From: Al Viro Subject: Re: [PULL REQUEST] Quota cleanups & fixes and ext3 fixes for 2.6.34-rc1 Date: Fri, 5 Mar 2010 23:01:17 +0000 Message-ID: <20100305230117.GZ30031@ZenIV.linux.org.uk> References: <20100305003304.GE4887@quack.suse.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Jan Kara , LKML , linux-fsdevel@vger.kernel.org, dushistov@mail.ru, joel.becker@oracle.com, shaggy@linux.vnet.ibm.com, linux-ext4@vger.kernel.org To: Linus Torvalds Return-path: Received: from zeniv.linux.org.uk ([195.92.253.2]:56666 "EHLO ZenIV.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755358Ab0CEXB1 (ORCPT ); Fri, 5 Mar 2010 18:01:27 -0500 Content-Disposition: inline In-Reply-To: Sender: linux-ext4-owner@vger.kernel.org List-ID: On Fri, Mar 05, 2010 at 01:24:35PM -0800, Linus Torvalds wrote: > > > On Fri, 5 Mar 2010, Jan Kara wrote: > > > > could you please pull from > > > > git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6.git for_linus > > > > You'll get lots of quota cleanups from Christoph which touch filesystems > > all over the place but it's just function renaming and moving quota calls > > from VFS into filesystems themselves. > > Gaah. This interacted with the VFS stuff (the do_filp_open() cleanip in > particular) I merged from Al earlier today. > > It really looked very trivial, so I fixed it up. Al had split up > do_filp_open() into several helper functions, and one of your changes was > to the old pre-split one. I felt confident enough about it that I just did > the obvious merge resolution, but you guys should check it out to be sure. It's OK as merge resolution, but I really wonder if that patch (taking quota initialization to filesystems) is a good idea. It adds boilerplate code that is easy to miss. And looking at what's done in e.g. ext2, I'd say that places where it's added are lousy - if we want to do that on directory modifications, we ought to take it into places that actually create directory entries. Not to mention that we never shrink directories there, so it should be just one call in ext2_add_link() instead of all that pile. Looks like a good case for a followup series, if nothing else.