?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Simplify "setcc + select $0<->$1" into "setne/seteq".
[smatch.git] / parse.c
blob1c32a2e4576ba8b228dd1585a55800c2fe60c36c
1 /*
2 * Stupid C parser, version 1e-6.
3 *
4 * Let's see how hard this is to do.
5 *
6 * Copyright (C) 2003 Transmeta Corp.
7 * 2003-2004 Linus Torvalds
8 *
9 * Licensed under the Open Software License version 1.1
10 */
11
12 #include <stdarg.h>
13 #include <stdlib.h>
14 #include <stdio.h>
15 #include <string.h>
16 #include <ctype.h>
17 #include <unistd.h>
18 #include <fcntl.h>
19 #include <limits.h>
20
21 #include "lib.h"
22 #include "token.h"
23 #include "parse.h"
24 #include "symbol.h"
25 #include "scope.h"
26 #include "expression.h"
27 #include "target.h"
28
29 #define warn_on_mixed (1)
30
31 static struct symbol_list **function_symbol_list;
32 struct symbol_list *function_computed_target_list;
33 struct statement_list *function_computed_goto_list;
34
35 // Add a symbol to the list of function-local symbols
36 #define fn_local_symbol(x) add_symbol(function_symbol_list, (x))
37
38 static struct token *statement(struct token *token, struct statement **tree);
39 static struct token *external_declaration(struct token *token, struct symbol_list **list);
40
41 static int match_idents(struct token *token, ...)
42 {
43 va_list args;
44
45 if (token_type(token) != TOKEN_IDENT)
46 return 0;
47
48 va_start(args, token);
49 for (;;) {
50 struct ident * next = va_arg(args, struct ident *);
51 if (!next)
52 return 0;
53 if (token->ident == next)
54 return 1;
55 }
56 }
57
58
59 struct statement *alloc_statement(struct position pos, int type)
60 {
61 struct statement *stmt = __alloc_statement(0);
62 stmt->type = type;
63 stmt->pos = pos;
64 return stmt;
65 }
66
67 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list);
68
69 static struct symbol * indirect(struct position pos, struct ctype *ctype, int type)
70 {
71 struct symbol *sym = alloc_symbol(pos, type);
72
73 sym->ctype.base_type = ctype->base_type;
74 sym->ctype.modifiers = ctype->modifiers & ~MOD_STORAGE;
75
76 ctype->base_type = sym;
77 ctype->modifiers &= MOD_STORAGE;
78 return sym;
79 }
80
81 static struct symbol *lookup_or_create_symbol(enum namespace ns, enum type type, struct token *token)
82 {
83 struct symbol *sym = lookup_symbol(token->ident, ns);
84 if (!sym) {
85 sym = alloc_symbol(token->pos, type);
86 bind_symbol(sym, token->ident, ns);
87 if (type == SYM_LABEL)
88 fn_local_symbol(sym);
89 }
90 return sym;
91 }
92
93 /*
94 * NOTE! NS_LABEL is not just a different namespace,
95 * it also ends up using function scope instead of the
96 * regular symbol scope.
97 */
98 struct symbol *label_symbol(struct token *token)
99 {
100 return lookup_or_create_symbol(NS_LABEL, SYM_LABEL, token);
101 }
102
103 struct token *struct_union_enum_specifier(enum type type,
104 struct token *token, struct ctype *ctype,
105 struct token *(*parse)(struct token *, struct symbol *))
106 {
107 struct symbol *sym;
108
109 ctype->modifiers = 0;
110 if (token_type(token) == TOKEN_IDENT) {
111 sym = lookup_symbol(token->ident, NS_STRUCT);
112 if (!sym ||
113 (sym->scope != block_scope &&
114 (match_op(token->next,';') || match_op(token->next,'{')))) {
115 // Either a new symbol, or else an out-of-scope
116 // symbol being redefined.
117 sym = alloc_symbol(token->pos, type);
118 bind_symbol(sym, token->ident, NS_STRUCT);
119 }
120 if (sym->type != type)
121 error_die(token->pos, "invalid tag applied to %s", show_typename (sym));
122 token = token->next;
123 ctype->base_type = sym;
124 if (match_op(token, '{')) {
125 // The following test is actually wrong for empty
126 // structs, but (1) they are not C99, (2) gcc does
127 // the same thing, and (3) it's easier.
128 if (sym->symbol_list)
129 error_die(token->pos, "redefinition of %s", show_typename (sym));
130 token = parse(token->next, sym);
131 token = expect(token, '}', "at end of struct-union-enum-specifier");
132 }
133 return token;
134 }
135
136 // private struct/union/enum type
137 if (!match_op(token, '{')) {
138 warning(token->pos, "expected declaration");
139 ctype->base_type = &bad_ctype;
140 return token;
141 }
142
143 sym = alloc_symbol(token->pos, type);
144 token = parse(token->next, sym);
145 ctype->base_type = sym;
146 return expect(token, '}', "at end of specifier");
147 }
148
149 static struct token *parse_struct_declaration(struct token *token, struct symbol *sym)
150 {
151 return struct_declaration_list(token, &sym->symbol_list);
152 }
153
154 struct token *struct_or_union_specifier(enum type type, struct token *token, struct ctype *ctype)
155 {
156 return struct_union_enum_specifier(type, token, ctype, parse_struct_declaration);
157 }
158
159 typedef struct {
160 int x;
161 unsigned long long y;
162 } Num;
163
164 static void upper_boundary(Num *n, Num *v)
165 {
166 if (n->x > v->x)
167 return;
168 if (n->x < v->x) {
169 *n = *v;
170 return;
171 }
172 if (n->y < v->y)
173 n->y = v->y;
174 }
175
176 static void lower_boundary(Num *n, Num *v)
177 {
178 if (n->x < v->x)
179 return;
180 if (n->x > v->x) {
181 *n = *v;
182 return;
183 }
184 if (n->y > v->y)
185 n->y = v->y;
186 }
187
188 static int type_is_ok(struct symbol *type, Num *upper, Num *lower)
189 {
190 int shift = type->bit_size;
191 int is_unsigned = type->ctype.modifiers & MOD_UNSIGNED;
192
193 if (!is_unsigned)
194 shift--;
195 if (upper->x == 0 && upper->y >> shift)
196 return 0;
197 if (lower->x == 0 || (!is_unsigned && (~lower->y >> shift) == 0))
198 return 1;
199 return 0;
200 }
201
202 static struct token *parse_enum_declaration(struct token *token, struct symbol *parent)
203 {
204 unsigned long long lastval = 0;
205 struct symbol *ctype = NULL, *base_type = NULL;
206 Num upper = {-1, 0}, lower = {1, 0};
207
208 while (token_type(token) == TOKEN_IDENT) {
209 struct token *next = token->next;
210 struct symbol *sym;
211
212 sym = alloc_symbol(token->pos, SYM_ENUM);
213 bind_symbol(sym, token->ident, NS_SYMBOL);
214
215 if (match_op(next, '=')) {
216 struct expression *expr;
217 next = constant_expression(next->next, &expr);
218 lastval = get_expression_value(expr);
219 ctype = expr->ctype;
220 } else if (!ctype) {
221 ctype = &int_ctype;
222 } else if (is_int_type(ctype)) {
223 lastval++;
224 } else {
225 error_die(token->pos, "can't increment the last enum member");
226 }
227
228 sym->value = lastval;
229 sym->ctype.base_type = ctype;
230
231 if (base_type != &bad_ctype) {
232 if (ctype->type == SYM_NODE)
233 ctype = ctype->ctype.base_type;
234 if (ctype->type == SYM_ENUM)
235 ctype = ctype->ctype.base_type;
236 /*
237 * base_type rules:
238 * - if all enum's are of the same type, then
239 * the base_type is that type (two first
240 * cases)
241 * - if enums are of different types, they
242 * all have to be integer types, and the
243 * base type is "int_ctype".
244 * - otherwise the base_type is "bad_ctype".
245 */
246 if (!base_type) {
247 base_type = ctype;
248 } else if (ctype == base_type) {
249 /* nothing */
250 } else if (is_int_type(base_type) && is_int_type(ctype)) {
251 base_type = &int_ctype;
252 } else
253 base_type = &bad_ctype;
254 }
255 if (is_int_type(base_type)) {
256 Num v = {.y = lastval};
257 if (ctype->ctype.modifiers & MOD_UNSIGNED)
258 v.x = 0;
259 else if ((long long)lastval >= 0)
260 v.x = 0;
261 else
262 v.x = -1;
263 upper_boundary(&upper, &v);
264 lower_boundary(&lower, &v);
265 }
266 token = next;
267 if (!match_op(token, ','))
268 break;
269 token = token->next;
270 }
271 if (!base_type)
272 base_type = &bad_ctype;
273 else if (!is_int_type(base_type))
274 base_type = base_type;
275 else if (type_is_ok(base_type, &upper, &lower))
276 base_type = base_type;
277 else if (type_is_ok(&int_ctype, &upper, &lower))
278 base_type = &int_ctype;
279 else if (type_is_ok(&uint_ctype, &upper, &lower))
280 base_type = &uint_ctype;
281 else if (type_is_ok(&long_ctype, &upper, &lower))
282 base_type = &long_ctype;
283 else if (type_is_ok(&ulong_ctype, &upper, &lower))
284 base_type = &ulong_ctype;
285 else if (type_is_ok(&llong_ctype, &upper, &lower))
286 base_type = &llong_ctype;
287 else if (type_is_ok(&ullong_ctype, &upper, &lower))
288 base_type = &ullong_ctype;
289 else
290 base_type = &bad_ctype;
291 parent->ctype.base_type = base_type;
292 parent->ctype.modifiers |= (base_type->ctype.modifiers & MOD_UNSIGNED);
293 return token;
294 }
295
296 struct token *enum_specifier(struct token *token, struct ctype *ctype)
297 {
298 return struct_union_enum_specifier(SYM_ENUM, token, ctype, parse_enum_declaration);
299 }
300
301 struct token *typeof_specifier(struct token *token, struct ctype *ctype)
302 {
303 struct symbol *sym;
304
305 if (!match_op(token, '(')) {
306 warning(token->pos, "expected '(' after typeof");
307 return token;
308 }
309 if (lookup_type(token->next)) {
310 token = typename(token->next, &sym);
311 *ctype = sym->ctype;
312 } else {
313 struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
314 token = parse_expression(token->next, &typeof_sym->initializer);
315
316 ctype->modifiers = 0;
317 ctype->base_type = typeof_sym;
318 }
319 return expect(token, ')', "after typeof");
320 }
321
322 static const char * handle_attribute(struct ctype *ctype, struct ident *attribute, struct expression *expr)
323 {
324 if (attribute == &packed_ident ||
325 attribute == &__packed___ident) {
326 ctype->alignment = 1;
327 return NULL;
328 }
329 if (attribute == &aligned_ident ||
330 attribute == &__aligned___ident) {
331 int alignment = max_alignment;
332 if (expr)
333 alignment = get_expression_value(expr);
334 ctype->alignment = alignment;
335 return NULL;
336 }
337 if (attribute == &nocast_ident) {
338 ctype->modifiers |= MOD_NOCAST;
339 return NULL;
340 }
341 if (attribute == &noderef_ident) {
342 ctype->modifiers |= MOD_NODEREF;
343 return NULL;
344 }
345 if (attribute == &safe_ident) {
346 ctype->modifiers |= MOD_SAFE;
347 return NULL;
348 }
349 if (attribute == &force_ident) {
350 ctype->modifiers |= MOD_FORCE;
351 return NULL;
352 }
353 if (attribute == &bitwise_ident) {
354 if (Wbitwise)
355 ctype->modifiers |= MOD_BITWISE;
356 return NULL;
357 }
358 if (attribute == &address_space_ident) {
359 if (!expr)
360 return "expected address space number";
361 ctype->as = get_expression_value(expr);
362 return NULL;
363 }
364 if (attribute == &context_ident) {
365 if (expr && expr->type == EXPR_COMMA) {
366 int input = get_expression_value(expr->left);
367 int output = get_expression_value(expr->right);
368 ctype->in_context = input;
369 ctype->out_context = output;
370 return NULL;
371 }
372 return "expected context input/output values";
373 }
374 if (attribute == &mode_ident ||
375 attribute == &__mode___ident) {
376 if (expr && expr->type == EXPR_SYMBOL) {
377 struct ident *ident = expr->symbol_name;
378
379 /*
380 * Match against __QI__/__HI__/__SI__/__DI__
381 *
382 * FIXME! This is broken - we don't actually get
383 * the type information updated properly at this
384 * stage for some reason.
385 */
386 if (ident == &__QI___ident ||
387 ident == &QI_ident) {
388 ctype->modifiers |= MOD_CHAR;
389 ctype->base_type = ctype_integer(ctype->modifiers);
390 return NULL;
391 }
392 if (ident == &__HI___ident ||
393 ident == &HI_ident) {
394 ctype->modifiers |= MOD_SHORT;
395 ctype->base_type = ctype_integer(ctype->modifiers);
396 return NULL;
397 }
398 if (ident == &__SI___ident ||
399 ident == &SI_ident) {
400 /* Nothing? */
401 return NULL;
402 }
403 if (ident == &__DI___ident ||
404 ident == &DI_ident) {
405 ctype->modifiers |= MOD_LONGLONG;
406 ctype->base_type = ctype_integer(ctype->modifiers);
407 return NULL;
408 }
409 if (ident == &__word___ident ||
410 ident == &word_ident) {
411 ctype->modifiers |= MOD_LONG;
412 ctype->base_type = ctype_integer(ctype->modifiers);
413 return NULL;
414 }
415 return "unknown mode attribute";
416 }
417 return "expected attribute mode symbol";
418 }
419
420 /* Throw away for now.. */
421 if (attribute == &format_ident ||
422 attribute == &__format___ident)
423 return NULL;
424 if (attribute == &section_ident ||
425 attribute == &__section___ident)
426 return NULL;
427 if (attribute == &unused_ident ||
428 attribute == &__unused___ident)
429 return NULL;
430 if (attribute == &const_ident ||
431 attribute == &__const_ident ||
432 attribute == &__const___ident)
433 return NULL;
434 if (attribute == &noreturn_ident ||
435 attribute == &__noreturn___ident)
436 return NULL;
437 if (attribute == &regparm_ident)
438 return NULL;
439 if (attribute == &weak_ident)
440 return NULL;
441 if (attribute == &alias_ident)
442 return NULL;
443 if (attribute == &pure_ident)
444 return NULL;
445 if (attribute == &always_inline_ident)
446 return NULL;
447 if (attribute == &syscall_linkage_ident)
448 return NULL;
449 if (attribute == &visibility_ident)
450 return NULL;
451 if (attribute == &model_ident ||
452 attribute == &__model___ident)
453 return NULL;
454
455 return "unknown attribute";
456 }
457
458 static struct token *attribute_specifier(struct token *token, struct ctype *ctype)
459 {
460 ctype->modifiers = 0;
461 token = expect(token, '(', "after attribute");
462 token = expect(token, '(', "after attribute");
463
464 for (;;) {
465 const char *error;
466 struct ident *attribute_name;
467 struct expression *attribute_expr;
468
469 if (eof_token(token))
470 break;
471 if (match_op(token, ';'))
472 break;
473 if (token_type(token) != TOKEN_IDENT)
474 break;
475 attribute_name = token->ident;
476 token = token->next;
477 attribute_expr = NULL;
478 if (match_op(token, '('))
479 token = parens_expression(token, &attribute_expr, "in attribute");
480 error = handle_attribute(ctype, attribute_name, attribute_expr);
481 if (error)
482 warning(token->pos, "attribute '%s': %s", show_ident(attribute_name), error);
483 if (!match_op(token, ','))
484 break;
485 token = token->next;
486 }
487
488 token = expect(token, ')', "after attribute");
489 token = expect(token, ')', "after attribute");
490 return token;
491 }
492
493 struct symbol * ctype_integer(unsigned long spec)
494 {
495 static struct symbol *const integer_ctypes[][3] = {
496 { &llong_ctype, &sllong_ctype, &ullong_ctype },
497 { &long_ctype, &slong_ctype, &ulong_ctype },
498 { &short_ctype, &sshort_ctype, &ushort_ctype },
499 { &char_ctype, &schar_ctype, &uchar_ctype },
500 { &int_ctype, &sint_ctype, &uint_ctype },
501 };
502 struct symbol *const (*ctype)[3];
503 int sub;
504
505 ctype = integer_ctypes;
506 if (!(spec & MOD_LONGLONG)) {
507 ctype++;
508 if (!(spec & MOD_LONG)) {
509 ctype++;
510 if (!(spec & MOD_SHORT)) {
511 ctype++;
512 if (!(spec & MOD_CHAR))
513 ctype++;
514 }
515 }
516 }
517
518 sub = ((spec & MOD_UNSIGNED)
519 ? 2
520 : ((spec & MOD_EXPLICITLY_SIGNED)
521 ? 1
522 : 0));
523
524 return ctype[0][sub];
525 }
526
527 struct symbol * ctype_fp(unsigned long spec)
528 {
529 if (spec & MOD_LONGLONG)
530 return &ldouble_ctype;
531 if (spec & MOD_LONG)
532 return &double_ctype;
533 return &float_ctype;
534 }
535
536 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype)
537 {
538 unsigned long mod = thistype->modifiers;
539
540 if (mod) {
541 unsigned long old = ctype->modifiers;
542 unsigned long extra = 0, dup, conflict;
543
544 if (mod & old & MOD_LONG) {
545 extra = MOD_LONGLONG | MOD_LONG;
546 mod &= ~MOD_LONG;
547 old &= ~MOD_LONG;
548 }
549 dup = (mod & old) | (extra & old) | (extra & mod);
550 if (dup)
551 warning(pos, "Just how %sdo you want this type to be?",
552 modifier_string(dup));
553
554 conflict = !(~mod & ~old & (MOD_LONG | MOD_SHORT));
555 if (conflict)
556 warning(pos, "You cannot have both long and short modifiers.");
557
558 conflict = !(~mod & ~old & (MOD_SIGNED | MOD_UNSIGNED));
559 if (conflict)
560 warning(pos, "You cannot have both signed and unsigned modifiers.");
561
562 // Only one storage modifier allowed, except that "inline" doesn't count.
563 conflict = (mod | old) & (MOD_STORAGE & ~MOD_INLINE);
564 conflict &= (conflict - 1);
565 if (conflict)
566 warning(pos, "multiple storage classes");
567
568 ctype->modifiers = old | mod | extra;
569 }
570
571 /* Context mask and value */
572 ctype->in_context += thistype->in_context;
573 ctype->out_context += thistype->out_context;
574
575 /* Alignment */
576 if (thistype->alignment & (thistype->alignment-1)) {
577 warning(pos, "I don't like non-power-of-2 alignments");
578 thistype->alignment = 0;
579 }
580 if (thistype->alignment > ctype->alignment)
581 ctype->alignment = thistype->alignment;
582
583 /* Address space */
584 ctype->as = thistype->as;
585 }
586
587 static void check_modifiers(struct position *pos, struct symbol *s, unsigned long mod)
588 {
589 unsigned long banned, wrong;
590 unsigned long this_mod = s->ctype.modifiers;
591 const unsigned long BANNED_SIZE = MOD_LONG | MOD_LONGLONG | MOD_SHORT;
592 const unsigned long BANNED_SIGN = MOD_SIGNED | MOD_UNSIGNED;
593
594 if (this_mod & (MOD_STRUCTOF | MOD_UNIONOF | MOD_ENUMOF))
595 banned = BANNED_SIZE | BANNED_SIGN;
596 else if (this_mod & MOD_SPECIALBITS)
597 banned = 0;
598 else if (s->ctype.base_type == &fp_type)
599 banned = BANNED_SIGN;
600 else if (s->ctype.base_type == &int_type || !s->ctype.base_type || is_int_type (s))
601 banned = 0;
602 else {
603 // label_type
604 // void_type
605 // bad_type
606 // vector_type <-- whatever that is
607 banned = BANNED_SIZE | BANNED_SIGN;
608 }
609
610 wrong = mod & banned;
611 if (wrong)
612 warning(*pos, "modifier %sis invalid in this context",
613 modifier_string (wrong));
614 }
615
616
617 static struct token *declaration_specifiers(struct token *next, struct ctype *ctype, int qual)
618 {
619 struct token *token;
620
621 while ( (token = next) != NULL ) {
622 struct ctype thistype;
623 struct ident *ident;
624 struct symbol *s, *type;
625 unsigned long mod;
626
627 next = token->next;
628 if (token_type(token) != TOKEN_IDENT)
629 break;
630 ident = token->ident;
631
632 s = lookup_symbol(ident, NS_TYPEDEF);
633 if (!s)
634 break;
635 thistype = s->ctype;
636 mod = thistype.modifiers;
637 if (qual && (mod & ~(MOD_ATTRIBUTE | MOD_CONST | MOD_VOLATILE)))
638 break;
639 if (mod & MOD_SPECIALBITS) {
640 if (mod & MOD_STRUCTOF)
641 next = struct_or_union_specifier(SYM_STRUCT, next, &thistype);
642 else if (mod & MOD_UNIONOF)
643 next = struct_or_union_specifier(SYM_UNION, next, &thistype);
644 else if (mod & MOD_ENUMOF)
645 next = enum_specifier(next, &thistype);
646 else if (mod & MOD_ATTRIBUTE)
647 next = attribute_specifier(next, &thistype);
648 else if (mod & MOD_TYPEOF)
649 next = typeof_specifier(next, &thistype);
650 mod = thistype.modifiers;
651 }
652 type = thistype.base_type;
653 if (type) {
654 if (qual)
655 break;
656 if (ctype->base_type)
657 break;
658 /* User types only mix with qualifiers */
659 if (mod & MOD_USERTYPE) {
660 if (ctype->modifiers & MOD_SPECIFIER)
661 break;
662 }
663 ctype->base_type = type;
664 }
665
666 check_modifiers(&token->pos, s, ctype->modifiers);
667 apply_ctype(token->pos, &thistype, ctype);
668 }
669
670 /* Turn the "virtual types" into real types with real sizes etc */
671 if (!ctype->base_type) {
672 struct symbol *base = &incomplete_ctype;
673
674 /*
675 * If we have modifiers, we'll default to an integer
676 * type, and "ctype_integer()" will turn this into
677 * a specific one.
678 */
679 if (ctype->modifiers & MOD_SPECIFIER)
680 base = &int_type;
681 ctype->base_type = base;
682 }
683 if (ctype->base_type == &int_type) {
684 ctype->base_type = ctype_integer(ctype->modifiers);
685 ctype->modifiers &= ~MOD_SPECIFIER;
686 } else if (ctype->base_type == &fp_type) {
687 ctype->base_type = ctype_fp(ctype->modifiers);
688 ctype->modifiers &= ~MOD_SPECIFIER;
689 }
690 if (ctype->modifiers & MOD_BITWISE) {
691 struct symbol *type;
692 ctype->modifiers &= ~(MOD_BITWISE | MOD_SPECIFIER);
693 if (!is_int_type(ctype->base_type)) {
694 warning(token->pos, "invalid modifier");
695 return token;
696 }
697 type = alloc_symbol(token->pos, SYM_BASETYPE);
698 *type = *ctype->base_type;
699 type->ctype.base_type = ctype->base_type;
700 type->type = SYM_RESTRICT;
701 type->ctype.modifiers &= ~MOD_SPECIFIER;
702 ctype->base_type = type;
703 }
704 return token;
705 }
706
707 static struct token *abstract_array_declarator(struct token *token, struct symbol *sym)
708 {
709 struct expression *expr = NULL;
710
711 token = parse_expression(token, &expr);
712 sym->array_size = expr;
713 return token;
714 }
715
716 static struct token *parameter_type_list(struct token *, struct symbol *);
717 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p);
718
719 static struct token *handle_attributes(struct token *token, struct ctype *ctype)
720 {
721 for (;;) {
722 if (token_type(token) != TOKEN_IDENT)
723 break;
724 if (match_idents(token, &__attribute___ident, &__attribute_ident, NULL)) {
725 struct ctype thistype = { 0, };
726 token = attribute_specifier(token->next, &thistype);
727 apply_ctype(token->pos, &thistype, ctype);
728 continue;
729 }
730 if (match_idents(token, &asm_ident, &__asm_ident, &__asm___ident)) {
731 struct expression *expr;
732 token = expect(token->next, '(', "after asm");
733 token = parse_expression(token->next, &expr);
734 token = expect(token, ')', "after asm");
735 continue;
736 }
737 break;
738 }
739 return token;
740 }
741
742 static struct token *direct_declarator(struct token *token, struct symbol *decl, struct ident **p)
743 {
744 struct ctype *ctype = &decl->ctype;
745
746 if (p && token_type(token) == TOKEN_IDENT) {
747 *p = token->ident;
748 token = token->next;
749 }
750
751 for (;;) {
752 token = handle_attributes(token, ctype);
753
754 if (token_type(token) != TOKEN_SPECIAL)
755 return token;
756
757 /*
758 * This can be either a parameter list or a grouping.
759 * For the direct (non-abstract) case, we know if must be
760 * a parameter list if we already saw the identifier.
761 * For the abstract case, we know if must be a parameter
762 * list if it is empty or starts with a type.
763 */
764 if (token->special == '(') {
765 struct symbol *sym;
766 struct token *next = token->next;
767 int fn = (p && *p) || match_op(next, ')') || lookup_type(next);
768
769 if (!fn) {
770 struct symbol *base_type = ctype->base_type;
771 token = declarator(next, decl, p);
772 token = expect(token, ')', "in nested declarator");
773 while (ctype->base_type != base_type)
774 ctype = &ctype->base_type->ctype;
775 p = NULL;
776 continue;
777 }
778
779 sym = indirect(token->pos, ctype, SYM_FN);
780 token = parameter_type_list(next, sym);
781 token = expect(token, ')', "in function declarator");
782 continue;
783 }
784 if (token->special == '[') {
785 struct symbol *array = indirect(token->pos, ctype, SYM_ARRAY);
786 token = abstract_array_declarator(token->next, array);
787 token = expect(token, ']', "in abstract_array_declarator");
788 ctype = &array->ctype;
789 continue;
790 }
791 break;
792 }
793 return token;
794 }
795
796 static struct token *pointer(struct token *token, struct ctype *ctype)
797 {
798 unsigned long modifiers;
799 struct symbol *base_type;
800
801 modifiers = ctype->modifiers & ~(MOD_TYPEDEF | MOD_ATTRIBUTE);
802 base_type = ctype->base_type;
803 ctype->modifiers = modifiers;
804
805 while (match_op(token,'*')) {
806 struct symbol *ptr = alloc_symbol(token->pos, SYM_PTR);
807 ptr->ctype.modifiers = modifiers & ~MOD_STORAGE;
808 ptr->ctype.as = ctype->as;
809 ptr->ctype.in_context += ctype->in_context;
810 ptr->ctype.out_context += ctype->out_context;
811 ptr->ctype.base_type = base_type;
812
813 base_type = ptr;
814 ctype->modifiers = modifiers & MOD_STORAGE;
815 ctype->base_type = base_type;
816 ctype->as = 0;
817 ctype->in_context = 0;
818 ctype->out_context = 0;
819
820 token = declaration_specifiers(token->next, ctype, 1);
821 modifiers = ctype->modifiers;
822 }
823 return token;
824 }
825
826 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p)
827 {
828 token = pointer(token, &sym->ctype);
829 return direct_declarator(token, sym, p);
830 }
831
832 static struct token *handle_bitfield(struct token *token, struct symbol *decl)
833 {
834 struct ctype *ctype = &decl->ctype;
835 struct expression *expr;
836 struct symbol *bitfield;
837 long long width;
838
839 if (!is_int_type(ctype->base_type)) {
840 warning(token->pos, "invalid bitfield specifier for type %s.",
841 show_typename(ctype->base_type));
842 // Parse this to recover gracefully.
843 return conditional_expression(token->next, &expr);
844 }
845
846 bitfield = indirect(token->pos, ctype, SYM_BITFIELD);
847 token = conditional_expression(token->next, &expr);
848 width = get_expression_value(expr);
849 bitfield->bit_size = width;
850
851 if (width < 0 || width > INT_MAX) {
852 warning(token->pos, "invalid bitfield width, %lld.", width);
853 width = -1;
854 } else if (decl->ident && width == 0) {
855 warning(token->pos, "invalid named zero-width bitfield `%s'",
856 show_ident(decl->ident));
857 width = -1;
858 } else if (decl->ident) {
859 struct symbol *base_type = bitfield->ctype.base_type;
860 int is_signed = !(base_type->ctype.modifiers & MOD_UNSIGNED);
861 if (width == 1 && is_signed) {
862 // Valid values are either {-1;0} or {0}, depending on integer
863 // representation. The latter makes for very efficient code...
864 warning(token->pos, "dubious one-bit signed bitfield");
865 }
866 if (Wdefault_bitfield_sign &&
867 base_type->type != SYM_ENUM &&
868 !(base_type->ctype.modifiers & MOD_EXPLICITLY_SIGNED) &&
869 is_signed) {
870 // The sign of bitfields is unspecified by default.
871 warning (token->pos, "dubious bitfield without explicit `signed' or `unsigned'");
872 }
873 }
874 bitfield->bit_size = width;
875 return token;
876 }
877
878 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list)
879 {
880 while (!match_op(token, '}')) {
881 struct ctype ctype = {0, };
882
883 token = declaration_specifiers(token, &ctype, 0);
884 for (;;) {
885 struct ident *ident = NULL;
886 struct symbol *decl = alloc_symbol(token->pos, SYM_NODE);
887 decl->ctype = ctype;
888 token = declarator(token, decl, &ident);
889 decl->ident = ident;
890 if (match_op(token, ':')) {
891 token = handle_bitfield(token, decl);
892 token = handle_attributes(token, &decl->ctype);
893 }
894 add_symbol(list, decl);
895 if (!match_op(token, ','))
896 break;
897 token = token->next;
898 }
899 if (!match_op(token, ';')) {
900 warning(token->pos, "expected ; at end of declaration");
901 break;
902 }
903 token = token->next;
904 }
905 return token;
906 }
907
908 static struct token *parameter_declaration(struct token *token, struct symbol **tree)
909 {
910 struct ident *ident = NULL;
911 struct symbol *sym;
912 struct ctype ctype = { 0, };
913
914 token = declaration_specifiers(token, &ctype, 0);
915 sym = alloc_symbol(token->pos, SYM_NODE);
916 sym->ctype = ctype;
917 *tree = sym;
918 token = declarator(token, sym, &ident);
919 sym->ident = ident;
920 return token;
921 }
922
923 struct token *typename(struct token *token, struct symbol **p)
924 {
925 struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
926 *p = sym;
927 token = declaration_specifiers(token, &sym->ctype, 0);
928 return declarator(token, sym, NULL);
929 }
930
931 struct token *expression_statement(struct token *token, struct expression **tree)
932 {
933 token = parse_expression(token, tree);
934 return expect(token, ';', "at end of statement");
935 }
936
937 static struct token *parse_asm_operands(struct token *token, struct statement *stmt)
938 {
939 struct expression *expr;
940
941 /* Allow empty operands */
942 if (match_op(token->next, ':') || match_op(token->next, ')'))
943 return token->next;
944 do {
945 if (match_op(token->next, '[') &&
946 token_type(token->next->next) == TOKEN_IDENT &&
947 match_op(token->next->next->next, ']'))
948 token = token->next->next->next;
949 token = primary_expression(token->next, &expr);
950 token = parens_expression(token, &expr, "in asm parameter");
951 } while (match_op(token, ','));
952 return token;
953 }
954
955 static struct token *parse_asm_clobbers(struct token *token, struct statement *stmt)
956 {
957 struct expression *expr;
958
959 do {
960 token = primary_expression(token->next, &expr);
961 } while (match_op(token, ','));
962 return token;
963 }
964
965 static struct token *parse_asm(struct token *token, struct statement *stmt)
966 {
967 struct expression *expr;
968
969 stmt->type = STMT_ASM;
970 if (match_idents(token, &__volatile___ident, &__volatile_ident, &volatile_ident, NULL)) {
971 token = token->next;
972 }
973 token = expect(token, '(', "after asm");
974 token = parse_expression(token->next, &expr);
975 if (match_op(token, ':'))
976 token = parse_asm_operands(token, stmt);
977 if (match_op(token, ':'))
978 token = parse_asm_operands(token, stmt);
979 if (match_op(token, ':'))
980 token = parse_asm_clobbers(token, stmt);
981 token = expect(token, ')', "after asm");
982 return expect(token, ';', "at end of asm-statement");
983 }
984
985 /* Make a statement out of an expression */
986 static struct statement *make_statement(struct expression *expr)
987 {
988 struct statement *stmt;
989
990 if (!expr)
991 return NULL;
992 stmt = alloc_statement(expr->pos, STMT_EXPRESSION);
993 stmt->expression = expr;
994 return stmt;
995 }
996
997 /*
998 * All iterators have two symbols associated with them:
999 * the "continue" and "break" symbols, which are targets
1000 * for continue and break statements respectively.
1001 *
1002 * They are in a special name-space, but they follow
1003 * all the normal visibility rules, so nested iterators
1004 * automatically work right.
1005 */
1006 static void start_iterator(struct statement *stmt)
1007 {
1008 struct symbol *cont, *brk;
1009
1010 start_symbol_scope();
1011 cont = alloc_symbol(stmt->pos, SYM_NODE);
1012 bind_symbol(cont, &continue_ident, NS_ITERATOR);
1013 brk = alloc_symbol(stmt->pos, SYM_NODE);
1014 bind_symbol(brk, &break_ident, NS_ITERATOR);
1015
1016 stmt->type = STMT_ITERATOR;
1017 stmt->iterator_break = brk;
1018 stmt->iterator_continue = cont;
1019 fn_local_symbol(brk);
1020 fn_local_symbol(cont);
1021 }
1022
1023 static void end_iterator(struct statement *stmt)
1024 {
1025 end_symbol_scope();
1026 }
1027
1028 static struct statement *start_function(struct symbol *sym)
1029 {
1030 struct symbol *ret;
1031 struct statement *stmt = alloc_statement(sym->pos, STMT_COMPOUND);
1032
1033 start_function_scope();
1034 ret = alloc_symbol(sym->pos, SYM_NODE);
1035 ret->ctype = sym->ctype.base_type->ctype;
1036 ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
1037 ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
1038 bind_symbol(ret, &return_ident, NS_ITERATOR);
1039 stmt->ret = ret;
1040 fn_local_symbol(ret);
1041
1042 // Currently parsed symbol for __func__/__FUNCTION__/__PRETTY_FUNCTION__
1043 current_fn = sym;
1044
1045 return stmt;
1046 }
1047
1048 static void end_function(struct symbol *sym)
1049 {
1050 current_fn = NULL;
1051 end_function_scope();
1052 }
1053
1054 /*
1055 * A "switch()" statement, like an iterator, has a
1056 * the "break" symbol associated with it. It works
1057 * exactly like the iterator break - it's the target
1058 * for any break-statements in scope, and means that
1059 * "break" handling doesn't even need to know whether
1060 * it's breaking out of an iterator or a switch.
1061 *
1062 * In addition, the "case" symbol is a marker for the
1063 * case/default statements to find the switch statement
1064 * that they are associated with.
1065 */
1066 static void start_switch(struct statement *stmt)
1067 {
1068 struct symbol *brk, *switch_case;
1069
1070 start_symbol_scope();
1071 brk = alloc_symbol(stmt->pos, SYM_NODE);
1072 bind_symbol(brk, &break_ident, NS_ITERATOR);
1073
1074 switch_case = alloc_symbol(stmt->pos, SYM_NODE);
1075 bind_symbol(switch_case, &case_ident, NS_ITERATOR);
1076 switch_case->stmt = stmt;
1077
1078 stmt->type = STMT_SWITCH;
1079 stmt->switch_break = brk;
1080 stmt->switch_case = switch_case;
1081
1082 fn_local_symbol(brk);
1083 fn_local_symbol(switch_case);
1084 }
1085
1086 static void end_switch(struct statement *stmt)
1087 {
1088 if (!stmt->switch_case->symbol_list)
1089 warning(stmt->pos, "switch with no cases");
1090 end_symbol_scope();
1091 }
1092
1093 static void add_case_statement(struct statement *stmt)
1094 {
1095 struct symbol *target = lookup_symbol(&case_ident, NS_ITERATOR);
1096 struct symbol *sym;
1097
1098 if (!target) {
1099 warning(stmt->pos, "not in switch scope");
1100 stmt->type = STMT_NONE;
1101 return;
1102 }
1103 sym = alloc_symbol(stmt->pos, SYM_NODE);
1104 add_symbol(&target->symbol_list, sym);
1105 sym->stmt = stmt;
1106 stmt->case_label = sym;
1107 fn_local_symbol(sym);
1108 }
1109
1110 static struct token *parse_return_statement(struct token *token, struct statement *stmt)
1111 {
1112 struct symbol *target = lookup_symbol(&return_ident, NS_ITERATOR);
1113
1114 if (!target)
1115 error_die(token->pos, "internal error: return without a function target");
1116 stmt->type = STMT_RETURN;
1117 stmt->ret_target = target;
1118 return expression_statement(token->next, &stmt->ret_value);
1119 }
1120
1121 static struct token *parse_for_statement(struct token *token, struct statement *stmt)
1122 {
1123 struct symbol_list *syms;
1124 struct expression *e1, *e2, *e3;
1125 struct statement *iterator;
1126
1127 start_iterator(stmt);
1128 token = expect(token->next, '(', "after 'for'");
1129
1130 syms = NULL;
1131 e1 = NULL;
1132 /* C99 variable declaration? */
1133 if (lookup_type(token)) {
1134 token = external_declaration(token, &syms);
1135 } else {
1136 token = parse_expression(token, &e1);
1137 token = expect(token, ';', "in 'for'");
1138 }
1139 token = parse_expression(token, &e2);
1140 token = expect(token, ';', "in 'for'");
1141 token = parse_expression(token, &e3);
1142 token = expect(token, ')', "in 'for'");
1143 token = statement(token, &iterator);
1144
1145 stmt->iterator_syms = syms;
1146 stmt->iterator_pre_statement = make_statement(e1);
1147 stmt->iterator_pre_condition = e2;
1148 stmt->iterator_post_statement = make_statement(e3);
1149 stmt->iterator_post_condition = e2;
1150 stmt->iterator_statement = iterator;
1151 end_iterator(stmt);
1152
1153 return token;
1154 }
1155
1156 struct token *parse_while_statement(struct token *token, struct statement *stmt)
1157 {
1158 struct expression *expr;
1159 struct statement *iterator;
1160
1161 start_iterator(stmt);
1162 token = parens_expression(token->next, &expr, "after 'while'");
1163 token = statement(token, &iterator);
1164
1165 stmt->iterator_pre_condition = expr;
1166 stmt->iterator_post_condition = expr;
1167 stmt->iterator_statement = iterator;
1168 end_iterator(stmt);
1169
1170 return token;
1171 }
1172
1173 struct token *parse_do_statement(struct token *token, struct statement *stmt)
1174 {
1175 struct expression *expr;
1176 struct statement *iterator;
1177
1178 start_iterator(stmt);
1179 token = statement(token->next, &iterator);
1180 if (token_type(token) == TOKEN_IDENT && token->ident == &while_ident)
1181 token = token->next;
1182 else
1183 warning(token->pos, "expected 'while' after 'do'");
1184 token = parens_expression(token, &expr, "after 'do-while'");
1185
1186 stmt->iterator_post_condition = expr;
1187 stmt->iterator_statement = iterator;
1188 end_iterator(stmt);
1189
1190 return expect(token, ';', "after statement");
1191 }
1192
1193 static struct token *statement(struct token *token, struct statement **tree)
1194 {
1195 struct statement *stmt = alloc_statement(token->pos, STMT_NONE);
1196
1197 *tree = stmt;
1198 if (token_type(token) == TOKEN_IDENT) {
1199 if (token->ident == &if_ident) {
1200 stmt->type = STMT_IF;
1201 token = parens_expression(token->next, &stmt->if_conditional, "after if");
1202 token = statement(token, &stmt->if_true);
1203 if (token_type(token) != TOKEN_IDENT)
1204 return token;
1205 if (token->ident != &else_ident)
1206 return token;
1207 return statement(token->next, &stmt->if_false);
1208 }
1209
1210 if (token->ident == &return_ident)
1211 return parse_return_statement(token, stmt);
1212
1213 if (token->ident == &break_ident || token->ident == &continue_ident) {
1214 struct symbol *target = lookup_symbol(token->ident, NS_ITERATOR);
1215 stmt->type = STMT_GOTO;
1216 stmt->goto_label = target;
1217 if (!target)
1218 warning(stmt->pos, "break/continue not in iterator scope");
1219 return expect(token->next, ';', "at end of statement");
1220 }
1221 if (token->ident == &default_ident) {
1222 token = token->next;
1223 goto default_statement;
1224 }
1225 if (token->ident == &case_ident) {
1226 token = parse_expression(token->next, &stmt->case_expression);
1227 if (match_op(token, SPECIAL_ELLIPSIS))
1228 token = parse_expression(token->next, &stmt->case_to);
1229 default_statement:
1230 stmt->type = STMT_CASE;
1231 token = expect(token, ':', "after default/case");
1232 add_case_statement(stmt);
1233 return statement(token, &stmt->case_statement);
1234 }
1235 if (token->ident == &switch_ident) {
1236 stmt->type = STMT_SWITCH;
1237 start_switch(stmt);
1238 token = parens_expression(token->next, &stmt->switch_expression, "after 'switch'");
1239 token = statement(token, &stmt->switch_statement);
1240 end_switch(stmt);
1241 return token;
1242 }
1243 if (token->ident == &for_ident)
1244 return parse_for_statement(token, stmt);
1245
1246 if (token->ident == &while_ident)
1247 return parse_while_statement(token, stmt);
1248
1249 if (token->ident == &do_ident)
1250 return parse_do_statement(token, stmt);
1251
1252 if (token->ident == &goto_ident) {
1253 stmt->type = STMT_GOTO;
1254 token = token->next;
1255 if (match_op(token, '*')) {
1256 token = parse_expression(token->next, &stmt->goto_expression);
1257 add_statement(&function_computed_goto_list, stmt);
1258 } else if (token_type(token) == TOKEN_IDENT) {
1259 stmt->goto_label = label_symbol(token);
1260 token = token->next;
1261 } else {
1262 warning(token->pos, "Expected identifier or goto expression");
1263 }
1264 return expect(token, ';', "at end of statement");
1265 }
1266 if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident, NULL)) {
1267 return parse_asm(token->next, stmt);
1268 }
1269 if (token->ident == &__context___ident) {
1270 stmt->type = STMT_INTERNAL;
1271 token = parse_expression(token->next, &stmt->expression);
1272 return expect(token, ';', "at end of statement");
1273 }
1274 if (match_op(token->next, ':')) {
1275 stmt->type = STMT_LABEL;
1276 stmt->label_identifier = label_symbol(token);
1277 return statement(token->next->next, &stmt->label_statement);
1278 }
1279 }
1280
1281 if (match_op(token, '{')) {
1282 stmt->type = STMT_COMPOUND;
1283 start_symbol_scope();
1284 token = compound_statement(token->next, stmt);
1285 end_symbol_scope();
1286
1287 return expect(token, '}', "at end of compound statement");
1288 }
1289
1290 stmt->type = STMT_EXPRESSION;
1291 return expression_statement(token, &stmt->expression);
1292 }
1293
1294 static struct token * statement_list(struct token *token, struct statement_list **list, struct symbol_list **syms)
1295 {
1296 for (;;) {
1297 struct statement * stmt;
1298 if (eof_token(token))
1299 break;
1300 if (match_op(token, '}'))
1301 break;
1302 if (lookup_type(token)) {
1303 if (warn_on_mixed && *list)
1304 warning(token->pos, "mixing declarations and code");
1305 token = external_declaration(token, syms);
1306 continue;
1307 }
1308 token = statement(token, &stmt);
1309 add_statement(list, stmt);
1310 }
1311 return token;
1312 }
1313
1314 static struct token *parameter_type_list(struct token *token, struct symbol *fn)
1315 {
1316 struct symbol_list **list = &fn->arguments;
1317
1318 if (match_op(token, ')')) {
1319 // No warning for "void oink ();"
1320 // Bug or feature: warns for "void oink () __attribute__ ((noreturn));"
1321 if (!match_op(token->next, ';'))
1322 warning(token->pos, "non-ANSI function declaration");
1323 return token;
1324 }
1325
1326 for (;;) {
1327 struct symbol *sym;
1328
1329 if (match_op(token, SPECIAL_ELLIPSIS)) {
1330 if (!*list)
1331 warning(token->pos, "variadic functions must have one named argument");
1332 fn->variadic = 1;
1333 token = token->next;
1334 break;
1335 }
1336
1337 sym = alloc_symbol(token->pos, SYM_NODE);
1338 token = parameter_declaration(token, &sym);
1339 if (sym->ctype.base_type == &void_ctype) {
1340 /* Special case: (void) */
1341 if (!*list && !sym->ident)
1342 break;
1343 warning(token->pos, "void parameter");
1344 }
1345 add_symbol(list, sym);
1346 if (!match_op(token, ','))
1347 break;
1348 token = token->next;
1349
1350 }
1351 return token;
1352 }
1353
1354 struct token *compound_statement(struct token *token, struct statement *stmt)
1355 {
1356 token = statement_list(token, &stmt->stmts, &stmt->syms);
1357 return token;
1358 }
1359
1360 static struct expression *identifier_expression(struct token *token)
1361 {
1362 struct expression *expr = alloc_expression(token->pos, EXPR_IDENTIFIER);
1363 expr->expr_ident = token->ident;
1364 return expr;
1365 }
1366
1367 static struct expression *index_expression(struct expression *from, struct expression *to)
1368 {
1369 int idx_from, idx_to;
1370 struct expression *expr = alloc_expression(from->pos, EXPR_INDEX);
1371
1372 idx_from = get_expression_value(from);
1373 idx_to = idx_from;
1374 if (to) {
1375 idx_to = get_expression_value(to);
1376 if (idx_to < idx_from || idx_from < 0)
1377 warning(from->pos, "nonsense array initializer index range");
1378 }
1379 expr->idx_from = idx_from;
1380 expr->idx_to = idx_to;
1381 return expr;
1382 }
1383
1384 static struct token *single_initializer(struct expression **ep, struct token *token)
1385 {
1386 int expect_equal = 0;
1387 struct token *next = token->next;
1388 struct expression **tail = ep;
1389 int nested;
1390
1391 *ep = NULL;
1392
1393 if ((token_type(token) == TOKEN_IDENT) && match_op(next, ':')) {
1394 struct expression *expr = identifier_expression(token);
1395 warning(token->pos, "obsolete struct initializer, use C99 syntax");
1396 token = initializer(&expr->ident_expression, next->next);
1397 if (expr->ident_expression)
1398 *ep = expr;
1399 return token;
1400 }
1401
1402 for (tail = ep, nested = 0; ; nested++, next = token->next) {
1403 if (match_op(token, '.') && (token_type(next) == TOKEN_IDENT)) {
1404 struct expression *expr = identifier_expression(next);
1405 *tail = expr;
1406 tail = &expr->ident_expression;
1407 expect_equal = 1;
1408 token = next->next;
1409 } else if (match_op(token, '[')) {
1410 struct expression *from = NULL, *to = NULL, *expr;
1411 token = constant_expression(token->next, &from);
1412 if (match_op(token, SPECIAL_ELLIPSIS))
1413 token = constant_expression(token->next, &to);
1414 expr = index_expression(from, to);
1415 *tail = expr;
1416 tail = &expr->idx_expression;
1417 token = expect(token, ']', "at end of initializer index");
1418 if (nested)
1419 expect_equal = 1;
1420 } else {
1421 break;
1422 }
1423 }
1424 if (nested && !expect_equal) {
1425 if (!match_op(token, '='))
1426 warning(token->pos, "obsolete array initializer, use C99 syntax");
1427 else
1428 expect_equal = 1;
1429 }
1430 if (expect_equal)
1431 token = expect(token, '=', "at end of initializer index");
1432
1433 token = initializer(tail, token);
1434 if (!*tail)
1435 *ep = NULL;
1436 return token;
1437 }
1438
1439 static struct token *initializer_list(struct expression_list **list, struct token *token)
1440 {
1441 struct expression *expr;
1442
1443 for (;;) {
1444 token = single_initializer(&expr, token);
1445 if (!expr)
1446 break;
1447 add_expression(list, expr);
1448 if (!match_op(token, ','))
1449 break;
1450 token = token->next;
1451 }
1452 return token;
1453 }
1454
1455 struct token *initializer(struct expression **tree, struct token *token)
1456 {
1457 if (match_op(token, '{')) {
1458 struct expression *expr = alloc_expression(token->pos, EXPR_INITIALIZER);
1459 *tree = expr;
1460 token = initializer_list(&expr->expr_list, token->next);
1461 return expect(token, '}', "at end of initializer");
1462 }
1463 return assignment_expression(token, tree);
1464 }
1465
1466 static void declare_argument(struct symbol *sym, struct symbol *fn)
1467 {
1468 if (!sym->ident) {
1469 warning(sym->pos, "no identifier for function argument");
1470 return;
1471 }
1472 bind_symbol(sym, sym->ident, NS_SYMBOL);
1473 }
1474
1475 static struct token *parse_function_body(struct token *token, struct symbol *decl,
1476 struct symbol_list **list)
1477 {
1478 struct symbol *base_type = decl->ctype.base_type;
1479 struct statement *stmt, **p;
1480 struct symbol *arg;
1481
1482 if (decl->ctype.modifiers & MOD_INLINE) {
1483 function_symbol_list = &decl->inline_symbol_list;
1484 p = &base_type->inline_stmt;
1485 } else {
1486 function_symbol_list = &decl->symbol_list;
1487 p = &base_type->stmt;
1488 }
1489 function_computed_target_list = NULL;
1490 function_computed_goto_list = NULL;
1491
1492 if (decl->ctype.modifiers & MOD_EXTERN) {
1493 if (!(decl->ctype.modifiers & MOD_INLINE))
1494 warning(decl->pos, "function with external linkage has definition");
1495 }
1496 if (!(decl->ctype.modifiers & MOD_STATIC))
1497 decl->ctype.modifiers |= MOD_EXTERN;
1498
1499 stmt = start_function(decl);
1500
1501 *p = stmt;
1502 FOR_EACH_PTR (base_type->arguments, arg) {
1503 declare_argument(arg, base_type);
1504 } END_FOR_EACH_PTR(arg);
1505
1506 token = compound_statement(token->next, stmt);
1507
1508 end_function(decl);
1509 if (!(decl->ctype.modifiers & MOD_INLINE))
1510 add_symbol(list, decl);
1511 check_declaration(decl);
1512 function_symbol_list = NULL;
1513 if (function_computed_goto_list) {
1514 if (!function_computed_target_list)
1515 warning(decl->pos, "function has computed goto but no targets?");
1516 else {
1517 struct statement *stmt;
1518 FOR_EACH_PTR(function_computed_goto_list, stmt) {
1519 stmt->target_list = function_computed_target_list;
1520 } END_FOR_EACH_PTR(stmt);
1521 }
1522 }
1523 return expect(token, '}', "at end of function");
1524 }
1525
1526 static void promote_k_r_types(struct symbol *arg)
1527 {
1528 struct symbol *base = arg->ctype.base_type;
1529 if (base && base->ctype.base_type == &int_type && (base->ctype.modifiers & (MOD_CHAR | MOD_SHORT))) {
1530 arg->ctype.base_type = &int_ctype;
1531 }
1532 }
1533
1534 static void apply_k_r_types(struct symbol_list *argtypes, struct symbol *fn)
1535 {
1536 struct symbol_list *real_args = fn->ctype.base_type->arguments;
1537 struct symbol *arg;
1538
1539 FOR_EACH_PTR(real_args, arg) {
1540 struct symbol *type;
1541
1542 /* This is quadratic in the number of arguments. We _really_ don't care */
1543 FOR_EACH_PTR(argtypes, type) {
1544 if (type->ident == arg->ident)
1545 goto match;
1546 } END_FOR_EACH_PTR(type);
1547 warning(arg->pos, "missing type declaration for parameter '%s'", show_ident(arg->ident));
1548 continue;
1549 match:
1550 type->used = 1;
1551 /* "char" and "short" promote to "int" */
1552 promote_k_r_types(type);
1553
1554 arg->ctype = type->ctype;
1555 } END_FOR_EACH_PTR(arg);
1556
1557 FOR_EACH_PTR(argtypes, arg) {
1558 if (!arg->used)
1559 warning(arg->pos, "nonsensical parameter declaration '%s'", show_ident(arg->ident));
1560 } END_FOR_EACH_PTR(arg);
1561
1562 }
1563
1564 static struct token *parse_k_r_arguments(struct token *token, struct symbol *decl,
1565 struct symbol_list **list)
1566 {
1567 struct symbol_list *args = NULL;
1568
1569 warning(token->pos, "non-ANSI function declaration");
1570 do {
1571 token = external_declaration(token, &args);
1572 } while (lookup_type(token));
1573
1574 apply_k_r_types(args, decl);
1575
1576 if (!match_op(token, '{')) {
1577 warning(token->pos, "expected function body");
1578 return token;
1579 }
1580 return parse_function_body(token, decl, list);
1581 }
1582
1583
1584 static struct token *external_declaration(struct token *token, struct symbol_list **list)
1585 {
1586 struct ident *ident = NULL;
1587 struct symbol *decl;
1588 struct ctype ctype = { 0, };
1589 struct symbol *base_type;
1590 int is_typedef;
1591
1592 /* Top-level inline asm? */
1593 if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident, NULL)) {
1594 struct symbol *anon = alloc_symbol(token->pos, SYM_NODE);
1595 struct symbol *fn = alloc_symbol(token->pos, SYM_FN);
1596 struct statement *stmt;
1597
1598 anon->ctype.base_type = fn;
1599 function_symbol_list = &anon->symbol_list;
1600 stmt = start_function(anon);
1601 token = parse_asm(token->next, stmt);
1602 end_function(anon);
1603 function_symbol_list = NULL;
1604 add_symbol(list, anon);
1605 return token;
1606 }
1607
1608 /* Parse declaration-specifiers, if any */
1609 token = declaration_specifiers(token, &ctype, 0);
1610 decl = alloc_symbol(token->pos, SYM_NODE);
1611 decl->ctype = ctype;
1612 token = declarator(token, decl, &ident);
1613
1614 /* Just a type declaration? */
1615 if (!ident)
1616 return expect(token, ';', "end of type declaration");
1617
1618 /* type define declaration? */
1619 is_typedef = (ctype.modifiers & MOD_TYPEDEF) != 0;
1620
1621 /* Typedef's don't have meaningful storage */
1622 if (is_typedef) {
1623 ctype.modifiers &= ~MOD_STORAGE;
1624 decl->ctype.modifiers &= ~MOD_STORAGE;
1625 decl->ctype.modifiers |= MOD_USERTYPE;
1626 }
1627
1628 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1629
1630 base_type = decl->ctype.base_type;
1631 if (!is_typedef && base_type && base_type->type == SYM_FN) {
1632 /* K&R argument declaration? */
1633 if (lookup_type(token))
1634 return parse_k_r_arguments(token, decl, list);
1635 if (match_op(token, '{'))
1636 return parse_function_body(token, decl, list);
1637
1638 if (!(decl->ctype.modifiers & MOD_STATIC))
1639 decl->ctype.modifiers |= MOD_EXTERN;
1640 } else if (!is_typedef && base_type == &void_ctype && !(decl->ctype.modifiers & MOD_EXTERN)) {
1641 warning(token->pos, "void declaration");
1642 }
1643
1644 for (;;) {
1645 if (!is_typedef && match_op(token, '=')) {
1646 if (decl->ctype.modifiers & MOD_EXTERN) {
1647 warning(decl->pos, "symbol with external linkage has initializer");
1648 decl->ctype.modifiers &= ~MOD_EXTERN;
1649 }
1650 token = initializer(&decl->initializer, token->next);
1651 }
1652 if (!is_typedef) {
1653 if (!(decl->ctype.modifiers & (MOD_EXTERN | MOD_INLINE))) {
1654 add_symbol(list, decl);
1655 if (function_symbol_list)
1656 fn_local_symbol(decl);
1657 }
1658 }
1659 check_declaration(decl);
1660
1661 if (!match_op(token, ','))
1662 break;
1663
1664 token = token->next;
1665 ident = NULL;
1666 decl = alloc_symbol(token->pos, SYM_NODE);
1667 decl->ctype = ctype;
1668 token = declaration_specifiers(token, &decl->ctype, 1);
1669 token = declarator(token, decl, &ident);
1670 if (!ident) {
1671 warning(token->pos, "expected identifier name in type definition");
1672 return token;
1673 }
1674
1675 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1676
1677 /* Function declarations are automatically extern unless specifically static */
1678 base_type = decl->ctype.base_type;
1679 if (!is_typedef && base_type && base_type->type == SYM_FN) {
1680 if (!(decl->ctype.modifiers & MOD_STATIC))
1681 decl->ctype.modifiers |= MOD_EXTERN;
1682 }
1683 }
1684 return expect(token, ';', "at end of declaration");
1685 }
1686
1687 struct symbol_list *translation_unit(struct token *token)
1688 {
1689 while (!eof_token(token))
1690 token = external_declaration(token, &translation_unit_used_list);
1691 // They aren't needed any more
1692 clear_token_alloc();
1693
1694 /* Evaluate the symbol list */
1695 evaluate_symbol_list(translation_unit_used_list);
1696 return translation_unit_used_list;
1697 }
Static analysis for C