----- rkrcmar@xxxxxxxxxx wrote:
> 2018-05-22 17:16+0300, Liran Alon:
> > Intel SDM specifies that INVVPID type 0 (individual-address
> > invalidation) should fail in case VPID specified in INVVPID_DESC
> > passed as operand is 0.
> >
> > Reported-by: Eyal Moscovici <eyal.moscovici@xxxxxxxxxx>
> > Reviewed-by: Darren Kenny <darren.kenny@xxxxxxxxxx>
> > Signed-off-by: Liran Alon <liran.alon@xxxxxxxxxx>
> > ---
> > arch/x86/kvm/vmx.c | 3 ++-
> > 1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> > index ebbed33b748f..33fb512539a3 100644
> > --- a/arch/x86/kvm/vmx.c
> > +++ b/arch/x86/kvm/vmx.c
> > @@ -8320,7 +8320,8 @@ static int handle_invvpid(struct kvm_vcpu
> *vcpu)
> >
> > switch (type) {
> > case VMX_VPID_EXTENT_INDIVIDUAL_ADDR:
> > - if (is_noncanonical_address(operand.gla, vcpu)) {
> > + if (!operand.vpid ||
>
> We did case fall-though to handle this, so this patch just makes the
> check twice and I have squashed it with [4/4] as the result seemed
> more
> logical that way.
>
> Please see kvm/queue, thanks.
Oops, missed that. Checked kvm/queue and it seems fine.
Thanks.
