From: Darren Hart Subject: Re: [PATCH] mke2fs: do not change root dir ownership Date: Mon, 13 May 2013 16:09:51 -0700 Message-ID: <519172BF.5010608@linux.intel.com> References: <1367875316-3089-1-git-send-email-vapier@gentoo.org> <20130513133731.GC4845@thunk.org> <201305131212.28657.vapier@gentoo.org> <20130513210613.GB20102@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Mike Frysinger , linux-ext4@vger.kernel.org, Robert Yang To: Theodore Ts'o Return-path: Received: from mga09.intel.com ([134.134.136.24]:46254 "EHLO mga09.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755136Ab3EMXJz (ORCPT ); Mon, 13 May 2013 19:09:55 -0400 In-Reply-To: <20130513210613.GB20102@thunk.org> Sender: linux-ext4-owner@vger.kernel.org List-ID: On 05/13/2013 02:06 PM, Theodore Ts'o wrote: > On Mon, May 13, 2013 at 12:12:27PM -0400, Mike Frysinger wrote: >> nifty. we already have to mount the fs to populate it, so putting a chown in >> there isn't a big deal in the current setup. >> >> now if only e2fsprogs would integrate the `genext2fs` project :). > > Darren Hart created a shell script which uses debugfs to do all of the > dirty work, called populate-extfs.sh which I believe Linaro and the > Yocto project is using instead of genext2fs. > > I'm not entirely sure this the latest version, but so you can get a > flavor of the script, see: > > http://patches.openembedded.org/patch/45415/ > > Darren, if you think the script is pretty stable, I'd be happy to drop > the latest version of it into the contrib directory of e2fsprogs. Let me sync with Robert (on Cc) who has been doing the integration to see what he thinks about that. We consider the script to be a temporary mechanism. Now that the symlink support has been added to debugfs, we are working to validate that everything works via the script. Once complete, the second stage of this effort is to move some of the logic from debugfs into libe2fs and make it available to mke2fs as well as debugfs. Then we would want to add the -i (initial dir) argument I had proposed originally to be able to format and populate an image in a single step. I provide that context in case it impacts Mike's plans. Always good to know what others are planning :-) -- Darren Hart Intel Open Source Technology Center Yocto Project - Technical Lead - Linux Kernel