avcodec/vp8dsp: add VP7 idct and loop filter

Signed-off-by: Peter Ross <pross@xvid.org>
Reviewed-by: "Ronald S. Bultje" <rsbultje@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavcodec/arm/vp8dsp.h

@@ -21,8 +21,8 @@
 
 #include "libavcodec/vp8dsp.h"
 
-void ff_vp8dsp_init_armv6(VP8DSPContext *dsp);
-void ff_vp8dsp_init_neon(VP8DSPContext *dsp);
+void ff_vp8dsp_init_armv6(VP8DSPContext *dsp, int vp7);
+void ff_vp8dsp_init_neon(VP8DSPContext *dsp, int vp7);
 
 #define VP8_LF_Y(hv, inner, opt)                                             \
     void ff_vp8_##hv##_loop_filter16##inner##_##opt(uint8_t *dst,            \

libavcodec/arm/vp8dsp_init_arm.c

@@ -23,12 +23,12 @@
 #include "libavcodec/vp8dsp.h"
 #include "vp8dsp.h"
 
-av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp)
+av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp, int vp7)
 {
     int cpu_flags = av_get_cpu_flags();
 
     if (have_armv6(cpu_flags))
-        ff_vp8dsp_init_armv6(dsp);
+        ff_vp8dsp_init_armv6(dsp, vp7);
     if (have_neon(cpu_flags))
-        ff_vp8dsp_init_neon(dsp);
+        ff_vp8dsp_init_neon(dsp, vp7);
 }

libavcodec/arm/vp8dsp_init_armv6.c

@@ -40,8 +40,9 @@ VP8_BILIN(16, armv6);
 VP8_BILIN(8,  armv6);
 VP8_BILIN(4,  armv6);
 
-av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp)
+av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp, int vp7)
 {
+    if (!vp7) {
     dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_armv6;
     dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6;
 
@@ -62,6 +63,7 @@ av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp)
 
     dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_armv6;
     dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_armv6;
+    }
 
     dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_armv6;
     dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_armv6;

libavcodec/arm/vp8dsp_init_neon.c

@@ -39,8 +39,9 @@ VP8_BILIN(16, neon);
 VP8_BILIN(8,  neon);
 VP8_BILIN(4,  neon);
 
-av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp)
+av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp, int vp7)
 {
+    if (!vp7) {
     dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_neon;
 
     dsp->vp8_idct_add       = ff_vp8_idct_add_neon;
@@ -60,6 +61,7 @@ av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp)
 
     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;

libavcodec/vp8.c

@@ -2010,7 +2010,7 @@ av_cold int ff_vp8_decode_init(AVCodecContext *avctx)
 
     ff_videodsp_init(&s->vdsp, 8);
     ff_h264_pred_init(&s->hpc, AV_CODEC_ID_VP8, 8, 1);
-    ff_vp8dsp_init(&s->vp8dsp);
+    ff_vp8dsp_init(&s->vp8dsp, 0);
 
     if ((ret = vp8_init_frames(s)) < 0) {
         ff_vp8_decode_free(avctx);

libavcodec/vp8dsp.h

@@ -88,9 +88,9 @@ void ff_put_vp8_pixels8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
 void ff_put_vp8_pixels4_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
                           int h, int x, int y);
 
-void ff_vp8dsp_init(VP8DSPContext *c);
-void ff_vp8dsp_init_x86(VP8DSPContext *c);
-void ff_vp8dsp_init_arm(VP8DSPContext *c);
+void ff_vp8dsp_init(VP8DSPContext *c, int vp7);
+void ff_vp8dsp_init_x86(VP8DSPContext *c, int vp7);
+void ff_vp8dsp_init_arm(VP8DSPContext *c, int vp7);
 void ff_vp8dsp_init_ppc(VP8DSPContext *c);
 
 #endif /* AVCODEC_VP8DSP_H */

libavcodec/x86/vp8dsp_init.c

@@ -315,18 +315,22 @@ DECLARE_LOOP_FILTER(sse4)
     c->put_vp8_bilinear_pixels_tab[IDX][2][2] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT
 
 
-av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
+av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c, int vp7)
 {
 #if HAVE_YASM
     int cpu_flags = av_get_cpu_flags();
 
     if (EXTERNAL_MMX(cpu_flags)) {
+        if (!vp7) {
         c->vp8_idct_dc_add    = ff_vp8_idct_dc_add_mmx;
         c->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_mmx;
+        }
 #if ARCH_X86_32
+        if (!vp7) {
         c->vp8_idct_dc_add4y  = ff_vp8_idct_dc_add4y_mmx;
         c->vp8_idct_add       = ff_vp8_idct_add_mmx;
         c->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_mmx;
+        }
         c->put_vp8_epel_pixels_tab[0][0][0]     =
         c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_mmx;
 #endif
@@ -334,6 +338,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         c->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_mmx;
 
 #if ARCH_X86_32
+        if (!vp7) {
         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmx;
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmx;
 
@@ -346,6 +351,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_mmx;
         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_mmx;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_mmx;
+        }
 #endif
     }
 
@@ -360,6 +366,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         VP8_BILINEAR_MC_FUNC(0, 16, mmxext);
         VP8_BILINEAR_MC_FUNC(1,  8, mmxext);
 
+        if (!vp7) {
         c->vp8_v_loop_filter_simple   = ff_vp8_v_loop_filter_simple_mmxext;
         c->vp8_h_loop_filter_simple   = ff_vp8_h_loop_filter_simple_mmxext;
 
@@ -372,12 +379,15 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_mmxext;
         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_mmxext;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_mmxext;
+        }
 #endif
     }
 
     if (EXTERNAL_SSE(cpu_flags)) {
+        if (!vp7) {
         c->vp8_idct_add                         = ff_vp8_idct_add_sse;
         c->vp8_luma_dc_wht                      = ff_vp8_luma_dc_wht_sse;
+        }
         c->put_vp8_epel_pixels_tab[0][0][0]     =
         c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse;
     }
@@ -388,6 +398,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         VP8_BILINEAR_MC_FUNC(0, 16, sse2);
         VP8_BILINEAR_MC_FUNC(1, 8, sse2);
 
+        if (!vp7) {
         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2;
 
         c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2;
@@ -395,9 +406,11 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
 
         c->vp8_v_loop_filter16y       = ff_vp8_v_loop_filter16y_mbedge_sse2;
         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_sse2;
+        }
     }
 
     if (EXTERNAL_SSE2(cpu_flags)) {
+        if (!vp7) {
         c->vp8_idct_dc_add4y          = ff_vp8_idct_dc_add4y_sse2;
 
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2;
@@ -407,6 +420,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
 
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_sse2;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_sse2;
+        }
     }
 
     if (EXTERNAL_SSSE3(cpu_flags)) {
@@ -417,6 +431,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         VP8_BILINEAR_MC_FUNC(1, 8, ssse3);
         VP8_BILINEAR_MC_FUNC(2, 4, ssse3);
 
+        if (!vp7) {
         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_ssse3;
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_ssse3;
 
@@ -429,14 +444,17 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_ssse3;
         c->vp8_v_loop_filter8uv       = ff_vp8_v_loop_filter8uv_mbedge_ssse3;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_ssse3;
+        }
     }
 
     if (EXTERNAL_SSE4(cpu_flags)) {
+        if (!vp7) {
         c->vp8_idct_dc_add                  = ff_vp8_idct_dc_add_sse4;
 
         c->vp8_h_loop_filter_simple   = ff_vp8_h_loop_filter_simple_sse4;
         c->vp8_h_loop_filter16y       = ff_vp8_h_loop_filter16y_mbedge_sse4;
         c->vp8_h_loop_filter8uv       = ff_vp8_h_loop_filter8uv_mbedge_sse4;
+        }
     }
 #endif /* HAVE_YASM */
 }