From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail191.messagelabs.com (mail191.messagelabs.com [216.82.242.19]) by kanga.kvack.org (Postfix) with ESMTP id 311E06B004F for ; Sun, 12 Jul 2009 16:22:31 -0400 (EDT) MIME-Version: 1.0 Message-ID: Date: Sun, 12 Jul 2009 13:39:07 -0700 (PDT) From: Dan Magenheimer Subject: RE: [Xen-devel] Re: [RFC PATCH 0/4] (Take 2): transcendent memory ("tmem") for Linux In-Reply-To: <4A5A1A51.2080301@redhat.com> Content-Type: text/plain; charset=Windows-1252 Content-Transfer-Encoding: quoted-printable Sender: owner-linux-mm@kvack.org To: Avi Kivity Cc: npiggin@suse.de, akpm@osdl.org, xen-devel@lists.xensource.com, tmem-devel@oss.oracle.com, kurt.hackel@oracle.com, Rusty Russell , jeremy@goop.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, sunil.mushran@oracle.com, chris.mason@oracle.com, Anthony Liguori , Schwidefsky , dave.mccracken@oracle.com, Marcelo Tosatti , alan@lxorguk.ukuu.org.uk, Balbir Singh List-ID: > CMM2 and tmem are not any different in this regard; both require OS=20 > modification, and both make information available to the=20 > hypervisor. In=20 > fact CMM2 is much more intrusive (but on the other hand provides much=20 > more information). > > > For those that believe it will be pervasive in the > > future, finding the right balance is a critical step > > in operating system evolution. >=20 > You're arguing for CMM2 here IMO. I'm arguing that both are a good thing and a step in the right direction. In some ways, tmem is a bigger step and in some ways CMM2 is a bigger step. > My take on this is that precache (predecache?) / preswap can be=20 > implemented even without tmem by using write-through backing for the=20 > virtual disk. For swap this is actually slight;y more efficient than=20 > tmem preswap, for preuncache slightly less efficient (since=20 > there will=20 > be some double caching). So I'm more interested in other use=20 > cases of tmem/CMM2. >=20 > Right, the transient uses of tmem when applied to disk objects=20 > (swap/pagecache) are very similar to disk caches. Which is=20 > why you can=20 > get a very similar effect when caching your virtual disks;=20 > this can be=20 > done without any guest modification. Write-through backing and virtual disk cacheing offer a similar effect, but it is far from the same. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org