1 files changed, 122 insertions, 250 deletions

diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index 8f8ed7d20bac..ac3e81da6d4e 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -22,6 +22,7 @@
 #include "ctree.h"
 #include "delayed-ref.h"
 #include "transaction.h"
+#include "qgroup.h"
 
 struct kmem_cache *btrfs_delayed_ref_head_cachep;
 struct kmem_cache *btrfs_delayed_tree_ref_cachep;
@@ -84,87 +85,6 @@ static int comp_data_refs(struct btrfs_delayed_data_ref *ref2,
 	return 0;
 }
 
-/*
- * entries in the rb tree are ordered by the byte number of the extent,
- * type of the delayed backrefs and content of delayed backrefs.
- */
-static int comp_entry(struct btrfs_delayed_ref_node *ref2,
-		      struct btrfs_delayed_ref_node *ref1,
-		      bool compare_seq)
-{
-	if (ref1->bytenr < ref2->bytenr)
-		return -1;
-	if (ref1->bytenr > ref2->bytenr)
-		return 1;
-	if (ref1->is_head && ref2->is_head)
-		return 0;
-	if (ref2->is_head)
-		return -1;
-	if (ref1->is_head)
-		return 1;
-	if (ref1->type < ref2->type)
-		return -1;
-	if (ref1->type > ref2->type)
-		return 1;
-	if (ref1->no_quota > ref2->no_quota)
-		return 1;
-	if (ref1->no_quota < ref2->no_quota)
-		return -1;
-	/* merging of sequenced refs is not allowed */
-	if (compare_seq) {
-		if (ref1->seq < ref2->seq)
-			return -1;
-		if (ref1->seq > ref2->seq)
-			return 1;
-	}
-	if (ref1->type == BTRFS_TREE_BLOCK_REF_KEY ||
-	    ref1->type == BTRFS_SHARED_BLOCK_REF_KEY) {
-		return comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref2),
-				      btrfs_delayed_node_to_tree_ref(ref1),
-				      ref1->type);
-	} else if (ref1->type == BTRFS_EXTENT_DATA_REF_KEY ||
-		   ref1->type == BTRFS_SHARED_DATA_REF_KEY) {
-		return comp_data_refs(btrfs_delayed_node_to_data_ref(ref2),
-				      btrfs_delayed_node_to_data_ref(ref1));
-	}
-	BUG();
-	return 0;
-}
-
-/*
- * insert a new ref into the rbtree.  This returns any existing refs
- * for the same (bytenr,parent) tuple, or NULL if the new node was properly
- * inserted.
- */
-static struct btrfs_delayed_ref_node *tree_insert(struct rb_root *root,
-						  struct rb_node *node)
-{
-	struct rb_node **p = &root->rb_node;
-	struct rb_node *parent_node = NULL;
-	struct btrfs_delayed_ref_node *entry;
-	struct btrfs_delayed_ref_node *ins;
-	int cmp;
-
-	ins = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
-	while (*p) {
-		parent_node = *p;
-		entry = rb_entry(parent_node, struct btrfs_delayed_ref_node,
-				 rb_node);
-
-		cmp = comp_entry(entry, ins, 1);
-		if (cmp < 0)
-			p = &(*p)->rb_left;
-		else if (cmp > 0)
-			p = &(*p)->rb_right;
-		else
-			return entry;
-	}
-
-	rb_link_node(node, parent_node, p);
-	rb_insert_color(node, root);
-	return NULL;
-}
-
 /* insert a new ref to head ref rbtree */
 static struct btrfs_delayed_ref_head *htree_insert(struct rb_root *root,
 						   struct rb_node *node)
@@ -268,7 +188,7 @@ static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
 		rb_erase(&head->href_node, &delayed_refs->href_root);
 	} else {
 		assert_spin_locked(&head->lock);
-		rb_erase(&ref->rb_node, &head->ref_root);
+		list_del(&ref->list);
 	}
 	ref->in_tree = 0;
 	btrfs_put_delayed_ref(ref);
@@ -277,99 +197,6 @@ static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
 		trans->delayed_ref_updates--;
 }
 
-static int merge_ref(struct btrfs_trans_handle *trans,
-		     struct btrfs_delayed_ref_root *delayed_refs,
-		     struct btrfs_delayed_ref_head *head,
-		     struct btrfs_delayed_ref_node *ref, u64 seq)
-{
-	struct rb_node *node;
-	int mod = 0;
-	int done = 0;
-
-	node = rb_next(&ref->rb_node);
-	while (!done && node) {
-		struct btrfs_delayed_ref_node *next;
-
-		next = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
-		node = rb_next(node);
-		if (seq && next->seq >= seq)
-			break;
-		if (comp_entry(ref, next, 0))
-			continue;
-
-		if (ref->action == next->action) {
-			mod = next->ref_mod;
-		} else {
-			if (ref->ref_mod < next->ref_mod) {
-				struct btrfs_delayed_ref_node *tmp;
-
-				tmp = ref;
-				ref = next;
-				next = tmp;
-				done = 1;
-			}
-			mod = -next->ref_mod;
-		}
-
-		drop_delayed_ref(trans, delayed_refs, head, next);
-		ref->ref_mod += mod;
-		if (ref->ref_mod == 0) {
-			drop_delayed_ref(trans, delayed_refs, head, ref);
-			done = 1;
-		} else {
-			/*
-			 * You can't have multiples of the same ref on a tree
-			 * block.
-			 */
-			WARN_ON(ref->type == BTRFS_TREE_BLOCK_REF_KEY ||
-				ref->type == BTRFS_SHARED_BLOCK_REF_KEY);
-		}
-	}
-	return done;
-}
-
-void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
-			      struct btrfs_fs_info *fs_info,
-			      struct btrfs_delayed_ref_root *delayed_refs,
-			      struct btrfs_delayed_ref_head *head)
-{
-	struct rb_node *node;
-	u64 seq = 0;
-
-	assert_spin_locked(&head->lock);
-	/*
-	 * We don't have too much refs to merge in the case of delayed data
-	 * refs.
-	 */
-	if (head->is_data)
-		return;
-
-	spin_lock(&fs_info->tree_mod_seq_lock);
-	if (!list_empty(&fs_info->tree_mod_seq_list)) {
-		struct seq_list *elem;
-
-		elem = list_first_entry(&fs_info->tree_mod_seq_list,
-					struct seq_list, list);
-		seq = elem->seq;
-	}
-	spin_unlock(&fs_info->tree_mod_seq_lock);
-
-	node = rb_first(&head->ref_root);
-	while (node) {
-		struct btrfs_delayed_ref_node *ref;
-
-		ref = rb_entry(node, struct btrfs_delayed_ref_node,
-			       rb_node);
-		/* We can't merge refs that are outside of our seq count */
-		if (seq && ref->seq >= seq)
-			break;
-		if (merge_ref(trans, delayed_refs, head, ref, seq))
-			node = rb_first(&head->ref_root);
-		else
-			node = rb_next(&ref->rb_node);
-	}
-}
-
 int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info,
 			    struct btrfs_delayed_ref_root *delayed_refs,
 			    u64 seq)
@@ -443,45 +270,71 @@ again:
 }
 
 /*
- * helper function to update an extent delayed ref in the
- * rbtree.  existing and update must both have the same
- * bytenr and parent
+ * Helper to insert the ref_node to the tail or merge with tail.
  *
- * This may free existing if the update cancels out whatever
- * operation it was doing.
+ * Return 0 for insert.
+ * Return >0 for merge.
  */
-static noinline void
-update_existing_ref(struct btrfs_trans_handle *trans,
-		    struct btrfs_delayed_ref_root *delayed_refs,
-		    struct btrfs_delayed_ref_head *head,
-		    struct btrfs_delayed_ref_node *existing,
-		    struct btrfs_delayed_ref_node *update)
+static int
+add_delayed_ref_tail_merge(struct btrfs_trans_handle *trans,
+			   struct btrfs_delayed_ref_root *root,
+			   struct btrfs_delayed_ref_head *href,
+			   struct btrfs_delayed_ref_node *ref)
 {
-	if (update->action != existing->action) {
-		/*
-		 * this is effectively undoing either an add or a
-		 * drop.  We decrement the ref_mod, and if it goes
-		 * down to zero we just delete the entry without
-		 * every changing the extent allocation tree.
-		 */
-		existing->ref_mod--;
-		if (existing->ref_mod == 0)
-			drop_delayed_ref(trans, delayed_refs, head, existing);
-		else
-			WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY ||
-				existing->type == BTRFS_SHARED_BLOCK_REF_KEY);
+	struct btrfs_delayed_ref_node *exist;
+	int mod;
+	int ret = 0;
+
+	spin_lock(&href->lock);
+	/* Check whether we can merge the tail node with ref */
+	if (list_empty(&href->ref_list))
+		goto add_tail;
+	exist = list_entry(href->ref_list.prev, struct btrfs_delayed_ref_node,
+			   list);
+	/* No need to compare bytenr nor is_head */
+	if (exist->type != ref->type || exist->no_quota != ref->no_quota ||
+	    exist->seq != ref->seq)
+		goto add_tail;
+
+	if ((exist->type == BTRFS_TREE_BLOCK_REF_KEY ||
+	     exist->type == BTRFS_SHARED_BLOCK_REF_KEY) &&
+	    comp_tree_refs(btrfs_delayed_node_to_tree_ref(exist),
+			   btrfs_delayed_node_to_tree_ref(ref),
+			   ref->type))
+		goto add_tail;
+	if ((exist->type == BTRFS_EXTENT_DATA_REF_KEY ||
+	     exist->type == BTRFS_SHARED_DATA_REF_KEY) &&
+	    comp_data_refs(btrfs_delayed_node_to_data_ref(exist),
+			   btrfs_delayed_node_to_data_ref(ref)))
+		goto add_tail;
+
+	/* Now we are sure we can merge */
+	ret = 1;
+	if (exist->action == ref->action) {
+		mod = ref->ref_mod;
 	} else {
-		WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY ||
-			existing->type == BTRFS_SHARED_BLOCK_REF_KEY);
-		/*
-		 * the action on the existing ref matches
-		 * the action on the ref we're trying to add.
-		 * Bump the ref_mod by one so the backref that
-		 * is eventually added/removed has the correct
-		 * reference count
-		 */
-		existing->ref_mod += update->ref_mod;
+		/* Need to change action */
+		if (exist->ref_mod < ref->ref_mod) {
+			exist->action = ref->action;
+			mod = -exist->ref_mod;
+			exist->ref_mod = ref->ref_mod;
+		} else
+			mod = -ref->ref_mod;
 	}
+	exist->ref_mod += mod;
+
+	/* remove existing tail if its ref_mod is zero */
+	if (exist->ref_mod == 0)
+		drop_delayed_ref(trans, root, href, exist);
+	spin_unlock(&href->lock);
+	return ret;
+
+add_tail:
+	list_add_tail(&ref->list, &href->ref_list);
+	atomic_inc(&root->num_entries);
+	trans->delayed_ref_updates++;
+	spin_unlock(&href->lock);
+	return ret;
 }
 
 /*
@@ -568,12 +421,14 @@ 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, u64 bytenr,
-		     u64 num_bytes, int action, int is_data)
+		     struct btrfs_delayed_ref_node *ref,
+		     struct btrfs_qgroup_extent_record *qrecord,
+		     u64 bytenr, u64 num_bytes, int action, int is_data)
 {
 	struct btrfs_delayed_ref_head *existing;
 	struct btrfs_delayed_ref_head *head_ref = NULL;
 	struct btrfs_delayed_ref_root *delayed_refs;
+	struct btrfs_qgroup_extent_record *qexisting;
 	int count_mod = 1;
 	int must_insert_reserved = 0;
 
@@ -618,10 +473,22 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 	head_ref = btrfs_delayed_node_to_head(ref);
 	head_ref->must_insert_reserved = must_insert_reserved;
 	head_ref->is_data = is_data;
-	head_ref->ref_root = RB_ROOT;
+	INIT_LIST_HEAD(&head_ref->ref_list);
 	head_ref->processing = 0;
 	head_ref->total_ref_mod = count_mod;
 
+	/* Record qgroup extent info if provided */
+	if (qrecord) {
+		qrecord->bytenr = bytenr;
+		qrecord->num_bytes = num_bytes;
+		qrecord->old_roots = NULL;
+
+		qexisting = btrfs_qgroup_insert_dirty_extent(delayed_refs,
+							     qrecord);
+		if (qexisting)
+			kfree(qrecord);
+	}
+
 	spin_lock_init(&head_ref->lock);
 	mutex_init(&head_ref->mutex);
 
@@ -659,10 +526,10 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 		     u64 num_bytes, u64 parent, u64 ref_root, int level,
 		     int action, int no_quota)
 {
-	struct btrfs_delayed_ref_node *existing;
 	struct btrfs_delayed_tree_ref *full_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	u64 seq = 0;
+	int ret;
 
 	if (action == BTRFS_ADD_DELAYED_EXTENT)
 		action = BTRFS_ADD_DELAYED_REF;
@@ -693,21 +560,14 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 
 	trace_add_delayed_tree_ref(ref, full_ref, action);
 
-	spin_lock(&head_ref->lock);
-	existing = tree_insert(&head_ref->ref_root, &ref->rb_node);
-	if (existing) {
-		update_existing_ref(trans, delayed_refs, head_ref, existing,
-				    ref);
-		/*
-		 * we've updated the existing ref, free the newly
-		 * allocated ref
-		 */
+	ret = add_delayed_ref_tail_merge(trans, delayed_refs, head_ref, ref);
+
+	/*
+	 * XXX: memory should be freed at the same level allocated.
+	 * But bad practice is anywhere... Follow it now. Need cleanup.
+	 */
+	if (ret > 0)
 		kmem_cache_free(btrfs_delayed_tree_ref_cachep, full_ref);
-	} else {
-		atomic_inc(&delayed_refs->num_entries);
-		trans->delayed_ref_updates++;
-	}
-	spin_unlock(&head_ref->lock);
 }
 
 /*
@@ -721,10 +581,10 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 		     u64 num_bytes, u64 parent, u64 ref_root, u64 owner,
 		     u64 offset, int action, int no_quota)
 {
-	struct btrfs_delayed_ref_node *existing;
 	struct btrfs_delayed_data_ref *full_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	u64 seq = 0;
+	int ret;
 
 	if (action == BTRFS_ADD_DELAYED_EXTENT)
 		action = BTRFS_ADD_DELAYED_REF;
@@ -758,21 +618,10 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 
 	trace_add_delayed_data_ref(ref, full_ref, action);
 
-	spin_lock(&head_ref->lock);
-	existing = tree_insert(&head_ref->ref_root, &ref->rb_node);
-	if (existing) {
-		update_existing_ref(trans, delayed_refs, head_ref, existing,
-				    ref);
-		/*
-		 * we've updated the existing ref, free the newly
-		 * allocated ref
-		 */
+	ret = add_delayed_ref_tail_merge(trans, delayed_refs, head_ref, ref);
+
+	if (ret > 0)
 		kmem_cache_free(btrfs_delayed_data_ref_cachep, full_ref);
-	} else {
-		atomic_inc(&delayed_refs->num_entries);
-		trans->delayed_ref_updates++;
-	}
-	spin_unlock(&head_ref->lock);
 }
 
 /*
@@ -790,6 +639,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	struct btrfs_delayed_tree_ref *ref;
 	struct btrfs_delayed_ref_head *head_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
+	struct btrfs_qgroup_extent_record *record = NULL;
 
 	if (!is_fstree(ref_root) || !fs_info->quota_enabled)
 		no_quota = 0;
@@ -800,9 +650,13 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 		return -ENOMEM;
 
 	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
-	if (!head_ref) {
-		kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
-		return -ENOMEM;
+	if (!head_ref)
+		goto free_ref;
+
+	if (fs_info->quota_enabled && is_fstree(ref_root)) {
+		record = kmalloc(sizeof(*record), GFP_NOFS);
+		if (!record)
+			goto free_head_ref;
 	}
 
 	head_ref->extent_op = extent_op;
@@ -814,7 +668,7 @@ 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,
+	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
 					bytenr, num_bytes, action, 0);
 
 	add_delayed_tree_ref(fs_info, trans, head_ref, &ref->node, bytenr,
@@ -823,6 +677,13 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	spin_unlock(&delayed_refs->lock);
 
 	return 0;
+
+free_head_ref:
+	kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
+free_ref:
+	kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
+
+	return -ENOMEM;
 }
 
 /*
@@ -839,6 +700,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 	struct btrfs_delayed_data_ref *ref;
 	struct btrfs_delayed_ref_head *head_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
+	struct btrfs_qgroup_extent_record *record = NULL;
 
 	if (!is_fstree(ref_root) || !fs_info->quota_enabled)
 		no_quota = 0;
@@ -854,6 +716,16 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 		return -ENOMEM;
 	}
 
+	if (fs_info->quota_enabled && is_fstree(ref_root)) {
+		record = kmalloc(sizeof(*record), GFP_NOFS);
+		if (!record) {
+			kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
+			kmem_cache_free(btrfs_delayed_ref_head_cachep,
+					head_ref);
+			return -ENOMEM;
+		}
+	}
+
 	head_ref->extent_op = extent_op;
 
 	delayed_refs = &trans->transaction->delayed_refs;
@@ -863,7 +735,7 @@ 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,
+	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
 					bytenr, num_bytes, action, 1);
 
 	add_delayed_data_ref(fs_info, trans, head_ref, &ref->node, bytenr,
@@ -891,9 +763,9 @@ 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, bytenr,
-				   num_bytes, BTRFS_UPDATE_DELAYED_HEAD,
-				   extent_op->is_data);
+	add_delayed_ref_head(fs_info, trans, &head_ref->node, NULL, bytenr,
+			     num_bytes, BTRFS_UPDATE_DELAYED_HEAD,
+			     extent_op->is_data);
 
 	spin_unlock(&delayed_refs->lock);
 	return 0;