[openib-general] [PATCH] [MAD] changes to ib_create_send_mad

[Hefty, Sean]

>> E.g. if you look at the SA, the ACK (GetTable) needs to go to the
agent
>> using the lookup tables.  The TID carried in the ACK was actually set
by the
>> sender of the ACK, and cannot be used to route to the SA's agent.
>
>Right now, OpenSM registers the SA MAD agent with all methods set in
the
>method mask (overkill) so I'm not sure what is not happening correctly
>yet.

This sounds fine.

>> Can you describe in more detail what the exact failure that you're
seeing is
>> and what you were expecting?
>
>When the ACK is sent, I do not get a send failure. This is expected
>based on what you said at the SA side. I think it (TID matching) would
>only happen in the MAD layer automatically on the SA client side so you
>would get a send failure if the ACK didn't come back in the other
>direction.

Are you referring to RMPP direction switch?  If so, that is not yet
implemented.
I.e. The code will not generate or process ACKs to ACKs, which are
required for
RMPP direction switch.  What the code does now is to match RMPP requests
with
RMPP responses, but in a manner similar to receiver initiated RMPP.

I don't think that the routing algorithms in the MAD layer will need to
change
in order to support direction switch.  Also, I don't think that
supporting
direction switch on the initiator's side is difficult.  It is a little
more
challenging on the responder side.  On the responder side, the MAD layer
would need to know if a received MAD was going to generate an RMPP
response.
I'm not sure how to do this in a generic fashion, without the MAD layer
being protocol aware.

- Sean