From: Josh Hunt Subject: Re: ext4_da_update_reserve_space warning fix not in stable Date: Wed, 7 Jan 2015 13:49:46 -0600 Message-ID: <54AD8DDA.6090902@akamai.com> References: <54AD772B.2050003@akamai.com> <20150107192105.GE2802@quack.suse.cz> Mime-Version: 1.0 Content-Type: text/plain; charset="windows-1252"; format=flowed Content-Transfer-Encoding: 7bit Cc: "linux-ext4@vger.kernel.org" , "tytso@mit.edu" , "adilger.kernel@dilger.ca" To: Jan Kara Return-path: Received: from prod-mail-xrelay08.akamai.com ([96.6.114.112]:20000 "EHLO prod-mail-xrelay08.akamai.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751528AbbAGTvT (ORCPT ); Wed, 7 Jan 2015 14:51:19 -0500 In-Reply-To: <20150107192105.GE2802@quack.suse.cz> Sender: linux-ext4-owner@vger.kernel.org List-ID: On 01/07/2015 01:21 PM, Jan Kara wrote: > On Wed 07-01-15 12:12:59, Josh Hunt wrote: >> We are hitting the following warning in the field pretty frequently >> on the latest 3.10 stable kernel: >> [11708.763067] WARNING: at fs/ext4/inode.c:363 >> ext4_da_update_reserve_space+0x261/0x280 [ext4]() >> >> The issue was resolved via "ext4: fix warning in >> ext4_da_update_reserve_space()" (7d7345322d60ed), but the fix was >> not submitted as a stable update for 3.10. I was wondering if >> there's a reason why this was not done, or if it was just >> overlooked? >> >> The patch appears to have a dependency on "quota: provide interface >> for readding allocated space into reserved space" (1c8924eb106c1ac), >> so I think both would need to be submitted if you feel it's >> possible. > Since the problem is relatively harmless, I think we didn't bother > porting it to 3.10. Feel free to port those two patches to 3.10 and submit > them to stable@vger.kernel.org. Thanks for the reply Jan. I've ported them locally to 3.10 and they seem to apply cleanly as-is. I can submit the request to stable as long as you believe there's no harm in doing so, or any other dependency which may be required. My testing of a patched 3.10 with the two above patches shows the warning no longer triggers as you report in your commit log. Thanks Josh