From owner-freebsd-hackers@FreeBSD.ORG  Mon Jan 12 09:30:03 2004
Return-Path: <owner-freebsd-hackers@FreeBSD.ORG>
Delivered-To: freebsd-hackers@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 7E3E716A4CE
	for <hackers@freebsd.org>; Mon, 12 Jan 2004 09:30:03 -0800 (PST)
Received: from mx2.ngi.de (mx2.ngi.de [213.191.74.84])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 6368643D69
	for <hackers@freebsd.org>; Mon, 12 Jan 2004 09:29:56 -0800 (PST)
	(envelope-from gerald.heinig@ngi.de)
Received: (qmail 18097 invoked from network); 12 Jan 2004 17:11:48 -0000
Received: from unknown (HELO mx2.ngi.de) ([213.191.74.86])
          (envelope-sender <gerald.heinig@ngi.de>)
          by 0 (qmail-ldap-1.03) with SMTP
          for <hackers@freebsd.org>; 12 Jan 2004 17:11:48 -0000
From: <gerald.heinig@ngi.de>
To: hackers@freebsd.org
Date: Mon, 12 Jan 2004 18:10:21 +0100
X-Mailer: NGI Webmail
Message-Id: <20040112172956.6368643D69@mx1.FreeBSD.org>
Subject: ping duplicates under load
X-BeenThere: freebsd-hackers@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
List-Id: Technical Discussions relating to FreeBSD
	<freebsd-hackers.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-hackers>,
	<mailto:freebsd-hackers-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-hackers>
List-Post: <mailto:freebsd-hackers@freebsd.org>
List-Help: <mailto:freebsd-hackers-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-hackers>,
	<mailto:freebsd-hackers-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Mon, 12 Jan 2004 17:30:03 -0000

Hi all,

I've noticed a very weird phenomenon with ping which I can't explain.

I have two machines with 2 NICs.
When I started a ping -f with larger packet size (-s 3052) with NIC 1
and simultaneously hammered NIC 2 with UDP using ttcp, I got duplicate
ping replies (on NIC 1, of course). Not many: only about 1 in 10000 to
100000, but still, I can't explain it.
Since I had a similar problem a while back with another driver I
developed (for OS X) and that turned out to be due to lost interrupts
and my ISR not terminating quickly enough, I decided to limit the number
of iterations of my ISR to 3.
That improved the situation somewhat: I now only get 1 duplicate ping in
1E8.
The real kicker is: I see this effect when my driver is the one doing
the UDP receive. The ping duplicates also happen with other card
drivers, so it's not a simple transmit/receive ring problem with my
driver. When I use a different card driver to do the UDP receives, the
effect goes away.

It seems like the receive interrupt activity on my driver affects the
ping replies of the _other_ drivers.

Does anyone have any idea what could be going on?

Cheers,

Gerald