From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH v5] KVM: VMX: Execute WBINVD to keep data consistency with assigned devices Date: Tue, 29 Jun 2010 15:37:53 +0300 Message-ID: <4C29E921.90304@redhat.com> References: <4C286CCE.10309@redhat.com> <1277781419-13227-1-git-send-email-sheng@linux.intel.com> <4C29BF58.20107@redhat.com> <4C29CBD1.1060604@siemens.com> <4C29CE31.4080101@redhat.com> <20100629123232.GD1788@amd.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Jan Kiszka , Sheng Yang , Marcelo Tosatti , "kvm@vger.kernel.org" , "Yaozu (Eddie) Dong" To: "Roedel, Joerg" Return-path: Received: from mx1.redhat.com ([209.132.183.28]:3250 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754591Ab0F2MiI (ORCPT ); Tue, 29 Jun 2010 08:38:08 -0400 In-Reply-To: <20100629123232.GD1788@amd.com> Sender: kvm-owner@vger.kernel.org List-ID: On 06/29/2010 03:32 PM, Roedel, Joerg wrote: > On Tue, Jun 29, 2010 at 06:42:57AM -0400, Avi Kivity wrote: > >> On 06/29/2010 01:32 PM, Jan Kiszka wrote: >> > >> Correctness is more important than performance. Since we don't know >> whether the guest needs it or not, we have to enable it. The user may >> disable it if he likes. >> > Can't this code only be enabled if VT-d hardware is detected that does > not support the snoop force bit? So the user does not have to struggle > with configuration options that are hard to understand for a > non-developer. > > It's already this way, that's why we test for KVM_IOMMU_CACHE_COHERENCY. Jan wants an override for older hardware which doesn't support a coherent iommu (the wbinvd ipi on task migrations, on processors that don't support wbinvd exits, is particularly expensive). -- error compiling committee.c: too many arguments to function