Line Draw into SAM3x8e DMA buffer (Arduino DUE)

line.cpp

// Assume there is a currX and currY global state which is the current x,y position
// Assume that currBuffer points to a valid properly sized buffer in RAM for DMA
//
// Buffer size can be calculated as n * 4 bytes where n is max(xdelta, ydelta) 
// where xdelta and ydelta are the extents of the line to be drawn
//
// xd and yd is the destination point
//
// The output DACs are 12 bit but for speed this outputs in steps of 4 to cut the point generation by 4 times
// Effective resolution is 1024 points in X and Y
//
// The routine uses a SWAR (SIMD within a register) technique of operating on two 16 bit values within a 32 bit register
//
// So in summary the words are 16 bit and only the low 12 bits are used for the DAC. And then I'm only stepping in 10 bit resolution

// The extra OR with 0x10000000 is a hardware flag to use the other DAC
// y is in the low word and x is in the high word
#define DAC_PACK_COORD(_x, _y) (((0x0FFF & (_y)) << 2) | ((0x0FFF & (_x)) << 18) | 0x10000000)  

void line(uint16_t xd, uint16_t yd) {

  uint16_t x1 = currX;
  uint16_t y1 = currY;

  uint16_t x2 = xd;
  uint16_t y2 = yd;

  // updated the new currX and currY to be the end position
  currX = x2;
  currY = y2;

  uint16_t dx;  // abs(deltax)
  uint16_t dy;  // abs(deltay)
  uint16_t hx;  // counter used when moving predominatnly in the x direction
  uint16_t hy;  // counter used when moving predominantly in the y direction
  uint16_t ddx; // accumulator to decide when to move in the 
  uint16_t ddy; // accumulator


  // packs both x and y into a 32 bit word suitable for DMA to DAC on Arduino DUE
  uint32_t coord = DAC_PACK_COORD(x1,y1);

  if (x1 > x2) {
    dx = x1-x2;
    if (y1 > y2) {
      dy = y1-y2;
      if (dy > dx) {
        hy = dy;
        ddy = dy;
        while(hy)
        {
          ddy = ddy - dx;
          if (ddy < 0) {
            coord = coord - (1 << 18);  // Occasionally stepping by 1 in X   (16 to shift to upper word and 2 more for increments of 4)
            ddy = ddy + dy;
          } 
          currBuffer[bufferCounter++] = coord;
          coord = coord - (1<<2);       // Always stepping by 1 in Y  (increments of 4 because of 10 bit stepping)
          hy = hy  - 1;
        }
      } else {
        hx = dx;
        ddx = dx;
        while(hx)
        {
          ddx = ddx - dy;   
          if (ddx < 0) {
            coord = coord - (1<<2);
            ddx = ddx + dx;
          }
          currBuffer[bufferCounter++] = coord;
          coord = coord - (1 << 18);
          hx = hx - 1;
        }     
      }
    } else {
      dy = y2-y1;
      if (dy > dx) {
        hy = dy;
        ddy = dy;        
        while(hy)
        {
          ddy = ddy - dx;
          if (ddy < 0) {
            coord = coord - (1<<18);
            ddy = ddy + dy;
          } 
          currBuffer[bufferCounter++] = coord;
          coord = coord + (1<<2);
          hy = hy  - 1;
        }
      } else {
        hx = dx;
        ddx = dx;        
        while(hx)
        {
          ddx = ddx - dy;   
          if (ddx < 0) {
            coord = coord + (1<<2);
            ddx = ddx + dx;
          }
          currBuffer[bufferCounter++] = coord;
          coord = coord - (1<<18);
          hx = hx - 1;
        }     
      }
    }
  } else { 
    dx = x2-x1;
    if (y1 > y2) {
      dy = y1-y2;
      if (dy > dx) {
        hy = dy;
        ddy = dy;
        while(hy)
        {
          ddy = ddy - dx;
          if (ddy < 0) {
            coord = coord + (1<<18);
            ddy = ddy + dy;
          } 
          currBuffer[bufferCounter++] = coord;
          coord = coord - (1<<2);
          hy = hy  - 1;
        }
      } else {
        hx = dx;
        ddx = dx;
        while(hx)
        {
          ddx = ddx - dy;   
          if (ddx < 0) {
            coord = coord - (1<<2);
            ddx = ddx + dx;
          }
          currBuffer[bufferCounter++] = coord;
          coord = coord + (1<<18);
          hx = hx - 1;
        }     
      }
    } else {
      dy = y2-y1;
      if (dy > dx) {
        hy = dy;
        ddy = dy;
        while(hy)
        {
          ddy = ddy - dx;
          if (ddy < 0) {
            coord = coord + (1<<18);
            ddy = ddy + dy;
          } 
          currBuffer[bufferCounter++] = coord;
          coord = coord + (1<<2);
          hy = hy  - 1;
        }
      } else {
        hx = dx;
        ddx = dx;
        while(hx)
        {
          ddx = ddx - dy;   
          if (ddx < 0) {
            coord = coord + (1<<2);
            ddx = ddx + dx;
          }
          currBuffer[bufferCounter++] = coord;
          coord = coord + (1<<18);
          hx = hx - 1;
        }     
      }
    }
  }
};