parse.c

   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 Linus Torvalds
   8  *
   9  *  Licensed under the Open Software License version 1.1
  10  */
  11
  12 #include <stdarg.h>
  13 #include <stdlib.h>
  14 #include <stdio.h>
  15 #include <string.h>
  16 #include <ctype.h>
  17 #include <unistd.h>
  18 #include <fcntl.h>
  19
  20 #include "lib.h"
  21 #include "token.h"
  22 #include "parse.h"
  23 #include "symbol.h"
  24 #include "scope.h"
  25 #include "expression.h"
  26 #include "target.h"
  27
  28 static struct symbol_list **function_symbol_list;
  29
  30 // Add a symbol to the list of function-local symbols
  31 #define fn_local_symbol(x) add_symbol(function_symbol_list, (x))
  32
  33 static struct token *statement(struct token *token, struct statement **tree);
  34 static struct token *external_declaration(struct token *token, struct symbol_list **list);
  35
  36 struct statement *alloc_statement(struct position pos, int type)
  37 {
  38         struct statement *stmt = __alloc_statement(0);
  39         stmt->type = type;
  40         stmt->pos = pos;
  41         return stmt;
  42 }
  43
  44 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list);
  45
  46 static struct symbol * indirect(struct position pos, struct ctype *ctype, int type)
  47 {
  48         struct symbol *sym = alloc_symbol(pos, type);
  49
  50         sym->ctype.base_type = ctype->base_type;
  51         sym->ctype.modifiers = ctype->modifiers & ~MOD_STORAGE;
  52
  53         ctype->base_type = sym;
  54         ctype->modifiers &= MOD_STORAGE;
  55         return sym;
  56 }
  57
  58 static struct symbol *lookup_or_create_symbol(enum namespace ns, enum type type, struct token *token)
  59 {
  60         struct symbol *sym = lookup_symbol(token->ident, ns);
  61         if (!sym) {
  62                 sym = alloc_symbol(token->pos, type);
  63                 sym->ident = token->ident;
  64                 bind_symbol(sym, token->ident, ns);
  65                 if (type == SYM_LABEL)
  66                         fn_local_symbol(sym);
  67         }
  68         return sym;
  69 }
  70
  71 /*
  72  * NOTE! NS_LABEL is not just a different namespace,
  73  * it also ends up using function scope instead of the
  74  * regular symbol scope.
  75  */
  76 struct symbol *label_symbol(struct token *token)
  77 {
  78         return lookup_or_create_symbol(NS_LABEL, SYM_LABEL, token);
  79 }
  80
  81 struct token *struct_union_enum_specifier(enum namespace ns, enum type type,
  82         struct token *token, struct ctype *ctype,
  83         struct token *(*parse)(struct token *, struct symbol *))
  84 {
  85         struct symbol *sym;
  86
  87         ctype->modifiers = 0;
  88         if (token_type(token) == TOKEN_IDENT) {
  89                 sym = lookup_or_create_symbol(ns, type, token);
  90                 token = token->next;
  91                 ctype->base_type = sym;
  92                 if (match_op(token, '{')) {
  93                         token = parse(token->next, sym);
  94                         token = expect(token, '}', "at end of struct-union-enum-specifier");
  95                 }
  96                 return token;
  97         }
  98
  99         // private struct/union/enum type
 100         if (!match_op(token, '{')) {
 101                 warn(token->pos, "expected declaration");
 102                 ctype->base_type = &bad_type;
 103                 return token;
 104         }
 105
 106         sym = alloc_symbol(token->pos, type);
 107         token = parse(token->next, sym);
 108         ctype->base_type = sym;
 109         return expect(token, '}', "at end of specifier");
 110 }
 111
 112 static struct token *parse_struct_declaration(struct token *token, struct symbol *sym)
 113 {
 114         return struct_declaration_list(token, &sym->symbol_list);
 115 }
 116
 117 struct token *struct_or_union_specifier(enum type type, struct token *token, struct ctype *ctype)
 118 {
 119         return struct_union_enum_specifier(NS_STRUCT, type, token, ctype, parse_struct_declaration);
 120 }
 121
 122 static struct token *parse_enum_declaration(struct token *token, struct symbol *parent)
 123 {
 124         int nextval = 0;
 125         while (token_type(token) == TOKEN_IDENT) {
 126                 struct token *next = token->next;
 127                 struct symbol *sym;
 128
 129                 sym = alloc_symbol(token->pos, SYM_ENUM);
 130                 bind_symbol(sym, token->ident, NS_SYMBOL);
 131                 sym->ctype.base_type = parent;
 132
 133                 if (match_op(next, '=')) {
 134                         struct expression *expr;
 135                         next = constant_expression(next->next, &expr);
 136                         nextval = get_expression_value(expr);
 137                 }
 138                 sym->value = nextval;
 139
 140                 token = next;
 141                 if (!match_op(token, ','))
 142                         break;
 143                 token = token->next;
 144                 nextval = nextval + 1;
 145         }
 146         return token;
 147 }
 148
 149 struct token *enum_specifier(struct token *token, struct ctype *ctype)
 150 {
 151         return struct_union_enum_specifier(NS_ENUM, SYM_ENUM, token, ctype, parse_enum_declaration);
 152 }
 153
 154 struct token *typeof_specifier(struct token *token, struct ctype *ctype)
 155 {
 156         struct symbol *sym;
 157
 158         if (!match_op(token, '(')) {
 159                 warn(token->pos, "expected '(' after typeof");
 160                 return token;
 161         }
 162         if (lookup_type(token->next)) {
 163                 token = typename(token->next, &sym);
 164                 *ctype = sym->ctype;
 165         } else {
 166                 struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
 167                 token = parse_expression(token->next, &typeof_sym->initializer);
 168
 169                 ctype->modifiers = 0;
 170                 ctype->base_type = typeof_sym;
 171         }
 172         return expect(token, ')', "after typeof");
 173 }
 174
 175 static const char * handle_attribute(struct ctype *ctype, struct ident *attribute, struct expression *expr)
 176 {
 177         if (match_string_ident(attribute, "packed") ||
 178             match_string_ident(attribute, "__packed__")) {
 179                 ctype->alignment = 1;
 180                 return NULL;
 181         }
 182         if (match_string_ident(attribute, "aligned") ||
 183             match_string_ident(attribute, "__aligned__")) {
 184                 int alignment = MAX_ALIGNMENT;
 185                 if (expr)
 186                         alignment = get_expression_value(expr);
 187                 ctype->alignment = alignment;
 188                 return NULL;
 189         }
 190         if (match_string_ident(attribute, "nocast")) {
 191                 ctype->modifiers |= MOD_NOCAST;
 192                 return NULL;
 193         }
 194         if (match_string_ident(attribute, "noderef")) {
 195                 ctype->modifiers |= MOD_NODEREF;
 196                 return NULL;
 197         }
 198         if (match_string_ident(attribute, "address_space")) {
 199                 if (!expr)
 200                         return "expected address space number";
 201                 ctype->as = get_expression_value(expr);
 202                 return NULL;
 203         }
 204         if (match_string_ident(attribute, "context")) {
 205                 if (expr->type == EXPR_COMMA) {
 206                         int mask = get_expression_value(expr->left);
 207                         int value = get_expression_value(expr->right);
 208                         if (value & ~mask)
 209                                 return "nonsense attribute types";
 210                         ctype->contextmask |= mask;
 211                         ctype->context |= value;
 212                         return NULL;
 213                 }
 214                 return "expected context mask and value";
 215         }
 216
 217         /* Throw away for now.. */
 218         if (match_string_ident(attribute, "format") ||
 219             match_string_ident(attribute, "__format__"))
 220                 return NULL;
 221         if (match_string_ident(attribute, "section") ||
 222             match_string_ident(attribute, "__section__"))
 223                 return NULL;
 224         if (match_string_ident(attribute, "unused") ||
 225             match_string_ident(attribute, "__unused__"))
 226                 return NULL;
 227         if (match_string_ident(attribute, "const") ||
 228             match_string_ident(attribute, "__const__"))
 229                 return NULL;
 230         if (match_string_ident(attribute, "noreturn"))
 231                 return NULL;
 232         if (match_string_ident(attribute, "regparm"))
 233                 return NULL;
 234         if (match_string_ident(attribute, "weak"))
 235                 return NULL;
 236
 237         return "unknown attribute";
 238 }
 239
 240 struct token *attribute_specifier(struct token *token, struct ctype *ctype)
 241 {
 242         ctype->modifiers = 0;
 243         token = expect(token, '(', "after attribute");
 244         token = expect(token, '(', "after attribute");
 245
 246         for (;;) {
 247                 const char *error;
 248                 struct ident *attribute_name;
 249                 struct expression *attribute_expr;
 250
 251                 if (eof_token(token))
 252                         break;
 253                 if (match_op(token, ';'))
 254                         break;
 255                 if (token_type(token) != TOKEN_IDENT)
 256                         break;
 257                 attribute_name = token->ident;
 258                 token = token->next;
 259                 attribute_expr = NULL;
 260                 if (match_op(token, '('))
 261                         token = parens_expression(token, &attribute_expr, "in attribute");
 262                 error = handle_attribute(ctype, attribute_name, attribute_expr);
 263                 if (error)
 264                         warn(token->pos, "attribute '%s': %s", show_ident(attribute_name), error);
 265                 if (!match_op(token, ','))
 266                         break;
 267                 token = token->next;
 268         }
 269
 270         token = expect(token, ')', "after attribute");
 271         token = expect(token, ')', "after attribute");
 272         return token;
 273 }
 274
 275 #define MOD_SPECIALBITS (MOD_STRUCTOF | MOD_UNIONOF | MOD_ENUMOF | MOD_ATTRIBUTE | MOD_TYPEOF)
 276 #define MOD_SPECIFIER (MOD_CHAR | MOD_SHORT | MOD_LONG | MOD_LONGLONG | MOD_SIGNED | MOD_UNSIGNED)
 277
 278 struct symbol * ctype_integer(unsigned int spec)
 279 {
 280         static struct symbol *const integer_ctypes[][2] = {
 281                 { &llong_ctype, &ullong_ctype },
 282                 { &long_ctype,  &ulong_ctype  },
 283                 { &short_ctype, &ushort_ctype },
 284                 { &char_ctype,  &uchar_ctype  },
 285                 { &int_ctype,   &uint_ctype   },
 286         };
 287         struct symbol *const (*ctype)[2];
 288
 289         ctype = integer_ctypes;
 290         if (!(spec & MOD_LONGLONG)) {
 291                 ctype++;
 292                 if (!(spec & MOD_LONG)) {
 293                         ctype++;
 294                         if (!(spec & MOD_SHORT)) {
 295                                 ctype++;
 296                                 if (!(spec & MOD_CHAR))
 297                                         ctype++;
 298                         }
 299                 }
 300         }
 301         return ctype[0][(spec & MOD_UNSIGNED) != 0];
 302 }
 303
 304 struct symbol * ctype_fp(unsigned int spec)
 305 {
 306         if (spec & MOD_LONGLONG)
 307                 return &ldouble_ctype;
 308         if (spec & MOD_LONG)
 309                 return &double_ctype;
 310         return &float_ctype;
 311 }
 312
 313 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype)
 314 {
 315         unsigned long mod = thistype->modifiers;
 316
 317         if (mod) {
 318                 unsigned long old = ctype->modifiers;
 319                 unsigned long extra = 0, dup;
 320
 321                 if (mod & old & MOD_LONG) {
 322                         extra = MOD_LONGLONG | MOD_LONG;
 323                         mod &= ~MOD_LONG;
 324                         old &= ~MOD_LONG;
 325                 }
 326                 dup = (mod & old) | (extra & old) | (extra & mod);
 327                 if (dup)
 328                         warn(pos, "Just how %sdo you want this type to be?",
 329                                 modifier_string(dup));
 330                 ctype->modifiers = old | mod | extra;
 331         }
 332
 333         /* Context mask and value */
 334         if ((ctype->context ^ thistype->context) & (ctype->contextmask & thistype->contextmask)) {
 335                 warn(pos, "inconsistend attribute types");
 336                 thistype->context = 0;
 337                 thistype->contextmask = 0;
 338         }
 339         ctype->context |= thistype->context;
 340         ctype->contextmask |= thistype->contextmask;
 341
 342         /* Alignment */
 343         if (thistype->alignment & (thistype->alignment-1)) {
 344                 warn(pos, "I don't like non-power-of-2 alignments");
 345                 thistype->alignment = 0;
 346         }
 347         if (thistype->alignment > ctype->alignment)
 348                 ctype->alignment = thistype->alignment;
 349
 350         /* Address space */
 351         ctype->as = thistype->as;
 352 }
 353
 354
 355 static struct token *declaration_specifiers(struct token *next, struct ctype *ctype, int qual)
 356 {
 357         struct token *token;
 358
 359         while ( (token = next) != NULL ) {
 360                 struct ctype thistype;
 361                 struct ident *ident;
 362                 struct symbol *s, *type;
 363                 unsigned long mod;
 364
 365                 next = token->next;
 366                 if (token_type(token) != TOKEN_IDENT)
 367                         break;
 368                 ident = token->ident;
 369
 370                 s = lookup_symbol(ident, NS_TYPEDEF);
 371                 if (!s)
 372                         break;
 373                 thistype = s->ctype;
 374                 mod = thistype.modifiers;
 375                 if (qual && (mod & ~(MOD_ATTRIBUTE | MOD_CONST | MOD_VOLATILE)))
 376                         break;
 377                 if (mod & MOD_SPECIALBITS) {
 378                         if (mod & MOD_STRUCTOF)
 379                                 next = struct_or_union_specifier(SYM_STRUCT, next, &thistype);
 380                         else if (mod & MOD_UNIONOF)
 381                                 next = struct_or_union_specifier(SYM_UNION, next, &thistype);
 382                         else if (mod & MOD_ENUMOF)
 383                                 next = enum_specifier(next, &thistype);
 384                         else if (mod & MOD_ATTRIBUTE)
 385                                 next = attribute_specifier(next, &thistype);
 386                         else if (mod & MOD_TYPEOF)
 387                                 next = typeof_specifier(next, &thistype);
 388                         mod = thistype.modifiers;
 389                 }
 390                 type = thistype.base_type;
 391                 if (type) {
 392                         if (qual)
 393                                 break;
 394                         if (type != ctype->base_type) {
 395                                 if (ctype->base_type)
 396                                         break;
 397                                 ctype->base_type = type;
 398                         }
 399                 }
 400
 401                 apply_ctype(token->pos, &thistype, ctype);
 402         }
 403
 404         /* Turn the "virtual types" into real types with real sizes etc */
 405         if (!ctype->base_type && (ctype->modifiers & MOD_SPECIFIER))
 406                 ctype->base_type = &int_type;
 407
 408         if (ctype->base_type == &int_type) {
 409                 ctype->base_type = ctype_integer(ctype->modifiers & MOD_SPECIFIER);
 410                 ctype->modifiers &= ~MOD_SPECIFIER;
 411                 return token;
 412         }
 413         if (ctype->base_type == &fp_type) {
 414                 ctype->base_type = ctype_fp(ctype->modifiers & MOD_SPECIFIER);
 415                 ctype->modifiers &= ~MOD_SPECIFIER;
 416                 return token;
 417         }
 418         return token;
 419 }
 420
 421 static struct token *abstract_array_declarator(struct token *token, struct symbol *sym)
 422 {
 423         struct expression *expr = NULL;
 424
 425         token = parse_expression(token, &expr);
 426         sym->array_size = expr;
 427         return token;
 428 }
 429
 430 static struct token *parameter_type_list(struct token *, struct symbol *);
 431 static struct token *declarator(struct token *token, struct symbol **tree, struct ident **p);
 432
 433 static struct token *direct_declarator(struct token *token, struct symbol **tree, struct ident **p)
 434 {
 435         struct ctype *ctype = &(*tree)->ctype;
 436
 437         if (p && token_type(token) == TOKEN_IDENT) {
 438                 *p = token->ident;
 439                 token = token->next;
 440         }
 441
 442         for (;;) {
 443                 if (match_ident(token, &__attribute___ident) || match_ident(token, &__attribute_ident)) {
 444                         struct ctype thistype = { 0, };
 445                         token = attribute_specifier(token->next, &thistype);
 446                         apply_ctype(token->pos, &thistype, ctype);
 447                         continue;
 448                 }
 449                 if (token_type(token) != TOKEN_SPECIAL)
 450                         return token;
 451
 452                 /*
 453                  * This can be either a parameter list or a grouping.
 454                  * For the direct (non-abstract) case, we know if must be
 455                  * a paramter list if we already saw the identifier.
 456                  * For the abstract case, we know if must be a parameter
 457                  * list if it is empty or starts with a type.
 458                  */
 459                 if (token->special == '(') {
 460                         struct symbol *sym;
 461                         struct token *next = token->next;
 462                         int fn = (p && *p) || match_op(next, ')') || lookup_type(next);
 463
 464                         if (!fn) {
 465                                 struct symbol *base_type = ctype->base_type;
 466                                 token = declarator(next, tree, p);
 467                                 token = expect(token, ')', "in nested declarator");
 468                                 while (ctype->base_type != base_type)
 469                                         ctype = &ctype->base_type->ctype;
 470                                 p = NULL;
 471                                 continue;
 472                         }
 473
 474                         sym = indirect(token->pos, ctype, SYM_FN);
 475                         token = parameter_type_list(next, sym);
 476                         token = expect(token, ')', "in function declarator");
 477                         continue;
 478                 }
 479                 if (token->special == '[') {
 480                         struct symbol *array = indirect(token->pos, ctype, SYM_ARRAY);
 481                         token = abstract_array_declarator(token->next, array);
 482                         token = expect(token, ']', "in abstract_array_declarator");
 483                         ctype = &array->ctype;
 484                         continue;
 485                 }
 486                 if (token->special == ':') {
 487                         struct symbol *bitfield;
 488                         struct expression *expr;
 489                         bitfield = indirect(token->pos, ctype, SYM_BITFIELD);
 490                         token = conditional_expression(token->next, &expr);
 491                         bitfield->fieldwidth = get_expression_value(expr);
 492                         continue;
 493                 }
 494                 break;
 495         }
 496         if (p) {
 497                 (*tree)->ident = *p;
 498         }
 499         return token;
 500 }
 501
 502 static struct token *pointer(struct token *token, struct ctype *ctype)
 503 {
 504         unsigned long modifiers;
 505         struct symbol *base_type;
 506
 507         modifiers = ctype->modifiers & ~(MOD_TYPEDEF | MOD_ATTRIBUTE);
 508         base_type = ctype->base_type;
 509         ctype->modifiers = modifiers;
 510
 511         while (match_op(token,'*')) {
 512                 struct symbol *ptr = alloc_symbol(token->pos, SYM_PTR);
 513                 ptr->ctype.modifiers = modifiers & ~MOD_STORAGE;
 514                 ptr->ctype.as = ctype->as;
 515                 ptr->ctype.context = ctype->context;
 516                 ptr->ctype.contextmask = ctype->contextmask;
 517                 ptr->ctype.base_type = base_type;
 518
 519                 base_type = ptr;
 520                 ctype->modifiers = modifiers & MOD_STORAGE;
 521                 ctype->base_type = base_type;
 522                 ctype->as = 0;
 523                 ctype->context = 0;
 524                 ctype->contextmask = 0;
 525
 526                 token = declaration_specifiers(token->next, ctype, 1);
 527         }
 528         return token;
 529 }
 530
 531 static struct token *declarator(struct token *token, struct symbol **tree, struct ident **p)
 532 {
 533         token = pointer(token, &(*tree)->ctype);
 534         return direct_declarator(token, tree, p);
 535 }
 536
 537 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list)
 538 {
 539         while (!match_op(token, '}')) {
 540                 struct ctype ctype = {0, };
 541
 542                 token = declaration_specifiers(token, &ctype, 0);
 543                 for (;;) {
 544                         struct ident *ident = NULL;
 545                         struct symbol *decl = alloc_symbol(token->pos, SYM_NODE);
 546                         decl->ctype = ctype;
 547                         token = pointer(token, &decl->ctype);
 548                         token = direct_declarator(token, &decl, &ident);
 549                         if (match_op(token, ':')) {
 550                                 struct expression *expr;
 551                                 token = parse_expression(token->next, &expr);
 552                         }
 553                         add_symbol(list, decl);
 554                         if (!match_op(token, ','))
 555                                 break;
 556                         token = token->next;
 557                 }
 558                 if (!match_op(token, ';'))
 559                         break;
 560                 token = token->next;
 561         }
 562         return token;
 563 }
 564
 565 static struct token *parameter_declaration(struct token *token, struct symbol **tree)
 566 {
 567         struct ident *ident = NULL;
 568         struct symbol *sym;
 569         struct ctype ctype = { 0, };
 570
 571         token = declaration_specifiers(token, &ctype, 0);
 572         sym = alloc_symbol(token->pos, SYM_NODE);
 573         sym->ctype = ctype;
 574         *tree = sym;
 575         token = pointer(token, &sym->ctype);
 576         token = direct_declarator(token, tree, &ident);
 577         return token;
 578 }
 579
 580 struct token *typename(struct token *token, struct symbol **p)
 581 {
 582         struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
 583         *p = sym;
 584         token = declaration_specifiers(token, &sym->ctype, 0);
 585         return declarator(token, &sym, NULL);
 586 }
 587
 588 struct token *expression_statement(struct token *token, struct expression **tree)
 589 {
 590         token = parse_expression(token, tree);
 591         return expect(token, ';', "at end of statement");
 592 }
 593
 594 static struct token *parse_asm_operands(struct token *token, struct statement *stmt)
 595 {
 596         struct expression *expr;
 597
 598         /* Allow empty operands */
 599         if (match_op(token->next, ':') || match_op(token->next, ')'))
 600                 return token->next;
 601         do {
 602                 token = primary_expression(token->next, &expr);
 603                 token = parens_expression(token, &expr, "in asm parameter");
 604         } while (match_op(token, ','));
 605         return token;
 606 }
 607
 608 static struct token *parse_asm_clobbers(struct token *token, struct statement *stmt)
 609 {
 610         struct expression *expr;
 611
 612         do {
 613                 token = primary_expression(token->next, &expr);
 614         } while (match_op(token, ','));
 615         return token;
 616 }
 617
 618 /* Make a statement out of an expression */
 619 static struct statement *make_statement(struct expression *expr)
 620 {
 621         struct statement *stmt;
 622
 623         if (!expr)
 624                 return NULL;
 625         stmt = alloc_statement(expr->pos, STMT_EXPRESSION);
 626         stmt->expression = expr;
 627         return stmt;
 628 }
 629
 630 /*
 631  * All iterators have two symbols associated with them:
 632  * the "continue" and "break" symbols, which are targets
 633  * for continue and break statements respectively.
 634  *
 635  * They are in a special name-space, but they follow
 636  * all the normal visibility rules, so nested iterators
 637  * automatically work right.
 638  */
 639 static void start_iterator(struct statement *stmt)
 640 {
 641         struct symbol *cont, *brk;
 642
 643         start_symbol_scope();
 644         cont = alloc_symbol(stmt->pos, SYM_NODE);
 645         cont->ident = &continue_ident;
 646         bind_symbol(cont, &continue_ident, NS_ITERATOR);
 647         brk = alloc_symbol(stmt->pos, SYM_NODE);
 648         brk->ident = &break_ident;
 649         bind_symbol(brk, &break_ident, NS_ITERATOR);
 650
 651         stmt->type = STMT_ITERATOR;
 652         stmt->iterator_break = brk;
 653         stmt->iterator_continue = cont;
 654         fn_local_symbol(brk);
 655         fn_local_symbol(cont);
 656 }
 657
 658 static void end_iterator(struct statement *stmt)
 659 {
 660         end_symbol_scope();
 661 }
 662
 663 static struct statement *start_function(struct symbol *sym)
 664 {
 665         struct symbol *ret;
 666         struct statement *stmt = alloc_statement(sym->pos, STMT_COMPOUND);
 667
 668         start_function_scope();
 669         ret = alloc_symbol(sym->pos, SYM_NODE);
 670         ret->ident = &return_ident;
 671         ret->ctype = sym->ctype.base_type->ctype;
 672         ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
 673         ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
 674         bind_symbol(ret, &return_ident, NS_ITERATOR);
 675         stmt->ret = ret;
 676
 677         fn_local_symbol(ret);
 678         return stmt;
 679 }
 680
 681 static void end_function(struct symbol *sym)
 682 {
 683         end_function_scope();
 684 }
 685
 686 /*
 687  * A "switch()" statement, like an iterator, has a
 688  * the "break" symbol associated with it. It works
 689  * exactly like the iterator break - it's the target
 690  * for any break-statements in scope, and means that
 691  * "break" handling doesn't even need to know whether
 692  * it's breaking out of an iterator or a switch.
 693  *
 694  * In addition, the "case" symbol is a marker for the
 695  * case/default statements to find the switch statement
 696  * that they are associated with.
 697  */
 698 static void start_switch(struct statement *stmt)
 699 {
 700         struct symbol *brk, *switch_case;
 701
 702         start_symbol_scope();
 703         brk = alloc_symbol(stmt->pos, SYM_NODE);
 704         brk->ident = &break_ident;
 705         bind_symbol(brk, &break_ident, NS_ITERATOR);
 706
 707         switch_case = alloc_symbol(stmt->pos, SYM_NODE);
 708         switch_case->ident = &case_ident;
 709         bind_symbol(switch_case, &case_ident, NS_ITERATOR);
 710         switch_case->stmt = stmt;
 711
 712         stmt->type = STMT_SWITCH;
 713         stmt->switch_break = brk;
 714         stmt->switch_case = switch_case;
 715
 716         fn_local_symbol(brk);
 717         fn_local_symbol(switch_case);
 718 }
 719
 720 static void end_switch(struct statement *stmt)
 721 {
 722         if (!stmt->switch_case->symbol_list)
 723                 warn(stmt->pos, "switch with no cases");
 724         end_symbol_scope();
 725 }
 726
 727 static void add_case_statement(struct statement *stmt)
 728 {
 729         struct symbol *target = lookup_symbol(&case_ident, NS_ITERATOR);
 730         struct symbol *sym;
 731
 732         if (!target) {
 733                 warn(stmt->pos, "not in switch scope");
 734                 return;
 735         }
 736         sym = alloc_symbol(stmt->pos, SYM_NODE);
 737         add_symbol(&target->symbol_list, sym);
 738         sym->stmt = stmt;
 739         stmt->case_label = sym;
 740         fn_local_symbol(sym);
 741 }
 742
 743 static struct token *parse_return_statement(struct token *token, struct statement *stmt)
 744 {
 745         struct symbol *target = lookup_symbol(&return_ident, NS_ITERATOR);
 746
 747         if (!target)
 748                 error(token->pos, "internal error: return without a function target");
 749         stmt->type = STMT_RETURN;
 750         stmt->ret_target = target;
 751         return expression_statement(token->next, &stmt->ret_value);
 752 }
 753
 754 static struct token *parse_for_statement(struct token *token, struct statement *stmt)
 755 {
 756         struct symbol_list *syms;
 757         struct expression *e1, *e2, *e3;
 758         struct statement *iterator;
 759
 760         start_iterator(stmt);
 761         token = expect(token->next, '(', "after 'for'");
 762
 763         syms = NULL;
 764         e1 = NULL;
 765         /* C99 variable declaration? */
 766         if (lookup_type(token)) {
 767                 token = external_declaration(token, &syms);
 768         } else {
 769                 token = parse_expression(token, &e1);
 770                 token = expect(token, ';', "in 'for'");
 771         }
 772         token = parse_expression(token, &e2);
 773         token = expect(token, ';', "in 'for'");
 774         token = parse_expression(token, &e3);
 775         token = expect(token, ')', "in 'for'");
 776         token = statement(token, &iterator);
 777
 778         stmt->iterator_syms = syms;
 779         stmt->iterator_pre_statement = make_statement(e1);
 780         stmt->iterator_pre_condition = e2;
 781         stmt->iterator_post_statement = make_statement(e3);
 782         stmt->iterator_post_condition = e2;
 783         stmt->iterator_statement = iterator;
 784         end_iterator(stmt);
 785
 786         return token;
 787 }
 788
 789 struct token *parse_while_statement(struct token *token, struct statement *stmt)
 790 {
 791         struct expression *expr;
 792         struct statement *iterator;
 793
 794         start_iterator(stmt);
 795         token = parens_expression(token->next, &expr, "after 'while'");
 796         token = statement(token, &iterator);
 797
 798         stmt->iterator_pre_condition = expr;
 799         stmt->iterator_post_condition = expr;
 800         stmt->iterator_statement = iterator;
 801         end_iterator(stmt);
 802
 803         return token;
 804 }
 805
 806 struct token *parse_do_statement(struct token *token, struct statement *stmt)
 807 {
 808         struct expression *expr;
 809         struct statement *iterator;
 810
 811         start_iterator(stmt);
 812         token = statement(token->next, &iterator);
 813         if (token_type(token) == TOKEN_IDENT && token->ident == &while_ident)
 814                 token = token->next;
 815         else
 816                 warn(token->pos, "expected 'while' after 'do'");
 817         token = parens_expression(token, &expr, "after 'do-while'");
 818
 819         stmt->iterator_post_condition = expr;
 820         stmt->iterator_statement = iterator;
 821         end_iterator(stmt);
 822
 823         return expect(token, ';', "after statement");
 824 }
 825
 826 static struct token *statement(struct token *token, struct statement **tree)
 827 {
 828         struct statement *stmt = alloc_statement(token->pos, STMT_NONE);
 829
 830         *tree = stmt;
 831         if (token_type(token) == TOKEN_IDENT) {
 832                 if (token->ident == &if_ident) {
 833                         stmt->type = STMT_IF;
 834                         token = parens_expression(token->next, &stmt->if_conditional, "after if");
 835                         token = statement(token, &stmt->if_true);
 836                         if (token_type(token) != TOKEN_IDENT)
 837                                 return token;
 838                         if (token->ident != &else_ident)
 839                                 return token;
 840                         return statement(token->next, &stmt->if_false);
 841                 }
 842
 843                 if (token->ident == &return_ident)
 844                         return parse_return_statement(token, stmt);
 845
 846                 if (token->ident == &break_ident || token->ident == &continue_ident) {
 847                         struct symbol *target = lookup_symbol(token->ident, NS_ITERATOR);
 848                         stmt->type = STMT_GOTO;
 849                         stmt->goto_label = target;
 850                         if (!target)
 851                                 warn(stmt->pos, "break/continue not in iterator scope");
 852                         return expect(token->next, ';', "at end of statement");
 853                 }
 854                 if (token->ident == &default_ident) {
 855                         token = token->next;
 856                         goto default_statement;
 857                 }
 858                 if (token->ident == &case_ident) {
 859                         token = parse_expression(token->next, &stmt->case_expression);
 860                         if (match_op(token, SPECIAL_ELLIPSIS))
 861                                 token = parse_expression(token->next, &stmt->case_to);
 862 default_statement:
 863                         stmt->type = STMT_CASE;
 864                         token = expect(token, ':', "after default/case");
 865                         add_case_statement(stmt);
 866                         return statement(token, &stmt->case_statement);
 867                 }
 868                 if (token->ident == &switch_ident) {
 869                         stmt->type = STMT_SWITCH;
 870                         start_switch(stmt);
 871                         token = parens_expression(token->next, &stmt->switch_expression, "after 'switch'");
 872                         token = statement(token, &stmt->switch_statement);
 873                         end_switch(stmt);
 874                         return token;
 875                 }
 876                 if (token->ident == &for_ident)
 877                         return parse_for_statement(token, stmt);
 878
 879                 if (token->ident == &while_ident)
 880                         return parse_while_statement(token, stmt);
 881
 882                 if (token->ident == &do_ident)
 883                         return parse_do_statement(token, stmt);
 884
 885                 if (token->ident == &goto_ident) {
 886                         stmt->type = STMT_GOTO;
 887                         token = token->next;
 888                         if (match_op(token, '*')) {
 889                                 token = parse_expression(token->next, &stmt->goto_expression);
 890                         } else if (token_type(token) == TOKEN_IDENT) {
 891                                 stmt->goto_label = label_symbol(token);
 892                                 token = token->next;
 893                         } else {
 894                                 warn(token->pos, "Expected identifier or goto expression");
 895                         }
 896                         return expect(token, ';', "at end of statement");
 897                 }
 898                 if (token->ident == &asm_ident || token->ident == &__asm___ident || token->ident == &__asm_ident) {
 899                         struct expression *expr;
 900                         stmt->type = STMT_ASM;
 901                         token = token->next;
 902                         if (token_type(token) == TOKEN_IDENT) {
 903                                 if (token->ident == &__volatile___ident || token->ident == &volatile_ident)
 904                                         token = token->next;
 905                         }
 906                         token = expect(token, '(', "after asm");
 907                         token = parse_expression(token->next, &expr);
 908                         if (match_op(token, ':'))
 909                                 token = parse_asm_operands(token, stmt);
 910                         if (match_op(token, ':'))
 911                                 token = parse_asm_operands(token, stmt);
 912                         if (match_op(token, ':'))
 913                                 token = parse_asm_clobbers(token, stmt);
 914                         token = expect(token, ')', "after asm");
 915                         return expect(token, ';', "at end of asm-statement");
 916                 }
 917                 if (match_op(token->next, ':')) {
 918                         stmt->type = STMT_LABEL;
 919                         stmt->label_identifier = label_symbol(token);
 920                         return statement(token->next->next, &stmt->label_statement);
 921                 }
 922         }
 923
 924         if (match_op(token, '{')) {
 925                 stmt->type = STMT_COMPOUND;
 926                 start_symbol_scope();
 927                 token = compound_statement(token->next, stmt);
 928                 end_symbol_scope();
 929
 930                 return expect(token, '}', "at end of compound statement");
 931         }
 932
 933         stmt->type = STMT_EXPRESSION;
 934         return expression_statement(token, &stmt->expression);
 935 }
 936
 937 struct token * statement_list(struct token *token, struct statement_list **list)
 938 {
 939         for (;;) {
 940                 struct statement * stmt;
 941                 if (eof_token(token))
 942                         break;
 943                 if (match_op(token, '}'))
 944                         break;
 945                 token = statement(token, &stmt);
 946                 add_statement(list, stmt);
 947         }
 948         return token;
 949 }
 950
 951 static struct token *parameter_type_list(struct token *token, struct symbol *fn)
 952 {
 953         struct symbol_list **list = &fn->arguments;
 954         for (;;) {
 955                 struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
 956
 957                 if (match_op(token, SPECIAL_ELLIPSIS)) {
 958                         fn->variadic = 1;
 959                         token = token->next;
 960                         break;
 961                 }
 962
 963                 if (!lookup_type(token)) {
 964                         warn(token->pos, "non-ANSI parameter list");
 965                         break;
 966                 }
 967                 token = parameter_declaration(token, &sym);
 968                 /* Special case: (void) */
 969                 if (!*list && !sym->ident && sym->ctype.base_type == &void_ctype)
 970                         break;
 971                 add_symbol(list, sym);
 972                 if (!match_op(token, ','))
 973                         break;
 974                 token = token->next;
 975
 976         }
 977         return token;
 978 }
 979
 980 struct token *compound_statement(struct token *token, struct statement *stmt)
 981 {
 982         while (!eof_token(token)) {
 983                 if (!lookup_type(token))
 984                         break;
 985                 token = external_declaration(token, &stmt->syms);
 986         }
 987         token = statement_list(token, &stmt->stmts);
 988         return token;
 989 }
 990
 991 static struct expression *identifier_expression(struct token *token)
 992 {
 993         struct expression *expr = alloc_expression(token->pos, EXPR_IDENTIFIER);
 994         expr->expr_ident = token->ident;
 995         return expr;
 996 }
 997
 998 static struct expression *index_expression(struct expression *from, struct expression *to)
 999 {
1000         int idx_from, idx_to;
1001         struct expression *expr = alloc_expression(from->pos, EXPR_INDEX);
1002
1003         idx_from = get_expression_value(from);
1004         idx_to = idx_from;
1005         if (to) {
1006                 idx_to = get_expression_value(to);
1007                 if (idx_to < idx_from || idx_from < 0)
1008                         warn(from->pos, "nonsense array initializer index range");
1009         }
1010         expr->idx_from = idx_from;
1011         expr->idx_to = idx_to;
1012         return expr;
1013 }
1014
1015 static struct token *initializer_list(struct expression_list **list, struct token *token)
1016 {
1017         for (;;) {
1018                 struct token *next = token->next;
1019                 struct expression *expr;
1020
1021                 if (match_op(token, '.') && (token_type(next) == TOKEN_IDENT) && match_op(next->next, '=')) {
1022                         add_expression(list, identifier_expression(next));
1023                         token = next->next->next;
1024                 } else if ((token_type(token) == TOKEN_IDENT) && match_op(next, ':')) {
1025                         add_expression(list, identifier_expression(token));
1026                         token = next->next;
1027                 } else if (match_op(token, '[')) {
1028                         struct expression *from = NULL, *to = NULL;
1029                         token = constant_expression(token->next, &from);
1030                         if (match_op(token, SPECIAL_ELLIPSIS))
1031                                 token = constant_expression(token->next, &to);
1032                         add_expression(list, index_expression(from, to));
1033                         token = expect(token, ']', "at end of initializer index");
1034                         token = expect(token, '=', "at end of initializer index");
1035                 }
1036
1037                 expr = NULL;
1038                 token = initializer(&expr, token);
1039                 if (!expr)
1040                         break;
1041                 add_expression(list, expr);
1042                 if (!match_op(token, ','))
1043                         break;
1044                 token = token->next;
1045         }
1046         return token;
1047 }
1048
1049 struct token *initializer(struct expression **tree, struct token *token)
1050 {
1051         if (match_op(token, '{')) {
1052                 struct expression *expr = alloc_expression(token->pos, EXPR_INITIALIZER);
1053                 *tree = expr;
1054                 token = initializer_list(&expr->expr_list, token->next);
1055                 return expect(token, '}', "at end of initializer");
1056         }
1057         return assignment_expression(token, tree);
1058 }
1059
1060 static void declare_argument(struct symbol *sym, struct symbol *fn)
1061 {
1062         if (!sym->ident) {
1063                 warn(sym->pos, "no identifier for function argument");
1064                 return;
1065         }
1066         bind_symbol(sym, sym->ident, NS_SYMBOL);
1067 }
1068
1069 static struct token *parse_function_body(struct token *token, struct symbol *decl,
1070         struct symbol_list **list)
1071 {
1072         struct symbol *base_type = decl->ctype.base_type;
1073         struct statement *stmt;
1074         struct symbol *arg;
1075
1076         function_symbol_list = &decl->symbol_list;
1077         if (decl->ctype.modifiers & MOD_EXTERN) {
1078                 if (!(decl->ctype.modifiers & MOD_INLINE))
1079                         warn(decl->pos, "function with external linkage has definition");
1080         }
1081         if (!(decl->ctype.modifiers & MOD_STATIC))
1082                 decl->ctype.modifiers |= MOD_EXTERN;
1083
1084         stmt = start_function(decl);
1085
1086         base_type->stmt = stmt;
1087         FOR_EACH_PTR (base_type->arguments, arg) {
1088                 declare_argument(arg, base_type);
1089         } END_FOR_EACH_PTR;
1090
1091         token = compound_statement(token->next, stmt);
1092
1093         end_function(decl);
1094         if (!(decl->ctype.modifiers & MOD_INLINE))
1095                 add_symbol(list, decl);
1096         check_declaration(decl);
1097         function_symbol_list = NULL;
1098         return expect(token, '}', "at end of function");
1099 }
1100
1101 static struct token *external_declaration(struct token *token, struct symbol_list **list)
1102 {
1103         struct ident *ident = NULL;
1104         struct symbol *decl;
1105         struct ctype ctype = { 0, };
1106         struct symbol *base_type;
1107         int is_typedef;
1108
1109         /* Parse declaration-specifiers, if any */
1110         token = declaration_specifiers(token, &ctype, 0);
1111         decl = alloc_symbol(token->pos, SYM_NODE);
1112         decl->ctype = ctype;
1113         token = pointer(token, &decl->ctype);
1114         token = declarator(token, &decl, &ident);
1115
1116         /* Just a type declaration? */
1117         if (!ident)
1118                 return expect(token, ';', "end of type declaration");
1119
1120         decl->ident = ident;
1121
1122         /* type define declaration? */
1123         is_typedef = (ctype.modifiers & MOD_TYPEDEF) != 0;
1124
1125         /* Typedef's don't have meaningful storage */
1126         if (is_typedef) {
1127                 ctype.modifiers &= ~MOD_STORAGE;
1128                 decl->ctype.modifiers &= ~MOD_STORAGE;
1129         }
1130
1131         bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1132
1133         base_type = decl->ctype.base_type;
1134         if (!is_typedef && base_type && base_type->type == SYM_FN) {
1135                 if (match_op(token, '{'))
1136                         return parse_function_body(token, decl, list);
1137
1138                 if (!(decl->ctype.modifiers & MOD_STATIC))
1139                         decl->ctype.modifiers |= MOD_EXTERN;
1140         }
1141
1142         for (;;) {
1143                 if (!is_typedef && match_op(token, '=')) {
1144                         if (decl->ctype.modifiers & MOD_EXTERN) {
1145                                 warn(decl->pos, "symbol with external linkage has initializer");
1146                                 decl->ctype.modifiers &= ~MOD_EXTERN;
1147                         }
1148                         token = initializer(&decl->initializer, token->next);
1149                 }
1150                 if (!is_typedef) {
1151                         if (!(decl->ctype.modifiers & (MOD_EXTERN | MOD_INLINE))) {
1152                                 add_symbol(list, decl);
1153                                 if (function_symbol_list)
1154                                         fn_local_symbol(decl);
1155                         }
1156                 }
1157                 check_declaration(decl);
1158
1159                 if (!match_op(token, ','))
1160                         break;
1161
1162                 ident = NULL;
1163                 decl = alloc_symbol(token->pos, SYM_NODE);
1164                 decl->ctype = ctype;
1165                 token = pointer(token, &decl->ctype);
1166                 token = declarator(token->next, &decl, &ident);
1167                 if (!ident) {
1168                         warn(token->pos, "expected identifier name in type definition");
1169                         return token;
1170                 }
1171
1172                 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1173
1174                 /* Function declarations are automatically extern unless specifically static */
1175                 base_type = decl->ctype.base_type;
1176                 if (!is_typedef && base_type && base_type->type == SYM_FN) {
1177                         if (!(decl->ctype.modifiers & MOD_STATIC))
1178                                 decl->ctype.modifiers |= MOD_EXTERN;
1179                 }
1180         }
1181         return expect(token, ';', "at end of declaration");
1182 }
1183
1184 void translation_unit(struct token *token, struct symbol_list **list)
1185 {
1186         while (!eof_token(token))
1187                 token = external_declaration(token, list);
1188         // They aren't needed any more
1189         clear_token_alloc();
1190 }