Hi,

"J. MacPhail" <jrm@kw.igs.net> wrote:
<snip>
> >
> > You can't really compile anything since gcc randomly crashes, a
> > normal boot leaves a half dozen or so core files in /, etc.  It was
> > kind of ugly and usable only if you were dedicated since the
> > solution to a random crash was try running the command again, and
> > again, and again, until it worked.  Or, rebuild it from scratch.
> >  
> > Oh, and X starts up only once in a while.
> 
> So we can expect floating point operations to show up in libm.so,
> libc.so, at least one other library (libcrypt.so?), and other programs
> that would call these but are statically linked.
> 

Well, and programs that use them internally.

Of the top of my head

top
tcsh
perl

 
> I don't understand this.  The info for gcc-3.3.3, M680x0 Options, says
> 
> `-msoft-float'
>      Generate output containing library calls for floating point.
>      *Warning:* the requisite libraries are not available for all m68k
>      targets.  Normally the facilities of the machine's usual C
>      compiler are used, but this can't be done directly in
>      cross-compilation.  You must make your own arrangements to provide
>      suitable library functions for cross-compilation.  The embedded
>      targets `m68k-*-aout' and `m68k-*-coff' do provide software
>      floating point support.

My experience is that -msoft-float works fine in the compile step, 
but, come ld time ld fails finding the proper routines to impliment
the functions that gcc has called.

I just tried on a 060 running 1.6.2 and got the following:

#include <stdio.h>
int main (int argc, char **argv)
{
        double a=1.0;
        double b=7.0;

        double c;

        c = a/b;
        printf("%lf %lf %lf\n",a,b,c);
        return 0;
}

lilith: {6} cc -msoft-float -o t1 t1.c
lilith: {7} ./t1
1.000000 7.000000 nan

lilith: {9} cc -o t1 t1.c
lilith: {10} ./t1
1.000000 7.000000 0.142857


So, ld worked but the -msoft-float produced unusable code.

> 
> And that still leaves me uncertain whether or not native compilation
> with -msoft-float is supposed to just work.

On my system (040 with 1.6.2 with the -msoft-float option when built)
if I do not use -msoft-float I have an output with fmove, fdiv, etc 
in it.  OTOH, if I do use 
-msoft-float the only fpu looking thing I see is a call to __divdf3.

So native compulation (with divdf3 et al in libc) does seem to work.
It also works as a cross compiler since I do my full builds
on a macppc system (thanks John Klos!).  Of course this is 2.95.3.

> 
> 
> And surprises.  Even if all your stuff is perfect, maybe the
> -msoftfloat option to gcc has some unexpected side-effect.  (That
> could potentially make it hard to list all binaries containing
> floating point operations....)

That's correct, the code generated with -msoft-float, on the surface,
does not really look the code generated when -msoft-float is not
used.

Thanks!

cheers

bruce


> 
> 
> > I'll try to do the diff next time I build it.
> 
> Great!
> 
> -- 
> John