?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Make "sparse()" handle multiple input files on the command line
[smatch.git] / parse.c
blob1c54244b3b11a490b2ae107d3ee0a584b5d71c62
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 "allocate.h"
23 #include "token.h"
24 #include "parse.h"
25 #include "symbol.h"
26 #include "scope.h"
27 #include "expression.h"
28 #include "target.h"
29
30 #define warn_on_mixed (1)
31
32 static struct symbol_list **function_symbol_list;
33 struct symbol_list *function_computed_target_list;
34 struct statement_list *function_computed_goto_list;
35
36 static struct token *statement(struct token *token, struct statement **tree);
37 static struct token *external_declaration(struct token *token, struct symbol_list **list);
38
39 // Add a symbol to the list of function-local symbols
40 static void fn_local_symbol(struct symbol *sym)
41 {
42 if (function_symbol_list)
43 add_symbol(function_symbol_list, sym);
44 }
45
46 static int match_idents(struct token *token, ...)
47 {
48 va_list args;
49
50 if (token_type(token) != TOKEN_IDENT)
51 return 0;
52
53 va_start(args, token);
54 for (;;) {
55 struct ident * next = va_arg(args, struct ident *);
56 if (!next)
57 return 0;
58 if (token->ident == next)
59 return 1;
60 }
61 }
62
63
64 struct statement *alloc_statement(struct position pos, int type)
65 {
66 struct statement *stmt = __alloc_statement(0);
67 stmt->type = type;
68 stmt->pos = pos;
69 return stmt;
70 }
71
72 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list);
73
74 static struct symbol * indirect(struct position pos, struct ctype *ctype, int type)
75 {
76 struct symbol *sym = alloc_symbol(pos, type);
77
78 sym->ctype.base_type = ctype->base_type;
79 sym->ctype.modifiers = ctype->modifiers & ~MOD_STORAGE;
80
81 ctype->base_type = sym;
82 ctype->modifiers &= MOD_STORAGE;
83 return sym;
84 }
85
86 static struct symbol *lookup_or_create_symbol(enum namespace ns, enum type type, struct token *token)
87 {
88 struct symbol *sym = lookup_symbol(token->ident, ns);
89 if (!sym) {
90 sym = alloc_symbol(token->pos, type);
91 bind_symbol(sym, token->ident, ns);
92 if (type == SYM_LABEL)
93 fn_local_symbol(sym);
94 }
95 return sym;
96 }
97
98 /*
99 * NOTE! NS_LABEL is not just a different namespace,
100 * it also ends up using function scope instead of the
101 * regular symbol scope.
102 */
103 struct symbol *label_symbol(struct token *token)
104 {
105 return lookup_or_create_symbol(NS_LABEL, SYM_LABEL, token);
106 }
107
108 static struct token *struct_union_enum_specifier(enum type type,
109 struct token *token, struct ctype *ctype,
110 struct token *(*parse)(struct token *, struct symbol *))
111 {
112 struct symbol *sym;
113
114 ctype->modifiers = 0;
115 if (token_type(token) == TOKEN_IDENT) {
116 sym = lookup_symbol(token->ident, NS_STRUCT);
117 if (!sym ||
118 (sym->scope != block_scope &&
119 (match_op(token->next,';') || match_op(token->next,'{')))) {
120 // Either a new symbol, or else an out-of-scope
121 // symbol being redefined.
122 sym = alloc_symbol(token->pos, type);
123 bind_symbol(sym, token->ident, NS_STRUCT);
124 }
125 if (sym->type != type)
126 error_die(token->pos, "invalid tag applied to %s", show_typename (sym));
127 token = token->next;
128 ctype->base_type = sym;
129 if (match_op(token, '{')) {
130 // The following test is actually wrong for empty
131 // structs, but (1) they are not C99, (2) gcc does
132 // the same thing, and (3) it's easier.
133 if (sym->symbol_list)
134 error_die(token->pos, "redefinition of %s", show_typename (sym));
135 token = parse(token->next, sym);
136 token = expect(token, '}', "at end of struct-union-enum-specifier");
137
138 // Mark the structure as needing re-examination
139 sym->examined = 0;
140 }
141 return token;
142 }
143
144 // private struct/union/enum type
145 if (!match_op(token, '{')) {
146 warning(token->pos, "expected declaration");
147 ctype->base_type = &bad_ctype;
148 return token;
149 }
150
151 sym = alloc_symbol(token->pos, type);
152 token = parse(token->next, sym);
153 ctype->base_type = sym;
154 return expect(token, '}', "at end of specifier");
155 }
156
157 static struct token *parse_struct_declaration(struct token *token, struct symbol *sym)
158 {
159 return struct_declaration_list(token, &sym->symbol_list);
160 }
161
162 static struct token *struct_or_union_specifier(enum type type, struct token *token, struct ctype *ctype)
163 {
164 return struct_union_enum_specifier(type, token, ctype, parse_struct_declaration);
165 }
166
167 typedef struct {
168 int x;
169 unsigned long long y;
170 } Num;
171
172 static void upper_boundary(Num *n, Num *v)
173 {
174 if (n->x > v->x)
175 return;
176 if (n->x < v->x) {
177 *n = *v;
178 return;
179 }
180 if (n->y < v->y)
181 n->y = v->y;
182 }
183
184 static void lower_boundary(Num *n, Num *v)
185 {
186 if (n->x < v->x)
187 return;
188 if (n->x > v->x) {
189 *n = *v;
190 return;
191 }
192 if (n->y > v->y)
193 n->y = v->y;
194 }
195
196 static int type_is_ok(struct symbol *type, Num *upper, Num *lower)
197 {
198 int shift = type->bit_size;
199 int is_unsigned = type->ctype.modifiers & MOD_UNSIGNED;
200
201 if (!is_unsigned)
202 shift--;
203 if (upper->x == 0 && upper->y >> shift)
204 return 0;
205 if (lower->x == 0 || (!is_unsigned && (~lower->y >> shift) == 0))
206 return 1;
207 return 0;
208 }
209
210 static struct token *parse_enum_declaration(struct token *token, struct symbol *parent)
211 {
212 unsigned long long lastval = 0;
213 struct symbol *ctype = NULL, *base_type = NULL;
214 Num upper = {-1, 0}, lower = {1, 0};
215
216 parent->examined = 1;
217 parent->ctype.base_type = &int_ctype;
218 while (token_type(token) == TOKEN_IDENT) {
219 struct expression *expr = NULL;
220 struct token *next = token->next;
221 struct symbol *sym;
222
223 sym = alloc_symbol(token->pos, SYM_NODE);
224 bind_symbol(sym, token->ident, NS_SYMBOL);
225 sym->ctype.modifiers &= ~MOD_ADDRESSABLE;
226
227 if (match_op(next, '=')) {
228 next = constant_expression(next->next, &expr);
229 lastval = get_expression_value(expr);
230 ctype = &void_ctype;
231 if (expr && expr->ctype)
232 ctype = expr->ctype;
233 } else if (!ctype) {
234 ctype = &int_ctype;
235 } else if (is_int_type(ctype)) {
236 lastval++;
237 } else {
238 error_die(token->pos, "can't increment the last enum member");
239 }
240
241 if (!expr) {
242 expr = alloc_expression(token->pos, EXPR_VALUE);
243 expr->value = lastval;
244 }
245
246 sym->initializer = expr;
247 sym->ctype.base_type = parent;
248
249 if (base_type != &bad_ctype) {
250 if (ctype->type == SYM_NODE)
251 ctype = ctype->ctype.base_type;
252 if (ctype->type == SYM_ENUM) {
253 if (ctype == parent)
254 ctype = base_type;
255 else
256 ctype = ctype->ctype.base_type;
257 }
258 /*
259 * base_type rules:
260 * - if all enum's are of the same type, then
261 * the base_type is that type (two first
262 * cases)
263 * - if enums are of different types, they
264 * all have to be integer types, and the
265 * base type is "int_ctype".
266 * - otherwise the base_type is "bad_ctype".
267 */
268 if (!base_type) {
269 base_type = ctype;
270 } else if (ctype == base_type) {
271 /* nothing */
272 } else if (is_int_type(base_type) && is_int_type(ctype)) {
273 base_type = &int_ctype;
274 } else
275 base_type = &bad_ctype;
276 }
277 if (is_int_type(base_type)) {
278 Num v = {.y = lastval};
279 if (ctype->ctype.modifiers & MOD_UNSIGNED)
280 v.x = 0;
281 else if ((long long)lastval >= 0)
282 v.x = 0;
283 else
284 v.x = -1;
285 upper_boundary(&upper, &v);
286 lower_boundary(&lower, &v);
287 }
288 token = next;
289 if (!match_op(token, ','))
290 break;
291 token = token->next;
292 }
293 if (!base_type)
294 base_type = &bad_ctype;
295 else if (!is_int_type(base_type))
296 base_type = base_type;
297 else if (type_is_ok(base_type, &upper, &lower))
298 base_type = base_type;
299 else if (type_is_ok(&int_ctype, &upper, &lower))
300 base_type = &int_ctype;
301 else if (type_is_ok(&uint_ctype, &upper, &lower))
302 base_type = &uint_ctype;
303 else if (type_is_ok(&long_ctype, &upper, &lower))
304 base_type = &long_ctype;
305 else if (type_is_ok(&ulong_ctype, &upper, &lower))
306 base_type = &ulong_ctype;
307 else if (type_is_ok(&llong_ctype, &upper, &lower))
308 base_type = &llong_ctype;
309 else if (type_is_ok(&ullong_ctype, &upper, &lower))
310 base_type = &ullong_ctype;
311 else
312 base_type = &bad_ctype;
313 parent->ctype.base_type = base_type;
314 parent->ctype.modifiers |= (base_type->ctype.modifiers & MOD_UNSIGNED);
315 parent->examined = 0;
316 return token;
317 }
318
319 static struct token *enum_specifier(struct token *token, struct ctype *ctype)
320 {
321 return struct_union_enum_specifier(SYM_ENUM, token, ctype, parse_enum_declaration);
322 }
323
324 static struct token *typeof_specifier(struct token *token, struct ctype *ctype)
325 {
326 struct symbol *sym;
327
328 if (!match_op(token, '(')) {
329 warning(token->pos, "expected '(' after typeof");
330 return token;
331 }
332 if (lookup_type(token->next)) {
333 token = typename(token->next, &sym);
334 *ctype = sym->ctype;
335 } else {
336 struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
337 token = parse_expression(token->next, &typeof_sym->initializer);
338
339 ctype->modifiers = 0;
340 ctype->base_type = typeof_sym;
341 }
342 return expect(token, ')', "after typeof");
343 }
344
345 static const char * handle_attribute(struct ctype *ctype, struct ident *attribute, struct expression *expr)
346 {
347 if (attribute == &packed_ident ||
348 attribute == &__packed___ident) {
349 ctype->alignment = 1;
350 return NULL;
351 }
352 if (attribute == &aligned_ident ||
353 attribute == &__aligned___ident) {
354 int alignment = max_alignment;
355 if (expr)
356 alignment = get_expression_value(expr);
357 ctype->alignment = alignment;
358 return NULL;
359 }
360 if (attribute == &nocast_ident) {
361 ctype->modifiers |= MOD_NOCAST;
362 return NULL;
363 }
364 if (attribute == &noderef_ident) {
365 ctype->modifiers |= MOD_NODEREF;
366 return NULL;
367 }
368 if (attribute == &safe_ident) {
369 ctype->modifiers |= MOD_SAFE;
370 return NULL;
371 }
372 if (attribute == &force_ident) {
373 ctype->modifiers |= MOD_FORCE;
374 return NULL;
375 }
376 if (attribute == &bitwise_ident ||
377 attribute == &__bitwise___ident) {
378 if (Wbitwise)
379 ctype->modifiers |= MOD_BITWISE;
380 return NULL;
381 }
382 if (attribute == &address_space_ident) {
383 if (!expr)
384 return "expected address space number";
385 ctype->as = get_expression_value(expr);
386 return NULL;
387 }
388 if (attribute == &context_ident) {
389 if (expr && expr->type == EXPR_COMMA) {
390 int input = get_expression_value(expr->left);
391 int output = get_expression_value(expr->right);
392 ctype->in_context = input;
393 ctype->out_context = output;
394 return NULL;
395 }
396 return "expected context input/output values";
397 }
398 if (attribute == &mode_ident ||
399 attribute == &__mode___ident) {
400 if (expr && expr->type == EXPR_SYMBOL) {
401 struct ident *ident = expr->symbol_name;
402
403 /*
404 * Match against __QI__/__HI__/__SI__/__DI__
405 *
406 * FIXME! This is broken - we don't actually get
407 * the type information updated properly at this
408 * stage for some reason.
409 */
410 if (ident == &__QI___ident ||
411 ident == &QI_ident) {
412 ctype->modifiers |= MOD_CHAR;
413 return NULL;
414 }
415 if (ident == &__HI___ident ||
416 ident == &HI_ident) {
417 ctype->modifiers |= MOD_SHORT;
418 return NULL;
419 }
420 if (ident == &__SI___ident ||
421 ident == &SI_ident) {
422 /* Nothing? */
423 return NULL;
424 }
425 if (ident == &__DI___ident ||
426 ident == &DI_ident) {
427 ctype->modifiers |= MOD_LONGLONG;
428 return NULL;
429 }
430 if (ident == &__word___ident ||
431 ident == &word_ident) {
432 ctype->modifiers |= MOD_LONG;
433 return NULL;
434 }
435 return "unknown mode attribute";
436 }
437 return "expected attribute mode symbol";
438 }
439
440 /* Throw away for now.. */
441 if (attribute == &__transparent_union___ident) {
442 if (Wtransparent_union)
443 return "ignoring attribute __transparent_union__";
444 return NULL;
445 }
446 if (attribute == &nothrow_ident ||
447 attribute == &__nothrow_ident ||
448 attribute == &__nothrow___ident)
449 return NULL;
450 if (attribute == &__malloc___ident)
451 return NULL;
452 if (attribute == &nonnull_ident ||
453 attribute == &__nonnull_ident ||
454 attribute == &__nonnull___ident)
455 return NULL;
456 if (attribute == &format_ident ||
457 attribute == &__format___ident ||
458 attribute == &__format_arg___ident)
459 return NULL;
460 if (attribute == &section_ident ||
461 attribute == &__section___ident)
462 return NULL;
463 if (attribute == &unused_ident ||
464 attribute == &__unused___ident)
465 return NULL;
466 if (attribute == &const_ident ||
467 attribute == &__const_ident ||
468 attribute == &__const___ident)
469 return NULL;
470 if (attribute == &noreturn_ident ||
471 attribute == &__noreturn___ident)
472 return NULL;
473 if (attribute == &regparm_ident)
474 return NULL;
475 if (attribute == &weak_ident ||
476 attribute == &__weak___ident)
477 return NULL;
478 if (attribute == &alias_ident)
479 return NULL;
480 if (attribute == &pure_ident ||
481 attribute == &__pure___ident)
482 return NULL;
483 if (attribute == &always_inline_ident)
484 return NULL;
485 if (attribute == &syscall_linkage_ident)
486 return NULL;
487 if (attribute == &visibility_ident)
488 return NULL;
489 if (attribute == &deprecated_ident ||
490 attribute == &__deprecated___ident)
491 return NULL;
492 if (attribute == &noinline_ident)
493 return NULL;
494 if (attribute == &__used___ident)
495 return NULL;
496 if (attribute == &warn_unused_result_ident ||
497 attribute == &__warn_unused_result___ident)
498 return NULL;
499 if (attribute == &model_ident ||
500 attribute == &__model___ident)
501 return NULL;
502
503 return "unknown attribute";
504 }
505
506 static struct token *attribute_specifier(struct token *token, struct ctype *ctype)
507 {
508 ctype->modifiers = 0;
509 token = expect(token, '(', "after attribute");
510 token = expect(token, '(', "after attribute");
511
512 for (;;) {
513 const char *error;
514 struct ident *attribute_name;
515 struct expression *attribute_expr;
516
517 if (eof_token(token))
518 break;
519 if (match_op(token, ';'))
520 break;
521 if (token_type(token) != TOKEN_IDENT)
522 break;
523 attribute_name = token->ident;
524 token = token->next;
525 attribute_expr = NULL;
526 if (match_op(token, '('))
527 token = parens_expression(token, &attribute_expr, "in attribute");
528 error = handle_attribute(ctype, attribute_name, attribute_expr);
529 if (error)
530 warning(token->pos, "attribute '%s': %s", show_ident(attribute_name), error);
531 if (!match_op(token, ','))
532 break;
533 token = token->next;
534 }
535
536 token = expect(token, ')', "after attribute");
537 token = expect(token, ')', "after attribute");
538 return token;
539 }
540
541 struct symbol * ctype_integer(unsigned long spec)
542 {
543 static struct symbol *const integer_ctypes[][3] = {
544 { &llong_ctype, &sllong_ctype, &ullong_ctype },
545 { &long_ctype, &slong_ctype, &ulong_ctype },
546 { &short_ctype, &sshort_ctype, &ushort_ctype },
547 { &char_ctype, &schar_ctype, &uchar_ctype },
548 { &int_ctype, &sint_ctype, &uint_ctype },
549 };
550 struct symbol *const (*ctype)[3];
551 int sub;
552
553 ctype = integer_ctypes;
554 if (!(spec & MOD_LONGLONG)) {
555 ctype++;
556 if (!(spec & MOD_LONG)) {
557 ctype++;
558 if (!(spec & MOD_SHORT)) {
559 ctype++;
560 if (!(spec & MOD_CHAR))
561 ctype++;
562 }
563 }
564 }
565
566 sub = ((spec & MOD_UNSIGNED)
567 ? 2
568 : ((spec & MOD_EXPLICITLY_SIGNED)
569 ? 1
570 : 0));
571
572 return ctype[0][sub];
573 }
574
575 struct symbol * ctype_fp(unsigned long spec)
576 {
577 if (spec & MOD_LONGLONG)
578 return &ldouble_ctype;
579 if (spec & MOD_LONG)
580 return &double_ctype;
581 return &float_ctype;
582 }
583
584 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype)
585 {
586 unsigned long mod = thistype->modifiers;
587
588 if (mod) {
589 unsigned long old = ctype->modifiers;
590 unsigned long extra = 0, dup, conflict;
591
592 if (mod & old & MOD_LONG) {
593 extra = MOD_LONGLONG | MOD_LONG;
594 mod &= ~MOD_LONG;
595 old &= ~MOD_LONG;
596 }
597 dup = (mod & old) | (extra & old) | (extra & mod);
598 if (dup)
599 warning(pos, "Just how %sdo you want this type to be?",
600 modifier_string(dup));
601
602 conflict = !(~mod & ~old & (MOD_LONG | MOD_SHORT));
603 if (conflict)
604 warning(pos, "You cannot have both long and short modifiers.");
605
606 conflict = !(~mod & ~old & (MOD_SIGNED | MOD_UNSIGNED));
607 if (conflict)
608 warning(pos, "You cannot have both signed and unsigned modifiers.");
609
610 // Only one storage modifier allowed, except that "inline" doesn't count.
611 conflict = (mod | old) & (MOD_STORAGE & ~MOD_INLINE);
612 conflict &= (conflict - 1);
613 if (conflict)
614 warning(pos, "multiple storage classes");
615
616 ctype->modifiers = old | mod | extra;
617 }
618
619 /* Context mask and value */
620 ctype->in_context += thistype->in_context;
621 ctype->out_context += thistype->out_context;
622
623 /* Alignment */
624 if (thistype->alignment & (thistype->alignment-1)) {
625 warning(pos, "I don't like non-power-of-2 alignments");
626 thistype->alignment = 0;
627 }
628 if (thistype->alignment > ctype->alignment)
629 ctype->alignment = thistype->alignment;
630
631 /* Address space */
632 ctype->as = thistype->as;
633 }
634
635 static void check_modifiers(struct position *pos, struct symbol *s, unsigned long mod)
636 {
637 unsigned long banned, wrong;
638 unsigned long this_mod = s->ctype.modifiers;
639 const unsigned long BANNED_SIZE = MOD_LONG | MOD_LONGLONG | MOD_SHORT;
640 const unsigned long BANNED_SIGN = MOD_SIGNED | MOD_UNSIGNED;
641
642 if (this_mod & (MOD_STRUCTOF | MOD_UNIONOF | MOD_ENUMOF))
643 banned = BANNED_SIZE | BANNED_SIGN;
644 else if (this_mod & MOD_SPECIALBITS)
645 banned = 0;
646 else if (s->ctype.base_type == &fp_type)
647 banned = BANNED_SIGN;
648 else if (s->ctype.base_type == &int_type || !s->ctype.base_type || is_int_type (s))
649 banned = 0;
650 else {
651 // label_type
652 // void_type
653 // bad_type
654 // vector_type <-- whatever that is
655 banned = BANNED_SIZE | BANNED_SIGN;
656 }
657
658 wrong = mod & banned;
659 if (wrong)
660 warning(*pos, "modifier %sis invalid in this context",
661 modifier_string (wrong));
662 }
663
664
665 static struct token *declaration_specifiers(struct token *next, struct ctype *ctype, int qual)
666 {
667 struct token *token;
668
669 while ( (token = next) != NULL ) {
670 struct ctype thistype;
671 struct ident *ident;
672 struct symbol *s, *type;
673 unsigned long mod;
674
675 next = token->next;
676 if (token_type(token) != TOKEN_IDENT)
677 break;
678 ident = token->ident;
679
680 s = lookup_symbol(ident, NS_TYPEDEF);
681 if (!s)
682 break;
683 thistype = s->ctype;
684 mod = thistype.modifiers;
685 if (qual && (mod & ~(MOD_ATTRIBUTE | MOD_CONST | MOD_VOLATILE)))
686 break;
687 if (mod & MOD_SPECIALBITS) {
688 if (mod & MOD_STRUCTOF)
689 next = struct_or_union_specifier(SYM_STRUCT, next, &thistype);
690 else if (mod & MOD_UNIONOF)
691 next = struct_or_union_specifier(SYM_UNION, next, &thistype);
692 else if (mod & MOD_ENUMOF)
693 next = enum_specifier(next, &thistype);
694 else if (mod & MOD_ATTRIBUTE)
695 next = attribute_specifier(next, &thistype);
696 else if (mod & MOD_TYPEOF)
697 next = typeof_specifier(next, &thistype);
698 mod = thistype.modifiers;
699 }
700 type = thistype.base_type;
701 if (type) {
702 if (qual)
703 break;
704 if (ctype->base_type)
705 break;
706 /* User types only mix with qualifiers */
707 if (mod & MOD_USERTYPE) {
708 if (ctype->modifiers & MOD_SPECIFIER)
709 break;
710 }
711 ctype->base_type = type;
712 }
713
714 check_modifiers(&token->pos, s, ctype->modifiers);
715 apply_ctype(token->pos, &thistype, ctype);
716 }
717
718 /* Turn the "virtual types" into real types with real sizes etc */
719 if (!ctype->base_type) {
720 struct symbol *base = &incomplete_ctype;
721
722 /*
723 * If we have modifiers, we'll default to an integer
724 * type, and "ctype_integer()" will turn this into
725 * a specific one.
726 */
727 if (ctype->modifiers & MOD_SPECIFIER)
728 base = &int_type;
729 ctype->base_type = base;
730 }
731 if (ctype->base_type == &int_type) {
732 ctype->base_type = ctype_integer(ctype->modifiers);
733 ctype->modifiers &= ~MOD_SPECIFIER;
734 } else if (ctype->base_type == &fp_type) {
735 ctype->base_type = ctype_fp(ctype->modifiers);
736 ctype->modifiers &= ~MOD_SPECIFIER;
737 }
738 if (ctype->modifiers & MOD_BITWISE) {
739 struct symbol *type;
740 ctype->modifiers &= ~(MOD_BITWISE | MOD_SPECIFIER);
741 if (!is_int_type(ctype->base_type)) {
742 warning(token->pos, "invalid modifier");
743 return token;
744 }
745 type = alloc_symbol(token->pos, SYM_BASETYPE);
746 *type = *ctype->base_type;
747 type->ctype.base_type = ctype->base_type;
748 type->type = SYM_RESTRICT;
749 type->ctype.modifiers &= ~MOD_SPECIFIER;
750 ctype->base_type = type;
751 }
752 return token;
753 }
754
755 static struct token *abstract_array_declarator(struct token *token, struct symbol *sym)
756 {
757 struct expression *expr = NULL;
758
759 token = parse_expression(token, &expr);
760 sym->array_size = expr;
761 return token;
762 }
763
764 static struct token *parameter_type_list(struct token *, struct symbol *, struct ident **p);
765 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p);
766
767 static struct token *handle_attributes(struct token *token, struct ctype *ctype)
768 {
769 for (;;) {
770 if (token_type(token) != TOKEN_IDENT)
771 break;
772 if (match_idents(token, &__attribute___ident, &__attribute_ident, NULL)) {
773 struct ctype thistype = { 0, };
774 token = attribute_specifier(token->next, &thistype);
775 apply_ctype(token->pos, &thistype, ctype);
776 continue;
777 }
778 if (match_idents(token, &asm_ident, &__asm_ident, &__asm___ident)) {
779 struct expression *expr;
780 token = expect(token->next, '(', "after asm");
781 token = parse_expression(token->next, &expr);
782 token = expect(token, ')', "after asm");
783 continue;
784 }
785 break;
786 }
787 return token;
788 }
789
790 static struct token *direct_declarator(struct token *token, struct symbol *decl, struct ident **p)
791 {
792 struct ctype *ctype = &decl->ctype;
793
794 if (p && token_type(token) == TOKEN_IDENT) {
795 *p = token->ident;
796 token = token->next;
797 }
798
799 for (;;) {
800 token = handle_attributes(token, ctype);
801
802 if (token_type(token) != TOKEN_SPECIAL)
803 return token;
804
805 /*
806 * This can be either a parameter list or a grouping.
807 * For the direct (non-abstract) case, we know if must be
808 * a parameter list if we already saw the identifier.
809 * For the abstract case, we know if must be a parameter
810 * list if it is empty or starts with a type.
811 */
812 if (token->special == '(') {
813 struct symbol *sym;
814 struct token *next = token->next;
815 int fn = (p && *p) || match_op(next, ')') || lookup_type(next);
816
817 if (!fn) {
818 struct symbol *base_type = ctype->base_type;
819 token = declarator(next, decl, p);
820 token = expect(token, ')', "in nested declarator");
821 while (ctype->base_type != base_type)
822 ctype = &ctype->base_type->ctype;
823 p = NULL;
824 continue;
825 }
826
827 sym = indirect(token->pos, ctype, SYM_FN);
828 token = parameter_type_list(next, sym, p);
829 token = expect(token, ')', "in function declarator");
830 continue;
831 }
832 if (token->special == '[') {
833 struct symbol *array = indirect(token->pos, ctype, SYM_ARRAY);
834 token = abstract_array_declarator(token->next, array);
835 token = expect(token, ']', "in abstract_array_declarator");
836 ctype = &array->ctype;
837 continue;
838 }
839 break;
840 }
841 return token;
842 }
843
844 static struct token *pointer(struct token *token, struct ctype *ctype)
845 {
846 unsigned long modifiers;
847 struct symbol *base_type;
848
849 modifiers = ctype->modifiers & ~(MOD_TYPEDEF | MOD_ATTRIBUTE);
850 base_type = ctype->base_type;
851 ctype->modifiers = modifiers;
852
853 while (match_op(token,'*')) {
854 struct symbol *ptr = alloc_symbol(token->pos, SYM_PTR);
855 ptr->ctype.modifiers = modifiers & ~MOD_STORAGE;
856 ptr->ctype.as = ctype->as;
857 ptr->ctype.in_context += ctype->in_context;
858 ptr->ctype.out_context += ctype->out_context;
859 ptr->ctype.base_type = base_type;
860
861 base_type = ptr;
862 ctype->modifiers = modifiers & MOD_STORAGE;
863 ctype->base_type = base_type;
864 ctype->as = 0;
865 ctype->in_context = 0;
866 ctype->out_context = 0;
867
868 token = declaration_specifiers(token->next, ctype, 1);
869 modifiers = ctype->modifiers;
870 }
871 return token;
872 }
873
874 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p)
875 {
876 token = pointer(token, &sym->ctype);
877 return direct_declarator(token, sym, p);
878 }
879
880 static struct token *handle_bitfield(struct token *token, struct symbol *decl)
881 {
882 struct ctype *ctype = &decl->ctype;
883 struct expression *expr;
884 struct symbol *bitfield;
885 long long width;
886
887 if (!is_int_type(ctype->base_type)) {
888 warning(token->pos, "invalid bitfield specifier for type %s.",
889 show_typename(ctype->base_type));
890 // Parse this to recover gracefully.
891 return conditional_expression(token->next, &expr);
892 }
893
894 bitfield = indirect(token->pos, ctype, SYM_BITFIELD);
895 token = conditional_expression(token->next, &expr);
896 width = get_expression_value(expr);
897 bitfield->bit_size = width;
898
899 if (width < 0 || width > INT_MAX) {
900 warning(token->pos, "invalid bitfield width, %lld.", width);
901 width = -1;
902 } else if (decl->ident && width == 0) {
903 warning(token->pos, "invalid named zero-width bitfield `%s'",
904 show_ident(decl->ident));
905 width = -1;
906 } else if (decl->ident) {
907 struct symbol *base_type = bitfield->ctype.base_type;
908 int is_signed = !(base_type->ctype.modifiers & MOD_UNSIGNED);
909 if (width == 1 && is_signed) {
910 // Valid values are either {-1;0} or {0}, depending on integer
911 // representation. The latter makes for very efficient code...
912 warning(token->pos, "dubious one-bit signed bitfield");
913 }
914 if (Wdefault_bitfield_sign &&
915 base_type->type != SYM_ENUM &&
916 !(base_type->ctype.modifiers & MOD_EXPLICITLY_SIGNED) &&
917 is_signed) {
918 // The sign of bitfields is unspecified by default.
919 warning (token->pos, "dubious bitfield without explicit `signed' or `unsigned'");
920 }
921 }
922 bitfield->bit_size = width;
923 return token;
924 }
925
926 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list)
927 {
928 while (!match_op(token, '}')) {
929 struct ctype ctype = {0, };
930
931 token = declaration_specifiers(token, &ctype, 0);
932 for (;;) {
933 struct ident *ident = NULL;
934 struct symbol *decl = alloc_symbol(token->pos, SYM_NODE);
935 decl->ctype = ctype;
936 token = declarator(token, decl, &ident);
937 decl->ident = ident;
938 if (match_op(token, ':')) {
939 token = handle_bitfield(token, decl);
940 token = handle_attributes(token, &decl->ctype);
941 }
942 add_symbol(list, decl);
943 if (!match_op(token, ','))
944 break;
945 token = token->next;
946 }
947 if (!match_op(token, ';')) {
948 warning(token->pos, "expected ; at end of declaration");
949 break;
950 }
951 token = token->next;
952 }
953 return token;
954 }
955
956 static struct token *parameter_declaration(struct token *token, struct symbol **tree)
957 {
958 struct ident *ident = NULL;
959 struct symbol *sym;
960 struct ctype ctype = { 0, };
961
962 token = declaration_specifiers(token, &ctype, 0);
963 sym = alloc_symbol(token->pos, SYM_NODE);
964 sym->ctype = ctype;
965 *tree = sym;
966 token = declarator(token, sym, &ident);
967 sym->ident = ident;
968 return token;
969 }
970
971 struct token *typename(struct token *token, struct symbol **p)
972 {
973 struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
974 *p = sym;
975 token = declaration_specifiers(token, &sym->ctype, 0);
976 return declarator(token, sym, NULL);
977 }
978
979 static struct token *expression_statement(struct token *token, struct expression **tree)
980 {
981 token = parse_expression(token, tree);
982 return expect(token, ';', "at end of statement");
983 }
984
985 static struct token *parse_asm_operands(struct token *token, struct statement *stmt,
986 struct expression_list **inout)
987 {
988 struct expression *expr;
989
990 /* Allow empty operands */
991 if (match_op(token->next, ':') || match_op(token->next, ')'))
992 return token->next;
993 do {
994 struct ident *ident = NULL;
995 if (match_op(token->next, '[') &&
996 token_type(token->next->next) == TOKEN_IDENT &&
997 match_op(token->next->next->next, ']')) {
998 ident = token->next->next->ident;
999 token = token->next->next->next;
1000 }
1001 add_expression(inout, (struct expression *)ident); /* UGGLEE!!! */
1002 token = primary_expression(token->next, &expr);
1003 add_expression(inout, expr);
1004 token = parens_expression(token, &expr, "in asm parameter");
1005 add_expression(inout, expr);
1006 } while (match_op(token, ','));
1007 return token;
1008 }
1009
1010 static struct token *parse_asm_clobbers(struct token *token, struct statement *stmt,
1011 struct expression_list **clobbers)
1012 {
1013 struct expression *expr;
1014
1015 do {
1016 token = primary_expression(token->next, &expr);
1017 add_expression(clobbers, expr);
1018 } while (match_op(token, ','));
1019 return token;
1020 }
1021
1022 static struct token *parse_asm(struct token *token, struct statement *stmt)
1023 {
1024 stmt->type = STMT_ASM;
1025 if (match_idents(token, &__volatile___ident, &__volatile_ident, &volatile_ident, NULL)) {
1026 token = token->next;
1027 }
1028 token = expect(token, '(', "after asm");
1029 token = parse_expression(token, &stmt->asm_string);
1030 if (match_op(token, ':'))
1031 token = parse_asm_operands(token, stmt, &stmt->asm_outputs);
1032 if (match_op(token, ':'))
1033 token = parse_asm_operands(token, stmt, &stmt->asm_inputs);
1034 if (match_op(token, ':'))
1035 token = parse_asm_clobbers(token, stmt, &stmt->asm_clobbers);
1036 token = expect(token, ')', "after asm");
1037 return expect(token, ';', "at end of asm-statement");
1038 }
1039
1040 /* Make a statement out of an expression */
1041 static struct statement *make_statement(struct expression *expr)
1042 {
1043 struct statement *stmt;
1044
1045 if (!expr)
1046 return NULL;
1047 stmt = alloc_statement(expr->pos, STMT_EXPRESSION);
1048 stmt->expression = expr;
1049 return stmt;
1050 }
1051
1052 /*
1053 * All iterators have two symbols associated with them:
1054 * the "continue" and "break" symbols, which are targets
1055 * for continue and break statements respectively.
1056 *
1057 * They are in a special name-space, but they follow
1058 * all the normal visibility rules, so nested iterators
1059 * automatically work right.
1060 */
1061 static void start_iterator(struct statement *stmt)
1062 {
1063 struct symbol *cont, *brk;
1064
1065 start_symbol_scope();
1066 cont = alloc_symbol(stmt->pos, SYM_NODE);
1067 bind_symbol(cont, &continue_ident, NS_ITERATOR);
1068 brk = alloc_symbol(stmt->pos, SYM_NODE);
1069 bind_symbol(brk, &break_ident, NS_ITERATOR);
1070
1071 stmt->type = STMT_ITERATOR;
1072 stmt->iterator_break = brk;
1073 stmt->iterator_continue = cont;
1074 fn_local_symbol(brk);
1075 fn_local_symbol(cont);
1076 }
1077
1078 static void end_iterator(struct statement *stmt)
1079 {
1080 end_symbol_scope();
1081 }
1082
1083 static struct statement *start_function(struct symbol *sym)
1084 {
1085 struct symbol *ret;
1086 struct statement *stmt = alloc_statement(sym->pos, STMT_COMPOUND);
1087
1088 start_function_scope();
1089 ret = alloc_symbol(sym->pos, SYM_NODE);
1090 ret->ctype = sym->ctype.base_type->ctype;
1091 ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
1092 ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
1093 bind_symbol(ret, &return_ident, NS_ITERATOR);
1094 stmt->ret = ret;
1095 fn_local_symbol(ret);
1096
1097 // Currently parsed symbol for __func__/__FUNCTION__/__PRETTY_FUNCTION__
1098 current_fn = sym;
1099
1100 return stmt;
1101 }
1102
1103 static void end_function(struct symbol *sym)
1104 {
1105 current_fn = NULL;
1106 end_function_scope();
1107 }
1108
1109 /*
1110 * A "switch()" statement, like an iterator, has a
1111 * the "break" symbol associated with it. It works
1112 * exactly like the iterator break - it's the target
1113 * for any break-statements in scope, and means that
1114 * "break" handling doesn't even need to know whether
1115 * it's breaking out of an iterator or a switch.
1116 *
1117 * In addition, the "case" symbol is a marker for the
1118 * case/default statements to find the switch statement
1119 * that they are associated with.
1120 */
1121 static void start_switch(struct statement *stmt)
1122 {
1123 struct symbol *brk, *switch_case;
1124
1125 start_symbol_scope();
1126 brk = alloc_symbol(stmt->pos, SYM_NODE);
1127 bind_symbol(brk, &break_ident, NS_ITERATOR);
1128
1129 switch_case = alloc_symbol(stmt->pos, SYM_NODE);
1130 bind_symbol(switch_case, &case_ident, NS_ITERATOR);
1131 switch_case->stmt = stmt;
1132
1133 stmt->type = STMT_SWITCH;
1134 stmt->switch_break = brk;
1135 stmt->switch_case = switch_case;
1136
1137 fn_local_symbol(brk);
1138 fn_local_symbol(switch_case);
1139 }
1140
1141 static void end_switch(struct statement *stmt)
1142 {
1143 if (!stmt->switch_case->symbol_list)
1144 warning(stmt->pos, "switch with no cases");
1145 end_symbol_scope();
1146 }
1147
1148 static void add_case_statement(struct statement *stmt)
1149 {
1150 struct symbol *target = lookup_symbol(&case_ident, NS_ITERATOR);
1151 struct symbol *sym;
1152
1153 if (!target) {
1154 warning(stmt->pos, "not in switch scope");
1155 stmt->type = STMT_NONE;
1156 return;
1157 }
1158 sym = alloc_symbol(stmt->pos, SYM_NODE);
1159 add_symbol(&target->symbol_list, sym);
1160 sym->stmt = stmt;
1161 stmt->case_label = sym;
1162 fn_local_symbol(sym);
1163 }
1164
1165 static struct token *parse_return_statement(struct token *token, struct statement *stmt)
1166 {
1167 struct symbol *target = lookup_symbol(&return_ident, NS_ITERATOR);
1168
1169 if (!target)
1170 error_die(token->pos, "internal error: return without a function target");
1171 stmt->type = STMT_RETURN;
1172 stmt->ret_target = target;
1173 return expression_statement(token->next, &stmt->ret_value);
1174 }
1175
1176 static struct token *parse_for_statement(struct token *token, struct statement *stmt)
1177 {
1178 struct symbol_list *syms;
1179 struct expression *e1, *e2, *e3;
1180 struct statement *iterator;
1181
1182 start_iterator(stmt);
1183 token = expect(token->next, '(', "after 'for'");
1184
1185 syms = NULL;
1186 e1 = NULL;
1187 /* C99 variable declaration? */
1188 if (lookup_type(token)) {
1189 token = external_declaration(token, &syms);
1190 } else {
1191 token = parse_expression(token, &e1);
1192 token = expect(token, ';', "in 'for'");
1193 }
1194 token = parse_expression(token, &e2);
1195 token = expect(token, ';', "in 'for'");
1196 token = parse_expression(token, &e3);
1197 token = expect(token, ')', "in 'for'");
1198 token = statement(token, &iterator);
1199
1200 stmt->iterator_syms = syms;
1201 stmt->iterator_pre_statement = make_statement(e1);
1202 stmt->iterator_pre_condition = e2;
1203 stmt->iterator_post_statement = make_statement(e3);
1204 stmt->iterator_post_condition = e2;
1205 stmt->iterator_statement = iterator;
1206 end_iterator(stmt);
1207
1208 return token;
1209 }
1210
1211 static struct token *parse_while_statement(struct token *token, struct statement *stmt)
1212 {
1213 struct expression *expr;
1214 struct statement *iterator;
1215
1216 start_iterator(stmt);
1217 token = parens_expression(token->next, &expr, "after 'while'");
1218 token = statement(token, &iterator);
1219
1220 stmt->iterator_pre_condition = expr;
1221 stmt->iterator_post_condition = expr;
1222 stmt->iterator_statement = iterator;
1223 end_iterator(stmt);
1224
1225 return token;
1226 }
1227
1228 static struct token *parse_do_statement(struct token *token, struct statement *stmt)
1229 {
1230 struct expression *expr;
1231 struct statement *iterator;
1232
1233 start_iterator(stmt);
1234 token = statement(token->next, &iterator);
1235 if (token_type(token) == TOKEN_IDENT && token->ident == &while_ident)
1236 token = token->next;
1237 else
1238 warning(token->pos, "expected 'while' after 'do'");
1239 token = parens_expression(token, &expr, "after 'do-while'");
1240
1241 stmt->iterator_post_condition = expr;
1242 stmt->iterator_statement = iterator;
1243 end_iterator(stmt);
1244
1245 return expect(token, ';', "after statement");
1246 }
1247
1248 static struct token *statement(struct token *token, struct statement **tree)
1249 {
1250 struct statement *stmt = alloc_statement(token->pos, STMT_NONE);
1251
1252 *tree = stmt;
1253 if (token_type(token) == TOKEN_IDENT) {
1254 if (token->ident == &if_ident) {
1255 stmt->type = STMT_IF;
1256 token = parens_expression(token->next, &stmt->if_conditional, "after if");
1257 token = statement(token, &stmt->if_true);
1258 if (token_type(token) != TOKEN_IDENT)
1259 return token;
1260 if (token->ident != &else_ident)
1261 return token;
1262 return statement(token->next, &stmt->if_false);
1263 }
1264
1265 if (token->ident == &return_ident)
1266 return parse_return_statement(token, stmt);
1267
1268 if (token->ident == &break_ident || token->ident == &continue_ident) {
1269 struct symbol *target = lookup_symbol(token->ident, NS_ITERATOR);
1270 stmt->type = STMT_GOTO;
1271 stmt->goto_label = target;
1272 if (!target)
1273 warning(stmt->pos, "break/continue not in iterator scope");
1274 return expect(token->next, ';', "at end of statement");
1275 }
1276 if (token->ident == &default_ident) {
1277 token = token->next;
1278 goto default_statement;
1279 }
1280 if (token->ident == &case_ident) {
1281 token = parse_expression(token->next, &stmt->case_expression);
1282 if (match_op(token, SPECIAL_ELLIPSIS))
1283 token = parse_expression(token->next, &stmt->case_to);
1284 default_statement:
1285 stmt->type = STMT_CASE;
1286 token = expect(token, ':', "after default/case");
1287 add_case_statement(stmt);
1288 return statement(token, &stmt->case_statement);
1289 }
1290 if (token->ident == &switch_ident) {
1291 stmt->type = STMT_SWITCH;
1292 start_switch(stmt);
1293 token = parens_expression(token->next, &stmt->switch_expression, "after 'switch'");
1294 token = statement(token, &stmt->switch_statement);
1295 end_switch(stmt);
1296 return token;
1297 }
1298 if (token->ident == &for_ident)
1299 return parse_for_statement(token, stmt);
1300
1301 if (token->ident == &while_ident)
1302 return parse_while_statement(token, stmt);
1303
1304 if (token->ident == &do_ident)
1305 return parse_do_statement(token, stmt);
1306
1307 if (token->ident == &goto_ident) {
1308 stmt->type = STMT_GOTO;
1309 token = token->next;
1310 if (match_op(token, '*')) {
1311 token = parse_expression(token->next, &stmt->goto_expression);
1312 add_statement(&function_computed_goto_list, stmt);
1313 } else if (token_type(token) == TOKEN_IDENT) {
1314 stmt->goto_label = label_symbol(token);
1315 token = token->next;
1316 } else {
1317 warning(token->pos, "Expected identifier or goto expression");
1318 }
1319 return expect(token, ';', "at end of statement");
1320 }
1321 if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident, NULL)) {
1322 return parse_asm(token->next, stmt);
1323 }
1324 if (token->ident == &__context___ident) {
1325 stmt->type = STMT_CONTEXT;
1326 token = parse_expression(token->next, &stmt->expression);
1327 return expect(token, ';', "at end of statement");
1328 }
1329 if (token->ident == &__range___ident) {
1330 stmt->type = STMT_RANGE;
1331 token = assignment_expression(token->next, &stmt->range_expression);
1332 token = expect(token, ',', "after range expression");
1333 token = assignment_expression(token, &stmt->range_low);
1334 token = expect(token, ',', "after low range");
1335 token = assignment_expression(token, &stmt->range_high);
1336 return expect(token, ';', "after range statement");
1337 }
1338 if (match_op(token->next, ':')) {
1339 stmt->type = STMT_LABEL;
1340 stmt->label_identifier = label_symbol(token);
1341 return statement(token->next->next, &stmt->label_statement);
1342 }
1343 }
1344
1345 if (match_op(token, '{')) {
1346 stmt->type = STMT_COMPOUND;
1347 start_symbol_scope();
1348 token = compound_statement(token->next, stmt);
1349 end_symbol_scope();
1350
1351 return expect(token, '}', "at end of compound statement");
1352 }
1353
1354 stmt->type = STMT_EXPRESSION;
1355 return expression_statement(token, &stmt->expression);
1356 }
1357
1358 static struct token * statement_list(struct token *token, struct statement_list **list, struct symbol_list **syms)
1359 {
1360 for (;;) {
1361 struct statement * stmt;
1362 if (eof_token(token))
1363 break;
1364 if (match_op(token, '}'))
1365 break;
1366 if (lookup_type(token)) {
1367 if (warn_on_mixed && *list)
1368 warning(token->pos, "mixing declarations and code");
1369 token = external_declaration(token, syms);
1370 continue;
1371 }
1372 token = statement(token, &stmt);
1373 add_statement(list, stmt);
1374 }
1375 return token;
1376 }
1377
1378 static struct token *parameter_type_list(struct token *token, struct symbol *fn, struct ident **p)
1379 {
1380 struct symbol_list **list = &fn->arguments;
1381
1382 if (match_op(token, ')')) {
1383 // No warning for "void oink ();"
1384 // Bug or feature: warns for "void oink () __attribute__ ((noreturn));"
1385 if (p && !match_op(token->next, ';'))
1386 warning(token->pos, "non-ANSI function declaration of function '%s'", show_ident(*p));
1387 return token;
1388 }
1389
1390 for (;;) {
1391 struct symbol *sym;
1392
1393 if (match_op(token, SPECIAL_ELLIPSIS)) {
1394 if (!*list)
1395 warning(token->pos, "variadic functions must have one named argument");
1396 fn->variadic = 1;
1397 token = token->next;
1398 break;
1399 }
1400
1401 sym = alloc_symbol(token->pos, SYM_NODE);
1402 token = parameter_declaration(token, &sym);
1403 if (sym->ctype.base_type == &void_ctype) {
1404 /* Special case: (void) */
1405 if (!*list && !sym->ident)
1406 break;
1407 warning(token->pos, "void parameter");
1408 }
1409 add_symbol(list, sym);
1410 if (!match_op(token, ','))
1411 break;
1412 token = token->next;
1413
1414 }
1415 return token;
1416 }
1417
1418 struct token *compound_statement(struct token *token, struct statement *stmt)
1419 {
1420 token = statement_list(token, &stmt->stmts, &stmt->syms);
1421 return token;
1422 }
1423
1424 static struct expression *identifier_expression(struct token *token)
1425 {
1426 struct expression *expr = alloc_expression(token->pos, EXPR_IDENTIFIER);
1427 expr->expr_ident = token->ident;
1428 return expr;
1429 }
1430
1431 static struct expression *index_expression(struct expression *from, struct expression *to)
1432 {
1433 int idx_from, idx_to;
1434 struct expression *expr = alloc_expression(from->pos, EXPR_INDEX);
1435
1436 idx_from = get_expression_value(from);
1437 idx_to = idx_from;
1438 if (to) {
1439 idx_to = get_expression_value(to);
1440 if (idx_to < idx_from || idx_from < 0)
1441 warning(from->pos, "nonsense array initializer index range");
1442 }
1443 expr->idx_from = idx_from;
1444 expr->idx_to = idx_to;
1445 return expr;
1446 }
1447
1448 static struct token *single_initializer(struct expression **ep, struct token *token)
1449 {
1450 int expect_equal = 0;
1451 struct token *next = token->next;
1452 struct expression **tail = ep;
1453 int nested;
1454
1455 *ep = NULL;
1456
1457 if ((token_type(token) == TOKEN_IDENT) && match_op(next, ':')) {
1458 struct expression *expr = identifier_expression(token);
1459 warning(token->pos, "obsolete struct initializer, use C99 syntax");
1460 token = initializer(&expr->ident_expression, next->next);
1461 if (expr->ident_expression)
1462 *ep = expr;
1463 return token;
1464 }
1465
1466 for (tail = ep, nested = 0; ; nested++, next = token->next) {
1467 if (match_op(token, '.') && (token_type(next) == TOKEN_IDENT)) {
1468 struct expression *expr = identifier_expression(next);
1469 *tail = expr;
1470 tail = &expr->ident_expression;
1471 expect_equal = 1;
1472 token = next->next;
1473 } else if (match_op(token, '[')) {
1474 struct expression *from = NULL, *to = NULL, *expr;
1475 token = constant_expression(token->next, &from);
1476 if (!from) {
1477 warning(token->pos, "Expected constant expression");
1478 break;
1479 }
1480 if (match_op(token, SPECIAL_ELLIPSIS))
1481 token = constant_expression(token->next, &to);
1482 expr = index_expression(from, to);
1483 *tail = expr;
1484 tail = &expr->idx_expression;
1485 token = expect(token, ']', "at end of initializer index");
1486 if (nested)
1487 expect_equal = 1;
1488 } else {
1489 break;
1490 }
1491 }
1492 if (nested && !expect_equal) {
1493 if (!match_op(token, '='))
1494 warning(token->pos, "obsolete array initializer, use C99 syntax");
1495 else
1496 expect_equal = 1;
1497 }
1498 if (expect_equal)
1499 token = expect(token, '=', "at end of initializer index");
1500
1501 token = initializer(tail, token);
1502 if (!*tail)
1503 *ep = NULL;
1504 return token;
1505 }
1506
1507 static struct token *initializer_list(struct expression_list **list, struct token *token)
1508 {
1509 struct expression *expr;
1510
1511 for (;;) {
1512 token = single_initializer(&expr, token);
1513 if (!expr)
1514 break;
1515 add_expression(list, expr);
1516 if (!match_op(token, ','))
1517 break;
1518 token = token->next;
1519 }
1520 return token;
1521 }
1522
1523 struct token *initializer(struct expression **tree, struct token *token)
1524 {
1525 if (match_op(token, '{')) {
1526 struct expression *expr = alloc_expression(token->pos, EXPR_INITIALIZER);
1527 *tree = expr;
1528 token = initializer_list(&expr->expr_list, token->next);
1529 return expect(token, '}', "at end of initializer");
1530 }
1531 return assignment_expression(token, tree);
1532 }
1533
1534 static void declare_argument(struct symbol *sym, struct symbol *fn)
1535 {
1536 if (!sym->ident) {
1537 warning(sym->pos, "no identifier for function argument");
1538 return;
1539 }
1540 bind_symbol(sym, sym->ident, NS_SYMBOL);
1541 }
1542
1543 static struct token *parse_function_body(struct token *token, struct symbol *decl,
1544 struct symbol_list **list)
1545 {
1546 struct symbol_list **old_symbol_list;
1547 struct symbol *base_type = decl->ctype.base_type;
1548 struct statement *stmt, **p;
1549 struct symbol *arg;
1550
1551 old_symbol_list = function_symbol_list;
1552 if (decl->ctype.modifiers & MOD_INLINE) {
1553 function_symbol_list = &decl->inline_symbol_list;
1554 p = &base_type->inline_stmt;
1555 } else {
1556 function_symbol_list = &decl->symbol_list;
1557 p = &base_type->stmt;
1558 }
1559 function_computed_target_list = NULL;
1560 function_computed_goto_list = NULL;
1561
1562 if (decl->ctype.modifiers & MOD_EXTERN) {
1563 if (!(decl->ctype.modifiers & MOD_INLINE))
1564 warning(decl->pos, "function '%s' with external linkage has definition", show_ident(decl->ident));
1565 }
1566 if (!(decl->ctype.modifiers & MOD_STATIC))
1567 decl->ctype.modifiers |= MOD_EXTERN;
1568
1569 stmt = start_function(decl);
1570
1571 *p = stmt;
1572 FOR_EACH_PTR (base_type->arguments, arg) {
1573 declare_argument(arg, base_type);
1574 } END_FOR_EACH_PTR(arg);
1575
1576 token = compound_statement(token->next, stmt);
1577
1578 end_function(decl);
1579 if (!(decl->ctype.modifiers & MOD_INLINE))
1580 add_symbol(list, decl);
1581 check_declaration(decl);
1582 function_symbol_list = old_symbol_list;
1583 if (function_computed_goto_list) {
1584 if (!function_computed_target_list)
1585 warning(decl->pos, "function '%s' has computed goto but no targets?", show_ident(decl->ident));
1586 else {
1587 struct statement *stmt;
1588 FOR_EACH_PTR(function_computed_goto_list, stmt) {
1589 stmt->target_list = function_computed_target_list;
1590 } END_FOR_EACH_PTR(stmt);
1591 }
1592 }
1593 return expect(token, '}', "at end of function");
1594 }
1595
1596 static void promote_k_r_types(struct symbol *arg)
1597 {
1598 struct symbol *base = arg->ctype.base_type;
1599 if (base && base->ctype.base_type == &int_type && (base->ctype.modifiers & (MOD_CHAR | MOD_SHORT))) {
1600 arg->ctype.base_type = &int_ctype;
1601 }
1602 }
1603
1604 static void apply_k_r_types(struct symbol_list *argtypes, struct symbol *fn)
1605 {
1606 struct symbol_list *real_args = fn->ctype.base_type->arguments;
1607 struct symbol *arg;
1608
1609 FOR_EACH_PTR(real_args, arg) {
1610 struct symbol *type;
1611
1612 /* This is quadratic in the number of arguments. We _really_ don't care */
1613 FOR_EACH_PTR(argtypes, type) {
1614 if (type->ident == arg->ident)
1615 goto match;
1616 } END_FOR_EACH_PTR(type);
1617 warning(arg->pos, "missing type declaration for parameter '%s'", show_ident(arg->ident));
1618 continue;
1619 match:
1620 type->used = 1;
1621 /* "char" and "short" promote to "int" */
1622 promote_k_r_types(type);
1623
1624 arg->ctype = type->ctype;
1625 } END_FOR_EACH_PTR(arg);
1626
1627 FOR_EACH_PTR(argtypes, arg) {
1628 if (!arg->used)
1629 warning(arg->pos, "nonsensical parameter declaration '%s'", show_ident(arg->ident));
1630 } END_FOR_EACH_PTR(arg);
1631
1632 }
1633
1634 static struct token *parse_k_r_arguments(struct token *token, struct symbol *decl,
1635 struct symbol_list **list)
1636 {
1637 struct symbol_list *args = NULL;
1638
1639 warning(token->pos, "non-ANSI function declaration of function '%s'", show_ident(decl->ident));
1640 do {
1641 token = external_declaration(token, &args);
1642 } while (lookup_type(token));
1643
1644 apply_k_r_types(args, decl);
1645
1646 if (!match_op(token, '{')) {
1647 warning(token->pos, "expected function body");
1648 return token;
1649 }
1650 return parse_function_body(token, decl, list);
1651 }
1652
1653
1654 static struct token *external_declaration(struct token *token, struct symbol_list **list)
1655 {
1656 struct ident *ident = NULL;
1657 struct symbol *decl;
1658 struct ctype ctype = { 0, };
1659 struct symbol *base_type;
1660 int is_typedef;
1661
1662 /* Top-level inline asm? */
1663 if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident, NULL)) {
1664 struct symbol_list **old_symbol_list;
1665 struct symbol *anon = alloc_symbol(token->pos, SYM_NODE);
1666 struct symbol *fn = alloc_symbol(token->pos, SYM_FN);
1667 struct statement *stmt;
1668
1669 anon->ctype.base_type = fn;
1670 old_symbol_list = function_symbol_list;
1671 function_symbol_list = &anon->symbol_list;
1672 stmt = start_function(anon);
1673 token = parse_asm(token->next, stmt);
1674 end_function(anon);
1675 function_symbol_list = old_symbol_list;
1676 add_symbol(list, anon);
1677 return token;
1678 }
1679
1680 /* Parse declaration-specifiers, if any */
1681 token = declaration_specifiers(token, &ctype, 0);
1682 decl = alloc_symbol(token->pos, SYM_NODE);
1683 decl->ctype = ctype;
1684 token = declarator(token, decl, &ident);
1685
1686 /* Just a type declaration? */
1687 if (!ident)
1688 return expect(token, ';', "end of type declaration");
1689
1690 /* type define declaration? */
1691 is_typedef = (ctype.modifiers & MOD_TYPEDEF) != 0;
1692
1693 /* Typedef's don't have meaningful storage */
1694 if (is_typedef) {
1695 ctype.modifiers &= ~MOD_STORAGE;
1696 decl->ctype.modifiers &= ~MOD_STORAGE;
1697 decl->ctype.modifiers |= MOD_USERTYPE;
1698 }
1699
1700 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1701
1702 base_type = decl->ctype.base_type;
1703 if (!is_typedef && base_type && base_type->type == SYM_FN) {
1704 /* K&R argument declaration? */
1705 if (lookup_type(token))
1706 return parse_k_r_arguments(token, decl, list);
1707 if (match_op(token, '{'))
1708 return parse_function_body(token, decl, list);
1709
1710 if (!(decl->ctype.modifiers & MOD_STATIC))
1711 decl->ctype.modifiers |= MOD_EXTERN;
1712 } else if (!is_typedef && base_type == &void_ctype && !(decl->ctype.modifiers & MOD_EXTERN)) {
1713 warning(token->pos, "void declaration");
1714 }
1715
1716 for (;;) {
1717 if (!is_typedef && match_op(token, '=')) {
1718 if (decl->ctype.modifiers & MOD_EXTERN) {
1719 warning(decl->pos, "symbol with external linkage has initializer");
1720 decl->ctype.modifiers &= ~MOD_EXTERN;
1721 }
1722 token = initializer(&decl->initializer, token->next);
1723 }
1724 if (!is_typedef) {
1725 if (!(decl->ctype.modifiers & (MOD_EXTERN | MOD_INLINE))) {
1726 add_symbol(list, decl);
1727 fn_local_symbol(decl);
1728 }
1729 }
1730 check_declaration(decl);
1731
1732 if (!match_op(token, ','))
1733 break;
1734
1735 token = token->next;
1736 ident = NULL;
1737 decl = alloc_symbol(token->pos, SYM_NODE);
1738 decl->ctype = ctype;
1739 token = declaration_specifiers(token, &decl->ctype, 1);
1740 token = declarator(token, decl, &ident);
1741 if (!ident) {
1742 warning(token->pos, "expected identifier name in type definition");
1743 return token;
1744 }
1745
1746 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1747
1748 /* Function declarations are automatically extern unless specifically static */
1749 base_type = decl->ctype.base_type;
1750 if (!is_typedef && base_type && base_type->type == SYM_FN) {
1751 if (!(decl->ctype.modifiers & MOD_STATIC))
1752 decl->ctype.modifiers |= MOD_EXTERN;
1753 }
1754 }
1755 return expect(token, ';', "at end of declaration");
1756 }
1757
1758 struct symbol_list *translation_unit(struct token *token)
1759 {
1760 while (!eof_token(token))
1761 token = external_declaration(token, &translation_unit_used_list);
1762 // They aren't needed any more
1763 clear_token_alloc();
1764
1765 /* Evaluate the symbol list */
1766 evaluate_symbol_list(translation_unit_used_list);
1767 return translation_unit_used_list;
1768 }
Static analysis for C