On Mon, Jan 04, 2016 at 02:46:09PM -0500, Thomas F Herbert wrote: > > > On 1/4/16 1:20 PM, Ben Pfaff wrote: > >On Wed, Dec 30, 2015 at 07:29:15PM -0500, Thomas F Herbert wrote: > >> > >>On 11/9/15 3:58 PM, Ben Pfaff wrote: > >>>On Fri, Oct 02, 2015 at 05:31:58PM -0400, Thomas F Herbert wrote: > >>>>From: "Thomas F. Herbert" <[email protected]> > >>>> > >>>>ToDo: The flow structure needs to be updated to hold both the inner and > >>>>outer > >>>>tpid since these are encoded by the kernel module. > >>>I think I'd like to see the version with both TPIDs encoded. That's the > >>>way I've always envisioned this working. > >>I have been working on the changes to add the inner and outer tpids to the > >>flow struct this week. I changed flow struct to place two vlan structs right > >>after the dl_type with the outer first followed by the inner. This is done > >>so the miniflow would still work with few modifications because dl_type and > >>outer tci are sequential. However, this raises the issue as to whether the > >>outer tpid and the inner vlan, tci and tpid should also be added to the > >>miniflow. What are your thoughts? > >I don't understand the question. A miniflow and a flow represent the > >same data, only a miniflow has a compressed format. > Yes, they represent or should represent the same data but that wasn't my > question. Maybe the question will be clearer when I submit the patch. The > miniflow_extract function requires some changes that wasn't done for > previous versions of this patch. The current miniflow extract compresses > only the outer vlan_tci, dl_type. For this version, I am writing a > miniflow_push_vlan() macro to compress both the tpid and tci into the > miniflow and call the same function to compress the inner tpid and tci as > well.
OK. I think it will be clearer from the patch. _______________________________________________ dev mailing list [email protected] http://openvswitch.org/mailman/listinfo/dev
