[ovs-dev] [PATCH] datapath-windows: Removed hardcoded names for internal/external vports

[Nithin Raju]

> On Sep 14, 2015, at 7:51 PM, Alin Serdean <aserdean at cloudbasesolutions.com> wrote:
> 
> For this to work in the case for the internal port change you also need to change the following lines:
> https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_openvswitch_ovs_blob_master_datapath-2Dwindows_ovsext_Vport.c-23L2141-2DL2149&d=BQIGaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=pNHQcdr7B40b4h6Yb7FIedI1dnBsxdDuTLBYD3JqV80&m=zCHx69GiiUmOAzrRY67SE4QUGabdsfedEcpjTaoZODw&s=OQ7O90tnlLIaxihwmrdXcIPtNCta97dPUYjSX8t0-Rk&e= 

Thanks Alin for pointing this out.

The idea behind the checks above is that we want to be able to easily identify which of the vports represents the “internal adapter”. So, we don’t overwrite the vport->ovsType value to be OVS_VPORT_TYPE_NETDEV and preserve it to be OVS_VPORT_TYPE_INTERNAL.

The second check is that we need to identify ports of type “internal” to be ports that are internal to a bridge and we don’t forward packets to that port.

To implement Sorin’s change, we can probably nuke both the strcmp’s, and start treating hyper-v switch internal port to be a Netdev, but that would mean changes in other places as well.

> IMO we should not change the internal port name at the moment because in the case of multiple switches/supporting multiple adapters in the extension we will need to remove it.

This sounds like a good idea, IMO too.

-- Nithin