From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jamal Hadi Salim Subject: Re: [PATCH v2 net-next 1/2] cls_bpf: introduce integrated actions Date: Fri, 18 Sep 2015 08:13:44 -0400 Message-ID: <55FBFFF8.3050101@mojatatu.com> References: <1442383543-4720-1-git-send-email-ast@plumgrid.com> <1442383543-4720-2-git-send-email-ast@plumgrid.com> <55FAB40E.304@mojatatu.com> <55FABC7B.6060502@iogearbox.net> <55FAD9F4.1010603@plumgrid.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Cc: John Fastabend , netdev@vger.kernel.org To: Alexei Starovoitov , Daniel Borkmann , "David S. Miller" Return-path: Received: from mail-ig0-f171.google.com ([209.85.213.171]:35538 "EHLO mail-ig0-f171.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753425AbbIRMNr (ORCPT ); Fri, 18 Sep 2015 08:13:47 -0400 Received: by igbkq10 with SMTP id kq10so16420717igb.0 for ; Fri, 18 Sep 2015 05:13:46 -0700 (PDT) In-Reply-To: <55FAD9F4.1010603@plumgrid.com> Sender: netdev-owner@vger.kernel.org List-ID: On 09/17/15 11:19, Alexei Starovoitov wrote: > misread as bpf program now executes the actions and bypasses > tcf_exts_exec() ? Well, that may be interesting idea for > the future, And above is precisely why i raised the concern. You are already bypassing tcf_exts_exec with that patch. It is a big jump. It is kind of hard to continue the discussion because i notice Dave just took in the patches. Please dont go the above path of fully fledged bypass. The architecture is about small tools that come together to provide complex processing. ebpf may be the best classifier today - but by no means the only one or guaranteed that nothing better will exist for speficic use cases. If there is something in the core that needs improvement for the benefit of all, then lets do that. Example i find the classid metadata interesting but flinching at ACT_REDIRECT. cheers, jamal