From mboxrd@z Thu Jan 1 00:00:00 1970 From: Willy Tarreau Subject: Re: [PATCH] bonding: move IPv6 support into a separate kernel module Date: Sun, 22 Feb 2009 09:27:17 +0100 Message-ID: <20090222082717.GS5038@1wt.eu> References: <499F4B61.3050508@hp.com> <20090221.235901.244910753.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: brian.haley@hp.com, arvidjaar@mail.ru, vladislav.yasevich@hp.com, chuck.lever@oracle.com, tytso@mit.edu, Valdis.Kletnieks@vt.edu, rjw@sisk.pl, netdev@vger.kernel.org, bonding-devel@lists.sourceforge.net, jamagallon@ono.com, linux-kernel@vger.kernel.org, fubar@us.ibm.com To: David Miller Return-path: Received: from 1wt.eu ([62.212.114.60]:2411 "EHLO 1wt.eu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751969AbZBVI2O (ORCPT ); Sun, 22 Feb 2009 03:28:14 -0500 Content-Disposition: inline In-Reply-To: <20090221.235901.244910753.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Sat, Feb 21, 2009 at 11:59:01PM -0800, David Miller wrote: > From: Brian Haley > Date: Fri, 20 Feb 2009 19:31:29 -0500 > > > [Possible fix for bonding IPv6 regression reported by Andrey Borzenkov, > > tried to keep all Cc's] > > > > This patch moves the IPv6 bonding code into a separate kernel module > > called bonding_ipv6 if either bonding or IPv6 are built as modules. > > If both are built into the kernel then this is as well. Bonding_ipv6.ko > > registers an "send_unsol_na" function pointer for the unsolicited > > advertisement function to be called on a failover - the default action > > is to do nothing. The notifier callbacks are now registered in this > > module and not in the base bonding module. > > > > Signed-off-by: Brian Haley > > Thanks for taking the time to work on this Brian. > > I wonder if we aren't just trading one evil for another. > > Right now just configuring bonding will get the bonding > module loaded and the ipv6 facilities will be visible. > > Now with your change, the user has to explicitly load > the module. That's extremely user-unfriendly. Especially since bonding is working between layers 1 and 2 and provides a protocol-agnostic ethernet-like interface. It would seem very confusing to load ethernet drivers depending on the protocols we expect to run on top of them :-/ Willy