[openib-general] [RFC] [PATCH 2/7] ib_multicast 2.6.20: add ib_multicast module to track join requests from the same port

[Roland Dreier]

    Eitan> Hi Or, Maybe I did not explain myself right.  The idea is
    Eitan> not to implement it in the mad.c code but rather to
    Eitan> implement it at the lowest level: The problem with a new
    Eitan> API is that a single ULP/applications which does direct
    Eitan> umad or QP1 access will break the reference count.

    Eitan> Implementing at the lowest level - I.e. by sniffing QP1
    Eitan> packets - would be enforced for all applications/ULPs.

Does not seem very feasible -- for one thing you would have to attach
a privilege level to every MAD somehow, and keep that all the way
through the stack.  And then as far as I can tell you would
essentially need a complete local copy of all GS handling to decide
what to do with each QP1 message.

 - R.