On Fri, Apr 23, 2004 at  8:38am, Christopher SEKIYA wrote:
> Subject: Re: panic: cache error -- on R5k INDY

Heyya Chris... How'z been?

> Math time :)
> 
> On Thu, Apr 22, 2004 at 01:50:48PM -0500, sgimips NetBSD list wrote:
> 
> > hpc0 at gio0 addr 0x1fb80000: SGI HPC3
> 
> (snip)
> 
> > dsclock0 at hpc0 offset 0x60000
> 
> (snip)
> 
> > bus error: cpu_stat 00000203 addr 0887fe40, gio_stat 00000000 addr 1fbc4003
> > panic: cache error @ EPC 0x882b0010 ErrCtl 0x3 CacheErr 0xa01934f3

Interesting... dsclock.  That'll be a neat comparison to the latest note below.

> Does the panic happen immediately after that bus error?

It happens right away... i.e. the scrolling  of the messages is immediate.

> Could you try a 2.0-branch kernel and see if it blows up?  Maybe the IP12 code
> is interfering with something ...

These are MAIN branch kernels.   You mean cvs pull from the netbsd-2-0 branch and
build a kernel from there?  If so, you know the proper cvs command to pull that? :)
I thought the default MAIN was already pretty much netbsd-2-0.

> (the mc bus_error handler should be dealing with the bus error, though)

Unh huh :)

One more note to make things interesting... I can ssh as a regular user into
the machine, then su root... and it seems fine.   I can't ssh in as root or
the machine slams to a halt, and sits there until the watchdog resets the
hardware back to ARCS.

Pfun stough :-D

-scott