From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932398AbbAIOiX (ORCPT ); Fri, 9 Jan 2015 09:38:23 -0500 Received: from mx1.redhat.com ([209.132.183.28]:39072 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750989AbbAIOiU (ORCPT ); Fri, 9 Jan 2015 09:38:20 -0500 From: =?UTF-8?q?Radim=20Kr=C4=8Dm=C3=A1=C5=99?= To: linux-kernel@vger.kernel.org Cc: kvm@vger.kernel.org, Paolo Bonzini , Gleb Natapov , Feng Wu Subject: [PATCH] KVM: x86: amend APIC lowest priority arbitration Date: Fri, 9 Jan 2015 15:37:26 +0100 Message-Id: <1420814246-19033-1-git-send-email-rkrcmar@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Lowest priority should take the task priority into account. SDM 10.6.2.4 Lowest Priority Delivery Mode. (Too long to quote; second and last paragraphs are relevant.) Before this patch, we strived to have the same amount of handled lowest-priority interrupts on all VCPUs. This is only a complication, but kept for compatibility. Real modern Intels can't send lowest priority IPIs and the chipset directs external ones using processors' TPR. AMD still has rough edges. Signed-off-by: Radim Krčmář --- arch/x86/kvm/lapic.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c index a688fbffb34e..5b9d8c589bba 100644 --- a/arch/x86/kvm/lapic.c +++ b/arch/x86/kvm/lapic.c @@ -833,7 +833,15 @@ static int __apic_accept_irq(struct kvm_lapic *apic, int delivery_mode, int kvm_apic_compare_prio(struct kvm_vcpu *vcpu1, struct kvm_vcpu *vcpu2) { - return vcpu1->arch.apic_arb_prio - vcpu2->arch.apic_arb_prio; + /* XXX: AMD (2:16.6.2 Lowest Priority Messages and Arbitration) + * - uses the APR register (which also considers ISR and IRR), + * - chooses the highest APIC ID when APRs are identical, + * - and allows a focus processor. + * XXX: pseudo-balancing with apic_arb_prio is a KVM-specific feature + */ + int tpr = kvm_apic_get_reg(vcpu1->arch.apic, APIC_TASKPRI) - + kvm_apic_get_reg(vcpu2->arch.apic, APIC_TASKPRI); + return tpr ? : vcpu1->arch.apic_arb_prio - vcpu2->arch.apic_arb_prio; } static void kvm_ioapic_send_eoi(struct kvm_lapic *apic, int vector) -- 2.2.0