From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Wed, 3 Jan 2001 16:39:43 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Wed, 3 Jan 2001 16:39:34 -0500 Received: from a203-167-249-89.reverse.clear.net.nz ([203.167.249.89]:23565 "HELO metastasis.f00f.org") by vger.kernel.org with SMTP id ; Wed, 3 Jan 2001 16:39:28 -0500 Date: Thu, 4 Jan 2001 10:39:24 +1300 From: Chris Wedgwood To: Chris Mason Cc: Linus Torvalds , Marcelo Tosatti , Alexander Viro , Kernel Mailing List Subject: Re: [PATCH] filemap_fdatasync & related changes Message-ID: <20010104103924.B28974@metastasis.f00f.org> In-Reply-To: <466640000.978548970@tiny> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <466640000.978548970@tiny>; from mason@suse.com on Wed, Jan 03, 2001 at 02:09:30PM -0500 X-No-Archive: Yes Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Jan 03, 2001 at 02:09:30PM -0500, Chris Mason wrote: The problem with [mf]sync is that we really want to write the page before returning back to the application. We don't _want_ to -- we MUST. Some applications assume fsync and fdatasync work correctly; if they do not the all bests are off. This includes many MTAs, MDAs and databases (msync also being important). Looping somewhere before returning to the application is by far better than returning when some pages have yet to be flushed. --cw - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/