Comment 43 for bug 1097213

Responding to comment #42

The crashes reported don't seem to have reproducible symptoms and are triggered by test code which is hotplugging cpus, so it seemed reasonable to hope cache coherency bugs in the cpu power down code might be the cause.

As regard to "most of time, there is no any panic pop up. just kernel hang without any output", I wasn't aware that this was what people are seeing, the only problems I'd seen and what Naresh and Basil's latest comments seem to be about, are crashes after repeatedly running the cpu_hotplug_latency_with{out}_load test.