Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932827AbeAHTSE (ORCPT + 1 other); Mon, 8 Jan 2018 14:18:04 -0500 Received: from mail-io0-f193.google.com ([209.85.223.193]:43362 "EHLO mail-io0-f193.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932462AbeAHTSB (ORCPT ); Mon, 8 Jan 2018 14:18:01 -0500 X-Google-Smtp-Source: ACJfBovvXhBCf3CswXwZ3K8lAMJ/F0IIK9VxIYRTiK6QhiuG6LIKi6oTYiqJDahaETNtt08XGrfLnihMXSS3+GApUY4= MIME-Version: 1.0 In-Reply-To: <1515434925-10250-4-git-send-email-pbonzini@redhat.com> References: <1515434925-10250-1-git-send-email-pbonzini@redhat.com> <1515434925-10250-4-git-send-email-pbonzini@redhat.com> From: Jim Mattson Date: Mon, 8 Jan 2018 11:18:00 -0800 Message-ID: Subject: Re: [PATCH 3/7] kvm: vmx: pass MSR_IA32_SPEC_CTRL and MSR_IA32_PRED_CMD down to the guest To: Paolo Bonzini Cc: LKML , kvm list , aliguori@amazon.com, Tom Lendacky , dwmw@amazon.co.uk, bp@alien8.de Content-Type: text/plain; charset="UTF-8" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Return-Path: Guest usage of MSR_IA32_SPEC_CTRL and MSR_IA32_PRED_CMD should be predicated on guest_cpuid_has(vcpu, X86_FEATURE_SPEC_CTRL). On Mon, Jan 8, 2018 at 10:08 AM, Paolo Bonzini wrote: > Direct access to MSR_IA32_SPEC_CTRL and MSR_IA32_PRED_CMD is important > for performance. Allow load/store of MSR_IA32_SPEC_CTRL, restore guest > IBRS on VM entry and set it to 0 on VM exit (because Linux does not use > it yet). > > Signed-off-by: Paolo Bonzini > --- > arch/x86/kvm/vmx.c | 32 ++++++++++++++++++++++++++++++++ > 1 file changed, 32 insertions(+) > > diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c > index 669f5f74857d..d00bcad7336e 100644 > --- a/arch/x86/kvm/vmx.c > +++ b/arch/x86/kvm/vmx.c > @@ -120,6 +120,8 @@ > module_param_named(preemption_timer, enable_preemption_timer, bool, S_IRUGO); > #endif > > +static bool __read_mostly have_spec_ctrl; > + > #define KVM_GUEST_CR0_MASK (X86_CR0_NW | X86_CR0_CD) > #define KVM_VM_CR0_ALWAYS_ON_UNRESTRICTED_GUEST (X86_CR0_WP | X86_CR0_NE) > #define KVM_VM_CR0_ALWAYS_ON \ > @@ -609,6 +611,8 @@ struct vcpu_vmx { > u64 msr_host_kernel_gs_base; > u64 msr_guest_kernel_gs_base; > #endif > + u64 spec_ctrl; > + > u32 vm_entry_controls_shadow; > u32 vm_exit_controls_shadow; > u32 secondary_exec_control; > @@ -3361,6 +3365,9 @@ static int vmx_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info) > case MSR_IA32_TSC: > msr_info->data = guest_read_tsc(vcpu); > break; > + case MSR_IA32_SPEC_CTRL: > + msr_info->data = to_vmx(vcpu)->spec_ctrl; > + break; I have: if (!have_spec_ctrl || (!msr_info->host_initiated && !guest_cpuid_has(vcpu, X86_FEATURE_SPEC_CTRL))) return 1; msr_info->data = to_vmx(vcpu)->msr_ia32_spec_ctrl; break; > case MSR_IA32_SYSENTER_CS: > msr_info->data = vmcs_read32(GUEST_SYSENTER_CS); > break; > @@ -3500,6 +3507,9 @@ static int vmx_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info) > case MSR_IA32_TSC: > kvm_write_tsc(vcpu, msr_info); > break; > + case MSR_IA32_SPEC_CTRL: > + to_vmx(vcpu)->spec_ctrl = msr_info->data; > + break; I have: if (!have_spec_ctrl || (!msr_info->host_initiated && !guest_cpuid_has(vcpu, X86_FEATURE_SPEC_CTRL))) return 1; to_vmx(vcpu)->msr_ia32_spec_ctrl = data; break; > case MSR_IA32_CR_PAT: > if (vmcs_config.vmentry_ctrl & VM_ENTRY_LOAD_IA32_PAT) { > if (!kvm_mtrr_valid(vcpu, MSR_IA32_CR_PAT, data)) > @@ -7062,6 +7072,17 @@ static __init int hardware_setup(void) > goto out; > } > > + /* > + * FIXME: this is only needed until SPEC_CTRL is supported > + * by upstream Linux in cpufeatures, then it can be replaced > + * with static_cpu_has. > + */ > + have_spec_ctrl = cpu_has_spec_ctrl(); > + if (have_spec_ctrl) > + pr_info("kvm: SPEC_CTRL available\n"); > + else > + pr_info("kvm: SPEC_CTRL not available\n"); > + > if (boot_cpu_has(X86_FEATURE_NX)) > kvm_enable_efer_bits(EFER_NX); > > @@ -7131,6 +7152,8 @@ static __init int hardware_setup(void) > vmx_disable_intercept_for_msr(MSR_IA32_SYSENTER_CS, false); > vmx_disable_intercept_for_msr(MSR_IA32_SYSENTER_ESP, false); > vmx_disable_intercept_for_msr(MSR_IA32_SYSENTER_EIP, false); > + vmx_disable_intercept_for_msr(MSR_IA32_SPEC_CTRL, false); > + vmx_disable_intercept_for_msr(MSR_IA32_PRED_CMD, false); I have a lot of changes to MSR permission bitmap handling, but these intercepts should only be disabled when guest_cpuid_has(vcpu, X86_FEATURE_SPEC_CTRL). > memcpy(vmx_msr_bitmap_legacy_x2apic_apicv, > vmx_msr_bitmap_legacy, PAGE_SIZE); > @@ -9597,6 +9620,9 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu) > > pt_guest_enter(vmx); > > + if (have_spec_ctrl && vmx->spec_ctrl != 0) > + wrmsrl(MSR_IA32_SPEC_CTRL, vmx->spec_ctrl); > + Here, I have: /* * If the guest is allowed to write to MSR_IA32_SPEC_CTRL, * store it on VM-exit. */ if (guest_cpuid_has(vcpu, X86_FEATURE_SPEC_CTRL)) add_autostore_msr(vmx, MSR_IA32_SPEC_CTRL); else clear_autostore_msr(vmx, MSR_IA32_SPEC_CTRL); /* * If the guest's IA32_SPEC_CTRL MSR doesn't match the host's * IA32_SPEC_CTRL MSR, then add the MSR to the atomic switch * MSRs, so that the guest value will be loaded on VM-entry * and the host value will be loaded on VM-exit. */ if (vmx->msr_ia32_spec_ctrl != spec_ctrl_enabled()) add_atomic_switch_msr(vmx, MSR_IA32_SPEC_CTRL, vmx->msr_ia32_spec_ctrl, spec_ctrl_enabled()); else clear_atomic_switch_msr(vmx, MSR_IA32_SPEC_CTRL); > atomic_switch_perf_msrs(vmx); > > vmx_arm_hv_timer(vcpu); > @@ -9707,6 +9733,12 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu) > #endif > ); > > + if (have_spec_ctrl) { > + rdmsrl(MSR_IA32_SPEC_CTRL, vmx->spec_ctrl); > + if (vmx->spec_ctrl) > + wrmsrl(MSR_IA32_SPEC_CTRL, 0); > + } > + I know the VM-exit MSR load and store lists are probably slower, but I'm a little uncomfortable restoring the host's IA32_SPEC_CTRL MSR late if the guest has it clear and the host has it set. > /* MSR_IA32_DEBUGCTLMSR is zeroed on vmexit. Restore it if needed */ > if (vmx->host_debugctlmsr) > update_debugctlmsr(vmx->host_debugctlmsr); > -- > 1.8.3.1 > >