From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sergei Shtylyov Subject: Re: [PATCH net-next] mpls: Don't accept multipath configuration until the support is complete Date: Mon, 2 Nov 2015 22:49:08 +0300 Message-ID: <5637BE34.10304@cogentembedded.com> References: <1446133748-13738-1-git-send-email-roopa@cumulusnetworks.com> <87egg8ryia.fsf@x220.int.ebiederm.org> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Cc: rshearma@brocade.com, netdev@vger.kernel.org, Roopa Prabhu To: "Eric W. Biederman" , David Miller Return-path: Received: from mail-lb0-f182.google.com ([209.85.217.182]:33190 "EHLO mail-lb0-f182.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751711AbbKBTtN (ORCPT ); Mon, 2 Nov 2015 14:49:13 -0500 Received: by lbbec13 with SMTP id ec13so96489656lbb.0 for ; Mon, 02 Nov 2015 11:49:12 -0800 (PST) In-Reply-To: <87egg8ryia.fsf@x220.int.ebiederm.org> Sender: netdev-owner@vger.kernel.org List-ID: Hello. On 11/02/2015 10:29 PM, Eric W. Biederman wrote: > Currently the multipath code has a nasty failure mode in that it will > fail to notice link down or administrative device down and will > instead black hole packets instead of sending them to their nexthop > destination. > > Half the point of multipath is to gracefully handle forwarding path > failures and as the current code does not handle forwarding failures the > current code is dangerous to use. > > As mpls multipath has never been exported to userspace and as the > implementation was not complete before the merge window disable the mpls > multipath code by rejecting all multipath configuration requests. This > will give us another kernel development cycle to cleanly sort out the > issues, without any bad precedents to worry about. > > Signed-off-by: "Eric W. Biederman" > --- > net/mpls/af_mpls.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/net/mpls/af_mpls.c b/net/mpls/af_mpls.c > index c70d750148b6..893cd2dc3979 100644 > --- a/net/mpls/af_mpls.c > +++ b/net/mpls/af_mpls.c > @@ -1162,6 +1162,8 @@ static int rtm_to_route_config(struct sk_buff *skb, struct nlmsghdr *nlh, > { > cfg->rc_mp = nla_data(nla); > cfg->rc_mp_len = nla_len(nla); > + /* Fail until multipath support is complete */ > + goto errout; > break; Forgot to delete *break*? > } > default: > MBR, Sergei