From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Roedel, Joerg" Subject: Re: [PATCH 0/3] KVM: Introduce VCPU-wide notion of guest-mode V2 Date: Wed, 1 Dec 2010 11:03:59 +0100 Message-ID: <20101201100359.GG2258@amd.com> References: <4CF3D093.40604@redhat.com> <1291049509-3430-1-git-send-email-joerg.roedel@amd.com> <20101201080149.GA29486@fermat.math.technion.ac.il> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Cc: Avi Kivity , Marcelo Tosatti , "kvm@vger.kernel.org" , "linux-kernel@vger.kernel.org" To: Nadav Har'El Return-path: Received: from db3ehsobe006.messaging.microsoft.com ([213.199.154.144]:48384 "EHLO DB3EHSOBE006.bigfish.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751755Ab0LAKEE (ORCPT ); Wed, 1 Dec 2010 05:04:04 -0500 Content-Disposition: inline In-Reply-To: <20101201080149.GA29486@fermat.math.technion.ac.il> Sender: kvm-owner@vger.kernel.org List-ID: On Wed, Dec 01, 2010 at 03:01:49AM -0500, Nadav Har'El wrote: > On Mon, Nov 29, 2010, Joerg Roedel wrote about "[PATCH 0/3] KVM: Introduce VCPU-wide notion of guest-mode V2": > > Hi Avi, Hi Marcelo, > > > > here is the re-spin I promised. The change to V1 are essentially the > > renames: > > > > kvm_vcpu_enter_gm -> enter_guest_mode > > kvm_vcpu_leave_gm -> leave_guest_mode > > kvm_vcpu_is_gm -> is_guest_mode > > I like this concept, and will be happy to change the nested VMX code to use > it as well. > > One small thing: After the name change, it might not be obvious on first > sight that these functions refer to the state of the vcpu, not the state > of the actual CPU (which, if you think about it, is never in guest mode while > KVM code is running ;-)). I think that a short comment before the definition > of these functions might be useful - perhaps saying that they pertain to a > hypervisor running in the vcpu (i.e., nested virtualization). Yes, right. Thats a good thing. I sent a follow-on patch adding the comments. Btw, another idea which came up recently was to concentrate the actuall vmexit emulation at a single point. Every code place which does the exit directly today will be changed to only set a request-bit and the real exit is then done later. Your code might already do this, I havn't checked. In fact the idea is from the neste-VMX patchset for Xen :) This would fit very well in the generic code because it already has request-bit infrastructure. What do you think, can nested VMX also make use of that too? Joerg -- AMD Operating System Research Center Advanced Micro Devices GmbH Einsteinring 24 85609 Dornach General Managers: Alberto Bozzo, Andrew Bowd Registration: Dornach, Landkr. Muenchen; Registerger. Muenchen, HRB Nr. 43632