Hi James,

On 9/7/21 4:14 PM, James Prestwood wrote:
> There were two places where SAE returned a fatal error code rather
> than silently drop the frame as the spec requires. These two cases
> return -EAGAIN which ultimately drops the frame in netdev.
> ---
>   src/sae.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/src/sae.c b/src/sae.c
> index 62fd6c88..fbd0298d 100644
> --- a/src/sae.c
> +++ b/src/sae.c
> @@ -678,7 +678,7 @@ static int sae_process_commit(struct sae_sm *sm, const uint8_t *from,
>   	if (l_ecc_scalars_are_equal(sm->p_scalar, sm->scalar) ||
>   			l_ecc_points_are_equal(sm->p_element, sm->element)) {
>   		l_warn("peer scalar or element matched own, discarding frame");
> -		return -ENOMSG;
> +		return -EAGAIN;

Hmm, this one seems like it should remain unchanged?

>   	}
>   
>   	sm->sc++;
> @@ -1185,7 +1185,7 @@ static int sae_verify_accepted(struct sae_sm *sm, uint16_t trans,
>   	/* spec does not specify what to do here, so print and discard */
>   	if (trans != SAE_STATE_CONFIRMED) {
>   		l_error("received transaction %u in accepted state", trans);
> -		return -EBADMSG;
> +		return -EAGAIN;

I guess you probably just want 'return 0' here and a quote from 12.4.8.6.1?

>   	}
>   
>   	if (sm->sync > SAE_SYNC_MAX)
> 

Regards,
-Denis