radeon_vtxfmt.c

来自「mesa-6.5-minigui源码」· C语言 代码 · 共 1,086 行 · 第 1/3 页

/* $XFree86: xc/lib/GL/mesa/src/drv/radeon/radeon_vtxfmt.c,v 1.6 2003/05/06 23:52:08 daenzer Exp $ */
/**************************************************************************

Copyright 2000, 2001 ATI Technologies Inc., Ontario, Canada, and
                     Tungsten Graphics Inc., Cedar Park, Texas.

All Rights Reserved.

Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
"Software"), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:

The above copyright notice and this permission notice (including the
next paragraph) shall be included in all copies or substantial
portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

**************************************************************************/

/*
 * Authors:
 *   Keith Whitwell <[e-mail]>
 */
#include "glheader.h"
#include "imports.h"
#include "api_noop.h"
#include "api_arrayelt.h"
#include "context.h"
#include "mtypes.h"
#include "enums.h"
#include "glapi.h"
#include "colormac.h"
#include "light.h"
#include "state.h"
#include "vtxfmt.h"

#include "tnl/tnl.h"
#include "tnl/t_context.h"
#include "tnl/t_array_api.h"

#include "radeon_context.h"
#include "radeon_state.h"
#include "radeon_ioctl.h"
#include "radeon_tex.h"
#include "radeon_tcl.h"
#include "radeon_swtcl.h"
#include "radeon_vtxfmt.h"

#define VERT_ATTRIB_TEX(u)	(VERT_ATTRIB_TEX0 + (u))

#include "dispatch.h"

static void radeonVtxfmtFlushVertices( GLcontext *, GLuint );

static void count_func( const char *name,  struct dynfn *l )
{
   int i = 0;
   struct dynfn *f;
   foreach (f, l) i++;
   if (i) fprintf(stderr, "%s: %d\n", name, i );
}

static void count_funcs( radeonContextPtr rmesa )
{
   count_func( "Vertex2f", &rmesa->vb.dfn_cache.Vertex2f );
   count_func( "Vertex2fv", &rmesa->vb.dfn_cache.Vertex2fv );
   count_func( "Vertex3f", &rmesa->vb.dfn_cache.Vertex3f );
   count_func( "Vertex3fv", &rmesa->vb.dfn_cache.Vertex3fv );
   count_func( "Color4ub", &rmesa->vb.dfn_cache.Color4ub );
   count_func( "Color4ubv", &rmesa->vb.dfn_cache.Color4ubv );
   count_func( "Color3ub", &rmesa->vb.dfn_cache.Color3ub );
   count_func( "Color3ubv", &rmesa->vb.dfn_cache.Color3ubv );
   count_func( "Color4f", &rmesa->vb.dfn_cache.Color4f );
   count_func( "Color4fv", &rmesa->vb.dfn_cache.Color4fv );
   count_func( "Color3f", &rmesa->vb.dfn_cache.Color3f );
   count_func( "Color3fv", &rmesa->vb.dfn_cache.Color3fv );
   count_func( "SecondaryColor3f", &rmesa->vb.dfn_cache.SecondaryColor3fEXT );
   count_func( "SecondaryColor3fv", &rmesa->vb.dfn_cache.SecondaryColor3fvEXT );
   count_func( "SecondaryColor3ub", &rmesa->vb.dfn_cache.SecondaryColor3ubEXT );
   count_func( "SecondaryColor3ubv", &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT );
   count_func( "Normal3f", &rmesa->vb.dfn_cache.Normal3f );
   count_func( "Normal3fv", &rmesa->vb.dfn_cache.Normal3fv );
   count_func( "TexCoord2f", &rmesa->vb.dfn_cache.TexCoord2f );
   count_func( "TexCoord2fv", &rmesa->vb.dfn_cache.TexCoord2fv );
   count_func( "TexCoord1f", &rmesa->vb.dfn_cache.TexCoord1f );
   count_func( "TexCoord1fv", &rmesa->vb.dfn_cache.TexCoord1fv );
   count_func( "MultiTexCoord2fARB", &rmesa->vb.dfn_cache.MultiTexCoord2fARB );
   count_func( "MultiTexCoord2fvARB", &rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
   count_func( "MultiTexCoord1fARB", &rmesa->vb.dfn_cache.MultiTexCoord1fARB );
   count_func( "MultiTexCoord1fvARB", &rmesa->vb.dfn_cache.MultiTexCoord1fvARB );
}


void radeon_copy_to_current( GLcontext *ctx ) 
{
   GLuint unit;
   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);

   assert(ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT);

   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_N0) {
      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][0] = rmesa->vb.normalptr[0];
      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][1] = rmesa->vb.normalptr[1];
      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][2] = rmesa->vb.normalptr[2];
   }

   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_PKCOLOR) {
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->red );
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->green );
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->blue );
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->alpha );
   } 
   
   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_FPCOLOR) {
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] = rmesa->vb.floatcolorptr[0];
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] = rmesa->vb.floatcolorptr[1];
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] = rmesa->vb.floatcolorptr[2];
   }

   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_FPALPHA)
      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = rmesa->vb.floatcolorptr[3];
      
   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_PKSPEC) {
      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][0] = UBYTE_TO_FLOAT( rmesa->vb.specptr->red );
      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][1] = UBYTE_TO_FLOAT( rmesa->vb.specptr->green );
      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][2] = UBYTE_TO_FLOAT( rmesa->vb.specptr->blue );
   } 

   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
      if (rmesa->vb.vertex_format & RADEON_ST_BIT(unit)) {
	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][0] = rmesa->vb.texcoordptr[unit][0];
	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][1] = rmesa->vb.texcoordptr[unit][1];
	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][2] = 0.0F;
	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][3] = 1.0F;
      }
   }

   ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT;
}

static GLboolean discreet_gl_prim[GL_POLYGON+1] = {
   1,				/* 0 points */
   1,				/* 1 lines */
   0,				/* 2 line_strip */
   0,				/* 3 line_loop */
   1,				/* 4 tris */
   0,				/* 5 tri_fan */
   0,				/* 6 tri_strip */
   1,				/* 7 quads */
   0,				/* 8 quadstrip */
   0,				/* 9 poly */
};

static void flush_prims( radeonContextPtr rmesa )
{
   int i,j;
   struct radeon_dma_region tmp = rmesa->dma.current;
   
   tmp.buf->refcount++;
   tmp.aos_size = rmesa->vb.vertex_size;
   tmp.aos_stride = rmesa->vb.vertex_size;
   tmp.aos_start = GET_START(&tmp);

   rmesa->dma.current.ptr = rmesa->dma.current.start += 
      (rmesa->vb.initial_counter - rmesa->vb.counter) * rmesa->vb.vertex_size * 4; 

   rmesa->tcl.vertex_format = rmesa->vb.vertex_format;
   rmesa->tcl.aos_components[0] = &tmp;
   rmesa->tcl.nr_aos_components = 1;
   rmesa->dma.flush = NULL;

   /* Optimize the primitive list:
    */
   if (rmesa->vb.nrprims > 1) {
      for (j = 0, i = 1 ; i < rmesa->vb.nrprims; i++) {
	 int pj = rmesa->vb.primlist[j].prim & 0xf;
	 int pi = rmesa->vb.primlist[i].prim & 0xf;
      
	 if (pj == pi && discreet_gl_prim[pj] &&
	     rmesa->vb.primlist[i].start == rmesa->vb.primlist[j].end) {
	    rmesa->vb.primlist[j].end = rmesa->vb.primlist[i].end;
	 }
	 else {
	    j++;
	    if (j != i) rmesa->vb.primlist[j] = rmesa->vb.primlist[i];
	 }
      }
      rmesa->vb.nrprims = j+1;
   }

   for (i = 0 ; i < rmesa->vb.nrprims; i++) {
      if (RADEON_DEBUG & DEBUG_PRIMS)
	 fprintf(stderr, "vtxfmt prim %d: %s %d..%d\n", i,
		 _mesa_lookup_enum_by_nr( rmesa->vb.primlist[i].prim & 
					  PRIM_MODE_MASK ),
		 rmesa->vb.primlist[i].start,
		 rmesa->vb.primlist[i].end);

      radeonEmitPrimitive( rmesa->glCtx,
			   rmesa->vb.primlist[i].start,
			   rmesa->vb.primlist[i].end,
			   rmesa->vb.primlist[i].prim );
   }

   rmesa->vb.nrprims = 0;
   radeonReleaseDmaRegion( rmesa, &tmp, __FUNCTION__ );
}


static void start_prim( radeonContextPtr rmesa, GLuint mode )
{
   if (RADEON_DEBUG & DEBUG_VFMT)
      fprintf(stderr, "%s %d\n", __FUNCTION__, rmesa->vb.initial_counter - rmesa->vb.counter);

   rmesa->vb.primlist[rmesa->vb.nrprims].start = rmesa->vb.initial_counter - rmesa->vb.counter;
   rmesa->vb.primlist[rmesa->vb.nrprims].prim = mode;
}

static void note_last_prim( radeonContextPtr rmesa, GLuint flags )
{
   if (RADEON_DEBUG & DEBUG_VFMT)
      fprintf(stderr, "%s %d\n", __FUNCTION__, rmesa->vb.initial_counter - rmesa->vb.counter);

   if (rmesa->vb.prim[0] != GL_POLYGON+1) {
      rmesa->vb.primlist[rmesa->vb.nrprims].prim |= flags;
      rmesa->vb.primlist[rmesa->vb.nrprims].end = rmesa->vb.initial_counter - rmesa->vb.counter;

      if (++(rmesa->vb.nrprims) == RADEON_MAX_PRIMS)
	 flush_prims( rmesa );
   }
}


static void copy_vertex( radeonContextPtr rmesa, GLuint n, GLfloat *dst )
{
   GLuint i;
   GLfloat *src = (GLfloat *)(rmesa->dma.current.address + 
			      rmesa->dma.current.ptr + 
			      (rmesa->vb.primlist[rmesa->vb.nrprims].start + n) * 
			      rmesa->vb.vertex_size * 4);

   if (RADEON_DEBUG & DEBUG_VFMT) 
      fprintf(stderr, "copy_vertex %d\n", rmesa->vb.primlist[rmesa->vb.nrprims].start + n);

   for (i = 0 ; i < rmesa->vb.vertex_size; i++) {
      dst[i] = src[i];
   }
}

/* NOTE: This actually reads the copied vertices back from uncached
 * memory.  Could also use the counter/notify mechanism to populate
 * tmp on the fly as vertices are generated.  
 */
static GLuint copy_dma_verts( radeonContextPtr rmesa, GLfloat (*tmp)[RADEON_MAX_VERTEX_SIZE] )
{
   GLuint ovf, i;
   GLuint nr = (rmesa->vb.initial_counter - rmesa->vb.counter) - rmesa->vb.primlist[rmesa->vb.nrprims].start;

   if (RADEON_DEBUG & DEBUG_VFMT)
      fprintf(stderr, "%s %d verts\n", __FUNCTION__, nr);

   switch( rmesa->vb.prim[0] )
   {
   case GL_POINTS:
      return 0;
   case GL_LINES:
      ovf = nr&1;
      for (i = 0 ; i < ovf ; i++)
	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
      return i;
   case GL_TRIANGLES:
      ovf = nr%3;
      for (i = 0 ; i < ovf ; i++)
	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
      return i;
   case GL_QUADS:
      ovf = nr&3;
      for (i = 0 ; i < ovf ; i++)
	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
      return i;
   case GL_LINE_STRIP:
      if (nr == 0) 
	 return 0;
      copy_vertex( rmesa, nr-1, tmp[0] );
      return 1;
   case GL_LINE_LOOP:
   case GL_TRIANGLE_FAN:
   case GL_POLYGON:
      if (nr == 0) 
	 return 0;
      else if (nr == 1) {
	 copy_vertex( rmesa, 0, tmp[0] );
	 return 1;
      } else {
	 copy_vertex( rmesa, 0, tmp[0] );
	 copy_vertex( rmesa, nr-1, tmp[1] );
	 return 2;
      }
   case GL_TRIANGLE_STRIP:
      ovf = MIN2(nr, 2);
      for (i = 0 ; i < ovf ; i++)
	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
      return i;
   case GL_QUAD_STRIP:
      switch (nr) {
      case 0: ovf = 0; break;
      case 1: ovf = 1; break;
      default: ovf = 2 + (nr&1); break;
      }
      for (i = 0 ; i < ovf ; i++)
	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
      return i;
   default:
      assert(0);
      return 0;
   }
}

static void VFMT_FALLBACK_OUTSIDE_BEGIN_END( const char *caller )
{
   GET_CURRENT_CONTEXT(ctx);
   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);

   if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS))
      fprintf(stderr, "%s from %s\n", __FUNCTION__, caller);

   if (ctx->Driver.NeedFlush) 
      radeonVtxfmtFlushVertices( ctx, ctx->Driver.NeedFlush );

   if (ctx->NewState)
      _mesa_update_state( ctx ); /* clear state so fell_back sticks */

   _tnl_wakeup_exec( ctx );
   ctx->Driver.FlushVertices = radeonFlushVertices;

   assert( rmesa->dma.flush == 0 );
   rmesa->vb.fell_back = GL_TRUE;
   rmesa->vb.installed = GL_FALSE;
}


static void VFMT_FALLBACK( const char *caller )
{
   GET_CURRENT_CONTEXT(ctx);
   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
   GLfloat tmp[3][RADEON_MAX_VERTEX_SIZE];
   GLuint i, prim;
   GLuint ind = rmesa->vb.vertex_format;
   GLuint nrverts;
   GLfloat alpha = 1.0;
   GLuint unit;