From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jamal Hadi Salim Subject: Re: [PATCH net-next 1/1] net_sched: Introduce skbmod action Date: Tue, 19 Jul 2016 09:56:02 -0400 Message-ID: <578E3172.9020601@mojatatu.com> References: <1468744884-8246-1-git-send-email-jhs@emojatatu.com> <20160718041933.GB36253@ast-mbp.thefacebook.com> <578C7C5C.2070804@mojatatu.com> <578CA4EB.7060703@iogearbox.net> <578CAA96.2090501@mojatatu.com> <578E2960.7080709@iogearbox.net> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Cc: davem@davemloft.net, netdev@vger.kernel.org, xiyou.wangcong@gmail.com, nikolay@cumulusnetworks.com To: Daniel Borkmann , Alexei Starovoitov Return-path: Received: from mail-lf0-f68.google.com ([209.85.215.68]:35884 "EHLO mail-lf0-f68.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753867AbcGSN4H (ORCPT ); Tue, 19 Jul 2016 09:56:07 -0400 Received: by mail-lf0-f68.google.com with SMTP id 33so1366658lfw.3 for ; Tue, 19 Jul 2016 06:56:06 -0700 (PDT) In-Reply-To: <578E2960.7080709@iogearbox.net> Sender: netdev-owner@vger.kernel.org List-ID: On 16-07-19 09:21 AM, Daniel Borkmann wrote: > True, the 32 bit chunks are more generic and as such you need to put more > effort in user space to handle them, but at the same time gain more > flexibility > w/o having to have a module for each and every proto. I dont see anything wrong with using pedit as a first step; even if you did what Cong said he would do _i wont use it_ given the choice against skbmod. I think we are going in circles now in this discussion. You probably didnt mean to say module per protocol above since we only have one action module [no different than what ebtables or openvswitch does. It may have more justifiable extensions in the future]. > But apart from this, > neither pedit nor tcf_skbmod_run() here handle checksum complete, so you'll > potentially get false positives wrt csum corruption and drops as a result > when using either of the two. > pedit maybe tricky. Any suggestions? On tcf_skbmod_run, mostly ignorance: while doing only ethernet updates; is it still needed to do the checksum complete? cheers, jamal