?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Introduce top level parsing for asm parsing.
[smatch.git] / parse.c
blob94b3e0f438c9c5fe66c9390d8c88742d1edf9ce7
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 * Copyright (C) 2004 Christopher Li
9 *
10 * Licensed under the Open Software License version 1.1
11 */
12
13 #include <stdarg.h>
14 #include <stdlib.h>
15 #include <stdio.h>
16 #include <string.h>
17 #include <ctype.h>
18 #include <unistd.h>
19 #include <fcntl.h>
20 #include <limits.h>
21
22 #include "lib.h"
23 #include "allocate.h"
24 #include "token.h"
25 #include "parse.h"
26 #include "symbol.h"
27 #include "scope.h"
28 #include "expression.h"
29 #include "target.h"
30
31 #define warn_on_mixed (1)
32
33 static struct symbol_list **function_symbol_list;
34 struct symbol_list *function_computed_target_list;
35 struct statement_list *function_computed_goto_list;
36
37 static struct token *statement(struct token *token, struct statement **tree);
38 static struct token *handle_attributes(struct token *token, struct ctype *ctype);
39
40 static struct token *struct_specifier(struct token *token, struct ctype *ctype);
41 static struct token *union_specifier(struct token *token, struct ctype *ctype);
42 static struct token *enum_specifier(struct token *token, struct ctype *ctype);
43 static struct token *attribute_specifier(struct token *token, struct ctype *ctype);
44 static struct token *typeof_specifier(struct token *token, struct ctype *ctype);
45
46 static struct token *parse_if_statement(struct token *token, struct statement *stmt);
47 static struct token *parse_return_statement(struct token *token, struct statement *stmt);
48 static struct token *parse_loop_iterator(struct token *token, struct statement *stmt);
49 static struct token *parse_default_statement(struct token *token, struct statement *stmt);
50 static struct token *parse_case_statement(struct token *token, struct statement *stmt);
51 static struct token *parse_switch_statement(struct token *token, struct statement *stmt);
52 static struct token *parse_for_statement(struct token *token, struct statement *stmt);
53 static struct token *parse_while_statement(struct token *token, struct statement *stmt);
54 static struct token *parse_do_statement(struct token *token, struct statement *stmt);
55 static struct token *parse_goto_statement(struct token *token, struct statement *stmt);
56 static struct token *parse_context_statement(struct token *token, struct statement *stmt);
57 static struct token *parse_range_statement(struct token *token, struct statement *stmt);
58 static struct token *parse_asm(struct token *token, struct statement *stmt);
59 static struct token *toplevel_asm_declaration(struct token *token, struct symbol_list **list);
60
61
62 static struct symbol_op modifier_op = {
63 .type = KW_MODIFIER,
64 };
65
66 static struct symbol_op qualifier_op = {
67 .type = KW_QUALIFIER,
68 };
69
70 static struct symbol_op typeof_op = {
71 .type = KW_TYPEOF,
72 .declarator = typeof_specifier,
73 };
74
75 static struct symbol_op attribute_op = {
76 .type = KW_ATTRIBUTE,
77 .declarator = attribute_specifier,
78 };
79
80 static struct symbol_op struct_op = {
81 .type = KW_SPECIFIER,
82 .declarator = struct_specifier,
83 };
84
85 static struct symbol_op union_op = {
86 .type = KW_SPECIFIER,
87 .declarator = union_specifier,
88 };
89
90 static struct symbol_op enum_op = {
91 .type = KW_SPECIFIER,
92 .declarator = enum_specifier,
93 };
94
95
96
97 static struct symbol_op if_op = {
98 .statement = parse_if_statement,
99 };
100
101 static struct symbol_op return_op = {
102 .statement = parse_return_statement,
103 };
104
105 static struct symbol_op loop_iter_op = {
106 .statement = parse_loop_iterator,
107 };
108
109 static struct symbol_op default_op = {
110 .statement = parse_default_statement,
111 };
112
113 static struct symbol_op case_op = {
114 .statement = parse_case_statement,
115 };
116
117 static struct symbol_op switch_op = {
118 .statement = parse_switch_statement,
119 };
120
121 static struct symbol_op for_op = {
122 .statement = parse_for_statement,
123 };
124
125 static struct symbol_op while_op = {
126 .statement = parse_while_statement,
127 };
128
129 static struct symbol_op do_op = {
130 .statement = parse_do_statement,
131 };
132
133 static struct symbol_op goto_op = {
134 .statement = parse_goto_statement,
135 };
136
137 static struct symbol_op context_op = {
138 .statement = parse_context_statement,
139 };
140
141 static struct symbol_op range_op = {
142 .statement = parse_range_statement,
143 };
144
145 static struct symbol_op asm_op = {
146 .statement = parse_asm,
147 .toplevel = toplevel_asm_declaration,
148 };
149
150 static struct init_keyword {
151 const char *name;
152 enum namespace ns;
153 unsigned long modifiers;
154 struct symbol_op *op;
155 } keyword_table[] = {
156 /* Type qualifiers */
157 { "const", NS_TYPEDEF, MOD_CONST, .op = &qualifier_op },
158 { "__const", NS_TYPEDEF, MOD_CONST, .op = &qualifier_op },
159 { "__const__", NS_TYPEDEF, MOD_CONST, .op = &qualifier_op },
160 { "volatile", NS_TYPEDEF, MOD_VOLATILE, .op = &qualifier_op },
161 { "__volatile", NS_TYPEDEF, MOD_VOLATILE, .op = &qualifier_op },
162 { "__volatile__", NS_TYPEDEF, MOD_VOLATILE, .op = &qualifier_op },
163
164 /* Typedef.. */
165 { "typedef", NS_TYPEDEF, MOD_TYPEDEF, .op = &modifier_op },
166
167 /* Extended types */
168 { "typeof", NS_TYPEDEF, .op = &typeof_op },
169 { "__typeof", NS_TYPEDEF, .op = &typeof_op },
170 { "__typeof__", NS_TYPEDEF, .op = &typeof_op },
171
172 { "__attribute", NS_TYPEDEF, .op = &attribute_op },
173 { "__attribute__", NS_TYPEDEF, .op = &attribute_op },
174
175 { "struct", NS_TYPEDEF, .op = &struct_op },
176 { "union", NS_TYPEDEF, .op = &union_op },
177 { "enum", NS_TYPEDEF, .op = &enum_op },
178
179 { "inline", NS_TYPEDEF, MOD_INLINE, .op = &modifier_op },
180 { "__inline", NS_TYPEDEF, MOD_INLINE, .op = &modifier_op },
181 { "__inline__", NS_TYPEDEF, MOD_INLINE, .op = &modifier_op },
182
183 /* Ignored for now.. */
184 { "restrict", NS_TYPEDEF, .op = &qualifier_op},
185 { "__restrict", NS_TYPEDEF, .op = &qualifier_op},
186
187 /* Statement */
188 { "if", NS_KEYWORD, .op = &if_op },
189 { "return", NS_KEYWORD, .op = &return_op },
190 { "break", NS_KEYWORD, .op = &loop_iter_op },
191 { "continue", NS_KEYWORD, .op = &loop_iter_op },
192 { "default", NS_KEYWORD, .op = &default_op },
193 { "case", NS_KEYWORD, .op = &case_op },
194 { "switch", NS_KEYWORD, .op = &switch_op },
195 { "for", NS_KEYWORD, .op = &for_op },
196 { "while", NS_KEYWORD, .op = &while_op },
197 { "do", NS_KEYWORD, .op = &do_op },
198 { "goto", NS_KEYWORD, .op = &goto_op },
199 { "__context__",NS_KEYWORD, .op = &context_op },
200 { "__range__", NS_KEYWORD, .op = &range_op },
201 { "asm", NS_KEYWORD, .op = &asm_op },
202 { "__asm", NS_KEYWORD, .op = &asm_op },
203 { "__asm__", NS_KEYWORD, .op = &asm_op },
204
205 };
206
207 void init_parser(int stream)
208 {
209 int i;
210 for (i = 0; i < sizeof keyword_table/sizeof keyword_table[0]; i++) {
211 struct init_keyword *ptr = keyword_table + i;
212 struct symbol *sym = create_symbol(stream, ptr->name, SYM_KEYWORD, ptr->ns);
213 sym->ident->keyword = 1;
214 sym->ctype.modifiers = ptr->modifiers;
215 sym->op = ptr->op;
216 }
217 }
218
219 // Add a symbol to the list of function-local symbols
220 static void fn_local_symbol(struct symbol *sym)
221 {
222 if (function_symbol_list)
223 add_symbol(function_symbol_list, sym);
224 }
225
226 static int match_idents(struct token *token, ...)
227 {
228 va_list args;
229
230 if (token_type(token) != TOKEN_IDENT)
231 return 0;
232
233 va_start(args, token);
234 for (;;) {
235 struct ident * next = va_arg(args, struct ident *);
236 if (!next)
237 return 0;
238 if (token->ident == next)
239 return 1;
240 }
241 }
242
243
244 struct statement *alloc_statement(struct position pos, int type)
245 {
246 struct statement *stmt = __alloc_statement(0);
247 stmt->type = type;
248 stmt->pos = pos;
249 return stmt;
250 }
251
252 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list);
253
254 static int apply_modifiers(struct position pos, struct ctype *ctype)
255 {
256 struct symbol *base;
257
258 while ((base = ctype->base_type)) {
259 switch (base->type) {
260 case SYM_FN:
261 case SYM_ENUM:
262 case SYM_ARRAY:
263 case SYM_BITFIELD:
264 case SYM_PTR:
265 ctype = &base->ctype;
266 continue;
267 }
268 break;
269 }
270
271 /* Turn the "virtual types" into real types with real sizes etc */
272 if (ctype->base_type == &int_type) {
273 ctype->base_type = ctype_integer(ctype->modifiers);
274 ctype->modifiers &= ~MOD_SPECIFIER;
275 } else if (ctype->base_type == &fp_type) {
276 ctype->base_type = ctype_fp(ctype->modifiers);
277 ctype->modifiers &= ~MOD_SPECIFIER;
278 }
279
280 if (ctype->modifiers & MOD_BITWISE) {
281 struct symbol *type;
282 ctype->modifiers &= ~(MOD_BITWISE | MOD_SPECIFIER);
283 if (!is_int_type(ctype->base_type)) {
284 sparse_error(pos, "invalid modifier");
285 return 1;
286 }
287 type = alloc_symbol(pos, SYM_BASETYPE);
288 *type = *ctype->base_type;
289 type->ctype.base_type = ctype->base_type;
290 type->type = SYM_RESTRICT;
291 type->ctype.modifiers &= ~MOD_SPECIFIER;
292 ctype->base_type = type;
293 create_fouled(type);
294 }
295 return 0;
296 }
297
298 static struct symbol * alloc_indirect_symbol(struct position pos, struct ctype *ctype, int type)
299 {
300 struct symbol *sym = alloc_symbol(pos, type);
301
302 sym->ctype.base_type = ctype->base_type;
303 sym->ctype.modifiers = ctype->modifiers & ~MOD_STORAGE;
304
305 ctype->base_type = sym;
306 ctype->modifiers &= MOD_STORAGE;
307 return sym;
308 }
309
310 static struct symbol *lookup_or_create_symbol(enum namespace ns, enum type type, struct token *token)
311 {
312 struct symbol *sym = lookup_symbol(token->ident, ns);
313 if (!sym) {
314 sym = alloc_symbol(token->pos, type);
315 bind_symbol(sym, token->ident, ns);
316 if (type == SYM_LABEL)
317 fn_local_symbol(sym);
318 }
319 return sym;
320 }
321
322 /*
323 * NOTE! NS_LABEL is not just a different namespace,
324 * it also ends up using function scope instead of the
325 * regular symbol scope.
326 */
327 struct symbol *label_symbol(struct token *token)
328 {
329 return lookup_or_create_symbol(NS_LABEL, SYM_LABEL, token);
330 }
331
332 static struct token *struct_union_enum_specifier(enum type type,
333 struct token *token, struct ctype *ctype,
334 struct token *(*parse)(struct token *, struct symbol *))
335 {
336 struct symbol *sym;
337 struct position *repos;
338
339 ctype->modifiers = 0;
340 token = handle_attributes(token, ctype);
341 if (token_type(token) == TOKEN_IDENT) {
342 sym = lookup_symbol(token->ident, NS_STRUCT);
343 if (!sym ||
344 (sym->scope != block_scope &&
345 (match_op(token->next,';') || match_op(token->next,'{')))) {
346 // Either a new symbol, or else an out-of-scope
347 // symbol being redefined.
348 sym = alloc_symbol(token->pos, type);
349 bind_symbol(sym, token->ident, NS_STRUCT);
350 }
351 if (sym->type != type)
352 error_die(token->pos, "invalid tag applied to %s", show_typename (sym));
353 ctype->base_type = sym;
354 repos = &token->pos;
355 token = token->next;
356 if (match_op(token, '{')) {
357 // The following test is actually wrong for empty
358 // structs, but (1) they are not C99, (2) gcc does
359 // the same thing, and (3) it's easier.
360 if (sym->symbol_list)
361 error_die(token->pos, "redefinition of %s", show_typename (sym));
362 sym->pos = *repos;
363 token = parse(token->next, sym);
364 token = expect(token, '}', "at end of struct-union-enum-specifier");
365
366 // Mark the structure as needing re-examination
367 sym->examined = 0;
368 }
369 return token;
370 }
371
372 // private struct/union/enum type
373 if (!match_op(token, '{')) {
374 sparse_error(token->pos, "expected declaration");
375 ctype->base_type = &bad_ctype;
376 return token;
377 }
378
379 sym = alloc_symbol(token->pos, type);
380 token = parse(token->next, sym);
381 ctype->base_type = sym;
382 return expect(token, '}', "at end of specifier");
383 }
384
385 static struct token *parse_struct_declaration(struct token *token, struct symbol *sym)
386 {
387 return struct_declaration_list(token, &sym->symbol_list);
388 }
389
390 static struct token *struct_specifier(struct token *token, struct ctype *ctype)
391 {
392 return struct_union_enum_specifier(SYM_STRUCT, token, ctype, parse_struct_declaration);
393 }
394
395 static struct token *union_specifier(struct token *token, struct ctype *ctype)
396 {
397 return struct_union_enum_specifier(SYM_UNION, token, ctype, parse_struct_declaration);
398 }
399
400
401 typedef struct {
402 int x;
403 unsigned long long y;
404 } Num;
405
406 static void upper_boundary(Num *n, Num *v)
407 {
408 if (n->x > v->x)
409 return;
410 if (n->x < v->x) {
411 *n = *v;
412 return;
413 }
414 if (n->y < v->y)
415 n->y = v->y;
416 }
417
418 static void lower_boundary(Num *n, Num *v)
419 {
420 if (n->x < v->x)
421 return;
422 if (n->x > v->x) {
423 *n = *v;
424 return;
425 }
426 if (n->y > v->y)
427 n->y = v->y;
428 }
429
430 static int type_is_ok(struct symbol *type, Num *upper, Num *lower)
431 {
432 int shift = type->bit_size;
433 int is_unsigned = type->ctype.modifiers & MOD_UNSIGNED;
434
435 if (!is_unsigned)
436 shift--;
437 if (upper->x == 0 && upper->y >> shift)
438 return 0;
439 if (lower->x == 0 || (!is_unsigned && (~lower->y >> shift) == 0))
440 return 1;
441 return 0;
442 }
443
444 static struct symbol *bigger_enum_type(struct symbol *s1, struct symbol *s2)
445 {
446 if (s1->bit_size < s2->bit_size) {
447 s1 = s2;
448 } else if (s1->bit_size == s2->bit_size) {
449 if (s2->ctype.modifiers & MOD_UNSIGNED)
450 s1 = s2;
451 }
452 if (s1->bit_size < bits_in_int)
453 return &int_ctype;
454 return s1;
455 }
456
457 static void cast_enum_list(struct symbol_list *list, struct symbol *base_type)
458 {
459 struct symbol *sym;
460
461 FOR_EACH_PTR(list, sym) {
462 struct expression *expr = sym->initializer;
463 struct symbol *ctype;
464 if (expr->type != EXPR_VALUE)
465 continue;
466 ctype = expr->ctype;
467 if (ctype->bit_size == base_type->bit_size)
468 continue;
469 cast_value(expr, base_type, expr, ctype);
470 } END_FOR_EACH_PTR(sym);
471 }
472
473 static struct token *parse_enum_declaration(struct token *token, struct symbol *parent)
474 {
475 unsigned long long lastval = 0;
476 struct symbol *ctype = NULL, *base_type = NULL;
477 Num upper = {-1, 0}, lower = {1, 0};
478 struct symbol_list *entries = NULL;
479
480 parent->examined = 1;
481 parent->ctype.base_type = &int_ctype;
482 while (token_type(token) == TOKEN_IDENT) {
483 struct expression *expr = NULL;
484 struct token *next = token->next;
485 struct symbol *sym;
486
487 sym = alloc_symbol(token->pos, SYM_NODE);
488 bind_symbol(sym, token->ident, NS_SYMBOL);
489 sym->ctype.modifiers &= ~MOD_ADDRESSABLE;
490
491 if (match_op(next, '=')) {
492 next = constant_expression(next->next, &expr);
493 lastval = get_expression_value(expr);
494 ctype = &void_ctype;
495 if (expr && expr->ctype)
496 ctype = expr->ctype;
497 } else if (!ctype) {
498 ctype = &int_ctype;
499 } else if (is_int_type(ctype)) {
500 lastval++;
501 } else {
502 error_die(token->pos, "can't increment the last enum member");
503 }
504
505 if (!expr) {
506 expr = alloc_expression(token->pos, EXPR_VALUE);
507 expr->value = lastval;
508 expr->ctype = ctype;
509 }
510
511 sym->initializer = expr;
512 sym->ctype.base_type = parent;
513 add_ptr_list(&entries, sym);
514
515 if (base_type != &bad_ctype) {
516 if (ctype->type == SYM_NODE)
517 ctype = ctype->ctype.base_type;
518 if (ctype->type == SYM_ENUM) {
519 if (ctype == parent)
520 ctype = base_type;
521 else
522 ctype = ctype->ctype.base_type;
523 }
524 /*
525 * base_type rules:
526 * - if all enum's are of the same type, then
527 * the base_type is that type (two first
528 * cases)
529 * - if enums are of different types, they
530 * all have to be integer types, and the
531 * base type is at least "int_ctype".
532 * - otherwise the base_type is "bad_ctype".
533 */
534 if (!base_type) {
535 base_type = ctype;
536 } else if (ctype == base_type) {
537 /* nothing */
538 } else if (is_int_type(base_type) && is_int_type(ctype)) {
539 base_type = bigger_enum_type(base_type, ctype);
540 } else
541 base_type = &bad_ctype;
542 parent->ctype.base_type = base_type;
543 }
544 if (is_int_type(base_type)) {
545 Num v = {.y = lastval};
546 if (ctype->ctype.modifiers & MOD_UNSIGNED)
547 v.x = 0;
548 else if ((long long)lastval >= 0)
549 v.x = 0;
550 else
551 v.x = -1;
552 upper_boundary(&upper, &v);
553 lower_boundary(&lower, &v);
554 }
555 token = next;
556 if (!match_op(token, ','))
557 break;
558 token = token->next;
559 }
560 if (!base_type) {
561 sparse_error(token->pos, "bad enum definition");
562 base_type = &bad_ctype;
563 }
564 else if (!is_int_type(base_type))
565 base_type = base_type;
566 else if (type_is_ok(base_type, &upper, &lower))
567 base_type = base_type;
568 else if (type_is_ok(&int_ctype, &upper, &lower))
569 base_type = &int_ctype;
570 else if (type_is_ok(&uint_ctype, &upper, &lower))
571 base_type = &uint_ctype;
572 else if (type_is_ok(&long_ctype, &upper, &lower))
573 base_type = &long_ctype;
574 else if (type_is_ok(&ulong_ctype, &upper, &lower))
575 base_type = &ulong_ctype;
576 else if (type_is_ok(&llong_ctype, &upper, &lower))
577 base_type = &llong_ctype;
578 else if (type_is_ok(&ullong_ctype, &upper, &lower))
579 base_type = &ullong_ctype;
580 else
581 base_type = &bad_ctype;
582 parent->ctype.base_type = base_type;
583 parent->ctype.modifiers |= (base_type->ctype.modifiers & MOD_UNSIGNED);
584 parent->examined = 0;
585
586 cast_enum_list(entries, base_type);
587 free_ptr_list(&entries);
588
589 return token;
590 }
591
592 static struct token *enum_specifier(struct token *token, struct ctype *ctype)
593 {
594 struct token *ret = struct_union_enum_specifier(SYM_ENUM, token, ctype, parse_enum_declaration);
595
596 ctype = &ctype->base_type->ctype;
597 if (!ctype->base_type)
598 ctype->base_type = &incomplete_ctype;
599
600 return ret;
601 }
602
603 static struct token *typeof_specifier(struct token *token, struct ctype *ctype)
604 {
605 struct symbol *sym;
606
607 if (!match_op(token, '(')) {
608 sparse_error(token->pos, "expected '(' after typeof");
609 return token;
610 }
611 if (lookup_type(token->next)) {
612 token = typename(token->next, &sym);
613 *ctype = sym->ctype;
614 } else {
615 struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
616 token = parse_expression(token->next, &typeof_sym->initializer);
617
618 ctype->modifiers = 0;
619 ctype->base_type = typeof_sym;
620 }
621 return expect(token, ')', "after typeof");
622 }
623
624 static const char * handle_attribute(struct ctype *ctype, struct ident *attribute, struct expression *expr)
625 {
626 if (attribute == &packed_ident ||
627 attribute == &__packed___ident) {
628 ctype->alignment = 1;
629 return NULL;
630 }
631 if (attribute == &aligned_ident ||
632 attribute == &__aligned___ident) {
633 int alignment = max_alignment;
634 if (expr)
635 alignment = get_expression_value(expr);
636 ctype->alignment = alignment;
637 return NULL;
638 }
639 if (attribute == &nocast_ident) {
640 ctype->modifiers |= MOD_NOCAST;
641 return NULL;
642 }
643 if (attribute == &noderef_ident) {
644 ctype->modifiers |= MOD_NODEREF;
645 return NULL;
646 }
647 if (attribute == &safe_ident) {
648 ctype->modifiers |= MOD_SAFE;
649 return NULL;
650 }
651 if (attribute == &force_ident) {
652 ctype->modifiers |= MOD_FORCE;
653 return NULL;
654 }
655 if (attribute == &bitwise_ident ||
656 attribute == &__bitwise___ident) {
657 if (Wbitwise)
658 ctype->modifiers |= MOD_BITWISE;
659 return NULL;
660 }
661 if (attribute == &address_space_ident) {
662 if (!expr)
663 return "expected address space number";
664 ctype->as = get_expression_value(expr);
665 return NULL;
666 }
667 if (attribute == &context_ident) {
668 if (expr && expr->type == EXPR_COMMA) {
669 struct context *context = alloc_context();
670 if(expr->left->type == EXPR_COMMA) {
671 context->context = expr->left->left;
672 context->in = get_expression_value(
673 expr->left->right);
674 } else {
675 context->context = NULL;
676 context->in = get_expression_value(expr->left);
677 }
678 context->out = get_expression_value(expr->right);
679 add_ptr_list(&ctype->contexts, context);
680 return NULL;
681 }
682 return "expected context input/output values";
683 }
684 if (attribute == &mode_ident ||
685 attribute == &__mode___ident) {
686 if (expr && expr->type == EXPR_SYMBOL) {
687 struct ident *ident = expr->symbol_name;
688
689 /*
690 * Match against __QI__/__HI__/__SI__/__DI__
691 *
692 * FIXME! This is broken - we don't actually get
693 * the type information updated properly at this
694 * stage for some reason.
695 */
696 if (ident == &__QI___ident ||
697 ident == &QI_ident) {
698 ctype->modifiers |= MOD_CHAR;
699 return NULL;
700 }
701 if (ident == &__HI___ident ||
702 ident == &HI_ident) {
703 ctype->modifiers |= MOD_SHORT;
704 return NULL;
705 }
706 if (ident == &__SI___ident ||
707 ident == &SI_ident) {
708 /* Nothing? */
709 return NULL;
710 }
711 if (ident == &__DI___ident ||
712 ident == &DI_ident) {
713 ctype->modifiers |= MOD_LONGLONG;
714 return NULL;
715 }
716 if (ident == &__word___ident ||
717 ident == &word_ident) {
718 ctype->modifiers |= MOD_LONG;
719 return NULL;
720 }
721 return "unknown mode attribute";
722 }
723 return "expected attribute mode symbol";
724 }
725
726 /* Throw away for now.. */
727 if (attribute == &__transparent_union___ident) {
728 if (Wtransparent_union)
729 return "ignoring attribute __transparent_union__";
730 return NULL;
731 }
732 if (attribute == &nothrow_ident ||
733 attribute == &__nothrow_ident ||
734 attribute == &__nothrow___ident)
735 return NULL;
736 if (attribute == &__malloc___ident)
737 return NULL;
738 if (attribute == &nonnull_ident ||
739 attribute == &__nonnull_ident ||
740 attribute == &__nonnull___ident)
741 return NULL;
742 if (attribute == &format_ident ||
743 attribute == &__format___ident ||
744 attribute == &__format_arg___ident)
745 return NULL;
746 if (attribute == &section_ident ||
747 attribute == &__section___ident)
748 return NULL;
749 if (attribute == &unused_ident ||
750 attribute == &__unused___ident)
751 return NULL;
752 if (attribute == &const_ident ||
753 attribute == &__const_ident ||
754 attribute == &__const___ident)
755 return NULL;
756 if (attribute == &noreturn_ident ||
757 attribute == &__noreturn___ident)
758 return NULL;
759 if (attribute == &no_instrument_function_ident ||
760 attribute == &__no_instrument_function___ident)
761 return NULL;
762 if (attribute == &sentinel_ident ||
763 attribute == &__sentinel___ident)
764 return NULL;
765 if (attribute == &regparm_ident)
766 return NULL;
767 if (attribute == &weak_ident ||
768 attribute == &__weak___ident)
769 return NULL;
770 if (attribute == &alias_ident ||
771 attribute == &__alias___ident)
772 return NULL;
773 if (attribute == &pure_ident ||
774 attribute == &__pure___ident)
775 return NULL;
776 if (attribute == &always_inline_ident)
777 return NULL;
778 if (attribute == &syscall_linkage_ident)
779 return NULL;
780 if (attribute == &visibility_ident ||
781 attribute == &__visibility___ident)
782 return NULL;
783 if (attribute == &deprecated_ident ||
784 attribute == &__deprecated___ident)
785 return NULL;
786 if (attribute == &noinline_ident)
787 return NULL;
788 if (attribute == &__used___ident)
789 return NULL;
790 if (attribute == &warn_unused_result_ident ||
791 attribute == &__warn_unused_result___ident)
792 return NULL;
793 if (attribute == &model_ident ||
794 attribute == &__model___ident)
795 return NULL;
796
797 return "unknown attribute";
798 }
799
800 static struct token *attribute_specifier(struct token *token, struct ctype *ctype)
801 {
802 ctype->modifiers = 0;
803 token = expect(token, '(', "after attribute");
804 token = expect(token, '(', "after attribute");
805
806 for (;;) {
807 const char *error_str;
808 struct ident *attribute_name;
809 struct expression *attribute_expr;
810
811 if (eof_token(token))
812 break;
813 if (match_op(token, ';'))
814 break;
815 if (token_type(token) != TOKEN_IDENT)
816 break;
817 attribute_name = token->ident;
818 token = token->next;
819 attribute_expr = NULL;
820 if (match_op(token, '('))
821 token = parens_expression(token, &attribute_expr, "in attribute");
822 error_str = handle_attribute(ctype, attribute_name, attribute_expr);
823 if (error_str)
824 sparse_error(token->pos, "attribute '%s': %s", show_ident(attribute_name), error_str);
825 if (!match_op(token, ','))
826 break;
827 token = token->next;
828 }
829
830 token = expect(token, ')', "after attribute");
831 token = expect(token, ')', "after attribute");
832 return token;
833 }
834
835 struct symbol * ctype_integer(unsigned long spec)
836 {
837 static struct symbol *const integer_ctypes[][3] = {
838 { &llong_ctype, &sllong_ctype, &ullong_ctype },
839 { &long_ctype, &slong_ctype, &ulong_ctype },
840 { &short_ctype, &sshort_ctype, &ushort_ctype },
841 { &char_ctype, &schar_ctype, &uchar_ctype },
842 { &int_ctype, &sint_ctype, &uint_ctype },
843 };
844 struct symbol *const (*ctype)[3];
845 int sub;
846
847 ctype = integer_ctypes;
848 if (!(spec & MOD_LONGLONG)) {
849 ctype++;
850 if (!(spec & MOD_LONG)) {
851 ctype++;
852 if (!(spec & MOD_SHORT)) {
853 ctype++;
854 if (!(spec & MOD_CHAR))
855 ctype++;
856 }
857 }
858 }
859
860 sub = ((spec & MOD_UNSIGNED)
861 ? 2
862 : ((spec & MOD_EXPLICITLY_SIGNED)
863 ? 1
864 : 0));
865
866 return ctype[0][sub];
867 }
868
869 struct symbol * ctype_fp(unsigned long spec)
870 {
871 if (spec & MOD_LONGLONG)
872 return &ldouble_ctype;
873 if (spec & MOD_LONG)
874 return &double_ctype;
875 return &float_ctype;
876 }
877
878 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype)
879 {
880 unsigned long mod = thistype->modifiers;
881
882 if (mod) {
883 unsigned long old = ctype->modifiers;
884 unsigned long extra = 0, dup, conflict;
885
886 if (mod & old & MOD_LONG) {
887 extra = MOD_LONGLONG | MOD_LONG;
888 mod &= ~MOD_LONG;
889 old &= ~MOD_LONG;
890 }
891 dup = (mod & old) | (extra & old) | (extra & mod);
892 if (dup)
893 sparse_error(pos, "Just how %sdo you want this type to be?",
894 modifier_string(dup));
895
896 conflict = !(~mod & ~old & (MOD_LONG | MOD_SHORT));
897 if (conflict)
898 sparse_error(pos, "You cannot have both long and short modifiers.");
899
900 conflict = !(~mod & ~old & (MOD_SIGNED | MOD_UNSIGNED));
901 if (conflict)
902 sparse_error(pos, "You cannot have both signed and unsigned modifiers.");
903
904 // Only one storage modifier allowed, except that "inline" doesn't count.
905 conflict = (mod | old) & (MOD_STORAGE & ~MOD_INLINE);
906 conflict &= (conflict - 1);
907 if (conflict)
908 sparse_error(pos, "multiple storage classes");
909
910 ctype->modifiers = old | mod | extra;
911 }
912
913 /* Context */
914 concat_ptr_list((struct ptr_list *)thistype->contexts,
915 (struct ptr_list **)&ctype->contexts);
916
917 /* Alignment */
918 if (thistype->alignment & (thistype->alignment-1)) {
919 warning(pos, "I don't like non-power-of-2 alignments");
920 thistype->alignment = 0;
921 }
922 if (thistype->alignment > ctype->alignment)
923 ctype->alignment = thistype->alignment;
924
925 /* Address space */
926 if (thistype->as)
927 ctype->as = thistype->as;
928 }
929
930 static void check_modifiers(struct position *pos, struct symbol *s, unsigned long mod)
931 {
932 unsigned long banned, wrong;
933 const unsigned long BANNED_SIZE = MOD_LONG | MOD_LONGLONG | MOD_SHORT;
934 const unsigned long BANNED_SIGN = MOD_SIGNED | MOD_UNSIGNED;
935
936 if (s->type == SYM_KEYWORD)
937 banned = s->op->type == KW_SPECIFIER ? (BANNED_SIZE | BANNED_SIGN) : 0;
938 else if (s->ctype.base_type == &fp_type)
939 banned = BANNED_SIGN;
940 else if (s->ctype.base_type == &int_type || !s->ctype.base_type || is_int_type (s))
941 banned = 0;
942 else {
943 // label_type
944 // void_type
945 // bad_type
946 // vector_type <-- whatever that is
947 banned = BANNED_SIZE | BANNED_SIGN;
948 }
949
950 wrong = mod & banned;
951 if (wrong)
952 sparse_error(*pos, "modifier %sis invalid in this context",
953 modifier_string (wrong));
954 }
955
956 static struct token *declaration_specifiers(struct token *next, struct ctype *ctype, int qual)
957 {
958 struct token *token;
959
960 while ( (token = next) != NULL ) {
961 struct ctype thistype;
962 struct ident *ident;
963 struct symbol *s, *type;
964 unsigned long mod;
965
966 next = token->next;
967 if (token_type(token) != TOKEN_IDENT)
968 break;
969 ident = token->ident;
970
971 s = lookup_symbol(ident, NS_TYPEDEF);
972 if (!s)
973 break;
974 thistype = s->ctype;
975 mod = thistype.modifiers;
976 if (qual) {
977 if (s->type != SYM_KEYWORD)
978 break;
979 if (!(s->op->type & (KW_ATTRIBUTE | KW_QUALIFIER)))
980 break;
981 }
982 if (s->type == SYM_KEYWORD && s->op->declarator) {
983 next = s->op->declarator(next, &thistype);
984 mod = thistype.modifiers;
985 }
986 type = thistype.base_type;
987 if (type) {
988 if (qual)
989 break;
990 if (ctype->base_type)
991 break;
992 /* User types only mix with qualifiers */
993 if (mod & MOD_USERTYPE) {
994 if (ctype->modifiers & MOD_SPECIFIER)
995 break;
996 }
997 ctype->base_type = type;
998 }
999
1000 check_modifiers(&token->pos, s, ctype->modifiers);
1001 apply_ctype(token->pos, &thistype, ctype);
1002 }
1003
1004 if (!ctype->base_type) {
1005 struct symbol *base = &incomplete_ctype;
1006
1007 /*
1008 * If we have modifiers, we'll default to an integer
1009 * type, and "ctype_integer()" will turn this into
1010 * a specific one.
1011 */
1012 if (ctype->modifiers & MOD_SPECIFIER)
1013 base = &int_type;
1014 ctype->base_type = base;
1015 }
1016 return token;
1017 }
1018
1019 static struct token *abstract_array_declarator(struct token *token, struct symbol *sym)
1020 {
1021 struct expression *expr = NULL;
1022
1023 token = parse_expression(token, &expr);
1024 sym->array_size = expr;
1025 return token;
1026 }
1027
1028 static struct token *parameter_type_list(struct token *, struct symbol *, struct ident **p);
1029 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p);
1030
1031 static struct token *handle_attributes(struct token *token, struct ctype *ctype)
1032 {
1033 for (;;) {
1034 if (token_type(token) != TOKEN_IDENT)
1035 break;
1036 if (match_idents(token, &__attribute___ident, &__attribute_ident, NULL)) {
1037 struct ctype thistype = { 0, };
1038 token = attribute_specifier(token->next, &thistype);
1039 apply_ctype(token->pos, &thistype, ctype);
1040 continue;
1041 }
1042 if (match_idents(token, &asm_ident, &__asm_ident, &__asm___ident, NULL)) {
1043 struct expression *expr;
1044 token = expect(token->next, '(', "after asm");
1045 token = parse_expression(token->next, &expr);
1046 token = expect(token, ')', "after asm");
1047 continue;
1048 }
1049 break;
1050 }
1051 return token;
1052 }
1053
1054 static struct token *direct_declarator(struct token *token, struct symbol *decl, struct ident **p)
1055 {
1056 struct ctype *ctype = &decl->ctype;
1057
1058 if (p && token_type(token) == TOKEN_IDENT) {
1059 *p = token->ident;
1060 token = token->next;
1061 }
1062
1063 for (;;) {
1064 token = handle_attributes(token, ctype);
1065
1066 if (token_type(token) != TOKEN_SPECIAL)
1067 return token;
1068
1069 /*
1070 * This can be either a parameter list or a grouping.
1071 * For the direct (non-abstract) case, we know if must be
1072 * a parameter list if we already saw the identifier.
1073 * For the abstract case, we know if must be a parameter
1074 * list if it is empty or starts with a type.
1075 */
1076 if (token->special == '(') {
1077 struct symbol *sym;
1078 struct token *next = token->next;
1079 int fn = (p && *p) || match_op(next, ')') || lookup_type(next);
1080
1081 if (!fn) {
1082 struct symbol *base_type = ctype->base_type;
1083 token = declarator(next, decl, p);
1084 token = expect(token, ')', "in nested declarator");
1085 while (ctype->base_type != base_type)
1086 ctype = &ctype->base_type->ctype;
1087 p = NULL;
1088 continue;
1089 }
1090
1091 sym = alloc_indirect_symbol(token->pos, ctype, SYM_FN);
1092 token = parameter_type_list(next, sym, p);
1093 token = expect(token, ')', "in function declarator");
1094 continue;
1095 }
1096 if (token->special == '[') {
1097 struct symbol *array = alloc_indirect_symbol(token->pos, ctype, SYM_ARRAY);
1098 token = abstract_array_declarator(token->next, array);
1099 token = expect(token, ']', "in abstract_array_declarator");
1100 ctype = &array->ctype;
1101 continue;
1102 }
1103 break;
1104 }
1105 return token;
1106 }
1107
1108 static struct token *pointer(struct token *token, struct ctype *ctype)
1109 {
1110 unsigned long modifiers;
1111 struct symbol *base_type;
1112
1113 modifiers = ctype->modifiers & ~MOD_TYPEDEF;
1114 base_type = ctype->base_type;
1115 ctype->modifiers = modifiers;
1116
1117 while (match_op(token,'*')) {
1118 struct symbol *ptr = alloc_symbol(token->pos, SYM_PTR);
1119 ptr->ctype.modifiers = modifiers & ~MOD_STORAGE;
1120 ptr->ctype.as = ctype->as;
1121 concat_ptr_list((struct ptr_list *)ctype->contexts,
1122 (struct ptr_list **)&ptr->ctype.contexts);
1123 ptr->ctype.base_type = base_type;
1124
1125 base_type = ptr;
1126 ctype->modifiers = modifiers & MOD_STORAGE;
1127 ctype->base_type = base_type;
1128 ctype->as = 0;
1129 free_ptr_list(&ctype->contexts);
1130
1131 token = declaration_specifiers(token->next, ctype, 1);
1132 modifiers = ctype->modifiers;
1133 }
1134 return token;
1135 }
1136
1137 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p)
1138 {
1139 token = pointer(token, &sym->ctype);
1140 return direct_declarator(token, sym, p);
1141 }
1142
1143 static struct token *handle_bitfield(struct token *token, struct symbol *decl)
1144 {
1145 struct ctype *ctype = &decl->ctype;
1146 struct expression *expr;
1147 struct symbol *bitfield;
1148 long long width;
1149
1150 if (ctype->base_type != &int_type && !is_int_type(ctype->base_type)) {
1151 sparse_error(token->pos, "invalid bitfield specifier for type %s.",
1152 show_typename(ctype->base_type));
1153 // Parse this to recover gracefully.
1154 return conditional_expression(token->next, &expr);
1155 }
1156
1157 bitfield = alloc_indirect_symbol(token->pos, ctype, SYM_BITFIELD);
1158 token = conditional_expression(token->next, &expr);
1159 width = get_expression_value(expr);
1160 bitfield->bit_size = width;
1161
1162 if (width < 0 || width > INT_MAX) {
1163 sparse_error(token->pos, "invalid bitfield width, %lld.", width);
1164 width = -1;
1165 } else if (decl->ident && width == 0) {
1166 sparse_error(token->pos, "invalid named zero-width bitfield `%s'",
1167 show_ident(decl->ident));
1168 width = -1;
1169 } else if (decl->ident) {
1170 struct symbol *base_type = bitfield->ctype.base_type;
1171 struct symbol *bitfield_type = base_type == &int_type ? bitfield : base_type;
1172 int is_signed = !(bitfield_type->ctype.modifiers & MOD_UNSIGNED);
1173 if (Wone_bit_signed_bitfield && width == 1 && is_signed) {
1174 // Valid values are either {-1;0} or {0}, depending on integer
1175 // representation. The latter makes for very efficient code...
1176 sparse_error(token->pos, "dubious one-bit signed bitfield");
1177 }
1178 if (Wdefault_bitfield_sign &&
1179 bitfield_type->type != SYM_ENUM &&
1180 !(bitfield_type->ctype.modifiers & MOD_EXPLICITLY_SIGNED) &&
1181 is_signed) {
1182 // The sign of bitfields is unspecified by default.
1183 sparse_error(token->pos, "dubious bitfield without explicit `signed' or `unsigned'");
1184 }
1185 }
1186 bitfield->bit_size = width;
1187 return token;
1188 }
1189
1190 static struct token *declaration_list(struct token *token, struct symbol_list **list)
1191 {
1192 struct ctype ctype = {0, };
1193
1194 token = declaration_specifiers(token, &ctype, 0);
1195 for (;;) {
1196 struct ident *ident = NULL;
1197 struct symbol *decl = alloc_symbol(token->pos, SYM_NODE);
1198 decl->ctype = ctype;
1199 token = declarator(token, decl, &ident);
1200 decl->ident = ident;
1201 if (match_op(token, ':')) {
1202 token = handle_bitfield(token, decl);
1203 token = handle_attributes(token, &decl->ctype);
1204 }
1205 apply_modifiers(token->pos, &decl->ctype);
1206 add_symbol(list, decl);
1207 if (!match_op(token, ','))
1208 break;
1209 token = token->next;
1210 }
1211 return token;
1212 }
1213
1214 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list)
1215 {
1216 while (!match_op(token, '}')) {
1217 if (!match_op(token, ';'))
1218 token = declaration_list(token, list);
1219 if (!match_op(token, ';')) {
1220 sparse_error(token->pos, "expected ; at end of declaration");
1221 break;
1222 }
1223 token = token->next;
1224 }
1225 return token;
1226 }
1227
1228 static struct token *parameter_declaration(struct token *token, struct symbol **tree)
1229 {
1230 struct ident *ident = NULL;
1231 struct symbol *sym;
1232 struct ctype ctype = { 0, };
1233
1234 token = declaration_specifiers(token, &ctype, 0);
1235 sym = alloc_symbol(token->pos, SYM_NODE);
1236 sym->ctype = ctype;
1237 *tree = sym;
1238 token = declarator(token, sym, &ident);
1239 sym->ident = ident;
1240 apply_modifiers(token->pos, &sym->ctype);
1241 return token;
1242 }
1243
1244 struct token *typename(struct token *token, struct symbol **p)
1245 {
1246 struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
1247 *p = sym;
1248 token = declaration_specifiers(token, &sym->ctype, 0);
1249 token = declarator(token, sym, NULL);
1250 apply_modifiers(token->pos, &sym->ctype);
1251 return token;
1252 }
1253
1254 static struct token *expression_statement(struct token *token, struct expression **tree)
1255 {
1256 token = parse_expression(token, tree);
1257 return expect(token, ';', "at end of statement");
1258 }
1259
1260 static struct token *parse_asm_operands(struct token *token, struct statement *stmt,
1261 struct expression_list **inout)
1262 {
1263 struct expression *expr;
1264
1265 /* Allow empty operands */
1266 if (match_op(token->next, ':') || match_op(token->next, ')'))
1267 return token->next;
1268 do {
1269 struct ident *ident = NULL;
1270 if (match_op(token->next, '[') &&
1271 token_type(token->next->next) == TOKEN_IDENT &&
1272 match_op(token->next->next->next, ']')) {
1273 ident = token->next->next->ident;
1274 token = token->next->next->next;
1275 }
1276 add_expression(inout, (struct expression *)ident); /* UGGLEE!!! */
1277 token = primary_expression(token->next, &expr);
1278 add_expression(inout, expr);
1279 token = parens_expression(token, &expr, "in asm parameter");
1280 add_expression(inout, expr);
1281 } while (match_op(token, ','));
1282 return token;
1283 }
1284
1285 static struct token *parse_asm_clobbers(struct token *token, struct statement *stmt,
1286 struct expression_list **clobbers)
1287 {
1288 struct expression *expr;
1289
1290 do {
1291 token = primary_expression(token->next, &expr);
1292 add_expression(clobbers, expr);
1293 } while (match_op(token, ','));
1294 return token;
1295 }
1296
1297 static struct token *parse_asm(struct token *token, struct statement *stmt)
1298 {
1299 token = token->next;
1300 stmt->type = STMT_ASM;
1301 if (match_idents(token, &__volatile___ident, &__volatile_ident, &volatile_ident, NULL)) {
1302 token = token->next;
1303 }
1304 token = expect(token, '(', "after asm");
1305 token = parse_expression(token, &stmt->asm_string);
1306 if (match_op(token, ':'))
1307 token = parse_asm_operands(token, stmt, &stmt->asm_outputs);
1308 if (match_op(token, ':'))
1309 token = parse_asm_operands(token, stmt, &stmt->asm_inputs);
1310 if (match_op(token, ':'))
1311 token = parse_asm_clobbers(token, stmt, &stmt->asm_clobbers);
1312 token = expect(token, ')', "after asm");
1313 return expect(token, ';', "at end of asm-statement");
1314 }
1315
1316 /* Make a statement out of an expression */
1317 static struct statement *make_statement(struct expression *expr)
1318 {
1319 struct statement *stmt;
1320
1321 if (!expr)
1322 return NULL;
1323 stmt = alloc_statement(expr->pos, STMT_EXPRESSION);
1324 stmt->expression = expr;
1325 return stmt;
1326 }
1327
1328 /*
1329 * All iterators have two symbols associated with them:
1330 * the "continue" and "break" symbols, which are targets
1331 * for continue and break statements respectively.
1332 *
1333 * They are in a special name-space, but they follow
1334 * all the normal visibility rules, so nested iterators
1335 * automatically work right.
1336 */
1337 static void start_iterator(struct statement *stmt)
1338 {
1339 struct symbol *cont, *brk;
1340
1341 start_symbol_scope();
1342 cont = alloc_symbol(stmt->pos, SYM_NODE);
1343 bind_symbol(cont, &continue_ident, NS_ITERATOR);
1344 brk = alloc_symbol(stmt->pos, SYM_NODE);
1345 bind_symbol(brk, &break_ident, NS_ITERATOR);
1346
1347 stmt->type = STMT_ITERATOR;
1348 stmt->iterator_break = brk;
1349 stmt->iterator_continue = cont;
1350 fn_local_symbol(brk);
1351 fn_local_symbol(cont);
1352 }
1353
1354 static void end_iterator(struct statement *stmt)
1355 {
1356 end_symbol_scope();
1357 }
1358
1359 static struct statement *start_function(struct symbol *sym)
1360 {
1361 struct symbol *ret;
1362 struct statement *stmt = alloc_statement(sym->pos, STMT_COMPOUND);
1363
1364 start_function_scope();
1365 ret = alloc_symbol(sym->pos, SYM_NODE);
1366 ret->ctype = sym->ctype.base_type->ctype;
1367 ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
1368 ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
1369 bind_symbol(ret, &return_ident, NS_ITERATOR);
1370 stmt->ret = ret;
1371 fn_local_symbol(ret);
1372
1373 // Currently parsed symbol for __func__/__FUNCTION__/__PRETTY_FUNCTION__
1374 current_fn = sym;
1375
1376 return stmt;
1377 }
1378
1379 static void end_function(struct symbol *sym)
1380 {
1381 current_fn = NULL;
1382 end_function_scope();
1383 }
1384
1385 /*
1386 * A "switch()" statement, like an iterator, has a
1387 * the "break" symbol associated with it. It works
1388 * exactly like the iterator break - it's the target
1389 * for any break-statements in scope, and means that
1390 * "break" handling doesn't even need to know whether
1391 * it's breaking out of an iterator or a switch.
1392 *
1393 * In addition, the "case" symbol is a marker for the
1394 * case/default statements to find the switch statement
1395 * that they are associated with.
1396 */
1397 static void start_switch(struct statement *stmt)
1398 {
1399 struct symbol *brk, *switch_case;
1400
1401 start_symbol_scope();
1402 brk = alloc_symbol(stmt->pos, SYM_NODE);
1403 bind_symbol(brk, &break_ident, NS_ITERATOR);
1404
1405 switch_case = alloc_symbol(stmt->pos, SYM_NODE);
1406 bind_symbol(switch_case, &case_ident, NS_ITERATOR);
1407 switch_case->stmt = stmt;
1408
1409 stmt->type = STMT_SWITCH;
1410 stmt->switch_break = brk;
1411 stmt->switch_case = switch_case;
1412
1413 fn_local_symbol(brk);
1414 fn_local_symbol(switch_case);
1415 }
1416
1417 static void end_switch(struct statement *stmt)
1418 {
1419 if (!stmt->switch_case->symbol_list)
1420 warning(stmt->pos, "switch with no cases");
1421 end_symbol_scope();
1422 }
1423
1424 static void add_case_statement(struct statement *stmt)
1425 {
1426 struct symbol *target = lookup_symbol(&case_ident, NS_ITERATOR);
1427 struct symbol *sym;
1428
1429 if (!target) {
1430 sparse_error(stmt->pos, "not in switch scope");
1431 stmt->type = STMT_NONE;
1432 return;
1433 }
1434 sym = alloc_symbol(stmt->pos, SYM_NODE);
1435 add_symbol(&target->symbol_list, sym);
1436 sym->stmt = stmt;
1437 stmt->case_label = sym;
1438 fn_local_symbol(sym);
1439 }
1440
1441 static struct token *parse_return_statement(struct token *token, struct statement *stmt)
1442 {
1443 struct symbol *target = lookup_symbol(&return_ident, NS_ITERATOR);
1444
1445 if (!target)
1446 error_die(token->pos, "internal error: return without a function target");
1447 stmt->type = STMT_RETURN;
1448 stmt->ret_target = target;
1449 return expression_statement(token->next, &stmt->ret_value);
1450 }
1451
1452 static struct token *parse_for_statement(struct token *token, struct statement *stmt)
1453 {
1454 struct symbol_list *syms;
1455 struct expression *e1, *e2, *e3;
1456 struct statement *iterator;
1457
1458 start_iterator(stmt);
1459 token = expect(token->next, '(', "after 'for'");
1460
1461 syms = NULL;
1462 e1 = NULL;
1463 /* C99 variable declaration? */
1464 if (lookup_type(token)) {
1465 token = external_declaration(token, &syms);
1466 } else {
1467 token = parse_expression(token, &e1);
1468 token = expect(token, ';', "in 'for'");
1469 }
1470 token = parse_expression(token, &e2);
1471 token = expect(token, ';', "in 'for'");
1472 token = parse_expression(token, &e3);
1473 token = expect(token, ')', "in 'for'");
1474 token = statement(token, &iterator);
1475
1476 stmt->iterator_syms = syms;
1477 stmt->iterator_pre_statement = make_statement(e1);
1478 stmt->iterator_pre_condition = e2;
1479 stmt->iterator_post_statement = make_statement(e3);
1480 stmt->iterator_post_condition = NULL;
1481 stmt->iterator_statement = iterator;
1482 end_iterator(stmt);
1483
1484 return token;
1485 }
1486
1487 static struct token *parse_while_statement(struct token *token, struct statement *stmt)
1488 {
1489 struct expression *expr;
1490 struct statement *iterator;
1491
1492 start_iterator(stmt);
1493 token = parens_expression(token->next, &expr, "after 'while'");
1494 token = statement(token, &iterator);
1495
1496 stmt->iterator_pre_condition = expr;
1497 stmt->iterator_post_condition = NULL;
1498 stmt->iterator_statement = iterator;
1499 end_iterator(stmt);
1500
1501 return token;
1502 }
1503
1504 static struct token *parse_do_statement(struct token *token, struct statement *stmt)
1505 {
1506 struct expression *expr;
1507 struct statement *iterator;
1508
1509 start_iterator(stmt);
1510 token = statement(token->next, &iterator);
1511 if (token_type(token) == TOKEN_IDENT && token->ident == &while_ident)
1512 token = token->next;
1513 else
1514 sparse_error(token->pos, "expected 'while' after 'do'");
1515 token = parens_expression(token, &expr, "after 'do-while'");
1516
1517 stmt->iterator_post_condition = expr;
1518 stmt->iterator_statement = iterator;
1519 end_iterator(stmt);
1520
1521 if (iterator && iterator->type != STMT_COMPOUND && Wdo_while)
1522 warning(iterator->pos, "do-while statement is not a compound statement");
1523
1524 return expect(token, ';', "after statement");
1525 }
1526
1527 static struct token *parse_if_statement(struct token *token, struct statement *stmt)
1528 {
1529 stmt->type = STMT_IF;
1530 token = parens_expression(token->next, &stmt->if_conditional, "after if");
1531 token = statement(token, &stmt->if_true);
1532 if (token_type(token) != TOKEN_IDENT)
1533 return token;
1534 if (token->ident != &else_ident)
1535 return token;
1536 return statement(token->next, &stmt->if_false);
1537 }
1538
1539 static inline struct token *case_statement(struct token *token, struct statement *stmt)
1540 {
1541 stmt->type = STMT_CASE;
1542 token = expect(token, ':', "after default/case");
1543 add_case_statement(stmt);
1544 return statement(token, &stmt->case_statement);
1545 }
1546
1547 static struct token *parse_case_statement(struct token *token, struct statement *stmt)
1548 {
1549 token = parse_expression(token->next, &stmt->case_expression);
1550 if (match_op(token, SPECIAL_ELLIPSIS))
1551 token = parse_expression(token->next, &stmt->case_to);
1552 return case_statement(token, stmt);
1553 }
1554
1555 static struct token *parse_default_statement(struct token *token, struct statement *stmt)
1556 {
1557 return case_statement(token->next, stmt);
1558 }
1559
1560 static struct token *parse_loop_iterator(struct token *token, struct statement *stmt)
1561 {
1562 struct symbol *target = lookup_symbol(token->ident, NS_ITERATOR);
1563 stmt->type = STMT_GOTO;
1564 stmt->goto_label = target;
1565 if (!target)
1566 sparse_error(stmt->pos, "break/continue not in iterator scope");
1567 return expect(token->next, ';', "at end of statement");
1568 }
1569
1570 static struct token *parse_switch_statement(struct token *token, struct statement *stmt)
1571 {
1572 stmt->type = STMT_SWITCH;
1573 start_switch(stmt);
1574 token = parens_expression(token->next, &stmt->switch_expression, "after 'switch'");
1575 token = statement(token, &stmt->switch_statement);
1576 end_switch(stmt);
1577 return token;
1578 }
1579
1580 static struct token *parse_goto_statement(struct token *token, struct statement *stmt)
1581 {
1582 stmt->type = STMT_GOTO;
1583 token = token->next;
1584 if (match_op(token, '*')) {
1585 token = parse_expression(token->next, &stmt->goto_expression);
1586 add_statement(&function_computed_goto_list, stmt);
1587 } else if (token_type(token) == TOKEN_IDENT) {
1588 stmt->goto_label = label_symbol(token);
1589 token = token->next;
1590 } else {
1591 sparse_error(token->pos, "Expected identifier or goto expression");
1592 }
1593 return expect(token, ';', "at end of statement");
1594 }
1595
1596 static struct token *parse_context_statement(struct token *token, struct statement *stmt)
1597 {
1598 stmt->type = STMT_CONTEXT;
1599 token = parse_expression(token->next, &stmt->expression);
1600 if(stmt->expression->type == EXPR_PREOP
1601 && stmt->expression->op == '('
1602 && stmt->expression->unop->type == EXPR_COMMA) {
1603 struct expression *expr;
1604 expr = stmt->expression->unop;
1605 stmt->context = expr->left;
1606 stmt->expression = expr->right;
1607 }
1608 return expect(token, ';', "at end of statement");
1609 }
1610
1611 static struct token *parse_range_statement(struct token *token, struct statement *stmt)
1612 {
1613 stmt->type = STMT_RANGE;
1614 token = assignment_expression(token->next, &stmt->range_expression);
1615 token = expect(token, ',', "after range expression");
1616 token = assignment_expression(token, &stmt->range_low);
1617 token = expect(token, ',', "after low range");
1618 token = assignment_expression(token, &stmt->range_high);
1619 return expect(token, ';', "after range statement");
1620 }
1621
1622 static struct token *statement(struct token *token, struct statement **tree)
1623 {
1624 struct statement *stmt = alloc_statement(token->pos, STMT_NONE);
1625
1626 *tree = stmt;
1627 if (token_type(token) == TOKEN_IDENT) {
1628 struct symbol *s = lookup_keyword(token->ident, NS_KEYWORD);
1629 if (s && s->op->statement)
1630 return s->op->statement(token, stmt);
1631
1632 if (match_op(token->next, ':')) {
1633 stmt->type = STMT_LABEL;
1634 stmt->label_identifier = label_symbol(token);
1635 return statement(token->next->next, &stmt->label_statement);
1636 }
1637 }
1638
1639 if (match_op(token, '{')) {
1640 stmt->type = STMT_COMPOUND;
1641 start_symbol_scope();
1642 token = compound_statement(token->next, stmt);
1643 end_symbol_scope();
1644
1645 return expect(token, '}', "at end of compound statement");
1646 }
1647
1648 stmt->type = STMT_EXPRESSION;
1649 return expression_statement(token, &stmt->expression);
1650 }
1651
1652 static struct token * statement_list(struct token *token, struct statement_list **list)
1653 {
1654 int seen_statement = 0;
1655 for (;;) {
1656 struct statement * stmt;
1657 if (eof_token(token))
1658 break;
1659 if (match_op(token, '}'))
1660 break;
1661 if (lookup_type(token)) {
1662 if (seen_statement) {
1663 warning(token->pos, "mixing declarations and code");
1664 seen_statement = 0;
1665 }
1666 stmt = alloc_statement(token->pos, STMT_DECLARATION);
1667 token = external_declaration(token, &stmt->declaration);
1668 } else {
1669 seen_statement = warn_on_mixed;
1670 token = statement(token, &stmt);
1671 }
1672 add_statement(list, stmt);
1673 }
1674 return token;
1675 }
1676
1677 static struct token *parameter_type_list(struct token *token, struct symbol *fn, struct ident **p)
1678 {
1679 struct symbol_list **list = &fn->arguments;
1680
1681 if (match_op(token, ')')) {
1682 // No warning for "void oink ();"
1683 // Bug or feature: warns for "void oink () __attribute__ ((noreturn));"
1684 if (p && !match_op(token->next, ';'))
1685 warning(token->pos, "non-ANSI function declaration of function '%s'", show_ident(*p));
1686 return token;
1687 }
1688
1689 for (;;) {
1690 struct symbol *sym;
1691
1692 if (match_op(token, SPECIAL_ELLIPSIS)) {
1693 if (!*list)
1694 warning(token->pos, "variadic functions must have one named argument");
1695 fn->variadic = 1;
1696 token = token->next;
1697 break;
1698 }
1699
1700 sym = alloc_symbol(token->pos, SYM_NODE);
1701 token = parameter_declaration(token, &sym);
1702 if (sym->ctype.base_type == &void_ctype) {
1703 /* Special case: (void) */
1704 if (!*list && !sym->ident)
1705 break;
1706 warning(token->pos, "void parameter");
1707 }
1708 add_symbol(list, sym);
1709 if (!match_op(token, ','))
1710 break;
1711 token = token->next;
1712
1713 }
1714 return token;
1715 }
1716
1717 struct token *compound_statement(struct token *token, struct statement *stmt)
1718 {
1719 token = statement_list(token, &stmt->stmts);
1720 return token;
1721 }
1722
1723 static struct expression *identifier_expression(struct token *token)
1724 {
1725 struct expression *expr = alloc_expression(token->pos, EXPR_IDENTIFIER);
1726 expr->expr_ident = token->ident;
1727 return expr;
1728 }
1729
1730 static struct expression *index_expression(struct expression *from, struct expression *to)
1731 {
1732 int idx_from, idx_to;
1733 struct expression *expr = alloc_expression(from->pos, EXPR_INDEX);
1734
1735 idx_from = get_expression_value(from);
1736 idx_to = idx_from;
1737 if (to) {
1738 idx_to = get_expression_value(to);
1739 if (idx_to < idx_from || idx_from < 0)
1740 warning(from->pos, "nonsense array initializer index range");
1741 }
1742 expr->idx_from = idx_from;
1743 expr->idx_to = idx_to;
1744 return expr;
1745 }
1746
1747 static struct token *single_initializer(struct expression **ep, struct token *token)
1748 {
1749 int expect_equal = 0;
1750 struct token *next = token->next;
1751 struct expression **tail = ep;
1752 int nested;
1753
1754 *ep = NULL;
1755
1756 if ((token_type(token) == TOKEN_IDENT) && match_op(next, ':')) {
1757 struct expression *expr = identifier_expression(token);
1758 warning(token->pos, "obsolete struct initializer, use C99 syntax");
1759 token = initializer(&expr->ident_expression, next->next);
1760 if (expr->ident_expression)
1761 *ep = expr;
1762 return token;
1763 }
1764
1765 for (tail = ep, nested = 0; ; nested++, next = token->next) {
1766 if (match_op(token, '.') && (token_type(next) == TOKEN_IDENT)) {
1767 struct expression *expr = identifier_expression(next);
1768 *tail = expr;
1769 tail = &expr->ident_expression;
1770 expect_equal = 1;
1771 token = next->next;
1772 } else if (match_op(token, '[')) {
1773 struct expression *from = NULL, *to = NULL, *expr;
1774 token = constant_expression(token->next, &from);
1775 if (!from) {
1776 sparse_error(token->pos, "Expected constant expression");
1777 break;
1778 }
1779 if (match_op(token, SPECIAL_ELLIPSIS))
1780 token = constant_expression(token->next, &to);
1781 expr = index_expression(from, to);
1782 *tail = expr;
1783 tail = &expr->idx_expression;
1784 token = expect(token, ']', "at end of initializer index");
1785 if (nested)
1786 expect_equal = 1;
1787 } else {
1788 break;
1789 }
1790 }
1791 if (nested && !expect_equal) {
1792 if (!match_op(token, '='))
1793 warning(token->pos, "obsolete array initializer, use C99 syntax");
1794 else
1795 expect_equal = 1;
1796 }
1797 if (expect_equal)
1798 token = expect(token, '=', "at end of initializer index");
1799
1800 token = initializer(tail, token);
1801 if (!*tail)
1802 *ep = NULL;
1803 return token;
1804 }
1805
1806 static struct token *initializer_list(struct expression_list **list, struct token *token)
1807 {
1808 struct expression *expr;
1809
1810 for (;;) {
1811 token = single_initializer(&expr, token);
1812 if (!expr)
1813 break;
1814 add_expression(list, expr);
1815 if (!match_op(token, ','))
1816 break;
1817 token = token->next;
1818 }
1819 return token;
1820 }
1821
1822 struct token *initializer(struct expression **tree, struct token *token)
1823 {
1824 if (match_op(token, '{')) {
1825 struct expression *expr = alloc_expression(token->pos, EXPR_INITIALIZER);
1826 *tree = expr;
1827 token = initializer_list(&expr->expr_list, token->next);
1828 return expect(token, '}', "at end of initializer");
1829 }
1830 return assignment_expression(token, tree);
1831 }
1832
1833 static void declare_argument(struct symbol *sym, struct symbol *fn)
1834 {
1835 if (!sym->ident) {
1836 sparse_error(sym->pos, "no identifier for function argument");
1837 return;
1838 }
1839 bind_symbol(sym, sym->ident, NS_SYMBOL);
1840 }
1841
1842 static struct token *parse_function_body(struct token *token, struct symbol *decl,
1843 struct symbol_list **list)
1844 {
1845 struct symbol_list **old_symbol_list;
1846 struct symbol *base_type = decl->ctype.base_type;
1847 struct statement *stmt, **p;
1848 struct symbol *arg;
1849
1850 old_symbol_list = function_symbol_list;
1851 if (decl->ctype.modifiers & MOD_INLINE) {
1852 function_symbol_list = &decl->inline_symbol_list;
1853 p = &base_type->inline_stmt;
1854 } else {
1855 function_symbol_list = &decl->symbol_list;
1856 p = &base_type->stmt;
1857 }
1858 function_computed_target_list = NULL;
1859 function_computed_goto_list = NULL;
1860
1861 if (decl->ctype.modifiers & MOD_EXTERN) {
1862 if (!(decl->ctype.modifiers & MOD_INLINE))
1863 warning(decl->pos, "function '%s' with external linkage has definition", show_ident(decl->ident));
1864 }
1865 if (!(decl->ctype.modifiers & MOD_STATIC))
1866 decl->ctype.modifiers |= MOD_EXTERN;
1867
1868 stmt = start_function(decl);
1869
1870 *p = stmt;
1871 FOR_EACH_PTR (base_type->arguments, arg) {
1872 declare_argument(arg, base_type);
1873 } END_FOR_EACH_PTR(arg);
1874
1875 token = compound_statement(token->next, stmt);
1876
1877 end_function(decl);
1878 if (!(decl->ctype.modifiers & MOD_INLINE))
1879 add_symbol(list, decl);
1880 check_declaration(decl);
1881 function_symbol_list = old_symbol_list;
1882 if (function_computed_goto_list) {
1883 if (!function_computed_target_list)
1884 warning(decl->pos, "function '%s' has computed goto but no targets?", show_ident(decl->ident));
1885 else {
1886 struct statement *stmt;
1887 FOR_EACH_PTR(function_computed_goto_list, stmt) {
1888 stmt->target_list = function_computed_target_list;
1889 } END_FOR_EACH_PTR(stmt);
1890 }
1891 }
1892 return expect(token, '}', "at end of function");
1893 }
1894
1895 static void promote_k_r_types(struct symbol *arg)
1896 {
1897 struct symbol *base = arg->ctype.base_type;
1898 if (base && base->ctype.base_type == &int_type && (base->ctype.modifiers & (MOD_CHAR | MOD_SHORT))) {
1899 arg->ctype.base_type = &int_ctype;
1900 }
1901 }
1902
1903 static void apply_k_r_types(struct symbol_list *argtypes, struct symbol *fn)
1904 {
1905 struct symbol_list *real_args = fn->ctype.base_type->arguments;
1906 struct symbol *arg;
1907
1908 FOR_EACH_PTR(real_args, arg) {
1909 struct symbol *type;
1910
1911 /* This is quadratic in the number of arguments. We _really_ don't care */
1912 FOR_EACH_PTR(argtypes, type) {
1913 if (type->ident == arg->ident)
1914 goto match;
1915 } END_FOR_EACH_PTR(type);
1916 sparse_error(arg->pos, "missing type declaration for parameter '%s'", show_ident(arg->ident));
1917 continue;
1918 match:
1919 type->used = 1;
1920 /* "char" and "short" promote to "int" */
1921 promote_k_r_types(type);
1922
1923 arg->ctype = type->ctype;
1924 } END_FOR_EACH_PTR(arg);
1925
1926 FOR_EACH_PTR(argtypes, arg) {
1927 if (!arg->used)
1928 warning(arg->pos, "nonsensical parameter declaration '%s'", show_ident(arg->ident));
1929 } END_FOR_EACH_PTR(arg);
1930
1931 }
1932
1933 static struct token *parse_k_r_arguments(struct token *token, struct symbol *decl,
1934 struct symbol_list **list)
1935 {
1936 struct symbol_list *args = NULL;
1937
1938 warning(token->pos, "non-ANSI definition of function '%s'", show_ident(decl->ident));
1939 do {
1940 token = declaration_list(token, &args);
1941 if (!match_op(token, ';')) {
1942 sparse_error(token->pos, "expected ';' at end of parameter declaration");
1943 break;
1944 }
1945 token = token->next;
1946 } while (lookup_type(token));
1947
1948 apply_k_r_types(args, decl);
1949
1950 if (!match_op(token, '{')) {
1951 sparse_error(token->pos, "expected function body");
1952 return token;
1953 }
1954 return parse_function_body(token, decl, list);
1955 }
1956
1957 static struct token *toplevel_asm_declaration(struct token *token, struct symbol_list **list)
1958 {
1959 struct symbol *anon = alloc_symbol(token->pos, SYM_NODE);
1960 struct symbol *fn = alloc_symbol(token->pos, SYM_FN);
1961 struct statement *stmt;
1962
1963 anon->ctype.base_type = fn;
1964 stmt = alloc_statement(token->pos, STMT_NONE);
1965 fn->stmt = stmt;
1966
1967 token = parse_asm(token, stmt);
1968
1969 add_symbol(list, anon);
1970 return token;
1971 }
1972
1973 struct token *external_declaration(struct token *token, struct symbol_list **list)
1974 {
1975 struct ident *ident = NULL;
1976 struct symbol *decl;
1977 struct ctype ctype = { 0, };
1978 struct symbol *base_type;
1979 int is_typedef;
1980
1981 /* Top-level inline asm? */
1982 if (token_type(token) == TOKEN_IDENT) {
1983 struct symbol *s = lookup_keyword(token->ident, NS_KEYWORD);
1984 if (s && s->op->toplevel)
1985 return s->op->toplevel(token, list);
1986 }
1987
1988 /* Parse declaration-specifiers, if any */
1989 token = declaration_specifiers(token, &ctype, 0);
1990 decl = alloc_symbol(token->pos, SYM_NODE);
1991 decl->ctype = ctype;
1992 token = declarator(token, decl, &ident);
1993 apply_modifiers(token->pos, &decl->ctype);
1994
1995 /* Just a type declaration? */
1996 if (!ident)
1997 return expect(token, ';', "end of type declaration");
1998
1999 /* type define declaration? */
2000 is_typedef = (ctype.modifiers & MOD_TYPEDEF) != 0;
2001
2002 /* Typedef's don't have meaningful storage */
2003 if (is_typedef) {
2004 ctype.modifiers &= ~MOD_STORAGE;
2005 decl->ctype.modifiers &= ~MOD_STORAGE;
2006 decl->ctype.modifiers |= MOD_USERTYPE;
2007 }
2008
2009 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
2010
2011 base_type = decl->ctype.base_type;
2012
2013 if (is_typedef) {
2014 if (base_type && !base_type->ident)
2015 base_type->ident = ident;
2016 } else if (base_type && base_type->type == SYM_FN) {
2017 /* K&R argument declaration? */
2018 if (lookup_type(token))
2019 return parse_k_r_arguments(token, decl, list);
2020 if (match_op(token, '{'))
2021 return parse_function_body(token, decl, list);
2022
2023 if (!(decl->ctype.modifiers & MOD_STATIC))
2024 decl->ctype.modifiers |= MOD_EXTERN;
2025 } else if (base_type == &void_ctype && !(decl->ctype.modifiers & MOD_EXTERN)) {
2026 sparse_error(token->pos, "void declaration");
2027 }
2028
2029 for (;;) {
2030 if (!is_typedef && match_op(token, '=')) {
2031 if (decl->ctype.modifiers & MOD_EXTERN) {
2032 warning(decl->pos, "symbol with external linkage has initializer");
2033 decl->ctype.modifiers &= ~MOD_EXTERN;
2034 }
2035 token = initializer(&decl->initializer, token->next);
2036 }
2037 if (!is_typedef) {
2038 if (!(decl->ctype.modifiers & (MOD_EXTERN | MOD_INLINE))) {
2039 add_symbol(list, decl);
2040 fn_local_symbol(decl);
2041 }
2042 }
2043 check_declaration(decl);
2044
2045 if (!match_op(token, ','))
2046 break;
2047
2048 token = token->next;
2049 ident = NULL;
2050 decl = alloc_symbol(token->pos, SYM_NODE);
2051 decl->ctype = ctype;
2052 token = declaration_specifiers(token, &decl->ctype, 1);
2053 token = declarator(token, decl, &ident);
2054 apply_modifiers(token->pos, &decl->ctype);
2055 if (!ident) {
2056 sparse_error(token->pos, "expected identifier name in type definition");
2057 return token;
2058 }
2059
2060 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
2061
2062 /* Function declarations are automatically extern unless specifically static */
2063 base_type = decl->ctype.base_type;
2064 if (!is_typedef && base_type && base_type->type == SYM_FN) {
2065 if (!(decl->ctype.modifiers & MOD_STATIC))
2066 decl->ctype.modifiers |= MOD_EXTERN;
2067 }
2068 }
2069 return expect(token, ';', "at end of declaration");
2070 }
Static analysis for C