On Wed, Jan 2, 2013 at 5:00 PM, Ethan Jackson <et...@nicira.com> wrote:
> Sounds good. I've applied the following minor changes to the patch which I
> think is good enough for now. Let me know if there's anthing else that needs
> to change before we do the appropriate refactoring.
There's one other thing that I had mentioned before that probably got
lost in the noise of the other conversation:
On Wed, Dec 26, 2012 at 8:16 PM, Ethan Jackson <et...@nicira.com> wrote:
> +ofproto_receive(const struct dpif_backer *backer, struct ofpbuf *packet,
> + const struct nlattr *key, size_t key_len,
> + struct flow *flow, struct ofproto_dpif **ofproto,
> + uint32_t *odp_in_port, ovs_be16 *initial_tci)
> +
[...]
> + port = odp_port_to_ofport(backer, flow->in_port);
> + if (!port) {
> + VLOG_INFO_RL(&rl, "received packet on an unassociated port %"PRIu32,
> + flow->in_port);
I think we only want to log this in the miss upcall case. In the
sFlow and trace cases it's possible for it to happen in normal
situations.
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
