KVM: MMU: Use different shadows when EFER.NXE changes

A pte that is shadowed when the guest EFER.NXE=1 is not valid when EFER.NXE=0; if bit 63 is set, the pte should cause a fault, and since the shadow EFER always has NX enabled, this won't happen. Fix by using a different shadow page table for different EFER.NXE bits. This allows vcpus to run correctly with different values of EFER.NXE, and for transitions on this bit to be handled correctly without requiring a full flush. Signed-off-by: Avi Kivity <avi@redhat.com>
git-mirror · Jun 10, 2009 · 9645bb5 · 9645bb5
1 parent 82725b2
commit 9645bb5
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 0 deletions.
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
@@ -185,6 +185,7 @@ union kvm_mmu_page_role {
 		unsigned access:3;
 		unsigned invalid:1;
 		unsigned cr4_pge:1;
+		unsigned nxe:1;
 	};
 };
 

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
@@ -523,6 +523,9 @@ static void set_efer(struct kvm_vcpu *vcpu, u64 efer)
 	efer |= vcpu->arch.shadow_efer & EFER_LMA;
 
 	vcpu->arch.shadow_efer = efer;
+
+	vcpu->arch.mmu.base_role.nxe = (efer & EFER_NX) && !tdp_enabled;
+	kvm_mmu_reset_context(vcpu);
 }
 
 void kvm_enable_efer_bits(u64 mask)