&fs_info->treelog_bg_lock);
3742         spin_lock(&fs_info->relocation_bg_lock);
3743
3744         if (ret)
3745                 goto out;
3746
3747         ASSERT(!ffe_ctl->for_treelog ||
3748                block_group->start == fs_info->treelog_bg ||
3749                fs_info->treelog_bg == 0);
3750         ASSERT(!ffe_ctl->for_data_reloc ||
3751                block_group->start == fs_info->data_reloc_bg ||
3752                fs_info->data_reloc_bg == 0);
3753
3754         if (block_group->ro ||
3755             (!ffe_ctl->for_data_reloc &&
3756              test_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC, &block_group->runtime_flags))) {
3757                 ret = 1;
3758                 goto out;
3759         }
3760
3761         /*
3762          * Do not allow currently using block group to be tree-log dedicated
3763          * block group.
3764          */
3765         if (ffe_ctl->for_treelog && !fs_info->treelog_bg &&
3766             (block_group->used || block_group->reserved)) {
3767                 ret = 1;
3768                 goto out;
3769         }
3770
3771         /*
3772          * Do not allow currently used block group to be the data relocation
3773          * dedicated block group.
3774          */
3775         if (ffe_ctl->for_data_reloc && !fs_info->data_reloc_bg &&
3776             (block_group->used || block_group->reserved)) {
3777                 ret = 1;
3778                 goto out;
3779         }
3780
3781         WARN_ON_ONCE(block_group->alloc_offset > block_group->zone_capacity);
3782         avail = block_group->zone_capacity - block_group->alloc_offset;
3783         if (avail < num_bytes) {
3784                 if (ffe_ctl->max_extent_size < avail) {
3785                         /*
3786                          * With sequential allocator, free space is always
3787                          * contiguous
3788                          */
3789                         ffe_ctl->max_extent_size = avail;
3790                         ffe_ctl->total_free_space = avail;
3791                 }
3792                 ret = 1;
3793                 goto out;
3794         }
3795
3796         if (ffe_ctl->for_treelog && !fs_info->treelog_bg)
3797                 fs_info->treelog_bg = block_group->start;
3798
3799         if (ffe_ctl->for_data_reloc) {
3800                 if (!fs_info->data_reloc_bg)
3801                         fs_info->data_reloc_bg = block_group->start;
3802                 /*
3803                  * Do not allow allocations from this block group, unless it is
3804                  * for data relocation. Compared to increasing the ->ro, setting
3805                  * the ->zoned_data_reloc_ongoing flag still allows nocow
3806                  * writers to come in. See btrfs_inc_nocow_writers().
3807                  *
3808                  * We need to disable an allocation to avoid an allocation of
3809                  * regular (non-relocation data) extent. With mix of relocation
3810                  * extents and regular extents, we can dispatch WRITE commands
3811                  * (for relocation extents) and ZONE APPEND commands (for
3812                  * regular extents) at the same time to the same zone, which
3813                  * easily break the write pointer.
3814                  *
3815                  * Also, this flag avoids this block group to be zone finished.
3816                  */
3817                 set_bit(BLOCK_GROUP_FLAG_ZONED_DATA_RELOC, &block_group->runtime_flags);
3818         }
3819
3820         ffe_ctl->found_offset = start + block_group->alloc_offset;
3821         block_group->alloc_offset += num_bytes;
3822         spin_lock(&ctl->tree_lock);
3823         ctl->free_space -= num_bytes;
3824         spin_unlock(&ctl->tree_lock);
3825
3826         /*
3827          * We do not check if found_offset is aligned to stripesize. The
3828          * address is anyway rewritten when using zone append writing.
3829          */
3830
3831         ffe_ctl->search_start = ffe_ctl->found_offset;
3832
3833 out:
3834         if (ret && ffe_ctl->for_treelog)
3835                 fs_info->treelog_bg = 0;
3836         if (ret && ffe_ctl->for_data_reloc)
3837                 fs_info->data_reloc_bg = 0;
3838         spin_unlock(&fs_info->relocation_bg_lock);
3839         spin_unlock(&fs_info->treelog_bg_lock);
3840         spin_unlock(&block_group->lock);
3841         spin_unlock(&space_info->lock);
3842         return ret;
3843 }
3844
3845 static int do_allocation(struct btrfs_block_group *block_group,
3846                          struct find_free_extent_ctl *ffe_ctl,
3847                          struct btrfs_block_group **bg_ret)
3848 {
3849         switch (ffe_ctl->policy) {
3850         case BTRFS_EXTENT_ALLOC_CLUSTERED:
3851                 return do_allocation_clustered(block_group, ffe_ctl, bg_ret);
3852         case BTRFS_EXTENT_ALLOC_ZONED:
3853                 return do_allocation_zoned(block_group, ffe_ctl, bg_ret);
3854         default:
3855                 BUG();
3856         }
3857 }
3858
3859 static void release_block_group(struct btrfs_block_group *block_group,
3860                                 struct find_free_extent_ctl *ffe_ctl,
3861                                 int delalloc)
3862 {
3863         switch (ffe_ctl->policy) {
3864         case BTRFS_EXTENT_ALLOC_CLUSTERED:
3865                 ffe_ctl->retry_uncached = false;
3866                 break;
3867         case BTRFS_EXTENT_ALLOC_ZONED:
3868                 /* Nothing to do */
3869                 break;
3870         default:
3871                 BUG();
3872         }
3873
3874         BUG_ON(btrfs_bg_flags_to_raid_index(block_group->flags) !=
3875                ffe_ctl->index);
3876         btrfs_release_block_group(block_group, delalloc);
3877 }
3878
3879 static void found_extent_clustered(struct find_free_extent_ctl *ffe_ctl,
3880                                    struct btrfs_key *ins)
3881 {
3882         struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
3883
3884         if (!ffe_ctl->use_cluster && last_ptr) {
3885                 spin_lock(&last_ptr->lock);
3886                 last_ptr->window_start = ins->objectid;
3887                 spin_unlock(&last_ptr->lock);
3888         }
3889 }
3890
3891 static void found_extent(struct find_free_extent_ctl *ffe_ctl,
3892                          struct btrfs_key *ins)
3893 {
3894         switch (ffe_ctl->policy) {
3895         case BTRFS_EXTENT_ALLOC_CLUSTERED:
3896                 found_extent_clustered(ffe_ctl, ins);
3897                 break;
3898         case BTRFS_EXTENT_ALLOC_ZONED:
3899                 /* Nothing to do */
3900                 break;
3901         default:
3902                 BUG();
3903         }
3904 }
3905
3906 static int can_allocate_chunk_zoned(struct btrfs_fs_info *fs_info,
3907                                     struct find_free_extent_ctl *ffe_ctl)
3908 {
3909         /* Block group's activeness is not a requirement for METADATA block groups. */
3910         if (!(ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA))
3911                 return 0;
3912
3913         /* If we can activate new zone, just allocate a chunk and use it */
3914         if (btrfs_can_activate_zone(fs_info->fs_devices, ffe_ctl->flags))
3915                 return 0;
3916
3917         /*
3918          * We already reached the max active zones. Try to finish one block
3919          * group to make a room for a new block group. This is only possible
3920          * for a data block group because btrfs_zone_finish() may need to wait
3921          * for a running transaction which can cause a deadlock for metadata
3922          * allocation.
3923          */
3924         if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA) {
3925                 int ret = btrfs_zone_finish_one_bg(fs_info);
3926
3927                 if (ret == 1)
3928                         return 0;
3929                 else if (ret < 0)
3930                         return ret;
3931         }
3932
3933         /*
3934          * If we have enough free space left in an already active block group
3935          * and we can't activate any other zone now, do not allow allocating a
3936          * new chunk and let find_free_extent() retry with a smaller size.
3937          */
3938         if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size)
3939                 return -ENOSPC;
3940
3941         /*
3942          * Even min_alloc_size is not left in any block groups. Since we cannot
3943          * activate a new block group, allocating it may not help. Let's tell a
3944          * caller to try again and hope it progress something by writing some
3945          * parts of the region. That is only possible for data block groups,
3946          * where a part of the region can be written.
3947          */
3948         if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA)
3949                 return -EAGAIN;
3950
3951         /*
3952          * We cannot activate a new block group and no enough space left in any
3953          * block groups. So, allocating a new block group may not help. But,
3954          * there is nothing to do anyway, so let's go with it.
3955          */
3956         return 0;
3957 }
3958
3959 static int can_allocate_chunk(struct btrfs_fs_info *fs_info,
3960                               struct find_free_extent_ctl *ffe_ctl)
3961 {
3962         switch (ffe_ctl->policy) {
3963         case BTRFS_EXTENT_ALLOC_CLUSTERED:
3964                 return 0;
3965         case BTRFS_EXTENT_ALLOC_ZONED:
3966                 return can_allocate_chunk_zoned(fs_info, ffe_ctl);
3967         default:
3968                 BUG();
3969         }
3970 }
3971
3972 /*
3973  * Return >0 means caller needs to re-search for free extent
3974  * Return 0 means we have the needed free extent.
3975  * Return <0 means we failed to locate any free extent.
3976  */
3977 static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
3978                                         struct btrfs_key *ins,
3979                                         struct find_free_extent_ctl *ffe_ctl,
3980                                         bool full_search)
3981 {
3982         struct btrfs_root *root = fs_info->chunk_root;
3983         int ret;
3984
3985         if ((ffe_ctl->loop == LOOP_CACHING_NOWAIT) &&
3986             ffe_ctl->have_caching_bg && !ffe_ctl->orig_have_caching_bg)
3987                 ffe_ctl->orig_have_caching_bg = true;
3988
3989         if (ins->objectid) {
3990                 found_extent(ffe_ctl, ins);
3991                 return 0;
3992         }
3993
3994         if (ffe_ctl->loop >= LOOP_CACHING_WAIT && ffe_ctl->have_caching_bg)
3995                 return 1;
3996
3997         ffe_ctl->index++;
3998         if (ffe_ctl->index < BTRFS_NR_RAID_TYPES)
3999                 return 1;
4000
4001         /* See the comments for btrfs_loop_type for an explanation of the phases. */
4002         if (ffe_ctl->loop < LOOP_NO_EMPTY_SIZE) {
4003                 ffe_ctl->index = 0;
4004                 /*
4005                  * We want to skip the LOOP_CACHING_WAIT step if we don't have
4006                  * any uncached bgs and we've already done a full search
4007                  * through.
4008                  */
4009                 if (ffe_ctl->loop == LOOP_CACHING_NOWAIT &&
4010                     (!ffe_ctl->orig_have_caching_bg && full_search))
4011                         ffe_ctl->loop++;
4012                 ffe_ctl->loop++;
4013
4014                 if (ffe_ctl->loop == LOOP_ALLOC_CHUNK) {
4015                         struct btrfs_trans_handle *trans;
4016                         int exist = 0;
4017
4018                         /* Check if allocation policy allows to create a new chunk */
4019                         ret = can_allocate_chunk(fs_info, ffe_ctl);
4020                         if (ret)
4021                                 return ret;
4022
4023                         trans = current->journal_info;
4024                         if (trans)
4025                                 exist = 1;
4026                         else
4027                                 trans = btrfs_join_transaction(root);
4028
4029                         if (IS_ERR(trans)) {
4030                                 ret = PTR_ERR(trans);
4031                                 return ret;
4032                         }
4033
4034                         ret = btrfs_chunk_alloc(trans, ffe_ctl->flags,
4035                                                 CHUNK_ALLOC_FORCE_FOR_EXTENT);
4036
4037                         /* Do not bail out on ENOSPC since we can do more. */
4038                         if (ret == -ENOSPC) {
4039                                 ret = 0;
4040                                 ffe_ctl->loop++;
4041                         }
4042                         else if (ret < 0)
4043                                 btrfs_abort_transaction(trans, ret);
4044                         else
4045                                 ret = 0;
4046                         if (!exist)
4047                                 btrfs_end_transaction(trans);
4048                         if (ret)
4049                                 return ret;
4050                 }
4051
4052                 if (ffe_ctl->loop == LOOP_NO_EMPTY_SIZE) {
4053                         if (ffe_ctl->policy != BTRFS_EXTENT_ALLOC_CLUSTERED)
4054                                 return -ENOSPC;
4055
4056                         /*
4057                          * Don't loop again if we already have no empty_size and
4058                          * no empty_cluster.
4059                          */
4060                         if (ffe_ctl->empty_size == 0 &&
4061                             ffe_ctl->empty_cluster == 0)
4062                                 return -ENOSPC;
4063                         ffe_ctl->empty_size = 0;
4064                         ffe_ctl->empty_cluster = 0;
4065                 }
4066                 return 1;
4067         }
4068         return -ENOSPC;
4069 }
4070
4071 static bool find_free_extent_check_size_class(struct find_free_extent_ctl *ffe_ctl,
4072                                               struct btrfs_block_group *bg)
4073 {
4074         if (ffe_ctl->policy == BTRFS_EXTENT_ALLOC_ZONED)
4075                 return true;
4076         if (!btrfs_block_group_should_use_size_class(bg))
4077                 return true;
4078         if (ffe_ctl->loop >= LOOP_WRONG_SIZE_CLASS)
4079                 return true;
4080         if (ffe_ctl->loop >= LOOP_UNSET_SIZE_CLASS &&
4081             bg->size_class == BTRFS_BG_SZ_NONE)
4082                 return true;
4083         return ffe_ctl->size_class == bg->size_class;
4084 }
4085
4086 static int prepare_allocation_clustered(struct btrfs_fs_info *fs_info,
4087                                         struct find_free_extent_ctl *ffe_ctl,
4088                                         struct btrfs_space_info *space_info,
4089                                         struct btrfs_key *ins)
4090 {
4091         /*
4092          * If our free space is heavily fragmented we may not be able to make
4093          * big contiguous allocations, so instead of doing the expensive search
4094          * for free space, simply return ENOSPC with our max_extent_size so we
4095          * can go ahead and search for a more manageable chunk.
4096          *
4097          * If our max_extent_size is large enough for our allocation simply
4098          * disable clustering since we will likely not be able to find enough
4099          * space to create a cluster and induce latency trying.
4100          */
4101         if (space_info->max_extent_size) {
4102                 spin_lock(&space_info->lock);
4103                 if (space_info->max_extent_size &&
4104                     ffe_ctl->num_bytes > space_info->max_extent_size) {
4105                         ins->offset = space_info->max_extent_size;
4106                         spin_unlock(&space_info->lock);
4107                         return -ENOSPC;
4108                 } else if (space_info->max_extent_size) {
4109                         ffe_ctl->use_cluster = false;
4110                 }
4111                 spin_unlock(&space_info->lock);
4112         }
4113
4114         ffe_ctl->last_ptr = fetch_cluster_info(fs_info, space_info,
4115                                                &ffe_ctl->empty_cluster);
4116         if (ffe_ctl->last_ptr) {
4117                 struct btrfs_free_cluster *last_ptr = ffe_ctl->last_ptr;
4118
4119                 spin_lock(&last_ptr->lock);
4120                 if (last_ptr->block_group)
4121                         ffe_ctl->hint_byte = last_ptr->window_start;
4122                 if (last_ptr->fragmented) {
4123                         /*
4124                          * We still set window_start so we can keep track of the
4125                          * last place we found an allocation to try and save
4126                          * some time.
4127                          */
4128                         ffe_ctl->hint_byte = last_ptr->window_start;
4129                         ffe_ctl->use_cluster = false;
4130                 }
4131                 spin_unlock(&last_ptr->lock);
4132         }
4133
4134         return 0;
4135 }
4136
4137 static int prepare_allocation(struct btrfs_fs_info *fs_info,
4138                               struct find_free_extent_ctl *ffe_ctl,
4139                               struct btrfs_space_info *space_info,
4140                               struct btrfs_key *ins)
4141 {
4142         switch (ffe_ctl->policy) {
4143         case BTRFS_EXTENT_ALLOC_CLUSTERED:
4144                 return prepare_allocation_clustered(fs_info, ffe_ctl,
4145                                                     space_info, ins);
4146         case BTRFS_EXTENT_ALLOC_ZONED:
4147                 if (ffe_ctl->for_treelog) {
4148                         spin_lock(&fs_info->treelog_bg_lock);
4149                         if (fs_info->treelog_bg)
4150                                 ffe_ctl->hint_byte = fs_info->treelog_bg;
4151                         spin_unlock(&fs_info->treelog_bg_lock);
4152                 }
4153                 if (ffe_ctl->for_data_reloc) {
4154                         spin_lock(&fs_info->relocation_bg_lock);
4155                         if (fs_info->data_reloc_bg)
4156                                 ffe_ctl->hint_byte = fs_info->data_reloc_bg;
4157                         spin_unlock(&fs_info->relocation_bg_lock);
4158                 }
4159                 return 0;
4160         default:
4161                 BUG();
4162         }
4163 }
4164
4165 /*
4166  * walks the btree of allocated extents and find a hole of a given size.
4167  * The key ins is changed to record the hole:
4168  * ins->objectid == start position
4169  * ins->flags = BTRFS_EXTENT_ITEM_KEY
4170  * ins->offset == the size of the hole.
4171  * Any available blocks before search_start are skipped.
4172  *
4173  * If there is no suitable free space, we will record the max size of
4174  * the free space extent currently.
4175  *
4176  * The overall logic and call chain:
4177  *
4178  * find_free_extent()
4179  * |- Iterate through all block groups
4180  * |  |- Get a valid block group
4181  * |  |- Try to do clustered allocation in that block group
4182  * |  |- Try to do unclustered allocation in that block group
4183  * |  |- Check if the result is valid
4184  * |  |  |- If valid, then exit
4185  * |  |- Jump to next block group
4186  * |
4187  * |- Push harder to find free extents
4188  *    |- If not found, re-iterate all block groups
4189  */
4190 static noinline int find_free_extent(struct btrfs_root *root,
4191                                      struct btrfs_key *ins,
4192                                      struct find_free_extent_ctl *ffe_ctl)
4193 {
4194         struct btrfs_fs_info *fs_info = root->fs_info;
4195         int ret = 0;
4196         int cache_block_group_error = 0;
4197         struct btrfs_block_group *block_group = NULL;
4198         struct btrfs_space_info *space_info;
4199         bool full_search = false;
4200
4201         WARN_ON(ffe_ctl->num_bytes < fs_info->sectorsize);
4202
4203         ffe_ctl->search_start = 0;
4204         /* For clustered allocation */
4205         ffe_ctl->empty_cluster = 0;
4206         ffe_ctl->last_ptr = NULL;
4207         ffe_ctl->use_cluster = true;
4208         ffe_ctl->have_caching_bg = false;
4209         ffe_ctl->orig_have_caching_bg = false;
4210         ffe_ctl->index = btrfs_bg_flags_to_raid_index(ffe_ctl->flags);
4211         ffe_ctl->loop = 0;
4212         ffe_ctl->retry_uncached = false;
4213         ffe_ctl->cached = 0;
4214         ffe_ctl->max_extent_size = 0;
4215         ffe_ctl->total_free_space = 0;
4216         ffe_ctl->found_offset = 0;
4217         ffe_ctl->policy = BTRFS_EXTENT_ALLOC_CLUSTERED;
4218         ffe_ctl->size_class = btrfs_calc_block_group_size_class(ffe_ctl->num_bytes);
4219
4220         if (btrfs_is_zoned(fs_info))
4221                 ffe_ctl->policy = BTRFS_EXTENT_ALLOC_ZONED;
4222
4223         ins->type = BTRFS_EXTENT_ITEM_KEY;
4224         ins->objectid = 0;
4225         ins->offset = 0;
4226
4227         trace_find_free_extent(root, ffe_ctl);
4228
4229         space_info = btrfs_find_space_info(fs_info, ffe_ctl->flags);
4230         if (!space_info) {
4231                 btrfs_err(fs_info, "No space info for %llu", ffe_ctl->flags);
4232                 return -ENOSPC;
4233         }
4234
4235         ret = prepare_allocation(fs_info, ffe_ctl, space_info, ins);
4236         if (ret < 0)
4237                 return ret;
4238
4239         ffe_ctl->search_start = max(ffe_ctl->search_start,
4240                                     first_logical_byte(fs_info));
4241         ffe_ctl->search_start = max(ffe_ctl->search_start, ffe_ctl->hint_byte);
4242         if (ffe_ctl->search_start == ffe_ctl->hint_byte) {
4243                 block_group = btrfs_lookup_block_group(fs_info,
4244                                                        ffe_ctl->search_start);
4245                 /*
4246                  * we don't want to use the block group if it doesn't match our
4247                  * allocation bits, or if its not cached.
4248                  *
4249                  * However if we are re-searching with an ideal block group
4250                  * picked out then we don't care that the block group is cached.
4251                  */
4252                 if (block_group && block_group_bits(block_group, ffe_ctl->flags) &&
4253                     block_group->cached != BTRFS_CACHE_NO) {
4254                         down_read(&space_info->groups_sem);
4255                         if (list_empty(&block_group->list) ||
4256                             block_group->ro) {
4257                                 /*
4258                                  * someone is removing this block group,
4259                                  * we can't jump into the have_block_group
4260                                  * target because our list pointers are not
4261                                  * valid
4262                                  */
4263                                 btrfs_put_block_group(block_group);
4264                                 up_read(&space_info->groups_sem);
4265                         } else {
4266                                 ffe_ctl->index = btrfs_bg_flags_to_raid_index(
4267                                                         block_group->flags);
4268                                 btrfs_lock_block_group(block_group,
4269                                                        ffe_ctl->delalloc);
4270                                 ffe_ctl->hinted = true;
4271                                 goto have_block_group;
4272                         }
4273                 } else if (block_group) {
4274                         btrfs_put_block_group(block_group);
4275                 }
4276         }
4277 search:
4278         trace_find_free_extent_search_loop(root, ffe_ctl);
4279         ffe_ctl->have_caching_bg = false;
4280         if (ffe_ctl->index == btrfs_bg_flags_to_raid_index(ffe_ctl->flags) ||
4281             ffe_ctl->index == 0)
4282                 full_search = true;
4283         down_read(&space_info->groups_sem);
4284         list_for_each_entry(block_group,
4285                             &space_info->block_groups[ffe_ctl->index], list) {
4286                 struct btrfs_block_group *bg_ret;
4287
4288                 ffe_ctl->hinted = false;
4289                 /* If the block group is read-only, we can skip it entirely. */
4290                 if (unlikely(block_group->ro)) {
4291                         if (ffe_ctl->for_treelog)
4292                                 btrfs_clear_treelog_bg(block_group);
4293                         if (ffe_ctl->for_data_reloc)
4294                                 btrfs_clear_data_reloc_bg(block_group);
4295                         continue;
4296                 }
4297
4298                 btrfs_grab_block_group(block_group, ffe_ctl->delalloc);
4299                 ffe_ctl->search_start = block_group->start;
4300
4301                 /*
4302                  * this can happen if we end up cycling through all the
4303                  * raid types, but we want to make sure we only allocate
4304                  * for the proper type.
4305                  */
4306                 if (!block_group_bits(block_group, ffe_ctl->flags)) {
4307                         u64 extra = BTRFS_BLOCK_GROUP_DUP |
4308                                 BTRFS_BLOCK_GROUP_RAID1_MASK |
4309                                 BTRFS_BLOCK_GROUP_RAID56_MASK |
4310                                 BTRFS_BLOCK_GROUP_RAID10;
4311
4312                         /*
4313                          * if they asked for extra copies and this block group
4314                          * doesn't provide them, bail.  This does allow us to
4315                          * fill raid0 from raid1.
4316                          */
4317                         if ((ffe_ctl->flags & extra) && !(block_group->flags & extra))
4318                                 goto loop;
4319
4320                         /*
4321                          * This block group has different flags than we want.
4322                          * It's possible that we have MIXED_GROUP flag but no
4323                          * block group is mixed.  Just skip such block group.
4324                          */
4325                         btrfs_release_block_group(block_group, ffe_ctl->delalloc);
4326                         continue;
4327                 }
4328
4329 have_block_group:
4330                 trace_find_free_extent_have_block_group(root, ffe_ctl, block_group);
4331                 ffe_ctl->cached = btrfs_block_group_done(block_group);
4332                 if (unlikely(!ffe_ctl->cached)) {
4333                         ffe_ctl->have_caching_bg = true;
4334                         ret = btrfs_cache_block_group(block_group, false);
4335
4336                         /*
4337                          * If we get ENOMEM here or something else we want to
4338                          * try other block groups, because it may not be fatal.
4339                          * However if we can't find anything else we need to
4340                          * save our return here so that we return the actual
4341                          * error that caused problems, not ENOSPC.
4342                          */
4343                         if (ret < 0) {
4344                                 if (!cache_block_group_error)
4345                                         cache_block_group_error = ret;
4346                                 ret = 0;
4347                                 goto loop;
4348                         }
4349                         ret = 0;
4350                 }
4351
4352                 if (unlikely(block_group->cached == BTRFS_CACHE_ERROR)) {
4353                         if (!cache_block_group_error)
4354                                 cache_block_group_error = -EIO;
4355                         goto loop;
4356                 }
4357
4358                 if (!find_free_extent_check_size_class(ffe_ctl, block_group))
4359                         goto loop;
4360
4361                 bg_ret = NULL;
4362                 ret = do_allocation(block_group, ffe_ctl, &bg_ret);
4363                 if (ret > 0)
4364                         goto loop;
4365
4366                 if (bg_ret && bg_ret != block_group) {
4367                         btrfs_release_block_group(block_group, ffe_ctl->delalloc);
4368                         block_group = bg_ret;
4369                 }
4370
4371                 /* Checks */
4372                 ffe_ctl->search_start = round_up(ffe_ctl->found_offset,
4373                                                  fs_info->stripesize);
4374
4375                 /* move on to the next group */
4376                 if (ffe_ctl->search_start + ffe_ctl->num_bytes >
4377                     block_group->start + block_group->length) {
4378                         btrfs_add_free_space_unused(block_group,
4379                                             ffe_ctl->found_offset,
4380                                             ffe_ctl->num_bytes);
4381                         goto loop;
4382                 }
4383
4384                 if (ffe_ctl->found_offset < ffe_ctl->search_start)
4385                         btrfs_add_free_space_unused(block_group,
4386                                         ffe_ctl->found_offset,
4387                                         ffe_ctl->search_start - ffe_ctl->found_offset);
4388
4389                 ret = btrfs_add_reserved_bytes(block_group, ffe_ctl->ram_bytes,
4390                                                ffe_ctl->num_bytes,
4391                                                ffe_ctl->delalloc,
4392                                                ffe_ctl->loop >= LOOP_WRONG_SIZE_CLASS);
4393                 if (ret == -EAGAIN) {
4394                         btrfs_add_free_space_unused(block_group,
4395                                         ffe_ctl->found_offset,
4396                                         ffe_ctl->num_bytes);
4397                         goto loop;
4398                 }
4399                 btrfs_inc_block_group_reservations(block_group);
4400
4401                 /* we are all good, lets return */
4402                 ins->objectid = ffe_ctl->search_start;
4403                 ins->offset = ffe_ctl->num_bytes;
4404
4405                 trace_btrfs_reserve_extent(block_group, ffe_ctl);
4406                 btrfs_release_block_group(block_group, ffe_ctl->delalloc);
4407                 break;
4408 loop:
4409                 if (!ffe_ctl->cached && ffe_ctl->loop > LOOP_CACHING_NOWAIT &&
4410                     !ffe_ctl->retry_uncached) {
4411                         ffe_ctl->retry_uncached = true;
4412                         btrfs_wait_block_group_cache_progress(block_group,
4413                                                 ffe_ctl->num_bytes +
4414                                                 ffe_ctl->empty_cluster +
4415                                                 ffe_ctl->empty_size);
4416                         goto have_block_group;
4417                 }
4418                 release_block_group(block_group, ffe_ctl, ffe_ctl->delalloc);
4419                 cond_resched();
4420         }
4421         up_read(&space_info->groups_sem);
4422
4423         ret = find_free_extent_update_loop(fs_info, ins, ffe_ctl, full_search);
4424         if (ret > 0)
4425                 goto search;
4426
4427         if (ret == -ENOSPC && !cache_block_group_error) {
4428                 /*
4429                  * Use ffe_ctl->total_free_space as fallback if we can't find
4430                  * any contiguous hole.
4431                  */
4432                 if (!ffe_ctl->max_extent_size)
4433                         ffe_ctl->max_extent_size = ffe_ctl->total_free_space;
4434                 spin_lock(&space_info->lock);
4435                 space_info->max_extent_size = ffe_ctl->max_extent_size;
4436                 spin_unlock(&space_info->lock);
4437                 ins->offset = ffe_ctl->max_extent_size;
4438         } else if (ret == -ENOSPC) {
4439                 ret = cache_block_group_error;
4440         }
4441         return ret;
4442 }
4443
4444 /*
4445  * btrfs_reserve_extent - entry point to the extent allocator. Tries to find a
4446  *                        hole that is at least as big as @num_bytes.
4447  *
4448  * @root           -    The root that will contain this extent
4449  *
4450  * @ram_bytes      -    The amount of space in ram that @num_bytes take. This
4451  *                      is used for accounting purposes. This value differs
4452  *                      from @num_bytes only in the case of compressed extents.
4453  *
4454  * @num_bytes      -    Number of bytes to allocate on-disk.
4455  *
4456  * @min_alloc_size -    Indicates the minimum amount of space that the
4457  *                      allocator should try to satisfy. In some cases
4458  *                      @num_bytes may be larger than what is required and if
4459  *                      the filesystem is fragmented then allocation fails.
4460  *                      However, the presence of @min_alloc_size gives a
4461  *                      chance to try and satisfy the smaller allocation.
4462  *
4463  * @empty_size     -    A hint that you plan on doing more COW. This is the
4464  *                      size in bytes the allocator should try to find free
4465  *                      next to the block it returns.  This is just a hint and
4466  *                      may be ignored by the allocator.
4467  *
4468  * @hint_byte      -    Hint to the allocator to start searching above the byte
4469  *                      address passed. It might be ignored.
4470  *
4471  * @ins            -    This key is modified to record the found hole. It will
4472  *                      have the following values:
4473  *                      ins->objectid == start position
4474  *                      ins->flags = BTRFS_EXTENT_ITEM_KEY
4475  *                      ins->offset == the size of the hole.
4476  *
4477  * @is_data        -    Boolean flag indicating whether an extent is
4478  *                      allocated for data (true) or metadata (false)
4479  *
4480  * @delalloc       -    Boolean flag indicating whether this allocation is for
4481  *                      delalloc or not. If 'true' data_rwsem of block groups
4482  *                      is going to be acquired.
4483  *
4484  *
4485  * Returns 0 when an allocation succeeded or < 0 when an error occurred. In
4486  * case -ENOSPC is returned then @ins->offset will contain the size of the
4487  * largest available hole the allocator managed to find.
4488  */
4489 int btrfs_reserve_extent(struct btrfs_root *root, u64 ram_bytes,
4490                          u64 num_bytes, u64 min_alloc_size,
4491                          u64 empty_size, u64 hint_byte,
4492                          struct btrfs_key *ins, int is_data, int delalloc)
4493 {
4494         struct btrfs_fs_info *fs_info = root->fs_info;
4495         struct find_free_extent_ctl ffe_ctl = {};
4496         bool final_tried = num_bytes == min_alloc_size;
4497         u64 flags;
4498         int ret;
4499         bool for_treelog = (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID);
4500         bool for_data_reloc = (btrfs_is_data_reloc_root(root) && is_data);
4501
4502         flags = get_alloc_profile_by_root(root, is_data);
4503 again:
4504         WARN_ON(num_bytes < fs_info->sectorsize);
4505
4506         ffe_ctl.ram_bytes = ram_bytes;
4507         ffe_ctl.num_bytes = num_bytes;
4508         ffe_ctl.min_alloc_size = min_alloc_size;
4509         ffe_ctl.empty_size = empty_size;
4510         ffe_ctl.flags = flags;
4511         ffe_ctl.delalloc = delalloc;
4512         ffe_ctl.hint_byte = hint_byte;
4513         ffe_ctl.for_treelog = for_treelog;
4514         ffe_ctl.for_data_reloc = for_data_reloc;
4515
4516         ret = find_free_extent(root, ins, &ffe_ctl);
4517         if (!ret && !is_data) {
4518                 btrfs_dec_block_group_reservations(fs_info, ins->objectid);
4519         } else if (ret == -ENOSPC) {
4520                 if (!final_tried && ins->offset) {
4521                         num_bytes = min(num_bytes >> 1, ins->offset);
4522                         num_bytes = round_down(num_bytes,
4523                                                fs_info->sectorsize);
4524                         num_bytes = max(num_bytes, min_alloc_size);
4525                         ram_bytes = num_bytes;
4526                         if (num_bytes == min_alloc_size)
4527                                 final_tried = true;
4528                         goto again;
4529                 } else if (btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
4530                         struct btrfs_space_info *sinfo;
4531
4532                         sinfo = btrfs_find_space_info(fs_info, flags);
4533                         btrfs_err(fs_info,
4534         "allocation failed flags %llu, wanted %llu tree-log %d, relocation: %d",
4535                                   flags, num_bytes, for_treelog, for_data_reloc);
4536                         if (sinfo)
4537                                 btrfs_dump_space_info(fs_info, sinfo,
4538                                                       num_bytes, 1);
4539                 }
4540         }
4541
4542         return ret;
4543 }
4544
4545 int btrfs_free_reserved_extent(struct btrfs_fs_info *fs_info,
4546                                u64 start, u64 len, int delalloc)
4547 {
4548         struct btrfs_block_group *cache;
4549
4550         cache = btrfs_lookup_block_group(fs_info, start);
4551         if (!cache) {
4552                 btrfs_err(fs_info, "Unable to find block group for %llu",
4553                           start);
4554                 return -ENOSPC;
4555         }
4556
4557         btrfs_add_free_space(cache, start, len);
4558         btrfs_free_reserved_bytes(cache, len, delalloc);
4559         trace_btrfs_reserved_extent_free(fs_info, start, len);
4560
4561         btrfs_put_block_group(cache);
4562         return 0;
4563 }
4564
4565 int btrfs_pin_reserved_extent(struct btrfs_trans_handle *trans, u64 start,
4566                               u64 len)
4567 {
4568         struct btrfs_block_group *cache;
4569         int ret = 0;
4570
4571         cache = btrfs_lookup_block_group(trans->fs_info, start);
4572         if (!cache) {
4573                 btrfs_err(trans->fs_info, "unable to find block group for %llu",
4574                           start);
4575                 return -ENOSPC;
4576         }
4577
4578         ret = pin_down_extent(trans, cache, start, len, 1);
4579         btrfs_put_block_group(cache);
4580         return ret;
4581 }
4582
4583 static int alloc_reserved_extent(struct btrfs_trans_handle *trans, u64 bytenr,
4584                                  u64 num_bytes)
4585 {
4586         struct btrfs_fs_info *fs_info = trans->fs_info;
4587         int ret;
4588
4589         ret = remove_from_free_space_tree(trans, bytenr, num_bytes);
4590         if (ret)
4591                 return ret;
4592
4593         ret = btrfs_update_block_group(trans, bytenr, num_bytes, true);
4594         if (ret) {
4595                 ASSERT(!ret);
4596                 btrfs_err(fs_info, "update block group failed for %llu %llu",
4597                           bytenr, num_bytes);
4598                 return ret;
4599         }
4600
4601         trace_btrfs_reserved_extent_alloc(fs_info, bytenr, num_bytes);
4602         return 0;
4603 }
4604
4605 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
4606                                       u64 parent, u64 root_objectid,
4607                                       u64 flags, u64 owner, u64 offset,
4608                                       struct btrfs_key *ins, int ref_mod)
4609 {
4610         struct btrfs_fs_info *fs_info = trans->fs_info;
4611         struct btrfs_root *extent_root;
4612         int ret;
4613         struct btrfs_extent_item *extent_item;
4614         struct btrfs_extent_inline_ref *iref;
4615         struct btrfs_path *path;
4616         struct extent_buffer *leaf;
4617         int type;
4618         u32 size;
4619
4620         if (parent > 0)
4621                 type = BTRFS_SHARED_DATA_REF_KEY;
4622         else
4623                 type = BTRFS_EXTENT_DATA_REF_KEY;
4624
4625         size = sizeof(*extent_item) + btrfs_extent_inline_ref_size(type);
4626
4627         path = btrfs_alloc_path();
4628         if (!path)
4629                 return -ENOMEM;
4630
4631         extent_root = btrfs_extent_root(fs_info, ins->objectid);
4632         ret = btrfs_insert_empty_item(trans, extent_root, path, ins, size);
4633         if (ret) {
4634                 btrfs_free_path(path);
4635                 return ret;
4636         }
4637
4638         leaf = path->nodes[0];
4639         extent_item = btrfs_item_ptr(leaf, path->slots[0],
4640                                      struct btrfs_extent_item);
4641         btrfs_set_extent_refs(leaf, extent_item, ref_mod);
4642         btrfs_set_extent_generation(leaf, extent_item, trans->transid);
4643         btrfs_set_extent_flags(leaf, extent_item,
4644                                flags | BTRFS_EXTENT_FLAG_DATA);
4645
4646         iref = (struct btrfs_extent_inline_ref *)(extent_item + 1);
4647         btrfs_set_extent_inline_ref_type(leaf, iref, type);
4648         if (parent > 0) {
4649                 struct btrfs_shared_data_ref *ref;
4650                 ref = (struct btrfs_shared_data_ref *)(iref + 1);
4651                 btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
4652                 btrfs_set_shared_data_ref_count(leaf, ref, ref_mod);
4653         } else {
4654                 struct btrfs_extent_data_ref *ref;
4655                 ref = (struct btrfs_extent_data_ref *)(&iref->offset);
4656                 btrfs_set_extent_data_ref_root(leaf, ref, root_objectid);
4657                 btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
4658                 btrfs_set_extent_data_ref_offset(leaf, ref, offset);
4659                 btrfs_set_extent_data_ref_count(leaf, ref, ref_mod);
4660         }
4661
4662         btrfs_mark_buffer_dirty(path->nodes[0]);
4663         btrfs_free_path(path);
4664
4665         return alloc_reserved_extent(trans, ins->objectid, ins->offset);
4666 }
4667
4668 static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
4669                                      struct btrfs_delayed_ref_node *node,
4670                                      struct btrfs_delayed_extent_op *extent_op)
4671 {
4672         struct btrfs_fs_info *fs_info = trans->fs_info;
4673         struct btrfs_root *extent_root;
4674         int ret;
4675         struct btrfs_extent_item *extent_item;
4676         struct btrfs_key extent_key;
4677         struct btrfs_tree_block_info *block_info;
4678         struct btrfs_extent_inline_ref *iref;
4679         struct btrfs_path *path;
4680         struct extent_buffer *leaf;
4681         struct btrfs_delayed_tree_ref *ref;
4682         u32 size = sizeof(*extent_item) + sizeof(*iref);
4683         u64 flags = extent_op->flags_to_set;
4684         bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
4685
4686         ref = btrfs_delayed_node_to_tree_ref(node);
4687
4688         extent_key.objectid = node->bytenr;
4689         if (skinny_metadata) {
4690                 extent_key.offset = ref->level;
4691                 extent_key.type = BTRFS_METADATA_ITEM_KEY;
4692         } else {
4693                 extent_key.offset = node->num_bytes;
4694                 extent_key.type = BTRFS_EXTENT_ITEM_KEY;
4695                 size += sizeof(*block_info);
4696         }
4697
4698         path = btrfs_alloc_path();
4699         if (!path)
4700                 return -ENOMEM;
4701
4702         extent_root = btrfs_extent_root(fs_info, extent_key.objectid);
4703         ret = btrfs_insert_empty_item(trans, extent_root, path, &extent_key,
4704                                       size);
4705         if (ret) {
4706                 btrfs_free_path(path);
4707                 return ret;
4708         }
4709
4710         leaf = path->nodes[0];
4711         extent_item = btrfs_item_ptr(leaf, path->slots[0],
4712                                      struct btrfs_extent_item);
4713         btrfs_set_extent_refs(leaf, extent_item, 1);
4714         btrfs_set_extent_generation(leaf, extent_item, trans->transid);
4715         btrfs_set_extent_flags(leaf, extent_item,
4716                                flags | BTRFS_EXTENT_FLAG_TREE_BLOCK);
4717
4718         if (skinny_metadata) {
4719                 iref = (struct btrfs_extent_inline_ref *)(extent_item + 1);
4720         } else {
4721                 block_info = (struct btrfs_tree_block_info *)(extent_item + 1);
4722                 btrfs_set_tree_block_key(leaf, block_info, &extent_op->key);
4723                 btrfs_set_tree_block_level(leaf, block_info, ref->level);
4724                 iref = (struct btrfs_extent_inline_ref *)(block_info + 1);
4725         }
4726
4727         if (node->type == BTRFS_SHARED_BLOCK_REF_KEY) {
4728                 btrfs_set_extent_inline_ref_type(leaf, iref,
4729                                                  BTRFS_SHARED_BLOCK_REF_KEY);
4730                 btrfs_set_extent_inline_ref_offset(leaf, iref, ref->parent);
4731         } else {
4732                 btrfs_set_extent_inline_ref_type(leaf, iref,
4733                                                  BTRFS_TREE_BLOCK_REF_KEY);
4734                 btrfs_set_extent_inline_ref_offset(leaf, iref, ref->root);
4735         }
4736
4737         btrfs_mark_buffer_dirty(leaf);
4738         btrfs_free_path(path);
4739
4740         return alloc_reserved_extent(trans, node->bytenr, fs_info->nodesize);
4741 }
4742
4743 int btrfs_alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
4744                                      struct btrfs_root *root, u64 owner,
4745                                      u64 offset, u64 ram_bytes,
4746                                      struct btrfs_key *ins)
4747 {
4748         struct btrfs_ref generic_ref = { 0 };
4749
4750         BUG_ON(root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID);
4751
4752         btrfs_init_generic_ref(&generic_ref, BTRFS_ADD_DELAYED_EXTENT,
4753                                ins->objectid, ins->offset, 0);
4754         btrfs_init_data_ref(&generic_ref, root->root_key.objectid, owner,
4755                             offset, 0, false);
4756         btrfs_ref_tree_mod(root->fs_info, &generic_ref);
4757
4758         return btrfs_add_delayed_data_ref(trans, &generic_ref, ram_bytes);
4759 }
4760
4761 /*
4762  * this is used by the tree logging recovery code.  It records that
4763  * an extent has been allocated and makes sure to clear the free
4764  * space cache bits as well
4765  */
4766 int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
4767                                    u64 root_objectid, u64 owner, u64 offset,
4768                                    struct btrfs_key *ins)
4769 {
4770         struct btrfs_fs_info *fs_info = trans->fs_info;
4771         int ret;
4772         struct btrfs_block_group *block_group;
4773         struct btrfs_space_info *space_info;
4774
4775         /*
4776          * Mixed block groups will exclude before processing the log so we only
4777          * need to do the exclude dance if this fs isn't mixed.
4778          */
4779         if (!btrfs_fs_incompat(fs_info, MIXED_GROUPS)) {
4780                 ret = __exclude_logged_extent(fs_info, ins->objectid,
4781                                               ins->offset);
4782                 if (ret)
4783                         return ret;
4784         }
4785
4786         block_group = btrfs_lookup_block_group(fs_info, ins->objectid);
4787         if (!block_group)
4788                 return -EINVAL;
4789
4790         space_info = block_group->space_info;
4791         spin_lock(&space_info->lock);
4792         spin_lock(&block_group->lock);
4793         space_info->bytes_reserved += ins->offset;
4794         block_group->reserved += ins->offset;
4795         spin_unlock(&block_group->lock);
4796         spin_unlock(&space_info->lock);
4797
4798         ret = alloc_reserved_file_extent(trans, 0, root_objectid, 0, owner,
4799                                          offset, ins, 1);
4800         if (ret)
4801                 btrfs_pin_extent(trans, ins->objectid, ins->offset, 1);
4802         btrfs_put_block_group(block_group);
4803         return ret;
4804 }
4805
4806 static struct extent_buffer *
4807 btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct btrfs_root *root,
4808                       u64 bytenr, int level, u64 owner,
4809                       enum btrfs_lock_nesting nest)
4810 {
4811         struct btrfs_fs_info *fs_info = root->fs_info;
4812         struct extent_buffer *buf;
4813         u64 lockdep_owner = owner;
4814
4815         buf = btrfs_find_create_tree_block(fs_info, bytenr, owner, level);
4816         if (IS_ERR(buf))
4817                 return buf;
4818
4819         /*
4820          * Extra safety check in case the extent tree is corrupted and extent
4821          * allocator chooses to use a tree block which is already used and
4822          * locked.
4823          */
4824         if (buf->lock_owner == current->pid) {
4825                 btrfs_err_rl(fs_info,
4826 "tree block %llu owner %llu already locked by pid=%d, extent tree corruption detected",
4827                         buf->start, btrfs_header_owner(buf), current->pid);
4828                 free_extent_buffer(buf);
4829                 return ERR_PTR(-EUCLEAN);
4830         }
4831
4832         /*
4833          * The reloc trees are just snapshots, so we need them to appear to be
4834          * just like any other fs tree WRT lockdep.
4835          *
4836          * The exception however is in replace_path() in relocation, where we
4837          * hold the lock on the original fs root and then search for the reloc
4838          * root.  At that point we need to make sure any reloc root buffers are
4839          * set to the BTRFS_TREE_RELOC_OBJECTID lockdep class in order to make
4840          * lockdep happy.
4841          */
4842         if (lockdep_owner == BTRFS_TREE_RELOC_OBJECTID &&
4843             !test_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &root->state))
4844                 lockdep_owner = BTRFS_FS_TREE_OBJECTID;
4845
4846         /* btrfs_clear_buffer_dirty() accesses generation field. */
4847         btrfs_set_header_generation(buf, trans->transid);
4848
4849         /*
4850          * This needs to stay, because we could allocate a freed block from an
4851          * old tree into a new tree, so we need to make sure this new block is
4852          * set to the appropriate level and owner.
4853          */
4854         btrfs_set_buffer_lockdep_class(lockdep_owner, buf, level);
4855
4856         __btrfs_tree_lock(buf, nest);
4857         btrfs_clear_buffer_dirty(trans, buf);
4858         clear_bit(EXTENT_BUFFER_STALE, &buf->bflags);
4859         clear_bit(EXTENT_BUFFER_NO_CHECK, &buf->bflags);
4860
4861         set_extent_buffer_uptodate(buf);
4862
4863         memzero_extent_buffer(buf, 0, sizeof(struct btrfs_header));
4864         btrfs_set_header_level(buf, level);
4865         btrfs_set_header_bytenr(buf, buf->start);
4866         btrfs_set_header_generation(buf, trans->transid);
4867         btrfs_set_header_backref_rev(buf, BTRFS_MIXED_BACKREF_REV);
4868         btrfs_set_header_owner(buf, owner);
4869         write_extent_buffer_fsid(buf, fs_info->fs_devices->metadata_uuid);
4870         write_extent_buffer_chunk_tree_uuid(buf, fs_info->chunk_tree_uuid);
4871         if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
4872                 buf->log_index = root->log_transid % 2;
4873                 /*
4874                  * we allow two log transactions at a time, use different
4875                  * EXTENT bit to differentiate dirty pages.
4876                  */
4877                 if (buf->log_index == 0)
4878                         set_extent_bit(&root->dirty_log_pages, buf->start,
4879                                        buf->start + buf->len - 1,
4880                                        EXTENT_DIRTY, NULL);
4881                 else
4882                         set_extent_bit(&root->dirty_log_pages, buf->start,
4883                                        buf->start + buf->len - 1,
4884                                        EXTENT_NEW, NULL);
4885         } else {
4886                 buf->log_index = -1;
4887                 set_extent_bit(&trans->transaction->dirty_pages, buf->start,
4888                                buf->start + buf->len - 1, EXTENT_DIRTY, NULL);
4889         }
4890         /* this returns a buffer locked for blocking */
4891         return buf;
4892 }
4893
4894 /*
4895  * finds a free extent and does all the dirty work required for allocation
4896  * returns the tree buffer or an ERR_PTR on error.
4897  */
4898 struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
4899                                              struct btrfs_root *root,
4900                                              u64 parent, u64 root_objectid,
4901                                              const struct btrfs_disk_key *key,
4902                                              int level, u64 hint,
4903                                              u64 empty_size,
4904                                              enum btrfs_lock_nesting nest)
4905 {
4906         struct btrfs_fs_info *fs_info = root->fs_info;
4907         struct btrfs_key ins;
4908         struct btrfs_block_rsv *block_rsv;
4909         struct extent_buffer *buf;
4910         struct btrfs_delayed_extent_op *extent_op;
4911         struct btrfs_ref generic_ref = { 0 };
4912         u64 flags = 0;
4913         int ret;
4914         u32 blocksize = fs_info->nodesize;
4915         bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
4916
4917 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
4918         if (btrfs_is_testing(fs_info)) {
4919                 buf = btrfs_init_new_buffer(trans, root, root->alloc_bytenr,
4920                                             level, root_objectid, nest);
4921                 if (!IS_ERR(buf))
4922                         root->alloc_bytenr += blocksize;
4923                 return buf;
4924         }
4925 #endif
4926
4927         block_rsv = btrfs_use_block_rsv(trans, root, blocksize);
4928         if (IS_ERR(block_rsv))
4929                 return ERR_CAST(block_rsv);
4930
4931         ret = btrfs_reserve_extent(root, blocksize, blocksize, blocksize,
4932                                    empty_size, hint, &ins, 0, 0);
4933         if (ret)
4934                 goto out_unuse;
4935
4936         buf = btrfs_init_new_buffer(trans, root, ins.objectid, level,
4937                                     root_objectid, nest);
4938         if (IS_ERR(buf)) {
4939                 ret = PTR_ERR(buf);
4940                 goto out_free_reserved;
4941         }
4942
4943         if (root_objectid == BTRFS_TREE_RELOC_OBJECTID) {
4944                 if (parent == 0)
4945                         parent = ins.objectid;
4946                 flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
4947         } else
4948                 BUG_ON(parent > 0);
4949
4950         if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
4951                 extent_op = btrfs_alloc_delayed_extent_op();
4952                 if (!extent_op) {
4953                         ret = -ENOMEM;
4954                         goto out_free_buf;
4955                 }
4956                 if (key)
4957                         memcpy(&extent_op->key, key, sizeof(extent_op->key));
4958                 else
4959                         memset(&extent_op->key, 0, sizeof(extent_op->key));
4960                 extent_op->flags_to_set = flags;
4961                 extent_op->update_key = skinny_metadata ? false : true;
4962                 extent_op->update_flags = true;
4963                 extent_op->level = level;
4964
4965                 btrfs_init_generic_ref(&generic_ref, BTRFS_ADD_DELAYED_EXTENT,
4966                                        ins.objectid, ins.offset, parent);
4967                 btrfs_init_tree_ref(&generic_ref, level, root_objectid,
4968                                     root->root_key.objectid, false);
4969                 btrfs_ref_tree_mod(fs_info, &generic_ref);
4970                 ret = btrfs_add_delayed_tree_ref(trans, &generic_ref, extent_op);
4971                 if (ret)
4972                         goto out_free_delayed;
4973         }
4974         return buf;
4975
4976 out_free_delayed:
4977         btrfs_free_delayed_extent_op(extent_op);
4978 out_free_buf:
4979         btrfs_tree_unlock(buf);
4980         free_extent_buffer(buf);
4981 out_free_reserved:
4982         btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 0);
4983 out_unuse:
4984         btrfs_unuse_block_rsv(fs_info, block_rsv, blocksize);
4985         return ERR_PTR(ret);
4986 }
4987
4988 struct walk_control {
4989         u64 refs[BTRFS_MAX_LEVEL];
4990         u64 flags[BTRFS_MAX_LEVEL];
4991         struct btrfs_key update_progress;
4992         struct btrfs_key drop_progress;
4993         int drop_level;
4994         int stage;
4995         int level;
4996         int shared_level;
4997         int update_ref;
4998         int keep_locks;
4999         int reada_slot;
5000         int reada_count;
5001         int restarted;
5002 };
5003
5004 #define DROP_REFERENCE  1
5005 #define UPDATE_BACKREF  2
5006
5007 static noinline void reada_walk_down(struct btrfs_trans_handle *trans,
5008                                      struct btrfs_root *root,
5009                                      struct walk_control *wc,
5010                                      struct btrfs_path *path)
5011 {
5012         struct btrfs_fs_info *fs_info = root->fs_info;
5013         u64 bytenr;
5014         u64 generation;
5015         u64 refs;
5016         u64 flags;
5017         u32 nritems;
5018         struct btrfs_key key;
5019         struct extent_buffer *eb;
5020         int ret;
5021         int slot;
5022         int nread = 0;
5023
5024         if (path->slots[wc->level] < wc->reada_slot) {
5025                 wc->reada_count = wc->reada_count * 2 / 3;
5026                 wc->reada_count = max(wc->reada_count, 2);
5027         } else {
5028                 wc->reada_count = wc->reada_count * 3 / 2;
5029                 wc->reada_count = min_t(int, wc->reada_count,
5030                                         BTRFS_NODEPTRS_PER_BLOCK(fs_info));
5031         }
5032
5033         eb = path->nodes[wc->level];
5034         nritems = btrfs_header_nritems(eb);
5035
5036         for (slot = path->slots[wc->level]; slot < nritems; slot++) {
5037                 if (nread >= wc->reada_count)
5038                         break;
5039
5040                 cond_resched();
5041                 bytenr = btrfs_node_blockptr(eb, slot);
5042                 generation = btrfs_node_ptr_generation(eb, slot);
5043
5044                 if (slot == path->slots[wc->level])
5045                         goto reada;
5046
5047                 if (wc->stage == UPDATE_BACKREF &&
5048                     generation <= root->root_key.offset)
5049                         continue;
5050
5051                 /* We don't lock the tree block, it's OK to be racy here */
5052                 ret = btrfs_lookup_extent_info(trans, fs_info, bytenr,
5053                                                wc->level - 1, 1, &refs,
5054                                                &flags);
5055                 /* We don't care about errors in readahead. */
5056                 if (ret < 0)
5057                         continue;
5058                 BUG_ON(refs == 0);
5059
5060                 if (wc->stage == DROP_REFERENCE) {
5061                         if (refs == 1)
5062                                 goto reada;
5063
5064                         if (wc->level == 1 &&
5065                             (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF))
5066                                 continue;
5067                         if (!wc->update_ref ||
5068                             generation <= root->root_key.offset)
5069                                 continue;
5070                         btrfs_node_key_to_cpu(eb, &key, slot);
5071                         ret = btrfs_comp_cpu_keys(&key,
5072                                                   &wc->update_progress);
5073                         if (ret < 0)
5074                                 continue;
5075                 } else {
5076                         if (wc->level == 1 &&
5077                             (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF))
5078                                 continue;
5079                 }
5080 reada:
5081                 btrfs_readahead_node_child(eb, slot);
5082                 nread++;
5083         }
5084         wc->reada_slot = slot;
5085 }
5086
5087 /*
5088  * helper to process tree block while walking down the tree.
5089  *
5090  * when wc->stage == UPDATE_BACKREF, this function updates
5091  * back refs for pointers in the block.
5092  *
5093  * NOTE: return value 1 means we should stop walking down.
5094  */
5095 static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
5096                                    struct btrfs_root *root,
5097                                    struct btrfs_path *path,
5098                                    struct walk_control *wc, int lookup_info)
5099 {
5100         struct btrfs_fs_info *fs_info = root->fs_info;
5101         int level = wc->level;
5102         struct extent_buffer *eb = path->nodes[level];
5103         u64 flag = BTRFS_BLOCK_FLAG_FULL_BACKREF;
5104         int ret;
5105
5106         if (wc->stage == UPDATE_BACKREF &&
5107             btrfs_header_owner(eb) != root->root_key.objectid)
5108                 return 1;
5109
5110         /*
5111          * when reference count of tree block is 1, it won't increase
5112          * again. once full backref flag is set, we never clear it.
5113          */
5114         if (lookup_info &&
5115             ((wc->stage == DROP_REFERENCE && wc->refs[level] != 1) ||
5116              (wc->stage == UPDATE_BACKREF && !(wc->flags[level] & flag)))) {
5117                 BUG_ON(!path->locks[level]);
5118                 ret = btrfs_lookup_extent_info(trans, fs_info,
5119                                                eb->start, level, 1,
5120                                                &wc->refs[level],
5121                                                &wc->flags[level]);
5122                 BUG_ON(ret == -ENOMEM);
5123                 if (ret)
5124                         return ret;
5125                 BUG_ON(wc->refs[level] == 0);
5126         }
5127
5128         if (wc->stage == DROP_REFERENCE) {
5129                 if (wc->refs[level] > 1)
5130                         return 1;
5131
5132                 if (path->locks[level] && !wc->keep_locks) {
5133                         btrfs_tree_unlock_rw(eb, path->locks[level]);
5134                         path->locks[level] = 0;
5135                 }
5136                 return 0;
5137         }
5138
5139         /* wc->stage == UPDATE_BACKREF */
5140         if (!(wc->flags[level] & flag)) {
5141                 BUG_ON(!path->locks[level]);
5142                 ret = btrfs_inc_ref(trans, root, eb, 1);
5143                 BUG_ON(ret); /* -ENOMEM */
5144                 ret = btrfs_dec_ref(trans, root, eb, 0);
5145                 BUG_ON(ret); /* -ENOMEM */
5146                 ret = btrfs_set_disk_extent_flags(trans, eb, flag);
5147                 BUG_ON(ret); /* -ENOMEM */
5148                 wc->flags[level] |= flag;
5149         }
5150
5151         /*
5152          * the block is shared by multiple trees, so it's not good to
5153          * keep the tree lock
5154          */
5155         if (path->locks[level] && level > 0) {
5156                 btrfs_tree_unlock_rw(eb, path->locks[level]);
5157                 path->locks[level] = 0;
5158         }
5159         return 0;
5160 }
5161
5162 /*
5163  * This is used to verify a ref exists for this root to deal with a bug where we
5164  * would have a drop_progress key that hadn't been updated properly.
5165  */
5166 static int check_ref_exists(struct btrfs_trans_handle *trans,
5167                             struct btrfs_root *root, u64 bytenr, u64 parent,
5168                             int level)
5169 {
5170         struct btrfs_path *path;
5171         struct btrfs_extent_inline_ref *iref;
5172         int ret;
5173
5174         path = btrfs_alloc_path();
5175         if (!path)
5176                 return -ENOMEM;
5177
5178         ret = lookup_extent_backref(trans, path, &iref, bytenr,
5179                                     root->fs_info->nodesize, parent,
5180                                     root->root_key.objectid, level, 0);
5181         btrfs_free_path(path);
5182         if (ret == -ENOENT)
5183                 return 0;
5184         if (ret < 0)
5185                 return ret;
5186         return 1;
5187 }
5188
5189 /*
5190  * helper to process tree block pointer.
5191  *
5192  * when wc->stage == DROP_REFERENCE, this function checks
5193  * reference count of the block pointed to. if the block
5194  * is shared and we need update back refs for the subtree
5195  * rooted at the block, this function changes wc->stage to
5196  * UPDATE_BACKREF. if the block is shared and there is no
5197  * need to update back, this function drops the reference
5198  * to the block.
5199  *
5200  * NOTE: return value 1 means we should stop walking down.
5201  */
5202 static noinline int do_walk_down(struct btrfs_trans_handle *trans,
5203                                  struct btrfs_root *root,
5204                                  struct btrfs_path *path,
5205                                  struct walk_control *wc, int *lookup_info)
5206 {
5207         struct btrfs_fs_info *fs_info = root->fs_info;
5208         u64 bytenr;
5209         u64 generation;
5210         u64 parent;
5211         struct btrfs_tree_parent_check check = { 0 };
5212         struct btrfs_key key;
5213         struct btrfs_ref ref = { 0 };
5214         struct extent_buffer *next;
5215         int level = wc->level;
5216         int reada = 0;
5217         int ret = 0;
5218         bool need_account = false;
5219
5220         generation = btrfs_node_ptr_generation(path->nodes[level],
5221                                                path->slots[level]);
5222         /*
5223          * if the lower level block was created before the snapshot
5224          * was created, we know there is no need to update back refs
5225          * for the subtree
5226          */
5227         if (wc->stage == UPDATE_BACKREF &&
5228             generation <= root->root_key.offset) {
5229                 *lookup_info = 1;
5230                 return 1;
5231         }
5232
5233         bytenr = btrfs_node_blockptr(path->nodes[level], path->slots[level]);
5234
5235         check.level = level - 1;
5236         check.transid = generation;
5237         check.owner_root = root->root_key.objectid;
5238         check.has_first_key = true;
5239         btrfs_node_key_to_cpu(path->nodes[level], &check.first_key,
5240                               path->slots[level]);
5241
5242         next = find_extent_buffer(fs_info, bytenr);
5243         if (!next) {
5244                 next = btrfs_find_create_tree_block(fs_info, bytenr,
5245                                 root->root_key.objectid, level - 1);
5246                 if (IS_ERR(next))
5247                         return PTR_ERR(next);
5248                 reada = 1;
5249         }
5250         btrfs_tree_lock(next);
5251
5252         ret = btrfs_lookup_extent_info(trans, fs_info, bytenr, level - 1, 1,
5253                                        &wc->refs[level - 1],
5254                                        &wc->flags[level - 1]);
5255         if (ret < 0)
5256                 goto out_unlock;
5257
5258         if (unlikely(wc->refs[level - 1] == 0)) {
5259                 btrfs_err(fs_info, "Missing references.");
5260                 ret = -EIO;
5261                 goto out_unlock;
5262         }
5263         *lookup_info = 0;
5264
5265         if (wc->stage == DROP_REFERENCE) {
5266                 if (wc->refs[level - 1] > 1) {
5267                         need_account = true;
5268                         if (level == 1 &&
5269                             (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF))
5270                                 goto skip;
5271
5272                         if (!wc->update_ref ||
5273                             generation <= root->root_key.offset)
5274                                 goto skip;
5275
5276                         btrfs_node_key_to_cpu(path->nodes[level], &key,
5277                                               path->slots[level]);
5278                         ret = btrfs_comp_cpu_keys(&key, &wc->update_progress);
5279                         if (ret < 0)
5280                                 goto skip;
5281
5282                         wc->stage = UPDATE_BACKREF;
5283                         wc->shared_level = level - 1;
5284                 }
5285         } else {
5286                 if (level == 1 &&
5287                     (wc->flags[0] & BTRFS_BLOCK_FLAG_FULL_BACKREF))
5288                         goto skip;
5289         }
5290
5291         if (!btrfs_buffer_uptodate(next, generation, 0)) {
5292                 btrfs_tree_unlock(next);
5293                 free_extent_buffer(next);
5294                 next = NULL;
5295                 *lookup_info = 1;
5296         }
5297
5298         if (!next) {
5299                 if (reada && level == 1)
5300                         reada_walk_down(trans, root, wc, path);
5301                 next = read_tree_block(fs_info, bytenr, &check);
5302                 if (IS_ERR(next)) {
5303                         return PTR_ERR(next);
5304                 } else if (!extent_buffer_uptodate(next)) {
5305                         free_extent_buffer(next);
5306                         return -EIO;
5307                 }
5308                 btrfs_tree_lock(next);
5309         }
5310
5311         level--;
5312         ASSERT(level == btrfs_header_level(next));
5313         if (level != btrfs_header_level(next)) {
5314                 btrfs_err(root->fs_info, "mismatched level");
5315                 ret = -EIO;
5316                 goto out_unlock;
5317         }
5318         path->nodes[level] = next;
5319         path->slots[level] = 0;
5320         path->locks[level] = BTRFS_WRITE_LOCK;
5321         wc->level = level;
5322         if (wc->level == 1)
5323                 wc->reada_slot = 0;
5324         return 0;
5325 skip:
5326         wc->refs[level - 1] = 0;
5327         wc->flags[level - 1] = 0;
5328         if (wc->stage == DROP_REFERENCE) {
5329                 if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
5330                         parent = path->nodes[level]->start;
5331                 } else {
5332                         ASSERT(root->root_key.objectid ==
5333                                btrfs_header_owner(path->nodes[level]));
5334                         if (root->root_key.objectid !=
5335                             btrfs_header_owner(path->nodes[level])) {
5336                                 btrfs_err(root->fs_info,
5337                                                 "mismatched block owner");
5338                                 ret = -EIO;
5339                                 goto out_unlock;
5340                         }
5341                         parent = 0;
5342                 }
5343
5344                 /*
5345                  * If we had a drop_progress we need to verify the refs are set
5346                  * as expected.  If we find our ref then we know that from here
5347                  * on out everything should be correct, and we can clear the
5348                  * ->restarted flag.
5349                  */
5350                 if (wc->restarted) {
5351                         ret = check_ref_exists(trans, root, bytenr, parent,
5352                                                level - 1);
5353                         if (ret < 0)
5354                                 goto out_unlock;
5355                         if (ret == 0)
5356                                 goto no_delete;
5357                         ret = 0;
5358                         wc->restarted = 0;
5359                 }
5360
5361                 /*
5362                  * Reloc tree doesn't contribute to qgroup numbers, and we have
5363                  * already accounted them at merge time (replace_path),
5364                  * thus we could skip expensive subtree trace here.
5365                  */
5366                 if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID &&
5367                     need_account) {
5368                         ret = btrfs_qgroup_trace_subtree(trans, next,
5369                                                          generation, level - 1);
5370                         if (ret) {
5371                                 btrfs_err_rl(fs_info,
5372                                              "Error %d accounting shared subtree. Quota is out of sync, rescan required.",
5373                                              ret);
5374                         }
5375                 }
5376
5377                 /*
5378                  * We need to update the next key in our walk control so we can
5379                  * update the drop_progress key accordingly.  We don't care if
5380                  * find_next_key doesn't find a key because that means we're at
5381                  * the end and are going to clean up now.
5382                  */
5383                 wc->drop_level = level;
5384                 find_next_key(path, level, &wc->drop_progress);
5385
5386                 btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, bytenr,
5387                                        fs_info->nodesize, parent);
5388                 btrfs_init_tree_ref(&ref, level - 1, root->root_key.objectid,
5389                                     0, false);
5390                 ret = btrfs_free_extent(trans, &ref);
5391                 if (ret)
5392                         goto out_unlock;
5393         }
5394 no_delete:
5395         *lookup_info = 1;
5396         ret = 1;
5397
5398 out_unlock:
5399         btrfs_tree_unlock(next);
5400         free_extent_buffer(next);
5401
5402         return ret;
5403 }
5404
5405 /*
5406  * helper to process tree block while walking up the tree.
5407  *
5408  * when wc->stage == DROP_REFERENCE, this function drops
5409  * reference count on the block.
5410  *
5411  * when wc->stage == UPDATE_BACKREF, this function changes
5412  * wc->stage back to DROP_REFERENCE if we changed wc->stage
5413  * to UPDATE_BACKREF previously while processing the block.
5414  *
5415  * NOTE: return value 1 means we should stop walking up.
5416  */
5417 static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
5418                                  struct btrfs_root *root,
5419                                  struct btrfs_path *path,
5420                                  struct walk_control *wc)
5421 {
5422         struct btrfs_fs_info *fs_info = root->fs_info;
5423         int ret;
5424         int level = wc->level;
5425         struct extent_buffer *eb = path->nodes[level];
5426         u64 parent = 0;
5427
5428         if (wc->stage == UPDATE_BACKREF) {
5429                 BUG_ON(wc->shared_level < level);
5430                 if (level < wc->shared_level)
5431                         goto out;
5432
5433                 ret = find_next_key(path, level + 1, &wc->update_progress);
5434                 if (ret > 0)
5435                         wc->update_ref = 0;
5436
5437                 wc->stage = DROP_REFERENCE;
5438                 wc->shared_level = -1;
5439                 path->slots[level] = 0;
5440
5441                 /*
5442                  * check reference count again if the block isn't locked.
5443                  * we should start walking down the tree again if reference
5444                  * count is one.
5445                  */
5446                 if (!path->locks[level]) {
5447                         BUG_ON(level == 0);
5448                         btrfs_tree_lock(eb);
5449                         path->locks[level] = BTRFS_WRITE_LOCK;
5450
5451                         ret = btrfs_lookup_extent_info(trans, fs_info,
5452                                                        eb->start, level, 1,
5453                                                        &wc->refs[level],
5454                                                        &wc->flags[level]);
5455                         if (ret < 0) {
5456                                 btrfs_tree_unlock_rw(eb, path->locks[level]);
5457                                 path->locks[level] = 0;
5458                                 return ret;
5459                         }
5460                         BUG_ON(wc->refs[level] == 0);
5461                         if (wc->refs[level] == 1) {
5462                                 btrfs_tree_unlock_rw(eb, path->locks[level]);
5463                                 path->locks[level] = 0;
5464                                 return 1;
5465                         }
5466                 }
5467         }
5468
5469         /* wc->stage == DROP_REFERENCE */
5470         BUG_ON(wc->refs[level] > 1 && !path->locks[level]);
5471
5472         if (wc->refs[level] == 1) {
5473                 if (level == 0) {
5474                         if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5475                                 ret = btrfs_dec_ref(trans, root, eb, 1);
5476                         else
5477                                 ret = btrfs_dec_ref(trans, root, eb, 0);
5478                         BUG_ON(ret); /* -ENOMEM */
5479                         if (is_fstree(root->root_key.objectid)) {
5480                                 ret = btrfs_qgroup_trace_leaf_items(trans, eb);
5481                                 if (ret) {
5482                                         btrfs_err_rl(fs_info,
5483         "error %d accounting leaf items, quota is out of sync, rescan required",
5484                                              ret);
5485                                 }
5486                         }
5487                 }
5488                 /* Make block locked assertion in btrfs_clear_buffer_dirty happy. */
5489                 if (!path->locks[level]) {
5490                         btrfs_tree_lock(eb);
5491                         path->locks[level] = BTRFS_WRITE_LOCK;
5492                 }
5493                 btrfs_clear_buffer_dirty(trans, eb);
5494         }
5495
5496         if (eb == root->node) {
5497                 if (wc->flags[level] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5498                         parent = eb->start;
5499                 else if (root->root_key.objectid != btrfs_header_owner(eb))
5500                         goto owner_mismatch;
5501         } else {
5502                 if (wc->flags[level + 1] & BTRFS_BLOCK_FLAG_FULL_BACKREF)
5503                         parent = path->nodes[level + 1]->start;
5504                 else if (root->root_key.objectid !=
5505                          btrfs_header_owner(path->nodes[level + 1]))
5506                         goto owner_mismatch;
5507         }
5508
5509         btrfs_free_tree_block(trans, btrfs_root_id(root), eb, parent,
5510                               wc->refs[level] == 1);
5511 out:
5512         wc->refs[level] = 0;
5513         wc->flags[level] = 0;
5514         return 0;
5515
5516 owner_mismatch:
5517         btrfs_err_rl(fs_info, "unexpected tree owner, have %llu expect %llu",
5518                      btrfs_header_owner(eb), root->root_key.objectid);
5519         return -EUCLEAN;
5520 }
5521
5522 static noinline int walk_down_tree(struct btrfs_trans_handle *trans,
5523                                    struct btrfs_root *root,
5524                                    struct btrfs_path *path,
5525                                    struct walk_control *wc)
5526 {
5527         int level = wc->level;
5528         int lookup_info = 1;
5529         int ret = 0;
5530
5531         while (level >= 0) {
5532                 ret = walk_down_proc(trans, root, path, wc, lookup_info);
5533                 if (ret)
5534                         break;
5535
5536                 if (level == 0)
5537                         break;
5538
5539                 if (path->slots[level] >=
5540                     btrfs_header_nritems(path->nodes[level]))
5541                         break;
5542
5543                 ret = do_walk_down(trans, root, path, wc, &lookup_info);
5544                 if (ret > 0) {
5545                         path->slots[level]++;
5546                         continue;
5547                 } else if (ret < 0)
5548                         break;
5549                 level = wc->level;
5550         }
5551         return (ret == 1) ? 0 : ret;
5552 }
5553
5554 static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
5555                                  struct btrfs_root *root,
5556                                  struct btrfs_path *path,
5557                                  struct walk_control *wc, int max_level)
5558 {
5559         int level = wc->level;
5560         int ret;
5561
5562         path->slots[level] = btrfs_header_nritems(path->nodes[level]);
5563         while (level < max_level && path->nodes[level]) {
5564                 wc->level = level;
5565                 if (path->slots[level] + 1 <
5566                     btrfs_header_nritems(path->nodes[level])) {
5567                         path->slots[level]++;
5568                         return 0;
5569                 } else {
5570                         ret = walk_up_proc(trans, root, path, wc);
5571                         if (ret > 0)
5572                                 return 0;
5573                         if (ret < 0)
5574                                 return ret;
5575
5576                         if (path->locks[level]) {
5577                                 btrfs_tree_unlock_rw(path->nodes[level],
5578                                                      path->locks[level]);
5579                                 path->locks[level] = 0;
5580                         }
5581                         free_extent_buffer(path->nodes[level]);
5582                         path->nodes[level] = NULL;
5583                         level++;
5584                 }
5585         }
5586         return 1;
5587 }
5588
5589 /*
5590  * drop a subvolume tree.
5591  *
5592  * this function traverses the tree freeing any blocks that only
5593  * referenced by the tree.
5594  *
5595  * when a shared tree block is found. this function decreases its
5596  * reference count by one. if update_ref is true, this function
5597  * also make sure backrefs for the shared block and all lower level
5598  * blocks are properly updated.
5599  *
5600  * If called with for_reloc == 0, may exit early with -EAGAIN
5601  */
5602 int btrfs_drop_snapshot(struct btrfs_root *root, int update_ref, int for_reloc)
5603 {
5604         const bool is_reloc_root = (root->root_key.objectid ==
5605                                     BTRFS_TREE_RELOC_OBJECTID);
5606         struct btrfs_fs_info *fs_info = root->fs_info;
5607         struct btrfs_path *path;
5608         struct btrfs_trans_handle *trans;
5609         struct btrfs_root *tree_root = fs_info->tree_root;
5610         struct btrfs_root_item *root_item = &root->root_item;
5611         struct walk_control *wc;
5612         struct btrfs_key key;
5613         int err = 0;
5614         int ret;
5615         int level;
5616         bool root_dropped = false;
5617         bool unfinished_drop = false;
5618
5619         btrfs_debug(fs_info, "Drop subvolume %llu", root->root_key.objectid);
5620
5621         path = btrfs_alloc_path();
5622         if (!path) {
5623                 err = -ENOMEM;
5624                 goto out;
5625         }
5626
5627         wc = kzalloc(sizeof(*wc), GFP_NOFS);
5628         if (!wc) {
5629                 btrfs_free_path(path);
5630                 err = -ENOMEM;
5631                 goto out;
5632         }
5633
5634         /*
5635          * Use join to avoid potential EINTR from transaction start. See
5636          * wait_reserve_ticket and the whole reservation callchain.
5637          */
5638         if (for_reloc)
5639                 trans = btrfs_join_transaction(tree_root);
5640         else
5641                 trans = btrfs_start_transaction(tree_root, 0);
5642         if (IS_ERR(trans)) {
5643                 err = PTR_ERR(trans);
5644                 goto out_free;
5645         }
5646
5647         err = btrfs_run_delayed_items(trans);
5648         if (err)
5649                 goto out_end_trans;
5650
5651         /*
5652          * This will help us catch people modifying the fs tree while we're
5653          * dropping it.  It is unsafe to mess with the fs tree while it's being
5654          * dropped as we unlock the root node and parent nodes as we walk down
5655          * the tree, assuming nothing will change.  If something does change
5656          * then we'll have stale information and drop references to blocks we've
5657          * already dropped.
5658          */
5659         set_bit(BTRFS_ROOT_DELETING, &root->state);
5660         unfinished_drop = test_bit(BTRFS_ROOT_UNFINISHED_DROP, &root->state);
5661
5662         if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
5663                 level = btrfs_header_level(root->node);
5664                 path->nodes[level] = btrfs_lock_root_node(root);
5665                 path->slots[level] = 0;
5666                 path->locks[level] = BTRFS_WRITE_LOCK;
5667                 memset(&wc->update_progress, 0,
5668                        sizeof(wc->update_progress));
5669         } else {
5670                 btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
5671                 memcpy(&wc->update_progress, &key,
5672                        sizeof(wc->update_progress));
5673
5674                 level = btrfs_root_drop_level(root_item);
5675                 BUG_ON(level == 0);
5676                 path->lowest_level = level;
5677                 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
5678                 path->lowest_level = 0;
5679                 if (ret < 0) {
5680                         err = ret;
5681                         goto out_end_trans;
5682                 }
5683                 WARN_ON(ret > 0);
5684
5685                 /*
5686                  * unlock our path, this is safe because only this
5687                  * function is allowed to delete this snapshot
5688                  */
5689                 btrfs_unlock_up_safe(path, 0);
5690
5691                 level = btrfs_header_level(root->node);
5692                 while (1) {
5693                         btrfs_tree_lock(path->nodes[level]);
5694                         path->locks[level] = BTRFS_WRITE_LOCK;
5695
5696                         ret = btrfs_lookup_extent_info(trans, fs_info,
5697                                                 path->nodes[level]->start,
5698                                                 level, 1, &wc->refs[level],
5699                                                 &wc->flags[level]);
5700                         if (ret < 0) {
5701                                 err = ret;
5702                                 goto out_end_trans;
5703                         }
5704                         BUG_ON(wc->refs[level] == 0);
5705
5706                         if (level == btrfs_root_drop_level(root_item))
5707                                 break;
5708
5709                         btrfs_tree_unlock(path->nodes[level]);
5710                         path->locks[level] = 0;
5711                         WARN_ON(wc->refs[level] != 1);
5712                         level--;
5713                 }
5714         }
5715
5716         wc->restarted = test_bit(BTRFS_ROOT_DEAD_TREE, &root->state);
5717         wc->level = level;
5718         wc->shared_level = -1;
5719         wc->stage = DROP_REFERENCE;
5720         wc->update_ref = update_ref;
5721         wc->keep_locks = 0;
5722         wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(fs_info);
5723
5724         while (1) {
5725
5726                 ret = walk_down_tree(trans, root, path, wc);
5727                 if (ret < 0) {
5728                         btrfs_abort_transaction(trans, ret);
5729                         err = ret;
5730                         break;
5731                 }
5732
5733                 ret = walk_up_tree(trans, root, path, wc, BTRFS_MAX_LEVEL);
5734                 if (ret < 0) {
5735                         btrfs_abort_transaction(trans, ret);
5736                         err = ret;
5737                         break;
5738                 }
5739
5740                 if (ret > 0) {
5741                         BUG_ON(wc->stage != DROP_REFERENCE);
5742                         break;
5743                 }
5744
5745                 if (wc->stage == DROP_REFERENCE) {
5746                         wc->drop_level = wc->level;
5747                         btrfs_node_key_to_cpu(path->nodes[wc->drop_level],
5748                                               &wc->drop_progress,
5749                                               path->slots[wc->drop_level]);
5750                 }
5751                 btrfs_cpu_key_to_disk(&root_item->drop_progress,
5752                                       &wc->drop_progress);
5753                 btrfs_set_root_drop_level(root_item, wc->drop_level);
5754
5755                 BUG_ON(wc->level == 0);
5756                 if (btrfs_should_end_transaction(trans) ||
5757                     (!for_reloc && btrfs_need_cleaner_sleep(fs_info))) {
5758                         ret = btrfs_update_root(trans, tree_root,
5759                                                 &root->root_key,
5760                                                 root_item);
5761                         if (ret) {
5762                                 btrfs_abort_transaction(trans, ret);
5763                                 err = ret;
5764                                 goto out_end_trans;
5765                         }
5766
5767                         if (!is_reloc_root)
5768                                 btrfs_set_last_root_drop_gen(fs_info, trans->transid);
5769
5770                         btrfs_end_transaction_throttle(trans);
5771                         if (!for_reloc && btrfs_need_cleaner_sleep(fs_info)) {
5772                                 btrfs_debug(fs_info,
5773                                             "drop snapshot early exit");
5774                                 err = -EAGAIN;
5775                                 goto out_free;
5776                         }
5777
5778                        /*
5779                         * Use join to avoid potential EINTR from transaction
5780                         * start. See wait_reserve_ticket and the whole
5781                         * reservation callchain.
5782                         */
5783                         if (for_reloc)
5784                                 trans = btrfs_join_transaction(tree_root);
5785                         else
5786                                 trans = btrfs_start_transaction(tree_root, 0);
5787                         if (IS_ERR(trans)) {
5788                                 err = PTR_ERR(trans);
5789                                 goto out_free;
5790                         }
5791                 }
5792         }
5793         btrfs_release_path(path);
5794         if (err)
5795                 goto out_end_trans;
5796
5797         ret = btrfs_del_root(trans, &root->root_key);
5798         if (ret) {
5799                 btrfs_abort_transaction(trans, ret);
5800                 err = ret;
5801                 goto out_end_trans;
5802         }
5803
5804         if (!is_reloc_root) {
5805                 ret = btrfs_find_root(tree_root, &root->root_key, path,
5806                                       NULL, NULL);
5807                 if (ret < 0) {
5808                         btrfs_abort_transaction(trans, ret);
5809                         err = ret;
5810                         goto out_end_trans;
5811                 } else if (ret > 0) {
5812                         /* if we fail to delete the orphan item this time
5813                          * around, it'll get picked up the next time.
5814                          *
5815                          * The most common failure here is just -ENOENT.
5816                          */
5817                         btrfs_del_orphan_item(trans, tree_root,
5818                                               root->root_key.objectid);
5819                 }
5820         }
5821
5822         /*
5823          * This subvolume is going to be completely dropped, and won't be
5824          * recorded as dirty roots, thus pertrans meta rsv will not be freed at
5825          * commit transaction time.  So free it here manually.
5826          */
5827         btrfs_qgroup_convert_reserved_meta(root, INT_MAX);
5828         btrfs_qgroup_free_meta_all_pertrans(root);
5829
5830         if (test_bit(BTRFS_ROOT_IN_RADIX, &root->state))
5831                 btrfs_add_dropped_root(trans, root);
5832         else
5833                 btrfs_put_root(root);
5834         root_dropped = true;
5835 out_end_trans:
5836         if (!is_reloc_root)
5837                 btrfs_set_last_root_drop_gen(fs_info, trans->transid);
5838
5839         btrfs_end_transaction_throttle(trans);
5840 out_free:
5841         kfree(wc);
5842         btrfs_free_path(path);
5843 out:
5844         /*
5845          * We were an unfinished drop root, check to see if there are any
5846          * pending, and if not clear and wake up any waiters.
5847          */
5848         if (!err && unfinished_drop)
5849                 btrfs_maybe_wake_unfinished_drop(fs_info);
5850
5851         /*
5852          * So if we need to stop dropping the snapshot for whatever reason we
5853          * need to make sure to add it back to the dead root list so that we
5854          * keep trying to do the work later.  This also cleans up roots if we
5855          * don't have it in the radix (like when we recover after a power fail
5856          * or unmount) so we don't leak memory.
5857          */
5858         if (!for_reloc && !root_dropped)
5859                 btrfs_add_dead_root(root);
5860         return err;
5861 }
5862
5863 /*
5864  * drop subtree rooted at tree block 'node'.
5865  *
5866  * NOTE: this function will unlock and release tree block 'node'
5867  * only used by relocation code
5868  */
5869 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
5870                         struct btrfs_root *root,
5871                         struct extent_buffer *node,
5872                         struct extent_buffer *parent)
5873 {
5874         struct btrfs_fs_info *fs_info = root->fs_info;
5875         struct btrfs_path *path;
5876         struct walk_control *wc;
5877         int level;
5878         int parent_level;
5879         int ret = 0;
5880         int wret;
5881
5882         BUG_ON(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
5883
5884         path = btrfs_alloc_path();
5885         if (!path)
5886                 return -ENOMEM;
5887
5888         wc = kzalloc(sizeof(*wc), GFP_NOFS);
5889         if (!wc) {
5890                 btrfs_free_path(path);
5891                 return -ENOMEM;
5892         }
5893
5894         btrfs_assert_tree_write_locked(parent);
5895         parent_level = btrfs_header_level(parent);
5896         atomic_inc(&parent->refs);
5897         path->nodes[parent_level] = parent;
5898         path->slots[parent_level] = btrfs_header_nritems(parent);
5899
5900         btrfs_assert_tree_write_locked(node);
5901         level = btrfs_header_level(node);
5902         path->nodes[level] = node;
5903         path->slots[level] = 0;
5904         path->locks[level] = BTRFS_WRITE_LOCK;
5905
5906         wc->refs[parent_level] = 1;
5907         wc->flags[parent_level] = BTRFS_BLOCK_FLAG_FULL_BACKREF;
5908         wc->level = level;
5909         wc->shared_level = -1;
5910         wc->stage = DROP_REFERENCE;
5911         wc->update_ref = 0;
5912         wc->keep_locks = 1;
5913         wc->reada_count = BTRFS_NODEPTRS_PER_BLOCK(fs_info);
5914
5915         while (1) {
5916                 wret = walk_down_tree(trans, root, path, wc);
5917                 if (wret < 0) {
5918                         ret = wret;
5919                         break;
5920                 }
5921
5922                 wret = walk_up_tree(trans, root, path, wc, parent_level);
5923                 if (wret < 0)
5924                         ret = wret;
5925                 if (wret != 0)
5926                         break;
5927         }
5928
5929         kfree(wc);
5930         btrfs_free_path(path);
5931         return ret;
5932 }
5933
5934 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
5935                                    u64 start, u64 end)
5936 {
5937         return unpin_extent_range(fs_info, start, end, false);
5938 }
5939
5940 /*
5941  * It used to be that old block groups would be left around forever.
5942  * Iterating over them would be enough to trim unused space.  Since we
5943  * now automatically remove them, we also need to iterate over unallocated
5944  * space.
5945  *
5946  * We don't want a transaction for this since the discard may take a
5947  * substantial amount of time.  We don't require that a transaction be
5948  * running, but we do need to take a running transaction into account
5949  * to ensure that we're not discarding chunks that were released or
5950  * allocated in the current transaction.
5951  *
5952  * Holding the chunks lock will prevent other threads from allocating
5953  * or releasing chunks, but it won't prevent a running transaction
5954  * from committing and releasing the memory that the pending chunks
5955  * list head uses.  For that, we need to take a reference to the
5956  * transaction and hold the commit root sem.  We only need to hold
5957  * it while performing the free space search since we have already
5958  * held back allocations.
5959  */
5960 static int btrfs_trim_free_extents(struct btrfs_device *device, u64 *trimmed)
5961 {
5962         u64 start = BTRFS_DEVICE_RANGE_RESERVED, len = 0, end = 0;
5963         int ret;
5964
5965         *trimmed = 0;
5966
5967         /* Discard not supported = nothing to do. */
5968         if (!bdev_max_discard_sectors(device->bdev))
5969                 return 0;
5970
5971         /* Not writable = nothing to do. */
5972         if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state))
5973                 return 0;
5974
5975         /* No free space = nothing to do. */
5976         if (device->total_bytes <= device->bytes_used)
5977                 return 0;
5978
5979         ret = 0;
5980
5981         while (1) {
5982                 struct btrfs_fs_info *fs_info = device->fs_info;
5983                 u64 bytes;
5984
5985                 ret = mutex_lock_interruptible(&fs_info->chunk_mutex);
5986                 if (ret)
5987                         break;
5988
5989                 find_first_clear_extent_bit(&device->alloc_state, start,
5990                                             &start, &end,
5991                                             CHUNK_TRIMMED | CHUNK_ALLOCATED);
5992
5993                 /* Check if there are any CHUNK_* bits left */
5994                 if (start > device->total_bytes) {
5995                         WARN_ON(IS_ENABLED(CONFIG_BTRFS_DEBUG));
5996                         btrfs_warn_in_rcu(fs_info,
5997 "ignoring attempt to trim beyond device size: offset %llu length %llu device %s device size %llu",
5998                                           start, end - start + 1,
5999                                           btrfs_dev_name(device),
6000                                           device->total_bytes);
6001                         mutex_unlock(&fs_info->chunk_mutex);
6002                         ret = 0;
6003                         break;
6004                 }
6005
6006                 /* Ensure we skip the reserved space on each device. */
6007                 start = max_t(u64, start, BTRFS_DEVICE_RANGE_RESERVED);
6008
6009                 /*
6010                  * If find_first_clear_extent_bit find a range that spans the
6011                  * end of the device it will set end to -1, in this case it's up
6012                  * to the caller to trim the value to the size of the device.
6013                  */
6014                 end = min(end, device->total_bytes - 1);
6015
6016                 len = end - start + 1;
6017
6018                 /* We didn't find any extents */
6019                 if (!len) {
6020                         mutex_unlock(&fs_info->chunk_mutex);
6021                         ret = 0;
6022                         break;
6023                 }
6024
6025                 ret = btrfs_issue_discard(device->bdev, start, len,
6026                                           &bytes);
6027                 if (!ret)
6028                         set_extent_bit(&device->alloc_state, start,
6029                                        start + bytes - 1, CHUNK_TRIMMED, NULL);
6030                 mutex_unlock(&fs_info->chunk_mutex);
6031
6032                 if (ret)
6033                         break;
6034
6035                 start += len;
6036                 *trimmed += bytes;
6037
6038                 if (fatal_signal_pending(current)) {
6039                         ret = -ERESTARTSYS;
6040                         break;
6041                 }
6042
6043                 cond_resched();
6044         }
6045
6046         return ret;
6047 }
6048
6049 /*
6050  * Trim the whole filesystem by:
6051  * 1) trimming the free space in each block group
6052  * 2) trimming the unallocated space on each device
6053  *
6054  * This will also continue trimming even if a block group or device encounters
6055  * an error.  The return value will be the last error, or 0 if nothing bad
6056  * happens.
6057  */
6058 int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
6059 {
6060         struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
6061         struct btrfs_block_group *cache = NULL;
6062         struct btrfs_device *device;
6063         u64 group_trimmed;
6064         u64 range_end = U64_MAX;
6065         u64 start;
6066         u64 end;
6067         u64 trimmed = 0;
6068         u64 bg_failed = 0;
6069         u64 dev_failed = 0;
6070         int bg_ret = 0;
6071         int dev_ret = 0;
6072         int ret = 0;
6073
6074         if (range->start == U64_MAX)
6075                 return -EINVAL;
6076
6077         /*
6078          * Check range overflow if range->len is set.
6079          * The default range->len is U64_MAX.
6080          */
6081         if (range->len != U64_MAX &&
6082             check_add_overflow(range->start, range->len, &range_end))
6083                 return -EINVAL;
6084
6085         cache = btrfs_lookup_first_block_group(fs_info, range->start);
6086         for (; cache; cache = btrfs_next_block_group(cache)) {
6087                 if (cache->start >= range_end) {
6088                         btrfs_put_block_group(cache);
6089                         break;
6090                 }
6091
6092                 start = max(range->start, cache->start);
6093                 end = min(range_end, cache->start + cache->length);
6094
6095                 if (end - start >= range->minlen) {
6096                         if (!btrfs_block_group_done(cache)) {
6097                                 ret = btrfs_cache_block_group(cache, true);
6098                                 if (ret) {
6099                                         bg_failed++;
6100                                         bg_ret = ret;
6101                                         continue;
6102                                 }
6103                         }
6104                         ret = btrfs_trim_block_group(cache,
6105                                                      &group_trimmed,
6106                                                      start,
6107                                                      end,
6108                                                      range->minlen);
6109
6110                         trimmed += group_trimmed;
6111                         if (ret) {
6112                                 bg_failed++;
6113                                 bg_ret = ret;
6114                                 continue;
6115                         }
6116                 }
6117         }
6118
6119         if (bg_failed)
6120                 btrfs_warn(fs_info,
6121                         "failed to trim %llu block group(s), last error %d",
6122                         bg_failed, bg_ret);
6123
6124         mutex_lock(&fs_devices->device_list_mutex);
6125         list_for_each_entry(device, &fs_devices->devices, dev_list) {
6126                 if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state))
6127                         continue;
6128
6129                 ret = btrfs_trim_free_extents(device, &group_trimmed);
6130                 if (ret) {
6131                         dev_failed++;
6132                         dev_ret = ret;
6133                         break;
6134                 }
6135
6136                 trimmed += group_trimmed;
6137         }
6138         mutex_unlock(&fs_devices->device_list_mutex);
6139
6140         if (dev_failed)
6141                 btrfs_warn(fs_info,
6142                         "failed to trim %llu device(s), last error %d",
6143                         dev_failed, dev_ret);
6144         range->len = trimmed;
6145         if (bg_ret)
6146                 return bg_ret;
6147         return dev_ret;
6148 }