Re: Aiee in 2.0.24

Leonard N. Zubkoff (
Mon, 4 Nov 1996 15:04:36 -0800

From: (Brad Pepers)
Date: Sun, 3 Nov 1996 22:41:12 -0700 (MST)

I got the first oops I've had in years today! I'm using 2.0.24
on a Micron PPro 200 system with 64Mb RAM, BusLogic SCSI, WD8013
ethernet. The error happens when doing a lot of NFS work. I was
able to reproduce the problem three times by doing three "cp -a"
commands at a time from a machine that had the / directory of the
Micron mounted. If I just did one or two "cp -a" commands, there
was no problem. When I strted up the third, I get these error
messages and the "rpc.nfsd" command dies. If I start rpc.nfsd up
again it will do a bit more work and die again. I keep doing this
until one of the "cp -a" commands is done and then no more errors!

The death shown is in ei_start_xmit+290 which seems an odd place.
Here is the dissassem output:

0x19fb97 <ei_start_xmit+287>: movl %esi,%edx
0x19fb99 <ei_start_xmit+289>: outb %al,(%dx)
0x19fb9a <ei_start_xmit+290>: outb %al,$0x80
0x19fb9c <ei_start_xmit+292>: movb 0x1a(%ebp),%al
0x19fb9f <ei_start_xmit+295>: testb %al,%al

I'm not sure what else to check! I've never had this problem before
but perhaps I haven't been hitting the NFS stuff so hard before...
Any idea what it could be?

Was this an OOPS or a "reserved" trap? You didn't include that part of the
kernel trap report. It sure looks like the P6 Local APIC Spurious Interrupt
Bug for which I recently posted a new workaround patch. I have a collection of
these from a few months ago when I had EPI pointing to precisely the same OUTB