From mboxrd@z Thu Jan 1 00:00:00 1970 From: Vivek Goyal Subject: Re: Questions on block drivers, REQ_FLUSH and REQ_FUA Date: Wed, 25 May 2011 15:15:23 -0400 Message-ID: <20110525191523.GC6724@redhat.com> References: <8E109506CAF2B94185C68859@nimrod.local> <20110524223220.GA379@redhat.com> <20110525085950.GC10146@htj.dyndns.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Alex Bligh , linux-fsdevel@vger.kernel.org To: Tejun Heo Return-path: Received: from mx1.redhat.com ([209.132.183.28]:54917 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754577Ab1EYTP3 (ORCPT ); Wed, 25 May 2011 15:15:29 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Wed, May 25, 2011 at 06:43:14PM +0200, Tejun Heo wrote: [..] > > * Write data, REQ_FUA - Write should be completed before FLUSH is > issued - ie. the write data should be on platter along with previously > completed writes on bio completion. Tejun, I have a confusion regarding the following statement. "along with previously completed writes on bio completion" I thought REQ_FUA just gurantees that bio/req carrying REQ_FUA flag is on disk platter. It does not gurantee anything about other requests which completed before this bio/req and might still be in non volatile caches of device. Thanks Vivek