Subject: Re: "NMI ... going to debugger" -- possible to store a backtrace?
To: Brian de Alwis <bsd@cs.ubc.ca>
From: Andrew Doran <ad@netbsd.org>
List: netbsd-users
Date: 07/06/2007 17:56:51
On Fri, Jul 06, 2007 at 10:46:01AM -0600, Brian de Alwis wrote:

> I've been dumped into the debugger about once a week because of
> messages like:
> 
> Jul  6 10:01:05 monolith /netbsd: NMI ... going to debugger
> 
> As I almost exclusively use X and have ddb.onpanic=0, I thought
> the machine had gone into a deep hang and would reboot.  I discovered
> this morning that it's in the debugger and could be resumed, and
> found the above message in /var/log/messages.
> 
> Is there anything I can do to get a trace?  Or rather, to dump the
> trace to a useful spot to look at later?

Not really, it (usually) indicates a serious hardware fault. You can get
information about the exception through the machine check registers, which
are processor dependent. For that to happen, someone first needs to write
the code. ;-)

Andrew