Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752557AbbETGls (ORCPT ); Wed, 20 May 2015 02:41:48 -0400 Received: from mail-wi0-f169.google.com ([209.85.212.169]:36843 "EHLO mail-wi0-f169.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750894AbbETGlq (ORCPT ); Wed, 20 May 2015 02:41:46 -0400 Message-ID: <555C2CA6.6040906@redhat.com> Date: Wed, 20 May 2015 08:41:42 +0200 From: Paolo Bonzini User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.5.0 MIME-Version: 1.0 To: "Zhang, Yang Z" , "Li, Liang Z" , "kvm@vger.kernel.org" , "linux-kernel@vger.kernel.org" CC: "gleb@kernel.or" , "tglx@linutronix.de" , "mingo@redhat.com" , "hpa@zytor.com" , "x86@kernel.org" Subject: Re: [PATCH] kvm/fpu: Enable eager restore kvm FPU for MPX References: <1432132539-6194-1-git-send-email-liang.z.li@intel.com> In-Reply-To: Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2256 Lines: 66 On 20/05/2015 07:20, Zhang, Yang Z wrote: > Li, Liang Z wrote on 2015-05-20: >> The MPX feature requires eager KVM FPU restore support. We have >> verified that MPX cannot work correctly with the current lazy KVM FPU >> restore mechanism. Eager KVM FPU restore should be enabled if the MPX >> feature is exposed to VM. >> >> Signed-off-by: Liang Li >> --- >> arch/x86/kvm/vmx.c | 2 ++ >> arch/x86/kvm/x86.c | 3 ++- >> 2 files changed, 4 insertions(+), 1 deletion(-) >> >> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c >> index f7b6168..e2cccbe 100644 >> --- a/arch/x86/kvm/vmx.c >> +++ b/arch/x86/kvm/vmx.c >> @@ -8445,6 +8445,8 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id) >> goto free_vmcs; >> } >> >> + if (vmx_mpx_supported()) >> + vmx_fpu_activate(&vmx->vcpu); >> return &vmx->vcpu; >> >> free_vmcs: > > Is it better to use guest_cpuid_has_mpx() instead of vmx_mpx_supported()? CPUID hasn't been set yet, so I think it is okay to key it on vmx_mpx_supported(). It will be deactivated soon afterwards. Or even do it unconditionally; just make sure to add a comment about it. >> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 5f38188..5993f5f >> 100644 >> --- a/arch/x86/kvm/x86.c >> +++ b/arch/x86/kvm/x86.c >> @@ -7060,7 +7060,8 @@ void kvm_put_guest_fpu(struct kvm_vcpu *vcpu) >> fpu_save_init(&vcpu->arch.guest_fpu); >> __kernel_fpu_end(); >> ++vcpu->stat.fpu_reload; >> - kvm_make_request(KVM_REQ_DEACTIVATE_FPU, vcpu); >> + if (!kvm_x86_ops->mpx_supported()) >> + kvm_make_request(KVM_REQ_DEACTIVATE_FPU, vcpu); >> trace_kvm_fpu(0); >> } This is a hotter path. Here it's definitely better to avoid the call to kvm_x86_ops->mpx_supported(). Especially because, with MPX enabled, you would call this on every userspace exit. Yang's suggestion of using CPUID is actually more valuable here. You could add a new field eager_fpu in kvm->arch and update it in kvm_update_cpuid. Thanks, Paolo -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/