pp: rework the way templating is done.

This refactoring simplifies the usage of the template: define the
profile and include the template is all that is required. It should now
be easier to add more instruction sets.

The HAVE_* flags are changed with TEMPLATE_PP_* setting to avoid messing
them up.

See the top comment in postprocess_template.c for details.

libpostproc/postprocess.c

@@ -538,85 +538,37 @@ static av_always_inline void do_a_deblock_C(uint8_t *src, int step, int stride,
 //Note: we have C, MMX, MMX2, 3DNOW version there is no 3DNOW+MMX2 one
 //Plain C versions
 //we always compile C for testing which needs bitexactness
-#define COMPILE_C
-
-#if HAVE_ALTIVEC
-#define COMPILE_ALTIVEC
-#endif //HAVE_ALTIVEC
-
-#if ARCH_X86 && HAVE_INLINE_ASM
-
-#if (HAVE_MMX_INLINE && !HAVE_AMD3DNOW_INLINE && !HAVE_MMXEXT_INLINE) || CONFIG_RUNTIME_CPUDETECT
-#define COMPILE_MMX
-#endif
-
-#if HAVE_MMXEXT_INLINE || CONFIG_RUNTIME_CPUDETECT
-#define COMPILE_MMX2
-#endif
-
-#if (HAVE_AMD3DNOW_INLINE && !HAVE_MMXEXT_INLINE) || CONFIG_RUNTIME_CPUDETECT
-#define COMPILE_3DNOW
-#endif
-#endif /* ARCH_X86 */
-
-#undef HAVE_MMX_INLINE
-#define HAVE_MMX_INLINE 0
-#undef HAVE_MMXEXT_INLINE
-#define HAVE_MMXEXT_INLINE 0
-#undef HAVE_AMD3DNOW_INLINE
-#define HAVE_AMD3DNOW_INLINE 0
-#undef HAVE_ALTIVEC
-#define HAVE_ALTIVEC 0
-
-#ifdef COMPILE_C
-#define RENAME(a) a ## _C
-#include "postprocess_template.c"
-#endif
-
-#ifdef COMPILE_ALTIVEC
-#undef RENAME
-#undef HAVE_ALTIVEC
-#define HAVE_ALTIVEC 1
-#define RENAME(a) a ## _altivec
-#include "postprocess_altivec_template.c"
-#include "postprocess_template.c"
-#endif
-
-//MMX versions
-#ifdef COMPILE_MMX
-#undef RENAME
-#undef HAVE_MMX_INLINE
-#define HAVE_MMX_INLINE 1
-#define RENAME(a) a ## _MMX
+#define TEMPLATE_PP_C 1
 #include "postprocess_template.c"
-#endif
 
-//MMX2 versions
-#ifdef COMPILE_MMX2
-#undef RENAME
-#undef HAVE_MMX_INLINE
-#undef HAVE_MMXEXT_INLINE
-#define HAVE_MMX_INLINE 1
-#define HAVE_MMXEXT_INLINE 1
-#define RENAME(a) a ## _MMX2
-#include "postprocess_template.c"
+#if HAVE_ALTIVEC
+#   define TEMPLATE_PP_ALTIVEC 1
+#   include "postprocess_altivec_template.c"
+#   include "postprocess_template.c"
 #endif
 
-//3DNOW versions
-#ifdef COMPILE_3DNOW
-#undef RENAME
-#undef HAVE_MMX_INLINE
-#undef HAVE_MMXEXT_INLINE
-#undef HAVE_AMD3DNOW_INLINE
-#define HAVE_MMX_INLINE 1
-#define HAVE_MMXEXT_INLINE 0
-#define HAVE_AMD3DNOW_INLINE 1
-#define RENAME(a) a ## _3DNow
-#include "postprocess_template.c"
+#if ARCH_X86 && HAVE_INLINE_ASM
+#    if CONFIG_RUNTIME_CPUDETECT
+#        define TEMPLATE_PP_MMX 1
+#        include "postprocess_template.c"
+#        define TEMPLATE_PP_MMXEXT 1
+#        include "postprocess_template.c"
+#        define TEMPLATE_PP_3DNOW 1
+#        include "postprocess_template.c"
+#    else
+#        if HAVE_MMXEXT_INLINE
+#            define TEMPLATE_PP_MMXEXT 1
+#            include "postprocess_template.c"
+#        elif HAVE_AMD3DNOW_INLINE
+#            define TEMPLATE_PP_3DNOW 1
+#            include "postprocess_template.c"
+#        elif HAVE_MMX_INLINE
+#            define TEMPLATE_PP_MMX 1
+#            include "postprocess_template.c"
+#        endif
+#    endif
 #endif
 
-// minor note: the HAVE_xyz is messed up after that line so do not use it.
-
 static inline void postProcess(const uint8_t src[], int srcStride, uint8_t dst[], int dstStride, int width, int height,
         const QP_STORE_T QPs[], int QPStride, int isColor, pp_mode *vm, pp_context *vc)
 {

libpostproc/postprocess_altivec_template.c

@@ -825,7 +825,7 @@ static inline void dering_altivec(uint8_t src[], int stride, PPContext *c) {
 #define doHorizDefFilter_altivec(a...) doHorizDefFilter_C(a)
 #define do_a_deblock_altivec(a...) do_a_deblock_C(a)
 
-static inline void RENAME(tempNoiseReducer)(uint8_t *src, int stride,
+static inline void tempNoiseReducer_altivec(uint8_t *src, int stride,
                                             uint8_t *tempBlurred, uint32_t *tempBlurredPast, int *maxNoise)
 {
     const vector signed char neg1 = vec_splat_s8(-1);