dsputil.c

君正早期ucos系统(只有早期的才不没有打包成库),MPLAYER,文件系统,图片解码,浏览,电子书,录音,想学ucos,识货的人就下吧 russblock fmradio explore set

                    + (b&0x03030303UL)\
                    + 0x02020202UL;\
        uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
                   + ((b&0xFCFCFCFCUL)>>2);\
        uint32_t l1,h1;\
\
        pixels+=line_size;\
        for(i=0; i<h; i+=2){\
            uint32_t a= AV_RN32(pixels  );\
            uint32_t b= AV_RN32(pixels+1);\
            l1=  (a&0x03030303UL)\
               + (b&0x03030303UL);\
            h1= ((a&0xFCFCFCFCUL)>>2)\
              + ((b&0xFCFCFCFCUL)>>2);\
            OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
            pixels+=line_size;\
            block +=line_size;\
            a= AV_RN32(pixels  );\
            b= AV_RN32(pixels+1);\
            l0=  (a&0x03030303UL)\
               + (b&0x03030303UL)\
               + 0x02020202UL;\
            h0= ((a&0xFCFCFCFCUL)>>2)\
              + ((b&0xFCFCFCFCUL)>>2);\
            OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
            pixels+=line_size;\
            block +=line_size;\
        }\
}\
\
static inline void OPNAME ## _pixels8_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
    int j;\
    for(j=0; j<2; j++){\
        int i;\
        const uint32_t a= AV_RN32(pixels  );\
        const uint32_t b= AV_RN32(pixels+1);\
        uint32_t l0=  (a&0x03030303UL)\
                    + (b&0x03030303UL)\
                    + 0x02020202UL;\
        uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
                   + ((b&0xFCFCFCFCUL)>>2);\
        uint32_t l1,h1;\
\
        pixels+=line_size;\
        for(i=0; i<h; i+=2){\
            uint32_t a= AV_RN32(pixels  );\
            uint32_t b= AV_RN32(pixels+1);\
            l1=  (a&0x03030303UL)\
               + (b&0x03030303UL);\
            h1= ((a&0xFCFCFCFCUL)>>2)\
              + ((b&0xFCFCFCFCUL)>>2);\
            OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
            pixels+=line_size;\
            block +=line_size;\
            a= AV_RN32(pixels  );\
            b= AV_RN32(pixels+1);\
            l0=  (a&0x03030303UL)\
               + (b&0x03030303UL)\
               + 0x02020202UL;\
            h0= ((a&0xFCFCFCFCUL)>>2)\
              + ((b&0xFCFCFCFCUL)>>2);\
            OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
            pixels+=line_size;\
            block +=line_size;\
        }\
        pixels+=4-line_size*(h+1);\
        block +=4-line_size*h;\
    }\
}\
\
static inline void OPNAME ## _no_rnd_pixels8_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\
{\
    int j;\
    for(j=0; j<2; j++){\
        int i;\
        const uint32_t a= AV_RN32(pixels  );\
        const uint32_t b= AV_RN32(pixels+1);\
        uint32_t l0=  (a&0x03030303UL)\
                    + (b&0x03030303UL)\
                    + 0x01010101UL;\
        uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
                   + ((b&0xFCFCFCFCUL)>>2);\
        uint32_t l1,h1;\
\
        pixels+=line_size;\
        for(i=0; i<h; i+=2){\
            uint32_t a= AV_RN32(pixels  );\
            uint32_t b= AV_RN32(pixels+1);\
            l1=  (a&0x03030303UL)\
               + (b&0x03030303UL);\
            h1= ((a&0xFCFCFCFCUL)>>2)\
              + ((b&0xFCFCFCFCUL)>>2);\
            OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
            pixels+=line_size;\
            block +=line_size;\
            a= AV_RN32(pixels  );\
            b= AV_RN32(pixels+1);\
            l0=  (a&0x03030303UL)\
               + (b&0x03030303UL)\
               + 0x01010101UL;\
            h0= ((a&0xFCFCFCFCUL)>>2)\
              + ((b&0xFCFCFCFCUL)>>2);\
            OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
            pixels+=line_size;\
            block +=line_size;\
        }\
        pixels+=4-line_size*(h+1);\
        block +=4-line_size*h;\
    }\
}\
\
CALL_2X_PIXELS(OPNAME ## _pixels16_c  , OPNAME ## _pixels8_c  , 8)\
CALL_2X_PIXELS(OPNAME ## _pixels16_x2_c , OPNAME ## _pixels8_x2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _pixels16_y2_c , OPNAME ## _pixels8_y2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_c, OPNAME ## _pixels8_xy2_c, 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_c  , OPNAME ## _pixels8_c         , 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_c , OPNAME ## _no_rnd_pixels8_x2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels8_y2_c , 8)\
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels8_xy2_c, 8)\

#define op_avg(a, b) a = rnd_avg32(a, b)
#endif
#define op_put(a, b) a = b

PIXOP2(avg, op_avg)
PIXOP2(put, op_put)
#undef op_avg
#undef op_put

#define avg2(a,b) ((a+b+1)>>1)
#define avg4(a,b,c,d) ((a+b+c+d+2)>>2)


#ifdef JZ4740_MXU_OPT
#define put_pixels2_mxu put_pixels2_c
#define avg_pixels2_mxu avg_pixels2_c
#define put_pixels2_xy2_mxu put_pixels2_xy2_c
#define avg_pixels2_xy2_mxu avg_pixels2_xy2_c
#define put_pixels4_xy2_mxu put_pixels4_xy2_c
#define avg_pixels4_xy2_mxu avg_pixels4_xy2_c

static void put_pixels4_mxu(uint8_t *dst, const uint8_t *src, int stride, int h){
  uint32_t src_aln,src_rs;
  int i;
  src_aln = (uint32_t)src & 0xfffffffc;
  src_rs = 4 - ((uint32_t)src & 3);

  dst -= stride;
  for(i=0; i<h; i++) {
    S32LDD(xr1,src_aln,0);
    S32LDD(xr2,src_aln,4);
    src_aln += stride;

    S32ALN(xr1,xr2,xr1,src_rs);  //xr1 <- src[3:0]  
    S32SDIV(xr1,dst,stride,0);
  }
}

static void avg_pixels4_mxu(uint8_t *dst, const uint8_t *src, int stride, int h){

  uint32_t src_aln,src_rs;
  int i;
  src_aln = (uint32_t)src & 0xfffffffc;
  src_rs = 4 - ((uint32_t)src & 3);

  dst -= stride;
  src_aln -= stride;
  for(i=0; i<4; i++) {
    S32LDIV(xr1,src_aln,stride,0);
    S32LDD(xr2,src_aln,4);
    S32LDIV(xr3, dst, stride, 0);

    S32ALN(xr1,xr2,xr1,src_rs);  //xr1 <- src[3:0]
    Q8AVGR(xr1,xr3,xr1);
  
    S32STD(xr1,dst,0);
  }
}

static void put_pixels8_mxu(uint8_t *dst, const uint8_t *src, int stride, int h){
  uint32_t src_aln,src_rs;
  int i;
  src_aln = (uint32_t)src & 0xfffffffc;
  src_rs = 4 - ((uint32_t)src & 3);

  dst -= stride;
  src_aln -= stride;
  for(i=0; i<h; i++) {
    S32LDIV(xr1,src_aln,stride,0);
    S32LDD(xr2,src_aln,4);
    S32LDD(xr3,src_aln,8);

    S32ALN(xr1,xr2,xr1,src_rs);  //xr1 <- src[3:0]
    S32ALN(xr2,xr3,xr2,src_rs);  //xr2 <- src[7:4]
  
    S32SDIV(xr1,dst,stride,0);
    S32STD(xr2,dst,4);
  }
}
#define put_no_rnd_pixels8_mxu put_pixels8_mxu

static void avg_pixels8_mxu(uint8_t *dst, const uint8_t *src, int stride, int h){

  uint32_t src_aln,src_rs;
  int i;
  src_aln = (uint32_t)src & 0xfffffffc;
  src_rs = 4 - ((uint32_t)src & 3);

  dst -= stride;
  src_aln -= stride;
  for(i=0; i<8; i++) {
    S32LDIV(xr1,src_aln,stride,0);
    S32LDD(xr2,src_aln,4);
    S32LDD(xr3,src_aln,8);

    S32LDIV(xr4, dst, stride, 0);
    S32LDD(xr5,dst,4);

    S32ALN(xr1,xr2,xr1,src_rs);  //xr1 <- src[3:0]
    S32ALN(xr2,xr3,xr2,src_rs);  //xr2 <- src[7:4]

    Q8AVGR(xr1,xr4,xr1);
    Q8AVGR(xr2,xr5,xr2);
  
    S32STD(xr1,dst,0);
    S32STD(xr2,dst,4);
  }
}
#define avg_no_rnd_pixels8_mxu avg_pixels8_mxu

static void put_pixels16_mxu(uint8_t *dst, const uint8_t *src, int stride, int h){

  uint32_t src_aln,src_rs;
  int i;
  src_aln = (uint32_t)src & 0xfffffffc;
  src_rs = 4 - ((uint32_t)src & 3);

  dst -= stride;
  src_aln -= stride;
  for(i=0; i<h; i++) {
    S32LDIV(xr1,src_aln,stride,0);
    S32LDD(xr2,src_aln,4);
    S32LDD(xr3,src_aln,8);
    S32LDD(xr4,src_aln,12);
    S32LDD(xr5,src_aln,16);

    S32ALN(xr1,xr2,xr1,src_rs);  //xr1 <- src[3:0]
    S32ALN(xr2,xr3,xr2,src_rs);  //xr2 <- src[7:4]
    S32ALN(xr3,xr4,xr3,src_rs);  //xr3 <- src[11:8]
    S32ALN(xr4,xr5,xr4,src_rs);  //xr4 <- src[16:12]
  
    S32SDIV(xr1,dst,stride,0);
    S32STD(xr2,dst,4);
    S32STD(xr3,dst,8);
    S32STD(xr4,dst,12);
  }
}
#define put_no_rnd_pixels16_mxu put_pixels16_mxu

static void avg_pixels16_mxu(uint8_t *dst, const uint8_t *src, int stride, int h){
  uint32_t src_aln,src_rs;
  int i;
  src_aln = (uint32_t)src & 0xfffffffc;
  src_rs = 4 - ((uint32_t)src & 3);

  dst -= stride;
  src_aln -= stride;
  for(i=0; i<h; i++) {
    S32LDIV(xr1,src_aln,stride,0);
    S32LDD(xr2,src_aln,4);
    S32LDD(xr3,src_aln,8);
    S32LDD(xr4,src_aln,12);
    S32LDD(xr5,src_aln,16);

    S32LDIV(xr6, dst, stride, 0);
    S32LDD(xr7,dst,4);
    S32LDD(xr8,dst,8);
    S32LDD(xr9,dst,12);

    S32ALN(xr1,xr2,xr1,src_rs);  //xr1 <- src[3:0]
    S32ALN(xr2,xr3,xr2,src_rs);  //xr2 <- src[7:4]
    S32ALN(xr3,xr4,xr3,src_rs);  //xr3 <- src[11:8]
    S32ALN(xr4,xr5,xr4,src_rs);  //xr2 <- src[16:12]
  
    Q8AVGR(xr1,xr6,xr1);
    Q8AVGR(xr2,xr7,xr2);
    Q8AVGR(xr3,xr8,xr3);
    Q8AVGR(xr4,xr9,xr4);

    S32STD(xr1,dst,0);
    S32STD(xr2,dst,4);
    S32STD(xr3,dst,8);
    S32STD(xr4,dst,12);
  }
}
#define avg_no_rnd_pixels16_mxu avg_pixels16_mxu

static void put_pixels4_l2_mxu(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, 
                                                int src_stride1, int src_stride2, int h){
    int i;
    uint32_t src_aln1, src_rs1, src_aln2, src_rs2;

    src_aln1 = (uint32_t)src1 & 0xfffffffc;
    src_rs1 = 4 - ((uint32_t)src1 & 3);
    src_aln2 = (uint32_t)src2 & 0xfffffffc;
    src_rs2 = 4 - ((uint32_t)src2 & 3);
    
    dst -= dst_stride;
    src_aln1 -= src_stride1;
    src_aln2 -= src_stride2;

    for(i=0; i<h; i++){
      S32LDIV(xr1, src_aln1, src_stride1, 0);
      S32LDD(xr2, src_aln1, 4);
      S32LDIV(xr4, src_aln2, src_stride2, 0);
      S32LDD(xr5, src_aln2, 4);

      S32ALN(xr1, xr2, xr1, src_rs1);
      S32ALN(xr4, xr5, xr4, src_rs2);
      Q8AVGR(xr1, xr4, xr1);
      S32SDIV(xr1, dst, dst_stride, 0);
    }
}

static void avg_pixels4_l2_mxu(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, 
                                                int src_stride1, int src_stride2, int h){
    int i;
    uint32_t src_aln1, src_rs1, src_aln2, src_rs2;

    src_aln1 = (uint32_t)src1 & 0xfffffffc;
    src_rs1 = 4 - ((uint32_t)src1 & 3);
    src_aln2 = (uint32_t)src2 & 0xfffffffc;
    src_rs2 = 4 - ((uint32_t)src2 & 3);
    
    dst -= dst_stride;
    src_aln1 -= src_stride1;
    src_aln2 -= src_stride2;

    for(i=0; i<h; i++){
      S32LDIV(xr1, src_aln1, src_stride1, 0);
      S32LDD(xr2, src_aln1, 4);
      S32LDIV(xr4, src_aln2, src_stride2, 0);
      S32LDD(xr5, src_aln2, 4);
      S32LDIV(xr3, dst, dst_stride, 0);

      S32ALN(xr1, xr2, xr1, src_rs1);
      S32ALN(xr4, xr5, xr4, src_rs2);
      Q8AVGR(xr1, xr4, xr1);
      Q8AVGR(xr1, xr3, xr1);
      S32STD(xr1, dst, 0);
    }
}


static void put_pixels8_l2_mxu(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, 
				      int dst_stride, int src_stride1, int src_stride2, int h){
    int i;
    uint32_t src_aln1, src_rs1, src_aln2, src_rs2;

    src_aln1 = (uint32_t)src1 & 0xfffffffc;
    src_rs1 = 4 - ((uint32_t)src1 & 3);
    src_aln2 = (uint32_t)src2 & 0xfffffffc;
    src_rs2 = 4 - ((uint32_t)src2 & 3);
    
    dst -= dst_stride;
    src_aln1 -= src_stride1;
    src_aln2 -= src_stride2;

    for(i=0; i<h; i++) {    
      S32LDIV(xr1, src_aln1, src_stride1, 0);
      S32LDD(xr2, src_aln1, 4);
      S32LDD(xr3, src_aln1, 8);
      S32LDIV(xr4, src_aln2, src_stride2, 0);
      S32LDD(xr5, src_aln2, 4);
      S32LDD(xr6, src_aln2, 8);