[XviD-devel] Linked errors cross-compiling for x64 Windows on x64 Linux

[Leo Izen]

On Wed, Feb 16, 2011 at 5:04 PM, Michael Militzer <michael at xvid.org> wrote:

> Hi,
>
> you can prevent the leading underscores by adding "-DNO_PREFIX" to your
> yasm/nasm command line.
>
>
I did that, and it build except for errors with:

  L: libxvidcore.dll
Creating library file: libxvidcore.dll.a
/usr/bin/../lib/gcc/x86_64-w64-mingw32/4.6.0/../../../../x86_64-w64-mingw32/lib/../lib/libmingw32.a(lib64_libmingw32_a-crt0_c.o):
In function `main':
/buildbot/mingw-w64/linux-x86_64-x86_64/build/build/mingw/obj/../../../build/mingw/mingw-w64-crt/crt/crt0_c.c:18:
undefined reference to `WinMain'
collect2: ld returned 1 exit status
make: *** [libxvidcore.dll] Error 1

Why is this?

Regards,
> Michael
>
>
> Quoting Leo Izen <leo.izen at gmail.com>:
>
> > On Wed, Feb 16, 2011 at 8:03 AM, Michael Militzer <michael at xvid.org>
> wrote:
> >
> >> Hi,
> >>
> >> all your undefined references are symbols defined in assembler .asm
> files.
> >> So I guess something did go wrong assembling these files (or the object
> >> output is broken etc.).
> >>
> >
> > Problem discovered: all the assembly symbols are prefixed by and
> underscore:
> > i.e.
> >
> > [Leo at chessman =build]$ x86_64-w64-mingw32-nm
> > quant/x86_asm/quantize_h263_mmx.o
> > 0000000000000000 r .rodata
> > 0000000000000000 t .rotext
> > 0000000000000000 t .text
> > 0000000000000840 T _dequant_h263_inter_mmx
> > 000000000000087e t _dequant_h263_inter_mmx.loop
> > 00000000000009c0 T _dequant_h263_inter_sse2
> > 0000000000000a1d t _dequant_h263_inter_sse2.loop
> > 0000000000000900 T _dequant_h263_inter_xmm
> > 0000000000000940 t _dequant_h263_inter_xmm.loop
> > 0000000000000580 T _dequant_h263_intra_mmx
> > 00000000000005be t _dequant_h263_intra_mmx.loop
> > 0000000000000720 T _dequant_h263_intra_sse2
> > 0000000000000777 t _dequant_h263_intra_sse2.loop
> > 0000000000000660 T _dequant_h263_intra_xmm
> > 000000000000069c t _dequant_h263_intra_xmm.loop
> > 00000000000002c0 T _quant_h263_inter_mmx
> > 0000000000000356 t _quant_h263_inter_mmx.done
> > 0000000000000300 t _quant_h263_inter_mmx.loop
> > 0000000000000380 t _quant_h263_inter_mmx.q1loop
> > 000000000000036b t _quant_h263_inter_mmx.q1routine
> > 00000000000003e0 T _quant_h263_inter_sse2
> > 00000000000004ac t _quant_h263_inter_sse2.qes2_done
> > 0000000000000440 t _quant_h263_inter_sse2.qes2_loop
> > 000000000000040f t _quant_h263_inter_sse2.qes2_not1
> > 00000000000004dd t _quant_h263_inter_sse2.qes2_q1_routine
> > 0000000000000500 t _quant_h263_inter_sse2.qes2_q1loop
> > 0000000000000000 T _quant_h263_intra_mmx
> > 0000000000000120 t _quant_h263_intra_mmx.end
> > 000000000000004e t _quant_h263_intra_mmx.loop
> > 00000000000000c1 t _quant_h263_intra_mmx.loop_low
> > 00000000000000af t _quant_h263_intra_mmx.low
> > 0000000000000140 T _quant_h263_intra_sse2
> > 00000000000002a5 t _quant_h263_intra_sse2.end
> > 000000000000019f t _quant_h263_intra_sse2.loop
> > 000000000000022e t _quant_h263_intra_sse2.loop_low
> > 000000000000021b t _quant_h263_intra_sse2.low
> > 0000000000000220 r mmx_div
> > 0000000000000020 r mmx_quant
> > 0000000000000120 r mmx_sub
> > 0000000000000000 r plus_one
> > [Leo at chessman =build]$
> >
> >
> >>
> >> Which assembler do you use? yasm or nasm? Make sure you have the latest
> >> (stable) version installed. If you already have the latest version, try
> >> yasm instead of nasm (or vice-versa).
> >>
> >>
> >  I am using yasm, from my package manager and the latest cvs of xvidcore.
> > However, seeing that I figured out that there's underscores above, I
> think a
> > better solution would be an addition to AFLAGS to not use an underscore.
> My
> > current AFLAGS do use -f win64 and the needed stuff to assemble for
> > x86_64-w64-mingw32, and from the nm above, the object files are valid.
> >
> > So...  is there an AFLAG that will let me remove the _prefix?
> >
> >
> >> I have successfully cross-compiled xvidcore to win64 with both yasm and
> >> nasm, but with older gcc (not 4.6). But it doesn't look like gcc is the
> >> problem here...
> >>
> >> Regards,
> >> Michael
> >>
> >>
>
>
>
> _______________________________________________
> Xvid-devel mailing list
> Xvid-devel at xvid.org
> http://list.xvid.org/mailman/listinfo/xvid-devel
>