VS3100 crash / machine check
Wolfgang Rupp
panic: Machine check
Stopped in ls at _ka410_halt: function "_ka410_halt()", entry-mask 0x0
cvtps r4, $3, $0, $56
db>
I was logged into the VAX over a telnet session and did "ls".
What does this error tell me?? The machine has been up for a few days.
Wolfgang
Brian Chase
> > 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 | b...@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/
Anders Magnusson
>
> panic: Machine check
> Stopped in ls at _ka410_halt: function "_ka410_halt()", entry-mask 0x0
> cvtps r4, $3, $0, $56
>
> db>
>
> What does this error tell me?? The machine has been up for a few days.
>
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.
A Machine check will also occur on some CPU types when adressing an
unexisting memory location, but not on 3100 machines. So you likely
have some problem with your memory or so.
-- RAgge
Wolfgang Rupp
> 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.
I don't remember, but it is possible that I halted the system to the
console and then rebooted without issuing an "init" at >>>. I always
wondered what good "init" at the console might do, anyway. (Not that
I knew if init resets the cache.) Fortunately there were no Machine
Checks since, and the system has been running for two days again -
with compilations, and all.
One of the two SCSI disks will be dead soon, it has already caused me
trouble. I *hope* there is no other hardware problem...
Wolfgang