[openib-general] gen2 dev branch

[Hal Rosenstock]

Roland Dreier wrote:
> I don't like the assumption that that every MAD on QP1
> will follow all the GSI rules (especially since CM doesn't really
> follow all the GSI ruls).  

As far as I know, CM follows all the GSI rules.

Are you referring to the CM being the only GS entity to use send 
rather than request/response methods (and the use of transaction IDs 
for message sequences rather than request/response pairs) ? 

In any case, there are several existence proofs of CMs running
on top of a interface similar to the one being proposed.

> I don't like building a general kernel GSI
> layer when the SA client is the only known consumer.

Not sure what you mean by known in the above. Are you 
referring to "phase 1" kernel consumers only in that statement ?

-- Hal