From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alexander Duyck Subject: Re: Missing IPv4 routes Date: Tue, 27 Oct 2015 13:29:10 -0700 Message-ID: <562FDE96.3030400@gmail.com> References: <562AA7D3.9030905@vultr.com> <562AB578.9090107@gmail.com> <562B8856.6070501@vultr.com> <562E46BA.7060904@gmail.com> <562E5B69.6080108@vultr.com> <562FD820.5050901@gameservers.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit To: Brian Rak , netdev@vger.kernel.org Return-path: Received: from mail-pa0-f47.google.com ([209.85.220.47]:34342 "EHLO mail-pa0-f47.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S965251AbbJ0U3L (ORCPT ); Tue, 27 Oct 2015 16:29:11 -0400 Received: by padhk11 with SMTP id hk11so232570989pad.1 for ; Tue, 27 Oct 2015 13:29:11 -0700 (PDT) In-Reply-To: <562FD820.5050901@gameservers.com> Sender: netdev-owner@vger.kernel.org List-ID: On 10/27/2015 01:01 PM, Brian Rak wrote: > (Existing email got kinda messy, starting over again): > > So, I'm having an issue with the kernel where if I add a bunch of > routes, I see some of them go "missing". They don't show up in the > 'ip -4 route' list, but they do show up if I do 'ip -4 route get X'. > > I managed to come up with a simple set of reproduction commands: > > ip link add veth0 type veth peer name veth1 > ip link set veth0 up > ip link set veth1 up > > ip route add 108.61.171.119/32 dev veth0 scope link > ip route add 108.61.171.141/32 dev veth1 scope link > ip route add 108.61.171.223/32 dev veth1 scope link > ip route add 108.61.171.250/32 dev veth1 scope link > ip route add 108.61.171.247/32 dev veth1 scope link > > ip route show > > In the route show, you'll see 108.61.171.250/32 and 108.61.171.247/32 > missing completely. > > I did a lot of bisecting, and traced it down to this commit: > > commit 8be33e955cb959dabc1a6eef0b7356fe8cf73fa6 > Author: Alexander Duyck > Date: Wed Mar 4 14:59:19 2015 -0800 > > fib_trie: Fib walk rcu should take a tnode and key instead of a > trie and a leaf > > The commit immediately prior to this one > (7289e6ddb633aaee6ccea2bd2e410654c47b29a6) works fine. > > I tried the off-by-one fix from > e55ffaf457bcc8ec4e9d9f56f955971f834d65b3, however this doesn't appear > to help at all. This code is a little above my head, so I don't > really understand what exactly is broken here. I'll take a look at it and see if I can come up with a fix by this afternoon. Thanks. - Alex