LoongArch: KVM: Implement kvm mmu operations

Implement LoongArch kvm mmu, it is used to switch gpa to hpa when guest
exit because of address translation exception.

This patch implement: allocating gpa page table, searching gpa from it,
and flushing guest gpa in the table.

Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Tested-by: Huacai Chen <chenhuacai@loongson.cn>
Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>

1 files changed, 914 insertions, 0 deletions

diff --git a/arch/loongarch/kvm/mmu.c b/arch/loongarch/kvm/mmu.c
new file mode 100644
index 000000000000..80480df5f550
--- /dev/null
+++ b/arch/loongarch/kvm/mmu.c
@@ -0,0 +1,914 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020-2023 Loongson Technology Corporation Limited
+ */
+
+#include <linux/highmem.h>
+#include <linux/hugetlb.h>
+#include <linux/kvm_host.h>
+#include <linux/page-flags.h>
+#include <linux/uaccess.h>
+#include <asm/mmu_context.h>
+#include <asm/pgalloc.h>
+#include <asm/tlb.h>
+#include <asm/kvm_mmu.h>
+
+static inline void kvm_ptw_prepare(struct kvm *kvm, kvm_ptw_ctx *ctx)
+{
+	ctx->level = kvm->arch.root_level;
+	/* pte table */
+	ctx->invalid_ptes  = kvm->arch.invalid_ptes;
+	ctx->pte_shifts    = kvm->arch.pte_shifts;
+	ctx->pgtable_shift = ctx->pte_shifts[ctx->level];
+	ctx->invalid_entry = ctx->invalid_ptes[ctx->level];
+	ctx->opaque        = kvm;
+}
+
+/*
+ * Mark a range of guest physical address space old (all accesses fault) in the
+ * VM's GPA page table to allow detection of commonly used pages.
+ */
+static int kvm_mkold_pte(kvm_pte_t *pte, phys_addr_t addr, kvm_ptw_ctx *ctx)
+{
+	if (kvm_pte_young(*pte)) {
+		*pte = kvm_pte_mkold(*pte);
+		return 1;
+	}
+
+	return 0;
+}
+
+/*
+ * Mark a range of guest physical address space clean (writes fault) in the VM's
+ * GPA page table to allow dirty page tracking.
+ */
+static int kvm_mkclean_pte(kvm_pte_t *pte, phys_addr_t addr, kvm_ptw_ctx *ctx)
+{
+	gfn_t offset;
+	kvm_pte_t val;
+
+	val = *pte;
+	/*
+	 * For kvm_arch_mmu_enable_log_dirty_pt_masked with mask, start and end
+	 * may cross hugepage, for first huge page parameter addr is equal to
+	 * start, however for the second huge page addr is base address of
+	 * this huge page, rather than start or end address
+	 */
+	if ((ctx->flag & _KVM_HAS_PGMASK) && !kvm_pte_huge(val)) {
+		offset = (addr >> PAGE_SHIFT) - ctx->gfn;
+		if (!(BIT(offset) & ctx->mask))
+			return 0;
+	}
+
+	/*
+	 * Need not split huge page now, just set write-proect pte bit
+	 * Split huge page until next write fault
+	 */
+	if (kvm_pte_dirty(val)) {
+		*pte = kvm_pte_mkclean(val);
+		return 1;
+	}
+
+	return 0;
+}
+
+/*
+ * Clear pte entry
+ */
+static int kvm_flush_pte(kvm_pte_t *pte, phys_addr_t addr, kvm_ptw_ctx *ctx)
+{
+	struct kvm *kvm;
+
+	kvm = ctx->opaque;
+	if (ctx->level)
+		kvm->stat.hugepages--;
+	else
+		kvm->stat.pages--;
+
+	*pte = ctx->invalid_entry;
+
+	return 1;
+}
+
+/*
+ * kvm_pgd_alloc() - Allocate and initialise a KVM GPA page directory.
+ *
+ * Allocate a blank KVM GPA page directory (PGD) for representing guest physical
+ * to host physical page mappings.
+ *
+ * Returns:	Pointer to new KVM GPA page directory.
+ *		NULL on allocation failure.
+ */
+kvm_pte_t *kvm_pgd_alloc(void)
+{
+	kvm_pte_t *pgd;
+
+	pgd = (kvm_pte_t *)__get_free_pages(GFP_KERNEL, 0);
+	if (pgd)
+		pgd_init((void *)pgd);
+
+	return pgd;
+}
+
+static void _kvm_pte_init(void *addr, unsigned long val)
+{
+	unsigned long *p, *end;
+
+	p = (unsigned long *)addr;
+	end = p + PTRS_PER_PTE;
+	do {
+		p[0] = val;
+		p[1] = val;
+		p[2] = val;
+		p[3] = val;
+		p[4] = val;
+		p += 8;
+		p[-3] = val;
+		p[-2] = val;
+		p[-1] = val;
+	} while (p != end);
+}
+
+/*
+ * Caller must hold kvm->mm_lock
+ *
+ * Walk the page tables of kvm to find the PTE corresponding to the
+ * address @addr. If page tables don't exist for @addr, they will be created
+ * from the MMU cache if @cache is not NULL.
+ */
+static kvm_pte_t *kvm_populate_gpa(struct kvm *kvm,
+				struct kvm_mmu_memory_cache *cache,
+				unsigned long addr, int level)
+{
+	kvm_ptw_ctx ctx;
+	kvm_pte_t *entry, *child;
+
+	kvm_ptw_prepare(kvm, &ctx);
+	child = kvm->arch.pgd;
+	while (ctx.level > level) {
+		entry = kvm_pgtable_offset(&ctx, child, addr);
+		if (kvm_pte_none(&ctx, entry)) {
+			if (!cache)
+				return NULL;
+
+			child = kvm_mmu_memory_cache_alloc(cache);
+			_kvm_pte_init(child, ctx.invalid_ptes[ctx.level - 1]);
+			kvm_set_pte(entry, __pa(child));
+		} else if (kvm_pte_huge(*entry)) {
+			return entry;
+		} else
+			child = (kvm_pte_t *)__va(PHYSADDR(*entry));
+		kvm_ptw_enter(&ctx);
+	}
+
+	entry = kvm_pgtable_offset(&ctx, child, addr);
+
+	return entry;
+}
+
+/*
+ * Page walker for VM shadow mmu at last level
+ * The last level is small pte page or huge pmd page
+ */
+static int kvm_ptw_leaf(kvm_pte_t *dir, phys_addr_t addr, phys_addr_t end, kvm_ptw_ctx *ctx)
+{
+	int ret;
+	phys_addr_t next, start, size;
+	struct list_head *list;
+	kvm_pte_t *entry, *child;
+
+	ret = 0;
+	start = addr;
+	child = (kvm_pte_t *)__va(PHYSADDR(*dir));
+	entry = kvm_pgtable_offset(ctx, child, addr);
+	do {
+		next = addr + (0x1UL << ctx->pgtable_shift);
+		if (!kvm_pte_present(ctx, entry))
+			continue;
+
+		ret |= ctx->ops(entry, addr, ctx);
+	} while (entry++, addr = next, addr < end);
+
+	if (kvm_need_flush(ctx)) {
+		size = 0x1UL << (ctx->pgtable_shift + PAGE_SHIFT - 3);
+		if (start + size == end) {
+			list = (struct list_head *)child;
+			list_add_tail(list, &ctx->list);
+			*dir = ctx->invalid_ptes[ctx->level + 1];
+		}
+	}
+
+	return ret;
+}
+
+/*
+ * Page walker for VM shadow mmu at page table dir level
+ */
+static int kvm_ptw_dir(kvm_pte_t *dir, phys_addr_t addr, phys_addr_t end, kvm_ptw_ctx *ctx)
+{
+	int ret;
+	phys_addr_t next, start, size;
+	struct list_head *list;
+	kvm_pte_t *entry, *child;
+
+	ret = 0;
+	start = addr;
+	child = (kvm_pte_t *)__va(PHYSADDR(*dir));
+	entry = kvm_pgtable_offset(ctx, child, addr);
+	do {
+		next = kvm_pgtable_addr_end(ctx, addr, end);
+		if (!kvm_pte_present(ctx, entry))
+			continue;
+
+		if (kvm_pte_huge(*entry)) {
+			ret |= ctx->ops(entry, addr, ctx);
+			continue;
+		}
+
+		kvm_ptw_enter(ctx);
+		if (ctx->level == 0)
+			ret |= kvm_ptw_leaf(entry, addr, next, ctx);
+		else
+			ret |= kvm_ptw_dir(entry, addr, next, ctx);
+		kvm_ptw_exit(ctx);
+	}  while (entry++, addr = next, addr < end);
+
+	if (kvm_need_flush(ctx)) {
+		size = 0x1UL << (ctx->pgtable_shift + PAGE_SHIFT - 3);
+		if (start + size == end) {
+			list = (struct list_head *)child;
+			list_add_tail(list, &ctx->list);
+			*dir = ctx->invalid_ptes[ctx->level + 1];
+		}
+	}
+
+	return ret;
+}
+
+/*
+ * Page walker for VM shadow mmu at page root table
+ */
+static int kvm_ptw_top(kvm_pte_t *dir, phys_addr_t addr, phys_addr_t end, kvm_ptw_ctx *ctx)
+{
+	int ret;
+	phys_addr_t next;
+	kvm_pte_t *entry;
+
+	ret = 0;
+	entry = kvm_pgtable_offset(ctx, dir, addr);
+	do {
+		next = kvm_pgtable_addr_end(ctx, addr, end);
+		if (!kvm_pte_present(ctx, entry))
+			continue;
+
+		kvm_ptw_enter(ctx);
+		ret |= kvm_ptw_dir(entry, addr, next, ctx);
+		kvm_ptw_exit(ctx);
+	}  while (entry++, addr = next, addr < end);
+
+	return ret;
+}
+
+/*
+ * kvm_flush_range() - Flush a range of guest physical addresses.
+ * @kvm:	KVM pointer.
+ * @start_gfn:	Guest frame number of first page in GPA range to flush.
+ * @end_gfn:	Guest frame number of last page in GPA range to flush.
+ * @lock:	Whether to hold mmu_lock or not
+ *
+ * Flushes a range of GPA mappings from the GPA page tables.
+ */
+static void kvm_flush_range(struct kvm *kvm, gfn_t start_gfn, gfn_t end_gfn, int lock)
+{
+	int ret;
+	kvm_ptw_ctx ctx;
+	struct list_head *pos, *temp;
+
+	ctx.ops = kvm_flush_pte;
+	ctx.flag = _KVM_FLUSH_PGTABLE;
+	kvm_ptw_prepare(kvm, &ctx);
+	INIT_LIST_HEAD(&ctx.list);
+
+	if (lock) {
+		spin_lock(&kvm->mmu_lock);
+		ret = kvm_ptw_top(kvm->arch.pgd, start_gfn << PAGE_SHIFT,
+					end_gfn << PAGE_SHIFT, &ctx);
+		spin_unlock(&kvm->mmu_lock);
+	} else
+		ret = kvm_ptw_top(kvm->arch.pgd, start_gfn << PAGE_SHIFT,
+					end_gfn << PAGE_SHIFT, &ctx);
+
+	/* Flush vpid for each vCPU individually */
+	if (ret)
+		kvm_flush_remote_tlbs(kvm);
+
+	/*
+	 * free pte table page after mmu_lock
+	 * the pte table page is linked together with ctx.list
+	 */
+	list_for_each_safe(pos, temp, &ctx.list) {
+		list_del(pos);
+		free_page((unsigned long)pos);
+	}
+}
+
+/*
+ * kvm_mkclean_gpa_pt() - Make a range of guest physical addresses clean.
+ * @kvm:	KVM pointer.
+ * @start_gfn:	Guest frame number of first page in GPA range to flush.
+ * @end_gfn:	Guest frame number of last page in GPA range to flush.
+ *
+ * Make a range of GPA mappings clean so that guest writes will fault and
+ * trigger dirty page logging.
+ *
+ * The caller must hold the @kvm->mmu_lock spinlock.
+ *
+ * Returns:	Whether any GPA mappings were modified, which would require
+ *		derived mappings (GVA page tables & TLB enties) to be
+ *		invalidated.
+ */
+static int kvm_mkclean_gpa_pt(struct kvm *kvm, gfn_t start_gfn, gfn_t end_gfn)
+{
+	kvm_ptw_ctx ctx;
+
+	ctx.ops = kvm_mkclean_pte;
+	ctx.flag = 0;
+	kvm_ptw_prepare(kvm, &ctx);
+	return kvm_ptw_top(kvm->arch.pgd, start_gfn << PAGE_SHIFT, end_gfn << PAGE_SHIFT, &ctx);
+}
+
+/*
+ * kvm_arch_mmu_enable_log_dirty_pt_masked() - write protect dirty pages
+ * @kvm:	The KVM pointer
+ * @slot:	The memory slot associated with mask
+ * @gfn_offset:	The gfn offset in memory slot
+ * @mask:	The mask of dirty pages at offset 'gfn_offset' in this memory
+ *		slot to be write protected
+ *
+ * Walks bits set in mask write protects the associated pte's. Caller must
+ * acquire @kvm->mmu_lock.
+ */
+void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
+		struct kvm_memory_slot *slot, gfn_t gfn_offset, unsigned long mask)
+{
+	kvm_ptw_ctx ctx;
+	gfn_t base_gfn = slot->base_gfn + gfn_offset;
+	gfn_t start = base_gfn + __ffs(mask);
+	gfn_t end = base_gfn + __fls(mask) + 1;
+
+	ctx.ops = kvm_mkclean_pte;
+	ctx.flag = _KVM_HAS_PGMASK;
+	ctx.mask = mask;
+	ctx.gfn = base_gfn;
+	kvm_ptw_prepare(kvm, &ctx);
+
+	kvm_ptw_top(kvm->arch.pgd, start << PAGE_SHIFT, end << PAGE_SHIFT, &ctx);
+}
+
+void kvm_arch_commit_memory_region(struct kvm *kvm,
+				   struct kvm_memory_slot *old,
+				   const struct kvm_memory_slot *new,
+				   enum kvm_mr_change change)
+{
+	int needs_flush;
+
+	/*
+	 * If dirty page logging is enabled, write protect all pages in the slot
+	 * ready for dirty logging.
+	 *
+	 * There is no need to do this in any of the following cases:
+	 * CREATE:	No dirty mappings will already exist.
+	 * MOVE/DELETE:	The old mappings will already have been cleaned up by
+	 *		kvm_arch_flush_shadow_memslot()
+	 */
+	if (change == KVM_MR_FLAGS_ONLY &&
+	    (!(old->flags & KVM_MEM_LOG_DIRTY_PAGES) &&
+	     new->flags & KVM_MEM_LOG_DIRTY_PAGES)) {
+		spin_lock(&kvm->mmu_lock);
+		/* Write protect GPA page table entries */
+		needs_flush = kvm_mkclean_gpa_pt(kvm, new->base_gfn,
+					new->base_gfn + new->npages);
+		spin_unlock(&kvm->mmu_lock);
+		if (needs_flush)
+			kvm_flush_remote_tlbs(kvm);
+	}
+}
+
+void kvm_arch_flush_shadow_all(struct kvm *kvm)
+{
+	kvm_flush_range(kvm, 0, kvm->arch.gpa_size >> PAGE_SHIFT, 0);
+}
+
+void kvm_arch_flush_shadow_memslot(struct kvm *kvm, struct kvm_memory_slot *slot)
+{
+	/*
+	 * The slot has been made invalid (ready for moving or deletion), so we
+	 * need to ensure that it can no longer be accessed by any guest vCPUs.
+	 */
+	kvm_flush_range(kvm, slot->base_gfn, slot->base_gfn + slot->npages, 1);
+}
+
+bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
+{
+	kvm_ptw_ctx ctx;
+
+	ctx.flag = 0;
+	ctx.ops = kvm_flush_pte;
+	kvm_ptw_prepare(kvm, &ctx);
+	INIT_LIST_HEAD(&ctx.list);
+
+	return kvm_ptw_top(kvm->arch.pgd, range->start << PAGE_SHIFT,
+			range->end << PAGE_SHIFT, &ctx);
+}
+
+bool kvm_set_spte_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
+{
+	unsigned long prot_bits;
+	kvm_pte_t *ptep;
+	kvm_pfn_t pfn = pte_pfn(range->arg.pte);
+	gpa_t gpa = range->start << PAGE_SHIFT;
+
+	ptep = kvm_populate_gpa(kvm, NULL, gpa, 0);
+	if (!ptep)
+		return false;
+
+	/* Replacing an absent or old page doesn't need flushes */
+	if (!kvm_pte_present(NULL, ptep) || !kvm_pte_young(*ptep)) {
+		kvm_set_pte(ptep, 0);
+		return false;
+	}
+
+	/* Fill new pte if write protected or page migrated */
+	prot_bits = _PAGE_PRESENT | __READABLE;
+	prot_bits |= _CACHE_MASK & pte_val(range->arg.pte);
+
+	/*
+	 * Set _PAGE_WRITE or _PAGE_DIRTY iff old and new pte both support
+	 * _PAGE_WRITE for map_page_fast if next page write fault
+	 * _PAGE_DIRTY since gpa has already recorded as dirty page
+	 */
+	prot_bits |= __WRITEABLE & *ptep & pte_val(range->arg.pte);
+	kvm_set_pte(ptep, kvm_pfn_pte(pfn, __pgprot(prot_bits)));
+
+	return true;
+}
+
+bool kvm_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
+{
+	kvm_ptw_ctx ctx;
+
+	ctx.flag = 0;
+	ctx.ops = kvm_mkold_pte;
+	kvm_ptw_prepare(kvm, &ctx);
+
+	return kvm_ptw_top(kvm->arch.pgd, range->start << PAGE_SHIFT,
+				range->end << PAGE_SHIFT, &ctx);
+}
+
+bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
+{
+	gpa_t gpa = range->start << PAGE_SHIFT;
+	kvm_pte_t *ptep = kvm_populate_gpa(kvm, NULL, gpa, 0);
+
+	if (ptep && kvm_pte_present(NULL, ptep) && kvm_pte_young(*ptep))
+		return true;
+
+	return false;
+}
+
+/*
+ * kvm_map_page_fast() - Fast path GPA fault handler.
+ * @vcpu:		vCPU pointer.
+ * @gpa:		Guest physical address of fault.
+ * @write:	Whether the fault was due to a write.
+ *
+ * Perform fast path GPA fault handling, doing all that can be done without
+ * calling into KVM. This handles marking old pages young (for idle page
+ * tracking), and dirtying of clean pages (for dirty page logging).
+ *
+ * Returns:	0 on success, in which case we can update derived mappings and
+ *		resume guest execution.
+ *		-EFAULT on failure due to absent GPA mapping or write to
+ *		read-only page, in which case KVM must be consulted.
+ */
+static int kvm_map_page_fast(struct kvm_vcpu *vcpu, unsigned long gpa, bool write)
+{
+	int ret = 0;
+	kvm_pfn_t pfn = 0;
+	kvm_pte_t *ptep, changed, new;
+	gfn_t gfn = gpa >> PAGE_SHIFT;
+	struct kvm *kvm = vcpu->kvm;
+	struct kvm_memory_slot *slot;
+
+	spin_lock(&kvm->mmu_lock);
+
+	/* Fast path - just check GPA page table for an existing entry */
+	ptep = kvm_populate_gpa(kvm, NULL, gpa, 0);
+	if (!ptep || !kvm_pte_present(NULL, ptep)) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	/* Track access to pages marked old */
+	new = *ptep;
+	if (!kvm_pte_young(new))
+		new = kvm_pte_mkyoung(new);
+		/* call kvm_set_pfn_accessed() after unlock */
+
+	if (write && !kvm_pte_dirty(new)) {
+		if (!kvm_pte_write(new)) {
+			ret = -EFAULT;
+			goto out;
+		}
+
+		if (kvm_pte_huge(new)) {
+			/*
+			 * Do not set write permission when dirty logging is
+			 * enabled for HugePages
+			 */
+			slot = gfn_to_memslot(kvm, gfn);
+			if (kvm_slot_dirty_track_enabled(slot)) {
+				ret = -EFAULT;
+				goto out;
+			}
+		}
+
+		/* Track dirtying of writeable pages */
+		new = kvm_pte_mkdirty(new);
+	}
+
+	changed = new ^ (*ptep);
+	if (changed) {
+		kvm_set_pte(ptep, new);
+		pfn = kvm_pte_pfn(new);
+	}
+	spin_unlock(&kvm->mmu_lock);
+
+	/*
+	 * Fixme: pfn may be freed after mmu_lock
+	 * kvm_try_get_pfn(pfn)/kvm_release_pfn pair to prevent this?
+	 */
+	if (kvm_pte_young(changed))
+		kvm_set_pfn_accessed(pfn);
+
+	if (kvm_pte_dirty(changed)) {
+		mark_page_dirty(kvm, gfn);
+		kvm_set_pfn_dirty(pfn);
+	}
+	return ret;
+out:
+	spin_unlock(&kvm->mmu_lock);
+	return ret;
+}
+
+static bool fault_supports_huge_mapping(struct kvm_memory_slot *memslot,
+				unsigned long hva, unsigned long map_size, bool write)
+{
+	size_t size;
+	gpa_t gpa_start;
+	hva_t uaddr_start, uaddr_end;
+
+	/* Disable dirty logging on HugePages */
+	if (kvm_slot_dirty_track_enabled(memslot) && write)
+		return false;
+
+	size = memslot->npages * PAGE_SIZE;
+	gpa_start = memslot->base_gfn << PAGE_SHIFT;
+	uaddr_start = memslot->userspace_addr;
+	uaddr_end = uaddr_start + size;
+
+	/*
+	 * Pages belonging to memslots that don't have the same alignment
+	 * within a PMD for userspace and GPA cannot be mapped with stage-2
+	 * PMD entries, because we'll end up mapping the wrong pages.
+	 *
+	 * Consider a layout like the following:
+	 *
+	 *    memslot->userspace_addr:
+	 *    +-----+--------------------+--------------------+---+
+	 *    |abcde|fgh  Stage-1 block  |    Stage-1 block tv|xyz|
+	 *    +-----+--------------------+--------------------+---+
+	 *
+	 *    memslot->base_gfn << PAGE_SIZE:
+	 *      +---+--------------------+--------------------+-----+
+	 *      |abc|def  Stage-2 block  |    Stage-2 block   |tvxyz|
+	 *      +---+--------------------+--------------------+-----+
+	 *
+	 * If we create those stage-2 blocks, we'll end up with this incorrect
+	 * mapping:
+	 *   d -> f
+	 *   e -> g
+	 *   f -> h
+	 */
+	if ((gpa_start & (map_size - 1)) != (uaddr_start & (map_size - 1)))
+		return false;
+
+	/*
+	 * Next, let's make sure we're not trying to map anything not covered
+	 * by the memslot. This means we have to prohibit block size mappings
+	 * for the beginning and end of a non-block aligned and non-block sized
+	 * memory slot (illustrated by the head and tail parts of the
+	 * userspace view above containing pages 'abcde' and 'xyz',
+	 * respectively).
+	 *
+	 * Note that it doesn't matter if we do the check using the
+	 * userspace_addr or the base_gfn, as both are equally aligned (per
+	 * the check above) and equally sized.
+	 */
+	return (hva & ~(map_size - 1)) >= uaddr_start &&
+		(hva & ~(map_size - 1)) + map_size <= uaddr_end;
+}
+
+/*
+ * Lookup the mapping level for @gfn in the current mm.
+ *
+ * WARNING!  Use of host_pfn_mapping_level() requires the caller and the end
+ * consumer to be tied into KVM's handlers for MMU notifier events!
+ *
+ * There are several ways to safely use this helper:
+ *
+ * - Check mmu_invalidate_retry_hva() after grabbing the mapping level, before
+ *   consuming it.  In this case, mmu_lock doesn't need to be held during the
+ *   lookup, but it does need to be held while checking the MMU notifier.
+ *
+ * - Hold mmu_lock AND ensure there is no in-progress MMU notifier invalidation
+ *   event for the hva.  This can be done by explicit checking the MMU notifier
+ *   or by ensuring that KVM already has a valid mapping that covers the hva.
+ *
+ * - Do not use the result to install new mappings, e.g. use the host mapping
+ *   level only to decide whether or not to zap an entry.  In this case, it's
+ *   not required to hold mmu_lock (though it's highly likely the caller will
+ *   want to hold mmu_lock anyways, e.g. to modify SPTEs).
+ *
+ * Note!  The lookup can still race with modifications to host page tables, but
+ * the above "rules" ensure KVM will not _consume_ the result of the walk if a
+ * race with the primary MMU occurs.
+ */
+static int host_pfn_mapping_level(struct kvm *kvm, gfn_t gfn,
+				const struct kvm_memory_slot *slot)
+{
+	int level = 0;
+	unsigned long hva;
+	unsigned long flags;
+	pgd_t pgd;
+	p4d_t p4d;
+	pud_t pud;
+	pmd_t pmd;
+
+	/*
+	 * Note, using the already-retrieved memslot and __gfn_to_hva_memslot()
+	 * is not solely for performance, it's also necessary to avoid the
+	 * "writable" check in __gfn_to_hva_many(), which will always fail on
+	 * read-only memslots due to gfn_to_hva() assuming writes.  Earlier
+	 * page fault steps have already verified the guest isn't writing a
+	 * read-only memslot.
+	 */
+	hva = __gfn_to_hva_memslot(slot, gfn);
+
+	/*
+	 * Disable IRQs to prevent concurrent tear down of host page tables,
+	 * e.g. if the primary MMU promotes a P*D to a huge page and then frees
+	 * the original page table.
+	 */
+	local_irq_save(flags);
+
+	/*
+	 * Read each entry once.  As above, a non-leaf entry can be promoted to
+	 * a huge page _during_ this walk.  Re-reading the entry could send the
+	 * walk into the weeks, e.g. p*d_large() returns false (sees the old
+	 * value) and then p*d_offset() walks into the target huge page instead
+	 * of the old page table (sees the new value).
+	 */
+	pgd = READ_ONCE(*pgd_offset(kvm->mm, hva));
+	if (pgd_none(pgd))
+		goto out;
+
+	p4d = READ_ONCE(*p4d_offset(&pgd, hva));
+	if (p4d_none(p4d) || !p4d_present(p4d))
+		goto out;
+
+	pud = READ_ONCE(*pud_offset(&p4d, hva));
+	if (pud_none(pud) || !pud_present(pud))
+		goto out;
+
+	pmd = READ_ONCE(*pmd_offset(&pud, hva));
+	if (pmd_none(pmd) || !pmd_present(pmd))
+		goto out;
+
+	if (kvm_pte_huge(pmd_val(pmd)))
+		level = 1;
+
+out:
+	local_irq_restore(flags);
+	return level;
+}
+
+/*
+ * Split huge page
+ */
+static kvm_pte_t *kvm_split_huge(struct kvm_vcpu *vcpu, kvm_pte_t *ptep, gfn_t gfn)
+{
+	int i;
+	kvm_pte_t val, *child;
+	struct kvm *kvm = vcpu->kvm;
+	struct kvm_mmu_memory_cache *memcache;
+
+	memcache = &vcpu->arch.mmu_page_cache;
+	child = kvm_mmu_memory_cache_alloc(memcache);
+	val = kvm_pte_mksmall(*ptep);
+	for (i = 0; i < PTRS_PER_PTE; i++) {
+		kvm_set_pte(child + i, val);
+		val += PAGE_SIZE;
+	}
+
+	/* The later kvm_flush_tlb_gpa() will flush hugepage tlb */
+	kvm_set_pte(ptep, __pa(child));
+
+	kvm->stat.hugepages--;
+	kvm->stat.pages += PTRS_PER_PTE;
+
+	return child + (gfn & (PTRS_PER_PTE - 1));
+}
+
+/*
+ * kvm_map_page() - Map a guest physical page.
+ * @vcpu:		vCPU pointer.
+ * @gpa:		Guest physical address of fault.
+ * @write:	Whether the fault was due to a write.
+ *
+ * Handle GPA faults by creating a new GPA mapping (or updating an existing
+ * one).
+ *
+ * This takes care of marking pages young or dirty (idle/dirty page tracking),
+ * asking KVM for the corresponding PFN, and creating a mapping in the GPA page
+ * tables. Derived mappings (GVA page tables and TLBs) must be handled by the
+ * caller.
+ *
+ * Returns:	0 on success
+ *		-EFAULT if there is no memory region at @gpa or a write was
+ *		attempted to a read-only memory region. This is usually handled
+ *		as an MMIO access.
+ */
+static int kvm_map_page(struct kvm_vcpu *vcpu, unsigned long gpa, bool write)
+{
+	bool writeable;
+	int srcu_idx, err, retry_no = 0, level;
+	unsigned long hva, mmu_seq, prot_bits;
+	kvm_pfn_t pfn;
+	kvm_pte_t *ptep, new_pte;
+	gfn_t gfn = gpa >> PAGE_SHIFT;
+	struct kvm *kvm = vcpu->kvm;
+	struct kvm_memory_slot *memslot;
+	struct kvm_mmu_memory_cache *memcache = &vcpu->arch.mmu_page_cache;
+
+	/* Try the fast path to handle old / clean pages */
+	srcu_idx = srcu_read_lock(&kvm->srcu);
+	err = kvm_map_page_fast(vcpu, gpa, write);
+	if (!err)
+		goto out;
+
+	memslot = gfn_to_memslot(kvm, gfn);
+	hva = gfn_to_hva_memslot_prot(memslot, gfn, &writeable);
+	if (kvm_is_error_hva(hva) || (write && !writeable)) {
+		err = -EFAULT;
+		goto out;
+	}
+
+	/* We need a minimum of cached pages ready for page table creation */
+	err = kvm_mmu_topup_memory_cache(memcache, KVM_MMU_CACHE_MIN_PAGES);
+	if (err)
+		goto out;
+
+retry:
+	/*
+	 * Used to check for invalidations in progress, of the pfn that is
+	 * returned by pfn_to_pfn_prot below.
+	 */
+	mmu_seq = kvm->mmu_invalidate_seq;
+	/*
+	 * Ensure the read of mmu_invalidate_seq isn't reordered with PTE reads in
+	 * gfn_to_pfn_prot() (which calls get_user_pages()), so that we don't
+	 * risk the page we get a reference to getting unmapped before we have a
+	 * chance to grab the mmu_lock without mmu_invalidate_retry() noticing.
+	 *
+	 * This smp_rmb() pairs with the effective smp_wmb() of the combination
+	 * of the pte_unmap_unlock() after the PTE is zapped, and the
+	 * spin_lock() in kvm_mmu_invalidate_invalidate_<page|range_end>() before
+	 * mmu_invalidate_seq is incremented.
+	 */
+	smp_rmb();
+
+	/* Slow path - ask KVM core whether we can access this GPA */
+	pfn = gfn_to_pfn_prot(kvm, gfn, write, &writeable);
+	if (is_error_noslot_pfn(pfn)) {
+		err = -EFAULT;
+		goto out;
+	}
+
+	/* Check if an invalidation has taken place since we got pfn */
+	spin_lock(&kvm->mmu_lock);
+	if (mmu_invalidate_retry_hva(kvm, mmu_seq, hva)) {
+		/*
+		 * This can happen when mappings are changed asynchronously, but
+		 * also synchronously if a COW is triggered by
+		 * gfn_to_pfn_prot().
+		 */
+		spin_unlock(&kvm->mmu_lock);
+		kvm_release_pfn_clean(pfn);
+		if (retry_no > 100) {
+			retry_no = 0;
+			schedule();
+		}
+		retry_no++;
+		goto retry;
+	}
+
+	/*
+	 * For emulated devices such virtio device, actual cache attribute is
+	 * determined by physical machine.
+	 * For pass through physical device, it should be uncachable
+	 */
+	prot_bits = _PAGE_PRESENT | __READABLE;
+	if (pfn_valid(pfn))
+		prot_bits |= _CACHE_CC;
+	else
+		prot_bits |= _CACHE_SUC;
+
+	if (writeable) {
+		prot_bits |= _PAGE_WRITE;
+		if (write)
+			prot_bits |= __WRITEABLE;
+	}
+
+	/* Disable dirty logging on HugePages */
+	level = 0;
+	if (!fault_supports_huge_mapping(memslot, hva, PMD_SIZE, write)) {
+		level = 0;
+	} else {
+		level = host_pfn_mapping_level(kvm, gfn, memslot);
+		if (level == 1) {
+			gfn = gfn & ~(PTRS_PER_PTE - 1);
+			pfn = pfn & ~(PTRS_PER_PTE - 1);
+		}
+	}
+
+	/* Ensure page tables are allocated */
+	ptep = kvm_populate_gpa(kvm, memcache, gpa, level);
+	new_pte = kvm_pfn_pte(pfn, __pgprot(prot_bits));
+	if (level == 1) {
+		new_pte = kvm_pte_mkhuge(new_pte);
+		/*
+		 * previous pmd entry is invalid_pte_table
+		 * there is invalid tlb with small page
+		 * need flush these invalid tlbs for current vcpu
+		 */
+		kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
+		++kvm->stat.hugepages;
+	}  else if (kvm_pte_huge(*ptep) && write)
+		ptep = kvm_split_huge(vcpu, ptep, gfn);
+	else
+		++kvm->stat.pages;
+	kvm_set_pte(ptep, new_pte);
+	spin_unlock(&kvm->mmu_lock);
+
+	if (prot_bits & _PAGE_DIRTY) {
+		mark_page_dirty_in_slot(kvm, memslot, gfn);
+		kvm_set_pfn_dirty(pfn);
+	}
+
+	kvm_set_pfn_accessed(pfn);
+	kvm_release_pfn_clean(pfn);
+out:
+	srcu_read_unlock(&kvm->srcu, srcu_idx);
+	return err;
+}
+
+int kvm_handle_mm_fault(struct kvm_vcpu *vcpu, unsigned long gpa, bool write)
+{
+	int ret;
+
+	ret = kvm_map_page(vcpu, gpa, write);
+	if (ret)
+		return ret;
+
+	/* Invalidate this entry in the TLB */
+	kvm_flush_tlb_gpa(vcpu, gpa);
+
+	return 0;
+}
+
+void kvm_arch_sync_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot)
+{
+}
+
+int kvm_arch_prepare_memory_region(struct kvm *kvm, const struct kvm_memory_slot *old,
+				   struct kvm_memory_slot *new, enum kvm_mr_change change)
+{
+	return 0;
+}
+
+void kvm_arch_flush_remote_tlbs_memslot(struct kvm *kvm,
+					const struct kvm_memory_slot *memslot)
+{
+	kvm_flush_remote_tlbs(kvm);
+}