aboutsummaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/i965/brw_wm_emit.c
blob: a705d8b34490013af3dd838346e78abe33514374 (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
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
/*
 Copyright (C) Intel Corp.  2006.  All Rights Reserved.
 Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
 develop this 3D driver.
 
 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 <keith@tungstengraphics.com>
  */

#include "util/u_math.h"
#include "tgsi/tgsi_info.h"

#include "brw_context.h"
#include "brw_wm.h"
#include "brw_debug.h"

/* Not quite sure how correct this is - need to understand horiz
 * vs. vertical strides a little better.
 */
static INLINE struct brw_reg sechalf( struct brw_reg reg )
{
   if (reg.vstride)
      reg.nr++;
   return reg;
}

/* Payload R0:
 *
 * R0.0 -- pixel mask, one bit for each of 4 pixels in 4 quads,
 *         corresponding to each of the 16 execution channels.
 * R0.1..8 -- ?
 * R1.0 -- triangle vertex 0.X
 * R1.1 -- triangle vertex 0.Y
 * R1.2 -- quad 0 x,y coords (2 packed uwords)
 * R1.3 -- quad 1 x,y coords (2 packed uwords)
 * R1.4 -- quad 2 x,y coords (2 packed uwords)
 * R1.5 -- quad 3 x,y coords (2 packed uwords)
 * R1.6 -- ?
 * R1.7 -- ?
 * R1.8 -- ?
 */


static void emit_pixel_xy(struct brw_compile *p,
			  const struct brw_reg *dst,
			  GLuint mask)
{
   struct brw_reg r1 = brw_vec1_grf(1, 0);
   struct brw_reg r1_uw = retype(r1, BRW_REGISTER_TYPE_UW);

   brw_set_compression_control(p, BRW_COMPRESSION_NONE);

   /* Calculate pixel centers by adding 1 or 0 to each of the
    * micro-tile coordinates passed in r1.
    */
   if (mask & BRW_WRITEMASK_X) {
      brw_ADD(p,
	      vec16(retype(dst[0], BRW_REGISTER_TYPE_UW)),
	      stride(suboffset(r1_uw, 4), 2, 4, 0),
	      brw_imm_v(0x10101010));
   }

   if (mask & BRW_WRITEMASK_Y) {
      brw_ADD(p,
	      vec16(retype(dst[1], BRW_REGISTER_TYPE_UW)),
	      stride(suboffset(r1_uw,5), 2, 4, 0),
	      brw_imm_v(0x11001100));
   }

   brw_set_compression_control(p, BRW_COMPRESSION_COMPRESSED);
}



static void emit_delta_xy(struct brw_compile *p,
			  const struct brw_reg *dst,
			  GLuint mask,
			  const struct brw_reg *arg0)
{
   struct brw_reg r1 = brw_vec1_grf(1, 0);

   /* Calc delta X,Y by subtracting origin in r1 from the pixel
    * centers.
    */
   if (mask & BRW_WRITEMASK_X) {
      brw_ADD(p,
	      dst[0],
	      retype(arg0[0], BRW_REGISTER_TYPE_UW),
	      negate(r1));
   }

   if (mask & BRW_WRITEMASK_Y) {
      brw_ADD(p,
	      dst[1],
	      retype(arg0[1], BRW_REGISTER_TYPE_UW),
	      negate(suboffset(r1,1)));

   }
}

static void emit_wpos_xy(struct brw_wm_compile *c,
			 const struct brw_reg *dst,
			 GLuint mask,
			 const struct brw_reg *arg0)
{
   struct brw_compile *p = &c->func;

   if (mask & BRW_WRITEMASK_X) {
      /* X' = X */
      brw_MOV(p,
	      dst[0],
	      retype(arg0[0], BRW_REGISTER_TYPE_W));
   }

   /* XXX: is this needed any more, or is this a NOOP?
    */
   if (mask & BRW_WRITEMASK_Y) {
#if 0
      /* Y' = height - 1 - Y */
      brw_ADD(p,
	      dst[1],
	      negate(retype(arg0[1], BRW_REGISTER_TYPE_W)),
	      brw_imm_d(c->key.drawable_height - 1));
#else
      brw_MOV(p,
	      dst[0],
	      retype(arg0[0], BRW_REGISTER_TYPE_W));
#endif
   }
}


static void emit_pixel_w( struct brw_compile *p,
			  const struct brw_reg *dst,
			  GLuint mask,
			  const struct brw_reg *arg0,
			  const struct brw_reg *deltas)
{
   /* Don't need this if all you are doing is interpolating color, for
    * instance.
    */
   if (mask & BRW_WRITEMASK_W) {      
      struct brw_reg interp3 = brw_vec1_grf(arg0[0].nr+1, 4);

      /* Calc 1/w - just linterp wpos[3] optimized by putting the
       * result straight into a message reg.
       */
      brw_LINE(p, brw_null_reg(), interp3, deltas[0]);
      brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), deltas[1]);

      /* Calc w */
      brw_math_16( p, dst[3],
		   BRW_MATH_FUNCTION_INV,
		   BRW_MATH_SATURATE_NONE,
		   2, brw_null_reg(),
		   BRW_MATH_PRECISION_FULL);
   }
}



static void emit_linterp( struct brw_compile *p, 
			 const struct brw_reg *dst,
			 GLuint mask,
			 const struct brw_reg *arg0,
			 const struct brw_reg *deltas )
{
   struct brw_reg interp[4];
   GLuint nr = arg0[0].nr;
   GLuint i;

   interp[0] = brw_vec1_grf(nr, 0);
   interp[1] = brw_vec1_grf(nr, 4);
   interp[2] = brw_vec1_grf(nr+1, 0);
   interp[3] = brw_vec1_grf(nr+1, 4);

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 brw_LINE(p, brw_null_reg(), interp[i], deltas[0]);
	 brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]);
      }
   }
}


static void emit_pinterp( struct brw_compile *p, 
			  const struct brw_reg *dst,
			  GLuint mask,
			  const struct brw_reg *arg0,
			  const struct brw_reg *deltas,
			  const struct brw_reg *w)
{
   struct brw_reg interp[4];
   GLuint nr = arg0[0].nr;
   GLuint i;

   interp[0] = brw_vec1_grf(nr, 0);
   interp[1] = brw_vec1_grf(nr, 4);
   interp[2] = brw_vec1_grf(nr+1, 0);
   interp[3] = brw_vec1_grf(nr+1, 4);

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 brw_LINE(p, brw_null_reg(), interp[i], deltas[0]);
	 brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]);
      }
   }
   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 brw_MUL(p, dst[i], dst[i], w[3]);
      }
   }
}


static void emit_cinterp( struct brw_compile *p, 
			 const struct brw_reg *dst,
			 GLuint mask,
			 const struct brw_reg *arg0 )
{
   struct brw_reg interp[4];
   GLuint nr = arg0[0].nr;
   GLuint i;

   interp[0] = brw_vec1_grf(nr, 0);
   interp[1] = brw_vec1_grf(nr, 4);
   interp[2] = brw_vec1_grf(nr+1, 0);
   interp[3] = brw_vec1_grf(nr+1, 4);

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
         brw_MOV(p, dst[i], suboffset(interp[i],3));	/* TODO: optimize away like other moves */
      }
   }
}

/* Sets the destination channels to 1.0 or 0.0 according to glFrontFacing. */
static void emit_frontfacing( struct brw_compile *p,
			      const struct brw_reg *dst,
			      GLuint mask )
{
   struct brw_reg r1_6ud = retype(brw_vec1_grf(1, 6), BRW_REGISTER_TYPE_UD);
   GLuint i;

   if (!(mask & BRW_WRITEMASK_XYZW))
      return;

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 brw_MOV(p, dst[i], brw_imm_f(0.0));
      }
   }

   /* bit 31 is "primitive is back face", so checking < (1 << 31) gives
    * us front face
    */
   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, r1_6ud, brw_imm_ud(1 << 31));
   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 brw_MOV(p, dst[i], brw_imm_f(1.0));
      }
   }
   brw_set_predicate_control_flag_value(p, 0xff);
}

/* For OPCODE_DDX and OPCODE_DDY, per channel of output we've got input
 * looking like:
 *
 * arg0: q0.tl q0.tr q0.bl q0.br q1.tl q1.tr q1.bl q1.br
 *
 * and we're trying to produce:
 *
 *           DDX                     DDY
 * dst: (q0.tr - q0.tl)     (q0.tl - q0.bl)
 *      (q0.tr - q0.tl)     (q0.tr - q0.br)
 *      (q0.br - q0.bl)     (q0.tl - q0.bl)
 *      (q0.br - q0.bl)     (q0.tr - q0.br)
 *      (q1.tr - q1.tl)     (q1.tl - q1.bl)
 *      (q1.tr - q1.tl)     (q1.tr - q1.br)
 *      (q1.br - q1.bl)     (q1.tl - q1.bl)
 *      (q1.br - q1.bl)     (q1.tr - q1.br)
 *
 * and add two more quads if in 16-pixel dispatch mode.
 *
 * For DDX, it ends up being easy: width = 2, horiz=0 gets us the same result
 * for each pair, and vertstride = 2 jumps us 2 elements after processing a
 * pair. But for DDY, it's harder, as we want to produce the pairs swizzled
 * between each other.  We could probably do it like ddx and swizzle the right
 * order later, but bail for now and just produce
 * ((q0.tl - q0.bl)x4 (q1.tl - q1.bl)x4)
 */
void emit_ddxy(struct brw_compile *p,
	       const struct brw_reg *dst,
	       GLuint mask,
	       GLboolean is_ddx,
	       const struct brw_reg *arg0)
{
   int i;
   struct brw_reg src0, src1;

   if (mask & SATURATE)
      brw_set_saturate(p, 1);
   for (i = 0; i < 4; i++ ) {
      if (mask & (1<<i)) {
	 if (is_ddx) {
	    src0 = brw_reg(arg0[i].file, arg0[i].nr, 1,
			   BRW_REGISTER_TYPE_F,
			   BRW_VERTICAL_STRIDE_2,
			   BRW_WIDTH_2,
			   BRW_HORIZONTAL_STRIDE_0,
			   BRW_SWIZZLE_XYZW, BRW_WRITEMASK_XYZW);
	    src1 = brw_reg(arg0[i].file, arg0[i].nr, 0,
			   BRW_REGISTER_TYPE_F,
			   BRW_VERTICAL_STRIDE_2,
			   BRW_WIDTH_2,
			   BRW_HORIZONTAL_STRIDE_0,
			   BRW_SWIZZLE_XYZW, BRW_WRITEMASK_XYZW);
	 } else {
	    src0 = brw_reg(arg0[i].file, arg0[i].nr, 0,
			   BRW_REGISTER_TYPE_F,
			   BRW_VERTICAL_STRIDE_4,
			   BRW_WIDTH_4,
			   BRW_HORIZONTAL_STRIDE_0,
			   BRW_SWIZZLE_XYZW, BRW_WRITEMASK_XYZW);
	    src1 = brw_reg(arg0[i].file, arg0[i].nr, 2,
			   BRW_REGISTER_TYPE_F,
			   BRW_VERTICAL_STRIDE_4,
			   BRW_WIDTH_4,
			   BRW_HORIZONTAL_STRIDE_0,
			   BRW_SWIZZLE_XYZW, BRW_WRITEMASK_XYZW);
	 }
	 brw_ADD(p, dst[i], src0, negate(src1));
      }
   }
   if (mask & SATURATE)
      brw_set_saturate(p, 0);
}

static void emit_alu1( struct brw_compile *p, 
		       struct brw_instruction *(*func)(struct brw_compile *, 
						       struct brw_reg, 
						       struct brw_reg),
		       const struct brw_reg *dst,
		       GLuint mask,
		       const struct brw_reg *arg0 )
{
   GLuint i;

   if (mask & SATURATE)
      brw_set_saturate(p, 1);

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 func(p, dst[i], arg0[i]);
      }
   }

   if (mask & SATURATE)
      brw_set_saturate(p, 0);
}


static void emit_alu2( struct brw_compile *p, 
		       struct brw_instruction *(*func)(struct brw_compile *, 
						       struct brw_reg, 
						       struct brw_reg, 
						       struct brw_reg),
		       const struct brw_reg *dst,
		       GLuint mask,
		       const struct brw_reg *arg0,
		       const struct brw_reg *arg1 )
{
   GLuint i;

   if (mask & SATURATE)
      brw_set_saturate(p, 1);

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 func(p, dst[i], arg0[i], arg1[i]);
      }
   }

   if (mask & SATURATE)
      brw_set_saturate(p, 0);
}


static void emit_mad( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1,
		      const struct brw_reg *arg2 )
{
   GLuint i;

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 brw_MUL(p, dst[i], arg0[i], arg1[i]);

	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_ADD(p, dst[i], dst[i], arg2[i]);
	 brw_set_saturate(p, 0);
      }
   }
}

static void emit_trunc( struct brw_compile *p,
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0)
{
   GLuint i;

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {
	 brw_RNDZ(p, dst[i], arg0[i]);
      }
   }
}

static void emit_lrp( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1,
		      const struct brw_reg *arg2 )
{
   GLuint i;

   /* Uses dst as a temporary:
    */
   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {	
	 /* Can I use the LINE instruction for this? 
	  */
	 brw_ADD(p, dst[i], negate(arg0[i]), brw_imm_f(1.0));
	 brw_MUL(p, brw_null_reg(), dst[i], arg2[i]);

	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MAC(p, dst[i], arg0[i], arg1[i]);
	 brw_set_saturate(p, 0);
      }
   }
}

static void emit_sop( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      GLuint cond,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   GLuint i;

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {	
	 brw_MOV(p, dst[i], brw_imm_f(0));
	 brw_CMP(p, brw_null_reg(), cond, arg0[i], arg1[i]);
	 brw_MOV(p, dst[i], brw_imm_f(1.0));
	 brw_set_predicate_control_flag_value(p, 0xff);
      }
   }
}

static void emit_slt( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   emit_sop(p, dst, mask, BRW_CONDITIONAL_L, arg0, arg1);
}

static void emit_sle( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   emit_sop(p, dst, mask, BRW_CONDITIONAL_LE, arg0, arg1);
}

static void emit_sgt( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   emit_sop(p, dst, mask, BRW_CONDITIONAL_G, arg0, arg1);
}

static void emit_sge( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   emit_sop(p, dst, mask, BRW_CONDITIONAL_GE, arg0, arg1);
}

static void emit_seq( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   emit_sop(p, dst, mask, BRW_CONDITIONAL_EQ, arg0, arg1);
}

static void emit_sne( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   emit_sop(p, dst, mask, BRW_CONDITIONAL_NEQ, arg0, arg1);
}

static void emit_cmp( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1,
		      const struct brw_reg *arg2 )
{
   GLuint i;

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {	
	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MOV(p, dst[i], arg2[i]);
	 brw_set_saturate(p, 0);

	 brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0[i], brw_imm_f(0));

	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MOV(p, dst[i], arg1[i]);
	 brw_set_saturate(p, 0);
	 brw_set_predicate_control_flag_value(p, 0xff);
      }
   }
}

static void emit_max( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   GLuint i;

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {	
	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MOV(p, dst[i], arg0[i]);
	 brw_set_saturate(p, 0);

	 brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0[i], arg1[i]);

	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MOV(p, dst[i], arg1[i]);
	 brw_set_saturate(p, 0);
	 brw_set_predicate_control_flag_value(p, 0xff);
      }
   }
}

static void emit_min( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   GLuint i;

   for (i = 0; i < 4; i++) {
      if (mask & (1<<i)) {	
	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MOV(p, dst[i], arg1[i]);
	 brw_set_saturate(p, 0);

	 brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0[i], arg1[i]);

	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MOV(p, dst[i], arg0[i]);
	 brw_set_saturate(p, 0);
	 brw_set_predicate_control_flag_value(p, 0xff);
      }
   }
}


static void emit_dp3( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   int dst_chan = ffs(mask & BRW_WRITEMASK_XYZW) - 1;

   if (!(mask & BRW_WRITEMASK_XYZW))
      return; /* Do not emit dead code */

   assert(util_is_power_of_two(mask & BRW_WRITEMASK_XYZW));

   brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]);
   brw_MAC(p, brw_null_reg(), arg0[1], arg1[1]);

   brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
   brw_MAC(p, dst[dst_chan], arg0[2], arg1[2]);
   brw_set_saturate(p, 0);
}


static void emit_dp4( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   int dst_chan = ffs(mask & BRW_WRITEMASK_XYZW) - 1;

   if (!(mask & BRW_WRITEMASK_XYZW))
      return; /* Do not emit dead code */

   assert(util_is_power_of_two(mask & BRW_WRITEMASK_XYZW));

   brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]);
   brw_MAC(p, brw_null_reg(), arg0[1], arg1[1]);
   brw_MAC(p, brw_null_reg(), arg0[2], arg1[2]);

   brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
   brw_MAC(p, dst[dst_chan], arg0[3], arg1[3]);
   brw_set_saturate(p, 0);
}


static void emit_dph( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   const int dst_chan = ffs(mask & BRW_WRITEMASK_XYZW) - 1;

   if (!(mask & BRW_WRITEMASK_XYZW))
      return; /* Do not emit dead code */

   assert(util_is_power_of_two(mask & BRW_WRITEMASK_XYZW));

   brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]);
   brw_MAC(p, brw_null_reg(), arg0[1], arg1[1]);
   brw_MAC(p, dst[dst_chan], arg0[2], arg1[2]);

   brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
   brw_ADD(p, dst[dst_chan], dst[dst_chan], arg1[3]);
   brw_set_saturate(p, 0);
}


static void emit_xpd( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0,
		      const struct brw_reg *arg1 )
{
   GLuint i;

   assert(!(mask & BRW_WRITEMASK_W) == BRW_WRITEMASK_X);
   
   for (i = 0 ; i < 3; i++) {
      if (mask & (1<<i)) {
	 GLuint i2 = (i+2)%3;
	 GLuint i1 = (i+1)%3;

	 brw_MUL(p, brw_null_reg(), negate(arg0[i2]), arg1[i1]);

	 brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
	 brw_MAC(p, dst[i], arg0[i1], arg1[i2]);
	 brw_set_saturate(p, 0);
      }
   }
}


static void emit_math1( struct brw_compile *p, 
			GLuint function,
			const struct brw_reg *dst,
			GLuint mask,
			const struct brw_reg *arg0 )
{
   int dst_chan = ffs(mask & BRW_WRITEMASK_XYZW) - 1;

   if (!(mask & BRW_WRITEMASK_XYZW))
      return; /* Do not emit dead code */

   assert(util_is_power_of_two(mask & BRW_WRITEMASK_XYZW));

   brw_MOV(p, brw_message_reg(2), arg0[0]);

   /* Send two messages to perform all 16 operations:
    */
   brw_math_16(p, 
	       dst[dst_chan],
	       function,
	       (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
	       2,
	       brw_null_reg(),
	       BRW_MATH_PRECISION_FULL);
}


static void emit_math2( struct brw_compile *p, 
			GLuint function,
			const struct brw_reg *dst,
			GLuint mask,
			const struct brw_reg *arg0,
			const struct brw_reg *arg1)
{
   int dst_chan = ffs(mask & BRW_WRITEMASK_XYZW) - 1;

   if (!(mask & BRW_WRITEMASK_XYZW))
      return; /* Do not emit dead code */

   assert(util_is_power_of_two(mask & BRW_WRITEMASK_XYZW));

   brw_push_insn_state(p);

   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
   brw_MOV(p, brw_message_reg(2), arg0[0]);
   brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
   brw_MOV(p, brw_message_reg(4), sechalf(arg0[0]));

   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
   brw_MOV(p, brw_message_reg(3), arg1[0]);
   brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
   brw_MOV(p, brw_message_reg(5), sechalf(arg1[0]));

   
   /* Send two messages to perform all 16 operations:
    */
   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
   brw_math(p, 
	    dst[dst_chan],
	    function,
	    (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
	    2,
	    brw_null_reg(),
	    BRW_MATH_DATA_VECTOR,
	    BRW_MATH_PRECISION_FULL);

   brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
   brw_math(p, 
	    offset(dst[dst_chan],1),
	    function,
	    (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
	    4,
	    brw_null_reg(),
	    BRW_MATH_DATA_VECTOR,
	    BRW_MATH_PRECISION_FULL);
   
   brw_pop_insn_state(p);
}
		     


static void emit_tex( struct brw_wm_compile *c,
		      const struct brw_wm_instruction *inst,
		      struct brw_reg *dst,
		      GLuint dst_flags,
		      struct brw_reg *arg )
{
   struct brw_compile *p = &c->func;
   GLuint msgLength, responseLength;
   GLuint i, nr;
   GLuint emit;
   GLuint msg_type;
   GLboolean shadow = FALSE;

   /* How many input regs are there?
    */
   switch (inst->tex_target) {
   case TGSI_TEXTURE_1D:
      emit = BRW_WRITEMASK_X;
      nr = 1;
      break;
   case TGSI_TEXTURE_SHADOW1D:
      emit = BRW_WRITEMASK_XW;
      nr = 4;
      shadow = TRUE;
      break;
   case TGSI_TEXTURE_2D:
      emit = BRW_WRITEMASK_XY;
      nr = 2;
      break;
   case TGSI_TEXTURE_SHADOW2D:
   case TGSI_TEXTURE_SHADOWRECT:
      emit = BRW_WRITEMASK_XYW;
      nr = 4;
      shadow = TRUE;
      break;
   case TGSI_TEXTURE_3D:
   case TGSI_TEXTURE_CUBE:
      emit = BRW_WRITEMASK_XYZ;
      nr = 3;
      break;
   default:
      /* unexpected target */
      abort();
   }

   msgLength = 1;

   for (i = 0; i < nr; i++) {
      static const GLuint swz[4] = {0,1,2,2};
      if (emit & (1<<i)) 
	 brw_MOV(p, brw_message_reg(msgLength+1), arg[swz[i]]);
      else
	 brw_MOV(p, brw_message_reg(msgLength+1), brw_imm_f(0));
      msgLength += 2;
   }

   responseLength = 8;		/* always */

   if (BRW_IS_IGDNG(p->brw)) {
       if (shadow)
           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_COMPARE_IGDNG;
       else
           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_IGDNG;
   } else {
       if (shadow)
           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_COMPARE;
       else
           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE;
   }

   brw_SAMPLE(p, 
	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
	      1,
	      retype(c->payload.depth[0].hw_reg, BRW_REGISTER_TYPE_UW),
              SURF_INDEX_TEXTURE(inst->tex_unit),
	      inst->tex_unit,	  /* sampler */
	      inst->writemask,
	      msg_type, 
	      responseLength,
	      msgLength,
	      0,	
	      1,
	      BRW_SAMPLER_SIMD_MODE_SIMD16);	
}


static void emit_txb( struct brw_wm_compile *c,
		      const struct brw_wm_instruction *inst,
		      struct brw_reg *dst,
		      GLuint dst_flags,
		      struct brw_reg *arg )
{
   struct brw_compile *p = &c->func;
   GLuint msgLength;
   GLuint msg_type;
   /* Shadow ignored for txb.
    */
   switch (inst->tex_target) {
   case TGSI_TEXTURE_1D:
   case TGSI_TEXTURE_SHADOW1D:
      brw_MOV(p, brw_message_reg(2), arg[0]);
      brw_MOV(p, brw_message_reg(4), brw_imm_f(0));
      brw_MOV(p, brw_message_reg(6), brw_imm_f(0));
      break;
   case TGSI_TEXTURE_2D:
   case TGSI_TEXTURE_RECT:
   case TGSI_TEXTURE_SHADOW2D:
   case TGSI_TEXTURE_SHADOWRECT:
      brw_MOV(p, brw_message_reg(2), arg[0]);
      brw_MOV(p, brw_message_reg(4), arg[1]);
      brw_MOV(p, brw_message_reg(6), brw_imm_f(0));
      break;
   case TGSI_TEXTURE_3D:
   case TGSI_TEXTURE_CUBE:
      brw_MOV(p, brw_message_reg(2), arg[0]);
      brw_MOV(p, brw_message_reg(4), arg[1]);
      brw_MOV(p, brw_message_reg(6), arg[2]);
      break;
   default:
      /* unexpected target */
      abort();
   }

   brw_MOV(p, brw_message_reg(8), arg[3]);
   msgLength = 9;

   if (BRW_IS_IGDNG(p->brw))
       msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS_IGDNG;
   else
       msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS;

   brw_SAMPLE(p, 
	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
	      1,
	      retype(c->payload.depth[0].hw_reg, BRW_REGISTER_TYPE_UW),
              SURF_INDEX_TEXTURE(inst->tex_unit),
	      inst->tex_unit,	  /* sampler */
	      inst->writemask,
	      msg_type,
	      8,		/* responseLength */
	      msgLength,
	      0,	
	      1,
	      BRW_SAMPLER_SIMD_MODE_SIMD16);	
}


static void emit_lit( struct brw_compile *p, 
		      const struct brw_reg *dst,
		      GLuint mask,
		      const struct brw_reg *arg0 )
{
   assert((mask & BRW_WRITEMASK_XW) == 0);

   if (mask & BRW_WRITEMASK_Y) {
      brw_set_saturate(p, (mask & SATURATE) ? 1 : 0);
      brw_MOV(p, dst[1], arg0[0]);
      brw_set_saturate(p, 0);
   }

   if (mask & BRW_WRITEMASK_Z) {
      emit_math2(p, BRW_MATH_FUNCTION_POW,
		 &dst[2],
		 BRW_WRITEMASK_X | (mask & SATURATE),
		 &arg0[1],
		 &arg0[3]);
   }

   /* Ordinarily you'd use an iff statement to skip or shortcircuit
    * some of the POW calculations above, but 16-wide iff statements
    * seem to lock c1 hardware, so this is a nasty workaround:
    */
   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_LE, arg0[0], brw_imm_f(0));
   {
      if (mask & BRW_WRITEMASK_Y) 
	 brw_MOV(p, dst[1], brw_imm_f(0));

      if (mask & BRW_WRITEMASK_Z) 
	 brw_MOV(p, dst[2], brw_imm_f(0)); 
   }
   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
}


/* Kill pixel - set execution mask to zero for those pixels which
 * fail.
 */
static void emit_kil( struct brw_wm_compile *c,
		      struct brw_reg *arg0)
{
   struct brw_compile *p = &c->func;
   struct brw_reg r0uw = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
   GLuint i;
   
   /* XXX - usually won't need 4 compares!
    */
   for (i = 0; i < 4; i++) {
      brw_push_insn_state(p);
      brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_GE, arg0[i], brw_imm_f(0));   
      brw_set_predicate_control_flag_value(p, 0xff);
      brw_set_compression_control(p, BRW_COMPRESSION_NONE);
      brw_AND(p, r0uw, brw_flag_reg(), r0uw);
      brw_pop_insn_state(p);
   }
}

/* KILLP kills the pixels that are currently executing, not based on a test
 * of the arguments.
 */
static void emit_killp( struct brw_wm_compile *c )
{
   struct brw_compile *p = &c->func;
   struct brw_reg r0uw = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);

   brw_push_insn_state(p);
   brw_set_mask_control(p, BRW_MASK_DISABLE);
   brw_NOT(p, c->emit_mask_reg, brw_mask_reg(1)); //IMASK
   brw_AND(p, r0uw, c->emit_mask_reg, r0uw);
   brw_pop_insn_state(p);
}

static void fire_fb_write( struct brw_wm_compile *c,
			   GLuint base_reg,
			   GLuint nr,
			   GLuint target,
			   GLuint eot )
{
   struct brw_compile *p = &c->func;
   
   /* Pass through control information:
    */
/*  mov (8) m1.0<1>:ud   r1.0<8;8,1>:ud   { Align1 NoMask } */
   {
      brw_push_insn_state(p);
      brw_set_mask_control(p, BRW_MASK_DISABLE); /* ? */
      brw_set_compression_control(p, BRW_COMPRESSION_NONE);
      brw_MOV(p, 
	       brw_message_reg(base_reg + 1),
	       brw_vec8_grf(1, 0));
      brw_pop_insn_state(p);
   }

   /* Send framebuffer write message: */
/*  send (16) null.0<1>:uw m0               r0.0<8;8,1>:uw   0x85a04000:ud    { Align1 EOT } */
   brw_fb_WRITE(p,
		retype(vec16(brw_null_reg()), BRW_REGISTER_TYPE_UW),
		base_reg,
		retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW),
		target,		
		nr,
		0, 
		eot);
}


static void emit_aa( struct brw_wm_compile *c,
		     struct brw_reg *arg1,
		     GLuint reg )
{
   struct brw_compile *p = &c->func;
   GLuint comp = c->key.aa_dest_stencil_reg / 2;
   GLuint off = c->key.aa_dest_stencil_reg % 2;
   struct brw_reg aa = offset(arg1[comp], off);

   brw_push_insn_state(p);
   brw_set_compression_control(p, BRW_COMPRESSION_NONE); /* ?? */
   brw_MOV(p, brw_message_reg(reg), aa);
   brw_pop_insn_state(p);
}


/* Post-fragment-program processing.  Send the results to the
 * framebuffer.
 * \param arg0  the fragment color
 * \param arg1  the pass-through depth value
 * \param arg2  the shader-computed depth value
 */
static void emit_fb_write( struct brw_wm_compile *c,
			   struct brw_reg *arg0,
			   struct brw_reg *arg1,
			   struct brw_reg *arg2,
			   GLuint target,
			   GLuint eot)
{
   struct brw_compile *p = &c->func;
   GLuint nr = 2;
   GLuint channel;

   /* Reserve a space for AA - may not be needed:
    */
   if (c->key.aa_dest_stencil_reg)
      nr += 1;

   /* I don't really understand how this achieves the color interleave
    * (ie RGBARGBA) in the result:  [Do the saturation here]
    */
   {
      brw_push_insn_state(p);
      
      for (channel = 0; channel < 4; channel++) {
	 /*  mov (8) m2.0<1>:ud   r28.0<8;8,1>:ud  { Align1 } */
	 /*  mov (8) m6.0<1>:ud   r29.0<8;8,1>:ud  { Align1 SecHalf } */

	 brw_set_compression_control(p, BRW_COMPRESSION_NONE);
	 brw_MOV(p,
		 brw_message_reg(nr + channel),
		 arg0[channel]);
       
	 brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
	 brw_MOV(p,
		 brw_message_reg(nr + channel + 4),
		 sechalf(arg0[channel]));
      }

      /* skip over the regs populated above:
       */
      nr += 8;
   
      brw_pop_insn_state(p);
   }

   if (c->key.source_depth_to_render_target)
   {
      if (c->key.computes_depth) 
	 brw_MOV(p, brw_message_reg(nr), arg2[2]);
      else 
	 brw_MOV(p, brw_message_reg(nr), arg1[1]); /* ? */

      nr += 2;
   }

   if (c->key.dest_depth_reg)
   {
      GLuint comp = c->key.dest_depth_reg / 2;
      GLuint off = c->key.dest_depth_reg % 2;

      if (off != 0) {
         brw_push_insn_state(p);
         brw_set_compression_control(p, BRW_COMPRESSION_NONE);

         brw_MOV(p, brw_message_reg(nr), offset(arg1[comp],1));
         /* 2nd half? */
         brw_MOV(p, brw_message_reg(nr+1), arg1[comp+1]);
         brw_pop_insn_state(p);
      }
      else {
         brw_MOV(p, brw_message_reg(nr), arg1[comp]);
      }
      nr += 2;
   }

   if (!c->key.runtime_check_aads_emit) {
      if (c->key.aa_dest_stencil_reg)
	 emit_aa(c, arg1, 2);

      fire_fb_write(c, 0, nr, target, eot);
   }
   else {
      struct brw_reg v1_null_ud = vec1(retype(brw_null_reg(), BRW_REGISTER_TYPE_UD));
      struct brw_reg ip = brw_ip_reg();
      struct brw_instruction *jmp;
      
      brw_set_compression_control(p, BRW_COMPRESSION_NONE);
      brw_set_conditionalmod(p, BRW_CONDITIONAL_Z);
      brw_AND(p, 
	      v1_null_ud, 
	      get_element_ud(brw_vec8_grf(1,0), 6), 
	      brw_imm_ud(1<<26)); 

      jmp = brw_JMPI(p, ip, ip, brw_imm_d(0));
      {
	 emit_aa(c, arg1, 2);
	 fire_fb_write(c, 0, nr, target, eot);
	 /* note - thread killed in subroutine */
      }
      brw_land_fwd_jump(p, jmp);

      /* ELSE: Shuffle up one register to fill in the hole left for AA:
       */
      fire_fb_write(c, 1, nr-1, target, eot);
   }
}


/**
 * Move a GPR to scratch memory. 
 */
static void emit_spill( struct brw_wm_compile *c,
			struct brw_reg reg,
			GLuint slot )
{
   struct brw_compile *p = &c->func;

   /*
     mov (16) m2.0<1>:ud   r2.0<8;8,1>:ud   { Align1 Compr }
   */
   brw_MOV(p, brw_message_reg(2), reg);

   /*
     mov (1) r0.2<1>:d    0x00000080:d     { Align1 NoMask }
     send (16) null.0<1>:uw m1               r0.0<8;8,1>:uw   0x053003ff:ud    { Align1 }
   */
   brw_dp_WRITE_16(p, 
		   retype(vec16(brw_vec8_grf(0, 0)), BRW_REGISTER_TYPE_UW),
		   slot);
}


/**
 * Load a GPR from scratch memory. 
 */
static void emit_unspill( struct brw_wm_compile *c,
			  struct brw_reg reg,
			  GLuint slot )
{
   struct brw_compile *p = &c->func;

   /* Slot 0 is the undef value.
    */
   if (slot == 0) {
      brw_MOV(p, reg, brw_imm_f(0));
      return;
   }

   /*
     mov (1) r0.2<1>:d    0x000000c0:d     { Align1 NoMask }
     send (16) r110.0<1>:uw m1               r0.0<8;8,1>:uw   0x041243ff:ud    { Align1 }
   */

   brw_dp_READ_16(p,
		  retype(vec16(reg), BRW_REGISTER_TYPE_UW),
		  slot);
}


/**
 * Retrieve up to 4 GEN4 register pairs for the given wm reg:
 * Args with unspill_reg != 0 will be loaded from scratch memory.
 */
static void get_argument_regs( struct brw_wm_compile *c,
			       struct brw_wm_ref *arg[],
			       struct brw_reg *regs )
{
   GLuint i;

   for (i = 0; i < 4; i++) {
      if (arg[i]) {
	 if (arg[i]->unspill_reg)
	    emit_unspill(c,
			 brw_vec8_grf(arg[i]->unspill_reg, 0),
			 arg[i]->value->spill_slot);

	 regs[i] = arg[i]->hw_reg;
      }
      else {
	 regs[i] = brw_null_reg();
      }
   }
}


/**
 * For values that have a spill_slot!=0, write those regs to scratch memory.
 */
static void spill_values( struct brw_wm_compile *c,
			  struct brw_wm_value *values,
			  GLuint nr )
{
   GLuint i;

   for (i = 0; i < nr; i++)
      if (values[i].spill_slot) 
	 emit_spill(c, values[i].hw_reg, values[i].spill_slot);
}


/* Emit the fragment program instructions here.
 */
void brw_wm_emit( struct brw_wm_compile *c )
{
   struct brw_compile *p = &c->func;
   GLuint insn;

   brw_set_compression_control(p, BRW_COMPRESSION_COMPRESSED);

   /* Check if any of the payload regs need to be spilled:
    */
   spill_values(c, c->payload.depth, 4);
   spill_values(c, c->creg, c->nr_creg);
   spill_values(c, c->payload.input_interp, PIPE_MAX_SHADER_INPUTS);
   

   for (insn = 0; insn < c->nr_insns; insn++) {

      struct brw_wm_instruction *inst = &c->instruction[insn];
      struct brw_reg args[3][4], dst[4];
      GLuint i, dst_flags;
      
      /* Get argument regs:
       */
      for (i = 0; i < 3; i++) 
	 get_argument_regs(c, inst->src[i], args[i]);

      /* Get dest regs:
       */
      for (i = 0; i < 4; i++)
	 if (inst->dst[i])
	    dst[i] = inst->dst[i]->hw_reg;
	 else
	    dst[i] = brw_null_reg();
      
      /* Flags
       */
      dst_flags = inst->writemask;
      if (inst->saturate) 
	 dst_flags |= SATURATE;

      switch (inst->opcode) {
	 /* Generated instructions for calculating triangle interpolants:
	  */
      case WM_PIXELXY:
	 emit_pixel_xy(p, dst, dst_flags);
	 break;

      case WM_DELTAXY:
	 emit_delta_xy(p, dst, dst_flags, args[0]);
	 break;

      case WM_WPOSXY:
	 emit_wpos_xy(c, dst, dst_flags, args[0]);
	 break;

      case WM_PIXELW:
	 emit_pixel_w(p, dst, dst_flags, args[0], args[1]);
	 break;

      case WM_LINTERP:
	 emit_linterp(p, dst, dst_flags, args[0], args[1]);
	 break;

      case WM_PINTERP:
	 emit_pinterp(p, dst, dst_flags, args[0], args[1], args[2]);
	 break;

      case WM_CINTERP:
	 emit_cinterp(p, dst, dst_flags, args[0]);
	 break;

      case WM_FB_WRITE:
	 emit_fb_write(c, args[0], args[1], args[2], inst->target, inst->eot);
	 break;

      case WM_FRONTFACING:
	 emit_frontfacing(p, dst, dst_flags);
	 break;

	 /* Straightforward arithmetic:
	  */
      case TGSI_OPCODE_ADD:
	 emit_alu2(p, brw_ADD, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_FRC:
	 emit_alu1(p, brw_FRC, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_FLR:
	 emit_alu1(p, brw_RNDD, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_DDX:
	 emit_ddxy(p, dst, dst_flags, GL_TRUE, args[0]);
	 break;

      case TGSI_OPCODE_DDY:
	 emit_ddxy(p, dst, dst_flags, GL_FALSE, args[0]);
	 break;

      case TGSI_OPCODE_DP3:
	 emit_dp3(p, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_DP4:
	 emit_dp4(p, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_DPH:
	 emit_dph(p, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_TRUNC:
	 emit_trunc(p, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_LRP:
	 emit_lrp(p, dst, dst_flags, args[0], args[1], args[2]);
	 break;

      case TGSI_OPCODE_MAD:	
	 emit_mad(p, dst, dst_flags, args[0], args[1], args[2]);
	 break;

      case TGSI_OPCODE_MOV:
	 emit_alu1(p, brw_MOV, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_MUL:
	 emit_alu2(p, brw_MUL, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_XPD:
	 emit_xpd(p, dst, dst_flags, args[0], args[1]);
	 break;

	 /* Higher math functions:
	  */
      case TGSI_OPCODE_RCP:
	 emit_math1(p, BRW_MATH_FUNCTION_INV, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_RSQ:
	 emit_math1(p, BRW_MATH_FUNCTION_RSQ, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_SIN:
	 emit_math1(p, BRW_MATH_FUNCTION_SIN, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_COS:
	 emit_math1(p, BRW_MATH_FUNCTION_COS, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_EX2:
	 emit_math1(p, BRW_MATH_FUNCTION_EXP, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_LG2:
	 emit_math1(p, BRW_MATH_FUNCTION_LOG, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_SCS:
	 /* There is an scs math function, but it would need some
	  * fixup for 16-element execution.
	  */
	 if (dst_flags & BRW_WRITEMASK_X)
	    emit_math1(p, BRW_MATH_FUNCTION_COS, dst, (dst_flags&SATURATE)|BRW_WRITEMASK_X, args[0]);
	 if (dst_flags & BRW_WRITEMASK_Y)
	    emit_math1(p, BRW_MATH_FUNCTION_SIN, dst+1, (dst_flags&SATURATE)|BRW_WRITEMASK_X, args[0]);
	 break;

      case TGSI_OPCODE_POW:
	 emit_math2(p, BRW_MATH_FUNCTION_POW, dst, dst_flags, args[0], args[1]);
	 break;

	 /* Comparisons:
	  */
      case TGSI_OPCODE_CMP:
	 emit_cmp(p, dst, dst_flags, args[0], args[1], args[2]);
	 break;

      case TGSI_OPCODE_MAX:
	 emit_max(p, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_MIN:
	 emit_min(p, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_SLT:
	 emit_slt(p, dst, dst_flags, args[0], args[1]);
	 break;

      case TGSI_OPCODE_SLE:
	 emit_sle(p, dst, dst_flags, args[0], args[1]);
	break;
      case TGSI_OPCODE_SGT:
	 emit_sgt(p, dst, dst_flags, args[0], args[1]);
	break;
      case TGSI_OPCODE_SGE:
	 emit_sge(p, dst, dst_flags, args[0], args[1]);
	 break;
      case TGSI_OPCODE_SEQ:
	 emit_seq(p, dst, dst_flags, args[0], args[1]);
	break;
      case TGSI_OPCODE_SNE:
	 emit_sne(p, dst, dst_flags, args[0], args[1]);
	break;

      case TGSI_OPCODE_LIT:
	 emit_lit(p, dst, dst_flags, args[0]);
	 break;

	 /* Texturing operations:
	  */
      case TGSI_OPCODE_TEX:
	 emit_tex(c, inst, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_TXB:
	 emit_txb(c, inst, dst, dst_flags, args[0]);
	 break;

      case TGSI_OPCODE_KIL:
	 emit_kil(c, args[0]);
	 break;

      case TGSI_OPCODE_KILP:
	 emit_killp(c);
	 break;

      default:
	 debug_printf("Unsupported opcode %i (%s) in fragment shader\n",
		      inst->opcode, 
		      tgsi_get_opcode_info(inst->opcode)->mnemonic);
      }
      
      for (i = 0; i < 4; i++)
	if (inst->dst[i] && inst->dst[i]->spill_slot) 
	   emit_spill(c, 
		      inst->dst[i]->hw_reg, 
		      inst->dst[i]->spill_slot);
   }

   if (BRW_DEBUG & DEBUG_WM) {
      int i;

      debug_printf("wm-native:\n");
      for (i = 0; i < p->nr_insn; i++)
	 brw_disasm(stderr, &p->store[i]);
      debug_printf("\n");
   }
}