From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932195AbbLNScV (ORCPT ); Mon, 14 Dec 2015 13:32:21 -0500 Received: from mx1.redhat.com ([209.132.183.28]:33171 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932090AbbLNScR (ORCPT ); Mon, 14 Dec 2015 13:32:17 -0500 Date: Mon, 14 Dec 2015 16:32:14 -0200 From: Eduardo Habkost To: "Raj, Ashok" Cc: Borislav Petkov , kvm@vger.kernel.org, Tony Luck , Gong Chen , Gleb Natapov , linux-kernel@vger.kernel.org, qemu-devel@nongnu.org, Andi Kleen , Paolo Bonzini Subject: Re: [Qemu-devel] [Patch V2 1/2] x86, mce: Basic support to add LMCE support to QEMU Message-ID: <20151214183214.GG3774@thinpad.lan.raisama.net> References: <1449776482-26070-1-git-send-email-ashok.raj@intel.com> <20151214162356.GA5314@thinpad.lan.raisama.net> <20151214163738.GJ11630@pd.tnic> <20151214191145.GB18012@otc-brkl-03.jf.intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20151214191145.GB18012@otc-brkl-03.jf.intel.com> X-Fnord: you can see the fnord User-Agent: Mutt/1.5.23 (2014-03-12) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Dec 14, 2015 at 02:11:46PM -0500, Raj, Ashok wrote: > On Mon, Dec 14, 2015 at 05:37:38PM +0100, Borislav Petkov wrote: > > > > ... and obviously LMCE is vendor-specific so it cannot be enabled on > > !Intel guests with a define like that. mce_init() in qemu should check > > vendor too. > > > > The same mistake was done with SER_P but that's much harder to change, > > as we discussed previously. > > > > This is mostly harmless.. since the MCG_CAP space is shared and has no > conflict between vendors. Also just the CAP being set has no effect. > > The Guest OS needs to opt-in and the SIGBUS indicating SRAR are the only > ones that are treated special treatment. Also Intel was the only one > broadcasting MCE's.. so we are like the rest now :-) If the feature won't be enabled by default, I believe it will be OK to not make it conditional on CPUID vendor (as we would be simply doing that the user asked for). But if it's going to be enabled by default, I would like to get some assurance that there won't be conflicts between vendors in the MCG_CAP bits. -- Eduardo