Re: gcc 2.8.1 - ported!

[LE COAT Francois <lecoat@ief.ief.u-psud.fr>]

Hello Christian,

Christian Felsch wrote:
> Today I've uploaded my new port of gcc to this server:
>   ftp://ftp.rz.tu-harburg.de/pub/software/systems/atari/gnu/gcc/2.8.1
> 
> It should be installable like the port of gcc 2.7.2.3.

I have problem with this port of GCC 2.8.1 for compiling programs
that require floating point computation with the -m68060 option.
For example This program :

#include <stdio.h>
#include <math.h>
void main()
	{
	double x;
	printf("enter x in degrees : ");
	scanf("%lf",&x);
	printf("\ncos(%lf) = %lf\n",x,cos(x*3.141592654/180.0));
	getc(stdin);
	}

With this makefile :

LIBPATH = d:\gnu\gnucc\lib
INCPATH = d:\gnu\gnucc\include
GCCEXEC = d:\gnu\gnucc\
GNULIB = d:\gnu\gnucc\lib
GNUINC = d:\gnu\gnucc\include
TEMP = c:\tmp

CFLAGS=		-c -m68060 -I$(INCPATH) -D__M68881__
LFLAGS =	-o hello.prg
CC =            d:\gnu\gnucc\gcc

# Make's implicit rules for making a .o file from a .c file...
#
.c.o :
	$(CC) $(CFLAGS) $*.c


OBJS = hello.o


povray:	$(OBJS)
	$(CC) $(LFLAGS) $(OBJS)

produce the following logfile (output errors) :

d:\gnu\gnucc\gcc -c -m68060 -Id:\gnu\gnucc\include -D__M68881__ hello.c
hello.c: In function `main':
hello.c:4: warning: return type of `main' is not `int'
/c/tmp/cc210000.s: Assembler messages:
/c/tmp/cc210000.s:26: Error: invalid instruction for this architecture;
needs 68040 or 68060 -- statement `fdmoved a6@(-8),fp0' ignored
/c/tmp/cc210000.s:27: Error: invalid instruction for this architecture;
needs 68040 or 68060 -- statement `fdmuld #0r3.1415926540000001,fp0'
ignored
/c/tmp/cc210000.s:28: Error: invalid instruction for this architecture;
needs 68040 or 68060 -- statement `fdmovex fp0,fp1' ignored
/c/tmp/cc210000.s:29: Error: invalid instruction for this architecture;
needs 68040 or 68060 -- statement `fddivd #0r180,fp1' ignored
/c/tmp/cc210000.s:30: Error: invalid instruction for this architecture;
needs fpu (68040, 68060 or 68881/68882) -- statement `fmoved fp1,sp@-'
ignored
/c/tmp/cc210000.s:33: Error: invalid instruction for this architecture;
needs fpu (68040, 68060 or 68881/68882) -- statement `fmoved fp0,sp@-'
ignored
/c/tmp/cc210000.s:88: Error: invalid instruction for this architecture;
needs fpu (68040, 68060 or 68881/68882) -- statement `fmovem #0x4,sp@-'
ignored
/c/tmp/cc210000.s:89: Error: invalid instruction for this architecture;
needs 68040 or 68060 -- statement `fdmoved a6@(8),fp1' ignored
/c/tmp/cc210000.s:91: Error: invalid instruction for this architecture;
needs fpu (68040, 68060 or 68881/68882) -- statement `fcosx fp1,fp2'
ignored
/c/tmp/cc210000.s:93: Error: invalid instruction for this architecture;
needs fpu (68040, 68060 or 68881/68882) -- statement `fmoved
fp2,a6@(-8)' ignored
/c/tmp/cc210000.s:98: Error: invalid instruction for this architecture;
needs 68040 or 68060 -- statement `fdmoved sp@+,fp0' ignored
/c/tmp/cc210000.s:102: Error: invalid instruction for this architecture;
needs fpu (68040, 68060 or 68881/68882) -- statement `fmovem sp@+,#0x20'
ignored
make: *** [hello.o] Error 1

__________________________________________________________________________

I am wondering wether there is something to add to the specs file.
Is there special library that need to be used for 68060 ? I use
the inline definition of mathematical functions included in
math-68881.h. I've updated this file with the one included in the
GCC 2.8.1 archive distribution, but it is still not working ...
Is there something to do ?

Thanks,

-- Francois LE COAT
E-mail : mailto:lecoat@ief.u-psud.fr
WEB : http://www.ief.u-psud.fr/~lecoat
Eureka 2.12 Author (2D Graph Describer 3D Modeller)