From:   Like Xu <like.xu.linux@gmail.com>
To:     Sean Christopherson <seanjc@google.com>
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Ravi Bangoria <ravi.bangoria@amd.com>, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: [PATCH 1/5] KVM: x86/pmu: Emulate CTR overflow directly in kvm_pmu_handle_event()
Date:   Fri, 10 Mar 2023 18:53:42 +0800
Message-Id: <20230310105346.12302-2-likexu@tencent.com>
In-Reply-To: <20230310105346.12302-1-likexu@tencent.com>
References: <20230310105346.12302-1-likexu@tencent.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Like Xu <likexu@tencent.com>

More and more vPMU emulations are deferred to kvm_pmu_handle_event()
as it reduces the overhead of repeated execution. The reprogram_counter()
is only responsible for creating the required perf_event, and naturally
doesn't include the counter overflow part from emulated instruction events.

Prior to this change, the assignment of pmc->prev_counter always occurred
after pmc was enabled, and to keep the same semantics, pmc->prev_counter
always needed to be reset after it has taken effect.

No functional change intended.

Signed-off-by: Like Xu <likexu@tencent.com>
---
 arch/x86/kvm/pmu.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c
index a1a79b5f49d7..d1c89a6625a0 100644
--- a/arch/x86/kvm/pmu.c
+++ b/arch/x86/kvm/pmu.c
@@ -418,9 +418,6 @@ static void reprogram_counter(struct kvm_pmc *pmc)
 	if (!event_is_allowed(pmc))
 		goto reprogram_complete;
 
-	if (pmc->counter < pmc->prev_counter)
-		__kvm_perf_overflow(pmc, false);
-
 	if (eventsel & ARCH_PERFMON_EVENTSEL_PIN_CONTROL)
 		printk_once("kvm pmu: pin control bit is ignored\n");
 
@@ -458,6 +455,13 @@ static void reprogram_counter(struct kvm_pmc *pmc)
 
 reprogram_complete:
 	clear_bit(pmc->idx, (unsigned long *)&pmc_to_pmu(pmc)->reprogram_pmi);
+}
+
+static inline void kvm_pmu_handle_pmc_overflow(struct kvm_pmc *pmc)
+{
+	if (pmc->counter < pmc->prev_counter)
+		__kvm_perf_overflow(pmc, false);
+
 	pmc->prev_counter = 0;
 }
 
@@ -475,6 +479,7 @@ void kvm_pmu_handle_event(struct kvm_vcpu *vcpu)
 		}
 
 		reprogram_counter(pmc);
+		kvm_pmu_handle_pmc_overflow(pmc);
 	}
 
 	/*
-- 
2.39.2