Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Wanpeng Li <kernellwp@gmail.com>
To:     linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Sean Christopherson <seanjc@google.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>
Subject: [PATCH v4] KVM: Avoid expensive "should kick" operation for running vCPU
Date:   Wed, 20 Oct 2021 03:00:53 -0700
Message-Id: <1634724053-73627-1-git-send-email-wanpengli@tencent.com>
Precedence: bulk

From: Wanpeng Li <wanpengli@tencent.com>

Avoid the moderately expensive "should kick" operation if this pCPU
is currently running the target vCPU.

Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
---
v3 -> v4:
 * check running vCPU in a separate patch
v2 -> v3:
 * use kvm_arch_vcpu_get_wait()
v1 -> v2:
 * move checking running vCPU logic to kvm_vcpu_kick
 * check rcuwait_active(&vcpu->wait) etc

 virt/kvm/kvm_main.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 7851f3a1b5f7..4a4684e55ef5 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -3325,11 +3325,22 @@ void kvm_vcpu_kick(struct kvm_vcpu *vcpu)
 	 * vCPU also requires it to leave IN_GUEST_MODE.
 	 */
 	me = get_cpu();
+
+	/*
+	 * avoid the moderately expensive "should kick" operation if this pCPU
+	 * is currently running the target vcpu, in which case it's a KVM bug
+	 * if the vCPU is in the inner run loop.
+	 */
+	if (vcpu == __this_cpu_read(kvm_running_vcpu) &&
+	    !WARN_ON_ONCE(vcpu->mode == IN_GUEST_MODE))
+		goto out;
+
 	if (kvm_arch_vcpu_should_kick(vcpu)) {
 		cpu = READ_ONCE(vcpu->cpu);
 		if (cpu != me && (unsigned)cpu < nr_cpu_ids && cpu_online(cpu))
 			smp_send_reschedule(cpu);
 	}
+out:
 	put_cpu();
 }
 EXPORT_SYMBOL_GPL(kvm_vcpu_kick);
-- 
2.25.1