From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751149AbZG2ODT (ORCPT ); Wed, 29 Jul 2009 10:03:19 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750848AbZG2ODT (ORCPT ); Wed, 29 Jul 2009 10:03:19 -0400 Received: from mx2.redhat.com ([66.187.237.31]:36416 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750794AbZG2ODS (ORCPT ); Wed, 29 Jul 2009 10:03:18 -0400 Message-ID: <4A7057BE.6020900@redhat.com> Date: Wed, 29 Jul 2009 17:07:58 +0300 From: Avi Kivity User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1b3pre) Gecko/20090513 Fedora/3.0-2.3.beta2.fc11 Lightning/1.0pre Thunderbird/3.0b2 MIME-Version: 1.0 To: Joerg Roedel CC: Alexander Graf , kvm@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH 0/12] Nested SVM cleanups References: <1248872192-30881-1-git-send-email-joerg.roedel@amd.com> <4A70492E.2080705@redhat.com> <20090729130944.GE18313@amd.com> <4A704F5B.9000800@redhat.com> <20090729133840.GG18313@amd.com> <4A705327.8050101@redhat.com> <20090729134857.GL18313@amd.com> <4A7055D4.7000905@redhat.com> <20090729135859.GN18313@amd.com> In-Reply-To: <20090729135859.GN18313@amd.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 07/29/2009 04:58 PM, Joerg Roedel wrote: > On Wed, Jul 29, 2009 at 04:59:48PM +0300, Avi Kivity wrote: > >> On 07/29/2009 04:48 PM, Joerg Roedel wrote: >> >>> Performance isn't too bad even without this. At least if the host >>> supports nested paging :) But shadow nested paging should give nested >>> svm a real boost, thats very true. >>> >> Well, any workload that benefits from npt would benefit from nnpt a >> lot more. >> > > Most complicated part will be the two dimensional software page walker. > That will be a somewhat intrusive change (at least if I do it without > code duplication). But lets see how challenging this will become... > I think if you change the memory accessors to go through kvm_read_guest_virt() again, you'll get it almost for free. You'll need to supply a context so the accessors know which level you're accessing. -- error compiling committee.c: too many arguments to function