btrfs: remove delayed_ref_node from ref_head

This is just excessive information in the ref_head, and makes the code
complicated.  It is a relic from when we had the heads and the refs in
the same tree, which is no longer the case.  With this removal I've
cleaned up a bunch of the cruft around this old assumption as well.

Signed-off-by: Josef Bacik <jbacik@fb.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>

6 files changed, 119 insertions, 175 deletions

diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index b517ef1477ea..33cba1abf8b6 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -1178,7 +1178,7 @@ again:
 		head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
 		if (head) {
 			if (!mutex_trylock(&head->mutex)) {
-				refcount_inc(&head->node.refs);
+				refcount_inc(&head->refs);
 				spin_unlock(&delayed_refs->lock);
 
 				btrfs_release_path(path);
@@ -1189,7 +1189,7 @@ again:
 				 */
 				mutex_lock(&head->mutex);
 				mutex_unlock(&head->mutex);
-				btrfs_put_delayed_ref(&head->node);
+				btrfs_put_delayed_ref_head(head);
 				goto again;
 			}
 			spin_unlock(&delayed_refs->lock);
diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index 93ffa898df6d..b9b41c838da4 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -96,15 +96,15 @@ static struct btrfs_delayed_ref_head *htree_insert(struct rb_root *root,
 	u64 bytenr;
 
 	ins = rb_entry(node, struct btrfs_delayed_ref_head, href_node);
-	bytenr = ins->node.bytenr;
+	bytenr = ins->bytenr;
 	while (*p) {
 		parent_node = *p;
 		entry = rb_entry(parent_node, struct btrfs_delayed_ref_head,
 				 href_node);
 
-		if (bytenr < entry->node.bytenr)
+		if (bytenr < entry->bytenr)
 			p = &(*p)->rb_left;
-		else if (bytenr > entry->node.bytenr)
+		else if (bytenr > entry->bytenr)
 			p = &(*p)->rb_right;
 		else
 			return entry;
@@ -133,15 +133,15 @@ find_ref_head(struct rb_root *root, u64 bytenr,
 	while (n) {
 		entry = rb_entry(n, struct btrfs_delayed_ref_head, href_node);
 
-		if (bytenr < entry->node.bytenr)
+		if (bytenr < entry->bytenr)
 			n = n->rb_left;
-		else if (bytenr > entry->node.bytenr)
+		else if (bytenr > entry->bytenr)
 			n = n->rb_right;
 		else
 			return entry;
 	}
 	if (entry && return_bigger) {
-		if (bytenr > entry->node.bytenr) {
+		if (bytenr > entry->bytenr) {
 			n = rb_next(&entry->href_node);
 			if (!n)
 				n = rb_first(root);
@@ -164,17 +164,17 @@ int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans,
 	if (mutex_trylock(&head->mutex))
 		return 0;
 
-	refcount_inc(&head->node.refs);
+	refcount_inc(&head->refs);
 	spin_unlock(&delayed_refs->lock);
 
 	mutex_lock(&head->mutex);
 	spin_lock(&delayed_refs->lock);
-	if (!head->node.in_tree) {
+	if (RB_EMPTY_NODE(&head->href_node)) {
 		mutex_unlock(&head->mutex);
-		btrfs_put_delayed_ref(&head->node);
+		btrfs_put_delayed_ref_head(head);
 		return -EAGAIN;
 	}
-	btrfs_put_delayed_ref(&head->node);
+	btrfs_put_delayed_ref_head(head);
 	return 0;
 }
 
@@ -183,15 +183,10 @@ static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
 				    struct btrfs_delayed_ref_head *head,
 				    struct btrfs_delayed_ref_node *ref)
 {
-	if (btrfs_delayed_ref_is_head(ref)) {
-		head = btrfs_delayed_node_to_head(ref);
-		rb_erase(&head->href_node, &delayed_refs->href_root);
-	} else {
-		assert_spin_locked(&head->lock);
-		list_del(&ref->list);
-		if (!list_empty(&ref->add_list))
-			list_del(&ref->add_list);
-	}
+	assert_spin_locked(&head->lock);
+	list_del(&ref->list);
+	if (!list_empty(&ref->add_list))
+		list_del(&ref->add_list);
 	ref->in_tree = 0;
 	btrfs_put_delayed_ref(ref);
 	atomic_dec(&delayed_refs->num_entries);
@@ -380,8 +375,8 @@ again:
 	head->processing = 1;
 	WARN_ON(delayed_refs->num_heads_ready == 0);
 	delayed_refs->num_heads_ready--;
-	delayed_refs->run_delayed_start = head->node.bytenr +
-		head->node.num_bytes;
+	delayed_refs->run_delayed_start = head->bytenr +
+		head->num_bytes;
 	return head;
 }
 
@@ -469,20 +464,16 @@ add_tail:
  */
 static noinline void
 update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
-			 struct btrfs_delayed_ref_node *existing,
-			 struct btrfs_delayed_ref_node *update,
+			 struct btrfs_delayed_ref_head *existing,
+			 struct btrfs_delayed_ref_head *update,
 			 int *old_ref_mod_ret)
 {
-	struct btrfs_delayed_ref_head *existing_ref;
-	struct btrfs_delayed_ref_head *ref;
 	int old_ref_mod;
 
-	existing_ref = btrfs_delayed_node_to_head(existing);
-	ref = btrfs_delayed_node_to_head(update);
-	BUG_ON(existing_ref->is_data != ref->is_data);
+	BUG_ON(existing->is_data != update->is_data);
 
-	spin_lock(&existing_ref->lock);
-	if (ref->must_insert_reserved) {
+	spin_lock(&existing->lock);
+	if (update->must_insert_reserved) {
 		/* if the extent was freed and then
 		 * reallocated before the delayed ref
 		 * entries were processed, we can end up
@@ -490,7 +481,7 @@ update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
 		 * the must_insert_reserved flag set.
 		 * Set it again here
 		 */
-		existing_ref->must_insert_reserved = ref->must_insert_reserved;
+		existing->must_insert_reserved = update->must_insert_reserved;
 
 		/*
 		 * update the num_bytes so we make sure the accounting
@@ -500,22 +491,22 @@ update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
 
 	}
 
-	if (ref->extent_op) {
-		if (!existing_ref->extent_op) {
-			existing_ref->extent_op = ref->extent_op;
+	if (update->extent_op) {
+		if (!existing->extent_op) {
+			existing->extent_op = update->extent_op;
 		} else {
-			if (ref->extent_op->update_key) {
-				memcpy(&existing_ref->extent_op->key,
-				       &ref->extent_op->key,
-				       sizeof(ref->extent_op->key));
-				existing_ref->extent_op->update_key = true;
+			if (update->extent_op->update_key) {
+				memcpy(&existing->extent_op->key,
+				       &update->extent_op->key,
+				       sizeof(update->extent_op->key));
+				existing->extent_op->update_key = true;
 			}
-			if (ref->extent_op->update_flags) {
-				existing_ref->extent_op->flags_to_set |=
-					ref->extent_op->flags_to_set;
-				existing_ref->extent_op->update_flags = true;
+			if (update->extent_op->update_flags) {
+				existing->extent_op->flags_to_set |=
+					update->extent_op->flags_to_set;
+				existing->extent_op->update_flags = true;
 			}
-			btrfs_free_delayed_extent_op(ref->extent_op);
+			btrfs_free_delayed_extent_op(update->extent_op);
 		}
 	}
 	/*
@@ -523,23 +514,23 @@ update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
 	 * only need the lock for this case cause we could be processing it
 	 * currently, for refs we just added we know we're a-ok.
 	 */
-	old_ref_mod = existing_ref->total_ref_mod;
+	old_ref_mod = existing->total_ref_mod;
 	if (old_ref_mod_ret)
 		*old_ref_mod_ret = old_ref_mod;
 	existing->ref_mod += update->ref_mod;
-	existing_ref->total_ref_mod += update->ref_mod;
+	existing->total_ref_mod += update->ref_mod;
 
 	/*
 	 * If we are going to from a positive ref mod to a negative or vice
 	 * versa we need to make sure to adjust pending_csums accordingly.
 	 */
-	if (existing_ref->is_data) {
-		if (existing_ref->total_ref_mod >= 0 && old_ref_mod < 0)
+	if (existing->is_data) {
+		if (existing->total_ref_mod >= 0 && old_ref_mod < 0)
 			delayed_refs->pending_csums -= existing->num_bytes;
-		if (existing_ref->total_ref_mod < 0 && old_ref_mod >= 0)
+		if (existing->total_ref_mod < 0 && old_ref_mod >= 0)
 			delayed_refs->pending_csums += existing->num_bytes;
 	}
-	spin_unlock(&existing_ref->lock);
+	spin_unlock(&existing->lock);
 }
 
 /*
@@ -550,14 +541,13 @@ update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
 static noinline struct btrfs_delayed_ref_head *
 add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 		     struct btrfs_trans_handle *trans,
-		     struct btrfs_delayed_ref_node *ref,
+		     struct btrfs_delayed_ref_head *head_ref,
 		     struct btrfs_qgroup_extent_record *qrecord,
 		     u64 bytenr, u64 num_bytes, u64 ref_root, u64 reserved,
 		     int action, int is_data, int *qrecord_inserted_ret,
 		     int *old_ref_mod, int *new_ref_mod)
 {
 	struct btrfs_delayed_ref_head *existing;
-	struct btrfs_delayed_ref_head *head_ref = NULL;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	int count_mod = 1;
 	int must_insert_reserved = 0;
@@ -593,26 +583,21 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 
 	delayed_refs = &trans->transaction->delayed_refs;
 
-	/* first set the basic ref node struct up */
-	refcount_set(&ref->refs, 1);
-	ref->bytenr = bytenr;
-	ref->num_bytes = num_bytes;
-	ref->ref_mod = count_mod;
-	ref->type  = 0;
-	ref->action  = 0;
-	ref->is_head = 1;
-	ref->in_tree = 1;
-	ref->seq = 0;
-
-	head_ref = btrfs_delayed_node_to_head(ref);
+	refcount_set(&head_ref->refs, 1);
+	head_ref->bytenr = bytenr;
+	head_ref->num_bytes = num_bytes;
+	head_ref->ref_mod = count_mod;
 	head_ref->must_insert_reserved = must_insert_reserved;
 	head_ref->is_data = is_data;
 	INIT_LIST_HEAD(&head_ref->ref_list);
 	INIT_LIST_HEAD(&head_ref->ref_add_list);
+	RB_CLEAR_NODE(&head_ref->href_node);
 	head_ref->processing = 0;
 	head_ref->total_ref_mod = count_mod;
 	head_ref->qgroup_reserved = 0;
 	head_ref->qgroup_ref_root = 0;
+	spin_lock_init(&head_ref->lock);
+	mutex_init(&head_ref->mutex);
 
 	/* Record qgroup extent info if provided */
 	if (qrecord) {
@@ -632,17 +617,14 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 			qrecord_inserted = 1;
 	}
 
-	spin_lock_init(&head_ref->lock);
-	mutex_init(&head_ref->mutex);
-
-	trace_add_delayed_ref_head(fs_info, ref, head_ref, action);
+	trace_add_delayed_ref_head(fs_info, head_ref, action);
 
 	existing = htree_insert(&delayed_refs->href_root,
 				&head_ref->href_node);
 	if (existing) {
 		WARN_ON(ref_root && reserved && existing->qgroup_ref_root
 			&& existing->qgroup_reserved);
-		update_existing_head_ref(delayed_refs, &existing->node, ref,
+		update_existing_head_ref(delayed_refs, existing, head_ref,
 					 old_ref_mod);
 		/*
 		 * we've updated the existing ref, free the newly
@@ -821,7 +803,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	 * insert both the head node and the new ref without dropping
 	 * the spin lock
 	 */
-	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
+	head_ref = add_delayed_ref_head(fs_info, trans, head_ref, record,
 					bytenr, num_bytes, 0, 0, action, 0,
 					&qrecord_inserted, old_ref_mod,
 					new_ref_mod);
@@ -888,7 +870,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 	 * insert both the head node and the new ref without dropping
 	 * the spin lock
 	 */
-	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
+	head_ref = add_delayed_ref_head(fs_info, trans, head_ref, record,
 					bytenr, num_bytes, ref_root, reserved,
 					action, 1, &qrecord_inserted,
 					old_ref_mod, new_ref_mod);
@@ -920,7 +902,7 @@ int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
 	delayed_refs = &trans->transaction->delayed_refs;
 	spin_lock(&delayed_refs->lock);
 
-	add_delayed_ref_head(fs_info, trans, &head_ref->node, NULL, bytenr,
+	add_delayed_ref_head(fs_info, trans, head_ref, NULL, bytenr,
 			     num_bytes, 0, 0, BTRFS_UPDATE_DELAYED_HEAD,
 			     extent_op->is_data, NULL, NULL, NULL);
 
diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h
index ce88e4ac5276..1ce11858d727 100644
--- a/fs/btrfs/delayed-ref.h
+++ b/fs/btrfs/delayed-ref.h
@@ -26,15 +26,6 @@
 #define BTRFS_ADD_DELAYED_EXTENT 3 /* record a full extent allocation */
 #define BTRFS_UPDATE_DELAYED_HEAD 4 /* not changing ref count on head ref */
 
-/*
- * XXX: Qu: I really hate the design that ref_head and tree/data ref shares the
- * same ref_node structure.
- * Ref_head is in a higher logic level than tree/data ref, and duplicated
- * bytenr/num_bytes in ref_node is really a waste or memory, they should be
- * referred from ref_head.
- * This gets more disgusting after we use list to store tree/data ref in
- * ref_head. Must clean this mess up later.
- */
 struct btrfs_delayed_ref_node {
 	/*data/tree ref use list, stored in ref_head->ref_list. */
 	struct list_head list;
@@ -91,8 +82,9 @@ struct btrfs_delayed_extent_op {
  * reference count modifications we've queued up.
  */
 struct btrfs_delayed_ref_head {
-	struct btrfs_delayed_ref_node node;
-
+	u64 bytenr;
+	u64 num_bytes;
+	refcount_t refs;
 	/*
 	 * the mutex is held while running the refs, and it is also
 	 * held when checking the sum of reference modifications.
@@ -116,6 +108,14 @@ struct btrfs_delayed_ref_head {
 	int total_ref_mod;
 
 	/*
+	 * This is the current outstanding mod references for this bytenr.  This
+	 * is used with lookup_extent_info to get an accurate reference count
+	 * for a bytenr, so it is adjusted as delayed refs are run so that any
+	 * on disk reference count + ref_mod is accurate.
+	 */
+	int ref_mod;
+
+	/*
 	 * For qgroup reserved space freeing.
 	 *
 	 * ref_root and reserved will be recorded after
@@ -234,15 +234,18 @@ static inline void btrfs_put_delayed_ref(struct btrfs_delayed_ref_node *ref)
 		case BTRFS_SHARED_DATA_REF_KEY:
 			kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
 			break;
-		case 0:
-			kmem_cache_free(btrfs_delayed_ref_head_cachep, ref);
-			break;
 		default:
 			BUG();
 		}
 	}
 }
 
+static inline void btrfs_put_delayed_ref_head(struct btrfs_delayed_ref_head *head)
+{
+	if (refcount_dec_and_test(&head->refs))
+		kmem_cache_free(btrfs_delayed_ref_head_cachep, head);
+}
+
 int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 			       struct btrfs_trans_handle *trans,
 			       u64 bytenr, u64 num_bytes, u64 parent,
@@ -283,35 +286,17 @@ int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info,
 			    u64 seq);
 
 /*
- * a node might live in a head or a regular ref, this lets you
- * test for the proper type to use.
- */
-static int btrfs_delayed_ref_is_head(struct btrfs_delayed_ref_node *node)
-{
-	return node->is_head;
-}
-
-/*
  * helper functions to cast a node into its container
  */
 static inline struct btrfs_delayed_tree_ref *
 btrfs_delayed_node_to_tree_ref(struct btrfs_delayed_ref_node *node)
 {
-	WARN_ON(btrfs_delayed_ref_is_head(node));
 	return container_of(node, struct btrfs_delayed_tree_ref, node);
 }
 
 static inline struct btrfs_delayed_data_ref *
 btrfs_delayed_node_to_data_ref(struct btrfs_delayed_ref_node *node)
 {
-	WARN_ON(btrfs_delayed_ref_is_head(node));
 	return container_of(node, struct btrfs_delayed_data_ref, node);
 }
-
-static inline struct btrfs_delayed_ref_head *
-btrfs_delayed_node_to_head(struct btrfs_delayed_ref_node *node)
-{
-	WARN_ON(!btrfs_delayed_ref_is_head(node));
-	return container_of(node, struct btrfs_delayed_ref_head, node);
-}
 #endif
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index f971d5680e6f..69ce738c00d0 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -4121,12 +4121,12 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 		head = rb_entry(node, struct btrfs_delayed_ref_head,
 				href_node);
 		if (!mutex_trylock(&head->mutex)) {
-			refcount_inc(&head->node.refs);
+			refcount_inc(&head->refs);
 			spin_unlock(&delayed_refs->lock);
 
 			mutex_lock(&head->mutex);
 			mutex_unlock(&head->mutex);
-			btrfs_put_delayed_ref(&head->node);
+			btrfs_put_delayed_ref_head(head);
 			spin_lock(&delayed_refs->lock);
 			continue;
 		}
@@ -4147,16 +4147,16 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 		if (head->processing == 0)
 			delayed_refs->num_heads_ready--;
 		atomic_dec(&delayed_refs->num_entries);
-		head->node.in_tree = 0;
 		rb_erase(&head->href_node, &delayed_refs->href_root);
+		RB_CLEAR_NODE(&head->href_node);
 		spin_unlock(&head->lock);
 		spin_unlock(&delayed_refs->lock);
 		mutex_unlock(&head->mutex);
 
 		if (pin_bytes)
-			btrfs_pin_extent(fs_info, head->node.bytenr,
-					 head->node.num_bytes, 1);
-		btrfs_put_delayed_ref(&head->node);
+			btrfs_pin_extent(fs_info, head->bytenr,
+					 head->num_bytes, 1);
+		btrfs_put_delayed_ref_head(head);
 		cond_resched();
 		spin_lock(&delayed_refs->lock);
 	}
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 628ae71094f2..423d89145bac 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -912,7 +912,7 @@ search_again:
 	head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
 	if (head) {
 		if (!mutex_trylock(&head->mutex)) {
-			refcount_inc(&head->node.refs);
+			refcount_inc(&head->refs);
 			spin_unlock(&delayed_refs->lock);
 
 			btrfs_release_path(path);
@@ -923,7 +923,7 @@ search_again:
 			 */
 			mutex_lock(&head->mutex);
 			mutex_unlock(&head->mutex);
-			btrfs_put_delayed_ref(&head->node);
+			btrfs_put_delayed_ref_head(head);
 			goto search_again;
 		}
 		spin_lock(&head->lock);
@@ -932,7 +932,7 @@ search_again:
 		else
 			BUG_ON(num_refs == 0);
 
-		num_refs += head->node.ref_mod;
+		num_refs += head->ref_mod;
 		spin_unlock(&head->lock);
 		mutex_unlock(&head->mutex);
 	}
@@ -2337,7 +2337,7 @@ static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
 
 static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 				 struct btrfs_fs_info *fs_info,
-				 struct btrfs_delayed_ref_node *node,
+				 struct btrfs_delayed_ref_head *head,
 				 struct btrfs_delayed_extent_op *extent_op)
 {
 	struct btrfs_key key;
@@ -2359,14 +2359,14 @@ static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 	if (!path)
 		return -ENOMEM;
 
-	key.objectid = node->bytenr;
+	key.objectid = head->bytenr;
 
 	if (metadata) {
 		key.type = BTRFS_METADATA_ITEM_KEY;
 		key.offset = extent_op->level;
 	} else {
 		key.type = BTRFS_EXTENT_ITEM_KEY;
-		key.offset = node->num_bytes;
+		key.offset = head->num_bytes;
 	}
 
 again:
@@ -2383,17 +2383,17 @@ again:
 				path->slots[0]--;
 				btrfs_item_key_to_cpu(path->nodes[0], &key,
 						      path->slots[0]);
-				if (key.objectid == node->bytenr &&
+				if (key.objectid == head->bytenr &&
 				    key.type == BTRFS_EXTENT_ITEM_KEY &&
-				    key.offset == node->num_bytes)
+				    key.offset == head->num_bytes)
 					ret = 0;
 			}
 			if (ret > 0) {
 				btrfs_release_path(path);
 				metadata = 0;
 
-				key.objectid = node->bytenr;
-				key.offset = node->num_bytes;
+				key.objectid = head->bytenr;
+				key.offset = head->num_bytes;
 				key.type = BTRFS_EXTENT_ITEM_KEY;
 				goto again;
 			}
@@ -2562,7 +2562,7 @@ static int cleanup_extent_op(struct btrfs_trans_handle *trans,
 		return 0;
 	}
 	spin_unlock(&head->lock);
-	ret = run_delayed_extent_op(trans, fs_info, &head->node, extent_op);
+	ret = run_delayed_extent_op(trans, fs_info, head, extent_op);
 	btrfs_free_delayed_extent_op(extent_op);
 	return ret ? ret : 1;
 }
@@ -2597,39 +2597,37 @@ static int cleanup_ref_head(struct btrfs_trans_handle *trans,
 		spin_unlock(&delayed_refs->lock);
 		return 1;
 	}
-	head->node.in_tree = 0;
 	delayed_refs->num_heads--;
 	rb_erase(&head->href_node, &delayed_refs->href_root);
+	RB_CLEAR_NODE(&head->href_node);
 	spin_unlock(&delayed_refs->lock);
 	spin_unlock(&head->lock);
 	atomic_dec(&delayed_refs->num_entries);
 
-	trace_run_delayed_ref_head(fs_info, &head->node, head,
-				   head->node.action);
+	trace_run_delayed_ref_head(fs_info, head, 0);
 
 	if (head->total_ref_mod < 0) {
 		struct btrfs_block_group_cache *cache;
 
-		cache = btrfs_lookup_block_group(fs_info, head->node.bytenr);
+		cache = btrfs_lookup_block_group(fs_info, head->bytenr);
 		ASSERT(cache);
 		percpu_counter_add(&cache->space_info->total_bytes_pinned,
-				   -head->node.num_bytes);
+				   -head->num_bytes);
 		btrfs_put_block_group(cache);
 
 		if (head->is_data) {
 			spin_lock(&delayed_refs->lock);
-			delayed_refs->pending_csums -= head->node.num_bytes;
+			delayed_refs->pending_csums -= head->num_bytes;
 			spin_unlock(&delayed_refs->lock);
 		}
 	}
 
 	if (head->must_insert_reserved) {
-		btrfs_pin_extent(fs_info, head->node.bytenr,
-				 head->node.num_bytes, 1);
+		btrfs_pin_extent(fs_info, head->bytenr,
+				 head->num_bytes, 1);
 		if (head->is_data) {
-			ret = btrfs_del_csums(trans, fs_info,
-					      head->node.bytenr,
-					      head->node.num_bytes);
+			ret = btrfs_del_csums(trans, fs_info, head->bytenr,
+					      head->num_bytes);
 		}
 	}
 
@@ -2637,7 +2635,7 @@ static int cleanup_ref_head(struct btrfs_trans_handle *trans,
 	btrfs_qgroup_free_delayed_ref(fs_info, head->qgroup_ref_root,
 				      head->qgroup_reserved);
 	btrfs_delayed_ref_unlock(head);
-	btrfs_put_delayed_ref(&head->node);
+	btrfs_put_delayed_ref_head(head);
 	return 0;
 }
 
@@ -2751,10 +2749,10 @@ static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 		switch (ref->action) {
 		case BTRFS_ADD_DELAYED_REF:
 		case BTRFS_ADD_DELAYED_EXTENT:
-			locked_ref->node.ref_mod -= ref->ref_mod;
+			locked_ref->ref_mod -= ref->ref_mod;
 			break;
 		case BTRFS_DROP_DELAYED_REF:
-			locked_ref->node.ref_mod += ref->ref_mod;
+			locked_ref->ref_mod += ref->ref_mod;
 			break;
 		default:
 			WARN_ON(1);
@@ -3087,33 +3085,16 @@ again:
 			spin_unlock(&delayed_refs->lock);
 			goto out;
 		}
+		head = rb_entry(node, struct btrfs_delayed_ref_head,
+				href_node);
+		refcount_inc(&head->refs);
+		spin_unlock(&delayed_refs->lock);
 
-		while (node) {
-			head = rb_entry(node, struct btrfs_delayed_ref_head,
-					href_node);
-			if (btrfs_delayed_ref_is_head(&head->node)) {
-				struct btrfs_delayed_ref_node *ref;
-
-				ref = &head->node;
-				refcount_inc(&ref->refs);
-
-				spin_unlock(&delayed_refs->lock);
-				/*
-				 * Mutex was contended, block until it's
-				 * released and try again
-				 */
-				mutex_lock(&head->mutex);
-				mutex_unlock(&head->mutex);
+		/* Mutex was contended, block until it's released and retry. */
+		mutex_lock(&head->mutex);
+		mutex_unlock(&head->mutex);
 
-				btrfs_put_delayed_ref(ref);
-				cond_resched();
-				goto again;
-			} else {
-				WARN_ON(1);
-			}
-			node = rb_next(node);
-		}
-		spin_unlock(&delayed_refs->lock);
+		btrfs_put_delayed_ref_head(head);
 		cond_resched();
 		goto again;
 	}
@@ -3171,7 +3152,7 @@ static noinline int check_delayed_ref(struct btrfs_root *root,
 	}
 
 	if (!mutex_trylock(&head->mutex)) {
-		refcount_inc(&head->node.refs);
+		refcount_inc(&head->refs);
 		spin_unlock(&delayed_refs->lock);
 
 		btrfs_release_path(path);
@@ -3182,7 +3163,7 @@ static noinline int check_delayed_ref(struct btrfs_root *root,
 		 */
 		mutex_lock(&head->mutex);
 		mutex_unlock(&head->mutex);
-		btrfs_put_delayed_ref(&head->node);
+		btrfs_put_delayed_ref_head(head);
 		return -EAGAIN;
 	}
 	spin_unlock(&delayed_refs->lock);
@@ -7235,9 +7216,8 @@ static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans,
 	 * at this point we have a head with no other entries.  Go
 	 * ahead and process it.
 	 */
-	head->node.in_tree = 0;
 	rb_erase(&head->href_node, &delayed_refs->href_root);
-
+	RB_CLEAR_NODE(&head->href_node);
 	atomic_dec(&delayed_refs->num_entries);
 
 	/*
@@ -7256,7 +7236,7 @@ static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans,
 		ret = 1;
 
 	mutex_unlock(&head->mutex);
-	btrfs_put_delayed_ref(&head->node);
+	btrfs_put_delayed_ref_head(head);
 	return ret;
 out:
 	spin_unlock(&head->lock);
diff --git a/include/trace/events/btrfs.h b/include/trace/events/btrfs.h
index 77437f545c63..bfe2f23b578c 100644
--- a/include/trace/events/btrfs.h
+++ b/include/trace/events/btrfs.h
@@ -798,11 +798,10 @@ DEFINE_EVENT(btrfs_delayed_data_ref,  run_delayed_data_ref,
 DECLARE_EVENT_CLASS(btrfs_delayed_ref_head,
 
 	TP_PROTO(const struct btrfs_fs_info *fs_info,
-		 const struct btrfs_delayed_ref_node *ref,
 		 const struct btrfs_delayed_ref_head *head_ref,
 		 int action),
 
-	TP_ARGS(fs_info, ref, head_ref, action),
+	TP_ARGS(fs_info, head_ref, action),
 
 	TP_STRUCT__entry_btrfs(
 		__field(	u64,  bytenr		)
@@ -812,8 +811,8 @@ DECLARE_EVENT_CLASS(btrfs_delayed_ref_head,
 	),
 
 	TP_fast_assign_btrfs(fs_info,
-		__entry->bytenr		= ref->bytenr;
-		__entry->num_bytes	= ref->num_bytes;
+		__entry->bytenr		= head_ref->bytenr;
+		__entry->num_bytes	= head_ref->num_bytes;
 		__entry->action		= action;
 		__entry->is_data	= head_ref->is_data;
 	),
@@ -828,21 +827,19 @@ DECLARE_EVENT_CLASS(btrfs_delayed_ref_head,
 DEFINE_EVENT(btrfs_delayed_ref_head,  add_delayed_ref_head,
 
 	TP_PROTO(const struct btrfs_fs_info *fs_info,
-		 const struct btrfs_delayed_ref_node *ref,
 		 const struct btrfs_delayed_ref_head *head_ref,
 		 int action),
 
-	TP_ARGS(fs_info, ref, head_ref, action)
+	TP_ARGS(fs_info, head_ref, action)
 );
 
 DEFINE_EVENT(btrfs_delayed_ref_head,  run_delayed_ref_head,
 
 	TP_PROTO(const struct btrfs_fs_info *fs_info,
-		 const struct btrfs_delayed_ref_node *ref,
 		 const struct btrfs_delayed_ref_head *head_ref,
 		 int action),
 
-	TP_ARGS(fs_info, ref, head_ref, action)
+	TP_ARGS(fs_info, head_ref, action)
 );
 
 #define show_chunk_type(type)					\