I still don't understand why we even need b_error on the submission side, we should be able to just pass the error on the stack. But we can sort this out later, and the change looks ok, so: Reviewed-by: Christoph Hellwig <hch at lst.de>