[HARDWARE] Has anyone looked into USB networking for a small rack of motherboards.

Joe Zbiciak j-zbiciak1 at ti.com
Sun Nov 1 02:59:17 EST 1998

'Timothy Baldwin' said previously:

| > SLIP (and any other TCP/IP encapsulation technique) need bidirectional
| > links, because TCP needs a bidirectional link.   (How else would
| > something as simple as the standard three-way-handshake work?)
| Nonsense
| TCP uses IP and will work if IP packets can be sent to and received from
| the remote host. SLIP is a pure encapsulation protocol (unlike PPP) and
| therefore can work with unidirectional links.
| The IP layer will take care of this, by routing packets round the ring.

SLIP is indeed merely an IP-encapsulation protocol, and if all of the
required routes exist between any two given nodes, then yes, the IP
routing layer will do all of the dirty work.  

I'm slapping myself for not looking more closely through the RFC
before opening my big mouth.  (Interested parties can read RFC 1055
at   http://www.cis.ohio-state.edu/htbin/rfc/rfc1055.html   ).  I was
under the impression that SLIP required some amount of feedback from
the host it was communicating with, eg. for negotiating parameters,
etc. but it does not.  PPP does, though.  I had remembered that SLIP
was much more limited than PPP, but I had forgotten just _HOW_ much
more limited.  :-)

Every SLIP implementation I've ever seen expects to send and receive IP
frames from the same host, and so I've always assumed that the SLIP
protocol required at least a bi-directional link.  It does not; this
imagined requirement is due to the implementations, not due to the
protocol.  It merely specifies how to frame an IP packet.  (And at
that, the required framing is very, very simple.)

Does anyone know of a _complete_ unidirectional SLIP implementation? :-)  
A simple high-level implementation of "send_packet" and "recv_packet"
do exist in the RFC that can be used as a starting point.

BTW, Tim, thank you for slapping me about a bit.  I was severely
lacking a clue there for my past couple posts.  :-)



 +------ Joseph Zbiciak -----+
 | - - j-zbiciak1 at ti.com - - |  "The meaning of a value is determined 
 |-Texas Instruments, Dallas-|   by how it is used."
 | - #include <disclaim.h> - |                          -- Ousterhout

To unsubscribe, send 'unsubscribe hardware' to majordomo at lists.distributed.net

More information about the Hardware mailing list