From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Re: [PATCH] Fix 2.6.34-rc1 regression in disable_ipv6 support Date: Thu, 9 Dec 2010 14:51:43 -0800 Message-ID: <20101209145143.5c45cf60@nehalam> References: <4D00F58A.2050307@hp.com> <20101209111611.1d2e6e2b@nehalam> <20101209.122033.183046393.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: ebiederm@xmission.com, brian.haley@hp.com, netdev@vger.kernel.org, maheshkelkar@gmail.com, lorenzo@google.com, yoshfuji@linux-ipv6.org, stable@kernel.org To: David Miller Return-path: Received: from mail.vyatta.com ([76.74.103.46]:57296 "EHLO mail.vyatta.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752114Ab0LIWvs (ORCPT ); Thu, 9 Dec 2010 17:51:48 -0500 In-Reply-To: <20101209.122033.183046393.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, 09 Dec 2010 12:20:33 -0800 (PST) David Miller wrote: > From: Stephen Hemminger > Date: Thu, 9 Dec 2010 11:16:11 -0800 > > > No but since removing address propagates up to user space daemons > > like Quagga please analyze and fix the problem, don't just look > > for band aid. > > Stephen, we lived with the previous behavior for 12+ years. > > You broke stuff that did work before your change. > > Putting the onus on Eric to fix it exactly how you want it to > be fixed is therefore not appropriate. > > You seem to be putting exactly zero effort into trying to reproduce > the problem yourself and fixing a bug you introduced. And hey we > have a standard way to deal with a regression when the guilty party > is uncooperative, revert. > > There are therefore three choices: > > 1) Revert. And this is the one I'm favoring because of how you are > handling this issue. The responsibility to resolve this regression > is your's not Eric's. > > Frankly, Eric is being incredibly nice by working on trying to fix > a bug which you introduced. > > 2) Accept Eric's proposed fix. > > 3) Figure out the real bug yourself and fix the problem the way you > find acceptable in a reasonable, short, amount of time. > > Loopback has always been special, especially on ipv6. When we don't > have a device to point something at, we point it at loopback. > > Also destination cache entries which still have references when they > get zapped get pointed at loopback. Quit being a grinch. I am working on it, just don't know the answer. I want to try a couple solutions, so far Eric's looks okay, just want to make sure that it doesn't break anything. You are over reacting. Doing on the fly re-enabling of ipv6 is a corner case. The problem was only discovered a couple of days ago, it is not like the world is burning down. --