2 * Copyright (C) 2012, 2013
5 * Permission is hereby granted, free of charge, to any person obtaining a copy of
6 * this software and associated documentation files (the "Software"), to deal in
7 * the Software without restriction, including without limitation the rights to
8 * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
9 * of the Software, and to permit persons to whom the Software is furnished to do
10 * so, subject to the following conditions:
12 * The above copyright notice and this permission notice shall be included in all
13 * copies or substantial portions of the Software.
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 THE
18 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
29 #define FOLD_STRING_UNTRANSLATE_HTSIZE 1024
30 #define FOLD_STRING_DOTRANSLATE_HTSIZE 1024
33 * There is two stages to constant folding in GMQCC: there is the parse
34 * stage constant folding, where, witht he help of the AST, operator
35 * usages can be constant folded. Then there is the constant folding
36 * in the IR for things like eliding if statements, can occur.
38 * This file is thus, split into two parts.
41 #define isfloat(X) (((ast_expression*)(X))->vtype == TYPE_FLOAT)
42 #define isvector(X) (((ast_expression*)(X))->vtype == TYPE_VECTOR)
43 #define isstring(X) (((ast_expression*)(X))->vtype == TYPE_STRING)
44 #define isfloats(X,Y) (isfloat (X) && isfloat (Y))
47 * Implementation of basic vector math for vec3_t, for trivial constant
50 * TODO: gcc/clang hinting for autovectorization
52 static GMQCC_INLINE vec3_t vec3_add(vec3_t a, vec3_t b) {
60 static GMQCC_INLINE vec3_t vec3_sub(vec3_t a, vec3_t b) {
68 static GMQCC_INLINE vec3_t vec3_neg(vec3_t a) {
76 static GMQCC_INLINE vec3_t vec3_or(vec3_t a, vec3_t b) {
78 out.x = (qcfloat_t)(((qcint_t)a.x) | ((qcint_t)b.x));
79 out.y = (qcfloat_t)(((qcint_t)a.y) | ((qcint_t)b.y));
80 out.z = (qcfloat_t)(((qcint_t)a.z) | ((qcint_t)b.z));
84 static GMQCC_INLINE vec3_t vec3_orvf(vec3_t a, qcfloat_t b) {
86 out.x = (qcfloat_t)(((qcint_t)a.x) | ((qcint_t)b));
87 out.y = (qcfloat_t)(((qcint_t)a.y) | ((qcint_t)b));
88 out.z = (qcfloat_t)(((qcint_t)a.z) | ((qcint_t)b));
92 static GMQCC_INLINE vec3_t vec3_and(vec3_t a, vec3_t b) {
94 out.x = (qcfloat_t)(((qcint_t)a.x) & ((qcint_t)b.x));
95 out.y = (qcfloat_t)(((qcint_t)a.y) & ((qcint_t)b.y));
96 out.z = (qcfloat_t)(((qcint_t)a.z) & ((qcint_t)b.z));
100 static GMQCC_INLINE vec3_t vec3_andvf(vec3_t a, qcfloat_t b) {
102 out.x = (qcfloat_t)(((qcint_t)a.x) & ((qcint_t)b));
103 out.y = (qcfloat_t)(((qcint_t)a.y) & ((qcint_t)b));
104 out.z = (qcfloat_t)(((qcint_t)a.z) & ((qcint_t)b));
108 static GMQCC_INLINE vec3_t vec3_xor(vec3_t a, vec3_t b) {
110 out.x = (qcfloat_t)(((qcint_t)a.x) ^ ((qcint_t)b.x));
111 out.y = (qcfloat_t)(((qcint_t)a.y) ^ ((qcint_t)b.y));
112 out.z = (qcfloat_t)(((qcint_t)a.z) ^ ((qcint_t)b.z));
116 static GMQCC_INLINE vec3_t vec3_xorvf(vec3_t a, qcfloat_t b) {
118 out.x = (qcfloat_t)(((qcint_t)a.x) ^ ((qcint_t)b));
119 out.y = (qcfloat_t)(((qcint_t)a.y) ^ ((qcint_t)b));
120 out.z = (qcfloat_t)(((qcint_t)a.z) ^ ((qcint_t)b));
124 static GMQCC_INLINE vec3_t vec3_not(vec3_t a) {
126 out.x = (qcfloat_t)(~((qcint_t)a.x));
127 out.y = (qcfloat_t)(~((qcint_t)a.y));
128 out.z = (qcfloat_t)(~((qcint_t)a.z));
132 static GMQCC_INLINE qcfloat_t vec3_mulvv(vec3_t a, vec3_t b) {
133 return (a.x * b.x + a.y * b.y + a.z * b.z);
136 static GMQCC_INLINE vec3_t vec3_mulvf(vec3_t a, qcfloat_t b) {
144 static GMQCC_INLINE bool vec3_cmp(vec3_t a, vec3_t b) {
150 static GMQCC_INLINE vec3_t vec3_create(float x, float y, float z) {
158 static GMQCC_INLINE qcfloat_t vec3_notf(vec3_t a) {
159 return (!a.x && !a.y && !a.z);
162 static GMQCC_INLINE bool vec3_pbool(vec3_t a) {
163 return (a.x && a.y && a.z);
166 static GMQCC_INLINE vec3_t vec3_cross(vec3_t a, vec3_t b) {
168 out.x = a.y * b.z - a.z * b.y;
169 out.y = a.z * b.x - a.x * b.z;
170 out.z = a.x * b.y - a.y * b.x;
174 static lex_ctx_t fold_ctx(fold_t *fold) {
176 if (fold->parser->lex)
177 return parser_ctx(fold->parser);
179 memset(&ctx, 0, sizeof(ctx));
183 static GMQCC_INLINE bool fold_immediate_true(fold_t *fold, ast_value *v) {
184 switch (v->expression.vtype) {
186 return !!v->constval.vfloat;
188 return !!v->constval.vint;
190 if (OPTS_FLAG(CORRECT_LOGIC))
191 return vec3_pbool(v->constval.vvec);
192 return !!(v->constval.vvec.x);
194 if (!v->constval.vstring)
196 if (OPTS_FLAG(TRUE_EMPTY_STRINGS))
198 return !!v->constval.vstring[0];
200 compile_error(fold_ctx(fold), "internal error: fold_immediate_true on invalid type");
203 return !!v->constval.vfunc;
206 /* Handy macros to determine if an ast_value can be constant folded. */
207 #define fold_can_1(X) \
208 (ast_istype(((ast_expression*)(X)), ast_value) && (X)->hasvalue && ((X)->cvq == CV_CONST) && \
209 ((ast_expression*)(X))->vtype != TYPE_FUNCTION)
211 #define fold_can_2(X, Y) (fold_can_1(X) && fold_can_1(Y))
212 #define fold_can_div(X) (fold_immvalue_float(X) != 0.0f)
214 #define fold_immvalue_float(E) ((E)->constval.vfloat)
215 #define fold_immvalue_vector(E) ((E)->constval.vvec)
216 #define fold_immvalue_string(E) ((E)->constval.vstring)
219 # define fold_infinity_float INFINITY
221 # define fold_infinity_float (1.0 / 0.0)
222 #endif /*! INFINITY */
224 #define fold_infinity_vector \
226 fold_infinity_float, \
227 fold_infinity_float, \
228 fold_infinity_float \
231 fold_t *fold_init(parser_t *parser) {
232 fold_t *fold = (fold_t*)mem_a(sizeof(fold_t));
233 fold->parser = parser;
234 fold->imm_float = NULL;
235 fold->imm_vector = NULL;
236 fold->imm_string = NULL;
237 fold->imm_string_untranslate = util_htnew(FOLD_STRING_UNTRANSLATE_HTSIZE);
238 fold->imm_string_dotranslate = util_htnew(FOLD_STRING_DOTRANSLATE_HTSIZE);
241 * prime the tables with common constant values at constant
244 (void)fold_constgen_float (fold, 0.0f);
245 (void)fold_constgen_float (fold, 1.0f);
246 (void)fold_constgen_float (fold, -1.0f);
247 (void)fold_constgen_float (fold, fold_infinity_float); /* +inf */
249 (void)fold_constgen_vector(fold, vec3_create(0.0f, 0.0f, 0.0f));
250 (void)fold_constgen_vector(fold, vec3_create(-1.0f, -1.0f, -1.0f));
251 (void)fold_constgen_vector(fold, fold_infinity_vector); /* +inf */
256 bool fold_generate(fold_t *fold, ir_builder *ir) {
257 /* generate globals for immediate folded values */
261 for (i = 0; i < vec_size(fold->imm_float); ++i)
262 if (!ast_global_codegen ((cur = fold->imm_float[i]), ir, false)) goto err;
263 for (i = 0; i < vec_size(fold->imm_vector); ++i)
264 if (!ast_global_codegen((cur = fold->imm_vector[i]), ir, false)) goto err;
265 for (i = 0; i < vec_size(fold->imm_string); ++i)
266 if (!ast_global_codegen((cur = fold->imm_string[i]), ir, false)) goto err;
271 con_out("failed to generate global %s\n", cur->name);
272 ir_builder_delete(ir);
276 void fold_cleanup(fold_t *fold) {
279 for (i = 0; i < vec_size(fold->imm_float); ++i) ast_delete(fold->imm_float[i]);
280 for (i = 0; i < vec_size(fold->imm_vector); ++i) ast_delete(fold->imm_vector[i]);
281 for (i = 0; i < vec_size(fold->imm_string); ++i) ast_delete(fold->imm_string[i]);
283 vec_free(fold->imm_float);
284 vec_free(fold->imm_vector);
285 vec_free(fold->imm_string);
287 util_htdel(fold->imm_string_untranslate);
288 util_htdel(fold->imm_string_dotranslate);
293 ast_expression *fold_constgen_float(fold_t *fold, qcfloat_t value) {
294 ast_value *out = NULL;
297 for (i = 0; i < vec_size(fold->imm_float); i++) {
298 if (fold->imm_float[i]->constval.vfloat == value)
299 return (ast_expression*)fold->imm_float[i];
302 out = ast_value_new(fold_ctx(fold), "#IMMEDIATE", TYPE_FLOAT);
304 out->hasvalue = true;
305 out->constval.vfloat = value;
307 vec_push(fold->imm_float, out);
309 return (ast_expression*)out;
312 ast_expression *fold_constgen_vector(fold_t *fold, vec3_t value) {
316 for (i = 0; i < vec_size(fold->imm_vector); i++) {
317 if (vec3_cmp(fold->imm_vector[i]->constval.vvec, value))
318 return (ast_expression*)fold->imm_vector[i];
321 out = ast_value_new(fold_ctx(fold), "#IMMEDIATE", TYPE_VECTOR);
323 out->hasvalue = true;
324 out->constval.vvec = value;
326 vec_push(fold->imm_vector, out);
328 return (ast_expression*)out;
331 ast_expression *fold_constgen_string(fold_t *fold, const char *str, bool translate) {
332 hash_table_t *table = (translate) ? fold->imm_string_untranslate : fold->imm_string_dotranslate;
333 ast_value *out = NULL;
334 size_t hash = util_hthash(table, str);
336 if ((out = (ast_value*)util_htgeth(table, str, hash)))
337 return (ast_expression*)out;
341 util_snprintf(name, sizeof(name), "dotranslate_%lu", (unsigned long)(fold->parser->translated++));
342 out = ast_value_new(parser_ctx(fold->parser), name, TYPE_STRING);
343 out->expression.flags |= AST_FLAG_INCLUDE_DEF; /* def needs to be included for translatables */
345 out = ast_value_new(fold_ctx(fold), "#IMMEDIATE", TYPE_STRING);
348 out->hasvalue = true;
350 out->constval.vstring = parser_strdup(str);
352 vec_push(fold->imm_string, out);
353 util_htseth(table, str, hash, out);
355 return (ast_expression*)out;
359 static GMQCC_INLINE ast_expression *fold_op_mul_vec(fold_t *fold, vec3_t vec, ast_value *sel, const char *set) {
361 * vector-component constant folding works by matching the component sets
362 * to eliminate expensive operations on whole-vectors (3 components at runtime).
363 * to achive this effect in a clean manner this function generalizes the
364 * values through the use of a set paramater, which is used as an indexing method
365 * for creating the elided ast binary expression.
367 * Consider 'n 0 0' where y, and z need to be tested for 0, and x is
368 * used as the value in a binary operation generating an INSTR_MUL instruction,
369 * to acomplish the indexing of the correct component value we use set[0], set[1], set[2]
370 * as x, y, z, where the values of those operations return 'x', 'y', 'z'. Because
371 * of how ASCII works we can easily deliniate:
372 * vec.z is the same as set[2]-'x' for when set[2] is 'z', 'z'-'x' results in a
373 * literal value of 2, using this 2, we know that taking the address of vec->x (float)
374 * and indxing it with this literal will yeild the immediate address of that component
376 * Of course more work needs to be done to generate the correct index for the ast_member_new
377 * call, which is no problem: set[0]-'x' suffices that job.
379 qcfloat_t x = (&vec.x)[set[0]-'x'];
380 qcfloat_t y = (&vec.x)[set[1]-'x'];
381 qcfloat_t z = (&vec.x)[set[2]-'x'];
385 ++opts_optimizationcount[OPTIM_VECTOR_COMPONENTS];
386 out = (ast_expression*)ast_member_new(fold_ctx(fold), (ast_expression*)sel, set[0]-'x', NULL);
387 out->node.keep = false;
388 ((ast_member*)out)->rvalue = true;
390 return (ast_expression*)ast_binary_new(fold_ctx(fold), INSTR_MUL_F, fold_constgen_float(fold, x), out);
396 static GMQCC_INLINE ast_expression *fold_op_neg(fold_t *fold, ast_value *a) {
399 return fold_constgen_float(fold, -fold_immvalue_float(a));
400 } else if (isvector(a)) {
402 return fold_constgen_vector(fold, vec3_neg(fold_immvalue_vector(a)));
407 static GMQCC_INLINE ast_expression *fold_op_not(fold_t *fold, ast_value *a) {
410 return fold_constgen_float(fold, !fold_immvalue_float(a));
411 } else if (isvector(a)) {
413 return fold_constgen_float(fold, vec3_notf(fold_immvalue_vector(a)));
414 } else if (isstring(a)) {
416 if (OPTS_FLAG(TRUE_EMPTY_STRINGS))
417 return fold_constgen_float(fold, !fold_immvalue_string(a));
419 return fold_constgen_float(fold, !fold_immvalue_string(a) || !*fold_immvalue_string(a));
425 static GMQCC_INLINE ast_expression *fold_op_add(fold_t *fold, ast_value *a, ast_value *b) {
427 if (fold_can_2(a, b))
428 return fold_constgen_float(fold, fold_immvalue_float(a) + fold_immvalue_float(b));
429 } else if (isvector(a)) {
430 if (fold_can_2(a, b))
431 return fold_constgen_vector(fold, vec3_add(fold_immvalue_vector(a), fold_immvalue_vector(b)));
436 static GMQCC_INLINE ast_expression *fold_op_sub(fold_t *fold, ast_value *a, ast_value *b) {
438 if (fold_can_2(a, b))
439 return fold_constgen_float(fold, fold_immvalue_float(a) - fold_immvalue_float(b));
440 } else if (isvector(a)) {
441 if (fold_can_2(a, b))
442 return fold_constgen_vector(fold, vec3_sub(fold_immvalue_vector(a), fold_immvalue_vector(b)));
447 static GMQCC_INLINE ast_expression *fold_op_mul(fold_t *fold, ast_value *a, ast_value *b) {
450 if (fold_can_2(a, b))
451 return fold_constgen_vector(fold, vec3_mulvf(fold_immvalue_vector(b), fold_immvalue_float(a)));
453 if (fold_can_2(a, b))
454 return fold_constgen_float(fold, fold_immvalue_float(a) * fold_immvalue_float(b));
456 } else if (isvector(a)) {
458 if (fold_can_2(a, b))
459 return fold_constgen_vector(fold, vec3_mulvf(fold_immvalue_vector(a), fold_immvalue_float(b)));
461 if (fold_can_2(a, b)) {
462 return fold_constgen_float(fold, vec3_mulvv(fold_immvalue_vector(a), fold_immvalue_vector(b)));
463 } else if (OPTS_OPTIMIZATION(OPTIM_VECTOR_COMPONENTS) && fold_can_1(a)) {
465 if ((out = fold_op_mul_vec(fold, fold_immvalue_vector(a), b, "xyz"))) return out;
466 if ((out = fold_op_mul_vec(fold, fold_immvalue_vector(a), b, "yxz"))) return out;
467 if ((out = fold_op_mul_vec(fold, fold_immvalue_vector(a), b, "zxy"))) return out;
468 } else if (OPTS_OPTIMIZATION(OPTIM_VECTOR_COMPONENTS) && fold_can_1(b)) {
470 if ((out = fold_op_mul_vec(fold, fold_immvalue_vector(b), a, "xyz"))) return out;
471 if ((out = fold_op_mul_vec(fold, fold_immvalue_vector(b), a, "yxz"))) return out;
472 if ((out = fold_op_mul_vec(fold, fold_immvalue_vector(b), a, "zxy"))) return out;
479 static GMQCC_INLINE ast_expression *fold_op_div(fold_t *fold, ast_value *a, ast_value *b) {
481 if (fold_can_2(a, b)) {
483 return fold_constgen_float(fold, fold_immvalue_float(a) / fold_immvalue_float(b));
485 return (ast_expression*)fold->imm_float[3]; /* inf */
486 } else if (fold_can_1(b)) {
487 return (ast_expression*)ast_binary_new(
491 fold_constgen_float(fold, 1.0f / fold_immvalue_float(b))
494 } else if (isvector(a)) {
495 if (fold_can_2(a, b)) {
496 if (fold_can_div(b)) {
497 return fold_constgen_vector(fold, vec3_mulvf(fold_immvalue_vector(a), 1.0f / fold_immvalue_float(b)));
500 return (ast_expression*)fold->imm_vector[2]; /* inf */
503 return (ast_expression*)ast_binary_new(
508 ? (ast_expression*)fold_constgen_float(fold, 1.0f / fold_immvalue_float(b))
509 : (ast_expression*)ast_binary_new(
512 (ast_expression*)fold->imm_float[1],
521 static GMQCC_INLINE ast_expression *fold_op_mod(fold_t *fold, ast_value *a, ast_value *b) {
522 if (fold_can_2(a, b)) {
524 return fold_constgen_float(fold, (qcfloat_t)(((qcint_t)fold_immvalue_float(a)) % ((qcint_t)fold_immvalue_float(b))));
526 return (ast_expression*)fold->imm_float[3]; /* inf */
531 static GMQCC_INLINE ast_expression *fold_op_bor(fold_t *fold, ast_value *a, ast_value *b) {
533 if (fold_can_2(a, b))
534 return fold_constgen_float(fold, (qcfloat_t)(((qcint_t)fold_immvalue_float(a)) | ((qcint_t)fold_immvalue_float(b))));
537 if (fold_can_2(a, b))
538 return fold_constgen_vector(fold, vec3_or(fold_immvalue_vector(a), fold_immvalue_vector(b)));
540 if (fold_can_2(a, b))
541 return fold_constgen_vector(fold, vec3_orvf(fold_immvalue_vector(a), fold_immvalue_float(b)));
547 static GMQCC_INLINE ast_expression *fold_op_band(fold_t *fold, ast_value *a, ast_value *b) {
549 if (fold_can_2(a, b))
550 return fold_constgen_float(fold, (qcfloat_t)(((qcint_t)fold_immvalue_float(a)) & ((qcint_t)fold_immvalue_float(b))));
553 if (fold_can_2(a, b))
554 return fold_constgen_vector(fold, vec3_and(fold_immvalue_vector(a), fold_immvalue_vector(b)));
556 if (fold_can_2(a, b))
557 return fold_constgen_vector(fold, vec3_andvf(fold_immvalue_vector(a), fold_immvalue_float(b)));
563 static GMQCC_INLINE ast_expression *fold_op_xor(fold_t *fold, ast_value *a, ast_value *b) {
565 if (fold_can_2(a, b))
566 return fold_constgen_float(fold, (qcfloat_t)(((qcint_t)fold_immvalue_float(a)) ^ ((qcint_t)fold_immvalue_float(b))));
569 if (fold_can_2(a, b))
570 return fold_constgen_vector(fold, vec3_xor(fold_immvalue_vector(a), fold_immvalue_vector(b)));
572 if (fold_can_2(a, b))
573 return fold_constgen_vector(fold, vec3_xorvf(fold_immvalue_vector(a), fold_immvalue_float(b)));
579 static GMQCC_INLINE ast_expression *fold_op_lshift(fold_t *fold, ast_value *a, ast_value *b) {
580 if (fold_can_2(a, b) && isfloats(a, b))
581 return fold_constgen_float(fold, (qcfloat_t)((qcuint_t)(fold_immvalue_float(a)) << (qcuint_t)(fold_immvalue_float(b))));
585 static GMQCC_INLINE ast_expression *fold_op_rshift(fold_t *fold, ast_value *a, ast_value *b) {
586 if (fold_can_2(a, b) && isfloats(a, b))
587 return fold_constgen_float(fold, (qcfloat_t)((qcuint_t)(fold_immvalue_float(a)) >> (qcuint_t)(fold_immvalue_float(b))));
591 static GMQCC_INLINE ast_expression *fold_op_andor(fold_t *fold, ast_value *a, ast_value *b, float expr) {
592 if (fold_can_2(a, b)) {
593 if (OPTS_FLAG(PERL_LOGIC)) {
594 if (fold_immediate_true(fold, a))
595 return (ast_expression*)b;
597 return fold_constgen_float (
599 ((expr) ? (fold_immediate_true(fold, a) || fold_immediate_true(fold, b))
600 : (fold_immediate_true(fold, a) && fold_immediate_true(fold, b)))
609 static GMQCC_INLINE ast_expression *fold_op_tern(fold_t *fold, ast_value *a, ast_value *b, ast_value *c) {
611 return fold_immediate_true(fold, a)
613 : (ast_expression*)c;
618 static GMQCC_INLINE ast_expression *fold_op_exp(fold_t *fold, ast_value *a, ast_value *b) {
619 if (fold_can_2(a, b))
620 return fold_constgen_float(fold, (qcfloat_t)powf(fold_immvalue_float(a), fold_immvalue_float(b)));
624 static GMQCC_INLINE ast_expression *fold_op_lteqgt(fold_t *fold, ast_value *a, ast_value *b) {
625 if (fold_can_2(a,b)) {
626 if (fold_immvalue_float(a) < fold_immvalue_float(b)) return (ast_expression*)fold->imm_float[2];
627 if (fold_immvalue_float(a) == fold_immvalue_float(b)) return (ast_expression*)fold->imm_float[0];
628 if (fold_immvalue_float(a) > fold_immvalue_float(b)) return (ast_expression*)fold->imm_float[1];
633 static GMQCC_INLINE ast_expression *fold_op_cmp(fold_t *fold, ast_value *a, ast_value *b, bool ne) {
634 if (fold_can_2(a, b)) {
635 return fold_constgen_float(
637 (ne) ? (fold_immvalue_float(a) != fold_immvalue_float(b))
638 : (fold_immvalue_float(a) == fold_immvalue_float(b))
644 static GMQCC_INLINE ast_expression *fold_op_bnot(fold_t *fold, ast_value *a) {
647 return fold_constgen_float(fold, ~((qcint_t)fold_immvalue_float(a)));
651 return fold_constgen_vector(fold, vec3_not(fold_immvalue_vector(a)));
657 static GMQCC_INLINE ast_expression *fold_op_cross(fold_t *fold, ast_value *a, ast_value *b) {
658 if (fold_can_2(a, b))
659 return fold_constgen_vector(fold, vec3_cross(fold_immvalue_vector(a), fold_immvalue_vector(b)));
663 ast_expression *fold_op(fold_t *fold, const oper_info *info, ast_expression **opexprs) {
664 ast_value *a = (ast_value*)opexprs[0];
665 ast_value *b = (ast_value*)opexprs[1];
666 ast_value *c = (ast_value*)opexprs[2];
667 ast_expression *e = NULL;
669 /* can a fold operation be applied to this operator usage? */
673 switch(info->operands) {
674 case 3: if(!c) return NULL;
675 case 2: if(!b) return NULL;
678 compile_error(fold_ctx(fold), "internal error: fold_op no operands to fold\n");
684 * we could use a boolean and default case but ironically gcc produces
685 * invalid broken assembly from that operation. clang/tcc get it right,
686 * but interestingly ignore compiling this to a jump-table when I do that,
687 * this happens to be the most efficent method, since you have per-level
688 * granularity on the pointer check happening only for the case you check
689 * it in. Opposed to the default method which would involve a boolean and
690 * pointer check after wards.
692 #define fold_op_case(ARGS, ARGS_OPID, OP, ARGS_FOLD) \
693 case opid##ARGS ARGS_OPID: \
694 if ((e = fold_op_##OP ARGS_FOLD)) { \
695 ++opts_optimizationcount[OPTIM_CONST_FOLD]; \
700 fold_op_case(2, ('-', 'P'), neg, (fold, a));
701 fold_op_case(2, ('!', 'P'), not, (fold, a));
702 fold_op_case(1, ('+'), add, (fold, a, b));
703 fold_op_case(1, ('-'), sub, (fold, a, b));
704 fold_op_case(1, ('*'), mul, (fold, a, b));
705 fold_op_case(1, ('/'), div, (fold, a, b));
706 fold_op_case(1, ('%'), mod, (fold, a, b));
707 fold_op_case(1, ('|'), bor, (fold, a, b));
708 fold_op_case(1, ('&'), band, (fold, a, b));
709 fold_op_case(1, ('^'), xor, (fold, a, b));
710 fold_op_case(2, ('<', '<'), lshift, (fold, a, b));
711 fold_op_case(2, ('>', '>'), rshift, (fold, a, b));
712 fold_op_case(2, ('|', '|'), andor, (fold, a, b, true));
713 fold_op_case(2, ('&', '&'), andor, (fold, a, b, false));
714 fold_op_case(2, ('?', ':'), tern, (fold, a, b, c));
715 fold_op_case(2, ('*', '*'), exp, (fold, a, b));
716 fold_op_case(3, ('<','=','>'), lteqgt, (fold, a, b));
717 fold_op_case(2, ('!', '='), cmp, (fold, a, b, true));
718 fold_op_case(2, ('=', '='), cmp, (fold, a, b, false));
719 fold_op_case(2, ('~', 'P'), bnot, (fold, a));
720 fold_op_case(2, ('>', '<'), cross, (fold, a, b));
723 compile_error(fold_ctx(fold), "internal error: attempted to constant-fold for unsupported operator");
728 * Constant folding for compiler intrinsics, simaler approach to operator
729 * folding, primarly: individual functions for each intrinsics to fold,
730 * and a generic selection function.
732 static GMQCC_INLINE ast_expression *fold_intrin_mod(fold_t *fold, ast_value *lhs, ast_value *rhs) {
733 return fold_constgen_float(
736 fold_immvalue_float(lhs),
737 fold_immvalue_float(rhs)
742 static GMQCC_INLINE ast_expression *fold_intrin_pow(fold_t *fold, ast_value *lhs, ast_value *rhs) {
743 return fold_constgen_float(
746 fold_immvalue_float(lhs),
747 fold_immvalue_float(rhs)
752 static GMQCC_INLINE ast_expression *fold_intrin_exp(fold_t *fold, ast_value *value) {
753 return fold_constgen_float(fold, exp(fold_immvalue_float(value)));
756 static GMQCC_INLINE ast_expression *fold_intrin_isnan(fold_t *fold, ast_value *value) {
757 return fold_constgen_float(fold, isnan(fold_immvalue_float(value)) != 0.0f);
760 static GMQCC_INLINE ast_expression *fold_intrin_fabs(fold_t *fold, ast_value *value) {
761 return fold_constgen_float(fold, fabs(fold_immvalue_float(value)));
764 ast_expression *fold_intrin(fold_t *fold, const char *intrin, ast_expression **arg) {
765 if (!strcmp(intrin, "mod")) return fold_intrin_mod (fold, (ast_value*)arg[0], (ast_value*)arg[1]);
766 if (!strcmp(intrin, "pow")) return fold_intrin_pow (fold, (ast_value*)arg[0], (ast_value*)arg[1]);
767 if (!strcmp(intrin, "exp")) return fold_intrin_exp (fold, (ast_value*)arg[0]);
768 if (!strcmp(intrin, "isnan")) return fold_intrin_isnan(fold, (ast_value*)arg[0]);
769 if (!strcmp(intrin, "fabs")) return fold_intrin_fabs (fold, (ast_value*)arg[0]);
775 * These are all the actual constant folding methods that happen in between
776 * the AST/IR stage of the compiler , i.e eliminating branches for const
777 * expressions, which is the only supported thing so far. We undefine the
778 * testing macros here because an ir_value is differant than an ast_value.
784 #undef fold_immvalue_float
785 #undef fold_immvalue_string
786 #undef fold_immvalue_vector
790 #define isfloat(X) ((X)->vtype == TYPE_FLOAT)
791 /*#define isstring(X) ((X)->vtype == TYPE_STRING)*/
792 /*#define isvector(X) ((X)->vtype == TYPE_VECTOR)*/
793 #define fold_immvalue_float(X) ((X)->constval.vfloat)
794 #define fold_immvalue_vector(X) ((X)->constval.vvec)
795 /*#define fold_immvalue_string(X) ((X)->constval.vstring)*/
796 #define fold_can_1(X) ((X)->hasvalue && (X)->cvq == CV_CONST)
797 /*#define fold_can_2(X,Y) (fold_can_1(X) && fold_can_1(Y))*/
799 ast_expression *fold_superfluous(ast_expression *left, ast_expression *right, int op) {
802 if (!ast_istype(left, ast_value) || !fold_can_1((load = (ast_value*)right)))
808 if (fold_immvalue_float(load) == 1.0f) {
809 ++opts_optimizationcount[OPTIM_PEEPHOLE];
810 return (ast_expression*)left;
817 if (fold_immvalue_float(load) == 0.0f) {
818 ++opts_optimizationcount[OPTIM_PEEPHOLE];
819 return (ast_expression*)left;
824 if (vec3_cmp(fold_immvalue_vector(load), vec3_create(1, 1, 1))) {
825 ++opts_optimizationcount[OPTIM_PEEPHOLE];
826 return (ast_expression*)left;
832 if (vec3_cmp(fold_immvalue_vector(load), vec3_create(0, 0, 0))) {
833 ++opts_optimizationcount[OPTIM_PEEPHOLE];
834 return (ast_expression*)left;
842 static GMQCC_INLINE int fold_cond(ir_value *condval, ast_function *func, ast_ifthen *branch) {
843 if (isfloat(condval) && fold_can_1(condval) && OPTS_OPTIMIZATION(OPTIM_CONST_FOLD_DCE)) {
844 ast_expression_codegen *cgen;
847 bool istrue = (fold_immvalue_float(condval) != 0.0f && branch->on_true);
848 bool isfalse = (fold_immvalue_float(condval) == 0.0f && branch->on_false);
849 ast_expression *path = (istrue) ? branch->on_true :
850 (isfalse) ? branch->on_false : NULL;
853 * no path to take implies that the evaluation is if(0) and there
854 * is no else block. so eliminate all the code.
856 ++opts_optimizationcount[OPTIM_CONST_FOLD_DCE];
860 if (!(elide = ir_function_create_block(ast_ctx(branch), func->ir_func, ast_function_label(func, ((istrue) ? "ontrue" : "onfalse")))))
862 if (!(*(cgen = path->codegen))((ast_expression*)path, func, false, &dummy))
864 if (!ir_block_create_jump(func->curblock, ast_ctx(branch), elide))
867 * now the branch has been eliminated and the correct block for the constant evaluation
868 * is expanded into the current block for the function.
870 func->curblock = elide;
871 ++opts_optimizationcount[OPTIM_CONST_FOLD_DCE];
874 return -1; /* nothing done */
877 int fold_cond_ternary(ir_value *condval, ast_function *func, ast_ternary *branch) {
878 return fold_cond(condval, func, (ast_ifthen*)branch);
881 int fold_cond_ifthen(ir_value *condval, ast_function *func, ast_ifthen *branch) {
882 return fold_cond(condval, func, branch);