summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri/i965/brw_context.h
blob: 1e21e9214a7fd6762f5a2e7473701c045ac7e2f5 (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
/*
 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>
  */


#ifndef BRWCONTEXT_INC
#define BRWCONTEXT_INC

#include "intel_context.h"
#include "brw_structs.h"
#include "main/imports.h"

#ifdef __cplusplus
extern "C" {
#endif

/* Glossary:
 *
 * URB - uniform resource buffer.  A mid-sized buffer which is
 * partitioned between the fixed function units and used for passing
 * values (vertices, primitives, constants) between them.
 *
 * CURBE - constant URB entry.  An urb region (entry) used to hold
 * constant values which the fixed function units can be instructed to
 * preload into the GRF when spawning a thread.
 *
 * VUE - vertex URB entry.  An urb entry holding a vertex and usually
 * a vertex header.  The header contains control information and
 * things like primitive type, Begin/end flags and clip codes.  
 *
 * PUE - primitive URB entry.  An urb entry produced by the setup (SF)
 * unit holding rasterization and interpolation parameters.
 *
 * GRF - general register file.  One of several register files
 * addressable by programmed threads.  The inputs (r0, payload, curbe,
 * urb) of the thread are preloaded to this area before the thread is
 * spawned.  The registers are individually 8 dwords wide and suitable
 * for general usage.  Registers holding thread input values are not
 * special and may be overwritten.
 *
 * MRF - message register file.  Threads communicate (and terminate)
 * by sending messages.  Message parameters are placed in contiguous
 * MRF registers.  All program output is via these messages.  URB
 * entries are populated by sending a message to the shared URB
 * function containing the new data, together with a control word,
 * often an unmodified copy of R0.
 *
 * R0 - GRF register 0.  Typically holds control information used when
 * sending messages to other threads.
 *
 * EU or GEN4 EU: The name of the programmable subsystem of the
 * i965 hardware.  Threads are executed by the EU, the registers
 * described above are part of the EU architecture.
 *
 * Fixed function units:
 *
 * CS - Command streamer.  Notional first unit, little software
 * interaction.  Holds the URB entries used for constant data, ie the
 * CURBEs.
 *
 * VF/VS - Vertex Fetch / Vertex Shader.  The fixed function part of
 * this unit is responsible for pulling vertices out of vertex buffers
 * in vram and injecting them into the processing pipe as VUEs.  If
 * enabled, it first passes them to a VS thread which is a good place
 * for the driver to implement any active vertex shader.
 *
 * GS - Geometry Shader.  This corresponds to a new DX10 concept.  If
 * enabled, incoming strips etc are passed to GS threads in individual
 * line/triangle/point units.  The GS thread may perform arbitary
 * computation and emit whatever primtives with whatever vertices it
 * chooses.  This makes GS an excellent place to implement GL's
 * unfilled polygon modes, though of course it is capable of much
 * more.  Additionally, GS is used to translate away primitives not
 * handled by latter units, including Quads and Lineloops.
 *
 * CS - Clipper.  Mesa's clipping algorithms are imported to run on
 * this unit.  The fixed function part performs cliptesting against
 * the 6 fixed clipplanes and makes descisions on whether or not the
 * incoming primitive needs to be passed to a thread for clipping.
 * User clip planes are handled via cooperation with the VS thread.
 *
 * SF - Strips Fans or Setup: Triangles are prepared for
 * rasterization.  Interpolation coefficients are calculated.
 * Flatshading and two-side lighting usually performed here.
 *
 * WM - Windower.  Interpolation of vertex attributes performed here.
 * Fragment shader implemented here.  SIMD aspects of EU taken full
 * advantage of, as pixels are processed in blocks of 16.
 *
 * CC - Color Calculator.  No EU threads associated with this unit.
 * Handles blending and (presumably) depth and stencil testing.
 */


#define BRW_MAX_CURBE                    (32*16)

struct brw_context;
struct brw_instruction;
struct brw_vs_prog_key;
struct brw_wm_prog_key;
struct brw_wm_prog_data;

enum brw_state_id {
   BRW_STATE_URB_FENCE,
   BRW_STATE_FRAGMENT_PROGRAM,
   BRW_STATE_VERTEX_PROGRAM,
   BRW_STATE_INPUT_DIMENSIONS,
   BRW_STATE_CURBE_OFFSETS,
   BRW_STATE_REDUCED_PRIMITIVE,
   BRW_STATE_PRIMITIVE,
   BRW_STATE_CONTEXT,
   BRW_STATE_WM_INPUT_DIMENSIONS,
   BRW_STATE_PSP,
   BRW_STATE_SURFACES,
   BRW_STATE_VS_BINDING_TABLE,
   BRW_STATE_GS_BINDING_TABLE,
   BRW_STATE_PS_BINDING_TABLE,
   BRW_STATE_INDICES,
   BRW_STATE_VERTICES,
   BRW_STATE_BATCH,
   BRW_STATE_NR_WM_SURFACES,
   BRW_STATE_NR_VS_SURFACES,
   BRW_STATE_INDEX_BUFFER,
   BRW_STATE_VS_CONSTBUF,
   BRW_STATE_PROGRAM_CACHE,
   BRW_STATE_STATE_BASE_ADDRESS,
   BRW_STATE_SOL_INDICES,
};

#define BRW_NEW_URB_FENCE               (1 << BRW_STATE_URB_FENCE)
#define BRW_NEW_FRAGMENT_PROGRAM        (1 << BRW_STATE_FRAGMENT_PROGRAM)
#define BRW_NEW_VERTEX_PROGRAM          (1 << BRW_STATE_VERTEX_PROGRAM)
#define BRW_NEW_INPUT_DIMENSIONS        (1 << BRW_STATE_INPUT_DIMENSIONS)
#define BRW_NEW_CURBE_OFFSETS           (1 << BRW_STATE_CURBE_OFFSETS)
#define BRW_NEW_REDUCED_PRIMITIVE       (1 << BRW_STATE_REDUCED_PRIMITIVE)
#define BRW_NEW_PRIMITIVE               (1 << BRW_STATE_PRIMITIVE)
#define BRW_NEW_CONTEXT                 (1 << BRW_STATE_CONTEXT)
#define BRW_NEW_WM_INPUT_DIMENSIONS     (1 << BRW_STATE_WM_INPUT_DIMENSIONS)
#define BRW_NEW_PSP                     (1 << BRW_STATE_PSP)
#define BRW_NEW_SURFACES		(1 << BRW_STATE_SURFACES)
#define BRW_NEW_VS_BINDING_TABLE	(1 << BRW_STATE_VS_BINDING_TABLE)
#define BRW_NEW_GS_BINDING_TABLE	(1 << BRW_STATE_GS_BINDING_TABLE)
#define BRW_NEW_PS_BINDING_TABLE	(1 << BRW_STATE_PS_BINDING_TABLE)
#define BRW_NEW_INDICES			(1 << BRW_STATE_INDICES)
#define BRW_NEW_VERTICES		(1 << BRW_STATE_VERTICES)
/**
 * Used for any batch entry with a relocated pointer that will be used
 * by any 3D rendering.
 */
#define BRW_NEW_BATCH                  (1 << BRW_STATE_BATCH)
/** \see brw.state.depth_region */
#define BRW_NEW_INDEX_BUFFER           (1 << BRW_STATE_INDEX_BUFFER)
#define BRW_NEW_VS_CONSTBUF            (1 << BRW_STATE_VS_CONSTBUF)
#define BRW_NEW_PROGRAM_CACHE		(1 << BRW_STATE_PROGRAM_CACHE)
#define BRW_NEW_STATE_BASE_ADDRESS	(1 << BRW_STATE_STATE_BASE_ADDRESS)
#define BRW_NEW_SOL_INDICES		(1 << BRW_STATE_SOL_INDICES)

struct brw_state_flags {
   /** State update flags signalled by mesa internals */
   GLuint mesa;
   /**
    * State update flags signalled as the result of brw_tracked_state updates
    */
   GLuint brw;
   /** State update flags signalled by brw_state_cache.c searches */
   GLuint cache;
};

enum state_struct_type {
   AUB_TRACE_VS_STATE =			1,
   AUB_TRACE_GS_STATE =			2,
   AUB_TRACE_CLIP_STATE =		3,
   AUB_TRACE_SF_STATE =			4,
   AUB_TRACE_WM_STATE =			5,
   AUB_TRACE_CC_STATE =			6,
   AUB_TRACE_CLIP_VP_STATE =		7,
   AUB_TRACE_SF_VP_STATE =		8,
   AUB_TRACE_CC_VP_STATE =		0x9,
   AUB_TRACE_SAMPLER_STATE =		0xa,
   AUB_TRACE_KERNEL_INSTRUCTIONS =	0xb,
   AUB_TRACE_SCRATCH_SPACE =		0xc,
   AUB_TRACE_SAMPLER_DEFAULT_COLOR =    0xd,

   AUB_TRACE_SCISSOR_STATE =		0x15,
   AUB_TRACE_BLEND_STATE =		0x16,
   AUB_TRACE_DEPTH_STENCIL_STATE =	0x17,

   /* Not written to .aub files the same way the structures above are. */
   AUB_TRACE_NO_TYPE =			0x100,
   AUB_TRACE_BINDING_TABLE =		0x101,
   AUB_TRACE_SURFACE_STATE =		0x102,
   AUB_TRACE_VS_CONSTANTS =		0x103,
   AUB_TRACE_WM_CONSTANTS =		0x104,
};

/** Subclass of Mesa vertex program */
struct brw_vertex_program {
   struct gl_vertex_program program;
   GLuint id;
   bool use_const_buffer;
};


/** Subclass of Mesa fragment program */
struct brw_fragment_program {
   struct gl_fragment_program program;
   GLuint id;  /**< serial no. to identify frag progs, never re-used */
};

struct brw_shader {
   struct gl_shader base;

   /** Shader IR transformed for native compile, at link time. */
   struct exec_list *ir;
};

struct brw_shader_program {
   struct gl_shader_program base;
};

enum param_conversion {
   PARAM_NO_CONVERT,
   PARAM_CONVERT_F2I,
   PARAM_CONVERT_F2U,
   PARAM_CONVERT_F2B,
   PARAM_CONVERT_ZERO,
};

/* Data about a particular attempt to compile a program.  Note that
 * there can be many of these, each in a different GL state
 * corresponding to a different brw_wm_prog_key struct, with different
 * compiled programs:
 */
struct brw_wm_prog_data {
   GLuint curb_read_length;
   GLuint urb_read_length;

   GLuint first_curbe_grf;
   GLuint first_curbe_grf_16;
   GLuint reg_blocks;
   GLuint reg_blocks_16;
   GLuint total_scratch;

   GLuint nr_params;       /**< number of float params/constants */
   GLuint nr_pull_params;
   bool error;
   int dispatch_width;
   uint32_t prog_offset_16;

   /**
    * Mask of which interpolation modes are required by the fragment shader.
    * Used in hardware setup on gen6+.
    */
   uint32_t barycentric_interp_modes;

   /* Pointer to tracked values (only valid once
    * _mesa_load_state_parameters has been called at runtime).
    */
   const float *param[MAX_UNIFORMS * 4]; /* should be: BRW_MAX_CURBE */
   enum param_conversion param_convert[MAX_UNIFORMS * 4];
   const float *pull_param[MAX_UNIFORMS * 4];
   enum param_conversion pull_param_convert[MAX_UNIFORMS * 4];
};

/**
 * Enum representing the i965-specific vertex results that don't correspond
 * exactly to any element of gl_vert_result.  The values of this enum are
 * assigned such that they don't conflict with gl_vert_result.
 */
typedef enum
{
   BRW_VERT_RESULT_NDC = VERT_RESULT_MAX,
   BRW_VERT_RESULT_HPOS_DUPLICATE,
   BRW_VERT_RESULT_PAD,
   /*
    * It's actually not a vert_result but just a _mark_ to let sf aware that
    * he need do something special to handle gl_PointCoord builtin variable
    * correctly. see compile_sf_prog() for more info.
    */
   BRW_VERT_RESULT_PNTC,
   BRW_VERT_RESULT_MAX
} brw_vert_result;


/**
 * Data structure recording the relationship between the gl_vert_result enum
 * and "slots" within the vertex URB entry (VUE).  A "slot" is defined as a
 * single octaword within the VUE (128 bits).
 *
 * Note that each BRW register contains 256 bits (2 octawords), so when
 * accessing the VUE in URB_NOSWIZZLE mode, each register corresponds to two
 * consecutive VUE slots.  When accessing the VUE in URB_INTERLEAVED mode (as
 * in a vertex shader), each register corresponds to a single VUE slot, since
 * it contains data for two separate vertices.
 */
struct brw_vue_map {
   /**
    * Map from gl_vert_result value to VUE slot.  For gl_vert_results that are
    * not stored in a slot (because they are not written, or because
    * additional processing is applied before storing them in the VUE), the
    * value is -1.
    */
   int vert_result_to_slot[BRW_VERT_RESULT_MAX];

   /**
    * Map from VUE slot to gl_vert_result value.  For slots that do not
    * directly correspond to a gl_vert_result, the value comes from
    * brw_vert_result.
    *
    * For slots that are not in use, the value is BRW_VERT_RESULT_MAX (this
    * simplifies code that uses the value stored in slot_to_vert_result to
    * create a bit mask).
    */
   int slot_to_vert_result[BRW_VERT_RESULT_MAX];

   /**
    * Total number of VUE slots in use
    */
   int num_slots;
};

/**
 * Convert a VUE slot number into a byte offset within the VUE.
 */
static inline GLuint brw_vue_slot_to_offset(GLuint slot)
{
   return 16*slot;
}

/**
 * Convert a vert_result into a byte offset within the VUE.
 */
static inline GLuint brw_vert_result_to_offset(struct brw_vue_map *vue_map,
                                               GLuint vert_result)
{
   return brw_vue_slot_to_offset(vue_map->vert_result_to_slot[vert_result]);
}


struct brw_sf_prog_data {
   GLuint urb_read_length;
   GLuint total_grf;

   /* Each vertex may have upto 12 attributes, 4 components each,
    * except WPOS which requires only 2.  (11*4 + 2) == 44 ==> 11
    * rows.
    *
    * Actually we use 4 for each, so call it 12 rows.
    */
   GLuint urb_entry_size;
};

struct brw_clip_prog_data {
   GLuint curb_read_length;	/* user planes? */
   GLuint clip_mode;
   GLuint urb_read_length;
   GLuint total_grf;
};

struct brw_gs_prog_data {
   GLuint urb_read_length;
   GLuint total_grf;

   /**
    * Gen6 transform feedback: Amount by which the streaming vertex buffer
    * indices should be incremented each time the GS is invoked.
    */
   unsigned svbi_postincrement_value;
};

struct brw_vs_prog_data {
   struct brw_vue_map vue_map;

   GLuint curb_read_length;
   GLuint urb_read_length;
   GLuint total_grf;
   GLbitfield64 outputs_written;
   GLuint nr_params;       /**< number of float params/constants */
   GLuint nr_pull_params; /**< number of dwords referenced by pull_param[] */
   GLuint total_scratch;

   GLbitfield64 inputs_read;

   /* Used for calculating urb partitions:
    */
   GLuint urb_entry_size;

   const float *param[MAX_UNIFORMS * 4]; /* should be: BRW_MAX_CURBE */
   const float *pull_param[MAX_UNIFORMS * 4];

   bool uses_new_param_layout;
   bool uses_vertexid;
   bool userclip;

   int num_surfaces;
};


/* Size == 0 if output either not written, or always [0,0,0,1]
 */
struct brw_vs_ouput_sizes {
   GLubyte output_size[VERT_RESULT_MAX];
};


/** Number of texture sampler units */
#define BRW_MAX_TEX_UNIT 16

/** Max number of render targets in a shader */
#define BRW_MAX_DRAW_BUFFERS 8

/**
 * Max number of binding table entries used for stream output.
 *
 * From the OpenGL 3.0 spec, table 6.44 (Transform Feedback State), the
 * minimum value of MAX_TRANSFORM_FEEDBACK_INTERLEAVED_COMPONENTS is 64.
 *
 * On Gen6, the size of transform feedback data is limited not by the number
 * of components but by the number of binding table entries we set aside.  We
 * use one binding table entry for a float, one entry for a vector, and one
 * entry per matrix column.  Since the only way we can communicate our
 * transform feedback capabilities to the client is via
 * MAX_TRANSFORM_FEEDBACK_INTERLEAVED_COMPONENTS, we need to plan for the
 * worst case, in which all the varyings are floats, so we use up one binding
 * table entry per component.  Therefore we need to set aside at least 64
 * binding table entries for use by transform feedback.
 *
 * Note: since we don't currently pack varyings, it is currently impossible
 * for the client to actually use up all of these binding table entries--if
 * all of their varyings were floats, they would run out of varying slots and
 * fail to link.  But that's a bug, so it seems prudent to go ahead and
 * allocate the number of binding table entries we will need once the bug is
 * fixed.
 */
#define BRW_MAX_SOL_BINDINGS 64

/** Maximum number of actual buffers used for stream output */
#define BRW_MAX_SOL_BUFFERS 4

/**
 * Helpers to create Surface Binding Table indexes for draw buffers,
 * textures, and constant buffers.
 *
 * Shader threads access surfaces via numeric handles, rather than directly
 * using pointers.  The binding table maps these numeric handles to the
 * address of the actual buffer.
 *
 * For example, a shader might ask to sample from "surface 7."  In this case,
 * bind[7] would contain a pointer to a texture.
 *
 * Currently, our WM binding tables are (arbitrarily) programmed as follows:
 *
 *    +-------------------------------+
 *    |   0 | Draw buffer 0           |
 *    |   . |     .                   |
 *    |   : |     :                   |
 *    |   7 | Draw buffer 7           |
 *    |-----|-------------------------|
 *    |   8 | WM Pull Constant Buffer |
 *    |-----|-------------------------|
 *    |   9 | Texture 0               |
 *    |   . |     .                   |
 *    |   : |     :                   |
 *    |  24 | Texture 15              |
 *    +-------------------------------+
 *
 * Our VS binding tables are programmed as follows:
 *
 *    +-----+-------------------------+
 *    |   0 | VS Pull Constant Buffer |
 *    +-----+-------------------------+
 *    |   1 | Texture 0               |
 *    |   . |     .                   |
 *    |   : |     :                   |
 *    |  16 | Texture 15              |
 *    +-------------------------------+
 *
 * Our (gen6) GS binding tables are programmed as follows:
 *
 *    +-----+-------------------------+
 *    |   0 | SOL Binding 0           |
 *    |   . |     .                   |
 *    |   : |     :                   |
 *    |  63 | SOL Binding 63          |
 *    +-----+-------------------------+
 *
 * Note that nothing actually uses the SURF_INDEX_DRAW macro, so it has to be
 * the identity function or things will break.  We do want to keep draw buffers
 * first so we can use headerless render target writes for RT 0.
 */
#define SURF_INDEX_DRAW(d)           (d)
#define SURF_INDEX_FRAG_CONST_BUFFER (BRW_MAX_DRAW_BUFFERS + 1)
#define SURF_INDEX_TEXTURE(t)        (BRW_MAX_DRAW_BUFFERS + 2 + (t))

/** Maximum size of the binding table. */
#define BRW_MAX_WM_SURFACES          SURF_INDEX_TEXTURE(BRW_MAX_TEX_UNIT)

#define SURF_INDEX_VERT_CONST_BUFFER (0)
#define SURF_INDEX_VS_TEXTURE(t)     (SURF_INDEX_VERT_CONST_BUFFER + 1 + (t))
#define BRW_MAX_VS_SURFACES          SURF_INDEX_VS_TEXTURE(BRW_MAX_TEX_UNIT)

#define SURF_INDEX_SOL_BINDING(t)    ((t))
#define BRW_MAX_GS_SURFACES          SURF_INDEX_SOL_BINDING(BRW_MAX_SOL_BINDINGS)

enum brw_cache_id {
   BRW_BLEND_STATE,
   BRW_DEPTH_STENCIL_STATE,
   BRW_COLOR_CALC_STATE,
   BRW_CC_VP,
   BRW_CC_UNIT,
   BRW_WM_PROG,
   BRW_BLORP_BLIT_PROG,
   BRW_SAMPLER,
   BRW_WM_UNIT,
   BRW_SF_PROG,
   BRW_SF_VP,
   BRW_SF_UNIT, /* scissor state on gen6 */
   BRW_VS_UNIT,
   BRW_VS_PROG,
   BRW_GS_UNIT,
   BRW_GS_PROG,
   BRW_CLIP_VP,
   BRW_CLIP_UNIT,
   BRW_CLIP_PROG,

   BRW_MAX_CACHE
};

struct brw_cache_item {
   /**
    * Effectively part of the key, cache_id identifies what kind of state
    * buffer is involved, and also which brw->state.dirty.cache flag should
    * be set when this cache item is chosen.
    */
   enum brw_cache_id cache_id;
   /** 32-bit hash of the key data */
   GLuint hash;
   GLuint key_size;		/* for variable-sized keys */
   GLuint aux_size;
   const void *key;

   uint32_t offset;
   uint32_t size;

   struct brw_cache_item *next;
};   



struct brw_cache {
   struct brw_context *brw;

   struct brw_cache_item **items;
   drm_intel_bo *bo;
   GLuint size, n_items;

   uint32_t next_offset;
   bool bo_used_by_gpu;
};


/* Considered adding a member to this struct to document which flags
 * an update might raise so that ordering of the state atoms can be
 * checked or derived at runtime.  Dropped the idea in favor of having
 * a debug mode where the state is monitored for flags which are
 * raised that have already been tested against.
 */
struct brw_tracked_state {
   struct brw_state_flags dirty;
   void (*emit)( struct brw_context *brw );
};

/* Flags for brw->state.cache.
 */
#define CACHE_NEW_BLEND_STATE            (1<<BRW_BLEND_STATE)
#define CACHE_NEW_DEPTH_STENCIL_STATE    (1<<BRW_DEPTH_STENCIL_STATE)
#define CACHE_NEW_COLOR_CALC_STATE       (1<<BRW_COLOR_CALC_STATE)
#define CACHE_NEW_CC_VP                  (1<<BRW_CC_VP)
#define CACHE_NEW_CC_UNIT                (1<<BRW_CC_UNIT)
#define CACHE_NEW_WM_PROG                (1<<BRW_WM_PROG)
#define CACHE_NEW_SAMPLER                (1<<BRW_SAMPLER)
#define CACHE_NEW_WM_UNIT                (1<<BRW_WM_UNIT)
#define CACHE_NEW_SF_PROG                (1<<BRW_SF_PROG)
#define CACHE_NEW_SF_VP                  (1<<BRW_SF_VP)
#define CACHE_NEW_SF_UNIT                (1<<BRW_SF_UNIT)
#define CACHE_NEW_VS_UNIT                (1<<BRW_VS_UNIT)
#define CACHE_NEW_VS_PROG                (1<<BRW_VS_PROG)
#define CACHE_NEW_GS_UNIT                (1<<BRW_GS_UNIT)
#define CACHE_NEW_GS_PROG                (1<<BRW_GS_PROG)
#define CACHE_NEW_CLIP_VP                (1<<BRW_CLIP_VP)
#define CACHE_NEW_CLIP_UNIT              (1<<BRW_CLIP_UNIT)
#define CACHE_NEW_CLIP_PROG              (1<<BRW_CLIP_PROG)

struct brw_cached_batch_item {
   struct header *header;
   GLuint sz;
   struct brw_cached_batch_item *next;
};
   


/* Protect against a future where VERT_ATTRIB_MAX > 32.  Wouldn't life
 * be easier if C allowed arrays of packed elements?
 */
#define ATTRIB_BIT_DWORDS  ((VERT_ATTRIB_MAX+31)/32)

struct brw_vertex_buffer {
   /** Buffer object containing the uploaded vertex data */
   drm_intel_bo *bo;
   uint32_t offset;
   /** Byte stride between elements in the uploaded array */
   GLuint stride;
};
struct brw_vertex_element {
   const struct gl_client_array *glarray;

   int buffer;

   /** The corresponding Mesa vertex attribute */
   gl_vert_attrib attrib;
   /** Size of a complete element */
   GLuint element_size;
   /** Offset of the first element within the buffer object */
   unsigned int offset;
};



struct brw_vertex_info {
   GLuint sizes[ATTRIB_BIT_DWORDS * 2]; /* sizes:2[VERT_ATTRIB_MAX] */
};

struct brw_query_object {
   struct gl_query_object Base;

   /** Last query BO associated with this query. */
   drm_intel_bo *bo;
   /** First index in bo with query data for this object. */
   int first_index;
   /** Last index in bo with query data for this object. */
   int last_index;
};


/**
 * brw_context is derived from intel_context.
 */
struct brw_context 
{
   struct intel_context intel;  /**< base class, must be first field */
   GLuint primitive; /**< Hardware primitive, such as _3DPRIM_TRILIST. */

   bool emit_state_always;
   bool has_surface_tile_offset;
   bool has_compr4;
   bool has_negative_rhw_bug;
   bool has_aa_line_parameters;
   bool has_pln;
   bool precompile;

   struct {
      struct brw_state_flags dirty;
   } state;

   struct brw_cache cache;
   struct brw_cached_batch_item *cached_batch_items;

   struct {
      struct brw_vertex_element inputs[VERT_ATTRIB_MAX];
      struct brw_vertex_buffer buffers[VERT_ATTRIB_MAX];
      struct {
	      uint32_t handle;
	      uint32_t offset;
	      uint32_t stride;
      } current_buffers[VERT_ATTRIB_MAX];

      struct brw_vertex_element *enabled[VERT_ATTRIB_MAX];
      GLuint nr_enabled;
      GLuint nr_buffers, nr_current_buffers;

      /* Summary of size and varying of active arrays, so we can check
       * for changes to this state:
       */
      struct brw_vertex_info info;
      unsigned int min_index, max_index;

      /* Offset from start of vertex buffer so we can avoid redefining
       * the same VB packed over and over again.
       */
      unsigned int start_vertex_bias;
   } vb;

   struct {
      /**
       * Index buffer for this draw_prims call.
       *
       * Updates are signaled by BRW_NEW_INDICES.
       */
      const struct _mesa_index_buffer *ib;

      /* Updates are signaled by BRW_NEW_INDEX_BUFFER. */
      drm_intel_bo *bo;
      GLuint type;

      /* Offset to index buffer index to use in CMD_3D_PRIM so that we can
       * avoid re-uploading the IB packet over and over if we're actually
       * referencing the same index buffer.
       */
      unsigned int start_vertex_offset;
   } ib;

   /* Active vertex program: 
    */
   const struct gl_vertex_program *vertex_program;
   const struct gl_fragment_program *fragment_program;

   /* hw-dependent 3DSTATE_VF_STATISTICS opcode */
   uint32_t CMD_VF_STATISTICS;
   /* hw-dependent 3DSTATE_PIPELINE_SELECT opcode */
   uint32_t CMD_PIPELINE_SELECT;

   /**
    * Platform specific constants containing the maximum number of threads
    * for each pipeline stage.
    */
   int max_vs_threads;
   int max_gs_threads;
   int max_wm_threads;

   /* BRW_NEW_URB_ALLOCATIONS:
    */
   struct {
      GLuint vsize;		/* vertex size plus header in urb registers */
      GLuint csize;		/* constant buffer size in urb registers */
      GLuint sfsize;		/* setup data size in urb registers */

      bool constrained;

      GLuint max_vs_entries;	/* Maximum number of VS entries */
      GLuint max_gs_entries;	/* Maximum number of GS entries */

      GLuint nr_vs_entries;
      GLuint nr_gs_entries;
      GLuint nr_clip_entries;
      GLuint nr_sf_entries;
      GLuint nr_cs_entries;

      /* gen6:
       * The length of each URB entry owned by the VS (or GS), as
       * a number of 1024-bit (128-byte) rows.  Should be >= 1.
       *
       * gen7: Same meaning, but in 512-bit (64-byte) rows.
       */
      GLuint vs_size;
      GLuint gs_size;

      GLuint vs_start;
      GLuint gs_start;
      GLuint clip_start;
      GLuint sf_start;
      GLuint cs_start;
      GLuint size; /* Hardware URB size, in KB. */

      /* gen6: True if the most recently sent _3DSTATE_URB message allocated
       * URB space for the GS.
       */
      bool gen6_gs_previously_active;
   } urb;

   
   /* BRW_NEW_CURBE_OFFSETS: 
    */
   struct {
      GLuint wm_start;  /**< pos of first wm const in CURBE buffer */
      GLuint wm_size;   /**< number of float[4] consts, multiple of 16 */
      GLuint clip_start;
      GLuint clip_size;
      GLuint vs_start;
      GLuint vs_size;
      GLuint total_size;

      drm_intel_bo *curbe_bo;
      /** Offset within curbe_bo of space for current curbe entry */
      GLuint curbe_offset;
      /** Offset within curbe_bo of space for next curbe entry */
      GLuint curbe_next_offset;

      /**
       * Copy of the last set of CURBEs uploaded.  Frequently we'll end up
       * in brw_curbe.c with the same set of constant data to be uploaded,
       * so we'd rather not upload new constants in that case (it can cause
       * a pipeline bubble since only up to 4 can be pipelined at a time).
       */
      GLfloat *last_buf;
      /**
       * Allocation for where to calculate the next set of CURBEs.
       * It's a hot enough path that malloc/free of that data matters.
       */
      GLfloat *next_buf;
      GLuint last_bufsz;
   } curbe;

   /** SAMPLER_STATE count and offset */
   struct {
      GLuint count;
      uint32_t offset;
   } sampler;

   struct {
      struct brw_vs_prog_data *prog_data;
      int8_t *constant_map; /* variable array following prog_data */

      drm_intel_bo *scratch_bo;
      drm_intel_bo *const_bo;
      /** Offset in the program cache to the VS program */
      uint32_t prog_offset;
      uint32_t state_offset;

      uint32_t push_const_offset; /* Offset in the batchbuffer */
      int push_const_size; /* in 256-bit register increments */

      /** @{ register allocator */

      struct ra_regs *regs;

      /**
       * Array of the ra classes for the unaligned contiguous register
       * block sizes used.
       */
      int *classes;

      /**
       * Mapping for register-allocated objects in *regs to the first
       * GRF for that object.
      */
      uint8_t *ra_reg_to_grf;
      /** @} */

      uint32_t bind_bo_offset;
      uint32_t surf_offset[BRW_MAX_VS_SURFACES];
   } vs;

   struct {
      struct brw_gs_prog_data *prog_data;

      bool prog_active;
      /** Offset in the program cache to the CLIP program pre-gen6 */
      uint32_t prog_offset;
      uint32_t state_offset;

      uint32_t bind_bo_offset;
      uint32_t surf_offset[BRW_MAX_GS_SURFACES];
   } gs;

   struct {
      struct brw_clip_prog_data *prog_data;

      /** Offset in the program cache to the CLIP program pre-gen6 */
      uint32_t prog_offset;

      /* Offset in the batch to the CLIP state on pre-gen6. */
      uint32_t state_offset;

      /* As of gen6, this is the offset in the batch to the CLIP VP,
       * instead of vp_bo.
       */
      uint32_t vp_offset;
   } clip;


   struct {
      struct brw_sf_prog_data *prog_data;

      /** Offset in the program cache to the CLIP program pre-gen6 */
      uint32_t prog_offset;
      uint32_t state_offset;
      uint32_t vp_offset;
   } sf;

   struct {
      struct brw_wm_prog_data *prog_data;
      struct brw_wm_compile *compile_data;

      /** Input sizes, calculated from active vertex program.
       * One bit per fragment program input attribute.
       */
      GLbitfield input_size_masks[4];

      /** offsets in the batch to sampler default colors (texture border color)
       */
      uint32_t sdc_offset[BRW_MAX_TEX_UNIT];

      GLuint render_surf;

      drm_intel_bo *scratch_bo;

      /** Offset in the program cache to the WM program */
      uint32_t prog_offset;

      uint32_t state_offset; /* offset in batchbuffer to pre-gen6 WM state */

      drm_intel_bo *const_bo; /* pull constant buffer. */
      /**
       * This is offset in the batch to the push constants on gen6.
       *
       * Pre-gen6, push constants live in the CURBE.
       */
      uint32_t push_const_offset;

      /** Binding table of pointers to surf_bo entries */
      uint32_t bind_bo_offset;
      uint32_t surf_offset[BRW_MAX_WM_SURFACES];

      /** @{ register allocator */

      struct ra_regs *regs;

      /** Array of the ra classes for the unaligned contiguous
       * register block sizes used.
       */
      int *classes;

      /**
       * Mapping for register-allocated objects in *regs to the first
       * GRF for that object.
      */
      uint8_t *ra_reg_to_grf;

      /**
       * ra class for the aligned pairs we use for PLN, which doesn't
       * appear in *classes.
       */
      int aligned_pairs_class;

      /** @} */
   } wm;


   struct {
      uint32_t state_offset;
      uint32_t blend_state_offset;
      uint32_t depth_stencil_state_offset;
      uint32_t vp_offset;
   } cc;

   struct {
      struct brw_query_object *obj;
      drm_intel_bo *bo;
      int index;
      bool active;
   } query;
   /* Used to give every program string a unique id
    */
   GLuint program_id;

   int num_atoms;
   const struct brw_tracked_state **atoms;

   /* If (INTEL_DEBUG & DEBUG_BATCH) */
   struct {
      uint32_t offset;
      uint32_t size;
      enum state_struct_type type;
   } *state_batch_list;
   int state_batch_count;

   struct brw_sol_state {
      uint32_t svbi_0_starting_index;
      uint32_t svbi_0_max_index;
      uint32_t offset_0_batch_start;
      uint32_t primitives_generated;
      uint32_t primitives_written;
   } sol;

   uint32_t render_target_format[MESA_FORMAT_COUNT];
   bool format_supported_as_render_target[MESA_FORMAT_COUNT];

   /* PrimitiveRestart */
   struct {
      bool in_progress;
      bool enable_cut_index;
   } prim_restart;
};



#define BRW_PACKCOLOR8888(r,g,b,a)  ((r<<24) | (g<<16) | (b<<8) | a)

struct brw_instruction_info {
    char    *name;
    int	    nsrc;
    int	    ndst;
    bool is_arith;
};
extern const struct brw_instruction_info brw_opcodes[128];

/*======================================================================
 * brw_vtbl.c
 */
void brwInitVtbl( struct brw_context *brw );

/*======================================================================
 * brw_context.c
 */
bool brwCreateContext(int api,
		      const struct gl_config *mesaVis,
		      __DRIcontext *driContextPriv,
		      void *sharedContextPrivate);

/*======================================================================
 * brw_queryobj.c
 */
void brw_init_queryobj_functions(struct dd_function_table *functions);
void brw_prepare_query_begin(struct brw_context *brw);
void brw_emit_query_begin(struct brw_context *brw);
void brw_emit_query_end(struct brw_context *brw);

/*======================================================================
 * brw_state_dump.c
 */
void brw_debug_batch(struct intel_context *intel);

/*======================================================================
 * brw_tex.c
 */
void brw_validate_textures( struct brw_context *brw );


/*======================================================================
 * brw_program.c
 */
void brwInitFragProgFuncs( struct dd_function_table *functions );

int brw_get_scratch_size(int size);
void brw_get_scratch_bo(struct intel_context *intel,
			drm_intel_bo **scratch_bo, int size);


/* brw_urb.c
 */
void brw_upload_urb_fence(struct brw_context *brw);

/* brw_curbe.c
 */
void brw_upload_cs_urb_state(struct brw_context *brw);

/* brw_disasm.c */
int brw_disasm (FILE *file, struct brw_instruction *inst, int gen);

/* brw_vs.c */
gl_clip_plane *brw_select_clip_planes(struct gl_context *ctx);

/* brw_wm_surface_state.c */
void brw_init_surface_formats(struct brw_context *brw);
void
brw_update_sol_surface(struct brw_context *brw,
                       struct gl_buffer_object *buffer_obj,
                       uint32_t *out_offset, unsigned num_vector_components,
                       unsigned stride_dwords, unsigned offset_dwords);

/* gen6_sol.c */
void
brw_begin_transform_feedback(struct gl_context *ctx, GLenum mode,
			     struct gl_transform_feedback_object *obj);
void
brw_end_transform_feedback(struct gl_context *ctx,
                           struct gl_transform_feedback_object *obj);

/* gen7_sol_state.c */
void
gen7_end_transform_feedback(struct gl_context *ctx,
			    struct gl_transform_feedback_object *obj);

/* brw_blorp_blit.cpp */
GLbitfield
brw_blorp_framebuffer(struct intel_context *intel,
                      GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
                      GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
                      GLbitfield mask, GLenum filter);

/* gen6_multisample_state.c */
void
gen6_emit_3dstate_multisample(struct brw_context *brw,
                              unsigned num_samples);
void
gen6_emit_3dstate_sample_mask(struct brw_context *brw,
                              unsigned num_samples);



/*======================================================================
 * Inline conversion functions.  These are better-typed than the
 * macros used previously:
 */
static INLINE struct brw_context *
brw_context( struct gl_context *ctx )
{
   return (struct brw_context *)ctx;
}

static INLINE struct brw_vertex_program *
brw_vertex_program(struct gl_vertex_program *p)
{
   return (struct brw_vertex_program *) p;
}

static INLINE const struct brw_vertex_program *
brw_vertex_program_const(const struct gl_vertex_program *p)
{
   return (const struct brw_vertex_program *) p;
}

static INLINE struct brw_fragment_program *
brw_fragment_program(struct gl_fragment_program *p)
{
   return (struct brw_fragment_program *) p;
}

static INLINE const struct brw_fragment_program *
brw_fragment_program_const(const struct gl_fragment_program *p)
{
   return (const struct brw_fragment_program *) p;
}

static inline
float convert_param(enum param_conversion conversion, const float *param)
{
   union {
      float f;
      uint32_t u;
      int32_t i;
   } fi;

   switch (conversion) {
   case PARAM_NO_CONVERT:
      return *param;
   case PARAM_CONVERT_F2I:
      fi.i = *param;
      return fi.f;
   case PARAM_CONVERT_F2U:
      fi.u = *param;
      return fi.f;
   case PARAM_CONVERT_F2B:
      if (*param != 0.0)
	 fi.i = 1;
      else
	 fi.i = 0;
      return fi.f;
   case PARAM_CONVERT_ZERO:
      return 0.0;
   default:
      return *param;
   }
}

/**
 * Pre-gen6, the register file of the EUs was shared between threads,
 * and each thread used some subset allocated on a 16-register block
 * granularity.  The unit states wanted these block counts.
 */
static inline int
brw_register_blocks(int reg_count)
{
   return ALIGN(reg_count, 16) / 16 - 1;
}

static inline uint32_t
brw_program_reloc(struct brw_context *brw, uint32_t state_offset,
		  uint32_t prog_offset)
{
   struct intel_context *intel = &brw->intel;

   if (intel->gen >= 5) {
      /* Using state base address. */
      return prog_offset;
   }

   drm_intel_bo_emit_reloc(intel->batch.bo,
			   state_offset,
			   brw->cache.bo,
			   prog_offset,
			   I915_GEM_DOMAIN_INSTRUCTION, 0);

   return brw->cache.bo->offset + prog_offset;
}

bool brw_do_cubemap_normalize(struct exec_list *instructions);

#ifdef __cplusplus
}
#endif

#endif