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;
Content-Type: text/plain;
        charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 11.1 \(3445.4.7\))
Subject: Re: [PATCH v2 2/3] KVM: X86: Implement PV sched yield hypercall
From:   Liran Alon <liran.alon@oracle.com>
In-Reply-To: <1559004795-19927-3-git-send-email-wanpengli@tencent.com>
Date:   Wed, 29 May 2019 15:28:35 +0300
Cc:     linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
        Paolo Bonzini <pbonzini@redhat.com>,
        =?utf-8?B?UmFkaW0gS3LEjW3DocWZ?= <rkrcmar@redhat.com>
Content-Transfer-Encoding: quoted-printable
Message-Id: <D119BC9B-3097-4453-BC17-5AE532EDB995@oracle.com>
References: <1559004795-19927-1-git-send-email-wanpengli@tencent.com>
 <1559004795-19927-3-git-send-email-wanpengli@tencent.com>
To:     Wanpeng Li <kernellwp@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


> On 28 May 2019, at 3:53, Wanpeng Li <kernellwp@gmail.com> wrote:
>=20
> From: Wanpeng Li <wanpengli@tencent.com>
>=20
> The target vCPUs are in runnable state after vcpu_kick and suitable=20
> as a yield target. This patch implements the sched yield hypercall.
>=20
> 17% performace increase of ebizzy benchmark can be observed in an=20
> over-subscribe environment. (w/ kvm-pv-tlb disabled, testing TLB flush=20=

> call-function IPI-many since call-function is not easy to be trigged=20=

> by userspace workload).
>=20
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Radim Kr=C4=8Dm=C3=A1=C5=99 <rkrcmar@redhat.com>
> Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
> ---
> arch/x86/kvm/x86.c | 24 ++++++++++++++++++++++++
> 1 file changed, 24 insertions(+)
>=20
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index e7e57de..2ceef51 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -7172,6 +7172,26 @@ void kvm_vcpu_deactivate_apicv(struct kvm_vcpu =
*vcpu)
> 	kvm_x86_ops->refresh_apicv_exec_ctrl(vcpu);
> }
>=20
> +void kvm_sched_yield(struct kvm *kvm, u64 dest_id)
> +{
> +	struct kvm_vcpu *target;
> +	struct kvm_apic_map *map;
> +
> +	rcu_read_lock();
> +	map =3D rcu_dereference(kvm->arch.apic_map);
> +
> +	if (unlikely(!map))
> +		goto out;
> +

We should have a bounds-check here on =E2=80=9Cdest_id=E2=80=9D.

-Liran

> +	if (map->phys_map[dest_id]->vcpu) {
> +		target =3D map->phys_map[dest_id]->vcpu;
> +		kvm_vcpu_yield_to(target);
> +	}
> +
> +out:
> +	rcu_read_unlock();
> +}
> +
> int kvm_emulate_hypercall(struct kvm_vcpu *vcpu)
> {
> 	unsigned long nr, a0, a1, a2, a3, ret;
> @@ -7218,6 +7238,10 @@ int kvm_emulate_hypercall(struct kvm_vcpu =
*vcpu)
> 	case KVM_HC_SEND_IPI:
> 		ret =3D kvm_pv_send_ipi(vcpu->kvm, a0, a1, a2, a3, =
op_64_bit);
> 		break;
> +	case KVM_HC_SCHED_YIELD:
> +		kvm_sched_yield(vcpu->kvm, a0);
> +		ret =3D 0;
> +		break;
> 	default:
> 		ret =3D -KVM_ENOSYS;
> 		break;
> --=20
> 2.7.4
>=20