[openib-general] Re: CMA: compliancy issue?

[Or Gerlitz]

On 5/8/06, Sean Hefty <mshefty at ichips.intel.com> wrote:

> I think that it makes more sense to give the user the CONNECT_RESPONSE event,
> and have them call rdma_accept() or rdma_reject(), versus allowing them to call
> rdma_reject() after an ESTABLISHED event.  There is already code for this; it's
> just tied to the existence of a QP off the rdma_cm_id.

I agree that it does not make sense to have REJ following ESTABLISHED,
so it should be either as it is now or change the cma to deliver RESP
have the ULP call accept or reject and based on the user call send RTU
or REJ, such that in the active side there is always one event of the
set {REJ, RESPONSE, CONNECT_ERROR} and ESTABLISHED in delivered only
in the passive side.

>From iSER point of view, this approach is fine, and it would allow for
some future flexibility to reject the REP. We prefer to implement it
only for 2.6.19, that is when 2.6.18-rc1 is out.

Or.