summaryrefslogtreecommitdiff
path: root/src/mesa/pipe/softpipe/sp_tex_sample.c
blob: 64cb94d94419f00a1e34b1e639dc1fbb3624f809 (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
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
/**************************************************************************
 * 
 * Copyright 2007 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, 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 TUNGSTEN GRAPHICS 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.
 * 
 **************************************************************************/

/**
 * Texture sampling
 *
 * Authors:
 *   Brian Paul
 */


#include "sp_context.h"
#include "sp_surface.h"
#include "sp_tex_sample.h"
#include "pipe/p_context.h"
#include "pipe/p_defines.h"
#include "pipe/p_util.h"
#include "pipe/tgsi/exec/tgsi_exec.h"


/*
 * Note, the FRAC macro has to work perfectly.  Otherwise you'll sometimes
 * see 1-pixel bands of improperly weighted linear-filtered textures.
 * The tests/texwrap.c demo is a good test.
 * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0.
 * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x).
 */
#define FRAC(f)  ((f) - ifloor(f))


/**
 * Linear interpolation macro
 */
#define LERP(T, A, B)  ( (A) + (T) * ((B) - (A)) )


/**
 * Do 2D/biliner interpolation of float values.
 * v00, v10, v01 and v11 are typically four texture samples in a square/box.
 * a and b are the horizontal and vertical interpolants.
 * It's important that this function is inlined when compiled with
 * optimization!  If we find that's not true on some systems, convert
 * to a macro.
 */
static INLINE float
lerp_2d(float a, float b,
        float v00, float v10, float v01, float v11)
{
   const float temp0 = LERP(a, v00, v10);
   const float temp1 = LERP(a, v01, v11);
   return LERP(b, temp0, temp1);
}


/**
 * If A is a signed integer, A % B doesn't give the right value for A < 0
 * (in terms of texture repeat).  Just casting to unsigned fixes that.
 */
#define REMAINDER(A, B) ((unsigned) (A) % (unsigned) (B))


/**
 * Apply texture coord wrapping mode and return integer texture index.
 * \param wrapMode  PIPE_TEX_WRAP_x
 * \param s  the texcoord
 * \param size  the texture image size
 * \return  integer texture index
 */
static INLINE int
nearest_texcoord(unsigned wrapMode, float s, unsigned size)
{
   int i;
   switch (wrapMode) {
   case PIPE_TEX_WRAP_REPEAT:
      /* s limited to [0,1) */
      /* i limited to [0,size-1] */
      i = ifloor(s * size);
      i = REMAINDER(i, size);
      return i;
   case PIPE_TEX_WRAP_CLAMP:
      /* s limited to [0,1] */
      /* i limited to [0,size-1] */
      if (s <= 0.0F)
         i = 0;
      else if (s >= 1.0F)
         i = size - 1;
      else
         i = ifloor(s * size);
      return i;
   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
      {
         /* s limited to [min,max] */
         /* i limited to [0, size-1] */
         const float min = 1.0F / (2.0F * size);
         const float max = 1.0F - min;
         if (s < min)
            i = 0;
         else if (s > max)
            i = size - 1;
         else
            i = ifloor(s * size);
      }
      return i;
   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
      {
         /* s limited to [min,max] */
         /* i limited to [-1, size] */
         const float min = -1.0F / (2.0F * size);
         const float max = 1.0F - min;
         if (s <= min)
            i = -1;
         else if (s >= max)
            i = size;
         else
            i = ifloor(s * size);
      }
      return i;
   case PIPE_TEX_WRAP_MIRROR_REPEAT:
      {
         const float min = 1.0F / (2.0F * size);
         const float max = 1.0F - min;
         const int flr = ifloor(s);
         float u;
         if (flr & 1)
            u = 1.0F - (s - (float) flr);
         else
            u = s - (float) flr;
         if (u < min)
            i = 0;
         else if (u > max)
            i = size - 1;
         else
            i = ifloor(u * size);
      }
      return i;
   case PIPE_TEX_WRAP_MIRROR_CLAMP:
      {
         /* s limited to [0,1] */
         /* i limited to [0,size-1] */
         const float u = FABSF(s);
         if (u <= 0.0F)
            i = 0;
         else if (u >= 1.0F)
            i = size - 1;
         else
            i = ifloor(u * size);
      }
      return i;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
      {
         /* s limited to [min,max] */
         /* i limited to [0, size-1] */
         const float min = 1.0F / (2.0F * size);
         const float max = 1.0F - min;
         const float u = FABSF(s);
         if (u < min)
            i = 0;
         else if (u > max)
            i = size - 1;
         else
            i = ifloor(u * size);
      }
      return i;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
      {
         /* s limited to [min,max] */
         /* i limited to [0, size-1] */
         const float min = -1.0F / (2.0F * size);
         const float max = 1.0F - min;
         const float u = FABSF(s);
         if (u < min)
            i = -1;
         else if (u > max)
            i = size;
         else
            i = ifloor(u * size);
      }
      return i;
   default:
      assert(0);
      return 0;
   }
}


/**
 * Used to compute texel locations for linear sampling.
 * \param wrapMode  PIPE_TEX_WRAP_x
 * \param s  the texcoord
 * \param size  the texture image size
 * \param i0  returns first texture index
 * \param i1  returns second texture index (usually *i0 + 1)
 * \param a  returns blend factor/weight between texture indexes
 */
static INLINE void
linear_texcoord(unsigned wrapMode, float s, unsigned size,
                int *i0, int *i1, float *a)
{
   float u;
   switch (wrapMode) {
   case PIPE_TEX_WRAP_REPEAT:
      u = s * size - 0.5F;
      *i0 = REMAINDER(ifloor(u), size);
      *i1 = REMAINDER(*i0 + 1, size);
      break;
   case PIPE_TEX_WRAP_CLAMP:
      if (s <= 0.0F)
         u = 0.0F;
      else if (s >= 1.0F)
         u = (float) size;
      else
         u = s * size;
      u -= 0.5F;
      *i0 = ifloor(u);
      *i1 = *i0 + 1;
      break;
   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
      if (s <= 0.0F)
         u = 0.0F;
      else if (s >= 1.0F)
         u = (float) size;
      else
         u = s * size;
      u -= 0.5F;
      *i0 = ifloor(u);
      *i1 = *i0 + 1;
      if (*i0 < 0)
         *i0 = 0;
      if (*i1 >= (int) size)
         *i1 = size - 1;
      break;
   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
      {
         const float min = -1.0F / (2.0F * size);
         const float max = 1.0F - min;
         if (s <= min)
            u = min * size;
         else if (s >= max)
            u = max * size;
         else
            u = s * size;
         u -= 0.5F;
         *i0 = ifloor(u);
         *i1 = *i0 + 1;
      }
      break;
   case PIPE_TEX_WRAP_MIRROR_REPEAT:
      {
         const int flr = ifloor(s);
         if (flr & 1)
            u = 1.0F - (s - (float) flr);
         else
            u = s - (float) flr;
         u = (u * size) - 0.5F;
         *i0 = ifloor(u);
         *i1 = *i0 + 1;
         if (*i0 < 0)
            *i0 = 0;
         if (*i1 >= (int) size)
            *i1 = size - 1;
      }
      break;
   case PIPE_TEX_WRAP_MIRROR_CLAMP:
      u = FABSF(s);
      if (u >= 1.0F)
         u = (float) size;
      else
         u *= size;
      u -= 0.5F;
      *i0 = ifloor(u);
      *i1 = *i0 + 1;
      break;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
      u = FABSF(s);
      if (u >= 1.0F)
         u = (float) size;
      else
         u *= size;
      u -= 0.5F;
      *i0 = ifloor(u);
      *i1 = *i0 + 1;
      if (*i0 < 0)
         *i0 = 0;
      if (*i1 >= (int) size)
         *i1 = size - 1;
      break;
   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
      {
         const float min = -1.0F / (2.0F * size);
         const float max = 1.0F - min;
         u = FABSF(s);
         if (u <= min)
            u = min * size;
         else if (u >= max)
            u = max * size;
         else
            u *= size;
         u -= 0.5F;
         *i0 = ifloor(u);
         *i1 = *i0 + 1;
      }
      break;
   default:
      assert(0);
   }
   *a = FRAC(u);
}


static unsigned
choose_cube_face(float rx, float ry, float rz, float *newS, float *newT)
{
   /*
      major axis
      direction     target                             sc     tc    ma
      ----------    -------------------------------    ---    ---   ---
       +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx
       -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx
       +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry
       -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry
       +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz
       -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz
   */
   const float arx = FABSF(rx), ary = FABSF(ry), arz = FABSF(rz);
   unsigned face;
   float sc, tc, ma;

   if (arx > ary && arx > arz) {
      if (rx >= 0.0F) {
         face = PIPE_TEX_FACE_POS_X;
         sc = -rz;
         tc = -ry;
         ma = arx;
      }
      else {
         face = PIPE_TEX_FACE_NEG_X;
         sc = rz;
         tc = -ry;
         ma = arx;
      }
   }
   else if (ary > arx && ary > arz) {
      if (ry >= 0.0F) {
         face = PIPE_TEX_FACE_POS_Y;
         sc = rx;
         tc = rz;
         ma = ary;
      }
      else {
         face = PIPE_TEX_FACE_NEG_Y;
         sc = rx;
         tc = -rz;
         ma = ary;
      }
   }
   else {
      if (rz > 0.0F) {
         face = PIPE_TEX_FACE_POS_Z;
         sc = rx;
         tc = -ry;
         ma = arz;
      }
      else {
         face = PIPE_TEX_FACE_NEG_Z;
         sc = -rx;
         tc = -ry;
         ma = arz;
      }
   }

   *newS = ( sc / ma + 1.0F ) * 0.5F;
   *newT = ( tc / ma + 1.0F ) * 0.5F;

   return face;
}


/**
 * Examine the quad's texture coordinates to compute the partial
 * derivatives w.r.t X and Y, then compute lambda (level of detail).
 *
 * This is only done for fragment shaders, not vertex shaders.
 */
static float
compute_lambda(struct tgsi_sampler *sampler,
               const float s[QUAD_SIZE],
               const float t[QUAD_SIZE],
               const float p[QUAD_SIZE],
               float lodbias)
{
   float rho, lambda;

   assert(s);
   {
      float dsdx = s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT];
      float dsdy = s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT];
      dsdx = FABSF(dsdx);
      dsdy = FABSF(dsdy);
      rho = MAX2(dsdx, dsdy);
      if (sampler->state->normalized_coords)
         rho *= sampler->texture->width0;
   }
   if (t) {
      float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT];
      float dtdy = t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT];
      float max;
      dtdx = FABSF(dtdx);
      dtdy = FABSF(dtdy);
      max = MAX2(dtdx, dtdy);
      if (sampler->state->normalized_coords)
         max *= sampler->texture->height0;
      rho = MAX2(rho, max);
   }
   if (p) {
      float dpdx = p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT];
      float dpdy = p[QUAD_TOP_LEFT]     - p[QUAD_BOTTOM_LEFT];
      float max;
      dpdx = FABSF(dpdx);
      dpdy = FABSF(dpdy);
      max = MAX2(dpdx, dpdy);
      if (sampler->state->normalized_coords)
         max *= sampler->texture->depth0;
      rho = MAX2(rho, max);
   }

   lambda = LOG2(rho);

   lambda += lodbias + sampler->state->lod_bias;
   lambda = CLAMP(lambda, sampler->state->min_lod, sampler->state->max_lod);

   return lambda;
}



/**
 * Do several things here:
 * 1. Compute lambda from the texcoords, if needed
 * 2. Determine if we're minifying or magnifying
 * 3. If minifying, choose mipmap levels
 * 4. Return image filter to use within mipmap images
 */
static void
choose_mipmap_levels(struct tgsi_sampler *sampler,
                     const float s[QUAD_SIZE],
                     const float t[QUAD_SIZE],
                     const float p[QUAD_SIZE],
                     float lodbias,
                     unsigned *level0, unsigned *level1, float *levelBlend,
                     unsigned *imgFilter)
{
   if (sampler->state->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) {
      /* no mipmap selection needed */
      *imgFilter = sampler->state->mag_img_filter;
      *level0 = *level1 = sampler->texture->first_level;
   }
   else {
      float lambda;

      if (1)
         /* fragment shader */
         lambda = compute_lambda(sampler, s, t, p, lodbias);
      else
         /* vertex shader */
         lambda = lodbias; /* not really a bias, but absolute LOD */

      if (lambda < 0.0) { /* XXX threshold depends on the filter */
         /* magnifying */
         *imgFilter = sampler->state->mag_img_filter;
         *level0 = *level1 = sampler->texture->first_level;
      }
      else {
         /* minifying */
         *imgFilter = sampler->state->min_img_filter;

         /* choose mipmap level(s) and compute the blend factor between them */
         if (sampler->state->min_mip_filter == PIPE_TEX_MIPFILTER_NEAREST) {
            /* Nearest mipmap level */
            const int lvl = (int) (lambda + 0.5);
            *level0 =
            *level1 = CLAMP(lvl,
                            (int) sampler->texture->first_level,
                            (int) sampler->texture->last_level);
         }
         else {
            /* Linear interpolation between mipmap levels */
            const int lvl = (int) lambda;
            *level0 = CLAMP(lvl,
                            (int) sampler->texture->first_level,
                            (int) sampler->texture->last_level);
            *level1 = CLAMP(lvl + 1,
                            (int) sampler->texture->first_level,
                            (int) sampler->texture->last_level);
            *levelBlend = FRAC(lambda);  /* blending weight between levels */
         }
      }
   }
}



/**
 * Given the texture face, level, zslice, x and y values, compute
 * the cache entry position/index where we'd hope to find the
 * cached texture tile.
 * This is basically a direct-map cache.
 * XXX There's probably lots of ways in which we can improve
 * texture caching....
 */
static unsigned
compute_cache_pos(unsigned face, unsigned level, unsigned zslice,
                  int tx, int ty)
{
#if 01
   unsigned entry = tx + ty * 2 + zslice * 4 + face + level;
   return entry % TEX_CACHE_NUM_ENTRIES;
#else
   return 0;
#endif
}


/**
 * Get a texel from a texture, using the texture tile cache.
 *
 * \param face  the cube face in 0..5
 * \param level  the mipmap level
 * \param x  the x coord of texel within 2D image
 * \param y  the y coord of texel within 2D image
 * \param zslice  which slice of a 3D texture
 * \param rgba  the quad to put the texel/color into
 * \param j  which element of the rgba quad to write to
 */
static void
get_texel(struct tgsi_sampler *sampler,
          unsigned face, unsigned level, int x, int y, unsigned zslice,
          float rgba[NUM_CHANNELS][QUAD_SIZE], unsigned j)
{
   int tx = x / TEX_CACHE_TILE_SIZE;
   int ty = y / TEX_CACHE_TILE_SIZE;
   unsigned entry = compute_cache_pos(face, level, zslice, tx, ty);

   if (tx != sampler->cache[entry].x ||
       ty != sampler->cache[entry].y ||
       face != sampler->cache[entry].face ||
       level != sampler->cache[entry].level ||
       zslice != sampler->cache[entry].zslice) {
      /* entry is not what's expected */
      struct pipe_context *pipe = (struct pipe_context *) sampler->pipe;
      struct pipe_surface *ps
         = pipe->get_tex_surface(pipe, sampler->texture, face, level, zslice);

      /*
      printf("cache miss (%d, %d) face %u\n", tx, ty, face);
      */

      assert(ps->width == sampler->texture->level[level].width);
      assert(ps->height == sampler->texture->level[level].height);
      sampler->cache[entry].x = tx;
      sampler->cache[entry].y = ty;
      sampler->cache[entry].level = level;
      sampler->cache[entry].face = face;
      sampler->cache[entry].zslice = zslice;
      ps->get_tile(ps,
                   tx * TEX_CACHE_TILE_SIZE,
                   ty * TEX_CACHE_TILE_SIZE,
                   TEX_CACHE_TILE_SIZE, TEX_CACHE_TILE_SIZE,
                   (float *) sampler->cache[entry].data);

      pipe_surface_reference(&ps, NULL);
   }
   else {
      /*
      printf("cache hit (%d, %d)\n", x, y);
      */
   }

   /* get the texel from cache entry */
   tx = x % TEX_CACHE_TILE_SIZE;
   ty = y % TEX_CACHE_TILE_SIZE;
   if (sampler->texture->format == PIPE_FORMAT_U_Z16 ||
       sampler->texture->format == PIPE_FORMAT_U_Z32 ||
       sampler->texture->format == PIPE_FORMAT_S8_Z24) {
      /* get_tile() returned one float per texel */
      float *src = (float *) sampler->cache[entry].data;
      rgba[0][j] =
      rgba[1][j] =
      rgba[2][j] =
      rgba[3][j] = src[ty * TEX_CACHE_TILE_SIZE + tx];
   }
   else {
      /* get_tile() returned four floats per texel */
      rgba[0][j] = sampler->cache[entry].data[ty][tx][0];
      rgba[1][j] = sampler->cache[entry].data[ty][tx][1];
      rgba[2][j] = sampler->cache[entry].data[ty][tx][2];
      rgba[3][j] = sampler->cache[entry].data[ty][tx][3];
   }
}



/**
 * Common code for sampling 1D/2D/cube textures.
 * Could probably extend for 3D...
 */
static void
sp_get_samples_2d_common(struct tgsi_sampler *sampler,
                         const float s[QUAD_SIZE],
                         const float t[QUAD_SIZE],
                         const float p[QUAD_SIZE],
                         float lodbias,
                         float rgba[NUM_CHANNELS][QUAD_SIZE],
                         const unsigned faces[4])
{
   unsigned level0, level1, j, imgFilter;
   int width, height;
   float levelBlend;

   choose_mipmap_levels(sampler, s, t, p, lodbias,
                        &level0, &level1, &levelBlend, &imgFilter);

   if (sampler->state->normalized_coords) {
      width = sampler->texture->level[level0].width;
      height = sampler->texture->level[level0].height;
   }
   else {
      width = height = 1.0;
   }

   assert(width > 0);

   switch (imgFilter) {
   case PIPE_TEX_FILTER_NEAREST:
      for (j = 0; j < QUAD_SIZE; j++) {
         int x = nearest_texcoord(sampler->state->wrap_s, s[j], width);
         int y = nearest_texcoord(sampler->state->wrap_t, t[j], height);
         get_texel(sampler, faces[j], level0, x, y, 0, rgba, j);

         if (level0 != level1) {
            /* get texels from second mipmap level and blend */
            float rgba2[4][4];
            unsigned c;
            x = x / 2;
            y = y / 2;
            get_texel(sampler, faces[j], level1, x, y, 0, rgba2, j);
            for (c = 0; c < NUM_CHANNELS; c++) {
               rgba[c][j] = LERP(levelBlend, rgba[c][j], rgba2[c][j]);
            }
         }
      }
      break;
   case PIPE_TEX_FILTER_LINEAR:
      for (j = 0; j < QUAD_SIZE; j++) {
         float tx[4][4], a, b;
         int x0, y0, x1, y1, c;
         linear_texcoord(sampler->state->wrap_s, s[j], width,  &x0, &x1, &a);
         linear_texcoord(sampler->state->wrap_t, t[j], height, &y0, &y1, &b);
         get_texel(sampler, faces[j], level0, x0, y0, 0, tx, 0);
         get_texel(sampler, faces[j], level0, x1, y0, 0, tx, 1);
         get_texel(sampler, faces[j], level0, x0, y1, 0, tx, 2);
         get_texel(sampler, faces[j], level0, x1, y1, 0, tx, 3);
         for (c = 0; c < 4; c++) {
            rgba[c][j] = lerp_2d(a, b, tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
         }

         if (level0 != level1) {
            /* get texels from second mipmap level and blend */
            float rgba2[4][4];
            x0 = x0 / 2;
            y0 = y0 / 2;
            x1 = x1 / 2;
            y1 = y1 / 2;
            get_texel(sampler, faces[j], level1, x0, y0, 0, tx, 0);
            get_texel(sampler, faces[j], level1, x1, y0, 0, tx, 1);
            get_texel(sampler, faces[j], level1, x0, y1, 0, tx, 2);
            get_texel(sampler, faces[j], level1, x1, y1, 0, tx, 3);
            for (c = 0; c < 4; c++) {
               rgba2[c][j] = lerp_2d(a, b,
                                     tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
            }

            for (c = 0; c < NUM_CHANNELS; c++) {
               rgba[c][j] = LERP(levelBlend, rgba[c][j], rgba2[c][j]);
            }
         }
      }
      break;
   default:
      assert(0);
   }
}


static void
sp_get_samples_1d(struct tgsi_sampler *sampler,
                  const float s[QUAD_SIZE],
                  const float t[QUAD_SIZE],
                  const float p[QUAD_SIZE],
                  float lodbias,
                  float rgba[NUM_CHANNELS][QUAD_SIZE])
{
   static const unsigned faces[4] = {0, 0, 0, 0};
   static const float tzero[4] = {0, 0, 0, 0};
   sp_get_samples_2d_common(sampler, s, tzero, NULL, lodbias, rgba, faces);
}


static void
sp_get_samples_2d(struct tgsi_sampler *sampler,
                  const float s[QUAD_SIZE],
                  const float t[QUAD_SIZE],
                  const float p[QUAD_SIZE],
                  float lodbias,
                  float rgba[NUM_CHANNELS][QUAD_SIZE])
{
   static const unsigned faces[4] = {0, 0, 0, 0};
   sp_get_samples_2d_common(sampler, s, t, NULL, lodbias, rgba, faces);
}


static void
sp_get_samples_3d(struct tgsi_sampler *sampler,
                  const float s[QUAD_SIZE],
                  const float t[QUAD_SIZE],
                  const float p[QUAD_SIZE],
                  float lodbias,
                  float rgba[NUM_CHANNELS][QUAD_SIZE])
{
   /* get/map pipe_surfaces corresponding to 3D tex slices */
   unsigned level0, level1, j, imgFilter;
   int width, height, depth;
   float levelBlend;
   const uint face = 0;

   choose_mipmap_levels(sampler, s, t, p, lodbias,
                        &level0, &level1, &levelBlend, &imgFilter);

   if (sampler->state->normalized_coords) {
      width = sampler->texture->level[level0].width;
      height = sampler->texture->level[level0].height;
      depth = sampler->texture->level[level0].depth;
   }
   else {
      width = height = depth = 1.0;
   }

   assert(width > 0);
   assert(height > 0);
   assert(depth > 0);

   switch (imgFilter) {
   case PIPE_TEX_FILTER_NEAREST:
      for (j = 0; j < QUAD_SIZE; j++) {
         int x = nearest_texcoord(sampler->state->wrap_s, s[j], width);
         int y = nearest_texcoord(sampler->state->wrap_t, t[j], height);
         int z = nearest_texcoord(sampler->state->wrap_r, p[j], depth);
         get_texel(sampler, face, level0, x, y, z, rgba, j);

         if (level0 != level1) {
            /* get texels from second mipmap level and blend */
            float rgba2[4][4];
            unsigned c;
            x /= 2;
            y /= 2;
            z /= 2;
            get_texel(sampler, face, level1, x, y, z, rgba2, j);
            for (c = 0; c < NUM_CHANNELS; c++) {
               rgba[c][j] = LERP(levelBlend, rgba2[c][j], rgba[c][j]);
            }
         }
      }
      break;
   case PIPE_TEX_FILTER_LINEAR:
      for (j = 0; j < QUAD_SIZE; j++) {
         float texel0[4][4], texel1[4][4];
         float xw, yw, zw; /* interpolation weights */
         int x0, x1, y0, y1, z0, z1, c;
         linear_texcoord(sampler->state->wrap_s, s[j], width,  &x0, &x1, &xw);
         linear_texcoord(sampler->state->wrap_t, t[j], height, &y0, &y1, &yw);
         linear_texcoord(sampler->state->wrap_r, p[j], depth,  &z0, &z1, &zw);
         get_texel(sampler, face, level0, x0, y0, z0, texel0, 0);
         get_texel(sampler, face, level0, x1, y0, z0, texel0, 1);
         get_texel(sampler, face, level0, x0, y1, z0, texel0, 2);
         get_texel(sampler, face, level0, x1, y1, z0, texel0, 3);
         get_texel(sampler, face, level0, x0, y0, z1, texel1, 0);
         get_texel(sampler, face, level0, x1, y0, z1, texel1, 1);
         get_texel(sampler, face, level0, x0, y1, z1, texel1, 2);
         get_texel(sampler, face, level0, x1, y1, z1, texel1, 3);

         /* 3D lerp */
         for (c = 0; c < 4; c++) {
            float ctemp0[4][4], ctemp1[4][4];
            ctemp0[c][j] = lerp_2d(xw, yw,
                                   texel0[c][0], texel0[c][1],
                                   texel0[c][2], texel0[c][3]);
            ctemp1[c][j] = lerp_2d(xw, yw,
                                   texel1[c][0], texel1[c][1],
                                   texel1[c][2], texel1[c][3]);
            rgba[c][j] = LERP(zw, ctemp0[c][j], ctemp1[c][j]);
         }

         if (level0 != level1) {
            /* get texels from second mipmap level and blend */
            float rgba2[4][4];
            x0 /= 2;
            y0 /= 2;
            z0 /= 2;
            x1 /= 2;
            y1 /= 2;
            z1 /= 2;
            get_texel(sampler, face, level1, x0, y0, z0, texel0, 0);
            get_texel(sampler, face, level1, x1, y0, z0, texel0, 1);
            get_texel(sampler, face, level1, x0, y1, z0, texel0, 2);
            get_texel(sampler, face, level1, x1, y1, z0, texel0, 3);
            get_texel(sampler, face, level1, x0, y0, z1, texel1, 0);
            get_texel(sampler, face, level1, x1, y0, z1, texel1, 1);
            get_texel(sampler, face, level1, x0, y1, z1, texel1, 2);
            get_texel(sampler, face, level1, x1, y1, z1, texel1, 3);

            /* 3D lerp */
            for (c = 0; c < 4; c++) {
               float ctemp0[4][4], ctemp1[4][4];
               ctemp0[c][j] = lerp_2d(xw, yw,
                                      texel0[c][0], texel0[c][1],
                                      texel0[c][2], texel0[c][3]);
               ctemp1[c][j] = lerp_2d(xw, yw,
                                      texel1[c][0], texel1[c][1],
                                      texel1[c][2], texel1[c][3]);
               rgba2[c][j] = LERP(zw, ctemp0[c][j], ctemp1[c][j]);
            }

            /* blend mipmap levels */
            for (c = 0; c < NUM_CHANNELS; c++) {
               rgba[c][j] = LERP(levelBlend, rgba[c][j], rgba2[c][j]);
            }
         }
      }
      break;
   default:
      assert(0);
   }
}


static void
sp_get_samples_cube(struct tgsi_sampler *sampler,
                    const float s[QUAD_SIZE],
                    const float t[QUAD_SIZE],
                    const float p[QUAD_SIZE],
                    float lodbias,
                    float rgba[NUM_CHANNELS][QUAD_SIZE])
{
   unsigned faces[QUAD_SIZE], j;
   float ssss[4], tttt[4];
   for (j = 0; j < QUAD_SIZE; j++) {
      faces[j] = choose_cube_face(s[j], t[j], p[j], ssss + j, tttt + j);
   }
   sp_get_samples_2d_common(sampler, ssss, tttt, NULL, lodbias, rgba, faces);
}


/**
 * Called via tgsi_sampler::get_samples()
 * Use the sampler's state setting to get a filtered RGBA value
 * from the sampler's texture (mipmap tree).
 *
 * XXX we can implement many versions of this function, each
 * tightly coded for a specific combination of sampler state
 * (nearest + repeat), (bilinear mipmap + clamp), etc.
 *
 * The update_samplers() function in st_atom_sampler.c could create
 * a new tgsi_sampler object for each state combo it finds....
 */
void
sp_get_samples(struct tgsi_sampler *sampler,
               const float s[QUAD_SIZE],
               const float t[QUAD_SIZE],
               const float p[QUAD_SIZE],
               float lodbias,
               float rgba[NUM_CHANNELS][QUAD_SIZE])
{
   if (!sampler->texture)
      return;

   switch (sampler->texture->target) {
   case PIPE_TEXTURE_1D:
      sp_get_samples_1d(sampler, s, t, p, lodbias, rgba);
      break;
   case PIPE_TEXTURE_2D:
      sp_get_samples_2d(sampler, s, t, p, lodbias, rgba);
      break;
   case PIPE_TEXTURE_3D:
      sp_get_samples_3d(sampler, s, t, p, lodbias, rgba);
      break;
   case PIPE_TEXTURE_CUBE:
      sp_get_samples_cube(sampler, s, t, p, lodbias, rgba);
      break;
   default:
      assert(0);
   }
}