On2 VP7 decoder

Further performance improvements and security fixes by
Vittorio Giovara, Luca Barbato and Diego Biurrun.

Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Signed-off-by: Diego Biurrun <diego@biurrun.de>

Changelog

@@ -13,6 +13,7 @@ version <next>:
 - OpenEXR image decoder
 - support encoding and decoding 4-channel SGI images
 - support decoding 16-bit RLE SGI images
+- VP7 video decoder
 
 
 version 10:

configure

@@ -1847,6 +1847,7 @@ vp5_decoder_select="h264chroma hpeldsp videodsp vp3dsp"
 vp6_decoder_select="h264chroma hpeldsp huffman videodsp vp3dsp"
 vp6a_decoder_select="vp6_decoder"
 vp6f_decoder_select="vp6_decoder"
+vp7_decoder_select="h264pred videodsp"
 vp8_decoder_select="h264pred videodsp"
 vp9_decoder_select="videodsp"
 webp_decoder_select="vp8_decoder"

doc/general.texi

@@ -650,6 +650,8 @@ following image formats are supported:
     @tab fourcc: VP50
 @item On2 VP6                @tab     @tab  X
     @tab fourcc: VP60,VP61,VP62
+@item On2 VP7                @tab     @tab  X
+    @tab fourcc: VP70,VP71
 @item VP8                    @tab  E  @tab  X
     @tab fourcc: VP80, encoding supported through external library libvpx
 @item VP9                    @tab  E  @tab  X

libavcodec/Makefile

@@ -386,6 +386,7 @@ OBJS-$(CONFIG_VP5_DECODER)             += vp5.o vp56.o vp56data.o vp56dsp.o \
                                           vp56rac.o
 OBJS-$(CONFIG_VP6_DECODER)             += vp6.o vp56.o vp56data.o vp56dsp.o \
                                           vp6dsp.o vp56rac.o
+OBJS-$(CONFIG_VP7_DECODER)             += vp8.o vp8dsp.o vp56rac.o
 OBJS-$(CONFIG_VP8_DECODER)             += vp8.o vp8dsp.o vp56rac.o
 OBJS-$(CONFIG_VP9_DECODER)             += vp9.o vp9data.o vp9dsp.o \
                                           vp9block.o vp9prob.o vp9mvs.o vp56rac.o

libavcodec/allcodecs.c

@@ -261,6 +261,7 @@ void avcodec_register_all(void)
     REGISTER_DECODER(VP6,               vp6);
     REGISTER_DECODER(VP6A,              vp6a);
     REGISTER_DECODER(VP6F,              vp6f);
+    REGISTER_DECODER(VP7,               vp7);
     REGISTER_DECODER(VP8,               vp8);
     REGISTER_DECODER(VP9,               vp9);
     REGISTER_DECODER(VQA,               vqa);

libavcodec/arm/Makefile

@@ -31,6 +31,7 @@ OBJS-$(CONFIG_MLP_DECODER)             += arm/mlpdsp_init_arm.o
 OBJS-$(CONFIG_VC1_DECODER)             += arm/vc1dsp_init_arm.o
 OBJS-$(CONFIG_VORBIS_DECODER)          += arm/vorbisdsp_init_arm.o
 OBJS-$(CONFIG_VP6_DECODER)             += arm/vp6dsp_init_arm.o
+OBJS-$(CONFIG_VP7_DECODER)             += arm/vp8dsp_init_arm.o
 OBJS-$(CONFIG_VP8_DECODER)             += arm/vp8dsp_init_arm.o
 OBJS-$(CONFIG_RV30_DECODER)            += arm/rv34dsp_init_arm.o
 OBJS-$(CONFIG_RV40_DECODER)            += arm/rv34dsp_init_arm.o        \
@@ -55,6 +56,9 @@ ARMV6-OBJS-$(CONFIG_HPELDSP)           += arm/hpeldsp_init_armv6.o      \
 ARMV6-OBJS-$(CONFIG_MPEGAUDIODSP)      += arm/mpegaudiodsp_fixed_armv6.o
 
 ARMV6-OBJS-$(CONFIG_MLP_DECODER)       += arm/mlpdsp_armv6.o
+ARMV6-OBJS-$(CONFIG_VP7_DECODER)       += arm/vp8_armv6.o               \
+                                          arm/vp8dsp_init_armv6.o       \
+                                          arm/vp8dsp_armv6.o
 ARMV6-OBJS-$(CONFIG_VP8_DECODER)       += arm/vp8_armv6.o               \
                                           arm/vp8dsp_init_armv6.o       \
                                           arm/vp8dsp_armv6.o
@@ -102,5 +106,7 @@ NEON-OBJS-$(CONFIG_VC1_DECODER)        += arm/vc1dsp_init_neon.o        \
                                           arm/vc1dsp_neon.o
 NEON-OBJS-$(CONFIG_VORBIS_DECODER)     += arm/vorbisdsp_neon.o
 NEON-OBJS-$(CONFIG_VP6_DECODER)        += arm/vp6dsp_neon.o
+NEON-OBJS-$(CONFIG_VP7_DECODER)        += arm/vp8dsp_init_neon.o        \
+                                          arm/vp8dsp_neon.o
 NEON-OBJS-$(CONFIG_VP8_DECODER)        += arm/vp8dsp_init_neon.o        \
                                           arm/vp8dsp_neon.o

libavcodec/arm/h264pred_init_arm.c

@@ -56,10 +56,11 @@ static av_cold void h264_pred_init_neon(H264PredContext *h, int codec_id,
 
     h->pred8x8[VERT_PRED8x8     ] = ff_pred8x8_vert_neon;
     h->pred8x8[HOR_PRED8x8      ] = ff_pred8x8_hor_neon;
-    if (codec_id != AV_CODEC_ID_VP8)
+    if (codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8)
         h->pred8x8[PLANE_PRED8x8] = ff_pred8x8_plane_neon;
     h->pred8x8[DC_128_PRED8x8   ] = ff_pred8x8_128_dc_neon;
-    if (codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP8) {
+    if (codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7 &&
+        codec_id != AV_CODEC_ID_VP8) {
         h->pred8x8[DC_PRED8x8     ] = ff_pred8x8_dc_neon;
         h->pred8x8[LEFT_DC_PRED8x8] = ff_pred8x8_left_dc_neon;
         h->pred8x8[TOP_DC_PRED8x8 ] = ff_pred8x8_top_dc_neon;
@@ -75,7 +76,8 @@ static av_cold void h264_pred_init_neon(H264PredContext *h, int codec_id,
     h->pred16x16[LEFT_DC_PRED8x8] = ff_pred16x16_left_dc_neon;
     h->pred16x16[TOP_DC_PRED8x8 ] = ff_pred16x16_top_dc_neon;
     h->pred16x16[DC_128_PRED8x8 ] = ff_pred16x16_128_dc_neon;
-    if (codec_id != AV_CODEC_ID_SVQ3 && codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP8)
+    if (codec_id != AV_CODEC_ID_SVQ3 && codec_id != AV_CODEC_ID_RV40 &&
+        codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8)
         h->pred16x16[PLANE_PRED8x8  ] = ff_pred16x16_plane_neon;
 }
 

libavcodec/arm/vp8dsp.h

@@ -21,6 +21,8 @@
 
 #include "libavcodec/vp8dsp.h"
 
+void ff_vp78dsp_init_armv6(VP8DSPContext *dsp);
+void ff_vp78dsp_init_neon(VP8DSPContext *dsp);
 void ff_vp8dsp_init_armv6(VP8DSPContext *dsp);
 void ff_vp8dsp_init_neon(VP8DSPContext *dsp);
 

libavcodec/arm/vp8dsp_init_arm.c

@@ -23,6 +23,16 @@
 #include "libavcodec/vp8dsp.h"
 #include "vp8dsp.h"
 
+av_cold void ff_vp78dsp_init_arm(VP8DSPContext *dsp)
+{
+    int cpu_flags = av_get_cpu_flags();
+
+    if (have_armv6(cpu_flags))
+        ff_vp78dsp_init_armv6(dsp);
+    if (have_neon(cpu_flags))
+        ff_vp78dsp_init_neon(dsp);
+}
+
 av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp)
 {
     int cpu_flags = av_get_cpu_flags();

libavcodec/arm/vp8dsp_init_armv6.c

@@ -40,29 +40,8 @@ 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_vp78dsp_init_armv6(VP8DSPContext *dsp)
 {
-    dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_armv6;
-    dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6;
-
-    dsp->vp8_idct_add       = ff_vp8_idct_add_armv6;
-    dsp->vp8_idct_dc_add    = ff_vp8_idct_dc_add_armv6;
-    dsp->vp8_idct_dc_add4y  = ff_vp8_idct_dc_add4y_armv6;
-    dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_armv6;
-
-    dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_armv6;
-    dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_armv6;
-    dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_armv6;
-    dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_armv6;
-
-    dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_armv6;
-    dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_armv6;
-    dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_armv6;
-    dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_armv6;
-
-    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;
     dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_armv6;
@@ -118,3 +97,27 @@ av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp)
     dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_armv6;
     dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_armv6;
 }
+
+av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp)
+{
+    dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_armv6;
+    dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6;
+
+    dsp->vp8_idct_add       = ff_vp8_idct_add_armv6;
+    dsp->vp8_idct_dc_add    = ff_vp8_idct_dc_add_armv6;
+    dsp->vp8_idct_dc_add4y  = ff_vp8_idct_dc_add4y_armv6;
+    dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_armv6;
+
+    dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_armv6;
+    dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_armv6;
+    dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_armv6;
+    dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_armv6;
+
+    dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_armv6;
+    dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_armv6;
+    dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_armv6;
+    dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_armv6;
+
+    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;
+}

libavcodec/arm/vp8dsp_init_neon.c

@@ -39,28 +39,8 @@ 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_vp78dsp_init_neon(VP8DSPContext *dsp)
 {
-    dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_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;
@@ -114,3 +94,26 @@ av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp)
     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;
 }
+
+av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp)
+{
+    dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_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;
+}

libavcodec/avcodec.h

@@ -288,6 +288,7 @@ enum AVCodecID {
     AV_CODEC_ID_BRENDER_PIX,
     AV_CODEC_ID_PAF_VIDEO,
     AV_CODEC_ID_EXR,
+    AV_CODEC_ID_VP7,
 
     /* various PCM "codecs" */
     AV_CODEC_ID_FIRST_AUDIO = 0x10000,     ///< A dummy id pointing at the start of audio codecs

libavcodec/codec_desc.c

@@ -1099,6 +1099,13 @@ static const AVCodecDescriptor codec_descriptors[] = {
         .long_name = NULL_IF_CONFIG_SMALL("Amazing Studio Packed Animation File Video"),
         .props     = AV_CODEC_PROP_LOSSY,
     },
+    {
+        .id        = AV_CODEC_ID_VP7,
+        .type      = AVMEDIA_TYPE_VIDEO,
+        .name      = "vp7",
+        .long_name = NULL_IF_CONFIG_SMALL("On2 VP7"),
+        .props     = AV_CODEC_PROP_LOSSY,
+    },
 
     /* image codecs */
     {

libavcodec/h264pred.c

@@ -410,7 +410,7 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
 
 #define H264_PRED(depth) \
     if(codec_id != AV_CODEC_ID_RV40){\
-        if(codec_id == AV_CODEC_ID_VP8) {\
+        if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {\
             h->pred4x4[VERT_PRED       ]= FUNCD(pred4x4_vertical_vp8);\
             h->pred4x4[HOR_PRED        ]= FUNCD(pred4x4_horizontal_vp8);\
         } else {\
@@ -425,15 +425,14 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
         h->pred4x4[DIAG_DOWN_RIGHT_PRED]= FUNCC(pred4x4_down_right        , depth);\
         h->pred4x4[VERT_RIGHT_PRED     ]= FUNCC(pred4x4_vertical_right    , depth);\
         h->pred4x4[HOR_DOWN_PRED       ]= FUNCC(pred4x4_horizontal_down   , depth);\
-        if (codec_id == AV_CODEC_ID_VP8) {\
+        if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {\
             h->pred4x4[VERT_LEFT_PRED  ]= FUNCD(pred4x4_vertical_left_vp8);\
         } else\
             h->pred4x4[VERT_LEFT_PRED  ]= FUNCC(pred4x4_vertical_left     , depth);\
         h->pred4x4[HOR_UP_PRED         ]= FUNCC(pred4x4_horizontal_up     , depth);\
-        if(codec_id != AV_CODEC_ID_VP8) {\
+        if (codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8) {\
             h->pred4x4[LEFT_DC_PRED    ]= FUNCC(pred4x4_left_dc           , depth);\
             h->pred4x4[TOP_DC_PRED     ]= FUNCC(pred4x4_top_dc            , depth);\
-            h->pred4x4[DC_128_PRED     ]= FUNCC(pred4x4_128_dc            , depth);\
         } else {\
             h->pred4x4[TM_VP8_PRED     ]= FUNCD(pred4x4_tm_vp8);\
             h->pred4x4[DC_127_PRED     ]= FUNCC(pred4x4_127_dc            , depth);\
@@ -441,6 +440,8 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
             h->pred4x4[VERT_VP8_PRED   ]= FUNCC(pred4x4_vertical          , depth);\
             h->pred4x4[HOR_VP8_PRED    ]= FUNCC(pred4x4_horizontal        , depth);\
         }\
+        if (codec_id != AV_CODEC_ID_VP8)\
+            h->pred4x4[DC_128_PRED     ]= FUNCC(pred4x4_128_dc            , depth);\
     }else{\
         h->pred4x4[VERT_PRED           ]= FUNCC(pred4x4_vertical          , depth);\
         h->pred4x4[HOR_PRED            ]= FUNCC(pred4x4_horizontal        , depth);\
@@ -479,7 +480,7 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
         h->pred8x8[VERT_PRED8x8   ]= FUNCC(pred8x16_vertical              , depth);\
         h->pred8x8[HOR_PRED8x8    ]= FUNCC(pred8x16_horizontal            , depth);\
     }\
-    if (codec_id != AV_CODEC_ID_VP8) {\
+    if (codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8) {\
         if (chroma_format_idc <= 1) {\
             h->pred8x8[PLANE_PRED8x8]= FUNCC(pred8x8_plane                , depth);\
         } else {\
@@ -487,7 +488,8 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
         }\
     } else\
         h->pred8x8[PLANE_PRED8x8]= FUNCD(pred8x8_tm_vp8);\
-    if(codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP8){\
+    if (codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7 && \
+        codec_id != AV_CODEC_ID_VP8) {\
         if (chroma_format_idc <= 1) {\
             h->pred8x8[DC_PRED8x8     ]= FUNCC(pred8x8_dc                     , depth);\
             h->pred8x8[LEFT_DC_PRED8x8]= FUNCC(pred8x8_left_dc                , depth);\
@@ -509,7 +511,7 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
         h->pred8x8[DC_PRED8x8     ]= FUNCD(pred8x8_dc_rv40);\
         h->pred8x8[LEFT_DC_PRED8x8]= FUNCD(pred8x8_left_dc_rv40);\
         h->pred8x8[TOP_DC_PRED8x8 ]= FUNCD(pred8x8_top_dc_rv40);\
-        if (codec_id == AV_CODEC_ID_VP8) {\
+        if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {\
             h->pred8x8[DC_127_PRED8x8]= FUNCC(pred8x8_127_dc              , depth);\
             h->pred8x8[DC_129_PRED8x8]= FUNCC(pred8x8_129_dc              , depth);\
         }\
@@ -530,6 +532,7 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
     case AV_CODEC_ID_RV40:\
        h->pred16x16[PLANE_PRED8x8  ]= FUNCD(pred16x16_plane_rv40);\
        break;\
+    case AV_CODEC_ID_VP7:\
     case AV_CODEC_ID_VP8:\
        h->pred16x16[PLANE_PRED8x8  ]= FUNCD(pred16x16_tm_vp8);\
        h->pred16x16[DC_127_PRED8x8]= FUNCC(pred16x16_127_dc               , depth);\

libavcodec/h264pred.h

@@ -60,7 +60,7 @@
 #define VERT_VP8_PRED         10    ///< for VP8, #VERT_PRED is the average of
                                     ///< (left col+cur col x2+right col) / 4;
                                     ///< this is the "unaveraged" one
-#define HOR_VP8_PRED          11    ///< unaveraged version of #HOR_PRED, see
+#define HOR_VP8_PRED          14    ///< unaveraged version of #HOR_PRED, see
                                     ///< #VERT_VP8_PRED for details
 #define DC_127_PRED           12
 #define DC_129_PRED           13

libavcodec/ppc/Makefile

@@ -13,6 +13,7 @@ OBJS-$(CONFIG_VP3DSP)                  += ppc/vp3dsp_altivec.o
 
 OBJS-$(CONFIG_VC1_DECODER)             += ppc/vc1dsp_altivec.o
 OBJS-$(CONFIG_VORBIS_DECODER)          += ppc/vorbisdsp_altivec.o
+OBJS-$(CONFIG_VP7_DECODER)             += ppc/vp8dsp_altivec.o
 OBJS-$(CONFIG_VP8_DECODER)             += ppc/vp8dsp_altivec.o
 
 ALTIVEC-OBJS-$(CONFIG_DSPUTIL)         += ppc/dsputil_altivec.o         \

libavcodec/ppc/vp8dsp_altivec.c

@@ -311,7 +311,8 @@ static void put_vp8_pixels16_altivec(uint8_t *dst, ptrdiff_t dstride, uint8_t *s
 
 #endif /* HAVE_ALTIVEC */
 
-av_cold void ff_vp8dsp_init_ppc(VP8DSPContext *c)
+
+av_cold void ff_vp78dsp_init_ppc(VP8DSPContext *c)
 {
 #if HAVE_ALTIVEC
     if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))

libavcodec/version.h

@@ -29,7 +29,7 @@
 #include "libavutil/version.h"
 
 #define LIBAVCODEC_VERSION_MAJOR 55
-#define LIBAVCODEC_VERSION_MINOR 42
+#define LIBAVCODEC_VERSION_MINOR 43
 #define LIBAVCODEC_VERSION_MICRO  0
 
 #define LIBAVCODEC_VERSION_INT  AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \

libavcodec/vp8.h

@@ -237,6 +237,7 @@ typedef struct VP8Context {
         uint8_t pred8x8c[3];
         uint8_t token[4][16][3][NUM_DCT_TOKENS - 1];
         uint8_t mvc[2][19];
+        uint8_t scan[16];
     } prob[2];
 
     VP8Macroblock *macroblocks_base;
@@ -270,6 +271,26 @@ typedef struct VP8Context {
      * 1 -> Macroblocks for entire frame alloced (sliced thread).
      */
     int mb_layout;
+
+    /**
+     * Fade bit present in bitstream (VP7)
+     */
+    int fade_present;
+
+    /**
+     * Interframe DC prediction (VP7)
+     * [0] VP56_FRAME_PREVIOUS
+     * [1] VP56_FRAME_GOLDEN
+     */
+    uint16_t inter_dc_pred[2][2];
+
+    /**
+     * Macroblock features (VP7)
+     */
+    uint8_t feature_enabled[4];
+    uint8_t feature_present_prob[4];
+    uint8_t feature_index_prob[4][3];
+    uint8_t feature_value[4][4];
 } VP8Context;
 
 int ff_vp8_decode_init(AVCodecContext *avctx);

libavcodec/vp8data.h

@@ -30,6 +30,13 @@
 #include "vp8.h"
 #include "h264pred.h"
 
+static const uint8_t vp7_pred4x4_mode[] = {
+    [DC_PRED8x8]    = DC_PRED,
+    [VERT_PRED8x8]  = TM_VP8_PRED,
+    [HOR_PRED8x8]   = TM_VP8_PRED,
+    [PLANE_PRED8x8] = TM_VP8_PRED,
+};
+
 static const uint8_t vp8_pred4x4_mode[] = {
     [DC_PRED8x8]    = DC_PRED,
     [VERT_PRED8x8]  = VERT_PRED,
@@ -51,6 +58,63 @@ static const int8_t vp8_pred16x16_tree_inter[4][2] = {
     { -PLANE_PRED8x8,   -MODE_I4x4 }, // '110', '111'
 };
 
+typedef struct VP7MVPred {
+    int8_t yoffset;
+    int8_t xoffset;
+    uint8_t subblock;
+    uint8_t score;
+} VP7MVPred;
+
+#define VP7_MV_PRED_COUNT 12
+static const VP7MVPred vp7_mv_pred[VP7_MV_PRED_COUNT] = {
+    { -1,  0, 12, 8 },
+    {  0, -1,  3, 8 },
+    { -1, -1, 15, 2 },
+    { -1,  1, 12, 2 },
+    { -2,  0, 12, 2 },
+    {  0, -2,  3, 2 },
+    { -1, -2, 15, 1 },
+    { -2, -1, 15, 1 },
+    { -2,  1, 12, 1 },
+    { -1,  2, 12, 1 },
+    { -2, -2, 15, 1 },
+    { -2,  2, 12, 1 },
+};
+
+static const int vp7_mode_contexts[31][4] = {
+    {   3,   3,   1, 246 },
+    {   7,  89,  66, 239 },
+    {  10,  90,  78, 238 },
+    {  14, 118,  95, 241 },
+    {  14, 123, 106, 238 },
+    {  20, 140, 109, 240 },
+    {  13, 155, 103, 238 },
+    {  21, 158,  99, 240 },
+    {  27,  82, 108, 232 },
+    {  19,  99, 123, 217 },
+    {  45, 139, 148, 236 },
+    {  50, 117, 144, 235 },
+    {  57, 128, 164, 238 },
+    {  69, 139, 171, 239 },
+    {  74, 154, 179, 238 },
+    { 112, 165, 186, 242 },
+    {  98, 143, 185, 245 },
+    { 105, 153, 190, 250 },
+    { 124, 167, 192, 245 },
+    { 131, 186, 203, 246 },
+    {  59, 184, 222, 224 },
+    { 148, 215, 214, 213 },
+    { 137, 211, 210, 219 },
+    { 190, 227, 128, 228 },
+    { 183, 228, 128, 228 },
+    { 194, 234, 128, 228 },
+    { 202, 236, 128, 228 },
+    { 205, 240, 128, 228 },
+    { 205, 244, 128, 228 },
+    { 225, 246, 128, 228 },
+    { 233, 251, 128, 228 },
+};
+
 static const int vp8_mode_contexts[6][4] = {
     {   7,   1,   1, 143 },
     {  14,  18,  14, 107 },
@@ -82,6 +146,10 @@ static const uint8_t vp8_mbsplit_prob[3] = {
     110, 111, 150
 };
 
+static const uint8_t vp7_submv_prob[3] = {
+    180, 162, 25
+};
+
 static const uint8_t vp8_submv_prob[5][3] = {
     { 147, 136,  18 },
     { 106, 145,   1 },
@@ -94,7 +162,7 @@ static const uint8_t vp8_pred16x16_prob_intra[4] = {
     145, 156, 163, 128
 };
 static const uint8_t vp8_pred16x16_prob_inter[4] = {
-    112, 86, 140, 37
+    112,  86, 140,  37
 };
 
 static const int8_t vp8_pred4x4_tree[9][2] = {
@@ -641,7 +709,7 @@ static const uint8_t vp8_token_update_probs[4][8][3][NUM_DCT_TOKENS - 1] = {
 };
 
 // fixme: copied from h264data.h
-static const uint8_t zigzag_scan[16] = {
+static const uint8_t zigzag_scan[16]={
     0 + 0 * 4, 1 + 0 * 4, 0 + 1 * 4, 0 + 2 * 4,
     1 + 1 * 4, 2 + 0 * 4, 3 + 0 * 4, 2 + 1 * 4,
     1 + 2 * 4, 0 + 3 * 4, 1 + 3 * 4, 2 + 2 * 4,
@@ -674,11 +742,22 @@ static const uint8_t vp8_mv_update_prob[2][19] = {
     { 237,
       246,
       253, 253, 254, 254, 254, 254, 254,
-      254, 254, 254, 254, 254, 250, 250, 252, 254, 254 },
+      254, 254, 254, 254, 254, 250, 250, 252, /* VP8 only: */ 254, 254 },
     { 231,
       243,
       245, 253, 254, 254, 254, 254, 254,
-      254, 254, 254, 254, 254, 251, 251, 254, 254, 254 }
+      254, 254, 254, 254, 254, 251, 251, 254, /* VP8 only: */ 254, 254 }
+};
+
+static const uint8_t vp7_mv_default_prob[2][17] = {
+    { 162,
+      128,
+      225, 146, 172, 147, 214,  39, 156,
+      247, 210, 135,  68, 138, 220, 239, 246 },
+    { 164,
+      128,
+      204, 170, 119, 235, 140, 230, 228,
+      244, 184, 201,  44, 173, 221, 239, 253 }
 };
 
 static const uint8_t vp8_mv_default_prob[2][19] = {
@@ -692,4 +771,68 @@ static const uint8_t vp8_mv_default_prob[2][19] = {
       128, 130, 130,  74, 148, 180, 203, 236, 254, 254 }
 };
 
+static const uint8_t vp7_feature_value_size[2][4] = {
+    { 7, 6, 0, 8 },
+    { 7, 6, 0, 5 },
+};
+
+static const int8_t vp7_feature_index_tree[4][2] =
+{
+    {  1,  2 },
+    { -0, -1 }, // '00', '01'
+    { -2, -3 }, // '10', '11'
+};
+
+static const uint16_t vp7_ydc_qlookup[] = {
+      4,   4,   5,   6,   6,   7,   8,   8,   9,  10,  11,  12,  13,  14,  15,
+     16,  17,  18,  19,  20,  21,  22,  23,  23,  24,  25,  26,  27,  28,  29,
+     30,  31,  32,  33,  33,  34,  35,  36,  36,  37,  38,  39,  39,  40,  41,
+     41,  42,  43,  43,  44,  45,  45,  46,  47,  48,  48,  49,  50,  51,  52,
+     53,  53,  54,  56,  57,  58,  59,  60,  62,  63,  65,  66,  68,  70,  72,
+     74,  76,  79,  81,  84,  87,  90,  93,  96, 100, 104, 108, 112, 116, 121,
+    126, 131, 136, 142, 148, 154, 160, 167, 174, 182, 189, 198, 206, 215, 224,
+    234, 244, 254, 265, 277, 288, 301, 313, 327, 340, 355, 370, 385, 401, 417,
+    434, 452, 470, 489, 509, 529, 550, 572,
+};
+
+static const uint16_t vp7_yac_qlookup[] = {
+       4,    4,   5,   5,   6,   6,   7,   8,   9,  10,  11,  12,   13,   15,
+      16,   17,  19,  20,  22,  23,  25,  26,  28,  29,  31,  32,   34,   35,
+      37,   38,  40,  41,  42,  44,  45,  46,  48,  49,  50,  51,   53,   54,
+      55,   56,  57,  58,  59,  61,  62,  63,  64,  65,  67,  68,   69,   70,
+      72,   73,  75,  76,  78,  80,  82,  84,  86,  88,  91,  93,   96,   99,
+     102,  105, 109, 112, 116, 121, 125, 130, 135, 140, 146, 152,  158,  165,
+     172,  180, 188, 196, 205, 214, 224, 234, 245, 256, 268, 281,  294,  308,
+     322,  337, 353, 369, 386, 404, 423, 443, 463, 484, 506, 529,  553,  578,
+     604,  631, 659, 688, 718, 749, 781, 814, 849, 885, 922, 960, 1000, 1041,
+    1083, 1127,
+};
+
+static const uint16_t vp7_y2dc_qlookup[] = {
+       7,    9,  11,  13,  15,  17,  19,  21,  23,  26,  28,  30,   33,   35,
+      37,   39,  42,  44,  46,  48,  51,  53,  55,  57,  59,  61,   63,   65,
+      67,   69,  70,  72,  74,  75,  77,  78,  80,  81,  83,  84,   85,   87,
+      88,   89,  90,  92,  93,  94,  95,  96,  97,  99, 100, 101,  102,  104,
+     105,  106, 108, 109, 111, 113, 114, 116, 118, 120, 123, 125,  128,  131,
+     134,  137, 140, 144, 148, 152, 156, 161, 166, 171, 176, 182,  188,  195,
+     202,  209, 217, 225, 234, 243, 253, 263, 274, 285, 297, 309,  322,  336,
+     350,  365, 381, 397, 414, 432, 450, 470, 490, 511, 533, 556,  579,  604,
+     630,  656, 684, 713, 742, 773, 805, 838, 873, 908, 945, 983, 1022, 1063,
+    1105, 1148,
+};
+
+static const uint16_t vp7_y2ac_qlookup[] = {
+       7,    9,   11,   13,   16,   18,   21,   24,   26,   29,   32,   35,
+      38,   41,   43,   46,   49,   52,   55,   58,   61,   64,   66,   69,
+      72,   74,   77,   79,   82,   84,   86,   88,   91,   93,   95,   97,
+      98,  100,  102,  104,  105,  107,  109,  110,  112,  113,  115,  116,
+     117,  119,  120,  122,  123,  125,  127,  128,  130,  132,  134,  136,
+     138,  141,  143,  146,  149,  152,  155,  158,  162,  166,  171,  175,
+     180,  185,  191,  197,  204,  210,  218,  226,  234,  243,  252,  262,
+     273,  284,  295,  308,  321,  335,  350,  365,  381,  398,  416,  435,
+     455,  476,  497,  520,  544,  569,  595,  622,  650,  680,  711,  743,
+     776,  811,  848,  885,  925,  965, 1008, 1052, 1097, 1144, 1193, 1244,
+    1297, 1351, 1407, 1466, 1526, 1588, 1652, 1719,
+};
+
 #endif /* AVCODEC_VP8DATA_H */

libavcodec/vp8dsp.c

@@ -1,6 +1,7 @@
 /*
  * Copyright (C) 2010 David Conrad
  * Copyright (C) 2010 Ronald S. Bultje
+ * Copyright (C) 2014 Peter Ross
  *
  * This file is part of Libav.
  *
@@ -29,7 +30,126 @@
 #include "mathops.h"
 #include "vp8dsp.h"
 
+#define MK_IDCT_DC_ADD4_C(name)                                               \
+static void name ## _idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16],     \
+                                      ptrdiff_t stride)                       \
+{                                                                             \
+    name ## _idct_dc_add_c(dst + stride * 0 + 0, block[0], stride);           \
+    name ## _idct_dc_add_c(dst + stride * 0 + 4, block[1], stride);           \
+    name ## _idct_dc_add_c(dst + stride * 4 + 0, block[2], stride);           \
+    name ## _idct_dc_add_c(dst + stride * 4 + 4, block[3], stride);           \
+}                                                                             \
+                                                                              \
+static void name ## _idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16],      \
+                                     ptrdiff_t stride)                        \
+{                                                                             \
+    name ## _idct_dc_add_c(dst +  0, block[0], stride);                       \
+    name ## _idct_dc_add_c(dst +  4, block[1], stride);                       \
+    name ## _idct_dc_add_c(dst +  8, block[2], stride);                       \
+    name ## _idct_dc_add_c(dst + 12, block[3], stride);                       \
+}
+
+#if CONFIG_VP7_DECODER
+static void vp7_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
+{
+    int i, a1, b1, c1, d1;
+    int16_t tmp[16];
+
+    for (i = 0; i < 4; i++) {
+        a1 = (dc[i * 4 + 0] + dc[i * 4 + 2]) * 23170;
+        b1 = (dc[i * 4 + 0] - dc[i * 4 + 2]) * 23170;
+        c1 = dc[i * 4 + 1] * 12540 - dc[i * 4 + 3] * 30274;
+        d1 = dc[i * 4 + 1] * 30274 + dc[i * 4 + 3] * 12540;
+        tmp[i * 4 + 0] = (a1 + d1) >> 14;
+        tmp[i * 4 + 3] = (a1 - d1) >> 14;
+        tmp[i * 4 + 1] = (b1 + c1) >> 14;
+        tmp[i * 4 + 2] = (b1 - c1) >> 14;
+    }
+
+    for (i = 0; i < 4; i++) {
+        a1 = (tmp[i + 0] + tmp[i + 8]) * 23170;
+        b1 = (tmp[i + 0] - tmp[i + 8]) * 23170;
+        c1 = tmp[i + 4] * 12540 - tmp[i + 12] * 30274;
+        d1 = tmp[i + 4] * 30274 + tmp[i + 12] * 12540;
+        dc[i * 4 + 0] = 0;
+        dc[i * 4 + 1] = 0;
+        dc[i * 4 + 2] = 0;
+        dc[i * 4 + 3] = 0;
+        block[0][i][0] = (a1 + d1 + 0x20000) >> 18;
+        block[3][i][0] = (a1 - d1 + 0x20000) >> 18;
+        block[1][i][0] = (b1 + c1 + 0x20000) >> 18;
+        block[2][i][0] = (b1 - c1 + 0x20000) >> 18;
+    }
+}
+
+static void vp7_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16])
+{
+    int i, val = (23170 * (23170 * dc[0] >> 14) + 0x20000) >> 18;
+    dc[0] = 0;
+
+    for (i = 0; i < 4; i++) {
+        block[i][0][0] = val;
+        block[i][1][0] = val;
+        block[i][2][0] = val;
+        block[i][3][0] = val;
+    }
+}
+
+static void vp7_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
+{
+    int i, a1, b1, c1, d1;
+    int16_t tmp[16];
+
+    for (i = 0; i < 4; i++) {
+        a1 = (block[i * 4 + 0] + block[i * 4 + 2]) * 23170;
+        b1 = (block[i * 4 + 0] - block[i * 4 + 2]) * 23170;
+        c1 = block[i * 4 + 1] * 12540 - block[i * 4 + 3] * 30274;
+        d1 = block[i * 4 + 1] * 30274 + block[i * 4 + 3] * 12540;
+        block[i * 4 + 0] = 0;
+        block[i * 4 + 1] = 0;
+        block[i * 4 + 2] = 0;
+        block[i * 4 + 3] = 0;
+        tmp[i * 4 + 0] = (a1 + d1) >> 14;
+        tmp[i * 4 + 3] = (a1 - d1) >> 14;
+        tmp[i * 4 + 1] = (b1 + c1) >> 14;
+        tmp[i * 4 + 2] = (b1 - c1) >> 14;
+    }
+
+    for (i = 0; i < 4; i++) {
+        a1 = (tmp[i + 0] + tmp[i + 8]) * 23170;
+        b1 = (tmp[i + 0] - tmp[i + 8]) * 23170;
+        c1 = tmp[i + 4] * 12540 - tmp[i + 12] * 30274;
+        d1 = tmp[i + 4] * 30274 + tmp[i + 12] * 12540;
+        dst[0 * stride + i] = av_clip_uint8(dst[0 * stride + i] +
+                                            ((a1 + d1 + 0x20000) >> 18));
+        dst[3 * stride + i] = av_clip_uint8(dst[3 * stride + i] +
+                                            ((a1 - d1 + 0x20000) >> 18));
+        dst[1 * stride + i] = av_clip_uint8(dst[1 * stride + i] +
+                                            ((b1 + c1 + 0x20000) >> 18));
+        dst[2 * stride + i] = av_clip_uint8(dst[2 * stride + i] +
+                                            ((b1 - c1 + 0x20000) >> 18));
+    }
+}
+
+static void vp7_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
+{
+    int i, dc = (23170 * (23170 * block[0] >> 14) + 0x20000) >> 18;
+    block[0] = 0;
+
+    for (i = 0; i < 4; i++) {
+        dst[0] = av_clip_uint8(dst[0] + dc);
+        dst[1] = av_clip_uint8(dst[1] + dc);
+        dst[2] = av_clip_uint8(dst[2] + dc);
+        dst[3] = av_clip_uint8(dst[3] + dc);
+        dst   += stride;
+    }
+}
+
+MK_IDCT_DC_ADD4_C(vp7)
+#endif /* CONFIG_VP7_DECODER */
+
 // TODO: Maybe add dequant
+#if CONFIG_VP8_DECODER
 static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
 {
     int i, t0, t1, t2, t3;
@@ -128,23 +248,8 @@ static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
     }
 }
 
-static void vp8_idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16],
-                                 ptrdiff_t stride)
-{
-    vp8_idct_dc_add_c(dst + stride * 0 + 0, block[0], stride);
-    vp8_idct_dc_add_c(dst + stride * 0 + 4, block[1], stride);
-    vp8_idct_dc_add_c(dst + stride * 4 + 0, block[2], stride);
-    vp8_idct_dc_add_c(dst + stride * 4 + 4, block[3], stride);
-}
-
-static void vp8_idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16],
-                                ptrdiff_t stride)
-{
-    vp8_idct_dc_add_c(dst +  0, block[0], stride);
-    vp8_idct_dc_add_c(dst +  4, block[1], stride);
-    vp8_idct_dc_add_c(dst +  8, block[2], stride);
-    vp8_idct_dc_add_c(dst + 12, block[3], stride);
-}
+MK_IDCT_DC_ADD4_C(vp8)
+#endif /* CONFIG_VP8_DECODER */
 
 // because I like only having two parameters to pass functions...
 #define LOAD_PIXELS                                                           \
@@ -160,7 +265,7 @@ static void vp8_idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16],
 #define clip_int8(n) (cm[n + 0x80] - 0x80)
 
 static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride,
-                                           int is4tap)
+                                           int is4tap, int is_vp7)
 {
     LOAD_PIXELS
     int a, f1, f2;
@@ -176,7 +281,11 @@ static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride,
     // We deviate from the spec here with c(a+3) >> 3
     // since that's what libvpx does.
     f1 = FFMIN(a + 4, 127) >> 3;
-    f2 = FFMIN(a + 3, 127) >> 3;
+
+    if (is_vp7)
+        f2 = f1 - ((a & 7) == 4);
+    else
+        f2 = FFMIN(a + 3, 127) >> 3;
 
     // Despite what the spec says, we do need to clamp here to
     // be bitexact with libvpx.
@@ -185,13 +294,33 @@ static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride,
 
     // only used for _inner on blocks without high edge variance
     if (!is4tap) {
-        a = (f1 + 1) >> 1;
+        a              = (f1 + 1) >> 1;
         p[-2 * stride] = cm[p1 + a];
         p[ 1 * stride] = cm[q1 - a];
     }
 }
 
-static av_always_inline int simple_limit(uint8_t *p, ptrdiff_t stride, int flim)
+static av_always_inline void vp7_filter_common(uint8_t *p, ptrdiff_t stride,
+                                               int is4tap)
+{
+    filter_common(p, stride, is4tap, IS_VP7);
+}
+
+static av_always_inline void vp8_filter_common(uint8_t *p, ptrdiff_t stride,
+                                               int is4tap)
+{
+    filter_common(p, stride, is4tap, IS_VP8);
+}
+
+static av_always_inline int vp7_simple_limit(uint8_t *p, ptrdiff_t stride,
+                                             int flim)
+{
+    LOAD_PIXELS
+    return FFABS(p0 - q0) <= flim;
+}
+
+static av_always_inline int vp8_simple_limit(uint8_t *p, ptrdiff_t stride,
+                                             int flim)
 {
     LOAD_PIXELS
     return 2 * FFABS(p0 - q0) + (FFABS(p1 - q1) >> 1) <= flim;
@@ -201,19 +330,21 @@ static av_always_inline int simple_limit(uint8_t *p, ptrdiff_t stride, int flim)
  * E - limit at the macroblock edge
  * I - limit for interior difference
  */
-static av_always_inline int normal_limit(uint8_t *p, ptrdiff_t stride,
-                                         int E, int I)
-{
-    LOAD_PIXELS
-    return simple_limit(p, stride, E) &&
-           FFABS(p3 - p2) <= I &&
-           FFABS(p2 - p1) <= I &&
-           FFABS(p1 - p0) <= I &&
-           FFABS(q3 - q2) <= I &&
-           FFABS(q2 - q1) <= I &&
-           FFABS(q1 - q0) <= I;
+#define NORMAL_LIMIT(vpn)                                                     \
+static av_always_inline int vp ## vpn ## _normal_limit(uint8_t *p,            \
+                                                       ptrdiff_t stride,      \
+                                                       int E, int I)          \
+{                                                                             \
+    LOAD_PIXELS                                                               \
+    return vp ## vpn ## _simple_limit(p, stride, E) &&                        \
+           FFABS(p3 - p2) <= I && FFABS(p2 - p1) <= I &&                      \
+           FFABS(p1 - p0) <= I && FFABS(q3 - q2) <= I &&                      \
+           FFABS(q2 - q1) <= I && FFABS(q1 - q0) <= I;                        \
 }
 
+NORMAL_LIMIT(7)
+NORMAL_LIMIT(8)
+
 // high edge variance
 static av_always_inline int hev(uint8_t *p, ptrdiff_t stride, int thresh)
 {
@@ -243,82 +374,91 @@ static av_always_inline void filter_mbedge(uint8_t *p, ptrdiff_t stride)
     p[ 2 * stride] = cm[q2 - a2];
 }
 
-#define LOOP_FILTER(dir, size, stridea, strideb, maybe_inline)                \
+#define LOOP_FILTER(vpn, dir, size, stridea, strideb, maybe_inline)           \
 static maybe_inline                                                           \
-void vp8_ ## dir ## _loop_filter ## size ## _c(uint8_t *dst,                  \
-                                               ptrdiff_t stride,              \
-                                               int flim_E, int flim_I,        \
-                                               int hev_thresh)                \
+void vpn ## _ ## dir ## _loop_filter ## size ## _c(uint8_t *dst,              \
+                                                   ptrdiff_t stride,          \
+                                                   int flim_E, int flim_I,    \
+                                                   int hev_thresh)            \
 {                                                                             \
     int i;                                                                    \
     for (i = 0; i < size; i++)                                                \
-        if (normal_limit(dst + i * stridea, strideb, flim_E, flim_I)) {       \
+        if (vpn ## _normal_limit(dst + i * stridea, strideb,                  \
+                                 flim_E, flim_I)) {                           \
             if (hev(dst + i * stridea, strideb, hev_thresh))                  \
-                filter_common(dst + i * stridea, strideb, 1);                 \
+                vpn ## _filter_common(dst + i * stridea, strideb, 1);         \
             else                                                              \
                 filter_mbedge(dst + i * stridea, strideb);                    \
         }                                                                     \
 }                                                                             \
                                                                               \
 static maybe_inline                                                           \
-void vp8_ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst,            \
-                                                     ptrdiff_t stride,        \
-                                                     int flim_E, int flim_I,  \
-                                                     int hev_thresh)          \
+void vpn ## _ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst,        \
+                                                         ptrdiff_t stride,    \
+                                                         int flim_E,          \
+                                                         int flim_I,          \
+                                                         int hev_thresh)      \
 {                                                                             \
     int i;                                                                    \
     for (i = 0; i < size; i++)                                                \
-        if (normal_limit(dst + i * stridea, strideb, flim_E, flim_I)) {       \
+        if (vpn ## _normal_limit(dst + i * stridea, strideb,                  \
+                                 flim_E, flim_I)) {                           \
             int hv = hev(dst + i * stridea, strideb, hev_thresh);             \
             if (hv)                                                           \
-                filter_common(dst + i * stridea, strideb, 1);                 \
+                vpn ## _filter_common(dst + i * stridea, strideb, 1);         \
             else                                                              \
-                filter_common(dst + i * stridea, strideb, 0);                 \
+                vpn ## _filter_common(dst + i * stridea, strideb, 0);         \
         }                                                                     \
 }
 
-LOOP_FILTER(v, 16, 1, stride, )
-LOOP_FILTER(h, 16, stride, 1, )
-
-#define UV_LOOP_FILTER(dir, stridea, strideb)                                 \
-LOOP_FILTER(dir, 8, stridea, strideb, av_always_inline)                       \
-static void vp8_ ## dir ## _loop_filter8uv_c(uint8_t *dstU, uint8_t *dstV,    \
-                                             ptrdiff_t stride, int fE,        \
-                                             int fI, int hev_thresh)          \
+#define UV_LOOP_FILTER(vpn, dir, stridea, strideb)                            \
+LOOP_FILTER(vpn, dir, 8, stridea, strideb, av_always_inline)                  \
+static void vpn ## _ ## dir ## _loop_filter8uv_c(uint8_t *dstU,               \
+                                                 uint8_t *dstV,               \
+                                                 ptrdiff_t stride, int fE,    \
+                                                 int fI, int hev_thresh)      \
 {                                                                             \
-    vp8_ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh);         \
-    vp8_ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh);         \
+    vpn ## _ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh);     \
+    vpn ## _ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh);     \
 }                                                                             \
                                                                               \
-static void vp8_ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU,             \
-                                                   uint8_t *dstV,             \
-                                                   ptrdiff_t stride, int fE,  \
-                                                   int fI, int hev_thresh)    \
+static void vpn ## _ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU,         \
+                                                       uint8_t *dstV,         \
+                                                       ptrdiff_t stride,      \
+                                                       int fE, int fI,        \
+                                                       int hev_thresh)        \
 {                                                                             \
-    vp8_ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, hev_thresh);   \
-    vp8_ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, hev_thresh);   \
+    vpn ## _ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI,            \
+                                             hev_thresh);                     \
+    vpn ## _ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI,            \
+                                             hev_thresh);                     \
 }
 
-UV_LOOP_FILTER(v, 1, stride)
-UV_LOOP_FILTER(h, stride, 1)
-
-static void vp8_v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim)
-{
-    int i;
-
-    for (i = 0; i < 16; i++)
-        if (simple_limit(dst + i, stride, flim))
-            filter_common(dst + i, stride, 1);
+#define LOOP_FILTER_SIMPLE(vpn)                                               \
+static void vpn ## _v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride,    \
+                                           int flim)                          \
+{                                                                             \
+    int i;                                                                    \
+    for (i = 0; i < 16; i++)                                                  \
+        if (vpn ## _simple_limit(dst + i, stride, flim))                      \
+            vpn ## _filter_common(dst + i, stride, 1);                        \
+}                                                                             \
+                                                                              \
+static void vpn ## _h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride,    \
+                                           int flim)                          \
+{                                                                             \
+    int i;                                                                    \
+    for (i = 0; i < 16; i++)                                                  \
+        if (vpn ## _simple_limit(dst + i * stride, 1, flim))                  \
+            vpn ## _filter_common(dst + i * stride, 1, 1);                    \
 }
 
-static void vp8_h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim)
-{
-    int i;
-
-    for (i = 0; i < 16; i++)
-        if (simple_limit(dst + i * stride, 1, flim))
-            filter_common(dst + i * stride, 1, 1);
-}
+#define LOOP_FILTERS(vpn)                \
+    LOOP_FILTER(vpn, v, 16, 1, stride, ) \
+    LOOP_FILTER(vpn, h, 16, stride, 1, ) \
+    UV_LOOP_FILTER(vpn, v, 1, stride)    \
+    UV_LOOP_FILTER(vpn, h, stride, 1)    \
+    LOOP_FILTER_SIMPLE(vpn)              \
 
 static const uint8_t subpel_filters[7][6] = {
     { 0,  6, 123,  12,  1, 0 },
@@ -507,7 +647,7 @@ VP8_BILINEAR(16)
 VP8_BILINEAR(8)
 VP8_BILINEAR(4)
 
-#define VP8_MC_FUNC(IDX, SIZE)                                                \
+#define VP78_MC_FUNC(IDX, SIZE)                                               \
     dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c;   \
     dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c;  \
     dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c;  \
@@ -518,8 +658,8 @@ VP8_BILINEAR(4)
     dsp->put_vp8_epel_pixels_tab[IDX][2][1] = put_vp8_epel ## SIZE ## _h4v6_c; \
     dsp->put_vp8_epel_pixels_tab[IDX][2][2] = put_vp8_epel ## SIZE ## _h6v6_c
 
-#define VP8_BILINEAR_MC_FUNC(IDX, SIZE)                                       \
-    dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
+#define VP78_BILINEAR_MC_FUNC(IDX, SIZE)                                      \
+    dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels   ## SIZE ## _c; \
     dsp->put_vp8_bilinear_pixels_tab[IDX][0][1] = put_vp8_bilinear ## SIZE ## _h_c; \
     dsp->put_vp8_bilinear_pixels_tab[IDX][0][2] = put_vp8_bilinear ## SIZE ## _h_c; \
     dsp->put_vp8_bilinear_pixels_tab[IDX][1][0] = put_vp8_bilinear ## SIZE ## _v_c; \
@@ -529,6 +669,54 @@ VP8_BILINEAR(4)
     dsp->put_vp8_bilinear_pixels_tab[IDX][2][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
     dsp->put_vp8_bilinear_pixels_tab[IDX][2][2] = put_vp8_bilinear ## SIZE ## _hv_c
 
+av_cold void ff_vp78dsp_init(VP8DSPContext *dsp)
+{
+    VP78_MC_FUNC(0, 16);
+    VP78_MC_FUNC(1, 8);
+    VP78_MC_FUNC(2, 4);
+
+    VP78_BILINEAR_MC_FUNC(0, 16);
+    VP78_BILINEAR_MC_FUNC(1, 8);
+    VP78_BILINEAR_MC_FUNC(2, 4);
+
+    if (ARCH_ARM)
+        ff_vp78dsp_init_arm(dsp);
+    if (ARCH_PPC)
+        ff_vp78dsp_init_ppc(dsp);
+    if (ARCH_X86)
+        ff_vp78dsp_init_x86(dsp);
+}
+
+#if CONFIG_VP7_DECODER
+LOOP_FILTERS(vp7)
+
+av_cold void ff_vp7dsp_init(VP8DSPContext *dsp)
+{
+    dsp->vp8_luma_dc_wht    = vp7_luma_dc_wht_c;
+    dsp->vp8_luma_dc_wht_dc = vp7_luma_dc_wht_dc_c;
+    dsp->vp8_idct_add       = vp7_idct_add_c;
+    dsp->vp8_idct_dc_add    = vp7_idct_dc_add_c;
+    dsp->vp8_idct_dc_add4y  = vp7_idct_dc_add4y_c;
+    dsp->vp8_idct_dc_add4uv = vp7_idct_dc_add4uv_c;
+
+    dsp->vp8_v_loop_filter16y = vp7_v_loop_filter16_c;
+    dsp->vp8_h_loop_filter16y = vp7_h_loop_filter16_c;
+    dsp->vp8_v_loop_filter8uv = vp7_v_loop_filter8uv_c;
+    dsp->vp8_h_loop_filter8uv = vp7_h_loop_filter8uv_c;
+
+    dsp->vp8_v_loop_filter16y_inner = vp7_v_loop_filter16_inner_c;
+    dsp->vp8_h_loop_filter16y_inner = vp7_h_loop_filter16_inner_c;
+    dsp->vp8_v_loop_filter8uv_inner = vp7_v_loop_filter8uv_inner_c;
+    dsp->vp8_h_loop_filter8uv_inner = vp7_h_loop_filter8uv_inner_c;
+
+    dsp->vp8_v_loop_filter_simple = vp7_v_loop_filter_simple_c;
+    dsp->vp8_h_loop_filter_simple = vp7_h_loop_filter_simple_c;
+}
+#endif /* CONFIG_VP7_DECODER */
+
+#if CONFIG_VP8_DECODER
+LOOP_FILTERS(vp8)
+
 av_cold void ff_vp8dsp_init(VP8DSPContext *dsp)
 {
     dsp->vp8_luma_dc_wht    = vp8_luma_dc_wht_c;
@@ -551,18 +739,9 @@ av_cold void ff_vp8dsp_init(VP8DSPContext *dsp)
     dsp->vp8_v_loop_filter_simple = vp8_v_loop_filter_simple_c;
     dsp->vp8_h_loop_filter_simple = vp8_h_loop_filter_simple_c;
 
-    VP8_MC_FUNC(0, 16);
-    VP8_MC_FUNC(1, 8);
-    VP8_MC_FUNC(2, 4);
-
-    VP8_BILINEAR_MC_FUNC(0, 16);
-    VP8_BILINEAR_MC_FUNC(1, 8);
-    VP8_BILINEAR_MC_FUNC(2, 4);
-
     if (ARCH_ARM)
         ff_vp8dsp_init_arm(dsp);
-    if (ARCH_PPC)
-        ff_vp8dsp_init_ppc(dsp);
     if (ARCH_X86)
         ff_vp8dsp_init_x86(dsp);
 }
+#endif /* CONFIG_VP8_DECODER */

libavcodec/vp8dsp.h

@@ -88,9 +88,18 @@ 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_vp7dsp_init(VP8DSPContext *c);
+
+void ff_vp78dsp_init(VP8DSPContext *c);
+void ff_vp78dsp_init_arm(VP8DSPContext *c);
+void ff_vp78dsp_init_ppc(VP8DSPContext *c);
+void ff_vp78dsp_init_x86(VP8DSPContext *c);
+
 void ff_vp8dsp_init(VP8DSPContext *c);
-void ff_vp8dsp_init_x86(VP8DSPContext *c);
 void ff_vp8dsp_init_arm(VP8DSPContext *c);
-void ff_vp8dsp_init_ppc(VP8DSPContext *c);
+void ff_vp8dsp_init_x86(VP8DSPContext *c);
+
+#define IS_VP7 1
+#define IS_VP8 0
 
 #endif /* AVCODEC_VP8DSP_H */

libavcodec/x86/Makefile

@@ -37,6 +37,7 @@ OBJS-$(CONFIG_TRUEHD_DECODER)          += x86/mlpdsp.o
 OBJS-$(CONFIG_VC1_DECODER)             += x86/vc1dsp_init.o
 OBJS-$(CONFIG_VORBIS_DECODER)          += x86/vorbisdsp_init.o
 OBJS-$(CONFIG_VP6_DECODER)             += x86/vp6dsp_init.o
+OBJS-$(CONFIG_VP7_DECODER)             += x86/vp8dsp_init.o
 OBJS-$(CONFIG_VP8_DECODER)             += x86/vp8dsp_init.o
 OBJS-$(CONFIG_VP9_DECODER)             += x86/vp9dsp_init.o
 
@@ -92,6 +93,8 @@ YASM-OBJS-$(CONFIG_RV40_DECODER)       += x86/rv34dsp.o                 \
 YASM-OBJS-$(CONFIG_VC1_DECODER)        += x86/vc1dsp.o
 YASM-OBJS-$(CONFIG_VORBIS_DECODER)     += x86/vorbisdsp.o
 YASM-OBJS-$(CONFIG_VP6_DECODER)        += x86/vp6dsp.o
+YASM-OBJS-$(CONFIG_VP7_DECODER)        += x86/vp8dsp.o                  \
+                                          x86/vp8dsp_loopfilter.o
 YASM-OBJS-$(CONFIG_VP8_DECODER)        += x86/vp8dsp.o                  \
                                           x86/vp8dsp_loopfilter.o
 YASM-OBJS-$(CONFIG_VP9_DECODER)        += x86/vp9dsp.o

libavcodec/x86/h264_intrapred_init.c

@@ -195,7 +195,7 @@ av_cold void ff_h264_pred_init_x86(H264PredContext *h, int codec_id,
                 h->pred8x8  [VERT_PRED8x8     ] = ff_pred8x8_vertical_8_mmx;
                 h->pred8x8  [HOR_PRED8x8      ] = ff_pred8x8_horizontal_8_mmx;
             }
-            if (codec_id == AV_CODEC_ID_VP8) {
+            if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {
                 h->pred16x16[PLANE_PRED8x8    ] = ff_pred16x16_tm_vp8_8_mmx;
                 h->pred8x8  [PLANE_PRED8x8    ] = ff_pred8x8_tm_vp8_8_mmx;
                 h->pred4x4  [TM_VP8_PRED      ] = ff_pred4x4_tm_vp8_8_mmx;
@@ -231,7 +231,8 @@ av_cold void ff_h264_pred_init_x86(H264PredContext *h, int codec_id,
             h->pred4x4  [VERT_RIGHT_PRED        ] = ff_pred4x4_vertical_right_8_mmxext;
             h->pred4x4  [HOR_DOWN_PRED          ] = ff_pred4x4_horizontal_down_8_mmxext;
             h->pred4x4  [DC_PRED                ] = ff_pred4x4_dc_8_mmxext;
-            if (codec_id == AV_CODEC_ID_VP8 || codec_id == AV_CODEC_ID_H264) {
+            if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8 ||
+                codec_id == AV_CODEC_ID_H264) {
                 h->pred4x4  [DIAG_DOWN_LEFT_PRED] = ff_pred4x4_down_left_8_mmxext;
             }
             if (codec_id == AV_CODEC_ID_SVQ3 || codec_id == AV_CODEC_ID_H264) {
@@ -246,7 +247,7 @@ av_cold void ff_h264_pred_init_x86(H264PredContext *h, int codec_id,
                     h->pred8x8[DC_PRED8x8       ] = ff_pred8x8_dc_8_mmxext;
                 }
             }
-            if (codec_id == AV_CODEC_ID_VP8) {
+            if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {
                 h->pred16x16[PLANE_PRED8x8      ] = ff_pred16x16_tm_vp8_8_mmxext;
                 h->pred8x8  [DC_PRED8x8         ] = ff_pred8x8_dc_rv40_8_mmxext;
                 h->pred8x8  [PLANE_PRED8x8      ] = ff_pred8x8_tm_vp8_8_mmxext;
@@ -276,7 +277,7 @@ av_cold void ff_h264_pred_init_x86(H264PredContext *h, int codec_id,
             h->pred8x8l [VERT_RIGHT_PRED      ] = ff_pred8x8l_vertical_right_8_sse2;
             h->pred8x8l [VERT_LEFT_PRED       ] = ff_pred8x8l_vertical_left_8_sse2;
             h->pred8x8l [HOR_DOWN_PRED        ] = ff_pred8x8l_horizontal_down_8_sse2;
-            if (codec_id == AV_CODEC_ID_VP8) {
+            if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {
                 h->pred16x16[PLANE_PRED8x8    ] = ff_pred16x16_tm_vp8_8_sse2;
                 h->pred8x8  [PLANE_PRED8x8    ] = ff_pred8x8_tm_vp8_8_sse2;
             } else {
@@ -307,7 +308,7 @@ av_cold void ff_h264_pred_init_x86(H264PredContext *h, int codec_id,
             h->pred8x8l [VERT_LEFT_PRED       ] = ff_pred8x8l_vertical_left_8_ssse3;
             h->pred8x8l [HOR_UP_PRED          ] = ff_pred8x8l_horizontal_up_8_ssse3;
             h->pred8x8l [HOR_DOWN_PRED        ] = ff_pred8x8l_horizontal_down_8_ssse3;
-            if (codec_id == AV_CODEC_ID_VP8) {
+            if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {
                 h->pred8x8  [PLANE_PRED8x8    ] = ff_pred8x8_tm_vp8_8_ssse3;
                 h->pred4x4  [TM_VP8_PRED      ] = ff_pred4x4_tm_vp8_8_ssse3;
             } else {

libavcodec/x86/vp8dsp_init.c

@@ -315,25 +315,69 @@ 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_vp78dsp_init_x86(VP8DSPContext *c)
 {
 #if HAVE_YASM
     int cpu_flags = av_get_cpu_flags();
 
     if (EXTERNAL_MMX(cpu_flags)) {
-        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
-        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
         c->put_vp8_epel_pixels_tab[1][0][0]     =
         c->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_mmx;
+    }
+
+    /* note that 4-tap width=16 functions are missing because w=16
+     * is only used for luma, and luma is always a copy or sixtap. */
+    if (EXTERNAL_MMXEXT(cpu_flags)) {
+        VP8_MC_FUNC(2, 4, mmxext);
+        VP8_BILINEAR_MC_FUNC(2, 4, mmxext);
+#if ARCH_X86_32
+        VP8_LUMA_MC_FUNC(0, 16, mmxext);
+        VP8_MC_FUNC(1, 8, mmxext);
+        VP8_BILINEAR_MC_FUNC(0, 16, mmxext);
+        VP8_BILINEAR_MC_FUNC(1,  8, mmxext);
+#endif
+    }
+
+    if (EXTERNAL_SSE(cpu_flags)) {
+        c->put_vp8_epel_pixels_tab[0][0][0]     =
+        c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse;
+    }
+
+    if (EXTERNAL_SSE2(cpu_flags) && (cpu_flags & AV_CPU_FLAG_SSE2SLOW)) {
+        VP8_LUMA_MC_FUNC(0, 16, sse2);
+        VP8_MC_FUNC(1, 8, sse2);
+        VP8_BILINEAR_MC_FUNC(0, 16, sse2);
+        VP8_BILINEAR_MC_FUNC(1, 8, sse2);
+    }
+
+    if (EXTERNAL_SSSE3(cpu_flags)) {
+        VP8_LUMA_MC_FUNC(0, 16, ssse3);
+        VP8_MC_FUNC(1, 8, ssse3);
+        VP8_MC_FUNC(2, 4, ssse3);
+        VP8_BILINEAR_MC_FUNC(0, 16, ssse3);
+        VP8_BILINEAR_MC_FUNC(1, 8, ssse3);
+        VP8_BILINEAR_MC_FUNC(2, 4, ssse3);
+    }
+#endif /* HAVE_YASM */
+}
 
+av_cold void ff_vp8dsp_init_x86(VP8DSPContext *c)
+{
+#if HAVE_YASM
+    int cpu_flags = av_get_cpu_flags();
+
+    if (EXTERNAL_MMX(cpu_flags)) {
+        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
+        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->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;
 
@@ -352,14 +396,7 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
     /* note that 4-tap width=16 functions are missing because w=16
      * is only used for luma, and luma is always a copy or sixtap. */
     if (EXTERNAL_MMXEXT(cpu_flags)) {
-        VP8_MC_FUNC(2, 4, mmxext);
-        VP8_BILINEAR_MC_FUNC(2, 4, mmxext);
 #if ARCH_X86_32
-        VP8_LUMA_MC_FUNC(0, 16, mmxext);
-        VP8_MC_FUNC(1, 8, mmxext);
-        VP8_BILINEAR_MC_FUNC(0, 16, mmxext);
-        VP8_BILINEAR_MC_FUNC(1,  8, mmxext);
-
         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;
 
@@ -378,16 +415,9 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
     if (EXTERNAL_SSE(cpu_flags)) {
         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;
     }
 
     if (EXTERNAL_SSE2(cpu_flags) && (cpu_flags & AV_CPU_FLAG_SSE2SLOW)) {
-        VP8_LUMA_MC_FUNC(0, 16, sse2);
-        VP8_MC_FUNC(1, 8, sse2);
-        VP8_BILINEAR_MC_FUNC(0, 16, sse2);
-        VP8_BILINEAR_MC_FUNC(1, 8, sse2);
-
         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;
@@ -410,13 +440,6 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c)
     }
 
     if (EXTERNAL_SSSE3(cpu_flags)) {
-        VP8_LUMA_MC_FUNC(0, 16, ssse3);
-        VP8_MC_FUNC(1, 8, ssse3);
-        VP8_MC_FUNC(2, 4, ssse3);
-        VP8_BILINEAR_MC_FUNC(0, 16, ssse3);
-        VP8_BILINEAR_MC_FUNC(1, 8, ssse3);
-        VP8_BILINEAR_MC_FUNC(2, 4, ssse3);
-
         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;
 

libavformat/riff.c

@@ -251,6 +251,8 @@ const AVCodecTag ff_codec_bmp_tags[] = {
     { AV_CODEC_ID_VP6A,         MKTAG('V', 'P', '6', 'A') },
     { AV_CODEC_ID_VP6F,         MKTAG('V', 'P', '6', 'F') },
     { AV_CODEC_ID_VP6F,         MKTAG('F', 'L', 'V', '4') },
+    { AV_CODEC_ID_VP7,          MKTAG('V', 'P', '7', '0') },
+    { AV_CODEC_ID_VP7,          MKTAG('V', 'P', '7', '1') },
     { AV_CODEC_ID_VP8,          MKTAG('V', 'P', '8', '0') },
     { AV_CODEC_ID_VP9,          MKTAG('V', 'P', '9', '0') },
     { AV_CODEC_ID_ASV1,         MKTAG('A', 'S', 'V', '1') },