On Wed, Jul 04, 2012 at 05:50:43PM +0900, Simon Horman wrote: > Signed-off-by: Simon Horman <ho...@verge.net.au>
This seems reasonable. I believe that we can test the behavior here by adding a few more lines to the "ovs-ofctl parse-nxm" and "ovs-ofctl parse-oxm" tests in tests/ovs-ofctl.at. If you do that to validate it, I'll commit it. Also, here, VLAN_CFI is just one bit, and vlan_tci is in network byte order, so you can just write "flow->vlan_tci & htons(VLAN_CFI). (As-is, in fact, VLAN_CFI != htons(VLAN_CFI) on x86, so it will never be true, right?) > @@ -878,6 +878,8 @@ mf_are_prereqs_ok(const struct mf_field *mf, const struct > flow *flow) > return flow->dl_type == htons(ETH_TYPE_IP); > case MFP_IPV6: > return flow->dl_type == htons(ETH_TYPE_IPV6); > + case MFP_VLAN_VID: > + return (flow->vlan_tci & VLAN_CFI) == htons(VLAN_CFI); > case MFP_VLAN_TPID: > return (flow->vlan_tpid == htons(ETH_TYPE_VLAN) || > flow->vlan_tpid == htons(ETH_TYPE_VLAN_8021AD)); Thanks, Ben. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
