There seems to be a lot of confusion about the SNFMVDM testcase so
here is some background information:

To start with, this testcase (and all the .exe's used) have been
running since OS/2 version 1.2.  They have not been recompiled, and
have run successfully on every release since.

On the PCM Compatibility Table, there have been several thousand system
models that have been tested, and all of these passed this testcase.

From the information you provided, there appear to be approximately
5 or 6 manufacturers that are having trouble with this testcase on
one of their models.  However it does not indicate how many other
models each of these PCM's have, that have passed this testcase.

As we saw with the VME problem last year, the testkit does what it was
intended to do.  That is to find out if there is a basic compatability
problem with the PCM model and OS/2.  And while the VME problem did
not affect every system, it certainly affected more than the 6 or so
we have information on concerning MAT015.EXE within SNFMVDM.

The analysis of MAT015.EXE has NOT shown that the testcase is wrong.
The testcase code should execute correctly, but on these systems, it
does not, that is being investigated.  Any fix made to correct this
problem would be in either OS/2 or possibly in the PCM's systems.

The SNFMVDM testcase has proved beneficial in finding basic
compatibility problems which have resulted in changes to the OS/2 code,
as well as changes to the PCM systems themselves.

Changing a testcase that has been successful in finding a problem does
not fix the problem, it only goes around it.  So instead of IBM or the
PCM finding out there is a problem, the user finds it the hard way.