>>
>>
>> + /* Metadata. */
>> if (md) {
>> - flow->tunnel = md->tunnel;
>> - flow->in_port = md->in_port;
>> - flow->skb_priority = md->skb_priority;
>> - flow->pkt_mark = md->pkt_mark;
>> - flow->recirc_id = md->recirc_id;
>> - flow->dp_hash = md->dp_hash;
>> + if (md->tunnel.ip_dst) {
>> + miniflow_push_words(mf, tunnel, &md->tunnel,
>> + sizeof md->tunnel / 4);
>> + }
>> + miniflow_push_uint32_check(mf, skb_priority, md->skb_priority);
>> + miniflow_push_uint32_check(mf, pkt_mark, md->pkt_mark);
>> + miniflow_push_uint32_check(mf, recirc_id, md->recirc_id);
>> + miniflow_push_uint32(mf, in_port, odp_to_u32(md->in_port.odp_port));
>> }
>
> ...
>
>> }
>> }
>> + if (md) {
>> + miniflow_push_uint32_check(mf, dp_hash, md->dp_hash);
>> + }
>> + out:
>> + dst->map = mf.map;
>> }
>
> Was there a particular reason for shifting the dp_hash part to the end of the
> function here, rather than having it grouped with the other metadata fields?
The only reason is that we push the miniflow data in order. I actually proposed
moving the dp_hash up with the other metadata in struct flow, so that it could
be pushed together with the other metadata, but did not follow through. I'll
send a patch to do that.
Jarno
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev
