KVM: nVMX: update last_nonleaf_level when initializing nested EPT
authorLadi Prosek <lprosek@redhat.com>
Thu, 5 Oct 2017 09:10:22 +0000 (11:10 +0200)
committerPaolo Bonzini <pbonzini@redhat.com>
Tue, 10 Oct 2017 13:31:18 +0000 (15:31 +0200)
The function updates context->root_level but didn't call
update_last_nonleaf_level so the previous and potentially wrong value
was used for page walks.  For example, a zero value of last_nonleaf_level
would allow a potential out-of-bounds access in arch/x86/mmu/paging_tmpl.h's
walk_addr_generic function (CVE-2017-12188).

Fixes: 155a97a3d7c78b46cef6f1a973c831bc5a4f82bb
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/mmu.c

index 106d4a029a8a933ad3865a11a5e2c1a11810e714..3c25f20115bc79b65545567d63c8e6a4129e5119 100644 (file)
@@ -4555,6 +4555,7 @@ void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly,
 
        update_permission_bitmask(vcpu, context, true);
        update_pkru_bitmask(vcpu, context, true);
+       update_last_nonleaf_level(vcpu, context);
        reset_rsvds_bits_mask_ept(vcpu, context, execonly);
        reset_ept_shadow_zero_bits_mask(vcpu, context, execonly);
 }