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
|
/*
* Copyright © 2014 Broadcom
*
* 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 AUTHORS OR COPYRIGHT HOLDERS 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.
*/
#ifndef VC4_QIR_H
#define VC4_QIR_H
#include <assert.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdbool.h>
#include <stdint.h>
#include <string.h>
#include "util/macros.h"
#include "compiler/nir/nir.h"
#include "util/list.h"
#include "util/u_math.h"
#include "vc4_screen.h"
#include "vc4_qpu_defines.h"
#include "vc4_qpu.h"
#include "kernel/vc4_packet.h"
#include "pipe/p_state.h"
struct nir_builder;
enum qfile {
QFILE_NULL,
QFILE_TEMP,
QFILE_VARY,
QFILE_UNIF,
QFILE_VPM,
QFILE_TLB_COLOR_WRITE,
QFILE_TLB_COLOR_WRITE_MS,
QFILE_TLB_Z_WRITE,
QFILE_TLB_STENCIL_SETUP,
/* Payload registers that aren't in the physical register file, so we
* can just use the corresponding qpu_reg at qpu_emit time.
*/
QFILE_FRAG_X,
QFILE_FRAG_Y,
QFILE_FRAG_REV_FLAG,
QFILE_QPU_ELEMENT,
/**
* Stores an immediate value in the index field that will be used
* directly by qpu_load_imm().
*/
QFILE_LOAD_IMM,
/**
* Stores an immediate value in the index field that can be turned
* into a small immediate field by qpu_encode_small_immediate().
*/
QFILE_SMALL_IMM,
};
struct qreg {
enum qfile file;
uint32_t index;
int pack;
};
static inline struct qreg qir_reg(enum qfile file, uint32_t index)
{
return (struct qreg){file, index};
}
enum qop {
QOP_UNDEF,
QOP_MOV,
QOP_FMOV,
QOP_MMOV,
QOP_FADD,
QOP_FSUB,
QOP_FMUL,
QOP_V8MULD,
QOP_V8MIN,
QOP_V8MAX,
QOP_V8ADDS,
QOP_V8SUBS,
QOP_MUL24,
QOP_FMIN,
QOP_FMAX,
QOP_FMINABS,
QOP_FMAXABS,
QOP_ADD,
QOP_SUB,
QOP_SHL,
QOP_SHR,
QOP_ASR,
QOP_MIN,
QOP_MAX,
QOP_AND,
QOP_OR,
QOP_XOR,
QOP_NOT,
QOP_FTOI,
QOP_ITOF,
QOP_RCP,
QOP_RSQ,
QOP_EXP2,
QOP_LOG2,
QOP_VW_SETUP,
QOP_VR_SETUP,
QOP_TLB_COLOR_READ,
QOP_MS_MASK,
QOP_VARY_ADD_C,
QOP_FRAG_Z,
QOP_FRAG_W,
/** Texture x coordinate parameter write */
QOP_TEX_S,
/** Texture y coordinate parameter write */
QOP_TEX_T,
/** Texture border color parameter or cube map z coordinate write */
QOP_TEX_R,
/** Texture LOD bias parameter write */
QOP_TEX_B,
/**
* Texture-unit 4-byte read with address provided direct in S
* cooordinate.
*
* The first operand is the offset from the start of the UBO, and the
* second is the uniform that has the UBO's base pointer.
*/
QOP_TEX_DIRECT,
/**
* Signal of texture read being necessary and then reading r4 into
* the destination
*/
QOP_TEX_RESULT,
/**
* Insert the signal for switching threads in a threaded fragment
* shader. No value can be live in an accumulator across a thrsw.
*
* At the QPU level, this will have several delay slots before the
* switch happens. Those slots are the responsibility of the
* scheduler.
*/
QOP_THRSW,
/* 32-bit immediate loaded to each SIMD channel */
QOP_LOAD_IMM,
/* 32-bit immediate divided into 16 2-bit unsigned int values and
* loaded to each corresponding SIMD channel.
*/
QOP_LOAD_IMM_U2,
/* 32-bit immediate divided into 16 2-bit signed int values and
* loaded to each corresponding SIMD channel.
*/
QOP_LOAD_IMM_I2,
QOP_ROT_MUL,
/* Jumps to block->successor[0] if the qinst->cond (as a
* QPU_COND_BRANCH_*) passes, or block->successor[1] if not. Note
* that block->successor[1] may be unset if the condition is ALWAYS.
*/
QOP_BRANCH,
/* Emits an ADD from src[0] to src[1], where src[0] must be a
* QOP_LOAD_IMM result and src[1] is a QUNIFORM_UNIFORMS_ADDRESS,
* required by the kernel as part of its branch validation.
*/
QOP_UNIFORMS_RESET,
};
struct queued_qpu_inst {
struct list_head link;
uint64_t inst;
};
struct qinst {
struct list_head link;
enum qop op;
struct qreg dst;
struct qreg *src;
bool sf;
bool cond_is_exec_mask;
uint8_t cond;
};
enum qstage {
/**
* Coordinate shader, runs during binning, before the VS, and just
* outputs position.
*/
QSTAGE_COORD,
QSTAGE_VERT,
QSTAGE_FRAG,
};
enum quniform_contents {
/**
* Indicates that a constant 32-bit value is copied from the program's
* uniform contents.
*/
QUNIFORM_CONSTANT,
/**
* Indicates that the program's uniform contents are used as an index
* into the GL uniform storage.
*/
QUNIFORM_UNIFORM,
/** @{
* Scaling factors from clip coordinates to relative to the viewport
* center.
*
* This is used by the coordinate and vertex shaders to produce the
* 32-bit entry consisting of 2 16-bit fields with 12.4 signed fixed
* point offsets from the viewport ccenter.
*/
QUNIFORM_VIEWPORT_X_SCALE,
QUNIFORM_VIEWPORT_Y_SCALE,
/** @} */
QUNIFORM_VIEWPORT_Z_OFFSET,
QUNIFORM_VIEWPORT_Z_SCALE,
QUNIFORM_USER_CLIP_PLANE,
/**
* A reference to a texture config parameter 0 uniform.
*
* This is a uniform implicitly loaded with a QPU_W_TMU* write, which
* defines texture type, miplevels, and such. It will be found as a
* parameter to the first QOP_TEX_[STRB] instruction in a sequence.
*/
QUNIFORM_TEXTURE_CONFIG_P0,
/**
* A reference to a texture config parameter 1 uniform.
*
* This is a uniform implicitly loaded with a QPU_W_TMU* write, which
* defines texture width, height, filters, and wrap modes. It will be
* found as a parameter to the second QOP_TEX_[STRB] instruction in a
* sequence.
*/
QUNIFORM_TEXTURE_CONFIG_P1,
/** A reference to a texture config parameter 2 cubemap stride uniform */
QUNIFORM_TEXTURE_CONFIG_P2,
QUNIFORM_TEXTURE_FIRST_LEVEL,
QUNIFORM_TEXTURE_MSAA_ADDR,
QUNIFORM_UBO_ADDR,
QUNIFORM_TEXRECT_SCALE_X,
QUNIFORM_TEXRECT_SCALE_Y,
QUNIFORM_TEXTURE_BORDER_COLOR,
QUNIFORM_BLEND_CONST_COLOR_X,
QUNIFORM_BLEND_CONST_COLOR_Y,
QUNIFORM_BLEND_CONST_COLOR_Z,
QUNIFORM_BLEND_CONST_COLOR_W,
QUNIFORM_BLEND_CONST_COLOR_RGBA,
QUNIFORM_BLEND_CONST_COLOR_AAAA,
QUNIFORM_STENCIL,
QUNIFORM_ALPHA_REF,
QUNIFORM_SAMPLE_MASK,
/* Placeholder uniform that will be updated by the kernel when used by
* an instruction writing to QPU_W_UNIFORMS_ADDRESS.
*/
QUNIFORM_UNIFORMS_ADDRESS,
};
struct vc4_varying_slot {
uint8_t slot;
uint8_t swizzle;
};
struct vc4_compiler_ubo_range {
/**
* offset in bytes from the start of the ubo where this range is
* uploaded.
*
* Only set once used is set.
*/
uint32_t dst_offset;
/**
* offset in bytes from the start of the gallium uniforms where the
* data comes from.
*/
uint32_t src_offset;
/** size in bytes of this ubo range */
uint32_t size;
/**
* Set if this range is used by the shader for indirect uniforms
* access.
*/
bool used;
};
struct vc4_key {
struct vc4_uncompiled_shader *shader_state;
struct {
enum pipe_format format;
uint8_t swizzle[4];
union {
struct {
unsigned compare_mode:1;
unsigned compare_func:3;
unsigned wrap_s:3;
unsigned wrap_t:3;
bool force_first_level:1;
};
struct {
uint16_t msaa_width, msaa_height;
};
};
} tex[VC4_MAX_TEXTURE_SAMPLERS];
uint8_t ucp_enables;
};
struct vc4_fs_key {
struct vc4_key base;
enum pipe_format color_format;
bool depth_enabled;
bool stencil_enabled;
bool stencil_twoside;
bool stencil_full_writemasks;
bool is_points;
bool is_lines;
bool alpha_test;
bool point_coord_upper_left;
bool light_twoside;
bool msaa;
bool sample_coverage;
bool sample_alpha_to_coverage;
bool sample_alpha_to_one;
uint8_t alpha_test_func;
uint8_t logicop_func;
uint32_t point_sprite_mask;
struct pipe_rt_blend_state blend;
};
struct vc4_vs_key {
struct vc4_key base;
const struct vc4_fs_inputs *fs_inputs;
enum pipe_format attr_formats[8];
bool is_coord;
bool per_vertex_point_size;
bool clamp_color;
};
/** A basic block of QIR intructions. */
struct qblock {
struct list_head link;
struct list_head instructions;
struct list_head qpu_inst_list;
struct set *predecessors;
struct qblock *successors[2];
int index;
/* Instruction IPs for the first and last instruction of the block.
* Set by vc4_qpu_schedule.c.
*/
uint32_t start_qpu_ip;
uint32_t end_qpu_ip;
/* Instruction IP for the branch instruction of the block. Set by
* vc4_qpu_schedule.c.
*/
uint32_t branch_qpu_ip;
/** @{ used by vc4_qir_live_variables.c */
BITSET_WORD *def;
BITSET_WORD *use;
BITSET_WORD *live_in;
BITSET_WORD *live_out;
int start_ip, end_ip;
/** @} */
};
struct vc4_compile {
struct vc4_context *vc4;
nir_shader *s;
nir_function_impl *impl;
struct exec_list *cf_node_list;
/**
* Mapping from nir_register * or nir_ssa_def * to array of struct
* qreg for the values.
*/
struct hash_table *def_ht;
/* For each temp, the instruction generating its value. */
struct qinst **defs;
uint32_t defs_array_size;
/**
* Inputs to the shader, arranged by TGSI declaration order.
*
* Not all fragment shader QFILE_VARY reads are present in this array.
*/
struct qreg *inputs;
struct qreg *outputs;
bool msaa_per_sample_output;
struct qreg color_reads[VC4_MAX_SAMPLES];
struct qreg sample_colors[VC4_MAX_SAMPLES];
uint32_t inputs_array_size;
uint32_t outputs_array_size;
uint32_t uniforms_array_size;
struct vc4_compiler_ubo_range *ubo_ranges;
uint32_t ubo_ranges_array_size;
/** Number of uniform areas declared in ubo_ranges. */
uint32_t num_uniform_ranges;
/** Number of uniform areas used for indirect addressed loads. */
uint32_t num_ubo_ranges;
uint32_t next_ubo_dst_offset;
/* State for whether we're executing on each channel currently. 0 if
* yes, otherwise a block number + 1 that the channel jumped to.
*/
struct qreg execute;
struct qreg line_x, point_x, point_y;
/** boolean (~0 -> true) if the fragment has been discarded. */
struct qreg discard;
struct qreg payload_FRAG_Z;
struct qreg payload_FRAG_W;
uint8_t vattr_sizes[8];
/**
* Array of the VARYING_SLOT_* of all FS QFILE_VARY reads.
*
* This includes those that aren't part of the VPM varyings, like
* point/line coordinates.
*/
struct vc4_varying_slot *input_slots;
uint32_t num_input_slots;
uint32_t input_slots_array_size;
/**
* An entry per outputs[] in the VS indicating what the VARYING_SLOT_*
* of the output is. Used to emit from the VS in the order that the
* FS needs.
*/
struct vc4_varying_slot *output_slots;
struct pipe_shader_state *shader_state;
struct vc4_key *key;
struct vc4_fs_key *fs_key;
struct vc4_vs_key *vs_key;
/* Live ranges of temps. */
int *temp_start, *temp_end;
uint32_t *uniform_data;
enum quniform_contents *uniform_contents;
uint32_t uniform_array_size;
uint32_t num_uniforms;
uint32_t num_outputs;
uint32_t num_texture_samples;
uint32_t output_position_index;
uint32_t output_color_index;
uint32_t output_point_size_index;
uint32_t output_sample_mask_index;
struct qreg undef;
enum qstage stage;
uint32_t num_temps;
struct list_head blocks;
int next_block_index;
struct qblock *cur_block;
struct qblock *loop_cont_block;
struct qblock *loop_break_block;
struct list_head qpu_inst_list;
uint64_t *qpu_insts;
uint32_t qpu_inst_count;
uint32_t qpu_inst_size;
uint32_t num_inputs;
/**
* Number of inputs from num_inputs remaining to be queued to the read
* FIFO in the VS/CS.
*/
uint32_t num_inputs_remaining;
/* Number of inputs currently in the read FIFO for the VS/CS */
uint32_t num_inputs_in_fifo;
/** Next offset in the VPM to read from in the VS/CS */
uint32_t vpm_read_offset;
uint32_t program_id;
uint32_t variant_id;
bool failed;
};
/* Special nir_load_input intrinsic index for loading the current TLB
* destination color.
*/
#define VC4_NIR_TLB_COLOR_READ_INPUT 2000000000
#define VC4_NIR_MS_MASK_OUTPUT 2000000000
struct vc4_compile *qir_compile_init(void);
void qir_compile_destroy(struct vc4_compile *c);
struct qblock *qir_new_block(struct vc4_compile *c);
void qir_set_emit_block(struct vc4_compile *c, struct qblock *block);
void qir_link_blocks(struct qblock *predecessor, struct qblock *successor);
struct qblock *qir_entry_block(struct vc4_compile *c);
struct qblock *qir_exit_block(struct vc4_compile *c);
struct qinst *qir_inst(enum qop op, struct qreg dst,
struct qreg src0, struct qreg src1);
struct qinst *qir_inst4(enum qop op, struct qreg dst,
struct qreg a,
struct qreg b,
struct qreg c,
struct qreg d);
void qir_remove_instruction(struct vc4_compile *c, struct qinst *qinst);
struct qreg qir_uniform(struct vc4_compile *c,
enum quniform_contents contents,
uint32_t data);
void qir_schedule_instructions(struct vc4_compile *c);
void qir_reorder_uniforms(struct vc4_compile *c);
void qir_emit_uniform_stream_resets(struct vc4_compile *c);
struct qreg qir_emit_def(struct vc4_compile *c, struct qinst *inst);
struct qinst *qir_emit_nondef(struct vc4_compile *c, struct qinst *inst);
struct qreg qir_get_temp(struct vc4_compile *c);
void qir_calculate_live_intervals(struct vc4_compile *c);
int qir_get_op_nsrc(enum qop qop);
bool qir_reg_equals(struct qreg a, struct qreg b);
bool qir_has_side_effects(struct vc4_compile *c, struct qinst *inst);
bool qir_has_side_effect_reads(struct vc4_compile *c, struct qinst *inst);
bool qir_is_mul(struct qinst *inst);
bool qir_is_raw_mov(struct qinst *inst);
bool qir_is_tex(struct qinst *inst);
bool qir_is_float_input(struct qinst *inst);
bool qir_depends_on_flags(struct qinst *inst);
bool qir_writes_r4(struct qinst *inst);
struct qreg qir_follow_movs(struct vc4_compile *c, struct qreg reg);
uint8_t qir_channels_written(struct qinst *inst);
void qir_dump(struct vc4_compile *c);
void qir_dump_inst(struct vc4_compile *c, struct qinst *inst);
const char *qir_get_stage_name(enum qstage stage);
void qir_validate(struct vc4_compile *c);
void qir_optimize(struct vc4_compile *c);
bool qir_opt_algebraic(struct vc4_compile *c);
bool qir_opt_constant_folding(struct vc4_compile *c);
bool qir_opt_copy_propagation(struct vc4_compile *c);
bool qir_opt_dead_code(struct vc4_compile *c);
bool qir_opt_peephole_sf(struct vc4_compile *c);
bool qir_opt_small_immediates(struct vc4_compile *c);
bool qir_opt_vpm(struct vc4_compile *c);
void vc4_nir_lower_blend(nir_shader *s, struct vc4_compile *c);
void vc4_nir_lower_io(nir_shader *s, struct vc4_compile *c);
nir_ssa_def *vc4_nir_get_swizzled_channel(struct nir_builder *b,
nir_ssa_def **srcs, int swiz);
void vc4_nir_lower_txf_ms(nir_shader *s, struct vc4_compile *c);
void qir_lower_uniforms(struct vc4_compile *c);
uint32_t qpu_schedule_instructions(struct vc4_compile *c);
void qir_SF(struct vc4_compile *c, struct qreg src);
static inline struct qreg
qir_uniform_ui(struct vc4_compile *c, uint32_t ui)
{
return qir_uniform(c, QUNIFORM_CONSTANT, ui);
}
static inline struct qreg
qir_uniform_f(struct vc4_compile *c, float f)
{
return qir_uniform(c, QUNIFORM_CONSTANT, fui(f));
}
#define QIR_ALU0(name) \
static inline struct qreg \
qir_##name(struct vc4_compile *c) \
{ \
return qir_emit_def(c, qir_inst(QOP_##name, c->undef, \
c->undef, c->undef)); \
} \
static inline struct qinst * \
qir_##name##_dest(struct vc4_compile *c, struct qreg dest) \
{ \
return qir_emit_nondef(c, qir_inst(QOP_##name, dest, \
c->undef, c->undef)); \
}
#define QIR_ALU1(name) \
static inline struct qreg \
qir_##name(struct vc4_compile *c, struct qreg a) \
{ \
return qir_emit_def(c, qir_inst(QOP_##name, c->undef, \
a, c->undef)); \
} \
static inline struct qinst * \
qir_##name##_dest(struct vc4_compile *c, struct qreg dest, \
struct qreg a) \
{ \
return qir_emit_nondef(c, qir_inst(QOP_##name, dest, a, \
c->undef)); \
}
#define QIR_ALU2(name) \
static inline struct qreg \
qir_##name(struct vc4_compile *c, struct qreg a, struct qreg b) \
{ \
return qir_emit_def(c, qir_inst(QOP_##name, c->undef, a, b)); \
} \
static inline struct qinst * \
qir_##name##_dest(struct vc4_compile *c, struct qreg dest, \
struct qreg a, struct qreg b) \
{ \
return qir_emit_nondef(c, qir_inst(QOP_##name, dest, a, b)); \
}
#define QIR_NODST_1(name) \
static inline struct qinst * \
qir_##name(struct vc4_compile *c, struct qreg a) \
{ \
return qir_emit_nondef(c, qir_inst(QOP_##name, c->undef, \
a, c->undef)); \
}
#define QIR_NODST_2(name) \
static inline struct qinst * \
qir_##name(struct vc4_compile *c, struct qreg a, struct qreg b) \
{ \
return qir_emit_nondef(c, qir_inst(QOP_##name, c->undef, \
a, b)); \
}
#define QIR_PAYLOAD(name) \
static inline struct qreg \
qir_##name(struct vc4_compile *c) \
{ \
struct qreg *payload = &c->payload_##name; \
if (payload->file != QFILE_NULL) \
return *payload; \
*payload = qir_get_temp(c); \
struct qinst *inst = qir_inst(QOP_##name, *payload, \
c->undef, c->undef); \
struct qblock *entry = qir_entry_block(c); \
list_add(&inst->link, &entry->instructions); \
c->defs[payload->index] = inst; \
return *payload; \
}
QIR_ALU1(MOV)
QIR_ALU1(FMOV)
QIR_ALU1(MMOV)
QIR_ALU2(FADD)
QIR_ALU2(FSUB)
QIR_ALU2(FMUL)
QIR_ALU2(V8MULD)
QIR_ALU2(V8MIN)
QIR_ALU2(V8MAX)
QIR_ALU2(V8ADDS)
QIR_ALU2(V8SUBS)
QIR_ALU2(MUL24)
QIR_ALU2(FMIN)
QIR_ALU2(FMAX)
QIR_ALU2(FMINABS)
QIR_ALU2(FMAXABS)
QIR_ALU1(FTOI)
QIR_ALU1(ITOF)
QIR_ALU2(ADD)
QIR_ALU2(SUB)
QIR_ALU2(SHL)
QIR_ALU2(SHR)
QIR_ALU2(ASR)
QIR_ALU2(MIN)
QIR_ALU2(MAX)
QIR_ALU2(AND)
QIR_ALU2(OR)
QIR_ALU2(XOR)
QIR_ALU1(NOT)
QIR_ALU1(RCP)
QIR_ALU1(RSQ)
QIR_ALU1(EXP2)
QIR_ALU1(LOG2)
QIR_ALU1(VARY_ADD_C)
QIR_NODST_2(TEX_S)
QIR_NODST_2(TEX_T)
QIR_NODST_2(TEX_R)
QIR_NODST_2(TEX_B)
QIR_NODST_2(TEX_DIRECT)
QIR_PAYLOAD(FRAG_Z)
QIR_PAYLOAD(FRAG_W)
QIR_ALU0(TEX_RESULT)
QIR_ALU0(TLB_COLOR_READ)
QIR_NODST_1(MS_MASK)
static inline struct qreg
qir_SEL(struct vc4_compile *c, uint8_t cond, struct qreg src0, struct qreg src1)
{
struct qreg t = qir_get_temp(c);
struct qinst *a = qir_MOV_dest(c, t, src0);
struct qinst *b = qir_MOV_dest(c, t, src1);
a->cond = cond;
b->cond = qpu_cond_complement(cond);
return t;
}
static inline struct qreg
qir_UNPACK_8_F(struct vc4_compile *c, struct qreg src, int i)
{
struct qreg t = qir_FMOV(c, src);
c->defs[t.index]->src[0].pack = QPU_UNPACK_8A + i;
return t;
}
static inline struct qreg
qir_UNPACK_8_I(struct vc4_compile *c, struct qreg src, int i)
{
struct qreg t = qir_MOV(c, src);
c->defs[t.index]->src[0].pack = QPU_UNPACK_8A + i;
return t;
}
static inline struct qreg
qir_UNPACK_16_F(struct vc4_compile *c, struct qreg src, int i)
{
struct qreg t = qir_FMOV(c, src);
c->defs[t.index]->src[0].pack = QPU_UNPACK_16A + i;
return t;
}
static inline struct qreg
qir_UNPACK_16_I(struct vc4_compile *c, struct qreg src, int i)
{
struct qreg t = qir_MOV(c, src);
c->defs[t.index]->src[0].pack = QPU_UNPACK_16A + i;
return t;
}
static inline void
qir_PACK_8_F(struct vc4_compile *c, struct qreg dest, struct qreg val, int chan)
{
assert(!dest.pack);
dest.pack = QPU_PACK_MUL_8A + chan;
qir_emit_nondef(c, qir_inst(QOP_MMOV, dest, val, c->undef));
}
static inline struct qreg
qir_PACK_8888_F(struct vc4_compile *c, struct qreg val)
{
struct qreg dest = qir_MMOV(c, val);
c->defs[dest.index]->dst.pack = QPU_PACK_MUL_8888;
return dest;
}
static inline struct qreg
qir_POW(struct vc4_compile *c, struct qreg x, struct qreg y)
{
return qir_EXP2(c, qir_FMUL(c,
y,
qir_LOG2(c, x)));
}
static inline void
qir_VPM_WRITE(struct vc4_compile *c, struct qreg val)
{
qir_MOV_dest(c, qir_reg(QFILE_VPM, 0), val);
}
static inline struct qreg
qir_LOAD_IMM(struct vc4_compile *c, uint32_t val)
{
return qir_emit_def(c, qir_inst(QOP_LOAD_IMM, c->undef,
qir_reg(QFILE_LOAD_IMM, val), c->undef));
}
static inline struct qreg
qir_LOAD_IMM_U2(struct vc4_compile *c, uint32_t val)
{
return qir_emit_def(c, qir_inst(QOP_LOAD_IMM_U2, c->undef,
qir_reg(QFILE_LOAD_IMM, val),
c->undef));
}
static inline struct qreg
qir_LOAD_IMM_I2(struct vc4_compile *c, uint32_t val)
{
return qir_emit_def(c, qir_inst(QOP_LOAD_IMM_I2, c->undef,
qir_reg(QFILE_LOAD_IMM, val),
c->undef));
}
/** Shifts the multiply output to the right by rot channels */
static inline struct qreg
qir_ROT_MUL(struct vc4_compile *c, struct qreg val, uint32_t rot)
{
return qir_emit_def(c, qir_inst(QOP_ROT_MUL, c->undef,
val,
qir_reg(QFILE_LOAD_IMM,
QPU_SMALL_IMM_MUL_ROT + rot)));
}
static inline struct qinst *
qir_MOV_cond(struct vc4_compile *c, uint8_t cond,
struct qreg dest, struct qreg src)
{
struct qinst *mov = qir_MOV_dest(c, dest, src);
mov->cond = cond;
return mov;
}
static inline struct qinst *
qir_BRANCH(struct vc4_compile *c, uint8_t cond)
{
struct qinst *inst = qir_inst(QOP_BRANCH, c->undef, c->undef, c->undef);
inst->cond = cond;
qir_emit_nondef(c, inst);
return inst;
}
#define qir_for_each_block(block, c) \
list_for_each_entry(struct qblock, block, &c->blocks, link)
#define qir_for_each_block_rev(block, c) \
list_for_each_entry_rev(struct qblock, block, &c->blocks, link)
/* Loop over the non-NULL members of the successors array. */
#define qir_for_each_successor(succ, block) \
for (struct qblock *succ = block->successors[0]; \
succ != NULL; \
succ = (succ == block->successors[1] ? NULL : \
block->successors[1]))
#define qir_for_each_inst(inst, block) \
list_for_each_entry(struct qinst, inst, &block->instructions, link)
#define qir_for_each_inst_rev(inst, block) \
list_for_each_entry_rev(struct qinst, inst, &block->instructions, link)
#define qir_for_each_inst_safe(inst, block) \
list_for_each_entry_safe(struct qinst, inst, &block->instructions, link)
#define qir_for_each_inst_inorder(inst, c) \
qir_for_each_block(_block, c) \
qir_for_each_inst(inst, _block)
#endif /* VC4_QIR_H */
|