From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jamal Hadi Salim Subject: Re: [PATCH net-next] net, sched: add clsact qdisc Date: Thu, 7 Jan 2016 09:29:52 -0500 Message-ID: <568E7660.1070801@mojatatu.com> References: <20160107035334.GA99020@ast-mbp.thefacebook.com> <568E395F.8020504@stressinduktion.org> <568E52E8.1010804@iogearbox.net> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Cc: davem@davemloft.net, john.fastabend@gmail.com, eric.dumazet@gmail.com, netdev@vger.kernel.org To: Daniel Borkmann , Hannes Frederic Sowa , Alexei Starovoitov Return-path: Received: from mail-io0-f169.google.com ([209.85.223.169]:34455 "EHLO mail-io0-f169.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750722AbcAGO37 (ORCPT ); Thu, 7 Jan 2016 09:29:59 -0500 Received: by mail-io0-f169.google.com with SMTP id 1so206730299ion.1 for ; Thu, 07 Jan 2016 06:29:58 -0800 (PST) In-Reply-To: <568E52E8.1010804@iogearbox.net> Sender: netdev-owner@vger.kernel.org List-ID: On 16-01-07 06:58 AM, Daniel Borkmann wrote: > On 01/07/2016 11:09 AM, Hannes Frederic Sowa wrote: > > Idea was that this is done before we pick txq as stated. F.e., could also > be that we end up not having enqueue handler, thus moving this further down > (not sure if there's a good place?), might make it all more scattered resp. > complex to cover all parts. > From a cursory glance (Will comment a lot more later, definetely bt netdev11 time): I like the idea - actually i think that this is where John was heading initially. Both efforts complement each other. This scheme could be used to compensate for the queue lock avoidance and do a direct txmit (and let the hardware do the scheduling). The classid selection could clearly be done at this early stage. And a few of the standard skb metadatum could be used to map to hardware offloading. cheers, jamal