KVM: X86: Implement PV sched yield hypercall

The target vCPUs are in runnable state after vcpu_kick and suitable
as a yield target. This patch implements the sched yield hypercall.

17% performance increasement of ebizzy benchmark can be observed in an
over-subscribe environment. (w/ kvm-pv-tlb disabled, testing TLB flush
call-function IPI-many since call-function is not easy to be trigged
by userspace workload).

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Liran Alon <liran.alon@oracle.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 721af7b..2e302e9 100644 (file)
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -7193,6 +7193,23 @@ void kvm_vcpu_deactivate_apicv(struct kvm_vcpu *vcpu)
        kvm_x86_ops->refresh_apicv_exec_ctrl(vcpu);
 }
 
+static void kvm_sched_yield(struct kvm *kvm, unsigned long dest_id)
+{
+       struct kvm_vcpu *target = NULL;
+       struct kvm_apic_map *map;
+
+       rcu_read_lock();
+       map = rcu_dereference(kvm->arch.apic_map);
+
+       if (likely(map) && dest_id <= map->max_apic_id && map->phys_map[dest_id])
+               target = map->phys_map[dest_id]->vcpu;
+
+       rcu_read_unlock();
+
+       if (target)
+               kvm_vcpu_yield_to(target);
+}
+
 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu)
 {
        unsigned long nr, a0, a1, a2, a3, ret;
@@ -7239,6 +7256,10 @@ int kvm_emulate_hypercall(struct kvm_vcpu *vcpu)
        case KVM_HC_SEND_IPI:
                ret = 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 = 0;
+               break;
        default:
                ret = -KVM_ENOSYS;
                break;