[ovs-dev] Support for in_phy_port

[Ben Pfaff]

On Mon, Nov 25, 2013 at 05:22:47PM -0800, Jesse Gross wrote:
> On Mon, Nov 25, 2013 at 4:04 PM, Simon Horman <horms at verge.net.au> wrote:
> > On Mon, Nov 25, 2013 at 10:09:30AM -0800, Ben Pfaff wrote:
> >> On Mon, Nov 25, 2013 at 10:09:55PM +0900, Simon Horman wrote:
> >> > as far as I can tell no one is actively working on the following item in
> >> > OPENFLOW-1.1+. So I have made a start it.
> >> >
> >> >     * The new in_phy_port field in OFPT_PACKET_IN needs some kind of
> >> >       implementation.  It has a sensible interpretation for tunnels
> >> >       but in general the physical port is not in the datapath for OVS
> >> >       so the value is not necessarily meaningful.  We might have to
> >> >       just fix it as the same as in_port.
> >> >       [required for OF1.1; optional for OF1.2+]
> >>
> >> Sounds good!  I hope you're planning to do something simple.
> >
> > My main plan is to allow communication of the in_phy_port field
> > between ovs-vswtichd and the datapath by adding a new netlink key.
> > Then to expose that in packet_in messages. I also have it in mind
> > to allow matching on the in_phy_port, but probably later.
> >
> > As for determining the in_phy_port. My plan is to determine the vport that
> > tunneled packets arrive on in their encapsulated form and use that as the
> > in_phy_port. I plan to not set the in_phy_port for non-tunnelled packets;
> > to set it to the same as in_port for non-tunnelled; or some combination of
> > the two depending on how the code pans out.
> 
> How do you plan on getting the physical vport? It seems a little
> challenging because the port might not be attached to OVS at all or at
> the very least it is likely not attached to the same bridge.

That's one main reason I haven't bothered with this: it seems unlikely
to be useful.