From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH 0/2] Get rid of ndo_xmit_flush Date: Mon, 01 Sep 2014 15:35:59 -0700 (PDT) Message-ID: <20140901.153559.1200380977792272217.davem@davemloft.net> References: <1409601942.21965.23.camel@localhost> <20140901.145646.174847768842584425.davem@davemloft.net> <1409610669.21965.55.camel@localhost> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, therbert@google.com, jhs@mojatatu.com, edumazet@google.com, jeffrey.t.kirsher@intel.com, rusty@rustcorp.com.au, dborkman@redhat.com, brouer@redhat.com, john.r.fastabend@intel.com To: hannes@stressinduktion.org Return-path: Received: from shards.monkeyblade.net ([149.20.54.216]:44814 "EHLO shards.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752031AbaIAWgB (ORCPT ); Mon, 1 Sep 2014 18:36:01 -0400 In-Reply-To: <1409610669.21965.55.camel@localhost> Sender: netdev-owner@vger.kernel.org List-ID: From: Hannes Frederic Sowa Date: Tue, 02 Sep 2014 00:31:09 +0200 > I am afraid that TX-lock is per nic-queue and it won't work out that > easy because skbs for different queues can reside in a Qdisc. Batching will help mostly when netif_queue_start() triggers, when we are in steady state stopping and re-starting the txq. What we really care about are the qdisc <--> txq one to one relationship which is by far the most common case. In the multiple txq to qdisc case, we still can benefit because packets for one txq will still tend to bunch op and give deferral opportunities.