[openib-general] RFC on CM error handling

[Sean Hefty]

>  Actually, I was thinking about returning an error from the REP/REQ
>handler, this I believe is suppose to destroy the communication identifier,
>which I presume will generate a REJ? This is at least the behaviour I would
>like to see. In this case, since the communication identifier is destroyed
>there wouldn't be any other state transitions.

Some thoughts about destroying cm_id's based on a callback return value.

Given the current code structure, it's possible for a user to receive
multiple simultaneous callbacks.  For example, a user could be notified of a
reply and reject (sent as a result of an rtu timeout) at the same time.  The
CM will handle the state transitions properly to fail calls into the API.
But I can see where it would be easy for a user to end up trying to destroy
the cm_id multiple times, or receiving the second callback after they had
already destroyed their context.

I'm not seeing an easy way to handle this condition.  I think that I need to
serialize all callbacks for a given cm_id.

- Sean