[openib-general] CMA: compliancy issue?

[Or Gerlitz]

Michael S. Tsirkin wrote:

> CMA currently sends RTU from cma_rep_recv, before notifying
> the user.

> I think the same problem might affect other ULPs as well: CMA
> should notify the user *before* responsing with CM message, and
> should supply a way for the user to reject the connection,
> with or without destroying the CMA ID.

Please read a little deeper into the documentation of rdma_create_qp and 
rdma_accept in rdma_cm.h and also see the code.

The design of the CMA say that if the ULP wants the CMA to do the QP 
states transitions it should call rdma_create_qp and later the CMA will 
not deliver up event of REP but instead modify the QP state to RTR, RTS 
and send an RTU. And vise versa, if the ULP want to get a callback on 
REP it should ***not*** call rdma_create_qp and rather do the state 
changes by itself.

Or.