[openib-general] Re: RDMA Generic Connection Management

[Roland Dreier]

    James> For example when SRP's srp_add_one() function is called, I
    James> don't see a synchronization primitive setup to do
    James> this. Suppose for example that a thread was in SRP's CM
    James> callback handler, srp_cm_handler(), just before the call to
    James> ib_modify_qp() on line 713 when the device removal
    James> callback, srp_remove_one(), is called. Currently, the SRP
    James> code will still call ib_modify_qp() even though the device
    James> has been removed.

That's a good catch.  I'll add a little code to SRP to fix it.  I also
plan on fixing the user MAD problems later this week.

user verbs is definitely broken and and needs to be fixed, although I
worked around the worst problems by taking a reference on the
low-level driver module so at least it can't be unloaded.

I think the SA client is OK, since the ib_unregister_event_handler and
ib_unregister_mad_agent calls in the remove method should wait until
everything is cleaned up.  Similarly I think the MAD module is OK.

 - R.