From mboxrd@z Thu Jan 1 00:00:00 1970 From: ebiederm@xmission.com (Eric W. Biederman) Subject: Re: [PATCH] Fix 2.6.34-rc1 regression in disable_ipv6 support Date: Thu, 09 Dec 2010 11:22:56 -0800 Message-ID: References: <4CFD0AF0.9090809@hp.com> <4D000B6D.1060708@hp.com> <20101208154932.5bc4b254@nehalam> <4D00F58A.2050307@hp.com> <20101209082703.6e5519e7@nehalam> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Brian Haley , David Miller , netdev@vger.kernel.org, Mahesh Kelkar , Lorenzo Colitti , YOSHIFUJI Hideaki , stable@kernel.org To: Stephen Hemminger Return-path: Received: from out02.mta.xmission.com ([166.70.13.232]:55568 "EHLO out02.mta.xmission.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753975Ab0LITXC (ORCPT ); Thu, 9 Dec 2010 14:23:02 -0500 In-Reply-To: <20101209082703.6e5519e7@nehalam> (Stephen Hemminger's message of "Thu, 9 Dec 2010 08:27:03 -0800") Sender: netdev-owner@vger.kernel.org List-ID: Stephen Hemminger writes: > On Thu, 09 Dec 2010 10:28:10 -0500 > Brian Haley wrote: > >> On 12/08/2010 11:16 PM, Eric W. Biederman wrote: >> /* If just doing link down, and address is permanent >> and not link-local, then retain it. */ >> if (!how && >> (ifa->flags&IFA_F_PERMANENT) && >> !(ipv6_addr_type(&ifa->addr) & IPV6_ADDR_LINKLOCAL)) { >> list_move_tail(&ifa->if_list, &keep_list); >> >> /* If not doing DAD on this address, just keep it. */ >> if ((dev->flags&(IFF_NOARP|IFF_LOOPBACK)) || >> idev->cnf.accept_dad <= 0 || >> (ifa->flags & IFA_F_NODAD)) > > I think the problem is on coming back up, not on the down step. Oh it is. All addresses that you keep break if you down the loopback interface, no matter which interface those addresses are on. Stephen the cause of the regression in 2.6.34-rc1 that you introduced that breaks the disable_ipv6 functionality in practice is removing the loopback address from the loopback interface. So I sent a partial revert. It is safe to do a partial revert because the loopback address is always reprogrammed when we bring the interface back up. But that reprogramming only works if it doesn't error out with -EEXIST. So by all means properly fix the ancient bug that breaks usage of all local ipv6 addresses when the loopback interface is brought down, and we can remove the regression fix. However complaining about a partial revert to fix a regression you introduced because it fixes a problem deep within the ipv6 networking stack that the smallest modicum of testing would have revealed on your part before you broke things seems inappropriate. Please let's get the disable_ipv6 functionality working again (where in practice we don't care about preserving addresses). Then let's take our time and tack and fix whatever this is properly. Eric