On Fri, 3 Mar 2000, Anders Magnusson wrote:

> > My VS3100M38 running 1.4L just crashed with:
> > 
> > panic: Machine check
> > Stopped in ls at _ka410_halt: function "_ka410_halt()", entry-mask 0x0
> > cvtps r4, $3, $0, $56
> > 
> > db>

> A "Machine check" can be any type of unexpected (hardware) inconsistency
> in the machine. Normally it is parity error somewhere, like in the cache
> or on some bus. Unfortunately Machine checks are not handled very gracefully
> in the current code; and it is quite hard to test that code.

And not that's it's necessarily relevant in the particular case, but
historically I have noticed strange behavior on my VS3100's if I boot
between different versions and sometimes just slightly different compiles
of the NetBSD/vax kernel.  If it's a warm restart from a halt, then I've
often found the kernel to report errors with the cache (I don't recall
whether it was the primary or secondary or both).  If I power cycled the
system, everything came up fine.  So I think cache, once enabled, may be
left in some inconsistent state between halts.

Again, this may not be relevant at all to your problem.  You may really
have somehthing wrong with your hardware.

-brian.
--- Brian Chase | bdc@world.std.com | http://world.std.com/~bdc/ -----
CHESS IS A FUN SPORT WHEN PLAYED WITH SHOTGUNS.  -- MegaHAL.
http://ciips.ee.uwa.edu.au/~hutch/hal/HAL/