On 15.03.2017 15:40, Jim Mattson wrote:
> It is never correct for a VMX instruction to fail with "invalid VMCS"
> if there is, in fact, a current VMCS. Reads from unbacked addresses
> return all 1's, which means that an unbacked VMCS will not have the
> correct VMCS revision ID (i.e. VMCS12_REVISION).
>
> Fixes: 63846663eac78 ("KVM: nVMX: Implement VMPTRLD")
> Signed-off-by: Jim Mattson <jmattson@xxxxxxxxxx>
> ---
> arch/x86/kvm/vmx.c | 9 ++++++++-
> 1 file changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 98e82ee1e699..2c21181c8a70 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -7604,7 +7604,14 @@ static int handle_vmptrld(struct kvm_vcpu *vcpu)
> struct page *page;
> page = nested_get_page(vcpu, vmptr);
> if (page == NULL) {
> - nested_vmx_failInvalid(vcpu);
> + /*
> + * Reads from an unbacked page return all 1s,
> + * which means that the 32 bits located at the
> + * given physical address won't match the required
> + * VMCS12_REVISION identifier.
> + */
> + nested_vmx_failValid(vcpu,
> + VMXERR_VMPTRLD_INCORRECT_VMCS_REVISION_ID);
> return kvm_skip_emulated_instruction(vcpu);
> }
> new_vmcs12 = kmap(page);
>
Reviewed-by: David Hildenbrand <david@xxxxxxxxxx>
--
Thanks,
David
