From owner-freebsd-net  Sun Oct  6 17: 0:42 2002
Delivered-To: freebsd-net@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 99FE537B404
	for <freebsd-net@FreeBSD.ORG>; Sun,  6 Oct 2002 17:00:41 -0700 (PDT)
Received: from out0.mx.nwbl.wi.voyager.net (out0.mx.nwbl.wi.voyager.net [169.207.3.118])
	by mx1.FreeBSD.org (Postfix) with ESMTP id CF35D43E75
	for <freebsd-net@FreeBSD.ORG>; Sun,  6 Oct 2002 17:00:40 -0700 (PDT)
	(envelope-from silby@silby.com)
Received: from pop4.nwbl.wi.voyager.net (pop4.nwbl.wi.voyager.net [169.207.2.83])
	by out0.mx.nwbl.wi.voyager.net (Postfix) with ESMTP
	id 0E4DF83690; Sun,  6 Oct 2002 19:00:40 -0500 (CDT)
Received: from [10.1.1.6] (d45.as8.nwbl0.wi.voyager.net [169.207.132.45])
	by pop4.nwbl.wi.voyager.net (8.10.2/8.10.2) with ESMTP id g9700do56831;
	Sun, 6 Oct 2002 19:00:39 -0500 (CDT)
Date: Sun, 6 Oct 2002 19:05:04 -0500 (CDT)
From: Mike Silbersack <silby@silby.com>
To: Luigi Rizzo <rizzo@icir.org>
Cc: Steve Francis <steve@expertcity.com>, <freebsd-net@FreeBSD.ORG>
Subject: Re: Help with net.inet.ip.intr_queue_maxlen
In-Reply-To: <20021006161652.A15352@carp.icir.org>
Message-ID: <20021006190135.J14377-100000@patrocles.silby.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-freebsd-net@FreeBSD.ORG
Precedence: bulk
List-ID: <freebsd-net.FreeBSD.ORG>
List-Archive: <http://docs.freebsd.org/mail/> (Web Archive)
List-Help: <mailto:majordomo@FreeBSD.ORG?subject=help> (List Instructions)
List-Subscribe: <mailto:majordomo@FreeBSD.ORG?subject=subscribe%20freebsd-net>
List-Unsubscribe: <mailto:majordomo@FreeBSD.ORG?subject=unsubscribe%20freebsd-net>
X-Loop: FreeBSD.org


On Sun, 6 Oct 2002, Luigi Rizzo wrote:

> large queues to build up in other drivers. E.g. the "xl"
> driver was recently reported to take a long time every second to
> read some info (statistics ?) from the NIC; at 8kpps, 10ms
> is roughly 80 packets, which would easily explain the drops you
> are seeing.

The assertion that statistics retrieval was taking up lots of time turned
out to be completely false.  In the past, the statistics update also
triggered a MII poll, which is why the statistics interrupt seemed to be
the culprit.

But yes, the MII poll was taking 10ms every second until I recently
committed Harti Brandt's changes to -current.  Now it "only" takes 1ms,
which is still too long, but hard to reduce further.  The MFC will have to
wait until after 4.7-release is tagged, unfortunately.

Mike "Silby" Silbersack


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message