Merge remote-tracking branch 'qatar/master'

* qatar/master:
  ppc: fix build with altivec disabled
  vp3: move idct and loop filter pointers to new vp3dsp context
  build: add CONFIG_VP3DSP, reduce repetition in OBJS lists
  tscc2: do not add/subtract 128 bias during DCT
  tscc2: fix typo in DCT
  configure: clarify external library section of help output
  configure: mark libfdk-aac as nonfree
  configure: cosmetics: drop some unnecessary backslashes
  os_support: K&R formatting cosmetics

Conflicts:
	configure
	libavcodec/vp3.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

21 files changed, 232 insertions, 182 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index bbe188a..45dee19 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -59,6 +59,7 @@ OBJS-$(CONFIG_SINEWIN)                 += sinewin.o
 OBJS-$(CONFIG_VAAPI)                   += vaapi.o
 OBJS-$(CONFIG_VDA)                     += vda.o
 OBJS-$(CONFIG_VDPAU)                   += vdpau.o
+OBJS-$(CONFIG_VP3DSP)                  += vp3dsp.o
 
 # decoders/encoders/hardware accelerators
 OBJS-$(CONFIG_A64MULTI_ENCODER)        += a64multienc.o elbg.o
@@ -484,11 +485,11 @@ OBJS-$(CONFIG_VORBIS_DECODER)          += vorbisdec.o vorbis.o \
                                           vorbis_data.o xiph.o
 OBJS-$(CONFIG_VORBIS_ENCODER)          += vorbisenc.o vorbis.o \
                                           vorbis_data.o
-OBJS-$(CONFIG_VP3_DECODER)             += vp3.o vp3dsp.o
+OBJS-$(CONFIG_VP3_DECODER)             += vp3.o
 OBJS-$(CONFIG_VP5_DECODER)             += vp5.o vp56.o vp56data.o vp56dsp.o \
-                                          vp3dsp.o vp56rac.o
+                                          vp56rac.o
 OBJS-$(CONFIG_VP6_DECODER)             += vp6.o vp56.o vp56data.o vp56dsp.o \
-                                          vp3dsp.o vp6dsp.o vp56rac.o
+                                          vp6dsp.o vp56rac.o
 OBJS-$(CONFIG_VP8_DECODER)             += vp8.o vp8dsp.o vp56rac.o
 OBJS-$(CONFIG_VQA_DECODER)             += vqavideo.o
 OBJS-$(CONFIG_WAVPACK_DECODER)         += wavpack.o
diff --git a/libavcodec/arm/Makefile b/libavcodec/arm/Makefile
index d2bdd50..a8e531c 100644
--- a/libavcodec/arm/Makefile
+++ b/libavcodec/arm/Makefile
@@ -11,6 +11,7 @@ ARMV6-OBJS-$(CONFIG_AC3DSP)            += arm/ac3dsp_armv6.o
 OBJS-$(CONFIG_MPEGAUDIODSP)            += arm/mpegaudiodsp_init_arm.o
 ARMV6-OBJS-$(CONFIG_MPEGAUDIODSP)      += arm/mpegaudiodsp_fixed_armv6.o
 
+OBJS-$(CONFIG_VP3DSP)                  += arm/vp3dsp_init_arm.o
 OBJS-$(CONFIG_VP5_DECODER)             += arm/vp56dsp_init_arm.o
 OBJS-$(CONFIG_VP6_DECODER)             += arm/vp56dsp_init_arm.o
 OBJS-$(CONFIG_VP8_DECODER)             += arm/vp8dsp_init_arm.o
@@ -75,13 +76,11 @@ NEON-OBJS-$(CONFIG_RV40_DECODER)       += arm/rv34dsp_init_neon.o       \
                                           arm/rv40dsp_neon.o            \
                                           arm/h264cmc_neon.o            \
 
-NEON-OBJS-$(CONFIG_VP3_DECODER)        += arm/vp3dsp_neon.o
+NEON-OBJS-$(CONFIG_VP3DSP)             += arm/vp3dsp_neon.o
 
 NEON-OBJS-$(CONFIG_VP5_DECODER)        += arm/vp56dsp_neon.o            \
-                                          arm/vp3dsp_neon.o             \
 
 NEON-OBJS-$(CONFIG_VP6_DECODER)        += arm/vp56dsp_neon.o            \
-                                          arm/vp3dsp_neon.o             \
 
 NEON-OBJS-$(CONFIG_VP8_DECODER)        += arm/vp8dsp_init_neon.o        \
                                           arm/vp8dsp_neon.o
diff --git a/libavcodec/arm/dsputil_init_neon.c b/libavcodec/arm/dsputil_init_neon.c
index ef5a8df..5533a28 100644
--- a/libavcodec/arm/dsputil_init_neon.c
+++ b/libavcodec/arm/dsputil_init_neon.c
@@ -29,11 +29,6 @@ void ff_simple_idct_neon(DCTELEM *data);
 void ff_simple_idct_put_neon(uint8_t *dest, int line_size, DCTELEM *data);
 void ff_simple_idct_add_neon(uint8_t *dest, int line_size, DCTELEM *data);
 
-void ff_vp3_idct_neon(DCTELEM *data);
-void ff_vp3_idct_put_neon(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_vp3_idct_add_neon(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_vp3_idct_dc_add_neon(uint8_t *dest, int line_size, const DCTELEM *data);
-
 void ff_clear_block_neon(DCTELEM *block);
 void ff_clear_blocks_neon(DCTELEM *blocks);
 
@@ -147,9 +142,6 @@ void ff_avg_h264_chroma_mc8_neon(uint8_t *, uint8_t *, int, int, int, int);
 void ff_avg_h264_chroma_mc4_neon(uint8_t *, uint8_t *, int, int, int, int);
 void ff_avg_h264_chroma_mc2_neon(uint8_t *, uint8_t *, int, int, int, int);
 
-void ff_vp3_v_loop_filter_neon(uint8_t *, int, int *);
-void ff_vp3_h_loop_filter_neon(uint8_t *, int, int *);
-
 void ff_vector_fmul_window_neon(float *dst, const float *src0,
                                 const float *src1, const float *win, int len);
 void ff_vector_fmul_scalar_neon(float *dst, const float *src, float mul,
@@ -186,13 +178,6 @@ void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
             c->idct_add              = ff_simple_idct_add_neon;
             c->idct                  = ff_simple_idct_neon;
             c->idct_permutation_type = FF_PARTTRANS_IDCT_PERM;
-        } else if ((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER ||
-                    CONFIG_VP6_DECODER) &&
-                   avctx->idct_algo == FF_IDCT_VP3) {
-            c->idct_put              = ff_vp3_idct_put_neon;
-            c->idct_add              = ff_vp3_idct_add_neon;
-            c->idct                  = ff_vp3_idct_neon;
-            c->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
         }
     }
 
@@ -319,12 +304,6 @@ void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
         c->avg_h264_qpel_pixels_tab[1][15] = ff_avg_h264_qpel8_mc33_neon;
     }
 
-    if (CONFIG_VP3_DECODER) {
-        c->vp3_v_loop_filter = ff_vp3_v_loop_filter_neon;
-        c->vp3_h_loop_filter = ff_vp3_h_loop_filter_neon;
-        c->vp3_idct_dc_add   = ff_vp3_idct_dc_add_neon;
-    }
-
     c->vector_fmul_window         = ff_vector_fmul_window_neon;
     c->vector_fmul_scalar         = ff_vector_fmul_scalar_neon;
     c->butterflies_float          = ff_butterflies_float_neon;
diff --git a/libavcodec/arm/vp3dsp_init_arm.c b/libavcodec/arm/vp3dsp_init_arm.c
new file mode 100644
index 0000000..90fc34b
--- /dev/null
+++ b/libavcodec/arm/vp3dsp_init_arm.c
@@ -0,0 +1,45 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stdint.h>
+
+#include "libavutil/attributes.h"
+#include "libavutil/cpu.h"
+#include "libavutil/arm/cpu.h"
+#include "libavcodec/vp3dsp.h"
+
+void ff_vp3_idct_put_neon(uint8_t *dest, int line_size, DCTELEM *data);
+void ff_vp3_idct_add_neon(uint8_t *dest, int line_size, DCTELEM *data);
+void ff_vp3_idct_dc_add_neon(uint8_t *dest, int line_size, const DCTELEM *data);
+
+void ff_vp3_v_loop_filter_neon(uint8_t *, int, int *);
+void ff_vp3_h_loop_filter_neon(uint8_t *, int, int *);
+
+av_cold void ff_vp3dsp_init_arm(VP3DSPContext *c, int flags)
+{
+    int cpu_flags = av_get_cpu_flags();
+
+    if (have_neon(cpu_flags)) {
+        c->idct_put      = ff_vp3_idct_put_neon;
+        c->idct_add      = ff_vp3_idct_add_neon;
+        c->idct_dc_add   = ff_vp3_idct_dc_add_neon;
+        c->v_loop_filter = ff_vp3_v_loop_filter_neon;
+        c->h_loop_filter = ff_vp3_h_loop_filter_neon;
+        c->idct_perm     = FF_TRANSPOSE_IDCT_PERM;
+    }
+}
diff --git a/libavcodec/arm/vp3dsp_neon.S b/libavcodec/arm/vp3dsp_neon.S
index 70cfd29..0c88562 100644
--- a/libavcodec/arm/vp3dsp_neon.S
+++ b/libavcodec/arm/vp3dsp_neon.S
@@ -260,32 +260,6 @@ endfunc
 VP3_IDCT_END row
 VP3_IDCT_END col
 
-function ff_vp3_idct_neon, export=1
-    mov             ip,  lr
-    mov             r2,  r0
-    bl              vp3_idct_start_neon
-    bl              vp3_idct_end_row_neon
-    mov             r3,  #8
-    bl              vp3_idct_core_neon
-    bl              vp3_idct_end_col_neon
-    mov             lr,  ip
-    vpop            {d8-d15}
-
-    vshr.s16        q8,  q8,  #4
-    vshr.s16        q9,  q9,  #4
-    vshr.s16        q10, q10, #4
-    vshr.s16        q11, q11, #4
-    vshr.s16        q12, q12, #4
-    vst1.64         {d16-d19}, [r0,:128]!
-    vshr.s16        q13, q13, #4
-    vshr.s16        q14, q14, #4
-    vst1.64         {d20-d23}, [r0,:128]!
-    vshr.s16        q15, q15, #4
-    vst1.64         {d24-d27}, [r0,:128]!
-    vst1.64         {d28-d31}, [r0,:128]!
-    bx              lr
-endfunc
-
 function ff_vp3_idct_put_neon, export=1
     mov             ip,  lr
     bl              vp3_idct_start_neon
diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 3577926..ef31431 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -2870,12 +2870,6 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
             c->idct_add= ff_jref_idct_add;
             c->idct    = ff_j_rev_dct;
             c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
-        }else if((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER || CONFIG_VP6_DECODER ) &&
-                avctx->idct_algo==FF_IDCT_VP3){
-            c->idct_put= ff_vp3_idct_put_c;
-            c->idct_add= ff_vp3_idct_add_c;
-            c->idct    = ff_vp3_idct_c;
-            c->idct_permutation_type= FF_NO_IDCT_PERM;
         }else if(avctx->idct_algo==FF_IDCT_WMV2){
             c->idct_put= ff_wmv2_idct_put_c;
             c->idct_add= ff_wmv2_idct_add_c;
@@ -3037,12 +3031,6 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
         c->h263_v_loop_filter= h263_v_loop_filter_c;
     }
 
-    if (CONFIG_VP3_DECODER) {
-        c->vp3_h_loop_filter= ff_vp3_h_loop_filter_c;
-        c->vp3_v_loop_filter= ff_vp3_v_loop_filter_c;
-        c->vp3_idct_dc_add= ff_vp3_idct_dc_add_c;
-    }
-
     c->h261_loop_filter= h261_loop_filter_c;
 
     c->try_8x8basis= try_8x8basis_c;
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index af11048..7b533cc 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -108,15 +108,6 @@ PUTAVG_PIXELS(14)
 #define ff_put_pixels16x16_c ff_put_pixels16x16_8_c
 #define ff_avg_pixels16x16_c ff_avg_pixels16x16_8_c
 
-/* VP3 DSP functions */
-void ff_vp3_idct_c(DCTELEM *block/* align 16*/);
-void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
-void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
-void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
-
-void ff_vp3_v_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
-void ff_vp3_h_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
-
 /* EA functions */
 void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
 
@@ -401,10 +392,6 @@ typedef struct DSPContext {
     void (*x8_v_loop_filter)(uint8_t *src, int stride, int qscale);
     void (*x8_h_loop_filter)(uint8_t *src, int stride, int qscale);
 
-    void (*vp3_idct_dc_add)(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
-    void (*vp3_v_loop_filter)(uint8_t *src, int stride, int *bounding_values);
-    void (*vp3_h_loop_filter)(uint8_t *src, int stride, int *bounding_values);
-
     /* assume len is a multiple of 4, and arrays are 16-byte aligned */
     void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
     void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
diff --git a/libavcodec/ppc/Makefile b/libavcodec/ppc/Makefile
index 31f4fb8..e5d1d39 100644
--- a/libavcodec/ppc/Makefile
+++ b/libavcodec/ppc/Makefile
@@ -1,14 +1,13 @@
 OBJS                                   += ppc/dsputil_ppc.o             \
 
+OBJS-$(CONFIG_VP3DSP)                  += ppc/vp3dsp_altivec.o
+
 FFT-OBJS-$(HAVE_GNU_AS)                += ppc/fft_altivec_s.o
 ALTIVEC-OBJS-$(CONFIG_FFT)             += ppc/fft_altivec.o             \
                                           $(FFT-OBJS-yes)
 ALTIVEC-OBJS-$(CONFIG_H264DSP)         += ppc/h264_altivec.o
 ALTIVEC-OBJS-$(CONFIG_MPEGAUDIODSP)    += ppc/mpegaudiodec_altivec.o
 ALTIVEC-OBJS-$(CONFIG_VC1_DECODER)     += ppc/vc1dsp_altivec.o
-ALTIVEC-OBJS-$(CONFIG_VP3_DECODER)     += ppc/vp3dsp_altivec.o
-ALTIVEC-OBJS-$(CONFIG_VP5_DECODER)     += ppc/vp3dsp_altivec.o
-ALTIVEC-OBJS-$(CONFIG_VP6_DECODER)     += ppc/vp3dsp_altivec.o
 ALTIVEC-OBJS-$(CONFIG_VP8_DECODER)     += ppc/vp8dsp_altivec.o
 
 ALTIVEC-OBJS                           += ppc/dsputil_altivec.o         \
diff --git a/libavcodec/ppc/dsputil_altivec.h b/libavcodec/ppc/dsputil_altivec.h
index 6c87782..0b5e404 100644
--- a/libavcodec/ppc/dsputil_altivec.h
+++ b/libavcodec/ppc/dsputil_altivec.h
@@ -36,10 +36,6 @@ void ff_gmc1_altivec(uint8_t *dst, uint8_t *src, int stride, int h,
 void ff_idct_put_altivec(uint8_t *dest, int line_size, int16_t *block);
 void ff_idct_add_altivec(uint8_t *dest, int line_size, int16_t *block);
 
-void ff_vp3_idct_altivec(DCTELEM *block);
-void ff_vp3_idct_put_altivec(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_vp3_idct_add_altivec(uint8_t *dest, int line_size, DCTELEM *block);
-
 void ff_dsputil_h264_init_ppc(DSPContext* c, AVCodecContext *avctx);
 
 void ff_dsputil_init_altivec(DSPContext* c, AVCodecContext *avctx);
diff --git a/libavcodec/ppc/dsputil_ppc.c b/libavcodec/ppc/dsputil_ppc.c
index 195aa20..c6fdc8e 100644
--- a/libavcodec/ppc/dsputil_ppc.c
+++ b/libavcodec/ppc/dsputil_ppc.c
@@ -193,12 +193,6 @@ void ff_dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
                 c->idct_put = ff_idct_put_altivec;
                 c->idct_add = ff_idct_add_altivec;
                 c->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
-            }else if((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER || CONFIG_VP6_DECODER) &&
-                     avctx->idct_algo==FF_IDCT_VP3){
-                c->idct_put = ff_vp3_idct_put_altivec;
-                c->idct_add = ff_vp3_idct_add_altivec;
-                c->idct     = ff_vp3_idct_altivec;
-                c->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
             }
         }
 
diff --git a/libavcodec/ppc/vp3dsp_altivec.c b/libavcodec/ppc/vp3dsp_altivec.c
index 950e5c7..ac00c93 100644
--- a/libavcodec/ppc/vp3dsp_altivec.c
+++ b/libavcodec/ppc/vp3dsp_altivec.c
@@ -18,6 +18,13 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#include "config.h"
+#include "libavutil/attributes.h"
+#include "libavutil/cpu.h"
+#include "libavcodec/vp3dsp.h"
+
+#if HAVE_ALTIVEC
+
 #include "libavutil/ppc/types_altivec.h"
 #include "libavutil/ppc/util_altivec.h"
 #include "libavcodec/dsputil.h"
@@ -107,25 +114,7 @@ static inline vec_s16 M16(vec_s16 a, vec_s16 C)
 #define ADD8(a) vec_add(a, eight)
 #define SHIFT4(a) vec_sra(a, four)
 
-void ff_vp3_idct_altivec(DCTELEM block[64])
-{
-    IDCT_START
-
-    IDCT_1D(NOP, NOP)
-    TRANSPOSE8(b0, b1, b2, b3, b4, b5, b6, b7);
-    IDCT_1D(ADD8, SHIFT4)
-
-    vec_st(b0, 0x00, block);
-    vec_st(b1, 0x10, block);
-    vec_st(b2, 0x20, block);
-    vec_st(b3, 0x30, block);
-    vec_st(b4, 0x40, block);
-    vec_st(b5, 0x50, block);
-    vec_st(b6, 0x60, block);
-    vec_st(b7, 0x70, block);
-}
-
-void ff_vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64])
+static void vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64])
 {
     vec_u8 t;
     IDCT_START
@@ -153,7 +142,7 @@ void ff_vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64])
     PUT(b7)
 }
 
-void ff_vp3_idct_add_altivec(uint8_t *dst, int stride, DCTELEM block[64])
+static void vp3_idct_add_altivec(uint8_t *dst, int stride, DCTELEM block[64])
 {
     LOAD_ZERO;
     vec_u8 t, vdst;
@@ -183,3 +172,16 @@ void ff_vp3_idct_add_altivec(uint8_t *dst, int stride, DCTELEM block[64])
     ADD(b6)     dst += stride;
     ADD(b7)
 }
+
+#endif /* HAVE_ALTIVEC */
+
+av_cold void ff_vp3dsp_init_ppc(VP3DSPContext *c, int flags)
+{
+#if HAVE_ALTIVEC
+    if (av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) {
+        c->idct_put  = vp3_idct_put_altivec;
+        c->idct_add  = vp3_idct_add_altivec;
+        c->idct_perm = FF_TRANSPOSE_IDCT_PERM;
+    }
+#endif
+}
diff --git a/libavcodec/tscc2.c b/libavcodec/tscc2.c
index 9c8d234..a8fd652 100644
--- a/libavcodec/tscc2.c
+++ b/libavcodec/tscc2.c
@@ -91,11 +91,11 @@ static av_cold int init_vlcs(TSCC2Context *c)
 #define DCT1D(d0, d1, d2, d3, s0, s1, s2, s3, OP) \
     OP(d0, 5 * ((s0) + (s1) + (s2)) + 2 * (s3));  \
     OP(d1, 5 * ((s0) - (s2) - (s3)) + 2 * (s1));  \
-    OP(d2, 5 * ((s0) - (s2) + (s3)) - 2 * (s3));  \
+    OP(d2, 5 * ((s0) - (s2) + (s3)) - 2 * (s1));  \
     OP(d3, 5 * ((s0) - (s1) + (s2)) - 2 * (s3));  \
 
 #define COL_OP(a, b)  a = b
-#define ROW_OP(a, b)  a = (((b) + 0x20) >> 6) + 0x80
+#define ROW_OP(a, b)  a = ((b) + 0x20) >> 6
 
 static void tscc2_idct4_put(int *in, int q[3], uint8_t *dst, int stride)
 {
@@ -158,7 +158,7 @@ static int tscc2_decode_mb(TSCC2Context *c, int *q, int vlc_set,
             }
             dc          = (dc + prev_dc) & 0xFF;
             prev_dc     = dc;
-            c->block[0] = dc - 0x80;
+            c->block[0] = dc;
 
             nc = get_vlc2(gb, c->nc_vlc[vlc_set].table, 9, 1);
             if (nc == -1)
diff --git a/libavcodec/vp3.c b/libavcodec/vp3.c
index 81f6b89..a3bfd74 100644
--- a/libavcodec/vp3.c
+++ b/libavcodec/vp3.c
@@ -40,6 +40,7 @@
 #include "get_bits.h"
 
 #include "vp3data.h"
+#include "vp3dsp.h"
 #include "xiph.h"
 #include "thread.h"
 
@@ -135,6 +136,7 @@ typedef struct Vp3DecodeContext {
     AVFrame current_frame;
     int keyframe;
     DSPContext dsp;
+    VP3DSPContext vp3dsp;
     int flipped_image;
     int last_slice_end;
     int skip_loop_filter;
@@ -1302,14 +1304,14 @@ static void apply_loop_filter(Vp3DecodeContext *s, int plane, int ystart, int ye
             {
                 /* do not perform left edge filter for left columns frags */
                 if (x > 0) {
-                    s->dsp.vp3_h_loop_filter(
+                    s->vp3dsp.h_loop_filter(
                         plane_data + 8*x,
                         stride, bounding_values);
                 }
 
                 /* do not perform top edge filter for top row fragments */
                 if (y > 0) {
-                    s->dsp.vp3_v_loop_filter(
+                    s->vp3dsp.v_loop_filter(
                         plane_data + 8*x,
                         stride, bounding_values);
                 }
@@ -1319,7 +1321,7 @@ static void apply_loop_filter(Vp3DecodeContext *s, int plane, int ystart, int ye
                  * in this frame (it will be filtered in next iteration) */
                 if ((x < width - 1) &&
                     (s->all_fragments[fragment + 1].coding_method == MODE_COPY)) {
-                    s->dsp.vp3_h_loop_filter(
+                    s->vp3dsp.h_loop_filter(
                         plane_data + 8*x + 8,
                         stride, bounding_values);
                 }
@@ -1329,7 +1331,7 @@ static void apply_loop_filter(Vp3DecodeContext *s, int plane, int ystart, int ye
                  * in this frame (it will be filtered in the next row) */
                 if ((y < height - 1) &&
                     (s->all_fragments[fragment + width].coding_method == MODE_COPY)) {
-                    s->dsp.vp3_v_loop_filter(
+                    s->vp3dsp.v_loop_filter(
                         plane_data + 8*x + 8*stride,
                         stride, bounding_values);
                 }
@@ -1574,20 +1576,18 @@ static void render_slice(Vp3DecodeContext *s, int slice)
 
                     if (s->all_fragments[i].coding_method == MODE_INTRA) {
                         vp3_dequant(s, s->all_fragments + i, plane, 0, block);
-                        if(s->avctx->idct_algo!=FF_IDCT_VP3)
-                            block[0] += 128<<3;
-                        s->dsp.idct_put(
+                        s->vp3dsp.idct_put(
                             output_plane + first_pixel,
                             stride,
                             block);
                     } else {
                         if (vp3_dequant(s, s->all_fragments + i, plane, 1, block)) {
-                        s->dsp.idct_add(
+                        s->vp3dsp.idct_add(
                             output_plane + first_pixel,
                             stride,
                             block);
                         } else {
-                            s->dsp.vp3_idct_dc_add(output_plane + first_pixel, stride, block);
+                            s->vp3dsp.idct_dc_add(output_plane + first_pixel, stride, block);
                         }
                     }
                 } else {
@@ -1670,10 +1670,10 @@ static av_cold int vp3_decode_init(AVCodecContext *avctx)
     if (avctx->codec_id != CODEC_ID_THEORA)
         avctx->pix_fmt = PIX_FMT_YUV420P;
     avctx->chroma_sample_location = AVCHROMA_LOC_CENTER;
-    if(avctx->idct_algo==FF_IDCT_AUTO)
-        avctx->idct_algo=FF_IDCT_VP3;
     ff_dsputil_init(&s->dsp, avctx);
+    ff_vp3dsp_init(&s->vp3dsp, avctx->flags);
 
+    ff_init_scantable_permutation(s->dsp.idct_permutation, s->vp3dsp.idct_perm);
     ff_init_scantable(s->dsp.idct_permutation, &s->scantable, ff_zigzag_direct);
 
     /* initialize to an impossible value which will force a recalculation
diff --git a/libavcodec/vp3dsp.c b/libavcodec/vp3dsp.c
index 9fded0f..4c5ff6a 100644
--- a/libavcodec/vp3dsp.c
+++ b/libavcodec/vp3dsp.c
@@ -24,8 +24,10 @@
  * source code.
  */
 
+#include "libavutil/attributes.h"
 #include "avcodec.h"
 #include "dsputil.h"
+#include "vp3dsp.h"
 
 #define IdctAdjustBeforeShift 8
 #define xC1S7 64277
@@ -210,19 +212,16 @@ static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int
     }
 }
 
-void ff_vp3_idct_c(DCTELEM *block/* align 16*/){
-    idct(NULL, 0, block, 0);
-}
-
-void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
+static void vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
     idct(dest, line_size, block, 1);
 }
 
-void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
+static void vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
     idct(dest, line_size, block, 2);
 }
 
-void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/){
+static void vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size,
+                              const DCTELEM *block/*align 16*/){
     int i, dc = (block[0] + 15) >> 5;
 
     for(i = 0; i < 8; i++){
@@ -238,7 +237,8 @@ void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM
     }
 }
 
-void ff_vp3_v_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_values)
+static void vp3_v_loop_filter_c(uint8_t *first_pixel, int stride,
+                                int *bounding_values)
 {
     unsigned char *end;
     int filter_value;
@@ -254,7 +254,8 @@ void ff_vp3_v_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_valu
     }
 }
 
-void ff_vp3_h_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_values)
+static void vp3_h_loop_filter_c(uint8_t *first_pixel, int stride,
+                                int *bounding_values)
 {
     unsigned char *end;
     int filter_value;
@@ -268,3 +269,21 @@ void ff_vp3_h_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_valu
         first_pixel[ 0] = av_clip_uint8(first_pixel[ 0] - filter_value);
     }
 }
+
+av_cold void ff_vp3dsp_init(VP3DSPContext *c, int flags)
+{
+    c->idct_put      = vp3_idct_put_c;
+    c->idct_add      = vp3_idct_add_c;
+    c->idct_dc_add   = vp3_idct_dc_add_c;
+    c->v_loop_filter = vp3_v_loop_filter_c;
+    c->h_loop_filter = vp3_h_loop_filter_c;
+
+    c->idct_perm = FF_NO_IDCT_PERM;
+
+    if (ARCH_ARM)
+        ff_vp3dsp_init_arm(c, flags);
+    if (ARCH_PPC)
+        ff_vp3dsp_init_ppc(c, flags);
+    if (ARCH_X86)
+        ff_vp3dsp_init_x86(c, flags);
+}
diff --git a/libavcodec/vp3dsp.h b/libavcodec/vp3dsp.h
new file mode 100644
index 0000000..a14dec1
--- /dev/null
+++ b/libavcodec/vp3dsp.h
@@ -0,0 +1,40 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_VP3DSP_H
+#define AVCODEC_VP3DSP_H
+
+#include <stdint.h>
+#include "dsputil.h"
+
+typedef struct VP3DSPContext {
+    void (*idct_put)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*idct_add)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*idct_dc_add)(uint8_t *dest, int line_size, const DCTELEM *block);
+    void (*v_loop_filter)(uint8_t *src, int stride, int *bounding_values);
+    void (*h_loop_filter)(uint8_t *src, int stride, int *bounding_values);
+
+    int idct_perm;
+} VP3DSPContext;
+
+void ff_vp3dsp_init(VP3DSPContext *c, int flags);
+void ff_vp3dsp_init_arm(VP3DSPContext *c, int flags);
+void ff_vp3dsp_init_ppc(VP3DSPContext *c, int flags);
+void ff_vp3dsp_init_x86(VP3DSPContext *c, int flags);
+
+#endif /* AVCODEC_VP3DSP_H */
diff --git a/libavcodec/vp56.c b/libavcodec/vp56.c
index c6e32af..f98d22c 100644
--- a/libavcodec/vp56.c
+++ b/libavcodec/vp56.c
@@ -411,7 +411,7 @@ static void vp56_decode_mb(VP56Context *s, int row, int col, int is_alpha)
         case VP56_MB_INTRA:
             for (b=0; b<b_max; b++) {
                 plane = ff_vp56_b2p[b+ab];
-                s->dsp.idct_put(frame_current->data[plane] + s->block_offset[b],
+                s->vp3dsp.idct_put(frame_current->data[plane] + s->block_offset[b],
                                 s->stride[plane], s->block_coeff[b]);
             }
             break;
@@ -424,7 +424,7 @@ static void vp56_decode_mb(VP56Context *s, int row, int col, int is_alpha)
                 s->dsp.put_pixels_tab[1][0](frame_current->data[plane] + off,
                                             frame_ref->data[plane] + off,
                                             s->stride[plane], 8);
-                s->dsp.idct_add(frame_current->data[plane] + off,
+                s->vp3dsp.idct_add(frame_current->data[plane] + off,
                                 s->stride[plane], s->block_coeff[b]);
             }
             break;
@@ -442,7 +442,7 @@ static void vp56_decode_mb(VP56Context *s, int row, int col, int is_alpha)
                 plane = ff_vp56_b2p[b+ab];
                 vp56_mc(s, b, plane, frame_ref->data[plane], s->stride[plane],
                         16*col+x_off, 16*row+y_off);
-                s->dsp.idct_add(frame_current->data[plane] + s->block_offset[b],
+                s->vp3dsp.idct_add(frame_current->data[plane] + s->block_offset[b],
                                 s->stride[plane], s->block_coeff[b]);
             }
             break;
@@ -666,10 +666,10 @@ av_cold void ff_vp56_init(AVCodecContext *avctx, int flip, int has_alpha)
     s->avctx = avctx;
     avctx->pix_fmt = has_alpha ? PIX_FMT_YUVA420P : PIX_FMT_YUV420P;
 
-    if (avctx->idct_algo == FF_IDCT_AUTO)
-        avctx->idct_algo = FF_IDCT_VP3;
     ff_dsputil_init(&s->dsp, avctx);
+    ff_vp3dsp_init(&s->vp3dsp, avctx->flags);
     ff_vp56dsp_init(&s->vp56dsp, avctx->codec->id);
+    ff_init_scantable_permutation(s->dsp.idct_permutation, s->vp3dsp.idct_perm);
     ff_init_scantable(s->dsp.idct_permutation, &s->scantable,ff_zigzag_direct);
 
     for (i=0; i<4; i++) {
diff --git a/libavcodec/vp56.h b/libavcodec/vp56.h
index e135718..12f9380 100644
--- a/libavcodec/vp56.h
+++ b/libavcodec/vp56.h
@@ -30,6 +30,7 @@
 #include "dsputil.h"
 #include "get_bits.h"
 #include "bytestream.h"
+#include "vp3dsp.h"
 #include "vp56dsp.h"
 
 typedef struct vp56_context VP56Context;
@@ -91,6 +92,7 @@ typedef struct {
 struct vp56_context {
     AVCodecContext *avctx;
     DSPContext dsp;
+    VP3DSPContext vp3dsp;
     VP56DSPContext vp56dsp;
     ScanTable scantable;
     AVFrame frames[4];
diff --git a/libavcodec/x86/Makefile b/libavcodec/x86/Makefile
index 8acbd07..e18074a 100644
--- a/libavcodec/x86/Makefile
+++ b/libavcodec/x86/Makefile
@@ -1,5 +1,6 @@
 OBJS-$(CONFIG_MLP_DECODER)             += x86/mlpdsp.o
 OBJS-$(CONFIG_TRUEHD_DECODER)          += x86/mlpdsp.o
+OBJS-$(CONFIG_VP3DSP)                  += x86/vp3dsp_init.o
 OBJS-$(CONFIG_XMM_CLOBBER_TEST)        += x86/w64xmmtest.o
 
 MMX-OBJS                               += x86/dsputil_mmx.o             \
@@ -67,10 +68,8 @@ YASM-OBJS-$(CONFIG_RV40_DECODER)       += x86/rv34dsp.o                 \
                                           x86/rv40dsp.o
 YASM-OBJS-$(CONFIG_V210_DECODER)       += x86/v210.o
 YASM-OBJS-$(CONFIG_VC1_DECODER)        += x86/vc1dsp_yasm.o
-YASM-OBJS-$(CONFIG_VP3_DECODER)        += x86/vp3dsp.o
-YASM-OBJS-$(CONFIG_VP5_DECODER)        += x86/vp3dsp.o
-YASM-OBJS-$(CONFIG_VP6_DECODER)        += x86/vp3dsp.o                  \
-                                          x86/vp56dsp.o
+YASM-OBJS-$(CONFIG_VP3DSP)             += x86/vp3dsp.o
+YASM-OBJS-$(CONFIG_VP6_DECODER)        += x86/vp56dsp.o
 YASM-OBJS-$(CONFIG_VP8_DECODER)        += x86/vp8dsp.o
 
 YASM-OBJS                              += x86/dsputil_yasm.o            \
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index e829bbd..8049bde 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -2574,20 +2574,6 @@ static void vector_clipf_sse(float *dst, const float *src,
     );
 }
 
-void ff_vp3_idct_mmx(int16_t *input_data);
-void ff_vp3_idct_put_mmx(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_vp3_idct_add_mmx(uint8_t *dest, int line_size, DCTELEM *block);
-
-void ff_vp3_idct_dc_add_mmx2(uint8_t *dest, int line_size,
-                             const DCTELEM *block);
-
-void ff_vp3_v_loop_filter_mmx2(uint8_t *src, int stride, int *bounding_values);
-void ff_vp3_h_loop_filter_mmx2(uint8_t *src, int stride, int *bounding_values);
-
-void ff_vp3_idct_sse2(int16_t *input_data);
-void ff_vp3_idct_put_sse2(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_vp3_idct_add_sse2(uint8_t *dest, int line_size, DCTELEM *block);
-
 int32_t ff_scalarproduct_int16_mmx2(const int16_t *v1, const int16_t *v2,
                                     int order);
 int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2,
@@ -2782,14 +2768,7 @@ static void dsputil_init_mmx2(DSPContext *c, AVCodecContext *avctx,
             c->avg_pixels_tab[0][3] = avg_pixels16_xy2_mmx2;
             c->avg_pixels_tab[1][3] = avg_pixels8_xy2_mmx2;
         }
-
-        if (CONFIG_VP3_DECODER && HAVE_YASM) {
-            c->vp3_v_loop_filter = ff_vp3_v_loop_filter_mmx2;
-            c->vp3_h_loop_filter = ff_vp3_h_loop_filter_mmx2;
-        }
     }
-    if (CONFIG_VP3_DECODER && HAVE_YASM)
-        c->vp3_idct_dc_add = ff_vp3_idct_dc_add_mmx2;
 
     if (CONFIG_VP3_DECODER && (avctx->codec_id == CODEC_ID_VP3 ||
                                avctx->codec_id == CODEC_ID_THEORA)) {
@@ -3165,20 +3144,6 @@ void ff_dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx)
                 }
                 c->idct_permutation_type = FF_LIBMPEG2_IDCT_PERM;
 #endif
-            } else if ((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER ||
-                        CONFIG_VP6_DECODER) &&
-                       idct_algo == FF_IDCT_VP3 && HAVE_YASM) {
-                if (mm_flags & AV_CPU_FLAG_SSE2) {
-                    c->idct_put              = ff_vp3_idct_put_sse2;
-                    c->idct_add              = ff_vp3_idct_add_sse2;
-                    c->idct                  = ff_vp3_idct_sse2;
-                    c->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
-                } else {
-                    c->idct_put              = ff_vp3_idct_put_mmx;
-                    c->idct_add              = ff_vp3_idct_add_mmx;
-                    c->idct                  = ff_vp3_idct_mmx;
-                    c->idct_permutation_type = FF_PARTTRANS_IDCT_PERM;
-                }
             } else if (idct_algo == FF_IDCT_CAVS) {
                     c->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
             } else if (idct_algo == FF_IDCT_XVIDMMX) {
diff --git a/libavcodec/x86/vp3dsp.asm b/libavcodec/x86/vp3dsp.asm
index 99621fb..0e0bd29 100644
--- a/libavcodec/x86/vp3dsp.asm
+++ b/libavcodec/x86/vp3dsp.asm
@@ -524,10 +524,6 @@ cglobal vp3_h_loop_filter_mmx2, 3, 4
 %endmacro
 
 %macro vp3_idct_funcs 3
-cglobal vp3_idct_%1, 1, 1, %2
-    VP3_IDCT_%1   r0
-    RET
-
 cglobal vp3_idct_put_%1, 3, %3, %2
     VP3_IDCT_%1   r2
 %if ARCH_X86_64
diff --git a/libavcodec/x86/vp3dsp_init.c b/libavcodec/x86/vp3dsp_init.c
new file mode 100644
index 0000000..3ae2a90
--- /dev/null
+++ b/libavcodec/x86/vp3dsp_init.c
@@ -0,0 +1,65 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stdint.h>
+
+#include "libavutil/attributes.h"
+#include "libavutil/cpu.h"
+#include "libavcodec/avcodec.h"
+#include "libavcodec/vp3dsp.h"
+#include "config.h"
+
+void ff_vp3_idct_put_mmx(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_vp3_idct_add_mmx(uint8_t *dest, int line_size, DCTELEM *block);
+
+void ff_vp3_idct_put_sse2(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_vp3_idct_add_sse2(uint8_t *dest, int line_size, DCTELEM *block);
+
+void ff_vp3_idct_dc_add_mmx2(uint8_t *dest, int line_size,
+                             const DCTELEM *block);
+
+void ff_vp3_v_loop_filter_mmx2(uint8_t *src, int stride, int *bounding_values);
+void ff_vp3_h_loop_filter_mmx2(uint8_t *src, int stride, int *bounding_values);
+
+av_cold void ff_vp3dsp_init_x86(VP3DSPContext *c, int flags)
+{
+#if HAVE_YASM
+    int cpuflags = av_get_cpu_flags();
+
+    if (HAVE_MMX && cpuflags & AV_CPU_FLAG_MMX) {
+        c->idct_put  = ff_vp3_idct_put_mmx;
+        c->idct_add  = ff_vp3_idct_add_mmx;
+        c->idct_perm = FF_PARTTRANS_IDCT_PERM;
+    }
+
+    if (HAVE_MMX2 && cpuflags & AV_CPU_FLAG_MMX2) {
+        c->idct_dc_add = ff_vp3_idct_dc_add_mmx2;
+
+        if (!(flags & CODEC_FLAG_BITEXACT)) {
+            c->v_loop_filter = ff_vp3_v_loop_filter_mmx2;
+            c->h_loop_filter = ff_vp3_h_loop_filter_mmx2;
+        }
+    }
+
+    if (cpuflags & AV_CPU_FLAG_SSE2) {
+        c->idct_put  = ff_vp3_idct_put_sse2;
+        c->idct_add  = ff_vp3_idct_add_sse2;
+        c->idct_perm = FF_TRANSPOSE_IDCT_PERM;
+    }
+#endif
+}