From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner+w=401wt.eu-S1422672AbXDLF0v@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1422672AbXDLF0v (ORCPT <rfc822;w@1wt.eu>);
	Thu, 12 Apr 2007 01:26:51 -0400
Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1422669AbXDLF0v
	(ORCPT <rfc822;linux-kernel-outgoing>);
	Thu, 12 Apr 2007 01:26:51 -0400
Received: from mga02.intel.com ([134.134.136.20]:48488 "EHLO mga02.intel.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1422664AbXDLF0u (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Thu, 12 Apr 2007 01:26:50 -0400
X-ExtLoop1: 1
X-IronPort-AV: i="4.14,399,1170662400"; 
   d="scan'208"; a="226051487:sNHT28494914"
Subject: Re: [PATCH] NET: [UPDATED] Multiqueue network device support
	implementation.
From: Zhu Yi <yi.zhu@intel.com>
To: Patrick McHardy <kaber@trash.net>
Cc: "Waskiewicz Jr, Peter P" <peter.p.waskiewicz.jr@intel.com>,
       davem@davemloft.net, netdev@vger.kernel.org,
       linux-kernel@vger.kernel.org, jgarzik@pobox.com,
       cramerj <cramerj@intel.com>,
       "Kok, Auke-jan H" <auke-jan.h.kok@intel.com>,
       "Leech, Christopher" <christopher.leech@intel.com>
In-Reply-To: <461D14E1.6020100@trash.net>
References: <D5C1322C3E673F459512FB59E0DDC32902A1A535@orsmsx414.amr.corp.intel.com>
	 <461D14E1.6020100@trash.net>
Content-Type: text/plain
Organization: Intel Corp.
Date: Thu, 12 Apr 2007 13:24:23 +0800
Message-Id: <1176355463.16123.243.camel@debian.sh.intel.com>
Mime-Version: 1.0
X-Mailer: Evolution 2.6.3 
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
X-Mailing-List: linux-kernel@vger.kernel.org

On Wed, 2007-04-11 at 19:03 +0200, Patrick McHardy wrote:
> 
> You bring up a good point, it would be good to hear the opinion from
> one of the wireless people on this since they have their own
> multiqueue scheduler in the wireless-dev tree. 

The one in the wireless-dev is pretty much like this one. It existed
only because there was not such a multiqueue aware qdisc available at
that time.

The requirement for wireless is the same as the strict PRIO with an
addition that the dequeued SKB's corresponding NIC hardware queue must
be active (this is also true for other devices I think, otherwise it has
to be requeued which leads a busy or dead loop in the end). In other
words, the dequeue method should select the SKB with the highest
priority from all the ACTIVE hardware queues (not all queues). The
wireless hardware then schedules all the packets from its 4 hardware TX
queues based on the priority and network environment.

Thanks,
-yi