Subject: Re: 4.99.36 instability on amd64
To: Paul Goyette <>
From: Quentin Garnier <>
List: current-users
Date: 11/16/2007 02:30:35
On Thu, Nov 15, 2007 at 05:11:38PM -0800, Paul Goyette wrote:
> On Wed, 14 Nov 2007, Paul Goyette wrote:
> >I've just gotten the following crash:
> >
> >panic: kernel diagnostic assertion "amap == NULL || 
> >mutex_owned(&amap->amap_l)" failed: file "/usr/src/sys/uvm/uvm_fault.c", 
> >line 896
> >kernel: protection fault trap, code=0
> >
> >Unfortunately the keyboard (USB) was totally unresponsive at this point so 
> >I could not get a traceback.
> >
> >Nothing special was going on at the time, other than that I was trying to 
> >log out of an xterm (with display going to a remote X server).
> Hmmm, this seems to be almost repeatable!  Same circumstances as 
> yesterday - running an xterm with display pointing to a remote display 
> (which was "forwarded" via ssh -x); the local Xserver was not even 
> running at the time.  Today, I got a little bit more information on the 
> console:
> Stopped in pid 7904.1 (xterm) at netbsd:sme_events_worker + 0x111: andl 
> $-0x2, 0x9c(%r13)
> Still unable to get a backtrace since I don't have a legacy PS2 keyboard 
> attached (trying to find the antique beast in the bone pile).

You can get a core dump, though, if you tell ddb not to kick in on a
panic.  You can also configure ddb to print the trace when it enters.

Quentin Garnier