?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
introduce: get_dinfo()
[smatch.git] / parse.c
blobb5aced14cd51c3341f503f4abbaa4b0cba0dbb0d
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 static struct symbol_list **function_symbol_list;
32 struct symbol_list *function_computed_target_list;
33 struct statement_list *function_computed_goto_list;
34
35 static struct token *statement(struct token *token, struct statement **tree);
36 static struct token *handle_attributes(struct token *token, struct decl_state *ctx, unsigned int keywords);
37
38 typedef struct token *declarator_t(struct token *, struct decl_state *);
39 static declarator_t
40 struct_specifier, union_specifier, enum_specifier,
41 attribute_specifier, typeof_specifier, parse_asm_declarator,
42 typedef_specifier, inline_specifier, auto_specifier,
43 register_specifier, static_specifier, extern_specifier,
44 thread_specifier, const_qualifier, volatile_qualifier;
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_statement(struct token *token, struct statement *stmt);
59 static struct token *toplevel_asm_declaration(struct token *token, struct symbol_list **list);
60
61 typedef struct token *attr_t(struct token *, struct symbol *,
62 struct decl_state *);
63
64 static attr_t
65 attribute_packed, attribute_aligned, attribute_modifier,
66 attribute_address_space, attribute_context,
67 attribute_designated_init,
68 attribute_transparent_union, ignore_attribute,
69 attribute_mode, attribute_force;
70
71 typedef struct symbol *to_mode_t(struct symbol *);
72
73 static to_mode_t
74 to_QI_mode, to_HI_mode, to_SI_mode, to_DI_mode, to_TI_mode, to_word_mode;
75
76 enum {
77 Set_T = 1,
78 Set_S = 2,
79 Set_Char = 4,
80 Set_Int = 8,
81 Set_Double = 16,
82 Set_Float = 32,
83 Set_Signed = 64,
84 Set_Unsigned = 128,
85 Set_Short = 256,
86 Set_Long = 512,
87 Set_Vlong = 1024,
88 Set_Any = Set_T | Set_Short | Set_Long | Set_Signed | Set_Unsigned
89 };
90
91 enum {
92 CInt = 0, CSInt, CUInt, CReal, CChar, CSChar, CUChar
93 };
94
95 enum {
96 SNone = 0, STypedef, SAuto, SRegister, SExtern, SStatic, SForced
97 };
98
99 static struct symbol_op typedef_op = {
100 .type = KW_MODIFIER,
101 .declarator = typedef_specifier,
102 };
103
104 static struct symbol_op inline_op = {
105 .type = KW_MODIFIER,
106 .declarator = inline_specifier,
107 };
108
109 static struct symbol_op auto_op = {
110 .type = KW_MODIFIER,
111 .declarator = auto_specifier,
112 };
113
114 static struct symbol_op register_op = {
115 .type = KW_MODIFIER,
116 .declarator = register_specifier,
117 };
118
119 static struct symbol_op static_op = {
120 .type = KW_MODIFIER,
121 .declarator = static_specifier,
122 };
123
124 static struct symbol_op extern_op = {
125 .type = KW_MODIFIER,
126 .declarator = extern_specifier,
127 };
128
129 static struct symbol_op thread_op = {
130 .type = KW_MODIFIER,
131 .declarator = thread_specifier,
132 };
133
134 static struct symbol_op const_op = {
135 .type = KW_QUALIFIER,
136 .declarator = const_qualifier,
137 };
138
139 static struct symbol_op volatile_op = {
140 .type = KW_QUALIFIER,
141 .declarator = volatile_qualifier,
142 };
143
144 static struct symbol_op restrict_op = {
145 .type = KW_QUALIFIER,
146 };
147
148 static struct symbol_op typeof_op = {
149 .type = KW_SPECIFIER,
150 .declarator = typeof_specifier,
151 .test = Set_Any,
152 .set = Set_S|Set_T,
153 };
154
155 static struct symbol_op attribute_op = {
156 .type = KW_ATTRIBUTE,
157 .declarator = attribute_specifier,
158 };
159
160 static struct symbol_op struct_op = {
161 .type = KW_SPECIFIER,
162 .declarator = struct_specifier,
163 .test = Set_Any,
164 .set = Set_S|Set_T,
165 };
166
167 static struct symbol_op union_op = {
168 .type = KW_SPECIFIER,
169 .declarator = union_specifier,
170 .test = Set_Any,
171 .set = Set_S|Set_T,
172 };
173
174 static struct symbol_op enum_op = {
175 .type = KW_SPECIFIER,
176 .declarator = enum_specifier,
177 .test = Set_Any,
178 .set = Set_S|Set_T,
179 };
180
181 static struct symbol_op spec_op = {
182 .type = KW_SPECIFIER | KW_EXACT,
183 .test = Set_Any,
184 .set = Set_S|Set_T,
185 };
186
187 static struct symbol_op char_op = {
188 .type = KW_SPECIFIER,
189 .test = Set_T|Set_Long|Set_Short,
190 .set = Set_T|Set_Char,
191 .class = CChar,
192 };
193
194 static struct symbol_op int_op = {
195 .type = KW_SPECIFIER,
196 .test = Set_T,
197 .set = Set_T|Set_Int,
198 };
199
200 static struct symbol_op double_op = {
201 .type = KW_SPECIFIER,
202 .test = Set_T|Set_Signed|Set_Unsigned|Set_Short|Set_Vlong,
203 .set = Set_T|Set_Double,
204 .class = CReal,
205 };
206
207 static struct symbol_op float_op = {
208 .type = KW_SPECIFIER | KW_SHORT,
209 .test = Set_T|Set_Signed|Set_Unsigned|Set_Short|Set_Long,
210 .set = Set_T|Set_Float,
211 .class = CReal,
212 };
213
214 static struct symbol_op short_op = {
215 .type = KW_SPECIFIER | KW_SHORT,
216 .test = Set_S|Set_Char|Set_Float|Set_Double|Set_Long|Set_Short,
217 .set = Set_Short,
218 };
219
220 static struct symbol_op signed_op = {
221 .type = KW_SPECIFIER,
222 .test = Set_S|Set_Float|Set_Double|Set_Signed|Set_Unsigned,
223 .set = Set_Signed,
224 .class = CSInt,
225 };
226
227 static struct symbol_op unsigned_op = {
228 .type = KW_SPECIFIER,
229 .test = Set_S|Set_Float|Set_Double|Set_Signed|Set_Unsigned,
230 .set = Set_Unsigned,
231 .class = CUInt,
232 };
233
234 static struct symbol_op long_op = {
235 .type = KW_SPECIFIER | KW_LONG,
236 .test = Set_S|Set_Char|Set_Float|Set_Short|Set_Vlong,
237 .set = Set_Long,
238 };
239
240 static struct symbol_op if_op = {
241 .statement = parse_if_statement,
242 };
243
244 static struct symbol_op return_op = {
245 .statement = parse_return_statement,
246 };
247
248 static struct symbol_op loop_iter_op = {
249 .statement = parse_loop_iterator,
250 };
251
252 static struct symbol_op default_op = {
253 .statement = parse_default_statement,
254 };
255
256 static struct symbol_op case_op = {
257 .statement = parse_case_statement,
258 };
259
260 static struct symbol_op switch_op = {
261 .statement = parse_switch_statement,
262 };
263
264 static struct symbol_op for_op = {
265 .statement = parse_for_statement,
266 };
267
268 static struct symbol_op while_op = {
269 .statement = parse_while_statement,
270 };
271
272 static struct symbol_op do_op = {
273 .statement = parse_do_statement,
274 };
275
276 static struct symbol_op goto_op = {
277 .statement = parse_goto_statement,
278 };
279
280 static struct symbol_op __context___op = {
281 .statement = parse_context_statement,
282 };
283
284 static struct symbol_op range_op = {
285 .statement = parse_range_statement,
286 };
287
288 static struct symbol_op asm_op = {
289 .type = KW_ASM,
290 .declarator = parse_asm_declarator,
291 .statement = parse_asm_statement,
292 .toplevel = toplevel_asm_declaration,
293 };
294
295 static struct symbol_op packed_op = {
296 .attribute = attribute_packed,
297 };
298
299 static struct symbol_op aligned_op = {
300 .attribute = attribute_aligned,
301 };
302
303 static struct symbol_op attr_mod_op = {
304 .attribute = attribute_modifier,
305 };
306
307 static struct symbol_op attr_force_op = {
308 .attribute = attribute_force,
309 };
310
311 static struct symbol_op address_space_op = {
312 .attribute = attribute_address_space,
313 };
314
315 static struct symbol_op mode_op = {
316 .attribute = attribute_mode,
317 };
318
319 static struct symbol_op context_op = {
320 .attribute = attribute_context,
321 };
322
323 static struct symbol_op designated_init_op = {
324 .attribute = attribute_designated_init,
325 };
326
327 static struct symbol_op transparent_union_op = {
328 .attribute = attribute_transparent_union,
329 };
330
331 static struct symbol_op ignore_attr_op = {
332 .attribute = ignore_attribute,
333 };
334
335 static struct symbol_op mode_QI_op = {
336 .type = KW_MODE,
337 .to_mode = to_QI_mode
338 };
339
340 static struct symbol_op mode_HI_op = {
341 .type = KW_MODE,
342 .to_mode = to_HI_mode
343 };
344
345 static struct symbol_op mode_SI_op = {
346 .type = KW_MODE,
347 .to_mode = to_SI_mode
348 };
349
350 static struct symbol_op mode_DI_op = {
351 .type = KW_MODE,
352 .to_mode = to_DI_mode
353 };
354
355 static struct symbol_op mode_TI_op = {
356 .type = KW_MODE,
357 .to_mode = to_TI_mode
358 };
359
360 static struct symbol_op mode_word_op = {
361 .type = KW_MODE,
362 .to_mode = to_word_mode
363 };
364
365 static struct init_keyword {
366 const char *name;
367 enum namespace ns;
368 unsigned long modifiers;
369 struct symbol_op *op;
370 struct symbol *type;
371 } keyword_table[] = {
372 /* Type qualifiers */
373 { "const", NS_TYPEDEF, .op = &const_op },
374 { "__const", NS_TYPEDEF, .op = &const_op },
375 { "__const__", NS_TYPEDEF, .op = &const_op },
376 { "volatile", NS_TYPEDEF, .op = &volatile_op },
377 { "__volatile", NS_TYPEDEF, .op = &volatile_op },
378 { "__volatile__", NS_TYPEDEF, .op = &volatile_op },
379
380 /* Typedef.. */
381 { "typedef", NS_TYPEDEF, .op = &typedef_op },
382
383 /* Type specifiers */
384 { "void", NS_TYPEDEF, .type = &void_ctype, .op = &spec_op},
385 { "char", NS_TYPEDEF, .op = &char_op },
386 { "short", NS_TYPEDEF, .op = &short_op },
387 { "int", NS_TYPEDEF, .op = &int_op },
388 { "long", NS_TYPEDEF, .op = &long_op },
389 { "float", NS_TYPEDEF, .op = &float_op },
390 { "double", NS_TYPEDEF, .op = &double_op },
391 { "signed", NS_TYPEDEF, .op = &signed_op },
392 { "__signed", NS_TYPEDEF, .op = &signed_op },
393 { "__signed__", NS_TYPEDEF, .op = &signed_op },
394 { "unsigned", NS_TYPEDEF, .op = &unsigned_op },
395 { "_Bool", NS_TYPEDEF, .type = &bool_ctype, .op = &spec_op },
396
397 /* Predeclared types */
398 { "__builtin_va_list", NS_TYPEDEF, .type = &ptr_ctype, .op = &spec_op },
399 { "__builtin_ms_va_list", NS_TYPEDEF, .type = &ptr_ctype, .op = &spec_op },
400 { "__int128_t", NS_TYPEDEF, .type = &lllong_ctype, .op = &spec_op },
401 { "__uint128_t",NS_TYPEDEF, .type = &ulllong_ctype, .op = &spec_op },
402
403 /* Extended types */
404 { "typeof", NS_TYPEDEF, .op = &typeof_op },
405 { "__typeof", NS_TYPEDEF, .op = &typeof_op },
406 { "__typeof__", NS_TYPEDEF, .op = &typeof_op },
407
408 { "__attribute", NS_TYPEDEF, .op = &attribute_op },
409 { "__attribute__", NS_TYPEDEF, .op = &attribute_op },
410
411 { "struct", NS_TYPEDEF, .op = &struct_op },
412 { "union", NS_TYPEDEF, .op = &union_op },
413 { "enum", NS_TYPEDEF, .op = &enum_op },
414
415 { "inline", NS_TYPEDEF, .op = &inline_op },
416 { "__inline", NS_TYPEDEF, .op = &inline_op },
417 { "__inline__", NS_TYPEDEF, .op = &inline_op },
418
419 /* Ignored for now.. */
420 { "restrict", NS_TYPEDEF, .op = &restrict_op},
421 { "__restrict", NS_TYPEDEF, .op = &restrict_op},
422
423 /* Storage class */
424 { "auto", NS_TYPEDEF, .op = &auto_op },
425 { "register", NS_TYPEDEF, .op = &register_op },
426 { "static", NS_TYPEDEF, .op = &static_op },
427 { "extern", NS_TYPEDEF, .op = &extern_op },
428 { "__thread", NS_TYPEDEF, .op = &thread_op },
429
430 /* Statement */
431 { "if", NS_KEYWORD, .op = &if_op },
432 { "return", NS_KEYWORD, .op = &return_op },
433 { "break", NS_KEYWORD, .op = &loop_iter_op },
434 { "continue", NS_KEYWORD, .op = &loop_iter_op },
435 { "default", NS_KEYWORD, .op = &default_op },
436 { "case", NS_KEYWORD, .op = &case_op },
437 { "switch", NS_KEYWORD, .op = &switch_op },
438 { "for", NS_KEYWORD, .op = &for_op },
439 { "while", NS_KEYWORD, .op = &while_op },
440 { "do", NS_KEYWORD, .op = &do_op },
441 { "goto", NS_KEYWORD, .op = &goto_op },
442 { "__context__",NS_KEYWORD, .op = &__context___op },
443 { "__range__", NS_KEYWORD, .op = &range_op },
444 { "asm", NS_KEYWORD, .op = &asm_op },
445 { "__asm", NS_KEYWORD, .op = &asm_op },
446 { "__asm__", NS_KEYWORD, .op = &asm_op },
447
448 /* Attribute */
449 { "packed", NS_KEYWORD, .op = &packed_op },
450 { "__packed__", NS_KEYWORD, .op = &packed_op },
451 { "aligned", NS_KEYWORD, .op = &aligned_op },
452 { "__aligned__",NS_KEYWORD, .op = &aligned_op },
453 { "nocast", NS_KEYWORD, MOD_NOCAST, .op = &attr_mod_op },
454 { "noderef", NS_KEYWORD, MOD_NODEREF, .op = &attr_mod_op },
455 { "safe", NS_KEYWORD, MOD_SAFE, .op = &attr_mod_op },
456 { "force", NS_KEYWORD, .op = &attr_force_op },
457 { "bitwise", NS_KEYWORD, MOD_BITWISE, .op = &attr_mod_op },
458 { "__bitwise__",NS_KEYWORD, MOD_BITWISE, .op = &attr_mod_op },
459 { "address_space",NS_KEYWORD, .op = &address_space_op },
460 { "mode", NS_KEYWORD, .op = &mode_op },
461 { "context", NS_KEYWORD, .op = &context_op },
462 { "designated_init", NS_KEYWORD, .op = &designated_init_op },
463 { "__transparent_union__", NS_KEYWORD, .op = &transparent_union_op },
464 { "noreturn", NS_KEYWORD, MOD_NORETURN, .op = &attr_mod_op },
465 { "__noreturn__", NS_KEYWORD, MOD_NORETURN, .op = &attr_mod_op },
466
467 { "__mode__", NS_KEYWORD, .op = &mode_op },
468 { "QI", NS_KEYWORD, MOD_CHAR, .op = &mode_QI_op },
469 { "__QI__", NS_KEYWORD, MOD_CHAR, .op = &mode_QI_op },
470 { "HI", NS_KEYWORD, MOD_SHORT, .op = &mode_HI_op },
471 { "__HI__", NS_KEYWORD, MOD_SHORT, .op = &mode_HI_op },
472 { "SI", NS_KEYWORD, .op = &mode_SI_op },
473 { "__SI__", NS_KEYWORD, .op = &mode_SI_op },
474 { "DI", NS_KEYWORD, MOD_LONGLONG, .op = &mode_DI_op },
475 { "__DI__", NS_KEYWORD, MOD_LONGLONG, .op = &mode_DI_op },
476 { "TI", NS_KEYWORD, MOD_LONGLONGLONG, .op = &mode_TI_op },
477 { "__TI__", NS_KEYWORD, MOD_LONGLONGLONG, .op = &mode_TI_op },
478 { "word", NS_KEYWORD, MOD_LONG, .op = &mode_word_op },
479 { "__word__", NS_KEYWORD, MOD_LONG, .op = &mode_word_op },
480 };
481
482 const char *ignored_attributes[] = {
483 "alias",
484 "__alias__",
485 "alloc_size",
486 "__alloc_size__",
487 "always_inline",
488 "__always_inline__",
489 "bounded",
490 "__bounded__",
491 "cdecl",
492 "__cdecl__",
493 "cold",
494 "__cold__",
495 "const",
496 "__const",
497 "__const__",
498 "constructor",
499 "__constructor__",
500 "deprecated",
501 "__deprecated__",
502 "destructor",
503 "__destructor__",
504 "dllexport",
505 "__dllexport__",
506 "dllimport",
507 "__dllimport__",
508 "externally_visible",
509 "__externally_visible__",
510 "fastcall",
511 "__fastcall__",
512 "format",
513 "__format__",
514 "format_arg",
515 "__format_arg__",
516 "hot",
517 "__hot__",
518 "malloc",
519 "__malloc__",
520 "model",
521 "__model__",
522 "ms_abi",
523 "__ms_abi__",
524 "no_instrument_function",
525 "__no_instrument_function__",
526 "noinline",
527 "__noinline__",
528 "nonnull",
529 "__nonnull",
530 "__nonnull__",
531 "nothrow",
532 "__nothrow",
533 "__nothrow__",
534 "pure",
535 "__pure__",
536 "regparm",
537 "__regparm__",
538 "section",
539 "__section__",
540 "sentinel",
541 "__sentinel__",
542 "signal",
543 "__signal__",
544 "stdcall",
545 "__stdcall__",
546 "syscall_linkage",
547 "__syscall_linkage__",
548 "sysv_abi",
549 "__sysv_abi__",
550 "unused",
551 "__unused__",
552 "used",
553 "__used__",
554 "visibility",
555 "__visibility__",
556 "warn_unused_result",
557 "__warn_unused_result__",
558 "warning",
559 "__warning__",
560 "weak",
561 "__weak__",
562 };
563
564
565 void init_parser(int stream)
566 {
567 int i;
568 for (i = 0; i < ARRAY_SIZE(keyword_table); i++) {
569 struct init_keyword *ptr = keyword_table + i;
570 struct symbol *sym = create_symbol(stream, ptr->name, SYM_KEYWORD, ptr->ns);
571 sym->ident->keyword = 1;
572 if (ptr->ns == NS_TYPEDEF)
573 sym->ident->reserved = 1;
574 sym->ctype.modifiers = ptr->modifiers;
575 sym->ctype.base_type = ptr->type;
576 sym->op = ptr->op;
577 }
578
579 for (i = 0; i < ARRAY_SIZE(ignored_attributes); i++) {
580 const char * name = ignored_attributes[i];
581 struct symbol *sym = create_symbol(stream, name, SYM_KEYWORD,
582 NS_KEYWORD);
583 sym->ident->keyword = 1;
584 sym->op = &ignore_attr_op;
585 }
586 }
587
588
589 // Add a symbol to the list of function-local symbols
590 static void fn_local_symbol(struct symbol *sym)
591 {
592 if (function_symbol_list)
593 add_symbol(function_symbol_list, sym);
594 }
595
596 static int SENTINEL_ATTR match_idents(struct token *token, ...)
597 {
598 va_list args;
599 struct ident * next;
600
601 if (token_type(token) != TOKEN_IDENT)
602 return 0;
603
604 va_start(args, token);
605 do {
606 next = va_arg(args, struct ident *);
607 } while (next && token->ident != next);
608 va_end(args);
609
610 return next && token->ident == next;
611 }
612
613
614 struct statement *alloc_statement(struct position pos, int type)
615 {
616 struct statement *stmt = __alloc_statement(0);
617 stmt->type = type;
618 stmt->pos = pos;
619 return stmt;
620 }
621
622 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list);
623
624 static void apply_modifiers(struct position pos, struct decl_state *ctx)
625 {
626 struct symbol *ctype;
627 if (!ctx->mode)
628 return;
629 ctype = ctx->mode->to_mode(ctx->ctype.base_type);
630 if (!ctype)
631 sparse_error(pos, "don't know how to apply mode to %s",
632 show_typename(ctx->ctype.base_type));
633 else
634 ctx->ctype.base_type = ctype;
635
636 }
637
638 static struct symbol * alloc_indirect_symbol(struct position pos, struct ctype *ctype, int type)
639 {
640 struct symbol *sym = alloc_symbol(pos, type);
641
642 sym->ctype.base_type = ctype->base_type;
643 sym->ctype.modifiers = ctype->modifiers;
644
645 ctype->base_type = sym;
646 ctype->modifiers = 0;
647 return sym;
648 }
649
650 /*
651 * NOTE! NS_LABEL is not just a different namespace,
652 * it also ends up using function scope instead of the
653 * regular symbol scope.
654 */
655 struct symbol *label_symbol(struct token *token)
656 {
657 struct symbol *sym = lookup_symbol(token->ident, NS_LABEL);
658 if (!sym) {
659 sym = alloc_symbol(token->pos, SYM_LABEL);
660 bind_symbol(sym, token->ident, NS_LABEL);
661 fn_local_symbol(sym);
662 }
663 return sym;
664 }
665
666 static struct token *struct_union_enum_specifier(enum type type,
667 struct token *token, struct decl_state *ctx,
668 struct token *(*parse)(struct token *, struct symbol *))
669 {
670 struct symbol *sym;
671 struct position *repos;
672
673 token = handle_attributes(token, ctx, KW_ATTRIBUTE);
674 if (token_type(token) == TOKEN_IDENT) {
675 sym = lookup_symbol(token->ident, NS_STRUCT);
676 if (!sym ||
677 (is_outer_scope(sym->scope) &&
678 (match_op(token->next,';') || match_op(token->next,'{')))) {
679 // Either a new symbol, or else an out-of-scope
680 // symbol being redefined.
681 sym = alloc_symbol(token->pos, type);
682 bind_symbol(sym, token->ident, NS_STRUCT);
683 }
684 if (sym->type != type)
685 error_die(token->pos, "invalid tag applied to %s", show_typename (sym));
686 ctx->ctype.base_type = sym;
687 repos = &token->pos;
688 token = token->next;
689 if (match_op(token, '{')) {
690 // The following test is actually wrong for empty
691 // structs, but (1) they are not C99, (2) gcc does
692 // the same thing, and (3) it's easier.
693 if (sym->symbol_list)
694 error_die(token->pos, "redefinition of %s", show_typename (sym));
695 sym->pos = *repos;
696 token = parse(token->next, sym);
697 token = expect(token, '}', "at end of struct-union-enum-specifier");
698
699 // Mark the structure as needing re-examination
700 sym->examined = 0;
701 sym->endpos = token->pos;
702 }
703 return token;
704 }
705
706 // private struct/union/enum type
707 if (!match_op(token, '{')) {
708 sparse_error(token->pos, "expected declaration");
709 ctx->ctype.base_type = &bad_ctype;
710 return token;
711 }
712
713 sym = alloc_symbol(token->pos, type);
714 token = parse(token->next, sym);
715 ctx->ctype.base_type = sym;
716 token = expect(token, '}', "at end of specifier");
717 sym->endpos = token->pos;
718
719 return token;
720 }
721
722 static struct token *parse_struct_declaration(struct token *token, struct symbol *sym)
723 {
724 struct symbol *field, *last = NULL;
725 struct token *res;
726 res = struct_declaration_list(token, &sym->symbol_list);
727 FOR_EACH_PTR(sym->symbol_list, field) {
728 if (!field->ident) {
729 struct symbol *base = field->ctype.base_type;
730 if (base && base->type == SYM_BITFIELD)
731 continue;
732 }
733 if (last)
734 last->next_subobject = field;
735 last = field;
736 } END_FOR_EACH_PTR(field);
737 return res;
738 }
739
740 static struct token *parse_union_declaration(struct token *token, struct symbol *sym)
741 {
742 return struct_declaration_list(token, &sym->symbol_list);
743 }
744
745 static struct token *struct_specifier(struct token *token, struct decl_state *ctx)
746 {
747 return struct_union_enum_specifier(SYM_STRUCT, token, ctx, parse_struct_declaration);
748 }
749
750 static struct token *union_specifier(struct token *token, struct decl_state *ctx)
751 {
752 return struct_union_enum_specifier(SYM_UNION, token, ctx, parse_union_declaration);
753 }
754
755
756 typedef struct {
757 int x;
758 unsigned long long y;
759 } Num;
760
761 static void upper_boundary(Num *n, Num *v)
762 {
763 if (n->x > v->x)
764 return;
765 if (n->x < v->x) {
766 *n = *v;
767 return;
768 }
769 if (n->y < v->y)
770 n->y = v->y;
771 }
772
773 static void lower_boundary(Num *n, Num *v)
774 {
775 if (n->x < v->x)
776 return;
777 if (n->x > v->x) {
778 *n = *v;
779 return;
780 }
781 if (n->y > v->y)
782 n->y = v->y;
783 }
784
785 static int type_is_ok(struct symbol *type, Num *upper, Num *lower)
786 {
787 int shift = type->bit_size;
788 int is_unsigned = type->ctype.modifiers & MOD_UNSIGNED;
789
790 if (!is_unsigned)
791 shift--;
792 if (upper->x == 0 && upper->y >> shift)
793 return 0;
794 if (lower->x == 0 || (!is_unsigned && (~lower->y >> shift) == 0))
795 return 1;
796 return 0;
797 }
798
799 static struct symbol *bigger_enum_type(struct symbol *s1, struct symbol *s2)
800 {
801 if (s1->bit_size < s2->bit_size) {
802 s1 = s2;
803 } else if (s1->bit_size == s2->bit_size) {
804 if (s2->ctype.modifiers & MOD_UNSIGNED)
805 s1 = s2;
806 }
807 if (s1->bit_size < bits_in_int)
808 return &int_ctype;
809 return s1;
810 }
811
812 static void cast_enum_list(struct symbol_list *list, struct symbol *base_type)
813 {
814 struct symbol *sym;
815
816 FOR_EACH_PTR(list, sym) {
817 struct expression *expr = sym->initializer;
818 struct symbol *ctype;
819 if (expr->type != EXPR_VALUE)
820 continue;
821 ctype = expr->ctype;
822 if (ctype->bit_size == base_type->bit_size)
823 continue;
824 cast_value(expr, base_type, expr, ctype);
825 } END_FOR_EACH_PTR(sym);
826 }
827
828 static struct token *parse_enum_declaration(struct token *token, struct symbol *parent)
829 {
830 unsigned long long lastval = 0;
831 struct symbol *ctype = NULL, *base_type = NULL;
832 Num upper = {-1, 0}, lower = {1, 0};
833
834 parent->examined = 1;
835 parent->ctype.base_type = &int_ctype;
836 while (token_type(token) == TOKEN_IDENT) {
837 struct expression *expr = NULL;
838 struct token *next = token->next;
839 struct symbol *sym;
840
841 if (match_op(next, '=')) {
842 next = constant_expression(next->next, &expr);
843 lastval = get_expression_value(expr);
844 ctype = &void_ctype;
845 if (expr && expr->ctype)
846 ctype = expr->ctype;
847 } else if (!ctype) {
848 ctype = &int_ctype;
849 } else if (is_int_type(ctype)) {
850 lastval++;
851 } else {
852 error_die(token->pos, "can't increment the last enum member");
853 }
854
855 if (!expr) {
856 expr = alloc_expression(token->pos, EXPR_VALUE);
857 expr->value = lastval;
858 expr->ctype = ctype;
859 }
860
861 sym = alloc_symbol(token->pos, SYM_NODE);
862 bind_symbol(sym, token->ident, NS_SYMBOL);
863 sym->ctype.modifiers &= ~MOD_ADDRESSABLE;
864 sym->initializer = expr;
865 sym->enum_member = 1;
866 sym->ctype.base_type = parent;
867 add_ptr_list(&parent->symbol_list, sym);
868
869 if (base_type != &bad_ctype) {
870 if (ctype->type == SYM_NODE)
871 ctype = ctype->ctype.base_type;
872 if (ctype->type == SYM_ENUM) {
873 if (ctype == parent)
874 ctype = base_type;
875 else
876 ctype = ctype->ctype.base_type;
877 }
878 /*
879 * base_type rules:
880 * - if all enums are of the same type, then
881 * the base_type is that type (two first
882 * cases)
883 * - if enums are of different types, they
884 * all have to be integer types, and the
885 * base type is at least "int_ctype".
886 * - otherwise the base_type is "bad_ctype".
887 */
888 if (!base_type) {
889 base_type = ctype;
890 } else if (ctype == base_type) {
891 /* nothing */
892 } else if (is_int_type(base_type) && is_int_type(ctype)) {
893 base_type = bigger_enum_type(base_type, ctype);
894 } else
895 base_type = &bad_ctype;
896 parent->ctype.base_type = base_type;
897 }
898 if (is_int_type(base_type)) {
899 Num v = {.y = lastval};
900 if (ctype->ctype.modifiers & MOD_UNSIGNED)
901 v.x = 0;
902 else if ((long long)lastval >= 0)
903 v.x = 0;
904 else
905 v.x = -1;
906 upper_boundary(&upper, &v);
907 lower_boundary(&lower, &v);
908 }
909 token = next;
910
911 sym->endpos = token->pos;
912
913 if (!match_op(token, ','))
914 break;
915 token = token->next;
916 }
917 if (!base_type) {
918 sparse_error(token->pos, "bad enum definition");
919 base_type = &bad_ctype;
920 }
921 else if (!is_int_type(base_type))
922 base_type = base_type;
923 else if (type_is_ok(base_type, &upper, &lower))
924 base_type = base_type;
925 else if (type_is_ok(&int_ctype, &upper, &lower))
926 base_type = &int_ctype;
927 else if (type_is_ok(&uint_ctype, &upper, &lower))
928 base_type = &uint_ctype;
929 else if (type_is_ok(&long_ctype, &upper, &lower))
930 base_type = &long_ctype;
931 else if (type_is_ok(&ulong_ctype, &upper, &lower))
932 base_type = &ulong_ctype;
933 else if (type_is_ok(&llong_ctype, &upper, &lower))
934 base_type = &llong_ctype;
935 else if (type_is_ok(&ullong_ctype, &upper, &lower))
936 base_type = &ullong_ctype;
937 else
938 base_type = &bad_ctype;
939 parent->ctype.base_type = base_type;
940 parent->ctype.modifiers |= (base_type->ctype.modifiers & MOD_UNSIGNED);
941 parent->examined = 0;
942
943 cast_enum_list(parent->symbol_list, base_type);
944
945 return token;
946 }
947
948 static struct token *enum_specifier(struct token *token, struct decl_state *ctx)
949 {
950 struct token *ret = struct_union_enum_specifier(SYM_ENUM, token, ctx, parse_enum_declaration);
951 struct ctype *ctype = &ctx->ctype.base_type->ctype;
952
953 if (!ctype->base_type)
954 ctype->base_type = &incomplete_ctype;
955
956 return ret;
957 }
958
959 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype);
960
961 static struct token *typeof_specifier(struct token *token, struct decl_state *ctx)
962 {
963 struct symbol *sym;
964
965 if (!match_op(token, '(')) {
966 sparse_error(token->pos, "expected '(' after typeof");
967 return token;
968 }
969 if (lookup_type(token->next)) {
970 token = typename(token->next, &sym, NULL);
971 ctx->ctype.base_type = sym->ctype.base_type;
972 apply_ctype(token->pos, &sym->ctype, &ctx->ctype);
973 } else {
974 struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
975 token = parse_expression(token->next, &typeof_sym->initializer);
976
977 typeof_sym->endpos = token->pos;
978 if (!typeof_sym->initializer) {
979 sparse_error(token->pos, "expected expression after the '(' token");
980 typeof_sym = &bad_ctype;
981 }
982 ctx->ctype.base_type = typeof_sym;
983 }
984 return expect(token, ')', "after typeof");
985 }
986
987 static struct token *ignore_attribute(struct token *token, struct symbol *attr, struct decl_state *ctx)
988 {
989 struct expression *expr = NULL;
990 if (match_op(token, '('))
991 token = parens_expression(token, &expr, "in attribute");
992 return token;
993 }
994
995 static struct token *attribute_packed(struct token *token, struct symbol *attr, struct decl_state *ctx)
996 {
997 if (!ctx->ctype.alignment)
998 ctx->ctype.alignment = 1;
999 return token;
1000 }
1001
1002 static struct token *attribute_aligned(struct token *token, struct symbol *attr, struct decl_state *ctx)
1003 {
1004 int alignment = max_alignment;
1005 struct expression *expr = NULL;
1006
1007 if (match_op(token, '(')) {
1008 token = parens_expression(token, &expr, "in attribute");
1009 if (expr)
1010 alignment = const_expression_value(expr);
1011 }
1012 if (alignment & (alignment-1)) {
1013 warning(token->pos, "I don't like non-power-of-2 alignments");
1014 return token;
1015 } else if (alignment > ctx->ctype.alignment)
1016 ctx->ctype.alignment = alignment;
1017 return token;
1018 }
1019
1020 static void apply_qualifier(struct position *pos, struct ctype *ctx, unsigned long qual)
1021 {
1022 if (ctx->modifiers & qual)
1023 warning(*pos, "duplicate %s", modifier_string(qual));
1024 ctx->modifiers |= qual;
1025 }
1026
1027 static struct token *attribute_modifier(struct token *token, struct symbol *attr, struct decl_state *ctx)
1028 {
1029 apply_qualifier(&token->pos, &ctx->ctype, attr->ctype.modifiers);
1030 return token;
1031 }
1032
1033 static struct token *attribute_address_space(struct token *token, struct symbol *attr, struct decl_state *ctx)
1034 {
1035 struct expression *expr = NULL;
1036 int as;
1037 token = expect(token, '(', "after address_space attribute");
1038 token = conditional_expression(token, &expr);
1039 if (expr) {
1040 as = const_expression_value(expr);
1041 if (Waddress_space && as)
1042 ctx->ctype.as = as;
1043 }
1044 token = expect(token, ')', "after address_space attribute");
1045 return token;
1046 }
1047
1048 static struct symbol *to_QI_mode(struct symbol *ctype)
1049 {
1050 if (ctype->ctype.base_type != &int_type)
1051 return NULL;
1052 if (ctype == &char_ctype)
1053 return ctype;
1054 return ctype->ctype.modifiers & MOD_UNSIGNED ? &uchar_ctype
1055 : &schar_ctype;
1056 }
1057
1058 static struct symbol *to_HI_mode(struct symbol *ctype)
1059 {
1060 if (ctype->ctype.base_type != &int_type)
1061 return NULL;
1062 return ctype->ctype.modifiers & MOD_UNSIGNED ? &ushort_ctype
1063 : &sshort_ctype;
1064 }
1065
1066 static struct symbol *to_SI_mode(struct symbol *ctype)
1067 {
1068 if (ctype->ctype.base_type != &int_type)
1069 return NULL;
1070 return ctype->ctype.modifiers & MOD_UNSIGNED ? &uint_ctype
1071 : &sint_ctype;
1072 }
1073
1074 static struct symbol *to_DI_mode(struct symbol *ctype)
1075 {
1076 if (ctype->ctype.base_type != &int_type)
1077 return NULL;
1078 return ctype->ctype.modifiers & MOD_UNSIGNED ? &ullong_ctype
1079 : &sllong_ctype;
1080 }
1081
1082 static struct symbol *to_TI_mode(struct symbol *ctype)
1083 {
1084 if (ctype->ctype.base_type != &int_type)
1085 return NULL;
1086 return ctype->ctype.modifiers & MOD_UNSIGNED ? &ulllong_ctype
1087 : &slllong_ctype;
1088 }
1089
1090 static struct symbol *to_word_mode(struct symbol *ctype)
1091 {
1092 if (ctype->ctype.base_type != &int_type)
1093 return NULL;
1094 return ctype->ctype.modifiers & MOD_UNSIGNED ? &ulong_ctype
1095 : &slong_ctype;
1096 }
1097
1098 static struct token *attribute_mode(struct token *token, struct symbol *attr, struct decl_state *ctx)
1099 {
1100 token = expect(token, '(', "after mode attribute");
1101 if (token_type(token) == TOKEN_IDENT) {
1102 struct symbol *mode = lookup_keyword(token->ident, NS_KEYWORD);
1103 if (mode && mode->op->type == KW_MODE)
1104 ctx->mode = mode->op;
1105 else
1106 sparse_error(token->pos, "unknown mode attribute %s\n", show_ident(token->ident));
1107 token = token->next;
1108 } else
1109 sparse_error(token->pos, "expect attribute mode symbol\n");
1110 token = expect(token, ')', "after mode attribute");
1111 return token;
1112 }
1113
1114 static struct token *attribute_context(struct token *token, struct symbol *attr, struct decl_state *ctx)
1115 {
1116 struct context *context = alloc_context();
1117 struct expression *args[3];
1118 int argc = 0;
1119
1120 token = expect(token, '(', "after context attribute");
1121 while (!match_op(token, ')')) {
1122 struct expression *expr = NULL;
1123 token = conditional_expression(token, &expr);
1124 if (!expr)
1125 break;
1126 if (argc < 3)
1127 args[argc++] = expr;
1128 if (!match_op(token, ','))
1129 break;
1130 token = token->next;
1131 }
1132
1133 switch(argc) {
1134 case 0:
1135 sparse_error(token->pos, "expected context input/output values");
1136 break;
1137 case 1:
1138 context->in = get_expression_value(args[0]);
1139 break;
1140 case 2:
1141 context->in = get_expression_value(args[0]);
1142 context->out = get_expression_value(args[1]);
1143 break;
1144 case 3:
1145 context->context = args[0];
1146 context->in = get_expression_value(args[1]);
1147 context->out = get_expression_value(args[2]);
1148 break;
1149 }
1150
1151 if (argc)
1152 add_ptr_list(&ctx->ctype.contexts, context);
1153
1154 token = expect(token, ')', "after context attribute");
1155 return token;
1156 }
1157
1158 static struct token *attribute_designated_init(struct token *token, struct symbol *attr, struct decl_state *ctx)
1159 {
1160 if (ctx->ctype.base_type && ctx->ctype.base_type->type == SYM_STRUCT)
1161 ctx->ctype.base_type->designated_init = 1;
1162 else
1163 warning(token->pos, "attribute designated_init applied to non-structure type");
1164 return token;
1165 }
1166
1167 static struct token *attribute_transparent_union(struct token *token, struct symbol *attr, struct decl_state *ctx)
1168 {
1169 if (Wtransparent_union)
1170 warning(token->pos, "ignoring attribute __transparent_union__");
1171 return token;
1172 }
1173
1174 static struct token *recover_unknown_attribute(struct token *token)
1175 {
1176 struct expression *expr = NULL;
1177
1178 sparse_error(token->pos, "attribute '%s': unknown attribute", show_ident(token->ident));
1179 token = token->next;
1180 if (match_op(token, '('))
1181 token = parens_expression(token, &expr, "in attribute");
1182 return token;
1183 }
1184
1185 static struct token *attribute_specifier(struct token *token, struct decl_state *ctx)
1186 {
1187 token = expect(token, '(', "after attribute");
1188 token = expect(token, '(', "after attribute");
1189
1190 for (;;) {
1191 struct ident *attribute_name;
1192 struct symbol *attr;
1193
1194 if (eof_token(token))
1195 break;
1196 if (match_op(token, ';'))
1197 break;
1198 if (token_type(token) != TOKEN_IDENT)
1199 break;
1200 attribute_name = token->ident;
1201 attr = lookup_keyword(attribute_name, NS_KEYWORD);
1202 if (attr && attr->op->attribute)
1203 token = attr->op->attribute(token->next, attr, ctx);
1204 else
1205 token = recover_unknown_attribute(token);
1206
1207 if (!match_op(token, ','))
1208 break;
1209 token = token->next;
1210 }
1211
1212 token = expect(token, ')', "after attribute");
1213 token = expect(token, ')', "after attribute");
1214 return token;
1215 }
1216
1217 static const char *storage_class[] =
1218 {
1219 [STypedef] = "typedef",
1220 [SAuto] = "auto",
1221 [SExtern] = "extern",
1222 [SStatic] = "static",
1223 [SRegister] = "register",
1224 [SForced] = "[force]"
1225 };
1226
1227 static unsigned long storage_modifiers(struct decl_state *ctx)
1228 {
1229 static unsigned long mod[] =
1230 {
1231 [SAuto] = MOD_AUTO,
1232 [SExtern] = MOD_EXTERN,
1233 [SStatic] = MOD_STATIC,
1234 [SRegister] = MOD_REGISTER
1235 };
1236 return mod[ctx->storage_class] | (ctx->is_inline ? MOD_INLINE : 0)
1237 | (ctx->is_tls ? MOD_TLS : 0);
1238 }
1239
1240 static void set_storage_class(struct position *pos, struct decl_state *ctx, int class)
1241 {
1242 /* __thread can be used alone, or with extern or static */
1243 if (ctx->is_tls && (class != SStatic && class != SExtern)) {
1244 sparse_error(*pos, "__thread can only be used alone, or with "
1245 "extern or static");
1246 return;
1247 }
1248
1249 if (!ctx->storage_class) {
1250 ctx->storage_class = class;
1251 return;
1252 }
1253 if (ctx->storage_class == class)
1254 sparse_error(*pos, "duplicate %s", storage_class[class]);
1255 else
1256 sparse_error(*pos, "multiple storage classes");
1257 }
1258
1259 static struct token *typedef_specifier(struct token *next, struct decl_state *ctx)
1260 {
1261 set_storage_class(&next->pos, ctx, STypedef);
1262 return next;
1263 }
1264
1265 static struct token *auto_specifier(struct token *next, struct decl_state *ctx)
1266 {
1267 set_storage_class(&next->pos, ctx, SAuto);
1268 return next;
1269 }
1270
1271 static struct token *register_specifier(struct token *next, struct decl_state *ctx)
1272 {
1273 set_storage_class(&next->pos, ctx, SRegister);
1274 return next;
1275 }
1276
1277 static struct token *static_specifier(struct token *next, struct decl_state *ctx)
1278 {
1279 set_storage_class(&next->pos, ctx, SStatic);
1280 return next;
1281 }
1282
1283 static struct token *extern_specifier(struct token *next, struct decl_state *ctx)
1284 {
1285 set_storage_class(&next->pos, ctx, SExtern);
1286 return next;
1287 }
1288
1289 static struct token *thread_specifier(struct token *next, struct decl_state *ctx)
1290 {
1291 /* This GCC extension can be used alone, or with extern or static */
1292 if (!ctx->storage_class || ctx->storage_class == SStatic
1293 || ctx->storage_class == SExtern) {
1294 ctx->is_tls = 1;
1295 } else {
1296 sparse_error(next->pos, "__thread can only be used alone, or "
1297 "with extern or static");
1298 }
1299
1300 return next;
1301 }
1302
1303 static struct token *attribute_force(struct token *token, struct symbol *attr, struct decl_state *ctx)
1304 {
1305 set_storage_class(&token->pos, ctx, SForced);
1306 return token;
1307 }
1308
1309 static struct token *inline_specifier(struct token *next, struct decl_state *ctx)
1310 {
1311 ctx->is_inline = 1;
1312 return next;
1313 }
1314
1315 static struct token *const_qualifier(struct token *next, struct decl_state *ctx)
1316 {
1317 apply_qualifier(&next->pos, &ctx->ctype, MOD_CONST);
1318 return next;
1319 }
1320
1321 static struct token *volatile_qualifier(struct token *next, struct decl_state *ctx)
1322 {
1323 apply_qualifier(&next->pos, &ctx->ctype, MOD_VOLATILE);
1324 return next;
1325 }
1326
1327 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype)
1328 {
1329 unsigned long mod = thistype->modifiers;
1330
1331 if (mod)
1332 apply_qualifier(&pos, ctype, mod);
1333
1334 /* Context */
1335 concat_ptr_list((struct ptr_list *)thistype->contexts,
1336 (struct ptr_list **)&ctype->contexts);
1337
1338 /* Alignment */
1339 if (thistype->alignment > ctype->alignment)
1340 ctype->alignment = thistype->alignment;
1341
1342 /* Address space */
1343 if (thistype->as)
1344 ctype->as = thistype->as;
1345 }
1346
1347 static void specifier_conflict(struct position pos, int what, struct ident *new)
1348 {
1349 const char *old;
1350 if (what & (Set_S | Set_T))
1351 goto Catch_all;
1352 if (what & Set_Char)
1353 old = "char";
1354 else if (what & Set_Double)
1355 old = "double";
1356 else if (what & Set_Float)
1357 old = "float";
1358 else if (what & Set_Signed)
1359 old = "signed";
1360 else if (what & Set_Unsigned)
1361 old = "unsigned";
1362 else if (what & Set_Short)
1363 old = "short";
1364 else if (what & Set_Long)
1365 old = "long";
1366 else
1367 old = "long long";
1368 sparse_error(pos, "impossible combination of type specifiers: %s %s",
1369 old, show_ident(new));
1370 return;
1371
1372 Catch_all:
1373 sparse_error(pos, "two or more data types in declaration specifiers");
1374 }
1375
1376 static struct symbol * const int_types[] =
1377 {&short_ctype, &int_ctype, &long_ctype, &llong_ctype};
1378 static struct symbol * const signed_types[] =
1379 {&sshort_ctype, &sint_ctype, &slong_ctype, &sllong_ctype,
1380 &slllong_ctype};
1381 static struct symbol * const unsigned_types[] =
1382 {&ushort_ctype, &uint_ctype, &ulong_ctype, &ullong_ctype,
1383 &ulllong_ctype};
1384 static struct symbol * const real_types[] =
1385 {&float_ctype, &double_ctype, &ldouble_ctype};
1386 static struct symbol * const char_types[] =
1387 {&char_ctype, &schar_ctype, &uchar_ctype};
1388 static struct symbol * const * const types[] = {
1389 int_types + 1, signed_types + 1, unsigned_types + 1,
1390 real_types + 1, char_types, char_types + 1, char_types + 2
1391 };
1392
1393 struct symbol *ctype_integer(int size, int want_unsigned)
1394 {
1395 return types[want_unsigned ? CUInt : CInt][size];
1396 }
1397
1398 static struct token *handle_qualifiers(struct token *t, struct decl_state *ctx)
1399 {
1400 while (token_type(t) == TOKEN_IDENT) {
1401 struct symbol *s = lookup_symbol(t->ident, NS_TYPEDEF);
1402 if (!s)
1403 break;
1404 if (s->type != SYM_KEYWORD)
1405 break;
1406 if (!(s->op->type & (KW_ATTRIBUTE | KW_QUALIFIER)))
1407 break;
1408 t = t->next;
1409 if (s->op->declarator)
1410 t = s->op->declarator(t, ctx);
1411 }
1412 return t;
1413 }
1414
1415 static struct token *declaration_specifiers(struct token *token, struct decl_state *ctx)
1416 {
1417 int seen = 0;
1418 int class = CInt;
1419 int size = 0;
1420
1421 while (token_type(token) == TOKEN_IDENT) {
1422 struct symbol *s = lookup_symbol(token->ident,
1423 NS_TYPEDEF | NS_SYMBOL);
1424 if (!s || !(s->namespace & NS_TYPEDEF))
1425 break;
1426 if (s->type != SYM_KEYWORD) {
1427 if (seen & Set_Any)
1428 break;
1429 seen |= Set_S | Set_T;
1430 ctx->ctype.base_type = s->ctype.base_type;
1431 apply_ctype(token->pos, &s->ctype, &ctx->ctype);
1432 token = token->next;
1433 continue;
1434 }
1435 if (s->op->type & KW_SPECIFIER) {
1436 if (seen & s->op->test) {
1437 specifier_conflict(token->pos,
1438 seen & s->op->test,
1439 token->ident);
1440 break;
1441 }
1442 seen |= s->op->set;
1443 class += s->op->class;
1444 if (s->op->type & KW_SHORT) {
1445 size = -1;
1446 } else if (s->op->type & KW_LONG && size++) {
1447 if (class == CReal) {
1448 specifier_conflict(token->pos,
1449 Set_Vlong,
1450 &double_ident);
1451 break;
1452 }
1453 seen |= Set_Vlong;
1454 }
1455 }
1456 token = token->next;
1457 if (s->op->declarator)
1458 token = s->op->declarator(token, ctx);
1459 if (s->op->type & KW_EXACT) {
1460 ctx->ctype.base_type = s->ctype.base_type;
1461 ctx->ctype.modifiers |= s->ctype.modifiers;
1462 }
1463 }
1464
1465 if (!(seen & Set_S)) { /* not set explicitly? */
1466 struct symbol *base = &incomplete_ctype;
1467 if (seen & Set_Any)
1468 base = types[class][size];
1469 ctx->ctype.base_type = base;
1470 }
1471
1472 if (ctx->ctype.modifiers & MOD_BITWISE) {
1473 struct symbol *type;
1474 ctx->ctype.modifiers &= ~MOD_BITWISE;
1475 if (!is_int_type(ctx->ctype.base_type)) {
1476 sparse_error(token->pos, "invalid modifier");
1477 return token;
1478 }
1479 type = alloc_symbol(token->pos, SYM_BASETYPE);
1480 *type = *ctx->ctype.base_type;
1481 type->ctype.modifiers &= ~MOD_SPECIFIER;
1482 type->ctype.base_type = ctx->ctype.base_type;
1483 type->type = SYM_RESTRICT;
1484 ctx->ctype.base_type = type;
1485 create_fouled(type);
1486 }
1487 return token;
1488 }
1489
1490 static struct token *abstract_array_declarator(struct token *token, struct symbol *sym)
1491 {
1492 struct expression *expr = NULL;
1493
1494 if (match_idents(token, &restrict_ident, &__restrict_ident, NULL))
1495 token = token->next;
1496 token = parse_expression(token, &expr);
1497 sym->array_size = expr;
1498 return token;
1499 }
1500
1501 static struct token *parameter_type_list(struct token *, struct symbol *);
1502 static struct token *identifier_list(struct token *, struct symbol *);
1503 static struct token *declarator(struct token *token, struct decl_state *ctx);
1504
1505 static struct token *skip_attribute(struct token *token)
1506 {
1507 token = token->next;
1508 if (match_op(token, '(')) {
1509 int depth = 1;
1510 token = token->next;
1511 while (depth && !eof_token(token)) {
1512 if (token_type(token) == TOKEN_SPECIAL) {
1513 if (token->special == '(')
1514 depth++;
1515 else if (token->special == ')')
1516 depth--;
1517 }
1518 token = token->next;
1519 }
1520 }
1521 return token;
1522 }
1523
1524 static struct token *skip_attributes(struct token *token)
1525 {
1526 struct symbol *keyword;
1527 for (;;) {
1528 if (token_type(token) != TOKEN_IDENT)
1529 break;
1530 keyword = lookup_keyword(token->ident, NS_KEYWORD | NS_TYPEDEF);
1531 if (!keyword || keyword->type != SYM_KEYWORD)
1532 break;
1533 if (!(keyword->op->type & KW_ATTRIBUTE))
1534 break;
1535 token = expect(token->next, '(', "after attribute");
1536 token = expect(token, '(', "after attribute");
1537 for (;;) {
1538 if (eof_token(token))
1539 break;
1540 if (match_op(token, ';'))
1541 break;
1542 if (token_type(token) != TOKEN_IDENT)
1543 break;
1544 token = skip_attribute(token);
1545 if (!match_op(token, ','))
1546 break;
1547 token = token->next;
1548 }
1549 token = expect(token, ')', "after attribute");
1550 token = expect(token, ')', "after attribute");
1551 }
1552 return token;
1553 }
1554
1555 static struct token *handle_attributes(struct token *token, struct decl_state *ctx, unsigned int keywords)
1556 {
1557 struct symbol *keyword;
1558 for (;;) {
1559 if (token_type(token) != TOKEN_IDENT)
1560 break;
1561 keyword = lookup_keyword(token->ident, NS_KEYWORD | NS_TYPEDEF);
1562 if (!keyword || keyword->type != SYM_KEYWORD)
1563 break;
1564 if (!(keyword->op->type & keywords))
1565 break;
1566 token = keyword->op->declarator(token->next, ctx);
1567 keywords &= KW_ATTRIBUTE;
1568 }
1569 return token;
1570 }
1571
1572 static int is_nested(struct token *token, struct token **p,
1573 int prefer_abstract)
1574 {
1575 /*
1576 * This can be either a parameter list or a grouping.
1577 * For the direct (non-abstract) case, we know if must be
1578 * a parameter list if we already saw the identifier.
1579 * For the abstract case, we know if must be a parameter
1580 * list if it is empty or starts with a type.
1581 */
1582 struct token *next = token->next;
1583
1584 *p = next = skip_attributes(next);
1585
1586 if (token_type(next) == TOKEN_IDENT) {
1587 if (lookup_type(next))
1588 return !prefer_abstract;
1589 return 1;
1590 }
1591
1592 if (match_op(next, ')') || match_op(next, SPECIAL_ELLIPSIS))
1593 return 0;
1594
1595 return 1;
1596 }
1597
1598 enum kind {
1599 Empty, K_R, Proto, Bad_Func,
1600 };
1601
1602 static enum kind which_func(struct token *token,
1603 struct ident **n,
1604 int prefer_abstract)
1605 {
1606 struct token *next = token->next;
1607
1608 if (token_type(next) == TOKEN_IDENT) {
1609 if (lookup_type(next))
1610 return Proto;
1611 /* identifier list not in definition; complain */
1612 if (prefer_abstract)
1613 warning(token->pos,
1614 "identifier list not in definition");
1615 return K_R;
1616 }
1617
1618 if (token_type(next) != TOKEN_SPECIAL)
1619 return Bad_Func;
1620
1621 if (next->special == ')') {
1622 /* don't complain about those */
1623 if (!n || match_op(next->next, ';'))
1624 return Empty;
1625 warning(next->pos,
1626 "non-ANSI function declaration of function '%s'",
1627 show_ident(*n));
1628 return Empty;
1629 }
1630
1631 if (next->special == SPECIAL_ELLIPSIS) {
1632 warning(next->pos,
1633 "variadic functions must have one named argument");
1634 return Proto;
1635 }
1636
1637 return Bad_Func;
1638 }
1639
1640 static struct token *direct_declarator(struct token *token, struct decl_state *ctx)
1641 {
1642 struct ctype *ctype = &ctx->ctype;
1643 struct token *next;
1644 struct ident **p = ctx->ident;
1645
1646 if (ctx->ident && token_type(token) == TOKEN_IDENT) {
1647 *ctx->ident = token->ident;
1648 token = token->next;
1649 } else if (match_op(token, '(') &&
1650 is_nested(token, &next, ctx->prefer_abstract)) {
1651 struct symbol *base_type = ctype->base_type;
1652 if (token->next != next)
1653 next = handle_attributes(token->next, ctx,
1654 KW_ATTRIBUTE);
1655 token = declarator(next, ctx);
1656 token = expect(token, ')', "in nested declarator");
1657 while (ctype->base_type != base_type)
1658 ctype = &ctype->base_type->ctype;
1659 p = NULL;
1660 }
1661
1662 if (match_op(token, '(')) {
1663 enum kind kind = which_func(token, p, ctx->prefer_abstract);
1664 struct symbol *fn;
1665 fn = alloc_indirect_symbol(token->pos, ctype, SYM_FN);
1666 token = token->next;
1667 if (kind == K_R)
1668 token = identifier_list(token, fn);
1669 else if (kind == Proto)
1670 token = parameter_type_list(token, fn);
1671 token = expect(token, ')', "in function declarator");
1672 fn->endpos = token->pos;
1673 return token;
1674 }
1675
1676 while (match_op(token, '[')) {
1677 struct symbol *array;
1678 array = alloc_indirect_symbol(token->pos, ctype, SYM_ARRAY);
1679 token = abstract_array_declarator(token->next, array);
1680 token = expect(token, ']', "in abstract_array_declarator");
1681 array->endpos = token->pos;
1682 ctype = &array->ctype;
1683 }
1684 return token;
1685 }
1686
1687 static struct token *pointer(struct token *token, struct decl_state *ctx)
1688 {
1689 while (match_op(token,'*')) {
1690 struct symbol *ptr = alloc_symbol(token->pos, SYM_PTR);
1691 ptr->ctype.modifiers = ctx->ctype.modifiers;
1692 ptr->ctype.base_type = ctx->ctype.base_type;
1693 ptr->ctype.as = ctx->ctype.as;
1694 ptr->ctype.contexts = ctx->ctype.contexts;
1695 ctx->ctype.modifiers = 0;
1696 ctx->ctype.base_type = ptr;
1697 ctx->ctype.as = 0;
1698 ctx->ctype.contexts = NULL;
1699
1700 token = handle_qualifiers(token->next, ctx);
1701 ctx->ctype.base_type->endpos = token->pos;
1702 }
1703 return token;
1704 }
1705
1706 static struct token *declarator(struct token *token, struct decl_state *ctx)
1707 {
1708 token = pointer(token, ctx);
1709 return direct_declarator(token, ctx);
1710 }
1711
1712 static struct token *handle_bitfield(struct token *token, struct decl_state *ctx)
1713 {
1714 struct ctype *ctype = &ctx->ctype;
1715 struct expression *expr;
1716 struct symbol *bitfield;
1717 long long width;
1718
1719 if (ctype->base_type != &int_type && !is_int_type(ctype->base_type)) {
1720 sparse_error(token->pos, "invalid bitfield specifier for type %s.",
1721 show_typename(ctype->base_type));
1722 // Parse this to recover gracefully.
1723 return conditional_expression(token->next, &expr);
1724 }
1725
1726 bitfield = alloc_indirect_symbol(token->pos, ctype, SYM_BITFIELD);
1727 token = conditional_expression(token->next, &expr);
1728 width = const_expression_value(expr);
1729 bitfield->bit_size = width;
1730
1731 if (width < 0 || width > INT_MAX) {
1732 sparse_error(token->pos, "invalid bitfield width, %lld.", width);
1733 width = -1;
1734 } else if (*ctx->ident && width == 0) {
1735 sparse_error(token->pos, "invalid named zero-width bitfield `%s'",
1736 show_ident(*ctx->ident));
1737 width = -1;
1738 } else if (*ctx->ident) {
1739 struct symbol *base_type = bitfield->ctype.base_type;
1740 struct symbol *bitfield_type = base_type == &int_type ? bitfield : base_type;
1741 int is_signed = !(bitfield_type->ctype.modifiers & MOD_UNSIGNED);
1742 if (Wone_bit_signed_bitfield && width == 1 && is_signed) {
1743 // Valid values are either {-1;0} or {0}, depending on integer
1744 // representation. The latter makes for very efficient code...
1745 sparse_error(token->pos, "dubious one-bit signed bitfield");
1746 }
1747 if (Wdefault_bitfield_sign &&
1748 bitfield_type->type != SYM_ENUM &&
1749 !(bitfield_type->ctype.modifiers & MOD_EXPLICITLY_SIGNED) &&
1750 is_signed) {
1751 // The sign of bitfields is unspecified by default.
1752 warning(token->pos, "dubious bitfield without explicit `signed' or `unsigned'");
1753 }
1754 }
1755 bitfield->bit_size = width;
1756 bitfield->endpos = token->pos;
1757 return token;
1758 }
1759
1760 static struct token *declaration_list(struct token *token, struct symbol_list **list)
1761 {
1762 struct decl_state ctx = {.prefer_abstract = 0};
1763 struct ctype saved;
1764 unsigned long mod;
1765
1766 token = declaration_specifiers(token, &ctx);
1767 mod = storage_modifiers(&ctx);
1768 saved = ctx.ctype;
1769 for (;;) {
1770 struct symbol *decl = alloc_symbol(token->pos, SYM_NODE);
1771 ctx.ident = &decl->ident;
1772
1773 token = declarator(token, &ctx);
1774 if (match_op(token, ':'))
1775 token = handle_bitfield(token, &ctx);
1776
1777 token = handle_attributes(token, &ctx, KW_ATTRIBUTE);
1778 apply_modifiers(token->pos, &ctx);
1779
1780 decl->ctype = ctx.ctype;
1781 decl->ctype.modifiers |= mod;
1782 decl->endpos = token->pos;
1783 add_symbol(list, decl);
1784 if (!match_op(token, ','))
1785 break;
1786 token = token->next;
1787 ctx.ctype = saved;
1788 }
1789 return token;
1790 }
1791
1792 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list)
1793 {
1794 while (!match_op(token, '}')) {
1795 if (!match_op(token, ';'))
1796 token = declaration_list(token, list);
1797 if (!match_op(token, ';')) {
1798 sparse_error(token->pos, "expected ; at end of declaration");
1799 break;
1800 }
1801 token = token->next;
1802 }
1803 return token;
1804 }
1805
1806 static struct token *parameter_declaration(struct token *token, struct symbol *sym)
1807 {
1808 struct decl_state ctx = {.prefer_abstract = 1};
1809
1810 token = declaration_specifiers(token, &ctx);
1811 ctx.ident = &sym->ident;
1812 token = declarator(token, &ctx);
1813 token = handle_attributes(token, &ctx, KW_ATTRIBUTE);
1814 apply_modifiers(token->pos, &ctx);
1815 sym->ctype = ctx.ctype;
1816 sym->ctype.modifiers |= storage_modifiers(&ctx);
1817 sym->endpos = token->pos;
1818 return token;
1819 }
1820
1821 struct token *typename(struct token *token, struct symbol **p, int *forced)
1822 {
1823 struct decl_state ctx = {.prefer_abstract = 1};
1824 int class;
1825 struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
1826 *p = sym;
1827 token = declaration_specifiers(token, &ctx);
1828 token = declarator(token, &ctx);
1829 apply_modifiers(token->pos, &ctx);
1830 sym->ctype = ctx.ctype;
1831 sym->endpos = token->pos;
1832 class = ctx.storage_class;
1833 if (forced) {
1834 *forced = 0;
1835 if (class == SForced) {
1836 *forced = 1;
1837 class = 0;
1838 }
1839 }
1840 if (class)
1841 warning(sym->pos, "storage class in typename (%s %s)",
1842 storage_class[class], show_typename(sym));
1843 return token;
1844 }
1845
1846 static struct token *expression_statement(struct token *token, struct expression **tree)
1847 {
1848 token = parse_expression(token, tree);
1849 return expect(token, ';', "at end of statement");
1850 }
1851
1852 static struct token *parse_asm_operands(struct token *token, struct statement *stmt,
1853 struct expression_list **inout)
1854 {
1855 struct expression *expr;
1856
1857 /* Allow empty operands */
1858 if (match_op(token->next, ':') || match_op(token->next, ')'))
1859 return token->next;
1860 do {
1861 struct ident *ident = NULL;
1862 if (match_op(token->next, '[') &&
1863 token_type(token->next->next) == TOKEN_IDENT &&
1864 match_op(token->next->next->next, ']')) {
1865 ident = token->next->next->ident;
1866 token = token->next->next->next;
1867 }
1868 add_expression(inout, (struct expression *)ident); /* UGGLEE!!! */
1869 token = primary_expression(token->next, &expr);
1870 add_expression(inout, expr);
1871 token = parens_expression(token, &expr, "in asm parameter");
1872 add_expression(inout, expr);
1873 } while (match_op(token, ','));
1874 return token;
1875 }
1876
1877 static struct token *parse_asm_clobbers(struct token *token, struct statement *stmt,
1878 struct expression_list **clobbers)
1879 {
1880 struct expression *expr;
1881
1882 do {
1883 token = primary_expression(token->next, &expr);
1884 add_expression(clobbers, expr);
1885 } while (match_op(token, ','));
1886 return token;
1887 }
1888
1889 static struct token *parse_asm_statement(struct token *token, struct statement *stmt)
1890 {
1891 token = token->next;
1892 stmt->type = STMT_ASM;
1893 if (match_idents(token, &__volatile___ident, &__volatile_ident, &volatile_ident, NULL)) {
1894 token = token->next;
1895 }
1896 token = expect(token, '(', "after asm");
1897 token = parse_expression(token, &stmt->asm_string);
1898 if (match_op(token, ':'))
1899 token = parse_asm_operands(token, stmt, &stmt->asm_outputs);
1900 if (match_op(token, ':'))
1901 token = parse_asm_operands(token, stmt, &stmt->asm_inputs);
1902 if (match_op(token, ':'))
1903 token = parse_asm_clobbers(token, stmt, &stmt->asm_clobbers);
1904 token = expect(token, ')', "after asm");
1905 return expect(token, ';', "at end of asm-statement");
1906 }
1907
1908 static struct token *parse_asm_declarator(struct token *token, struct decl_state *ctx)
1909 {
1910 struct expression *expr;
1911 token = expect(token, '(', "after asm");
1912 token = parse_expression(token->next, &expr);
1913 token = expect(token, ')', "after asm");
1914 return token;
1915 }
1916
1917 /* Make a statement out of an expression */
1918 static struct statement *make_statement(struct expression *expr)
1919 {
1920 struct statement *stmt;
1921
1922 if (!expr)
1923 return NULL;
1924 stmt = alloc_statement(expr->pos, STMT_EXPRESSION);
1925 stmt->expression = expr;
1926 return stmt;
1927 }
1928
1929 /*
1930 * All iterators have two symbols associated with them:
1931 * the "continue" and "break" symbols, which are targets
1932 * for continue and break statements respectively.
1933 *
1934 * They are in a special name-space, but they follow
1935 * all the normal visibility rules, so nested iterators
1936 * automatically work right.
1937 */
1938 static void start_iterator(struct statement *stmt)
1939 {
1940 struct symbol *cont, *brk;
1941
1942 start_symbol_scope();
1943 cont = alloc_symbol(stmt->pos, SYM_NODE);
1944 bind_symbol(cont, &continue_ident, NS_ITERATOR);
1945 brk = alloc_symbol(stmt->pos, SYM_NODE);
1946 bind_symbol(brk, &break_ident, NS_ITERATOR);
1947
1948 stmt->type = STMT_ITERATOR;
1949 stmt->iterator_break = brk;
1950 stmt->iterator_continue = cont;
1951 fn_local_symbol(brk);
1952 fn_local_symbol(cont);
1953 }
1954
1955 static void end_iterator(struct statement *stmt)
1956 {
1957 end_symbol_scope();
1958 }
1959
1960 static struct statement *start_function(struct symbol *sym)
1961 {
1962 struct symbol *ret;
1963 struct statement *stmt = alloc_statement(sym->pos, STMT_COMPOUND);
1964
1965 start_function_scope();
1966 ret = alloc_symbol(sym->pos, SYM_NODE);
1967 ret->ctype = sym->ctype.base_type->ctype;
1968 ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_TLS | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
1969 ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
1970 bind_symbol(ret, &return_ident, NS_ITERATOR);
1971 stmt->ret = ret;
1972 fn_local_symbol(ret);
1973
1974 // Currently parsed symbol for __func__/__FUNCTION__/__PRETTY_FUNCTION__
1975 current_fn = sym;
1976
1977 return stmt;
1978 }
1979
1980 static void end_function(struct symbol *sym)
1981 {
1982 current_fn = NULL;
1983 end_function_scope();
1984 }
1985
1986 /*
1987 * A "switch()" statement, like an iterator, has a
1988 * the "break" symbol associated with it. It works
1989 * exactly like the iterator break - it's the target
1990 * for any break-statements in scope, and means that
1991 * "break" handling doesn't even need to know whether
1992 * it's breaking out of an iterator or a switch.
1993 *
1994 * In addition, the "case" symbol is a marker for the
1995 * case/default statements to find the switch statement
1996 * that they are associated with.
1997 */
1998 static void start_switch(struct statement *stmt)
1999 {
2000 struct symbol *brk, *switch_case;
2001
2002 start_symbol_scope();
2003 brk = alloc_symbol(stmt->pos, SYM_NODE);
2004 bind_symbol(brk, &break_ident, NS_ITERATOR);
2005
2006 switch_case = alloc_symbol(stmt->pos, SYM_NODE);
2007 bind_symbol(switch_case, &case_ident, NS_ITERATOR);
2008 switch_case->stmt = stmt;
2009
2010 stmt->type = STMT_SWITCH;
2011 stmt->switch_break = brk;
2012 stmt->switch_case = switch_case;
2013
2014 fn_local_symbol(brk);
2015 fn_local_symbol(switch_case);
2016 }
2017
2018 static void end_switch(struct statement *stmt)
2019 {
2020 if (!stmt->switch_case->symbol_list)
2021 warning(stmt->pos, "switch with no cases");
2022 end_symbol_scope();
2023 }
2024
2025 static void add_case_statement(struct statement *stmt)
2026 {
2027 struct symbol *target = lookup_symbol(&case_ident, NS_ITERATOR);
2028 struct symbol *sym;
2029
2030 if (!target) {
2031 sparse_error(stmt->pos, "not in switch scope");
2032 stmt->type = STMT_NONE;
2033 return;
2034 }
2035 sym = alloc_symbol(stmt->pos, SYM_NODE);
2036 add_symbol(&target->symbol_list, sym);
2037 sym->stmt = stmt;
2038 stmt->case_label = sym;
2039 fn_local_symbol(sym);
2040 }
2041
2042 static struct token *parse_return_statement(struct token *token, struct statement *stmt)
2043 {
2044 struct symbol *target = lookup_symbol(&return_ident, NS_ITERATOR);
2045
2046 if (!target)
2047 error_die(token->pos, "internal error: return without a function target");
2048 stmt->type = STMT_RETURN;
2049 stmt->ret_target = target;
2050 return expression_statement(token->next, &stmt->ret_value);
2051 }
2052
2053 static struct token *parse_for_statement(struct token *token, struct statement *stmt)
2054 {
2055 struct symbol_list *syms;
2056 struct expression *e1, *e2, *e3;
2057 struct statement *iterator;
2058
2059 start_iterator(stmt);
2060 token = expect(token->next, '(', "after 'for'");
2061
2062 syms = NULL;
2063 e1 = NULL;
2064 /* C99 variable declaration? */
2065 if (lookup_type(token)) {
2066 token = external_declaration(token, &syms);
2067 } else {
2068 token = parse_expression(token, &e1);
2069 token = expect(token, ';', "in 'for'");
2070 }
2071 token = parse_expression(token, &e2);
2072 token = expect(token, ';', "in 'for'");
2073 token = parse_expression(token, &e3);
2074 token = expect(token, ')', "in 'for'");
2075 token = statement(token, &iterator);
2076
2077 stmt->iterator_syms = syms;
2078 stmt->iterator_pre_statement = make_statement(e1);
2079 stmt->iterator_pre_condition = e2;
2080 stmt->iterator_post_statement = make_statement(e3);
2081 stmt->iterator_post_condition = NULL;
2082 stmt->iterator_statement = iterator;
2083 end_iterator(stmt);
2084
2085 return token;
2086 }
2087
2088 static struct token *parse_while_statement(struct token *token, struct statement *stmt)
2089 {
2090 struct expression *expr;
2091 struct statement *iterator;
2092
2093 start_iterator(stmt);
2094 token = parens_expression(token->next, &expr, "after 'while'");
2095 token = statement(token, &iterator);
2096
2097 stmt->iterator_pre_condition = expr;
2098 stmt->iterator_post_condition = NULL;
2099 stmt->iterator_statement = iterator;
2100 end_iterator(stmt);
2101
2102 return token;
2103 }
2104
2105 static struct token *parse_do_statement(struct token *token, struct statement *stmt)
2106 {
2107 struct expression *expr;
2108 struct statement *iterator;
2109
2110 start_iterator(stmt);
2111 token = statement(token->next, &iterator);
2112 if (token_type(token) == TOKEN_IDENT && token->ident == &while_ident)
2113 token = token->next;
2114 else
2115 sparse_error(token->pos, "expected 'while' after 'do'");
2116 token = parens_expression(token, &expr, "after 'do-while'");
2117
2118 stmt->iterator_post_condition = expr;
2119 stmt->iterator_statement = iterator;
2120 end_iterator(stmt);
2121
2122 if (iterator && iterator->type != STMT_COMPOUND && Wdo_while)
2123 warning(iterator->pos, "do-while statement is not a compound statement");
2124
2125 return expect(token, ';', "after statement");
2126 }
2127
2128 static struct token *parse_if_statement(struct token *token, struct statement *stmt)
2129 {
2130 stmt->type = STMT_IF;
2131 token = parens_expression(token->next, &stmt->if_conditional, "after if");
2132 token = statement(token, &stmt->if_true);
2133 if (token_type(token) != TOKEN_IDENT)
2134 return token;
2135 if (token->ident != &else_ident)
2136 return token;
2137 return statement(token->next, &stmt->if_false);
2138 }
2139
2140 static inline struct token *case_statement(struct token *token, struct statement *stmt)
2141 {
2142 stmt->type = STMT_CASE;
2143 token = expect(token, ':', "after default/case");
2144 add_case_statement(stmt);
2145 return statement(token, &stmt->case_statement);
2146 }
2147
2148 static struct token *parse_case_statement(struct token *token, struct statement *stmt)
2149 {
2150 token = parse_expression(token->next, &stmt->case_expression);
2151 if (match_op(token, SPECIAL_ELLIPSIS))
2152 token = parse_expression(token->next, &stmt->case_to);
2153 return case_statement(token, stmt);
2154 }
2155
2156 static struct token *parse_default_statement(struct token *token, struct statement *stmt)
2157 {
2158 return case_statement(token->next, stmt);
2159 }
2160
2161 static struct token *parse_loop_iterator(struct token *token, struct statement *stmt)
2162 {
2163 struct symbol *target = lookup_symbol(token->ident, NS_ITERATOR);
2164 stmt->type = STMT_GOTO;
2165 stmt->goto_label = target;
2166 if (!target)
2167 sparse_error(stmt->pos, "break/continue not in iterator scope");
2168 return expect(token->next, ';', "at end of statement");
2169 }
2170
2171 static struct token *parse_switch_statement(struct token *token, struct statement *stmt)
2172 {
2173 stmt->type = STMT_SWITCH;
2174 start_switch(stmt);
2175 token = parens_expression(token->next, &stmt->switch_expression, "after 'switch'");
2176 token = statement(token, &stmt->switch_statement);
2177 end_switch(stmt);
2178 return token;
2179 }
2180
2181 static struct token *parse_goto_statement(struct token *token, struct statement *stmt)
2182 {
2183 stmt->type = STMT_GOTO;
2184 token = token->next;
2185 if (match_op(token, '*')) {
2186 token = parse_expression(token->next, &stmt->goto_expression);
2187 add_statement(&function_computed_goto_list, stmt);
2188 } else if (token_type(token) == TOKEN_IDENT) {
2189 stmt->goto_label = label_symbol(token);
2190 token = token->next;
2191 } else {
2192 sparse_error(token->pos, "Expected identifier or goto expression");
2193 }
2194 return expect(token, ';', "at end of statement");
2195 }
2196
2197 static struct token *parse_context_statement(struct token *token, struct statement *stmt)
2198 {
2199 stmt->type = STMT_CONTEXT;
2200 token = parse_expression(token->next, &stmt->expression);
2201 if(stmt->expression->type == EXPR_PREOP
2202 && stmt->expression->op == '('
2203 && stmt->expression->unop->type == EXPR_COMMA) {
2204 struct expression *expr;
2205 expr = stmt->expression->unop;
2206 stmt->context = expr->left;
2207 stmt->expression = expr->right;
2208 }
2209 return expect(token, ';', "at end of statement");
2210 }
2211
2212 static struct token *parse_range_statement(struct token *token, struct statement *stmt)
2213 {
2214 stmt->type = STMT_RANGE;
2215 token = assignment_expression(token->next, &stmt->range_expression);
2216 token = expect(token, ',', "after range expression");
2217 token = assignment_expression(token, &stmt->range_low);
2218 token = expect(token, ',', "after low range");
2219 token = assignment_expression(token, &stmt->range_high);
2220 return expect(token, ';', "after range statement");
2221 }
2222
2223 static struct token *statement(struct token *token, struct statement **tree)
2224 {
2225 struct statement *stmt = alloc_statement(token->pos, STMT_NONE);
2226
2227 *tree = stmt;
2228 if (token_type(token) == TOKEN_IDENT) {
2229 struct symbol *s = lookup_keyword(token->ident, NS_KEYWORD);
2230 if (s && s->op->statement)
2231 return s->op->statement(token, stmt);
2232
2233 if (match_op(token->next, ':')) {
2234 stmt->type = STMT_LABEL;
2235 stmt->label_identifier = label_symbol(token);
2236 token = skip_attributes(token->next->next);
2237 return statement(token, &stmt->label_statement);
2238 }
2239 }
2240
2241 if (match_op(token, '{')) {
2242 stmt->type = STMT_COMPOUND;
2243 start_symbol_scope();
2244 token = compound_statement(token->next, stmt);
2245 end_symbol_scope();
2246
2247 return expect(token, '}', "at end of compound statement");
2248 }
2249
2250 stmt->type = STMT_EXPRESSION;
2251 return expression_statement(token, &stmt->expression);
2252 }
2253
2254 /* gcc extension - __label__ ident-list; in the beginning of compound stmt */
2255 static struct token *label_statement(struct token *token)
2256 {
2257 while (token_type(token) == TOKEN_IDENT) {
2258 struct symbol *sym = alloc_symbol(token->pos, SYM_LABEL);
2259 /* it's block-scope, but we want label namespace */
2260 bind_symbol(sym, token->ident, NS_SYMBOL);
2261 sym->namespace = NS_LABEL;
2262 fn_local_symbol(sym);
2263 token = token->next;
2264 if (!match_op(token, ','))
2265 break;
2266 token = token->next;
2267 }
2268 return expect(token, ';', "at end of label declaration");
2269 }
2270
2271 static struct token * statement_list(struct token *token, struct statement_list **list)
2272 {
2273 int seen_statement = 0;
2274 while (token_type(token) == TOKEN_IDENT &&
2275 token->ident == &__label___ident)
2276 token = label_statement(token->next);
2277 for (;;) {
2278 struct statement * stmt;
2279 if (eof_token(token))
2280 break;
2281 if (match_op(token, '}'))
2282 break;
2283 if (lookup_type(token)) {
2284 if (seen_statement) {
2285 warning(token->pos, "mixing declarations and code");
2286 seen_statement = 0;
2287 }
2288 stmt = alloc_statement(token->pos, STMT_DECLARATION);
2289 token = external_declaration(token, &stmt->declaration);
2290 } else {
2291 seen_statement = Wdeclarationafterstatement;
2292 token = statement(token, &stmt);
2293 }
2294 add_statement(list, stmt);
2295 }
2296 return token;
2297 }
2298
2299 static struct token *identifier_list(struct token *token, struct symbol *fn)
2300 {
2301 struct symbol_list **list = &fn->arguments;
2302 for (;;) {
2303 struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
2304 sym->ident = token->ident;
2305 token = token->next;
2306 sym->endpos = token->pos;
2307 sym->ctype.base_type = &incomplete_ctype;
2308 add_symbol(list, sym);
2309 if (!match_op(token, ',') ||
2310 token_type(token->next) != TOKEN_IDENT ||
2311 lookup_type(token->next))
2312 break;
2313 token = token->next;
2314 }
2315 return token;
2316 }
2317
2318 static struct token *parameter_type_list(struct token *token, struct symbol *fn)
2319 {
2320 struct symbol_list **list = &fn->arguments;
2321
2322 for (;;) {
2323 struct symbol *sym;
2324
2325 if (match_op(token, SPECIAL_ELLIPSIS)) {
2326 fn->variadic = 1;
2327 token = token->next;
2328 break;
2329 }
2330
2331 sym = alloc_symbol(token->pos, SYM_NODE);
2332 token = parameter_declaration(token, sym);
2333 if (sym->ctype.base_type == &void_ctype) {
2334 /* Special case: (void) */
2335 if (!*list && !sym->ident)
2336 break;
2337 warning(token->pos, "void parameter");
2338 }
2339 add_symbol(list, sym);
2340 if (!match_op(token, ','))
2341 break;
2342 token = token->next;
2343 }
2344 return token;
2345 }
2346
2347 struct token *compound_statement(struct token *token, struct statement *stmt)
2348 {
2349 token = statement_list(token, &stmt->stmts);
2350 return token;
2351 }
2352
2353 static struct expression *identifier_expression(struct token *token)
2354 {
2355 struct expression *expr = alloc_expression(token->pos, EXPR_IDENTIFIER);
2356 expr->expr_ident = token->ident;
2357 return expr;
2358 }
2359
2360 static struct expression *index_expression(struct expression *from, struct expression *to)
2361 {
2362 int idx_from, idx_to;
2363 struct expression *expr = alloc_expression(from->pos, EXPR_INDEX);
2364
2365 idx_from = const_expression_value(from);
2366 idx_to = idx_from;
2367 if (to) {
2368 idx_to = const_expression_value(to);
2369 if (idx_to < idx_from || idx_from < 0)
2370 warning(from->pos, "nonsense array initializer index range");
2371 }
2372 expr->idx_from = idx_from;
2373 expr->idx_to = idx_to;
2374 return expr;
2375 }
2376
2377 static struct token *single_initializer(struct expression **ep, struct token *token)
2378 {
2379 int expect_equal = 0;
2380 struct token *next = token->next;
2381 struct expression **tail = ep;
2382 int nested;
2383
2384 *ep = NULL;
2385
2386 if ((token_type(token) == TOKEN_IDENT) && match_op(next, ':')) {
2387 struct expression *expr = identifier_expression(token);
2388 if (Wold_initializer)
2389 warning(token->pos, "obsolete struct initializer, use C99 syntax");
2390 token = initializer(&expr->ident_expression, next->next);
2391 if (expr->ident_expression)
2392 *ep = expr;
2393 return token;
2394 }
2395
2396 for (tail = ep, nested = 0; ; nested++, next = token->next) {
2397 if (match_op(token, '.') && (token_type(next) == TOKEN_IDENT)) {
2398 struct expression *expr = identifier_expression(next);
2399 *tail = expr;
2400 tail = &expr->ident_expression;
2401 expect_equal = 1;
2402 token = next->next;
2403 } else if (match_op(token, '[')) {
2404 struct expression *from = NULL, *to = NULL, *expr;
2405 token = constant_expression(token->next, &from);
2406 if (!from) {
2407 sparse_error(token->pos, "Expected constant expression");
2408 break;
2409 }
2410 if (match_op(token, SPECIAL_ELLIPSIS))
2411 token = constant_expression(token->next, &to);
2412 expr = index_expression(from, to);
2413 *tail = expr;
2414 tail = &expr->idx_expression;
2415 token = expect(token, ']', "at end of initializer index");
2416 if (nested)
2417 expect_equal = 1;
2418 } else {
2419 break;
2420 }
2421 }
2422 if (nested && !expect_equal) {
2423 if (!match_op(token, '='))
2424 warning(token->pos, "obsolete array initializer, use C99 syntax");
2425 else
2426 expect_equal = 1;
2427 }
2428 if (expect_equal)
2429 token = expect(token, '=', "at end of initializer index");
2430
2431 token = initializer(tail, token);
2432 if (!*tail)
2433 *ep = NULL;
2434 return token;
2435 }
2436
2437 static struct token *initializer_list(struct expression_list **list, struct token *token)
2438 {
2439 struct expression *expr;
2440
2441 for (;;) {
2442 token = single_initializer(&expr, token);
2443 if (!expr)
2444 break;
2445 add_expression(list, expr);
2446 if (!match_op(token, ','))
2447 break;
2448 token = token->next;
2449 }
2450 return token;
2451 }
2452
2453 struct token *initializer(struct expression **tree, struct token *token)
2454 {
2455 if (match_op(token, '{')) {
2456 struct expression *expr = alloc_expression(token->pos, EXPR_INITIALIZER);
2457 *tree = expr;
2458 token = initializer_list(&expr->expr_list, token->next);
2459 return expect(token, '}', "at end of initializer");
2460 }
2461 return assignment_expression(token, tree);
2462 }
2463
2464 static void declare_argument(struct symbol *sym, struct symbol *fn)
2465 {
2466 if (!sym->ident) {
2467 sparse_error(sym->pos, "no identifier for function argument");
2468 return;
2469 }
2470 bind_symbol(sym, sym->ident, NS_SYMBOL);
2471 }
2472
2473 static struct token *parse_function_body(struct token *token, struct symbol *decl,
2474 struct symbol_list **list)
2475 {
2476 struct symbol_list **old_symbol_list;
2477 struct symbol *base_type = decl->ctype.base_type;
2478 struct statement *stmt, **p;
2479 struct symbol *prev;
2480 struct symbol *arg;
2481
2482 old_symbol_list = function_symbol_list;
2483 if (decl->ctype.modifiers & MOD_INLINE) {
2484 function_symbol_list = &decl->inline_symbol_list;
2485 p = &base_type->inline_stmt;
2486 } else {
2487 function_symbol_list = &decl->symbol_list;
2488 p = &base_type->stmt;
2489 }
2490 function_computed_target_list = NULL;
2491 function_computed_goto_list = NULL;
2492
2493 if (decl->ctype.modifiers & MOD_EXTERN) {
2494 if (!(decl->ctype.modifiers & MOD_INLINE))
2495 warning(decl->pos, "function '%s' with external linkage has definition", show_ident(decl->ident));
2496 }
2497 if (!(decl->ctype.modifiers & MOD_STATIC))
2498 decl->ctype.modifiers |= MOD_EXTERN;
2499
2500 stmt = start_function(decl);
2501
2502 *p = stmt;
2503 FOR_EACH_PTR (base_type->arguments, arg) {
2504 declare_argument(arg, base_type);
2505 } END_FOR_EACH_PTR(arg);
2506
2507 token = compound_statement(token->next, stmt);
2508
2509 end_function(decl);
2510 if (!(decl->ctype.modifiers & MOD_INLINE))
2511 add_symbol(list, decl);
2512 check_declaration(decl);
2513 decl->definition = decl;
2514 prev = decl->same_symbol;
2515 if (prev && prev->definition) {
2516 warning(decl->pos, "multiple definitions for function '%s'",
2517 show_ident(decl->ident));
2518 info(prev->definition->pos, " the previous one is here");
2519 } else {
2520 while (prev) {
2521 prev->definition = decl;
2522 prev = prev->same_symbol;
2523 }
2524 }
2525 function_symbol_list = old_symbol_list;
2526 if (function_computed_goto_list) {
2527 if (!function_computed_target_list)
2528 warning(decl->pos, "function '%s' has computed goto but no targets?", show_ident(decl->ident));
2529 else {
2530 FOR_EACH_PTR(function_computed_goto_list, stmt) {
2531 stmt->target_list = function_computed_target_list;
2532 } END_FOR_EACH_PTR(stmt);
2533 }
2534 }
2535 return expect(token, '}', "at end of function");
2536 }
2537
2538 static void promote_k_r_types(struct symbol *arg)
2539 {
2540 struct symbol *base = arg->ctype.base_type;
2541 if (base && base->ctype.base_type == &int_type && (base->ctype.modifiers & (MOD_CHAR | MOD_SHORT))) {
2542 arg->ctype.base_type = &int_ctype;
2543 }
2544 }
2545
2546 static void apply_k_r_types(struct symbol_list *argtypes, struct symbol *fn)
2547 {
2548 struct symbol_list *real_args = fn->ctype.base_type->arguments;
2549 struct symbol *arg;
2550
2551 FOR_EACH_PTR(real_args, arg) {
2552 struct symbol *type;
2553
2554 /* This is quadratic in the number of arguments. We _really_ don't care */
2555 FOR_EACH_PTR(argtypes, type) {
2556 if (type->ident == arg->ident)
2557 goto match;
2558 } END_FOR_EACH_PTR(type);
2559 sparse_error(arg->pos, "missing type declaration for parameter '%s'", show_ident(arg->ident));
2560 continue;
2561 match:
2562 type->used = 1;
2563 /* "char" and "short" promote to "int" */
2564 promote_k_r_types(type);
2565
2566 arg->ctype = type->ctype;
2567 } END_FOR_EACH_PTR(arg);
2568
2569 FOR_EACH_PTR(argtypes, arg) {
2570 if (!arg->used)
2571 warning(arg->pos, "nonsensical parameter declaration '%s'", show_ident(arg->ident));
2572 } END_FOR_EACH_PTR(arg);
2573
2574 }
2575
2576 static struct token *parse_k_r_arguments(struct token *token, struct symbol *decl,
2577 struct symbol_list **list)
2578 {
2579 struct symbol_list *args = NULL;
2580
2581 warning(token->pos, "non-ANSI definition of function '%s'", show_ident(decl->ident));
2582 do {
2583 token = declaration_list(token, &args);
2584 if (!match_op(token, ';')) {
2585 sparse_error(token->pos, "expected ';' at end of parameter declaration");
2586 break;
2587 }
2588 token = token->next;
2589 } while (lookup_type(token));
2590
2591 apply_k_r_types(args, decl);
2592
2593 if (!match_op(token, '{')) {
2594 sparse_error(token->pos, "expected function body");
2595 return token;
2596 }
2597 return parse_function_body(token, decl, list);
2598 }
2599
2600 static struct token *toplevel_asm_declaration(struct token *token, struct symbol_list **list)
2601 {
2602 struct symbol *anon = alloc_symbol(token->pos, SYM_NODE);
2603 struct symbol *fn = alloc_symbol(token->pos, SYM_FN);
2604 struct statement *stmt;
2605
2606 anon->ctype.base_type = fn;
2607 stmt = alloc_statement(token->pos, STMT_NONE);
2608 fn->stmt = stmt;
2609
2610 token = parse_asm_statement(token, stmt);
2611
2612 add_symbol(list, anon);
2613 return token;
2614 }
2615
2616 struct token *external_declaration(struct token *token, struct symbol_list **list)
2617 {
2618 struct ident *ident = NULL;
2619 struct symbol *decl;
2620 struct decl_state ctx = { .ident = &ident };
2621 struct ctype saved;
2622 struct symbol *base_type;
2623 unsigned long mod;
2624 int is_typedef;
2625
2626 /* Top-level inline asm? */
2627 if (token_type(token) == TOKEN_IDENT) {
2628 struct symbol *s = lookup_keyword(token->ident, NS_KEYWORD);
2629 if (s && s->op->toplevel)
2630 return s->op->toplevel(token, list);
2631 }
2632
2633 /* Parse declaration-specifiers, if any */
2634 token = declaration_specifiers(token, &ctx);
2635 mod = storage_modifiers(&ctx);
2636 decl = alloc_symbol(token->pos, SYM_NODE);
2637 /* Just a type declaration? */
2638 if (match_op(token, ';')) {
2639 apply_modifiers(token->pos, &ctx);
2640 return token->next;
2641 }
2642
2643 saved = ctx.ctype;
2644 token = declarator(token, &ctx);
2645 token = handle_attributes(token, &ctx, KW_ATTRIBUTE | KW_ASM);
2646 apply_modifiers(token->pos, &ctx);
2647
2648 decl->ctype = ctx.ctype;
2649 decl->ctype.modifiers |= mod;
2650 decl->endpos = token->pos;
2651
2652 /* Just a type declaration? */
2653 if (!ident) {
2654 warning(token->pos, "missing identifier in declaration");
2655 return expect(token, ';', "at the end of type declaration");
2656 }
2657
2658 /* type define declaration? */
2659 is_typedef = ctx.storage_class == STypedef;
2660
2661 /* Typedefs don't have meaningful storage */
2662 if (is_typedef)
2663 decl->ctype.modifiers |= MOD_USERTYPE;
2664
2665 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
2666
2667 base_type = decl->ctype.base_type;
2668
2669 if (is_typedef) {
2670 if (base_type && !base_type->ident) {
2671 switch (base_type->type) {
2672 case SYM_STRUCT:
2673 case SYM_UNION:
2674 case SYM_ENUM:
2675 case SYM_RESTRICT:
2676 base_type->ident = ident;
2677 }
2678 }
2679 } else if (base_type && base_type->type == SYM_FN) {
2680 /* K&R argument declaration? */
2681 if (lookup_type(token))
2682 return parse_k_r_arguments(token, decl, list);
2683 if (match_op(token, '{'))
2684 return parse_function_body(token, decl, list);
2685
2686 if (!(decl->ctype.modifiers & MOD_STATIC))
2687 decl->ctype.modifiers |= MOD_EXTERN;
2688 } else if (base_type == &void_ctype && !(decl->ctype.modifiers & MOD_EXTERN)) {
2689 sparse_error(token->pos, "void declaration");
2690 }
2691
2692 for (;;) {
2693 if (!is_typedef && match_op(token, '=')) {
2694 if (decl->ctype.modifiers & MOD_EXTERN) {
2695 warning(decl->pos, "symbol with external linkage has initializer");
2696 decl->ctype.modifiers &= ~MOD_EXTERN;
2697 }
2698 token = initializer(&decl->initializer, token->next);
2699 }
2700 if (!is_typedef) {
2701 if (!(decl->ctype.modifiers & (MOD_EXTERN | MOD_INLINE))) {
2702 add_symbol(list, decl);
2703 fn_local_symbol(decl);
2704 }
2705 }
2706 check_declaration(decl);
2707 if (decl->same_symbol)
2708 decl->definition = decl->same_symbol->definition;
2709
2710 if (!match_op(token, ','))
2711 break;
2712
2713 token = token->next;
2714 ident = NULL;
2715 decl = alloc_symbol(token->pos, SYM_NODE);
2716 ctx.ctype = saved;
2717 token = handle_attributes(token, &ctx, KW_ATTRIBUTE);
2718 token = declarator(token, &ctx);
2719 token = handle_attributes(token, &ctx, KW_ATTRIBUTE | KW_ASM);
2720 apply_modifiers(token->pos, &ctx);
2721 decl->ctype = ctx.ctype;
2722 decl->ctype.modifiers |= mod;
2723 decl->endpos = token->pos;
2724 if (!ident) {
2725 sparse_error(token->pos, "expected identifier name in type definition");
2726 return token;
2727 }
2728
2729 if (is_typedef)
2730 decl->ctype.modifiers |= MOD_USERTYPE;
2731
2732 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
2733
2734 /* Function declarations are automatically extern unless specifically static */
2735 base_type = decl->ctype.base_type;
2736 if (!is_typedef && base_type && base_type->type == SYM_FN) {
2737 if (!(decl->ctype.modifiers & MOD_STATIC))
2738 decl->ctype.modifiers |= MOD_EXTERN;
2739 }
2740 }
2741 return expect(token, ';', "at end of declaration");
2742 }
Static analysis for C