Subject: Re: 1.6.1 stability problems?
To: Steven M. Bellovin <smb@research.att.com>
From: David Maxwell <david@vex.net>
List: netbsd-users
Date: 02/07/2003 20:08:57
On Fri, Feb 07, 2003 at 06:47:55PM -0500, Steven M. Bellovin wrote:
> >could you provide a backtrace of a crash? Otherwise, I'm afraid, this
> 
> I wish I could, but I don't know how to get one.  The machine locks up 
> hard, and the only way I know to reset it -- this is a laptop -- is to 
> lean on the power button for several seconds.  ctl+alt+esc doesn't work 
> at that point, as best I can tell, though since I'm in X I'm not 100% 

I have a NEC Versa LX running a 'few days old' -current. I have some
similar issues...

Cardbus activities won't hang the machine, but sometimes the card isn't
recognized and the network interface (tlp0) doesn't appear.

Using apm -S will take the machine to a light sleep that I can recover
from.

Using apm -z will suspend the machine, but when I wake it up, the first
disk access will hang the machine, with a couple ATA timeout and DMA
errors, after which the hang is complete - can't get to ddb anymore.

With some tips from Manuel, I single-stepped after waking the machine,
and found that the hang happens on an inb instruction, which I believe
is part of the bus_space_read_1() in ic/wdc.c/wdcwait() - it seems like
my pciide controller is not reset properly after the suspend.

When the machine hangs, the HD light is on solid as well. 

Does any of that match what you've seen?

-- 
David Maxwell, david@vex.net|david@maxwell.net -->
Any sufficiently advanced Common Sense will seem like magic... 
					      - me