[PATCH 09/12] btrfs: clear defragmented inodes using postorder in btrfs_cleanup_defrag_inodes()

[David Sterba <dsterba@suse.com>]

btrfs_cleanup_defrag_inodes() is not called frequently, only in remount
or unmount, but the way it frees the inodes in fs_info->defrag_inodes
is inefficient. Each time it needs to locate first node, remove it,
potentially rebalance tree until it's done. This allows to do a
conditional reschedule.

For cleanups the rbtree_postorder_for_each_entry_safe() iterator is
convenient but if the reschedule happens and unlocks fs_info->defrag_inodes_lock
we can't be sure that the tree is in the same state. If that happens,
restart the iteration from the beginning.

The cleanup operation is kmem_cache_free() which will likely take the
fast path for most objects.

Signed-off-by: David Sterba <dsterba@suse.com>
---
 fs/btrfs/defrag.c | 17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)

diff --git a/fs/btrfs/defrag.c b/fs/btrfs/defrag.c
index 41d67065d02b..bd1769dd134c 100644
--- a/fs/btrfs/defrag.c
+++ b/fs/btrfs/defrag.c
@@ -212,19 +212,18 @@ static struct inode_defrag *btrfs_pick_defrag_inode(
 
 void btrfs_cleanup_defrag_inodes(struct btrfs_fs_info *fs_info)
 {
-	struct inode_defrag *defrag;
-	struct rb_node *node;
+	struct inode_defrag *defrag, *next;
 
 	spin_lock(&fs_info->defrag_inodes_lock);
-	node = rb_first(&fs_info->defrag_inodes);
-	while (node) {
-		rb_erase(node, &fs_info->defrag_inodes);
-		defrag = rb_entry(node, struct inode_defrag, rb_node);
+again:
+	rbtree_postorder_for_each_entry_safe(defrag, next, &fs_info->defrag_inodes,
+					     rb_node) {
 		kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 
-		cond_resched_lock(&fs_info->defrag_inodes_lock);
-
-		node = rb_first(&fs_info->defrag_inodes);
+		if (cond_resched_lock(&fs_info->defrag_inodes_lock)) {
+			/* Rescheduled and unlocked. */
+			goto again;
+		}
 	}
 	spin_unlock(&fs_info->defrag_inodes_lock);
 }
-- 
2.45.0