From mboxrd@z Thu Jan 1 00:00:00 1970 From: Marcelo Tosatti Subject: Re: [PATCH] KVM: MMU: Disassociate direct maps from guest levels Date: Tue, 23 Mar 2010 10:59:44 -0300 Message-ID: <20100323135944.GB15811@amt.cnet> References: <1268554972-9827-1-git-send-email-avi@redhat.com> <20100317181625.GA10372@amt.cnet> <4BA89D23.1090005@redhat.com> <4BA89FCB.7000708@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Joerg Roedel , kvm@vger.kernel.org To: Avi Kivity Return-path: Received: from mx1.redhat.com ([209.132.183.28]:23786 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753520Ab0CWOAq (ORCPT ); Tue, 23 Mar 2010 10:00:46 -0400 Content-Disposition: inline In-Reply-To: <4BA89FCB.7000708@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: On Tue, Mar 23, 2010 at 01:02:35PM +0200, Avi Kivity wrote: > On 03/23/2010 12:51 PM, Avi Kivity wrote: > >On 03/17/2010 08:16 PM, Marcelo Tosatti wrote: > >>On Sun, Mar 14, 2010 at 10:22:52AM +0200, Avi Kivity wrote: > >>>Direct maps are linear translations for a section of memory, used for > >>>real mode or with large pages. As such, they are independent > >>>of the guest > >>>levels. > >>> > >>>Teach the mmu about this by making page->role.glevels = 0 for > >>>direct maps. > >>>This allows direct maps to be shared among real mode and the > >>>various paging > >>>modes. > >>> > >>>@@ -1328,6 +1328,8 @@ static struct kvm_mmu_page > >>>*kvm_mmu_get_page(struct kvm_vcpu *vcpu, > >>> role = vcpu->arch.mmu.base_role; > >>> role.level = level; > >>> role.direct = direct; > >>>+ if (role.direct) > >>>+ role.glevels = 0; > >>> role.access = access; > >>> if (vcpu->arch.mmu.root_level<= PT32_ROOT_LEVEL) { > >>> quadrant = gaddr>> (PAGE_SHIFT + (PT64_PT_BITS * level)); > >>>-- > >>>1.7.0.2 > >>Isnt this what happens already, since for tdp base_role.glevels is not > >>initialized? > > > >Correct. I was thinking about the nested npt case, which will use > >role.glevels. Joerg, please incorporate this into your nnpt > >patchset. > > > > Actually, incorrect. This patch is also applicable to non-tdp > (which is the same case as nnpt). It will allow reusing real-mode > direct maps and large page mappings from the various modes. It > won't make any performance difference, it's more a documentation aid > that direct maps aren't dependent on the guest at all, except for > permissions. Sure, makes sense.