Re: [manet] [babel] RPF for meshes

[Donald Eastlake <[e-mail]>]

Hi Juliusz,

On Fri, Jul 29, 2022 at 2:47 PM Juliusz Chroboczek <[e-mail]> wrote:
>
> Hi,
>
> At today's joint MANET-Babel meeting, somebody (I don't recall who)
> mentioned that we might be able to solve the multicast problem by looking
> at the Ethernet header.  That got me thinking, and perhaps the following
> modification to RTP that works with MAC addresses instead of interfaces
> could be made to work.
>
> When you receive a multicast packet S->G, you lookup S in your unicast
> routing table; this gives you a next-hop address A.  You perform address
> resolution on A, this yields a MAC address M.  If the layer 2 source
> address of the packet is not M, you drop the packet.
>
> The procedure above clearly avoids loops if your unicast routing graph is
> loop-free, since it enforces that multicast packets follow the reverse
> unicast path.

I think that's a reasonable idea.

Thanks,
Donald

PS: I assume you meant RPF, not RTP, in the message body.
===============================
 Donald E. Eastlake 3rd   [phone] (cell)
 2386 Panoramic Circle, Apopka, FL 32703 USA
 [e-mail]

> I'm sure it breaks something, but what?
>
> -- Juliusz
>
> _______________________________________________
> babel mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/babel