From patchwork Mon Aug 24 08:54:26 2009 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Gleb Natapov X-Patchwork-Id: 43576 Received: from vger.kernel.org (vger.kernel.org [209.132.176.167]) by demeter.kernel.org (8.14.2/8.14.2) with ESMTP id n7O8t0lM005378 for ; Mon, 24 Aug 2009 08:55:02 GMT Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752073AbZHXIyk (ORCPT ); Mon, 24 Aug 2009 04:54:40 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751816AbZHXIyj (ORCPT ); Mon, 24 Aug 2009 04:54:39 -0400 Received: from mx1.redhat.com ([209.132.183.28]:45426 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752047AbZHXIy2 (ORCPT ); Mon, 24 Aug 2009 04:54:28 -0400 Received: from int-mx06.intmail.prod.int.phx2.redhat.com (int-mx06.intmail.prod.int.phx2.redhat.com [10.5.11.19]) by mx1.redhat.com (8.13.8/8.13.8) with ESMTP id n7O8sURM007055 for ; Mon, 24 Aug 2009 04:54:30 -0400 Received: from dhcp-1-237.tlv.redhat.com (dhcp-1-237.tlv.redhat.com [10.35.1.237]) by int-mx06.intmail.prod.int.phx2.redhat.com (8.13.8/8.13.8) with ESMTP id n7O8sSwt002887; Mon, 24 Aug 2009 04:54:29 -0400 Received: by dhcp-1-237.tlv.redhat.com (Postfix, from userid 13519) id 9CD6118D470; Mon, 24 Aug 2009 11:54:26 +0300 (IDT) From: Gleb Natapov To: avi@redhat.com Cc: kvm@vger.kernel.org Subject: [PATCH v4 9/9] Drop kvm->irq_lock lock from irq injection path. Date: Mon, 24 Aug 2009 11:54:26 +0300 Message-Id: <1251104066-23504-10-git-send-email-gleb@redhat.com> In-Reply-To: <1251104066-23504-1-git-send-email-gleb@redhat.com> References: <1251104066-23504-1-git-send-email-gleb@redhat.com> X-Scanned-By: MIMEDefang 2.67 on 10.5.11.19 Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org The only thing it protects now is interrupt injection into lapic and this can work lockless. Even now with kvm->irq_lock in place access to lapic is not entirely serialized since vcpu access doesn't take kvm->irq_lock. Signed-off-by: Gleb Natapov --- arch/ia64/kvm/kvm-ia64.c | 2 -- arch/x86/kvm/i8254.c | 2 -- arch/x86/kvm/lapic.c | 2 -- arch/x86/kvm/x86.c | 2 -- virt/kvm/eventfd.c | 2 -- virt/kvm/irq_comm.c | 6 +----- virt/kvm/kvm_main.c | 2 -- 7 files changed, 1 insertions(+), 17 deletions(-) diff --git a/arch/ia64/kvm/kvm-ia64.c b/arch/ia64/kvm/kvm-ia64.c index 4a98314..f534e0f 100644 --- a/arch/ia64/kvm/kvm-ia64.c +++ b/arch/ia64/kvm/kvm-ia64.c @@ -982,10 +982,8 @@ long kvm_arch_vm_ioctl(struct file *filp, goto out; if (irqchip_in_kernel(kvm)) { __s32 status; - mutex_lock(&kvm->irq_lock); status = kvm_set_irq(kvm, KVM_USERSPACE_IRQ_SOURCE_ID, irq_event.irq, irq_event.level); - mutex_unlock(&kvm->irq_lock); if (ioctl == KVM_IRQ_LINE_STATUS) { irq_event.status = status; if (copy_to_user(argp, &irq_event, diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c index 82ad523..b857ca3 100644 --- a/arch/x86/kvm/i8254.c +++ b/arch/x86/kvm/i8254.c @@ -688,10 +688,8 @@ static void __inject_pit_timer_intr(struct kvm *kvm) struct kvm_vcpu *vcpu; int i; - mutex_lock(&kvm->irq_lock); kvm_set_irq(kvm, kvm->arch.vpit->irq_source_id, 0, 1); kvm_set_irq(kvm, kvm->arch.vpit->irq_source_id, 0, 0); - mutex_unlock(&kvm->irq_lock); /* * Provides NMI watchdog support via Virtual Wire mode. diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c index f24d4d0..e41e948 100644 --- a/arch/x86/kvm/lapic.c +++ b/arch/x86/kvm/lapic.c @@ -501,9 +501,7 @@ static void apic_send_ipi(struct kvm_lapic *apic) irq.trig_mode, irq.level, irq.dest_mode, irq.delivery_mode, irq.vector); - mutex_lock(&apic->vcpu->kvm->irq_lock); kvm_irq_delivery_to_apic(apic->vcpu->kvm, apic, &irq); - mutex_unlock(&apic->vcpu->kvm->irq_lock); } static u32 apic_get_tmcct(struct kvm_lapic *apic) diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index f49b2a1..fe68745 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -2268,10 +2268,8 @@ long kvm_arch_vm_ioctl(struct file *filp, goto out; if (irqchip_in_kernel(kvm)) { __s32 status; - mutex_lock(&kvm->irq_lock); status = kvm_set_irq(kvm, KVM_USERSPACE_IRQ_SOURCE_ID, irq_event.irq, irq_event.level); - mutex_unlock(&kvm->irq_lock); if (ioctl == KVM_IRQ_LINE_STATUS) { irq_event.status = status; if (copy_to_user(argp, &irq_event, diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c index 99017e8..95954ad 100644 --- a/virt/kvm/eventfd.c +++ b/virt/kvm/eventfd.c @@ -61,10 +61,8 @@ irqfd_inject(struct work_struct *work) struct _irqfd *irqfd = container_of(work, struct _irqfd, inject); struct kvm *kvm = irqfd->kvm; - mutex_lock(&kvm->irq_lock); kvm_set_irq(kvm, KVM_USERSPACE_IRQ_SOURCE_ID, irqfd->gsi, 1); kvm_set_irq(kvm, KVM_USERSPACE_IRQ_SOURCE_ID, irqfd->gsi, 0); - mutex_unlock(&kvm->irq_lock); } /* diff --git a/virt/kvm/irq_comm.c b/virt/kvm/irq_comm.c index 036a431..47c0194 100644 --- a/virt/kvm/irq_comm.c +++ b/virt/kvm/irq_comm.c @@ -82,8 +82,6 @@ int kvm_irq_delivery_to_apic(struct kvm *kvm, struct kvm_lapic *src, int i, r = -1; struct kvm_vcpu *vcpu, *lowest = NULL; - WARN_ON(!mutex_is_locked(&kvm->irq_lock)); - if (irq->dest_mode == 0 && irq->dest_id == 0xff && kvm_is_dm_lowest_prio(irq)) printk(KERN_INFO "kvm: apic: phys broadcast and lowest prio\n"); @@ -138,7 +136,7 @@ static int kvm_set_msi(struct kvm_kernel_irq_routing_entry *e, return kvm_irq_delivery_to_apic(kvm, NULL, &irq); } -/* This should be called with the kvm->irq_lock mutex held +/* * Return value: * < 0 Interrupt was ignored (masked or not delivered for other reasons) * = 0 Interrupt was coalesced (previous irq is still pending) @@ -153,8 +151,6 @@ int kvm_set_irq(struct kvm *kvm, int irq_source_id, u32 irq, int level) trace_kvm_set_irq(irq, level, irq_source_id); - WARN_ON(!mutex_is_locked(&kvm->irq_lock)); - /* Not possible to detect if the guest uses the PIC or the * IOAPIC. So set the bit in both. The guest will ignore * writes to the unused one. diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 783fa7c..1cb3cc4 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -137,7 +137,6 @@ static void kvm_assigned_dev_interrupt_work_handler(struct work_struct *work) interrupt_work); kvm = assigned_dev->kvm; - mutex_lock(&kvm->irq_lock); spin_lock_irq(&assigned_dev->assigned_dev_lock); if (assigned_dev->irq_requested_type & KVM_DEV_IRQ_HOST_MSIX) { struct kvm_guest_msix_entry *guest_entries = @@ -156,7 +155,6 @@ static void kvm_assigned_dev_interrupt_work_handler(struct work_struct *work) assigned_dev->guest_irq, 1); spin_unlock_irq(&assigned_dev->assigned_dev_lock); - mutex_unlock(&assigned_dev->kvm->irq_lock); } static irqreturn_t kvm_assigned_dev_intr(int irq, void *dev_id)