[openib-general] mapping between IP address and device name

[Hal Rosenstock]

On Fri, 2005-06-24 at 15:14, Talpey, Thomas wrote:
> At 02:27 PM 6/24/2005, Hal Rosenstock wrote:
> >On Fri, 2005-06-24 at 13:51, Talpey, Thomas wrote:
> >> mentioned earlier. For better or for worse, the ATS approach is easily
> >> administered and does not impact any protocol layers outside of its
> >> own. I think of it as ARP for IB.
> >
> >reverse ARP for IB
> 
> Actually, I did mean ARP. It's only necessary to answer the questions
> "please send this to address foo" (for making connections),

This is the ARP side (get the link layer address (GID + QPN) for an IP
address) I'm connecting to.

>  along with
> "what address sent this request" (for accepting connections and export
> checking).

I think this is the RARP side (get the IP address that corresponds to a
partial link layer address (GID)). This was the side being discussed on
the list, right ?

Do I have this backwards ?

> There is no need for RARP, which would answer the question "who is
> supposed to belong to this MAC address". In fact, I don't want to know
> the answer to that! :-) That's link layer stuff, three floors down.

I think we are just discussing terminology. Perhaps I should
have said reverse ARP for IP (not IB). You are calling that ARP for IB.

Doesn't RARP say for a given MAC (hardware) address give me the IP
address and that's exactly what ATS is doing (where the MAC address is
equivalent to the GID) in the case under discussion ? ARP is the other
way 'round: for an IP address give me the link layer address.

-- Hal