> On Mar 7, 2016, at 11:18 AM, Ben Pfaff <b...@ovn.org> wrote:
>
> +/* Modifies 'match' so that the MPLS TTL is wildcarded. */
> +void
> +match_set_any_mpls_ttl(struct match *match, int idx)
> +{
> + match->wc.masks.mpls_lse[idx] &= ~htonl(MPLS_TTL_MASK);
> + flow_set_mpls_ttl(&match->flow, idx, 0);
> +}
> +
> +/* Modifies 'match' so that it matches only packets with an MPLS header whose
> + * TTL equals 'mpls_ttl' */
> +void
> +match_set_mpls_ttl(struct match *match, int idx, uint8_t mpls_ttl)
> +{
> + match->wc.masks.mpls_lse[idx] |= htonl(MPLS_TTL_MASK);
> + flow_set_mpls_ttl(&match->flow, idx, mpls_ttl);
> +}
Do you think it's worth documenting the "idx" arguments for these two functions?
> + /* "mpls_ttl".
> + *
> + * The outermost MPLS label's time-to-level (TTL) field, or 0 if no MPLS
Shouldn't that be "time-to-live"? Or is that how MPLS indicates that it wants
to have a serious talk.
Acked-by: Justin Pettit <jpet...@ovn.org>
--Justin
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
