Merge branch 'kvm-updates/2.6.29' of git://git.kernel.org/pub/scm/virt/kvm/kvm

* 'kvm-updates/2.6.29' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: VMX: Flush volatile msrs before emulating rdmsr
  KVM: Fix assigned devices circular locking dependency
  KVM: x86: fix LAPIC pending count calculation
  KVM: Fix INTx for device assignment
  KVM: MMU: Map device MMIO as UC in EPT
  KVM: x86: disable kvmclock on non constant TSC hosts
  KVM: PIT: fix i8254 pending count read
  KVM: Fix racy in kvm_free_assigned_irq
  KVM: Add kvm_arch_sync_events to sync with asynchronize events
  KVM: mmu_notifiers release method
  KVM: Avoid using CONFIG_ in userspace visible headers
  KVM: ia64: fix fp fault/trap handler

arch/ia64/include/asm/kvm.h

@@ -25,6 +25,10 @@
 
 #include <linux/ioctl.h>
 
+/* Select x86 specific features in <linux/kvm.h> */
+#define __KVM_HAVE_IOAPIC
+#define __KVM_HAVE_DEVICE_ASSIGNMENT
+
 /* Architectural interrupt line count. */
 #define KVM_NR_INTERRUPTS 256
 

arch/ia64/kvm/kvm-ia64.c

@@ -1337,6 +1337,10 @@ static void kvm_release_vm_pages(struct kvm *kvm)
 	}
 }
 
+void kvm_arch_sync_events(struct kvm *kvm)
+{
+}
+
 void kvm_arch_destroy_vm(struct kvm *kvm)
 {
 	kvm_iommu_unmap_guest(kvm);

arch/ia64/kvm/process.c

@@ -455,11 +455,16 @@ fpswa_ret_t vmm_fp_emulate(int fp_fault, void *bundle, unsigned long *ipsr,
 	if (!vmm_fpswa_interface)
 		return (fpswa_ret_t) {-1, 0, 0, 0};
 
+	memset(&fp_state, 0, sizeof(fp_state_t));
+
 	/*
-	 * Just let fpswa driver to use hardware fp registers.
-	 * No fp register is valid in memory.
+	 * compute fp_state.  only FP registers f6 - f11 are used by the
+	 * vmm, so set those bits in the mask and set the low volatile
+	 * pointer to point to these registers.
 	 */
-	memset(&fp_state, 0, sizeof(fp_state_t));
+	fp_state.bitmask_low64 = 0xfc0;  /* bit6..bit11 */
+
+	fp_state.fp_state_low_volatile = (fp_state_low_volatile_t *) &regs->f6;
 
    /*
 	 * unsigned long (*EFI_FPSWA) (
@@ -545,10 +550,6 @@ void reflect_interruption(u64 ifa, u64 isr, u64 iim,
 		status = vmm_handle_fpu_swa(0, regs, isr);
 		if (!status)
 			return ;
-		else if (-EAGAIN == status) {
-			vcpu_decrement_iip(vcpu);
-			return ;
-		}
 		break;
 	}
 

arch/powerpc/kvm/powerpc.c

@@ -125,6 +125,10 @@ static void kvmppc_free_vcpus(struct kvm *kvm)
 	}
 }
 
+void kvm_arch_sync_events(struct kvm *kvm)
+{
+}
+
 void kvm_arch_destroy_vm(struct kvm *kvm)
 {
 	kvmppc_free_vcpus(kvm);

arch/s390/kvm/kvm-s390.c

@@ -212,6 +212,10 @@ static void kvm_free_vcpus(struct kvm *kvm)
 	}
 }
 
+void kvm_arch_sync_events(struct kvm *kvm)
+{
+}
+
 void kvm_arch_destroy_vm(struct kvm *kvm)
 {
 	kvm_free_vcpus(kvm);

arch/x86/include/asm/kvm.h

@@ -9,6 +9,13 @@
 #include <linux/types.h>
 #include <linux/ioctl.h>
 
+/* Select x86 specific features in <linux/kvm.h> */
+#define __KVM_HAVE_PIT
+#define __KVM_HAVE_IOAPIC
+#define __KVM_HAVE_DEVICE_ASSIGNMENT
+#define __KVM_HAVE_MSI
+#define __KVM_HAVE_USER_NMI
+
 /* Architectural interrupt line count. */
 #define KVM_NR_INTERRUPTS 256
 

arch/x86/kvm/i8254.c

@@ -207,7 +207,7 @@ static int __pit_timer_fn(struct kvm_kpit_state *ps)
 	hrtimer_add_expires_ns(&pt->timer, pt->period);
 	pt->scheduled = hrtimer_get_expires_ns(&pt->timer);
 	if (pt->period)
-		ps->channels[0].count_load_time = hrtimer_get_expires(&pt->timer);
+		ps->channels[0].count_load_time = ktime_get();
 
 	return (pt->period == 0 ? 0 : 1);
 }

arch/x86/kvm/irq.c

@@ -87,13 +87,6 @@ void kvm_inject_pending_timer_irqs(struct kvm_vcpu *vcpu)
 }
 EXPORT_SYMBOL_GPL(kvm_inject_pending_timer_irqs);
 
-void kvm_timer_intr_post(struct kvm_vcpu *vcpu, int vec)
-{
-	kvm_apic_timer_intr_post(vcpu, vec);
-	/* TODO: PIT, RTC etc. */
-}
-EXPORT_SYMBOL_GPL(kvm_timer_intr_post);
-
 void __kvm_migrate_timers(struct kvm_vcpu *vcpu)
 {
 	__kvm_migrate_apic_timer(vcpu);

arch/x86/kvm/irq.h

@@ -89,7 +89,6 @@ static inline int irqchip_in_kernel(struct kvm *kvm)
 
 void kvm_pic_reset(struct kvm_kpic_state *s);
 
-void kvm_timer_intr_post(struct kvm_vcpu *vcpu, int vec);
 void kvm_inject_pending_timer_irqs(struct kvm_vcpu *vcpu);
 void kvm_inject_apic_timer_irqs(struct kvm_vcpu *vcpu);
 void kvm_apic_nmi_wd_deliver(struct kvm_vcpu *vcpu);

arch/x86/kvm/lapic.c

@@ -35,6 +35,12 @@
 #include "kvm_cache_regs.h"
 #include "irq.h"
 
+#ifndef CONFIG_X86_64
+#define mod_64(x, y) ((x) - (y) * div64_u64(x, y))
+#else
+#define mod_64(x, y) ((x) % (y))
+#endif
+
 #define PRId64 "d"
 #define PRIx64 "llx"
 #define PRIu64 "u"
@@ -511,52 +517,22 @@ static void apic_send_ipi(struct kvm_lapic *apic)
 
 static u32 apic_get_tmcct(struct kvm_lapic *apic)
 {
-	u64 counter_passed;
-	ktime_t passed, now;
+	ktime_t remaining;
+	s64 ns;
 	u32 tmcct;
 
 	ASSERT(apic != NULL);
 
-	now = apic->timer.dev.base->get_time();
-	tmcct = apic_get_reg(apic, APIC_TMICT);
-
 	/* if initial count is 0, current count should also be 0 */
-	if (tmcct == 0)
+	if (apic_get_reg(apic, APIC_TMICT) == 0)
 		return 0;
 
-	if (unlikely(ktime_to_ns(now) <=
-		ktime_to_ns(apic->timer.last_update))) {
-		/* Wrap around */
-		passed = ktime_add(( {
-				    (ktime_t) {
-				    .tv64 = KTIME_MAX -
-				    (apic->timer.last_update).tv64}; }
-				   ), now);
-		apic_debug("time elapsed\n");
-	} else
-		passed = ktime_sub(now, apic->timer.last_update);
-
-	counter_passed = div64_u64(ktime_to_ns(passed),
-				   (APIC_BUS_CYCLE_NS * apic->timer.divide_count));
+	remaining = hrtimer_expires_remaining(&apic->timer.dev);
+	if (ktime_to_ns(remaining) < 0)
+		remaining = ktime_set(0, 0);
 
-	if (counter_passed > tmcct) {
-		if (unlikely(!apic_lvtt_period(apic))) {
-			/* one-shot timers stick at 0 until reset */
-			tmcct = 0;
-		} else {
-			/*
-			 * periodic timers reset to APIC_TMICT when they
-			 * hit 0. The while loop simulates this happening N
-			 * times. (counter_passed %= tmcct) would also work,
-			 * but might be slower or not work on 32-bit??
-			 */
-			while (counter_passed > tmcct)
-				counter_passed -= tmcct;
-			tmcct -= counter_passed;
-		}
-	} else {
-		tmcct -= counter_passed;
-	}
+	ns = mod_64(ktime_to_ns(remaining), apic->timer.period);
+	tmcct = div64_u64(ns, (APIC_BUS_CYCLE_NS * apic->timer.divide_count));
 
 	return tmcct;
 }
@@ -653,8 +629,6 @@ static void start_apic_timer(struct kvm_lapic *apic)
 {
 	ktime_t now = apic->timer.dev.base->get_time();
 
-	apic->timer.last_update = now;
-
 	apic->timer.period = apic_get_reg(apic, APIC_TMICT) *
 		    APIC_BUS_CYCLE_NS * apic->timer.divide_count;
 	atomic_set(&apic->timer.pending, 0);
@@ -1110,16 +1084,6 @@ void kvm_inject_apic_timer_irqs(struct kvm_vcpu *vcpu)
 	}
 }
 
-void kvm_apic_timer_intr_post(struct kvm_vcpu *vcpu, int vec)
-{
-	struct kvm_lapic *apic = vcpu->arch.apic;
-
-	if (apic && apic_lvt_vector(apic, APIC_LVTT) == vec)
-		apic->timer.last_update = ktime_add_ns(
-				apic->timer.last_update,
-				apic->timer.period);
-}
-
 int kvm_get_apic_interrupt(struct kvm_vcpu *vcpu)
 {
 	int vector = kvm_apic_has_interrupt(vcpu);

arch/x86/kvm/lapic.h

@@ -12,7 +12,6 @@ struct kvm_lapic {
 		atomic_t pending;
 		s64 period;	/* unit: ns */
 		u32 divide_count;
-		ktime_t last_update;
 		struct hrtimer dev;
 	} timer;
 	struct kvm_vcpu *vcpu;
@@ -42,7 +41,6 @@ void kvm_set_apic_base(struct kvm_vcpu *vcpu, u64 data);
 void kvm_apic_post_state_restore(struct kvm_vcpu *vcpu);
 int kvm_lapic_enabled(struct kvm_vcpu *vcpu);
 int kvm_lapic_find_highest_irr(struct kvm_vcpu *vcpu);
-void kvm_apic_timer_intr_post(struct kvm_vcpu *vcpu, int vec);
 
 void kvm_lapic_set_vapic_addr(struct kvm_vcpu *vcpu, gpa_t vapic_addr);
 void kvm_lapic_sync_from_vapic(struct kvm_vcpu *vcpu);

arch/x86/kvm/mmu.c

@@ -1698,8 +1698,13 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte,
 	if (largepage)
 		spte |= PT_PAGE_SIZE_MASK;
 	if (mt_mask) {
+		if (!kvm_is_mmio_pfn(pfn)) {
 			mt_mask = get_memory_type(vcpu, gfn) <<
 				kvm_x86_ops->get_mt_mask_shift();
+			mt_mask |= VMX_EPT_IGMT_BIT;
+		} else
+			mt_mask = MTRR_TYPE_UNCACHABLE <<
+				kvm_x86_ops->get_mt_mask_shift();
 		spte |= mt_mask;
 	}
 

arch/x86/kvm/svm.c

@@ -1600,7 +1600,6 @@ static void svm_intr_assist(struct kvm_vcpu *vcpu)
 	/* Okay, we can deliver the interrupt: grab it and update PIC state. */
 	intr_vector = kvm_cpu_get_interrupt(vcpu);
 	svm_inject_irq(svm, intr_vector);
-	kvm_timer_intr_post(vcpu, intr_vector);
 out:
 	update_cr8_intercept(vcpu);
 }

arch/x86/kvm/vmx.c

@@ -903,6 +903,7 @@ static int vmx_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata)
 		data = vmcs_readl(GUEST_SYSENTER_ESP);
 		break;
 	default:
+		vmx_load_host_state(to_vmx(vcpu));
 		msr = find_msr_entry(to_vmx(vcpu), msr_index);
 		if (msr) {
 			data = msr->data;
@@ -3285,7 +3286,6 @@ static void vmx_intr_assist(struct kvm_vcpu *vcpu)
 	}
 	if (vcpu->arch.interrupt.pending) {
 		vmx_inject_irq(vcpu, vcpu->arch.interrupt.nr);
-		kvm_timer_intr_post(vcpu, vcpu->arch.interrupt.nr);
 		if (kvm_cpu_has_interrupt(vcpu))
 			enable_irq_window(vcpu);
 	}
@@ -3687,8 +3687,7 @@ static int __init vmx_init(void)
 	if (vm_need_ept()) {
 		bypass_guest_pf = 0;
 		kvm_mmu_set_base_ptes(VMX_EPT_READABLE_MASK |
-			VMX_EPT_WRITABLE_MASK |
-			VMX_EPT_IGMT_BIT);
+			VMX_EPT_WRITABLE_MASK);
 		kvm_mmu_set_mask_ptes(0ull, 0ull, 0ull, 0ull,
 				VMX_EPT_EXECUTABLE_MASK,
 				VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT);

arch/x86/kvm/x86.c

@@ -967,7 +967,6 @@ int kvm_dev_ioctl_check_extension(long ext)
 	case KVM_CAP_MMU_SHADOW_CACHE_CONTROL:
 	case KVM_CAP_SET_TSS_ADDR:
 	case KVM_CAP_EXT_CPUID:
-	case KVM_CAP_CLOCKSOURCE:
 	case KVM_CAP_PIT:
 	case KVM_CAP_NOP_IO_DELAY:
 	case KVM_CAP_MP_STATE:
@@ -992,6 +991,9 @@ int kvm_dev_ioctl_check_extension(long ext)
 	case KVM_CAP_IOMMU:
 		r = iommu_found();
 		break;
+	case KVM_CAP_CLOCKSOURCE:
+		r = boot_cpu_has(X86_FEATURE_CONSTANT_TSC);
+		break;
 	default:
 		r = 0;
 		break;
@@ -4127,9 +4129,13 @@ static void kvm_free_vcpus(struct kvm *kvm)
 
 }
 
-void kvm_arch_destroy_vm(struct kvm *kvm)
+void kvm_arch_sync_events(struct kvm *kvm)
 {
 	kvm_free_all_assigned_devices(kvm);
+}
+
+void kvm_arch_destroy_vm(struct kvm *kvm)
+{
 	kvm_iommu_unmap_guest(kvm);
 	kvm_free_pit(kvm);
 	kfree(kvm->arch.vpic);

include/linux/kvm.h

@@ -58,10 +58,10 @@ struct kvm_irqchip {
 	__u32 pad;
         union {
 		char dummy[512];  /* reserving space */
-#ifdef CONFIG_X86
+#ifdef __KVM_HAVE_PIT
 		struct kvm_pic_state pic;
 #endif
-#if defined(CONFIG_X86) || defined(CONFIG_IA64)
+#ifdef __KVM_HAVE_IOAPIC
 		struct kvm_ioapic_state ioapic;
 #endif
 	} chip;
@@ -384,16 +384,16 @@ struct kvm_trace_rec {
 #define KVM_CAP_MP_STATE 14
 #define KVM_CAP_COALESCED_MMIO 15
 #define KVM_CAP_SYNC_MMU 16  /* Changes to host mmap are reflected in guest */
-#if defined(CONFIG_X86)||defined(CONFIG_IA64)
+#ifdef __KVM_HAVE_DEVICE_ASSIGNMENT
 #define KVM_CAP_DEVICE_ASSIGNMENT 17
 #endif
 #define KVM_CAP_IOMMU 18
-#if defined(CONFIG_X86)
+#ifdef __KVM_HAVE_MSI
 #define KVM_CAP_DEVICE_MSI 20
 #endif
 /* Bug in KVM_SET_USER_MEMORY_REGION fixed: */
 #define KVM_CAP_DESTROY_MEMORY_REGION_WORKS 21
-#if defined(CONFIG_X86)
+#ifdef __KVM_HAVE_USER_NMI
 #define KVM_CAP_USER_NMI 22
 #endif
 

include/linux/kvm_host.h

@@ -285,6 +285,7 @@ void kvm_free_physmem(struct kvm *kvm);
 struct  kvm *kvm_arch_create_vm(void);
 void kvm_arch_destroy_vm(struct kvm *kvm);
 void kvm_free_all_assigned_devices(struct kvm *kvm);
+void kvm_arch_sync_events(struct kvm *kvm);
 
 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
 int kvm_cpu_has_interrupt(struct kvm_vcpu *v);

virt/kvm/iommu.c

@@ -73,14 +73,13 @@ static int kvm_iommu_map_memslots(struct kvm *kvm)
 {
 	int i, r = 0;
 
-	down_read(&kvm->slots_lock);
 	for (i = 0; i < kvm->nmemslots; i++) {
 		r = kvm_iommu_map_pages(kvm, kvm->memslots[i].base_gfn,
 					kvm->memslots[i].npages);
 		if (r)
 			break;
 	}
-	up_read(&kvm->slots_lock);
+
 	return r;
 }
 
@@ -190,12 +189,11 @@ static void kvm_iommu_put_pages(struct kvm *kvm,
 static int kvm_iommu_unmap_memslots(struct kvm *kvm)
 {
 	int i;
-	down_read(&kvm->slots_lock);
+
 	for (i = 0; i < kvm->nmemslots; i++) {
 		kvm_iommu_put_pages(kvm, kvm->memslots[i].base_gfn,
 				    kvm->memslots[i].npages);
 	}
-	up_read(&kvm->slots_lock);
 
 	return 0;
 }

virt/kvm/kvm_main.c

@@ -173,7 +173,6 @@ static void kvm_assigned_dev_interrupt_work_handler(struct work_struct *work)
 		assigned_dev->host_irq_disabled = false;
 	}
 	mutex_unlock(&assigned_dev->kvm->lock);
-	kvm_put_kvm(assigned_dev->kvm);
 }
 
 static irqreturn_t kvm_assigned_dev_intr(int irq, void *dev_id)
@@ -181,8 +180,6 @@ static irqreturn_t kvm_assigned_dev_intr(int irq, void *dev_id)
 	struct kvm_assigned_dev_kernel *assigned_dev =
 		(struct kvm_assigned_dev_kernel *) dev_id;
 
-	kvm_get_kvm(assigned_dev->kvm);
-
 	schedule_work(&assigned_dev->interrupt_work);
 
 	disable_irq_nosync(irq);
@@ -213,6 +210,7 @@ static void kvm_assigned_dev_ack_irq(struct kvm_irq_ack_notifier *kian)
 	}
 }
 
+/* The function implicit hold kvm->lock mutex due to cancel_work_sync() */
 static void kvm_free_assigned_irq(struct kvm *kvm,
 				  struct kvm_assigned_dev_kernel *assigned_dev)
 {
@@ -228,11 +226,24 @@ static void kvm_free_assigned_irq(struct kvm *kvm,
 	if (!assigned_dev->irq_requested_type)
 		return;
 
-	if (cancel_work_sync(&assigned_dev->interrupt_work))
-		/* We had pending work. That means we will have to take
-		 * care of kvm_put_kvm.
+	/*
+	 * In kvm_free_device_irq, cancel_work_sync return true if:
+	 * 1. work is scheduled, and then cancelled.
+	 * 2. work callback is executed.
+	 *
+	 * The first one ensured that the irq is disabled and no more events
+	 * would happen. But for the second one, the irq may be enabled (e.g.
+	 * for MSI). So we disable irq here to prevent further events.
+	 *
+	 * Notice this maybe result in nested disable if the interrupt type is
+	 * INTx, but it's OK for we are going to free it.
+	 *
+	 * If this function is a part of VM destroy, please ensure that till
+	 * now, the kvm state is still legal for probably we also have to wait
+	 * interrupt_work done.
 	 */
-		kvm_put_kvm(kvm);
+	disable_irq_nosync(assigned_dev->host_irq);
+	cancel_work_sync(&assigned_dev->interrupt_work);
 
 	free_irq(assigned_dev->host_irq, (void *)assigned_dev);
 
@@ -285,8 +296,8 @@ static int assigned_device_update_intx(struct kvm *kvm,
 
 	if (irqchip_in_kernel(kvm)) {
 		if (!msi2intx &&
-		    adev->irq_requested_type & KVM_ASSIGNED_DEV_HOST_MSI) {
-			free_irq(adev->host_irq, (void *)kvm);
+		    (adev->irq_requested_type & KVM_ASSIGNED_DEV_HOST_MSI)) {
+			free_irq(adev->host_irq, (void *)adev);
 			pci_disable_msi(adev->dev);
 		}
 
@@ -455,6 +466,7 @@ static int kvm_vm_ioctl_assign_device(struct kvm *kvm,
 	struct kvm_assigned_dev_kernel *match;
 	struct pci_dev *dev;
 
+	down_read(&kvm->slots_lock);
 	mutex_lock(&kvm->lock);
 
 	match = kvm_find_assigned_dev(&kvm->arch.assigned_dev_head,
@@ -516,6 +528,7 @@ static int kvm_vm_ioctl_assign_device(struct kvm *kvm,
 
 out:
 	mutex_unlock(&kvm->lock);
+	up_read(&kvm->slots_lock);
 	return r;
 out_list_del:
 	list_del(&match->list);
@@ -527,6 +540,7 @@ static int kvm_vm_ioctl_assign_device(struct kvm *kvm,
 out_free:
 	kfree(match);
 	mutex_unlock(&kvm->lock);
+	up_read(&kvm->slots_lock);
 	return r;
 }
 #endif
@@ -789,11 +803,19 @@ static int kvm_mmu_notifier_clear_flush_young(struct mmu_notifier *mn,
 	return young;
 }
 
+static void kvm_mmu_notifier_release(struct mmu_notifier *mn,
+				     struct mm_struct *mm)
+{
+	struct kvm *kvm = mmu_notifier_to_kvm(mn);
+	kvm_arch_flush_shadow(kvm);
+}
+
 static const struct mmu_notifier_ops kvm_mmu_notifier_ops = {
 	.invalidate_page	= kvm_mmu_notifier_invalidate_page,
 	.invalidate_range_start	= kvm_mmu_notifier_invalidate_range_start,
 	.invalidate_range_end	= kvm_mmu_notifier_invalidate_range_end,
 	.clear_flush_young	= kvm_mmu_notifier_clear_flush_young,
+	.release		= kvm_mmu_notifier_release,
 };
 #endif /* CONFIG_MMU_NOTIFIER && KVM_ARCH_WANT_MMU_NOTIFIER */
 
@@ -883,6 +905,7 @@ static void kvm_destroy_vm(struct kvm *kvm)
 {
 	struct mm_struct *mm = kvm->mm;
 
+	kvm_arch_sync_events(kvm);
 	spin_lock(&kvm_lock);
 	list_del(&kvm->vm_list);
 	spin_unlock(&kvm_lock);