linux-kernel - [PATCH v11 11/11] KVM: x86/pmu: Reduce the overhead of LBR passthrough or cancellation

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20200514083054.62538-12-like.xu@linux.intel.com>
Date:   Thu, 14 May 2020 16:30:54 +0800
From:   Like Xu <like.xu@...ux.intel.com>
To:     Peter Zijlstra <peterz@...radead.org>,
        Paolo Bonzini <pbonzini@...hat.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
        Sean Christopherson <sean.j.christopherson@...el.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Joerg Roedel <joro@...tes.org>,
        Thomas Gleixner <tglx@...utronix.de>, ak@...ux.intel.com,
        wei.w.wang@...el.com, Like Xu <like.xu@...ux.intel.com>
Subject: [PATCH v11 11/11] KVM: x86/pmu: Reduce the overhead of LBR passthrough or cancellation

After passthrough the LBR msrs to the guest, there is no need to call
intel_pmu_intercept_lbr_msrs() again and again, and vice versa.

If host reclaims the LBR between two availability checks, the interception
state and LBR records can be safely preserved due to native save/restore
support from guest LBR event.

Signed-off-by: Like Xu <like.xu@...ux.intel.com>
---
 arch/x86/include/asm/kvm_host.h | 3 +++
 arch/x86/kvm/pmu.c              | 1 +
 arch/x86/kvm/vmx/pmu_intel.c    | 9 +++++++++
 3 files changed, 13 insertions(+)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 57b281c4b196..dd51250c5688 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -511,6 +511,9 @@ struct kvm_pmu {
 	 * The records may be inaccurate if the host reclaims the LBR.
 	 */
 	struct perf_event *lbr_event;
+
+	/* A flag to reduce the overhead of LBR pass-through or cancellation. */
+	bool lbr_already_available;
 };
 
 struct kvm_pmu_ops;
diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c
index d0dece055605..583ecb5f5f7c 100644
--- a/arch/x86/kvm/pmu.c
+++ b/arch/x86/kvm/pmu.c
@@ -437,6 +437,7 @@ void kvm_pmu_init(struct kvm_vcpu *vcpu)
 	init_irq_work(&pmu->irq_work, kvm_pmi_trigger_fn);
 	pmu->event_count = 0;
 	pmu->need_cleanup = false;
+	pmu->lbr_already_available = false;
 	kvm_pmu_refresh(vcpu);
 }
 
diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c
index db185dca903d..408d80c9418b 100644
--- a/arch/x86/kvm/vmx/pmu_intel.c
+++ b/arch/x86/kvm/vmx/pmu_intel.c
@@ -316,6 +316,7 @@ static void intel_pmu_free_lbr_event(struct kvm_vcpu *vcpu)
 
 	perf_event_release_kernel(event);
 	intel_pmu_intercept_lbr_msrs(vcpu, true);
+	pmu->lbr_already_available = false;
 	pmu->event_count--;
 	pmu->lbr_event = NULL;
 }
@@ -653,10 +654,18 @@ static bool intel_pmu_lbr_is_availabile(struct kvm_vcpu *vcpu)
 	if (!pmu->lbr_event)
 		return false;
 
+	if (pmu->lbr_already_available && event_is_oncpu(pmu->lbr_event))
+		return true;
+
+	if (!pmu->lbr_already_available && !event_is_oncpu(pmu->lbr_event))
+		return false;
+
 	if (event_is_oncpu(pmu->lbr_event)) {
 		intel_pmu_intercept_lbr_msrs(vcpu, false);
+		pmu->lbr_already_available = true;
 	} else {
 		intel_pmu_intercept_lbr_msrs(vcpu, true);
+		pmu->lbr_already_available = false;
 		return false;
 	}
 
-- 
2.21.3