1 files changed, 316 insertions, 249 deletions

diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index 04edf69be87..e25564bfcb4 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -36,16 +36,23 @@ static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb,
 				u64 extent_item_pos,
 				struct extent_inode_elem **eie)
 {
-	u64 data_offset;
-	u64 data_len;
+	u64 offset = 0;
 	struct extent_inode_elem *e;
 
-	data_offset = btrfs_file_extent_offset(eb, fi);
-	data_len = btrfs_file_extent_num_bytes(eb, fi);
+	if (!btrfs_file_extent_compression(eb, fi) &&
+	    !btrfs_file_extent_encryption(eb, fi) &&
+	    !btrfs_file_extent_other_encoding(eb, fi)) {
+		u64 data_offset;
+		u64 data_len;
 
-	if (extent_item_pos < data_offset ||
-	    extent_item_pos >= data_offset + data_len)
-		return 1;
+		data_offset = btrfs_file_extent_offset(eb, fi);
+		data_len = btrfs_file_extent_num_bytes(eb, fi);
+
+		if (extent_item_pos < data_offset ||
+		    extent_item_pos >= data_offset + data_len)
+			return 1;
+		offset = extent_item_pos - data_offset;
+	}
 
 	e = kmalloc(sizeof(*e), GFP_NOFS);
 	if (!e)
@@ -53,12 +60,22 @@ static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb,
 
 	e->next = *eie;
 	e->inum = key->objectid;
-	e->offset = key->offset + (extent_item_pos - data_offset);
+	e->offset = key->offset + offset;
 	*eie = e;
 
 	return 0;
 }
 
+static void free_inode_elem_list(struct extent_inode_elem *eie)
+{
+	struct extent_inode_elem *eie_next;
+
+	for (; eie; eie = eie_next) {
+		eie_next = eie->next;
+		kfree(eie);
+	}
+}
+
 static int find_extent_in_eb(struct extent_buffer *eb, u64 wanted_disk_byte,
 				u64 extent_item_pos,
 				struct extent_inode_elem **eie)
@@ -112,6 +129,26 @@ struct __prelim_ref {
 	u64 wanted_disk_byte;
 };
 
+static struct kmem_cache *btrfs_prelim_ref_cache;
+
+int __init btrfs_prelim_ref_init(void)
+{
+	btrfs_prelim_ref_cache = kmem_cache_create("btrfs_prelim_ref",
+					sizeof(struct __prelim_ref),
+					0,
+					SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
+					NULL);
+	if (!btrfs_prelim_ref_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void btrfs_prelim_ref_exit(void)
+{
+	if (btrfs_prelim_ref_cache)
+		kmem_cache_destroy(btrfs_prelim_ref_cache);
+}
+
 /*
  * the rules for all callers of this function are:
  * - obtaining the parent is the goal
@@ -153,12 +190,15 @@ struct __prelim_ref {
 
 static int __add_prelim_ref(struct list_head *head, u64 root_id,
 			    struct btrfs_key *key, int level,
-			    u64 parent, u64 wanted_disk_byte, int count)
+			    u64 parent, u64 wanted_disk_byte, int count,
+			    gfp_t gfp_mask)
 {
 	struct __prelim_ref *ref;
 
-	/* in case we're adding delayed refs, we're holding the refs spinlock */
-	ref = kmalloc(sizeof(*ref), GFP_ATOMIC);
+	if (root_id == BTRFS_DATA_RELOC_TREE_OBJECTID)
+		return 0;
+
+	ref = kmem_cache_alloc(btrfs_prelim_ref_cache, gfp_mask);
 	if (!ref)
 		return -ENOMEM;
 
@@ -179,18 +219,20 @@ static int __add_prelim_ref(struct list_head *head, u64 root_id,
 }
 
 static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
-				struct ulist *parents, int level,
-				struct btrfs_key *key_for_search, u64 time_seq,
-				u64 wanted_disk_byte,
-				const u64 *extent_item_pos)
+			   struct ulist *parents, struct __prelim_ref *ref,
+			   int level, u64 time_seq, const u64 *extent_item_pos,
+			   u64 total_refs)
 {
 	int ret = 0;
 	int slot;
 	struct extent_buffer *eb;
 	struct btrfs_key key;
+	struct btrfs_key *key_for_search = &ref->key_for_search;
 	struct btrfs_file_extent_item *fi;
-	struct extent_inode_elem *eie = NULL;
+	struct extent_inode_elem *eie = NULL, *old = NULL;
 	u64 disk_byte;
+	u64 wanted_disk_byte = ref->wanted_disk_byte;
+	u64 count = 0;
 
 	if (level != 0) {
 		eb = path->nodes[level];
@@ -208,7 +250,7 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0]))
 		ret = btrfs_next_old_leaf(root, path, time_seq);
 
-	while (!ret) {
+	while (!ret && count < total_refs) {
 		eb = path->nodes[0];
 		slot = path->slots[0];
 
@@ -223,6 +265,8 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 
 		if (disk_byte == wanted_disk_byte) {
 			eie = NULL;
+			old = NULL;
+			count++;
 			if (extent_item_pos) {
 				ret = check_extent_in_eb(&key, eb, fi,
 						*extent_item_pos,
@@ -230,23 +274,28 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 				if (ret < 0)
 					break;
 			}
-			if (!ret) {
-				ret = ulist_add(parents, eb->start,
-						(uintptr_t)eie, GFP_NOFS);
-				if (ret < 0)
-					break;
-				if (!extent_item_pos) {
-					ret = btrfs_next_old_leaf(root, path,
-							time_seq);
-					continue;
-				}
+			if (ret > 0)
+				goto next;
+			ret = ulist_add_merge(parents, eb->start,
+					      (uintptr_t)eie,
+					      (u64 *)&old, GFP_NOFS);
+			if (ret < 0)
+				break;
+			if (!ret && extent_item_pos) {
+				while (old->next)
+					old = old->next;
+				old->next = eie;
 			}
+			eie = NULL;
 		}
+next:
 		ret = btrfs_next_old_item(root, path, time_seq);
 	}
 
 	if (ret > 0)
 		ret = 0;
+	else if (ret < 0)
+		free_inode_elem_list(eie);
 	return ret;
 }
 
@@ -255,54 +304,59 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
  * to a logical address
  */
 static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
-					int search_commit_root,
-					u64 time_seq,
-					struct __prelim_ref *ref,
-					struct ulist *parents,
-					const u64 *extent_item_pos)
+				  struct btrfs_path *path, u64 time_seq,
+				  struct __prelim_ref *ref,
+				  struct ulist *parents,
+				  const u64 *extent_item_pos, u64 total_refs)
 {
-	struct btrfs_path *path;
 	struct btrfs_root *root;
 	struct btrfs_key root_key;
 	struct extent_buffer *eb;
 	int ret = 0;
 	int root_level;
 	int level = ref->level;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-	path->search_commit_root = !!search_commit_root;
+	int index;
 
 	root_key.objectid = ref->root_id;
 	root_key.type = BTRFS_ROOT_ITEM_KEY;
 	root_key.offset = (u64)-1;
+
+	index = srcu_read_lock(&fs_info->subvol_srcu);
+
 	root = btrfs_read_fs_root_no_name(fs_info, &root_key);
 	if (IS_ERR(root)) {
+		srcu_read_unlock(&fs_info->subvol_srcu, index);
 		ret = PTR_ERR(root);
 		goto out;
 	}
 
-	root_level = btrfs_old_root_level(root, time_seq);
+	if (path->search_commit_root)
+		root_level = btrfs_header_level(root->commit_root);
+	else
+		root_level = btrfs_old_root_level(root, time_seq);
 
-	if (root_level + 1 == level)
+	if (root_level + 1 == level) {
+		srcu_read_unlock(&fs_info->subvol_srcu, index);
 		goto out;
+	}
 
 	path->lowest_level = level;
 	ret = btrfs_search_old_slot(root, &ref->key_for_search, path, time_seq);
+
+	/* root node has been locked, we can release @subvol_srcu safely here */
+	srcu_read_unlock(&fs_info->subvol_srcu, index);
+
 	pr_debug("search slot in root %llu (level %d, ref count %d) returned "
 		 "%d for key (%llu %u %llu)\n",
-		 (unsigned long long)ref->root_id, level, ref->count, ret,
-		 (unsigned long long)ref->key_for_search.objectid,
-		 ref->key_for_search.type,
-		 (unsigned long long)ref->key_for_search.offset);
+		 ref->root_id, level, ref->count, ret,
+		 ref->key_for_search.objectid, ref->key_for_search.type,
+		 ref->key_for_search.offset);
 	if (ret < 0)
 		goto out;
 
 	eb = path->nodes[level];
 	while (!eb) {
-		if (!level) {
-			WARN_ON(1);
+		if (WARN_ON(!level)) {
 			ret = 1;
 			goto out;
 		}
@@ -310,11 +364,11 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 		eb = path->nodes[level];
 	}
 
-	ret = add_all_parents(root, path, parents, level, &ref->key_for_search,
-				time_seq, ref->wanted_disk_byte,
-				extent_item_pos);
+	ret = add_all_parents(root, path, parents, ref, level, time_seq,
+			      extent_item_pos, total_refs);
 out:
-	btrfs_free_path(path);
+	path->lowest_level = 0;
+	btrfs_release_path(path);
 	return ret;
 }
 
@@ -322,9 +376,9 @@ out:
  * resolve all indirect backrefs from the list
  */
 static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
-				   int search_commit_root, u64 time_seq,
+				   struct btrfs_path *path, u64 time_seq,
 				   struct list_head *head,
-				   const u64 *extent_item_pos)
+				   const u64 *extent_item_pos, u64 total_refs)
 {
 	int err;
 	int ret = 0;
@@ -349,13 +403,18 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 			continue;
 		if (ref->count == 0)
 			continue;
-		err = __resolve_indirect_ref(fs_info, search_commit_root,
-					     time_seq, ref, parents,
-					     extent_item_pos);
-		if (err) {
-			if (ret == 0)
-				ret = err;
+		err = __resolve_indirect_ref(fs_info, path, time_seq, ref,
+					     parents, extent_item_pos,
+					     total_refs);
+		/*
+		 * we can only tolerate ENOENT,otherwise,we should catch error
+		 * and return directly.
+		 */
+		if (err == -ENOENT) {
 			continue;
+		} else if (err) {
+			ret = err;
+			goto out;
 		}
 
 		/* we put the first parent into the ref at hand */
@@ -363,14 +422,15 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 		node = ulist_next(parents, &uiter);
 		ref->parent = node ? node->val : 0;
 		ref->inode_list = node ?
-			(struct extent_inode_elem *)(uintptr_t)node->aux : 0;
+			(struct extent_inode_elem *)(uintptr_t)node->aux : NULL;
 
 		/* additional parents require new refs being added here */
 		while ((node = ulist_next(parents, &uiter))) {
-			new_ref = kmalloc(sizeof(*new_ref), GFP_NOFS);
+			new_ref = kmem_cache_alloc(btrfs_prelim_ref_cache,
+						   GFP_NOFS);
 			if (!new_ref) {
 				ret = -ENOMEM;
-				break;
+				goto out;
 			}
 			memcpy(new_ref, ref, sizeof(*ref));
 			new_ref->parent = node->val;
@@ -380,7 +440,7 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 		}
 		ulist_reinit(parents);
 	}
-
+out:
 	ulist_free(parents);
 	return ret;
 }
@@ -424,7 +484,10 @@ static int __add_missing_keys(struct btrfs_fs_info *fs_info,
 		BUG_ON(!ref->wanted_disk_byte);
 		eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
 				     fs_info->tree_root->leafsize, 0);
-		BUG_ON(!eb);
+		if (!eb || !extent_buffer_uptodate(eb)) {
+			free_extent_buffer(eb);
+			return -EIO;
+		}
 		btrfs_tree_read_lock(eb);
 		if (btrfs_header_level(eb) == 0)
 			btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
@@ -446,7 +509,7 @@ static int __add_missing_keys(struct btrfs_fs_info *fs_info,
  *           having a parent).
  * mode = 2: merge identical parents
  */
-static int __merge_refs(struct list_head *head, int mode)
+static void __merge_refs(struct list_head *head, int mode)
 {
 	struct list_head *pos1;
 
@@ -488,11 +551,10 @@ static int __merge_refs(struct list_head *head, int mode)
 			ref1->count += ref2->count;
 
 			list_del(&ref2->list);
-			kfree(ref2);
+			kmem_cache_free(btrfs_prelim_ref_cache, ref2);
 		}
 
 	}
-	return 0;
 }
 
 /*
@@ -500,7 +562,7 @@ static int __merge_refs(struct list_head *head, int mode)
  * smaller or equal that seq to the list
  */
 static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
-			      struct list_head *prefs)
+			      struct list_head *prefs, u64 *total_refs)
 {
 	struct btrfs_delayed_extent_op *extent_op = head->extent_op;
 	struct rb_node *n = &head->node.rb_node;
@@ -512,14 +574,13 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 	if (extent_op && extent_op->update_key)
 		btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
 
-	while ((n = rb_prev(n))) {
+	spin_lock(&head->lock);
+	n = rb_first(&head->ref_root);
+	while (n) {
 		struct btrfs_delayed_ref_node *node;
 		node = rb_entry(n, struct btrfs_delayed_ref_node,
 				rb_node);
-		if (node->bytenr != head->node.bytenr)
-			break;
-		WARN_ON(node->is_head);
-
+		n = rb_next(n);
 		if (node->seq > seq)
 			continue;
 
@@ -537,6 +598,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 		default:
 			BUG_ON(1);
 		}
+		*total_refs += (node->ref_mod * sgn);
 		switch (node->type) {
 		case BTRFS_TREE_BLOCK_REF_KEY: {
 			struct btrfs_delayed_tree_ref *ref;
@@ -544,7 +606,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 			ref = btrfs_delayed_node_to_tree_ref(node);
 			ret = __add_prelim_ref(prefs, ref->root, &op_key,
 					       ref->level + 1, 0, node->bytenr,
-					       node->ref_mod * sgn);
+					       node->ref_mod * sgn, GFP_ATOMIC);
 			break;
 		}
 		case BTRFS_SHARED_BLOCK_REF_KEY: {
@@ -554,7 +616,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 			ret = __add_prelim_ref(prefs, ref->root, NULL,
 					       ref->level + 1, ref->parent,
 					       node->bytenr,
-					       node->ref_mod * sgn);
+					       node->ref_mod * sgn, GFP_ATOMIC);
 			break;
 		}
 		case BTRFS_EXTENT_DATA_REF_KEY: {
@@ -566,7 +628,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 			key.offset = ref->offset;
 			ret = __add_prelim_ref(prefs, ref->root, &key, 0, 0,
 					       node->bytenr,
-					       node->ref_mod * sgn);
+					       node->ref_mod * sgn, GFP_ATOMIC);
 			break;
 		}
 		case BTRFS_SHARED_DATA_REF_KEY: {
@@ -579,16 +641,17 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 			key.offset = ref->offset;
 			ret = __add_prelim_ref(prefs, ref->root, &key, 0,
 					       ref->parent, node->bytenr,
-					       node->ref_mod * sgn);
+					       node->ref_mod * sgn, GFP_ATOMIC);
 			break;
 		}
 		default:
 			WARN_ON(1);
 		}
-		BUG_ON(ret);
+		if (ret)
+			break;
 	}
-
-	return 0;
+	spin_unlock(&head->lock);
+	return ret;
 }
 
 /*
@@ -596,12 +659,14 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
  */
 static int __add_inline_refs(struct btrfs_fs_info *fs_info,
 			     struct btrfs_path *path, u64 bytenr,
-			     int *info_level, struct list_head *prefs)
+			     int *info_level, struct list_head *prefs,
+			     u64 *total_refs)
 {
 	int ret = 0;
 	int slot;
 	struct extent_buffer *leaf;
 	struct btrfs_key key;
+	struct btrfs_key found_key;
 	unsigned long ptr;
 	unsigned long end;
 	struct btrfs_extent_item *ei;
@@ -619,17 +684,22 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
 
 	ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
 	flags = btrfs_extent_flags(leaf, ei);
+	*total_refs += btrfs_extent_refs(leaf, ei);
+	btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
 	ptr = (unsigned long)(ei + 1);
 	end = (unsigned long)ei + item_size;
 
-	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+	if (found_key.type == BTRFS_EXTENT_ITEM_KEY &&
+	    flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
 		struct btrfs_tree_block_info *info;
 
 		info = (struct btrfs_tree_block_info *)ptr;
 		*info_level = btrfs_tree_block_level(leaf, info);
 		ptr += sizeof(struct btrfs_tree_block_info);
 		BUG_ON(ptr > end);
+	} else if (found_key.type == BTRFS_METADATA_ITEM_KEY) {
+		*info_level = found_key.offset;
 	} else {
 		BUG_ON(!(flags & BTRFS_EXTENT_FLAG_DATA));
 	}
@@ -647,7 +717,7 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
 		case BTRFS_SHARED_BLOCK_REF_KEY:
 			ret = __add_prelim_ref(prefs, 0, NULL,
 						*info_level + 1, offset,
-						bytenr, 1);
+						bytenr, 1, GFP_NOFS);
 			break;
 		case BTRFS_SHARED_DATA_REF_KEY: {
 			struct btrfs_shared_data_ref *sdref;
@@ -656,13 +726,13 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
 			sdref = (struct btrfs_shared_data_ref *)(iref + 1);
 			count = btrfs_shared_data_ref_count(leaf, sdref);
 			ret = __add_prelim_ref(prefs, 0, NULL, 0, offset,
-					       bytenr, count);
+					       bytenr, count, GFP_NOFS);
 			break;
 		}
 		case BTRFS_TREE_BLOCK_REF_KEY:
 			ret = __add_prelim_ref(prefs, offset, NULL,
 					       *info_level + 1, 0,
-					       bytenr, 1);
+					       bytenr, 1, GFP_NOFS);
 			break;
 		case BTRFS_EXTENT_DATA_REF_KEY: {
 			struct btrfs_extent_data_ref *dref;
@@ -677,13 +747,14 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
 			key.offset = btrfs_extent_data_ref_offset(leaf, dref);
 			root = btrfs_extent_data_ref_root(leaf, dref);
 			ret = __add_prelim_ref(prefs, root, &key, 0, 0,
-					       bytenr, count);
+					       bytenr, count, GFP_NOFS);
 			break;
 		}
 		default:
 			WARN_ON(1);
 		}
-		BUG_ON(ret);
+		if (ret)
+			return ret;
 		ptr += btrfs_extent_inline_ref_size(type);
 	}
 
@@ -727,7 +798,7 @@ static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
 		case BTRFS_SHARED_BLOCK_REF_KEY:
 			ret = __add_prelim_ref(prefs, 0, NULL,
 						info_level + 1, key.offset,
-						bytenr, 1);
+						bytenr, 1, GFP_NOFS);
 			break;
 		case BTRFS_SHARED_DATA_REF_KEY: {
 			struct btrfs_shared_data_ref *sdref;
@@ -737,13 +808,13 @@ static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
 					      struct btrfs_shared_data_ref);
 			count = btrfs_shared_data_ref_count(leaf, sdref);
 			ret = __add_prelim_ref(prefs, 0, NULL, 0, key.offset,
-						bytenr, count);
+						bytenr, count, GFP_NOFS);
 			break;
 		}
 		case BTRFS_TREE_BLOCK_REF_KEY:
 			ret = __add_prelim_ref(prefs, key.offset, NULL,
 					       info_level + 1, 0,
-					       bytenr, 1);
+					       bytenr, 1, GFP_NOFS);
 			break;
 		case BTRFS_EXTENT_DATA_REF_KEY: {
 			struct btrfs_extent_data_ref *dref;
@@ -759,13 +830,15 @@ static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
 			key.offset = btrfs_extent_data_ref_offset(leaf, dref);
 			root = btrfs_extent_data_ref_root(leaf, dref);
 			ret = __add_prelim_ref(prefs, root, &key, 0, 0,
-					       bytenr, count);
+					       bytenr, count, GFP_NOFS);
 			break;
 		}
 		default:
 			WARN_ON(1);
 		}
-		BUG_ON(ret);
+		if (ret)
+			return ret;
+
 	}
 
 	return ret;
@@ -790,22 +863,29 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 	struct btrfs_delayed_ref_head *head;
 	int info_level = 0;
 	int ret;
-	int search_commit_root = (trans == BTRFS_BACKREF_SEARCH_COMMIT_ROOT);
 	struct list_head prefs_delayed;
 	struct list_head prefs;
 	struct __prelim_ref *ref;
+	struct extent_inode_elem *eie = NULL;
+	u64 total_refs = 0;
 
 	INIT_LIST_HEAD(&prefs);
 	INIT_LIST_HEAD(&prefs_delayed);
 
 	key.objectid = bytenr;
-	key.type = BTRFS_EXTENT_ITEM_KEY;
 	key.offset = (u64)-1;
+	if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
+		key.type = BTRFS_METADATA_ITEM_KEY;
+	else
+		key.type = BTRFS_EXTENT_ITEM_KEY;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-	path->search_commit_root = !!search_commit_root;
+	if (!trans) {
+		path->search_commit_root = 1;
+		path->skip_locking = 1;
+	}
 
 	/*
 	 * grab both a lock on the path and a lock on the delayed ref head.
@@ -820,7 +900,11 @@ again:
 		goto out;
 	BUG_ON(ret == 0);
 
-	if (trans != BTRFS_BACKREF_SEARCH_COMMIT_ROOT) {
+#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
+	if (trans && likely(trans->type != __TRANS_DUMMY)) {
+#else
+	if (trans) {
+#endif
 		/*
 		 * look if there are updates for this ref queued and lock the
 		 * head
@@ -844,15 +928,15 @@ again:
 				btrfs_put_delayed_ref(&head->node);
 				goto again;
 			}
+			spin_unlock(&delayed_refs->lock);
 			ret = __add_delayed_refs(head, time_seq,
-						 &prefs_delayed);
+						 &prefs_delayed, &total_refs);
 			mutex_unlock(&head->mutex);
-			if (ret) {
-				spin_unlock(&delayed_refs->lock);
+			if (ret)
 				goto out;
-			}
+		} else {
+			spin_unlock(&delayed_refs->lock);
 		}
-		spin_unlock(&delayed_refs->lock);
 	}
 
 	if (path->slots[0]) {
@@ -864,9 +948,11 @@ again:
 		slot = path->slots[0];
 		btrfs_item_key_to_cpu(leaf, &key, slot);
 		if (key.objectid == bytenr &&
-		    key.type == BTRFS_EXTENT_ITEM_KEY) {
+		    (key.type == BTRFS_EXTENT_ITEM_KEY ||
+		     key.type == BTRFS_METADATA_ITEM_KEY)) {
 			ret = __add_inline_refs(fs_info, path, bytenr,
-						&info_level, &prefs);
+						&info_level, &prefs,
+						&total_refs);
 			if (ret)
 				goto out;
 			ret = __add_keyed_refs(fs_info, path, bytenr,
@@ -883,46 +969,50 @@ again:
 	if (ret)
 		goto out;
 
-	ret = __merge_refs(&prefs, 1);
-	if (ret)
-		goto out;
+	__merge_refs(&prefs, 1);
 
-	ret = __resolve_indirect_refs(fs_info, search_commit_root, time_seq,
-				      &prefs, extent_item_pos);
+	ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs,
+				      extent_item_pos, total_refs);
 	if (ret)
 		goto out;
 
-	ret = __merge_refs(&prefs, 2);
-	if (ret)
-		goto out;
+	__merge_refs(&prefs, 2);
 
 	while (!list_empty(&prefs)) {
 		ref = list_first_entry(&prefs, struct __prelim_ref, list);
-		list_del(&ref->list);
 		WARN_ON(ref->count < 0);
-		if (ref->count && ref->root_id && ref->parent == 0) {
+		if (roots && ref->count && ref->root_id && ref->parent == 0) {
 			/* no parent == root of tree */
 			ret = ulist_add(roots, ref->root_id, 0, GFP_NOFS);
-			BUG_ON(ret < 0);
+			if (ret < 0)
+				goto out;
 		}
 		if (ref->count && ref->parent) {
-			struct extent_inode_elem *eie = NULL;
-			if (extent_item_pos && !ref->inode_list) {
+			if (extent_item_pos && !ref->inode_list &&
+			    ref->level == 0) {
 				u32 bsz;
 				struct extent_buffer *eb;
 				bsz = btrfs_level_size(fs_info->extent_root,
-							info_level);
+							ref->level);
 				eb = read_tree_block(fs_info->extent_root,
 							   ref->parent, bsz, 0);
-				BUG_ON(!eb);
+				if (!eb || !extent_buffer_uptodate(eb)) {
+					free_extent_buffer(eb);
+					ret = -EIO;
+					goto out;
+				}
 				ret = find_extent_in_eb(eb, bytenr,
 							*extent_item_pos, &eie);
-				ref->inode_list = eie;
 				free_extent_buffer(eb);
+				if (ret < 0)
+					goto out;
+				ref->inode_list = eie;
 			}
 			ret = ulist_add_merge(refs, ref->parent,
 					      (uintptr_t)ref->inode_list,
 					      (u64 *)&eie, GFP_NOFS);
+			if (ret < 0)
+				goto out;
 			if (!ret && extent_item_pos) {
 				/*
 				 * we've recorded that parent, so we must extend
@@ -933,9 +1023,10 @@ again:
 					eie = eie->next;
 				eie->next = ref->inode_list;
 			}
-			BUG_ON(ret < 0);
+			eie = NULL;
 		}
-		kfree(ref);
+		list_del(&ref->list);
+		kmem_cache_free(btrfs_prelim_ref_cache, ref);
 	}
 
 out:
@@ -943,15 +1034,16 @@ out:
 	while (!list_empty(&prefs)) {
 		ref = list_first_entry(&prefs, struct __prelim_ref, list);
 		list_del(&ref->list);
-		kfree(ref);
+		kmem_cache_free(btrfs_prelim_ref_cache, ref);
 	}
 	while (!list_empty(&prefs_delayed)) {
 		ref = list_first_entry(&prefs_delayed, struct __prelim_ref,
 				       list);
 		list_del(&ref->list);
-		kfree(ref);
+		kmem_cache_free(btrfs_prelim_ref_cache, ref);
 	}
-
+	if (ret < 0)
+		free_inode_elem_list(eie);
 	return ret;
 }
 
@@ -959,7 +1051,6 @@ static void free_leaf_list(struct ulist *blocks)
 {
 	struct ulist_node *node = NULL;
 	struct extent_inode_elem *eie;
-	struct extent_inode_elem *eie_next;
 	struct ulist_iterator uiter;
 
 	ULIST_ITER_INIT(&uiter);
@@ -967,10 +1058,7 @@ static void free_leaf_list(struct ulist *blocks)
 		if (!node->aux)
 			continue;
 		eie = (struct extent_inode_elem *)(uintptr_t)node->aux;
-		for (; eie; eie = eie_next) {
-			eie_next = eie->next;
-			kfree(eie);
-		}
+		free_inode_elem_list(eie);
 		node->aux = 0;
 	}
 
@@ -990,22 +1078,14 @@ static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
 				u64 time_seq, struct ulist **leafs,
 				const u64 *extent_item_pos)
 {
-	struct ulist *tmp;
 	int ret;
 
-	tmp = ulist_alloc(GFP_NOFS);
-	if (!tmp)
-		return -ENOMEM;
 	*leafs = ulist_alloc(GFP_NOFS);
-	if (!*leafs) {
-		ulist_free(tmp);
+	if (!*leafs)
 		return -ENOMEM;
-	}
 
 	ret = find_parent_nodes(trans, fs_info, bytenr,
-				time_seq, *leafs, tmp, extent_item_pos);
-	ulist_free(tmp);
-
+				time_seq, *leafs, NULL, extent_item_pos);
 	if (ret < 0 && ret != -ENOENT) {
 		free_leaf_list(*leafs);
 		return ret;
@@ -1027,9 +1107,9 @@ static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
  *
  * returns 0 on success, < 0 on error.
  */
-int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
-				struct btrfs_fs_info *fs_info, u64 bytenr,
-				u64 time_seq, struct ulist **roots)
+static int __btrfs_find_all_roots(struct btrfs_trans_handle *trans,
+				  struct btrfs_fs_info *fs_info, u64 bytenr,
+				  u64 time_seq, struct ulist **roots)
 {
 	struct ulist *tmp;
 	struct ulist_node *node = NULL;
@@ -1058,42 +1138,25 @@ int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
 		if (!node)
 			break;
 		bytenr = node->val;
+		cond_resched();
 	}
 
 	ulist_free(tmp);
 	return 0;
 }
 
-
-static int __inode_info(u64 inum, u64 ioff, u8 key_type,
-			struct btrfs_root *fs_root, struct btrfs_path *path,
-			struct btrfs_key *found_key)
+int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
+			 struct btrfs_fs_info *fs_info, u64 bytenr,
+			 u64 time_seq, struct ulist **roots)
 {
 	int ret;
-	struct btrfs_key key;
-	struct extent_buffer *eb;
 
-	key.type = key_type;
-	key.objectid = inum;
-	key.offset = ioff;
-
-	ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
-	if (ret < 0)
-		return ret;
-
-	eb = path->nodes[0];
-	if (ret && path->slots[0] >= btrfs_header_nritems(eb)) {
-		ret = btrfs_next_leaf(fs_root, path);
-		if (ret)
-			return ret;
-		eb = path->nodes[0];
-	}
-
-	btrfs_item_key_to_cpu(eb, found_key, path->slots[0]);
-	if (found_key->type != key.type || found_key->objectid != key.objectid)
-		return 1;
-
-	return 0;
+	if (!trans)
+		down_read(&fs_info->commit_root_sem);
+	ret = __btrfs_find_all_roots(trans, fs_info, bytenr, time_seq, roots);
+	if (!trans)
+		up_read(&fs_info->commit_root_sem);
+	return ret;
 }
 
 /*
@@ -1103,16 +1166,16 @@ int inode_item_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
 			struct btrfs_path *path)
 {
 	struct btrfs_key key;
-	return __inode_info(inum, ioff, BTRFS_INODE_ITEM_KEY, fs_root, path,
-				&key);
+	return btrfs_find_item(fs_root, path, inum, ioff,
+			BTRFS_INODE_ITEM_KEY, &key);
 }
 
 static int inode_ref_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
 				struct btrfs_path *path,
 				struct btrfs_key *found_key)
 {
-	return __inode_info(inum, ioff, BTRFS_INODE_REF_KEY, fs_root, path,
-				found_key);
+	return btrfs_find_item(fs_root, path, inum, ioff,
+			BTRFS_INODE_REF_KEY, found_key);
 }
 
 int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
@@ -1183,6 +1246,20 @@ int btrfs_find_one_extref(struct btrfs_root *root, u64 inode_objectid,
 	return ret;
 }
 
+/*
+ * this iterates to turn a name (from iref/extref) into a full filesystem path.
+ * Elements of the path are separated by '/' and the path is guaranteed to be
+ * 0-terminated. the path is only given within the current file system.
+ * Therefore, it never starts with a '/'. the caller is responsible to provide
+ * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
+ * the start point of the resulting string is returned. this pointer is within
+ * dest, normally.
+ * in case the path buffer would overflow, the pointer is decremented further
+ * as if output was written to the buffer, though no more output is actually
+ * generated. that way, the caller can determine how much space would be
+ * required for the path to fit into the buffer. in that case, the returned
+ * value will be smaller than dest. callers must check this!
+ */
 char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 			u32 name_len, unsigned long name_off,
 			struct extent_buffer *eb_in, u64 parent,
@@ -1252,32 +1329,6 @@ char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 }
 
 /*
- * this iterates to turn a btrfs_inode_ref into a full filesystem path. elements
- * of the path are separated by '/' and the path is guaranteed to be
- * 0-terminated. the path is only given within the current file system.
- * Therefore, it never starts with a '/'. the caller is responsible to provide
- * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
- * the start point of the resulting string is returned. this pointer is within
- * dest, normally.
- * in case the path buffer would overflow, the pointer is decremented further
- * as if output was written to the buffer, though no more output is actually
- * generated. that way, the caller can determine how much space would be
- * required for the path to fit into the buffer. in that case, the returned
- * value will be smaller than dest. callers must check this!
- */
-char *btrfs_iref_to_path(struct btrfs_root *fs_root,
-			 struct btrfs_path *path,
-			 struct btrfs_inode_ref *iref,
-			 struct extent_buffer *eb_in, u64 parent,
-			 char *dest, u32 size)
-{
-	return btrfs_ref_to_path(fs_root, path,
-				 btrfs_inode_ref_name_len(eb_in, iref),
-				 (unsigned long)(iref + 1),
-				 eb_in, parent, dest, size);
-}
-
-/*
  * this makes the path point to (logical EXTENT_ITEM *)
  * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
  * tree blocks and <0 on error.
@@ -1288,29 +1339,38 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
 {
 	int ret;
 	u64 flags;
+	u64 size = 0;
 	u32 item_size;
 	struct extent_buffer *eb;
 	struct btrfs_extent_item *ei;
 	struct btrfs_key key;
 
-	key.type = BTRFS_EXTENT_ITEM_KEY;
+	if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
+		key.type = BTRFS_METADATA_ITEM_KEY;
+	else
+		key.type = BTRFS_EXTENT_ITEM_KEY;
 	key.objectid = logical;
 	key.offset = (u64)-1;
 
 	ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
 	if (ret < 0)
 		return ret;
-	ret = btrfs_previous_item(fs_info->extent_root, path,
-					0, BTRFS_EXTENT_ITEM_KEY);
-	if (ret < 0)
-		return ret;
 
+	ret = btrfs_previous_extent_item(fs_info->extent_root, path, 0);
+	if (ret) {
+		if (ret > 0)
+			ret = -ENOENT;
+		return ret;
+	}
 	btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
-	if (found_key->type != BTRFS_EXTENT_ITEM_KEY ||
-	    found_key->objectid > logical ||
-	    found_key->objectid + found_key->offset <= logical) {
-		pr_debug("logical %llu is not within any extent\n",
-			 (unsigned long long)logical);
+	if (found_key->type == BTRFS_METADATA_ITEM_KEY)
+		size = fs_info->extent_root->leafsize;
+	else if (found_key->type == BTRFS_EXTENT_ITEM_KEY)
+		size = found_key->offset;
+
+	if (found_key->objectid > logical ||
+	    found_key->objectid + size <= logical) {
+		pr_debug("logical %llu is not within any extent\n", logical);
 		return -ENOENT;
 	}
 
@@ -1323,11 +1383,8 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
 
 	pr_debug("logical %llu is at position %llu within the extent (%llu "
 		 "EXTENT_ITEM %llu) flags %#llx size %u\n",
-		 (unsigned long long)logical,
-		 (unsigned long long)(logical - found_key->objectid),
-		 (unsigned long long)found_key->objectid,
-		 (unsigned long long)found_key->offset,
-		 (unsigned long long)flags, item_size);
+		 logical, logical - found_key->objectid, found_key->objectid,
+		 found_key->offset, flags, item_size);
 
 	WARN_ON(!flags_ret);
 	if (flags_ret) {
@@ -1352,9 +1409,10 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
  * returns <0 on error
  */
 static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
-				struct btrfs_extent_item *ei, u32 item_size,
-				struct btrfs_extent_inline_ref **out_eiref,
-				int *out_type)
+				   struct btrfs_key *key,
+				   struct btrfs_extent_item *ei, u32 item_size,
+				   struct btrfs_extent_inline_ref **out_eiref,
+				   int *out_type)
 {
 	unsigned long end;
 	u64 flags;
@@ -1364,19 +1422,26 @@ static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
 		/* first call */
 		flags = btrfs_extent_flags(eb, ei);
 		if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
-			info = (struct btrfs_tree_block_info *)(ei + 1);
-			*out_eiref =
-				(struct btrfs_extent_inline_ref *)(info + 1);
+			if (key->type == BTRFS_METADATA_ITEM_KEY) {
+				/* a skinny metadata extent */
+				*out_eiref =
+				     (struct btrfs_extent_inline_ref *)(ei + 1);
+			} else {
+				WARN_ON(key->type != BTRFS_EXTENT_ITEM_KEY);
+				info = (struct btrfs_tree_block_info *)(ei + 1);
+				*out_eiref =
+				   (struct btrfs_extent_inline_ref *)(info + 1);
+			}
 		} else {
 			*out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
 		}
 		*ptr = (unsigned long)*out_eiref;
-		if ((void *)*ptr >= (void *)ei + item_size)
+		if ((unsigned long)(*ptr) >= (unsigned long)ei + item_size)
 			return -ENOENT;
 	}
 
 	end = (unsigned long)ei + item_size;
-	*out_eiref = (struct btrfs_extent_inline_ref *)*ptr;
+	*out_eiref = (struct btrfs_extent_inline_ref *)(*ptr);
 	*out_type = btrfs_extent_inline_ref_type(eb, *out_eiref);
 
 	*ptr += btrfs_extent_inline_ref_size(*out_type);
@@ -1395,8 +1460,8 @@ static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
  * <0 on error.
  */
 int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
-				struct btrfs_extent_item *ei, u32 item_size,
-				u64 *out_root, u8 *out_level)
+			    struct btrfs_key *key, struct btrfs_extent_item *ei,
+			    u32 item_size, u64 *out_root, u8 *out_level)
 {
 	int ret;
 	int type;
@@ -1407,8 +1472,8 @@ int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
 		return 1;
 
 	while (1) {
-		ret = __get_extent_inline_ref(ptr, eb, ei, item_size,
-						&eiref, &type);
+		ret = __get_extent_inline_ref(ptr, eb, key, ei, item_size,
+					      &eiref, &type);
 		if (ret < 0)
 			return ret;
 
@@ -1464,9 +1529,7 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 				iterate_extent_inodes_t *iterate, void *ctx)
 {
 	int ret;
-	struct list_head data_refs = LIST_HEAD_INIT(data_refs);
-	struct list_head shared_refs = LIST_HEAD_INIT(shared_refs);
-	struct btrfs_trans_handle *trans;
+	struct btrfs_trans_handle *trans = NULL;
 	struct ulist *refs = NULL;
 	struct ulist *roots = NULL;
 	struct ulist_node *ref_node = NULL;
@@ -1478,13 +1541,13 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 	pr_debug("resolving all inodes for extent %llu\n",
 			extent_item_objectid);
 
-	if (search_commit_root) {
-		trans = BTRFS_BACKREF_SEARCH_COMMIT_ROOT;
-	} else {
+	if (!search_commit_root) {
 		trans = btrfs_join_transaction(fs_info->extent_root);
 		if (IS_ERR(trans))
 			return PTR_ERR(trans);
 		btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
+	} else {
+		down_read(&fs_info->commit_root_sem);
 	}
 
 	ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
@@ -1495,15 +1558,15 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 
 	ULIST_ITER_INIT(&ref_uiter);
 	while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
-		ret = btrfs_find_all_roots(trans, fs_info, ref_node->val,
-					   tree_mod_seq_elem.seq, &roots);
+		ret = __btrfs_find_all_roots(trans, fs_info, ref_node->val,
+					     tree_mod_seq_elem.seq, &roots);
 		if (ret)
 			break;
 		ULIST_ITER_INIT(&root_uiter);
 		while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
 			pr_debug("root %llu references leaf %llu, data list "
 				 "%#llx\n", root_node->val, ref_node->val,
-				 (long long)ref_node->aux);
+				 ref_node->aux);
 			ret = iterate_leaf_refs((struct extent_inode_elem *)
 						(uintptr_t)ref_node->aux,
 						root_node->val,
@@ -1511,15 +1574,15 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 						iterate, ctx);
 		}
 		ulist_free(roots);
-		roots = NULL;
 	}
 
 	free_leaf_list(refs);
-	ulist_free(roots);
 out:
 	if (!search_commit_root) {
 		btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
 		btrfs_end_transaction(trans, fs_info->extent_root);
+	} else {
+		up_read(&fs_info->commit_root_sem);
 	}
 
 	return ret;
@@ -1570,7 +1633,6 @@ static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
 	struct btrfs_key found_key;
 
 	while (!ret) {
-		path->leave_spinning = 1;
 		ret = inode_ref_info(inum, parent ? parent+1 : 0, fs_root, path,
 				     &found_key);
 		if (ret < 0)
@@ -1583,23 +1645,25 @@ static int iterate_inode_refs(u64 inum, struct btrfs_root *fs_root,
 
 		parent = found_key.offset;
 		slot = path->slots[0];
-		eb = path->nodes[0];
-		/* make sure we can use eb after releasing the path */
-		atomic_inc(&eb->refs);
+		eb = btrfs_clone_extent_buffer(path->nodes[0]);
+		if (!eb) {
+			ret = -ENOMEM;
+			break;
+		}
+		extent_buffer_get(eb);
 		btrfs_tree_read_lock(eb);
 		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
 		btrfs_release_path(path);
 
-		item = btrfs_item_nr(eb, slot);
+		item = btrfs_item_nr(slot);
 		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
 
 		for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
 			name_len = btrfs_inode_ref_name_len(eb, iref);
 			/* path must be released before calling iterate()! */
 			pr_debug("following ref at offset %u for inode %llu in "
-				 "tree %llu\n", cur,
-				 (unsigned long long)found_key.objectid,
-				 (unsigned long long)fs_root->objectid);
+				 "tree %llu\n", cur, found_key.objectid,
+				 fs_root->objectid);
 			ret = iterate(parent, name_len,
 				      (unsigned long)(iref + 1), eb, ctx);
 			if (ret)
@@ -1644,17 +1708,20 @@ static int iterate_inode_extrefs(u64 inum, struct btrfs_root *fs_root,
 		++found;
 
 		slot = path->slots[0];
-		eb = path->nodes[0];
-		/* make sure we can use eb after releasing the path */
-		atomic_inc(&eb->refs);
+		eb = btrfs_clone_extent_buffer(path->nodes[0]);
+		if (!eb) {
+			ret = -ENOMEM;
+			break;
+		}
+		extent_buffer_get(eb);
 
 		btrfs_tree_read_lock(eb);
 		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
 		btrfs_release_path(path);
 
 		leaf = path->nodes[0];
-		item_size = btrfs_item_size_nr(leaf, path->slots[0]);
-		ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
+		item_size = btrfs_item_size_nr(leaf, slot);
+		ptr = btrfs_item_ptr_offset(leaf, slot);
 		cur_offset = 0;
 
 		while (cur_offset < item_size) {