Re: Mupen64 FTBFS on Sparc
On Tuesday 09 May 2006 11:12, Eddy Petrişor wrote:
> > I try the build with -02 (instead of -DX86 -03):
>
> Does that means that I am absolved of building it on ppc? :-)
Well, if we can get it to work on Sparc, it'll probably work on PPC too.
But...
> > (snip)
> > r4300/x86/gcop0.o r4300/x86/gcop0.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/assemble.o r4300/x86/assemble.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/gcop1.o r4300/x86/gcop1.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/gcop1_s.o r4300/x86/gcop1_s.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/gcop1_d.o r4300/x86/gcop1_d.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/gtlb.o r4300/x86/gtlb.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/gregimm.o r4300/x86/gregimm.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/gspecial.o r4300/x86/gspecial.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/gcop1_w.o r4300/x86/gcop1_w.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/debug.o r4300/x86/debug.c
> > gcc -O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall
> > -pipe -c -o r4300/x86/rjump.o r4300/x86/rjump.c
> > {standard input}: Assembler messages:
> > {standard input}:60: Error: Illegal operands
> > {standard input}:64: Error: Illegal operands
> > {standard input}:93: Error: Illegal operands
> > make[1]: *** [r4300/x86/rjump.o] Error 1
> > make[1]: Leaving directory `/tmp/buildd/mupen64-0.5.dfsg'
> > make: *** [build-stamp] Error 2
> > pbuilder: Failed autobuilding of package
>
> What I find weird is that all of these are under an x86 directory!?
> Isn't this arch dependant code?
Ah, yes, the wonders of emulator code. I know there's a Mac OS X port, so I
assumed that the code would Do The Right Thing on non-x86 arches once X86 was
undefined. Guess not. Some tinkering and googling revealed the _BIG_ENDIAN
define and a forum thread on compiling Mupen64 for PPC[1], so I think I can
get it to work.
If anyone out there wants to hack at it, try these CFLAGS on any non-x86 arch
and post the build logs:
CFLAGS
=-O2 -fomit-frame-pointer -ffast-math -fno-strict-aliasing -fPIC -Wall -pipe -DCOMPARE_CORE -D_BIG_ENDIAN
[1] http://www.emutalk.net/archive/index.php/t-23288.html
--
Ryan Schultz
"Rise above oneself and grasp the world." -- Archimedes
Reply to: