From owner-freebsd-bugs  Tue Apr 18 04:22:01 1995
Return-Path: bugs-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id EAA00792
          for bugs-outgoing; Tue, 18 Apr 1995 04:22:01 -0700
Received: from mail.Germany.EU.net (mail.Germany.EU.net [192.76.144.65])
          by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id EAA00784
          for <freebsd-bugs@freebsd.org>; Tue, 18 Apr 1995 04:21:56 -0700
From: Bernard.Steiner@Germany.EU.net
Received: by mail.Germany.EU.net with ESMTP (8.6.5:29/EUnetD-2.5.1.d) via EUnet
	id NAA29020; Tue, 18 Apr 1995 13:23:04 +0200
Message-Id: <199504181123.NAA03844@qwerty.Germany.EU.net>
Received: from localhost.Germany.EU.net
	by qwerty.Germany.EU.net with SMTP (8.6.4/EUnetDlan-1.14-1.2.0)
	via EUnet for [mail.germany.eu.net]
	id NAA03844; Tue, 18 Apr 1995 13:23:03 +0200
To: davidg@Root.COM
cc: freebsd-bugs@FreeBSD.org
Subject: Re: couple of bugs in 2.0R 
In-reply-to: Your message of Mon, 17 Apr 1995 19:28:49 PDT.
             <199504180228.TAA00214@corbin.Root.COM> 
Date: Tue, 18 Apr 1995 13:23:01 +0200
Sender: bugs-owner@FreeBSD.org
Precedence: bulk

     >solution), and now I don't get any more of those
     >stray irq 11
     >ed0 timeout
     >messages. Unless solved, can we get a flag for ed that sort of increases 
    the
     >watchdog timer significantly ?
     
        Umm, that would increase the timeout to 20 seconds. This is not desirea
    ble
     and should never be necessary. I don't see how increasing it above what it
     currently is could affect the problem at all (the interrupt should occur i
    n
     less than 100ms no matter what happens).

Hmm. Very strange, indeed. Note, however, that the kernel reported the
problems in the exact sequence given above, i.e.
first stray irq 11 and then ed0 timeout; both at a snap of a finger's time
from ifconfig ed0 up.

It works now with the watchdog at 20; I've re-tried the original kernel with a
watchdog set to 2, and that failed.

Is there anything I should do in order to help track this down, or just set
the watchdog to 20 which doesn't make sense but works (note: there wasn't even
a two second latency in accessing the ethernet) ?

Thanks,
	Bernard