Re: big regression in ext4 performance?