X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=fs%2Fbtrfs%2Fctree.c;h=0d1d966b0fe45d20e6ff75f44a6dea64f66eb54e;hb=cc106eb35ed4abea675bce0d8fe40a46ff0b4a72;hp=efce173a935c5d284595a7bf6e8a6c2266a7e34e;hpb=ce9adaa5a792c2099a83246265eb4055bc38b6b8;p=safe%2Fjmp%2Flinux-2.6 diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index efce173..0d1d966 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -1,5 +1,5 @@ /* - * Copyright (C) 2007 Oracle. All rights reserved. + * Copyright (C) 2007,2008 Oracle. All rights reserved. * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public @@ -17,10 +17,12 @@ */ #include +#include #include "ctree.h" #include "disk-io.h" #include "transaction.h" #include "print-tree.h" +#include "locking.h" static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, int level); @@ -29,47 +31,156 @@ static int split_leaf(struct btrfs_trans_handle *trans, struct btrfs_root struct btrfs_path *path, int data_size, int extend); static int push_node_left(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *dst, - struct extent_buffer *src); + struct extent_buffer *src, int empty); static int balance_node_right(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *dst_buf, struct extent_buffer *src_buf); static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, int level, int slot); +static int setup_items_for_insert(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_path *path, + struct btrfs_key *cpu_key, u32 *data_size, + u32 total_data, u32 total_size, int nr); -inline void btrfs_init_path(struct btrfs_path *p) -{ - memset(p, 0, sizeof(*p)); -} struct btrfs_path *btrfs_alloc_path(void) { struct btrfs_path *path; - path = kmem_cache_alloc(btrfs_path_cachep, GFP_NOFS); - if (path) { - btrfs_init_path(path); + path = kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS); + if (path) path->reada = 1; - } return path; } +/* + * set all locked nodes in the path to blocking locks. This should + * be done before scheduling + */ +noinline void btrfs_set_path_blocking(struct btrfs_path *p) +{ + int i; + for (i = 0; i < BTRFS_MAX_LEVEL; i++) { + if (p->nodes[i] && p->locks[i]) + btrfs_set_lock_blocking(p->nodes[i]); + } +} + +/* + * reset all the locked nodes in the patch to spinning locks. + * + * held is used to keep lockdep happy, when lockdep is enabled + * we set held to a blocking lock before we go around and + * retake all the spinlocks in the path. You can safely use NULL + * for held + */ +noinline void btrfs_clear_path_blocking(struct btrfs_path *p, + struct extent_buffer *held) +{ + int i; + +#ifdef CONFIG_DEBUG_LOCK_ALLOC + /* lockdep really cares that we take all of these spinlocks + * in the right order. If any of the locks in the path are not + * currently blocking, it is going to complain. So, make really + * really sure by forcing the path to blocking before we clear + * the path blocking. + */ + if (held) + btrfs_set_lock_blocking(held); + btrfs_set_path_blocking(p); +#endif + + for (i = BTRFS_MAX_LEVEL - 1; i >= 0; i--) { + if (p->nodes[i] && p->locks[i]) + btrfs_clear_lock_blocking(p->nodes[i]); + } + +#ifdef CONFIG_DEBUG_LOCK_ALLOC + if (held) + btrfs_clear_lock_blocking(held); +#endif +} + +/* this also releases the path */ void btrfs_free_path(struct btrfs_path *p) { btrfs_release_path(NULL, p); kmem_cache_free(btrfs_path_cachep, p); } -void btrfs_release_path(struct btrfs_root *root, struct btrfs_path *p) +/* + * path release drops references on the extent buffers in the path + * and it drops any locks held by this path + * + * It is safe to call this on paths that no locks or extent buffers held. + */ +noinline void btrfs_release_path(struct btrfs_root *root, struct btrfs_path *p) { int i; + for (i = 0; i < BTRFS_MAX_LEVEL; i++) { + p->slots[i] = 0; if (!p->nodes[i]) - break; + continue; + if (p->locks[i]) { + btrfs_tree_unlock(p->nodes[i]); + p->locks[i] = 0; + } free_extent_buffer(p->nodes[i]); + p->nodes[i] = NULL; } - memset(p, 0, sizeof(*p)); } +/* + * safely gets a reference on the root node of a tree. A lock + * is not taken, so a concurrent writer may put a different node + * at the root of the tree. See btrfs_lock_root_node for the + * looping required. + * + * The extent buffer returned by this has a reference taken, so + * it won't disappear. It may stop being the root of the tree + * at any time because there are no locks held. + */ +struct extent_buffer *btrfs_root_node(struct btrfs_root *root) +{ + struct extent_buffer *eb; + spin_lock(&root->node_lock); + eb = root->node; + extent_buffer_get(eb); + spin_unlock(&root->node_lock); + return eb; +} + +/* loop around taking references on and locking the root node of the + * tree until you end up with a lock on the root. A locked buffer + * is returned, with a reference held. + */ +struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root) +{ + struct extent_buffer *eb; + + while (1) { + eb = btrfs_root_node(root); + btrfs_tree_lock(eb); + + spin_lock(&root->node_lock); + if (eb == root->node) { + spin_unlock(&root->node_lock); + break; + } + spin_unlock(&root->node_lock); + + btrfs_tree_unlock(eb); + free_extent_buffer(eb); + } + return eb; +} + +/* cowonly root (everything not a reference counted cow subvolume), just get + * put onto a simple dirty list. transaction.c walks this to make sure they + * get properly updated on disk. + */ static void add_root_to_dirty_list(struct btrfs_root *root) { if (root->track_dirty && list_empty(&root->dirty_list)) { @@ -78,6 +189,11 @@ static void add_root_to_dirty_list(struct btrfs_root *root) } } +/* + * used by snapshot creation to make a copy of a root for a tree with + * a given objectid. The buffer with the new root node is returned in + * cow_ret, and this func returns zero on success or a negative error code. + */ int btrfs_copy_root(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf, @@ -87,15 +203,7 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans, u32 nritems; int ret = 0; int level; - struct btrfs_key first_key; - struct btrfs_root *new_root; - - new_root = kmalloc(sizeof(*new_root), GFP_NOFS); - if (!new_root) - return -ENOMEM; - - memcpy(new_root, root, sizeof(*new_root)); - new_root->root_key.objectid = new_root_objectid; + struct btrfs_disk_key disk_key; WARN_ON(root->ref_cows && trans->transid != root->fs_info->running_transaction->transid); @@ -103,32 +211,37 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans, level = btrfs_header_level(buf); nritems = btrfs_header_nritems(buf); - if (nritems) { - if (level == 0) - btrfs_item_key_to_cpu(buf, &first_key, 0); - else - btrfs_node_key_to_cpu(buf, &first_key, 0); - } else { - first_key.objectid = 0; - } - cow = __btrfs_alloc_free_block(trans, new_root, buf->len, - new_root_objectid, - trans->transid, first_key.objectid, - level, buf->start, 0); - if (IS_ERR(cow)) { - kfree(new_root); + if (level == 0) + btrfs_item_key(buf, &disk_key, 0); + else + btrfs_node_key(buf, &disk_key, 0); + + cow = btrfs_alloc_free_block(trans, root, buf->len, 0, + new_root_objectid, &disk_key, level, + buf->start, 0); + if (IS_ERR(cow)) return PTR_ERR(cow); - } copy_extent_buffer(cow, buf, 0, 0, cow->len); btrfs_set_header_bytenr(cow, cow->start); btrfs_set_header_generation(cow, trans->transid); - btrfs_set_header_owner(cow, new_root_objectid); - btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN); + btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV); + btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN | + BTRFS_HEADER_FLAG_RELOC); + if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID) + btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC); + else + btrfs_set_header_owner(cow, new_root_objectid); + + write_extent_buffer(cow, root->fs_info->fsid, + (unsigned long)btrfs_header_fsid(cow), + BTRFS_FSID_SIZE); WARN_ON(btrfs_header_generation(buf) > trans->transid); - ret = btrfs_inc_ref(trans, new_root, buf); - kfree(new_root); + if (new_root_objectid == BTRFS_TREE_RELOC_OBJECTID) + ret = btrfs_inc_ref(trans, root, cow, 1); + else + ret = btrfs_inc_ref(trans, root, cow, 0); if (ret) return ret; @@ -138,128 +251,304 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans, return 0; } -int __btrfs_cow_block(struct btrfs_trans_handle *trans, +/* + * check if the tree block can be shared by multiple trees + */ +int btrfs_block_can_be_shared(struct btrfs_root *root, + struct extent_buffer *buf) +{ + /* + * Tree blocks not in refernece counted trees and tree roots + * are never shared. If a block was allocated after the last + * snapshot and the block was not allocated by tree relocation, + * we know the block is not shared. + */ + if (root->ref_cows && + buf != root->node && buf != root->commit_root && + (btrfs_header_generation(buf) <= + btrfs_root_last_snapshot(&root->root_item) || + btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC))) + return 1; +#ifdef BTRFS_COMPAT_EXTENT_TREE_V0 + if (root->ref_cows && + btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV) + return 1; +#endif + return 0; +} + +static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct extent_buffer *buf, + struct extent_buffer *cow, + int *last_ref) +{ + u64 refs; + u64 owner; + u64 flags; + u64 new_flags = 0; + int ret; + + /* + * Backrefs update rules: + * + * Always use full backrefs for extent pointers in tree block + * allocated by tree relocation. + * + * If a shared tree block is no longer referenced by its owner + * tree (btrfs_header_owner(buf) == root->root_key.objectid), + * use full backrefs for extent pointers in tree block. + * + * If a tree block is been relocating + * (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID), + * use full backrefs for extent pointers in tree block. + * The reason for this is some operations (such as drop tree) + * are only allowed for blocks use full backrefs. + */ + + if (btrfs_block_can_be_shared(root, buf)) { + ret = btrfs_lookup_extent_info(trans, root, buf->start, + buf->len, &refs, &flags); + BUG_ON(ret); + BUG_ON(refs == 0); + } else { + refs = 1; + if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID || + btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV) + flags = BTRFS_BLOCK_FLAG_FULL_BACKREF; + else + flags = 0; + } + + owner = btrfs_header_owner(buf); + BUG_ON(owner == BTRFS_TREE_RELOC_OBJECTID && + !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)); + + if (refs > 1) { + if ((owner == root->root_key.objectid || + root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) && + !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) { + ret = btrfs_inc_ref(trans, root, buf, 1); + BUG_ON(ret); + + if (root->root_key.objectid == + BTRFS_TREE_RELOC_OBJECTID) { + ret = btrfs_dec_ref(trans, root, buf, 0); + BUG_ON(ret); + ret = btrfs_inc_ref(trans, root, cow, 1); + BUG_ON(ret); + } + new_flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF; + } else { + + if (root->root_key.objectid == + BTRFS_TREE_RELOC_OBJECTID) + ret = btrfs_inc_ref(trans, root, cow, 1); + else + ret = btrfs_inc_ref(trans, root, cow, 0); + BUG_ON(ret); + } + if (new_flags != 0) { + ret = btrfs_set_disk_extent_flags(trans, root, + buf->start, + buf->len, + new_flags, 0); + BUG_ON(ret); + } + } else { + if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) { + if (root->root_key.objectid == + BTRFS_TREE_RELOC_OBJECTID) + ret = btrfs_inc_ref(trans, root, cow, 1); + else + ret = btrfs_inc_ref(trans, root, cow, 0); + BUG_ON(ret); + ret = btrfs_dec_ref(trans, root, buf, 1); + BUG_ON(ret); + } + clean_tree_block(trans, root, buf); + *last_ref = 1; + } + return 0; +} + +/* + * does the dirty work in cow of a single block. The parent block (if + * supplied) is updated to point to the new cow copy. The new buffer is marked + * dirty and returned locked. If you modify the block it needs to be marked + * dirty again. + * + * search_start -- an allocation hint for the new block + * + * empty_size -- a hint that you plan on doing more cow. This is the size in + * bytes the allocator should try to find free next to the block it returns. + * This is just a hint and may be ignored by the allocator. + */ +static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf, struct extent_buffer *parent, int parent_slot, struct extent_buffer **cow_ret, u64 search_start, u64 empty_size) { - u64 root_gen; + struct btrfs_disk_key disk_key; struct extent_buffer *cow; - u32 nritems; - int ret = 0; - int different_trans = 0; int level; - struct btrfs_key first_key; + int last_ref = 0; + int unlock_orig = 0; + u64 parent_start; + + if (*cow_ret == buf) + unlock_orig = 1; + + btrfs_assert_tree_locked(buf); - if (root->ref_cows) { - root_gen = trans->transid; - } else { - root_gen = 0; - } WARN_ON(root->ref_cows && trans->transid != root->fs_info->running_transaction->transid); WARN_ON(root->ref_cows && trans->transid != root->last_trans); level = btrfs_header_level(buf); - nritems = btrfs_header_nritems(buf); - if (nritems) { - if (level == 0) - btrfs_item_key_to_cpu(buf, &first_key, 0); + + if (level == 0) + btrfs_item_key(buf, &disk_key, 0); + else + btrfs_node_key(buf, &disk_key, 0); + + if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) { + if (parent) + parent_start = parent->start; else - btrfs_node_key_to_cpu(buf, &first_key, 0); - } else { - first_key.objectid = 0; - } - cow = __btrfs_alloc_free_block(trans, root, buf->len, - root->root_key.objectid, - root_gen, first_key.objectid, level, - search_start, empty_size); + parent_start = 0; + } else + parent_start = 0; + + cow = btrfs_alloc_free_block(trans, root, buf->len, parent_start, + root->root_key.objectid, &disk_key, + level, search_start, empty_size); if (IS_ERR(cow)) return PTR_ERR(cow); + /* cow is set to blocking by btrfs_init_new_buffer */ + copy_extent_buffer(cow, buf, 0, 0, cow->len); btrfs_set_header_bytenr(cow, cow->start); btrfs_set_header_generation(cow, trans->transid); - btrfs_set_header_owner(cow, root->root_key.objectid); - btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN); + btrfs_set_header_backref_rev(cow, BTRFS_MIXED_BACKREF_REV); + btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN | + BTRFS_HEADER_FLAG_RELOC); + if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) + btrfs_set_header_flag(cow, BTRFS_HEADER_FLAG_RELOC); + else + btrfs_set_header_owner(cow, root->root_key.objectid); - WARN_ON(btrfs_header_generation(buf) > trans->transid); - if (btrfs_header_generation(buf) != trans->transid) { - different_trans = 1; - ret = btrfs_inc_ref(trans, root, buf); - if (ret) - return ret; - } else { - clean_tree_block(trans, root, buf); - } + write_extent_buffer(cow, root->fs_info->fsid, + (unsigned long)btrfs_header_fsid(cow), + BTRFS_FSID_SIZE); + + update_ref_for_cow(trans, root, buf, cow, &last_ref); + + if (root->ref_cows) + btrfs_reloc_cow_block(trans, root, buf, cow); if (buf == root->node) { - root_gen = btrfs_header_generation(buf); + WARN_ON(parent && parent != buf); + if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID || + btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV) + parent_start = buf->start; + else + parent_start = 0; + + spin_lock(&root->node_lock); root->node = cow; extent_buffer_get(cow); - if (buf != root->commit_root) { - btrfs_free_extent(trans, root, buf->start, - buf->len, root->root_key.objectid, - root_gen, 0, 0, 1); - } + spin_unlock(&root->node_lock); + + btrfs_free_tree_block(trans, root, buf, parent_start, + last_ref); free_extent_buffer(buf); add_root_to_dirty_list(root); } else { - root_gen = btrfs_header_generation(parent); + if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) + parent_start = parent->start; + else + parent_start = 0; + + WARN_ON(trans->transid != btrfs_header_generation(parent)); btrfs_set_node_blockptr(parent, parent_slot, cow->start); - WARN_ON(trans->transid == 0); btrfs_set_node_ptr_generation(parent, parent_slot, trans->transid); btrfs_mark_buffer_dirty(parent); - WARN_ON(btrfs_header_generation(parent) != trans->transid); - btrfs_free_extent(trans, root, buf->start, buf->len, - btrfs_header_owner(parent), root_gen, - 0, 0, 1); + btrfs_free_tree_block(trans, root, buf, parent_start, + last_ref); } + if (unlock_orig) + btrfs_tree_unlock(buf); free_extent_buffer(buf); btrfs_mark_buffer_dirty(cow); *cow_ret = cow; return 0; } -int btrfs_cow_block(struct btrfs_trans_handle *trans, +static inline int should_cow_block(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct extent_buffer *buf) +{ + if (btrfs_header_generation(buf) == trans->transid && + !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) && + !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID && + btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC))) + return 0; + return 1; +} + +/* + * cows a single block, see __btrfs_cow_block for the real work. + * This version of it has extra checks so that a block isn't cow'd more than + * once per transaction, as long as it hasn't been written yet + */ +noinline int btrfs_cow_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf, struct extent_buffer *parent, int parent_slot, struct extent_buffer **cow_ret) { u64 search_start; - u64 header_trans; int ret; if (trans->transaction != root->fs_info->running_transaction) { - printk(KERN_CRIT "trans %Lu running %Lu\n", trans->transid, + printk(KERN_CRIT "trans %llu running %llu\n", + (unsigned long long)trans->transid, + (unsigned long long) root->fs_info->running_transaction->transid); WARN_ON(1); } if (trans->transid != root->fs_info->generation) { - printk(KERN_CRIT "trans %Lu running %Lu\n", trans->transid, - root->fs_info->generation); + printk(KERN_CRIT "trans %llu running %llu\n", + (unsigned long long)trans->transid, + (unsigned long long)root->fs_info->generation); WARN_ON(1); } - header_trans = btrfs_header_generation(buf); - spin_lock(&root->fs_info->hash_lock); - if (header_trans == trans->transid && - !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) { + if (!should_cow_block(trans, root, buf)) { *cow_ret = buf; - spin_unlock(&root->fs_info->hash_lock); return 0; } - spin_unlock(&root->fs_info->hash_lock); + search_start = buf->start & ~((u64)(1024 * 1024 * 1024) - 1); + + if (parent) + btrfs_set_lock_blocking(parent); + btrfs_set_lock_blocking(buf); + ret = __btrfs_cow_block(trans, root, buf, parent, parent_slot, cow_ret, search_start, 0); return ret; } +/* + * helper function for defrag to decide if two blocks pointed to by a + * node are actually close by + */ static int close_blocks(u64 blocknr, u64 other, u32 blocksize) { if (blocknr < other && other - (blocknr + blocksize) < 32768) @@ -278,30 +567,42 @@ static int comp_keys(struct btrfs_disk_key *disk, struct btrfs_key *k2) btrfs_disk_key_to_cpu(&k1, disk); - if (k1.objectid > k2->objectid) + return btrfs_comp_cpu_keys(&k1, k2); +} + +/* + * same as comp_keys only with two btrfs_key's + */ +int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2) +{ + if (k1->objectid > k2->objectid) return 1; - if (k1.objectid < k2->objectid) + if (k1->objectid < k2->objectid) return -1; - if (k1.type > k2->type) + if (k1->type > k2->type) return 1; - if (k1.type < k2->type) + if (k1->type < k2->type) return -1; - if (k1.offset > k2->offset) + if (k1->offset > k2->offset) return 1; - if (k1.offset < k2->offset) + if (k1->offset < k2->offset) return -1; return 0; } - +/* + * this is used by the defrag code to go through all the + * leaves pointed to by a node and reallocate them so that + * disk order is close to key order + */ int btrfs_realloc_node(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *parent, int start_slot, int cache_only, u64 *last_ret, struct btrfs_key *progress) { struct extent_buffer *cur; - struct extent_buffer *tmp; u64 blocknr; + u64 gen; u64 search_start = *last_ret; u64 last_block = 0; u64 other; @@ -319,16 +620,10 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, if (cache_only && parent_level != 1) return 0; - if (trans->transaction != root->fs_info->running_transaction) { - printk(KERN_CRIT "trans %Lu running %Lu\n", trans->transid, - root->fs_info->running_transaction->transid); + if (trans->transaction != root->fs_info->running_transaction) WARN_ON(1); - } - if (trans->transid != root->fs_info->generation) { - printk(KERN_CRIT "trans %Lu running %Lu\n", trans->transid, - root->fs_info->generation); + if (trans->transid != root->fs_info->generation) WARN_ON(1); - } parent_nritems = btrfs_header_nritems(parent); blocksize = btrfs_level_size(root, parent_level - 1); @@ -337,6 +632,8 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, if (parent_nritems == 1) return 0; + btrfs_set_lock_blocking(parent); + for (i = start_slot; i < end_slot; i++) { int close = 1; @@ -354,6 +651,7 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, progress_passed = 1; blocknr = btrfs_node_blockptr(parent, i); + gen = btrfs_node_ptr_generation(parent, i); if (last_block == 0) last_block = blocknr; @@ -361,7 +659,7 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, other = btrfs_node_blockptr(parent, i - 1); close = close_blocks(blocknr, other, blocksize); } - if (close && i < end_slot - 2) { + if (!close && i < end_slot - 2) { other = btrfs_node_blockptr(parent, i + 1); close = close_blocks(blocknr, other, blocksize); } @@ -377,7 +675,7 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, cur = btrfs_find_tree_block(root, blocknr, blocksize); if (cur) - uptodate = btrfs_buffer_uptodate(cur); + uptodate = btrfs_buffer_uptodate(cur, gen); else uptodate = 0; if (!cur || !uptodate) { @@ -387,29 +685,30 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, } if (!cur) { cur = read_tree_block(root, blocknr, - blocksize); + blocksize, gen); } else if (!uptodate) { - btrfs_read_buffer(cur); + btrfs_read_buffer(cur, gen); } } if (search_start == 0) search_start = last_block; - btrfs_verify_block_csum(root, cur); + btrfs_tree_lock(cur); + btrfs_set_lock_blocking(cur); err = __btrfs_cow_block(trans, root, cur, parent, i, - &tmp, search_start, + &cur, search_start, min(16 * blocksize, (end_slot - i) * blocksize)); if (err) { + btrfs_tree_unlock(cur); free_extent_buffer(cur); break; } - search_start = tmp->start; - last_block = tmp->start; + search_start = cur->start; + last_block = cur->start; *last_ret = search_start; - if (parent_level == 1) - btrfs_clear_buffer_defrag(tmp); - free_extent_buffer(tmp); + btrfs_tree_unlock(cur); + free_extent_buffer(cur); } if (parent->map_token) { unmap_extent_buffer(parent, parent->map_token, @@ -433,6 +732,10 @@ static inline unsigned int leaf_data_end(struct btrfs_root *root, return btrfs_item_offset_nr(leaf, nr - 1); } +/* + * extra debugging checks to make sure all the items in a key are + * well formed and in the proper order + */ static int check_node(struct btrfs_root *root, struct btrfs_path *path, int level) { @@ -473,6 +776,10 @@ static int check_node(struct btrfs_root *root, struct btrfs_path *path, return 0; } +/* + * extra checking to make sure all the items in a leaf are + * well formed and in the proper order + */ static int check_leaf(struct btrfs_root *root, struct btrfs_path *path, int level) { @@ -502,51 +809,18 @@ static int check_leaf(struct btrfs_root *root, struct btrfs_path *path, BUG_ON(btrfs_node_blockptr(parent, parent_slot) != btrfs_header_bytenr(leaf)); } -#if 0 - for (i = 0; nritems > 1 && i < nritems - 2; i++) { - btrfs_item_key_to_cpu(leaf, &cpukey, i + 1); - btrfs_item_key(leaf, &leaf_key, i); - if (comp_keys(&leaf_key, &cpukey) >= 0) { - btrfs_print_leaf(root, leaf); - printk("slot %d offset bad key\n", i); - BUG_ON(1); - } - if (btrfs_item_offset_nr(leaf, i) != - btrfs_item_end_nr(leaf, i + 1)) { - btrfs_print_leaf(root, leaf); - printk("slot %d offset bad\n", i); - BUG_ON(1); - } - if (i == 0) { - if (btrfs_item_offset_nr(leaf, i) + - btrfs_item_size_nr(leaf, i) != - BTRFS_LEAF_DATA_SIZE(root)) { - btrfs_print_leaf(root, leaf); - printk("slot %d first offset bad\n", i); - BUG_ON(1); - } - } - } - if (nritems > 0) { - if (btrfs_item_size_nr(leaf, nritems - 1) > 4096) { - btrfs_print_leaf(root, leaf); - printk("slot %d bad size \n", nritems - 1); - BUG_ON(1); - } - } -#endif if (slot != 0 && slot < nritems - 1) { btrfs_item_key(leaf, &leaf_key, slot); btrfs_item_key_to_cpu(leaf, &cpukey, slot - 1); if (comp_keys(&leaf_key, &cpukey) <= 0) { btrfs_print_leaf(root, leaf); - printk("slot %d offset bad key\n", slot); + printk(KERN_CRIT "slot %d offset bad key\n", slot); BUG_ON(1); } if (btrfs_item_offset_nr(leaf, slot - 1) != btrfs_item_end_nr(leaf, slot)) { btrfs_print_leaf(root, leaf); - printk("slot %d offset bad\n", slot); + printk(KERN_CRIT "slot %d offset bad\n", slot); BUG_ON(1); } } @@ -557,7 +831,7 @@ static int check_leaf(struct btrfs_root *root, struct btrfs_path *path, if (btrfs_item_offset_nr(leaf, slot) != btrfs_item_end_nr(leaf, slot + 1)) { btrfs_print_leaf(root, leaf); - printk("slot %d offset bad\n", slot); + printk(KERN_CRIT "slot %d offset bad\n", slot); BUG_ON(1); } } @@ -566,20 +840,10 @@ static int check_leaf(struct btrfs_root *root, struct btrfs_path *path, return 0; } -static int noinline check_block(struct btrfs_root *root, +static noinline int check_block(struct btrfs_root *root, struct btrfs_path *path, int level) { return 0; -#if 0 - struct extent_buffer *buf = path->nodes[level]; - - if (memcmp_extent_buffer(buf, root->fs_info->fsid, - (unsigned long)btrfs_header_fsid(buf), - BTRFS_FSID_SIZE)) { - printk("warning bad block %Lu\n", buf->start); - return 1; - } -#endif if (level == 0) return check_leaf(root, path, level); return check_node(root, path, level); @@ -595,9 +859,10 @@ static int noinline check_block(struct btrfs_root *root, * * slot may point to max if the key is bigger than all of the keys */ -static int generic_bin_search(struct extent_buffer *eb, unsigned long p, - int item_size, struct btrfs_key *key, - int max, int *slot) +static noinline int generic_bin_search(struct extent_buffer *eb, + unsigned long p, + int item_size, struct btrfs_key *key, + int max, int *slot) { int low = 0; int high = max; @@ -612,7 +877,7 @@ static int generic_bin_search(struct extent_buffer *eb, unsigned long p, unsigned long map_len = 0; int err; - while(low < high) { + while (low < high) { mid = (low + high) / 2; offset = p + mid * item_size; @@ -623,7 +888,8 @@ static int generic_bin_search(struct extent_buffer *eb, unsigned long p, unmap_extent_buffer(eb, map_token, KM_USER0); map_token = NULL; } - err = map_extent_buffer(eb, offset, + + err = map_private_extent_buffer(eb, offset, sizeof(struct btrfs_disk_key), &map_token, &kaddr, &map_start, &map_len, KM_USER0); @@ -683,18 +949,54 @@ static int bin_search(struct extent_buffer *eb, struct btrfs_key *key, return -1; } -static struct extent_buffer *read_node_slot(struct btrfs_root *root, +int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key, + int level, int *slot) +{ + return bin_search(eb, key, level, slot); +} + +static void root_add_used(struct btrfs_root *root, u32 size) +{ + spin_lock(&root->accounting_lock); + btrfs_set_root_used(&root->root_item, + btrfs_root_used(&root->root_item) + size); + spin_unlock(&root->accounting_lock); +} + +static void root_sub_used(struct btrfs_root *root, u32 size) +{ + spin_lock(&root->accounting_lock); + btrfs_set_root_used(&root->root_item, + btrfs_root_used(&root->root_item) - size); + spin_unlock(&root->accounting_lock); +} + +/* given a node and slot number, this reads the blocks it points to. The + * extent buffer is returned with a reference taken (but unlocked). + * NULL is returned on error. + */ +static noinline struct extent_buffer *read_node_slot(struct btrfs_root *root, struct extent_buffer *parent, int slot) { + int level = btrfs_header_level(parent); if (slot < 0) return NULL; if (slot >= btrfs_header_nritems(parent)) return NULL; + + BUG_ON(level == 0); + return read_tree_block(root, btrfs_node_blockptr(parent, slot), - btrfs_level_size(root, btrfs_header_level(parent) - 1)); + btrfs_level_size(root, level - 1), + btrfs_node_ptr_generation(parent, slot)); } -static int balance_level(struct btrfs_trans_handle *trans, +/* + * node level balancing, used to make sure nodes are in proper order for + * item deletion. We balance from the top down, so we have to make sure + * that a deletion won't leave an node completely empty later on. + */ +static noinline int balance_level(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, int level) { @@ -713,6 +1015,8 @@ static int balance_level(struct btrfs_trans_handle *trans, return 0; mid = path->nodes[level]; + + WARN_ON(!path->locks[level]); WARN_ON(btrfs_header_generation(mid) != trans->transid); orig_ptr = btrfs_node_blockptr(mid, orig_slot); @@ -734,22 +1038,34 @@ static int balance_level(struct btrfs_trans_handle *trans, /* promote the child to a root */ child = read_node_slot(root, mid, 0); BUG_ON(!child); + btrfs_tree_lock(child); + btrfs_set_lock_blocking(child); ret = btrfs_cow_block(trans, root, child, mid, 0, &child); - BUG_ON(ret); + if (ret) { + btrfs_tree_unlock(child); + free_extent_buffer(child); + goto enospc; + } + spin_lock(&root->node_lock); root->node = child; + spin_unlock(&root->node_lock); + add_root_to_dirty_list(root); + btrfs_tree_unlock(child); + + path->locks[level] = 0; path->nodes[level] = NULL; clean_tree_block(trans, root, mid); - wait_on_tree_block_writeback(root, mid); + btrfs_tree_unlock(mid); /* once for the path */ free_extent_buffer(mid); - ret = btrfs_free_extent(trans, root, mid->start, mid->len, - root->root_key.objectid, - btrfs_header_generation(mid), 0, 0, 1); + + root_sub_used(root, mid->len); + btrfs_free_tree_block(trans, root, mid, 0, 1); /* once for the root ptr */ free_extent_buffer(mid); - return ret; + return 0; } if (btrfs_header_nritems(mid) > BTRFS_NODEPTRS_PER_BLOCK(root) / 4) @@ -760,6 +1076,8 @@ static int balance_level(struct btrfs_trans_handle *trans, left = read_node_slot(root, parent, pslot - 1); if (left) { + btrfs_tree_lock(left); + btrfs_set_lock_blocking(left); wret = btrfs_cow_block(trans, root, left, parent, pslot - 1, &left); if (wret) { @@ -769,6 +1087,8 @@ static int balance_level(struct btrfs_trans_handle *trans, } right = read_node_slot(root, parent, pslot + 1); if (right) { + btrfs_tree_lock(right); + btrfs_set_lock_blocking(right); wret = btrfs_cow_block(trans, root, right, parent, pslot + 1, &right); if (wret) { @@ -780,7 +1100,7 @@ static int balance_level(struct btrfs_trans_handle *trans, /* first, try to make some room in the middle buffer */ if (left) { orig_slot += btrfs_header_nritems(left); - wret = push_node_left(trans, root, left, mid); + wret = push_node_left(trans, root, left, mid, 1); if (wret < 0) ret = wret; if (btrfs_header_nritems(mid) < 2) @@ -791,28 +1111,20 @@ static int balance_level(struct btrfs_trans_handle *trans, * then try to empty the right most buffer into the middle */ if (right) { - wret = push_node_left(trans, root, mid, right); + wret = push_node_left(trans, root, mid, right, 1); if (wret < 0 && wret != -ENOSPC) ret = wret; if (btrfs_header_nritems(right) == 0) { - u64 bytenr = right->start; - u64 generation = btrfs_header_generation(parent); - u32 blocksize = right->len; - clean_tree_block(trans, root, right); - wait_on_tree_block_writeback(root, right); - free_extent_buffer(right); - right = NULL; + btrfs_tree_unlock(right); wret = del_ptr(trans, root, path, level + 1, pslot + 1); if (wret) ret = wret; - wret = btrfs_free_extent(trans, root, bytenr, - blocksize, - btrfs_header_owner(parent), - generation, 0, 0, 1); - if (wret) - ret = wret; + root_sub_used(root, right->len); + btrfs_free_tree_block(trans, root, right, 0, 1); + free_extent_buffer(right); + right = NULL; } else { struct btrfs_disk_key right_key; btrfs_node_key(right, &right_key, 0); @@ -836,25 +1148,23 @@ static int balance_level(struct btrfs_trans_handle *trans, ret = wret; goto enospc; } + if (wret == 1) { + wret = push_node_left(trans, root, left, mid, 1); + if (wret < 0) + ret = wret; + } BUG_ON(wret == 1); } if (btrfs_header_nritems(mid) == 0) { - /* we've managed to empty the middle node, drop it */ - u64 root_gen = btrfs_header_generation(parent); - u64 bytenr = mid->start; - u32 blocksize = mid->len; clean_tree_block(trans, root, mid); - wait_on_tree_block_writeback(root, mid); - free_extent_buffer(mid); - mid = NULL; + btrfs_tree_unlock(mid); wret = del_ptr(trans, root, path, level + 1, pslot); if (wret) ret = wret; - wret = btrfs_free_extent(trans, root, bytenr, blocksize, - btrfs_header_owner(parent), - root_gen, 0, 0, 1); - if (wret) - ret = wret; + root_sub_used(root, mid->len); + btrfs_free_tree_block(trans, root, mid, 0, 1); + free_extent_buffer(mid); + mid = NULL; } else { /* update the parent key to reflect our changes */ struct btrfs_disk_key mid_key; @@ -867,11 +1177,14 @@ static int balance_level(struct btrfs_trans_handle *trans, if (left) { if (btrfs_header_nritems(left) > orig_slot) { extent_buffer_get(left); + /* left was locked after cow */ path->nodes[level] = left; path->slots[level + 1] -= 1; path->slots[level] = orig_slot; - if (mid) + if (mid) { + btrfs_tree_unlock(mid); free_extent_buffer(mid); + } } else { orig_slot -= btrfs_header_nritems(left); path->slots[level] = orig_slot; @@ -883,15 +1196,23 @@ static int balance_level(struct btrfs_trans_handle *trans, btrfs_node_blockptr(path->nodes[level], path->slots[level])) BUG(); enospc: - if (right) + if (right) { + btrfs_tree_unlock(right); free_extent_buffer(right); - if (left) + } + if (left) { + if (path->nodes[level] != left) + btrfs_tree_unlock(left); free_extent_buffer(left); + } return ret; } -/* returns zero if the push worked, non-zero otherwise */ -static int noinline push_nodes_for_insert(struct btrfs_trans_handle *trans, +/* Node balancing for insertion. Here we only split or push nodes around + * when they are completely full. This is also done top down, so we + * have to be pessimistic. + */ +static noinline int push_nodes_for_insert(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, int level) { @@ -924,6 +1245,10 @@ static int noinline push_nodes_for_insert(struct btrfs_trans_handle *trans, /* first, try to make some room in the middle buffer */ if (left) { u32 left_nr; + + btrfs_tree_lock(left); + btrfs_set_lock_blocking(left); + left_nr = btrfs_header_nritems(left); if (left_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) { wret = 1; @@ -934,7 +1259,7 @@ static int noinline push_nodes_for_insert(struct btrfs_trans_handle *trans, wret = 1; else { wret = push_node_left(trans, root, - left, mid); + left, mid, 0); } } if (wret < 0) @@ -949,24 +1274,31 @@ static int noinline push_nodes_for_insert(struct btrfs_trans_handle *trans, path->nodes[level] = left; path->slots[level + 1] -= 1; path->slots[level] = orig_slot; + btrfs_tree_unlock(mid); free_extent_buffer(mid); } else { orig_slot -= btrfs_header_nritems(left); path->slots[level] = orig_slot; + btrfs_tree_unlock(left); free_extent_buffer(left); } return 0; } + btrfs_tree_unlock(left); free_extent_buffer(left); } - right= read_node_slot(root, parent, pslot + 1); + right = read_node_slot(root, parent, pslot + 1); /* * then try to empty the right most buffer into the middle */ if (right) { u32 right_nr; + + btrfs_tree_lock(right); + btrfs_set_lock_blocking(right); + right_nr = btrfs_header_nritems(right); if (right_nr >= BTRFS_NODEPTRS_PER_BLOCK(root) - 1) { wret = 1; @@ -995,29 +1327,33 @@ static int noinline push_nodes_for_insert(struct btrfs_trans_handle *trans, path->slots[level + 1] += 1; path->slots[level] = orig_slot - btrfs_header_nritems(mid); + btrfs_tree_unlock(mid); free_extent_buffer(mid); } else { + btrfs_tree_unlock(right); free_extent_buffer(right); } return 0; } + btrfs_tree_unlock(right); free_extent_buffer(right); } return 1; } /* - * readahead one full node of leaves + * readahead one full node of leaves, finding things that are close + * to the block in 'slot', and triggering ra on them. */ -static void reada_for_search(struct btrfs_root *root, struct btrfs_path *path, +static void reada_for_search(struct btrfs_root *root, + struct btrfs_path *path, int level, int slot, u64 objectid) { struct extent_buffer *node; struct btrfs_disk_key disk_key; u32 nritems; u64 search; - u64 lowest_read; - u64 highest_read; + u64 target; u64 nread = 0; int direction = path->reada; struct extent_buffer *eb; @@ -1032,6 +1368,7 @@ static void reada_for_search(struct btrfs_root *root, struct btrfs_path *path, return; node = path->nodes[level]; + search = btrfs_node_blockptr(node, slot); blocksize = btrfs_level_size(root, level - 1); eb = btrfs_find_tree_block(root, search, blocksize); @@ -1040,12 +1377,11 @@ static void reada_for_search(struct btrfs_root *root, struct btrfs_path *path, return; } - highest_read = search; - lowest_read = search; + target = search; nritems = btrfs_header_nritems(node); nr = slot; - while(1) { + while (1) { if (direction < 0) { if (nr == 0) break; @@ -1061,34 +1397,302 @@ static void reada_for_search(struct btrfs_root *root, struct btrfs_path *path, break; } search = btrfs_node_blockptr(node, nr); - if ((search >= lowest_read && search <= highest_read) || - (search < lowest_read && lowest_read - search <= 32768) || - (search > highest_read && search - highest_read <= 32768)) { - readahead_tree_block(root, search, blocksize); + if ((search <= target && target - search <= 65536) || + (search > target && search - target <= 65536)) { + readahead_tree_block(root, search, blocksize, + btrfs_node_ptr_generation(node, nr)); nread += blocksize; } nscan++; - if (path->reada < 2 && (nread > (256 * 1024) || nscan > 32)) + if ((nread > 65536 || nscan > 32)) break; - if(nread > (1024 * 1024) || nscan > 128) - break; - - if (search < lowest_read) - lowest_read = search; - if (search > highest_read) - highest_read = search; } } + /* - * look for key in the tree. path is filled in with nodes along the way - * if key is found, we return zero and you can find the item in the leaf - * level of the path (level 0) - * - * If the key isn't found, the path points to the slot where it should - * be inserted, and 1 is returned. If there are other errors during the - * search a negative error number is returned. - * - * if ins_len > 0, nodes and leaves will be split as we walk down the + * returns -EAGAIN if it had to drop the path, or zero if everything was in + * cache + */ +static noinline int reada_for_balance(struct btrfs_root *root, + struct btrfs_path *path, int level) +{ + int slot; + int nritems; + struct extent_buffer *parent; + struct extent_buffer *eb; + u64 gen; + u64 block1 = 0; + u64 block2 = 0; + int ret = 0; + int blocksize; + + parent = path->nodes[level + 1]; + if (!parent) + return 0; + + nritems = btrfs_header_nritems(parent); + slot = path->slots[level + 1]; + blocksize = btrfs_level_size(root, level); + + if (slot > 0) { + block1 = btrfs_node_blockptr(parent, slot - 1); + gen = btrfs_node_ptr_generation(parent, slot - 1); + eb = btrfs_find_tree_block(root, block1, blocksize); + if (eb && btrfs_buffer_uptodate(eb, gen)) + block1 = 0; + free_extent_buffer(eb); + } + if (slot + 1 < nritems) { + block2 = btrfs_node_blockptr(parent, slot + 1); + gen = btrfs_node_ptr_generation(parent, slot + 1); + eb = btrfs_find_tree_block(root, block2, blocksize); + if (eb && btrfs_buffer_uptodate(eb, gen)) + block2 = 0; + free_extent_buffer(eb); + } + if (block1 || block2) { + ret = -EAGAIN; + + /* release the whole path */ + btrfs_release_path(root, path); + + /* read the blocks */ + if (block1) + readahead_tree_block(root, block1, blocksize, 0); + if (block2) + readahead_tree_block(root, block2, blocksize, 0); + + if (block1) { + eb = read_tree_block(root, block1, blocksize, 0); + free_extent_buffer(eb); + } + if (block2) { + eb = read_tree_block(root, block2, blocksize, 0); + free_extent_buffer(eb); + } + } + return ret; +} + + +/* + * when we walk down the tree, it is usually safe to unlock the higher layers + * in the tree. The exceptions are when our path goes through slot 0, because + * operations on the tree might require changing key pointers higher up in the + * tree. + * + * callers might also have set path->keep_locks, which tells this code to keep + * the lock if the path points to the last slot in the block. This is part of + * walking through the tree, and selecting the next slot in the higher block. + * + * lowest_unlock sets the lowest level in the tree we're allowed to unlock. so + * if lowest_unlock is 1, level 0 won't be unlocked + */ +static noinline void unlock_up(struct btrfs_path *path, int level, + int lowest_unlock) +{ + int i; + int skip_level = level; + int no_skips = 0; + struct extent_buffer *t; + + for (i = level; i < BTRFS_MAX_LEVEL; i++) { + if (!path->nodes[i]) + break; + if (!path->locks[i]) + break; + if (!no_skips && path->slots[i] == 0) { + skip_level = i + 1; + continue; + } + if (!no_skips && path->keep_locks) { + u32 nritems; + t = path->nodes[i]; + nritems = btrfs_header_nritems(t); + if (nritems < 1 || path->slots[i] >= nritems - 1) { + skip_level = i + 1; + continue; + } + } + if (skip_level < i && i >= lowest_unlock) + no_skips = 1; + + t = path->nodes[i]; + if (i >= lowest_unlock && i > skip_level && path->locks[i]) { + btrfs_tree_unlock(t); + path->locks[i] = 0; + } + } +} + +/* + * This releases any locks held in the path starting at level and + * going all the way up to the root. + * + * btrfs_search_slot will keep the lock held on higher nodes in a few + * corner cases, such as COW of the block at slot zero in the node. This + * ignores those rules, and it should only be called when there are no + * more updates to be done higher up in the tree. + */ +noinline void btrfs_unlock_up_safe(struct btrfs_path *path, int level) +{ + int i; + + if (path->keep_locks) + return; + + for (i = level; i < BTRFS_MAX_LEVEL; i++) { + if (!path->nodes[i]) + continue; + if (!path->locks[i]) + continue; + btrfs_tree_unlock(path->nodes[i]); + path->locks[i] = 0; + } +} + +/* + * helper function for btrfs_search_slot. The goal is to find a block + * in cache without setting the path to blocking. If we find the block + * we return zero and the path is unchanged. + * + * If we can't find the block, we set the path blocking and do some + * reada. -EAGAIN is returned and the search must be repeated. + */ +static int +read_block_for_search(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_path *p, + struct extent_buffer **eb_ret, int level, int slot, + struct btrfs_key *key) +{ + u64 blocknr; + u64 gen; + u32 blocksize; + struct extent_buffer *b = *eb_ret; + struct extent_buffer *tmp; + int ret; + + blocknr = btrfs_node_blockptr(b, slot); + gen = btrfs_node_ptr_generation(b, slot); + blocksize = btrfs_level_size(root, level - 1); + + tmp = btrfs_find_tree_block(root, blocknr, blocksize); + if (tmp && btrfs_buffer_uptodate(tmp, gen)) { + /* + * we found an up to date block without sleeping, return + * right away + */ + *eb_ret = tmp; + return 0; + } + + /* + * reduce lock contention at high levels + * of the btree by dropping locks before + * we read. Don't release the lock on the current + * level because we need to walk this node to figure + * out which blocks to read. + */ + btrfs_unlock_up_safe(p, level + 1); + btrfs_set_path_blocking(p); + + if (tmp) + free_extent_buffer(tmp); + if (p->reada) + reada_for_search(root, p, level, slot, key->objectid); + + btrfs_release_path(NULL, p); + + ret = -EAGAIN; + tmp = read_tree_block(root, blocknr, blocksize, 0); + if (tmp) { + /* + * If the read above didn't mark this buffer up to date, + * it will never end up being up to date. Set ret to EIO now + * and give up so that our caller doesn't loop forever + * on our EAGAINs. + */ + if (!btrfs_buffer_uptodate(tmp, 0)) + ret = -EIO; + free_extent_buffer(tmp); + } + return ret; +} + +/* + * helper function for btrfs_search_slot. This does all of the checks + * for node-level blocks and does any balancing required based on + * the ins_len. + * + * If no extra work was required, zero is returned. If we had to + * drop the path, -EAGAIN is returned and btrfs_search_slot must + * start over + */ +static int +setup_nodes_for_search(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_path *p, + struct extent_buffer *b, int level, int ins_len) +{ + int ret; + if ((p->search_for_split || ins_len > 0) && btrfs_header_nritems(b) >= + BTRFS_NODEPTRS_PER_BLOCK(root) - 3) { + int sret; + + sret = reada_for_balance(root, p, level); + if (sret) + goto again; + + btrfs_set_path_blocking(p); + sret = split_node(trans, root, p, level); + btrfs_clear_path_blocking(p, NULL); + + BUG_ON(sret > 0); + if (sret) { + ret = sret; + goto done; + } + b = p->nodes[level]; + } else if (ins_len < 0 && btrfs_header_nritems(b) < + BTRFS_NODEPTRS_PER_BLOCK(root) / 2) { + int sret; + + sret = reada_for_balance(root, p, level); + if (sret) + goto again; + + btrfs_set_path_blocking(p); + sret = balance_level(trans, root, p, level); + btrfs_clear_path_blocking(p, NULL); + + if (sret) { + ret = sret; + goto done; + } + b = p->nodes[level]; + if (!b) { + btrfs_release_path(NULL, p); + goto again; + } + BUG_ON(btrfs_header_nritems(b) == 1); + } + return 0; + +again: + ret = -EAGAIN; +done: + return ret; +} + +/* + * look for key in the tree. path is filled in with nodes along the way + * if key is found, we return zero and you can find the item in the leaf + * level of the path (level 0) + * + * If the key isn't found, the path points to the slot where it should + * be inserted, and 1 is returned. If there are other errors during the + * search a negative error number is returned. + * + * if ins_len > 0, nodes and leaves will be split as we walk down the * tree. if ins_len < 0, nodes will be merged as we walk down the tree (if * possible) */ @@ -1097,99 +1701,173 @@ int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root ins_len, int cow) { struct extent_buffer *b; - u64 bytenr; - u64 ptr_gen; int slot; int ret; + int err; int level; - int should_reada = p->reada; + int lowest_unlock = 1; u8 lowest_level = 0; lowest_level = p->lowest_level; - WARN_ON(lowest_level && ins_len); + WARN_ON(lowest_level && ins_len > 0); WARN_ON(p->nodes[0] != NULL); - WARN_ON(!mutex_is_locked(&root->fs_info->fs_mutex)); + + if (ins_len < 0) + lowest_unlock = 2; + again: - b = root->node; - extent_buffer_get(b); + if (p->search_commit_root) { + b = root->commit_root; + extent_buffer_get(b); + if (!p->skip_locking) + btrfs_tree_lock(b); + } else { + if (p->skip_locking) + b = btrfs_root_node(root); + else + b = btrfs_lock_root_node(root); + } + while (b) { level = btrfs_header_level(b); + + /* + * setup the path here so we can release it under lock + * contention with the cow code + */ + p->nodes[level] = b; + if (!p->skip_locking) + p->locks[level] = 1; + if (cow) { - int wret; - wret = btrfs_cow_block(trans, root, b, - p->nodes[level + 1], - p->slots[level + 1], - &b); - if (wret) { - free_extent_buffer(b); - return wret; + /* + * if we don't really need to cow this block + * then we don't want to set the path blocking, + * so we test it here + */ + if (!should_cow_block(trans, root, b)) + goto cow_done; + + btrfs_set_path_blocking(p); + + err = btrfs_cow_block(trans, root, b, + p->nodes[level + 1], + p->slots[level + 1], &b); + if (err) { + ret = err; + goto done; } } +cow_done: BUG_ON(!cow && ins_len); if (level != btrfs_header_level(b)) WARN_ON(1); level = btrfs_header_level(b); + p->nodes[level] = b; + if (!p->skip_locking) + p->locks[level] = 1; + + btrfs_clear_path_blocking(p, NULL); + + /* + * we have a lock on b and as long as we aren't changing + * the tree, there is no way to for the items in b to change. + * It is safe to drop the lock on our parent before we + * go through the expensive btree search on b. + * + * If cow is true, then we might be changing slot zero, + * which may require changing the parent. So, we can't + * drop the lock until after we know which slot we're + * operating on. + */ + if (!cow) + btrfs_unlock_up_safe(p, level + 1); + ret = check_block(root, p, level); - if (ret) - return -1; + if (ret) { + ret = -1; + goto done; + } + ret = bin_search(b, key, level, &slot); + if (level != 0) { - if (ret && slot > 0) + int dec = 0; + if (ret && slot > 0) { + dec = 1; slot -= 1; + } p->slots[level] = slot; - if (ins_len > 0 && btrfs_header_nritems(b) >= - BTRFS_NODEPTRS_PER_BLOCK(root) - 1) { - int sret = split_node(trans, root, p, level); - BUG_ON(sret > 0); - if (sret) - return sret; - b = p->nodes[level]; - slot = p->slots[level]; - } else if (ins_len < 0) { - int sret = balance_level(trans, root, p, - level); - if (sret) - return sret; - b = p->nodes[level]; - if (!b) { - btrfs_release_path(NULL, p); - goto again; - } - slot = p->slots[level]; - BUG_ON(btrfs_header_nritems(b) == 1); + err = setup_nodes_for_search(trans, root, p, b, level, + ins_len); + if (err == -EAGAIN) + goto again; + if (err) { + ret = err; + goto done; } - /* this is only true while dropping a snapshot */ - if (level == lowest_level) - break; - bytenr = btrfs_node_blockptr(b, slot); - ptr_gen = btrfs_node_ptr_generation(b, slot); - if (should_reada) - reada_for_search(root, p, level, slot, - key->objectid); - b = read_tree_block(root, bytenr, - btrfs_level_size(root, level - 1)); - if (ptr_gen != btrfs_header_generation(b)) { - printk("block %llu bad gen wanted %llu " - "found %llu\n", - (unsigned long long)b->start, - (unsigned long long)ptr_gen, - (unsigned long long)btrfs_header_generation(b)); + b = p->nodes[level]; + slot = p->slots[level]; + + unlock_up(p, level, lowest_unlock); + + if (level == lowest_level) { + if (dec) + p->slots[level]++; + goto done; + } + + err = read_block_for_search(trans, root, p, + &b, level, slot, key); + if (err == -EAGAIN) + goto again; + if (err) { + ret = err; + goto done; + } + + if (!p->skip_locking) { + btrfs_clear_path_blocking(p, NULL); + err = btrfs_try_spin_lock(b); + + if (!err) { + btrfs_set_path_blocking(p); + btrfs_tree_lock(b); + btrfs_clear_path_blocking(p, b); + } } } else { p->slots[level] = slot; - if (ins_len > 0 && btrfs_leaf_free_space(root, b) < - sizeof(struct btrfs_item) + ins_len) { - int sret = split_leaf(trans, root, key, - p, ins_len, ret == 0); - BUG_ON(sret > 0); - if (sret) - return sret; + if (ins_len > 0 && + btrfs_leaf_free_space(root, b) < ins_len) { + btrfs_set_path_blocking(p); + err = split_leaf(trans, root, key, + p, ins_len, ret == 0); + btrfs_clear_path_blocking(p, NULL); + + BUG_ON(err > 0); + if (err) { + ret = err; + goto done; + } } - return ret; + if (!p->search_for_split) + unlock_up(p, level, lowest_unlock); + goto done; } } - return 1; + ret = 1; +done: + /* + * we don't really know what they plan on doing with the path + * from here on, so for now just mark it as blocking + */ + if (!p->leave_spinning) + btrfs_set_path_blocking(p); + if (ret < 0) + btrfs_release_path(root, p); + return ret; } /* @@ -1224,6 +1902,41 @@ static int fixup_low_keys(struct btrfs_trans_handle *trans, } /* + * update item key. + * + * This function isn't completely safe. It's the caller's responsibility + * that the new key won't break the order + */ +int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_path *path, + struct btrfs_key *new_key) +{ + struct btrfs_disk_key disk_key; + struct extent_buffer *eb; + int slot; + + eb = path->nodes[0]; + slot = path->slots[0]; + if (slot > 0) { + btrfs_item_key(eb, &disk_key, slot - 1); + if (comp_keys(&disk_key, new_key) >= 0) + return -1; + } + if (slot < btrfs_header_nritems(eb) - 1) { + btrfs_item_key(eb, &disk_key, slot + 1); + if (comp_keys(&disk_key, new_key) <= 0) + return -1; + } + + btrfs_cpu_key_to_disk(&disk_key, new_key); + btrfs_set_item_key(eb, &disk_key, slot); + btrfs_mark_buffer_dirty(eb); + if (slot == 0) + fixup_low_keys(trans, root, path, &disk_key, 1); + return 0; +} + +/* * try to push data from one node into the next node left in the * tree. * @@ -1232,7 +1945,7 @@ static int fixup_low_keys(struct btrfs_trans_handle *trans, */ static int push_node_left(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *dst, - struct extent_buffer *src) + struct extent_buffer *src, int empty) { int push_items = 0; int src_nritems; @@ -1245,17 +1958,31 @@ static int push_node_left(struct btrfs_trans_handle *trans, WARN_ON(btrfs_header_generation(src) != trans->transid); WARN_ON(btrfs_header_generation(dst) != trans->transid); - if (push_items <= 0) { + if (!empty && src_nritems <= 8) return 1; - } - if (src_nritems < push_items) - push_items = src_nritems; + if (push_items <= 0) + return 1; + + if (empty) { + push_items = min(src_nritems, push_items); + if (push_items < src_nritems) { + /* leave at least 8 pointers in the node if + * we aren't going to empty it + */ + if (src_nritems - push_items < 8) { + if (push_items <= 8) + return 1; + push_items -= 8; + } + } + } else + push_items = min(src_nritems - 8, push_items); copy_extent_buffer(dst, src, btrfs_node_key_ptr_offset(dst_nritems), btrfs_node_key_ptr_offset(0), - push_items * sizeof(struct btrfs_key_ptr)); + push_items * sizeof(struct btrfs_key_ptr)); if (push_items < src_nritems) { memmove_extent_buffer(src, btrfs_node_key_ptr_offset(0), @@ -1267,6 +1994,7 @@ static int push_node_left(struct btrfs_trans_handle *trans, btrfs_set_header_nritems(dst, dst_nritems + push_items); btrfs_mark_buffer_dirty(src); btrfs_mark_buffer_dirty(dst); + return ret; } @@ -1299,6 +2027,9 @@ static int balance_node_right(struct btrfs_trans_handle *trans, if (push_items <= 0) return 1; + if (src_nritems < 4) + return 1; + max_push = src_nritems / 2 + 1; /* don't try to empty the node */ if (max_push >= src_nritems) @@ -1315,13 +2046,14 @@ static int balance_node_right(struct btrfs_trans_handle *trans, copy_extent_buffer(dst, src, btrfs_node_key_ptr_offset(0), btrfs_node_key_ptr_offset(src_nritems - push_items), - push_items * sizeof(struct btrfs_key_ptr)); + push_items * sizeof(struct btrfs_key_ptr)); btrfs_set_header_nritems(src, src_nritems - push_items); btrfs_set_header_nritems(dst, dst_nritems + push_items); btrfs_mark_buffer_dirty(src); btrfs_mark_buffer_dirty(dst); + return ret; } @@ -1332,74 +2064,71 @@ static int balance_node_right(struct btrfs_trans_handle *trans, * * returns zero on success or < 0 on failure. */ -static int noinline insert_new_root(struct btrfs_trans_handle *trans, +static noinline int insert_new_root(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, int level) { - u64 root_gen; u64 lower_gen; struct extent_buffer *lower; struct extent_buffer *c; + struct extent_buffer *old; struct btrfs_disk_key lower_key; BUG_ON(path->nodes[level]); BUG_ON(path->nodes[level-1] != root->node); - if (root->ref_cows) - root_gen = trans->transid; - else - root_gen = 0; - lower = path->nodes[level-1]; if (level == 1) btrfs_item_key(lower, &lower_key, 0); else btrfs_node_key(lower, &lower_key, 0); - c = __btrfs_alloc_free_block(trans, root, root->nodesize, - root->root_key.objectid, - root_gen, lower_key.objectid, level, - root->node->start, 0); + c = btrfs_alloc_free_block(trans, root, root->nodesize, 0, + root->root_key.objectid, &lower_key, + level, root->node->start, 0); if (IS_ERR(c)) return PTR_ERR(c); - memset_extent_buffer(c, 0, 0, root->nodesize); + + root_add_used(root, root->nodesize); + + memset_extent_buffer(c, 0, 0, sizeof(struct btrfs_header)); btrfs_set_header_nritems(c, 1); btrfs_set_header_level(c, level); btrfs_set_header_bytenr(c, c->start); btrfs_set_header_generation(c, trans->transid); + btrfs_set_header_backref_rev(c, BTRFS_MIXED_BACKREF_REV); btrfs_set_header_owner(c, root->root_key.objectid); write_extent_buffer(c, root->fs_info->fsid, (unsigned long)btrfs_header_fsid(c), BTRFS_FSID_SIZE); + + write_extent_buffer(c, root->fs_info->chunk_tree_uuid, + (unsigned long)btrfs_header_chunk_tree_uuid(c), + BTRFS_UUID_SIZE); + btrfs_set_node_key(c, &lower_key, 0); btrfs_set_node_blockptr(c, 0, lower->start); lower_gen = btrfs_header_generation(lower); - WARN_ON(lower_gen == 0); + WARN_ON(lower_gen != trans->transid); btrfs_set_node_ptr_generation(c, 0, lower_gen); btrfs_mark_buffer_dirty(c); - /* the super has an extra ref to root->node */ - free_extent_buffer(root->node); + spin_lock(&root->node_lock); + old = root->node; root->node = c; + spin_unlock(&root->node_lock); + + /* the super has an extra ref to root->node */ + free_extent_buffer(old); + add_root_to_dirty_list(root); extent_buffer_get(c); path->nodes[level] = c; + path->locks[level] = 1; path->slots[level] = 0; - - if (root->ref_cows && lower_gen != trans->transid) { - struct btrfs_path *back_path = btrfs_alloc_path(); - int ret; - ret = btrfs_insert_extent_backref(trans, - root->fs_info->extent_root, - path, lower->start, - root->root_key.objectid, - trans->transid, 0, 0); - BUG_ON(ret); - btrfs_free_path(back_path); - } return 0; } @@ -1420,10 +2149,10 @@ static int insert_ptr(struct btrfs_trans_handle *trans, struct btrfs_root int nritems; BUG_ON(!path->nodes[level]); + btrfs_assert_tree_locked(path->nodes[level]); lower = path->nodes[level]; nritems = btrfs_header_nritems(lower); - if (slot > nritems) - BUG(); + BUG_ON(slot > nritems); if (nritems == BTRFS_NODEPTRS_PER_BLOCK(root)) BUG(); if (slot != nritems) { @@ -1450,10 +2179,10 @@ static int insert_ptr(struct btrfs_trans_handle *trans, struct btrfs_root * * returns 0 on success and < 0 on failure */ -static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root - *root, struct btrfs_path *path, int level) +static noinline int split_node(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, int level) { - u64 root_gen; struct extent_buffer *c; struct extent_buffer *split; struct btrfs_disk_key disk_key; @@ -1473,38 +2202,37 @@ static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root ret = push_nodes_for_insert(trans, root, path, level); c = path->nodes[level]; if (!ret && btrfs_header_nritems(c) < - BTRFS_NODEPTRS_PER_BLOCK(root) - 1) + BTRFS_NODEPTRS_PER_BLOCK(root) - 3) return 0; if (ret < 0) return ret; } c_nritems = btrfs_header_nritems(c); - if (root->ref_cows) - root_gen = trans->transid; - else - root_gen = 0; - - btrfs_node_key(c, &disk_key, 0); - split = __btrfs_alloc_free_block(trans, root, root->nodesize, - root->root_key.objectid, - root_gen, - btrfs_disk_key_objectid(&disk_key), - level, c->start, 0); + mid = (c_nritems + 1) / 2; + btrfs_node_key(c, &disk_key, mid); + + split = btrfs_alloc_free_block(trans, root, root->nodesize, 0, + root->root_key.objectid, + &disk_key, level, c->start, 0); if (IS_ERR(split)) return PTR_ERR(split); - btrfs_set_header_flags(split, btrfs_header_flags(c)); + root_add_used(root, root->nodesize); + + memset_extent_buffer(split, 0, 0, sizeof(struct btrfs_header)); btrfs_set_header_level(split, btrfs_header_level(c)); btrfs_set_header_bytenr(split, split->start); btrfs_set_header_generation(split, trans->transid); + btrfs_set_header_backref_rev(split, BTRFS_MIXED_BACKREF_REV); btrfs_set_header_owner(split, root->root_key.objectid); - btrfs_set_header_flags(split, 0); write_extent_buffer(split, root->fs_info->fsid, (unsigned long)btrfs_header_fsid(split), BTRFS_FSID_SIZE); + write_extent_buffer(split, root->fs_info->chunk_tree_uuid, + (unsigned long)btrfs_header_chunk_tree_uuid(split), + BTRFS_UUID_SIZE); - mid = (c_nritems + 1) / 2; copy_extent_buffer(split, c, btrfs_node_key_ptr_offset(0), @@ -1517,7 +2245,6 @@ static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root btrfs_mark_buffer_dirty(c); btrfs_mark_buffer_dirty(split); - btrfs_node_key(split, &disk_key, 0); wret = insert_ptr(trans, root, path, &disk_key, split->start, path->slots[level + 1] + 1, level + 1); @@ -1526,10 +2253,12 @@ static int split_node(struct btrfs_trans_handle *trans, struct btrfs_root if (path->slots[level] >= mid) { path->slots[level] -= mid; + btrfs_tree_unlock(c); free_extent_buffer(c); path->nodes[level] = split; path->slots[level + 1] += 1; } else { + btrfs_tree_unlock(split); free_extent_buffer(split); } return ret; @@ -1560,93 +2289,66 @@ static int leaf_space_used(struct extent_buffer *l, int start, int nr) * the start of the leaf data. IOW, how much room * the leaf has left for both items and data */ -int btrfs_leaf_free_space(struct btrfs_root *root, struct extent_buffer *leaf) +noinline int btrfs_leaf_free_space(struct btrfs_root *root, + struct extent_buffer *leaf) { int nritems = btrfs_header_nritems(leaf); int ret; ret = BTRFS_LEAF_DATA_SIZE(root) - leaf_space_used(leaf, 0, nritems); if (ret < 0) { - printk("leaf free space ret %d, leaf data size %lu, used %d nritems %d\n", + printk(KERN_CRIT "leaf free space ret %d, leaf data size %lu, " + "used %d nritems %d\n", ret, (unsigned long) BTRFS_LEAF_DATA_SIZE(root), leaf_space_used(leaf, 0, nritems), nritems); } return ret; } -/* - * push some data in the path leaf to the right, trying to free up at - * least data_size bytes. returns zero if the push worked, nonzero otherwise - * - * returns 1 if the push failed because the other node didn't have enough - * room, 0 if everything worked out and < 0 if there were major errors. - */ -static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root - *root, struct btrfs_path *path, int data_size, - int empty) +static noinline int __push_leaf_right(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + int data_size, int empty, + struct extent_buffer *right, + int free_space, u32 left_nritems) { struct extent_buffer *left = path->nodes[0]; - struct extent_buffer *right; - struct extent_buffer *upper; + struct extent_buffer *upper = path->nodes[1]; struct btrfs_disk_key disk_key; int slot; u32 i; - int free_space; int push_space = 0; int push_items = 0; struct btrfs_item *item; - u32 left_nritems; u32 nr; u32 right_nritems; u32 data_end; u32 this_item_size; - int ret; + + if (empty) + nr = 0; + else + nr = 1; + + if (path->slots[0] >= left_nritems) + push_space += data_size; slot = path->slots[1]; - if (!path->nodes[1]) { - return 1; - } - upper = path->nodes[1]; - if (slot >= btrfs_header_nritems(upper) - 1) - return 1; + i = left_nritems - 1; + while (i >= nr) { + item = btrfs_item_nr(left, i); - right = read_tree_block(root, btrfs_node_blockptr(upper, slot + 1), - root->leafsize); - free_space = btrfs_leaf_free_space(root, right); - if (free_space < data_size + sizeof(struct btrfs_item)) { - free_extent_buffer(right); - return 1; - } - - /* cow and double check */ - ret = btrfs_cow_block(trans, root, right, upper, - slot + 1, &right); - if (ret) { - free_extent_buffer(right); - return 1; - } - free_space = btrfs_leaf_free_space(root, right); - if (free_space < data_size + sizeof(struct btrfs_item)) { - free_extent_buffer(right); - return 1; - } - - left_nritems = btrfs_header_nritems(left); - if (left_nritems == 0) { - free_extent_buffer(right); - return 1; - } - - if (empty) - nr = 0; - else - nr = 1; - - i = left_nritems - 1; - while (i >= nr) { - item = btrfs_item_nr(left, i); + if (!empty && push_items > 0) { + if (path->slots[0] > i) + break; + if (path->slots[0] == i) { + int space = btrfs_leaf_free_space(root, left); + if (space + push_space * 2 > free_space) + break; + } + } if (path->slots[0] == i) - push_space += data_size + sizeof(*item); + push_space += data_size; if (!left->map_token) { map_extent_buffer(left, (unsigned long)item, @@ -1659,6 +2361,7 @@ static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root this_item_size = btrfs_item_size(left, item); if (this_item_size + sizeof(*item) + push_space > free_space) break; + push_items++; push_space += this_item_size + sizeof(*item); if (i == 0) @@ -1670,10 +2373,8 @@ static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root left->map_token = NULL; } - if (push_items == 0) { - free_extent_buffer(right); - return 1; - } + if (push_items == 0) + goto out_unlock; if (!empty && push_items == left_nritems) WARN_ON(1); @@ -1732,6 +2433,9 @@ static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root if (left_nritems) btrfs_mark_buffer_dirty(left); + else + clean_tree_block(trans, root, left); + btrfs_mark_buffer_dirty(right); btrfs_item_key(right, &disk_key, 0); @@ -1741,33 +2445,101 @@ static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root /* then fixup the leaf pointer in the path */ if (path->slots[0] >= left_nritems) { path->slots[0] -= left_nritems; + if (btrfs_header_nritems(path->nodes[0]) == 0) + clean_tree_block(trans, root, path->nodes[0]); + btrfs_tree_unlock(path->nodes[0]); free_extent_buffer(path->nodes[0]); path->nodes[0] = right; path->slots[1] += 1; } else { + btrfs_tree_unlock(right); free_extent_buffer(right); } return 0; + +out_unlock: + btrfs_tree_unlock(right); + free_extent_buffer(right); + return 1; +} + +/* + * push some data in the path leaf to the right, trying to free up at + * least data_size bytes. returns zero if the push worked, nonzero otherwise + * + * returns 1 if the push failed because the other node didn't have enough + * room, 0 if everything worked out and < 0 if there were major errors. + */ +static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root + *root, struct btrfs_path *path, int data_size, + int empty) +{ + struct extent_buffer *left = path->nodes[0]; + struct extent_buffer *right; + struct extent_buffer *upper; + int slot; + int free_space; + u32 left_nritems; + int ret; + + if (!path->nodes[1]) + return 1; + + slot = path->slots[1]; + upper = path->nodes[1]; + if (slot >= btrfs_header_nritems(upper) - 1) + return 1; + + btrfs_assert_tree_locked(path->nodes[1]); + + right = read_node_slot(root, upper, slot + 1); + btrfs_tree_lock(right); + btrfs_set_lock_blocking(right); + + free_space = btrfs_leaf_free_space(root, right); + if (free_space < data_size) + goto out_unlock; + + /* cow and double check */ + ret = btrfs_cow_block(trans, root, right, upper, + slot + 1, &right); + if (ret) + goto out_unlock; + + free_space = btrfs_leaf_free_space(root, right); + if (free_space < data_size) + goto out_unlock; + + left_nritems = btrfs_header_nritems(left); + if (left_nritems == 0) + goto out_unlock; + + return __push_leaf_right(trans, root, path, data_size, empty, + right, free_space, left_nritems); +out_unlock: + btrfs_tree_unlock(right); + free_extent_buffer(right); + return 1; } + /* * push some data in the path leaf to the left, trying to free up at * least data_size bytes. returns zero if the push worked, nonzero otherwise */ -static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root - *root, struct btrfs_path *path, int data_size, - int empty) +static noinline int __push_leaf_left(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, int data_size, + int empty, struct extent_buffer *left, + int free_space, int right_nritems) { struct btrfs_disk_key disk_key; struct extent_buffer *right = path->nodes[0]; - struct extent_buffer *left; int slot; int i; - int free_space; int push_space = 0; int push_items = 0; struct btrfs_item *item; u32 old_left_nritems; - u32 right_nritems; u32 nr; int ret = 0; int wret; @@ -1775,38 +2547,6 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root u32 old_left_item_size; slot = path->slots[1]; - if (slot == 0) - return 1; - if (!path->nodes[1]) - return 1; - - right_nritems = btrfs_header_nritems(right); - if (right_nritems == 0) { - return 1; - } - - left = read_tree_block(root, btrfs_node_blockptr(path->nodes[1], - slot - 1), root->leafsize); - free_space = btrfs_leaf_free_space(root, left); - if (free_space < data_size + sizeof(struct btrfs_item)) { - free_extent_buffer(left); - return 1; - } - - /* cow and double check */ - ret = btrfs_cow_block(trans, root, left, - path->nodes[1], slot - 1, &left); - if (ret) { - /* we hit -ENOSPC, but it isn't fatal here */ - free_extent_buffer(left); - return 1; - } - - free_space = btrfs_leaf_free_space(root, left); - if (free_space < data_size + sizeof(struct btrfs_item)) { - free_extent_buffer(left); - return 1; - } if (empty) nr = right_nritems; @@ -1823,8 +2563,18 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root KM_USER1); } + if (!empty && push_items > 0) { + if (path->slots[0] < i) + break; + if (path->slots[0] == i) { + int space = btrfs_leaf_free_space(root, right); + if (space + push_space * 2 > free_space) + break; + } + } + if (path->slots[0] == i) - push_space += data_size + sizeof(*item); + push_space += data_size; this_item_size = btrfs_item_size(right, item); if (this_item_size + sizeof(*item) + push_space > free_space) @@ -1840,8 +2590,8 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root } if (push_items == 0) { - free_extent_buffer(left); - return 1; + ret = 1; + goto out; } if (!empty && push_items == btrfs_header_nritems(right)) WARN_ON(1); @@ -1853,7 +2603,7 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root push_items * sizeof(struct btrfs_item)); push_space = BTRFS_LEAF_DATA_SIZE(root) - - btrfs_item_offset_nr(right, push_items -1); + btrfs_item_offset_nr(right, push_items - 1); copy_extent_buffer(left, right, btrfs_leaf_data(left) + leaf_data_end(root, left) - push_space, @@ -1861,7 +2611,7 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root btrfs_item_offset_nr(right, push_items - 1), push_space); old_left_nritems = btrfs_header_nritems(left); - BUG_ON(old_left_nritems < 0); + BUG_ON(old_left_nritems <= 0); old_left_item_size = btrfs_item_offset_nr(left, old_left_nritems - 1); for (i = old_left_nritems; i < old_left_nritems + push_items; i++) { @@ -1888,7 +2638,8 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root /* fixup right node */ if (push_items > right_nritems) { - printk("push items %d nr %u\n", push_items, right_nritems); + printk(KERN_CRIT "push items %d nr %u\n", push_items, + right_nritems); WARN_ON(1); } @@ -1930,6 +2681,8 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root btrfs_mark_buffer_dirty(left); if (right_nritems) btrfs_mark_buffer_dirty(right); + else + clean_tree_block(trans, root, right); btrfs_item_key(right, &disk_key, 0); wret = fixup_low_keys(trans, root, path, &disk_key, 1); @@ -1939,161 +2692,103 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root /* then fixup the leaf pointer in the path */ if (path->slots[0] < push_items) { path->slots[0] += old_left_nritems; + btrfs_tree_unlock(path->nodes[0]); free_extent_buffer(path->nodes[0]); path->nodes[0] = left; path->slots[1] -= 1; } else { + btrfs_tree_unlock(left); free_extent_buffer(left); path->slots[0] -= push_items; } BUG_ON(path->slots[0] < 0); return ret; +out: + btrfs_tree_unlock(left); + free_extent_buffer(left); + return ret; } /* - * split the path's leaf in two, making sure there is at least data_size - * available for the resulting leaf level of the path. - * - * returns 0 if all went well and < 0 on failure. + * push some data in the path leaf to the left, trying to free up at + * least data_size bytes. returns zero if the push worked, nonzero otherwise */ -static int split_leaf(struct btrfs_trans_handle *trans, struct btrfs_root - *root, struct btrfs_key *ins_key, - struct btrfs_path *path, int data_size, int extend) +static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root + *root, struct btrfs_path *path, int data_size, + int empty) { - u64 root_gen; - struct extent_buffer *l; - u32 nritems; - int mid; + struct extent_buffer *right = path->nodes[0]; + struct extent_buffer *left; int slot; - struct extent_buffer *right; - int space_needed = data_size + sizeof(struct btrfs_item); - int data_copy_size; - int rt_data_off; - int i; + int free_space; + u32 right_nritems; int ret = 0; - int wret; - int double_split; - int num_doubles = 0; - struct btrfs_disk_key disk_key; - if (extend) - space_needed = data_size; + slot = path->slots[1]; + if (slot == 0) + return 1; + if (!path->nodes[1]) + return 1; - if (root->ref_cows) - root_gen = trans->transid; - else - root_gen = 0; + right_nritems = btrfs_header_nritems(right); + if (right_nritems == 0) + return 1; - /* first try to make some room by pushing left and right */ - if (ins_key->type != BTRFS_DIR_ITEM_KEY) { - wret = push_leaf_right(trans, root, path, data_size, 0); - if (wret < 0) { - return wret; - } - if (wret) { - wret = push_leaf_left(trans, root, path, data_size, 0); - if (wret < 0) - return wret; - } - l = path->nodes[0]; + btrfs_assert_tree_locked(path->nodes[1]); - /* did the pushes work? */ - if (btrfs_leaf_free_space(root, l) >= space_needed) - return 0; - } + left = read_node_slot(root, path->nodes[1], slot - 1); + btrfs_tree_lock(left); + btrfs_set_lock_blocking(left); - if (!path->nodes[1]) { - ret = insert_new_root(trans, root, path, 1); - if (ret) - return ret; + free_space = btrfs_leaf_free_space(root, left); + if (free_space < data_size) { + ret = 1; + goto out; } -again: - double_split = 0; - l = path->nodes[0]; - slot = path->slots[0]; - nritems = btrfs_header_nritems(l); - mid = (nritems + 1)/ 2; - btrfs_item_key(l, &disk_key, 0); - - right = __btrfs_alloc_free_block(trans, root, root->leafsize, - root->root_key.objectid, - root_gen, disk_key.objectid, 0, - l->start, 0); - if (IS_ERR(right)) { - BUG_ON(1); - return PTR_ERR(right); + /* cow and double check */ + ret = btrfs_cow_block(trans, root, left, + path->nodes[1], slot - 1, &left); + if (ret) { + /* we hit -ENOSPC, but it isn't fatal here */ + ret = 1; + goto out; } - memset_extent_buffer(right, 0, 0, sizeof(struct btrfs_header)); - btrfs_set_header_bytenr(right, right->start); - btrfs_set_header_generation(right, trans->transid); - btrfs_set_header_owner(right, root->root_key.objectid); - btrfs_set_header_level(right, 0); - write_extent_buffer(right, root->fs_info->fsid, - (unsigned long)btrfs_header_fsid(right), - BTRFS_FSID_SIZE); - if (mid <= slot) { - if (nritems == 1 || - leaf_space_used(l, mid, nritems - mid) + space_needed > - BTRFS_LEAF_DATA_SIZE(root)) { - if (slot >= nritems) { - btrfs_cpu_key_to_disk(&disk_key, ins_key); - btrfs_set_header_nritems(right, 0); - wret = insert_ptr(trans, root, path, - &disk_key, right->start, - path->slots[1] + 1, 1); - if (wret) - ret = wret; - free_extent_buffer(path->nodes[0]); - path->nodes[0] = right; - path->slots[0] = 0; - path->slots[1] += 1; - btrfs_mark_buffer_dirty(right); - return ret; - } - mid = slot; - if (mid != nritems && - leaf_space_used(l, mid, nritems - mid) + - space_needed > BTRFS_LEAF_DATA_SIZE(root)) { - double_split = 1; - } - } - } else { - if (leaf_space_used(l, 0, mid + 1) + space_needed > - BTRFS_LEAF_DATA_SIZE(root)) { - if (!extend && slot == 0) { - btrfs_cpu_key_to_disk(&disk_key, ins_key); - btrfs_set_header_nritems(right, 0); - wret = insert_ptr(trans, root, path, - &disk_key, - right->start, - path->slots[1], 1); - if (wret) - ret = wret; - free_extent_buffer(path->nodes[0]); - path->nodes[0] = right; - path->slots[0] = 0; - if (path->slots[1] == 0) { - wret = fixup_low_keys(trans, root, - path, &disk_key, 1); - if (wret) - ret = wret; - } - btrfs_mark_buffer_dirty(right); - return ret; - } else if (extend && slot == 0) { - mid = 1; - } else { - mid = slot; - if (mid != nritems && - leaf_space_used(l, mid, nritems - mid) + - space_needed > BTRFS_LEAF_DATA_SIZE(root)) { - double_split = 1; - } - } - } + free_space = btrfs_leaf_free_space(root, left); + if (free_space < data_size) { + ret = 1; + goto out; } + + return __push_leaf_left(trans, root, path, data_size, + empty, left, free_space, right_nritems); +out: + btrfs_tree_unlock(left); + free_extent_buffer(left); + return ret; +} + +/* + * split the path's leaf in two, making sure there is at least data_size + * available for the resulting leaf level of the path. + * + * returns 0 if all went well and < 0 on failure. + */ +static noinline int copy_for_split(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct extent_buffer *l, + struct extent_buffer *right, + int slot, int mid, int nritems) +{ + int data_copy_size; + int rt_data_off; + int i; + int ret = 0; + int wret; + struct btrfs_disk_key disk_key; + nritems = nritems - mid; btrfs_set_header_nritems(right, nritems); data_copy_size = btrfs_item_end_nr(l, mid) - leaf_data_end(root, l); @@ -2144,32 +2839,408 @@ again: BUG_ON(path->slots[0] != slot); if (mid <= slot) { + btrfs_tree_unlock(path->nodes[0]); free_extent_buffer(path->nodes[0]); path->nodes[0] = right; path->slots[0] -= mid; path->slots[1] += 1; - } else + } else { + btrfs_tree_unlock(right); free_extent_buffer(right); + } BUG_ON(path->slots[0] < 0); - if (double_split) { - BUG_ON(num_doubles != 0); - num_doubles++; - goto again; - } return ret; } -int btrfs_truncate_item(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_path *path, - u32 new_size, int from_end) -{ - int ret = 0; - int slot; - int slot_orig; - struct extent_buffer *leaf; +/* + * split the path's leaf in two, making sure there is at least data_size + * available for the resulting leaf level of the path. + * + * returns 0 if all went well and < 0 on failure. + */ +static noinline int split_leaf(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_key *ins_key, + struct btrfs_path *path, int data_size, + int extend) +{ + struct btrfs_disk_key disk_key; + struct extent_buffer *l; + u32 nritems; + int mid; + int slot; + struct extent_buffer *right; + int ret = 0; + int wret; + int split; + int num_doubles = 0; + + l = path->nodes[0]; + slot = path->slots[0]; + if (extend && data_size + btrfs_item_size_nr(l, slot) + + sizeof(struct btrfs_item) > BTRFS_LEAF_DATA_SIZE(root)) + return -EOVERFLOW; + + /* first try to make some room by pushing left and right */ + if (data_size && ins_key->type != BTRFS_DIR_ITEM_KEY) { + wret = push_leaf_right(trans, root, path, data_size, 0); + if (wret < 0) + return wret; + if (wret) { + wret = push_leaf_left(trans, root, path, data_size, 0); + if (wret < 0) + return wret; + } + l = path->nodes[0]; + + /* did the pushes work? */ + if (btrfs_leaf_free_space(root, l) >= data_size) + return 0; + } + + if (!path->nodes[1]) { + ret = insert_new_root(trans, root, path, 1); + if (ret) + return ret; + } +again: + split = 1; + l = path->nodes[0]; + slot = path->slots[0]; + nritems = btrfs_header_nritems(l); + mid = (nritems + 1) / 2; + + if (mid <= slot) { + if (nritems == 1 || + leaf_space_used(l, mid, nritems - mid) + data_size > + BTRFS_LEAF_DATA_SIZE(root)) { + if (slot >= nritems) { + split = 0; + } else { + mid = slot; + if (mid != nritems && + leaf_space_used(l, mid, nritems - mid) + + data_size > BTRFS_LEAF_DATA_SIZE(root)) { + split = 2; + } + } + } + } else { + if (leaf_space_used(l, 0, mid) + data_size > + BTRFS_LEAF_DATA_SIZE(root)) { + if (!extend && data_size && slot == 0) { + split = 0; + } else if ((extend || !data_size) && slot == 0) { + mid = 1; + } else { + mid = slot; + if (mid != nritems && + leaf_space_used(l, mid, nritems - mid) + + data_size > BTRFS_LEAF_DATA_SIZE(root)) { + split = 2 ; + } + } + } + } + + if (split == 0) + btrfs_cpu_key_to_disk(&disk_key, ins_key); + else + btrfs_item_key(l, &disk_key, mid); + + right = btrfs_alloc_free_block(trans, root, root->leafsize, 0, + root->root_key.objectid, + &disk_key, 0, l->start, 0); + if (IS_ERR(right)) + return PTR_ERR(right); + + root_add_used(root, root->leafsize); + + memset_extent_buffer(right, 0, 0, sizeof(struct btrfs_header)); + btrfs_set_header_bytenr(right, right->start); + btrfs_set_header_generation(right, trans->transid); + btrfs_set_header_backref_rev(right, BTRFS_MIXED_BACKREF_REV); + btrfs_set_header_owner(right, root->root_key.objectid); + btrfs_set_header_level(right, 0); + write_extent_buffer(right, root->fs_info->fsid, + (unsigned long)btrfs_header_fsid(right), + BTRFS_FSID_SIZE); + + write_extent_buffer(right, root->fs_info->chunk_tree_uuid, + (unsigned long)btrfs_header_chunk_tree_uuid(right), + BTRFS_UUID_SIZE); + + if (split == 0) { + if (mid <= slot) { + btrfs_set_header_nritems(right, 0); + wret = insert_ptr(trans, root, path, + &disk_key, right->start, + path->slots[1] + 1, 1); + if (wret) + ret = wret; + + btrfs_tree_unlock(path->nodes[0]); + free_extent_buffer(path->nodes[0]); + path->nodes[0] = right; + path->slots[0] = 0; + path->slots[1] += 1; + } else { + btrfs_set_header_nritems(right, 0); + wret = insert_ptr(trans, root, path, + &disk_key, + right->start, + path->slots[1], 1); + if (wret) + ret = wret; + btrfs_tree_unlock(path->nodes[0]); + free_extent_buffer(path->nodes[0]); + path->nodes[0] = right; + path->slots[0] = 0; + if (path->slots[1] == 0) { + wret = fixup_low_keys(trans, root, + path, &disk_key, 1); + if (wret) + ret = wret; + } + } + btrfs_mark_buffer_dirty(right); + return ret; + } + + ret = copy_for_split(trans, root, path, l, right, slot, mid, nritems); + BUG_ON(ret); + + if (split == 2) { + BUG_ON(num_doubles != 0); + num_doubles++; + goto again; + } + + return ret; +} + +static noinline int setup_leaf_for_split(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, int ins_len) +{ + struct btrfs_key key; + struct extent_buffer *leaf; + struct btrfs_file_extent_item *fi; + u64 extent_len = 0; + u32 item_size; + int ret; + + leaf = path->nodes[0]; + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + + BUG_ON(key.type != BTRFS_EXTENT_DATA_KEY && + key.type != BTRFS_EXTENT_CSUM_KEY); + + if (btrfs_leaf_free_space(root, leaf) >= ins_len) + return 0; + + item_size = btrfs_item_size_nr(leaf, path->slots[0]); + if (key.type == BTRFS_EXTENT_DATA_KEY) { + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + extent_len = btrfs_file_extent_num_bytes(leaf, fi); + } + btrfs_release_path(root, path); + + path->keep_locks = 1; + path->search_for_split = 1; + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + path->search_for_split = 0; + if (ret < 0) + goto err; + + ret = -EAGAIN; + leaf = path->nodes[0]; + /* if our item isn't there or got smaller, return now */ + if (ret > 0 || item_size != btrfs_item_size_nr(leaf, path->slots[0])) + goto err; + + /* the leaf has changed, it now has room. return now */ + if (btrfs_leaf_free_space(root, path->nodes[0]) >= ins_len) + goto err; + + if (key.type == BTRFS_EXTENT_DATA_KEY) { + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + if (extent_len != btrfs_file_extent_num_bytes(leaf, fi)) + goto err; + } + + btrfs_set_path_blocking(path); + ret = split_leaf(trans, root, &key, path, ins_len, 1); + if (ret) + goto err; + + path->keep_locks = 0; + btrfs_unlock_up_safe(path, 1); + return 0; +err: + path->keep_locks = 0; + return ret; +} + +static noinline int split_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *new_key, + unsigned long split_offset) +{ + struct extent_buffer *leaf; + struct btrfs_item *item; + struct btrfs_item *new_item; + int slot; + char *buf; + u32 nritems; + u32 item_size; + u32 orig_offset; + struct btrfs_disk_key disk_key; + + leaf = path->nodes[0]; + BUG_ON(btrfs_leaf_free_space(root, leaf) < sizeof(struct btrfs_item)); + + btrfs_set_path_blocking(path); + + item = btrfs_item_nr(leaf, path->slots[0]); + orig_offset = btrfs_item_offset(leaf, item); + item_size = btrfs_item_size(leaf, item); + + buf = kmalloc(item_size, GFP_NOFS); + if (!buf) + return -ENOMEM; + + read_extent_buffer(leaf, buf, btrfs_item_ptr_offset(leaf, + path->slots[0]), item_size); + + slot = path->slots[0] + 1; + nritems = btrfs_header_nritems(leaf); + if (slot != nritems) { + /* shift the items */ + memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + 1), + btrfs_item_nr_offset(slot), + (nritems - slot) * sizeof(struct btrfs_item)); + } + + btrfs_cpu_key_to_disk(&disk_key, new_key); + btrfs_set_item_key(leaf, &disk_key, slot); + + new_item = btrfs_item_nr(leaf, slot); + + btrfs_set_item_offset(leaf, new_item, orig_offset); + btrfs_set_item_size(leaf, new_item, item_size - split_offset); + + btrfs_set_item_offset(leaf, item, + orig_offset + item_size - split_offset); + btrfs_set_item_size(leaf, item, split_offset); + + btrfs_set_header_nritems(leaf, nritems + 1); + + /* write the data for the start of the original item */ + write_extent_buffer(leaf, buf, + btrfs_item_ptr_offset(leaf, path->slots[0]), + split_offset); + + /* write the data for the new item */ + write_extent_buffer(leaf, buf + split_offset, + btrfs_item_ptr_offset(leaf, slot), + item_size - split_offset); + btrfs_mark_buffer_dirty(leaf); + + BUG_ON(btrfs_leaf_free_space(root, leaf) < 0); + kfree(buf); + return 0; +} + +/* + * This function splits a single item into two items, + * giving 'new_key' to the new item and splitting the + * old one at split_offset (from the start of the item). + * + * The path may be released by this operation. After + * the split, the path is pointing to the old item. The + * new item is going to be in the same node as the old one. + * + * Note, the item being split must be smaller enough to live alone on + * a tree block with room for one extra struct btrfs_item + * + * This allows us to split the item in place, keeping a lock on the + * leaf the entire time. + */ +int btrfs_split_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *new_key, + unsigned long split_offset) +{ + int ret; + ret = setup_leaf_for_split(trans, root, path, + sizeof(struct btrfs_item)); + if (ret) + return ret; + + ret = split_item(trans, root, path, new_key, split_offset); + return ret; +} + +/* + * This function duplicate a item, giving 'new_key' to the new item. + * It guarantees both items live in the same tree leaf and the new item + * is contiguous with the original item. + * + * This allows us to split file extent in place, keeping a lock on the + * leaf the entire time. + */ +int btrfs_duplicate_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *new_key) +{ + struct extent_buffer *leaf; + int ret; + u32 item_size; + + leaf = path->nodes[0]; + item_size = btrfs_item_size_nr(leaf, path->slots[0]); + ret = setup_leaf_for_split(trans, root, path, + item_size + sizeof(struct btrfs_item)); + if (ret) + return ret; + + path->slots[0]++; + ret = setup_items_for_insert(trans, root, path, new_key, &item_size, + item_size, item_size + + sizeof(struct btrfs_item), 1); + BUG_ON(ret); + + leaf = path->nodes[0]; + memcpy_extent_buffer(leaf, + btrfs_item_ptr_offset(leaf, path->slots[0]), + btrfs_item_ptr_offset(leaf, path->slots[0] - 1), + item_size); + return 0; +} + +/* + * make the item pointed to by the path smaller. new_size indicates + * how small to make it, and from_end tells us if we just chop bytes + * off the end of the item or if we shift the item to chop bytes off + * the front. + */ +int btrfs_truncate_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + u32 new_size, int from_end) +{ + int ret = 0; + int slot; + int slot_orig; + struct extent_buffer *leaf; struct btrfs_item *item; u32 nritems; unsigned int data_end; @@ -2245,8 +3316,8 @@ int btrfs_truncate_item(struct btrfs_trans_handle *trans, BTRFS_FILE_EXTENT_INLINE) { ptr = btrfs_item_ptr_offset(leaf, slot); memmove_extent_buffer(leaf, ptr, - (unsigned long)fi, - offsetof(struct btrfs_file_extent_item, + (unsigned long)fi, + offsetof(struct btrfs_file_extent_item, disk_bytenr)); } } @@ -2274,6 +3345,9 @@ int btrfs_truncate_item(struct btrfs_trans_handle *trans, return ret; } +/* + * make the item pointed to by the path bigger, data_size is the new size. + */ int btrfs_extend_item(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, u32 data_size) @@ -2305,7 +3379,8 @@ int btrfs_extend_item(struct btrfs_trans_handle *trans, BUG_ON(slot < 0); if (slot >= nritems) { btrfs_print_leaf(root, leaf); - printk("slot %d too large, nritems %d\n", slot, nritems); + printk(KERN_CRIT "slot %d too large, nritems %d\n", + slot, nritems); BUG_ON(1); } @@ -2344,76 +3419,205 @@ int btrfs_extend_item(struct btrfs_trans_handle *trans, btrfs_set_item_size(leaf, item, old_size + data_size); btrfs_mark_buffer_dirty(leaf); - ret = 0; + ret = 0; + if (btrfs_leaf_free_space(root, leaf) < 0) { + btrfs_print_leaf(root, leaf); + BUG(); + } + return ret; +} + +/* + * Given a key and some data, insert items into the tree. + * This does all the path init required, making room in the tree if needed. + * Returns the number of keys that were inserted. + */ +int btrfs_insert_some_items(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *cpu_key, u32 *data_size, + int nr) +{ + struct extent_buffer *leaf; + struct btrfs_item *item; + int ret = 0; + int slot; + int i; + u32 nritems; + u32 total_data = 0; + u32 total_size = 0; + unsigned int data_end; + struct btrfs_disk_key disk_key; + struct btrfs_key found_key; + + for (i = 0; i < nr; i++) { + if (total_size + data_size[i] + sizeof(struct btrfs_item) > + BTRFS_LEAF_DATA_SIZE(root)) { + break; + nr = i; + } + total_data += data_size[i]; + total_size += data_size[i] + sizeof(struct btrfs_item); + } + BUG_ON(nr == 0); + + ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1); + if (ret == 0) + return -EEXIST; + if (ret < 0) + goto out; + + leaf = path->nodes[0]; + + nritems = btrfs_header_nritems(leaf); + data_end = leaf_data_end(root, leaf); + + if (btrfs_leaf_free_space(root, leaf) < total_size) { + for (i = nr; i >= 0; i--) { + total_data -= data_size[i]; + total_size -= data_size[i] + sizeof(struct btrfs_item); + if (total_size < btrfs_leaf_free_space(root, leaf)) + break; + } + nr = i; + } + + slot = path->slots[0]; + BUG_ON(slot < 0); + + if (slot != nritems) { + unsigned int old_data = btrfs_item_end_nr(leaf, slot); + + item = btrfs_item_nr(leaf, slot); + btrfs_item_key_to_cpu(leaf, &found_key, slot); + + /* figure out how many keys we can insert in here */ + total_data = data_size[0]; + for (i = 1; i < nr; i++) { + if (btrfs_comp_cpu_keys(&found_key, cpu_key + i) <= 0) + break; + total_data += data_size[i]; + } + nr = i; + + if (old_data < data_end) { + btrfs_print_leaf(root, leaf); + printk(KERN_CRIT "slot %d old_data %d data_end %d\n", + slot, old_data, data_end); + BUG_ON(1); + } + /* + * item0..itemN ... dataN.offset..dataN.size .. data0.size + */ + /* first correct the data pointers */ + WARN_ON(leaf->map_token); + for (i = slot; i < nritems; i++) { + u32 ioff; + + item = btrfs_item_nr(leaf, i); + if (!leaf->map_token) { + map_extent_buffer(leaf, (unsigned long)item, + sizeof(struct btrfs_item), + &leaf->map_token, &leaf->kaddr, + &leaf->map_start, &leaf->map_len, + KM_USER1); + } + + ioff = btrfs_item_offset(leaf, item); + btrfs_set_item_offset(leaf, item, ioff - total_data); + } + if (leaf->map_token) { + unmap_extent_buffer(leaf, leaf->map_token, KM_USER1); + leaf->map_token = NULL; + } + + /* shift the items */ + memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr), + btrfs_item_nr_offset(slot), + (nritems - slot) * sizeof(struct btrfs_item)); + + /* shift the data */ + memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) + + data_end - total_data, btrfs_leaf_data(leaf) + + data_end, old_data - data_end); + data_end = old_data; + } else { + /* + * this sucks but it has to be done, if we are inserting at + * the end of the leaf only insert 1 of the items, since we + * have no way of knowing whats on the next leaf and we'd have + * to drop our current locks to figure it out + */ + nr = 1; + } + + /* setup the item for the new data */ + for (i = 0; i < nr; i++) { + btrfs_cpu_key_to_disk(&disk_key, cpu_key + i); + btrfs_set_item_key(leaf, &disk_key, slot + i); + item = btrfs_item_nr(leaf, slot + i); + btrfs_set_item_offset(leaf, item, data_end - data_size[i]); + data_end -= data_size[i]; + btrfs_set_item_size(leaf, item, data_size[i]); + } + btrfs_set_header_nritems(leaf, nritems + nr); + btrfs_mark_buffer_dirty(leaf); + + ret = 0; + if (slot == 0) { + btrfs_cpu_key_to_disk(&disk_key, cpu_key); + ret = fixup_low_keys(trans, root, path, &disk_key, 1); + } + if (btrfs_leaf_free_space(root, leaf) < 0) { btrfs_print_leaf(root, leaf); BUG(); } +out: + if (!ret) + ret = nr; return ret; } /* - * Given a key and some data, insert an item into the tree. - * This does all the path init required, making room in the tree if needed. + * this is a helper for btrfs_insert_empty_items, the main goal here is + * to save stack depth by doing the bulk of the work in a function + * that doesn't call btrfs_search_slot */ -int btrfs_insert_empty_items(struct btrfs_trans_handle *trans, - struct btrfs_root *root, - struct btrfs_path *path, - struct btrfs_key *cpu_key, u32 *data_size, - int nr) +static noinline_for_stack int +setup_items_for_insert(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_path *path, + struct btrfs_key *cpu_key, u32 *data_size, + u32 total_data, u32 total_size, int nr) { - struct extent_buffer *leaf; struct btrfs_item *item; - int ret = 0; - int slot; - int slot_orig; int i; u32 nritems; - u32 total_size = 0; - u32 total_data = 0; unsigned int data_end; struct btrfs_disk_key disk_key; + int ret; + struct extent_buffer *leaf; + int slot; - for (i = 0; i < nr; i++) { - total_data += data_size[i]; - } - - /* create a root if there isn't one */ - if (!root->node) - BUG(); - - total_size = total_data + (nr - 1) * sizeof(struct btrfs_item); - ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1); - if (ret == 0) { - return -EEXIST; - } - if (ret < 0) - goto out; - - slot_orig = path->slots[0]; leaf = path->nodes[0]; + slot = path->slots[0]; nritems = btrfs_header_nritems(leaf); data_end = leaf_data_end(root, leaf); - if (btrfs_leaf_free_space(root, leaf) < - sizeof(struct btrfs_item) + total_size) { + if (btrfs_leaf_free_space(root, leaf) < total_size) { btrfs_print_leaf(root, leaf); - printk("not enough freespace need %u have %d\n", + printk(KERN_CRIT "not enough freespace need %u have %d\n", total_size, btrfs_leaf_free_space(root, leaf)); BUG(); } - slot = path->slots[0]; - BUG_ON(slot < 0); - if (slot != nritems) { - int i; unsigned int old_data = btrfs_item_end_nr(leaf, slot); if (old_data < data_end) { btrfs_print_leaf(root, leaf); - printk("slot %d old_data %d data_end %d\n", + printk(KERN_CRIT "slot %d old_data %d data_end %d\n", slot, old_data, data_end); BUG_ON(1); } @@ -2463,19 +3667,58 @@ int btrfs_insert_empty_items(struct btrfs_trans_handle *trans, data_end -= data_size[i]; btrfs_set_item_size(leaf, item, data_size[i]); } + btrfs_set_header_nritems(leaf, nritems + nr); - btrfs_mark_buffer_dirty(leaf); ret = 0; if (slot == 0) { + struct btrfs_disk_key disk_key; btrfs_cpu_key_to_disk(&disk_key, cpu_key); ret = fixup_low_keys(trans, root, path, &disk_key, 1); } + btrfs_unlock_up_safe(path, 1); + btrfs_mark_buffer_dirty(leaf); if (btrfs_leaf_free_space(root, leaf) < 0) { btrfs_print_leaf(root, leaf); BUG(); } + return ret; +} + +/* + * Given a key and some data, insert items into the tree. + * This does all the path init required, making room in the tree if needed. + */ +int btrfs_insert_empty_items(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *cpu_key, u32 *data_size, + int nr) +{ + struct extent_buffer *leaf; + int ret = 0; + int slot; + int i; + u32 total_size = 0; + u32 total_data = 0; + + for (i = 0; i < nr; i++) + total_data += data_size[i]; + + total_size = total_data + (nr * sizeof(struct btrfs_item)); + ret = btrfs_search_slot(trans, root, cpu_key, path, total_size, 1); + if (ret == 0) + return -EEXIST; + if (ret < 0) + goto out; + + leaf = path->nodes[0]; + slot = path->slots[0]; + BUG_ON(slot < 0); + + ret = setup_items_for_insert(trans, root, path, cpu_key, data_size, + total_data, total_size, nr); out: return ret; @@ -2510,9 +3753,8 @@ int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root /* * delete the pointer from a given node. * - * If the delete empties a node, the node is removed from the tree, - * continuing all the way the root if required. The root is converted into - * a leaf if all the nodes are emptied. + * the tree should have been previously balanced so the deletion does not + * empty a node. */ static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, int level, int slot) @@ -2523,7 +3765,7 @@ static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root, int wret; nritems = btrfs_header_nritems(parent); - if (slot != nritems -1) { + if (slot != nritems - 1) { memmove_extent_buffer(parent, btrfs_node_key_ptr_offset(slot), btrfs_node_key_ptr_offset(slot + 1), @@ -2549,6 +3791,39 @@ static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root, } /* + * a helper function to delete the leaf pointed to by path->slots[1] and + * path->nodes[1]. + * + * This deletes the pointer in path->nodes[1] and frees the leaf + * block extent. zero is returned if it all worked out, < 0 otherwise. + * + * The path must have already been setup for deleting the leaf, including + * all the proper balancing. path->nodes[1] must be locked. + */ +static noinline int btrfs_del_leaf(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_path *path, + struct extent_buffer *leaf) +{ + int ret; + + WARN_ON(btrfs_header_generation(leaf) != trans->transid); + ret = del_ptr(trans, root, path, 1, path->slots[1]); + if (ret) + return ret; + + /* + * btrfs_free_extent is expensive, we want to make sure we + * aren't holding any locks when we call it + */ + btrfs_unlock_up_safe(path, 0); + + root_sub_used(root, leaf->len); + + btrfs_free_tree_block(trans, root, leaf, 0, 1); + return 0; +} +/* * delete the item at the leaf level in path. If that empties * the leaf, remove it from the tree */ @@ -2573,7 +3848,6 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, nritems = btrfs_header_nritems(leaf); if (slot + nr != nritems) { - int i; int data_end = leaf_data_end(root, leaf); memmove_extent_buffer(leaf, btrfs_leaf_data(leaf) + @@ -2614,18 +3888,10 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, if (leaf == root->node) { btrfs_set_header_level(leaf, 0); } else { - u64 root_gen = btrfs_header_generation(path->nodes[1]); + btrfs_set_path_blocking(path); clean_tree_block(trans, root, leaf); - wait_on_tree_block_writeback(root, leaf); - wret = del_ptr(trans, root, path, 1, path->slots[1]); - if (wret) - ret = wret; - wret = btrfs_free_extent(trans, root, - leaf->start, leaf->len, - btrfs_header_owner(path->nodes[1]), - root_gen, 0, 0, 1); - if (wret) - ret = wret; + ret = btrfs_del_leaf(trans, root, path, leaf); + BUG_ON(ret); } } else { int used = leaf_space_used(leaf, 0, nritems); @@ -2640,7 +3906,7 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, } /* delete the leaf if it is mostly empty */ - if (used < BTRFS_LEAF_DATA_SIZE(root) / 4) { + if (used < BTRFS_LEAF_DATA_SIZE(root) / 3) { /* push_leaf_left fixes the path. * make sure the path still points to our leaf * for possible call to del_ptr below @@ -2648,6 +3914,7 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, slot = path->slots[1]; extent_buffer_get(leaf); + btrfs_set_path_blocking(path); wret = push_leaf_left(trans, root, path, 1, 1); if (wret < 0 && wret != -ENOSPC) ret = wret; @@ -2660,29 +3927,18 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, } if (btrfs_header_nritems(leaf) == 0) { - u64 root_gen; - u64 bytenr = leaf->start; - u32 blocksize = leaf->len; - - root_gen = btrfs_header_generation( - path->nodes[1]); - - clean_tree_block(trans, root, leaf); - wait_on_tree_block_writeback(root, leaf); - - wret = del_ptr(trans, root, path, 1, slot); - if (wret) - ret = wret; - + path->slots[1] = slot; + ret = btrfs_del_leaf(trans, root, path, leaf); + BUG_ON(ret); free_extent_buffer(leaf); - wret = btrfs_free_extent(trans, root, bytenr, - blocksize, - btrfs_header_owner(path->nodes[1]), - root_gen, 0, 0, 1); - if (wret) - ret = wret; } else { - btrfs_mark_buffer_dirty(leaf); + /* if we're still in the path, make sure + * we're dirty. Otherwise, one of the + * push_leaf functions must have already + * dirtied this buffer + */ + if (path->nodes[0] == leaf) + btrfs_mark_buffer_dirty(leaf); free_extent_buffer(leaf); } } else { @@ -2693,122 +3949,449 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, } /* - * walk up the tree as far as required to find the previous leaf. + * search the tree again to find a leaf with lesser keys * returns 0 if it found something or 1 if there are no lesser leaves. * returns < 0 on io errors. + * + * This may release the path, and so you may lose any locks held at the + * time you call it. */ int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path) { - u64 bytenr; + struct btrfs_key key; + struct btrfs_disk_key found_key; + int ret; + + btrfs_item_key_to_cpu(path->nodes[0], &key, 0); + + if (key.offset > 0) + key.offset--; + else if (key.type > 0) + key.type--; + else if (key.objectid > 0) + key.objectid--; + else + return 1; + + btrfs_release_path(root, path); + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) + return ret; + btrfs_item_key(path->nodes[0], &found_key, 0); + ret = comp_keys(&found_key, &key); + if (ret < 0) + return 0; + return 1; +} + +/* + * A helper function to walk down the tree starting at min_key, and looking + * for nodes or leaves that are either in cache or have a minimum + * transaction id. This is used by the btree defrag code, and tree logging + * + * This does not cow, but it does stuff the starting key it finds back + * into min_key, so you can call btrfs_search_slot with cow=1 on the + * key and get a writable path. + * + * This does lock as it descends, and path->keep_locks should be set + * to 1 by the caller. + * + * This honors path->lowest_level to prevent descent past a given level + * of the tree. + * + * min_trans indicates the oldest transaction that you are interested + * in walking through. Any nodes or leaves older than min_trans are + * skipped over (without reading them). + * + * returns zero if something useful was found, < 0 on error and 1 if there + * was nothing in the tree that matched the search criteria. + */ +int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key, + struct btrfs_key *max_key, + struct btrfs_path *path, int cache_only, + u64 min_trans) +{ + struct extent_buffer *cur; + struct btrfs_key found_key; + int slot; + int sret; + u32 nritems; + int level; + int ret = 1; + + WARN_ON(!path->keep_locks); +again: + cur = btrfs_lock_root_node(root); + level = btrfs_header_level(cur); + WARN_ON(path->nodes[level]); + path->nodes[level] = cur; + path->locks[level] = 1; + + if (btrfs_header_generation(cur) < min_trans) { + ret = 1; + goto out; + } + while (1) { + nritems = btrfs_header_nritems(cur); + level = btrfs_header_level(cur); + sret = bin_search(cur, min_key, level, &slot); + + /* at the lowest level, we're done, setup the path and exit */ + if (level == path->lowest_level) { + if (slot >= nritems) + goto find_next_key; + ret = 0; + path->slots[level] = slot; + btrfs_item_key_to_cpu(cur, &found_key, slot); + goto out; + } + if (sret && slot > 0) + slot--; + /* + * check this node pointer against the cache_only and + * min_trans parameters. If it isn't in cache or is too + * old, skip to the next one. + */ + while (slot < nritems) { + u64 blockptr; + u64 gen; + struct extent_buffer *tmp; + struct btrfs_disk_key disk_key; + + blockptr = btrfs_node_blockptr(cur, slot); + gen = btrfs_node_ptr_generation(cur, slot); + if (gen < min_trans) { + slot++; + continue; + } + if (!cache_only) + break; + + if (max_key) { + btrfs_node_key(cur, &disk_key, slot); + if (comp_keys(&disk_key, max_key) >= 0) { + ret = 1; + goto out; + } + } + + tmp = btrfs_find_tree_block(root, blockptr, + btrfs_level_size(root, level - 1)); + + if (tmp && btrfs_buffer_uptodate(tmp, gen)) { + free_extent_buffer(tmp); + break; + } + if (tmp) + free_extent_buffer(tmp); + slot++; + } +find_next_key: + /* + * we didn't find a candidate key in this node, walk forward + * and find another one + */ + if (slot >= nritems) { + path->slots[level] = slot; + btrfs_set_path_blocking(path); + sret = btrfs_find_next_key(root, path, min_key, level, + cache_only, min_trans); + if (sret == 0) { + btrfs_release_path(root, path); + goto again; + } else { + goto out; + } + } + /* save our key for returning back */ + btrfs_node_key_to_cpu(cur, &found_key, slot); + path->slots[level] = slot; + if (level == path->lowest_level) { + ret = 0; + unlock_up(path, level, 1); + goto out; + } + btrfs_set_path_blocking(path); + cur = read_node_slot(root, cur, slot); + + btrfs_tree_lock(cur); + + path->locks[level - 1] = 1; + path->nodes[level - 1] = cur; + unlock_up(path, level, 1); + btrfs_clear_path_blocking(path, NULL); + } +out: + if (ret == 0) + memcpy(min_key, &found_key, sizeof(found_key)); + btrfs_set_path_blocking(path); + return ret; +} + +/* + * this is similar to btrfs_next_leaf, but does not try to preserve + * and fixup the path. It looks for and returns the next key in the + * tree based on the current path and the cache_only and min_trans + * parameters. + * + * 0 is returned if another key is found, < 0 if there are any errors + * and 1 is returned if there are no higher keys in the tree + * + * path->keep_locks should be set to 1 on the search made before + * calling this function. + */ +int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path, + struct btrfs_key *key, int level, + int cache_only, u64 min_trans) +{ int slot; - int level = 1; struct extent_buffer *c; - struct extent_buffer *next = NULL; - while(level < BTRFS_MAX_LEVEL) { + WARN_ON(!path->keep_locks); + while (level < BTRFS_MAX_LEVEL) { if (!path->nodes[level]) return 1; - slot = path->slots[level]; + slot = path->slots[level] + 1; c = path->nodes[level]; - if (slot == 0) { - level++; - if (level == BTRFS_MAX_LEVEL) +next: + if (slot >= btrfs_header_nritems(c)) { + int ret; + int orig_lowest; + struct btrfs_key cur_key; + if (level + 1 >= BTRFS_MAX_LEVEL || + !path->nodes[level + 1]) return 1; - continue; + + if (path->locks[level + 1]) { + level++; + continue; + } + + slot = btrfs_header_nritems(c) - 1; + if (level == 0) + btrfs_item_key_to_cpu(c, &cur_key, slot); + else + btrfs_node_key_to_cpu(c, &cur_key, slot); + + orig_lowest = path->lowest_level; + btrfs_release_path(root, path); + path->lowest_level = level; + ret = btrfs_search_slot(NULL, root, &cur_key, path, + 0, 0); + path->lowest_level = orig_lowest; + if (ret < 0) + return ret; + + c = path->nodes[level]; + slot = path->slots[level]; + if (ret == 0) + slot++; + goto next; } - slot--; - bytenr = btrfs_node_blockptr(c, slot); - if (next) - free_extent_buffer(next); + if (level == 0) + btrfs_item_key_to_cpu(c, key, slot); + else { + u64 blockptr = btrfs_node_blockptr(c, slot); + u64 gen = btrfs_node_ptr_generation(c, slot); - next = read_tree_block(root, bytenr, - btrfs_level_size(root, level - 1)); - break; - } - path->slots[level] = slot; - while(1) { - level--; - c = path->nodes[level]; - free_extent_buffer(c); - slot = btrfs_header_nritems(next); - if (slot != 0) - slot--; - path->nodes[level] = next; - path->slots[level] = slot; - if (!level) - break; - next = read_tree_block(root, btrfs_node_blockptr(next, slot), - btrfs_level_size(root, level - 1)); + if (cache_only) { + struct extent_buffer *cur; + cur = btrfs_find_tree_block(root, blockptr, + btrfs_level_size(root, level - 1)); + if (!cur || !btrfs_buffer_uptodate(cur, gen)) { + slot++; + if (cur) + free_extent_buffer(cur); + goto next; + } + free_extent_buffer(cur); + } + if (gen < min_trans) { + slot++; + goto next; + } + btrfs_node_key_to_cpu(c, key, slot); + } + return 0; } - return 0; + return 1; } /* - * walk up the tree as far as required to find the next leaf. + * search the tree again to find a leaf with greater keys * returns 0 if it found something or 1 if there are no greater leaves. * returns < 0 on io errors. */ int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path) { int slot; - int level = 1; - u64 bytenr; + int level; struct extent_buffer *c; - struct extent_buffer *next = NULL; + struct extent_buffer *next; + struct btrfs_key key; + u32 nritems; + int ret; + int old_spinning = path->leave_spinning; + int force_blocking = 0; - while(level < BTRFS_MAX_LEVEL) { - if (!path->nodes[level]) - return 1; + nritems = btrfs_header_nritems(path->nodes[0]); + if (nritems == 0) + return 1; + + /* + * we take the blocks in an order that upsets lockdep. Using + * blocking mode is the only way around it. + */ +#ifdef CONFIG_DEBUG_LOCK_ALLOC + force_blocking = 1; +#endif + + btrfs_item_key_to_cpu(path->nodes[0], &key, nritems - 1); +again: + level = 1; + next = NULL; + btrfs_release_path(root, path); + + path->keep_locks = 1; + + if (!force_blocking) + path->leave_spinning = 1; + + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + path->keep_locks = 0; + + if (ret < 0) + return ret; + + nritems = btrfs_header_nritems(path->nodes[0]); + /* + * by releasing the path above we dropped all our locks. A balance + * could have added more items next to the key that used to be + * at the very end of the block. So, check again here and + * advance the path if there are now more items available. + */ + if (nritems > 0 && path->slots[0] < nritems - 1) { + if (ret == 0) + path->slots[0]++; + ret = 0; + goto done; + } + + while (level < BTRFS_MAX_LEVEL) { + if (!path->nodes[level]) { + ret = 1; + goto done; + } slot = path->slots[level] + 1; c = path->nodes[level]; if (slot >= btrfs_header_nritems(c)) { level++; - if (level == BTRFS_MAX_LEVEL) - return 1; + if (level == BTRFS_MAX_LEVEL) { + ret = 1; + goto done; + } continue; } - bytenr = btrfs_node_blockptr(c, slot); - if (next) + if (next) { + btrfs_tree_unlock(next); free_extent_buffer(next); + } - if (path->reada) - reada_for_search(root, path, level, slot, 0); + next = c; + ret = read_block_for_search(NULL, root, path, &next, level, + slot, &key); + if (ret == -EAGAIN) + goto again; + + if (ret < 0) { + btrfs_release_path(root, path); + goto done; + } - next = read_tree_block(root, bytenr, - btrfs_level_size(root, level -1)); + if (!path->skip_locking) { + ret = btrfs_try_spin_lock(next); + if (!ret) { + btrfs_set_path_blocking(path); + btrfs_tree_lock(next); + if (!force_blocking) + btrfs_clear_path_blocking(path, next); + } + if (force_blocking) + btrfs_set_lock_blocking(next); + } break; } path->slots[level] = slot; - while(1) { + while (1) { level--; c = path->nodes[level]; + if (path->locks[level]) + btrfs_tree_unlock(c); + free_extent_buffer(c); path->nodes[level] = next; path->slots[level] = 0; + if (!path->skip_locking) + path->locks[level] = 1; + if (!level) break; - if (path->reada) - reada_for_search(root, path, level, 0, 0); - next = read_tree_block(root, btrfs_node_blockptr(next, 0), - btrfs_level_size(root, level - 1)); + + ret = read_block_for_search(NULL, root, path, &next, level, + 0, &key); + if (ret == -EAGAIN) + goto again; + + if (ret < 0) { + btrfs_release_path(root, path); + goto done; + } + + if (!path->skip_locking) { + btrfs_assert_tree_locked(path->nodes[level]); + ret = btrfs_try_spin_lock(next); + if (!ret) { + btrfs_set_path_blocking(path); + btrfs_tree_lock(next); + if (!force_blocking) + btrfs_clear_path_blocking(path, next); + } + if (force_blocking) + btrfs_set_lock_blocking(next); + } } - return 0; + ret = 0; +done: + unlock_up(path, 0, 1); + path->leave_spinning = old_spinning; + if (!old_spinning) + btrfs_set_path_blocking(path); + + return ret; } +/* + * this uses btrfs_prev_leaf to walk backwards in the tree, and keeps + * searching until it gets past min_objectid or finds an item of 'type' + * + * returns 0 if something is found, 1 if nothing was found and < 0 on error + */ int btrfs_previous_item(struct btrfs_root *root, struct btrfs_path *path, u64 min_objectid, int type) { struct btrfs_key found_key; struct extent_buffer *leaf; + u32 nritems; int ret; - while(1) { + while (1) { if (path->slots[0] == 0) { + btrfs_set_path_blocking(path); ret = btrfs_prev_leaf(root, path); if (ret != 0) return ret; @@ -2816,10 +4399,20 @@ int btrfs_previous_item(struct btrfs_root *root, path->slots[0]--; } leaf = path->nodes[0]; + nritems = btrfs_header_nritems(leaf); + if (nritems == 0) + return 1; + if (path->slots[0] == nritems) + path->slots[0]--; + btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); + if (found_key.objectid < min_objectid) + break; if (found_key.type == type) return 0; + if (found_key.objectid == min_objectid && + found_key.type < type) + break; } return 1; } -