basic type mechanism working

1 files changed, 49 insertions, 156 deletions

diff --git a/ffmpeg/libavcodec/dsputil.c b/ffmpeg/libavcodec/dsputil.c
index b56af0e..0e9e347 100644
--- a/ffmpeg/libavcodec/dsputil.c
+++ b/ffmpeg/libavcodec/dsputil.c
@@ -27,6 +27,7 @@
  * DSP utils
  */
 
+#include "libavutil/attributes.h"
 #include "libavutil/imgutils.h"
 #include "libavutil/internal.h"
 #include "avcodec.h"
@@ -68,9 +69,6 @@ const uint8_t ff_zigzag248_direct[64] = {
     53, 61, 54, 62, 39, 47, 55, 63,
 };
 
-/* not permutated inverse zigzag_direct + 1 for MMX quantizer */
-DECLARE_ALIGNED(16, uint16_t, ff_inv_zigzag_direct16)[64];
-
 const uint8_t ff_alternate_horizontal_scan[64] = {
     0,  1,   2,  3,  8,  9, 16, 17,
     10, 11,  4,  5,  6,  7, 15, 14,
@@ -107,7 +105,9 @@ static const uint8_t simple_mmx_permutation[64]={
 
 static const uint8_t idct_sse2_row_perm[8] = {0, 4, 1, 5, 2, 6, 3, 7};
 
-void ff_init_scantable(uint8_t *permutation, ScanTable *st, const uint8_t *src_scantable){
+av_cold void ff_init_scantable(uint8_t *permutation, ScanTable *st,
+                               const uint8_t *src_scantable)
+{
     int i;
     int end;
 
@@ -128,8 +128,8 @@ void ff_init_scantable(uint8_t *permutation, ScanTable *st, const uint8_t *src_s
     }
 }
 
-void ff_init_scantable_permutation(uint8_t *idct_permutation,
-                                   int idct_permutation_type)
+av_cold void ff_init_scantable_permutation(uint8_t *idct_permutation,
+                                           int idct_permutation_type)
 {
     int i;
 
@@ -797,7 +797,7 @@ static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int
 
 #define QPEL_MC(r, OPNAME, RND, OP) \
 static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<h; i++)\
     {\
@@ -816,7 +816,7 @@ static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstS
 \
 static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     const int w=8;\
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<w; i++)\
     {\
@@ -843,7 +843,7 @@ static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstS
 }\
 \
 static void OPNAME ## mpeg4_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     \
     for(i=0; i<h; i++)\
@@ -870,7 +870,7 @@ static void OPNAME ## mpeg4_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dst
 }\
 \
 static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     const int w=16;\
     for(i=0; i<w; i++)\
@@ -1329,19 +1329,19 @@ QPEL_MC(0, avg_       , _       , op_avg)
 
 void ff_put_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)
 {
-  put_pixels8_8_c(dst, src, stride, 8);
+    put_pixels8_8_c(dst, src, stride, 8);
 }
 void ff_avg_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)
 {
-  avg_pixels8_8_c(dst, src, stride, 8);
+    avg_pixels8_8_c(dst, src, stride, 8);
 }
 void ff_put_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)
 {
-  put_pixels16_8_c(dst, src, stride, 16);
+    put_pixels16_8_c(dst, src, stride, 16);
 }
 void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)
 {
-  avg_pixels16_8_c(dst, src, stride, 16);
+    avg_pixels16_8_c(dst, src, stride, 16);
 }
 
 #define put_qpel8_mc00_c  ff_put_pixels8x8_c
@@ -1352,7 +1352,7 @@ void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)
 #define put_no_rnd_qpel16_mc00_c ff_put_pixels16x16_c
 
 static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
     int i;
 
     for(i=0; i<h; i++){
@@ -1434,7 +1434,7 @@ DIRAC_MC(avg)
 #endif
 
 static void wmv2_mspel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int w){
-    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
     int i;
 
     for(i=0; i<w; i++){
@@ -1513,107 +1513,6 @@ 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 void h261_loop_filter_c(uint8_t *src, int stride){
-    int x,y,xy,yz;
-    int temp[64];
-
-    for(x=0; x<8; x++){
-        temp[x      ] = 4*src[x           ];
-        temp[x + 7*8] = 4*src[x + 7*stride];
-    }
-    for(y=1; y<7; y++){
-        for(x=0; x<8; x++){
-            xy = y * stride + x;
-            yz = y * 8 + x;
-            temp[yz] = src[xy - stride] + 2*src[xy] + src[xy + stride];
-        }
-    }
-
-    for(y=0; y<8; y++){
-        src[  y*stride] = (temp[  y*8] + 2)>>2;
-        src[7+y*stride] = (temp[7+y*8] + 2)>>2;
-        for(x=1; x<7; x++){
-            xy = y * stride + x;
-            yz = y * 8 + x;
-            src[xy] = (temp[yz-1] + 2*temp[yz] + temp[yz+1] + 8)>>4;
-        }
-    }
-}
-
 static inline int pix_abs16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int s, i;
@@ -1955,7 +1854,7 @@ void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){
 
 static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){
     long i;
-    for(i=0; i<=w-sizeof(long); i+=sizeof(long)){
+    for (i = 0; i <= w - (int)sizeof(long); i += sizeof(long)) {
         long a = *(long*)(src+i);
         long b = *(long*)(dst+i);
         *(long*)(dst+i) = ((a&pb_7f) + (b&pb_7f)) ^ ((a^b)&pb_80);
@@ -1980,7 +1879,7 @@ static void diff_bytes_c(uint8_t *dst, const uint8_t *src1, const uint8_t *src2,
         }
     }else
 #endif
-    for(i=0; i<=w-sizeof(long); i+=sizeof(long)){
+    for (i = 0; i <= w - (int)sizeof(long); i += sizeof(long)) {
         long a = *(long*)(src1+i);
         long b = *(long*)(src2+i);
         *(long*)(dst+i) = ((a|pb_80) - (b&pb_7f)) ^ ((a^b^pb_80)&pb_80);
@@ -2596,19 +2495,6 @@ static int32_t scalarproduct_and_madd_int16_c(int16_t *v1, const int16_t *v2, co
     return res;
 }
 
-static void apply_window_int16_c(int16_t *output, const int16_t *input,
-                                 const int16_t *window, unsigned int len)
-{
-    int i;
-    int len2 = len >> 1;
-
-    for (i = 0; i < len2; i++) {
-        int16_t w       = window[i];
-        output[i]       = (MUL16(input[i],       w) + (1 << 14)) >> 15;
-        output[len-i-1] = (MUL16(input[len-i-1], w) + (1 << 14)) >> 15;
-    }
-}
-
 static void vector_clip_int32_c(int32_t *dst, const int32_t *src, int32_t min,
                                 int32_t max, unsigned int len)
 {
@@ -2625,12 +2511,12 @@ static void vector_clip_int32_c(int32_t *dst, const int32_t *src, int32_t min,
     } while (len > 0);
 }
 
-static void ff_jref_idct_put(uint8_t *dest, int line_size, int16_t *block)
+static void jref_idct_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct (block);
     put_pixels_clamped_c(block, dest, line_size);
 }
-static void ff_jref_idct_add(uint8_t *dest, int line_size, int16_t *block)
+static void jref_idct_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct (block);
     add_pixels_clamped_c(block, dest, line_size);
@@ -2675,8 +2561,6 @@ av_cold void ff_dsputil_static_init(void)
     for(i=0;i<512;i++) {
         ff_squareTbl[i] = (i - 256) * (i - 256);
     }
-
-    for(i=0; i<64; i++) ff_inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1;
 }
 
 int ff_check_alignment(void){
@@ -2744,10 +2628,15 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
             c->idct_add              = ff_simple_idct_add_10;
             c->idct                  = ff_simple_idct_10;
             c->idct_permutation_type = FF_NO_IDCT_PERM;
+        } else if (avctx->bits_per_raw_sample == 12) {
+            c->idct_put              = ff_simple_idct_put_12;
+            c->idct_add              = ff_simple_idct_add_12;
+            c->idct                  = ff_simple_idct_12;
+            c->idct_permutation_type = FF_NO_IDCT_PERM;
         } else {
         if(avctx->idct_algo==FF_IDCT_INT){
-            c->idct_put= ff_jref_idct_put;
-            c->idct_add= ff_jref_idct_add;
+            c->idct_put= jref_idct_put;
+            c->idct_add= jref_idct_add;
             c->idct    = ff_j_rev_dct;
             c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
         }else if(avctx->idct_algo==FF_IDCT_FAAN){
@@ -2891,20 +2780,12 @@ 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->h261_loop_filter= h261_loop_filter_c;
-
     c->try_8x8basis= try_8x8basis_c;
     c->add_8x8basis= add_8x8basis_c;
 
     c->vector_clipf = vector_clipf_c;
     c->scalarproduct_int16 = scalarproduct_int16_c;
     c->scalarproduct_and_madd_int16 = scalarproduct_and_madd_int16_c;
-    c->apply_window_int16 = apply_window_int16_c;
     c->vector_clip_int32 = vector_clip_int32_c;
 
     c->shrink[0]= av_image_copy_plane;
@@ -2919,13 +2800,13 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
 #define FUNC(f, depth) f ## _ ## depth
 #define FUNCC(f, depth) f ## _ ## depth ## _c
 
-#define BIT_DEPTH_FUNCS(depth) \
-    c->get_pixels                    = FUNCC(get_pixels,   depth);
-
     c->draw_edges                    = FUNCC(draw_edges, 8);
     c->clear_block                   = FUNCC(clear_block, 8);
     c->clear_blocks                  = FUNCC(clear_blocks, 8);
 
+#define BIT_DEPTH_FUNCS(depth) \
+    c->get_pixels                    = FUNCC(get_pixels,   depth);
+
     switch (avctx->bits_per_raw_sample) {
     case 9:
     case 10:
@@ -2941,13 +2822,20 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
     }
 
 
-    if (HAVE_MMX)        ff_dsputil_init_mmx   (c, avctx);
-    if (ARCH_ARM)        ff_dsputil_init_arm   (c, avctx);
-    if (HAVE_VIS)        ff_dsputil_init_vis   (c, avctx);
-    if (ARCH_ALPHA)      ff_dsputil_init_alpha (c, avctx);
-    if (ARCH_PPC)        ff_dsputil_init_ppc   (c, avctx);
-    if (ARCH_SH4)        ff_dsputil_init_sh4   (c, avctx);
-    if (ARCH_BFIN)       ff_dsputil_init_bfin  (c, avctx);
+    if (ARCH_ALPHA)
+        ff_dsputil_init_alpha(c, avctx);
+    if (ARCH_ARM)
+        ff_dsputil_init_arm(c, avctx);
+    if (ARCH_BFIN)
+        ff_dsputil_init_bfin(c, avctx);
+    if (ARCH_PPC)
+        ff_dsputil_init_ppc(c, avctx);
+    if (ARCH_SH4)
+        ff_dsputil_init_sh4(c, avctx);
+    if (HAVE_VIS)
+        ff_dsputil_init_vis(c, avctx);
+    if (ARCH_X86)
+        ff_dsputil_init_x86(c, avctx);
 
     ff_init_scantable_permutation(c->idct_permutation,
                                   c->idct_permutation_type);
@@ -2957,3 +2845,8 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx)
 {
     ff_dsputil_init(c, avctx);
 }
+
+av_cold void avpriv_dsputil_init(DSPContext *c, AVCodecContext *avctx)
+{
+    ff_dsputil_init(c, avctx);
+}