Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Wanpeng Li <kernellwp@gmail.com>
To:     linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        =?UTF-8?q?Radim=20Kr=C4=8Dm=C3=A1=C5=99?= <rkrcmar@redhat.com>,
        Sean Christopherson <sean.j.christopherson@intel.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>
Subject: [PATCH v2] KVM: LAPIC: Loose fluctuation filter for auto tune lapic_timer_advance_ns
Date:   Thu, 26 Sep 2019 08:54:03 +0800
Message-Id: <1569459243-21950-1-git-send-email-wanpengli@tencent.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

From: Wanpeng Li <wanpengli@tencent.com>

5000 guest cycles delta is easy to encounter on desktop, per-vCPU 
lapic_timer_advance_ns always keeps at 1000ns initial value, lets 
loose fluctuation filter a bit to make auto tune can make some 
progress.

Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
---
v1 -> v2:
 * separate defines for ns vs cycles

 arch/x86/kvm/lapic.c | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index 3a3a685..7bfc3c0 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -66,9 +66,10 @@
 #define X2APIC_BROADCAST		0xFFFFFFFFul
 
 static bool lapic_timer_advance_dynamic __read_mostly;
-#define LAPIC_TIMER_ADVANCE_ADJUST_MIN 100
-#define LAPIC_TIMER_ADVANCE_ADJUST_MAX 5000
-#define LAPIC_TIMER_ADVANCE_ADJUST_INIT 1000
+#define LAPIC_TIMER_ADVANCE_EXPIRE_MIN	100
+#define LAPIC_TIMER_ADVANCE_EXPIRE_MAX	10000
+#define LAPIC_TIMER_ADVANCE_NS_INIT	1000
+#define LAPIC_TIMER_ADVANCE_NS_MAX     5000
 /* step-by-step approximation to mitigate fluctuation */
 #define LAPIC_TIMER_ADVANCE_ADJUST_STEP 8
 
@@ -1488,8 +1489,8 @@ static inline void adjust_lapic_timer_advance(struct kvm_vcpu *vcpu,
 	u64 ns;
 
 	/* Do not adjust for tiny fluctuations or large random spikes. */
-	if (abs(advance_expire_delta) > LAPIC_TIMER_ADVANCE_ADJUST_MAX ||
-	    abs(advance_expire_delta) < LAPIC_TIMER_ADVANCE_ADJUST_MIN)
+	if (abs(advance_expire_delta) > LAPIC_TIMER_ADVANCE_EXPIRE_MAX ||
+	    abs(advance_expire_delta) < LAPIC_TIMER_ADVANCE_EXPIRE_MIN)
 		return;
 
 	/* too early */
@@ -1504,8 +1505,8 @@ static inline void adjust_lapic_timer_advance(struct kvm_vcpu *vcpu,
 		timer_advance_ns += ns/LAPIC_TIMER_ADVANCE_ADJUST_STEP;
 	}
 
-	if (unlikely(timer_advance_ns > LAPIC_TIMER_ADVANCE_ADJUST_MAX))
-		timer_advance_ns = LAPIC_TIMER_ADVANCE_ADJUST_INIT;
+	if (unlikely(timer_advance_ns > LAPIC_TIMER_ADVANCE_NS_MAX))
+		timer_advance_ns = LAPIC_TIMER_ADVANCE_NS_INIT;
 	apic->lapic_timer.timer_advance_ns = timer_advance_ns;
 }
 
@@ -2302,7 +2303,7 @@ int kvm_create_lapic(struct kvm_vcpu *vcpu, int timer_advance_ns)
 		     HRTIMER_MODE_ABS_HARD);
 	apic->lapic_timer.timer.function = apic_timer_fn;
 	if (timer_advance_ns == -1) {
-		apic->lapic_timer.timer_advance_ns = LAPIC_TIMER_ADVANCE_ADJUST_INIT;
+		apic->lapic_timer.timer_advance_ns = LAPIC_TIMER_ADVANCE_NS_INIT;
 		lapic_timer_advance_dynamic = true;
 	} else {
 		apic->lapic_timer.timer_advance_ns = timer_advance_ns;
-- 
2.7.4