>Ideally the best approach would be to have a mux at the ib_mad level. We could >allow a user space application to intercept all outbound MADs for a given class >and/or attribute. Unlike the present "snooping" of mads, this would literally >be a interception. This would provide a number of key advantages:
I agree that this is a good idea, and I mentioned something similar to this before on the list. The idea was rejected in favor of using standard SA redirection. That said, it may not be the best approach in all situations. It restricts any solution to the MAD layer only. If you look at how IB ACM approaches the problem, it combines the name/address resolution (ARP) with route resolution (PR query) without sending QP 1 traffic. A solution limited to snooping MADs would be less efficient. For these two patches, if user space can determine the address mapping that they want for their connection and/or the path that should be used, do we allow the user to specify either or both? I'd like to keep this discussion somewhat separate from how those were obtained. - Sean -- 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