Merge tag 'kvm-4.15-2' of git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull KVM updates from Radim Krčmář:
 "Trimmed second batch of KVM changes for Linux 4.15:

   - GICv4 Support for KVM/ARM

   - re-introduce support for CPUs without virtual NMI (cc stable) and
     allow testing of KVM without virtual NMI on available CPUs

   - fix long-standing performance issues with assigned devices on AMD
     (cc stable)"

* tag 'kvm-4.15-2' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (30 commits)
  kvm: vmx: Allow disabling virtual NMI support
  kvm: vmx: Reinstate support for CPUs without virtual NMI
  KVM: SVM: obey guest PAT
  KVM: arm/arm64: Don't queue VLPIs on INV/INVALL
  KVM: arm/arm64: Fix GICv4 ITS initialization issues
  KVM: arm/arm64: GICv4: Theory of operations
  KVM: arm/arm64: GICv4: Enable VLPI support
  KVM: arm/arm64: GICv4: Prevent userspace from changing doorbell affinity
  KVM: arm/arm64: GICv4: Prevent a VM using GICv4 from being saved
  KVM: arm/arm64: GICv4: Enable virtual cpuif if VLPIs can be delivered
  KVM: arm/arm64: GICv4: Hook vPE scheduling into vgic flush/sync
  KVM: arm/arm64: GICv4: Use the doorbell interrupt as an unblocking source
  KVM: arm/arm64: GICv4: Add doorbell interrupt handling
  KVM: arm/arm64: GICv4: Use pending_last as a scheduling hint
  KVM: arm/arm64: GICv4: Handle INVALL applied to a vPE
  KVM: arm/arm64: GICv4: Propagate property updates to VLPIs
  KVM: arm/arm64: GICv4: Handle MOVALL applied to a vPE
  KVM: arm/arm64: GICv4: Handle CLEAR applied to a VLPI
  KVM: arm/arm64: GICv4: Propagate affinity changes to the physical ITS
  KVM: arm/arm64: GICv4: Unmap VLPI when freeing an LPI
  ...

Documentation/admin-guide/kernel-parameters.txt

@@ -1890,6 +1890,10 @@
 			[KVM,ARM] Trap guest accesses to GICv3 common
 			system registers
 
+	kvm-arm.vgic_v4_enable=
+			[KVM,ARM] Allow use of GICv4 for direct injection of
+			LPIs.
+
 	kvm-intel.ept=	[KVM,Intel] Disable extended page tables
 			(virtualized MMU) support on capable Intel chips.
 			Default is 1 (enabled)

Documentation/virtual/kvm/devices/arm-vgic-its.txt

@@ -64,6 +64,8 @@ Groups:
     -EINVAL: Inconsistent restored data
     -EFAULT: Invalid guest ram access
     -EBUSY:  One or more VCPUS are running
+    -EACCES: The virtual ITS is backed by a physical GICv4 ITS, and the
+	     state is not available
 
   KVM_DEV_ARM_VGIC_GRP_ITS_REGS
   Attributes:

arch/arm/kvm/Kconfig

@@ -4,6 +4,7 @@
 #
 
 source "virt/kvm/Kconfig"
+source "virt/lib/Kconfig"
 
 menuconfig VIRTUALIZATION
 	bool "Virtualization"
@@ -23,6 +24,8 @@ config KVM
 	select PREEMPT_NOTIFIERS
 	select ANON_INODES
 	select ARM_GIC
+	select ARM_GIC_V3
+	select ARM_GIC_V3_ITS
 	select HAVE_KVM_CPU_RELAX_INTERCEPT
 	select HAVE_KVM_ARCH_TLB_FLUSH_ALL
 	select KVM_MMIO
@@ -36,6 +39,8 @@ config KVM
 	select HAVE_KVM_IRQCHIP
 	select HAVE_KVM_IRQ_ROUTING
 	select HAVE_KVM_MSI
+	select IRQ_BYPASS_MANAGER
+	select HAVE_KVM_IRQ_BYPASS
 	depends on ARM_VIRT_EXT && ARM_LPAE && ARM_ARCH_TIMER
 	---help---
 	  Support hosting virtualized guest machines.

arch/arm/kvm/Makefile

@@ -32,6 +32,7 @@ obj-y += $(KVM)/arm/vgic/vgic-init.o
 obj-y += $(KVM)/arm/vgic/vgic-irqfd.o
 obj-y += $(KVM)/arm/vgic/vgic-v2.o
 obj-y += $(KVM)/arm/vgic/vgic-v3.o
+obj-y += $(KVM)/arm/vgic/vgic-v4.o
 obj-y += $(KVM)/arm/vgic/vgic-mmio.o
 obj-y += $(KVM)/arm/vgic/vgic-mmio-v2.o
 obj-y += $(KVM)/arm/vgic/vgic-mmio-v3.o

arch/arm64/kvm/Kconfig

@@ -4,6 +4,7 @@
 #
 
 source "virt/kvm/Kconfig"
+source "virt/lib/Kconfig"
 
 menuconfig VIRTUALIZATION
 	bool "Virtualization"
@@ -36,6 +37,8 @@ config KVM
 	select HAVE_KVM_MSI
 	select HAVE_KVM_IRQCHIP
 	select HAVE_KVM_IRQ_ROUTING
+	select IRQ_BYPASS_MANAGER
+	select HAVE_KVM_IRQ_BYPASS
 	---help---
 	  Support hosting virtualized guest machines.
 	  We don't support KVM with 16K page tables yet, due to the multiple

arch/arm64/kvm/Makefile

@@ -27,6 +27,7 @@ kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-init.o
 kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-irqfd.o
 kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-v2.o
 kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-v3.o
+kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-v4.o
 kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-mmio.o
 kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-mmio-v2.o
 kvm-$(CONFIG_KVM_ARM_HOST) += $(KVM)/arm/vgic/vgic-mmio-v3.o

arch/x86/kvm/svm.c

@@ -3671,6 +3671,13 @@ static int svm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr)
 	u32 ecx = msr->index;
 	u64 data = msr->data;
 	switch (ecx) {
+	case MSR_IA32_CR_PAT:
+		if (!kvm_mtrr_valid(vcpu, MSR_IA32_CR_PAT, data))
+			return 1;
+		vcpu->arch.pat = data;
+		svm->vmcb->save.g_pat = data;
+		mark_dirty(svm->vmcb, VMCB_NPT);
+		break;
 	case MSR_IA32_TSC:
 		kvm_write_tsc(vcpu, msr);
 		break;

arch/x86/kvm/vmx.c

@@ -70,6 +70,9 @@ MODULE_DEVICE_TABLE(x86cpu, vmx_cpu_id);
 static bool __read_mostly enable_vpid = 1;
 module_param_named(vpid, enable_vpid, bool, 0444);
 
+static bool __read_mostly enable_vnmi = 1;
+module_param_named(vnmi, enable_vnmi, bool, S_IRUGO);
+
 static bool __read_mostly flexpriority_enabled = 1;
 module_param_named(flexpriority, flexpriority_enabled, bool, S_IRUGO);
 
@@ -202,6 +205,10 @@ struct loaded_vmcs {
 	bool nmi_known_unmasked;
 	unsigned long vmcs_host_cr3;	/* May not match real cr3 */
 	unsigned long vmcs_host_cr4;	/* May not match real cr4 */
+	/* Support for vnmi-less CPUs */
+	int soft_vnmi_blocked;
+	ktime_t entry_time;
+	s64 vnmi_blocked_time;
 	struct list_head loaded_vmcss_on_cpu_link;
 };
 
@@ -1291,6 +1298,11 @@ static inline bool cpu_has_vmx_invpcid(void)
 		SECONDARY_EXEC_ENABLE_INVPCID;
 }
 
+static inline bool cpu_has_virtual_nmis(void)
+{
+	return vmcs_config.pin_based_exec_ctrl & PIN_BASED_VIRTUAL_NMIS;
+}
+
 static inline bool cpu_has_vmx_wbinvd_exit(void)
 {
 	return vmcs_config.cpu_based_2nd_exec_ctrl &
@@ -1348,11 +1360,6 @@ static inline bool nested_cpu_has2(struct vmcs12 *vmcs12, u32 bit)
 		(vmcs12->secondary_vm_exec_control & bit);
 }
 
-static inline bool nested_cpu_has_virtual_nmis(struct vmcs12 *vmcs12)
-{
-	return vmcs12->pin_based_vm_exec_control & PIN_BASED_VIRTUAL_NMIS;
-}
-
 static inline bool nested_cpu_has_preemption_timer(struct vmcs12 *vmcs12)
 {
 	return vmcs12->pin_based_vm_exec_control &
@@ -3712,9 +3719,9 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 				&_vmexit_control) < 0)
 		return -EIO;
 
-	min = PIN_BASED_EXT_INTR_MASK | PIN_BASED_NMI_EXITING |
-		PIN_BASED_VIRTUAL_NMIS;
-	opt = PIN_BASED_POSTED_INTR | PIN_BASED_VMX_PREEMPTION_TIMER;
+	min = PIN_BASED_EXT_INTR_MASK | PIN_BASED_NMI_EXITING;
+	opt = PIN_BASED_VIRTUAL_NMIS | PIN_BASED_POSTED_INTR |
+		 PIN_BASED_VMX_PREEMPTION_TIMER;
 	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PINBASED_CTLS,
 				&_pin_based_exec_control) < 0)
 		return -EIO;
@@ -5232,6 +5239,10 @@ static u32 vmx_pin_based_exec_ctrl(struct vcpu_vmx *vmx)
 
 	if (!kvm_vcpu_apicv_active(&vmx->vcpu))
 		pin_based_exec_ctrl &= ~PIN_BASED_POSTED_INTR;
+
+	if (!enable_vnmi)
+		pin_based_exec_ctrl &= ~PIN_BASED_VIRTUAL_NMIS;
+
 	/* Enable the preemption timer dynamically */
 	pin_based_exec_ctrl &= ~PIN_BASED_VMX_PREEMPTION_TIMER;
 	return pin_based_exec_ctrl;
@@ -5666,7 +5677,8 @@ static void enable_irq_window(struct kvm_vcpu *vcpu)
 
 static void enable_nmi_window(struct kvm_vcpu *vcpu)
 {
-	if (vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) & GUEST_INTR_STATE_STI) {
+	if (!enable_vnmi ||
+	    vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) & GUEST_INTR_STATE_STI) {
 		enable_irq_window(vcpu);
 		return;
 	}
@@ -5706,6 +5718,19 @@ static void vmx_inject_nmi(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 
+	if (!enable_vnmi) {
+		/*
+		 * Tracking the NMI-blocked state in software is built upon
+		 * finding the next open IRQ window. This, in turn, depends on
+		 * well-behaving guests: They have to keep IRQs disabled at
+		 * least as long as the NMI handler runs. Otherwise we may
+		 * cause NMI nesting, maybe breaking the guest. But as this is
+		 * highly unlikely, we can live with the residual risk.
+		 */
+		vmx->loaded_vmcs->soft_vnmi_blocked = 1;
+		vmx->loaded_vmcs->vnmi_blocked_time = 0;
+	}
+
 	++vcpu->stat.nmi_injections;
 	vmx->loaded_vmcs->nmi_known_unmasked = false;
 
@@ -5724,6 +5749,8 @@ static bool vmx_get_nmi_mask(struct kvm_vcpu *vcpu)
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	bool masked;
 
+	if (!enable_vnmi)
+		return vmx->loaded_vmcs->soft_vnmi_blocked;
 	if (vmx->loaded_vmcs->nmi_known_unmasked)
 		return false;
 	masked = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) & GUEST_INTR_STATE_NMI;
@@ -5735,20 +5762,31 @@ static void vmx_set_nmi_mask(struct kvm_vcpu *vcpu, bool masked)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 
-	vmx->loaded_vmcs->nmi_known_unmasked = !masked;
-	if (masked)
-		vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO,
-			      GUEST_INTR_STATE_NMI);
-	else
-		vmcs_clear_bits(GUEST_INTERRUPTIBILITY_INFO,
-				GUEST_INTR_STATE_NMI);
+	if (!enable_vnmi) {
+		if (vmx->loaded_vmcs->soft_vnmi_blocked != masked) {
+			vmx->loaded_vmcs->soft_vnmi_blocked = masked;
+			vmx->loaded_vmcs->vnmi_blocked_time = 0;
+		}
+	} else {
+		vmx->loaded_vmcs->nmi_known_unmasked = !masked;
+		if (masked)
+			vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO,
+				      GUEST_INTR_STATE_NMI);
+		else
+			vmcs_clear_bits(GUEST_INTERRUPTIBILITY_INFO,
+					GUEST_INTR_STATE_NMI);
+	}
 }
 
 static int vmx_nmi_allowed(struct kvm_vcpu *vcpu)
 {
 	if (to_vmx(vcpu)->nested.nested_run_pending)
 		return 0;
 
+	if (!enable_vnmi &&
+	    to_vmx(vcpu)->loaded_vmcs->soft_vnmi_blocked)
+		return 0;
+
 	return	!(vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) &
 		  (GUEST_INTR_STATE_MOV_SS | GUEST_INTR_STATE_STI
 		   | GUEST_INTR_STATE_NMI));
@@ -6476,6 +6514,7 @@ static int handle_ept_violation(struct kvm_vcpu *vcpu)
 	 * AAK134, BY25.
 	 */
 	if (!(to_vmx(vcpu)->idt_vectoring_info & VECTORING_INFO_VALID_MASK) &&
+			enable_vnmi &&
 			(exit_qualification & INTR_INFO_UNBLOCK_NMI))
 		vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO, GUEST_INTR_STATE_NMI);
 
@@ -6535,6 +6574,7 @@ static int handle_ept_misconfig(struct kvm_vcpu *vcpu)
 
 static int handle_nmi_window(struct kvm_vcpu *vcpu)
 {
+	WARN_ON_ONCE(!enable_vnmi);
 	vmcs_clear_bits(CPU_BASED_VM_EXEC_CONTROL,
 			CPU_BASED_VIRTUAL_NMI_PENDING);
 	++vcpu->stat.nmi_window_exits;
@@ -6758,6 +6798,9 @@ static __init int hardware_setup(void)
 	if (!cpu_has_vmx_flexpriority())
 		flexpriority_enabled = 0;
 
+	if (!cpu_has_virtual_nmis())
+		enable_vnmi = 0;
+
 	/*
 	 * set_apic_access_page_addr() is used to reload apic access
 	 * page upon invalidation.  No need to do anything if not
@@ -6962,7 +7005,7 @@ static struct loaded_vmcs *nested_get_current_vmcs02(struct vcpu_vmx *vmx)
 	}
 
 	/* Create a new VMCS */
-	item = kmalloc(sizeof(struct vmcs02_list), GFP_KERNEL);
+	item = kzalloc(sizeof(struct vmcs02_list), GFP_KERNEL);
 	if (!item)
 		return NULL;
 	item->vmcs02.vmcs = alloc_vmcs();
@@ -7979,6 +8022,7 @@ static int handle_pml_full(struct kvm_vcpu *vcpu)
 	 * "blocked by NMI" bit has to be set before next VM entry.
 	 */
 	if (!(to_vmx(vcpu)->idt_vectoring_info & VECTORING_INFO_VALID_MASK) &&
+			enable_vnmi &&
 			(exit_qualification & INTR_INFO_UNBLOCK_NMI))
 		vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO,
 				GUEST_INTR_STATE_NMI);
@@ -8823,6 +8867,25 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu)
 		return 0;
 	}
 
+	if (unlikely(!enable_vnmi &&
+		     vmx->loaded_vmcs->soft_vnmi_blocked)) {
+		if (vmx_interrupt_allowed(vcpu)) {
+			vmx->loaded_vmcs->soft_vnmi_blocked = 0;
+		} else if (vmx->loaded_vmcs->vnmi_blocked_time > 1000000000LL &&
+			   vcpu->arch.nmi_pending) {
+			/*
+			 * This CPU don't support us in finding the end of an
+			 * NMI-blocked window if the guest runs with IRQs
+			 * disabled. So we pull the trigger after 1 s of
+			 * futile waiting, but inform the user about this.
+			 */
+			printk(KERN_WARNING "%s: Breaking out of NMI-blocked "
+			       "state on VCPU %d after 1 s timeout\n",
+			       __func__, vcpu->vcpu_id);
+			vmx->loaded_vmcs->soft_vnmi_blocked = 0;
+		}
+	}
+
 	if (exit_reason < kvm_vmx_max_exit_handlers
 	    && kvm_vmx_exit_handlers[exit_reason])
 		return kvm_vmx_exit_handlers[exit_reason](vcpu);
@@ -9105,33 +9168,38 @@ static void vmx_recover_nmi_blocking(struct vcpu_vmx *vmx)
 
 	idtv_info_valid = vmx->idt_vectoring_info & VECTORING_INFO_VALID_MASK;
 
-	if (vmx->loaded_vmcs->nmi_known_unmasked)
-		return;
-	/*
-	 * Can't use vmx->exit_intr_info since we're not sure what
-	 * the exit reason is.
-	 */
-	exit_intr_info = vmcs_read32(VM_EXIT_INTR_INFO);
-	unblock_nmi = (exit_intr_info & INTR_INFO_UNBLOCK_NMI) != 0;
-	vector = exit_intr_info & INTR_INFO_VECTOR_MASK;
-	/*
-	 * SDM 3: 27.7.1.2 (September 2008)
-	 * Re-set bit "block by NMI" before VM entry if vmexit caused by
-	 * a guest IRET fault.
-	 * SDM 3: 23.2.2 (September 2008)
-	 * Bit 12 is undefined in any of the following cases:
-	 *  If the VM exit sets the valid bit in the IDT-vectoring
-	 *   information field.
-	 *  If the VM exit is due to a double fault.
-	 */
-	if ((exit_intr_info & INTR_INFO_VALID_MASK) && unblock_nmi &&
-	    vector != DF_VECTOR && !idtv_info_valid)
-		vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO,
-			      GUEST_INTR_STATE_NMI);
-	else
-		vmx->loaded_vmcs->nmi_known_unmasked =
-			!(vmcs_read32(GUEST_INTERRUPTIBILITY_INFO)
-			  & GUEST_INTR_STATE_NMI);
+	if (enable_vnmi) {
+		if (vmx->loaded_vmcs->nmi_known_unmasked)
+			return;
+		/*
+		 * Can't use vmx->exit_intr_info since we're not sure what
+		 * the exit reason is.
+		 */
+		exit_intr_info = vmcs_read32(VM_EXIT_INTR_INFO);
+		unblock_nmi = (exit_intr_info & INTR_INFO_UNBLOCK_NMI) != 0;
+		vector = exit_intr_info & INTR_INFO_VECTOR_MASK;
+		/*
+		 * SDM 3: 27.7.1.2 (September 2008)
+		 * Re-set bit "block by NMI" before VM entry if vmexit caused by
+		 * a guest IRET fault.
+		 * SDM 3: 23.2.2 (September 2008)
+		 * Bit 12 is undefined in any of the following cases:
+		 *  If the VM exit sets the valid bit in the IDT-vectoring
+		 *   information field.
+		 *  If the VM exit is due to a double fault.
+		 */
+		if ((exit_intr_info & INTR_INFO_VALID_MASK) && unblock_nmi &&
+		    vector != DF_VECTOR && !idtv_info_valid)
+			vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO,
+				      GUEST_INTR_STATE_NMI);
+		else
+			vmx->loaded_vmcs->nmi_known_unmasked =
+				!(vmcs_read32(GUEST_INTERRUPTIBILITY_INFO)
+				  & GUEST_INTR_STATE_NMI);
+	} else if (unlikely(vmx->loaded_vmcs->soft_vnmi_blocked))
+		vmx->loaded_vmcs->vnmi_blocked_time +=
+			ktime_to_ns(ktime_sub(ktime_get(),
+					      vmx->loaded_vmcs->entry_time));
 }
 
 static void __vmx_complete_interrupts(struct kvm_vcpu *vcpu,
@@ -9248,6 +9316,11 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu)
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	unsigned long debugctlmsr, cr3, cr4;
 
+	/* Record the guest's net vcpu time for enforced NMI injections. */
+	if (unlikely(!enable_vnmi &&
+		     vmx->loaded_vmcs->soft_vnmi_blocked))
+		vmx->loaded_vmcs->entry_time = ktime_get();
+
 	/* Don't enter VMX if guest state is invalid, let the exit handler
 	   start emulation until we arrive back to a valid state */
 	if (vmx->emulation_required)