Btrfs: kill BUG_ON in run_delayed_tree_ref
authorLiu Bo <bo.li.liu@oracle.com>
Thu, 15 Sep 2016 02:19:05 +0000 (19:19 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 16 Jan 2019 21:16:07 +0000 (22:16 +0100)
commit 02794222c4132ac003e7281fb71f4ec1645ffc87 upstream.

In a corrupted btrfs image, we can come across this BUG_ON and
get an unreponsive system, but if we return errors instead,
its caller can handle everything gracefully by aborting the current
transaction.

Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fs/btrfs/extent-tree.c

index abfd232..1454938 100644 (file)
@@ -2342,7 +2342,13 @@ static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
                ins.type = BTRFS_EXTENT_ITEM_KEY;
        }
 
-       BUG_ON(node->ref_mod != 1);
+       if (node->ref_mod != 1) {
+               btrfs_err(root->fs_info,
+       "btree block(%llu) has %d references rather than 1: action %d ref_root %llu parent %llu",
+                         node->bytenr, node->ref_mod, node->action, ref_root,
+                         parent);
+               return -EIO;
+       }
        if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
                BUG_ON(!extent_op || !extent_op->update_flags);
                ret = alloc_reserved_tree_block(trans, root,