[openib-general] Re: IPoIB

[Hal Rosenstock]

On Mon, 2005-04-04 at 18:15, Roland Dreier wrote:
> I think it should "just work" with the current setup.
> ipoib_hard_header() will look at the skb it's passed, and if there's
> no neighbour struct, it will just save off the destination link
> address.  Then ipoib_start_xmit() will look at the destination address
> and handle multicast link addresses correctly.

That's good to hear. There are going to be some other changes for this.
At a quick glance, ipoib_main.c::ipoib_start_xmit drops a unicast link
level response if it is not ARP. RARP is also possible there, right ?

I'm not sure the Linux code above this is set up to support the larger
link level address needed by IPoIB either.

-- Hal