Roland Dreier wrote:
Bob> There is already a RDMA device independent API being Bob> developed for the kernel by people on this list. It is Bob> starting with the kDAPL code base, which was designed to Bob> support both IBA and iWarp devices.I believe kDAPL-based layers are an OK short-term solution, but I don't think anything like this should be proposed for merging in the Linus kernel. If we need another abstraction layer on top of our existing abstraction layer, that just says to me that we should fix the current abstraction layer.
My personal view is that any higher level abstractions could combine common functionality or simplify the interface, possibly with a loss of lower level control. For example, an interface could be created that allowed the user to operate with more abstract addressing, hiding SA query and CM interactions. Such an interface may not be needed yet, but if there were multiple RDMA transports, it could be valuable. However, I do not think this necessarily translates into the need to abstract concepts like PDs, QPs, and CQs.
- Sean _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
