Conference 7.286::fddi

93.0. "Bit twisting" by HAMPS::PRUDEN_G () Wed Jul 18 1990 10:32

    I have a question about who will do the bit twisting of MAC addresses.
    With Ethernet using a different ordering of MAC address bits (canonical) to
    FDDI and .5 the MAC address in ARP packets will not automatically
    match up between FDDI and Ethernet. After some discussion with various
    people it would appear that FDDI stations will automatically twist
    bits so that the ARP packet contains canonical ordering. This will
    mean that bridges do not have to twist them inside the ARP packet
    between FDDI and Ethernet.
    
    However, will existing 802.5 TCP/IP implementations interwork with
    this? This would require them to twist the bits at the end station
    like FDDI. The questions I have therefore are :
    
    1. Am I correct in assuming that our TCP/IP drivers on FDDI twist
    the MAC address before sending it out on FDDI ?
    
    2. Do existing Token Ring end station implementations of IP twist the
    MAC address before putting it in ARP?
    
    3. If they don't will the .5 to FDDI bridge have to do it or will
    there be a scrap in the standards ?
                                        
    4. How is this handled in OSI (ES-IS 9542) i.e. did ISO define the
    bit ordering and therefore its done in end stations ? 
    
                                     

1. Yes.
2. I don't think so, but this is being changed.
3. I suspect .5 to FDDI bridges will have to do some hackery to deal with
   old .5 nodes.
4. OSI has this taken care of.

	paul