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] KVM: VMX: Dont' deliver posted IRQ if vCPU == this vCPU and vCPU is IN_GUEST_MODE
Date:   Thu,  6 Jan 2022 04:12:51 -0800
Message-Id: <1641471171-34232-1-git-send-email-wanpengli@tencent.com>
Precedence: bulk

From: Wanpeng Li <wanpengli@tencent.com>

Commit fdba608f15e2 (KVM: VMX: Wake vCPU when delivering posted IRQ even 
if vCPU == this vCPU) fixes wakeup event is missing when it is not from 
synchronous kvm context by dropping vcpu == running_vcpu checking completely.
However, it will break the original goal to optimise timer fastpath, let's 
move the checking under vCPU is IN_GUEST_MODE to restore the performance.

Suggested-by: Chao Gao <chao.gao@intel.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
---
 arch/x86/kvm/vmx/vmx.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index fe06b02..71e8afc 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -3932,7 +3932,8 @@ static inline void kvm_vcpu_trigger_posted_interrupt(struct kvm_vcpu *vcpu,
 		 * which has no effect is safe here.
 		 */
 
-		apic->send_IPI_mask(get_cpu_mask(vcpu->cpu), pi_vec);
+		if (vcpu != kvm_get_running_vcpu())
+			apic->send_IPI_mask(get_cpu_mask(vcpu->cpu), pi_vec);
 		return;
 	}
 #endif
-- 
2.7.4