On 7/13/26 6:06 AM, Naveen Yerramneni wrote:
> - Skip VLAN tagging for traffic mirrored to vtap NF ports
> - Only perform overlay redirection for traffic coming from inline NF
> ports. This is not required for vtap ports since traffic is consumed
> at the NF.
>
> Signed-off-by: Naveen Yerramneni <[email protected]>
> Acked-by: Sragdhara Datta Chaudhuri <[email protected]>
> Acked-by: Aditya Mehakare <[email protected]>
> Assisted-by: Claude Opus 4.7, Cursor
> ---
Hi Naveen,
Thanks for the new version!
> controller/physical.c | 35 +++++++++++++++++++++++++++--------
> 1 file changed, 27 insertions(+), 8 deletions(-)
>
> diff --git a/controller/physical.c b/controller/physical.c
> index 89ff2785e..26939b728 100644
> --- a/controller/physical.c
> +++ b/controller/physical.c
> @@ -2182,6 +2182,23 @@ enforce_tunneling_for_multichassis_ports(
> vector_destroy(&tuns);
> }
>
> +/* Check if port is a vtap network function port.
> + * Vtap ports have is-nf option set to true but no nf-linked-port option.
> + */
> +static bool
> +is_nf_vtap_port(const struct sbrec_port_binding *binding)
> +{
> + bool is_nf = smap_get_bool(&binding->options, "is-nf", false);
> + return is_nf && !smap_get(&binding->options, "nf-linked-port");
> +}
> +
> +static bool
> +is_nf_inline_port(const struct sbrec_port_binding *binding)
> +{
> + bool is_nf = smap_get_bool(&binding->options, "is-nf", false);
> + return is_nf && smap_get(&binding->options, "nf-linked-port");
> +}
> +
> static void
> consider_port_binding(const struct physical_ctx *ctx,
> const struct sbrec_port_binding *binding,
> @@ -2515,19 +2532,22 @@ consider_port_binding(const struct physical_ctx *ctx,
> * Deliver the packet to the local vif. */
> ofpbuf_clear(ofpacts_p);
> match_outport_dp_and_port_keys(&match, dp_key, port_key);
> - if (tag) {
> + bool should_skip_vlan = is_nf_vtap_port(binding);
> + if (tag && !should_skip_vlan) {
> /* For containers sitting behind a local vif, tag the packets
> - * before delivering them. */
> + * before delivering them. Skip VLAN tagging for vtap network
> + * function ports. */
> ofpact_put_push_vlan(
> ofpacts_p, localnet_port ? &localnet_port->options : NULL,
> tag);
> }
> ofpact_put_OUTPUT(ofpacts_p)->port = ofport;
> - if (tag) {
> + if (tag && !should_skip_vlan) {
> /* Revert the tag added to the packets headed to containers
> * in the previous step. If we don't do this, the packets
> * that are to be broadcasted to a VM in the same logical
> - * switch will also contain the tag. */
> + * switch will also contain the tag. Skip VLAN stripping for vtap
> + * network function ports. */
> ofpact_put_STRIP_VLAN(ofpacts_p);
> }
> ofctrl_add_flow(flow_table, OFTABLE_LOG_TO_PHY, 100,
> @@ -2630,10 +2650,9 @@ consider_port_binding(const struct physical_ctx *ctx,
> }
>
> /* Packets egressing from network function ports need to be sent to
> the
> - * source. */
> - if (is_nf && localnet_port) {
> - put_redirect_overlay_to_source_from_nf_port(
> - binding,
> + * source */
Nit: missing period at end of sentence.
I took care of this and applied the patch to main.
Regards,
Dumitru
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
