diff options
Diffstat (limited to 'fs/btrfs/relocation.c')
| -rw-r--r-- | fs/btrfs/relocation.c | 2389 |
1 files changed, 1378 insertions, 1011 deletions
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 3bbae80c752f..5bfefc3e9c06 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -18,13 +18,25 @@ #include "btrfs_inode.h" #include "async-thread.h" #include "free-space-cache.h" -#include "inode-map.h" #include "qgroup.h" #include "print-tree.h" #include "delalloc-space.h" #include "block-group.h" #include "backref.h" #include "misc.h" +#include "subpage.h" +#include "zoned.h" +#include "inode-item.h" +#include "space-info.h" +#include "fs.h" +#include "accessors.h" +#include "extent-tree.h" +#include "root-tree.h" +#include "file-item.h" +#include "relocation.h" +#include "super.h" +#include "tree-checker.h" +#include "raid-stripe-tree.h" /* * Relocation overview @@ -78,10 +90,15 @@ * map address of tree root to tree */ struct mapping_node { - struct { - struct rb_node rb_node; - u64 bytenr; - }; /* Use rb_simle_node for search/insert */ + union { + /* Use rb_simple_node for search/insert */ + struct { + struct rb_node rb_node; + u64 bytenr; + }; + + struct rb_simple_node simple_node; + }; void *data; }; @@ -94,13 +111,19 @@ struct mapping_tree { * present a tree block to process */ struct tree_block { - struct { - struct rb_node rb_node; - u64 bytenr; - }; /* Use rb_simple_node for search/insert */ + union { + /* Use rb_simple_node for search/insert */ + struct { + struct rb_node rb_node; + u64 bytenr; + }; + + struct rb_simple_node simple_node; + }; + u64 owner; struct btrfs_key key; - unsigned int level:8; - unsigned int key_ready:1; + u8 level; + bool key_ready; }; #define MAX_EXTENTS 128 @@ -110,6 +133,13 @@ struct file_extent_cluster { u64 end; u64 boundary[MAX_EXTENTS]; unsigned int nr; + u64 owning_root; +}; + +/* Stages of data relocation. */ +enum reloc_stage { + MOVE_DATA_EXTENTS, + UPDATE_DATA_PTRS }; struct reloc_control { @@ -143,16 +173,12 @@ struct reloc_control { u64 search_start; u64 extents_found; - unsigned int stage:8; - unsigned int create_reloc_tree:1; - unsigned int merge_reloc_tree:1; - unsigned int found_file_extent:1; + enum reloc_stage stage; + bool create_reloc_tree; + bool merge_reloc_tree; + bool found_file_extent; }; -/* stages of data relocation */ -#define MOVE_DATA_EXTENTS 0 -#define UPDATE_DATA_PTRS 1 - static void mark_block_processed(struct reloc_control *rc, struct btrfs_backref_node *node) { @@ -162,19 +188,13 @@ static void mark_block_processed(struct reloc_control *rc, in_range(node->bytenr, rc->block_group->start, rc->block_group->length)) { blocksize = rc->extent_root->fs_info->nodesize; - set_extent_bits(&rc->processed_blocks, node->bytenr, - node->bytenr + blocksize - 1, EXTENT_DIRTY); + btrfs_set_extent_bit(&rc->processed_blocks, node->bytenr, + node->bytenr + blocksize - 1, EXTENT_DIRTY, + NULL); } node->processed = 1; } - -static void mapping_tree_init(struct mapping_tree *tree) -{ - tree->rb_root = RB_ROOT; - spin_lock_init(&tree->lock); -} - /* * walk up backref nodes until reach node presents tree root */ @@ -186,8 +206,8 @@ static struct btrfs_backref_node *walk_up_backref( int idx = *index; while (!list_empty(&node->upper)) { - edge = list_entry(node->upper.next, - struct btrfs_backref_edge, list[LOWER]); + edge = list_first_entry(&node->upper, struct btrfs_backref_edge, + list[LOWER]); edges[idx++] = edge; node = edge->node[UPPER]; } @@ -213,8 +233,8 @@ static struct btrfs_backref_node *walk_down_backref( idx--; continue; } - edge = list_entry(edge->list[LOWER].next, - struct btrfs_backref_edge, list[LOWER]); + edge = list_first_entry(&edge->list[LOWER], struct btrfs_backref_edge, + list[LOWER]); edges[idx - 1] = edge; *index = idx; return edge->node[UPPER]; @@ -223,71 +243,7 @@ static struct btrfs_backref_node *walk_down_backref( return NULL; } -static void update_backref_node(struct btrfs_backref_cache *cache, - struct btrfs_backref_node *node, u64 bytenr) -{ - struct rb_node *rb_node; - rb_erase(&node->rb_node, &cache->rb_root); - node->bytenr = bytenr; - rb_node = rb_simple_insert(&cache->rb_root, node->bytenr, &node->rb_node); - if (rb_node) - btrfs_backref_panic(cache->fs_info, bytenr, -EEXIST); -} - -/* - * update backref cache after a transaction commit - */ -static int update_backref_cache(struct btrfs_trans_handle *trans, - struct btrfs_backref_cache *cache) -{ - struct btrfs_backref_node *node; - int level = 0; - - if (cache->last_trans == 0) { - cache->last_trans = trans->transid; - return 0; - } - - if (cache->last_trans == trans->transid) - return 0; - - /* - * detached nodes are used to avoid unnecessary backref - * lookup. transaction commit changes the extent tree. - * so the detached nodes are no longer useful. - */ - while (!list_empty(&cache->detached)) { - node = list_entry(cache->detached.next, - struct btrfs_backref_node, list); - btrfs_backref_cleanup_node(cache, node); - } - - while (!list_empty(&cache->changed)) { - node = list_entry(cache->changed.next, - struct btrfs_backref_node, list); - list_del_init(&node->list); - BUG_ON(node->pending); - update_backref_node(cache, node, node->new_bytenr); - } - - /* - * some nodes can be left in the pending list if there were - * errors during processing the pending nodes. - */ - for (level = 0; level < BTRFS_MAX_LEVEL; level++) { - list_for_each_entry(node, &cache->pending[level], list) { - BUG_ON(!node->pending); - if (node->bytenr == node->new_bytenr) - continue; - update_backref_node(cache, node, node->new_bytenr); - } - } - - cache->last_trans = 0; - return 1; -} - -static bool reloc_root_is_dead(struct btrfs_root *root) +static bool reloc_root_is_dead(const struct btrfs_root *root) { /* * Pair with set_bit/clear_bit in clean_dirty_subvols and @@ -308,7 +264,7 @@ static bool reloc_root_is_dead(struct btrfs_root *root) * from no reloc root. But btrfs_should_ignore_reloc_root() below is a * special case. */ -static bool have_reloc_root(struct btrfs_root *root) +static bool have_reloc_root(const struct btrfs_root *root) { if (reloc_root_is_dead(root)) return false; @@ -317,31 +273,30 @@ static bool have_reloc_root(struct btrfs_root *root) return true; } -int btrfs_should_ignore_reloc_root(struct btrfs_root *root) +bool btrfs_should_ignore_reloc_root(const struct btrfs_root *root) { struct btrfs_root *reloc_root; if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) - return 0; + return false; /* This root has been merged with its reloc tree, we can ignore it */ if (reloc_root_is_dead(root)) - return 1; + return true; reloc_root = root->reloc_root; if (!reloc_root) - return 0; + return false; if (btrfs_header_generation(reloc_root->commit_root) == root->fs_info->running_transaction->transid) - return 0; + return false; /* - * if there is reloc tree and it was created in previous - * transaction backref lookup can find the reloc tree, - * so backref node for the fs tree root is useless for - * relocation. + * If there is reloc tree and it was created in previous transaction + * backref lookup can find the reloc tree, so backref node for the fs + * tree root is useless for relocation. */ - return 1; + return true; } /* @@ -359,7 +314,7 @@ struct btrfs_root *find_reloc_root(struct btrfs_fs_info *fs_info, u64 bytenr) rb_node = rb_simple_search(&rc->reloc_root_tree.rb_root, bytenr); if (rb_node) { node = rb_entry(rb_node, struct mapping_node, rb_node); - root = (struct btrfs_root *)node->data; + root = node->data; } spin_unlock(&rc->reloc_root_tree.lock); return btrfs_grab_root(root); @@ -398,19 +353,13 @@ static bool handle_useless_nodes(struct reloc_control *rc, if (cur == node) ret = true; - /* The node is the lowest node */ - if (cur->lowest) { - list_del_init(&cur->lower); - cur->lowest = 0; - } - /* Cleanup the lower edges */ while (!list_empty(&cur->lower)) { struct btrfs_backref_edge *edge; struct btrfs_backref_node *lower; - edge = list_entry(cur->lower.next, - struct btrfs_backref_edge, list[UPPER]); + edge = list_first_entry(&cur->lower, struct btrfs_backref_edge, + list[UPPER]); list_del(&edge->list[UPPER]); list_del(&edge->list[LOWER]); lower = edge->node[LOWER]; @@ -429,7 +378,6 @@ static bool handle_useless_nodes(struct reloc_control *rc, * cache to avoid unnecessary backref lookup. */ if (cur->level > 0) { - list_add(&cur->list, &cache->detached); cur->detached = 1; } else { rb_erase(&cur->rb_node, &cache->rb_root); @@ -454,6 +402,7 @@ static bool handle_useless_nodes(struct reloc_control *rc, * cached. */ static noinline_for_stack struct btrfs_backref_node *build_backref_tree( + struct btrfs_trans_handle *trans, struct reloc_control *rc, struct btrfs_key *node_key, int level, u64 bytenr) { @@ -465,34 +414,31 @@ static noinline_for_stack struct btrfs_backref_node *build_backref_tree( struct btrfs_backref_node *node = NULL; struct btrfs_backref_edge *edge; int ret; - int err = 0; - iter = btrfs_backref_iter_alloc(rc->extent_root->fs_info, GFP_NOFS); + iter = btrfs_backref_iter_alloc(rc->extent_root->fs_info); if (!iter) return ERR_PTR(-ENOMEM); path = btrfs_alloc_path(); if (!path) { - err = -ENOMEM; + ret = -ENOMEM; goto out; } node = btrfs_backref_alloc_node(cache, bytenr, level); if (!node) { - err = -ENOMEM; + ret = -ENOMEM; goto out; } - node->lowest = 1; cur = node; /* Breadth-first search to build backref cache */ do { - ret = btrfs_backref_add_tree_node(cache, path, iter, node_key, - cur); - if (ret < 0) { - err = ret; + ret = btrfs_backref_add_tree_node(trans, cache, path, iter, + node_key, cur); + if (ret < 0) goto out; - } + edge = list_first_entry_or_null(&cache->pending_edge, struct btrfs_backref_edge, list[UPPER]); /* @@ -507,19 +453,18 @@ static noinline_for_stack struct btrfs_backref_node *build_backref_tree( /* Finish the upper linkage of newly added edges/nodes */ ret = btrfs_backref_finish_upper_links(cache, node); - if (ret < 0) { - err = ret; + if (ret < 0) goto out; - } if (handle_useless_nodes(rc, node)) node = NULL; out: - btrfs_backref_iter_free(iter); + btrfs_free_path(iter->path); + kfree(iter); btrfs_free_path(path); - if (err) { + if (ret) { btrfs_backref_error_cleanup(cache, node); - return ERR_PTR(err); + return ERR_PTR(ret); } ASSERT(!node || !node->detached); ASSERT(list_empty(&cache->useless_node) && @@ -528,98 +473,9 @@ out: } /* - * helper to add backref node for the newly created snapshot. - * the backref node is created by cloning backref node that - * corresponds to root of source tree - */ -static int clone_backref_node(struct btrfs_trans_handle *trans, - struct reloc_control *rc, - struct btrfs_root *src, - struct btrfs_root *dest) -{ - struct btrfs_root *reloc_root = src->reloc_root; - struct btrfs_backref_cache *cache = &rc->backref_cache; - struct btrfs_backref_node *node = NULL; - struct btrfs_backref_node *new_node; - struct btrfs_backref_edge *edge; - struct btrfs_backref_edge *new_edge; - struct rb_node *rb_node; - - if (cache->last_trans > 0) - update_backref_cache(trans, cache); - - rb_node = rb_simple_search(&cache->rb_root, src->commit_root->start); - if (rb_node) { - node = rb_entry(rb_node, struct btrfs_backref_node, rb_node); - if (node->detached) - node = NULL; - else - BUG_ON(node->new_bytenr != reloc_root->node->start); - } - - if (!node) { - rb_node = rb_simple_search(&cache->rb_root, - reloc_root->commit_root->start); - if (rb_node) { - node = rb_entry(rb_node, struct btrfs_backref_node, - rb_node); - BUG_ON(node->detached); - } - } - - if (!node) - return 0; - - new_node = btrfs_backref_alloc_node(cache, dest->node->start, - node->level); - if (!new_node) - return -ENOMEM; - - new_node->lowest = node->lowest; - new_node->checked = 1; - new_node->root = btrfs_grab_root(dest); - ASSERT(new_node->root); - - if (!node->lowest) { - list_for_each_entry(edge, &node->lower, list[UPPER]) { - new_edge = btrfs_backref_alloc_edge(cache); - if (!new_edge) - goto fail; - - btrfs_backref_link_edge(new_edge, edge->node[LOWER], - new_node, LINK_UPPER); - } - } else { - list_add_tail(&new_node->lower, &cache->leaves); - } - - rb_node = rb_simple_insert(&cache->rb_root, new_node->bytenr, - &new_node->rb_node); - if (rb_node) - btrfs_backref_panic(trans->fs_info, new_node->bytenr, -EEXIST); - - if (!new_node->lowest) { - list_for_each_entry(new_edge, &new_node->lower, list[UPPER]) { - list_add_tail(&new_edge->list[LOWER], - &new_edge->node[LOWER]->upper); - } - } - return 0; -fail: - while (!list_empty(&new_node->lower)) { - new_edge = list_entry(new_node->lower.next, - struct btrfs_backref_edge, list[UPPER]); - list_del(&new_edge->list[UPPER]); - btrfs_backref_free_edge(cache, new_edge); - } - btrfs_backref_free_node(cache, new_node); - return -ENOMEM; -} - -/* * helper to add 'address of tree root -> reloc tree' mapping */ -static int __must_check __add_reloc_root(struct btrfs_root *root) +static int __add_reloc_root(struct btrfs_root *root) { struct btrfs_fs_info *fs_info = root->fs_info; struct rb_node *rb_node; @@ -634,13 +490,13 @@ static int __must_check __add_reloc_root(struct btrfs_root *root) node->data = root; spin_lock(&rc->reloc_root_tree.lock); - rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root, - node->bytenr, &node->rb_node); + rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root, &node->simple_node); spin_unlock(&rc->reloc_root_tree.lock); if (rb_node) { - btrfs_panic(fs_info, -EEXIST, + btrfs_err(fs_info, "Duplicate root found for start=%llu while inserting into relocation tree", node->bytenr); + return -EEXIST; } list_add_tail(&root->root_list, &rc->reloc_roots); @@ -655,7 +511,7 @@ static void __del_reloc_root(struct btrfs_root *root) { struct btrfs_fs_info *fs_info = root->fs_info; struct rb_node *rb_node; - struct mapping_node *node = NULL; + struct mapping_node AUTO_KFREE(node); struct reloc_control *rc = fs_info->reloc_ctl; bool put_ref = false; @@ -669,9 +525,7 @@ static void __del_reloc_root(struct btrfs_root *root) RB_CLEAR_NODE(&node->rb_node); } spin_unlock(&rc->reloc_root_tree.lock); - if (!node) - return; - BUG_ON((struct btrfs_root *)node->data != root); + ASSERT(!node || (struct btrfs_root *)node->data == root); } /* @@ -690,7 +544,6 @@ static void __del_reloc_root(struct btrfs_root *root) spin_unlock(&fs_info->trans_lock); if (put_ref) btrfs_put_root(root); - kfree(node); } /* @@ -719,8 +572,7 @@ static int __update_reloc_root(struct btrfs_root *root) spin_lock(&rc->reloc_root_tree.lock); node->bytenr = root->node->start; - rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root, - node->bytenr, &node->rb_node); + rb_node = rb_simple_insert(&rc->reloc_root_tree.rb_root, &node->simple_node); spin_unlock(&rc->reloc_root_tree.lock); if (rb_node) btrfs_backref_panic(fs_info, node->bytenr, -EEXIST); @@ -733,24 +585,45 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info = root->fs_info; struct btrfs_root *reloc_root; struct extent_buffer *eb; - struct btrfs_root_item *root_item; + struct btrfs_root_item AUTO_KFREE(root_item); struct btrfs_key root_key; - int ret; + int ret = 0; root_item = kmalloc(sizeof(*root_item), GFP_NOFS); - BUG_ON(!root_item); + if (!root_item) + return ERR_PTR(-ENOMEM); root_key.objectid = BTRFS_TREE_RELOC_OBJECTID; root_key.type = BTRFS_ROOT_ITEM_KEY; root_key.offset = objectid; - if (root->root_key.objectid == objectid) { + if (btrfs_root_id(root) == objectid) { u64 commit_root_gen; + /* + * Relocation will wait for cleaner thread, and any half-dropped + * subvolume will be fully cleaned up at mount time. + * So here we shouldn't hit a subvolume with non-zero drop_progress. + * + * If this isn't the case, error out since it can make us attempt to + * drop references for extents that were already dropped before. + */ + if (unlikely(btrfs_disk_key_objectid(&root->root_item.drop_progress))) { + struct btrfs_key cpu_key; + + btrfs_disk_key_to_cpu(&cpu_key, &root->root_item.drop_progress); + btrfs_err(fs_info, + "cannot relocate partially dropped subvolume %llu, drop progress key " BTRFS_KEY_FMT, + objectid, BTRFS_KEY_FMT_VALUE(&cpu_key)); + return ERR_PTR(-EUCLEAN); + } + /* called by btrfs_init_reloc_root */ ret = btrfs_copy_root(trans, root, root->commit_root, &eb, BTRFS_TREE_RELOC_OBJECTID); - BUG_ON(ret); + if (ret) + return ERR_PTR(ret); + /* * Set the last_snapshot field to the generation of the commit * root - like this ctree.c:btrfs_block_can_be_shared() behaves @@ -771,19 +644,25 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans, */ ret = btrfs_copy_root(trans, root, root->node, &eb, BTRFS_TREE_RELOC_OBJECTID); - BUG_ON(ret); + if (ret) + return ERR_PTR(ret); } + /* + * We have changed references at this point, we must abort the + * transaction if anything fails (i.e. 'goto abort'). + */ + memcpy(root_item, &root->root_item, sizeof(*root_item)); btrfs_set_root_bytenr(root_item, eb->start); btrfs_set_root_level(root_item, btrfs_header_level(eb)); btrfs_set_root_generation(root_item, trans->transid); - if (root->root_key.objectid == objectid) { + if (btrfs_root_id(root) == objectid) { btrfs_set_root_refs(root_item, 0); memset(&root_item->drop_progress, 0, sizeof(struct btrfs_disk_key)); - root_item->drop_level = 0; + btrfs_set_root_drop_level(root_item, 0); } btrfs_tree_unlock(eb); @@ -791,14 +670,21 @@ static struct btrfs_root *create_reloc_root(struct btrfs_trans_handle *trans, ret = btrfs_insert_root(trans, fs_info->tree_root, &root_key, root_item); - BUG_ON(ret); - kfree(root_item); + if (ret) + goto abort; reloc_root = btrfs_read_tree_root(fs_info->tree_root, &root_key); - BUG_ON(IS_ERR(reloc_root)); + if (IS_ERR(reloc_root)) { + ret = PTR_ERR(reloc_root); + goto abort; + } set_bit(BTRFS_ROOT_SHAREABLE, &reloc_root->state); - reloc_root->last_trans = trans->transid; + btrfs_set_root_last_trans(reloc_root, trans->transid); return reloc_root; + +abort: + btrfs_abort_transaction(trans, ret); + return ERR_PTR(ret); } /* @@ -838,7 +724,7 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans, */ if (root->reloc_root) { reloc_root = root->reloc_root; - reloc_root->last_trans = trans->transid; + btrfs_set_root_last_trans(reloc_root, trans->transid); return 0; } @@ -846,8 +732,7 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans, * We are merging reloc roots, we do not need new reloc trees. Also * reloc trees never need their own reloc tree. */ - if (!rc->create_reloc_tree || - root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) + if (!rc->create_reloc_tree || btrfs_root_id(root) == BTRFS_TREE_RELOC_OBJECTID) return 0; if (!trans->reloc_reserved) { @@ -855,12 +740,19 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans, trans->block_rsv = rc->block_rsv; clear_rsv = 1; } - reloc_root = create_reloc_root(trans, root, root->root_key.objectid); + reloc_root = create_reloc_root(trans, root, btrfs_root_id(root)); if (clear_rsv) trans->block_rsv = rsv; + if (IS_ERR(reloc_root)) + return PTR_ERR(reloc_root); ret = __add_reloc_root(reloc_root); - BUG_ON(ret < 0); + ASSERT(ret != -EEXIST); + if (ret) { + /* Pairs with create_reloc_root */ + btrfs_put_root(reloc_root); + return ret; + } root->reloc_root = btrfs_grab_root(reloc_root); return 0; } @@ -877,7 +769,7 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans, int ret; if (!have_reloc_root(root)) - goto out; + return 0; reloc_root = root->reloc_root; root_item = &reloc_root->root_item; @@ -890,7 +782,7 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans, btrfs_grab_root(reloc_root); /* root->reloc_root will stay until current relocation finished */ - if (fs_info->reloc_ctl->merge_reloc_tree && + if (fs_info->reloc_ctl && fs_info->reloc_ctl->merge_reloc_tree && btrfs_root_refs(root_item) == 0) { set_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state); /* @@ -910,64 +802,8 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans, ret = btrfs_update_root(trans, fs_info->tree_root, &reloc_root->root_key, root_item); - BUG_ON(ret); btrfs_put_root(reloc_root); -out: - return 0; -} - -/* - * helper to find first cached inode with inode number >= objectid - * in a subvolume - */ -static struct inode *find_next_inode(struct btrfs_root *root, u64 objectid) -{ - struct rb_node *node; - struct rb_node *prev; - struct btrfs_inode *entry; - struct inode *inode; - - spin_lock(&root->inode_lock); -again: - node = root->inode_tree.rb_node; - prev = NULL; - while (node) { - prev = node; - entry = rb_entry(node, struct btrfs_inode, rb_node); - - if (objectid < btrfs_ino(entry)) - node = node->rb_left; - else if (objectid > btrfs_ino(entry)) - node = node->rb_right; - else - break; - } - if (!node) { - while (prev) { - entry = rb_entry(prev, struct btrfs_inode, rb_node); - if (objectid <= btrfs_ino(entry)) { - node = prev; - break; - } - prev = rb_next(prev); - } - } - while (node) { - entry = rb_entry(node, struct btrfs_inode, rb_node); - inode = igrab(&entry->vfs_inode); - if (inode) { - spin_unlock(&root->inode_lock); - return inode; - } - - objectid = btrfs_ino(entry) + 1; - if (cond_resched_lock(&root->inode_lock)) - goto again; - - node = rb_next(node); - } - spin_unlock(&root->inode_lock); - return NULL; + return ret; } /* @@ -977,7 +813,7 @@ static int get_new_location(struct inode *reloc_inode, u64 *new_bytenr, u64 bytenr, u64 num_bytes) { struct btrfs_root *root = BTRFS_I(reloc_inode)->root; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_file_extent_item *fi; struct extent_buffer *leaf; int ret; @@ -986,15 +822,13 @@ static int get_new_location(struct inode *reloc_inode, u64 *new_bytenr, if (!path) return -ENOMEM; - bytenr -= BTRFS_I(reloc_inode)->index_cnt; + bytenr -= BTRFS_I(reloc_inode)->reloc_block_group_start; ret = btrfs_lookup_file_extent(NULL, root, path, btrfs_ino(BTRFS_I(reloc_inode)), bytenr, 0); if (ret < 0) - goto out; - if (ret > 0) { - ret = -ENOENT; - goto out; - } + return ret; + if (ret > 0) + return -ENOENT; leaf = path->nodes[0]; fi = btrfs_item_ptr(leaf, path->slots[0], @@ -1005,16 +839,11 @@ static int get_new_location(struct inode *reloc_inode, u64 *new_bytenr, btrfs_file_extent_encryption(leaf, fi) || btrfs_file_extent_other_encoding(leaf, fi)); - if (num_bytes != btrfs_file_extent_disk_num_bytes(leaf, fi)) { - ret = -EINVAL; - goto out; - } + if (num_bytes != btrfs_file_extent_disk_num_bytes(leaf, fi)) + return -EINVAL; *new_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); - ret = 0; -out: - btrfs_free_path(path); - return ret; + return 0; } /* @@ -1030,7 +859,7 @@ int replace_file_extents(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info = root->fs_info; struct btrfs_key key; struct btrfs_file_extent_item *fi; - struct inode *inode = NULL; + struct btrfs_inode *inode = NULL; u64 parent; u64 bytenr; u64 new_bytenr = 0; @@ -1040,13 +869,12 @@ int replace_file_extents(struct btrfs_trans_handle *trans, u32 i; int ret = 0; int first = 1; - int dirty = 0; if (rc->stage != UPDATE_DATA_PTRS) return 0; /* reloc trees always use full backref */ - if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) + if (btrfs_root_id(root) == BTRFS_TREE_RELOC_OBJECTID) parent = leaf->start; else parent = 0; @@ -1072,33 +900,40 @@ int replace_file_extents(struct btrfs_trans_handle *trans, continue; /* - * if we are modifying block in fs tree, wait for readpage + * if we are modifying block in fs tree, wait for read_folio * to complete and drop the extent cache */ - if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) { + if (btrfs_root_id(root) != BTRFS_TREE_RELOC_OBJECTID) { if (first) { - inode = find_next_inode(root, key.objectid); + inode = btrfs_find_first_inode(root, key.objectid); first = 0; - } else if (inode && btrfs_ino(BTRFS_I(inode)) < key.objectid) { + } else if (inode && btrfs_ino(inode) < key.objectid) { btrfs_add_delayed_iput(inode); - inode = find_next_inode(root, key.objectid); + inode = btrfs_find_first_inode(root, key.objectid); } - if (inode && btrfs_ino(BTRFS_I(inode)) == key.objectid) { + if (inode && btrfs_ino(inode) == key.objectid) { + struct extent_state *cached_state = NULL; + end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); WARN_ON(!IS_ALIGNED(key.offset, fs_info->sectorsize)); WARN_ON(!IS_ALIGNED(end, fs_info->sectorsize)); end--; - ret = try_lock_extent(&BTRFS_I(inode)->io_tree, - key.offset, end); - if (!ret) + /* Take mmap lock to serialize with reflinks. */ + if (!down_read_trylock(&inode->i_mmap_lock)) + continue; + ret = btrfs_try_lock_extent(&inode->io_tree, key.offset, + end, &cached_state); + if (!ret) { + up_read(&inode->i_mmap_lock); continue; + } - btrfs_drop_extent_cache(BTRFS_I(inode), - key.offset, end, 1); - unlock_extent(&BTRFS_I(inode)->io_tree, - key.offset, end); + btrfs_drop_extent_map_range(inode, key.offset, end, true); + btrfs_unlock_extent(&inode->io_tree, key.offset, end, + &cached_state); + up_read(&inode->i_mmap_lock); } } @@ -1113,41 +948,44 @@ int replace_file_extents(struct btrfs_trans_handle *trans, } btrfs_set_file_extent_disk_bytenr(leaf, fi, new_bytenr); - dirty = 1; key.offset -= btrfs_file_extent_offset(leaf, fi); - btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, new_bytenr, - num_bytes, parent); - ref.real_root = root->root_key.objectid; - btrfs_init_data_ref(&ref, btrfs_header_owner(leaf), - key.objectid, key.offset); + ref.action = BTRFS_ADD_DELAYED_REF; + ref.bytenr = new_bytenr; + ref.num_bytes = num_bytes; + ref.parent = parent; + ref.owning_root = btrfs_root_id(root); + ref.ref_root = btrfs_header_owner(leaf); + btrfs_init_data_ref(&ref, key.objectid, key.offset, + btrfs_root_id(root), false); ret = btrfs_inc_extent_ref(trans, &ref); - if (ret) { + if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); break; } - btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, bytenr, - num_bytes, parent); - ref.real_root = root->root_key.objectid; - btrfs_init_data_ref(&ref, btrfs_header_owner(leaf), - key.objectid, key.offset); + ref.action = BTRFS_DROP_DELAYED_REF; + ref.bytenr = bytenr; + ref.num_bytes = num_bytes; + ref.parent = parent; + ref.owning_root = btrfs_root_id(root); + ref.ref_root = btrfs_header_owner(leaf); + btrfs_init_data_ref(&ref, key.objectid, key.offset, + btrfs_root_id(root), false); ret = btrfs_free_extent(trans, &ref); - if (ret) { + if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); break; } } - if (dirty) - btrfs_mark_buffer_dirty(leaf); if (inode) btrfs_add_delayed_iput(inode); return ret; } -static noinline_for_stack -int memcmp_node_keys(struct extent_buffer *eb, int slot, - struct btrfs_path *path, int level) +static noinline_for_stack int memcmp_node_keys(const struct extent_buffer *eb, + int slot, const struct btrfs_path *path, + int level) { struct btrfs_disk_key key1; struct btrfs_disk_key key2; @@ -1187,8 +1025,8 @@ int replace_path(struct btrfs_trans_handle *trans, struct reloc_control *rc, int ret; int slot; - BUG_ON(src->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID); - BUG_ON(dest->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID); + ASSERT(btrfs_root_id(src) == BTRFS_TREE_RELOC_OBJECTID); + ASSERT(btrfs_root_id(dest) != BTRFS_TREE_RELOC_OBJECTID); last_snapshot = btrfs_root_last_snapshot(&src->root_item); again: @@ -1196,7 +1034,6 @@ again: btrfs_node_key_to_cpu(path->nodes[lowest_level], &key, slot); eb = btrfs_lock_root_node(dest); - btrfs_set_lock_blocking_write(eb); level = btrfs_header_level(eb); if (level < lowest_level) { @@ -1206,10 +1043,14 @@ again: } if (cow) { - ret = btrfs_cow_block(trans, dest, eb, NULL, 0, &eb); - BUG_ON(ret); + ret = btrfs_cow_block(trans, dest, eb, NULL, 0, &eb, + BTRFS_NESTING_COW); + if (ret) { + btrfs_tree_unlock(eb); + free_extent_buffer(eb); + return ret; + } } - btrfs_set_lock_blocking_write(eb); if (next_key) { next_key->objectid = (u64)-1; @@ -1219,12 +1060,10 @@ again: parent = eb; while (1) { - struct btrfs_key first_key; - level = btrfs_header_level(parent); - BUG_ON(level < lowest_level); + ASSERT(level >= lowest_level); - ret = btrfs_bin_search(parent, &key, &slot); + ret = btrfs_bin_search(parent, 0, &key, &slot); if (ret < 0) break; if (ret && slot > 0) @@ -1236,7 +1075,6 @@ again: old_bytenr = btrfs_node_blockptr(parent, slot); blocksize = fs_info->nodesize; old_ptr_gen = btrfs_node_ptr_generation(parent, slot); - btrfs_node_key_to_cpu(parent, &first_key, slot); if (level <= max_level) { eb = path->nodes[level]; @@ -1261,23 +1099,22 @@ again: break; } - eb = read_tree_block(fs_info, old_bytenr, old_ptr_gen, - level - 1, &first_key); + eb = btrfs_read_node_slot(parent, slot); if (IS_ERR(eb)) { ret = PTR_ERR(eb); break; - } else if (!extent_buffer_uptodate(eb)) { - ret = -EIO; - free_extent_buffer(eb); - break; } btrfs_tree_lock(eb); if (cow) { ret = btrfs_cow_block(trans, dest, eb, parent, - slot, &eb); - BUG_ON(ret); + slot, &eb, + BTRFS_NESTING_COW); + if (ret) { + btrfs_tree_unlock(eb); + free_extent_buffer(eb); + break; + } } - btrfs_set_lock_blocking_write(eb); btrfs_tree_unlock(parent); free_extent_buffer(parent); @@ -1298,9 +1135,15 @@ again: btrfs_release_path(path); path->lowest_level = level; + set_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &src->state); ret = btrfs_search_slot(trans, src, &key, path, 0, 1); + clear_bit(BTRFS_ROOT_RESET_LOCKDEP_CLASS, &src->state); path->lowest_level = 0; - BUG_ON(ret); + if (ret) { + if (ret > 0) + ret = -ENOENT; + break; + } /* * Info qgroup to trace both subtrees. @@ -1316,7 +1159,7 @@ again: * The real subtree rescan is delayed until we have new * CoW on the subtree root node before transaction commit. */ - ret = btrfs_qgroup_add_swapped_blocks(trans, dest, + ret = btrfs_qgroup_add_swapped_blocks(dest, rc->block_group, parent, slot, path->nodes[level], path->slots[level], last_snapshot); @@ -1327,40 +1170,65 @@ again: */ btrfs_set_node_blockptr(parent, slot, new_bytenr); btrfs_set_node_ptr_generation(parent, slot, new_ptr_gen); - btrfs_mark_buffer_dirty(parent); btrfs_set_node_blockptr(path->nodes[level], path->slots[level], old_bytenr); btrfs_set_node_ptr_generation(path->nodes[level], path->slots[level], old_ptr_gen); - btrfs_mark_buffer_dirty(path->nodes[level]); - btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, old_bytenr, - blocksize, path->nodes[level]->start); - ref.skip_qgroup = true; - btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid); + ref.action = BTRFS_ADD_DELAYED_REF; + ref.bytenr = old_bytenr; + ref.num_bytes = blocksize; + ref.parent = path->nodes[level]->start; + ref.owning_root = btrfs_root_id(src); + ref.ref_root = btrfs_root_id(src); + btrfs_init_tree_ref(&ref, level - 1, 0, true); ret = btrfs_inc_extent_ref(trans, &ref); - BUG_ON(ret); - btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, new_bytenr, - blocksize, 0); - ref.skip_qgroup = true; - btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + break; + } + + ref.action = BTRFS_ADD_DELAYED_REF; + ref.bytenr = new_bytenr; + ref.num_bytes = blocksize; + ref.parent = 0; + ref.owning_root = btrfs_root_id(dest); + ref.ref_root = btrfs_root_id(dest); + btrfs_init_tree_ref(&ref, level - 1, 0, true); ret = btrfs_inc_extent_ref(trans, &ref); - BUG_ON(ret); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + break; + } - btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, new_bytenr, - blocksize, path->nodes[level]->start); - btrfs_init_tree_ref(&ref, level - 1, src->root_key.objectid); - ref.skip_qgroup = true; + /* We don't know the real owning_root, use 0. */ + ref.action = BTRFS_DROP_DELAYED_REF; + ref.bytenr = new_bytenr; + ref.num_bytes = blocksize; + ref.parent = path->nodes[level]->start; + ref.owning_root = 0; + ref.ref_root = btrfs_root_id(src); + btrfs_init_tree_ref(&ref, level - 1, 0, true); ret = btrfs_free_extent(trans, &ref); - BUG_ON(ret); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + break; + } - btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF, old_bytenr, - blocksize, 0); - btrfs_init_tree_ref(&ref, level - 1, dest->root_key.objectid); - ref.skip_qgroup = true; + /* We don't know the real owning_root, use 0. */ + ref.action = BTRFS_DROP_DELAYED_REF; + ref.bytenr = old_bytenr; + ref.num_bytes = blocksize; + ref.parent = 0; + ref.owning_root = 0; + ref.ref_root = btrfs_root_id(dest); + btrfs_init_tree_ref(&ref, level - 1, 0, true); ret = btrfs_free_extent(trans, &ref); - BUG_ON(ret); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + break; + } btrfs_unlock_up_safe(path, 0); @@ -1416,10 +1284,8 @@ static noinline_for_stack int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path, int *level) { - struct btrfs_fs_info *fs_info = root->fs_info; struct extent_buffer *eb = NULL; int i; - u64 bytenr; u64 ptr_gen = 0; u64 last_snapshot; u32 nritems; @@ -1427,8 +1293,6 @@ int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path, last_snapshot = btrfs_root_last_snapshot(&root->root_item); for (i = *level; i > 0; i--) { - struct btrfs_key first_key; - eb = path->nodes[i]; nritems = btrfs_header_nritems(eb); while (path->slots[i] < nritems) { @@ -1448,16 +1312,9 @@ int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path, return 0; } - bytenr = btrfs_node_blockptr(eb, path->slots[i]); - btrfs_node_key_to_cpu(eb, &first_key, path->slots[i]); - eb = read_tree_block(fs_info, bytenr, ptr_gen, i - 1, - &first_key); - if (IS_ERR(eb)) { + eb = btrfs_read_node_slot(eb, path->slots[i]); + if (IS_ERR(eb)) return PTR_ERR(eb); - } else if (!extent_buffer_uptodate(eb)) { - free_extent_buffer(eb); - return -EIO; - } BUG_ON(btrfs_header_level(eb) != i - 1); path->nodes[i - 1] = eb; path->slots[i - 1] = 0; @@ -1470,35 +1327,38 @@ int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path, * [min_key, max_key) */ static int invalidate_extent_cache(struct btrfs_root *root, - struct btrfs_key *min_key, - struct btrfs_key *max_key) + const struct btrfs_key *min_key, + const struct btrfs_key *max_key) { struct btrfs_fs_info *fs_info = root->fs_info; - struct inode *inode = NULL; + struct btrfs_inode *inode = NULL; u64 objectid; u64 start, end; u64 ino; objectid = min_key->objectid; while (1) { + struct extent_state *cached_state = NULL; + cond_resched(); - iput(inode); + if (inode) + iput(&inode->vfs_inode); if (objectid > max_key->objectid) break; - inode = find_next_inode(root, objectid); + inode = btrfs_find_first_inode(root, objectid); if (!inode) break; - ino = btrfs_ino(BTRFS_I(inode)); + ino = btrfs_ino(inode); if (ino > max_key->objectid) { - iput(inode); + iput(&inode->vfs_inode); break; } objectid = ino + 1; - if (!S_ISREG(inode->i_mode)) + if (!S_ISREG(inode->vfs_inode.i_mode)) continue; if (unlikely(min_key->objectid == ino)) { @@ -1530,10 +1390,10 @@ static int invalidate_extent_cache(struct btrfs_root *root, end = (u64)-1; } - /* the lock_extent waits for readpage to complete */ - lock_extent(&BTRFS_I(inode)->io_tree, start, end); - btrfs_drop_extent_cache(BTRFS_I(inode), start, end, 1); - unlock_extent(&BTRFS_I(inode)->io_tree, start, end); + /* the lock_extent waits for read_folio to complete */ + btrfs_lock_extent(&inode->io_tree, start, end, &cached_state); + btrfs_drop_extent_map_range(inode, start, end, true); + btrfs_unlock_extent(&inode->io_tree, start, end, &cached_state); } return 0; } @@ -1559,28 +1419,33 @@ static int find_next_key(struct btrfs_path *path, int level, /* * Insert current subvolume into reloc_control::dirty_subvol_roots */ -static void insert_dirty_subvol(struct btrfs_trans_handle *trans, - struct reloc_control *rc, - struct btrfs_root *root) +static int insert_dirty_subvol(struct btrfs_trans_handle *trans, + struct reloc_control *rc, + struct btrfs_root *root) { struct btrfs_root *reloc_root = root->reloc_root; struct btrfs_root_item *reloc_root_item; + int ret; /* @root must be a subvolume tree root with a valid reloc tree */ - ASSERT(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID); + ASSERT(btrfs_root_id(root) != BTRFS_TREE_RELOC_OBJECTID); ASSERT(reloc_root); reloc_root_item = &reloc_root->root_item; memset(&reloc_root_item->drop_progress, 0, sizeof(reloc_root_item->drop_progress)); - reloc_root_item->drop_level = 0; + btrfs_set_root_drop_level(reloc_root_item, 0); btrfs_set_root_refs(reloc_root_item, 0); - btrfs_update_reloc_root(trans, root); + ret = btrfs_update_reloc_root(trans, root); + if (ret) + return ret; if (list_empty(&root->reloc_dirty_list)) { btrfs_grab_root(root); list_add_tail(&root->reloc_dirty_list, &rc->dirty_subvol_roots); } + + return 0; } static int clean_dirty_subvols(struct reloc_control *rc) @@ -1592,7 +1457,7 @@ static int clean_dirty_subvols(struct reloc_control *rc) list_for_each_entry_safe(root, next, &rc->dirty_subvol_roots, reloc_dirty_list) { - if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) { + if (btrfs_root_id(root) != BTRFS_TREE_RELOC_OBJECTID) { /* Merged subvolume, cleanup its reloc root */ struct btrfs_root *reloc_root = root->reloc_root; @@ -1610,7 +1475,7 @@ static int clean_dirty_subvols(struct reloc_control *rc) * ->reloc_root. If it fails however we must * drop the ref ourselves. */ - ret2 = btrfs_drop_snapshot(reloc_root, 0, 1); + ret2 = btrfs_drop_snapshot(reloc_root, false, true); if (ret2 < 0) { btrfs_put_root(reloc_root); if (!ret) @@ -1620,7 +1485,7 @@ static int clean_dirty_subvols(struct reloc_control *rc) btrfs_put_root(root); } else { /* Orphan reloc tree, just clean it up */ - ret2 = btrfs_drop_snapshot(root, 0, 1); + ret2 = btrfs_drop_snapshot(root, false, true); if (ret2 < 0) { btrfs_put_root(root); if (!ret) @@ -1646,11 +1511,11 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, struct btrfs_root_item *root_item; struct btrfs_path *path; struct extent_buffer *leaf; + int reserve_level; int level; int max_level; int replaced = 0; - int ret; - int err = 0; + int ret = 0; u32 min_reserved; path = btrfs_alloc_path(); @@ -1663,13 +1528,13 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, if (btrfs_disk_key_objectid(&root_item->drop_progress) == 0) { level = btrfs_root_level(root_item); - atomic_inc(&reloc_root->node->refs); + refcount_inc(&reloc_root->node->refs); path->nodes[level] = reloc_root->node; path->slots[level] = 0; } else { btrfs_disk_key_to_cpu(&key, &root_item->drop_progress); - level = root_item->drop_level; + level = btrfs_root_drop_level(root_item); BUG_ON(level == 0); path->lowest_level = level; ret = btrfs_search_slot(NULL, reloc_root, &key, path, 0, 0); @@ -1686,19 +1551,27 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, btrfs_unlock_up_safe(path, 0); } - min_reserved = fs_info->nodesize * (BTRFS_MAX_LEVEL - 1) * 2; + /* + * In merge_reloc_root(), we modify the upper level pointer to swap the + * tree blocks between reloc tree and subvolume tree. Thus for tree + * block COW, we COW at most from level 1 to root level for each tree. + * + * Thus the needed metadata size is at most root_level * nodesize, + * and * 2 since we have two trees to COW. + */ + reserve_level = max_t(int, 1, btrfs_root_level(root_item)); + min_reserved = fs_info->nodesize * reserve_level * 2; memset(&next_key, 0, sizeof(next_key)); while (1) { - ret = btrfs_block_rsv_refill(root, rc->block_rsv, min_reserved, - BTRFS_RESERVE_FLUSH_ALL); - if (ret) { - err = ret; + ret = btrfs_block_rsv_refill(fs_info, rc->block_rsv, + min_reserved, + BTRFS_RESERVE_FLUSH_LIMIT); + if (ret) goto out; - } trans = btrfs_start_transaction(root, 0); if (IS_ERR(trans)) { - err = PTR_ERR(trans); + ret = PTR_ERR(trans); trans = NULL; goto out; } @@ -1713,17 +1586,15 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, * btrfs_update_reloc_root() and update our root item * appropriately. */ - reloc_root->last_trans = trans->transid; + btrfs_set_root_last_trans(reloc_root, trans->transid); trans->block_rsv = rc->block_rsv; replaced = 0; max_level = level; ret = walk_down_reloc_tree(reloc_root, path, &level); - if (ret < 0) { - err = ret; + if (ret < 0) goto out; - } if (ret > 0) break; @@ -1734,11 +1605,8 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, ret = replace_path(trans, rc, root, reloc_root, path, &next_key, level, max_level); } - if (ret < 0) { - err = ret; + if (ret < 0) goto out; - } - if (ret > 0) { level = ret; btrfs_node_key_to_cpu(path->nodes[level], &key, @@ -1757,7 +1625,7 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, */ btrfs_node_key(path->nodes[level], &root_item->drop_progress, path->slots[level]); - root_item->drop_level = level; + btrfs_set_root_drop_level(root_item, level); btrfs_end_transaction_throttle(trans); trans = NULL; @@ -1773,16 +1641,18 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc, * relocated and the block is tree root. */ leaf = btrfs_lock_root_node(root); - ret = btrfs_cow_block(trans, root, leaf, NULL, 0, &leaf); + ret = btrfs_cow_block(trans, root, leaf, NULL, 0, &leaf, + BTRFS_NESTING_COW); btrfs_tree_unlock(leaf); free_extent_buffer(leaf); - if (ret < 0) - err = ret; out: btrfs_free_path(path); - if (err == 0) - insert_dirty_subvol(trans, rc, root); + if (ret == 0) { + ret = insert_dirty_subvol(trans, rc, root); + if (ret) + btrfs_abort_transaction(trans, ret); + } if (trans) btrfs_end_transaction_throttle(trans); @@ -1792,7 +1662,7 @@ out: if (replaced && rc->stage == UPDATE_DATA_PTRS) invalidate_extent_cache(root, &key, &next_key); - return err; + return ret; } static noinline_for_stack @@ -1814,7 +1684,7 @@ int prepare_to_merge(struct reloc_control *rc, int err) again: if (!err) { num_bytes = rc->merging_rsv_size; - ret = btrfs_block_rsv_add(root, rc->block_rsv, num_bytes, + ret = btrfs_block_rsv_add(fs_info, rc->block_rsv, num_bytes, BTRFS_RESERVE_FLUSH_ALL); if (ret) err = ret; @@ -1837,17 +1707,59 @@ again: } } - rc->merge_reloc_tree = 1; + rc->merge_reloc_tree = true; while (!list_empty(&rc->reloc_roots)) { - reloc_root = list_entry(rc->reloc_roots.next, - struct btrfs_root, root_list); + reloc_root = list_first_entry(&rc->reloc_roots, + struct btrfs_root, root_list); list_del_init(&reloc_root->root_list); root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset, false); - BUG_ON(IS_ERR(root)); - BUG_ON(root->reloc_root != reloc_root); + if (IS_ERR(root)) { + /* + * Even if we have an error we need this reloc root + * back on our list so we can clean up properly. + */ + list_add(&reloc_root->root_list, &reloc_roots); + btrfs_abort_transaction(trans, (int)PTR_ERR(root)); + if (!err) + err = PTR_ERR(root); + break; + } + + if (unlikely(root->reloc_root != reloc_root)) { + if (root->reloc_root) { + btrfs_err(fs_info, +"reloc tree mismatch, root %lld has reloc root key (%lld %u %llu) gen %llu, expect reloc root key (%lld %u %llu) gen %llu", + btrfs_root_id(root), + btrfs_root_id(root->reloc_root), + root->reloc_root->root_key.type, + root->reloc_root->root_key.offset, + btrfs_root_generation( + &root->reloc_root->root_item), + btrfs_root_id(reloc_root), + reloc_root->root_key.type, + reloc_root->root_key.offset, + btrfs_root_generation( + &reloc_root->root_item)); + } else { + btrfs_err(fs_info, +"reloc tree mismatch, root %lld has no reloc root, expect reloc root key (%lld %u %llu) gen %llu", + btrfs_root_id(root), + btrfs_root_id(reloc_root), + reloc_root->root_key.type, + reloc_root->root_key.offset, + btrfs_root_generation( + &reloc_root->root_item)); + } + list_add(&reloc_root->root_list, &reloc_roots); + btrfs_put_root(root); + btrfs_abort_transaction(trans, -EUCLEAN); + if (!err) + err = -EUCLEAN; + break; + } /* * set reference count to 1, so btrfs_recover_relocation @@ -1855,16 +1767,27 @@ again: */ if (!err) btrfs_set_root_refs(&reloc_root->root_item, 1); - btrfs_update_reloc_root(trans, root); + ret = btrfs_update_reloc_root(trans, root); + /* + * Even if we have an error we need this reloc root back on our + * list so we can clean up properly. + */ list_add(&reloc_root->root_list, &reloc_roots); btrfs_put_root(root); + + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + if (!err) + err = ret; + break; + } } list_splice(&reloc_roots, &rc->reloc_roots); if (!err) - btrfs_commit_transaction(trans); + err = btrfs_commit_transaction(trans); else btrfs_end_transaction(trans); return err; @@ -1903,14 +1826,31 @@ again: while (!list_empty(&reloc_roots)) { found = 1; - reloc_root = list_entry(reloc_roots.next, - struct btrfs_root, root_list); + reloc_root = list_first_entry(&reloc_roots, struct btrfs_root, root_list); root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset, false); if (btrfs_root_refs(&reloc_root->root_item) > 0) { - BUG_ON(IS_ERR(root)); - BUG_ON(root->reloc_root != reloc_root); + if (WARN_ON(IS_ERR(root))) { + /* + * For recovery we read the fs roots on mount, + * and if we didn't find the root then we marked + * the reloc root as a garbage root. For normal + * relocation obviously the root should exist in + * memory. However there's no reason we can't + * handle the error properly here just in case. + */ + ret = PTR_ERR(root); + goto out; + } + if (WARN_ON(root->reloc_root != reloc_root)) { + /* + * This can happen if on-disk metadata has some + * corruption, e.g. bad reloc tree key offset. + */ + ret = -EINVAL; + goto out; + } ret = merge_reloc_root(rc, root); btrfs_put_root(root); if (ret) { @@ -1988,12 +1928,31 @@ static int record_reloc_root_in_trans(struct btrfs_trans_handle *trans, struct btrfs_root *root; int ret; - if (reloc_root->last_trans == trans->transid) + if (btrfs_get_root_last_trans(reloc_root) == trans->transid) return 0; root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset, false); - BUG_ON(IS_ERR(root)); - BUG_ON(root->reloc_root != reloc_root); + + /* + * This should succeed, since we can't have a reloc root without having + * already looked up the actual root and created the reloc root for this + * root. + * + * However if there's some sort of corruption where we have a ref to a + * reloc root without a corresponding root this could return ENOENT. + */ + if (IS_ERR(root)) { + DEBUG_WARN("error %ld reading root for reloc root", PTR_ERR(root)); + return PTR_ERR(root); + } + if (unlikely(root->reloc_root != reloc_root)) { + DEBUG_WARN("unexpected reloc root found"); + btrfs_err(fs_info, + "root %llu has two reloc roots associated with it", + reloc_root->root_key.offset); + btrfs_put_root(root); + return -EUCLEAN; + } ret = btrfs_record_root_in_trans(trans, root); btrfs_put_root(root); @@ -2009,45 +1968,74 @@ struct btrfs_root *select_reloc_root(struct btrfs_trans_handle *trans, struct btrfs_backref_node *next; struct btrfs_root *root; int index = 0; + int ret; - next = node; - while (1) { - cond_resched(); - next = walk_up_backref(next, edges, &index); - root = next->root; - BUG_ON(!root); - BUG_ON(!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)); + next = walk_up_backref(node, edges, &index); + root = next->root; - if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) { - record_reloc_root_in_trans(trans, root); - break; - } + /* + * If there is no root, then our references for this block are + * incomplete, as we should be able to walk all the way up to a block + * that is owned by a root. + * + * This path is only for SHAREABLE roots, so if we come upon a + * non-SHAREABLE root then we have backrefs that resolve improperly. + * + * Both of these cases indicate file system corruption, or a bug in the + * backref walking code. + */ + if (unlikely(!root)) { + btrfs_err(trans->fs_info, + "bytenr %llu doesn't have a backref path ending in a root", + node->bytenr); + return ERR_PTR(-EUCLEAN); + } + if (unlikely(!test_bit(BTRFS_ROOT_SHAREABLE, &root->state))) { + btrfs_err(trans->fs_info, + "bytenr %llu has multiple refs with one ending in a non-shareable root", + node->bytenr); + return ERR_PTR(-EUCLEAN); + } - btrfs_record_root_in_trans(trans, root); - root = root->reloc_root; + if (btrfs_root_id(root) == BTRFS_TREE_RELOC_OBJECTID) { + ret = record_reloc_root_in_trans(trans, root); + if (ret) + return ERR_PTR(ret); + goto found; + } - if (next->new_bytenr != root->node->start) { - BUG_ON(next->new_bytenr); - BUG_ON(!list_empty(&next->list)); - next->new_bytenr = root->node->start; - btrfs_put_root(next->root); - next->root = btrfs_grab_root(root); - ASSERT(next->root); - list_add_tail(&next->list, - &rc->backref_cache.changed); - mark_block_processed(rc, next); - break; - } + ret = btrfs_record_root_in_trans(trans, root); + if (ret) + return ERR_PTR(ret); + root = root->reloc_root; - WARN_ON(1); - root = NULL; - next = walk_down_backref(edges, &index); - if (!next || next->level <= node->level) - break; - } + /* + * We could have raced with another thread which failed, so + * root->reloc_root may not be set, return ENOENT in this case. + */ if (!root) - return NULL; + return ERR_PTR(-ENOENT); + if (unlikely(next->new_bytenr)) { + /* + * We just created the reloc root, so we shouldn't have + * ->new_bytenr set yet. If it is then we have multiple roots + * pointing at the same bytenr which indicates corruption, or + * we've made a mistake in the backref walking code. + */ + ASSERT(next->new_bytenr == 0); + btrfs_err(trans->fs_info, + "bytenr %llu possibly has multiple roots pointing at the same bytenr %llu", + node->bytenr, next->bytenr); + return ERR_PTR(-EUCLEAN); + } + + next->new_bytenr = root->node->start; + btrfs_put_root(next->root); + next->root = btrfs_grab_root(root); + ASSERT(next->root); + mark_block_processed(rc, next); +found: next = node; /* setup backref node path for btrfs_reloc_cow_block */ while (1) { @@ -2082,13 +2070,19 @@ struct btrfs_root *select_one_root(struct btrfs_backref_node *node) cond_resched(); next = walk_up_backref(next, edges, &index); root = next->root; - BUG_ON(!root); + + /* + * This can occur if we have incomplete extent refs leading all + * the way up a particular path, in this case return -EUCLEAN. + */ + if (unlikely(!root)) + return ERR_PTR(-EUCLEAN); /* No other choice for non-shareable tree */ if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) return root; - if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) + if (btrfs_root_id(root) != BTRFS_TREE_RELOC_OBJECTID) fs_root = root; if (next != node) @@ -2104,9 +2098,8 @@ struct btrfs_root *select_one_root(struct btrfs_backref_node *node) return fs_root; } -static noinline_for_stack -u64 calcu_metadata_size(struct reloc_control *rc, - struct btrfs_backref_node *node, int reserve) +static noinline_for_stack u64 calcu_metadata_size(struct reloc_control *rc, + struct btrfs_backref_node *node) { struct btrfs_fs_info *fs_info = rc->extent_root->fs_info; struct btrfs_backref_node *next = node; @@ -2115,12 +2108,12 @@ u64 calcu_metadata_size(struct reloc_control *rc, u64 num_bytes = 0; int index = 0; - BUG_ON(reserve && node->processed); + BUG_ON(node->processed); while (next) { cond_resched(); while (1) { - if (next->processed && (reserve || next != node)) + if (next->processed) break; num_bytes += fs_info->nodesize; @@ -2128,8 +2121,8 @@ u64 calcu_metadata_size(struct reloc_control *rc, if (list_empty(&next->upper)) break; - edge = list_entry(next->upper.next, - struct btrfs_backref_edge, list[LOWER]); + edge = list_first_entry(&next->upper, struct btrfs_backref_edge, + list[LOWER]); edges[index++] = edge; next = edge->node[UPPER]; } @@ -2138,17 +2131,11 @@ u64 calcu_metadata_size(struct reloc_control *rc, return num_bytes; } -static int reserve_metadata_space(struct btrfs_trans_handle *trans, - struct reloc_control *rc, - struct btrfs_backref_node *node) +static int refill_metadata_space(struct btrfs_trans_handle *trans, + struct reloc_control *rc, u64 num_bytes) { - struct btrfs_root *root = rc->extent_root; - struct btrfs_fs_info *fs_info = root->fs_info; - u64 num_bytes; + struct btrfs_fs_info *fs_info = trans->fs_info; int ret; - u64 tmp; - - num_bytes = calcu_metadata_size(rc, node, 1) * 2; trans->block_rsv = rc->block_rsv; rc->reserved_bytes += num_bytes; @@ -2158,10 +2145,11 @@ static int reserve_metadata_space(struct btrfs_trans_handle *trans, * If we get an enospc just kick back -EAGAIN so we know to drop the * transaction and try to refill when we can flush all the things. */ - ret = btrfs_block_rsv_refill(root, rc->block_rsv, num_bytes, - BTRFS_RESERVE_FLUSH_LIMIT); + ret = btrfs_block_rsv_refill(fs_info, rc->block_rsv, num_bytes, + BTRFS_RESERVE_FLUSH_LIMIT); if (ret) { - tmp = fs_info->nodesize * RELOCATION_RESERVED_NODES; + u64 tmp = fs_info->nodesize * RELOCATION_RESERVED_NODES; + while (tmp <= rc->reserved_bytes) tmp <<= 1; /* @@ -2179,6 +2167,16 @@ static int reserve_metadata_space(struct btrfs_trans_handle *trans, return 0; } +static int reserve_metadata_space(struct btrfs_trans_handle *trans, + struct reloc_control *rc, + struct btrfs_backref_node *node) +{ + u64 num_bytes; + + num_bytes = calcu_metadata_size(rc, node) * 2; + return refill_metadata_space(trans, rc, num_bytes); +} + /* * relocate a block tree, and then update pointers in upper level * blocks that reference the block to point to the new location. @@ -2192,7 +2190,6 @@ static int do_relocation(struct btrfs_trans_handle *trans, struct btrfs_key *key, struct btrfs_path *path, int lowest) { - struct btrfs_fs_info *fs_info = rc->extent_root->fs_info; struct btrfs_backref_node *upper; struct btrfs_backref_edge *edge; struct btrfs_backref_edge *edges[BTRFS_MAX_LEVEL - 1]; @@ -2200,32 +2197,32 @@ static int do_relocation(struct btrfs_trans_handle *trans, struct extent_buffer *eb; u32 blocksize; u64 bytenr; - u64 generation; int slot; - int ret; - int err = 0; + int ret = 0; - BUG_ON(lowest && node->eb); + /* + * If we are lowest then this is the first time we're processing this + * block, and thus shouldn't have an eb associated with it yet. + */ + ASSERT(!lowest || !node->eb); path->lowest_level = node->level + 1; rc->backref_cache.path[node->level] = node; list_for_each_entry(edge, &node->upper, list[LOWER]) { - struct btrfs_key first_key; - struct btrfs_ref ref = { 0 }; - cond_resched(); upper = edge->node[UPPER]; root = select_reloc_root(trans, rc, upper, edges); - BUG_ON(!root); + if (IS_ERR(root)) { + ret = PTR_ERR(root); + goto next; + } if (upper->eb && !upper->locked) { if (!lowest) { - ret = btrfs_bin_search(upper->eb, key, &slot); - if (ret < 0) { - err = ret; + ret = btrfs_bin_search(upper->eb, 0, key, &slot); + if (ret < 0) goto next; - } BUG_ON(ret); bytenr = btrfs_node_blockptr(upper->eb, slot); if (node->eb->start == bytenr) @@ -2237,10 +2234,8 @@ static int do_relocation(struct btrfs_trans_handle *trans, if (!upper->eb) { ret = btrfs_search_slot(trans, root, key, path, 0, 1); if (ret) { - if (ret < 0) - err = ret; - else - err = -ENOENT; + if (ret > 0) + ret = -ENOENT; btrfs_release_path(path); break; @@ -2259,22 +2254,20 @@ static int do_relocation(struct btrfs_trans_handle *trans, slot = path->slots[upper->level]; btrfs_release_path(path); } else { - ret = btrfs_bin_search(upper->eb, key, &slot); - if (ret < 0) { - err = ret; + ret = btrfs_bin_search(upper->eb, 0, key, &slot); + if (ret < 0) goto next; - } BUG_ON(ret); } bytenr = btrfs_node_blockptr(upper->eb, slot); if (lowest) { - if (bytenr != node->bytenr) { + if (unlikely(bytenr != node->bytenr)) { btrfs_err(root->fs_info, "lowest leaf/node mismatch: bytenr %llu node->bytenr %llu slot %d upper %llu", bytenr, node->bytenr, slot, upper->eb->start); - err = -EIO; + ret = -EIO; goto next; } } else { @@ -2283,68 +2276,73 @@ static int do_relocation(struct btrfs_trans_handle *trans, } blocksize = root->fs_info->nodesize; - generation = btrfs_node_ptr_generation(upper->eb, slot); - btrfs_node_key_to_cpu(upper->eb, &first_key, slot); - eb = read_tree_block(fs_info, bytenr, generation, - upper->level - 1, &first_key); + eb = btrfs_read_node_slot(upper->eb, slot); if (IS_ERR(eb)) { - err = PTR_ERR(eb); - goto next; - } else if (!extent_buffer_uptodate(eb)) { - free_extent_buffer(eb); - err = -EIO; + ret = PTR_ERR(eb); goto next; } btrfs_tree_lock(eb); - btrfs_set_lock_blocking_write(eb); if (!node->eb) { ret = btrfs_cow_block(trans, root, eb, upper->eb, - slot, &eb); + slot, &eb, BTRFS_NESTING_COW); btrfs_tree_unlock(eb); free_extent_buffer(eb); - if (ret < 0) { - err = ret; + if (ret < 0) goto next; - } - BUG_ON(node->eb != eb); + /* + * We've just COWed this block, it should have updated + * the correct backref node entry. + */ + ASSERT(node->eb == eb); } else { + struct btrfs_ref ref = { + .action = BTRFS_ADD_DELAYED_REF, + .bytenr = node->eb->start, + .num_bytes = blocksize, + .parent = upper->eb->start, + .owning_root = btrfs_header_owner(upper->eb), + .ref_root = btrfs_header_owner(upper->eb), + }; + btrfs_set_node_blockptr(upper->eb, slot, node->eb->start); btrfs_set_node_ptr_generation(upper->eb, slot, trans->transid); - btrfs_mark_buffer_dirty(upper->eb); + btrfs_mark_buffer_dirty(trans, upper->eb); - btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, - node->eb->start, blocksize, - upper->eb->start); - ref.real_root = root->root_key.objectid; btrfs_init_tree_ref(&ref, node->level, - btrfs_header_owner(upper->eb)); + btrfs_root_id(root), false); ret = btrfs_inc_extent_ref(trans, &ref); - BUG_ON(ret); - - ret = btrfs_drop_subtree(trans, root, eb, upper->eb); - BUG_ON(ret); + if (!ret) + ret = btrfs_drop_subtree(trans, root, eb, + upper->eb); + if (unlikely(ret)) + btrfs_abort_transaction(trans, ret); } next: if (!upper->pending) btrfs_backref_drop_node_buffer(upper); else btrfs_backref_unlock_node_buffer(upper); - if (err) + if (ret) break; } - if (!err && node->pending) { + if (!ret && node->pending) { btrfs_backref_drop_node_buffer(node); - list_move_tail(&node->list, &rc->backref_cache.changed); + list_del_init(&node->list); node->pending = 0; } path->lowest_level = 0; - BUG_ON(err == -ENOSPC); - return err; + + /* + * We should have allocated all of our space in the block rsv and thus + * shouldn't ENOSPC. + */ + ASSERT(ret != -ENOSPC); + return ret; } static int link_to_upper(struct btrfs_trans_handle *trans, @@ -2370,8 +2368,8 @@ static int finish_pending_nodes(struct btrfs_trans_handle *trans, for (level = 0; level < BTRFS_MAX_LEVEL; level++) { while (!list_empty(&cache->pending[level])) { - node = list_entry(cache->pending[level].next, - struct btrfs_backref_node, list); + node = list_first_entry(&cache->pending[level], + struct btrfs_backref_node, list); list_move_tail(&node->list, &list); BUG_ON(!node->pending); @@ -2409,8 +2407,8 @@ static void update_processed_blocks(struct reloc_control *rc, if (list_empty(&next->upper)) break; - edge = list_entry(next->upper.next, - struct btrfs_backref_edge, list[LOWER]); + edge = list_first_entry(&next->upper, struct btrfs_backref_edge, + list[LOWER]); edges[index++] = edge; next = edge->node[UPPER]; } @@ -2422,8 +2420,8 @@ static int tree_block_processed(u64 bytenr, struct reloc_control *rc) { u32 blocksize = rc->extent_root->fs_info->nodesize; - if (test_range_bit(&rc->processed_blocks, bytenr, - bytenr + blocksize - 1, EXTENT_DIRTY, 1, NULL)) + if (btrfs_test_range_bit(&rc->processed_blocks, bytenr, + bytenr + blocksize - 1, EXTENT_DIRTY, NULL)) return 1; return 0; } @@ -2431,13 +2429,17 @@ static int tree_block_processed(u64 bytenr, struct reloc_control *rc) static int get_tree_block_key(struct btrfs_fs_info *fs_info, struct tree_block *block) { + struct btrfs_tree_parent_check check = { + .level = block->level, + .owner_root = block->owner, + .transid = block->key.offset + }; struct extent_buffer *eb; - eb = read_tree_block(fs_info, block->bytenr, block->key.offset, - block->level, NULL); - if (IS_ERR(eb)) { + eb = read_tree_block(fs_info, block->bytenr, &check); + if (IS_ERR(eb)) return PTR_ERR(eb); - } else if (!extent_buffer_uptodate(eb)) { + if (unlikely(!extent_buffer_uptodate(eb))) { free_extent_buffer(eb); return -EIO; } @@ -2446,7 +2448,7 @@ static int get_tree_block_key(struct btrfs_fs_info *fs_info, else btrfs_node_key_to_cpu(eb, &block->key, 0); free_extent_buffer(eb); - block->key_ready = 1; + block->key_ready = true; return 0; } @@ -2475,28 +2477,62 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans, BUG_ON(node->processed); root = select_one_root(node); - if (root == ERR_PTR(-ENOENT)) { - update_processed_blocks(rc, node); + if (IS_ERR(root)) { + ret = PTR_ERR(root); + + /* See explanation in select_one_root for the -EUCLEAN case. */ + ASSERT(ret == -ENOENT); + if (ret == -ENOENT) { + ret = 0; + update_processed_blocks(rc, node); + } goto out; } if (root) { if (test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) { - BUG_ON(node->new_bytenr); - BUG_ON(!list_empty(&node->list)); - btrfs_record_root_in_trans(trans, root); + /* + * This block was the root block of a root, and this is + * the first time we're processing the block and thus it + * should not have had the ->new_bytenr modified. + * + * However in the case of corruption we could have + * multiple refs pointing to the same block improperly, + * and thus we would trip over these checks. ASSERT() + * for the developer case, because it could indicate a + * bug in the backref code, however error out for a + * normal user in the case of corruption. + */ + ASSERT(node->new_bytenr == 0); + if (unlikely(node->new_bytenr)) { + btrfs_err(root->fs_info, + "bytenr %llu has improper references to it", + node->bytenr); + ret = -EUCLEAN; + goto out; + } + ret = btrfs_record_root_in_trans(trans, root); + if (ret) + goto out; + /* + * Another thread could have failed, need to check if we + * have reloc_root actually set. + */ + if (!root->reloc_root) { + ret = -ENOENT; + goto out; + } root = root->reloc_root; node->new_bytenr = root->node->start; btrfs_put_root(node->root); node->root = btrfs_grab_root(root); ASSERT(node->root); - list_add_tail(&node->list, &rc->backref_cache.changed); } else { - path->lowest_level = node->level; - ret = btrfs_search_slot(trans, root, key, path, 0, 1); - btrfs_release_path(path); - if (ret > 0) - ret = 0; + btrfs_err(root->fs_info, + "bytenr %llu resolved to a non-shareable root", + node->bytenr); + ret = -EUCLEAN; + goto out; } if (!ret) update_processed_blocks(rc, node); @@ -2504,11 +2540,50 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans, ret = do_relocation(trans, rc, node, key, path, 1); } out: - if (ret || node->level == 0 || node->cowonly) + if (ret || node->level == 0) btrfs_backref_cleanup_node(&rc->backref_cache, node); return ret; } +static int relocate_cowonly_block(struct btrfs_trans_handle *trans, + struct reloc_control *rc, struct tree_block *block, + struct btrfs_path *path) +{ + struct btrfs_fs_info *fs_info = trans->fs_info; + struct btrfs_root *root; + u64 num_bytes; + int nr_levels; + int ret; + + root = btrfs_get_fs_root(fs_info, block->owner, true); + if (IS_ERR(root)) + return PTR_ERR(root); + + nr_levels = max(btrfs_header_level(root->node) - block->level, 0) + 1; + + num_bytes = fs_info->nodesize * nr_levels; + ret = refill_metadata_space(trans, rc, num_bytes); + if (ret) { + btrfs_put_root(root); + return ret; + } + path->lowest_level = block->level; + if (root == root->fs_info->chunk_root) + btrfs_reserve_chunk_metadata(trans, false); + + ret = btrfs_search_slot(trans, root, &block->key, path, 0, 1); + path->lowest_level = 0; + btrfs_release_path(path); + + if (root == root->fs_info->chunk_root) + btrfs_trans_release_chunk_metadata(trans); + if (ret > 0) + ret = 0; + btrfs_put_root(root); + + return ret; +} + /* * relocate a list of blocks */ @@ -2521,300 +2596,427 @@ int relocate_tree_blocks(struct btrfs_trans_handle *trans, struct btrfs_path *path; struct tree_block *block; struct tree_block *next; - int ret; - int err = 0; + int ret = 0; path = btrfs_alloc_path(); if (!path) { - err = -ENOMEM; + ret = -ENOMEM; goto out_free_blocks; } /* Kick in readahead for tree blocks with missing keys */ rbtree_postorder_for_each_entry_safe(block, next, blocks, rb_node) { if (!block->key_ready) - readahead_tree_block(fs_info, block->bytenr); + btrfs_readahead_tree_block(fs_info, block->bytenr, + block->owner, 0, + block->level); } /* Get first keys */ rbtree_postorder_for_each_entry_safe(block, next, blocks, rb_node) { if (!block->key_ready) { - err = get_tree_block_key(fs_info, block); - if (err) + ret = get_tree_block_key(fs_info, block); + if (ret) goto out_free_path; } } /* Do tree relocation */ rbtree_postorder_for_each_entry_safe(block, next, blocks, rb_node) { - node = build_backref_tree(rc, &block->key, + /* + * For COWonly blocks, or the data reloc tree, we only need to + * COW down to the block, there's no need to generate a backref + * tree. + */ + if (block->owner && + (!btrfs_is_fstree(block->owner) || + block->owner == BTRFS_DATA_RELOC_TREE_OBJECTID)) { + ret = relocate_cowonly_block(trans, rc, block, path); + if (ret) + break; + continue; + } + + node = build_backref_tree(trans, rc, &block->key, block->level, block->bytenr); if (IS_ERR(node)) { - err = PTR_ERR(node); + ret = PTR_ERR(node); goto out; } ret = relocate_tree_block(trans, rc, node, &block->key, path); - if (ret < 0) { - err = ret; + if (ret < 0) break; - } } out: - err = finish_pending_nodes(trans, rc, path, err); + ret = finish_pending_nodes(trans, rc, path, ret); out_free_path: btrfs_free_path(path); out_free_blocks: free_block_list(blocks); - return err; + return ret; } -static noinline_for_stack -int prealloc_file_extent_cluster(struct inode *inode, - struct file_extent_cluster *cluster) +static noinline_for_stack int prealloc_file_extent_cluster(struct reloc_control *rc) { + const struct file_extent_cluster *cluster = &rc->cluster; + struct btrfs_inode *inode = BTRFS_I(rc->data_inode); u64 alloc_hint = 0; u64 start; u64 end; - u64 offset = BTRFS_I(inode)->index_cnt; + u64 offset = inode->reloc_block_group_start; u64 num_bytes; - int nr = 0; + int nr; int ret = 0; u64 prealloc_start = cluster->start - offset; u64 prealloc_end = cluster->end - offset; - u64 cur_offset; - struct extent_changeset *data_reserved = NULL; + u64 cur_offset = prealloc_start; - BUG_ON(cluster->start != cluster->boundary[0]); - inode_lock(inode); + /* + * For blocksize < folio size case (either bs < page size or large folios), + * beyond i_size, all blocks are filled with zero. + * + * If the current cluster covers the above range, btrfs_do_readpage() + * will skip the read, and relocate_one_folio() will later writeback + * the padding zeros as new data, causing data corruption. + * + * Here we have to invalidate the cache covering our cluster. + */ + ret = filemap_invalidate_inode(&inode->vfs_inode, true, prealloc_start, + prealloc_end); + if (ret < 0) + return ret; - ret = btrfs_check_data_free_space(inode, &data_reserved, prealloc_start, - prealloc_end + 1 - prealloc_start); + BUG_ON(cluster->start != cluster->boundary[0]); + ret = btrfs_alloc_data_chunk_ondemand(inode, + prealloc_end + 1 - prealloc_start); if (ret) - goto out; + return ret; + + btrfs_inode_lock(inode, 0); + for (nr = 0; nr < cluster->nr; nr++) { + struct extent_state *cached_state = NULL; - cur_offset = prealloc_start; - while (nr < cluster->nr) { start = cluster->boundary[nr] - offset; if (nr + 1 < cluster->nr) end = cluster->boundary[nr + 1] - 1 - offset; else end = cluster->end - offset; - lock_extent(&BTRFS_I(inode)->io_tree, start, end); + btrfs_lock_extent(&inode->io_tree, start, end, &cached_state); num_bytes = end + 1 - start; - if (cur_offset < start) - btrfs_free_reserved_data_space(inode, data_reserved, - cur_offset, start - cur_offset); - ret = btrfs_prealloc_file_range(inode, 0, start, + ret = btrfs_prealloc_file_range(&inode->vfs_inode, 0, start, num_bytes, num_bytes, end + 1, &alloc_hint); cur_offset = end + 1; - unlock_extent(&BTRFS_I(inode)->io_tree, start, end); + btrfs_unlock_extent(&inode->io_tree, start, end, &cached_state); if (ret) break; - nr++; } + btrfs_inode_unlock(inode, 0); + if (cur_offset < prealloc_end) - btrfs_free_reserved_data_space(inode, data_reserved, - cur_offset, prealloc_end + 1 - cur_offset); -out: - inode_unlock(inode); - extent_changeset_free(data_reserved); + btrfs_free_reserved_data_space_noquota(inode, + prealloc_end + 1 - cur_offset); return ret; } -static noinline_for_stack -int setup_extent_mapping(struct inode *inode, u64 start, u64 end, - u64 block_start) +static noinline_for_stack int setup_relocation_extent_mapping(struct reloc_control *rc) { - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct btrfs_inode *inode = BTRFS_I(rc->data_inode); struct extent_map *em; + struct extent_state *cached_state = NULL; + u64 offset = inode->reloc_block_group_start; + u64 start = rc->cluster.start - offset; + u64 end = rc->cluster.end - offset; int ret = 0; - em = alloc_extent_map(); + em = btrfs_alloc_extent_map(); if (!em) return -ENOMEM; em->start = start; em->len = end + 1 - start; - em->block_len = em->len; - em->block_start = block_start; - set_bit(EXTENT_FLAG_PINNED, &em->flags); + em->disk_bytenr = rc->cluster.start; + em->disk_num_bytes = em->len; + em->ram_bytes = em->len; + em->flags |= EXTENT_FLAG_PINNED; + + btrfs_lock_extent(&inode->io_tree, start, end, &cached_state); + ret = btrfs_replace_extent_map_range(inode, em, false); + btrfs_unlock_extent(&inode->io_tree, start, end, &cached_state); + btrfs_free_extent_map(em); - lock_extent(&BTRFS_I(inode)->io_tree, start, end); - while (1) { - write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em, 0); - write_unlock(&em_tree->lock); - if (ret != -EEXIST) { - free_extent_map(em); - break; - } - btrfs_drop_extent_cache(BTRFS_I(inode), start, end, 0); - } - unlock_extent(&BTRFS_I(inode)->io_tree, start, end); return ret; } /* - * Allow error injection to test balance cancellation + * Allow error injection to test balance/relocation cancellation */ -int btrfs_should_cancel_balance(struct btrfs_fs_info *fs_info) +noinline int btrfs_should_cancel_balance(const struct btrfs_fs_info *fs_info) { - return atomic_read(&fs_info->balance_cancel_req); + return atomic_read(&fs_info->balance_cancel_req) || + atomic_read(&fs_info->reloc_cancel_req) || + fatal_signal_pending(current); } ALLOW_ERROR_INJECTION(btrfs_should_cancel_balance, TRUE); -static int relocate_file_extent_cluster(struct inode *inode, - struct file_extent_cluster *cluster) -{ - struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); - u64 page_start; - u64 page_end; - u64 offset = BTRFS_I(inode)->index_cnt; - unsigned long index; - unsigned long last_index; - struct page *page; - struct file_ra_state *ra; +static u64 get_cluster_boundary_end(const struct file_extent_cluster *cluster, + int cluster_nr) +{ + /* Last extent, use cluster end directly */ + if (cluster_nr >= cluster->nr - 1) + return cluster->end; + + /* Use next boundary start*/ + return cluster->boundary[cluster_nr + 1] - 1; +} + +static int relocate_one_folio(struct reloc_control *rc, + struct file_ra_state *ra, + int *cluster_nr, u64 *file_offset_ret) +{ + const struct file_extent_cluster *cluster = &rc->cluster; + struct inode *inode = rc->data_inode; + struct btrfs_fs_info *fs_info = inode_to_fs_info(inode); + const u64 orig_file_offset = *file_offset_ret; + u64 offset = BTRFS_I(inode)->reloc_block_group_start; + const pgoff_t last_index = (cluster->end - offset) >> PAGE_SHIFT; + const pgoff_t index = orig_file_offset >> PAGE_SHIFT; gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); - int nr = 0; - int ret = 0; + struct folio *folio; + u64 folio_start; + u64 folio_end; + u64 cur; + int ret; + const bool use_rst = btrfs_need_stripe_tree_update(fs_info, rc->block_group->flags); - if (!cluster->nr) - return 0; + ASSERT(index <= last_index); +again: + folio = filemap_lock_folio(inode->i_mapping, index); + if (IS_ERR(folio)) { - ra = kzalloc(sizeof(*ra), GFP_NOFS); - if (!ra) - return -ENOMEM; + /* + * On relocation we're doing readahead on the relocation inode, + * but if the filesystem is backed by a RAID stripe tree we can + * get ENOENT (e.g. due to preallocated extents not being + * mapped in the RST) from the lookup. + * + * But readahead doesn't handle the error and submits invalid + * reads to the device, causing a assertion failures. + */ + if (!use_rst) + page_cache_sync_readahead(inode->i_mapping, ra, NULL, + index, last_index + 1 - index); + folio = __filemap_get_folio(inode->i_mapping, index, + FGP_LOCK | FGP_ACCESSED | FGP_CREAT, + mask); + if (IS_ERR(folio)) + return PTR_ERR(folio); + } - ret = prealloc_file_extent_cluster(inode, cluster); - if (ret) - goto out; + if (folio_test_readahead(folio) && !use_rst) + page_cache_async_readahead(inode->i_mapping, ra, NULL, + folio, last_index + 1 - index); - file_ra_state_init(ra, inode->i_mapping); + if (!folio_test_uptodate(folio)) { + btrfs_read_folio(NULL, folio); + folio_lock(folio); + if (unlikely(!folio_test_uptodate(folio))) { + ret = -EIO; + goto release_folio; + } + if (folio->mapping != inode->i_mapping) { + folio_unlock(folio); + folio_put(folio); + goto again; + } + } - ret = setup_extent_mapping(inode, cluster->start - offset, - cluster->end - offset, cluster->start); - if (ret) - goto out; + /* + * We could have lost folio private when we dropped the lock to read the + * folio above, make sure we set_folio_extent_mapped() here so we have any + * of the subpage blocksize stuff we need in place. + */ + ret = set_folio_extent_mapped(folio); + if (ret < 0) + goto release_folio; + + folio_start = folio_pos(folio); + folio_end = folio_start + folio_size(folio) - 1; - index = (cluster->start - offset) >> PAGE_SHIFT; - last_index = (cluster->end - offset) >> PAGE_SHIFT; - while (index <= last_index) { + /* + * Start from the cluster, as for subpage case, the cluster can start + * inside the folio. + */ + cur = max(folio_start, cluster->boundary[*cluster_nr] - offset); + while (cur <= folio_end) { + struct extent_state *cached_state = NULL; + u64 extent_start = cluster->boundary[*cluster_nr] - offset; + u64 extent_end = get_cluster_boundary_end(cluster, + *cluster_nr) - offset; + u64 clamped_start = max(folio_start, extent_start); + u64 clamped_end = min(folio_end, extent_end); + u32 clamped_len = clamped_end + 1 - clamped_start; + + /* Reserve metadata for this range */ ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), - PAGE_SIZE); + clamped_len, clamped_len, + false); if (ret) - goto out; + goto release_folio; - page = find_lock_page(inode->i_mapping, index); - if (!page) { - page_cache_sync_readahead(inode->i_mapping, - ra, NULL, index, - last_index + 1 - index); - page = find_or_create_page(inode->i_mapping, index, - mask); - if (!page) { - btrfs_delalloc_release_metadata(BTRFS_I(inode), - PAGE_SIZE, true); - btrfs_delalloc_release_extents(BTRFS_I(inode), - PAGE_SIZE); - ret = -ENOMEM; - goto out; - } + /* Mark the range delalloc and dirty for later writeback */ + btrfs_lock_extent(&BTRFS_I(inode)->io_tree, clamped_start, + clamped_end, &cached_state); + ret = btrfs_set_extent_delalloc(BTRFS_I(inode), clamped_start, + clamped_end, 0, &cached_state); + if (ret) { + btrfs_clear_extent_bit(&BTRFS_I(inode)->io_tree, + clamped_start, clamped_end, + EXTENT_LOCKED | EXTENT_BOUNDARY, + &cached_state); + btrfs_delalloc_release_metadata(BTRFS_I(inode), + clamped_len, true); + btrfs_delalloc_release_extents(BTRFS_I(inode), + clamped_len); + goto release_folio; } + btrfs_folio_set_dirty(fs_info, folio, clamped_start, clamped_len); - if (PageReadahead(page)) { - page_cache_async_readahead(inode->i_mapping, - ra, NULL, page, index, - last_index + 1 - index); + /* + * Set the boundary if it's inside the folio. + * Data relocation requires the destination extents to have the + * same size as the source. + * EXTENT_BOUNDARY bit prevents current extent from being merged + * with previous extent. + */ + if (in_range(cluster->boundary[*cluster_nr] - offset, + folio_start, folio_size(folio))) { + u64 boundary_start = cluster->boundary[*cluster_nr] - + offset; + u64 boundary_end = boundary_start + + fs_info->sectorsize - 1; + + btrfs_set_extent_bit(&BTRFS_I(inode)->io_tree, + boundary_start, boundary_end, + EXTENT_BOUNDARY, NULL); } - - if (!PageUptodate(page)) { - btrfs_readpage(NULL, page); - lock_page(page); - if (!PageUptodate(page)) { - unlock_page(page); - put_page(page); - btrfs_delalloc_release_metadata(BTRFS_I(inode), - PAGE_SIZE, true); - btrfs_delalloc_release_extents(BTRFS_I(inode), - PAGE_SIZE); - ret = -EIO; - goto out; - } + btrfs_unlock_extent(&BTRFS_I(inode)->io_tree, clamped_start, clamped_end, + &cached_state); + btrfs_delalloc_release_extents(BTRFS_I(inode), clamped_len); + cur += clamped_len; + + /* Crossed extent end, go to next extent */ + if (cur >= extent_end) { + (*cluster_nr)++; + /* Just finished the last extent of the cluster, exit. */ + if (*cluster_nr >= cluster->nr) + break; } + } + folio_unlock(folio); + folio_put(folio); + + balance_dirty_pages_ratelimited(inode->i_mapping); + btrfs_throttle(fs_info); + if (btrfs_should_cancel_balance(fs_info)) + ret = -ECANCELED; + *file_offset_ret = folio_end + 1; + return ret; - page_start = page_offset(page); - page_end = page_start + PAGE_SIZE - 1; - - lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end); +release_folio: + folio_unlock(folio); + folio_put(folio); + return ret; +} - set_page_extent_mapped(page); +static int relocate_file_extent_cluster(struct reloc_control *rc) +{ + struct inode *inode = rc->data_inode; + const struct file_extent_cluster *cluster = &rc->cluster; + u64 offset = BTRFS_I(inode)->reloc_block_group_start; + u64 cur_file_offset = cluster->start - offset; + struct file_ra_state AUTO_KFREE(ra); + int cluster_nr = 0; + int ret = 0; - if (nr < cluster->nr && - page_start + offset == cluster->boundary[nr]) { - set_extent_bits(&BTRFS_I(inode)->io_tree, - page_start, page_end, - EXTENT_BOUNDARY); - nr++; - } + if (!cluster->nr) + return 0; - ret = btrfs_set_extent_delalloc(inode, page_start, page_end, 0, - NULL); - if (ret) { - unlock_page(page); - put_page(page); - btrfs_delalloc_release_metadata(BTRFS_I(inode), - PAGE_SIZE, true); - btrfs_delalloc_release_extents(BTRFS_I(inode), - PAGE_SIZE); + ra = kzalloc(sizeof(*ra), GFP_NOFS); + if (!ra) + return -ENOMEM; - clear_extent_bits(&BTRFS_I(inode)->io_tree, - page_start, page_end, - EXTENT_LOCKED | EXTENT_BOUNDARY); - goto out; + ret = prealloc_file_extent_cluster(rc); + if (ret) + return ret; - } - set_page_dirty(page); + file_ra_state_init(ra, inode->i_mapping); - unlock_extent(&BTRFS_I(inode)->io_tree, - page_start, page_end); - unlock_page(page); - put_page(page); + ret = setup_relocation_extent_mapping(rc); + if (ret) + return ret; - index++; - btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE); - balance_dirty_pages_ratelimited(inode->i_mapping); - btrfs_throttle(fs_info); - if (btrfs_should_cancel_balance(fs_info)) { - ret = -ECANCELED; - goto out; - } + while (cur_file_offset < cluster->end - offset) { + ret = relocate_one_folio(rc, ra, &cluster_nr, &cur_file_offset); + if (ret) + break; } - WARN_ON(nr != cluster->nr); -out: - kfree(ra); + if (ret == 0) + WARN_ON(cluster_nr != cluster->nr); return ret; } -static noinline_for_stack -int relocate_data_extent(struct inode *inode, struct btrfs_key *extent_key, - struct file_extent_cluster *cluster) +static noinline_for_stack int relocate_data_extent(struct reloc_control *rc, + const struct btrfs_key *extent_key) { + struct inode *inode = rc->data_inode; + struct file_extent_cluster *cluster = &rc->cluster; int ret; + struct btrfs_root *root = BTRFS_I(inode)->root; if (cluster->nr > 0 && extent_key->objectid != cluster->end + 1) { - ret = relocate_file_extent_cluster(inode, cluster); + ret = relocate_file_extent_cluster(rc); if (ret) return ret; cluster->nr = 0; } - if (!cluster->nr) + /* + * Under simple quotas, we set root->relocation_src_root when we find + * the extent. If adjacent extents have different owners, we can't merge + * them while relocating. Handle this by storing the owning root that + * started a cluster and if we see an extent from a different root break + * cluster formation (just like the above case of non-adjacent extents). + * + * Without simple quotas, relocation_src_root is always 0, so we should + * never see a mismatch, and it should have no effect on relocation + * clusters. + */ + if (cluster->nr > 0 && cluster->owning_root != root->relocation_src_root) { + u64 tmp = root->relocation_src_root; + + /* + * root->relocation_src_root is the state that actually affects + * the preallocation we do here, so set it to the root owning + * the cluster we need to relocate. + */ + root->relocation_src_root = cluster->owning_root; + ret = relocate_file_extent_cluster(rc); + if (ret) + return ret; + cluster->nr = 0; + /* And reset it back for the current extent's owning root. */ + root->relocation_src_root = tmp; + } + + if (!cluster->nr) { cluster->start = extent_key->objectid; + cluster->owning_root = root->relocation_src_root; + } else BUG_ON(cluster->nr >= MAX_EXTENTS); cluster->end = extent_key->objectid + extent_key->offset - 1; @@ -2822,7 +3024,7 @@ int relocate_data_extent(struct inode *inode, struct btrfs_key *extent_key, cluster->nr++; if (cluster->nr >= MAX_EXTENTS) { - ret = relocate_file_extent_cluster(inode, cluster); + ret = relocate_file_extent_cluster(rc); if (ret) return ret; cluster->nr = 0; @@ -2835,7 +3037,7 @@ int relocate_data_extent(struct inode *inode, struct btrfs_key *extent_key, * the major work is getting the generation and level of the block */ static int add_tree_block(struct reloc_control *rc, - struct btrfs_key *extent_key, + const struct btrfs_key *extent_key, struct btrfs_path *path, struct rb_root *blocks) { @@ -2847,27 +3049,65 @@ static int add_tree_block(struct reloc_control *rc, u32 item_size; int level = -1; u64 generation; + u64 owner = 0; eb = path->nodes[0]; - item_size = btrfs_item_size_nr(eb, path->slots[0]); + item_size = btrfs_item_size(eb, path->slots[0]); if (extent_key->type == BTRFS_METADATA_ITEM_KEY || item_size >= sizeof(*ei) + sizeof(*bi)) { + unsigned long ptr = 0, end; + ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item); + end = (unsigned long)ei + item_size; if (extent_key->type == BTRFS_EXTENT_ITEM_KEY) { bi = (struct btrfs_tree_block_info *)(ei + 1); level = btrfs_tree_block_level(eb, bi); + ptr = (unsigned long)(bi + 1); } else { level = (int)extent_key->offset; + ptr = (unsigned long)(ei + 1); } generation = btrfs_extent_generation(eb, ei); - } else if (unlikely(item_size == sizeof(struct btrfs_extent_item_v0))) { - btrfs_print_v0_err(eb->fs_info); - btrfs_handle_fs_error(eb->fs_info, -EINVAL, NULL); - return -EINVAL; + + /* + * We're reading random blocks without knowing their owner ahead + * of time. This is ok most of the time, as all reloc roots and + * fs roots have the same lock type. However normal trees do + * not, and the only way to know ahead of time is to read the + * inline ref offset. We know it's an fs root if + * + * 1. There's more than one ref. + * 2. There's a SHARED_DATA_REF_KEY set. + * 3. FULL_BACKREF is set on the flags. + * + * Otherwise it's safe to assume that the ref offset == the + * owner of this block, so we can use that when calling + * read_tree_block. + */ + if (btrfs_extent_refs(eb, ei) == 1 && + !(btrfs_extent_flags(eb, ei) & + BTRFS_BLOCK_FLAG_FULL_BACKREF) && + ptr < end) { + struct btrfs_extent_inline_ref *iref; + int type; + + iref = (struct btrfs_extent_inline_ref *)ptr; + type = btrfs_get_extent_inline_ref_type(eb, iref, + BTRFS_REF_TYPE_BLOCK); + if (type == BTRFS_REF_TYPE_INVALID) + return -EINVAL; + if (type == BTRFS_TREE_BLOCK_REF_KEY) + owner = btrfs_extent_inline_ref_offset(eb, iref); + } } else { - BUG(); + btrfs_print_leaf(eb); + btrfs_err(rc->block_group->fs_info, + "unrecognized tree backref at tree block %llu slot %u", + eb->start, path->slots[0]); + btrfs_release_path(path); + return -EUCLEAN; } btrfs_release_path(path); @@ -2882,9 +3122,10 @@ static int add_tree_block(struct reloc_control *rc, block->key.objectid = rc->extent_root->fs_info->nodesize; block->key.offset = generation; block->level = level; - block->key_ready = 0; + block->key_ready = false; + block->owner = owner; - rb_node = rb_simple_insert(blocks, block->bytenr, &block->rb_node); + rb_node = rb_simple_insert(blocks, &block->simple_node); if (rb_node) btrfs_backref_panic(rc->extent_root->fs_info, block->bytenr, -EEXIST); @@ -2900,7 +3141,7 @@ static int __add_tree_block(struct reloc_control *rc, struct rb_root *blocks) { struct btrfs_fs_info *fs_info = rc->extent_root->fs_info; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_key key; int ret; bool skinny = btrfs_fs_incompat(fs_info, SKINNY_METADATA); @@ -2924,11 +3165,11 @@ again: key.offset = blocksize; } - path->search_commit_root = 1; - path->skip_locking = 1; + path->search_commit_root = true; + path->skip_locking = true; ret = btrfs_search_slot(NULL, rc->extent_root, &key, path, 0, 0); if (ret < 0) - goto out; + return ret; if (ret > 0 && skinny) { if (path->slots[0]) { @@ -2955,31 +3196,29 @@ again: "tree block extent item (%llu) is not found in extent tree", bytenr); WARN_ON(1); - ret = -EINVAL; - goto out; + return -EINVAL; } - ret = add_tree_block(rc, &key, path, blocks); -out: - btrfs_free_path(path); - return ret; + return add_tree_block(rc, &key, path, blocks); } -static int delete_block_group_cache(struct btrfs_fs_info *fs_info, - struct btrfs_block_group *block_group, +static int delete_block_group_cache(struct btrfs_block_group *block_group, struct inode *inode, u64 ino) { + struct btrfs_fs_info *fs_info = block_group->fs_info; struct btrfs_root *root = fs_info->tree_root; struct btrfs_trans_handle *trans; + struct btrfs_inode *btrfs_inode; int ret = 0; if (inode) goto truncate; - inode = btrfs_iget(fs_info->sb, ino, root); - if (IS_ERR(inode)) + btrfs_inode = btrfs_iget(ino, root); + if (IS_ERR(btrfs_inode)) return -ENOENT; + inode = &btrfs_inode->vfs_inode; truncate: ret = btrfs_check_trunc_cache_free_space(fs_info, @@ -3021,11 +3260,16 @@ static int delete_v1_space_cache(struct extent_buffer *leaf, return 0; for (i = 0; i < btrfs_header_nritems(leaf); i++) { + u8 type; + btrfs_item_key_to_cpu(leaf, &key, i); if (key.type != BTRFS_EXTENT_DATA_KEY) continue; ei = btrfs_item_ptr(leaf, i, struct btrfs_file_extent_item); - if (btrfs_file_extent_type(leaf, ei) == BTRFS_FILE_EXTENT_REG && + type = btrfs_file_extent_type(leaf, ei); + + if ((type == BTRFS_FILE_EXTENT_REG || + type == BTRFS_FILE_EXTENT_PREALLOC) && btrfs_file_extent_disk_bytenr(leaf, ei) == data_bytenr) { found = true; space_cache_ino = key.objectid; @@ -3034,38 +3278,40 @@ static int delete_v1_space_cache(struct extent_buffer *leaf, } if (!found) return -ENOENT; - ret = delete_block_group_cache(leaf->fs_info, block_group, NULL, - space_cache_ino); + ret = delete_block_group_cache(block_group, NULL, space_cache_ino); return ret; } /* * helper to find all tree blocks that reference a given data extent */ -static noinline_for_stack -int add_data_references(struct reloc_control *rc, - struct btrfs_key *extent_key, - struct btrfs_path *path, - struct rb_root *blocks) +static noinline_for_stack int add_data_references(struct reloc_control *rc, + const struct btrfs_key *extent_key, + struct btrfs_path *path, + struct rb_root *blocks) { - struct btrfs_fs_info *fs_info = rc->extent_root->fs_info; - struct ulist *leaves = NULL; + struct btrfs_backref_walk_ctx ctx = { 0 }; struct ulist_iterator leaf_uiter; struct ulist_node *ref_node = NULL; - const u32 blocksize = fs_info->nodesize; + const u32 blocksize = rc->extent_root->fs_info->nodesize; int ret = 0; btrfs_release_path(path); - ret = btrfs_find_all_leafs(NULL, fs_info, extent_key->objectid, - 0, &leaves, NULL, true); + + ctx.bytenr = extent_key->objectid; + ctx.skip_inode_ref_list = true; + ctx.fs_info = rc->extent_root->fs_info; + + ret = btrfs_find_all_leafs(&ctx); if (ret < 0) return ret; ULIST_ITER_INIT(&leaf_uiter); - while ((ref_node = ulist_next(leaves, &leaf_uiter))) { + while ((ref_node = ulist_next(ctx.refs, &leaf_uiter))) { + struct btrfs_tree_parent_check check = { 0 }; struct extent_buffer *eb; - eb = read_tree_block(fs_info, ref_node->val, 0, 0, NULL); + eb = read_tree_block(ctx.fs_info, ref_node->val, &check); if (IS_ERR(eb)) { ret = PTR_ERR(eb); break; @@ -3081,7 +3327,7 @@ int add_data_references(struct reloc_control *rc, } if (ret < 0) free_block_list(blocks); - ulist_free(leaves); + ulist_free(ctx.refs); return ret; } @@ -3100,6 +3346,8 @@ int find_next_extent(struct reloc_control *rc, struct btrfs_path *path, last = rc->block_group->start + rc->block_group->length; while (1) { + bool block_found; + cond_resched(); if (rc->search_start >= last) { ret = 1; @@ -3110,8 +3358,8 @@ int find_next_extent(struct reloc_control *rc, struct btrfs_path *path, key.type = BTRFS_EXTENT_ITEM_KEY; key.offset = 0; - path->search_commit_root = 1; - path->skip_locking = 1; + path->search_commit_root = true; + path->skip_locking = true; ret = btrfs_search_slot(NULL, rc->extent_root, &key, path, 0, 0); if (ret < 0) @@ -3150,11 +3398,11 @@ next: goto next; } - ret = find_first_extent_bit(&rc->processed_blocks, - key.objectid, &start, &end, - EXTENT_DIRTY, NULL); + block_found = btrfs_find_first_extent_bit(&rc->processed_blocks, + key.objectid, &start, &end, + EXTENT_DIRTY, NULL); - if (ret == 0 && start <= key.objectid) { + if (block_found && start <= key.objectid) { btrfs_release_path(path); rc->search_start = end + 1; } else { @@ -3189,20 +3437,6 @@ static void unset_reloc_control(struct reloc_control *rc) mutex_unlock(&fs_info->reloc_mutex); } -static int check_extent_flags(u64 flags) -{ - if ((flags & BTRFS_EXTENT_FLAG_DATA) && - (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)) - return 1; - if (!(flags & BTRFS_EXTENT_FLAG_DATA) && - !(flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)) - return 1; - if ((flags & BTRFS_EXTENT_FLAG_DATA) && - (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) - return 1; - return 0; -} - static noinline_for_stack int prepare_to_relocate(struct reloc_control *rc) { @@ -3222,13 +3456,13 @@ int prepare_to_relocate(struct reloc_control *rc) rc->reserved_bytes = 0; rc->block_rsv->size = rc->extent_root->fs_info->nodesize * RELOCATION_RESERVED_NODES; - ret = btrfs_block_rsv_refill(rc->extent_root, + ret = btrfs_block_rsv_refill(rc->extent_root->fs_info, rc->block_rsv, rc->block_rsv->size, BTRFS_RESERVE_FLUSH_ALL); if (ret) return ret; - rc->create_reloc_tree = 1; + rc->create_reloc_tree = true; set_reloc_control(rc); trans = btrfs_join_transaction(rc->extent_root); @@ -3241,8 +3475,12 @@ int prepare_to_relocate(struct reloc_control *rc) */ return PTR_ERR(trans); } - btrfs_commit_transaction(trans); - return 0; + + ret = btrfs_commit_transaction(trans); + if (ret) + unset_reloc_control(rc); + + return ret; } static noinline_for_stack int relocate_block_group(struct reloc_control *rc) @@ -3251,10 +3489,9 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc) struct rb_root blocks = RB_ROOT; struct btrfs_key key; struct btrfs_trans_handle *trans = NULL; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_extent_item *ei; u64 flags; - u32 item_size; int ret; int err = 0; int progress = 0; @@ -3272,9 +3509,9 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc) while (1) { rc->reserved_bytes = 0; - ret = btrfs_block_rsv_refill(rc->extent_root, - rc->block_rsv, rc->block_rsv->size, - BTRFS_RESERVE_FLUSH_ALL); + ret = btrfs_block_rsv_refill(fs_info, rc->block_rsv, + rc->block_rsv->size, + BTRFS_RESERVE_FLUSH_ALL); if (ret) { err = ret; break; @@ -3287,11 +3524,9 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc) break; } restart: - if (update_backref_cache(trans, &rc->backref_cache)) { - btrfs_end_transaction(trans); - trans = NULL; - continue; - } + if (rc->backref_cache.last_trans != trans->transid) + btrfs_backref_release_cache(&rc->backref_cache); + rc->backref_cache.last_trans = trans->transid; ret = find_next_extent(rc, path, &key); if (ret < 0) @@ -3303,18 +3538,21 @@ restart: ei = btrfs_item_ptr(path->nodes[0], path->slots[0], struct btrfs_extent_item); - item_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]); - if (item_size >= sizeof(*ei)) { - flags = btrfs_extent_flags(path->nodes[0], ei); - ret = check_extent_flags(flags); - BUG_ON(ret); - } else if (unlikely(item_size == sizeof(struct btrfs_extent_item_v0))) { - err = -EINVAL; - btrfs_print_v0_err(trans->fs_info); - btrfs_abort_transaction(trans, err); - break; - } else { - BUG(); + flags = btrfs_extent_flags(path->nodes[0], ei); + + /* + * If we are relocating a simple quota owned extent item, we + * need to note the owner on the reloc data root so that when + * we allocate the replacement item, we can attribute it to the + * correct eventual owner (rather than the reloc data root). + */ + if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE) { + struct btrfs_root *root = BTRFS_I(rc->data_inode)->root; + u64 owning_root_id = btrfs_get_extent_owner_root(fs_info, + path->nodes[0], + path->slots[0]); + + root->relocation_src_root = owning_root_id; } if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) { @@ -3349,9 +3587,8 @@ restart: if (rc->stage == MOVE_DATA_EXTENTS && (flags & BTRFS_EXTENT_FLAG_DATA)) { - rc->found_file_extent = 1; - ret = relocate_data_extent(rc->data_inode, - &key, &rc->cluster); + rc->found_file_extent = true; + ret = relocate_data_extent(rc, &key); if (ret < 0) { err = ret; break; @@ -3372,7 +3609,7 @@ restart: } btrfs_release_path(path); - clear_extent_bits(&rc->processed_blocks, 0, (u64)-1, EXTENT_DIRTY); + btrfs_clear_extent_bit(&rc->processed_blocks, 0, (u64)-1, EXTENT_DIRTY, NULL); if (trans) { btrfs_end_transaction_throttle(trans); @@ -3380,13 +3617,12 @@ restart: } if (!err) { - ret = relocate_file_extent_cluster(rc->data_inode, - &rc->cluster); + ret = relocate_file_extent_cluster(rc); if (ret < 0) err = ret; } - rc->create_reloc_tree = 0; + rc->create_reloc_tree = false; set_reloc_control(rc); btrfs_backref_release_cache(&rc->backref_cache); @@ -3404,7 +3640,7 @@ restart: merge_reloc_roots(rc); - rc->merge_reloc_tree = 0; + rc->merge_reloc_tree = false; unset_reloc_control(rc); btrfs_block_rsv_release(fs_info, rc->block_rsv, (u64)-1, NULL); @@ -3414,20 +3650,21 @@ restart: err = PTR_ERR(trans); goto out_free; } - btrfs_commit_transaction(trans); + ret = btrfs_commit_transaction(trans); + if (ret && !err) + err = ret; out_free: ret = clean_dirty_subvols(rc); if (ret < 0 && !err) err = ret; btrfs_free_block_rsv(fs_info, rc->block_rsv); - btrfs_free_path(path); return err; } static int __insert_orphan_inode(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 objectid) { - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_inode_item *item; struct extent_buffer *leaf; int ret; @@ -3438,7 +3675,7 @@ static int __insert_orphan_inode(struct btrfs_trans_handle *trans, ret = btrfs_insert_empty_inode(trans, root, path, objectid); if (ret) - goto out; + return ret; leaf = path->nodes[0]; item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_inode_item); @@ -3448,25 +3685,50 @@ static int __insert_orphan_inode(struct btrfs_trans_handle *trans, btrfs_set_inode_mode(leaf, item, S_IFREG | 0600); btrfs_set_inode_flags(leaf, item, BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC); - btrfs_mark_buffer_dirty(leaf); + return 0; +} + +static void delete_orphan_inode(struct btrfs_trans_handle *trans, + struct btrfs_root *root, u64 objectid) +{ + BTRFS_PATH_AUTO_FREE(path); + struct btrfs_key key; + int ret = 0; + + path = btrfs_alloc_path(); + if (!path) { + ret = -ENOMEM; + goto out; + } + + key.objectid = objectid; + key.type = BTRFS_INODE_ITEM_KEY; + key.offset = 0; + ret = btrfs_search_slot(trans, root, &key, path, -1, 1); + if (ret) { + if (ret > 0) + ret = -ENOENT; + goto out; + } + ret = btrfs_del_item(trans, root, path); out: - btrfs_free_path(path); - return ret; + if (ret) + btrfs_abort_transaction(trans, ret); } /* * helper to create inode for data relocation. * the inode is in data relocation tree and its link count is 0 */ -static noinline_for_stack -struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info, - struct btrfs_block_group *group) +static noinline_for_stack struct inode *create_reloc_inode( + const struct btrfs_block_group *group) { - struct inode *inode = NULL; + struct btrfs_fs_info *fs_info = group->fs_info; + struct btrfs_inode *inode = NULL; struct btrfs_trans_handle *trans; struct btrfs_root *root; u64 objectid; - int err = 0; + int ret = 0; root = btrfs_grab_root(fs_info->data_reloc_root); trans = btrfs_start_transaction(root, 6); @@ -3475,28 +3737,76 @@ struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info, return ERR_CAST(trans); } - err = btrfs_find_free_objectid(root, &objectid); - if (err) + ret = btrfs_get_free_objectid(root, &objectid); + if (ret) goto out; - err = __insert_orphan_inode(trans, root, objectid); - BUG_ON(err); + ret = __insert_orphan_inode(trans, root, objectid); + if (ret) + goto out; - inode = btrfs_iget(fs_info->sb, objectid, root); - BUG_ON(IS_ERR(inode)); - BTRFS_I(inode)->index_cnt = group->start; + inode = btrfs_iget(objectid, root); + if (IS_ERR(inode)) { + delete_orphan_inode(trans, root, objectid); + ret = PTR_ERR(inode); + inode = NULL; + goto out; + } + inode->reloc_block_group_start = group->start; - err = btrfs_orphan_add(trans, BTRFS_I(inode)); + ret = btrfs_orphan_add(trans, inode); out: btrfs_put_root(root); btrfs_end_transaction(trans); btrfs_btree_balance_dirty(fs_info); - if (err) { + if (ret) { if (inode) - iput(inode); - inode = ERR_PTR(err); + iput(&inode->vfs_inode); + return ERR_PTR(ret); + } + return &inode->vfs_inode; +} + +/* + * Mark start of chunk relocation that is cancellable. Check if the cancellation + * has been requested meanwhile and don't start in that case. + * NOTE: if this returns an error, reloc_chunk_end() must not be called. + * + * Return: + * 0 success + * -EINPROGRESS operation is already in progress, that's probably a bug + * -ECANCELED cancellation request was set before the operation started + */ +static int reloc_chunk_start(struct btrfs_fs_info *fs_info) +{ + if (test_and_set_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags)) { + /* This should not happen */ + btrfs_err(fs_info, "reloc already running, cannot start"); + return -EINPROGRESS; } - return inode; + + if (atomic_read(&fs_info->reloc_cancel_req) > 0) { + btrfs_info(fs_info, "chunk relocation canceled on start"); + /* On cancel, clear all requests. */ + clear_and_wake_up_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags); + atomic_set(&fs_info->reloc_cancel_req, 0); + return -ECANCELED; + } + return 0; +} + +/* + * Mark end of chunk relocation that is cancellable and wake any waiters. + * NOTE: call only if a previous call to reloc_chunk_start() succeeded. + */ +static void reloc_chunk_end(struct btrfs_fs_info *fs_info) +{ + ASSERT(test_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags)); + /* Requested after start, clear bit first so any waiters can continue */ + if (atomic_read(&fs_info->reloc_cancel_req) > 0) + btrfs_info(fs_info, "chunk relocation canceled during operation"); + clear_and_wake_up_bit(BTRFS_FS_RELOC_RUNNING, &fs_info->flags); + atomic_set(&fs_info->reloc_cancel_req, 0); } static struct reloc_control *alloc_reloc_control(struct btrfs_fs_info *fs_info) @@ -3509,10 +3819,10 @@ static struct reloc_control *alloc_reloc_control(struct btrfs_fs_info *fs_info) INIT_LIST_HEAD(&rc->reloc_roots); INIT_LIST_HEAD(&rc->dirty_subvol_roots); - btrfs_backref_init_cache(fs_info, &rc->backref_cache, 1); - mapping_tree_init(&rc->reloc_root_tree); - extent_io_tree_init(fs_info, &rc->processed_blocks, - IO_TREE_RELOC_BLOCKS, NULL); + btrfs_backref_init_cache(fs_info, &rc->backref_cache, true); + rc->reloc_root_tree.rb_root = RB_ROOT; + spin_lock_init(&rc->reloc_root_tree.lock); + btrfs_extent_io_tree_init(fs_info, &rc->processed_blocks, IO_TREE_RELOC_BLOCKS); return rc; } @@ -3531,19 +3841,17 @@ static void free_reloc_control(struct reloc_control *rc) /* * Print the block group being relocated */ -static void describe_relocation(struct btrfs_fs_info *fs_info, - struct btrfs_block_group *block_group) +static void describe_relocation(struct btrfs_block_group *block_group) { - char buf[128] = {'\0'}; + char buf[128] = "NONE"; btrfs_describe_block_groups(block_group->flags, buf, sizeof(buf)); - btrfs_info(fs_info, - "relocating block group %llu flags %s", + btrfs_info(block_group->fs_info, "relocating block group %llu flags %s", block_group->start, buf); } -static const char *stage_to_string(int stage) +static const char *stage_to_string(enum reloc_stage stage) { if (stage == MOVE_DATA_EXTENTS) return "move data extents"; @@ -3555,21 +3863,45 @@ static const char *stage_to_string(int stage) /* * function to relocate all extents in a block group. */ -int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start) +int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start, + bool verbose) { struct btrfs_block_group *bg; - struct btrfs_root *extent_root = fs_info->extent_root; + struct btrfs_root *extent_root = btrfs_extent_root(fs_info, group_start); struct reloc_control *rc; struct inode *inode; struct btrfs_path *path; int ret; - int rw = 0; - int err = 0; + bool bg_is_ro = false; + + /* + * This only gets set if we had a half-deleted snapshot on mount. We + * cannot allow relocation to start while we're still trying to clean up + * these pending deletions. + */ + ret = wait_on_bit(&fs_info->flags, BTRFS_FS_UNFINISHED_DROPS, TASK_INTERRUPTIBLE); + if (ret) + return ret; + + /* We may have been woken up by close_ctree, so bail if we're closing. */ + if (btrfs_fs_closing(fs_info)) + return -EINTR; bg = btrfs_lookup_block_group(fs_info, group_start); if (!bg) return -ENOENT; + /* + * Relocation of a data block group creates ordered extents. Without + * sb_start_write(), we can freeze the filesystem while unfinished + * ordered extents are left. Such ordered extents can cause a deadlock + * e.g. when syncfs() is waiting for their completion but they can't + * finish because they block when joining a transaction, due to the + * fact that the freeze locks are being held in write mode. + */ + if (bg->flags & BTRFS_BLOCK_GROUP_DATA) + ASSERT(sb_write_started(fs_info->sb)); + if (btrfs_pinned_by_swapfile(fs_info, bg)) { btrfs_put_block_group(bg); return -ETXTBSY; @@ -3581,19 +3913,21 @@ int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start) return -ENOMEM; } + ret = reloc_chunk_start(fs_info); + if (ret < 0) + goto out_put_bg; + rc->extent_root = extent_root; rc->block_group = bg; ret = btrfs_inc_block_group_ro(rc->block_group, true); - if (ret) { - err = ret; + if (ret) goto out; - } - rw = 1; + bg_is_ro = true; path = btrfs_alloc_path(); if (!path) { - err = -ENOMEM; + ret = -ENOMEM; goto out; } @@ -3601,38 +3935,36 @@ int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start) btrfs_free_path(path); if (!IS_ERR(inode)) - ret = delete_block_group_cache(fs_info, rc->block_group, inode, 0); + ret = delete_block_group_cache(rc->block_group, inode, 0); else ret = PTR_ERR(inode); - if (ret && ret != -ENOENT) { - err = ret; + if (ret && ret != -ENOENT) goto out; - } - rc->data_inode = create_reloc_inode(fs_info, rc->block_group); + rc->data_inode = create_reloc_inode(rc->block_group); if (IS_ERR(rc->data_inode)) { - err = PTR_ERR(rc->data_inode); + ret = PTR_ERR(rc->data_inode); rc->data_inode = NULL; goto out; } - describe_relocation(fs_info, rc->block_group); + if (verbose) + describe_relocation(rc->block_group); btrfs_wait_block_group_reservations(rc->block_group); btrfs_wait_nocow_writers(rc->block_group); - btrfs_wait_ordered_roots(fs_info, U64_MAX, - rc->block_group->start, - rc->block_group->length); + btrfs_wait_ordered_roots(fs_info, U64_MAX, rc->block_group); + + ret = btrfs_zone_finish(rc->block_group); + WARN_ON(ret && ret != -EAGAIN); while (1) { - int finishes_stage; + enum reloc_stage finishes_stage; mutex_lock(&fs_info->cleaner_mutex); ret = relocate_block_group(rc); mutex_unlock(&fs_info->cleaner_mutex); - if (ret < 0) - err = ret; finishes_stage = rc->stage; /* @@ -3645,35 +3977,41 @@ int btrfs_relocate_block_group(struct btrfs_fs_info *fs_info, u64 group_start) * out of the loop if we hit an error. */ if (rc->stage == MOVE_DATA_EXTENTS && rc->found_file_extent) { - ret = btrfs_wait_ordered_range(rc->data_inode, 0, - (u64)-1); - if (ret) - err = ret; + int wb_ret; + + wb_ret = btrfs_wait_ordered_range(BTRFS_I(rc->data_inode), 0, + (u64)-1); + if (wb_ret && ret == 0) + ret = wb_ret; invalidate_mapping_pages(rc->data_inode->i_mapping, 0, -1); rc->stage = UPDATE_DATA_PTRS; } - if (err < 0) + if (ret < 0) goto out; if (rc->extents_found == 0) break; - btrfs_info(fs_info, "found %llu extents, stage: %s", - rc->extents_found, stage_to_string(finishes_stage)); + if (verbose) + btrfs_info(fs_info, "found %llu extents, stage: %s", + rc->extents_found, + stage_to_string(finishes_stage)); } WARN_ON(rc->block_group->pinned > 0); WARN_ON(rc->block_group->reserved > 0); WARN_ON(rc->block_group->used > 0); out: - if (err && rw) + if (ret && bg_is_ro) btrfs_dec_block_group_ro(rc->block_group); iput(rc->data_inode); - btrfs_put_block_group(rc->block_group); + reloc_chunk_end(fs_info); +out_put_bg: + btrfs_put_block_group(bg); free_reloc_control(rc); - return err; + return ret; } static noinline_for_stack int mark_garbage_root(struct btrfs_root *root) @@ -3688,7 +4026,7 @@ static noinline_for_stack int mark_garbage_root(struct btrfs_root *root) memset(&root->root_item.drop_progress, 0, sizeof(root->root_item.drop_progress)); - root->root_item.drop_level = 0; + btrfs_set_root_drop_level(&root->root_item, 0); btrfs_set_root_refs(&root->root_item, 0); ret = btrfs_update_root(trans, fs_info->tree_root, &root->root_key, &root->root_item); @@ -3705,9 +4043,8 @@ static noinline_for_stack int mark_garbage_root(struct btrfs_root *root) * this function resumes merging reloc trees with corresponding fs trees. * this is important for keeping the sharing of tree blocks */ -int btrfs_recover_relocation(struct btrfs_root *root) +int btrfs_recover_relocation(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_info *fs_info = root->fs_info; LIST_HEAD(reloc_roots); struct btrfs_key key; struct btrfs_root *fs_root; @@ -3716,8 +4053,8 @@ int btrfs_recover_relocation(struct btrfs_root *root) struct extent_buffer *leaf; struct reloc_control *rc = NULL; struct btrfs_trans_handle *trans; - int ret; - int err = 0; + int ret2; + int ret = 0; path = btrfs_alloc_path(); if (!path) @@ -3731,15 +4068,14 @@ int btrfs_recover_relocation(struct btrfs_root *root) while (1) { ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); - if (ret < 0) { - err = ret; + if (ret < 0) goto out; - } if (ret > 0) { if (path->slots[0] == 0) break; path->slots[0]--; } + ret = 0; leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); btrfs_release_path(path); @@ -3748,9 +4084,9 @@ int btrfs_recover_relocation(struct btrfs_root *root) key.type != BTRFS_ROOT_ITEM_KEY) break; - reloc_root = btrfs_read_tree_root(root, &key); + reloc_root = btrfs_read_tree_root(fs_info->tree_root, &key); if (IS_ERR(reloc_root)) { - err = PTR_ERR(reloc_root); + ret = PTR_ERR(reloc_root); goto out; } @@ -3762,15 +4098,12 @@ int btrfs_recover_relocation(struct btrfs_root *root) reloc_root->root_key.offset, false); if (IS_ERR(fs_root)) { ret = PTR_ERR(fs_root); - if (ret != -ENOENT) { - err = ret; + if (ret != -ENOENT) goto out; - } ret = mark_garbage_root(reloc_root); - if (ret < 0) { - err = ret; + if (ret < 0) goto out; - } + ret = 0; } else { btrfs_put_root(fs_root); } @@ -3788,25 +4121,28 @@ int btrfs_recover_relocation(struct btrfs_root *root) rc = alloc_reloc_control(fs_info); if (!rc) { - err = -ENOMEM; + ret = -ENOMEM; goto out; } - rc->extent_root = fs_info->extent_root; + ret = reloc_chunk_start(fs_info); + if (ret < 0) + goto out_end; + + rc->extent_root = btrfs_extent_root(fs_info, 0); set_reloc_control(rc); trans = btrfs_join_transaction(rc->extent_root); if (IS_ERR(trans)) { - err = PTR_ERR(trans); + ret = PTR_ERR(trans); goto out_unset; } - rc->merge_reloc_tree = 1; + rc->merge_reloc_tree = true; while (!list_empty(&reloc_roots)) { - reloc_root = list_entry(reloc_roots.next, - struct btrfs_root, root_list); + reloc_root = list_first_entry(&reloc_roots, struct btrfs_root, root_list); list_del(&reloc_root->root_list); if (btrfs_root_refs(&reloc_root->root_item) == 0) { @@ -3818,20 +4154,26 @@ int btrfs_recover_relocation(struct btrfs_root *root) fs_root = btrfs_get_fs_root(fs_info, reloc_root->root_key.offset, false); if (IS_ERR(fs_root)) { - err = PTR_ERR(fs_root); + ret = PTR_ERR(fs_root); list_add_tail(&reloc_root->root_list, &reloc_roots); btrfs_end_transaction(trans); goto out_unset; } - err = __add_reloc_root(reloc_root); - BUG_ON(err < 0); /* -ENOMEM or logic error */ + ret = __add_reloc_root(reloc_root); + ASSERT(ret != -EEXIST); + if (ret) { + list_add_tail(&reloc_root->root_list, &reloc_roots); + btrfs_put_root(fs_root); + btrfs_end_transaction(trans); + goto out_unset; + } fs_root->reloc_root = btrfs_grab_root(reloc_root); btrfs_put_root(fs_root); } - err = btrfs_commit_transaction(trans); - if (err) + ret = btrfs_commit_transaction(trans); + if (ret) goto out_unset; merge_reloc_roots(rc); @@ -3840,30 +4182,32 @@ int btrfs_recover_relocation(struct btrfs_root *root) trans = btrfs_join_transaction(rc->extent_root); if (IS_ERR(trans)) { - err = PTR_ERR(trans); + ret = PTR_ERR(trans); goto out_clean; } - err = btrfs_commit_transaction(trans); + ret = btrfs_commit_transaction(trans); out_clean: - ret = clean_dirty_subvols(rc); - if (ret < 0 && !err) - err = ret; + ret2 = clean_dirty_subvols(rc); + if (ret2 < 0 && !ret) + ret = ret2; out_unset: unset_reloc_control(rc); + reloc_chunk_end(fs_info); +out_end: free_reloc_control(rc); out: free_reloc_roots(&reloc_roots); btrfs_free_path(path); - if (err == 0) { + if (ret == 0) { /* cleanup orphan inode in data relocation tree */ fs_root = btrfs_grab_root(fs_info->data_reloc_root); ASSERT(fs_root); - err = btrfs_orphan_cleanup(fs_root); + ret = btrfs_orphan_cleanup(fs_root); btrfs_put_root(fs_root); } - return err; + return ret; } /* @@ -3872,27 +4216,27 @@ out: * cloning checksum properly handles the nodatasum extents. * it also saves CPU time to re-calculate the checksum. */ -int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len) +int btrfs_reloc_clone_csums(struct btrfs_ordered_extent *ordered) { - struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); - struct btrfs_ordered_sum *sums; - struct btrfs_ordered_extent *ordered; - int ret; - u64 disk_bytenr; - u64 new_bytenr; + struct btrfs_inode *inode = ordered->inode; + struct btrfs_fs_info *fs_info = inode->root->fs_info; + u64 disk_bytenr = ordered->file_offset + inode->reloc_block_group_start; + struct btrfs_root *csum_root = btrfs_csum_root(fs_info, disk_bytenr); LIST_HEAD(list); + int ret; - ordered = btrfs_lookup_ordered_extent(inode, file_pos); - BUG_ON(ordered->file_offset != file_pos || ordered->num_bytes != len); - - disk_bytenr = file_pos + BTRFS_I(inode)->index_cnt; - ret = btrfs_lookup_csums_range(fs_info->csum_root, disk_bytenr, - disk_bytenr + len - 1, &list, 0); - if (ret) - goto out; + ret = btrfs_lookup_csums_list(csum_root, disk_bytenr, + disk_bytenr + ordered->num_bytes - 1, + &list, false); + if (ret < 0) { + btrfs_mark_ordered_extent_error(ordered); + return ret; + } while (!list_empty(&list)) { - sums = list_entry(list.next, struct btrfs_ordered_sum, list); + struct btrfs_ordered_sum *sums = + list_first_entry(&list, struct btrfs_ordered_sum, list); + list_del_init(&sums->list); /* @@ -3907,18 +4251,16 @@ int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len) * disk_len vs real len like with real inodes since it's all * disk length. */ - new_bytenr = ordered->disk_bytenr + sums->bytenr - disk_bytenr; - sums->bytenr = new_bytenr; - + sums->logical = ordered->disk_bytenr + sums->logical - disk_bytenr; btrfs_add_ordered_sum(ordered, sums); } -out: - btrfs_put_ordered_extent(ordered); - return ret; + + return 0; } int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct extent_buffer *buf, + struct btrfs_root *root, + const struct extent_buffer *buf, struct extent_buffer *cow) { struct btrfs_fs_info *fs_info = root->fs_info; @@ -3932,24 +4274,32 @@ int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans, if (!rc) return 0; - BUG_ON(rc->stage == UPDATE_DATA_PTRS && - root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID); + BUG_ON(rc->stage == UPDATE_DATA_PTRS && btrfs_is_data_reloc_root(root)); level = btrfs_header_level(buf); if (btrfs_header_generation(buf) <= btrfs_root_last_snapshot(&root->root_item)) first_cow = 1; - if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID && - rc->create_reloc_tree) { + if (btrfs_root_id(root) == BTRFS_TREE_RELOC_OBJECTID && rc->create_reloc_tree) { WARN_ON(!first_cow && level == 0); node = rc->backref_cache.path[level]; - BUG_ON(node->bytenr != buf->start && - node->new_bytenr != buf->start); + + /* + * If node->bytenr != buf->start and node->new_bytenr != + * buf->start then we've got the wrong backref node for what we + * expected to see here and the cache is incorrect. + */ + if (unlikely(node->bytenr != buf->start && node->new_bytenr != buf->start)) { + btrfs_err(fs_info, +"bytenr %llu was found but our backref cache was expecting %llu or %llu", + buf->start, node->bytenr, node->new_bytenr); + return -EUCLEAN; + } btrfs_backref_drop_node_buffer(node); - atomic_inc(&cow->refs); + refcount_inc(&cow->refs); node->eb = cow; node->new_bytenr = cow->start; @@ -4034,16 +4384,33 @@ int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans, } new_root = pending->snap; - reloc_root = create_reloc_root(trans, root->reloc_root, - new_root->root_key.objectid); + reloc_root = create_reloc_root(trans, root->reloc_root, btrfs_root_id(new_root)); if (IS_ERR(reloc_root)) return PTR_ERR(reloc_root); ret = __add_reloc_root(reloc_root); - BUG_ON(ret < 0); + ASSERT(ret != -EEXIST); + if (ret) { + /* Pairs with create_reloc_root */ + btrfs_put_root(reloc_root); + return ret; + } new_root->reloc_root = btrfs_grab_root(reloc_root); + return 0; +} - if (rc->create_reloc_tree) - ret = clone_backref_node(trans, rc, root, reloc_root); - return ret; +/* + * Get the current bytenr for the block group which is being relocated. + * + * Return U64_MAX if no running relocation. + */ +u64 btrfs_get_reloc_bg_bytenr(const struct btrfs_fs_info *fs_info) +{ + u64 logical = U64_MAX; + + lockdep_assert_held(&fs_info->reloc_mutex); + + if (fs_info->reloc_ctl && fs_info->reloc_ctl->block_group) + logical = fs_info->reloc_ctl->block_group->start; + return logical; } |