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

Pull kvm updates from Paolo Bonzini:
 "ARM64:

   - Enable the per-vcpu dirty-ring tracking mechanism, together with an
     option to keep the good old dirty log around for pages that are
     dirtied by something other than a vcpu.

   - Switch to the relaxed parallel fault handling, using RCU to delay
     page table reclaim and giving better performance under load.

   - Relax the MTE ABI, allowing a VMM to use the MAP_SHARED mapping
     option, which multi-process VMMs such as crosvm rely on (see merge
     commit 382b5b87a97d: "Fix a number of issues with MTE, such as
     races on the tags being initialised vs the PG_mte_tagged flag as
     well as the lack of support for VM_SHARED when KVM is involved.
     Patches from Catalin Marinas and Peter Collingbourne").

   - Merge the pKVM shadow vcpu state tracking that allows the
     hypervisor to have its own view of a vcpu, keeping that state
     private.

   - Add support for the PMUv3p5 architecture revision, bringing support
     for 64bit counters on systems that support it, and fix the
     no-quite-compliant CHAIN-ed counter support for the machines that
     actually exist out there.

   - Fix a handful of minor issues around 52bit VA/PA support (64kB
     pages only) as a prefix of the oncoming support for 4kB and 16kB
     pages.

   - Pick a small set of documentation and spelling fixes, because no
     good merge window would be complete without those.

  s390:

   - Second batch of the lazy destroy patches

   - First batch of KVM changes for kernel virtual != physical address
     support

   - Removal of a unused function

  x86:

   - Allow compiling out SMM support

   - Cleanup and documentation of SMM state save area format

   - Preserve interrupt shadow in SMM state save area

   - Respond to generic signals during slow page faults

   - Fixes and optimizations for the non-executable huge page errata
     fix.

   - Reprogram all performance counters on PMU filter change

   - Cleanups to Hyper-V emulation and tests

   - Process Hyper-V TLB flushes from a nested guest (i.e. from a L2
     guest running on top of a L1 Hyper-V hypervisor)

   - Advertise several new Intel features

   - x86 Xen-for-KVM:

      - Allow the Xen runstate information to cross a page boundary

      - Allow XEN_RUNSTATE_UPDATE flag behaviour to be configured

      - Add support for 32-bit guests in SCHEDOP_poll

   - Notable x86 fixes and cleanups:

      - One-off fixes for various emulation flows (SGX, VMXON, NRIPS=0).

      - Reinstate IBPB on emulated VM-Exit that was incorrectly dropped
        a few years back when eliminating unnecessary barriers when
        switching between vmcs01 and vmcs02.

      - Clean up vmread_error_trampoline() to make it more obvious that
        params must be passed on the stack, even for x86-64.

      - Let userspace set all supported bits in MSR_IA32_FEAT_CTL
        irrespective of the current guest CPUID.

      - Fudge around a race with TSC refinement that results in KVM
        incorrectly thinking a guest needs TSC scaling when running on a
        CPU with a constant TSC, but no hardware-enumerated TSC
        frequency.

      - Advertise (on AMD) that the SMM_CTL MSR is not supported

      - Remove unnecessary exports

  Generic:

   - Support for responding to signals during page faults; introduces
     new FOLL_INTERRUPTIBLE flag that was reviewed by mm folks

  Selftests:

   - Fix an inverted check in the access tracking perf test, and restore
     support for asserting that there aren't too many idle pages when
     running on bare metal.

   - Fix build errors that occur in certain setups (unsure exactly what
     is unique about the problematic setup) due to glibc overriding
     static_assert() to a variant that requires a custom message.

   - Introduce actual atomics for clear/set_bit() in selftests

   - Add support for pinning vCPUs in dirty_log_perf_test.

   - Rename the so called "perf_util" framework to "memstress".

   - Add a lightweight psuedo RNG for guest use, and use it to randomize
     the access pattern and write vs. read percentage in the memstress
     tests.

   - Add a common ucall implementation; code dedup and pre-work for
     running SEV (and beyond) guests in selftests.

   - Provide a common constructor and arch hook, which will eventually
     be used by x86 to automatically select the right hypercall (AMD vs.
     Intel).

   - A bunch of added/enabled/fixed selftests for ARM64, covering
     memslots, breakpoints, stage-2 faults and access tracking.

   - x86-specific selftest changes:

      - Clean up x86's page table management.

      - Clean up and enhance the "smaller maxphyaddr" test, and add a
        related test to cover generic emulation failure.

      - Clean up the nEPT support checks.

      - Add X86_PROPERTY_* framework to retrieve multi-bit CPUID values.

      - Fix an ordering issue in the AMX test introduced by recent
        conversions to use kvm_cpu_has(), and harden the code to guard
        against similar bugs in the future. Anything that tiggers
        caching of KVM's supported CPUID, kvm_cpu_has() in this case,
        effectively hides opt-in XSAVE features if the caching occurs
        before the test opts in via prctl().

  Documentation:

   - Remove deleted ioctls from documentation

   - Clean up the docs for the x86 MSR filter.

   - Various fixes"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (361 commits)
  KVM: x86: Add proper ReST tables for userspace MSR exits/flags
  KVM: selftests: Allocate ucall pool from MEM_REGION_DATA
  KVM: arm64: selftests: Align VA space allocator with TTBR0
  KVM: arm64: Fix benign bug with incorrect use of VA_BITS
  KVM: arm64: PMU: Fix period computation for 64bit counters with 32bit overflow
  KVM: x86: Advertise that the SMM_CTL MSR is not supported
  KVM: x86: remove unnecessary exports
  KVM: selftests: Fix spelling mistake "probabalistic" -> "probabilistic"
  tools: KVM: selftests: Convert clear/set_bit() to actual atomics
  tools: Drop "atomic_" prefix from atomic test_and_set_bit()
  tools: Drop conflicting non-atomic test_and_{clear,set}_bit() helpers
  KVM: selftests: Use non-atomic clear/set bit helpers in KVM tests
  perf tools: Use dedicated non-atomic clear/set bit helpers
  tools: Take @bit as an "unsigned long" in {clear,set}_bit() helpers
  KVM: arm64: selftests: Enable single-step without a "full" ucall()
  KVM: x86: fix APICv/x2AVIC disabled when vm reboot by itself
  KVM: Remove stale comment about KVM_REQ_UNHALT
  KVM: Add missing arch for KVM_CREATE_DEVICE and KVM_{SET,GET}_DEVICE_ATTR
  KVM: Reference to kvm_userspace_memory_region in doc and comments
  KVM: Delete all references to removed KVM_SET_MEMORY_ALIAS ioctl
  ...

16 files changed, 1741 insertions, 438 deletions

diff --git a/arch/arm64/kvm/hyp/hyp-constants.c b/arch/arm64/kvm/hyp/hyp-constants.c
index b3742a6691e8..b257a3b4bfc5 100644
--- a/arch/arm64/kvm/hyp/hyp-constants.c
+++ b/arch/arm64/kvm/hyp/hyp-constants.c
@@ -2,9 +2,12 @@
 
 #include <linux/kbuild.h>
 #include <nvhe/memory.h>
+#include <nvhe/pkvm.h>
 
 int main(void)
 {
 	DEFINE(STRUCT_HYP_PAGE_SIZE,	sizeof(struct hyp_page));
+	DEFINE(PKVM_HYP_VM_SIZE,	sizeof(struct pkvm_hyp_vm));
+	DEFINE(PKVM_HYP_VCPU_SIZE,	sizeof(struct pkvm_hyp_vcpu));
 	return 0;
 }
diff --git a/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h b/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h
index 80e99836eac7..b7bdbe63deed 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/mem_protect.h
@@ -8,8 +8,10 @@
 #define __KVM_NVHE_MEM_PROTECT__
 #include <linux/kvm_host.h>
 #include <asm/kvm_hyp.h>
+#include <asm/kvm_mmu.h>
 #include <asm/kvm_pgtable.h>
 #include <asm/virt.h>
+#include <nvhe/pkvm.h>
 #include <nvhe/spinlock.h>
 
 /*
@@ -43,30 +45,45 @@ static inline enum pkvm_page_state pkvm_getstate(enum kvm_pgtable_prot prot)
 	return prot & PKVM_PAGE_STATE_PROT_MASK;
 }
 
-struct host_kvm {
+struct host_mmu {
 	struct kvm_arch arch;
 	struct kvm_pgtable pgt;
 	struct kvm_pgtable_mm_ops mm_ops;
 	hyp_spinlock_t lock;
 };
-extern struct host_kvm host_kvm;
+extern struct host_mmu host_mmu;
 
-extern const u8 pkvm_hyp_id;
+/* This corresponds to page-table locking order */
+enum pkvm_component_id {
+	PKVM_ID_HOST,
+	PKVM_ID_HYP,
+};
+
+extern unsigned long hyp_nr_cpus;
 
 int __pkvm_prot_finalize(void);
 int __pkvm_host_share_hyp(u64 pfn);
 int __pkvm_host_unshare_hyp(u64 pfn);
+int __pkvm_host_donate_hyp(u64 pfn, u64 nr_pages);
+int __pkvm_hyp_donate_host(u64 pfn, u64 nr_pages);
 
 bool addr_is_memory(phys_addr_t phys);
 int host_stage2_idmap_locked(phys_addr_t addr, u64 size, enum kvm_pgtable_prot prot);
 int host_stage2_set_owner_locked(phys_addr_t addr, u64 size, u8 owner_id);
 int kvm_host_prepare_stage2(void *pgt_pool_base);
+int kvm_guest_prepare_stage2(struct pkvm_hyp_vm *vm, void *pgd);
 void handle_host_mem_abort(struct kvm_cpu_context *host_ctxt);
 
+int hyp_pin_shared_mem(void *from, void *to);
+void hyp_unpin_shared_mem(void *from, void *to);
+void reclaim_guest_pages(struct pkvm_hyp_vm *vm, struct kvm_hyp_memcache *mc);
+int refill_memcache(struct kvm_hyp_memcache *mc, unsigned long min_pages,
+		    struct kvm_hyp_memcache *host_mc);
+
 static __always_inline void __load_host_stage2(void)
 {
 	if (static_branch_likely(&kvm_protected_mode_initialized))
-		__load_stage2(&host_kvm.arch.mmu, &host_kvm.arch);
+		__load_stage2(&host_mmu.arch.mmu, &host_mmu.arch);
 	else
 		write_sysreg(0, vttbr_el2);
 }
diff --git a/arch/arm64/kvm/hyp/include/nvhe/memory.h b/arch/arm64/kvm/hyp/include/nvhe/memory.h
index 592b7edb3edb..ab205c4d6774 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/memory.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/memory.h
@@ -38,6 +38,10 @@ static inline phys_addr_t hyp_virt_to_phys(void *addr)
 #define hyp_page_to_virt(page)	__hyp_va(hyp_page_to_phys(page))
 #define hyp_page_to_pool(page)	(((struct hyp_page *)page)->pool)
 
+/*
+ * Refcounting for 'struct hyp_page'.
+ * hyp_pool::lock must be held if atomic access to the refcount is required.
+ */
 static inline int hyp_page_count(void *addr)
 {
 	struct hyp_page *p = hyp_virt_to_page(addr);
@@ -45,4 +49,27 @@ static inline int hyp_page_count(void *addr)
 	return p->refcount;
 }
 
+static inline void hyp_page_ref_inc(struct hyp_page *p)
+{
+	BUG_ON(p->refcount == USHRT_MAX);
+	p->refcount++;
+}
+
+static inline void hyp_page_ref_dec(struct hyp_page *p)
+{
+	BUG_ON(!p->refcount);
+	p->refcount--;
+}
+
+static inline int hyp_page_ref_dec_and_test(struct hyp_page *p)
+{
+	hyp_page_ref_dec(p);
+	return (p->refcount == 0);
+}
+
+static inline void hyp_set_page_refcounted(struct hyp_page *p)
+{
+	BUG_ON(p->refcount);
+	p->refcount = 1;
+}
 #endif /* __KVM_HYP_MEMORY_H */
diff --git a/arch/arm64/kvm/hyp/include/nvhe/mm.h b/arch/arm64/kvm/hyp/include/nvhe/mm.h
index 42d8eb9bfe72..d5ec972b5c1e 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/mm.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/mm.h
@@ -13,9 +13,13 @@
 extern struct kvm_pgtable pkvm_pgtable;
 extern hyp_spinlock_t pkvm_pgd_lock;
 
+int hyp_create_pcpu_fixmap(void);
+void *hyp_fixmap_map(phys_addr_t phys);
+void hyp_fixmap_unmap(void);
+
 int hyp_create_idmap(u32 hyp_va_bits);
 int hyp_map_vectors(void);
-int hyp_back_vmemmap(phys_addr_t phys, unsigned long size, phys_addr_t back);
+int hyp_back_vmemmap(phys_addr_t back);
 int pkvm_cpu_set_vector(enum arm64_hyp_spectre_vector slot);
 int pkvm_create_mappings(void *from, void *to, enum kvm_pgtable_prot prot);
 int pkvm_create_mappings_locked(void *from, void *to, enum kvm_pgtable_prot prot);
@@ -24,16 +28,4 @@ int __pkvm_create_private_mapping(phys_addr_t phys, size_t size,
 				  unsigned long *haddr);
 int pkvm_alloc_private_va_range(size_t size, unsigned long *haddr);
 
-static inline void hyp_vmemmap_range(phys_addr_t phys, unsigned long size,
-				     unsigned long *start, unsigned long *end)
-{
-	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct hyp_page *p = hyp_phys_to_page(phys);
-
-	*start = (unsigned long)p;
-	*end = *start + nr_pages * sizeof(struct hyp_page);
-	*start = ALIGN_DOWN(*start, PAGE_SIZE);
-	*end = ALIGN(*end, PAGE_SIZE);
-}
-
 #endif /* __KVM_HYP_MM_H */
diff --git a/arch/arm64/kvm/hyp/include/nvhe/pkvm.h b/arch/arm64/kvm/hyp/include/nvhe/pkvm.h
new file mode 100644
index 000000000000..82b3d62538a6
--- /dev/null
+++ b/arch/arm64/kvm/hyp/include/nvhe/pkvm.h
@@ -0,0 +1,68 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2021 Google LLC
+ * Author: Fuad Tabba <tabba@google.com>
+ */
+
+#ifndef __ARM64_KVM_NVHE_PKVM_H__
+#define __ARM64_KVM_NVHE_PKVM_H__
+
+#include <asm/kvm_pkvm.h>
+
+#include <nvhe/gfp.h>
+#include <nvhe/spinlock.h>
+
+/*
+ * Holds the relevant data for maintaining the vcpu state completely at hyp.
+ */
+struct pkvm_hyp_vcpu {
+	struct kvm_vcpu vcpu;
+
+	/* Backpointer to the host's (untrusted) vCPU instance. */
+	struct kvm_vcpu *host_vcpu;
+};
+
+/*
+ * Holds the relevant data for running a protected vm.
+ */
+struct pkvm_hyp_vm {
+	struct kvm kvm;
+
+	/* Backpointer to the host's (untrusted) KVM instance. */
+	struct kvm *host_kvm;
+
+	/* The guest's stage-2 page-table managed by the hypervisor. */
+	struct kvm_pgtable pgt;
+	struct kvm_pgtable_mm_ops mm_ops;
+	struct hyp_pool pool;
+	hyp_spinlock_t lock;
+
+	/*
+	 * The number of vcpus initialized and ready to run.
+	 * Modifying this is protected by 'vm_table_lock'.
+	 */
+	unsigned int nr_vcpus;
+
+	/* Array of the hyp vCPU structures for this VM. */
+	struct pkvm_hyp_vcpu *vcpus[];
+};
+
+static inline struct pkvm_hyp_vm *
+pkvm_hyp_vcpu_to_hyp_vm(struct pkvm_hyp_vcpu *hyp_vcpu)
+{
+	return container_of(hyp_vcpu->vcpu.kvm, struct pkvm_hyp_vm, kvm);
+}
+
+void pkvm_hyp_vm_table_init(void *tbl);
+
+int __pkvm_init_vm(struct kvm *host_kvm, unsigned long vm_hva,
+		   unsigned long pgd_hva);
+int __pkvm_init_vcpu(pkvm_handle_t handle, struct kvm_vcpu *host_vcpu,
+		     unsigned long vcpu_hva);
+int __pkvm_teardown_vm(pkvm_handle_t handle);
+
+struct pkvm_hyp_vcpu *pkvm_load_hyp_vcpu(pkvm_handle_t handle,
+					 unsigned int vcpu_idx);
+void pkvm_put_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu);
+
+#endif /* __ARM64_KVM_NVHE_PKVM_H__ */
diff --git a/arch/arm64/kvm/hyp/include/nvhe/spinlock.h b/arch/arm64/kvm/hyp/include/nvhe/spinlock.h
index 4652fd04bdbe..7c7ea8c55405 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/spinlock.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/spinlock.h
@@ -28,9 +28,17 @@ typedef union hyp_spinlock {
 	};
 } hyp_spinlock_t;
 
+#define __HYP_SPIN_LOCK_INITIALIZER \
+	{ .__val = 0 }
+
+#define __HYP_SPIN_LOCK_UNLOCKED \
+	((hyp_spinlock_t) __HYP_SPIN_LOCK_INITIALIZER)
+
+#define DEFINE_HYP_SPINLOCK(x)	hyp_spinlock_t x = __HYP_SPIN_LOCK_UNLOCKED
+
 #define hyp_spin_lock_init(l)						\
 do {									\
-	*(l) = (hyp_spinlock_t){ .__val = 0 };				\
+	*(l) = __HYP_SPIN_LOCK_UNLOCKED;				\
 } while (0)
 
 static inline void hyp_spin_lock(hyp_spinlock_t *lock)
diff --git a/arch/arm64/kvm/hyp/nvhe/cache.S b/arch/arm64/kvm/hyp/nvhe/cache.S
index 0c367eb5f4e2..85936c17ae40 100644
--- a/arch/arm64/kvm/hyp/nvhe/cache.S
+++ b/arch/arm64/kvm/hyp/nvhe/cache.S
@@ -12,3 +12,14 @@ SYM_FUNC_START(__pi_dcache_clean_inval_poc)
 	ret
 SYM_FUNC_END(__pi_dcache_clean_inval_poc)
 SYM_FUNC_ALIAS(dcache_clean_inval_poc, __pi_dcache_clean_inval_poc)
+
+SYM_FUNC_START(__pi_icache_inval_pou)
+alternative_if ARM64_HAS_CACHE_DIC
+	isb
+	ret
+alternative_else_nop_endif
+
+	invalidate_icache_by_line x0, x1, x2, x3
+	ret
+SYM_FUNC_END(__pi_icache_inval_pou)
+SYM_FUNC_ALIAS(icache_inval_pou, __pi_icache_inval_pou)
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index 3cea4b6ac23e..728e01d4536b 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -15,17 +15,93 @@
 
 #include <nvhe/mem_protect.h>
 #include <nvhe/mm.h>
+#include <nvhe/pkvm.h>
 #include <nvhe/trap_handler.h>
 
 DEFINE_PER_CPU(struct kvm_nvhe_init_params, kvm_init_params);
 
 void __kvm_hyp_host_forward_smc(struct kvm_cpu_context *host_ctxt);
 
+static void flush_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
+{
+	struct kvm_vcpu *host_vcpu = hyp_vcpu->host_vcpu;
+
+	hyp_vcpu->vcpu.arch.ctxt	= host_vcpu->arch.ctxt;
+
+	hyp_vcpu->vcpu.arch.sve_state	= kern_hyp_va(host_vcpu->arch.sve_state);
+	hyp_vcpu->vcpu.arch.sve_max_vl	= host_vcpu->arch.sve_max_vl;
+
+	hyp_vcpu->vcpu.arch.hw_mmu	= host_vcpu->arch.hw_mmu;
+
+	hyp_vcpu->vcpu.arch.hcr_el2	= host_vcpu->arch.hcr_el2;
+	hyp_vcpu->vcpu.arch.mdcr_el2	= host_vcpu->arch.mdcr_el2;
+	hyp_vcpu->vcpu.arch.cptr_el2	= host_vcpu->arch.cptr_el2;
+
+	hyp_vcpu->vcpu.arch.iflags	= host_vcpu->arch.iflags;
+	hyp_vcpu->vcpu.arch.fp_state	= host_vcpu->arch.fp_state;
+
+	hyp_vcpu->vcpu.arch.debug_ptr	= kern_hyp_va(host_vcpu->arch.debug_ptr);
+	hyp_vcpu->vcpu.arch.host_fpsimd_state = host_vcpu->arch.host_fpsimd_state;
+
+	hyp_vcpu->vcpu.arch.vsesr_el2	= host_vcpu->arch.vsesr_el2;
+
+	hyp_vcpu->vcpu.arch.vgic_cpu.vgic_v3 = host_vcpu->arch.vgic_cpu.vgic_v3;
+}
+
+static void sync_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
+{
+	struct kvm_vcpu *host_vcpu = hyp_vcpu->host_vcpu;
+	struct vgic_v3_cpu_if *hyp_cpu_if = &hyp_vcpu->vcpu.arch.vgic_cpu.vgic_v3;
+	struct vgic_v3_cpu_if *host_cpu_if = &host_vcpu->arch.vgic_cpu.vgic_v3;
+	unsigned int i;
+
+	host_vcpu->arch.ctxt		= hyp_vcpu->vcpu.arch.ctxt;
+
+	host_vcpu->arch.hcr_el2		= hyp_vcpu->vcpu.arch.hcr_el2;
+	host_vcpu->arch.cptr_el2	= hyp_vcpu->vcpu.arch.cptr_el2;
+
+	host_vcpu->arch.fault		= hyp_vcpu->vcpu.arch.fault;
+
+	host_vcpu->arch.iflags		= hyp_vcpu->vcpu.arch.iflags;
+	host_vcpu->arch.fp_state	= hyp_vcpu->vcpu.arch.fp_state;
+
+	host_cpu_if->vgic_hcr		= hyp_cpu_if->vgic_hcr;
+	for (i = 0; i < hyp_cpu_if->used_lrs; ++i)
+		host_cpu_if->vgic_lr[i] = hyp_cpu_if->vgic_lr[i];
+}
+
 static void handle___kvm_vcpu_run(struct kvm_cpu_context *host_ctxt)
 {
-	DECLARE_REG(struct kvm_vcpu *, vcpu, host_ctxt, 1);
+	DECLARE_REG(struct kvm_vcpu *, host_vcpu, host_ctxt, 1);
+	int ret;
 
-	cpu_reg(host_ctxt, 1) =  __kvm_vcpu_run(kern_hyp_va(vcpu));
+	host_vcpu = kern_hyp_va(host_vcpu);
+
+	if (unlikely(is_protected_kvm_enabled())) {
+		struct pkvm_hyp_vcpu *hyp_vcpu;
+		struct kvm *host_kvm;
+
+		host_kvm = kern_hyp_va(host_vcpu->kvm);
+		hyp_vcpu = pkvm_load_hyp_vcpu(host_kvm->arch.pkvm.handle,
+					      host_vcpu->vcpu_idx);
+		if (!hyp_vcpu) {
+			ret = -EINVAL;
+			goto out;
+		}
+
+		flush_hyp_vcpu(hyp_vcpu);
+
+		ret = __kvm_vcpu_run(&hyp_vcpu->vcpu);
+
+		sync_hyp_vcpu(hyp_vcpu);
+		pkvm_put_hyp_vcpu(hyp_vcpu);
+	} else {
+		/* The host is fully trusted, run its vCPU directly. */
+		ret = __kvm_vcpu_run(host_vcpu);
+	}
+
+out:
+	cpu_reg(host_ctxt, 1) =  ret;
 }
 
 static void handle___kvm_adjust_pc(struct kvm_cpu_context *host_ctxt)
@@ -191,6 +267,33 @@ static void handle___pkvm_vcpu_init_traps(struct kvm_cpu_context *host_ctxt)
 	__pkvm_vcpu_init_traps(kern_hyp_va(vcpu));
 }
 
+static void handle___pkvm_init_vm(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(struct kvm *, host_kvm, host_ctxt, 1);
+	DECLARE_REG(unsigned long, vm_hva, host_ctxt, 2);
+	DECLARE_REG(unsigned long, pgd_hva, host_ctxt, 3);
+
+	host_kvm = kern_hyp_va(host_kvm);
+	cpu_reg(host_ctxt, 1) = __pkvm_init_vm(host_kvm, vm_hva, pgd_hva);
+}
+
+static void handle___pkvm_init_vcpu(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(pkvm_handle_t, handle, host_ctxt, 1);
+	DECLARE_REG(struct kvm_vcpu *, host_vcpu, host_ctxt, 2);
+	DECLARE_REG(unsigned long, vcpu_hva, host_ctxt, 3);
+
+	host_vcpu = kern_hyp_va(host_vcpu);
+	cpu_reg(host_ctxt, 1) = __pkvm_init_vcpu(handle, host_vcpu, vcpu_hva);
+}
+
+static void handle___pkvm_teardown_vm(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(pkvm_handle_t, handle, host_ctxt, 1);
+
+	cpu_reg(host_ctxt, 1) = __pkvm_teardown_vm(handle);
+}
+
 typedef void (*hcall_t)(struct kvm_cpu_context *);
 
 #define HANDLE_FUNC(x)	[__KVM_HOST_SMCCC_FUNC_##x] = (hcall_t)handle_##x
@@ -220,6 +323,9 @@ static const hcall_t host_hcall[] = {
 	HANDLE_FUNC(__vgic_v3_save_aprs),
 	HANDLE_FUNC(__vgic_v3_restore_aprs),
 	HANDLE_FUNC(__pkvm_vcpu_init_traps),
+	HANDLE_FUNC(__pkvm_init_vm),
+	HANDLE_FUNC(__pkvm_init_vcpu),
+	HANDLE_FUNC(__pkvm_teardown_vm),
 };
 
 static void handle_host_hcall(struct kvm_cpu_context *host_ctxt)
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-smp.c b/arch/arm64/kvm/hyp/nvhe/hyp-smp.c
index 9f54833af400..04d194583f1e 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-smp.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-smp.c
@@ -23,6 +23,8 @@ u64 cpu_logical_map(unsigned int cpu)
 	return hyp_cpu_logical_map[cpu];
 }
 
+unsigned long __ro_after_init kvm_arm_hyp_percpu_base[NR_CPUS];
+
 unsigned long __hyp_per_cpu_offset(unsigned int cpu)
 {
 	unsigned long *cpu_base_array;
diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
index 07f9dc9848ef..552653fa18be 100644
--- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c
+++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
@@ -21,21 +21,33 @@
 
 #define KVM_HOST_S2_FLAGS (KVM_PGTABLE_S2_NOFWB | KVM_PGTABLE_S2_IDMAP)
 
-extern unsigned long hyp_nr_cpus;
-struct host_kvm host_kvm;
+struct host_mmu host_mmu;
 
 static struct hyp_pool host_s2_pool;
 
-const u8 pkvm_hyp_id = 1;
+static DEFINE_PER_CPU(struct pkvm_hyp_vm *, __current_vm);
+#define current_vm (*this_cpu_ptr(&__current_vm))
+
+static void guest_lock_component(struct pkvm_hyp_vm *vm)
+{
+	hyp_spin_lock(&vm->lock);
+	current_vm = vm;
+}
+
+static void guest_unlock_component(struct pkvm_hyp_vm *vm)
+{
+	current_vm = NULL;
+	hyp_spin_unlock(&vm->lock);
+}
 
 static void host_lock_component(void)
 {
-	hyp_spin_lock(&host_kvm.lock);
+	hyp_spin_lock(&host_mmu.lock);
 }
 
 static void host_unlock_component(void)
 {
-	hyp_spin_unlock(&host_kvm.lock);
+	hyp_spin_unlock(&host_mmu.lock);
 }
 
 static void hyp_lock_component(void)
@@ -79,6 +91,11 @@ static void host_s2_put_page(void *addr)
 	hyp_put_page(&host_s2_pool, addr);
 }
 
+static void host_s2_free_removed_table(void *addr, u32 level)
+{
+	kvm_pgtable_stage2_free_removed(&host_mmu.mm_ops, addr, level);
+}
+
 static int prepare_s2_pool(void *pgt_pool_base)
 {
 	unsigned long nr_pages, pfn;
@@ -90,9 +107,10 @@ static int prepare_s2_pool(void *pgt_pool_base)
 	if (ret)
 		return ret;
 
-	host_kvm.mm_ops = (struct kvm_pgtable_mm_ops) {
+	host_mmu.mm_ops = (struct kvm_pgtable_mm_ops) {
 		.zalloc_pages_exact = host_s2_zalloc_pages_exact,
 		.zalloc_page = host_s2_zalloc_page,
+		.free_removed_table = host_s2_free_removed_table,
 		.phys_to_virt = hyp_phys_to_virt,
 		.virt_to_phys = hyp_virt_to_phys,
 		.page_count = hyp_page_count,
@@ -111,7 +129,7 @@ static void prepare_host_vtcr(void)
 	parange = kvm_get_parange(id_aa64mmfr0_el1_sys_val);
 	phys_shift = id_aa64mmfr0_parange_to_phys_shift(parange);
 
-	host_kvm.arch.vtcr = kvm_get_vtcr(id_aa64mmfr0_el1_sys_val,
+	host_mmu.arch.vtcr = kvm_get_vtcr(id_aa64mmfr0_el1_sys_val,
 					  id_aa64mmfr1_el1_sys_val, phys_shift);
 }
 
@@ -119,45 +137,170 @@ static bool host_stage2_force_pte_cb(u64 addr, u64 end, enum kvm_pgtable_prot pr
 
 int kvm_host_prepare_stage2(void *pgt_pool_base)
 {
-	struct kvm_s2_mmu *mmu = &host_kvm.arch.mmu;
+	struct kvm_s2_mmu *mmu = &host_mmu.arch.mmu;
 	int ret;
 
 	prepare_host_vtcr();
-	hyp_spin_lock_init(&host_kvm.lock);
-	mmu->arch = &host_kvm.arch;
+	hyp_spin_lock_init(&host_mmu.lock);
+	mmu->arch = &host_mmu.arch;
 
 	ret = prepare_s2_pool(pgt_pool_base);
 	if (ret)
 		return ret;
 
-	ret = __kvm_pgtable_stage2_init(&host_kvm.pgt, mmu,
-					&host_kvm.mm_ops, KVM_HOST_S2_FLAGS,
+	ret = __kvm_pgtable_stage2_init(&host_mmu.pgt, mmu,
+					&host_mmu.mm_ops, KVM_HOST_S2_FLAGS,
 					host_stage2_force_pte_cb);
 	if (ret)
 		return ret;
 
-	mmu->pgd_phys = __hyp_pa(host_kvm.pgt.pgd);
-	mmu->pgt = &host_kvm.pgt;
+	mmu->pgd_phys = __hyp_pa(host_mmu.pgt.pgd);
+	mmu->pgt = &host_mmu.pgt;
 	atomic64_set(&mmu->vmid.id, 0);
 
 	return 0;
 }
 
+static bool guest_stage2_force_pte_cb(u64 addr, u64 end,
+				      enum kvm_pgtable_prot prot)
+{
+	return true;
+}
+
+static void *guest_s2_zalloc_pages_exact(size_t size)
+{
+	void *addr = hyp_alloc_pages(&current_vm->pool, get_order(size));
+
+	WARN_ON(size != (PAGE_SIZE << get_order(size)));
+	hyp_split_page(hyp_virt_to_page(addr));
+
+	return addr;
+}
+
+static void guest_s2_free_pages_exact(void *addr, unsigned long size)
+{
+	u8 order = get_order(size);
+	unsigned int i;
+
+	for (i = 0; i < (1 << order); i++)
+		hyp_put_page(&current_vm->pool, addr + (i * PAGE_SIZE));
+}
+
+static void *guest_s2_zalloc_page(void *mc)
+{
+	struct hyp_page *p;
+	void *addr;
+
+	addr = hyp_alloc_pages(&current_vm->pool, 0);
+	if (addr)
+		return addr;
+
+	addr = pop_hyp_memcache(mc, hyp_phys_to_virt);
+	if (!addr)
+		return addr;
+
+	memset(addr, 0, PAGE_SIZE);
+	p = hyp_virt_to_page(addr);
+	memset(p, 0, sizeof(*p));
+	p->refcount = 1;
+
+	return addr;
+}
+
+static void guest_s2_get_page(void *addr)
+{
+	hyp_get_page(&current_vm->pool, addr);
+}
+
+static void guest_s2_put_page(void *addr)
+{
+	hyp_put_page(&current_vm->pool, addr);
+}
+
+static void clean_dcache_guest_page(void *va, size_t size)
+{
+	__clean_dcache_guest_page(hyp_fixmap_map(__hyp_pa(va)), size);
+	hyp_fixmap_unmap();
+}
+
+static void invalidate_icache_guest_page(void *va, size_t size)
+{
+	__invalidate_icache_guest_page(hyp_fixmap_map(__hyp_pa(va)), size);
+	hyp_fixmap_unmap();
+}
+
+int kvm_guest_prepare_stage2(struct pkvm_hyp_vm *vm, void *pgd)
+{
+	struct kvm_s2_mmu *mmu = &vm->kvm.arch.mmu;
+	unsigned long nr_pages;
+	int ret;
+
+	nr_pages = kvm_pgtable_stage2_pgd_size(vm->kvm.arch.vtcr) >> PAGE_SHIFT;
+	ret = hyp_pool_init(&vm->pool, hyp_virt_to_pfn(pgd), nr_pages, 0);
+	if (ret)
+		return ret;
+
+	hyp_spin_lock_init(&vm->lock);
+	vm->mm_ops = (struct kvm_pgtable_mm_ops) {
+		.zalloc_pages_exact	= guest_s2_zalloc_pages_exact,
+		.free_pages_exact	= guest_s2_free_pages_exact,
+		.zalloc_page		= guest_s2_zalloc_page,
+		.phys_to_virt		= hyp_phys_to_virt,
+		.virt_to_phys		= hyp_virt_to_phys,
+		.page_count		= hyp_page_count,
+		.get_page		= guest_s2_get_page,
+		.put_page		= guest_s2_put_page,
+		.dcache_clean_inval_poc	= clean_dcache_guest_page,
+		.icache_inval_pou	= invalidate_icache_guest_page,
+	};
+
+	guest_lock_component(vm);
+	ret = __kvm_pgtable_stage2_init(mmu->pgt, mmu, &vm->mm_ops, 0,
+					guest_stage2_force_pte_cb);
+	guest_unlock_component(vm);
+	if (ret)
+		return ret;
+
+	vm->kvm.arch.mmu.pgd_phys = __hyp_pa(vm->pgt.pgd);
+
+	return 0;
+}
+
+void reclaim_guest_pages(struct pkvm_hyp_vm *vm, struct kvm_hyp_memcache *mc)
+{
+	void *addr;
+
+	/* Dump all pgtable pages in the hyp_pool */
+	guest_lock_component(vm);
+	kvm_pgtable_stage2_destroy(&vm->pgt);
+	vm->kvm.arch.mmu.pgd_phys = 0ULL;
+	guest_unlock_component(vm);
+
+	/* Drain the hyp_pool into the memcache */
+	addr = hyp_alloc_pages(&vm->pool, 0);
+	while (addr) {
+		memset(hyp_virt_to_page(addr), 0, sizeof(struct hyp_page));
+		push_hyp_memcache(mc, addr, hyp_virt_to_phys);
+		WARN_ON(__pkvm_hyp_donate_host(hyp_virt_to_pfn(addr), 1));
+		addr = hyp_alloc_pages(&vm->pool, 0);
+	}
+}
+
 int __pkvm_prot_finalize(void)
 {
-	struct kvm_s2_mmu *mmu = &host_kvm.arch.mmu;
+	struct kvm_s2_mmu *mmu = &host_mmu.arch.mmu;
 	struct kvm_nvhe_init_params *params = this_cpu_ptr(&kvm_init_params);
 
 	if (params->hcr_el2 & HCR_VM)
 		return -EPERM;
 
 	params->vttbr = kvm_get_vttbr(mmu);
-	params->vtcr = host_kvm.arch.vtcr;
+	params->vtcr = host_mmu.arch.vtcr;
 	params->hcr_el2 |= HCR_VM;
 	kvm_flush_dcache_to_poc(params, sizeof(*params));
 
 	write_sysreg(params->hcr_el2, hcr_el2);
-	__load_stage2(&host_kvm.arch.mmu, &host_kvm.arch);
+	__load_stage2(&host_mmu.arch.mmu, &host_mmu.arch);
 
 	/*
 	 * Make sure to have an ISB before the TLB maintenance below but only
@@ -175,7 +318,7 @@ int __pkvm_prot_finalize(void)
 
 static int host_stage2_unmap_dev_all(void)
 {
-	struct kvm_pgtable *pgt = &host_kvm.pgt;
+	struct kvm_pgtable *pgt = &host_mmu.pgt;
 	struct memblock_region *reg;
 	u64 addr = 0;
 	int i, ret;
@@ -195,7 +338,7 @@ struct kvm_mem_range {
 	u64 end;
 };
 
-static bool find_mem_range(phys_addr_t addr, struct kvm_mem_range *range)
+static struct memblock_region *find_mem_range(phys_addr_t addr, struct kvm_mem_range *range)
 {
 	int cur, left = 0, right = hyp_memblock_nr;
 	struct memblock_region *reg;
@@ -218,18 +361,28 @@ static bool find_mem_range(phys_addr_t addr, struct kvm_mem_range *range)
 		} else {
 			range->start = reg->base;
 			range->end = end;
-			return true;
+			return reg;
 		}
 	}
 
-	return false;
+	return NULL;
 }
 
 bool addr_is_memory(phys_addr_t phys)
 {
 	struct kvm_mem_range range;
 
-	return find_mem_range(phys, &range);
+	return !!find_mem_range(phys, &range);
+}
+
+static bool addr_is_allowed_memory(phys_addr_t phys)
+{
+	struct memblock_region *reg;
+	struct kvm_mem_range range;
+
+	reg = find_mem_range(phys, &range);
+
+	return reg && !(reg->flags & MEMBLOCK_NOMAP);
 }
 
 static bool is_in_mem_range(u64 addr, struct kvm_mem_range *range)
@@ -250,8 +403,8 @@ static bool range_is_memory(u64 start, u64 end)
 static inline int __host_stage2_idmap(u64 start, u64 end,
 				      enum kvm_pgtable_prot prot)
 {
-	return kvm_pgtable_stage2_map(&host_kvm.pgt, start, end - start, start,
-				      prot, &host_s2_pool);
+	return kvm_pgtable_stage2_map(&host_mmu.pgt, start, end - start, start,
+				      prot, &host_s2_pool, 0);
 }
 
 /*
@@ -263,7 +416,7 @@ static inline int __host_stage2_idmap(u64 start, u64 end,
 #define host_stage2_try(fn, ...)					\
 	({								\
 		int __ret;						\
-		hyp_assert_lock_held(&host_kvm.lock);			\
+		hyp_assert_lock_held(&host_mmu.lock);			\
 		__ret = fn(__VA_ARGS__);				\
 		if (__ret == -ENOMEM) {					\
 			__ret = host_stage2_unmap_dev_all();		\
@@ -286,8 +439,8 @@ static int host_stage2_adjust_range(u64 addr, struct kvm_mem_range *range)
 	u32 level;
 	int ret;
 
-	hyp_assert_lock_held(&host_kvm.lock);
-	ret = kvm_pgtable_get_leaf(&host_kvm.pgt, addr, &pte, &level);
+	hyp_assert_lock_held(&host_mmu.lock);
+	ret = kvm_pgtable_get_leaf(&host_mmu.pgt, addr, &pte, &level);
 	if (ret)
 		return ret;
 
@@ -319,7 +472,7 @@ int host_stage2_idmap_locked(phys_addr_t addr, u64 size,
 
 int host_stage2_set_owner_locked(phys_addr_t addr, u64 size, u8 owner_id)
 {
-	return host_stage2_try(kvm_pgtable_stage2_set_owner, &host_kvm.pgt,
+	return host_stage2_try(kvm_pgtable_stage2_set_owner, &host_mmu.pgt,
 			       addr, size, &host_s2_pool, owner_id);
 }
 
@@ -348,7 +501,7 @@ static bool host_stage2_force_pte_cb(u64 addr, u64 end, enum kvm_pgtable_prot pr
 static int host_stage2_idmap(u64 addr)
 {
 	struct kvm_mem_range range;
-	bool is_memory = find_mem_range(addr, &range);
+	bool is_memory = !!find_mem_range(addr, &range);
 	enum kvm_pgtable_prot prot;
 	int ret;
 
@@ -380,12 +533,6 @@ void handle_host_mem_abort(struct kvm_cpu_context *host_ctxt)
 	BUG_ON(ret && ret != -EAGAIN);
 }
 
-/* This corresponds to locking order */
-enum pkvm_component_id {
-	PKVM_ID_HOST,
-	PKVM_ID_HYP,
-};
-
 struct pkvm_mem_transition {
 	u64				nr_pages;
 
@@ -399,6 +546,9 @@ struct pkvm_mem_transition {
 				/* Address in the completer's address space */
 				u64	completer_addr;
 			} host;
+			struct {
+				u64	completer_addr;
+			} hyp;
 		};
 	} initiator;
 
@@ -412,23 +562,24 @@ struct pkvm_mem_share {
 	const enum kvm_pgtable_prot		completer_prot;
 };
 
+struct pkvm_mem_donation {
+	const struct pkvm_mem_transition	tx;
+};
+
 struct check_walk_data {
 	enum pkvm_page_state	desired;
 	enum pkvm_page_state	(*get_page_state)(kvm_pte_t pte);
 };
 
-static int __check_page_state_visitor(u64 addr, u64 end, u32 level,
-				      kvm_pte_t *ptep,
-				      enum kvm_pgtable_walk_flags flag,
-				      void * const arg)
+static int __check_page_state_visitor(const struct kvm_pgtable_visit_ctx *ctx,
+				      enum kvm_pgtable_walk_flags visit)
 {
-	struct check_walk_data *d = arg;
-	kvm_pte_t pte = *ptep;
+	struct check_walk_data *d = ctx->arg;
 
-	if (kvm_pte_valid(pte) && !addr_is_memory(kvm_pte_to_phys(pte)))
+	if (kvm_pte_valid(ctx->old) && !addr_is_allowed_memory(kvm_pte_to_phys(ctx->old)))
 		return -EINVAL;
 
-	return d->get_page_state(pte) == d->desired ? 0 : -EPERM;
+	return d->get_page_state(ctx->old) == d->desired ? 0 : -EPERM;
 }
 
 static int check_page_state_range(struct kvm_pgtable *pgt, u64 addr, u64 size,
@@ -459,8 +610,8 @@ static int __host_check_page_state_range(u64 addr, u64 size,
 		.get_page_state	= host_get_page_state,
 	};
 
-	hyp_assert_lock_held(&host_kvm.lock);
-	return check_page_state_range(&host_kvm.pgt, addr, size, &d);
+	hyp_assert_lock_held(&host_mmu.lock);
+	return check_page_state_range(&host_mmu.pgt, addr, size, &d);
 }
 
 static int __host_set_page_state_range(u64 addr, u64 size,
@@ -511,6 +662,46 @@ static int host_initiate_unshare(u64 *completer_addr,
 	return __host_set_page_state_range(addr, size, PKVM_PAGE_OWNED);
 }
 
+static int host_initiate_donation(u64 *completer_addr,
+				  const struct pkvm_mem_transition *tx)
+{
+	u8 owner_id = tx->completer.id;
+	u64 size = tx->nr_pages * PAGE_SIZE;
+
+	*completer_addr = tx->initiator.host.completer_addr;
+	return host_stage2_set_owner_locked(tx->initiator.addr, size, owner_id);
+}
+
+static bool __host_ack_skip_pgtable_check(const struct pkvm_mem_transition *tx)
+{
+	return !(IS_ENABLED(CONFIG_NVHE_EL2_DEBUG) ||
+		 tx->initiator.id != PKVM_ID_HYP);
+}
+
+static int __host_ack_transition(u64 addr, const struct pkvm_mem_transition *tx,
+				 enum pkvm_page_state state)
+{
+	u64 size = tx->nr_pages * PAGE_SIZE;
+
+	if (__host_ack_skip_pgtable_check(tx))
+		return 0;
+
+	return __host_check_page_state_range(addr, size, state);
+}
+
+static int host_ack_donation(u64 addr, const struct pkvm_mem_transition *tx)
+{
+	return __host_ack_transition(addr, tx, PKVM_NOPAGE);
+}
+
+static int host_complete_donation(u64 addr, const struct pkvm_mem_transition *tx)
+{
+	u64 size = tx->nr_pages * PAGE_SIZE;
+	u8 host_id = tx->completer.id;
+
+	return host_stage2_set_owner_locked(addr, size, host_id);
+}
+
 static enum pkvm_page_state hyp_get_page_state(kvm_pte_t pte)
 {
 	if (!kvm_pte_valid(pte))
@@ -531,6 +722,27 @@ static int __hyp_check_page_state_range(u64 addr, u64 size,
 	return check_page_state_range(&pkvm_pgtable, addr, size, &d);
 }
 
+static int hyp_request_donation(u64 *completer_addr,
+				const struct pkvm_mem_transition *tx)
+{
+	u64 size = tx->nr_pages * PAGE_SIZE;
+	u64 addr = tx->initiator.addr;
+
+	*completer_addr = tx->initiator.hyp.completer_addr;
+	return __hyp_check_page_state_range(addr, size, PKVM_PAGE_OWNED);
+}
+
+static int hyp_initiate_donation(u64 *completer_addr,
+				 const struct pkvm_mem_transition *tx)
+{
+	u64 size = tx->nr_pages * PAGE_SIZE;
+	int ret;
+
+	*completer_addr = tx->initiator.hyp.completer_addr;
+	ret = kvm_pgtable_hyp_unmap(&pkvm_pgtable, tx->initiator.addr, size);
+	return (ret != size) ? -EFAULT : 0;
+}
+
 static bool __hyp_ack_skip_pgtable_check(const struct pkvm_mem_transition *tx)
 {
 	return !(IS_ENABLED(CONFIG_NVHE_EL2_DEBUG) ||
@@ -555,6 +767,9 @@ static int hyp_ack_unshare(u64 addr, const struct pkvm_mem_transition *tx)
 {
 	u64 size = tx->nr_pages * PAGE_SIZE;
 
+	if (tx->initiator.id == PKVM_ID_HOST && hyp_page_count((void *)addr))
+		return -EBUSY;
+
 	if (__hyp_ack_skip_pgtable_check(tx))
 		return 0;
 
@@ -562,6 +777,16 @@ static int hyp_ack_unshare(u64 addr, const struct pkvm_mem_transition *tx)
 					    PKVM_PAGE_SHARED_BORROWED);
 }
 
+static int hyp_ack_donation(u64 addr, const struct pkvm_mem_transition *tx)
+{
+	u64 size = tx->nr_pages * PAGE_SIZE;
+
+	if (__hyp_ack_skip_pgtable_check(tx))
+		return 0;
+
+	return __hyp_check_page_state_range(addr, size, PKVM_NOPAGE);
+}
+
 static int hyp_complete_share(u64 addr, const struct pkvm_mem_transition *tx,
 			      enum kvm_pgtable_prot perms)
 {
@@ -580,6 +805,15 @@ static int hyp_complete_unshare(u64 addr, const struct pkvm_mem_transition *tx)
 	return (ret != size) ? -EFAULT : 0;
 }
 
+static int hyp_complete_donation(u64 addr,
+				 const struct pkvm_mem_transition *tx)
+{
+	void *start = (void *)addr, *end = start + (tx->nr_pages * PAGE_SIZE);
+	enum kvm_pgtable_prot prot = pkvm_mkstate(PAGE_HYP, PKVM_PAGE_OWNED);
+
+	return pkvm_create_mappings_locked(start, end, prot);
+}
+
 static int check_share(struct pkvm_mem_share *share)
 {
 	const struct pkvm_mem_transition *tx = &share->tx;
@@ -732,6 +966,94 @@ static int do_unshare(struct pkvm_mem_share *share)
 	return WARN_ON(__do_unshare(share));
 }
 
+static int check_donation(struct pkvm_mem_donation *donation)
+{
+	const struct pkvm_mem_transition *tx = &donation->tx;
+	u64 completer_addr;
+	int ret;
+
+	switch (tx->initiator.id) {
+	case PKVM_ID_HOST:
+		ret = host_request_owned_transition(&completer_addr, tx);
+		break;
+	case PKVM_ID_HYP:
+		ret = hyp_request_donation(&completer_addr, tx);
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	if (ret)
+		return ret;
+
+	switch (tx->completer.id) {
+	case PKVM_ID_HOST:
+		ret = host_ack_donation(completer_addr, tx);
+		break;
+	case PKVM_ID_HYP:
+		ret = hyp_ack_donation(completer_addr, tx);
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	return ret;
+}
+
+static int __do_donate(struct pkvm_mem_donation *donation)
+{
+	const struct pkvm_mem_transition *tx = &donation->tx;
+	u64 completer_addr;
+	int ret;
+
+	switch (tx->initiator.id) {
+	case PKVM_ID_HOST:
+		ret = host_initiate_donation(&completer_addr, tx);
+		break;
+	case PKVM_ID_HYP:
+		ret = hyp_initiate_donation(&completer_addr, tx);
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	if (ret)
+		return ret;
+
+	switch (tx->completer.id) {
+	case PKVM_ID_HOST:
+		ret = host_complete_donation(completer_addr, tx);
+		break;
+	case PKVM_ID_HYP:
+		ret = hyp_complete_donation(completer_addr, tx);
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	return ret;
+}
+
+/*
+ * do_donate():
+ *
+ * The page owner transfers ownership to another component, losing access
+ * as a consequence.
+ *
+ * Initiator: OWNED	=> NOPAGE
+ * Completer: NOPAGE	=> OWNED
+ */
+static int do_donate(struct pkvm_mem_donation *donation)
+{
+	int ret;
+
+	ret = check_donation(donation);
+	if (ret)
+		return ret;
+
+	return WARN_ON(__do_donate(donation));
+}
+
 int __pkvm_host_share_hyp(u64 pfn)
 {
 	int ret;
@@ -797,3 +1119,112 @@ int __pkvm_host_unshare_hyp(u64 pfn)
 
 	return ret;
 }
+
+int __pkvm_host_donate_hyp(u64 pfn, u64 nr_pages)
+{
+	int ret;
+	u64 host_addr = hyp_pfn_to_phys(pfn);
+	u64 hyp_addr = (u64)__hyp_va(host_addr);
+	struct pkvm_mem_donation donation = {
+		.tx	= {
+			.nr_pages	= nr_pages,
+			.initiator	= {
+				.id	= PKVM_ID_HOST,
+				.addr	= host_addr,
+				.host	= {
+					.completer_addr = hyp_addr,
+				},
+			},
+			.completer	= {
+				.id	= PKVM_ID_HYP,
+			},
+		},
+	};
+
+	host_lock_component();
+	hyp_lock_component();
+
+	ret = do_donate(&donation);
+
+	hyp_unlock_component();
+	host_unlock_component();
+
+	return ret;
+}
+
+int __pkvm_hyp_donate_host(u64 pfn, u64 nr_pages)
+{
+	int ret;
+	u64 host_addr = hyp_pfn_to_phys(pfn);
+	u64 hyp_addr = (u64)__hyp_va(host_addr);
+	struct pkvm_mem_donation donation = {
+		.tx	= {
+			.nr_pages	= nr_pages,
+			.initiator	= {
+				.id	= PKVM_ID_HYP,
+				.addr	= hyp_addr,
+				.hyp	= {
+					.completer_addr = host_addr,
+				},
+			},
+			.completer	= {
+				.id	= PKVM_ID_HOST,
+			},
+		},
+	};
+
+	host_lock_component();
+	hyp_lock_component();
+
+	ret = do_donate(&donation);
+
+	hyp_unlock_component();
+	host_unlock_component();
+
+	return ret;
+}
+
+int hyp_pin_shared_mem(void *from, void *to)
+{
+	u64 cur, start = ALIGN_DOWN((u64)from, PAGE_SIZE);
+	u64 end = PAGE_ALIGN((u64)to);
+	u64 size = end - start;
+	int ret;
+
+	host_lock_component();
+	hyp_lock_component();
+
+	ret = __host_check_page_state_range(__hyp_pa(start), size,
+					    PKVM_PAGE_SHARED_OWNED);
+	if (ret)
+		goto unlock;
+
+	ret = __hyp_check_page_state_range(start, size,
+					   PKVM_PAGE_SHARED_BORROWED);
+	if (ret)
+		goto unlock;
+
+	for (cur = start; cur < end; cur += PAGE_SIZE)
+		hyp_page_ref_inc(hyp_virt_to_page(cur));
+
+unlock:
+	hyp_unlock_component();
+	host_unlock_component();
+
+	return ret;
+}
+
+void hyp_unpin_shared_mem(void *from, void *to)
+{
+	u64 cur, start = ALIGN_DOWN((u64)from, PAGE_SIZE);
+	u64 end = PAGE_ALIGN((u64)to);
+
+	host_lock_component();
+	hyp_lock_component();
+
+	for (cur = start; cur < end; cur += PAGE_SIZE)
+		hyp_page_ref_dec(hyp_virt_to_page(cur));
+
+	hyp_unlock_component();
+	host_unlock_component();
+}
diff --git a/arch/arm64/kvm/hyp/nvhe/mm.c b/arch/arm64/kvm/hyp/nvhe/mm.c
index 96193cb31a39..318298eb3d6b 100644
--- a/arch/arm64/kvm/hyp/nvhe/mm.c
+++ b/arch/arm64/kvm/hyp/nvhe/mm.c
@@ -14,6 +14,7 @@
 #include <nvhe/early_alloc.h>
 #include <nvhe/gfp.h>
 #include <nvhe/memory.h>
+#include <nvhe/mem_protect.h>
 #include <nvhe/mm.h>
 #include <nvhe/spinlock.h>
 
@@ -25,6 +26,12 @@ unsigned int hyp_memblock_nr;
 
 static u64 __io_map_base;
 
+struct hyp_fixmap_slot {
+	u64 addr;
+	kvm_pte_t *ptep;
+};
+static DEFINE_PER_CPU(struct hyp_fixmap_slot, fixmap_slots);
+
 static int __pkvm_create_mappings(unsigned long start, unsigned long size,
 				  unsigned long phys, enum kvm_pgtable_prot prot)
 {
@@ -129,13 +136,36 @@ int pkvm_create_mappings(void *from, void *to, enum kvm_pgtable_prot prot)
 	return ret;
 }
 
-int hyp_back_vmemmap(phys_addr_t phys, unsigned long size, phys_addr_t back)
+int hyp_back_vmemmap(phys_addr_t back)
 {
-	unsigned long start, end;
+	unsigned long i, start, size, end = 0;
+	int ret;
 
-	hyp_vmemmap_range(phys, size, &start, &end);
+	for (i = 0; i < hyp_memblock_nr; i++) {
+		start = hyp_memory[i].base;
+		start = ALIGN_DOWN((u64)hyp_phys_to_page(start), PAGE_SIZE);
+		/*
+		 * The begining of the hyp_vmemmap region for the current
+		 * memblock may already be backed by the page backing the end
+		 * the previous region, so avoid mapping it twice.
+		 */
+		start = max(start, end);
+
+		end = hyp_memory[i].base + hyp_memory[i].size;
+		end = PAGE_ALIGN((u64)hyp_phys_to_page(end));
+		if (start >= end)
+			continue;
+
+		size = end - start;
+		ret = __pkvm_create_mappings(start, size, back, PAGE_HYP);
+		if (ret)
+			return ret;
+
+		memset(hyp_phys_to_virt(back), 0, size);
+		back += size;
+	}
 
-	return __pkvm_create_mappings(start, end - start, back, PAGE_HYP);
+	return 0;
 }
 
 static void *__hyp_bp_vect_base;
@@ -189,6 +219,102 @@ int hyp_map_vectors(void)
 	return 0;
 }
 
+void *hyp_fixmap_map(phys_addr_t phys)
+{
+	struct hyp_fixmap_slot *slot = this_cpu_ptr(&fixmap_slots);
+	kvm_pte_t pte, *ptep = slot->ptep;
+
+	pte = *ptep;
+	pte &= ~kvm_phys_to_pte(KVM_PHYS_INVALID);
+	pte |= kvm_phys_to_pte(phys) | KVM_PTE_VALID;
+	WRITE_ONCE(*ptep, pte);
+	dsb(ishst);
+
+	return (void *)slot->addr;
+}
+
+static void fixmap_clear_slot(struct hyp_fixmap_slot *slot)
+{
+	kvm_pte_t *ptep = slot->ptep;
+	u64 addr = slot->addr;
+
+	WRITE_ONCE(*ptep, *ptep & ~KVM_PTE_VALID);
+
+	/*
+	 * Irritatingly, the architecture requires that we use inner-shareable
+	 * broadcast TLB invalidation here in case another CPU speculates
+	 * through our fixmap and decides to create an "amalagamation of the
+	 * values held in the TLB" due to the apparent lack of a
+	 * break-before-make sequence.
+	 *
+	 * https://lore.kernel.org/kvm/20221017115209.2099-1-will@kernel.org/T/#mf10dfbaf1eaef9274c581b81c53758918c1d0f03
+	 */
+	dsb(ishst);
+	__tlbi_level(vale2is, __TLBI_VADDR(addr, 0), (KVM_PGTABLE_MAX_LEVELS - 1));
+	dsb(ish);
+	isb();
+}
+
+void hyp_fixmap_unmap(void)
+{
+	fixmap_clear_slot(this_cpu_ptr(&fixmap_slots));
+}
+
+static int __create_fixmap_slot_cb(const struct kvm_pgtable_visit_ctx *ctx,
+				   enum kvm_pgtable_walk_flags visit)
+{
+	struct hyp_fixmap_slot *slot = per_cpu_ptr(&fixmap_slots, (u64)ctx->arg);
+
+	if (!kvm_pte_valid(ctx->old) || ctx->level != KVM_PGTABLE_MAX_LEVELS - 1)
+		return -EINVAL;
+
+	slot->addr = ctx->addr;
+	slot->ptep = ctx->ptep;
+
+	/*
+	 * Clear the PTE, but keep the page-table page refcount elevated to
+	 * prevent it from ever being freed. This lets us manipulate the PTEs
+	 * by hand safely without ever needing to allocate memory.
+	 */
+	fixmap_clear_slot(slot);
+
+	return 0;
+}
+
+static int create_fixmap_slot(u64 addr, u64 cpu)
+{
+	struct kvm_pgtable_walker walker = {
+		.cb	= __create_fixmap_slot_cb,
+		.flags	= KVM_PGTABLE_WALK_LEAF,
+		.arg = (void *)cpu,
+	};
+
+	return kvm_pgtable_walk(&pkvm_pgtable, addr, PAGE_SIZE, &walker);
+}
+
+int hyp_create_pcpu_fixmap(void)
+{
+	unsigned long addr, i;
+	int ret;
+
+	for (i = 0; i < hyp_nr_cpus; i++) {
+		ret = pkvm_alloc_private_va_range(PAGE_SIZE, &addr);
+		if (ret)
+			return ret;
+
+		ret = kvm_pgtable_hyp_map(&pkvm_pgtable, addr, PAGE_SIZE,
+					  __hyp_pa(__hyp_bss_start), PAGE_HYP);
+		if (ret)
+			return ret;
+
+		ret = create_fixmap_slot(addr, i);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
 int hyp_create_idmap(u32 hyp_va_bits)
 {
 	unsigned long start, end;
@@ -213,3 +339,36 @@ int hyp_create_idmap(u32 hyp_va_bits)
 
 	return __pkvm_create_mappings(start, end - start, start, PAGE_HYP_EXEC);
 }
+
+static void *admit_host_page(void *arg)
+{
+	struct kvm_hyp_memcache *host_mc = arg;
+
+	if (!host_mc->nr_pages)
+		return NULL;
+
+	/*
+	 * The host still owns the pages in its memcache, so we need to go
+	 * through a full host-to-hyp donation cycle to change it. Fortunately,
+	 * __pkvm_host_donate_hyp() takes care of races for us, so if it
+	 * succeeds we're good to go.
+	 */
+	if (__pkvm_host_donate_hyp(hyp_phys_to_pfn(host_mc->head), 1))
+		return NULL;
+
+	return pop_hyp_memcache(host_mc, hyp_phys_to_virt);
+}
+
+/* Refill our local memcache by poping pages from the one provided by the host. */
+int refill_memcache(struct kvm_hyp_memcache *mc, unsigned long min_pages,
+		    struct kvm_hyp_memcache *host_mc)
+{
+	struct kvm_hyp_memcache tmp = *host_mc;
+	int ret;
+
+	ret =  __topup_hyp_memcache(mc, min_pages, admit_host_page,
+				    hyp_virt_to_phys, &tmp);
+	*host_mc = tmp;
+
+	return ret;
+}
diff --git a/arch/arm64/kvm/hyp/nvhe/page_alloc.c b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
index d40f0b30b534..803ba3222e75 100644
--- a/arch/arm64/kvm/hyp/nvhe/page_alloc.c
+++ b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
@@ -93,11 +93,16 @@ static inline struct hyp_page *node_to_page(struct list_head *node)
 static void __hyp_attach_page(struct hyp_pool *pool,
 			      struct hyp_page *p)
 {
+	phys_addr_t phys = hyp_page_to_phys(p);
 	unsigned short order = p->order;
 	struct hyp_page *buddy;
 
 	memset(hyp_page_to_virt(p), 0, PAGE_SIZE << p->order);
 
+	/* Skip coalescing for 'external' pages being freed into the pool. */
+	if (phys < pool->range_start || phys >= pool->range_end)
+		goto insert;
+
 	/*
 	 * Only the first struct hyp_page of a high-order page (otherwise known
 	 * as the 'head') should have p->order set. The non-head pages should
@@ -116,6 +121,7 @@ static void __hyp_attach_page(struct hyp_pool *pool,
 		p = min(p, buddy);
 	}
 
+insert:
 	/* Mark the new head, and insert it */
 	p->order = order;
 	page_add_to_list(p, &pool->free_area[order]);
@@ -144,25 +150,6 @@ static struct hyp_page *__hyp_extract_page(struct hyp_pool *pool,
 	return p;
 }
 
-static inline void hyp_page_ref_inc(struct hyp_page *p)
-{
-	BUG_ON(p->refcount == USHRT_MAX);
-	p->refcount++;
-}
-
-static inline int hyp_page_ref_dec_and_test(struct hyp_page *p)
-{
-	BUG_ON(!p->refcount);
-	p->refcount--;
-	return (p->refcount == 0);
-}
-
-static inline void hyp_set_page_refcounted(struct hyp_page *p)
-{
-	BUG_ON(p->refcount);
-	p->refcount = 1;
-}
-
 static void __hyp_put_page(struct hyp_pool *pool, struct hyp_page *p)
 {
 	if (hyp_page_ref_dec_and_test(p))
@@ -249,10 +236,8 @@ int hyp_pool_init(struct hyp_pool *pool, u64 pfn, unsigned int nr_pages,
 
 	/* Init the vmemmap portion */
 	p = hyp_phys_to_page(phys);
-	for (i = 0; i < nr_pages; i++) {
-		p[i].order = 0;
+	for (i = 0; i < nr_pages; i++)
 		hyp_set_page_refcounted(&p[i]);
-	}
 
 	/* Attach the unused pages to the buddy tree */
 	for (i = reserved_pages; i < nr_pages; i++)
diff --git a/arch/arm64/kvm/hyp/nvhe/pkvm.c b/arch/arm64/kvm/hyp/nvhe/pkvm.c
index 85d3b7ae720f..a06ece14a6d8 100644
--- a/arch/arm64/kvm/hyp/nvhe/pkvm.c
+++ b/arch/arm64/kvm/hyp/nvhe/pkvm.c
@@ -7,8 +7,17 @@
 #include <linux/kvm_host.h>
 #include <linux/mm.h>
 #include <nvhe/fixed_config.h>
+#include <nvhe/mem_protect.h>
+#include <nvhe/memory.h>
+#include <nvhe/pkvm.h>
 #include <nvhe/trap_handler.h>
 
+/* Used by icache_is_vpipt(). */
+unsigned long __icache_flags;
+
+/* Used by kvm_get_vttbr(). */
+unsigned int kvm_arm_vmid_bits;
+
 /*
  * Set trap register values based on features in ID_AA64PFR0.
  */
@@ -183,3 +192,430 @@ void __pkvm_vcpu_init_traps(struct kvm_vcpu *vcpu)
 	pvm_init_traps_aa64mmfr0(vcpu);
 	pvm_init_traps_aa64mmfr1(vcpu);
 }
+
+/*
+ * Start the VM table handle at the offset defined instead of at 0.
+ * Mainly for sanity checking and debugging.
+ */
+#define HANDLE_OFFSET 0x1000
+
+static unsigned int vm_handle_to_idx(pkvm_handle_t handle)
+{
+	return handle - HANDLE_OFFSET;
+}
+
+static pkvm_handle_t idx_to_vm_handle(unsigned int idx)
+{
+	return idx + HANDLE_OFFSET;
+}
+
+/*
+ * Spinlock for protecting state related to the VM table. Protects writes
+ * to 'vm_table' and 'nr_table_entries' as well as reads and writes to
+ * 'last_hyp_vcpu_lookup'.
+ */
+static DEFINE_HYP_SPINLOCK(vm_table_lock);
+
+/*
+ * The table of VM entries for protected VMs in hyp.
+ * Allocated at hyp initialization and setup.
+ */
+static struct pkvm_hyp_vm **vm_table;
+
+void pkvm_hyp_vm_table_init(void *tbl)
+{
+	WARN_ON(vm_table);
+	vm_table = tbl;
+}
+
+/*
+ * Return the hyp vm structure corresponding to the handle.
+ */
+static struct pkvm_hyp_vm *get_vm_by_handle(pkvm_handle_t handle)
+{
+	unsigned int idx = vm_handle_to_idx(handle);
+
+	if (unlikely(idx >= KVM_MAX_PVMS))
+		return NULL;
+
+	return vm_table[idx];
+}
+
+struct pkvm_hyp_vcpu *pkvm_load_hyp_vcpu(pkvm_handle_t handle,
+					 unsigned int vcpu_idx)
+{
+	struct pkvm_hyp_vcpu *hyp_vcpu = NULL;
+	struct pkvm_hyp_vm *hyp_vm;
+
+	hyp_spin_lock(&vm_table_lock);
+	hyp_vm = get_vm_by_handle(handle);
+	if (!hyp_vm || hyp_vm->nr_vcpus <= vcpu_idx)
+		goto unlock;
+
+	hyp_vcpu = hyp_vm->vcpus[vcpu_idx];
+	hyp_page_ref_inc(hyp_virt_to_page(hyp_vm));
+unlock:
+	hyp_spin_unlock(&vm_table_lock);
+	return hyp_vcpu;
+}
+
+void pkvm_put_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
+{
+	struct pkvm_hyp_vm *hyp_vm = pkvm_hyp_vcpu_to_hyp_vm(hyp_vcpu);
+
+	hyp_spin_lock(&vm_table_lock);
+	hyp_page_ref_dec(hyp_virt_to_page(hyp_vm));
+	hyp_spin_unlock(&vm_table_lock);
+}
+
+static void unpin_host_vcpu(struct kvm_vcpu *host_vcpu)
+{
+	if (host_vcpu)
+		hyp_unpin_shared_mem(host_vcpu, host_vcpu + 1);
+}
+
+static void unpin_host_vcpus(struct pkvm_hyp_vcpu *hyp_vcpus[],
+			     unsigned int nr_vcpus)
+{
+	int i;
+
+	for (i = 0; i < nr_vcpus; i++)
+		unpin_host_vcpu(hyp_vcpus[i]->host_vcpu);
+}
+
+static void init_pkvm_hyp_vm(struct kvm *host_kvm, struct pkvm_hyp_vm *hyp_vm,
+			     unsigned int nr_vcpus)
+{
+	hyp_vm->host_kvm = host_kvm;
+	hyp_vm->kvm.created_vcpus = nr_vcpus;
+	hyp_vm->kvm.arch.vtcr = host_mmu.arch.vtcr;
+}
+
+static int init_pkvm_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu,
+			      struct pkvm_hyp_vm *hyp_vm,
+			      struct kvm_vcpu *host_vcpu,
+			      unsigned int vcpu_idx)
+{
+	int ret = 0;
+
+	if (hyp_pin_shared_mem(host_vcpu, host_vcpu + 1))
+		return -EBUSY;
+
+	if (host_vcpu->vcpu_idx != vcpu_idx) {
+		ret = -EINVAL;
+		goto done;
+	}
+
+	hyp_vcpu->host_vcpu = host_vcpu;
+
+	hyp_vcpu->vcpu.kvm = &hyp_vm->kvm;
+	hyp_vcpu->vcpu.vcpu_id = READ_ONCE(host_vcpu->vcpu_id);
+	hyp_vcpu->vcpu.vcpu_idx = vcpu_idx;
+
+	hyp_vcpu->vcpu.arch.hw_mmu = &hyp_vm->kvm.arch.mmu;
+	hyp_vcpu->vcpu.arch.cflags = READ_ONCE(host_vcpu->arch.cflags);
+done:
+	if (ret)
+		unpin_host_vcpu(host_vcpu);
+	return ret;
+}
+
+static int find_free_vm_table_entry(struct kvm *host_kvm)
+{
+	int i;
+
+	for (i = 0; i < KVM_MAX_PVMS; ++i) {
+		if (!vm_table[i])
+			return i;
+	}
+
+	return -ENOMEM;
+}
+
+/*
+ * Allocate a VM table entry and insert a pointer to the new vm.
+ *
+ * Return a unique handle to the protected VM on success,
+ * negative error code on failure.
+ */
+static pkvm_handle_t insert_vm_table_entry(struct kvm *host_kvm,
+					   struct pkvm_hyp_vm *hyp_vm)
+{
+	struct kvm_s2_mmu *mmu = &hyp_vm->kvm.arch.mmu;
+	int idx;
+
+	hyp_assert_lock_held(&vm_table_lock);
+
+	/*
+	 * Initializing protected state might have failed, yet a malicious
+	 * host could trigger this function. Thus, ensure that 'vm_table'
+	 * exists.
+	 */
+	if (unlikely(!vm_table))
+		return -EINVAL;
+
+	idx = find_free_vm_table_entry(host_kvm);
+	if (idx < 0)
+		return idx;
+
+	hyp_vm->kvm.arch.pkvm.handle = idx_to_vm_handle(idx);
+
+	/* VMID 0 is reserved for the host */
+	atomic64_set(&mmu->vmid.id, idx + 1);
+
+	mmu->arch = &hyp_vm->kvm.arch;
+	mmu->pgt = &hyp_vm->pgt;
+
+	vm_table[idx] = hyp_vm;
+	return hyp_vm->kvm.arch.pkvm.handle;
+}
+
+/*
+ * Deallocate and remove the VM table entry corresponding to the handle.
+ */
+static void remove_vm_table_entry(pkvm_handle_t handle)
+{
+	hyp_assert_lock_held(&vm_table_lock);
+	vm_table[vm_handle_to_idx(handle)] = NULL;
+}
+
+static size_t pkvm_get_hyp_vm_size(unsigned int nr_vcpus)
+{
+	return size_add(sizeof(struct pkvm_hyp_vm),
+		size_mul(sizeof(struct pkvm_hyp_vcpu *), nr_vcpus));
+}
+
+static void *map_donated_memory_noclear(unsigned long host_va, size_t size)
+{
+	void *va = (void *)kern_hyp_va(host_va);
+
+	if (!PAGE_ALIGNED(va))
+		return NULL;
+
+	if (__pkvm_host_donate_hyp(hyp_virt_to_pfn(va),
+				   PAGE_ALIGN(size) >> PAGE_SHIFT))
+		return NULL;
+
+	return va;
+}
+
+static void *map_donated_memory(unsigned long host_va, size_t size)
+{
+	void *va = map_donated_memory_noclear(host_va, size);
+
+	if (va)
+		memset(va, 0, size);
+
+	return va;
+}
+
+static void __unmap_donated_memory(void *va, size_t size)
+{
+	WARN_ON(__pkvm_hyp_donate_host(hyp_virt_to_pfn(va),
+				       PAGE_ALIGN(size) >> PAGE_SHIFT));
+}
+
+static void unmap_donated_memory(void *va, size_t size)
+{
+	if (!va)
+		return;
+
+	memset(va, 0, size);
+	__unmap_donated_memory(va, size);
+}
+
+static void unmap_donated_memory_noclear(void *va, size_t size)
+{
+	if (!va)
+		return;
+
+	__unmap_donated_memory(va, size);
+}
+
+/*
+ * Initialize the hypervisor copy of the protected VM state using the
+ * memory donated by the host.
+ *
+ * Unmaps the donated memory from the host at stage 2.
+ *
+ * host_kvm: A pointer to the host's struct kvm.
+ * vm_hva: The host va of the area being donated for the VM state.
+ *	   Must be page aligned.
+ * pgd_hva: The host va of the area being donated for the stage-2 PGD for
+ *	    the VM. Must be page aligned. Its size is implied by the VM's
+ *	    VTCR.
+ *
+ * Return a unique handle to the protected VM on success,
+ * negative error code on failure.
+ */
+int __pkvm_init_vm(struct kvm *host_kvm, unsigned long vm_hva,
+		   unsigned long pgd_hva)
+{
+	struct pkvm_hyp_vm *hyp_vm = NULL;
+	size_t vm_size, pgd_size;
+	unsigned int nr_vcpus;
+	void *pgd = NULL;
+	int ret;
+
+	ret = hyp_pin_shared_mem(host_kvm, host_kvm + 1);
+	if (ret)
+		return ret;
+
+	nr_vcpus = READ_ONCE(host_kvm->created_vcpus);
+	if (nr_vcpus < 1) {
+		ret = -EINVAL;
+		goto err_unpin_kvm;
+	}
+
+	vm_size = pkvm_get_hyp_vm_size(nr_vcpus);
+	pgd_size = kvm_pgtable_stage2_pgd_size(host_mmu.arch.vtcr);
+
+	ret = -ENOMEM;
+
+	hyp_vm = map_donated_memory(vm_hva, vm_size);
+	if (!hyp_vm)
+		goto err_remove_mappings;
+
+	pgd = map_donated_memory_noclear(pgd_hva, pgd_size);
+	if (!pgd)
+		goto err_remove_mappings;
+
+	init_pkvm_hyp_vm(host_kvm, hyp_vm, nr_vcpus);
+
+	hyp_spin_lock(&vm_table_lock);
+	ret = insert_vm_table_entry(host_kvm, hyp_vm);
+	if (ret < 0)
+		goto err_unlock;
+
+	ret = kvm_guest_prepare_stage2(hyp_vm, pgd);
+	if (ret)
+		goto err_remove_vm_table_entry;
+	hyp_spin_unlock(&vm_table_lock);
+
+	return hyp_vm->kvm.arch.pkvm.handle;
+
+err_remove_vm_table_entry:
+	remove_vm_table_entry(hyp_vm->kvm.arch.pkvm.handle);
+err_unlock:
+	hyp_spin_unlock(&vm_table_lock);
+err_remove_mappings:
+	unmap_donated_memory(hyp_vm, vm_size);
+	unmap_donated_memory(pgd, pgd_size);
+err_unpin_kvm:
+	hyp_unpin_shared_mem(host_kvm, host_kvm + 1);
+	return ret;
+}
+
+/*
+ * Initialize the hypervisor copy of the protected vCPU state using the
+ * memory donated by the host.
+ *
+ * handle: The handle for the protected vm.
+ * host_vcpu: A pointer to the corresponding host vcpu.
+ * vcpu_hva: The host va of the area being donated for the vcpu state.
+ *	     Must be page aligned. The size of the area must be equal to
+ *	     the page-aligned size of 'struct pkvm_hyp_vcpu'.
+ * Return 0 on success, negative error code on failure.
+ */
+int __pkvm_init_vcpu(pkvm_handle_t handle, struct kvm_vcpu *host_vcpu,
+		     unsigned long vcpu_hva)
+{
+	struct pkvm_hyp_vcpu *hyp_vcpu;
+	struct pkvm_hyp_vm *hyp_vm;
+	unsigned int idx;
+	int ret;
+
+	hyp_vcpu = map_donated_memory(vcpu_hva, sizeof(*hyp_vcpu));
+	if (!hyp_vcpu)
+		return -ENOMEM;
+
+	hyp_spin_lock(&vm_table_lock);
+
+	hyp_vm = get_vm_by_handle(handle);
+	if (!hyp_vm) {
+		ret = -ENOENT;
+		goto unlock;
+	}
+
+	idx = hyp_vm->nr_vcpus;
+	if (idx >= hyp_vm->kvm.created_vcpus) {
+		ret = -EINVAL;
+		goto unlock;
+	}
+
+	ret = init_pkvm_hyp_vcpu(hyp_vcpu, hyp_vm, host_vcpu, idx);
+	if (ret)
+		goto unlock;
+
+	hyp_vm->vcpus[idx] = hyp_vcpu;
+	hyp_vm->nr_vcpus++;
+unlock:
+	hyp_spin_unlock(&vm_table_lock);
+
+	if (ret)
+		unmap_donated_memory(hyp_vcpu, sizeof(*hyp_vcpu));
+
+	return ret;
+}
+
+static void
+teardown_donated_memory(struct kvm_hyp_memcache *mc, void *addr, size_t size)
+{
+	size = PAGE_ALIGN(size);
+	memset(addr, 0, size);
+
+	for (void *start = addr; start < addr + size; start += PAGE_SIZE)
+		push_hyp_memcache(mc, start, hyp_virt_to_phys);
+
+	unmap_donated_memory_noclear(addr, size);
+}
+
+int __pkvm_teardown_vm(pkvm_handle_t handle)
+{
+	struct kvm_hyp_memcache *mc;
+	struct pkvm_hyp_vm *hyp_vm;
+	struct kvm *host_kvm;
+	unsigned int idx;
+	size_t vm_size;
+	int err;
+
+	hyp_spin_lock(&vm_table_lock);
+	hyp_vm = get_vm_by_handle(handle);
+	if (!hyp_vm) {
+		err = -ENOENT;
+		goto err_unlock;
+	}
+
+	if (WARN_ON(hyp_page_count(hyp_vm))) {
+		err = -EBUSY;
+		goto err_unlock;
+	}
+
+	host_kvm = hyp_vm->host_kvm;
+
+	/* Ensure the VMID is clean before it can be reallocated */
+	__kvm_tlb_flush_vmid(&hyp_vm->kvm.arch.mmu);
+	remove_vm_table_entry(handle);
+	hyp_spin_unlock(&vm_table_lock);
+
+	/* Reclaim guest pages (including page-table pages) */
+	mc = &host_kvm->arch.pkvm.teardown_mc;
+	reclaim_guest_pages(hyp_vm, mc);
+	unpin_host_vcpus(hyp_vm->vcpus, hyp_vm->nr_vcpus);
+
+	/* Push the metadata pages to the teardown memcache */
+	for (idx = 0; idx < hyp_vm->nr_vcpus; ++idx) {
+		struct pkvm_hyp_vcpu *hyp_vcpu = hyp_vm->vcpus[idx];
+
+		teardown_donated_memory(mc, hyp_vcpu, sizeof(*hyp_vcpu));
+	}
+
+	vm_size = pkvm_get_hyp_vm_size(hyp_vm->kvm.created_vcpus);
+	teardown_donated_memory(mc, hyp_vm, vm_size);
+	hyp_unpin_shared_mem(host_kvm, host_kvm + 1);
+	return 0;
+
+err_unlock:
+	hyp_spin_unlock(&vm_table_lock);
+	return err;
+}
diff --git a/arch/arm64/kvm/hyp/nvhe/setup.c b/arch/arm64/kvm/hyp/nvhe/setup.c
index e8d4ea2fcfa0..110f04627785 100644
--- a/arch/arm64/kvm/hyp/nvhe/setup.c
+++ b/arch/arm64/kvm/hyp/nvhe/setup.c
@@ -16,6 +16,7 @@
 #include <nvhe/memory.h>
 #include <nvhe/mem_protect.h>
 #include <nvhe/mm.h>
+#include <nvhe/pkvm.h>
 #include <nvhe/trap_handler.h>
 
 unsigned long hyp_nr_cpus;
@@ -24,6 +25,7 @@ unsigned long hyp_nr_cpus;
 			 (unsigned long)__per_cpu_start)
 
 static void *vmemmap_base;
+static void *vm_table_base;
 static void *hyp_pgt_base;
 static void *host_s2_pgt_base;
 static struct kvm_pgtable_mm_ops pkvm_pgtable_mm_ops;
@@ -31,16 +33,20 @@ static struct hyp_pool hpool;
 
 static int divide_memory_pool(void *virt, unsigned long size)
 {
-	unsigned long vstart, vend, nr_pages;
+	unsigned long nr_pages;
 
 	hyp_early_alloc_init(virt, size);
 
-	hyp_vmemmap_range(__hyp_pa(virt), size, &vstart, &vend);
-	nr_pages = (vend - vstart) >> PAGE_SHIFT;
+	nr_pages = hyp_vmemmap_pages(sizeof(struct hyp_page));
 	vmemmap_base = hyp_early_alloc_contig(nr_pages);
 	if (!vmemmap_base)
 		return -ENOMEM;
 
+	nr_pages = hyp_vm_table_pages();
+	vm_table_base = hyp_early_alloc_contig(nr_pages);
+	if (!vm_table_base)
+		return -ENOMEM;
+
 	nr_pages = hyp_s1_pgtable_pages();
 	hyp_pgt_base = hyp_early_alloc_contig(nr_pages);
 	if (!hyp_pgt_base)
@@ -78,7 +84,7 @@ static int recreate_hyp_mappings(phys_addr_t phys, unsigned long size,
 	if (ret)
 		return ret;
 
-	ret = hyp_back_vmemmap(phys, size, hyp_virt_to_phys(vmemmap_base));
+	ret = hyp_back_vmemmap(hyp_virt_to_phys(vmemmap_base));
 	if (ret)
 		return ret;
 
@@ -138,20 +144,17 @@ static int recreate_hyp_mappings(phys_addr_t phys, unsigned long size,
 	}
 
 	/*
-	 * Map the host's .bss and .rodata sections RO in the hypervisor, but
-	 * transfer the ownership from the host to the hypervisor itself to
-	 * make sure it can't be donated or shared with another entity.
+	 * Map the host sections RO in the hypervisor, but transfer the
+	 * ownership from the host to the hypervisor itself to make sure they
+	 * can't be donated or shared with another entity.
 	 *
 	 * The ownership transition requires matching changes in the host
 	 * stage-2. This will be done later (see finalize_host_mappings()) once
 	 * the hyp_vmemmap is addressable.
 	 */
 	prot = pkvm_mkstate(PAGE_HYP_RO, PKVM_PAGE_SHARED_OWNED);
-	ret = pkvm_create_mappings(__start_rodata, __end_rodata, prot);
-	if (ret)
-		return ret;
-
-	ret = pkvm_create_mappings(__hyp_bss_end, __bss_stop, prot);
+	ret = pkvm_create_mappings(&kvm_vgic_global_state,
+				   &kvm_vgic_global_state + 1, prot);
 	if (ret)
 		return ret;
 
@@ -186,33 +189,20 @@ static void hpool_put_page(void *addr)
 	hyp_put_page(&hpool, addr);
 }
 
-static int finalize_host_mappings_walker(u64 addr, u64 end, u32 level,
-					 kvm_pte_t *ptep,
-					 enum kvm_pgtable_walk_flags flag,
-					 void * const arg)
+static int fix_host_ownership_walker(const struct kvm_pgtable_visit_ctx *ctx,
+				     enum kvm_pgtable_walk_flags visit)
 {
-	struct kvm_pgtable_mm_ops *mm_ops = arg;
 	enum kvm_pgtable_prot prot;
 	enum pkvm_page_state state;
-	kvm_pte_t pte = *ptep;
 	phys_addr_t phys;
 
-	if (!kvm_pte_valid(pte))
+	if (!kvm_pte_valid(ctx->old))
 		return 0;
 
-	/*
-	 * Fix-up the refcount for the page-table pages as the early allocator
-	 * was unable to access the hyp_vmemmap and so the buddy allocator has
-	 * initialised the refcount to '1'.
-	 */
-	mm_ops->get_page(ptep);
-	if (flag != KVM_PGTABLE_WALK_LEAF)
-		return 0;
-
-	if (level != (KVM_PGTABLE_MAX_LEVELS - 1))
+	if (ctx->level != (KVM_PGTABLE_MAX_LEVELS - 1))
 		return -EINVAL;
 
-	phys = kvm_pte_to_phys(pte);
+	phys = kvm_pte_to_phys(ctx->old);
 	if (!addr_is_memory(phys))
 		return -EINVAL;
 
@@ -220,10 +210,10 @@ static int finalize_host_mappings_walker(u64 addr, u64 end, u32 level,
 	 * Adjust the host stage-2 mappings to match the ownership attributes
 	 * configured in the hypervisor stage-1.
 	 */
-	state = pkvm_getstate(kvm_pgtable_hyp_pte_prot(pte));
+	state = pkvm_getstate(kvm_pgtable_hyp_pte_prot(ctx->old));
 	switch (state) {
 	case PKVM_PAGE_OWNED:
-		return host_stage2_set_owner_locked(phys, PAGE_SIZE, pkvm_hyp_id);
+		return host_stage2_set_owner_locked(phys, PAGE_SIZE, PKVM_ID_HYP);
 	case PKVM_PAGE_SHARED_OWNED:
 		prot = pkvm_mkstate(PKVM_HOST_MEM_PROT, PKVM_PAGE_SHARED_BORROWED);
 		break;
@@ -237,12 +227,25 @@ static int finalize_host_mappings_walker(u64 addr, u64 end, u32 level,
 	return host_stage2_idmap_locked(phys, PAGE_SIZE, prot);
 }
 
-static int finalize_host_mappings(void)
+static int fix_hyp_pgtable_refcnt_walker(const struct kvm_pgtable_visit_ctx *ctx,
+					 enum kvm_pgtable_walk_flags visit)
+{
+	/*
+	 * Fix-up the refcount for the page-table pages as the early allocator
+	 * was unable to access the hyp_vmemmap and so the buddy allocator has
+	 * initialised the refcount to '1'.
+	 */
+	if (kvm_pte_valid(ctx->old))
+		ctx->mm_ops->get_page(ctx->ptep);
+
+	return 0;
+}
+
+static int fix_host_ownership(void)
 {
 	struct kvm_pgtable_walker walker = {
-		.cb	= finalize_host_mappings_walker,
-		.flags	= KVM_PGTABLE_WALK_LEAF | KVM_PGTABLE_WALK_TABLE_POST,
-		.arg	= pkvm_pgtable.mm_ops,
+		.cb	= fix_host_ownership_walker,
+		.flags	= KVM_PGTABLE_WALK_LEAF,
 	};
 	int i, ret;
 
@@ -258,6 +261,18 @@ static int finalize_host_mappings(void)
 	return 0;
 }
 
+static int fix_hyp_pgtable_refcnt(void)
+{
+	struct kvm_pgtable_walker walker = {
+		.cb	= fix_hyp_pgtable_refcnt_walker,
+		.flags	= KVM_PGTABLE_WALK_LEAF | KVM_PGTABLE_WALK_TABLE_POST,
+		.arg	= pkvm_pgtable.mm_ops,
+	};
+
+	return kvm_pgtable_walk(&pkvm_pgtable, 0, BIT(pkvm_pgtable.ia_bits),
+				&walker);
+}
+
 void __noreturn __pkvm_init_finalise(void)
 {
 	struct kvm_host_data *host_data = this_cpu_ptr(&kvm_host_data);
@@ -287,10 +302,19 @@ void __noreturn __pkvm_init_finalise(void)
 	};
 	pkvm_pgtable.mm_ops = &pkvm_pgtable_mm_ops;
 
-	ret = finalize_host_mappings();
+	ret = fix_host_ownership();
+	if (ret)
+		goto out;
+
+	ret = fix_hyp_pgtable_refcnt();
+	if (ret)
+		goto out;
+
+	ret = hyp_create_pcpu_fixmap();
 	if (ret)
 		goto out;
 
+	pkvm_hyp_vm_table_init(vm_table_base);
 out:
 	/*
 	 * We tail-called to here from handle___pkvm_init() and will not return,
diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
index cdf8e76b0be1..b11cf2c618a6 100644
--- a/arch/arm64/kvm/hyp/pgtable.c
+++ b/arch/arm64/kvm/hyp/pgtable.c
@@ -49,35 +49,38 @@
 #define KVM_INVALID_PTE_OWNER_MASK	GENMASK(9, 2)
 #define KVM_MAX_OWNER_ID		1
 
+/*
+ * Used to indicate a pte for which a 'break-before-make' sequence is in
+ * progress.
+ */
+#define KVM_INVALID_PTE_LOCKED		BIT(10)
+
 struct kvm_pgtable_walk_data {
-	struct kvm_pgtable		*pgt;
 	struct kvm_pgtable_walker	*walker;
 
 	u64				addr;
 	u64				end;
 };
 
-#define KVM_PHYS_INVALID (-1ULL)
-
 static bool kvm_phys_is_valid(u64 phys)
 {
 	return phys < BIT(id_aa64mmfr0_parange_to_phys_shift(ID_AA64MMFR0_EL1_PARANGE_MAX));
 }
 
-static bool kvm_block_mapping_supported(u64 addr, u64 end, u64 phys, u32 level)
+static bool kvm_block_mapping_supported(const struct kvm_pgtable_visit_ctx *ctx, u64 phys)
 {
-	u64 granule = kvm_granule_size(level);
+	u64 granule = kvm_granule_size(ctx->level);
 
-	if (!kvm_level_supports_block_mapping(level))
+	if (!kvm_level_supports_block_mapping(ctx->level))
 		return false;
 
-	if (granule > (end - addr))
+	if (granule > (ctx->end - ctx->addr))
 		return false;
 
 	if (kvm_phys_is_valid(phys) && !IS_ALIGNED(phys, granule))
 		return false;
 
-	return IS_ALIGNED(addr, granule);
+	return IS_ALIGNED(ctx->addr, granule);
 }
 
 static u32 kvm_pgtable_idx(struct kvm_pgtable_walk_data *data, u32 level)
@@ -88,7 +91,7 @@ static u32 kvm_pgtable_idx(struct kvm_pgtable_walk_data *data, u32 level)
 	return (data->addr >> shift) & mask;
 }
 
-static u32 __kvm_pgd_page_idx(struct kvm_pgtable *pgt, u64 addr)
+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;
@@ -96,11 +99,6 @@ static u32 __kvm_pgd_page_idx(struct kvm_pgtable *pgt, u64 addr)
 	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 = {
@@ -108,7 +106,7 @@ static u32 kvm_pgd_pages(u32 ia_bits, u32 start_level)
 		.start_level	= start_level,
 	};
 
-	return __kvm_pgd_page_idx(&pgt, -1ULL) + 1;
+	return kvm_pgd_page_idx(&pgt, -1ULL) + 1;
 }
 
 static bool kvm_pte_table(kvm_pte_t pte, u32 level)
@@ -122,16 +120,6 @@ static bool kvm_pte_table(kvm_pte_t pte, u32 level)
 	return FIELD_GET(KVM_PTE_TYPE, pte) == KVM_PTE_TYPE_TABLE;
 }
 
-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, struct kvm_pgtable_mm_ops *mm_ops)
 {
 	return mm_ops->phys_to_virt(kvm_pte_to_phys(pte));
@@ -142,16 +130,13 @@ static void kvm_clear_pte(kvm_pte_t *ptep)
 	WRITE_ONCE(*ptep, 0);
 }
 
-static void kvm_set_table_pte(kvm_pte_t *ptep, kvm_pte_t *childp,
-			      struct kvm_pgtable_mm_ops *mm_ops)
+static kvm_pte_t kvm_init_table_pte(kvm_pte_t *childp, struct kvm_pgtable_mm_ops *mm_ops)
 {
-	kvm_pte_t old = *ptep, pte = kvm_phys_to_pte(mm_ops->virt_to_phys(childp));
+	kvm_pte_t pte = kvm_phys_to_pte(mm_ops->virt_to_phys(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);
+	return pte;
 }
 
 static kvm_pte_t kvm_init_valid_leaf_pte(u64 pa, kvm_pte_t attr, u32 level)
@@ -172,36 +157,47 @@ static kvm_pte_t kvm_init_invalid_leaf_owner(u8 owner_id)
 	return FIELD_PREP(KVM_INVALID_PTE_OWNER_MASK, owner_id);
 }
 
-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)
+static int kvm_pgtable_visitor_cb(struct kvm_pgtable_walk_data *data,
+				  const struct kvm_pgtable_visit_ctx *ctx,
+				  enum kvm_pgtable_walk_flags visit)
 {
 	struct kvm_pgtable_walker *walker = data->walker;
-	return walker->cb(addr, data->end, level, ptep, flag, walker->arg);
+
+	/* Ensure the appropriate lock is held (e.g. RCU lock for stage-2 MMU) */
+	WARN_ON_ONCE(kvm_pgtable_walk_shared(ctx) && !kvm_pgtable_walk_lock_held());
+	return walker->cb(ctx, visit);
 }
 
 static int __kvm_pgtable_walk(struct kvm_pgtable_walk_data *data,
-			      kvm_pte_t *pgtable, u32 level);
+			      struct kvm_pgtable_mm_ops *mm_ops, kvm_pteref_t pgtable, u32 level);
 
 static inline int __kvm_pgtable_visit(struct kvm_pgtable_walk_data *data,
-				      kvm_pte_t *ptep, u32 level)
+				      struct kvm_pgtable_mm_ops *mm_ops,
+				      kvm_pteref_t pteref, 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;
+	kvm_pte_t *ptep = kvm_dereference_pteref(data->walker, pteref);
+	struct kvm_pgtable_visit_ctx ctx = {
+		.ptep	= ptep,
+		.old	= READ_ONCE(*ptep),
+		.arg	= data->walker->arg,
+		.mm_ops	= mm_ops,
+		.addr	= data->addr,
+		.end	= data->end,
+		.level	= level,
+		.flags	= flags,
+	};
+	int ret = 0;
+	kvm_pteref_t childp;
+	bool table = kvm_pte_table(ctx.old, level);
 
-	if (table && (flags & KVM_PGTABLE_WALK_TABLE_PRE)) {
-		ret = kvm_pgtable_visitor_cb(data, addr, level, ptep,
-					     KVM_PGTABLE_WALK_TABLE_PRE);
-	}
+	if (table && (ctx.flags & KVM_PGTABLE_WALK_TABLE_PRE))
+		ret = kvm_pgtable_visitor_cb(data, &ctx, 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 (!table && (ctx.flags & KVM_PGTABLE_WALK_LEAF)) {
+		ret = kvm_pgtable_visitor_cb(data, &ctx, KVM_PGTABLE_WALK_LEAF);
+		ctx.old = READ_ONCE(*ptep);
+		table = kvm_pte_table(ctx.old, level);
 	}
 
 	if (ret)
@@ -213,22 +209,20 @@ static inline int __kvm_pgtable_visit(struct kvm_pgtable_walk_data *data,
 		goto out;
 	}
 
-	childp = kvm_pte_follow(pte, data->pgt->mm_ops);
-	ret = __kvm_pgtable_walk(data, childp, level + 1);
+	childp = (kvm_pteref_t)kvm_pte_follow(ctx.old, mm_ops);
+	ret = __kvm_pgtable_walk(data, mm_ops, 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);
-	}
+	if (ctx.flags & KVM_PGTABLE_WALK_TABLE_POST)
+		ret = kvm_pgtable_visitor_cb(data, &ctx, 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)
+			      struct kvm_pgtable_mm_ops *mm_ops, kvm_pteref_t pgtable, u32 level)
 {
 	u32 idx;
 	int ret = 0;
@@ -237,12 +231,12 @@ static int __kvm_pgtable_walk(struct kvm_pgtable_walk_data *data,
 		return -EINVAL;
 
 	for (idx = kvm_pgtable_idx(data, level); idx < PTRS_PER_PTE; ++idx) {
-		kvm_pte_t *ptep = &pgtable[idx];
+		kvm_pteref_t pteref = &pgtable[idx];
 
 		if (data->addr >= data->end)
 			break;
 
-		ret = __kvm_pgtable_visit(data, ptep, level);
+		ret = __kvm_pgtable_visit(data, mm_ops, pteref, level);
 		if (ret)
 			break;
 	}
@@ -250,11 +244,10 @@ static int __kvm_pgtable_walk(struct kvm_pgtable_walk_data *data,
 	return ret;
 }
 
-static int _kvm_pgtable_walk(struct kvm_pgtable_walk_data *data)
+static int _kvm_pgtable_walk(struct kvm_pgtable *pgt, 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)
@@ -263,10 +256,10 @@ static int _kvm_pgtable_walk(struct kvm_pgtable_walk_data *data)
 	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];
+	for (idx = kvm_pgd_page_idx(pgt, data->addr); data->addr < data->end; ++idx) {
+		kvm_pteref_t pteref = &pgt->pgd[idx * PTRS_PER_PTE];
 
-		ret = __kvm_pgtable_walk(data, ptep, pgt->start_level);
+		ret = __kvm_pgtable_walk(data, pgt->mm_ops, pteref, pgt->start_level);
 		if (ret)
 			break;
 	}
@@ -278,13 +271,20 @@ 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,
 	};
+	int r;
 
-	return _kvm_pgtable_walk(&walk_data);
+	r = kvm_pgtable_walk_begin(walker);
+	if (r)
+		return r;
+
+	r = _kvm_pgtable_walk(pgt, &walk_data);
+	kvm_pgtable_walk_end(walker);
+
+	return r;
 }
 
 struct leaf_walk_data {
@@ -292,13 +292,13 @@ struct leaf_walk_data {
 	u32		level;
 };
 
-static int leaf_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
-		       enum kvm_pgtable_walk_flags flag, void * const arg)
+static int leaf_walker(const struct kvm_pgtable_visit_ctx *ctx,
+		       enum kvm_pgtable_walk_flags visit)
 {
-	struct leaf_walk_data *data = arg;
+	struct leaf_walk_data *data = ctx->arg;
 
-	data->pte   = *ptep;
-	data->level = level;
+	data->pte   = ctx->old;
+	data->level = ctx->level;
 
 	return 0;
 }
@@ -329,7 +329,6 @@ int kvm_pgtable_get_leaf(struct kvm_pgtable *pgt, u64 addr,
 struct hyp_map_data {
 	u64				phys;
 	kvm_pte_t			attr;
-	struct kvm_pgtable_mm_ops	*mm_ops;
 };
 
 static int hyp_set_prot_attr(enum kvm_pgtable_prot prot, kvm_pte_t *ptep)
@@ -383,47 +382,49 @@ enum kvm_pgtable_prot kvm_pgtable_hyp_pte_prot(kvm_pte_t pte)
 	return prot;
 }
 
-static bool hyp_map_walker_try_leaf(u64 addr, u64 end, u32 level,
-				    kvm_pte_t *ptep, struct hyp_map_data *data)
+static bool hyp_map_walker_try_leaf(const struct kvm_pgtable_visit_ctx *ctx,
+				    struct hyp_map_data *data)
 {
-	kvm_pte_t new, old = *ptep;
-	u64 granule = kvm_granule_size(level), phys = data->phys;
+	kvm_pte_t new;
+	u64 granule = kvm_granule_size(ctx->level), phys = data->phys;
 
-	if (!kvm_block_mapping_supported(addr, end, phys, level))
+	if (!kvm_block_mapping_supported(ctx, phys))
 		return false;
 
 	data->phys += granule;
-	new = kvm_init_valid_leaf_pte(phys, data->attr, level);
-	if (old == new)
+	new = kvm_init_valid_leaf_pte(phys, data->attr, ctx->level);
+	if (ctx->old == new)
 		return true;
-	if (!kvm_pte_valid(old))
-		data->mm_ops->get_page(ptep);
-	else if (WARN_ON((old ^ new) & ~KVM_PTE_LEAF_ATTR_HI_SW))
+	if (!kvm_pte_valid(ctx->old))
+		ctx->mm_ops->get_page(ctx->ptep);
+	else if (WARN_ON((ctx->old ^ new) & ~KVM_PTE_LEAF_ATTR_HI_SW))
 		return false;
 
-	smp_store_release(ptep, new);
+	smp_store_release(ctx->ptep, new);
 	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)
+static int hyp_map_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			  enum kvm_pgtable_walk_flags visit)
 {
-	kvm_pte_t *childp;
-	struct hyp_map_data *data = arg;
-	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
+	kvm_pte_t *childp, new;
+	struct hyp_map_data *data = ctx->arg;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
 
-	if (hyp_map_walker_try_leaf(addr, end, level, ptep, arg))
+	if (hyp_map_walker_try_leaf(ctx, data))
 		return 0;
 
-	if (WARN_ON(level == KVM_PGTABLE_MAX_LEVELS - 1))
+	if (WARN_ON(ctx->level == KVM_PGTABLE_MAX_LEVELS - 1))
 		return -EINVAL;
 
 	childp = (kvm_pte_t *)mm_ops->zalloc_page(NULL);
 	if (!childp)
 		return -ENOMEM;
 
-	kvm_set_table_pte(ptep, childp, mm_ops);
-	mm_ops->get_page(ptep);
+	new = kvm_init_table_pte(childp, mm_ops);
+	mm_ops->get_page(ctx->ptep);
+	smp_store_release(ctx->ptep, new);
+
 	return 0;
 }
 
@@ -433,7 +434,6 @@ int kvm_pgtable_hyp_map(struct kvm_pgtable *pgt, u64 addr, u64 size, u64 phys,
 	int ret;
 	struct hyp_map_data map_data = {
 		.phys	= ALIGN_DOWN(phys, PAGE_SIZE),
-		.mm_ops	= pgt->mm_ops,
 	};
 	struct kvm_pgtable_walker walker = {
 		.cb	= hyp_map_walker,
@@ -451,44 +451,39 @@ int kvm_pgtable_hyp_map(struct kvm_pgtable *pgt, u64 addr, u64 size, u64 phys,
 	return ret;
 }
 
-struct hyp_unmap_data {
-	u64				unmapped;
-	struct kvm_pgtable_mm_ops	*mm_ops;
-};
-
-static int hyp_unmap_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
-			    enum kvm_pgtable_walk_flags flag, void * const arg)
+static int hyp_unmap_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			    enum kvm_pgtable_walk_flags visit)
 {
-	kvm_pte_t pte = *ptep, *childp = NULL;
-	u64 granule = kvm_granule_size(level);
-	struct hyp_unmap_data *data = arg;
-	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
+	kvm_pte_t *childp = NULL;
+	u64 granule = kvm_granule_size(ctx->level);
+	u64 *unmapped = ctx->arg;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
 
-	if (!kvm_pte_valid(pte))
+	if (!kvm_pte_valid(ctx->old))
 		return -EINVAL;
 
-	if (kvm_pte_table(pte, level)) {
-		childp = kvm_pte_follow(pte, mm_ops);
+	if (kvm_pte_table(ctx->old, ctx->level)) {
+		childp = kvm_pte_follow(ctx->old, mm_ops);
 
 		if (mm_ops->page_count(childp) != 1)
 			return 0;
 
-		kvm_clear_pte(ptep);
+		kvm_clear_pte(ctx->ptep);
 		dsb(ishst);
-		__tlbi_level(vae2is, __TLBI_VADDR(addr, 0), level);
+		__tlbi_level(vae2is, __TLBI_VADDR(ctx->addr, 0), ctx->level);
 	} else {
-		if (end - addr < granule)
+		if (ctx->end - ctx->addr < granule)
 			return -EINVAL;
 
-		kvm_clear_pte(ptep);
+		kvm_clear_pte(ctx->ptep);
 		dsb(ishst);
-		__tlbi_level(vale2is, __TLBI_VADDR(addr, 0), level);
-		data->unmapped += granule;
+		__tlbi_level(vale2is, __TLBI_VADDR(ctx->addr, 0), ctx->level);
+		*unmapped += granule;
 	}
 
 	dsb(ish);
 	isb();
-	mm_ops->put_page(ptep);
+	mm_ops->put_page(ctx->ptep);
 
 	if (childp)
 		mm_ops->put_page(childp);
@@ -498,12 +493,10 @@ static int hyp_unmap_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
 
 u64 kvm_pgtable_hyp_unmap(struct kvm_pgtable *pgt, u64 addr, u64 size)
 {
-	struct hyp_unmap_data unmap_data = {
-		.mm_ops	= pgt->mm_ops,
-	};
+	u64 unmapped = 0;
 	struct kvm_pgtable_walker walker = {
 		.cb	= hyp_unmap_walker,
-		.arg	= &unmap_data,
+		.arg	= &unmapped,
 		.flags	= KVM_PGTABLE_WALK_LEAF | KVM_PGTABLE_WALK_TABLE_POST,
 	};
 
@@ -511,7 +504,7 @@ u64 kvm_pgtable_hyp_unmap(struct kvm_pgtable *pgt, u64 addr, u64 size)
 		return 0;
 
 	kvm_pgtable_walk(pgt, addr, size, &walker);
-	return unmap_data.unmapped;
+	return unmapped;
 }
 
 int kvm_pgtable_hyp_init(struct kvm_pgtable *pgt, u32 va_bits,
@@ -519,7 +512,7 @@ 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 *)mm_ops->zalloc_page(NULL);
+	pgt->pgd = (kvm_pteref_t)mm_ops->zalloc_page(NULL);
 	if (!pgt->pgd)
 		return -ENOMEM;
 
@@ -532,19 +525,18 @@ int kvm_pgtable_hyp_init(struct kvm_pgtable *pgt, u32 va_bits,
 	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)
+static int hyp_free_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			   enum kvm_pgtable_walk_flags visit)
 {
-	struct kvm_pgtable_mm_ops *mm_ops = arg;
-	kvm_pte_t pte = *ptep;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
 
-	if (!kvm_pte_valid(pte))
+	if (!kvm_pte_valid(ctx->old))
 		return 0;
 
-	mm_ops->put_page(ptep);
+	mm_ops->put_page(ctx->ptep);
 
-	if (kvm_pte_table(pte, level))
-		mm_ops->put_page(kvm_pte_follow(pte, mm_ops));
+	if (kvm_pte_table(ctx->old, ctx->level))
+		mm_ops->put_page(kvm_pte_follow(ctx->old, mm_ops));
 
 	return 0;
 }
@@ -554,11 +546,10 @@ void kvm_pgtable_hyp_destroy(struct kvm_pgtable *pgt)
 	struct kvm_pgtable_walker walker = {
 		.cb	= hyp_free_walker,
 		.flags	= KVM_PGTABLE_WALK_LEAF | KVM_PGTABLE_WALK_TABLE_POST,
-		.arg	= pgt->mm_ops,
 	};
 
 	WARN_ON(kvm_pgtable_walk(pgt, 0, BIT(pgt->ia_bits), &walker));
-	pgt->mm_ops->put_page(pgt->pgd);
+	pgt->mm_ops->put_page(kvm_dereference_pteref(&walker, pgt->pgd));
 	pgt->pgd = NULL;
 }
 
@@ -573,8 +564,6 @@ struct stage2_map_data {
 	struct kvm_s2_mmu		*mmu;
 	void				*memcache;
 
-	struct kvm_pgtable_mm_ops	*mm_ops;
-
 	/* Force mappings to page granularity */
 	bool				force_pte;
 };
@@ -682,19 +671,92 @@ static bool stage2_pte_is_counted(kvm_pte_t pte)
 	return !!pte;
 }
 
-static void stage2_put_pte(kvm_pte_t *ptep, struct kvm_s2_mmu *mmu, u64 addr,
-			   u32 level, struct kvm_pgtable_mm_ops *mm_ops)
+static bool stage2_pte_is_locked(kvm_pte_t pte)
+{
+	return !kvm_pte_valid(pte) && (pte & KVM_INVALID_PTE_LOCKED);
+}
+
+static bool stage2_try_set_pte(const struct kvm_pgtable_visit_ctx *ctx, kvm_pte_t new)
+{
+	if (!kvm_pgtable_walk_shared(ctx)) {
+		WRITE_ONCE(*ctx->ptep, new);
+		return true;
+	}
+
+	return cmpxchg(ctx->ptep, ctx->old, new) == ctx->old;
+}
+
+/**
+ * stage2_try_break_pte() - Invalidates a pte according to the
+ *			    'break-before-make' requirements of the
+ *			    architecture.
+ *
+ * @ctx: context of the visited pte.
+ * @mmu: stage-2 mmu
+ *
+ * Returns: true if the pte was successfully broken.
+ *
+ * If the removed pte was valid, performs the necessary serialization and TLB
+ * invalidation for the old value. For counted ptes, drops the reference count
+ * on the containing table page.
+ */
+static bool stage2_try_break_pte(const struct kvm_pgtable_visit_ctx *ctx,
+				 struct kvm_s2_mmu *mmu)
+{
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
+
+	if (stage2_pte_is_locked(ctx->old)) {
+		/*
+		 * Should never occur if this walker has exclusive access to the
+		 * page tables.
+		 */
+		WARN_ON(!kvm_pgtable_walk_shared(ctx));
+		return false;
+	}
+
+	if (!stage2_try_set_pte(ctx, KVM_INVALID_PTE_LOCKED))
+		return false;
+
+	/*
+	 * Perform the appropriate TLB invalidation based on the evicted pte
+	 * value (if any).
+	 */
+	if (kvm_pte_table(ctx->old, ctx->level))
+		kvm_call_hyp(__kvm_tlb_flush_vmid, mmu);
+	else if (kvm_pte_valid(ctx->old))
+		kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, mmu, ctx->addr, ctx->level);
+
+	if (stage2_pte_is_counted(ctx->old))
+		mm_ops->put_page(ctx->ptep);
+
+	return true;
+}
+
+static void stage2_make_pte(const struct kvm_pgtable_visit_ctx *ctx, kvm_pte_t new)
+{
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
+
+	WARN_ON(!stage2_pte_is_locked(*ctx->ptep));
+
+	if (stage2_pte_is_counted(new))
+		mm_ops->get_page(ctx->ptep);
+
+	smp_store_release(ctx->ptep, new);
+}
+
+static void stage2_put_pte(const struct kvm_pgtable_visit_ctx *ctx, struct kvm_s2_mmu *mmu,
+			   struct kvm_pgtable_mm_ops *mm_ops)
 {
 	/*
 	 * Clear the existing PTE, and perform break-before-make with
 	 * TLB maintenance if it was valid.
 	 */
-	if (kvm_pte_valid(*ptep)) {
-		kvm_clear_pte(ptep);
-		kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, mmu, addr, level);
+	if (kvm_pte_valid(ctx->old)) {
+		kvm_clear_pte(ctx->ptep);
+		kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, mmu, ctx->addr, ctx->level);
 	}
 
-	mm_ops->put_page(ptep);
+	mm_ops->put_page(ctx->ptep);
 }
 
 static bool stage2_pte_cacheable(struct kvm_pgtable *pgt, kvm_pte_t pte)
@@ -708,44 +770,42 @@ static bool stage2_pte_executable(kvm_pte_t pte)
 	return !(pte & KVM_PTE_LEAF_ATTR_HI_S2_XN);
 }
 
-static bool stage2_leaf_mapping_allowed(u64 addr, u64 end, u32 level,
+static bool stage2_leaf_mapping_allowed(const struct kvm_pgtable_visit_ctx *ctx,
 					struct stage2_map_data *data)
 {
-	if (data->force_pte && (level < (KVM_PGTABLE_MAX_LEVELS - 1)))
+	if (data->force_pte && (ctx->level < (KVM_PGTABLE_MAX_LEVELS - 1)))
 		return false;
 
-	return kvm_block_mapping_supported(addr, end, data->phys, level);
+	return kvm_block_mapping_supported(ctx, data->phys);
 }
 
-static int stage2_map_walker_try_leaf(u64 addr, u64 end, u32 level,
-				      kvm_pte_t *ptep,
+static int stage2_map_walker_try_leaf(const struct kvm_pgtable_visit_ctx *ctx,
 				      struct stage2_map_data *data)
 {
-	kvm_pte_t new, old = *ptep;
-	u64 granule = kvm_granule_size(level), phys = data->phys;
+	kvm_pte_t new;
+	u64 granule = kvm_granule_size(ctx->level), phys = data->phys;
 	struct kvm_pgtable *pgt = data->mmu->pgt;
-	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
 
-	if (!stage2_leaf_mapping_allowed(addr, end, level, data))
+	if (!stage2_leaf_mapping_allowed(ctx, data))
 		return -E2BIG;
 
 	if (kvm_phys_is_valid(phys))
-		new = kvm_init_valid_leaf_pte(phys, data->attr, level);
+		new = kvm_init_valid_leaf_pte(phys, data->attr, ctx->level);
 	else
 		new = kvm_init_invalid_leaf_owner(data->owner_id);
 
-	if (stage2_pte_is_counted(old)) {
-		/*
-		 * Skip updating the PTE if we are trying to recreate the exact
-		 * same mapping or only change the access permissions. Instead,
-		 * the vCPU will exit one more time from guest if still needed
-		 * and then go through the path of relaxing permissions.
-		 */
-		if (!stage2_pte_needs_update(old, new))
-			return -EAGAIN;
+	/*
+	 * Skip updating the PTE if we are trying to recreate the exact
+	 * same mapping or only change the access permissions. Instead,
+	 * the vCPU will exit one more time from guest if still needed
+	 * and then go through the path of relaxing permissions.
+	 */
+	if (!stage2_pte_needs_update(ctx->old, new))
+		return -EAGAIN;
 
-		stage2_put_pte(ptep, data->mmu, addr, level, mm_ops);
-	}
+	if (!stage2_try_break_pte(ctx, data->mmu))
+		return -EAGAIN;
 
 	/* Perform CMOs before installation of the guest stage-2 PTE */
 	if (mm_ops->dcache_clean_inval_poc && stage2_pte_cacheable(pgt, new))
@@ -755,56 +815,43 @@ static int stage2_map_walker_try_leaf(u64 addr, u64 end, u32 level,
 	if (mm_ops->icache_inval_pou && stage2_pte_executable(new))
 		mm_ops->icache_inval_pou(kvm_pte_follow(new, mm_ops), granule);
 
-	smp_store_release(ptep, new);
-	if (stage2_pte_is_counted(new))
-		mm_ops->get_page(ptep);
+	stage2_make_pte(ctx, new);
+
 	if (kvm_phys_is_valid(phys))
 		data->phys += granule;
 	return 0;
 }
 
-static int stage2_map_walk_table_pre(u64 addr, u64 end, u32 level,
-				     kvm_pte_t *ptep,
+static int stage2_map_walk_table_pre(const struct kvm_pgtable_visit_ctx *ctx,
 				     struct stage2_map_data *data)
 {
-	if (data->anchor)
-		return 0;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
+	kvm_pte_t *childp = kvm_pte_follow(ctx->old, mm_ops);
+	int ret;
 
-	if (!stage2_leaf_mapping_allowed(addr, end, level, data))
+	if (!stage2_leaf_mapping_allowed(ctx, data))
 		return 0;
 
-	data->childp = kvm_pte_follow(*ptep, data->mm_ops);
-	kvm_clear_pte(ptep);
+	ret = stage2_map_walker_try_leaf(ctx, data);
+	if (ret)
+		return ret;
 
-	/*
-	 * Invalidate the whole stage-2, as we may have numerous leaf
-	 * entries below us which would otherwise need invalidating
-	 * individually.
-	 */
-	kvm_call_hyp(__kvm_tlb_flush_vmid, data->mmu);
-	data->anchor = ptep;
+	mm_ops->free_removed_table(childp, ctx->level);
 	return 0;
 }
 
-static int stage2_map_walk_leaf(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
+static int stage2_map_walk_leaf(const struct kvm_pgtable_visit_ctx *ctx,
 				struct stage2_map_data *data)
 {
-	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
-	kvm_pte_t *childp, pte = *ptep;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
+	kvm_pte_t *childp, new;
 	int ret;
 
-	if (data->anchor) {
-		if (stage2_pte_is_counted(pte))
-			mm_ops->put_page(ptep);
-
-		return 0;
-	}
-
-	ret = stage2_map_walker_try_leaf(addr, end, level, ptep, data);
+	ret = stage2_map_walker_try_leaf(ctx, data);
 	if (ret != -E2BIG)
 		return ret;
 
-	if (WARN_ON(level == KVM_PGTABLE_MAX_LEVELS - 1))
+	if (WARN_ON(ctx->level == KVM_PGTABLE_MAX_LEVELS - 1))
 		return -EINVAL;
 
 	if (!data->memcache)
@@ -814,99 +861,62 @@ static int stage2_map_walk_leaf(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
 	if (!childp)
 		return -ENOMEM;
 
+	if (!stage2_try_break_pte(ctx, data->mmu)) {
+		mm_ops->put_page(childp);
+		return -EAGAIN;
+	}
+
 	/*
 	 * 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 (stage2_pte_is_counted(pte))
-		stage2_put_pte(ptep, data->mmu, addr, level, mm_ops);
-
-	kvm_set_table_pte(ptep, childp, mm_ops);
-	mm_ops->get_page(ptep);
+	new = kvm_init_table_pte(childp, mm_ops);
+	stage2_make_pte(ctx, new);
 
 	return 0;
 }
 
-static int stage2_map_walk_table_post(u64 addr, u64 end, u32 level,
-				      kvm_pte_t *ptep,
-				      struct stage2_map_data *data)
-{
-	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
-	kvm_pte_t *childp;
-	int ret = 0;
-
-	if (!data->anchor)
-		return 0;
-
-	if (data->anchor == ptep) {
-		childp = data->childp;
-		data->anchor = NULL;
-		data->childp = NULL;
-		ret = stage2_map_walk_leaf(addr, end, level, ptep, data);
-	} else {
-		childp = kvm_pte_follow(*ptep, mm_ops);
-	}
-
-	mm_ops->put_page(childp);
-	mm_ops->put_page(ptep);
-
-	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 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 it replaces the entry and calls
+ * kvm_pgtable_mm_ops::free_removed_table() to tear down the 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.
+ * Otherwise, the LEAF callback performs the mapping at the existing leaves
+ * instead.
  */
-static int stage2_map_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
-			     enum kvm_pgtable_walk_flags flag, void * const arg)
+static int stage2_map_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			     enum kvm_pgtable_walk_flags visit)
 {
-	struct stage2_map_data *data = arg;
+	struct stage2_map_data *data = ctx->arg;
 
-	switch (flag) {
+	switch (visit) {
 	case KVM_PGTABLE_WALK_TABLE_PRE:
-		return stage2_map_walk_table_pre(addr, end, level, ptep, data);
+		return stage2_map_walk_table_pre(ctx, 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 stage2_map_walk_leaf(ctx, data);
+	default:
+		return -EINVAL;
 	}
-
-	return -EINVAL;
 }
 
 int kvm_pgtable_stage2_map(struct kvm_pgtable *pgt, u64 addr, u64 size,
 			   u64 phys, enum kvm_pgtable_prot prot,
-			   void *mc)
+			   void *mc, enum kvm_pgtable_walk_flags flags)
 {
 	int ret;
 	struct stage2_map_data map_data = {
 		.phys		= ALIGN_DOWN(phys, PAGE_SIZE),
 		.mmu		= pgt->mmu,
 		.memcache	= mc,
-		.mm_ops		= pgt->mm_ops,
 		.force_pte	= pgt->force_pte_cb && pgt->force_pte_cb(addr, addr + size, prot),
 	};
 	struct kvm_pgtable_walker walker = {
 		.cb		= stage2_map_walker,
-		.flags		= KVM_PGTABLE_WALK_TABLE_PRE |
-				  KVM_PGTABLE_WALK_LEAF |
-				  KVM_PGTABLE_WALK_TABLE_POST,
+		.flags		= flags |
+				  KVM_PGTABLE_WALK_TABLE_PRE |
+				  KVM_PGTABLE_WALK_LEAF,
 		.arg		= &map_data,
 	};
 
@@ -930,15 +940,13 @@ int kvm_pgtable_stage2_set_owner(struct kvm_pgtable *pgt, u64 addr, u64 size,
 		.phys		= KVM_PHYS_INVALID,
 		.mmu		= pgt->mmu,
 		.memcache	= mc,
-		.mm_ops		= pgt->mm_ops,
 		.owner_id	= owner_id,
 		.force_pte	= true,
 	};
 	struct kvm_pgtable_walker walker = {
 		.cb		= stage2_map_walker,
 		.flags		= KVM_PGTABLE_WALK_TABLE_PRE |
-				  KVM_PGTABLE_WALK_LEAF |
-				  KVM_PGTABLE_WALK_TABLE_POST,
+				  KVM_PGTABLE_WALK_LEAF,
 		.arg		= &map_data,
 	};
 
@@ -949,30 +957,29 @@ int kvm_pgtable_stage2_set_owner(struct kvm_pgtable *pgt, u64 addr, u64 size,
 	return ret;
 }
 
-static int stage2_unmap_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
-			       enum kvm_pgtable_walk_flags flag,
-			       void * const arg)
+static int stage2_unmap_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			       enum kvm_pgtable_walk_flags visit)
 {
-	struct kvm_pgtable *pgt = arg;
+	struct kvm_pgtable *pgt = ctx->arg;
 	struct kvm_s2_mmu *mmu = pgt->mmu;
-	struct kvm_pgtable_mm_ops *mm_ops = pgt->mm_ops;
-	kvm_pte_t pte = *ptep, *childp = NULL;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
+	kvm_pte_t *childp = NULL;
 	bool need_flush = false;
 
-	if (!kvm_pte_valid(pte)) {
-		if (stage2_pte_is_counted(pte)) {
-			kvm_clear_pte(ptep);
-			mm_ops->put_page(ptep);
+	if (!kvm_pte_valid(ctx->old)) {
+		if (stage2_pte_is_counted(ctx->old)) {
+			kvm_clear_pte(ctx->ptep);
+			mm_ops->put_page(ctx->ptep);
 		}
 		return 0;
 	}
 
-	if (kvm_pte_table(pte, level)) {
-		childp = kvm_pte_follow(pte, mm_ops);
+	if (kvm_pte_table(ctx->old, ctx->level)) {
+		childp = kvm_pte_follow(ctx->old, mm_ops);
 
 		if (mm_ops->page_count(childp) != 1)
 			return 0;
-	} else if (stage2_pte_cacheable(pgt, pte)) {
+	} else if (stage2_pte_cacheable(pgt, ctx->old)) {
 		need_flush = !stage2_has_fwb(pgt);
 	}
 
@@ -981,11 +988,11 @@ static int stage2_unmap_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
 	 * block entry and rely on the remaining portions being faulted
 	 * back lazily.
 	 */
-	stage2_put_pte(ptep, mmu, addr, level, mm_ops);
+	stage2_put_pte(ctx, mmu, mm_ops);
 
 	if (need_flush && mm_ops->dcache_clean_inval_poc)
-		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
-					       kvm_granule_size(level));
+		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(ctx->old, mm_ops),
+					       kvm_granule_size(ctx->level));
 
 	if (childp)
 		mm_ops->put_page(childp);
@@ -1009,21 +1016,19 @@ struct stage2_attr_data {
 	kvm_pte_t			attr_clr;
 	kvm_pte_t			pte;
 	u32				level;
-	struct kvm_pgtable_mm_ops	*mm_ops;
 };
 
-static int stage2_attr_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
-			      enum kvm_pgtable_walk_flags flag,
-			      void * const arg)
+static int stage2_attr_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			      enum kvm_pgtable_walk_flags visit)
 {
-	kvm_pte_t pte = *ptep;
-	struct stage2_attr_data *data = arg;
-	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
+	kvm_pte_t pte = ctx->old;
+	struct stage2_attr_data *data = ctx->arg;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
 
-	if (!kvm_pte_valid(pte))
+	if (!kvm_pte_valid(ctx->old))
 		return 0;
 
-	data->level = level;
+	data->level = ctx->level;
 	data->pte = pte;
 	pte &= ~data->attr_clr;
 	pte |= data->attr_set;
@@ -1039,10 +1044,12 @@ static int stage2_attr_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
 		 * stage-2 PTE if we are going to add executable permission.
 		 */
 		if (mm_ops->icache_inval_pou &&
-		    stage2_pte_executable(pte) && !stage2_pte_executable(*ptep))
+		    stage2_pte_executable(pte) && !stage2_pte_executable(ctx->old))
 			mm_ops->icache_inval_pou(kvm_pte_follow(pte, mm_ops),
-						  kvm_granule_size(level));
-		WRITE_ONCE(*ptep, pte);
+						  kvm_granule_size(ctx->level));
+
+		if (!stage2_try_set_pte(ctx, pte))
+			return -EAGAIN;
 	}
 
 	return 0;
@@ -1051,19 +1058,18 @@ static int stage2_attr_walker(u64 addr, u64 end, u32 level, kvm_pte_t *ptep,
 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)
+				    u32 *level, enum kvm_pgtable_walk_flags flags)
 {
 	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,
-		.mm_ops		= pgt->mm_ops,
 	};
 	struct kvm_pgtable_walker walker = {
 		.cb		= stage2_attr_walker,
 		.arg		= &data,
-		.flags		= KVM_PGTABLE_WALK_LEAF,
+		.flags		= flags | KVM_PGTABLE_WALK_LEAF,
 	};
 
 	ret = kvm_pgtable_walk(pgt, addr, size, &walker);
@@ -1082,14 +1088,14 @@ 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);
+					NULL, NULL, 0);
 }
 
 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);
+				 &pte, NULL, 0);
 	dsb(ishst);
 	return pte;
 }
@@ -1098,7 +1104,7 @@ 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);
+				 &pte, NULL, 0);
 	/*
 	 * "But where's the TLBI?!", you scream.
 	 * "Over in the core code", I sigh.
@@ -1111,7 +1117,7 @@ kvm_pte_t kvm_pgtable_stage2_mkold(struct kvm_pgtable *pgt, u64 addr)
 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);
+	stage2_update_leaf_attrs(pgt, addr, 1, 0, 0, &pte, NULL, 0);
 	return pte & KVM_PTE_LEAF_ATTR_LO_S2_AF;
 }
 
@@ -1134,26 +1140,25 @@ int kvm_pgtable_stage2_relax_perms(struct kvm_pgtable *pgt, u64 addr,
 	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);
+	ret = stage2_update_leaf_attrs(pgt, addr, 1, set, clr, NULL, &level,
+				       KVM_PGTABLE_WALK_SHARED);
 	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)
+static int stage2_flush_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			       enum kvm_pgtable_walk_flags visit)
 {
-	struct kvm_pgtable *pgt = arg;
+	struct kvm_pgtable *pgt = ctx->arg;
 	struct kvm_pgtable_mm_ops *mm_ops = pgt->mm_ops;
-	kvm_pte_t pte = *ptep;
 
-	if (!kvm_pte_valid(pte) || !stage2_pte_cacheable(pgt, pte))
+	if (!kvm_pte_valid(ctx->old) || !stage2_pte_cacheable(pgt, ctx->old))
 		return 0;
 
 	if (mm_ops->dcache_clean_inval_poc)
-		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
-					       kvm_granule_size(level));
+		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(ctx->old, mm_ops),
+					       kvm_granule_size(ctx->level));
 	return 0;
 }
 
@@ -1184,7 +1189,7 @@ int __kvm_pgtable_stage2_init(struct kvm_pgtable *pgt, struct kvm_s2_mmu *mmu,
 	u32 start_level = VTCR_EL2_TGRAN_SL0_BASE - sl0;
 
 	pgd_sz = kvm_pgd_pages(ia_bits, start_level) * PAGE_SIZE;
-	pgt->pgd = mm_ops->zalloc_pages_exact(pgd_sz);
+	pgt->pgd = (kvm_pteref_t)mm_ops->zalloc_pages_exact(pgd_sz);
 	if (!pgt->pgd)
 		return -ENOMEM;
 
@@ -1200,20 +1205,27 @@ int __kvm_pgtable_stage2_init(struct kvm_pgtable *pgt, struct kvm_s2_mmu *mmu,
 	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)
+size_t kvm_pgtable_stage2_pgd_size(u64 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;
+
+	return kvm_pgd_pages(ia_bits, start_level) * PAGE_SIZE;
+}
+
+static int stage2_free_walker(const struct kvm_pgtable_visit_ctx *ctx,
+			      enum kvm_pgtable_walk_flags visit)
 {
-	struct kvm_pgtable_mm_ops *mm_ops = arg;
-	kvm_pte_t pte = *ptep;
+	struct kvm_pgtable_mm_ops *mm_ops = ctx->mm_ops;
 
-	if (!stage2_pte_is_counted(pte))
+	if (!stage2_pte_is_counted(ctx->old))
 		return 0;
 
-	mm_ops->put_page(ptep);
+	mm_ops->put_page(ctx->ptep);
 
-	if (kvm_pte_table(pte, level))
-		mm_ops->put_page(kvm_pte_follow(pte, mm_ops));
+	if (kvm_pte_table(ctx->old, ctx->level))
+		mm_ops->put_page(kvm_pte_follow(ctx->old, mm_ops));
 
 	return 0;
 }
@@ -1225,11 +1237,33 @@ void kvm_pgtable_stage2_destroy(struct kvm_pgtable *pgt)
 		.cb	= stage2_free_walker,
 		.flags	= KVM_PGTABLE_WALK_LEAF |
 			  KVM_PGTABLE_WALK_TABLE_POST,
-		.arg	= pgt->mm_ops,
 	};
 
 	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;
-	pgt->mm_ops->free_pages_exact(pgt->pgd, pgd_sz);
+	pgt->mm_ops->free_pages_exact(kvm_dereference_pteref(&walker, pgt->pgd), pgd_sz);
 	pgt->pgd = NULL;
 }
+
+void kvm_pgtable_stage2_free_removed(struct kvm_pgtable_mm_ops *mm_ops, void *pgtable, u32 level)
+{
+	kvm_pteref_t ptep = (kvm_pteref_t)pgtable;
+	struct kvm_pgtable_walker walker = {
+		.cb	= stage2_free_walker,
+		.flags	= KVM_PGTABLE_WALK_LEAF |
+			  KVM_PGTABLE_WALK_TABLE_POST,
+	};
+	struct kvm_pgtable_walk_data data = {
+		.walker	= &walker,
+
+		/*
+		 * At this point the IPA really doesn't matter, as the page
+		 * table being traversed has already been removed from the stage
+		 * 2. Set an appropriate range to cover the entire page table.
+		 */
+		.addr	= 0,
+		.end	= kvm_granule_size(level),
+	};
+
+	WARN_ON(__kvm_pgtable_walk(&data, mm_ops, ptep, level + 1));
+}
diff --git a/arch/arm64/kvm/hyp/vhe/Makefile b/arch/arm64/kvm/hyp/vhe/Makefile
index 96bec0ecf9dd..3b9e5464b5b3 100644
--- a/arch/arm64/kvm/hyp/vhe/Makefile
+++ b/arch/arm64/kvm/hyp/vhe/Makefile
@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0
 #
-# Makefile for Kernel-based Virtual Machine module, HYP/nVHE part
+# Makefile for Kernel-based Virtual Machine module, HYP/VHE part
 #
 
 asflags-y := -D__KVM_VHE_HYPERVISOR__