From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Graf Subject: Re: [PATCH net-next v2] net: Add fib rules at vrf device create Date: Wed, 9 Dec 2015 14:04:22 +0100 Message-ID: <20151209130422.GB9900@pox.localdomain> References: <1449607685-8910-1-git-send-email-dsa@cumulusnetworks.com> <20151208.214411.2137019448940208712.davem@davemloft.net> <56679B2F.5080700@cumulusnetworks.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: David Miller , netdev@vger.kernel.org, shm@cumulusnetworks.com To: David Ahern Return-path: Received: from mail-wm0-f46.google.com ([74.125.82.46]:36587 "EHLO mail-wm0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751592AbbLINEY (ORCPT ); Wed, 9 Dec 2015 08:04:24 -0500 Received: by wmww144 with SMTP id w144so221340641wmw.1 for ; Wed, 09 Dec 2015 05:04:23 -0800 (PST) Content-Disposition: inline In-Reply-To: <56679B2F.5080700@cumulusnetworks.com> Sender: netdev-owner@vger.kernel.org List-ID: On 12/08/15 at 08:08pm, David Ahern wrote: > On 12/8/15 7:44 PM, David Miller wrote: > >Unfortunately it's too late for this, you should have considered this > >issue fully when VRF first went into an upstream release. > > > >If I add your change, the user experience is _worse_. > > > >Users on older kernels have to use the full sequence, then if they > >upgrade their kernels to one with this patch then the 'ru add' et > >al. commands will fail. > > > > Perhaps this is another fug (feature bug), they do not fail: > > root@kenny-jessie2:~# ip ru ls > 0: from all lookup local > 32760: from all iif vrf-green lookup vrf-green > 32761: from all oif vrf-green lookup vrf-green > 32762: from all iif vrf-blue lookup vrf-blue > 32763: from all oif vrf-blue lookup vrf-blue > 32764: from all iif vrf-red lookup vrf-red > 32765: from all oif vrf-red lookup vrf-red > 32766: from all lookup main > 32767: from all lookup default > > root@kenny-jessie2:~# ip ru add oif vrf-red lookup vrf-red This is not a bug. FIB rules can include forward jumps so it's perfectly valid to have the same rule in the entire chain multiple times. Just like it is for iptables and multiple chains.