Subject: Re: netbsd 3.0_STABLE instabilities
To: None <port-sparc@netbsd.org>
From: Andreas_Hallmann <hallmann@ahatec.de>
List: port-sparc
Date: 02/09/2006 17:48:07
Dear Bruce,

thankx but uwe's fix is already pulled up to the netbsd-3 branch.
It's made netbsd-3 useable on my SS1/SS2/SS4/SS5.

But this is a new issue.

So my question remains ...

`` How to get a back trace of a kernel unable to dump core ??? ``

Somebody must have done it, just a hint please

	Andreas

On Thu, Feb 09, 2006 at 12:56:44PM +0000, Bruce O'Neel wrote:
> 
> Hi,
> 
> I think the key fix was:
> 
> http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/sparc/sparc/cpu.c?f=h
> Vers 1.193
> 
> Compute c_totalsize for split caches correctly: it's max, not sum (and
> we even have that documented in cache.h).  Initialize c_nlines for
> split caches.
> 
> This should fix srmmu cache flush routines for split I/D caches that
> has been broken since cache.c revision 1.79 started using c_nlines.
> 
> ------------------------------------------------------------
> 
> This fix, at least for my MicrosparcII (in a SS4) fixed the random
> crashes.
> 
> My knowledge of a SS2 is pretty much non-existant, sorry.
> 
> 
> cheers
> 
> bruce
> 
> 
> On Wed, Feb 08, 2006 at 10:19:19PM +0100, Andreas_Hallmann wrote:
> > Hi all,
> > 
> > after upgrading my gateway host, an SS2, from 1.6.2 to netbsd 3.0 I have a few problems here.
> > 
> > Mostly I notice kernel crashes followed by a reboot, but lately I also got data access exceptions
> > followed by a watchdog reset.
> > 
> > This system is without any daught in perfect condition and run 1.6.2 without any unusual events
> > in the last years.
> > 
> > Other NetBSD 3.0 systems here, SS4 and 2 Cycle Turbo Sparcs, are also hit by random reboots.
> > But the problem seems to appear more likely under stress ( i.e a bulk build will do on the SS2)
> > but later 3.0_stable generic kernels are either fixed or much more stress resistent on turbo sparcs.
> > 
> > Unfortunately those reboots don't leave a core dump.
> > 
> > My big question:
> > 
> > (1) How to get at least a back-trace?
> > 
> > 	Andreas Hallmann
> > -- 
> > NetBSD: If you happen to have any problem with your uptime.
> > 
> 
> -- 
> edoneel@sdf.lonestar.org
> SDF Public Access UNIX System - http://sdf.lonestar.org
> 

-- 
NetBSD: If you happen to have any problem with your uptime.