-
-
Notifications
You must be signed in to change notification settings - Fork 37
/
mojoshader.c
4006 lines (3403 loc) · 134 KB
/
mojoshader.c
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
/**
* MojoShader; generate shader programs from bytecode of compiled
* Direct3D shaders.
*
* Please see the file LICENSE.txt in the source's root directory.
*
* This file written by Ryan C. Gordon.
*/
// !!! FIXME: this file really needs to be split up.
// !!! FIXME: I keep changing coding styles for symbols and typedefs.
// !!! FIXME: rules from MSDN about temp registers we probably don't check.
// - There are limited temporaries: vs_1_1 has 12 (ps_1_1 has _2_!).
// - SM2 apparently was variable, between 12 and 32. Shader Model 3 has 32.
// - A maximum of three temp registers can be used in a single instruction.
#define __MOJOSHADER_INTERNAL__ 1
#include "profiles/mojoshader_profile.h"
// Deal with register lists... !!! FIXME: I sort of hate this.
static void free_reglist(MOJOSHADER_free f, void *d, RegisterList *item)
{
while (item != NULL)
{
RegisterList *next = item->next;
f(item, d);
item = next;
} // while
} // free_reglist
static inline const RegisterList *reglist_exists(RegisterList *prev,
const RegisterType regtype,
const int regnum)
{
return (reglist_find(prev, regtype, regnum));
} // reglist_exists
static inline int register_was_written(Context *ctx, const RegisterType rtype,
const int regnum)
{
RegisterList *reg = reglist_find(&ctx->used_registers, rtype, regnum);
return (reg && reg->written);
} // register_was_written
static inline int get_defined_register(Context *ctx, const RegisterType rtype,
const int regnum)
{
return (reglist_exists(&ctx->defined_registers, rtype, regnum) != NULL);
} // get_defined_register
static void add_attribute_register(Context *ctx, const RegisterType rtype,
const int regnum, const MOJOSHADER_usage usage,
const int index, const int writemask, int flags)
{
RegisterList *item = reglist_insert(ctx, &ctx->attributes, rtype, regnum);
item->usage = usage;
item->index = index;
item->writemask = writemask;
item->misc = flags;
if ((rtype == REG_TYPE_OUTPUT) && (usage == MOJOSHADER_USAGE_POINTSIZE))
ctx->uses_pointsize = 1; // note that we have to check this later.
else if ((rtype == REG_TYPE_OUTPUT) && (usage == MOJOSHADER_USAGE_FOG))
ctx->uses_fog = 1; // note that we have to check this later.
} // add_attribute_register
static inline TextureType cvtMojoToD3DSamplerType(const MOJOSHADER_samplerType type)
{
return (TextureType) (((int) type) + 2);
} // cvtMojoToD3DSamplerType
static inline MOJOSHADER_samplerType cvtD3DToMojoSamplerType(const TextureType type)
{
return (MOJOSHADER_samplerType) (((int) type) - 2);
} // cvtD3DToMojoSamplerType
static inline void add_sampler(Context *ctx, const int regnum,
TextureType ttype, const int texbem)
{
const RegisterType rtype = REG_TYPE_SAMPLER;
// !!! FIXME: make sure it doesn't exist?
// !!! FIXME: (ps_1_1 assume we can add it multiple times...)
RegisterList *item = reglist_insert(ctx, &ctx->samplers, rtype, regnum);
if (ctx->samplermap != NULL)
{
unsigned int i;
for (i = 0; i < ctx->samplermap_count; i++)
{
if (ctx->samplermap[i].index == regnum)
{
ttype = cvtMojoToD3DSamplerType(ctx->samplermap[i].type);
break;
} // if
} // for
} // if
item->index = (int) ttype;
item->misc |= texbem;
} // add_sampler
static inline void adjust_token_position(Context *ctx, const int incr)
{
ctx->tokens += incr;
ctx->tokencount -= incr;
ctx->current_position += incr * sizeof (uint32);
} // adjust_token_position
// Generate emitter declarations for each profile with this macro...
#define PREDECLARE_PROFILE(prof) \
void emit_##prof##_start(Context *ctx, const char *profilestr); \
void emit_##prof##_end(Context *ctx); \
void emit_##prof##_phase(Context *ctx); \
void emit_##prof##_finalize(Context *ctx); \
void emit_##prof##_global(Context *ctx, RegisterType regtype, int regnum);\
void emit_##prof##_array(Context *ctx, VariableList *var); \
void emit_##prof##_const_array(Context *ctx, const ConstantsList *clist, \
int base, int size); \
void emit_##prof##_uniform(Context *ctx, RegisterType regtype, int regnum,\
const VariableList *var); \
void emit_##prof##_sampler(Context *ctx, int stage, TextureType ttype, \
int tb); \
void emit_##prof##_attribute(Context *ctx, RegisterType regtype, \
int regnum, MOJOSHADER_usage usage, \
int index, int wmask, int flags); \
void emit_##prof##_NOP(Context *ctx); \
void emit_##prof##_MOV(Context *ctx); \
void emit_##prof##_ADD(Context *ctx); \
void emit_##prof##_SUB(Context *ctx); \
void emit_##prof##_MAD(Context *ctx); \
void emit_##prof##_MUL(Context *ctx); \
void emit_##prof##_RCP(Context *ctx); \
void emit_##prof##_RSQ(Context *ctx); \
void emit_##prof##_DP3(Context *ctx); \
void emit_##prof##_DP4(Context *ctx); \
void emit_##prof##_MIN(Context *ctx); \
void emit_##prof##_MAX(Context *ctx); \
void emit_##prof##_SLT(Context *ctx); \
void emit_##prof##_SGE(Context *ctx); \
void emit_##prof##_EXP(Context *ctx); \
void emit_##prof##_LOG(Context *ctx); \
void emit_##prof##_LIT(Context *ctx); \
void emit_##prof##_DST(Context *ctx); \
void emit_##prof##_LRP(Context *ctx); \
void emit_##prof##_FRC(Context *ctx); \
void emit_##prof##_M4X4(Context *ctx); \
void emit_##prof##_M4X3(Context *ctx); \
void emit_##prof##_M3X4(Context *ctx); \
void emit_##prof##_M3X3(Context *ctx); \
void emit_##prof##_M3X2(Context *ctx); \
void emit_##prof##_CALL(Context *ctx); \
void emit_##prof##_CALLNZ(Context *ctx); \
void emit_##prof##_LOOP(Context *ctx); \
void emit_##prof##_ENDLOOP(Context *ctx); \
void emit_##prof##_LABEL(Context *ctx); \
void emit_##prof##_DCL(Context *ctx); \
void emit_##prof##_POW(Context *ctx); \
void emit_##prof##_CRS(Context *ctx); \
void emit_##prof##_SGN(Context *ctx); \
void emit_##prof##_ABS(Context *ctx); \
void emit_##prof##_NRM(Context *ctx); \
void emit_##prof##_SINCOS(Context *ctx); \
void emit_##prof##_REP(Context *ctx); \
void emit_##prof##_ENDREP(Context *ctx); \
void emit_##prof##_IF(Context *ctx); \
void emit_##prof##_IFC(Context *ctx); \
void emit_##prof##_ELSE(Context *ctx); \
void emit_##prof##_ENDIF(Context *ctx); \
void emit_##prof##_BREAK(Context *ctx); \
void emit_##prof##_BREAKC(Context *ctx); \
void emit_##prof##_MOVA(Context *ctx); \
void emit_##prof##_DEFB(Context *ctx); \
void emit_##prof##_DEFI(Context *ctx); \
void emit_##prof##_TEXCRD(Context *ctx); \
void emit_##prof##_TEXKILL(Context *ctx); \
void emit_##prof##_TEXLD(Context *ctx); \
void emit_##prof##_TEXBEM(Context *ctx); \
void emit_##prof##_TEXBEML(Context *ctx); \
void emit_##prof##_TEXREG2AR(Context *ctx); \
void emit_##prof##_TEXREG2GB(Context *ctx); \
void emit_##prof##_TEXM3X2PAD(Context *ctx); \
void emit_##prof##_TEXM3X2TEX(Context *ctx); \
void emit_##prof##_TEXM3X3PAD(Context *ctx); \
void emit_##prof##_TEXM3X3TEX(Context *ctx); \
void emit_##prof##_TEXM3X3SPEC(Context *ctx); \
void emit_##prof##_TEXM3X3VSPEC(Context *ctx); \
void emit_##prof##_EXPP(Context *ctx); \
void emit_##prof##_LOGP(Context *ctx); \
void emit_##prof##_CND(Context *ctx); \
void emit_##prof##_DEF(Context *ctx); \
void emit_##prof##_TEXREG2RGB(Context *ctx); \
void emit_##prof##_TEXDP3TEX(Context *ctx); \
void emit_##prof##_TEXM3X2DEPTH(Context *ctx); \
void emit_##prof##_TEXDP3(Context *ctx); \
void emit_##prof##_TEXM3X3(Context *ctx); \
void emit_##prof##_TEXDEPTH(Context *ctx); \
void emit_##prof##_CMP(Context *ctx); \
void emit_##prof##_BEM(Context *ctx); \
void emit_##prof##_DP2ADD(Context *ctx); \
void emit_##prof##_DSX(Context *ctx); \
void emit_##prof##_DSY(Context *ctx); \
void emit_##prof##_TEXLDD(Context *ctx); \
void emit_##prof##_SETP(Context *ctx); \
void emit_##prof##_TEXLDL(Context *ctx); \
void emit_##prof##_BREAKP(Context *ctx); \
void emit_##prof##_RESERVED(Context *ctx); \
void emit_##prof##_RET(Context *ctx); \
const char *get_##prof##_varname(Context *ctx, RegisterType rt, \
int regnum); \
const char *get_##prof##_const_array_varname(Context *ctx, \
int base, int size);
// Check for profile support...
#define AT_LEAST_ONE_PROFILE 0
#if !SUPPORT_PROFILE_BYTECODE
#define PROFILE_EMITTER_BYTECODE(op)
#else
#undef AT_LEAST_ONE_PROFILE
#define AT_LEAST_ONE_PROFILE 1
#define PROFILE_EMITTER_BYTECODE(op) emit_BYTECODE_##op,
PREDECLARE_PROFILE(BYTECODE)
#endif
#if !SUPPORT_PROFILE_D3D
#define PROFILE_EMITTER_D3D(op)
#else
#undef AT_LEAST_ONE_PROFILE
#define AT_LEAST_ONE_PROFILE 1
#define PROFILE_EMITTER_D3D(op) emit_D3D_##op,
PREDECLARE_PROFILE(D3D)
#endif
#if !SUPPORT_PROFILE_HLSL
#define PROFILE_EMITTER_HLSL(op)
#else
#undef AT_LEAST_ONE_PROFILE
#define AT_LEAST_ONE_PROFILE 1
#define PROFILE_EMITTER_HLSL(op) emit_HLSL_##op,
PREDECLARE_PROFILE(HLSL)
#endif
#if !SUPPORT_PROFILE_GLSL
#define PROFILE_EMITTER_GLSL(op)
#else
#undef AT_LEAST_ONE_PROFILE
#define AT_LEAST_ONE_PROFILE 1
#define PROFILE_EMITTER_GLSL(op) emit_GLSL_##op,
PREDECLARE_PROFILE(GLSL)
#endif
#if !SUPPORT_PROFILE_METAL
#define PROFILE_EMITTER_METAL(op)
#else
#undef AT_LEAST_ONE_PROFILE
#define AT_LEAST_ONE_PROFILE 1
#define PROFILE_EMITTER_METAL(op) emit_METAL_##op,
PREDECLARE_PROFILE(METAL)
#endif
#if !SUPPORT_PROFILE_ARB1
#define PROFILE_EMITTER_ARB1(op)
#else
#undef AT_LEAST_ONE_PROFILE
#define AT_LEAST_ONE_PROFILE 1
#define PROFILE_EMITTER_ARB1(op) emit_ARB1_##op,
PREDECLARE_PROFILE(ARB1)
#endif
#if !SUPPORT_PROFILE_SPIRV
#define PROFILE_EMITTER_SPIRV(op)
#else
#undef AT_LEAST_ONE_PROFILE
#define AT_LEAST_ONE_PROFILE 1
#define PROFILE_EMITTER_SPIRV(op) emit_SPIRV_##op,
PREDECLARE_PROFILE(SPIRV)
#endif
#if !AT_LEAST_ONE_PROFILE
#error No profiles are supported. Fix your build.
#endif
#define DEFINE_PROFILE(prof) { \
MOJOSHADER_PROFILE_##prof, \
emit_##prof##_start, \
emit_##prof##_end, \
emit_##prof##_phase, \
emit_##prof##_global, \
emit_##prof##_array, \
emit_##prof##_const_array, \
emit_##prof##_uniform, \
emit_##prof##_sampler, \
emit_##prof##_attribute, \
emit_##prof##_finalize, \
get_##prof##_varname, \
get_##prof##_const_array_varname, \
},
static const Profile profiles[] =
{
#if SUPPORT_PROFILE_D3D
DEFINE_PROFILE(D3D)
#endif
#if SUPPORT_PROFILE_BYTECODE
DEFINE_PROFILE(BYTECODE)
#endif
#if SUPPORT_PROFILE_HLSL
DEFINE_PROFILE(HLSL)
#endif
#if SUPPORT_PROFILE_GLSL
DEFINE_PROFILE(GLSL)
#endif
#if SUPPORT_PROFILE_ARB1
DEFINE_PROFILE(ARB1)
#endif
#if SUPPORT_PROFILE_METAL
DEFINE_PROFILE(METAL)
#endif
#if SUPPORT_PROFILE_SPIRV
DEFINE_PROFILE(SPIRV)
#endif
};
#undef DEFINE_PROFILE
// This is for profiles that extend other profiles...
static const struct { const char *from; const char *to; } profileMap[] =
{
{ MOJOSHADER_PROFILE_GLSPIRV, MOJOSHADER_PROFILE_SPIRV },
{ MOJOSHADER_PROFILE_GLSLES, MOJOSHADER_PROFILE_GLSL },
{ MOJOSHADER_PROFILE_GLSLES3, MOJOSHADER_PROFILE_GLSL },
{ MOJOSHADER_PROFILE_GLSL120, MOJOSHADER_PROFILE_GLSL },
{ MOJOSHADER_PROFILE_NV2, MOJOSHADER_PROFILE_ARB1 },
{ MOJOSHADER_PROFILE_NV3, MOJOSHADER_PROFILE_ARB1 },
{ MOJOSHADER_PROFILE_NV4, MOJOSHADER_PROFILE_ARB1 },
};
// The PROFILE_EMITTER_* items MUST be in the same order as profiles[]!
#define PROFILE_EMITTERS(op) { \
PROFILE_EMITTER_D3D(op) \
PROFILE_EMITTER_BYTECODE(op) \
PROFILE_EMITTER_HLSL(op) \
PROFILE_EMITTER_GLSL(op) \
PROFILE_EMITTER_ARB1(op) \
PROFILE_EMITTER_METAL(op) \
PROFILE_EMITTER_SPIRV(op) \
}
static int parse_destination_token(Context *ctx, DestArgInfo *info)
{
// !!! FIXME: recheck against the spec for ranges (like RASTOUT values, etc).
if (ctx->tokencount == 0)
{
fail(ctx, "Out of tokens in destination parameter");
return 0;
} // if
const uint32 token = SWAP32(*(ctx->tokens));
const int reserved1 = (int) ((token >> 14) & 0x3); // bits 14 through 15
const int reserved2 = (int) ((token >> 31) & 0x1); // bit 31
info->token = ctx->tokens;
info->regnum = (int) (token & 0x7ff); // bits 0 through 10
info->relative = (int) ((token >> 13) & 0x1); // bit 13
info->orig_writemask = (int) ((token >> 16) & 0xF); // bits 16 through 19
info->result_mod = (int) ((token >> 20) & 0xF); // bits 20 through 23
info->result_shift = (int) ((token >> 24) & 0xF); // bits 24 through 27 abc
info->regtype = (RegisterType) (((token >> 28) & 0x7) | ((token >> 8) & 0x18)); // bits 28-30, 11-12
int writemask;
if (isscalar(ctx, ctx->shader_type, info->regtype, info->regnum))
writemask = 0x1; // just x.
else
writemask = info->orig_writemask;
set_dstarg_writemask(info, writemask); // bits 16 through 19.
// all the REG_TYPE_CONSTx types are the same register type, it's just
// split up so its regnum can be > 2047 in the bytecode. Clean it up.
if (info->regtype == REG_TYPE_CONST2)
{
info->regtype = REG_TYPE_CONST;
info->regnum += 2048;
} // else if
else if (info->regtype == REG_TYPE_CONST3)
{
info->regtype = REG_TYPE_CONST;
info->regnum += 4096;
} // else if
else if (info->regtype == REG_TYPE_CONST4)
{
info->regtype = REG_TYPE_CONST;
info->regnum += 6144;
} // else if
// swallow token for now, for multiple calls in a row.
adjust_token_position(ctx, 1);
if (reserved1 != 0x0)
fail(ctx, "Reserved bit #1 in destination token must be zero");
if (reserved2 != 0x1)
fail(ctx, "Reserved bit #2 in destination token must be one");
if (info->relative)
{
if (!shader_is_vertex(ctx))
fail(ctx, "Relative addressing in non-vertex shader");
if (!shader_version_atleast(ctx, 3, 0))
fail(ctx, "Relative addressing in vertex shader version < 3.0");
if ((!ctx->ctab.have_ctab) && (!ctx->ignores_ctab))
{
// it's hard to do this efficiently without!
fail(ctx, "relative addressing unsupported without a CTAB");
} // if
// !!! FIXME: I don't have a shader that has a relative dest currently.
fail(ctx, "Relative addressing of dest tokens is unsupported");
return 2;
} // if
const int s = info->result_shift;
if (s != 0)
{
if (!shader_is_pixel(ctx))
fail(ctx, "Result shift scale in non-pixel shader");
if (shader_version_atleast(ctx, 2, 0))
fail(ctx, "Result shift scale in pixel shader version >= 2.0");
if ( ! (((s >= 1) && (s <= 3)) || ((s >= 0xD) && (s <= 0xF))) )
fail(ctx, "Result shift scale isn't 1 to 3, or 13 to 15.");
} // if
if (info->result_mod & MOD_PP) // Partial precision (pixel shaders only)
{
if (!shader_is_pixel(ctx))
fail(ctx, "Partial precision result mod in non-pixel shader");
} // if
if (info->result_mod & MOD_CENTROID) // Centroid (pixel shaders only)
{
if (!shader_is_pixel(ctx))
fail(ctx, "Centroid result mod in non-pixel shader");
else if (!ctx->centroid_allowed) // only on DCL opcodes!
fail(ctx, "Centroid modifier not allowed here");
} // if
if (/*(info->regtype < 0) ||*/ (info->regtype > REG_TYPE_MAX))
fail(ctx, "Register type is out of range");
if (!isfail(ctx))
set_used_register(ctx, info->regtype, info->regnum, 1);
return 1;
} // parse_destination_token
static void determine_constants_arrays(Context *ctx)
{
// Only process this stuff once. This is called after all DEF* opcodes
// could have been parsed.
if (ctx->determined_constants_arrays)
return;
ctx->determined_constants_arrays = 1;
if (ctx->constant_count <= 1)
return; // nothing to sort or group.
// Sort the linked list into an array for easier tapdancing...
ConstantsList **array = (ConstantsList **) alloca(sizeof (ConstantsList *) * (ctx->constant_count + 1));
ConstantsList *item = ctx->constants;
int i;
for (i = 0; i < ctx->constant_count; i++)
{
if (item == NULL)
{
fail(ctx, "BUG: mismatched constant list and count");
return;
} // if
array[i] = item;
item = item->next;
} // for
array[ctx->constant_count] = NULL;
// bubble sort ftw.
int sorted;
do
{
sorted = 1;
for (i = 0; i < ctx->constant_count-1; i++)
{
if (array[i]->constant.index > array[i+1]->constant.index)
{
ConstantsList *tmp = array[i];
array[i] = array[i+1];
array[i+1] = tmp;
sorted = 0;
} // if
} // for
} while (!sorted);
// okay, sorted. While we're here, let's redo the linked list in order...
for (i = 0; i < ctx->constant_count; i++)
array[i]->next = array[i+1];
ctx->constants = array[0];
// now figure out the groupings of constants and add to ctx->variables...
int start = -1;
int prev = -1;
int count = 0;
const int hi = ctx->constant_count;
for (i = 0; i <= hi; i++)
{
if (array[i] && (array[i]->constant.type != MOJOSHADER_UNIFORM_FLOAT))
continue; // we only care about REG_TYPE_CONST for array groups.
if (start == -1)
{
prev = start = i; // first REG_TYPE_CONST we've seen. Mark it!
continue;
} // if
// not a match (or last item in the array)...see if we had a
// contiguous set before this point...
if ( (array[i]) && (array[i]->constant.index == (array[prev]->constant.index + 1)) )
count++;
else
{
if (count > 0) // multiple constants in the set?
{
VariableList *var;
var = (VariableList *) Malloc(ctx, sizeof (VariableList));
if (var == NULL)
break;
var->type = MOJOSHADER_UNIFORM_FLOAT;
var->index = array[start]->constant.index;
var->count = (array[prev]->constant.index - var->index) + 1;
var->constant = array[start];
var->used = 0;
var->emit_position = -1;
var->next = ctx->variables;
ctx->variables = var;
} // if
start = i; // set this as new start of sequence.
} // if
prev = i;
} // for
} // determine_constants_arrays
static void shader_model_1_input_usage(const int regnum, MOJOSHADER_usage *usage, int *index)
{
*index = 0;
switch (regnum) // these are hardcoded for Shader Model 1: v0 is POSITION, v1 is BLENDWEIGHT, etc.
{
case 0: *usage = MOJOSHADER_USAGE_POSITION; break;
case 1: *usage = MOJOSHADER_USAGE_BLENDWEIGHT; break;
case 2: *usage = MOJOSHADER_USAGE_BLENDINDICES; break;
case 3: *usage = MOJOSHADER_USAGE_NORMAL; break;
case 4: *usage = MOJOSHADER_USAGE_POINTSIZE; break;
case 5: *usage = MOJOSHADER_USAGE_COLOR; break; // diffuse
case 6: *usage = MOJOSHADER_USAGE_COLOR; *index = 1; break; // specular
case 7: *usage = MOJOSHADER_USAGE_TEXCOORD; break;
case 8: *usage = MOJOSHADER_USAGE_TEXCOORD; *index = 1; break;
case 9: *usage = MOJOSHADER_USAGE_TEXCOORD; *index = 2; break;
case 10: *usage = MOJOSHADER_USAGE_TEXCOORD; *index = 3; break;
case 11: *usage = MOJOSHADER_USAGE_TEXCOORD; *index = 4; break;
case 12: *usage = MOJOSHADER_USAGE_TEXCOORD; *index = 5; break;
case 13: *usage = MOJOSHADER_USAGE_TEXCOORD; *index = 6; break;
case 14: *usage = MOJOSHADER_USAGE_TEXCOORD; *index = 7; break;
case 15: *usage = MOJOSHADER_USAGE_POSITION; *index = 1; break;
case 16: *usage = MOJOSHADER_USAGE_NORMAL; *index = 1; break;
default: *usage = MOJOSHADER_USAGE_UNKNOWN; break;
} // switch
} // shader_model_1_input_usage
static int adjust_swizzle(const Context *ctx, const RegisterType regtype,
const int regnum, const int swizzle)
{
if (regtype != REG_TYPE_INPUT) // !!! FIXME: maybe lift this later?
return swizzle;
else if (ctx->swizzles_count == 0)
return swizzle;
MOJOSHADER_usage usage = MOJOSHADER_USAGE_UNKNOWN;
int index = 0;
// Shader Model 1 didn't have to predeclare attribute variables, so
// reglist_find won't be able to look them up at this point, but
// their usages don't change.
if (!shader_version_atleast(ctx, 2, 0))
shader_model_1_input_usage(regnum, &usage, &index);
else
{
const RegisterList *reg = reglist_find(&ctx->attributes, regtype, regnum);
if (reg == NULL)
return swizzle;
usage = reg->usage;
index = reg->index;
} // else
if (usage == MOJOSHADER_USAGE_UNKNOWN)
return swizzle;
size_t i;
for (i = 0; i < ctx->swizzles_count; i++)
{
const MOJOSHADER_swizzle *swiz = &ctx->swizzles[i];
if ((swiz->usage == usage) && (swiz->index == index))
{
return ( (((int)(swiz->swizzles[((swizzle >> 0) & 0x3)])) << 0) |
(((int)(swiz->swizzles[((swizzle >> 2) & 0x3)])) << 2) |
(((int)(swiz->swizzles[((swizzle >> 4) & 0x3)])) << 4) |
(((int)(swiz->swizzles[((swizzle >> 6) & 0x3)])) << 6) );
} // if
} // for
return swizzle;
} // adjust_swizzle
static int parse_source_token(Context *ctx, SourceArgInfo *info)
{
int retval = 1;
if (ctx->tokencount == 0)
{
fail(ctx, "Out of tokens in source parameter");
return 0;
} // if
const uint32 token = SWAP32(*(ctx->tokens));
const int reserved1 = (int) ((token >> 14) & 0x3); // bits 14 through 15
const int reserved2 = (int) ((token >> 31) & 0x1); // bit 31
info->token = ctx->tokens;
info->regnum = (int) (token & 0x7ff); // bits 0 through 10
info->relative = (int) ((token >> 13) & 0x1); // bit 13
const int swizzle = (int) ((token >> 16) & 0xFF); // bits 16 through 23
info->src_mod = (SourceMod) ((token >> 24) & 0xF); // bits 24 through 27
info->regtype = (RegisterType) (((token >> 28) & 0x7) | ((token >> 8) & 0x18)); // bits 28-30, 11-12
// all the REG_TYPE_CONSTx types are the same register type, it's just
// split up so its regnum can be > 2047 in the bytecode. Clean it up.
if (info->regtype == REG_TYPE_CONST2)
{
info->regtype = REG_TYPE_CONST;
info->regnum += 2048;
} // else if
else if (info->regtype == REG_TYPE_CONST3)
{
info->regtype = REG_TYPE_CONST;
info->regnum += 4096;
} // else if
else if (info->regtype == REG_TYPE_CONST4)
{
info->regtype = REG_TYPE_CONST;
info->regnum += 6144;
} // else if
info->swizzle = adjust_swizzle(ctx, info->regtype, info->regnum, swizzle);
info->swizzle_x = ((info->swizzle >> 0) & 0x3);
info->swizzle_y = ((info->swizzle >> 2) & 0x3);
info->swizzle_z = ((info->swizzle >> 4) & 0x3);
info->swizzle_w = ((info->swizzle >> 6) & 0x3);
// swallow token for now, for multiple calls in a row.
adjust_token_position(ctx, 1);
if (reserved1 != 0x0)
fail(ctx, "Reserved bits #1 in source token must be zero");
if (reserved2 != 0x1)
fail(ctx, "Reserved bit #2 in source token must be one");
if ((info->relative) && (ctx->tokencount == 0))
{
fail(ctx, "Out of tokens in relative source parameter");
info->relative = 0; // don't try to process it.
} // if
if (info->relative)
{
if ( (shader_is_pixel(ctx)) && (!shader_version_atleast(ctx, 3, 0)) )
fail(ctx, "Relative addressing in pixel shader version < 3.0");
// Shader Model 1 doesn't have an extra token to specify the
// relative register: it's always a0.x.
if (!shader_version_atleast(ctx, 2, 0))
{
info->relative_regnum = 0;
info->relative_regtype = REG_TYPE_ADDRESS;
info->relative_component = 0;
} // if
else // Shader Model 2 and later...
{
const uint32 reltoken = SWAP32(*(ctx->tokens));
// swallow token for now, for multiple calls in a row.
adjust_token_position(ctx, 1);
const int relswiz = (int) ((reltoken >> 16) & 0xFF);
info->relative_regnum = (int) (reltoken & 0x7ff);
info->relative_regtype = (RegisterType)
(((reltoken >> 28) & 0x7) |
((reltoken >> 8) & 0x18));
if (((reltoken >> 31) & 0x1) == 0)
fail(ctx, "bit #31 in relative address must be set");
if ((reltoken & 0xF00E000) != 0) // usused bits.
fail(ctx, "relative address reserved bit must be zero");
switch (info->relative_regtype)
{
case REG_TYPE_LOOP:
case REG_TYPE_ADDRESS:
break;
default:
fail(ctx, "invalid register for relative address");
break;
} // switch
if (info->relative_regnum != 0) // true for now.
fail(ctx, "invalid register for relative address");
if ( (info->relative_regtype != REG_TYPE_LOOP) && !replicate_swizzle(relswiz) )
fail(ctx, "relative address needs replicate swizzle");
info->relative_component = (relswiz & 0x3);
retval++;
} // else
if (info->regtype == REG_TYPE_INPUT)
{
if ( (shader_is_pixel(ctx)) || (!shader_version_atleast(ctx, 3, 0)) )
fail(ctx, "relative addressing of input registers not supported in this shader model");
ctx->have_relative_input_registers = 1;
} // if
else if (info->regtype == REG_TYPE_CONST)
{
// figure out what array we're in...
if (!ctx->ignores_ctab)
{
if (!ctx->ctab.have_ctab) // hard to do efficiently without!
fail(ctx, "relative addressing unsupported without a CTAB");
else
{
determine_constants_arrays(ctx);
VariableList *var;
const int reltarget = info->regnum;
for (var = ctx->variables; var != NULL; var = var->next)
{
const int lo = var->index;
if ( (reltarget >= lo) && (reltarget < (lo + var->count)) )
break; // match!
} // for
if (var == NULL)
fail(ctx, "relative addressing of indeterminate array");
else
{
var->used = 1;
info->relative_array = var;
set_used_register(ctx, info->relative_regtype, info->relative_regnum, 0);
} // else
} // else
} // if
} // else if
else
{
fail(ctx, "relative addressing of invalid register");
} // else
} // if
switch (info->src_mod)
{
case SRCMOD_NONE:
case SRCMOD_ABSNEGATE:
case SRCMOD_ABS:
case SRCMOD_NEGATE:
break; // okay in any shader model.
// apparently these are only legal in Shader Model 1.x ...
case SRCMOD_BIASNEGATE:
case SRCMOD_BIAS:
case SRCMOD_SIGNNEGATE:
case SRCMOD_SIGN:
case SRCMOD_COMPLEMENT:
case SRCMOD_X2NEGATE:
case SRCMOD_X2:
case SRCMOD_DZ:
case SRCMOD_DW:
if (shader_version_atleast(ctx, 2, 0))
fail(ctx, "illegal source mod for this Shader Model.");
break;
case SRCMOD_NOT: // !!! FIXME: I _think_ this is right...
if (shader_version_atleast(ctx, 2, 0))
{
if (info->regtype != REG_TYPE_PREDICATE
&& info->regtype != REG_TYPE_CONSTBOOL)
fail(ctx, "NOT only allowed on bool registers.");
} // if
break;
default:
fail(ctx, "Unknown source modifier");
} // switch
// !!! FIXME: docs say this for sm3 ... check these!
// "The negate modifier cannot be used on second source register of these
// instructions: m3x2 - ps, m3x3 - ps, m3x4 - ps, m4x3 - ps, and
// m4x4 - ps."
// "If any version 3 shader reads from one or more constant float
// registers (c#), one of the following must be true.
// All of the constant floating-point registers must use the abs modifier.
// None of the constant floating-point registers can use the abs modifier.
if (!isfail(ctx))
{
RegisterList *reg;
reg = set_used_register(ctx, info->regtype, info->regnum, 0);
// !!! FIXME: this test passes if you write to the register
// !!! FIXME: in this same instruction, because we parse the
// !!! FIXME: destination token first.
// !!! FIXME: Microsoft's shader validation explicitly checks temp
// !!! FIXME: registers for this...do they check other writable ones?
if ((info->regtype == REG_TYPE_TEMP) && (reg) && (!reg->written))
failf(ctx, "Temp register r%d used uninitialized", info->regnum);
} // if
return retval;
} // parse_source_token
static int parse_predicated_token(Context *ctx)
{
SourceArgInfo *arg = &ctx->predicate_arg;
parse_source_token(ctx, arg);
if (arg->regtype != REG_TYPE_PREDICATE)
fail(ctx, "Predicated instruction but not predicate register!");
if ((arg->src_mod != SRCMOD_NONE) && (arg->src_mod != SRCMOD_NOT))
fail(ctx, "Predicated instruction register is not NONE or NOT");
if ( !no_swizzle(arg->swizzle) && !replicate_swizzle(arg->swizzle) )
fail(ctx, "Predicated instruction register has wrong swizzle");
if (arg->relative) // I'm pretty sure this is illegal...?
fail(ctx, "relative addressing in predicated token");
return 1;
} // parse_predicated_token
static int parse_args_NULL(Context *ctx)
{
return 1;
} // parse_args_NULL
static int parse_args_DEF(Context *ctx)
{
parse_destination_token(ctx, &ctx->dest_arg);
if (ctx->dest_arg.regtype != REG_TYPE_CONST)
fail(ctx, "DEF using non-CONST register");
if (ctx->dest_arg.relative) // I'm pretty sure this is illegal...?
fail(ctx, "relative addressing in DEF");
ctx->dwords[0] = SWAP32(ctx->tokens[0]);
ctx->dwords[1] = SWAP32(ctx->tokens[1]);
ctx->dwords[2] = SWAP32(ctx->tokens[2]);
ctx->dwords[3] = SWAP32(ctx->tokens[3]);
return 6;
} // parse_args_DEF
static int parse_args_DEFI(Context *ctx)
{
parse_destination_token(ctx, &ctx->dest_arg);
if (ctx->dest_arg.regtype != REG_TYPE_CONSTINT)
fail(ctx, "DEFI using non-CONSTING register");
if (ctx->dest_arg.relative) // I'm pretty sure this is illegal...?
fail(ctx, "relative addressing in DEFI");
ctx->dwords[0] = SWAP32(ctx->tokens[0]);
ctx->dwords[1] = SWAP32(ctx->tokens[1]);
ctx->dwords[2] = SWAP32(ctx->tokens[2]);
ctx->dwords[3] = SWAP32(ctx->tokens[3]);
return 6;
} // parse_args_DEFI
static int parse_args_DEFB(Context *ctx)
{
parse_destination_token(ctx, &ctx->dest_arg);
if (ctx->dest_arg.regtype != REG_TYPE_CONSTBOOL)
fail(ctx, "DEFB using non-CONSTBOOL register");
if (ctx->dest_arg.relative) // I'm pretty sure this is illegal...?
fail(ctx, "relative addressing in DEFB");
ctx->dwords[0] = *(ctx->tokens) ? 1 : 0;
return 3;
} // parse_args_DEFB
static int valid_texture_type(const uint32 ttype)
{
switch ((const TextureType) ttype)
{
case TEXTURE_TYPE_2D:
case TEXTURE_TYPE_CUBE:
case TEXTURE_TYPE_VOLUME:
return 1; // it's okay.
} // switch
return 0;
} // valid_texture_type
// !!! FIXME: this function is kind of a mess.
static int parse_args_DCL(Context *ctx)
{
int unsupported = 0;
const uint32 token = SWAP32(*(ctx->tokens));
const int reserved1 = (int) ((token >> 31) & 0x1); // bit 31
uint32 reserved_mask = 0x00000000;
if (reserved1 != 0x1)
fail(ctx, "Bit #31 in DCL token must be one");
ctx->centroid_allowed = 1;
adjust_token_position(ctx, 1);
parse_destination_token(ctx, &ctx->dest_arg);
ctx->centroid_allowed = 0;
if (ctx->dest_arg.result_shift != 0) // I'm pretty sure this is illegal...?
fail(ctx, "shift scale in DCL");
if (ctx->dest_arg.relative) // I'm pretty sure this is illegal...?
fail(ctx, "relative addressing in DCL");
const RegisterType regtype = ctx->dest_arg.regtype;
const int regnum = ctx->dest_arg.regnum;
if ( (shader_is_pixel(ctx)) && (shader_version_atleast(ctx, 3, 0)) )
{
if (regtype == REG_TYPE_INPUT)
{
const uint32 usage = (token & 0xF);
const uint32 index = ((token >> 16) & 0xF);
reserved_mask = 0x7FF0FFE0;
ctx->dwords[0] = usage;
ctx->dwords[1] = index;
} // if
else if (regtype == REG_TYPE_MISCTYPE)
{
const MiscTypeType mt = (MiscTypeType) regnum;
if (mt == MISCTYPE_TYPE_POSITION)
reserved_mask = 0x7FFFFFFF;
else if (mt == MISCTYPE_TYPE_FACE)
{
reserved_mask = 0x7FFFFFFF;
if (!writemask_xyzw(ctx->dest_arg.orig_writemask))
fail(ctx, "DCL face writemask must be full");
if (ctx->dest_arg.result_mod != 0)
fail(ctx, "DCL face result modifier must be zero");
if (ctx->dest_arg.result_shift != 0)
fail(ctx, "DCL face shift scale must be zero");
} // else if
else
{
unsupported = 1;
} // else
ctx->dwords[0] = (uint32) MOJOSHADER_USAGE_UNKNOWN;
ctx->dwords[1] = 0;
} // else if
else if (regtype == REG_TYPE_TEXTURE)
{
const uint32 usage = (token & 0xF);
const uint32 index = ((token >> 16) & 0xF);
if (usage == MOJOSHADER_USAGE_TEXCOORD)
{
if (index > 7)
fail(ctx, "DCL texcoord usage must have 0-7 index");
} // if