[openib-general] [ANNOUNCE] Initial trunk checkin of ISERinitiator

[Roland Dreier]

    Grant> Let me rephrase my question since your answer doesn't
    Grant> really address what I was asking: Any reason why kDAPL
    Grant> might not get pushed upstream?

Yes.  In fact, I am quite sure that kDAPL will not go upstream.  The
recent discussions trying to make sense of the kDAPL event handling
mess just convince me even more the kDAPL is quite broken.

I do agree that we need some abstraction of connection to cover both
IB and iWARP, but I think it would be much more productive to try and
build a sane connection API from scratch.

    Grant> If kDAPL for any reason doesn't get pushed upstream to
    Grant> kernel.org, we effectively don't have iSER or NFS/RDMA in
    Grant> linux. Since I think without them, linux won't be
    Grant> competitive in the commercial market place.

This I disagree with.  Linux is such a huge part of the RDMA world
that I would tend to believe the converse: without robust Linux
implementations, iSER and NFS/RDMA won't be competitive.

 - R.