Subject: Re: -msoft-float m68k build
To: Aaron J. Grier <agrier@poofygoof.com>
From: Ian Main <imain@stemwinder.org>
List: current-users
Date: 07/24/2002 11:11:34
hrrm, yes, that's no good.  I guess MKSOFTFLOAT hasn't been tested on 
m68k build..

On Wed, Jul 24, 2002 at 10:47:33AM -0700, Aaron J. Grier wrote:
> On Wed, Jul 24, 2002 at 10:01:07AM -0700, Ian Main wrote:
> 
> > So I co'd -current last night and am trying to cross compile a release
> > for the mac68k on my i386 using -msoft-float (MKSOFTFLOAT=yes).  I'm
> > doing this so I can put it on my LC475 with broken FPU.
> > 
> > However, I'm getting the following error:
> > 
> > /usr/src/tools/obj/tools.NetBSD-1.5ZC-i386/bin/m68k--netbsdelf-gcc -O0 -g -DALL_STATE -DUSG_COMPAT -msoft-float -Wall -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Wno-uninitialized -Wreturn-type -Wpointer-arith -Wswitch -Wshadow  -Werror  -D_LIBC -DLIBC_SCCS -DSYSLIBC_SCCS -D_REENTRANT -I/usr/src/lib/libc/include -DHESIOD -DINET6 -DNLS -DYP -D__DBINTERFACE_PRIVATE -I/usr/src/lib/libc/../../libexec/ld.elf_so -I/usr/src/lib/libc/dlfcn -DI18NMODULE_MAJOR=4 -DWITH_RUNE -I/usr/src/lib/libc -DRESOLVSORT -I. -DPOSIX_MISTAKE -DPORTMAP -DFLOATING_POINT -nostdinc -isystem /usr/mac68k_build/usr/include  -c /usr/src/lib/libc/arch/m68k/gen/ldexp_881.c
> > /usr/src/lib/libc/arch/m68k/gen/ldexp_881.c: In function `ldexp':
> > /usr/src/lib/libc/arch/m68k/gen/ldexp_881.c:59: inconsistent operand constraints in an `asm'
> > *** Error code 1
> 
> ldexp_881.c contains the following snippet:
> 
>         __asm ("fscalel %2,%1"
>                 : "=f" (temp)
>                 : "0" (value), "g" (exp));
> 
> fscalel is a floating point op, which obviously you're trying to
> avoid...
> 
> I'm wondering why that directory is even being visited, since almost all
> the assembly files in there use floating point operations, which is
> something you're trying to avoid.
> 
> -- 
>   Aaron J. Grier | "Not your ordinary poofy goof." | agrier@poofygoof.com