From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?iso-8859-2?Q?Micha=B3_Miros=B3aw?= Subject: Re: [PATCH 06/10] net: remove NETIF_F_ALL_TX_OFFLOADS Date: Thu, 14 Jul 2011 23:54:01 +0200 Message-ID: <20110714215401.GA17618@rere.qmqm.pl> References: <7750f18707f8ebc674a830555b0227c49cbfd608.1310601401.git.mirq-linux@rere.qmqm.pl> <20110714.144314.1353889855984029341.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-2 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: netdev@vger.kernel.org, fubar@us.ibm.com, andy@greyhouse.net, kaber@trash.net To: David Miller Return-path: Received: from rere.qmqm.pl ([89.167.52.164]:55321 "EHLO rere.qmqm.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754938Ab1GNVyD (ORCPT ); Thu, 14 Jul 2011 17:54:03 -0400 Content-Disposition: inline In-Reply-To: <20110714.144314.1353889855984029341.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, Jul 14, 2011 at 02:43:14PM -0700, David Miller wrote: > From: Micha=B3 Miros=B3aw > Date: Thu, 14 Jul 2011 02:10:29 +0200 (CEST) >=20 > > There is no software fallback implemented for SCTP or FCoE checksum= ming, > > and so it should not be passed on by software devices like bridge o= r bonding. > >=20 > > For VLAN devices, this is different. First, the driver for underlyi= ng device > > should be prepared to get offloaded packets even when the feature i= s disabled > > (especially if it advertises it in vlan_features). Second, devices = under > > VLANs do not get replaced without tearing down the VLAN first. > >=20 > > This fixes a mess I accidentally introduced while converting bondin= g to > > ndo_fix_features. > >=20 > > NETIF_F_SOFT_FEATURES are removed from BOND_VLAN_FEATURES because t= hey > > are unused as of commit 712ae51afd. > >=20 > > Signed-off-by: Micha=B3 Miros=B3aw >=20 > This is not the answer. >=20 > We have to implement software fallbacks for the checksum offloads. >=20 > That is easier than having this ever growing set of exceptions for > feature bit propagation, which at best will end up being a steaming > pile of unmaintainable poo. I agree about the fallbacks, but this patch is about fixing a regressio= n that's going to hit 3.0. Now all bonding devices always advertise e.g. SCTP_CSUM. Sorry if I was unclear about this. Best Regards, Micha=B3 Miros=B3aw