From e679376911d016b670c8cfc1645c178f77e8d1d3 Mon Sep 17 00:00:00 2001 From: Arne Jansen Date: Tue, 13 Sep 2011 11:18:10 +0200 Subject: Btrfs: add helper for tree enumeration Often no exact match is wanted but just the next lower or higher item. There's a lot of duplicated code throughout btrfs to deal with the corner cases. This patch adds a helper function that can facilitate searching. Signed-off-by: Arne Jansen --- fs/btrfs/ctree.c | 74 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 74 insertions(+) (limited to 'fs/btrfs/ctree.c') diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 8206b390058..c82a9e4a953 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -2721,6 +2721,80 @@ done: return ret; } +/* + * helper to use instead of search slot if no exact match is needed but + * instead the next or previous item should be returned. + * When find_higher is true, the next higher item is returned, the next lower + * otherwise. + * When return_any and find_higher are both true, and no higher item is found, + * return the next lower instead. + * When return_any is true and find_higher is false, and no lower item is found, + * return the next higher instead. + * It returns 0 if any item is found, 1 if none is found (tree empty), and + * < 0 on error + */ +int btrfs_search_slot_for_read(struct btrfs_root *root, + struct btrfs_key *key, struct btrfs_path *p, + int find_higher, int return_any) +{ + int ret; + struct extent_buffer *leaf; + +again: + ret = btrfs_search_slot(NULL, root, key, p, 0, 0); + if (ret <= 0) + return ret; + /* + * a return value of 1 means the path is at the position where the + * item should be inserted. Normally this is the next bigger item, + * but in case the previous item is the last in a leaf, path points + * to the first free slot in the previous leaf, i.e. at an invalid + * item. + */ + leaf = p->nodes[0]; + + if (find_higher) { + if (p->slots[0] >= btrfs_header_nritems(leaf)) { + ret = btrfs_next_leaf(root, p); + if (ret <= 0) + return ret; + if (!return_any) + return 1; + /* + * no higher item found, return the next + * lower instead + */ + return_any = 0; + find_higher = 0; + btrfs_release_path(p); + goto again; + } + } else { + if (p->slots[0] == 0) { + ret = btrfs_prev_leaf(root, p); + if (ret < 0) + return ret; + if (!ret) { + p->slots[0] = btrfs_header_nritems(leaf) - 1; + return 0; + } + if (!return_any) + return 1; + /* + * no lower item found, return the next + * higher instead + */ + return_any = 0; + find_higher = 1; + btrfs_release_path(p); + goto again; + } else { + --p->slots[0]; + } + } + return 0; +} + /* * adjust the pointers going up the tree, starting at level * making sure the right key of each node is points to 'key'. -- cgit v1.2.3-18-g5258 From 7069830a9e381e33d44ded45095f764844c71d24 Mon Sep 17 00:00:00 2001 From: Alexander Block Date: Tue, 5 Jun 2012 21:07:48 +0200 Subject: Btrfs: add btrfs_compare_trees function This function is used to find the differences between two trees. The tree compare skips whole subtrees if it detects shared tree blocks and thus is pretty fast. Signed-off-by: Alexander Block Reviewed-by: David Sterba Reviewed-by: Arne Jansen Reviewed-by: Jan Schmidt Reviewed-by: Alex Lyakas --- fs/btrfs/ctree.c | 425 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 425 insertions(+) (limited to 'fs/btrfs/ctree.c') diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index c82a9e4a953..4c10fd19d48 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -5005,6 +5005,431 @@ out: return ret; } +static void tree_move_down(struct btrfs_root *root, + struct btrfs_path *path, + int *level, int root_level) +{ + path->nodes[*level - 1] = read_node_slot(root, path->nodes[*level], + path->slots[*level]); + path->slots[*level - 1] = 0; + (*level)--; +} + +static int tree_move_next_or_upnext(struct btrfs_root *root, + struct btrfs_path *path, + int *level, int root_level) +{ + int ret = 0; + int nritems; + nritems = btrfs_header_nritems(path->nodes[*level]); + + path->slots[*level]++; + + while (path->slots[*level] == nritems) { + if (*level == root_level) + return -1; + + /* move upnext */ + path->slots[*level] = 0; + free_extent_buffer(path->nodes[*level]); + path->nodes[*level] = NULL; + (*level)++; + path->slots[*level]++; + + nritems = btrfs_header_nritems(path->nodes[*level]); + ret = 1; + } + return ret; +} + +/* + * Returns 1 if it had to move up and next. 0 is returned if it moved only next + * or down. + */ +static int tree_advance(struct btrfs_root *root, + struct btrfs_path *path, + int *level, int root_level, + int allow_down, + struct btrfs_key *key) +{ + int ret; + + if (*level == 0 || !allow_down) { + ret = tree_move_next_or_upnext(root, path, level, root_level); + } else { + tree_move_down(root, path, level, root_level); + ret = 0; + } + if (ret >= 0) { + if (*level == 0) + btrfs_item_key_to_cpu(path->nodes[*level], key, + path->slots[*level]); + else + btrfs_node_key_to_cpu(path->nodes[*level], key, + path->slots[*level]); + } + return ret; +} + +static int tree_compare_item(struct btrfs_root *left_root, + struct btrfs_path *left_path, + struct btrfs_path *right_path, + char *tmp_buf) +{ + int cmp; + int len1, len2; + unsigned long off1, off2; + + len1 = btrfs_item_size_nr(left_path->nodes[0], left_path->slots[0]); + len2 = btrfs_item_size_nr(right_path->nodes[0], right_path->slots[0]); + if (len1 != len2) + return 1; + + off1 = btrfs_item_ptr_offset(left_path->nodes[0], left_path->slots[0]); + off2 = btrfs_item_ptr_offset(right_path->nodes[0], + right_path->slots[0]); + + read_extent_buffer(left_path->nodes[0], tmp_buf, off1, len1); + + cmp = memcmp_extent_buffer(right_path->nodes[0], tmp_buf, off2, len1); + if (cmp) + return 1; + return 0; +} + +#define ADVANCE 1 +#define ADVANCE_ONLY_NEXT -1 + +/* + * This function compares two trees and calls the provided callback for + * every changed/new/deleted item it finds. + * If shared tree blocks are encountered, whole subtrees are skipped, making + * the compare pretty fast on snapshotted subvolumes. + * + * This currently works on commit roots only. As commit roots are read only, + * we don't do any locking. The commit roots are protected with transactions. + * Transactions are ended and rejoined when a commit is tried in between. + * + * This function checks for modifications done to the trees while comparing. + * If it detects a change, it aborts immediately. + */ +int btrfs_compare_trees(struct btrfs_root *left_root, + struct btrfs_root *right_root, + btrfs_changed_cb_t changed_cb, void *ctx) +{ + int ret; + int cmp; + struct btrfs_trans_handle *trans = NULL; + struct btrfs_path *left_path = NULL; + struct btrfs_path *right_path = NULL; + struct btrfs_key left_key; + struct btrfs_key right_key; + char *tmp_buf = NULL; + int left_root_level; + int right_root_level; + int left_level; + int right_level; + int left_end_reached; + int right_end_reached; + int advance_left; + int advance_right; + u64 left_blockptr; + u64 right_blockptr; + u64 left_start_ctransid; + u64 right_start_ctransid; + u64 ctransid; + + left_path = btrfs_alloc_path(); + if (!left_path) { + ret = -ENOMEM; + goto out; + } + right_path = btrfs_alloc_path(); + if (!right_path) { + ret = -ENOMEM; + goto out; + } + + tmp_buf = kmalloc(left_root->leafsize, GFP_NOFS); + if (!tmp_buf) { + ret = -ENOMEM; + goto out; + } + + left_path->search_commit_root = 1; + left_path->skip_locking = 1; + right_path->search_commit_root = 1; + right_path->skip_locking = 1; + + spin_lock(&left_root->root_times_lock); + left_start_ctransid = btrfs_root_ctransid(&left_root->root_item); + spin_unlock(&left_root->root_times_lock); + + spin_lock(&right_root->root_times_lock); + right_start_ctransid = btrfs_root_ctransid(&right_root->root_item); + spin_unlock(&right_root->root_times_lock); + + trans = btrfs_join_transaction(left_root); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + trans = NULL; + goto out; + } + + /* + * Strategy: Go to the first items of both trees. Then do + * + * If both trees are at level 0 + * Compare keys of current items + * If left < right treat left item as new, advance left tree + * and repeat + * If left > right treat right item as deleted, advance right tree + * and repeat + * If left == right do deep compare of items, treat as changed if + * needed, advance both trees and repeat + * If both trees are at the same level but not at level 0 + * Compare keys of current nodes/leafs + * If left < right advance left tree and repeat + * If left > right advance right tree and repeat + * If left == right compare blockptrs of the next nodes/leafs + * If they match advance both trees but stay at the same level + * and repeat + * If they don't match advance both trees while allowing to go + * deeper and repeat + * If tree levels are different + * Advance the tree that needs it and repeat + * + * Advancing a tree means: + * If we are at level 0, try to go to the next slot. If that's not + * possible, go one level up and repeat. Stop when we found a level + * where we could go to the next slot. We may at this point be on a + * node or a leaf. + * + * If we are not at level 0 and not on shared tree blocks, go one + * level deeper. + * + * If we are not at level 0 and on shared tree blocks, go one slot to + * the right if possible or go up and right. + */ + + left_level = btrfs_header_level(left_root->commit_root); + left_root_level = left_level; + left_path->nodes[left_level] = left_root->commit_root; + extent_buffer_get(left_path->nodes[left_level]); + + right_level = btrfs_header_level(right_root->commit_root); + right_root_level = right_level; + right_path->nodes[right_level] = right_root->commit_root; + extent_buffer_get(right_path->nodes[right_level]); + + if (left_level == 0) + btrfs_item_key_to_cpu(left_path->nodes[left_level], + &left_key, left_path->slots[left_level]); + else + btrfs_node_key_to_cpu(left_path->nodes[left_level], + &left_key, left_path->slots[left_level]); + if (right_level == 0) + btrfs_item_key_to_cpu(right_path->nodes[right_level], + &right_key, right_path->slots[right_level]); + else + btrfs_node_key_to_cpu(right_path->nodes[right_level], + &right_key, right_path->slots[right_level]); + + left_end_reached = right_end_reached = 0; + advance_left = advance_right = 0; + + while (1) { + /* + * We need to make sure the transaction does not get committed + * while we do anything on commit roots. This means, we need to + * join and leave transactions for every item that we process. + */ + if (trans && btrfs_should_end_transaction(trans, left_root)) { + btrfs_release_path(left_path); + btrfs_release_path(right_path); + + ret = btrfs_end_transaction(trans, left_root); + trans = NULL; + if (ret < 0) + goto out; + } + /* now rejoin the transaction */ + if (!trans) { + trans = btrfs_join_transaction(left_root); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + trans = NULL; + goto out; + } + + spin_lock(&left_root->root_times_lock); + ctransid = btrfs_root_ctransid(&left_root->root_item); + spin_unlock(&left_root->root_times_lock); + if (ctransid != left_start_ctransid) + left_start_ctransid = 0; + + spin_lock(&right_root->root_times_lock); + ctransid = btrfs_root_ctransid(&right_root->root_item); + spin_unlock(&right_root->root_times_lock); + if (ctransid != right_start_ctransid) + right_start_ctransid = 0; + + if (!left_start_ctransid || !right_start_ctransid) { + WARN(1, KERN_WARNING + "btrfs: btrfs_compare_tree detected " + "a change in one of the trees while " + "iterating. This is probably a " + "bug.\n"); + ret = -EIO; + goto out; + } + + /* + * the commit root may have changed, so start again + * where we stopped + */ + left_path->lowest_level = left_level; + right_path->lowest_level = right_level; + ret = btrfs_search_slot(NULL, left_root, + &left_key, left_path, 0, 0); + if (ret < 0) + goto out; + ret = btrfs_search_slot(NULL, right_root, + &right_key, right_path, 0, 0); + if (ret < 0) + goto out; + } + + if (advance_left && !left_end_reached) { + ret = tree_advance(left_root, left_path, &left_level, + left_root_level, + advance_left != ADVANCE_ONLY_NEXT, + &left_key); + if (ret < 0) + left_end_reached = ADVANCE; + advance_left = 0; + } + if (advance_right && !right_end_reached) { + ret = tree_advance(right_root, right_path, &right_level, + right_root_level, + advance_right != ADVANCE_ONLY_NEXT, + &right_key); + if (ret < 0) + right_end_reached = ADVANCE; + advance_right = 0; + } + + if (left_end_reached && right_end_reached) { + ret = 0; + goto out; + } else if (left_end_reached) { + if (right_level == 0) { + ret = changed_cb(left_root, right_root, + left_path, right_path, + &right_key, + BTRFS_COMPARE_TREE_DELETED, + ctx); + if (ret < 0) + goto out; + } + advance_right = ADVANCE; + continue; + } else if (right_end_reached) { + if (left_level == 0) { + ret = changed_cb(left_root, right_root, + left_path, right_path, + &left_key, + BTRFS_COMPARE_TREE_NEW, + ctx); + if (ret < 0) + goto out; + } + advance_left = ADVANCE; + continue; + } + + if (left_level == 0 && right_level == 0) { + cmp = btrfs_comp_cpu_keys(&left_key, &right_key); + if (cmp < 0) { + ret = changed_cb(left_root, right_root, + left_path, right_path, + &left_key, + BTRFS_COMPARE_TREE_NEW, + ctx); + if (ret < 0) + goto out; + advance_left = ADVANCE; + } else if (cmp > 0) { + ret = changed_cb(left_root, right_root, + left_path, right_path, + &right_key, + BTRFS_COMPARE_TREE_DELETED, + ctx); + if (ret < 0) + goto out; + advance_right = ADVANCE; + } else { + ret = tree_compare_item(left_root, left_path, + right_path, tmp_buf); + if (ret) { + ret = changed_cb(left_root, right_root, + left_path, right_path, + &left_key, + BTRFS_COMPARE_TREE_CHANGED, + ctx); + if (ret < 0) + goto out; + } + advance_left = ADVANCE; + advance_right = ADVANCE; + } + } else if (left_level == right_level) { + cmp = btrfs_comp_cpu_keys(&left_key, &right_key); + if (cmp < 0) { + advance_left = ADVANCE; + } else if (cmp > 0) { + advance_right = ADVANCE; + } else { + left_blockptr = btrfs_node_blockptr( + left_path->nodes[left_level], + left_path->slots[left_level]); + right_blockptr = btrfs_node_blockptr( + right_path->nodes[right_level], + right_path->slots[right_level]); + if (left_blockptr == right_blockptr) { + /* + * As we're on a shared block, don't + * allow to go deeper. + */ + advance_left = ADVANCE_ONLY_NEXT; + advance_right = ADVANCE_ONLY_NEXT; + } else { + advance_left = ADVANCE; + advance_right = ADVANCE; + } + } + } else if (left_level < right_level) { + advance_right = ADVANCE; + } else { + advance_left = ADVANCE; + } + } + +out: + btrfs_free_path(left_path); + btrfs_free_path(right_path); + kfree(tmp_buf); + + if (trans) { + if (!ret) + ret = btrfs_end_transaction(trans, left_root); + else + btrfs_end_transaction(trans, left_root); + } + + 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 -- cgit v1.2.3-18-g5258