Subject: Re: Unexpected Machine Check After Halt
To: Matthew Jacob <firstname.lastname@example.org>
From: None <email@example.com>
Date: 09/24/1997 10:26:16
> Okay- Jason and I have seen some similar nonsense on the AlphaStation 500s
> we just got- they just plain hang on reboot. Our current theory is that
> an interrupt is coming in at an inopportune time. I tried using shutdown_hook
> to quiesce a couple of interrupts, but I didn't manage to solve the
> problem. I tried a couple of other desultory attempts, but haven't
> cracked that nut.
> I also see a vaguely similar problem on my AlphaPC164 when I have the
> Qlogic 2100 (of latest vintage) inserted- when I halt or reboot, something
> generates a PCI Master Abort cycle, and bad (vector 0x660) trap kills
> the prom - you have to hit the reset button.
> On very rare occasions, I've managed to wedge the 8200 during a reboot
> cycle such that all sorts of this kind of spew occurs- basically, I think
> it's the PAL throwing up. I hope your carpets are cleanable.
> I'll file this latest one in my list of things to try and explain.
> What's this platform? An AXPPCI33?
We have experienced this problem (at BBN) on the AXPpci33. It appears to be
due to the fact that the system's console code is not particularly tolerant
of having the OS change interrupt controller configuration, at least in some
versions of the console. We were able to eliminate the problem on the AXPpci33
by upgrading to the latest (November, 1996, I believe) SRM console for the
AXPpci33. Alternately, I suppose that ICU configuration could be checkpointed
during boot and restored during shutdown. I think this is a lot of kludging
to get around console braindamage, however.