Re: panic at boot on 4/200

[foo bar <tokenalt%gmail.com@localhost>]

On Wed, Jan 7, 2026 at 4:23 PM David Brownlee <abs%absd.org@localhost> wrote:
>
> On Sun, 4 Jan 2026 at 17:06, foo bar <tokenalt%gmail.com@localhost> wrote:
> >
> > On Wed, Dec 31, 2025 at 10:06 PM foo bar <tokenalt%gmail.com@localhost> wrote:
> > >
> > > Hello all
> > >
> > > I have recently gotten my 3200 & 4200 CPU boards working in a 4/110
> > > case and while the 3200 works fine the 4200 does not. The kernel I'm
> > > booting on the 4200 will boot fine on a 4300 but on the 4200 it panics
> > > with mutex errors. I don't know if this is bad hardware or if there's
> > > something different between the two CPU boards. Unfortunately the
> > > ethernet does not work on the 4200 board so testing new kernels is
> > > slow.
> >
> > So after doing many test boots I've found netbsd 1.5.x will boot on it
> > but everything else, including sunos 4.1.1, will fail in various ways.
> > The fact that even sunos doesn't work on it leads me to suspect that
> > this is bad hardware. A shame since these boards are getting rare
> > these days. If anyone has suggestions I'm all ears and sorry for the
> > noise.
>
> (Some random questions)
>
> Could it be bad memory or an issue with one of the connections (uh P2?) ?
>
> Does the PROM have a real memory tester? If you can boot NetBSD-1.5 on
> it I'm assuming a userland memory tester would work.. or even compile
> a custom version of boot which just allocates and tests all memory?
> Might be able to compile a current kernel with a chunk of unused
> memory at one end to move things around...
>
> For curiosity you could test various current kernels between 1.5 and
> 1.6 to find out where something broke. Without a working network how
> are you getting kernels across - something like a blue scsi?
>
> Actually, if the ethernet is bad could it potentially be causing the
> issue - panic on attempting to attach it?
>
> I'm not familiar with the 3200 board - could it be running the
> bus/memory slower and avoiding the issue?
>
> Good hunting if you are still inclined :)

I'm positive the memory board is good. Not only do the CPU boards have
a thorough memory test in diagnostics mode but I've also used the
memory board with the 3200 CPU board for many hours running SunOS and
NetBSD. I am actually suspecting it's the cache that's bad as all the
kernels that have problems suddenly start getting bad data right
around when the cache is enabled. But if it is the cache presumably
that would also affect 1.5.x so I don't know.

I'm booting from a Sun-3 SCSI HBA connected to a zuluscsi.

The ethernet doesn't work I believe because of some corrosion on the
board in that area. The source of the corrosion appears to actually be
the double sided tape Sun used to hold the 20MHz crystal to the board.
I have a Sun3/E SCSI/Ethernet board that also has corrosion around the
20MHz crystal so anyone out there with VME boards check for corrosion
around that crystal.

I will do some more tests on it in a few weeks but first I have some
other hardware to work on.