The IP over ATM Mailing List Archive by date

Cell Relay Retreat>List Archive>month:1995-Feb> msg00174



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Fixed-length format for (In)ATMARP?

  • From: Mark Laubach <laubach@terra.com21.com>
  • Date: Tue, 28 Feb 1995 09:54:41 -0800 (PST)
  • CC: Craig Partridge <craig@aland.bbn.com>, Carl Marcinik <carlm@fore.com>, ip-atm@matmos.hpl.hp.com, atmiol@sun4.iol.unh.edu, gja@thumper.bellcore.com

> So perhaps we get:
> 
>      ar$sha   20 octets  source ATM number
>      ar$ssa   20 octets  source ATM subaddress
>      ar$spa   16 octets  source protocol address
>      ar$tha   20 octets  target ATM number
>      ar$tsa   20 octets  target ATM subaddress
>      ar$tpa   16 octets  target protocol address

Yup, with specific rules.

> I'd prefer it if we set ar$spln = 0 (or ar$tpln = 0) to indicate
> null protocol addresses, rather than a zero-filled but non-zero length
> address.

That's an opinion to use the length.  Other comments?
 
> Its just ARP that has the minor problem. However, it
> would be sufficient to say that ar$spln == 4 for IPv4 addresses
> and ar$spln == 16 for IPv6 addresses, given ar$op == 0x800 in
> both cases.

Ugh, that's ugly.  It works though.  Imagine an IPv4-only host doing an
ARP and getting an IPv6 address back....yuk.  But, that's a discussion for
some other WG.... 

> (aside: either 4, 16, or 20 bytes protocol address fields will
>  still use 3 cells per ATMARP when all fields contain information).

Thanks for the arithmetic check!

mark