1 files changed, 767 insertions, 459 deletions

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index aee38623b768..9335e1bf72ad 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -148,7 +148,7 @@ module_param(oos_shadow, bool, 0644);
 #define PT64_PERM_MASK (PT_PRESENT_MASK | PT_WRITABLE_MASK | PT_USER_MASK \
 			| PT64_NX_MASK)
 
-#define RMAP_EXT 4
+#define PTE_LIST_EXT 4
 
 #define ACC_EXEC_MASK    1
 #define ACC_WRITE_MASK   PT_WRITABLE_MASK
@@ -164,16 +164,16 @@ module_param(oos_shadow, bool, 0644);
 
 #define SHADOW_PT_INDEX(addr, level) PT64_INDEX(addr, level)
 
-struct kvm_rmap_desc {
-	u64 *sptes[RMAP_EXT];
-	struct kvm_rmap_desc *more;
+struct pte_list_desc {
+	u64 *sptes[PTE_LIST_EXT];
+	struct pte_list_desc *more;
 };
 
 struct kvm_shadow_walk_iterator {
 	u64 addr;
 	hpa_t shadow_addr;
-	int level;
 	u64 *sptep;
+	int level;
 	unsigned index;
 };
 
@@ -182,32 +182,68 @@ struct kvm_shadow_walk_iterator {
 	     shadow_walk_okay(&(_walker));			\
 	     shadow_walk_next(&(_walker)))
 
-typedef void (*mmu_parent_walk_fn) (struct kvm_mmu_page *sp, u64 *spte);
+#define for_each_shadow_entry_lockless(_vcpu, _addr, _walker, spte)	\
+	for (shadow_walk_init(&(_walker), _vcpu, _addr);		\
+	     shadow_walk_okay(&(_walker)) &&				\
+		({ spte = mmu_spte_get_lockless(_walker.sptep); 1; });	\
+	     __shadow_walk_next(&(_walker), spte))
 
-static struct kmem_cache *pte_chain_cache;
-static struct kmem_cache *rmap_desc_cache;
+static struct kmem_cache *pte_list_desc_cache;
 static struct kmem_cache *mmu_page_header_cache;
 static struct percpu_counter kvm_total_used_mmu_pages;
 
-static u64 __read_mostly shadow_trap_nonpresent_pte;
-static u64 __read_mostly shadow_notrap_nonpresent_pte;
 static u64 __read_mostly shadow_nx_mask;
 static u64 __read_mostly shadow_x_mask;	/* mutual exclusive with nx_mask */
 static u64 __read_mostly shadow_user_mask;
 static u64 __read_mostly shadow_accessed_mask;
 static u64 __read_mostly shadow_dirty_mask;
+static u64 __read_mostly shadow_mmio_mask;
 
-static inline u64 rsvd_bits(int s, int e)
+static void mmu_spte_set(u64 *sptep, u64 spte);
+
+void kvm_mmu_set_mmio_spte_mask(u64 mmio_mask)
 {
-	return ((1ULL << (e - s + 1)) - 1) << s;
+	shadow_mmio_mask = mmio_mask;
+}
+EXPORT_SYMBOL_GPL(kvm_mmu_set_mmio_spte_mask);
+
+static void mark_mmio_spte(u64 *sptep, u64 gfn, unsigned access)
+{
+	access &= ACC_WRITE_MASK | ACC_USER_MASK;
+
+	trace_mark_mmio_spte(sptep, gfn, access);
+	mmu_spte_set(sptep, shadow_mmio_mask | access | gfn << PAGE_SHIFT);
 }
 
-void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte)
+static bool is_mmio_spte(u64 spte)
 {
-	shadow_trap_nonpresent_pte = trap_pte;
-	shadow_notrap_nonpresent_pte = notrap_pte;
+	return (spte & shadow_mmio_mask) == shadow_mmio_mask;
+}
+
+static gfn_t get_mmio_spte_gfn(u64 spte)
+{
+	return (spte & ~shadow_mmio_mask) >> PAGE_SHIFT;
+}
+
+static unsigned get_mmio_spte_access(u64 spte)
+{
+	return (spte & ~shadow_mmio_mask) & ~PAGE_MASK;
+}
+
+static bool set_mmio_spte(u64 *sptep, gfn_t gfn, pfn_t pfn, unsigned access)
+{
+	if (unlikely(is_noslot_pfn(pfn))) {
+		mark_mmio_spte(sptep, gfn, access);
+		return true;
+	}
+
+	return false;
+}
+
+static inline u64 rsvd_bits(int s, int e)
+{
+	return ((1ULL << (e - s + 1)) - 1) << s;
 }
-EXPORT_SYMBOL_GPL(kvm_mmu_set_nonpresent_ptes);
 
 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
 		u64 dirty_mask, u64 nx_mask, u64 x_mask)
@@ -220,11 +256,6 @@ void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
 }
 EXPORT_SYMBOL_GPL(kvm_mmu_set_mask_ptes);
 
-static bool is_write_protection(struct kvm_vcpu *vcpu)
-{
-	return kvm_read_cr0_bits(vcpu, X86_CR0_WP);
-}
-
 static int is_cpuid_PSE36(void)
 {
 	return 1;
@@ -237,8 +268,7 @@ static int is_nx(struct kvm_vcpu *vcpu)
 
 static int is_shadow_present_pte(u64 pte)
 {
-	return pte != shadow_trap_nonpresent_pte
-		&& pte != shadow_notrap_nonpresent_pte;
+	return pte & PT_PRESENT_MASK && !is_mmio_spte(pte);
 }
 
 static int is_large_pte(u64 pte)
@@ -246,11 +276,6 @@ static int is_large_pte(u64 pte)
 	return pte & PT_PAGE_SIZE_MASK;
 }
 
-static int is_writable_pte(unsigned long pte)
-{
-	return pte & PT_WRITABLE_MASK;
-}
-
 static int is_dirty_gpte(unsigned long pte)
 {
 	return pte & PT_DIRTY_MASK;
@@ -282,26 +307,154 @@ static gfn_t pse36_gfn_delta(u32 gpte)
 	return (gpte & PT32_DIR_PSE36_MASK) << shift;
 }
 
+#ifdef CONFIG_X86_64
 static void __set_spte(u64 *sptep, u64 spte)
 {
-	set_64bit(sptep, spte);
+	*sptep = spte;
 }
 
-static u64 __xchg_spte(u64 *sptep, u64 new_spte)
+static void __update_clear_spte_fast(u64 *sptep, u64 spte)
 {
-#ifdef CONFIG_X86_64
-	return xchg(sptep, new_spte);
+	*sptep = spte;
+}
+
+static u64 __update_clear_spte_slow(u64 *sptep, u64 spte)
+{
+	return xchg(sptep, spte);
+}
+
+static u64 __get_spte_lockless(u64 *sptep)
+{
+	return ACCESS_ONCE(*sptep);
+}
+
+static bool __check_direct_spte_mmio_pf(u64 spte)
+{
+	/* It is valid if the spte is zapped. */
+	return spte == 0ull;
+}
 #else
-	u64 old_spte;
+union split_spte {
+	struct {
+		u32 spte_low;
+		u32 spte_high;
+	};
+	u64 spte;
+};
 
-	do {
-		old_spte = *sptep;
-	} while (cmpxchg64(sptep, old_spte, new_spte) != old_spte);
+static void count_spte_clear(u64 *sptep, u64 spte)
+{
+	struct kvm_mmu_page *sp =  page_header(__pa(sptep));
 
-	return old_spte;
-#endif
+	if (is_shadow_present_pte(spte))
+		return;
+
+	/* Ensure the spte is completely set before we increase the count */
+	smp_wmb();
+	sp->clear_spte_count++;
+}
+
+static void __set_spte(u64 *sptep, u64 spte)
+{
+	union split_spte *ssptep, sspte;
+
+	ssptep = (union split_spte *)sptep;
+	sspte = (union split_spte)spte;
+
+	ssptep->spte_high = sspte.spte_high;
+
+	/*
+	 * If we map the spte from nonpresent to present, We should store
+	 * the high bits firstly, then set present bit, so cpu can not
+	 * fetch this spte while we are setting the spte.
+	 */
+	smp_wmb();
+
+	ssptep->spte_low = sspte.spte_low;
 }
 
+static void __update_clear_spte_fast(u64 *sptep, u64 spte)
+{
+	union split_spte *ssptep, sspte;
+
+	ssptep = (union split_spte *)sptep;
+	sspte = (union split_spte)spte;
+
+	ssptep->spte_low = sspte.spte_low;
+
+	/*
+	 * If we map the spte from present to nonpresent, we should clear
+	 * present bit firstly to avoid vcpu fetch the old high bits.
+	 */
+	smp_wmb();
+
+	ssptep->spte_high = sspte.spte_high;
+	count_spte_clear(sptep, spte);
+}
+
+static u64 __update_clear_spte_slow(u64 *sptep, u64 spte)
+{
+	union split_spte *ssptep, sspte, orig;
+
+	ssptep = (union split_spte *)sptep;
+	sspte = (union split_spte)spte;
+
+	/* xchg acts as a barrier before the setting of the high bits */
+	orig.spte_low = xchg(&ssptep->spte_low, sspte.spte_low);
+	orig.spte_high = ssptep->spte_high = sspte.spte_high;
+	count_spte_clear(sptep, spte);
+
+	return orig.spte;
+}
+
+/*
+ * The idea using the light way get the spte on x86_32 guest is from
+ * gup_get_pte(arch/x86/mm/gup.c).
+ * The difference is we can not catch the spte tlb flush if we leave
+ * guest mode, so we emulate it by increase clear_spte_count when spte
+ * is cleared.
+ */
+static u64 __get_spte_lockless(u64 *sptep)
+{
+	struct kvm_mmu_page *sp =  page_header(__pa(sptep));
+	union split_spte spte, *orig = (union split_spte *)sptep;
+	int count;
+
+retry:
+	count = sp->clear_spte_count;
+	smp_rmb();
+
+	spte.spte_low = orig->spte_low;
+	smp_rmb();
+
+	spte.spte_high = orig->spte_high;
+	smp_rmb();
+
+	if (unlikely(spte.spte_low != orig->spte_low ||
+	      count != sp->clear_spte_count))
+		goto retry;
+
+	return spte.spte;
+}
+
+static bool __check_direct_spte_mmio_pf(u64 spte)
+{
+	union split_spte sspte = (union split_spte)spte;
+	u32 high_mmio_mask = shadow_mmio_mask >> 32;
+
+	/* It is valid if the spte is zapped. */
+	if (spte == 0ull)
+		return true;
+
+	/* It is valid if the spte is being zapped. */
+	if (sspte.spte_low == 0ull &&
+	    (sspte.spte_high & high_mmio_mask) == high_mmio_mask)
+		return true;
+
+	return false;
+}
+#endif
+
 static bool spte_has_volatile_bits(u64 spte)
 {
 	if (!shadow_accessed_mask)
@@ -322,12 +475,30 @@ static bool spte_is_bit_cleared(u64 old_spte, u64 new_spte, u64 bit_mask)
 	return (old_spte & bit_mask) && !(new_spte & bit_mask);
 }
 
-static void update_spte(u64 *sptep, u64 new_spte)
+/* Rules for using mmu_spte_set:
+ * Set the sptep from nonpresent to present.
+ * Note: the sptep being assigned *must* be either not present
+ * or in a state where the hardware will not attempt to update
+ * the spte.
+ */
+static void mmu_spte_set(u64 *sptep, u64 new_spte)
+{
+	WARN_ON(is_shadow_present_pte(*sptep));
+	__set_spte(sptep, new_spte);
+}
+
+/* Rules for using mmu_spte_update:
+ * Update the state bits, it means the mapped pfn is not changged.
+ */
+static void mmu_spte_update(u64 *sptep, u64 new_spte)
 {
 	u64 mask, old_spte = *sptep;
 
 	WARN_ON(!is_rmap_spte(new_spte));
 
+	if (!is_shadow_present_pte(old_spte))
+		return mmu_spte_set(sptep, new_spte);
+
 	new_spte |= old_spte & shadow_dirty_mask;
 
 	mask = shadow_accessed_mask;
@@ -335,9 +506,9 @@ static void update_spte(u64 *sptep, u64 new_spte)
 		mask |= shadow_dirty_mask;
 
 	if (!spte_has_volatile_bits(old_spte) || (new_spte & mask) == mask)
-		__set_spte(sptep, new_spte);
+		__update_clear_spte_fast(sptep, new_spte);
 	else
-		old_spte = __xchg_spte(sptep, new_spte);
+		old_spte = __update_clear_spte_slow(sptep, new_spte);
 
 	if (!shadow_accessed_mask)
 		return;
@@ -348,6 +519,64 @@ static void update_spte(u64 *sptep, u64 new_spte)
 		kvm_set_pfn_dirty(spte_to_pfn(old_spte));
 }
 
+/*
+ * Rules for using mmu_spte_clear_track_bits:
+ * It sets the sptep from present to nonpresent, and track the
+ * state bits, it is used to clear the last level sptep.
+ */
+static int mmu_spte_clear_track_bits(u64 *sptep)
+{
+	pfn_t pfn;
+	u64 old_spte = *sptep;
+
+	if (!spte_has_volatile_bits(old_spte))
+		__update_clear_spte_fast(sptep, 0ull);
+	else
+		old_spte = __update_clear_spte_slow(sptep, 0ull);
+
+	if (!is_rmap_spte(old_spte))
+		return 0;
+
+	pfn = spte_to_pfn(old_spte);
+	if (!shadow_accessed_mask || old_spte & shadow_accessed_mask)
+		kvm_set_pfn_accessed(pfn);
+	if (!shadow_dirty_mask || (old_spte & shadow_dirty_mask))
+		kvm_set_pfn_dirty(pfn);
+	return 1;
+}
+
+/*
+ * Rules for using mmu_spte_clear_no_track:
+ * Directly clear spte without caring the state bits of sptep,
+ * it is used to set the upper level spte.
+ */
+static void mmu_spte_clear_no_track(u64 *sptep)
+{
+	__update_clear_spte_fast(sptep, 0ull);
+}
+
+static u64 mmu_spte_get_lockless(u64 *sptep)
+{
+	return __get_spte_lockless(sptep);
+}
+
+static void walk_shadow_page_lockless_begin(struct kvm_vcpu *vcpu)
+{
+	rcu_read_lock();
+	atomic_inc(&vcpu->kvm->arch.reader_counter);
+
+	/* Increase the counter before walking shadow page table */
+	smp_mb__after_atomic_inc();
+}
+
+static void walk_shadow_page_lockless_end(struct kvm_vcpu *vcpu)
+{
+	/* Decrease the counter after walking shadow page table finished */
+	smp_mb__before_atomic_dec();
+	atomic_dec(&vcpu->kvm->arch.reader_counter);
+	rcu_read_unlock();
+}
+
 static int mmu_topup_memory_cache(struct kvm_mmu_memory_cache *cache,
 				  struct kmem_cache *base_cache, int min)
 {
@@ -397,12 +626,8 @@ static int mmu_topup_memory_caches(struct kvm_vcpu *vcpu)
 {
 	int r;
 
-	r = mmu_topup_memory_cache(&vcpu->arch.mmu_pte_chain_cache,
-				   pte_chain_cache, 4);
-	if (r)
-		goto out;
-	r = mmu_topup_memory_cache(&vcpu->arch.mmu_rmap_desc_cache,
-				   rmap_desc_cache, 4 + PTE_PREFETCH_NUM);
+	r = mmu_topup_memory_cache(&vcpu->arch.mmu_pte_list_desc_cache,
+				   pte_list_desc_cache, 8 + PTE_PREFETCH_NUM);
 	if (r)
 		goto out;
 	r = mmu_topup_memory_cache_page(&vcpu->arch.mmu_page_cache, 8);
@@ -416,8 +641,8 @@ out:
 
 static void mmu_free_memory_caches(struct kvm_vcpu *vcpu)
 {
-	mmu_free_memory_cache(&vcpu->arch.mmu_pte_chain_cache, pte_chain_cache);
-	mmu_free_memory_cache(&vcpu->arch.mmu_rmap_desc_cache, rmap_desc_cache);
+	mmu_free_memory_cache(&vcpu->arch.mmu_pte_list_desc_cache,
+				pte_list_desc_cache);
 	mmu_free_memory_cache_page(&vcpu->arch.mmu_page_cache);
 	mmu_free_memory_cache(&vcpu->arch.mmu_page_header_cache,
 				mmu_page_header_cache);
@@ -433,26 +658,15 @@ static void *mmu_memory_cache_alloc(struct kvm_mmu_memory_cache *mc,
 	return p;
 }
 
-static struct kvm_pte_chain *mmu_alloc_pte_chain(struct kvm_vcpu *vcpu)
-{
-	return mmu_memory_cache_alloc(&vcpu->arch.mmu_pte_chain_cache,
-				      sizeof(struct kvm_pte_chain));
-}
-
-static void mmu_free_pte_chain(struct kvm_pte_chain *pc)
+static struct pte_list_desc *mmu_alloc_pte_list_desc(struct kvm_vcpu *vcpu)
 {
-	kmem_cache_free(pte_chain_cache, pc);
+	return mmu_memory_cache_alloc(&vcpu->arch.mmu_pte_list_desc_cache,
+				      sizeof(struct pte_list_desc));
 }
 
-static struct kvm_rmap_desc *mmu_alloc_rmap_desc(struct kvm_vcpu *vcpu)
+static void mmu_free_pte_list_desc(struct pte_list_desc *pte_list_desc)
 {
-	return mmu_memory_cache_alloc(&vcpu->arch.mmu_rmap_desc_cache,
-				      sizeof(struct kvm_rmap_desc));
-}
-
-static void mmu_free_rmap_desc(struct kvm_rmap_desc *rd)
-{
-	kmem_cache_free(rmap_desc_cache, rd);
+	kmem_cache_free(pte_list_desc_cache, pte_list_desc);
 }
 
 static gfn_t kvm_mmu_page_get_gfn(struct kvm_mmu_page *sp, int index)
@@ -498,6 +712,7 @@ static void account_shadowed(struct kvm *kvm, gfn_t gfn)
 		linfo = lpage_info_slot(gfn, slot, i);
 		linfo->write_count += 1;
 	}
+	kvm->arch.indirect_shadow_pages++;
 }
 
 static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn)
@@ -513,6 +728,7 @@ static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn)
 		linfo->write_count -= 1;
 		WARN_ON(linfo->write_count < 0);
 	}
+	kvm->arch.indirect_shadow_pages--;
 }
 
 static int has_wrprotected_page(struct kvm *kvm,
@@ -588,67 +804,42 @@ static int mapping_level(struct kvm_vcpu *vcpu, gfn_t large_gfn)
 }
 
 /*
- * Take gfn and return the reverse mapping to it.
- */
-
-static unsigned long *gfn_to_rmap(struct kvm *kvm, gfn_t gfn, int level)
-{
-	struct kvm_memory_slot *slot;
-	struct kvm_lpage_info *linfo;
-
-	slot = gfn_to_memslot(kvm, gfn);
-	if (likely(level == PT_PAGE_TABLE_LEVEL))
-		return &slot->rmap[gfn - slot->base_gfn];
-
-	linfo = lpage_info_slot(gfn, slot, level);
-
-	return &linfo->rmap_pde;
-}
-
-/*
- * Reverse mapping data structures:
+ * Pte mapping structures:
  *
- * If rmapp bit zero is zero, then rmapp point to the shadw page table entry
- * that points to page_address(page).
+ * If pte_list bit zero is zero, then pte_list point to the spte.
  *
- * If rmapp bit zero is one, (then rmap & ~1) points to a struct kvm_rmap_desc
- * containing more mappings.
+ * If pte_list bit zero is one, (then pte_list & ~1) points to a struct
+ * pte_list_desc containing more mappings.
  *
- * Returns the number of rmap entries before the spte was added or zero if
+ * Returns the number of pte entries before the spte was added or zero if
  * the spte was not added.
  *
  */
-static int rmap_add(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn)
+static int pte_list_add(struct kvm_vcpu *vcpu, u64 *spte,
+			unsigned long *pte_list)
 {
-	struct kvm_mmu_page *sp;
-	struct kvm_rmap_desc *desc;
-	unsigned long *rmapp;
+	struct pte_list_desc *desc;
 	int i, count = 0;
 
-	if (!is_rmap_spte(*spte))
-		return count;
-	sp = page_header(__pa(spte));
-	kvm_mmu_page_set_gfn(sp, spte - sp->spt, gfn);
-	rmapp = gfn_to_rmap(vcpu->kvm, gfn, sp->role.level);
-	if (!*rmapp) {
-		rmap_printk("rmap_add: %p %llx 0->1\n", spte, *spte);
-		*rmapp = (unsigned long)spte;
-	} else if (!(*rmapp & 1)) {
-		rmap_printk("rmap_add: %p %llx 1->many\n", spte, *spte);
-		desc = mmu_alloc_rmap_desc(vcpu);
-		desc->sptes[0] = (u64 *)*rmapp;
+	if (!*pte_list) {
+		rmap_printk("pte_list_add: %p %llx 0->1\n", spte, *spte);
+		*pte_list = (unsigned long)spte;
+	} else if (!(*pte_list & 1)) {
+		rmap_printk("pte_list_add: %p %llx 1->many\n", spte, *spte);
+		desc = mmu_alloc_pte_list_desc(vcpu);
+		desc->sptes[0] = (u64 *)*pte_list;
 		desc->sptes[1] = spte;
-		*rmapp = (unsigned long)desc | 1;
+		*pte_list = (unsigned long)desc | 1;
 		++count;
 	} else {
-		rmap_printk("rmap_add: %p %llx many->many\n", spte, *spte);
-		desc = (struct kvm_rmap_desc *)(*rmapp & ~1ul);
-		while (desc->sptes[RMAP_EXT-1] && desc->more) {
+		rmap_printk("pte_list_add: %p %llx many->many\n", spte, *spte);
+		desc = (struct pte_list_desc *)(*pte_list & ~1ul);
+		while (desc->sptes[PTE_LIST_EXT-1] && desc->more) {
 			desc = desc->more;
-			count += RMAP_EXT;
+			count += PTE_LIST_EXT;
 		}
-		if (desc->sptes[RMAP_EXT-1]) {
-			desc->more = mmu_alloc_rmap_desc(vcpu);
+		if (desc->sptes[PTE_LIST_EXT-1]) {
+			desc->more = mmu_alloc_pte_list_desc(vcpu);
 			desc = desc->more;
 		}
 		for (i = 0; desc->sptes[i]; ++i)
@@ -658,59 +849,78 @@ static int rmap_add(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn)
 	return count;
 }
 
-static void rmap_desc_remove_entry(unsigned long *rmapp,
-				   struct kvm_rmap_desc *desc,
-				   int i,
-				   struct kvm_rmap_desc *prev_desc)
+static u64 *pte_list_next(unsigned long *pte_list, u64 *spte)
+{
+	struct pte_list_desc *desc;
+	u64 *prev_spte;
+	int i;
+
+	if (!*pte_list)
+		return NULL;
+	else if (!(*pte_list & 1)) {
+		if (!spte)
+			return (u64 *)*pte_list;
+		return NULL;
+	}
+	desc = (struct pte_list_desc *)(*pte_list & ~1ul);
+	prev_spte = NULL;
+	while (desc) {
+		for (i = 0; i < PTE_LIST_EXT && desc->sptes[i]; ++i) {
+			if (prev_spte == spte)
+				return desc->sptes[i];
+			prev_spte = desc->sptes[i];
+		}
+		desc = desc->more;
+	}
+	return NULL;
+}
+
+static void
+pte_list_desc_remove_entry(unsigned long *pte_list, struct pte_list_desc *desc,
+			   int i, struct pte_list_desc *prev_desc)
 {
 	int j;
 
-	for (j = RMAP_EXT - 1; !desc->sptes[j] && j > i; --j)
+	for (j = PTE_LIST_EXT - 1; !desc->sptes[j] && j > i; --j)
 		;
 	desc->sptes[i] = desc->sptes[j];
 	desc->sptes[j] = NULL;
 	if (j != 0)
 		return;
 	if (!prev_desc && !desc->more)
-		*rmapp = (unsigned long)desc->sptes[0];
+		*pte_list = (unsigned long)desc->sptes[0];
 	else
 		if (prev_desc)
 			prev_desc->more = desc->more;
 		else
-			*rmapp = (unsigned long)desc->more | 1;
-	mmu_free_rmap_desc(desc);
+			*pte_list = (unsigned long)desc->more | 1;
+	mmu_free_pte_list_desc(desc);
 }
 
-static void rmap_remove(struct kvm *kvm, u64 *spte)
+static void pte_list_remove(u64 *spte, unsigned long *pte_list)
 {
-	struct kvm_rmap_desc *desc;
-	struct kvm_rmap_desc *prev_desc;
-	struct kvm_mmu_page *sp;
-	gfn_t gfn;
-	unsigned long *rmapp;
+	struct pte_list_desc *desc;
+	struct pte_list_desc *prev_desc;
 	int i;
 
-	sp = page_header(__pa(spte));
-	gfn = kvm_mmu_page_get_gfn(sp, spte - sp->spt);
-	rmapp = gfn_to_rmap(kvm, gfn, sp->role.level);
-	if (!*rmapp) {
-		printk(KERN_ERR "rmap_remove: %p 0->BUG\n", spte);
+	if (!*pte_list) {
+		printk(KERN_ERR "pte_list_remove: %p 0->BUG\n", spte);
 		BUG();
-	} else if (!(*rmapp & 1)) {
-		rmap_printk("rmap_remove:  %p 1->0\n", spte);
-		if ((u64 *)*rmapp != spte) {
-			printk(KERN_ERR "rmap_remove:  %p 1->BUG\n", spte);
+	} else if (!(*pte_list & 1)) {
+		rmap_printk("pte_list_remove:  %p 1->0\n", spte);
+		if ((u64 *)*pte_list != spte) {
+			printk(KERN_ERR "pte_list_remove:  %p 1->BUG\n", spte);
 			BUG();
 		}
-		*rmapp = 0;
+		*pte_list = 0;
 	} else {
-		rmap_printk("rmap_remove:  %p many->many\n", spte);
-		desc = (struct kvm_rmap_desc *)(*rmapp & ~1ul);
+		rmap_printk("pte_list_remove:  %p many->many\n", spte);
+		desc = (struct pte_list_desc *)(*pte_list & ~1ul);
 		prev_desc = NULL;
 		while (desc) {
-			for (i = 0; i < RMAP_EXT && desc->sptes[i]; ++i)
+			for (i = 0; i < PTE_LIST_EXT && desc->sptes[i]; ++i)
 				if (desc->sptes[i] == spte) {
-					rmap_desc_remove_entry(rmapp,
+					pte_list_desc_remove_entry(pte_list,
 							       desc, i,
 							       prev_desc);
 					return;
@@ -718,62 +928,80 @@ static void rmap_remove(struct kvm *kvm, u64 *spte)
 			prev_desc = desc;
 			desc = desc->more;
 		}
-		pr_err("rmap_remove: %p many->many\n", spte);
+		pr_err("pte_list_remove: %p many->many\n", spte);
 		BUG();
 	}
 }
 
-static int set_spte_track_bits(u64 *sptep, u64 new_spte)
+typedef void (*pte_list_walk_fn) (u64 *spte);
+static void pte_list_walk(unsigned long *pte_list, pte_list_walk_fn fn)
 {
-	pfn_t pfn;
-	u64 old_spte = *sptep;
+	struct pte_list_desc *desc;
+	int i;
 
-	if (!spte_has_volatile_bits(old_spte))
-		__set_spte(sptep, new_spte);
-	else
-		old_spte = __xchg_spte(sptep, new_spte);
+	if (!*pte_list)
+		return;
 
-	if (!is_rmap_spte(old_spte))
-		return 0;
+	if (!(*pte_list & 1))
+		return fn((u64 *)*pte_list);
 
-	pfn = spte_to_pfn(old_spte);
-	if (!shadow_accessed_mask || old_spte & shadow_accessed_mask)
-		kvm_set_pfn_accessed(pfn);
-	if (!shadow_dirty_mask || (old_spte & shadow_dirty_mask))
-		kvm_set_pfn_dirty(pfn);
-	return 1;
+	desc = (struct pte_list_desc *)(*pte_list & ~1ul);
+	while (desc) {
+		for (i = 0; i < PTE_LIST_EXT && desc->sptes[i]; ++i)
+			fn(desc->sptes[i]);
+		desc = desc->more;
+	}
 }
 
-static void drop_spte(struct kvm *kvm, u64 *sptep, u64 new_spte)
+/*
+ * Take gfn and return the reverse mapping to it.
+ */
+static unsigned long *gfn_to_rmap(struct kvm *kvm, gfn_t gfn, int level)
 {
-	if (set_spte_track_bits(sptep, new_spte))
-		rmap_remove(kvm, sptep);
+	struct kvm_memory_slot *slot;
+	struct kvm_lpage_info *linfo;
+
+	slot = gfn_to_memslot(kvm, gfn);
+	if (likely(level == PT_PAGE_TABLE_LEVEL))
+		return &slot->rmap[gfn - slot->base_gfn];
+
+	linfo = lpage_info_slot(gfn, slot, level);
+
+	return &linfo->rmap_pde;
+}
+
+static int rmap_add(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn)
+{
+	struct kvm_mmu_page *sp;
+	unsigned long *rmapp;
+
+	sp = page_header(__pa(spte));
+	kvm_mmu_page_set_gfn(sp, spte - sp->spt, gfn);
+	rmapp = gfn_to_rmap(vcpu->kvm, gfn, sp->role.level);
+	return pte_list_add(vcpu, spte, rmapp);
 }
 
 static u64 *rmap_next(struct kvm *kvm, unsigned long *rmapp, u64 *spte)
 {
-	struct kvm_rmap_desc *desc;
-	u64 *prev_spte;
-	int i;
+	return pte_list_next(rmapp, spte);
+}
 
-	if (!*rmapp)
-		return NULL;
-	else if (!(*rmapp & 1)) {
-		if (!spte)
-			return (u64 *)*rmapp;
-		return NULL;
-	}
-	desc = (struct kvm_rmap_desc *)(*rmapp & ~1ul);
-	prev_spte = NULL;
-	while (desc) {
-		for (i = 0; i < RMAP_EXT && desc->sptes[i]; ++i) {
-			if (prev_spte == spte)
-				return desc->sptes[i];
-			prev_spte = desc->sptes[i];
-		}
-		desc = desc->more;
-	}
-	return NULL;
+static void rmap_remove(struct kvm *kvm, u64 *spte)
+{
+	struct kvm_mmu_page *sp;
+	gfn_t gfn;
+	unsigned long *rmapp;
+
+	sp = page_header(__pa(spte));
+	gfn = kvm_mmu_page_get_gfn(sp, spte - sp->spt);
+	rmapp = gfn_to_rmap(kvm, gfn, sp->role.level);
+	pte_list_remove(spte, rmapp);
+}
+
+static void drop_spte(struct kvm *kvm, u64 *sptep)
+{
+	if (mmu_spte_clear_track_bits(sptep))
+		rmap_remove(kvm, sptep);
 }
 
 static int rmap_write_protect(struct kvm *kvm, u64 gfn)
@@ -790,7 +1018,7 @@ static int rmap_write_protect(struct kvm *kvm, u64 gfn)
 		BUG_ON(!(*spte & PT_PRESENT_MASK));
 		rmap_printk("rmap_write_protect: spte %p %llx\n", spte, *spte);
 		if (is_writable_pte(*spte)) {
-			update_spte(spte, *spte & ~PT_WRITABLE_MASK);
+			mmu_spte_update(spte, *spte & ~PT_WRITABLE_MASK);
 			write_protected = 1;
 		}
 		spte = rmap_next(kvm, rmapp, spte);
@@ -807,8 +1035,7 @@ static int rmap_write_protect(struct kvm *kvm, u64 gfn)
 			BUG_ON((*spte & (PT_PAGE_SIZE_MASK|PT_PRESENT_MASK)) != (PT_PAGE_SIZE_MASK|PT_PRESENT_MASK));
 			pgprintk("rmap_write_protect(large): spte %p %llx %lld\n", spte, *spte, gfn);
 			if (is_writable_pte(*spte)) {
-				drop_spte(kvm, spte,
-					  shadow_trap_nonpresent_pte);
+				drop_spte(kvm, spte);
 				--kvm->stat.lpages;
 				spte = NULL;
 				write_protected = 1;
@@ -829,7 +1056,7 @@ static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp,
 	while ((spte = rmap_next(kvm, rmapp, NULL))) {
 		BUG_ON(!(*spte & PT_PRESENT_MASK));
 		rmap_printk("kvm_rmap_unmap_hva: spte %p %llx\n", spte, *spte);
-		drop_spte(kvm, spte, shadow_trap_nonpresent_pte);
+		drop_spte(kvm, spte);
 		need_tlb_flush = 1;
 	}
 	return need_tlb_flush;
@@ -851,7 +1078,7 @@ static int kvm_set_pte_rmapp(struct kvm *kvm, unsigned long *rmapp,
 		rmap_printk("kvm_set_pte_rmapp: spte %p %llx\n", spte, *spte);
 		need_flush = 1;
 		if (pte_write(*ptep)) {
-			drop_spte(kvm, spte, shadow_trap_nonpresent_pte);
+			drop_spte(kvm, spte);
 			spte = rmap_next(kvm, rmapp, NULL);
 		} else {
 			new_spte = *spte &~ (PT64_BASE_ADDR_MASK);
@@ -860,7 +1087,8 @@ static int kvm_set_pte_rmapp(struct kvm *kvm, unsigned long *rmapp,
 			new_spte &= ~PT_WRITABLE_MASK;
 			new_spte &= ~SPTE_HOST_WRITEABLE;
 			new_spte &= ~shadow_accessed_mask;
-			set_spte_track_bits(spte, new_spte);
+			mmu_spte_clear_track_bits(spte);
+			mmu_spte_set(spte, new_spte);
 			spte = rmap_next(kvm, rmapp, spte);
 		}
 	}
@@ -1032,151 +1260,89 @@ static inline void kvm_mod_used_mmu_pages(struct kvm *kvm, int nr)
 	percpu_counter_add(&kvm_total_used_mmu_pages, nr);
 }
 
-static void kvm_mmu_free_page(struct kvm *kvm, struct kvm_mmu_page *sp)
+/*
+ * Remove the sp from shadow page cache, after call it,
+ * we can not find this sp from the cache, and the shadow
+ * page table is still valid.
+ * It should be under the protection of mmu lock.
+ */
+static void kvm_mmu_isolate_page(struct kvm_mmu_page *sp)
 {
 	ASSERT(is_empty_shadow_page(sp->spt));
 	hlist_del(&sp->hash_link);
-	list_del(&sp->link);
-	free_page((unsigned long)sp->spt);
 	if (!sp->role.direct)
 		free_page((unsigned long)sp->gfns);
-	kmem_cache_free(mmu_page_header_cache, sp);
-	kvm_mod_used_mmu_pages(kvm, -1);
 }
 
-static unsigned kvm_page_table_hashfn(gfn_t gfn)
+/*
+ * Free the shadow page table and the sp, we can do it
+ * out of the protection of mmu lock.
+ */
+static void kvm_mmu_free_page(struct kvm_mmu_page *sp)
 {
-	return gfn & ((1 << KVM_MMU_HASH_SHIFT) - 1);
+	list_del(&sp->link);
+	free_page((unsigned long)sp->spt);
+	kmem_cache_free(mmu_page_header_cache, sp);
 }
 
-static struct kvm_mmu_page *kvm_mmu_alloc_page(struct kvm_vcpu *vcpu,
-					       u64 *parent_pte, int direct)
+static unsigned kvm_page_table_hashfn(gfn_t gfn)
 {
-	struct kvm_mmu_page *sp;
-
-	sp = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_header_cache, sizeof *sp);
-	sp->spt = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache, PAGE_SIZE);
-	if (!direct)
-		sp->gfns = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache,
-						  PAGE_SIZE);
-	set_page_private(virt_to_page(sp->spt), (unsigned long)sp);
-	list_add(&sp->link, &vcpu->kvm->arch.active_mmu_pages);
-	bitmap_zero(sp->slot_bitmap, KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS);
-	sp->multimapped = 0;
-	sp->parent_pte = parent_pte;
-	kvm_mod_used_mmu_pages(vcpu->kvm, +1);
-	return sp;
+	return gfn & ((1 << KVM_MMU_HASH_SHIFT) - 1);
 }
 
 static void mmu_page_add_parent_pte(struct kvm_vcpu *vcpu,
 				    struct kvm_mmu_page *sp, u64 *parent_pte)
 {
-	struct kvm_pte_chain *pte_chain;
-	struct hlist_node *node;
-	int i;
-
 	if (!parent_pte)
 		return;
-	if (!sp->multimapped) {
-		u64 *old = sp->parent_pte;
 
-		if (!old) {
-			sp->parent_pte = parent_pte;
-			return;
-		}
-		sp->multimapped = 1;
-		pte_chain = mmu_alloc_pte_chain(vcpu);
-		INIT_HLIST_HEAD(&sp->parent_ptes);
-		hlist_add_head(&pte_chain->link, &sp->parent_ptes);
-		pte_chain->parent_ptes[0] = old;
-	}
-	hlist_for_each_entry(pte_chain, node, &sp->parent_ptes, link) {
-		if (pte_chain->parent_ptes[NR_PTE_CHAIN_ENTRIES-1])
-			continue;
-		for (i = 0; i < NR_PTE_CHAIN_ENTRIES; ++i)
-			if (!pte_chain->parent_ptes[i]) {
-				pte_chain->parent_ptes[i] = parent_pte;
-				return;
-			}
-	}
-	pte_chain = mmu_alloc_pte_chain(vcpu);
-	BUG_ON(!pte_chain);
-	hlist_add_head(&pte_chain->link, &sp->parent_ptes);
-	pte_chain->parent_ptes[0] = parent_pte;
+	pte_list_add(vcpu, parent_pte, &sp->parent_ptes);
 }
 
 static void mmu_page_remove_parent_pte(struct kvm_mmu_page *sp,
 				       u64 *parent_pte)
 {
-	struct kvm_pte_chain *pte_chain;
-	struct hlist_node *node;
-	int i;
-
-	if (!sp->multimapped) {
-		BUG_ON(sp->parent_pte != parent_pte);
-		sp->parent_pte = NULL;
-		return;
-	}
-	hlist_for_each_entry(pte_chain, node, &sp->parent_ptes, link)
-		for (i = 0; i < NR_PTE_CHAIN_ENTRIES; ++i) {
-			if (!pte_chain->parent_ptes[i])
-				break;
-			if (pte_chain->parent_ptes[i] != parent_pte)
-				continue;
-			while (i + 1 < NR_PTE_CHAIN_ENTRIES
-				&& pte_chain->parent_ptes[i + 1]) {
-				pte_chain->parent_ptes[i]
-					= pte_chain->parent_ptes[i + 1];
-				++i;
-			}
-			pte_chain->parent_ptes[i] = NULL;
-			if (i == 0) {
-				hlist_del(&pte_chain->link);
-				mmu_free_pte_chain(pte_chain);
-				if (hlist_empty(&sp->parent_ptes)) {
-					sp->multimapped = 0;
-					sp->parent_pte = NULL;
-				}
-			}
-			return;
-		}
-	BUG();
+	pte_list_remove(parent_pte, &sp->parent_ptes);
 }
 
-static void mmu_parent_walk(struct kvm_mmu_page *sp, mmu_parent_walk_fn fn)
+static void drop_parent_pte(struct kvm_mmu_page *sp,
+			    u64 *parent_pte)
 {
-	struct kvm_pte_chain *pte_chain;
-	struct hlist_node *node;
-	struct kvm_mmu_page *parent_sp;
-	int i;
-
-	if (!sp->multimapped && sp->parent_pte) {
-		parent_sp = page_header(__pa(sp->parent_pte));
-		fn(parent_sp, sp->parent_pte);
-		return;
-	}
-
-	hlist_for_each_entry(pte_chain, node, &sp->parent_ptes, link)
-		for (i = 0; i < NR_PTE_CHAIN_ENTRIES; ++i) {
-			u64 *spte = pte_chain->parent_ptes[i];
+	mmu_page_remove_parent_pte(sp, parent_pte);
+	mmu_spte_clear_no_track(parent_pte);
+}
 
-			if (!spte)
-				break;
-			parent_sp = page_header(__pa(spte));
-			fn(parent_sp, spte);
-		}
+static struct kvm_mmu_page *kvm_mmu_alloc_page(struct kvm_vcpu *vcpu,
+					       u64 *parent_pte, int direct)
+{
+	struct kvm_mmu_page *sp;
+	sp = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_header_cache,
+					sizeof *sp);
+	sp->spt = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache, PAGE_SIZE);
+	if (!direct)
+		sp->gfns = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache,
+						  PAGE_SIZE);
+	set_page_private(virt_to_page(sp->spt), (unsigned long)sp);
+	list_add(&sp->link, &vcpu->kvm->arch.active_mmu_pages);
+	bitmap_zero(sp->slot_bitmap, KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS);
+	sp->parent_ptes = 0;
+	mmu_page_add_parent_pte(vcpu, sp, parent_pte);
+	kvm_mod_used_mmu_pages(vcpu->kvm, +1);
+	return sp;
 }
 
-static void mark_unsync(struct kvm_mmu_page *sp, u64 *spte);
+static void mark_unsync(u64 *spte);
 static void kvm_mmu_mark_parents_unsync(struct kvm_mmu_page *sp)
 {
-	mmu_parent_walk(sp, mark_unsync);
+	pte_list_walk(&sp->parent_ptes, mark_unsync);
 }
 
-static void mark_unsync(struct kvm_mmu_page *sp, u64 *spte)
+static void mark_unsync(u64 *spte)
 {
+	struct kvm_mmu_page *sp;
 	unsigned int index;
 
+	sp = page_header(__pa(spte));
 	index = spte - sp->spt;
 	if (__test_and_set_bit(index, sp->unsync_child_bitmap))
 		return;
@@ -1185,15 +1351,6 @@ static void mark_unsync(struct kvm_mmu_page *sp, u64 *spte)
 	kvm_mmu_mark_parents_unsync(sp);
 }
 
-static void nonpaging_prefetch_page(struct kvm_vcpu *vcpu,
-				    struct kvm_mmu_page *sp)
-{
-	int i;
-
-	for (i = 0; i < PT64_ENT_PER_PAGE; ++i)
-		sp->spt[i] = shadow_trap_nonpresent_pte;
-}
-
 static int nonpaging_sync_page(struct kvm_vcpu *vcpu,
 			       struct kvm_mmu_page *sp)
 {
@@ -1475,6 +1632,14 @@ static void mmu_sync_children(struct kvm_vcpu *vcpu,
 	}
 }
 
+static void init_shadow_page_table(struct kvm_mmu_page *sp)
+{
+	int i;
+
+	for (i = 0; i < PT64_ENT_PER_PAGE; ++i)
+		sp->spt[i] = 0ull;
+}
+
 static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
 					     gfn_t gfn,
 					     gva_t gaddr,
@@ -1537,10 +1702,7 @@ static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
 
 		account_shadowed(vcpu->kvm, gfn);
 	}
-	if (shadow_trap_nonpresent_pte != shadow_notrap_nonpresent_pte)
-		vcpu->arch.mmu.prefetch_page(vcpu, sp);
-	else
-		nonpaging_prefetch_page(vcpu, sp);
+	init_shadow_page_table(sp);
 	trace_kvm_mmu_get_page(sp, true);
 	return sp;
 }
@@ -1572,21 +1734,28 @@ static bool shadow_walk_okay(struct kvm_shadow_walk_iterator *iterator)
 	if (iterator->level < PT_PAGE_TABLE_LEVEL)
 		return false;
 
-	if (iterator->level == PT_PAGE_TABLE_LEVEL)
-		if (is_large_pte(*iterator->sptep))
-			return false;
-
 	iterator->index = SHADOW_PT_INDEX(iterator->addr, iterator->level);
 	iterator->sptep	= ((u64 *)__va(iterator->shadow_addr)) + iterator->index;
 	return true;
 }
 
-static void shadow_walk_next(struct kvm_shadow_walk_iterator *iterator)
+static void __shadow_walk_next(struct kvm_shadow_walk_iterator *iterator,
+			       u64 spte)
 {
-	iterator->shadow_addr = *iterator->sptep & PT64_BASE_ADDR_MASK;
+	if (is_last_spte(spte, iterator->level)) {
+		iterator->level = 0;
+		return;
+	}
+
+	iterator->shadow_addr = spte & PT64_BASE_ADDR_MASK;
 	--iterator->level;
 }
 
+static void shadow_walk_next(struct kvm_shadow_walk_iterator *iterator)
+{
+	return __shadow_walk_next(iterator, *iterator->sptep);
+}
+
 static void link_shadow_page(u64 *sptep, struct kvm_mmu_page *sp)
 {
 	u64 spte;
@@ -1594,13 +1763,13 @@ static void link_shadow_page(u64 *sptep, struct kvm_mmu_page *sp)
 	spte = __pa(sp->spt)
 		| PT_PRESENT_MASK | PT_ACCESSED_MASK
 		| PT_WRITABLE_MASK | PT_USER_MASK;
-	__set_spte(sptep, spte);
+	mmu_spte_set(sptep, spte);
 }
 
 static void drop_large_spte(struct kvm_vcpu *vcpu, u64 *sptep)
 {
 	if (is_large_pte(*sptep)) {
-		drop_spte(vcpu->kvm, sptep, shadow_trap_nonpresent_pte);
+		drop_spte(vcpu->kvm, sptep);
 		kvm_flush_remote_tlbs(vcpu->kvm);
 	}
 }
@@ -1622,38 +1791,39 @@ static void validate_direct_spte(struct kvm_v