i965/fs: Migrate Gen4 send dependency workarounds to the IR builder.
[mesa.git] / src / mesa / drivers / dri / i965 / brw_fs.h
1 /*
2 * Copyright © 2010 Intel Corporation
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
21 * IN THE SOFTWARE.
22 *
23 * Authors:
24 * Eric Anholt <eric@anholt.net>
25 *
26 */
27
28 #pragma once
29
30 #include "brw_shader.h"
31 #include "brw_ir_fs.h"
32 #include "brw_fs_builder.h"
33
34 extern "C" {
35
36 #include <sys/types.h>
37
38 #include "main/macros.h"
39 #include "main/shaderobj.h"
40 #include "main/uniforms.h"
41 #include "program/prog_parameter.h"
42 #include "program/prog_print.h"
43 #include "program/prog_optimize.h"
44 #include "util/register_allocate.h"
45 #include "program/hash_table.h"
46 #include "brw_context.h"
47 #include "brw_eu.h"
48 #include "brw_wm.h"
49 #include "intel_asm_annotation.h"
50 }
51 #include "glsl/glsl_types.h"
52 #include "glsl/ir.h"
53 #include "glsl/nir/nir.h"
54 #include "program/sampler.h"
55
56 struct bblock_t;
57 namespace {
58 struct acp_entry;
59 }
60
61 namespace brw {
62 class fs_live_variables;
63 }
64
65 /**
66 * The fragment shader front-end.
67 *
68 * Translates either GLSL IR or Mesa IR (for ARB_fragment_program) into FS IR.
69 */
70 class fs_visitor : public backend_shader
71 {
72 public:
73 const fs_reg reg_null_f;
74 const fs_reg reg_null_d;
75 const fs_reg reg_null_ud;
76
77 fs_visitor(struct brw_context *brw,
78 void *mem_ctx,
79 gl_shader_stage stage,
80 const void *key,
81 struct brw_stage_prog_data *prog_data,
82 struct gl_shader_program *shader_prog,
83 struct gl_program *prog,
84 unsigned dispatch_width);
85
86 ~fs_visitor();
87
88 fs_reg vgrf(const glsl_type *const type);
89 fs_reg vgrf(int num_components);
90 void import_uniforms(fs_visitor *v);
91 void setup_uniform_clipplane_values();
92 void compute_clip_distance();
93
94 uint32_t gather_channel(int orig_chan, uint32_t sampler);
95 void swizzle_result(ir_texture_opcode op, int dest_components,
96 fs_reg orig_val, uint32_t sampler);
97
98 fs_inst *emit(fs_inst *inst);
99 void emit(exec_list list);
100
101 fs_inst *emit(enum opcode opcode);
102 fs_inst *emit(enum opcode opcode, const fs_reg &dst);
103 fs_inst *emit(enum opcode opcode, const fs_reg &dst, const fs_reg &src0);
104 fs_inst *emit(enum opcode opcode, const fs_reg &dst, const fs_reg &src0,
105 const fs_reg &src1);
106 fs_inst *emit(enum opcode opcode, const fs_reg &dst,
107 const fs_reg &src0, const fs_reg &src1, const fs_reg &src2);
108 fs_inst *emit(enum opcode opcode, const fs_reg &dst,
109 fs_reg src[], int sources);
110
111 fs_inst *MOV(const fs_reg &dst, const fs_reg &src);
112 fs_inst *NOT(const fs_reg &dst, const fs_reg &src);
113 fs_inst *RNDD(const fs_reg &dst, const fs_reg &src);
114 fs_inst *RNDE(const fs_reg &dst, const fs_reg &src);
115 fs_inst *RNDZ(const fs_reg &dst, const fs_reg &src);
116 fs_inst *FRC(const fs_reg &dst, const fs_reg &src);
117 fs_inst *ADD(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
118 fs_inst *MUL(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
119 fs_inst *MACH(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
120 fs_inst *MAC(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
121 fs_inst *SHL(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
122 fs_inst *SHR(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
123 fs_inst *ASR(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
124 fs_inst *AND(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
125 fs_inst *OR(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
126 fs_inst *XOR(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
127 fs_inst *IF(enum brw_predicate predicate);
128 fs_inst *IF(const fs_reg &src0, const fs_reg &src1,
129 enum brw_conditional_mod condition);
130 fs_inst *CMP(fs_reg dst, fs_reg src0, fs_reg src1,
131 enum brw_conditional_mod condition);
132 fs_inst *LRP(const fs_reg &dst, const fs_reg &a, const fs_reg &y,
133 const fs_reg &x);
134 fs_inst *BFREV(const fs_reg &dst, const fs_reg &value);
135 fs_inst *BFE(const fs_reg &dst, const fs_reg &bits, const fs_reg &offset,
136 const fs_reg &value);
137 fs_inst *BFI1(const fs_reg &dst, const fs_reg &bits, const fs_reg &offset);
138 fs_inst *BFI2(const fs_reg &dst, const fs_reg &bfi1_dst,
139 const fs_reg &insert, const fs_reg &base);
140 fs_inst *FBH(const fs_reg &dst, const fs_reg &value);
141 fs_inst *FBL(const fs_reg &dst, const fs_reg &value);
142 fs_inst *CBIT(const fs_reg &dst, const fs_reg &value);
143 fs_inst *MAD(const fs_reg &dst, const fs_reg &c, const fs_reg &b,
144 const fs_reg &a);
145 fs_inst *ADDC(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
146 fs_inst *SUBB(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
147 fs_inst *SEL(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1);
148
149 int type_size(const struct glsl_type *type);
150 fs_inst *get_instruction_generating_reg(fs_inst *start,
151 fs_inst *end,
152 const fs_reg &reg);
153
154 fs_inst *LOAD_PAYLOAD(const fs_reg &dst, fs_reg *src, int sources,
155 int header_size);
156
157 exec_list VARYING_PULL_CONSTANT_LOAD(const fs_reg &dst,
158 const fs_reg &surf_index,
159 const fs_reg &varying_offset,
160 uint32_t const_offset);
161 void DEP_RESOLVE_MOV(const brw::fs_builder &bld, int grf);
162
163 bool run_fs();
164 bool run_vs();
165 bool run_cs();
166 void optimize();
167 void allocate_registers();
168 void assign_binding_table_offsets();
169 void setup_payload_gen4();
170 void setup_payload_gen6();
171 void setup_vs_payload();
172 void setup_cs_payload();
173 void fixup_3src_null_dest();
174 void assign_curb_setup();
175 void calculate_urb_setup();
176 void assign_urb_setup();
177 void assign_vs_urb_setup();
178 bool assign_regs(bool allow_spilling);
179 void assign_regs_trivial();
180 void get_used_mrfs(bool *mrf_used);
181 void setup_payload_interference(struct ra_graph *g, int payload_reg_count,
182 int first_payload_node);
183 void setup_mrf_hack_interference(struct ra_graph *g,
184 int first_mrf_hack_node);
185 int choose_spill_reg(struct ra_graph *g);
186 void spill_reg(int spill_reg);
187 void split_virtual_grfs();
188 bool compact_virtual_grfs();
189 void move_uniform_array_access_to_pull_constants();
190 void assign_constant_locations();
191 void demote_pull_constants();
192 void invalidate_live_intervals();
193 void calculate_live_intervals();
194 void calculate_register_pressure();
195 bool opt_algebraic();
196 bool opt_redundant_discard_jumps();
197 bool opt_cse();
198 bool opt_cse_local(bblock_t *block);
199 bool opt_copy_propagate();
200 bool try_copy_propagate(fs_inst *inst, int arg, acp_entry *entry);
201 bool try_constant_propagate(fs_inst *inst, acp_entry *entry);
202 bool opt_copy_propagate_local(void *mem_ctx, bblock_t *block,
203 exec_list *acp);
204 bool opt_register_renaming();
205 bool register_coalesce();
206 bool compute_to_mrf();
207 bool eliminate_find_live_channel();
208 bool dead_code_eliminate();
209 bool remove_duplicate_mrf_writes();
210
211 bool opt_sampler_eot();
212 bool virtual_grf_interferes(int a, int b);
213 void schedule_instructions(instruction_scheduler_mode mode);
214 void insert_gen4_send_dependency_workarounds();
215 void insert_gen4_pre_send_dependency_workarounds(bblock_t *block,
216 fs_inst *inst);
217 void insert_gen4_post_send_dependency_workarounds(bblock_t *block,
218 fs_inst *inst);
219 void vfail(const char *msg, va_list args);
220 void fail(const char *msg, ...);
221 void no16(const char *msg, ...);
222 void lower_uniform_pull_constant_loads();
223 bool lower_load_payload();
224 bool lower_integer_multiplication();
225 bool opt_combine_constants();
226
227 void emit_dummy_fs();
228 void emit_repclear_shader();
229 fs_reg *emit_fragcoord_interpolation(bool pixel_center_integer,
230 bool origin_upper_left);
231 fs_inst *emit_linterp(const fs_reg &attr, const fs_reg &interp,
232 glsl_interp_qualifier interpolation_mode,
233 bool is_centroid, bool is_sample);
234 fs_reg *emit_frontfacing_interpolation();
235 fs_reg *emit_samplepos_setup();
236 fs_reg *emit_sampleid_setup();
237 void emit_general_interpolation(fs_reg attr, const char *name,
238 const glsl_type *type,
239 glsl_interp_qualifier interpolation_mode,
240 int location, bool mod_centroid,
241 bool mod_sample);
242 fs_reg *emit_vs_system_value(int location);
243 void emit_interpolation_setup_gen4();
244 void emit_interpolation_setup_gen6();
245 void compute_sample_position(fs_reg dst, fs_reg int_sample_pos);
246 fs_reg rescale_texcoord(fs_reg coordinate, int coord_components,
247 bool is_rect, uint32_t sampler, int texunit);
248 fs_inst *emit_texture_gen4(ir_texture_opcode op, fs_reg dst,
249 fs_reg coordinate, int coord_components,
250 fs_reg shadow_comp,
251 fs_reg lod, fs_reg lod2, int grad_components,
252 uint32_t sampler);
253 fs_inst *emit_texture_gen4_simd16(ir_texture_opcode op, fs_reg dst,
254 fs_reg coordinate, int vector_elements,
255 fs_reg shadow_c, fs_reg lod,
256 uint32_t sampler);
257 fs_inst *emit_texture_gen5(ir_texture_opcode op, fs_reg dst,
258 fs_reg coordinate, int coord_components,
259 fs_reg shadow_comp,
260 fs_reg lod, fs_reg lod2, int grad_components,
261 fs_reg sample_index, uint32_t sampler,
262 bool has_offset);
263 fs_inst *emit_texture_gen7(ir_texture_opcode op, fs_reg dst,
264 fs_reg coordinate, int coord_components,
265 fs_reg shadow_comp,
266 fs_reg lod, fs_reg lod2, int grad_components,
267 fs_reg sample_index, fs_reg mcs, fs_reg sampler,
268 fs_reg offset_value);
269 void emit_texture(ir_texture_opcode op,
270 const glsl_type *dest_type,
271 fs_reg coordinate, int components,
272 fs_reg shadow_c,
273 fs_reg lod, fs_reg dpdy, int grad_components,
274 fs_reg sample_index,
275 fs_reg offset,
276 fs_reg mcs,
277 int gather_component,
278 bool is_cube_array,
279 bool is_rect,
280 uint32_t sampler,
281 fs_reg sampler_reg,
282 int texunit);
283 fs_reg emit_mcs_fetch(fs_reg coordinate, int components, fs_reg sampler);
284 void emit_gen6_gather_wa(uint8_t wa, fs_reg dst);
285 void resolve_source_modifiers(fs_reg *src);
286 fs_reg fix_math_operand(fs_reg src);
287 fs_inst *emit_math(enum opcode op, fs_reg dst, fs_reg src0);
288 fs_inst *emit_math(enum opcode op, fs_reg dst, fs_reg src0, fs_reg src1);
289 fs_inst *emit_lrp(const fs_reg &dst, const fs_reg &x, const fs_reg &y,
290 const fs_reg &a);
291 void emit_discard_jump();
292 /** Copy any live channel from \p src to the first channel of \p dst. */
293 void emit_uniformize(const fs_reg &dst, const fs_reg &src);
294 bool try_replace_with_sel();
295 bool opt_peephole_sel();
296 bool opt_peephole_predicated_break();
297 bool opt_saturate_propagation();
298 bool opt_cmod_propagation();
299 bool opt_zero_samples();
300 void emit_unspill(bblock_t *block, fs_inst *inst, fs_reg reg,
301 uint32_t spill_offset, int count);
302 void emit_spill(bblock_t *block, fs_inst *inst, fs_reg reg,
303 uint32_t spill_offset, int count);
304
305 void emit_nir_code();
306 void nir_setup_inputs(nir_shader *shader);
307 void nir_setup_outputs(nir_shader *shader);
308 void nir_setup_uniforms(nir_shader *shader);
309 void nir_setup_uniform(nir_variable *var);
310 void nir_setup_builtin_uniform(nir_variable *var);
311 void nir_emit_system_values(nir_shader *shader);
312 void nir_emit_impl(nir_function_impl *impl);
313 void nir_emit_cf_list(exec_list *list);
314 void nir_emit_if(nir_if *if_stmt);
315 void nir_emit_loop(nir_loop *loop);
316 void nir_emit_block(nir_block *block);
317 void nir_emit_instr(nir_instr *instr);
318 void nir_emit_alu(nir_alu_instr *instr);
319 void nir_emit_intrinsic(nir_intrinsic_instr *instr);
320 void nir_emit_texture(nir_tex_instr *instr);
321 void nir_emit_jump(nir_jump_instr *instr);
322 fs_reg get_nir_src(nir_src src);
323 fs_reg get_nir_dest(nir_dest dest);
324 void emit_percomp(fs_inst *inst, unsigned wr_mask);
325
326 bool optimize_frontfacing_ternary(nir_alu_instr *instr,
327 const fs_reg &result);
328
329 void setup_color_payload(fs_reg *dst, fs_reg color, unsigned components,
330 unsigned exec_size, bool use_2nd_half);
331 void emit_alpha_test();
332 fs_inst *emit_single_fb_write(fs_reg color1, fs_reg color2,
333 fs_reg src0_alpha, unsigned components,
334 unsigned exec_size, bool use_2nd_half = false);
335 void emit_fb_writes();
336 void emit_urb_writes();
337 void emit_cs_terminate();
338
339 void emit_shader_time_begin();
340 void emit_shader_time_end();
341 fs_inst *SHADER_TIME_ADD(enum shader_time_shader_type type, fs_reg value);
342
343 void emit_untyped_atomic(unsigned atomic_op, unsigned surf_index,
344 fs_reg dst, fs_reg offset, fs_reg src0,
345 fs_reg src1);
346
347 void emit_untyped_surface_read(unsigned surf_index, fs_reg dst,
348 fs_reg offset);
349
350 void resolve_ud_negate(fs_reg *reg);
351
352 fs_reg get_timestamp(fs_inst **out_mov);
353
354 struct brw_reg interp_reg(int location, int channel);
355 int implied_mrf_writes(fs_inst *inst);
356
357 virtual void dump_instructions();
358 virtual void dump_instructions(const char *name);
359 void dump_instruction(backend_instruction *inst);
360 void dump_instruction(backend_instruction *inst, FILE *file);
361
362 const void *const key;
363 const struct brw_sampler_prog_key_data *key_tex;
364
365 struct brw_stage_prog_data *prog_data;
366 unsigned int sanity_param_count;
367
368 int *param_size;
369
370 int *virtual_grf_start;
371 int *virtual_grf_end;
372 brw::fs_live_variables *live_intervals;
373
374 int *regs_live_at_ip;
375
376 /** Number of uniform variable components visited. */
377 unsigned uniforms;
378
379 /** Total number of direct uniforms we can get from NIR */
380 unsigned num_direct_uniforms;
381
382 /** Byte-offset for the next available spot in the scratch space buffer. */
383 unsigned last_scratch;
384
385 /**
386 * Array mapping UNIFORM register numbers to the pull parameter index,
387 * or -1 if this uniform register isn't being uploaded as a pull constant.
388 */
389 int *pull_constant_loc;
390
391 /**
392 * Array mapping UNIFORM register numbers to the push parameter index,
393 * or -1 if this uniform register isn't being uploaded as a push constant.
394 */
395 int *push_constant_loc;
396
397 fs_reg frag_depth;
398 fs_reg sample_mask;
399 fs_reg outputs[VARYING_SLOT_MAX];
400 unsigned output_components[VARYING_SLOT_MAX];
401 fs_reg dual_src_output;
402 bool do_dual_src;
403 int first_non_payload_grf;
404 /** Either BRW_MAX_GRF or GEN7_MRF_HACK_START */
405 unsigned max_grf;
406
407 fs_reg *nir_locals;
408 fs_reg *nir_globals;
409 fs_reg nir_inputs;
410 fs_reg nir_outputs;
411 fs_reg *nir_system_values;
412
413 /** @{ debug annotation info */
414 const char *current_annotation;
415 const void *base_ir;
416 /** @} */
417
418 bool failed;
419 char *fail_msg;
420 bool simd16_unsupported;
421 char *no16_msg;
422
423 /* Result of last visit() method. Still used by emit_texture() */
424 fs_reg result;
425
426 /** Register numbers for thread payload fields. */
427 struct {
428 uint8_t source_depth_reg;
429 uint8_t source_w_reg;
430 uint8_t aa_dest_stencil_reg;
431 uint8_t dest_depth_reg;
432 uint8_t sample_pos_reg;
433 uint8_t sample_mask_in_reg;
434 uint8_t barycentric_coord_reg[BRW_WM_BARYCENTRIC_INTERP_MODE_COUNT];
435
436 /** The number of thread payload registers the hardware will supply. */
437 uint8_t num_regs;
438 } payload;
439
440 bool source_depth_to_render_target;
441 bool runtime_check_aads_emit;
442
443 fs_reg pixel_x;
444 fs_reg pixel_y;
445 fs_reg wpos_w;
446 fs_reg pixel_w;
447 fs_reg delta_xy[BRW_WM_BARYCENTRIC_INTERP_MODE_COUNT];
448 fs_reg shader_start_time;
449 fs_reg userplane[MAX_CLIP_PLANES];
450
451 unsigned grf_used;
452 bool spilled_any_registers;
453
454 const unsigned dispatch_width; /**< 8 or 16 */
455
456 unsigned promoted_constants;
457 brw::fs_builder bld;
458 };
459
460 /**
461 * The fragment shader code generator.
462 *
463 * Translates FS IR to actual i965 assembly code.
464 */
465 class fs_generator
466 {
467 public:
468 fs_generator(struct brw_context *brw,
469 void *mem_ctx,
470 const void *key,
471 struct brw_stage_prog_data *prog_data,
472 struct gl_program *fp,
473 unsigned promoted_constants,
474 bool runtime_check_aads_emit,
475 const char *stage_abbrev);
476 ~fs_generator();
477
478 void enable_debug(const char *shader_name);
479 int generate_code(const cfg_t *cfg, int dispatch_width);
480 const unsigned *get_assembly(unsigned int *assembly_size);
481
482 private:
483 void fire_fb_write(fs_inst *inst,
484 struct brw_reg payload,
485 struct brw_reg implied_header,
486 GLuint nr);
487 void generate_fb_write(fs_inst *inst, struct brw_reg payload);
488 void generate_urb_write(fs_inst *inst, struct brw_reg payload);
489 void generate_cs_terminate(fs_inst *inst, struct brw_reg payload);
490 void generate_blorp_fb_write(fs_inst *inst);
491 void generate_linterp(fs_inst *inst, struct brw_reg dst,
492 struct brw_reg *src);
493 void generate_tex(fs_inst *inst, struct brw_reg dst, struct brw_reg src,
494 struct brw_reg sampler_index);
495 void generate_math_gen6(fs_inst *inst,
496 struct brw_reg dst,
497 struct brw_reg src0,
498 struct brw_reg src1);
499 void generate_math_gen4(fs_inst *inst,
500 struct brw_reg dst,
501 struct brw_reg src);
502 void generate_math_g45(fs_inst *inst,
503 struct brw_reg dst,
504 struct brw_reg src);
505 void generate_ddx(enum opcode op, struct brw_reg dst, struct brw_reg src);
506 void generate_ddy(enum opcode op, struct brw_reg dst, struct brw_reg src,
507 bool negate_value);
508 void generate_scratch_write(fs_inst *inst, struct brw_reg src);
509 void generate_scratch_read(fs_inst *inst, struct brw_reg dst);
510 void generate_scratch_read_gen7(fs_inst *inst, struct brw_reg dst);
511 void generate_uniform_pull_constant_load(fs_inst *inst, struct brw_reg dst,
512 struct brw_reg index,
513 struct brw_reg offset);
514 void generate_uniform_pull_constant_load_gen7(fs_inst *inst,
515 struct brw_reg dst,
516 struct brw_reg surf_index,
517 struct brw_reg offset);
518 void generate_varying_pull_constant_load(fs_inst *inst, struct brw_reg dst,
519 struct brw_reg index,
520 struct brw_reg offset);
521 void generate_varying_pull_constant_load_gen7(fs_inst *inst,
522 struct brw_reg dst,
523 struct brw_reg index,
524 struct brw_reg offset);
525 void generate_mov_dispatch_to_flags(fs_inst *inst);
526
527 void generate_pixel_interpolator_query(fs_inst *inst,
528 struct brw_reg dst,
529 struct brw_reg src,
530 struct brw_reg msg_data,
531 unsigned msg_type);
532
533 void generate_set_omask(fs_inst *inst,
534 struct brw_reg dst,
535 struct brw_reg sample_mask);
536
537 void generate_set_sample_id(fs_inst *inst,
538 struct brw_reg dst,
539 struct brw_reg src0,
540 struct brw_reg src1);
541
542 void generate_set_simd4x2_offset(fs_inst *inst,
543 struct brw_reg dst,
544 struct brw_reg offset);
545 void generate_discard_jump(fs_inst *inst);
546
547 void generate_pack_half_2x16_split(fs_inst *inst,
548 struct brw_reg dst,
549 struct brw_reg x,
550 struct brw_reg y);
551 void generate_unpack_half_2x16_split(fs_inst *inst,
552 struct brw_reg dst,
553 struct brw_reg src);
554
555 void generate_shader_time_add(fs_inst *inst,
556 struct brw_reg payload,
557 struct brw_reg offset,
558 struct brw_reg value);
559
560 bool patch_discard_jumps_to_fb_writes();
561
562 struct brw_context *brw;
563 const struct brw_device_info *devinfo;
564
565 struct brw_codegen *p;
566 const void * const key;
567 struct brw_stage_prog_data * const prog_data;
568
569 const struct gl_program *prog;
570
571 unsigned dispatch_width; /**< 8 or 16 */
572
573 exec_list discard_halt_patches;
574 unsigned promoted_constants;
575 bool runtime_check_aads_emit;
576 bool debug_flag;
577 const char *shader_name;
578 const char *stage_abbrev;
579 void *mem_ctx;
580 };
581
582 bool brw_do_channel_expressions(struct exec_list *instructions);
583 bool brw_do_vector_splitting(struct exec_list *instructions);
584 void brw_setup_tex_for_precompile(struct brw_context *brw,
585 struct brw_sampler_prog_key_data *tex,
586 struct gl_program *prog);