vp8: arm: separate ARMv6 functions from NEON

This is a preparation for complete ARMv6 optimisations.

Signed-off-by: Mans Rullgard <mans@mansr.com>

1 files changed, 5 insertions, 139 deletions

diff --git a/libavcodec/arm/vp8dsp_init_arm.c b/libavcodec/arm/vp8dsp_init_arm.c
index 29c64d4079..603f68cd24 100644
--- a/libavcodec/arm/vp8dsp_init_arm.c
+++ b/libavcodec/arm/vp8dsp_init_arm.c
@@ -20,148 +20,14 @@
 
 #include "libavutil/arm/cpu.h"
 #include "libavcodec/vp8dsp.h"
-
-void ff_vp8_luma_dc_wht_neon(DCTELEM block[4][4][16], DCTELEM dc[16]);
-void ff_vp8_luma_dc_wht_dc_neon(DCTELEM block[4][4][16], DCTELEM dc[16]);
-
-void ff_vp8_idct_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add4y_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add4uv_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
-
-void ff_vp8_v_loop_filter16_neon(uint8_t *dst, ptrdiff_t stride,
-                                 int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_h_loop_filter16_neon(uint8_t *dst, ptrdiff_t stride,
-                                 int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_v_loop_filter8uv_neon(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
-                                  int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_h_loop_filter8uv_neon(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
-                                  int flim_E, int flim_I, int hev_thresh);
-
-void ff_vp8_v_loop_filter16_inner_neon(uint8_t *dst, ptrdiff_t stride,
-                                       int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_h_loop_filter16_inner_neon(uint8_t *dst, ptrdiff_t stride,
-                                       int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_v_loop_filter8uv_inner_neon(uint8_t *dstU, uint8_t *dstV,
-                                        ptrdiff_t stride, int flim_E, int flim_I,
-                                        int hev_thresh);
-void ff_vp8_h_loop_filter8uv_inner_neon(uint8_t *dstU, uint8_t *dstV,
-                                        ptrdiff_t stride, int flim_E, int flim_I,
-                                        int hev_thresh);
-
-void ff_vp8_v_loop_filter16_simple_neon(uint8_t *dst, ptrdiff_t stride, int flim);
-void ff_vp8_h_loop_filter16_simple_neon(uint8_t *dst, ptrdiff_t stride, int flim);
-
-
-#define VP8_MC(n)                                                       \
-    void ff_put_vp8_##n##_neon(uint8_t *dst, ptrdiff_t dststride,       \
-                               uint8_t *src, ptrdiff_t srcstride,       \
-                               int h, int x, int y)
-
-#define VP8_EPEL(w)                             \
-    VP8_MC(pixels ## w);                        \
-    VP8_MC(epel ## w ## _h4);                   \
-    VP8_MC(epel ## w ## _h6);                   \
-    VP8_MC(epel ## w ## _v4);                   \
-    VP8_MC(epel ## w ## _h4v4);                 \
-    VP8_MC(epel ## w ## _h6v4);                 \
-    VP8_MC(epel ## w ## _v6);                   \
-    VP8_MC(epel ## w ## _h4v6);                 \
-    VP8_MC(epel ## w ## _h6v6)
-
-VP8_EPEL(16);
-VP8_EPEL(8);
-VP8_EPEL(4);
-
-VP8_MC(bilin16_h);
-VP8_MC(bilin16_v);
-VP8_MC(bilin16_hv);
-VP8_MC(bilin8_h);
-VP8_MC(bilin8_v);
-VP8_MC(bilin8_hv);
-VP8_MC(bilin4_h);
-VP8_MC(bilin4_v);
-VP8_MC(bilin4_hv);
+#include "vp8dsp.h"
 
 av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp)
 {
     int cpu_flags = av_get_cpu_flags();
 
-    if (have_neon(cpu_flags)) {
-        dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_neon;
-        dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_neon;
-
-        dsp->vp8_idct_add       = ff_vp8_idct_add_neon;
-        dsp->vp8_idct_dc_add    = ff_vp8_idct_dc_add_neon;
-        dsp->vp8_idct_dc_add4y  = ff_vp8_idct_dc_add4y_neon;
-        dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_neon;
-
-        dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_neon;
-        dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_neon;
-        dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_neon;
-        dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_neon;
-
-        dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_neon;
-        dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_neon;
-        dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_neon;
-        dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_neon;
-
-        dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_neon;
-        dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_neon;
-
-        dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_neon;
-        dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_neon;
-        dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_neon;
-        dsp->put_vp8_epel_pixels_tab[0][2][2] = ff_put_vp8_epel16_h6v6_neon;
-
-        dsp->put_vp8_epel_pixels_tab[1][0][0] = ff_put_vp8_pixels8_neon;
-        dsp->put_vp8_epel_pixels_tab[1][0][1] = ff_put_vp8_epel8_h4_neon;
-        dsp->put_vp8_epel_pixels_tab[1][0][2] = ff_put_vp8_epel8_h6_neon;
-        dsp->put_vp8_epel_pixels_tab[1][1][0] = ff_put_vp8_epel8_v4_neon;
-        dsp->put_vp8_epel_pixels_tab[1][1][1] = ff_put_vp8_epel8_h4v4_neon;
-        dsp->put_vp8_epel_pixels_tab[1][1][2] = ff_put_vp8_epel8_h6v4_neon;
-        dsp->put_vp8_epel_pixels_tab[1][2][0] = ff_put_vp8_epel8_v6_neon;
-        dsp->put_vp8_epel_pixels_tab[1][2][1] = ff_put_vp8_epel8_h4v6_neon;
-        dsp->put_vp8_epel_pixels_tab[1][2][2] = ff_put_vp8_epel8_h6v6_neon;
-
-        dsp->put_vp8_epel_pixels_tab[2][0][0] = ff_put_vp8_pixels4_neon;
-        dsp->put_vp8_epel_pixels_tab[2][0][1] = ff_put_vp8_epel4_h4_neon;
-        dsp->put_vp8_epel_pixels_tab[2][0][2] = ff_put_vp8_epel4_h6_neon;
-        dsp->put_vp8_epel_pixels_tab[2][1][0] = ff_put_vp8_epel4_v4_neon;
-        dsp->put_vp8_epel_pixels_tab[2][1][1] = ff_put_vp8_epel4_h4v4_neon;
-        dsp->put_vp8_epel_pixels_tab[2][1][2] = ff_put_vp8_epel4_h6v4_neon;
-        dsp->put_vp8_epel_pixels_tab[2][2][0] = ff_put_vp8_epel4_v6_neon;
-        dsp->put_vp8_epel_pixels_tab[2][2][1] = ff_put_vp8_epel4_h4v6_neon;
-        dsp->put_vp8_epel_pixels_tab[2][2][2] = ff_put_vp8_epel4_h6v6_neon;
-
-        dsp->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][0][1] = ff_put_vp8_bilin16_h_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][0][2] = ff_put_vp8_bilin16_h_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][1][0] = ff_put_vp8_bilin16_v_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][1][1] = ff_put_vp8_bilin16_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][1][2] = ff_put_vp8_bilin16_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][2][0] = ff_put_vp8_bilin16_v_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][2][1] = ff_put_vp8_bilin16_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[0][2][2] = ff_put_vp8_bilin16_hv_neon;
-
-        dsp->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][0][1] = ff_put_vp8_bilin8_h_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][0][2] = ff_put_vp8_bilin8_h_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][1][0] = ff_put_vp8_bilin8_v_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][1][1] = ff_put_vp8_bilin8_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][1][2] = ff_put_vp8_bilin8_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][2][0] = ff_put_vp8_bilin8_v_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][2][1] = ff_put_vp8_bilin8_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[1][2][2] = ff_put_vp8_bilin8_hv_neon;
-
-        dsp->put_vp8_bilinear_pixels_tab[2][0][0] = ff_put_vp8_pixels4_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][0][1] = ff_put_vp8_bilin4_h_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][0][2] = ff_put_vp8_bilin4_h_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][1][0] = ff_put_vp8_bilin4_v_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][1][1] = ff_put_vp8_bilin4_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][1][2] = ff_put_vp8_bilin4_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][2][0] = ff_put_vp8_bilin4_v_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_neon;
-        dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_neon;
-    }
+    if (have_armv6(cpu_flags))
+        ff_vp8dsp_init_armv6(dsp);
+    if (have_neon(cpu_flags))
+        ff_vp8dsp_init_neon(dsp);
 }