From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Sandeen Subject: Re: RFC, 32-bit compat handlers for EXT4_IOC_GROUP_ADD Date: Mon, 08 Dec 2008 10:31:03 -0600 Message-ID: <493D4BC7.1010903@redhat.com> References: <49385BC5.2070703@redhat.com> <20081206010204.GY3186@webber.adilger.int> <20081206202409.GI1323@mit.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Andreas Dilger , ext4 development To: Theodore Tso Return-path: Received: from mx2.redhat.com ([66.187.237.31]:60883 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753212AbYLHQb2 (ORCPT ); Mon, 8 Dec 2008 11:31:28 -0500 In-Reply-To: <20081206202409.GI1323@mit.edu> Sender: linux-ext4-owner@vger.kernel.org List-ID: Theodore Tso wrote: > On Fri, Dec 05, 2008 at 06:02:04PM -0700, Andreas Dilger wrote: >>> How far out of "dev" are we? I'm leaning towards saying "oh well, would >>> have been nicer the other way" but going ahead and just putting the >>> compat handler into the kernel. >> I would be OK with changing to the "proper" struct layout. Not being able >> to resize with an older e2fsprogs + newer kernel isn't going to cause any >> serious problems (unlike e.g. not being able to mount or e2fsck "/"). >> >> If we are seriously worried about compatibility, we could add the compat >> handler for 32-bit kernels (should have a different IOC number anyways >> because of the struct size) and add some arbitrary check like: >> >> #ifdef LINUX_KERNEL_VERSION > KERNEL_VERSION(2,6,40) >> #warning remove this old compat code >> #endif > > Given that a bunch of distro's have shipped e2fsprogs 1.41.x which we > advertised as being ext4 compatibility, I think we need to keep the > compatibility code. If we want to add the complexity for the 32-bit > side, with a 2-3 year timeout, that seems like a reasonable > compromise. > > - Ted I tend to agree, unfortunately... I'll send the current compat patch, then, and when/if I get motivated, add another cleaner interface+number I guess... -Eric