source: rtems/c/src/lib/libbsp/m68k/idp/README @ 9b4422a2

This board support package has not been tested with multiprocessor
or the timing support.  The uniprocessor IDP package was tested though
with a fairly large application (although bugs may exist).  The
$RTEMS_ROOT/src/tests/Makefile should therefore have the samples and
the tests directories compiled first with the tmtests (timing) and
multiprocessor directories optional.

Let me know if you have any problems or bug fixes.  Bug fixes are greatly
appreciated.  I do not work for RTEMS or am a member of the RTEMS support
group in any way, however.  I am just one of many that appreciate
"free" software and enjoy contributing when possible :).  You can do it too!
 
                -- doug mcbride
                   mcbride@rodin.colorado.edu

==============================================================
Notes: Make sure that -msoft-float is defined when compiling in the
newlib-beta-rtems/newlib/msoft-float directory subtree.  You also probably
want to add the following line to line 413 of
newlib-beta-rtems/newlib/msoft-float/libc/stdio/vfprintf.c:

#define INTEGER_ONLY

That allows you to pass most of the paranoia test in the samples
directory of RTEMS although you can't see the floating point values actually
printed (software floating point tends to make that difficult anyway).
In order to pass the whole paranoia test (with one flaw), however, I had to
comment out the following line in milestone 140 (why?):

/*  printf ("Testing X^((X + 1) / (X - 1)) vs. exp(2) = %.17e as X -> 1.\n",
    Exp2); */