/**

  * $Id: rendcore.c,v 1.16 2007-01-31 10:32:25 nkeynes Exp $

  *

  * PVR2 renderer core.

  *

  * Copyright (c) 2005 Nathan Keynes.

  *

  * This program is free software; you can redistribute it and/or modify

  * it under the terms of the GNU General Public License as published by

  * the Free Software Foundation; either version 2 of the License, or

  * (at your option) any later version.

  *

  * This program 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 General Public License for more details.

  */

 #include <sys/time.h>

 #include "pvr2/pvr2.h"

 #include "asic.h"

 int pvr2_poly_depthmode[8] = { GL_NEVER, GL_LESS, GL_EQUAL, GL_LEQUAL,

 				      GL_GREATER, GL_NOTEQUAL, GL_GEQUAL, 

 				      GL_ALWAYS };

 int pvr2_poly_srcblend[8] = { 

     GL_ZERO, GL_ONE, GL_DST_COLOR, GL_ONE_MINUS_DST_COLOR,

     GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_DST_ALPHA, 

     GL_ONE_MINUS_DST_ALPHA };

 int pvr2_poly_dstblend[8] = {

     GL_ZERO, GL_ONE, GL_SRC_COLOR, GL_ONE_MINUS_SRC_COLOR,

     GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_DST_ALPHA,

     GL_ONE_MINUS_DST_ALPHA };

 int pvr2_poly_texblend[4] = {

     GL_REPLACE, 

     GL_MODULATE,  

     GL_DECAL, 

     GL_MODULATE 

 };

 int pvr2_render_colour_format[8] = {

     COLFMT_ARGB1555, COLFMT_RGB565, COLFMT_ARGB4444, COLFMT_ARGB1555,

     COLFMT_RGB888, COLFMT_ARGB8888, COLFMT_ARGB8888, COLFMT_ARGB4444 };

 #define CULL_NONE 0

 #define CULL_SMALL 1

 #define CULL_CCW 2

 #define CULL_CW 3

 #define SEGMENT_END         0x80000000

 #define SEGMENT_ZCLEAR      0x40000000

 #define SEGMENT_SORT_TRANS  0x20000000

 #define SEGMENT_START       0x10000000

 #define SEGMENT_X(c)        (((c) >> 2) & 0x3F)

 #define SEGMENT_Y(c)        (((c) >> 8) & 0x3F)

 #define NO_POINTER          0x80000000

 extern char *video_base;

 gboolean pvr2_force_fragment_alpha = FALSE;

 gboolean pvr2_debug_render = FALSE;

 struct tile_segment {

     uint32_t control;

     pvraddr_t opaque_ptr;

     pvraddr_t opaquemod_ptr;

     pvraddr_t trans_ptr;

     pvraddr_t transmod_ptr;

     pvraddr_t punchout_ptr;

 };

 void render_print_tilelist( FILE *f, uint32_t tile_entry );

 /**

  * Convert a half-float (16-bit) FP number to a regular 32-bit float.

  * Source is 1-bit sign, 5-bit exponent, 10-bit mantissa.

  * TODO: Check the correctness of this.

  */

 float halftofloat( uint16_t half )

 {

     union {

         float f;

         uint32_t i;

     } temp;

     /* int e = ((half & 0x7C00) >> 10) - 15 + 127;

     temp.i = ((half & 0x8000) << 16) | (e << 23) |

     ((half & 0x03FF) << 13); */

     temp.i = ((uint32_t)half)<<16;

     return temp.f;

 }

 /**

  * Setup the GL context for the supplied polygon context.

  * @param context pointer to 3 or 5 words of polygon context

  * @param modified boolean flag indicating that the modified

  *  version should be used, rather than the normal version.

  */

 void render_set_context( uint32_t *context, int render_mode )

 {

     uint32_t poly1 = context[0], poly2, texture;

     if( render_mode == RENDER_FULLMOD ) {

 	poly2 = context[3];

 	texture = context[4];

     } else {

 	poly2 = context[1];

 	texture = context[2];

     }

     if( POLY1_DEPTH_ENABLE(poly1) ) {

 	glEnable( GL_DEPTH_TEST );

 	glDepthFunc( POLY1_DEPTH_MODE(poly1) );

     } else {

 	glDisable( GL_DEPTH_TEST );

     }

     switch( POLY1_CULL_MODE(poly1) ) {

     case CULL_NONE:

     case CULL_SMALL:

 	glDisable( GL_CULL_FACE );

 	break;

     case CULL_CCW:

 	glEnable( GL_CULL_FACE );

 	glFrontFace( GL_CW );

 	break;

     case CULL_CW:

 	glEnable( GL_CULL_FACE );

 	glFrontFace( GL_CCW );

 	break;

     }

     if( POLY1_SPECULAR(poly1) ) {

 	glEnable(GL_COLOR_SUM);

     } else {

 	glDisable(GL_COLOR_SUM);

     }

     pvr2_force_fragment_alpha = POLY2_ALPHA_ENABLE(poly2) ? FALSE : TRUE;

     if( POLY1_TEXTURED(poly1) ) {

 	int width = POLY2_TEX_WIDTH(poly2);

 	int height = POLY2_TEX_HEIGHT(poly2);

 	glEnable(GL_TEXTURE_2D);

 	texcache_get_texture( (texture&0x000FFFFF)<<3, width, height, texture );

 	switch( POLY2_TEX_BLEND(poly2) ) {

 	case 0: /* Replace */

 	    glTexEnvi( GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE );

 	    break;

 	case 2:/* Decal */

 	    glTexEnvi( GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_DECAL );

 	    break;

 	case 1: /* Modulate RGB */

 	    /* This is not directly supported by opengl (other than by mucking

 	     * with the texture format), but we get the same effect by forcing

 	     * the fragment alpha to 1.0 and using GL_MODULATE.

 	     */

 	    pvr2_force_fragment_alpha = TRUE;

 	case 3: /* Modulate RGBA */

 	    glTexEnvi( GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_MODULATE );

 	    break;

 	}

 	glTexEnvi( GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, POLY2_TEX_BLEND(poly2) );

 	if( POLY2_TEX_CLAMP_U(poly2) ) {

 	    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP );

 	} else {

 	    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT );

 	}	    

 	if( POLY2_TEX_CLAMP_V(poly2) ) {

 	    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP );

 	} else {

 	    glTexParameteri( GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT );

 	}

     } else {

 	glDisable( GL_TEXTURE_2D );

     }

     glShadeModel( POLY1_SHADE_MODEL(poly1) );

     int srcblend = POLY2_SRC_BLEND(poly2);

     int destblend = POLY2_DEST_BLEND(poly2);

     glBlendFunc( srcblend, destblend );

     if( POLY2_SRC_BLEND_TARGET(poly2) || POLY2_DEST_BLEND_TARGET(poly2) ) {

 	ERROR( "Accumulation buffer not supported" );

     }

 }

 #define FARGB_A(x) (((float)(((x)>>24)+1))/256.0)

 #define FARGB_R(x) (((float)((((x)>>16)&0xFF)+1))/256.0)

 #define FARGB_G(x) (((float)((((x)>>8)&0xFF)+1))/256.0)

 #define FARGB_B(x) (((float)(((x)&0xFF)+1))/256.0)

 void render_unpack_vertexes( struct vertex_unpacked *out, uint32_t poly1, 

 			     uint32_t *vertexes, int num_vertexes,

 			     int vertex_size, int render_mode )

 {

     int m = 0, i;

     if( render_mode == RENDER_FULLMOD ) {

 	m = (vertex_size - 3)/2;

     }

     for( i=0; i<num_vertexes; i++ ) {

 	float *vertexf = (float *)vertexes;

 	int k = m + 3;

 	out[i].x = vertexf[0];

 	out[i].y = vertexf[1];

 	out[i].z = vertexf[2];

     	if( POLY1_TEXTURED(poly1) ) {

 	    if( POLY1_UV16(poly1) ) {

 		out[i].u = halftofloat(vertexes[k]>>16);

 		out[i].v = halftofloat(vertexes[k]);

 		k++;

 	    } else {

 		out[i].u = vertexf[k];

 		out[i].v = vertexf[k+1];

 		k+=2;

 	    }

 	} else {

 	    out[i].u = 0;

 	    out[i].v = 0;

 	}

 	uint32_t argb = vertexes[k++];

 	out[i].rgba[0] = FARGB_R(argb);

 	out[i].rgba[1] = FARGB_G(argb);

         out[i].rgba[2] = FARGB_B(argb);

 	out[i].rgba[3] = FARGB_A(argb);

 	if( POLY1_SPECULAR(poly1) ) {

 	    uint32_t offset = vertexes[k++];

 	    out[i].offset_rgba[0] = FARGB_R(argb);

 	    out[i].offset_rgba[1] = FARGB_G(argb);

 	    out[i].offset_rgba[2] = FARGB_B(argb);

 	    out[i].offset_rgba[3] = FARGB_A(argb);

 	}

 	vertexes += vertex_size;

     }

 }

 /**

  * Unpack the vertexes for a quad, calculating the values for the last

  * vertex.

  * FIXME: Integrate this with rendbkg somehow

  */

 void render_unpack_quad( struct vertex_unpacked *unpacked, uint32_t poly1, 

 			 uint32_t *vertexes, int vertex_size,

 			 int render_mode )

 {

     int i;

     struct vertex_unpacked diff0, diff1;

     render_unpack_vertexes( unpacked, poly1, vertexes, 3, vertex_size, render_mode );

     diff0.x = unpacked[0].x - unpacked[1].x;

     diff0.y = unpacked[0].y - unpacked[1].y;

     diff1.x = unpacked[2].x - unpacked[1].x;

     diff1.y = unpacked[2].y - unpacked[1].y;

     float detxy = ((diff1.y) * (diff0.x)) - ((diff0.y) * (diff1.x));

     float *vertexf = (float *)(vertexes+(vertex_size*3));

     if( detxy == 0 ) {

 	memcpy( &unpacked[3], &unpacked[2], sizeof(struct vertex_unpacked) );

 	unpacked[3].x = vertexf[0];

 	unpacked[3].y = vertexf[1];

 	return;

     }	

     unpacked[3].x = vertexf[0];

     unpacked[3].y = vertexf[1];

     float t = ((unpacked[3].x - unpacked[1].x) * diff1.y -

 	       (unpacked[3].y - unpacked[1].y) * diff1.x) / detxy;

     float s = ((unpacked[3].y - unpacked[1].y) * diff0.x -

 	       (unpacked[3].x - unpacked[1].x) * diff0.y) / detxy;

     diff0.z = (1/unpacked[0].z) - (1/unpacked[1].z);

     diff1.z = (1/unpacked[2].z) - (1/unpacked[1].z);

     unpacked[3].z = 1/((1/unpacked[1].z) + (t*diff0.z) + (s*diff1.z));

     diff0.u = unpacked[0].u - unpacked[1].u;

     diff0.v = unpacked[0].v - unpacked[1].v;

     diff1.u = unpacked[2].u - unpacked[1].u;

     diff1.v = unpacked[2].v - unpacked[1].v;

     unpacked[3].u = unpacked[1].u + (t*diff0.u) + (s*diff1.u);

     unpacked[3].v = unpacked[1].v + (t*diff0.v) + (s*diff1.v);

     if( !POLY1_GOURAUD_SHADED(poly1) ) {

 	memcpy( unpacked[3].rgba, unpacked[2].rgba, sizeof(unpacked[2].rgba) );

 	memcpy( unpacked[3].offset_rgba, unpacked[2].offset_rgba, sizeof(unpacked[2].offset_rgba) );

     } else {

 	for( i=0; i<4; i++ ) {

 	    float d0 = unpacked[0].rgba[i] - unpacked[1].rgba[i];

 	    float d1 = unpacked[2].rgba[i] - unpacked[1].rgba[i];

 	    unpacked[3].rgba[i] = unpacked[1].rgba[i] + (t*d0) + (s*d1);

 	    d0 = unpacked[0].offset_rgba[i] - unpacked[1].offset_rgba[i];

 	    d1 = unpacked[2].offset_rgba[i] - unpacked[1].offset_rgba[i];

 	    unpacked[3].offset_rgba[i] = unpacked[1].offset_rgba[i] + (t*d0) + (s*d1);

 	}

     }    

 }

 void render_unpacked_vertex_array( uint32_t poly1, struct vertex_unpacked *vertexes[], 

 				   int num_vertexes ) {

     int i;

     glBegin( GL_TRIANGLE_STRIP );

     for( i=0; i<num_vertexes; i++ ) {

 	if( POLY1_TEXTURED(poly1) ) {

 	    glTexCoord2f( vertexes[i]->u, vertexes[i]->v );

 	}

 	if( pvr2_force_fragment_alpha ) {

 	    glColor4f( vertexes[i]->rgba[0], vertexes[i]->rgba[1], vertexes[i]->rgba[2], 1.0 );

 	} else {

 	    glColor4f( vertexes[i]->rgba[0], vertexes[i]->rgba[1], vertexes[i]->rgba[2],

 		       vertexes[i]->rgba[3] );

 	}

 	if( POLY1_SPECULAR(poly1) ) {

 	    glSecondaryColor3fEXT( vertexes[i]->offset_rgba[0],

 				   vertexes[i]->offset_rgba[1],

 				   vertexes[i]->offset_rgba[2] );

 	}

 	glVertex3f( vertexes[i]->x, vertexes[i]->y, 1/vertexes[i]->z );

     }

     glEnd();

 }

 void render_quad_vertexes( uint32_t poly1, uint32_t *vertexes, int vertex_size, int render_mode )

 {

     struct vertex_unpacked unpacked[4];

     struct vertex_unpacked *pt[4] = {&unpacked[0], &unpacked[1], &unpacked[3], &unpacked[2]};

     render_unpack_quad( unpacked, poly1, vertexes, vertex_size, render_mode );

     render_unpacked_vertex_array( poly1, pt, 4 );

 }

 void render_vertex_array( uint32_t poly1, uint32_t *vert_array[], int num_vertexes, int vertex_size,

 			  int render_mode ) 

 {

     int i, m=0;

     if( render_mode == RENDER_FULLMOD ) {

 	m = (vertex_size - 3)/2;

     }

     glBegin( GL_TRIANGLE_STRIP );

     for( i=0; i<num_vertexes; i++ ) {

 	uint32_t *vertexes = vert_array[i];

 	float *vertexf = (float *)vert_array[i];

 	uint32_t argb;

 	int k = m + 3;

 	if( POLY1_TEXTURED(poly1) ) {

 	    if( POLY1_UV16(poly1) ) {

 		glTexCoord2f( halftofloat(vertexes[k]>>16),

 			      halftofloat(vertexes[k]) );

 		k++;

 	    } else {

 		glTexCoord2f( vertexf[k], vertexf[k+1] );

 		k+=2;

 	    }

 	}

 	argb = vertexes[k++];

 	if( pvr2_force_fragment_alpha ) {

 	    glColor4ub( (GLubyte)(argb >> 16), (GLubyte)(argb >> 8), 

 			(GLubyte)argb, 0xFF );

 	} else {

 	    glColor4ub( (GLubyte)(argb >> 16), (GLubyte)(argb >> 8), 

 			(GLubyte)argb, (GLubyte)(argb >> 24) );

 	}

 	if( POLY1_SPECULAR(poly1) ) {

 	    uint32_t spec = vertexes[k++];

 	    glSecondaryColor3ubEXT( (GLubyte)(spec >> 16), (GLubyte)(spec >> 8), 

 				 (GLubyte)spec );

 	}

 	glVertex3f( vertexf[0], vertexf[1], 1/vertexf[2] );

 	vertexes += vertex_size;

     }

     glEnd();

 }

 void render_vertexes( uint32_t poly1, uint32_t *vertexes, int num_vertexes, int vertex_size,

 		      int render_mode )

 {

     uint32_t *vert_array[num_vertexes];

     int i;

     for( i=0; i<num_vertexes; i++ ) {

 	vert_array[i] = vertexes;

 	vertexes += vertex_size;

     }

     render_vertex_array( poly1, vert_array, num_vertexes, vertex_size, render_mode );

 }

 /**

  * Render a simple (not auto-sorted) tile

  */

 void render_tile( pvraddr_t tile_entry, int render_mode, gboolean cheap_modifier_mode ) {

     uint32_t poly_bank = MMIO_READ(PVR2,RENDER_POLYBASE);

     uint32_t *tile_list = (uint32_t *)(video_base+tile_entry);

     do {

 	uint32_t entry = *tile_list++;

 	if( entry >> 28 == 0x0F ) {

 	    break;

 	} else if( entry >> 28 == 0x0E ) {

 	    tile_list = (uint32_t *)(video_base + (entry&0x007FFFFF));

 	} else {

 	    uint32_t *polygon = (uint32_t *)(video_base + poly_bank + ((entry & 0x000FFFFF) << 2));

 	    int is_modified = entry & 0x01000000;

 	    int vertex_length = (entry >> 21) & 0x07;

 	    int context_length = 3;

 	    if( is_modified && !cheap_modifier_mode ) {

 		context_length = 5;

 		vertex_length *= 2 ;

 	    }

 	    vertex_length += 3;

 	    if( (entry & 0xE0000000) == 0x80000000 ) {

 		/* Triangle(s) */

 		int strip_count = ((entry >> 25) & 0x0F)+1;

 		int polygon_length = 3 * vertex_length + context_length;

 		int i;

 		for( i=0; i<strip_count; i++ ) {

 		    render_set_context( polygon, render_mode );

 		    render_vertexes( *polygon, polygon+context_length, 3, vertex_length,

 				     render_mode );

 		    polygon += polygon_length;

 		}

 	    } else if( (entry & 0xE0000000) == 0xA0000000 ) {

 		/* Sprite(s) */

 		int strip_count = ((entry >> 25) & 0x0F)+1;

 		int polygon_length = 4 * vertex_length + context_length;

 		int i;

 		for( i=0; i<strip_count; i++ ) {

 		    render_set_context( polygon, render_mode );

 		    render_quad_vertexes( *polygon, polygon+context_length, vertex_length,

 					  render_mode );

 		    polygon += polygon_length;

 		}

 	    } else {

 		/* Polygon */

 		int i, first=-1, last = -1;

 		for( i=0; i<6; i++ ) {

 		    if( entry & (0x40000000>>i) ) {

 			if( first == -1 ) first = i;

 			last = i;

 		    }

 		}

 		if( first != -1 ) {

 		    first = 0;

 		    render_set_context(polygon, render_mode);

 		    render_vertexes( *polygon, polygon+context_length + (first*vertex_length),

 				     (last-first+3), vertex_length, render_mode );

 		}

 	    }

 	}

     } while( 1 );

 }

 void pvr2_render_tilebuffer( int width, int height, int clipx1, int clipy1, 

 			int clipx2, int clipy2 ) {

     pvraddr_t segmentbase = MMIO_READ( PVR2, RENDER_TILEBASE );

     int tile_sort;

     gboolean cheap_shadow;

     int obj_config = MMIO_READ( PVR2, RENDER_OBJCFG );

     int isp_config = MMIO_READ( PVR2, RENDER_ISPCFG );

     int shadow_cfg = MMIO_READ( PVR2, RENDER_SHADOW );

     if( (obj_config & 0x00200000) == 0 ) {

 	if( isp_config & 1 ) {

 	    tile_sort = 0;

 	} else {

 	    tile_sort = 2;

 	}

     } else {

 	tile_sort = 1;

     }

     cheap_shadow = shadow_cfg & 0x100 ? TRUE : FALSE;

     struct tile_segment *segment = (struct tile_segment *)(video_base + segmentbase);

     struct timeval tv_start, tv_end;

     gettimeofday(&tv_start, NULL);

     glEnable( GL_SCISSOR_TEST );

     do {

 	// fwrite_dump32v( (uint32_t *)segment, sizeof(struct tile_segment), 6, stderr );

 	int tilex = SEGMENT_X(segment->control);

 	int tiley = SEGMENT_Y(segment->control);

 	int x1 = tilex << 5;

 	int y1 = tiley << 5;

 	if( x1 + 32 <= clipx1 ||

 	    y1 + 32 <= clipy1 ||

 	    x1 >= clipx2 ||

 	    y1 >= clipy2 ) {

 	    /* Tile completely clipped, skip */

 	    continue;

 	}

 	/* Set a scissor on the visible part of the tile */

 	int w = MIN(x1+32, clipx2) - x1;

 	int h = MIN(y1+32, clipy2) - y1;

 	x1 = MAX(x1,clipx1);

 	y1 = MAX(y1,clipy1);

 	glScissor( x1, height-y1-h, w, h );

 	if( (segment->opaque_ptr & NO_POINTER) == 0 ) {

 	    if( pvr2_debug_render ) {

 		fprintf( stderr, "Tile %d,%d Opaque\n", tilex, tiley );

 		render_print_tilelist( stderr, segment->opaque_ptr );

 	    }

 	    if( (segment->opaquemod_ptr & NO_POINTER) == 0 ) {

 		/* TODO */

 	    }

 	    render_tile( segment->opaque_ptr, RENDER_NORMAL, cheap_shadow );

 	}

 	if( (segment->trans_ptr & NO_POINTER) == 0 ) {

 	    if( pvr2_debug_render ) {

 		fprintf( stderr, "Tile %d,%d Trans\n", tilex, tiley );

 		render_print_tilelist( stderr, segment->trans_ptr );

 	    }

 	    if( (segment->transmod_ptr & NO_POINTER) == 0 ) {

 		/* TODO */

 	    } 

 	    if( tile_sort == 2 || 

 		(tile_sort == 1 && ((segment->control & SEGMENT_SORT_TRANS)==0)) ) {

 		render_autosort_tile( segment->trans_ptr, RENDER_NORMAL, cheap_shadow );

 	    } else {

 		render_tile( segment->trans_ptr, RENDER_NORMAL, cheap_shadow );

 	    }

 	}

 	if( (segment->punchout_ptr & NO_POINTER) == 0 ) {

 	    if( pvr2_debug_render ) {

 		fprintf( stderr, "Tile %d,%d Punchout\n", tilex, tiley );

 		render_print_tilelist( stderr, segment->punchout_ptr );

 	    }

 	    render_tile( segment->punchout_ptr, RENDER_NORMAL, cheap_shadow );

 	}

     } while( ((segment++)->control & SEGMENT_END) == 0 );

     glDisable( GL_SCISSOR_TEST );

     gettimeofday(&tv_end, NULL);

     timersub(&tv_end,&tv_start, &tv_start);

 }

 static float render_find_maximum_tile_z( pvraddr_t tile_entry, float inputz )

 {

     uint32_t poly_bank = MMIO_READ(PVR2,RENDER_POLYBASE);

     uint32_t *tile_list = (uint32_t *)(video_base+tile_entry);

     int shadow_cfg = MMIO_READ( PVR2, RENDER_SHADOW ) & 0x100;

     int i, j;

     float z = inputz;

     do {

 	uint32_t entry = *tile_list++;

 	if( entry >> 28 == 0x0F ) {

 	    break;

 	} else if( entry >> 28 == 0x0E ) {

 	    tile_list = (uint32_t *)(video_base + (entry&0x007FFFFF));

 	} else {

 	    uint32_t *polygon = (uint32_t *)(video_base + poly_bank + ((entry & 0x000FFFFF) << 2));

 	    int is_modified = entry & 0x01000000;

 	    int vertex_length = (entry >> 21) & 0x07;

 	    int context_length = 3;

 	    if( (entry & 0x01000000) && shadow_cfg ) {

 		context_length = 5;

 		vertex_length *= 2 ;

 	    }

 	    vertex_length += 3;

 	    if( (entry & 0xE0000000) == 0x80000000 ) {

 		/* Triangle(s) */

 		int strip_count = ((entry >> 25) & 0x0F)+1;

 		float *vertexz = (float *)(polygon+context_length+2);

 		for( i=0; i<strip_count; i++ ) {

 		    for( j=0; j<3; j++ ) {

 			if( *vertexz > z ) {

 			    z = *vertexz;

 			}

 			vertexz += vertex_length;

 		    }

 		    vertexz += context_length;

 		}

 	    } else if( (entry & 0xE0000000) == 0xA0000000 ) {

 		/* Sprite(s) */

 		int strip_count = ((entry >> 25) & 0x0F)+1;

 		int polygon_length = 4 * vertex_length + context_length;

 		int i;

 		float *vertexz = (float *)(polygon+context_length+2);

 		for( i=0; i<strip_count; i++ ) {

 		    for( j=0; j<4; j++ ) {

 			if( *vertexz > z ) {

 			    z = *vertexz;

 			}

 			vertexz += vertex_length;

 		    }

 		    vertexz+=context_length;

 		}

 	    } else {

 		/* Polygon */

 		int i, first=-1, last = -1;

 		float *vertexz = (float *)polygon+context_length+2;

 		for( i=0; i<6; i++ ) {

 		    if( (entry & (0x40000000>>i)) && *vertexz > z ) {

 			z = *vertexz;

 		    }

 		    vertexz += vertex_length;

 		}

 	    }

 	}

     } while(1);

     return z;

 }

 /**

  * Scan through the scene to determine the largest z value (in order to set up

  * an appropriate near clip plane).

  */

 float pvr2_render_find_maximum_z( )

 {

     pvraddr_t segmentbase = MMIO_READ( PVR2, RENDER_TILEBASE );

     float maximumz = MMIO_READF( PVR2, RENDER_FARCLIP ); /* Initialize to the far clip plane */

     struct tile_segment *segment = (struct tile_segment *)(video_base + segmentbase);

     do {

 	if( (segment->opaque_ptr & NO_POINTER) == 0 ) {

 	    maximumz = render_find_maximum_tile_z(segment->opaque_ptr, maximumz);

 	}

 	if( (segment->opaquemod_ptr & NO_POINTER) == 0 ) {

 	    maximumz = render_find_maximum_tile_z(segment->opaquemod_ptr, maximumz);

 	}

 	if( (segment->trans_ptr & NO_POINTER) == 0 ) {

 	    maximumz = render_find_maximum_tile_z(segment->trans_ptr, maximumz);

 	}

 	if( (segment->transmod_ptr & NO_POINTER) == 0 ) {

 	    maximumz = render_find_maximum_tile_z(segment->transmod_ptr, maximumz);

 	}

 	if( (segment->punchout_ptr & NO_POINTER) == 0 ) {

 	    maximumz = render_find_maximum_tile_z(segment->punchout_ptr, maximumz);

 	}

     } while( ((segment++)->control & SEGMENT_END) == 0 );

     return 1/maximumz;

 }

 /**

  * Scan the segment info to determine the width and height of the render (in 

  * pixels).

  * @param x,y output values to receive the width and height info.

  */

 void pvr2_render_getsize( int *x, int *y ) 

 {

     pvraddr_t segmentbase = MMIO_READ( PVR2, RENDER_TILEBASE );

     int maxx = 0, maxy = 0;

     struct tile_segment *segment = (struct tile_segment *)(video_base + segmentbase);

     do {

 	int tilex = SEGMENT_X(segment->control);

 	int tiley = SEGMENT_Y(segment->control);

 	if( tilex > maxx ) {

 	    maxx = tilex;

 	} 

 	if( tiley > maxy ) {

 	    maxy = tiley;

 	}

     } while( ((segment++)->control & SEGMENT_END) == 0 );

     *x = (maxx+1)<<5;

     *y = (maxy+1)<<5;

 }

 void render_print_vertexes( FILE *f, uint32_t poly1, uint32_t *vert_array[], 

 			    int num_vertexes, int vertex_size )

 {

     char buf[256], *p;

     int i,j, k;

     for( i=0; i<num_vertexes; i++ ) {

 	p = buf;

 	float *vertf = (float *)vert_array[i];

 	uint32_t *verti = (uint32_t *)vert_array[i];

 	p += sprintf( p, "  V %9.5f,%9.5f,%9.5f  ", vertf[0], vertf[1], vertf[2] );

 	k = 3;

 	if( POLY1_TEXTURED(poly1) ) {

 	    if( POLY1_UV16(poly1) ) {

 		p += sprintf( p, "uv=%9.5f,%9.5f  ",

 			       halftofloat(verti[k]>>16),

 			       halftofloat(verti[k]) );

 		k++;

 	    } else {

 		p += sprintf( p, "uv=%9.5f,%9.5f  ", vertf[k], vertf[k+1] );

 		k+=2;

 	    }

 	}

 	p += sprintf( p, "%08X ", verti[k++] );

 	if( POLY1_SPECULAR(poly1) ) {

 	    p += sprintf( p, "%08X", verti[k++] );

 	}

 	p += sprintf( p, "\n" );

 	fprintf( f, buf );

     }

 }

 void render_print_polygon( FILE *f, uint32_t entry )

 {

     uint32_t poly_bank = MMIO_READ(PVR2,RENDER_POLYBASE);

     int shadow_cfg = MMIO_READ( PVR2, RENDER_SHADOW ) & 0x100;

     int i, j;

     if( entry >> 28 == 0x0F ) {

 	fprintf( f, "EOT\n" );

     } else if( entry >> 28 == 0x0E ) {

 	fprintf( f, "LINK %08X\n", entry &0x7FFFFF );

     } else {

 	uint32_t *polygon = (uint32_t *)(video_base + poly_bank + ((entry & 0x000FFFFF) << 2));

 	int is_modified = entry & 0x01000000;

 	int vertex_length = (entry >> 21) & 0x07;

 	int context_length = 3;

 	if( (entry & 0x01000000) && shadow_cfg ) {

 	    context_length = 5;

 	    vertex_length *= 2 ;

 	}

 	vertex_length += 3;

 	if( (entry & 0xE0000000) == 0x80000000 ) {

 	    /* Triangle(s) */

 	    int strip_count = ((entry >> 25) & 0x0F)+1;

 	    for( i=0; i<strip_count; i++ ) {

 		fprintf( f, "TRI  %08X %08X %08X\n", polygon[0], polygon[1], polygon[2] ); 

 		uint32_t *array[3];

 		array[0] = polygon + context_length;

 		array[1] = array[0] + vertex_length;

 		array[2] = array[1] + vertex_length;

 		render_print_vertexes( f, *polygon, array, 3, vertex_length );

 		polygon = array[2] + vertex_length;

 	    }

 	} else if( (entry & 0xE0000000) == 0xA0000000 ) {

 	    /* Sprite(s) */

 	    int strip_count = ((entry >> 25) & 0x0F)+1;

 	    for( i=0; i<strip_count; i++ ) {

 		fprintf( f, "QUAD %08X %08X %08X\n", polygon[0], polygon[1], polygon[2] ); 

 		uint32_t *array[4];

 		array[0] = polygon + context_length;

 		array[1] = array[0] + vertex_length;

 		array[2] = array[1] + vertex_length;

 		array[3] = array[2] + vertex_length;

 		render_print_vertexes( f, *polygon, array, 4, vertex_length );

 		polygon = array[3] + vertex_length;

 	    }

 	} else {

 	    /* Polygon */

 	    int last = -1;

 	    float *vertexz = (float *)polygon+context_length+2;

 	    uint32_t *array[8];

 	    for( i=0; i<6; i++ ) {

 		if( entry & (0x40000000>>i) ) {

 		    last = i;

 		}

 	    }

 	    fprintf( f, "POLY %08X %08X %08X\n", polygon[0], polygon[1], polygon[2] );

 	    for( i=0; i<last+2; i++ ) {

 		array[i] = polygon + context_length + vertex_length*i;

 	    }

 	    render_print_vertexes( f, *polygon, array, last+2, vertex_length );

 	}

     }

 }

 void render_print_tilelist( FILE *f, uint32_t tile_entry )

 {

     uint32_t *tile_list = (uint32_t *)(video_base+tile_entry);

     do {

 	uint32_t entry = *tile_list++;

 	if( entry >> 28 == 0x0F ) {

 	    break;

 	} else if( entry >> 28 == 0x0E ) {

 	    tile_list = (uint32_t *)(video_base + (entry&0x007FFFFF));

 	} else {

 	    render_print_polygon(f, entry);

 	}

     } while( 1 );

 }