>But that isn't really the point, IB CM already provides this API so, >why corrupt the RDMA CM abstraction with this?
The RDMA CM abstraction does not work on larger cluster sizes. Either we can fix the RDMA CM or force all RDMA CM applications to be re-written and become transport aware. This allows the librdmacm to implement a fix and communicate the necessary information to the kernel for connection establishment using generic mechanisms and an existing RDMA CM API. (One alternative is for the kernel RDMA CM to communicate directly with the IB ACM or other external service, but I'd like to see much wider testing, acceptance, and standardization before attempting that.) I could not find an efficient way of implementing the entire solution in user space because of the event reporting and port space handling. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html