Return to BSD News archive
Newsgroups: comp.unix.bsd
Path: sserve!manuel!munnari.oz.au!uunet!usc!rpi!psinntp!psinntp!dg-rtp!ponds!rivers
From: rivers@ponds.uucp (Thomas David Rivers)
Subject: More on NMI problems.
Message-ID: <1992Sep2.123248.17492@ponds.uucp>
Date: Wed, 2 Sep 1992 12:32:48 GMT
Lines: 33
Well, I got much advice on my NMI problems; indicating that I probably
did, in fact, have a bad memory chip on my machine.
So; I grabbed testext.exe from SIMTEL20 and let it have at the machine,
to determine which chip was bad, etc....
It ran for 38+ hours, without a single error (this program uses
protected mode read/writes of words and double words, randomly scattered
through available memory.) Other people had sent me mail indicating
this program found their problem, etc...
So, at this point, I'm beginning to return to my previous thought, that
something else is causing the parity error (perhaps some funny DMA, or
some timing problem with the drivers?)
When I booted 386bsd again, it took less than an hour to reproduce the
problem. It's very strange, because I never get more than one NMI
line in /var/log/messages (you would think there would be several.)
Also, I never got these from version 0.0.
Also, after reading the descriptions of '386 traps, and noticing the
comment that some CGA-emulation cards used the NMI to simulate the
CGA registers, I pulled out my CGA-on-hercules emulation card, but that
didn't solve it.
My next step is to rip out the math co-processor, then I'll be left
with a hercules card, a 4-port clone card and an IDE controller (pretty
basic)...
- Dave Rivers -
(rivers@ponds.uucp)
p.s. Thanks to all the people who responded!