[openib-general] Re: [PATCH 9 of 18] ipath - char devices for diagnostics and lightweight subnet management

[Roland Dreier]

    Bryan> We have customers who use our driver who do not want a full
    Bryan> IB stack present, for example in embedded environments.

I think it's fine that your low-level driver can work without ib_core,
ib_mad and the rest loaded.  But I still (after all this discussion)
don't understand why you need to have two SMA implementations to
handle this along with the code to switch between the two modes like:

 > +	list_for_each_entry(dd, &ipath_dev_list, ipath_list) {
 > +		if (!(dd->ipath_flags & IPATH_INITTED))
 > +			continue;
 > +		*dd->ipath_statusp &= ~IPATH_STATUS_SMA;
 > +		if (ipath_verbs_registered)
 > +			*dd->ipath_statusp |= IPATH_STATUS_OIB_SMA;
 > +	}

You also have all the functions like recv_subn_get_nodeinfo() etc. for
handling SM queries.  Presumably all this is duplicated in the
userspace SMA.  Why can't you get down to one NodeInfo query handler?

 - R.