summaryrefslogtreecommitdiffstats
path: root/src/mesa/drivers/dri/r300/r300_render.c
blob: acdecb54ec2d5bcb02240d07ed6b96e893bf21a6 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
/**************************************************************************

Copyright (C) 2004 Nicolai Haehnle.

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
on the rights to use, copy, modify, merge, publish, distribute, sub
license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL
ATI, VA LINUX SYSTEMS AND/OR THEIR 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:
 *   Nicolai Haehnle <prefect_@gmx.net>
 */

#include "glheader.h"
#include "state.h"
#include "imports.h"
#include "enums.h"
#include "macros.h"
#include "context.h"
#include "dd.h"
#include "simple_list.h"

#include "api_arrayelt.h"
#include "swrast/swrast.h"
#include "swrast_setup/swrast_setup.h"
#include "array_cache/acache.h"
#include "tnl/tnl.h"
#include "tnl/t_vp_build.h"

#include "radeon_reg.h"
#include "radeon_macros.h"
#include "radeon_ioctl.h"
#include "radeon_state.h"
#include "r300_context.h"
#include "r300_ioctl.h"
#include "r300_state.h"
#include "r300_reg.h"
#include "r300_program.h"
#include "r300_tex.h"
#include "r300_maos.h"
#include "r300_emit.h"

extern int future_hw_tcl_on;

/**********************************************************************
*                     Hardware rasterization
*
* When we fell back to software TCL, we still try to use the
* rasterization hardware for rendering.
**********************************************************************/

static int r300_get_primitive_type(r300ContextPtr rmesa, GLcontext *ctx, int prim)
{
	int type=-1;

	switch (prim & PRIM_MODE_MASK) {
	case GL_POINTS:
	        type=R300_VAP_VF_CNTL__PRIM_POINTS;
      		break;
	case GL_LINES:
	        type=R300_VAP_VF_CNTL__PRIM_LINES;
      		break;
	case GL_LINE_STRIP:
	        type=R300_VAP_VF_CNTL__PRIM_LINE_STRIP;
      		break;
	case GL_LINE_LOOP:
		type=R300_VAP_VF_CNTL__PRIM_LINE_LOOP;
      		break;
    	case GL_TRIANGLES:
	        type=R300_VAP_VF_CNTL__PRIM_TRIANGLES;
      		break;
   	case GL_TRIANGLE_STRIP:
	        type=R300_VAP_VF_CNTL__PRIM_TRIANGLE_STRIP;
      		break;
   	case GL_TRIANGLE_FAN:
	        type=R300_VAP_VF_CNTL__PRIM_TRIANGLE_FAN;
      		break;
	case GL_QUADS:
	        type=R300_VAP_VF_CNTL__PRIM_QUADS;
      		break;
	case GL_QUAD_STRIP:
	        type=R300_VAP_VF_CNTL__PRIM_QUAD_STRIP;
      		break;
	case GL_POLYGON:
		type=R300_VAP_VF_CNTL__PRIM_POLYGON;
		break;
   	default:
 		fprintf(stderr, "%s:%s Do not know how to handle primitive %02x - help me !\n",
			__FILE__, __FUNCTION__,
			prim & PRIM_MODE_MASK);
		return -1;
         	break;
   	}
   return type;
}

static int r300_get_num_verts(r300ContextPtr rmesa,
	GLcontext *ctx,
	int num_verts,
	int prim)
{
	int verts_off=0;
	char *name="UNKNOWN";

	switch (prim & PRIM_MODE_MASK) {
	case GL_POINTS:
   		name="P";
		verts_off = 0;
      		break;
	case GL_LINES:
   		name="L";
		verts_off = num_verts % 2;
      		break;
	case GL_LINE_STRIP:
   		name="LS";
		if(num_verts < 2)
			verts_off = num_verts;
      		break;
	case GL_LINE_LOOP:
   		name="LL";
		if(num_verts < 2)
			verts_off = num_verts;
      		break;
    	case GL_TRIANGLES:
   		name="T";
		verts_off = num_verts % 3;
      		break;
   	case GL_TRIANGLE_STRIP:
   		name="TS";
		if(num_verts < 3)
			verts_off = num_verts;
      		break;
   	case GL_TRIANGLE_FAN:
   		name="TF";
		if(num_verts < 3)
			verts_off = num_verts;
      		break;
	case GL_QUADS:
   		name="Q";
		verts_off = num_verts % 4;
      		break;
	case GL_QUAD_STRIP:
   		name="QS";
		if(num_verts < 4)
			verts_off = num_verts;
		else
			verts_off = num_verts % 2;
      		break;
	case GL_POLYGON:
		name="P";
		if(num_verts < 3)
			verts_off = num_verts;
		break;
   	default:
 		fprintf(stderr, "%s:%s Do not know how to handle primitive %02x - help me !\n",
			__FILE__, __FUNCTION__,
			prim & PRIM_MODE_MASK);
		return -1;
         	break;
   	}

	if (RADEON_DEBUG & DEBUG_VERTS) {
		if (num_verts - verts_off == 0) {
			WARN_ONCE("user error: Need more than %d vertices to draw primitive %s !\n", num_verts, name);
			return 0;
		}

		if (verts_off > 0) {
			WARN_ONCE("user error: %d is not a valid number of vertices for primitive %s !\n", num_verts, name);
		}
	}

	return num_verts - verts_off;
}

/* Immediate implementation has been removed from CVS. */

/* vertex buffer implementation */

static void inline fire_EB(PREFIX unsigned long addr, int vertex_count, int type, int elt_size)
{
	LOCAL_VARS
	unsigned long addr_a;
	unsigned long t_addr;
	unsigned long magic_1, magic_2;
	GLcontext *ctx;
	ctx = rmesa->radeon.glCtx; 
	
	assert(elt_size == 2 || elt_size == 4);
	
	if(addr & (elt_size-1)){
		WARN_ONCE("Badly aligned buffer\n");
		return ;
	}
#ifdef OPTIMIZE_ELTS
	addr_a = 0;
	
	magic_1 = (addr % 32) / 4;
	t_addr = addr & (~0x1d);
	magic_2 = (vertex_count + 1 + (t_addr & 0x2)) / 2 + magic_1;
	
	check_space(6);
	
	start_packet3(RADEON_CP_PACKET3_3D_DRAW_INDX_2, 0);
	if(elt_size == 4){
		e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (vertex_count<<16) | type | R300_VAP_VF_CNTL__INDEX_SIZE_32bit);
	} else {
		e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (vertex_count<<16) | type);
	}

	start_packet3(RADEON_CP_PACKET3_INDX_BUFFER, 2);
	if(elt_size == 4){
		e32(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
		e32(addr /*& 0xffffffe3*/);
	} else {
		e32(R300_EB_UNK1 | (magic_1 << 16) | R300_EB_UNK2);
		e32(t_addr);
	}
	
	if(elt_size == 4){
		e32(vertex_count /*+ addr_a/4*/); /* Total number of dwords needed? */
	} else {
		e32(magic_2); /* Total number of dwords needed? */
	}
	//cp_delay(PASS_PREFIX 1);
#if 0
	fprintf(stderr, "magic_1 %d\n", magic_1);
	fprintf(stderr, "t_addr %x\n", t_addr);
	fprintf(stderr, "magic_2 %d\n", magic_2);
	exit(1);
#endif
#else
	(void)magic_2, (void)magic_1, (void)t_addr;
	
	addr_a = 0;
	
	check_space(6);
	
	start_packet3(RADEON_CP_PACKET3_3D_DRAW_INDX_2, 0);
	if(elt_size == 4){
		e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (vertex_count<<16) | type | R300_VAP_VF_CNTL__INDEX_SIZE_32bit);
	} else {
		e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (vertex_count<<16) | type);
	}

	start_packet3(RADEON_CP_PACKET3_INDX_BUFFER, 2);
	e32(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
	e32(addr /*& 0xffffffe3*/);
	
	if(elt_size == 4){
		e32(vertex_count /*+ addr_a/4*/); /* Total number of dwords needed? */
	} else {
		e32((vertex_count+1)/2 /*+ addr_a/4*/); /* Total number of dwords needed? */
	}
	//cp_delay(PASS_PREFIX 1);
#endif	
}

static void r300_render_vb_primitive(r300ContextPtr rmesa,
	GLcontext *ctx,
	int start,
	int end,
	int prim)
{
   int type, num_verts;

   type=r300_get_primitive_type(rmesa, ctx, prim);
   num_verts=r300_get_num_verts(rmesa, ctx, end-start, prim);

   if(type<0 || num_verts <= 0)return;

   if(rmesa->state.VB.Elts){
	r300EmitAOS(rmesa, rmesa->state.aos_count, /*0*/start);
#if 0
	LOCAL_VARS
	int i;
	start_index32_packet(num_verts, type);
	for(i=0; i < num_verts; i++)
		e32(((unsigned long *)rmesa->state.VB.Elts)[i]/*rmesa->state.Elts[start+i]*/); /* start ? */
#else
	if(num_verts == 1){
		//start_index32_packet(num_verts, type);
		//e32(rmesa->state.Elts[start]);
		return;
	}
	
	if(num_verts > 65535){ /* not implemented yet */
		WARN_ONCE("Too many elts\n");
		return;
	}
	
	r300EmitElts(ctx, rmesa->state.VB.Elts, num_verts, rmesa->state.VB.elt_size);
	fire_EB(PASS_PREFIX rmesa->state.elt_dma.aos_offset, num_verts, type, rmesa->state.VB.elt_size);
#endif
   }else{
	   r300EmitAOS(rmesa, rmesa->state.aos_count, start);
	   fire_AOS(PASS_PREFIX num_verts, type);
   }
}

#if 0
void dump_array(struct r300_dma_region *rvb, int count)
{
	int *out = (int *)(rvb->address + rvb->start);
	int i, ci;
	
	fprintf(stderr, "stride %d:", rvb->aos_stride);
	for (i=0; i < count; i++) {
		fprintf(stderr, "{");
		if (rvb->aos_format == AOS_FORMAT_FLOAT)
			for (ci=0; ci < rvb->aos_size; ci++)
				fprintf(stderr, "%f ", ((float *)out)[ci]);
		else
			for (ci=0; ci < rvb->aos_size; ci++)
				fprintf(stderr, "%d ", ((unsigned char *)out)[ci]);
		fprintf(stderr, "}");
		
		out += rvb->aos_stride;
	}

	fprintf(stderr, "\n");
}
		
void dump_dt(struct dt *dt, int count)
{
	int *out = dt->data;
	int i, ci;
	
	fprintf(stderr, "stride %d", dt->stride);
	
	for (i=0; i < count; i++){
		fprintf(stderr, "{");
		if (dt->type == GL_FLOAT)
			for (ci=0; ci < dt->size; ci++)
				fprintf(stderr, "%f ", ((float *)out)[ci]);
		else
			for (ci=0; ci < dt->size; ci++)
				fprintf(stderr, "%d ", ((unsigned char *)out)[ci]);
		fprintf(stderr, "}");
		
		out = (int *)((char *)out + dt->stride);
	}
	
	fprintf(stderr, "\n");
}
#endif

GLboolean r300_run_vb_render(GLcontext *ctx,
				 struct tnl_pipeline_stage *stage)
{
	r300ContextPtr rmesa = R300_CONTEXT(ctx);
	struct radeon_vertex_buffer *VB = &rmesa->state.VB;
	int i;
	LOCAL_VARS
   
	if (RADEON_DEBUG & DEBUG_PRIMS)
		fprintf(stderr, "%s\n", __FUNCTION__);

	if (stage) {
 		TNLcontext *tnl = TNL_CONTEXT(ctx);
		radeon_vb_to_rvb(rmesa, VB, &tnl->vb);
	}
	
	r300UpdateShaders(rmesa);
	if (rmesa->state.VB.LockCount == 0 || 1) {
 	  	r300ReleaseArrays(ctx);
		r300EmitArrays(ctx, GL_FALSE);

		r300UpdateShaderStates(rmesa);
	} else {
		/* TODO: Figure out why do we need these. */
		R300_STATECHANGE(rmesa, vir[0]);
		R300_STATECHANGE(rmesa, vir[1]);
		R300_STATECHANGE(rmesa, vic);
		R300_STATECHANGE(rmesa, vof);
		
#if 0		
		fprintf(stderr, "dt:\n");
		for(i=0; i < VERT_ATTRIB_MAX; i++){
			fprintf(stderr, "dt %d:", i);
			dump_dt(&rmesa->state.VB.AttribPtr[i], VB->Count);
		}
		
		fprintf(stderr, "before:\n");
		for(i=0; i < rmesa->state.aos_count; i++){
			fprintf(stderr, "aos %d:", i);
			dump_array(&rmesa->state.aos[i], VB->Count);
		}
#endif
#if 0
 	  	r300ReleaseArrays(ctx);
		r300EmitArrays(ctx, GL_FALSE);
			
		fprintf(stderr, "after:\n");
		for(i=0; i < rmesa->state.aos_count; i++){
			fprintf(stderr, "aos %d:", i);
			dump_array(&rmesa->state.aos[i], VB->Count);
		}
#endif
	}
	
	reg_start(R300_RB3D_DSTCACHE_CTLSTAT,0);
	e32(0x0000000a);

	reg_start(0x4f18,0);
	e32(0x00000003);
#if 0
	reg_start(R300_VAP_PVS_WAITIDLE,0);
		e32(0x00000000);
#endif
	r300EmitState(rmesa);
	
	for(i=0; i < VB->PrimitiveCount; i++){
		GLuint prim = VB->Primitive[i].mode;
		GLuint start = VB->Primitive[i].start;
		GLuint length = VB->Primitive[i].count;
		
		r300_render_vb_primitive(rmesa, ctx, start, start + length, prim);
	}

	reg_start(R300_RB3D_DSTCACHE_CTLSTAT,0);
	e32(0x0000000a/*0x2*/);

	reg_start(0x4f18,0);
	e32(0x00000003/*0x1*/);

#ifdef USER_BUFFERS
	r300UseArrays(ctx);
#endif
	return GL_FALSE;
}

#define FALLBACK_IF(expr) \
do {										\
	if (expr) {								\
		if (1 || RADEON_DEBUG & DEBUG_FALLBACKS)			\
			WARN_ONCE("fallback:%s\n", #expr);			\
		return R300_FALLBACK_RAST;							\
	}									\
} while(0)

int r300Fallback(GLcontext *ctx)
{
	int i;

	//FALLBACK_IF(ctx->RenderMode != GL_RENDER);  // We do not do SELECT or FEEDBACK (yet ?)
	
#if 0 /* These should work now.. */
	FALLBACK_IF(ctx->Color.DitherFlag);
	FALLBACK_IF(ctx->Color.AlphaEnabled); // GL_ALPHA_TEST
	FALLBACK_IF(ctx->Color.BlendEnabled); // GL_BLEND
	FALLBACK_IF(ctx->Polygon.OffsetFill); // GL_POLYGON_OFFSET_FILL
#endif
	FALLBACK_IF(ctx->Polygon.OffsetPoint); // GL_POLYGON_OFFSET_POINT
	FALLBACK_IF(ctx->Polygon.OffsetLine); // GL_POLYGON_OFFSET_LINE
	//FALLBACK_IF(ctx->Stencil.Enabled); // GL_STENCIL_TEST
	
	//FALLBACK_IF(ctx->Fog.Enabled); // GL_FOG disable as swtcl doesnt seem to support this
	//FALLBACK_IF(ctx->Polygon.SmoothFlag); // GL_POLYGON_SMOOTH disabling to get blender going
	FALLBACK_IF(ctx->Polygon.StippleFlag); // GL_POLYGON_STIPPLE
	FALLBACK_IF(ctx->Multisample.Enabled); // GL_MULTISAMPLE_ARB
	
	
	FALLBACK_IF(ctx->Line.StippleFlag);
	
	/* HW doesnt appear to directly support these */
	FALLBACK_IF(ctx->Line.SmoothFlag); // GL_LINE_SMOOTH
	FALLBACK_IF(ctx->Point.SmoothFlag); // GL_POINT_SMOOTH
	/* Rest could be done with vertex fragments */
	if (ctx->Extensions.NV_point_sprite || ctx->Extensions.ARB_point_sprite)
		FALLBACK_IF(ctx->Point.PointSprite); // GL_POINT_SPRITE_NV

	for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
		if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_RECT_BIT)
			return R300_FALLBACK_TCL;
		
	return R300_FALLBACK_NONE;
}

/**
 * Called by the pipeline manager to render a batch of primitives.
 * We can return true to pass on to the next stage (i.e. software
 * rasterization) or false to indicate that the pipeline has finished
 * after we render something.
 */
static GLboolean r300_run_render(GLcontext *ctx,
				 struct tnl_pipeline_stage *stage)
{

	if (RADEON_DEBUG & DEBUG_PRIMS)
		fprintf(stderr, "%s\n", __FUNCTION__);

	if (r300Fallback(ctx) >= R300_FALLBACK_RAST)
		return GL_TRUE;

	return r300_run_vb_render(ctx, stage);
}

const struct tnl_pipeline_stage _r300_render_stage = {
	"r300 hw rasterize",
	NULL,
	NULL,
	NULL,
	NULL,
	r300_run_render		/* run */
};

static GLboolean r300_run_tcl_render(GLcontext *ctx,
				 struct tnl_pipeline_stage *stage)
{
	r300ContextPtr rmesa = R300_CONTEXT(ctx);
	struct r300_vertex_program *vp;
   
   	hw_tcl_on=future_hw_tcl_on;
   
	if (RADEON_DEBUG & DEBUG_PRIMS)
		fprintf(stderr, "%s\n", __FUNCTION__);
	if(hw_tcl_on == GL_FALSE)
		return GL_TRUE;
	
	if (r300Fallback(ctx) >= R300_FALLBACK_TCL) {
		hw_tcl_on = GL_FALSE;
		return GL_TRUE;
	}
	
	r300UpdateShaders(rmesa);

	vp = (struct r300_vertex_program *)CURRENT_VERTEX_SHADER(ctx);
#if 0 /* Draw every second request with software arb vp */
	vp->native++;
	vp->native &= 1;
	//vp->native = GL_FALSE;
#endif

#if 0 /* You dont want to know what this does... */
	TNLcontext *tnl = TNL_CONTEXT(ctx);
	struct tnl_cache *cache;
	struct tnl_cache_item *c;
	
	cache = tnl->vp_cache;
	c = cache->items[0xc000cc0e % cache->size];
	
	if(c && c->data == vp)
		vp->native = GL_FALSE;
	
#endif
#if 0
	vp->native = GL_FALSE;
#endif
	if (vp->native == GL_FALSE) {
		hw_tcl_on = GL_FALSE;
		return GL_TRUE;
	}
	//r300UpdateShaderStates(rmesa);
	
	return r300_run_vb_render(ctx, stage);
}

const struct tnl_pipeline_stage _r300_tcl_stage = {
	"r300 tcl",
	NULL,
	NULL,
	NULL,
	NULL,
	r300_run_tcl_render	/* run */
};

/* R300 texture rectangle expects coords in 0..1 range, not 0..dimension
 * as in the extension spec.  Need to translate here.
 *
 * Note that swrast expects 0..dimension, so if a fallback is active,
 * don't do anything.  (Maybe need to configure swrast to match hw)
 */
struct texrect_stage_data {
   GLvector4f texcoord[MAX_TEXTURE_UNITS];
};

#define TEXRECT_STAGE_DATA(stage) ((struct texrect_stage_data *)stage->privatePtr)


static GLboolean run_texrect_stage( GLcontext *ctx,
				    struct tnl_pipeline_stage *stage )
{
   struct texrect_stage_data *store = TEXRECT_STAGE_DATA(stage);
   r300ContextPtr rmesa = R300_CONTEXT(ctx);
   TNLcontext *tnl = TNL_CONTEXT(ctx);
   struct vertex_buffer *VB = &tnl->vb;
   GLuint i;

   if (rmesa->radeon.Fallback)
      return GL_TRUE;

   for (i = 0 ; i < ctx->Const.MaxTextureUnits ; i++) {
      if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_RECT_BIT) {
	 struct gl_texture_object *texObj = ctx->Texture.Unit[i].CurrentRect;
	 struct gl_texture_image *texImage = texObj->Image[0][texObj->BaseLevel];
	 const GLfloat iw = 1.0/texImage->Width;
	 const GLfloat ih = 1.0/texImage->Height;
	 GLfloat *in = (GLfloat *)VB->TexCoordPtr[i]->data;
	 GLint instride = VB->TexCoordPtr[i]->stride;
	 GLfloat (*out)[4] = store->texcoord[i].data;
	 GLint j;

	 store->texcoord[i].size = VB->TexCoordPtr[i]->size;
	 for (j = 0 ; j < VB->Count ; j++) {
	    switch (VB->TexCoordPtr[i]->size) {
	    case 4:
	       out[j][3] = in[3];
	    /* fallthrough */
	    case 3:
	       out[j][2] = in[2];
	    /* fallthrough */
	    default:
	       out[j][0] = in[0] * iw;
	       out[j][1] = in[1] * ih;
	    }
	    in = (GLfloat *)((GLubyte *)in + instride);
	 }

	 VB->AttribPtr[VERT_ATTRIB_TEX0+i] = VB->TexCoordPtr[i] = &store->texcoord[i];
      }
   }

   return GL_TRUE;
}


/* Called the first time stage->run() is invoked.
 */
static GLboolean alloc_texrect_data( GLcontext *ctx,
				     struct tnl_pipeline_stage *stage )
{
   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
   struct texrect_stage_data *store;
   GLuint i;

   stage->privatePtr = CALLOC(sizeof(*store));
   store = TEXRECT_STAGE_DATA(stage);
   if (!store)
      return GL_FALSE;

   for (i = 0 ; i < ctx->Const.MaxTextureUnits ; i++)
      _mesa_vector4f_alloc( &store->texcoord[i], 0, VB->Size, 32 );

   return GL_TRUE;
}

static void free_texrect_data( struct tnl_pipeline_stage *stage )
{
   struct texrect_stage_data *store = TEXRECT_STAGE_DATA(stage);
   GLuint i;

   if (store) {
      for (i = 0 ; i < MAX_TEXTURE_UNITS ; i++)
	 if (store->texcoord[i].data)
	    _mesa_vector4f_free( &store->texcoord[i] );
      FREE( store );
      stage->privatePtr = NULL;
   }
}

const struct tnl_pipeline_stage _r300_texrect_stage =
{
   "r300 texrect stage",			/* name */
   NULL,
   alloc_texrect_data,
   free_texrect_data,
   NULL,
   run_texrect_stage
};