dsputil: Split off H.263 bits into their own H263DSPContext

configure

@@ -1384,6 +1384,7 @@ CONFIG_EXTRA="
     gcrypt
     golomb
     gplv3
+    h263dsp
     h264chroma
     h264dsp
     h264pred
@@ -1598,8 +1599,8 @@ g2m_decoder_deps="zlib"
 g2m_decoder_select="dsputil"
 h261_decoder_select="error_resilience mpegvideo"
 h261_encoder_select="aandcttables mpegvideoenc"
-h263_decoder_select="error_resilience h263_parser mpegvideo"
-h263_encoder_select="aandcttables mpegvideoenc"
+h263_decoder_select="error_resilience h263_parser h263dsp mpegvideo"
+h263_encoder_select="aandcttables h263dsp mpegvideoenc"
 h263i_decoder_select="h263_decoder"
 h263p_encoder_select="h263_encoder"
 h264_decoder_select="golomb h264chroma h264dsp h264pred h264qpel videodsp"
@@ -1665,9 +1666,9 @@ qcelp_decoder_select="lsp"
 qdm2_decoder_select="mdct rdft mpegaudiodsp"
 ra_144_encoder_select="audio_frame_queue lpc"
 ralf_decoder_select="golomb"
-rv10_decoder_select="error_resilience h263_decoder"
+rv10_decoder_select="error_resilience h263_decoder h263dsp"
 rv10_encoder_select="h263_encoder"
-rv20_decoder_select="error_resilience h263_decoder"
+rv20_decoder_select="error_resilience h263_decoder h263dsp"
 rv20_encoder_select="h263_encoder"
 rv30_decoder_select="error_resilience golomb h264chroma h264pred h264qpel mpegvideo videodsp"
 rv40_decoder_select="error_resilience golomb h264chroma h264pred h264qpel mpegvideo videodsp"

libavcodec/Makefile

@@ -41,6 +41,7 @@ FFT-OBJS-$(CONFIG_HARDCODED_TABLES)    += cos_tables.o cos_fixed_tables.o
 OBJS-$(CONFIG_FFT)                     += avfft.o fft_fixed.o fft_float.o \
                                           $(FFT-OBJS-yes)
 OBJS-$(CONFIG_GOLOMB)                  += golomb.o
+OBJS-$(CONFIG_H263DSP)                 += h263dsp.o
 OBJS-$(CONFIG_H264CHROMA)              += h264chroma.o
 OBJS-$(CONFIG_H264DSP)                 += h264dsp.o h264idct.o
 OBJS-$(CONFIG_H264PRED)                += h264pred.o

libavcodec/dsputil.c

@@ -1409,80 +1409,6 @@ static void put_mspel8_mc22_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)
     wmv2_mspel8_v_lowpass(dst, halfH+8, stride, 8, 8);
 }
 
-static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
-    if(CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
-    int x;
-    const int strength= ff_h263_loop_filter_strength[qscale];
-
-    for(x=0; x<8; x++){
-        int d1, d2, ad1;
-        int p0= src[x-2*stride];
-        int p1= src[x-1*stride];
-        int p2= src[x+0*stride];
-        int p3= src[x+1*stride];
-        int d = (p0 - p3 + 4*(p2 - p1)) / 8;
-
-        if     (d<-2*strength) d1= 0;
-        else if(d<-  strength) d1=-2*strength - d;
-        else if(d<   strength) d1= d;
-        else if(d< 2*strength) d1= 2*strength - d;
-        else                   d1= 0;
-
-        p1 += d1;
-        p2 -= d1;
-        if(p1&256) p1= ~(p1>>31);
-        if(p2&256) p2= ~(p2>>31);
-
-        src[x-1*stride] = p1;
-        src[x+0*stride] = p2;
-
-        ad1= FFABS(d1)>>1;
-
-        d2= av_clip((p0-p3)/4, -ad1, ad1);
-
-        src[x-2*stride] = p0 - d2;
-        src[x+  stride] = p3 + d2;
-    }
-    }
-}
-
-static void h263_h_loop_filter_c(uint8_t *src, int stride, int qscale){
-    if(CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
-    int y;
-    const int strength= ff_h263_loop_filter_strength[qscale];
-
-    for(y=0; y<8; y++){
-        int d1, d2, ad1;
-        int p0= src[y*stride-2];
-        int p1= src[y*stride-1];
-        int p2= src[y*stride+0];
-        int p3= src[y*stride+1];
-        int d = (p0 - p3 + 4*(p2 - p1)) / 8;
-
-        if     (d<-2*strength) d1= 0;
-        else if(d<-  strength) d1=-2*strength - d;
-        else if(d<   strength) d1= d;
-        else if(d< 2*strength) d1= 2*strength - d;
-        else                   d1= 0;
-
-        p1 += d1;
-        p2 -= d1;
-        if(p1&256) p1= ~(p1>>31);
-        if(p2&256) p2= ~(p2>>31);
-
-        src[y*stride-1] = p1;
-        src[y*stride+0] = p2;
-
-        ad1= FFABS(d1)>>1;
-
-        d2= av_clip((p0-p3)/4, -ad1, ad1);
-
-        src[y*stride-2] = p0 - d2;
-        src[y*stride+1] = p3 + d2;
-    }
-    }
-}
-
 static inline int pix_abs16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int s, i;
@@ -2701,11 +2627,6 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->bswap_buf= bswap_buf;
     c->bswap16_buf = bswap16_buf;
 
-    if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
-        c->h263_h_loop_filter= h263_h_loop_filter_c;
-        c->h263_v_loop_filter= h263_v_loop_filter_c;
-    }
-
     c->try_8x8basis= try_8x8basis_c;
     c->add_8x8basis= add_8x8basis_c;
 

libavcodec/dsputil.h

@@ -202,9 +202,6 @@ typedef struct DSPContext {
     void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w);
     void (*bswap16_buf)(uint16_t *dst, const uint16_t *src, int len);
 
-    void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
-    void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
-
     /* assume len is a multiple of 8, and arrays are 16-byte aligned */
     void (*vector_clipf)(float *dst /* align 16 */, const float *src /* align 16 */, float min, float max, int len /* align 16 */);
 

libavcodec/h263.c

@@ -152,8 +152,8 @@ void ff_h263_loop_filter(MpegEncContext * s){
     */
     if (!IS_SKIP(s->current_picture.mb_type[xy])) {
         qp_c= s->qscale;
-        s->dsp.h263_v_loop_filter(dest_y+8*linesize  , linesize, qp_c);
-        s->dsp.h263_v_loop_filter(dest_y+8*linesize+8, linesize, qp_c);
+        s->h263dsp.h263_v_loop_filter(dest_y + 8 * linesize,     linesize, qp_c);
+        s->h263dsp.h263_v_loop_filter(dest_y + 8 * linesize + 8, linesize, qp_c);
     }else
         qp_c= 0;
 
@@ -172,15 +172,15 @@ void ff_h263_loop_filter(MpegEncContext * s){
 
         if(qp_tc){
             const int chroma_qp= s->chroma_qscale_table[qp_tc];
-            s->dsp.h263_v_loop_filter(dest_y  ,   linesize, qp_tc);
-            s->dsp.h263_v_loop_filter(dest_y+8,   linesize, qp_tc);
+            s->h263dsp.h263_v_loop_filter(dest_y,     linesize, qp_tc);
+            s->h263dsp.h263_v_loop_filter(dest_y + 8, linesize, qp_tc);
 
-            s->dsp.h263_v_loop_filter(dest_cb , uvlinesize, chroma_qp);
-            s->dsp.h263_v_loop_filter(dest_cr , uvlinesize, chroma_qp);
+            s->h263dsp.h263_v_loop_filter(dest_cb, uvlinesize, chroma_qp);
+            s->h263dsp.h263_v_loop_filter(dest_cr, uvlinesize, chroma_qp);
         }
 
         if(qp_tt)
-            s->dsp.h263_h_loop_filter(dest_y-8*linesize+8  ,   linesize, qp_tt);
+            s->h263dsp.h263_h_loop_filter(dest_y - 8 * linesize + 8, linesize, qp_tt);
 
         if(s->mb_x){
             if (qp_tt || IS_SKIP(s->current_picture.mb_type[xy - 1 - s->mb_stride]))
@@ -190,17 +190,17 @@ void ff_h263_loop_filter(MpegEncContext * s){
 
             if(qp_dt){
                 const int chroma_qp= s->chroma_qscale_table[qp_dt];
-                s->dsp.h263_h_loop_filter(dest_y -8*linesize  ,   linesize, qp_dt);
-                s->dsp.h263_h_loop_filter(dest_cb-8*uvlinesize, uvlinesize, chroma_qp);
-                s->dsp.h263_h_loop_filter(dest_cr-8*uvlinesize, uvlinesize, chroma_qp);
+                s->h263dsp.h263_h_loop_filter(dest_y  - 8 * linesize,   linesize,   qp_dt);
+                s->h263dsp.h263_h_loop_filter(dest_cb - 8 * uvlinesize, uvlinesize, chroma_qp);
+                s->h263dsp.h263_h_loop_filter(dest_cr - 8 * uvlinesize, uvlinesize, chroma_qp);
             }
         }
     }
 
     if(qp_c){
-        s->dsp.h263_h_loop_filter(dest_y +8,   linesize, qp_c);
+        s->h263dsp.h263_h_loop_filter(dest_y + 8, linesize, qp_c);
         if(s->mb_y + 1 == s->mb_height)
-            s->dsp.h263_h_loop_filter(dest_y+8*linesize+8,   linesize, qp_c);
+            s->h263dsp.h263_h_loop_filter(dest_y + 8 * linesize + 8, linesize, qp_c);
     }
 
     if(s->mb_x){
@@ -211,12 +211,12 @@ void ff_h263_loop_filter(MpegEncContext * s){
             qp_lc = s->current_picture.qscale_table[xy - 1];
 
         if(qp_lc){
-            s->dsp.h263_h_loop_filter(dest_y,   linesize, qp_lc);
+            s->h263dsp.h263_h_loop_filter(dest_y, linesize, qp_lc);
             if(s->mb_y + 1 == s->mb_height){
                 const int chroma_qp= s->chroma_qscale_table[qp_lc];
-                s->dsp.h263_h_loop_filter(dest_y +8*  linesize,   linesize, qp_lc);
-                s->dsp.h263_h_loop_filter(dest_cb             , uvlinesize, chroma_qp);
-                s->dsp.h263_h_loop_filter(dest_cr             , uvlinesize, chroma_qp);
+                s->h263dsp.h263_h_loop_filter(dest_y + 8 * linesize, linesize, qp_lc);
+                s->h263dsp.h263_h_loop_filter(dest_cb, uvlinesize, chroma_qp);
+                s->h263dsp.h263_h_loop_filter(dest_cr, uvlinesize, chroma_qp);
             }
         }
     }

libavcodec/h263data.h

@@ -272,11 +272,6 @@ uint8_t ff_mba_length[7]={
       6,   7,   9,  11,  13,  14,  14
 };
 
-const uint8_t ff_h263_loop_filter_strength[32]={
-//  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
-    0, 1, 1, 2, 2, 3, 3, 4, 4, 4, 5, 5, 6, 6, 7, 7, 7, 8, 8, 8, 9, 9, 9,10,10,10,11,11,11,12,12,12
-};
-
 const AVRational ff_h263_pixel_aspect[16]={
  {0, 1},
  {1, 1},

libavcodec/h263dec.c

@@ -116,6 +116,7 @@ av_cold int ff_h263_decode_init(AVCodecContext *avctx)
         if ((ret = ff_MPV_common_init(s)) < 0)
             return ret;
 
+    ff_h263dsp_init(&s->h263dsp);
     ff_h263_decode_init_vlc();
 
     return 0;

libavcodec/h263dsp.c

@@ -0,0 +1,108 @@
+/*
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; 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/common.h"
+#include "config.h"
+#include "h263dsp.h"
+
+const uint8_t ff_h263_loop_filter_strength[32]={
+//  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
+    0, 1, 1, 2, 2, 3, 3, 4, 4, 4, 5, 5, 6, 6, 7, 7, 7, 8, 8, 8, 9, 9, 9,10,10,10,11,11,11,12,12,12
+};
+
+static void h263_h_loop_filter_c(uint8_t *src, int stride, int qscale){
+    int y;
+    const int strength= ff_h263_loop_filter_strength[qscale];
+
+    for(y=0; y<8; y++){
+        int d1, d2, ad1;
+        int p0= src[y*stride-2];
+        int p1= src[y*stride-1];
+        int p2= src[y*stride+0];
+        int p3= src[y*stride+1];
+        int d = (p0 - p3 + 4*(p2 - p1)) / 8;
+
+        if     (d<-2*strength) d1= 0;
+        else if(d<-  strength) d1=-2*strength - d;
+        else if(d<   strength) d1= d;
+        else if(d< 2*strength) d1= 2*strength - d;
+        else                   d1= 0;
+
+        p1 += d1;
+        p2 -= d1;
+        if(p1&256) p1= ~(p1>>31);
+        if(p2&256) p2= ~(p2>>31);
+
+        src[y*stride-1] = p1;
+        src[y*stride+0] = p2;
+
+        ad1= FFABS(d1)>>1;
+
+        d2= av_clip((p0-p3)/4, -ad1, ad1);
+
+        src[y*stride-2] = p0 - d2;
+        src[y*stride+1] = p3 + d2;
+    }
+}
+
+static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
+    int x;
+    const int strength= ff_h263_loop_filter_strength[qscale];
+
+    for(x=0; x<8; x++){
+        int d1, d2, ad1;
+        int p0= src[x-2*stride];
+        int p1= src[x-1*stride];
+        int p2= src[x+0*stride];
+        int p3= src[x+1*stride];
+        int d = (p0 - p3 + 4*(p2 - p1)) / 8;
+
+        if     (d<-2*strength) d1= 0;
+        else if(d<-  strength) d1=-2*strength - d;
+        else if(d<   strength) d1= d;
+        else if(d< 2*strength) d1= 2*strength - d;
+        else                   d1= 0;
+
+        p1 += d1;
+        p2 -= d1;
+        if(p1&256) p1= ~(p1>>31);
+        if(p2&256) p2= ~(p2>>31);
+
+        src[x-1*stride] = p1;
+        src[x+0*stride] = p2;
+
+        ad1= FFABS(d1)>>1;
+
+        d2= av_clip((p0-p3)/4, -ad1, ad1);
+
+        src[x-2*stride] = p0 - d2;
+        src[x+  stride] = p3 + d2;
+    }
+}
+
+av_cold void ff_h263dsp_init(H263DSPContext *ctx)
+{
+    ctx->h263_h_loop_filter = h263_h_loop_filter_c;
+    ctx->h263_v_loop_filter = h263_v_loop_filter_c;
+
+    if (ARCH_X86)
+        ff_h263dsp_init_x86(ctx);
+}

libavcodec/h263dsp.h

@@ -0,0 +1,34 @@
+/*
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_H263DSP_H
+#define AVCODEC_H263DSP_H
+
+#include <stdint.h>
+
+extern const uint8_t ff_h263_loop_filter_strength[32];
+
+typedef struct H263DSPContext {
+    void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
+    void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
+} H263DSPContext;
+
+void ff_h263dsp_init(H263DSPContext *ctx);
+void ff_h263dsp_init_x86(H263DSPContext *ctx);
+
+#endif /* AVCODEC_H263DSP_H */

libavcodec/mpegvideo.h

@@ -32,6 +32,7 @@
 #include "dsputil.h"
 #include "error_resilience.h"
 #include "get_bits.h"
+#include "h263dsp.h"
 #include "hpeldsp.h"
 #include "put_bits.h"
 #include "ratecontrol.h"
@@ -384,6 +385,7 @@ typedef struct MpegEncContext {
     DSPContext dsp;             ///< pointers for accelerated dsp functions
     HpelDSPContext hdsp;
     VideoDSPContext vdsp;
+    H263DSPContext h263dsp;
     int f_code;                 ///< forward MV resolution
     int b_code;                 ///< backward MV resolution for B Frames (mpeg4)
     int16_t (*p_mv_table_base)[2];
@@ -896,7 +898,6 @@ void ff_mpeg1_encode_slice_header(MpegEncContext *s);
 
 extern const uint8_t ff_aic_dc_scale_table[32];
 extern const uint8_t ff_h263_chroma_qscale_table[32];
-extern const uint8_t ff_h263_loop_filter_strength[32];
 
 /* rv10.c */
 void ff_rv10_encode_picture_header(MpegEncContext *s, int picture_number);

libavcodec/mpegvideo_enc.c

@@ -707,6 +707,7 @@ av_cold int ff_MPV_encode_init(AVCodecContext *avctx)
     if (ARCH_X86)
         ff_MPV_encode_init_x86(s);
 
+    ff_h263dsp_init(&s->h263dsp);
     if (!s->dct_quantize)
         s->dct_quantize = ff_dct_quantize_c;
     if (!s->denoise_dct)

libavcodec/rv10.c

@@ -484,6 +484,7 @@ static av_cold int rv10_decode_init(AVCodecContext *avctx)
     if ((ret = ff_MPV_common_init(s)) < 0)
         return ret;
 
+    ff_h263dsp_init(&s->h263dsp);
     ff_h263_decode_init_vlc();
 
     /* init rv vlc */

libavcodec/x86/Makefile

@@ -12,6 +12,7 @@ OBJS-$(CONFIG_ENCODERS)                += x86/dsputilenc_mmx.o          \
                                           x86/fdct.o                    \
                                           x86/motion_est.o
 OBJS-$(CONFIG_FFT)                     += x86/fft_init.o
+OBJS-$(CONFIG_H263DSP)                 += x86/h263dsp_init.o
 OBJS-$(CONFIG_H264CHROMA)              += x86/h264chroma_init.o
 OBJS-$(CONFIG_H264DSP)                 += x86/h264dsp_init.o
 OBJS-$(CONFIG_H264PRED)                += x86/h264_intrapred_init.o
@@ -59,8 +60,7 @@ YASM-OBJS-$(CONFIG_DSPUTIL)            += x86/dsputil.o                 \
                                           x86/qpel.o
 YASM-OBJS-$(CONFIG_ENCODERS)           += x86/dsputilenc.o
 YASM-OBJS-$(CONFIG_FFT)                += x86/fft.o
-YASM-OBJS-$(CONFIG_H263_DECODER)       += x86/h263_loopfilter.o
-YASM-OBJS-$(CONFIG_H263_ENCODER)       += x86/h263_loopfilter.o
+YASM-OBJS-$(CONFIG_H263DSP)            += x86/h263_loopfilter.o
 YASM-OBJS-$(CONFIG_H264CHROMA)         += x86/h264_chromamc.o           \
                                           x86/h264_chromamc_10bit.o
 YASM-OBJS-$(CONFIG_H264DSP)            += x86/h264_deblock.o            \

libavcodec/x86/dsputil_init.c

@@ -68,9 +68,6 @@ void ff_put_no_rnd_mpeg4_qpel8_v_lowpass_mmxext(uint8_t *dst, uint8_t *src,
 #define ff_put_no_rnd_pixels16_mmxext ff_put_pixels16_mmxext
 #define ff_put_no_rnd_pixels8_mmxext ff_put_pixels8_mmxext
 
-void ff_h263_v_loop_filter_mmx(uint8_t *src, int stride, int qscale);
-void ff_h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale);
-
 int32_t ff_scalarproduct_int16_mmxext(const int16_t *v1, const int16_t *v2,
                                       int order);
 int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2,
@@ -566,11 +563,6 @@ static av_cold void dsputil_init_mmx(DSPContext *c, AVCodecContext *avctx,
 #endif /* HAVE_MMX_INLINE */
 
 #if HAVE_MMX_EXTERNAL
-    if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
-        c->h263_v_loop_filter = ff_h263_v_loop_filter_mmx;
-        c->h263_h_loop_filter = ff_h263_h_loop_filter_mmx;
-    }
-
     c->vector_clip_int32 = ff_vector_clip_int32_mmx;
 #endif /* HAVE_MMX_EXTERNAL */
 }

libavcodec/x86/h263dsp_init.c

@@ -0,0 +1,39 @@
+/*
+ * Copyright (c) 2013 Diego Biurrun <diego@biurrun.de>
+ *
+ * This file is part of Libav.
+ *
+ * Libav 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.
+ *
+ * Libav 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 Libav; 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/x86/cpu.h"
+#include "libavcodec/h263dsp.h"
+
+void ff_h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale);
+void ff_h263_v_loop_filter_mmx(uint8_t *src, int stride, int qscale);
+
+av_cold void ff_h263dsp_init_x86(H263DSPContext *c)
+{
+    int cpu_flags = av_get_cpu_flags();
+
+    if (EXTERNAL_MMX(cpu_flags)) {
+        c->h263_h_loop_filter = ff_h263_h_loop_filter_mmx;
+        c->h263_v_loop_filter = ff_h263_v_loop_filter_mmx;
+    }
+}