| Message ID | 20181126154732.23025-5-vkuznets@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | x86/kvm/hyper-v: Implement Direct Mode for synthetic timers | expand |
On 26/11/18 16:47, Vitaly Kuznetsov wrote: > stimers_pending optimization only helps us to avoid multiple > kvm_make_request() calls. This doesn't happen very often and these > calls are very cheap in the first place, remove open-coded version of > stimer_mark_pending() from kvm_hv_notify_acked_sint(). I wouldn't say very cheap, but still relatively cheap compared to a vmexit. Paolo
On Mon, Nov 26, 2018 at 04:47:32PM +0100, Vitaly Kuznetsov wrote: > stimers_pending optimization only helps us to avoid multiple > kvm_make_request() calls. This doesn't happen very often and these > calls are very cheap in the first place, remove open-coded version of > stimer_mark_pending() from kvm_hv_notify_acked_sint(). Frankly speaking, I've yet to see a guest that configures more than one SynIC timer. So it was indeed a bit of overengineering in the first place. > Suggested-by: Paolo Bonzini <pbonzini@redhat.com> > Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com> > --- > arch/x86/kvm/hyperv.c | 12 +++--------- > 1 file changed, 3 insertions(+), 9 deletions(-) Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c index 9533133be566..e6a2a085644a 100644 --- a/arch/x86/kvm/hyperv.c +++ b/arch/x86/kvm/hyperv.c @@ -206,7 +206,7 @@ static void kvm_hv_notify_acked_sint(struct kvm_vcpu *vcpu, u32 sint) struct kvm_vcpu_hv_synic *synic = vcpu_to_synic(vcpu); struct kvm_vcpu_hv *hv_vcpu = vcpu_to_hv_vcpu(vcpu); struct kvm_vcpu_hv_stimer *stimer; - int gsi, idx, stimers_pending; + int gsi, idx; trace_kvm_hv_notify_acked_sint(vcpu->vcpu_id, sint); @@ -214,19 +214,13 @@ static void kvm_hv_notify_acked_sint(struct kvm_vcpu *vcpu, u32 sint) synic_clear_sint_msg_pending(synic, sint); /* Try to deliver pending Hyper-V SynIC timers messages */ - stimers_pending = 0; for (idx = 0; idx < ARRAY_SIZE(hv_vcpu->stimer); idx++) { stimer = &hv_vcpu->stimer[idx]; if (stimer->msg_pending && stimer->config.enable && !stimer->config.direct_mode && - stimer->config.sintx == sint) { - set_bit(stimer->index, - hv_vcpu->stimer_pending_bitmap); - stimers_pending++; - } + stimer->config.sintx == sint) + stimer_mark_pending(stimer, false); } - if (stimers_pending) - kvm_make_request(KVM_REQ_HV_STIMER, vcpu); idx = srcu_read_lock(&kvm->irq_srcu); gsi = atomic_read(&synic->sint_to_gsi[sint]);
stimers_pending optimization only helps us to avoid multiple kvm_make_request() calls. This doesn't happen very often and these calls are very cheap in the first place, remove open-coded version of stimer_mark_pending() from kvm_hv_notify_acked_sint(). Suggested-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com> --- arch/x86/kvm/hyperv.c | 12 +++--------- 1 file changed, 3 insertions(+), 9 deletions(-)