Code 25 (port redirection only) raises a separate issue that I think is orthogonal to REJ handling. The CM spec always allows the CM to run on a different port from the destination of a connection so if we want to implement the whole spec we do need a way for the consumer to specify a primary and alternate path for the connection itself as well as a CM path.
It looks like one way to handle this is to add a new field to struct ib_cm_req_param: struct ib_sa_path_rec *cm_path. A user could easily set this to the same path record referenced by the *primary_path field. This would only be a small change to existing users.
I'm still not sure how to handle a failed path between the CMs. And even if you could detect a failed path, recovering from it is difficult without violating the REQ max_cm_retries. Is anyone planning on using APM?
- Sean _______________________________________________ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general