From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from verein.lst.de ([213.95.11.211]:59204 "EHLO newverein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753822AbdBEUNz (ORCPT ); Sun, 5 Feb 2017 15:13:55 -0500 Date: Sun, 5 Feb 2017 21:13:53 +0100 From: Christoph Hellwig Subject: Re: [PATCH 4/4] xfs: allocate direct I/O COW blocks in iomap_begin Message-ID: <20170205201353.GA12914@lst.de> References: <1485985359-971-1-git-send-email-hch@lst.de> <1485985359-971-5-git-send-email-hch@lst.de> <20170203005523.GQ9134@birch.djwong.org> <20170203105359.GA30020@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20170203105359.GA30020@lst.de> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: "Darrick J. Wong" Cc: Christoph Hellwig , linux-xfs@vger.kernel.org On Fri, Feb 03, 2017 at 11:53:59AM +0100, Christoph Hellwig wrote: > > I think this is incorrect -- we create an unwritten extent in the cow > > fork above (BMAP_COWFORK | BMAP_PREALLOC) but we no longer convert the > > part we're actually writing to a real extent. This means that > > _reflink_cow won't actually remap anything that we've written. > > It's not correct and I messed up when rebasing - the version I tested > still had it and worked. Sigh.. Or rather I hadn't finished rebasing and the version I had was totally b0rked. Proper one in progress now..