Merge branch 'kvm-updates/2.6.35' of git://git.kernel.org/pub/scm/virt/kvm/kvm

* 'kvm-updates/2.6.35' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: read apic->irr with ioapic lock held
  KVM: ia64: Add missing spin_unlock in kvm_arch_hardware_enable()
  KVM: Fix order passed to iommu_unmap
  KVM: MMU: Remove user access when allowing kernel access to gpte.w=0 page
  KVM: MMU: invalidate and flush on spte small->large page size change
  KVM: SVM: Implement workaround for Erratum 383
  KVM: SVM: Handle MCEs early in the vmexit process
  KVM: powerpc: fix init/exit annotation

arch/ia64/kvm/kvm-ia64.c

@@ -144,6 +144,7 @@ int kvm_arch_hardware_enable(void *garbage)
 				VP_INIT_ENV : VP_INIT_ENV_INITALIZE,
 			__pa(kvm_vm_buffer), KVM_VM_BUFFER_BASE, &tmp_base);
 	if (status != 0) {
+		spin_unlock(&vp_lock);
 		printk(KERN_WARNING"kvm: Failed to Enable VT Support!!!!\n");
 		return -EINVAL;
 	}

arch/powerpc/kvm/e500.c

@@ -164,7 +164,7 @@ static int __init kvmppc_e500_init(void)
 	return kvm_init(NULL, sizeof(struct kvmppc_vcpu_e500), 0, THIS_MODULE);
 }
 
-static void __init kvmppc_e500_exit(void)
+static void __exit kvmppc_e500_exit(void)
 {
 	kvmppc_booke_exit();
 }

arch/x86/include/asm/msr-index.h

@@ -110,6 +110,7 @@
 #define MSR_AMD64_PATCH_LOADER		0xc0010020
 #define MSR_AMD64_OSVW_ID_LENGTH	0xc0010140
 #define MSR_AMD64_OSVW_STATUS		0xc0010141
+#define MSR_AMD64_DC_CFG		0xc0011022
 #define MSR_AMD64_IBSFETCHCTL		0xc0011030
 #define MSR_AMD64_IBSFETCHLINAD		0xc0011031
 #define MSR_AMD64_IBSFETCHPHYSAD	0xc0011032

arch/x86/kvm/mmu.c

@@ -1815,6 +1815,9 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
 
 		spte |= PT_WRITABLE_MASK;
 
+		if (!tdp_enabled && !(pte_access & ACC_WRITE_MASK))
+			spte &= ~PT_USER_MASK;
+
 		/*
 		 * Optimization: for pte sync, if spte was writable the hash
 		 * lookup is unnecessary (and expensive). Write protection
@@ -1870,6 +1873,8 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
 
 			child = page_header(pte & PT64_BASE_ADDR_MASK);
 			mmu_page_remove_parent_pte(child, sptep);
+			__set_spte(sptep, shadow_trap_nonpresent_pte);
+			kvm_flush_remote_tlbs(vcpu->kvm);
 		} else if (pfn != spte_to_pfn(*sptep)) {
 			pgprintk("hfn old %lx new %lx\n",
 				 spte_to_pfn(*sptep), pfn);

arch/x86/kvm/svm.c

@@ -28,6 +28,7 @@
 #include <linux/ftrace_event.h>
 #include <linux/slab.h>
 
+#include <asm/tlbflush.h>
 #include <asm/desc.h>
 
 #include <asm/virtext.h>
@@ -56,6 +57,8 @@ MODULE_LICENSE("GPL");
 
 #define DEBUGCTL_RESERVED_BITS (~(0x3fULL))
 
+static bool erratum_383_found __read_mostly;
+
 static const u32 host_save_user_msrs[] = {
 #ifdef CONFIG_X86_64
 	MSR_STAR, MSR_LSTAR, MSR_CSTAR, MSR_SYSCALL_MASK, MSR_KERNEL_GS_BASE,
@@ -374,6 +377,31 @@ static void svm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr,
 	svm->vmcb->control.event_inj_err = error_code;
 }
 
+static void svm_init_erratum_383(void)
+{
+	u32 low, high;
+	int err;
+	u64 val;
+
+	/* Only Fam10h is affected */
+	if (boot_cpu_data.x86 != 0x10)
+		return;
+
+	/* Use _safe variants to not break nested virtualization */
+	val = native_read_msr_safe(MSR_AMD64_DC_CFG, &err);
+	if (err)
+		return;
+
+	val |= (1ULL << 47);
+
+	low  = lower_32_bits(val);
+	high = upper_32_bits(val);
+
+	native_write_msr_safe(MSR_AMD64_DC_CFG, low, high);
+
+	erratum_383_found = true;
+}
+
 static int has_svm(void)
 {
 	const char *msg;
@@ -429,6 +457,8 @@ static int svm_hardware_enable(void *garbage)
 
 	wrmsrl(MSR_VM_HSAVE_PA, page_to_pfn(sd->save_area) << PAGE_SHIFT);
 
+	svm_init_erratum_383();
+
 	return 0;
 }
 
@@ -1410,8 +1440,59 @@ static int nm_interception(struct vcpu_svm *svm)
 	return 1;
 }
 
-static int mc_interception(struct vcpu_svm *svm)
+static bool is_erratum_383(void)
 {
+	int err, i;
+	u64 value;
+
+	if (!erratum_383_found)
+		return false;
+
+	value = native_read_msr_safe(MSR_IA32_MC0_STATUS, &err);
+	if (err)
+		return false;
+
+	/* Bit 62 may or may not be set for this mce */
+	value &= ~(1ULL << 62);
+
+	if (value != 0xb600000000010015ULL)
+		return false;
+
+	/* Clear MCi_STATUS registers */
+	for (i = 0; i < 6; ++i)
+		native_write_msr_safe(MSR_IA32_MCx_STATUS(i), 0, 0);
+
+	value = native_read_msr_safe(MSR_IA32_MCG_STATUS, &err);
+	if (!err) {
+		u32 low, high;
+
+		value &= ~(1ULL << 2);
+		low    = lower_32_bits(value);
+		high   = upper_32_bits(value);
+
+		native_write_msr_safe(MSR_IA32_MCG_STATUS, low, high);
+	}
+
+	/* Flush tlb to evict multi-match entries */
+	__flush_tlb_all();
+
+	return true;
+}
+
+static void svm_handle_mce(struct vcpu_svm *svm)
+{
+	if (is_erratum_383()) {
+		/*
+		 * Erratum 383 triggered. Guest state is corrupt so kill the
+		 * guest.
+		 */
+		pr_err("KVM: Guest triggered AMD Erratum 383\n");
+
+		set_bit(KVM_REQ_TRIPLE_FAULT, &svm->vcpu.requests);
+
+		return;
+	}
+
 	/*
 	 * On an #MC intercept the MCE handler is not called automatically in
 	 * the host. So do it by hand here.
@@ -1420,6 +1501,11 @@ static int mc_interception(struct vcpu_svm *svm)
 		"int $0x12\n");
 	/* not sure if we ever come back to this point */
 
+	return;
+}
+
+static int mc_interception(struct vcpu_svm *svm)
+{
 	return 1;
 }
 
@@ -3088,6 +3174,14 @@ static void svm_vcpu_run(struct kvm_vcpu *vcpu)
 		vcpu->arch.regs_avail &= ~(1 << VCPU_EXREG_PDPTR);
 		vcpu->arch.regs_dirty &= ~(1 << VCPU_EXREG_PDPTR);
 	}
+
+	/*
+	 * We need to handle MC intercepts here before the vcpu has a chance to
+	 * change the physical cpu
+	 */
+	if (unlikely(svm->vmcb->control.exit_code ==
+		     SVM_EXIT_EXCP_BASE + MC_VECTOR))
+		svm_handle_mce(svm);
 }
 
 #undef R

virt/kvm/ioapic.c

@@ -192,12 +192,13 @@ static int ioapic_deliver(struct kvm_ioapic *ioapic, int irq)
 
 int kvm_ioapic_set_irq(struct kvm_ioapic *ioapic, int irq, int level)
 {
-	u32 old_irr = ioapic->irr;
+	u32 old_irr;
 	u32 mask = 1 << irq;
 	union kvm_ioapic_redirect_entry entry;
 	int ret = 1;
 
 	spin_lock(&ioapic->lock);
+	old_irr = ioapic->irr;
 	if (irq >= 0 && irq < IOAPIC_NUM_PINS) {
 		entry = ioapic->redirtbl[irq];
 		level ^= entry.fields.polarity;

virt/kvm/iommu.c

@@ -271,7 +271,7 @@ static void kvm_iommu_put_pages(struct kvm *kvm,
 		pfn  = phys >> PAGE_SHIFT;
 
 		/* Unmap address from IO address space */
-		order       = iommu_unmap(domain, gfn_to_gpa(gfn), PAGE_SIZE);
+		order       = iommu_unmap(domain, gfn_to_gpa(gfn), 0);
 		unmap_pages = 1ULL << order;
 
 		/* Unpin all pages we just unmapped to not leak any memory */