From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [RFC] relaxed barrier semantics Date: Fri, 30 Jul 2010 09:11:38 +0200 Message-ID: <20100730071138.GA6411__9664.82668086446$1280473936$gmane$org@lst.de> References: <4C4FF136.5000205@kernel.org> <20100728090025.GA9252@lst.de> <4C4FF592.9090800@kernel.org> <20100728092859.GA11096@lst.de> <20100729014431.GD4506@thunk.org> <4C51DA1F.2040701@redhat.com> <20100729194904.GA17098@lst.de> <4C51DCF1.3010507@redhat.com> <25F5E16E-968D-4FEF-8187-70453985B19B@dilger.ca> <20100729230406.GI4506@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from verein.lst.de ([213.95.11.210]:35276 "EHLO verein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755135Ab0G3HMJ (ORCPT ); Fri, 30 Jul 2010 03:12:09 -0400 Content-Disposition: inline In-Reply-To: <20100729230406.GI4506@thunk.org> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Ted Ts'o , Andreas Dilger , Ric Wheeler , Christoph Hellwig , Tejun Heo , Vivek Goyal On Thu, Jul 29, 2010 at 07:04:06PM -0400, Ted Ts'o wrote: > My understanding is that "everything FUA" can be a performance > disaster. That's because it bypasses the track buffer, and things get > written directly to disk. So there is no possibility to reorder > buffers so that they get written in one disk rotation. Depending on > the disk, it might even be that if you send N sequential sectors all > tagged with FUA, it could be slower than sending the N sectors > followed by a cache flush or SYNCHRONIZE_CACHE command. Not sure why the discussion is drifting in this direction again, but no one suggested to switch eweryone to forcefully use a FUA only primitive. If we offer a WRITE_FUA primitive to those who can make use of it, it won't mean the the cache flush primitive will go away - we will need it to implement fsync anyway. >