callout_reset page fault panic

[Scott Long]

John Baldwin wrote:
> On Wednesday 24 May 2006 16:22, Pawel Worach wrote:
> 
>>John Baldwin wrote:
>>
>>>On Saturday 20 May 2006 09:22, Pawel Worach wrote:
>>>
>>>>One day old CURRENT, i368 UP. Died while installing some ports and 
>>>>running mplayer. vmcore and kernel available (minidumps kick ass!).
>>>>
>>>>Unread portion of the kernel message buffer:
>>>>kernel trap 12 with interrupts disabled
>>>
>>>I think the previous 'struct callout' on the callout wheel
>>>is unmapped, maybe a kernel module was unloaded and it forgot
>>>to do a callout_stop() or callout_drain() or somehow did a
>>>callout_reset() after doing the stop() or drain()?
>>>
>>
>>If nothing was unloaded automagically it's unlikely.
> 
> 
> It could have been free'd perhaps and the page later subsequently unmapped 
> from KVA?  (Not sure if that can happen.)
> 

That's how these kind of panics usually happen, actually.

Scott