KVM: MMU: Update shadow ptes on partial guest pte writes

A guest partial guest pte write will leave shadow_trap_nonpresent_pte
in spte, which generates a vmexit at the next guest access through that pte.

This patch improves this by reading the full guest pte in advance and thus
being able to update the spte and eliminate the vmexit.

This helps pae guests which use two 32-bit writes to set a single 64-bit pte.

[truncation fix by Eric]

Signed-off-by: Yaozu (Eddie) Dong <eddie.dong@intel.com>
Signed-off-by: Feng (Eric) Liu <eric.e.liu@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

2 files changed, 18 insertions, 12 deletions

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index e55af12e11b7..28f9a44060cc 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -1329,8 +1329,7 @@ static void mmu_pte_write_zap_pte(struct kvm_vcpu *vcpu,
 static void mmu_pte_write_new_pte(struct kvm_vcpu *vcpu,
 				  struct kvm_mmu_page *sp,
 				  u64 *spte,
-				  const void *new, int bytes,
-				  int offset_in_pte)
+				  const void *new)
 {
 	if (sp->role.level != PT_PAGE_TABLE_LEVEL) {
 		++vcpu->kvm->stat.mmu_pde_zapped;
@@ -1339,9 +1338,9 @@ static void mmu_pte_write_new_pte(struct kvm_vcpu *vcpu,
 
 	++vcpu->kvm->stat.mmu_pte_updated;
 	if (sp->role.glevels == PT32_ROOT_LEVEL)
-		paging32_update_pte(vcpu, sp, spte, new, bytes, offset_in_pte);
+		paging32_update_pte(vcpu, sp, spte, new);
 	else
-		paging64_update_pte(vcpu, sp, spte, new, bytes, offset_in_pte);
+		paging64_update_pte(vcpu, sp, spte, new);
 }
 
 static bool need_remote_flush(u64 old, u64 new)
@@ -1423,7 +1422,7 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 	struct hlist_node *node, *n;
 	struct hlist_head *bucket;
 	unsigned index;
-	u64 entry;
+	u64 entry, gentry;
 	u64 *spte;
 	unsigned offset = offset_in_page(gpa);
 	unsigned pte_size;
@@ -1433,6 +1432,7 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 	int level;
 	int flooded = 0;
 	int npte;
+	int r;
 
 	pgprintk("%s: gpa %llx bytes %d\n", __FUNCTION__, gpa, bytes);
 	mmu_guess_page_from_pte_write(vcpu, gpa, new, bytes);
@@ -1496,11 +1496,20 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 				continue;
 		}
 		spte = &sp->spt[page_offset / sizeof(*spte)];
+		if ((gpa & (pte_size - 1)) || (bytes < pte_size)) {
+			gentry = 0;
+			r = kvm_read_guest_atomic(vcpu->kvm,
+						  gpa & ~(u64)(pte_size - 1),
+						  &gentry, pte_size);
+			new = (const void *)&gentry;
+			if (r < 0)
+				new = NULL;
+		}
 		while (npte--) {
 			entry = *spte;
 			mmu_pte_write_zap_pte(vcpu, sp, spte);
-			mmu_pte_write_new_pte(vcpu, sp, spte, new, bytes,
-					      page_offset & (pte_size - 1));
+			if (new)
+				mmu_pte_write_new_pte(vcpu, sp, spte, new);
 			mmu_pte_write_flush_tlb(vcpu, entry, *spte);
 			++spte;
 		}
diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h
index ecc0856268c4..c2fd2b96144f 100644
--- a/arch/x86/kvm/paging_tmpl.h
+++ b/arch/x86/kvm/paging_tmpl.h
@@ -243,8 +243,7 @@ err:
 }
 
 static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page,
-			      u64 *spte, const void *pte, int bytes,
-			      int offset_in_pte)
+			      u64 *spte, const void *pte)
 {
 	pt_element_t gpte;
 	unsigned pte_access;
@@ -252,12 +251,10 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page,
 
 	gpte = *(const pt_element_t *)pte;
 	if (~gpte & (PT_PRESENT_MASK | PT_ACCESSED_MASK)) {
-		if (!offset_in_pte && !is_present_pte(gpte))
+		if (!is_present_pte(gpte))
 			set_shadow_pte(spte, shadow_notrap_nonpresent_pte);
 		return;
 	}
-	if (bytes < sizeof(pt_element_t))
-		return;
 	pgprintk("%s: gpte %llx spte %p\n", __FUNCTION__, (u64)gpte, spte);
 	pte_access = page->role.access & FNAME(gpte_access)(vcpu, gpte);
 	if (gpte_to_gfn(gpte) != vcpu->arch.update_pte.gfn)