"Ronald S. Bultje" <[email protected]> writes:
> From: "Ronald S. Bultje" <[email protected]>
>
> ---
> libswscale/x86/swscale_mmx.c | 15 +++++++++------
> 1 file changed, 9 insertions(+), 6 deletions(-)
>
> diff --git a/libswscale/x86/swscale_mmx.c b/libswscale/x86/swscale_mmx.c
> index 99b3262..37d786a 100644
> --- a/libswscale/x86/swscale_mmx.c
> +++ b/libswscale/x86/swscale_mmx.c
> @@ -329,12 +329,12 @@ void ff_sws_init_swScale_mmx(SwsContext *c)
> case 8: ASSIGN_SCALE_FUNC2(hscalefn, 8, opt1, opt2); break; \
> default: ASSIGN_SCALE_FUNC2(hscalefn, X, opt1, opt2); break; \
> }
> -#define ASSIGN_VSCALEX_FUNC(vscalefn, opt, do_16_case) \
> +#define ASSIGN_VSCALEX_FUNC(vscalefn, opt, do_16_case, condition_8bit) \
> switch(c->dstBpc){ \
> case 16: do_16_case;
> break; \
> case 10: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2planeX_10_ ## opt;
> break; \
> case 9: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2planeX_9_ ## opt;
> break; \
> - default: vscalefn = ff_yuv2planeX_8_ ## opt;
> break; \
> + default: if (condition_8bit) vscalefn = ff_yuv2planeX_8_ ## opt;
> break; \
This makes me shiver. Then again, the original code does too.
> }
> #define ASSIGN_VSCALE_FUNC(vscalefn, opt1, opt2, opt2chk) \
> switch(c->dstBpc){ \
> @@ -386,7 +386,7 @@ switch(c->dstBpc){ \
> }
> }
> if (cpu_flags & AV_CPU_FLAG_MMX2) {
> - ASSIGN_VSCALEX_FUNC(c->yuv2planeX, mmx2,);
> + ASSIGN_VSCALEX_FUNC(c->yuv2planeX, mmx2, , 1);
> }
> #endif
> #define ASSIGN_SSE_SCALE_FUNC(hscalefn, filtersize, opt1, opt2) \
> @@ -400,7 +400,8 @@ switch(c->dstBpc){ \
> if (cpu_flags & AV_CPU_FLAG_SSE2) {
> ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, sse2, sse2);
> ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, sse2, sse2);
> - ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse2,);
> + ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse2, ,
> + HAVE_ALIGNED_STACK || ARCH_X86_64);
ARCH_X86_64 sets HAVE_ALIGNED_STACK, no need to check for the former
explicitly.
--
Måns Rullgård
[email protected]
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
