Subject: Re: ex0 weird error
To: Peter Seebach <seebs@plethora.net>
From: Andrew Brown <atatat@atatdot.net>
List: current-users
Date: 12/14/2000 00:46:46
>stray eb164 irq 1
>...
>stray eb164 irq 1
>...
>stray eb164 irq 1
>
>After a bit of this, ex0 died completely.  This happened under an abusive
>load; I'm performance testing some changes in BSD/OS's NFS client code, and
>I was running a couple of simultaneous programs manipulating 2GB files.
>I have temporarily switched to an Intel card, but I'm curious as to:
>
>1.  Whether this is a known issue of some sort.
>2.  Whether it's likely to be alpha-specific.
>3.  Whether anyone is interested in persuing this.
>
>If no one cares much, I'll probably send-pr it and forget about it.

known, afaik.  i asked someone about this a few weeks ago and was told
that there was a fix in a pr somewhere.  your "halting" problem was
probably accompanied by a message of the form "stray eb164 irq 1;
stopped logging", yes?  the offending interrupt gets disabled at that
point, and that means that ex0's interrupts are no longer handled.
therefore, all traffic grinds to a halt.

i "fixed" that (for that reboot, at least) by breaking into ddb and
doing

   call eb164_intr_enable(0x3)

you probably wanna use 0x1 for that.  after that, my network spun back
into action and it never logged anything again.  of course...i've
rebooted since then.

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior@daemon.org             * "ah!  i see you have the internet
twofsonet@graffiti.com (Andrew Brown)                that goes *ping*!"
andrew@crossbar.com       * "information is power -- share the wealth."