dsputil: Separate h264 qpel

The sh4 optimizations are removed, because the code is
100% identical to the C code, so it is unlikely to
provide any real practical benefit.

Signed-off-by: Diego Biurrun <diego@biurrun.de>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

3 files changed, 131 insertions, 89 deletions

diff --git a/libavcodec/x86/Makefile b/libavcodec/x86/Makefile
index 0bade86375..9b8b6531d0 100644
--- a/libavcodec/x86/Makefile
+++ b/libavcodec/x86/Makefile
@@ -7,6 +7,7 @@ OBJS-$(CONFIG_DNXHD_ENCODER)           += x86/dnxhdenc.o
 OBJS-$(CONFIG_FFT)                     += x86/fft_init.o
 OBJS-$(CONFIG_H264DSP)                 += x86/h264dsp_init.o
 OBJS-$(CONFIG_H264PRED)                += x86/h264_intrapred_init.o
+OBJS-$(CONFIG_H264QPEL)                += x86/h264_qpel.o
 OBJS-$(CONFIG_LPC)                     += x86/lpc.o
 OBJS-$(CONFIG_MLP_DECODER)             += x86/mlpdsp.o
 OBJS-$(CONFIG_MPEGAUDIODSP)            += x86/mpegaudiodec.o
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 2521b9f1cc..f9da04f6ea 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -1690,7 +1690,10 @@ static void gmc_mmx(uint8_t *dst, uint8_t *src,
 }
 #endif /* HAVE_INLINE_ASM */
 
-#include "h264_qpel.c"
+void ff_put_pixels16_sse2(uint8_t *block, const uint8_t *pixels,
+                          int line_size, int h);
+void ff_avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels,
+                          int line_size, int h);
 
 void ff_put_h264_chroma_mc8_rnd_mmx  (uint8_t *dst, uint8_t *src,
                                       int stride, int h, int x, int y);
@@ -1882,22 +1885,6 @@ void ff_vector_clip_int32_sse4    (int32_t *dst, const int32_t *src,
         c->PFX ## _pixels_tab IDX [3] = PFX ## _pixels ## SIZE ## _xy2_ ## CPU; \
     } while (0)
 
-#define H264_QPEL_FUNCS(x, y, CPU)                                                            \
-    do {                                                                                      \
-        c->put_h264_qpel_pixels_tab[0][x + y * 4] = put_h264_qpel16_mc ## x ## y ## _ ## CPU; \
-        c->put_h264_qpel_pixels_tab[1][x + y * 4] = put_h264_qpel8_mc  ## x ## y ## _ ## CPU; \
-        c->avg_h264_qpel_pixels_tab[0][x + y * 4] = avg_h264_qpel16_mc ## x ## y ## _ ## CPU; \
-        c->avg_h264_qpel_pixels_tab[1][x + y * 4] = avg_h264_qpel8_mc  ## x ## y ## _ ## CPU; \
-    } while (0)
-
-#define H264_QPEL_FUNCS_10(x, y, CPU)                                                               \
-    do {                                                                                            \
-        c->put_h264_qpel_pixels_tab[0][x + y * 4] = ff_put_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \
-        c->put_h264_qpel_pixels_tab[1][x + y * 4] = ff_put_h264_qpel8_mc  ## x ## y ## _10_ ## CPU; \
-        c->avg_h264_qpel_pixels_tab[0][x + y * 4] = ff_avg_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \
-        c->avg_h264_qpel_pixels_tab[1][x + y * 4] = ff_avg_h264_qpel8_mc  ## x ## y ## _10_ ## CPU; \
-    } while (0)
-
 static void dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx, int mm_flags)
 {
     const int high_bit_depth = avctx->bits_per_raw_sample > 8;
@@ -2014,26 +2001,6 @@ static void dsputil_init_mmxext(DSPContext *c, AVCodecContext *avctx,
 #endif /* HAVE_INLINE_ASM */
 
 #if HAVE_MMXEXT_EXTERNAL
-    if (CONFIG_H264QPEL) {
-        if (!high_bit_depth) {
-            SET_QPEL_FUNCS(put_h264_qpel, 0, 16, mmxext, );
-            SET_QPEL_FUNCS(put_h264_qpel, 1,  8, mmxext, );
-            SET_QPEL_FUNCS(put_h264_qpel, 2,  4, mmxext, );
-            SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, mmxext, );
-            SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, mmxext, );
-            SET_QPEL_FUNCS(avg_h264_qpel, 2,  4, mmxext, );
-        } else if (bit_depth == 10) {
-#if !ARCH_X86_64
-            SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_mmxext, ff_);
-            SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_mmxext, ff_);
-            SET_QPEL_FUNCS(put_h264_qpel, 1,  8, 10_mmxext, ff_);
-            SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, 10_mmxext, ff_);
-#endif
-            SET_QPEL_FUNCS(put_h264_qpel, 2, 4,  10_mmxext, ff_);
-            SET_QPEL_FUNCS(avg_h264_qpel, 2, 4,  10_mmxext, ff_);
-        }
-    }
-
     if (!high_bit_depth && CONFIG_H264CHROMA) {
         c->avg_h264_chroma_pixels_tab[0] = ff_avg_h264_chroma_mc8_rnd_mmxext;
         c->avg_h264_chroma_pixels_tab[1] = ff_avg_h264_chroma_mc4_mmxext;
@@ -2148,36 +2115,10 @@ static void dsputil_init_sse2(DSPContext *c, AVCodecContext *avctx,
             c->put_pixels_tab[0][0]        = ff_put_pixels16_sse2;
             c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_sse2;
             c->avg_pixels_tab[0][0]        = ff_avg_pixels16_sse2;
-            if (CONFIG_H264QPEL)
-                H264_QPEL_FUNCS(0, 0, sse2);
         }
     }
 
-    if (!high_bit_depth && CONFIG_H264QPEL) {
-        H264_QPEL_FUNCS(0, 1, sse2);
-        H264_QPEL_FUNCS(0, 2, sse2);
-        H264_QPEL_FUNCS(0, 3, sse2);
-        H264_QPEL_FUNCS(1, 1, sse2);
-        H264_QPEL_FUNCS(1, 2, sse2);
-        H264_QPEL_FUNCS(1, 3, sse2);
-        H264_QPEL_FUNCS(2, 1, sse2);
-        H264_QPEL_FUNCS(2, 2, sse2);
-        H264_QPEL_FUNCS(2, 3, sse2);
-        H264_QPEL_FUNCS(3, 1, sse2);
-        H264_QPEL_FUNCS(3, 2, sse2);
-        H264_QPEL_FUNCS(3, 3, sse2);
-    }
-
     if (bit_depth == 10) {
-        if (CONFIG_H264QPEL) {
-            SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_sse2, ff_);
-            SET_QPEL_FUNCS(put_h264_qpel, 1,  8, 10_sse2, ff_);
-            SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_sse2, ff_);
-            SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, 10_sse2, ff_);
-            H264_QPEL_FUNCS_10(1, 0, sse2_cache64);
-            H264_QPEL_FUNCS_10(2, 0, sse2_cache64);
-            H264_QPEL_FUNCS_10(3, 0, sse2_cache64);
-        }
         if (CONFIG_H264CHROMA) {
             c->put_h264_chroma_pixels_tab[0] = ff_put_h264_chroma_mc8_10_sse2;
             c->avg_h264_chroma_pixels_tab[0] = ff_avg_h264_chroma_mc8_10_sse2;
@@ -2205,27 +2146,7 @@ static void dsputil_init_ssse3(DSPContext *c, AVCodecContext *avctx,
 {
 #if HAVE_SSSE3_EXTERNAL
     const int high_bit_depth = avctx->bits_per_raw_sample > 8;
-    const int bit_depth      = avctx->bits_per_raw_sample;
 
-    if (!high_bit_depth && CONFIG_H264QPEL) {
-        H264_QPEL_FUNCS(1, 0, ssse3);
-        H264_QPEL_FUNCS(1, 1, ssse3);
-        H264_QPEL_FUNCS(1, 2, ssse3);
-        H264_QPEL_FUNCS(1, 3, ssse3);
-        H264_QPEL_FUNCS(2, 0, ssse3);
-        H264_QPEL_FUNCS(2, 1, ssse3);
-        H264_QPEL_FUNCS(2, 2, ssse3);
-        H264_QPEL_FUNCS(2, 3, ssse3);
-        H264_QPEL_FUNCS(3, 0, ssse3);
-        H264_QPEL_FUNCS(3, 1, ssse3);
-        H264_QPEL_FUNCS(3, 2, ssse3);
-        H264_QPEL_FUNCS(3, 3, ssse3);
-    }
-    if (bit_depth == 10 && CONFIG_H264QPEL) {
-        H264_QPEL_FUNCS_10(1, 0, ssse3_cache64);
-        H264_QPEL_FUNCS_10(2, 0, ssse3_cache64);
-        H264_QPEL_FUNCS_10(3, 0, ssse3_cache64);
-    }
     if (!high_bit_depth && CONFIG_H264CHROMA) {
         c->put_h264_chroma_pixels_tab[0] = ff_put_h264_chroma_mc8_rnd_ssse3;
         c->avg_h264_chroma_pixels_tab[0] = ff_avg_h264_chroma_mc8_rnd_ssse3;
@@ -2262,12 +2183,6 @@ static void dsputil_init_avx(DSPContext *c, AVCodecContext *avctx, int mm_flags)
     if (bit_depth == 10) {
         // AVX implies !cache64.
         // TODO: Port cache(32|64) detection from x264.
-        if (CONFIG_H264QPEL) {
-            H264_QPEL_FUNCS_10(1, 0, sse2);
-            H264_QPEL_FUNCS_10(2, 0, sse2);
-            H264_QPEL_FUNCS_10(3, 0, sse2);
-        }
-
         if (CONFIG_H264CHROMA) {
             c->put_h264_chroma_pixels_tab[0] = ff_put_h264_chroma_mc8_10_avx;
             c->avg_h264_chroma_pixels_tab[0] = ff_avg_h264_chroma_mc8_10_avx;
diff --git a/libavcodec/x86/h264_qpel.c b/libavcodec/x86/h264_qpel.c
index bc56d091e1..bebf5a5f3d 100644
--- a/libavcodec/x86/h264_qpel.c
+++ b/libavcodec/x86/h264_qpel.c
@@ -22,6 +22,7 @@
 #include "libavutil/cpu.h"
 #include "libavutil/x86/asm.h"
 #include "libavcodec/dsputil.h"
+#include "libavcodec/h264qpel.h"
 #include "libavcodec/mpegvideo.h"
 #include "dsputil_mmx.h"
 
@@ -490,3 +491,128 @@ QPEL16(mmxext)
 #endif
 
 #endif /* HAVE_YASM */
+
+#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX)                          \
+    do {                                                                     \
+    c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \
+    c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU; \
+    } while (0)
+
+#define H264_QPEL_FUNCS(x, y, CPU)                                                            \
+    do {                                                                                      \
+        c->put_h264_qpel_pixels_tab[0][x + y * 4] = put_h264_qpel16_mc ## x ## y ## _ ## CPU; \
+        c->put_h264_qpel_pixels_tab[1][x + y * 4] = put_h264_qpel8_mc  ## x ## y ## _ ## CPU; \
+        c->avg_h264_qpel_pixels_tab[0][x + y * 4] = avg_h264_qpel16_mc ## x ## y ## _ ## CPU; \
+        c->avg_h264_qpel_pixels_tab[1][x + y * 4] = avg_h264_qpel8_mc  ## x ## y ## _ ## CPU; \
+    } while (0)
+
+#define H264_QPEL_FUNCS_10(x, y, CPU)                                                               \
+    do {                                                                                            \
+        c->put_h264_qpel_pixels_tab[0][x + y * 4] = ff_put_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \
+        c->put_h264_qpel_pixels_tab[1][x + y * 4] = ff_put_h264_qpel8_mc  ## x ## y ## _10_ ## CPU; \
+        c->avg_h264_qpel_pixels_tab[0][x + y * 4] = ff_avg_h264_qpel16_mc ## x ## y ## _10_ ## CPU; \
+        c->avg_h264_qpel_pixels_tab[1][x + y * 4] = ff_avg_h264_qpel8_mc  ## x ## y ## _10_ ## CPU; \
+    } while (0)
+
+void ff_h264qpel_init_x86(H264QpelContext *c, int bit_depth)
+{
+    int high_bit_depth = bit_depth > 8;
+    int mm_flags = av_get_cpu_flags();
+
+#if HAVE_MMXEXT_EXTERNAL
+    if (!high_bit_depth) {
+        SET_QPEL_FUNCS(put_h264_qpel, 0, 16, mmxext, );
+        SET_QPEL_FUNCS(put_h264_qpel, 1,  8, mmxext, );
+        SET_QPEL_FUNCS(put_h264_qpel, 2,  4, mmxext, );
+        SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, mmxext, );
+        SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, mmxext, );
+        SET_QPEL_FUNCS(avg_h264_qpel, 2,  4, mmxext, );
+    } else if (bit_depth == 10) {
+#if !ARCH_X86_64
+        SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_mmxext, ff_);
+        SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_mmxext, ff_);
+        SET_QPEL_FUNCS(put_h264_qpel, 1,  8, 10_mmxext, ff_);
+        SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, 10_mmxext, ff_);
+#endif
+        SET_QPEL_FUNCS(put_h264_qpel, 2, 4,  10_mmxext, ff_);
+        SET_QPEL_FUNCS(avg_h264_qpel, 2, 4,  10_mmxext, ff_);
+    }
+#endif
+
+#if HAVE_SSE2_EXTERNAL
+    if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW) && !high_bit_depth) {
+        // these functions are slower than mmx on AMD, but faster on Intel
+        H264_QPEL_FUNCS(0, 0, sse2);
+    }
+
+    if (!high_bit_depth) {
+        H264_QPEL_FUNCS(0, 1, sse2);
+        H264_QPEL_FUNCS(0, 2, sse2);
+        H264_QPEL_FUNCS(0, 3, sse2);
+        H264_QPEL_FUNCS(1, 1, sse2);
+        H264_QPEL_FUNCS(1, 2, sse2);
+        H264_QPEL_FUNCS(1, 3, sse2);
+        H264_QPEL_FUNCS(2, 1, sse2);
+        H264_QPEL_FUNCS(2, 2, sse2);
+        H264_QPEL_FUNCS(2, 3, sse2);
+        H264_QPEL_FUNCS(3, 1, sse2);
+        H264_QPEL_FUNCS(3, 2, sse2);
+        H264_QPEL_FUNCS(3, 3, sse2);
+    }
+
+    if (bit_depth == 10) {
+        SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_sse2, ff_);
+        SET_QPEL_FUNCS(put_h264_qpel, 1,  8, 10_sse2, ff_);
+        SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_sse2, ff_);
+        SET_QPEL_FUNCS(avg_h264_qpel, 1,  8, 10_sse2, ff_);
+        H264_QPEL_FUNCS_10(1, 0, sse2_cache64);
+        H264_QPEL_FUNCS_10(2, 0, sse2_cache64);
+        H264_QPEL_FUNCS_10(3, 0, sse2_cache64);
+    }
+#endif
+
+#if HAVE_SSSE3_EXTERNAL
+    if (!high_bit_depth) {
+        H264_QPEL_FUNCS(1, 0, ssse3);
+        H264_QPEL_FUNCS(1, 1, ssse3);
+        H264_QPEL_FUNCS(1, 2, ssse3);
+        H264_QPEL_FUNCS(1, 3, ssse3);
+        H264_QPEL_FUNCS(2, 0, ssse3);
+        H264_QPEL_FUNCS(2, 1, ssse3);
+        H264_QPEL_FUNCS(2, 2, ssse3);
+        H264_QPEL_FUNCS(2, 3, ssse3);
+        H264_QPEL_FUNCS(3, 0, ssse3);
+        H264_QPEL_FUNCS(3, 1, ssse3);
+        H264_QPEL_FUNCS(3, 2, ssse3);
+        H264_QPEL_FUNCS(3, 3, ssse3);
+    }
+
+    if (bit_depth == 10) {
+        H264_QPEL_FUNCS_10(1, 0, ssse3_cache64);
+        H264_QPEL_FUNCS_10(2, 0, ssse3_cache64);
+        H264_QPEL_FUNCS_10(3, 0, ssse3_cache64);
+    }
+#endif
+
+#if HAVE_AVX_EXTERNAL
+    if (bit_depth == 10) {
+        H264_QPEL_FUNCS_10(1, 0, sse2);
+        H264_QPEL_FUNCS_10(2, 0, sse2);
+        H264_QPEL_FUNCS_10(3, 0, sse2);
+    }
+#endif
+}