From mboxrd@z Thu Jan 1 00:00:00 1970 From: Radim =?utf-8?B?S3LEjW3DocWZ?= Subject: Re: [PATCH] KVM: x86: Add MSR_AMD64_DC_CFG to the list of ignored MSRs Date: Wed, 12 Apr 2017 21:16:40 +0200 Message-ID: <20170412191639.GF20145@potion> References: <20170406132220.17178-1-lprosek@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: kvm@vger.kernel.org, joro@8bytes.org To: Ladi Prosek Return-path: Received: from mx1.redhat.com ([209.132.183.28]:35256 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754206AbdDLTQo (ORCPT ); Wed, 12 Apr 2017 15:16:44 -0400 Content-Disposition: inline In-Reply-To: <20170406132220.17178-1-lprosek@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: 2017-04-06 15:22+0200, Ladi Prosek: > Hyper-V writes 0x800000000000 to MSR_AMD64_DC_CFG when running on AMD CPUs > as recommended in erratum 383, analogous to our svm_init_erratum_383. > > By ignoring the MSR, this patch enables running Hyper-V in L1 on AMD. > > Signed-off-by: Ladi Prosek > --- Dirty, but I don't see a reason to overengineer this ... Applied to kvm/queue, thanks. > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > @@ -2417,6 +2418,7 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info) > case MSR_FAM10H_MMIO_CONF_BASE: > case MSR_AMD64_BU_CFG2: > case MSR_IA32_PERF_CTL: > + case MSR_AMD64_DC_CFG: This will change L1 KVM's behavior to set erratum_383_found, but L0 will kill it anyway after hitting the errata in L2.