Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull kvm updates from Paolo Bonzini:
 "S390:

   - ultravisor communication device driver

   - fix TEID on terminating storage key ops

  RISC-V:

   - Added Sv57x4 support for G-stage page table

   - Added range based local HFENCE functions

   - Added remote HFENCE functions based on VCPU requests

   - Added ISA extension registers in ONE_REG interface

   - Updated KVM RISC-V maintainers entry to cover selftests support

  ARM:

   - Add support for the ARMv8.6 WFxT extension

   - Guard pages for the EL2 stacks

   - Trap and emulate AArch32 ID registers to hide unsupported features

   - Ability to select and save/restore the set of hypercalls exposed to
     the guest

   - Support for PSCI-initiated suspend in collaboration with userspace

   - GICv3 register-based LPI invalidation support

   - Move host PMU event merging into the vcpu data structure

   - GICv3 ITS save/restore fixes

   - The usual set of small-scale cleanups and fixes

  x86:

   - New ioctls to get/set TSC frequency for a whole VM

   - Allow userspace to opt out of hypercall patching

   - Only do MSR filtering for MSRs accessed by rdmsr/wrmsr

  AMD SEV improvements:

   - Add KVM_EXIT_SHUTDOWN metadata for SEV-ES

   - V_TSC_AUX support

  Nested virtualization improvements for AMD:

   - Support for "nested nested" optimizations (nested vVMLOAD/VMSAVE,
     nested vGIF)

   - Allow AVIC to co-exist with a nested guest running

   - Fixes for LBR virtualizations when a nested guest is running, and
     nested LBR virtualization support

   - PAUSE filtering for nested hypervisors

  Guest support:

   - Decoupling of vcpu_is_preempted from PV spinlocks"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (199 commits)
  KVM: x86: Fix the intel_pt PMI handling wrongly considered from guest
  KVM: selftests: x86: Sync the new name of the test case to .gitignore
  Documentation: kvm: reorder ARM-specific section about KVM_SYSTEM_EVENT_SUSPEND
  x86, kvm: use correct GFP flags for preemption disabled
  KVM: LAPIC: Drop pending LAPIC timer injection when canceling the timer
  x86/kvm: Alloc dummy async #PF token outside of raw spinlock
  KVM: x86: avoid calling x86 emulator without a decoded instruction
  KVM: SVM: Use kzalloc for sev ioctl interfaces to prevent kernel data leak
  x86/fpu: KVM: Set the base guest FPU uABI size to sizeof(struct kvm_xsave)
  s390/uv_uapi: depend on CONFIG_S390
  KVM: selftests: x86: Fix test failure on arch lbr capable platforms
  KVM: LAPIC: Trace LAPIC timer expiration on every vmentry
  KVM: s390: selftest: Test suppression indication on key prot exception
  KVM: s390: Don't indicate suppression on dirtying, failing memop
  selftests: drivers/s390x: Add uvdevice tests
  drivers/s390/char: Add Ultravisor io device
  MAINTAINERS: Update KVM RISC-V entry to cover selftests support
  RISC-V: KVM: Introduce ISA extension register
  RISC-V: KVM: Cleanup stale TLB entries when host CPU changes
  RISC-V: KVM: Add remote HFENCE functions based on VCPU requests
  ...

97 files changed, 5238 insertions, 2053 deletions

diff --git a/arch/arm64/include/asm/barrier.h b/arch/arm64/include/asm/barrier.h
index 62217be36217..9f3e2c3d2ca0 100644
--- a/arch/arm64/include/asm/barrier.h
+++ b/arch/arm64/include/asm/barrier.h
@@ -16,7 +16,11 @@
 
 #define sev()		asm volatile("sev" : : : "memory")
 #define wfe()		asm volatile("wfe" : : : "memory")
+#define wfet(val)	asm volatile("msr s0_3_c1_c0_0, %0"	\
+				     : : "r" (val) : "memory")
 #define wfi()		asm volatile("wfi" : : : "memory")
+#define wfit(val)	asm volatile("msr s0_3_c1_c0_1, %0"	\
+				     : : "r" (val) : "memory")
 
 #define isb()		asm volatile("isb" : : : "memory")
 #define dmb(opt)	asm volatile("dmb " #opt : : : "memory")
diff --git a/arch/arm64/include/asm/cputype.h b/arch/arm64/include/asm/cputype.h
index 92331c07c2d1..8aa0d276a636 100644
--- a/arch/arm64/include/asm/cputype.h
+++ b/arch/arm64/include/asm/cputype.h
@@ -118,6 +118,10 @@
 
 #define APPLE_CPU_PART_M1_ICESTORM	0x022
 #define APPLE_CPU_PART_M1_FIRESTORM	0x023
+#define APPLE_CPU_PART_M1_ICESTORM_PRO	0x024
+#define APPLE_CPU_PART_M1_FIRESTORM_PRO	0x025
+#define APPLE_CPU_PART_M1_ICESTORM_MAX	0x028
+#define APPLE_CPU_PART_M1_FIRESTORM_MAX	0x029
 
 #define MIDR_CORTEX_A53 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A53)
 #define MIDR_CORTEX_A57 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A57)
@@ -164,6 +168,10 @@
 #define MIDR_HISI_TSV110 MIDR_CPU_MODEL(ARM_CPU_IMP_HISI, HISI_CPU_PART_TSV110)
 #define MIDR_APPLE_M1_ICESTORM MIDR_CPU_MODEL(ARM_CPU_IMP_APPLE, APPLE_CPU_PART_M1_ICESTORM)
 #define MIDR_APPLE_M1_FIRESTORM MIDR_CPU_MODEL(ARM_CPU_IMP_APPLE, APPLE_CPU_PART_M1_FIRESTORM)
+#define MIDR_APPLE_M1_ICESTORM_PRO MIDR_CPU_MODEL(ARM_CPU_IMP_APPLE, APPLE_CPU_PART_M1_ICESTORM_PRO)
+#define MIDR_APPLE_M1_FIRESTORM_PRO MIDR_CPU_MODEL(ARM_CPU_IMP_APPLE, APPLE_CPU_PART_M1_FIRESTORM_PRO)
+#define MIDR_APPLE_M1_ICESTORM_MAX MIDR_CPU_MODEL(ARM_CPU_IMP_APPLE, APPLE_CPU_PART_M1_ICESTORM_MAX)
+#define MIDR_APPLE_M1_FIRESTORM_MAX MIDR_CPU_MODEL(ARM_CPU_IMP_APPLE, APPLE_CPU_PART_M1_FIRESTORM_MAX)
 
 /* Fujitsu Erratum 010001 affects A64FX 1.0 and 1.1, (v0r0 and v1r0) */
 #define MIDR_FUJITSU_ERRATUM_010001		MIDR_FUJITSU_A64FX
diff --git a/arch/arm64/include/asm/esr.h b/arch/arm64/include/asm/esr.h
index 8f236de7359c..15b34fbfca66 100644
--- a/arch/arm64/include/asm/esr.h
+++ b/arch/arm64/include/asm/esr.h
@@ -135,7 +135,10 @@
 #define ESR_ELx_CV		(UL(1) << 24)
 #define ESR_ELx_COND_SHIFT	(20)
 #define ESR_ELx_COND_MASK	(UL(0xF) << ESR_ELx_COND_SHIFT)
-#define ESR_ELx_WFx_ISS_TI	(UL(1) << 0)
+#define ESR_ELx_WFx_ISS_RN	(UL(0x1F) << 5)
+#define ESR_ELx_WFx_ISS_RV	(UL(1) << 2)
+#define ESR_ELx_WFx_ISS_TI	(UL(3) << 0)
+#define ESR_ELx_WFx_ISS_WFxT	(UL(2) << 0)
 #define ESR_ELx_WFx_ISS_WFI	(UL(0) << 0)
 #define ESR_ELx_WFx_ISS_WFE	(UL(1) << 0)
 #define ESR_ELx_xVC_IMM_MASK	((UL(1) << 16) - 1)
@@ -148,7 +151,8 @@
 #define DISR_EL1_ESR_MASK	(ESR_ELx_AET | ESR_ELx_EA | ESR_ELx_FSC)
 
 /* ESR value templates for specific events */
-#define ESR_ELx_WFx_MASK	(ESR_ELx_EC_MASK | ESR_ELx_WFx_ISS_TI)
+#define ESR_ELx_WFx_MASK	(ESR_ELx_EC_MASK |			\
+				 (ESR_ELx_WFx_ISS_TI & ~ESR_ELx_WFx_ISS_WFxT))
 #define ESR_ELx_WFx_WFI_VAL	((ESR_ELx_EC_WFx << ESR_ELx_EC_SHIFT) |	\
 				 ESR_ELx_WFx_ISS_WFI)
 
diff --git a/arch/arm64/include/asm/hwcap.h b/arch/arm64/include/asm/hwcap.h
index 9f0ce004fdbc..aa443d8f8cfb 100644
--- a/arch/arm64/include/asm/hwcap.h
+++ b/arch/arm64/include/asm/hwcap.h
@@ -117,6 +117,7 @@
 #define KERNEL_HWCAP_SME_B16F32		__khwcap2_feature(SME_B16F32)
 #define KERNEL_HWCAP_SME_F32F32		__khwcap2_feature(SME_F32F32)
 #define KERNEL_HWCAP_SME_FA64		__khwcap2_feature(SME_FA64)
+#define KERNEL_HWCAP_WFXT		__khwcap2_feature(WFXT)
 
 /*
  * This yields a mask that user programs can use to figure out what
diff --git a/arch/arm64/include/asm/kvm_arm.h b/arch/arm64/include/asm/kvm_arm.h
index 13ae232ec4a1..8aa8492dafc0 100644
--- a/arch/arm64/include/asm/kvm_arm.h
+++ b/arch/arm64/include/asm/kvm_arm.h
@@ -80,11 +80,12 @@
  * FMO:		Override CPSR.F and enable signaling with VF
  * SWIO:	Turn set/way invalidates into set/way clean+invalidate
  * PTW:		Take a stage2 fault if a stage1 walk steps in device memory
+ * TID3:	Trap EL1 reads of group 3 ID registers
  */
 #define HCR_GUEST_FLAGS (HCR_TSC | HCR_TSW | HCR_TWE | HCR_TWI | HCR_VM | \
 			 HCR_BSU_IS | HCR_FB | HCR_TACR | \
 			 HCR_AMO | HCR_SWIO | HCR_TIDCP | HCR_RW | HCR_TLOR | \
-			 HCR_FMO | HCR_IMO | HCR_PTW )
+			 HCR_FMO | HCR_IMO | HCR_PTW | HCR_TID3 )
 #define HCR_VIRT_EXCP_MASK (HCR_VSE | HCR_VI | HCR_VF)
 #define HCR_HOST_NVHE_FLAGS (HCR_RW | HCR_API | HCR_APK | HCR_ATA)
 #define HCR_HOST_NVHE_PROTECTED_FLAGS (HCR_HOST_NVHE_FLAGS | HCR_TSC)
diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h
index d5b0386ef765..2e277f2ed671 100644
--- a/arch/arm64/include/asm/kvm_asm.h
+++ b/arch/arm64/include/asm/kvm_asm.h
@@ -169,6 +169,7 @@ struct kvm_nvhe_init_params {
 	unsigned long tcr_el2;
 	unsigned long tpidr_el2;
 	unsigned long stack_hyp_va;
+	unsigned long stack_pa;
 	phys_addr_t pgd_pa;
 	unsigned long hcr_el2;
 	unsigned long vttbr;
diff --git a/arch/arm64/include/asm/kvm_emulate.h b/arch/arm64/include/asm/kvm_emulate.h
index 08233172e7a9..0e66edd3aff2 100644
--- a/arch/arm64/include/asm/kvm_emulate.h
+++ b/arch/arm64/include/asm/kvm_emulate.h
@@ -87,13 +87,6 @@ static inline void vcpu_reset_hcr(struct kvm_vcpu *vcpu)
 
 	if (vcpu_el1_is_32bit(vcpu))
 		vcpu->arch.hcr_el2 &= ~HCR_RW;
-	else
-		/*
-		 * TID3: trap feature register accesses that we virtualise.
-		 * For now this is conditional, since no AArch32 feature regs
-		 * are currently virtualised.
-		 */
-		vcpu->arch.hcr_el2 |= HCR_TID3;
 
 	if (cpus_have_const_cap(ARM64_MISMATCHED_CACHE_TYPE) ||
 	    vcpu_el1_is_32bit(vcpu))
diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index d5888dedf02a..47a1e25e25bb 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -46,6 +46,7 @@
 #define KVM_REQ_RECORD_STEAL	KVM_ARCH_REQ(3)
 #define KVM_REQ_RELOAD_GICv4	KVM_ARCH_REQ(4)
 #define KVM_REQ_RELOAD_PMU	KVM_ARCH_REQ(5)
+#define KVM_REQ_SUSPEND		KVM_ARCH_REQ(6)
 
 #define KVM_DIRTY_LOG_MANUAL_CAPS   (KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE | \
 				     KVM_DIRTY_LOG_INITIALLY_SET)
@@ -101,15 +102,25 @@ struct kvm_s2_mmu {
 struct kvm_arch_memory_slot {
 };
 
+/**
+ * struct kvm_smccc_features: Descriptor of the hypercall services exposed to the guests
+ *
+ * @std_bmap: Bitmap of standard secure service calls
+ * @std_hyp_bmap: Bitmap of standard hypervisor service calls
+ * @vendor_hyp_bmap: Bitmap of vendor specific hypervisor service calls
+ */
+struct kvm_smccc_features {
+	unsigned long std_bmap;
+	unsigned long std_hyp_bmap;
+	unsigned long vendor_hyp_bmap;
+};
+
 struct kvm_arch {
 	struct kvm_s2_mmu mmu;
 
 	/* VTCR_EL2 value for this VM */
 	u64    vtcr;
 
-	/* The maximum number of vCPUs depends on the used GIC model */
-	int max_vcpus;
-
 	/* Interrupt controller */
 	struct vgic_dist	vgic;
 
@@ -136,6 +147,8 @@ struct kvm_arch {
 	 */
 #define KVM_ARCH_FLAG_REG_WIDTH_CONFIGURED		3
 #define KVM_ARCH_FLAG_EL1_32BIT				4
+	/* PSCI SYSTEM_SUSPEND enabled for the guest */
+#define KVM_ARCH_FLAG_SYSTEM_SUSPEND_ENABLED		5
 
 	unsigned long flags;
 
@@ -150,6 +163,9 @@ struct kvm_arch {
 
 	u8 pfr0_csv2;
 	u8 pfr0_csv3;
+
+	/* Hypercall features firmware registers' descriptor */
+	struct kvm_smccc_features smccc_feat;
 };
 
 struct kvm_vcpu_fault_info {
@@ -254,14 +270,8 @@ struct kvm_cpu_context {
 	struct kvm_vcpu *__hyp_running_vcpu;
 };
 
-struct kvm_pmu_events {
-	u32 events_host;
-	u32 events_guest;
-};
-
 struct kvm_host_data {
 	struct kvm_cpu_context host_ctxt;
-	struct kvm_pmu_events pmu_events;
 };
 
 struct kvm_host_psci_config {
@@ -368,8 +378,8 @@ struct kvm_vcpu_arch {
 		u32	mdscr_el1;
 	} guest_debug_preserved;
 
-	/* vcpu power-off state */
-	bool power_off;
+	/* vcpu power state */
+	struct kvm_mp_state mp_state;
 
 	/* Don't run the guest (internal implementation need) */
 	bool pause;
@@ -455,6 +465,7 @@ struct kvm_vcpu_arch {
 #define KVM_ARM64_FP_FOREIGN_FPSTATE	(1 << 14)
 #define KVM_ARM64_ON_UNSUPPORTED_CPU	(1 << 15) /* Physical CPU not in supported_cpus */
 #define KVM_ARM64_HOST_SME_ENABLED	(1 << 16) /* SME enabled for EL0 */
+#define KVM_ARM64_WFIT			(1 << 17) /* WFIT instruction trapped */
 
 #define KVM_GUESTDBG_VALID_MASK (KVM_GUESTDBG_ENABLE | \
 				 KVM_GUESTDBG_USE_SW_BP | \
@@ -687,10 +698,11 @@ int kvm_handle_cp14_64(struct kvm_vcpu *vcpu);
 int kvm_handle_cp15_32(struct kvm_vcpu *vcpu);
 int kvm_handle_cp15_64(struct kvm_vcpu *vcpu);
 int kvm_handle_sys_reg(struct kvm_vcpu *vcpu);
+int kvm_handle_cp10_id(struct kvm_vcpu *vcpu);
 
 void kvm_reset_sys_regs(struct kvm_vcpu *vcpu);
 
-void kvm_sys_reg_table_init(void);
+int kvm_sys_reg_table_init(void);
 
 /* MMIO helpers */
 void kvm_mmio_write_buf(void *buf, unsigned int len, unsigned long data);
@@ -799,9 +811,6 @@ void kvm_arch_vcpu_put_debug_state_flags(struct kvm_vcpu *vcpu);
 #ifdef CONFIG_KVM
 void kvm_set_pmu_events(u32 set, struct perf_event_attr *attr);
 void kvm_clr_pmu_events(u32 clr);
-
-void kvm_vcpu_pmu_restore_guest(struct kvm_vcpu *vcpu);
-void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu);
 #else
 static inline void kvm_set_pmu_events(u32 set, struct perf_event_attr *attr) {}
 static inline void kvm_clr_pmu_events(u32 clr) {}
@@ -833,8 +842,6 @@ bool kvm_arm_vcpu_is_finalized(struct kvm_vcpu *vcpu);
 #define kvm_has_mte(kvm)					\
 	(system_supports_mte() &&				\
 	 test_bit(KVM_ARCH_FLAG_MTE_ENABLED, &(kvm)->arch.flags))
-#define kvm_vcpu_has_pmu(vcpu)					\
-	(test_bit(KVM_ARM_VCPU_PMU_V3, (vcpu)->arch.features))
 
 int kvm_trng_call(struct kvm_vcpu *vcpu);
 #ifdef CONFIG_KVM
@@ -845,4 +852,7 @@ void __init kvm_hyp_reserve(void);
 static inline void kvm_hyp_reserve(void) { }
 #endif
 
+void kvm_arm_vcpu_power_off(struct kvm_vcpu *vcpu);
+bool kvm_arm_vcpu_stopped(struct kvm_vcpu *vcpu);
+
 #endif /* __ARM64_KVM_HOST_H__ */
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index 74735a864eee..b208da3bebec 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -154,6 +154,9 @@ static __always_inline unsigned long __kern_hyp_va(unsigned long v)
 int kvm_share_hyp(void *from, void *to);
 void kvm_unshare_hyp(void *from, void *to);
 int create_hyp_mappings(void *from, void *to, enum kvm_pgtable_prot prot);
+int __create_hyp_mappings(unsigned long start, unsigned long size,
+			  unsigned long phys, enum kvm_pgtable_prot prot);
+int hyp_alloc_private_va_range(size_t size, unsigned long *haddr);
 int create_hyp_io_mappings(phys_addr_t phys_addr, size_t size,
 			   void __iomem **kaddr,
 			   void __iomem **haddr);
diff --git a/arch/arm64/include/uapi/asm/hwcap.h b/arch/arm64/include/uapi/asm/hwcap.h
index b0256cec63b5..4bb2cc8ac446 100644
--- a/arch/arm64/include/uapi/asm/hwcap.h
+++ b/arch/arm64/include/uapi/asm/hwcap.h
@@ -87,5 +87,6 @@
 #define HWCAP2_SME_B16F32	(1 << 28)
 #define HWCAP2_SME_F32F32	(1 << 29)
 #define HWCAP2_SME_FA64		(1 << 30)
+#define HWCAP2_WFXT		(1UL << 31)
 
 #endif /* _UAPI__ASM_HWCAP_H */
diff --git a/arch/arm64/include/uapi/asm/kvm.h b/arch/arm64/include/uapi/asm/kvm.h
index ab585359242d..3bb134355874 100644
--- a/arch/arm64/include/uapi/asm/kvm.h
+++ b/arch/arm64/include/uapi/asm/kvm.h
@@ -334,6 +334,40 @@ struct kvm_arm_copy_mte_tags {
 #define KVM_ARM64_SVE_VLS_WORDS	\
 	((KVM_ARM64_SVE_VQ_MAX - KVM_ARM64_SVE_VQ_MIN) / 64 + 1)
 
+/* Bitmap feature firmware registers */
+#define KVM_REG_ARM_FW_FEAT_BMAP		(0x0016 << KVM_REG_ARM_COPROC_SHIFT)
+#define KVM_REG_ARM_FW_FEAT_BMAP_REG(r)		(KVM_REG_ARM64 | KVM_REG_SIZE_U64 | \
+						KVM_REG_ARM_FW_FEAT_BMAP |	\
+						((r) & 0xffff))
+
+#define KVM_REG_ARM_STD_BMAP			KVM_REG_ARM_FW_FEAT_BMAP_REG(0)
+
+enum {
+	KVM_REG_ARM_STD_BIT_TRNG_V1_0	= 0,
+#ifdef __KERNEL__
+	KVM_REG_ARM_STD_BMAP_BIT_COUNT,
+#endif
+};
+
+#define KVM_REG_ARM_STD_HYP_BMAP		KVM_REG_ARM_FW_FEAT_BMAP_REG(1)
+
+enum {
+	KVM_REG_ARM_STD_HYP_BIT_PV_TIME	= 0,
+#ifdef __KERNEL__
+	KVM_REG_ARM_STD_HYP_BMAP_BIT_COUNT,
+#endif
+};
+
+#define KVM_REG_ARM_VENDOR_HYP_BMAP		KVM_REG_ARM_FW_FEAT_BMAP_REG(2)
+
+enum {
+	KVM_REG_ARM_VENDOR_HYP_BIT_FUNC_FEAT	= 0,
+	KVM_REG_ARM_VENDOR_HYP_BIT_PTP		= 1,
+#ifdef __KERNEL__
+	KVM_REG_ARM_VENDOR_HYP_BMAP_BIT_COUNT,
+#endif
+};
+
 /* Device Control API: ARM VGIC */
 #define KVM_DEV_ARM_VGIC_GRP_ADDR	0
 #define KVM_DEV_ARM_VGIC_GRP_DIST_REGS	1
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 4ccddf382e5b..42ea2bd856c6 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -237,6 +237,7 @@ static const struct arm64_ftr_bits ftr_id_aa64isar2[] = {
 	ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_PTR_AUTH),
 		       FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR2_GPA3_SHIFT, 4, 0),
 	ARM64_FTR_BITS(FTR_VISIBLE, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64ISAR2_RPRES_SHIFT, 4, 0),
+	ARM64_FTR_BITS(FTR_VISIBLE, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64ISAR2_WFXT_SHIFT, 4, 0),
 	ARM64_FTR_END,
 };
 
@@ -2517,6 +2518,17 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
 		.cpu_enable = fa64_kernel_enable,
 	},
 #endif /* CONFIG_ARM64_SME */
+	{
+		.desc = "WFx with timeout",
+		.capability = ARM64_HAS_WFXT,
+		.type = ARM64_CPUCAP_SYSTEM_FEATURE,
+		.sys_reg = SYS_ID_AA64ISAR2_EL1,
+		.sign = FTR_UNSIGNED,
+		.field_pos = ID_AA64ISAR2_WFXT_SHIFT,
+		.field_width = 4,
+		.matches = has_cpuid_feature,
+		.min_field_value = ID_AA64ISAR2_WFXT_SUPPORTED,
+	},
 	{},
 };
 
@@ -2650,6 +2662,7 @@ static const struct arm64_cpu_capabilities arm64_elf_hwcaps[] = {
 	HWCAP_CAP(SYS_ID_AA64MMFR0_EL1, ID_AA64MMFR0_ECV_SHIFT, 4, FTR_UNSIGNED, 1, CAP_HWCAP, KERNEL_HWCAP_ECV),
 	HWCAP_CAP(SYS_ID_AA64MMFR1_EL1, ID_AA64MMFR1_AFP_SHIFT, 4, FTR_UNSIGNED, 1, CAP_HWCAP, KERNEL_HWCAP_AFP),
 	HWCAP_CAP(SYS_ID_AA64ISAR2_EL1, ID_AA64ISAR2_RPRES_SHIFT, 4, FTR_UNSIGNED, 1, CAP_HWCAP, KERNEL_HWCAP_RPRES),
+	HWCAP_CAP(SYS_ID_AA64ISAR2_EL1, ID_AA64ISAR2_WFXT_SHIFT, 4, FTR_UNSIGNED, ID_AA64ISAR2_WFXT_SUPPORTED, CAP_HWCAP, KERNEL_HWCAP_WFXT),
 #ifdef CONFIG_ARM64_SME
 	HWCAP_CAP(SYS_ID_AA64PFR1_EL1, ID_AA64PFR1_SME_SHIFT, 4, FTR_UNSIGNED, ID_AA64PFR1_SME, CAP_HWCAP, KERNEL_HWCAP_SME),
 	HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_FA64_SHIFT, 1, FTR_UNSIGNED, ID_AA64SMFR0_FA64, CAP_HWCAP, KERNEL_HWCAP_SME_FA64),
diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c
index 8a8136a096ac..8eff0a34ffd4 100644
--- a/arch/arm64/kernel/cpuinfo.c
+++ b/arch/arm64/kernel/cpuinfo.c
@@ -106,6 +106,7 @@ static const char *const hwcap_str[] = {
 	[KERNEL_HWCAP_SME_B16F32]	= "smeb16f32",
 	[KERNEL_HWCAP_SME_F32F32]	= "smef32f32",
 	[KERNEL_HWCAP_SME_FA64]		= "smefa64",
+	[KERNEL_HWCAP_WFXT]		= "wfxt",
 };
 
 #ifdef CONFIG_COMPAT
diff --git a/arch/arm64/kvm/Makefile b/arch/arm64/kvm/Makefile
index 261644b1a6bb..aa127ae9f675 100644
--- a/arch/arm64/kvm/Makefile
+++ b/arch/arm64/kvm/Makefile
@@ -13,7 +13,7 @@ obj-$(CONFIG_KVM) += hyp/
 kvm-y += arm.o mmu.o mmio.o psci.o hypercalls.o pvtime.o \
 	 inject_fault.o va_layout.o handle_exit.o \
 	 guest.o debug.o reset.o sys_regs.o \
-	 vgic-sys-reg-v3.o fpsimd.o pmu.o pkvm.o \
+	 vgic-sys-reg-v3.o fpsimd.o pkvm.o \
 	 arch_timer.o trng.o vmid.o \
 	 vgic/vgic.o vgic/vgic-init.o \
 	 vgic/vgic-irqfd.o vgic/vgic-v2.o \
@@ -22,7 +22,7 @@ kvm-y += arm.o mmu.o mmio.o psci.o hypercalls.o pvtime.o \
 	 vgic/vgic-mmio-v3.o vgic/vgic-kvm-device.o \
 	 vgic/vgic-its.o vgic/vgic-debug.o
 
-kvm-$(CONFIG_HW_PERF_EVENTS)  += pmu-emul.o
+kvm-$(CONFIG_HW_PERF_EVENTS)  += pmu-emul.o pmu.o
 
 always-y := hyp_constants.h hyp-constants.s
 
diff --git a/arch/arm64/kvm/arch_timer.c b/arch/arm64/kvm/arch_timer.c
index 6e542e2eae32..4e39ace073af 100644
--- a/arch/arm64/kvm/arch_timer.c
+++ b/arch/arm64/kvm/arch_timer.c
@@ -208,18 +208,16 @@ static irqreturn_t kvm_arch_timer_handler(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static u64 kvm_timer_compute_delta(struct arch_timer_context *timer_ctx)
+static u64 kvm_counter_compute_delta(struct arch_timer_context *timer_ctx,
+				     u64 val)
 {
-	u64 cval, now;
-
-	cval = timer_get_cval(timer_ctx);
-	now = kvm_phys_timer_read() - timer_get_offset(timer_ctx);
+	u64 now = kvm_phys_timer_read() - timer_get_offset(timer_ctx);
 
-	if (now < cval) {
+	if (now < val) {
 		u64 ns;
 
 		ns = cyclecounter_cyc2ns(timecounter->cc,
-					 cval - now,
+					 val - now,
 					 timecounter->mask,
 					 &timecounter->frac);
 		return ns;
@@ -228,6 +226,11 @@ static u64 kvm_timer_compute_delta(struct arch_timer_context *timer_ctx)
 	return 0;
 }
 
+static u64 kvm_timer_compute_delta(struct arch_timer_context *timer_ctx)
+{
+	return kvm_counter_compute_delta(timer_ctx, timer_get_cval(timer_ctx));
+}
+
 static bool kvm_timer_irq_can_fire(struct arch_timer_context *timer_ctx)
 {
 	WARN_ON(timer_ctx && timer_ctx->loaded);
@@ -236,6 +239,20 @@ static bool kvm_timer_irq_can_fire(struct arch_timer_context *timer_ctx)
 		  (ARCH_TIMER_CTRL_IT_MASK | ARCH_TIMER_CTRL_ENABLE)) == ARCH_TIMER_CTRL_ENABLE);
 }
 
+static bool vcpu_has_wfit_active(struct kvm_vcpu *vcpu)
+{
+	return (cpus_have_final_cap(ARM64_HAS_WFXT) &&
+		(vcpu->arch.flags & KVM_ARM64_WFIT));
+}
+
+static u64 wfit_delay_ns(struct kvm_vcpu *vcpu)
+{
+	struct arch_timer_context *ctx = vcpu_vtimer(vcpu);
+	u64 val = vcpu_get_reg(vcpu, kvm_vcpu_sys_get_rt(vcpu));
+
+	return kvm_counter_compute_delta(ctx, val);
+}
+
 /*
  * Returns the earliest expiration time in ns among guest timers.
  * Note that it will return 0 if none of timers can fire.
@@ -253,6 +270,9 @@ static u64 kvm_timer_earliest_exp(struct kvm_vcpu *vcpu)
 			min_delta = min(min_delta, kvm_timer_compute_delta(ctx));
 	}
 
+	if (vcpu_has_wfit_active(vcpu))
+		min_delta = min(min_delta, wfit_delay_ns(vcpu));
+
 	/* If none of timers can fire, then return 0 */
 	if (min_delta == ULLONG_MAX)
 		return 0;
@@ -350,15 +370,9 @@ static bool kvm_timer_should_fire(struct arch_timer_context *timer_ctx)
 	return cval <= now;
 }
 
-bool kvm_timer_is_pending(struct kvm_vcpu *vcpu)
+int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
 {
-	struct timer_map map;
-
-	get_timer_map(vcpu, &map);
-
-	return kvm_timer_should_fire(map.direct_vtimer) ||
-	       kvm_timer_should_fire(map.direct_ptimer) ||
-	       kvm_timer_should_fire(map.emul_ptimer);
+	return vcpu_has_wfit_active(vcpu) && wfit_delay_ns(vcpu) == 0;
 }
 
 /*
@@ -484,7 +498,8 @@ static void kvm_timer_blocking(struct kvm_vcpu *vcpu)
 	 */
 	if (!kvm_timer_irq_can_fire(map.direct_vtimer) &&
 	    !kvm_timer_irq_can_fire(map.direct_ptimer) &&
-	    !kvm_timer_irq_can_fire(map.emul_ptimer))
+	    !kvm_timer_irq_can_fire(map.emul_ptimer) &&
+	    !vcpu_has_wfit_active(vcpu))
 		return;
 
 	/*
diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
index cedc3ba2c098..400bb0fe2745 100644
--- a/arch/arm64/kvm/arm.c
+++ b/arch/arm64/kvm/arm.c
@@ -97,6 +97,10 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
 		}
 		mutex_unlock(&kvm->lock);
 		break;
+	case KVM_CAP_ARM_SYSTEM_SUSPEND:
+		r = 0;
+		set_bit(KVM_ARCH_FLAG_SYSTEM_SUSPEND_ENABLED, &kvm->arch.flags);
+		break;
 	default:
 		r = -EINVAL;
 		break;
@@ -153,9 +157,10 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 	kvm_vgic_early_init(kvm);
 
 	/* The maximum number of VCPUs is limited by the host's GIC model */
-	kvm->arch.max_vcpus = kvm_arm_default_max_vcpus();
+	kvm->max_vcpus = kvm_arm_default_max_vcpus();
 
 	set_default_spectre(kvm);
+	kvm_arm_init_hypercalls(kvm);
 
 	return ret;
 out_free_stage2_pgd:
@@ -210,6 +215,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_SET_GUEST_DEBUG:
 	case KVM_CAP_VCPU_ATTRIBUTES:
 	case KVM_CAP_PTP_KVM:
+	case KVM_CAP_ARM_SYSTEM_SUSPEND:
 		r = 1;
 		break;
 	case KVM_CAP_SET_GUEST_DEBUG2:
@@ -230,7 +236,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_MAX_VCPUS:
 	case KVM_CAP_MAX_VCPU_ID:
 		if (kvm)
-			r = kvm->arch.max_vcpus;
+			r = kvm->max_vcpus;
 		else
 			r = kvm_arm_default_max_vcpus();
 		break;
@@ -306,7 +312,7 @@ int kvm_arch_vcpu_precreate(struct kvm *kvm, unsigned int id)
 	if (irqchip_in_kernel(kvm) && vgic_initialized(kvm))
 		return -EBUSY;
 
-	if (id >= kvm->arch.max_vcpus)
+	if (id >= kvm->max_vcpus)
 		return -EINVAL;
 
 	return 0;
@@ -356,11 +362,6 @@ void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu)
 	kvm_arm_vcpu_destroy(vcpu);
 }
 
-int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
-{
-	return kvm_timer_is_pending(vcpu);
-}
-
 void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
 {
 
@@ -432,20 +433,34 @@ void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
 	vcpu->cpu = -1;
 }
 
-static void vcpu_power_off(struct kvm_vcpu *vcpu)
+void kvm_arm_vcpu_power_off(struct kvm_vcpu *vcpu)
 {
-	vcpu->arch.power_off = true;
+	vcpu->arch.mp_state.mp_state = KVM_MP_STATE_STOPPED;
 	kvm_make_request(KVM_REQ_SLEEP, vcpu);
 	kvm_vcpu_kick(vcpu);
 }
 
+bool kvm_arm_vcpu_stopped(struct kvm_vcpu *vcpu)
+{
+	return vcpu->arch.mp_state.mp_state == KVM_MP_STATE_STOPPED;
+}
+
+static void kvm_arm_vcpu_suspend(struct kvm_vcpu *vcpu)
+{
+	vcpu->arch.mp_state.mp_state = KVM_MP_STATE_SUSPENDED;
+	kvm_make_request(KVM_REQ_SUSPEND, vcpu);
+	kvm_vcpu_kick(vcpu);
+}
+
+static bool kvm_arm_vcpu_suspended(struct kvm_vcpu *vcpu)
+{
+	return vcpu->arch.mp_state.mp_state == KVM_MP_STATE_SUSPENDED;
+}
+
 int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
 				    struct kvm_mp_state *mp_state)
 {
-	if (vcpu->arch.power_off)
-		mp_state->mp_state = KVM_MP_STATE_STOPPED;
-	else
-		mp_state->mp_state = KVM_MP_STATE_RUNNABLE;
+	*mp_state = vcpu->arch.mp_state;
 
 	return 0;
 }
@@ -457,10 +472,13 @@ int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
 
 	switch (mp_state->mp_state) {
 	case KVM_MP_STATE_RUNNABLE:
-		vcpu->arch.power_off = false;
+		vcpu->arch.mp_state = *mp_state;
 		break;
 	case KVM_MP_STATE_STOPPED:
-		vcpu_power_off(vcpu);
+		kvm_arm_vcpu_power_off(vcpu);
+		break;
+	case KVM_MP_STATE_SUSPENDED:
+		kvm_arm_vcpu_suspend(vcpu);
 		break;
 	default:
 		ret = -EINVAL;
@@ -480,7 +498,7 @@ int kvm_arch_vcpu_runnable(struct kvm_vcpu *v)
 {
 	bool irq_lines = *vcpu_hcr(v) & (HCR_VI | HCR_VF);
 	return ((irq_lines || kvm_vgic_vcpu_pending_irq(v))
-		&& !v->arch.power_off && !v->arch.pause);
+		&& !kvm_arm_vcpu_stopped(v) && !v->arch.pause);
 }
 
 bool kvm_arch_vcpu_in_kernel(struct kvm_vcpu *vcpu)
@@ -592,15 +610,15 @@ void kvm_arm_resume_guest(struct kvm *kvm)
 	}
 }
 
-static void vcpu_req_sleep(struct kvm_vcpu *vcpu)
+static void kvm_vcpu_sleep(struct kvm_vcpu *vcpu)
 {
 	struct rcuwait *wait = kvm_arch_vcpu_get_wait(vcpu);
 
 	rcuwait_wait_event(wait,
-			   (!vcpu->arch.power_off) &&(!vcpu->arch.pause),
+			   (!kvm_arm_vcpu_stopped(vcpu)) && (!vcpu->arch.pause),
 			   TASK_INTERRUPTIBLE);
 
-	if (vcpu->arch.power_off || vcpu->arch.pause) {
+	if (kvm_arm_vcpu_stopped(vcpu) || vcpu->arch.pause) {
 		/* Awaken to handle a signal, request we sleep again later. */
 		kvm_make_request(KVM_REQ_SLEEP, vcpu);
 	}
@@ -639,6 +657,7 @@ void kvm_vcpu_wfi(struct kvm_vcpu *vcpu)
 	preempt_enable();
 
 	kvm_vcpu_halt(vcpu);
+	vcpu->arch.flags &= ~KVM_ARM64_WFIT;
 	kvm_clear_request(KVM_REQ_UNHALT, vcpu);
 
 	preempt_disable();
@@ -646,11 +665,53 @@ void kvm_vcpu_wfi(struct kvm_vcpu *vcpu)
 	preempt_enable();
 }
 
-static void check_vcpu_requests(struct kvm_vcpu *vcpu)
+static int kvm_vcpu_suspend(struct kvm_vcpu *vcpu)
+{
+	if (!kvm_arm_vcpu_suspended(vcpu))
+		return 1;
+
+	kvm_vcpu_wfi(vcpu);
+
+	/*
+	 * The suspend state is sticky; we do not leave it until userspace
+	 * explicitly marks the vCPU as runnable. Request that we suspend again
+	 * later.
+	 */
+	kvm_make_request(KVM_REQ_SUSPEND, vcpu);
+
+	/*
+	 * Check to make sure the vCPU is actually runnable. If so, exit to
+	 * userspace informing it of the wakeup condition.
+	 */
+	if (kvm_arch_vcpu_runnable(vcpu)) {
+		memset(&vcpu->run->system_event, 0, sizeof(vcpu->run->system_event));
+		vcpu->run->system_event.type = KVM_SYSTEM_EVENT_WAKEUP;
+		vcpu->run->exit_reason = KVM_EXIT_SYSTEM_EVENT;
+		return 0;
+	}
+
+	/*
+	 * Otherwise, we were unblocked to process a different event, such as a
+	 * pending signal. Return 1 and allow kvm_arch_vcpu_ioctl_run() to
+	 * process the event.
+	 */
+	return 1;
+}
+
+/**
+ * check_vcpu_requests - check and handle pending vCPU requests
+ * @vcpu:	the VCPU pointer
+ *
+ * Return: 1 if we should enter the guest
+ *	   0 if we should exit to userspace
+ *	   < 0 if we should exit to userspace, where the return value indicates
+ *	   an error
+ */
+static int check_vcpu_requests(struct kvm_vcpu *vcpu)
 {
 	if (kvm_request_pending(vcpu)) {
 		if (kvm_check_request(KVM_REQ_SLEEP, vcpu))
-			vcpu_req_sleep(vcpu);
+			kvm_vcpu_sleep(vcpu);
 
 		if (kvm_check_request(KVM_REQ_VCPU_RESET, vcpu))
 			kvm_reset_vcpu(vcpu);
@@ -675,7 +736,12 @@ static void check_vcpu_requests(struct kvm_vcpu *vcpu)
 		if (kvm_check_request(KVM_REQ_RELOAD_PMU, vcpu))
 			kvm_pmu_handle_pmcr(vcpu,
 					    __vcpu_sys_reg(vcpu, PMCR_EL0));
+
+		if (kvm_check_request(KVM_REQ_SUSPEND, vcpu))
+			return kvm_vcpu_suspend(vcpu);
 	}
+
+	return 1;
 }
 
 static bool vcpu_mode_is_bad_32bit(struct kvm_vcpu *vcpu)
@@ -792,7 +858,8 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 		if (!ret)
 			ret = 1;
 
-		check_vcpu_requests(vcpu);
+		if (ret > 0)
+			ret = check_vcpu_requests(vcpu);
 
 		/*
 		 * Preparing the interrupts to be injected also
@@ -816,6 +883,8 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 
 		kvm_vgic_flush_hwstate(vcpu);
 
+		kvm_pmu_update_vcpu_events(vcpu);
+
 		/*
 		 * Ensure we set mode to IN_GUEST_MODE after we disable
 		 * interrupts and before the final VCPU requests check.
@@ -1125,9 +1194,9 @@ static int kvm_arch_vcpu_ioctl_vcpu_init(struct kvm_vcpu *vcpu,
 	 * Handle the "start in power-off" case.
 	 */
 	if (test_bit(KVM_ARM_VCPU_POWER_OFF, vcpu->arch.features))
-		vcpu_power_off(vcpu);
+		kvm_arm_vcpu_power_off(vcpu);
 	else
-		vcpu->arch.power_off = false;
+		vcpu->arch.mp_state.mp_state = KVM_MP_STATE_RUNNABLE;
 
 	return 0;
 }
@@ -1485,7 +1554,6 @@ static void cpu_prepare_hyp_mode(int cpu)
 	tcr |= (idmap_t0sz & GENMASK(TCR_TxSZ_WIDTH - 1, 0)) << TCR_T0SZ_OFFSET;
 	params->tcr_el2 = tcr;
 
-	params->stack_hyp_va = kern_hyp_va(per_cpu(kvm_arm_hyp_stack_page, cpu) + PAGE_SIZE);
 	params->pgd_pa = kvm_mmu_get_httbr();
 	if (is_protected_kvm_enabled())
 		params->hcr_el2 = HCR_HOST_NVHE_PROTECTED_FLAGS;
@@ -1763,8 +1831,6 @@ static int init_subsystems(void)
 
 	kvm_register_perf_callbacks(NULL);
 
-	kvm_sys_reg_table_init();
-
 out:
 	if (err || !is_protected_kvm_enabled())
 		on_each_cpu(_kvm_arch_hardware_disable, NULL, 1);
@@ -1935,14 +2001,46 @@ static int init_hyp_mode(void)
 	 * Map the Hyp stack pages
 	 */
 	for_each_possible_cpu(cpu) {
+		struct kvm_nvhe_init_params *params = per_cpu_ptr_nvhe_sym(kvm_init_params, cpu);
 		char *stack_page = (char *)per_cpu(kvm_arm_hyp_stack_page, cpu);
-		err = create_hyp_mappings(stack_page, stack_page + PAGE_SIZE,
-					  PAGE_HYP);
+		unsigned long hyp_addr;
+
+		/*
+		 * Allocate a contiguous HYP private VA range for the stack
+		 * and guard page. The allocation is also aligned based on
+		 * the order of its size.
+		 */
+		err = hyp_alloc_private_va_range(PAGE_SIZE * 2, &hyp_addr);
+		if (err) {
+			kvm_err("Cannot allocate hyp stack guard page\n");
+			goto out_err;
+		}
 
+		/*
+		 * Since the stack grows downwards, map the stack to the page
+		 * at the higher address and leave the lower guard page
+		 * unbacked.
+		 *
+		 * Any valid stack address now has the PAGE_SHIFT bit as 1
+		 * and addresses corresponding to the guard page have the
+		 * PAGE_SHIFT bit as 0 - this is used for overflow detection.
+		 */
+		err = __create_hyp_mappings(hyp_addr + PAGE_SIZE, PAGE_SIZE,
+					    __pa(stack_page), PAGE_HYP);
 		if (err) {
 			kvm_err("Cannot map hyp stack\n");
 			goto out_err;
 		}
+
+		/*
+		 * Save the stack PA in nvhe_init_params. This will be needed
+		 * to recreate the stack mapping in protected nVHE mode.
+		 * __hyp_pa() won't do the right thing there, since the stack
+		 * has been mapped in the flexible private VA space.
+		 */
+		params->stack_pa = __pa(stack_page);
+
+		params->stack_hyp_va = hyp_addr + (2 * PAGE_SIZE);
 	}
 
 	for_each_possible_cpu(cpu) {
@@ -2091,6 +2189,12 @@ int kvm_arch_init(void *opaque)
 		return -ENODEV;
 	}
 
+	err = kvm_sys_reg_table_init();
+	if (err) {
+		kvm_info("Error initializing system register tables");
+		return err;
+	}
+
 	in_hyp_mode = is_kernel_in_hyp_mode();
 
 	if (cpus_have_final_cap(ARM64_WORKAROUND_DEVICE_LOAD_ACQUIRE) ||
diff --git a/arch/arm64/kvm/guest.c b/arch/arm64/kvm/guest.c
index 7e15b03fbdf8..8c607199cad1 100644
--- a/arch/arm64/kvm/guest.c
+++ b/arch/arm64/kvm/guest.c
@@ -18,7 +18,7 @@
 #include <linux/string.h>
 #include <linux/vmalloc.h>
 #include <linux/fs.h>
-#include <kvm/arm_psci.h>
+#include <kvm/arm_hypercalls.h>
 #include <asm/cputype.h>
 #include <linux/uaccess.h>
 #include <asm/fpsimd.h>
@@ -756,7 +756,9 @@ int kvm_arm_get_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
 
 	switch (reg->id & KVM_REG_ARM_COPROC_MASK) {
 	case KVM_REG_ARM_CORE:	return get_core_reg(vcpu, reg);
-	case KVM_REG_ARM_FW:	return kvm_arm_get_fw_reg(vcpu, reg);
+	case KVM_REG_ARM_FW:
+	case KVM_REG_ARM_FW_FEAT_BMAP:
+		return kvm_arm_get_fw_reg(vcpu, reg);
 	case KVM_REG_ARM64_SVE:	return get_sve_reg(vcpu, reg);
 	}
 
@@ -774,7 +776,9 @@ int kvm_arm_set_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
 
 	switch (reg->id & KVM_REG_ARM_COPROC_MASK) {
 	case KVM_REG_ARM_CORE:	return set_core_reg(vcpu, reg);
-	case KVM_REG_ARM_FW:	return kvm_arm_set_fw_reg(vcpu, reg);
+	case KVM_REG_ARM_FW:
+	case KVM_REG_ARM_FW_FEAT_BMAP:
+		return kvm_arm_set_fw_reg(vcpu, reg);
 	case KVM_REG_ARM64_SVE:	return set_sve_reg(vcpu, reg);
 	}
 
diff --git a/arch/arm64/kvm/handle_exit.c b/arch/arm64/kvm/handle_exit.c
index 0b829292dc54..f66c0142b335 100644
--- a/arch/arm64/kvm/handle_exit.c
+++ b/arch/arm64/kvm/handle_exit.c
@@ -80,24 +80,51 @@ static int handle_no_fpsimd(struct kvm_vcpu *vcpu)
  *
  * @vcpu:	the vcpu pointer
  *
- * WFE: Yield the CPU and come back to this vcpu when the scheduler
+ * WFE[T]: Yield the CPU and come back to this vcpu when the scheduler
  * decides to.
  * WFI: Simply call kvm_vcpu_halt(), which will halt execution of
  * world-switches and schedule other host processes until there is an
  * incoming IRQ or FIQ to the VM.
+ * WFIT: Same as WFI, with a timed wakeup implemented as a background timer
+ *
+ * WF{I,E}T can immediately return if the deadline has already expired.
  */
 static int kvm_handle_wfx(struct kvm_vcpu *vcpu)
 {
-	if (kvm_vcpu_get_esr(vcpu) & ESR_ELx_WFx_ISS_WFE) {
+	u64 esr = kvm_vcpu_get_esr(vcpu);
+
+	if (esr & ESR_ELx_WFx_ISS_WFE) {
 		trace_kvm_wfx_arm64(*vcpu_pc(vcpu), true);
 		vcpu->stat.wfe_exit_stat++;
-		kvm_vcpu_on_spin(vcpu, vcpu_mode_priv(vcpu));
 	} else {
 		trace_kvm_wfx_arm64(*vcpu_pc(vcpu), false);
 		vcpu->stat.wfi_exit_stat++;
-		kvm_vcpu_wfi(vcpu);
 	}
 
+	if (esr & ESR_ELx_WFx_ISS_WFxT) {
+		if (esr & ESR_ELx_WFx_ISS_RV) {
+			u64 val, now;
+
+			now = kvm_arm_timer_get_reg(vcpu, KVM_REG_ARM_TIMER_CNT);
+			val = vcpu_get_reg(vcpu, kvm_vcpu_sys_get_rt(vcpu));
+
+			if (now >= val)
+				goto out;
+		} else {
+			/* Treat WFxT as WFx if RN is invalid */
+			esr &= ~ESR_ELx_WFx_ISS_WFxT;
+		}
+	}
+
+	if (esr & ESR_ELx_WFx_ISS_WFE) {
+		kvm_vcpu_on_spin(vcpu, vcpu_mode_priv(vcpu));
+	} else {
+		if (esr & ESR_ELx_WFx_ISS_WFxT)
+			vcpu->arch.flags |= KVM_ARM64_WFIT;
+
+		kvm_vcpu_wfi(vcpu);
+	}
+out:
 	kvm_incr_pc(vcpu);
 
 	return 1;
@@ -169,6 +196,7 @@ static exit_handle_fn arm_exit_handlers[] = {
 	[ESR_ELx_EC_CP15_64]	= kvm_handle_cp15_64,
 	[ESR_ELx_EC_CP14_MR]	= kvm_handle_cp14_32,
 	[ESR_ELx_EC_CP14_LS]	= kvm_handle_cp14_load_store,
+	[ESR_ELx_EC_CP10_ID]	= kvm_handle_cp10_id,
 	[ESR_ELx_EC_CP14_64]	= kvm_handle_cp14_64,
 	[ESR_ELx_EC_HVC32]	= handle_hvc,
 	[ESR_ELx_EC_SMC32]	= handle_smc,
@@ -297,13 +325,8 @@ void __noreturn __cold nvhe_hyp_panic_handler(u64 esr, u64 spsr,
 	u64 elr_in_kimg = __phys_to_kimg(elr_phys);
 	u64 hyp_offset = elr_in_kimg - kaslr_offset() - elr_virt;
 	u64 mode = spsr & PSR_MODE_MASK;
+	u64 panic_addr = elr_virt + hyp_offset;
 
-	/*
-	 * The nVHE hyp symbols are not included by kallsyms to avoid issues
-	 * with aliasing. That means that the symbols cannot be printed with the
-	 * "%pS" format specifier, so fall back to the vmlinux address if
-	 * there's no better option.
-	 */
 	if (mode != PSR_MODE_EL2t && mode != PSR_MODE_EL2h) {
 		kvm_err("Invalid host exception to nVHE hyp!\n");
 	} else if (ESR_ELx_EC(esr) == ESR_ELx_EC_BRK64 &&
@@ -323,9 +346,11 @@ void __noreturn __cold nvhe_hyp_panic_handler(u64 esr, u64 spsr,
 		if (file)
 			kvm_err("nVHE hyp BUG at: %s:%u!\n", file, line);
 		else
-			kvm_err("nVHE hyp BUG at: %016llx!\n", elr_virt + hyp_offset);
+			kvm_err("nVHE hyp BUG at: [<%016llx>] %pB!\n", panic_addr,
+					(void *)panic_addr);
 	} else {
-		kvm_err("nVHE hyp panic at: %016llx!\n", elr_virt + hyp_offset);
+		kvm_err("nVHE hyp panic at: [<%016llx>] %pB!\n", panic_addr,
+				(void *)panic_addr);
 	}
 
 	/*
diff --git a/arch/arm64/kvm/hyp/include/nvhe/mm.h b/arch/arm64/kvm/hyp/include/nvhe/mm.h
index 2d08510c6cc1..42d8eb9bfe72 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/mm.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/mm.h
@@ -19,8 +19,10 @@ int hyp_back_vmemmap(phys_addr_t phys, unsigned long size, phys_addr_t back);
 int pkvm_cpu_set_vector(enum arm64_hyp_spectre_vector slot);
 int pkvm_create_mappings(void *from, void *to, enum kvm_pgtable_prot prot);
 int pkvm_create_mappings_locked(void *from, void *to, enum kvm_pgtable_prot prot);
-unsigned long __pkvm_create_private_mapping(phys_addr_t phys, size_t size,
-					    enum kvm_pgtable_prot prot);
+int __pkvm_create_private_mapping(phys_addr_t phys, size_t size,
+				  enum kvm_pgtable_prot prot,
+				  unsigned long *haddr);
+int pkvm_alloc_private_va_range(size_t size, unsigned long *haddr);
 
 static inline void hyp_vmemmap_range(phys_addr_t phys, unsigned long size,
 				     unsigned long *start, unsigned long *end)
diff --git a/arch/arm64/kvm/hyp/nvhe/host.S b/arch/arm64/kvm/hyp/nvhe/host.S
index 727c979b2b69..ea6a397b64a6 100644
--- a/arch/arm64/kvm/hyp/nvhe/host.S
+++ b/arch/arm64/kvm/hyp/nvhe/host.S
@@ -80,7 +80,7 @@ SYM_FUNC_START(__hyp_do_panic)
 	mov	lr, #(PSR_F_BIT | PSR_I_BIT | PSR_A_BIT | PSR_D_BIT |\
 		      PSR_MODE_EL1h)
 	msr	spsr_el2, lr
-	ldr	lr, =nvhe_hyp_panic_handler
+	adr_l	lr, nvhe_hyp_panic_handler
 	hyp_kimg_va lr, x6
 	msr	elr_el2, lr
 
@@ -125,13 +125,11 @@ alternative_else_nop_endif
 	add	sp, sp, #16
 	/*
 	 * Compute the idmap address of __kvm_handle_stub_hvc and
-	 * jump there. Since we use kimage_voffset, do not use the
-	 * HYP VA for __kvm_handle_stub_hvc, but the kernel VA instead
-	 * (by loading it from the constant pool).
+	 * jump there.
 	 *
 	 * Preserve x0-x4, which may contain stub parameters.
 	 */
-	ldr	x5, =__kvm_handle_stub_hvc
+	adr_l	x5, __kvm_handle_stub_hvc
 	hyp_pa	x5, x6
 	br	x5
 SYM_FUNC_END(__host_hvc)
@@ -153,6 +151,18 @@ SYM_FUNC_END(__host_hvc)
 
 .macro invalid_host_el2_vect
 	.align 7
+
+	/*
+	 * Test whether the SP has overflowed, without corrupting a GPR.
+	 * nVHE hypervisor stacks are aligned so that the PAGE_SHIFT bit
+	 * of SP should always be 1.
+	 */
+	add	sp, sp, x0			// sp' = sp + x0
+	sub	x0, sp, x0			// x0' = sp' - x0 = (sp + x0) - x0 = sp
+	tbz	x0, #PAGE_SHIFT, .L__hyp_sp_overflow\@
+	sub	x0, sp, x0			// x0'' = sp' - x0' = (sp + x0) - sp = x0
+	sub	sp, sp, x0			// sp'' = sp' - x0 = (sp + x0) - x0 = sp
+
 	/* If a guest is loaded, panic out of it. */
 	stp	x0, x1, [sp, #-16]!
 	get_loaded_vcpu x0, x1
@@ -165,6 +175,18 @@ SYM_FUNC_END(__host_hvc)
 	 * been partially clobbered by __host_enter.
 	 */
 	b	hyp_panic
+
+.L__hyp_sp_overflow\@:
+	/*
+	 * Reset SP to the top of the stack, to allow handling the hyp_panic.
+	 * This corrupts the stack but is ok, since we won't be attempting
+	 * any unwinding here.
+	 */
+	ldr_this_cpu	x0, kvm_init_params + NVHE_INIT_STACK_HYP_VA, x1
+	mov	sp, x0
+
+	b	hyp_panic_bad_stack
+	ASM_BUG()
 .endm
 
 .macro invalid_host_el1_vect
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index 5e2197db0d32..3cea4b6ac23e 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -160,7 +160,23 @@ static void handle___pkvm_create_private_mapping(struct kvm_cpu_context *host_ct
 	DECLARE_REG(size_t, size, host_ctxt, 2);
 	DECLARE_REG(enum kvm_pgtable_prot, prot, host_ctxt, 3);
 
-	cpu_reg(host_ctxt, 1) = __pkvm_create_private_mapping(phys, size, prot);
+	/*
+	 * __pkvm_create_private_mapping() populates a pointer with the
+	 * hypervisor start address of the allocation.
+	 *
+	 * However, handle___pkvm_create_private_mapping() hypercall crosses the
+	 * EL1/EL2 boundary so the pointer would not be valid in this context.
+	 *
+	 * Instead pass the allocation address as the return value (or return
+	 * ERR_PTR() on failure).
+	 */
+	unsigned long haddr;
+	int err = __pkvm_create_private_mapping(phys, size, prot, &haddr);
+
+	if (err)
+		haddr = (unsigned long)ERR_PTR(err);
+
+	cpu_reg(host_ctxt, 1) = haddr;
 }
 
 static void handle___pkvm_prot_finalize(struct kvm_cpu_context *host_ctxt)
diff --git a/arch/arm64/kvm/hyp/nvhe/mm.c b/arch/arm64/kvm/hyp/nvhe/mm.c
index cdbe8e246418..96193cb31a39 100644
--- a/arch/arm64/kvm/hyp/nvhe/mm.c
+++ b/arch/arm64/kvm/hyp/nvhe/mm.c
@@ -37,36 +37,60 @@ static int __pkvm_create_mappings(unsigned long start, unsigned long size,
 	return err;
 }
 
-unsigned long __pkvm_create_private_mapping(phys_addr_t phys, size_t size,
-					    enum kvm_pgtable_prot prot)
+/**
+ * pkvm_alloc_private_va_range - Allocates a private VA range.
+ * @size:	The size of the VA range to reserve.
+ * @haddr:	The hypervisor virtual start address of the allocation.
+ *
+ * The private virtual address (VA) range is allocated above __io_map_base
+ * and aligned based on the order of @size.
+ *
+ * Return: 0 on success or negative error code on failure.
+ */
+int pkvm_alloc_private_va_range(size_t size, unsigned long *haddr)
 {
-	unsigned long addr;
-	int err;
+	unsigned long base, addr;
+	int ret = 0;
 
 	hyp_spin_lock(&pkvm_pgd_lock);
 
-	size = PAGE_ALIGN(size + offset_in_page(phys));
-	addr = __io_map_base;
-	__io_map_base += size;
+	/* Align the allocation based on the order of its size */
+	addr = ALIGN(__io_map_base, PAGE_SIZE << get_order(size));
 
-	/* Are we overflowing on the vmemmap ? */
-	if (__io_map_base > __hyp_vmemmap) {
-		__io_map_base -= size;
-		addr = (unsigned long)ERR_PTR(-ENOMEM);
-		goto out;
-	}
+	/* The allocated size is always a multiple of PAGE_SIZE */
+	base = addr + PAGE_ALIGN(size);
 
-	err = kvm_pgtable_hyp_map(&pkvm_pgtable, addr, size, phys, prot);
-	if (err) {
-		addr = (unsigned long)ERR_PTR(err);
-		goto out;
+	/* Are we overflowing on the vmemmap ? */
+	if (!addr || base > __hyp_vmemmap)
+		ret = -ENOMEM;
+	else {
+		__io_map_base = base;
+		*haddr = addr;
 	}
 
-	addr = addr + offset_in_page(phys);
-out:
 	hyp_spin_unlock(&pkvm_pgd_lock);
 
-	return addr;
+	return ret;
+}
+
+int __pkvm_create_private_mapping(phys_addr_t phys, size_t size,
+				  enum kvm_pgtable_prot prot,
+				  unsigned long *haddr)
+{
+	unsigned long addr;
+	int err;
+
+	size = PAGE_ALIGN(size + offset_in_page(phys));
+	err = pkvm_alloc_private_va_range(size, &addr);
+	if (err)
+		return err;
+
+	err = __pkvm_create_mappings(addr, size, phys, prot);
+	if (err)
+		return err;
+
+	*haddr = addr + offset_in_page(phys);
+	return err;
 }
 
 int pkvm_create_mappings_locked(void *from, void *to, enum kvm_pgtable_prot prot)
@@ -146,7 +170,8 @@ int pkvm_cpu_set_vector(enum arm64_hyp_spectre_vector slot)
 int hyp_map_vectors(void)
 {
 	phys_addr_t phys;
-	void *bp_base;
+	unsigned long bp_base;
+	int ret;
 
 	if (!kvm_system_needs_idmapped_vectors()) {
 		__hyp_bp_vect_base = __bp_harden_hyp_vecs;
@@ -154,13 +179,12 @@ int hyp_map_vectors(void)
 	}
 
 	phys = __hyp_pa(__bp_harden_hyp_vecs);
-	bp_base = (void *)__pkvm_create_private_mapping(phys,
-							__BP_HARDEN_HYP_VECS_SZ,
-							PAGE_HYP_EXEC);
-	if (IS_ERR_OR_NULL(bp_base))
-		return PTR_ERR(bp_base);
+	ret = __pkvm_create_private_mapping(phys, __BP_HARDEN_HYP_VECS_SZ,
+					    PAGE_HYP_EXEC, &bp_base);
+	if (ret)
+		return ret;
 
-	__hyp_bp_vect_base = bp_base;
+	__hyp_bp_vect_base = (void *)bp_base;
 
 	return 0;
 }
diff --git a/arch/arm64/kvm/hyp/nvhe/setup.c b/arch/arm64/kvm/hyp/nvhe/setup.c
index 27af337f9fea..e8d4ea2fcfa0 100644
--- a/arch/arm64/kvm/hyp/nvhe/setup.c
+++ b/arch/arm64/kvm/hyp/nvhe/setup.c
@@ -99,17 +99,42 @@ static int recreate_hyp_mappings(phys_addr_t phys, unsigned long size,
 		return ret;
 
 	for (i = 0; i < hyp_nr_cpus; i++) {
+		struct kvm_nvhe_init_params *params = per_cpu_ptr(&kvm_init_params, i);
+		unsigned long hyp_addr;
+
 		start = (void *)kern_hyp_va(per_cpu_base[i]);
 		end = start + PAGE_ALIGN(hyp_percpu_size);
 		ret = pkvm_create_mappings(start, end, PAGE_HYP);
 		if (ret)
 			return ret;
 
-		end = (void *)per_cpu_ptr(&kvm_init_params, i)->stack_hyp_va;
-		start = end - PAGE_SIZE;
-		ret = pkvm_create_mappings(start, end, PAGE_HYP);
+		/*
+		 * Allocate a contiguous HYP private VA range for the stack
+		 * and guard page. The allocation is also aligned based on
+		 * the order of its size.
+		 */
+		ret = pkvm_alloc_private_va_range(PAGE_SIZE * 2, &hyp_addr);
+		if (ret)
+			return ret;
+
+		/*
+		 * Since the stack grows downwards, map the stack to the page
+		 * at the higher address and leave the lower guard page
+		 * unbacked.
+		 *
+		 * Any valid stack address now has the PAGE_SHIFT bit as 1
+		 * and addresses corresponding to the guard page have the
+		 * PAGE_SHIFT bit as 0 - this is used for overflow detection.
+		 */
+		hyp_spin_lock(&pkvm_pgd_lock);
+		ret = kvm_pgtable_hyp_map(&pkvm_pgtable, hyp_addr + PAGE_SIZE,
+					PAGE_SIZE, params->stack_pa, PAGE_HYP);
+		hyp_spin_unlock(&pkvm_pgd_lock);
 		if (ret)
 			return ret;
+
+		/* Update stack_hyp_va to end of the stack's private VA range */
+		params->stack_hyp_va = hyp_addr + (2 * PAGE_SIZE);
 	}
 
 	/*
diff --git a/arch/arm64/kvm/hyp/nvhe/switch.c b/arch/arm64/kvm/hyp/nvhe/switch.c
index caace61ea459..6db801db8f27 100644
--- a/arch/arm64/kvm/hyp/nvhe/switch.c
+++ b/arch/arm64/kvm/hyp/nvhe/switch.c
@@ -150,16 +150,13 @@ static void __hyp_vgic_restore_state(struct kvm_vcpu *vcpu)
 	}
 }
 
-/**
+/*
  * Disable host events, enable guest events
  */
-static bool __pmu_switch_to_guest(struct kvm_cpu_context *host_ctxt)
+#ifdef CONFIG_HW_PERF_EVENTS
+static bool __pmu_switch_to_guest(struct kvm_vcpu *vcpu)
 {
-	struct kvm_host_data *host;
-	struct kvm_pmu_events *pmu;
-
-	host = container_of(host_ctxt, struct kvm_host_data, host_ctxt);
-	pmu = &host->pmu_events;
+	struct kvm_pmu_events *pmu = &vcpu->arch.pmu.events;
 
 	if (pmu->events_host)
 		write_sysreg(pmu->events_host, pmcntenclr_el0);
@@ -170,16 +167,12 @@ static bool __pmu_switch_to_guest(struct kvm_cpu_context *host_ctxt)
 	return (pmu->events_host || pmu->events_guest);
 }
 
-/**
+/*
  * Disable guest events, enable host events
  */
-static void __pmu_switch_to_host(struct kvm_cpu_context *host_ctxt)
+static void __pmu_switch_to_host(struct kvm_vcpu *vcpu)
 {
-	struct kvm_host_data *host;
-	struct kvm_pmu_events *pmu;
-
-	host = container_of(host_ctxt, struct kvm_host_data, host_ctxt);
-	pmu = &host->pmu_events;
+	struct kvm_pmu_events *pmu = &vcpu->arch.pmu.events;
 
 	if (pmu->events_guest)
 		write_sysreg(pmu->events_guest, pmcntenclr_el0);
@@ -187,8 +180,12 @@ static void __pmu_switch_to_host(struct kvm_cpu_context *host_ctxt)
 	if (pmu->events_host)
 		write_sysreg(pmu->events_host, pmcntenset_el0);
 }
+#else
+#define __pmu_switch_to_guest(v)	({ false; })
+#define __pmu_switch_to_host(v)		do {} while (0)
+#endif
 
-/**
+/*
  * Handler for protected VM MSR, MRS or System instruction execution in AArch64.
  *
  * Returns true if the hypervisor has handled the exit, and control should go
@@ -205,23 +202,6 @@ static bool kvm_handle_pvm_sys64(struct kvm_vcpu *vcpu, u64 *exit_code)
 		kvm_handle_pvm_sysreg(vcpu, exit_code));
 }
 
-/**
- * Handler for protected floating-point and Advanced SIMD accesses.
- *
- * Returns true if the hypervisor has handled the exit, and control should go
- * back to the guest, or false if it hasn't.
- */
-static bool kvm_handle_pvm_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code)
-{
-	/* Linux guests assume support for floating-point and Advanced SIMD. */
-	BUILD_BUG_ON(!FIELD_GET(ARM64_FEATURE_MASK(ID_AA64PFR0_FP),
-				PVM_ID_AA64PFR0_ALLOW));
-	BUILD_BUG_ON(!FIELD_GET(ARM64_FEATURE_MASK(ID_AA64PFR0_ASIMD),
-				PVM_ID_AA64PFR0_ALLOW));
-
-	return kvm_hyp_handle_fpsimd(vcpu, exit_code);
-}
-
 static const exit_handler_fn hyp_exit_handlers[] = {
 	[0 ... ESR_ELx_EC_MAX]		= NULL,
 	[ESR_ELx_EC_CP15_32]		= kvm_hyp_handle_cp15_32,
@@ -237,7 +217,7 @@ static const exit_handler_fn pvm_exit_handlers[] = {
 	[0 ... ESR_ELx_EC_MAX]		= NULL,
 	[ESR_ELx_EC_SYS64]		= kvm_handle_pvm_sys64,
 	[ESR_ELx_EC_SVE]		= kvm_handle_pvm_restricted,
-	[ESR_ELx_EC_FP_ASIMD]		= kvm_handle_pvm_fpsimd,
+	[ESR_ELx_EC_FP_ASIMD]		= kvm_hyp_handle_fpsimd,
 	[ESR_ELx_EC_IABT_LOW]		= kvm_hyp_handle_iabt_low,
 	[ESR_ELx_EC_DABT_LOW]		= kvm_hyp_handle_dabt_low,
 	[ESR_ELx_EC_PAC]		= kvm_hyp_handle_ptrauth,
@@ -304,7 +284,7 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	host_ctxt->__hyp_running_vcpu = vcpu;
 	guest_ctxt = &vcpu->arch.ctxt;
 
-	pmu_switch_needed = __pmu_switch_to_guest(host_ctxt);
+	pmu_switch_needed = __pmu_switch_to_guest(vcpu);
 
 	__sysreg_save_state_nvhe(host_ctxt);
 	/*
@@ -366,7 +346,7 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	__debug_restore_host_buffers_nvhe(vcpu);
 
 	if (pmu_switch_needed)
-		__pmu_switch_to_host(host_ctxt);
+		__pmu_switch_to_host(vcpu);
 
 	/* Returning to host will clear PSR.I, remask PMR if needed */
 	if (system_uses_irq_prio_masking())
@@ -377,7 +357,7 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	return exit_code;
 }
 
-void __noreturn hyp_panic(void)
+asmlinkage void __noreturn hyp_panic(void)
 {
 	u64 spsr = read_sysreg_el2(SYS_SPSR);
 	u64 elr = read_sysreg_el2(SYS_ELR);
@@ -399,6 +379,11 @@ void __noreturn hyp_panic(void)
 	unreachable();
 }
 
+asmlinkage void __noreturn hyp_panic_bad_stack(void)
+{
+	hyp_panic();
+}
+
 asmlinkage void kvm_unexpected_el2_exception(void)
 {
 	return __kvm_unexpected_el2_exception();
diff --git a/arch/arm64/kvm/hyp/nvhe/sys_regs.c b/arch/arm64/kvm/hyp/nvhe/sys_regs.c
index 619f94fc95fa..b6d86e423319 100644
--- a/arch/arm64/kvm/hyp/nvhe/sys_regs.c
+++ b/arch/arm64/kvm/hyp/nvhe/sys_regs.c
@@ -90,9 +90,6 @@ static u64 get_pvm_id_aa64pfr0(const struct kvm_vcpu *vcpu)
 	u64 set_mask = 0;
 	u64 allow_mask = PVM_ID_AA64PFR0_ALLOW;
 
-	if (!vcpu_has_sve(vcpu))
-		allow_mask &= ~ARM64_FEATURE_MASK(ID_AA64PFR0_SVE);
-
 	set_mask |= get_restricted_features_unsigned(id_aa64pfr0_el1_sys_val,
 		PVM_ID_AA64PFR0_RESTRICT_UNSIGNED);
 
diff --git a/arch/arm64/kvm/hypercalls.c b/arch/arm64/kvm/hypercalls.c
index 202b8c455724..c9f401fa01a9 100644
--- a/arch/arm64/kvm/hypercalls.c
+++ b/arch/arm64/kvm/hypercalls.c
@@ -9,6 +9,13 @@
 #include <kvm/arm_hypercalls.h>
 #include <kvm/arm_psci.h>
 
+#define KVM_ARM_SMCCC_STD_FEATURES				\
+	GENMASK(KVM_REG_ARM_STD_BMAP_BIT_COUNT - 1, 0)
+#define KVM_ARM_SMCCC_STD_HYP_FEATURES				\
+	GENMASK(KVM_REG_ARM_STD_HYP_BMAP_BIT_COUNT - 1, 0)
+#define KVM_ARM_SMCCC_VENDOR_HYP_FEATURES			\
+	GENMASK(KVM_REG_ARM_VENDOR_HYP_BMAP_BIT_COUNT - 1, 0)
+
 static void kvm_ptp_get_time(struct kvm_vcpu *vcpu, u64 *val)
 {
 	struct system_time_snapshot systime_snapshot;
@@ -58,13 +65,73 @@ static void kvm_ptp_get_time(struct kvm_vcpu *vcpu, u64 *val)
 	val[3] = lower_32_bits(cycles);
 }
 
+static bool kvm_hvc_call_default_allowed(u32 func_id)
+{
+	switch (func_id) {
+	/*
+	 * List of function-ids that are not gated with the bitmapped
+	 * feature firmware registers, and are to be allowed for
+	 * servicing the call by default.
+	 */
+	case ARM_SMCCC_VERSION_FUNC_ID:
+	case ARM_SMCCC_ARCH_FEATURES_FUNC_ID:
+		return true;
+	default:
+		/* PSCI 0.2 and up is in the 0:0x1f range */
+		if (ARM_SMCCC_OWNER_NUM(func_id) == ARM_SMCCC_OWNER_STANDARD &&
+		    ARM_SMCCC_FUNC_NUM(func_id) <= 0x1f)
+			return true;
+
+		/*
+		 * KVM's PSCI 0.1 doesn't comply with SMCCC, and has
+		 * its own function-id base and range
+		 */
+		if (func_id >= KVM_PSCI_FN(0) && func_id <= KVM_PSCI_FN(3))
+			return true;
+
+		return false;
+	}
+}
+
+static bool kvm_hvc_call_allowed(struct kvm_vcpu *vcpu, u32 func_id)
+{
+	struct kvm_smccc_features *smccc_feat = &vcpu->kvm->arch.smccc_feat;
+
+	switch (func_id) {
+	case ARM_SMCCC_TRNG_VERSION:
+	case ARM_SMCCC_TRNG_FEATURES:
+	case ARM_SMCCC_TRNG_GET_UUID:
+	case ARM_SMCCC_TRNG_RND32:
+	case ARM_SMCCC_TRNG_RND64:
+		return test_bit(KVM_REG_ARM_STD_BIT_TRNG_V1_0,
+				&smccc_feat->std_bmap);
+	case ARM_SMCCC_HV_PV_TIME_FEATURES:
+	case ARM_SMCCC_HV_PV_TIME_ST:
+		return test_bit(KVM_REG_ARM_STD_HYP_BIT_PV_TIME,
+				&smccc_feat->std_hyp_bmap);
+	case ARM_SMCCC_VENDOR_HYP_KVM_FEATURES_FUNC_ID:
+	case ARM_SMCCC_VENDOR_HYP_CALL_UID_FUNC_ID:
+		return test_bit(KVM_REG_ARM_VENDOR_HYP_BIT_FUNC_FEAT,
+				&smccc_feat->vendor_hyp_bmap);
+	case ARM_SMCCC_VENDOR_HYP_KVM_PTP_FUNC_ID:
+		return test_bit(KVM_REG_ARM_VENDOR_HYP_BIT_PTP,
+				&smccc_feat->vendor_hyp_bmap);
+	default:
+		return kvm_hvc_call_default_allowed(func_id);
+	}
+}
+
 int kvm_hvc_call_handler(struct kvm_vcpu *vcpu)
 {
+	struct kvm_smccc_features *smccc_feat = &vcpu->kvm->arch.smccc_feat;
 	u32 func_id = smccc_get_function(vcpu);
 	u64 val[4] = {SMCCC_RET_NOT_SUPPORTED};
 	u32 feature;
 	gpa_t gpa;
 
+	if (!kvm_hvc_call_allowed(vcpu, func_id))
+		goto out;
+
 	switch (func_id) {
 	case ARM_SMCCC_VERSION_FUNC_ID:
 		val[0] = ARM_SMCCC_VERSION_1_1;
@@ -120,7 +187,9 @@ int kvm_hvc_call_handler(struct kvm_vcpu *vcpu)
 			}
 			break;
 		case ARM_SMCCC_HV_PV_TIME_FEATURES:
-			val[0] = SMCCC_RET_SUCCESS;
+			if (test_bit(KVM_REG_ARM_STD_HYP_BIT_PV_TIME,
+				     &smccc_feat->std_hyp_bmap))
+				val[0] = SMCCC_RET_SUCCESS;
 			break;
 		}
 		break;
@@ -139,8 +208,7 @@ int kvm_hvc_call_handler(struct kvm_vcpu *vcpu)
 		val[3] = ARM_SMCCC_VENDOR_HYP_UID_KVM_REG_3;
 		break;
 	case ARM_SMCCC_VENDOR_HYP_KVM_FEATURES_FUNC_ID:
-		val[0] = BIT(ARM_SMCCC_KVM_FUNC_FEATURES);
-		val[0] |= BIT(ARM_SMCCC_KVM_FUNC_PTP);
+		val[0] = smccc_feat->vendor_hyp_bmap;
 		break;
 	case ARM_SMCCC_VENDOR_HYP_KVM_PTP_FUNC_ID:
 		kvm_ptp_get_time(vcpu, val);
@@ -155,6 +223,259 @@ int kvm_hvc_call_handler(struct kvm_vcpu *vcpu)
 		return kvm_psci_call(vcpu);
 	}
 
+out:
 	smccc_set_retval(vcpu, val[0], val[1], val[2], val[3]);
 	return 1;
 }
+
+static const u64 kvm_arm_fw_reg_ids[] = {
+	KVM_REG_ARM_PSCI_VERSION,
+	KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1,
+	KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2,
+	KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3,
+	KVM_REG_ARM_STD_BMAP,
+	KVM_REG_ARM_STD_HYP_BMAP,
+	KVM_REG_ARM_VENDOR_HYP_BMAP,
+};
+
+void kvm_arm_init_hypercalls(struct kvm *kvm)
+{
+	struct kvm_smccc_features *smccc_feat = &kvm->arch.smccc_feat;
+
+	smccc_feat->std_bmap = KVM_ARM_SMCCC_STD_FEATURES;
+	smccc_feat->std_hyp_bmap = KVM_ARM_SMCCC_STD_HYP_FEATURES;
+	smccc_feat->vendor_hyp_bmap = KVM_ARM_SMCCC_VENDOR_HYP_FEATURES;
+}
+
+int kvm_arm_get_fw_num_regs(struct kvm_vcpu *vcpu)
+{
+	return ARRAY_SIZE(kvm_arm_fw_reg_ids);
+}
+
+int kvm_arm_copy_fw_reg_indices(struct kvm_vcpu *vcpu, u64 __user *uindices)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(kvm_arm_fw_reg_ids); i++) {
+		if (put_user(kvm_arm_fw_reg_ids[i], uindices++))
+			return -EFAULT;
+	}
+
+	return 0;
+}
+
+#define KVM_REG_FEATURE_LEVEL_MASK	GENMASK(3, 0)
+
+/*
+ * Convert the workaround level into an easy-to-compare number, where higher
+ * values mean better protection.
+ */
+static int get_kernel_wa_level(u64 regid)
+{
+	switch (regid) {
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
+		switch (arm64_get_spectre_v2_state()) {
+		case SPECTRE_VULNERABLE:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_NOT_AVAIL;
+		case SPECTRE_MITIGATED:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_AVAIL;
+		case SPECTRE_UNAFFECTED:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_NOT_REQUIRED;
+		}
+		return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_NOT_AVAIL;
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2:
+		switch (arm64_get_spectre_v4_state()) {
+		case SPECTRE_MITIGATED:
+			/*
+			 * As for the hypercall discovery, we pretend we
+			 * don't have any FW mitigation if SSBS is there at
+			 * all times.
+			 */
+			if (cpus_have_final_cap(ARM64_SSBS))
+				return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
+			fallthrough;
+		case SPECTRE_UNAFFECTED:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_REQUIRED;
+		case SPECTRE_VULNERABLE:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
+		}
+		break;
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3:
+		switch (arm64_get_spectre_bhb_state()) {
+		case SPECTRE_VULNERABLE:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_NOT_AVAIL;
+		case SPECTRE_MITIGATED:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_AVAIL;
+		case SPECTRE_UNAFFECTED:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_NOT_REQUIRED;
+		}
+		return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_NOT_AVAIL;
+	}
+
+	return -EINVAL;
+}
+
+int kvm_arm_get_fw_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
+{
+	struct kvm_smccc_features *smccc_feat = &vcpu->kvm->arch.smccc_feat;
+	void __user *uaddr = (void __user *)(long)reg->addr;
+	u64 val;
+
+	switch (reg->id) {
+	case KVM_REG_ARM_PSCI_VERSION:
+		val = kvm_psci_version(vcpu);
+		break;
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2:
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3:
+		val = get_kernel_wa_level(reg->id) & KVM_REG_FEATURE_LEVEL_MASK;
+		break;
+	case KVM_REG_ARM_STD_BMAP:
+		val = READ_ONCE(smccc_feat->std_bmap);
+		break;
+	case KVM_REG_ARM_STD_HYP_BMAP:
+		val = READ_ONCE(smccc_feat->std_hyp_bmap);
+		break;
+	case KVM_REG_ARM_VENDOR_HYP_BMAP:
+		val = READ_ONCE(smccc_feat->vendor_hyp_bmap);
+		break;
+	default:
+		return -ENOENT;
+	}
+
+	if (copy_to_user(uaddr, &val, KVM_REG_SIZE(reg->id)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static int kvm_arm_set_fw_reg_bmap(struct kvm_vcpu *vcpu, u64 reg_id, u64 val)
+{
+	int ret = 0;
+	struct kvm *kvm = vcpu->kvm;
+	struct kvm_smccc_features *smccc_feat = &kvm->arch.smccc_feat;
+	unsigned long *fw_reg_bmap, fw_reg_features;
+
+	switch (reg_id) {
+	case KVM_REG_ARM_STD_BMAP:
+		fw_reg_bmap = &smccc_feat->std_bmap;
+		fw_reg_features = KVM_ARM_SMCCC_STD_FEATURES;
+		break;
+	case KVM_REG_ARM_STD_HYP_BMAP:
+		fw_reg_bmap = &smccc_feat->std_hyp_bmap;
+		fw_reg_features = KVM_ARM_SMCCC_STD_HYP_FEATURES;
+		break;
+	case KVM_REG_ARM_VENDOR_HYP_BMAP:
+		fw_reg_bmap = &smccc_feat->vendor_hyp_bmap;
+		fw_reg_features = KVM_ARM_SMCCC_VENDOR_HYP_FEATURES;
+		break;
+	default:
+		return -ENOENT;
+	}
+
+	/* Check for unsupported bit */
+	if (val & ~fw_reg_features)
+		return -EINVAL;
+
+	mutex_lock(&kvm->lock);
+
+	if (test_bit(KVM_ARCH_FLAG_HAS_RAN_ONCE, &kvm->arch.flags) &&
+	    val != *fw_reg_bmap) {
+		ret = -EBUSY;
+		goto out;
+	}
+
+	WRITE_ONCE(*fw_reg_bmap, val);
+out:
+	mutex_unlock(&kvm->lock);
+	return ret;
+}
+
+int kvm_arm_set_fw_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
+{
+	void __user *uaddr = (void __user *)(long)reg->addr;
+	u64 val;
+	int wa_level;
+
+	if (copy_from_user(&val, uaddr, KVM_REG_SIZE(reg->id)))
+		return -EFAULT;
+
+	switch (reg->id) {
+	case KVM_REG_ARM_PSCI_VERSION:
+	{
+		bool wants_02;
+
+		wants_02 = test_bit(KVM_ARM_VCPU_PSCI_0_2, vcpu->arch.features);
+
+		switch (val) {
+		case KVM_ARM_PSCI_0_1:
+			if (wants_02)
+				return -EINVAL;
+			vcpu->kvm->arch.psci_version = val;
+			return 0;
+		case KVM_ARM_PSCI_0_2:
+		case KVM_ARM_PSCI_1_0:
+		case KVM_ARM_PSCI_1_1:
+			if (!wants_02)
+				return -EINVAL;
+			vcpu->kvm->arch.psci_version = val;
+			return 0;
+		}
+		break;
+	}
+
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3:
+		if (val & ~KVM_REG_FEATURE_LEVEL_MASK)
+			return -EINVAL;
+
+		if (get_kernel_wa_level(reg->id) < val)
+			return -EINVAL;
+
+		return 0;
+
+	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2:
+		if (val & ~(KVM_REG_FEATURE_LEVEL_MASK |
+			    KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_ENABLED))
+			return -EINVAL;
+
+		/* The enabled bit must not be set unless the level is AVAIL. */
+		if ((val & KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_ENABLED) &&
+		    (val & KVM_REG_FEATURE_LEVEL_MASK) != KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL)
+			return -EINVAL;
+
+		/*
+		 * Map all the possible incoming states to the only two we
+		 * really want to deal with.
+		 */
+		switch (val & KVM_REG_FEATURE_LEVEL_MASK) {
+		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL:
+		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_UNKNOWN:
+			wa_level = KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
+			break;
+		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL:
+		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_REQUIRED:
+			wa_level = KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_REQUIRED;
+			break;
+		default:
+			return -EINVAL;
+		}
+
+		/*
+		 * We can deal with NOT_AVAIL on NOT_REQUIRED, but not the
+		 * other way around.
+		 */
+		if (get_kernel_wa_level(reg->id) < wa_level)
+			return -EINVAL;
+
+		return 0;
+	case KVM_REG_ARM_STD_BMAP:
+	case KVM_REG_ARM_STD_HYP_BMAP:
+	case KVM_REG_ARM_VENDOR_HYP_BMAP:
+		return kvm_arm_set_fw_reg_bmap(vcpu, reg->id, val);
+	default:
+		return -ENOENT;
+	}
+
+	return -EINVAL;
+}
diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index 5400fc020164..f5651a05b6a8 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -258,8 +258,8 @@ static bool kvm_host_owns_hyp_mappings(void)
 	return true;
 }
 
-static int __create_hyp_mappings(unsigned long start, unsigned long size,
-				 unsigned long phys, enum kvm_pgtable_prot prot)
+int __create_hyp_mappings(unsigned long start, unsigned long size,
+			  unsigned long phys, enum kvm_pgtable_prot prot)
 {
 	int err;
 
@@ -457,23 +457,22 @@ int create_hyp_mappings(void *from, void *to, enum kvm_pgtable_prot prot)
 	return 0;
 }
 
-static int __create_hyp_private_mapping(phys_addr_t phys_addr, size_t size,
-					unsigned long *haddr,
-					enum kvm_pgtable_prot prot)
+
+/**
+ * hyp_alloc_private_va_range - Allocates a private VA range.
+ * @size:	The size of the VA range to reserve.
+ * @haddr:	The hypervisor virtual start address of the allocation.
+ *
+ * The private virtual address (VA) range is allocated below io_map_base
+ * and aligned based on the order of @size.
+ *
+ * Return: 0 on success or negative error code on failure.
+ */
+int hyp_alloc_private_va_range(size_t size, unsigned long *haddr)
 {
 	unsigned long base;
 	int ret = 0;
 
-	if (!kvm_host_owns_hyp_mappings()) {
-		base = kvm_call_hyp_nvhe(__pkvm_create_private_mapping,
-					 phys_addr, size, prot);
-		if (IS_ERR_OR_NULL((void *)base))
-			return PTR_ERR((void *)base);
-		*haddr = base;
-
-		return 0;
-	}
-
 	mutex_lock(&kvm_hyp_pgd_mutex);
 
 	/*
@@ -484,8 +483,10 @@ static int __create_hyp_private_mapping(phys_addr_t phys_addr, size_t size,
 	 *
 	 * The allocated size is always a multiple of PAGE_SIZE.
 	 */
-	size = PAGE_ALIGN(size + offset_in_page(phys_addr));
-	base = io_map_base - size;
+	base = io_map_base - PAGE_ALIGN(size);
+
+	/* Align the allocation based on the order of its size */
+	base = ALIGN_DOWN(base, PAGE_SIZE << get_order(size));
 
 	/*
 	 * Verify that BIT(VA_BITS - 1) hasn't been flipped by
@@ -495,19 +496,40 @@ static int __create_hyp_private_mapping(phys_addr_t phys_addr, size_t size,
 	if ((base ^ io_map_base) & BIT(VA_BITS - 1))
 		ret = -ENOMEM;
 	else
-		io_map_base = base;
+		*haddr = io_map_base = base;
 
 	mutex_unlock(&kvm_hyp_pgd_mutex);
 
+	return ret;
+}
+
+static int __create_hyp_private_mapping(phys_addr_t phys_addr, size_t size,
+					unsigned long *haddr,
+					enum kvm_pgtable_prot prot)
+{
+	unsigned long addr;
+	int ret = 0;
+
+	if (!kvm_host_owns_hyp_mappings()) {
+		addr = kvm_call_hyp_nvhe(__pkvm_create_private_mapping,
+					 phys_addr, size, prot);
+		if (IS_ERR_VALUE(addr))
+			return addr;
+		*haddr = addr;
+
+		return 0;
+	}
+
+	size = PAGE_ALIGN(size + offset_in_page(phys_addr));
+	ret = hyp_alloc_private_va_range(size, &addr);
 	if (ret)
-		goto out;
+		return ret;
 
-	ret = __create_hyp_mappings(base, size, phys_addr, prot);
+	ret = __create_hyp_mappings(addr, size, phys_addr, prot);
 	if (ret)
-		goto out;
+		return ret;
 
-	*haddr = base + offset_in_page(phys_addr);
-out:
+	*haddr = addr + offset_in_page(phys_addr);
 	return ret;
 }
 
diff --git a/arch/arm64/kvm/pmu-emul.c b/arch/arm64/kvm/pmu-emul.c
index 3dc990ac4f44..11c43bed5f97 100644
--- a/arch/arm64/kvm/pmu-emul.c
+++ b/arch/arm64/kvm/pmu-emul.c
@@ -774,8 +774,7 @@ void kvm_host_pmu_init(struct arm_pmu *pmu)
 {
 	struct arm_pmu_entry *entry;
 
-	if (pmu->pmuver == 0 || pmu->pmuver == ID_AA64DFR0_PMUVER_IMP_DEF ||
-	    is_protected_kvm_enabled())
+	if (pmu->pmuver == 0 || pmu->pmuver == ID_AA64DFR0_PMUVER_IMP_DEF)
 		return;
 
 	mutex_lock(&arm_pmus_lock);
diff --git a/arch/arm64/kvm/pmu.c b/arch/arm64/kvm/pmu.c
index 03a6c1f4a09a..7887133d15f0 100644
--- a/arch/arm64/kvm/pmu.c
+++ b/arch/arm64/kvm/pmu.c
@@ -5,7 +5,8 @@
  */
 #include <linux/kvm_host.h>
 #include <linux/perf_event.h>
-#include <asm/kvm_hyp.h>
+
+static DEFINE_PER_CPU(struct kvm_pmu_events, kvm_pmu_events);
 
 /*
  * Given the perf event attributes and system type, determine
@@ -25,21 +26,26 @@ static bool kvm_pmu_switch_needed(struct perf_event_attr *attr)
 	return (attr->exclude_host != attr->exclude_guest);
 }
 
+struct kvm_pmu_events *kvm_get_pmu_events(void)
+{
+	return this_cpu_ptr(&kvm_pmu_events);
+}
+
 /*
  * Add events to track that we may want to switch at guest entry/exit
  * time.
  */
 void kvm_set_pmu_events(u32 set, struct perf_event_attr *attr)
 {
-	struct kvm_host_data *ctx = this_cpu_ptr_hyp_sym(kvm_host_data);
+	struct kvm_pmu_events *pmu = kvm_get_pmu_events();
 
-	if (!kvm_arm_support_pmu_v3() || !ctx || !kvm_pmu_switch_needed(attr))
+	if (!kvm_arm_support_pmu_v3() || !pmu || !kvm_pmu_switch_needed(attr))
 		return;
 
 	if (!attr->exclude_host)
-		ctx->pmu_events.events_host |= set;
+		pmu->events_host |= set;
 	if (!attr->exclude_guest)
-		ctx->pmu_events.events_guest |= set;
+		pmu->events_guest |= set;
 }
 
 /*
@@ -47,13 +53,13 @@ void kvm_set_pmu_events(u32 set, struct perf_event_attr *attr)
  */
 void kvm_clr_pmu_events(u32 clr)
 {
-	struct kvm_host_data *ctx = this_cpu_ptr_hyp_sym(kvm_host_data);
+	struct kvm_pmu_events *pmu = kvm_get_pmu_events();
 
-	if (!kvm_arm_support_pmu_v3() || !ctx)
+	if (!kvm_arm_support_pmu_v3() || !pmu)
 		return;
 
-	ctx->pmu_events.events_host &= ~clr;
-	ctx->pmu_events.events_guest &= ~clr;
+	pmu->events_host &= ~clr;
+	pmu->events_guest &= ~clr;
 }
 
 #define PMEVTYPER_READ_CASE(idx)				\
@@ -169,16 +175,16 @@ static void kvm_vcpu_pmu_disable_el0(unsigned long events)
  */
 void kvm_vcpu_pmu_restore_guest(struct kvm_vcpu *vcpu)
 {
-	struct kvm_host_data *host;
+	struct kvm_pmu_events *pmu;
 	u32 events_guest, events_host;
 
 	if (!kvm_arm_support_pmu_v3() || !has_vhe())
 		return;
 
 	preempt_disable();
-	host = this_cpu_ptr_hyp_sym(kvm_host_data);
-	events_guest = host->pmu_events.events_guest;
-	events_host = host->pmu_events.events_host;
+	pmu = kvm_get_pmu_events();
+	events_guest = pmu->events_guest;
+	events_host = pmu->events_host;
 
 	kvm_vcpu_pmu_enable_el0(events_guest);
 	kvm_vcpu_pmu_disable_el0(events_host);
@@ -190,15 +196,15 @@ void kvm_vcpu_pmu_restore_guest(struct kvm_vcpu *vcpu)
  */
 void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu)
 {
-	struct kvm_host_data *host;
+	struct kvm_pmu_events *pmu;
 	u32 events_guest, events_host;
 
 	if (!kvm_arm_support_pmu_v3() || !has_vhe())
 		return;
 
-	host = this_cpu_ptr_hyp_sym(kvm_host_data);
-	events_guest = host->pmu_events.events_guest;
-	events_host = host->pmu_events.events_host;
+	pmu = kvm_get_pmu_events();
+	events_guest = pmu->events_guest;
+	events_host = pmu->events_host;
 
 	kvm_vcpu_pmu_enable_el0(events_host);
 	kvm_vcpu_pmu_disable_el0(events_guest);
diff --git a/arch/arm64/kvm/psci.c b/arch/arm64/kvm/psci.c
index 708d80e8e60d..7fbc4c1b9df0 100644
--- a/arch/arm64/kvm/psci.c
+++ b/arch/arm64/kvm/psci.c
@@ -51,13 +51,6 @@ static unsigned long kvm_psci_vcpu_suspend(struct kvm_vcpu *vcpu)
 	return PSCI_RET_SUCCESS;
 }
 
-static void kvm_psci_vcpu_off(struct kvm_vcpu *vcpu)
-{
-	vcpu->arch.power_off = true;
-	kvm_make_request(KVM_REQ_SLEEP, vcpu);
-	kvm_vcpu_kick(vcpu);
-}
-
 static inline bool kvm_psci_valid_affinity(struct kvm_vcpu *vcpu,
 					   unsigned long affinity)
 {
@@ -83,7 +76,7 @@ static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
 	 */
 	if (!vcpu)
 		return PSCI_RET_INVALID_PARAMS;
-	if (!vcpu->arch.power_off) {
+	if (!kvm_arm_vcpu_stopped(vcpu)) {
 		if (kvm_psci_version(source_vcpu) != KVM_ARM_PSCI_0_1)
 			return PSCI_RET_ALREADY_ON;
 		else
@@ -107,12 +100,12 @@ static unsigned long kvm_psci_vcpu_on(struct kvm_vcpu *source_vcpu)
 	kvm_make_request(KVM_REQ_VCPU_RESET, vcpu);
 
 	/*
-	 * Make sure the reset request is observed if the change to
-	 * power_off is observed.
+	 * Make sure the reset request is observed if the RUNNABLE mp_state is
+	 * observed.
 	 */
 	smp_wmb();
 
-	vcpu->arch.power_off = false;
+	vcpu->arch.mp_state.mp_state = KVM_MP_STATE_RUNNABLE;
 	kvm_vcpu_wake_up(vcpu);
 
 	return PSCI_RET_SUCCESS;
@@ -150,7 +143,7 @@ static unsigned long kvm_psci_vcpu_affinity_info(struct kvm_vcpu *vcpu)
 		mpidr = kvm_vcpu_get_mpidr_aff(tmp);
 		if ((mpidr & target_affinity_mask) == target_affinity) {
 			matching_cpus++;
-			if (!tmp->arch.power_off)
+			if (!kvm_arm_vcpu_stopped(tmp))
 				return PSCI_0_2_AFFINITY_LEVEL_ON;
 		}
 	}
@@ -176,7 +169,7 @@ static void kvm_prepare_system_event(struct kvm_vcpu *vcpu, u32 type, u64 flags)
 	 * re-initialized.
 	 */
 	kvm_for_each_vcpu(i, tmp, vcpu->kvm)
-		tmp->arch.power_off = true;
+		tmp->arch.mp_state.mp_state = KVM_MP_STATE_STOPPED;
 	kvm_make_all_cpus_request(vcpu->kvm, KVM_REQ_SLEEP);
 
 	memset(&vcpu->run->system_event, 0, sizeof(vcpu->run->system_event));
@@ -202,6 +195,15 @@ static void kvm_psci_system_reset2(struct kvm_vcpu *vcpu)
 				 KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2);
 }
 
+static void kvm_psci_system_suspend(struct kvm_vcpu *vcpu)
+{
+	struct kvm_run *run = vcpu->run;
+
+	memset(&run->system_event, 0, sizeof(vcpu->run->system_event));
+	run->system_event.type = KVM_SYSTEM_EVENT_SUSPEND;
+	run->exit_reason = KVM_EXIT_SYSTEM_EVENT;
+}
+
 static void kvm_psci_narrow_to_32bit(struct kvm_vcpu *vcpu)
 {
 	int i;
@@ -245,7 +247,7 @@ static int kvm_psci_0_2_call(struct kvm_vcpu *vcpu)
 		val = kvm_psci_vcpu_suspend(vcpu);
 		break;
 	case PSCI_0_2_FN_CPU_OFF:
-		kvm_psci_vcpu_off(vcpu);
+		kvm_arm_vcpu_power_off(vcpu);
 		val = PSCI_RET_SUCCESS;
 		break;
 	case PSCI_0_2_FN_CPU_ON:
@@ -305,9 +307,10 @@ static int kvm_psci_0_2_call(struct kvm_vcpu *vcpu)
 
 static int kvm_psci_1_x_call(struct kvm_vcpu *vcpu, u32 minor)
 {
+	unsigned long val = PSCI_RET_NOT_SUPPORTED;
 	u32 psci_fn = smccc_get_function(vcpu);
+	struct kvm *kvm = vcpu->kvm;
 	u32 arg;
-	unsigned long val;
 	int ret = 1;
 
 	switch(psci_fn) {
@@ -320,6 +323,8 @@ static int kvm_psci_1_x_call(struct kvm_vcpu *vcpu, u32 minor)
 		if (val)
 			break;
 
+		val = PSCI_RET_NOT_SUPPORTED;
+
 		switch(arg) {
 		case PSCI_0_2_FN_PSCI_VERSION:
 		case PSCI_0_2_FN_CPU_SUSPEND:
@@ -336,18 +341,32 @@ static int kvm_psci_1_x_call(struct kvm_vcpu *vcpu, u32 minor)
 		case ARM_SMCCC_VERSION_FUNC_ID:
 			val = 0;
 			break;
+		case PSCI_1_0_FN_SYSTEM_SUSPEND:
+		case PSCI_1_0_FN64_SYSTEM_SUSPEND:
+			if (test_bit(KVM_ARCH_FLAG_SYSTEM_SUSPEND_ENABLED, &kvm->arch.flags))
+				val = 0;
+			break;
 		case PSCI_1_1_FN_SYSTEM_RESET2:
 		case PSCI_1_1_FN64_SYSTEM_RESET2:
-			if (minor >= 1) {
+			if (minor >= 1)
 				val = 0;
-				break;
-			}
-			fallthrough;
-		default:
-			val = PSCI_RET_NOT_SUPPORTED;
 			break;
 		}
 		break;
+	case PSCI_1_0_FN_SYSTEM_SUSPEND:
+		kvm_psci_narrow_to_32bit(vcpu);
+		fallthrough;
+	case PSCI_1_0_FN64_SYSTEM_SUSPEND:
+		/*
+		 * Return directly to userspace without changing the vCPU's
+		 * registers. Userspace depends on reading the SMCCC parameters
+		 * to implement SYSTEM_SUSPEND.
+		 */
+		if (test_bit(KVM_ARCH_FLAG_SYSTEM_SUSPEND_ENABLED, &kvm->arch.flags)) {
+			kvm_psci_system_suspend(vcpu);
+			return 0;
+		}
+		break;
 	case PSCI_1_1_FN_SYSTEM_RESET2:
 		kvm_psci_narrow_to_32bit(vcpu);
 		fallthrough;
@@ -365,7 +384,7 @@ static int kvm_psci_1_x_call(struct kvm_vcpu *vcpu, u32 minor)
 			val = PSCI_RET_INVALID_PARAMS;
 			break;
 		}
-		fallthrough;
+		break;
 	default:
 		return kvm_psci_0_2_call(vcpu);
 	}
@@ -382,7 +401,7 @@ static int kvm_psci_0_1_call(struct kvm_vcpu *vcpu)
 
 	switch (psci_fn) {
 	case KVM_PSCI_FN_CPU_OFF:
-		kvm_psci_vcpu_off(vcpu);
+		kvm_arm_vcpu_power_off(vcpu);
 		val = PSCI_RET_SUCCESS;
 		break;
 	case KVM_PSCI_FN_CPU_ON:
@@ -437,186 +456,3 @@ int kvm_psci_call(struct kvm_vcpu *vcpu)
 		return -EINVAL;
 	}
 }
-
-int kvm_arm_get_fw_num_regs(struct kvm_vcpu *vcpu)
-{
-	return 4;		/* PSCI version and three workaround registers */
-}
-
-int kvm_arm_copy_fw_reg_indices(struct kvm_vcpu *vcpu, u64 __user *uindices)
-{
-	if (put_user(KVM_REG_ARM_PSCI_VERSION, uindices++))
-		return -EFAULT;
-
-	if (put_user(KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1, uindices++))
-		return -EFAULT;
-
-	if (put_user(KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2, uindices++))
-		return -EFAULT;
-
-	if (put_user(KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3, uindices++))
-		return -EFAULT;
-
-	return 0;
-}
-
-#define KVM_REG_FEATURE_LEVEL_WIDTH	4
-#define KVM_REG_FEATURE_LEVEL_MASK	(BIT(KVM_REG_FEATURE_LEVEL_WIDTH) - 1)
-
-/*
- * Convert the workaround level into an easy-to-compare number, where higher
- * values mean better protection.
- */
-static int get_kernel_wa_level(u64 regid)
-{
-	switch (regid) {
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
-		switch (arm64_get_spectre_v2_state()) {
-		case SPECTRE_VULNERABLE:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_NOT_AVAIL;
-		case SPECTRE_MITIGATED:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_AVAIL;
-		case SPECTRE_UNAFFECTED:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_NOT_REQUIRED;
-		}
-		return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_NOT_AVAIL;
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2:
-		switch (arm64_get_spectre_v4_state()) {
-		case SPECTRE_MITIGATED:
-			/*
-			 * As for the hypercall discovery, we pretend we
-			 * don't have any FW mitigation if SSBS is there at
-			 * all times.
-			 */
-			if (cpus_have_final_cap(ARM64_SSBS))
-				return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
-			fallthrough;
-		case SPECTRE_UNAFFECTED:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_REQUIRED;
-		case SPECTRE_VULNERABLE:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
-		}
-		break;
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3:
-		switch (arm64_get_spectre_bhb_state()) {
-		case SPECTRE_VULNERABLE:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_NOT_AVAIL;
-		case SPECTRE_MITIGATED:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_AVAIL;
-		case SPECTRE_UNAFFECTED:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_NOT_REQUIRED;
-		}
-		return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3_NOT_AVAIL;
-	}
-
-	return -EINVAL;
-}
-
-int kvm_arm_get_fw_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
-{
-	void __user *uaddr = (void __user *)(long)reg->addr;
-	u64 val;
-
-	switch (reg->id) {
-	case KVM_REG_ARM_PSCI_VERSION:
-		val = kvm_psci_version(vcpu);
-		break;
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2:
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3:
-		val = get_kernel_wa_level(reg->id) & KVM_REG_FEATURE_LEVEL_MASK;
-		break;
-	default:
-		return -ENOENT;
-	}
-
-	if (copy_to_user(uaddr, &val, KVM_REG_SIZE(reg->id)))
-		return -EFAULT;
-
-	return 0;
-}
-
-int kvm_arm_set_fw_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
-{
-	void __user *uaddr = (void __user *)(long)reg->addr;
-	u64 val;
-	int wa_level;
-
-	if (copy_from_user(&val, uaddr, KVM_REG_SIZE(reg->id)))
-		return -EFAULT;
-
-	switch (reg->id) {
-	case KVM_REG_ARM_PSCI_VERSION:
-	{
-		bool wants_02;
-
-		wants_02 = test_bit(KVM_ARM_VCPU_PSCI_0_2, vcpu->arch.features);
-
-		switch (val) {
-		case KVM_ARM_PSCI_0_1:
-			if (wants_02)
-				return -EINVAL;
-			vcpu->kvm->arch.psci_version = val;
-			return 0;
-		case KVM_ARM_PSCI_0_2:
-		case KVM_ARM_PSCI_1_0:
-		case KVM_ARM_PSCI_1_1:
-			if (!wants_02)
-				return -EINVAL;
-			vcpu->kvm->arch.psci_version = val;
-			return 0;
-		}
-		break;
-	}
-
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_3:
-		if (val & ~KVM_REG_FEATURE_LEVEL_MASK)
-			return -EINVAL;
-
-		if (get_kernel_wa_level(reg->id) < val)
-			return -EINVAL;
-
-		return 0;
-
-	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2:
-		if (val & ~(KVM_REG_FEATURE_LEVEL_MASK |
-			    KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_ENABLED))
-			return -EINVAL;
-
-		/* The enabled bit must not be set unless the level is AVAIL. */
-		if ((val & KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_ENABLED) &&
-		    (val & KVM_REG_FEATURE_LEVEL_MASK) != KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL)
-			return -EINVAL;
-
-		/*
-		 * Map all the possible incoming states to the only two we
-		 * really want to deal with.
-		 */
-		switch (val & KVM_REG_FEATURE_LEVEL_MASK) {
-		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL:
-		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_UNKNOWN:
-			wa_level = KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
-			break;
-		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL:
-		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_REQUIRED:
-			wa_level = KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_REQUIRED;
-			break;
-		default:
-			return -EINVAL;
-		}
-
-		/*
-		 * We can deal with NOT_AVAIL on NOT_REQUIRED, but not the
-		 * other way around.
-		 */
-		if (get_kernel_wa_level(reg->id) < wa_level)
-			return -EINVAL;
-
-		return 0;
-	default:
-		return -ENOENT;
-	}
-
-	return -EINVAL;
-}
diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index 18b403b58b53..c06c0477fab5 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -1145,6 +1145,8 @@ static u64 read_id_reg(const struct kvm_vcpu *vcpu,
 		if (!vcpu_has_ptrauth(vcpu))
 			val &= ~(ARM64_FEATURE_MASK(ID_AA64ISAR2_APA3) |
 				 ARM64_FEATURE_MASK(ID_AA64ISAR2_GPA3));
+		if (!cpus_have_final_cap(ARM64_HAS_WFXT))
+			val &= ~ARM64_FEATURE_MASK(ID_AA64ISAR2_WFXT);
 		break;
 	case SYS_ID_AA64DFR0_EL1:
 		/* Limit debug to ARMv8.0 */
@@ -2020,20 +2022,22 @@ static const struct sys_reg_desc cp14_64_regs[] = {
 	{ Op1( 0), CRm( 2), .access = trap_raz_wi },
 };
 
+#define CP15_PMU_SYS_REG(_map, _Op1, _CRn, _CRm, _Op2)			\
+	AA32(_map),							\
+	Op1(_Op1), CRn(_CRn), CRm(_CRm), Op2(_Op2),			\
+	.visibility = pmu_visibility
+
 /* Macro to expand the PMEVCNTRn register */
 #define PMU_PMEVCNTR(n)							\
-	/* PMEVCNTRn */							\
-	{ Op1(0), CRn(0b1110),						\
-	  CRm((0b1000 | (((n) >> 3) & 0x3))), Op2(((n) & 0x7)),		\
-	  access_pmu_evcntr }
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 0b1110,				\
+	  (0b1000 | (((n) >> 3) & 0x3)), ((n) & 0x7)),			\
+	  .access = access_pmu_evcntr }
 
 /* Macro to expand the PMEVTYPERn register */
 #define PMU_PMEVTYPER(n)						\
-	/* PMEVTYPERn */						\
-	{ Op1(0), CRn(0b1110),						\
-	  CRm((0b1100 | (((n) >> 3) & 0x3))), Op2(((n) & 0x7)),		\
-	  access_pmu_evtyper }
-
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 0b1110,				\
+	  (0b1100 | (((n) >> 3) & 0x3)), ((n) & 0x7)),			\
+	  .access = access_pmu_evtyper }
 /*
  * Trapped cp15 registers. TTBR0/TTBR1 get a double encoding,
  * depending on the way they are accessed (as a 32bit or a 64bit
@@ -2073,25 +2077,25 @@ static const struct sys_reg_desc cp15_regs[] = {
 	{ Op1( 0), CRn( 7), CRm(14), Op2( 2), access_dcsw },
 
 	/* PMU */
-	{ Op1( 0), CRn( 9), CRm(12), Op2( 0), access_pmcr },
-	{ Op1( 0), CRn( 9), CRm(12), Op2( 1), access_pmcnten },
-	{ Op1( 0), CRn( 9), CRm(12), Op2( 2), access_pmcnten },
-	{ Op1( 0), CRn( 9), CRm(12), Op2( 3), access_pmovs },
-	{ Op1( 0), CRn( 9), CRm(12), Op2( 4), access_pmswinc },
-	{ Op1( 0), CRn( 9), CRm(12), Op2( 5), access_pmselr },
-	{ AA32(LO), Op1( 0), CRn( 9), CRm(12), Op2( 6), access_pmceid },
-	{ AA32(LO), Op1( 0), CRn( 9), CRm(12), Op2( 7), access_pmceid },
-	{ Op1( 0), CRn( 9), CRm(13), Op2( 0), access_pmu_evcntr },
-	{ Op1( 0), CRn( 9), CRm(13), Op2( 1), access_pmu_evtyper },
-	{ Op1( 0), CRn( 9), CRm(13), Op2( 2), access_pmu_evcntr },
-	{ Op1( 0), CRn( 9), CRm(14), Op2( 0), access_pmuserenr },
-	{ Op1( 0), CRn( 9), CRm(14), Op2( 1), access_pminten },
-	{ Op1( 0), CRn( 9), CRm(14), Op2( 2), access_pminten },
-	{ Op1( 0), CRn( 9), CRm(14), Op2( 3), access_pmovs },
-	{ AA32(HI), Op1( 0), CRn( 9), CRm(14), Op2( 4), access_pmceid },
-	{ AA32(HI), Op1( 0), CRn( 9), CRm(14), Op2( 5), access_pmceid },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 12, 0), .access = access_pmcr },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 12, 1), .access = access_pmcnten },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 12, 2), .access = access_pmcnten },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 12, 3), .access = access_pmovs },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 12, 4), .access = access_pmswinc },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 12, 5), .access = access_pmselr },
+	{ CP15_PMU_SYS_REG(LO,     0, 9, 12, 6), .access = access_pmceid },
+	{ CP15_PMU_SYS_REG(LO,     0, 9, 12, 7), .access = access_pmceid },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 13, 0), .access = access_pmu_evcntr },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 13, 1), .access = access_pmu_evtyper },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 13, 2), .access = access_pmu_evcntr },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 14, 0), .access = access_pmuserenr },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 14, 1), .access = access_pminten },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 14, 2), .access = access_pminten },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 14, 3), .access = access_pmovs },
+	{ CP15_PMU_SYS_REG(HI,     0, 9, 14, 4), .access = access_pmceid },
+	{ CP15_PMU_SYS_REG(HI,     0, 9, 14, 5), .access = access_pmceid },
 	/* PMMIR */
-	{ Op1( 0), CRn( 9), CRm(14), Op2( 6), trap_raz_wi },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 9, 14, 6), .access = trap_raz_wi },
 
 	/* PRRR/MAIR0 */
 	{ AA32(LO), Op1( 0), CRn(10), CRm( 2), Op2( 0), access_vm_reg, NULL, MAIR_EL1 },
@@ -2176,7 +2180,7 @@ static const struct sys_reg_desc cp15_regs[] = {
 	PMU_PMEVTYPER(29),
 	PMU_PMEVTYPER(30),
 	/* PMCCFILTR */
-	{ Op1(0), CRn(14), CRm(15), Op2(7), access_pmu_evtyper },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 14, 15, 7), .access = access_pmu_evtyper },
 
 	{ Op1(1), CRn( 0), CRm( 0), Op2(0), access_ccsidr },
 	{ Op1(1), CRn( 0), CRm( 0), Op2(1), access_clidr },
@@ -2185,7 +2189,7 @@ static const struct sys_reg_desc cp15_regs[] = {
 
 static const struct sys_reg_desc cp15_64_regs[] = {
 	{ Op1( 0), CRn( 0), CRm( 2), Op2( 0), access_vm_reg, NULL, TTBR0_EL1 },
-	{ Op1( 0), CRn( 0), CRm( 9), Op2( 0), access_pmu_evcntr },
+	{ CP15_PMU_SYS_REG(DIRECT, 0, 0, 9, 0), .access = access_pmu_evcntr },
 	{ Op1( 0), CRn( 0), CRm(12), Op2( 0), access_gic_sgi }, /* ICC_SGI1R */
 	{ Op1( 1), CRn( 0), CRm( 2), Op2( 0), access_vm_reg, NULL, TTBR1_EL1 },
 	{ Op1( 1), CRn( 0), CRm(12), Op2( 0), access_gic_sgi }, /* ICC_ASGI1R */
@@ -2193,25 +2197,24 @@ static const struct sys_reg_desc cp15_64_regs[] = {
 	{ SYS_DESC(SYS_AARCH32_CNTP_CVAL),    access_arch_timer },
 };
 
-static int check_sysreg_table(const struct sys_reg_desc *table, unsigned int n,
-			      bool is_32)
+static bool check_sysreg_table(const struct sys_reg_desc *table, unsigned int n,
+			       bool is_32)
 {
 	unsigned int i;
 
 	for (i = 0; i < n; i++) {
 		if (!is_32 && table[i].reg && !table[i].reset) {
-			kvm_err("sys_reg table %p entry %d has lacks reset\n",
-				table, i);
-			return 1;
+			kvm_err("sys_reg table %pS entry %d lacks reset\n", &table[i], i);
+			return false;
 		}
 
 		if (i && cmp_sys_reg(&table[i-1], &table[i]) >= 0) {
-			kvm_err("sys_reg table %p out of order (%d)\n", table, i - 1);
-			return 1;
+			kvm_err("sys_reg table %pS entry %d out of order\n", &table[i - 1], i - 1);
+			return false;
 		}
 	}
 
-	return 0;
+	return true;
 }
 
 int kvm_handle_cp14_load_store(struct kvm_vcpu *vcpu)
@@ -2252,27 +2255,27 @@ static void perform_access(struct kvm_vcpu *vcpu,
  * @table: array of trap descriptors
  * @num: size of the trap descriptor array
  *
- * Return 0 if the access has been handled, and -1 if not.
+ * Return true if the access has been handled, false if not.
  */
-static int emulate_cp(struct kvm_vcpu *vcpu,
-		      struct sys_reg_params *params,
-		      const struct sys_reg_desc *table,
-		      size_t num)
+static bool emulate_cp(struct kvm_vcpu *vcpu,
+		       struct sys_reg_params *params,
+		       const struct sys_reg_desc *table,
+		       size_t num)
 {
 	const struct sys_reg_desc *r;
 
 	if (!table)
-		return -1;	/* Not handled */
+		return false;	/* Not handled */
 
 	r = find_reg(params, table, num);
 
 	if (r) {
 		perform_access(vcpu, params, r);
-		return 0;
+		return true;
 	}
 
 	/* Not handled */
-	return -1;
+	return false;
 }
 
 static void unhandled_cp_access(struct kvm_vcpu *vcpu,
@@ -2336,7 +2339,7 @@ static int kvm_handle_cp_64(struct kvm_vcpu *vcpu,
 	 * potential register operation in the case of a read and return
 	 * with success.
 	 */
-	if (!emulate_cp(vcpu, &params, global, nr_global)) {
+	if (emulate_cp(vcpu, &params, global, nr_global)) {
 		/* Split up the value between registers for the read side */
 		if (!params.is_write) {
 			vcpu_set_reg(vcpu, Rt, lower_32_bits(params.regval));
@@ -2350,34 +2353,144 @@ static int kvm_handle_cp_64(struct kvm_vcpu *vcpu,
 	return 1;
 }
 
+static bool emulate_sys_reg(struct kvm_vcpu *vcpu, struct sys_reg_params *params);
+
+/*
+ * The CP10 ID registers are architecturally mapped to AArch64 feature
+ * registers. Abuse that fact so we can rely on the AArch64 handler for accesses
+ * from AArch32.
+ */
+static bool kvm_esr_cp10_id_to_sys64(u64 esr, struct sys_reg_params *params)
+{
+	u8 reg_id = (esr >> 10) & 0xf;
+	bool valid;
+
+	params->is_write = ((esr & 1) == 0);
+	params->Op0 = 3;
+	params->Op1 = 0;
+	params->CRn = 0;
+	params->CRm = 3;
+
+	/* CP10 ID registers are read-only */
+	valid = !params->is_write;
+
+	switch (reg_id) {
+	/* MVFR0 */
+	case 0b0111:
+		params->Op2 = 0;
+		break;
+	/* MVFR1 */
+	case 0b0110:
+		params->Op2 = 1;
+		break;
+	/* MVFR2 */
+	case 0b0101:
+		params->Op2 = 2;
+		break;
+	default:
+		valid = false;
+	}
+
+	if (valid)
+		return true;
+
+	kvm_pr_unimpl("Unhandled cp10 register %s: %u\n",
+		      params->is_write ? "write" : "read", reg_id);
+	return false;
+}
+
+/**
+ * kvm_handle_cp10_id() - Handles a VMRS trap on guest access to a 'Media and
+ *			  VFP Register' from AArch32.
+ * @vcpu: The vCPU pointer
+ *
+ * MVFR{0-2} are architecturally mapped to the AArch64 MVFR{0-2}_EL1 registers.
+ * Work out the correct AArch64 system register encoding and reroute to the
+ * AArch64 system register emulation.
+ */
+int kvm_handle_cp10_id(struct kvm_vcpu *vcpu)
+{
+	int Rt = kvm_vcpu_sys_get_rt(vcpu);
+	u64 esr = kvm_vcpu_get_esr(vcpu);
+	struct sys_reg_params params;
+
+	/* UNDEF on any unhandled register access */
+	if (!kvm_esr_cp10_id_to_sys64(esr, &params)) {
+		kvm_inject_undefined(vcpu);
+		return 1;
+	}
+
+	if (emulate_sys_reg(vcpu, &params))
+		vcpu_set_reg(vcpu, Rt, params.regval);
+
+	return 1;
+}
+
+/**
+ * kvm_emulate_cp15_id_reg() - Handles an MRC trap on a guest CP15 access where
+ *			       CRn=0, which corresponds to the AArch32 feature
+ *			       registers.
+ * @vcpu: the vCPU pointer
+ * @params: the system register access parameters.
+ *
+ * Our cp15 system register tables do not enumerate the AArch32 feature
+ * registers. Conveniently, our AArch64 table does, and the AArch32 system
+ * register encoding can be trivially remapped into the AArch64 for the feature
+ * registers: Append op0=3, leaving op1, CRn, CRm, and op2 the same.
+ *
+ * According to DDI0487G.b G7.3.1, paragraph "Behavior of VMSAv8-32 32-bit
+ * System registers with (coproc=0b1111, CRn==c0)", read accesses from this
+ * range are either UNKNOWN or RES0. Rerouting remains architectural as we
+ * treat undefined registers in this range as RAZ.
+ */
+static int kvm_emulate_cp15_id_reg(struct kvm_vcpu *vcpu,
+				   struct sys_reg_params *params)
+{
+	int Rt = kvm_vcpu_sys_get_rt(vcpu);
+
+	/* Treat impossible writes to RO registers as UNDEFINED */
+	if (params->is_write) {
+		unhandled_cp_access(vcpu, params);
+		return 1;
+	}
+
+	params->Op0 = 3;
+
+	/*
+	 * All registers where CRm > 3 are known to be UNKNOWN/RAZ from AArch32.
+	 * Avoid conflicting with future expansion of AArch64 feature registers
+	 * and simply treat them as RAZ here.
+	 */
+	if (params->CRm > 3)
+		params->regval = 0;
+	else if (!emulate_sys_reg(vcpu, params))
+		return 1;
+
+	vcpu_set_reg(vcpu, Rt, params->regval);
+	return 1;
+}
+
 /**
  * kvm_handle_cp_32 -- handles a mrc/mcr trap on a guest CP14/CP15 access
  * @vcpu: The VCPU pointer
  * @run:  The kvm_run struct
  */
 static int kvm_handle_cp_32(struct kvm_vcpu *vcpu,
+			    struct sys_reg_params *params,
 			    const struct sys_reg_desc *global,
 			    size_t nr_global)
 {
-	struct sys_reg_params params;
-	u64 esr = kvm_vcpu_get_esr(vcpu);
 	int Rt  = kvm_vcpu_sys_get_rt(vcpu);
 
-	params.CRm = (esr >> 1) & 0xf;
-	params.regval = vcpu_get_reg(vcpu, Rt);
-	params.is_write = ((esr & 1) == 0);
-	params.CRn = (esr >> 10) & 0xf;
-	params.Op0 = 0;
-	params.Op1 = (esr >> 14) & 0x7;
-	params.Op2 = (esr >> 17) & 0x7;
+	params->regval = vcpu_get_reg(vcpu, Rt);
 
-	if (!emulate_cp(vcpu, &params, global, nr_global)) {
-		if (!params.is_write)
-			vcpu_set_reg(vcpu, Rt, params.regval);
+	if (emulate_cp(vcpu, params, global, nr_global)) {
+		if (!params->is_write)
+			vcpu_set_reg(vcpu, Rt, params->regval);
 		return 1;
 	}
 
-	unhandled_cp_access(vcpu, &params);
+	unhandled_cp_access(vcpu, params);
 	return 1;
 }
 
@@ -2388,7 +2501,20 @@ int kvm_handle_cp15_64(struct kvm_vcpu *vcpu)
 
 int kvm_handle_cp15_32(struct kvm_vcpu *vcpu)
 {
-	return kvm_handle_cp_32(vcpu, cp15_regs, ARRAY_SIZE(cp15_regs));
+	struct sys_reg_params params;
+
+	params = esr_cp1x_32_to_params(kvm_vcpu_get_esr(vcpu));
+
+	/*
+	 * Certain AArch32 ID registers are handled by rerouting to the AArch64
+	 * system register table. Registers in the ID range where CRm=0 are
+	 * excluded from this scheme as they do not trivially map into AArch64
+	 * system register encodings.
+	 */
+	if (params.Op1 == 0 && params.CRn == 0 && params.CRm)
+		return kvm_emulate_cp15_id_reg(vcpu, &params);
+
+	return kvm_handle_cp_32(vcpu, &params, cp15_regs, ARRAY_SIZE(cp15_regs));
 }
 
 int kvm_handle_cp14_64(struct kvm_vcpu *vcpu)
@@ -2398,7 +2524,11 @@ int kvm_handle_cp14_64(struct kvm_vcpu *vcpu)
 
 int kvm_handle_cp14_32(struct kvm_vcpu *vcpu)
 {
-	return kvm_handle_cp_32(vcpu, cp14_regs, ARRAY_SIZE(cp14_regs));
+	struct sys_reg_params params;
+
+	params = esr_cp1x_32_to_params(kvm_vcpu_get_esr(vcpu));
+
+	return kvm_handle_cp_32(vcpu, &params, cp14_regs, ARRAY_SIZE(cp14_regs));
 }
 
 static bool is_imp_def_sys_reg(struct sys_reg_params *params)
@@ -2407,7 +2537,14 @@ static bool is_imp_def_sys_reg(struct sys_reg_params *params)
 	return params->Op0 == 3 && (params->CRn & 0b1011) == 0b1011;
 }
 
-static int emulate_sys_reg(struct kvm_vcpu *vcpu,
+/**
+ * emulate_sys_reg - Emulate a guest access to an AArch64 system register
+ * @vcpu: The VCPU pointer
+ * @params: Decoded system register parameters
+ *
+ * Return: true if the system register access was successful, false otherwise.
+ */
+static bool emulate_sys_reg(struct kvm_vcpu *vcpu,
 			   struct sys_reg_params *params)
 {
 	const struct sys_reg_desc *r;
@@ -2416,7 +2553,10 @@ static int emulate_sys_reg(struct kvm_vcpu *vcpu,
 
 	if (likely(r)) {
 		perform_access(vcpu, params, r);
-	} else if (is_imp_def_sys_reg(params)) {
+		return true;
+	}
+
+	if (is_imp_def_sys_reg(params)) {
 		kvm_inject_undefined(vcpu);
 	} else {
 		print_sys_reg_msg(params,
@@ -2424,7 +2564,7 @@ static int emulate_sys_reg(struct kvm_vcpu *vcpu,
 				  *vcpu_pc(vcpu), *vcpu_cpsr(vcpu));
 		kvm_inject_undefined(vcpu);
 	}
-	return 1;
+	return false;
 }
 
 /**
@@ -2452,18 +2592,18 @@ int kvm_handle_sys_reg(struct kvm_vcpu *vcpu)
 	struct sys_reg_params params;
 	unsigned long esr = kvm_vcpu_get_esr(vcpu);
 	int Rt = kvm_vcpu_sys_get_rt(vcpu);
-	int ret;
 
 	trace_kvm_handle_sys_reg(esr);
 
 	params = esr_sys64_to_params(esr);
 	params.regval = vcpu_get_reg(vcpu, Rt);
 
-	ret = emulate_sys_reg(vcpu, &params);
+	if (!emulate_sys_reg(vcpu, &params))
+		return 1;
 
 	if (!params.is_write)
 		vcpu_set_reg(vcpu, Rt, params.regval);
-	return ret;
+	return 1;
 }
 
 /******************************************************************************
@@ -2866,18 +3006,22 @@ int kvm_arm_copy_sys_reg_indices(struct kvm_vcpu *vcpu, u64 __user *uindices)
 	return write_demux_regids(uindices);
 }
 
-void kvm_sys_reg_table_init(void)
+int kvm_sys_reg_table_init(void)
 {
+	bool valid = true;
 	unsigned int i;
 	struct sys_reg_desc clidr;
 
 	/* Make sure tables are unique and in order. */
-	BUG_ON(check_sysreg_table(sys_reg_descs, ARRAY_SIZE(sys_reg_descs), false));
-	BUG_ON(check_sysreg_table(cp14_regs, ARRAY_SIZE(cp14_regs), true));
-	BUG_ON(check_sysreg_table(cp14_64_regs, ARRAY_SIZE(cp14_64_regs), true));
-	BUG_ON(check_sysreg_table(cp15_regs, ARRAY_SIZE(cp15_regs), true));
-	BUG_ON(check_sysreg_table(cp15_64_regs, ARRAY_SIZE(cp15_64_regs), true));
-	BUG_ON(check_sysreg_table(invariant_sys_regs, ARRAY_SIZE(invariant_sys_regs), false));
+	valid &= check_sysreg_table(sys_reg_descs, ARRAY_SIZE(sys_reg_descs), false);
+	valid &= check_sysreg_table(cp14_regs, ARRAY_SIZE(cp14_regs), true);
+	valid &= check_sysreg_table(cp14_64_regs, ARRAY_SIZE(cp14_64_regs), true);
+	valid &= check_sysreg_table(cp15_regs, ARRAY_SIZE(cp15_regs), true);
+	valid &= check_sysreg_table(cp15_64_regs, ARRAY_SIZE(cp15_64_regs), true);
+	valid &= check_sysreg_table(invariant_sys_regs, ARRAY_SIZE(invariant_sys_regs), false);
+
+	if (!valid)
+		return -EINVAL;
 
 	/* We abuse the reset function to overwrite the table itself. */
 	for (i = 0; i < ARRAY_SIZE(invariant_sys_regs); i++)
@@ -2900,4 +3044,6 @@ void kvm_sys_reg_table_init(void)
 			break;
 	/* Clear all higher bits. */
 	cache_levels &= (1 << (i*3))-1;
+
+	return 0;
 }
diff --git a/arch/arm64/kvm/sys_regs.h b/arch/arm64/kvm/sys_regs.h
index cc0cc95a0280..aee8ea054f0d 100644
--- a/arch/arm64/kvm/sys_regs.h
+++ b/arch/arm64/kvm/sys_regs.h
@@ -35,12 +35,19 @@ struct sys_reg_params {
 				  .Op2 = ((esr) >> 17) & 0x7,                  \
 				  .is_write = !((esr) & 1) })
 
+#define esr_cp1x_32_to_params(esr)						\
+	((struct sys_reg_params){ .Op1 = ((esr) >> 14) & 0x7,			\
+				  .CRn = ((esr) >> 10) & 0xf,			\
+				  .CRm = ((esr) >> 1) & 0xf,			\
+				  .Op2 = ((esr) >> 17) & 0x7,			\
+				  .is_write = !((esr) & 1) })
+
 struct sys_reg_desc {
 	/* Sysreg string for debug */
 	const char *name;
 
 	enum {
-		AA32_ZEROHIGH,
+		AA32_DIRECT,
 		AA32_LO,
 		AA32_HI,
 	} aarch32_map;
diff --git a/arch/arm64/kvm/vgic/vgic-init.c b/arch/arm64/kvm/vgic/vgic-init.c
index fc00304fe7d8..f6d4f4052555 100644
--- a/arch/arm64/kvm/vgic/vgic-init.c
+++ b/arch/arm64/kvm/vgic/vgic-init.c
@@ -98,11 +98,11 @@ int kvm_vgic_create(struct kvm *kvm, u32 type)
 	ret = 0;
 
 	if (type == KVM_DEV_TYPE_ARM_VGIC_V2)
-		kvm->arch.max_vcpus = VGIC_V2_MAX_CPUS;
+		kvm->max_vcpus = VGIC_V2_MAX_CPUS;
 	else
-		kvm->arch.max_vcpus = VGIC_V3_MAX_CPUS;
+		kvm->max_vcpus = VGIC_V3_MAX_CPUS;
 
-	if (atomic_read(&kvm->online_vcpus) > kvm->arch.max_vcpus) {
+	if (atomic_read(&kvm->online_vcpus) > kvm->max_vcpus) {
 		ret = -E2BIG;
 		goto out_unlock;
 	}
@@ -319,7 +319,12 @@ int vgic_init(struct kvm *kvm)
 
 	vgic_debug_init(kvm);
 
-	dist->implementation_rev = 2;
+	/*
+	 * If userspace didn't set the GIC implementation revision,
+	 * default to the latest and greatest. You know want it.
+	 */
+	if (!dist->implementation_rev)
+		dist->implementation_rev = KVM_VGIC_IMP_REV_LATEST;
 	dist->initialized = true;
 
 out:
diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c
index 2e13402be3bd..9d3299a70242 100644
--- a/arch/arm64/kvm/vgic/vgic-its.c
+++ b/arch/arm64/kvm/vgic/vgic-its.c
@@ -683,7 +683,7 @@ int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
 	if (!vcpu)
 		return E_ITS_INT_UNMAPPED_INTERRUPT;
 
-	if (!vcpu->arch.vgic_cpu.lpis_enabled)
+	if (!vgic_lpis_enabled(vcpu))
 		return -EBUSY;
 
 	vgic_its_cache_translation(kvm, its, devid, eventid, ite->irq);
@@ -894,6 +894,18 @@ static int vgic_its_cmd_handle_movi(struct kvm *kvm, struct vgic_its *its,
 	return update_affinity(ite->irq, vcpu);
 }
 
+static bool __is_visible_gfn_locked(struct vgic_its *its, gpa_t gpa)
+{
+	gfn_t gfn = gpa >> PAGE_SHIFT;
+	int idx;
+	bool ret;
+
+	idx = srcu_read_lock(&its->dev->kvm->srcu);
+	ret = kvm_is_visible_gfn(its->dev->kvm, gfn);
+	srcu_read_unlock(&its->dev->kvm->srcu, idx);
+	return ret;
+}
+
 /*
  * Check whether an ID can be stored into the corresponding guest table.
  * For a direct table this is pretty easy, but gets a bit nasty for
@@ -908,9 +920,7 @@ static bool vgic_its_check_id(struct vgic_its *its, u64 baser, u32 id,
 	u64 indirect_ptr, type = GITS_BASER_TYPE(baser);
 	phys_addr_t base = GITS_BASER_ADDR_48_to_52(baser);
 	int esz = GITS_BASER_ENTRY_SIZE(baser);
-	int index, idx;
-	gfn_t gfn;
-	bool ret;
+	int index;
 
 	switch (type) {
 	case GITS_BASER_TYPE_DEVICE:
@@ -933,12 +943,11 @@ static bool vgic_its_check_id(struct vgic_its *its, u64 baser, u32 id,
 			return false;
 
 		addr = base + id * esz;
-		gfn = addr >> PAGE_SHIFT;
 
 		if (eaddr)
 			*eaddr = addr;
 
-		goto out;
+		return __is_visible_gfn_locked(its, addr);
 	}
 
 	/* calculate and check the index into the 1st level */
@@ -964,27 +973,42 @@ static bool vgic_its_check_id(struct vgic_its *its, u64 baser, u32 id,
 	/* Find the address of the actual entry */
 	index = id % (SZ_64K / esz);
 	indirect_ptr += index * esz;
-	gfn = indirect_ptr >> PAGE_SHIFT;
 
 	if (eaddr)
 		*eaddr = indirect_ptr;
 
-out:
-	idx = srcu_read_lock(&its->dev->kvm->srcu);
-	ret = kvm_is_visible_gfn(its->dev->kvm, gfn);
-	srcu_read_unlock(&its->dev->kvm->srcu, idx);
-	return ret;
+	return __is_visible_gfn_locked(its, indirect_ptr);
 }
 
+/*
+ * Check whether an event ID can be stored in the corresponding Interrupt
+ * Translation Table, which starts at device->itt_addr.
+ */
+static bool vgic_its_check_event_id(struct vgic_its *its, struct its_device *device,
+		u32 event_id)
+{
+	const struct vgic_its_abi *abi = vgic_its_get_abi(its);
+	int ite_esz = abi->ite_esz;
+	gpa_t gpa;
+
+	/* max table size is: BIT_ULL(device->num_eventid_bits) * ite_esz */
+	if (event_id >= BIT_ULL(device->num_eventid_bits))
+		return false;
+
+	gpa = device->itt_addr + event_id * ite_esz;
+	return __is_visible_gfn_locked(its, gpa);
+}
+
+/*
+ * Add a new collection into the ITS collection table.
+ * Returns 0 on success, and a negative error value for generic errors.
+ */
 static int vgic_its_alloc_collection(struct vgic_its *its,
 				     struct its_collection **colp,
 				     u32 coll_id)
 {
 	struct its_collection *collection;
 
-	if (!vgic_its_check_id(its, its->baser_coll_table, coll_id, NULL))
-		return E_ITS_MAPC_COLLECTION_OOR;
-
 	collection = kzalloc(sizeof(*collection), GFP_KERNEL_ACCOUNT);
 	if (!collection)
 		return -ENOMEM;
@@ -1061,7 +1085,7 @@ static int vgic_its_cmd_handle_mapi(struct kvm *kvm, struct vgic_its *its,
 	if (!device)
 		return E_ITS_MAPTI_UNMAPPED_DEVICE;
 
-	if (event_id >= BIT_ULL(device->num_eventid_bits))
+	if (!vgic_its_check_event_id(its, device, event_id))
 		return E_ITS_MAPTI_ID_OOR;
 
 	if (its_cmd_get_command(its_cmd) == GITS_CMD_MAPTI)
@@ -1078,7 +1102,12 @@ static int vgic_its_cmd_handle_mapi(struct kvm *kvm, struct vgic_its *its,
 
 	collection = find_collection(its, coll_id);
 	if (!collection) {
-		int ret = vgic_its_alloc_collection(its, &collection, coll_id);
+		int ret;
+
+		if (!vgic_its_check_id(its, its->baser_coll_table, coll_id, NULL))
+			return E_ITS_MAPC_COLLECTION_OOR;
+
+		ret = vgic_its_alloc_collection(its, &collection, coll_id);
 		if (ret)
 			return ret;
 		new_coll = collection;
@@ -1233,6 +1262,10 @@ static int vgic_its_cmd_handle_mapc(struct kvm *kvm, struct vgic_its *its,
 		if (!collection) {
 			int ret;
 
+			if (!vgic_its_check_id(its, its->baser_coll_table,
+						coll_id, NULL))
+				return E_ITS_MAPC_COLLECTION_OOR;
+
 			ret = vgic_its_alloc_collection(its, &collection,
 							coll_id);
 			if (ret)
@@ -1272,6 +1305,11 @@ static int vgic_its_cmd_handle_clear(struct kvm *kvm, struct vgic_its *its,
 	return 0;
 }
 
+int vgic_its_inv_lpi(struct kvm *kvm, struct vgic_irq *irq)
+{
+	return update_lpi_config(kvm, irq, NULL, true);
+}
+
 /*
  * The INV command syncs the configuration bits from the memory table.
  * Must be called with the its_lock mutex held.
@@ -1288,7 +1326,41 @@ static int vgic_its_cmd_handle_inv(struct kvm *kvm, struct vgic_its *its,
 	if (!ite)
 		return E_ITS_INV_UNMAPPED_INTERRUPT;
 
-	return update_lpi_config(kvm, ite->irq, NULL, true);
+	return vgic_its_inv_lpi(kvm, ite->irq);
+}
+
+/**
+ * vgic_its_invall - invalidate all LPIs targetting a given vcpu
+ * @vcpu: the vcpu for which the RD is targetted by an invalidation
+ *
+ * Contrary to the INVALL command, this targets a RD instead of a
+ * collection, and we don't need to hold the its_lock, since no ITS is
+ * involved here.
+ */
+int vgic_its_invall(struct kvm_vcpu *vcpu)
+{
+	struct kvm *kvm = vcpu->kvm;
+	int irq_count, i = 0;
+	u32 *intids;
+
+	irq_count = vgic_copy_lpi_list(kvm, vcpu, &intids);
+	if (irq_count < 0)
+		return irq_count;
+
+	for (i = 0; i < irq_count; i++) {
+		struct vgic_irq *irq = vgic_get_irq(kvm, NULL, intids[i]);
+		if (!irq)
+			continue;
+		update_lpi_config(kvm, irq, vcpu, false);
+		vgic_put_irq(kvm, irq);
+	}
+
+	kfree(intids);
+
+	if (vcpu->arch.vgic_cpu.vgic_v3.its_vpe.its_vm)
+		its_invall_vpe(&vcpu->arch.vgic_cpu.vgic_v3.its_vpe);
+
+	return 0;
 }
 
 /*
@@ -1305,32 +1377,13 @@ static int vgic_its_cmd_handle_invall(struct kvm *kvm, struct vgic_its *its,
 	u32 coll_id = its_cmd_get_collection(its_cmd);
 	struct its_collection *collection;
 	struct kvm_vcpu *vcpu;
-	struct vgic_irq *irq;
-	u32 *intids;
-	int irq_count, i;
 
 	collection = find_collection(its, coll_id);
 	if (!its_is_collection_mapped(collection))
 		return E_ITS_INVALL_UNMAPPED_COLLECTION;
 
 	vcpu = kvm_get_vcpu(kvm, collection->target_addr);
-
-	irq_count = vgic_copy_lpi_list(kvm, vcpu, &intids);
-	if (irq_count < 0)
-		return irq_count;
-
-	for (i = 0; i < irq_count; i++) {
-		irq = vgic_get_irq(kvm, NULL, intids[i]);
-		if (!irq)
-			continue;
-		update_lpi_config(kvm, irq, vcpu, false);
-		vgic_put_irq(kvm, irq);
-	}
-
-	kfree(intids);
-
-	if (vcpu->arch.vgic_cpu.vgic_v3.its_vpe.its_vm)
-		its_invall_vpe(&vcpu->arch.vgic_cpu.vgic_v3.its_vpe);
+	vgic_its_invall(vcpu);
 
 	return 0;
 }
@@ -2175,6 +2228,9 @@ static int vgic_its_restore_ite(struct vgic_its *its, u32 event_id,
 	if (!collection)
 		return -EINVAL;
 
+	if (!vgic_its_check_event_id(its, dev, event_id))
+		return -EINVAL;
+
 	ite = vgic_its_alloc_ite(dev, collection, event_id);
 	if (IS_ERR(ite))
 		return PTR_ERR(ite);
@@ -2183,8 +2239,10 @@ static int vgic_its_restore_ite(struct vgic_its *its, u32 event_id,
 		vcpu = kvm_get_vcpu(kvm, collection->target_addr);
 
 	irq = vgic_add_lpi(kvm, lpi_id, vcpu);
-	if (IS_ERR(irq))
+	if (IS_ERR(irq)) {
+		its_free_ite(kvm, ite);
 		return PTR_ERR(irq);
+	}
 	ite->irq = irq;
 
 	return offset;
@@ -2296,6 +2354,7 @@ static int vgic_its_restore_dte(struct vgic_its *its, u32 id,
 				void *ptr, void *opaque)
 {
 	struct its_device *dev;
+	u64 baser = its->baser_device_table;
 	gpa_t itt_addr;
 	u8 num_eventid_bits;
 	u64 entry = *(u64 *)ptr;
@@ -2316,6 +2375,9 @@ static int vgic_its_restore_dte(struct vgic_its *its, u32 id,
 	/* dte entry is valid */
 	offset = (entry & KVM_ITS_DTE_NEXT_MASK) >> KVM_ITS_DTE_NEXT_SHIFT;
 
+	if (!vgic_its_check_id(its, baser, id, NULL))
+		return -EINVAL;
+
 	dev = vgic_its_alloc_device(its, id, itt_addr, num_eventid_bits);
 	if (IS_ERR(dev))
 		return PTR_ERR(dev);
@@ -2445,6 +2507,9 @@ static int vgic_its_restore_device_tables(struct vgic_its *its)
 	if (ret > 0)
 		ret = 0;
 
+	if (ret < 0)
+		vgic_its_free_device_list(its->dev->kvm, its);
+
 	return ret;
 }
 
@@ -2461,6 +2526,11 @@ static int vgic_its_save_cte(struct vgic_its *its,
 	return kvm_write_guest_lock(its->dev->kvm, gpa, &val, esz);
 }
 
+/*
+ * Restore a collection entry into the ITS collection table.
+ * Return +1 on success, 0 if the entry was invalid (which should be
+ * interpreted as end-of-table), and a negative error value for generic errors.
+ */
 static int vgic_its_restore_cte(struct vgic_its *its, gpa_t gpa, int esz)
 {
 	struct its_collection *collection;
@@ -2487,6 +2557,10 @@ static int vgic_its_restore_cte(struct vgic_its *its, gpa_t gpa, int esz)
 	collection = find_collection(its, coll_id);
 	if (collection)
 		return -EEXIST;
+
+	if (!vgic_its_check_id(its, its->baser_coll_table, coll_id, NULL))
+		return -EINVAL;
+
 	ret = vgic_its_alloc_collection(its, &collection, coll_id);
 	if (ret)
 		return ret;
@@ -2566,6 +2640,9 @@ static int vgic_its_restore_collection_table(struct vgic_its *its)
 	if (ret > 0)
 		return 0;
 
+	if (ret < 0)
+		vgic_its_free_collection_list(its->dev->kvm, its);
+
 	return ret;
 }
 
@@ -2597,7 +2674,10 @@ static int vgic_its_restore_tables_v0(struct vgic_its *its)
 	if (ret)
 		return ret;
 
-	return vgic_its_restore_device_tables(its);
+	ret = vgic_its_restore_device_tables(its);
+	if (ret)
+		vgic_its_free_collection_list(its->dev->kvm, its);
+	return ret;
 }
 
 static int vgic_its_commit_v0(struct vgic_its *its)
diff --git a/arch/arm64/kvm/vgic/vgic-mmio-v2.c b/arch/arm64/kvm/vgic/vgic-mmio-v2.c
index 12e4c223e6b8..77a67e9d3d14 100644
--- a/arch/arm64/kvm/vgic/vgic-mmio-v2.c
+++ b/arch/arm64/kvm/vgic/vgic-mmio-v2.c
@@ -73,9 +73,13 @@ static int vgic_mmio_uaccess_write_v2_misc(struct kvm_vcpu *vcpu,
 					   gpa_t addr, unsigned int len,
 					   unsigned long val)
 {
+	struct vgic_dist *dist = &vcpu->kvm->arch.vgic;
+	u32 reg;
+
 	switch (addr & 0x0c) {
 	case GIC_DIST_IIDR:
-		if (val != vgic_mmio_read_v2_misc(vcpu, addr, len))
+		reg = vgic_mmio_read_v2_misc(vcpu, addr, len);
+		if ((reg ^ val) & ~GICD_IIDR_REVISION_MASK)
 			return -EINVAL;
 
 		/*
@@ -87,8 +91,16 @@ static int vgic_mmio_uaccess_write_v2_misc(struct kvm_vcpu *vcpu,
 		 * migration from old kernels to new kernels with legacy
 		 * userspace.
 		 */
-		vcpu->kvm->arch.vgic.v2_groups_user_writable = true;
-		return 0;
+		reg = FIELD_GET(GICD_IIDR_REVISION_MASK, reg);
+		switch (reg) {
+		case KVM_VGIC_IMP_REV_2:
+		case KVM_VGIC_IMP_REV_3:
+			vcpu->kvm->arch.vgic.v2_groups_user_writable = true;
+			dist->implementation_rev = reg;
+			return 0;
+		default:
+			return -EINVAL;
+		}
 	}
 
 	vgic_mmio_write_v2_misc(vcpu, addr, len, val);
diff --git a/arch/arm64/kvm/vgic/vgic-mmio-v3.c b/arch/arm64/kvm/vgic/vgic-mmio-v3.c
index 58e40b4874f8..f7aa7bcd6fb8 100644
--- a/arch/arm64/kvm/vgic/vgic-mmio-v3.c
+++ b/arch/arm64/kvm/vgic/vgic-mmio-v3.c
@@ -155,13 +155,27 @@ static int vgic_mmio_uaccess_write_v3_misc(struct kvm_vcpu *vcpu,
 					   unsigned long val)
 {
 	struct vgic_dist *dist = &vcpu->kvm->arch.vgic;
+	u32 reg;
 
 	switch (addr & 0x0c) {
 	case GICD_TYPER2:
-	case GICD_IIDR:
 		if (val != vgic_mmio_read_v3_misc(vcpu, addr, len))
 			return -EINVAL;
 		return 0;
+	case GICD_IIDR:
+		reg = vgic_mmio_read_v3_misc(vcpu, addr, len);
+		if ((reg ^ val) & ~GICD_IIDR_REVISION_MASK)
+			return -EINVAL;
+
+		reg = FIELD_GET(GICD_IIDR_REVISION_MASK, reg);
+		switch (reg) {
+		case KVM_VGIC_IMP_REV_2:
+		case KVM_VGIC_IMP_REV_3:
+			dist->implementation_rev = reg;
+			return 0;
+		default:
+			return -EINVAL;
+		}
 	case GICD_CTLR:
 		/* Not a GICv4.1? No HW SGIs */
 		if (!kvm_vgic_global_state.has_gicv4_1)
@@ -221,34 +235,58 @@ static void vgic_mmio_write_irouter(struct kvm_vcpu *vcpu,
 	vgic_put_irq(vcpu->kvm, irq);
 }
 
+bool vgic_lpis_enabled(struct kvm_vcpu *vcpu)
+{
+	struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu;
+
+	return atomic_read(&vgic_cpu->ctlr) == GICR_CTLR_ENABLE_LPIS;
+}
+
 static unsigned long vgic_mmio_read_v3r_ctlr(struct kvm_vcpu *vcpu,
 					     gpa_t addr, unsigned int len)
 {
 	struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu;
+	unsigned long val;
 
-	return vgic_cpu->lpis_enabled ? GICR_CTLR_ENABLE_LPIS : 0;
-}
+	val = atomic_read(&vgic_cpu->ctlr);
+	if (vgic_get_implementation_rev(vcpu) >= KVM_VGIC_IMP_REV_3)
+		val |= GICR_CTLR_IR | GICR_CTLR_CES;
 
+	return val;
+}
 
 static void vgic_mmio_write_v3r_ctlr(struct kvm_vcpu *vcpu,
 				     gpa_t addr, unsigned int len,
 				     unsigned long val)
 {
 	struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu;
-	bool was_enabled = vgic_cpu->lpis_enabled;
+	u32 ctlr;
 
 	if (!vgic_has_its(vcpu->kvm))
 		return;
 
-	vgic_cpu->lpis_enabled = val & GICR_CTLR_ENABLE_LPIS;
+	if (!(val & GICR_CTLR_ENABLE_LPIS)) {
+		/*
+		 * Don't disable if RWP is set, as there already an
+		 * ongoing disable. Funky guest...
+		 */
+		ctlr = atomic_cmpxchg_acquire(&vgic_cpu->ctlr,
+					      GICR_CTLR_ENABLE_LPIS,
+					      GICR_CTLR_RWP);
+		if (ctlr != GICR_CTLR_ENABLE_LPIS)
+			return;
 
-	if (was_enabled && !vgic_cpu->lpis_enabled) {
 		vgic_flush_pending_lpis(vcpu);
 		vgic_its_invalidate_cache(vcpu->kvm);
-	}
+		atomic_set_release(&vgic_cpu->ctlr, 0);
+	} else {
+		ctlr = atomic_cmpxchg_acquire(&vgic_cpu->ctlr, 0,
+					      GICR_CTLR_ENABLE_LPIS);
+		if (ctlr != 0)
+			return;
 
-	if (!was_enabled && vgic_cpu->lpis_enabled)
 		vgic_enable_lpis(vcpu);
+	}
 }
 
 static bool vgic_mmio_vcpu_rdist_is_last(struct kvm_vcpu *vcpu)
@@ -478,11 +516,10 @@ static void vgic_mmio_write_propbase(struct kvm_vcpu *vcpu,
 				     unsigned long val)
 {
 	struct vgic_dist *dist = &vcpu->kvm->arch.vgic;
-	struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu;
 	u64 old_propbaser, propbaser;
 
 	/* Storing a value with LPIs already enabled is undefined */
-	if (vgic_cpu->lpis_enabled)
+	if (vgic_lpis_enabled(vcpu))
 		return;
 
 	do {
@@ -513,7 +550,7 @@ static void vgic_mmio_write_pendbase(struct kvm_vcpu *vcpu,
 	u64 old_pendbaser, pendbaser;
 
 	/* Storing a value with LPIs already enabled is undefined */
-	if (vgic_cpu->lpis_enabled)
+	if (vgic_lpis_enabled(vcpu))
 		return;
 
 	do {
@@ -525,6 +562,63 @@ static void vgic_mmio_write_pendbase(struct kvm_vcpu *vcpu,
 			   pendbaser) != old_pendbaser);
 }
 
+static unsigned long vgic_mmio_read_sync(struct kvm_vcpu *vcpu,
+					 gpa_t addr, unsigned int len)
+{
+	return !!atomic_read(&vcpu->arch.vgic_cpu.syncr_busy);
+}
+
+static void vgic_set_rdist_busy(struct kvm_vcpu *vcpu, bool busy)
+{
+	if (busy) {
+		atomic_inc(&vcpu->arch.vgic_cpu.syncr_busy);
+		smp_mb__after_atomic();
+	} else {
+		smp_mb__before_atomic();
+		atomic_dec(&vcpu->arch.vgic_cpu.syncr_busy);
+	}
+}
+
+static void vgic_mmio_write_invlpi(struct kvm_vcpu *vcpu,
+				   gpa_t addr, unsigned int len,
+				   unsigned long val)
+{
+	struct vgic_irq *irq;
+
+	/*
+	 * If the guest wrote only to the upper 32bit part of the
+	 * register, drop the write on the floor, as it is only for
+	 * vPEs (which we don't support for obvious reasons).
+	 *
+	 * Also discard the access if LPIs are not enabled.
+	 */
+	if ((addr & 4) || !vgic_lpis_enabled(vcpu))
+		return;
+
+	vgic_set_rdist_busy(vcpu, true);
+
+	irq = vgic_get_irq(vcpu->kvm, NULL, lower_32_bits(val));
+	if (irq) {
+		vgic_its_inv_lpi(vcpu->kvm, irq);
+		vgic_put_irq(vcpu->kvm, irq);
+	}
+
+	vgic_set_rdist_busy(vcpu, false);
+}
+
+static void vgic_mmio_write_invall(struct kvm_vcpu *vcpu,
+				   gpa_t addr, unsigned int len,
+				   unsigned long val)
+{
+	/* See vgic_mmio_write_invlpi() for the early return rationale */
+	if ((addr & 4) || !vgic_lpis_enabled(vcpu))
+		return;
+
+	vgic_set_rdist_busy(vcpu, true);
+	vgic_its_invall(vcpu);
+	vgic_set_rdist_busy(vcpu, false);
+}
+
 /*
  * The GICv3 per-IRQ registers are split to control PPIs and SGIs in the
  * redistributors, while SPIs are covered by registers in the distributor
@@ -630,6 +724,15 @@ static const struct vgic_register_region vgic_v3_rd_registers[] = {
 	REGISTER_DESC_WITH_LENGTH(GICR_PENDBASER,
 		vgic_mmio_read_pendbase, vgic_mmio_write_pendbase, 8,
 		VGIC_ACCESS_64bit | VGIC_ACCESS_32bit),
+	REGISTER_DESC_WITH_LENGTH(GICR_INVLPIR,
+		vgic_mmio_read_raz, vgic_mmio_write_invlpi, 8,
+		VGIC_ACCESS_64bit | VGIC_ACCESS_32bit),
+	REGISTER_DESC_WITH_LENGTH(GICR_INVALLR,
+		vgic_mmio_read_raz, vgic_mmio_write_invall, 8,
+		VGIC_ACCESS_64bit | VGIC_ACCESS_32bit),
+	REGISTER_DESC_WITH_LENGTH(GICR_SYNCR,
+		vgic_mmio_read_sync, vgic_mmio_write_wi, 4,
+		VGIC_ACCESS_32bit),
 	REGISTER_DESC_WITH_LENGTH(GICR_IDREGS,
 		vgic_mmio_read_v3_idregs, vgic_mmio_write_wi, 48,
 		VGIC_ACCESS_32bit),
diff --git a/arch/arm64/kvm/vgic/vgic-v3.c b/arch/arm64/kvm/vgic/vgic-v3.c
index b549af8b1dc2..826ff6f2a4e7 100644
--- a/arch/arm64/kvm/vgic/vgic-v3.c
+++ b/arch/arm64/kvm/vgic/vgic-v3.c
@@ -612,6 +612,10 @@ early_param("kvm-arm.vgic_v4_enable", early_gicv4_enable);
 static const struct midr_range broken_seis[] = {
 	MIDR_ALL_VERSIONS(MIDR_APPLE_M1_ICESTORM),
 	MIDR_ALL_VERSIONS(MIDR_APPLE_M1_FIRESTORM),
+	MIDR_ALL_VERSIONS(MIDR_APPLE_M1_ICESTORM_PRO),
+	MIDR_ALL_VERSIONS(MIDR_APPLE_M1_FIRESTORM_PRO),
+	MIDR_ALL_VERSIONS(MIDR_APPLE_M1_ICESTORM_MAX),
+	MIDR_ALL_VERSIONS(MIDR_APPLE_M1_FIRESTORM_MAX),
 	{},
 };
 
diff --git a/arch/arm64/kvm/vgic/vgic.h b/arch/arm64/kvm/vgic/vgic.h
index 3fd6c86a7ef3..4c6bdd321faa 100644
--- a/arch/arm64/kvm/vgic/vgic.h
+++ b/arch/arm64/kvm/vgic/vgic.h
@@ -98,6 +98,11 @@
 #define DEBUG_SPINLOCK_BUG_ON(p)
 #endif
 
+static inline u32 vgic_get_implementation_rev(struct kvm_vcpu *vcpu)
+{
+	return vcpu->kvm->arch.vgic.implementation_rev;
+}
+
 /* Requires the irq_lock to be held by the caller. */
 static inline bool irq_is_pending(struct vgic_irq *irq)
 {
@@ -308,6 +313,7 @@ static inline bool vgic_dist_overlap(struct kvm *kvm, gpa_t base, size_t size)
 		(base < d->vgic_dist_base + KVM_VGIC_V3_DIST_SIZE);
 }
 
+bool vgic_lpis_enabled(struct kvm_vcpu *vcpu);
 int vgic_copy_lpi_list(struct kvm *kvm, struct kvm_vcpu *vcpu, u32 **intid_ptr);
 int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
 			 u32 devid, u32 eventid, struct vgic_irq **irq);
@@ -317,6 +323,10 @@ void vgic_lpi_translation_cache_init(struct kvm *kvm);
 void vgic_lpi_translation_cache_destroy(struct kvm *kvm);
 void vgic_its_invalidate_cache(struct kvm *kvm);
 
+/* GICv4.1 MMIO interface */
+int vgic_its_inv_lpi(struct kvm *kvm, struct vgic_irq *irq);
+int vgic_its_invall(struct kvm_vcpu *vcpu);
+
 bool vgic_supports_direct_msis(struct kvm *kvm);
 int vgic_v4_init(struct kvm *kvm);
 void vgic_v4_teardown(struct kvm *kvm);
diff --git a/arch/arm64/lib/delay.c b/arch/arm64/lib/delay.c
index 1688af0a4c97..5b7890139bc2 100644
--- a/arch/arm64/lib/delay.c
+++ b/arch/arm64/lib/delay.c
@@ -27,7 +27,17 @@ void __delay(unsigned long cycles)
 {
 	cycles_t start = get_cycles();
 
-	if (arch_timer_evtstrm_available()) {
+	if (cpus_have_const_cap(ARM64_HAS_WFXT)) {
+		u64 end = start + cycles;
+
+		/*
+		 * Start with WFIT. If an interrupt makes us resume
+		 * early, use a WFET loop to complete the delay.
+		 */
+		wfit(end);
+		while ((get_cycles() - start) < cycles)
+			wfet(end);
+	} else 	if (arch_timer_evtstrm_available()) {
 		const cycles_t timer_evt_period =
 			USECS_TO_CYCLES(ARCH_TIMER_EVT_STREAM_PERIOD_US);
 
diff --git a/arch/arm64/tools/cpucaps b/arch/arm64/tools/cpucaps
index e52b289a27c2..507b20373953 100644
--- a/arch/arm64/tools/cpucaps
+++ b/arch/arm64/tools/cpucaps
@@ -38,6 +38,7 @@ HAS_STAGE2_FWB
 HAS_SYSREG_GIC_CPUIF
 HAS_TLB_RANGE
 HAS_VIRT_HOST_EXTN
+HAS_WFXT
 HW_DBM
 KVM_PROTECTED_MODE
 MISMATCHED_CACHE_TYPE
diff --git a/arch/riscv/include/asm/csr.h b/arch/riscv/include/asm/csr.h
index e935f27b10fd..cc40521e438b 100644
--- a/arch/riscv/include/asm/csr.h
+++ b/arch/riscv/include/asm/csr.h
@@ -117,6 +117,7 @@
 #define HGATP_MODE_SV32X4	_AC(1, UL)
 #define HGATP_MODE_SV39X4	_AC(8, UL)
 #define HGATP_MODE_SV48X4	_AC(9, UL)
+#define HGATP_MODE_SV57X4	_AC(10, UL)
 
 #define HGATP32_MODE_SHIFT	31
 #define HGATP32_VMID_SHIFT	22
diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
index cd4bbcecb0fb..319c8aeb42af 100644
--- a/arch/riscv/include/asm/kvm_host.h
+++ b/arch/riscv/include/asm/kvm_host.h
@@ -12,12 +12,12 @@
 #include <linux/types.h>
 #include <linux/kvm.h>
 #include <linux/kvm_types.h>
+#include <linux/spinlock.h>
 #include <asm/csr.h>
 #include <asm/kvm_vcpu_fp.h>
 #include <asm/kvm_vcpu_timer.h>
 
-#define KVM_MAX_VCPUS			\
-	((HGATP_VMID_MASK >> HGATP_VMID_SHIFT) + 1)
+#define KVM_MAX_VCPUS			1024
 
 #define KVM_HALT_POLL_NS_DEFAULT	500000
 
@@ -27,6 +27,31 @@
 	KVM_ARCH_REQ_FLAGS(0, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
 #define KVM_REQ_VCPU_RESET		KVM_ARCH_REQ(1)
 #define KVM_REQ_UPDATE_HGATP		KVM_ARCH_REQ(2)
+#define KVM_REQ_FENCE_I			\
+	KVM_ARCH_REQ_FLAGS(3, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
+#define KVM_REQ_HFENCE_GVMA_VMID_ALL	KVM_REQ_TLB_FLUSH
+#define KVM_REQ_HFENCE_VVMA_ALL		\
+	KVM_ARCH_REQ_FLAGS(4, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
+#define KVM_REQ_HFENCE			\
+	KVM_ARCH_REQ_FLAGS(5, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
+
+enum kvm_riscv_hfence_type {
+	KVM_RISCV_HFENCE_UNKNOWN = 0,
+	KVM_RISCV_HFENCE_GVMA_VMID_GPA,
+	KVM_RISCV_HFENCE_VVMA_ASID_GVA,
+	KVM_RISCV_HFENCE_VVMA_ASID_ALL,
+	KVM_RISCV_HFENCE_VVMA_GVA,
+};
+
+struct kvm_riscv_hfence {
+	enum kvm_riscv_hfence_type type;
+	unsigned long asid;
+	unsigned long order;
+	gpa_t addr;
+	gpa_t size;
+};
+
+#define KVM_RISCV_VCPU_MAX_HFENCE	64
 
 struct kvm_vm_stat {
 	struct kvm_vm_stat_generic generic;
@@ -54,10 +79,10 @@ struct kvm_vmid {
 };
 
 struct kvm_arch {
-	/* stage2 vmid */
+	/* G-stage vmid */
 	struct kvm_vmid vmid;
 
-	/* stage2 page table */
+	/* G-stage page table */
 	pgd_t *pgd;
 	phys_addr_t pgd_phys;
 
@@ -141,6 +166,9 @@ struct kvm_vcpu_arch {
 	/* VCPU ran at least once */
 	bool ran_atleast_once;
 
+	/* Last Host CPU on which Guest VCPU exited */
+	int last_exit_cpu;
+
 	/* ISA feature bits (similar to MISA) */
 	unsigned long isa;
 
@@ -179,6 +207,12 @@ struct kvm_vcpu_arch {
 	/* VCPU Timer */
 	struct kvm_vcpu_timer timer;
 
+	/* HFENCE request queue */
+	spinlock_t hfence_lock;
+	unsigned long hfence_head;
+	unsigned long hfence_tail;
+	struct kvm_riscv_hfence hfence_queue[KVM_RISCV_VCPU_MAX_HFENCE];
+
 	/* MMIO instruction details */
 	struct kvm_mmio_decode mmio_decode;
 
@@ -201,27 +235,71 @@ static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
 
 #define KVM_ARCH_WANT_MMU_NOTIFIER
 
-void __kvm_riscv_hfence_gvma_vmid_gpa(unsigned long gpa_divby_4,
-				      unsigned long vmid);
-void __kvm_riscv_hfence_gvma_vmid(unsigned long vmid);
-void __kvm_riscv_hfence_gvma_gpa(unsigned long gpa_divby_4);
-void __kvm_riscv_hfence_gvma_all(void);
-
-int kvm_riscv_stage2_map(struct kvm_vcpu *vcpu,
+#define KVM_RISCV_GSTAGE_TLB_MIN_ORDER		12
+
+void kvm_riscv_local_hfence_gvma_vmid_gpa(unsigned long vmid,
+					  gpa_t gpa, gpa_t gpsz,
+					  unsigned long order);
+void kvm_riscv_local_hfence_gvma_vmid_all(unsigned long vmid);
+void kvm_riscv_local_hfence_gvma_gpa(gpa_t gpa, gpa_t gpsz,
+				     unsigned long order);
+void kvm_riscv_local_hfence_gvma_all(void);
+void kvm_riscv_local_hfence_vvma_asid_gva(unsigned long vmid,
+					  unsigned long asid,
+					  unsigned long gva,
+					  unsigned long gvsz,
+					  unsigned long order);
+void kvm_riscv_local_hfence_vvma_asid_all(unsigned long vmid,
+					  unsigned long asid);
+void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
+				     unsigned long gva, unsigned long gvsz,
+				     unsigned long order);
+void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
+
+void kvm_riscv_local_tlb_sanitize(struct kvm_vcpu *vcpu);
+
+void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
+void kvm_riscv_hfence_gvma_vmid_all_process(struct kvm_vcpu *vcpu);
+void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
+void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu);
+
+void kvm_riscv_fence_i(struct kvm *kvm,
+		       unsigned long hbase, unsigned long hmask);
+void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask,
+				    gpa_t gpa, gpa_t gpsz,
+				    unsigned long order);
+void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask);
+void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask,
+				    unsigned long gva, unsigned long gvsz,
+				    unsigned long order, unsigned long asid);
+void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask,
+				    unsigned long asid);
+void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
+			       unsigned long hbase, unsigned long hmask,
+			       unsigned long gva, unsigned long gvsz,
+			       unsigned long order);
+void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
+			       unsigned long hbase, unsigned long hmask);
+
+int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
 			 struct kvm_memory_slot *memslot,
 			 gpa_t gpa, unsigned long hva, bool is_write);
-int kvm_riscv_stage2_alloc_pgd(struct kvm *kvm);
-void kvm_riscv_stage2_free_pgd(struct kvm *kvm);
-void kvm_riscv_stage2_update_hgatp(struct kvm_vcpu *vcpu);
-void kvm_riscv_stage2_mode_detect(void);
-unsigned long kvm_riscv_stage2_mode(void);
-int kvm_riscv_stage2_gpa_bits(void);
-
-void kvm_riscv_stage2_vmid_detect(void);
-unsigned long kvm_riscv_stage2_vmid_bits(void);
-int kvm_riscv_stage2_vmid_init(struct kvm *kvm);
-bool kvm_riscv_stage2_vmid_ver_changed(struct kvm_vmid *vmid);
-void kvm_riscv_stage2_vmid_update(struct kvm_vcpu *vcpu);
+int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
+void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
+void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
+void kvm_riscv_gstage_mode_detect(void);
+unsigned long kvm_riscv_gstage_mode(void);
+int kvm_riscv_gstage_gpa_bits(void);
+
+void kvm_riscv_gstage_vmid_detect(void);
+unsigned long kvm_riscv_gstage_vmid_bits(void);
+int kvm_riscv_gstage_vmid_init(struct kvm *kvm);
+bool kvm_riscv_gstage_vmid_ver_changed(struct kvm_vmid *vmid);
+void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu);
 
 void __kvm_riscv_unpriv_trap(void);
 
diff --git a/arch/riscv/include/uapi/asm/kvm.h b/arch/riscv/include/uapi/asm/kvm.h
index f808ad1ce500..6119368ba6d5 100644
--- a/arch/riscv/include/uapi/asm/kvm.h
+++ b/arch/riscv/include/uapi/asm/kvm.h
@@ -82,6 +82,23 @@ struct kvm_riscv_timer {
 	__u64 state;
 };
 
+/*
+ * ISA extension IDs specific to KVM. This is not the same as the host ISA
+ * extension IDs as that is internal to the host and should not be exposed
+ * to the guest. This should always be contiguous to keep the mapping simple
+ * in KVM implementation.
+ */
+enum KVM_RISCV_ISA_EXT_ID {
+	KVM_RISCV_ISA_EXT_A = 0,
+	KVM_RISCV_ISA_EXT_C,
+	KVM_RISCV_ISA_EXT_D,
+	KVM_RISCV_ISA_EXT_F,
+	KVM_RISCV_ISA_EXT_H,
+	KVM_RISCV_ISA_EXT_I,
+	KVM_RISCV_ISA_EXT_M,
+	KVM_RISCV_ISA_EXT_MAX,
+};
+
 /* Possible states for kvm_riscv_timer */
 #define KVM_RISCV_TIMER_STATE_OFF	0
 #define KVM_RISCV_TIMER_STATE_ON	1
@@ -123,6 +140,9 @@ struct kvm_riscv_timer {
 #define KVM_REG_RISCV_FP_D_REG(name)	\
 		(offsetof(struct __riscv_d_ext_state, name) / sizeof(__u64))
 
+/* ISA Extension registers are mapped as type 7 */
+#define KVM_REG_RISCV_ISA_EXT		(0x07 << KVM_REG_RISCV_TYPE_SHIFT)
+
 #endif
 
 #endif /* __LINUX_KVM_RISCV_H */
diff --git a/arch/riscv/kvm/main.c b/arch/riscv/kvm/main.c
index 2e5ca43c8c49..1549205fe5fe 100644
--- a/arch/riscv/kvm/main.c
+++ b/arch/riscv/kvm/main.c
@@ -89,13 +89,13 @@ int kvm_arch_init(void *opaque)
 		return -ENODEV;
 	}
 
-	kvm_riscv_stage2_mode_detect();
+	kvm_riscv_gstage_mode_detect();
 
-	kvm_riscv_stage2_vmid_detect();
+	kvm_riscv_gstage_vmid_detect();
 
 	kvm_info("hypervisor extension available\n");
 
-	switch (kvm_riscv_stage2_mode()) {
+	switch (kvm_riscv_gstage_mode()) {
 	case HGATP_MODE_SV32X4:
 		str = "Sv32x4";
 		break;
@@ -105,12 +105,15 @@ int kvm_arch_init(void *opaque)
 	case HGATP_MODE_SV48X4:
 		str = "Sv48x4";
 		break;
+	case HGATP_MODE_SV57X4:
+		str = "Sv57x4";
+		break;
 	default:
 		return -ENODEV;
 	}
 	kvm_info("using %s G-stage page table format\n", str);
 
-	kvm_info("VMID %ld bits available\n", kvm_riscv_stage2_vmid_bits());
+	kvm_info("VMID %ld bits available\n", kvm_riscv_gstage_vmid_bits());
 
 	return 0;
 }
diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
index f80a34fbf102..1c00695ebee7 100644
--- a/arch/riscv/kvm/mmu.c
+++ b/arch/riscv/kvm/mmu.c
@@ -18,53 +18,52 @@
 #include <asm/csr.h>
 #include <asm/page.h>
 #include <asm/pgtable.h>
-#include <asm/sbi.h>
 
 #ifdef CONFIG_64BIT
-static unsigned long stage2_mode = (HGATP_MODE_SV39X4 << HGATP_MODE_SHIFT);
-static unsigned long stage2_pgd_levels = 3;
-#define stage2_index_bits	9
+static unsigned long gstage_mode = (HGATP_MODE_SV39X4 << HGATP_MODE_SHIFT);
+static unsigned long gstage_pgd_levels = 3;
+#define gstage_index_bits	9
 #else
-static unsigned long stage2_mode = (HGATP_MODE_SV32X4 << HGATP_MODE_SHIFT);
-static unsigned long stage2_pgd_levels = 2;
-#define stage2_index_bits	10
+static unsigned long gstage_mode = (HGATP_MODE_SV32X4 << HGATP_MODE_SHIFT);
+static unsigned long gstage_pgd_levels = 2;
+#define gstage_index_bits	10
 #endif
 
-#define stage2_pgd_xbits	2
-#define stage2_pgd_size	(1UL << (HGATP_PAGE_SHIFT + stage2_pgd_xbits))
-#define stage2_gpa_bits	(HGATP_PAGE_SHIFT + \
-			 (stage2_pgd_levels * stage2_index_bits) + \
-			 stage2_pgd_xbits)
-#define stage2_gpa_size	((gpa_t)(1ULL << stage2_gpa_bits))
+#define gstage_pgd_xbits	2
+#define gstage_pgd_size	(1UL << (HGATP_PAGE_SHIFT + gstage_pgd_xbits))
+#define gstage_gpa_bits	(HGATP_PAGE_SHIFT + \
+			 (gstage_pgd_levels * gstage_index_bits) + \
+			 gstage_pgd_xbits)
+#define gstage_gpa_size	((gpa_t)(1ULL << gstage_gpa_bits))
 
-#define stage2_pte_leaf(__ptep)	\
+#define gstage_pte_leaf(__ptep)	\
 	(pte_val(*(__ptep)) & (_PAGE_READ | _PAGE_WRITE | _PAGE_EXEC))
 
-static inline unsigned long stage2_pte_index(gpa_t addr, u32 level)
+static inline unsigned long gstage_pte_index(gpa_t addr, u32 level)
 {
 	unsigned long mask;
-	unsigned long shift = HGATP_PAGE_SHIFT + (stage2_index_bits * level);
+	unsigned long shift = HGATP_PAGE_SHIFT + (gstage_index_bits * level);
 
-	if (level == (stage2_pgd_levels - 1))
-		mask = (PTRS_PER_PTE * (1UL << stage2_pgd_xbits)) - 1;
+	if (level == (gstage_pgd_levels - 1))
+		mask = (PTRS_PER_PTE * (1UL << gstage_pgd_xbits)) - 1;
 	else
 		mask = PTRS_PER_PTE - 1;
 
 	return (addr >> shift) & mask;
 }
 
-static inline unsigned long stage2_pte_page_vaddr(pte_t pte)
+static inline unsigned long gstage_pte_page_vaddr(pte_t pte)
 {
 	return (unsigned long)pfn_to_virt(pte_val(pte) >> _PAGE_PFN_SHIFT);
 }
 
-static int stage2_page_size_to_level(unsigned long page_size, u32 *out_level)
+static int gstage_page_size_to_level(unsigned long page_size, u32 *out_level)
 {
 	u32 i;
 	unsigned long psz = 1UL << 12;
 
-	for (i = 0; i < stage2_pgd_levels; i++) {
-		if (page_size == (psz << (i * stage2_index_bits))) {
+	for (i = 0; i < gstage_pgd_levels; i++) {
+		if (page_size == (psz << (i * gstage_index_bits))) {
 			*out_level = i;
 			return 0;
 		}
@@ -73,27 +72,39 @@ static int stage2_page_size_to_level(unsigned long page_size, u32 *out_level)
 	return -EINVAL;
 }
 
-static int stage2_level_to_page_size(u32 level, unsigned long *out_pgsize)
+static int gstage_level_to_page_order(u32 level, unsigned long *out_pgorder)
 {
-	if (stage2_pgd_levels < level)
+	if (gstage_pgd_levels < level)
 		return -EINVAL;
 
-	*out_pgsize = 1UL << (12 + (level * stage2_index_bits));
+	*out_pgorder = 12 + (level * gstage_index_bits);
+	return 0;
+}
+
+static int gstage_level_to_page_size(u32 level, unsigned long *out_pgsize)
+{
+	int rc;
+	unsigned long page_order = PAGE_SHIFT;
+
+	rc = gstage_level_to_page_order(level, &page_order);
+	if (rc)
+		return rc;
 
+	*out_pgsize = BIT(page_order);
 	return 0;
 }
 
-static bool stage2_get_leaf_entry(struct kvm *kvm, gpa_t addr,
+static bool gstage_get_leaf_entry(struct kvm *kvm, gpa_t addr,
 				  pte_t **ptepp, u32 *ptep_level)
 {
 	pte_t *ptep;
-	u32 current_level = stage2_pgd_levels - 1;
+	u32 current_level = gstage_pgd_levels - 1;
 
 	*ptep_level = current_level;
 	ptep = (pte_t *)kvm->arch.pgd;
-	ptep = &ptep[stage2_pte_index(addr, current_level)];
+	ptep = &ptep[gstage_pte_index(addr, current_level)];
 	while (ptep && pte_val(*ptep)) {
-		if (stage2_pte_leaf(ptep)) {
+		if (gstage_pte_leaf(ptep)) {
 			*ptep_level = current_level;
 			*ptepp = ptep;
 			return true;
@@ -102,8 +113,8 @@ static bool stage2_get_leaf_entry(struct kvm *kvm, gpa_t addr,
 		if (current_level) {
 			current_level--;
 			*ptep_level = current_level;
-			ptep = (pte_t *)stage2_pte_page_vaddr(*ptep);
-			ptep = &ptep[stage2_pte_index(addr, current_level)];
+			ptep = (pte_t *)gstage_pte_page_vaddr(*ptep);
+			ptep = &ptep[gstage_pte_index(addr, current_level)];
 		} else {
 			ptep = NULL;
 		}
@@ -112,38 +123,30 @@ static bool stage2_get_leaf_entry(struct kvm *kvm, gpa_t addr,
 	return false;
 }
 
-static void stage2_remote_tlb_flush(struct kvm *kvm, u32 level, gpa_t addr)
+static void gstage_remote_tlb_flush(struct kvm *kvm, u32 level, gpa_t addr)
 {
-	unsigned long size = PAGE_SIZE;
-	struct kvm_vmid *vmid = &kvm->arch.vmid;
+	unsigned long order = PAGE_SHIFT;
 
-	if (stage2_level_to_page_size(level, &size))
+	if (gstage_level_to_page_order(level, &order))
 		return;
-	addr &= ~(size - 1);
+	addr &= ~(BIT(order) - 1);
 
-	/*
-	 * TODO: Instead of cpu_online_mask, we should only target CPUs
-	 * where the Guest/VM is running.
-	 */
-	preempt_disable();
-	sbi_remote_hfence_gvma_vmid(cpu_online_mask, addr, size,
-				    READ_ONCE(vmid->vmid));
-	preempt_enable();
+	kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0, addr, BIT(order), order);
 }
 
-static int stage2_set_pte(struct kvm *kvm, u32 level,
+static int gstage_set_pte(struct kvm *kvm, u32 level,
 			   struct kvm_mmu_memory_cache *pcache,
 			   gpa_t addr, const pte_t *new_pte)
 {
-	u32 current_level = stage2_pgd_levels - 1;
+	u32 current_level = gstage_pgd_levels - 1;
 	pte_t *next_ptep = (pte_t *)kvm->arch.pgd;
-	pte_t *ptep = &next_ptep[stage2_pte_index(addr, current_level)];
+	pte_t *ptep = &next_ptep[gstage_pte_index(addr, current_level)];
 
 	if (current_level < level)
 		return -EINVAL;
 
 	while (current_level != level) {
-		if (stage2_pte_leaf(ptep))
+		if (gstage_pte_leaf(ptep))
 			return -EEXIST;
 
 		if (!pte_val(*ptep)) {
@@ -155,23 +158,23 @@ static int stage2_set_pte(struct kvm *kvm, u32 level,
 			*ptep = pfn_pte(PFN_DOWN(__pa(next_ptep)),
 					__pgprot(_PAGE_TABLE));
 		} else {
-			if (stage2_pte_leaf(ptep))
+			if (gstage_pte_leaf(ptep))
 				return -EEXIST;
-			next_ptep = (pte_t *)stage2_pte_page_vaddr(*ptep);
+			next_ptep = (pte_t *)gstage_pte_page_vaddr(*ptep);
 		}
 
 		current_level--;
-		ptep = &next_ptep[stage2_pte_index(addr, current_level)];
+		ptep = &next_ptep[gstage_pte_index(addr, current_level)];
 	}
 
 	*ptep = *new_pte;
-	if (stage2_pte_leaf(ptep))
-		stage2_remote_tlb_flush(kvm, current_level, addr);
+	if (gstage_pte_leaf(ptep))
+		gstage_remote_tlb_flush(kvm, current_level, addr);
 
 	return 0;
 }
 
-static int stage2_map_page(struct kvm *kvm,
+static int gstage_map_page(struct kvm *kvm,
 			   struct kvm_mmu_memory_cache *pcache,
 			   gpa_t gpa, phys_addr_t hpa,
 			   unsigned long page_size,
@@ -182,7 +185,7 @@ static int stage2_map_page(struct kvm *kvm,
 	pte_t new_pte;
 	pgprot_t prot;
 
-	ret = stage2_page_size_to_level(page_size, &level);
+	ret = gstage_page_size_to_level(page_size, &level);
 	if (ret)
 		return ret;
 
@@ -193,9 +196,9 @@ static int stage2_map_page(struct kvm *kvm,
 	 *    PTE so that software can update these bits.
 	 *
 	 * We support both options mentioned above. To achieve this, we
-	 * always set 'A' and 'D' PTE bits at time of creating stage2
+	 * always set 'A' and 'D' PTE bits at time of creating G-stage
 	 * mapping. To support KVM dirty page logging with both options
-	 * mentioned above, we will write-protect stage2 PTEs to track
+	 * mentioned above, we will write-protect G-stage PTEs to track
 	 * dirty pages.
 	 */
 
@@ -213,24 +216,24 @@ static int stage2_map_page(struct kvm *kvm,
 	new_pte = pfn_pte(PFN_DOWN(hpa), prot);
 	new_pte = pte_mkdirty(new_pte);
 
-	return stage2_set_pte(kvm, level, pcache, gpa, &new_pte);
+	return gstage_set_pte(kvm, level, pcache, gpa, &new_pte);
 }
 
-enum stage2_op {
-	STAGE2_OP_NOP = 0,	/* Nothing */
-	STAGE2_OP_CLEAR,	/* Clear/Unmap */
-	STAGE2_OP_WP,		/* Write-protect */
+enum gstage_op {
+	GSTAGE_OP_NOP = 0,	/* Nothing */
+	GSTAGE_OP_CLEAR,	/* Clear/Unmap */
+	GSTAGE_OP_WP,		/* Write-protect */
 };
 
-static void stage2_op_pte(struct kvm *kvm, gpa_t addr,
-			  pte_t *ptep, u32 ptep_level, enum stage2_op op)
+static void gstage_op_pte(struct kvm *kvm, gpa_t addr,
+			  pte_t *ptep, u32 ptep_level, enum gstage_op op)
 {
 	int i, ret;
 	pte_t *next_ptep;
 	u32 next_ptep_level;
 	unsigned long next_page_size, page_size;
 
-	ret = stage2_level_to_page_size(ptep_level, &page_size);
+	ret = gstage_level_to_page_size(ptep_level, &page_size);
 	if (ret)
 		return;
 
@@ -239,31 +242,31 @@ static void stage2_op_pte(struct kvm *kvm, gpa_t addr,
 	if (!pte_val(*ptep))
 		return;
 
-	if (ptep_level && !stage2_pte_leaf(ptep)) {
-		next_ptep = (pte_t *)stage2_pte_page_vaddr(*ptep);
+	if (ptep_level && !gstage_pte_leaf(ptep)) {
+		next_ptep = (pte_t *)gstage_pte_page_vaddr(*ptep);
 		next_ptep_level = ptep_level - 1;
-		ret = stage2_level_to_page_size(next_ptep_level,
+		ret = gstage_level_to_page_size(next_ptep_level,
 						&next_page_size);
 		if (ret)
 			return;
 
-		if (op == STAGE2_OP_CLEAR)
+		if (op == GSTAGE_OP_CLEAR)
 			set_pte(ptep, __pte(0));
 		for (i = 0; i < PTRS_PER_PTE; i++)
-			stage2_op_pte(kvm, addr + i * next_page_size,
+			gstage_op_pte(kvm, addr + i * next_page_size,
 					&next_ptep[i], next_ptep_level, op);
-		if (op == STAGE2_OP_CLEAR)
+		if (op == GSTAGE_OP_CLEAR)
 			put_page(virt_to_page(next_ptep));
 	} else {
-		if (op == STAGE2_OP_CLEAR)
+		if (op == GSTAGE_OP_CLEAR)
 			set_pte(ptep, __pte(0));
-		else if (op == STAGE2_OP_WP)
+		else if (op == GSTAGE_OP_WP)
 			set_pte(ptep, __pte(pte_val(*ptep) & ~_PAGE_WRITE));
-		stage2_remote_tlb_flush(kvm, ptep_level, addr);
+		gstage_remote_tlb_flush(kvm, ptep_level, addr);
 	}
 }
 
-static void stage2_unmap_range(struct kvm *kvm, gpa_t start,
+static void gstage_unmap_range(struct kvm *kvm, gpa_t start,
 			       gpa_t size, bool may_block)
 {
 	int ret;
@@ -274,9 +277,9 @@ static void stage2_unmap_range(struct kvm *kvm, gpa_t start,
 	gpa_t addr = start, end = start + size;
 
 	while (addr < end) {
-		found_leaf = stage2_get_leaf_entry(kvm, addr,
+		found_leaf = gstage_get_leaf_entry(kvm, addr,
 						   &ptep, &ptep_level);
-		ret = stage2_level_to_page_size(ptep_level, &page_size);
+		ret = gstage_level_to_page_size(ptep_level, &page_size);
 		if (ret)
 			break;
 
@@ -284,8 +287,8 @@ static void stage2_unmap_range(struct kvm *kvm, gpa_t start,
 			goto next;
 
 		if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
-			stage2_op_pte(kvm, addr, ptep,
-				      ptep_level, STAGE2_OP_CLEAR);
+			gstage_op_pte(kvm, addr, ptep,
+				      ptep_level, GSTAGE_OP_CLEAR);
 
 next:
 		addr += page_size;
@@ -299,7 +302,7 @@ next:
 	}
 }
 
-static void stage2_wp_range(struct kvm *kvm, gpa_t start, gpa_t end)
+static void gstage_wp_range(struct kvm *kvm, gpa_t start, gpa_t end)
 {
 	int ret;
 	pte_t *ptep;
@@ -309,9 +312,9 @@ static void stage2_wp_range(struct kvm *kvm, gpa_t start, gpa_t end)
 	unsigned long page_size;
 
 	while (addr < end) {
-		found_leaf = stage2_get_leaf_entry(kvm, addr,
+		found_leaf = gstage_get_leaf_entry(kvm, addr,
 						   &ptep, &ptep_level);
-		ret = stage2_level_to_page_size(ptep_level, &page_size);
+		ret = gstage_level_to_page_size(ptep_level, &page_size);
 		if (ret)
 			break;
 
@@ -319,15 +322,15 @@ static void stage2_wp_range(struct kvm *kvm, gpa_t start, gpa_t end)
 			goto next;
 
 		if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
-			stage2_op_pte(kvm, addr, ptep,
-				      ptep_level, STAGE2_OP_WP);
+			gstage_op_pte(kvm, addr, ptep,
+				      ptep_level, GSTAGE_OP_WP);
 
 next:
 		addr += page_size;
 	}
 }
 
-static void stage2_wp_memory_region(struct kvm *kvm, int slot)
+static void gstage_wp_memory_region(struct kvm *kvm, int slot)
 {
 	struct kvm_memslots *slots = kvm_memslots(kvm);
 	struct kvm_memory_slot *memslot = id_to_memslot(slots, slot);
@@ -335,12 +338,12 @@ static void stage2_wp_memory_region(struct kvm *kvm, int slot)
 	phys_addr_t end = (memslot->base_gfn + memslot->npages) << PAGE_SHIFT;
 
 	spin_lock(&kvm->mmu_lock);
-	stage2_wp_range(kvm, start, end);
+	gstage_wp_range(kvm, start, end);
 	spin_unlock(&kvm->mmu_lock);
 	kvm_flush_remote_tlbs(kvm);
 }
 
-static int stage2_ioremap(struct kvm *kvm, gpa_t gpa, phys_addr_t hpa,
+static int gstage_ioremap(struct kvm *kvm, gpa_t gpa, phys_addr_t hpa,
 			  unsigned long size, bool writable)
 {
 	pte_t pte;
@@ -361,12 +364,12 @@ static int stage2_ioremap(struct kvm *kvm, gpa_t gpa, phys_addr_t hpa,
 		if (!writable)
 			pte = pte_wrprotect(pte);
 
-		ret = kvm_mmu_topup_memory_cache(&pcache, stage2_pgd_levels);
+		ret = kvm_mmu_topup_memory_cache(&pcache, gstage_pgd_levels);
 		if (ret)
 			goto out;
 
 		spin_lock(&kvm->mmu_lock);
-		ret = stage2_set_pte(kvm, 0, &pcache, addr, &pte);
+		ret = gstage_set_pte(kvm, 0, &pcache, addr, &pte);
 		spin_unlock(&kvm->mmu_lock);
 		if (ret)
 			goto out;
@@ -388,7 +391,7 @@ void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
 	phys_addr_t start = (base_gfn +  __ffs(mask)) << PAGE_SHIFT;
 	phys_addr_t end = (base_gfn + __fls(mask) + 1) << PAGE_SHIFT;
 
-	stage2_wp_range(kvm, start, end);
+	gstage_wp_range(kvm, start, end);
 }
 
 void kvm_arch_sync_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot)
@@ -411,7 +414,7 @@ void kvm_arch_memslots_updated(struct kvm *kvm, u64 gen)
 
 void kvm_arch_flush_shadow_all(struct kvm *kvm)
 {
-	kvm_riscv_stage2_free_pgd(kvm);
+	kvm_riscv_gstage_free_pgd(kvm);
 }
 
 void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
@@ -421,7 +424,7 @@ void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
 	phys_addr_t size = slot->npages << PAGE_SHIFT;
 
 	spin_lock(&kvm->mmu_lock);
-	stage2_unmap_range(kvm, gpa, size, false);
+	gstage_unmap_range(kvm, gpa, size, false);
 	spin_unlock(&kvm->mmu_lock);
 }
 
@@ -436,7 +439,7 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
 	 * the memory slot is write protected.
 	 */
 	if (change != KVM_MR_DELETE && new->flags & KVM_MEM_LOG_DIRTY_PAGES)
-		stage2_wp_memory_region(kvm, new->id);
+		gstage_wp_memory_region(kvm, new->id);
 }
 
 int kvm_arch_prepare_memory_region(struct kvm *kvm,
@@ -458,7 +461,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
 	 * space addressable by the KVM guest GPA space.
 	 */
 	if ((new->base_gfn + new->npages) >=
-	    (stage2_gpa_size >> PAGE_SHIFT))
+	    (gstage_gpa_size >> PAGE_SHIFT))
 		return -EFAULT;
 
 	hva = new->userspace_addr;
@@ -514,7 +517,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
 				goto out;
 			}
 
-			ret = stage2_ioremap(kvm, gpa, pa,
+			ret = gstage_ioremap(kvm, gpa, pa,
 					     vm_end - vm_start, writable);
 			if (ret)
 				break;
@@ -527,7 +530,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
 
 	spin_lock(&kvm->mmu_lock);
 	if (ret)
-		stage2_unmap_range(kvm, base_gpa, size, false);
+		gstage_unmap_range(kvm, base_gpa, size, false);
 	spin_unlock(&kvm->mmu_lock);
 
 out:
@@ -540,7 +543,7 @@ bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
 	if (!kvm->arch.pgd)
 		return false;
 
-	stage2_unmap_range(kvm, range->start << PAGE_SHIFT,
+	gstage_unmap_range(kvm, range->start << PAGE_SHIFT,
 			   (range->end - range->start) << PAGE_SHIFT,
 			   range->may_block);
 	return false;
@@ -556,10 +559,10 @@ bool kvm_set_spte_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
 
 	WARN_ON(range->end - range->start != 1);
 
-	ret = stage2_map_page(kvm, NULL, range->start << PAGE_SHIFT,
+	ret = gstage_map_page(kvm, NULL, range->start << PAGE_SHIFT,
 			      __pfn_to_phys(pfn), PAGE_SIZE, true, true);
 	if (ret) {
-		kvm_debug("Failed to map stage2 page (error %d)\n", ret);
+		kvm_debug("Failed to map G-stage page (error %d)\n", ret);
 		return true;
 	}
 
@@ -577,7 +580,7 @@ bool kvm_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
 
 	WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PGDIR_SIZE);
 
-	if (!stage2_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
+	if (!gstage_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
 				   &ptep, &ptep_level))
 		return false;
 
@@ -595,14 +598,14 @@ bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
 
 	WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PGDIR_SIZE);
 
-	if (!stage2_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
+	if (!gstage_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
 				   &ptep, &ptep_level))
 		return false;
 
 	return pte_young(*ptep);
 }
 
-int kvm_riscv_stage2_map(struct kvm_vcpu *vcpu,
+int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
 			 struct kvm_memory_slot *memslot,
 			 gpa_t gpa, unsigned long hva, bool is_write)
 {
@@ -648,9 +651,9 @@ int kvm_riscv_stage2_map(struct kvm_vcpu *vcpu,
 	}
 
 	/* We need minimum second+third level pages */
-	ret = kvm_mmu_topup_memory_cache(pcache, stage2_pgd_levels);
+	ret = kvm_mmu_topup_memory_cache(pcache, gstage_pgd_levels);
 	if (ret) {
-		kvm_err("Failed to topup stage2 cache\n");
+		kvm_err("Failed to topup G-stage cache\n");
 		return ret;
 	}
 
@@ -680,15 +683,15 @@ int kvm_riscv_stage2_map(struct kvm_vcpu *vcpu,
 	if (writeable) {
 		kvm_set_pfn_dirty(hfn);
 		mark_page_dirty(kvm, gfn);
-		ret = stage2_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
+		ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
 				      vma_pagesize, false, true);
 	} else {
-		ret = stage2_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
+		ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
 				      vma_pagesize, true, true);
 	}
 
 	if (ret)
-		kvm_err("Failed to map in stage2\n");
+		kvm_err("Failed to map in G-stage\n");
 
 out_unlock:
 	spin_unlock(&kvm->mmu_lock);
@@ -697,7 +700,7 @@ out_unlock:
 	return ret;
 }
 
-int kvm_riscv_stage2_alloc_pgd(struct kvm *kvm)
+int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm)
 {
 	struct page *pgd_page;
 
@@ -707,7 +710,7 @@ int kvm_riscv_stage2_alloc_pgd(struct kvm *kvm)
 	}
 
 	pgd_page = alloc_pages(GFP_KERNEL | __GFP_ZERO,
-				get_order(stage2_pgd_size));
+				get_order(gstage_pgd_size));
 	if (!pgd_page)
 		return -ENOMEM;
 	kvm->arch.pgd = page_to_virt(pgd_page);
@@ -716,13 +719,13 @@ int kvm_riscv_stage2_alloc_pgd(struct kvm *kvm)
 	return 0;
 }
 
-void kvm_riscv_stage2_free_pgd(struct kvm *kvm)
+void kvm_riscv_gstage_free_pgd(struct kvm *kvm)
 {
 	void *pgd = NULL;
 
 	spin_lock(&kvm->mmu_lock);
 	if (kvm->arch.pgd) {
-		stage2_unmap_range(kvm, 0UL, stage2_gpa_size, false);
+		gstage_unmap_range(kvm, 0UL, gstage_gpa_size, false);
 		pgd = READ_ONCE(kvm->arch.pgd);
 		kvm->arch.pgd = NULL;
 		kvm->arch.pgd_phys = 0;
@@ -730,12 +733,12 @@ void kvm_riscv_stage2_free_pgd(struct kvm *kvm)
 	spin_unlock(&kvm->mmu_lock);
 
 	if (pgd)
-		free_pages((unsigned long)pgd, get_order(stage2_pgd_size));
+		free_pages((unsigned long)pgd, get_order(gstage_pgd_size));
 }
 
-void kvm_riscv_stage2_update_hgatp(struct kvm_vcpu *vcpu)
+void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu)
 {
-	unsigned long hgatp = stage2_mode;
+	unsigned long hgatp = gstage_mode;
 	struct kvm_arch *k = &vcpu->kvm->arch;
 
 	hgatp |= (READ_ONCE(k->vmid.vmid) << HGATP_VMID_SHIFT) &
@@ -744,31 +747,40 @@ void kvm_riscv_stage2_update_hgatp(struct kvm_vcpu *vcpu)
 
 	csr_write(CSR_HGATP, hgatp);
 
-	if (!kvm_riscv_stage2_vmid_bits())
-		__kvm_riscv_hfence_gvma_all();
+	if (!kvm_riscv_gstage_vmid_bits())
+		kvm_riscv_local_hfence_gvma_all();
 }
 
-void kvm_riscv_stage2_mode_detect(void)
+void kvm_riscv_gstage_mode_detect(void)
 {
 #ifdef CONFIG_64BIT
-	/* Try Sv48x4 stage2 mode */
+	/* Try Sv57x4 G-stage mode */
+	csr_write(CSR_HGATP, HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
+	if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV57X4) {
+		gstage_mode = (HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
+		gstage_pgd_levels = 5;
+		goto skip_sv48x4_test;
+	}
+
+	/* Try Sv48x4 G-stage mode */
 	csr_write(CSR_HGATP, HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
 	if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV48X4) {
-		stage2_mode = (HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
-		stage2_pgd_levels = 4;
+		gstage_mode = (HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
+		gstage_pgd_levels = 4;
 	}
-	csr_write(CSR_HGATP, 0);
+skip_sv48x4_test:
 
-	__kvm_riscv_hfence_gvma_all();
+	csr_write(CSR_HGATP, 0);
+	kvm_riscv_local_hfence_gvma_all();
 #endif
 }
 
-unsigned long kvm_riscv_stage2_mode(void)
+unsigned long kvm_riscv_gstage_mode(void)
 {
-	return stage2_mode >> HGATP_MODE_SHIFT;
+	return gstage_mode >> HGATP_MODE_SHIFT;
 }
 
-int kvm_riscv_stage2_gpa_bits(void)
+int kvm_riscv_gstage_gpa_bits(void)
 {
-	return stage2_gpa_bits;
+	return gstage_gpa_bits;
 }
diff --git a/arch/riscv/kvm/tlb.S b/arch/riscv/kvm/tlb.S
deleted file mode 100644
index 899f75d60bad..000000000000
--- a/arch/riscv/kvm/tlb.S
+++ /dev/null
@@ -1,74 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-/*
- * Copyright (C) 2019 Western Digital Corporation or its affiliates.
- *
- * Authors:
- *     Anup Patel <anup.patel@wdc.com>
- */
-
-#include <linux/linkage.h>
-#include <asm/asm.h>
-
-	.text
-	.altmacro
-	.option norelax
-
-	/*
-	 * Instruction encoding of hfence.gvma is:
-	 * HFENCE.GVMA rs1, rs2
-	 * HFENCE.GVMA zero, rs2
-	 * HFENCE.GVMA rs1
-	 * HFENCE.GVMA
-	 *
-	 * rs1!=zero and rs2!=zero ==> HFENCE.GVMA rs1, rs2
-	 * rs1==zero and rs2!=zero ==> HFENCE.GVMA zero, rs2
-	 * rs1!=zero and rs2==zero ==> HFENCE.GVMA rs1
-	 * rs1==zero and rs2==zero ==> HFENCE.GVMA
-	 *
-	 * Instruction encoding of HFENCE.GVMA is:
-	 * 0110001 rs2(5) rs1(5) 000 00000 1110011
-	 */
-
-ENTRY(__kvm_riscv_hfence_gvma_vmid_gpa)
-	/*
-	 * rs1 = a0 (GPA >> 2)
-	 * rs2 = a1 (VMID)
-	 * HFENCE.GVMA a0, a1
-	 * 0110001 01011 01010 000 00000 1110011
-	 */
-	.word 0x62b50073
-	ret
-ENDPROC(__kvm_riscv_hfence_gvma_vmid_gpa)
-
-ENTRY(__kvm_riscv_hfence_gvma_vmid)
-	/*
-	 * rs1 = zero
-	 * rs2 = a0 (VMID)
-	 * HFENCE.GVMA zero, a0
-	 * 0110001 01010 00000 000 00000 1110011
-	 */
-	.word 0x62a00073
-	ret
-ENDPROC(__kvm_riscv_hfence_gvma_vmid)
-
-ENTRY(__kvm_riscv_hfence_gvma_gpa)
-	/*
-	 * rs1 = a0 (GPA >> 2)
-	 * rs2 = zero
-	 * HFENCE.GVMA a0
-	 * 0110001 00000 01010 000 00000 1110011
-	 */
-	.word 0x62050073
-	ret
-ENDPROC(__kvm_riscv_hfence_gvma_gpa)
-
-ENTRY(__kvm_riscv_hfence_gvma_all)
-	/*
-	 * rs1 = zero
-	 * rs2 = zero
-	 * HFENCE.GVMA
-	 * 0110001 00000 00000 000 00000 1110011
-	 */
-	.word 0x62000073
-	ret
-ENDPROC(__kvm_riscv_hfence_gvma_all)
diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
new file mode 100644
index 000000000000..1a76d0b1907d
--- /dev/null
+++ b/arch/riscv/kvm/tlb.c
@@ -0,0 +1,461 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2022 Ventana Micro Systems Inc.
+ */
+
+#include <linux/bitmap.h>
+#include <linux/cpumask.h>
+#include <linux/errno.h>
+#include <linux/err.h>
+#include <linux/module.h>
+#include <linux/smp.h>
+#include <linux/kvm_host.h>
+#include <asm/cacheflush.h>
+#include <asm/csr.h>
+
+/*
+ * Instruction encoding of hfence.gvma is:
+ * HFENCE.GVMA rs1, rs2
+ * HFENCE.GVMA zero, rs2
+ * HFENCE.GVMA rs1
+ * HFENCE.GVMA
+ *
+ * rs1!=zero and rs2!=zero ==> HFENCE.GVMA rs1, rs2
+ * rs1==zero and rs2!=zero ==> HFENCE.GVMA zero, rs2
+ * rs1!=zero and rs2==zero ==> HFENCE.GVMA rs1
+ * rs1==zero and rs2==zero ==> HFENCE.GVMA
+ *
+ * Instruction encoding of HFENCE.GVMA is:
+ * 0110001 rs2(5) rs1(5) 000 00000 1110011
+ */
+
+void kvm_riscv_local_hfence_gvma_vmid_gpa(unsigned long vmid,
+					  gpa_t gpa, gpa_t gpsz,
+					  unsigned long order)
+{
+	gpa_t pos;
+
+	if (PTRS_PER_PTE < (gpsz >> order)) {
+		kvm_riscv_local_hfence_gvma_vmid_all(vmid);
+		return;
+	}
+
+	for (pos = gpa; pos < (gpa + gpsz); pos += BIT(order)) {
+		/*
+		 * rs1 = a0 (GPA >> 2)
+		 * rs2 = a1 (VMID)
+		 * HFENCE.GVMA a0, a1
+		 * 0110001 01011 01010 000 00000 1110011
+		 */
+		asm volatile ("srli a0, %0, 2\n"
+			      "add a1, %1, zero\n"
+			      ".word 0x62b50073\n"
+			      :: "r" (pos), "r" (vmid)
+			      : "a0", "a1", "memory");
+	}
+}
+
+void kvm_riscv_local_hfence_gvma_vmid_all(unsigned long vmid)
+{
+	/*
+	 * rs1 = zero
+	 * rs2 = a0 (VMID)
+	 * HFENCE.GVMA zero, a0
+	 * 0110001 01010 00000 000 00000 1110011
+	 */
+	asm volatile ("add a0, %0, zero\n"
+		      ".word 0x62a00073\n"
+		      :: "r" (vmid) : "a0", "memory");
+}
+
+void kvm_riscv_local_hfence_gvma_gpa(gpa_t gpa, gpa_t gpsz,
+				     unsigned long order)
+{
+	gpa_t pos;
+
+	if (PTRS_PER_PTE < (gpsz >> order)) {
+		kvm_riscv_local_hfence_gvma_all();
+		return;
+	}
+
+	for (pos = gpa; pos < (gpa + gpsz); pos += BIT(order)) {
+		/*
+		 * rs1 = a0 (GPA >> 2)
+		 * rs2 = zero
+		 * HFENCE.GVMA a0
+		 * 0110001 00000 01010 000 00000 1110011
+		 */
+		asm volatile ("srli a0, %0, 2\n"
+			      ".word 0x62050073\n"
+			      :: "r" (pos) : "a0", "memory");
+	}
+}
+
+void kvm_riscv_local_hfence_gvma_all(void)
+{
+	/*
+	 * rs1 = zero
+	 * rs2 = zero
+	 * HFENCE.GVMA
+	 * 0110001 00000 00000 000 00000 1110011
+	 */
+	asm volatile (".word 0x62000073" ::: "memory");
+}
+
+/*
+ * Instruction encoding of hfence.gvma is:
+ * HFENCE.VVMA rs1, rs2
+ * HFENCE.VVMA zero, rs2
+ * HFENCE.VVMA rs1
+ * HFENCE.VVMA
+ *
+ * rs1!=zero and rs2!=zero ==> HFENCE.VVMA rs1, rs2
+ * rs1==zero and rs2!=zero ==> HFENCE.VVMA zero, rs2
+ * rs1!=zero and rs2==zero ==> HFENCE.VVMA rs1
+ * rs1==zero and rs2==zero ==> HFENCE.VVMA
+ *
+ * Instruction encoding of HFENCE.VVMA is:
+ * 0010001 rs2(5) rs1(5) 000 00000 1110011
+ */
+
+void kvm_riscv_local_hfence_vvma_asid_gva(unsigned long vmid,
+					  unsigned long asid,
+					  unsigned long gva,
+					  unsigned long gvsz,
+					  unsigned long order)
+{
+	unsigned long pos, hgatp;
+
+	if (PTRS_PER_PTE < (gvsz >> order)) {
+		kvm_riscv_local_hfence_vvma_asid_all(vmid, asid);
+		return;
+	}
+
+	hgatp = csr_swap(CSR_HGATP, vmid << HGATP_VMID_SHIFT);
+
+	for (pos = gva; pos < (gva + gvsz); pos += BIT(order)) {
+		/*
+		 * rs1 = a0 (GVA)
+		 * rs2 = a1 (ASID)
+		 * HFENCE.VVMA a0, a1
+		 * 0010001 01011 01010 000 00000 1110011
+		 */
+		asm volatile ("add a0, %0, zero\n"
+			      "add a1, %1, zero\n"
+			      ".word 0x22b50073\n"
+			      :: "r" (pos), "r" (asid)
+			      : "a0", "a1", "memory");
+	}
+
+	csr_write(CSR_HGATP, hgatp);
+}
+
+void kvm_riscv_local_hfence_vvma_asid_all(unsigned long vmid,
+					  unsigned long asid)
+{
+	unsigned long hgatp;
+
+	hgatp = csr_swap(CSR_HGATP, vmid << HGATP_VMID_SHIFT);
+
+	/*
+	 * rs1 = zero
+	 * rs2 = a0 (ASID)
+	 * HFENCE.VVMA zero, a0
+	 * 0010001 01010 00000 000 00000 1110011
+	 */
+	asm volatile ("add a0, %0, zero\n"
+		      ".word 0x22a00073\n"
+		      :: "r" (asid) : "a0", "memory");
+
+	csr_write(CSR_HGATP, hgatp);
+}
+
+void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
+				     unsigned long gva, unsigned long gvsz,
+				     unsigned long order)
+{
+	unsigned long pos, hgatp;
+
+	if (PTRS_PER_PTE < (gvsz >> order)) {
+		kvm_riscv_local_hfence_vvma_all(vmid);
+		return;
+	}
+
+	hgatp = csr_swap(CSR_HGATP, vmid << HGATP_VMID_SHIFT);
+
+	for (pos = gva; pos < (gva + gvsz); pos += BIT(order)) {
+		/*
+		 * rs1 = a0 (GVA)
+		 * rs2 = zero
+		 * HFENCE.VVMA a0
+		 * 0010001 00000 01010 000 00000 1110011
+		 */
+		asm volatile ("add a0, %0, zero\n"
+			      ".word 0x22050073\n"
+			      :: "r" (pos) : "a0", "memory");
+	}
+
+	csr_write(CSR_HGATP, hgatp);
+}
+
+void kvm_riscv_local_hfence_vvma_all(unsigned long vmid)
+{
+	unsigned long hgatp;
+
+	hgatp = csr_swap(CSR_HGATP, vmid << HGATP_VMID_SHIFT);
+
+	/*
+	 * rs1 = zero
+	 * rs2 = zero
+	 * HFENCE.VVMA
+	 * 0010001 00000 00000 000 00000 1110011
+	 */
+	asm volatile (".word 0x22000073" ::: "memory");
+
+	csr_write(CSR_HGATP, hgatp);
+}
+
+void kvm_riscv_local_tlb_sanitize(struct kvm_vcpu *vcpu)
+{
+	unsigned long vmid;
+
+	if (!kvm_riscv_gstage_vmid_bits() ||
+	    vcpu->arch.last_exit_cpu == vcpu->cpu)
+		return;
+
+	/*
+	 * On RISC-V platforms with hardware VMID support, we share same
+	 * VMID for all VCPUs of a particular Guest/VM. This means we might
+	 * have stale G-stage TLB entries on the current Host CPU due to
+	 * some other VCPU of the same Guest which ran previously on the
+	 * current Host CPU.
+	 *
+	 * To cleanup stale TLB entries, we simply flush all G-stage TLB
+	 * entries by VMID whenever underlying Host CPU changes for a VCPU.
+	 */
+
+	vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
+	kvm_riscv_local_hfence_gvma_vmid_all(vmid);
+}
+
+void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu)
+{
+	local_flush_icache_all();
+}
+
+void kvm_riscv_hfence_gvma_vmid_all_process(struct kvm_vcpu *vcpu)
+{
+	struct kvm_vmid *vmid;
+
+	vmid = &vcpu->kvm->arch.vmid;
+	kvm_riscv_local_hfence_gvma_vmid_all(READ_ONCE(vmid->vmid));
+}
+
+void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu)
+{
+	struct kvm_vmid *vmid;
+
+	vmid = &vcpu->kvm->arch.vmid;
+	kvm_riscv_local_hfence_vvma_all(READ_ONCE(vmid->vmid));
+}
+
+static bool vcpu_hfence_dequeue(struct kvm_vcpu *vcpu,
+				struct kvm_riscv_hfence *out_data)
+{
+	bool ret = false;
+	struct kvm_vcpu_arch *varch = &vcpu->arch;
+
+	spin_lock(&varch->hfence_lock);
+
+	if (varch->hfence_queue[varch->hfence_head].type) {
+		memcpy(out_data, &varch->hfence_queue[varch->hfence_head],
+		       sizeof(*out_data));
+		varch->hfence_queue[varch->hfence_head].type = 0;
+
+		varch->hfence_head++;
+		if (varch->hfence_head == KVM_RISCV_VCPU_MAX_HFENCE)
+			varch->hfence_head = 0;
+
+		ret = true;
+	}
+
+	spin_unlock(&varch->hfence_lock);
+
+	return ret;
+}
+
+static bool vcpu_hfence_enqueue(struct kvm_vcpu *vcpu,
+				const struct kvm_riscv_hfence *data)
+{
+	bool ret = false;
+	struct kvm_vcpu_arch *varch = &vcpu->arch;
+
+	spin_lock(&varch->hfence_lock);
+
+	if (!varch->hfence_queue[varch->hfence_tail].type) {
+		memcpy(&varch->hfence_queue[varch->hfence_tail],
+		       data, sizeof(*data));
+
+		varch->hfence_tail++;
+		if (varch->hfence_tail == KVM_RISCV_VCPU_MAX_HFENCE)
+			varch->hfence_tail = 0;
+
+		ret = true;
+	}
+
+	spin_unlock(&varch->hfence_lock);
+
+	return ret;
+}
+
+void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu)
+{
+	struct kvm_riscv_hfence d = { 0 };
+	struct kvm_vmid *v = &vcpu->kvm->arch.vmid;
+
+	while (vcpu_hfence_dequeue(vcpu, &d)) {
+		switch (d.type) {
+		case KVM_RISCV_HFENCE_UNKNOWN:
+			break;
+		case KVM_RISCV_HFENCE_GVMA_VMID_GPA:
+			kvm_riscv_local_hfence_gvma_vmid_gpa(
+						READ_ONCE(v->vmid),
+						d.addr, d.size, d.order);
+			break;
+		case KVM_RISCV_HFENCE_VVMA_ASID_GVA:
+			kvm_riscv_local_hfence_vvma_asid_gva(
+						READ_ONCE(v->vmid), d.asid,
+						d.addr, d.size, d.order);
+			break;
+		case KVM_RISCV_HFENCE_VVMA_ASID_ALL:
+			kvm_riscv_local_hfence_vvma_asid_all(
+						READ_ONCE(v->vmid), d.asid);
+			break;
+		case KVM_RISCV_HFENCE_VVMA_GVA:
+			kvm_riscv_local_hfence_vvma_gva(
+						READ_ONCE(v->vmid),
+						d.addr, d.size, d.order);
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+static void make_xfence_request(struct kvm *kvm,
+				unsigned long hbase, unsigned long hmask,
+				unsigned int req, unsigned int fallback_req,
+				const struct kvm_riscv_hfence *data)
+{
+	unsigned long i;
+	struct kvm_vcpu *vcpu;
+	unsigned int actual_req = req;
+	DECLARE_BITMAP(vcpu_mask, KVM_MAX_VCPUS);
+
+	bitmap_clear(vcpu_mask, 0, KVM_MAX_VCPUS);
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		if (hbase != -1UL) {
+			if (vcpu->vcpu_id < hbase)
+				continue;
+			if (!(hmask & (1UL << (vcpu->vcpu_id - hbase))))
+				continue;
+		}
+
+		bitmap_set(vcpu_mask, i, 1);
+
+		if (!data || !data->type)
+			continue;
+
+		/*
+		 * Enqueue hfence data to VCPU hfence queue. If we don't
+		 * have space in the VCPU hfence queue then fallback to
+		 * a more conservative hfence request.
+		 */
+		if (!vcpu_hfence_enqueue(vcpu, data))
+			actual_req = fallback_req;
+	}
+
+	kvm_make_vcpus_request_mask(kvm, actual_req, vcpu_mask);
+}
+
+void kvm_riscv_fence_i(struct kvm *kvm,
+		       unsigned long hbase, unsigned long hmask)
+{
+	make_xfence_request(kvm, hbase, hmask, KVM_REQ_FENCE_I,
+			    KVM_REQ_FENCE_I, NULL);
+}
+
+void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask,
+				    gpa_t gpa, gpa_t gpsz,
+				    unsigned long order)
+{
+	struct kvm_riscv_hfence data;
+
+	data.type = KVM_RISCV_HFENCE_GVMA_VMID_GPA;
+	data.asid = 0;
+	data.addr = gpa;
+	data.size = gpsz;
+	data.order = order;
+	make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
+			    KVM_REQ_HFENCE_GVMA_VMID_ALL, &data);
+}
+
+void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask)
+{
+	make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_GVMA_VMID_ALL,
+			    KVM_REQ_HFENCE_GVMA_VMID_ALL, NULL);
+}
+
+void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask,
+				    unsigned long gva, unsigned long gvsz,
+				    unsigned long order, unsigned long asid)
+{
+	struct kvm_riscv_hfence data;
+
+	data.type = KVM_RISCV_HFENCE_VVMA_ASID_GVA;
+	data.asid = asid;
+	data.addr = gva;
+	data.size = gvsz;
+	data.order = order;
+	make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
+			    KVM_REQ_HFENCE_VVMA_ALL, &data);
+}
+
+void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
+				    unsigned long hbase, unsigned long hmask,
+				    unsigned long asid)
+{
+	struct kvm_riscv_hfence data;
+
+	data.type = KVM_RISCV_HFENCE_VVMA_ASID_ALL;
+	data.asid = asid;
+	data.addr = data.size = data.order = 0;
+	make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
+			    KVM_REQ_HFENCE_VVMA_ALL, &data);
+}
+
+void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
+			       unsigned long hbase, unsigned long hmask,
+			       unsigned long gva, unsigned long gvsz,
+			       unsigned long order)
+{
+	struct kvm_riscv_hfence data;
+
+	data.type = KVM_RISCV_HFENCE_VVMA_GVA;
+	data.asid = 0;
+	data.addr = gva;
+	data.size = gvsz;
+	data.order = order;
+	make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
+			    KVM_REQ_HFENCE_VVMA_ALL, &data);
+}
+
+void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
+			       unsigned long hbase, unsigned long hmask)
+{
+	make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_VVMA_ALL,
+			    KVM_REQ_HFENCE_VVMA_ALL, NULL);
+}
diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index 7461f964d20a..7f4ad5e4373a 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -67,6 +67,8 @@ static void kvm_riscv_reset_vcpu(struct kvm_vcpu *vcpu)
 	if (loaded)
 		kvm_arch_vcpu_put(vcpu);
 
+	vcpu->arch.last_exit_cpu = -1;
+
 	memcpy(csr, reset_csr, sizeof(*csr));
 
 	memcpy(cntx, reset_cntx, sizeof(*cntx));
@@ -78,6 +80,10 @@ static void kvm_riscv_reset_vcpu(struct kvm_vcpu *vcpu)
 	WRITE_ONCE(vcpu->arch.irqs_pending, 0);
 	WRITE_ONCE(vcpu->arch.irqs_pending_mask, 0);
 
+	vcpu->arch.hfence_head = 0;
+	vcpu->arch.hfence_tail = 0;
+	memset(vcpu->arch.hfence_queue, 0, sizeof(vcpu->arch.hfence_queue));
+
 	/* Reset the guest CSRs for hotplug usecase */
 	if (loaded)
 		kvm_arch_vcpu_load(vcpu, smp_processor_id());
@@ -101,6 +107,9 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
 	/* Setup ISA features available to VCPU */
 	vcpu->arch.isa = riscv_isa_extension_base(NULL) & KVM_RISCV_ISA_ALLOWED;
 
+	/* Setup VCPU hfence queue */
+	spin_lock_init(&vcpu->arch.hfence_lock);
+
 	/* Setup reset state of shadow SSTATUS and HSTATUS CSRs */
 	cntx = &vcpu->arch.guest_reset_context;
 	cntx->sstatus = SR_SPP | SR_SPIE;
@@ -137,7 +146,7 @@ void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu)
 	/* Cleanup VCPU timer */
 	kvm_riscv_vcpu_timer_deinit(vcpu);
 
-	/* Free unused pages pre-allocated for Stage2 page table mappings */
+	/* Free unused pages pre-allocated for G-stage page table mappings */
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_page_cache);
 }
 
@@ -365,6 +374,101 @@ static int kvm_riscv_vcpu_set_reg_csr(struct kvm_vcpu *vcpu,
 	return 0;
 }
 
+/* Mapping between KVM ISA Extension ID & Host ISA extension ID */
+static unsigned long kvm_isa_ext_arr[] = {
+	RISCV_ISA_EXT_a,
+	RISCV_ISA_EXT_c,
+	RISCV_ISA_EXT_d,
+	RISCV_ISA_EXT_f,
+	RISCV_ISA_EXT_h,
+	RISCV_ISA_EXT_i,
+	RISCV_ISA_EXT_m,
+};
+
+static int kvm_riscv_vcpu_get_reg_isa_ext(struct kvm_vcpu *vcpu,
+					  const struct kvm_one_reg *reg)
+{
+	unsigned long __user *uaddr =
+			(unsigned long __user *)(unsigned long)reg->addr;
+	unsigned long reg_num = reg->id & ~(KVM_REG_ARCH_MASK |
+					    KVM_REG_SIZE_MASK |
+					    KVM_REG_RISCV_ISA_EXT);
+	unsigned long reg_val = 0;
+	unsigned long host_isa_ext;
+
+	if (KVM_REG_SIZE(reg->id) != sizeof(unsigned long))
+		return -EINVAL;
+
+	if (reg_num >= KVM_RISCV_ISA_EXT_MAX || reg_num >= ARRAY_SIZE(kvm_isa_ext_arr))
+		return -EINVAL;
+
+	host_isa_ext = kvm_isa_ext_arr[reg_num];
+	if (__riscv_isa_extension_available(&vcpu->arch.isa, host_isa_ext))
+		reg_val = 1; /* Mark the given extension as available */
+
+	if (copy_to_user(uaddr, &reg_val, KVM_REG_SIZE(reg->id)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static int kvm_riscv_vcpu_set_reg_isa_ext(struct kvm_vcpu *vcpu,
+					  const struct kvm_one_reg *reg)
+{
+	unsigned long __user *uaddr =
+			(unsigned long __user *)(unsigned long)reg->addr;
+	unsigned long reg_num = reg->id & ~(KVM_REG_ARCH_MASK |
+					    KVM_REG_SIZE_MASK |
+					    KVM_REG_RISCV_ISA_EXT);
+	unsigned long reg_val;
+	unsigned long host_isa_ext;
+	unsigned long host_isa_ext_mask;
+
+	if (KVM_REG_SIZE(reg->id) != sizeof(unsigned long))
+		return -EINVAL;
+
+	if (reg_num >= KVM_RISCV_ISA_EXT_MAX || reg_num >= ARRAY_SIZE(kvm_isa_ext_arr))
+		return -EINVAL;
+
+	if (copy_from_user(&reg_val, uaddr, KVM_REG_SIZE(reg->id)))
+		return -EFAULT;
+
+	host_isa_ext = kvm_isa_ext_arr[reg_num];
+	if (!__riscv_isa_extension_available(NULL, host_isa_ext))
+		return	-EOPNOTSUPP;
+
+	if (host_isa_ext >= RISCV_ISA_EXT_BASE &&
+	    host_isa_ext < RISCV_ISA_EXT_MAX) {
+		/*
+		 * Multi-letter ISA extension. Currently there is no provision
+		 * to enable/disable the multi-letter ISA extensions for guests.
+		 * Return success if the request is to enable any ISA extension
+		 * that is available in the hardware.
+		 * Return -EOPNOTSUPP otherwise.
+		 */
+		if (!reg_val)
+			return -EOPNOTSUPP;
+		else
+			return 0;
+	}
+
+	/* Single letter base ISA extension */
+	if (!vcpu->arch.ran_atleast_once) {
+		host_isa_ext_mask = BIT_MASK(host_isa_ext);
+		if (!reg_val && (host_isa_ext_mask & KVM_RISCV_ISA_DISABLE_ALLOWED))
+			vcpu->arch.isa &= ~host_isa_ext_mask;
+		else
+			vcpu->arch.isa |= host_isa_ext_mask;
+		vcpu->arch.isa &= riscv_isa_extension_base(NULL);
+		vcpu->arch.isa &= KVM_RISCV_ISA_ALLOWED;
+		kvm_riscv_vcpu_fp_reset(vcpu);
+	} else {
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
 static int kvm_riscv_vcpu_set_reg(struct kvm_vcpu *vcpu,
 				  const struct kvm_one_reg *reg)
 {
@@ -382,6 +486,8 @@ static int kvm_riscv_vcpu_set_reg(struct kvm_vcpu *vcpu,
 	else if ((reg->id & KVM_REG_RISCV_TYPE_MASK) == KVM_REG_RISCV_FP_D)
 		return kvm_riscv_vcpu_set_reg_fp(vcpu, reg,
 						 KVM_REG_RISCV_FP_D);
+	else if ((reg->id & KVM_REG_RISCV_TYPE_MASK) == KVM_REG_RISCV_ISA_EXT)
+		return kvm_riscv_vcpu_set_reg_isa_ext(vcpu, reg);
 
 	return -EINVAL;
 }
@@ -403,6 +509,8 @@ static int kvm_riscv_vcpu_get_reg(struct kvm_vcpu *vcpu,
 	else if ((reg->id & KVM_REG_RISCV_TYPE_MASK) == KVM_REG_RISCV_FP_D)
 		return kvm_riscv_vcpu_get_reg_fp(vcpu, reg,
 						 KVM_REG_RISCV_FP_D);
+	else if ((reg->id & KVM_REG_RISCV_TYPE_MASK) == KVM_REG_RISCV_ISA_EXT)
+		return kvm_riscv_vcpu_get_reg_isa_ext(vcpu, reg);
 
 	return -EINVAL;
 }
@@ -635,7 +743,7 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	csr_write(CSR_HVIP, csr->hvip);
 	csr_write(CSR_VSATP, csr->vsatp);
 
-	kvm_riscv_stage2_update_hgatp(vcpu);
+	kvm_riscv_gstage_update_hgatp(vcpu);
 
 	kvm_riscv_vcpu_timer_restore(vcpu);
 
@@ -690,10 +798,23 @@ static void kvm_riscv_check_vcpu_requests(struct kvm_vcpu *vcpu)
 			kvm_riscv_reset_vcpu(vcpu);
 
 		if (kvm_check_request(KVM_REQ_UPDATE_HGATP, vcpu))
-			kvm_riscv_stage2_update_hgatp(vcpu);
+			kvm_riscv_gstage_update_hgatp(vcpu);
 
-		if (kvm_check_request(KVM_REQ_TLB_FLUSH, vcpu))
-			__kvm_riscv_hfence_gvma_all();
+		if (kvm_check_request(KVM_REQ_FENCE_I, vcpu))
+			kvm_riscv_fence_i_process(vcpu);
+
+		/*
+		 * The generic KVM_REQ_TLB_FLUSH is same as
+		 * KVM_REQ_HFENCE_GVMA_VMID_ALL
+		 */
+		if (kvm_check_request(KVM_REQ_HFENCE_GVMA_VMID_ALL, vcpu))
+			kvm_riscv_hfence_gvma_vmid_all_process(vcpu);
+
+		if (kvm_check_request(KVM_REQ_HFENCE_VVMA_ALL, vcpu))
+			kvm_riscv_hfence_vvma_all_process(vcpu);
+
+		if (kvm_check_request(KVM_REQ_HFENCE, vcpu))
+			kvm_riscv_hfence_process(vcpu);
 	}
 }
 
@@ -715,6 +836,7 @@ static void noinstr kvm_riscv_vcpu_enter_exit(struct kvm_vcpu *vcpu)
 {
 	guest_state_enter_irqoff();
 	__kvm_riscv_switch_to(&vcpu->arch);
+	vcpu->arch.last_exit_cpu = vcpu->cpu;
 	guest_state_exit_irqoff();
 }
 
@@ -762,7 +884,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 		/* Check conditions before entering the guest */
 		cond_resched();
 
-		kvm_riscv_stage2_vmid_update(vcpu);
+		kvm_riscv_gstage_vmid_update(vcpu);
 
 		kvm_riscv_check_vcpu_requests(vcpu);
 
@@ -800,7 +922,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 		kvm_riscv_update_hvip(vcpu);
 
 		if (ret <= 0 ||
-		    kvm_riscv_stage2_vmid_ver_changed(&vcpu->kvm->arch.vmid) ||
+		    kvm_riscv_gstage_vmid_ver_changed(&vcpu->kvm->arch.vmid) ||
 		    kvm_request_pending(vcpu)) {
 			vcpu->mode = OUTSIDE_GUEST_MODE;
 			local_irq_enable();
@@ -809,6 +931,14 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
 			continue;
 		}
 
+		/*
+		 * Cleanup stale TLB enteries
+		 *
+		 * Note: This should be done after G-stage VMID has been
+		 * updated using kvm_riscv_gstage_vmid_ver_changed()
+		 */
+		kvm_riscv_local_tlb_sanitize(vcpu);
+
 		guest_timing_enter_irqoff();
 
 		kvm_riscv_vcpu_enter_exit(vcpu);
diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
index a72c15d4b42a..dbb09afd7546 100644
--- a/arch/riscv/kvm/vcpu_exit.c
+++ b/arch/riscv/kvm/vcpu_exit.c
@@ -412,7 +412,7 @@ static int emulate_store(struct kvm_vcpu *vcpu, struct kvm_run *run,
 	return 0;
 }
 
-static int stage2_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
+static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
 			     struct kvm_cpu_trap *trap)
 {
 	struct kvm_memory_slot *memslot;
@@ -440,7 +440,7 @@ static int stage2_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
 		};
 	}
 
-	ret = kvm_riscv_stage2_map(vcpu, memslot, fault_addr, hva,
+	ret = kvm_riscv_gstage_map(vcpu, memslot, fault_addr, hva,
 		(trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false);
 	if (ret < 0)
 		return ret;
@@ -686,7 +686,7 @@ int kvm_riscv_vcpu_exit(struct kvm_vcpu *vcpu, struct kvm_run *run,
 	case EXC_LOAD_GUEST_PAGE_FAULT:
 	case EXC_STORE_GUEST_PAGE_FAULT:
 		if (vcpu->arch.guest_context.hstatus & HSTATUS_SPV)
-			ret = stage2_page_fault(vcpu, run, trap);
+			ret = gstage_page_fault(vcpu, run, trap);
 		break;
 	case EXC_SUPERVISOR_SYSCALL:
 		if (vcpu->arch.guest_context.hstatus & HSTATUS_SPV)
diff --git a/arch/riscv/kvm/vcpu_sbi_replace.c b/arch/riscv/kvm/vcpu_sbi_replace.c
index 0f217365c287..4c034d8a606a 100644
--- a/arch/riscv/kvm/vcpu_sbi_replace.c
+++ b/arch/riscv/kvm/vcpu_sbi_replace.c
@@ -81,43 +81,41 @@ static int kvm_sbi_ext_rfence_handler(struct kvm_vcpu *vcpu, struct kvm_run *run
 				      struct kvm_cpu_trap *utrap, bool *exit)
 {
 	int ret = 0;
-	unsigned long i;
-	struct cpumask cm;
-	struct kvm_vcpu *tmp;
 	struct kvm_cpu_context *cp = &vcpu->arch.guest_context;
 	unsigned long hmask = cp->a0;
 	unsigned long hbase = cp->a1;
 	unsigned long funcid = cp->a6;
 
-	cpumask_clear(&cm);
-	kvm_for_each_vcpu(i, tmp, vcpu->kvm) {
-		if (hbase != -1UL) {
-			if (tmp->vcpu_id < hbase)
-				continue;
-			if (!(hmask & (1UL << (tmp->vcpu_id - hbase))))
-				continue;
-		}
-		if (tmp->cpu < 0)
-			continue;
-		cpumask_set_cpu(tmp->cpu, &cm);
-	}
-
 	switch (funcid) {
 	case SBI_EXT_RFENCE_REMOTE_FENCE_I:
-		ret = sbi_remote_fence_i(&cm);
+		kvm_riscv_fence_i(vcpu->kvm, hbase, hmask);
 		break;
 	case SBI_EXT_RFENCE_REMOTE_SFENCE_VMA:
-		ret = sbi_remote_hfence_vvma(&cm, cp->a2, cp->a3);
+		if (cp->a2 == 0 && cp->a3 == 0)
+			kvm_riscv_hfence_vvma_all(vcpu->kvm, hbase, hmask);
+		else
+			kvm_riscv_hfence_vvma_gva(vcpu->kvm, hbase, hmask,
+						  cp->a2, cp->a3, PAGE_SHIFT);
 		break;
 	case SBI_EXT_RFENCE_REMOTE_SFENCE_VMA_ASID:
-		ret = sbi_remote_hfence_vvma_asid(&cm, cp->a2,
-						  cp->a3, cp->a4);
+		if (cp->a2 == 0 && cp->a3 == 0)
+			kvm_riscv_hfence_vvma_asid_all(vcpu->kvm,
+						       hbase, hmask, cp->a4);
+		else
+			kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm,
+						       hbase, hmask,
+						       cp->a2, cp->a3,
+						       PAGE_SHIFT, cp->a4);
 		break;
 	case SBI_EXT_RFENCE_REMOTE_HFENCE_GVMA:
 	case SBI_EXT_RFENCE_REMOTE_HFENCE_GVMA_VMID:
 	case SBI_EXT_RFENCE_REMOTE_HFENCE_VVMA:
 	case SBI_EXT_RFENCE_REMOTE_HFENCE_VVMA_ASID:
-	/* TODO: implement for nested hypervisor case */
+		/*
+		 * Until nested virtualization is implemented, the
+		 * SBI HFENCE calls should be treated as NOPs
+		 */
+		break;
 	default:
 		ret = -EOPNOTSUPP;
 	}
diff --git a/arch/riscv/kvm/vcpu_sbi_v01.c b/arch/riscv/kvm/vcpu_sbi_v01.c
index da4d6c99c2cf..8a91a14e7139 100644
--- a/arch/riscv/kvm/vcpu_sbi_v01.c
+++ b/arch/riscv/kvm/vcpu_sbi_v01.c
@@ -23,7 +23,6 @@ static int kvm_sbi_ext_v01_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
 	int i, ret = 0;
 	u64 next_cycle;
 	struct kvm_vcpu *rvcpu;
-	struct cpumask cm;
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_cpu_context *cp = &vcpu->arch.guest_context;
 
@@ -80,19 +79,29 @@ static int kvm_sbi_ext_v01_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
 		if (utrap->scause)
 			break;
 
-		cpumask_clear(&cm);
-		for_each_set_bit(i, &hmask, BITS_PER_LONG) {
-			rvcpu = kvm_get_vcpu_by_id(vcpu->kvm, i);
-			if (rvcpu->cpu < 0)
-				continue;
-			cpumask_set_cpu(rvcpu->cpu, &cm);
-		}
 		if (cp->a7 == SBI_EXT_0_1_REMOTE_FENCE_I)
-			ret = sbi_remote_fence_i(&cm);
-		else if (cp->a7 == SBI_EXT_0_1_REMOTE_SFENCE_VMA)
-			ret = sbi_remote_hfence_vvma(&cm, cp->a1, cp->a2);
-		else
-			ret = sbi_remote_hfence_vvma_asid(&cm, cp->a1, cp->a2, cp->a3);
+			kvm_riscv_fence_i(vcpu->kvm, 0, hmask);
+		else if (cp->a7 == SBI_EXT_0_1_REMOTE_SFENCE_VMA) {
+			if (cp->a1 == 0 && cp->a2 == 0)
+				kvm_riscv_hfence_vvma_all(vcpu->kvm,
+							  0, hmask);
+			else
+				kvm_riscv_hfence_vvma_gva(vcpu->kvm,
+							  0, hmask,
+							  cp->a1, cp->a2,
+							  PAGE_SHIFT);
+		} else {
+			if (cp->a1 == 0 && cp->a2 == 0)
+				kvm_riscv_hfence_vvma_asid_all(vcpu->kvm,
+							       0, hmask,
+							       cp->a3);
+			else
+				kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm,
+							       0, hmask,
+							       cp->a1, cp->a2,
+							       PAGE_SHIFT,
+							       cp->a3);
+		}
 		break;
 	default:
 		ret = -EINVAL;
diff --git a/arch/riscv/kvm/vm.c b/arch/riscv/kvm/vm.c
index c768f75279ef..945a2bf5e3f6 100644
--- a/arch/riscv/kvm/vm.c
+++ b/arch/riscv/kvm/vm.c
@@ -31,13 +31,13 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 {
 	int r;
 
-	r = kvm_riscv_stage2_alloc_pgd(kvm);
+	r = kvm_riscv_gstage_alloc_pgd(kvm);
 	if (r)
 		return r;
 
-	r = kvm_riscv_stage2_vmid_init(kvm);
+	r = kvm_riscv_gstage_vmid_init(kvm);
 	if (r) {
-		kvm_riscv_stage2_free_pgd(kvm);
+		kvm_riscv_gstage_free_pgd(kvm);
 		return r;
 	}
 
@@ -75,7 +75,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 		r = KVM_USER_MEM_SLOTS;
 		break;
 	case KVM_CAP_VM_GPA_BITS:
-		r = kvm_riscv_stage2_gpa_bits();
+		r = kvm_riscv_gstage_gpa_bits();
 		break;
 	default:
 		r = 0;
diff --git a/arch/riscv/kvm/vmid.c b/arch/riscv/kvm/vmid.c
index 2fa4f7b1813d..9f764df125db 100644
--- a/arch/riscv/kvm/vmid.c
+++ b/arch/riscv/kvm/vmid.c
@@ -11,16 +11,16 @@
 #include <linux/errno.h>
 #include <linux/err.h>
 #include <linux/module.h>
+#include <linux/smp.h>
 #include <linux/kvm_host.h>
 #include <asm/csr.h>
-#include <asm/sbi.h>
 
 static unsigned long vmid_version = 1;
 static unsigned long vmid_next;
 static unsigned long vmid_bits;
 static DEFINE_SPINLOCK(vmid_lock);
 
-void kvm_riscv_stage2_vmid_detect(void)
+void kvm_riscv_gstage_vmid_detect(void)
 {
 	unsigned long old;
 
@@ -33,19 +33,19 @@ void kvm_riscv_stage2_vmid_detect(void)
 	csr_write(CSR_HGATP, old);
 
 	/* We polluted local TLB so flush all guest TLB */
-	__kvm_riscv_hfence_gvma_all();
+	kvm_riscv_local_hfence_gvma_all();
 
 	/* We don't use VMID bits if they are not sufficient */
 	if ((1UL << vmid_bits) < num_possible_cpus())
 		vmid_bits = 0;
 }
 
-unsigned long kvm_riscv_stage2_vmid_bits(void)
+unsigned long kvm_riscv_gstage_vmid_bits(void)
 {
 	return vmid_bits;
 }
 
-int kvm_riscv_stage2_vmid_init(struct kvm *kvm)
+int kvm_riscv_gstage_vmid_init(struct kvm *kvm)
 {
 	/* Mark the initial VMID and VMID version invalid */
 	kvm->arch.vmid.vmid_version = 0;
@@ -54,7 +54,7 @@ int kvm_riscv_stage2_vmid_init(struct kvm *kvm)
 	return 0;
 }
 
-bool kvm_riscv_stage2_vmid_ver_changed(struct kvm_vmid *vmid)
+bool kvm_riscv_gstage_vmid_ver_changed(struct kvm_vmid *vmid)
 {
 	if (!vmid_bits)
 		return false;
@@ -63,13 +63,18 @@ bool kvm_riscv_stage2_vmid_ver_changed(struct kvm_vmid *vmid)
 			READ_ONCE(vmid_version));
 }
 
-void kvm_riscv_stage2_vmid_update(struct kvm_vcpu *vcpu)
+static void __local_hfence_gvma_all(void *info)
+{
+	kvm_riscv_local_hfence_gvma_all();
+}
+
+void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu)
 {
 	unsigned long i;
 	struct kvm_vcpu *v;
 	struct kvm_vmid *vmid = &vcpu->kvm->arch.vmid;
 
-	if (!kvm_riscv_stage2_vmid_ver_changed(vmid))
+	if (!kvm_riscv_gstage_vmid_ver_changed(vmid))
 		return;
 
 	spin_lock(&vmid_lock);
@@ -78,7 +83,7 @@ void kvm_riscv_stage2_vmid_update(struct kvm_vcpu *vcpu)
 	 * We need to re-check the vmid_version here to ensure that if
 	 * another vcpu already allocated a valid vmid for this vm.
 	 */
-	if (!kvm_riscv_stage2_vmid_ver_changed(vmid)) {
+	if (!kvm_riscv_gstage_vmid_ver_changed(vmid)) {
 		spin_unlock(&vmid_lock);
 		return;
 	}
@@ -96,12 +101,13 @@ void kvm_riscv_stage2_vmid_update(struct kvm_vcpu *vcpu)
 		 * instances is invalid and we have force VMID re-assignement
 		 * for all Guest instances. The Guest instances that were not
 		 * running will automatically pick-up new VMIDs because will
-		 * call kvm_riscv_stage2_vmid_update() whenever they enter
+		 * call kvm_riscv_gstage_vmid_update() whenever they enter
 		 * in-kernel run loop. For Guest instances that are already
 		 * running, we force VM exits on all host CPUs using IPI and
 		 * flush all Guest TLBs.
 		 */
-		sbi_remote_hfence_gvma(cpu_online_mask, 0, 0);
+		on_each_cpu_mask(cpu_online_mask, __local_hfence_gvma_all,
+				 NULL, 1);
 	}
 
 	vmid->vmid = vmid_next;
@@ -112,7 +118,7 @@ void kvm_riscv_stage2_vmid_update(struct kvm_vcpu *vcpu)
 
 	spin_unlock(&vmid_lock);
 
-	/* Request stage2 page table update for all VCPUs */
+	/* Request G-stage page table update for all VCPUs */
 	kvm_for_each_vcpu(i, v, vcpu->kvm)
 		kvm_make_request(KVM_REQ_UPDATE_HGATP, v);
 }
diff --git a/arch/s390/include/asm/uv.h b/arch/s390/include/asm/uv.h
index a2d376b8bce3..cfea7b77a5b8 100644
--- a/arch/s390/include/asm/uv.h
+++ b/arch/s390/include/asm/uv.h
@@ -2,7 +2,7 @@
 /*
  * Ultravisor Interfaces
  *
- * Copyright IBM Corp. 2019
+ * Copyright IBM Corp. 2019, 2022
  *
  * Author(s):
  *	Vasily Gorbik <gor@linux.ibm.com>
@@ -52,6 +52,7 @@
 #define UVC_CMD_UNPIN_PAGE_SHARED	0x0342
 #define UVC_CMD_SET_SHARED_ACCESS	0x1000
 #define UVC_CMD_REMOVE_SHARED_ACCESS	0x1001
+#define UVC_CMD_RETR_ATTEST		0x1020
 
 /* Bits in installed uv calls */
 enum uv_cmds_inst {
@@ -76,6 +77,7 @@ enum uv_cmds_inst {
 	BIT_UVC_CMD_UNSHARE_ALL = 20,
 	BIT_UVC_CMD_PIN_PAGE_SHARED = 21,
 	BIT_UVC_CMD_UNPIN_PAGE_SHARED = 22,
+	BIT_UVC_CMD_RETR_ATTEST = 28,
 };
 
 enum uv_feat_ind {
@@ -219,6 +221,25 @@ struct uv_cb_share {
 	u64 reserved28;
 } __packed __aligned(8);
 
+/* Retrieve Attestation Measurement */
+struct uv_cb_attest {
+	struct uv_cb_header header;	/* 0x0000 */
+	u64 reserved08[2];		/* 0x0008 */
+	u64 arcb_addr;			/* 0x0018 */
+	u64 cont_token;			/* 0x0020 */
+	u8  reserved28[6];		/* 0x0028 */
+	u16 user_data_len;		/* 0x002e */
+	u8  user_data[256];		/* 0x0030 */
+	u32 reserved130[3];		/* 0x0130 */
+	u32 meas_len;			/* 0x013c */
+	u64 meas_addr;			/* 0x0140 */
+	u8  config_uid[16];		/* 0x0148 */
+	u32 reserved158;		/* 0x0158 */
+	u32 add_data_len;		/* 0x015c */
+	u64 add_data_addr;		/* 0x0160 */
+	u64 reserved168[4];		/* 0x0168 */
+} __packed __aligned(8);
+
 static inline int __uv_call(unsigned long r1, unsigned long r2)
 {
 	int cc;
diff --git a/arch/s390/include/uapi/asm/uvdevice.h b/arch/s390/include/uapi/asm/uvdevice.h
new file mode 100644
index 000000000000..10a5ac918e02
--- /dev/null
+++ b/arch/s390/include/uapi/asm/uvdevice.h
@@ -0,0 +1,51 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+/*
+ *  Copyright IBM Corp. 2022
+ *  Author(s): Steffen Eiden <seiden@linux.ibm.com>
+ */
+#ifndef __S390_ASM_UVDEVICE_H
+#define __S390_ASM_UVDEVICE_H
+
+#include <linux/types.h>
+
+struct uvio_ioctl_cb {
+	__u32 flags;
+	__u16 uv_rc;			/* UV header rc value */
+	__u16 uv_rrc;			/* UV header rrc value */
+	__u64 argument_addr;		/* Userspace address of uvio argument */
+	__u32 argument_len;
+	__u8  reserved14[0x40 - 0x14];	/* must be zero */
+};
+
+#define UVIO_ATT_USER_DATA_LEN		0x100
+#define UVIO_ATT_UID_LEN		0x10
+struct uvio_attest {
+	__u64 arcb_addr;				/* 0x0000 */
+	__u64 meas_addr;				/* 0x0008 */
+	__u64 add_data_addr;				/* 0x0010 */
+	__u8  user_data[UVIO_ATT_USER_DATA_LEN];	/* 0x0018 */
+	__u8  config_uid[UVIO_ATT_UID_LEN];		/* 0x0118 */
+	__u32 arcb_len;					/* 0x0128 */
+	__u32 meas_len;					/* 0x012c */
+	__u32 add_data_len;				/* 0x0130 */
+	__u16 user_data_len;				/* 0x0134 */
+	__u16 reserved136;				/* 0x0136 */
+};
+
+/*
+ * The following max values define an upper length for the IOCTL in/out buffers.
+ * However, they do not represent the maximum the Ultravisor allows which is
+ * often way smaller. By allowing larger buffer sizes we hopefully do not need
+ * to update the code with every machine update. It is therefore possible for
+ * userspace to request more memory than actually used by kernel/UV.
+ */
+#define UVIO_ATT_ARCB_MAX_LEN		0x100000
+#define UVIO_ATT_MEASUREMENT_MAX_LEN	0x8000
+#define UVIO_ATT_ADDITIONAL_MAX_LEN	0x8000
+
+#define UVIO_DEVICE_NAME "uv"
+#define UVIO_TYPE_UVC 'u'
+
+#define UVIO_IOCTL_ATT _IOWR(UVIO_TYPE_UVC, 0x01, struct uvio_ioctl_cb)
+
+#endif /* __S390_ASM_UVDEVICE_H */
diff --git a/arch/s390/kvm/gaccess.c b/arch/s390/kvm/gaccess.c
index d53a183c2005..227ed0009354 100644
--- a/arch/s390/kvm/gaccess.c
+++ b/arch/s390/kvm/gaccess.c
@@ -491,8 +491,8 @@ enum prot_type {
 	PROT_TYPE_IEP  = 4,
 };
 
-static int trans_exc(struct kvm_vcpu *vcpu, int code, unsigned long gva,
-		     u8 ar, enum gacc_mode mode, enum prot_type prot)
+static int trans_exc_ending(struct kvm_vcpu *vcpu, int code, unsigned long gva, u8 ar,
+			    enum gacc_mode mode, enum prot_type prot, bool terminate)
 {
 	struct kvm_s390_pgm_info *pgm = &vcpu->arch.pgm;
 	struct trans_exc_code_bits *tec;
@@ -520,6 +520,11 @@ static int trans_exc(struct kvm_vcpu *vcpu, int code, unsigned long gva,
 			tec->b61 = 1;
 			break;
 		}
+		if (terminate) {
+			tec->b56 = 0;
+			tec->b60 = 0;
+			tec->b61 = 0;
+		}
 		fallthrough;
 	case PGM_ASCE_TYPE:
 	case PGM_PAGE_TRANSLATION:
@@ -552,6 +557,12 @@ static int trans_exc(struct kvm_vcpu *vcpu, int code, unsigned long gva,
 	return code;
 }
 
+static int trans_exc(struct kvm_vcpu *vcpu, int code, unsigned long gva, u8 ar,
+		     enum gacc_mode mode, enum prot_type prot)
+{
+	return trans_exc_ending(vcpu, code, gva, ar, mode, prot, false);
+}
+
 static int get_vcpu_asce(struct kvm_vcpu *vcpu, union asce *asce,
 			 unsigned long ga, u8 ar, enum gacc_mode mode)
 {
@@ -1109,8 +1120,11 @@ int access_guest_with_key(struct kvm_vcpu *vcpu, unsigned long ga, u8 ar,
 		data += fragment_len;
 		ga = kvm_s390_logical_to_effective(vcpu, ga + fragment_len);
 	}
-	if (rc > 0)
-		rc = trans_exc(vcpu, rc, ga, ar, mode, prot);
+	if (rc > 0) {
+		bool terminate = (mode == GACC_STORE) && (idx > 0);
+
+		rc = trans_exc_ending(vcpu, rc, ga, ar, mode, prot, terminate);
+	}
 out_unlock:
 	if (need_ipte_lock)
 		ipte_unlock(vcpu);
diff --git a/arch/x86/include/asm/cpufeatures.h b/arch/x86/include/asm/cpufeatures.h
index 21bb78dfd41d..393f2bbb5e3a 100644
--- a/arch/x86/include/asm/cpufeatures.h
+++ b/arch/x86/include/asm/cpufeatures.h
@@ -407,6 +407,7 @@
 #define X86_FEATURE_SEV			(19*32+ 1) /* AMD Secure Encrypted Virtualization */
 #define X86_FEATURE_VM_PAGE_FLUSH	(19*32+ 2) /* "" VM Page Flush MSR is supported */
 #define X86_FEATURE_SEV_ES		(19*32+ 3) /* AMD Secure Encrypted Virtualization - Encrypted State */
+#define X86_FEATURE_V_TSC_AUX		(19*32+ 9) /* "" Virtual TSC_AUX */
 #define X86_FEATURE_SME_COHERENT	(19*32+10) /* "" AMD hardware-enforced cache coherency */
 
 /*
diff --git a/arch/x86/include/asm/kvm-x86-ops.h b/arch/x86/include/asm/kvm-x86-ops.h
index 1a6d7e3f6c32..da47f60a4650 100644
--- a/arch/x86/include/asm/kvm-x86-ops.h
+++ b/arch/x86/include/asm/kvm-x86-ops.h
@@ -127,6 +127,7 @@ KVM_X86_OP_OPTIONAL(migrate_timers)
 KVM_X86_OP(msr_filter_changed)
 KVM_X86_OP(complete_emulated_msr)
 KVM_X86_OP(vcpu_deliver_sipi_vector)
+KVM_X86_OP_OPTIONAL_RET0(vcpu_get_apicv_inhibit_reasons);
 
 #undef KVM_X86_OP
 #undef KVM_X86_OP_OPTIONAL
diff --git a/arch/x86/include/asm/kvm-x86-pmu-ops.h b/arch/x86/include/asm/kvm-x86-pmu-ops.h
new file mode 100644
index 000000000000..fdfd8e06fee6
--- /dev/null
+++ b/arch/x86/include/asm/kvm-x86-pmu-ops.h
@@ -0,0 +1,31 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#if !defined(KVM_X86_PMU_OP) || !defined(KVM_X86_PMU_OP_OPTIONAL)
+BUILD_BUG_ON(1)
+#endif
+
+/*
+ * KVM_X86_PMU_OP() and KVM_X86_PMU_OP_OPTIONAL() are used to help generate
+ * both DECLARE/DEFINE_STATIC_CALL() invocations and
+ * "static_call_update()" calls.
+ *
+ * KVM_X86_PMU_OP_OPTIONAL() can be used for those functions that can have
+ * a NULL definition, for example if "static_call_cond()" will be used
+ * at the call sites.
+ */
+KVM_X86_PMU_OP(pmc_perf_hw_id)
+KVM_X86_PMU_OP(pmc_is_enabled)
+KVM_X86_PMU_OP(pmc_idx_to_pmc)
+KVM_X86_PMU_OP(rdpmc_ecx_to_pmc)
+KVM_X86_PMU_OP(msr_idx_to_pmc)
+KVM_X86_PMU_OP(is_valid_rdpmc_ecx)
+KVM_X86_PMU_OP(is_valid_msr)
+KVM_X86_PMU_OP(get_msr)
+KVM_X86_PMU_OP(set_msr)
+KVM_X86_PMU_OP(refresh)
+KVM_X86_PMU_OP(init)
+KVM_X86_PMU_OP(reset)
+KVM_X86_PMU_OP_OPTIONAL(deliver_pmi)
+KVM_X86_PMU_OP_OPTIONAL(cleanup)
+
+#undef KVM_X86_PMU_OP
+#undef KVM_X86_PMU_OP_OPTIONAL
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 4ff36610af6a..959d66b9be94 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -281,11 +281,11 @@ struct kvm_kernel_irq_routing_entry;
 /*
  * kvm_mmu_page_role tracks the properties of a shadow page (where shadow page
  * also includes TDP pages) to determine whether or not a page can be used in
- * the given MMU context.  This is a subset of the overall kvm_mmu_role to
+ * the given MMU context.  This is a subset of the overall kvm_cpu_role to
  * minimize the size of kvm_memory_slot.arch.gfn_track, i.e. allows allocating
  * 2 bytes per gfn instead of 4 bytes per gfn.
  *
- * Indirect upper-level shadow pages are tracked for write-protection via
+ * Upper-level shadow pages having gptes are tracked for write-protection via
  * gfn_track.  As above, gfn_track is a 16 bit counter, so KVM must not create
  * more than 2^16-1 upper-level shadow pages at a single gfn, otherwise
  * gfn_track will overflow and explosions will ensure.
@@ -331,7 +331,8 @@ union kvm_mmu_page_role {
 		unsigned smap_andnot_wp:1;
 		unsigned ad_disabled:1;
 		unsigned guest_mode:1;
-		unsigned :6;
+		unsigned passthrough:1;
+		unsigned :5;
 
 		/*
 		 * This is left at the top of the word so that
@@ -367,8 +368,6 @@ union kvm_mmu_extended_role {
 	struct {
 		unsigned int valid:1;
 		unsigned int execonly:1;
-		unsigned int cr0_pg:1;
-		unsigned int cr4_pae:1;
 		unsigned int cr4_pse:1;
 		unsigned int cr4_pke:1;
 		unsigned int cr4_smap:1;
@@ -378,7 +377,7 @@ union kvm_mmu_extended_role {
 	};
 };
 
-union kvm_mmu_role {
+union kvm_cpu_role {
 	u64 as_u64;
 	struct {
 		union kvm_mmu_page_role base;
@@ -438,19 +437,8 @@ struct kvm_mmu {
 			 struct kvm_mmu_page *sp);
 	void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva, hpa_t root_hpa);
 	struct kvm_mmu_root_info root;
-	union kvm_mmu_role mmu_role;
-	u8 root_level;
-	u8 shadow_root_level;
-	u8 ept_ad;
-	bool direct_map;
-	struct kvm_mmu_root_info prev_roots[KVM_MMU_NUM_PREV_ROOTS];
-
-	/*
-	 * Bitmap; bit set = permission fault
-	 * Byte index: page fault error code [4:1]
-	 * Bit index: pte permissions in ACC_* format
-	 */
-	u8 permissions[16];
+	union kvm_cpu_role cpu_role;
+	union kvm_mmu_page_role root_role;
 
 	/*
 	* The pkru_mask indicates if protection key checks are needed.  It
@@ -460,6 +448,15 @@ struct kvm_mmu {
 	*/
 	u32 pkru_mask;
 
+	struct kvm_mmu_root_info prev_roots[KVM_MMU_NUM_PREV_ROOTS];
+
+	/*
+	 * Bitmap; bit set = permission fault
+	 * Byte index: page fault error code [4:1]
+	 * Bit index: pte permissions in ACC_* format
+	 */
+	u8 permissions[16];
+
 	u64 *pae_root;
 	u64 *pml4_root;
 	u64 *pml5_root;
@@ -607,16 +604,21 @@ struct kvm_vcpu_hv {
 struct kvm_vcpu_xen {
 	u64 hypercall_rip;
 	u32 current_runstate;
-	bool vcpu_info_set;
-	bool vcpu_time_info_set;
-	bool runstate_set;
-	struct gfn_to_hva_cache vcpu_info_cache;
-	struct gfn_to_hva_cache vcpu_time_info_cache;
-	struct gfn_to_hva_cache runstate_cache;
+	u8 upcall_vector;
+	struct gfn_to_pfn_cache vcpu_info_cache;
+	struct gfn_to_pfn_cache vcpu_time_info_cache;
+	struct gfn_to_pfn_cache runstate_cache;
 	u64 last_steal;
 	u64 runstate_entry_time;
 	u64 runstate_times[4];
 	unsigned long evtchn_pending_sel;
+	u32 vcpu_id; /* The Xen / ACPI vCPU ID */
+	u32 timer_virq;
+	u64 timer_expires; /* In guest epoch */
+	atomic_t timer_pending;
+	struct hrtimer timer;
+	int poll_evtchn;
+	struct timer_list poll_timer;
 };
 
 struct kvm_vcpu_arch {
@@ -753,8 +755,7 @@ struct kvm_vcpu_arch {
 	gpa_t time;
 	struct pvclock_vcpu_time_info hv_clock;
 	unsigned int hw_tsc_khz;
-	struct gfn_to_hva_cache pv_time;
-	bool pv_time_enabled;
+	struct gfn_to_pfn_cache pv_time;
 	/* set guest stopped flag in pvclock flags field */
 	bool pvclock_set_guest_stopped_request;
 
@@ -1024,9 +1025,12 @@ struct msr_bitmap_range {
 
 /* Xen emulation context */
 struct kvm_xen {
+	u32 xen_version;
 	bool long_mode;
 	u8 upcall_vector;
 	struct gfn_to_pfn_cache shinfo_cache;
+	struct idr evtchn_ports;
+	unsigned long poll_mask[BITS_TO_LONGS(KVM_MAX_VCPUS)];
 };
 
 enum kvm_irqchip_mode {
@@ -1119,6 +1123,8 @@ struct kvm_arch {
 	u64 cur_tsc_generation;
 	int nr_vcpus_matched_tsc;
 
+	u32 default_tsc_khz;
+
 	seqcount_raw_spinlock_t pvclock_sc;
 	bool use_master_clock;
 	u64 master_kernel_ns;
@@ -1263,7 +1269,12 @@ struct kvm_vm_stat {
 
 struct kvm_vcpu_stat {
 	struct kvm_vcpu_stat_generic generic;
+	u64 pf_taken;
 	u64 pf_fixed;
+	u64 pf_emulate;
+	u64 pf_spurious;
+	u64 pf_fast;
+	u64 pf_mmio_spte_created;
 	u64 pf_guest;
 	u64 tlb_flush;
 	u64 invlpg;
@@ -1455,8 +1466,6 @@ struct kvm_x86_ops {
 	int cpu_dirty_log_size;
 	void (*update_cpu_dirty_logging)(struct kvm_vcpu *vcpu);
 
-	/* pmu operations of sub-arch */
-	const struct kvm_pmu_ops *pmu_ops;
 	const struct kvm_x86_nested_ops *nested_ops;
 
 	void (*vcpu_blocking)(struct kvm_vcpu *vcpu);
@@ -1499,11 +1508,18 @@ struct kvm_x86_ops {
 	int (*complete_emulated_msr)(struct kvm_vcpu *vcpu, int err);
 
 	void (*vcpu_deliver_sipi_vector)(struct kvm_vcpu *vcpu, u8 vector);
+
+	/*
+	 * Returns vCPU specific APICv inhibit reasons
+	 */
+	unsigned long (*vcpu_get_apicv_inhibit_reasons)(struct kvm_vcpu *vcpu);
 };
 
 struct kvm_x86_nested_ops {
 	void (*leave_nested)(struct kvm_vcpu *vcpu);
 	int (*check_events)(struct kvm_vcpu *vcpu);
+	bool (*handle_page_fault_workaround)(struct kvm_vcpu *vcpu,
+					     struct x86_exception *fault);
 	bool (*hv_timer_pending)(struct kvm_vcpu *vcpu);
 	void (*triple_fault)(struct kvm_vcpu *vcpu);
 	int (*get_state)(struct kvm_vcpu *vcpu,
@@ -1528,6 +1544,7 @@ struct kvm_x86_init_ops {
 	unsigned int (*handle_intel_pt_intr)(void);
 
 	struct kvm_x86_ops *runtime_ops;
+	struct kvm_pmu_ops *pmu_ops;
 };
 
 struct kvm_arch_async_pf {
@@ -1549,20 +1566,6 @@ extern struct kvm_x86_ops kvm_x86_ops;
 #define KVM_X86_OP_OPTIONAL_RET0 KVM_X86_OP
 #include <asm/kvm-x86-ops.h>
 
-static inline void kvm_ops_static_call_update(void)
-{
-#define __KVM_X86_OP(func) \
-	static_call_update(kvm_x86_##func, kvm_x86_ops.func);
-#define KVM_X86_OP(func) \
-	WARN_ON(!kvm_x86_ops.func); __KVM_X86_OP(func)
-#define KVM_X86_OP_OPTIONAL __KVM_X86_OP
-#define KVM_X86_OP_OPTIONAL_RET0(func) \
-	static_call_update(kvm_x86_##func, (void *)kvm_x86_ops.func ? : \
-					   (void *)__static_call_return0);
-#include <asm/kvm-x86-ops.h>
-#undef __KVM_X86_OP
-}
-
 #define __KVM_HAVE_ARCH_VM_ALLOC
 static inline struct kvm *kvm_arch_alloc_vm(void)
 {
@@ -1800,6 +1803,7 @@ gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
 				struct x86_exception *exception);
 
 bool kvm_apicv_activated(struct kvm *kvm);
+bool kvm_vcpu_apicv_activated(struct kvm_vcpu *vcpu);
 void kvm_vcpu_update_apicv(struct kvm_vcpu *vcpu);
 void __kvm_set_or_clear_apicv_inhibit(struct kvm *kvm,
 				      enum kvm_apicv_inhibit reason, bool set);
@@ -1989,6 +1993,7 @@ int memslot_rmap_alloc(struct kvm_memory_slot *slot, unsigned long npages);
 	 KVM_X86_QUIRK_CD_NW_CLEARED |		\
 	 KVM_X86_QUIRK_LAPIC_MMIO_HOLE |	\
 	 KVM_X86_QUIRK_OUT_7E_INC_RIP |		\
-	 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT)
+	 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT |	\
+	 KVM_X86_QUIRK_FIX_HYPERCALL_INSN)
 
 #endif /* _ASM_X86_KVM_HOST_H */
diff --git a/arch/x86/include/asm/uaccess.h b/arch/x86/include/asm/uaccess.h
index f78e2b3501a1..35f222aa66bf 100644
--- a/arch/x86/include/asm/uaccess.h
+++ b/arch/x86/include/asm/uaccess.h
@@ -382,6 +382,103 @@ do {									\
 
 #endif // CONFIG_CC_HAS_ASM_GOTO_OUTPUT
 
+#ifdef CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
+#define __try_cmpxchg_user_asm(itype, ltype, _ptr, _pold, _new, label)	({ \
+	bool success;							\
+	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
+	__typeof__(*(_ptr)) __old = *_old;				\
+	__typeof__(*(_ptr)) __new = (_new);				\
+	asm_volatile_goto("\n"						\
+		     "1: " LOCK_PREFIX "cmpxchg"itype" %[new], %[ptr]\n"\
+		     _ASM_EXTABLE_UA(1b, %l[label])			\
+		     : CC_OUT(z) (success),				\
+		       [ptr] "+m" (*_ptr),				\
+		       [old] "+a" (__old)				\
+		     : [new] ltype (__new)				\
+		     : "memory"						\
+		     : label);						\
+	if (unlikely(!success))						\
+		*_old = __old;						\
+	likely(success);					})
+
+#ifdef CONFIG_X86_32
+#define __try_cmpxchg64_user_asm(_ptr, _pold, _new, label)	({	\
+	bool success;							\
+	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
+	__typeof__(*(_ptr)) __old = *_old;				\
+	__typeof__(*(_ptr)) __new = (_new);				\
+	asm_volatile_goto("\n"						\
+		     "1: " LOCK_PREFIX "cmpxchg8b %[ptr]\n"		\
+		     _ASM_EXTABLE_UA(1b, %l[label])			\
+		     : CC_OUT(z) (success),				\
+		       "+A" (__old),					\
+		       [ptr] "+m" (*_ptr)				\
+		     : "b" ((u32)__new),				\
+		       "c" ((u32)((u64)__new >> 32))			\
+		     : "memory"						\
+		     : label);						\
+	if (unlikely(!success))						\
+		*_old = __old;						\
+	likely(success);					})
+#endif // CONFIG_X86_32
+#else  // !CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
+#define __try_cmpxchg_user_asm(itype, ltype, _ptr, _pold, _new, label)	({ \
+	int __err = 0;							\
+	bool success;							\
+	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
+	__typeof__(*(_ptr)) __old = *_old;				\
+	__typeof__(*(_ptr)) __new = (_new);				\
+	asm volatile("\n"						\
+		     "1: " LOCK_PREFIX "cmpxchg"itype" %[new], %[ptr]\n"\
+		     CC_SET(z)						\
+		     "2:\n"						\
+		     _ASM_EXTABLE_TYPE_REG(1b, 2b, EX_TYPE_EFAULT_REG,	\
+					   %[errout])			\
+		     : CC_OUT(z) (success),				\
+		       [errout] "+r" (__err),				\
+		       [ptr] "+m" (*_ptr),				\
+		       [old] "+a" (__old)				\
+		     : [new] ltype (__new)				\
+		     : "memory", "cc");					\
+	if (unlikely(__err))						\
+		goto label;						\
+	if (unlikely(!success))						\
+		*_old = __old;						\
+	likely(success);					})
+
+#ifdef CONFIG_X86_32
+/*
+ * Unlike the normal CMPXCHG, hardcode ECX for both success/fail and error.
+ * There are only six GPRs available and four (EAX, EBX, ECX, and EDX) are
+ * hardcoded by CMPXCHG8B, leaving only ESI and EDI.  If the compiler uses
+ * both ESI and EDI for the memory operand, compilation will fail if the error
+ * is an input+output as there will be no register available for input.
+ */
+#define __try_cmpxchg64_user_asm(_ptr, _pold, _new, label)	({	\
+	int __result;							\
+	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
+	__typeof__(*(_ptr)) __old = *_old;				\
+	__typeof__(*(_ptr)) __new = (_new);				\
+	asm volatile("\n"						\
+		     "1: " LOCK_PREFIX "cmpxchg8b %[ptr]\n"		\
+		     "mov $0, %%ecx\n\t"				\
+		     "setz %%cl\n"					\
+		     "2:\n"						\
+		     _ASM_EXTABLE_TYPE_REG(1b, 2b, EX_TYPE_EFAULT_REG, %%ecx) \
+		     : [result]"=c" (__result),				\
+		       "+A" (__old),					\
+		       [ptr] "+m" (*_ptr)				\
+		     : "b" ((u32)__new),				\
+		       "c" ((u32)((u64)__new >> 32))			\
+		     : "memory", "cc");					\
+	if (unlikely(__result < 0))					\
+		goto label;						\
+	if (unlikely(!__result))					\
+		*_old = __old;						\
+	likely(__result);					})
+#endif // CONFIG_X86_32
+#endif // CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
+
 /* FIXME: this hack is definitely wrong -AK */
 struct __large_struct { unsigned long buf[100]; };
 #define __m(x) (*(struct __large_struct __user *)(x))
@@ -474,6 +571,51 @@ do {										\
 } while (0)
 #endif // CONFIG_CC_HAS_ASM_GOTO_OUTPUT
 
+extern void __try_cmpxchg_user_wrong_size(void);
+
+#ifndef CONFIG_X86_32
+#define __try_cmpxchg64_user_asm(_ptr, _oldp, _nval, _label)		\
+	__try_cmpxchg_user_asm("q", "r", (_ptr), (_oldp), (_nval), _label)
+#endif
+
+/*
+ * Force the pointer to u<size> to match the size expected by the asm helper.
+ * clang/LLVM compiles all cases and only discards the unused paths after
+ * processing errors, which breaks i386 if the pointer is an 8-byte value.
+ */
+#define unsafe_try_cmpxchg_user(_ptr, _oldp, _nval, _label) ({			\
+	bool __ret;								\
+	__chk_user_ptr(_ptr);							\
+	switch (sizeof(*(_ptr))) {						\
+	case 1:	__ret = __try_cmpxchg_user_asm("b", "q",			\
+					       (__force u8 *)(_ptr), (_oldp),	\
+					       (_nval), _label);		\
+		break;								\
+	case 2:	__ret = __try_cmpxchg_user_asm("w", "r",			\
+					       (__force u16 *)(_ptr), (_oldp),	\
+					       (_nval), _label);		\
+		break;								\
+	case 4:	__ret = __try_cmpxchg_user_asm("l", "r",			\
+					       (__force u32 *)(_ptr), (_oldp),	\
+					       (_nval), _label);		\
+		break;								\
+	case 8:	__ret = __try_cmpxchg64_user_asm((__force u64 *)(_ptr), (_oldp),\
+						 (_nval), _label);		\
+		break;								\
+	default: __try_cmpxchg_user_wrong_size();				\
+	}									\
+	__ret;						})
+
+/* "Returns" 0 on success, 1 on failure, -EFAULT if the access faults. */
+#define __try_cmpxchg_user(_ptr, _oldp, _nval, _label)	({		\
+	int __ret = -EFAULT;						\
+	__uaccess_begin_nospec();					\
+	__ret = !unsafe_try_cmpxchg_user(_ptr, _oldp, _nval, _label);	\
+_label:									\
+	__uaccess_end();						\
+	__ret;								\
+							})
+
 /*
  * We want the unsafe accessors to always be inlined and use
  * the error labels - thus the macro games.
diff --git a/arch/x86/include/asm/vmx.h b/arch/x86/include/asm/vmx.h
index 0ffaa3156a4e..6c343c6a1855 100644
--- a/arch/x86/include/asm/vmx.h
+++ b/arch/x86/include/asm/vmx.h
@@ -543,16 +543,14 @@ enum vm_entry_failure_code {
 #define EPT_VIOLATION_ACC_READ_BIT	0
 #define EPT_VIOLATION_ACC_WRITE_BIT	1
 #define EPT_VIOLATION_ACC_INSTR_BIT	2
-#define EPT_VIOLATION_READABLE_BIT	3
-#define EPT_VIOLATION_WRITABLE_BIT	4
-#define EPT_VIOLATION_EXECUTABLE_BIT	5
+#define EPT_VIOLATION_RWX_SHIFT		3
+#define EPT_VIOLATION_GVA_IS_VALID_BIT	7
 #define EPT_VIOLATION_GVA_TRANSLATED_BIT 8
 #define EPT_VIOLATION_ACC_READ		(1 << EPT_VIOLATION_ACC_READ_BIT)
 #define EPT_VIOLATION_ACC_WRITE		(1 << EPT_VIOLATION_ACC_WRITE_BIT)
 #define EPT_VIOLATION_ACC_INSTR		(1 << EPT_VIOLATION_ACC_INSTR_BIT)
-#define EPT_VIOLATION_READABLE		(1 << EPT_VIOLATION_READABLE_BIT)
-#define EPT_VIOLATION_WRITABLE		(1 << EPT_VIOLATION_WRITABLE_BIT)
-#define EPT_VIOLATION_EXECUTABLE	(1 << EPT_VIOLATION_EXECUTABLE_BIT)
+#define EPT_VIOLATION_RWX_MASK		(VMX_EPT_RWX_MASK << EPT_VIOLATION_RWX_SHIFT)
+#define EPT_VIOLATION_GVA_IS_VALID	(1 << EPT_VIOLATION_GVA_IS_VALID_BIT)
 #define EPT_VIOLATION_GVA_TRANSLATED	(1 << EPT_VIOLATION_GVA_TRANSLATED_BIT)
 
 /*
diff --git a/arch/x86/include/uapi/asm/kvm.h b/arch/x86/include/uapi/asm/kvm.h
index bf6e96011dfe..21614807a2cb 100644
--- a/arch/x86/include/uapi/asm/kvm.h
+++ b/arch/x86/include/uapi/asm/kvm.h
@@ -428,11 +428,12 @@ struct kvm_sync_regs {
 	struct kvm_vcpu_events events;
 };
 
-#define KVM_X86_QUIRK_LINT0_REENABLED	   (1 << 0)
-#define KVM_X86_QUIRK_CD_NW_CLEARED	   (1 << 1)
-#define KVM_X86_QUIRK_LAPIC_MMIO_HOLE	   (1 << 2)
-#define KVM_X86_QUIRK_OUT_7E_INC_RIP	   (1 << 3)
-#define KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT (1 << 4)
+#define KVM_X86_QUIRK_LINT0_REENABLED		(1 << 0)
+#define KVM_X86_QUIRK_CD_NW_CLEARED		(1 << 1)
+#define KVM_X86_QUIRK_LAPIC_MMIO_HOLE		(1 << 2)
+#define KVM_X86_QUIRK_OUT_7E_INC_RIP		(1 << 3)
+#define KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT	(1 << 4)
+#define KVM_X86_QUIRK_FIX_HYPERCALL_INSN	(1 << 5)
 
 #define KVM_STATE_NESTED_FORMAT_VMX	0
 #define KVM_STATE_NESTED_FORMAT_SVM	1
diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c
index b14533af7676..9b698215d261 100644
--- a/arch/x86/kernel/asm-offsets_64.c
+++ b/arch/x86/kernel/asm-offsets_64.c
@@ -5,7 +5,7 @@
 
 #include <asm/ia32.h>
 
-#if defined(CONFIG_KVM_GUEST) && defined(CONFIG_PARAVIRT_SPINLOCKS)
+#if defined(CONFIG_KVM_GUEST)
 #include <asm/kvm_para.h>
 #endif
 
@@ -20,7 +20,7 @@ int main(void)
 	BLANK();
 #endif
 
-#if defined(CONFIG_KVM_GUEST) && defined(CONFIG_PARAVIRT_SPINLOCKS)
+#if defined(CONFIG_KVM_GUEST)
 	OFFSET(KVM_STEAL_TIME_preempted, kvm_steal_time, preempted);
 	BLANK();
 #endif
diff --git a/arch/x86/kernel/fpu/core.c b/arch/x86/kernel/fpu/core.c
index e28ab0ecc537..0fdc807ae13f 100644
--- a/arch/x86/kernel/fpu/core.c
+++ b/arch/x86/kernel/fpu/core.c
@@ -14,6 +14,8 @@
 #include <asm/traps.h>
 #include <asm/irq_regs.h>
 
+#include <uapi/asm/kvm.h>
+
 #include <linux/hardirq.h>
 #include <linux/pkeys.h>
 #include <linux/vmalloc.h>
@@ -232,7 +234,20 @@ bool fpu_alloc_guest_fpstate(struct fpu_guest *gfpu)
 	gfpu->fpstate		= fpstate;
 	gfpu->xfeatures		= fpu_user_cfg.default_features;
 	gfpu->perm		= fpu_user_cfg.default_features;
-	gfpu->uabi_size		= fpu_user_cfg.default_size;
+
+	/*
+	 * KVM sets the FP+SSE bits in the XSAVE header when copying FPU state
+	 * to userspace, even when XSAVE is unsupported, so that restoring FPU
+	 * state on a different CPU that does support XSAVE can cleanly load
+	 * the incoming state using its natural XSAVE.  In other words, KVM's
+	 * uABI size may be larger than this host's default size.  Conversely,
+	 * the default size should never be larger than KVM's base uABI size;
+	 * all features that can expand the uABI size must be opt-in.
+	 */
+	gfpu->uabi_size		= sizeof(struct kvm_xsave);
+	if (WARN_ON_ONCE(fpu_user_cfg.default_size > gfpu->uabi_size))
+		gfpu->uabi_size = fpu_user_cfg.default_size;
+
 	fpu_init_guest_permissions(gfpu);
 
 	return true;
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index 8b1c45c9cda8..1a3658f7e6d9 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -191,7 +191,7 @@ void kvm_async_pf_task_wake(u32 token)
 {
 	u32 key = hash_32(token, KVM_TASK_SLEEP_HASHBITS);
 	struct kvm_task_sleep_head *b = &async_pf_sleepers[key];
-	struct kvm_task_sleep_node *n;
+	struct kvm_task_sleep_node *n, *dummy = NULL;
 
 	if (token == ~0) {
 		apf_task_wake_all();
@@ -203,28 +203,41 @@ again:
 	n = _find_apf_task(b, token);
 	if (!n) {
 		/*
-		 * async PF was not yet handled.
-		 * Add dummy entry for the token.
+		 * Async #PF not yet handled, add a dummy entry for the token.
+		 * Allocating the token must be down outside of the raw lock
+		 * as the allocator is preemptible on PREEMPT_RT kernels.
 		 */
-		n = kzalloc(sizeof(*n), GFP_ATOMIC);
-		if (!n) {
+		if (!dummy) {
+			raw_spin_unlock(&b->lock);
+			dummy = kzalloc(sizeof(*dummy), GFP_ATOMIC);
+
 			/*
-			 * Allocation failed! Busy wait while other cpu
-			 * handles async PF.
+			 * Continue looping on allocation failure, eventually
+			 * the async #PF will be handled and allocating a new
+			 * node will be unnecessary.
+			 */
+			if (!dummy)
+				cpu_relax();
+
+			/*
+			 * Recheck for async #PF completion before enqueueing
+			 * the dummy token to avoid duplicate list entries.
 			 */
-			raw_spin_unlock(&b->lock);
-			cpu_relax();
 			goto again;
 		}
-		n->token = token;
-		n->cpu = smp_processor_id();
-		init_swait_queue_head(&n->wq);
-		hlist_add_head(&n->link, &b->list);
+		dummy->token = token;
+		dummy->cpu = smp_processor_id();
+		init_swait_queue_head(&dummy->wq);
+		hlist_add_head(&dummy->link, &b->list);
+		dummy = NULL;
 	} else {
 		apf_task_wake_one(n);
 	}
 	raw_spin_unlock(&b->lock);
-	return;
+
+	/* A dummy token might be allocated and ultimately not used.  */
+	if (dummy)
+		kfree(dummy);
 }
 EXPORT_SYMBOL_GPL(kvm_async_pf_task_wake);
 
@@ -765,6 +778,42 @@ static void kvm_crash_shutdown(struct pt_regs *regs)
 }
 #endif
 
+#if defined(CONFIG_X86_32) || !defined(CONFIG_SMP)
+bool __kvm_vcpu_is_preempted(long cpu);
+
+__visible bool __kvm_vcpu_is_preempted(long cpu)
+{
+	struct kvm_steal_time *src = &per_cpu(steal_time, cpu);
+
+	return !!(src->preempted & KVM_VCPU_PREEMPTED);
+}
+PV_CALLEE_SAVE_REGS_THUNK(__kvm_vcpu_is_preempted);
+
+#else
+
+#include <asm/asm-offsets.h>
+
+extern bool __raw_callee_save___kvm_vcpu_is_preempted(long);
+
+/*
+ * Hand-optimize version for x86-64 to avoid 8 64-bit register saving and
+ * restoring to/from the stack.
+ */
+asm(
+".pushsection .text;"
+".global __raw_callee_save___kvm_vcpu_is_preempted;"
+".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
+"__raw_callee_save___kvm_vcpu_is_preempted:"
+ASM_ENDBR
+"movq	__per_cpu_offset(,%rdi,8), %rax;"
+"cmpb	$0, " __stringify(KVM_STEAL_TIME_preempted) "+steal_time(%rax);"
+"setne	%al;"
+ASM_RET
+".size __raw_callee_save___kvm_vcpu_is_preempted, .-__raw_callee_save___kvm_vcpu_is_preempted;"
+".popsection");
+
+#endif
+
 static void __init kvm_guest_init(void)
 {
 	int i;
@@ -777,6 +826,9 @@ static void __init kvm_guest_init(void)
 	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
 		has_steal_clock = 1;
 		static_call_update(pv_steal_clock, kvm_steal_clock);
+
+		pv_ops.lock.vcpu_is_preempted =
+			PV_CALLEE_SAVE(__kvm_vcpu_is_preempted);
 	}
 
 	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
@@ -1018,40 +1070,6 @@ static void kvm_wait(u8 *ptr, u8 val)
 	}
 }
 
-#ifdef CONFIG_X86_32
-__visible bool __kvm_vcpu_is_preempted(long cpu)
-{
-	struct kvm_steal_time *src = &per_cpu(steal_time, cpu);
-
-	return !!(src->preempted & KVM_VCPU_PREEMPTED);
-}
-PV_CALLEE_SAVE_REGS_THUNK(__kvm_vcpu_is_preempted);
-
-#else
-
-#include <asm/asm-offsets.h>
-
-extern bool __raw_callee_save___kvm_vcpu_is_preempted(long);
-
-/*
- * Hand-optimize version for x86-64 to avoid 8 64-bit register saving and
- * restoring to/from the stack.
- */
-asm(
-".pushsection .text;"
-".global __raw_callee_save___kvm_vcpu_is_preempted;"
-".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
-"__raw_callee_save___kvm_vcpu_is_preempted:"
-ASM_ENDBR
-"movq	__per_cpu_offset(,%rdi,8), %rax;"
-"cmpb	$0, " __stringify(KVM_STEAL_TIME_preempted) "+steal_time(%rax);"
-"setne	%al;"
-ASM_RET
-".size __raw_callee_save___kvm_vcpu_is_preempted, .-__raw_callee_save___kvm_vcpu_is_preempted;"
-".popsection");
-
-#endif
-
 /*
  * Setup pv_lock_ops to exploit KVM_FEATURE_PV_UNHALT if present.
  */
@@ -1095,10 +1113,6 @@ void __init kvm_spinlock_init(void)
 	pv_ops.lock.wait = kvm_wait;
 	pv_ops.lock.kick = kvm_kick_cpu;
 
-	if (kvm_para_has_feature(KVM_FEATURE_STEAL_TIME)) {
-		pv_ops.lock.vcpu_is_preempted =
-			PV_CALLEE_SAVE(__kvm_vcpu_is_preempted);
-	}
 	/*
 	 * When PV spinlock is enabled which is preferred over
 	 * virt_spin_lock(), virt_spin_lock_key's value is meaningless.
diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c
index c5caa7311bd8..16333ba1904b 100644
--- a/arch/x86/kernel/kvmclock.c
+++ b/arch/x86/kernel/kvmclock.c
@@ -239,7 +239,7 @@ static void __init kvmclock_init_mem(void)
 
 static int __init kvm_setup_vsyscall_timeinfo(void)
 {
-	if (!kvm_para_available() || !kvmclock)
+	if (!kvm_para_available() || !kvmclock || nopv)
 		return 0;
 
 	kvmclock_init_mem();
diff --git a/arch/x86/kvm/i8259.c b/arch/x86/kvm/i8259.c
index be99dc86293d..e1bb6218bb96 100644
--- a/arch/x86/kvm/i8259.c
+++ b/arch/x86/kvm/i8259.c
@@ -252,7 +252,6 @@ int kvm_pic_read_irq(struct kvm *kvm)
 				 */
 				irq2 = 7;
 			intno = s->pics[1].irq_base + irq2;
-			irq = irq2 + 8;
 		} else
 			intno = s->pics[0].irq_base + irq;
 	} else {
diff --git a/arch/x86/kvm/irq.c b/arch/x86/kvm/irq.c
index 172b05343cfd..f371f1292ca3 100644
--- a/arch/x86/kvm/irq.c
+++ b/arch/x86/kvm/irq.c
@@ -22,10 +22,14 @@
  */
 int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
 {
+	int r = 0;
+
 	if (lapic_in_kernel(vcpu))
-		return apic_has_pending_timer(vcpu);
+		r = apic_has_pending_timer(vcpu);
+	if (kvm_xen_timer_enabled(vcpu))
+		r += kvm_xen_has_pending_timer(vcpu);
 
-	return 0;
+	return r;
 }
 EXPORT_SYMBOL(kvm_cpu_has_pending_timer);
 
@@ -143,6 +147,8 @@ void kvm_inject_pending_timer_irqs(struct kvm_vcpu *vcpu)
 {
 	if (lapic_in_kernel(vcpu))
 		kvm_inject_apic_timer_irqs(vcpu);
+	if (kvm_xen_timer_enabled(vcpu))
+		kvm_xen_inject_timer_irqs(vcpu);
 }
 EXPORT_SYMBOL_GPL(kvm_inject_pending_timer_irqs);
 
diff --git a/arch/x86/kvm/irq_comm.c b/arch/x86/kvm/irq_comm.c
index 6e0dab04320e..0687162c4f22 100644
--- a/arch/x86/kvm/irq_comm.c
+++ b/arch/x86/kvm/irq_comm.c
@@ -181,7 +181,7 @@ int kvm_arch_set_irq_inatomic(struct kvm_kernel_irq_routing_entry *e,
 		if (!level)
 			return -1;
 
-		return kvm_xen_set_evtchn_fast(e, kvm);
+		return kvm_xen_set_evtchn_fast(&e->xen_evtchn, kvm);
 #endif
 	default:
 		break;
diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index 66b0eb0bda94..f1bdac3f5aa8 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -1548,6 +1548,7 @@ static void cancel_apic_timer(struct kvm_lapic *apic)
 	if (apic->lapic_timer.hv_timer_in_use)
 		cancel_hv_timer(apic);
 	preempt_enable();
+	atomic_set(&apic->lapic_timer.pending, 0);
 }
 
 static void apic_update_lvtt(struct kvm_lapic *apic)
@@ -1648,10 +1649,10 @@ static void __kvm_wait_lapic_expire(struct kvm_vcpu *vcpu)
 	tsc_deadline = apic->lapic_timer.expired_tscdeadline;
 	apic->lapic_timer.expired_tscdeadline = 0;
 	guest_tsc = kvm_read_l1_tsc(vcpu, rdtsc());
-	apic->lapic_timer.advance_expire_delta = guest_tsc - tsc_deadline;
+	trace_kvm_wait_lapic_expire(vcpu->vcpu_id, guest_tsc - tsc_deadline);
 
 	if (lapic_timer_advance_dynamic) {
-		adjust_lapic_timer_advance(vcpu, apic->lapic_timer.advance_expire_delta);
+		adjust_lapic_timer_advance(vcpu, guest_tsc - tsc_deadline);
 		/*
 		 * If the timer fired early, reread the TSC to account for the
 		 * overhead of the above adjustment to avoid waiting longer
diff --git a/arch/x86/kvm/lapic.h b/arch/x86/kvm/lapic.h
index 4e4f8a22754f..65bb2a8cf145 100644
--- a/arch/x86/kvm/lapic.h
+++ b/arch/x86/kvm/lapic.h
@@ -38,7 +38,6 @@ struct kvm_timer {
 	u64 tscdeadline;
 	u64 expired_tscdeadline;
 	u32 timer_advance_ns;
-	s64 advance_expire_delta;
 	atomic_t pending;			/* accumulated triggered timers */
 	bool hv_timer_in_use;
 };
diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
index a335e7f1f69e..f8192864b496 100644
--- a/arch/x86/kvm/mmu.h
+++ b/arch/x86/kvm/mmu.h
@@ -89,7 +89,27 @@ static inline gfn_t kvm_mmu_max_gfn(void)
 	return (1ULL << (max_gpa_bits - PAGE_SHIFT)) - 1;
 }
 
+static inline u8 kvm_get_shadow_phys_bits(void)
+{
+	/*
+	 * boot_cpu_data.x86_phys_bits is reduced when MKTME or SME are detected
+	 * in CPU detection code, but the processor treats those reduced bits as
+	 * 'keyID' thus they are not reserved bits. Therefore KVM needs to look at
+	 * the physical address bits reported by CPUID.
+	 */
+	if (likely(boot_cpu_data.extended_cpuid_level >= 0x80000008))
+		return cpuid_eax(0x80000008) & 0xff;
+
+	/*
+	 * Quite weird to have VMX or SVM but not MAXPHYADDR; probably a VM with
+	 * custom CPUID.  Proceed with whatever the kernel found since these features
+	 * aren't virtualizable (SME/SEV also require CPUIDs higher than 0x80000008).
+	 */
+	return boot_cpu_data.x86_phys_bits;
+}
+
 void kvm_mmu_set_mmio_spte_mask(u64 mmio_value, u64 mmio_mask, u64 access_mask);
+void kvm_mmu_set_me_spte_mask(u64 me_value, u64 me_mask);
 void kvm_mmu_set_ept_masks(bool has_ad_bits, bool has_exec_only);
 
 void kvm_init_mmu(struct kvm_vcpu *vcpu);
@@ -138,94 +158,7 @@ static inline void kvm_mmu_load_pgd(struct kvm_vcpu *vcpu)
 		return;
 
 	static_call(kvm_x86_load_mmu_pgd)(vcpu, root_hpa,
-					  vcpu->arch.mmu->shadow_root_level);
-}
-
-struct kvm_page_fault {
-	/* arguments to kvm_mmu_do_page_fault.  */
-	const gpa_t addr;
-	const u32 error_code;
-	const bool prefetch;
-
-	/* Derived from error_code.  */
-	const bool exec;
-	const bool write;
-	const bool present;
-	const bool rsvd;
-	const bool user;
-
-	/* Derived from mmu and global state.  */
-	const bool is_tdp;
-	const bool nx_huge_page_workaround_enabled;
-
-	/*
-	 * Whether a >4KB mapping can be created or is forbidden due to NX
-	 * hugepages.
-	 */
-	bool huge_page_disallowed;
-
-	/*
-	 * Maximum page size that can be created for this fault; input to
-	 * FNAME(fetch), __direct_map and kvm_tdp_mmu_map.
-	 */
-	u8 max_level;
-
-	/*
-	 * Page size that can be created based on the max_level and the
-	 * page size used by the host mapping.
-	 */
-	u8 req_level;
-
-	/*
-	 * Page size that will be created based on the req_level and
-	 * huge_page_disallowed.
-	 */
-	u8 goal_level;
-
-	/* Shifted addr, or result of guest page table walk if addr is a gva.  */
-	gfn_t gfn;
-
-	/* The memslot containing gfn. May be NULL. */
-	struct kvm_memory_slot *slot;
-
-	/* Outputs of kvm_faultin_pfn.  */
-	kvm_pfn_t pfn;
-	hva_t hva;
-	bool map_writable;
-};
-
-int kvm_tdp_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);
-
-extern int nx_huge_pages;
-static inline bool is_nx_huge_page_enabled(void)
-{
-	return READ_ONCE(nx_huge_pages);
-}
-
-static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
-					u32 err, bool prefetch)
-{
-	struct kvm_page_fault fault = {
-		.addr = cr2_or_gpa,
-		.error_code = err,
-		.exec = err & PFERR_FETCH_MASK,
-		.write = err & PFERR_WRITE_MASK,
-		.present = err & PFERR_PRESENT_MASK,
-		.rsvd = err & PFERR_RSVD_MASK,
-		.user = err & PFERR_USER_MASK,
-		.prefetch = prefetch,
-		.is_tdp = likely(vcpu->arch.mmu->page_fault == kvm_tdp_page_fault),
-		.nx_huge_page_workaround_enabled = is_nx_huge_page_enabled(),
-
-		.max_level = KVM_MAX_HUGEPAGE_LEVEL,
-		.req_level = PG_LEVEL_4K,
-		.goal_level = PG_LEVEL_4K,
-	};
-#ifdef CONFIG_RETPOLINE
-	if (fault.is_tdp)
-		return kvm_tdp_page_fault(vcpu, &fault);
-#endif
-	return vcpu->arch.mmu->page_fault(vcpu, &fault);
+					  vcpu->arch.mmu->root_role.level);
 }
 
 /*
diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index 45e1573f8f1d..f4653688fa6d 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -193,11 +193,12 @@ struct kvm_mmu_role_regs {
 
 /*
  * Yes, lot's of underscores.  They're a hint that you probably shouldn't be
- * reading from the role_regs.  Once the mmu_role is constructed, it becomes
+ * reading from the role_regs.  Once the root_role is constructed, it becomes
  * the single source of truth for the MMU's state.
  */
 #define BUILD_MMU_ROLE_REGS_ACCESSOR(reg, name, flag)			\
-static inline bool __maybe_unused ____is_##reg##_##name(struct kvm_mmu_role_regs *regs)\
+static inline bool __maybe_unused					\
+____is_##reg##_##name(const struct kvm_mmu_role_regs *regs)		\
 {									\
 	return !!(regs->reg & flag);					\
 }
@@ -221,17 +222,26 @@ BUILD_MMU_ROLE_REGS_ACCESSOR(efer, lma, EFER_LMA);
 #define BUILD_MMU_ROLE_ACCESSOR(base_or_ext, reg, name)		\
 static inline bool __maybe_unused is_##reg##_##name(struct kvm_mmu *mmu)	\
 {								\
-	return !!(mmu->mmu_role. base_or_ext . reg##_##name);	\
+	return !!(mmu->cpu_role. base_or_ext . reg##_##name);	\
 }
-BUILD_MMU_ROLE_ACCESSOR(ext,  cr0, pg);
 BUILD_MMU_ROLE_ACCESSOR(base, cr0, wp);
 BUILD_MMU_ROLE_ACCESSOR(ext,  cr4, pse);
-BUILD_MMU_ROLE_ACCESSOR(ext,  cr4, pae);
 BUILD_MMU_ROLE_ACCESSOR(ext,  cr4, smep);
 BUILD_MMU_ROLE_ACCESSOR(ext,  cr4, smap);
 BUILD_MMU_ROLE_ACCESSOR(ext,  cr4, pke);
 BUILD_MMU_ROLE_ACCESSOR(ext,  cr4, la57);
 BUILD_MMU_ROLE_ACCESSOR(base, efer, nx);
+BUILD_MMU_ROLE_ACCESSOR(ext,  efer, lma);
+
+static inline bool is_cr0_pg(struct kvm_mmu *mmu)
+{
+        return mmu->cpu_role.base.level > 0;
+}
+
+static inline bool is_cr4_pae(struct kvm_mmu *mmu)
+{
+        return !mmu->cpu_role.base.has_4_byte_gpte;
+}
 
 static struct kvm_mmu_role_regs vcpu_to_role_regs(struct kvm_vcpu *vcpu)
 {
@@ -244,19 +254,6 @@ static struct kvm_mmu_role_regs vcpu_to_role_regs(struct kvm_vcpu *vcpu)
 	return regs;
 }
 
-static int role_regs_to_root_level(struct kvm_mmu_role_regs *regs)
-{
-	if (!____is_cr0_pg(regs))
-		return 0;
-	else if (____is_efer_lma(regs))
-		return ____is_cr4_la57(regs) ? PT64_ROOT_5LEVEL :
-					       PT64_ROOT_4LEVEL;
-	else if (____is_cr4_pae(regs))
-		return PT32E_ROOT_LEVEL;
-	else
-		return PT32_ROOT_LEVEL;
-}
-
 static inline bool kvm_available_flush_tlb_with_range(void)
 {
 	return kvm_x86_ops.tlb_remote_flush_with_range;
@@ -714,6 +711,9 @@ static void mmu_free_pte_list_desc(struct pte_list_desc *pte_list_desc)
 
 static gfn_t kvm_mmu_page_get_gfn(struct kvm_mmu_page *sp, int index)
 {
+	if (sp->role.passthrough)
+		return sp->gfn;
+
 	if (!sp->role.direct)
 		return sp->gfns[index];
 
@@ -722,6 +722,11 @@ static gfn_t kvm_mmu_page_get_gfn(struct kvm_mmu_page *sp, int index)
 
 static void kvm_mmu_page_set_gfn(struct kvm_mmu_page *sp, int index, gfn_t gfn)
 {
+	if (sp->role.passthrough) {
+		WARN_ON_ONCE(gfn != sp->gfn);
+		return;
+	}
+
 	if (!sp->role.direct) {
 		sp->gfns[index] = gfn;
 		return;
@@ -1478,9 +1483,11 @@ static bool slot_rmap_walk_okay(struct slot_rmap_walk_iterator *iterator)
 
 static void slot_rmap_walk_next(struct slot_rmap_walk_iterator *iterator)
 {
-	if (++iterator->rmap <= iterator->end_rmap) {
+	while (++iterator->rmap <= iterator->end_rmap) {
 		iterator->gfn += (1UL << KVM_HPAGE_GFN_SHIFT(iterator->level));
-		return;
+
+		if (iterator->rmap->val)
+			return;
 	}
 
 	if (++iterator->level > iterator->end_level) {
@@ -1833,27 +1840,35 @@ static bool kvm_mmu_prepare_zap_page(struct kvm *kvm, struct kvm_mmu_page *sp,
 static void kvm_mmu_commit_zap_page(struct kvm *kvm,
 				    struct list_head *invalid_list);
 
+static bool sp_has_gptes(struct kvm_mmu_page *sp)
+{
+	if (sp->role.direct)
+		return false;
+
+	if (sp->role.passthrough)
+		return false;
+
+	return true;
+}
+
 #define for_each_valid_sp(_kvm, _sp, _list)				\
 	hlist_for_each_entry(_sp, _list, hash_link)			\
 		if (is_obsolete_sp((_kvm), (_sp))) {			\
 		} else
 
-#define for_each_gfn_indirect_valid_sp(_kvm, _sp, _gfn)			\
+#define for_each_gfn_valid_sp_with_gptes(_kvm, _sp, _gfn)		\
 	for_each_valid_sp(_kvm, _sp,					\
 	  &(_kvm)->arch.mmu_page_hash[kvm_page_table_hashfn(_gfn)])	\
-		if ((_sp)->gfn != (_gfn) || (_sp)->role.direct) {} else
+		if ((_sp)->gfn != (_gfn) || !sp_has_gptes(_sp)) {} else
 
-static bool kvm_sync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
+static int kvm_sync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
 			 struct list_head *invalid_list)
 {
 	int ret = vcpu->arch.mmu->sync_page(vcpu, sp);
 
-	if (ret < 0) {
+	if (ret < 0)
 		kvm_mmu_prepare_zap_page(vcpu->kvm, sp, invalid_list);
-		return false;
-	}
-
-	return !!ret;
+	return ret;
 }
 
 static bool kvm_mmu_remote_flush_or_zap(struct kvm *kvm,
@@ -1975,7 +1990,7 @@ static int mmu_sync_children(struct kvm_vcpu *vcpu,
 
 		for_each_sp(pages, sp, parents, i) {
 			kvm_unlink_unsync_page(vcpu->kvm, sp);
-			flush |= kvm_sync_page(vcpu, sp, &invalid_list);
+			flush |= kvm_sync_page(vcpu, sp, &invalid_list) > 0;
 			mmu_pages_clear_parents(&parents);
 		}
 		if (need_resched() || rwlock_needbreak(&vcpu->kvm->mmu_lock)) {
@@ -2011,15 +2026,16 @@ static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
 					     int direct,
 					     unsigned int access)
 {
-	bool direct_mmu = vcpu->arch.mmu->direct_map;
+	bool direct_mmu = vcpu->arch.mmu->root_role.direct;
 	union kvm_mmu_page_role role;
 	struct hlist_head *sp_list;
 	unsigned quadrant;
 	struct kvm_mmu_page *sp;
+	int ret;
 	int collisions = 0;
 	LIST_HEAD(invalid_list);
 
-	role = vcpu->arch.mmu->mmu_role.base;
+	role = vcpu->arch.mmu->root_role;
 	role.level = level;
 	role.direct = direct;
 	role.access = access;
@@ -2028,6 +2044,8 @@ static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
 		quadrant &= (1 << ((PT32_PT_BITS - PT64_PT_BITS) * level)) - 1;
 		role.quadrant = quadrant;
 	}
+	if (level <= vcpu->arch.mmu->cpu_role.base.level)
+		role.passthrough = 0;
 
 	sp_list = &vcpu->kvm->arch.mmu_page_hash[kvm_page_table_hashfn(gfn)];
 	for_each_valid_sp(vcpu->kvm, sp, sp_list) {
@@ -2068,11 +2086,13 @@ static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
 			 * If the sync fails, the page is zapped.  If so, break
 			 * in order to rebuild it.
 			 */
-			if (!kvm_sync_page(vcpu, sp, &invalid_list))
+			ret = kvm_sync_page(vcpu, sp, &invalid_list);
+			if (ret < 0)
 				break;
 
 			WARN_ON(!list_empty(&invalid_list));
-			kvm_flush_remote_tlbs(vcpu->kvm);
+			if (ret > 0)
+				kvm_flush_remote_tlbs(vcpu->kvm);
 		}
 
 		__clear_sp_write_flooding_count(sp);
@@ -2089,7 +2109,7 @@ trace_get_page:
 	sp->gfn = gfn;
 	sp->role = role;
 	hlist_add_head(&sp->hash_link, sp_list);
-	if (!direct) {
+	if (sp_has_gptes(sp)) {
 		account_shadowed(vcpu->kvm, sp);
 		if (level == PG_LEVEL_4K && kvm_vcpu_write_protect_gfn(vcpu, gfn))
 			kvm_flush_remote_tlbs_with_address(vcpu->kvm, gfn, 1);
@@ -2109,11 +2129,11 @@ static void shadow_walk_init_using_root(struct kvm_shadow_walk_iterator *iterato
 {
 	iterator->addr = addr;
 	iterator->shadow_addr = root;
-	iterator->level = vcpu->arch.mmu->shadow_root_level;
+	iterator->level = vcpu->arch.mmu->root_role.level;
 
 	if (iterator->level >= PT64_ROOT_4LEVEL &&
-	    vcpu->arch.mmu->root_level < PT64_ROOT_4LEVEL &&
-	    !vcpu->arch.mmu->direct_map)
+	    vcpu->arch.mmu->cpu_role.base.level < PT64_ROOT_4LEVEL &&
+	    !vcpu->arch.mmu->root_role.direct)
 		iterator->level = PT32E_ROOT_LEVEL;
 
 	if (iterator->level == PT32E_ROOT_LEVEL) {
@@ -2298,7 +2318,7 @@ static bool __kvm_mmu_prepare_zap_page(struct kvm *kvm,
 	/* Zapping children means active_mmu_pages has become unstable. */
 	list_unstable = *nr_zapped;
 
-	if (!sp->role.invalid && !sp->role.direct)
+	if (!sp->role.invalid && sp_has_gptes(sp))
 		unaccount_shadowed(kvm, sp);
 
 	if (sp->unsync)
@@ -2478,7 +2498,7 @@ int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn)
 	pgprintk("%s: looking for gfn %llx\n", __func__, gfn);
 	r = 0;
 	write_lock(&kvm->mmu_lock);
-	for_each_gfn_indirect_valid_sp(kvm, sp, gfn) {
+	for_each_gfn_valid_sp_with_gptes(kvm, sp, gfn) {
 		pgprintk("%s: gfn %llx role %x\n", __func__, gfn,
 			 sp->role.word);
 		r = 1;
@@ -2495,7 +2515,7 @@ static int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva)
 	gpa_t gpa;
 	int r;
 
-	if (vcpu->arch.mmu->direct_map)
+	if (vcpu->arch.mmu->root_role.direct)
 		return 0;
 
 	gpa = kvm_mmu_gva_to_gpa_read(vcpu, gva, NULL);
@@ -2540,7 +2560,7 @@ int mmu_try_to_unsync_pages(struct kvm *kvm, const struct kvm_memory_slot *slot,
 	 * that case, KVM must complete emulation of the guest TLB flush before
 	 * allowing shadow pages to become unsync (writable by the guest).
 	 */
-	for_each_gfn_indirect_valid_sp(kvm, sp, gfn) {
+	for_each_gfn_valid_sp_with_gptes(kvm, sp, gfn) {
 		if (!can_unsync)
 			return -EPERM;
 
@@ -2642,6 +2662,7 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, struct kvm_memory_slot *slot,
 		 *sptep, write_fault, gfn);
 
 	if (unlikely(is_noslot_pfn(pfn))) {
+		vcpu->stat.pf_mmio_spte_created++;
 		mark_mmio_spte(vcpu, sptep, gfn, pte_access);
 		return RET_PF_EMULATE;
 	}
@@ -2962,7 +2983,6 @@ static int __direct_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 		return ret;
 
 	direct_pte_prefetch(vcpu, it.sptep);
-	++vcpu->stat.pf_fixed;
 	return ret;
 }
 
@@ -2989,14 +3009,12 @@ static int kvm_handle_bad_page(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn)
 	return -EFAULT;
 }
 
-static bool handle_abnormal_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
-				unsigned int access, int *ret_val)
+static int handle_abnormal_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
+			       unsigned int access)
 {
 	/* The pfn is invalid, report the error! */
-	if (unlikely(is_error_pfn(fault->pfn))) {
-		*ret_val = kvm_handle_bad_page(vcpu, fault->gfn, fault->pfn);
-		return true;
-	}
+	if (unlikely(is_error_pfn(fault->pfn)))
+		return kvm_handle_bad_page(vcpu, fault->gfn, fault->pfn);
 
 	if (unlikely(!fault->slot)) {
 		gva_t gva = fault->is_tdp ? 0 : fault->addr;
@@ -3013,44 +3031,48 @@ static bool handle_abnormal_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fa
 		 * and only if L1's MAXPHYADDR is inaccurate with respect to
 		 * the hardware's).
 		 */
-		if (unlikely(!shadow_mmio_value) ||
-		    unlikely(fault->gfn > kvm_mmu_max_gfn())) {
-			*ret_val = RET_PF_EMULATE;
-			return true;
-		}
+		if (unlikely(!enable_mmio_caching) ||
+		    unlikely(fault->gfn > kvm_mmu_max_gfn()))
+			return RET_PF_EMULATE;
 	}
 
-	return false;
+	return RET_PF_CONTINUE;
 }
 
 static bool page_fault_can_be_fast(struct kvm_page_fault *fault)
 {
 	/*
-	 * Do not fix the mmio spte with invalid generation number which
-	 * need to be updated by slow page fault path.
+	 * Page faults with reserved bits set, i.e. faults on MMIO SPTEs, only
+	 * reach the common page fault handler if the SPTE has an invalid MMIO
+	 * generation number.  Refreshing the MMIO generation needs to go down
+	 * the slow path.  Note, EPT Misconfigs do NOT set the PRESENT flag!
 	 */
 	if (fault->rsvd)
 		return false;
 
-	/* See if the page fault is due to an NX violation */
-	if (unlikely(fault->exec && fault->present))
-		return false;
-
 	/*
 	 * #PF can be fast if:
-	 * 1. The shadow page table entry is not present, which could mean that
-	 *    the fault is potentially caused by access tracking (if enabled).
-	 * 2. The shadow page table entry is present and the fault
-	 *    is caused by write-protect, that means we just need change the W
-	 *    bit of the spte which can be done out of mmu-lock.
 	 *
-	 * However, if access tracking is disabled we know that a non-present
-	 * page must be a genuine page fault where we have to create a new SPTE.
-	 * So, if access tracking is disabled, we return true only for write
-	 * accesses to a present page.
+	 * 1. The shadow page table entry is not present and A/D bits are
+	 *    disabled _by KVM_, which could mean that the fault is potentially
+	 *    caused by access tracking (if enabled).  If A/D bits are enabled
+	 *    by KVM, but disabled by L1 for L2, KVM is forced to disable A/D
+	 *    bits for L2 and employ access tracking, but the fast page fault
+	 *    mechanism only supports direct MMUs.
+	 * 2. The shadow page table entry is present, the access is a write,
+	 *    and no reserved bits are set (MMIO SPTEs cannot be "fixed"), i.e.
+	 *    the fault was caused by a write-protection violation.  If the
+	 *    SPTE is MMU-writable (determined later), the fault can be fixed
+	 *    by setting the Writable bit, which can be done out of mmu_lock.
 	 */
+	if (!fault->present)
+		return !kvm_ad_enabled();
 
-	return shadow_acc_track_mask != 0 || (fault->write && fault->present);
+	/*
+	 * Note, instruction fetches and writes are mutually exclusive, ignore
+	 * the "exec" flag.
+	 */
+	return fault->write;
 }
 
 /*
@@ -3165,13 +3187,25 @@ static int fast_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 
 		new_spte = spte;
 
-		if (is_access_track_spte(spte))
+		/*
+		 * KVM only supports fixing page faults outside of MMU lock for
+		 * direct MMUs, nested MMUs are always indirect, and KVM always
+		 * uses A/D bits for non-nested MMUs.  Thus, if A/D bits are
+		 * enabled, the SPTE can't be an access-tracked SPTE.
+		 */
+		if (unlikely(!kvm_ad_enabled()) && is_access_track_spte(spte))
 			new_spte = restore_acc_track_spte(new_spte);
 
 		/*
-		 * Currently, to simplify the code, write-protection can
-		 * be removed in the fast path only if the SPTE was
-		 * write-protected for dirty-logging or access tracking.
+		 * To keep things simple, only SPTEs that are MMU-writable can
+		 * be made fully writable outside of mmu_lock, e.g. only SPTEs
+		 * that were write-protected for dirty-logging or access
+		 * tracking are handled here.  Don't bother checking if the
+		 * SPTE is writable to prioritize running with A/D bits enabled.
+		 * The is_access_allowed() check above handles the common case
+		 * of the fault being spurious, and the SPTE is known to be
+		 * shadow-present, i.e. except for access tracking restoration
+		 * making the new SPTE writable, the check is wasteful.
 		 */
 		if (fault->write && is_mmu_writable_spte(spte)) {
 			new_spte |= PT_WRITABLE_MASK;
@@ -3217,6 +3251,9 @@ static int fast_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 	trace_fast_page_fault(vcpu, fault, sptep, spte, ret);
 	walk_shadow_page_lockless_end(vcpu);
 
+	if (ret != RET_PF_INVALID)
+		vcpu->stat.pf_fast++;
+
 	return ret;
 }
 
@@ -3303,7 +3340,7 @@ void kvm_mmu_free_guest_mode_roots(struct kvm *kvm, struct kvm_mmu *mmu)
 	 * This should not be called while L2 is active, L2 can't invalidate
 	 * _only_ its own roots, e.g. INVVPID unconditionally exits.
 	 */
-	WARN_ON_ONCE(mmu->mmu_role.base.guest_mode);
+	WARN_ON_ONCE(mmu->root_role.guest_mode);
 
 	for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++) {
 		root_hpa = mmu->prev_roots[i].hpa;
@@ -3346,7 +3383,7 @@ static hpa_t mmu_alloc_root(struct kvm_vcpu *vcpu, gfn_t gfn, gva_t gva,
 static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
 {
 	struct kvm_mmu *mmu = vcpu->arch.mmu;
-	u8 shadow_root_level = mmu->shadow_root_level;
+	u8 shadow_root_level = mmu->root_role.level;
 	hpa_t root;
 	unsigned i;
 	int r;
@@ -3470,7 +3507,7 @@ static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)
 	 * On SVM, reading PDPTRs might access guest memory, which might fault
 	 * and thus might sleep.  Grab the PDPTRs before acquiring mmu_lock.
 	 */
-	if (mmu->root_level == PT32E_ROOT_LEVEL) {
+	if (mmu->cpu_role.base.level == PT32E_ROOT_LEVEL) {
 		for (i = 0; i < 4; ++i) {
 			pdptrs[i] = mmu->get_pdptr(vcpu, i);
 			if (!(pdptrs[i] & PT_PRESENT_MASK))
@@ -3494,9 +3531,9 @@ static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)
 	 * Do we shadow a long mode page table? If so we need to
 	 * write-protect the guests page table root.
 	 */
-	if (mmu->root_level >= PT64_ROOT_4LEVEL) {
+	if (mmu->cpu_role.base.level >= PT64_ROOT_4LEVEL) {
 		root = mmu_alloc_root(vcpu, root_gfn, 0,
-				      mmu->shadow_root_level, false);
+				      mmu->root_role.level, false);
 		mmu->root.hpa = root;
 		goto set_root_pgd;
 	}
@@ -3511,8 +3548,8 @@ static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)
 	 * or a PAE 3-level page table. In either case we need to be aware that
 	 * the shadow page table may be a PAE or a long mode page table.
 	 */
-	pm_mask = PT_PRESENT_MASK | shadow_me_mask;
-	if (mmu->shadow_root_level >= PT64_ROOT_4LEVEL) {
+	pm_mask = PT_PRESENT_MASK | shadow_me_value;
+	if (mmu->root_role.level >= PT64_ROOT_4LEVEL) {
 		pm_mask |= PT_ACCESSED_MASK | PT_WRITABLE_MASK | PT_USER_MASK;
 
 		if (WARN_ON_ONCE(!mmu->pml4_root)) {
@@ -3521,7 +3558,7 @@ static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)
 		}
 		mmu->pml4_root[0] = __pa(mmu->pae_root) | pm_mask;
 
-		if (mmu->shadow_root_level == PT64_ROOT_5LEVEL) {
+		if (mmu->root_role.level == PT64_ROOT_5LEVEL) {
 			if (WARN_ON_ONCE(!mmu->pml5_root)) {
 				r = -EIO;
 				goto out_unlock;
@@ -3533,7 +3570,7 @@ static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)
 	for (i = 0; i < 4; ++i) {
 		WARN_ON_ONCE(IS_VALID_PAE_ROOT(mmu->pae_root[i]));
 
-		if (mmu->root_level == PT32E_ROOT_LEVEL) {
+		if (mmu->cpu_role.base.level == PT32E_ROOT_LEVEL) {
 			if (!(pdptrs[i] & PT_PRESENT_MASK)) {
 				mmu->pae_root[i] = INVALID_PAE_ROOT;
 				continue;
@@ -3546,9 +3583,9 @@ static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)
 		mmu->pae_root[i] = root | pm_mask;
 	}
 
-	if (mmu->shadow_root_level == PT64_ROOT_5LEVEL)
+	if (mmu->root_role.level == PT64_ROOT_5LEVEL)
 		mmu->root.hpa = __pa(mmu->pml5_root);
-	else if (mmu->shadow_root_level == PT64_ROOT_4LEVEL)
+	else if (mmu->root_role.level == PT64_ROOT_4LEVEL)
 		mmu->root.hpa = __pa(mmu->pml4_root);
 	else
 		mmu->root.hpa = __pa(mmu->pae_root);
@@ -3564,7 +3601,7 @@ out_unlock:
 static int mmu_alloc_special_roots(struct kvm_vcpu *vcpu)
 {
 	struct kvm_mmu *mmu = vcpu->arch.mmu;
-	bool need_pml5 = mmu->shadow_root_level > PT64_ROOT_4LEVEL;
+	bool need_pml5 = mmu->root_role.level > PT64_ROOT_4LEVEL;
 	u64 *pml5_root = NULL;
 	u64 *pml4_root = NULL;
 	u64 *pae_root;
@@ -3575,8 +3612,9 @@ static int mmu_alloc_special_roots(struct kvm_vcpu *vcpu)
 	 * equivalent level in the guest's NPT to shadow.  Allocate the tables
 	 * on demand, as running a 32-bit L1 VMM on 64-bit KVM is very rare.
 	 */
-	if (mmu->direct_map || mmu->root_level >= PT64_ROOT_4LEVEL ||
-	    mmu->shadow_root_level < PT64_ROOT_4LEVEL)
+	if (mmu->root_role.direct ||
+	    mmu->cpu_role.base.level >= PT64_ROOT_4LEVEL ||
+	    mmu->root_role.level < PT64_ROOT_4LEVEL)
 		return 0;
 
 	/*
@@ -3672,7 +3710,7 @@ void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
 	int i;
 	struct kvm_mmu_page *sp;
 
-	if (vcpu->arch.mmu->direct_map)
+	if (vcpu->arch.mmu->root_role.direct)
 		return;
 
 	if (!VALID_PAGE(vcpu->arch.mmu->root.hpa))
@@ -3680,7 +3718,7 @@ void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
 
 	vcpu_clear_mmio_info(vcpu, MMIO_GVA_ANY);
 
-	if (vcpu->arch.mmu->root_level >= PT64_ROOT_4LEVEL) {
+	if (vcpu->arch.mmu->cpu_role.base.level >= PT64_ROOT_4LEVEL) {
 		hpa_t root = vcpu->arch.mmu->root.hpa;
 		sp = to_shadow_page(root);
 
@@ -3902,14 +3940,33 @@ static bool kvm_arch_setup_async_pf(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
 
 	arch.token = alloc_apf_token(vcpu);
 	arch.gfn = gfn;
-	arch.direct_map = vcpu->arch.mmu->direct_map;
+	arch.direct_map = vcpu->arch.mmu->root_role.direct;
 	arch.cr3 = vcpu->arch.mmu->get_guest_pgd(vcpu);
 
 	return kvm_setup_async_pf(vcpu, cr2_or_gpa,
 				  kvm_vcpu_gfn_to_hva(vcpu, gfn), &arch);
 }
 
-static bool kvm_faultin_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault, int *r)
+void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu, struct kvm_async_pf *work)
+{
+	int r;
+
+	if ((vcpu->arch.mmu->root_role.direct != work->arch.direct_map) ||
+	      work->wakeup_all)
+		return;
+
+	r = kvm_mmu_reload(vcpu);
+	if (unlikely(r))
+		return;
+
+	if (!vcpu->arch.mmu->root_role.direct &&
+	      work->arch.cr3 != vcpu->arch.mmu->get_guest_pgd(vcpu))
+		return;
+
+	kvm_mmu_do_page_fault(vcpu, work->cr2_or_gpa, 0, true);
+}
+
+static int kvm_faultin_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 {
 	struct kvm_memory_slot *slot = fault->slot;
 	bool async;
@@ -3920,7 +3977,7 @@ static bool kvm_faultin_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
 	 * be zapped before KVM inserts a new MMIO SPTE for the gfn.
 	 */
 	if (slot && (slot->flags & KVM_MEMSLOT_INVALID))
-		goto out_retry;
+		return RET_PF_RETRY;
 
 	if (!kvm_is_visible_memslot(slot)) {
 		/* Don't expose private memslots to L2. */
@@ -3928,7 +3985,7 @@ static bool kvm_faultin_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
 			fault->slot = NULL;
 			fault->pfn = KVM_PFN_NOSLOT;
 			fault->map_writable = false;
-			return false;
+			return RET_PF_CONTINUE;
 		}
 		/*
 		 * If the APIC access page exists but is disabled, go directly
@@ -3937,10 +3994,8 @@ static bool kvm_faultin_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
 		 * when the AVIC is re-enabled.
 		 */
 		if (slot && slot->id == APIC_ACCESS_PAGE_PRIVATE_MEMSLOT &&
-		    !kvm_apicv_activated(vcpu->kvm)) {
-			*r = RET_PF_EMULATE;
-			return true;
-		}
+		    !kvm_apicv_activated(vcpu->kvm))
+			return RET_PF_EMULATE;
 	}
 
 	async = false;
@@ -3948,26 +4003,23 @@ static bool kvm_faultin_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
 					  fault->write, &fault->map_writable,
 					  &fault->hva);
 	if (!async)
-		return false; /* *pfn has correct page already */
+		return RET_PF_CONTINUE; /* *pfn has correct page already */
 
 	if (!fault->prefetch && kvm_can_do_async_pf(vcpu)) {
 		trace_kvm_try_async_get_page(fault->addr, fault->gfn);
 		if (kvm_find_async_pf_gfn(vcpu, fault->gfn)) {
 			trace_kvm_async_pf_doublefault(fault->addr, fault->gfn);
 			kvm_make_request(KVM_REQ_APF_HALT, vcpu);
-			goto out_retry;
-		} else if (kvm_arch_setup_async_pf(vcpu, fault->addr, fault->gfn))
-			goto out_retry;
+			return RET_PF_RETRY;
+		} else if (kvm_arch_setup_async_pf(vcpu, fault->addr, fault->gfn)) {
+			return RET_PF_RETRY;
+		}
 	}
 
 	fault->pfn = __gfn_to_pfn_memslot(slot, fault->gfn, false, NULL,
 					  fault->write, &fault->map_writable,
 					  &fault->hva);
-	return false;
-
-out_retry:
-	*r = RET_PF_RETRY;
-	return true;
+	return RET_PF_CONTINUE;
 }
 
 /*
@@ -4022,10 +4074,12 @@ static int direct_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault
 	mmu_seq = vcpu->kvm->mmu_notifier_seq;
 	smp_rmb();
 
-	if (kvm_faultin_pfn(vcpu, fault, &r))
+	r = kvm_faultin_pfn(vcpu, fault);
+	if (r != RET_PF_CONTINUE)
 		return r;
 
-	if (handle_abnormal_pfn(vcpu, fault, ACC_ALL, &r))
+	r = handle_abnormal_pfn(vcpu, fault, ACC_ALL);
+	if (r != RET_PF_CONTINUE)
 		return r;
 
 	r = RET_PF_RETRY;
@@ -4120,7 +4174,6 @@ static void nonpaging_init_context(struct kvm_mmu *context)
 	context->gva_to_gpa = nonpaging_gva_to_gpa;
 	context->sync_page = nonpaging_sync_page;
 	context->invlpg = NULL;
-	context->direct_map = true;
 }
 
 static inline bool is_root_usable(struct kvm_mmu_root_info *root, gpa_t pgd,
@@ -4214,7 +4267,7 @@ static bool fast_pgd_switch(struct kvm *kvm, struct kvm_mmu *mmu,
 void kvm_mmu_new_pgd(struct kvm_vcpu *vcpu, gpa_t new_pgd)
 {
 	struct kvm_mmu *mmu = vcpu->arch.mmu;
-	union kvm_mmu_page_role new_role = mmu->mmu_role.base;
+	union kvm_mmu_page_role new_role = mmu->root_role;
 
 	if (!fast_pgd_switch(vcpu->kvm, mmu, new_pgd, new_role)) {
 		/* kvm_mmu_ensure_valid_pgd will set up a new root.  */
@@ -4391,12 +4444,12 @@ static bool guest_can_use_gbpages(struct kvm_vcpu *vcpu)
 			     guest_cpuid_has(vcpu, X86_FEATURE_GBPAGES);
 }
 
-static void reset_rsvds_bits_mask(struct kvm_vcpu *vcpu,
-				  struct kvm_mmu *context)
+static void reset_guest_rsvds_bits_mask(struct kvm_vcpu *vcpu,
+					struct kvm_mmu *context)
 {
 	__reset_rsvds_bits_mask(&context->guest_rsvd_check,
 				vcpu->arch.reserved_gpa_bits,
-				context->root_level, is_efer_nx(context),
+				context->cpu_role.base.level, is_efer_nx(context),
 				guest_can_use_gbpages(vcpu),
 				is_cr4_pse(context),
 				guest_cpuid_is_amd_or_hygon(vcpu));
@@ -4461,16 +4514,6 @@ static inline u64 reserved_hpa_bits(void)
 static void reset_shadow_zero_bits_mask(struct kvm_vcpu *vcpu,
 					struct kvm_mmu *context)
 {
-	/*
-	 * KVM uses NX when TDP is disabled to handle a variety of scenarios,
-	 * notably for huge SPTEs if iTLB multi-hit mitigation is enabled and
-	 * to generate correct permissions for CR0.WP=0/CR4.SMEP=1/EFER.NX=0.
-	 * The iTLB multi-hit workaround can be toggled at any time, so assume
-	 * NX can be used by any non-nested shadow MMU to avoid having to reset
-	 * MMU contexts.  Note, KVM forces EFER.NX=1 when TDP is disabled.
-	 */
-	bool uses_nx = is_efer_nx(context) || !tdp_enabled;
-
 	/* @amd adds a check on bit of SPTEs, which KVM shouldn't use anyways. */
 	bool is_amd = true;
 	/* KVM doesn't use 2-level page tables for the shadow MMU. */
@@ -4478,19 +4521,28 @@ static void reset_shadow_zero_bits_mask(struct kvm_vcpu *vcpu,
 	struct rsvd_bits_validate *shadow_zero_check;
 	int i;
 
-	WARN_ON_ONCE(context->shadow_root_level < PT32E_ROOT_LEVEL);
+	WARN_ON_ONCE(context->root_role.level < PT32E_ROOT_LEVEL);
 
 	shadow_zero_check = &context->shadow_zero_check;
 	__reset_rsvds_bits_mask(shadow_zero_check, reserved_hpa_bits(),
-				context->shadow_root_level, uses_nx,
+				context->root_role.level,
+				context->root_role.efer_nx,
 				guest_can_use_gbpages(vcpu), is_pse, is_amd);
 
 	if (!shadow_me_mask)
 		return;
 
-	for (i = context->shadow_root_level; --i >= 0;) {
-		shadow_zero_check->rsvd_bits_mask[0][i] &= ~shadow_me_mask;
-		shadow_zero_check->rsvd_bits_mask[1][i] &= ~shadow_me_mask;
+	for (i = context->root_role.level; --i >= 0;) {
+		/*
+		 * So far shadow_me_value is a constant during KVM's life
+		 * time.  Bits in shadow_me_value are allowed to be set.
+		 * Bits in shadow_me_mask but not in shadow_me_value are
+		 * not allowed to be set.
+		 */
+		shadow_zero_check->rsvd_bits_mask[0][i] |= shadow_me_mask;
+		shadow_zero_check->rsvd_bits_mask[1][i] |= shadow_me_mask;
+		shadow_zero_check->rsvd_bits_mask[0][i] &= ~shadow_me_value;
+		shadow_zero_check->rsvd_bits_mask[1][i] &= ~shadow_me_value;
 	}
 
 }
@@ -4515,7 +4567,7 @@ reset_tdp_shadow_zero_bits_mask(struct kvm_mmu *context)
 
 	if (boot_cpu_is_amd())
 		__reset_rsvds_bits_mask(shadow_zero_check, reserved_hpa_bits(),
-					context->shadow_root_level, false,
+					context->root_role.level, false,
 					boot_cpu_has(X86_FEATURE_GBPAGES),
 					false, true);
 	else
@@ -4526,7 +4578,7 @@ reset_tdp_shadow_zero_bits_mask(struct kvm_mmu *context)
 	if (!shadow_me_mask)
 		return;
 
-	for (i = context->shadow_root_level; --i >= 0;) {
+	for (i = context->root_role.level; --i >= 0;) {
 		shadow_zero_check->rsvd_bits_mask[0][i] &= ~shadow_me_mask;
 		shadow_zero_check->rsvd_bits_mask[1][i] &= ~shadow_me_mask;
 	}
@@ -4700,7 +4752,7 @@ static void reset_guest_paging_metadata(struct kvm_vcpu *vcpu,
 	if (!is_cr0_pg(mmu))
 		return;
 
-	reset_rsvds_bits_mask(vcpu, mmu);
+	reset_guest_rsvds_bits_mask(vcpu, mmu);
 	update_permission_bitmask(mmu, false);
 	update_pkru_bitmask(mmu);
 }
@@ -4711,7 +4763,6 @@ static void paging64_init_context(struct kvm_mmu *context)
 	context->gva_to_gpa = paging64_gva_to_gpa;
 	context->sync_page = paging64_sync_page;
 	context->invlpg = paging64_invlpg;
-	context->direct_map = false;
 }
 
 static void paging32_init_context(struct kvm_mmu *context)
@@ -4720,51 +4771,45 @@ static void paging32_init_context(struct kvm_mmu *context)
 	context->gva_to_gpa = paging32_gva_to_gpa;
 	context->sync_page = paging32_sync_page;
 	context->invlpg = paging32_invlpg;
-	context->direct_map = false;
-}
-
-static union kvm_mmu_extended_role kvm_calc_mmu_role_ext(struct kvm_vcpu *vcpu,
-							 struct kvm_mmu_role_regs *regs)
-{
-	union kvm_mmu_extended_role ext = {0};
-
-	if (____is_cr0_pg(regs)) {
-		ext.cr0_pg = 1;
-		ext.cr4_pae = ____is_cr4_pae(regs);
-		ext.cr4_smep = ____is_cr4_smep(regs);
-		ext.cr4_smap = ____is_cr4_smap(regs);
-		ext.cr4_pse = ____is_cr4_pse(regs);
-
-		/* PKEY and LA57 are active iff long mode is active. */
-		ext.cr4_pke = ____is_efer_lma(regs) && ____is_cr4_pke(regs);
-		ext.cr4_la57 = ____is_efer_lma(regs) && ____is_cr4_la57(regs);
-		ext.efer_lma = ____is_efer_lma(regs);
-	}
-
-	ext.valid = 1;
-
-	return ext;
 }
 
-static union kvm_mmu_role kvm_calc_mmu_role_common(struct kvm_vcpu *vcpu,
-						   struct kvm_mmu_role_regs *regs,
-						   bool base_only)
+static union kvm_cpu_role
+kvm_calc_cpu_role(struct kvm_vcpu *vcpu, const struct kvm_mmu_role_regs *regs)
 {
-	union kvm_mmu_role role = {0};
+	union kvm_cpu_role role = {0};
 
 	role.base.access = ACC_ALL;
-	if (____is_cr0_pg(regs)) {
-		role.base.efer_nx = ____is_efer_nx(regs);
-		role.base.cr0_wp = ____is_cr0_wp(regs);
-	}
 	role.base.smm = is_smm(vcpu);
 	role.base.guest_mode = is_guest_mode(vcpu);
+	role.ext.valid = 1;
 
-	if (base_only)
+	if (!____is_cr0_pg(regs)) {
+		role.base.direct = 1;
 		return role;
+	}
 
-	role.ext = kvm_calc_mmu_role_ext(vcpu, regs);
+	role.base.efer_nx = ____is_efer_nx(regs);
+	role.base.cr0_wp = ____is_cr0_wp(regs);
+	role.base.smep_andnot_wp = ____is_cr4_smep(regs) && !____is_cr0_wp(regs);
+	role.base.smap_andnot_wp = ____is_cr4_smap(regs) && !____is_cr0_wp(regs);
+	role.base.has_4_byte_gpte = !____is_cr4_pae(regs);
 
+	if (____is_efer_lma(regs))
+		role.base.level = ____is_cr4_la57(regs) ? PT64_ROOT_5LEVEL
+							: PT64_ROOT_4LEVEL;
+	else if (____is_cr4_pae(regs))
+		role.base.level = PT32E_ROOT_LEVEL;
+	else
+		role.base.level = PT32_ROOT_LEVEL;
+
+	role.ext.cr4_smep = ____is_cr4_smep(regs);
+	role.ext.cr4_smap = ____is_cr4_smap(regs);
+	role.ext.cr4_pse = ____is_cr4_pse(regs);
+
+	/* PKEY and LA57 are active iff long mode is active. */
+	role.ext.cr4_pke = ____is_efer_lma(regs) && ____is_cr4_pke(regs);
+	role.ext.cr4_la57 = ____is_efer_lma(regs) && ____is_cr4_la57(regs);
+	role.ext.efer_lma = ____is_efer_lma(regs);
 	return role;
 }
 
@@ -4781,40 +4826,43 @@ static inline int kvm_mmu_get_tdp_level(struct kvm_vcpu *vcpu)
 	return max_tdp_level;
 }
 
-static union kvm_mmu_role
+static union kvm_mmu_page_role
 kvm_calc_tdp_mmu_root_page_role(struct kvm_vcpu *vcpu,
-				struct kvm_mmu_role_regs *regs, bool base_only)
+				union kvm_cpu_role cpu_role)
 {
-	union kvm_mmu_role role = kvm_calc_mmu_role_common(vcpu, regs, base_only);
+	union kvm_mmu_page_role role = {0};
 
-	role.base.ad_disabled = (shadow_accessed_mask == 0);
-	role.base.level = kvm_mmu_get_tdp_level(vcpu);
-	role.base.direct = true;
-	role.base.has_4_byte_gpte = false;
+	role.access = ACC_ALL;
+	role.cr0_wp = true;
+	role.efer_nx = true;
+	role.smm = cpu_role.base.smm;
+	role.guest_mode = cpu_role.base.guest_mode;
+	role.ad_disabled = !kvm_ad_enabled();
+	role.level = kvm_mmu_get_tdp_level(vcpu);
+	role.direct = true;
+	role.has_4_byte_gpte = false;
 
 	return role;
 }
 
-static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu)
+static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu,
+			     union kvm_cpu_role cpu_role)
 {
 	struct kvm_mmu *context = &vcpu->arch.root_mmu;
-	struct kvm_mmu_role_regs regs = vcpu_to_role_regs(vcpu);
-	union kvm_mmu_role new_role =
-		kvm_calc_tdp_mmu_root_page_role(vcpu, &regs, false);
+	union kvm_mmu_page_role root_role = kvm_calc_tdp_mmu_root_page_role(vcpu, cpu_role);
 
-	if (new_role.as_u64 == context->mmu_role.as_u64)
+	if (cpu_role.as_u64 == context->cpu_role.as_u64 &&
+	    root_role.word == context->root_role.word)
 		return;
 
-	context->mmu_role.as_u64 = new_role.as_u64;
+	context->cpu_role.as_u64 = cpu_role.as_u64;
+	context->root_role.word = root_role.word;
 	context->page_fault = kvm_tdp_page_fault;
 	context->sync_page = nonpaging_sync_page;
 	context->invlpg = NULL;
-	context->shadow_root_level = kvm_mmu_get_tdp_level(vcpu);
-	context->direct_map = true;
 	context->get_guest_pgd = get_cr3;
 	context->get_pdptr = kvm_pdptr_read;
 	context->inject_page_fault = kvm_inject_page_fault;
-	context->root_level = role_regs_to_root_level(&regs);
 
 	if (!is_cr0_pg(context))
 		context->gva_to_gpa = nonpaging_gva_to_gpa;
@@ -4827,46 +4875,16 @@ static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu)
 	reset_tdp_shadow_zero_bits_mask(context);
 }
 
-static union kvm_mmu_role
-kvm_calc_shadow_root_page_role_common(struct kvm_vcpu *vcpu,
-				      struct kvm_mmu_role_regs *regs, bool base_only)
-{
-	union kvm_mmu_role role = kvm_calc_mmu_role_common(vcpu, regs, base_only);
-
-	role.base.smep_andnot_wp = role.ext.cr4_smep && !____is_cr0_wp(regs);
-	role.base.smap_andnot_wp = role.ext.cr4_smap && !____is_cr0_wp(regs);
-	role.base.has_4_byte_gpte = ____is_cr0_pg(regs) && !____is_cr4_pae(regs);
-
-	return role;
-}
-
-static union kvm_mmu_role
-kvm_calc_shadow_mmu_root_page_role(struct kvm_vcpu *vcpu,
-				   struct kvm_mmu_role_regs *regs, bool base_only)
-{
-	union kvm_mmu_role role =
-		kvm_calc_shadow_root_page_role_common(vcpu, regs, base_only);
-
-	role.base.direct = !____is_cr0_pg(regs);
-
-	if (!____is_efer_lma(regs))
-		role.base.level = PT32E_ROOT_LEVEL;
-	else if (____is_cr4_la57(regs))
-		role.base.level = PT64_ROOT_5LEVEL;
-	else
-		role.base.level = PT64_ROOT_4LEVEL;
-
-	return role;
-}
-
 static void shadow_mmu_init_context(struct kvm_vcpu *vcpu, struct kvm_mmu *context,
-				    struct kvm_mmu_role_regs *regs,
-				    union kvm_mmu_role new_role)
+				    union kvm_cpu_role cpu_role,
+				    union kvm_mmu_page_role root_role)
 {
-	if (new_role.as_u64 == context->mmu_role.as_u64)
+	if (cpu_role.as_u64 == context->cpu_role.as_u64 &&
+	    root_role.word == context->root_role.word)
 		return;
 
-	context->mmu_role.as_u64 = new_role.as_u64;
+	context->cpu_role.as_u64 = cpu_role.as_u64;
+	context->root_role.word = root_role.word;
 
 	if (!is_cr0_pg(context))
 		nonpaging_init_context(context);
@@ -4874,35 +4892,34 @@ static void shadow_mmu_init_context(struct kvm_vcpu *vcpu, struct kvm_mmu *conte
 		paging64_init_context(context);
 	else
 		paging32_init_context(context);
-	context->root_level = role_regs_to_root_level(regs);
 
 	reset_guest_paging_metadata(vcpu, context);
-	context->shadow_root_level = new_role.base.level;
-
 	reset_shadow_zero_bits_mask(vcpu, context);
 }
 
 static void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu,
-				struct kvm_mmu_role_regs *regs)
+				union kvm_cpu_role cpu_role)
 {
 	struct kvm_mmu *context = &vcpu->arch.root_mmu;
-	union kvm_mmu_role new_role =
-		kvm_calc_shadow_mmu_root_page_role(vcpu, regs, false);
+	union kvm_mmu_page_role root_role;
 
-	shadow_mmu_init_context(vcpu, context, regs, new_role);
-}
+	root_role = cpu_role.base;
 
-static union kvm_mmu_role
-kvm_calc_shadow_npt_root_page_role(struct kvm_vcpu *vcpu,
-				   struct kvm_mmu_role_regs *regs)
-{
-	union kvm_mmu_role role =
-		kvm_calc_shadow_root_page_role_common(vcpu, regs, false);
+	/* KVM uses PAE paging whenever the guest isn't using 64-bit paging. */
+	root_role.level = max_t(u32, root_role.level, PT32E_ROOT_LEVEL);
 
-	role.base.direct = false;
-	role.base.level = kvm_mmu_get_tdp_level(vcpu);
+	/*
+	 * KVM forces EFER.NX=1 when TDP is disabled, reflect it in the MMU role.
+	 * KVM uses NX when TDP is disabled to handle a variety of scenarios,
+	 * notably for huge SPTEs if iTLB multi-hit mitigation is enabled and
+	 * to generate correct permissions for CR0.WP=0/CR4.SMEP=1/EFER.NX=0.
+	 * The iTLB multi-hit workaround can be toggled at any time, so assume
+	 * NX can be used by any non-nested shadow MMU to avoid having to reset
+	 * MMU contexts.
+	 */
+	root_role.efer_nx = true;
 
-	return role;
+	shadow_mmu_init_context(vcpu, context, cpu_role, root_role);
 }
 
 void kvm_init_shadow_npt_mmu(struct kvm_vcpu *vcpu, unsigned long cr0,
@@ -4914,24 +4931,34 @@ void kvm_init_shadow_npt_mmu(struct kvm_vcpu *vcpu, unsigned long cr0,
 		.cr4 = cr4 & ~X86_CR4_PKE,
 		.efer = efer,
 	};
-	union kvm_mmu_role new_role;
+	union kvm_cpu_role cpu_role = kvm_calc_cpu_role(vcpu, &regs);
+	union kvm_mmu_page_role root_role;
+
+	/* NPT requires CR0.PG=1. */
+	WARN_ON_ONCE(cpu_role.base.direct);
 
-	new_role = kvm_calc_shadow_npt_root_page_role(vcpu, &regs);
+	root_role = cpu_role.base;
+	root_role.level = kvm_mmu_get_tdp_level(vcpu);
+	if (root_role.level == PT64_ROOT_5LEVEL &&
+	    cpu_role.base.level == PT64_ROOT_4LEVEL)
+		root_role.passthrough = 1;
 
-	shadow_mmu_init_context(vcpu, context, &regs, new_role);
+	shadow_mmu_init_context(vcpu, context, cpu_role, root_role);
 	kvm_mmu_new_pgd(vcpu, nested_cr3);
 }
 EXPORT_SYMBOL_GPL(kvm_init_shadow_npt_mmu);
 
-static union kvm_mmu_role
+static union kvm_cpu_role
 kvm_calc_shadow_ept_root_page_role(struct kvm_vcpu *vcpu, bool accessed_dirty,
 				   bool execonly, u8 level)
 {
-	union kvm_mmu_role role = {0};
-
-	/* SMM flag is inherited from root_mmu */
-	role.base.smm = vcpu->arch.root_mmu.mmu_role.base.smm;
+	union kvm_cpu_role role = {0};
 
+	/*
+	 * KVM does not support SMM transfer monitors, and consequently does not
+	 * support the "entry to SMM" control either.  role.base.smm is always 0.
+	 */
+	WARN_ON_ONCE(is_smm(vcpu));
 	role.base.level = level;
 	role.base.has_4_byte_gpte = false;
 	role.base.direct = false;
@@ -4939,7 +4966,6 @@ kvm_calc_shadow_ept_root_page_role(struct kvm_vcpu *vcpu, bool accessed_dirty,
 	role.base.guest_mode = true;
 	role.base.access = ACC_ALL;
 
-	/* EPT, and thus nested EPT, does not consume CR0, CR4, nor EFER. */
 	role.ext.word = 0;
 	role.ext.execonly = execonly;
 	role.ext.valid = 1;
@@ -4953,22 +4979,20 @@ void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly,
 {
 	struct kvm_mmu *context = &vcpu->arch.guest_mmu;
 	u8 level = vmx_eptp_page_walk_level(new_eptp);
-	union kvm_mmu_role new_role =
+	union kvm_cpu_role new_mode =
 		kvm_calc_shadow_ept_root_page_role(vcpu, accessed_dirty,
 						   execonly, level);
 
-	if (new_role.as_u64 != context->mmu_role.as_u64) {
-		context->mmu_role.as_u64 = new_role.as_u64;
+	if (new_mode.as_u64 != context->cpu_role.as_u64) {
+		/* EPT, and thus nested EPT, does not consume CR0, CR4, nor EFER. */
+		context->cpu_role.as_u64 = new_mode.as_u64;
+		context->root_role.word = new_mode.base.word;
 
-		context->shadow_root_level = level;
-
-		context->ept_ad = accessed_dirty;
 		context->page_fault = ept_page_fault;
 		context->gva_to_gpa = ept_gva_to_gpa;
 		context->sync_page = ept_sync_page;
 		context->invlpg = ept_invlpg;
-		context->root_level = level;
-		context->direct_map = false;
+
 		update_permission_bitmask(context, true);
 		context->pkru_mask = 0;
 		reset_rsvds_bits_mask_ept(vcpu, context, execonly, huge_page_level);
@@ -4979,49 +5003,30 @@ void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly,
 }
 EXPORT_SYMBOL_GPL(kvm_init_shadow_ept_mmu);
 
-static void init_kvm_softmmu(struct kvm_vcpu *vcpu)
+static void init_kvm_softmmu(struct kvm_vcpu *vcpu,
+			     union kvm_cpu_role cpu_role)
 {
 	struct kvm_mmu *context = &vcpu->arch.root_mmu;
-	struct kvm_mmu_role_regs regs = vcpu_to_role_regs(vcpu);
 
-	kvm_init_shadow_mmu(vcpu, &regs);
+	kvm_init_shadow_mmu(vcpu, cpu_role);
 
 	context->get_guest_pgd     = get_cr3;
 	context->get_pdptr         = kvm_pdptr_read;
 	context->inject_page_fault = kvm_inject_page_fault;
 }
 
-static union kvm_mmu_role
-kvm_calc_nested_mmu_role(struct kvm_vcpu *vcpu, struct kvm_mmu_role_regs *regs)
+static void init_kvm_nested_mmu(struct kvm_vcpu *vcpu,
+				union kvm_cpu_role new_mode)
 {
-	union kvm_mmu_role role;
-
-	role = kvm_calc_shadow_root_page_role_common(vcpu, regs, false);
-
-	/*
-	 * Nested MMUs are used only for walking L2's gva->gpa, they never have
-	 * shadow pages of their own and so "direct" has no meaning.   Set it
-	 * to "true" to try to detect bogus usage of the nested MMU.
-	 */
-	role.base.direct = true;
-	role.base.level = role_regs_to_root_level(regs);
-	return role;
-}
-
-static void init_kvm_nested_mmu(struct kvm_vcpu *vcpu)
-{
-	struct kvm_mmu_role_regs regs = vcpu_to_role_regs(vcpu);
-	union kvm_mmu_role new_role = kvm_calc_nested_mmu_role(vcpu, &regs);
 	struct kvm_mmu *g_context = &vcpu->arch.nested_mmu;
 
-	if (new_role.as_u64 == g_context->mmu_role.as_u64)
+	if (new_mode.as_u64 == g_context->cpu_role.as_u64)
 		return;
 
-	g_context->mmu_role.as_u64 = new_role.as_u64;
+	g_context->cpu_role.as_u64   = new_mode.as_u64;
 	g_context->get_guest_pgd     = get_cr3;
 	g_context->get_pdptr         = kvm_pdptr_read;
 	g_context->inject_page_fault = kvm_inject_page_fault;
-	g_context->root_level        = new_role.base.level;
 
 	/*
 	 * L2 page tables are never shadowed, so there is no need to sync
@@ -5051,12 +5056,15 @@ static void init_kvm_nested_mmu(struct kvm_vcpu *vcpu)
 
 void kvm_init_mmu(struct kvm_vcpu *vcpu)
 {
+	struct kvm_mmu_role_regs regs = vcpu_to_role_regs(vcpu);
+	union kvm_cpu_role cpu_role = kvm_calc_cpu_role(vcpu, &regs);
+
 	if (mmu_is_nested(vcpu))
-		init_kvm_nested_mmu(vcpu);
+		init_kvm_nested_mmu(vcpu, cpu_role);
 	else if (tdp_enabled)
-		init_kvm_tdp_mmu(vcpu);
+		init_kvm_tdp_mmu(vcpu, cpu_role);
 	else
-		init_kvm_softmmu(vcpu);
+		init_kvm_softmmu(vcpu, cpu_role);
 }
 EXPORT_SYMBOL_GPL(kvm_init_mmu);
 
@@ -5074,9 +5082,12 @@ void kvm_mmu_after_set_cpuid(struct kvm_vcpu *vcpu)
 	 * problem is swept under the rug; KVM's CPUID API is horrific and
 	 * it's all but impossible to solve it without introducing a new API.
 	 */
-	vcpu->arch.root_mmu.mmu_role.ext.valid = 0;
-	vcpu->arch.guest_mmu.mmu_role.ext.valid = 0;
-	vcpu->arch.nested_mmu.mmu_role.ext.valid = 0;
+	vcpu->arch.root_mmu.root_role.word = 0;
+	vcpu->arch.guest_mmu.root_role.word = 0;
+	vcpu->arch.nested_mmu.root_role.word = 0;
+	vcpu->arch.root_mmu.cpu_role.ext.valid = 0;
+	vcpu->arch.guest_mmu.cpu_role.ext.valid = 0;
+	vcpu->arch.nested_mmu.cpu_role.ext.valid = 0;
 	kvm_mmu_reset_context(vcpu);
 
 	/*
@@ -5097,13 +5108,13 @@ int kvm_mmu_load(struct kvm_vcpu *vcpu)
 {
 	int r;
 
-	r = mmu_topup_memory_caches(vcpu, !vcpu->arch.mmu->direct_map);
+	r = mmu_topup_memory_caches(vcpu, !vcpu->arch.mmu->root_role.direct);
 	if (r)
 		goto out;
 	r = mmu_alloc_special_roots(vcpu);
 	if (r)
 		goto out;
-	if (vcpu->arch.mmu->direct_map)
+	if (vcpu->arch.mmu->root_role.direct)
 		r = mmu_alloc_direct_roots(vcpu);
 	else
 		r = mmu_alloc_shadow_roots(vcpu);
@@ -5330,7 +5341,7 @@ static void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 
 	++vcpu->kvm->stat.mmu_pte_write;
 
-	for_each_gfn_indirect_valid_sp(vcpu->kvm, sp, gfn) {
+	for_each_gfn_valid_sp_with_gptes(vcpu->kvm, sp, gfn) {
 		if (detect_write_misaligned(sp, gpa, bytes) ||
 		      detect_write_flooding(sp)) {
 			kvm_mmu_prepare_zap_page(vcpu->kvm, sp, &invalid_list);
@@ -5356,11 +5367,11 @@ static void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 	write_unlock(&vcpu->kvm->mmu_lock);
 }
 
-int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, u64 error_code,
+int noinline kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, u64 error_code,
 		       void *insn, int insn_len)
 {
 	int r, emulation_type = EMULTYPE_PF;
-	bool direct = vcpu->arch.mmu->direct_map;
+	bool direct = vcpu->arch.mmu->root_role.direct;
 
 	if (WARN_ON(!VALID_PAGE(vcpu->arch.mmu->root.hpa)))
 		return RET_PF_RETRY;
@@ -5391,7 +5402,7 @@ int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, u64 error_code,
 	 * paging in both guests. If true, we simply unprotect the page
 	 * and resume the guest.
 	 */
-	if (vcpu->arch.mmu->direct_map &&
+	if (vcpu->arch.mmu->root_role.direct &&
 	    (error_code & PFERR_NESTED_GUEST_PAGE) == PFERR_NESTED_GUEST_PAGE) {
 		kvm_mmu_unprotect_page(vcpu->kvm, gpa_to_gfn(cr2_or_gpa));
 		return 1;
@@ -5625,7 +5636,7 @@ static int __kvm_mmu_create(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu)
 	if (!tdp_enabled)
 		set_memory_decrypted((unsigned long)mmu->pae_root, 1);
 	else
-		WARN_ON_ONCE(shadow_me_mask);
+		WARN_ON_ONCE(shadow_me_value);
 
 	for (i = 0; i < 4; ++i)
 		mmu->pae_root[i] = INVALID_PAE_ROOT;
@@ -6287,7 +6298,7 @@ int kvm_mmu_vendor_module_init(void)
 	 */
 	BUILD_BUG_ON(sizeof(union kvm_mmu_page_role) != sizeof(u32));
 	BUILD_BUG_ON(sizeof(union kvm_mmu_extended_role) != sizeof(u32));
-	BUILD_BUG_ON(sizeof(union kvm_mmu_role) != sizeof(u64));
+	BUILD_BUG_ON(sizeof(union kvm_cpu_role) != sizeof(u64));
 
 	kvm_mmu_reset_all_pte_masks();
 
diff --git a/arch/x86/kvm/mmu/mmu_internal.h b/arch/x86/kvm/mmu/mmu_internal.h
index 1bff453f7cbe..bd2a26897b97 100644
--- a/arch/x86/kvm/mmu/mmu_internal.h
+++ b/arch/x86/kvm/mmu/mmu_internal.h
@@ -140,9 +140,72 @@ void kvm_flush_remote_tlbs_with_address(struct kvm *kvm,
 					u64 start_gfn, u64 pages);
 unsigned int pte_list_count(struct kvm_rmap_head *rmap_head);
 
+extern int nx_huge_pages;
+static inline bool is_nx_huge_page_enabled(void)
+{
+	return READ_ONCE(nx_huge_pages);
+}
+
+struct kvm_page_fault {
+	/* arguments to kvm_mmu_do_page_fault.  */
+	const gpa_t addr;
+	const u32 error_code;
+	const bool prefetch;
+
+	/* Derived from error_code.  */
+	const bool exec;
+	const bool write;
+	const bool present;
+	const bool rsvd;
+	const bool user;
+
+	/* Derived from mmu and global state.  */
+	const bool is_tdp;
+	const bool nx_huge_page_workaround_enabled;
+
+	/*
+	 * Whether a >4KB mapping can be created or is forbidden due to NX
+	 * hugepages.
+	 */
+	bool huge_page_disallowed;
+
+	/*
+	 * Maximum page size that can be created for this fault; input to
+	 * FNAME(fetch), __direct_map and kvm_tdp_mmu_map.
+	 */
+	u8 max_level;
+
+	/*
+	 * Page size that can be created based on the max_level and the
+	 * page size used by the host mapping.
+	 */
+	u8 req_level;
+
+	/*
+	 * Page size that will be created based on the req_level and
+	 * huge_page_disallowed.
+	 */
+	u8 goal_level;
+
+	/* Shifted addr, or result of guest page table walk if addr is a gva.  */
+	gfn_t gfn;
+
+	/* The memslot containing gfn. May be NULL. */
+	struct kvm_memory_slot *slot;
+
+	/* Outputs of kvm_faultin_pfn.  */
+	kvm_pfn_t pfn;
+	hva_t hva;
+	bool map_writable;
+};
+
+int kvm_tdp_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);
+
 /*
- * Return values of handle_mmio_page_fault, mmu.page_fault, and fast_page_fault().
+ * Return values of handle_mmio_page_fault(), mmu.page_fault(), fast_page_fault(),
+ * and of course kvm_mmu_do_page_fault().
  *
+ * RET_PF_CONTINUE: So far, so good, keep handling the page fault.
  * RET_PF_RETRY: let CPU fault again on the address.
  * RET_PF_EMULATE: mmio page fault, emulate the instruction directly.
  * RET_PF_INVALID: the spte is invalid, let the real page fault path update it.
@@ -151,15 +214,71 @@ unsigned int pte_list_count(struct kvm_rmap_head *rmap_head);
  *
  * Any names added to this enum should be exported to userspace for use in
  * tracepoints via TRACE_DEFINE_ENUM() in mmutrace.h
+ *
+ * Note, all values must be greater than or equal to zero so as not to encroach
+ * on -errno return values.  Somewhat arbitrarily use '0' for CONTINUE, which
+ * will allow for efficient machine code when checking for CONTINUE, e.g.
+ * "TEST %rax, %rax, JNZ", as all "stop!" values are non-zero.
  */
 enum {
-	RET_PF_RETRY = 0,
+	RET_PF_CONTINUE = 0,
+	RET_PF_RETRY,
 	RET_PF_EMULATE,
 	RET_PF_INVALID,
 	RET_PF_FIXED,
 	RET_PF_SPURIOUS,
 };
 
+static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
+					u32 err, bool prefetch)
+{
+	struct kvm_page_fault fault = {
+		.addr = cr2_or_gpa,
+		.error_code = err,
+		.exec = err & PFERR_FETCH_MASK,
+		.write = err & PFERR_WRITE_MASK,
+		.present = err & PFERR_PRESENT_MASK,
+		.rsvd = err & PFERR_RSVD_MASK,
+		.user = err & PFERR_USER_MASK,
+		.prefetch = prefetch,
+		.is_tdp = likely(vcpu->arch.mmu->page_fault == kvm_tdp_page_fault),
+		.nx_huge_page_workaround_enabled = is_nx_huge_page_enabled(),
+
+		.max_level = KVM_MAX_HUGEPAGE_LEVEL,
+		.req_level = PG_LEVEL_4K,
+		.goal_level = PG_LEVEL_4K,
+	};
+	int r;
+
+	/*
+	 * Async #PF "faults", a.k.a. prefetch faults, are not faults from the
+	 * guest perspective and have already been counted at the time of the
+	 * original fault.
+	 */
+	if (!prefetch)
+		vcpu->stat.pf_taken++;
+
+	if (IS_ENABLED(CONFIG_RETPOLINE) && fault.is_tdp)
+		r = kvm_tdp_page_fault(vcpu, &fault);
+	else
+		r = vcpu->arch.mmu->page_fault(vcpu, &fault);
+
+	/*
+	 * Similar to above, prefetch faults aren't truly spurious, and the
+	 * async #PF path doesn't do emulation.  Do count faults that are fixed
+	 * by the async #PF handler though, otherwise they'll never be counted.
+	 */
+	if (r == RET_PF_FIXED)
+		vcpu->stat.pf_fixed++;
+	else if (prefetch)
+		;
+	else if (r == RET_PF_EMULATE)
+		vcpu->stat.pf_emulate++;
+	else if (r == RET_PF_SPURIOUS)
+		vcpu->stat.pf_spurious++;
+	return r;
+}
+
 int kvm_mmu_max_mapping_level(struct kvm *kvm,
 			      const struct kvm_memory_slot *slot, gfn_t gfn,
 			      kvm_pfn_t pfn, int max_level);
diff --git a/arch/x86/kvm/mmu/mmutrace.h b/arch/x86/kvm/mmu/mmutrace.h
index 12247b96af01..ae86820cef69 100644
--- a/arch/x86/kvm/mmu/mmutrace.h
+++ b/arch/x86/kvm/mmu/mmutrace.h
@@ -54,6 +54,7 @@
 	{ PFERR_RSVD_MASK, "RSVD" },	\
 	{ PFERR_FETCH_MASK, "F" }
 
+TRACE_DEFINE_ENUM(RET_PF_CONTINUE);
 TRACE_DEFINE_ENUM(RET_PF_RETRY);
 TRACE_DEFINE_ENUM(RET_PF_EMULATE);
 TRACE_DEFINE_ENUM(RET_PF_INVALID);
diff --git a/arch/x86/kvm/mmu/paging_tmpl.h b/arch/x86/kvm/mmu/paging_tmpl.h
index 01fee5f67ac3..db80f7ccaa4e 100644
--- a/arch/x86/kvm/mmu/paging_tmpl.h
+++ b/arch/x86/kvm/mmu/paging_tmpl.h
@@ -63,7 +63,7 @@
 	#define PT_LEVEL_BITS PT64_LEVEL_BITS
 	#define PT_GUEST_DIRTY_SHIFT 9
 	#define PT_GUEST_ACCESSED_SHIFT 8
-	#define PT_HAVE_ACCESSED_DIRTY(mmu) ((mmu)->ept_ad)
+	#define PT_HAVE_ACCESSED_DIRTY(mmu) (!(mmu)->cpu_role.base.ad_disabled)
 	#ifdef CONFIG_X86_64
 	#define CMPXCHG "cmpxchgq"
 	#endif
@@ -144,42 +144,6 @@ static bool FNAME(is_rsvd_bits_set)(struct kvm_mmu *mmu, u64 gpte, int level)
 	       FNAME(is_bad_mt_xwr)(&mmu->guest_rsvd_check, gpte);
 }
 
-static int FNAME(cmpxchg_gpte)(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
-			       pt_element_t __user *ptep_user, unsigned index,
-			       pt_element_t orig_pte, pt_element_t new_pte)
-{
-	signed char r;
-
-	if (!user_access_begin(ptep_user, sizeof(pt_element_t)))
-		return -EFAULT;
-
-#ifdef CMPXCHG
-	asm volatile("1:" LOCK_PREFIX CMPXCHG " %[new], %[ptr]\n"
-		     "setnz %b[r]\n"
-		     "2:"
-		     _ASM_EXTABLE_TYPE_REG(1b, 2b, EX_TYPE_EFAULT_REG, %k[r])
-		     : [ptr] "+m" (*ptep_user),
-		       [old] "+a" (orig_pte),
-		       [r] "=q" (r)
-		     : [new] "r" (new_pte)
-		     : "memory");
-#else
-	asm volatile("1:" LOCK_PREFIX "cmpxchg8b %[ptr]\n"
-		     "setnz %b[r]\n"
-		     "2:"
-		     _ASM_EXTABLE_TYPE_REG(1b, 2b, EX_TYPE_EFAULT_REG, %k[r])
-		     : [ptr] "+m" (*ptep_user),
-		       [old] "+A" (orig_pte),
-		       [r] "=q" (r)
-		     : [new_lo] "b" ((u32)new_pte),
-		       [new_hi] "c" ((u32)(new_pte >> 32))
-		     : "memory");
-#endif
-
-	user_access_end();
-	return r;
-}
-
 static bool FNAME(prefetch_invalid_gpte)(struct kvm_vcpu *vcpu,
 				  struct kvm_mmu_page *sp, u64 *spte,
 				  u64 gpte)
@@ -187,7 +151,7 @@ static bool FNAME(prefetch_invalid_gpte)(struct kvm_vcpu *vcpu,
 	if (!FNAME(is_present_gpte)(gpte))
 		goto no_present;
 
-	/* if accessed bit is not supported prefetch non accessed gpte */
+	/* Prefetch only accessed entries (unless A/D bits are disabled). */
 	if (PT_HAVE_ACCESSED_DIRTY(vcpu->arch.mmu) &&
 	    !(gpte & PT_GUEST_ACCESSED_MASK))
 		goto no_present;
@@ -278,7 +242,7 @@ static int FNAME(update_accessed_dirty_bits)(struct kvm_vcpu *vcpu,
 		if (unlikely(!walker->pte_writable[level - 1]))
 			continue;
 
-		ret = FNAME(cmpxchg_gpte)(vcpu, mmu, ptep_user, index, orig_pte, pte);
+		ret = __try_cmpxchg_user(ptep_user, &orig_pte, pte, fault);
 		if (ret)
 			return ret;
 
@@ -317,7 +281,7 @@ static inline bool FNAME(is_last_gpte)(struct kvm_mmu *mmu,
 	 * is not reserved and does not indicate a large page at this level,
 	 * so clear PT_PAGE_SIZE_MASK in gpte if that is the case.
 	 */
-	gpte &= level - (PT32_ROOT_LEVEL + mmu->mmu_role.ext.cr4_pse);
+	gpte &= level - (PT32_ROOT_LEVEL + mmu->cpu_role.ext.cr4_pse);
 #endif
 	/*
 	 * PG_LEVEL_4K always terminates.  The RHS has bit 7 set
@@ -355,7 +319,7 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
 
 	trace_kvm_mmu_pagetable_walk(addr, access);
 retry_walk:
-	walker->level = mmu->root_level;
+	walker->level = mmu->cpu_role.base.level;
 	pte           = mmu->get_guest_pgd(vcpu);
 	have_ad       = PT_HAVE_ACCESSED_DIRTY(mmu);
 
@@ -515,14 +479,21 @@ error:
 	 * The other bits are set to 0.
 	 */
 	if (!(errcode & PFERR_RSVD_MASK)) {
-		vcpu->arch.exit_qualification &= 0x180;
+		vcpu->arch.exit_qualification &= (EPT_VIOLATION_GVA_IS_VALID |
+						  EPT_VIOLATION_GVA_TRANSLATED);
 		if (write_fault)
 			vcpu->arch.exit_qualification |= EPT_VIOLATION_ACC_WRITE;
 		if (user_fault)
 			vcpu->arch.exit_qualification |= EPT_VIOLATION_ACC_READ;
 		if (fetch_fault)
 			vcpu->arch.exit_qualification |= EPT_VIOLATION_ACC_INSTR;
-		vcpu->arch.exit_qualification |= (pte_access & 0x7) << 3;
+
+		/*
+		 * Note, pte_access holds the raw RWX bits from the EPTE, not
+		 * ACC_*_MASK flags!
+		 */
+		vcpu->arch.exit_qualification |= (pte_access & VMX_EPT_RWX_MASK) <<
+						 EPT_VIOLATION_RWX_SHIFT;
 	}
 #endif
 	walker->fault.address = addr;
@@ -650,7 +621,7 @@ static int FNAME(fetch)(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
 	WARN_ON_ONCE(gw->gfn != base_gfn);
 	direct_access = gw->pte_access;
 
-	top_level = vcpu->arch.mmu->root_level;
+	top_level = vcpu->arch.mmu->cpu_role.base.level;
 	if (top_level == PT32E_ROOT_LEVEL)
 		top_level = PT32_ROOT_LEVEL;
 	/*
@@ -752,7 +723,6 @@ static int FNAME(fetch)(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault,
 		return ret;
 
 	FNAME(pte_prefetch)(vcpu, gw, it.sptep);
-	++vcpu->stat.pf_fixed;
 	return ret;
 
 out_gpte_changed:
@@ -867,10 +837,12 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault
 	mmu_seq = vcpu->kvm->mmu_notifier_seq;
 	smp_rmb();
 
-	if (kvm_faultin_pfn(vcpu, fault, &r))
+	r = kvm_faultin_pfn(vcpu, fault);
+	if (r != RET_PF_CONTINUE)
 		return r;
 
-	if (handle_abnormal_pfn(vcpu, fault, walker.pte_access, &r))
+	r = handle_abnormal_pfn(vcpu, fault, walker.pte_access);
+	if (r != RET_PF_CONTINUE)
 		return r;
 
 	/*
@@ -1017,7 +989,7 @@ static gpa_t FNAME(gva_to_gpa)(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
  */
 static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
 {
-	union kvm_mmu_page_role mmu_role = vcpu->arch.mmu->mmu_role.base;
+	union kvm_mmu_page_role root_role = vcpu->arch.mmu->root_role;
 	int i;
 	bool host_writable;
 	gpa_t first_pte_gpa;
@@ -1036,6 +1008,7 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
 		.level = 0xf,
 		.access = 0x7,
 		.quadrant = 0x3,
+		.passthrough = 0x1,
 	};
 
 	/*
@@ -1045,7 +1018,7 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
 	 * reserved bits checks will be wrong, etc...
 	 */
 	if (WARN_ON_ONCE(sp->role.direct ||
-			 (sp->role.word ^ mmu_role.word) & ~sync_role_ign.word))
+			 (sp->role.word ^ root_role.word) & ~sync_role_ign.word))
 		return -1;
 
 	first_pte_gpa = FNAME(get_level1_sp_gpa)(sp);
diff --git a/arch/x86/kvm/mmu/spte.c b/arch/x86/kvm/mmu/spte.c
index e5c0b6db6f2c..b5960bbde7f7 100644
--- a/arch/x86/kvm/mmu/spte.c
+++ b/arch/x86/kvm/mmu/spte.c
@@ -19,7 +19,7 @@
 #include <asm/memtype.h>
 #include <asm/vmx.h>
 
-static bool __read_mostly enable_mmio_caching = true;
+bool __read_mostly enable_mmio_caching = true;
 module_param_named(mmio_caching, enable_mmio_caching, bool, 0444);
 
 u64 __read_mostly shadow_host_writable_mask;
@@ -33,6 +33,7 @@ u64 __read_mostly shadow_mmio_value;
 u64 __read_mostly shadow_mmio_mask;
 u64 __read_mostly shadow_mmio_access_mask;
 u64 __read_mostly shadow_present_mask;
+u64 __read_mostly shadow_me_value;
 u64 __read_mostly shadow_me_mask;
 u64 __read_mostly shadow_acc_track_mask;
 
@@ -167,8 +168,8 @@ bool make_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
 	else
 		pte_access &= ~ACC_WRITE_MASK;
 
-	if (!kvm_is_mmio_pfn(pfn))
-		spte |= shadow_me_mask;
+	if (shadow_me_value && !kvm_is_mmio_pfn(pfn))
+		spte |= shadow_me_value;
 
 	spte |= (u64)pfn << PAGE_SHIFT;
 
@@ -284,7 +285,7 @@ u64 make_nonleaf_spte(u64 *child_pt, bool ad_disabled)
 	u64 spte = SPTE_MMU_PRESENT_MASK;
 
 	spte |= __pa(child_pt) | shadow_present_mask | PT_WRITABLE_MASK |
-		shadow_user_mask | shadow_x_mask | shadow_me_mask;
+		shadow_user_mask | shadow_x_mask | shadow_me_value;
 
 	if (ad_disabled)
 		spte |= SPTE_TDP_AD_DISABLED_MASK;
@@ -310,25 +311,6 @@ u64 kvm_mmu_changed_pte_notifier_make_spte(u64 old_spte, kvm_pfn_t new_pfn)
 	return new_spte;
 }
 
-static u8 kvm_get_shadow_phys_bits(void)
-{
-	/*
-	 * boot_cpu_data.x86_phys_bits is reduced when MKTME or SME are detected
-	 * in CPU detection code, but the processor treats those reduced bits as
-	 * 'keyID' thus they are not reserved bits. Therefore KVM needs to look at
-	 * the physical address bits reported by CPUID.
-	 */
-	if (likely(boot_cpu_data.extended_cpuid_level >= 0x80000008))
-		return cpuid_eax(0x80000008) & 0xff;
-
-	/*
-	 * Quite weird to have VMX or SVM but not MAXPHYADDR; probably a VM with
-	 * custom CPUID.  Proceed with whatever the kernel found since these features
-	 * aren't virtualizable (SME/SEV also require CPUIDs higher than 0x80000008).
-	 */
-	return boot_cpu_data.x86_phys_bits;
-}
-
 u64 mark_spte_for_access_track(u64 spte)
 {
 	if (spte_ad_enabled(spte))
@@ -379,12 +361,26 @@ void kvm_mmu_set_mmio_spte_mask(u64 mmio_value, u64 mmio_mask, u64 access_mask)
 	    WARN_ON(mmio_value && (REMOVED_SPTE & mmio_mask) == mmio_value))
 		mmio_value = 0;
 
+	if (!mmio_value)
+		enable_mmio_caching = false;
+
 	shadow_mmio_value = mmio_value;
 	shadow_mmio_mask  = mmio_mask;
 	shadow_mmio_access_mask = access_mask;
 }
 EXPORT_SYMBOL_GPL(kvm_mmu_set_mmio_spte_mask);
 
+void kvm_mmu_set_me_spte_mask(u64 me_value, u64 me_mask)
+{
+	/* shadow_me_value must be a subset of shadow_me_mask */
+	if (WARN_ON(me_value & ~me_mask))
+		me_value = me_mask = 0;
+
+	shadow_me_value = me_value;
+	shadow_me_mask = me_mask;
+}
+EXPORT_SYMBOL_GPL(kvm_mmu_set_me_spte_mask);
+
 void kvm_mmu_set_ept_masks(bool has_ad_bits, bool has_exec_only)
 {
 	shadow_user_mask	= VMX_EPT_READABLE_MASK;
@@ -394,8 +390,6 @@ void kvm_mmu_set_ept_masks(bool has_ad_bits, bool has_exec_only)
 	shadow_x_mask		= VMX_EPT_EXECUTABLE_MASK;
 	shadow_present_mask	= has_exec_only ? 0ull : VMX_EPT_READABLE_MASK;
 	shadow_acc_track_mask	= VMX_EPT_RWX_MASK;
-	shadow_me_mask		= 0ull;
-
 	shadow_host_writable_mask = EPT_SPTE_HOST_WRITABLE;
 	shadow_mmu_writable_mask  = EPT_SPTE_MMU_WRITABLE;
 
@@ -446,7 +440,8 @@ void kvm_mmu_reset_all_pte_masks(void)
 	shadow_x_mask		= 0;
 	shadow_present_mask	= PT_PRESENT_MASK;
 	shadow_acc_track_mask	= 0;
-	shadow_me_mask		= sme_me_mask;
+	shadow_me_mask		= 0;
+	shadow_me_value		= 0;
 
 	shadow_host_writable_mask = DEFAULT_SPTE_HOST_WRITABLE;
 	shadow_mmu_writable_mask  = DEFAULT_SPTE_MMU_WRITABLE;
diff --git a/arch/x86/kvm/mmu/spte.h b/arch/x86/kvm/mmu/spte.h
index 80ab0f5cff01..0127bb6e3c7d 100644
--- a/arch/x86/kvm/mmu/spte.h
+++ b/arch/x86/kvm/mmu/spte.h
@@ -5,6 +5,8 @@
 
 #include "mmu_internal.h"
 
+extern bool __read_mostly enable_mmio_caching;
+
 /*
  * A MMU present SPTE is backed by actual memory and may or may not be present
  * in hardware.  E.g. MMIO SPTEs are not considered present.  Use bit 11, as it
@@ -149,6 +151,7 @@ extern u64 __read_mostly shadow_mmio_value;
 extern u64 __read_mostly shadow_mmio_mask;
 extern u64 __read_mostly shadow_mmio_access_mask;
 extern u64 __read_mostly shadow_present_mask;
+extern u64 __read_mostly shadow_me_value;
 extern u64 __read_mostly shadow_me_mask;
 
 /*
@@ -204,7 +207,7 @@ extern u64 __read_mostly shadow_nonpresent_or_rsvd_lower_gfn_mask;
 static inline bool is_mmio_spte(u64 spte)
 {
 	return (spte & shadow_mmio_mask) == shadow_mmio_value &&
-	       likely(shadow_mmio_value);
+	       likely(enable_mmio_caching);
 }
 
 static inline bool is_shadow_present_pte(u64 pte)
@@ -212,6 +215,17 @@ static inline bool is_shadow_present_pte(u64 pte)
 	return !!(pte & SPTE_MMU_PRESENT_MASK);
 }
 
+/*
+ * Returns true if A/D bits are supported in hardware and are enabled by KVM.
+ * When enabled, KVM uses A/D bits for all non-nested MMUs.  Because L1 can
+ * disable A/D bits in EPTP12, SP and SPTE variants are needed to handle the
+ * scenario where KVM is using A/D bits for L1, but not L2.
+ */
+static inline bool kvm_ad_enabled(void)
+{
+	return !!shadow_accessed_mask;
+}
+
 static inline bool sp_ad_disabled(struct kvm_mmu_page *sp)
 {
 	return sp->role.ad_disabled;
diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
index 922b06bf4b94..841feaa48be5 100644
--- a/arch/x86/kvm/mmu/tdp_mmu.c
+++ b/arch/x86/kvm/mmu/tdp_mmu.c
@@ -310,7 +310,7 @@ static void tdp_mmu_init_child_sp(struct kvm_mmu_page *child_sp,
 
 hpa_t kvm_tdp_mmu_get_vcpu_root_hpa(struct kvm_vcpu *vcpu)
 {
-	union kvm_mmu_page_role role = vcpu->arch.mmu->mmu_role.base;
+	union kvm_mmu_page_role role = vcpu->arch.mmu->root_role;
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_mmu_page *root;
 
@@ -1100,6 +1100,7 @@ static int tdp_mmu_map_handle_target_level(struct kvm_vcpu *vcpu,
 
 	/* If a MMIO SPTE is installed, the MMIO will need to be emulated. */
 	if (unlikely(is_mmio_spte(new_spte))) {
+		vcpu->stat.pf_mmio_spte_created++;
 		trace_mark_mmio_spte(rcu_dereference(iter->sptep), iter->gfn,
 				     new_spte);
 		ret = RET_PF_EMULATE;
@@ -1108,13 +1109,6 @@ static int tdp_mmu_map_handle_target_level(struct kvm_vcpu *vcpu,
 				       rcu_dereference(iter->sptep));
 	}
 
-	/*
-	 * Increase pf_fixed in both RET_PF_EMULATE and RET_PF_FIXED to be
-	 * consistent with legacy MMU behavior.
-	 */
-	if (ret != RET_PF_SPURIOUS)
-		vcpu->stat.pf_fixed++;
-
 	return ret;
 }
 
@@ -1136,7 +1130,7 @@ static int tdp_mmu_link_sp(struct kvm *kvm, struct tdp_iter *iter,
 			   struct kvm_mmu_page *sp, bool account_nx,
 			   bool shared)
 {
-	u64 spte = make_nonleaf_spte(sp->spt, !shadow_accessed_mask);
+	u64 spte = make_nonleaf_spte(sp->spt, !kvm_ad_enabled());
 	int ret = 0;
 
 	if (shared) {
@@ -1859,7 +1853,7 @@ int kvm_tdp_mmu_get_walk(struct kvm_vcpu *vcpu, u64 addr, u64 *sptes,
 	gfn_t gfn = addr >> PAGE_SHIFT;
 	int leaf = -1;
 
-	*root_level = vcpu->arch.mmu->shadow_root_level;
+	*root_level = vcpu->arch.mmu->root_role.level;
 
 	tdp_mmu_for_each_pte(iter, mmu, gfn, gfn + 1) {
 		leaf = iter.level;
diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c
index 0604bc29f0b8..3f868fed9114 100644
--- a/arch/x86/kvm/pmu.c
+++ b/arch/x86/kvm/pmu.c
@@ -49,6 +49,32 @@
  *        * AMD:   [0 .. AMD64_NUM_COUNTERS-1] <=> gp counters
  */
 
+static struct kvm_pmu_ops kvm_pmu_ops __read_mostly;
+
+#define KVM_X86_PMU_OP(func)					     \
+	DEFINE_STATIC_CALL_NULL(kvm_x86_pmu_##func,			     \
+				*(((struct kvm_pmu_ops *)0)->func));
+#define KVM_X86_PMU_OP_OPTIONAL KVM_X86_PMU_OP
+#include <asm/kvm-x86-pmu-ops.h>
+
+void kvm_pmu_ops_update(const struct kvm_pmu_ops *pmu_ops)
+{
+	memcpy(&kvm_pmu_ops, pmu_ops, sizeof(kvm_pmu_ops));
+
+#define __KVM_X86_PMU_OP(func) \
+	static_call_update(kvm_x86_pmu_##func, kvm_pmu_ops.func);
+#define KVM_X86_PMU_OP(func) \
+	WARN_ON(!kvm_pmu_ops.func); __KVM_X86_PMU_OP(func)
+#define KVM_X86_PMU_OP_OPTIONAL __KVM_X86_PMU_OP
+#include <asm/kvm-x86-pmu-ops.h>
+#undef __KVM_X86_PMU_OP
+}
+
+static inline bool pmc_is_enabled(struct kvm_pmc *pmc)
+{
+	return static_call(kvm_x86_pmu_pmc_is_enabled)(pmc);
+}
+
 static void kvm_pmi_trigger_fn(struct irq_work *irq_work)
 {
 	struct kvm_pmu *pmu = container_of(irq_work, struct kvm_pmu, irq_work);
@@ -216,7 +242,7 @@ void reprogram_gp_counter(struct kvm_pmc *pmc, u64 eventsel)
 			  ARCH_PERFMON_EVENTSEL_CMASK |
 			  HSW_IN_TX |
 			  HSW_IN_TX_CHECKPOINTED))) {
-		config = kvm_x86_ops.pmu_ops->pmc_perf_hw_id(pmc);
+		config = static_call(kvm_x86_pmu_pmc_perf_hw_id)(pmc);
 		if (config != PERF_COUNT_HW_MAX)
 			type = PERF_TYPE_HARDWARE;
 	}
@@ -266,7 +292,7 @@ void reprogram_fixed_counter(struct kvm_pmc *pmc, u8 ctrl, int idx)
 
 	pmc->current_config = (u64)ctrl;
 	pmc_reprogram_counter(pmc, PERF_TYPE_HARDWARE,
-			      kvm_x86_ops.pmu_ops->pmc_perf_hw_id(pmc),
+			      static_call(kvm_x86_pmu_pmc_perf_hw_id)(pmc),
 			      !(en_field & 0x2), /* exclude user */
 			      !(en_field & 0x1), /* exclude kernel */
 			      pmi);
@@ -275,7 +301,7 @@ EXPORT_SYMBOL_GPL(reprogram_fixed_counter);
 
 void reprogram_counter(struct kvm_pmu *pmu, int pmc_idx)
 {
-	struct kvm_pmc *pmc = kvm_x86_ops.pmu_ops->pmc_idx_to_pmc(pmu, pmc_idx);
+	struct kvm_pmc *pmc = static_call(kvm_x86_pmu_pmc_idx_to_pmc)(pmu, pmc_idx);
 
 	if (!pmc)
 		return;
@@ -297,7 +323,7 @@ void kvm_pmu_handle_event(struct kvm_vcpu *vcpu)
 	int bit;
 
 	for_each_set_bit(bit, pmu->reprogram_pmi, X86_PMC_IDX_MAX) {
-		struct kvm_pmc *pmc = kvm_x86_ops.pmu_ops->pmc_idx_to_pmc(pmu, bit);
+		struct kvm_pmc *pmc = static_call(kvm_x86_pmu_pmc_idx_to_pmc)(pmu, bit);
 
 		if (unlikely(!pmc || !pmc->perf_event)) {
 			clear_bit(bit, pmu->reprogram_pmi);
@@ -319,7 +345,7 @@ void kvm_pmu_handle_event(struct kvm_vcpu *vcpu)
 /* check if idx is a valid index to access PMU */
 bool kvm_pmu_is_valid_rdpmc_ecx(struct kvm_vcpu *vcpu, unsigned int idx)
 {
-	return kvm_x86_ops.pmu_ops->is_valid_rdpmc_ecx(vcpu, idx);
+	return static_call(kvm_x86_pmu_is_valid_rdpmc_ecx)(vcpu, idx);
 }
 
 bool is_vmware_backdoor_pmc(u32 pmc_idx)
@@ -369,7 +395,7 @@ int kvm_pmu_rdpmc(struct kvm_vcpu *vcpu, unsigned idx, u64 *data)
 	if (is_vmware_backdoor_pmc(idx))
 		return kvm_pmu_rdpmc_vmware(vcpu, idx, data);
 
-	pmc = kvm_x86_ops.pmu_ops->rdpmc_ecx_to_pmc(vcpu, idx, &mask);
+	pmc = static_call(kvm_x86_pmu_rdpmc_ecx_to_pmc)(vcpu, idx, &mask);
 	if (!pmc)
 		return 1;
 
@@ -385,22 +411,21 @@ int kvm_pmu_rdpmc(struct kvm_vcpu *vcpu, unsigned idx, u64 *data)
 void kvm_pmu_deliver_pmi(struct kvm_vcpu *vcpu)
 {
 	if (lapic_in_kernel(vcpu)) {
-		if (kvm_x86_ops.pmu_ops->deliver_pmi)
-			kvm_x86_ops.pmu_ops->deliver_pmi(vcpu);
+		static_call_cond(kvm_x86_pmu_deliver_pmi)(vcpu);
 		kvm_apic_local_deliver(vcpu->arch.apic, APIC_LVTPC);
 	}
 }
 
 bool kvm_pmu_is_valid_msr(struct kvm_vcpu *vcpu, u32 msr)
 {
-	return kvm_x86_ops.pmu_ops->msr_idx_to_pmc(vcpu, msr) ||
-		kvm_x86_ops.pmu_ops->is_valid_msr(vcpu, msr);
+	return static_call(kvm_x86_pmu_msr_idx_to_pmc)(vcpu, msr) ||
+		static_call(kvm_x86_pmu_is_valid_msr)(vcpu, msr);
 }
 
 static void kvm_pmu_mark_pmc_in_use(struct kvm_vcpu *vcpu, u32 msr)
 {
 	struct kvm_pmu *pmu = vcpu_to_pmu(vcpu);
-	struct kvm_pmc *pmc = kvm_x86_ops.pmu_ops->msr_idx_to_pmc(vcpu, msr);
+	struct kvm_pmc *pmc = static_call(kvm_x86_pmu_msr_idx_to_pmc)(vcpu, msr);
 
 	if (pmc)
 		__set_bit(pmc->idx, pmu->pmc_in_use);
@@ -408,13 +433,13 @@ static void kvm_pmu_mark_pmc_in_use(struct kvm_vcpu *vcpu, u32 msr)
 
 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 {
-	return kvm_x86_ops.pmu_ops->get_msr(vcpu, msr_info);
+	return static_call(kvm_x86_pmu_get_msr)(vcpu, msr_info);
 }
 
 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 {
 	kvm_pmu_mark_pmc_in_use(vcpu, msr_info->index);
-	return kvm_x86_ops.pmu_ops->set_msr(vcpu, msr_info);
+	return static_call(kvm_x86_pmu_set_msr)(vcpu, msr_info);
 }
 
 /* refresh PMU settings. This function generally is called when underlying
@@ -423,7 +448,7 @@ int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
  */
 void kvm_pmu_refresh(struct kvm_vcpu *vcpu)
 {
-	kvm_x86_ops.pmu_ops->refresh(vcpu);
+	static_call(kvm_x86_pmu_refresh)(vcpu);
 }
 
 void kvm_pmu_reset(struct kvm_vcpu *vcpu)
@@ -431,7 +456,7 @@ void kvm_pmu_reset(struct kvm_vcpu *vcpu)
 	struct kvm_pmu *pmu = vcpu_to_pmu(vcpu);
 
 	irq_work_sync(&pmu->irq_work);
-	kvm_x86_ops.pmu_ops->reset(vcpu);
+	static_call(kvm_x86_pmu_reset)(vcpu);
 }
 
 void kvm_pmu_init(struct kvm_vcpu *vcpu)
@@ -439,7 +464,7 @@ void kvm_pmu_init(struct kvm_vcpu *vcpu)
 	struct kvm_pmu *pmu = vcpu_to_pmu(vcpu);
 
 	memset(pmu, 0, sizeof(*pmu));
-	kvm_x86_ops.pmu_ops->init(vcpu);
+	static_call(kvm_x86_pmu_init)(vcpu);
 	init_irq_work(&pmu->irq_work, kvm_pmi_trigger_fn);
 	pmu->event_count = 0;
 	pmu->need_cleanup = false;
@@ -471,14 +496,13 @@ void kvm_pmu_cleanup(struct kvm_vcpu *vcpu)
 		      pmu->pmc_in_use, X86_PMC_IDX_MAX);
 
 	for_each_set_bit(i, bitmask, X86_PMC_IDX_MAX) {
-		pmc = kvm_x86_ops.pmu_ops->pmc_idx_to_pmc(pmu, i);
+		pmc = static_call(kvm_x86_pmu_pmc_idx_to_pmc)(pmu, i);
 
 		if (pmc && pmc->perf_event && !pmc_speculative_in_use(pmc))
 			pmc_stop_counter(pmc);
 	}
 
-	if (kvm_x86_ops.pmu_ops->cleanup)
-		kvm_x86_ops.pmu_ops->cleanup(vcpu);
+	static_call_cond(kvm_x86_pmu_cleanup)(vcpu);
 
 	bitmap_zero(pmu->pmc_in_use, X86_PMC_IDX_MAX);
 }
@@ -508,7 +532,7 @@ static inline bool eventsel_match_perf_hw_id(struct kvm_pmc *pmc,
 	unsigned int config;
 
 	pmc->eventsel &= (ARCH_PERFMON_EVENTSEL_EVENT | ARCH_PERFMON_EVENTSEL_UMASK);
-	config = kvm_x86_ops.pmu_ops->pmc_perf_hw_id(pmc);
+	config = static_call(kvm_x86_pmu_pmc_perf_hw_id)(pmc);
 	pmc->eventsel = old_eventsel;
 	return config == perf_hw_id;
 }
@@ -536,7 +560,7 @@ void kvm_pmu_trigger_event(struct kvm_vcpu *vcpu, u64 perf_hw_id)
 	int i;
 
 	for_each_set_bit(i, pmu->all_valid_pmc_idx, X86_PMC_IDX_MAX) {
-		pmc = kvm_x86_ops.pmu_ops->pmc_idx_to_pmc(pmu, i);
+		pmc = static_call(kvm_x86_pmu_pmc_idx_to_pmc)(pmu, i);
 
 		if (!pmc || !pmc_is_enabled(pmc) || !pmc_speculative_in_use(pmc))
 			continue;
diff --git a/arch/x86/kvm/pmu.h b/arch/x86/kvm/pmu.h
index 22992b049d38..e745f443b6a8 100644
--- a/arch/x86/kvm/pmu.h
+++ b/arch/x86/kvm/pmu.h
@@ -39,6 +39,8 @@ struct kvm_pmu_ops {
 	void (*cleanup)(struct kvm_vcpu *vcpu);
 };
 
+void kvm_pmu_ops_update(const struct kvm_pmu_ops *pmu_ops);
+
 static inline u64 pmc_bitmask(struct kvm_pmc *pmc)
 {
 	struct kvm_pmu *pmu = pmc_to_pmu(pmc);
@@ -86,11 +88,6 @@ static inline bool pmc_is_fixed(struct kvm_pmc *pmc)
 	return pmc->type == KVM_PMC_FIXED;
 }
 
-static inline bool pmc_is_enabled(struct kvm_pmc *pmc)
-{
-	return kvm_x86_ops.pmu_ops->pmc_is_enabled(pmc);
-}
-
 static inline bool kvm_valid_perf_global_ctrl(struct kvm_pmu *pmu,
 						 u64 data)
 {
diff --git a/arch/x86/kvm/svm/avic.c b/arch/x86/kvm/svm/avic.c
index 421619540ff9..54fe03714f8a 100644
--- a/arch/x86/kvm/svm/avic.c
+++ b/arch/x86/kvm/svm/avic.c
@@ -165,9 +165,8 @@ free_avic:
 	return err;
 }
 
-void avic_init_vmcb(struct vcpu_svm *svm)
+void avic_init_vmcb(struct vcpu_svm *svm, struct vmcb *vmcb)
 {
-	struct vmcb *vmcb = svm->vmcb;
 	struct kvm_svm *kvm_svm = to_kvm_svm(svm->vcpu.kvm);
 	phys_addr_t bpa = __sme_set(page_to_phys(svm->avic_backing_page));
 	phys_addr_t lpa = __sme_set(page_to_phys(kvm_svm->avic_logical_id_table_page));
@@ -285,11 +284,77 @@ void avic_ring_doorbell(struct kvm_vcpu *vcpu)
 	put_cpu();
 }
 
-static void avic_kick_target_vcpus(struct kvm *kvm, struct kvm_lapic *source,
-				   u32 icrl, u32 icrh)
+/*
+ * A fast-path version of avic_kick_target_vcpus(), which attempts to match
+ * destination APIC ID to vCPU without looping through all vCPUs.
+ */
+static int avic_kick_target_vcpus_fast(struct kvm *kvm, struct kvm_lapic *source,
+				       u32 icrl, u32 icrh, u32 index)
 {
+	u32 dest, apic_id;
 	struct kvm_vcpu *vcpu;
+	int dest_mode = icrl & APIC_DEST_MASK;
+	int shorthand = icrl & APIC_SHORT_MASK;
+	struct kvm_svm *kvm_svm = to_kvm_svm(kvm);
+	u32 *avic_logical_id_table = page_address(kvm_svm->avic_logical_id_table_page);
+
+	if (shorthand != APIC_DEST_NOSHORT)
+		return -EINVAL;
+
+	/*
+	 * The AVIC incomplete IPI #vmexit info provides index into
+	 * the physical APIC ID table, which can be used to derive
+	 * guest physical APIC ID.
+	 */
+	if (dest_mode == APIC_DEST_PHYSICAL) {
+		apic_id = index;
+	} else {
+		if (!apic_x2apic_mode(source)) {
+			/* For xAPIC logical mode, the index is for logical APIC table. */
+			apic_id = avic_logical_id_table[index] & 0x1ff;
+		} else {
+			return -EINVAL;
+		}
+	}
+
+	/*
+	 * Assuming vcpu ID is the same as physical apic ID,
+	 * and use it to retrieve the target vCPU.
+	 */
+	vcpu = kvm_get_vcpu_by_id(kvm, apic_id);
+	if (!vcpu)
+		return -EINVAL;
+
+	if (apic_x2apic_mode(vcpu->arch.apic))
+		dest = icrh;
+	else
+		dest = GET_APIC_DEST_FIELD(icrh);
+
+	/*
+	 * Try matching the destination APIC ID with the vCPU.
+	 */
+	if (kvm_apic_match_dest(vcpu, source, shorthand, dest, dest_mode)) {
+		vcpu->arch.apic->irr_pending = true;
+		svm_complete_interrupt_delivery(vcpu,
+						icrl & APIC_MODE_MASK,
+						icrl & APIC_INT_LEVELTRIG,
+						icrl & APIC_VECTOR_MASK);
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static void avic_kick_target_vcpus(struct kvm *kvm, struct kvm_lapic *source,
+				   u32 icrl, u32 icrh, u32 index)
+{
 	unsigned long i;
+	struct kvm_vcpu *vcpu;
+
+	if (!avic_kick_target_vcpus_fast(kvm, source, icrl, icrh, index))
+		return;
+
+	trace_kvm_avic_kick_vcpu_slowpath(icrh, icrl, index);
 
 	/*
 	 * Wake any target vCPUs that are blocking, i.e. waiting for a wake
@@ -316,7 +381,7 @@ int avic_incomplete_ipi_interception(struct kvm_vcpu *vcpu)
 	u32 icrh = svm->vmcb->control.exit_info_1 >> 32;
 	u32 icrl = svm->vmcb->control.exit_info_1;
 	u32 id = svm->vmcb->control.exit_info_2 >> 32;
-	u32 index = svm->vmcb->control.exit_info_2 & 0xFF;
+	u32 index = svm->vmcb->control.exit_info_2 & 0x1FF;
 	struct kvm_lapic *apic = vcpu->arch.apic;
 
 	trace_kvm_avic_incomplete_ipi(vcpu->vcpu_id, icrh, icrl, id, index);
@@ -343,7 +408,7 @@ int avic_incomplete_ipi_interception(struct kvm_vcpu *vcpu)
 		 * set the appropriate IRR bits on the valid target
 		 * vcpus. So, we just need to kick the appropriate vcpu.
 		 */
-		avic_kick_target_vcpus(vcpu->kvm, apic, icrl, icrh);
+		avic_kick_target_vcpus(vcpu->kvm, apic, icrl, icrh, index);
 		break;
 	case AVIC_IPI_FAILURE_INVALID_TARGET:
 		break;
@@ -357,6 +422,13 @@ int avic_incomplete_ipi_interception(struct kvm_vcpu *vcpu)
 	return 1;
 }
 
+unsigned long avic_vcpu_get_apicv_inhibit_reasons(struct kvm_vcpu *vcpu)
+{
+	if (is_guest_mode(vcpu))
+		return APICV_INHIBIT_REASON_NESTED;
+	return 0;
+}
+
 static u32 *avic_get_logical_id_entry(struct kvm_vcpu *vcpu, u32 ldr, bool flat)
 {
 	struct kvm_svm *kvm_svm = to_kvm_svm(vcpu->kvm);
diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
index 96bab464967f..bed5e1692cef 100644
--- a/arch/x86/kvm/svm/nested.c
+++ b/arch/x86/kvm/svm/nested.c
@@ -36,40 +36,45 @@ static void nested_svm_inject_npf_exit(struct kvm_vcpu *vcpu,
 				       struct x86_exception *fault)
 {
 	struct vcpu_svm *svm = to_svm(vcpu);
+	struct vmcb *vmcb = svm->vmcb;
 
-	if (svm->vmcb->control.exit_code != SVM_EXIT_NPF) {
+	if (vmcb->control.exit_code != SVM_EXIT_NPF) {
 		/*
 		 * TODO: track the cause of the nested page fault, and
 		 * correctly fill in the high bits of exit_info_1.
 		 */
-		svm->vmcb->control.exit_code = SVM_EXIT_NPF;
-		svm->vmcb->control.exit_code_hi = 0;
-		svm->vmcb->control.exit_info_1 = (1ULL << 32);
-		svm->vmcb->control.exit_info_2 = fault->address;
+		vmcb->control.exit_code = SVM_EXIT_NPF;
+		vmcb->control.exit_code_hi = 0;
+		vmcb->control.exit_info_1 = (1ULL << 32);
+		vmcb->control.exit_info_2 = fault->address;
 	}
 
-	svm->vmcb->control.exit_info_1 &= ~0xffffffffULL;
-	svm->vmcb->control.exit_info_1 |= fault->error_code;
+	vmcb->control.exit_info_1 &= ~0xffffffffULL;
+	vmcb->control.exit_info_1 |= fault->error_code;
 
 	nested_svm_vmexit(svm);
 }
 
-static void svm_inject_page_fault_nested(struct kvm_vcpu *vcpu, struct x86_exception *fault)
+static bool nested_svm_handle_page_fault_workaround(struct kvm_vcpu *vcpu,
+						    struct x86_exception *fault)
 {
-       struct vcpu_svm *svm = to_svm(vcpu);
-       WARN_ON(!is_guest_mode(vcpu));
+	struct vcpu_svm *svm = to_svm(vcpu);
+	struct vmcb *vmcb = svm->vmcb;
+
+ 	WARN_ON(!is_guest_mode(vcpu));
 
 	if (vmcb12_is_intercept(&svm->nested.ctl,
 				INTERCEPT_EXCEPTION_OFFSET + PF_VECTOR) &&
-	    !svm->nested.nested_run_pending) {
-               svm->vmcb->control.exit_code = SVM_EXIT_EXCP_BASE + PF_VECTOR;
-               svm->vmcb->control.exit_code_hi = 0;
-               svm->vmcb->control.exit_info_1 = fault->error_code;
-               svm->vmcb->control.exit_info_2 = fault->address;
-               nested_svm_vmexit(svm);
-       } else {
-               kvm_inject_page_fault(vcpu, fault);
-       }
+	    !WARN_ON_ONCE(svm->nested.nested_run_pending)) {
+	     	vmcb->control.exit_code = SVM_EXIT_EXCP_BASE + PF_VECTOR;
+		vmcb->control.exit_code_hi = 0;
+		vmcb->control.exit_info_1 = fault->error_code;
+		vmcb->control.exit_info_2 = fault->address;
+		nested_svm_vmexit(svm);
+		return true;
+	}
+
+	return false;
 }
 
 static u64 nested_svm_get_tdp_pdptr(struct kvm_vcpu *vcpu, int index)
@@ -121,6 +126,20 @@ static void nested_svm_uninit_mmu_context(struct kvm_vcpu *vcpu)
 	vcpu->arch.walk_mmu = &vcpu->arch.root_mmu;
 }
 
+static bool nested_vmcb_needs_vls_intercept(struct vcpu_svm *svm)
+{
+	if (!svm->v_vmload_vmsave_enabled)
+		return true;
+
+	if (!nested_npt_enabled(svm))
+		return true;
+
+	if (!(svm->nested.ctl.virt_ext & VIRTUAL_VMLOAD_VMSAVE_ENABLE_MASK))
+		return true;
+
+	return false;
+}
+
 void recalc_intercepts(struct vcpu_svm *svm)
 {
 	struct vmcb_control_area *c, *h;
@@ -162,8 +181,17 @@ void recalc_intercepts(struct vcpu_svm *svm)
 	if (!intercept_smi)
 		vmcb_clr_intercept(c, INTERCEPT_SMI);
 
-	vmcb_set_intercept(c, INTERCEPT_VMLOAD);
-	vmcb_set_intercept(c, INTERCEPT_VMSAVE);
+	if (nested_vmcb_needs_vls_intercept(svm)) {
+		/*
+		 * If the virtual VMLOAD/VMSAVE is not enabled for the L2,
+		 * we must intercept these instructions to correctly
+		 * emulate them in case L1 doesn't intercept them.
+		 */
+		vmcb_set_intercept(c, INTERCEPT_VMLOAD);
+		vmcb_set_intercept(c, INTERCEPT_VMSAVE);
+	} else {
+		WARN_ON(!(c->virt_ext & VIRTUAL_VMLOAD_VMSAVE_ENABLE_MASK));
+	}
 }
 
 /*
@@ -413,6 +441,10 @@ void nested_sync_control_from_vmcb02(struct vcpu_svm *svm)
 		 */
 		mask &= ~V_IRQ_MASK;
 	}
+
+	if (nested_vgif_enabled(svm))
+		mask |= V_GIF_MASK;
+
 	svm->nested.ctl.int_ctl        &= ~mask;
 	svm->nested.ctl.int_ctl        |= svm->vmcb->control.int_ctl & mask;
 }
@@ -454,11 +486,6 @@ static void nested_save_pending_event_to_vmcb12(struct vcpu_svm *svm,
 	vmcb12->control.exit_int_info = exit_int_info;
 }
 
-static inline bool nested_npt_enabled(struct vcpu_svm *svm)
-{
-	return svm->nested.ctl.nested_ctl & SVM_NESTED_CTL_NP_ENABLE;
-}
-
 static void nested_svm_transition_tlb_flush(struct kvm_vcpu *vcpu)
 {
 	/*
@@ -515,6 +542,8 @@ void nested_vmcb02_compute_g_pat(struct vcpu_svm *svm)
 static void nested_vmcb02_prepare_save(struct vcpu_svm *svm, struct vmcb *vmcb12)
 {
 	bool new_vmcb12 = false;
+	struct vmcb *vmcb01 = svm->vmcb01.ptr;
+	struct vmcb *vmcb02 = svm->nested.vmcb02.ptr;
 
 	nested_vmcb02_compute_g_pat(svm);
 
@@ -526,18 +555,18 @@ static void nested_vmcb02_prepare_save(struct vcpu_svm *svm, struct vmcb *vmcb12
 	}
 
 	if (unlikely(new_vmcb12 || vmcb_is_dirty(vmcb12, VMCB_SEG))) {
-		svm->vmcb->save.es = vmcb12->save.es;
-		svm->vmcb->save.cs = vmcb12->save.cs;
-		svm->vmcb->save.ss = vmcb12->save.ss;
-		svm->vmcb->save.ds = vmcb12->save.ds;
-		svm->vmcb->save.cpl = vmcb12->save.cpl;
-		vmcb_mark_dirty(svm->vmcb, VMCB_SEG);
+		vmcb02->save.es = vmcb12->save.es;
+		vmcb02->save.cs = vmcb12->save.cs;
+		vmcb02->save.ss = vmcb12->save.ss;
+		vmcb02->save.ds = vmcb12->save.ds;
+		vmcb02->save.cpl = vmcb12->save.cpl;
+		vmcb_mark_dirty(vmcb02, VMCB_SEG);
 	}
 
 	if (unlikely(new_vmcb12 || vmcb_is_dirty(vmcb12, VMCB_DT))) {
-		svm->vmcb->save.gdtr = vmcb12->save.gdtr;
-		svm->vmcb->save.idtr = vmcb12->save.idtr;
-		vmcb_mark_dirty(svm->vmcb, VMCB_DT);
+		vmcb02->save.gdtr = vmcb12->save.gdtr;
+		vmcb02->save.idtr = vmcb12->save.idtr;
+		vmcb_mark_dirty(vmcb02, VMCB_DT);
 	}
 
 	kvm_set_rflags(&svm->vcpu, vmcb12->save.rflags | X86_EFLAGS_FIXED);
@@ -554,47 +583,59 @@ static void nested_vmcb02_prepare_save(struct vcpu_svm *svm, struct vmcb *vmcb12
 	kvm_rip_write(&svm->vcpu, vmcb12->save.rip);
 
 	/* In case we don't even reach vcpu_run, the fields are not updated */
-	svm->vmcb->save.rax = vmcb12->save.rax;
-	svm->vmcb->save.rsp = vmcb12->save.rsp;
-	svm->vmcb->save.rip = vmcb12->save.rip;
+	vmcb02->save.rax = vmcb12->save.rax;
+	vmcb02->save.rsp = vmcb12->save.rsp;
+	vmcb02->save.rip = vmcb12->save.rip;
 
 	/* These bits will be set properly on the first execution when new_vmc12 is true */
 	if (unlikely(new_vmcb12 || vmcb_is_dirty(vmcb12, VMCB_DR))) {
-		svm->vmcb->save.dr7 = svm->nested.save.dr7 | DR7_FIXED_1;
+		vmcb02->save.dr7 = svm->nested.save.dr7 | DR7_FIXED_1;
 		svm->vcpu.arch.dr6  = svm->nested.save.dr6 | DR6_ACTIVE_LOW;
-		vmcb_mark_dirty(svm->vmcb, VMCB_DR);
+		vmcb_mark_dirty(vmcb02, VMCB_DR);
+	}
+
+	if (unlikely(svm->lbrv_enabled && (svm->nested.ctl.virt_ext & LBR_CTL_ENABLE_MASK))) {
+		/*
+		 * Reserved bits of DEBUGCTL are ignored.  Be consistent with
+		 * svm_set_msr's definition of reserved bits.
+		 */
+		svm_copy_lbrs(vmcb02, vmcb12);
+		vmcb02->save.dbgctl &= ~DEBUGCTL_RESERVED_BITS;
+		svm_update_lbrv(&svm->vcpu);
+
+	} else if (unlikely(vmcb01->control.virt_ext & LBR_CTL_ENABLE_MASK)) {
+		svm_copy_lbrs(vmcb02, vmcb01);
 	}
 }
 
 static void nested_vmcb02_prepare_control(struct vcpu_svm *svm)
 {
-	const u32 int_ctl_vmcb01_bits =
-		V_INTR_MASKING_MASK | V_GIF_MASK | V_GIF_ENABLE_MASK;
-
-	const u32 int_ctl_vmcb12_bits = V_TPR_MASK | V_IRQ_INJECTION_BITS_MASK;
+	u32 int_ctl_vmcb01_bits = V_INTR_MASKING_MASK;
+	u32 int_ctl_vmcb12_bits = V_TPR_MASK | V_IRQ_INJECTION_BITS_MASK;
 
 	struct kvm_vcpu *vcpu = &svm->vcpu;
+	struct vmcb *vmcb01 = svm->vmcb01.ptr;
+	struct vmcb *vmcb02 = svm->nested.vmcb02.ptr;
 
 	/*
 	 * Filled at exit: exit_code, exit_code_hi, exit_info_1, exit_info_2,
 	 * exit_int_info, exit_int_info_err, next_rip, insn_len, insn_bytes.
 	 */
 
-	/*
-	 * Also covers avic_vapic_bar, avic_backing_page, avic_logical_id,
-	 * avic_physical_id.
-	 */
-	WARN_ON(kvm_apicv_activated(svm->vcpu.kvm));
+	if (svm->vgif_enabled && (svm->nested.ctl.int_ctl & V_GIF_ENABLE_MASK))
+		int_ctl_vmcb12_bits |= (V_GIF_MASK | V_GIF_ENABLE_MASK);
+	else
+		int_ctl_vmcb01_bits |= (V_GIF_MASK | V_GIF_ENABLE_MASK);
 
 	/* Copied from vmcb01.  msrpm_base can be overwritten later.  */
-	svm->vmcb->control.nested_ctl = svm->vmcb01.ptr->control.nested_ctl;
-	svm->vmcb->control.iopm_base_pa = svm->vmcb01.ptr->control.iopm_base_pa;
-	svm->vmcb->control.msrpm_base_pa = svm->vmcb01.ptr->control.msrpm_base_pa;
+	vmcb02->control.nested_ctl = vmcb01->control.nested_ctl;
+	vmcb02->control.iopm_base_pa = vmcb01->control.iopm_base_pa;
+	vmcb02->control.msrpm_base_pa = vmcb01->control.msrpm_base_pa;
 
 	/* Done at vmrun: asid.  */
 
 	/* Also overwritten later if necessary.  */
-	svm->vmcb->control.tlb_ctl = TLB_CONTROL_DO_NOTHING;
+	vmcb02->control.tlb_ctl = TLB_CONTROL_DO_NOTHING;
 
 	/* nested_cr3.  */
 	if (nested_npt_enabled(svm))
@@ -605,21 +646,53 @@ static void nested_vmcb02_prepare_control(struct vcpu_svm *svm)
 			svm->nested.ctl.tsc_offset,
 			svm->tsc_ratio_msr);
 
-	svm->vmcb->control.tsc_offset = vcpu->arch.tsc_offset;
+	vmcb02->control.tsc_offset = vcpu->arch.tsc_offset;
 
 	if (svm->tsc_ratio_msr != kvm_default_tsc_scaling_ratio) {
 		WARN_ON(!svm->tsc_scaling_enabled);
 		nested_svm_update_tsc_ratio_msr(vcpu);
 	}
 
-	svm->vmcb->control.int_ctl             =
+	vmcb02->control.int_ctl             =
 		(svm->nested.ctl.int_ctl & int_ctl_vmcb12_bits) |
-		(svm->vmcb01.ptr->control.int_ctl & int_ctl_vmcb01_bits);
-
-	svm->vmcb->control.int_vector          = svm->nested.ctl.int_vector;
-	svm->vmcb->control.int_state           = svm->nested.ctl.int_state;
-	svm->vmcb->control.event_inj           = svm->nested.ctl.event_inj;
-	svm->vmcb->control.event_inj_err       = svm->nested.ctl.event_inj_err;
+		(vmcb01->control.int_ctl & int_ctl_vmcb01_bits);
+
+	vmcb02->control.int_vector          = svm->nested.ctl.int_vector;
+	vmcb02->control.int_state           = svm->nested.ctl.int_state;
+	vmcb02->control.event_inj           = svm->nested.ctl.event_inj;
+	vmcb02->control.event_inj_err       = svm->nested.ctl.event_inj_err;
+
+	vmcb02->control.virt_ext            = vmcb01->control.virt_ext &
+					      LBR_CTL_ENABLE_MASK;
+	if (svm->lbrv_enabled)
+		vmcb02->control.virt_ext  |=
+			(svm->nested.ctl.virt_ext & LBR_CTL_ENABLE_MASK);
+
+	if (!nested_vmcb_needs_vls_intercept(svm))
+		vmcb02->control.virt_ext |= VIRTUAL_VMLOAD_VMSAVE_ENABLE_MASK;
+
+	if (kvm_pause_in_guest(svm->vcpu.kvm)) {
+		/* use guest values since host doesn't use them */
+		vmcb02->control.pause_filter_count =
+				svm->pause_filter_enabled ?
+				svm->nested.ctl.pause_filter_count : 0;
+
+		vmcb02->control.pause_filter_thresh =
+				svm->pause_threshold_enabled ?
+				svm->nested.ctl.pause_filter_thresh : 0;
+
+	} else if (!vmcb12_is_intercept(&svm->nested.ctl, INTERCEPT_PAUSE)) {
+		/* use host values when guest doesn't use them */
+		vmcb02->control.pause_filter_count = vmcb01->control.pause_filter_count;
+		vmcb02->control.pause_filter_thresh = vmcb01->control.pause_filter_thresh;
+	} else {
+		/*
+		 * Intercept every PAUSE otherwise and
+		 * ignore both host and guest values
+		 */
+		vmcb02->control.pause_filter_count = 0;
+		vmcb02->control.pause_filter_thresh = 0;
+	}
 
 	nested_svm_transition_tlb_flush(vcpu);
 
@@ -680,14 +753,14 @@ int enter_svm_guest_mode(struct kvm_vcpu *vcpu, u64 vmcb12_gpa,
 	if (ret)
 		return ret;
 
-	if (!npt_enabled)
-		vcpu->arch.mmu->inject_page_fault = svm_inject_page_fault_nested;
-
 	if (!from_vmrun)
 		kvm_make_request(KVM_REQ_GET_NESTED_STATE_PAGES, vcpu);
 
 	svm_set_gif(svm, true);
 
+	if (kvm_vcpu_apicv_active(vcpu))
+		kvm_make_request(KVM_REQ_APICV_UPDATE, vcpu);
+
 	return 0;
 }
 
@@ -698,6 +771,7 @@ int nested_svm_vmrun(struct kvm_vcpu *vcpu)
 	struct vmcb *vmcb12;
 	struct kvm_host_map map;
 	u64 vmcb12_gpa;
+	struct vmcb *vmcb01 = svm->vmcb01.ptr;
 
 	if (!svm->nested.hsave_msr) {
 		kvm_inject_gp(vcpu, 0);
@@ -741,14 +815,14 @@ int nested_svm_vmrun(struct kvm_vcpu *vcpu)
 	 * Since vmcb01 is not in use, we can use it to store some of the L1
 	 * state.
 	 */
-	svm->vmcb01.ptr->save.efer   = vcpu->arch.efer;
-	svm->vmcb01.ptr->save.cr0    = kvm_read_cr0(vcpu);
-	svm->vmcb01.ptr->save.cr4    = vcpu->arch.cr4;
-	svm->vmcb01.ptr->save.rflags = kvm_get_rflags(vcpu);
-	svm->vmcb01.ptr->save.rip    = kvm_rip_read(vcpu);
+	vmcb01->save.efer   = vcpu->arch.efer;
+	vmcb01->save.cr0    = kvm_read_cr0(vcpu);
+	vmcb01->save.cr4    = vcpu->arch.cr4;
+	vmcb01->save.rflags = kvm_get_rflags(vcpu);
+	vmcb01->save.rip    = kvm_rip_read(vcpu);
 
 	if (!npt_enabled)
-		svm->vmcb01.ptr->save.cr3 = kvm_read_cr3(vcpu);
+		vmcb01->save.cr3 = kvm_read_cr3(vcpu);
 
 	svm->nested.nested_run_pending = 1;
 
@@ -814,14 +888,12 @@ void svm_copy_vmloadsave_state(struct vmcb *to_vmcb, struct vmcb *from_vmcb)
 int nested_svm_vmexit(struct vcpu_svm *svm)
 {
 	struct kvm_vcpu *vcpu = &svm->vcpu;
+	struct vmcb *vmcb01 = svm->vmcb01.ptr;
+	struct vmcb *vmcb02 = svm->nested.vmcb02.ptr;
 	struct vmcb *vmcb12;
-	struct vmcb *vmcb = svm->vmcb;
 	struct kvm_host_map map;
 	int rc;
 
-	/* Triple faults in L2 should never escape. */
-	WARN_ON_ONCE(kvm_check_request(KVM_REQ_TRIPLE_FAULT, vcpu));
-
 	rc = kvm_vcpu_map(vcpu, gpa_to_gfn(svm->nested.vmcb12_gpa), &map);
 	if (rc) {
 		if (rc == -EINVAL)
@@ -843,57 +915,68 @@ int nested_svm_vmexit(struct vcpu_svm *svm)
 
 	/* Give the current vmcb to the guest */
 
-	vmcb12->save.es     = vmcb->save.es;
-	vmcb12->save.cs     = vmcb->save.cs;
-	vmcb12->save.ss     = vmcb->save.ss;
-	vmcb12->save.ds     = vmcb->save.ds;
-	vmcb12->save.gdtr   = vmcb->save.gdtr;
-	vmcb12->save.idtr   = vmcb->save.idtr;
+	vmcb12->save.es     = vmcb02->save.es;
+	vmcb12->save.cs     = vmcb02->save.cs;
+	vmcb12->save.ss     = vmcb02->save.ss;
+	vmcb12->save.ds     = vmcb02->save.ds;
+	vmcb12->save.gdtr   = vmcb02->save.gdtr;
+	vmcb12->save.idtr   = vmcb02->save.idtr;
 	vmcb12->save.efer   = svm->vcpu.arch.efer;
 	vmcb12->save.cr0    = kvm_read_cr0(vcpu);
 	vmcb12->save.cr3    = kvm_read_cr3(vcpu);
-	vmcb12->save.cr2    = vmcb->save.cr2;
+	vmcb12->save.cr2    = vmcb02->save.cr2;
 	vmcb12->save.cr4    = svm->vcpu.arch.cr4;
 	vmcb12->save.rflags = kvm_get_rflags(vcpu);
 	vmcb12->save.rip    = kvm_rip_read(vcpu);
 	vmcb12->save.rsp    = kvm_rsp_read(vcpu);
 	vmcb12->save.rax    = kvm_rax_read(vcpu);
-	vmcb12->save.dr7    = vmcb->save.dr7;
+	vmcb12->save.dr7    = vmcb02->save.dr7;
 	vmcb12->save.dr6    = svm->vcpu.arch.dr6;
-	vmcb12->save.cpl    = vmcb->save.cpl;
+	vmcb12->save.cpl    = vmcb02->save.cpl;
 
-	vmcb12->control.int_state         = vmcb->control.int_state;
-	vmcb12->control.exit_code         = vmcb->control.exit_code;
-	vmcb12->control.exit_code_hi      = vmcb->control.exit_code_hi;
-	vmcb12->control.exit_info_1       = vmcb->control.exit_info_1;
-	vmcb12->control.exit_info_2       = vmcb->control.exit_info_2;
+	vmcb12->control.int_state         = vmcb02->control.int_state;
+	vmcb12->control.exit_code         = vmcb02->control.exit_code;
+	vmcb12->control.exit_code_hi      = vmcb02->control.exit_code_hi;
+	vmcb12->control.exit_info_1       = vmcb02->control.exit_info_1;
+	vmcb12->control.exit_info_2       = vmcb02->control.exit_info_2;
 
 	if (vmcb12->control.exit_code != SVM_EXIT_ERR)
 		nested_save_pending_event_to_vmcb12(svm, vmcb12);
 
 	if (svm->nrips_enabled)
-		vmcb12->control.next_rip  = vmcb->control.next_rip;
+		vmcb12->control.next_rip  = vmcb02->control.next_rip;
 
 	vmcb12->control.int_ctl           = svm->nested.ctl.int_ctl;
 	vmcb12->control.tlb_ctl           = svm->nested.ctl.tlb_ctl;
 	vmcb12->control.event_inj         = svm->nested.ctl.event_inj;
 	vmcb12->control.event_inj_err     = svm->nested.ctl.event_inj_err;
 
+	if (!kvm_pause_in_guest(vcpu->kvm) && vmcb02->control.pause_filter_count)
+		vmcb01->control.pause_filter_count = vmcb02->control.pause_filter_count;
+
 	nested_svm_copy_common_state(svm->nested.vmcb02.ptr, svm->vmcb01.ptr);
 
 	svm_switch_vmcb(svm, &svm->vmcb01);
 
+	if (unlikely(svm->lbrv_enabled && (svm->nested.ctl.virt_ext & LBR_CTL_ENABLE_MASK))) {
+		svm_copy_lbrs(vmcb12, vmcb02);
+		svm_update_lbrv(vcpu);
+	} else if (unlikely(vmcb01->control.virt_ext & LBR_CTL_ENABLE_MASK)) {
+		svm_copy_lbrs(vmcb01, vmcb02);
+		svm_update_lbrv(vcpu);
+	}
+
 	/*
 	 * On vmexit the  GIF is set to false and
 	 * no event can be injected in L1.
 	 */
 	svm_set_gif(svm, false);
-	svm->vmcb->control.exit_int_info = 0;
+	vmcb01->control.exit_int_info = 0;
 
 	svm->vcpu.arch.tsc_offset = svm->vcpu.arch.l1_tsc_offset;
-	if (svm->vmcb->control.tsc_offset != svm->vcpu.arch.tsc_offset) {
-		svm->vmcb->control.tsc_offset = svm->vcpu.arch.tsc_offset;
-		vmcb_mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
+	if (vmcb01->control.tsc_offset != svm->vcpu.arch.tsc_offset) {
+		vmcb01->control.tsc_offset = svm->vcpu.arch.tsc_offset;
+		vmcb_mark_dirty(vmcb01, VMCB_INTERCEPTS);
 	}
 
 	if (svm->tsc_ratio_msr != kvm_default_tsc_scaling_ratio) {
@@ -907,13 +990,13 @@ int nested_svm_vmexit(struct vcpu_svm *svm)
 	/*
 	 * Restore processor state that had been saved in vmcb01
 	 */
-	kvm_set_rflags(vcpu, svm->vmcb->save.rflags);
-	svm_set_efer(vcpu, svm->vmcb->save.efer);
-	svm_set_cr0(vcpu, svm->vmcb->save.cr0 | X86_CR0_PE);
-	svm_set_cr4(vcpu, svm->vmcb->save.cr4);
-	kvm_rax_write(vcpu, svm->vmcb->save.rax);
-	kvm_rsp_write(vcpu, svm->vmcb->save.rsp);
-	kvm_rip_write(vcpu, svm->vmcb->save.rip);
+	kvm_set_rflags(vcpu, vmcb01->save.rflags);
+	svm_set_efer(vcpu, vmcb01->save.efer);
+	svm_set_cr0(vcpu, vmcb01->save.cr0 | X86_CR0_PE);
+	svm_set_cr4(vcpu, vmcb01->save.cr4);
+	kvm_rax_write(vcpu, vmcb01->save.rax);
+	kvm_rsp_write(vcpu, vmcb01->save.rsp);
+	kvm_rip_write(vcpu, vmcb01->save.rip);
 
 	svm->vcpu.arch.dr7 = DR7_FIXED_1;
 	kvm_update_dr7(&svm->vcpu);
@@ -931,7 +1014,7 @@ int nested_svm_vmexit(struct vcpu_svm *svm)
 
 	nested_svm_uninit_mmu_context(vcpu);
 
-	rc = nested_svm_load_cr3(vcpu, svm->vmcb->save.cr3, false, true);
+	rc = nested_svm_load_cr3(vcpu, vmcb01->save.cr3, false, true);
 	if (rc)
 		return 1;
 
@@ -949,9 +1032,16 @@ int nested_svm_vmexit(struct vcpu_svm *svm)
 	 * right now so that it an be accounted for before we execute
 	 * L1's next instruction.
 	 */
-	if (unlikely(svm->vmcb->save.rflags & X86_EFLAGS_TF))
+	if (unlikely(vmcb01->save.rflags & X86_EFLAGS_TF))
 		kvm_queue_exception(&(svm->vcpu), DB_VECTOR);
 
+	/*
+	 * Un-inhibit the AVIC right away, so that other vCPUs can start
+	 * to benefit from it right away.
+	 */
+	if (kvm_apicv_activated(vcpu->kvm))
+		kvm_vcpu_update_apicv(vcpu);
+
 	return 0;
 }
 
@@ -1162,12 +1252,13 @@ static bool nested_exit_on_exception(struct vcpu_svm *svm)
 static void nested_svm_inject_exception_vmexit(struct vcpu_svm *svm)
 {
 	unsigned int nr = svm->vcpu.arch.exception.nr;
+	struct vmcb *vmcb = svm->vmcb;
 
-	svm->vmcb->control.exit_code = SVM_EXIT_EXCP_BASE + nr;
-	svm->vmcb->control.exit_code_hi = 0;
+	vmcb->control.exit_code = SVM_EXIT_EXCP_BASE + nr;
+	vmcb->control.exit_code_hi = 0;
 
 	if (svm->vcpu.arch.exception.has_error_code)
-		svm->vmcb->control.exit_info_1 = svm->vcpu.arch.exception.error_code;
+		vmcb->control.exit_info_1 = svm->vcpu.arch.exception.error_code;
 
 	/*
 	 * EXITINFO2 is undefined for all exception intercepts other
@@ -1175,11 +1266,11 @@ static void nested_svm_inject_exception_vmexit(struct vcpu_svm *svm)
 	 */
 	if (nr == PF_VECTOR) {
 		if (svm->vcpu.arch.exception.nested_apf)
-			svm->vmcb->control.exit_info_2 = svm->vcpu.arch.apf.nested_apf_token;
+			vmcb->control.exit_info_2 = svm->vcpu.arch.apf.nested_apf_token;
 		else if (svm->vcpu.arch.exception.has_payload)
-			svm->vmcb->control.exit_info_2 = svm->vcpu.arch.exception.payload;
+			vmcb->control.exit_info_2 = svm->vcpu.arch.exception.payload;
 		else
-			svm->vmcb->control.exit_info_2 = svm->vcpu.arch.cr2;
+			vmcb->control.exit_info_2 = svm->vcpu.arch.cr2;
 	} else if (nr == DB_VECTOR) {
 		/* See inject_pending_event.  */
 		kvm_deliver_exception_payload(&svm->vcpu);
@@ -1567,6 +1658,7 @@ static bool svm_get_nested_state_pages(struct kvm_vcpu *vcpu)
 struct kvm_x86_nested_ops svm_nested_ops = {
 	.leave_nested = svm_leave_nested,
 	.check_events = svm_check_nested_events,
+	.handle_page_fault_workaround = nested_svm_handle_page_fault_workaround,
 	.triple_fault = nested_svm_triple_fault,
 	.get_nested_state_pages = svm_get_nested_state_pages,
 	.get_state = svm_get_nested_state,
diff --git a/arch/x86/kvm/svm/pmu.c b/arch/x86/kvm/svm/pmu.c
index 16a5ebb420cf..136039fc6d01 100644
--- a/arch/x86/kvm/svm/pmu.c
+++ b/arch/x86/kvm/svm/pmu.c
@@ -342,7 +342,7 @@ static void amd_pmu_reset(struct kvm_vcpu *vcpu)
 	}
 }
 
-struct kvm_pmu_ops amd_pmu_ops = {
+struct kvm_pmu_ops amd_pmu_ops __initdata = {
 	.pmc_perf_hw_id = amd_pmc_perf_hw_id,
 	.pmc_is_enabled = amd_pmc_is_enabled,
 	.pmc_idx_to_pmc = amd_pmc_idx_to_pmc,
diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index 636c77ef55fc..51fd985cf21d 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -688,7 +688,7 @@ static int sev_launch_measure(struct kvm *kvm, struct kvm_sev_cmd *argp)
 		if (params.len > SEV_FW_BLOB_MAX_SIZE)
 			return -EINVAL;
 
-		blob = kmalloc(params.len, GFP_KERNEL_ACCOUNT);
+		blob = kzalloc(params.len, GFP_KERNEL_ACCOUNT);
 		if (!blob)
 			return -ENOMEM;
 
@@ -808,7 +808,7 @@ static int __sev_dbg_decrypt_user(struct kvm *kvm, unsigned long paddr,
 	if (!IS_ALIGNED(dst_paddr, 16) ||
 	    !IS_ALIGNED(paddr,     16) ||
 	    !IS_ALIGNED(size,      16)) {
-		tpage = (void *)alloc_page(GFP_KERNEL);
+		tpage = (void *)alloc_page(GFP_KERNEL | __GFP_ZERO);
 		if (!tpage)
 			return -ENOMEM;
 
@@ -1094,7 +1094,7 @@ static int sev_get_attestation_report(struct kvm *kvm, struct kvm_sev_cmd *argp)
 		if (params.len > SEV_FW_BLOB_MAX_SIZE)
 			return -EINVAL;
 
-		blob = kmalloc(params.len, GFP_KERNEL_ACCOUNT);
+		blob = kzalloc(params.len, GFP_KERNEL_ACCOUNT);
 		if (!blob)
 			return -ENOMEM;
 
@@ -1176,7 +1176,7 @@ static int sev_send_start(struct kvm *kvm, struct kvm_sev_cmd *argp)
 		return -EINVAL;
 
 	/* allocate the memory to hold the session data blob */
-	session_data = kmalloc(params.session_len, GFP_KERNEL_ACCOUNT);
+	session_data = kzalloc(params.session_len, GFP_KERNEL_ACCOUNT);
 	if (!session_data)
 		return -ENOMEM;
 
@@ -1300,11 +1300,11 @@ static int sev_send_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp)
 
 	/* allocate memory for header and transport buffer */
 	ret = -ENOMEM;
-	hdr = kmalloc(params.hdr_len, GFP_KERNEL_ACCOUNT);
+	hdr = kzalloc(params.hdr_len, GFP_KERNEL_ACCOUNT);
 	if (!hdr)
 		goto e_unpin;
 
-	trans_data = kmalloc(params.trans_len, GFP_KERNEL_ACCOUNT);
+	trans_data = kzalloc(params.trans_len, GFP_KERNEL_ACCOUNT);
 	if (!trans_data)
 		goto e_free_hdr;
 
@@ -2769,8 +2769,12 @@ static int sev_handle_vmgexit_msr_protocol(struct vcpu_svm *svm)
 		pr_info("SEV-ES guest requested termination: %#llx:%#llx\n",
 			reason_set, reason_code);
 
-		ret = -EINVAL;
-		break;
+		vcpu->run->exit_reason = KVM_EXIT_SYSTEM_EVENT;
+		vcpu->run->system_event.type = KVM_SYSTEM_EVENT_SEV_TERM;
+		vcpu->run->system_event.ndata = 1;
+		vcpu->run->system_event.data[0] = control->ghcb_gpa;
+
+		return 0;
 	}
 	default:
 		/* Error, keep GHCB MSR value as-is */
@@ -2953,6 +2957,14 @@ void sev_es_init_vmcb(struct vcpu_svm *svm)
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTBRANCHTOIP, 1, 1);
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTINTFROMIP, 1, 1);
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTINTTOIP, 1, 1);
+
+	if (boot_cpu_has(X86_FEATURE_V_TSC_AUX) &&
+	    (guest_cpuid_has(&svm->vcpu, X86_FEATURE_RDTSCP) ||
+	     guest_cpuid_has(&svm->vcpu, X86_FEATURE_RDPID))) {
+		set_msr_interception(vcpu, svm->msrpm, MSR_TSC_AUX, 1, 1);
+		if (guest_cpuid_has(&svm->vcpu, X86_FEATURE_RDTSCP))
+			svm_clr_intercept(svm, INTERCEPT_RDTSCP);
+	}
 }
 
 void sev_es_vcpu_reset(struct vcpu_svm *svm)
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index 17d334ef5430..200045f71df0 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -62,8 +62,6 @@ MODULE_DEVICE_TABLE(x86cpu, svm_cpu_id);
 #define SEG_TYPE_LDT 2
 #define SEG_TYPE_BUSY_TSS16 3
 
-#define DEBUGCTL_RESERVED_BITS (~(0x3fULL))
-
 static bool erratum_383_found __read_mostly;
 
 u32 msrpm_offsets[MSRPM_OFFSETS] __read_mostly;
@@ -101,6 +99,7 @@ static const struct svm_direct_access_msrs {
 	{ .index = MSR_EFER,				.always = false },
 	{ .index = MSR_IA32_CR_PAT,			.always = false },
 	{ .index = MSR_AMD64_SEV_ES_GHCB,		.always = true  },
+	{ .index = MSR_TSC_AUX,				.always = false },
 	{ .index = MSR_INVALID,				.always = false },
 };
 
@@ -172,7 +171,7 @@ static int vls = true;
 module_param(vls, int, 0444);
 
 /* enable/disable Virtual GIF */
-static int vgif = true;
+int vgif = true;
 module_param(vgif, int, 0444);
 
 /* enable/disable LBR virtualization */
@@ -189,6 +188,9 @@ module_param(tsc_scaling, int, 0444);
 static bool avic;
 module_param(avic, bool, 0444);
 
+static bool force_avic;
+module_param_unsafe(force_avic, bool, 0444);
+
 bool __read_mostly dump_invalid_vmcb;
 module_param(dump_invalid_vmcb, bool, 0644);
 
@@ -790,6 +792,17 @@ static void init_msrpm_offsets(void)
 	}
 }
 
+void svm_copy_lbrs(struct vmcb *to_vmcb, struct vmcb *from_vmcb)
+{
+	to_vmcb->save.dbgctl		= from_vmcb->save.dbgctl;
+	to_vmcb->save.br_from		= from_vmcb->save.br_from;
+	to_vmcb->save.br_to		= from_vmcb->save.br_to;
+	to_vmcb->save.last_excp_from	= from_vmcb->save.last_excp_from;
+	to_vmcb->save.last_excp_to	= from_vmcb->save.last_excp_to;
+
+	vmcb_mark_dirty(to_vmcb, VMCB_LBR);
+}
+
 static void svm_enable_lbrv(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_svm *svm = to_svm(vcpu);
@@ -799,6 +812,10 @@ static void svm_enable_lbrv(struct kvm_vcpu *vcpu)
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTBRANCHTOIP, 1, 1);
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTINTFROMIP, 1, 1);
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTINTTOIP, 1, 1);
+
+	/* Move the LBR msrs to the vmcb02 so that the guest can see them. */
+	if (is_guest_mode(vcpu))
+		svm_copy_lbrs(svm->vmcb, svm->vmcb01.ptr);
 }
 
 static void svm_disable_lbrv(struct kvm_vcpu *vcpu)
@@ -810,6 +827,67 @@ static void svm_disable_lbrv(struct kvm_vcpu *vcpu)
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTBRANCHTOIP, 0, 0);
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTINTFROMIP, 0, 0);
 	set_msr_interception(vcpu, svm->msrpm, MSR_IA32_LASTINTTOIP, 0, 0);
+
+	/*
+	 * Move the LBR msrs back to the vmcb01 to avoid copying them
+	 * on nested guest entries.
+	 */
+	if (is_guest_mode(vcpu))
+		svm_copy_lbrs(svm->vmcb01.ptr, svm->vmcb);
+}
+
+static int svm_get_lbr_msr(struct vcpu_svm *svm, u32 index)
+{
+	/*
+	 * If the LBR virtualization is disabled, the LBR msrs are always
+	 * kept in the vmcb01 to avoid copying them on nested guest entries.
+	 *
+	 * If nested, and the LBR virtualization is enabled/disabled, the msrs
+	 * are moved between the vmcb01 and vmcb02 as needed.
+	 */
+	struct vmcb *vmcb =
+		(svm->vmcb->control.virt_ext & LBR_CTL_ENABLE_MASK) ?
+			svm->vmcb : svm->vmcb01.ptr;
+
+	switch (index) {
+	case MSR_IA32_DEBUGCTLMSR:
+		return vmcb->save.dbgctl;
+	case MSR_IA32_LASTBRANCHFROMIP:
+		return vmcb->save.br_from;
+	case MSR_IA32_LASTBRANCHTOIP:
+		return vmcb->save.br_to;
+	case MSR_IA32_LASTINTFROMIP:
+		return vmcb->save.last_excp_from;
+	case MSR_IA32_LASTINTTOIP:
+		return vmcb->save.last_excp_to;
+	default:
+		KVM_BUG(false, svm->vcpu.kvm,
+			"%s: Unknown MSR 0x%x", __func__, index);
+		return 0;
+	}
+}
+
+void svm_update_lbrv(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	bool enable_lbrv = svm_get_lbr_msr(svm, MSR_IA32_DEBUGCTLMSR) &
+					   DEBUGCTLMSR_LBR;
+
+	bool current_enable_lbrv = !!(svm->vmcb->control.virt_ext &
+				      LBR_CTL_ENABLE_MASK);
+
+	if (unlikely(is_guest_mode(vcpu) && svm->lbrv_enabled))
+		if (unlikely(svm->nested.ctl.virt_ext & LBR_CTL_ENABLE_MASK))
+			enable_lbrv = true;
+
+	if (enable_lbrv == current_enable_lbrv)
+		return;
+
+	if (enable_lbrv)
+		svm_enable_lbrv(vcpu);
+	else
+		svm_disable_lbrv(vcpu);
 }
 
 void disable_nmi_singlestep(struct vcpu_svm *svm)
@@ -831,6 +909,9 @@ static void grow_ple_window(struct kvm_vcpu *vcpu)
 	struct vmcb_control_area *control = &svm->vmcb->control;
 	int old = control->pause_filter_count;
 
+	if (kvm_pause_in_guest(vcpu->kvm) || !old)
+		return;
+
 	control->pause_filter_count = __grow_ple_window(old,
 							pause_filter_count,
 							pause_filter_count_grow,
@@ -849,6 +930,9 @@ static void shrink_ple_window(struct kvm_vcpu *vcpu)
 	struct vmcb_control_area *control = &svm->vmcb->control;
 	int old = control->pause_filter_count;
 
+	if (kvm_pause_in_guest(vcpu->kvm) || !old)
+		return;
+
 	control->pause_filter_count =
 				__shrink_ple_window(old,
 						    pause_filter_count,
@@ -960,6 +1044,8 @@ static inline void init_vmcb_after_set_cpuid(struct kvm_vcpu *vcpu)
 
 		set_msr_interception(vcpu, svm->msrpm, MSR_IA32_SYSENTER_EIP, 0, 0);
 		set_msr_interception(vcpu, svm->msrpm, MSR_IA32_SYSENTER_ESP, 0, 0);
+
+		svm->v_vmload_vmsave_enabled = false;
 	} else {
 		/*
 		 * If hardware supports Virtual VMLOAD VMSAVE then enable it
@@ -979,8 +1065,9 @@ static inline void init_vmcb_after_set_cpuid(struct kvm_vcpu *vcpu)
 static void init_vmcb(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_svm *svm = to_svm(vcpu);
-	struct vmcb_control_area *control = &svm->vmcb->control;
-	struct vmcb_save_area *save = &svm->vmcb->save;
+	struct vmcb *vmcb = svm->vmcb01.ptr;
+	struct vmcb_control_area *control = &vmcb->control;
+	struct vmcb_save_area *save = &vmcb->save;
 
 	svm_set_intercept(svm, INTERCEPT_CR0_READ);
 	svm_set_intercept(svm, INTERCEPT_CR3_READ);
@@ -1104,7 +1191,7 @@ static void init_vmcb(struct kvm_vcpu *vcpu)
 		set_msr_interception(vcpu, svm->msrpm, MSR_IA32_SPEC_CTRL, 1, 1);
 
 	if (kvm_vcpu_apicv_active(vcpu))
-		avic_init_vmcb(svm);
+		avic_init_vmcb(svm, vmcb);
 
 	if (vgif) {
 		svm_clr_intercept(svm, INTERCEPT_STGI);
@@ -1122,10 +1209,10 @@ static void init_vmcb(struct kvm_vcpu *vcpu)
 		}
 	}
 
-	svm_hv_init_vmcb(svm->vmcb);
+	svm_hv_init_vmcb(vmcb);
 	init_vmcb_after_set_cpuid(vcpu);
 
-	vmcb_mark_all_dirty(svm->vmcb);
+	vmcb_mark_all_dirty(vmcb);
 
 	enable_gif(svm);
 }
@@ -1380,7 +1467,7 @@ static void svm_set_vintr(struct vcpu_svm *svm)
 	/*
 	 * The following fields are ignored when AVIC is enabled
 	 */
-	WARN_ON(kvm_apicv_activated(svm->vcpu.kvm));
+	WARN_ON(kvm_vcpu_apicv_activated(&svm->vcpu));
 
 	svm_set_intercept(svm, INTERCEPT_VINTR);
 
@@ -2142,7 +2229,7 @@ void svm_set_gif(struct vcpu_svm *svm, bool value)
 		 * Likewise, clear the VINTR intercept, we will set it
 		 * again while processing KVM_REQ_EVENT if needed.
 		 */
-		if (vgif_enabled(svm))
+		if (vgif)
 			svm_clr_intercept(svm, INTERCEPT_STGI);
 		if (svm_is_intercept(svm, INTERCEPT_VINTR))
 			svm_clear_vintr(svm);
@@ -2160,7 +2247,7 @@ void svm_set_gif(struct vcpu_svm *svm, bool value)
 		 * in use, we still rely on the VINTR intercept (rather than
 		 * STGI) to detect an open interrupt window.
 		*/
-		if (!vgif_enabled(svm))
+		if (!vgif)
 			svm_clear_vintr(svm);
 	}
 }
@@ -2575,25 +2662,12 @@ static int svm_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 	case MSR_TSC_AUX:
 		msr_info->data = svm->tsc_aux;
 		break;
-	/*
-	 * Nobody will change the following 5 values in the VMCB so we can
-	 * safely return them on rdmsr. They will always be 0 until LBRV is
-	 * implemented.
-	 */
 	case MSR_IA32_DEBUGCTLMSR:
-		msr_info->data = svm->vmcb->save.dbgctl;
-		break;
 	case MSR_IA32_LASTBRANCHFROMIP:
-		msr_info->data = svm->vmcb->save.br_from;
-		break;
 	case MSR_IA32_LASTBRANCHTOIP:
-		msr_info->data = svm->vmcb->save.br_to;
-		break;
 	case MSR_IA32_LASTINTFROMIP:
-		msr_info->data = svm->vmcb->save.last_excp_from;
-		break;
 	case MSR_IA32_LASTINTTOIP:
-		msr_info->data = svm->vmcb->save.last_excp_to;
+		msr_info->data = svm_get_lbr_msr(svm, msr_info->index);
 		break;
 	case MSR_VM_HSAVE_PA:
 		msr_info->data = svm->nested.hsave_msr;
@@ -2839,12 +2913,13 @@ static int svm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr)
 		if (data & DEBUGCTL_RESERVED_BITS)
 			return 1;
 
-		svm->vmcb->save.dbgctl = data;
-		vmcb_mark_dirty(svm->vmcb, VMCB_LBR);
-		if (data & (1ULL<<0))
-			svm_enable_lbrv(vcpu);
+		if (svm->vmcb->control.virt_ext & LBR_CTL_ENABLE_MASK)
+			svm->vmcb->save.dbgctl = data;
 		else
-			svm_disable_lbrv(vcpu);
+			svm->vmcb01.ptr->save.dbgctl = data;
+
+		svm_update_lbrv(vcpu);
+
 		break;
 	case MSR_VM_HSAVE_PA:
 		/*
@@ -2901,9 +2976,16 @@ static int interrupt_window_interception(struct kvm_vcpu *vcpu)
 	svm_clear_vintr(to_svm(vcpu));
 
 	/*
-	 * For AVIC, the only reason to end up here is ExtINTs.
+	 * If not running nested, for AVIC, the only reason to end up here is ExtINTs.
 	 * In this case AVIC was temporarily disabled for
 	 * requesting the IRQ window and we have to re-enable it.
+	 *
+	 * If running nested, still remove the VM wide AVIC inhibit to
+	 * support case in which the interrupt window was requested when the
+	 * vCPU was not running nested.
+
+	 * All vCPUs which run still run nested, will remain to have their
+	 * AVIC still inhibited due to per-cpu AVIC inhibition.
 	 */
 	kvm_clear_apicv_inhibit(vcpu->kvm, APICV_INHIBIT_REASON_IRQWIN);
 
@@ -2914,7 +2996,6 @@ static int interrupt_window_interception(struct kvm_vcpu *vcpu)
 static int pause_interception(struct kvm_vcpu *vcpu)
 {
 	bool in_kernel;
-
 	/*
 	 * CPL is not made available for an SEV-ES guest, therefore
 	 * vcpu->arch.preempted_in_kernel can never be true.  Just
@@ -2922,8 +3003,7 @@ static int pause_interception(struct kvm_vcpu *vcpu)
 	 */
 	in_kernel = !sev_es_guest(vcpu->kvm) && svm_get_cpl(vcpu) == 0;
 
-	if (!kvm_pause_in_guest(vcpu->kvm))
-		grow_ple_window(vcpu);
+	grow_ple_window(vcpu);
 
 	kvm_vcpu_on_spin(vcpu, in_kernel);
 	return kvm_skip_emulated_instruction(vcpu);
@@ -3496,14 +3576,20 @@ static void svm_enable_irq_window(struct kvm_vcpu *vcpu)
 	 * enabled, the STGI interception will not occur. Enable the irq
 	 * window under the assumption that the hardware will set the GIF.
 	 */
-	if (vgif_enabled(svm) || gif_set(svm)) {
+	if (vgif || gif_set(svm)) {
 		/*
 		 * IRQ window is not needed when AVIC is enabled,
 		 * unless we have pending ExtINT since it cannot be injected
-		 * via AVIC. In such case, we need to temporarily disable AVIC,
+		 * via AVIC. In such case, KVM needs to temporarily disable AVIC,
 		 * and fallback to injecting IRQ via V_IRQ.
+		 *
+		 * If running nested, AVIC is already locally inhibited
+		 * on this vCPU, therefore there is no need to request
+		 * the VM wide AVIC inhibition.
 		 */
-		kvm_set_apicv_inhibit(vcpu->kvm, APICV_INHIBIT_REASON_IRQWIN);
+		if (!is_guest_mode(vcpu))
+			kvm_set_apicv_inhibit(vcpu->kvm, APICV_INHIBIT_REASON_IRQWIN);
+
 		svm_set_vintr(svm);
 	}
 }
@@ -3516,7 +3602,7 @@ static void svm_enable_nmi_window(struct kvm_vcpu *vcpu)
 		return; /* IRET will cause a vm exit */
 
 	if (!gif_set(svm)) {
-		if (vgif_enabled(svm))
+		if (vgif)
 			svm_set_intercept(svm, INTERCEPT_STGI);
 		return; /* STGI will cause a vm exit */
 	}
@@ -3865,7 +3951,7 @@ static void svm_load_mmu_pgd(struct kvm_vcpu *vcpu, hpa_t root_hpa,
 		hv_track_root_tdp(vcpu, root_hpa);
 
 		cr3 = vcpu->arch.cr3;
-	} else if (vcpu->arch.mmu->shadow_root_level >= PT64_ROOT_4LEVEL) {
+	} else if (vcpu->arch.mmu->root_role.level >= PT64_ROOT_4LEVEL) {
 		cr3 = __sme_set(root_hpa) | kvm_get_active_pcid(vcpu);
 	} else {
 		/* PCID in the guest should be impossible with a 32-bit MMU. */
@@ -3946,6 +4032,17 @@ static void svm_vcpu_after_set_cpuid(struct kvm_vcpu *vcpu)
 			     guest_cpuid_has(vcpu, X86_FEATURE_NRIPS);
 
 	svm->tsc_scaling_enabled = tsc_scaling && guest_cpuid_has(vcpu, X86_FEATURE_TSCRATEMSR);
+	svm->lbrv_enabled = lbrv && guest_cpuid_has(vcpu, X86_FEATURE_LBRV);
+
+	svm->v_vmload_vmsave_enabled = vls && guest_cpuid_has(vcpu, X86_FEATURE_V_VMSAVE_VMLOAD);
+
+	svm->pause_filter_enabled = kvm_cpu_cap_has(X86_FEATURE_PAUSEFILTER) &&
+			guest_cpuid_has(vcpu, X86_FEATURE_PAUSEFILTER);
+
+	svm->pause_threshold_enabled = kvm_cpu_cap_has(X86_FEATURE_PFTHRESHOLD) &&
+			guest_cpuid_has(vcpu, X86_FEATURE_PFTHRESHOLD);
+
+	svm->vgif_enabled = vgif && guest_cpuid_has(vcpu, X86_FEATURE_VGIF);
 
 	svm_recalc_instruction_intercepts(vcpu, svm);
 
@@ -3963,13 +4060,6 @@ static void svm_vcpu_after_set_cpuid(struct kvm_vcpu *vcpu)
 		 */
 		if (guest_cpuid_has(vcpu, X86_FEATURE_X2APIC))
 			kvm_set_apicv_inhibit(kvm, APICV_INHIBIT_REASON_X2APIC);
-
-		/*
-		 * Currently, AVIC does not work with nested virtualization.
-		 * So, we disable AVIC when cpuid for SVM is set in the L1 guest.
-		 */
-		if (nested && guest_cpuid_has(vcpu, X86_FEATURE_SVM))
-			kvm_set_apicv_inhibit(kvm, APICV_INHIBIT_REASON_NESTED);
 	}
 	init_vmcb_after_set_cpuid(vcpu);
 }
@@ -4224,7 +4314,7 @@ static int svm_enter_smm(struct kvm_vcpu *vcpu, char *smstate)
 	svm->vmcb->save.rsp = vcpu->arch.regs[VCPU_REGS_RSP];
 	svm->vmcb->save.rip = vcpu->arch.regs[VCPU_REGS_RIP];
 
-	ret = nested_svm_vmexit(svm);
+	ret = nested_svm_simple_vmexit(svm, SVM_EXIT_SW);
 	if (ret)
 		return ret;
 
@@ -4321,7 +4411,7 @@ static void svm_enable_smi_window(struct kvm_vcpu *vcpu)
 	struct vcpu_svm *svm = to_svm(vcpu);
 
 	if (!gif_set(svm)) {
-		if (vgif_enabled(svm))
+		if (vgif)
 			svm_set_intercept(svm, INTERCEPT_STGI);
 		/* STGI will cause a vm exit */
 	} else {
@@ -4605,7 +4695,6 @@ static struct kvm_x86_ops svm_x86_ops __initdata = {
 
 	.sched_in = svm_sched_in,
 
-	.pmu_ops = &amd_pmu_ops,
 	.nested_ops = &svm_nested_ops,
 
 	.deliver_interrupt = svm_deliver_interrupt,
@@ -4633,6 +4722,7 @@ static struct kvm_x86_ops svm_x86_ops __initdata = {
 	.complete_emulated_msr = svm_complete_emulated_msr,
 
 	.vcpu_deliver_sipi_vector = svm_vcpu_deliver_sipi_vector,
+	.vcpu_get_apicv_inhibit_reasons = avic_vcpu_get_apicv_inhibit_reasons,
 };
 
 /*
@@ -4696,6 +4786,20 @@ static __init void svm_set_cpu_caps(void)
 		if (tsc_scaling)
 			kvm_cpu_cap_set(X86_FEATURE_TSCRATEMSR);
 
+		if (vls)
+			kvm_cpu_cap_set(X86_FEATURE_V_VMSAVE_VMLOAD);
+		if (lbrv)
+			kvm_cpu_cap_set(X86_FEATURE_LBRV);
+
+		if (boot_cpu_has(X86_FEATURE_PAUSEFILTER))
+			kvm_cpu_cap_set(X86_FEATURE_PAUSEFILTER);
+
+		if (boot_cpu_has(X86_FEATURE_PFTHRESHOLD))
+			kvm_cpu_cap_set(X86_FEATURE_PFTHRESHOLD);
+
+		if (vgif)
+			kvm_cpu_cap_set(X86_FEATURE_VGIF);
+
 		/* Nested VM can receive #VMEXIT instead of triggering #GP */
 		kvm_cpu_cap_set(X86_FEATURE_SVME_ADDR_CHK);
 	}
@@ -4789,6 +4893,9 @@ static __init int svm_hardware_setup(void)
 			  get_npt_level(), PG_LEVEL_1G);
 	pr_info("kvm: Nested Paging %sabled\n", npt_enabled ? "en" : "dis");
 
+	/* Setup shadow_me_value and shadow_me_mask */
+	kvm_mmu_set_me_spte_mask(sme_me_mask, sme_me_mask);
+
 	/* Note, SEV setup consumes npt_enabled. */
 	sev_hardware_setup();
 
@@ -4807,15 +4914,20 @@ static __init int svm_hardware_setup(void)
 			nrips = false;
 	}
 
-	enable_apicv = avic = avic && npt_enabled && boot_cpu_has(X86_FEATURE_AVIC);
+	enable_apicv = avic = avic && npt_enabled && (boot_cpu_has(X86_FEATURE_AVIC) || force_avic);
 
 	if (enable_apicv) {
-		pr_info("AVIC enabled\n");
+		if (!boot_cpu_has(X86_FEATURE_AVIC)) {
+			pr_warn("AVIC is not supported in CPUID but force enabled");
+			pr_warn("Your system might crash and burn");
+		} else
+			pr_info("AVIC enabled\n");
 
 		amd_iommu_register_ga_log_notifier(&avic_ga_log_notifier);
 	} else {
 		svm_x86_ops.vcpu_blocking = NULL;
 		svm_x86_ops.vcpu_unblocking = NULL;
+		svm_x86_ops.vcpu_get_apicv_inhibit_reasons = NULL;
 	}
 
 	if (vls) {
@@ -4880,6 +4992,7 @@ static struct kvm_x86_init_ops svm_init_ops __initdata = {
 	.check_processor_compatibility = svm_check_processor_compat,
 
 	.runtime_ops = &svm_x86_ops,
+	.pmu_ops = &amd_pmu_ops,
 };
 
 static int __init svm_init(void)
diff --git a/arch/x86/kvm/svm/svm.h b/arch/x86/kvm/svm/svm.h
index 2d83845b9032..21c5460e947a 100644
--- a/arch/x86/kvm/svm/svm.h
+++ b/arch/x86/kvm/svm/svm.h
@@ -29,10 +29,11 @@
 #define	IOPM_SIZE PAGE_SIZE * 3
 #define	MSRPM_SIZE PAGE_SIZE * 2
 
-#define MAX_DIRECT_ACCESS_MSRS	20
+#define MAX_DIRECT_ACCESS_MSRS	21
 #define MSRPM_OFFSETS	16
 extern u32 msrpm_offsets[MSRPM_OFFSETS] __read_mostly;
 extern bool npt_enabled;
+extern int vgif;
 extern bool intercept_smi;
 
 /*
@@ -231,9 +232,14 @@ struct vcpu_svm {
 	unsigned int3_injected;
 	unsigned long int3_rip;
 
-	/* cached guest cpuid flags for faster access */
+	/* optional nested SVM features that are enabled for this guest  */
 	bool nrips_enabled                : 1;
 	bool tsc_scaling_enabled          : 1;
+	bool v_vmload_vmsave_enabled      : 1;
+	bool lbrv_enabled                 : 1;
+	bool pause_filter_enabled         : 1;
+	bool pause_threshold_enabled      : 1;
+	bool vgif_enabled                 : 1;
 
 	u32 ldr_reg;
 	u32 dfr_reg;
@@ -452,44 +458,70 @@ static inline bool svm_is_intercept(struct vcpu_svm *svm, int bit)
 	return vmcb_is_intercept(&svm->vmcb->control, bit);
 }
 
-static inline bool vgif_enabled(struct vcpu_svm *svm)
+static inline bool nested_vgif_enabled(struct vcpu_svm *svm)
 {
-	return !!(svm->vmcb->control.int_ctl & V_GIF_ENABLE_MASK);
+	return svm->vgif_enabled && (svm->nested.ctl.int_ctl & V_GIF_ENABLE_MASK);
+}
+
+static inline struct vmcb *get_vgif_vmcb(struct vcpu_svm *svm)
+{
+	if (!vgif)
+		return NULL;
+
+	if (is_guest_mode(&svm->vcpu) && !nested_vgif_enabled(svm))
+		return svm->nested.vmcb02.ptr;
+	else
+		return svm->vmcb01.ptr;
 }
 
 static inline void enable_gif(struct vcpu_svm *svm)
 {
-	if (vgif_enabled(svm))
-		svm->vmcb->control.int_ctl |= V_GIF_MASK;
+	struct vmcb *vmcb = get_vgif_vmcb(svm);
+
+	if (vmcb)
+		vmcb->control.int_ctl |= V_GIF_MASK;
 	else
 		svm->vcpu.arch.hflags |= HF_GIF_MASK;
 }
 
 static inline void disable_gif(struct vcpu_svm *svm)
 {
-	if (vgif_enabled(svm))
-		svm->vmcb->control.int_ctl &= ~V_GIF_MASK;
+	struct vmcb *vmcb = get_vgif_vmcb(svm);
+
+	if (vmcb)
+		vmcb->control.int_ctl &= ~V_GIF_MASK;
 	else
 		svm->vcpu.arch.hflags &= ~HF_GIF_MASK;
 }
 
 static inline bool gif_set(struct vcpu_svm *svm)
 {
-	if (vgif_enabled(svm))
-		return !!(svm->vmcb->control.int_ctl & V_GIF_MASK);
+	struct vmcb *vmcb = get_vgif_vmcb(svm);
+
+	if (vmcb)
+		return !!(vmcb->control.int_ctl & V_GIF_MASK);
 	else
 		return !!(svm->vcpu.arch.hflags & HF_GIF_MASK);
 }
 
+static inline bool nested_npt_enabled(struct vcpu_svm *svm)
+{
+	return svm->nested.ctl.nested_ctl & SVM_NESTED_CTL_NP_ENABLE;
+}
+
 /* svm.c */
 #define MSR_INVALID				0xffffffffU
 
+#define DEBUGCTL_RESERVED_BITS (~(0x3fULL))
+
 extern bool dump_invalid_vmcb;
 
 u32 svm_msrpm_offset(u32 msr);
 u32 *svm_vcpu_alloc_msrpm(void);
 void svm_vcpu_init_msrpm(struct kvm_vcpu *vcpu, u32 *msrpm);
 void svm_vcpu_free_msrpm(u32 *msrpm);
+void svm_copy_lbrs(struct vmcb *to_vmcb, struct vmcb *from_vmcb);
+void svm_update_lbrv(struct kvm_vcpu *vcpu);
 
 int svm_set_efer(struct kvm_vcpu *vcpu, u64 efer);
 void svm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
@@ -574,7 +606,7 @@ extern struct kvm_x86_nested_ops svm_nested_ops;
 int avic_ga_log_notifier(u32 ga_tag);
 void avic_vm_destroy(struct kvm *kvm);
 int avic_vm_init(struct kvm *kvm);
-void avic_init_vmcb(struct vcpu_svm *svm);
+void avic_init_vmcb(struct vcpu_svm *svm, struct vmcb *vmcb);
 int avic_incomplete_ipi_interception(struct kvm_vcpu *vcpu);
 int avic_unaccelerated_access_interception(struct kvm_vcpu *vcpu);
 int avic_init_vcpu(struct vcpu_svm *svm);
@@ -592,6 +624,7 @@ int avic_pi_update_irte(struct kvm *kvm, unsigned int host_irq,
 void avic_vcpu_blocking(struct kvm_vcpu *vcpu);
 void avic_vcpu_unblocking(struct kvm_vcpu *vcpu);
 void avic_ring_doorbell(struct kvm_vcpu *vcpu);
+unsigned long avic_vcpu_get_apicv_inhibit_reasons(struct kvm_vcpu *vcpu);
 
 /* sev.c */
 
diff --git a/arch/x86/kvm/trace.h b/arch/x86/kvm/trace.h
index e3a24b8f04be..de4762517569 100644
--- a/arch/x86/kvm/trace.h
+++ b/arch/x86/kvm/trace.h
@@ -1459,6 +1459,26 @@ TRACE_EVENT(kvm_avic_ga_log,
 		  __entry->vmid, __entry->vcpuid)
 );
 
+TRACE_EVENT(kvm_avic_kick_vcpu_slowpath,
+	    TP_PROTO(u32 icrh, u32 icrl, u32 index),
+	    TP_ARGS(icrh, icrl, index),
+
+	TP_STRUCT__entry(
+		__field(u32, icrh)
+		__field(u32, icrl)
+		__field(u32, index)
+	),
+
+	TP_fast_assign(
+		__entry->icrh = icrh;
+		__entry->icrl = icrl;
+		__entry->index = index;
+	),
+
+	TP_printk("icrh:icrl=%#08x:%08x, index=%u",
+		  __entry->icrh, __entry->icrl, __entry->index)
+);
+
 TRACE_EVENT(kvm_hv_timer_state,
 		TP_PROTO(unsigned int vcpu_id, unsigned int hv_timer_in_use),
 		TP_ARGS(vcpu_id, hv_timer_in_use),
diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index 856c87563883..f5cb18e00e78 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -476,24 +476,23 @@ static int nested_vmx_check_exception(struct kvm_vcpu *vcpu, unsigned long *exit
 	return 0;
 }
 
-
-static void vmx_inject_page_fault_nested(struct kvm_vcpu *vcpu,
-		struct x86_exception *fault)
+static bool nested_vmx_handle_page_fault_workaround(struct kvm_vcpu *vcpu,
+						    struct x86_exception *fault)
 {
 	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
 
 	WARN_ON(!is_guest_mode(vcpu));
 
 	if (nested_vmx_is_page_fault_vmexit(vmcs12, fault->error_code) &&
-		!to_vmx(vcpu)->nested.nested_run_pending) {
+	    !WARN_ON_ONCE(to_vmx(vcpu)->nested.nested_run_pending)) {
 		vmcs12->vm_exit_intr_error_code = fault->error_code;
 		nested_vmx_vmexit(vcpu, EXIT_REASON_EXCEPTION_NMI,
 				  PF_VECTOR | INTR_TYPE_HARD_EXCEPTION |
 				  INTR_INFO_DELIVER_CODE_MASK | INTR_INFO_VALID_MASK,
 				  fault->address);
-	} else {
-		kvm_inject_page_fault(vcpu, fault);
+		return true;
 	}
+	return false;
 }
 
 static int nested_vmx_check_io_bitmap_controls(struct kvm_vcpu *vcpu,
@@ -2614,9 +2613,6 @@ static int prepare_vmcs02(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12,
 		vmcs_write64(GUEST_PDPTR3, vmcs12->guest_pdptr3);
 	}
 
-	if (!enable_ept)
-		vcpu->arch.walk_mmu->inject_page_fault = vmx_inject_page_fault_nested;
-
 	if ((vmcs12->vm_entry_controls & VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL) &&
 	    WARN_ON_ONCE(kvm_set_msr(vcpu, MSR_CORE_PERF_GLOBAL_CTRL,
 				     vmcs12->guest_ia32_perf_global_ctrl))) {
@@ -3695,12 +3691,34 @@ vmcs12_guest_cr4(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12)
 }
 
 static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
-				      struct vmcs12 *vmcs12)
+				      struct vmcs12 *vmcs12,
+				      u32 vm_exit_reason, u32 exit_intr_info)
 {
 	u32 idt_vectoring;
 	unsigned int nr;
 
-	if (vcpu->arch.exception.injected) {
+	/*
+	 * Per the SDM, VM-Exits due to double and triple faults are never
+	 * considered to occur during event delivery, even if the double/triple
+	 * fault is the result of an escalating vectoring issue.
+	 *
+	 * Note, the SDM qualifies the double fault behavior with "The original
+	 * event results in a double-fault exception".  It's unclear why the
+	 * qualification exists since exits due to double fault can occur only
+	 * while vectoring a different exception (injected events are never
+	 * subject to interception), i.e. there's _always_ an original event.
+	 *
+	 * The SDM also uses NMI as a confusing example for the "original event
+	 * causes the VM exit directly" clause.  NMI isn't special in any way,
+	 * the same rule applies to all events that cause an exit directly.
+	 * NMI is an odd choice for the example because NMIs can only occur on
+	 * instruction boundaries, i.e. they _can't_ occur during vectoring.
+	 */
+	if ((u16)vm_exit_reason == EXIT_REASON_TRIPLE_FAULT ||
+	    ((u16)vm_exit_reason == EXIT_REASON_EXCEPTION_NMI &&
+	     is_double_fault(exit_intr_info))) {
+		vmcs12->idt_vectoring_info_field = 0;
+	} else if (vcpu->arch.exception.injected) {
 		nr = vcpu->arch.exception.nr;
 		idt_vectoring = nr | VECTORING_INFO_VALID_MASK;
 
@@ -3733,6 +3751,8 @@ static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
 			idt_vectoring |= INTR_TYPE_EXT_INTR;
 
 		vmcs12->idt_vectoring_info_field = idt_vectoring;
+	} else {
+		vmcs12->idt_vectoring_info_field = 0;
 	}
 }
 
@@ -4202,12 +4222,12 @@ static void prepare_vmcs12(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12,
 	if (to_vmx(vcpu)->exit_reason.enclave_mode)
 		vmcs12->vm_exit_reason |= VMX_EXIT_REASONS_SGX_ENCLAVE_MODE;
 	vmcs12->exit_qualification = exit_qualification;
-	vmcs12->vm_exit_intr_info = exit_intr_info;
-
-	vmcs12->idt_vectoring_info_field = 0;
-	vmcs12->vm_exit_instruction_len = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
-	vmcs12->vmx_instruction_info = vmcs_read32(VMX_INSTRUCTION_INFO);
 
+	/*
+	 * On VM-Exit due to a failed VM-Entry, the VMCS isn't marked launched
+	 * and only EXIT_REASON and EXIT_QUALIFICATION are updated, all other
+	 * exit info fields are unmodified.
+	 */
 	if (!(vmcs12->vm_exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY)) {
 		vmcs12->launch_state = 1;
 
@@ -4219,7 +4239,12 @@ static void prepare_vmcs12(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12,
 		 * Transfer the event that L0 or L1 may wanted to inject into
 		 * L2 to IDT_VECTORING_INFO_FIELD.
 		 */
-		vmcs12_save_pending_event(vcpu, vmcs12);
+		vmcs12_save_pending_event(vcpu, vmcs12,
+					  vm_exit_reason, exit_intr_info);
+
+		vmcs12->vm_exit_intr_info = exit_intr_info;
+		vmcs12->vm_exit_instruction_len = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
+		vmcs12->vmx_instruction_info = vmcs_read32(VMX_INSTRUCTION_INFO);
 
 		/*
 		 * According to spec, there's no need to store the guest's
@@ -4518,9 +4543,6 @@ void nested_vmx_vmexit(struct kvm_vcpu *vcpu, u32 vm_exit_reason,
 	/* trying to cancel vmlaunch/vmresume is a bug */
 	WARN_ON_ONCE(vmx->nested.nested_run_pending);
 
-	/* Similarly, triple faults in L2 should never escape. */
-	WARN_ON_ONCE(kvm_check_request(KVM_REQ_TRIPLE_FAULT, vcpu));
-
 	if (kvm_check_request(KVM_REQ_GET_NESTED_STATE_PAGES, vcpu)) {
 		/*
 		 * KVM_REQ_GET_NESTED_STATE_PAGES is also used to map
@@ -6809,6 +6831,7 @@ __init int nested_vmx_hardware_setup(int (*exit_handlers[])(struct kvm_vcpu *))
 struct kvm_x86_nested_ops vmx_nested_ops = {
 	.leave_nested = vmx_leave_nested,
 	.check_events = vmx_check_nested_events,
+	.handle_page_fault_workaround = nested_vmx_handle_page_fault_workaround,
 	.hv_timer_pending = nested_vmx_preemption_timer_pending,
 	.triple_fault = nested_vmx_triple_fault,
 	.get_state = vmx_get_nested_state,
diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c
index b82b6709d7a8..37e9eb32e3d9 100644
--- a/arch/x86/kvm/vmx/pmu_intel.c
+++ b/arch/x86/kvm/vmx/pmu_intel.c
@@ -719,7 +719,7 @@ static void intel_pmu_cleanup(struct kvm_vcpu *vcpu)
 		intel_pmu_release_guest_lbr_event(vcpu);
 }
 
-struct kvm_pmu_ops intel_pmu_ops = {
+struct kvm_pmu_ops intel_pmu_ops __initdata = {
 	.pmc_perf_hw_id = intel_pmc_perf_hw_id,
 	.pmc_is_enabled = intel_pmc_is_enabled,
 	.pmc_idx_to_pmc = intel_pmc_idx_to_pmc,
diff --git a/arch/x86/kvm/vmx/posted_intr.c b/arch/x86/kvm/vmx/posted_intr.c
index 3834bb30ce54..07e5fcf5a5aa 100644
--- a/arch/x86/kvm/vmx/posted_intr.c
+++ b/arch/x86/kvm/vmx/posted_intr.c
@@ -202,16 +202,17 @@ void vmx_vcpu_pi_put(struct kvm_vcpu *vcpu)
 void pi_wakeup_handler(void)
 {
 	int cpu = smp_processor_id();
+	struct list_head *wakeup_list = &per_cpu(wakeup_vcpus_on_cpu, cpu);
+	raw_spinlock_t *spinlock = &per_cpu(wakeup_vcpus_on_cpu_lock, cpu);
 	struct vcpu_vmx *vmx;
 
-	raw_spin_lock(&per_cpu(wakeup_vcpus_on_cpu_lock, cpu));
-	list_for_each_entry(vmx, &per_cpu(wakeup_vcpus_on_cpu, cpu),
-			    pi_wakeup_list) {
+	raw_spin_lock(spinlock);
+	list_for_each_entry(vmx, wakeup_list, pi_wakeup_list) {
 
 		if (pi_test_on(&vmx->pi_desc))
 			kvm_vcpu_wake_up(&vmx->vcpu);
 	}
-	raw_spin_unlock(&per_cpu(wakeup_vcpus_on_cpu_lock, cpu));
+	raw_spin_unlock(spinlock);
 }
 
 void __init pi_init_cpu(int cpu)
@@ -311,7 +312,7 @@ int vmx_pi_update_irte(struct kvm *kvm, unsigned int host_irq,
 			continue;
 		}
 
-		vcpu_info.pi_desc_addr = __pa(&to_vmx(vcpu)->pi_desc);
+		vcpu_info.pi_desc_addr = __pa(vcpu_to_pi_desc(vcpu));
 		vcpu_info.vector = irq.vector;
 
 		trace_kvm_pi_irte_update(host_irq, vcpu->vcpu_id, e->gsi,
diff --git a/arch/x86/kvm/vmx/vmcs.h b/arch/x86/kvm/vmx/vmcs.h
index e325c290a816..2b9d7a7e83f7 100644
--- a/arch/x86/kvm/vmx/vmcs.h
+++ b/arch/x86/kvm/vmx/vmcs.h
@@ -104,6 +104,11 @@ static inline bool is_breakpoint(u32 intr_info)
 	return is_exception_n(intr_info, BP_VECTOR);
 }
 
+static inline bool is_double_fault(u32 intr_info)
+{
+	return is_exception_n(intr_info, DF_VECTOR);
+}
+
 static inline bool is_page_fault(u32 intr_info)
 {
 	return is_exception_n(intr_info, PF_VECTOR);
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index 610355b9ccce..f5aeade623d6 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -2444,7 +2444,7 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf,
 				&_cpu_based_exec_control) < 0)
 		return -EIO;
 #ifdef CONFIG_X86_64
-	if ((_cpu_based_exec_control & CPU_BASED_TPR_SHADOW))
+	if (_cpu_based_exec_control & CPU_BASED_TPR_SHADOW)
 		_cpu_based_exec_control &= ~CPU_BASED_CR8_LOAD_EXITING &
 					   ~CPU_BASED_CR8_STORE_EXITING;
 #endif
@@ -2948,7 +2948,7 @@ static void vmx_flush_tlb_current(struct kvm_vcpu *vcpu)
 
 	if (enable_ept)
 		ept_sync_context(construct_eptp(vcpu, root_hpa,
-						mmu->shadow_root_level));
+						mmu->root_role.level));
 	else
 		vpid_sync_context(vmx_get_current_vpid(vcpu));
 }
@@ -4385,7 +4385,7 @@ static void init_vmcs(struct vcpu_vmx *vmx)
 	if (cpu_has_secondary_exec_ctrls())
 		secondary_exec_controls_set(vmx, vmx_secondary_exec_control(vmx));
 
-	if (kvm_vcpu_apicv_active(&vmx->vcpu)) {
+	if (enable_apicv && lapic_in_kernel(&vmx->vcpu)) {
 		vmcs_write64(EOI_EXIT_BITMAP0, 0);
 		vmcs_write64(EOI_EXIT_BITMAP1, 0);
 		vmcs_write64(EOI_EXIT_BITMAP2, 0);
@@ -5410,9 +5410,7 @@ static int handle_ept_violation(struct kvm_vcpu *vcpu)
 	error_code |= (exit_qualification & EPT_VIOLATION_ACC_INSTR)
 		      ? PFERR_FETCH_MASK : 0;
 	/* ept page table entry is present? */
-	error_code |= (exit_qualification &
-		       (EPT_VIOLATION_READABLE | EPT_VIOLATION_WRITABLE |
-			EPT_VIOLATION_EXECUTABLE))
+	error_code |= (exit_qualification & EPT_VIOLATION_RWX_MASK)
 		      ? PFERR_PRESENT_MASK : 0;
 
 	error_code |= (exit_qualification & EPT_VIOLATION_GVA_TRANSLATED) != 0 ?
@@ -7823,7 +7821,6 @@ static struct kvm_x86_ops vmx_x86_ops __initdata = {
 	.cpu_dirty_log_size = PML_ENTITY_NUM,
 	.update_cpu_dirty_logging = vmx_update_cpu_dirty_logging,
 
-	.pmu_ops = &intel_pmu_ops,
 	.nested_ops = &vmx_nested_ops,
 
 	.pi_update_irte = vmx_pi_update_irte,
@@ -7856,7 +7853,7 @@ static unsigned int vmx_handle_intel_pt_intr(void)
 	struct kvm_vcpu *vcpu = kvm_get_running_vcpu();
 
 	/* '0' on failure so that the !PT case can use a RET0 static call. */
-	if (!kvm_arch_pmi_in_guest(vcpu))
+	if (!vcpu || !kvm_handling_nmi_from_guest(vcpu))
 		return 0;
 
 	kvm_make_request(KVM_REQ_PMI, vcpu);
@@ -7891,6 +7888,31 @@ static __init void vmx_setup_user_return_msrs(void)
 		kvm_add_user_return_msr(vmx_uret_msrs_list[i]);
 }
 
+static void __init vmx_setup_me_spte_mask(void)
+{
+	u64 me_mask = 0;
+
+	/*
+	 * kvm_get_shadow_phys_bits() returns shadow_phys_bits.  Use
+	 * the former to avoid exposing shadow_phys_bits.
+	 *
+	 * On pre-MKTME system, boot_cpu_data.x86_phys_bits equals to
+	 * shadow_phys_bits.  On MKTME and/or TDX capable systems,
+	 * boot_cpu_data.x86_phys_bits holds the actual physical address
+	 * w/o the KeyID bits, and shadow_phys_bits equals to MAXPHYADDR
+	 * reported by CPUID.  Those bits between are KeyID bits.
+	 */
+	if (boot_cpu_data.x86_phys_bits != kvm_get_shadow_phys_bits())
+		me_mask = rsvd_bits(boot_cpu_data.x86_phys_bits,
+			kvm_get_shadow_phys_bits() - 1);
+	/*
+	 * Unlike SME, host kernel doesn't support setting up any
+	 * MKTME KeyID on Intel platforms.  No memory encryption
+	 * bits should be included into the SPTE.
+	 */
+	kvm_mmu_set_me_spte_mask(0, me_mask);
+}
+
 static struct kvm_x86_init_ops vmx_init_ops __initdata;
 
 static __init int hardware_setup(void)
@@ -7993,6 +8015,12 @@ static __init int hardware_setup(void)
 		kvm_mmu_set_ept_masks(enable_ept_ad_bits,
 				      cpu_has_vmx_ept_execute_only());
 
+	/*
+	 * Setup shadow_me_value/shadow_me_mask to include MKTME KeyID
+	 * bits to shadow_zero_check.
+	 */
+	vmx_setup_me_spte_mask();
+
 	kvm_configure_mmu(enable_ept, 0, vmx_get_max_tdp_level(),
 			  ept_caps_to_lpage_level(vmx_capability.ept));
 
@@ -8077,6 +8105,7 @@ static struct kvm_x86_init_ops vmx_init_ops __initdata = {
 	.handle_intel_pt_intr = NULL,
 
 	.runtime_ops = &vmx_x86_ops,
+	.pmu_ops = &intel_pmu_ops,
 };
 
 static void vmx_cleanup_l1d_flush(void)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 4790f0d7d40b..b81ef4f497f4 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -266,7 +266,12 @@ const struct kvm_stats_header kvm_vm_stats_header = {
 
 const struct _kvm_stats_desc kvm_vcpu_stats_desc[] = {
 	KVM_GENERIC_VCPU_STATS(),
+	STATS_DESC_COUNTER(VCPU, pf_taken),
 	STATS_DESC_COUNTER(VCPU, pf_fixed),
+	STATS_DESC_COUNTER(VCPU, pf_emulate),
+	STATS_DESC_COUNTER(VCPU, pf_spurious),
+	STATS_DESC_COUNTER(VCPU, pf_fast),
+	STATS_DESC_COUNTER(VCPU, pf_mmio_spte_created),
 	STATS_DESC_COUNTER(VCPU, pf_guest),
 	STATS_DESC_COUNTER(VCPU, tlb_flush),
 	STATS_DESC_COUNTER(VCPU, invlpg),
@@ -748,6 +753,7 @@ void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault)
 }
 EXPORT_SYMBOL_GPL(kvm_inject_page_fault);
 
+/* Returns true if the page fault was immediately morphed into a VM-Exit. */
 bool kvm_inject_emulated_page_fault(struct kvm_vcpu *vcpu,
 				    struct x86_exception *fault)
 {
@@ -766,8 +772,26 @@ bool kvm_inject_emulated_page_fault(struct kvm_vcpu *vcpu,
 		kvm_mmu_invalidate_gva(vcpu, fault_mmu, fault->address,
 				       fault_mmu->root.hpa);
 
+	/*
+	 * A workaround for KVM's bad exception handling.  If KVM injected an
+	 * exception into L2, and L2 encountered a #PF while vectoring the
+	 * injected exception, manually check to see if L1 wants to intercept
+	 * #PF, otherwise queuing the #PF will lead to #DF or a lost exception.
+	 * In all other cases, defer the check to nested_ops->check_events(),
+	 * which will correctly handle priority (this does not).  Note, other
+	 * exceptions, e.g. #GP, are theoretically affected, #PF is simply the
+	 * most problematic, e.g. when L0 and L1 are both intercepting #PF for
+	 * shadow paging.
+	 *
+	 * TODO: Rewrite exception handling to track injected and pending
+	 *       (VM-Exit) exceptions separately.
+	 */
+	if (unlikely(vcpu->arch.exception.injected && is_guest_mode(vcpu)) &&
+	    kvm_x86_ops.nested_ops->handle_page_fault_workaround(vcpu, fault))
+		return true;
+
 	fault_mmu->inject_page_fault(vcpu, fault);
-	return fault->nested_page_fault;
+	return false;
 }
 EXPORT_SYMBOL_GPL(kvm_inject_emulated_page_fault);
 
@@ -961,11 +985,13 @@ void kvm_load_guest_xsave_state(struct kvm_vcpu *vcpu)
 			wrmsrl(MSR_IA32_XSS, vcpu->arch.ia32_xss);
 	}
 
+#ifdef CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS
 	if (static_cpu_has(X86_FEATURE_PKU) &&
-	    (kvm_read_cr4_bits(vcpu, X86_CR4_PKE) ||
-	     (vcpu->arch.xcr0 & XFEATURE_MASK_PKRU)) &&
-	    vcpu->arch.pkru != vcpu->arch.host_pkru)
+	    vcpu->arch.pkru != vcpu->arch.host_pkru &&
+	    ((vcpu->arch.xcr0 & XFEATURE_MASK_PKRU) ||
+	     kvm_read_cr4_bits(vcpu, X86_CR4_PKE)))
 		write_pkru(vcpu->arch.pkru);
+#endif /* CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS */
 }
 EXPORT_SYMBOL_GPL(kvm_load_guest_xsave_state);
 
@@ -974,13 +1000,15 @@ void kvm_load_host_xsave_state(struct kvm_vcpu *vcpu)
 	if (vcpu->arch.guest_state_protected)
 		return;
 
+#ifdef CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS
 	if (static_cpu_has(X86_FEATURE_PKU) &&
-	    (kvm_read_cr4_bits(vcpu, X86_CR4_PKE) ||
-	     (vcpu->arch.xcr0 & XFEATURE_MASK_PKRU))) {
+	    ((vcpu->arch.xcr0 & XFEATURE_MASK_PKRU) ||
+	     kvm_read_cr4_bits(vcpu, X86_CR4_PKE))) {
 		vcpu->arch.pkru = rdpkru();
 		if (vcpu->arch.pkru != vcpu->arch.host_pkru)
 			write_pkru(vcpu->arch.host_pkru);
 	}
+#endif /* CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS */
 
 	if (kvm_read_cr4_bits(vcpu, X86_CR4_OSXSAVE)) {
 
@@ -2249,14 +2277,13 @@ static void kvm_write_system_time(struct kvm_vcpu *vcpu, gpa_t system_time,
 	kvm_make_request(KVM_REQ_GLOBAL_CLOCK_UPDATE, vcpu);
 
 	/* we verify if the enable bit is set... */
-	vcpu->arch.pv_time_enabled = false;
-	if (!(system_time & 1))
-		return;
-
-	if (!kvm_gfn_to_hva_cache_init(vcpu->kvm,
-				       &vcpu->arch.pv_time, system_time & ~1ULL,
-				       sizeof(struct pvclock_vcpu_time_info)))
-		vcpu->arch.pv_time_enabled = true;
+	if (system_time & 1) {
+		kvm_gfn_to_pfn_cache_init(vcpu->kvm, &vcpu->arch.pv_time, vcpu,
+					  KVM_HOST_USES_PFN, system_time & ~1ULL,
+					  sizeof(struct pvclock_vcpu_time_info));
+	} else {
+		kvm_gfn_to_pfn_cache_destroy(vcpu->kvm, &vcpu->arch.pv_time);
+	}
 
 	return;
 }
@@ -2961,63 +2988,55 @@ u64 get_kvmclock_ns(struct kvm *kvm)
 	return data.clock;
 }
 
-static void kvm_setup_pvclock_page(struct kvm_vcpu *v,
-				   struct gfn_to_hva_cache *cache,
-				   unsigned int offset)
+static void kvm_setup_guest_pvclock(struct kvm_vcpu *v,
+				    struct gfn_to_pfn_cache *gpc,
+				    unsigned int offset)
 {
 	struct kvm_vcpu_arch *vcpu = &v->arch;
-	struct pvclock_vcpu_time_info guest_hv_clock;
+	struct pvclock_vcpu_time_info *guest_hv_clock;
+	unsigned long flags;
 
-	if (unlikely(kvm_read_guest_offset_cached(v->kvm, cache,
-		&guest_hv_clock, offset, sizeof(guest_hv_clock))))
-		return;
+	read_lock_irqsave(&gpc->lock, flags);
+	while (!kvm_gfn_to_pfn_cache_check(v->kvm, gpc, gpc->gpa,
+					   offset + sizeof(*guest_hv_clock))) {
+		read_unlock_irqrestore(&gpc->lock, flags);
 
-	/* This VCPU is paused, but it's legal for a guest to read another
+		if (kvm_gfn_to_pfn_cache_refresh(v->kvm, gpc, gpc->gpa,
+						 offset + sizeof(*guest_hv_clock)))
+			return;
+
+		read_lock_irqsave(&gpc->lock, flags);
+	}
+
+	guest_hv_clock = (void *)(gpc->khva + offset);
+
+	/*
+	 * This VCPU is paused, but it's legal for a guest to read another
 	 * VCPU's kvmclock, so we really have to follow the specification where
 	 * it says that version is odd if data is being modified, and even after
 	 * it is consistent.
-	 *
-	 * Version field updates must be kept separate.  This is because
-	 * kvm_write_guest_cached might use a "rep movs" instruction, and
-	 * writes within a string instruction are weakly ordered.  So there
-	 * are three writes overall.
-	 *
-	 * As a small optimization, only write the version field in the first
-	 * and third write.  The vcpu->pv_time cache is still valid, because the
-	 * version field is the first in the struct.
 	 */
-	BUILD_BUG_ON(offsetof(struct pvclock_vcpu_time_info, version) != 0);
-
-	if (guest_hv_clock.version & 1)
-		++guest_hv_clock.version;  /* first time write, random junk */
-
-	vcpu->hv_clock.version = guest_hv_clock.version + 1;
-	kvm_write_guest_offset_cached(v->kvm, cache,
-				      &vcpu->hv_clock, offset,
-				      sizeof(vcpu->hv_clock.version));
 
+	guest_hv_clock->version = vcpu->hv_clock.version = (guest_hv_clock->version + 1) | 1;
 	smp_wmb();
 
 	/* retain PVCLOCK_GUEST_STOPPED if set in guest copy */
-	vcpu->hv_clock.flags |= (guest_hv_clock.flags & PVCLOCK_GUEST_STOPPED);
+	vcpu->hv_clock.flags |= (guest_hv_clock->flags & PVCLOCK_GUEST_STOPPED);
 
 	if (vcpu->pvclock_set_guest_stopped_request) {
 		vcpu->hv_clock.flags |= PVCLOCK_GUEST_STOPPED;
 		vcpu->pvclock_set_guest_stopped_request = false;
 	}
 
-	trace_kvm_pvclock_update(v->vcpu_id, &vcpu->hv_clock);
+	memcpy(guest_hv_clock, &vcpu->hv_clock, sizeof(*guest_hv_clock));
+	smp_wmb();
 
-	kvm_write_guest_offset_cached(v->kvm, cache,
-				      &vcpu->hv_clock, offset,
-				      sizeof(vcpu->hv_clock));
+	guest_hv_clock->version = ++vcpu->hv_clock.version;
 
-	smp_wmb();
+	mark_page_dirty_in_slot(v->kvm, gpc->memslot, gpc->gpa >> PAGE_SHIFT);
+	read_unlock_irqrestore(&gpc->lock, flags);
 
-	vcpu->hv_clock.version++;
-	kvm_write_guest_offset_cached(v->kvm, cache,
-				     &vcpu->hv_clock, offset,
-				     sizeof(vcpu->hv_clock.version));
+	trace_kvm_pvclock_update(v->vcpu_id, &vcpu->hv_clock);
 }
 
 static int kvm_guest_time_update(struct kvm_vcpu *v)
@@ -3106,13 +3125,13 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 
 	vcpu->hv_clock.flags = pvclock_flags;
 
-	if (vcpu->pv_time_enabled)
-		kvm_setup_pvclock_page(v, &vcpu->pv_time, 0);
-	if (vcpu->xen.vcpu_info_set)
-		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_info_cache,
-				       offsetof(struct compat_vcpu_info, time));
-	if (vcpu->xen.vcpu_time_info_set)
-		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
+	if (vcpu->pv_time.active)
+		kvm_setup_guest_pvclock(v, &vcpu->pv_time, 0);
+	if (vcpu->xen.vcpu_info_cache.active)
+		kvm_setup_guest_pvclock(v, &vcpu->xen.vcpu_info_cache,
+					offsetof(struct compat_vcpu_info, time));
+	if (vcpu->xen.vcpu_time_info_cache.active)
+		kvm_setup_guest_pvclock(v, &vcpu->xen.vcpu_time_info_cache, 0);
 	kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
 	return 0;
 }
@@ -3300,7 +3319,7 @@ static int kvm_pv_enable_async_pf_int(struct kvm_vcpu *vcpu, u64 data)
 
 static void kvmclock_reset(struct kvm_vcpu *vcpu)
 {
-	vcpu->arch.pv_time_enabled = false;
+	kvm_gfn_to_pfn_cache_destroy(vcpu->kvm, &vcpu->arch.pv_time);
 	vcpu->arch.time = 0;
 }
 
@@ -4284,7 +4303,8 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 		r = KVM_XEN_HVM_CONFIG_HYPERCALL_MSR |
 		    KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL |
 		    KVM_XEN_HVM_CONFIG_SHARED_INFO |
-		    KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL;
+		    KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL |
+		    KVM_XEN_HVM_CONFIG_EVTCHN_SEND;
 		if (sched_info_on())
 			r |= KVM_XEN_HVM_CONFIG_RUNSTATE;
 		break;
@@ -4331,6 +4351,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 		r = boot_cpu_has(X86_FEATURE_XSAVE);
 		break;
 	case KVM_CAP_TSC_CONTROL:
+	case KVM_CAP_VM_TSC_CONTROL:
 		r = kvm_has_tsc_control;
 		break;
 	case KVM_CAP_X2APIC_API:
@@ -5102,7 +5123,7 @@ static int kvm_vcpu_ioctl_x86_set_xcrs(struct kvm_vcpu *vcpu,
  */
 static int kvm_set_guest_paused(struct kvm_vcpu *vcpu)
 {
-	if (!vcpu->arch.pv_time_enabled)
+	if (!vcpu->arch.pv_time.active)
 		return -EINVAL;
 	vcpu->arch.pvclock_set_guest_stopped_request = true;
 	kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
@@ -6186,7 +6207,7 @@ static int kvm_arch_suspend_notifier(struct kvm *kvm)
 
 	mutex_lock(&kvm->lock);
 	kvm_for_each_vcpu(i, vcpu, kvm) {
-		if (!vcpu->arch.pv_time_enabled)
+		if (!vcpu->arch.pv_time.active)
 			continue;
 
 		ret = kvm_set_guest_paused(vcpu);
@@ -6513,6 +6534,15 @@ set_pit2_out:
 		r = kvm_xen_hvm_set_attr(kvm, &xha);
 		break;
 	}
+	case KVM_XEN_HVM_EVTCHN_SEND: {
+		struct kvm_irq_routing_xen_evtchn uxe;
+
+		r = -EFAULT;
+		if (copy_from_user(&uxe, argp, sizeof(uxe)))
+			goto out;
+		r = kvm_xen_hvm_evtchn_send(kvm, &uxe);
+		break;
+	}
 #endif
 	case KVM_SET_CLOCK:
 		r = kvm_vm_ioctl_set_clock(kvm, argp);
@@ -6520,6 +6550,28 @@ set_pit2_out:
 	case KVM_GET_CLOCK:
 		r = kvm_vm_ioctl_get_clock(kvm, argp);
 		break;
+	case KVM_SET_TSC_KHZ: {
+		u32 user_tsc_khz;
+
+		r = -EINVAL;
+		user_tsc_khz = (u32)arg;
+
+		if (kvm_has_tsc_control &&
+		    user_tsc_khz >= kvm_max_guest_tsc_khz)
+			goto out;
+
+		if (user_tsc_khz == 0)
+			user_tsc_khz = tsc_khz;
+
+		WRITE_ONCE(kvm->arch.default_tsc_khz, user_tsc_khz);
+		r = 0;
+
+		goto out;
+	}
+	case KVM_GET_TSC_KHZ: {
+		r = READ_ONCE(kvm->arch.default_tsc_khz);
+		goto out;
+	}
 	case KVM_MEMORY_ENCRYPT_OP: {
 		r = -ENOTTY;
 		if (!kvm_x86_ops.mem_enc_ioctl)
@@ -7229,15 +7281,8 @@ static int emulator_write_emulated(struct x86_emulate_ctxt *ctxt,
 				   exception, &write_emultor);
 }
 
-#define CMPXCHG_TYPE(t, ptr, old, new) \
-	(cmpxchg((t *)(ptr), *(t *)(old), *(t *)(new)) == *(t *)(old))
-
-#ifdef CONFIG_X86_64
-#  define CMPXCHG64(ptr, old, new) CMPXCHG_TYPE(u64, ptr, old, new)
-#else
-#  define CMPXCHG64(ptr, old, new) \
-	(cmpxchg64((u64 *)(ptr), *(u64 *)(old), *(u64 *)(new)) == *(u64 *)(old))
-#endif
+#define emulator_try_cmpxchg_user(t, ptr, old, new) \
+	(__try_cmpxchg_user((t __user *)(ptr), (t *)(old), *(t *)(new), efault ## t))
 
 static int emulator_cmpxchg_emulated(struct x86_emulate_ctxt *ctxt,
 				     unsigned long addr,
@@ -7246,12 +7291,11 @@ static int emulator_cmpxchg_emulated(struct x86_emulate_ctxt *ctxt,
 				     unsigned int bytes,
 				     struct x86_exception *exception)
 {
-	struct kvm_host_map map;
 	struct kvm_vcpu *vcpu = emul_to_vcpu(ctxt);
 	u64 page_line_mask;
+	unsigned long hva;
 	gpa_t gpa;
-	char *kaddr;
-	bool exchanged;
+	int r;
 
 	/* guests cmpxchg8b have to be emulated atomically */
 	if (bytes > 8 || (bytes & (bytes - 1)))
@@ -7275,31 +7319,32 @@ static int emulator_cmpxchg_emulated(struct x86_emulate_ctxt *ctxt,
 	if (((gpa + bytes - 1) & page_line_mask) != (gpa & page_line_mask))
 		goto emul_write;
 
-	if (kvm_vcpu_map(vcpu, gpa_to_gfn(gpa), &map))
+	hva = kvm_vcpu_gfn_to_hva(vcpu, gpa_to_gfn(gpa));
+	if (kvm_is_error_hva(hva))
 		goto emul_write;
 
-	kaddr = map.hva + offset_in_page(gpa);
+	hva += offset_in_page(gpa);
 
 	switch (bytes) {
 	case 1:
-		exchanged = CMPXCHG_TYPE(u8, kaddr, old, new);
+		r = emulator_try_cmpxchg_user(u8, hva, old, new);
 		break;
 	case 2:
-		exchanged = CMPXCHG_TYPE(u16, kaddr, old, new);
+		r = emulator_try_cmpxchg_user(u16, hva, old, new);
 		break;
 	case 4:
-		exchanged = CMPXCHG_TYPE(u32, kaddr, old, new);
+		r = emulator_try_cmpxchg_user(u32, hva, old, new);
 		break;
 	case 8:
-		exchanged = CMPXCHG64(kaddr, old, new);
+		r = emulator_try_cmpxchg_user(u64, hva, old, new);
 		break;
 	default:
 		BUG();
 	}
 
-	kvm_vcpu_unmap(vcpu, &map, true);
-
-	if (!exchanged)
+	if (r < 0)
+		return X86EMUL_UNHANDLEABLE;
+	if (r)
 		return X86EMUL_CMPXCHG_FAILED;
 
 	kvm_page_track_write(vcpu, gpa, new, bytes);
@@ -8061,7 +8106,7 @@ static bool reexecute_instruction(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
 	    WARN_ON_ONCE(!(emulation_type & EMULTYPE_PF)))
 		return false;
 
-	if (!vcpu->arch.mmu->direct_map) {
+	if (!vcpu->arch.mmu->root_role.direct) {
 		/*
 		 * Write permission should be allowed since only
 		 * write access need to be emulated.
@@ -8094,7 +8139,7 @@ static bool reexecute_instruction(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
 	kvm_release_pfn_clean(pfn);
 
 	/* The instructions are well-emulated on direct mmu. */
-	if (vcpu->arch.mmu->direct_map) {
+	if (vcpu->arch.mmu->root_role.direct) {
 		unsigned int indirect_shadow_pages;
 
 		write_lock(&vcpu->kvm->mmu_lock);
@@ -8162,7 +8207,7 @@ static bool retry_instruction(struct x86_emulate_ctxt *ctxt,
 	vcpu->arch.last_retry_eip = ctxt->eip;
 	vcpu->arch.last_retry_addr = cr2_or_gpa;
 
-	if (!vcpu->arch.mmu->direct_map)
+	if (!vcpu->arch.mmu->root_role.direct)
 		gpa = kvm_mmu_gva_to_gpa_write(vcpu, cr2_or_gpa, NULL);
 
 	kvm_mmu_unprotect_page(vcpu->kvm, gpa_to_gfn(gpa));
@@ -8251,7 +8296,7 @@ int kvm_skip_emulated_instruction(struct kvm_vcpu *vcpu)
 }
 EXPORT_SYMBOL_GPL(kvm_skip_emulated_instruction);
 
-static bool kvm_vcpu_check_breakpoint(struct kvm_vcpu *vcpu, int *r)
+static bool kvm_vcpu_check_code_breakpoint(struct kvm_vcpu *vcpu, int *r)
 {
 	if (unlikely(vcpu->guest_debug & KVM_GUESTDBG_USE_HW_BP) &&
 	    (vcpu->arch.guest_debug_dr7 & DR7_BP_EN_MASK)) {
@@ -8320,25 +8365,23 @@ static bool is_vmware_backdoor_opcode(struct x86_emulate_ctxt *ctxt)
 }
 
 /*
- * Decode to be emulated instruction. Return EMULATION_OK if success.
+ * Decode an instruction for emulation.  The caller is responsible for handling
+ * code breakpoints.  Note, manually detecting code breakpoints is unnecessary
+ * (and wrong) when emulating on an intercepted fault-like exception[*], as
+ * code breakpoints have higher priority and thus have already been done by
+ * hardware.
+ *
+ * [*] Except #MC, which is higher priority, but KVM should never emulate in
+ *     response to a machine check.
  */
 int x86_decode_emulated_instruction(struct kvm_vcpu *vcpu, int emulation_type,
 				    void *insn, int insn_len)
 {
-	int r = EMULATION_OK;
 	struct x86_emulate_ctxt *ctxt = vcpu->arch.emulate_ctxt;
+	int r;
 
 	init_emulate_ctxt(vcpu);
 
-	/*
-	 * We will reenter on the same instruction since we do not set
-	 * complete_userspace_io. This does not handle watchpoints yet,
-	 * those would be handled in the emulate_ops.
-	 */
-	if (!(emulation_type & EMULTYPE_SKIP) &&
-	    kvm_vcpu_check_breakpoint(vcpu, &r))
-		return r;
-
 	r = x86_decode_insn(ctxt, insn, insn_len, emulation_type);
 
 	trace_kvm_emulate_insn_start(vcpu);
@@ -8371,6 +8414,15 @@ int x86_emulate_instruction(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
 	if (!(emulation_type & EMULTYPE_NO_DECODE)) {
 		kvm_clear_exception_queue(vcpu);
 
+		/*
+		 * Return immediately if RIP hits a code breakpoint, such #DBs
+		 * are fault-like and are higher priority than any faults on
+		 * the code fetch itself.
+		 */
+		if (!(emulation_type & EMULTYPE_SKIP) &&
+		    kvm_vcpu_check_code_breakpoint(vcpu, &r))
+			return r;
+
 		r = x86_decode_emulated_instruction(vcpu, emulation_type,
 						    insn, insn_len);
 		if (r != EMULATION_OK)  {
@@ -8442,7 +8494,7 @@ restart:
 		ctxt->exception.address = cr2_or_gpa;
 
 		/* With shadow page tables, cr2 contains a GVA or nGPA. */
-		if (vcpu->arch.mmu->direct_map) {
+		if (vcpu->arch.mmu->root_role.direct) {
 			ctxt->gpa_available = true;
 			ctxt->gpa_val = cr2_or_gpa;
 		}
@@ -8789,22 +8841,22 @@ static int kvmclock_cpu_online(unsigned int cpu)
 
 static void kvm_timer_init(void)
 {
-	max_tsc_khz = tsc_khz;
-
 	if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC)) {
-#ifdef CONFIG_CPU_FREQ
-		struct cpufreq_policy *policy;
-		int cpu;
-
-		cpu = get_cpu();
-		policy = cpufreq_cpu_get(cpu);
-		if (policy) {
-			if (policy->cpuinfo.max_freq)
-				max_tsc_khz = policy->cpuinfo.max_freq;
-			cpufreq_cpu_put(policy);
+		max_tsc_khz = tsc_khz;
+
+		if (IS_ENABLED(CONFIG_CPU_FREQ)) {
+			struct cpufreq_policy *policy;
+			int cpu;
+
+			cpu = get_cpu();
+			policy = cpufreq_cpu_get(cpu);
+			if (policy) {
+				if (policy->cpuinfo.max_freq)
+					max_tsc_khz = policy->cpuinfo.max_freq;
+				cpufreq_cpu_put(policy);
+			}
+			put_cpu();
 		}
-		put_cpu();
-#endif
 		cpufreq_register_notifier(&kvmclock_cpufreq_notifier_block,
 					  CPUFREQ_TRANSITION_NOTIFIER);
 	}
@@ -9089,6 +9141,14 @@ bool kvm_apicv_activated(struct kvm *kvm)
 }
 EXPORT_SYMBOL_GPL(kvm_apicv_activated);
 
+bool kvm_vcpu_apicv_activated(struct kvm_vcpu *vcpu)
+{
+	ulong vm_reasons = READ_ONCE(vcpu->kvm->arch.apicv_inhibit_reasons);
+	ulong vcpu_reasons = static_call(kvm_x86_vcpu_get_apicv_inhibit_reasons)(vcpu);
+
+	return (vm_reasons | vcpu_reasons) == 0;
+}
+EXPORT_SYMBOL_GPL(kvm_vcpu_apicv_activated);
 
 static void set_or_clear_apicv_inhibit(unsigned long *inhibits,
 				       enum kvm_apicv_inhibit reason, bool set)
@@ -9266,6 +9326,17 @@ static int emulator_fix_hypercall(struct x86_emulate_ctxt *ctxt)
 	char instruction[3];
 	unsigned long rip = kvm_rip_read(vcpu);
 
+	/*
+	 * If the quirk is disabled, synthesize a #UD and let the guest pick up
+	 * the pieces.
+	 */
+	if (!kvm_check_has_quirk(vcpu->kvm, KVM_X86_QUIRK_FIX_HYPERCALL_INSN)) {
+		ctxt->exception.error_code_valid = false;
+		ctxt->exception.vector = UD_VECTOR;
+		ctxt->have_exception = true;
+		return X86EMUL_PROPAGATE_FAULT;
+	}
+
 	static_call(kvm_x86_patch_hypercall)(vcpu, instruction);
 
 	return emulator_write_emulated(ctxt, rip, instruction, 3,
@@ -9763,7 +9834,8 @@ void kvm_vcpu_update_apicv(struct kvm_vcpu *vcpu)
 
 	down_read(&vcpu->kvm->arch.apicv_update_lock);
 
-	activate = kvm_apicv_activated(vcpu->kvm);
+	activate = kvm_vcpu_apicv_activated(vcpu);
+
 	if (vcpu->arch.apicv_active == activate)
 		goto out;
 
@@ -10171,7 +10243,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 		 * per-VM state, and responsing vCPUs must wait for the update
 		 * to complete before servicing KVM_REQ_APICV_UPDATE.
 		 */
-		WARN_ON_ONCE(kvm_apicv_activated(vcpu->kvm) != kvm_vcpu_apicv_active(vcpu));
+		WARN_ON_ONCE(kvm_vcpu_apicv_activated(vcpu) != kvm_vcpu_apicv_active(vcpu));
 
 		exit_fastpath = static_call(kvm_x86_vcpu_run)(vcpu);
 		if (likely(exit_fastpath != EXIT_FASTPATH_REENTER_GUEST))
@@ -10250,14 +10322,6 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 	 */
 	guest_timing_exit_irqoff();
 
-	if (lapic_in_kernel(vcpu)) {
-		s64 delta = vcpu->arch.apic->lapic_timer.advance_expire_delta;
-		if (delta != S64_MIN) {
-			trace_kvm_wait_lapic_expire(vcpu->vcpu_id, delta);
-			vcpu->arch.apic->lapic_timer.advance_expire_delta = S64_MIN;
-		}
-	}
-
 	local_irq_enable();
 	preempt_enable();
 
@@ -10368,6 +10432,9 @@ static int vcpu_run(struct kvm_vcpu *vcpu)
 			break;
 
 		kvm_clear_request(KVM_REQ_UNBLOCK, vcpu);
+		if (kvm_xen_has_pending_events(vcpu))
+			kvm_xen_inject_pending_events(vcpu);
+
 		if (kvm_cpu_has_pending_timer(vcpu))
 			kvm_inject_pending_timer_irqs(vcpu);
 
@@ -11263,9 +11330,10 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
 
 	vcpu->arch.arch_capabilities = kvm_get_arch_capabilities();
 	vcpu->arch.msr_platform_info = MSR_PLATFORM_INFO_CPUID_FAULT;
+	kvm_xen_init_vcpu(vcpu);
 	kvm_vcpu_mtrr_init(vcpu);
 	vcpu_load(vcpu);
-	kvm_set_tsc_khz(vcpu, max_tsc_khz);
+	kvm_set_tsc_khz(vcpu, vcpu->kvm->arch.default_tsc_khz);
 	kvm_vcpu_reset(vcpu, false);
 	kvm_init_mmu(vcpu);
 	vcpu_put(vcpu);
@@ -11320,6 +11388,7 @@ void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu)
 	free_cpumask_var(vcpu->arch.wbinvd_dirty_mask);
 	fpu_free_guest_fpstate(&vcpu->arch.guest_fpu);
 
+	kvm_xen_destroy_vcpu(vcpu);
 	kvm_hv_vcpu_uninit(vcpu);
 	kvm_pmu_destroy(vcpu);
 	kfree(vcpu->arch.mce_banks);
@@ -11581,6 +11650,24 @@ void kvm_arch_hardware_disable(void)
 	drop_user_return_notifiers();
 }
 
+static inline void kvm_ops_update(struct kvm_x86_init_ops *ops)
+{
+	memcpy(&kvm_x86_ops, ops->runtime_ops, sizeof(kvm_x86_ops));
+
+#define __KVM_X86_OP(func) \
+	static_call_update(kvm_x86_##func, kvm_x86_ops.func);
+#define KVM_X86_OP(func) \
+	WARN_ON(!kvm_x86_ops.func); __KVM_X86_OP(func)
+#define KVM_X86_OP_OPTIONAL __KVM_X86_OP
+#define KVM_X86_OP_OPTIONAL_RET0(func) \
+	static_call_update(kvm_x86_##func, (void *)kvm_x86_ops.func ? : \
+					   (void *)__static_call_return0);
+#include <asm/kvm-x86-ops.h>
+#undef __KVM_X86_OP
+
+	kvm_pmu_ops_update(ops->pmu_ops);
+}
+
 int kvm_arch_hardware_setup(void *opaque)
 {
 	struct kvm_x86_init_ops *ops = opaque;
@@ -11595,8 +11682,7 @@ int kvm_arch_hardware_setup(void *opaque)
 	if (r != 0)
 		return r;
 
-	memcpy(&kvm_x86_ops, ops->runtime_ops, sizeof(kvm_x86_ops));
-	kvm_ops_static_call_update();
+	kvm_ops_update(ops);
 
 	kvm_register_perf_callbacks(ops->handle_intel_pt_intr);
 
@@ -11712,6 +11798,7 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 	pvclock_update_vm_gtod_copy(kvm);
 	raw_spin_unlock_irqrestore(&kvm->arch.tsc_write_lock, flags);
 
+	kvm->arch.default_tsc_khz = max_tsc_khz ? : tsc_khz;
 	kvm->arch.guest_can_read_msr_platform_info = true;
 	kvm->arch.enable_pmu = enable_pmu;
 
@@ -11747,20 +11834,15 @@ static void kvm_unload_vcpu_mmu(struct kvm_vcpu *vcpu)
 	vcpu_put(vcpu);
 }
 
-static void kvm_free_vcpus(struct kvm *kvm)
+static void kvm_unload_vcpu_mmus(struct kvm *kvm)
 {
 	unsigned long i;
 	struct kvm_vcpu *vcpu;
 
-	/*
-	 * Unpin any mmu pages first.
-	 */
 	kvm_for_each_vcpu(i, vcpu, kvm) {
 		kvm_clear_async_pf_completion_queue(vcpu);
 		kvm_unload_vcpu_mmu(vcpu);
 	}
-
-	kvm_destroy_vcpus(kvm);
 }
 
 void kvm_arch_sync_events(struct kvm *kvm)
@@ -11866,11 +11948,12 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
 		__x86_set_memory_region(kvm, TSS_PRIVATE_MEMSLOT, 0, 0);
 		mutex_unlock(&kvm->slots_lock);
 	}
+	kvm_unload_vcpu_mmus(kvm);
 	static_call_cond(kvm_x86_vm_destroy)(kvm);
 	kvm_free_msr_filter(srcu_dereference_check(kvm->arch.msr_filter, &kvm->srcu, 1));
 	kvm_pic_destroy(kvm);
 	kvm_ioapic_destroy(kvm);
-	kvm_free_vcpus(kvm);
+	kvm_destroy_vcpus(kvm);
 	kvfree(rcu_dereference_check(kvm->arch.apic_map, 1));
 	kfree(srcu_dereference_check(kvm->arch.pmu_event_filter, &kvm->srcu, 1));
 	kvm_mmu_uninit_vm(kvm);
@@ -12193,6 +12276,12 @@ static inline bool kvm_vcpu_has_events(struct kvm_vcpu *vcpu)
 	    kvm_x86_ops.nested_ops->hv_timer_pending(vcpu))
 		return true;
 
+	if (kvm_xen_has_pending_events(vcpu))
+		return true;
+
+	if (kvm_test_request(KVM_REQ_TRIPLE_FAULT, vcpu))
+		return true;
+
 	return false;
 }
 
@@ -12290,25 +12379,6 @@ void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
 }
 EXPORT_SYMBOL_GPL(kvm_set_rflags);
 
-void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu, struct kvm_async_pf *work)
-{
-	int r;
-
-	if ((vcpu->arch.mmu->direct_map != work->arch.direct_map) ||
-	      work->wakeup_all)
-		return;
-
-	r = kvm_mmu_reload(vcpu);
-	if (unlikely(r))
-		return;
-
-	if (!vcpu->arch.mmu->direct_map &&
-	      work->arch.cr3 != vcpu->arch.mmu->get_guest_pgd(vcpu))
-		return;
-
-	kvm_mmu_do_page_fault(vcpu, work->cr2_or_gpa, 0, true);
-}
-
 static inline u32 kvm_async_pf_hash_fn(gfn_t gfn)
 {
 	BUILD_BUG_ON(!is_power_of_2(ASYNC_PF_PER_VCPU));
@@ -13000,6 +13070,7 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_pi_irte_update);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_avic_unaccelerated_access);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_avic_incomplete_ipi);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_avic_ga_log);
+EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_avic_kick_vcpu_slowpath);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_apicv_accept_irq);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_enter);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_exit);
diff --git a/arch/x86/kvm/xen.c b/arch/x86/kvm/xen.c
index bf6cc25eee76..610beba35907 100644
--- a/arch/x86/kvm/xen.c
+++ b/arch/x86/kvm/xen.c
@@ -9,17 +9,25 @@
 #include "x86.h"
 #include "xen.h"
 #include "hyperv.h"
+#include "lapic.h"
 
+#include <linux/eventfd.h>
 #include <linux/kvm_host.h>
 #include <linux/sched/stat.h>
 
 #include <trace/events/kvm.h>
 #include <xen/interface/xen.h>
 #include <xen/interface/vcpu.h>
+#include <xen/interface/version.h>
 #include <xen/interface/event_channel.h>
+#include <xen/interface/sched.h>
 
 #include "trace.h"
 
+static int kvm_xen_set_evtchn(struct kvm_xen_evtchn *xe, struct kvm *kvm);
+static int kvm_xen_setattr_evtchn(struct kvm *kvm, struct kvm_xen_hvm_attr *data);
+static bool kvm_xen_hcall_evtchn_send(struct kvm_vcpu *vcpu, u64 param, u64 *r);
+
 DEFINE_STATIC_KEY_DEFERRED_FALSE(kvm_xen_enabled, HZ);
 
 static int kvm_xen_shared_info_init(struct kvm *kvm, gfn_t gfn)
@@ -102,6 +110,66 @@ out:
 	return ret;
 }
 
+void kvm_xen_inject_timer_irqs(struct kvm_vcpu *vcpu)
+{
+	if (atomic_read(&vcpu->arch.xen.timer_pending) > 0) {
+		struct kvm_xen_evtchn e;
+
+		e.vcpu_id = vcpu->vcpu_id;
+		e.vcpu_idx = vcpu->vcpu_idx;
+		e.port = vcpu->arch.xen.timer_virq;
+		e.priority = KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL;
+
+		kvm_xen_set_evtchn(&e, vcpu->kvm);
+
+		vcpu->arch.xen.timer_expires = 0;
+		atomic_set(&vcpu->arch.xen.timer_pending, 0);
+	}
+}
+
+static enum hrtimer_restart xen_timer_callback(struct hrtimer *timer)
+{
+	struct kvm_vcpu *vcpu = container_of(timer, struct kvm_vcpu,
+					     arch.xen.timer);
+	if (atomic_read(&vcpu->arch.xen.timer_pending))
+		return HRTIMER_NORESTART;
+
+	atomic_inc(&vcpu->arch.xen.timer_pending);
+	kvm_make_request(KVM_REQ_UNBLOCK, vcpu);
+	kvm_vcpu_kick(vcpu);
+
+	return HRTIMER_NORESTART;
+}
+
+static void kvm_xen_start_timer(struct kvm_vcpu *vcpu, u64 guest_abs, s64 delta_ns)
+{
+	atomic_set(&vcpu->arch.xen.timer_pending, 0);
+	vcpu->arch.xen.timer_expires = guest_abs;
+
+	if (delta_ns <= 0) {
+		xen_timer_callback(&vcpu->arch.xen.timer);
+	} else {
+		ktime_t ktime_now = ktime_get();
+		hrtimer_start(&vcpu->arch.xen.timer,
+			      ktime_add_ns(ktime_now, delta_ns),
+			      HRTIMER_MODE_ABS_HARD);
+	}
+}
+
+static void kvm_xen_stop_timer(struct kvm_vcpu *vcpu)
+{
+	hrtimer_cancel(&vcpu->arch.xen.timer);
+	vcpu->arch.xen.timer_expires = 0;
+	atomic_set(&vcpu->arch.xen.timer_pending, 0);
+}
+
+static void kvm_xen_init_timer(struct kvm_vcpu *vcpu)
+{
+	hrtimer_init(&vcpu->arch.xen.timer, CLOCK_MONOTONIC,
+		     HRTIMER_MODE_ABS_HARD);
+	vcpu->arch.xen.timer.function = xen_timer_callback;
+}
+
 static void kvm_xen_update_runstate(struct kvm_vcpu *v, int state)
 {
 	struct kvm_vcpu_xen *vx = &v->arch.xen;
@@ -133,27 +201,36 @@ static void kvm_xen_update_runstate(struct kvm_vcpu *v, int state)
 void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, int state)
 {
 	struct kvm_vcpu_xen *vx = &v->arch.xen;
-	struct gfn_to_hva_cache *ghc = &vx->runstate_cache;
-	struct kvm_memslots *slots = kvm_memslots(v->kvm);
-	bool atomic = (state == RUNSTATE_runnable);
-	uint64_t state_entry_time;
-	int __user *user_state;
-	uint64_t __user *user_times;
+	struct gfn_to_pfn_cache *gpc = &vx->runstate_cache;
+	uint64_t *user_times;
+	unsigned long flags;
+	size_t user_len;
+	int *user_state;
 
 	kvm_xen_update_runstate(v, state);
 
-	if (!vx->runstate_set)
+	if (!vx->runstate_cache.active)
 		return;
 
-	if (unlikely(slots->generation != ghc->generation || kvm_is_error_hva(ghc->hva)) &&
-	    kvm_gfn_to_hva_cache_init(v->kvm, ghc, ghc->gpa, ghc->len))
-		return;
+	if (IS_ENABLED(CONFIG_64BIT) && v->kvm->arch.xen.long_mode)
+		user_len = sizeof(struct vcpu_runstate_info);
+	else
+		user_len = sizeof(struct compat_vcpu_runstate_info);
 
-	/* We made sure it fits in a single page */
-	BUG_ON(!ghc->memslot);
+	read_lock_irqsave(&gpc->lock, flags);
+	while (!kvm_gfn_to_pfn_cache_check(v->kvm, gpc, gpc->gpa,
+					   user_len)) {
+		read_unlock_irqrestore(&gpc->lock, flags);
 
-	if (atomic)
-		pagefault_disable();
+		/* When invoked from kvm_sched_out() we cannot sleep */
+		if (state == RUNSTATE_runnable)
+			return;
+
+		if (kvm_gfn_to_pfn_cache_refresh(v->kvm, gpc, gpc->gpa, user_len))
+			return;
+
+		read_lock_irqsave(&gpc->lock, flags);
+	}
 
 	/*
 	 * The only difference between 32-bit and 64-bit versions of the
@@ -167,38 +244,33 @@ void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, int state)
 	 */
 	BUILD_BUG_ON(offsetof(struct vcpu_runstate_info, state) != 0);
 	BUILD_BUG_ON(offsetof(struct compat_vcpu_runstate_info, state) != 0);
-	user_state = (int __user *)ghc->hva;
-
 	BUILD_BUG_ON(sizeof(struct compat_vcpu_runstate_info) != 0x2c);
-
-	user_times = (uint64_t __user *)(ghc->hva +
-					 offsetof(struct compat_vcpu_runstate_info,
-						  state_entry_time));
 #ifdef CONFIG_X86_64
 	BUILD_BUG_ON(offsetof(struct vcpu_runstate_info, state_entry_time) !=
 		     offsetof(struct compat_vcpu_runstate_info, state_entry_time) + 4);
 	BUILD_BUG_ON(offsetof(struct vcpu_runstate_info, time) !=
 		     offsetof(struct compat_vcpu_runstate_info, time) + 4);
-
-	if (v->kvm->arch.xen.long_mode)
-		user_times = (uint64_t __user *)(ghc->hva +
-						 offsetof(struct vcpu_runstate_info,
-							  state_entry_time));
 #endif
+
+	user_state = gpc->khva;
+
+	if (IS_ENABLED(CONFIG_64BIT) && v->kvm->arch.xen.long_mode)
+		user_times = gpc->khva + offsetof(struct vcpu_runstate_info,
+						  state_entry_time);
+	else
+		user_times = gpc->khva + offsetof(struct compat_vcpu_runstate_info,
+						  state_entry_time);
+
 	/*
 	 * First write the updated state_entry_time at the appropriate
 	 * location determined by 'offset'.
 	 */
-	state_entry_time = vx->runstate_entry_time;
-	state_entry_time |= XEN_RUNSTATE_UPDATE;
-
 	BUILD_BUG_ON(sizeof_field(struct vcpu_runstate_info, state_entry_time) !=
-		     sizeof(state_entry_time));
+		     sizeof(user_times[0]));
 	BUILD_BUG_ON(sizeof_field(struct compat_vcpu_runstate_info, state_entry_time) !=
-		     sizeof(state_entry_time));
+		     sizeof(user_times[0]));
 
-	if (__put_user(state_entry_time, user_times))
-		goto out;
+	user_times[0] = vx->runstate_entry_time | XEN_RUNSTATE_UPDATE;
 	smp_wmb();
 
 	/*
@@ -212,8 +284,7 @@ void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, int state)
 	BUILD_BUG_ON(sizeof_field(struct compat_vcpu_runstate_info, state) !=
 		     sizeof(vx->current_runstate));
 
-	if (__put_user(vx->current_runstate, user_state))
-		goto out;
+	*user_state = vx->current_runstate;
 
 	/*
 	 * Write the actual runstate times immediately after the
@@ -228,42 +299,114 @@ void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, int state)
 	BUILD_BUG_ON(sizeof_field(struct vcpu_runstate_info, time) !=
 		     sizeof(vx->runstate_times));
 
-	if (__copy_to_user(user_times + 1, vx->runstate_times, sizeof(vx->runstate_times)))
-		goto out;
+	memcpy(user_times + 1, vx->runstate_times, sizeof(vx->runstate_times));
 	smp_wmb();
 
 	/*
 	 * Finally, clear the XEN_RUNSTATE_UPDATE bit in the guest's
 	 * runstate_entry_time field.
 	 */
-	state_entry_time &= ~XEN_RUNSTATE_UPDATE;
-	__put_user(state_entry_time, user_times);
+	user_times[0] &= ~XEN_RUNSTATE_UPDATE;
 	smp_wmb();
 
- out:
-	mark_page_dirty_in_slot(v->kvm, ghc->memslot, ghc->gpa >> PAGE_SHIFT);
+	read_unlock_irqrestore(&gpc->lock, flags);
 
-	if (atomic)
-		pagefault_enable();
+	mark_page_dirty_in_slot(v->kvm, gpc->memslot, gpc->gpa >> PAGE_SHIFT);
 }
 
-int __kvm_xen_has_interrupt(struct kvm_vcpu *v)
+static void kvm_xen_inject_vcpu_vector(struct kvm_vcpu *v)
+{
+	struct kvm_lapic_irq irq = { };
+	int r;
+
+	irq.dest_id = v->vcpu_id;
+	irq.vector = v->arch.xen.upcall_vector;
+	irq.dest_mode = APIC_DEST_PHYSICAL;
+	irq.shorthand = APIC_DEST_NOSHORT;
+	irq.delivery_mode = APIC_DM_FIXED;
+	irq.level = 1;
+
+	/* The fast version will always work for physical unicast */
+	WARN_ON_ONCE(!kvm_irq_delivery_to_apic_fast(v->kvm, NULL, &irq, &r, NULL));
+}
+
+/*
+ * On event channel delivery, the vcpu_info may not have been accessible.
+ * In that case, there are bits in vcpu->arch.xen.evtchn_pending_sel which
+ * need to be marked into the vcpu_info (and evtchn_upcall_pending set).
+ * Do so now that we can sleep in the context of the vCPU to bring the
+ * page in, and refresh the pfn cache for it.
+ */
+void kvm_xen_inject_pending_events(struct kvm_vcpu *v)
 {
 	unsigned long evtchn_pending_sel = READ_ONCE(v->arch.xen.evtchn_pending_sel);
-	bool atomic = in_atomic() || !task_is_running(current);
-	int err;
+	struct gfn_to_pfn_cache *gpc = &v->arch.xen.vcpu_info_cache;
+	unsigned long flags;
+
+	if (!evtchn_pending_sel)
+		return;
+
+	/*
+	 * Yes, this is an open-coded loop. But that's just what put_user()
+	 * does anyway. Page it in and retry the instruction. We're just a
+	 * little more honest about it.
+	 */
+	read_lock_irqsave(&gpc->lock, flags);
+	while (!kvm_gfn_to_pfn_cache_check(v->kvm, gpc, gpc->gpa,
+					   sizeof(struct vcpu_info))) {
+		read_unlock_irqrestore(&gpc->lock, flags);
+
+		if (kvm_gfn_to_pfn_cache_refresh(v->kvm, gpc, gpc->gpa,
+						 sizeof(struct vcpu_info)))
+			return;
+
+		read_lock_irqsave(&gpc->lock, flags);
+	}
+
+	/* Now gpc->khva is a valid kernel address for the vcpu_info */
+	if (IS_ENABLED(CONFIG_64BIT) && v->kvm->arch.xen.long_mode) {
+		struct vcpu_info *vi = gpc->khva;
+
+		asm volatile(LOCK_PREFIX "orq %0, %1\n"
+			     "notq %0\n"
+			     LOCK_PREFIX "andq %0, %2\n"
+			     : "=r" (evtchn_pending_sel),
+			       "+m" (vi->evtchn_pending_sel),
+			       "+m" (v->arch.xen.evtchn_pending_sel)
+			     : "0" (evtchn_pending_sel));
+		WRITE_ONCE(vi->evtchn_upcall_pending, 1);
+	} else {
+		u32 evtchn_pending_sel32 = evtchn_pending_sel;
+		struct compat_vcpu_info *vi = gpc->khva;
+
+		asm volatile(LOCK_PREFIX "orl %0, %1\n"
+			     "notl %0\n"
+			     LOCK_PREFIX "andl %0, %2\n"
+			     : "=r" (evtchn_pending_sel32),
+			       "+m" (vi->evtchn_pending_sel),
+			       "+m" (v->arch.xen.evtchn_pending_sel)
+			     : "0" (evtchn_pending_sel32));
+		WRITE_ONCE(vi->evtchn_upcall_pending, 1);
+	}
+	read_unlock_irqrestore(&gpc->lock, flags);
+
+	/* For the per-vCPU lapic vector, deliver it as MSI. */
+	if (v->arch.xen.upcall_vector)
+		kvm_xen_inject_vcpu_vector(v);
+
+	mark_page_dirty_in_slot(v->kvm, gpc->memslot, gpc->gpa >> PAGE_SHIFT);
+}
+
+int __kvm_xen_has_interrupt(struct kvm_vcpu *v)
+{
+	struct gfn_to_pfn_cache *gpc = &v->arch.xen.vcpu_info_cache;
+	unsigned long flags;
 	u8 rc = 0;
 
 	/*
 	 * If the global upcall vector (HVMIRQ_callback_vector) is set and
 	 * the vCPU's evtchn_upcall_pending flag is set, the IRQ is pending.
 	 */
-	struct gfn_to_hva_cache *ghc = &v->arch.xen.vcpu_info_cache;
-	struct kvm_memslots *slots = kvm_memslots(v->kvm);
-	bool ghc_valid = slots->generation == ghc->generation &&
-		!kvm_is_error_hva(ghc->hva) && ghc->memslot;
-
-	unsigned int offset = offsetof(struct vcpu_info, evtchn_upcall_pending);
 
 	/* No need for compat handling here */
 	BUILD_BUG_ON(offsetof(struct vcpu_info, evtchn_upcall_pending) !=
@@ -273,101 +416,35 @@ int __kvm_xen_has_interrupt(struct kvm_vcpu *v)
 	BUILD_BUG_ON(sizeof(rc) !=
 		     sizeof_field(struct compat_vcpu_info, evtchn_upcall_pending));
 
-	/*
-	 * For efficiency, this mirrors the checks for using the valid
-	 * cache in kvm_read_guest_offset_cached(), but just uses
-	 * __get_user() instead. And falls back to the slow path.
-	 */
-	if (!evtchn_pending_sel && ghc_valid) {
-		/* Fast path */
-		pagefault_disable();
-		err = __get_user(rc, (u8 __user *)ghc->hva + offset);
-		pagefault_enable();
-		if (!err)
-			return rc;
-	}
-
-	/* Slow path */
+	read_lock_irqsave(&gpc->lock, flags);
+	while (!kvm_gfn_to_pfn_cache_check(v->kvm, gpc, gpc->gpa,
+					   sizeof(struct vcpu_info))) {
+		read_unlock_irqrestore(&gpc->lock, flags);
 
-	/*
-	 * This function gets called from kvm_vcpu_block() after setting the
-	 * task to TASK_INTERRUPTIBLE, to see if it needs to wake immediately
-	 * from a HLT. So we really mustn't sleep. If the page ended up absent
-	 * at that point, just return 1 in order to trigger an immediate wake,
-	 * and we'll end up getting called again from a context where we *can*
-	 * fault in the page and wait for it.
-	 */
-	if (atomic)
-		return 1;
+		/*
+		 * This function gets called from kvm_vcpu_block() after setting the
+		 * task to TASK_INTERRUPTIBLE, to see if it needs to wake immediately
+		 * from a HLT. So we really mustn't sleep. If the page ended up absent
+		 * at that point, just return 1 in order to trigger an immediate wake,
+		 * and we'll end up getting called again from a context where we *can*
+		 * fault in the page and wait for it.
+		 */
+		if (in_atomic() || !task_is_running(current))
+			return 1;
 
-	if (!ghc_valid) {
-		err = kvm_gfn_to_hva_cache_init(v->kvm, ghc, ghc->gpa, ghc->len);
-		if (err || !ghc->memslot) {
+		if (kvm_gfn_to_pfn_cache_refresh(v->kvm, gpc, gpc->gpa,
+						 sizeof(struct vcpu_info))) {
 			/*
 			 * If this failed, userspace has screwed up the
 			 * vcpu_info mapping. No interrupts for you.
 			 */
 			return 0;
 		}
+		read_lock_irqsave(&gpc->lock, flags);
 	}
 
-	/*
-	 * Now we have a valid (protected by srcu) userspace HVA in
-	 * ghc->hva which points to the struct vcpu_info. If there
-	 * are any bits in the in-kernel evtchn_pending_sel then
-	 * we need to write those to the guest vcpu_info and set
-	 * its evtchn_upcall_pending flag. If there aren't any bits
-	 * to add, we only want to *check* evtchn_upcall_pending.
-	 */
-	if (evtchn_pending_sel) {
-		bool long_mode = v->kvm->arch.xen.long_mode;
-
-		if (!user_access_begin((void __user *)ghc->hva, sizeof(struct vcpu_info)))
-			return 0;
-
-		if (IS_ENABLED(CONFIG_64BIT) && long_mode) {
-			struct vcpu_info __user *vi = (void __user *)ghc->hva;
-
-			/* Attempt to set the evtchn_pending_sel bits in the
-			 * guest, and if that succeeds then clear the same
-			 * bits in the in-kernel version. */
-			asm volatile("1:\t" LOCK_PREFIX "orq %0, %1\n"
-				     "\tnotq %0\n"
-				     "\t" LOCK_PREFIX "andq %0, %2\n"
-				     "2:\n"
-				     _ASM_EXTABLE_UA(1b, 2b)
-				     : "=r" (evtchn_pending_sel),
-				       "+m" (vi->evtchn_pending_sel),
-				       "+m" (v->arch.xen.evtchn_pending_sel)
-				     : "0" (evtchn_pending_sel));
-		} else {
-			struct compat_vcpu_info __user *vi = (void __user *)ghc->hva;
-			u32 evtchn_pending_sel32 = evtchn_pending_sel;
-
-			/* Attempt to set the evtchn_pending_sel bits in the
-			 * guest, and if that succeeds then clear the same
-			 * bits in the in-kernel version. */
-			asm volatile("1:\t" LOCK_PREFIX "orl %0, %1\n"
-				     "\tnotl %0\n"
-				     "\t" LOCK_PREFIX "andl %0, %2\n"
-				     "2:\n"
-				     _ASM_EXTABLE_UA(1b, 2b)
-				     : "=r" (evtchn_pending_sel32),
-				       "+m" (vi->evtchn_pending_sel),
-				       "+m" (v->arch.xen.evtchn_pending_sel)
-				     : "0" (evtchn_pending_sel32));
-		}
-		rc = 1;
-		unsafe_put_user(rc, (u8 __user *)ghc->hva + offset, err);
-
-	err:
-		user_access_end();
-
-		mark_page_dirty_in_slot(v->kvm, ghc->memslot, ghc->gpa >> PAGE_SHIFT);
-	} else {
-		__get_user(rc, (u8 __user *)ghc->hva + offset);
-	}
-
+	rc = ((struct vcpu_info *)gpc->khva)->evtchn_upcall_pending;
+	read_unlock_irqrestore(&gpc->lock, flags);
 	return rc;
 }
 
@@ -375,36 +452,51 @@ int kvm_xen_hvm_set_attr(struct kvm *kvm, struct kvm_xen_hvm_attr *data)
 {
 	int r = -ENOENT;
 
-	mutex_lock(&kvm->lock);
 
 	switch (data->type) {
 	case KVM_XEN_ATTR_TYPE_LONG_MODE:
 		if (!IS_ENABLED(CONFIG_64BIT) && data->u.long_mode) {
 			r = -EINVAL;
 		} else {
+			mutex_lock(&kvm->lock);
 			kvm->arch.xen.long_mode = !!data->u.long_mode;
+			mutex_unlock(&kvm->lock);
 			r = 0;
 		}
 		break;
 
 	case KVM_XEN_ATTR_TYPE_SHARED_INFO:
+		mutex_lock(&kvm->lock);
 		r = kvm_xen_shared_info_init(kvm, data->u.shared_info.gfn);
+		mutex_unlock(&kvm->lock);
 		break;
 
 	case KVM_XEN_ATTR_TYPE_UPCALL_VECTOR:
 		if (data->u.vector && data->u.vector < 0x10)
 			r = -EINVAL;
 		else {
+			mutex_lock(&kvm->lock);
 			kvm->arch.xen.upcall_vector = data->u.vector;
+			mutex_unlock(&kvm->lock);
 			r = 0;
 		}
 		break;
 
+	case KVM_XEN_ATTR_TYPE_EVTCHN:
+		r = kvm_xen_setattr_evtchn(kvm, data);
+		break;
+
+	case KVM_XEN_ATTR_TYPE_XEN_VERSION:
+		mutex_lock(&kvm->lock);
+		kvm->arch.xen.xen_version = data->u.xen_version;
+		mutex_unlock(&kvm->lock);
+		r = 0;
+		break;
+
 	default:
 		break;
 	}
 
-	mutex_unlock(&kvm->lock);
 	return r;
 }
 
@@ -433,6 +525,11 @@ int kvm_xen_hvm_get_attr(struct kvm *kvm, struct kvm_xen_hvm_attr *data)
 		r = 0;
 		break;
 
+	case KVM_XEN_ATTR_TYPE_XEN_VERSION:
+		data->u.xen_version = kvm->arch.xen.xen_version;
+		r = 0;
+		break;
+
 	default:
 		break;
 	}
@@ -457,48 +554,34 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 			     offsetof(struct compat_vcpu_info, time));
 
 		if (data->u.gpa == GPA_INVALID) {
-			vcpu->arch.xen.vcpu_info_set = false;
+			kvm_gfn_to_pfn_cache_destroy(vcpu->kvm, &vcpu->arch.xen.vcpu_info_cache);
 			r = 0;
 			break;
 		}
 
-		/* It must fit within a single page */
-		if ((data->u.gpa & ~PAGE_MASK) + sizeof(struct vcpu_info) > PAGE_SIZE) {
-			r = -EINVAL;
-			break;
-		}
-
-		r = kvm_gfn_to_hva_cache_init(vcpu->kvm,
+		r = kvm_gfn_to_pfn_cache_init(vcpu->kvm,
 					      &vcpu->arch.xen.vcpu_info_cache,
-					      data->u.gpa,
+					      NULL, KVM_HOST_USES_PFN, data->u.gpa,
 					      sizeof(struct vcpu_info));
-		if (!r) {
-			vcpu->arch.xen.vcpu_info_set = true;
+		if (!r)
 			kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
-		}
+
 		break;
 
 	case KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO:
 		if (data->u.gpa == GPA_INVALID) {
-			vcpu->arch.xen.vcpu_time_info_set = false;
+			kvm_gfn_to_pfn_cache_destroy(vcpu->kvm,
+						     &vcpu->arch.xen.vcpu_time_info_cache);
 			r = 0;
 			break;
 		}
 
-		/* It must fit within a single page */
-		if ((data->u.gpa & ~PAGE_MASK) + sizeof(struct pvclock_vcpu_time_info) > PAGE_SIZE) {
-			r = -EINVAL;
-			break;
-		}
-
-		r = kvm_gfn_to_hva_cache_init(vcpu->kvm,
+		r = kvm_gfn_to_pfn_cache_init(vcpu->kvm,
 					      &vcpu->arch.xen.vcpu_time_info_cache,
-					      data->u.gpa,
+					      NULL, KVM_HOST_USES_PFN, data->u.gpa,
 					      sizeof(struct pvclock_vcpu_time_info));
-		if (!r) {
-			vcpu->arch.xen.vcpu_time_info_set = true;
+		if (!r)
 			kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
-		}
 		break;
 
 	case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR:
@@ -507,24 +590,16 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 			break;
 		}
 		if (data->u.gpa == GPA_INVALID) {
-			vcpu->arch.xen.runstate_set = false;
+			kvm_gfn_to_pfn_cache_destroy(vcpu->kvm,
+						     &vcpu->arch.xen.runstate_cache);
 			r = 0;
 			break;
 		}
 
-		/* It must fit within a single page */
-		if ((data->u.gpa & ~PAGE_MASK) + sizeof(struct vcpu_runstate_info) > PAGE_SIZE) {
-			r = -EINVAL;
-			break;
-		}
-
-		r = kvm_gfn_to_hva_cache_init(vcpu->kvm,
+		r = kvm_gfn_to_pfn_cache_init(vcpu->kvm,
 					      &vcpu->arch.xen.runstate_cache,
-					      data->u.gpa,
+					      NULL, KVM_HOST_USES_PFN, data->u.gpa,
 					      sizeof(struct vcpu_runstate_info));
-		if (!r) {
-			vcpu->arch.xen.runstate_set = true;
-		}
 		break;
 
 	case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT:
@@ -622,6 +697,46 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 		r = 0;
 		break;
 
+	case KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID:
+		if (data->u.vcpu_id >= KVM_MAX_VCPUS)
+			r = -EINVAL;
+		else {
+			vcpu->arch.xen.vcpu_id = data->u.vcpu_id;
+			r = 0;
+		}
+		break;
+
+	case KVM_XEN_VCPU_ATTR_TYPE_TIMER:
+		if (data->u.timer.port) {
+			if (data->u.timer.priority != KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL) {
+				r = -EINVAL;
+				break;
+			}
+			vcpu->arch.xen.timer_virq = data->u.timer.port;
+			kvm_xen_init_timer(vcpu);
+
+			/* Restart the timer if it's set */
+			if (data->u.timer.expires_ns)
+				kvm_xen_start_timer(vcpu, data->u.timer.expires_ns,
+						    data->u.timer.expires_ns -
+						    get_kvmclock_ns(vcpu->kvm));
+		} else if (kvm_xen_timer_enabled(vcpu)) {
+			kvm_xen_stop_timer(vcpu);
+			vcpu->arch.xen.timer_virq = 0;
+		}
+
+		r = 0;
+		break;
+
+	case KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR:
+		if (data->u.vector && data->u.vector < 0x10)
+			r = -EINVAL;
+		else {
+			vcpu->arch.xen.upcall_vector = data->u.vector;
+			r = 0;
+		}
+		break;
+
 	default:
 		break;
 	}
@@ -639,7 +754,7 @@ int kvm_xen_vcpu_get_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 
 	switch (data->type) {
 	case KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO:
-		if (vcpu->arch.xen.vcpu_info_set)
+		if (vcpu->arch.xen.vcpu_info_cache.active)
 			data->u.gpa = vcpu->arch.xen.vcpu_info_cache.gpa;
 		else
 			data->u.gpa = GPA_INVALID;
@@ -647,7 +762,7 @@ int kvm_xen_vcpu_get_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 		break;
 
 	case KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO:
-		if (vcpu->arch.xen.vcpu_time_info_set)
+		if (vcpu->arch.xen.vcpu_time_info_cache.active)
 			data->u.gpa = vcpu->arch.xen.vcpu_time_info_cache.gpa;
 		else
 			data->u.gpa = GPA_INVALID;
@@ -659,7 +774,7 @@ int kvm_xen_vcpu_get_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 			r = -EOPNOTSUPP;
 			break;
 		}
-		if (vcpu->arch.xen.runstate_set) {
+		if (vcpu->arch.xen.runstate_cache.active) {
 			data->u.gpa = vcpu->arch.xen.runstate_cache.gpa;
 			r = 0;
 		}
@@ -697,6 +812,23 @@ int kvm_xen_vcpu_get_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 		r = -EINVAL;
 		break;
 
+	case KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID:
+		data->u.vcpu_id = vcpu->arch.xen.vcpu_id;
+		r = 0;
+		break;
+
+	case KVM_XEN_VCPU_ATTR_TYPE_TIMER:
+		data->u.timer.port = vcpu->arch.xen.timer_virq;
+		data->u.timer.priority = KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL;
+		data->u.timer.expires_ns = vcpu->arch.xen.timer_expires;
+		r = 0;
+		break;
+
+	case KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR:
+		data->u.vector = vcpu->arch.xen.upcall_vector;
+		r = 0;
+		break;
+
 	default:
 		break;
 	}
@@ -777,7 +909,11 @@ int kvm_xen_write_hypercall_page(struct kvm_vcpu *vcpu, u64 data)
 
 int kvm_xen_hvm_config(struct kvm *kvm, struct kvm_xen_hvm_config *xhc)
 {
-	if (xhc->flags & ~KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL)
+	/* Only some feature flags need to be *enabled* by userspace */
+	u32 permitted_flags = KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL |
+		KVM_XEN_HVM_CONFIG_EVTCHN_SEND;
+
+	if (xhc->flags & ~permitted_flags)
 		return -EINVAL;
 
 	/*
@@ -802,18 +938,6 @@ int kvm_xen_hvm_config(struct kvm *kvm, struct kvm_xen_hvm_config *xhc)
 	return 0;
 }
 
-void kvm_xen_init_vm(struct kvm *kvm)
-{
-}
-
-void kvm_xen_destroy_vm(struct kvm *kvm)
-{
-	kvm_gfn_to_pfn_cache_destroy(kvm, &kvm->arch.xen.shinfo_cache);
-
-	if (kvm->arch.xen_hvm_config.msr)
-		static_branch_slow_dec_deferred(&kvm_xen_enabled);
-}
-
 static int kvm_xen_hypercall_set_result(struct kvm_vcpu *vcpu, u64 result)
 {
 	kvm_rax_write(vcpu, result);
@@ -830,10 +954,268 @@ static int kvm_xen_hypercall_complete_userspace(struct kvm_vcpu *vcpu)
 	return kvm_xen_hypercall_set_result(vcpu, run->xen.u.hcall.result);
 }
 
+static bool wait_pending_event(struct kvm_vcpu *vcpu, int nr_ports,
+			       evtchn_port_t *ports)
+{
+	struct kvm *kvm = vcpu->kvm;
+	struct gfn_to_pfn_cache *gpc = &kvm->arch.xen.shinfo_cache;
+	unsigned long *pending_bits;
+	unsigned long flags;
+	bool ret = true;
+	int idx, i;
+
+	read_lock_irqsave(&gpc->lock, flags);
+	idx = srcu_read_lock(&kvm->srcu);
+	if (!kvm_gfn_to_pfn_cache_check(kvm, gpc, gpc->gpa, PAGE_SIZE))
+		goto out_rcu;
+
+	ret = false;
+	if (IS_ENABLED(CONFIG_64BIT) && kvm->arch.xen.long_mode) {
+		struct shared_info *shinfo = gpc->khva;
+		pending_bits = (unsigned long *)&shinfo->evtchn_pending;
+	} else {
+		struct compat_shared_info *shinfo = gpc->khva;
+		pending_bits = (unsigned long *)&shinfo->evtchn_pending;
+	}
+
+	for (i = 0; i < nr_ports; i++) {
+		if (test_bit(ports[i], pending_bits)) {
+			ret = true;
+			break;
+		}
+	}
+
+ out_rcu:
+	srcu_read_unlock(&kvm->srcu, idx);
+	read_unlock_irqrestore(&gpc->lock, flags);
+
+	return ret;
+}
+
+static bool kvm_xen_schedop_poll(struct kvm_vcpu *vcpu, bool longmode,
+				 u64 param, u64 *r)
+{
+	int idx, i;
+	struct sched_poll sched_poll;
+	evtchn_port_t port, *ports;
+	gpa_t gpa;
+
+	if (!longmode || !lapic_in_kernel(vcpu) ||
+	    !(vcpu->kvm->arch.xen_hvm_config.flags & KVM_XEN_HVM_CONFIG_EVTCHN_SEND))
+		return false;
+
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
+	gpa = kvm_mmu_gva_to_gpa_system(vcpu, param, NULL);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
+
+	if (!gpa || kvm_vcpu_read_guest(vcpu, gpa, &sched_poll,
+					sizeof(sched_poll))) {
+		*r = -EFAULT;
+		return true;
+	}
+
+	if (unlikely(sched_poll.nr_ports > 1)) {
+		/* Xen (unofficially) limits number of pollers to 128 */
+		if (sched_poll.nr_ports > 128) {
+			*r = -EINVAL;
+			return true;
+		}
+
+		ports = kmalloc_array(sched_poll.nr_ports,
+				      sizeof(*ports), GFP_KERNEL);
+		if (!ports) {
+			*r = -ENOMEM;
+			return true;
+		}
+	} else
+		ports = &port;
+
+	for (i = 0; i < sched_poll.nr_ports; i++) {
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
+		gpa = kvm_mmu_gva_to_gpa_system(vcpu,
+						(gva_t)(sched_poll.ports + i),
+						NULL);
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
+
+		if (!gpa || kvm_vcpu_read_guest(vcpu, gpa,
+						&ports[i], sizeof(port))) {
+			*r = -EFAULT;
+			goto out;
+		}
+	}
+
+	if (sched_poll.nr_ports == 1)
+		vcpu->arch.xen.poll_evtchn = port;
+	else
+		vcpu->arch.xen.poll_evtchn = -1;
+
+	set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.xen.poll_mask);
+
+	if (!wait_pending_event(vcpu, sched_poll.nr_ports, ports)) {
+		vcpu->arch.mp_state = KVM_MP_STATE_HALTED;
+
+		if (sched_poll.timeout)
+			mod_timer(&vcpu->arch.xen.poll_timer,
+				  jiffies + nsecs_to_jiffies(sched_poll.timeout));
+
+		kvm_vcpu_halt(vcpu);
+
+		if (sched_poll.timeout)
+			del_timer(&vcpu->arch.xen.poll_timer);
+
+		vcpu->arch.mp_state = KVM_MP_STATE_RUNNABLE;
+		kvm_clear_request(KVM_REQ_UNHALT, vcpu);
+	}
+
+	vcpu->arch.xen.poll_evtchn = 0;
+	*r = 0;
+out:
+	/* Really, this is only needed in case of timeout */
+	clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.xen.poll_mask);
+
+	if (unlikely(sched_poll.nr_ports > 1))
+		kfree(ports);
+	return true;
+}
+
+static void cancel_evtchn_poll(struct timer_list *t)
+{
+	struct kvm_vcpu *vcpu = from_timer(vcpu, t, arch.xen.poll_timer);
+
+	kvm_make_request(KVM_REQ_UNBLOCK, vcpu);
+	kvm_vcpu_kick(vcpu);
+}
+
+static bool kvm_xen_hcall_sched_op(struct kvm_vcpu *vcpu, bool longmode,
+				   int cmd, u64 param, u64 *r)
+{
+	switch (cmd) {
+	case SCHEDOP_poll:
+		if (kvm_xen_schedop_poll(vcpu, longmode, param, r))
+			return true;
+		fallthrough;
+	case SCHEDOP_yield:
+		kvm_vcpu_on_spin(vcpu, true);
+		*r = 0;
+		return true;
+	default:
+		break;
+	}
+
+	return false;
+}
+
+struct compat_vcpu_set_singleshot_timer {
+    uint64_t timeout_abs_ns;
+    uint32_t flags;
+} __attribute__((packed));
+
+static bool kvm_xen_hcall_vcpu_op(struct kvm_vcpu *vcpu, bool longmode, int cmd,
+				  int vcpu_id, u64 param, u64 *r)
+{
+	struct vcpu_set_singleshot_timer oneshot;
+	s64 delta;
+	gpa_t gpa;
+	int idx;
+
+	if (!kvm_xen_timer_enabled(vcpu))
+		return false;
+
+	switch (cmd) {
+	case VCPUOP_set_singleshot_timer:
+		if (vcpu->arch.xen.vcpu_id != vcpu_id) {
+			*r = -EINVAL;
+			return true;
+		}
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
+		gpa = kvm_mmu_gva_to_gpa_system(vcpu, param, NULL);
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
+
+		/*
+		 * The only difference for 32-bit compat is the 4 bytes of
+		 * padding after the interesting part of the structure. So
+		 * for a faithful emulation of Xen we have to *try* to copy
+		 * the padding and return -EFAULT if we can't. Otherwise we
+		 * might as well just have copied the 12-byte 32-bit struct.
+		 */
+		BUILD_BUG_ON(offsetof(struct compat_vcpu_set_singleshot_timer, timeout_abs_ns) !=
+			     offsetof(struct vcpu_set_singleshot_timer, timeout_abs_ns));
+		BUILD_BUG_ON(sizeof_field(struct compat_vcpu_set_singleshot_timer, timeout_abs_ns) !=
+			     sizeof_field(struct vcpu_set_singleshot_timer, timeout_abs_ns));
+		BUILD_BUG_ON(offsetof(struct compat_vcpu_set_singleshot_timer, flags) !=
+			     offsetof(struct vcpu_set_singleshot_timer, flags));
+		BUILD_BUG_ON(sizeof_field(struct compat_vcpu_set_singleshot_timer, flags) !=
+			     sizeof_field(struct vcpu_set_singleshot_timer, flags));
+
+		if (!gpa ||
+		    kvm_vcpu_read_guest(vcpu, gpa, &oneshot, longmode ? sizeof(oneshot) :
+					sizeof(struct compat_vcpu_set_singleshot_timer))) {
+			*r = -EFAULT;
+			return true;
+		}
+
+		delta = oneshot.timeout_abs_ns - get_kvmclock_ns(vcpu->kvm);
+		if ((oneshot.flags & VCPU_SSHOTTMR_future) && delta < 0) {
+			*r = -ETIME;
+			return true;
+		}
+
+		kvm_xen_start_timer(vcpu, oneshot.timeout_abs_ns, delta);
+		*r = 0;
+		return true;
+
+	case VCPUOP_stop_singleshot_timer:
+		if (vcpu->arch.xen.vcpu_id != vcpu_id) {
+			*r = -EINVAL;
+			return true;
+		}
+		kvm_xen_stop_timer(vcpu);
+		*r = 0;
+		return true;
+	}
+
+	return false;
+}
+
+static bool kvm_xen_hcall_set_timer_op(struct kvm_vcpu *vcpu, uint64_t timeout,
+				       u64 *r)
+{
+	if (!kvm_xen_timer_enabled(vcpu))
+		return false;
+
+	if (timeout) {
+		uint64_t guest_now = get_kvmclock_ns(vcpu->kvm);
+		int64_t delta = timeout - guest_now;
+
+		/* Xen has a 'Linux workaround' in do_set_timer_op() which
+		 * checks for negative absolute timeout values (caused by
+		 * integer overflow), and for values about 13 days in the
+		 * future (2^50ns) which would be caused by jiffies
+		 * overflow. For those cases, it sets the timeout 100ms in
+		 * the future (not *too* soon, since if a guest really did
+		 * set a long timeout on purpose we don't want to keep
+		 * churning CPU time by waking it up).
+		 */
+		if (unlikely((int64_t)timeout < 0 ||
+			     (delta > 0 && (uint32_t) (delta >> 50) != 0))) {
+			delta = 100 * NSEC_PER_MSEC;
+			timeout = guest_now + delta;
+		}
+
+		kvm_xen_start_timer(vcpu, timeout, delta);
+	} else {
+		kvm_xen_stop_timer(vcpu);
+	}
+
+	*r = 0;
+	return true;
+}
+
 int kvm_xen_hypercall(struct kvm_vcpu *vcpu)
 {
 	bool longmode;
-	u64 input, params[6];
+	u64 input, params[6], r = -ENOSYS;
+	bool handled = false;
 
 	input = (u64)kvm_register_read(vcpu, VCPU_REGS_RAX);
 
@@ -864,6 +1246,40 @@ int kvm_xen_hypercall(struct kvm_vcpu *vcpu)
 	trace_kvm_xen_hypercall(input, params[0], params[1], params[2],
 				params[3], params[4], params[5]);
 
+	switch (input) {
+	case __HYPERVISOR_xen_version:
+		if (params[0] == XENVER_version && vcpu->kvm->arch.xen.xen_version) {
+			r = vcpu->kvm->arch.xen.xen_version;
+			handled = true;
+		}
+		break;
+	case __HYPERVISOR_event_channel_op:
+		if (params[0] == EVTCHNOP_send)
+			handled = kvm_xen_hcall_evtchn_send(vcpu, params[1], &r);
+		break;
+	case __HYPERVISOR_sched_op:
+		handled = kvm_xen_hcall_sched_op(vcpu, longmode, params[0],
+						 params[1], &r);
+		break;
+	case __HYPERVISOR_vcpu_op:
+		handled = kvm_xen_hcall_vcpu_op(vcpu, longmode, params[0], params[1],
+						params[2], &r);
+		break;
+	case __HYPERVISOR_set_timer_op: {
+		u64 timeout = params[0];
+		/* In 32-bit mode, the 64-bit timeout is in two 32-bit params. */
+		if (!longmode)
+			timeout |= params[1] << 32;
+		handled = kvm_xen_hcall_set_timer_op(vcpu, timeout, &r);
+		break;
+	}
+	default:
+		break;
+	}
+
+	if (handled)
+		return kvm_xen_hypercall_set_result(vcpu, r);
+
 	vcpu->run->exit_reason = KVM_EXIT_XEN;
 	vcpu->run->xen.type = KVM_EXIT_XEN_HCALL;
 	vcpu->run->xen.u.hcall.longmode = longmode;
@@ -890,14 +1306,28 @@ static inline int max_evtchn_port(struct kvm *kvm)
 		return COMPAT_EVTCHN_2L_NR_CHANNELS;
 }
 
+static void kvm_xen_check_poller(struct kvm_vcpu *vcpu, int port)
+{
+	int poll_evtchn = vcpu->arch.xen.poll_evtchn;
+
+	if ((poll_evtchn == port || poll_evtchn == -1) &&
+	    test_and_clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.xen.poll_mask)) {
+		kvm_make_request(KVM_REQ_UNBLOCK, vcpu);
+		kvm_vcpu_kick(vcpu);
+	}
+}
+
 /*
- * This follows the kvm_set_irq() API, so it returns:
+ * The return value from this function is propagated to kvm_set_irq() API,
+ * so it returns:
  *  < 0   Interrupt was ignored (masked or not delivered for other reasons)
  *  = 0   Interrupt was coalesced (previous irq is still pending)
  *  > 0   Number of CPUs interrupt was delivered to
+ *
+ * It is also called directly from kvm_arch_set_irq_inatomic(), where the
+ * only check on its return value is a comparison with -EWOULDBLOCK'.
  */
-int kvm_xen_set_evtchn_fast(struct kvm_kernel_irq_routing_entry *e,
-			    struct kvm *kvm)
+int kvm_xen_set_evtchn_fast(struct kvm_xen_evtchn *xe, struct kvm *kvm)
 {
 	struct gfn_to_pfn_cache *gpc = &kvm->arch.xen.shinfo_cache;
 	struct kvm_vcpu *vcpu;
@@ -905,23 +1335,29 @@ int kvm_xen_set_evtchn_fast(struct kvm_kernel_irq_routing_entry *e,
 	unsigned long flags;
 	int port_word_bit;
 	bool kick_vcpu = false;
-	int idx;
-	int rc;
+	int vcpu_idx, idx, rc;
 
-	vcpu = kvm_get_vcpu_by_id(kvm, e->xen_evtchn.vcpu);
-	if (!vcpu)
-		return -1;
+	vcpu_idx = READ_ONCE(xe->vcpu_idx);
+	if (vcpu_idx >= 0)
+		vcpu = kvm_get_vcpu(kvm, vcpu_idx);
+	else {
+		vcpu = kvm_get_vcpu_by_id(kvm, xe->vcpu_id);
+		if (!vcpu)
+			return -EINVAL;
+		WRITE_ONCE(xe->vcpu_idx, kvm_vcpu_get_idx(vcpu));
+	}
 
-	if (!vcpu->arch.xen.vcpu_info_set)
-		return -1;
+	if (!vcpu->arch.xen.vcpu_info_cache.active)
+		return -EINVAL;
 
-	if (e->xen_evtchn.port >= max_evtchn_port(kvm))
-		return -1;
+	if (xe->port >= max_evtchn_port(kvm))
+		return -EINVAL;
 
 	rc = -EWOULDBLOCK;
-	read_lock_irqsave(&gpc->lock, flags);
 
 	idx = srcu_read_lock(&kvm->srcu);
+
+	read_lock_irqsave(&gpc->lock, flags);
 	if (!kvm_gfn_to_pfn_cache_check(kvm, gpc, gpc->gpa, PAGE_SIZE))
 		goto out_rcu;
 
@@ -929,12 +1365,12 @@ int kvm_xen_set_evtchn_fast(struct kvm_kernel_irq_routing_entry *e,
 		struct shared_info *shinfo = gpc->khva;
 		pending_bits = (unsigned long *)&shinfo->evtchn_pending;
 		mask_bits = (unsigned long *)&shinfo->evtchn_mask;
-		port_word_bit = e->xen_evtchn.port / 64;
+		port_word_bit = xe->port / 64;
 	} else {
 		struct compat_shared_info *shinfo = gpc->khva;
 		pending_bits = (unsigned long *)&shinfo->evtchn_pending;
 		mask_bits = (unsigned long *)&shinfo->evtchn_mask;
-		port_word_bit = e->xen_evtchn.port / 32;
+		port_word_bit = xe->port / 32;
 	}
 
 	/*
@@ -944,39 +1380,68 @@ int kvm_xen_set_evtchn_fast(struct kvm_kernel_irq_routing_entry *e,
 	 * already set, then we kick the vCPU in question to write to the
 	 * *real* evtchn_pending_sel in its own guest vcpu_info struct.
 	 */
-	if (test_and_set_bit(e->xen_evtchn.port, pending_bits)) {
+	if (test_and_set_bit(xe->port, pending_bits)) {
 		rc = 0; /* It was already raised */
-	} else if (test_bit(e->xen_evtchn.port, mask_bits)) {
-		rc = -1; /* Masked */
+	} else if (test_bit(xe->port, mask_bits)) {
+		rc = -ENOTCONN; /* Masked */
+		kvm_xen_check_poller(vcpu, xe->port);
 	} else {
-		rc = 1; /* Delivered. But was the vCPU waking already? */
-		if (!test_and_set_bit(port_word_bit, &vcpu->arch.xen.evtchn_pending_sel))
-			kick_vcpu = true;
+		rc = 1; /* Delivered to the bitmap in shared_info. */
+		/* Now switch to the vCPU's vcpu_info to set the index and pending_sel */
+		read_unlock_irqrestore(&gpc->lock, flags);
+		gpc = &vcpu->arch.xen.vcpu_info_cache;
+
+		read_lock_irqsave(&gpc->lock, flags);
+		if (!kvm_gfn_to_pfn_cache_check(kvm, gpc, gpc->gpa, sizeof(struct vcpu_info))) {
+			/*
+			 * Could not access the vcpu_info. Set the bit in-kernel
+			 * and prod the vCPU to deliver it for itself.
+			 */
+			if (!test_and_set_bit(port_word_bit, &vcpu->arch.xen.evtchn_pending_sel))
+				kick_vcpu = true;
+			goto out_rcu;
+		}
+
+		if (IS_ENABLED(CONFIG_64BIT) && kvm->arch.xen.long_mode) {
+			struct vcpu_info *vcpu_info = gpc->khva;
+			if (!test_and_set_bit(port_word_bit, &vcpu_info->evtchn_pending_sel)) {
+				WRITE_ONCE(vcpu_info->evtchn_upcall_pending, 1);
+				kick_vcpu = true;
+			}
+		} else {
+			struct compat_vcpu_info *vcpu_info = gpc->khva;
+			if (!test_and_set_bit(port_word_bit,
+					      (unsigned long *)&vcpu_info->evtchn_pending_sel)) {
+				WRITE_ONCE(vcpu_info->evtchn_upcall_pending, 1);
+				kick_vcpu = true;
+			}
+		}
+
+		/* For the per-vCPU lapic vector, deliver it as MSI. */
+		if (kick_vcpu && vcpu->arch.xen.upcall_vector) {
+			kvm_xen_inject_vcpu_vector(vcpu);
+			kick_vcpu = false;
+		}
 	}
 
  out_rcu:
-	srcu_read_unlock(&kvm->srcu, idx);
 	read_unlock_irqrestore(&gpc->lock, flags);
+	srcu_read_unlock(&kvm->srcu, idx);
 
 	if (kick_vcpu) {
-		kvm_make_request(KVM_REQ_EVENT, vcpu);
+		kvm_make_request(KVM_REQ_UNBLOCK, vcpu);
 		kvm_vcpu_kick(vcpu);
 	}
 
 	return rc;
 }
 
-/* This is the version called from kvm_set_irq() as the .set function */
-static int evtchn_set_fn(struct kvm_kernel_irq_routing_entry *e, struct kvm *kvm,
-			 int irq_source_id, int level, bool line_status)
+static int kvm_xen_set_evtchn(struct kvm_xen_evtchn *xe, struct kvm *kvm)
 {
 	bool mm_borrowed = false;
 	int rc;
 
-	if (!level)
-		return -1;
-
-	rc = kvm_xen_set_evtchn_fast(e, kvm);
+	rc = kvm_xen_set_evtchn_fast(xe, kvm);
 	if (rc != -EWOULDBLOCK)
 		return rc;
 
@@ -1020,7 +1485,7 @@ static int evtchn_set_fn(struct kvm_kernel_irq_routing_entry *e, struct kvm *kvm
 		struct gfn_to_pfn_cache *gpc = &kvm->arch.xen.shinfo_cache;
 		int idx;
 
-		rc = kvm_xen_set_evtchn_fast(e, kvm);
+		rc = kvm_xen_set_evtchn_fast(xe, kvm);
 		if (rc != -EWOULDBLOCK)
 			break;
 
@@ -1037,11 +1502,27 @@ static int evtchn_set_fn(struct kvm_kernel_irq_routing_entry *e, struct kvm *kvm
 	return rc;
 }
 
+/* This is the version called from kvm_set_irq() as the .set function */
+static int evtchn_set_fn(struct kvm_kernel_irq_routing_entry *e, struct kvm *kvm,
+			 int irq_source_id, int level, bool line_status)
+{
+	if (!level)
+		return -EINVAL;
+
+	return kvm_xen_set_evtchn(&e->xen_evtchn, kvm);
+}
+
+/*
+ * Set up an event channel interrupt from the KVM IRQ routing table.
+ * Used for e.g. PIRQ from passed through physical devices.
+ */
 int kvm_xen_setup_evtchn(struct kvm *kvm,
 			 struct kvm_kernel_irq_routing_entry *e,
 			 const struct kvm_irq_routing_entry *ue)
 
 {
+	struct kvm_vcpu *vcpu;
+
 	if (ue->u.xen_evtchn.port >= max_evtchn_port(kvm))
 		return -EINVAL;
 
@@ -1049,10 +1530,328 @@ int kvm_xen_setup_evtchn(struct kvm *kvm,
 	if (ue->u.xen_evtchn.priority != KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL)
 		return -EINVAL;
 
+	/*
+	 * Xen gives us interesting mappings from vCPU index to APIC ID,
+	 * which means kvm_get_vcpu_by_id() has to iterate over all vCPUs
+	 * to find it. Do that once at setup time, instead of every time.
+	 * But beware that on live update / live migration, the routing
+	 * table might be reinstated before the vCPU threads have finished
+	 * recreating their vCPUs.
+	 */
+	vcpu = kvm_get_vcpu_by_id(kvm, ue->u.xen_evtchn.vcpu);
+	if (vcpu)
+		e->xen_evtchn.vcpu_idx = kvm_vcpu_get_idx(vcpu);
+	else
+		e->xen_evtchn.vcpu_idx = -1;
+
 	e->xen_evtchn.port = ue->u.xen_evtchn.port;
-	e->xen_evtchn.vcpu = ue->u.xen_evtchn.vcpu;
+	e->xen_evtchn.vcpu_id = ue->u.xen_evtchn.vcpu;
 	e->xen_evtchn.priority = ue->u.xen_evtchn.priority;
 	e->set = evtchn_set_fn;
 
 	return 0;
 }
+
+/*
+ * Explicit event sending from userspace with KVM_XEN_HVM_EVTCHN_SEND ioctl.
+ */
+int kvm_xen_hvm_evtchn_send(struct kvm *kvm, struct kvm_irq_routing_xen_evtchn *uxe)
+{
+	struct kvm_xen_evtchn e;
+	int ret;
+
+	if (!uxe->port || uxe->port >= max_evtchn_port(kvm))
+		return -EINVAL;
+
+	/* We only support 2 level event channels for now */
+	if (uxe->priority != KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL)
+		return -EINVAL;
+
+	e.port = uxe->port;
+	e.vcpu_id = uxe->vcpu;
+	e.vcpu_idx = -1;
+	e.priority = uxe->priority;
+
+	ret = kvm_xen_set_evtchn(&e, kvm);
+
+	/*
+	 * None of that 'return 1 if it actually got delivered' nonsense.
+	 * We don't care if it was masked (-ENOTCONN) either.
+	 */
+	if (ret > 0 || ret == -ENOTCONN)
+		ret = 0;
+
+	return ret;
+}
+
+/*
+ * Support for *outbound* event channel events via the EVTCHNOP_send hypercall.
+ */
+struct evtchnfd {
+	u32 send_port;
+	u32 type;
+	union {
+		struct kvm_xen_evtchn port;
+		struct {
+			u32 port; /* zero */
+			struct eventfd_ctx *ctx;
+		} eventfd;
+	} deliver;
+};
+
+/*
+ * Update target vCPU or priority for a registered sending channel.
+ */
+static int kvm_xen_eventfd_update(struct kvm *kvm,
+				  struct kvm_xen_hvm_attr *data)
+{
+	u32 port = data->u.evtchn.send_port;
+	struct evtchnfd *evtchnfd;
+
+	if (!port || port >= max_evtchn_port(kvm))
+		return -EINVAL;
+
+	mutex_lock(&kvm->lock);
+	evtchnfd = idr_find(&kvm->arch.xen.evtchn_ports, port);
+	mutex_unlock(&kvm->lock);
+
+	if (!evtchnfd)
+		return -ENOENT;
+
+	/* For an UPDATE, nothing may change except the priority/vcpu */
+	if (evtchnfd->type != data->u.evtchn.type)
+		return -EINVAL;
+
+	/*
+	 * Port cannot change, and if it's zero that was an eventfd
+	 * which can't be changed either.
+	 */
+	if (!evtchnfd->deliver.port.port ||
+	    evtchnfd->deliver.port.port != data->u.evtchn.deliver.port.port)
+		return -EINVAL;
+
+	/* We only support 2 level event channels for now */
+	if (data->u.evtchn.deliver.port.priority != KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL)
+		return -EINVAL;
+
+	mutex_lock(&kvm->lock);
+	evtchnfd->deliver.port.priority = data->u.evtchn.deliver.port.priority;
+	if (evtchnfd->deliver.port.vcpu_id != data->u.evtchn.deliver.port.vcpu) {
+		evtchnfd->deliver.port.vcpu_id = data->u.evtchn.deliver.port.vcpu;
+		evtchnfd->deliver.port.vcpu_idx = -1;
+	}
+	mutex_unlock(&kvm->lock);
+	return 0;
+}
+
+/*
+ * Configure the target (eventfd or local port delivery) for sending on
+ * a given event channel.
+ */
+static int kvm_xen_eventfd_assign(struct kvm *kvm,
+				  struct kvm_xen_hvm_attr *data)
+{
+	u32 port = data->u.evtchn.send_port;
+	struct eventfd_ctx *eventfd = NULL;
+	struct evtchnfd *evtchnfd = NULL;
+	int ret = -EINVAL;
+
+	if (!port || port >= max_evtchn_port(kvm))
+		return -EINVAL;
+
+	evtchnfd = kzalloc(sizeof(struct evtchnfd), GFP_KERNEL);
+	if (!evtchnfd)
+		return -ENOMEM;
+
+	switch(data->u.evtchn.type) {
+	case EVTCHNSTAT_ipi:
+		/* IPI  must map back to the same port# */
+		if (data->u.evtchn.deliver.port.port != data->u.evtchn.send_port)
+			goto out; /* -EINVAL */
+		break;
+
+	case EVTCHNSTAT_interdomain:
+		if (data->u.evtchn.deliver.port.port) {
+			if (data->u.evtchn.deliver.port.port >= max_evtchn_port(kvm))
+				goto out; /* -EINVAL */
+		} else {
+			eventfd = eventfd_ctx_fdget(data->u.evtchn.deliver.eventfd.fd);
+			if (IS_ERR(eventfd)) {
+				ret = PTR_ERR(eventfd);
+				goto out;
+			}
+		}
+		break;
+
+	case EVTCHNSTAT_virq:
+	case EVTCHNSTAT_closed:
+	case EVTCHNSTAT_unbound:
+	case EVTCHNSTAT_pirq:
+	default: /* Unknown event channel type */
+		goto out; /* -EINVAL */
+	}
+
+	evtchnfd->send_port = data->u.evtchn.send_port;
+	evtchnfd->type = data->u.evtchn.type;
+	if (eventfd) {
+		evtchnfd->deliver.eventfd.ctx = eventfd;
+	} else {
+		/* We only support 2 level event channels for now */
+		if (data->u.evtchn.deliver.port.priority != KVM_IRQ_ROUTING_XEN_EVTCHN_PRIO_2LEVEL)
+			goto out; /* -EINVAL; */
+
+		evtchnfd->deliver.port.port = data->u.evtchn.deliver.port.port;
+		evtchnfd->deliver.port.vcpu_id = data->u.evtchn.deliver.port.vcpu;
+		evtchnfd->deliver.port.vcpu_idx = -1;
+		evtchnfd->deliver.port.priority = data->u.evtchn.deliver.port.priority;
+	}
+
+	mutex_lock(&kvm->lock);
+	ret = idr_alloc(&kvm->arch.xen.evtchn_ports, evtchnfd, port, port + 1,
+			GFP_KERNEL);
+	mutex_unlock(&kvm->lock);
+	if (ret >= 0)
+		return 0;
+
+	if (ret == -ENOSPC)
+		ret = -EEXIST;
+out:
+	if (eventfd)
+		eventfd_ctx_put(eventfd);
+	kfree(evtchnfd);
+	return ret;
+}
+
+static int kvm_xen_eventfd_deassign(struct kvm *kvm, u32 port)
+{
+	struct evtchnfd *evtchnfd;
+
+	mutex_lock(&kvm->lock);
+	evtchnfd = idr_remove(&kvm->arch.xen.evtchn_ports, port);
+	mutex_unlock(&kvm->lock);
+
+	if (!evtchnfd)
+		return -ENOENT;
+
+	if (kvm)
+		synchronize_srcu(&kvm->srcu);
+	if (!evtchnfd->deliver.port.port)
+		eventfd_ctx_put(evtchnfd->deliver.eventfd.ctx);
+	kfree(evtchnfd);
+	return 0;
+}
+
+static int kvm_xen_eventfd_reset(struct kvm *kvm)
+{
+	struct evtchnfd *evtchnfd;
+	int i;
+
+	mutex_lock(&kvm->lock);
+	idr_for_each_entry(&kvm->arch.xen.evtchn_ports, evtchnfd, i) {
+		idr_remove(&kvm->arch.xen.evtchn_ports, evtchnfd->send_port);
+		synchronize_srcu(&kvm->srcu);
+		if (!evtchnfd->deliver.port.port)
+			eventfd_ctx_put(evtchnfd->deliver.eventfd.ctx);
+		kfree(evtchnfd);
+	}
+	mutex_unlock(&kvm->lock);
+
+	return 0;
+}
+
+static int kvm_xen_setattr_evtchn(struct kvm *kvm, struct kvm_xen_hvm_attr *data)
+{
+	u32 port = data->u.evtchn.send_port;
+
+	if (data->u.evtchn.flags == KVM_XEN_EVTCHN_RESET)
+		return kvm_xen_eventfd_reset(kvm);
+
+	if (!port || port >= max_evtchn_port(kvm))
+		return -EINVAL;
+
+	if (data->u.evtchn.flags == KVM_XEN_EVTCHN_DEASSIGN)
+		return kvm_xen_eventfd_deassign(kvm, port);
+	if (data->u.evtchn.flags == KVM_XEN_EVTCHN_UPDATE)
+		return kvm_xen_eventfd_update(kvm, data);
+	if (data->u.evtchn.flags)
+		return -EINVAL;
+
+	return kvm_xen_eventfd_assign(kvm, data);
+}
+
+static bool kvm_xen_hcall_evtchn_send(struct kvm_vcpu *vcpu, u64 param, u64 *r)
+{
+	struct evtchnfd *evtchnfd;
+	struct evtchn_send send;
+	gpa_t gpa;
+	int idx;
+
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
+	gpa = kvm_mmu_gva_to_gpa_system(vcpu, param, NULL);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
+
+	if (!gpa || kvm_vcpu_read_guest(vcpu, gpa, &send, sizeof(send))) {
+		*r = -EFAULT;
+		return true;
+	}
+
+	/* The evtchn_ports idr is protected by vcpu->kvm->srcu */
+	evtchnfd = idr_find(&vcpu->kvm->arch.xen.evtchn_ports, send.port);
+	if (!evtchnfd)
+		return false;
+
+	if (evtchnfd->deliver.port.port) {
+		int ret = kvm_xen_set_evtchn(&evtchnfd->deliver.port, vcpu->kvm);
+		if (ret < 0 && ret != -ENOTCONN)
+			return false;
+	} else {
+		eventfd_signal(evtchnfd->deliver.eventfd.ctx, 1);
+	}
+
+	*r = 0;
+	return true;
+}
+
+void kvm_xen_init_vcpu(struct kvm_vcpu *vcpu)
+{
+	vcpu->arch.xen.vcpu_id = vcpu->vcpu_idx;
+	vcpu->arch.xen.poll_evtchn = 0;
+	timer_setup(&vcpu->arch.xen.poll_timer, cancel_evtchn_poll, 0);
+}
+
+void kvm_xen_destroy_vcpu(struct kvm_vcpu *vcpu)
+{
+	if (kvm_xen_timer_enabled(vcpu))
+		kvm_xen_stop_timer(vcpu);
+
+	kvm_gfn_to_pfn_cache_destroy(vcpu->kvm,
+				     &vcpu->arch.xen.runstate_cache);
+	kvm_gfn_to_pfn_cache_destroy(vcpu->kvm,
+				     &vcpu->arch.xen.vcpu_info_cache);
+	kvm_gfn_to_pfn_cache_destroy(vcpu->kvm,
+				     &vcpu->arch.xen.vcpu_time_info_cache);
+	del_timer_sync(&vcpu->arch.xen.poll_timer);
+}
+
+void kvm_xen_init_vm(struct kvm *kvm)
+{
+	idr_init(&kvm->arch.xen.evtchn_ports);
+}
+
+void kvm_xen_destroy_vm(struct kvm *kvm)
+{
+	struct evtchnfd *evtchnfd;
+	int i;
+
+	kvm_gfn_to_pfn_cache_destroy(kvm, &kvm->arch.xen.shinfo_cache);
+
+	idr_for_each_entry(&kvm->arch.xen.evtchn_ports, evtchnfd, i) {
+		if (!evtchnfd->deliver.port.port)
+			eventfd_ctx_put(evtchnfd->deliver.eventfd.ctx);
+		kfree(evtchnfd);
+	}
+	idr_destroy(&kvm->arch.xen.evtchn_ports);
+
+	if (kvm->arch.xen_hvm_config.msr)
+		static_branch_slow_dec_deferred(&kvm_xen_enabled);
+}
diff --git a/arch/x86/kvm/xen.h b/arch/x86/kvm/xen.h
index adbcc9ed59db..ee5c4ae0755c 100644
--- a/arch/x86/kvm/xen.h
+++ b/arch/x86/kvm/xen.h
@@ -15,16 +15,19 @@
 extern struct static_key_false_deferred kvm_xen_enabled;
 
 int __kvm_xen_has_interrupt(struct kvm_vcpu *vcpu);
+void kvm_xen_inject_pending_events(struct kvm_vcpu *vcpu);
 int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data);
 int kvm_xen_vcpu_get_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data);
 int kvm_xen_hvm_set_attr(struct kvm *kvm, struct kvm_xen_hvm_attr *data);
 int kvm_xen_hvm_get_attr(struct kvm *kvm, struct kvm_xen_hvm_attr *data);
+int kvm_xen_hvm_evtchn_send(struct kvm *kvm, struct kvm_irq_routing_xen_evtchn *evt);
 int kvm_xen_write_hypercall_page(struct kvm_vcpu *vcpu, u64 data);
 int kvm_xen_hvm_config(struct kvm *kvm, struct kvm_xen_hvm_config *xhc);
 void kvm_xen_init_vm(struct kvm *kvm);
 void kvm_xen_destroy_vm(struct kvm *kvm);
-
-int kvm_xen_set_evtchn_fast(struct kvm_kernel_irq_routing_entry *e,
+void kvm_xen_init_vcpu(struct kvm_vcpu *vcpu);
+void kvm_xen_destroy_vcpu(struct kvm_vcpu *vcpu);
+int kvm_xen_set_evtchn_fast(struct kvm_xen_evtchn *xe,
 			    struct kvm *kvm);
 int kvm_xen_setup_evtchn(struct kvm *kvm,
 			 struct kvm_kernel_irq_routing_entry *e,
@@ -46,11 +49,33 @@ static inline bool kvm_xen_hypercall_enabled(struct kvm *kvm)
 static inline int kvm_xen_has_interrupt(struct kvm_vcpu *vcpu)
 {
 	if (static_branch_unlikely(&kvm_xen_enabled.key) &&
-	    vcpu->arch.xen.vcpu_info_set && vcpu->kvm->arch.xen.upcall_vector)
+	    vcpu->arch.xen.vcpu_info_cache.active &&
+	    vcpu->kvm->arch.xen.upcall_vector)
 		return __kvm_xen_has_interrupt(vcpu);
 
 	return 0;
 }
+
+static inline bool kvm_xen_has_pending_events(struct kvm_vcpu *vcpu)
+{
+	return static_branch_unlikely(&kvm_xen_enabled.key) &&
+		vcpu->arch.xen.evtchn_pending_sel;
+}
+
+static inline bool kvm_xen_timer_enabled(struct kvm_vcpu *vcpu)
+{
+	return !!vcpu->arch.xen.timer_virq;
+}
+
+static inline int kvm_xen_has_pending_timer(struct kvm_vcpu *vcpu)
+{
+	if (kvm_xen_hypercall_enabled(vcpu->kvm) && kvm_xen_timer_enabled(vcpu))
+		return atomic_read(&vcpu->arch.xen.timer_pending);
+
+	return 0;
+}
+
+void kvm_xen_inject_timer_irqs(struct kvm_vcpu *vcpu);
 #else
 static inline int kvm_xen_write_hypercall_page(struct kvm_vcpu *vcpu, u64 data)
 {
@@ -65,6 +90,14 @@ static inline void kvm_xen_destroy_vm(struct kvm *kvm)
 {
 }
 
+static inline void kvm_xen_init_vcpu(struct kvm_vcpu *vcpu)
+{
+}
+
+static inline void kvm_xen_destroy_vcpu(struct kvm_vcpu *vcpu)
+{
+}
+
 static inline bool kvm_xen_msr_enabled(struct kvm *kvm)
 {
 	return false;
@@ -79,6 +112,29 @@ static inline int kvm_xen_has_interrupt(struct kvm_vcpu *vcpu)
 {
 	return 0;
 }
+
+static inline void kvm_xen_inject_pending_events(struct kvm_vcpu *vcpu)
+{
+}
+
+static inline bool kvm_xen_has_pending_events(struct kvm_vcpu *vcpu)
+{
+	return false;
+}
+
+static inline int kvm_xen_has_pending_timer(struct kvm_vcpu *vcpu)
+{
+	return 0;
+}
+
+static inline void kvm_xen_inject_timer_irqs(struct kvm_vcpu *vcpu)
+{
+}
+
+static inline bool kvm_xen_timer_enabled(struct kvm_vcpu *vcpu)
+{
+	return false;
+}
 #endif
 
 int kvm_xen_hypercall(struct kvm_vcpu *vcpu);