Re: [PATCH 3/5] KVM: nVMX: add limit check for expand-down segments

[Paolo Bonzini <pbonzini@redhat.com>]

On 20/08/2015 21:37, Eugene Korenevsky wrote:
> Add limit checking for expand-down data segments. For such segments, the
> effective limit specifies the last address that is not allowed to be accessed
> within the segment. I.e. offset <= limit means means limit exceeding.
> 
> Signed-off-by: Eugene Korenevsky <ekorenevsky@gmail.com>
> ---
>  arch/x86/kvm/vmx.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index faa05a4..4a4d677 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -6377,7 +6377,10 @@ static int vmx_protmode_seg_check(struct kvm_vcpu *vcpu,
>  	/* #GP(0)/#SS(0) if the segment is unusable. */
>  	exn = (s->unusable != 0);
>  	/* #GP(0)/#SS(0) if the memory operand is outside the segment limit. */
> -	exn = exn || (off + mem_op_size - 1 > s->limit);
> +	if (!(s->type & 8) && (s->type & 4)) /* expand-down segment? */
> +		exn = exn || (off <= s->limit);
> +	else
> +		exn = exn || (off + mem_op_size - 1 > s->limit);

For expand-down segments you also have to check against the size of the
segment (48 bits for 64-bit segments and positive addresses, 32 bits for
32-bit segments, 16 bits for 16-bit segments; see __linearize in emulate.c).

Paolo