/**************************************************************************
 *                                                                        *
 *               Copyright (C) 1994, Silicon Graphics, Inc.               *
 *                                                                        *
 *  These coded instructions, statements, and computer programs  contain  *
 *  unpublished  proprietary  information of Silicon Graphics, Inc., and  *
 *  are protected by Federal copyright law.  They  may  not be disclosed  *
 *  to  third  parties  or copied or duplicated in any form, in whole or  *
 *  in part, without the prior written consent of Silicon Graphics, Inc.  *
 *                                                                        *
 **************************************************************************/

/*
 *  Color Combiner Unit
 *
 *  This unit is basically a 'paint mixer', providing various ways of
 *  combining texture, shade, and various register colors.
 *
 *  8/16/94  RJM
 */
#include <stdio.h>

#include "cc.h"

#define POSEDGE         (save_clk && !save_clk_old)
#define NEGEDGE         (!save_clk && save_clk_old)


#ifndef NULL
#define NULL (void *)0
#endif

#ifndef MIN
#define MIN(x,y)		(((x) < (y)) ? (x) : (y))
#endif

#ifndef MAX
#define MAX(x,y)		(((x) > (y)) ? (x) : (y))
#endif

#ifndef ABS
#define ABS(x)			(((x) < 0) ? -(x) : (x))
#endif

#define CLAMP(low, val, high)	(MAX(MIN(val, high), low))

#define AKE_SIGN_EXTEND_9B(x)	((((x) & 0x180) == 0x180) ? \
				((x) | ~0x1ff) : (x))

#define SIGN_EXTEND_9B(x)	(((x) & 0x100) ? ((x) | (~0x1ff)) : (x))
#define SIGN_EXTEND_12B(x)	(((x) & 0x800) ? ((x) | (~0xfff)) : (x))
#define SIGN_EXTEND_17B(x)	(((x) & 0x10000) ? ((x) | (~0x1ffff)) : (x))


/***************************************************************************
 * four_to_one(), 4:1 mux
 ***************************************************************************
 */
static int
  four_to_one( int sel, int a, int b, int c, int d )
{
  switch(sel & 3)
  {
    case 0:
      return(a);
    case 1:
      return(b);
    case 2:
      return(c);
    case 3:
      return(d);
  }
}

/***************************************************************************
 * eight_to_one(), 8:1 mux
 ***************************************************************************
 */
static int
  eight_to_one( int sel, int i0, int i1, int i2, int i3, int i4, int i5, int i6, int i7 )
{
  int out0, out1;
  int sel0, sel1;
  sel0 = sel & 3;
  sel1 = (sel & 4) >> 2;

  out0 = four_to_one(sel0, i0, i1, i2, i3);
  out1 = four_to_one(sel0, i4, i5, i6, i7);
  return(sel1 ? out1 : out0);
}

/***************************************************************************
 * sixteen_to_one(), 16:1 mux
 ***************************************************************************
 */
static int
  sixteen_to_one( int sel, int i0, int i1, int i2, int i3, int i4, int i5, int i6, int i7,
                           int i8, int i9, int ia, int ib, int ic, int id, int ie, int ig )
{
  int out0, out1;
  int sel0, sel1;
  sel0 = sel & 0x7;
  sel1 = (sel & 8) >> 3;

  out0 = eight_to_one(sel0, i0, i1, i2, i3, i4, i5, i6, i7);
  out1 = eight_to_one(sel0, i8, i9, ia, ib, ic, id, ie, ig);
  return(sel1 ? out1 : out0);
}

/***************************************************************************
 * thirtytwo_to_one(), 32:1 mux
 ***************************************************************************
 */
static int
  thirtytwo_to_one( int sel, int i0, int i1, int i2, int i3, int i4, int i5, int i6, int i7,
                             int i8, int i9, int ia, int ib, int ic, int id, int ie, int ig,
                             int i10, int i11, int i12, int i13, int i14, int i15, int i16, int i17,
                             int i18, int i19, int i1a, int i1b, int i1c, int i1d, int i1e, int i1f)
{
  int out0, out1;
  int sel0, sel1;
  sel0 = sel & 0xf;
  sel1 = (sel & 0x10) >> 4;

  out0 = sixteen_to_one(sel0, i0, i1, i2, i3, i4, i5, i6, i7,
                            i8, i9, ia, ib, ic, id, ie, ig);
  out1 = sixteen_to_one(sel0, i10, i11, i12, i13, i14, i15, i16, i17,
                            i18, i19, i1a, i1b, i1c, i1d, i1e, i1f);
  return(sel1 ? out1 : out0);
}


/***************************************************************************
 *  Linear Interpolator for color combiner function
 ***************************************************************************/

static int
  do_rgb_lerp( int suba, int subb, int mul, int add)
{
  int tsuba, tsubb, tmula, tmulb, tadda, taddb;

  /* convert to sign extended 2's comp numbers */
  tsuba = AKE_SIGN_EXTEND_9B(suba);
  tsubb = AKE_SIGN_EXTEND_9B(subb);
  tmula = tsuba - tsubb;
  tmulb = SIGN_EXTEND_9B(mul);
  tadda = (tmula * tmulb);
  taddb = AKE_SIGN_EXTEND_9B(add);
  return(tadda + (taddb << 8) + 0x80); /* round, s8.8 */
}

/***************************************************************************
 *  Linear Interpolator for alpha combiner function
 ***************************************************************************/
static int
  do_alpha_lerp( int suba, int subb, int mul, int add)
{
  int tsuba, tsubb, tmula, tmulb, tadda, taddb;

  /* convert to sign extended 2's comp numbers */
  tsuba = AKE_SIGN_EXTEND_9B(suba);
  tsubb = AKE_SIGN_EXTEND_9B(subb);
  tmula = tsuba - tsubb;
  tmulb = SIGN_EXTEND_9B(mul);
  tadda = (tmula * tmulb);
  taddb = AKE_SIGN_EXTEND_9B(add);
  return ((tadda + (taddb << 8) + 0x80) >> 8); /* s,0.8 */
}

/***************************************************************************
 *  Do Akeley method clamping:
 *
 *  Carry one extra bit of precision through all computations,
 *  and never clamp until the last computation is done (e.g. never clamp
 *  intermediate values, especially those that continue to be interpolated).
 *  When done, clamp the result if the MSB (extra bit) is one.  In this case,
 *  clamp to all zeros if the next most significant bit is 1, and to all ones
 *  if the next most significant bit is zero.  (To undo the wrap.)
 *
 ***************************************************************************/
static int
  akeley_clamp9b(int val)
{
  if(val & 0x100)
  {
    if(val & 0x80) /* clamp to zero */
	return 0;
    else /* clamp to all ones */
	return 0xff;
  }
  else /* no clamp */
    return(val);
}

/***************************************************************************
 *  Color Combiner function
 ***************************************************************************/

cc(cc_t **pp0, cc_t **pp1)
{
  /* pointers to memory structure */
  cc_t *p0, *p1;
  int save_clk;
  int save_clk_old;

  /* input signals */
    int st_span;                /* start of span */
    int ncyc;                   /* number of cycles mode */

    int key_en;                 /* enable keying mode */
    int cvg_times_alpha;        /* modify pixel coverage with alpha */
    int alpha_cvg_select;       /* select source of pixel alpha */

    int cc_x_sel_0_r;           /* [3:0] cc mode */
    int cc_y_sel_0_r;           /* [3:0] */
    int cc_a_sel_0_r;           /* [4:0] */
    int cc_c_sel_0_r;           /* [2:0] */

    int cc_x_sel_0_a;           /* [2:0] cc mode */
    int cc_y_sel_0_a;           /* [2:0] */
    int cc_a_sel_0_a;           /* [2:0] */
    int cc_c_sel_0_a;           /* [2:0] */

    int cc_x_sel_1_r;           /* [3:0] cc mode */
    int cc_y_sel_1_r;           /* [3:0] */
    int cc_a_sel_1_r;           /* [4:0] */
    int cc_c_sel_1_r;           /* [2:0] */

    int cc_x_sel_1_a;           /* [2:0] cc mode */
    int cc_y_sel_1_a;           /* [2:0] */
    int cc_a_sel_1_a;           /* [2:0] */
    int cc_c_sel_1_a;           /* [2:0] */

    int st_r;                   /* [7:0], shade value */
    int st_g;                   /* [7:0], shade value */
    int st_b;                   /* [7:0], shade value */
    int st_a;                   /* [7:0], shade value */

    int tf_r;                   /* [8:0], texel value */
    int tf_g;                   /* [8:0], texel value */
    int tf_b;                   /* [8:0], texel value */
    int tf_a;                   /* [8:0], texel value */

    int tf_lod_frac;            /* [8:0], texel lod value */

    int prim_lod_frac;          /* [8:0] primitive LOD fraction */

    int prim_r;                 /* [7:0] primitive color */
    int prim_g;                 /* [7:0] primitive color */
    int prim_b;                 /* [7:0] primitive color */
    int prim_a;                 /* [7:0] primitive color */

    int env_r;                  /* [7:0] environment color */
    int env_g;                  /* [7:0] environment color */
    int env_b;                  /* [7:0] environment color */
    int env_a;                  /* [7:0] environment color */

    int center_r;               /* [7:0] color keying center */
    int center_g;               /* [7:0] color keying center */
    int center_b;               /* [7:0] color keying center */
    int scale_r;                /* [7:0] color keying scale */
    int scale_g;                /* [7:0] color keying scale */
    int scale_b;                /* [7:0] color keying scale */

    int width_r;                /* [11:0] color keying width */
    int width_g;                /* [11:0] color keying width */
    int width_b;                /* [11:0] color keying width */

    int k4_coeff;               /* [8:0] color conversion coefficient */
    int k5_coeff;               /* [8:0] color conversion coefficient */

    int noise;                  /* [8:0] color conversion coefficient */

    int cvg;                    /* [3:0], 1.3, pixel coverage */

  /* temporary signals */
  int sel;
  int key_mode;

  /* outputs of muxes */
  int   suba_r, suba_g, suba_b, suba_a;
  int   subb_r, subb_g, subb_b, subb_a;
  int   mul_r, mul_g, mul_b, mul_a;
  int   add_r, add_g, add_b, add_a;

  /* intermediate key values */
  int key_r, key_g, key_b, key;
  int clamp_a, key_or_a, pix_cvg;
  int a_minus_lf, lfrac, denom, cvg_x_a;
  int dither, dither_a, dither_st_a;

  /* 
   *  Get pointers, clocks
   */
  p0 = *pp0;
  p1 = *pp1;
  save_clk = p0->gclk;
  save_clk_old = p1->gclk_old;

  if(POSEDGE)
  {
    /* grab inputs from p0 */
    st_span = p0->st_span;
    ncyc = p0->ncyc;

    key_en = p0->key_en;
    cvg_times_alpha = p0->cvg_times_alpha;
    alpha_cvg_select = p0->alpha_cvg_select;

    cc_x_sel_0_r = p0->cc_x_sel_0_r;
    cc_y_sel_0_r = p0->cc_y_sel_0_r;
    cc_a_sel_0_r = p0->cc_a_sel_0_r;
    cc_c_sel_0_r = p0->cc_c_sel_0_r;

    cc_x_sel_0_a = p0->cc_x_sel_0_a;
    cc_y_sel_0_a = p0->cc_y_sel_0_a;
    cc_a_sel_0_a = p0->cc_a_sel_0_a;
    cc_c_sel_0_a = p0->cc_c_sel_0_a;

    cc_x_sel_1_r = p0->cc_x_sel_1_r;
    cc_y_sel_1_r = p0->cc_y_sel_1_r;
    cc_a_sel_1_r = p0->cc_a_sel_1_r;
    cc_c_sel_1_r = p0->cc_c_sel_1_r;

    cc_x_sel_1_a = p0->cc_x_sel_1_a;
    cc_y_sel_1_a = p0->cc_y_sel_1_a;
    cc_a_sel_1_a = p0->cc_a_sel_1_a;
    cc_c_sel_1_a = p0->cc_c_sel_1_a;

    st_r = p0->st_r;
    st_g = p0->st_g;
    st_b = p0->st_b;
    st_a = p0->st_a;


    tf_r = p0->tf_r;
    tf_g = p0->tf_g;
    tf_b = p0->tf_b;
    tf_a = p0->tf_a;


    tf_lod_frac = p0->tf_lod_frac;

    prim_lod_frac = p0->prim_lod_frac;

    prim_r = p0->prim_r;
    prim_g = p0->prim_g;
    prim_b = p0->prim_b;
    prim_a = p0->prim_a;

    env_r = p0->env_r;
    env_g = p0->env_g;
    env_b = p0->env_b;
    env_a = p0->env_a;

    center_r = p0->center_r;
    center_g = p0->center_g;
    center_b = p0->center_b;

    scale_r = p0->scale_r;
    scale_g = p0->scale_g;
    scale_b = p0->scale_b;

    width_r = p0->width_r;
    width_g = p0->width_g;
    width_b = p0->width_b;

    k4_coeff = p0->k4_coeff;
    k5_coeff = p0->k5_coeff;

    noise = p0->noise;

    cvg = p0->cvg;

  }

  if(POSEDGE)
  {
    /* transfer all next-clock register values to register outputs. */
    *pp0 = p1; /* swap */
    *pp1 = p0;
    p0 = *pp0; /* fix pointers */
    p1 = *pp1;
  }

  if(POSEDGE)
  {
    /* Update all next-clock register values */
 
    /* 
     *  generate cycle signal 
     */
    p0->cycle = !((p1->cycle || st_span) && ncyc);
    p0->cycle_d1 = p1->cycle;
    p0->cycle_d2 = p1->cycle_d1;

    /* 
     *  Select mux selects 
     */
    if(p1->cycle)
    {
      p0->cc_x_sel_r = cc_x_sel_1_r;
      p0->cc_y_sel_r = cc_y_sel_1_r;
      p0->cc_a_sel_r = cc_a_sel_1_r;
      p0->cc_c_sel_r = cc_c_sel_1_r;
      p0->cc_x_sel_a = cc_x_sel_1_a;
      p0->cc_y_sel_a = cc_y_sel_1_a;
      p0->cc_a_sel_a = cc_a_sel_1_a;
      p0->cc_c_sel_a = cc_c_sel_1_a;
    }
    else
    {
      p0->cc_x_sel_r = cc_x_sel_0_r;
      p0->cc_y_sel_r = cc_y_sel_0_r;
      p0->cc_a_sel_r = cc_a_sel_0_r;
      p0->cc_c_sel_r = cc_c_sel_0_r;
      p0->cc_x_sel_a = cc_x_sel_0_a;
      p0->cc_y_sel_a = cc_y_sel_0_a;
      p0->cc_a_sel_a = cc_a_sel_0_a;
      p0->cc_c_sel_a = cc_c_sel_0_a;
    }
  
    /* 
     *  delay shade alpha, texture and lod_frac 
     */
    p0->st_a_d1 = st_a;
    p0->tf_r_d1 = tf_r;
    p0->tf_g_d1 = tf_g;
    p0->tf_b_d1 = tf_b;
    p0->tf_a_d1 = tf_a;
    p0->tf_lod_frac_d1 = tf_lod_frac;
    
    /*
     *  two cycle delayed lod_frac values
     */

    p0->tf_lod_frac_d2 = p1->tf_lod_frac_d1;

    /*
     *  Get CC RGB mux outputs
     */

    /* subtract A input */
    suba_r = sixteen_to_one(p1->cc_x_sel_r,
		p1->cc_r >> 8,
		p1->tf_r_d1,
		tf_r,
		prim_r,
		st_r,
		env_r,
		ONE_POINT_OH,
		noise,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
    suba_g = sixteen_to_one(p1->cc_x_sel_r,
		p1->cc_g >> 8,
		p1->tf_g_d1,
		tf_g,
		prim_g,
		st_g,
		env_g,
		ONE_POINT_OH,
		noise,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
    suba_b = sixteen_to_one(p1->cc_x_sel_r,
		p1->cc_b >> 8,
		p1->tf_b_d1,
		tf_b,
		prim_b,
		st_b,
		env_b,
		ONE_POINT_OH,
		noise,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);

    /* subtract B input */
    subb_r = sixteen_to_one(p1->cc_y_sel_r,
		p1->cc_r >> 8,
		p1->tf_r_d1,
		tf_r,
		prim_r,
		st_r,
		env_r,
		center_r,
		k4_coeff,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
    subb_g = sixteen_to_one(p1->cc_y_sel_r,
		p1->cc_g >> 8,
		p1->tf_g_d1,
		tf_g,
		prim_g,
		st_g,
		env_g,
		center_g,
		k4_coeff,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
    subb_b = sixteen_to_one(p1->cc_y_sel_r,
		p1->cc_b >> 8,
		p1->tf_b_d1,
		tf_b,
		prim_b,
		st_b,
		env_b,
		center_b,
		k4_coeff,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
		
    /* multiplier input */
    mul_r = thirtytwo_to_one(p1->cc_a_sel_r, 
		p1->cc_r >> 8,
		p1->tf_r_d1,
		tf_r,
		prim_r,
		st_r,
		env_r,
		scale_r,
		p1->cc_a,
		p1->tf_a_d1,
		tf_a,
		prim_a,
		st_a,
		env_a,
		p1->tf_lod_frac_d1,
		prim_lod_frac,
		k5_coeff,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
    mul_g = thirtytwo_to_one(p1->cc_a_sel_r, 
		p1->cc_g >> 8,
		p1->tf_g_d1,
		tf_g,
		prim_g,
		st_g,
		env_g,
		scale_g,
		p1->cc_a,
		p1->tf_a_d1,
		tf_a,
		prim_a,
		st_a,
		env_a,
		p1->tf_lod_frac_d1,
		prim_lod_frac,
		k5_coeff,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
    mul_b = thirtytwo_to_one(p1->cc_a_sel_r, 
		p1->cc_b >> 8,
		p1->tf_b_d1,
		tf_b,
		prim_b,
		st_b,
		env_b,
		scale_b,
		p1->cc_a,
		p1->tf_a_d1,
		tf_a,
		prim_a,
		st_a,
		env_a,
		p1->tf_lod_frac_d1,
		prim_lod_frac,
		k5_coeff,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO,
		ZERO, ZERO, ZERO, ZERO);
		
    /* adder input */
    add_r = eight_to_one(p1->cc_c_sel_r,
		p1->cc_r >> 8,
		p1->tf_r_d1,
		tf_r,
		prim_r,
		st_r,
		env_r,
		ONE_POINT_OH,
		ZERO);
    add_g = eight_to_one(p1->cc_c_sel_r,
		p1->cc_g >> 8,
		p1->tf_g_d1,
		tf_g,
		prim_g,
		st_g,
		env_g,
		ONE_POINT_OH,
		ZERO);
    add_b = eight_to_one(p1->cc_c_sel_r,
		p1->cc_b >> 8,
		p1->tf_b_d1,
		tf_b,
		prim_b,
		st_b,
		env_b,
		ONE_POINT_OH,
		ZERO);

  
    /*
     *  Get CC Alpha mux outputs
     */

    suba_a = eight_to_one(p1->cc_x_sel_a,
		p1->cc_a,
		p1->tf_a_d1,
		tf_a,
		prim_a,
		st_a,
		env_a,
		ONE_POINT_OH,
		ZERO);

    subb_a = eight_to_one(p1->cc_y_sel_a,
		p1->cc_a,
		p1->tf_a_d1,
		tf_a,
		prim_a,
		st_a,
		env_a,
		ONE_POINT_OH,
		ZERO);

    mul_a = eight_to_one(p1->cc_a_sel_a,
		p1->tf_lod_frac_d1,
		p1->tf_a_d1,
		tf_a,
		prim_a,
		st_a,
		env_a,
		prim_lod_frac,
		ZERO);

    add_a = eight_to_one(p1->cc_c_sel_a,
		p1->cc_a,
		p1->tf_a_d1,
		tf_a,
		prim_a,
		st_a,
		env_a,
		ONE_POINT_OH,
		ZERO);

  
    /*
     *  Do Linear Interpolator
     */
    p0->cc_r = do_rgb_lerp( suba_r, subb_r, mul_r, add_r);
    p0->cc_g = do_rgb_lerp( suba_g, subb_g, mul_g, add_g);
    p0->cc_b = do_rgb_lerp( suba_b, subb_b, mul_b, add_b);
    p0->cc_a = do_alpha_lerp( suba_a, subb_a, mul_a, add_a);
  
    /*
     *  latch output of sub a mux
     */
    p0->suba_r = suba_r; /* bypass keying s,8.0 */
    p0->suba_g = suba_g; /* bypass keying s,8.0 */
    p0->suba_b = suba_b; /* bypass keying s,8.0 */
  	
    /*
     *  Add->-abs(key') + width for keying, select min key' as key, clamp.
     */

 
    key_r = (-ABS(SIGN_EXTEND_17B(p1->cc_r))) + (width_r<<4);
    key_g = (-ABS(SIGN_EXTEND_17B(p1->cc_g))) + (width_g<<4);
    key_b = (-ABS(SIGN_EXTEND_17B(p1->cc_b))) + (width_b<<4);
    key   = MIN(MIN(key_r, key_g), key_b);
    key   = CLAMP(0, key, 0xff);
 
    /*
     *  Do coverage * alpha, texture edge stuff
     *  ORDER DEPENDENT SECTION
     */
  	
    key_mode = (key_en && p1->cycle_d2);

    clamp_a = akeley_clamp9b(SIGN_EXTEND_9B(p1->cc_a));
    if (clamp_a == 0xFF) clamp_a = 0x100;

    if(p1->alpha_dither_sel & 2)
    {
      if(p1->alpha_dither_sel & 1)
        dither = 0;
      else
        dither = ((noise >> 6) & 7);
    }
    else
    {
      if(p1->alpha_dither_sel & 1)
        dither = (~p1->alpha_dither) & 0x7;
      else
        dither = p1->alpha_dither;
    }

    dither_a = clamp_a + dither;
    dither_st_a = p1->st_a_d1 + dither;

    key_or_a = key_mode ? key : (dither_a & 0x100) ? 0xff : dither_a;
  

    pix_cvg       = (clamp_a * cvg + 4) >> 3;
    pix_cvg	  = cvg_times_alpha ? pix_cvg : (cvg << 5);
    p0->pixel_cvg = pix_cvg >> 5; /* 1.3 fixed point */
    if (pix_cvg > 0xFF) pix_cvg = 0xFF;
    p0->pixel_a   = alpha_cvg_select ? pix_cvg : key_or_a;
    
    /*
     *  Do bypass and clamp logic
     */
  
    p0->pixel_r = key_mode ? 
  	SIGN_EXTEND_9B(p1->suba_r) : SIGN_EXTEND_17B(p1->cc_r) >> 8;
    p0->pixel_g = key_mode ? 
  	SIGN_EXTEND_9B(p1->suba_g) : SIGN_EXTEND_17B(p1->cc_g) >> 8;
    p0->pixel_b = key_mode ? 
  	SIGN_EXTEND_9B(p1->suba_b) : SIGN_EXTEND_17B(p1->cc_b) >> 8;
  
    p0->pixel_r = akeley_clamp9b(p0->pixel_r);
    p0->pixel_g = akeley_clamp9b(p0->pixel_g);
    p0->pixel_b = akeley_clamp9b(p0->pixel_b);

    p0->shade_a = (dither_st_a & 0x100) ? 0xff : dither_st_a;

  } /* posedge */

  /* save last clock state */
   p0->gclk_old = p1->gclk_old = save_clk;
}
  
  
  
  
/*
 *  cc_init():
 *
 */
cc_init(cc_t *p0, cc_t *p1)
{
  p0->gclk = p1->gclk = 0;
  p0->gclk_old = p1->gclk_old = 0;
}