Comment 29 for bug 999755

I did repeat the crash on a local HVM guest with the current precise-proposed kernel. That way I was able to dump the crashed VM. A run with the same kernel in a different guest that had 4 CPUs did not crash for several hours. 2 CPUs seems to make the issue less likely and could be a reason this is not observed more often. Currently running the loop with a 3.4 mainline kernel and 2 CPUs.