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:   Zhenzhong Duan <zhenzhong.duan@oracle.com>
To:     linux-kernel@vger.kernel.org
Cc:     pbonzini@redhat.com, rkrcmar@redhat.com,
        rafael.j.wysocki@intel.com, joao.m.martins@oracle.com,
        mtosatti@redhat.com, kvm@vger.kernel.org, linux-pm@vger.kernel.org,
        Zhenzhong Duan <zhenzhong.duan@oracle.com>
Subject: [PATCH RESEND v2 4/4] KVM: ensure vCPU halt_poll_us in right scope
Date:   Wed,  6 Nov 2019 19:55:02 +0800
Message-Id: <1573041302-4904-5-git-send-email-zhenzhong.duan@oracle.com>
In-Reply-To: <1573041302-4904-1-git-send-email-zhenzhong.duan@oracle.com>
References: <1573041302-4904-1-git-send-email-zhenzhong.duan@oracle.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

As user can adjust halt_poll_ns_grow_start and halt_poll_ns which
leads to vcpu->halt_poll_ns beyond the two boundaries. This patch
ensures vcpu->halt_poll_ns in that scope after growing or shrinking.

If halt_poll_ns_shrink is 0, shrink vcpu->halt_poll_ns to
halt_poll_ns_grow_start instead of 0. To disable poll we can set
halt_poll_ns to 0.

In case user wrongly set halt_poll_ns_grow_start > halt_poll_ns > 0,
halt_poll_ns take precedency and poll time is a fixed value of
halt_poll_ns.

This patch also simplifies branch check based on the guest haltpoll
code.

Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
---
 virt/kvm/kvm_main.c | 29 +++++++++++++----------------
 1 file changed, 13 insertions(+), 16 deletions(-)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 359516b..b4fca66 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2308,9 +2308,15 @@ static void shrink_halt_poll_ns(struct kvm_vcpu *vcpu)
 	old = val = vcpu->halt_poll_ns;
 	shrink = READ_ONCE(halt_poll_ns_shrink);
 	if (shrink == 0)
-		val = 0;
-	else
+		val = halt_poll_ns_grow_start;
+	else {
 		val /= shrink;
+		if (val < halt_poll_ns_grow_start)
+			val = halt_poll_ns_grow_start;
+	}
+
+	if (val > halt_poll_ns)
+		val = halt_poll_ns;
 
 	vcpu->halt_poll_ns = val;
 	trace_kvm_halt_poll_ns_shrink(vcpu->vcpu_id, val, old);
@@ -2385,21 +2391,12 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);
 
 	if (!kvm_arch_no_poll(vcpu)) {
-		if (!vcpu_valid_wakeup(vcpu)) {
+		/* we had a long block, shrink polling */
+		if (!vcpu_valid_wakeup(vcpu) || block_ns > halt_poll_ns)
 			shrink_halt_poll_ns(vcpu);
-		} else if (halt_poll_ns) {
-			if (block_ns <= vcpu->halt_poll_ns)
-				;
-			/* we had a long block, shrink polling */
-			else if (vcpu->halt_poll_ns && block_ns > halt_poll_ns)
-				shrink_halt_poll_ns(vcpu);
-			/* we had a short halt and our poll time is too small */
-			else if (vcpu->halt_poll_ns < halt_poll_ns &&
-				block_ns < halt_poll_ns)
-				grow_halt_poll_ns(vcpu);
-		} else {
-			vcpu->halt_poll_ns = 0;
-		}
+		/* we had a short block and our poll time is too small */
+		else if (block_ns > vcpu->halt_poll_ns)
+			grow_halt_poll_ns(vcpu);
 	}
 
 	trace_kvm_vcpu_wakeup(block_ns, waited, vcpu_valid_wakeup(vcpu));
-- 
1.8.3.1