linux/fs/btrfs/relocation.c
<<
>>
Prefs
   1// SPDX-License-Identifier: GPL-2.0
   2/*
   3 * Copyright (C) 2009 Oracle.  All rights reserved.
   4 */
   5
   6#include <linux/sched.h>
   7#include <linux/pagemap.h>
   8#include <linux/writeback.h>
   9#include <linux/blkdev.h>
  10#include <linux/rbtree.h>
  11#include <linux/slab.h>
  12#include <linux/error-injection.h>
  13#include "ctree.h"
  14#include "disk-io.h"
  15#include "transaction.h"
  16#include "volumes.h"
  17#include "locking.h"
  18#include "btrfs_inode.h"
  19#include "async-thread.h"
  20#include "free-space-cache.h"
  21#include "qgroup.h"
  22#include "print-tree.h"
  23#include "delalloc-space.h"
  24#include "block-group.h"
  25#include "backref.h"
  26#include "misc.h"
  27
  28/*
  29 * Relocation overview
  30 *
  31 * [What does relocation do]
  32 *
  33 * The objective of relocation is to relocate all extents of the target block
  34 * group to other block groups.
  35 * This is utilized by resize (shrink only), profile converting, compacting
  36 * space, or balance routine to spread chunks over devices.
  37 *
  38 *              Before          |               After
  39 * ------------------------------------------------------------------
  40 *  BG A: 10 data extents       | BG A: deleted
  41 *  BG B:  2 data extents       | BG B: 10 data extents (2 old + 8 relocated)
  42 *  BG C:  1 extents            | BG C:  3 data extents (1 old + 2 relocated)
  43 *
  44 * [How does relocation work]
  45 *
  46 * 1.   Mark the target block group read-only
  47 *      New extents won't be allocated from the target block group.
  48 *
  49 * 2.1  Record each extent in the target block group
  50 *      To build a proper map of extents to be relocated.
  51 *
  52 * 2.2  Build data reloc tree and reloc trees
  53 *      Data reloc tree will contain an inode, recording all newly relocated
  54 *      data extents.
  55 *      There will be only one data reloc tree for one data block group.
  56 *
  57 *      Reloc tree will be a special snapshot of its source tree, containing
  58 *      relocated tree blocks.
  59 *      Each tree referring to a tree block in target block group will get its
  60 *      reloc tree built.
  61 *
  62 * 2.3  Swap source tree with its corresponding reloc tree
  63 *      Each involved tree only refers to new extents after swap.
  64 *
  65 * 3.   Cleanup reloc trees and data reloc tree.
  66 *      As old extents in the target block group are still referenced by reloc
  67 *      trees, we need to clean them up before really freeing the target block
  68 *      group.
  69 *
  70 * The main complexity is in steps 2.2 and 2.3.
  71 *
  72 * The entry point of relocation is relocate_block_group() function.
  73 */
  74
  75#define RELOCATION_RESERVED_NODES       256
  76/*
  77 * map address of tree root to tree
  78 */
  79struct mapping_node {
  80        struct {
  81                struct rb_node rb_node;
  82                u64 bytenr;
  83        }; /* Use rb_simle_node for search/insert */
  84        void *data;
  85};
  86
  87struct mapping_tree {
  88        struct rb_root rb_root;
  89        spinlock_t lock;
  90};
  91
  92/*
  93 * present a tree block to process
  94 */
  95struct tree_block {
  96        struct {
  97                struct rb_node rb_node;
  98                u64 bytenr;
  99        }; /* Use rb_simple_node for search/insert */
 100        u64 owner;
 101        struct btrfs_key key;
 102        unsigned int level:8;
 103        unsigned int key_ready:1;
 104};
 105
 106#define MAX_EXTENTS 128
 107
 108struct file_extent_cluster {
 109        u64 start;
 110        u64 end;
 111        u64 boundary[MAX_EXTENTS];
 112        unsigned int nr;
 113};
 114
 115struct reloc_control {
 116        /* block group to relocate */
 117        struct btrfs_block_group *block_group;
 118        /* extent tree */
 119        struct btrfs_root *extent_root;
 120        /* inode for moving data */
 121        struct inode *data_inode;
 122
 123        struct btrfs_block_rsv *block_rsv;
 124
 125        struct btrfs_backref_cache backref_cache;
 126
 127        struct file_extent_cluster cluster;
 128        /* tree blocks have been processed */
 129        struct extent_io_tree processed_blocks;
 130        /* map start of tree root to corresponding reloc tree */
 131        struct mapping_tree reloc_root_tree;
 132        /* list of reloc trees */
 133        struct list_head reloc_roots;
 134        /* list of subvolume trees that get relocated */
 135        struct list_head dirty_subvol_roots;
 136        /* size of metadata reservation for merging reloc trees */
 137        u64 merging_rsv_size;
 138        /* size of relocated tree nodes */
 139        u64 nodes_relocated;
 140        /* reserved size for block group relocation*/
 141        u64 reserved_bytes;
 142
 143        u64 search_start;
 144        u64 extents_found;
 145
 146        unsigned int stage:8;
 147        unsigned int create_reloc_tree:1;
 148        unsigned int merge_reloc_tree:1;
 149        unsigned int found_file_extent:1;
 150};
 151
 152/* stages of data relocation */
 153#define MOVE_DATA_EXTENTS       0
 154#define UPDATE_DATA_PTRS        1
 155
 156static void mark_block_processed(struct reloc_control *rc,
 157                                 struct btrfs_backref_node *node)
 158{
 159        u32 blocksize;
 160
 161        if (node->level == 0 ||
 162            in_range(node->bytenr, rc->block_group->start,
 163                     rc->block_group->length)) {
 164                blocksize = rc->extent_root->fs_info->nodesize;
 165                set_extent_bits(&rc->processed_blocks, node->bytenr,
 166                                node->bytenr + blocksize - 1, EXTENT_DIRTY);
 167        }
 168        node->processed = 1;
 169}
 170
 171
 172static void mapping_tree_init(struct mapping_tree *tree)
 173{
 174        tree->rb_root = RB_ROOT;
 175        spin_lock_init(&tree->lock);
 176}
 177
 178/*
 179 * walk up backref nodes until reach node presents tree root
 180 */
 181static struct btrfs_backref_node *walk_up_backref(
 182                struct btrfs_backref_node *node,
 183                struct btrfs_backref_edge *edges[], int *index)
 184{
 185        struct btrfs_backref_edge *edge;
 186        int idx = *index;
 187
 188        while (!list_empty(&node->upper)) {
 189                edge = list_entry(node->upper.next,
 190                                  struct btrfs_backref_edge, list[LOWER]);
 191                edges[idx++] = edge;
 192                node = edge->node[UPPER];
 193        }
 194        BUG_ON(node->detached);
 195        *index = idx;
 196        return node;
 197}
 198
 199/*
 200 * walk down backref nodes to find start of next reference path
 201 */
 202static struct btrfs_backref_node *walk_down_backref(
 203                struct btrfs_backref_edge *edges[], int *index)
 204{
 205        struct btrfs_backref_edge *edge;
 206        struct btrfs_backref_node *lower;
 207        int idx = *index;
 208
 209        while (idx > 0) {
 210                edge = edges[idx - 1];
 211                lower = edge->node[LOWER];
 212                if (list_is_last(&edge->list[LOWER], &lower->upper)) {
 213                        idx--;
 214                        continue;
 215                }
 216                edge = list_entry(edge->list[LOWER].next,
 217                                  struct btrfs_backref_edge, list[LOWER]);
 218                edges[idx - 1] = edge;
 219                *index = idx;
 220                return edge->node[UPPER];
 221        }
 222        *index = 0;
 223        return NULL;
 224}
 225
 226static void update_backref_node(struct btrfs_backref_cache *cache,
 227                                struct btrfs_backref_node *node, u64 bytenr)
 228{
 229        struct rb_node *rb_node;
 230        rb_erase(&node->rb_node, &cache->rb_root);
 231        node->bytenr = bytenr;
 232        rb_node = rb_simple_insert(&cache->rb_root, node->bytenr, &node->rb_node);
 233        if (rb_node)
 234                btrfs_backref_panic(cache->fs_info, bytenr, -EEXIST);
 235}
 236
 237/*
 238 * update backref cache after a transaction commit
 239 */
 240static int update_backref_cache(struct btrfs_trans_handle *trans,
 241                                struct btrfs_backref_cache *cache)
 242{
 243        struct btrfs_backref_node *node;
 244        int level = 0;
 245
 246        if (cache->last_trans == 0) {
 247                cache->last_trans = trans->transid;
 248                return 0;
 249        }
 250
 251        if (cache->last_trans == trans->transid)
 252                return 0;
 253
 254        /*
 255         * detached nodes are used to avoid unnecessary backref
 256         * lookup. transaction commit changes the extent tree.
 257         * so the detached nodes are no longer useful.
 258         */
 259        while (!list_empty(&cache->detached)) {
 260                node = list_entry(cache->detached.next,
 261                                  struct btrfs_backref_node, list);
 262                btrfs_backref_cleanup_node(cache, node);
 263        }
 264
 265        while (!list_empty(&cache->changed)) {
 266                node = list_entry(cache->changed.next,
 267                                  struct btrfs_backref_node, list);
 268                list_del_init(&node->list);
 269                BUG_ON(node->pending);
 270                update_backref_node(cache, node, node->new_bytenr);
 271        }
 272
 273        /*
 274         * some nodes can be left in the pending list if there were
 275         * errors during processing the pending nodes.
 276         */
 277        for (level = 0; level < BTRFS_MAX_LEVEL; level++) {
 278                list_for_each_entry(node, &cache->pending[level], list) {
 279                        BUG_ON(!node->pending);
 280                        if (node->bytenr == node->new_bytenr)
 281                                continue;
 282                        update_backref_node(cache, node, node->new_bytenr);
 283                }
 284        }
 285
 286        cache->last_trans = 0;
 287        return 1;
 288}
 289
 290static bool reloc_root_is_dead(struct btrfs_root *root)
 291{
 292        /*
 293         * Pair with set_bit/clear_bit in clean_dirty_subvols and
 294         * btrfs_update_reloc_root. We need to see the updated bit before
 295         * trying to access reloc_root
 296         */
 297        smp_rmb();
 298        if (test_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state))
 299                return true;
 300        return false;
 301}
 302
 303/*
 304 * Check if this subvolume tree has valid reloc tree.
 305 *
 306 * Reloc tree after swap is considered dead, thus not considered as valid.
 307 * This is enough for most callers, as they don't distinguish dead reloc root
 308 * from no reloc root.  But btrfs_should_ignore_reloc_root() below is a
 309 * special case.
 310 */
 311static bool have_reloc_root(struct btrfs_root *root)
 312{
 313        if (reloc_root_is_dead(root))
 314                return false;
 315        if (!root->reloc_root)
 316                return false;
 317        return true;
 318}
 319
 320int btrfs_should_ignore_reloc_root(struct btrfs_root *root)
 321{
 322        struct btrfs_root *reloc_root;
 323
 324        if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
 325                return 0;
 326
 327        /* This root has been merged with its reloc tree, we can ignore it */
 328        if (reloc_root_is_dead(root))
 329                return 1;
 330
 331        reloc_root = root->reloc_root;
 332        if (!reloc_root)
 333                return 0;
 334
 335        if (btrfs_header_generation(reloc_root->commit_root) ==
 336            root->fs_info->running_transaction->transid)
 337                return 0;
 338        /*
 339         * if there is reloc tree and it was created in previous
 340         * transaction backref lookup can find the reloc tree,
 341         * so backref node for the fs tree root is useless for
 342         * relocation.
 343         */
 344        return 1;
 345}
 346
 347/*
 348 * find reloc tree by address of tree root
 349 */
 350struct btrfs_root *find_reloc_root(struct btrfs_fs_info *fs_info, u64 bytenr)
 351{
 352        struct reloc_control *rc = fs_info->reloc_ctl;
 353        struct rb_node *rb_node;
 354        struct mapping_node *node;
 355        struct btrfs_root *root = NULL;
 356
 357        ASSERT(rc);
 358        spin_lock(&rc->reloc_root_tree.lock);
 359        rb_node = rb_simple_search(&rc->reloc_root_tree.rb_root, bytenr);
 360        if (rb_node) {
 361                node = rb_entry(rb_node, struct mapping_node, rb_node);
 362                root = (struct btrfs_root *)node->data;
 363        }
 364        spin_unlock(&rc->reloc_root_tree.lock);
 365        return btrfs_grab_root(root);
 366}
 367
 368/*
 369 * For useless nodes, do two major clean ups:
 370 *
 371 * - Cleanup the children edges and nodes
 372 *   If child node is also orphan (no parent) during cleanup, then the child
 373 *   node will also be cleaned up.
 374 *
 375 * - Freeing up leaves (level 0), keeps nodes detached
 376 *   For nodes, the node is still cached as "detached"
 377 *
 378 * Return false if @node is not in the @useless_nodes list.
 379 * Return true if @node is in the @useless_nodes list.
 380 */
 381static bool handle_useless_nodes(struct reloc_control *rc,
 382                                 struct btrfs_backref_node *node)
 383{
 384        struct btrfs_backref_cache *cache = &rc->backref_cache;
 385        struct list_head *useless_node = &cache->useless_node;
 386        bool ret = false;
 387
 388        while (!list_empty(useless_node)) {
 389                struct btrfs_backref_node *cur;
 390
 391                cur = list_first_entry(useless_node, struct btrfs_backref_node,
 392                                 list);
 393                list_del_init(&cur->list);
 394
 395                /* Only tree root nodes can be added to @useless_nodes */
 396                ASSERT(list_empty(&cur->upper));
 397
 398                if (cur == node)
 399                        ret = true;
 400
 401                /* The node is the lowest node */
 402                if (cur->lowest) {
 403                        list_del_init(&cur->lower);
 404                        cur->lowest = 0;
 405                }
 406
 407                /* Cleanup the lower edges */
 408                while (!list_empty(&cur->lower)) {
 409                        struct btrfs_backref_edge *edge;
 410                        struct btrfs_backref_node *lower;
 411
 412                        edge = list_entry(cur->lower.next,
 413                                        struct btrfs_backref_edge, list[UPPER]);
 414                        list_del(&edge->list[UPPER]);
 415                        list_del(&edge->list[LOWER]);
 416                        lower = edge->node[LOWER];
 417                        btrfs_backref_free_edge(cache, edge);
 418
 419                        /* Child node is also orphan, queue for cleanup */
 420                        if (list_empty(&lower->upper))
 421                                list_add(&lower->list, useless_node);
 422                }
 423                /* Mark this block processed for relocation */
 424                mark_block_processed(rc, cur);
 425
 426                /*
 427                 * Backref nodes for tree leaves are deleted from the cache.
 428                 * Backref nodes for upper level tree blocks are left in the
 429                 * cache to avoid unnecessary backref lookup.
 430                 */
 431                if (cur->level > 0) {
 432                        list_add(&cur->list, &cache->detached);
 433                        cur->detached = 1;
 434                } else {
 435                        rb_erase(&cur->rb_node, &cache->rb_root);
 436                        btrfs_backref_free_node(cache, cur);
 437                }
 438        }
 439        return ret;
 440}
 441
 442/*
 443 * Build backref tree for a given tree block. Root of the backref tree
 444 * corresponds the tree block, leaves of the backref tree correspond roots of
 445 * b-trees that reference the tree block.
 446 *
 447 * The basic idea of this function is check backrefs of a given block to find
 448 * upper level blocks that reference the block, and then check backrefs of
 449 * these upper level blocks recursively. The recursion stops when tree root is
 450 * reached or backrefs for the block is cached.
 451 *
 452 * NOTE: if we find that backrefs for a block are cached, we know backrefs for
 453 * all upper level blocks that directly/indirectly reference the block are also
 454 * cached.
 455 */
 456static noinline_for_stack struct btrfs_backref_node *build_backref_tree(
 457                        struct reloc_control *rc, struct btrfs_key *node_key,
 458                        int level, u64 bytenr)
 459{
 460        struct btrfs_backref_iter *iter;
 461        struct btrfs_backref_cache *cache = &rc->backref_cache;
 462        /* For searching parent of TREE_BLOCK_REF */
 463        struct btrfs_path *path;
 464        struct btrfs_backref_node *cur;
 465        struct btrfs_backref_node *node = NULL;
 466        struct btrfs_backref_edge *edge;
 467        int ret;
 468        int err = 0;
 469
 470        iter = btrfs_backref_iter_alloc(rc->extent_root->fs_info, GFP_NOFS);
 471        if (!iter)
 472                return ERR_PTR(-ENOMEM);
 473        path = btrfs_alloc_path();
 474        if (!path) {
 475                err = -ENOMEM;
 476                goto out;
 477        }
 478
 479        node = btrfs_backref_alloc_node(cache, bytenr, level);
 480        if (!node) {
 481                err = -ENOMEM;
 482                goto out;
 483        }
 484
 485        node->lowest = 1;
 486        cur = node;
 487
 488        /* Breadth-first search to build backref cache */
 489        do {
 490                ret = btrfs_backref_add_tree_node(cache, path, iter, node_key,
 491                                                  cur);
 492                if (ret < 0) {
 493                        err = ret;
 494                        goto out;
 495                }
 496                edge = list_first_entry_or_null(&cache->pending_edge,
 497                                struct btrfs_backref_edge, list[UPPER]);
 498                /*
 499                 * The pending list isn't empty, take the first block to
 500                 * process
 501                 */
 502                if (edge) {
 503                        list_del_init(&edge->list[UPPER]);
 504                        cur = edge->node[UPPER];
 505                }
 506        } while (edge);
 507
 508        /* Finish the upper linkage of newly added edges/nodes */
 509        ret = btrfs_backref_finish_upper_links(cache, node);
 510        if (ret < 0) {
 511                err = ret;
 512                goto out;
 513        }
 514
 515        if (handle_useless_nodes(rc, node))
 516                node = NULL;
 517out:
 518        btrfs_backref_iter_free(iter);
 519        btrfs_free_path(path);
 520        if (err) {
 521                btrfs_backref_error_cleanup(cache, node);
 522                return ERR_PTR(err);
 523        }
 524        ASSERT(!node || !node->detached);
 525        ASSERT(list_empty(&cache->useless_node) &&
 526               list_empty(&cache->pending_edge));
 527        return node;
 528}
 529
 530/*
 531 * helper to add backref node for the newly created snapshot.
 532 * the backref node is created by cloning backref node that
 533 * corresponds to root of source tree
 534 */
 535static int clone_backref_node(struct btrfs_trans_handle *trans,
 536                              struct reloc_control *rc,
 537                              struct btrfs_root *src,
 538                              struct btrfs_root *dest)
 539{
 540        struct btrfs_root *reloc_root = src->reloc_root;
 541        struct btrfs_backref_cache *cache = &rc->backref_cache;
 542        struct btrfs_backref_node *node = NULL;
 543        struct btrfs_backref_node *new_node;
 544        struct btrfs_backref_edge *edge;
 545        struct btrfs_backref_edge *new_edge;
 546        struct rb_node *rb_node;
 547
 548        if (cache->last_trans > 0)
 549                update_backref_cache(trans, cache);
 550
 551        rb_node = rb_simple_search(&cache->rb_root, src->commit_root->start);
 552        if (rb_node) {
 553                node = rb_entry(rb_node, struct btrfs_backref_node, rb_node);
 554                if (node->detached)
 555                        node = NULL;
 556                else
 557                        BUG_ON(node->new_bytenr != reloc_root->node->start);
 558        }
 559
 560        if (!node) {
 561                rb_node = rb_simple_search(&cache->rb_root,
 562                                           reloc_root->commit_root->start);
 563                if (rb_node) {
 564                        node = rb_entry(rb_node, struct btrfs_backref_node,
 565                                        rb_node);
 566                        BUG_ON(node->detached);
 567                }
 568        }
 569
 570        if (!node)
 571                return 0;
 572
 573        new_node = btrfs_backref_alloc_node(cache, dest->node->start,
 574                                            node->level);
 575        if (!new_node)
 576                return -ENOMEM;
 577
 578        new_node->lowest = node->lowest;
 579        new_node->checked = 1;
 580        new_node->root = btrfs_grab_root(dest);
 581        ASSERT(new_node->root);
 582
 583        if (!node->lowest) {
 584                list_for_each_entry(edge, &node->lower, list[UPPER]) {
 585                        new_edge = btrfs_backref_alloc_edge(cache);
 586                        if (!new_edge)
 587                                goto fail;
 588
 589                        btrfs_backref_link_edge(new_edge, edge->node[LOWER],
 590                                                new_node, LINK_UPPER);
 591                }
 592        } else {
 593                list_add_tail(&new_node->lower, &cache->leaves);
 594        }
 595
 596        rb_node = rb_simple_insert(&cache->rb_root, new_node->bytenr,
 597                                   &new_node->rb_node);
 598        if (rb_node)
 599                btrfs_backref_panic(trans->fs_info, new_node->bytenr, -EEXIST);
 600
 601        if (!new_node->lowest) {
 602                list_for_each_entry(new_edge, &new_node->lower, list[UPPER]) {
 603                        list_add_tail(&new_edge->list[LOWER],
 604                                      &new_edge->node[LOWER]->upper);
 605                }
 606        }
 607        return 0;
 608fail:
 609        while (!list_empty(&new_node->lower)) {
 610                new_edge = list_entry(new_node->lower.next,
 611                                      struct btrfs_backref_edge, list[UPPER]);
 612                list_del(&new_edge->list[UPPER]);
 613                btrfs_backref_free_edge(cache, new_edge);
 614        }
 615        btrfs_backref_free_node(cache, new_node);
 616        return -ENOMEM;
 617}
 618
 619/*
 620 * helper to add 'address of tree root -> reloc tree' mapping
 621 */
 622static int __must_check __add_reloc_root(struct btrfs_root *root)
 623{
 624        struct btrfs_fs_info *fs_info = root->fs_info;
 625        struct rb_node *rb_node;
 626        struct mapping_node *node;
 627        struct reloc_control *rc = fs_info->reloc_ctl;
 628
 629        node = kmalloc(sizeof(*node), GFP_NOFS);
 630        if (!node)
 631                return -ENOMEM;
 632
 633        node->bytenr = root->commit_root->start;
 634        node->data = root;
 635
 636        spin_lock(&rc->reloc_root_tree.lock);
 637        rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root,
 638                                   node->bytenr, &node->rb_node);
 639        spin_unlock(&rc->reloc_root_tree.lock);
 640        if (rb_node) {
 641                btrfs_err(fs_info,
 642                            "Duplicate root found for start=%llu while inserting into relocation tree",
 643                            node->bytenr);
 644                return -EEXIST;
 645        }
 646
 647        list_add_tail(&root->root_list, &rc->reloc_roots);
 648        return 0;
 649}
 650
 651/*
 652 * helper to delete the 'address of tree root -> reloc tree'
 653 * mapping
 654 */
 655static void __del_reloc_root(struct btrfs_root *root)
 656{
 657        struct btrfs_fs_info *fs_info = root->fs_info;
 658        struct rb_node *rb_node;
 659        struct mapping_node *node = NULL;
 660        struct reloc_control *rc = fs_info->reloc_ctl;
 661        bool put_ref = false;
 662
 663        if (rc && root->node) {
 664                spin_lock(&rc->reloc_root_tree.lock);
 665                rb_node = rb_simple_search(&rc->reloc_root_tree.rb_root,
 666                                           root->commit_root->start);
 667                if (rb_node) {
 668                        node = rb_entry(rb_node, struct mapping_node, rb_node);
 669                        rb_erase(&node->rb_node, &rc->reloc_root_tree.rb_root);
 670                        RB_CLEAR_NODE(&node->rb_node);
 671                }
 672                spin_unlock(&rc->reloc_root_tree.lock);
 673                ASSERT(!node || (struct btrfs_root *)node->data == root);
 674        }
 675
 676        /*
 677         * We only put the reloc root here if it's on the list.  There's a lot
 678         * of places where the pattern is to splice the rc->reloc_roots, process
 679         * the reloc roots, and then add the reloc root back onto
 680         * rc->reloc_roots.  If we call __del_reloc_root while it's off of the
 681         * list we don't want the reference being dropped, because the guy
 682         * messing with the list is in charge of the reference.
 683         */
 684        spin_lock(&fs_info->trans_lock);
 685        if (!list_empty(&root->root_list)) {
 686                put_ref = true;
 687                list_del_init(&root->root_list);
 688        }
 689        spin_unlock(&fs_info->trans_lock);
 690        if (put_ref)
 691                btrfs_put_root(root);
 692        kfree(node);
 693}
 694
 695/*
 696 * helper to update the 'address of tree root -> reloc tree'
 697 * mapping
 698 */
 699static int __update_reloc_root(struct btrfs_root *root)
 700{
 701        struct btrfs_fs_info *fs_info = root->fs_info;
 702        struct rb_node *rb_node;
 703        struct mapping_node *node = NULL;
 704        struct reloc_control *rc = fs_info->reloc_ctl;
 705
 706        spin_lock(&rc->reloc_root_tree.lock);
 707        rb_node = rb_simple_search(&rc->reloc_root_tree.rb_root,
 708                                   root->commit_root->start);
 709        if (rb_node) {
 710                node = rb_entry(rb_node, struct mapping_node, rb_node);
 711                rb_erase(&node->rb_node, &rc->reloc_root_tree.rb_root);
 712        }
 713        spin_unlock(&rc->reloc_root_tree.lock);
 714
 715        if (!node)
 716                return 0;
 717        BUG_ON((struct btrfs_root *)node->data != root);
 718
 719        spin_lock(&rc->reloc_root_tree.lock);
 720        node->bytenr = root->node->start;
 721        rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root,
 722                                   node->bytenr, &node->rb_node);
 723        spin_unlock(&rc->reloc_root_tree.lock);
 724        if (rb_node)
 725                btrfs_backref_panic(fs_info, node->bytenr, -EEXIST);
 726        return 0;
 727}
 728
 729static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans,
 730                                        struct btrfs_root *root, u64 objectid)
 731{
 732        struct btrfs_fs_info *fs_info = root->fs_info;
 733        struct btrfs_root *reloc_root;
 734        struct extent_buffer *eb;
 735        struct btrfs_root_item *root_item;
 736        struct btrfs_key root_key;
 737        int ret = 0;
 738        bool must_abort = false;
 739
 740        root_item = kmalloc(sizeof(*root_item), GFP_NOFS);
 741        if (!root_item)
 742                return ERR_PTR(-ENOMEM);
 743
 744        root_key.objectid = BTRFS_TREE_RELOC_OBJECTID;
 745        root_key.type = BTRFS_ROOT_ITEM_KEY;
 746        root_key.offset = objectid;
 747
 748        if (root->root_key.objectid == objectid) {
 749                u64 commit_root_gen;
 750
 751                /* called by btrfs_init_reloc_root */
 752                ret = btrfs_copy_root(trans, root, root->commit_root, &eb,
 753                                      BTRFS_TREE_RELOC_OBJECTID);
 754                if (ret)
 755                        goto fail;
 756
 757                /*
 758                 * Set the last_snapshot field to the generation of the commit
 759                 * root - like this ctree.c:btrfs_block_can_be_shared() behaves
 760                 * correctly (returns true) when the relocation root is created
 761                 * either inside the critical section of a transaction commit
 762                 * (through transaction.c:qgroup_account_snapshot()) and when
 763                 * it's created before the transaction commit is started.
 764                 */
 765                commit_root_gen = btrfs_header_generation(root->commit_root);
 766                btrfs_set_root_last_snapshot(&root->root_item, commit_root_gen);
 767        } else {
 768                /*
 769                 * called by btrfs_reloc_post_snapshot_hook.
 770                 * the source tree is a reloc tree, all tree blocks
 771                 * modified after it was created have RELOC flag
 772                 * set in their headers. so it's OK to not update
 773                 * the 'last_snapshot'.
 774                 */
 775                ret = btrfs_copy_root(trans, root, root->node, &eb,
 776                                      BTRFS_TREE_RELOC_OBJECTID);
 777                if (ret)
 778                        goto fail;
 779        }
 780
 781        /*
 782         * We have changed references at this point, we must abort the
 783         * transaction if anything fails.
 784         */
 785        must_abort = true;
 786
 787        memcpy(root_item, &root->root_item, sizeof(*root_item));
 788        btrfs_set_root_bytenr(root_item, eb->start);
 789        btrfs_set_root_level(root_item, btrfs_header_level(eb));
 790        btrfs_set_root_generation(root_item, trans->transid);
 791
 792        if (root->root_key.objectid == objectid) {
 793                btrfs_set_root_refs(root_item, 0);
 794                memset(&root_item->drop_progress, 0,
 795                       sizeof(struct btrfs_disk_key));
 796                btrfs_set_root_drop_level(root_item, 0);
 797        }
 798
 799        btrfs_tree_unlock(eb);
 800        free_extent_buffer(eb);
 801
 802        ret = btrfs_insert_root(trans, fs_info->tree_root,
 803                                &root_key, root_item);
 804        if (ret)
 805                goto fail;
 806
 807        kfree(root_item);
 808
 809        reloc_root = btrfs_read_tree_root(fs_info->tree_root, &root_key);
 810        if (IS_ERR(reloc_root)) {
 811                ret = PTR_ERR(reloc_root);
 812                goto abort;
 813        }
 814        set_bit(BTRFS_ROOT_SHAREABLE, &reloc_root->state);
 815        reloc_root->last_trans = trans->transid;
 816        return reloc_root;
 817fail:
 818        kfree(root_item);
 819abort:
 820        if (must_abort)
 821                btrfs_abort_transaction(trans, ret);
 822        return ERR_PTR(ret);
 823}
 824
 825/*
 826 * create reloc tree for a given fs tree. reloc tree is just a
 827 * snapshot of the fs tree with special root objectid.
 828 *
 829 * The reloc_root comes out of here with two references, one for
 830 * root->reloc_root, and another for being on the rc->reloc_roots list.
 831 */
 832int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 833                          struct btrfs_root *root)
 834{
 835        struct btrfs_fs_info *fs_info = root->fs_info;
 836        struct btrfs_root *reloc_root;
 837        struct reloc_control *rc = fs_info->reloc_ctl;
 838        struct btrfs_block_rsv *rsv;
 839        int clear_rsv = 0;
 840        int ret;
 841
 842        if (!rc)
 843                return 0;
 844
 845        /*
 846         * The subvolume has reloc tree but the swap is finished, no need to
 847         * create/update the dead reloc tree
 848         */
 849        if (reloc_root_is_dead(root))
 850                return 0;
 851
 852        /*
 853         * This is subtle but important.  We do not do
 854         * record_root_in_transaction for reloc roots, instead we record their
 855         * corresponding fs root, and then here we update the last trans for the
 856         * reloc root.  This means that we have to do this for the entire life
 857         * of the reloc root, regardless of which stage of the relocation we are
 858         * in.
 859         */
 860        if (root->reloc_root) {
 861                reloc_root = root->reloc_root;
 862                reloc_root->last_trans = trans->transid;
 863                return 0;
 864        }
 865
 866        /*
 867         * We are merging reloc roots, we do not need new reloc trees.  Also
 868         * reloc trees never need their own reloc tree.
 869         */
 870        if (!rc->create_reloc_tree ||
 871            root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
 872                return 0;
 873
 874        if (!trans->reloc_reserved) {
 875                rsv = trans->block_rsv;
 876                trans->block_rsv = rc->block_rsv;
 877                clear_rsv = 1;
 878        }
 879        reloc_root = create_reloc_root(trans, root, root->root_key.objectid);
 880        if (clear_rsv)
 881                trans->block_rsv = rsv;
 882        if (IS_ERR(reloc_root))
 883                return PTR_ERR(reloc_root);
 884
 885        ret = __add_reloc_root(reloc_root);
 886        ASSERT(ret != -EEXIST);
 887        if (ret) {
 888                /* Pairs with create_reloc_root */
 889                btrfs_put_root(reloc_root);
 890                return ret;
 891        }
 892        root->reloc_root = btrfs_grab_root(reloc_root);
 893        return 0;
 894}
 895
 896/*
 897 * update root item of reloc tree
 898 */
 899int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
 900                            struct btrfs_root *root)
 901{
 902        struct btrfs_fs_info *fs_info = root->fs_info;
 903        struct btrfs_root *reloc_root;
 904        struct btrfs_root_item *root_item;
 905        int ret;
 906
 907        if (!have_reloc_root(root))
 908                return 0;
 909
 910        reloc_root = root->reloc_root;
 911        root_item = &reloc_root->root_item;
 912
 913        /*
 914         * We are probably ok here, but __del_reloc_root() will drop its ref of
 915         * the root.  We have the ref for root->reloc_root, but just in case
 916         * hold it while we update the reloc root.
 917         */
 918        btrfs_grab_root(reloc_root);
 919
 920        /* root->reloc_root will stay until current relocation finished */
 921        if (fs_info->reloc_ctl->merge_reloc_tree &&
 922            btrfs_root_refs(root_item) == 0) {
 923                set_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state);
 924                /*
 925                 * Mark the tree as dead before we change reloc_root so
 926                 * have_reloc_root will not touch it from now on.
 927                 */
 928                smp_wmb();
 929                __del_reloc_root(reloc_root);
 930        }
 931
 932        if (reloc_root->commit_root != reloc_root->node) {
 933                __update_reloc_root(reloc_root);
 934                btrfs_set_root_node(root_item, reloc_root->node);
 935                free_extent_buffer(reloc_root->commit_root);
 936                reloc_root->commit_root = btrfs_root_node(reloc_root);
 937        }
 938
 939        ret = btrfs_update_root(trans, fs_info->tree_root,
 940                                &reloc_root->root_key, root_item);
 941        btrfs_put_root(reloc_root);
 942        return ret;
 943}
 944
 945/*
 946 * helper to find first cached inode with inode number >= objectid
 947 * in a subvolume
 948 */
 949static struct inode *find_next_inode(struct btrfs_root *root, u64 objectid)
 950{
 951        struct rb_node *node;
 952        struct rb_node *prev;
 953        struct btrfs_inode *entry;
 954        struct inode *inode;
 955
 956        spin_lock(&root->inode_lock);
 957again:
 958        node = root->inode_tree.rb_node;
 959        prev = NULL;
 960        while (node) {
 961                prev = node;
 962                entry = rb_entry(node, struct btrfs_inode, rb_node);
 963
 964                if (objectid < btrfs_ino(entry))
 965                        node = node->rb_left;
 966                else if (objectid > btrfs_ino(entry))
 967                        node = node->rb_right;
 968                else
 969                        break;
 970        }
 971        if (!node) {
 972                while (prev) {
 973                        entry = rb_entry(prev, struct btrfs_inode, rb_node);
 974                        if (objectid <= btrfs_ino(entry)) {
 975                                node = prev;
 976                                break;
 977                        }
 978                        prev = rb_next(prev);
 979                }
 980        }
 981        while (node) {
 982                entry = rb_entry(node, struct btrfs_inode, rb_node);
 983                inode = igrab(&entry->vfs_inode);
 984                if (inode) {
 985                        spin_unlock(&root->inode_lock);
 986                        return inode;
 987                }
 988
 989                objectid = btrfs_ino(entry) + 1;
 990                if (cond_resched_lock(&root->inode_lock))
 991                        goto again;
 992
 993                node = rb_next(node);
 994        }
 995        spin_unlock(&root->inode_lock);
 996        return NULL;
 997}
 998
 999/*
1000 * get new location of data
1001 */
1002static int get_new_location(struct inode *reloc_inode, u64 *new_bytenr,
1003                            u64 bytenr, u64 num_bytes)
1004{
1005        struct btrfs_root *root = BTRFS_I(reloc_inode)->root;
1006        struct btrfs_path *path;
1007        struct btrfs_file_extent_item *fi;
1008        struct extent_buffer *leaf;
1009        int ret;
1010
1011        path = btrfs_alloc_path();
1012        if (!path)
1013                return -ENOMEM;
1014
1015        bytenr -= BTRFS_I(reloc_inode)->index_cnt;
1016        ret = btrfs_lookup_file_extent(NULL, root, path,
1017                        btrfs_ino(BTRFS_I(reloc_inode)), bytenr, 0);
1018        if (ret < 0)
1019                goto out;
1020        if (ret > 0) {
1021                ret = -ENOENT;
1022                goto out;
1023        }
1024
1025        leaf = path->nodes[0];
1026        fi = btrfs_item_ptr(leaf, path->slots[0],
1027                            struct btrfs_file_extent_item);
1028
1029        BUG_ON(btrfs_file_extent_offset(leaf, fi) ||
1030               btrfs_file_extent_compression(leaf, fi) ||
1031               btrfs_file_extent_encryption(leaf, fi) ||
1032               btrfs_file_extent_other_encoding(leaf, fi));
1033
1034        if (num_bytes != btrfs_file_extent_disk_num_bytes(leaf, fi)) {
1035                ret = -EINVAL;
1036                goto out;
1037        }
1038
1039        *new_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
1040        ret = 0;
1041out:
1042        btrfs_free_path(path);
1043        return ret;
1044}
1045
1046/*
1047 * update file extent items in the tree leaf to point to
1048 * the new locations.
1049 */
1050static noinline_for_stack
1051int replace_file_extents(struct btrfs_trans_handle *trans,
1052                         struct reloc_control *rc,
1053                         struct btrfs_root *root,
1054                         struct extent_buffer *leaf)
1055{
1056        struct btrfs_fs_info *fs_info = root->fs_info;
1057        struct btrfs_key key;
1058        struct btrfs_file_extent_item *fi;
1059        struct inode *inode = NULL;
1060        u64 parent;
1061        u64 bytenr;
1062        u64 new_bytenr = 0;
1063        u64 num_bytes;
1064        u64 end;
1065        u32 nritems;
1066        u32 i;
1067        int ret = 0;
1068        int first = 1;
1069        int dirty = 0;
1070
1071        if (rc->stage != UPDATE_DATA_PTRS)
1072                return 0;
1073
1074        /* reloc trees always use full backref */
1075        if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
1076                parent = leaf->start;
1077        else
1078                parent = 0;
1079
1080        nritems = btrfs_header_nritems(leaf);
1081        for (i = 0; i < nritems; i++) {
1082                struct btrfs_ref ref = { 0 };
1083
1084                cond_resched();
1085                btrfs_item_key_to_cpu(leaf, &key, i);
1086                if (key.type != BTRFS_EXTENT_DATA_KEY)
1087                        continue;
1088                fi = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
1089                if (btrfs_file_extent_type(leaf, fi) ==
1090                    BTRFS_FILE_EXTENT_INLINE)
1091                        continue;
1092                bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
1093                num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
1094                if (bytenr == 0)
1095                        continue;
1096                if (!in_range(bytenr, rc->block_group->start,
1097                              rc->block_group->length))
1098                        continue;
1099
1100                /*
1101                 * if we are modifying block in fs tree, wait for readpage
1102                 * to complete and drop the extent cache
1103                 */
1104                if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
1105                        if (first) {
1106                                inode = find_next_inode(root, key.objectid);
1107                                first = 0;
1108                        } else if (inode && btrfs_ino(BTRFS_I(inode)) < key.objectid) {
1109                                btrfs_add_delayed_iput(inode);
1110                                inode = find_next_inode(root, key.objectid);
1111                        }
1112                        if (inode && btrfs_ino(BTRFS_I(inode)) == key.objectid) {
1113                                end = key.offset +
1114                                      btrfs_file_extent_num_bytes(leaf, fi);
1115                                WARN_ON(!IS_ALIGNED(key.offset,
1116                                                    fs_info->sectorsize));
1117                                WARN_ON(!IS_ALIGNED(end, fs_info->sectorsize));
1118                                end--;
1119                                ret = try_lock_extent(&BTRFS_I(inode)->io_tree,
1120                                                      key.offset, end);
1121                                if (!ret)
1122                                        continue;
1123
1124                                btrfs_drop_extent_cache(BTRFS_I(inode),
1125                                                key.offset,     end, 1);
1126                                unlock_extent(&BTRFS_I(inode)->io_tree,
1127                                              key.offset, end);
1128                        }
1129                }
1130
1131                ret = get_new_location(rc->data_inode, &new_bytenr,
1132                                       bytenr, num_bytes);
1133                if (ret) {
1134                        /*
1135                         * Don't have to abort since we've not changed anything
1136                         * in the file extent yet.
1137                         */
1138                        break;
1139                }
1140
1141                btrfs_set_file_extent_disk_bytenr(leaf, fi, new_bytenr);
1142                dirty = 1;
1143
1144                key.offset -= btrfs_file_extent_offset(leaf, fi);
1145                btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, new_bytenr,
1146                                       num_bytes, parent);
1147                ref.real_root = root->root_key.objectid;
1148                btrfs_init_data_ref(&ref, btrfs_header_owner(leaf),
1149                                    key.objectid, key.offset);
1150                ret = btrfs_inc_extent_ref(trans, &ref);
1151                if (ret) {
1152                        btrfs_abort_transaction(trans, ret);
1153                        break;
1154                }
1155
1156                btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, bytenr,
1157                                       num_bytes, parent);
1158                ref.real_root = root->root_key.objectid;
1159                btrfs_init_data_ref(&ref, btrfs_header_owner(leaf),
1160                                    key.objectid, key.offset);
1161                ret = btrfs_free_extent(trans, &ref);
1162                if (ret) {
1163                        btrfs_abort_transaction(trans, ret);
1164                        break;
1165                }
1166        }
1167        if (dirty)
1168                btrfs_mark_buffer_dirty(leaf);
1169        if (inode)
1170                btrfs_add_delayed_iput(inode);
1171        return ret;
1172}
1173
1174static noinline_for_stack
1175int memcmp_node_keys(struct extent_buffer *eb, int slot,
1176                     struct btrfs_path *path, int level)
1177{
1178        struct btrfs_disk_key key1;
1179        struct btrfs_disk_key key2;
1180        btrfs_node_key(eb, &key1, slot);
1181        btrfs_node_key(path->nodes[level], &key2, path->slots[level]);
1182        return memcmp(&key1, &key2, sizeof(key1));
1183}
1184
1185/*
1186 * try to replace tree blocks in fs tree with the new blocks
1187 * in reloc tree. tree blocks haven't been modified since the
1188 * reloc tree was create can be replaced.
1189 *
1190 * if a block was replaced, level of the block + 1 is returned.
1191 * if no block got replaced, 0 is returned. if there are other
1192 * errors, a negative error number is returned.
1193 */
1194static noinline_for_stack
1195int replace_path(struct btrfs_trans_handle *trans, struct reloc_control *rc,
1196                 struct btrfs_root *dest, struct btrfs_root *src,
1197                 struct btrfs_path *path, struct btrfs_key *next_key,
1198                 int lowest_level, int max_level)
1199{
1200        struct btrfs_fs_info *fs_info = dest->fs_info;
1201        struct extent_buffer *eb;
1202        struct extent_buffer *parent;
1203        struct btrfs_ref ref = { 0 };
1204        struct btrfs_key key;
1205        u64 old_bytenr;
1206        u64 new_bytenr;
1207        u64 old_ptr_gen;
1208        u64 new_ptr_gen;
1209        u64 last_snapshot;
1210        u32 blocksize;
1211        int cow = 0;
1212        int level;
1213        int ret;
1214        int slot;
1215
1216        ASSERT(src->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID);
1217        ASSERT(dest->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
1218
1219        last_snapshot = btrfs_root_last_snapshot(&src->root_item);
1220again:
1221        slot = path->slots[lowest_level];
1222        btrfs_node_key_to_cpu(path->nodes[lowest_level], &key, slot);
1223
1224        eb = btrfs_lock_root_node(dest);
1225        level = btrfs_header_level(eb);
1226
1227        if (level < lowest_level) {
1228                btrfs_tree_unlock(eb);
1229                free_extent_buffer(eb);
1230                return 0;
1231        }
1232
1233        if (cow) {
1234                ret = btrfs_cow_block(trans, dest, eb, NULL, 0, &eb,
1235                                      BTRFS_NESTING_COW);
1236                if (ret) {
1237                        btrfs_tree_unlock(eb);
1238                        free_extent_buffer(eb);
1239                        return ret;
1240                }
1241        }
1242
1243        if (next_key) {
1244                next_key->objectid = (u64)-1;
1245                next_key->type = (u8)-1;
1246                next_key->offset = (u64)-1;
1247        }
1248
1249        parent = eb;
1250        while (1) {
1251                level = btrfs_header_level(parent);
1252                ASSERT(level >= lowest_level);
1253
1254                ret = btrfs_bin_search(parent, &key, &slot);
1255                if (ret < 0)
1256                        break;
1257                if (ret && slot > 0)
1258                        slot--;
1259
1260                if (next_key && slot + 1 < btrfs_header_nritems(parent))
1261                        btrfs_node_key_to_cpu(parent, next_key, slot + 1);
1262
1263                old_bytenr = btrfs_node_blockptr(parent, slot);
1264                blocksize = fs_info->nodesize;
1265                old_ptr_gen = btrfs_node_ptr_generation(parent, slot);
1266
1267                if (level <= max_level) {
1268                        eb = path->nodes[level];
1269                        new_bytenr = btrfs_node_blockptr(eb,
1270                                                        path->slots[level]);
1271                        new_ptr_gen = btrfs_node_ptr_generation(eb,
1272                                                        path->slots[level]);
1273                } else {
1274                        new_bytenr = 0;
1275                        new_ptr_gen = 0;
1276                }
1277
1278                if (WARN_ON(new_bytenr > 0 && new_bytenr == old_bytenr)) {
1279                        ret = level;
1280                        break;
1281                }
1282
1283                if (new_bytenr == 0 || old_ptr_gen > last_snapshot ||
1284                    memcmp_node_keys(parent, slot, path, level)) {
1285                        if (level <= lowest_level) {
1286                                ret = 0;
1287                                break;
1288                        }
1289
1290                        eb = btrfs_read_node_slot(parent, slot);
1291                        if (IS_ERR(eb)) {
1292                                ret = PTR_ERR(eb);
1293                                break;
1294                        }
1295                        btrfs_tree_lock(eb);
1296                        if (cow) {
1297                                ret = btrfs_cow_block(trans, dest, eb, parent,
1298                                                      slot, &eb,
1299                                                      BTRFS_NESTING_COW);
1300                                if (ret) {
1301                                        btrfs_tree_unlock(eb);
1302                                        free_extent_buffer(eb);
1303                                        break;
1304                                }
1305                        }
1306
1307                        btrfs_tree_unlock(parent);
1308                        free_extent_buffer(parent);
1309
1310                        parent = eb;
1311                        continue;
1312                }
1313
1314                if (!cow) {
1315                        btrfs_tree_unlock(parent);
1316                        free_extent_buffer(parent);
1317                        cow = 1;
1318                        goto again;
1319                }
1320
1321                btrfs_node_key_to_cpu(path->nodes[level], &key,
1322                                      path->slots[level]);
1323                btrfs_release_path(path);
1324
1325                path->lowest_level = level;
1326                ret = btrfs_search_slot(trans, src, &key, path, 0, 1);
1327                path->lowest_level = 0;
1328                if (ret) {
1329                        if (ret > 0)
1330                                ret = -ENOENT;
1331                        break;
1332                }
1333
1334                /*
1335                 * Info qgroup to trace both subtrees.
1336                 *
1337                 * We must trace both trees.
1338                 * 1) Tree reloc subtree
1339                 *    If not traced, we will leak data numbers
1340                 * 2) Fs subtree
1341                 *    If not traced, we will double count old data
1342                 *
1343                 * We don't scan the subtree right now, but only record
1344                 * the swapped tree blocks.
1345                 * The real subtree rescan is delayed until we have new
1346                 * CoW on the subtree root node before transaction commit.
1347                 */
1348                ret = btrfs_qgroup_add_swapped_blocks(trans, dest,
1349                                rc->block_group, parent, slot,
1350                                path->nodes[level], path->slots[level],
1351                                last_snapshot);
1352                if (ret < 0)
1353                        break;
1354                /*
1355                 * swap blocks in fs tree and reloc tree.
1356                 */
1357                btrfs_set_node_blockptr(parent, slot, new_bytenr);
1358                btrfs_set_node_ptr_generation(parent, slot, new_ptr_gen);
1359                btrfs_mark_buffer_dirty(parent);
1360
1361                btrfs_set_node_blockptr(path->nodes[level],
1362                                        path->slots[level], old_bytenr);
1363                btrfs_set_node_ptr_generation(path->nodes[level],
1364                                              path->slots[level], old_ptr_gen);
1365                btrfs_mark_buffer_dirty(path->nodes[level]);
1366
1367                btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, old_bytenr,
1368                                       blocksize, path->nodes[level]->start);
1369                ref.skip_qgroup = true;
1370                btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid);
1371                ret = btrfs_inc_extent_ref(trans, &ref);
1372                if (ret) {
1373                        btrfs_abort_transaction(trans, ret);
1374                        break;
1375                }
1376                btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, new_bytenr,
1377                                       blocksize, 0);
1378                ref.skip_qgroup = true;
1379                btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid);
1380                ret = btrfs_inc_extent_ref(trans, &ref);
1381                if (ret) {
1382                        btrfs_abort_transaction(trans, ret);
1383                        break;
1384                }
1385
1386                btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, new_bytenr,
1387                                       blocksize, path->nodes[level]->start);
1388                btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid);
1389                ref.skip_qgroup = true;
1390                ret = btrfs_free_extent(trans, &ref);
1391                if (ret) {
1392                        btrfs_abort_transaction(trans, ret);
1393                        break;
1394                }
1395
1396                btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, old_bytenr,
1397                                       blocksize, 0);
1398                btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid);
1399                ref.skip_qgroup = true;
1400                ret = btrfs_free_extent(trans, &ref);
1401                if (ret) {
1402                        btrfs_abort_transaction(trans, ret);
1403                        break;
1404                }
1405
1406                btrfs_unlock_up_safe(path, 0);
1407
1408                ret = level;
1409                break;
1410        }
1411        btrfs_tree_unlock(parent);
1412        free_extent_buffer(parent);
1413        return ret;
1414}
1415
1416/*
1417 * helper to find next relocated block in reloc tree
1418 */
1419static noinline_for_stack
1420int walk_up_reloc_tree(struct btrfs_root *root, struct btrfs_path *path,
1421                       int *level)
1422{
1423        struct extent_buffer *eb;
1424        int i;
1425        u64 last_snapshot;
1426        u32 nritems;
1427
1428        last_snapshot = btrfs_root_last_snapshot(&root->root_item);
1429
1430        for (i = 0; i < *level; i++) {
1431                free_extent_buffer(path->nodes[i]);
1432                path->nodes[i] = NULL;
1433        }
1434
1435        for (i = *level; i < BTRFS_MAX_LEVEL && path->nodes[i]; i++) {
1436                eb = path->nodes[i];
1437                nritems = btrfs_header_nritems(eb);
1438                while (path->slots[i] + 1 < nritems) {
1439                        path->slots[i]++;
1440                        if (btrfs_node_ptr_generation(eb, path->slots[i]) <=
1441                            last_snapshot)
1442                                continue;
1443
1444                        *level = i;
1445                        return 0;
1446                }
1447                free_extent_buffer(path->nodes[i]);
1448                path->nodes[i] = NULL;
1449        }
1450        return 1;
1451}
1452
1453/*
1454 * walk down reloc tree to find relocated block of lowest level
1455 */
1456static noinline_for_stack
1457int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path,
1458                         int *level)
1459{
1460        struct extent_buffer *eb = NULL;
1461        int i;
1462        u64 ptr_gen = 0;
1463        u64 last_snapshot;
1464        u32 nritems;
1465
1466        last_snapshot = btrfs_root_last_snapshot(&root->root_item);
1467
1468        for (i = *level; i > 0; i--) {
1469                eb = path->nodes[i];
1470                nritems = btrfs_header_nritems(eb);
1471                while (path->slots[i] < nritems) {
1472                        ptr_gen = btrfs_node_ptr_generation(eb, path->slots[i]);
1473                        if (ptr_gen > last_snapshot)
1474                                break;
1475                        path->slots[i]++;
1476                }
1477                if (path->slots[i] >= nritems) {
1478                        if (i == *level)
1479                                break;
1480                        *level = i + 1;
1481                        return 0;
1482                }
1483                if (i == 1) {
1484                        *level = i;
1485                        return 0;
1486                }
1487
1488                eb = btrfs_read_node_slot(eb, path->slots[i]);
1489                if (IS_ERR(eb))
1490                        return PTR_ERR(eb);
1491                BUG_ON(btrfs_header_level(eb) != i - 1);
1492                path->nodes[i - 1] = eb;
1493                path->slots[i - 1] = 0;
1494        }
1495        return 1;
1496}
1497
1498/*
1499 * invalidate extent cache for file extents whose key in range of
1500 * [min_key, max_key)
1501 */
1502static int invalidate_extent_cache(struct btrfs_root *root,
1503                                   struct btrfs_key *min_key,
1504                                   struct btrfs_key *max_key)
1505{
1506        struct btrfs_fs_info *fs_info = root->fs_info;
1507        struct inode *inode = NULL;
1508        u64 objectid;
1509        u64 start, end;
1510        u64 ino;
1511
1512        objectid = min_key->objectid;
1513        while (1) {
1514                cond_resched();
1515                iput(inode);
1516
1517                if (objectid > max_key->objectid)
1518                        break;
1519
1520                inode = find_next_inode(root, objectid);
1521                if (!inode)
1522                        break;
1523                ino = btrfs_ino(BTRFS_I(inode));
1524
1525                if (ino > max_key->objectid) {
1526                        iput(inode);
1527                        break;
1528                }
1529
1530                objectid = ino + 1;
1531                if (!S_ISREG(inode->i_mode))
1532                        continue;
1533
1534                if (unlikely(min_key->objectid == ino)) {
1535                        if (min_key->type > BTRFS_EXTENT_DATA_KEY)
1536                                continue;
1537                        if (min_key->type < BTRFS_EXTENT_DATA_KEY)
1538                                start = 0;
1539                        else {
1540                                start = min_key->offset;
1541                                WARN_ON(!IS_ALIGNED(start, fs_info->sectorsize));
1542                        }
1543                } else {
1544                        start = 0;
1545                }
1546
1547                if (unlikely(max_key->objectid == ino)) {
1548                        if (max_key->type < BTRFS_EXTENT_DATA_KEY)
1549                                continue;
1550                        if (max_key->type > BTRFS_EXTENT_DATA_KEY) {
1551                                end = (u64)-1;
1552                        } else {
1553                                if (max_key->offset == 0)
1554                                        continue;
1555                                end = max_key->offset;
1556                                WARN_ON(!IS_ALIGNED(end, fs_info->sectorsize));
1557                                end--;
1558                        }
1559                } else {
1560                        end = (u64)-1;
1561                }
1562
1563                /* the lock_extent waits for readpage to complete */
1564                lock_extent(&BTRFS_I(inode)->io_tree, start, end);
1565                btrfs_drop_extent_cache(BTRFS_I(inode), start, end, 1);
1566                unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
1567        }
1568        return 0;
1569}
1570
1571static int find_next_key(struct btrfs_path *path, int level,
1572                         struct btrfs_key *key)
1573
1574{
1575        while (level < BTRFS_MAX_LEVEL) {
1576                if (!path->nodes[level])
1577                        break;
1578                if (path->slots[level] + 1 <
1579                    btrfs_header_nritems(path->nodes[level])) {
1580                        btrfs_node_key_to_cpu(path->nodes[level], key,
1581                                              path->slots[level] + 1);
1582                        return 0;
1583                }
1584                level++;
1585        }
1586        return 1;
1587}
1588
1589/*
1590 * Insert current subvolume into reloc_control::dirty_subvol_roots
1591 */
1592static int insert_dirty_subvol(struct btrfs_trans_handle *trans,
1593                               struct reloc_control *rc,
1594                               struct btrfs_root *root)
1595{
1596        struct btrfs_root *reloc_root = root->reloc_root;
1597        struct btrfs_root_item *reloc_root_item;
1598        int ret;
1599
1600        /* @root must be a subvolume tree root with a valid reloc tree */
1601        ASSERT(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
1602        ASSERT(reloc_root);
1603
1604        reloc_root_item = &reloc_root->root_item;
1605        memset(&reloc_root_item->drop_progress, 0,
1606                sizeof(reloc_root_item->drop_progress));
1607        btrfs_set_root_drop_level(reloc_root_item, 0);
1608        btrfs_set_root_refs(reloc_root_item, 0);
1609        ret = btrfs_update_reloc_root(trans, root);
1610        if (ret)
1611                return ret;
1612
1613        if (list_empty(&root->reloc_dirty_list)) {
1614                btrfs_grab_root(root);
1615                list_add_tail(&root->reloc_dirty_list, &rc->dirty_subvol_roots);
1616        }
1617
1618        return 0;
1619}
1620
1621static int clean_dirty_subvols(struct reloc_control *rc)
1622{
1623        struct btrfs_root *root;
1624        struct btrfs_root *next;
1625        int ret = 0;
1626        int ret2;
1627
1628        list_for_each_entry_safe(root, next, &rc->dirty_subvol_roots,
1629                                 reloc_dirty_list) {
1630                if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
1631                        /* Merged subvolume, cleanup its reloc root */
1632                        struct btrfs_root *reloc_root = root->reloc_root;
1633
1634                        list_del_init(&root->reloc_dirty_list);
1635                        root->reloc_root = NULL;
1636                        /*
1637                         * Need barrier to ensure clear_bit() only happens after
1638                         * root->reloc_root = NULL. Pairs with have_reloc_root.
1639                         */
1640                        smp_wmb();
1641                        clear_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state);
1642                        if (reloc_root) {
1643                                /*
1644                                 * btrfs_drop_snapshot drops our ref we hold for
1645                                 * ->reloc_root.  If it fails however we must
1646                                 * drop the ref ourselves.
1647                                 */
1648                                ret2 = btrfs_drop_snapshot(reloc_root, 0, 1);
1649                                if (ret2 < 0) {
1650                                        btrfs_put_root(reloc_root);
1651                                        if (!ret)
1652                                                ret = ret2;
1653                                }
1654                        }
1655                        btrfs_put_root(root);
1656                } else {
1657                        /* Orphan reloc tree, just clean it up */
1658                        ret2 = btrfs_drop_snapshot(root, 0, 1);
1659                        if (ret2 < 0) {
1660                                btrfs_put_root(root);
1661                                if (!ret)
1662                                        ret = ret2;
1663                        }
1664                }
1665        }
1666        return ret;
1667}
1668
1669/*
1670 * merge the relocated tree blocks in reloc tree with corresponding
1671 * fs tree.
1672 */
1673static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
1674                                               struct btrfs_root *root)
1675{
1676        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
1677        struct btrfs_key key;
1678        struct btrfs_key next_key;
1679        struct btrfs_trans_handle *trans = NULL;
1680        struct btrfs_root *reloc_root;
1681        struct btrfs_root_item *root_item;
1682        struct btrfs_path *path;
1683        struct extent_buffer *leaf;
1684        int reserve_level;
1685        int level;
1686        int max_level;
1687        int replaced = 0;
1688        int ret = 0;
1689        u32 min_reserved;
1690
1691        path = btrfs_alloc_path();
1692        if (!path)
1693                return -ENOMEM;
1694        path->reada = READA_FORWARD;
1695
1696        reloc_root = root->reloc_root;
1697        root_item = &reloc_root->root_item;
1698
1699        if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) {
1700                level = btrfs_root_level(root_item);
1701                atomic_inc(&reloc_root->node->refs);
1702                path->nodes[level] = reloc_root->node;
1703                path->slots[level] = 0;
1704        } else {
1705                btrfs_disk_key_to_cpu(&key, &root_item->drop_progress);
1706
1707                level = btrfs_root_drop_level(root_item);
1708                BUG_ON(level == 0);
1709                path->lowest_level = level;
1710                ret = btrfs_search_slot(NULL, reloc_root, &key, path, 0, 0);
1711                path->lowest_level = 0;
1712                if (ret < 0) {
1713                        btrfs_free_path(path);
1714                        return ret;
1715                }
1716
1717                btrfs_node_key_to_cpu(path->nodes[level], &next_key,
1718                                      path->slots[level]);
1719                WARN_ON(memcmp(&key, &next_key, sizeof(key)));
1720
1721                btrfs_unlock_up_safe(path, 0);
1722        }
1723
1724        /*
1725         * In merge_reloc_root(), we modify the upper level pointer to swap the
1726         * tree blocks between reloc tree and subvolume tree.  Thus for tree
1727         * block COW, we COW at most from level 1 to root level for each tree.
1728         *
1729         * Thus the needed metadata size is at most root_level * nodesize,
1730         * and * 2 since we have two trees to COW.
1731         */
1732        reserve_level = max_t(int, 1, btrfs_root_level(root_item));
1733        min_reserved = fs_info->nodesize * reserve_level * 2;
1734        memset(&next_key, 0, sizeof(next_key));
1735
1736        while (1) {
1737                ret = btrfs_block_rsv_refill(root, rc->block_rsv, min_reserved,
1738                                             BTRFS_RESERVE_FLUSH_LIMIT);
1739                if (ret)
1740                        goto out;
1741                trans = btrfs_start_transaction(root, 0);
1742                if (IS_ERR(trans)) {
1743                        ret = PTR_ERR(trans);
1744                        trans = NULL;
1745                        goto out;
1746                }
1747
1748                /*
1749                 * At this point we no longer have a reloc_control, so we can't
1750                 * depend on btrfs_init_reloc_root to update our last_trans.
1751                 *
1752                 * But that's ok, we started the trans handle on our
1753                 * corresponding fs_root, which means it's been added to the
1754                 * dirty list.  At commit time we'll still call
1755                 * btrfs_update_reloc_root() and update our root item
1756                 * appropriately.
1757                 */
1758                reloc_root->last_trans = trans->transid;
1759                trans->block_rsv = rc->block_rsv;
1760
1761                replaced = 0;
1762                max_level = level;
1763
1764                ret = walk_down_reloc_tree(reloc_root, path, &level);
1765                if (ret < 0)
1766                        goto out;
1767                if (ret > 0)
1768                        break;
1769
1770                if (!find_next_key(path, level, &key) &&
1771                    btrfs_comp_cpu_keys(&next_key, &key) >= 0) {
1772                        ret = 0;
1773                } else {
1774                        ret = replace_path(trans, rc, root, reloc_root, path,
1775                                           &next_key, level, max_level);
1776                }
1777                if (ret < 0)
1778                        goto out;
1779                if (ret > 0) {
1780                        level = ret;
1781                        btrfs_node_key_to_cpu(path->nodes[level], &key,
1782                                              path->slots[level]);
1783                        replaced = 1;
1784                }
1785
1786                ret = walk_up_reloc_tree(reloc_root, path, &level);
1787                if (ret > 0)
1788                        break;
1789
1790                BUG_ON(level == 0);
1791                /*
1792                 * save the merging progress in the drop_progress.
1793                 * this is OK since root refs == 1 in this case.
1794                 */
1795                btrfs_node_key(path->nodes[level], &root_item->drop_progress,
1796                               path->slots[level]);
1797                btrfs_set_root_drop_level(root_item, level);
1798
1799                btrfs_end_transaction_throttle(trans);
1800                trans = NULL;
1801
1802                btrfs_btree_balance_dirty(fs_info);
1803
1804                if (replaced && rc->stage == UPDATE_DATA_PTRS)
1805                        invalidate_extent_cache(root, &key, &next_key);
1806        }
1807
1808        /*
1809         * handle the case only one block in the fs tree need to be
1810         * relocated and the block is tree root.
1811         */
1812        leaf = btrfs_lock_root_node(root);
1813        ret = btrfs_cow_block(trans, root, leaf, NULL, 0, &leaf,
1814                              BTRFS_NESTING_COW);
1815        btrfs_tree_unlock(leaf);
1816        free_extent_buffer(leaf);
1817out:
1818        btrfs_free_path(path);
1819
1820        if (ret == 0) {
1821                ret = insert_dirty_subvol(trans, rc, root);
1822                if (ret)
1823                        btrfs_abort_transaction(trans, ret);
1824        }
1825
1826        if (trans)
1827                btrfs_end_transaction_throttle(trans);
1828
1829        btrfs_btree_balance_dirty(fs_info);
1830
1831        if (replaced && rc->stage == UPDATE_DATA_PTRS)
1832                invalidate_extent_cache(root, &key, &next_key);
1833
1834        return ret;
1835}
1836
1837static noinline_for_stack
1838int prepare_to_merge(struct reloc_control *rc, int err)
1839{
1840        struct btrfs_root *root = rc->extent_root;
1841        struct btrfs_fs_info *fs_info = root->fs_info;
1842        struct btrfs_root *reloc_root;
1843        struct btrfs_trans_handle *trans;
1844        LIST_HEAD(reloc_roots);
1845        u64 num_bytes = 0;
1846        int ret;
1847
1848        mutex_lock(&fs_info->reloc_mutex);
1849        rc->merging_rsv_size += fs_info->nodesize * (BTRFS_MAX_LEVEL - 1) * 2;
1850        rc->merging_rsv_size += rc->nodes_relocated * 2;
1851        mutex_unlock(&fs_info->reloc_mutex);
1852
1853again:
1854        if (!err) {
1855                num_bytes = rc->merging_rsv_size;
1856                ret = btrfs_block_rsv_add(root, rc->block_rsv, num_bytes,
1857                                          BTRFS_RESERVE_FLUSH_ALL);
1858                if (ret)
1859                        err = ret;
1860        }
1861
1862        trans = btrfs_join_transaction(rc->extent_root);
1863        if (IS_ERR(trans)) {
1864                if (!err)
1865                        btrfs_block_rsv_release(fs_info, rc->block_rsv,
1866                                                num_bytes, NULL);
1867                return PTR_ERR(trans);
1868        }
1869
1870        if (!err) {
1871                if (num_bytes != rc->merging_rsv_size) {
1872                        btrfs_end_transaction(trans);
1873                        btrfs_block_rsv_release(fs_info, rc->block_rsv,
1874                                                num_bytes, NULL);
1875                        goto again;
1876                }
1877        }
1878
1879        rc->merge_reloc_tree = 1;
1880
1881        while (!list_empty(&rc->reloc_roots)) {
1882                reloc_root = list_entry(rc->reloc_roots.next,
1883                                        struct btrfs_root, root_list);
1884                list_del_init(&reloc_root->root_list);
1885
1886                root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset,
1887                                false);
1888                if (IS_ERR(root)) {
1889                        /*
1890                         * Even if we have an error we need this reloc root
1891                         * back on our list so we can clean up properly.
1892                         */
1893                        list_add(&reloc_root->root_list, &reloc_roots);
1894                        btrfs_abort_transaction(trans, (int)PTR_ERR(root));
1895                        if (!err)
1896                                err = PTR_ERR(root);
1897                        break;
1898                }
1899                ASSERT(root->reloc_root == reloc_root);
1900
1901                /*
1902                 * set reference count to 1, so btrfs_recover_relocation
1903                 * knows it should resumes merging
1904                 */
1905                if (!err)
1906                        btrfs_set_root_refs(&reloc_root->root_item, 1);
1907                ret = btrfs_update_reloc_root(trans, root);
1908
1909                /*
1910                 * Even if we have an error we need this reloc root back on our
1911                 * list so we can clean up properly.
1912                 */
1913                list_add(&reloc_root->root_list, &reloc_roots);
1914                btrfs_put_root(root);
1915
1916                if (ret) {
1917                        btrfs_abort_transaction(trans, ret);
1918                        if (!err)
1919                                err = ret;
1920                        break;
1921                }
1922        }
1923
1924        list_splice(&reloc_roots, &rc->reloc_roots);
1925
1926        if (!err)
1927                err = btrfs_commit_transaction(trans);
1928        else
1929                btrfs_end_transaction(trans);
1930        return err;
1931}
1932
1933static noinline_for_stack
1934void free_reloc_roots(struct list_head *list)
1935{
1936        struct btrfs_root *reloc_root, *tmp;
1937
1938        list_for_each_entry_safe(reloc_root, tmp, list, root_list)
1939                __del_reloc_root(reloc_root);
1940}
1941
1942static noinline_for_stack
1943void merge_reloc_roots(struct reloc_control *rc)
1944{
1945        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
1946        struct btrfs_root *root;
1947        struct btrfs_root *reloc_root;
1948        LIST_HEAD(reloc_roots);
1949        int found = 0;
1950        int ret = 0;
1951again:
1952        root = rc->extent_root;
1953
1954        /*
1955         * this serializes us with btrfs_record_root_in_transaction,
1956         * we have to make sure nobody is in the middle of
1957         * adding their roots to the list while we are
1958         * doing this splice
1959         */
1960        mutex_lock(&fs_info->reloc_mutex);
1961        list_splice_init(&rc->reloc_roots, &reloc_roots);
1962        mutex_unlock(&fs_info->reloc_mutex);
1963
1964        while (!list_empty(&reloc_roots)) {
1965                found = 1;
1966                reloc_root = list_entry(reloc_roots.next,
1967                                        struct btrfs_root, root_list);
1968
1969                root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset,
1970                                         false);
1971                if (btrfs_root_refs(&reloc_root->root_item) > 0) {
1972                        if (IS_ERR(root)) {
1973                                /*
1974                                 * For recovery we read the fs roots on mount,
1975                                 * and if we didn't find the root then we marked
1976                                 * the reloc root as a garbage root.  For normal
1977                                 * relocation obviously the root should exist in
1978                                 * memory.  However there's no reason we can't
1979                                 * handle the error properly here just in case.
1980                                 */
1981                                ASSERT(0);
1982                                ret = PTR_ERR(root);
1983                                goto out;
1984                        }
1985                        if (root->reloc_root != reloc_root) {
1986                                /*
1987                                 * This is actually impossible without something
1988                                 * going really wrong (like weird race condition
1989                                 * or cosmic rays).
1990                                 */
1991                                ASSERT(0);
1992                                ret = -EINVAL;
1993                                goto out;
1994                        }
1995                        ret = merge_reloc_root(rc, root);
1996                        btrfs_put_root(root);
1997                        if (ret) {
1998                                if (list_empty(&reloc_root->root_list))
1999                                        list_add_tail(&reloc_root->root_list,
2000                                                      &reloc_roots);
2001                                goto out;
2002                        }
2003                } else {
2004                        if (!IS_ERR(root)) {
2005                                if (root->reloc_root == reloc_root) {
2006                                        root->reloc_root = NULL;
2007                                        btrfs_put_root(reloc_root);
2008                                }
2009                                clear_bit(BTRFS_ROOT_DEAD_RELOC_TREE,
2010                                          &root->state);
2011                                btrfs_put_root(root);
2012                        }
2013
2014                        list_del_init(&reloc_root->root_list);
2015                        /* Don't forget to queue this reloc root for cleanup */
2016                        list_add_tail(&reloc_root->reloc_dirty_list,
2017                                      &rc->dirty_subvol_roots);
2018                }
2019        }
2020
2021        if (found) {
2022                found = 0;
2023                goto again;
2024        }
2025out:
2026        if (ret) {
2027                btrfs_handle_fs_error(fs_info, ret, NULL);
2028                free_reloc_roots(&reloc_roots);
2029
2030                /* new reloc root may be added */
2031                mutex_lock(&fs_info->reloc_mutex);
2032                list_splice_init(&rc->reloc_roots, &reloc_roots);
2033                mutex_unlock(&fs_info->reloc_mutex);
2034                free_reloc_roots(&reloc_roots);
2035        }
2036
2037        /*
2038         * We used to have
2039         *
2040         * BUG_ON(!RB_EMPTY_ROOT(&rc->reloc_root_tree.rb_root));
2041         *
2042         * here, but it's wrong.  If we fail to start the transaction in
2043         * prepare_to_merge() we will have only 0 ref reloc roots, none of which
2044         * have actually been removed from the reloc_root_tree rb tree.  This is
2045         * fine because we're bailing here, and we hold a reference on the root
2046         * for the list that holds it, so these roots will be cleaned up when we
2047         * do the reloc_dirty_list afterwards.  Meanwhile the root->reloc_root
2048         * will be cleaned up on unmount.
2049         *
2050         * The remaining nodes will be cleaned up by free_reloc_control.
2051         */
2052}
2053
2054static void free_block_list(struct rb_root *blocks)
2055{
2056        struct tree_block *block;
2057        struct rb_node *rb_node;
2058        while ((rb_node = rb_first(blocks))) {
2059                block = rb_entry(rb_node, struct tree_block, rb_node);
2060                rb_erase(rb_node, blocks);
2061                kfree(block);
2062        }
2063}
2064
2065static int record_reloc_root_in_trans(struct btrfs_trans_handle *trans,
2066                                      struct btrfs_root *reloc_root)
2067{
2068        struct btrfs_fs_info *fs_info = reloc_root->fs_info;
2069        struct btrfs_root *root;
2070        int ret;
2071
2072        if (reloc_root->last_trans == trans->transid)
2073                return 0;
2074
2075        root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset, false);
2076
2077        /*
2078         * This should succeed, since we can't have a reloc root without having
2079         * already looked up the actual root and created the reloc root for this
2080         * root.
2081         *
2082         * However if there's some sort of corruption where we have a ref to a
2083         * reloc root without a corresponding root this could return ENOENT.
2084         */
2085        if (IS_ERR(root)) {
2086                ASSERT(0);
2087                return PTR_ERR(root);
2088        }
2089        if (root->reloc_root != reloc_root) {
2090                ASSERT(0);
2091                btrfs_err(fs_info,
2092                          "root %llu has two reloc roots associated with it",
2093                          reloc_root->root_key.offset);
2094                btrfs_put_root(root);
2095                return -EUCLEAN;
2096        }
2097        ret = btrfs_record_root_in_trans(trans, root);
2098        btrfs_put_root(root);
2099
2100        return ret;
2101}
2102
2103static noinline_for_stack
2104struct btrfs_root *select_reloc_root(struct btrfs_trans_handle *trans,
2105                                     struct reloc_control *rc,
2106                                     struct btrfs_backref_node *node,
2107                                     struct btrfs_backref_edge *edges[])
2108{
2109        struct btrfs_backref_node *next;
2110        struct btrfs_root *root;
2111        int index = 0;
2112        int ret;
2113
2114        next = node;
2115        while (1) {
2116                cond_resched();
2117                next = walk_up_backref(next, edges, &index);
2118                root = next->root;
2119
2120                /*
2121                 * If there is no root, then our references for this block are
2122                 * incomplete, as we should be able to walk all the way up to a
2123                 * block that is owned by a root.
2124                 *
2125                 * This path is only for SHAREABLE roots, so if we come upon a
2126                 * non-SHAREABLE root then we have backrefs that resolve
2127                 * improperly.
2128                 *
2129                 * Both of these cases indicate file system corruption, or a bug
2130                 * in the backref walking code.
2131                 */
2132                if (!root) {
2133                        ASSERT(0);
2134                        btrfs_err(trans->fs_info,
2135                "bytenr %llu doesn't have a backref path ending in a root",
2136                                  node->bytenr);
2137                        return ERR_PTR(-EUCLEAN);
2138                }
2139                if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) {
2140                        ASSERT(0);
2141                        btrfs_err(trans->fs_info,
2142        "bytenr %llu has multiple refs with one ending in a non-shareable root",
2143                                  node->bytenr);
2144                        return ERR_PTR(-EUCLEAN);
2145                }
2146
2147                if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) {
2148                        ret = record_reloc_root_in_trans(trans, root);
2149                        if (ret)
2150                                return ERR_PTR(ret);
2151                        break;
2152                }
2153
2154                ret = btrfs_record_root_in_trans(trans, root);
2155                if (ret)
2156                        return ERR_PTR(ret);
2157                root = root->reloc_root;
2158
2159                /*
2160                 * We could have raced with another thread which failed, so
2161                 * root->reloc_root may not be set, return ENOENT in this case.
2162                 */
2163                if (!root)
2164                        return ERR_PTR(-ENOENT);
2165
2166                if (next->new_bytenr != root->node->start) {
2167                        /*
2168                         * We just created the reloc root, so we shouldn't have
2169                         * ->new_bytenr set and this shouldn't be in the changed
2170                         *  list.  If it is then we have multiple roots pointing
2171                         *  at the same bytenr which indicates corruption, or
2172                         *  we've made a mistake in the backref walking code.
2173                         */
2174                        ASSERT(next->new_bytenr == 0);
2175                        ASSERT(list_empty(&next->list));
2176                        if (next->new_bytenr || !list_empty(&next->list)) {
2177                                btrfs_err(trans->fs_info,
2178        "bytenr %llu possibly has multiple roots pointing at the same bytenr %llu",
2179                                          node->bytenr, next->bytenr);
2180                                return ERR_PTR(-EUCLEAN);
2181                        }
2182
2183                        next->new_bytenr = root->node->start;
2184                        btrfs_put_root(next->root);
2185                        next->root = btrfs_grab_root(root);
2186                        ASSERT(next->root);
2187                        list_add_tail(&next->list,
2188                                      &rc->backref_cache.changed);
2189                        mark_block_processed(rc, next);
2190                        break;
2191                }
2192
2193                WARN_ON(1);
2194                root = NULL;
2195                next = walk_down_backref(edges, &index);
2196                if (!next || next->level <= node->level)
2197                        break;
2198        }
2199        if (!root) {
2200                /*
2201                 * This can happen if there's fs corruption or if there's a bug
2202                 * in the backref lookup code.
2203                 */
2204                ASSERT(0);
2205                return ERR_PTR(-ENOENT);
2206        }
2207
2208        next = node;
2209        /* setup backref node path for btrfs_reloc_cow_block */
2210        while (1) {
2211                rc->backref_cache.path[next->level] = next;
2212                if (--index < 0)
2213                        break;
2214                next = edges[index]->node[UPPER];
2215        }
2216        return root;
2217}
2218
2219/*
2220 * Select a tree root for relocation.
2221 *
2222 * Return NULL if the block is not shareable. We should use do_relocation() in
2223 * this case.
2224 *
2225 * Return a tree root pointer if the block is shareable.
2226 * Return -ENOENT if the block is root of reloc tree.
2227 */
2228static noinline_for_stack
2229struct btrfs_root *select_one_root(struct btrfs_backref_node *node)
2230{
2231        struct btrfs_backref_node *next;
2232        struct btrfs_root *root;
2233        struct btrfs_root *fs_root = NULL;
2234        struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
2235        int index = 0;
2236
2237        next = node;
2238        while (1) {
2239                cond_resched();
2240                next = walk_up_backref(next, edges, &index);
2241                root = next->root;
2242
2243                /*
2244                 * This can occur if we have incomplete extent refs leading all
2245                 * the way up a particular path, in this case return -EUCLEAN.
2246                 */
2247                if (!root)
2248                        return ERR_PTR(-EUCLEAN);
2249
2250                /* No other choice for non-shareable tree */
2251                if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))
2252                        return root;
2253
2254                if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID)
2255                        fs_root = root;
2256
2257                if (next != node)
2258                        return NULL;
2259
2260                next = walk_down_backref(edges, &index);
2261                if (!next || next->level <= node->level)
2262                        break;
2263        }
2264
2265        if (!fs_root)
2266                return ERR_PTR(-ENOENT);
2267        return fs_root;
2268}
2269
2270static noinline_for_stack
2271u64 calcu_metadata_size(struct reloc_control *rc,
2272                        struct btrfs_backref_node *node, int reserve)
2273{
2274        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
2275        struct btrfs_backref_node *next = node;
2276        struct btrfs_backref_edge *edge;
2277        struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
2278        u64 num_bytes = 0;
2279        int index = 0;
2280
2281        BUG_ON(reserve && node->processed);
2282
2283        while (next) {
2284                cond_resched();
2285                while (1) {
2286                        if (next->processed && (reserve || next != node))
2287                                break;
2288
2289                        num_bytes += fs_info->nodesize;
2290
2291                        if (list_empty(&next->upper))
2292                                break;
2293
2294                        edge = list_entry(next->upper.next,
2295                                        struct btrfs_backref_edge, list[LOWER]);
2296                        edges[index++] = edge;
2297                        next = edge->node[UPPER];
2298                }
2299                next = walk_down_backref(edges, &index);
2300        }
2301        return num_bytes;
2302}
2303
2304static int reserve_metadata_space(struct btrfs_trans_handle *trans,
2305                                  struct reloc_control *rc,
2306                                  struct btrfs_backref_node *node)
2307{
2308        struct btrfs_root *root = rc->extent_root;
2309        struct btrfs_fs_info *fs_info = root->fs_info;
2310        u64 num_bytes;
2311        int ret;
2312        u64 tmp;
2313
2314        num_bytes = calcu_metadata_size(rc, node, 1) * 2;
2315
2316        trans->block_rsv = rc->block_rsv;
2317        rc->reserved_bytes += num_bytes;
2318
2319        /*
2320         * We are under a transaction here so we can only do limited flushing.
2321         * If we get an enospc just kick back -EAGAIN so we know to drop the
2322         * transaction and try to refill when we can flush all the things.
2323         */
2324        ret = btrfs_block_rsv_refill(root, rc->block_rsv, num_bytes,
2325                                BTRFS_RESERVE_FLUSH_LIMIT);
2326        if (ret) {
2327                tmp = fs_info->nodesize * RELOCATION_RESERVED_NODES;
2328                while (tmp <= rc->reserved_bytes)
2329                        tmp <<= 1;
2330                /*
2331                 * only one thread can access block_rsv at this point,
2332                 * so we don't need hold lock to protect block_rsv.
2333                 * we expand more reservation size here to allow enough
2334                 * space for relocation and we will return earlier in
2335                 * enospc case.
2336                 */
2337                rc->block_rsv->size = tmp + fs_info->nodesize *
2338                                      RELOCATION_RESERVED_NODES;
2339                return -EAGAIN;
2340        }
2341
2342        return 0;
2343}
2344
2345/*
2346 * relocate a block tree, and then update pointers in upper level
2347 * blocks that reference the block to point to the new location.
2348 *
2349 * if called by link_to_upper, the block has already been relocated.
2350 * in that case this function just updates pointers.
2351 */
2352static int do_relocation(struct btrfs_trans_handle *trans,
2353                         struct reloc_control *rc,
2354                         struct btrfs_backref_node *node,
2355                         struct btrfs_key *key,
2356                         struct btrfs_path *path, int lowest)
2357{
2358        struct btrfs_backref_node *upper;
2359        struct btrfs_backref_edge *edge;
2360        struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
2361        struct btrfs_root *root;
2362        struct extent_buffer *eb;
2363        u32 blocksize;
2364        u64 bytenr;
2365        int slot;
2366        int ret = 0;
2367
2368        /*
2369         * If we are lowest then this is the first time we're processing this
2370         * block, and thus shouldn't have an eb associated with it yet.
2371         */
2372        ASSERT(!lowest || !node->eb);
2373
2374        path->lowest_level = node->level + 1;
2375        rc->backref_cache.path[node->level] = node;
2376        list_for_each_entry(edge, &node->upper, list[LOWER]) {
2377                struct btrfs_ref ref = { 0 };
2378
2379                cond_resched();
2380
2381                upper = edge->node[UPPER];
2382                root = select_reloc_root(trans, rc, upper, edges);
2383                if (IS_ERR(root)) {
2384                        ret = PTR_ERR(root);
2385                        goto next;
2386                }
2387
2388                if (upper->eb && !upper->locked) {
2389                        if (!lowest) {
2390                                ret = btrfs_bin_search(upper->eb, key, &slot);
2391                                if (ret < 0)
2392                                        goto next;
2393                                BUG_ON(ret);
2394                                bytenr = btrfs_node_blockptr(upper->eb, slot);
2395                                if (node->eb->start == bytenr)
2396                                        goto next;
2397                        }
2398                        btrfs_backref_drop_node_buffer(upper);
2399                }
2400
2401                if (!upper->eb) {
2402                        ret = btrfs_search_slot(trans, root, key, path, 0, 1);
2403                        if (ret) {
2404                                if (ret > 0)
2405                                        ret = -ENOENT;
2406
2407                                btrfs_release_path(path);
2408                                break;
2409                        }
2410
2411                        if (!upper->eb) {
2412                                upper->eb = path->nodes[upper->level];
2413                                path->nodes[upper->level] = NULL;
2414                        } else {
2415                                BUG_ON(upper->eb != path->nodes[upper->level]);
2416                        }
2417
2418                        upper->locked = 1;
2419                        path->locks[upper->level] = 0;
2420
2421                        slot = path->slots[upper->level];
2422                        btrfs_release_path(path);
2423                } else {
2424                        ret = btrfs_bin_search(upper->eb, key, &slot);
2425                        if (ret < 0)
2426                                goto next;
2427                        BUG_ON(ret);
2428                }
2429
2430                bytenr = btrfs_node_blockptr(upper->eb, slot);
2431                if (lowest) {
2432                        if (bytenr != node->bytenr) {
2433                                btrfs_err(root->fs_info,
2434                "lowest leaf/node mismatch: bytenr %llu node->bytenr %llu slot %d upper %llu",
2435                                          bytenr, node->bytenr, slot,
2436                                          upper->eb->start);
2437                                ret = -EIO;
2438                                goto next;
2439                        }
2440                } else {
2441                        if (node->eb->start == bytenr)
2442                                goto next;
2443                }
2444
2445                blocksize = root->fs_info->nodesize;
2446                eb = btrfs_read_node_slot(upper->eb, slot);
2447                if (IS_ERR(eb)) {
2448                        ret = PTR_ERR(eb);
2449                        goto next;
2450                }
2451                btrfs_tree_lock(eb);
2452
2453                if (!node->eb) {
2454                        ret = btrfs_cow_block(trans, root, eb, upper->eb,
2455                                              slot, &eb, BTRFS_NESTING_COW);
2456                        btrfs_tree_unlock(eb);
2457                        free_extent_buffer(eb);
2458                        if (ret < 0)
2459                                goto next;
2460                        /*
2461                         * We've just COWed this block, it should have updated
2462                         * the correct backref node entry.
2463                         */
2464                        ASSERT(node->eb == eb);
2465                } else {
2466                        btrfs_set_node_blockptr(upper->eb, slot,
2467                                                node->eb->start);
2468                        btrfs_set_node_ptr_generation(upper->eb, slot,
2469                                                      trans->transid);
2470                        btrfs_mark_buffer_dirty(upper->eb);
2471
2472                        btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF,
2473                                               node->eb->start, blocksize,
2474                                               upper->eb->start);
2475                        ref.real_root = root->root_key.objectid;
2476                        btrfs_init_tree_ref(&ref, node->level,
2477                                            btrfs_header_owner(upper->eb));
2478                        ret = btrfs_inc_extent_ref(trans, &ref);
2479                        if (!ret)
2480                                ret = btrfs_drop_subtree(trans, root, eb,
2481                                                         upper->eb);
2482                        if (ret)
2483                                btrfs_abort_transaction(trans, ret);
2484                }
2485next:
2486                if (!upper->pending)
2487                        btrfs_backref_drop_node_buffer(upper);
2488                else
2489                        btrfs_backref_unlock_node_buffer(upper);
2490                if (ret)
2491                        break;
2492        }
2493
2494        if (!ret && node->pending) {
2495                btrfs_backref_drop_node_buffer(node);
2496                list_move_tail(&node->list, &rc->backref_cache.changed);
2497                node->pending = 0;
2498        }
2499
2500        path->lowest_level = 0;
2501
2502        /*
2503         * We should have allocated all of our space in the block rsv and thus
2504         * shouldn't ENOSPC.
2505         */
2506        ASSERT(ret != -ENOSPC);
2507        return ret;
2508}
2509
2510static int link_to_upper(struct btrfs_trans_handle *trans,
2511                         struct reloc_control *rc,
2512                         struct btrfs_backref_node *node,
2513                         struct btrfs_path *path)
2514{
2515        struct btrfs_key key;
2516
2517        btrfs_node_key_to_cpu(node->eb, &key, 0);
2518        return do_relocation(trans, rc, node, &key, path, 0);
2519}
2520
2521static int finish_pending_nodes(struct btrfs_trans_handle *trans,
2522                                struct reloc_control *rc,
2523                                struct btrfs_path *path, int err)
2524{
2525        LIST_HEAD(list);
2526        struct btrfs_backref_cache *cache = &rc->backref_cache;
2527        struct btrfs_backref_node *node;
2528        int level;
2529        int ret;
2530
2531        for (level = 0; level < BTRFS_MAX_LEVEL; level++) {
2532                while (!list_empty(&cache->pending[level])) {
2533                        node = list_entry(cache->pending[level].next,
2534                                          struct btrfs_backref_node, list);
2535                        list_move_tail(&node->list, &list);
2536                        BUG_ON(!node->pending);
2537
2538                        if (!err) {
2539                                ret = link_to_upper(trans, rc, node, path);
2540                                if (ret < 0)
2541                                        err = ret;
2542                        }
2543                }
2544                list_splice_init(&list, &cache->pending[level]);
2545        }
2546        return err;
2547}
2548
2549/*
2550 * mark a block and all blocks directly/indirectly reference the block
2551 * as processed.
2552 */
2553static void update_processed_blocks(struct reloc_control *rc,
2554                                    struct btrfs_backref_node *node)
2555{
2556        struct btrfs_backref_node *next = node;
2557        struct btrfs_backref_edge *edge;
2558        struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1];
2559        int index = 0;
2560
2561        while (next) {
2562                cond_resched();
2563                while (1) {
2564                        if (next->processed)
2565                                break;
2566
2567                        mark_block_processed(rc, next);
2568
2569                        if (list_empty(&next->upper))
2570                                break;
2571
2572                        edge = list_entry(next->upper.next,
2573                                        struct btrfs_backref_edge, list[LOWER]);
2574                        edges[index++] = edge;
2575                        next = edge->node[UPPER];
2576                }
2577                next = walk_down_backref(edges, &index);
2578        }
2579}
2580
2581static int tree_block_processed(u64 bytenr, struct reloc_control *rc)
2582{
2583        u32 blocksize = rc->extent_root->fs_info->nodesize;
2584
2585        if (test_range_bit(&rc->processed_blocks, bytenr,
2586                           bytenr + blocksize - 1, EXTENT_DIRTY, 1, NULL))
2587                return 1;
2588        return 0;
2589}
2590
2591static int get_tree_block_key(struct btrfs_fs_info *fs_info,
2592                              struct tree_block *block)
2593{
2594        struct extent_buffer *eb;
2595
2596        eb = read_tree_block(fs_info, block->bytenr, block->owner,
2597                             block->key.offset, block->level, NULL);
2598        if (IS_ERR(eb)) {
2599                return PTR_ERR(eb);
2600        } else if (!extent_buffer_uptodate(eb)) {
2601                free_extent_buffer(eb);
2602                return -EIO;
2603        }
2604        if (block->level == 0)
2605                btrfs_item_key_to_cpu(eb, &block->key, 0);
2606        else
2607                btrfs_node_key_to_cpu(eb, &block->key, 0);
2608        free_extent_buffer(eb);
2609        block->key_ready = 1;
2610        return 0;
2611}
2612
2613/*
2614 * helper function to relocate a tree block
2615 */
2616static int relocate_tree_block(struct btrfs_trans_handle *trans,
2617                                struct reloc_control *rc,
2618                                struct btrfs_backref_node *node,
2619                                struct btrfs_key *key,
2620                                struct btrfs_path *path)
2621{
2622        struct btrfs_root *root;
2623        int ret = 0;
2624
2625        if (!node)
2626                return 0;
2627
2628        /*
2629         * If we fail here we want to drop our backref_node because we are going
2630         * to start over and regenerate the tree for it.
2631         */
2632        ret = reserve_metadata_space(trans, rc, node);
2633        if (ret)
2634                goto out;
2635
2636        BUG_ON(node->processed);
2637        root = select_one_root(node);
2638        if (IS_ERR(root)) {
2639                ret = PTR_ERR(root);
2640
2641                /* See explanation in select_one_root for the -EUCLEAN case. */
2642                ASSERT(ret == -ENOENT);
2643                if (ret == -ENOENT) {
2644                        ret = 0;
2645                        update_processed_blocks(rc, node);
2646                }
2647                goto out;
2648        }
2649
2650        if (root) {
2651                if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) {
2652                        /*
2653                         * This block was the root block of a root, and this is
2654                         * the first time we're processing the block and thus it
2655                         * should not have had the ->new_bytenr modified and
2656                         * should have not been included on the changed list.
2657                         *
2658                         * However in the case of corruption we could have
2659                         * multiple refs pointing to the same block improperly,
2660                         * and thus we would trip over these checks.  ASSERT()
2661                         * for the developer case, because it could indicate a
2662                         * bug in the backref code, however error out for a
2663                         * normal user in the case of corruption.
2664                         */
2665                        ASSERT(node->new_bytenr == 0);
2666                        ASSERT(list_empty(&node->list));
2667                        if (node->new_bytenr || !list_empty(&node->list)) {
2668                                btrfs_err(root->fs_info,
2669                                  "bytenr %llu has improper references to it",
2670                                          node->bytenr);
2671                                ret = -EUCLEAN;
2672                                goto out;
2673                        }
2674                        ret = btrfs_record_root_in_trans(trans, root);
2675                        if (ret)
2676                                goto out;
2677                        /*
2678                         * Another thread could have failed, need to check if we
2679                         * have reloc_root actually set.
2680                         */
2681                        if (!root->reloc_root) {
2682                                ret = -ENOENT;
2683                                goto out;
2684                        }
2685                        root = root->reloc_root;
2686                        node->new_bytenr = root->node->start;
2687                        btrfs_put_root(node->root);
2688                        node->root = btrfs_grab_root(root);
2689                        ASSERT(node->root);
2690                        list_add_tail(&node->list, &rc->backref_cache.changed);
2691                } else {
2692                        path->lowest_level = node->level;
2693                        ret = btrfs_search_slot(trans, root, key, path, 0, 1);
2694                        btrfs_release_path(path);
2695                        if (ret > 0)
2696                                ret = 0;
2697                }
2698                if (!ret)
2699                        update_processed_blocks(rc, node);
2700        } else {
2701                ret = do_relocation(trans, rc, node, key, path, 1);
2702        }
2703out:
2704        if (ret || node->level == 0 || node->cowonly)
2705                btrfs_backref_cleanup_node(&rc->backref_cache, node);
2706        return ret;
2707}
2708
2709/*
2710 * relocate a list of blocks
2711 */
2712static noinline_for_stack
2713int relocate_tree_blocks(struct btrfs_trans_handle *trans,
2714                         struct reloc_control *rc, struct rb_root *blocks)
2715{
2716        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
2717        struct btrfs_backref_node *node;
2718        struct btrfs_path *path;
2719        struct tree_block *block;
2720        struct tree_block *next;
2721        int ret;
2722        int err = 0;
2723
2724        path = btrfs_alloc_path();
2725        if (!path) {
2726                err = -ENOMEM;
2727                goto out_free_blocks;
2728        }
2729
2730        /* Kick in readahead for tree blocks with missing keys */
2731        rbtree_postorder_for_each_entry_safe(block, next, blocks, rb_node) {
2732                if (!block->key_ready)
2733                        btrfs_readahead_tree_block(fs_info, block->bytenr,
2734                                                   block->owner, 0,
2735                                                   block->level);
2736        }
2737
2738        /* Get first keys */
2739        rbtree_postorder_for_each_entry_safe(block, next, blocks, rb_node) {
2740                if (!block->key_ready) {
2741                        err = get_tree_block_key(fs_info, block);
2742                        if (err)
2743                                goto out_free_path;
2744                }
2745        }
2746
2747        /* Do tree relocation */
2748        rbtree_postorder_for_each_entry_safe(block, next, blocks, rb_node) {
2749                node = build_backref_tree(rc, &block->key,
2750                                          block->level, block->bytenr);
2751                if (IS_ERR(node)) {
2752                        err = PTR_ERR(node);
2753                        goto out;
2754                }
2755
2756                ret = relocate_tree_block(trans, rc, node, &block->key,
2757                                          path);
2758                if (ret < 0) {
2759                        err = ret;
2760                        break;
2761                }
2762        }
2763out:
2764        err = finish_pending_nodes(trans, rc, path, err);
2765
2766out_free_path:
2767        btrfs_free_path(path);
2768out_free_blocks:
2769        free_block_list(blocks);
2770        return err;
2771}
2772
2773static noinline_for_stack int prealloc_file_extent_cluster(
2774                                struct btrfs_inode *inode,
2775                                struct file_extent_cluster *cluster)
2776{
2777        u64 alloc_hint = 0;
2778        u64 start;
2779        u64 end;
2780        u64 offset = inode->index_cnt;
2781        u64 num_bytes;
2782        int nr;
2783        int ret = 0;
2784        u64 prealloc_start = cluster->start - offset;
2785        u64 prealloc_end = cluster->end - offset;
2786        u64 cur_offset = prealloc_start;
2787
2788        BUG_ON(cluster->start != cluster->boundary[0]);
2789        ret = btrfs_alloc_data_chunk_ondemand(inode,
2790                                              prealloc_end + 1 - prealloc_start);
2791        if (ret)
2792                return ret;
2793
2794        /*
2795         * On a zoned filesystem, we cannot preallocate the file region.
2796         * Instead, we dirty and fiemap_write the region.
2797         */
2798        if (btrfs_is_zoned(inode->root->fs_info)) {
2799                struct btrfs_root *root = inode->root;
2800                struct btrfs_trans_handle *trans;
2801
2802                end = cluster->end - offset + 1;
2803                trans = btrfs_start_transaction(root, 1);
2804                if (IS_ERR(trans))
2805                        return PTR_ERR(trans);
2806
2807                inode->vfs_inode.i_ctime = current_time(&inode->vfs_inode);
2808                i_size_write(&inode->vfs_inode, end);
2809                ret = btrfs_update_inode(trans, root, inode);
2810                if (ret) {
2811                        btrfs_abort_transaction(trans, ret);
2812                        btrfs_end_transaction(trans);
2813                        return ret;
2814                }
2815
2816                return btrfs_end_transaction(trans);
2817        }
2818
2819        btrfs_inode_lock(&inode->vfs_inode, 0);
2820        for (nr = 0; nr < cluster->nr; nr++) {
2821                start = cluster->boundary[nr] - offset;
2822                if (nr + 1 < cluster->nr)
2823                        end = cluster->boundary[nr + 1] - 1 - offset;
2824                else
2825                        end = cluster->end - offset;
2826
2827                lock_extent(&inode->io_tree, start, end);
2828                num_bytes = end + 1 - start;
2829                ret = btrfs_prealloc_file_range(&inode->vfs_inode, 0, start,
2830                                                num_bytes, num_bytes,
2831                                                end + 1, &alloc_hint);
2832                cur_offset = end + 1;
2833                unlock_extent(&inode->io_tree, start, end);
2834                if (ret)
2835                        break;
2836        }
2837        btrfs_inode_unlock(&inode->vfs_inode, 0);
2838
2839        if (cur_offset < prealloc_end)
2840                btrfs_free_reserved_data_space_noquota(inode->root->fs_info,
2841                                               prealloc_end + 1 - cur_offset);
2842        return ret;
2843}
2844
2845static noinline_for_stack
2846int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
2847                         u64 block_start)
2848{
2849        struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
2850        struct extent_map *em;
2851        int ret = 0;
2852
2853        em = alloc_extent_map();
2854        if (!em)
2855                return -ENOMEM;
2856
2857        em->start = start;
2858        em->len = end + 1 - start;
2859        em->block_len = em->len;
2860        em->block_start = block_start;
2861        set_bit(EXTENT_FLAG_PINNED, &em->flags);
2862
2863        lock_extent(&BTRFS_I(inode)->io_tree, start, end);
2864        while (1) {
2865                write_lock(&em_tree->lock);
2866                ret = add_extent_mapping(em_tree, em, 0);
2867                write_unlock(&em_tree->lock);
2868                if (ret != -EEXIST) {
2869                        free_extent_map(em);
2870                        break;
2871                }
2872                btrfs_drop_extent_cache(BTRFS_I(inode), start, end, 0);
2873        }
2874        unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
2875        return ret;
2876}
2877
2878/*
2879 * Allow error injection to test balance/relocation cancellation
2880 */
2881noinline int btrfs_should_cancel_balance(struct btrfs_fs_info *fs_info)
2882{
2883        return atomic_read(&fs_info->balance_cancel_req) ||
2884                atomic_read(&fs_info->reloc_cancel_req) ||
2885                fatal_signal_pending(current);
2886}
2887ALLOW_ERROR_INJECTION(btrfs_should_cancel_balance, TRUE);
2888
2889static int relocate_file_extent_cluster(struct inode *inode,
2890                                        struct file_extent_cluster *cluster)
2891{
2892        struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
2893        u64 page_start;
2894        u64 page_end;
2895        u64 offset = BTRFS_I(inode)->index_cnt;
2896        unsigned long index;
2897        unsigned long last_index;
2898        struct page *page;
2899        struct file_ra_state *ra;
2900        gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
2901        int nr = 0;
2902        int ret = 0;
2903
2904        if (!cluster->nr)
2905                return 0;
2906
2907        ra = kzalloc(sizeof(*ra), GFP_NOFS);
2908        if (!ra)
2909                return -ENOMEM;
2910
2911        ret = prealloc_file_extent_cluster(BTRFS_I(inode), cluster);
2912        if (ret)
2913                goto out;
2914
2915        file_ra_state_init(ra, inode->i_mapping);
2916
2917        ret = setup_extent_mapping(inode, cluster->start - offset,
2918                                   cluster->end - offset, cluster->start);
2919        if (ret)
2920                goto out;
2921
2922        index = (cluster->start - offset) >> PAGE_SHIFT;
2923        last_index = (cluster->end - offset) >> PAGE_SHIFT;
2924        while (index <= last_index) {
2925                ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode),
2926                                PAGE_SIZE);
2927                if (ret)
2928                        goto out;
2929
2930                page = find_lock_page(inode->i_mapping, index);
2931                if (!page) {
2932                        page_cache_sync_readahead(inode->i_mapping,
2933                                                  ra, NULL, index,
2934                                                  last_index + 1 - index);
2935                        page = find_or_create_page(inode->i_mapping, index,
2936                                                   mask);
2937                        if (!page) {
2938                                btrfs_delalloc_release_metadata(BTRFS_I(inode),
2939                                                        PAGE_SIZE, true);
2940                                btrfs_delalloc_release_extents(BTRFS_I(inode),
2941                                                        PAGE_SIZE);
2942                                ret = -ENOMEM;
2943                                goto out;
2944                        }
2945                }
2946                ret = set_page_extent_mapped(page);
2947                if (ret < 0) {
2948                        btrfs_delalloc_release_metadata(BTRFS_I(inode),
2949                                                        PAGE_SIZE, true);
2950                        btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE);
2951                        unlock_page(page);
2952                        put_page(page);
2953                        goto out;
2954                }
2955
2956                if (PageReadahead(page)) {
2957                        page_cache_async_readahead(inode->i_mapping,
2958                                                   ra, NULL, page, index,
2959                                                   last_index + 1 - index);
2960                }
2961
2962                if (!PageUptodate(page)) {
2963                        btrfs_readpage(NULL, page);
2964                        lock_page(page);
2965                        if (!PageUptodate(page)) {
2966                                unlock_page(page);
2967                                put_page(page);
2968                                btrfs_delalloc_release_metadata(BTRFS_I(inode),
2969                                                        PAGE_SIZE, true);
2970                                btrfs_delalloc_release_extents(BTRFS_I(inode),
2971                                                               PAGE_SIZE);
2972                                ret = -EIO;
2973                                goto out;
2974                        }
2975                }
2976
2977                page_start = page_offset(page);
2978                page_end = page_start + PAGE_SIZE - 1;
2979
2980                lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end);
2981
2982                if (nr < cluster->nr &&
2983                    page_start + offset == cluster->boundary[nr]) {
2984                        set_extent_bits(&BTRFS_I(inode)->io_tree,
2985                                        page_start, page_end,
2986                                        EXTENT_BOUNDARY);
2987                        nr++;
2988                }
2989
2990                ret = btrfs_set_extent_delalloc(BTRFS_I(inode), page_start,
2991                                                page_end, 0, NULL);
2992                if (ret) {
2993                        unlock_page(page);
2994                        put_page(page);
2995                        btrfs_delalloc_release_metadata(BTRFS_I(inode),
2996                                                         PAGE_SIZE, true);
2997                        btrfs_delalloc_release_extents(BTRFS_I(inode),
2998                                                       PAGE_SIZE);
2999
3000                        clear_extent_bits(&BTRFS_I(inode)->io_tree,
3001                                          page_start, page_end,
3002                                          EXTENT_LOCKED | EXTENT_BOUNDARY);
3003                        goto out;
3004
3005                }
3006                set_page_dirty(page);
3007
3008                unlock_extent(&BTRFS_I(inode)->io_tree,
3009                              page_start, page_end);
3010                unlock_page(page);
3011                put_page(page);
3012
3013                index++;
3014                btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE);
3015                balance_dirty_pages_ratelimited(inode->i_mapping);
3016                btrfs_throttle(fs_info);
3017                if (btrfs_should_cancel_balance(fs_info)) {
3018                        ret = -ECANCELED;
3019                        goto out;
3020                }
3021        }
3022        WARN_ON(nr != cluster->nr);
3023        if (btrfs_is_zoned(fs_info) && !ret)
3024                ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
3025out:
3026        kfree(ra);
3027        return ret;
3028}
3029
3030static noinline_for_stack
3031int relocate_data_extent(struct inode *inode, struct btrfs_key *extent_key,
3032                         struct file_extent_cluster *cluster)
3033{
3034        int ret;
3035
3036        if (cluster->nr > 0 && extent_key->objectid != cluster->end + 1) {
3037                ret = relocate_file_extent_cluster(inode, cluster);
3038                if (ret)
3039                        return ret;
3040                cluster->nr = 0;
3041        }
3042
3043        if (!cluster->nr)
3044                cluster->start = extent_key->objectid;
3045        else
3046                BUG_ON(cluster->nr >= MAX_EXTENTS);
3047        cluster->end = extent_key->objectid + extent_key->offset - 1;
3048        cluster->boundary[cluster->nr] = extent_key->objectid;
3049        cluster->nr++;
3050
3051        if (cluster->nr >= MAX_EXTENTS) {
3052                ret = relocate_file_extent_cluster(inode, cluster);
3053                if (ret)
3054                        return ret;
3055                cluster->nr = 0;
3056        }
3057        return 0;
3058}
3059
3060/*
3061 * helper to add a tree block to the list.
3062 * the major work is getting the generation and level of the block
3063 */
3064static int add_tree_block(struct reloc_control *rc,
3065                          struct btrfs_key *extent_key,
3066                          struct btrfs_path *path,
3067                          struct rb_root *blocks)
3068{
3069        struct extent_buffer *eb;
3070        struct btrfs_extent_item *ei;
3071        struct btrfs_tree_block_info *bi;
3072        struct tree_block *block;
3073        struct rb_node *rb_node;
3074        u32 item_size;
3075        int level = -1;
3076        u64 generation;
3077        u64 owner = 0;
3078
3079        eb =  path->nodes[0];
3080        item_size = btrfs_item_size_nr(eb, path->slots[0]);
3081
3082        if (extent_key->type == BTRFS_METADATA_ITEM_KEY ||
3083            item_size >= sizeof(*ei) + sizeof(*bi)) {
3084                unsigned long ptr = 0, end;
3085
3086                ei = btrfs_item_ptr(eb, path->slots[0],
3087                                struct btrfs_extent_item);
3088                end = (unsigned long)ei + item_size;
3089                if (extent_key->type == BTRFS_EXTENT_ITEM_KEY) {
3090                        bi = (struct btrfs_tree_block_info *)(ei + 1);
3091                        level = btrfs_tree_block_level(eb, bi);
3092                        ptr = (unsigned long)(bi + 1);
3093                } else {
3094                        level = (int)extent_key->offset;
3095                        ptr = (unsigned long)(ei + 1);
3096                }
3097                generation = btrfs_extent_generation(eb, ei);
3098
3099                /*
3100                 * We're reading random blocks without knowing their owner ahead
3101                 * of time.  This is ok most of the time, as all reloc roots and
3102                 * fs roots have the same lock type.  However normal trees do
3103                 * not, and the only way to know ahead of time is to read the
3104                 * inline ref offset.  We know it's an fs root if
3105                 *
3106                 * 1. There's more than one ref.
3107                 * 2. There's a SHARED_DATA_REF_KEY set.
3108                 * 3. FULL_BACKREF is set on the flags.
3109                 *
3110                 * Otherwise it's safe to assume that the ref offset == the
3111                 * owner of this block, so we can use that when calling
3112                 * read_tree_block.
3113                 */
3114                if (btrfs_extent_refs(eb, ei) == 1 &&
3115                    !(btrfs_extent_flags(eb, ei) &
3116                      BTRFS_BLOCK_FLAG_FULL_BACKREF) &&
3117                    ptr < end) {
3118                        struct btrfs_extent_inline_ref *iref;
3119                        int type;
3120
3121                        iref = (struct btrfs_extent_inline_ref *)ptr;
3122                        type = btrfs_get_extent_inline_ref_type(eb, iref,
3123                                                        BTRFS_REF_TYPE_BLOCK);
3124                        if (type == BTRFS_REF_TYPE_INVALID)
3125                                return -EINVAL;
3126                        if (type == BTRFS_TREE_BLOCK_REF_KEY)
3127                                owner = btrfs_extent_inline_ref_offset(eb, iref);
3128                }
3129        } else if (unlikely(item_size == sizeof(struct btrfs_extent_item_v0))) {
3130                btrfs_print_v0_err(eb->fs_info);
3131                btrfs_handle_fs_error(eb->fs_info, -EINVAL, NULL);
3132                return -EINVAL;
3133        } else {
3134                BUG();
3135        }
3136
3137        btrfs_release_path(path);
3138
3139        BUG_ON(level == -1);
3140
3141        block = kmalloc(sizeof(*block), GFP_NOFS);
3142        if (!block)
3143                return -ENOMEM;
3144
3145        block->bytenr = extent_key->objectid;
3146        block->key.objectid = rc->extent_root->fs_info->nodesize;
3147        block->key.offset = generation;
3148        block->level = level;
3149        block->key_ready = 0;
3150        block->owner = owner;
3151
3152        rb_node = rb_simple_insert(blocks, block->bytenr, &block->rb_node);
3153        if (rb_node)
3154                btrfs_backref_panic(rc->extent_root->fs_info, block->bytenr,
3155                                    -EEXIST);
3156
3157        return 0;
3158}
3159
3160/*
3161 * helper to add tree blocks for backref of type BTRFS_SHARED_DATA_REF_KEY
3162 */
3163static int __add_tree_block(struct reloc_control *rc,
3164                            u64 bytenr, u32 blocksize,
3165                            struct rb_root *blocks)
3166{
3167        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
3168        struct btrfs_path *path;
3169        struct btrfs_key key;
3170        int ret;
3171        bool skinny = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
3172
3173        if (tree_block_processed(bytenr, rc))
3174                return 0;
3175
3176        if (rb_simple_search(blocks, bytenr))
3177                return 0;
3178
3179        path = btrfs_alloc_path();
3180        if (!path)
3181                return -ENOMEM;
3182again:
3183        key.objectid = bytenr;
3184        if (skinny) {
3185                key.type = BTRFS_METADATA_ITEM_KEY;
3186                key.offset = (u64)-1;
3187        } else {
3188                key.type = BTRFS_EXTENT_ITEM_KEY;
3189                key.offset = blocksize;
3190        }
3191
3192        path->search_commit_root = 1;
3193        path->skip_locking = 1;
3194        ret = btrfs_search_slot(NULL, rc->extent_root, &key, path, 0, 0);
3195        if (ret < 0)
3196                goto out;
3197
3198        if (ret > 0 && skinny) {
3199                if (path->slots[0]) {
3200                        path->slots[0]--;
3201                        btrfs_item_key_to_cpu(path->nodes[0], &key,
3202                                              path->slots[0]);
3203                        if (key.objectid == bytenr &&
3204                            (key.type == BTRFS_METADATA_ITEM_KEY ||
3205                             (key.type == BTRFS_EXTENT_ITEM_KEY &&
3206                              key.offset == blocksize)))
3207                                ret = 0;
3208                }
3209
3210                if (ret) {
3211                        skinny = false;
3212                        btrfs_release_path(path);
3213                        goto again;
3214                }
3215        }
3216        if (ret) {
3217                ASSERT(ret == 1);
3218                btrfs_print_leaf(path->nodes[0]);
3219                btrfs_err(fs_info,
3220             "tree block extent item (%llu) is not found in extent tree",
3221                     bytenr);
3222                WARN_ON(1);
3223                ret = -EINVAL;
3224                goto out;
3225        }
3226
3227        ret = add_tree_block(rc, &key, path, blocks);
3228out:
3229        btrfs_free_path(path);
3230        return ret;
3231}
3232
3233static int delete_block_group_cache(struct btrfs_fs_info *fs_info,
3234                                    struct btrfs_block_group *block_group,
3235                                    struct inode *inode,
3236                                    u64 ino)
3237{
3238        struct btrfs_root *root = fs_info->tree_root;
3239        struct btrfs_trans_handle *trans;
3240        int ret = 0;
3241
3242        if (inode)
3243                goto truncate;
3244
3245        inode = btrfs_iget(fs_info->sb, ino, root);
3246        if (IS_ERR(inode))
3247                return -ENOENT;
3248
3249truncate:
3250        ret = btrfs_check_trunc_cache_free_space(fs_info,
3251                                                 &fs_info->global_block_rsv);
3252        if (ret)
3253                goto out;
3254
3255        trans = btrfs_join_transaction(root);
3256        if (IS_ERR(trans)) {
3257                ret = PTR_ERR(trans);
3258                goto out;
3259        }
3260
3261        ret = btrfs_truncate_free_space_cache(trans, block_group, inode);
3262
3263        btrfs_end_transaction(trans);
3264        btrfs_btree_balance_dirty(fs_info);
3265out:
3266        iput(inode);
3267        return ret;
3268}
3269
3270/*
3271 * Locate the free space cache EXTENT_DATA in root tree leaf and delete the
3272 * cache inode, to avoid free space cache data extent blocking data relocation.
3273 */
3274static int delete_v1_space_cache(struct extent_buffer *leaf,
3275                                 struct btrfs_block_group *block_group,
3276                                 u64 data_bytenr)
3277{
3278        u64 space_cache_ino;
3279        struct btrfs_file_extent_item *ei;
3280        struct btrfs_key key;
3281        bool found = false;
3282        int i;
3283        int ret;
3284
3285        if (btrfs_header_owner(leaf) != BTRFS_ROOT_TREE_OBJECTID)
3286                return 0;
3287
3288        for (i = 0; i < btrfs_header_nritems(leaf); i++) {
3289                u8 type;
3290
3291                btrfs_item_key_to_cpu(leaf, &key, i);
3292                if (key.type != BTRFS_EXTENT_DATA_KEY)
3293                        continue;
3294                ei = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item);
3295                type = btrfs_file_extent_type(leaf, ei);
3296
3297                if ((type == BTRFS_FILE_EXTENT_REG ||
3298                     type == BTRFS_FILE_EXTENT_PREALLOC) &&
3299                    btrfs_file_extent_disk_bytenr(leaf, ei) == data_bytenr) {
3300                        found = true;
3301                        space_cache_ino = key.objectid;
3302                        break;
3303                }
3304        }
3305        if (!found)
3306                return -ENOENT;
3307        ret = delete_block_group_cache(leaf->fs_info, block_group, NULL,
3308                                        space_cache_ino);
3309        return ret;
3310}
3311
3312/*
3313 * helper to find all tree blocks that reference a given data extent
3314 */
3315static noinline_for_stack
3316int add_data_references(struct reloc_control *rc,
3317                        struct btrfs_key *extent_key,
3318                        struct btrfs_path *path,
3319                        struct rb_root *blocks)
3320{
3321        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
3322        struct ulist *leaves = NULL;
3323        struct ulist_iterator leaf_uiter;
3324        struct ulist_node *ref_node = NULL;
3325        const u32 blocksize = fs_info->nodesize;
3326        int ret = 0;
3327
3328        btrfs_release_path(path);
3329        ret = btrfs_find_all_leafs(NULL, fs_info, extent_key->objectid,
3330                                   0, &leaves, NULL, true);
3331        if (ret < 0)
3332                return ret;
3333
3334        ULIST_ITER_INIT(&leaf_uiter);
3335        while ((ref_node = ulist_next(leaves, &leaf_uiter))) {
3336                struct extent_buffer *eb;
3337
3338                eb = read_tree_block(fs_info, ref_node->val, 0, 0, 0, NULL);
3339                if (IS_ERR(eb)) {
3340                        ret = PTR_ERR(eb);
3341                        break;
3342                }
3343                ret = delete_v1_space_cache(eb, rc->block_group,
3344                                            extent_key->objectid);
3345                free_extent_buffer(eb);
3346                if (ret < 0)
3347                        break;
3348                ret = __add_tree_block(rc, ref_node->val, blocksize, blocks);
3349                if (ret < 0)
3350                        break;
3351        }
3352        if (ret < 0)
3353                free_block_list(blocks);
3354        ulist_free(leaves);
3355        return ret;
3356}
3357
3358/*
3359 * helper to find next unprocessed extent
3360 */
3361static noinline_for_stack
3362int find_next_extent(struct reloc_control *rc, struct btrfs_path *path,
3363                     struct btrfs_key *extent_key)
3364{
3365        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
3366        struct btrfs_key key;
3367        struct extent_buffer *leaf;
3368        u64 start, end, last;
3369        int ret;
3370
3371        last = rc->block_group->start + rc->block_group->length;
3372        while (1) {
3373                cond_resched();
3374                if (rc->search_start >= last) {
3375                        ret = 1;
3376                        break;
3377                }
3378
3379                key.objectid = rc->search_start;
3380                key.type = BTRFS_EXTENT_ITEM_KEY;
3381                key.offset = 0;
3382
3383                path->search_commit_root = 1;
3384                path->skip_locking = 1;
3385                ret = btrfs_search_slot(NULL, rc->extent_root, &key, path,
3386                                        0, 0);
3387                if (ret < 0)
3388                        break;
3389next:
3390                leaf = path->nodes[0];
3391                if (path->slots[0] >= btrfs_header_nritems(leaf)) {
3392                        ret = btrfs_next_leaf(rc->extent_root, path);
3393                        if (ret != 0)
3394                                break;
3395                        leaf = path->nodes[0];
3396                }
3397
3398                btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
3399                if (key.objectid >= last) {
3400                        ret = 1;
3401                        break;
3402                }
3403
3404                if (key.type != BTRFS_EXTENT_ITEM_KEY &&
3405                    key.type != BTRFS_METADATA_ITEM_KEY) {
3406                        path->slots[0]++;
3407                        goto next;
3408                }
3409
3410                if (key.type == BTRFS_EXTENT_ITEM_KEY &&
3411                    key.objectid + key.offset <= rc->search_start) {
3412                        path->slots[0]++;
3413                        goto next;
3414                }
3415
3416                if (key.type == BTRFS_METADATA_ITEM_KEY &&
3417                    key.objectid + fs_info->nodesize <=
3418                    rc->search_start) {
3419                        path->slots[0]++;
3420                        goto next;
3421                }
3422
3423                ret = find_first_extent_bit(&rc->processed_blocks,
3424                                            key.objectid, &start, &end,
3425                                            EXTENT_DIRTY, NULL);
3426
3427                if (ret == 0 && start <= key.objectid) {
3428                        btrfs_release_path(path);
3429                        rc->search_start = end + 1;
3430                } else {
3431                        if (key.type == BTRFS_EXTENT_ITEM_KEY)
3432                                rc->search_start = key.objectid + key.offset;
3433                        else
3434                                rc->search_start = key.objectid +
3435                                        fs_info->nodesize;
3436                        memcpy(extent_key, &key, sizeof(key));
3437                        return 0;
3438                }
3439        }
3440        btrfs_release_path(path);
3441        return ret;
3442}
3443
3444static void set_reloc_control(struct reloc_control *rc)
3445{
3446        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
3447
3448        mutex_lock(&fs_info->reloc_mutex);
3449        fs_info->reloc_ctl = rc;
3450        mutex_unlock(&fs_info->reloc_mutex);
3451}
3452
3453static void unset_reloc_control(struct reloc_control *rc)
3454{
3455        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
3456
3457        mutex_lock(&fs_info->reloc_mutex);
3458        fs_info->reloc_ctl = NULL;
3459        mutex_unlock(&fs_info->reloc_mutex);
3460}
3461
3462static noinline_for_stack
3463int prepare_to_relocate(struct reloc_control *rc)
3464{
3465        struct btrfs_trans_handle *trans;
3466        int ret;
3467
3468        rc->block_rsv = btrfs_alloc_block_rsv(rc->extent_root->fs_info,
3469                                              BTRFS_BLOCK_RSV_TEMP);
3470        if (!rc->block_rsv)
3471                return -ENOMEM;
3472
3473        memset(&rc->cluster, 0, sizeof(rc->cluster));
3474        rc->search_start = rc->block_group->start;
3475        rc->extents_found = 0;
3476        rc->nodes_relocated = 0;
3477        rc->merging_rsv_size = 0;
3478        rc->reserved_bytes = 0;
3479        rc->block_rsv->size = rc->extent_root->fs_info->nodesize *
3480                              RELOCATION_RESERVED_NODES;
3481        ret = btrfs_block_rsv_refill(rc->extent_root,
3482                                     rc->block_rsv, rc->block_rsv->size,
3483                                     BTRFS_RESERVE_FLUSH_ALL);
3484        if (ret)
3485                return ret;
3486
3487        rc->create_reloc_tree = 1;
3488        set_reloc_control(rc);
3489
3490        trans = btrfs_join_transaction(rc->extent_root);
3491        if (IS_ERR(trans)) {
3492                unset_reloc_control(rc);
3493                /*
3494                 * extent tree is not a ref_cow tree and has no reloc_root to
3495                 * cleanup.  And callers are responsible to free the above
3496                 * block rsv.
3497                 */
3498                return PTR_ERR(trans);
3499        }
3500        return btrfs_commit_transaction(trans);
3501}
3502
3503static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
3504{
3505        struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
3506        struct rb_root blocks = RB_ROOT;
3507        struct btrfs_key key;
3508        struct btrfs_trans_handle *trans = NULL;
3509        struct btrfs_path *path;
3510        struct btrfs_extent_item *ei;
3511        u64 flags;
3512        int ret;
3513        int err = 0;
3514        int progress = 0;
3515
3516        path = btrfs_alloc_path();
3517        if (!path)
3518                return -ENOMEM;
3519        path->reada = READA_FORWARD;
3520
3521        ret = prepare_to_relocate(rc);
3522        if (ret) {
3523                err = ret;
3524                goto out_free;
3525        }
3526
3527        while (1) {
3528                rc->reserved_bytes = 0;
3529                ret = btrfs_block_rsv_refill(rc->extent_root,
3530                                        rc->block_rsv, rc->block_rsv->size,
3531                                        BTRFS_RESERVE_FLUSH_ALL);
3532                if (ret) {
3533                        err = ret;
3534                        break;
3535                }
3536                progress++;
3537                trans = btrfs_start_transaction(rc->extent_root, 0);
3538                if (IS_ERR(trans)) {
3539                        err = PTR_ERR(trans);
3540                        trans = NULL;
3541                        break;
3542                }
3543restart:
3544                if (update_backref_cache(trans, &rc->backref_cache)) {
3545                        btrfs_end_transaction(trans);
3546                        trans = NULL;
3547                        continue;
3548                }
3549
3550                ret = find_next_extent(rc, path, &key);
3551                if (ret < 0)
3552                        err = ret;
3553                if (ret != 0)
3554                        break;
3555
3556                rc->extents_found++;
3557
3558                ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
3559                                    struct btrfs_extent_item);
3560                flags = btrfs_extent_flags(path->nodes[0], ei);
3561
3562                if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
3563                        ret = add_tree_block(rc, &key, path, &blocks);
3564                } else if (rc->stage == UPDATE_DATA_PTRS &&
3565                           (flags & BTRFS_EXTENT_FLAG_DATA)) {
3566                        ret = add_data_references(rc, &key, path, &blocks);
3567                } else {
3568                        btrfs_release_path(path);
3569                        ret = 0;
3570                }
3571                if (ret < 0) {
3572                        err = ret;
3573                        break;
3574                }
3575
3576                if (!RB_EMPTY_ROOT(&blocks)) {
3577                        ret = relocate_tree_blocks(trans, rc, &blocks);
3578                        if (ret < 0) {
3579                                if (ret != -EAGAIN) {
3580                                        err = ret;
3581                                        break;
3582                                }
3583                                rc->extents_found--;
3584                                rc->search_start = key.objectid;
3585                        }
3586                }
3587
3588                btrfs_end_transaction_throttle(trans);
3589                btrfs_btree_balance_dirty(fs_info);
3590                trans = NULL;
3591
3592                if (rc->stage == MOVE_DATA_EXTENTS &&
3593                    (flags & BTRFS_EXTENT_FLAG_DATA)) {
3594                        rc->found_file_extent = 1;
3595                        ret = relocate_data_extent(rc->data_inode,
3596                                                   &key, &rc->cluster);
3597                        if (ret < 0) {
3598                                err = ret;
3599                                break;
3600                        }
3601                }
3602                if (btrfs_should_cancel_balance(fs_info)) {
3603                        err = -ECANCELED;
3604                        break;
3605                }
3606        }
3607        if (trans && progress && err == -ENOSPC) {
3608                ret = btrfs_force_chunk_alloc(trans, rc->block_group->flags);
3609                if (ret == 1) {
3610                        err = 0;
3611                        progress = 0;
3612                        goto restart;
3613                }
3614        }
3615
3616        btrfs_release_path(path);
3617        clear_extent_bits(&rc->processed_blocks, 0, (u64)-1, EXTENT_DIRTY);
3618
3619        if (trans) {
3620                btrfs_end_transaction_throttle(trans);
3621                btrfs_btree_balance_dirty(fs_info);
3622        }
3623
3624        if (!err) {
3625                ret = relocate_file_extent_cluster(rc->data_inode,
3626                                                   &rc->cluster);
3627                if (ret < 0)
3628                        err = ret;
3629        }
3630
3631        rc->create_reloc_tree = 0;
3632        set_reloc_control(rc);
3633
3634        btrfs_backref_release_cache(&rc->backref_cache);
3635        btrfs_block_rsv_release(fs_info, rc->block_rsv, (u64)-1, NULL);
3636
3637        /*
3638         * Even in the case when the relocation is cancelled, we should all go
3639         * through prepare_to_merge() and merge_reloc_roots().
3640         *
3641         * For error (including cancelled balance), prepare_to_merge() will
3642         * mark all reloc trees orphan, then queue them for cleanup in
3643         * merge_reloc_roots()
3644         */
3645        err = prepare_to_merge(rc, err);
3646
3647        merge_reloc_roots(rc);
3648
3649        rc->merge_reloc_tree = 0;
3650        unset_reloc_control(rc);
3651        btrfs_block_rsv_release(fs_info, rc->block_rsv, (u64)-1, NULL);
3652
3653        /* get rid of pinned extents */
3654        trans = btrfs_join_transaction(rc->extent_root);
3655        if (IS_ERR(trans)) {
3656                err = PTR_ERR(trans);
3657                goto out_free;
3658        }
3659        ret = btrfs_commit_transaction(trans);
3660        if (ret && !err)
3661                err = ret;
3662out_free:
3663        ret = clean_dirty_subvols(rc);
3664        if (ret < 0 && !err)
3665                err = ret;
3666        btrfs_free_block_rsv(fs_info, rc->block_rsv);
3667        btrfs_free_path(path);
3668        return err;
3669}
3670
3671static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
3672                                 struct btrfs_root *root, u64 objectid)
3673{
3674        struct btrfs_path *path;
3675        struct btrfs_inode_item *item;
3676        struct extent_buffer *leaf;
3677        u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC;
3678        int ret;
3679
3680        if (btrfs_is_zoned(trans->fs_info))
3681                flags &= ~BTRFS_INODE_PREALLOC;
3682
3683        path = btrfs_alloc_path();
3684        if (!path)
3685                return -ENOMEM;
3686
3687        ret = btrfs_insert_empty_inode(trans, root, path, objectid);
3688        if (ret)
3689                goto out;
3690
3691        leaf = path->nodes[0];
3692        item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item);
3693        memzero_extent_buffer(leaf, (unsigned long)item, sizeof(*item));
3694        btrfs_set_inode_generation(leaf, item, 1);
3695        btrfs_set_inode_size(leaf, item, 0);
3696        btrfs_set_inode_mode(leaf, item, S_IFREG | 0600);
3697        btrfs_set_inode_flags(leaf, item, flags);
3698        btrfs_mark_buffer_dirty(leaf);
3699out:
3700        btrfs_free_path(path);
3701        return ret;
3702}
3703
3704static void delete_orphan_inode(struct btrfs_trans_handle *trans,
3705                                struct btrfs_root *root, u64 objectid)
3706{
3707        struct btrfs_path *path;
3708        struct btrfs_key key;
3709        int ret = 0;
3710
3711        path = btrfs_alloc_path();
3712        if (!path) {
3713                ret = -ENOMEM;
3714                goto out;
3715        }
3716
3717        key.objectid = objectid;
3718        key.type = BTRFS_INODE_ITEM_KEY;
3719        key.offset = 0;
3720        ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
3721        if (ret) {
3722                if (ret > 0)
3723                        ret = -ENOENT;
3724                goto out;
3725        }
3726        ret = btrfs_del_item(trans, root, path);
3727out:
3728        if (ret)
3729                btrfs_abort_transaction(trans, ret);
3730        btrfs_free_path(path);
3731}
3732
3733/*
3734 * helper to create inode for data relocation.
3735 * the inode is in data relocation tree and its link count is 0
3736 */
3737static noinline_for_stack
3738struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info,
3739                                 struct btrfs_block_group *group)
3740{
3741        struct inode *inode = NULL;
3742        struct btrfs_trans_handle *trans;
3743        struct btrfs_root *root;
3744        u64 objectid;
3745        int err = 0;
3746
3747        root = btrfs_grab_root(fs_info->data_reloc_root);
3748        trans = btrfs_start_transaction(root, 6);
3749        if (IS_ERR(trans)) {
3750                btrfs_put_root(root);
3751                return ERR_CAST(trans);
3752        }
3753
3754        err = btrfs_get_free_objectid(root, &objectid);
3755        if (err)
3756                goto out;
3757
3758        err = __insert_orphan_inode(trans, root, objectid);
3759        if (err)
3760                goto out;
3761
3762        inode = btrfs_iget(fs_info->sb, objectid, root);
3763        if (IS_ERR(inode)) {
3764                delete_orphan_inode(trans, root, objectid);
3765                err = PTR_ERR(inode);
3766                inode = NULL;
3767                goto out;
3768        }
3769        BTRFS_I(inode)->index_cnt = group->start;
3770
3771        err = btrfs_orphan_add(trans, BTRFS_I(inode));
3772out:
3773        btrfs_put_root(root);
3774        btrfs_end_transaction(trans);
3775        btrfs_btree_balance_dirty(fs_info);
3776        if (err) {
3777                if (inode)
3778                        iput(inode);
3779                inode = ERR_PTR(err);
3780        }
3781        return inode;
3782}
3783
3784/*
3785 * Mark start of chunk relocation that is cancellable. Check if the cancellation
3786 * has been requested meanwhile and don't start in that case.
3787 *
3788 * Return:
3789 *   0             success
3790 *   -EINPROGRESS  operation is already in progress, that's probably a bug
3791 *   -ECANCELED    cancellation request was set before the operation started
3792 *   -EAGAIN       can not start because there are ongoing send operations
3793 */
3794static int reloc_chunk_start(struct btrfs_fs_info *fs_info)
3795{
3796        spin_lock(&fs_info->send_reloc_lock);
3797        if (fs_info->send_in_progress) {
3798                btrfs_warn_rl(fs_info,
3799"cannot run relocation while send operations are in progress (%d in progress)",
3800                              fs_info->send_in_progress);
3801                spin_unlock(&fs_info->send_reloc_lock);
3802                return -EAGAIN;
3803        }
3804        if (test_and_set_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags)) {
3805                /* This should not happen */
3806                spin_unlock(&fs_info->send_reloc_lock);
3807                btrfs_err(fs_info, "reloc already running, cannot start");
3808                return -EINPROGRESS;
3809        }
3810        spin_unlock(&fs_info->send_reloc_lock);
3811
3812        if (atomic_read(&fs_info->reloc_cancel_req) > 0) {
3813                btrfs_info(fs_info, "chunk relocation canceled on start");
3814                /*
3815                 * On cancel, clear all requests but let the caller mark
3816                 * the end after cleanup operations.
3817                 */
3818                atomic_set(&fs_info->reloc_cancel_req, 0);
3819                return -ECANCELED;
3820        }
3821        return 0;
3822}
3823
3824/*
3825 * Mark end of chunk relocation that is cancellable and wake any waiters.
3826 */
3827static void reloc_chunk_end(struct btrfs_fs_info *fs_info)
3828{
3829        /* Requested after start, clear bit first so any waiters can continue */
3830        if (atomic_read(&fs_info->reloc_cancel_req) > 0)
3831                btrfs_info(fs_info, "chunk relocation canceled during operation");
3832        spin_lock(&fs_info->send_reloc_lock);
3833        clear_and_wake_up_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags);
3834        spin_unlock(&fs_info->send_reloc_lock);
3835        atomic_set(&fs_info->reloc_cancel_req, 0);
3836}
3837
3838static struct reloc_control *alloc_reloc_control(struct btrfs_fs_info *fs_info)
3839{
3840        struct reloc_control *rc;
3841
3842        rc = kzalloc(sizeof(*rc), GFP_NOFS);
3843        if (!rc)
3844                return NULL;
3845
3846        INIT_LIST_HEAD(&rc->reloc_roots);
3847        INIT_LIST_HEAD(&rc->dirty_subvol_roots);
3848        btrfs_backref_init_cache(fs_info, &rc->backref_cache, 1);
3849        mapping_tree_init(&rc->reloc_root_tree);
3850        extent_io_tree_init(fs_info, &rc->processed_blocks,
3851                            IO_TREE_RELOC_BLOCKS, NULL);
3852        return rc;
3853}
3854
3855static void free_reloc_control(struct reloc_control *rc)
3856{
3857        struct mapping_node *node, *tmp;
3858
3859        free_reloc_roots(&rc->reloc_roots);
3860        rbtree_postorder_for_each_entry_safe(node, tmp,
3861                        &rc->reloc_root_tree.rb_root, rb_node)
3862                kfree(node);
3863
3864        kfree(rc);
3865}
3866
3867/*
3868 * Print the block group being relocated
3869 */
3870static void describe_relocation(struct btrfs_fs_info *fs_info,
3871                                struct btrfs_block_group *block_group)
3872{
3873        char buf[128] = {'\0'};
3874
3875        btrfs_describe_block_groups(block_group->flags, buf, sizeof(buf));
3876
3877        btrfs_info(fs_info,
3878                   "relocating block group %llu flags %s",
3879                   block_group->start, buf);
3880}
3881
3882static const char *stage_to_string(int stage)
3883{
3884        if (stage == MOVE_DATA_EXTENTS)
3885                return "move data extents";
3886        if (stage == UPDATE_DATA_PTRS)
3887                return "update data pointers";
3888        return "unknown";
3889}
3890
3891/*
3892 * function to relocate all extents in a block group.
3893 */
3894int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start)
3895{
3896        struct btrfs_block_group *bg;
3897        struct btrfs_root *extent_root = fs_info->extent_root;
3898        struct reloc_control *rc;
3899        struct inode *inode;
3900        struct btrfs_path *path;
3901        int ret;
3902        int rw = 0;
3903        int err = 0;
3904
3905        bg = btrfs_lookup_block_group(fs_info, group_start);
3906        if (!bg)
3907                return -ENOENT;
3908
3909        if (btrfs_pinned_by_swapfile(fs_info, bg)) {
3910                btrfs_put_block_group(bg);
3911                return -ETXTBSY;
3912        }
3913
3914        rc = alloc_reloc_control(fs_info);
3915        if (!rc) {
3916                btrfs_put_block_group(bg);
3917                return -ENOMEM;
3918        }
3919
3920        ret = reloc_chunk_start(fs_info);
3921        if (ret < 0) {
3922                err = ret;
3923                goto out_put_bg;
3924        }
3925
3926        rc->extent_root = extent_root;
3927        rc->block_group = bg;
3928
3929        ret = btrfs_inc_block_group_ro(rc->block_group, true);
3930        if (ret) {
3931                err = ret;
3932                goto out;
3933        }
3934        rw = 1;
3935
3936        path = btrfs_alloc_path();
3937        if (!path) {
3938                err = -ENOMEM;
3939                goto out;
3940        }
3941
3942        inode = lookup_free_space_inode(rc->block_group, path);
3943        btrfs_free_path(path);
3944
3945        if (!IS_ERR(inode))
3946                ret = delete_block_group_cache(fs_info, rc->block_group, inode, 0);
3947        else
3948                ret = PTR_ERR(inode);
3949
3950        if (ret && ret != -ENOENT) {
3951                err = ret;
3952                goto out;
3953        }
3954
3955        rc->data_inode = create_reloc_inode(fs_info, rc->block_group);
3956        if (IS_ERR(rc->data_inode)) {
3957                err = PTR_ERR(rc->data_inode);
3958                rc->data_inode = NULL;
3959                goto out;
3960        }
3961
3962        describe_relocation(fs_info, rc->block_group);
3963
3964        btrfs_wait_block_group_reservations(rc->block_group);
3965        btrfs_wait_nocow_writers(rc->block_group);
3966        btrfs_wait_ordered_roots(fs_info, U64_MAX,
3967                                 rc->block_group->start,
3968                                 rc->block_group->length);
3969
3970        while (1) {
3971                int finishes_stage;
3972
3973                mutex_lock(&fs_info->cleaner_mutex);
3974                ret = relocate_block_group(rc);
3975                mutex_unlock(&fs_info->cleaner_mutex);
3976                if (ret < 0)
3977                        err = ret;
3978
3979                finishes_stage = rc->stage;
3980                /*
3981                 * We may have gotten ENOSPC after we already dirtied some
3982                 * extents.  If writeout happens while we're relocating a
3983                 * different block group we could end up hitting the
3984                 * BUG_ON(rc->stage == UPDATE_DATA_PTRS) in
3985                 * btrfs_reloc_cow_block.  Make sure we write everything out
3986                 * properly so we don't trip over this problem, and then break
3987                 * out of the loop if we hit an error.
3988                 */
3989                if (rc->stage == MOVE_DATA_EXTENTS && rc->found_file_extent) {
3990                        ret = btrfs_wait_ordered_range(rc->data_inode, 0,
3991                                                       (u64)-1);
3992                        if (ret)
3993                                err = ret;
3994                        invalidate_mapping_pages(rc->data_inode->i_mapping,
3995                                                 0, -1);
3996                        rc->stage = UPDATE_DATA_PTRS;
3997                }
3998
3999                if (err < 0)
4000                        goto out;
4001
4002                if (rc->extents_found == 0)
4003                        break;
4004
4005                btrfs_info(fs_info, "found %llu extents, stage: %s",
4006                           rc->extents_found, stage_to_string(finishes_stage));
4007        }
4008
4009        WARN_ON(rc->block_group->pinned > 0);
4010        WARN_ON(rc->block_group->reserved > 0);
4011        WARN_ON(rc->block_group->used > 0);
4012out:
4013        if (err && rw)
4014                btrfs_dec_block_group_ro(rc->block_group);
4015        iput(rc->data_inode);
4016out_put_bg:
4017        btrfs_put_block_group(bg);
4018        reloc_chunk_end(fs_info);
4019        free_reloc_control(rc);
4020        return err;
4021}
4022
4023static noinline_for_stack int mark_garbage_root(struct btrfs_root *root)
4024{
4025        struct btrfs_fs_info *fs_info = root->fs_info;
4026        struct btrfs_trans_handle *trans;
4027        int ret, err;
4028
4029        trans = btrfs_start_transaction(fs_info->tree_root, 0);
4030        if (IS_ERR(trans))
4031                return PTR_ERR(trans);
4032
4033        memset(&root->root_item.drop_progress, 0,
4034                sizeof(root->root_item.drop_progress));
4035        btrfs_set_root_drop_level(&root->root_item, 0);
4036        btrfs_set_root_refs(&root->root_item, 0);
4037        ret = btrfs_update_root(trans, fs_info->tree_root,
4038                                &root->root_key, &root->root_item);
4039
4040        err = btrfs_end_transaction(trans);
4041        if (err)
4042                return err;
4043        return ret;
4044}
4045
4046/*
4047 * recover relocation interrupted by system crash.
4048 *
4049 * this function resumes merging reloc trees with corresponding fs trees.
4050 * this is important for keeping the sharing of tree blocks
4051 */
4052int btrfs_recover_relocation(struct btrfs_root *root)
4053{
4054        struct btrfs_fs_info *fs_info = root->fs_info;
4055        LIST_HEAD(reloc_roots);
4056        struct btrfs_key key;
4057        struct btrfs_root *fs_root;
4058        struct btrfs_root *reloc_root;
4059        struct btrfs_path *path;
4060        struct extent_buffer *leaf;
4061        struct reloc_control *rc = NULL;
4062        struct btrfs_trans_handle *trans;
4063        int ret;
4064        int err = 0;
4065
4066        path = btrfs_alloc_path();
4067        if (!path)
4068                return -ENOMEM;
4069        path->reada = READA_BACK;
4070
4071        key.objectid = BTRFS_TREE_RELOC_OBJECTID;
4072        key.type = BTRFS_ROOT_ITEM_KEY;
4073        key.offset = (u64)-1;
4074
4075        while (1) {
4076                ret = btrfs_search_slot(NULL, fs_info->tree_root, &key,
4077                                        path, 0, 0);
4078                if (ret < 0) {
4079                        err = ret;
4080                        goto out;
4081                }
4082                if (ret > 0) {
4083                        if (path->slots[0] == 0)
4084                                break;
4085                        path->slots[0]--;
4086                }
4087                leaf = path->nodes[0];
4088                btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
4089                btrfs_release_path(path);
4090
4091                if (key.objectid != BTRFS_TREE_RELOC_OBJECTID ||
4092                    key.type != BTRFS_ROOT_ITEM_KEY)
4093                        break;
4094
4095                reloc_root = btrfs_read_tree_root(root, &key);
4096                if (IS_ERR(reloc_root)) {
4097                        err = PTR_ERR(reloc_root);
4098                        goto out;
4099                }
4100
4101                set_bit(BTRFS_ROOT_SHAREABLE, &reloc_root->state);
4102                list_add(&reloc_root->root_list, &reloc_roots);
4103
4104                if (btrfs_root_refs(&reloc_root->root_item) > 0) {
4105                        fs_root = btrfs_get_fs_root(fs_info,
4106                                        reloc_root->root_key.offset, false);
4107                        if (IS_ERR(fs_root)) {
4108                                ret = PTR_ERR(fs_root);
4109                                if (ret != -ENOENT) {
4110                                        err = ret;
4111                                        goto out;
4112                                }
4113                                ret = mark_garbage_root(reloc_root);
4114                                if (ret < 0) {
4115                                        err = ret;
4116                                        goto out;
4117                                }
4118                        } else {
4119                                btrfs_put_root(fs_root);
4120                        }
4121                }
4122
4123                if (key.offset == 0)
4124                        break;
4125
4126                key.offset--;
4127        }
4128        btrfs_release_path(path);
4129
4130        if (list_empty(&reloc_roots))
4131                goto out;
4132
4133        rc = alloc_reloc_control(fs_info);
4134        if (!rc) {
4135                err = -ENOMEM;
4136                goto out;
4137        }
4138
4139        ret = reloc_chunk_start(fs_info);
4140        if (ret < 0) {
4141                err = ret;
4142                goto out_end;
4143        }
4144
4145        rc->extent_root = fs_info->extent_root;
4146
4147        set_reloc_control(rc);
4148
4149        trans = btrfs_join_transaction(rc->extent_root);
4150        if (IS_ERR(trans)) {
4151                err = PTR_ERR(trans);
4152                goto out_unset;
4153        }
4154
4155        rc->merge_reloc_tree = 1;
4156
4157        while (!list_empty(&reloc_roots)) {
4158                reloc_root = list_entry(reloc_roots.next,
4159                                        struct btrfs_root, root_list);
4160                list_del(&reloc_root->root_list);
4161
4162                if (btrfs_root_refs(&reloc_root->root_item) == 0) {
4163                        list_add_tail(&reloc_root->root_list,
4164                                      &rc->reloc_roots);
4165                        continue;
4166                }
4167
4168                fs_root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset,
4169                                            false);
4170                if (IS_ERR(fs_root)) {
4171                        err = PTR_ERR(fs_root);
4172                        list_add_tail(&reloc_root->root_list, &reloc_roots);
4173                        btrfs_end_transaction(trans);
4174                        goto out_unset;
4175                }
4176
4177                err = __add_reloc_root(reloc_root);
4178                ASSERT(err != -EEXIST);
4179                if (err) {
4180                        list_add_tail(&reloc_root->root_list, &reloc_roots);
4181                        btrfs_put_root(fs_root);
4182                        btrfs_end_transaction(trans);
4183                        goto out_unset;
4184                }
4185                fs_root->reloc_root = btrfs_grab_root(reloc_root);
4186                btrfs_put_root(fs_root);
4187        }
4188
4189        err = btrfs_commit_transaction(trans);
4190        if (err)
4191                goto out_unset;
4192
4193        merge_reloc_roots(rc);
4194
4195        unset_reloc_control(rc);
4196
4197        trans = btrfs_join_transaction(rc->extent_root);
4198        if (IS_ERR(trans)) {
4199                err = PTR_ERR(trans);
4200                goto out_clean;
4201        }
4202        err = btrfs_commit_transaction(trans);
4203out_clean:
4204        ret = clean_dirty_subvols(rc);
4205        if (ret < 0 && !err)
4206                err = ret;
4207out_unset:
4208        unset_reloc_control(rc);
4209out_end:
4210        reloc_chunk_end(fs_info);
4211        free_reloc_control(rc);
4212out:
4213        free_reloc_roots(&reloc_roots);
4214
4215        btrfs_free_path(path);
4216
4217        if (err == 0) {
4218                /* cleanup orphan inode in data relocation tree */
4219                fs_root = btrfs_grab_root(fs_info->data_reloc_root);
4220                ASSERT(fs_root);
4221                err = btrfs_orphan_cleanup(fs_root);
4222                btrfs_put_root(fs_root);
4223        }
4224        return err;
4225}
4226
4227/*
4228 * helper to add ordered checksum for data relocation.
4229 *
4230 * cloning checksum properly handles the nodatasum extents.
4231 * it also saves CPU time to re-calculate the checksum.
4232 */
4233int btrfs_reloc_clone_csums(struct btrfs_inode *inode, u64 file_pos, u64 len)
4234{
4235        struct btrfs_fs_info *fs_info = inode->root->fs_info;
4236        struct btrfs_ordered_sum *sums;
4237        struct btrfs_ordered_extent *ordered;
4238        int ret;
4239        u64 disk_bytenr;
4240        u64 new_bytenr;
4241        LIST_HEAD(list);
4242
4243        ordered = btrfs_lookup_ordered_extent(inode, file_pos);
4244        BUG_ON(ordered->file_offset != file_pos || ordered->num_bytes != len);
4245
4246        disk_bytenr = file_pos + inode->index_cnt;
4247        ret = btrfs_lookup_csums_range(fs_info->csum_root, disk_bytenr,
4248                                       disk_bytenr + len - 1, &list, 0);
4249        if (ret)
4250                goto out;
4251
4252        while (!list_empty(&list)) {
4253                sums = list_entry(list.next, struct btrfs_ordered_sum, list);
4254                list_del_init(&sums->list);
4255
4256                /*
4257                 * We need to offset the new_bytenr based on where the csum is.
4258                 * We need to do this because we will read in entire prealloc
4259                 * extents but we may have written to say the middle of the
4260                 * prealloc extent, so we need to make sure the csum goes with
4261                 * the right disk offset.
4262                 *
4263                 * We can do this because the data reloc inode refers strictly
4264                 * to the on disk bytes, so we don't have to worry about
4265                 * disk_len vs real len like with real inodes since it's all
4266                 * disk length.
4267                 */
4268                new_bytenr = ordered->disk_bytenr + sums->bytenr - disk_bytenr;
4269                sums->bytenr = new_bytenr;
4270
4271                btrfs_add_ordered_sum(ordered, sums);
4272        }
4273out:
4274        btrfs_put_ordered_extent(ordered);
4275        return ret;
4276}
4277
4278int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
4279                          struct btrfs_root *root, struct extent_buffer *buf,
4280                          struct extent_buffer *cow)
4281{
4282        struct btrfs_fs_info *fs_info = root->fs_info;
4283        struct reloc_control *rc;
4284        struct btrfs_backref_node *node;
4285        int first_cow = 0;
4286        int level;
4287        int ret = 0;
4288
4289        rc = fs_info->reloc_ctl;
4290        if (!rc)
4291                return 0;
4292
4293        BUG_ON(rc->stage == UPDATE_DATA_PTRS &&
4294               root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID);
4295
4296        level = btrfs_header_level(buf);
4297        if (btrfs_header_generation(buf) <=
4298            btrfs_root_last_snapshot(&root->root_item))
4299                first_cow = 1;
4300
4301        if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID &&
4302            rc->create_reloc_tree) {
4303                WARN_ON(!first_cow && level == 0);
4304
4305                node = rc->backref_cache.path[level];
4306                BUG_ON(node->bytenr != buf->start &&
4307                       node->new_bytenr != buf->start);
4308
4309                btrfs_backref_drop_node_buffer(node);
4310                atomic_inc(&cow->refs);
4311                node->eb = cow;
4312                node->new_bytenr = cow->start;
4313
4314                if (!node->pending) {
4315                        list_move_tail(&node->list,
4316                                       &rc->backref_cache.pending[level]);
4317                        node->pending = 1;
4318                }
4319
4320                if (first_cow)
4321                        mark_block_processed(rc, node);
4322
4323                if (first_cow && level > 0)
4324                        rc->nodes_relocated += buf->len;
4325        }
4326
4327        if (level == 0 && first_cow && rc->stage == UPDATE_DATA_PTRS)
4328                ret = replace_file_extents(trans, rc, root, cow);
4329        return ret;
4330}
4331
4332/*
4333 * called before creating snapshot. it calculates metadata reservation
4334 * required for relocating tree blocks in the snapshot
4335 */
4336void btrfs_reloc_pre_snapshot(struct btrfs_pending_snapshot *pending,
4337                              u64 *bytes_to_reserve)
4338{
4339        struct btrfs_root *root = pending->root;
4340        struct reloc_control *rc = root->fs_info->reloc_ctl;
4341
4342        if (!rc || !have_reloc_root(root))
4343                return;
4344
4345        if (!rc->merge_reloc_tree)
4346                return;
4347
4348        root = root->reloc_root;
4349        BUG_ON(btrfs_root_refs(&root->root_item) == 0);
4350        /*
4351         * relocation is in the stage of merging trees. the space
4352         * used by merging a reloc tree is twice the size of
4353         * relocated tree nodes in the worst case. half for cowing
4354         * the reloc tree, half for cowing the fs tree. the space
4355         * used by cowing the reloc tree will be freed after the
4356         * tree is dropped. if we create snapshot, cowing the fs
4357         * tree may use more space than it frees. so we need
4358         * reserve extra space.
4359         */
4360        *bytes_to_reserve += rc->nodes_relocated;
4361}
4362
4363/*
4364 * called after snapshot is created. migrate block reservation
4365 * and create reloc root for the newly created snapshot
4366 *
4367 * This is similar to btrfs_init_reloc_root(), we come out of here with two
4368 * references held on the reloc_root, one for root->reloc_root and one for
4369 * rc->reloc_roots.
4370 */
4371int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
4372                               struct btrfs_pending_snapshot *pending)
4373{
4374        struct btrfs_root *root = pending->root;
4375        struct btrfs_root *reloc_root;
4376        struct btrfs_root *new_root;
4377        struct reloc_control *rc = root->fs_info->reloc_ctl;
4378        int ret;
4379
4380        if (!rc || !have_reloc_root(root))
4381                return 0;
4382
4383        rc = root->fs_info->reloc_ctl;
4384        rc->merging_rsv_size += rc->nodes_relocated;
4385
4386        if (rc->merge_reloc_tree) {
4387                ret = btrfs_block_rsv_migrate(&pending->block_rsv,
4388                                              rc->block_rsv,
4389                                              rc->nodes_relocated, true);
4390                if (ret)
4391                        return ret;
4392        }
4393
4394        new_root = pending->snap;
4395        reloc_root = create_reloc_root(trans, root->reloc_root,
4396                                       new_root->root_key.objectid);
4397        if (IS_ERR(reloc_root))
4398                return PTR_ERR(reloc_root);
4399
4400        ret = __add_reloc_root(reloc_root);
4401        ASSERT(ret != -EEXIST);
4402        if (ret) {
4403                /* Pairs with create_reloc_root */
4404                btrfs_put_root(reloc_root);
4405                return ret;
4406        }
4407        new_root->reloc_root = btrfs_grab_root(reloc_root);
4408
4409        if (rc->create_reloc_tree)
4410                ret = clone_backref_node(trans, rc, root, reloc_root);
4411        return ret;
4412}
4413
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.