Merge tag 'kvmarm-5.10' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm64 updates for Linux 5.10

- New page table code for both hypervisor and guest stage-2
- Introduction of a new EL2-private host context
- Allow EL2 to have its own private per-CPU variables
- Support of PMU event filtering
- Complete rework of the Spectre mitigation

30 files changed, 1943 insertions, 1909 deletions

diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
index 318c8f2df245..043756db8f6e 100644
--- a/arch/arm64/kvm/Kconfig
+++ b/arch/arm64/kvm/Kconfig
@@ -57,9 +57,6 @@ config KVM_ARM_PMU
 	  Adds support for a virtual Performance Monitoring Unit (PMU) in
 	  virtual machines.
 
-config KVM_INDIRECT_VECTORS
-	def_bool HARDEN_BRANCH_PREDICTOR || RANDOMIZE_BASE
-
 endif # KVM
 
 endif # VIRTUALIZATION
diff --git a/arch/arm64/kvm/Makefile b/arch/arm64/kvm/Makefile
index 99977c1972cc..1504c81fbf5d 100644
--- a/arch/arm64/kvm/Makefile
+++ b/arch/arm64/kvm/Makefile
@@ -13,7 +13,7 @@ obj-$(CONFIG_KVM) += hyp/
 kvm-y := $(KVM)/kvm_main.o $(KVM)/coalesced_mmio.o $(KVM)/eventfd.o \
 	 $(KVM)/vfio.o $(KVM)/irqchip.o \
 	 arm.o mmu.o mmio.o psci.o perf.o hypercalls.o pvtime.o \
-	 inject_fault.o regmap.o va_layout.o hyp.o handle_exit.o \
+	 inject_fault.o regmap.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 \
 	 aarch32.o arch_timer.o \
diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
index b588c3b5c2f0..f56122eedffc 100644
--- a/arch/arm64/kvm/arm.c
+++ b/arch/arm64/kvm/arm.c
@@ -46,8 +46,10 @@
 __asm__(".arch_extension	virt");
 #endif
 
-DEFINE_PER_CPU(kvm_host_data_t, kvm_host_data);
+DECLARE_KVM_HYP_PER_CPU(unsigned long, kvm_hyp_vector);
+
 static DEFINE_PER_CPU(unsigned long, kvm_arm_hyp_stack_page);
+unsigned long kvm_arm_hyp_percpu_base[NR_CPUS];
 
 /* The VMID used in the VTTBR */
 static atomic64_t kvm_vmid_gen = ATOMIC64_INIT(1);
@@ -145,6 +147,8 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
 {
 	int i;
 
+	bitmap_free(kvm->arch.pmu_filter);
+
 	kvm_vgic_destroy(kvm);
 
 	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
@@ -286,7 +290,7 @@ void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu)
 	if (vcpu->arch.has_run_once && unlikely(!irqchip_in_kernel(vcpu->kvm)))
 		static_branch_dec(&userspace_irqchip_in_use);
 
-	kvm_mmu_free_memory_caches(vcpu);
+	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_page_cache);
 	kvm_timer_vcpu_terminate(vcpu);
 	kvm_pmu_vcpu_destroy(vcpu);
 
@@ -1259,12 +1263,60 @@ long kvm_arch_vm_ioctl(struct file *filp,
 	}
 }
 
+static unsigned long nvhe_percpu_size(void)
+{
+	return (unsigned long)CHOOSE_NVHE_SYM(__per_cpu_end) -
+		(unsigned long)CHOOSE_NVHE_SYM(__per_cpu_start);
+}
+
+static unsigned long nvhe_percpu_order(void)
+{
+	unsigned long size = nvhe_percpu_size();
+
+	return size ? get_order(size) : 0;
+}
+
+static int kvm_map_vectors(void)
+{
+	/*
+	 * SV2  = ARM64_SPECTRE_V2
+	 * HEL2 = ARM64_HARDEN_EL2_VECTORS
+	 *
+	 * !SV2 + !HEL2 -> use direct vectors
+	 *  SV2 + !HEL2 -> use hardened vectors in place
+	 * !SV2 +  HEL2 -> allocate one vector slot and use exec mapping
+	 *  SV2 +  HEL2 -> use hardened vectors and use exec mapping
+	 */
+	if (cpus_have_const_cap(ARM64_SPECTRE_V2)) {
+		__kvm_bp_vect_base = kvm_ksym_ref(__bp_harden_hyp_vecs);
+		__kvm_bp_vect_base = kern_hyp_va(__kvm_bp_vect_base);
+	}
+
+	if (cpus_have_const_cap(ARM64_HARDEN_EL2_VECTORS)) {
+		phys_addr_t vect_pa = __pa_symbol(__bp_harden_hyp_vecs);
+		unsigned long size = __BP_HARDEN_HYP_VECS_SZ;
+
+		/*
+		 * Always allocate a spare vector slot, as we don't
+		 * know yet which CPUs have a BP hardening slot that
+		 * we can reuse.
+		 */
+		__kvm_harden_el2_vector_slot = atomic_inc_return(&arm64_el2_vector_last_slot);
+		BUG_ON(__kvm_harden_el2_vector_slot >= BP_HARDEN_EL2_SLOTS);
+		return create_hyp_exec_mappings(vect_pa, size,
+						&__kvm_bp_vect_base);
+	}
+
+	return 0;
+}
+
 static void cpu_init_hyp_mode(void)
 {
 	phys_addr_t pgd_ptr;
 	unsigned long hyp_stack_ptr;
 	unsigned long vector_ptr;
 	unsigned long tpidr_el2;
+	struct arm_smccc_res res;
 
 	/* Switch from the HYP stub to our own HYP init vector */
 	__hyp_set_vectors(kvm_get_idmap_vector());
@@ -1274,12 +1326,13 @@ static void cpu_init_hyp_mode(void)
 	 * kernel's mapping to the linear mapping, and store it in tpidr_el2
 	 * so that we can use adr_l to access per-cpu variables in EL2.
 	 */
-	tpidr_el2 = ((unsigned long)this_cpu_ptr(&kvm_host_data) -
-		     (unsigned long)kvm_ksym_ref(&kvm_host_data));
+	tpidr_el2 = (unsigned long)this_cpu_ptr_nvhe_sym(__per_cpu_start) -
+		    (unsigned long)kvm_ksym_ref(CHOOSE_NVHE_SYM(__per_cpu_start));
 
 	pgd_ptr = kvm_mmu_get_httbr();
 	hyp_stack_ptr = __this_cpu_read(kvm_arm_hyp_stack_page) + PAGE_SIZE;
-	vector_ptr = (unsigned long)kvm_get_hyp_vector();
+	hyp_stack_ptr = kern_hyp_va(hyp_stack_ptr);
+	vector_ptr = (unsigned long)kern_hyp_va(kvm_ksym_ref(__kvm_hyp_host_vector));
 
 	/*
 	 * Call initialization code, and switch to the full blown HYP code.
@@ -1288,14 +1341,16 @@ static void cpu_init_hyp_mode(void)
 	 * cpus_have_const_cap() wrapper.
 	 */
 	BUG_ON(!system_capabilities_finalized());
-	__kvm_call_hyp((void *)pgd_ptr, hyp_stack_ptr, vector_ptr, tpidr_el2);
+	arm_smccc_1_1_hvc(KVM_HOST_SMCCC_FUNC(__kvm_hyp_init),
+			  pgd_ptr, tpidr_el2, hyp_stack_ptr, vector_ptr, &res);
+	WARN_ON(res.a0 != SMCCC_RET_SUCCESS);
 
 	/*
 	 * Disabling SSBD on a non-VHE system requires us to enable SSBS
 	 * at EL2.
 	 */
 	if (this_cpu_has_cap(ARM64_SSBS) &&
-	    arm64_get_ssbd_state() == ARM64_SSBD_FORCE_DISABLE) {
+	    arm64_get_spectre_v4_state() == SPECTRE_VULNERABLE) {
 		kvm_call_hyp_nvhe(__kvm_enable_ssbs);
 	}
 }
@@ -1308,10 +1363,12 @@ static void cpu_hyp_reset(void)
 
 static void cpu_hyp_reinit(void)
 {
-	kvm_init_host_cpu_context(&this_cpu_ptr(&kvm_host_data)->host_ctxt);
+	kvm_init_host_cpu_context(&this_cpu_ptr_hyp_sym(kvm_host_data)->host_ctxt);
 
 	cpu_hyp_reset();
 
+	*this_cpu_ptr_hyp_sym(kvm_hyp_vector) = (unsigned long)kvm_get_hyp_vector();
+
 	if (is_kernel_in_hyp_mode())
 		kvm_timer_init_vhe();
 	else
@@ -1462,8 +1519,10 @@ static void teardown_hyp_mode(void)
 	int cpu;
 
 	free_hyp_pgds();
-	for_each_possible_cpu(cpu)
+	for_each_possible_cpu(cpu) {
 		free_page(per_cpu(kvm_arm_hyp_stack_page, cpu));
+		free_pages(kvm_arm_hyp_percpu_base[cpu], nvhe_percpu_order());
+	}
 }
 
 /**
@@ -1497,6 +1556,24 @@ static int init_hyp_mode(void)
 	}
 
 	/*
+	 * Allocate and initialize pages for Hypervisor-mode percpu regions.
+	 */
+	for_each_possible_cpu(cpu) {
+		struct page *page;
+		void *page_addr;
+
+		page = alloc_pages(GFP_KERNEL, nvhe_percpu_order());
+		if (!page) {
+			err = -ENOMEM;
+			goto out_err;
+		}
+
+		page_addr = page_address(page);
+		memcpy(page_addr, CHOOSE_NVHE_SYM(__per_cpu_start), nvhe_percpu_size());
+		kvm_arm_hyp_percpu_base[cpu] = (unsigned long)page_addr;
+	}
+
+	/*
 	 * Map the Hyp-code called directly from the host
 	 */
 	err = create_hyp_mappings(kvm_ksym_ref(__hyp_text_start),
@@ -1540,22 +1617,21 @@ static int init_hyp_mode(void)
 		}
 	}
 
+	/*
+	 * Map Hyp percpu pages
+	 */
 	for_each_possible_cpu(cpu) {
-		kvm_host_data_t *cpu_data;
+		char *percpu_begin = (char *)kvm_arm_hyp_percpu_base[cpu];
+		char *percpu_end = percpu_begin + nvhe_percpu_size();
 
-		cpu_data = per_cpu_ptr(&kvm_host_data, cpu);
-		err = create_hyp_mappings(cpu_data, cpu_data + 1, PAGE_HYP);
+		err = create_hyp_mappings(percpu_begin, percpu_end, PAGE_HYP);
 
 		if (err) {
-			kvm_err("Cannot map host CPU state: %d\n", err);
+			kvm_err("Cannot map hyp percpu region\n");
 			goto out_err;
 		}
 	}
 
-	err = hyp_map_aux_data();
-	if (err)
-		kvm_err("Cannot map host auxiliary data: %d\n", err);
-
 	return 0;
 
 out_err:
diff --git a/arch/arm64/kvm/hyp.S b/arch/arm64/kvm/hyp.S
deleted file mode 100644
index 3c79a1124af2..000000000000
--- a/arch/arm64/kvm/hyp.S
+++ /dev/null
@@ -1,34 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0-only */
-/*
- * Copyright (C) 2012,2013 - ARM Ltd
- * Author: Marc Zyngier <marc.zyngier@arm.com>
- */
-
-#include <linux/linkage.h>
-
-#include <asm/alternative.h>
-#include <asm/assembler.h>
-#include <asm/cpufeature.h>
-
-/*
- * u64 __kvm_call_hyp(void *hypfn, ...);
- *
- * This is not really a variadic function in the classic C-way and care must
- * be taken when calling this to ensure parameters are passed in registers
- * only, since the stack will change between the caller and the callee.
- *
- * Call the function with the first argument containing a pointer to the
- * function you wish to call in Hyp mode, and subsequent arguments will be
- * passed as x0, x1, and x2 (a maximum of 3 arguments in addition to the
- * function pointer can be passed).  The function being called must be mapped
- * in Hyp mode (see init_hyp_mode in arch/arm/kvm/arm.c).  Return values are
- * passed in x0.
- *
- * A function pointer with a value less than 0xfff has a special meaning,
- * and is used to implement hyp stubs in the same way as in
- * arch/arm64/kernel/hyp_stub.S.
- */
-SYM_FUNC_START(__kvm_call_hyp)
-	hvc	#0
-	ret
-SYM_FUNC_END(__kvm_call_hyp)
diff --git a/arch/arm64/kvm/hyp/Makefile b/arch/arm64/kvm/hyp/Makefile
index f54f0e89a71c..4a81eddabcd8 100644
--- a/arch/arm64/kvm/hyp/Makefile
+++ b/arch/arm64/kvm/hyp/Makefile
@@ -10,5 +10,4 @@ subdir-ccflags-y := -I$(incdir)				\
 		    -DDISABLE_BRANCH_PROFILING		\
 		    $(DISABLE_STACKLEAK_PLUGIN)
 
-obj-$(CONFIG_KVM) += vhe/ nvhe/
-obj-$(CONFIG_KVM_INDIRECT_VECTORS) += smccc_wa.o
+obj-$(CONFIG_KVM) += vhe/ nvhe/ pgtable.o smccc_wa.o
diff --git a/arch/arm64/kvm/hyp/entry.S b/arch/arm64/kvm/hyp/entry.S
index 76e7eaf4675e..b0afad7a99c6 100644
--- a/arch/arm64/kvm/hyp/entry.S
+++ b/arch/arm64/kvm/hyp/entry.S
@@ -7,7 +7,6 @@
 #include <linux/linkage.h>
 
 #include <asm/alternative.h>
-#include <asm/asm-offsets.h>
 #include <asm/assembler.h>
 #include <asm/fpsimdmacros.h>
 #include <asm/kvm.h>
@@ -16,66 +15,28 @@
 #include <asm/kvm_mmu.h>
 #include <asm/kvm_ptrauth.h>
 
-#define CPU_XREG_OFFSET(x)	(CPU_USER_PT_REGS + 8*x)
-#define CPU_SP_EL0_OFFSET	(CPU_XREG_OFFSET(30) + 8)
-
 	.text
 
 /*
- * We treat x18 as callee-saved as the host may use it as a platform
- * register (e.g. for shadow call stack).
- */
-.macro save_callee_saved_regs ctxt
-	str	x18,      [\ctxt, #CPU_XREG_OFFSET(18)]
-	stp	x19, x20, [\ctxt, #CPU_XREG_OFFSET(19)]
-	stp	x21, x22, [\ctxt, #CPU_XREG_OFFSET(21)]
-	stp	x23, x24, [\ctxt, #CPU_XREG_OFFSET(23)]
-	stp	x25, x26, [\ctxt, #CPU_XREG_OFFSET(25)]
-	stp	x27, x28, [\ctxt, #CPU_XREG_OFFSET(27)]
-	stp	x29, lr,  [\ctxt, #CPU_XREG_OFFSET(29)]
-.endm
-
-.macro restore_callee_saved_regs ctxt
-	// We require \ctxt is not x18-x28
-	ldr	x18,      [\ctxt, #CPU_XREG_OFFSET(18)]
-	ldp	x19, x20, [\ctxt, #CPU_XREG_OFFSET(19)]
-	ldp	x21, x22, [\ctxt, #CPU_XREG_OFFSET(21)]
-	ldp	x23, x24, [\ctxt, #CPU_XREG_OFFSET(23)]
-	ldp	x25, x26, [\ctxt, #CPU_XREG_OFFSET(25)]
-	ldp	x27, x28, [\ctxt, #CPU_XREG_OFFSET(27)]
-	ldp	x29, lr,  [\ctxt, #CPU_XREG_OFFSET(29)]
-.endm
-
-.macro save_sp_el0 ctxt, tmp
-	mrs	\tmp,	sp_el0
-	str	\tmp,	[\ctxt, #CPU_SP_EL0_OFFSET]
-.endm
-
-.macro restore_sp_el0 ctxt, tmp
-	ldr	\tmp,	  [\ctxt, #CPU_SP_EL0_OFFSET]
-	msr	sp_el0, \tmp
-.endm
-
-/*
- * u64 __guest_enter(struct kvm_vcpu *vcpu,
- *		     struct kvm_cpu_context *host_ctxt);
+ * u64 __guest_enter(struct kvm_vcpu *vcpu);
  */
 SYM_FUNC_START(__guest_enter)
 	// x0: vcpu
-	// x1: host context
-	// x2-x17: clobbered by macros
+	// x1-x17: clobbered by macros
 	// x29: guest context
 
-	// Store the host regs
+	adr_this_cpu x1, kvm_hyp_ctxt, x2
+
+	// Store the hyp regs
 	save_callee_saved_regs x1
 
-	// Save the host's sp_el0
+	// Save hyp's sp_el0
 	save_sp_el0	x1, x2
 
-	// Now the host state is stored if we have a pending RAS SError it must
-	// affect the host. If any asynchronous exception is pending we defer
-	// the guest entry. The DSB isn't necessary before v8.2 as any SError
-	// would be fatal.
+	// Now the hyp state is stored if we have a pending RAS SError it must
+	// affect the host or hyp. If any asynchronous exception is pending we
+	// defer the guest entry. The DSB isn't necessary before v8.2 as any
+	// SError would be fatal.
 alternative_if ARM64_HAS_RAS_EXTN
 	dsb	nshst
 	isb
@@ -86,6 +47,8 @@ alternative_else_nop_endif
 	ret
 
 1:
+	set_loaded_vcpu x0, x1, x2
+
 	add	x29, x0, #VCPU_CONTEXT
 
 	// Macro ptrauth_switch_to_guest format:
@@ -116,6 +79,26 @@ alternative_else_nop_endif
 	eret
 	sb
 
+SYM_INNER_LABEL(__guest_exit_panic, SYM_L_GLOBAL)
+	// x2-x29,lr: vcpu regs
+	// vcpu x0-x1 on the stack
+
+	// If the hyp context is loaded, go straight to hyp_panic
+	get_loaded_vcpu x0, x1
+	cbz	x0, hyp_panic
+
+	// The hyp context is saved so make sure it is restored to allow
+	// hyp_panic to run at hyp and, subsequently, panic to run in the host.
+	// This makes use of __guest_exit to avoid duplication but sets the
+	// return address to tail call into hyp_panic. As a side effect, the
+	// current state is saved to the guest context but it will only be
+	// accurate if the guest had been completely restored.
+	adr_this_cpu x0, kvm_hyp_ctxt, x1
+	adr	x1, hyp_panic
+	str	x1, [x0, #CPU_XREG_OFFSET(30)]
+
+	get_vcpu_ptr	x1, x0
+
 SYM_INNER_LABEL(__guest_exit, SYM_L_GLOBAL)
 	// x0: return code
 	// x1: vcpu
@@ -148,21 +131,23 @@ SYM_INNER_LABEL(__guest_exit, SYM_L_GLOBAL)
 	// Store the guest's sp_el0
 	save_sp_el0	x1, x2
 
-	get_host_ctxt	x2, x3
+	adr_this_cpu x2, kvm_hyp_ctxt, x3
 
-	// Macro ptrauth_switch_to_guest format:
-	// 	ptrauth_switch_to_host(guest cxt, host cxt, tmp1, tmp2, tmp3)
+	// Macro ptrauth_switch_to_hyp format:
+	// 	ptrauth_switch_to_hyp(guest cxt, host cxt, tmp1, tmp2, tmp3)
 	// The below macro to save/restore keys is not implemented in C code
 	// as it may cause Pointer Authentication key signing mismatch errors
 	// when this feature is enabled for kernel code.
-	ptrauth_switch_to_host x1, x2, x3, x4, x5
+	ptrauth_switch_to_hyp x1, x2, x3, x4, x5
 
-	// Restore the hosts's sp_el0
+	// Restore hyp's sp_el0
 	restore_sp_el0 x2, x3
 
-	// Now restore the host regs
+	// Now restore the hyp regs
 	restore_callee_saved_regs x2
 
+	set_loaded_vcpu xzr, x1, x2
+
 alternative_if ARM64_HAS_RAS_EXTN
 	// If we have the RAS extensions we can consume a pending error
 	// without an unmask-SError and isb. The ESB-instruction consumed any
diff --git a/arch/arm64/kvm/hyp/hyp-entry.S b/arch/arm64/kvm/hyp/hyp-entry.S
index 46b4dab933d0..0a5b36eb54b3 100644
--- a/arch/arm64/kvm/hyp/hyp-entry.S
+++ b/arch/arm64/kvm/hyp/hyp-entry.S
@@ -12,7 +12,6 @@
 #include <asm/cpufeature.h>
 #include <asm/kvm_arm.h>
 #include <asm/kvm_asm.h>
-#include <asm/kvm_mmu.h>
 #include <asm/mmu.h>
 
 .macro save_caller_saved_regs_vect
@@ -41,20 +40,6 @@
 
 	.text
 
-.macro do_el2_call
-	/*
-	 * Shuffle the parameters before calling the function
-	 * pointed to in x0. Assumes parameters in x[1,2,3].
-	 */
-	str	lr, [sp, #-16]!
-	mov	lr, x0
-	mov	x0, x1
-	mov	x1, x2
-	mov	x2, x3
-	blr	lr
-	ldr	lr, [sp], #16
-.endm
-
 el1_sync:				// Guest trapped into EL2
 
 	mrs	x0, esr_el2
@@ -63,44 +48,6 @@ el1_sync:				// Guest trapped into EL2
 	ccmp	x0, #ESR_ELx_EC_HVC32, #4, ne
 	b.ne	el1_trap
 
-#ifdef __KVM_NVHE_HYPERVISOR__
-	mrs	x1, vttbr_el2		// If vttbr is valid, the guest
-	cbnz	x1, el1_hvc_guest	// called HVC
-
-	/* Here, we're pretty sure the host called HVC. */
-	ldp	x0, x1, [sp], #16
-
-	/* Check for a stub HVC call */
-	cmp	x0, #HVC_STUB_HCALL_NR
-	b.hs	1f
-
-	/*
-	 * 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).
-	 *
-	 * Preserve x0-x4, which may contain stub parameters.
-	 */
-	ldr	x5, =__kvm_handle_stub_hvc
-	ldr_l	x6, kimage_voffset
-
-	/* x5 = __pa(x5) */
-	sub	x5, x5, x6
-	br	x5
-
-1:
-	/*
-	 * Perform the EL2 call
-	 */
-	kern_hyp_va	x0
-	do_el2_call
-
-	eret
-	sb
-#endif /* __KVM_NVHE_HYPERVISOR__ */
-
-el1_hvc_guest:
 	/*
 	 * Fastest possible path for ARM_SMCCC_ARCH_WORKAROUND_1.
 	 * The workaround has already been applied on the host,
@@ -116,35 +63,6 @@ el1_hvc_guest:
 			  ARM_SMCCC_ARCH_WORKAROUND_2)
 	cbnz	w1, el1_trap
 
-#ifdef CONFIG_ARM64_SSBD
-alternative_cb	arm64_enable_wa2_handling
-	b	wa2_end
-alternative_cb_end
-	get_vcpu_ptr	x2, x0
-	ldr	x0, [x2, #VCPU_WORKAROUND_FLAGS]
-
-	// Sanitize the argument and update the guest flags
-	ldr	x1, [sp, #8]			// Guest's x1
-	clz	w1, w1				// Murphy's device:
-	lsr	w1, w1, #5			// w1 = !!w1 without using
-	eor	w1, w1, #1			// the flags...
-	bfi	x0, x1, #VCPU_WORKAROUND_2_FLAG_SHIFT, #1
-	str	x0, [x2, #VCPU_WORKAROUND_FLAGS]
-
-	/* Check that we actually need to perform the call */
-	hyp_ldr_this_cpu x0, arm64_ssbd_callback_required, x2
-	cbz	x0, wa2_end
-
-	mov	w0, #ARM_SMCCC_ARCH_WORKAROUND_2
-	smc	#0
-
-	/* Don't leak data from the SMC call */
-	mov	x3, xzr
-wa2_end:
-	mov	x2, xzr
-	mov	x1, xzr
-#endif
-
 wa_epilogue:
 	mov	x0, xzr
 	add	sp, sp, #16
@@ -198,24 +116,7 @@ el2_error:
 	eret
 	sb
 
-#ifdef __KVM_NVHE_HYPERVISOR__
-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, =panic
-	msr	elr_el2, lr
-	eret
-	sb
-SYM_FUNC_END(__hyp_do_panic)
-#endif
-
-SYM_CODE_START(__hyp_panic)
-	get_host_ctxt x0, x1
-	b	hyp_panic
-SYM_CODE_END(__hyp_panic)
-
-.macro invalid_vector	label, target = __hyp_panic
+.macro invalid_vector	label, target = __guest_exit_panic
 	.align	2
 SYM_CODE_START(\label)
 	b \target
@@ -227,7 +128,6 @@ SYM_CODE_END(\label)
 	invalid_vector	el2t_irq_invalid
 	invalid_vector	el2t_fiq_invalid
 	invalid_vector	el2t_error_invalid
-	invalid_vector	el2h_sync_invalid
 	invalid_vector	el2h_irq_invalid
 	invalid_vector	el2h_fiq_invalid
 	invalid_vector	el1_fiq_invalid
@@ -257,10 +157,9 @@ check_preamble_length 661b, 662b
 .macro invalid_vect target
 	.align 7
 661:
-	b	\target
 	nop
+	stp	x0, x1, [sp, #-16]!
 662:
-	ldp	x0, x1, [sp], #16
 	b	\target
 
 check_preamble_length 661b, 662b
@@ -288,7 +187,6 @@ SYM_CODE_START(__kvm_hyp_vector)
 	valid_vect	el1_error		// Error 32-bit EL1
 SYM_CODE_END(__kvm_hyp_vector)
 
-#ifdef CONFIG_KVM_INDIRECT_VECTORS
 .macro hyp_ventry
 	.align 7
 1:	esb
@@ -338,4 +236,3 @@ SYM_CODE_START(__bp_harden_hyp_vecs)
 1:	.org __bp_harden_hyp_vecs + __BP_HARDEN_HYP_VECS_SZ
 	.org 1b
 SYM_CODE_END(__bp_harden_hyp_vecs)
-#endif
diff --git a/arch/arm64/kvm/hyp/include/hyp/debug-sr.h b/arch/arm64/kvm/hyp/include/hyp/debug-sr.h
index 5e28ea6aa097..4ebe9f558f3a 100644
--- a/arch/arm64/kvm/hyp/include/hyp/debug-sr.h
+++ b/arch/arm64/kvm/hyp/include/hyp/debug-sr.h
@@ -135,7 +135,7 @@ static inline void __debug_switch_to_guest_common(struct kvm_vcpu *vcpu)
 	if (!(vcpu->arch.flags & KVM_ARM64_DEBUG_DIRTY))
 		return;
 
-	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
 	guest_ctxt = &vcpu->arch.ctxt;
 	host_dbg = &vcpu->arch.host_debug_state.regs;
 	guest_dbg = kern_hyp_va(vcpu->arch.debug_ptr);
@@ -154,7 +154,7 @@ static inline void __debug_switch_to_host_common(struct kvm_vcpu *vcpu)
 	if (!(vcpu->arch.flags & KVM_ARM64_DEBUG_DIRTY))
 		return;
 
-	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
 	guest_ctxt = &vcpu->arch.ctxt;
 	host_dbg = &vcpu->arch.host_debug_state.regs;
 	guest_dbg = kern_hyp_va(vcpu->arch.debug_ptr);
diff --git a/arch/arm64/kvm/hyp/include/hyp/switch.h b/arch/arm64/kvm/hyp/include/hyp/switch.h
index 0261308bf944..313a8fa3c721 100644
--- a/arch/arm64/kvm/hyp/include/hyp/switch.h
+++ b/arch/arm64/kvm/hyp/include/hyp/switch.h
@@ -126,11 +126,6 @@ static inline void ___deactivate_traps(struct kvm_vcpu *vcpu)
 	}
 }
 
-static inline void __activate_vm(struct kvm_s2_mmu *mmu)
-{
-	__load_guest_stage2(mmu);
-}
-
 static inline bool __translate_far_to_hpfar(u64 far, u64 *hpfar)
 {
 	u64 par, tmp;
@@ -377,6 +372,8 @@ static inline bool esr_is_ptrauth_trap(u32 esr)
 	ctxt_sys_reg(ctxt, key ## KEYHI_EL1) = __val;                   \
 } while(0)
 
+DECLARE_PER_CPU(struct kvm_cpu_context, kvm_hyp_ctxt);
+
 static inline bool __hyp_handle_ptrauth(struct kvm_vcpu *vcpu)
 {
 	struct kvm_cpu_context *ctxt;
@@ -386,7 +383,7 @@ static inline bool __hyp_handle_ptrauth(struct kvm_vcpu *vcpu)
 	    !esr_is_ptrauth_trap(kvm_vcpu_get_esr(vcpu)))
 		return false;
 
-	ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+	ctxt = this_cpu_ptr(&kvm_hyp_ctxt);
 	__ptrauth_save_key(ctxt, APIA);
 	__ptrauth_save_key(ctxt, APIB);
 	__ptrauth_save_key(ctxt, APDA);
@@ -479,49 +476,15 @@ exit:
 	return false;
 }
 
-static inline bool __needs_ssbd_off(struct kvm_vcpu *vcpu)
-{
-	if (!cpus_have_final_cap(ARM64_SSBD))
-		return false;
-
-	return !(vcpu->arch.workaround_flags & VCPU_WORKAROUND_2_FLAG);
-}
-
-static inline void __set_guest_arch_workaround_state(struct kvm_vcpu *vcpu)
-{
-#ifdef CONFIG_ARM64_SSBD
-	/*
-	 * The host runs with the workaround always present. If the
-	 * guest wants it disabled, so be it...
-	 */
-	if (__needs_ssbd_off(vcpu) &&
-	    __hyp_this_cpu_read(arm64_ssbd_callback_required))
-		arm_smccc_1_1_smc(ARM_SMCCC_ARCH_WORKAROUND_2, 0, NULL);
-#endif
-}
-
-static inline void __set_host_arch_workaround_state(struct kvm_vcpu *vcpu)
-{
-#ifdef CONFIG_ARM64_SSBD
-	/*
-	 * If the guest has disabled the workaround, bring it back on.
-	 */
-	if (__needs_ssbd_off(vcpu) &&
-	    __hyp_this_cpu_read(arm64_ssbd_callback_required))
-		arm_smccc_1_1_smc(ARM_SMCCC_ARCH_WORKAROUND_2, 1, NULL);
-#endif
-}
-
 static inline void __kvm_unexpected_el2_exception(void)
 {
+	extern char __guest_exit_panic[];
 	unsigned long addr, fixup;
-	struct kvm_cpu_context *host_ctxt;
 	struct exception_table_entry *entry, *end;
 	unsigned long elr_el2 = read_sysreg(elr_el2);
 
 	entry = hyp_symbol_addr(__start___kvm_ex_table);
 	end = hyp_symbol_addr(__stop___kvm_ex_table);
-	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
 
 	while (entry < end) {
 		addr = (unsigned long)&entry->insn + entry->insn;
@@ -536,7 +499,8 @@ static inline void __kvm_unexpected_el2_exception(void)
 		return;
 	}
 
-	hyp_panic(host_ctxt);
+	/* Trigger a panic after restoring the hyp context. */
+	write_sysreg(__guest_exit_panic, elr_el2);
 }
 
 #endif /* __ARM64_KVM_HYP_SWITCH_H__ */
diff --git a/arch/arm64/kvm/hyp/nvhe/.gitignore b/arch/arm64/kvm/hyp/nvhe/.gitignore
new file mode 100644
index 000000000000..695d73d0249e
--- /dev/null
+++ b/arch/arm64/kvm/hyp/nvhe/.gitignore
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+hyp.lds
diff --git a/arch/arm64/kvm/hyp/nvhe/Makefile b/arch/arm64/kvm/hyp/nvhe/Makefile
index aef76487edc2..ddde15fe85f2 100644
--- a/arch/arm64/kvm/hyp/nvhe/Makefile
+++ b/arch/arm64/kvm/hyp/nvhe/Makefile
@@ -6,44 +6,50 @@
 asflags-y := -D__KVM_NVHE_HYPERVISOR__
 ccflags-y := -D__KVM_NVHE_HYPERVISOR__
 
-obj-y := timer-sr.o sysreg-sr.o debug-sr.o switch.o tlb.o hyp-init.o
+obj-y := timer-sr.o sysreg-sr.o debug-sr.o switch.o tlb.o hyp-init.o host.o hyp-main.o
 obj-y += ../vgic-v3-sr.o ../aarch32.o ../vgic-v2-cpuif-proxy.o ../entry.o \
 	 ../fpsimd.o ../hyp-entry.o
 
-obj-y := $(patsubst %.o,%.hyp.o,$(obj-y))
-extra-y := $(patsubst %.hyp.o,%.hyp.tmp.o,$(obj-y))
+##
+## Build rules for compiling nVHE hyp code
+## Output of this folder is `kvm_nvhe.o`, a partially linked object
+## file containing all nVHE hyp code and data.
+##
 
-$(obj)/%.hyp.tmp.o: $(src)/%.c FORCE
+hyp-obj := $(patsubst %.o,%.nvhe.o,$(obj-y))
+obj-y := kvm_nvhe.o
+extra-y := $(hyp-obj) kvm_nvhe.tmp.o hyp.lds
+
+# 1) Compile all source files to `.nvhe.o` object files. The file extension
+#    avoids file name clashes for files shared with VHE.
+$(obj)/%.nvhe.o: $(src)/%.c FORCE
 	$(call if_changed_rule,cc_o_c)
-$(obj)/%.hyp.tmp.o: $(src)/%.S FORCE
+$(obj)/%.nvhe.o: $(src)/%.S FORCE
 	$(call if_changed_rule,as_o_S)
-$(obj)/%.hyp.o: $(obj)/%.hyp.tmp.o FORCE
-	$(call if_changed,hypcopy)
 
-# Disable reordering functions by GCC (enabled at -O2).
-# This pass puts functions into '.text.*' sections to aid the linker
-# in optimizing ELF layout. See HYPCOPY comment below for more info.
-ccflags-y += $(call cc-option,-fno-reorder-functions)
+# 2) Compile linker script.
+$(obj)/hyp.lds: $(src)/hyp.lds.S FORCE
+	$(call if_changed_dep,cpp_lds_S)
+
+# 3) Partially link all '.nvhe.o' files and apply the linker script.
+#    Prefixes names of ELF sections with '.hyp', eg. '.hyp.text'.
+#    Note: The following rule assumes that the 'ld' rule puts LDFLAGS before
+#          the list of dependencies to form '-T $(obj)/hyp.lds'. This is to
+#          keep the dependency on the target while avoiding an error from
+#          GNU ld if the linker script is passed to it twice.
+LDFLAGS_kvm_nvhe.tmp.o := -r -T
+$(obj)/kvm_nvhe.tmp.o: $(obj)/hyp.lds $(addprefix $(obj)/,$(hyp-obj)) FORCE
+	$(call if_changed,ld)
+
+# 4) Produce the final 'kvm_nvhe.o', ready to be linked into 'vmlinux'.
+#    Prefixes names of ELF symbols with '__kvm_nvhe_'.
+$(obj)/kvm_nvhe.o: $(obj)/kvm_nvhe.tmp.o FORCE
+	$(call if_changed,hypcopy)
 
 # The HYPCOPY command uses `objcopy` to prefix all ELF symbol names
-# and relevant ELF section names to avoid clashes with VHE code/data.
-#
-# Hyp code is assumed to be in the '.text' section of the input object
-# files (with the exception of specialized sections such as
-# '.hyp.idmap.text'). This assumption may be broken by a compiler that
-# divides code into sections like '.text.unlikely' so as to optimize
-# ELF layout. HYPCOPY checks that no such sections exist in the input
-# using `objdump`, otherwise they would be linked together with other
-# kernel code and not memory-mapped correctly at runtime.
+# to avoid clashes with VHE code/data.
 quiet_cmd_hypcopy = HYPCOPY $@
-      cmd_hypcopy =							\
-	if $(OBJDUMP) -h $< | grep -F '.text.'; then			\
-		echo "$@: function reordering not supported in nVHE hyp code" >&2; \
-		/bin/false;						\
-	fi;								\
-	$(OBJCOPY) --prefix-symbols=__kvm_nvhe_				\
-		   --rename-section=.text=.hyp.text			\
-		   $< $@
+      cmd_hypcopy = $(OBJCOPY) --prefix-symbols=__kvm_nvhe_ $< $@
 
 # Remove ftrace and Shadow Call Stack CFLAGS.
 # This is equivalent to the 'notrace' and '__noscs' annotations.
diff --git a/arch/arm64/kvm/hyp/nvhe/host.S b/arch/arm64/kvm/hyp/nvhe/host.S
new file mode 100644
index 000000000000..ff9a0f547b9f
--- /dev/null
+++ b/arch/arm64/kvm/hyp/nvhe/host.S
@@ -0,0 +1,187 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2020 - Google Inc
+ * Author: Andrew Scull <ascull@google.com>
+ */
+
+#include <linux/linkage.h>
+
+#include <asm/assembler.h>
+#include <asm/kvm_asm.h>
+#include <asm/kvm_mmu.h>
+
+	.text
+
+SYM_FUNC_START(__host_exit)
+	stp	x0, x1, [sp, #-16]!
+
+	get_host_ctxt	x0, x1
+
+	ALTERNATIVE(nop, SET_PSTATE_PAN(1), ARM64_HAS_PAN, CONFIG_ARM64_PAN)
+
+	/* Store the host regs x2 and x3 */
+	stp	x2, x3,   [x0, #CPU_XREG_OFFSET(2)]
+
+	/* Retrieve the host regs x0-x1 from the stack */
+	ldp	x2, x3, [sp], #16	// x0, x1
+
+	/* Store the host regs x0-x1 and x4-x17 */
+	stp	x2, x3,   [x0, #CPU_XREG_OFFSET(0)]
+	stp	x4, x5,   [x0, #CPU_XREG_OFFSET(4)]
+	stp	x6, x7,   [x0, #CPU_XREG_OFFSET(6)]
+	stp	x8, x9,   [x0, #CPU_XREG_OFFSET(8)]
+	stp	x10, x11, [x0, #CPU_XREG_OFFSET(10)]
+	stp	x12, x13, [x0, #CPU_XREG_OFFSET(12)]
+	stp	x14, x15, [x0, #CPU_XREG_OFFSET(14)]
+	stp	x16, x17, [x0, #CPU_XREG_OFFSET(16)]
+
+	/* Store the host regs x18-x29, lr */
+	save_callee_saved_regs x0
+
+	/* Save the host context pointer in x29 across the function call */
+	mov	x29, x0
+	bl	handle_trap
+
+	/* Restore host regs x0-x17 */
+	ldp	x0, x1,   [x29, #CPU_XREG_OFFSET(0)]
+	ldp	x2, x3,   [x29, #CPU_XREG_OFFSET(2)]
+	ldp	x4, x5,   [x29, #CPU_XREG_OFFSET(4)]
+	ldp	x6, x7,   [x29, #CPU_XREG_OFFSET(6)]
+
+	/* x0-7 are use for panic arguments */
+__host_enter_for_panic:
+	ldp	x8, x9,   [x29, #CPU_XREG_OFFSET(8)]
+	ldp	x10, x11, [x29, #CPU_XREG_OFFSET(10)]
+	ldp	x12, x13, [x29, #CPU_XREG_OFFSET(12)]
+	ldp	x14, x15, [x29, #CPU_XREG_OFFSET(14)]
+	ldp	x16, x17, [x29, #CPU_XREG_OFFSET(16)]
+
+	/* Restore host regs x18-x29, lr */
+	restore_callee_saved_regs x29
+
+	/* Do not touch any register after this! */
+__host_enter_without_restoring:
+	eret
+	sb
+SYM_FUNC_END(__host_exit)
+
+/*
+ * void __noreturn __hyp_do_panic(bool restore_host, u64 spsr, u64 elr, u64 par);
+ */
+SYM_FUNC_START(__hyp_do_panic)
+	/* Load the format arguments into x1-7 */
+	mov	x6, x3
+	get_vcpu_ptr x7, x3
+
+	mrs	x3, esr_el2
+	mrs	x4, far_el2
+	mrs	x5, hpfar_el2
+
+	/* Prepare and exit to the host's panic funciton. */
+	mov	lr, #(PSR_F_BIT | PSR_I_BIT | PSR_A_BIT | PSR_D_BIT |\
+		      PSR_MODE_EL1h)
+	msr	spsr_el2, lr
+	ldr	lr, =panic
+	msr	elr_el2, lr
+
+	/*
+	 * Set the panic format string and enter the host, conditionally
+	 * restoring the host context.
+	 */
+	cmp	x0, xzr
+	ldr	x0, =__hyp_panic_string
+	b.eq	__host_enter_without_restoring
+	b	__host_enter_for_panic
+SYM_FUNC_END(__hyp_do_panic)
+
+.macro host_el1_sync_vect
+	.align 7
+.L__vect_start\@:
+	stp	x0, x1, [sp, #-16]!
+	mrs	x0, esr_el2
+	lsr	x0, x0, #ESR_ELx_EC_SHIFT
+	cmp	x0, #ESR_ELx_EC_HVC64
+	ldp	x0, x1, [sp], #16
+	b.ne	__host_exit
+
+	/* Check for a stub HVC call */
+	cmp	x0, #HVC_STUB_HCALL_NR
+	b.hs	__host_exit
+
+	/*
+	 * 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).
+	 *
+	 * Preserve x0-x4, which may contain stub parameters.
+	 */
+	ldr	x5, =__kvm_handle_stub_hvc
+	ldr_l	x6, kimage_voffset
+
+	/* x5 = __pa(x5) */
+	sub	x5, x5, x6
+	br	x5
+.L__vect_end\@:
+.if ((.L__vect_end\@ - .L__vect_start\@) > 0x80)
+	.error "host_el1_sync_vect larger than vector entry"
+.endif
+.endm
+
+.macro invalid_host_el2_vect
+	.align 7
+	/* If a guest is loaded, panic out of it. */
+	stp	x0, x1, [sp, #-16]!
+	get_loaded_vcpu x0, x1
+	cbnz	x0, __guest_exit_panic
+	add	sp, sp, #16
+
+	/*
+	 * The panic may not be clean if the exception is taken before the host
+	 * context has been saved by __host_exit or after the hyp context has
+	 * been partially clobbered by __host_enter.
+	 */
+	b	hyp_panic
+.endm
+
+.macro invalid_host_el1_vect
+	.align 7
+	mov	x0, xzr		/* restore_host = false */
+	mrs	x1, spsr_el2
+	mrs	x2, elr_el2
+	mrs	x3, par_el1
+	b	__hyp_do_panic
+.endm
+
+/*
+ * The host vector does not use an ESB instruction in order to avoid consuming
+ * SErrors that should only be consumed by the host. Guest entry is deferred by
+ * __guest_enter if there are any pending asynchronous exceptions so hyp will
+ * always return to the host without having consumerd host SErrors.
+ *
+ * CONFIG_KVM_INDIRECT_VECTORS is not applied to the host vectors because the
+ * host knows about the EL2 vectors already, and there is no point in hiding
+ * them.
+ */
+	.align 11
+SYM_CODE_START(__kvm_hyp_host_vector)
+	invalid_host_el2_vect			// Synchronous EL2t
+	invalid_host_el2_vect			// IRQ EL2t
+	invalid_host_el2_vect			// FIQ EL2t
+	invalid_host_el2_vect			// Error EL2t
+
+	invalid_host_el2_vect			// Synchronous EL2h
+	invalid_host_el2_vect			// IRQ EL2h
+	invalid_host_el2_vect			// FIQ EL2h
+	invalid_host_el2_vect			// Error EL2h
+
+	host_el1_sync_vect			// Synchronous 64-bit EL1
+	invalid_host_el1_vect			// IRQ 64-bit EL1
+	invalid_host_el1_vect			// FIQ 64-bit EL1
+	invalid_host_el1_vect			// Error 64-bit EL1
+
+	invalid_host_el1_vect			// Synchronous 32-bit EL1
+	invalid_host_el1_vect			// IRQ 32-bit EL1
+	invalid_host_el1_vect			// FIQ 32-bit EL1
+	invalid_host_el1_vect			// Error 32-bit EL1
+SYM_CODE_END(__kvm_hyp_host_vector)
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-init.S b/arch/arm64/kvm/hyp/nvhe/hyp-init.S
index d9434e90c06d..47224dc62c51 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-init.S
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-init.S
@@ -4,11 +4,13 @@
  * Author: Marc Zyngier <marc.zyngier@arm.com>
  */
 
+#include <linux/arm-smccc.h>
 #include <linux/linkage.h>
 
 #include <asm/alternative.h>
 #include <asm/assembler.h>
 #include <asm/kvm_arm.h>
+#include <asm/kvm_asm.h>
 #include <asm/kvm_mmu.h>
 #include <asm/pgtable-hwdef.h>
 #include <asm/sysreg.h>
@@ -44,27 +46,37 @@ __invalid:
 	b	.
 
 	/*
-	 * x0: HYP pgd
-	 * x1: HYP stack
-	 * x2: HYP vectors
-	 * x3: per-CPU offset
+	 * x0: SMCCC function ID
+	 * x1: HYP pgd
+	 * x2: per-CPU offset
+	 * x3: HYP stack
+	 * x4: HYP vectors
 	 */
 __do_hyp_init:
 	/* Check for a stub HVC call */
 	cmp	x0, #HVC_STUB_HCALL_NR
 	b.lo	__kvm_handle_stub_hvc
 
-	phys_to_ttbr x4, x0
+	/* Set tpidr_el2 for use by HYP to free a register */
+	msr	tpidr_el2, x2
+
+	mov	x2, #KVM_HOST_SMCCC_FUNC(__kvm_hyp_init)
+	cmp	x0, x2
+	b.eq	1f
+	mov	x0, #SMCCC_RET_NOT_SUPPORTED
+	eret
+
+1:	phys_to_ttbr x0, x1
 alternative_if ARM64_HAS_CNP
-	orr	x4, x4, #TTBR_CNP_BIT
+	orr	x0, x0, #TTBR_CNP_BIT
 alternative_else_nop_endif
-	msr	ttbr0_el2, x4
+	msr	ttbr0_el2, x0
 
-	mrs	x4, tcr_el1
-	mov_q	x5, TCR_EL2_MASK
-	and	x4, x4, x5
-	mov	x5, #TCR_EL2_RES1
-	orr	x4, x4, x5
+	mrs	x0, tcr_el1
+	mov_q	x1, TCR_EL2_MASK
+	and	x0, x0, x1
+	mov	x1, #TCR_EL2_RES1
+	orr	x0, x0, x1
 
 	/*
 	 * The ID map may be configured to use an extended virtual address
@@ -80,18 +92,18 @@ alternative_else_nop_endif
 	 *
 	 * So use the same T0SZ value we use for the ID map.
 	 */
-	ldr_l	x5, idmap_t0sz
-	bfi	x4, x5, TCR_T0SZ_OFFSET, TCR_TxSZ_WIDTH
+	ldr_l	x1, idmap_t0sz
+	bfi	x0, x1, TCR_T0SZ_OFFSET, TCR_TxSZ_WIDTH
 
 	/*
 	 * Set the PS bits in TCR_EL2.
 	 */
-	tcr_compute_pa_size x4, #TCR_EL2_PS_SHIFT, x5, x6
+	tcr_compute_pa_size x0, #TCR_EL2_PS_SHIFT, x1, x2
 
-	msr	tcr_el2, x4
+	msr	tcr_el2, x0
 
-	mrs	x4, mair_el1
-	msr	mair_el2, x4
+	mrs	x0, mair_el1
+	msr	mair_el2, x0
 	isb
 
 	/* Invalidate the stale TLBs from Bootloader */
@@ -103,25 +115,22 @@ alternative_else_nop_endif
 	 * as well as the EE bit on BE. Drop the A flag since the compiler
 	 * is allowed to generate unaligned accesses.
 	 */
-	mov_q	x4, (SCTLR_EL2_RES1 | (SCTLR_ELx_FLAGS & ~SCTLR_ELx_A))
-CPU_BE(	orr	x4, x4, #SCTLR_ELx_EE)
+	mov_q	x0, (SCTLR_EL2_RES1 | (SCTLR_ELx_FLAGS & ~SCTLR_ELx_A))
+CPU_BE(	orr	x0, x0, #SCTLR_ELx_EE)
 alternative_if ARM64_HAS_ADDRESS_AUTH
-	mov_q	x5, (SCTLR_ELx_ENIA | SCTLR_ELx_ENIB | \
+	mov_q	x1, (SCTLR_ELx_ENIA | SCTLR_ELx_ENIB | \
 		     SCTLR_ELx_ENDA | SCTLR_ELx_ENDB)
-	orr	x4, x4, x5
+	orr	x0, x0, x1
 alternative_else_nop_endif
-	msr	sctlr_el2, x4
+	msr	sctlr_el2, x0
 	isb
 
 	/* Set the stack and new vectors */
-	kern_hyp_va	x1
-	mov	sp, x1
-	msr	vbar_el2, x2
-
-	/* Set tpidr_el2 for use by HYP */
-	msr	tpidr_el2, x3
+	mov	sp, x3
+	msr	vbar_el2, x4
 
 	/* Hello, World! */
+	mov	x0, #SMCCC_RET_SUCCESS
 	eret
 SYM_CODE_END(__kvm_hyp_init)
 
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
new file mode 100644
index 000000000000..e2eafe2c93af
--- /dev/null
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -0,0 +1,117 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2020 - Google Inc
+ * Author: Andrew Scull <ascull@google.com>
+ */
+
+#include <hyp/switch.h>
+
+#include <asm/kvm_asm.h>
+#include <asm/kvm_emulate.h>
+#include <asm/kvm_host.h>
+#include <asm/kvm_hyp.h>
+#include <asm/kvm_mmu.h>
+
+#include <kvm/arm_hypercalls.h>
+
+static void handle_host_hcall(unsigned long func_id,
+			      struct kvm_cpu_context *host_ctxt)
+{
+	unsigned long ret = 0;
+
+	switch (func_id) {
+	case KVM_HOST_SMCCC_FUNC(__kvm_vcpu_run): {
+		unsigned long r1 = host_ctxt->regs.regs[1];
+		struct kvm_vcpu *vcpu = (struct kvm_vcpu *)r1;
+
+		ret = __kvm_vcpu_run(kern_hyp_va(vcpu));
+		break;
+	}
+	case KVM_HOST_SMCCC_FUNC(__kvm_flush_vm_context):
+		__kvm_flush_vm_context();
+		break;
+	case KVM_HOST_SMCCC_FUNC(__kvm_tlb_flush_vmid_ipa): {
+		unsigned long r1 = host_ctxt->regs.regs[1];
+		struct kvm_s2_mmu *mmu = (struct kvm_s2_mmu *)r1;
+		phys_addr_t ipa = host_ctxt->regs.regs[2];
+		int level = host_ctxt->regs.regs[3];
+
+		__kvm_tlb_flush_vmid_ipa(kern_hyp_va(mmu), ipa, level);
+		break;
+	}
+	case KVM_HOST_SMCCC_FUNC(__kvm_tlb_flush_vmid): {
+		unsigned long r1 = host_ctxt->regs.regs[1];
+		struct kvm_s2_mmu *mmu = (struct kvm_s2_mmu *)r1;
+
+		__kvm_tlb_flush_vmid(kern_hyp_va(mmu));
+		break;
+	}
+	case KVM_HOST_SMCCC_FUNC(__kvm_tlb_flush_local_vmid): {
+		unsigned long r1 = host_ctxt->regs.regs[1];
+		struct kvm_s2_mmu *mmu = (struct kvm_s2_mmu *)r1;
+
+		__kvm_tlb_flush_local_vmid(kern_hyp_va(mmu));
+		break;
+	}
+	case KVM_HOST_SMCCC_FUNC(__kvm_timer_set_cntvoff): {
+		u64 cntvoff = host_ctxt->regs.regs[1];
+
+		__kvm_timer_set_cntvoff(cntvoff);
+		break;
+	}
+	case KVM_HOST_SMCCC_FUNC(__kvm_enable_ssbs):
+		__kvm_enable_ssbs();
+		break;
+	case KVM_HOST_SMCCC_FUNC(__vgic_v3_get_ich_vtr_el2):
+		ret = __vgic_v3_get_ich_vtr_el2();
+		break;
+	case KVM_HOST_SMCCC_FUNC(__vgic_v3_read_vmcr):
+		ret = __vgic_v3_read_vmcr();
+		break;
+	case KVM_HOST_SMCCC_FUNC(__vgic_v3_write_vmcr): {
+		u32 vmcr = host_ctxt->regs.regs[1];
+
+		__vgic_v3_write_vmcr(vmcr);
+		break;
+	}
+	case KVM_HOST_SMCCC_FUNC(__vgic_v3_init_lrs):
+		__vgic_v3_init_lrs();
+		break;
+	case KVM_HOST_SMCCC_FUNC(__kvm_get_mdcr_el2):
+		ret = __kvm_get_mdcr_el2();
+		break;
+	case KVM_HOST_SMCCC_FUNC(__vgic_v3_save_aprs): {
+		unsigned long r1 = host_ctxt->regs.regs[1];
+		struct vgic_v3_cpu_if *cpu_if = (struct vgic_v3_cpu_if *)r1;
+
+		__vgic_v3_save_aprs(kern_hyp_va(cpu_if));
+		break;
+	}
+	case KVM_HOST_SMCCC_FUNC(__vgic_v3_restore_aprs): {
+		unsigned long r1 = host_ctxt->regs.regs[1];
+		struct vgic_v3_cpu_if *cpu_if = (struct vgic_v3_cpu_if *)r1;
+
+		__vgic_v3_restore_aprs(kern_hyp_va(cpu_if));
+		break;
+	}
+	default:
+		/* Invalid host HVC. */
+		host_ctxt->regs.regs[0] = SMCCC_RET_NOT_SUPPORTED;
+		return;
+	}
+
+	host_ctxt->regs.regs[0] = SMCCC_RET_SUCCESS;
+	host_ctxt->regs.regs[1] = ret;
+}
+
+void handle_trap(struct kvm_cpu_context *host_ctxt)
+{
+	u64 esr = read_sysreg_el2(SYS_ESR);
+	unsigned long func_id;
+
+	if (ESR_ELx_EC(esr) != ESR_ELx_EC_HVC64)
+		hyp_panic();
+
+	func_id = host_ctxt->regs.regs[0];
+	handle_host_hcall(func_id, host_ctxt);
+}
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp.lds.S b/arch/arm64/kvm/hyp/nvhe/hyp.lds.S
new file mode 100644
index 000000000000..bb2d986ff696
--- /dev/null
+++ b/arch/arm64/kvm/hyp/nvhe/hyp.lds.S
@@ -0,0 +1,19 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2020 Google LLC.
+ * Written by David Brazdil <dbrazdil@google.com>
+ *
+ * Linker script used for partial linking of nVHE EL2 object files.
+ */
+
+#include <asm/hyp_image.h>
+#include <asm-generic/vmlinux.lds.h>
+#include <asm/cache.h>
+#include <asm/memory.h>
+
+SECTIONS {
+	HYP_SECTION(.text)
+	HYP_SECTION_NAME(.data..percpu) : {
+		PERCPU_INPUT(L1_CACHE_BYTES)
+	}
+}
diff --git a/arch/arm64/kvm/hyp/nvhe/switch.c b/arch/arm64/kvm/hyp/nvhe/switch.c
index 0970442d2dbc..a457a0306e03 100644
--- a/arch/arm64/kvm/hyp/nvhe/switch.c
+++ b/arch/arm64/kvm/hyp/nvhe/switch.c
@@ -27,6 +27,11 @@
 #include <asm/processor.h>
 #include <asm/thread_info.h>
 
+/* Non-VHE specific context */
+DEFINE_PER_CPU(struct kvm_host_data, kvm_host_data);
+DEFINE_PER_CPU(struct kvm_cpu_context, kvm_hyp_ctxt);
+DEFINE_PER_CPU(unsigned long, kvm_hyp_vector);
+
 static void __activate_traps(struct kvm_vcpu *vcpu)
 {
 	u64 val;
@@ -42,6 +47,7 @@ static void __activate_traps(struct kvm_vcpu *vcpu)
 	}
 
 	write_sysreg(val, cptr_el2);
+	write_sysreg(__this_cpu_read(kvm_hyp_vector), vbar_el2);
 
 	if (cpus_have_final_cap(ARM64_WORKAROUND_SPECULATIVE_AT)) {
 		struct kvm_cpu_context *ctxt = &vcpu->arch.ctxt;
@@ -60,6 +66,7 @@ static void __activate_traps(struct kvm_vcpu *vcpu)
 
 static void __deactivate_traps(struct kvm_vcpu *vcpu)
 {
+	extern char __kvm_hyp_host_vector[];
 	u64 mdcr_el2;
 
 	___deactivate_traps(vcpu);
@@ -91,9 +98,10 @@ static void __deactivate_traps(struct kvm_vcpu *vcpu)
 	write_sysreg(mdcr_el2, mdcr_el2);
 	write_sysreg(HCR_HOST_NVHE_FLAGS, hcr_el2);
 	write_sysreg(CPTR_EL2_DEFAULT, cptr_el2);
+	write_sysreg(__kvm_hyp_host_vector, vbar_el2);
 }
 
-static void __deactivate_vm(struct kvm_vcpu *vcpu)
+static void __load_host_stage2(void)
 {
 	write_sysreg(0, vttbr_el2);
 }
@@ -173,9 +181,7 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 		pmr_sync();
 	}
 
-	vcpu = kern_hyp_va(vcpu);
-
-	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
 	host_ctxt->__hyp_running_vcpu = vcpu;
 	guest_ctxt = &vcpu->arch.ctxt;
 
@@ -194,7 +200,7 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	__sysreg32_restore_state(vcpu);
 	__sysreg_restore_state_nvhe(guest_ctxt);
 
-	__activate_vm(kern_hyp_va(vcpu->arch.hw_mmu));
+	__load_guest_stage2(kern_hyp_va(vcpu->arch.hw_mmu));
 	__activate_traps(vcpu);
 
 	__hyp_vgic_restore_state(vcpu);
@@ -202,24 +208,20 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 
 	__debug_switch_to_guest(vcpu);
 
-	__set_guest_arch_workaround_state(vcpu);
-
 	do {
 		/* Jump in the fire! */
-		exit_code = __guest_enter(vcpu, host_ctxt);
+		exit_code = __guest_enter(vcpu);
 
 		/* And we're baaack! */
 	} while (fixup_guest_exit(vcpu, &exit_code));
 
-	__set_host_arch_workaround_state(vcpu);
-
 	__sysreg_save_state_nvhe(guest_ctxt);
 	__sysreg32_save_state(vcpu);
 	__timer_disable_traps(vcpu);
 	__hyp_vgic_save_state(vcpu);
 
 	__deactivate_traps(vcpu);
-	__deactivate_vm(vcpu);
+	__load_host_stage2();
 
 	__sysreg_restore_state_nvhe(host_ctxt);
 
@@ -239,35 +241,31 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	if (system_uses_irq_prio_masking())
 		gic_write_pmr(GIC_PRIO_IRQOFF);
 
+	host_ctxt->__hyp_running_vcpu = NULL;
+
 	return exit_code;
 }
 
-void __noreturn hyp_panic(struct kvm_cpu_context *host_ctxt)
+void __noreturn hyp_panic(void)
 {
 	u64 spsr = read_sysreg_el2(SYS_SPSR);
 	u64 elr = read_sysreg_el2(SYS_ELR);
 	u64 par = read_sysreg(par_el1);
-	struct kvm_vcpu *vcpu = host_ctxt->__hyp_running_vcpu;
-	unsigned long str_va;
+	bool restore_host = true;
+	struct kvm_cpu_context *host_ctxt;
+	struct kvm_vcpu *vcpu;
+
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+	vcpu = host_ctxt->__hyp_running_vcpu;
 
-	if (read_sysreg(vttbr_el2)) {
+	if (vcpu) {
 		__timer_disable_traps(vcpu);
 		__deactivate_traps(vcpu);
-		__deactivate_vm(vcpu);
+		__load_host_stage2();
 		__sysreg_restore_state_nvhe(host_ctxt);
 	}
 
-	/*
-	 * Force the panic string to be loaded from the literal pool,
-	 * making sure it is a kernel address and not a PC-relative
-	 * reference.
-	 */
-	asm volatile("ldr %0, =%1" : "=r" (str_va) : "S" (__hyp_panic_string));
-
-	__hyp_do_panic(str_va,
-		       spsr, elr,
-		       read_sysreg(esr_el2), read_sysreg_el2(SYS_FAR),
-		       read_sysreg(hpfar_el2), par, vcpu);
+	__hyp_do_panic(restore_host, spsr, elr, par);
 	unreachable();
 }
 
diff --git a/arch/arm64/kvm/hyp/nvhe/tlb.c b/arch/arm64/kvm/hyp/nvhe/tlb.c
index 69eae608d670..544bca3072b7 100644
--- a/arch/arm64/kvm/hyp/nvhe/tlb.c
+++ b/arch/arm64/kvm/hyp/nvhe/tlb.c
@@ -54,7 +54,6 @@ void __kvm_tlb_flush_vmid_ipa(struct kvm_s2_mmu *mmu,
 	dsb(ishst);
 
 	/* Switch to requested VMID */
-	mmu = kern_hyp_va(mmu);
 	__tlb_switch_to_guest(mmu, &cxt);
 
 	/*
@@ -108,7 +107,6 @@ void __kvm_tlb_flush_vmid(struct kvm_s2_mmu *mmu)
 	dsb(ishst);
 
 	/* Switch to requested VMID */
-	mmu = kern_hyp_va(mmu);
 	__tlb_switch_to_guest(mmu, &cxt);
 
 	__tlbi(vmalls12e1is);
diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
new file mode 100644
index 000000000000..0cdf6e461cbd
--- /dev/null
+++ b/arch/arm64/kvm/hyp/pgtable.c
@@ -0,0 +1,892 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Stand-alone page-table allocator for hyp stage-1 and guest stage-2.
+ * No bombay mix was harmed in the writing of this file.
+ *
+ * Copyright (C) 2020 Google LLC
+ * Author: Will Deacon <will@kernel.org>
+ */
+
+#include <linux/bitfield.h>
+#include <asm/kvm_pgtable.h>
+
+#define KVM_PGTABLE_MAX_LEVELS		4U
+
+#define KVM_PTE_VALID			BIT(0)
+
+#define KVM_PTE_TYPE			BIT(1)
+#define KVM_PTE_TYPE_BLOCK		0
+#define KVM_PTE_TYPE_PAGE		1
+#define KVM_PTE_TYPE_TABLE		1
+
+#define KVM_PTE_ADDR_MASK		GENMASK(47, PAGE_SHIFT)
+#define KVM_PTE_ADDR_51_48		GENMASK(15, 12)
+
+#define KVM_PTE_LEAF_ATTR_LO		GENMASK(11, 2)
+
+#define KVM_PTE_LEAF_ATTR_LO_S1_ATTRIDX	GENMASK(4, 2)
+#define KVM_PTE_LEAF_ATTR_LO_S1_AP	GENMASK(7, 6)
+#define KVM_PTE_LEAF_ATTR_LO_S1_AP_RO	3
+#define KVM_PTE_LEAF_ATTR_LO_S1_AP_RW	1
+#define KVM_PTE_LEAF_ATTR_LO_S1_SH	GENMASK(9, 8)
+#define KVM_PTE_LEAF_ATTR_LO_S1_SH_IS	3
+#define KVM_PTE_LEAF_ATTR_LO_S1_AF	BIT(10)
+
+#define KVM_PTE_LEAF_ATTR_LO_S2_MEMATTR	GENMASK(5, 2)
+#define KVM_PTE_LEAF_ATTR_LO_S2_S2AP_R	BIT(6)
+#define KVM_PTE_LEAF_ATTR_LO_S2_S2AP_W	BIT(7)
+#define KVM_PTE_LEAF_ATTR_LO_S2_SH	GENMASK(9, 8)
+#define KVM_PTE_LEAF_ATTR_LO_S2_SH_IS	3
+#define KVM_PTE_LEAF_ATTR_LO_S2_AF	BIT(10)
+
+#define KVM_PTE_LEAF_ATTR_HI		GENMASK(63, 51)
+
+#define KVM_PTE_LEAF_ATTR_HI_S1_XN	BIT(54)
+
+#define KVM_PTE_LEAF_ATTR_HI_S2_XN	BIT(54)
+
+struct kvm_pgtable_walk_data {
+	struct kvm_pgtable		*pgt;
+	struct kvm_pgtable_walker	*walker;
+
+	u64				addr;
+	u64				end;
+};
+
+static u64 kvm_granule_shift(u32 level)
+{
+	/* Assumes KVM_PGTABLE_MAX_LEVELS is 4 */
+	return ARM64_HW_PGTABLE_LEVEL_SHIFT(level);
+}
+
+static u64 kvm_granule_size(u32 level)
+{
+	return BIT(kvm_granule_shift(level));
+}
+
+static bool kvm_block_mapping_supported(u64 addr, u64 end, u64 phys, u32 level)
+{
+	u64 granule = kvm_granule_size(level);
+
+	/*
+	 * Reject invalid block mappings and don't bother with 4TB mappings for
+	 * 52-bit PAs.
+	 */
+	if (level == 0 || (PAGE_SIZE != SZ_4K && level == 1))
+		return false;
+
+	if (granule > (end - addr))
+		return false;
+
+	return IS_ALIGNED(addr, granule) && IS_ALIGNED(phys, granule);
+}
+
+static u32 kvm_pgtable_idx(struct kvm_pgtable_walk_data *data, u32 level)
+{
+	u64 shift = kvm_granule_shift(level);
+	u64 mask = BIT(PAGE_SHIFT - 3) - 1;
+
+	return (data->addr >> shift) & mask;
+}
+
+static u32 __kvm_pgd_page_idx(struct kvm_pgtable *pgt, u64 addr)
+{
+	u64 shift = kvm_granule_shift(pgt->start_level - 1); /* May underflow */
+	u64 mask = BIT(pgt->ia_bits) - 1;
+
+	return (addr & mask) >> shift;
+}
+
+static u32 kvm_pgd_page_idx(struct kvm_pgtable_walk_data *data)
+{
+	return __kvm_pgd_page_idx(data->pgt, data->addr);
+}
+
+static u32 kvm_pgd_pages(u32 ia_bits, u32 start_level)
+{
+	struct kvm_pgtable pgt = {
+		.ia_bits	= ia_bits,
+		.start_level	= start_level,
+	};
+
+	return __kvm_pgd_page_idx(&pgt, -1ULL) + 1;
+}
+
+static bool kvm_pte_valid(kvm_pte_t pte)
+{
+	return pte & KVM_PTE_VALID;
+}
+
+static bool kvm_pte_table(kvm_pte_t pte, u32 level)
+{
+	if (level == KVM_PGTABLE_MAX_LEVELS - 1)
+		return false;
+
+	if (!kvm_pte_valid(pte))
+		return false;
+
+	return FIELD_GET(KVM_PTE_TYPE, pte) == KVM_PTE_TYPE_TABLE;
+}
+
+static u64 kvm_pte_to_phys(kvm_pte_t pte)
+{
+	u64 pa = pte & KVM_PTE_ADDR_MASK;
+
+	if (PAGE_SHIFT == 16)
+		pa |= FIELD_GET(KVM_PTE_ADDR_51_48, pte) << 48;
+
+	return pa;
+}
+
+static kvm_pte_t kvm_phys_to_pte(u64 pa)
+{
+	kvm_pte_t pte = pa & KVM_PTE_ADDR_MASK;
+
+	if (PAGE_SHIFT == 16)
+		pte |= FIELD_PREP(KVM_PTE_ADDR_51_48, pa >> 48);
+
+	return pte;
+}
+
+static kvm_pte_t *kvm_pte_follow(kvm_pte_t pte)
+{
+	return __va(kvm_pte_to_phys(pte));
+}
+
+static void kvm_set_invalid_pte(kvm_pte_t *ptep)
+{
+	kvm_pte_t pte = *ptep;
+	WRITE_ONCE(*ptep, pte & ~KVM_PTE_VALID);
+}
+
+static void kvm_set_table_pte(kvm_pte_t *ptep, kvm_pte_t *childp)
+{
+	kvm_pte_t old = *ptep, pte = kvm_phys_to_pte(__pa(childp));
+
+	pte |= FIELD_PREP(KVM_PTE_TYPE, KVM_PTE_TYPE_TABLE);
+	pte |= KVM_PTE_VALID;
+
+	WARN_ON(kvm_pte_valid(old));
+	smp_store_release(ptep, pte);
+}
+
+static bool kvm_set_valid_leaf_pte(kvm_pte_t *ptep, u64 pa, kvm_pte_t attr,
+				   u32 level)
+{
+	kvm_pte_t old = *ptep, pte = kvm_phys_to_pte(pa);
+	u64 type = (level == KVM_PGTABLE_MAX_LEVELS - 1) ? KVM_PTE_TYPE_PAGE :
+							   KVM_PTE_TYPE_BLOCK;
+
+	pte |= attr & (KVM_PTE_LEAF_ATTR_LO | KVM_PTE_LEAF_ATTR_HI);
+	pte |= FIELD_PREP(KVM_PTE_TYPE, type);
+	pte |= KVM_PTE_VALID;
+
+	/* Tolerate KVM recreating the exact same mapping. */
+	if (kvm_pte_valid(old))
+		return old == pte;
+
+	smp_store_release(ptep, pte);
+	return true;
+}
+
+static int kvm_pgtable_visitor_cb(struct kvm_pgtable_walk_data *data, u64 addr,
+				  u32 level, kvm_pte_t *ptep,
+				  enum kvm_pgtable_walk_flags flag)
+{
+	struct kvm_pgtable_walker *walker = data->walker;
+	return walker->cb(addr, data->end, level, ptep, flag, walker->arg);
+}
+
+static int __kvm_pgtable_walk(struct kvm_pgtable_walk_data *data,
+			      kvm_pte_t *pgtable, u32 level);
+
+static inline int __kvm_pgtable_visit(struct kvm_pgtable_walk_data *data,
+				      kvm_pte_t *ptep, u32 level)
+{
+	int ret = 0;
+	u64 addr = data->addr;
+	kvm_pte_t *childp, pte = *ptep;
+	bool table = kvm_pte_table(pte, level);
+	enum kvm_pgtable_walk_flags flags = data->walker->flags;
+
+	if (table && (flags & KVM_PGTABLE_WALK_TABLE_PRE)) {
+		ret = kvm_pgtable_visitor_cb(data, addr, level, ptep,
+					     KVM_PGTABLE_WALK_TABLE_PRE);
+	}
+
+	if (!table && (flags & KVM_PGTABLE_WALK_LEAF)) {
+		ret = kvm_pgtable_visitor_cb(data, addr, level, ptep,
+					     KVM_PGTABLE_WALK_LEAF);
+		pte = *ptep;
+		table = kvm_pte_table(pte, level);
+	}
+
+	if (ret)
+		goto out;
+
+	if (!table) {
+		data->addr += kvm_granule_size(level);
+		goto out;
+	}
+
+	childp = kvm_pte_follow(pte);
+	ret = __kvm_pgtable_walk(data, childp, level + 1);
+	if (ret)
+		goto out;
+
+	if (flags & KVM_PGTABLE_WALK_TABLE_POST) {
+		ret = kvm_pgtable_visitor_cb(data, addr, level, ptep,
+					     KVM_PGTABLE_WALK_TABLE_POST);
+	}
+
+out:
+	return ret;
+}
+
+static int __kvm_pgtable_walk(struct kvm_pgtable_walk_data *data,
+			      kvm_pte_t *pgtable, u32 level)
+{
+	u32 idx;
+	int ret = 0;
+
+	if (WARN_ON_ONCE(level >= KVM_PGTABLE_MAX_LEVELS))
+		return -EINVAL;
+
+	for (idx = kvm_pgtable_idx(data, level); idx < PTRS_PER_PTE; ++idx) {
+		kvm_pte_t *ptep = &pgtable[idx];
+
+		if (data->addr >= data->end)
+			break;
+
+		ret = __kvm_pgtable_visit(data, ptep, level);
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+static int _kvm_pgtable_walk(struct kvm_pgtable_walk_data *data)
+{
+	u32 idx;
+	int ret = 0;
+	struct kvm_pgtable *pgt = data->pgt;
+	u64 limit = BIT(pgt->ia_bits);
+
+	if (data->addr > limit || data->end > limit)
+		return -ERANGE;
+
+	if (!pgt->pgd)
+		return -EINVAL;
+
+	for (idx = kvm_pgd_page_idx(data); data->addr < data->end; ++idx) {
+		kvm_pte_t *ptep = &pgt->pgd[idx * PTRS_PER_PTE];
+
+		ret = __kvm_pgtable_walk(data, ptep, pgt->start_level);
+		if (ret)
+			break;
+	}
+
+	return ret;
+}
+
+int kvm_pgtable_walk(struct kvm_pgtable *pgt, u64 addr, u64 size,
+		     struct kvm_pgtable_walker *walker)
+{
+	struct kvm_pgtable_walk_data walk_data = {
+		.pgt	= pgt,
+		.addr	= ALIGN_DOWN(addr, PAGE_SIZE),
+		.end	= PAGE_ALIGN(walk_data.addr + size),
+		.walker	= walker,
+	};
+
+	return _kvm_pgtable_walk(&walk_data);
+}
+
+struct hyp_map_data {
+	u64		phys;
+	kvm_pte_t	attr;
+};
+
+static int hyp_map_set_prot_attr(enum kvm_pgtable_prot prot,
+				 struct hyp_map_data *data)
+{
+	bool device = prot & KVM_PGTABLE_PROT_DEVICE;
+	u32 mtype = device ? MT_DEVICE_nGnRE : MT_NORMAL;
+	kvm_pte_t attr = FIELD_PREP(KVM_PTE_LEAF_ATTR_LO_S1_ATTRIDX, mtype);
+	u32 sh = KVM_PTE_LEAF_ATTR_LO_S1_SH_IS;
+	u32 ap = (prot & KVM_PGTABLE_PROT_W) ? KVM_PTE_LEAF_ATTR_LO_S1_AP_RW :
+					       KVM_PTE_LEAF_ATTR_LO_S1_AP_RO;
+
+	if (!(prot & KVM_PGTABLE_PROT_R))
+		return -EINVAL;
+
+	if (prot & KVM_PGTABLE_PROT_X) {
+		if (prot & KVM_PGTABLE_PROT_W)
+			return -EINVAL;
+
+		if (device)
+			return -EINVAL;
+	} else {
+		attr |= KVM_PTE_LEAF_ATTR_HI_S1_XN;
+	}
+
+	attr |= FIELD_PREP(KVM_PTE_LEAF_ATTR_LO_S1_AP, ap);
+	attr |= FIELD_PREP(KVM_PTE_LEAF_ATTR_LO_S1_SH, sh);
+	attr |= KVM_PTE_LEAF_ATTR_LO_S1_AF;
+	data->attr = attr;
+	return 0;
+}
+
+static bool hyp_map_walker_try_leaf(u64 addr, u64 end, u32 level,
+				    kvm_pte_t *ptep, struct hyp_map_data *data)
+{
+	u64 granule = kvm_granule_size(level), phys = data->phys;
+
+	if (!kvm_block_mapping_supported(addr, end, phys, level))
+		return false;
+
+	WARN_ON(!kvm_set_valid_leaf_pte(ptep, phys, data->attr, level));
+	data->phys += granule;
+	return true;
+}
+
+static int hyp_map_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+			  enum kvm_pgtable_walk_flags flag, void * const arg)
+{
+	kvm_pte_t *childp;
+
+	if (hyp_map_walker_try_leaf(addr, end, level, ptep, arg))
+		return 0;
+
+	if (WARN_ON(level == KVM_PGTABLE_MAX_LEVELS - 1))
+		return -EINVAL;
+
+	childp = (kvm_pte_t *)get_zeroed_page(GFP_KERNEL);
+	if (!childp)
+		return -ENOMEM;
+
+	kvm_set_table_pte(ptep, childp);
+	return 0;
+}
+
+int kvm_pgtable_hyp_map(struct kvm_pgtable *pgt, u64 addr, u64 size, u64 phys,
+			enum kvm_pgtable_prot prot)
+{
+	int ret;
+	struct hyp_map_data map_data = {
+		.phys	= ALIGN_DOWN(phys, PAGE_SIZE),
+	};
+	struct kvm_pgtable_walker walker = {
+		.cb	= hyp_map_walker,
+		.flags	= KVM_PGTABLE_WALK_LEAF,
+		.arg	= &map_data,
+	};
+
+	ret = hyp_map_set_prot_attr(prot, &map_data);
+	if (ret)
+		return ret;
+
+	ret = kvm_pgtable_walk(pgt, addr, size, &walker);
+	dsb(ishst);
+	isb();
+	return ret;
+}
+
+int kvm_pgtable_hyp_init(struct kvm_pgtable *pgt, u32 va_bits)
+{
+	u64 levels = ARM64_HW_PGTABLE_LEVELS(va_bits);
+
+	pgt->pgd = (kvm_pte_t *)get_zeroed_page(GFP_KERNEL);
+	if (!pgt->pgd)
+		return -ENOMEM;
+
+	pgt->ia_bits		= va_bits;
+	pgt->start_level	= KVM_PGTABLE_MAX_LEVELS - levels;
+	pgt->mmu		= NULL;
+	return 0;
+}
+
+static int hyp_free_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+			   enum kvm_pgtable_walk_flags flag, void * const arg)
+{
+	free_page((unsigned long)kvm_pte_follow(*ptep));
+	return 0;
+}
+
+void kvm_pgtable_hyp_destroy(struct kvm_pgtable *pgt)
+{
+	struct kvm_pgtable_walker walker = {
+		.cb	= hyp_free_walker,
+		.flags	= KVM_PGTABLE_WALK_TABLE_POST,
+	};
+
+	WARN_ON(kvm_pgtable_walk(pgt, 0, BIT(pgt->ia_bits), &walker));
+	free_page((unsigned long)pgt->pgd);
+	pgt->pgd = NULL;
+}
+
+struct stage2_map_data {
+	u64				phys;
+	kvm_pte_t			attr;
+
+	kvm_pte_t			*anchor;
+
+	struct kvm_s2_mmu		*mmu;
+	struct kvm_mmu_memory_cache	*memcache;
+};
+
+static int stage2_map_set_prot_attr(enum kvm_pgtable_prot prot,
+				    struct stage2_map_data *data)
+{
+	bool device = prot & KVM_PGTABLE_PROT_DEVICE;
+	kvm_pte_t attr = device ? PAGE_S2_MEMATTR(DEVICE_nGnRE) :
+			    PAGE_S2_MEMATTR(NORMAL);
+	u32 sh = KVM_PTE_LEAF_ATTR_LO_S2_SH_IS;
+
+	if (!(prot & KVM_PGTABLE_PROT_X))
+		attr |= KVM_PTE_LEAF_ATTR_HI_S2_XN;
+	else if (device)
+		return -EINVAL;
+
+	if (prot & KVM_PGTABLE_PROT_R)
+		attr |= KVM_PTE_LEAF_ATTR_LO_S2_S2AP_R;
+
+	if (prot & KVM_PGTABLE_PROT_W)
+		attr |= KVM_PTE_LEAF_ATTR_LO_S2_S2AP_W;
+
+	attr |= FIELD_PREP(KVM_PTE_LEAF_ATTR_LO_S2_SH, sh);
+	attr |= KVM_PTE_LEAF_ATTR_LO_S2_AF;
+	data->attr = attr;
+	return 0;
+}
+
+static bool stage2_map_walker_try_leaf(u64 addr, u64 end, u32 level,
+				       kvm_pte_t *ptep,
+				       struct stage2_map_data *data)
+{
+	u64 granule = kvm_granule_size(level), phys = data->phys;
+
+	if (!kvm_block_mapping_supported(addr, end, phys, level))
+		return false;
+
+	if (kvm_set_valid_leaf_pte(ptep, phys, data->attr, level))
+		goto out;
+
+	/* There's an existing valid leaf entry, so perform break-before-make */
+	kvm_set_invalid_pte(ptep);
+	kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, data->mmu, addr, level);
+	kvm_set_valid_leaf_pte(ptep, phys, data->attr, level);
+out:
+	data->phys += granule;
+	return true;
+}
+
+static int stage2_map_walk_table_pre(u64 addr, u64 end, u32 level,
+				     kvm_pte_t *ptep,
+				     struct stage2_map_data *data)
+{
+	if (data->anchor)
+		return 0;
+
+	if (!kvm_block_mapping_supported(addr, end, data->phys, level))
+		return 0;
+
+	kvm_set_invalid_pte(ptep);
+	kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, data->mmu, addr, 0);
+	data->anchor = ptep;
+	return 0;
+}
+
+static int stage2_map_walk_leaf(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+				struct stage2_map_data *data)
+{
+	kvm_pte_t *childp, pte = *ptep;
+	struct page *page = virt_to_page(ptep);
+
+	if (data->anchor) {
+		if (kvm_pte_valid(pte))
+			put_page(page);
+
+		return 0;
+	}
+
+	if (stage2_map_walker_try_leaf(addr, end, level, ptep, data))
+		goto out_get_page;
+
+	if (WARN_ON(level == KVM_PGTABLE_MAX_LEVELS - 1))
+		return -EINVAL;
+
+	if (!data->memcache)
+		return -ENOMEM;
+
+	childp = kvm_mmu_memory_cache_alloc(data->memcache);
+	if (!childp)
+		return -ENOMEM;
+
+	/*
+	 * If we've run into an existing block mapping then replace it with
+	 * a table. Accesses beyond 'end' that fall within the new table
+	 * will be mapped lazily.
+	 */
+	if (kvm_pte_valid(pte)) {
+		kvm_set_invalid_pte(ptep);
+		kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, data->mmu, addr, level);
+		put_page(page);
+	}
+
+	kvm_set_table_pte(ptep, childp);
+
+out_get_page:
+	get_page(page);
+	return 0;
+}
+
+static int stage2_map_walk_table_post(u64 addr, u64 end, u32 level,
+				      kvm_pte_t *ptep,
+				      struct stage2_map_data *data)
+{
+	int ret = 0;
+
+	if (!data->anchor)
+		return 0;
+
+	free_page((unsigned long)kvm_pte_follow(*ptep));
+	put_page(virt_to_page(ptep));
+
+	if (data->anchor == ptep) {
+		data->anchor = NULL;
+		ret = stage2_map_walk_leaf(addr, end, level, ptep, data);
+	}
+
+	return ret;
+}
+
+/*
+ * This is a little fiddly, as we use all three of the walk flags. The idea
+ * is that the TABLE_PRE callback runs for table entries on the way down,
+ * looking for table entries which we could conceivably replace with a
+ * block entry for this mapping. If it finds one, then it sets the 'anchor'
+ * field in 'struct stage2_map_data' to point at the table entry, before
+ * clearing the entry to zero and descending into the now detached table.
+ *
+ * The behaviour of the LEAF callback then depends on whether or not the
+ * anchor has been set. If not, then we're not using a block mapping higher
+ * up the table and we perform the mapping at the existing leaves instead.
+ * If, on the other hand, the anchor _is_ set, then we drop references to
+ * all valid leaves so that the pages beneath the anchor can be freed.
+ *
+ * Finally, the TABLE_POST callback does nothing if the anchor has not
+ * been set, but otherwise frees the page-table pages while walking back up
+ * the page-table, installing the block entry when it revisits the anchor
+ * pointer and clearing the anchor to NULL.
+ */
+static int stage2_map_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+			     enum kvm_pgtable_walk_flags flag, void * const arg)
+{
+	struct stage2_map_data *data = arg;
+
+	switch (flag) {
+	case KVM_PGTABLE_WALK_TABLE_PRE:
+		return stage2_map_walk_table_pre(addr, end, level, ptep, data);
+	case KVM_PGTABLE_WALK_LEAF:
+		return stage2_map_walk_leaf(addr, end, level, ptep, data);
+	case KVM_PGTABLE_WALK_TABLE_POST:
+		return stage2_map_walk_table_post(addr, end, level, ptep, data);
+	}
+
+	return -EINVAL;
+}
+
+int kvm_pgtable_stage2_map(struct kvm_pgtable *pgt, u64 addr, u64 size,
+			   u64 phys, enum kvm_pgtable_prot prot,
+			   struct kvm_mmu_memory_cache *mc)
+{
+	int ret;
+	struct stage2_map_data map_data = {
+		.phys		= ALIGN_DOWN(phys, PAGE_SIZE),
+		.mmu		= pgt->mmu,
+		.memcache	= mc,
+	};
+	struct kvm_pgtable_walker walker = {
+		.cb		= stage2_map_walker,
+		.flags		= KVM_PGTABLE_WALK_TABLE_PRE |
+				  KVM_PGTABLE_WALK_LEAF |
+				  KVM_PGTABLE_WALK_TABLE_POST,
+		.arg		= &map_data,
+	};
+
+	ret = stage2_map_set_prot_attr(prot, &map_data);
+	if (ret)
+		return ret;
+
+	ret = kvm_pgtable_walk(pgt, addr, size, &walker);
+	dsb(ishst);
+	return ret;
+}
+
+static void stage2_flush_dcache(void *addr, u64 size)
+{
+	if (cpus_have_const_cap(ARM64_HAS_STAGE2_FWB))
+		return;
+
+	__flush_dcache_area(addr, size);
+}
+
+static bool stage2_pte_cacheable(kvm_pte_t pte)
+{
+	u64 memattr = FIELD_GET(KVM_PTE_LEAF_ATTR_LO_S2_MEMATTR, pte);
+	return memattr == PAGE_S2_MEMATTR(NORMAL);
+}
+
+static int stage2_unmap_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+			       enum kvm_pgtable_walk_flags flag,
+			       void * const arg)
+{
+	struct kvm_s2_mmu *mmu = arg;
+	kvm_pte_t pte = *ptep, *childp = NULL;
+	bool need_flush = false;
+
+	if (!kvm_pte_valid(pte))
+		return 0;
+
+	if (kvm_pte_table(pte, level)) {
+		childp = kvm_pte_follow(pte);
+
+		if (page_count(virt_to_page(childp)) != 1)
+			return 0;
+	} else if (stage2_pte_cacheable(pte)) {
+		need_flush = true;
+	}
+
+	/*
+	 * This is similar to the map() path in that we unmap the entire
+	 * block entry and rely on the remaining portions being faulted
+	 * back lazily.
+	 */
+	kvm_set_invalid_pte(ptep);
+	kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, mmu, addr, level);
+	put_page(virt_to_page(ptep));
+
+	if (need_flush) {
+		stage2_flush_dcache(kvm_pte_follow(pte),
+				    kvm_granule_size(level));
+	}
+
+	if (childp)
+		free_page((unsigned long)childp);
+
+	return 0;
+}
+
+int kvm_pgtable_stage2_unmap(struct kvm_pgtable *pgt, u64 addr, u64 size)
+{
+	struct kvm_pgtable_walker walker = {
+		.cb	= stage2_unmap_walker,
+		.arg	= pgt->mmu,
+		.flags	= KVM_PGTABLE_WALK_LEAF | KVM_PGTABLE_WALK_TABLE_POST,
+	};
+
+	return kvm_pgtable_walk(pgt, addr, size, &walker);
+}
+
+struct stage2_attr_data {
+	kvm_pte_t	attr_set;
+	kvm_pte_t	attr_clr;
+	kvm_pte_t	pte;
+	u32		level;
+};
+
+static int stage2_attr_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+			      enum kvm_pgtable_walk_flags flag,
+			      void * const arg)
+{
+	kvm_pte_t pte = *ptep;
+	struct stage2_attr_data *data = arg;
+
+	if (!kvm_pte_valid(pte))
+		return 0;
+
+	data->level = level;
+	data->pte = pte;
+	pte &= ~data->attr_clr;
+	pte |= data->attr_set;
+
+	/*
+	 * We may race with the CPU trying to set the access flag here,
+	 * but worst-case the access flag update gets lost and will be
+	 * set on the next access instead.
+	 */
+	if (data->pte != pte)
+		WRITE_ONCE(*ptep, pte);
+
+	return 0;
+}
+
+static int stage2_update_leaf_attrs(struct kvm_pgtable *pgt, u64 addr,
+				    u64 size, kvm_pte_t attr_set,
+				    kvm_pte_t attr_clr, kvm_pte_t *orig_pte,
+				    u32 *level)
+{
+	int ret;
+	kvm_pte_t attr_mask = KVM_PTE_LEAF_ATTR_LO | KVM_PTE_LEAF_ATTR_HI;
+	struct stage2_attr_data data = {
+		.attr_set	= attr_set & attr_mask,
+		.attr_clr	= attr_clr & attr_mask,
+	};
+	struct kvm_pgtable_walker walker = {
+		.cb		= stage2_attr_walker,
+		.arg		= &data,
+		.flags		= KVM_PGTABLE_WALK_LEAF,
+	};
+
+	ret = kvm_pgtable_walk(pgt, addr, size, &walker);
+	if (ret)
+		return ret;
+
+	if (orig_pte)
+		*orig_pte = data.pte;
+
+	if (level)
+		*level = data.level;
+	return 0;
+}
+
+int kvm_pgtable_stage2_wrprotect(struct kvm_pgtable *pgt, u64 addr, u64 size)
+{
+	return stage2_update_leaf_attrs(pgt, addr, size, 0,
+					KVM_PTE_LEAF_ATTR_LO_S2_S2AP_W,
+					NULL, NULL);
+}
+
+kvm_pte_t kvm_pgtable_stage2_mkyoung(struct kvm_pgtable *pgt, u64 addr)
+{
+	kvm_pte_t pte = 0;
+	stage2_update_leaf_attrs(pgt, addr, 1, KVM_PTE_LEAF_ATTR_LO_S2_AF, 0,
+				 &pte, NULL);
+	dsb(ishst);
+	return pte;
+}
+
+kvm_pte_t kvm_pgtable_stage2_mkold(struct kvm_pgtable *pgt, u64 addr)
+{
+	kvm_pte_t pte = 0;
+	stage2_update_leaf_attrs(pgt, addr, 1, 0, KVM_PTE_LEAF_ATTR_LO_S2_AF,
+				 &pte, NULL);
+	/*
+	 * "But where's the TLBI?!", you scream.
+	 * "Over in the core code", I sigh.
+	 *
+	 * See the '->clear_flush_young()' callback on the KVM mmu notifier.
+	 */
+	return pte;
+}
+
+bool kvm_pgtable_stage2_is_young(struct kvm_pgtable *pgt, u64 addr)
+{
+	kvm_pte_t pte = 0;
+	stage2_update_leaf_attrs(pgt, addr, 1, 0, 0, &pte, NULL);
+	return pte & KVM_PTE_LEAF_ATTR_LO_S2_AF;
+}
+
+int kvm_pgtable_stage2_relax_perms(struct kvm_pgtable *pgt, u64 addr,
+				   enum kvm_pgtable_prot prot)
+{
+	int ret;
+	u32 level;
+	kvm_pte_t set = 0, clr = 0;
+
+	if (prot & KVM_PGTABLE_PROT_R)
+		set |= KVM_PTE_LEAF_ATTR_LO_S2_S2AP_R;
+
+	if (prot & KVM_PGTABLE_PROT_W)
+		set |= KVM_PTE_LEAF_ATTR_LO_S2_S2AP_W;
+
+	if (prot & KVM_PGTABLE_PROT_X)
+		clr |= KVM_PTE_LEAF_ATTR_HI_S2_XN;
+
+	ret = stage2_update_leaf_attrs(pgt, addr, 1, set, clr, NULL, &level);
+	if (!ret)
+		kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, pgt->mmu, addr, level);
+	return ret;
+}
+
+static int stage2_flush_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+			       enum kvm_pgtable_walk_flags flag,
+			       void * const arg)
+{
+	kvm_pte_t pte = *ptep;
+
+	if (!kvm_pte_valid(pte) || !stage2_pte_cacheable(pte))
+		return 0;
+
+	stage2_flush_dcache(kvm_pte_follow(pte), kvm_granule_size(level));
+	return 0;
+}
+
+int kvm_pgtable_stage2_flush(struct kvm_pgtable *pgt, u64 addr, u64 size)
+{
+	struct kvm_pgtable_walker walker = {
+		.cb	= stage2_flush_walker,
+		.flags	= KVM_PGTABLE_WALK_LEAF,
+	};
+
+	if (cpus_have_const_cap(ARM64_HAS_STAGE2_FWB))
+		return 0;
+
+	return kvm_pgtable_walk(pgt, addr, size, &walker);
+}
+
+int kvm_pgtable_stage2_init(struct kvm_pgtable *pgt, struct kvm *kvm)
+{
+	size_t pgd_sz;
+	u64 vtcr = kvm->arch.vtcr;
+	u32 ia_bits = VTCR_EL2_IPA(vtcr);
+	u32 sl0 = FIELD_GET(VTCR_EL2_SL0_MASK, vtcr);
+	u32 start_level = VTCR_EL2_TGRAN_SL0_BASE - sl0;
+
+	pgd_sz = kvm_pgd_pages(ia_bits, start_level) * PAGE_SIZE;
+	pgt->pgd = alloc_pages_exact(pgd_sz, GFP_KERNEL | __GFP_ZERO);
+	if (!pgt->pgd)
+		return -ENOMEM;
+
+	pgt->ia_bits		= ia_bits;
+	pgt->start_level	= start_level;
+	pgt->mmu		= &kvm->arch.mmu;
+
+	/* Ensure zeroed PGD pages are visible to the hardware walker */
+	dsb(ishst);
+	return 0;
+}
+
+static int stage2_free_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+			      enum kvm_pgtable_walk_flags flag,
+			      void * const arg)
+{
+	kvm_pte_t pte = *ptep;
+
+	if (!kvm_pte_valid(pte))
+		return 0;
+
+	put_page(virt_to_page(ptep));
+
+	if (kvm_pte_table(pte, level))
+		free_page((unsigned long)kvm_pte_follow(pte));
+
+	return 0;
+}
+
+void kvm_pgtable_stage2_destroy(struct kvm_pgtable *pgt)
+{
+	size_t pgd_sz;
+	struct kvm_pgtable_walker walker = {
+		.cb	= stage2_free_walker,
+		.flags	= KVM_PGTABLE_WALK_LEAF |
+			  KVM_PGTABLE_WALK_TABLE_POST,
+	};
+
+	WARN_ON(kvm_pgtable_walk(pgt, 0, BIT(pgt->ia_bits), &walker));
+	pgd_sz = kvm_pgd_pages(pgt->ia_bits, pgt->start_level) * PAGE_SIZE;
+	free_pages_exact(pgt->pgd, pgd_sz);
+	pgt->pgd = NULL;
+}
diff --git a/arch/arm64/kvm/hyp/vhe/switch.c b/arch/arm64/kvm/hyp/vhe/switch.c
index c1da4f86ccac..fe69de16dadc 100644
--- a/arch/arm64/kvm/hyp/vhe/switch.c
+++ b/arch/arm64/kvm/hyp/vhe/switch.c
@@ -28,6 +28,11 @@
 
 const char __hyp_panic_string[] = "HYP panic:\nPS:%08llx PC:%016llx ESR:%08llx\nFAR:%016llx HPFAR:%016llx PAR:%016llx\nVCPU:%p\n";
 
+/* VHE specific context */
+DEFINE_PER_CPU(struct kvm_host_data, kvm_host_data);
+DEFINE_PER_CPU(struct kvm_cpu_context, kvm_hyp_ctxt);
+DEFINE_PER_CPU(unsigned long, kvm_hyp_vector);
+
 static void __activate_traps(struct kvm_vcpu *vcpu)
 {
 	u64 val;
@@ -59,7 +64,7 @@ static void __activate_traps(struct kvm_vcpu *vcpu)
 
 	write_sysreg(val, cpacr_el1);
 
-	write_sysreg(kvm_get_hyp_vector(), vbar_el1);
+	write_sysreg(__this_cpu_read(kvm_hyp_vector), vbar_el1);
 }
 NOKPROBE_SYMBOL(__activate_traps);
 
@@ -108,7 +113,7 @@ static int __kvm_vcpu_run_vhe(struct kvm_vcpu *vcpu)
 	struct kvm_cpu_context *guest_ctxt;
 	u64 exit_code;
 
-	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
 	host_ctxt->__hyp_running_vcpu = vcpu;
 	guest_ctxt = &vcpu->arch.ctxt;
 
@@ -120,28 +125,24 @@ static int __kvm_vcpu_run_vhe(struct kvm_vcpu *vcpu)
 	 * HCR_EL2.TGE.
 	 *
 	 * We have already configured the guest's stage 1 translation in
-	 * kvm_vcpu_load_sysregs_vhe above.  We must now call __activate_vm
-	 * before __activate_traps, because __activate_vm configures
-	 * stage 2 translation, and __activate_traps clear HCR_EL2.TGE
-	 * (among other things).
+	 * kvm_vcpu_load_sysregs_vhe above.  We must now call
+	 * __load_guest_stage2 before __activate_traps, because
+	 * __load_guest_stage2 configures stage 2 translation, and
+	 * __activate_traps clear HCR_EL2.TGE (among other things).
 	 */
-	__activate_vm(vcpu->arch.hw_mmu);
+	__load_guest_stage2(vcpu->arch.hw_mmu);
 	__activate_traps(vcpu);
 
 	sysreg_restore_guest_state_vhe(guest_ctxt);
 	__debug_switch_to_guest(vcpu);
 
-	__set_guest_arch_workaround_state(vcpu);
-
 	do {
 		/* Jump in the fire! */
-		exit_code = __guest_enter(vcpu, host_ctxt);
+		exit_code = __guest_enter(vcpu);
 
 		/* And we're baaack! */
 	} while (fixup_guest_exit(vcpu, &exit_code));
 
-	__set_host_arch_workaround_state(vcpu);
-
 	sysreg_save_guest_state_vhe(guest_ctxt);
 
 	__deactivate_traps(vcpu);
@@ -192,10 +193,12 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	return ret;
 }
 
-static void __hyp_call_panic(u64 spsr, u64 elr, u64 par,
-			     struct kvm_cpu_context *host_ctxt)
+static void __hyp_call_panic(u64 spsr, u64 elr, u64 par)
 {
+	struct kvm_cpu_context *host_ctxt;
 	struct kvm_vcpu *vcpu;
+
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
 	vcpu = host_ctxt->__hyp_running_vcpu;
 
 	__deactivate_traps(vcpu);
@@ -208,13 +211,13 @@ static void __hyp_call_panic(u64 spsr, u64 elr, u64 par,
 }
 NOKPROBE_SYMBOL(__hyp_call_panic);
 
-void __noreturn hyp_panic(struct kvm_cpu_context *host_ctxt)
+void __noreturn hyp_panic(void)
 {
 	u64 spsr = read_sysreg_el2(SYS_SPSR);
 	u64 elr = read_sysreg_el2(SYS_ELR);
 	u64 par = read_sysreg(par_el1);
 
-	__hyp_call_panic(spsr, elr, par, host_ctxt);
+	__hyp_call_panic(spsr, elr, par);
 	unreachable();
 }
 
diff --git a/arch/arm64/kvm/hyp/vhe/sysreg-sr.c b/arch/arm64/kvm/hyp/vhe/sysreg-sr.c
index 996471e4c138..2a0b8c88d74f 100644
--- a/arch/arm64/kvm/hyp/vhe/sysreg-sr.c
+++ b/arch/arm64/kvm/hyp/vhe/sysreg-sr.c
@@ -66,7 +66,7 @@ void kvm_vcpu_load_sysregs_vhe(struct kvm_vcpu *vcpu)
 	struct kvm_cpu_context *guest_ctxt = &vcpu->arch.ctxt;
 	struct kvm_cpu_context *host_ctxt;
 
-	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
 	__sysreg_save_user_state(host_ctxt);
 
 	/*
@@ -100,7 +100,7 @@ void kvm_vcpu_put_sysregs_vhe(struct kvm_vcpu *vcpu)
 	struct kvm_cpu_context *guest_ctxt = &vcpu->arch.ctxt;
 	struct kvm_cpu_context *host_ctxt;
 
-	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
 	deactivate_traps_vhe_put();
 
 	__sysreg_save_el1_state(guest_ctxt);
diff --git a/arch/arm64/kvm/hypercalls.c b/arch/arm64/kvm/hypercalls.c
index 550dfa3e53cd..9824025ccc5c 100644
--- a/arch/arm64/kvm/hypercalls.c
+++ b/arch/arm64/kvm/hypercalls.c
@@ -24,27 +24,36 @@ int kvm_hvc_call_handler(struct kvm_vcpu *vcpu)
 		feature = smccc_get_arg1(vcpu);
 		switch (feature) {
 		case ARM_SMCCC_ARCH_WORKAROUND_1:
-			switch (kvm_arm_harden_branch_predictor()) {
-			case KVM_BP_HARDEN_UNKNOWN:
+			switch (arm64_get_spectre_v2_state()) {
+			case SPECTRE_VULNERABLE:
 				break;
-			case KVM_BP_HARDEN_WA_NEEDED:
+			case SPECTRE_MITIGATED:
 				val = SMCCC_RET_SUCCESS;
 				break;
-			case KVM_BP_HARDEN_NOT_REQUIRED:
+			case SPECTRE_UNAFFECTED:
 				val = SMCCC_RET_NOT_REQUIRED;
 				break;
 			}
 			break;
 		case ARM_SMCCC_ARCH_WORKAROUND_2:
-			switch (kvm_arm_have_ssbd()) {
-			case KVM_SSBD_FORCE_DISABLE:
-			case KVM_SSBD_UNKNOWN:
+			switch (arm64_get_spectre_v4_state()) {
+			case SPECTRE_VULNERABLE:
 				break;
-			case KVM_SSBD_KERNEL:
-				val = SMCCC_RET_SUCCESS;
-				break;
-			case KVM_SSBD_FORCE_ENABLE:
-			case KVM_SSBD_MITIGATED:
+			case SPECTRE_MITIGATED:
+				/*
+				 * SSBS everywhere: Indicate no firmware
+				 * support, as the SSBS support will be
+				 * indicated to the guest and the default is
+				 * safe.
+				 *
+				 * Otherwise, expose a permanent mitigation
+				 * to the guest, and hide SSBS so that the
+				 * guest stays protected.
+				 */
+				if (cpus_have_final_cap(ARM64_SSBS))
+					break;
+				fallthrough;
+			case SPECTRE_UNAFFECTED:
 				val = SMCCC_RET_NOT_REQUIRED;
 				break;
 			}
diff --git a/arch/arm64/kvm/inject_fault.c b/arch/arm64/kvm/inject_fault.c
index ebfdfc27b2bd..34a96ab244fa 100644
--- a/arch/arm64/kvm/inject_fault.c
+++ b/arch/arm64/kvm/inject_fault.c
@@ -202,6 +202,7 @@ void kvm_inject_pabt(struct kvm_vcpu *vcpu, unsigned long addr)
 
 /**
  * kvm_inject_undefined - inject an undefined instruction into the guest
+ * @vcpu: The vCPU in which to inject the exception
  *
  * It is assumed that this code is called from the VCPU thread and that the
  * VCPU therefore is not currently executing guest code.
diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index 3d26b47a1343..19aacc7d64de 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -14,6 +14,7 @@
 #include <asm/cacheflush.h>
 #include <asm/kvm_arm.h>
 #include <asm/kvm_mmu.h>
+#include <asm/kvm_pgtable.h>
 #include <asm/kvm_ras.h>
 #include <asm/kvm_asm.h>
 #include <asm/kvm_emulate.h>
@@ -21,9 +22,7 @@
 
 #include "trace.h"
 
-static pgd_t *boot_hyp_pgd;
-static pgd_t *hyp_pgd;
-static pgd_t *merged_hyp_pgd;
+static struct kvm_pgtable *hyp_pgtable;
 static DEFINE_MUTEX(kvm_hyp_pgd_mutex);
 
 static unsigned long hyp_idmap_start;
@@ -32,16 +31,42 @@ static phys_addr_t hyp_idmap_vector;
 
 static unsigned long io_map_base;
 
-#define hyp_pgd_order get_order(PTRS_PER_PGD * sizeof(pgd_t))
 
-#define KVM_S2PTE_FLAG_IS_IOMAP		(1UL << 0)
-#define KVM_S2_FLAG_LOGGING_ACTIVE	(1UL << 1)
-
-static bool is_iomap(unsigned long flags)
+/*
+ * Release kvm_mmu_lock periodically if the memory region is large. Otherwise,
+ * we may see kernel panics with CONFIG_DETECT_HUNG_TASK,
+ * CONFIG_LOCKUP_DETECTOR, CONFIG_LOCKDEP. Additionally, holding the lock too
+ * long will also starve other vCPUs. We have to also make sure that the page
+ * tables are not freed while we released the lock.
+ */
+static int stage2_apply_range(struct kvm *kvm, phys_addr_t addr,
+			      phys_addr_t end,
+			      int (*fn)(struct kvm_pgtable *, u64, u64),
+			      bool resched)
 {
-	return flags & KVM_S2PTE_FLAG_IS_IOMAP;
+	int ret;
+	u64 next;
+
+	do {
+		struct kvm_pgtable *pgt = kvm->arch.mmu.pgt;
+		if (!pgt)
+			return -EINVAL;
+
+		next = stage2_pgd_addr_end(kvm, addr, end);
+		ret = fn(pgt, addr, next - addr);
+		if (ret)
+			break;
+
+		if (resched && next != end)
+			cond_resched_lock(&kvm->mmu_lock);
+	} while (addr = next, addr != end);
+
+	return ret;
 }
 
+#define stage2_apply_range_resched(kvm, addr, end, fn)			\
+	stage2_apply_range(kvm, addr, end, fn, true)
+
 static bool memslot_is_logging(struct kvm_memory_slot *memslot)
 {
 	return memslot->dirty_bitmap && !(memslot->flags & KVM_MEM_READONLY);
@@ -58,154 +83,11 @@ void kvm_flush_remote_tlbs(struct kvm *kvm)
 	kvm_call_hyp(__kvm_tlb_flush_vmid, &kvm->arch.mmu);
 }
 
-static void kvm_tlb_flush_vmid_ipa(struct kvm_s2_mmu *mmu, phys_addr_t ipa,
-				   int level)
-{
-	kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, mmu, ipa, level);
-}
-
-/*
- * D-Cache management functions. They take the page table entries by
- * value, as they are flushing the cache using the kernel mapping (or
- * kmap on 32bit).
- */
-static void kvm_flush_dcache_pte(pte_t pte)
-{
-	__kvm_flush_dcache_pte(pte);
-}
-
-static void kvm_flush_dcache_pmd(pmd_t pmd)
-{
-	__kvm_flush_dcache_pmd(pmd);
-}
-
-static void kvm_flush_dcache_pud(pud_t pud)
-{
-	__kvm_flush_dcache_pud(pud);
-}
-
 static bool kvm_is_device_pfn(unsigned long pfn)
 {
 	return !pfn_valid(pfn);
 }
 
-/**
- * stage2_dissolve_pmd() - clear and flush huge PMD entry
- * @mmu:	pointer to mmu structure to operate on
- * @addr:	IPA
- * @pmd:	pmd pointer for IPA
- *
- * Function clears a PMD entry, flushes addr 1st and 2nd stage TLBs.
- */
-static void stage2_dissolve_pmd(struct kvm_s2_mmu *mmu, phys_addr_t addr, pmd_t *pmd)
-{
-	if (!pmd_thp_or_huge(*pmd))
-		return;
-
-	pmd_clear(pmd);
-	kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PMD_LEVEL);
-	put_page(virt_to_page(pmd));
-}
-
-/**
- * stage2_dissolve_pud() - clear and flush huge PUD entry
- * @mmu:	pointer to mmu structure to operate on
- * @addr:	IPA
- * @pud:	pud pointer for IPA
- *
- * Function clears a PUD entry, flushes addr 1st and 2nd stage TLBs.
- */
-static void stage2_dissolve_pud(struct kvm_s2_mmu *mmu, phys_addr_t addr, pud_t *pudp)
-{
-	struct kvm *kvm = mmu->kvm;
-
-	if (!stage2_pud_huge(kvm, *pudp))
-		return;
-
-	stage2_pud_clear(kvm, pudp);
-	kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PUD_LEVEL);
-	put_page(virt_to_page(pudp));
-}
-
-static void clear_stage2_pgd_entry(struct kvm_s2_mmu *mmu, pgd_t *pgd, phys_addr_t addr)
-{
-	struct kvm *kvm = mmu->kvm;
-	p4d_t *p4d_table __maybe_unused = stage2_p4d_offset(kvm, pgd, 0UL);
-	stage2_pgd_clear(kvm, pgd);
-	kvm_tlb_flush_vmid_ipa(mmu, addr, S2_NO_LEVEL_HINT);
-	stage2_p4d_free(kvm, p4d_table);
-	put_page(virt_to_page(pgd));
-}
-
-static void clear_stage2_p4d_entry(struct kvm_s2_mmu *mmu, p4d_t *p4d, phys_addr_t addr)
-{
-	struct kvm *kvm = mmu->kvm;
-	pud_t *pud_table __maybe_unused = stage2_pud_offset(kvm, p4d, 0);
-	stage2_p4d_clear(kvm, p4d);
-	kvm_tlb_flush_vmid_ipa(mmu, addr, S2_NO_LEVEL_HINT);
-	stage2_pud_free(kvm, pud_table);
-	put_page(virt_to_page(p4d));
-}
-
-static void clear_stage2_pud_entry(struct kvm_s2_mmu *mmu, pud_t *pud, phys_addr_t addr)
-{
-	struct kvm *kvm = mmu->kvm;
-	pmd_t *pmd_table __maybe_unused = stage2_pmd_offset(kvm, pud, 0);
-
-	VM_BUG_ON(stage2_pud_huge(kvm, *pud));
-	stage2_pud_clear(kvm, pud);
-	kvm_tlb_flush_vmid_ipa(mmu, addr, S2_NO_LEVEL_HINT);
-	stage2_pmd_free(kvm, pmd_table);
-	put_page(virt_to_page(pud));
-}
-
-static void clear_stage2_pmd_entry(struct kvm_s2_mmu *mmu, pmd_t *pmd, phys_addr_t addr)
-{
-	pte_t *pte_table = pte_offset_kernel(pmd, 0);
-	VM_BUG_ON(pmd_thp_or_huge(*pmd));
-	pmd_clear(pmd);
-	kvm_tlb_flush_vmid_ipa(mmu, addr, S2_NO_LEVEL_HINT);
-	free_page((unsigned long)pte_table);
-	put_page(virt_to_page(pmd));
-}
-
-static inline void kvm_set_pte(pte_t *ptep, pte_t new_pte)
-{
-	WRITE_ONCE(*ptep, new_pte);
-	dsb(ishst);
-}
-
-static inline void kvm_set_pmd(pmd_t *pmdp, pmd_t new_pmd)
-{
-	WRITE_ONCE(*pmdp, new_pmd);
-	dsb(ishst);
-}
-
-static inline void kvm_pmd_populate(pmd_t *pmdp, pte_t *ptep)
-{
-	kvm_set_pmd(pmdp, kvm_mk_pmd(ptep));
-}
-
-static inline void kvm_pud_populate(pud_t *pudp, pmd_t *pmdp)
-{
-	WRITE_ONCE(*pudp, kvm_mk_pud(pmdp));
-	dsb(ishst);
-}
-
-static inline void kvm_p4d_populate(p4d_t *p4dp, pud_t *pudp)
-{
-	WRITE_ONCE(*p4dp, kvm_mk_p4d(pudp));
-	dsb(ishst);
-}
-
-static inline void kvm_pgd_populate(pgd_t *pgdp, p4d_t *p4dp)
-{
-#ifndef __PAGETABLE_P4D_FOLDED
-	WRITE_ONCE(*pgdp, kvm_mk_pgd(p4dp));
-	dsb(ishst);
-#endif
-}
-
 /*
  * Unmapping vs dcache management:
  *
@@ -223,120 +105,19 @@ static inline void kvm_pgd_populate(pgd_t *pgdp, p4d_t *p4dp)
  * end up writing old data to disk.
  *
  * This is why right after unmapping a page/section and invalidating
- * the corresponding TLBs, we call kvm_flush_dcache_p*() to make sure
- * the IO subsystem will never hit in the cache.
+ * the corresponding TLBs, we flush to make sure the IO subsystem will
+ * never hit in the cache.
  *
  * This is all avoided on systems that have ARM64_HAS_STAGE2_FWB, as
  * we then fully enforce cacheability of RAM, no matter what the guest
  * does.
  */
-static void unmap_stage2_ptes(struct kvm_s2_mmu *mmu, pmd_t *pmd,
-		       phys_addr_t addr, phys_addr_t end)
-{
-	phys_addr_t start_addr = addr;
-	pte_t *pte, *start_pte;
-
-	start_pte = pte = pte_offset_kernel(pmd, addr);
-	do {
-		if (!pte_none(*pte)) {
-			pte_t old_pte = *pte;
-
-			kvm_set_pte(pte, __pte(0));
-			kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PTE_LEVEL);
-
-			/* No need to invalidate the cache for device mappings */
-			if (!kvm_is_device_pfn(pte_pfn(old_pte)))
-				kvm_flush_dcache_pte(old_pte);
-
-			put_page(virt_to_page(pte));
-		}
-	} while (pte++, addr += PAGE_SIZE, addr != end);
-
-	if (stage2_pte_table_empty(mmu->kvm, start_pte))
-		clear_stage2_pmd_entry(mmu, pmd, start_addr);
-}
-
-static void unmap_stage2_pmds(struct kvm_s2_mmu *mmu, pud_t *pud,
-		       phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	phys_addr_t next, start_addr = addr;
-	pmd_t *pmd, *start_pmd;
-
-	start_pmd = pmd = stage2_pmd_offset(kvm, pud, addr);
-	do {
-		next = stage2_pmd_addr_end(kvm, addr, end);
-		if (!pmd_none(*pmd)) {
-			if (pmd_thp_or_huge(*pmd)) {
-				pmd_t old_pmd = *pmd;
-
-				pmd_clear(pmd);
-				kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PMD_LEVEL);
-
-				kvm_flush_dcache_pmd(old_pmd);
-
-				put_page(virt_to_page(pmd));
-			} else {
-				unmap_stage2_ptes(mmu, pmd, addr, next);
-			}
-		}
-	} while (pmd++, addr = next, addr != end);
-
-	if (stage2_pmd_table_empty(kvm, start_pmd))
-		clear_stage2_pud_entry(mmu, pud, start_addr);
-}
-
-static void unmap_stage2_puds(struct kvm_s2_mmu *mmu, p4d_t *p4d,
-		       phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	phys_addr_t next, start_addr = addr;
-	pud_t *pud, *start_pud;
-
-	start_pud = pud = stage2_pud_offset(kvm, p4d, addr);
-	do {
-		next = stage2_pud_addr_end(kvm, addr, end);
-		if (!stage2_pud_none(kvm, *pud)) {
-			if (stage2_pud_huge(kvm, *pud)) {
-				pud_t old_pud = *pud;
-
-				stage2_pud_clear(kvm, pud);
-				kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PUD_LEVEL);
-				kvm_flush_dcache_pud(old_pud);
-				put_page(virt_to_page(pud));
-			} else {
-				unmap_stage2_pmds(mmu, pud, addr, next);
-			}
-		}
-	} while (pud++, addr = next, addr != end);
-
-	if (stage2_pud_table_empty(kvm, start_pud))
-		clear_stage2_p4d_entry(mmu, p4d, start_addr);
-}
-
-static void unmap_stage2_p4ds(struct kvm_s2_mmu *mmu, pgd_t *pgd,
-		       phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	phys_addr_t next, start_addr = addr;
-	p4d_t *p4d, *start_p4d;
-
-	start_p4d = p4d = stage2_p4d_offset(kvm, pgd, addr);
-	do {
-		next = stage2_p4d_addr_end(kvm, addr, end);
-		if (!stage2_p4d_none(kvm, *p4d))
-			unmap_stage2_puds(mmu, p4d, addr, next);
-	} while (p4d++, addr = next, addr != end);
-
-	if (stage2_p4d_table_empty(kvm, start_p4d))
-		clear_stage2_pgd_entry(mmu, pgd, start_addr);
-}
-
 /**
  * unmap_stage2_range -- Clear stage2 page table entries to unmap a range
- * @kvm:   The VM pointer
+ * @mmu:   The KVM stage-2 MMU pointer
  * @start: The intermediate physical base address of the range to unmap
  * @size:  The size of the area to unmap
+ * @may_block: Whether or not we are permitted to block
  *
  * Clear a range of stage-2 mappings, lowering the various ref-counts.  Must
  * be called while holding mmu_lock (unless for freeing the stage2 pgd before
@@ -347,32 +128,12 @@ static void __unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64
 				 bool may_block)
 {
 	struct kvm *kvm = mmu->kvm;
-	pgd_t *pgd;
-	phys_addr_t addr = start, end = start + size;
-	phys_addr_t next;
+	phys_addr_t end = start + size;
 
 	assert_spin_locked(&kvm->mmu_lock);
 	WARN_ON(size & ~PAGE_MASK);
-
-	pgd = mmu->pgd + stage2_pgd_index(kvm, addr);
-	do {
-		/*
-		 * Make sure the page table is still active, as another thread
-		 * could have possibly freed the page table, while we released
-		 * the lock.
-		 */
-		if (!READ_ONCE(mmu->pgd))
-			break;
-		next = stage2_pgd_addr_end(kvm, addr, end);
-		if (!stage2_pgd_none(kvm, *pgd))
-			unmap_stage2_p4ds(mmu, pgd, addr, next);
-		/*
-		 * If the range is too large, release the kvm->mmu_lock
-		 * to prevent starvation and lockup detector warnings.
-		 */
-		if (may_block && next != end)
-			cond_resched_lock(&kvm->mmu_lock);
-	} while (pgd++, addr = next, addr != end);
+	WARN_ON(stage2_apply_range(kvm, start, end, kvm_pgtable_stage2_unmap,
+				   may_block));
 }
 
 static void unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64 size)
@@ -380,89 +141,13 @@ static void unmap_stage2_range(struct kvm_s2_mmu *mmu, phys_addr_t start, u64 si
 	__unmap_stage2_range(mmu, start, size, true);
 }
 
-static void stage2_flush_ptes(struct kvm_s2_mmu *mmu, pmd_t *pmd,
-			      phys_addr_t addr, phys_addr_t end)
-{
-	pte_t *pte;
-
-	pte = pte_offset_kernel(pmd, addr);
-	do {
-		if (!pte_none(*pte) && !kvm_is_device_pfn(pte_pfn(*pte)))
-			kvm_flush_dcache_pte(*pte);
-	} while (pte++, addr += PAGE_SIZE, addr != end);
-}
-
-static void stage2_flush_pmds(struct kvm_s2_mmu *mmu, pud_t *pud,
-			      phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	pmd_t *pmd;
-	phys_addr_t next;
-
-	pmd = stage2_pmd_offset(kvm, pud, addr);
-	do {
-		next = stage2_pmd_addr_end(kvm, addr, end);
-		if (!pmd_none(*pmd)) {
-			if (pmd_thp_or_huge(*pmd))
-				kvm_flush_dcache_pmd(*pmd);
-			else
-				stage2_flush_ptes(mmu, pmd, addr, next);
-		}
-	} while (pmd++, addr = next, addr != end);
-}
-
-static void stage2_flush_puds(struct kvm_s2_mmu *mmu, p4d_t *p4d,
-			      phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	pud_t *pud;
-	phys_addr_t next;
-
-	pud = stage2_pud_offset(kvm, p4d, addr);
-	do {
-		next = stage2_pud_addr_end(kvm, addr, end);
-		if (!stage2_pud_none(kvm, *pud)) {
-			if (stage2_pud_huge(kvm, *pud))
-				kvm_flush_dcache_pud(*pud);
-			else
-				stage2_flush_pmds(mmu, pud, addr, next);
-		}
-	} while (pud++, addr = next, addr != end);
-}
-
-static void stage2_flush_p4ds(struct kvm_s2_mmu *mmu, pgd_t *pgd,
-			      phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	p4d_t *p4d;
-	phys_addr_t next;
-
-	p4d = stage2_p4d_offset(kvm, pgd, addr);
-	do {
-		next = stage2_p4d_addr_end(kvm, addr, end);
-		if (!stage2_p4d_none(kvm, *p4d))
-			stage2_flush_puds(mmu, p4d, addr, next);
-	} while (p4d++, addr = next, addr != end);
-}
-
 static void stage2_flush_memslot(struct kvm *kvm,
 				 struct kvm_memory_slot *memslot)
 {
-	struct kvm_s2_mmu *mmu = &kvm->arch.mmu;
 	phys_addr_t addr = memslot->base_gfn << PAGE_SHIFT;
 	phys_addr_t end = addr + PAGE_SIZE * memslot->npages;
-	phys_addr_t next;
-	pgd_t *pgd;
 
-	pgd = mmu->pgd + stage2_pgd_index(kvm, addr);
-	do {
-		next = stage2_pgd_addr_end(kvm, addr, end);
-		if (!stage2_pgd_none(kvm, *pgd))
-			stage2_flush_p4ds(mmu, pgd, addr, next);
-
-		if (next != end)
-			cond_resched_lock(&kvm->mmu_lock);
-	} while (pgd++, addr = next, addr != end);
+	stage2_apply_range_resched(kvm, addr, end, kvm_pgtable_stage2_flush);
 }
 
 /**
@@ -489,338 +174,28 @@ static void stage2_flush_vm(struct kvm *kvm)
 	srcu_read_unlock(&kvm->srcu, idx);
 }
 
-static void clear_hyp_pgd_entry(pgd_t *pgd)
-{
-	p4d_t *p4d_table __maybe_unused = p4d_offset(pgd, 0UL);
-	pgd_clear(pgd);
-	p4d_free(NULL, p4d_table);
-	put_page(virt_to_page(pgd));
-}
-
-static void clear_hyp_p4d_entry(p4d_t *p4d)
-{
-	pud_t *pud_table __maybe_unused = pud_offset(p4d, 0UL);
-	VM_BUG_ON(p4d_huge(*p4d));
-	p4d_clear(p4d);
-	pud_free(NULL, pud_table);
-	put_page(virt_to_page(p4d));
-}
-
-static void clear_hyp_pud_entry(pud_t *pud)
-{
-	pmd_t *pmd_table __maybe_unused = pmd_offset(pud, 0);
-	VM_BUG_ON(pud_huge(*pud));
-	pud_clear(pud);
-	pmd_free(NULL, pmd_table);
-	put_page(virt_to_page(pud));
-}
-
-static void clear_hyp_pmd_entry(pmd_t *pmd)
-{
-	pte_t *pte_table = pte_offset_kernel(pmd, 0);
-	VM_BUG_ON(pmd_thp_or_huge(*pmd));
-	pmd_clear(pmd);
-	pte_free_kernel(NULL, pte_table);
-	put_page(virt_to_page(pmd));
-}
-
-static void unmap_hyp_ptes(pmd_t *pmd, phys_addr_t addr, phys_addr_t end)
-{
-	pte_t *pte, *start_pte;
-
-	start_pte = pte = pte_offset_kernel(pmd, addr);
-	do {
-		if (!pte_none(*pte)) {
-			kvm_set_pte(pte, __pte(0));
-			put_page(virt_to_page(pte));
-		}
-	} while (pte++, addr += PAGE_SIZE, addr != end);
-
-	if (hyp_pte_table_empty(start_pte))
-		clear_hyp_pmd_entry(pmd);
-}
-
-static void unmap_hyp_pmds(pud_t *pud, phys_addr_t addr, phys_addr_t end)
-{
-	phys_addr_t next;
-	pmd_t *pmd, *start_pmd;
-
-	start_pmd = pmd = pmd_offset(pud, addr);
-	do {
-		next = pmd_addr_end(addr, end);
-		/* Hyp doesn't use huge pmds */
-		if (!pmd_none(*pmd))
-			unmap_hyp_ptes(pmd, addr, next);
-	} while (pmd++, addr = next, addr != end);
-
-	if (hyp_pmd_table_empty(start_pmd))
-		clear_hyp_pud_entry(pud);
-}
-
-static void unmap_hyp_puds(p4d_t *p4d, phys_addr_t addr, phys_addr_t end)
-{
-	phys_addr_t next;
-	pud_t *pud, *start_pud;
-
-	start_pud = pud = pud_offset(p4d, addr);
-	do {
-		next = pud_addr_end(addr, end);
-		/* Hyp doesn't use huge puds */
-		if (!pud_none(*pud))
-			unmap_hyp_pmds(pud, addr, next);
-	} while (pud++, addr = next, addr != end);
-
-	if (hyp_pud_table_empty(start_pud))
-		clear_hyp_p4d_entry(p4d);
-}
-
-static void unmap_hyp_p4ds(pgd_t *pgd, phys_addr_t addr, phys_addr_t end)
-{
-	phys_addr_t next;
-	p4d_t *p4d, *start_p4d;
-
-	start_p4d = p4d = p4d_offset(pgd, addr);
-	do {
-		next = p4d_addr_end(addr, end);
-		/* Hyp doesn't use huge p4ds */
-		if (!p4d_none(*p4d))
-			unmap_hyp_puds(p4d, addr, next);
-	} while (p4d++, addr = next, addr != end);
-
-	if (hyp_p4d_table_empty(start_p4d))
-		clear_hyp_pgd_entry(pgd);
-}
-
-static unsigned int kvm_pgd_index(unsigned long addr, unsigned int ptrs_per_pgd)
-{
-	return (addr >> PGDIR_SHIFT) & (ptrs_per_pgd - 1);
-}
-
-static void __unmap_hyp_range(pgd_t *pgdp, unsigned long ptrs_per_pgd,
-			      phys_addr_t start, u64 size)
-{
-	pgd_t *pgd;
-	phys_addr_t addr = start, end = start + size;
-	phys_addr_t next;
-
-	/*
-	 * We don't unmap anything from HYP, except at the hyp tear down.
-	 * Hence, we don't have to invalidate the TLBs here.
-	 */
-	pgd = pgdp + kvm_pgd_index(addr, ptrs_per_pgd);
-	do {
-		next = pgd_addr_end(addr, end);
-		if (!pgd_none(*pgd))
-			unmap_hyp_p4ds(pgd, addr, next);
-	} while (pgd++, addr = next, addr != end);
-}
-
-static void unmap_hyp_range(pgd_t *pgdp, phys_addr_t start, u64 size)
-{
-	__unmap_hyp_range(pgdp, PTRS_PER_PGD, start, size);
-}
-
-static void unmap_hyp_idmap_range(pgd_t *pgdp, phys_addr_t start, u64 size)
-{
-	__unmap_hyp_range(pgdp, __kvm_idmap_ptrs_per_pgd(), start, size);
-}
-
 /**
  * free_hyp_pgds - free Hyp-mode page tables
- *
- * Assumes hyp_pgd is a page table used strictly in Hyp-mode and
- * therefore contains either mappings in the kernel memory area (above
- * PAGE_OFFSET), or device mappings in the idmap range.
- *
- * boot_hyp_pgd should only map the idmap range, and is only used in
- * the extended idmap case.
  */
 void free_hyp_pgds(void)
 {
-	pgd_t *id_pgd;
-
 	mutex_lock(&kvm_hyp_pgd_mutex);
-
-	id_pgd = boot_hyp_pgd ? boot_hyp_pgd : hyp_pgd;
-
-	if (id_pgd) {
-		/* In case we never called hyp_mmu_init() */
-		if (!io_map_base)
-			io_map_base = hyp_idmap_start;
-		unmap_hyp_idmap_range(id_pgd, io_map_base,
-				      hyp_idmap_start + PAGE_SIZE - io_map_base);
+	if (hyp_pgtable) {
+		kvm_pgtable_hyp_destroy(hyp_pgtable);
+		kfree(hyp_pgtable);
 	}
-
-	if (boot_hyp_pgd) {
-		free_pages((unsigned long)boot_hyp_pgd, hyp_pgd_order);
-		boot_hyp_pgd = NULL;
-	}
-
-	if (hyp_pgd) {
-		unmap_hyp_range(hyp_pgd, kern_hyp_va(PAGE_OFFSET),
-				(uintptr_t)high_memory - PAGE_OFFSET);
-
-		free_pages((unsigned long)hyp_pgd, hyp_pgd_order);
-		hyp_pgd = NULL;
-	}
-	if (merged_hyp_pgd) {
-		clear_page(merged_hyp_pgd);
-		free_page((unsigned long)merged_hyp_pgd);
-		merged_hyp_pgd = NULL;
-	}
-
 	mutex_unlock(&kvm_hyp_pgd_mutex);
 }
 
-static void create_hyp_pte_mappings(pmd_t *pmd, unsigned long start,
-				    unsigned long end, unsigned long pfn,
-				    pgprot_t prot)
-{
-	pte_t *pte;
-	unsigned long addr;
-
-	addr = start;
-	do {
-		pte = pte_offset_kernel(pmd, addr);
-		kvm_set_pte(pte, kvm_pfn_pte(pfn, prot));
-		get_page(virt_to_page(pte));
-		pfn++;
-	} while (addr += PAGE_SIZE, addr != end);
-}
-
-static int create_hyp_pmd_mappings(pud_t *pud, unsigned long start,
-				   unsigned long end, unsigned long pfn,
-				   pgprot_t prot)
-{
-	pmd_t *pmd;
-	pte_t *pte;
-	unsigned long addr, next;
-
-	addr = start;
-	do {
-		pmd = pmd_offset(pud, addr);
-
-		BUG_ON(pmd_sect(*pmd));
-
-		if (pmd_none(*pmd)) {
-			pte = pte_alloc_one_kernel(NULL);
-			if (!pte) {
-				kvm_err("Cannot allocate Hyp pte\n");
-				return -ENOMEM;
-			}
-			kvm_pmd_populate(pmd, pte);
-			get_page(virt_to_page(pmd));
-		}
-
-		next = pmd_addr_end(addr, end);
-
-		create_hyp_pte_mappings(pmd, addr, next, pfn, prot);
-		pfn += (next - addr) >> PAGE_SHIFT;
-	} while (addr = next, addr != end);
-
-	return 0;
-}
-
-static int create_hyp_pud_mappings(p4d_t *p4d, unsigned long start,
-				   unsigned long end, unsigned long pfn,
-				   pgprot_t prot)
-{
-	pud_t *pud;
-	pmd_t *pmd;
-	unsigned long addr, next;
-	int ret;
-
-	addr = start;
-	do {
-		pud = pud_offset(p4d, addr);
-
-		if (pud_none_or_clear_bad(pud)) {
-			pmd = pmd_alloc_one(NULL, addr);
-			if (!pmd) {
-				kvm_err("Cannot allocate Hyp pmd\n");
-				return -ENOMEM;
-			}
-			kvm_pud_populate(pud, pmd);
-			get_page(virt_to_page(pud));
-		}
-
-		next = pud_addr_end(addr, end);
-		ret = create_hyp_pmd_mappings(pud, addr, next, pfn, prot);
-		if (ret)
-			return ret;
-		pfn += (next - addr) >> PAGE_SHIFT;
-	} while (addr = next, addr != end);
-
-	return 0;
-}
-
-static int create_hyp_p4d_mappings(pgd_t *pgd, unsigned long start,
-				   unsigned long end, unsigned long pfn,
-				   pgprot_t prot)
+static int __create_hyp_mappings(unsigned long start, unsigned long size,
+				 unsigned long phys, enum kvm_pgtable_prot prot)
 {
-	p4d_t *p4d;
-	pud_t *pud;
-	unsigned long addr, next;
-	int ret;
-
-	addr = start;
-	do {
-		p4d = p4d_offset(pgd, addr);
-
-		if (p4d_none(*p4d)) {
-			pud = pud_alloc_one(NULL, addr);
-			if (!pud) {
-				kvm_err("Cannot allocate Hyp pud\n");
-				return -ENOMEM;
-			}
-			kvm_p4d_populate(p4d, pud);
-			get_page(virt_to_page(p4d));
-		}
-
-		next = p4d_addr_end(addr, end);
-		ret = create_hyp_pud_mappings(p4d, addr, next, pfn, prot);
-		if (ret)
-			return ret;
-		pfn += (next - addr) >> PAGE_SHIFT;
-	} while (addr = next, addr != end);
-
-	return 0;
-}
-
-static int __create_hyp_mappings(pgd_t *pgdp, unsigned long ptrs_per_pgd,
-				 unsigned long start, unsigned long end,
-				 unsigned long pfn, pgprot_t prot)
-{
-	pgd_t *pgd;
-	p4d_t *p4d;
-	unsigned long addr, next;
-	int err = 0;
+	int err;
 
 	mutex_lock(&kvm_hyp_pgd_mutex);
-	addr = start & PAGE_MASK;
-	end = PAGE_ALIGN(end);
-	do {
-		pgd = pgdp + kvm_pgd_index(addr, ptrs_per_pgd);
-
-		if (pgd_none(*pgd)) {
-			p4d = p4d_alloc_one(NULL, addr);
-			if (!p4d) {
-				kvm_err("Cannot allocate Hyp p4d\n");
-				err = -ENOMEM;
-				goto out;
-			}
-			kvm_pgd_populate(pgd, p4d);
-			get_page(virt_to_page(pgd));
-		}
-
-		next = pgd_addr_end(addr, end);
-		err = create_hyp_p4d_mappings(pgd, addr, next, pfn, prot);
-		if (err)
-			goto out;
-		pfn += (next - addr) >> PAGE_SHIFT;
-	} while (addr = next, addr != end);
-out:
+	err = kvm_pgtable_hyp_map(hyp_pgtable, start, size, phys, prot);
 	mutex_unlock(&kvm_hyp_pgd_mutex);
+
 	return err;
 }
 
@@ -845,7 +220,7 @@ static phys_addr_t kvm_kaddr_to_phys(void *kaddr)
  * in Hyp-mode mapping (modulo HYP_PAGE_OFFSET) to the same underlying
  * physical pages.
  */
-int create_hyp_mappings(void *from, void *to, pgprot_t prot)
+int create_hyp_mappings(void *from, void *to, enum kvm_pgtable_prot prot)
 {
 	phys_addr_t phys_addr;
 	unsigned long virt_addr;
@@ -862,9 +237,7 @@ int create_hyp_mappings(void *from, void *to, pgprot_t prot)
 		int err;
 
 		phys_addr = kvm_kaddr_to_phys(from + virt_addr - start);
-		err = __create_hyp_mappings(hyp_pgd, PTRS_PER_PGD,
-					    virt_addr, virt_addr + PAGE_SIZE,
-					    __phys_to_pfn(phys_addr),
+		err = __create_hyp_mappings(virt_addr, PAGE_SIZE, phys_addr,
 					    prot);
 		if (err)
 			return err;
@@ -874,9 +247,9 @@ int create_hyp_mappings(void *from, void *to, pgprot_t prot)
 }
 
 static int __create_hyp_private_mapping(phys_addr_t phys_addr, size_t size,
-					unsigned long *haddr, pgprot_t prot)
+					unsigned long *haddr,
+					enum kvm_pgtable_prot prot)
 {
-	pgd_t *pgd = hyp_pgd;
 	unsigned long base;
 	int ret = 0;
 
@@ -908,17 +281,11 @@ static int __create_hyp_private_mapping(phys_addr_t phys_addr, size_t size,
 	if (ret)
 		goto out;
 
-	if (__kvm_cpu_uses_extended_idmap())
-		pgd = boot_hyp_pgd;
-
-	ret = __create_hyp_mappings(pgd, __kvm_idmap_ptrs_per_pgd(),
-				    base, base + size,
-				    __phys_to_pfn(phys_addr), prot);
+	ret = __create_hyp_mappings(base, size, phys_addr, prot);
 	if (ret)
 		goto out;
 
 	*haddr = base + offset_in_page(phys_addr);
-
 out:
 	return ret;
 }
@@ -989,47 +356,48 @@ int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
  * @kvm:	The pointer to the KVM structure
  * @mmu:	The pointer to the s2 MMU structure
  *
- * Allocates only the stage-2 HW PGD level table(s) of size defined by
- * stage2_pgd_size(mmu->kvm).
- *
+ * Allocates only the stage-2 HW PGD level table(s).
  * Note we don't need locking here as this is only called when the VM is
  * created, which can only be done once.
  */
 int kvm_init_stage2_mmu(struct kvm *kvm, struct kvm_s2_mmu *mmu)
 {
-	phys_addr_t pgd_phys;
-	pgd_t *pgd;
-	int cpu;
+	int cpu, err;
+	struct kvm_pgtable *pgt;
 
-	if (mmu->pgd != NULL) {
+	if (mmu->pgt != NULL) {
 		kvm_err("kvm_arch already initialized?\n");
 		return -EINVAL;
 	}
 
-	/* Allocate the HW PGD, making sure that each page gets its own refcount */
-	pgd = alloc_pages_exact(stage2_pgd_size(kvm), GFP_KERNEL | __GFP_ZERO);
-	if (!pgd)
+	pgt = kzalloc(sizeof(*pgt), GFP_KERNEL);
+	if (!pgt)
 		return -ENOMEM;
 
-	pgd_phys = virt_to_phys(pgd);
-	if (WARN_ON(pgd_phys & ~kvm_vttbr_baddr_mask(kvm)))
-		return -EINVAL;
+	err = kvm_pgtable_stage2_init(pgt, kvm);
+	if (err)
+		goto out_free_pgtable;
 
 	mmu->last_vcpu_ran = alloc_percpu(typeof(*mmu->last_vcpu_ran));
 	if (!mmu->last_vcpu_ran) {
-		free_pages_exact(pgd, stage2_pgd_size(kvm));
-		return -ENOMEM;
+		err = -ENOMEM;
+		goto out_destroy_pgtable;
 	}
 
 	for_each_possible_cpu(cpu)
 		*per_cpu_ptr(mmu->last_vcpu_ran, cpu) = -1;
 
 	mmu->kvm = kvm;
-	mmu->pgd = pgd;
-	mmu->pgd_phys = pgd_phys;
+	mmu->pgt = pgt;
+	mmu->pgd_phys = __pa(pgt->pgd);
 	mmu->vmid.vmid_gen = 0;
-
 	return 0;
+
+out_destroy_pgtable:
+	kvm_pgtable_stage2_destroy(pgt);
+out_free_pgtable:
+	kfree(pgt);
+	return err;
 }
 
 static void stage2_unmap_memslot(struct kvm *kvm,
@@ -1102,363 +470,21 @@ void stage2_unmap_vm(struct kvm *kvm)
 void kvm_free_stage2_pgd(struct kvm_s2_mmu *mmu)
 {
 	struct kvm *kvm = mmu->kvm;
-	void *pgd = NULL;
+	struct kvm_pgtable *pgt = NULL;
 
 	spin_lock(&kvm->mmu_lock);
-	if (mmu->pgd) {
-		unmap_stage2_range(mmu, 0, kvm_phys_size(kvm));
-		pgd = READ_ONCE(mmu->pgd);
-		mmu->pgd = NULL;
-	}
-	spin_unlock(&kvm->mmu_lock);
-
-	/* Free the HW pgd, one page at a time */
-	if (pgd) {
-		free_pages_exact(pgd, stage2_pgd_size(kvm));
+	pgt = mmu->pgt;
+	if (pgt) {
+		mmu->pgd_phys = 0;
+		mmu->pgt = NULL;
 		free_percpu(mmu->last_vcpu_ran);
 	}
-}
-
-static p4d_t *stage2_get_p4d(struct kvm_s2_mmu *mmu, struct kvm_mmu_memory_cache *cache,
-			     phys_addr_t addr)
-{
-	struct kvm *kvm = mmu->kvm;
-	pgd_t *pgd;
-	p4d_t *p4d;
-
-	pgd = mmu->pgd + stage2_pgd_index(kvm, addr);
-	if (stage2_pgd_none(kvm, *pgd)) {
-		if (!cache)
-			return NULL;
-		p4d = kvm_mmu_memory_cache_alloc(cache);
-		stage2_pgd_populate(kvm, pgd, p4d);
-		get_page(virt_to_page(pgd));
-	}
-
-	return stage2_p4d_offset(kvm, pgd, addr);
-}
-
-static pud_t *stage2_get_pud(struct kvm_s2_mmu *mmu, struct kvm_mmu_memory_cache *cache,
-			     phys_addr_t addr)
-{
-	struct kvm *kvm = mmu->kvm;
-	p4d_t *p4d;
-	pud_t *pud;
-
-	p4d = stage2_get_p4d(mmu, cache, addr);
-	if (stage2_p4d_none(kvm, *p4d)) {
-		if (!cache)
-			return NULL;
-		pud = kvm_mmu_memory_cache_alloc(cache);
-		stage2_p4d_populate(kvm, p4d, pud);
-		get_page(virt_to_page(p4d));
-	}
-
-	return stage2_pud_offset(kvm, p4d, addr);
-}
-
-static pmd_t *stage2_get_pmd(struct kvm_s2_mmu *mmu, struct kvm_mmu_memory_cache *cache,
-			     phys_addr_t addr)
-{
-	struct kvm *kvm = mmu->kvm;
-	pud_t *pud;
-	pmd_t *pmd;
-
-	pud = stage2_get_pud(mmu, cache, addr);
-	if (!pud || stage2_pud_huge(kvm, *pud))
-		return NULL;
-
-	if (stage2_pud_none(kvm, *pud)) {
-		if (!cache)
-			return NULL;
-		pmd = kvm_mmu_memory_cache_alloc(cache);
-		stage2_pud_populate(kvm, pud, pmd);
-		get_page(virt_to_page(pud));
-	}
-
-	return stage2_pmd_offset(kvm, pud, addr);
-}
-
-static int stage2_set_pmd_huge(struct kvm_s2_mmu *mmu,
-			       struct kvm_mmu_memory_cache *cache,
-			       phys_addr_t addr, const pmd_t *new_pmd)
-{
-	pmd_t *pmd, old_pmd;
-
-retry:
-	pmd = stage2_get_pmd(mmu, cache, addr);
-	VM_BUG_ON(!pmd);
-
-	old_pmd = *pmd;
-	/*
-	 * Multiple vcpus faulting on the same PMD entry, can
-	 * lead to them sequentially updating the PMD with the
-	 * same value. Following the break-before-make
-	 * (pmd_clear() followed by tlb_flush()) process can
-	 * hinder forward progress due to refaults generated
-	 * on missing translations.
-	 *
-	 * Skip updating the page table if the entry is
-	 * unchanged.
-	 */
-	if (pmd_val(old_pmd) == pmd_val(*new_pmd))
-		return 0;
-
-	if (pmd_present(old_pmd)) {
-		/*
-		 * If we already have PTE level mapping for this block,
-		 * we must unmap it to avoid inconsistent TLB state and
-		 * leaking the table page. We could end up in this situation
-		 * if the memory slot was marked for dirty logging and was
-		 * reverted, leaving PTE level mappings for the pages accessed
-		 * during the period. So, unmap the PTE level mapping for this
-		 * block and retry, as we could have released the upper level
-		 * table in the process.
-		 *
-		 * Normal THP split/merge follows mmu_notifier callbacks and do
-		 * get handled accordingly.
-		 */
-		if (!pmd_thp_or_huge(old_pmd)) {
-			unmap_stage2_range(mmu, addr & S2_PMD_MASK, S2_PMD_SIZE);
-			goto retry;
-		}
-		/*
-		 * Mapping in huge pages should only happen through a
-		 * fault.  If a page is merged into a transparent huge
-		 * page, the individual subpages of that huge page
-		 * should be unmapped through MMU notifiers before we
-		 * get here.
-		 *
-		 * Merging of CompoundPages is not supported; they
-		 * should become splitting first, unmapped, merged,
-		 * and mapped back in on-demand.
-		 */
-		WARN_ON_ONCE(pmd_pfn(old_pmd) != pmd_pfn(*new_pmd));
-		pmd_clear(pmd);
-		kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PMD_LEVEL);
-	} else {
-		get_page(virt_to_page(pmd));
-	}
-
-	kvm_set_pmd(pmd, *new_pmd);
-	return 0;
-}
-
-static int stage2_set_pud_huge(struct kvm_s2_mmu *mmu,
-			       struct kvm_mmu_memory_cache *cache,
-			       phys_addr_t addr, const pud_t *new_pudp)
-{
-	struct kvm *kvm = mmu->kvm;
-	pud_t *pudp, old_pud;
-
-retry:
-	pudp = stage2_get_pud(mmu, cache, addr);
-	VM_BUG_ON(!pudp);
-
-	old_pud = *pudp;
-
-	/*
-	 * A large number of vcpus faulting on the same stage 2 entry,
-	 * can lead to a refault due to the stage2_pud_clear()/tlb_flush().
-	 * Skip updating the page tables if there is no change.
-	 */
-	if (pud_val(old_pud) == pud_val(*new_pudp))
-		return 0;
-
-	if (stage2_pud_present(kvm, old_pud)) {
-		/*
-		 * If we already have table level mapping for this block, unmap
-		 * the range for this block and retry.
-		 */
-		if (!stage2_pud_huge(kvm, old_pud)) {
-			unmap_stage2_range(mmu, addr & S2_PUD_MASK, S2_PUD_SIZE);
-			goto retry;
-		}
-
-		WARN_ON_ONCE(kvm_pud_pfn(old_pud) != kvm_pud_pfn(*new_pudp));
-		stage2_pud_clear(kvm, pudp);
-		kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PUD_LEVEL);
-	} else {
-		get_page(virt_to_page(pudp));
-	}
-
-	kvm_set_pud(pudp, *new_pudp);
-	return 0;
-}
-
-/*
- * stage2_get_leaf_entry - walk the stage2 VM page tables and return
- * true if a valid and present leaf-entry is found. A pointer to the
- * leaf-entry is returned in the appropriate level variable - pudpp,
- * pmdpp, ptepp.
- */
-static bool stage2_get_leaf_entry(struct kvm_s2_mmu *mmu, phys_addr_t addr,
-				  pud_t **pudpp, pmd_t **pmdpp, pte_t **ptepp)
-{
-	struct kvm *kvm = mmu->kvm;
-	pud_t *pudp;
-	pmd_t *pmdp;
-	pte_t *ptep;
-
-	*pudpp = NULL;
-	*pmdpp = NULL;
-	*ptepp = NULL;
-
-	pudp = stage2_get_pud(mmu, NULL, addr);
-	if (!pudp || stage2_pud_none(kvm, *pudp) || !stage2_pud_present(kvm, *pudp))
-		return false;
-
-	if (stage2_pud_huge(kvm, *pudp)) {
-		*pudpp = pudp;
-		return true;
-	}
-
-	pmdp = stage2_pmd_offset(kvm, pudp, addr);
-	if (!pmdp || pmd_none(*pmdp) || !pmd_present(*pmdp))
-		return false;
-
-	if (pmd_thp_or_huge(*pmdp)) {
-		*pmdpp = pmdp;
-		return true;
-	}
-
-	ptep = pte_offset_kernel(pmdp, addr);
-	if (!ptep || pte_none(*ptep) || !pte_present(*ptep))
-		return false;
-
-	*ptepp = ptep;
-	return true;
-}
-
-static bool stage2_is_exec(struct kvm_s2_mmu *mmu, phys_addr_t addr, unsigned long sz)
-{
-	pud_t *pudp;
-	pmd_t *pmdp;
-	pte_t *ptep;
-	bool found;
-
-	found = stage2_get_leaf_entry(mmu, addr, &pudp, &pmdp, &ptep);
-	if (!found)
-		return false;
-
-	if (pudp)
-		return sz <= PUD_SIZE && kvm_s2pud_exec(pudp);
-	else if (pmdp)
-		return sz <= PMD_SIZE && kvm_s2pmd_exec(pmdp);
-	else
-		return sz == PAGE_SIZE && kvm_s2pte_exec(ptep);
-}
-
-static int stage2_set_pte(struct kvm_s2_mmu *mmu,
-			  struct kvm_mmu_memory_cache *cache,
-			  phys_addr_t addr, const pte_t *new_pte,
-			  unsigned long flags)
-{
-	struct kvm *kvm = mmu->kvm;
-	pud_t *pud;
-	pmd_t *pmd;
-	pte_t *pte, old_pte;
-	bool iomap = flags & KVM_S2PTE_FLAG_IS_IOMAP;
-	bool logging_active = flags & KVM_S2_FLAG_LOGGING_ACTIVE;
-
-	VM_BUG_ON(logging_active && !cache);
-
-	/* Create stage-2 page table mapping - Levels 0 and 1 */
-	pud = stage2_get_pud(mmu, cache, addr);
-	if (!pud) {
-		/*
-		 * Ignore calls from kvm_set_spte_hva for unallocated
-		 * address ranges.
-		 */
-		return 0;
-	}
-
-	/*
-	 * While dirty page logging - dissolve huge PUD, then continue
-	 * on to allocate page.
-	 */
-	if (logging_active)
-		stage2_dissolve_pud(mmu, addr, pud);
-
-	if (stage2_pud_none(kvm, *pud)) {
-		if (!cache)
-			return 0; /* ignore calls from kvm_set_spte_hva */
-		pmd = kvm_mmu_memory_cache_alloc(cache);
-		stage2_pud_populate(kvm, pud, pmd);
-		get_page(virt_to_page(pud));
-	}
-
-	pmd = stage2_pmd_offset(kvm, pud, addr);
-	if (!pmd) {
-		/*
-		 * Ignore calls from kvm_set_spte_hva for unallocated
-		 * address ranges.
-		 */
-		return 0;
-	}
-
-	/*
-	 * While dirty page logging - dissolve huge PMD, then continue on to
-	 * allocate page.
-	 */
-	if (logging_active)
-		stage2_dissolve_pmd(mmu, addr, pmd);
-
-	/* Create stage-2 page mappings - Level 2 */
-	if (pmd_none(*pmd)) {
-		if (!cache)
-			return 0; /* ignore calls from kvm_set_spte_hva */
-		pte = kvm_mmu_memory_cache_alloc(cache);
-		kvm_pmd_populate(pmd, pte);
-		get_page(virt_to_page(pmd));
-	}
-
-	pte = pte_offset_kernel(pmd, addr);
-
-	if (iomap && pte_present(*pte))
-		return -EFAULT;
-
-	/* Create 2nd stage page table mapping - Level 3 */
-	old_pte = *pte;
-	if (pte_present(old_pte)) {
-		/* Skip page table update if there is no change */
-		if (pte_val(old_pte) == pte_val(*new_pte))
-			return 0;
-
-		kvm_set_pte(pte, __pte(0));
-		kvm_tlb_flush_vmid_ipa(mmu, addr, S2_PTE_LEVEL);
-	} else {
-		get_page(virt_to_page(pte));
-	}
-
-	kvm_set_pte(pte, *new_pte);
-	return 0;
-}
+	spin_unlock(&kvm->mmu_lock);
 
-#ifndef __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
-static int stage2_ptep_test_and_clear_young(pte_t *pte)
-{
-	if (pte_young(*pte)) {
-		*pte = pte_mkold(*pte);
-		return 1;
+	if (pgt) {
+		kvm_pgtable_stage2_destroy(pgt);
+		kfree(pgt);
 	}
-	return 0;
-}
-#else
-static int stage2_ptep_test_and_clear_young(pte_t *pte)
-{
-	return __ptep_test_and_clear_young(pte);
-}
-#endif
-
-static int stage2_pmdp_test_and_clear_young(pmd_t *pmd)
-{
-	return stage2_ptep_test_and_clear_young((pte_t *)pmd);
-}
-
-static int stage2_pudp_test_and_clear_young(pud_t *pud)
-{
-	return stage2_ptep_test_and_clear_young((pte_t *)pud);
 }
 
 /**
@@ -1468,169 +494,52 @@ static int stage2_pudp_test_and_clear_young(pud_t *pud)
  * @guest_ipa:	The IPA at which to insert the mapping
  * @pa:		The physical address of the device
  * @size:	The size of the mapping
+ * @writable:   Whether or not to create a writable mapping
  */
 int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
 			  phys_addr_t pa, unsigned long size, bool writable)
 {
-	phys_addr_t addr, end;
+	phys_addr_t addr;
 	int ret = 0;
-	unsigned long pfn;
 	struct kvm_mmu_memory_cache cache = { 0, __GFP_ZERO, NULL, };
+	struct kvm_pgtable *pgt = kvm->arch.mmu.pgt;
+	enum kvm_pgtable_prot prot = KVM_PGTABLE_PROT_DEVICE |
+				     KVM_PGTABLE_PROT_R |
+				     (writable ? KVM_PGTABLE_PROT_W : 0);
 
-	end = (guest_ipa + size + PAGE_SIZE - 1) & PAGE_MASK;
-	pfn = __phys_to_pfn(pa);
-
-	for (addr = guest_ipa; addr < end; addr += PAGE_SIZE) {
-		pte_t pte = kvm_pfn_pte(pfn, PAGE_S2_DEVICE);
-
-		if (writable)
-			pte = kvm_s2pte_mkwrite(pte);
+	size += offset_in_page(guest_ipa);
+	guest_ipa &= PAGE_MASK;
 
+	for (addr = guest_ipa; addr < guest_ipa + size; addr += PAGE_SIZE) {
 		ret = kvm_mmu_topup_memory_cache(&cache,
 						 kvm_mmu_cache_min_pages(kvm));
 		if (ret)
-			goto out;
+			break;
+
 		spin_lock(&kvm->mmu_lock);
-		ret = stage2_set_pte(&kvm->arch.mmu, &cache, addr, &pte,
-				     KVM_S2PTE_FLAG_IS_IOMAP);
+		ret = kvm_pgtable_stage2_map(pgt, addr, PAGE_SIZE, pa, prot,
+					     &cache);
 		spin_unlock(&kvm->mmu_lock);
 		if (ret)
-			goto out;
+			break;
 
-		pfn++;
+		pa += PAGE_SIZE;
 	}
 
-out:
 	kvm_mmu_free_memory_cache(&cache);
 	return ret;
 }
 
 /**
- * stage2_wp_ptes - write protect PMD range
- * @pmd:	pointer to pmd entry
- * @addr:	range start address
- * @end:	range end address
- */
-static void stage2_wp_ptes(pmd_t *pmd, phys_addr_t addr, phys_addr_t end)
-{
-	pte_t *pte;
-
-	pte = pte_offset_kernel(pmd, addr);
-	do {
-		if (!pte_none(*pte)) {
-			if (!kvm_s2pte_readonly(pte))
-				kvm_set_s2pte_readonly(pte);
-		}
-	} while (pte++, addr += PAGE_SIZE, addr != end);
-}
-
-/**
- * stage2_wp_pmds - write protect PUD range
- * kvm:		kvm instance for the VM
- * @pud:	pointer to pud entry
- * @addr:	range start address
- * @end:	range end address
- */
-static void stage2_wp_pmds(struct kvm_s2_mmu *mmu, pud_t *pud,
-			   phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	pmd_t *pmd;
-	phys_addr_t next;
-
-	pmd = stage2_pmd_offset(kvm, pud, addr);
-
-	do {
-		next = stage2_pmd_addr_end(kvm, addr, end);
-		if (!pmd_none(*pmd)) {
-			if (pmd_thp_or_huge(*pmd)) {
-				if (!kvm_s2pmd_readonly(pmd))
-					kvm_set_s2pmd_readonly(pmd);
-			} else {
-				stage2_wp_ptes(pmd, addr, next);
-			}
-		}
-	} while (pmd++, addr = next, addr != end);
-}
-
-/**
- * stage2_wp_puds - write protect P4D range
- * @p4d:	pointer to p4d entry
- * @addr:	range start address
- * @end:	range end address
- */
-static void  stage2_wp_puds(struct kvm_s2_mmu *mmu, p4d_t *p4d,
-			    phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	pud_t *pud;
-	phys_addr_t next;
-
-	pud = stage2_pud_offset(kvm, p4d, addr);
-	do {
-		next = stage2_pud_addr_end(kvm, addr, end);
-		if (!stage2_pud_none(kvm, *pud)) {
-			if (stage2_pud_huge(kvm, *pud)) {
-				if (!kvm_s2pud_readonly(pud))
-					kvm_set_s2pud_readonly(pud);
-			} else {
-				stage2_wp_pmds(mmu, pud, addr, next);
-			}
-		}
-	} while (pud++, addr = next, addr != end);
-}
-
-/**
- * stage2_wp_p4ds - write protect PGD range
- * @pgd:	pointer to pgd entry
- * @addr:	range start address
- * @end:	range end address
- */
-static void  stage2_wp_p4ds(struct kvm_s2_mmu *mmu, pgd_t *pgd,
-			    phys_addr_t addr, phys_addr_t end)
-{
-	struct kvm *kvm = mmu->kvm;
-	p4d_t *p4d;
-	phys_addr_t next;
-
-	p4d = stage2_p4d_offset(kvm, pgd, addr);
-	do {
-		next = stage2_p4d_addr_end(kvm, addr, end);
-		if (!stage2_p4d_none(kvm, *p4d))
-			stage2_wp_puds(mmu, p4d, addr, next);
-	} while (p4d++, addr = next, addr != end);
-}
-
-/**
  * stage2_wp_range() - write protect stage2 memory region range
- * @kvm:	The KVM pointer
+ * @mmu:        The KVM stage-2 MMU pointer
  * @addr:	Start address of range
  * @end:	End address of range
  */
 static void stage2_wp_range(struct kvm_s2_mmu *mmu, phys_addr_t addr, phys_addr_t end)
 {
 	struct kvm *kvm = mmu->kvm;
-	pgd_t *pgd;
-	phys_addr_t next;
-
-	pgd = mmu->pgd + stage2_pgd_index(kvm, addr);
-	do {
-		/*
-		 * Release kvm_mmu_lock periodically if the memory region is
-		 * large. Otherwise, we may see kernel panics with
-		 * CONFIG_DETECT_HUNG_TASK, CONFIG_LOCKUP_DETECTOR,
-		 * CONFIG_LOCKDEP. Additionally, holding the lock too long
-		 * will also starve other vCPUs. We have to also make sure
-		 * that the page tables are not freed while we released
-		 * the lock.
-		 */
-		cond_resched_lock(&kvm->mmu_lock);
-		if (!READ_ONCE(mmu->pgd))
-			break;
-		next = stage2_pgd_addr_end(kvm, addr, end);
-		if (stage2_pgd_present(kvm, *pgd))
-			stage2_wp_p4ds(mmu, pgd, addr, next);
-	} while (pgd++, addr = next, addr != end);
+	stage2_apply_range_resched(kvm, addr, end, kvm_pgtable_stage2_wrprotect);
 }
 
 /**
@@ -1833,20 +742,21 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 			  struct kvm_memory_slot *memslot, unsigned long hva,
 			  unsigned long fault_status)
 {
-	int ret;
+	int ret = 0;
 	bool write_fault, writable, force_pte = false;
-	bool exec_fault, needs_exec;
+	bool exec_fault;
+	bool device = false;
 	unsigned long mmu_seq;
-	gfn_t gfn = fault_ipa >> PAGE_SHIFT;
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_mmu_memory_cache *memcache = &vcpu->arch.mmu_page_cache;
 	struct vm_area_struct *vma;
 	short vma_shift;
+	gfn_t gfn;
 	kvm_pfn_t pfn;
-	pgprot_t mem_type = PAGE_S2;
 	bool logging_active = memslot_is_logging(memslot);
-	unsigned long vma_pagesize, flags = 0;
-	struct kvm_s2_mmu *mmu = vcpu->arch.hw_mmu;
+	unsigned long vma_pagesize;
+	enum kvm_pgtable_prot prot = KVM_PGTABLE_PROT_R;
+	struct kvm_pgtable *pgt;
 
 	write_fault = kvm_is_write_fault(vcpu);
 	exec_fault = kvm_vcpu_trap_is_exec_fault(vcpu);
@@ -1871,31 +781,41 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 	else
 		vma_shift = PAGE_SHIFT;
 
-	vma_pagesize = 1ULL << vma_shift;
 	if (logging_active ||
-	    (vma->vm_flags & VM_PFNMAP) ||
-	    !fault_supports_stage2_huge_mapping(memslot, hva, vma_pagesize)) {
+	    (vma->vm_flags & VM_PFNMAP)) {
 		force_pte = true;
-		vma_pagesize = PAGE_SIZE;
 		vma_shift = PAGE_SHIFT;
 	}
 
-	/*
-	 * The stage2 has a minimum of 2 level table (For arm64 see
-	 * kvm_arm_setup_stage2()). Hence, we are guaranteed that we can
-	 * use PMD_SIZE huge mappings (even when the PMD is folded into PGD).
-	 * As for PUD huge maps, we must make sure that we have at least
-	 * 3 levels, i.e, PMD is not folded.
-	 */
-	if (vma_pagesize == PMD_SIZE ||
-	    (vma_pagesize == PUD_SIZE && kvm_stage2_has_pmd(kvm)))
-		gfn = (fault_ipa & huge_page_mask(hstate_vma(vma))) >> PAGE_SHIFT;
+	if (vma_shift == PUD_SHIFT &&
+	    !fault_supports_stage2_huge_mapping(memslot, hva, PUD_SIZE))
+	       vma_shift = PMD_SHIFT;
+
+	if (vma_shift == PMD_SHIFT &&
+	    !fault_supports_stage2_huge_mapping(memslot, hva, PMD_SIZE)) {
+		force_pte = true;
+		vma_shift = PAGE_SHIFT;
+	}
+
+	vma_pagesize = 1UL << vma_shift;
+	if (vma_pagesize == PMD_SIZE || vma_pagesize == PUD_SIZE)
+		fault_ipa &= ~(vma_pagesize - 1);
+
+	gfn = fault_ipa >> PAGE_SHIFT;
 	mmap_read_unlock(current->mm);
 
-	/* We need minimum second+third level pages */
-	ret = kvm_mmu_topup_memory_cache(memcache, kvm_mmu_cache_min_pages(kvm));
-	if (ret)
-		return ret;
+	/*
+	 * Permission faults just need to update the existing leaf entry,
+	 * and so normally don't require allocations from the memcache. The
+	 * only exception to this is when dirty logging is enabled at runtime
+	 * and a write fault needs to collapse a block entry into a table.
+	 */
+	if (fault_status != FSC_PERM || (logging_active && write_fault)) {
+		ret = kvm_mmu_topup_memory_cache(memcache,
+						 kvm_mmu_cache_min_pages(kvm));
+		if (ret)
+			return ret;
+	}
 
 	mmu_seq = vcpu->kvm->mmu_notifier_seq;
 	/*
@@ -1918,28 +838,20 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 		return -EFAULT;
 
 	if (kvm_is_device_pfn(pfn)) {
-		mem_type = PAGE_S2_DEVICE;
-		flags |= KVM_S2PTE_FLAG_IS_IOMAP;
-	} else if (logging_active) {
-		/*
-		 * Faults on pages in a memslot with logging enabled
-		 * should not be mapped with huge pages (it introduces churn
-		 * and performance degradation), so force a pte mapping.
-		 */
-		flags |= KVM_S2_FLAG_LOGGING_ACTIVE;
-
+		device = true;
+	} else if (logging_active && !write_fault) {
 		/*
 		 * Only actually map the page as writable if this was a write
 		 * fault.
 		 */
-		if (!write_fault)
-			writable = false;
+		writable = false;
 	}
 
-	if (exec_fault && is_iomap(flags))
+	if (exec_fault && device)
 		return -ENOEXEC;
 
 	spin_lock(&kvm->mmu_lock);
+	pgt = vcpu->arch.hw_mmu->pgt;
 	if (mmu_notifier_retry(kvm, mmu_seq))
 		goto out_unlock;
 
@@ -1950,67 +862,31 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 	if (vma_pagesize == PAGE_SIZE && !force_pte)
 		vma_pagesize = transparent_hugepage_adjust(memslot, hva,
 							   &pfn, &fault_ipa);
-	if (writable)
+	if (writable) {
+		prot |= KVM_PGTABLE_PROT_W;
 		kvm_set_pfn_dirty(pfn);
+		mark_page_dirty(kvm, gfn);
+	}
 
-	if (fault_status != FSC_PERM && !is_iomap(flags))
+	if (fault_status != FSC_PERM && !device)
 		clean_dcache_guest_page(pfn, vma_pagesize);
 
-	if (exec_fault)
+	if (exec_fault) {
+		prot |= KVM_PGTABLE_PROT_X;
 		invalidate_icache_guest_page(pfn, vma_pagesize);
+	}
 
-	/*
-	 * If we took an execution fault we have made the
-	 * icache/dcache coherent above and should now let the s2
-	 * mapping be executable.
-	 *
-	 * Write faults (!exec_fault && FSC_PERM) are orthogonal to
-	 * execute permissions, and we preserve whatever we have.
-	 */
-	needs_exec = exec_fault ||
-		(fault_status == FSC_PERM &&
-		 stage2_is_exec(mmu, fault_ipa, vma_pagesize));
-
-	/*
-	 * If PUD_SIZE == PMD_SIZE, there is no real PUD level, and
-	 * all we have is a 2-level page table. Trying to map a PUD in
-	 * this case would be fatally wrong.
-	 */
-	if (PUD_SIZE != PMD_SIZE && vma_pagesize == PUD_SIZE) {
-		pud_t new_pud = kvm_pfn_pud(pfn, mem_type);
-
-		new_pud = kvm_pud_mkhuge(new_pud);
-		if (writable)
-			new_pud = kvm_s2pud_mkwrite(new_pud);
-
-		if (needs_exec)
-			new_pud = kvm_s2pud_mkexec(new_pud);
-
-		ret = stage2_set_pud_huge(mmu, memcache, fault_ipa, &new_pud);
-	} else if (vma_pagesize == PMD_SIZE) {
-		pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);
-
-		new_pmd = kvm_pmd_mkhuge(new_pmd);
-
-		if (writable)
-			new_pmd = kvm_s2pmd_mkwrite(new_pmd);
-
-		if (needs_exec)
-			new_pmd = kvm_s2pmd_mkexec(new_pmd);
+	if (device)
+		prot |= KVM_PGTABLE_PROT_DEVICE;
+	else if (cpus_have_const_cap(ARM64_HAS_CACHE_DIC))
+		prot |= KVM_PGTABLE_PROT_X;
 
-		ret = stage2_set_pmd_huge(mmu, memcache, fault_ipa, &new_pmd);
+	if (fault_status == FSC_PERM && !(logging_active && writable)) {
+		ret = kvm_pgtable_stage2_relax_perms(pgt, fault_ipa, prot);
 	} else {
-		pte_t new_pte = kvm_pfn_pte(pfn, mem_type);
-
-		if (writable) {
-			new_pte = kvm_s2pte_mkwrite(new_pte);
-			mark_page_dirty(kvm, gfn);
-		}
-
-		if (needs_exec)
-			new_pte = kvm_s2pte_mkexec(new_pte);
-
-		ret = stage2_set_pte(mmu, memcache, fault_ipa, &new_pte, flags);
+		ret = kvm_pgtable_stage2_map(pgt, fault_ipa, vma_pagesize,
+					     __pfn_to_phys(pfn), prot,
+					     memcache);
 	}
 
 out_unlock:
@@ -2020,46 +896,23 @@ out_unlock:
 	return ret;
 }
 
-/*
- * Resolve the access fault by making the page young again.
- * Note that because the faulting entry is guaranteed not to be
- * cached in the TLB, we don't need to invalidate anything.
- * Only the HW Access Flag updates are supported for Stage 2 (no DBM),
- * so there is no need for atomic (pte|pmd)_mkyoung operations.
- */
+/* Resolve the access fault by making the page young again. */
 static void handle_access_fault(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa)
 {
-	pud_t *pud;
-	pmd_t *pmd;
-	pte_t *pte;
-	kvm_pfn_t pfn;
-	bool pfn_valid = false;
+	pte_t pte;
+	kvm_pte_t kpte;
+	struct kvm_s2_mmu *mmu;
 
 	trace_kvm_access_fault(fault_ipa);
 
 	spin_lock(&vcpu->kvm->mmu_lock);
-
-	if (!stage2_get_leaf_entry(vcpu->arch.hw_mmu, fault_ipa, &pud, &pmd, &pte))
-		goto out;
-
-	if (pud) {		/* HugeTLB */
-		*pud = kvm_s2pud_mkyoung(*pud);
-		pfn = kvm_pud_pfn(*pud);
-		pfn_valid = true;
-	} else	if (pmd) {	/* THP, HugeTLB */
-		*pmd = pmd_mkyoung(*pmd);
-		pfn = pmd_pfn(*pmd);
-		pfn_valid = true;
-	} else {
-		*pte = pte_mkyoung(*pte);	/* Just a page... */
-		pfn = pte_pfn(*pte);
-		pfn_valid = true;
-	}
-
-out:
+	mmu = vcpu->arch.hw_mmu;
+	kpte = kvm_pgtable_stage2_mkyoung(mmu->pgt, fault_ipa);
 	spin_unlock(&vcpu->kvm->mmu_lock);
-	if (pfn_valid)
-		kvm_set_pfn_accessed(pfn);
+
+	pte = __pte(kpte);
+	if (pte_valid(pte))
+		kvm_set_pfn_accessed(pte_pfn(pte));
 }
 
 /**
@@ -2230,7 +1083,7 @@ static int kvm_unmap_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *dat
 int kvm_unmap_hva_range(struct kvm *kvm,
 			unsigned long start, unsigned long end, unsigned flags)
 {
-	if (!kvm->arch.mmu.pgd)
+	if (!kvm->arch.mmu.pgt)
 		return 0;
 
 	trace_kvm_unmap_hva_range(start, end);
@@ -2240,28 +1093,27 @@ int kvm_unmap_hva_range(struct kvm *kvm,
 
 static int kvm_set_spte_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
 {
-	pte_t *pte = (pte_t *)data;
+	kvm_pfn_t *pfn = (kvm_pfn_t *)data;
 
 	WARN_ON(size != PAGE_SIZE);
+
 	/*
-	 * We can always call stage2_set_pte with KVM_S2PTE_FLAG_LOGGING_ACTIVE
-	 * flag clear because MMU notifiers will have unmapped a huge PMD before
-	 * calling ->change_pte() (which in turn calls kvm_set_spte_hva()) and
-	 * therefore stage2_set_pte() never needs to clear out a huge PMD
-	 * through this calling path.
+	 * The MMU notifiers will have unmapped a huge PMD before calling
+	 * ->change_pte() (which in turn calls kvm_set_spte_hva()) and
+	 * therefore we never need to clear out a huge PMD through this
+	 * calling path and a memcache is not required.
 	 */
-	stage2_set_pte(&kvm->arch.mmu, NULL, gpa, pte, 0);
+	kvm_pgtable_stage2_map(kvm->arch.mmu.pgt, gpa, PAGE_SIZE,
+			       __pfn_to_phys(*pfn), KVM_PGTABLE_PROT_R, NULL);
 	return 0;
 }
 
-
 int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
 {
 	unsigned long end = hva + PAGE_SIZE;
 	kvm_pfn_t pfn = pte_pfn(pte);
-	pte_t stage2_pte;
 
-	if (!kvm->arch.mmu.pgd)
+	if (!kvm->arch.mmu.pgt)
 		return 0;
 
 	trace_kvm_set_spte_hva(hva);
@@ -2271,51 +1123,30 @@ int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
 	 * just like a translation fault and clean the cache to the PoC.
 	 */
 	clean_dcache_guest_page(pfn, PAGE_SIZE);
-	stage2_pte = kvm_pfn_pte(pfn, PAGE_S2);
-	handle_hva_to_gpa(kvm, hva, end, &kvm_set_spte_handler, &stage2_pte);
-
+	handle_hva_to_gpa(kvm, hva, end, &kvm_set_spte_handler, &pfn);
 	return 0;
 }
 
 static int kvm_age_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
 {
-	pud_t *pud;
-	pmd_t *pmd;
-	pte_t *pte;
+	pte_t pte;
+	kvm_pte_t kpte;
 
 	WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
-	if (!stage2_get_leaf_entry(&kvm->arch.mmu, gpa, &pud, &pmd, &pte))
-		return 0;
-
-	if (pud)
-		return stage2_pudp_test_and_clear_young(pud);
-	else if (pmd)
-		return stage2_pmdp_test_and_clear_young(pmd);
-	else
-		return stage2_ptep_test_and_clear_young(pte);
+	kpte = kvm_pgtable_stage2_mkold(kvm->arch.mmu.pgt, gpa);
+	pte = __pte(kpte);
+	return pte_valid(pte) && pte_young(pte);
 }
 
 static int kvm_test_age_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
 {
-	pud_t *pud;
-	pmd_t *pmd;
-	pte_t *pte;
-
 	WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
-	if (!stage2_get_leaf_entry(&kvm->arch.mmu, gpa, &pud, &pmd, &pte))
-		return 0;
-
-	if (pud)
-		return kvm_s2pud_young(*pud);
-	else if (pmd)
-		return pmd_young(*pmd);
-	else
-		return pte_young(*pte);
+	return kvm_pgtable_stage2_is_young(kvm->arch.mmu.pgt, gpa);
 }
 
 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end)
 {
-	if (!kvm->arch.mmu.pgd)
+	if (!kvm->arch.mmu.pgt)
 		return 0;
 	trace_kvm_age_hva(start, end);
 	return handle_hva_to_gpa(kvm, start, end, kvm_age_hva_handler, NULL);
@@ -2323,24 +1154,16 @@ int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end)
 
 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva)
 {
-	if (!kvm->arch.mmu.pgd)
+	if (!kvm->arch.mmu.pgt)
 		return 0;
 	trace_kvm_test_age_hva(hva);
 	return handle_hva_to_gpa(kvm, hva, hva + PAGE_SIZE,
 				 kvm_test_age_hva_handler, NULL);
 }
 
-void kvm_mmu_free_memory_caches(struct kvm_vcpu *vcpu)
-{
-	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_page_cache);
-}
-
 phys_addr_t kvm_mmu_get_httbr(void)
 {
-	if (__kvm_cpu_uses_extended_idmap())
-		return virt_to_phys(merged_hyp_pgd);
-	else
-		return virt_to_phys(hyp_pgd);
+	return __pa(hyp_pgtable->pgd);
 }
 
 phys_addr_t kvm_get_idmap_vector(void)
@@ -2348,15 +1171,11 @@ phys_addr_t kvm_get_idmap_vector(void)
 	return hyp_idmap_vector;
 }
 
-static int kvm_map_idmap_text(pgd_t *pgd)
+static int kvm_map_idmap_text(void)
 {
-	int err;
-
-	/* Create the idmap in the boot page tables */
-	err = 	__create_hyp_mappings(pgd, __kvm_idmap_ptrs_per_pgd(),
-				      hyp_idmap_start, hyp_idmap_end,
-				      __phys_to_pfn(hyp_idmap_start),
-				      PAGE_HYP_EXEC);
+	unsigned long size = hyp_idmap_end - hyp_idmap_start;
+	int err = __create_hyp_mappings(hyp_idmap_start, size, hyp_idmap_start,
+					PAGE_HYP_EXEC);
 	if (err)
 		kvm_err("Failed to idmap %lx-%lx\n",
 			hyp_idmap_start, hyp_idmap_end);
@@ -2367,6 +1186,7 @@ static int kvm_map_idmap_text(pgd_t *pgd)
 int kvm_mmu_init(void)
 {
 	int err;
+	u32 hyp_va_bits;
 
 	hyp_idmap_start = __pa_symbol(__hyp_idmap_text_start);
 	hyp_idmap_start = ALIGN_DOWN(hyp_idmap_start, PAGE_SIZE);
@@ -2380,6 +1200,8 @@ int kvm_mmu_init(void)
 	 */
 	BUG_ON((hyp_idmap_start ^ (hyp_idmap_end - 1)) & PAGE_MASK);
 
+	hyp_va_bits = 64 - ((idmap_t0sz & TCR_T0SZ_MASK) >> TCR_T0SZ_OFFSET);
+	kvm_debug("Using %u-bit virtual addresses at EL2\n", hyp_va_bits);
 	kvm_debug("IDMAP page: %lx\n", hyp_idmap_start);
 	kvm_debug("HYP VA range: %lx:%lx\n",
 		  kern_hyp_va(PAGE_OFFSET),
@@ -2397,43 +1219,30 @@ int kvm_mmu_init(void)
 		goto out;
 	}
 
-	hyp_pgd = (pgd_t *)__get_free_pages(GFP_KERNEL | __GFP_ZERO, hyp_pgd_order);
-	if (!hyp_pgd) {
-		kvm_err("Hyp mode PGD not allocated\n");
+	hyp_pgtable = kzalloc(sizeof(*hyp_pgtable), GFP_KERNEL);
+	if (!hyp_pgtable) {
+		kvm_err("Hyp mode page-table not allocated\n");
 		err = -ENOMEM;
 		goto out;
 	}
 
-	if (__kvm_cpu_uses_extended_idmap()) {
-		boot_hyp_pgd = (pgd_t *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
-							 hyp_pgd_order);
-		if (!boot_hyp_pgd) {
-			kvm_err("Hyp boot PGD not allocated\n");
-			err = -ENOMEM;
-			goto out;
-		}
-
-		err = kvm_map_idmap_text(boot_hyp_pgd);
-		if (err)
-			goto out;
+	err = kvm_pgtable_hyp_init(hyp_pgtable, hyp_va_bits);
+	if (err)
+		goto out_free_pgtable;
 
-		merged_hyp_pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
-		if (!merged_hyp_pgd) {
-			kvm_err("Failed to allocate extra HYP pgd\n");
-			goto out;
-		}
-		__kvm_extend_hypmap(boot_hyp_pgd, hyp_pgd, merged_hyp_pgd,
-				    hyp_idmap_start);
-	} else {
-		err = kvm_map_idmap_text(hyp_pgd);
-		if (err)
-			goto out;
-	}
+	err = kvm_map_idmap_text();
+	if (err)
+		goto out_destroy_pgtable;
 
 	io_map_base = hyp_idmap_start;
 	return 0;
+
+out_destroy_pgtable:
+	kvm_pgtable_hyp_destroy(hyp_pgtable);
+out_free_pgtable:
+	kfree(hyp_pgtable);
+	hyp_pgtable = NULL;
 out:
-	free_hyp_pgds();
 	return err;
 }
 
@@ -2537,7 +1346,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
 	spin_lock(&kvm->mmu_lock);
 	if (ret)
 		unmap_stage2_range(&kvm->arch.mmu, mem->guest_phys_addr, mem->memory_size);
-	else
+	else if (!cpus_have_final_cap(ARM64_HAS_STAGE2_FWB))
 		stage2_flush_memslot(kvm, memslot);
 	spin_unlock(&kvm->mmu_lock);
 out:
diff --git a/arch/arm64/kvm/pmu-emul.c b/arch/arm64/kvm/pmu-emul.c
index f0d0312c0a55..ee13c5eecd3d 100644
--- a/arch/arm64/kvm/pmu-emul.c
+++ b/arch/arm64/kvm/pmu-emul.c
@@ -20,6 +20,21 @@ static void kvm_pmu_stop_counter(struct kvm_vcpu *vcpu, struct kvm_pmc *pmc);
 
 #define PERF_ATTR_CFG1_KVM_PMU_CHAINED 0x1
 
+static u32 kvm_pmu_event_mask(struct kvm *kvm)
+{
+	switch (kvm->arch.pmuver) {
+	case 1:			/* ARMv8.0 */
+		return GENMASK(9, 0);
+	case 4:			/* ARMv8.1 */
+	case 5:			/* ARMv8.4 */
+	case 6:			/* ARMv8.5 */
+		return GENMASK(15, 0);
+	default:		/* Shouldn't be here, just for sanity */
+		WARN_ONCE(1, "Unknown PMU version %d\n", kvm->arch.pmuver);
+		return 0;
+	}
+}
+
 /**
  * kvm_pmu_idx_is_64bit - determine if select_idx is a 64bit counter
  * @vcpu: The vcpu pointer
@@ -100,7 +115,7 @@ static bool kvm_pmu_idx_has_chain_evtype(struct kvm_vcpu *vcpu, u64 select_idx)
 		return false;
 
 	reg = PMEVTYPER0_EL0 + select_idx;
-	eventsel = __vcpu_sys_reg(vcpu, reg) & ARMV8_PMU_EVTYPE_EVENT;
+	eventsel = __vcpu_sys_reg(vcpu, reg) & kvm_pmu_event_mask(vcpu->kvm);
 
 	return eventsel == ARMV8_PMUV3_PERFCTR_CHAIN;
 }
@@ -495,7 +510,7 @@ void kvm_pmu_software_increment(struct kvm_vcpu *vcpu, u64 val)
 
 		/* PMSWINC only applies to ... SW_INC! */
 		type = __vcpu_sys_reg(vcpu, PMEVTYPER0_EL0 + i);
-		type &= ARMV8_PMU_EVTYPE_EVENT;
+		type &= kvm_pmu_event_mask(vcpu->kvm);
 		if (type != ARMV8_PMUV3_PERFCTR_SW_INCR)
 			continue;
 
@@ -578,11 +593,21 @@ static void kvm_pmu_create_perf_event(struct kvm_vcpu *vcpu, u64 select_idx)
 	data = __vcpu_sys_reg(vcpu, reg);
 
 	kvm_pmu_stop_counter(vcpu, pmc);
-	eventsel = data & ARMV8_PMU_EVTYPE_EVENT;
+	if (pmc->idx == ARMV8_PMU_CYCLE_IDX)
+		eventsel = ARMV8_PMUV3_PERFCTR_CPU_CYCLES;
+	else
+		eventsel = data & kvm_pmu_event_mask(vcpu->kvm);
 
-	/* Software increment event does't need to be backed by a perf event */
-	if (eventsel == ARMV8_PMUV3_PERFCTR_SW_INCR &&
-	    pmc->idx != ARMV8_PMU_CYCLE_IDX)
+	/* Software increment event doesn't need to be backed by a perf event */
+	if (eventsel == ARMV8_PMUV3_PERFCTR_SW_INCR)
+		return;
+
+	/*
+	 * If we have a filter in place and that the event isn't allowed, do
+	 * not install a perf event either.
+	 */
+	if (vcpu->kvm->arch.pmu_filter &&
+	    !test_bit(eventsel, vcpu->kvm->arch.pmu_filter))
 		return;
 
 	memset(&attr, 0, sizeof(struct perf_event_attr));
@@ -594,8 +619,7 @@ static void kvm_pmu_create_perf_event(struct kvm_vcpu *vcpu, u64 select_idx)
 	attr.exclude_kernel = data & ARMV8_PMU_EXCLUDE_EL1 ? 1 : 0;
 	attr.exclude_hv = 1; /* Don't count EL2 events */
 	attr.exclude_host = 1; /* Don't count host events */
-	attr.config = (pmc->idx == ARMV8_PMU_CYCLE_IDX) ?
-		ARMV8_PMUV3_PERFCTR_CPU_CYCLES : eventsel;
+	attr.config = eventsel;
 
 	counter = kvm_pmu_get_pair_counter_value(vcpu, pmc);
 
@@ -679,17 +703,95 @@ static void kvm_pmu_update_pmc_chained(struct kvm_vcpu *vcpu, u64 select_idx)
 void kvm_pmu_set_counter_event_type(struct kvm_vcpu *vcpu, u64 data,
 				    u64 select_idx)
 {
-	u64 reg, event_type = data & ARMV8_PMU_EVTYPE_MASK;
+	u64 reg, mask;
+
+	mask  =  ARMV8_PMU_EVTYPE_MASK;
+	mask &= ~ARMV8_PMU_EVTYPE_EVENT;
+	mask |= kvm_pmu_event_mask(vcpu->kvm);
 
 	reg = (select_idx == ARMV8_PMU_CYCLE_IDX)
 	      ? PMCCFILTR_EL0 : PMEVTYPER0_EL0 + select_idx;
 
-	__vcpu_sys_reg(vcpu, reg) = event_type;
+	__vcpu_sys_reg(vcpu, reg) = data & mask;
 
 	kvm_pmu_update_pmc_chained(vcpu, select_idx);
 	kvm_pmu_create_perf_event(vcpu, select_idx);
 }
 
+static int kvm_pmu_probe_pmuver(void)
+{
+	struct perf_event_attr attr = { };
+	struct perf_event *event;
+	struct arm_pmu *pmu;
+	int pmuver = 0xf;
+
+	/*
+	 * Create a dummy event that only counts user cycles. As we'll never
+	 * leave this function with the event being live, it will never
+	 * count anything. But it allows us to probe some of the PMU
+	 * details. Yes, this is terrible.
+	 */
+	attr.type = PERF_TYPE_RAW;
+	attr.size = sizeof(attr);
+	attr.pinned = 1;
+	attr.disabled = 0;
+	attr.exclude_user = 0;
+	attr.exclude_kernel = 1;
+	attr.exclude_hv = 1;
+	attr.exclude_host = 1;
+	attr.config = ARMV8_PMUV3_PERFCTR_CPU_CYCLES;
+	attr.sample_period = GENMASK(63, 0);
+
+	event = perf_event_create_kernel_counter(&attr, -1, current,
+						 kvm_pmu_perf_overflow, &attr);
+
+	if (IS_ERR(event)) {
+		pr_err_once("kvm: pmu event creation failed %ld\n",
+			    PTR_ERR(event));
+		return 0xf;
+	}
+
+	if (event->pmu) {
+		pmu = to_arm_pmu(event->pmu);
+		if (pmu->pmuver)
+			pmuver = pmu->pmuver;
+	}
+
+	perf_event_disable(event);
+	perf_event_release_kernel(event);
+
+	return pmuver;
+}
+
+u64 kvm_pmu_get_pmceid(struct kvm_vcpu *vcpu, bool pmceid1)
+{
+	unsigned long *bmap = vcpu->kvm->arch.pmu_filter;
+	u64 val, mask = 0;
+	int base, i;
+
+	if (!pmceid1) {
+		val = read_sysreg(pmceid0_el0);
+		base = 0;
+	} else {
+		val = read_sysreg(pmceid1_el0);
+		base = 32;
+	}
+
+	if (!bmap)
+		return val;
+
+	for (i = 0; i < 32; i += 8) {
+		u64 byte;
+
+		byte = bitmap_get_value8(bmap, base + i);
+		mask |= byte << i;
+		byte = bitmap_get_value8(bmap, 0x4000 + base + i);
+		mask |= byte << (32 + i);
+	}
+
+	return val & mask;
+}
+
 bool kvm_arm_support_pmu_v3(void)
 {
 	/*
@@ -735,15 +837,6 @@ int kvm_arm_pmu_v3_enable(struct kvm_vcpu *vcpu)
 
 static int kvm_arm_pmu_v3_init(struct kvm_vcpu *vcpu)
 {
-	if (!kvm_arm_support_pmu_v3())
-		return -ENODEV;
-
-	if (!test_bit(KVM_ARM_VCPU_PMU_V3, vcpu->arch.features))
-		return -ENXIO;
-
-	if (vcpu->arch.pmu.created)
-		return -EBUSY;
-
 	if (irqchip_in_kernel(vcpu->kvm)) {
 		int ret;
 
@@ -796,6 +889,19 @@ static bool pmu_irq_is_valid(struct kvm *kvm, int irq)
 
 int kvm_arm_pmu_v3_set_attr(struct kvm_vcpu *vcpu, struct kvm_device_attr *attr)
 {
+	if (!kvm_arm_support_pmu_v3() ||
+	    !test_bit(KVM_ARM_VCPU_PMU_V3, vcpu->arch.features))
+		return -ENODEV;
+
+	if (vcpu->arch.pmu.created)
+		return -EBUSY;
+
+	if (!vcpu->kvm->arch.pmuver)
+		vcpu->kvm->arch.pmuver = kvm_pmu_probe_pmuver();
+
+	if (vcpu->kvm->arch.pmuver == 0xf)
+		return -ENODEV;
+
 	switch (attr->attr) {
 	case KVM_ARM_VCPU_PMU_V3_IRQ: {
 		int __user *uaddr = (int __user *)(long)attr->addr;
@@ -804,9 +910,6 @@ int kvm_arm_pmu_v3_set_attr(struct kvm_vcpu *vcpu, struct kvm_device_attr *attr)
 		if (!irqchip_in_kernel(vcpu->kvm))
 			return -EINVAL;
 
-		if (!test_bit(KVM_ARM_VCPU_PMU_V3, vcpu->arch.features))
-			return -ENODEV;
-
 		if (get_user(irq, uaddr))
 			return -EFAULT;
 
@@ -824,6 +927,53 @@ int kvm_arm_pmu_v3_set_attr(struct kvm_vcpu *vcpu, struct kvm_device_attr *attr)
 		vcpu->arch.pmu.irq_num = irq;
 		return 0;
 	}
+	case KVM_ARM_VCPU_PMU_V3_FILTER: {
+		struct kvm_pmu_event_filter __user *uaddr;
+		struct kvm_pmu_event_filter filter;
+		int nr_events;
+
+		nr_events = kvm_pmu_event_mask(vcpu->kvm) + 1;
+
+		uaddr = (struct kvm_pmu_event_filter __user *)(long)attr->addr;
+
+		if (copy_from_user(&filter, uaddr, sizeof(filter)))
+			return -EFAULT;
+
+		if (((u32)filter.base_event + filter.nevents) > nr_events ||
+		    (filter.action != KVM_PMU_EVENT_ALLOW &&
+		     filter.action != KVM_PMU_EVENT_DENY))
+			return -EINVAL;
+
+		mutex_lock(&vcpu->kvm->lock);
+
+		if (!vcpu->kvm->arch.pmu_filter) {
+			vcpu->kvm->arch.pmu_filter = bitmap_alloc(nr_events, GFP_KERNEL);
+			if (!vcpu->kvm->arch.pmu_filter) {
+				mutex_unlock(&vcpu->kvm->lock);
+				return -ENOMEM;
+			}
+
+			/*
+			 * The default depends on the first applied filter.
+			 * If it allows events, the default is to deny.
+			 * Conversely, if the first filter denies a set of
+			 * events, the default is to allow.
+			 */
+			if (filter.action == KVM_PMU_EVENT_ALLOW)
+				bitmap_zero(vcpu->kvm->arch.pmu_filter, nr_events);
+			else
+				bitmap_fill(vcpu->kvm->arch.pmu_filter, nr_events);
+		}
+
+		if (filter.action == KVM_PMU_EVENT_ALLOW)
+			bitmap_set(vcpu->kvm->arch.pmu_filter, filter.base_event, filter.nevents);
+		else
+			bitmap_clear(vcpu->kvm->arch.pmu_filter, filter.base_event, filter.nevents);
+
+		mutex_unlock(&vcpu->kvm->lock);
+
+		return 0;
+	}
 	case KVM_ARM_VCPU_PMU_V3_INIT:
 		return kvm_arm_pmu_v3_init(vcpu);
 	}
@@ -860,6 +1010,7 @@ int kvm_arm_pmu_v3_has_attr(struct kvm_vcpu *vcpu, struct kvm_device_attr *attr)
 	switch (attr->attr) {
 	case KVM_ARM_VCPU_PMU_V3_IRQ:
 	case KVM_ARM_VCPU_PMU_V3_INIT:
+	case KVM_ARM_VCPU_PMU_V3_FILTER:
 		if (kvm_arm_support_pmu_v3() &&
 		    test_bit(KVM_ARM_VCPU_PMU_V3, vcpu->arch.features))
 			return 0;
diff --git a/arch/arm64/kvm/pmu.c b/arch/arm64/kvm/pmu.c
index 3c224162b3dd..faf32a44ba04 100644
--- a/arch/arm64/kvm/pmu.c
+++ b/arch/arm64/kvm/pmu.c
@@ -31,9 +31,9 @@ static bool kvm_pmu_switch_needed(struct perf_event_attr *attr)
  */
 void kvm_set_pmu_events(u32 set, struct perf_event_attr *attr)
 {
-	struct kvm_host_data *ctx = this_cpu_ptr(&kvm_host_data);
+	struct kvm_host_data *ctx = this_cpu_ptr_hyp_sym(kvm_host_data);
 
-	if (!kvm_pmu_switch_needed(attr))
+	if (!ctx || !kvm_pmu_switch_needed(attr))
 		return;
 
 	if (!attr->exclude_host)
@@ -47,7 +47,10 @@ 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(&kvm_host_data);
+	struct kvm_host_data *ctx = this_cpu_ptr_hyp_sym(kvm_host_data);
+
+	if (!ctx)
+		return;
 
 	ctx->pmu_events.events_host &= ~clr;
 	ctx->pmu_events.events_guest &= ~clr;
@@ -173,7 +176,7 @@ void kvm_vcpu_pmu_restore_guest(struct kvm_vcpu *vcpu)
 		return;
 
 	preempt_disable();
-	host = this_cpu_ptr(&kvm_host_data);
+	host = this_cpu_ptr_hyp_sym(kvm_host_data);
 	events_guest = host->pmu_events.events_guest;
 	events_host = host->pmu_events.events_host;
 
@@ -193,7 +196,7 @@ void kvm_vcpu_pmu_restore_host(struct kvm_vcpu *vcpu)
 	if (!has_vhe())
 		return;
 
-	host = this_cpu_ptr(&kvm_host_data);
+	host = this_cpu_ptr_hyp_sym(kvm_host_data);
 	events_guest = host->pmu_events.events_guest;
 	events_host = host->pmu_events.events_host;
 
diff --git a/arch/arm64/kvm/psci.c b/arch/arm64/kvm/psci.c
index 83415e96b589..db4056ecccfd 100644
--- a/arch/arm64/kvm/psci.c
+++ b/arch/arm64/kvm/psci.c
@@ -425,27 +425,30 @@ static int get_kernel_wa_level(u64 regid)
 {
 	switch (regid) {
 	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
-		switch (kvm_arm_harden_branch_predictor()) {
-		case KVM_BP_HARDEN_UNKNOWN:
+		switch (arm64_get_spectre_v2_state()) {
+		case SPECTRE_VULNERABLE:
 			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_NOT_AVAIL;
-		case KVM_BP_HARDEN_WA_NEEDED:
+		case SPECTRE_MITIGATED:
 			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1_AVAIL;
-		case KVM_BP_HARDEN_NOT_REQUIRED:
+		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 (kvm_arm_have_ssbd()) {
-		case KVM_SSBD_FORCE_DISABLE:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
-		case KVM_SSBD_KERNEL:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL;
-		case KVM_SSBD_FORCE_ENABLE:
-		case KVM_SSBD_MITIGATED:
+		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 KVM_SSBD_UNKNOWN:
-		default:
-			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_UNKNOWN;
+		case SPECTRE_VULNERABLE:
+			return KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_NOT_AVAIL;
 		}
 	}
 
@@ -462,14 +465,8 @@ int kvm_arm_get_fw_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
 		val = kvm_psci_version(vcpu, vcpu->kvm);
 		break;
 	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_1:
-		val = get_kernel_wa_level(reg->id) & KVM_REG_FEATURE_LEVEL_MASK;
-		break;
 	case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2:
 		val = get_kernel_wa_level(reg->id) & KVM_REG_FEATURE_LEVEL_MASK;
-
-		if (val == KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL &&
-		    kvm_arm_get_vcpu_workaround_2_flag(vcpu))
-			val |= KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_ENABLED;
 		break;
 	default:
 		return -ENOENT;
@@ -527,34 +524,35 @@ int kvm_arm_set_fw_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
 			    KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_ENABLED))
 			return -EINVAL;
 
-		wa_level = val & KVM_REG_FEATURE_LEVEL_MASK;
-
-		if (get_kernel_wa_level(reg->id) < wa_level)
-			return -EINVAL;
-
 		/* The enabled bit must not be set unless the level is AVAIL. */
-		if (wa_level != KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL &&
-		    wa_level != val)
+		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;
 
-		/* Are we finished or do we need to check the enable bit ? */
-		if (kvm_arm_have_ssbd() != KVM_SSBD_KERNEL)
-			return 0;
-
 		/*
-		 * If this kernel supports the workaround to be switched on
-		 * or off, make sure it matches the requested setting.
+		 * Map all the possible incoming states to the only two we
+		 * really want to deal with.
 		 */
-		switch (wa_level) {
-		case KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_AVAIL:
-			kvm_arm_set_vcpu_workaround_2_flag(vcpu,
-			    val & KVM_REG_ARM_SMCCC_ARCH_WORKAROUND_2_ENABLED);
+		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:
-			kvm_arm_set_vcpu_workaround_2_flag(vcpu, true);
+			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;
diff --git a/arch/arm64/kvm/reset.c b/arch/arm64/kvm/reset.c
index ee33875c5c2a..f32490229a4c 100644
--- a/arch/arm64/kvm/reset.c
+++ b/arch/arm64/kvm/reset.c
@@ -319,10 +319,6 @@ int kvm_reset_vcpu(struct kvm_vcpu *vcpu)
 		vcpu->arch.reset_state.reset = false;
 	}
 
-	/* Default workaround setup is enabled (if supported) */
-	if (kvm_arm_have_ssbd() == KVM_SSBD_KERNEL)
-		vcpu->arch.workaround_flags |= VCPU_WORKAROUND_2_FLAG;
-
 	/* Reset timer */
 	ret = kvm_timer_vcpu_reset(vcpu);
 out:
@@ -339,7 +335,7 @@ u32 get_kvm_ipa_limit(void)
 
 int kvm_set_ipa_limit(void)
 {
-	unsigned int ipa_max, pa_max, va_max, parange, tgran_2;
+	unsigned int parange, tgran_2;
 	u64 mmfr0;
 
 	mmfr0 = read_sanitised_ftr_reg(SYS_ID_AA64MMFR0_EL1);
@@ -376,39 +372,11 @@ int kvm_set_ipa_limit(void)
 		break;
 	}
 
-	pa_max = id_aa64mmfr0_parange_to_phys_shift(parange);
-
-	/* Clamp the IPA limit to the PA size supported by the kernel */
-	ipa_max = (pa_max > PHYS_MASK_SHIFT) ? PHYS_MASK_SHIFT : pa_max;
-	/*
-	 * Since our stage2 table is dependent on the stage1 page table code,
-	 * we must always honor the following condition:
-	 *
-	 *  Number of levels in Stage1 >= Number of levels in Stage2.
-	 *
-	 * So clamp the ipa limit further down to limit the number of levels.
-	 * Since we can concatenate upto 16 tables at entry level, we could
-	 * go upto 4bits above the maximum VA addressable with the current
-	 * number of levels.
-	 */
-	va_max = PGDIR_SHIFT + PAGE_SHIFT - 3;
-	va_max += 4;
-
-	if (va_max < ipa_max)
-		ipa_max = va_max;
-
-	/*
-	 * If the final limit is lower than the real physical address
-	 * limit of the CPUs, report the reason.
-	 */
-	if (ipa_max < pa_max)
-		pr_info("kvm: Limiting the IPA size due to kernel %s Address limit\n",
-			(va_max < pa_max) ? "Virtual" : "Physical");
-
-	WARN(ipa_max < KVM_PHYS_SHIFT,
-	     "KVM IPA limit (%d bit) is smaller than default size\n", ipa_max);
-	kvm_ipa_limit = ipa_max;
-	kvm_info("IPA Size Limit: %dbits\n", kvm_ipa_limit);
+	kvm_ipa_limit = id_aa64mmfr0_parange_to_phys_shift(parange);
+	WARN(kvm_ipa_limit < KVM_PHYS_SHIFT,
+	     "KVM IPA Size Limit (%d bits) is smaller than default size\n",
+	     kvm_ipa_limit);
+	kvm_info("IPA Size Limit: %d bits\n", kvm_ipa_limit);
 
 	return 0;
 }
diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index 077293b5115f..3c203cb8c103 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -769,10 +769,7 @@ static bool access_pmceid(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
 	if (pmu_access_el0_disabled(vcpu))
 		return false;
 
-	if (!(p->Op2 & 1))
-		pmceid = read_sysreg(pmceid0_el0);
-	else
-		pmceid = read_sysreg(pmceid1_el0);
+	pmceid = kvm_pmu_get_pmceid(vcpu, (p->Op2 & 1));
 
 	p->regval = pmceid;
 
@@ -1131,6 +1128,9 @@ static u64 read_id_reg(const struct kvm_vcpu *vcpu,
 		if (!vcpu_has_sve(vcpu))
 			val &= ~(0xfUL << ID_AA64PFR0_SVE_SHIFT);
 		val &= ~(0xfUL << ID_AA64PFR0_AMU_SHIFT);
+		if (!(val & (0xfUL << ID_AA64PFR0_CSV2_SHIFT)) &&
+		    arm64_get_spectre_v2_state() == SPECTRE_UNAFFECTED)
+			val |= (1UL << ID_AA64PFR0_CSV2_SHIFT);
 	} else if (id == SYS_ID_AA64ISAR1_EL1 && !vcpu_has_ptrauth(vcpu)) {
 		val &= ~((0xfUL << ID_AA64ISAR1_APA_SHIFT) |
 			 (0xfUL << ID_AA64ISAR1_API_SHIFT) |
diff --git a/arch/arm64/kvm/vgic/vgic-debug.c b/arch/arm64/kvm/vgic/vgic-debug.c
index b13a9e3f99dd..f38c40a76251 100644
--- a/arch/arm64/kvm/vgic/vgic-debug.c
+++ b/arch/arm64/kvm/vgic/vgic-debug.c
@@ -260,34 +260,14 @@ static int vgic_debug_show(struct seq_file *s, void *v)
 	return 0;
 }
 
-static const struct seq_operations vgic_debug_seq_ops = {
+static const struct seq_operations vgic_debug_sops = {
 	.start = vgic_debug_start,
 	.next  = vgic_debug_next,
 	.stop  = vgic_debug_stop,
 	.show  = vgic_debug_show
 };
 
-static int debug_open(struct inode *inode, struct file *file)
-{
-	int ret;
-	ret = seq_open(file, &vgic_debug_seq_ops);
-	if (!ret) {
-		struct seq_file *seq;
-		/* seq_open will have modified file->private_data */
-		seq = file->private_data;
-		seq->private = inode->i_private;
-	}
-
-	return ret;
-};
-
-static const struct file_operations vgic_debug_fops = {
-	.owner   = THIS_MODULE,
-	.open    = debug_open,
-	.read    = seq_read,
-	.llseek  = seq_lseek,
-	.release = seq_release
-};
+DEFINE_SEQ_ATTRIBUTE(vgic_debug);
 
 void vgic_debug_init(struct kvm *kvm)
 {
diff --git a/arch/arm64/kvm/vgic/vgic-v3.c b/arch/arm64/kvm/vgic/vgic-v3.c
index 76e2d85789ed..9cdf39a94a63 100644
--- a/arch/arm64/kvm/vgic/vgic-v3.c
+++ b/arch/arm64/kvm/vgic/vgic-v3.c
@@ -662,7 +662,7 @@ void vgic_v3_load(struct kvm_vcpu *vcpu)
 	if (likely(cpu_if->vgic_sre))
 		kvm_call_hyp(__vgic_v3_write_vmcr, cpu_if->vgic_vmcr);
 
-	kvm_call_hyp(__vgic_v3_restore_aprs, kern_hyp_va(cpu_if));
+	kvm_call_hyp(__vgic_v3_restore_aprs, cpu_if);
 
 	if (has_vhe())
 		__vgic_v3_activate_traps(cpu_if);
@@ -686,7 +686,7 @@ void vgic_v3_put(struct kvm_vcpu *vcpu)
 
 	vgic_v3_vmcr_sync(vcpu);
 
-	kvm_call_hyp(__vgic_v3_save_aprs, kern_hyp_va(cpu_if));
+	kvm_call_hyp(__vgic_v3_save_aprs, cpu_if);
 
 	if (has_vhe())
 		__vgic_v3_deactivate_traps(cpu_if);