From mboxrd@z Thu Jan 1 00:00:00 1970 From: Zheng Liu Subject: Re: [PATCH 2/2] ext4: fix cpu_vs_disk conversions Date: Wed, 3 Apr 2013 21:04:43 +0800 Message-ID: <20130403130443.GA16014@gmail.com> References: <1364986711-3630-1-git-send-email-dmonakhov@openvz.org> <1364986711-3630-2-git-send-email-dmonakhov@openvz.org> <20130403123317.GD7741@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Dmitry Monakhov , linux-ext4@vger.kernel.org, wenqing.lz@taobao.com To: Theodore Ts'o Return-path: Received: from mail-pa0-f42.google.com ([209.85.220.42]:43265 "EHLO mail-pa0-f42.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758541Ab3DCMs2 (ORCPT ); Wed, 3 Apr 2013 08:48:28 -0400 Received: by mail-pa0-f42.google.com with SMTP id kq13so886347pab.1 for ; Wed, 03 Apr 2013 05:48:28 -0700 (PDT) Content-Disposition: inline In-Reply-To: <20130403123317.GD7741@thunk.org> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Wed, Apr 03, 2013 at 08:33:17AM -0400, Theodore Ts'o wrote: > On Wed, Apr 03, 2013 at 02:58:31PM +0400, Dmitry Monakhov wrote: > > All new features are broken on bigendian hosts due to lack of conversion: > > es_cache: https://lkml.org/lkml/2013/3/28/64 > > inode's csum and ext_to_ind_migrate are also broken. > > > > Testcase: make C=2 CF="-D__CHECK_ENDIAN__ > > > > Signed-off-by: Dmitry Monakhov > > Thanks for this comprehensive patch. I'm currently trying to decide > how much of this I should try to push in before the next 3.9-rcX > window, and how much can wait until the next merge window. > > We definitely want to fix the fs corruption bug for big-endian > systems. For the rest, I'm on the fence, since they are less likely > to bite people hard --- metadata checksum is still pretty new and not > fully supported, and the punch hole bug is again also pretty new > functionality. > > Also, pushing something to Linus now could potentially disrupt the > patches in the ext4 dev tree for the next merge window. For the zero > extents problem, there's no question what are priorities should be; > user dataloss trumps developer convenience any day. For the rest, > what do you all think? Are they likely to hit people hard enough that > it's worth trying to get this to Linus sooner, as opposed to having > the fixes for the rest of the big endian issues land in 3.10 and > 3.9.1? Hi Ted, I agree with you that we need to fix the big-endian bug in extent tree because it affects all people who use ext4 file system. Meanwhile I think we need to fix the problem in punching hole and xattr because that has been there. I have looked at Dmitry's patch, and it fixes some problems that is only in dev branch (e.g. migration). I think this problem can be fixed in 3.10 later. Meanwhile metadata_csum is still under ext4dev. So I think it also can be fixed in next merge window. Thanks, - Zheng