Some tracing printk() show that we are calling update_sample() with totally bogus arguments.
There are a few good calls, then I see rmid=-380863112 evt_type=-30689 first=0 That turns into a wild vrmid, and we fault accessing mbm_current->prev_msr -Tony
