Missing collisions on unmanaged repeater (this is a proble..

[hardware]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Hi all,

I have been working on a small computing system with two processors
(Motorolas MPC860C). In order to communicate with these, we are using an
ethernet connection. This is done by using the on-chip Fast Ethernet
Controller on the MPC860C to connect, via MII, to Intels LXT9863A
un-mananged repeater. This repeater has two MII ports, and 6 100Base-TX
ports. I then use one of the 100Base-TX ports to connect the system to a PC.

The MPC860Cs, the LXT9863A and various magnetics are placed on the same PCB.
The LXT9863A has been configured ancordingly to the datasheet provided by
Intel. The Inter-Repeater Backbone (IRB) connection is terminated and the
unused ports has also been terminated.

This setup is working. However when the traffic becomes to intensive,
packets are lost. Closer examination has established that the collision
signal from LXT9863A is never asserted. This is includes the LED signal
(100-COL), the MII-COL signal to the MPC860Cs and the IRB 100-COL signal.
Since the collision signal is never asserted, the software in the MPC860Cs
have to use higher layers of the protocol to find lost packets. When the
traffic becomes to frequent, the higher protocol layers cannot keep up, and
some packets are never transmitted.

What I would like to know is if anyone has experienced problems with the
LXT9863A, especially the missing collision signal. Likewise I would like to
know if thers is some problem with the functionality of the LXT9863A that I
have overlooked, that creates problems with the described setup.

Best regards,

Jens Seiersen
Hardware Developer
ETI A/S Denmark

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

hardware <hardware@eti.dk> wrote:
> Hi all,

> I have been working on a small computing system with two processors
> (Motorolas MPC860C). In order to communicate with these, we are using an
> ethernet connection. This is done by using the on-chip Fast Ethernet
> Controller on the MPC860C to connect, via MII, to Intels LXT9863A
> un-mananged repeater. This repeater has two MII ports, and 6 100Base-TX
> ports. I then use one of the 100Base-TX ports to connect the system to a PC.

> The MPC860Cs, the LXT9863A and various magnetics are placed on the same PCB.
> The LXT9863A has been configured ancordingly to the datasheet provided by
> Intel. The Inter-Repeater Backbone (IRB) connection is terminated and the
> unused ports has also been terminated.

> This setup is working. However when the traffic becomes to intensive,
> packets are lost. Closer examination has established that the collision
> signal from LXT9863A is never asserted. This is includes the LED signal
> (100-COL), the MII-COL signal to the MPC860Cs and the IRB 100-COL signal.
> Since the collision signal is never asserted, the software in the MPC860Cs
> have to use higher layers of the protocol to find lost packets. When the
> traffic becomes to frequent, the higher protocol layers cannot keep up, and
> some packets are never transmitted.

> What I would like to know is if anyone has experienced problems with the
> LXT9863A, especially the missing collision signal. Likewise I would like to
> know if thers is some problem with the functionality of the LXT9863A that I
> have overlooked, that creates problems with the described setup.

I don't have experiences you ask for, but i know that ethernet
is an unreliable media ( like almost any other media) and your
transport layer *has* to verify delivery.

Yes, you should fix the collision signal, but you should also
make shure your transport takes care of dropped packets.




--
Peter Håkanson
IPSec Sverige ( At Gothenburg Riverside )
Sorry about my e-mail address, but i'm trying to keep spam out,
remove "icke-reklam" if you feel for mailing me. Thanx.