[openib-general] [PATCH] Initial checkin of userspace MAD access

[Roland Dreier]

    Sean> Is anyone willing to work on porting opensm to this?  If
    Sean> not, I can start on this.  Otherwise, I will continue
    Sean> working on adding MAD error/overrun handling.

It would be great to work on that but we need to resolve how to handle
the SM classes first.

One option would be to extend the user_mad code to handle MADs
timeouts and have OpenSM only receive solicited MADs (and have OpenSM
register an agent with class == 0).  This is only a temporary solution
because ultimately OpenSM needs to receive SMInfo SMPs.  I think this
still requires some figuring for how to handle the DR SMI.

Another option is to revise the kernel MAD code so that it does not
need to register an agent for the SM classes (ie pass all MADs to
low-level driver first).

 - R.