[openib-general] Handling SM class (SMInfo vs. other queries)

[Roland Dreier]

    Hal> What are the other special cases for registration ?

Not just registration... I just meant that having an extra snoop_mad
entry point and a special issm bit and hard-coding different treatment
of SMInfo in the MAD layer starts to smell to me like the MAD layer is
at the wrong level of abstraction... (and I haven't thought through
the CM yet but I'm a little worried about how the solicited/unsolicited
distinction that the MAD layer makes will fit with the CM).

    Hal> Wish this part of the discussion occurred 10 days ago or so :-)

I first raised the issue at least two months ago:
    http://article.gmane.org/gmane.linux.drivers.openib/4217

    Roland> This also avoids having to add another agent for BMA to ib_smi.

    Hal> Adding another agent is relatively trivial as evidenced by PMA.

Sure, it's not a big deal; it's only a minor difference either way,
but giving the low-level driver first crack at all MADs does make the
ib_agent code simpler.

 - Roland