1 /**************************************************************************
2 *
3 * Copyright 2009 VMware, Inc.
4 * All Rights Reserved.
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sub license, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
13 *
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
16 * of the Software.
17 *
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21 * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
25 *
26 **************************************************************************/
27
28 /**
29 * LLVM control flow build helpers.
30 *
31 * @author Jose Fonseca <jfonseca@vmware.com>
32 */
33
34 #include "util/u_debug.h"
35 #include "util/u_memory.h"
36
37 #include "lp_bld_init.h"
38 #include "lp_bld_type.h"
39 #include "lp_bld_flow.h"
40
41
42 /**
43 * Insert a new block, right where builder is pointing to.
44 *
45 * This is useful important not only for aesthetic reasons, but also for
46 * performance reasons, as frequently run blocks should be laid out next to
47 * each other and fall-throughs maximized.
48 *
49 * See also llvm/lib/Transforms/Scalar/BasicBlockPlacement.cpp.
50 *
51 * Note: this function has no dependencies on the flow code and could
52 * be used elsewhere.
53 */
54 LLVMBasicBlockRef
lp_build_insert_new_block(struct gallivm_state * gallivm,const char * name)55 lp_build_insert_new_block(struct gallivm_state *gallivm, const char *name)
56 {
57 LLVMBasicBlockRef current_block;
58 LLVMBasicBlockRef next_block;
59 LLVMBasicBlockRef new_block;
60
61 /* get current basic block */
62 current_block = LLVMGetInsertBlock(gallivm->builder);
63
64 /* check if there's another block after this one */
65 next_block = LLVMGetNextBasicBlock(current_block);
66 if (next_block) {
67 /* insert the new block before the next block */
68 new_block = LLVMInsertBasicBlockInContext(gallivm->context, next_block, name);
69 }
70 else {
71 /* append new block after current block */
72 LLVMValueRef function = LLVMGetBasicBlockParent(current_block);
73 new_block = LLVMAppendBasicBlockInContext(gallivm->context, function, name);
74 }
75
76 return new_block;
77 }
78
79
80 /**
81 * Begin a "skip" block. Inside this block we can test a condition and
82 * skip to the end of the block if the condition is false.
83 */
84 void
lp_build_flow_skip_begin(struct lp_build_skip_context * skip,struct gallivm_state * gallivm)85 lp_build_flow_skip_begin(struct lp_build_skip_context *skip,
86 struct gallivm_state *gallivm)
87 {
88 skip->gallivm = gallivm;
89 /* create new basic block */
90 skip->block = lp_build_insert_new_block(gallivm, "skip");
91 }
92
93
94 /**
95 * Insert code to test a condition and branch to the end of the current
96 * skip block if the condition is true.
97 */
98 void
lp_build_flow_skip_cond_break(struct lp_build_skip_context * skip,LLVMValueRef cond)99 lp_build_flow_skip_cond_break(struct lp_build_skip_context *skip,
100 LLVMValueRef cond)
101 {
102 LLVMBasicBlockRef new_block;
103
104 new_block = lp_build_insert_new_block(skip->gallivm, "");
105
106 /* if cond is true, goto skip->block, else goto new_block */
107 LLVMBuildCondBr(skip->gallivm->builder, cond, skip->block, new_block);
108
109 LLVMPositionBuilderAtEnd(skip->gallivm->builder, new_block);
110 }
111
112
113 void
lp_build_flow_skip_end(struct lp_build_skip_context * skip)114 lp_build_flow_skip_end(struct lp_build_skip_context *skip)
115 {
116 /* goto block */
117 LLVMBuildBr(skip->gallivm->builder, skip->block);
118 LLVMPositionBuilderAtEnd(skip->gallivm->builder, skip->block);
119 }
120
121
122 /**
123 * Check if the mask predicate is zero. If so, jump to the end of the block.
124 */
125 void
lp_build_mask_check(struct lp_build_mask_context * mask)126 lp_build_mask_check(struct lp_build_mask_context *mask)
127 {
128 LLVMBuilderRef builder = mask->skip.gallivm->builder;
129 LLVMValueRef value;
130 LLVMValueRef cond;
131
132 value = lp_build_mask_value(mask);
133
134 /*
135 * XXX this doesn't quite generate the most efficient code possible, if
136 * the masks are vectors which have all bits set to the same value
137 * in each element.
138 * movmskps/pmovmskb would be more efficient to get the required value
139 * into ordinary reg (certainly with 8 floats).
140 * Not sure if llvm could figure that out on its own.
141 */
142
143 /* cond = (mask == 0) */
144 cond = LLVMBuildICmp(builder,
145 LLVMIntEQ,
146 LLVMBuildBitCast(builder, value, mask->reg_type, ""),
147 LLVMConstNull(mask->reg_type),
148 "");
149
150 /* if cond, goto end of block */
151 lp_build_flow_skip_cond_break(&mask->skip, cond);
152 }
153
154
155 /**
156 * Begin a section of code which is predicated on a mask.
157 * \param mask the mask context, initialized here
158 * \param flow the flow context
159 * \param type the type of the mask
160 * \param value storage for the mask
161 */
162 void
lp_build_mask_begin(struct lp_build_mask_context * mask,struct gallivm_state * gallivm,struct lp_type type,LLVMValueRef value)163 lp_build_mask_begin(struct lp_build_mask_context *mask,
164 struct gallivm_state *gallivm,
165 struct lp_type type,
166 LLVMValueRef value)
167 {
168 memset(mask, 0, sizeof *mask);
169
170 mask->reg_type = LLVMIntTypeInContext(gallivm->context, type.width * type.length);
171 mask->var_type = lp_build_int_vec_type(gallivm, type);
172 mask->var = lp_build_alloca(gallivm,
173 mask->var_type,
174 "execution_mask");
175
176 LLVMBuildStore(gallivm->builder, value, mask->var);
177
178 lp_build_flow_skip_begin(&mask->skip, gallivm);
179 }
180
181
182 LLVMValueRef
lp_build_mask_value(struct lp_build_mask_context * mask)183 lp_build_mask_value(struct lp_build_mask_context *mask)
184 {
185 return LLVMBuildLoad2(mask->skip.gallivm->builder, mask->var_type, mask->var, "");
186 }
187
188
189 /**
190 * Update boolean mask with given value (bitwise AND).
191 * Typically used to update the quad's pixel alive/killed mask
192 * after depth testing, alpha testing, TGSI_OPCODE_KILL_IF, etc.
193 */
194 void
lp_build_mask_update(struct lp_build_mask_context * mask,LLVMValueRef value)195 lp_build_mask_update(struct lp_build_mask_context *mask,
196 LLVMValueRef value)
197 {
198 value = LLVMBuildAnd(mask->skip.gallivm->builder,
199 lp_build_mask_value(mask),
200 value, "");
201 LLVMBuildStore(mask->skip.gallivm->builder, value, mask->var);
202 }
203
204 /*
205 * Update boolean mask with given value.
206 * Used for per-sample shading to force per-sample execution masks.
207 */
208 void
lp_build_mask_force(struct lp_build_mask_context * mask,LLVMValueRef value)209 lp_build_mask_force(struct lp_build_mask_context *mask,
210 LLVMValueRef value)
211 {
212 LLVMBuildStore(mask->skip.gallivm->builder, value, mask->var);
213 }
214
215 /**
216 * End section of code which is predicated on a mask.
217 */
218 LLVMValueRef
lp_build_mask_end(struct lp_build_mask_context * mask)219 lp_build_mask_end(struct lp_build_mask_context *mask)
220 {
221 lp_build_flow_skip_end(&mask->skip);
222 return lp_build_mask_value(mask);
223 }
224
225
226
227 void
lp_build_loop_begin(struct lp_build_loop_state * state,struct gallivm_state * gallivm,LLVMValueRef start)228 lp_build_loop_begin(struct lp_build_loop_state *state,
229 struct gallivm_state *gallivm,
230 LLVMValueRef start)
231
232 {
233 LLVMBuilderRef builder = gallivm->builder;
234
235 state->block = lp_build_insert_new_block(gallivm, "loop_begin");
236
237 state->counter_type = LLVMTypeOf(start);
238 state->counter_var = lp_build_alloca(gallivm, state->counter_type, "loop_counter");
239 state->gallivm = gallivm;
240
241 LLVMBuildStore(builder, start, state->counter_var);
242
243 LLVMBuildBr(builder, state->block);
244
245 LLVMPositionBuilderAtEnd(builder, state->block);
246
247 state->counter = LLVMBuildLoad2(builder, state->counter_type, state->counter_var, "");
248 }
249
250
251 void
lp_build_loop_end_cond(struct lp_build_loop_state * state,LLVMValueRef end,LLVMValueRef step,LLVMIntPredicate llvm_cond)252 lp_build_loop_end_cond(struct lp_build_loop_state *state,
253 LLVMValueRef end,
254 LLVMValueRef step,
255 LLVMIntPredicate llvm_cond)
256 {
257 LLVMBuilderRef builder = state->gallivm->builder;
258 LLVMValueRef next;
259 LLVMValueRef cond;
260 LLVMBasicBlockRef after_block;
261
262 if (!step)
263 step = LLVMConstInt(LLVMTypeOf(end), 1, 0);
264
265 next = LLVMBuildAdd(builder, state->counter, step, "");
266
267 LLVMBuildStore(builder, next, state->counter_var);
268
269 cond = LLVMBuildICmp(builder, llvm_cond, next, end, "");
270
271 after_block = lp_build_insert_new_block(state->gallivm, "loop_end");
272
273 LLVMBuildCondBr(builder, cond, after_block, state->block);
274
275 LLVMPositionBuilderAtEnd(builder, after_block);
276
277 state->counter = LLVMBuildLoad2(builder, state->counter_type, state->counter_var, "");
278 }
279
280 void
lp_build_loop_force_set_counter(struct lp_build_loop_state * state,LLVMValueRef end)281 lp_build_loop_force_set_counter(struct lp_build_loop_state *state,
282 LLVMValueRef end)
283 {
284 LLVMBuilderRef builder = state->gallivm->builder;
285 LLVMBuildStore(builder, end, state->counter_var);
286 }
287
288 void
lp_build_loop_force_reload_counter(struct lp_build_loop_state * state)289 lp_build_loop_force_reload_counter(struct lp_build_loop_state *state)
290 {
291 LLVMBuilderRef builder = state->gallivm->builder;
292 state->counter = LLVMBuildLoad2(builder, state->counter_type, state->counter_var, "");
293 }
294
295 void
lp_build_loop_end(struct lp_build_loop_state * state,LLVMValueRef end,LLVMValueRef step)296 lp_build_loop_end(struct lp_build_loop_state *state,
297 LLVMValueRef end,
298 LLVMValueRef step)
299 {
300 lp_build_loop_end_cond(state, end, step, LLVMIntNE);
301 }
302
303 /**
304 * Creates a c-style for loop,
305 * contrasts lp_build_loop as this checks condition on entry
306 * e.g. for(i = start; i cmp_op end; i += step)
307 * \param state the for loop state, initialized here
308 * \param gallivm the gallivm state
309 * \param start starting value of iterator
310 * \param cmp_op comparison operator used for comparing current value with end value
311 * \param end value used to compare against iterator
312 * \param step value added to iterator at end of each loop
313 */
314 void
lp_build_for_loop_begin(struct lp_build_for_loop_state * state,struct gallivm_state * gallivm,LLVMValueRef start,LLVMIntPredicate cmp_op,LLVMValueRef end,LLVMValueRef step)315 lp_build_for_loop_begin(struct lp_build_for_loop_state *state,
316 struct gallivm_state *gallivm,
317 LLVMValueRef start,
318 LLVMIntPredicate cmp_op,
319 LLVMValueRef end,
320 LLVMValueRef step)
321 {
322 LLVMBuilderRef builder = gallivm->builder;
323
324 assert(LLVMTypeOf(start) == LLVMTypeOf(end));
325 assert(LLVMTypeOf(start) == LLVMTypeOf(step));
326
327 state->begin = lp_build_insert_new_block(gallivm, "loop_begin");
328 state->step = step;
329 state->counter_type = LLVMTypeOf(start);
330 state->counter_var = lp_build_alloca(gallivm, state->counter_type, "loop_counter");
331 state->gallivm = gallivm;
332 state->cond = cmp_op;
333 state->end = end;
334
335 LLVMBuildStore(builder, start, state->counter_var);
336 LLVMBuildBr(builder, state->begin);
337
338 LLVMPositionBuilderAtEnd(builder, state->begin);
339 state->counter = LLVMBuildLoad2(builder, state->counter_type, state->counter_var, "");
340
341 state->body = lp_build_insert_new_block(gallivm, "loop_body");
342 LLVMPositionBuilderAtEnd(builder, state->body);
343 }
344
345 /**
346 * End the for loop.
347 */
348 void
lp_build_for_loop_end(struct lp_build_for_loop_state * state)349 lp_build_for_loop_end(struct lp_build_for_loop_state *state)
350 {
351 LLVMValueRef next, cond;
352 LLVMBuilderRef builder = state->gallivm->builder;
353
354 next = LLVMBuildAdd(builder, state->counter, state->step, "");
355 LLVMBuildStore(builder, next, state->counter_var);
356 LLVMBuildBr(builder, state->begin);
357
358 state->exit = lp_build_insert_new_block(state->gallivm, "loop_exit");
359
360 /*
361 * We build the comparison for the begin block here,
362 * if we build it earlier the output llvm ir is not human readable
363 * as the code produced is not in the standard begin -> body -> end order.
364 */
365 LLVMPositionBuilderAtEnd(builder, state->begin);
366 cond = LLVMBuildICmp(builder, state->cond, state->counter, state->end, "");
367 LLVMBuildCondBr(builder, cond, state->body, state->exit);
368
369 LLVMPositionBuilderAtEnd(builder, state->exit);
370 }
371
372
373 /*
374 Example of if/then/else building:
375
376 int x;
377 if (cond) {
378 x = 1 + 2;
379 }
380 else {
381 x = 2 + 3;
382 }
383
384 Is built with:
385
386 // x needs an alloca variable
387 x = lp_build_alloca(builder, type, "x");
388
389
390 lp_build_if(ctx, builder, cond);
391 LLVMBuildStore(LLVMBuildAdd(1, 2), x);
392 lp_build_else(ctx);
393 LLVMBuildStore(LLVMBuildAdd(2, 3). x);
394 lp_build_endif(ctx);
395
396 */
397
398
399
400 /**
401 * Begin an if/else/endif construct.
402 */
403 void
lp_build_if(struct lp_build_if_state * ifthen,struct gallivm_state * gallivm,LLVMValueRef condition)404 lp_build_if(struct lp_build_if_state *ifthen,
405 struct gallivm_state *gallivm,
406 LLVMValueRef condition)
407 {
408 LLVMBasicBlockRef block = LLVMGetInsertBlock(gallivm->builder);
409
410 memset(ifthen, 0, sizeof *ifthen);
411 ifthen->gallivm = gallivm;
412 ifthen->condition = condition;
413 ifthen->entry_block = block;
414
415 /* create endif/merge basic block for the phi functions */
416 ifthen->merge_block = lp_build_insert_new_block(gallivm, "endif-block");
417
418 /* create/insert true_block before merge_block */
419 ifthen->true_block =
420 LLVMInsertBasicBlockInContext(gallivm->context,
421 ifthen->merge_block,
422 "if-true-block");
423
424 /* successive code goes into the true block */
425 LLVMPositionBuilderAtEnd(gallivm->builder, ifthen->true_block);
426 }
427
428
429 /**
430 * Begin else-part of a conditional
431 */
432 void
lp_build_else(struct lp_build_if_state * ifthen)433 lp_build_else(struct lp_build_if_state *ifthen)
434 {
435 LLVMBuilderRef builder = ifthen->gallivm->builder;
436
437 /* Append an unconditional Br(anch) instruction on the true_block */
438 LLVMBuildBr(builder, ifthen->merge_block);
439
440 /* create/insert false_block before the merge block */
441 ifthen->false_block =
442 LLVMInsertBasicBlockInContext(ifthen->gallivm->context,
443 ifthen->merge_block,
444 "if-false-block");
445
446 /* successive code goes into the else block */
447 LLVMPositionBuilderAtEnd(builder, ifthen->false_block);
448 }
449
450
451 /**
452 * End a conditional.
453 */
454 void
lp_build_endif(struct lp_build_if_state * ifthen)455 lp_build_endif(struct lp_build_if_state *ifthen)
456 {
457 LLVMBuilderRef builder = ifthen->gallivm->builder;
458
459 /* Insert branch to the merge block from current block */
460 LLVMBuildBr(builder, ifthen->merge_block);
461
462 /*
463 * Now patch in the various branch instructions.
464 */
465
466 /* Insert the conditional branch instruction at the end of entry_block */
467 LLVMPositionBuilderAtEnd(builder, ifthen->entry_block);
468 if (ifthen->false_block) {
469 /* we have an else clause */
470 LLVMBuildCondBr(builder, ifthen->condition,
471 ifthen->true_block, ifthen->false_block);
472 }
473 else {
474 /* no else clause */
475 LLVMBuildCondBr(builder, ifthen->condition,
476 ifthen->true_block, ifthen->merge_block);
477 }
478
479 /* Resume building code at end of the ifthen->merge_block */
480 LLVMPositionBuilderAtEnd(builder, ifthen->merge_block);
481 }
482
483
484 static LLVMBuilderRef
create_builder_at_entry(struct gallivm_state * gallivm)485 create_builder_at_entry(struct gallivm_state *gallivm)
486 {
487 LLVMBuilderRef builder = gallivm->builder;
488 LLVMBasicBlockRef current_block = LLVMGetInsertBlock(builder);
489 LLVMValueRef function = LLVMGetBasicBlockParent(current_block);
490 LLVMBasicBlockRef first_block = LLVMGetEntryBasicBlock(function);
491 LLVMValueRef first_instr = LLVMGetFirstInstruction(first_block);
492 LLVMBuilderRef first_builder = LLVMCreateBuilderInContext(gallivm->context);
493
494 if (first_instr) {
495 LLVMPositionBuilderBefore(first_builder, first_instr);
496 } else {
497 LLVMPositionBuilderAtEnd(first_builder, first_block);
498 }
499
500 return first_builder;
501 }
502
503
504 /**
505 * Allocate a scalar (or vector) variable.
506 *
507 * Although not strictly part of control flow, control flow has deep impact in
508 * how variables should be allocated.
509 *
510 * The mem2reg optimization pass is the recommended way to dealing with mutable
511 * variables, and SSA. It looks for allocas and if it can handle them, it
512 * promotes them, but only looks for alloca instructions in the entry block of
513 * the function. Being in the entry block guarantees that the alloca is only
514 * executed once, which makes analysis simpler.
515 *
516 * See also:
517 * - http://www.llvm.org/docs/tutorial/OCamlLangImpl7.html#memory
518 */
519 LLVMValueRef
lp_build_alloca(struct gallivm_state * gallivm,LLVMTypeRef type,const char * name)520 lp_build_alloca(struct gallivm_state *gallivm,
521 LLVMTypeRef type,
522 const char *name)
523 {
524 LLVMBuilderRef builder = gallivm->builder;
525 LLVMBuilderRef first_builder = create_builder_at_entry(gallivm);
526 LLVMValueRef res;
527
528 res = LLVMBuildAlloca(first_builder, type, name);
529 LLVMBuildStore(builder, LLVMConstNull(type), res);
530
531 LLVMDisposeBuilder(first_builder);
532
533 return res;
534 }
535
536
537 /**
538 * Like lp_build_alloca, but do not zero-initialize the variable.
539 */
540 LLVMValueRef
lp_build_alloca_undef(struct gallivm_state * gallivm,LLVMTypeRef type,const char * name)541 lp_build_alloca_undef(struct gallivm_state *gallivm,
542 LLVMTypeRef type,
543 const char *name)
544 {
545 LLVMBuilderRef first_builder = create_builder_at_entry(gallivm);
546 LLVMValueRef res;
547
548 res = LLVMBuildAlloca(first_builder, type, name);
549
550 LLVMDisposeBuilder(first_builder);
551
552 return res;
553 }
554
555
556 /**
557 * Allocate an array of scalars/vectors.
558 *
559 * mem2reg pass is not capable of promoting structs or arrays to registers, but
560 * we still put it in the first block anyway as failure to put allocas in the
561 * first block may prevent the X86 backend from successfully align the stack as
562 * required.
563 *
564 * Also the scalarrepl pass is supposedly more powerful and can promote
565 * arrays in many cases.
566 *
567 * See also:
568 * - http://www.llvm.org/docs/tutorial/OCamlLangImpl7.html#memory
569 */
570 LLVMValueRef
lp_build_array_alloca(struct gallivm_state * gallivm,LLVMTypeRef type,LLVMValueRef count,const char * name)571 lp_build_array_alloca(struct gallivm_state *gallivm,
572 LLVMTypeRef type,
573 LLVMValueRef count,
574 const char *name)
575 {
576 LLVMBuilderRef first_builder = create_builder_at_entry(gallivm);
577 LLVMValueRef res;
578
579 res = LLVMBuildArrayAlloca(first_builder, type, count, name);
580
581 LLVMDisposeBuilder(first_builder);
582
583 return res;
584 }
585