From: awb@ed.ac.uk (Alan W Black) Subject: Re: Error...NMI received? Date: 26 Dec 1992 10:57:40 GMT
In article <BzsCpn.4v1@citrus.SAC.CA.US> ianj@citrus.SAC.CA.US ( Ian Justman ) writes:
> marc@r-node.gts.org (Marc Fournier - Admin) writes:
> : Hello...
> :
> : As it says...I get an error, well, I got it twice so far, that
> : goes to the effect of:
> :
> : Uhhuh. NMI received. Dazed and confused, but trying to continue
> :
> : What does this mean? :)
> I wouldn't mind knowing, too. With my full setup, I _ALWAYS_ get it,
> but otherwise, everything seems to be in order.
> --
> Born to void warranties! ianj@ijpc.UUCP
I'm not sure if this is related but we used to have a similar
problem of spurious NMI's under 386bsd. After much guessing
and little hindsight we found the problem to be that our
386 board (which is rather old) would get memory parity errors
when the processor was halted (HLT instruction) this happens
in the 386bsd kernel idle loop. We removed this instruction
and let the processor simply loop rather than halt, and we've
never had NMI's since.
It appears that some older boards do not refresh memory properly when
the processor is halted. I don't know if the Linux kernel halts
but it could easily do so. If it does remove that instruction
replacing it with a loop and see if it makes any difference.
Alan
Alan W Black 2 Buccleuch Place, Edinburgh, UK
Centre for Cognitive Science tel: (+44) -31 650 4627
University of Edinburgh email: awb@ed.ac.uk