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>

6 files changed, 72 insertions, 46 deletions

diff --git a/libavcodec/arm/Makefile b/libavcodec/arm/Makefile
index 3d34c4a754..9d5b6aab5b 100644
--- a/libavcodec/arm/Makefile
+++ b/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
diff --git a/libavcodec/arm/h264pred_init_arm.c b/libavcodec/arm/h264pred_init_arm.c
index 2e25c2a271..bbfe63f2aa 100644
--- a/libavcodec/arm/h264pred_init_arm.c
+++ b/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;
 }
 
diff --git a/libavcodec/arm/vp8dsp.h b/libavcodec/arm/vp8dsp.h
index ce00e4abf5..0d55e0ffc0 100644
--- a/libavcodec/arm/vp8dsp.h
+++ b/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);
 
diff --git a/libavcodec/arm/vp8dsp_init_arm.c b/libavcodec/arm/vp8dsp_init_arm.c
index b7897cd9fd..aa77dbab98 100644
--- a/libavcodec/arm/vp8dsp_init_arm.c
+++ b/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();
diff --git a/libavcodec/arm/vp8dsp_init_armv6.c b/libavcodec/arm/vp8dsp_init_armv6.c
index e15e191e30..febe4e71a2 100644
--- a/libavcodec/arm/vp8dsp_init_armv6.c
+++ b/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;
+}
diff --git a/libavcodec/arm/vp8dsp_init_neon.c b/libavcodec/arm/vp8dsp_init_neon.c
index 0468181f70..2b6c7750d3 100644
--- a/libavcodec/arm/vp8dsp_init_neon.c
+++ b/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;
+}