[PATCH] m68000 ELF + two m68k gas fixes

[Aaron J. Grier]

On Sat, Feb 02, 2002 at 04:47:39PM -0800, Jason R Thorpe wrote:
> On Fri, Feb 01, 2002 at 09:59:34AM -0800, Aaron J. Grier wrote:
> 
>  > On Fri, Feb 01, 2002 at 08:34:49AM -0800, Jason R Thorpe wrote:
>  > 
>  > >   -m68881 | -m68882 | -mno-68881 | -mno-68882\n\
>  > >   			target has/lacks floating-point coprocessor\n\
>  > > ! 			[default yes for 68020, 68030, and cpu32]\n"));
>  > 
>  > for what it's worth, I've always found this wording pretty ambiguous.
>  > "has/lacks" and "yes" don't seem to match up.  perhaps
>  > [default lacks for 68020, 68030, and cpu32] would be more clear?
> 
> Except if you look at the code, it's "default has for 68020, 68030,
> and cpu32".  :-)

eh?  I've been compiling with -mcpu32 for m68k-rtems since 2.95.2 and
never gotten floating point opcodes out of it.  however, -m68020 and
-m68030 ARE emitting floating point opcodes:

$ m68k-rtems-gcc -v
Reading specs from /usr/local/cross/lib/gcc-lib/m68k-rtems/3.1/specs
Configured with: 
Thread model: single
gcc version 3.1 20020131 (experimental)
$ cat float.c
double floatfunction(double foo, double bar) {
        return foo / bar;
}
$ m68k-rtems-gcc -mcpu32 float.c -c
$ m68k-rtems-nm float.o
         U __divdf3
00000000 T floatfunction
$ m68k-rtems-gcc -m68020 float.c -c
$ m68k-rtems-nm float.o
00000000 T floatfunction
$ m68k-rtems-gcc -m68030 float.c -c
$ m68k-rtems-nm float.o
00000000 T floatfunction

now do you see why I'm confused?

> I'm actually inclined to leave this part alone, since it doesn't
> really have anything to do with what this patch is trying to achieve.

looks like something needs to be updated...  "the behaviour observed
differs from the documentation."

-- 
  Aaron J. Grier  |   Frye Electronics, Tigard, OR   |  aaron@frye.com
     "In a few thousand years people will be scratching their heads
       wondering how on earth the first computer was invented and
          bootstrapped without a prior computer to do it with."
                    --  Chris Malcolm, on comp.arch