When you say " Running lqs2mem on the original suspend to disk image does not work", do you mean that you're getting an error? Or that it's creating an image that doesn't work in volatility? I've ran lqs2mem literally on hundreds of QEMU images with no problems. Can you post the output of your run? If I recall correctly, Juerg had to pad a certain section of memory in order to get the structures to line up. It's possible that later versions of QEMU/KVM changed so that padding isn't necessary any more.