From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ipmail05.adl6.internode.on.net ([150.101.137.143]:49126 "EHLO ipmail05.adl6.internode.on.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751033AbcK1EBw (ORCPT ); Sun, 27 Nov 2016 23:01:52 -0500 Date: Mon, 28 Nov 2016 15:01:06 +1100 From: Dave Chinner Subject: Re: [PATCH 0/3] clean up speculative preallocation tracking and tagging Message-ID: <20161128040106.GY28177@dastard> References: <1479819662-64416-1-git-send-email-bfoster@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1479819662-64416-1-git-send-email-bfoster@redhat.com> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: Brian Foster Cc: linux-xfs@vger.kernel.org On Tue, Nov 22, 2016 at 08:00:59AM -0500, Brian Foster wrote: > Hi all, > > These are a few patches that resulted from the following discussion with > regard to how to tag reflink inodes correctly for COW fork > preallocation: > > http://www.spinics.net/lists/linux-xfs/msg02155.html > > The problem with the current code is that xfs_reflink_reserve_cow() > doesn't consider preallocation due to start offset alignment. The > problem with the first pass patch above is that the same function > doesn't distinguish between extent preallocation and extent merge. > > The solution in this series pushes down the incorporation of > preallocation to the bmapi call, where it already has the additional > context to identify whether an extent allocation was widened due to > cowextszhint alignment. The callers are still responsible for defining > how much to preallocate, to throttle, retry if necessary, etc. > > Note that this series is based on top of Christoph's recent extent > lookup cleanup patches. This survives xfstests for me on a reflink=1 fs. > Thoughts, reviews, flames appreciated. All in all looks like a nice little cleanup. Tests out just fine here, too, so consider it: Reviewed-by: Dave Chinner Cheers, Dave. -- Dave Chinner david@fromorbit.com